From owner-vxwexplo-process Wed Sep 1 01:32:34 1999 From: Paul Curson Date: Wed Sep 1 01:32:37 PDT 1999 Subject: Re: Design Constraints using VxWorks Subroutines Many thanks for the responses so far. I should have emphasized that these functions may be called from a number of different tasks (hence, they are a bit like shared library routines), or else they may be multiple (task) instances of the same subroutine. Hence, the functions must be re-enterent; yet I need to preserve state data unique to each calling task. A further point is that I am actually modelling the system in OO terms (I don't need convincing that this is a good way forward). In many respects it therefore makes sense to use an OO implementation language (of which C++ is the only contender here on VxWorks), but I don't require many of the more esoteric capabilities C++ offers, and don't really need to take much advantage of inheritance. I've seen other projects use subsets of C++, but it appears that there is a great temptation to keep on broadening this subset as time goes by. If I can manage my data encapsulation nicely, then I won't need C++. Paul From owner-vxwexplo-process Wed Sep 1 02:06:03 1999 From: Marios Charalambous Date: Wed Sep 1 02:06:06 PDT 1999 Subject: vxWorks I wish to go on the mailing list for the vxWorks section ----------------------------------------------------- Marios Charalambous AMIEE Software Engineer marios.charalambous@postimpressions.co.uk From owner-vxwexplo-process Wed Sep 1 03:15:24 1999 From: David Laight Date: Wed Sep 1 03:15:29 PDT 1999 Subject: vxWorks and DHCP All, I posed this about six months ago - and have finally got around to trying some experiments... I have a system where the ethernet interface is an a PCMCIA card that will not normally be present. Network connection will normally be by dial-out PPP. However there are times when an ethernet card will be inserted, and I want to be able to dynamically bring up the interface and use DHCP to obtain the IP address (etc). As an experiment, I booted the system with the ethernet card present and a static (non-DHCP) IP configuration, then ran the following from the target shell: dhcpLibInit( 67, 68, 4, 5, 3600, 30 ); lease = dhcpInit( ifunit( "xircom0" ), 1 ); sp (dhcpBind, lease, 1 ); (you will need to build a vxworks image containing the dhcp code - don't define INCLUDE_DHCPC, just include a reference to the routines.) What I expect this to do is contact the DHCP server, obtain some parameters and set the interface "xircom0" to use them. All I actually see is the local IP address changed to 10.0.x.x. I suspect this is a 'random' address assigned by the DHCP code so it can send out some UDP datagrams. The lease seems to match 'struct leaseData' in h/dhcp/dhcpcCommonLib.h and has the following set: initFlag 1 waitFlag 0 autoConfig 1 leaseSem leaseGood 0 leaseType 0 prevState 12 MAX_STATES+1 ??? currState 0 INIT timer initEpoch time(0) timeout 20 numRetry 3 eventHookRtn 0 cacheHookRtn 0 leaseReqSpec.waitsecs 5 leaseReqSpec.lease 3600 All the rest of the structure (about 60 words) is zero. Given some things that are not set up, I don't think the code has got very far at all. One very worrying thing I've spotted is the definition of structure members (eg if_info.name in dhcpc.h) as: char name[ sizeof "enxx" ]; This is horrid - how big is the array? Certainly my "xircom" interface name won't fit! Does anyone know: a) whether the test I'm doing is likely to suceed b) why it is failing on my system. 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 Sep 1 04:57:03 1999 From: dcn_trn18@hss.hns.com Date: Wed Sep 1 04:57:06 PDT 1999 Subject: Re: hi - ifdef and undef both works in VxWorks too ... Hi, Sorry to disturb u once again. Remember i once asked a silly question on VxWorks. Again a problem with me. I am using Profile to do performance measurement of the stack code i had ported. In my program takes a lot of inputs and i am unable to do my job. In manual it is given how to do profile of a netserver function which has an infinite loop and it requires no input. I had to make inputs also and also i had to measure performance of lots of functions in the stack. Can anyone tell how to do input and performance measurement simultaneously in Profile Regards Manohar "G. Peddareddappa" on 07/20/99 10:34:49 AM To: Krishna Manohar/HSS@HSS cc: Subject: hi - ifdef and undef both works in VxWorks too ... -- ----------------------------------------------------------- | G. Peddareddappa, | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ | | Networking Division | E-mail: gpr@teil.soft.net | | Design & Development, | Ph: 91-80-841-0221 | | Tata Elxsi Ltd. | Fax: 91-80-841-0219 | ----------------------------------------------------------- From owner-vxwexplo-process Wed Sep 1 05:57:09 1999 From: David Laight Date: Wed Sep 1 05:57:13 PDT 1999 Subject: Re: Design Constraints using VxWorks Subroutines > Subject: Re: Design Constraints using VxWorks Subroutines > Submitted-by: Paul Curson > > Many thanks for the responses so far. I should have emphasized that these > functions may be called from a number of different tasks (hence, they are a > bit like shared library routines), or else they may be multiple (task) > instances of the same subroutine. Hence, the functions must be re-enterent; > yet I need to preserve state data unique to each calling task. > I would consider using 'task variables' - see taskVarAdd() etc. You will probably also need to add a task delete hook - to tidy up. In practise a single task variable pointing to a malloced data area should suffice (for a single set of routines). You can, of course, just use the task id (taskIdSelf()) to search a private list of data areas for the one for the calling task... David vxworks ---------------------------------------------------------------- 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 Sep 1 09:16:30 1999 From: David Gasul Date: Wed Sep 1 09:16:35 PDT 1999 Subject: RE: A period in shell causes a reboot 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_01BEF49E.A9450F80 Content-Type: text/plain; charset="iso-8859-1" Problem was found - we run 68LC040 which does not have a FPU, BUT VxWorks was incorrectly built to a 68040 which does have an FPU. This caused the shell to reboot each time it was presented with a '.'. Thanks to Leonid Rosenboim for suggesting the possible problem source. Regards, David Gasul Telegate Ltd. 7 Haplada St., 60218 Or-Yehuda, Israel Voice: +972-3-5384634, Fax: +972-3-5335877 e-mail: davidg@telegate.co.il WWW: http://www.telegate.co.il -----Original Message----- From: Dheerendra Talur [SMTP:dtalur@fore.com] Sent: Wed 01 September 1999 18:09 To: David Gasul Subject: Re: A period in shell causes a reboot - If you input a period (the character '.') in the shell, the system reboots? - The period doesn't have to be the only character input. For example the - input 123.123 also causes a reboot. The only exception I know of is the - input - ~. if you're in rlogin session. I tried it on my system (5.3.1) and there is no reboot. ------_=_NextPart_001_01BEF49E.A9450F80 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable RE: A period in shell causes a reboot

Problem was found = - = we run 68LC040 which does not have = a FPU, BUT VxWorks was incorrectly built to a 68040 which does have an = FPU. This caused the shell to reboot

each time it = was presented with a '.'. Thanks to Leonid Rosenboim for = suggesting

the possible problem = source.

Regards,

David Gasul

Telegate Ltd.

7 Haplada St., 60218 = Or-Yehuda, Israel

Voice: +972-3-5384634, = Fax: +972-3-5335877

e-mail: davidg@telegate.co.il

WWW: http://www.telegate.co.il

    -----Original Message-----

    From:   Dheerendra Talur [SMTP:dtalur@fore.com]

    Sent:   Wed 01 September 1999 18:09

    To:     David Gasul

    Subject:       <= /B> Re: A period in shell causes a = reboot

    - If = you input a period (the character '.') in the shell, the system = reboots?

    - The = period doesn't have to be the only character input. For example = the

    - input = 123.123 also causes a reboot. The only exception I know of is the

    - input =

    - ~. if = you're in rlogin session.

    I tried = it on my system (5.3.1) and there is no reboot.

------_=_NextPart_001_01BEF49E.A9450F80-- From owner-vxwexplo-process Wed Sep 1 09:30:05 1999 From: "Ganesh S" Date: Wed Sep 1 09:30:09 PDT 1999 Subject: Unable to get ARP to respond to a ARP request This is a multi-part message in MIME format. ------=_NextPart_000_0008_01BEF4C5.F553D840 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hello, My End driver receives an ARP request packet and the driver calls the End receive rotine and passes the packet to IP.=20 I find the packet sent to IP is alright, but i do not get a reply to it from IP. When i check the icmpstatShow i find that there is an error for " destination unreachable " which i am not expecting. Could any one explain why ? VxWorks Tornado Thanks, Ganesh ------=_NextPart_000_0008_01BEF4C5.F553D840 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hello,
 
My End driver receives an ARP = request packet and=20 the driver
calls the End receive rotine and = passes the=20 packet to IP.
I find the packet sent to IP is alright, but i do = not get a=20 reply to it
from IP. When i check the icmpstatShow i find that = there is an=20 error
for " destination unreachable " which i am = not=20 expecting.
 
Could any one explain why ?
 
VxWorks Tornado
 
Thanks,
 
Ganesh
 
------=_NextPart_000_0008_01BEF4C5.F553D840-- From owner-vxwexplo-process Wed Sep 1 09:42:28 1999 From: Wayne Bond Date: Wed Sep 1 09:42:32 PDT 1999 Subject: Shared Memory support in VxWorks Does anyone know if it is possible to run shared memory (based on WRS shared memory option VxMP) using a proprietary bus, i.e. memory that is not directly addressable? This would need some work on our behalf, but is there a) support for this, and b) Has anyone done it themselves? regards Wayne Bond ECI Telecom From owner-vxwexplo-process Wed Sep 1 09:46:13 1999 From: Erin Swieringa Date: Wed Sep 1 09:46:17 PDT 1999 Subject: Raw ethernet packet hooks It appears that as of Tornado II and the new END drivers we no longer have any method to access raw ethernet packets!!!??? Please tell me there's an alternative to etherInput/Output hooks for the END drivers. etherInput/OutputHook routines are still available, and in fact have additional parameters for MUX/END drivers, but it appears that the END drivers do not utilize them - is this true? - Erin Swieringa Electronic Theatre Controls From owner-vxwexplo-process Wed Sep 1 13:54:56 1999 From: Xinyang Qian Date: Wed Sep 1 13:55:00 PDT 1999 Subject: Is VxMP platform VxWorks BSP dependent ? Hi, I wonder to know, if we got the SMNet ready, can we just buy VxMP from WRS without modifying anything and use it ? I heard from WRS field that VxMP is still platform dependent, which means, still some BSP modification is necessary, can any one provide me some info on it ?? Thanks, -xinyang ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Xinyang Qian E-Mail: xinyang@eng.sun.com Sun Microsystems Phone: 408.774-8298 420 N Mary Ave, #04 Fax: 408.616-5810 Sunnyvale, CA 94086 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ From owner-vxwexplo-process Wed Sep 1 15:08:51 1999 From: rcw@DynRes.com Date: Wed Sep 1 15:08:55 PDT 1999 Subject: symbols, searching in vxworks Hello VxWorks 'rs, I am trying to use the function symFindByName to locate a program variable without success and am wondering where I'm going wrong. I can use sysFindByName to search the global vxworks symbol table for the _vxworks_ variables. As an example, the following works from a loaded program or from Tornado: symFindByName (sysSymTbl, "symFindByName", &address, &type) This returns the address of the vxworks function 'symFindByName' by searching through the global symbol table (sysSymTbl) If I load up my program ( ->ld Date: Wed Sep 1 17:12:34 PDT 1999 Subject: RE: symbols, searching in vxworks You have a configuration in which - vxWorks is built w/ or loads it's own symbols onto the target - you are using the host shell - you have not enabled 'symbol table synchronization' symFindByName() runs on the target and can only find symbols that are on the target. ld() runs on the host and only downloads the code, not the symbols. lkup() runs on the host, so has access to the symbols. You can enable 'symbol table synchronization' (see the target server configuration dialog), however it runs as a background task and is really designed for human speed interaction. It is really not workable for programmatic access to the symbols. Bottom line is that under Tornado 2, you really can't depend on programmatic use of the VxWorks managed symbol table. I tried doing so (at a former employer, so can't share the code) and ended up chucking that approach and writing my own independent facility that required routines to 'register' their symbols of interest with a manager. Other routines that wanted to know about those symbols could then obtain them from that manager. > -----Original Message----- > Sent: Wednesday, September 01, 1999 3:09 PM > To: vxworks_users@csg.lbl.gov > Subject: symbols, searching in vxworks > > Submitted-by: rcw@DynRes.com > > Hello VxWorks 'rs, > > I am trying to use the function symFindByName to locate a > program variable without success and am wondering where I'm > going wrong. > > I can use sysFindByName to search the global vxworks symbol > table for the _vxworks_ variables. As an example, the > following works from a loaded program or from Tornado: > > symFindByName (sysSymTbl, "symFindByName", &address, &type) > > This returns the address of the vxworks function > 'symFindByName' by searching through the global symbol table > (sysSymTbl) > > If I load up my program ( ->ld the address of variables there, the function returns an error. > > If I use the function lkup from the Tornado shell, I _can_ > find my variables. > > Can anyone shed any light on this? My bottom line is that I > want to provide a way for a program to "monitor" itself in an > easily reconfigureable way. My thought is to create a text > file that can be read > by the program and the program can then locate variable > locations by using the sysFindByName David H. Olson Staff Engineer Oresis Communications, Inc. 14670 NW Greenbrier Parkway Beaverton, OR 97006 503.533.0717 x246 Fax:503.533.8223 dolson@oresis.com From owner-vxwexplo-process Thu Sep 2 04:00:30 1999 From: daemon@csg.lbl.gov Date: Thu Sep 2 04:00:35 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Sep 2 04:00:22 PDT 1999 Subject: FS: Force and Motorola board Subject: =?iso-8859-1?Q?R=E9f._:_ftpLib_broken_for_small_files?= Subject: Re: Book on VXworks? Subject: Re: CSP library for VxWorks? Subject: Re: Q: TCP throughput per architecture Subject: Re: CSP library for VxWorks? Subject: DC netif driver Subject: c++: templates overhead Subject: upgrading GCC? Subject: 38 Device Drivers for VxWorks, HPRT, AIX, HPUX, & Solaris Subject: Flash : TrueFFS on RFA. Subject: Help! How to make a task deamon ? Subject: VxWorks timerLib Subject: Re: How to increase the Stack Size?? Subject: Re: upgrading GCC? Subject: help! Subject: PPC604 PCI Problem (would appreciate help) Subject: UDP [socket] performance issue in VxWorks ? Subject: Re: malloc & cache safe Subject: Speech Engines for WindRiver VxWorks Subject: vxworks PCI driver as a loadable module Subject: parallel I/O on mv2700 or mv260x Subject: Re: PPC604 PCI Problem (would appreciate help) Subject: question! Subject: indicating receive packets to upper layer Subject: Re: Booting via Disk, config ? Subject: Re: Not suited to High Availability??? Subject: TrueFFS - reference Subject: Re: Read from files which are being appended Subject: Re: Flash : TrueFFS on RFA. Subject: are static variables global? Subject: Re: Design Constraints using VxWorks Subroutines Subject: Re: are static variables global? Subject: How to handle removable SCSI units Subject: Re: UDP [socket] performance issue in VxWorks ? Subject: Any network issues with going to Tornado-2 Subject: Reprogramming ROM from outside the box. Subject: How do I get the number of messages in a pipe Subject: Re: IPPC from GreenSpring Subject: END 802.5 SNAP to RFC 894 question Subject: Scalability dilemma Subject: Re: END 802.5 SNAP to RFC 894 question Subject: Re: vxWorks OK for High Availability systems ??? Subject: How to wake up select Subject: Re: Tornado II project failed Subject: Re: A period in shell causes a reboot Subject: Re: question! Subject: Re : upgrading GCC? Subject: Re: Design Constraints using VxWorks Subroutines Subject: Re: are static variables global? Subject: Re: are static variables global? Subject: sysAuxClkConnect Subject: Task stack consumption Subject: Re: vxWorks OK for High Availability systems ??? Subject: PThreads available in VXWorks? Subject: Re: How to increase the Stack Size?? Subject: Re: How to delay a task in micro seconds range Subject: END driver development - please help Subject: Re: Telnet into vxworks target Subject: Re: vxWorks and DHCP Subject: Re: Design Constraints using VxWorks Subroutines Subject: Tornado II with Tornado I Subject: Re: Problem while downloading vxworks on the target m/c Subject: Telnet into vxworks target ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: FS: Force and Motorola board Date: Mon, 23 Aug 1999 20:58:11 -0700 From: Innovative Research Technologies Organization: Innovative Research Technologies Message-ID: <37C21853.DE6065B9@innovative-research.com> We have the following boards coming available: FORCE SPARC CPU-5CE W/64MB MOTOROLA 68K MVME177 W/16MB and 32MB MOTOROLA PPC MVME1604 W/16MB and 80MB E-mail if interested or for more info. Marty@innovative-research.com --------------------------- Newsgroups: comp.os.vxworks Subject: =?iso-8859-1?Q?R=E9f._:_ftpLib_broken_for_small_files?= Date: Thu, 26 Aug 1999 13:25:27 +0200 From: Jerome.SOUQUIERES@dematic-postal.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: > There is a bug in ftpLib where occasionally small file transfers fail. [...] > bug is still present in Tornado 1.1 and 2.0. I have source for ftpLib > and am presently looking further into it. Please tell us the results of your research when you're done. We have similar problems with ftpXfer() : our application loads many tiny files (a few dozens bytes) and the transfers fail very often. However, retrying a failed transfer once (sometimes twice) eventually works. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Book on VXworks? Date: Thu, 26 Aug 1999 07:03:24 -0400 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <37C51EFC.B940C40@delphi.com> References: <7g2je6$qd7@catapult.gatech.edu> <7g4kh5$ssi$1@nnrp1.dejanews.com> <37e8e823.87554044@news.zetnet.co.uk> <37C43430.4275BF25@efi.com> Try: http://bigsun.wbs.net/homepages/g/u/y/guylancaster/ Speaking only for myself, Joe Durusau Prasoon Kumar wrote: > Where is Guy Lacestor's page???? > > -- Prasoon > EFI, Inc. > > "Ag@whd" wrote: > > > On Tue, 27 Apr 1999 15:19:08 GMT, bwedding@my-dejanews.com penned > > these words: > > > > >In article <7g2je6$qd7@catapult.gatech.edu>, > > > ccastae@acmex.gatech.edu (AMAE) wrote: > > > > > >> Could someone please recommend to me an easy to read and comperhensive > > >> book on VXworks? > > >> I come from a unix/linux background with know knowledge of RTOSs. > > > > > >Unfortunately, the best book is the VxWorks Programmer's Guide. It starts > > >with a tutorial on RTOS and continues as deep as you want to go. For > > >something widely available, I recommend Micro C /OS-II which is available at > > >amazon.com or from the author at www.ucos-ii.com. It contains an excellent > > >tutorial on the internals of an RTOS as well as all source code. It > > >certainly isn't as extensive as VxWorks, but it is a good start. > > > > There is also Guy Lancaster's page (link from above) with a micro > > tcp/ip stack uCIP and there is a listserver for uCOS discussions too. > > > > Adrian > > > > WWW WWW Adrian Gothard > > WWW ww WWW White Horse Design > > WWWWWWWWWW > > WWWW WWWW whd@zetnet.co.uk, http://www.users.zetnet.co.uk/whd > > --- > > It's good to be me, yes, Heh heh heh, 'purrr', 'purrr'. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CSP library for VxWorks? Date: Thu, 26 Aug 1999 12:52:04 GMT From: rmeenaks@olf.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7q3d9f$l39$1@nnrp1.deja.com> References: <37C4F982.C72789D5@computer.org> In article <37C4F982.C72789D5@computer.org>, "Øyvind Teig" wrote: > Sirs > > I am used to programming in the "runnable" CSP subset > and programming language occam. At present for a Texas > TMS320C3x DSP. > > A connected processor to the DSP is a 386EX running > VxWorks. I'd like to have the same CSP subset > functionality in VxWorks. > > CSP is a process algebra where PROCESSES can only > communicate over CHANNELS. All state machines can > be modeled, also monitors, semaphores etc., but they > are not needed a primitive elements as such. > > I would expect such a library to contain something like > the SGS-Thomson C toolset for transputer did, with > functional groups for process control and channel > communication. > > It would constitute a clean message-based layer on > top of VxWorks. > > Does anybody know of any such library? You could try SPOC (Southampton Portable Occam Compiler) and maybe KROC (Kent Retargetable Occam Compiler). SPOC converts Occam-2 code into ANSI C, so any ANSI C compiler will work. KROC converts transputer assembler over to the target machine. You can find both at http://www.hensa.ac.uk/parallel. I would suspect that SPOC is what you are probably looking for. Ram Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Q: TCP throughput per architecture Date: Thu, 26 Aug 1999 13:14:30 GMT From: bill.beckwith@realtime-corba.com (Bill Beckwith) Organization: Objective Interface (http://www.ois.com) Message-ID: <8E2E5E0C2rwbnewsoiscom@sigma.ois.com> References: <379C620A.92A8FDC3@bitband.com> <37C4457B.A7AA5134@analogic.com> Smail Fenni wrote in <37C4457B.A7AA5134@analogic.com>: > >Well, I am currently doing such a performance analysis and also using UDP >packets. I don't have any final numbers yet but what I am getting is quite >deplorable (sub 2Mbps)! I am using an 860T with a vxWorks 5.3.1 . I have >gotten feedback from WRS however that their numbers were in the range of >28Mbps with a similar setup. Is anyone out there doing similar testing >(using 100BT ethernet) ? Yes. We develop a real-time CORBA ORB and benchmark the socket performance of various boards in order to define a baseline for the ORB performance. I have run small message latency tests on our MBX860 25 MHz board. I was getting a little over 3 millisec round-trip time between two threads for a 64 byte TCP packet through the loopback adapter. I get a little over half this time when the other system is a 143 MHz UltraSPARC running Solaris 2.5. I haven't had a chance to run our throughput tests on this board yet though. I'll post the results when we do. >Leonid Rosenboim wrote: > >> A friend asked me to recommend the absolutely best CPU architecture for >> an application which requires heavy TCP throughput. Also, he wonders how >> fast SENS TCP can go over multiple 100baseT interfaces with current >> VxWorks. In my experience the best CPU architecture for fast TCP is going to be the one that has the best integer performance. I.e. something like a high clock rate 750 board. The performance of the boards networking support chips is another significant factor. A good NIC chip set can reduce the CPU time consumed by the protocol stack. Thus are you really asking for the best board architecture for heavy TCP throughput? >> If you got any performance data you could share, it would enable us to >> make a sensible comparison. We are interested in Mbits/sec using large >> packets with a large number of concurrent connections. >> If you have such performance data handy, please also state the CPU type >> you are using and possibly other related info. Will do. - -- Bill --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CSP library for VxWorks? Date: Thu, 26 Aug 1999 15:37:35 +0200 From: "Øyvind Teig" Organization: Navia Maritime AS, division Autronica Message-ID: <37C5431E.E914D5A2@computer.org> References: <37C4F982.C72789D5@computer.org> <7q3d9f$l39$1@nnrp1.deja.com> rmeenaks@olf.com wrote: > In article <37C4F982.C72789D5@computer.org>, > "Øyvind Teig" wrote: > ... > > It would constitute a clean message-based layer on > > top of VxWorks. > > > > Does anybody know of any such library? > > You could try SPOC (Southampton Portable Occam Compiler) and maybe KROC > (Kent Retargetable Occam Compiler). SPOC converts Occam-2 code into > ANSI C, so any ANSI C compiler will work. KROC converts transputer > assembler over to the target machine. You can find both at > http://www.hensa.ac.uk/parallel. I would suspect that SPOC is what you > are probably looking for. > > Ram > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. Nice to know that somebody else than myself is suggesting SPoC. We already use it with success in one live product, and on this new one! But I can't see how that is the solution here, since the 386ex running VxWorks will have user code written in C only. I have described our SPoC experiences at: Real-Time Magazine: Non-preemptive occam in DSP real-time system http://www.realtime-info.be/encyc/magazine/98Q3/index983.htm WoTUG-22 conference and proceedings (ISBN 90 5199 480 X) Another side of SPoC: occam's ALTer ego dissected with PC-lint http://www.keele.ac.uk/depts/cs/events/wotug22/ WoTUG-21 conference and proceedings (ISBN 90 5199 391 9): PAR and STARTP Take the Tanks http://www.hensa.ac.uk/parallel/groups/wotug/wotug21/papers.html Oyvind Teig --------------------------- Newsgroups: comp.os.vxworks Subject: DC netif driver Date: 26 Aug 1999 06:35:23 -0800 From: "Sharon Veasie" Message-ID: <37c5429b@apathy.ibsystems.com> Reply-To: "Sharon Veasie" Sender: "Sharon Veasie" Does anyone know what the default property values are for the DC netif driver in order to create a boot image? Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: c++: templates overhead Date: Mon, 23 Aug 1999 18:10:51 -0400 From: Lei Zhang Organization: Computer Science Department, The Johns Hopkins University Message-ID: References: <37ad5095.14948058@news.cadvision.com> <37BB58FA.4C61@gte.net> <01beea3d$f2e5be80$e12ac222@my0250> <37BC2E1F.49F6506@mindspring.com> <7ph13t$19p@sis.cambridge.arm.com> <37BC559E.4910F2D3@mindspring.com> Folks, I am trying to convince my co-workers here not to use templates in our embedded application(telecom). I only have the vague idea that current implementation of templates incur large memory and run-time overhead. Would anybody care providing more detailed explaination? We are using VxWorks. Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: upgrading GCC? Date: Thu, 26 Aug 1999 18:18:05 -0700 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <37C5E74D.5E4A@despam.pangea.ca> Hi Folks, Has anyone any experience with upgrading the host tools to later versions? I am thinking particularly of GCC. I would really like to run our code tree through 2.9.5. Are there any gotchas? - -het - -- "you don't really appreciate how smart a moron is until you try to design a robot..." -a robotics designer Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: 38 Device Drivers for VxWorks, HPRT, AIX, HPUX, & Solaris Date: Wed, 25 Aug 1999 17:33:24 -0700 From: "Compware Corp." Organization: Compware Corp. Message-ID: <37C48B54.CC4B0062@compware.com> - --------------20D3B4D2A3F1103097B6B997 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit 38 "Off-The-Shelf" Device Drivers for "Industry Pack", PMC, PCMIP, CompactPCI, & VME Devices and Controllers Including: MIL1553, Serial-async(2-ch.), Serial-async(8-ch.), SDLC, BiSync, I EEE-488, Ethernet, Multi-Ethernet, FDDI, A-D, D-A, Digital-i/o, and more... Please visit our web site for information & pricing on Device Driver Packages for VxWorks, HPRT, AIX, HPUX, & Solaris -- http://www.compware.com http://www.sbs-mio.com Compware welcomes your requirements for new driver development. On-time delivery, typically at 50% of in-house development costs. Delivering "Off-The-Shelf" Unix & Real-Time Device Drivers & Turn-key Systems since 1985. - -- ------------------------------------------------- mailto:sales@compware.com http://www.compware.com ------------------------------------------------- - --------------20D3B4D2A3F1103097B6B997 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit 38  "Off-The-Shelf" Device Drivers for

"Industry Pack", PMC, PCMIP,  CompactPCI, & VME Devices and Controllers
 

 
Including:

MIL1553,
Serial-async(2-ch.), Serial-async(8-ch.), SDLC, BiSync, I
EEE-488,
Ethernet, Multi-Ethernet,
FDDI,
A-D, D-A, Digital-i/o, and more...
 
 
 
Please visit our web site for information & pricing on Device Driver Packages for VxWorks, HPRT, AIX, HPUX, & Solaris  --

http://www.compware.com

http://www.sbs-mio.com

Compware welcomes your requirements for new driver development.

On-time delivery, typically at 50% of in-house development costs.

Delivering "Off-The-Shelf" Unix & Real-Time Device Drivers & Turn-key Systems since 1985.

--
 -------------------------------------------------

  mailto:sales@compware.com

  http://www.compware.com
  ------------------------------------------------- - --------------20D3B4D2A3F1103097B6B997-- --------------------------- Newsgroups: comp.os.vxworks Subject: Flash : TrueFFS on RFA. Date: Fri, 27 Aug 1999 00:46:11 GMT From: philippe.barthelemy@usa.net Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7q4n4h$io5$1@nnrp1.deja.com> Hi, I have looked in the doc, but I did not manage to find a definitve answer on the following question : Can I use TrueFFS on a Resident Flash Array device ? ( "Resident Flash Array" as opposed to "Linear Flash Memory" ) Thanks to the one who will answer.... - --Philippe Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Help! How to make a task deamon ? Date: Tue, 24 Aug 1999 10:33:47 +0900 From: "°íÁ¤±¹" Organization: Pusan National University Message-ID: <7pssos$guf$1@jagalchi.cc.pusan.ac.kr> Hi, vxWorks users I am now studying RTOS and would like to know how to make a task daemon. But I don't know how to do it.. Could you please tell me how to do it in vxWorks ? Hope this helps. Jeong-Gook Koh --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks timerLib Date: Tue, 24 Aug 1999 20:32:05 GMT From: mduong@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7puvg5$fh7$1@nnrp1.deja.com> Hello, I have an application that needs to use the following routines: timer_create(...); timer_connect(...); timer_settime(....); After building the project, I download the LAC.out and get the following error messages: Unresolved External Symbols Errors while downloading C:/LAC.out: _timer_create _timer_connect _timer_settime I included the timers.h which has the the following codes: extern int timer_create (...); extern int timer_connect (...); extern int timer_settime (....); The linker couldn't find the these functions. Do I need to download any module prior to LAC.out or do I need to include some header files? Can anyone please help? Thanks, Mike Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to increase the Stack Size?? Date: Fri, 27 Aug 1999 02:11:28 GMT From: lingsiva@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7q4s49$m59$1@nnrp1.deja.com> References: <7q2qge$8nk$1@nnrp1.deja.com> <935658000.275567@news.vbs.at> > Why not just spawn a task? > > e. g. > > taskSpawn("tStack",100,0,81920,TaskStack,0,0,0,0,0,0,0,0,0,0) > > to spawn a task with your function using 80kByte of stack I already knew about this. But, as i am just 1 and half month old to this OS, i am not sure whether taskSpawn is the norm to start a parent task. In my case, TaskStack is the parent task and it spawns several other tasks. Am i correct in saying that to start a parent task of desired stack size, the only way is to spwan it from usrRoot() at the end of the vxworks booting. regards, s.shiva Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: upgrading GCC? Date: Thu, 26 Aug 1999 22:33:24 -0400 From: "Don Bowman" Organization: UUNET Canada News Transport Message-ID: <7q4tdq$qob$1@demon.uunet.ca> References: <37C5E74D.5E4A@despam.pangea.ca> Harvey Taylor wrote in message news:37C5E74D.5E4A@despam.pangea.ca... > Hi Folks, > Has anyone any experience with upgrading the host tools > to later versions? I am thinking particularly of GCC. > I would really like to run our code tree through 2.9.5. > Are there any gotchas? its not really that hard. Configure the new gcc with configure --target=XXX-wrs-vxworks [eg mips-wrs-vxworks]. Make sure you get the cross binutils [eg the asmips etc] instead of the native one while this builds. Compile it, and then install into the Tornado tree so that 'gcc' becomes ccXXX [eg ccmips], and the gcc-lib directory goes under host//lib/gcc-lib/ Make sure your GCC_EXEC_PREFIX points to this directory. You might run into a small trouble with libgcc.a, which can be rectified by pulling two floating point objects out of the lib, and linking them explicitly to your BSP. Then you're done. I've done this with MIPS, i960, PowerPC, x86. - --don --------------------------- Newsgroups: comp.os.vxworks Subject: help! Date: Sun, 22 Aug 1999 16:12:38 -0400 From: Lei Sun Organization: University of Central Florida Message-ID: <37C059B6.23760C57@pegasus.cc.ucf.edu> Hi: Does anybody has experience on how to use VEM multiprocessor system with vx-works on it? can anybody tell me how to configure these whole things? thanks ! lei sun --------------------------- Newsgroups: comp.os.vxworks Subject: PPC604 PCI Problem (would appreciate help) Date: Sun, 22 Aug 1999 14:28:25 -0700 From: jim Organization: We don't need no stinkin' organization! Message-ID: <37C06B79.32EC@ns.net> Reply-To: mailbox2@ns.net Dear Fellow Firmware Engineers, I'm using Tornado 2, VxWorks 5.4, with a Motorola PPC board, MTX604, which has 3 PCI slots, and a "Raven" PCI controller chip. The problem is our custom PCI card, which works just fine in other (Intel/NT) platforms. I can read the vendor/device ID PCI config registers, and find our card at bus=0, device=0x10, func=0. I can then read and write all of its config registers. However, no matter what I write into the IOBASE or MEMBASE register, to map the chip's register set to PCI IO or Memory space, I cannot see the chip registers. I tried both I/O and Memory: I/O: 1. Wrote 0x1000 to the IOBASE register at 0x14. 2. Wrote 0x1100 to the IOBASE register at 0x18. 3. Enabled I/O slave access with 0x01 bit to register 0x04. 4. Read memory at 0x80000000 + 0x1000 and don't get the chip registers. MEM: 1. Wrote 0x1000 to the MEMBASE register at 0x1C 2. Enabled Mem slave access with 0x02 bit to register 0x04. 3. Read memory at 0xC0000000 + 0x1000 and don't get the chip registers. According to the BSP docs, 0x80000000 = PCI I/O Space, and 0xC0000000 = PCI Memory Space. I can read the Raven PCI Bridge registers at the memory locations shown in the map, and the registers indicate that the above I/O and memory mapping (0x80000000 and 0xC0000000) is programmed. What am I doing wrong??? By all that is holy, this should work! Thanks, Jim. - ------------------------------------------------------------------- Here is my probe output for the PCI space: PCI BUS PROBE ------------------------- (only !=0 && !=0xFFFFFFFF) Bus[0x00] Device[0x00] ID = 0x48011057 Offset[0x00] = 0x48011057 Offset[0x04] = 0x02800006 Offset[0x08] = 0x06000003 Offset[0x10] = 0x00000001 Offset[0x14] = 0x3C000000 Offset[0x80] = 0x80008FFE Offset[0x84] = 0x800000F2 Offset[0x88] = 0x8FFF8FFF Offset[0x8C] = 0x800000E2 Offset[0x90] = 0xFFF0FFF0 Offset[0x98] = 0xFFF0FFF0 Bus[0x00] Device[0x0B] ID = 0x056510AD Offset[0x00] = 0x056510AD Offset[0x04] = 0x02000007 Offset[0x08] = 0x06010010 Offset[0x0C] = 0x00800000 Offset[0x40] = 0xEF000420 Offset[0x44] = 0x00780000 Offset[0x48] = 0x00000001 Offset[0x4C] = 0x00043300 Offset[0x80] = 0x800001E0 Offset[0xF0] = 0x0E920000 Offset[0xF4] = 0x00000101 Bus[0x00] Device[0x0C] ID = 0x000F1000 Offset[0x00] = 0x000F1000 Offset[0x04] = 0x02000000 Offset[0x08] = 0x01000004 Offset[0x10] = 0x00000001 Offset[0x3C] = 0x40110100 Offset[0x80] = 0x000000C0 Offset[0x84] = 0x0F000000 Offset[0x88] = 0x00060000 Offset[0x8C] = 0x02000080 Offset[0x90] = 0x10000000 Offset[0x94] = 0xFFFFFF00 Offset[0x98] = 0x4001F000 Offset[0xA0] = 0x50000000 Offset[0xC4] = 0x0F700000 Offset[0xCC] = 0x00000057 Offset[0xE0] = 0x98804A41 Offset[0xE4] = 0xDFFEEDF7 Offset[0xE8] = 0x3011260D Offset[0xEC] = 0xEDEEDBFF Offset[0xF0] = 0xA4194841 Offset[0xF4] = 0x7C7FEB3F Offset[0xF8] = 0xD01CC463 Offset[0xFC] = 0xE7F7FFFF Bus[0x00] Device[0x0E] ID = 0x00091011 Offset[0x00] = 0x00091011 Offset[0x04] = 0x02800005 Offset[0x08] = 0x02000022 Offset[0x0C] = 0x0000FF20 Offset[0x10] = 0x01020001 Offset[0x2C] = 0x34561057 Offset[0x3C] = 0x28140100 Bus[0x00] Device[0x10] ID = 0x1028103C <--- this is our card Offset[0x00] = 0x1028103C Offset[0x04] = 0x02100000 Offset[0x08] = 0x0C04000B Offset[0x14] = 0x00000001 Offset[0x18] = 0x00000001 Offset[0x2C] = 0x0004103C Offset[0x34] = 0x00000050 Offset[0x3C] = 0x00200100 Offset[0x40] = 0x00260000 Offset[0x50] = 0x00010001 TACHYON HBA BASE ADDRESS REGISTERS-------- (OUR CARD) BAR0[0x10]: Before=0x00000000 AfterWriteFF=0x00000000 These 3 are for registers: BAR1[0x14]: Before=0x00000001 AfterWriteFF=0xFFFFFF01 <-- lower I/O space BAR2[0x18]: Before=0x00000001 AfterWriteFF=0xFFFFFF01 <-- upper I/O space BAR3[0x1C]: Before=0x00000000 AfterWriteFF=0xFFFFFE00 <-- memory space BAR4[0x20]: Before=0x00000000 AfterWriteFF=0x00000000 BAR5[0x24]: Before=0x00000000 AfterWriteFF=0xFFFE0000 <-- ROM space RAVEN PCI CONTROLLER REGISTERS------------ MEM[0xFEFF0000]= 0x10574801 MEM[0xFEFF0004]= 0x00030000 MEM[0xFEFF0008]= 0x0020FFFE MEM[0xFEFF000C]= 0x00000000 MEM[0xFEFF0010]= 0x000000BE MEM[0xFEFF0014]= 0x00000000 MEM[0xFEFF0018]= 0x00000000 MEM[0xFEFF001C]= 0x00000000 MEM[0xFEFF0020]= 0x00000000 MEM[0xFEFF0024]= 0x00000000 MEM[0xFEFF0028]= 0x00000000 MEM[0xFEFF002C]= 0x00000A0F MEM[0xFEFF0030]= 0x0700FF01 MEM[0xFEFF0034]= 0x00000000 MEM[0xFEFF0038]= 0x00000000 MEM[0xFEFF003C]= 0x00000000 MEM[0xFEFF0040]= 0xFD00FDFF MEM[0xFEFF0044]= 0x030000C2 MEM[0xFEFF0048]= 0xFE00FE7F MEM[0xFEFF004C]= 0x020000C0 This is per Motorola BSP docs MEM[0xFEFF0050]= 0xC000FCFF <-- Memory Space MEM[0xFEFF0054]= 0x400000C2 (wraps so PCI address is 0-based) MEM[0xFEFF0058]= 0x8000BF7F <-- I/O Space MEM[0xFEFF005C]= 0x800000C0 (same gimmick: wraps to 0-based) MEM[0xFEFF0060]= 0x000FFFFF MEM[0xFEFF0064]= 0x0000FFFF MEM[0xFEFF0068]= 0x000FFFFF MEM[0xFEFF006C]= 0x0000FFFF MEM[0xFEFF0070]= 0x00000000 MEM[0xFEFF0074]= 0x00000000 MEM[0xFEFF0078]= 0x00000000 MEM[0xFEFF007C]= 0x00000000 - ----------------------------------------------------------- Here is the code: #include #include int ProbePci(void) { int bus; long data; long orig; int dev; int off; int bar; long * base; char * adr; STATUS status; printf("\nPCI BUS PROBE -------------------------\n"); for (bus=0; bus<3; bus++) { for (dev=0; dev<32; dev++) { pciConfigInLong(bus,dev,0,0,&data); if ((data != 0) && (data != 0xFFFFFFFF)) { printf("Bus[0x%02X] Device[0x%02X] ID = 0x%08X\n",bus,dev,data); for (off=0; off<256; off=off+4) { pciConfigInLong(0,dev,0,off,&data); if ((data != 0) && (data != 0xFFFFFFFF)) { printf(" Offset[0x%02X] = 0x%08X\n",off,data); } } } } } printf("\nTACHYON HBA BASE ADDRESS REGISTERS--------\n"); for (bar=0; bar<6; bar++) { off = bar * 4; pciConfigInLong(0,0x10,0,0x10+off,&orig); pciConfigOutLong(0,0x10,0,0x10+off,0xFFFFFFFF); pciConfigInLong(0,0x10,0,0x10+off,&data); pciConfigOutLong(0,0x10,0,0x10+off,orig); printf("BAR%i[0x%02X]: Before=0x%08X AfterWriteFF=0x%08X\n", bar,0x10+off,orig,data); } printf("\nRAVEN PCI CONTROLLER REGISTERS------------\n"); base = (long *)0xFEFF0000; for (off=0; off<32; off++) { adr = (char *)&base[off]; sysProbeErrClr(); status = vxMemProbe(adr,VX_READ,4,(char *)&data); if (status == OK) { printf("MEM[0x%08X]= 0x%08X\n",(long)(void *)adr,data); } } return(0); } int TachMemSetup(int membase) { int data; /* core reset procedure */ /* */ pciConfigOutByte(0,0x10,0,0x44,0x01); pciConfigInByte(0,0x10,0,0x44,&data); pciConfigInByte(0,0x10,0,0x44,&data); pciConfigInByte(0,0x10,0,0x44,&data); pciConfigInByte(0,0x10,0,0x44,&data); pciConfigOutByte(0,0x10,0,0x44,0x00); /* clear and disable device */ /* */ pciConfigOutLong(0,0x10,0,0x04,0xFFFF0000); /* memory mapped setup */ /* */ pciConfigOutLong(0,0x10,0,0x1C,membase); /* set cache line size of 8 */ /* */ pciConfigOutByte(0,0x10,0,0x0C,0x08); /* turn on Memory accesses and PCI Master */ /* */ pciConfigOutWord(0,0x10,0,0x04,0x0006); return(0); } int TachIoSetup(int iobase) { int data; /* core reset procedure */ /* */ pciConfigOutByte(0,0x10,0,0x44,0x01); pciConfigInByte(0,0x10,0,0x44,&data); pciConfigInByte(0,0x10,0,0x44,&data); pciConfigInByte(0,0x10,0,0x44,&data); pciConfigInByte(0,0x10,0,0x44,&data); pciConfigOutByte(0,0x10,0,0x44,0x00); /* clear and disable device */ /* */ pciConfigOutLong(0,0x10,0,0x04,0xFFFF0000); /* I/O mapped setup */ /* */ pciConfigOutLong(0,0x10,0,0x14,iobase + 0x01); pciConfigOutLong(0,0x10,0,0x18,iobase + 0x101); /* set cache line size of 8 */ /* */ pciConfigOutByte(0,0x10,0,0x0C,0x08); /* turn on I/O access and PCI Master */ /* */ pciConfigOutWord(0,0x10,0,0x04,0x0005); return(0); } int ProbeIo(long start, long len) { int off; long data; char * adr; char * base; base = (char *)(void *)start; for (off=0; off Organization: SBC Internet Services Message-ID: <37C22BFD.BA8873FD@gte.net> Reply-To: amalpani@gte.net Has anyone faced performance issues when dealing with UDP sockets under VxWorks ? I'm writing a real-time-voice-traffic application; hence small packets (approx 50 bytes) at high frequency rates (say 20 ms) are to be despatched onto/received from the network. I noticed high CPU utilization; and on closer inspection and some test coding it appeared as if the socket function call itself ("sendto" in this case) is soaking up the majority of the CPU time.... Also tried the zbuf (zero-copy buffer) socket library; but as expected it didnt help much since it was clearly mentioned in manual that memcpy's are likely to be used (rather than avoided) when dealing with small packets...the netTask does run up some CPU usage; but a huge amount of the calling tasks' CPU consumption seems be done by the socket function call (sendto) Also tried a VxWorks image with UDP checksumming turned off, but that did'nt help either Anyone seen this before or have any clue as to what is going on ? Any help would be greatly appreciated... Thanks. DSN --------------------------- Newsgroups: comp.os.vxworks Subject: Re: malloc & cache safe Date: Tue, 24 Aug 1999 04:06:18 -0400 From: "Don Bowman" Organization: UUNET Canada News Transport Message-ID: <7ptk0e$5qm$1@demon.uunet.ca> References: <7pj66k$dmb$1@nnrp1.deja.com> wrote in message news:7pj66k$dmb$1@nnrp1.deja.com... > I am writing a DMA Driver API . > The user who uses this API may pass a malloc'd (NOT cacheDmaMalloc'd) > buffer and request the API to make it Cache safe and as well return the > physical address. > I am not very sure whether malloc returns physical address or virtual > address. I have got answers from the newsgroup some saying malloc know > only about virtual address..others malloc always returns physical > address. But I am assuming a worst-case scenario that malloc returns > virtual address. > > I would like to know how to do this translation and whether can I > invalidate cache for a section starting at user passed address thru > user passed address + number of bytes. It depends on your BSP whether physical == virtual. I think all VxWorks architectures have this as true for memory, but not for the bus. Unfortunately to make something safe for DMA, you have to ensure it is aligned to a cache line at the beginning and also at the end. If not, you are sharing a cache line with something that might flush out on top. This requires mallocing more than you need up to the next cache line, and moving the pointer. This can be trouble. 'memalign' can help, which mallocs memory aligned to a given boundary. Malloc of course returns the virtual address [since the processor is capable of writing to it]. --------------------------- Newsgroups: comp.os.vxworks Subject: Speech Engines for WindRiver VxWorks Date: Mon, 30 Aug 1999 14:10:33 -0400 From: "Rick Sullivan" Message-ID: <7qeheu$8aq$1@autumn.news.rcn.net> Is anybody aware of speech engines (Speech Recognition and/or Text To Speech) that are available for WindRiver's VXWorks real time embedded operating system? Thanks in advance, Rick Sullivan Vox2, Inc. --------------------------- Newsgroups: comp.os.vxworks Subject: vxworks PCI driver as a loadable module Date: Mon, 30 Aug 1999 19:13:25 +0100 From: "Rui Prior" Message-ID: <37cac92d@readers.ip.pt> From previous posts to this group, I understand it is necessary to do the PCI initialization of a device in the sysLib.c file of the BSP. I'd like, however, to make my driver a dynamically loadable module, so my question is: isn't there really any way of delaying the PCI configuration of the device until runtime? If there is, how is it done? Thanks in advance for the answers. Rui Prior --------------------------- Newsgroups: comp.os.vxworks Subject: parallel I/O on mv2700 or mv260x Date: Mon, 30 Aug 1999 11:20:07 -0700 From: Stefan Schaal Organization: USC & ERATO (JST) Message-ID: <37CACB55.DEE5F1C@usc.edu> Reply-To: sschaal@usc.edu Hello, we try to set the parallel port I/O of our mv2700/mv2604 boards to READ. Unfortunately, we could find out how to program the super I/O chip pc87303, which controls the parallel I/O, to go into READ mode (by default it is WRITE). The pointers to the config/control/data registers are easily found in the mv2600.h header in the BSP. But it seems to be impossible to reconfigure the parallel I/O directly through these registers. Has anybody ever programmed the pc87303 super I/O chip? Any information how to deal with the parallel I/O would be highly appreciated! Best wishes, - -Stefan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PPC604 PCI Problem (would appreciate help) Date: Tue, 24 Aug 1999 10:25:21 GMT From: paulvroberts@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7ptrua$nkh$1@nnrp1.deja.com> References: <37C06B79.32EC@ns.net> In article <37C06B79.32EC@ns.net>, mailbox2@ns.net wrote: > Dear Fellow Firmware Engineers, > > I'm using Tornado 2, VxWorks 5.4, with a Motorola PPC board, > MTX604, which has 3 PCI slots, and a "Raven" PCI controller chip. > > The problem is our custom PCI card, which works just fine in > other (Intel/NT) platforms. I can read the vendor/device ID > PCI config registers, and find our card at bus=0, device=0x10, > func=0. I can then read and write all of its config registers. > > However, no matter what I write into the IOBASE or MEMBASE > register, to map the chip's register set to PCI IO or Memory > space, I cannot see the chip registers. > > I tried both I/O and Memory: > This might be barking up the wrong tree, but how did you check the registers? I found that the d() command under the Tornado shell gave funny results under certain circumstances on the MVME2700 (PowerPC750) - just lots of ff's - whereas the d() command under the VxWorks target shell worked fine. If you are using Tornado for your tests, try the same tests under the VxWorks target shell. Paul Roberts. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: question! Date: Mon, 30 Aug 1999 16:45:24 -0400 From: Lei Sun Organization: University of Central Florida Message-ID: <37CAED64.AE2ACEB4@pegasus.cc.ucf.edu> Hi: Can anybody tell me what is the "tick", how to convert it into time, suppose I have a MC68k cpu with 33MHz, what is the relationship between tick and clock rate? I want to use taskDelay() to delay a task by 1 sec, but I need to use "tick " as parameter, How do I know that how many ticks equal 1 sec. and also , what does the constant "CLOCKS_PER_SEC" and "_SC_CLK_TCK " represent for? thanks a lot! lei --------------------------- Newsgroups: comp.os.vxworks Subject: indicating receive packets to upper layer Date: Tue, 24 Aug 1999 16:26:26 +0200 From: Erez Bar-Tal Organization: Internet Gold, ISRAEL Message-ID: <37C2AB92.51CB09AC@galileo.co.il> Hi i am writing an END driver. can anybody pls tell me how to build the mBlk/cBlk/cluster structre (i dont mean how to use the netBuLib functions) but what parameters shuld i assign to the mBlk structure before passing it over to the protocol ? is the data passed should contain the ethernet header ? where pMblk->mBlkHdr.mData should points to ? etc. any help will be appriciated thanks Erez erezb@galileo.co.il --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting via Disk, config ? Date: Mon, 23 Aug 1999 08:17:50 -0500 From: Frank Minich Organization: NCP Message-ID: <37C149FE.12F@ti.com> References: Reply-To: fminich@ti.com Sounds like you're booting VxWorks on a PC. The PC has such limited NVRAM that the boot image can't store the boot parameters in NVRAM, so you always get as default the boot parms built into your boot image. We got tired of this so built a bogus NVRAM out of the last sector of the floppy. Now, when you change the parms they stay changed. Email me if you want more details. Frank Minich fminich@ti.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Not suited to High Availability??? Date: Mon, 30 Aug 1999 19:50:11 -0400 From: "Jeff Szczepanski" Organization: Time Warner Road Runner - Rochester NY Message-ID: References: <7q6997$lh9$1@nnrp1.deja.com> <7qelcb$630$1@nnrp1.deja.com> Without attacking point by point, I see some circular reasoning in your arguments. Especially your first two, somewhat conflicting responses on the drivers being 'inside' the protection ring of the kernel. I would spend time rebutting, but I think Leonid Rosenboim's post did a nicer job of explaining my points, and I will stand down on that basis. :-> Regards, Jeff Alex wrote in message news:7qelcb$630$1@nnrp1.deja.com... > In article , > "Jeff Szczepanski" wrote: > > I didn't read the article, but I would have to say that the author is > > working from some bad assumptions here. Was this written by someone > who is > > associated with LynxOS? :-> > > > > 1) The first assumption from this claim is that the kernel has to be > > completely bug free for the application. If the kernel crashed in the > > process model case you are hosed anyway. Extreme example: Windows NT > is > > fully process based & crashes fairly regularly! > > But I believe most drivers live in or at the NT & W95 kernel! > This is probably the cause for most of the MS OSs crashes (ignoring > application s/w for the moment). > > > 2) Drivers tend to be inside the protection ring of the process > model...How > > many real embedded systems do you know that don't have atleast one > custom > > driver for some hardware interface/mechanism? ie: Atleast some of the > > functionality will be inside of ring of protection. > > With a decent architecture you can divide & rule-keep the > functionality required at this level at a minimum, plus > keep the time spent at this level at a minimum. > > > 3) OK so process protection traps your application doing the real > work, and > > safely terminates it things go bad...now what? Restart the > application. > > First that's still not 'high availability' since it didn't prevent the > > failure > > By definition you can't stop the failure, as it is a bug! > The question is what & how to to if a failure occurs. > This is what gives you higher availability compared to other > architected OS. > > >and I can get the same functionality with a simple watchdog > timer > > and an automatic hardware reset when it expires. > > Not quite the same functionality. Restarting a process should > always be faster than resetting the hardware. > It should also be easier to recover the context of the > previous process, in order to try and continue where things > left off. > With todays PC h/w in embedded systems, the reboot time is > a big problem. You want to reset only as a last resort. > This is also where "high availability" comes in-your system > is up a bigger % of time compared to a h/w watchdog because > of the faster restart time. This means that a process based > OS can be available more than a thread based one! > > >On top of that, the > > watchdog will cover you from kernel issues, the process model alone > will > > not. > > Agreed, the process model by itself is not good enough. > Use a s/w watchdog in conjunction with a hardware one, > for maximum protection and "higher availability". > IMO Using processes rather than threads give you an advantage. > > > 4) Any additional aplpication protection/features from the process > model > > comes at a performance and/or code overhead cost. > > Not necessarily. > > > In summary, there are a set of trade-offs at play here. I think that > for > > most embedded applications, less sophisticated, simpler > applications/kernels > > are easier to implement robustly and to verify. > > Always tradeoffs, decisions, decisions... :) > > An important factor consider in these decisions (imo) is > the architecture of the OS. > > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: TrueFFS - reference Date: Tue, 24 Aug 1999 19:38:43 +0200 From: "RonenO" Organization: Unspecified Organization Message-ID: <7puhub$104e$1@news.ibm.net.il> Hi, I am looking for someone who use TrueFFS for references. Please contact me if you use it, Thank you, Ronen. RonenO@tundo.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Read from files which are being appended Date: Mon, 23 Aug 1999 14:41:18 GMT From: "Ken Hayber" Message-ID: References: <199908201038.LAA21641@castor.epid.eurotherm.co.uk> You probably have to do a fflush() or possibly ioctl(..., FIOFLUSH, ...) first. I tried this a while ago and gave up on it...performance was way too low. I am thinking about putting some kind of intelligent buffering on top of the dosFs stuff just for the two tasks.... Tony Mayes wrote in message news:199908201038.LAA21641@castor.epid.eurotherm.co.uk... > Hi Vxworks users > > Can anyone tell me.. > > If I have a file open from 2 threads, one of which wants to read, one > wants to append... > > after an append will the first (reader) be able to access the > appended bytes or must I do a freopen() to get that data? > > Tony > > > > > > Tony Mayes > > (The opinions expressed within are my own) > > Email address tony.mayes@epid.eurotherm.co.uk > Phone (0044)(0)1903 205277 Fax (0044)(0)1903 524016 > > Eurotherm Process Instrumentation Div, > Southdownview Way, Worthing, Sussex, UK, BN14 8NN --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Flash : TrueFFS on RFA. Date: Fri, 27 Aug 1999 16:20:31 GMT From: Kremer@wxs.nl (Harro Kremer) Organization: Planet Internet Message-ID: <37c6baaa.338035167@news.wxs.nl> References: <7q4n4h$io5$1@nnrp1.deja.com> philippe.barthelemy@usa.net wrote: >Hi, > >I have looked in the doc, but I did not manage to find a definitve >answer on the following question : > >Can I use TrueFFS on a Resident Flash Array device ? >( "Resident Flash Array" as opposed to "Linear Flash Memory" ) The WRS supplied BSP for the MBX860 development system uses the flashes as RFA device. Harro Kremer --------------------------- Newsgroups: comp.os.vxworks Subject: are static variables global? Date: Mon, 30 Aug 1999 20:55:15 -0700 (PDT) From: anon789@webtv.net Organization: WebTV Subscriber Message-ID: <12969-37CB5223-123@newsd-223.iap.bryant.webtv.net> For example, int the following source file, all tasks spawned with f as the entry point share the same variable i /*file: a.c*/ int i; static int j; void f() {...} int main() { taskSpawn(...,f,...); taskSpawn(...,f,...); } Question: does the static keyword change that? j --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Design Constraints using VxWorks Subroutines Date: Tue, 31 Aug 1999 11:38:53 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37CBF70D.9BD06714@lucent.com> References: <7qgrhl$2jf1@overload.lbl.gov> Paul, how about declaring your data and subroutines as static to limit the scoping to that source file? (odd, but this is the second time this topic came up today...) Yup, what you have is really one task, vxWorks, controlling a bunch of threads inside a large block of memory. Not at all like UNIX where each task runs inside its own distinct space. So globals only exist once, not once per invocation of a task. Declaring data that is really only needed inside a small subset of functions as static helps to limit its scope. If you need run time data per invocation of a single thread, then you should consider allocating space at run-time and pointing to that data with a stack variable. That pointer will always be unique per invocation, and it will be taken care of by the normal task context switching. Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: are static variables global? Date: Tue, 31 Aug 1999 10:55:25 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37CBFAED.888E1F37@ciprico.com> References: <12969-37CB5223-123@newsd-223.iap.bryant.webtv.net> <37CBD8CC.78830A5D@lucent.com> Douglas Fraser wrote: > > [....] > If you had an 'int i' at the start of every file, the linker > will > complain of multiple declarations of variable i. (I hope.) > I suppose it could just silently collapse them all to one > instance, but that would be rude! The linkers I'm familiar with _will_ collapse them, and the older ones will do it silently. It used to be the expected behavior, I think because header files would omit the "extern." Anyway, to the original poster: if you want per-thread variables, you should read about taskVarLib. Mike --------------------------- Newsgroups: comp.os.vxworks Subject: How to handle removable SCSI units Date: Thu, 26 Aug 1999 14:32:44 -0400 From: Lloyd Davisson Organization: Lockheed Martin Federal Systems, Owego, NY Message-ID: <37C5884C.15BB7A0C@lmco.com> Our application is running VxWorks embedded on a single board computer. We communicate with the target via the shell (not Tornado). The system includes a SCSI device that holds mulitple PCMCIA cards. The sequence scsiPhysDevCreate, scsiBlkDevCreate, dosFsDevInit is used to setup each PCMCIA card at powerup. We would like some way to remove a card during operation and replace it with a different card. We thought we needed to use scsiPhysDevDelete prior to insertion of the new card followed by the above sequence once the new card is present. However, scsiPhysDevDelete returns an error - closer reading of the manual (novel concept) indicates the function returns an error if the input parameter, pScsiPhysDev, is NULL or if SCSI_BLK_DEVs have been created on the device. Assuming scsiBlkDevCreate above creates SCSI_BLK_DEVs on the device, we obviously shouldn't be using scsiPhysDevDelete to remove the card. Hopefully, someone here can point out where my (admittedly) limited knowledge is leading me astray. How should we be handling these removable devices? Any help would be greatly appreciated. Lloyd Davisson --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP [socket] performance issue in VxWorks ? Date: Thu, 26 Aug 1999 15:00:49 -0400 From: "Bob Ferrara" Message-ID: References: <37C22BFD.BA8873FD@gte.net> <37C490B8.3ABAFD6A@nowhere.com> <7q28oc$8ot@chronicle.concentric.net> Is this under Tornado-2 or -1? More specifically, what network stack are you using: 4.3 BSD, SENS, 4.4 BSD...? Interested, BobF Dinesh Nambisan wrote in message <7q28oc$8ot@chronicle.concentric.net>... >Thanks for your input, appreciate it... We're using a Motorola PPC i860 [@ >50 Mhz ]based platform here... I'm not sure about expected performance here; >and I might be comparing apples and oranges in this example that I mention, >but I noticed that a WinNT client @ 233 Mhz doing the same consumes >equivalent CPU....somehow I felt that there was definitely >something that could be tweaked to improve performance significantly... Are >the transmit and >receive buffer caching for Ethernet that you mention configurable parameters >or does it require a modification to the source ?? > >Thanks again, >DSN > >PS> I used a friend's e-mail id to post the problem earlier, lest u are >wondering ;-) > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Any network issues with going to Tornado-2 Date: Thu, 26 Aug 1999 15:12:36 -0400 From: "Bob Ferrara" Message-ID: I am interested in hearing anyone's experience migrating their project/product to Tornado-2, which now uses the 4.4 BSD TCP/IP network stack. I am specifically interested in network related issues, such as: - - porting device drivers to Tornado-2 - - network reliability (i.e. closed connections, dropped packets, etc) - - network performance Thanks, BobF --------------------------- Newsgroups: comp.os.vxworks Subject: Reprogramming ROM from outside the box. Date: Thu, 26 Aug 1999 13:29:06 -0500 From: "Matthew A. Townsend" Organization: Rockwell International Message-ID: <37C58772.143A7464@collins.rockwell.com> We are looking into using VxWorks and Tornado for the MPC8260. What we need to know is if you can reprogram your application code into ROM from outside the box. We can write code to do it, but it would be prefered if VxWorks or Tornado had this capability built in. Thanks in Advance, - -- - --------------------------------------------------------------------- Matthew A. Townsend Rockwell Collins, Inc. Voice: (319) 295-9152 Government Systems, SATCOM Facsimile:(319) 295-4317 M/S 107-140, 400 Collins Road NE matownse@collins.rockwell.com Cedar Rapids, Iowa 52498 - --------------------------------------------------------------------- All comments herein are my personal opinion and not that of any other organization that may be permitting my usenet access. - --------------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: How do I get the number of messages in a pipe Date: Mon, 30 Aug 1999 16:27:07 +0200 From: Rickard Rickardsson Organization: Enator PerCom AB Message-ID: <37CA94BB.9F0FEC9@extrees6.ericsson.se> This is a multi-part message in MIME format. - --------------BEFBA77EE943B02AE3B1C71C Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I'm trying to use ioctl(fd, FIONMSGS, &nofMsgs) (where nofMsgs is an int) to get the number of messages in a pipe but it doesn't work. I can't find enough information about this in the Programmer's guide. The pipe is created and opened like this: pipeDevCreate("/pipe/foo", 50, 1000); fd=open("/pipe/foo", O_RDWR, 0444); /Rickard - --------------BEFBA77EE943B02AE3B1C71C Content-Type: text/x-vcard; charset=iso-8859-1; name="qhsrdrn.vcf" Content-Transfer-Encoding: base64 Content-Description: Card for Rickard Rickardsson Content-Disposition: attachment; filename="qhsrdrn.vcf" YmVnaW46dmNhcmQgCm46Umlja2FyZHNzb247Umlja2FyZAp0ZWw7ZmF4Ois0NiAzMSA2NSA4 NCAzMwp0ZWw7aG9tZTorNDYgMzEgNzc1IDc2IDY4CnRlbDt3b3JrOis0NiAzMSA2NSA4NCA3 MAp4LW1vemlsbGEtaHRtbDpGQUxTRQpvcmc6RW5hdG9yIFBlckNvbSBBQgphZHI6OztLdW5n c2xhZHVn5XJkc2dhdGFuIDE2QSBidjtHb3RoZW5idXJnOzs0MTQgNjk7U3dlZGVuCnZlcnNp b246Mi4xCmVtYWlsO2ludGVybmV0OnJpY2thcmQucmlja2FyZHNzb25AZW5hdG9yLnNlCngt bW96aWxsYS1jcHQ6OzI1NDg4CmZuOlJpY2thcmQgUmlja2FyZHNzb24KZW5kOnZjYXJkCg== - --------------BEFBA77EE943B02AE3B1C71C-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IPPC from GreenSpring Date: Thu, 26 Aug 1999 14:33:55 -0700 From: Scott Kamara Organization: AP Labs Message-ID: <37C5B2C3.746AE4D3@pop.sttl.uswest.net> References: <37C50253.9E9A519C@lin.foa.se> Reply-To: spk@uswest.net > > Is possible to have the EXTENDE_VME defined ? If the BSP is like the T1.0.1 BSP set, then EXTENDED_VME is defined by default. > > if a chassie is configered with 2 mvme2700-cards > Has anyone peeked into CPU#1 from CPU#2 mem and > into CPU#2 from CPU#1 with cpu nummers in bootrom sat to (0 and 1) > > When I have proccess numbers to 0 and 1 I get the message > WTX Error 0x100ca (AGENT_MEM_ACCESS_ERROR) > when I try to peek into CPU#2 from CPU#1 but if I peek from CPU#2 into > CPU#1 it works. By default only cpu #0 (processor 0) has its local ram mapped to the VME bus in A32 space. Therefore, peeking from cpu #0 to cpu #1 would not work. You may change these defaults by strategically modifying the Universe VME chip support code. I would suggest tailoring sysUniverseInit2():universe.c. - ----------------------- Scott Kamara AP Data Systems Seattle Engr. Office (206) 525-8592 scott@aplabs.com --------------------------- Newsgroups: comp.os.vxworks Subject: END 802.5 SNAP to RFC 894 question Date: Tue, 31 Aug 1999 17:10:20 -0500 From: John Beauford Organization: Motorola Message-ID: <37CC52CC.91D423F6@email.mot.com> Reply-To: qjb005@email.mot.com Hello, I am endeavoring to write an END based driver for a Texas Instruments TMS 380 token ring chip for VXWorks. The packets that the chip receives and sends on the physical lan media will be in the 802.2 LLC/SNAP format(RFC 1042 format). 1. Does the default IP implementation require RFC 894 format? In other words does the muxReceive routine take this format and will the mux send the driver a packet in this format for sending? 2. If so does this little packet filter to change the packet from the LLC/SNAP to the 894 style go in the driver or somewhere else? 3. Is this documented anywhere in WRS docs that I may have overlooked? Thanks for any help. - -- John Beauford --------------------------- Newsgroups: comp.os.vxworks Subject: Scalability dilemma Date: Tue, 31 Aug 1999 23:56:12 GMT From: mkesti@gv.net Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qhq2n$g6q$1@nnrp1.deja.com> I am attempting to design a system that can download application programs via FTP, store them in flash, and execute them. The idea is to deploy a (hopefully) unchanging ROM based vxWorks image that includes an application that loads and executes the actual application that can be field upgraded. With the help of the DejaNews archive of the news.group's discussions of loadModuleAt() and symFindByName(), I have succeeded in doing just that. I now have a dilemma that stems from the scalable nature of vxWorks, however. I can see that future versions of the downloadable application programs may require OS components that were not included in the ROM's vxWorks image because I had not forseen that I would need them. I'm wondering how others are managing this issue? - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END 802.5 SNAP to RFC 894 question Date: Tue, 31 Aug 1999 21:43:57 -0400 From: "Jeff Szczepanski" Organization: Time Warner Road Runner - Rochester NY Message-ID: <3u%y3.4527$w6.123380@typhoon.nyroc.rr.com> References: <37CC52CC.91D423F6@email.mot.com> For END drivers, the default physical layer support is for standard Ethernet. You can have your own routines for forming the physical layer address/header by implementing the 'endAddressForm()' and 'endAddrGet()' entry points on the driver (instead of the default version provided for Ethernet). This is discussed in some detail in the Network Protocol Toolkit User's Guide available in the support area of the WRS website. Regards, Jeff John Beauford wrote in message news:37CC52CC.91D423F6@email.mot.com... > Hello, > > I am endeavoring to write an END based driver for a > Texas Instruments TMS 380 token ring chip for VXWorks. The > packets that the chip receives and sends on the physical > lan media will be in the 802.2 LLC/SNAP format(RFC 1042 format). > > 1. Does the default IP implementation require RFC 894 format? > In other words does the muxReceive routine take this format and > will the mux send the driver a packet in this format for sending? > > 2. If so does this little packet filter to change the packet > from the LLC/SNAP to the 894 style go in the driver or > somewhere else? > > 3. Is this documented anywhere in WRS docs that I may have > overlooked? > > Thanks for any help. > -- > John Beauford --------------------------- Newsgroups: comp.os.vxworks,comp.os.lynx,comp.os.chorus Subject: Re: vxWorks OK for High Availability systems ??? Date: Sun, 29 Aug 1999 02:03:25 GMT From: denesyk@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qa4d8$4uc$1@nnrp1.deja.com> References: <7q6aur$mr4$1@nnrp1.deja.com> <7q6oh8$g27$1@relay1.dsccc.com> Hi Jeff, Thanks for the reply. Are you saying you never saw tramplers in Lynx itself, or in your application SW. If you mean in your application, my hat is off to you. The cPCI mag article made the point that it is quite typical/normal/believeable to have between 1 and 5 SW defects per 1000 lines of application code that are not detected prior to shipping the product. Process-based RTOS are (apparently) able to give us clever app developers mechanisms to create systems that are capable of tolerating misbehaving applications without significant/sustained loss of system function. Regards, Tom Jeff.T.Williams@usa.alcatel.com (Jeff Williams) wrote: > No opinion on vxworks. Have been using threaded code on Lynxos in > a hig availability environment for about 2 years. Haven't seen the > slightest hint of a code trampler. Haven't seen any indication that > our threaded code is less reliable than the non-threaded code. YMMV. > > Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: How to wake up select Date: Sun, 29 Aug 1999 00:05:31 +0100 From: georges Message-ID: <37C86B3B.841CB817@redjupiter.demon.co.uk> Hi Guys, How to wake up select when it is pending on a dead fd (killed by another task).? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II project failed Date: Wed, 01 Sep 1999 06:30:48 GMT From: Kremer@wxs.nl (Harro Kremer) Organization: Planet Internet Message-ID: <37ccc74a.153452963@news.wxs.nl> References: <7qeam9$3ml4@overload.lbl.gov> Erin Swieringa wrote: I had a similar problem. TII reported "This project is of an older version'. It took me a while to find the cause. After checking in and getting it out of the config management tool (Visual Source (un)Safe) the file got corrupted. Real cause: VSS assumed,. based on the extension, that the file was a text file. Setting it to binary removed the corrupting. About the project itself: I recreated it because it was a small project. Harro >I'm trying to upgrade to Tornado II. I had a workspace set up with one >project in it. I had some custom builds, etc. - all was working fine. I >checked all project files into our configuration manager to capture the >currently working state of the project (that'll teach me). Later I tried >reopening my workspace and all the specifics of my project had disappeared - >no files, no build specifications. When I try to get the properties of >anything in the workspace I get a message saying "call to >"::ProjectPropertySheet::dataIn" failed: invalid command name "::name". I >tried to go back to an earlier checked in version of my project files and >that also did not work. > >Anybody else run into this and found a way to recover a project? > >- Erin >Electronic Theatre Controls --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A period in shell causes a reboot Date: Wed, 01 Sep 1999 18:10:15 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <37CD4FE7.F1FD2D58@bitband.com> References: <7qgrhp$2jg1@overload.lbl.gov> Reply-To: leonid@bitband.com This problem is solved - it happened that an MV162LC which uses 68LC040 CPU was used with a mv162 BSP configured for 68040 CPU. 68LC040 CPU lacks hardware floating point support, and thus the incorrectly handled FP exception which caused reboot.

Reconfiguring the CPU to use the 68LC040 libraries solved the problem.

This problem could happen to anyone using the 68LC040, upgrading from VxWorks 5.2 to Tornado, which is when a LC-specific library first appeared.

Leonid

David Gasul wrote:

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------_=_NextPart_001_01BEF3AE.44F66C80
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Michael,

Here's what I was able to find out:

1.      The problem occurs with the target shell
2.      The problem occurs with shell, telnet and rlogin
3.      The problem occurs ONLY with MVME162 board (it doesn't occur with
MVME2403 board)
4.      VxWorks version is 5.3.1
5.      Kernel version is 2.5

Any ideas?

David Gasul
Telegate Ltd.
7 Haplada St., 60218 Or-Yehuda, Israel
Voice: +972-3-5384634, Fax: +972-3-5335877
e-mail: davidg@telegate.co.il <mailto:davidg@telegate.co.il>=20
WWW: http://www.telegate.co.il

        -----Original Message-----
        From:   Michael Charny [SMTP:mcharny@wrs.com]
        Sent:   Tue 31 August 1999 11:20
        To:     David Gasul
        Cc:     'vxwexplo@lbl.gov'; 'support@wrs.com'; yoel@wrs.com
        Subject:        Re: A period in shell causes a reboot

        David,=20

        The only character sequence that can bring reboot is ^X.=20
        I was not able to reproduce your problem nor with Tornado 1 shell
nor with Tornado 2 one=20
        (using pc486 and mbx860 targets).=20

        Some clues:=20
        Check console versus rlogin/telnet.=20
        If problem with console only, check with another terminal emulator
(for example 'tip' on Solaris=20
        versus 'Hyper Terminal' on Windows).=20
        Check whether you use tyMonitorTrapSet( ) in your application.=20

        Regards,=20
        Michael Charny=20
        Wind River Systems, Israel.=20

        PS:=20
        The problem may be tied with floating point.=20
        Refer to the appendix corresponding to your CPU type=20
        in the "VxWorks Programmer's Guide". Notice the Exception Message
area=20
        in the VxWorks System Memory Layout. Use shell's 'd' command to
check=20
        this area for exception, that cause reboot.=20

        David Gasul wrote:=20

                =A0=20

                Hi,=20

                Has anyone encountered the following problem:=20

                If you input a period (the character '.') in the shell, the
system reboots? The period doesn't have to be the only character input. =
For
example the input 123.123 also causes a reboot. The only exception I =
know of
is the input=20

                ~. if you're in rlogin session.=20

                Is this a feature and if so - can it be turned off?=20

                If this is a bug - is there a workaround/fix?=20

                VxWorks Tornado=20

                David Gasul=20

                Telegate Ltd.=20

                7 Haplada St., 60218 Or-Yehuda, Israel=20

                Voice: +972-3-5384634, Fax: +972-3-5335877=20

                e-mail: davidg@telegate.co.il <mailto:davidg@telegate.co.il>

                WWW: http://www.telegate.co.il <http://www.telegate.co.il>=20
=09

------_=_NextPart_001_01BEF3AE.44F66C80
Content-Type: text/html;
        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Diso-8859-1">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2232.0">
<TITLE>RE: A period in shell causes a reboot</TITLE>
</HEAD>
<BODY>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">Michael,</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">Here</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">'</FONT><FONT SIZE=3D2 FACE=3D"Arial">s what I was able =
to find out:</FONT></P>

<OL TYPE=3D1><DIV ALIGN=3DLEFT><LI><FONT SIZE=3D2 FACE=3D"Arial">The =
problem occurs with the target shell</FONT></LI></DIV>
<DIV ALIGN=3DLEFT><LI><FONT SIZE=3D2 FACE=3D"Arial">T</FONT><FONT =
SIZE=3D2 FACE=3D"Arial">he problem occurs with shell,</FONT> <FONT =
SIZE=3D2 FACE=3D"Arial">telnet and rlogin</FONT></LI></DIV>
<DIV ALIGN=3DLEFT><LI><FONT SIZE=3D2 FACE=3D"Arial">The problem occurs =
ONLY with MVME162 board (it doesn</FONT><FONT SIZE=3D2 =
FACE=3D"Arial">'</FONT><FONT SIZE=3D2 FACE=3D"Arial">t occur with =
MVME2403 board)</FONT></LI></DIV>
<DIV ALIGN=3DLEFT><LI><FONT SIZE=3D2 FACE=3D"Arial">VxWorks version is =
5.3.1</FONT></LI></DIV>
<DIV ALIGN=3DLEFT><LI><FONT SIZE=3D2 FACE=3D"Arial">Kernel version is =
2.5</FONT></LI></DIV>
<BR>
</OL>
<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">Any ideas?</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">David Gasul</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">Telegate Ltd.</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">7 Haplada St., 60218 =
Or-Yehuda, Israel</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">Voice: +972-3-5384634, =
Fax: +972-3-5335877</FONT></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">e-mail:</FONT> <A =
HREF=3D"mailto:davidg@telegate.co.il"><U><FONT COLOR=3D"#0000FF" =
SIZE=3D2 FACE=3D"Arial">davidg@telegate.co.il</FONT></U></A></P>

<P ALIGN=3DLEFT><FONT SIZE=3D2 FACE=3D"Arial">WWW: http:</FONT><FONT =
SIZE=3D2 FACE=3D"Arial">//www.telegate.co.il</FONT></P>
<UL>
<P ALIGN=3DLEFT><A NAME=3D"_MailData"><FONT SIZE=3D2 =
FACE=3D"Arial">-----Original Message-----</FONT></A></P>

<P ALIGN=3DLEFT><B><FONT SIZE=3D2 =
FACE=3D"Arial">From:&nbsp;&nbsp;</FONT> <FONT SIZE=3D2 =
FACE=3D"Arial">Michael Charny [SMTP:mcharny@wrs.com]</FONT></B></P>

<P ALIGN=3DLEFT><B><FONT SIZE=3D2 =
FACE=3D"Arial">Sent:&nbsp;&nbsp;</FONT></B> <FONT SIZE=3D2 =
FACE=3D"Arial">Tue 31 August 1999 11:20</FONT></P>

<P ALIGN=3DLEFT><B><FONT SIZE=3D2 =
FACE=3D"Arial">To:&nbsp;&nbsp;&nbsp;&nbsp;</FONT></B> <FONT SIZE=3D2 =
FACE=3D"Arial">David Gasul</FONT></P>

<P ALIGN=3DLEFT><B><FONT SIZE=3D2 =
FACE=3D"Arial">Cc:&nbsp;&nbsp;&nbsp;&nbsp;</FONT></B> <FONT SIZE=3D2 =
FACE=3D"Arial">'vxwexplo@lbl.gov'; 'support@wrs.com'; =
yoel@wrs.com</FONT></P>

<P ALIGN=3DLEFT><B><FONT SIZE=3D2 =
FACE=3D"Arial">Subject:&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</FONT>=
</B> <FONT SIZE=3D2 FACE=3D"Arial">Re: A period in shell causes a =
reboot</FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" FACE=3D"Arial">David, =
</FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" FACE=3D"Arial">The only =
character sequence that can bring reboot is ^X.<BR>
I was not able to reproduce your problem nor with Tornado 1 shell nor =
with Tornado 2 one<BR>
(using pc486 and mbx860 targets). </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" FACE=3D"Arial">Some clues:<BR>
Check console versus rlogin/telnet.<BR>
If problem with consol</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial">e =
only, check with another terminal emulator (for example 'tip' on =
Solaris<BR>
versus 'Hyper Terminal' on Windows).<BR>
Check whether you use tyMonitorTrapSet( ) in your application. =
</FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" FACE=3D"Arial">Regards,<BR>
Michael Charny<BR>
Wind River Systems, Israel. </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" FACE=3D"Arial">PS:<BR>
</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial">The problem may be tied =
with floating point.<BR>
Refer to the appendix corresponding to your CPU type<BR>
in the &quot;VxWorks Programmer's Guide&quot;. Notice the Exception =
Message area<BR>
in the VxWorks System Memory Layout. Use shell's 'd' command to =
check<BR>
this area fo</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial">r</FONT><FONT =
COLOR=3D"#000000" FACE=3D"Arial"> exception, that cause reboot. =
</FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" FACE=3D"Arial">David Gasul =
wrote: </FONT></P>
<UL>
<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" FACE=3D"Arial">=A0</FONT><FONT =
COLOR=3D"#000000" FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">Hi,</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial"> =
</FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">Has =
anyone encountered the following problem:</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">If you =
input a period (the character</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"></FONT> <FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">'.') in the shell, the system reboots? The period =
doesn't have to be the only character input. For example</FONT> <FONT =
COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">the</FONT><FONT =
COLOR=3D"#000000" FACE=3D"Arial"></FONT> <FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">input 123.123 also causes a reboot. The only =
exception I know of is the input</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">~. if =
you're in rlogin session.</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial"> =
</FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">Is this =
a feature and if so</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"></FONT> <FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">- can</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"></FONT> <FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">it be turned off?</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">If this =
is a bug</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial"></FONT> <FONT =
COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">- is</FONT><FONT =
COLOR=3D"#000000" FACE=3D"Arial"></FONT> <FONT COLOR=3D"#000000" =
SIZE=3D2 FACE=3D"Arial">there a workaround/fix?</FONT><FONT =
COLOR=3D"#000000" FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">VxWorks =
Tornado</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">David =
Gasul</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">Tele</FONT><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">gate Ltd.</FONT><FONT COLOR=3D"#000000" FACE=3D"Arial"> =
</FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">7 =
Haplada St., 60218 Or-Yehuda, Israel</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 FACE=3D"Arial">Voice: =
+972-3-5384634, Fax: +972-3-5335877</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">e-mail:</FONT><FONT COLOR=3D"#000000" =
FACE=3D"Arial"></FONT> <A =
HREF=3D"mailto:davidg@telegate.co.il"><U><FONT COLOR=3D"#0000FF" =
SIZE=3D2 FACE=3D"Arial">davidg@telegate.co.il</FONT></U></A><FONT =
COLOR=3D"#000000" FACE=3D"Arial"> </FONT></P>

<P ALIGN=3DLEFT><FONT COLOR=3D"#000000" SIZE=3D2 =
FACE=3D"Arial">WWW:</FONT> <A =
HREF=3D"http://www.telegate.co.il"><U><FONT COLOR=3D"#0000FF" SIZE=3D2 =
FACE=3D"Arial">http://www.telegate.co.il</FONT></U></A></P>
</UL>
<P ALIGN=3DLEFT></P>
</UL>
</BODY>
</HTML>
------_=_NextPart_001_01BEF3AE.44F66C80--

--------------------------- Newsgroups: comp.os.vxworks Subject: Re: question! Date: Mon, 30 Aug 1999 16:40:23 -0500 From: Marty Schrader Organization: Fermilab Experimental Astrophysics Group Message-ID: <37CAFA47.64540811@fnal.gov> References: <37CAED64.AE2ACEB4@pegasus.cc.ucf.edu> You'll need to spend some more time going over the VxWorks Programmer's Guide and Reference Manual. I know this is very dry material and you may have some trouble with the grammer and such, but it is absolutely necessary. In the meantime, here is a function I use to delay operations by a specified number of milliseconds, close to the nearest system clock tick: void sleep(int Milliseconds) { int FiddleTicks; /* calculate ticks per millisecond; this might change on the fly, so the calculation must be done every time this thing is invoked */ FiddleTicks = ((sysClkRateGet() * Milliseconds) / 1000); FiddleTicks = (FiddleTicks ? FiddleTicks : 1); taskDelay(FiddleTicks); } Note how the value of FiddleTicks gets rounded up to 1 if the number of system ticks per second -- returned by sysClkRateGet() -- is too small a value. You'll have to experiment with this kind of thing yourself to see if these functions have the timing resolution you need. If not, you'll either have to crank up the number of system ticks per second via sysClkRateSet(), or go to a secondary (hardware based) timer. Look around this newsgroup for other discussions on this topic. - --- Marty Schrader x6645 Fermilab Experimental Astrophysics Group (SDSS) consult0@fnal.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re : upgrading GCC? Date: Fri, 27 Aug 1999 11:43:53 +0200 From: Jerome.SOUQUIERES@dematic-postal.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: > Has anyone any experience with upgrading the host tools > to later versions? I am thinking particularly of GCC. > I would really like to run our code tree through 2.9.5. Yes, we built EGCS 1.1.2 (the predecessor of GCC 2.95) [host x86-win32, target ppc] a few months ago. It was not that easy to build the cross compiler because of some strange stuff in WindRiver include files but we finally succeed. Everything works well except that we can't use C++ libraries built into VxWorks 5.3.1 (such as iostream) because the C++ name mangling has changed between GCC 2.7.2 and EGCS 1.1.2. But on the other hand, we benefit from all the interesting features of EGCS 1.1.2 : namespaces, enhanced template functionnality, MUCH improved optimisation... It's really worth. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Design Constraints using VxWorks Subroutines Date: Wed, 1 Sep 1999 08:03:14 -0700 (PDT) From: silvermang@webtv.net (Greg Silverman) Organization: WebTV Subscriber Message-ID: <19689-37CD4032-17@newsd-222.iap.bryant.webtv.net> References: <37CBF70D.9BD06714@lucent.com> How does VxWorks handle the request to load a function that is already loaded? Suppose A and B are two programs that have nothing in common other than calling function C from a library L, .e.g, both link in L. If A loaded and spawned and then the same is done for B, does the loader recognize that function C is already loaded and just ignore the new request to load it? More generally, does it make sense to not link in libraries when linking an executable, and, instead, individually load all of your libraries before loading any tasks? Greg Silverman TRW Sunnyvale, Ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: are static variables global? Date: Wed, 01 Sep 1999 11:44:32 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37CD57EF.341E7391@ciprico.com> References: <12969-37CB5223-123@newsd-223.iap.bryant.webtv.net> <37CBD8CC.78830A5D@lucent.com> <37CBFAED.888E1F37@ciprico.com> <37CD31F7.1C6F6F15@lucent.com> Douglas Fraser wrote: > > Mike Beede wrote: > > > The linkers I'm familiar with _will_ collapse them, and > > the older ones will do it silently. It used to be the > > expected behavior, I think because header files would > > omit the "extern." > > For Tornado, certainly, you are correct. It is still rude > behavior because you have the possibility of unintended > interaction caused by variable name conflicts at the source > file level. [....] Actually I was talking about old Unix environments. I agree with your observation--I was only making a pointless historical reference. > > Anyway, to the original poster: if you want per-thread > > variables, you should read about taskVarLib. > > You can also resort to allocating space for run time data > and referencing that through a stack variable. That context > will switch for you so that your referenced space is visible > to the running task. I feel that is a more natural > expression > of access than taskVars. [....] If you really want per-task globals, that means you (may) have to pass a gratuitous pointer around through various levels of functions that don't (appear to) need it. That can be ugly, IMHO. On the other hand, task variables are ugly to start with. I guess it comes down to a matter of taste/judgement in the individual case: whether you'd like to smear the ugliness about or concentrate it in a couple places. Both are valid techniques, anyway. Regards, Mike Beede - -- mbeede@ciprico.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: are static variables global? Date: Wed, 01 Sep 1999 13:09:14 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37CD5DBA.2AA05535@lucent.com> References: <12969-37CB5223-123@newsd-223.iap.bryant.webtv.net> <37CBD8CC.78830A5D@lucent.com> <37CBFAED.888E1F37@ciprico.com> <37CD31F7.1C6F6F15@lucent.com> <37CD57EF.341E7391@ciprico.com> Mike Beede wrote: > > Actually I was talking about old Unix environments. I agree > with your observation--I was only making a pointless historical > reference. But as you note, it is ugly! And it really isn't pointless, as history has a habit of coming back to bite us. > I guess it comes down to a matter of taste/judgement > in the individual case: whether you'd like to smear the ugliness > about or concentrate it in a couple places. Both are valid > techniques, anyway. I agree, and the choice should be based on your design. It depends on how flat or how deep your call chain is and whether you can isolate the knowledge of that data and pass it just to those tasks/functions that need it. That was the basis of my malloc/free example. There you don't want to pass that info around by hand because it is to widely distributed. So, you write the proceedures that need the actual information into a library that uses taskVars to keep the per invocation data. Cheers. Doug --------------------------- Newsgroups: comp.os.vxworks Subject: sysAuxClkConnect Date: Tue, 31 Aug 1999 10:30:11 -0500 From: "JSC" Organization: NASA Message-ID: <7qgsaa$7qu@niven.ksc.nasa.gov> I have a PPC on a PCI chassis that seems to have some trouble cleaning up after itself. I connect a routine to the Aux Clock and run. Then after I kill all my tasks I do a sysAuxClkConnect(0,0) thinking that this will clear it from memory. Then when I restart my process it calls the old routine and the new routine when the clock is enabled. What do I do? --------------------------- Newsgroups: comp.os.vxworks Subject: Task stack consumption Date: Wed, 01 Sep 1999 09:28:05 +0200 From: Ildefonso Junquero Organization: SAINSEL Sistemas Navales S.A. Message-ID: <37CCD585.E14F8187@sainsel.es> Hi! I'm in a situation which i can't explain myself. In a usual multi-task application, there is a main task that spawns others, each one has its own stack, and the size of the stack if not supplied is set to 20000 bytes as default. My problem is that, many many times but not always, when my application begin to execute, the task tShell causes an Access Violation. If I increase the stack size for every task that spawns the main task, the problem is corrected, so i can check which is the task that overheads its stack with the command checkStack, but none of the tasks reaches the original stack size!!!, so it means that original size must be sufficient for all the tasks. Anyone has an answer to this trascendental question? Thanks folks! Ilde. - -- ______________________________________________________________________ Ildefonso Junquero e@mail : junquero@sainsel.es ___ __o Software Engineer ____ _'\<,_ Sainsel Sistemas Navales S.A. ______ (*)/ (*) C/ Manuel Velasco Pando, N. 7 41007 Sevilla (Spain) Go faster! Phone : +34-95-493 64 65 Use the byke! Fax : +34-95-493 64 33 ______________________________________________________________________ --------------------------- Newsgroups: comp.os.vxworks,comp.os.lynx,comp.os.chorus Subject: Re: vxWorks OK for High Availability systems ??? Date: Wed, 1 Sep 1999 19:05:12 GMT From: Dave McCooey Organization: Lynx Real-Time Systems Message-ID: <37CD78E8.F9A41D31@lynx.com> References: <7q6aur$mr4$1@nnrp1.deja.com> Sender: news@spam.Lynx.COM (Root on Lynx Servers) Although the OS being asked about here is vxWorks, it occurred to me that there may be some misunderstanding with regard to LynxOS. LynxOS is NOT a thread-based RTOS (using the definition from the CompactPCI Systems article). It is a process-based RTOS that supports multi-threaded application programs (i.e. multiple threads within a process). denesyk@my-deja.com wrote: > July / Aug issue of CompactPCI Systems, page 44 suggests that thread- > based RTOS's are not appropriate for systems with a requirement for > five 9's availability (99.999 % available). > > The argument goes: > Because tasks in a thread-based RTOS share a common memory space, such > RTOS's will not offer good mechanisms to protect the target from memory > stompers. > > Has someone out there got an opinion about the choice of vxWorks as > RTOS for products with a heavy HA requirement? > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: PThreads available in VXWorks? Date: Wed, 01 Sep 1999 23:44:21 GMT From: wazdakka@twcny.rr.com (Dan Bostwick) Organization: Time Warner Road Runner - Syracuse NY Message-ID: <37cdb8e4.942823@news-server> Our team is beginning development on VXWorks 5.3.1, using Tornado 1.0.1 (possible T2, not sure yet - there is a driver problem), and coding in C++. We will be using substantial tasking and shared memory access throughout the system, and many of us would like to make use a of PThreads library. The Programmer's Ref. certainly talks about POSIX, but never actually confirms or denying native PThread support. Certainly there is no "pthread.h" laying around that I can find. Does anyone out there have any experiance with PThreads under this environment? What packages are available? I have located a library by a company called "dot4," but don't know anything about the quality of their product. Can any shead any light on any of this (becuase, boy is it dark in here!) ? - - Dan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to increase the Stack Size?? Date: Tue, 31 Aug 1999 03:42:48 GMT From: lingsiva@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qfivm$rct$1@nnrp1.deja.com> References: <7q2qge$8nk$1@nnrp1.deja.com> <935658000.275567@news.vbs.at> <7q4s49$m59$1@nnrp1.deja.com> <936018959.454531@news.vbs.at> > HTH > -- > Werner Schiendl > International Applications > -- Thanks al lot for the information. Regards, s.shiva Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to delay a task in micro seconds range Date: Tue, 31 Aug 1999 08:53:15 GMT From: whd@toadytwo.zetnet.zet (Ag@whd) Organization: White Horse Design Message-ID: <37d8968d.3113921@news.zetnet.co.uk> References: <7o6021$t1l$1@nnrp1.deja.com> <935053462.976046@internet> <37BC491A.97EA7ECB@broadcom.com> <37BD7B6D.8AC39778@nowhere.com> Reply-To: whd@toadytwo.zetnet.zet On Fri, 20 Aug 1999 15:59:41 GMT, Charlie Grames nntp@news.boeing.com (Boeing NNTP News Access) penned these words: >3) Create a non-blocking processor delay loop, e.g.: > for (i = 0; i < n; i++); > >Option 3 is the least desirable because of its nonportability. We chose >option 1 on the MVME2700 and increased the system clock rate to 1000 >Hz. Doing so allows you to use select() and nanosleep() with 1 ms >resolution. To make it more "portable" you could calibrate such a loop by counting how many iterations can be performed say in one second, however this won't help if you must have a specific and precise usec delay since interrupts can make it at least the usec you desire, but quite possibly a great deal more, unless you disable interrupts (both in the calibrating loop and the delay loop). You may like to look at your BSP specific "sysTimestamp()" collection of routines which I am trying to find myself in the BSP - I can't find the source code or description of their parameters except that the routines ("sysTimestampGetFreq()" and "sysTimestampGetPeriod()" or names similar to these) return large numbers suggesting high resolution. Adrian WWW WWW Adrian Gothard WWW ww WWW White Horse Design WWWWWWWWWW WWWW WWWW whd@zetnet.co.uk, http://www.users.zetnet.co.uk/whd - --- It's good to be me, yes, Heh heh heh, 'purrr', 'purrr'. --------------------------- Newsgroups: comp.os.vxworks Subject: END driver development - please help Date: Fri, 27 Aug 1999 14:37:29 +0100 From: "Rui Prior" Message-ID: <37c6940e@readers.ip.pt> In pages 19 and 20 of the Network Protocol Toolkit user's guide (5.4), it says endLoad() must "create a private pool of memory using netBufLib API". But on page 54 of the same manual, it says this pool is "used internally by the MUX interface". After all is it private to the END driver, or is it to be used by the MUX? And what is it used for? Rui Prior --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Telnet into vxworks target Date: Wed, 1 Sep 1999 19:59:20 -0700 From: "Steve Doiel" Message-ID: <37cde784.0@news.pacifier.com> References: <7qgs02$1t8@netnews.hinet.net> Reply-To: "Steve Doiel" I can't speak for the i960CA platform, but this same function does not work on the MVME162 platform either. I found that any diagnostic "printf" statements in the code go to standard output. The VxWorks guide indicates that you can redirect the output to the host shell. My experience has been that this works briefly but then the system gets confused and you lose your shell session. I haven't tried this with Tornado 2. SteveD wrote in message news:7qgs02$1t8@netnews.hinet.net... > > > My target is vxWorks 5.3.1 at i960CA platform. > > VxWorks Programmer's Guide describes that telnet daemon > > would spwan tasks to service telent client and rediret > > stdin and stdout from console to this telnet session. > > But I've tried several times, it seems not work! > > Can anyone tell me any idea to redirect system STDIN and > > STDOUT to telnet session? > > Thank you! > Lin Houcheng > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks and DHCP Date: Thu, 02 Sep 1999 04:00:26 GMT From: "Jerry Johnson" Message-ID: References: <199909011015.LAA07730@dsl-2.tadpole.co.uk> When I first tried using DHCPC on vxworks, I got similar addresses back when my dhcp server was not able to broadcast a response back. I needed to fix my linux based dhcp server. David Laight wrote in message <199909011015.LAA07730@dsl-2.tadpole.co.uk>... >All, > >I posed this about six months ago - and have finally got around to trying some >experiments... > >I have a system where the ethernet interface is an a PCMCIA card that will not >normally be present. Network connection will normally be by dial-out PPP. >However there are times when an ethernet card will be inserted, and I want to be >able to dynamically bring up the interface and use DHCP to obtain the IP address >(etc). > >As an experiment, I booted the system with the ethernet card present and a >static (non-DHCP) IP configuration, then ran the following from the target >shell: > dhcpLibInit( 67, 68, 4, 5, 3600, 30 ); > lease = dhcpInit( ifunit( "xircom0" ), 1 ); > sp (dhcpBind, lease, 1 ); > >(you will need to build a vxworks image containing the dhcp code - don't define >INCLUDE_DHCPC, just include a reference to the routines.) > >What I expect this to do is contact the DHCP server, obtain some parameters and >set the interface "xircom0" to use them. > >All I actually see is the local IP address changed to 10.0.x.x. I suspect this >is a 'random' address assigned by the DHCP code so it can send out some UDP >datagrams. > >The lease seems to match 'struct leaseData' in h/dhcp/dhcpcCommonLib.h and has >the following set: > > initFlag 1 > waitFlag 0 > autoConfig 1 > leaseSem > leaseGood 0 > leaseType 0 > prevState 12 MAX_STATES+1 ??? > currState 0 INIT > timer > initEpoch time(0) > timeout 20 > numRetry 3 > eventHookRtn 0 > cacheHookRtn 0 > leaseReqSpec.waitsecs 5 > leaseReqSpec.lease 3600 > >All the rest of the structure (about 60 words) is zero. > >Given some things that are not set up, I don't think the code has got very far >at all. > >One very worrying thing I've spotted is the definition of structure members (eg >if_info.name in dhcpc.h) as: > > char name[ sizeof "enxx" ]; > >This is horrid - how big is the array? >Certainly my "xircom" interface name won't fit! > >Does anyone know: >a) whether the test I'm doing is likely to suceed >b) why it is failing on my system. > > David > >---------------------------------------------------------------- >David Laight email: dsl@tadpole.co.uk >Tadpole Technology plc phone: +44 1223 278 256 >Cambridge, UK fax: +44 1223 278 201 > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Design Constraints using VxWorks Subroutines Date: Wed, 1 Sep 1999 22:13:52 -0700 (PDT) From: silvermang@webtv.net (Greg Silverman) Organization: WebTV Subscriber Message-ID: <17697-37CE0790-68@newsd-223.iap.bryant.webtv.net> References: <37CD5F37.AB1B8583@lucent.com> Douglas=A0Fraser wrote - ----- If A and B are each linked to L, then the call to C is fully resolved in each of A and B. When A runs, it will execute the image of C linked to it and when B runs, it will execute the image of C linked to it. Yes, this represents wasted space! C is loaded twice. If you have the system symbol table initialized, then the symbol table will contain a reference to C in A after A is loaded. It will contain a reference to C in B after B is loaded. You can experiment with this by running 'lkup "C"' after each ld command. - ----- Ok. Then, what if function C has a local static variable? As in, /*file C.c*/ static int i=3D0; void C() { static int j=3D0; if(j++%2=3D=3D0) { ++i; } if(i%2=3D=3D0) { //do something } } Does each loaded image of C have it's own copy of j? They all have to share i. Greg Silverman TRW Sunnyvale, Ca --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado II with Tornado I Date: Thu, 02 Sep 1999 07:28:39 GMT From: matthew.pardoe@widney-aish.co.uk (Matt pardoe) Message-ID: <37ce25b3.155893292@news.bournemouth-net.co.uk> we are currently planning a move from Tornado I to Tornado II. We are running on Windows NT. Currently we have one project in support that was developed using tornado I, and a new project that we will be developing using Tornado II. As well as user interface changes in Tornado II I believe there are also library and code changes. What we would like to to at least for a while is to have both Tornado I and Tornado II installed. Has anyone tried this or know if it is possible. My concern is that Tornado I seems to save a fair amount of data in the Registry, and this my cause conflicts with Tornado II. I would also be interested in hearing of any other problems people may have have moving from Tornado I to Tornado II, or in installing Tornado II. Thanks Matt Pardoe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem while downloading vxworks on the target m/c Date: Thu, 2 Sep 1999 16:54:50 +0800 From: "Zha Junhai" Organization: Alcatel Message-ID: <37ce3e18.0@dnews.sbell.com.cn> References: <7qgrhp$2jf2@overload.lbl.gov> Please check you ftp server configuration or if you have compiled vxworks for download. T C Omprakaash wrote in message news:7qgrhp$2jf2@overload.lbl.gov... > Hi, > After running the WFTPD server on the windows host, > on the target machine (PC486 with vxworks boot floppy),I tried to > download the vxworks image from the host. > The ene0 and lo interfaces come up successfully. after displaying the > message "Loading...", it returns with the following error message > > Error Downloading file : errno = d0003. > > Where is the mistake ? > Is there some problem with configuration ? > > If someone has faced this problem before, please help me with the > solution. > > Regards > Omprakaash > --------------------------- Newsgroups: comp.os.vxworks Subject: Telnet into vxworks target Date: 31 Aug 1999 15:22:42 GMT From: houcheng@ms17.hinet.net () Organization: DCI HiNet Message-ID: <7qgs02$1t8@netnews.hinet.net> My target is vxWorks 5.3.1 at i960CA platform. VxWorks Programmer's Guide describes that telnet daemon would spwan tasks to service telent client and rediret stdin and stdout from console to this telnet session. But I've tried several times, it seems not work! Can anyone tell me any idea to redirect system STDIN and STDOUT to telnet session? Thank you! Lin Houcheng --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Thu Sep 2 04:45:41 1999 From: Paul Curson Date: Thu Sep 2 04:45:45 PDT 1999 Subject: Re: Design Constraints using VxWorks Subroutines the vxWorks Users Group Exploder wrote: > Submitted-by owner-vxwexplo-process Wed Sep 1 05:57:09 1999 > Submitted-by: David Laight > > > Subject: Re: Design Constraints using VxWorks Subroutines > > Submitted-by: Paul Curson > > > [......] > > You can, of course, just use the task id (taskIdSelf()) to search a private list > of data areas for the one for the calling task... Interesting idea. Just to clarify my understanding: Such a private list would itself have to be static or global to be persistent. It would probably take on the form of a linked list of structures to allow for growth, etc. This list would also require (semaphore) protection against multiple accesses during modification of the list size. The list could either be at 'task level', in which case a task delete hook could remove entries from the list. If the list existed within a called subroutine, it would be nice to have as a static. It wouldn't be so eary to delete entries because the function wouldn't know when a given task finishes - unless a special call was made to the function from within the task delete hook. Hmm. This approach might have some potential - especially at the 'called subroutine' level where task variables don't appear to be so useful (as I understand them) in managing static data. Paul. From owner-vxwexplo-process Thu Sep 2 09:26:59 1999 From: Dave Olson Date: Thu Sep 2 09:27:03 PDT 1999 Subject: RE: symbols, searching in vxworks It is indeed possible to use the ld() variant implemented on the target shell either, as you suggest via rlogin, or in the form '@ld' from the host shell. (The '@' character causes those commands implemented on both the host and the target to be interpreted by the target. Other examples include 'ls', and 'pwd'.) Unlike the host version, the target shell version of ld() also accepts a parameter that chooses whether or not the ld() command will stop a script from executing if there are undefined symbols in the downloaded object. The disadvantages of this approach are: - you must be running the target shell - the download occurs in two steps. First the entire object is copied into the target shell's memory; then the linking and relocation occurs. You must have room in your target memory for the entire module. Certain module formats can be huge if the module is compiled for debugging. A module of mine compiled for the PowerPC, hence in ELF format, that represented 250K text+data was 10 Meg in size. After the download, of course, that memory is available for your programs use, except that you also have used some of your memory to hold all the symbols whether you need them all or not. - if you're also trying to use the host facilities such as the host shell or debugger, you also need to have 'symbol table synchronization' enabled in order for the host to find out about the symbols in the module you downloaded. This represents another task running on the target and consuming target resources. Prior to the introduction of Tornado, we had made extensive use of the naturally available system symbol table. I've come to the conclusion that Tornado has made use of this resource quite difficult and no longer consider it available for programmatic use. David H. Olson Staff Engineer Oresis Communications, Inc. 14670 NW Greenbrier Parkway Beaverton, OR 97006 503.533.0717 x246 Fax:503.533.8223 dolson@oresis.com > -----Original Message----- > From: Alexander_R_Povolotsky@res.raytheon.com > [mailto:Alexander_R_Povolotsky@res.raytheon.com] > Sent: Thursday, September 02, 1999 6:45 AM > To: dolson@oresis.com; rcw@DynRes.com > Subject: RE: symbols, searching in vxworks > > > Hello, > > When connecting to the target via rlogin using the "target > shell", another > version of "ld ( )" > is used, which, I think, updates directly the symbol table, > residing on the > target. > If such version of "ld ( )" is used to load the application and the > application's task(s) is spawned > from the target shell, then, I think, programmatically run > symFindByName() > will not rely on > the 'symbol table synchronization' and will be quite robust. > > Regards, > Alex Povolotsky > > ---------------------- Forwarded by Alexander R > Povolotsky/RES/Raytheon/US > on 09/02/99 09:32 AM --------------------------- > > > vxwexplo@lbl.gov (the vxWorks Users Group Exploder) on > 09/01/99 08:12:36 PM > > To: vxworks_users@csg.lbl.gov > cc: (bcc: Alexander R Povolotsky/RES/Raytheon/US) > Subject: RE: symbols, searching in vxworks > > > > > Submitted-by owner-vxwexplo-process Wed Sep 1 17:12:30 1999 > Submitted-by: Dave Olson > > You have a configuration in which > - vxWorks is built w/ or loads it's own symbols onto > the target > - you are using the host shell > - you have not enabled 'symbol table synchronization' > > symFindByName() runs on the target and can only find symbols > that are on the target. ld() runs on the host and only > downloads the code, not the symbols. lkup() runs on > the host, so has access to the symbols. > > You can enable 'symbol table synchronization' (see the > target server configuration dialog), however it runs as > a background task and is really designed for human speed > interaction. It is really not workable for programmatic > access to the symbols. > > Bottom line is that under Tornado 2, you really can't depend > on programmatic use of the VxWorks managed symbol table. I > tried doing so (at a former employer, so can't share the code) > and ended up chucking that approach and writing my own > independent facility that required routines to 'register' > their symbols of interest with a manager. Other routines that > wanted to know about those symbols could then obtain them from > that manager. > > > > -----Original Message----- > > Sent: Wednesday, September 01, 1999 3:09 PM > > To: vxworks_users@csg.lbl.gov > > Subject: symbols, searching in vxworks > > > > Submitted-by: rcw@DynRes.com > > > > Hello VxWorks 'rs, > > > > I am trying to use the function symFindByName to locate a > > program variable without success and am wondering where I'm > > going wrong. > > > > I can use sysFindByName to search the global vxworks symbol > > table for the _vxworks_ variables. As an example, the > > following works from a loaded program or from Tornado: > > > > symFindByName (sysSymTbl, "symFindByName", &address, &type) > > > > This returns the address of the vxworks function > > 'symFindByName' by searching through the global symbol table > > (sysSymTbl) > > > > If I load up my program ( ->ld > the address of variables there, the function returns an error. > > > > If I use the function lkup from the Tornado shell, I _can_ > > find my variables. > > > > Can anyone shed any light on this? My bottom line is that I > > want to provide a way for a program to "monitor" itself in an > > easily reconfigureable way. My thought is to create a text > > file that can be read > > by the program and the program can then locate variable > > locations by using the sysFindByName > > > David H. Olson > Staff Engineer > Oresis Communications, Inc. > 14670 NW Greenbrier Parkway > Beaverton, OR 97006 > 503.533.0717 x246 Fax:503.533.8223 > dolson@oresis.com From owner-vxwexplo-process Thu Sep 2 14:11:28 1999 From: rcw@DynRes.com Date: Thu Sep 2 14:11:31 PDT 1999 Subject: RE: symbols, searching in vxworks A couple people have suggested my problem might be with the 'sync' option on the target server. Which, in fact I had not done. But when I added this option to the target server, I could no longer run tgtsvr without errors. When I start up tgtsvr, I get the following error messages: Fatal WTX error (0x10136), synchronization stopped Fatal WTX error (0x1012f), synchronization stopped Others have suggested that this effort might be futile. It may in fact be, but having started down this track, I'd like to see it a little further along. So, does anyone have any hints on what this message might mean? My configuration of the OS includes: include_net_sym_tbl include_sym_tbl include_sym_tbl_sync > When you create your target server in Tornado make sure that you > press ALL symbols and Target/Host symbol sync buttons in the middle > of the create target server window. VxWorks From owner-vxwexplo-process Thu Sep 2 14:16:26 1999 From: rcw@DynRes.com Date: Thu Sep 2 14:16:30 PDT 1999 Subject: RE: symbols, searching in vxworks - oops, never mind Right after I hit send on this message, I found what I was looking for on the Wind River support web site. In case anyone else is interested, this indicates that the WIND_REGISTRY environment variable has not been properly set (TSR 44169) -----Original Message----- To: 'VxWorks Exploder' Subject: RE: symbols, searching in vxworks A couple people have suggested my problem might be with the 'sync' option on the target server. Which, in fact I had not done. But when I added this option to the target server, I could no longer run tgtsvr without errors. When I start up tgtsvr, I get the following error messages: Fatal WTX error (0x10136), synchronization stopped Fatal WTX error (0x1012f), synchronization stopped Others have suggested that this effort might be futile. It may in fact be, but having started down this track, I'd like to see it a little further along. So, does anyone have any hints on what this message might mean? My configuration of the OS includes: include_net_sym_tbl include_sym_tbl include_sym_tbl_sync > When you create your target server in Tornado make sure that you > press ALL symbols and Target/Host symbol sync buttons in the middle > of the create target server window. VxWorks From owner-vxwexplo-process Thu Sep 2 14:38:56 1999 From: kapil arora Date: Thu Sep 2 14:39:00 PDT 1999 Subject: Re : VxWorks timerLib Hello, I have an application that needs to use the following routines: timer_create(...); timer_connect(...); timer_settime(....); After building the project, I download the LAC.out and get the following error messages: Unresolved External Symbols Errors while downloading C:/LAC.out: _timer_create _timer_connect _timer_settime I included the timers.h which has the the following codes: extern int timer_create (...); extern int timer_connect (...); extern int timer_settime (....); The linker couldn't find the these functions. Do I need to download any module prior to LAC.out or do I need to include some header files? Can anyone please help? Thanks, Mike Hi Mike, I think the vxWorks image on which you are downloading your application lacks POSIX support, because these timer routines are POSIX timer routines. And as there is a load time linking in VxWorks so you won't get error during compilation for these symbols. Rebuild the vxWorks image with following in the config.h file and you should be going. #define INCLUDE_POSIX_ALL (or lesser functionality if size is also a problem). Hope it works. Bye, Kapil. From owner-vxwexplo-process Thu Sep 2 22:22:52 1999 From: Chris Hann Date: Thu Sep 2 22:22:55 PDT 1999 Subject: RE: symbols, searching in vxworks --=====================_132600529==_.ALT Content-Type: text/plain; charset="us-ascii"; format=flowed >The disadvantages of this approach are: > - you must be running the target shell > - the download occurs in two steps. First the entire > object is copied into the target shell's memory; then > the linking and relocation occurs. You must have > room in your target memory for the entire module. > Certain module formats can be huge if the module is > compiled for debugging. Sorry to intrude. This is only true if you are using a netdrv; if you use nfs or a local drive of some sort, which supports standard disk operations, then the loader will only extract those elements of the object file which it needs. netdrv was designed to overlay access methods which cannot support standard disk i/o operations such as seek, the file locally and then accessed almost as a RAM drive. Chris Hann Wind River Systems --=====================_132600529==_.ALT Content-Type: text/html; charset="us-ascii"
The disadvantages of this approach are:
  - you must be running the target shell
  - the download occurs in two steps.  First the entire
    object is copied into the target shell's memory; then
    the linking and relocation occurs.  You must have
    room in your target memory for the entire module.
    Certain module formats can be huge if the module is
    compiled for debugging. 

Sorry to intrude.

This is only true if you are using a netdrv; if you use nfs or a local drive of some sort, which supports standard disk operations, then the loader will only extract those elements of the object file which it needs. netdrv was designed to overlay access methods which cannot support standard disk i/o operations such as seek, the file locally and then accessed almost as a RAM drive.

Chris Hann
Wind River Systems
--=====================_132600529==_.ALT-- From owner-vxwexplo-process Fri Sep 3 01:59:05 1999 From: Piero Piero Pompei Date: Fri Sep 3 01:59:08 PDT 1999 Subject: Re: comp.os.vxworks newsdigest From owner-vxwexplo-process Fri Sep 3 03:02:46 1999 From: Rainer Stelzer Date: Fri Sep 3 03:02:50 PDT 1999 Subject: Pascal for VxWorks Hi vxworkers, is somebody out there who u works with PASCAL on vxworks target ? We want to integrate old PASCAL sourcecode into a new project. thanks in advance regards rainer ===================================================== Gebrüder Stoye GmbH department: traffic control systems name: Rainer Stelzer adr: Longericher Strasse 177 city: Cologne, Germany zip: 50739 phone: ++49/221/17954-51 mailto: r.stelzer@stoye.de www: www.stoye.de ===================================================== From owner-vxwexplo-process Fri Sep 3 04:00:15 1999 From: daemon@csg.lbl.gov Date: Fri Sep 3 04:00:20 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Sep 3 04:00:11 PDT 1999 Subject: Re: Is VxMP platform VxWorks BSP dependent ? Subject: Re: Help! Turning echo off for standard output? Subject: Re: UDP [socket] performance issue in VxWorks ? Subject: MPC8260 support Subject: Re: Not suited to High Availability??? Subject: Re: Design Constraints using VxWorks Subroutines Subject: Re: END driver development - please help Subject: Q: number of TCP socket Subject: Re: Not suited to High Availability??? Subject: Re: END driver development - please help Subject: Re: END driver development - please help Subject: Re: Tornado II with Tornado I Subject: Re: Raw ethernet packet hooks Subject: Re: Design Constraints using VxWorks Subroutines Subject: Re: Raw ethernet packet hooks Subject: Re: vxworks PCI driver as a loadable module Subject: Re: END driver development - please help Subject: windsh startup file? Subject: Re: windsh startup file? Subject: Re: _udivsi3 / _mulsi3 Subject: statics global Subject: Re: Design Constraints using VxWorks Subroutines Subject: Re: Tornado II with Tornado I Subject: universe2 vme interface chip with pio board problem Subject: Re: Design Constraints using VxWorks Subroutines Subject: Re: PThreads available in VXWorks? Subject: Re: Raw ethernet packet hooks Subject: Re: Reprogramming ROM from outside the box. Subject: sntp server Subject: Re: symbols, searching in vxworks Subject: How to increase the Stack Size?? Subject: Crosswind under emacs Subject: Crosswind and .inputrc Subject: VxWorks porting to strongARM Subject: Re: Tornado 1.0.1 vs Tornado 2 Subject: Re : Auxilliary Clock Subject: Re: Crosswind under emacs Subject: How can I boot vxworks via diskonchip2000? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is VxMP platform VxWorks BSP dependent ? Date: Thu, 02 Sep 1999 11:37:13 GMT From: dixon@crd.ge.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qlnh8$aec$1@nnrp1.deja.com> References: <7qk50l$1en3@overload.lbl.gov> In article <7qk50l$1en3@overload.lbl.gov>, Xinyang Qian wrote: > > Hi, > > I wonder to know, if we got the SMNet ready, can we just buy VxMP from > WRS without modifying anything and use it ? I heard from WRS field that > VxMP is still platform dependent, which means, still some BSP modification > is necessary, can any one provide me some info on it ?? > > Thanks, > -xinyang > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > Xinyang Qian E-Mail: xinyang@eng.sun.com > Sun Microsystems Phone: 408.774-8298 > 420 N Mary Ave, #04 Fax: 408.616-5810 > Sunnyvale, CA 94086 > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ I can't generalize, but I know how the x86 version of VxMP works and it is platform dependent. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help! Turning echo off for standard output? Date: Thu, 26 Aug 1999 01:04:46 GMT From: Charlie Grames Organization: Boeing Message-ID: <37C492AE.C91460EE@nowhere.com> References: Sender: nntp@news.boeing.com (Boeing NNTP News Access) Andrew, Is standard out going to the default location (e.g., a console port), or is it redirected to a virtual console? I know that with the virtual console in Tornado for Windows NT, your inputs are always echoed, regardless of the terminal definition. In fact, if OPT_ECHO is set, the characters are echoed twice! Knowing what target you are using would also be helpful. Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com Andrew Paryzek wrote: > > Trying to turn off echo for standard output using 'ioctl'. blah = ioctl(1, > FIOGETOPTIONS, 0) gives me blah=0 always. I can do a FIOSETOPTIONS [ blah2 > = ioctl (1, FIOSETOPTIONS, OPT_ECHO) ] all I want, but when I try and read > the options, they are all 0. Even more annoying, being 0 would imply that > the echo is turned off, but yet it is not (since it shows up on standard out > ...). Its something simple right? > > Please throw me a bone. > > Andrew Paryzek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP [socket] performance issue in VxWorks ? Date: Thu, 26 Aug 1999 00:56:24 GMT From: Charlie Grames Organization: Boeing Message-ID: <37C490B8.3ABAFD6A@nowhere.com> References: <37C22BFD.BA8873FD@gte.net> Sender: nntp@news.boeing.com (Boeing NNTP News Access) We have experimented with transmitting UDP packets of various sizes, using a ping-pong test program. As you might expect, as the size of the packets goes down, throughput goes down, and CPU utilization goes up. You didn't mention what platform you are using. Platform type has a large bearing on performance. So do factors such as cache usage. We are using a Motorola MVME2700 and discovered that we could improve performance by allowing the Ethernet chip's receive and transmit buffers be cached. It might help to know what your expectations for throughtput are, as well. Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com Ajay Malpani wrote: > > Has anyone faced performance issues when dealing with UDP sockets under > VxWorks ? > I'm writing a real-time-voice-traffic application; hence small packets > (approx 50 bytes) at high > frequency rates (say 20 ms) are to be despatched onto/received from the > network. I noticed > high CPU utilization; and on closer inspection and some test coding it > appeared as if the socket function > call itself ("sendto" in this case) is soaking up the majority of the > CPU time.... Also tried the zbuf (zero-copy buffer) socket library; but > as expected it didnt help much since it was clearly mentioned in manual > that > memcpy's are likely to be used (rather than avoided) when dealing with > small packets...the netTask does run up some CPU usage; but a huge > amount of the calling tasks' CPU consumption seems be done by the > socket function call (sendto) Also tried a VxWorks image with UDP > checksumming turned off, but that > did'nt help either > Anyone seen this before or have any clue as to what is going on ? > Any help would be greatly appreciated... > > Thanks. > > DSN --------------------------- Newsgroups: comp.os.vxworks Subject: MPC8260 support Date: Thu, 2 Sep 1999 14:46:49 +0200 From: "Michael Walton" Organization: An Internet Solution Customer Message-ID: <7qlrrj$ith$1@news.is.co.za> Hi I'm a latecomer, but can anyone give a quick answer to these two Qs? 1. Motorola 8260-ADS - is it the same as the 8260VADS? 2. Does VxWorks have a Tornado 1.0.1 BSP for 8260-ADS? Mike --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Not suited to High Availability??? Date: 27 Aug 1999 15:08:07 -0700 From: gillies@cs.ubc.ca (Donald Gillies) Organization: Computer Science, University of British Columbia, Canada Message-ID: <7q7287$kro$1@cascade.cs.ubc.ca> References: <7q6997$lh9$1@nnrp1.deja.com> denesyk@my-deja.com writes: >A recent article in the Compact PCI Systems >(July-August '99, p44) claims that process-based >RTOS's are superior to thread-based RTOS's for >High Availability (HA) systems. and that >thread-based RTOS's cannot provide any real degree >of fault tolerance to application SW bugs. >It said only process-based RTOS's have the >mechanisms for 'trapping, isolating, containing, >and cleaning up after' bugged tasks. >Can HA system developers credibly claim 99.999% >availability if their boxes are xWorks-based? >Comments??? This just demonstrates that there are a lot of narrow-minded authors in the world. For example, I can name two programming languages that make it almost unnecessary to have hardware protection in your system. The CLU language and the safe subset of Cedar language (from Xerox) are algorithmic languages that provide fully type-safe execution, without a terrible performance penalty (other than array checking and garbage collection). When you get done adding up the costs of health monitors and debugging pointer faults and the runtime costs of process switching in a heavy-duty operating system with inter-process firewalls, you may wish that you'd used a type-safe language rather than relying upon a process-safe operating system. The context switch time on a process-safe operating system can be so high as to make the OS unusable. Some of us oldsters remember that in the early 1980's, type-safe languages were popular because workstations had a low-performance CPU ( Mhz 68000, Xerox wildflower). It was cheaper in a software engineering sense get your fault and error checking at compile time than to get it from the high-overhead runtime system and in the OS... - - Don --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Design Constraints using VxWorks Subroutines Date: Thu, 02 Sep 1999 09:23:56 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37CE7A6C.2E21E724@lucent.com> References: <37CD5F37.AB1B8583@lucent.com> <17697-37CE0790-68@newsd-223.iap.bryant.webtv.net> Greg, I have to think about that for a bit, but my gut reaction is that they have separate instances of i, for this reason. When you load a module, it allocates space for .text, .data, and .bss, and then resolves the symbols into those allocated spaces. Since i is staticly linked to module C at compile time, two instances of C in memory as a result of separate loads will have separate instances of i. I may take a look at this if my current work falls into place this week, but you can try this on your own. here is how: Add a function to C.c to print the address of i. Add a function to A that calls this function in C. Add a function to B that calls this function in C. Link A to C and load it. Link B to C and load it. Call the address_of_i function from A then B and compare the adresses. If you do this, post the results back here to the group. You could also add a function to each of A and B to print the address of function C to show that it also exists in two distinct locations. Doug. PS: Dynamic loading and unloading of modules is a very powerful tool. Like any powerful tool, it needs to be used thoughtfully. Greg Silverman wrote: > > Ok. Then, what if function C has a local static variable? As in, > > /*file C.c*/ > > static int i=0; > > void C() > { > static int j=0; > if(j++%2==0) { > ++i; > } > if(i%2==0) { > //do something > } > } > > Does each loaded image of C have it's own copy of j? They all have to > share i. > > Greg Silverman > TRW > Sunnyvale, Ca > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END driver development - please help Date: Fri, 27 Aug 1999 18:55:33 -0400 From: "Jeff Szczepanski" Organization: Time Warner Road Runner - Rochester NY Message-ID: References: <37c6940e@readers.ip.pt> Typically the END drivers are implemented using mBuf sharing where the driver has a private pool of buffers for direct receive into your cluster DMA buffers. Without copying, the cluster can be assembled into the proper mBuf format and transferred to the protocol stack via the MUX. When the stack is done with it, it will free the mBuf and associated cluster to return it to your private pool. There is a decent example of all this in the ln7990 sample driver in the VxWorks tree...The path is something like ./usr/wind/target/src/drv/end/... (I think!?). Regards, Jeff =========================== Jeffrey R. Szczepanski InSciTek Microsystems, Inc. 635 CrossKeys Office Park Fairport, NY 14450 Rui Prior wrote in message news:37c6940e@readers.ip.pt... > In pages 19 and 20 of the Network Protocol Toolkit user's guide (5.4), it > says endLoad() must "create a private pool of memory using netBufLib API". > But on page 54 of the same manual, it says this pool is "used internally by > the MUX interface". After all is it private to the END driver, or is it to > be used by the MUX? And what is it used for? > > Rui Prior > > --------------------------- Newsgroups: comp.os.vxworks Subject: Q: number of TCP socket Date: 25 Aug 1999 03:03:53 GMT From: hiranabe@esm.co.jp (Kenji Hiranabe) Organization: Eiwa System Management, Inc. /Fukui,Japan Message-ID: Hi. I'm working on a TCP server program with vxworks5.0.3, pc486. You can telnet to this server and type "END" to quit connection. When i repeat the above operation several times, I finally cannot conncet to the server. (Telnet says Trying .....) Do you find any problem ? or is there a max number of sockets ? Thanks in advance. ===================== 8< Cut Here !! 8< -------------------- #include #include #include #include #include #include #define BUF_SZ 256 #define SEND_MSG "connected." int testserver(int port) { char send_buf[BUF_SZ], recv_buf[BUF_SZ]; int lsd, sd; int f = 1; int len; int ret = 0; struct sockaddr_in sa; memset(&sa, 0, sizeof(sa)); sa.sin_family = AF_INET; sa.sin_addr.s_addr = htonl(INADDR_ANY); sa.sin_port = htons(port); if ((lsd = socket(AF_INET, SOCK_STREAM, 0)) == ERROR) { fprintf(stderr, "socket() failed\n"); return -1; } if (setsockopt(lsd, SOL_SOCKET, SO_REUSEADDR, (char*)&f, sizeof(f)) == ERROR) { fprintf(stderr, "setsockopt() failed\n"); return -1; } if (bind(lsd, (struct sockaddr*)&sa, sizeof(sa)) == ERROR) { fprintf(stderr, "bind() failed\n"); return -1; } if (listen(lsd, 5) == ERROR) { fprintf(stderr, "listen() failed\n"); return -1; } printf("accept sd = %d\n", lsd); sprintf(send_buf, "%s\n", SEND_MSG); while (1) { if ((sd = accept(lsd, (struct sockaddr*)&sa, &len)) == ERROR) { fprintf(stderr, "accept() failed\n"); return -1; } printf("connect sd = %d\n", sd); len = send(sd, send_buf, strlen(send_buf), 0); if (len == 0) { fprintf(stdout, "connection closed\n"); close(sd); break; } if (len < 0) { fprintf(stderr, "send() failed\n"); ret = -1; close(sd); break; } while (1) { memset(recv_buf, 0, sizeof(recv_buf)); len = recv(sd, recv_buf, sizeof(recv_buf), 0); if (len == 0) { fprintf(stdout, "connection closed\n"); break; } if (len < 0) { fprintf(stderr, "recv() failed\n"); ret = -1; break; } if (! strncmp(recv_buf, "END", 3)) { fprintf(stdout, "END request received.\n"); break; } } if (shutdown(sd, 2) == ERROR) printf("hogheoghe\n"); close(sd); if (ret) break; } close(lsd); return ret; } - -- Eiwa System Management, Inc. http://www.esm.co.jp/ Kenji Hiranabe E-Mail: hiranabe@NOSPAMesm.co.jp --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Not suited to High Availability??? Date: Fri, 27 Aug 1999 19:11:02 -0400 From: "Jeff Szczepanski" Organization: Time Warner Road Runner - Rochester NY Message-ID: References: <7q6997$lh9$1@nnrp1.deja.com> I didn't read the article, but I would have to say that the author is working from some bad assumptions here. Was this written by someone who is associated with LynxOS? :-> 1) The first assumption from this claim is that the kernel has to be completely bug free for the application. If the kernel crashed in the process model case you are hosed anyway. Extreme example: Windows NT is fully process based & crashes fairly regularly! 2) Drivers tend to be inside the protection ring of the process model...How many real embedded systems do you know that don't have atleast one custom driver for some hardware interface/mechanism? ie: Atleast some of the functionality will be inside of ring of protection. 3) OK so process protection traps your application doing the real work, and safely terminates it things go bad...now what? Restart the application. First that's still not 'high availability' since it didn't prevent the failure and I can get the same functionality with a simple watchdog timer and an automatic hardware reset when it expires. On top of that, the watchdog will cover you from kernel issues, the process model alone will not. 4) Any additional aplpication protection/features from the process model comes at a performance and/or code overhead cost. In summary, there are a set of trade-offs at play here. I think that for most embedded applications, less sophisticated, simpler applications/kernels are easier to implement robustly and to verify. - -Jeff wrote in message news:7q6997$lh9$1@nnrp1.deja.com... > A recent article in the Compact PCI Systems > (July-August '99, p44) claims that process-based > RTOS's are superior to thread-based RTOS's for > High Availability (HA) systems. and that > thread-based RTOS's cannot provide any real degree > of fault tolerance to application SW bugs. > > It said only process-based RTOS's have the > mechanisms for 'trapping, isolating, containing, > and cleaning up after' bugged tasks. > > Can HA system developers credibly claim 99.999% > availability if their boxes are xWorks-based? > > Comments??? > > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END driver development - please help Date: Fri, 27 Aug 1999 16:04:54 -0700 From: "News" Organization: EarthLink Network, Inc. Message-ID: <7q754h$e7e$1@birch.prod.itd.earthlink.net> References: <37c6940e@readers.ip.pt> It is true that in endLoad() you must create a private pool of memory using netBufLib API. When you pass a buffer up to the MUX, your driver calls muxReceive(), which, in turn, calls the stackRcvRtn() routine, then after the upper layers are done with the data, they will free the buffer back to the driver's memory pool. Refer to SENS for Tornado (page 200) for more info. Vu. Rui Prior wrote in message <37c6940e@readers.ip.pt>... >In pages 19 and 20 of the Network Protocol Toolkit user's guide (5.4), it >says endLoad() must "create a private pool of memory using netBufLib API". >But on page 54 of the same manual, it says this pool is "used internally by >the MUX interface". After all is it private to the END driver, or is it to >be used by the MUX? And what is it used for? > >Rui Prior > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END driver development - please help Date: Fri, 27 Aug 1999 16:10:53 -0700 From: "News" Organization: EarthLink Network, Inc. Message-ID: <7q75g0$epe$1@birch.prod.itd.earthlink.net> References: <37c6940e@readers.ip.pt> When you are switching to END drivers and MUX, there is no concept of mBuf anymore (this is BSD stuff). See SENS for Tornado for more details. Vu. Jeff Szczepanski wrote in message ... >Typically the END drivers are implemented using mBuf sharing where the >driver has a private pool of buffers for direct receive into your cluster >DMA buffers. Without copying, the cluster can be assembled into the proper >mBuf format and transferred to the protocol stack via the MUX. When the >stack is done with it, it will free the mBuf and associated cluster to >return it to your private pool. > >There is a decent example of all this in the ln7990 sample driver in the >VxWorks tree...The path is something like ./usr/wind/target/src/drv/end/... >(I think!?). > >Regards, >Jeff > >=========================== >Jeffrey R. Szczepanski >InSciTek Microsystems, Inc. >635 CrossKeys Office Park >Fairport, NY 14450 > > >Rui Prior wrote in message news:37c6940e@readers.ip.pt... >> In pages 19 and 20 of the Network Protocol Toolkit user's guide (5.4), it >> says endLoad() must "create a private pool of memory using netBufLib API". >> But on page 54 of the same manual, it says this pool is "used internally >by >> the MUX interface". After all is it private to the END driver, or is it to >> be used by the MUX? And what is it used for? >> >> Rui Prior >> >> > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II with Tornado I Date: Thu, 02 Sep 1999 18:12:09 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <37CEA1D9.5765E0EF@bitband.com> References: <37ce25b3.155893292@news.bournemouth-net.co.uk> Reply-To: leonid@bitband.com Unlike UNIX, Windows can not maintain multiple installations of Tornado on the same machine, so you will have to leave one or two dedicated NT machines with Tornado 1 on them, while installing Tornado 2 on other machines. WRS has this clearly documented on the README notes shipped with every CD! Assuming your company opted for PCs to save money, now its time to pay for that decision... An old proverb says that only rich people can afford buying cheap things. Leonid Matt pardoe wrote: > we are currently planning a move from Tornado I to Tornado II. We are > running on Windows NT. Currently we have one project in support that > was developed using tornado I, and a new project that we will be > developing using Tornado II. > > As well as user interface changes in Tornado II I believe there are > also library and code changes. > > What we would like to to at least for a while is to have both Tornado > I and Tornado II installed. Has anyone tried this or know if it is > possible. My concern is that Tornado I seems to save a fair amount > of data in the Registry, and this my cause conflicts with Tornado II. > > I would also be interested in hearing of any other problems people may > have have moving from Tornado I to Tornado II, or in installing > Tornado II. > > Thanks > > Matt Pardoe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Raw ethernet packet hooks Date: Thu, 02 Sep 1999 16:34:01 +0100 From: David Rigg Organization: GEC Marconi Avionics Ltd Message-ID: <37CE98E9.7F1D@gecm.com> References: <7qjmun$1eo4@overload.lbl.gov> <37ce3152.258865148@news.inter.net.il> Reply-To: david.rigg@gecm.com Ran Shalgi wrote: > Two options are available to you: > 1. If you wish to continue working with etherHooks, you can simply > switch back to old BSD 4.3 drivers. Play around with your kernel > configuration, remove references to END and add instead old sytle > drivers, and all will be well. > are you sure? we've been told by WRS that SENS is the only option in T2 and there's no going back to BSD :( - -- David Rigg \ _ / Mission Avionics Division, x__________\_(0)_/__________x GEC Marconi Avionics, Scotland @ @ (](_o_)[) @ @ Tel : (+44) (0)131 314 8223 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Design Constraints using VxWorks Subroutines Date: Thu, 2 Sep 1999 08:06:39 -0700 (PDT) From: silvermang@webtv.net (Greg Silverman) Organization: WebTV Subscriber Message-ID: <4519-37CE927F-23@newsd-221.iap.bryant.webtv.net> References: <37CE7A6C.2E21E724@lucent.com> Doug, Checkout the "are static variables global?" thread. Everyone there agrees that file scope variables are share by all tasks, e.,g., no multiple instaces. That thread did not addess te issue of local statics, though. The basic issue is, what happens when task is running and another task or image that contains the same file scope and local static variables? Do the values of those variables get reinitialized (this would corrupt the executing task) or do new instatnces get created? Greg Silverman --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Raw ethernet packet hooks Date: Thu, 02 Sep 1999 11:47:13 -0400 From: "Joseph G. Jaeger" Organization: Open Networks Engineering Message-ID: <37CE9C00.79D881C1@one.com> References: <7qjmun$1eo4@overload.lbl.gov> I've been able to to install an etherInputHook successfully, by undefining INCLUDE_END and providing two additional arguments to etherInputHookAdd(). The additional arguments are required because I'm still using the mux. The additional arguments are a char * to the name of the device and the unit number (see the etherLib references). However, the only a NULL char* seems to work as the device name and results in a successfully installed etherHook. I would assume valid device names would be something like "dc" or "cpm". Anybody have a better understanding of what's going on, here? Joe Erin Swieringa wrote: > It appears that as of Tornado II and the new END drivers we no longer have > any method to access raw ethernet packets!!!??? Please tell me there's an > alternative to etherInput/Output hooks for the END drivers. > etherInput/OutputHook routines are still available, and in fact have > additional parameters for MUX/END drivers, but it appears that the END > drivers do not utilize them - is this true? > > - Erin Swieringa > Electronic Theatre Controls - -- ====================================================================== OPEN NETWORKS ENGINEERING, INC. Joseph G. Jaeger 2725 South Industrial Highway, Suite 100 Voice: (734) 975-7328 Ann Arbor, MI 48104 FAX: (734) 975-6940 Web: http://www.one.com Email: mailto:jgj@one.com ====================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks PCI driver as a loadable module Date: Thu, 02 Sep 1999 10:57:04 -0500 From: Doug Bercot Organization: Raytheon Company Message-ID: <37CE9E50.24BDB38D@ftw.rsc.raytheon.com> References: <37cac92d@readers.ip.pt> For us, the rub was that you have to configure the MMU at boot time, so depending on how your PCI interface (BSP) works, you may not have access to any shared memory on the device unless you map it in sysPhysMemDesc at startup. Otherwise you can just allocate a hole in the memory map and move the PCI device into that hole after startup. You must also make sure that pciInt is installed for each PCI interrupt, and other than that (may be forgetting something...) you can re-configure the PCI device at any time. Rui Prior wrote: > From previous posts to this group, I understand it is necessary to do the > PCI initialization of a device in the sysLib.c file of the BSP. > I'd like, however, to make my driver a dynamically loadable module, so my > question is: isn't there really any way of delaying the PCI configuration of > the device until runtime? If there is, how is it done? > > Thanks in advance for the answers. > > Rui Prior --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END driver development - please help Date: Fri, 27 Aug 1999 20:14:50 -0400 From: "Jeff Szczepanski" Organization: Time Warner Road Runner - Rochester NY Message-ID: References: <37c6940e@readers.ip.pt> <7q75g0$epe$1@birch.prod.itd.earthlink.net> News wrote in message news:7q75g0$epe$1@birch.prod.itd.earthlink.net... > When you are switching to END drivers and MUX, there is no concept of mBuf > anymore (this is BSD stuff). See SENS for Tornado for more details. > > Vu. What? Of course it still is based on the BSD implementation. The SENS stuff is derived from a newer release of the BSD code that the original stack was. Wind River has made some enhancements. Relative to the buffers, the interface exposed to the driver is a little different. I guess they refer to those as "mBlks" now. In any case, its essentially the same idea with a slightly different implementation. BTW, the exact file to refer to is: /usr/wind/target/src/drv/end/ln7990End.c Regards, Jeff > > Jeff Szczepanski wrote in message ... > >Typically the END drivers are implemented using mBuf sharing where the > >driver has a private pool of buffers for direct receive into your cluster > >DMA buffers. Without copying, the cluster can be assembled into the proper > >mBuf format and transferred to the protocol stack via the MUX. When the > >stack is done with it, it will free the mBuf and associated cluster to > >return it to your private pool. > > > >There is a decent example of all this in the ln7990 sample driver in the > >VxWorks tree...The path is something like ./usr/wind/target/src/drv/end/... > >(I think!?). > > > >Regards, > >Jeff > > > >=========================== > >Jeffrey R. Szczepanski > >InSciTek Microsystems, Inc. > >635 CrossKeys Office Park > >Fairport, NY 14450 > > > > > >Rui Prior wrote in message > news:37c6940e@readers.ip.pt... > >> In pages 19 and 20 of the Network Protocol Toolkit user's guide (5.4), it > >> says endLoad() must "create a private pool of memory using netBufLib > API". > >> But on page 54 of the same manual, it says this pool is "used internally > >by > >> the MUX interface". After all is it private to the END driver, or is it > to > >> be used by the MUX? And what is it used for? > >> > >> Rui Prior > >> > >> > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: windsh startup file? Date: 2 Sep 1999 16:39:15 -0000 From: Kin Cho Organization: mail2news@nym.alias.net Message-ID: <19990902163915.470.qmail@nym.alias.net> Is there such a thing as a shell startup file in windsh? - -kin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: windsh startup file? Date: Thu, 02 Sep 1999 16:55:40 GMT From: HJohnson@ZoneAutomation.Com (Hugh Johnson) Organization: Zone Automation Message-ID: <5_xz3.380$6w4.16228@nuq-read.news.verio.net> References: <19990902163915.470.qmail@nym.alias.net> Yes. in your .wind directory (assuming you are Win9x/WinNT) you can add a file, named windsh.tcl. That should contain tcl script commands that you want your windsh to perform upon startup. In article <19990902163915.470.qmail@nym.alias.net>, Kin Cho wrote: >Is there such a thing as a shell startup file in windsh? > >-kin > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: _udivsi3 / _mulsi3 Date: Thu, 02 Sep 1999 13:15:05 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37CEB099.22E9FF83@lucent.com> References: <7ql27i$29t$1@news2.kornet.net> <37ce9468.284231503@news.inter.net.il> Another way to solve this is to include a couple of 'long long' variables in your vxWorks code (e.g. usrAppl or sysLib) and define a function the does some basic arithmetic on those variables. You don't ever need to cal the function, just define it, and that will force the vxWorks kernel to load those libraries. Doug Ran Shalgi wrote: > = > On Thu, 2 Sep 1999 14:43:25 +0900, "=C1=A4=C1=F8=B1=BA" wrote: > = > >Undefined symbols: > >___udivsi3 > >___mulsi3 > >Warning: object module may not be usable because of undefined symbols.= > > > >My Division and Multiplication pgm had some problem. The above messge = told > >me "Something wrong " > > > >Could you help me??? > > > = > You may have tried to use multiplication or division on a 64-bit > integer type, or something similar which is not covered in the > standard VxWorks kernel build. > = > You'll need to extact the missing .o files from your archives and load > them into your target. See http://www.deja.com/getdoc.xp?AN=3D11892944= 6 > for an example. In short, search for a file named libgcc.a on your > host machine and run "ar386 -tv libgcc.a" (replace ar386 with > arYOUR_ARCH). > = > Ran Shalgi > Expand Networks, Inc > http://www.expand.com > = > --- > Replace 127.0.0.1 with expand.com when replying by e-mail --------------------------- Newsgroups: comp.os.vxworks Subject: statics global Date: Thu, 02 Sep 1999 13:28:26 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37CEB3BA.21F57E14@lucent.com> Greg Silverman wrote: > > Doug, > > Checkout the "are static variables global?" thread. Everyone there > agrees that file scope variables are share by all tasks, e.,g., no > multiple instaces. That thread did not addess te issue of local statics, > though. > > The basic issue is, what happens when task is running and another task > or image that contains the same file scope and local static variables? > Do the values of those variables get reinitialized (this would corrupt > the executing task) or do new instatnces get created? > The variables retain their current values, there are no new instances, nor is the current instance reset, that I know from my own experience. The values for all defined globals get set at vxWorks boot, or ld() by copying .data values from ROM (boot) or the act of locating .data into allocated memory (ld). Undefined globals will generally be set to zero, but that depends on the implementation of the BSP. The above holds true for normal globals, file scope statics and local statics. This assumes that you are NOT running VxVMI (or what ever it is called). That is, virtual := physical. I am not familiar with the extended virtual memory manager. The point I made before specifically addresses your earlier message, where module A is linked separately with module C, and module B is linked separately with module C, and then A is loaded, then B is loaded. In that case I believe you will get two instances of module C with decoupled static elements. Doug > Greg Silverman --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Design Constraints using VxWorks Subroutines Date: Thu, 02 Sep 1999 12:29:20 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37CEB3F0.593DCFA8@ciprico.com> References: <37CE7A6C.2E21E724@lucent.com> <4519-37CE927F-23@newsd-221.iap.bryant.webtv.net> Greg Silverman wrote: > > The basic issue is, what happens when task is running and another task > or image that contains the same file scope and local static variables? > Do the values of those variables get reinitialized (this would corrupt > the executing task) or do new instatnces get created? The point about reinitialzation hosing things is a good one. Another good suggestion (sorry, I don't remember who brought it up) is that the text and data segments are not seperable by the loader--the linker might have mixed things up arbitrarily (though I doubt it does). Anyway, Greg is correct in his guess that you get multiple copies of file statics in this case. I wrote a quick test program to verify it. This is on Tornado 2, by the way. I suspect that this is why, in spite of documentation to the contrary, the dynamic loader doesn't alter existing links. (Perhaps the docs have been corrected for Tornado 2--I haven't looked). Since this is my week for pointless history, I'll mention that this wouldn't be a problem on a system with Real Dynamic Linking (tm) like Multics, since 99 44/100 of code was loaded without any pre-linking*. Reference to a symbol caused the linker to find the segment containing it, load it if necessary, and "snap" (we'd probably say "validate", "connect" or "make") that link. I hasten to point out that this is hardly a desirable scheme for an RTOS where you'd like to get a handle on timing constraints, but it is R33l K3wl. Mike * Okay, frequently-used code was processed by the "binder" or something like that to reduce the number of segments that needed to be loaded in each user's process and speed execution. I get the idea that you only used that when you were done with development and were going to install it as a public facility. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II with Tornado I Date: Thu, 02 Sep 1999 12:38:59 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37CEB633.3CAF290@ciprico.com> References: <37ce25b3.155893292@news.bournemouth-net.co.uk> <37CEA1D9.5765E0EF@bitband.com> Leonid Rosenboim wrote: > > Unlike UNIX, Windows can not maintain multiple installations of Tornado on > the same machine, so you will have to leave one or two dedicated NT > machines with Tornado 1 on them, while installing Tornado 2 on other > machines. WRS has this clearly documented on the README notes shipped with > every CD! > > Assuming your company opted for PCs to save money, now its time to pay for > that decision... A differing data point, though I'd like to give a ringing "amen, Brother" on the issue of NT vs. Unix and low cost vs. high productivity. We installed Blow II -- er . . . sorry . . . Tornado II on NT, and I was still able to run Tornado I. I gave it a different installation directory, and ended up with new entries in the start menu, etc. I ran each at least a couple of times, but didn't give Tornado I a severe workout before I deleted it (oddly enough, the WinBlowsDelete.exe program didn't remove the start menu items, though presumably I'd just get the Oscillating Bill's Tool animation if I selected them). I'd experiment briefly on an ablative PC, but it might work. On the other hand, perhaps I was actually running the Tornado II tools from the Tornado I GUI.... As for the WRS documentation, I've come to regard that as helpful hints, but far from authoritative. I'm not arguing with previous posters, just reporting my experience and opinion. Regards, Mike Beede - -- mbeede@ciprico.com --------------------------- Newsgroups: comp.os.vxworks Subject: universe2 vme interface chip with pio board problem Date: Thu, 02 Sep 1999 19:27:11 GMT From: ywu@imatron.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qmj1v$24$1@nnrp1.deja.com> Hi: I have a puzzle here. We have a MC68040 bus mater and a xycom pio board(MC68230) in the vme chassis. The pio board worked well with this bus master. Now we are switching to power pc 750 bus arbiter with universe2 chip as vme interface. The pio board is not functioning properly. Even the status register on the pio board doesn't give consistent value upon power pc read. The RTOS is vxWorks 5.3.1. Does anybody have similar problem beofre? Thanks----Yongjian Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Design Constraints using VxWorks Subroutines Date: Thu, 02 Sep 1999 16:38:09 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37CEE031.C4F10748@lucent.com> References: <37CE7A6C.2E21E724@lucent.com> <4519-37CE927F-23@newsd-221.iap.bryant.webtv.net> <37CEB3F0.593DCFA8@ciprico.com> Mike Beede wrote: > Anyway, Greg is correct in his guess that you get multiple copies > of file statics in this case. I wrote a quick test program to > verify it. This is on Tornado 2, by the way. I wrote some code to test it also. Yes, you do get multiple copies as I suspected. Also, starting a task does not reset the values of statics or globals, but loading the image that defines them does. (which make sense if you consider what actions are taking place.) > > I suspect that this is why, in spite of documentation to the contrary, > the dynamic loader doesn't alter existing links. (Perhaps the docs have > been corrected for Tornado 2--I haven't looked). The documentation for loadModule() and loadModuleAt() only claims to resolve _unresolved_ symbols according to the system symbol table. Resolved symbols are merely relocated according to the addresses allocated for the load. The target shell loader description is ambiguous, but I assume (perhaps wrongly ;-) that it uses the loadLib modules. Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PThreads available in VXWorks? Date: Thu, 02 Sep 1999 14:03:28 -0700 From: Eric Flynn Organization: Hewlett Packard Cupertino Site Message-ID: <37CEE620.37D7A1F8@wwg2.sid.hp.com> References: <37cdb8e4.942823@news-server> Reply-To: eflynn@wwg2.sid.hp.com Dan, I can answer this one, since I just dealt with this myself. There is no *native* pthread support in VxWorks. However, there is a company (mentioned on the WRS website) which provides third-party support for pthreads. They take your BSP and add pthreads to it for you. I was not allowed to give them our BSP, so I am limited to using VxWorks tasks. Good luck, Eric Dan Bostwick wrote: > Our team is beginning development on VXWorks 5.3.1, using Tornado > 1.0.1 (possible T2, not sure yet - there is a driver problem), and > coding in C++. We will be using substantial tasking and shared memory > access throughout the system, and many of us would like to make use a > of PThreads library. The Programmer's Ref. certainly talks about > POSIX, but never actually confirms or denying native PThread support. > Certainly there is no "pthread.h" laying around that I can find. > > Does anyone out there have any experiance with PThreads under this > environment? What packages are available? I have located a library > by a company called "dot4," but don't know anything about the quality > of their product. Can any shead any light on any of this (becuase, > boy is it dark in here!) ? > > - Dan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Raw ethernet packet hooks Date: Thu, 02 Sep 1999 14:01:22 -0700 From: Stephen Macmanus Organization: Wind River Systems Message-ID: <37CEE5A2.476@wrs.com> References: <7qjmun$1eo4@overload.lbl.gov> Erin Swieringa wrote: > > It appears that as of Tornado II and the new END drivers we no longer have > any method to access raw ethernet packets!!!??? Please tell me there's an > alternative to etherInput/Output hooks for the END drivers. > etherInput/OutputHook routines are still available, and in fact have > additional parameters for MUX/END drivers, but it appears that the END > drivers do not utilize them - is this true? > - Erin Swieringa > Electronic Theatre Controls This is not correct. The ether input hook mechanism works with either BSD or END drivers. The hook itself is implemented differently, so it is no longer called by individual END drivers, but will still work in that case. It is true that this hook mechanism is obsolete and will be removed in the future, but not until a later release provides a solid alternative, probably including support for multiple SNARF protocols, which is not currently possible (as mentioned in another reply). - -- - ------------------ Stephen Macmanus #include Software Engineer Wind River Systems --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Reprogramming ROM from outside the box. Date: Thu, 02 Sep 1999 15:30:37 -0600 From: Lee D Gibbons Organization: Lucent Technologies Message-ID: <37CEEC7D.B16DCFDC@lucent.com> References: <37C58772.143A7464@collins.rockwell.com> Matthew A. Townsend wrote: > > We are looking into using VxWorks and Tornado for the MPC8260. > > What we need to know is if you can reprogram your application code into > ROM from outside the box. We can write code to do it, but it would be > prefered if VxWorks or Tornado had this capability built in. > > I think the most you'll find is that Wind River does supply flash writing code in their various BSPs, including the one for Motorola's VADS 8260. There is nothing that's processor-specific in any of this code. If anything, it's target specific (eg, manipulating Vpp). When you're talking about firmware download on your particular target, you've probably got a good deal of customization to do. - -- Lee Douglas Gibbons L u c e n t T e c h n o l o g i e s ldgibbons@lucent.com Bell Labs Innovations --------------------------- Newsgroups: comp.os.vxworks Subject: sntp server Date: Tue, 31 Aug 1999 18:20:44 -0500 From: Jonathan Price Organization: UUNET Message-ID: <37CC634C.66BD1A5B@montereynets.com> Has anyone had any luck with sntpsClockSet()? The Reference Manual doesn't offer too much guidance. I can't figure out the exact functionality the hook routine should fulfill. There are three different request types(SNTPS_ID, SNTPS_RESOLUTION, and SNTPS_TIME), and the hook should handle requests of each type. I assume the first is to get the clockid of the reference clock(should this be a clockid_t struct?), the second is to retrieve the resolution of the reference clock, and the last is to retrieve time from the reference clock. Is this right? If so, can anyone offer some suggestions on how to approach this. If not, could someone point me in a better direction? Any code examples would be much appreciated. Thanks, Jonathan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: symbols, searching in vxworks Date: Fri, 3 Sep 1999 00:41:51 GMT From: Charlie Grames Organization: Boeing Message-ID: <37CF194F.14ED58D5@nowhere.com> References: Sender: nntp@news.boeing.com (Boeing NNTP News Access) Ron, Don't be discouraged from following your current path. What you are attempting with symFindByName() not only works but works well for us. If you now have symbol synchronization working, everything else should fall into place. Let me set the record straight on a few points: 1) You do NOT need the target shell to load modules from the target or to run "@ld" from the host. We have the target shell disabled for security reasons and do loads from both the target and the host with no problem. 2) Loading from the target occurs in two steps ONLY if you are using the default netDrv driver to load your files from a remote host. netDrv uses FTP or rsh to load the entire file into target memory, after which the loader does its job with the copy in memory. ANY time you open a file with netDrv, it loads the entire file into memory; this is not peculiar to loading files from the target. If you want to avoid this problem, use nfsDrv (refer to the Reference Manual) instead. Of course, this assumes you are running NFS on your host. 3) The behavior of symbol tables in VxWorks from the perspective of the target has not changed at all since pre-Tornado days. What HAS changed is that along with the introduction of host tools came a separate host symbol table that, by default, has nothing in common with the target version other than the symbols defined by the VxWorks kernel itself. Symbol synchronization assures that both tables remain the same regardless of what you do from either the target or the host to change them. I will admit that there can be drawbacks to symFindByName(): You have to configure a target-based symbol table, which takes memory, and you have to configure symbol synchronization to use host tools, which creates an additional task (but only when the target server is active and symbol synchronization has been requested). If you are creating a small embedded system, availability of system resources may make this approach impractical. If, however, creating a dynamic system is more important, using symFindByName() (and associated bits) is invaluable. With our systems, after we have reached a certain point in development, we configure our targets to load applications automatically at startup and attach host tools only when necessary. This assures that our applications run as close as possible to the way they will when deployed while still allowing us access with the host tools. It is a system that has worked quite well for us. Good luck! Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com P.S. Check out the routine symSyncTimeoutSet() in the Reference Manual. You may find it useful if you are working with large programs. rcw@DynRes.com wrote: > Right after I hit send on this message, I found what I was looking for on the > Wind River support web site. > In case anyone else is interested, this indicates that the WIND_REGISTRY > environment variable has not been properly set (TSR 44169) --------------------------- Newsgroups: comp.os.vxworks Subject: How to increase the Stack Size?? Date: Thu, 26 Aug 1999 07:31:26 GMT From: lingsiva@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7q2qge$8nk$1@nnrp1.deja.com> HI, I would like to know how to increase the stack size of a task. The problem right now i am facing is as follows TaskStack() { DWORD dwArray[0x4000]; .... .. .. } While executing TaskStack from tornado shell, when i checked the tornado object browsers StackCheck utility, i noticed that the actuall stack allocated to this task is 19980 bytes. But, the actual requirement is around 65000 bytes. Thus resulting in a stack overflow. I do not know how to increase the size of the stack or where to set that option or whehter i should specify while compilation. Could anyone out there tell me how to do this. Thanks in advance regards, s.shiva Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Crosswind under emacs Date: 3 Sep 1999 03:33:09 -0000 From: Kin Cho Organization: mail2news@nym.alias.net Message-ID: <19990903033309.26701.qmail@nym.alias.net> Is anyone running Crosswind (gdb) under emacs? What's the command line option to launch gdbppc? I'm using Tornado 1.0.1 on NT 4.0. Thanks. - -kin --------------------------- Newsgroups: comp.os.vxworks Subject: Crosswind and .inputrc Date: 3 Sep 1999 03:45:22 -0000 From: Kin Cho Organization: mail2news@nym.alias.net Message-ID: <19990903034522.24731.qmail@nym.alias.net> Is anyone successful in using .inputrc under Crosswind? I have this single line in my %HOME%/.inputrc: set editing-mode vi but it seems to have no effect. - -kin --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks porting to strongARM Date: Fri, 3 Sep 1999 13:46:59 +0900 From: "À̼®Áø" Organization: Electronics and Telecommunications Research Institute (ETRI) Message-ID: <7qnjtc$h80$1@news.etri.re.kr> At Tonado directory, I can find ARM porting supporting file, but strongARM. Where can I find strongARM & Vxworks supporting. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 1.0.1 vs Tornado 2 Date: Sun, 29 Aug 1999 06:52:02 -0700 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <37C93B02.37FE@despam.pangea.ca> References: <7qao2i$r0q$1@spinner.corpeast.baynetworks.com> Ofer Goren wrote: > We are currently working with Tornado 1.0.1 for solaris, with clearcase > control version. We want to upgrade to Tornado 2, and we want to know should > we upgrade to NT, or stay with Solaris? > If anyone works with Tornado 2, AND clearcase version control, we want to > know what is the best step to do, so please contribute to the disscusion. > Hi Ofer, We are currently working on NT boxes with T1.0.1. FWIW, Our IT dept. went through an evaluation of ClearCase and found that there were just too many problems on NT... - -het - -- "you don't really appreciate how smart a moron is until you try to design a robot..." -a robotics designer Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re : Auxilliary Clock Date: Wed, 1 Sep 1999 15:53:59 +0200 From: Jerome.SOUQUIERES@dematic-postal.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: >My target is a PCI PPC 603. I am trying to schedule processes on the >auxilliary clock. I use the sysAuxClkConnect function to schedule a routine. >I try to kill all of my processes and delete the routine scheduled on the >sysAuxClkConnect by passing NULL as the first argument. Then when I restart, >calling sysAuxClkConnect again with the same routine it appears to be >executed twice for every clock interrupt. I noticed a similar behaviour with sysIntConnect(); - - on mv147 target (68k), calling sysIntConnect() on a specific interrupt vector replaces the previous ISR. - - on mvme603 and PowerCore750 targets (ppc), calling sysIntConnect() adds the new ISR to a chain, letting the previous ones in place, which is (IHMO) not a so good idea. But the BSPs are coded this way so I have to cope with it (i.e write some kludge code). However, I don't know if sysAuxClkConnect() also suffers from this problem. - -- ...Jess... Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Crosswind under emacs Date: Fri, 03 Sep 1999 07:54:02 GMT From: David Hamel Organization: @Home Network Message-ID: <37CF7E9A.47793DB1@ntanet.net> References: <19990903033309.26701.qmail@nym.alias.net> Kin Cho wrote: > Is anyone running Crosswind (gdb) under emacs? > > What's the command line option to launch gdbppc? > How about... gdbppc ? If that doesn't work right away... C:\Tornado\host\x86-win32\bin to your PATH from the control panel. Once gdbppc is running, type "target wtx targetservername" from the gdb prompt where targetservername is the name of the targetserver that you had started from Tornado. --------------------------- Newsgroups: comp.os.vxworks Subject: How can I boot vxworks via diskonchip2000? Date: Fri, 3 Sep 1999 19:12:29 +0900 From: "moony" Organization: System Engineering Research Institute (SERI) Message-ID: <7qo72f$3ht$1@green.kreonet.re.kr> Hi.. I am doing booting vxwoks via diskonchip2000 now. I downloaded device driver from m-system web site and followed the procedure as manual in files. But I can't boot it. just continue to reboot. It is normal state to appear the screen, "OS load ........ ". but after that, computer reboots.. How can I do? Hmm.. I would like to know the procedure in detail.. what file should I modify and where should I add in files? Please help me.. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Fri Sep 3 07:30:17 1999 From: Anurag Srivastava Date: Fri Sep 3 07:30:20 PDT 1999 Subject: Long File Name support TFFS > Hi everybody, > > I need long file name support on TrueFFS. I am making DOS file system > with long file name support and it is working . Does this support is > provided in tffs for Tornado. If yes , then please send me the command > to do s. > > Thanks, > > Anurag. From owner-vxwexplo-process Fri Sep 3 08:57:34 1999 From: "Thomas Keith Buchanan" Date: Fri Sep 3 08:57:38 PDT 1999 Subject: Re: Not suited to High Availability??? > >A recent article in the Compact PCI Systems > >(July-August '99, p44) claims that process-based > >RTOS's are superior to thread-based RTOS's for > >High Availability (HA) systems. and that > >thread-based RTOS's cannot provide any real degree > >of fault tolerance to application SW bugs. > > >It said only process-based RTOS's have the > >mechanisms for 'trapping, isolating, containing, > >and cleaning up after' bugged tasks. > > >Can HA system developers credibly claim 99.999% > >availability if their boxes are xWorks-based? > > >Comments??? This is a very narrow view of the issue. For systems that require high-availability (very common among applications running over VxWorks), software QA (every bit as rigorous as hrdware QA) and formal development processes (or semi-formal) are the key ingredients for success. Only certain types of bugs are caught by process-based memory protection. Most are not and can be just as insidious. In most OSes, threads in the same process context can alter memory areas of other threads. Software can cause algorithmic errors or introduce schedule anomalies and on and on and on... Ideally, software should be perfect ;o). Certainly no embedded system should be delivered with the types of problems that can alter memory outside of its own data seg and heap. Quality requirements in the real-time embedded world (usually automaticall an HA environment) are very different from desktop/enterprise systems. After putting so much effort into quality (required anyway), why should I sacrifice performance for task-based memory protection? I rarely need to inherit an environment or have parent/child relationships. In an embedded system, I find the simple light-weight task model of VxWorks adequate. 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 Sep 3 22:22:34 1999 From: "Ganesh S" Date: Fri Sep 3 22:22:37 PDT 1999 Subject: Boot Problem with non-END driver This is a multi-part message in MIME format. ------=_NextPart_000_001A_01BEF6C4.62AA3F60 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hello, I have observed the following problem with a vxWorks target. I have configured the target to use the ene network driver to boot. This is not = an END driver. The boot image is created with the INCLUDE_END definition = because, I have an END driver in the end table which is under test.=20 After the target boots the image, it tries to attach again to the driver = configured to boot with. ( see usrNetwork.c ). If iam not missing anything, then i = find a problem here. See the code from the usrNetwork.c given below. The problem is that the boot device is non-END ( params.bootDev =3D = "ene") but by having defined INCLUDE_END because of an END device in = the end table, this code checks for the presence of the "ene" in the end = table which it cannot find and will fail. It returns an error which i think is not correct. I = think if it cannot find the boot device in the end table, it should not fail but try to = check its presence in the ifnet structure list as it normally would do for any other = ethernet interface and attach to it. i.e the #else part of the code = below should be executed. >>> code fragment from usrNetwork.c=20 /* booting via ethernet */ #ifdef INCLUDE_END /* Find the END_OBJ associated with it. */ pEnd =3D endFindByName(params.bootDev, params.unitNum); if (pEnd =3D=3D NULL) return (ERROR); if (muxIoctl(pEnd, EIOCGMIB2, (caddr_t) &endM2Tbl) =3D=3D ERROR) return (ERROR); if (ipAttach(params.unitNum, params.bootDev) !=3D OK) { logMsg("Failed to attach to device %s", (int)params.bootDev, 2, 3, 4, 5, 6); return (ERROR); } if (usrNetIfConfig (params.bootDev, params.unitNum, = params.ead, params.targetName, netmask) !=3DOK) return (ERROR); printf("Attached TCP/IP interface to %s unit %d\n" , params.other, params.unitNum); #else if (usrNetIfAttach (params.other, params.unitNum, params.ead) !=3D OK) return (ERROR); if (usrNetIfConfig (params.other, params.unitNum, params.ead, params.targetName, netmask) !=3D OK) return (ERROR); #endif /* INCLUDE_END */ Please let me know if i have been correct in my interpretation or am i = missing anything.=20 VxWorks Tornado Thanks, Ganesh ------=_NextPart_000_001A_01BEF6C4.62AA3F60 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
 
Hello,
 
I have observed the following problem with a vxWorks = target. I=20 have
configured the target to use the ene network driver = to boot.=20 This is not an END driver. The = boot image is=20 created with the INCLUDE_END definition because, I have an  END = driver in=20 the end table which is under test.
 
After the target boots the image, it = tries to=20 attach again to the driver configured
to boot with. ( see usrNetwork.c ). = If iam not=20 missing anything, then i find a
problem here. = See the code=20 from the usrNetwork.c given below.
 
The problem is that the boot device = is non-END (=20 params.bootDev =3D "ene")  but by having defined = INCLUDE_END=20 because of an END device in the end table, this code checks=20 for the presence of the "ene" in the end table which it=20 cannot
find and will fail. It returns an error which i = think is not=20 correct. I think if it cannot
find the boot device in the end table, it should not = fail but=20 try to check its presence
in the ifnet structure list as it normally would do = for any=20 other ethernet interface and attach to it. i.e=20 the #else part of the code below should be executed.
 
 
>>> code fragment from = usrNetwork.c=20
 
        /*=20 booting via ethernet */
#ifdef=20 INCLUDE_END
         &nbs= p; =20 /* Find the END_OBJ associated with it.=20 */
            = pEnd =3D=20 endFindByName(params.bootDev,=20 params.unitNum);
         = ;  =20 if (pEnd =3D=3D=20 NULL)
          &nbs= p;    =20 return=20 (ERROR);
          &= nbsp;=20 if (muxIoctl(pEnd, EIOCGMIB2, (caddr_t)=20 &endM2Tbl)
         &= nbsp;     =20 =3D=3D=20 ERROR)
          &nb= sp;    =20 return=20 (ERROR);
          &= nbsp;=20 if (ipAttach(params.unitNum, params.bootDev) !=3D=20 OK)
           =     =20 {
           &n= bsp;   =20 logMsg("Failed to attach to device=20 %s",
          =              = (int)params.bootDev, 2, 3, 4, 5,=20 6);
           =     =20 return=20 (ERROR);
          &= nbsp;    =20 }
            = if=20 (usrNetIfConfig (params.bootDev, params.unitNum,=20 params.ead,
         &nbs= p;            = ;         =20 params.targetName,
        &nb= sp;           &nbs= p;          =20 netmask)=20 !=3DOK)
          &n= bsp;    =20 return = (ERROR);
        printf("Attached TCP/IP interface to %s unit=20 %d\n"
          = ;    =20 , params.other,=20 params.unitNum);
#else
        = if=20 (usrNetIfAttach (params.other, params.unitNum,=20 params.ead)
         &nbs= p;         =20 !=3D = OK)
           = =20 return (ERROR);
 
        if = (usrNetIfConfig=20 (params.other, params.unitNum,=20 params.ead,
         &nbs= p;            = ;            =  =20 params.targetName, netmask) !=3D=20 OK)
           = return=20 (ERROR);
#endif /* INCLUDE_END */
 
Please let me know if i have been correct in my = interpretation=20 or am i missing anything.
 
VxWorks  Tornado
 
 
Thanks,
 
Ganesh
 
 
 
 
------=_NextPart_000_001A_01BEF6C4.62AA3F60-- From owner-vxwexplo-process Sat Sep 4 00:30:54 1999 From: attila.pogany@essnet.se Date: Sat Sep 4 00:30:58 PDT 1999 Subject: Not read: comp.os.vxworks newsdigest Your message was not read by:Pogany Attila @ BUDAPEST/ESSNET GROUP at:Sat, 04 Sep 1999 07:22:57 UT From owner-vxwexplo-process Sat Sep 4 04:00:18 1999 From: daemon@csg.lbl.gov Date: Sat Sep 4 04:00:22 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Sep 4 04:00:14 PDT 1999 Subject: Re: UDP [socket] performance issue in VxWorks ? Subject: Rational Rose RealTime Subject: Re: VxWorks porting to strongARM Subject: Ethernet Problems Subject: workQPanic? Subject: Re: MPC8260 support Subject: On scsiRdSecs() routine Subject: Center an java.awt.frame in PersonalJWorks Subject: Re: Tornado II with Tornado I Subject: start target server outside of Tornado? Subject: Re: universe2 vme interface chip with pio board problem Subject: Not suited to High Availability??? Subject: Re: Why stay with VXWorks? Subject: any vme pio suggestions? Subject: Re: float in shell Subject: Re: vxWorks OK for High Availability systems ??? Subject: Re: Awesome Silicon Valley Start-up Subject: Re: Tornado II with Tornado I Subject: T2, mv2303 & 100Mbit lan => dc no carrier Subject: Re: start target server outside of Tornado? Subject: Re: VxWorks porting to strongARM Subject: Re: Tornado II with Tornado I Subject: mpc860 i2c problem Subject: Re: vxWorks OK for High Availability systems ??? Subject: Re: start target server outside of Tornado? Subject: Message-based library on top of VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP [socket] performance issue in VxWorks ? Date: 26 Aug 1999 14:00:59 PDT From: "Dinesh Nambisan" Organization: Concentric Internet Services Message-ID: <7q49ub$p99@chronicle.concentric.net> References: <37C22BFD.BA8873FD@gte.net> <37C490B8.3ABAFD6A@nowhere.com> <7q28oc$8ot@chronicle.concentric.net> Tornado 1, SENS stack... Bob Ferrara wrote in message ... >Is this under Tornado-2 or -1? More specifically, what network >stack are you using: 4.3 BSD, SENS, 4.4 BSD...? > >Interested, >BobF > >Dinesh Nambisan wrote in message <7q28oc$8ot@chronicle.concentric.net>... >>Thanks for your input, appreciate it... We're using a Motorola PPC i860 [@ >>50 Mhz ]based platform here... I'm not sure about expected performance >here; >>and I might be comparing apples and oranges in this example that I mention, >>but I noticed that a WinNT client @ 233 Mhz doing the same consumes >>equivalent CPU....somehow I felt that there was definitely >>something that could be tweaked to improve performance significantly... Are >>the transmit and >>receive buffer caching for Ethernet that you mention configurable >parameters >>or does it require a modification to the source ?? >> >>Thanks again, >>DSN >> >>PS> I used a friend's e-mail id to post the problem earlier, lest u are >>wondering ;-) >> >> >> >> > > --------------------------- Newsgroups: comp.os.vxworks Subject: Rational Rose RealTime Date: Fri, 3 Sep 1999 08:18:33 -0400 From: "Thomas Fuda" Organization: Pitney Bowes, Inc. Message-ID: <7qoe65$721@news.ct.pb.com> Hello Is anybody out there using the Rational Rose RealTime toolset with Tornado? My company is considering adopting this toolset for use on future software development efforts. I'd like to get some opinions from people who have actually used it. TIA. Tom Fuda Pitney Bowes, Inc. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks porting to strongARM Date: Fri, 03 Sep 1999 08:55:36 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37CFC548.FA5FBF4C@lucent.com> References: <7qnjtc$h80$1@news.etri.re.kr> There is a BSP for the SA110 that supports the ebsa-285 that is available from Intel. I believe there is also another board with an SA1100 (Brutus??) that has a Tornado BSP. Doug "=C0=CC=BC=AE=C1=F8" wrote: > = > At Tonado directory, I can find ARM porting supporting file, but strong= ARM. > = > Where can I find strongARM & Vxworks supporting. --------------------------- Newsgroups: comp.os.vxworks Subject: Ethernet Problems Date: Fri, 03 Sep 1999 13:44:45 GMT From: "Steeve Gagne" Organization: MetroNet Communications Group Inc. Message-ID: My mpc860 bsp boot from flash He download a load image from TFTP and it work fine When I boot the load image, the board not answer to ping and telnet from exterior. On the Net, when I throw a echo request paquet from the developpement board, I catch the reply on the net but the board didn't receive it. The Ip addr, the subnet mask, the Mac Address All It's ok I suspect an interrupt problem. Any Ideas ?? Steeve Gagne Firmware Designer steevegagne@sympatico.ca --------------------------- Newsgroups: comp.os.vxworks Subject: workQPanic? Date: 3 Sep 99 09:45:42 -0400 From: reichold@ohstpy.mps.ohio-state.edu (Dennis M. Reichhold) Organization: The Ohio State University, Department of Physics Message-ID: Hi One of my CPU's (a 162) was running just fine yesterday morning. Then, in the afternoon, it crashed (I assume). Now whenever it reboots, it goes through most of the login script but stops (always at the same spot, unless I've cycled the power) with the error message: workQPanic: Kernel work queue overflow. and then starts the reboot process. When I cycle the power, it reads in an auto_settings.sav file, and somewhere during the execution of iocInit, I get the same error message, and the processor reboots. The host computer is running SunOS 4.1.4. Anyone know what's going on? -Dennis --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MPC8260 support Date: Fri, 03 Sep 1999 10:01:20 -0600 From: Lee D Gibbons Organization: Lucent Technologies Message-ID: <37CFF0D0.43FF7859@lucent.com> References: <7qlrrj$ith$1@news.is.co.za> Michael Walton wrote: > > Hi > > I'm a latecomer, but can anyone give a quick answer to these two Qs? > > 1. Motorola 8260-ADS - is it the same as the 8260VADS? > 2. Does VxWorks have a Tornado 1.0.1 BSP for 8260-ADS? > > Mike The BSP I've received for the VADS8260 is for Tornado II. - -- Lee Douglas Gibbons L u c e n t T e c h n o l o g i e s ldgibbons@lucent.com Bell Labs Innovations --------------------------- Newsgroups: comp.os.vxworks Subject: On scsiRdSecs() routine Date: Sat, 04 Sep 1999 01:22:08 +0900 From: Young Jin Nam Organization: Postech News Server Message-ID: <37CFF5B0.D327FAB@postech.ac.kr> Hello, I found that "using scsiRdSecs() routine in parallel on the single SCSI channel" degraded lots of disk read performance. "In parallel" means that more than one tasks do call the scsiRdSecs() routine at the "almost" same time. (do I need to serialize the calling routine?) Any comments on this problem will be very appreciated. Thanks in advance. Young Jin --------------------------- Newsgroups: comp.os.vxworks Subject: Center an java.awt.frame in PersonalJWorks Date: Fri, 3 Sep 1999 18:56:53 +0200 From: Matthias Schumann Message-ID: Sender: Matthias Schumann Hi there, I am trying to center the singleton java.awt.frame component provided by PersonalJava on the screen, but the frame does not react in any way to the java.awt.frame.setLocation(int,int) method. The location is set, meaning that you can read out the new values set with getLocation, but the actual location on the screen is still (0,0). Does anybody know how to move the frame on the screen? +Matt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II with Tornado I Date: Fri, 03 Sep 1999 12:55:38 -0700 From: Latch Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <0c2b02cf.d00b482d@usw-ex0102-013.remarq.com> References: <37ce25b3.155893292@news.bournemouth-net.co.uk> <37CEA1D9.5765E0EF@bitband.com> <37CEB633.3CAF290@ciprico.com> 1. Change the Tornado Directory Name(ex. tornado to tornado1) 2. Install Tornado II Now you can work with both in just changing directory names. My 2 installations happily coexist. Issues: Some of the low level API functions are not available anymore in Tornado II. Some libraries too (rdbLib) Nevertheless if you don't digg into low level stuff (I had no choice) everything should work just fine. To make VxSIM work is another issue altogheter. Have fun. lstoyanov@truetime.com * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: start target server outside of Tornado? Date: 3 Sep 1999 20:25:14 -0000 From: Kin Cho Organization: mail2news@nym.alias.net Message-ID: <19990903202514.19003.qmail@nym.alias.net> Is it possible to start target server that I've configured outside of Tornado? That's, I'd like to start a target server as a Startup program under NT. I'm using Tornado 1.0.1 under NT 4.0. Thanks! - -kin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: universe2 vme interface chip with pio board problem Date: Fri, 03 Sep 1999 15:29:03 -0500 From: Marty Schrader Organization: Fermilab Experimental Astrophysics Group Message-ID: <37D02F8F.3AB5ED04@fnal.gov> References: <7qmj1v$24$1@nnrp1.deja.com> Uh, this may seem like a silly question, but are you possibly running into a byte swapping problem? The Universe II has some built-in swapping under certain conditions, as I recall. - -- Marty Schrader x6645 Fermilab Experimental Astrophysics Group (SDSS) consult0@fnal.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Not suited to High Availability??? Date: Fri, 27 Aug 1999 15:02:08 GMT From: denesyk@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7q6998$lha$1@nnrp1.deja.com> A recent article in the Compact PCI Systems (July-August '99, p44) claims that process-based RTOS's are superior to thread-based RTOS's for High Availability (HA) systems. and that thread-based RTOS's cannot provide any real degree of fault tolerance to application SW bugs. It said only process-based RTOS's have the mechanisms for 'trapping, isolating, containing, and cleaning up after' bugged tasks. Can HA system developers credibly claim 99.999% availability if their boxes are xWorks-based? Comments??? Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why stay with VXWorks? Date: Fri, 03 Sep 1999 13:49:58 -0700 From: Latch Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <000b8d9b.de388e51@usw-ex0102-013.remarq.com> References: <19990728030931.25207.00003136@ng-cr1.aol.com> <7nt6e9$3h4$1@nnrp1.deja.com> + The VxWorks kernel is very solid. The architecture makes the whole thing very portable. The Posix interface makes it Unix compatible. The Tornado-tools are excellent. - - WindRiver goes the Microsoft(which everybody hates and mimics) way, try to cash on applications.They are NOT the best you can choose and most of the time very frustrating, Support is lousy, because support-people look on their position as a temporary as soon as they got more experience they move somewhere else. Customers are not equal. Have worked for a big company and small one I've noticed a huge gap in the relation scheme. VxWorks is pretty pricy - I will not recommend it for small budgets. Source code is not available ( unless you pay big bucks ) * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: any vme pio suggestions? Date: Fri, 03 Sep 1999 21:31:46 GMT From: ywu@imatron.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qpeo2$2gh$1@nnrp1.deja.com> Hi: Since I can't get my xycom parallel I/O board work together with universe II interfaced power pc bus master. I am looking for possible replacement of this PIO board. Does anybody have any suggestion about an PIO board that will work with power PC? thanks - ----Yongjian Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: float in shell Date: Fri, 03 Sep 1999 22:13:40 GMT From: ywu@imatron.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qph6c$4co$1@nnrp1.deja.com> References: <37CBAC0C.522D@ensem.u-nancy.fr> Yes, the shell defaults every variable to integet. you can cast you argument into float. it will work. like: > foo((float) 20.01) Hope this helps - ----Yongjian > Hi, > When I execute function with float or double as parameter from the > shell, the parameter is badly transmitted to the function. Is there a > way to overcome to this problem ? > Thanks > Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks,comp.os.lynx,comp.os.chorus Subject: Re: vxWorks OK for High Availability systems ??? Date: 27 Aug 1999 19:22:16 GMT From: Jeff.T.Williams@usa.alcatel.com (Jeff Williams) Organization: DSC Communications Corporation Message-ID: <7q6oh8$g27$1@relay1.dsccc.com> References: <7q6aur$mr4$1@nnrp1.deja.com> No opinion on vxworks. Have been using threaded code on Lynxos in a hig availability environment for about 2 years. Haven't seen the slightest hint of a code trampler. Haven't seen any indication that our threaded code is less reliable than the non-threaded code. YMMV. In article <7q6aur$mr4$1@nnrp1.deja.com>, denesyk@my-deja.com writes: |> July / Aug issue of CompactPCI Systems, page 44 suggests that thread- |> based RTOS's are not appropriate for systems with a requirement for |> five 9's availability (99.999 % available). |> |> The argument goes: |> Because tasks in a thread-based RTOS share a common memory space, such |> RTOS's will not offer good mechanisms to protect the target from memory |> stompers. |> |> Has someone out there got an opinion about the choice of vxWorks as |> RTOS for products with a heavy HA requirement? |> |> |> Sent via Deja.com http://www.deja.com/ |> Share what you know. Learn what you don't. - -- Jeff Williams - Alcatel USA. Did you know that there is enough sand in North Africa to cover the entire Sahara desert? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Awesome Silicon Valley Start-up Date: Fri, 03 Sep 1999 18:01:34 -0500 From: Marty Schrader Organization: Fermilab Experimental Astrophysics Group Message-ID: <37D0534E.62CE89AD@fnal.gov> References: <7o7dg2$sod@journal.concentric.net> <37A8B3AA.780C28DD@audesi.com> You guys can't possibly so silly as to think this is real, can you?!? This is another headhunter on a fishing expedition. He needs to expand his resume base. I may be wrong, but lots of experience in the Chicagoland market convince me otherwise... - -- Marty Schrader x6645 Fermilab Experimental Astrophysics Group (SDSS) consult0@fnal.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II with Tornado I Date: Fri, 03 Sep 1999 18:43:21 -0700 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <37D07939.79D9@despam.pangea.ca> References: <37ce25b3.155893292@news.bournemouth-net.co.uk> <37CEA1D9.5765E0EF@bitband.com> <37CEB633.3CAF290@ciprico.com> <0c2b02cf.d00b482d@usw-ex0102-013.remarq.com> Latch wrote: >[sensible stuff] Hey Latch, how is life in California? > Some of the low level API functions are not available > anymore in Tornado II. > What is missing? It will be interesting to see what VxWorks 6.0 brings... - -het - -- "you don't really appreciate how smart a moron is until you try to design a robot..." -a robotics designer Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: T2, mv2303 & 100Mbit lan => dc no carrier Date: 4 Sep 1999 00:29:04 GMT From: antles@spk.hp.com (Tom Antles) Organization: Hewlett Packard Corvallis Site Message-ID: <7qpp4g$glg$1@hpcvnews.cv.hp.com> We just got Tornado 2 running on our mv2303. The problem is, whenever we run on a 100Mbit network we get the "interrupt: dc0 - no carrier" message. We NEVER see this message when running on 10Mbit network. With lots of LAN traffic we get so many of these messages that the log task can not keep up. Ex: "logTask: 124 log messages lost." I have tried modifying the "#define DC_MODE" macro in config.h according to earlier postings to this group. I have tried each of the values: 0x8, 0xc, 0x18, 0x1c (see if_dc.h macros below) on both the 100Mbit network and a 10Mbit half duplex network. However the result is always the same: 100Mbit gives "interrupt: dc0 - no carrier", 10Mbit does not. Does anyone know what is going on here? macro defines from if_dc.h: /* Modes for 100-Mb/s Ethernet configuration */ #define DC_100_MB_FLAG 0x04 /* 100-Mb/s mode */ #define DC_21140_FLAG 0x08 /* Lance chip is 21140 */ #define DC_FULLDUPLEX_FLAG 0x10 /* Full Duplex Mode */ #define DC_SCRAMBLER_FLAG 0x20 /* MII/SYM in scrambler mode */ #define DC_PCS_FLAG 0x40 /* MII/SYM in symbol mode */ #define DC_PS_FLAG 0x80 /* Serial Port selected */ #define DC_ILOOPB_FLAG 0x100 /* Internal Loop Back Mode */ #define DC_ELOOPB_FLAG 0x200 /* External Loop Back Mode */ #define DC_HBE_FLAG 0x400 /* Heart Beat Enable Flag */ More interesting facts: 0) any kind of lan traffic directed to the mv2303 processor results in 1 or more "interrupt: dc0 - no carrier" messages. Ex: windShell commands, ftping files. 1) when pinging the vxWorks processor with 64 byte packets, I get exactly 1 "interrupt: dc0 - no carrier" message per ping. 2) By executing the ifShow command, (see below) there does not appear to be ill behavior of the network. dc (unit number 0): Flags: (0x8063) UP BROADCAST MULTICAST ARP RUNNING Type: ETHERNET_CSMACD Internet address: 15.2.2.156 Broadcast address: 15.255.255.255 Netmask 0xff000000 Subnetmask 0xff000000 Ethernet address is 08:00:3e:27:04:4d Metric is 0 Maximum Transfer Unit size is 1500 56754 packets received; 57319 packets sent 53781 multicast packets received 45326 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped 3) There is NO performance difference when running 100Mbit or 10Mbit. Each downloads a 15Mbyte file at the same transfer rate: 226 Transfer complete 15396383 bytes received in 31.75 seconds (473.51 Kbytes/s) 4) occassionaly, there must be a burst of unrequested lan traffic as we suddenly see 100's of "interrupt: dc0 - no carrier" messages. Tornado vxWorks - -- Tom Antles antles@spk.hp.com Phone: (509) 921-3590 Fax: (509) 921-3991 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: start target server outside of Tornado? Date: Fri, 3 Sep 1999 22:53:54 -0700 From: "John Finley" Organization: Kivala Systems, Inc. Message-ID: <7qqc2e$oer@dfw-ixnews15.ix.netcom.com> References: <19990903202514.19003.qmail@nym.alias.net> Kin Cho wrote in message news:19990903202514.19003.qmail@nym.alias.net... > Is it possible to start target server that I've configured > outside of Tornado? That's, I'd like to start a target server > as a Startup program under NT. First create a shortcut that starts a DOS box. Then, in Tornado, configure your target server so it starts up nicely. Select and copy the text down at the bottom of the target server configuration dialog. Then, back in the dos box "properties", paste the target server command. Drag the shortcut into your startup folder and you're done. I'm not sure how you will make sure the portmapper and registry start before the target server. John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Tornado Apps with Visual Basic, LabVIEW - -------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks porting to strongARM Date: Fri, 3 Sep 1999 23:00:49 -0700 From: "John Finley" Organization: Kivala Systems, Inc. Message-ID: <7qqcfc$ol6@dfw-ixnews15.ix.netcom.com> References: <7qnjtc$h80$1@news.etri.re.kr> <37CFC548.FA5FBF4C@lucent.com> Douglas Fraser wrote in message news:37CFC548.FA5FBF4C@lucent.com... There is a BSP for the SA110 that supports the ebsa-285 that is available from Intel. I believe there is also another board with an SA1100 (Brutus??) that has a Tornado BSP. Doug "À̼®Áø" wrote: > > At Tonado directory, I can find ARM porting supporting file, but strongARM. > > Where can I find strongARM & Vxworks supporting. Yup, Brutus appears well supported; the BSP, UGL, and PJWorks all work as shipped with brutus and his LCD/touchscreen. Incidentally, the CPU define is the same for SA110 and 1100. (I'm gettin' there, Doug...) John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Tornado Apps with Visual Basic, LabVIEW - -------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II with Tornado I Date: Fri, 3 Sep 1999 23:11:46 -0700 From: "John Finley" Organization: Kivala Systems, Inc. Message-ID: <7qqd3v$p5q@dfw-ixnews15.ix.netcom.com> References: <37ce25b3.155893292@news.bournemouth-net.co.uk> <37CEA1D9.5765E0EF@bitband.com> Leonid Rosenboim wrote in message news:37CEA1D9.5765E0EF@bitband.com... > Unlike UNIX, Windows can not maintain multiple installations of Tornado on > the same machine, so you will have to leave one or two dedicated NT > machines with Tornado 1 on them, while installing Tornado 2 on other > machines. WRS has this clearly documented on the README notes shipped with > every CD! > > Assuming your company opted for PCs to save money, now its time to pay for > that decision... > > An old proverb says that only rich people can afford buying cheap things. Hmmm. I have T1-x86, T1-PPC, T1-960, T2-beta-PPC, and T2-ARM running on the same PC. I haven't found anything that doesn't work yet. They each have an entirely separate installation tree, though, which, at a penny per MB, costs about ten bucks each. I have noticed that there are separate registry keys for T1 and T2, for whatever that's worth. John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Tornado Apps with Visual Basic, LabVIEW - -------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: mpc860 i2c problem Date: Tue, 31 Aug 1999 12:38:49 GMT From: bbilir@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qgicn$gql$1@nnrp1.deja.com> Is there anyone who encountered a Tx collision error in one master environment. Is this one of MPC860 i2c bugs? Thank you all Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks,comp.os.lynx,comp.os.chorus Subject: Re: vxWorks OK for High Availability systems ??? Date: Tue, 31 Aug 1999 13:49:01 +0100 From: "Kevin Pope" Message-ID: <936104058.4542.0.nnrp-06.d4e53aea@news.demon.co.uk> References: <7q6aur$mr4$1@nnrp1.deja.com> A process based message passing RTOS allows the individual memory space around each process to be protected via any CPU memory management. The message passing model then allows the operating system to perform IPC from memory space to memory space. One such operating system is OSE, details of which can be found at www.enea.com I work for OSE in the UK, so I cannot claim to be unbiased, but I suggest you check out the OSE web site for our distributed/HA products. Regards, Kevin. wrote in message news:7q6aur$mr4$1@nnrp1.deja.com... > July / Aug issue of CompactPCI Systems, page 44 suggests that thread- > based RTOS's are not appropriate for systems with a requirement for > five 9's availability (99.999 % available). > > The argument goes: > Because tasks in a thread-based RTOS share a common memory space, such > RTOS's will not offer good mechanisms to protect the target from memory > stompers. > > Has someone out there got an opinion about the choice of vxWorks as > RTOS for products with a heavy HA requirement? > > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: start target server outside of Tornado? Date: Sat, 04 Sep 1999 07:49:58 GMT From: David Hamel Organization: @Home Network Message-ID: <37D0CF26.684B515F@ntanet.net> References: <19990903202514.19003.qmail@nym.alias.net> <7qqc2e$oer@dfw-ixnews15.ix.netcom.com> John Finley wrote: > Kin Cho wrote in message > news:19990903202514.19003.qmail@nym.alias.net... > > Is it possible to start target server that I've configured > > outside of Tornado? That's, I'd like to start a target server > > as a Startup program under NT. > > First create a shortcut that starts a DOS box. Then, in > Tornado, configure your target server so it starts up > nicely. Select and copy the text down at the bottom of > the target server configuration dialog. Then, back in the > dos box "properties", paste the target server command. > Drag the shortcut into your startup folder and you're > done. > > I'm not sure how you will make sure the portmapper and > registry start before the target server. Run them as a services and never think about it again..... --------------------------- Newsgroups: comp.os.vxworks,Message,based,library Subject: Message-based library on top of VxWorks Date: Mon, 30 Aug 1999 09:50:54 +0200 From: "Øyvind Teig" Organization: Navia Maritime AS, division Autronica Message-ID: <37CA37DE.62FDAA02@computer.org> Would anybody know of a message-based library that runs on top of VxWorks or is easily ported to it? Oyvind Teig PS. This is a continuation of the "CSP library for VxWorks" thread in this group, where I got an excellent reply for source on top of pthreads and unix. I'd really like something closer to VxWorks. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sun Sep 5 04:00:15 1999 From: daemon@csg.lbl.gov Date: Sun Sep 5 04:00:19 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Sep 5 04:00:11 PDT 1999 Subject: To connect a serial mouse to MPC823ADS Subject: Re: vxWorks OK for High Availability systems ??? Subject: PCI busmaster DMA and cache Subject: Re: How to increase the Stack Size?? Subject: Re: _udivsi3 / _mulsi3 Subject: Re: Not suited to High Availability??? Subject: Re: Watchdog timer in VxWorks. Subject: float in shell Subject: Re: UDP [socket] performance issue in VxWorks ? Subject: Re: are static variables global? Subject: (More) Shared memory problems ... Subject: Re: Scalability dilemma Subject: Re: Scalability dilemma Subject: Re: checking for process over-run Subject: Re: UDP [socket] performance issue in VxWorks ? Subject: Auxilliary Clock Subject: Re: Long File Name support TFFS Subject: Re: Raw ethernet packet hooks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: To connect a serial mouse to MPC823ADS Date: 30 Aug 1999 11:14:14 GMT From: "Clark" Organization: DCI HiNet Message-ID: <01bef2d9$14445480$c89247d2@server2> Hi, I want to connect a serial mouse to MPC823ADS. Do you have any reference( HW&SW) about this application? RTOS:Vxwork Target:MPC823ADS Clark --------------------------- Newsgroups: comp.os.vxworks,comp.os.lynx,comp.os.chorus Subject: Re: vxWorks OK for High Availability systems ??? Date: 30 Aug 1999 12:45:12 GMT From: Jeff.T.Williams@usa.alcatel.com (Jeff Williams) Subject: Re: vxWorks OK for High Availability systems ??? Organization: DSC Communications Corporation Message-ID: <7qduco$b0e$1@relay1.dsccc.com> References: <7q6aur$mr4$1@nnrp1.deja.com> <7q6oh8$g27$1@relay1.dsccc.com> <7qa4d8$4uc$1@nnrp1.deja.com> Never saw tramplers in either the OS or the apps. I have seen plenty of bugs in my apps and some bugs in the OS (although Lynx is very responsive). I don't expect perfect code (if I can't write, why should anyone else do so:), but I do expect good support. If you design your system well, you can incorporate mechanisms for coping with a wide range of problems. The biggest problem with a multiprocessing system is trying to consider all of the interactions you can get (well, that's my biggest problem - YMMV). Things which you consider to be linear may normally be linear but may, on occasion, not be linear. Based on my experience (13 years, 3 companies), the article to which you refer is probably optimistic. Some defects (particularly design defects) are so subtle that the code may live in the field for years without detection. Then the correct set of circumstances arises and you suddenly have a problem that has to be solved immediately. Bottom line is that while a good OS and good tools help, it's really the guy behind the wheel that makes the difference. In article <7qa4d8$4uc$1@nnrp1.deja.com>, denesyk@my-deja.com writes: |> Hi Jeff, |> Thanks for the reply. |> Are you saying you never saw tramplers in Lynx itself, or in your |> application SW. If you mean in your application, my hat is off to you. |> |> The cPCI mag article made the point that it is quite |> typical/normal/believeable to have between 1 and 5 SW defects per 1000 |> lines of application code that are not detected prior to shipping the |> product. |> |> Process-based RTOS are (apparently) able to give us clever app |> developers mechanisms to create systems that are capable of tolerating |> misbehaving applications without significant/sustained loss of system |> function. |> Regards, |> Tom - -- Jeff Williams - Alcatel USA. Did you know that there is enough sand in North Africa to cover the entire Sahara desert? --------------------------- Newsgroups: comp.os.vxworks Subject: PCI busmaster DMA and cache Date: Mon, 30 Aug 1999 14:00:32 +0100 From: "Rui Prior" Message-ID: <37ca7fdf@readers.ip.pt> I'm writing a driver for a PCI card which performs busmaster DMA. On Linux there are no cache problems with PCI devices performing DMA. (Unlike ISA devices, for which buffers have to be allocated using a special flag.) On VxWorks, do I have to use cacheLib functions (cacheFlush(), cacheInvalidate()) in order to avoid coherence problems, or is it unnecessary? (BTW, the platform is x86) Rui Prior --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to increase the Stack Size?? Date: Mon, 30 Aug 1999 15:15:53 +0200 From: "Werner Schiendl" Organization: VBS-Newsserver Message-ID: <936018959.454531@news.vbs.at> References: <7q2qge$8nk$1@nnrp1.deja.com> <935658000.275567@news.vbs.at> <7q4s49$m59$1@nnrp1.deja.com> Hi, - -- snip -- > > In my case, TaskStack is the parent task and it spawns several other > tasks. > > Am i correct in saying that to start a parent task of desired stack > size, the only way is to spwan it from usrRoot() at the end of the > vxworks booting. > > regards, > s.shiva > - -- snip -- see $(WIND_BASE)/target/config/all/usrConfig.c, function usrInit The call to kernelInit (very last call in the function, line 219 in my installation) specifies the function that will be spawned as root task (the default is usrRoot) and the stack size (ROOT_STACK_SIZE). You should be able to change ROOT_STACK_SIZE to meet your requirements with the project facility (operating system components/kernel components/kernel) on the Params the of the Properties dialog. HTH - -- Werner Schiendl International Applications - -- mailto:ws-news@gmx.at - -- B&R Industrial Automation http://www.br-automation.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: _udivsi3 / _mulsi3 Date: Thu, 02 Sep 1999 15:24:52 GMT From: Ran@127.0.0.1 (Ran Shalgi) Organization: Only when absolutely necessary Message-ID: <37ce9468.284231503@news.inter.net.il> References: <7ql27i$29t$1@news2.kornet.net> On Thu, 2 Sep 1999 14:43:25 +0900, "Á¤Áø±º" wrote: >Undefined symbols: >___udivsi3 >___mulsi3 >Warning: object module may not be usable because of undefined symbols. > >My Division and Multiplication pgm had some problem. The above messge told >me "Something wrong " > >Could you help me??? > You may have tried to use multiplication or division on a 64-bit integer type, or something similar which is not covered in the standard VxWorks kernel build. You'll need to extact the missing .o files from your archives and load them into your target. See http://www.deja.com/getdoc.xp?AN=118929446 for an example. In short, search for a file named libgcc.a on your host machine and run "ar386 -tv libgcc.a" (replace ar386 with arYOUR_ARCH). Ran Shalgi Expand Networks, Inc http://www.expand.com - --- Replace 127.0.0.1 with expand.com when replying by e-mail --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Not suited to High Availability??? Date: Mon, 30 Aug 1999 19:17:44 GMT From: Alex Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qelcb$630$1@nnrp1.deja.com> References: <7q6997$lh9$1@nnrp1.deja.com> In article , "Jeff Szczepanski" wrote: > I didn't read the article, but I would have to say that the author is > working from some bad assumptions here. Was this written by someone who is > associated with LynxOS? :-> > > 1) The first assumption from this claim is that the kernel has to be > completely bug free for the application. If the kernel crashed in the > process model case you are hosed anyway. Extreme example: Windows NT is > fully process based & crashes fairly regularly! But I believe most drivers live in or at the NT & W95 kernel! This is probably the cause for most of the MS OSs crashes (ignoring application s/w for the moment). > 2) Drivers tend to be inside the protection ring of the process model...How > many real embedded systems do you know that don't have atleast one custom > driver for some hardware interface/mechanism? ie: Atleast some of the > functionality will be inside of ring of protection. With a decent architecture you can divide & rule-keep the functionality required at this level at a minimum, plus keep the time spent at this level at a minimum. > 3) OK so process protection traps your application doing the real work, and > safely terminates it things go bad...now what? Restart the application. > First that's still not 'high availability' since it didn't prevent the > failure By definition you can't stop the failure, as it is a bug! The question is what & how to to if a failure occurs. This is what gives you higher availability compared to other architected OS. >and I can get the same functionality with a simple watchdog timer > and an automatic hardware reset when it expires. Not quite the same functionality. Restarting a process should always be faster than resetting the hardware. It should also be easier to recover the context of the previous process, in order to try and continue where things left off. With todays PC h/w in embedded systems, the reboot time is a big problem. You want to reset only as a last resort. This is also where "high availability" comes in-your system is up a bigger % of time compared to a h/w watchdog because of the faster restart time. This means that a process based OS can be available more than a thread based one! >On top of that, the > watchdog will cover you from kernel issues, the process model alone will > not. Agreed, the process model by itself is not good enough. Use a s/w watchdog in conjunction with a hardware one, for maximum protection and "higher availability". IMO Using processes rather than threads give you an advantage. > 4) Any additional aplpication protection/features from the process model > comes at a performance and/or code overhead cost. Not necessarily. > In summary, there are a set of trade-offs at play here. I think that for > most embedded applications, less sophisticated, simpler applications/kernels > are easier to implement robustly and to verify. Always tradeoffs, decisions, decisions... :) An important factor consider in these decisions (imo) is the architecture of the OS. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Watchdog timer in VxWorks. Date: Tue, 31 Aug 1999 12:15:48 +0200 From: Kokou DJATH Organization: CIRIL, Nancy, France Message-ID: <37CBAB54.5092@ensem.u-nancy.fr> References: <7qeam3$3ml1@overload.lbl.gov> T C Omprakaash wrote: > > Hi, > The routine that is going to be attached to a watchdog timer has all > restrictions that apply to ISRs. In case I need to invoke a function > with some specified timeout but with no such restrictions, what do i do. > are there any other timer utilities in vxworks. even if i use POSIX > timers, the restirctions exist. is there any roundabout way to overcome > these restirctions. > > Regards > Omprakaash One way to overcome these restrictions is to use 2 functions with semaphore. The function which will be connected to the watchdog must just give the semaphore, and the second function which is running in the task context must be blocked by the semaphore. Ihope it will help you --------------------------- Newsgroups: comp.os.vxworks Subject: float in shell Date: Tue, 31 Aug 1999 12:18:52 +0200 From: Kokou DJATH Organization: CIRIL, Nancy, France Message-ID: <37CBAC0C.522D@ensem.u-nancy.fr> Hi, When I execute function with float or double as parameter from the shell, the parameter is badly transmitted to the function. Is there a way to overcome to this problem ? Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP [socket] performance issue in VxWorks ? Date: Tue, 31 Aug 1999 11:24:06 GMT From: bill.beckwith@realtime-corba.com (Bill Beckwith) Organization: Objective Interface (http://www.ois.com) Message-ID: <8E334B568rwbnewsoiscom@sigma.ois.com> References: <7qdc42$87b$1@nnrp1.deja.com> colombo_daniele@my-deja.com wrote in <7qdc42$87b$1@nnrp1.deja.com>: >I test UDP performance too, on two kind of board, a PPC860 and a 68360. >The tests seems work, but when i generate a burst of packet, the >receiver lose some packets without any errors. This is the definition of UDP. Packet delivery is not reliable or ordered. - -- Bill --------------------------- Newsgroups: comp.os.vxworks Subject: Re: are static variables global? Date: Tue, 31 Aug 1999 09:29:48 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37CBD8CC.78830A5D@lucent.com> References: <12969-37CB5223-123@newsd-223.iap.bryant.webtv.net> Declaring j as static will not give you multiple instances of j, it just controls the scope of the variable. Where i is of global scope, j has file scope only. The difference is that you could have a 'static int j;' at the start of every C source file in your project, and each file would have its own instance of j that would be visible to all functions in that source file, but not visible to any function outside that source file. If you had an 'int i' at the start of every file, the linker will complain of multiple declarations of variable i. (I hope.) I suppose it could just silently collapse them all to one instance, but that would be rude! Doug anon789@webtv.net wrote: > > For example, int the following source file, all tasks spawned with f as > the entry point share the same variable i > > /*file: a.c*/ > int i; > static int j; > > void f() {...} > > int main() > { > taskSpawn(...,f,...); > taskSpawn(...,f,...); > } > > Question: does the static keyword change that? j --------------------------- Newsgroups: comp.os.vxworks Subject: (More) Shared memory problems ... Date: Wed, 01 Sep 1999 08:27:08 +0100 From: John Fallone Organization: GEC Marconi Avionics Limited Message-ID: <37CCD54C.2651@gecm.com> Reply-To: john.fallone@gecm.com (delurk) Apologies if this is straightforward, but I think I am missing a bit of the picture here. [Fairly long winded explanation coming up ...] I am trying to set up a system with three powerpc boards. Processor 0 has ethernet to the rest of the LAN, processors 1 & 2 have no ethernet and use the shared memory network to boot via processor 0. I think I have the basic setup working, since I can get either processor 0 and 1 operating, or 0 and 2, however when I try to get all three running either 1 or 2 stops communicating. Processor 0 communicates on the VME bus using an SVC64, and has the shared memory configured for VME interrupts. Processors 1 & 2 have a Tundra Universe and use mailbox interrupts for shared memory comms. I have configured both processors to put their tundra registers at different addresses on the VME backplane. All BSPs have been configured for Software TAS. It is probably easier to explain the problem using a couple of scenarios to show the effect Scenario 1: boot processor 0, starts OK, with shared memory network up and running, showing one processor. boot processor 1, starts OK, shared memory network appears to be running as expected, both machines can ping each other ... so far so good boot processor 2, starts OK, shared memory network on all three processors show the three cards with the correct sets of parameters. Ping from processor 0 to processor 2 is working, however processor 1 can now no longer be pinged from either processor 2 or processor 0. Scenario 2: repeat the above, this time booting proc2 first and then proc 1, in this case it is processor 2 which drops out. I have checked the visibility of the register areas for both cards and it seems to be okay. Processor 0 can see the tundra registers (and hence the mailbox reg) of both of the other cards. Both of the other processors can see the shared memory anchor, and shared memory structures on the master. Final point, I have stuck a VME bus analyser on the backplane, and when pinging from the master to the other two procesors I can only see the mailbox write going out for the working processor. So this suggests (to me) that the master processor has lost some information about the way to notify the "dead" card about new messages. Think that's about all I know about this at the moment, any help greatly appreciated. cheers JF --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Scalability dilemma Date: Wed, 01 Sep 1999 10:51:33 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <37CCE915.33A6391B@bitband.com> References: <7qhq2n$g6q$1@nnrp1.deja.com> Reply-To: leonid@bitband.com mkesti@gv.net wrote: > snip] > > I now have a dilemma that stems from the scalable nature of vxWorks, > however. I can see that future versions of the downloadable > application programs may require OS components that were not > included in the ROM's vxWorks image because I had not forseen that > I would need them. > > I'm wondering how others are managing this issue? > Simple - you can extract any particular module from the VxWorks library by doing: arARCH xv $WIND_BASE/target/lib/lib${CPU}gnuvx.a anymodule.o which will put the specified object in your current directory, and can be later downloaded with loadModuleAt() etc. and later initialized as required by specific module. Good luck, Leonid P.S. This is in a way an FAQ, any tech support engineer could tell you that out the top of his head. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Scalability dilemma Date: Wed, 01 Sep 1999 10:51:51 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <37CCE927.5B3956DF@bitband.com> References: <7qhq2n$g6q$1@nnrp1.deja.com> Reply-To: leonid@bitband.com mkesti@gv.net wrote: > snip] > > I now have a dilemma that stems from the scalable nature of vxWorks, > however. I can see that future versions of the downloadable > application programs may require OS components that were not > included in the ROM's vxWorks image because I had not forseen that > I would need them. > > I'm wondering how others are managing this issue? > Simple - you can extract any particular module from the VxWorks library by doing: arARCH xv $WIND_BASE/target/lib/lib${CPU}gnuvx.a anymodule.o which will put the specified object in your current directory, and can be later downloaded with loadModuleAt() etc. and later initialized as required by specific module. Good luck, Leonid P.S. This is in a way an FAQ, any tech support engineer could tell you that out the top of his head. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: checking for process over-run Date: Sun, 29 Aug 1999 11:17:32 -0700 From: Douglas Jerome Organization: http://www.primenet.com/~jerome Message-ID: <37C9793C.8DBBF0BE@primenet.com> References: <37bfccbc.400273783@news.inter.net.il> > On Thu, 19 Aug 1999 18:04:26 -0800, rcw@DynRes.com wrote: > > >VxWorks users, > > > >I currently have a scheduling mechanism whereby a binary semaphore is > >given every 4ms (I'm using sysAuxClk running at 1KHz to do this). > > > >I have a task that must run cyclically so that task waits on that semaphore. > > > >But I would like to know if the task has run longer than expected. I was > >thinking something like a semPeek would be just what I needed: I could > >look to see if the semaphore was already available. If it > >was I could report an over-run and if it wasn't, just wait on it. > > > >But of course there is no semPeek(). > > > >Does anyone have any clean solutions for this type of problem? > > > >Ron Wagner In the code that gives the semaphore, check that a flag is clear. If so, then set the flag and give the semaphore. If the flag is still set then the task has overrun. In the task that takes the semaphore, clear the flag before taking the semaphore. I have a scheduler that should run on vxWorks, it is at http://www.primenet.com/~jerome/software.html - -- Douglas Jerome http://www.primenet.com/~jerome --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP [socket] performance issue in VxWorks ? Date: Wed, 01 Sep 1999 10:16:55 GMT From: colombo_daniele@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7qiuei$9fv$1@nnrp1.deja.com> References: <7qdc42$87b$1@nnrp1.deja.com> <8E334B568rwbnewsoiscom@sigma.ois.com> In article <8E334B568rwbnewsoiscom@sigma.ois.com>, bill.beckwith@realtime-corba.com (Bill Beckwith) wrote: > > This is the definition of UDP. Packet delivery is > not reliable or ordered. > > -- Bill > I know that, only with TCP is guaranteed, but is possible to tune when UDP/IP start lose ? Maybe by changing something in configuration files ? Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Auxilliary Clock Date: Wed, 1 Sep 1999 07:29:50 -0500 From: "Tom Manning" Organization: JSC Message-ID: <7qj644$t7@niven.ksc.nasa.gov> Let me try to restate my problem. My target is a PCI PPC 603. I am trying to schedule processes on the auxilliary clock. I use the sysAuxClkConnect function to schedule a routine. I try to kill all of my processes and delete the routine scheduled on the sysAuxClkConnect by passing NULL as the first argument. Then when I restart, calling sysAuxClkConnect again with the same routine it appears to be executed twice for every clock interrupt. Inside of the routine I log the taskID of the task which called sysAuxClkConnect. The first time I run the software of course there is only one task ID which is displayed. Each consecutive time I run I get another task ID displayed. So the second time there are two task IDs, the third time there are three task IDs, and so on. What am I doing wrong? Thanks, Tom --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Long File Name support TFFS Date: Sun, 05 Sep 1999 10:16:38 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <37D226E6.D19E4A6F@bitband.com> References: <7qooll$1en7@overload.lbl.gov> Reply-To: leonid@bitband.com Assuming you are using the old dosFs, which is shipped with Tornado 2, then you are out of luck: TrueFFS is very nosy about the DosFs structure stored on its volumes, and the only safe way to use it is to format a TrueFFS volume only with the format function supplied with TrueFFS. Never use dosFsMkfs() to format a TrueFFS volume. DosFs long names are non-standard, and can be created only with dosFsMkfs(), and are known to cause file data to dissapear when used with TrueFFS, because TrueFFS's idea of the meaning of some fields in the boot block and DosFs's idea vary. This has been documented in the SPR database. Your solution is to either go to DosFs 2 which supports standard VFAT long names, or to do away with TrueFFS. Frankly I do not see why anyone would still bother with TrueFFS at all, you can get lower price per Megabyte with ATA-based Compact Flash cards, and not worry about any software issues, and remain vedor independent, buy the CF cards where it is cheaper. HTH, Leonid Anurag Srivastava wrote: > > Hi everybody, > > > > I need long file name support on TrueFFS. I am making DOS file system > > with long file name support and it is working . Does this support is > > provided in tffs for Tornado. If yes , then please send me the command > > to do s. > > > > Thanks, > > > > Anurag. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Raw ethernet packet hooks Date: Sun, 05 Sep 1999 09:17:48 GMT From: Ran@127.0.0.1 (Ran Shalgi) Organization: Only when absolutely necessary Message-ID: <37d2348f.521838373@news.inter.net.il> References: <7qjmun$1eo4@overload.lbl.gov> <37ce3152.258865148@news.inter.net.il> <37CE98E9.7F1D@gecm.com> On Thu, 02 Sep 1999 16:34:01 +0100, David Rigg wrote: >Ran Shalgi wrote: > >> Two options are available to you: >> 1. If you wish to continue working with etherHooks, you can simply >> switch back to old BSD 4.3 drivers. Play around with your kernel >> configuration, remove references to END and add instead old sytle >> drivers, and all will be well. >> > >are you sure? >we've been told by WRS that SENS is the only option in T2 >and there's no going back to BSD :( I tried it and it worked. The stack is still SENS, but the driver is BSD-style. Apparently, the new T2 SENS stack still retains backward compatibility with older BSD drivers, probably because WRS doesn't expect us all to re-write all our drivers from scratch overnight. Of course, some of the new features offered by SENS will not be active if you work through an older driver. Ran Shalgi Expand Networks, Inc http://www.expand.com - --- Replace 127.0.0.1 with expand.com when replying by e-mail --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Sep 6 02:06:34 1999 From: "tim allen" Date: Mon Sep 6 02:06:38 PDT 1999 Subject: Re: Auxilliary Clock >I use the sysAuxClkConnect function to schedule a routine. >I try to kill all of my processes and delete the routine scheduled on the >sysAuxClkConnect by passing NULL as the first argument. Then when I >restart, calling sysAuxClkConnect again with the same routine it appears to >be executed twice for every clock interrupt. You'll probably find that the sysAuxClkConnect function includes a call to sysHwInit2(). This is the vxWorks way of actioning the second phase of hardware initialisation. If you have source to your auxiliary clock driver then you can check this. sysHwInit2 is a BSP function (in sysLib.c) which, amongst other things, connects system interrupts, including the auxiliary clock. This means that each time you call sysAuxClkConnect, _all_ your system interrupts are connected again and they will all be called twice. If you have access to the source, then you can move the sysHwInit2 call, or prevent it from being called more than once. Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com From owner-vxwexplo-process Mon Sep 6 02:29:58 1999 From: David Laight Date: Mon Sep 6 02:30:02 PDT 1999 Subject: Re: VxWorks porting to strongARM John worte: > Subject: Re: VxWorks porting to strongARM > Date: Fri, 3 Sep 1999 23:00:49 -0700 > From: "John Finley" > Organization: Kivala Systems, Inc. > Message-ID: <7qqcfc$ol6@dfw-ixnews15.ix.netcom.com> > References: <7qnjtc$h80$1@news.etri.re.kr> <37CFC548.FA5FBF4C@lucent.com> > ... > > Where can I find strongARM & Vxworks supporting. > > Yup, Brutus appears well supported; the BSP, UGL, > and PJWorks all work as shipped with brutus and his > LCD/touchscreen. Incidentally, the CPU define is > the same for SA110 and 1100. (I'm gettin' there, > Doug...) I can't agree with 'well supported': 1) The FP routines are very very very slow. 2) The interrupt handler scheme isn't designed to allow drivers to attach to all the cascaded (second level) interrupts. The interrupt scheme needs to allow for external devices (eg SA1101 and UCBV1200) that provide multiple interript sources - including simple GPIO pin interrupts. I'm up to 144 possible interrupt sources (including all gpio pins.) 3) The timer code drifts badly. Unfortunately the 60Hz tick has to be exactly 60Hz to keep the vxworks RTC functions accurate. (You can't actually do 60Hz! 100Hz gives much better POSIX timers anyway.) 4) MMU setup is inappropriate for a simple RTOS. The SA1101 has its registers on 0x400 byte boundaries - you need to use 1Mb segments (not 4k pages) to map these efficiently. In fact 1Mb pages are fine for a simple SA1100 system. (You have to use the mmu to get DRAM at address 0.) The list goes on... David > > John > > - -------------------------------------------------- > John Finley Kivala Systems, Inc. > Project Manager (619) 689-0032 > john@kivala.com http://www.kivala.com > - -------------------------------------------------- > VxTool - Tornado Apps with Visual Basic, LabVIEW > - -------------------------------------------------- ---------------------------------------------------------------- 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 Sep 6 03:03:41 1999 From: David Laight Date: Mon Sep 6 03:03:44 PDT 1999 Subject: multiple 'int i;' All, There seems to be some confusion - and some comletely incorrect statements - about what 'int i;' actually generates. int i; generates a COMMON data item size 4 (typically); extern int i; generates an undefined symbol int i=0; generates a definition and allocates space (use nmxxx -f svsv xxx.o to get the symbol type displayed) When you link modules together (ld -r) multiple definitions are (normally) errorred and COMMON items are merged. A COMMON item will be discarded if a definition exists. When you generate a fully fixed up file (eg an a.out program, of a vxworks boot image) the space required by any remaining COMMON items is allocated in the bss (ie as zeros). If you dymamically load an object module onto an existing image (vxworks ld, unix dlopen(), unix kernel modload() etc) then COMMON items will be fixed up against the matching name in the image if one exists, if not the loader will allocate space in the bss of the module being loaded. This can cause problems - especially if you haven't been policing the namespace properly as the sharing of the data might not be what was expected. As well as having two parts of the system sharing a data item, the following problems also occur: 1) if the new module expects the symbol to be larger that the space originally allocated, then other data (following the item) will get stomped on. 2) diagnostic tools (eg crash, adb -k etc) that have to build a symbol table that matches the system image can fail to determine the location of common data. I find it useful to avoid these issues by never generating COMMON data items - except when they are definitely needed (eg errno). 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 Sep 6 04:00:09 1999 From: daemon@csg.lbl.gov Date: Mon Sep 6 04:00:14 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Sep 6 04:00:04 PDT 1999 Subject: Re: On scsiRdSecs() routine Subject: Splitting Vxworks into 2 Sections Subject: Re: Awesome Silicon Valley Start-up Subject: Re: NET+ARM Ethernet & ARM Processor ??? Subject: Re: question! Subject: Re: T2, mv2303 & 100Mbit lan => dc no carrier Subject: Father task should block until all childred tasks have terminated ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: On scsiRdSecs() routine Date: Sun, 05 Sep 1999 15:25:05 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <37D26F31.D642D335@bitband.com> References: <37CFF5B0.D327FAB@postech.ac.kr> Reply-To: leonid@bitband.com If you are doing parallel reads, you should keep in mind that there is a single SCSI bus, and possibly a single disk, each could become your performance bottleneck. Also, since each each task would access a different part of the disk, the disk would spend most of its time seeking between the areas used by each of the two tasks, instead of reading actual data. HTH, Leonid http://www.bitband.com Young Jin Nam wrote: > Hello, > > I found that "using scsiRdSecs() routine in parallel on the single > SCSI channel" degraded > lots of disk read performance. "In parallel" means that more than > one tasks do call the > scsiRdSecs() routine at the "almost" same time. (do I need to > serialize the calling routine?) > > Any comments on this problem will be very appreciated. Thanks in > advance. > > Young Jin --------------------------- Newsgroups: comp.os.vxworks Subject: Splitting Vxworks into 2 Sections Date: Sun, 5 Sep 1999 15:20:14 -0500 From: "Philip Searcy" Organization: MindSpring Enterprises Message-ID: <7quj2s$pdp$1@nntp5.atl.mindspring.net> Reply-To: "Philip Searcy" The hardware I'm working with forces the boot flash to be 2M from the top of memory. Unfortunately, we are supposed to provide 16M of flash space for VxWorks and applications to be run ROM resident. So, one idea is to put some or all of VxWorks in the 2M space and the rest somewhere else in lower memory. How do I do this? If I build the application into VxWorks, how do I make the build put it in specific places in memory? If I build the application separately from VxWorks, how do I burn it into flash with external references to VxWorks routines resolved? I was thinking of building VxWorks in 2 stages. First build romStart, the assembly language part of the boot, separately from the rest of VxWorks. Then build the rest of VxWorks without romStart. I could then burn the boot part of flash with romStart and the rest could go in the lower flash. Any ideas on how I could do this? First of all, romStart would need to be modified to jump to the lower flash. Also I would have to modify the VxWorks build somehow to build without romStart. Any ideas would be appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Awesome Silicon Valley Start-up Date: Fri, 03 Sep 1999 13:52:24 -0700 From: Latch Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <000b8d9b.dedad383@usw-ex0102-013.remarq.com> References: <7o7dg2$sod@journal.concentric.net> Tell me more details. Latchesar Stoyanov (Sr. Software Designer) lstoyanov@truetime.com * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NET+ARM Ethernet & ARM Processor ??? Date: Fri, 03 Sep 1999 13:58:15 -0700 From: Latch Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <1415c574.e061c8b8@usw-ex0102-013.remarq.com> References: <7nhqh4$6u2@overload.lbl.gov> Yes There is ARM7TDMI BSP for VxWorks available as well as for pSOS. * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: question! Date: Fri, 3 Sep 1999 01:01:43 GMT From: Charlie Grames Organization: Boeing Message-ID: <37CF1DF7.30C4CC46@nowhere.com> References: <37CAED64.AE2ACEB4@pegasus.cc.ucf.edu> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Lei, I expect than "CLOCKS_PER_SEC" and "_SC_CLK_TCK" are BSP specific, and I cannot comment on them. To delay for one second, use: taskDelay(sysClkRateGet()); sysClkRateGet() returns the number of ticks per second. By default, its value is 60. Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com Lei Sun wrote: > > Hi: > Can anybody tell me what is the "tick", how to convert it into > time, suppose I have a MC68k cpu with 33MHz, what is the relationship > between tick and clock rate? I want to use taskDelay() > to delay a task by 1 sec, but I need to use "tick " as parameter, How do > I know that how many ticks equal 1 sec. > and also , what does the constant "CLOCKS_PER_SEC" and "_SC_CLK_TCK > " represent for? > thanks a lot! > > lei --------------------------- Newsgroups: comp.os.vxworks Subject: Re: T2, mv2303 & 100Mbit lan => dc no carrier Date: Mon, 06 Sep 1999 09:39:46 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. Message-ID: <37D36FC2.EE147F37@bitband.com> References: <7qpp4g$glg$1@hpcvnews.cv.hp.com> Reply-To: leonid@bitband.com Tom Antles wrote: > We just got Tornado 2 running on our mv2303. The problem is, whenever we > run on a 100Mbit network we get the "interrupt: dc0 - no carrier" message. > We NEVER see this message when running on 10Mbit network. My guess would be that the driver does not properly initialize the PHY unit for full-duplex operation, so you could try to configure your 100Mbps switch to set that particular port to half-duplex, and the problem may be circumvented. The ":no carrier" error is when the PHY does not hear itself when transmitting, and it is impossinble for it to hear itself when on a full duplex connection. Hope this helps, Leonid http://www.bitband.com --------------------------- Newsgroups: comp.os.vxworks Subject: Father task should block until all childred tasks have terminated Date: Mon, 06 Sep 1999 12:40:16 +0200 From: "Øyvind Teig" Organization: Navia Maritime AS, division Autronica Message-ID: <37D39A10.D56DB8E4@computer.org> Father task should block until all childred tasks have terminated ========================================= That is the semantics of CSP, and we need it for our implementation. How is this best done in VxWorks? Signal? Semaphore? Active polling Other? A join would be perfect! |====================|===================================|====|====| | Oyvind Teig | oyvind.teig@autronica.no | | | | Navia Maritime AS | oyvind.teig@computer.org | | | | division Autronica | |Tel:|Fax:| | 7005 | http://www.navia.no | +47| +47| | Trondheim | http://www.autronica.no |7358|7391| | Norway | http://www.autronica-maritime.com |1268|9320| |====================|===================================|====|====| --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Sep 6 08:24:44 1999 From: David Laight Date: Mon Sep 6 08:24:47 PDT 1999 Subject: re: DHCP tests I've finally managed to get my dhcp test: dhcpBind( dhcpInit( ifunit( "xircom", 0 ) ) ); to work. It only required three patches to the vxWorks binary (good job the code isn't readonly...). The 'problem' is that our DHCP server complies with rfc2131 not the obsolete rfc1541. rfc1541 REQUIRES the dhcp options buffer to be 312 bytes long, whereas rfc2131 places no such restriction - allowing the buffer to be any length (subject to the negotiated maximum message size). This means that rfc2131 isn't a proper superset of rfc1541. The VxWorks dhcp code rejects messages where the options buffer isn't at least 64 bytes long (the length of this field in bootp). Patching the length in the compare instructions (IP message length in align_msg, IP message length and UDP message length in dhcpcBodyCheck) fixes the code. I also noticed that the code will always reject the responses to the retransmittions of its DHCPDISCOVER message - the xid field in the LEASE_DATA structure doesn't match that sent in the request. I see two mails to wrs support coming... 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 Sep 6 09:14:19 1999 From: David Laight Date: Mon Sep 6 09:14:23 PDT 1999 Subject: Re: T2, mv2303 & 100Mbit lan => dc no carrier > Subject: Re: T2, mv2303 & 100Mbit lan => dc no carrier > Date: Mon, 06 Sep 1999 09:39:46 +0200 > From: Leonid Rosenboim > Organization: BitBand Technologies Ltd. > Message-ID: <37D36FC2.EE147F37@bitband.com> > References: <7qpp4g$glg$1@hpcvnews.cv.hp.com> > Reply-To: leonid@bitband.com > > My guess would be that the driver does not properly initialize the PHY unit > for full-duplex operation, so you could try to configure your 100Mbps switch > to set that particular port to half-duplex, and the problem may be > circumvented. It is more likely that the PHY unit behaves differently at 10M and 100M and that the driver is not allowing for this. Certainly the Natsemi DP83840 (on suns sbus fast ethernet card (FEPS)) returns transmitted frames at 10M but not at 100M (so the device driver receives frames sent to the broadcast address only at 10M). I didn't ever read the transmit status from the feps - so can't say whether it reported 'carrier' at all. (The driver was a common AMD lance/sun FEPS one - with 95% common code.) The 'no carrier' errors at 100M are certainly consistent with my observed behaviour of the 83840 PHY. Note that I forced HDX - the ethernet chipset has to be programmed to match the PHY if you want to run FDX. This is somewhat difficult, the standard MII registers defined in 8802-3 don't let you read the current link speed! I don't have the specs for the dec chipset (any more). > > The ":no carrier" error is when the PHY does not hear itself when > transmitting, and it is impossible for it to hear itself when on a full > duplex connection. More correctly, it will be reported by the ethernet chipset when the PHY doesn't report 'receive carrier' during a transmit. This error is an indication that either the receive pair, or transmit pair is broken between the ethernet chipset and the PHY. Breaks in the collision pair are indicated by (lack of) the SQE test. > > Hope this helps, > > Leonid Sounds like this driver is: a) buggy b) being far to verbose How you extract a fix from WRS is anybodies guess... If you have the driver source, remove the logMsg call and rebuild your vxWorks image. 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 Tue Sep 7 05:44:32 1999 From: Alexander_R_Povolotsky@res.raytheon.com Date: Tue Sep 7 05:44:36 PDT 1999 Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) Hello, Leaving subject(s) of relevance of RTOS features, such as: Virtual Memory, full POSIX compliance (including threads support) , "per task memory address protection", and "OS memory address space separation from the Task/User memory address space" vs the "ease" of implementing High Availability/Hot Swap ASIDE ;-) , it is the fact that LynxOS does support Hot Swap for Compact PCI and that VxWorks (yet) doesn't. --------------------------------------------------------------------------- ----------------------------------------------- LynxExpress Newsletter - Volume 15 (http://www.lynx.com/): Motorola Computer Group stole the show at CT Expo with a dramatic demonstration of the CPX8200 computer system. MCG's CPX8000 delivers "five-nines" capability (99.999%; less than five minutes and 15 seconds of downtime per year), and provides the computer power for applications used in "carrier-grade" telecom systems. The demonstration included a Microsoft Windows 2000-based full hot-swap system is the first real-time system that does hot swapping and failover of CompactPCI processor cards, done in combination with MCG's MCP750 intelligent I/O processors and LynxOS real-time operating system from Lynx Real-Time Systems. The CPX8000 is based on an open-architecture CompactPCI platform, which enables any active module to be exchanged for repair or upgrade while the system continues to operate. From owner-vxwexplo-process Tue Sep 7 07:30:03 1999 From: mmenge@dspt.com Date: Tue Sep 7 07:30:07 PDT 1999 Subject: Tornado 2 Mark Menge@DSPT 09/07/99 10:29 AM Dear Tornadians, and vxWorkers: 1. I have gone through the pain of upgrading to Tornado 2. However, before a project is finished, I have to be able to write a single batch file that gets all of the source code from the source code control and builds it all automatically. My problem is that all project information is stored in the .wpj file. How can I generate the generated files (prjComps.h prjComps.c and Makefile etc.) from an external batch file? Can I use tcl? From owner-vxwexplo-process Tue Sep 7 09:50:03 1999 From: David Laight Date: Tue Sep 7 09:50:08 PDT 1999 Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) 'Hot Swapping' is an interesting idea, and, although possible to demonstrate, may be more difficult to achieve reliably in real life. I did some work with hot-swapping of PCI cards under UnixWare 7. It was possible, but in practise you could swap a working card for another working one. Swapping out a broken card is MUCH more difficult, particulararly if it had failed catestropically while being accessed. The PCI bus is not resilient to device failures - eg a bus transciever dying. The added complexity of switching the the bus signals, probbaly means that a system that support hot-swapping is actually more likely to fail than one which does not. A board with a non-critical failure (eg a line transceiver being blown) can almost certainly be replaced. The suggestion that a MSW-2000 system is doing hot-swap of cpu cards seems amazing! A can't imaging a cpu card dying without trashing at least some of the memory associated with its current activity. In something with a very restricted application (eg a comms switch) it might be possible to stash away the important parameters for existing connections so that an alternate cpu (or a rebooted cpu) could continue in a relatively seamless way, but for a more general stateful application it seems a little pie in the sky. DEMONSTRATING something is, as I said earlier an entirly different ball game... David > it is the fact that LynxOS does support Hot Swap for Compact PCI and that > VxWorks (yet) doesn't. > --------------------------------------------------------------------------- > LynxExpress Newsletter - Volume 15 (http://www.lynx.com/): > Motorola Computer Group stole the show at CT Expo with a dramatic > demonstration of the CPX8200 computer system. MCG's CPX8000 delivers > "five-nines" capability (99.999%; less than five minutes and 15 seconds of > downtime per year), and provides the computer power for applications used in > "carrier-grade" telecom systems. > The demonstration included a Microsoft Windows 2000-based full hot-swap > system is the first real-time system that does hot swapping and failover of > CompactPCI processor cards, done in combination with MCG's MCP750 > intelligent I/O processors and LynxOS real-time operating system from Lynx > Real-Time Systems. The CPX8000 is based on an open-architecture CompactPCI > platform, which enables any active module to be exchanged for repair or > upgrade while the system continues to operate. > ---------------------------------------------------------------- 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 Tue Sep 7 11:02:25 1999 From: Tom Antles Date: Tue Sep 7 11:02:29 PDT 1999 Subject: Re[2]: dc no carrier vxWorks , Tornado ron Heald wrote: > Hi Tom, > > I saw your recent posting to com.os.vxworks concerning the message > "interrupt: dc0 - no carrier". I have the same problem on our new and > first mv2700 with the one exception that ours has only a 10Mbit > connection. This must mean that you have full duplex 10Mbit connection. > > In our case the whole thing is seems benign. I've been using the > machine for a couple of months now with no problems except the > messages. The things you said about losing log messages and your > "More interesting facts" match exactly for me. > several things here: 1) I believe this IS mostly benign. I say 'mostly' for the following reasons: 1) our serial port becomes unusable in some situations because of the volume of dc no carrier messages. This isn't quite benign. 2) logTask is being flooded with messages to the point it is in some cases losing 100+ messages. If it can't keep up with the messages, it would seem that this must affect system performance in some way, even if it is interrupt driven. 2) I don't understand why disabling duplex operation on the mv2303 using the DC_MODE bits doesn't fix our problem, since the switch is supposed to be autosensing. 3) Even if we use the mode bits to set to 10Mbit vs 100Mbit, the ftp performance is exactly the same. It all points to the DC_MODE bits not really working in this case. I wonder if autosensing overrides DC_MODE bits in normal operation. > We also have another mv2700 hosted on an NT machine with a 10Mbit > connection which has NOT seen the problem at all. I use a Solaris > host. because it is not Not full duplex? > > I submitted TSR# 137083 to WRS on this. They say this problem is > SPR# 22196 which has been around since Tornado II beta. Their answer > was to send an "unofficial" patch. That SPR is not listed on > WindSurf. I've asked them why it's not listed a couple of times now, > and never gotten a satisfactory answer. I haven't installed the > patch (I'm troubled by the "unofficial") so I can't say what it > does. > > You mentioned some earlier postings to the news group on this same > problem. I've been unable to find them. What I did was use windsurf and searched on 'no carrier'. What I found was the folks largely recommended changing the DC_MODE value to solve the problem. As I already said, this didn't work for us. > > Any ideas or suggestions you might have would be greatly appreciated. I filed a TSR with windriver with no response yet. I will ask them all these question and try to get to the bottom of it + ask them about this "unofficial" patch and how to get it.  > > Ron Heald > National Radio Astronomy Observatory > Socorro, New Mexico > (505)835-7286 > Tom Antles antles@spk.hp.com Phone: (509) 921-3590 Fax: (509) 921-3991 From owner-vxwexplo-process Wed Sep 8 01:37:59 1999 From: "Tony Mayes" Date: Wed Sep 8 01:38:03 PDT 1999 Subject: Re: (VxWorks) Not suited to High Availability Date sent: Tue, 7 Sep 1999 09:50:10 -0700 (PDT) To: vxworks_users@csg.lbl.gov From: vxwexplo@lbl.gov (the vxWorks Users Group Exploder) Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) > Submitted-by owner-vxwexplo-process Tue Sep 7 09:50:03 1999 > Submitted-by: David Laight > > 'Hot Swapping' is an interesting idea, and, although possible to demonstrate, > may be more difficult to achieve reliably in real life. > > I agree having designed several such systems, in fact in our most recent product we use vxworks in a system which provides hot swap of the CPU stack, via our own backplane chassis. This controls what actually has to be changed and allows us to handle catastrophic internal board failures by removal / replacement of the failing CPU stack. In this case, each CPU has its own I/O, comms, memory etc failure of any component will allow, via our apps software and hardware, the other CPU stack to take over. The hot swap is them merely a matter of power control on the insertion / removal of each CPU stack. May not be relevent to the Telecomms market but works fine in the controls market where we operate. BTW. I am not sure that all the points scoring by Windriver opposition is particularly useful, we chose VxWorks over LynxOs in this app. for other reasons having used LynxOs previously for other apps. Tony Mayes (The opinions expressed within are my own) Email address tony.mayes@epid.eurotherm.co.uk Phone (0044)(0)1903 205277 Fax (0044)(0)1903 524016 Eurotherm Process Instrumentation Div, Southdownview Way, Worthing, Sussex, UK, BN14 8NN From owner-vxwexplo-process Wed Sep 8 04:00:14 1999 From: daemon@csg.lbl.gov Date: Wed Sep 8 04:00:18 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Sep 8 04:00:09 PDT 1999 Subject: Re: T2, mv2303 & 100Mbit lan => dc no carrier Subject: snmpdInit parameters ignored? Subject: Re: Father task should block until all childred tasks have terminated Subject: Re: Why stay with VXWorks? Subject: scatter-gather - please help Subject: How do I change the System Time and Date Subject: Re: VxWorks porting to strongARM Subject: Why 8260 set itself to ZERO wait cycle?? Subject: Re: Why 8260 set itself to ZERO wait cycle?? Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) Subject: Re: Tornado 1.0.1 vs Tornado 2 Subject: #pragma align/pack with gcc/vxworks Subject: Re: multiple 'int i;' Subject: How do use sys596Port() to add multicast addresses ? Subject: Re: tornado operator new Subject: Re: a makefile question Subject: Re: workQPanic? Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) Subject: Re: BSP problem Subject: BSP problem Subject: "fixunssfsi" Problem... Subject: VME backplane interrupts not working properly Subject: Need driver for RTL8139A Subject: Re: BSP problem ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: T2, mv2303 & 100Mbit lan => dc no carrier Date: Mon, 06 Sep 1999 13:20:42 +0200 From: Tomas Larsson Organization: National Defence Research Establishment Message-ID: <37D3A38A.7F4D6D7D@lin.foa.se> References: <7qpp4g$glg$1@hpcvnews.cv.hp.com> Tom Antles wrote: > We just got Tornado 2 running on our mv2303. The problem is, whenever we > run on a 100Mbit network we get the "interrupt: dc0 - no carrier" message. > We NEVER see this message when running on 10Mbit network. > > With lots of LAN traffic we get so many of these messages that the log task > can not keep up. Ex: "logTask: 124 log messages lost." > > I have tried modifying the "#define DC_MODE" macro in config.h according to > earlier postings to this group. I have tried each of the values: 0x8, 0xc, > 0x18, 0x1c (see if_dc.h macros below) on both the 100Mbit network and a > 10Mbit half duplex network. However the result is always the same: 100Mbit > gives "interrupt: dc0 - no carrier", 10Mbit does not. > > Does anyone know what is going on here? > > macro defines from if_dc.h: > > /* Modes for 100-Mb/s Ethernet configuration */ > #define DC_100_MB_FLAG 0x04 /* 100-Mb/s mode */ > #define DC_21140_FLAG 0x08 /* Lance chip is 21140 */ > #define DC_FULLDUPLEX_FLAG 0x10 /* Full Duplex Mode */ > #define DC_SCRAMBLER_FLAG 0x20 /* MII/SYM in scrambler mode */ > #define DC_PCS_FLAG 0x40 /* MII/SYM in symbol mode */ > #define DC_PS_FLAG 0x80 /* Serial Port selected */ > #define DC_ILOOPB_FLAG 0x100 /* Internal Loop Back Mode */ > #define DC_ELOOPB_FLAG 0x200 /* External Loop Back Mode */ > #define DC_HBE_FLAG 0x400 /* Heart Beat Enable Flag */ > > More interesting facts: > > 0) any kind of lan traffic directed to the mv2303 processor results > in 1 or more "interrupt: dc0 - no carrier" messages. Ex: > windShell commands, ftping files. > > 1) when pinging the vxWorks processor with 64 byte packets, I get > exactly 1 "interrupt: dc0 - no carrier" message per ping. > > 2) By executing the ifShow command, (see below) there does not > appear to be ill behavior of the network. > > dc (unit number 0): > Flags: (0x8063) UP BROADCAST MULTICAST ARP RUNNING > Type: ETHERNET_CSMACD > Internet address: 15.2.2.156 > Broadcast address: 15.255.255.255 > Netmask 0xff000000 Subnetmask 0xff000000 > Ethernet address is 08:00:3e:27:04:4d > Metric is 0 > Maximum Transfer Unit size is 1500 > 56754 packets received; 57319 packets sent > 53781 multicast packets received > 45326 multicast packets sent > 0 input errors; 0 output errors > 0 collisions; 0 dropped > > 3) There is NO performance difference when running 100Mbit or > 10Mbit. Each downloads a 15Mbyte file at the same transfer rate: > > 226 Transfer complete > 15396383 bytes received in 31.75 seconds (473.51 Kbytes/s) > > 4) occassionaly, there must be a burst of unrequested lan traffic as > we suddenly see 100's of "interrupt: dc0 - no carrier" messages. > > Tornado vxWorks > > -- > Tom Antles > antles@spk.hp.com > Phone: (509) 921-3590 > Fax: (509) 921-3991 Hi Tom ! We had the same problem when we were booting over the 100 Mbit There is a patch at WRS fixing this problem. Get the file spr22196.tar from WRS support directory And install it Hope this will help Tomas --------------------------- Newsgroups: comp.os.vxworks Subject: snmpdInit parameters ignored? Date: Mon, 6 Sep 1999 12:50:00 +0100 From: "Will Fookes" Organization: NDS UK Message-ID: <7r08k2$ia$1@ndsukns01.ndsuk.com> I have developed an SNMP Agent using WindNet 1.0 with VxWorks 5.3.1 and this has been working happily for some time now. However, we have reached a point in our development where we have a dozen (or so) different tasks running, and want to tidy up the task priorities and stack sizes. I find that the SNMP demon (tSnmpd) is running at a priority of 150 with stack size 28664, and I want to change this. OK, so I see that the priority and stacksize are parameters to snmpdInit() - see snmpdInit.h - but are ZERO when this function is called from usrNetwork.c (actually, in my system this code has been copied elsewhere and is only invoked after the system has been started - but I don't believe that this is relevant) - so I reasoned that maybe the system uses default values in this special case. Not so, it would seem: any priority and stacksize values passed to snmpdInit() appear to be totally ignored. So the question is: how can I establish the required stacksize for tSnmpd? - I can change the priority in snmpIoMain(), is there anything I can do about the stack? Will --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Father task should block until all childred tasks have terminated Date: 6 Sep 1999 12:05:50 GMT From: "Alex Goodyear" Organization: JET Joint Undertaking Message-ID: <01bef860$34954820$5371ef91@9600217-agood.jet.uk> References: <37D39A10.D56DB8E4@computer.org> Hi, I would use a counting semaphore ... /* * Global semaphore used by each terminating child CSP task. */ SEM_ID childFinished; /* * A child task. */ void child (void) { /* * Normal child like activities. */ semGive (childFinished); } #define NUM_CHILDREN 5 /* * Parent task. */ void parent (void) { int childCount = 0; /* * Use a FIFO if you are more interested in speed than the order that tasks will be blocked in. */ if ((childFinished = semCCreate (SEM_Q_FIFO, NUM_CHILDREN)) == NULL) { /* * We're not in the mood for creating children :-( */ return; } /* * Spawn children, the fun part ;-) */ /* * Now wait for all the children to complete their chores. */ do { semTake (childFinished, WAIT_FOREVER); } while (++childCount < NUM_CHILDREN); /* * Tidy up after children (usual parental type stuff). */ semDelete (childFinshed); } Hope this is helpful, Alex Goodyear. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why stay with VXWorks? Date: Mon, 6 Sep 1999 14:43:32 +0200 From: "David" Organization: Tasking Software BV Message-ID: <7r0ctm$p9$1@empoli.tasking.nl> References: <19990728030931.25207.00003136@ng-cr1.aol.com> <7nt6e9$3h4$1@nnrp1.deja.com> <000b8d9b.de388e51@usw-ex0102-013.remarq.com> what is the purpose of this posting? Latch wrote in message <000b8d9b.de388e51@usw-ex0102-013.remarq.com>... >+ >The VxWorks kernel is very solid. >The architecture makes the whole thing very portable. >The Posix interface makes it Unix compatible. >The Tornado-tools are excellent. > >- >WindRiver goes the Microsoft(which everybody hates and > mimics) way, try to cash on applications.They are NOT the > best you can choose and most of the time very frustrating, >Support is lousy, because support-people look on their > position as a temporary as soon as they got more experience > they move somewhere else. >Customers are not equal. Have worked for a big company and >small one I've noticed a huge gap in the relation scheme. >VxWorks is pretty pricy - I will not recommend it for small >budgets. >Source code is not available ( unless you pay big bucks ) > > >* Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * >The fastest and easiest way to search and participate in Usenet - Free! > --------------------------- Newsgroups: comp.os.vxworks Subject: scatter-gather - please help Date: Mon, 6 Sep 1999 15:22:42 +0100 From: "Rui Prior" Message-ID: <37d3ce29@readers.ip.pt> Hi. As I couldn't find previous posts on the subject here goes a question: I want do do scatter-gahter reception in an END driver. How do I append a clBlk-cluster to an mBlk that already contains data (ie, is already linked to at least one clBlk-cluster construct)? The manual isn't clear as to if I can use netMblkClJoin()... Rui Prior --------------------------- Newsgroups: comp.os.vxworks Subject: How do I change the System Time and Date Date: Mon, 06 Sep 1999 17:39:52 +0200 From: Carlos David Organization: Ericsson Eurolab Germany Message-ID: <37D3E048.866C1DBA@eed.ericsson.se> Hi, Im working with VxWorks 5.3.1 doing some Year 2000 testing. Our system consists of Solaris works which collect and store data and VxWorks boards delivering the data in realtime. My problem is that in order to test the year 2000 compliance of the system I have to change the date and time in both the Solaris boards and the VxWorks. For Solaris I know how but unfortunately, for VxWorks there is not such a command as the "date" command for Solaris. Can anyone help me? Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks porting to strongARM Date: Mon, 6 Sep 1999 09:40:20 -0700 From: "John Finley" Organization: Kivala Systems, Inc. Message-ID: <7r0qlf$lem@dfw-ixnews5.ix.netcom.com> References: <199909060929.KAA04626@dsl-2.tadpole.co.uk> I stand corrected. Very corrected, actually. "Seems to work", which is the extent of my few-weeks experience with the board, and "well supported" are indeed very different. Extensions to your list at any time would be welcome and appreciated. Thanks David, John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Tornado Apps with Visual Basic, LabVIEW - -------------------------------------------------- David Laight wrote in message news:199909060929.KAA04626@dsl-2.tadpole.co.uk... > John worte: > > Yup, Brutus appears well supported; the BSP, UGL, > > and PJWorks all work as shipped with brutus and his > > LCD/touchscreen. Incidentally, the CPU define is > > the same for SA110 and 1100. (I'm gettin' there, > > Doug...) > > I can't agree with 'well supported': > 1) The FP routines are very very very slow. > 2) The interrupt handler scheme isn't designed to allow drivers to attach to > all the cascaded (second level) interrupts. > The interrupt scheme needs to allow for external devices (eg SA1101 and > UCBV1200) that provide multiple interript sources - including simple GPIO > pin interrupts. > I'm up to 144 possible interrupt sources (including all gpio pins.) > 3) The timer code drifts badly. Unfortunately the 60Hz tick has to be exactly > 60Hz to keep the vxworks RTC functions accurate. > (You can't actually do 60Hz! 100Hz gives much better POSIX timers anyway.) > 4) MMU setup is inappropriate for a simple RTOS. The SA1101 has its registers > on 0x400 byte boundaries - you need to use 1Mb segments (not 4k pages) to > map these efficiently. In fact 1Mb pages are fine for a simple SA1100 > system. (You have to use the mmu to get DRAM at address 0.) > The list goes on... --------------------------- Newsgroups: comp.os.vxworks Subject: Why 8260 set itself to ZERO wait cycle?? Date: Mon, 06 Sep 1999 16:56:25 GMT From: Xiang Gao Organization: Bell Solutions Message-ID: <37D3F244.40184F85@netpointcorp.com> Good Day, After PORESET and HRESET configuration, I assume 8260 should set itself a very long wait state to start fetching instruction from address 0x100. ... But what I found from my board is a ZERO wait state. Can anyone tell me what's going wrong? I guess the HRESET configuration does nothing to this. This should be a default setting in 8260. Actually when my 8260 reading the HRESET config. word, the read cycles are very long. Any recommend will be very helpful. Thanks in advanced, Xiang --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Why 8260 set itself to ZERO wait cycle?? Date: Mon, 06 Sep 1999 23:14:15 GMT From: "Matthew P. Downs" Organization: @Home Network Message-ID: References: <37D3F244.40184F85@netpointcorp.com> First check with Motorola, they have a ton of open issues with the 8260. We have found a lot of stange interations. However, we wrote our own BSP, using it before WindRiver had support. Xiang Gao wrote in message <37D3F244.40184F85@netpointcorp.com>... >Good Day, > >After PORESET and HRESET configuration, I assume 8260 should set itself >a very long >wait state to start fetching instruction from address 0x100. ... But >what I found from my >board is a ZERO wait state. Can anyone tell me what's going wrong? I >guess the HRESET configuration does nothing to this. This should be a >default setting in 8260. >Actually when my 8260 reading the HRESET config. word, the read cycles >are very long. > >Any recommend will be very helpful. > >Thanks in advanced, > >Xiang > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) Date: Tue, 07 Sep 1999 14:43:25 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37D55CCD.ECC61F14@lucent.com> References: <199909071649.RAA10975@dsl-2.tadpole.co.uk> > The suggestion that a MSW-2000 system is doing hot-swap of cpu cards seems > amazing! A can't imaging a cpu card dying without trashing at least some of the > memory associated with its current activity. I am confused by the fact that it says MSW-2000 and LynxOS in the same posting. Particularly since the computer in question uses PPC-750 CPUs, which are not supported in W2000 that I am aware of. I have to assume that there was a hierarchy here with some iBox (Ouch, Steve Jobs will squawk at that!) running W2K (why 2 K?) driving a few PPC-750 slots running LynxOS. Face it, no box running MSW-2000 is going to achieve five nines. Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 1.0.1 vs Tornado 2 Date: Tue, 07 Sep 1999 11:12:46 -0700 From: Bryan Wilcutt Organization: AG Communication Systems Message-ID: <37D5559E.B82B7D88@agcs.com> References: <7qao2i$r0q$1@spinner.corpeast.baynetworks.com> Reply-To: wilcuttb@agcs.com We tried this on a subproject for Iridium [Motorola]. It can be done, however converting your VOBs may prove to be difficult with the differences in platforms and the size of the VOBs. I'd recommend keeping your VOBs on Solaris unless you REALLY have to move over. Also, if your Solaris machines are faster than your NT machines would be another good reason not to switch-- ClearCase has to be 1) the most powerful CM tool around and 2) the world's absolute slowest application on the planet! Remember, ClearCase was specifically designed to NFS over a Unix platform so it provides much more stability in this environment. G'luck Bw Ofer Goren wrote: > Hi. > We are currently working with Tornado 1.0.1 for solaris, with clearcase > control version. We want to upgrade to Tornado 2, and we want to know should > we upgrade to NT, or stay with Solaris? > If anyone works with Tornado 2, AND clearcase version control, we want to > know what is the best step to do, so please contribute to the disscusion. > > Thanks, > > Ofer goren > NortelNetworks Israel --------------------------- Newsgroups: comp.os.vxworks Subject: #pragma align/pack with gcc/vxworks Date: Tue, 07 Sep 1999 11:16:12 -0700 From: Bryan Wilcutt Organization: AG Communication Systems Message-ID: <37D5566C.9DEFEEDE@agcs.com> Reply-To: wilcuttb@agcs.com Pragma align and pack seem to have 'some' effect under C (gnu) but NONE under C++ (gnu). These are the compilers that ship with VxWorks. Some of the old online gnu manuals state pragma is no longer supported but this contradicts what is mentioned in the WindRiver manuals. Anyone have any ideas? Bw --------------------------- Newsgroups: comp.os.vxworks Subject: Re: multiple 'int i;' Date: Tue, 07 Sep 1999 17:09:08 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37D57EF4.EE48B608@lucent.com> References: <199909061003.LAA04692@dsl-2.tadpole.co.uk> <37D55B66.11868807@nowhere.com> Charlie Grames wrote: > > I'm afraid this just added to the confusion. If I recall correctly, the real > question at hand was: How are declarations like "int i" and "static int j" > handled when they exist in two different modules, both of which are dynamically > linked (loaded)? > > For VxWorks, the answer depends entirely on whether or not a symbol "exists" > (i.e., has storage defined) at link time. Let's start with the following two > files, which differ only in the name of the routine: < great big snip> Charlie, You have it right, for vxWorks. I also wrote a similar set of files to demonstrate this, even spawned them as tasks to show that the variables were persistent across invocations. Static variables should be separate across modules. By definition, a static outside of a function definition has file scope, not global scope. Static variables inside a function definition are persistent, like a global, but have function scope. If you have a global 'j' in file A.c and a static 'j' in file B.c and a static 'j' in function L in file B.c, then: j is visible everywhere except... file B, where the static 'j' is visible to all functions in B.c except.... function L, which has its own static 'j'. Just as locals in a function hide globals, statics hide globals at the file level. And of course, statics in a function have local scope and as such behave with the same scoping rules. Doug --------------------------- Newsgroups: comp.os.vxworks Subject: How do use sys596Port() to add multicast addresses ? Date: Tue, 07 Sep 1999 17:17:54 -0400 From: "Joseph G. Jaeger" Organization: Open Networks Engineering Message-ID: <37D58102.244E0C00@one.com> I need to send a command to the 82596 Ethernet controller on an MVME167 board. From the documentation on if_ei library, I see that the following two functions, below, are supported by the driver. void sys596Port (int unit, int cmd, UINT32 addr); void sys596ChanAtn (int unit); I'd like to use this interface to specify multicast addresses to the driver. I'm passing SCB_C_CUSTART as cmd and the address of an initialized CFD structure as addr (the unit value is pulled from the ifnet struct corresponding to the ei driver). The result I keep getting is a message from the ei driver stating that the "command field is frozen". Is this the correct use of these routines? If not, does anyone have a clue as to how to do this? NOTE: I'm not using an END driver and the solution must work with 5.2, 5.3.1 and 5.4). Joe - -- ====================================================================== OPEN NETWORKS ENGINEERING, INC. Joseph G. Jaeger 2725 South Industrial Highway, Suite 100 Voice: (734) 975-7328 Ann Arbor, MI 48104 FAX: (734) 975-6940 Web: http://www.one.com Email: mailto:jgj@one.com ====================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tornado operator new Date: Tue, 07 Sep 1999 14:24:36 -0700 From: Eric Flynn Organization: Hewlett Packard Cupertino Site Message-ID: <37D58294.CBB40FA2@wwg2.sid.hp.com> References: Reply-To: eflynn@wwg2.sid.hp.com Kandru, The first statement creates an array of 10 integers. The second statement creates one integer and initializes it to the value 10. Eric Kandru, Raja wrote: > Tornado Users, > > I am relatively new to Tornado and VxWorks. > I am wondering what's the difference between the two function (new) calls: > > int *i = new int[10]; > int *i = new int(10); > > thanks in advance, > Raja Kandru > Lucent Technologies > kandru@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: a makefile question Date: Tue, 7 Sep 1999 12:49:06 +0100 From: "Mark Setchell (Delta Plus Ltd)" Organization: UUNET WorldCom server (post doesn't reflect views of UUNET WorldCom Message-ID: <7r2u6h$4me$1@lure.pipex.net> References: <37D4EBC0.EB87F7C9@126.com> It evaluates to the current target - or in simple terms, normally the bit before the colon on the line above. - -- Mark Setchell Delta Plus Ltd, 24 St Albans Rd, Cambridge, England CB4 2HG Tel: +44 1223 462801 Fax: +44 1223 462802 URL: http://www.dpl.dial.pipex.com Email: dpl@dial.pipex.com zhaobo wrote in message <37D4EBC0.EB87F7C9@126.com>... >May anyone tell the meaning of symbol "$@" in gcc makefile? thanks. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: workQPanic? Date: Tue, 07 Sep 1999 14:02:00 +0100 From: David Machin Organization: GEC-Marconi Message-ID: <37D50CC8.C3B7D1C7@REMOVE-THIS.gecm.com> References: Hi Dennis The work queue overflow is caused by events being placed on the work-queue at a higher rate than the kernel can process them, and thus the work queue fills and VxWorks stops. In your case I would think the most likely cause is a failed device constantly asserting an interrupt - I'd guess that the point in initialisation where the problem happens is the point where this interrupt gets unmasked. Hope this helps. Regards Dave Machin Dennis M. Reichhold wrote: > > Hi > > One of my CPU's (a 162) was running just fine yesterday morning. > Then, in the afternoon, it crashed (I assume). Now whenever it > reboots, it goes through most of the login script but stops (always at > the same spot, unless I've cycled the power) with the error message: > > workQPanic: Kernel work queue overflow. > > and then starts the reboot process. When I cycle the power, it reads > in an auto_settings.sav file, and somewhere during the execution of > iocInit, I get the same error message, and the processor reboots. The > host computer is running SunOS 4.1.4. Anyone know what's going on? > > -Dennis --------------------------- Newsgroups: comp.os.vxworks Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) Date: Tue, 7 Sep 1999 19:18:53 -0400 From: "Jeff Szczepanski" Organization: Time Warner Road Runner - Rochester NY Message-ID: References: <7r3kn2$3fe1@overload.lbl.gov> > it is the fact that LynxOS does support Hot Swap for Compact PCI and that > VxWorks (yet) doesn't. I wouldn't really say that this is really 'a kernel' feature, but more an issue of having a BSP supporting it. As part of any of the VxWorks architecture ports, there is really no reason you couldn't add this functionality relatively easily since it's the drivers for the board(s) and PCI bridge controller that needs to handle it. In other words, the PCI configuration structure is fairly simple (elegant) and as a result it would be really straight forward to code up the support for this on your own. Admittedly it would be nice to have on hand, but not one of the critical parameters to consider in an OS selection.....The real issue for the OS selection is the process/task model, performance, interrupt latency, determinism, VM support, POSIX, etc. etc. compared to the needs of your application. Regards, Jeff wrote in message news:7r3kn2$3fe1@overload.lbl.gov... > Hello, > > Leaving subject(s) of relevance of RTOS features, such as: > Virtual Memory, full POSIX compliance (including threads support) , > "per task memory address protection", > and "OS memory address space separation from the Task/User memory address > space" > vs the "ease" of implementing High Availability/Hot Swap > ASIDE ;-) , > it is the fact that LynxOS does support Hot Swap for Compact PCI and that > VxWorks (yet) doesn't. > -------------------------------------------------------------------------- - - > ----------------------------------------------- > LynxExpress Newsletter - Volume 15 (http://www.lynx.com/): > Motorola Computer Group stole the show at CT Expo with a dramatic > demonstration > of the CPX8200 computer system. MCG's CPX8000 delivers "five-nines" > capability > (99.999%; less than five minutes and 15 seconds of downtime per year), and > provides the computer power for applications used in "carrier-grade" > telecom > systems. The demonstration included a Microsoft Windows 2000-based full > hot-swap > system is the first real-time system that does hot swapping and failover of > CompactPCI processor cards, done in combination with MCG's MCP750 > intelligent > I/O processors and LynxOS real-time operating system from Lynx Real-Time > Systems. The CPX8000 is based on an open-architecture CompactPCI platform, > which enables any active module to be exchanged for repair or upgrade while > the > system continues to operate. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP problem Date: Tue, 7 Sep 1999 19:24:33 -0400 From: "Jeff Szczepanski" Organization: Time Warner Road Runner - Rochester NY Message-ID: <45hB3.5687$w6.208838@typhoon.nyroc.rr.com> References: <7r34l0$g4p$1@nnrp1.deja.com> Just guessing..... One of the key things happening when you call kernelInit() from a BSP perspective is that interrupts get enabled. If the vectors are not installed correctly, setup right, or a pending interrupt is not cleared properly as they occur (ie: Hardware or driver issues)....then a hang or a 'walk into the weeds' is likely. Regards, Jeff - -- =========================== Jeffrey R. Szczepanski InSciTek Microsystems, Inc. 635 CrossKeys Office Park Fairport, NY 14450 wrote in message news:7r34l0$g4p$1@nnrp1.deja.com... > Hello, > I have a really serious problem to get a running vxWorks.res_rom_nosym > on our custom board with MPC850. > > The problem is: > usrInit() calls kernelInit() and usrRoot() is not called > I found out that the function windSpawn() jumps to a wrong address. > > What is my fault??? > > any help will be appreciate > > Oliver > > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: BSP problem Date: Tue, 07 Sep 1999 13:40:49 GMT From: oliver_fuchs@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7r34l0$g4p$1@nnrp1.deja.com> Hello, I have a really serious problem to get a running vxWorks.res_rom_nosym on our custom board with MPC850. The problem is: usrInit() calls kernelInit() and usrRoot() is not called I found out that the function windSpawn() jumps to a wrong address. What is my fault??? any help will be appreciate Oliver Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: "fixunssfsi" Problem... Date: Wed, 8 Sep 1999 13:12:00 +0900 From: "Young W. Cho" Organization: Electronics and Telecommunications Research Institute (ETRI) Message-ID: <7r4ntj$boq$1@news.etri.re.kr> Hi all, Is there anybody who knows what cause to be the following warning message ? When I compiled the original files under X86 environment, there was no warning. I am trying to port my application into strongARM board. Please help me out. - -> ld < swm Undefined symbols: ___fixunssfsi Warning: object module may not be usable because of undefined symbols. value = 4786672 = 0x4909f0 - -> Young W. Cho (ywcho@netscape.net) --------------------------- Newsgroups: comp.os.vxworks Subject: VME backplane interrupts not working properly Date: Tue, 7 Sep 1999 08:19:36 -0700 From: "Joseph R Hansen" Organization: NAWCWPNS Message-ID: <7r3ah9$2ok1@news.nawcwpns.navy.mil> Fellow developers, I am in the midst of developing a system which requires a Motorola MVME-162, running VxWorks 5.3.1, to service an interrupt from a Datum IRIG board. The Datum retrieves the IRIG-B time, saves it to registers, then generates a level-5 VME interrupt, which is serviced by the Motorola. The Motorola board then reads the IRIG-B time over the VME backplane. The interrupt occurs at a 100-Hz rate. I have written and tested the software in a five-slot BusTronics VME chassis. However, the delivery system uses a 21-slot Dawn VME chassis, and this is not working. For test purposes, I have removed all other boards from the chassis. I have placed jumpers in all slots except the first two. The Motorola is in slot 1, and the Datum is in slot 2. Finally, the Motorola is configured to be the bus master. When I spawn my main task, it initializes the Datum board, and installs the ISR. As soon as the proper interrupt level is enabled, VxWorks gives me a nonsensical "zero divide" error at a nonsensical address. I even changed interrupt levels to see if there were conflicts with anything else, but had the same results. I swapped in an identical Dawn VME chassis to make sure that the first was not fried somehow, but there was no change. I connected the IACQ jumper on the back of slot 2 for testing purposes. The system then did work to a degree, but the Motorola was irregular at seeing and handling the interrupts. However, it did not error out. Finally, I verified that the terminating resistors were on slots 1 and 21. I don't remember the exact values, but they were there. I think two of them were orange-orange-brown, and the other two were yellow-blue-brown, but I could be mistaken. At this point, I'm stumped. I think that this is a VME problem, rather than VxWorks, since I had it working in one chassis. Unfortunately, I have to deliver the system in the "non-working" chassis Any and all suggestions greatly appreciated, Joseph Hansen Dept. of Navy NAWCWD Code 471500D China Lake, CA 93555 HansenJR@navair.navy.mil (760) 939-6999 --------------------------- Newsgroups: comp.os.vxworks Subject: Need driver for RTL8139A Date: Wed, 08 Sep 1999 15:55:28 +0800 From: Richard Tseng Organization: Cyber Express Communications (using Airnews.net!) Message-ID: <2F48453F251C4B48.DADC22E28FB2FA45.48E41F2F685B4AE2@lp.airnews.net> Hello, I'm looking for a VxWorks x86 driver for the Realtek RTL8139A ethernet chip. Would anyone know where to download and/or purchase one ? Thanks, Richard Tseng We Software Ltd rtseng@wesoft.com.hk --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP problem Date: Wed, 08 Sep 1999 09:20:22 GMT From: oliver_fuchs@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7r59of$3o3$1@nnrp1.deja.com> References: <7r34l0$g4p$1@nnrp1.deja.com> <45hB3.5687$w6.208838@typhoon.nyroc.rr.com> Thanks for the hint. This was my 1st. idea too. I checked it already. I put some LED outs into the ppc860Intr.c. So I'm sure that ppc860IntrInit is called during pre kernel init and ppc860IntrDeMux or ppc860CpmIntrDeMux are not called. I have traced the wrong jump by means of a logic analyzer. The result is: the function windSpawn fetches a wrong address from activeQHead windSpawn: stwu r1, 0xfff8(r1) + 0x004 mfspr r0, LR + 0x008 stw r0, 0xc(r1) + 0x00c lis r9, Hi(activeQHead) + 0x010 addi r9, r9, Lo(activeQHead) + 0x014 addi r4, r3, 0x20 (32) + 0x018 or r3, r9, r9 + 0x01c lwz r9, 0xc(r3) + 0x020 lwz r0, 0x10(r9) + 0x024 mtspr LR, r0 + 0x028 li r5, 0x0 (0) + 0x02c blrl // last valid address In article <45hB3.5687$w6.208838@typhoon.nyroc.rr.com>, "Jeff Szczepanski" wrote: > Just guessing..... > > One of the key things happening when you call kernelInit() from a BSP > perspective is that interrupts get enabled. If the vectors are not installed > correctly, setup right, or a pending interrupt is not cleared properly as > they occur (ie: Hardware or driver issues)....then a hang or a 'walk into > the weeds' is likely. > > Regards, > Jeff > > -- > =========================== > Jeffrey R. Szczepanski > InSciTek Microsystems, Inc. > 635 CrossKeys Office Park > Fairport, NY 14450 > > > wrote in message > news:7r34l0$g4p$1@nnrp1.deja.com... > > Hello, > > I have a really serious problem to get a running vxWorks.res_rom_nosym > > on our custom board with MPC850. > > > > The problem is: > > usrInit() calls kernelInit() and usrRoot() is not called > > I found out that the function windSpawn() jumps to a wrong address. > > > > What is my fault??? > > > > any help will be appreciate > > > > Oliver > > > > > > Sent via Deja.com http://www.deja.com/ > > Share what you know. Learn what you don't. > > Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Wed Sep 8 08:02:24 1999 From: David Laight Date: Wed Sep 8 08:02:28 PDT 1999 Subject: Re: "fixunssfsi" problem Young, ___fixunssfsi is the routine to convert a float to an unsigned int. I don't know why it (and fixunsdfsi) are absent from the vxworks image - maybe there are just no callers... The x86 will (probably) be using hardware FP so won't have the soft float routines - and in particular definitly won't have the horrid (very slow) C routines that WRS ship with vxWorks for ARM. I've attached the assembler for these two routines (compile the file twice with and without -DFLOAT) from my assembler FP routines. This particular routine is untested - because there are no callers to it, but looks fine. David > Subject: "fixunssfsi" Problem... > Date: Wed, 8 Sep 1999 13:12:00 +0900 > From: "Young W. Cho" > Organization: Electronics and Telecommunications Research Institute (ETRI) > Message-ID: <7r4ntj$boq$1@news.etri.re.kr> > > Hi all, > > Is there anybody who knows what cause to be the following warning message ? > When I compiled the original files under X86 environment, there was no > warning. > I am trying to port my application into strongARM board. > Please help me out. > > > - -> ld < swm > Undefined symbols: > ___fixunssfsi #ifdef FLOAT #define ___fixunsdfsi ___fixunssfsi #define MNT_SZ 23 /* mantissa bits in first word */ #define r2 r1 #else #define MNT_SZ 20 #endif #define SGN_BIT 0x80000000 /* sign bit */ /**************************************************************************** double to unsigned int r0 = (unsigned int)( r0:r1 ) */ .global ___fixunsdfsi ___fixunsdfsi: movs r12,r0,asr #MNT_SZ /* get sign and exponent */ movmi r12,#0 /* convert -ve to zero */ /* extract top 32 bits of mantissa */ mov r0,r0,lsl #31 - MNT_SZ #ifndef FLOAT mov r1,r1,lsr #MNT_SZ + 1 orr r0,r0,r1 #endif orr r0,r0,#SGN_BIT /* add hidden bit */ #ifdef FLOAT rsbs r12,r12,#0x9e /* unbiassed and negated */ #else sub r12,r12,#0x400 rsbs r12,r12,#0x1e /* unbiassed and negated */ #endif bmi 10$ mov r0,r0,lsr r12 mov pc,lr 10$: / overflow */ mvn r0,#0 mov pc,lr ---------------------------------------------------------------- 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 Sep 8 08:08:40 1999 From: Alexander_R_Povolotsky@res.raytheon.com Date: Wed Sep 8 08:08:44 PDT 1999 Subject: Could anyone share his/her experience of using "Look!" on Tornado Hi , Could anyone share his/her experience of using "Look!" on Tornado 2 ? For those who doesn't know - "Look!" is the dynamic C++ browser supplied by WRS (as an optional product) on Tornado 2 platform. I am interested to know, in particular, how well it is integrated with WRS's GNU GDB debugger, i.e. - is it possible to insert break points out of "Look!" (and get stopped on those inside the "Look!'); is it also possible to do "step" debugging inside the "Look" ? Also how useful the "Look!" is in finding dynamic memory allocation/deallocation problems ? Any other useful features ? Best Regards, From owner-vxwexplo-process Wed Sep 8 11:32:59 1999 From: "Wenholz, Bruce" Date: Wed Sep 8 11:33:03 PDT 1999 Subject: chkdsk on VxWorks Hi, We are encountering some fat corruption and directory corruption on our target devices while running vxWorks 5.3.1 on a 486. I think the problems are specifically concerned with program failures causing the system to just stop and then the dosFs file system is not written completely. Does any one know about a chkdsk source that I could adapt to run on our target platform under vxWorks. ===========8<--------------------------------->8========== Bruce Wenholz St. Jude Medical, CRMD 818-493-4207 15900 Valley View Court Email: bwenholz@pacesetter.com Sylmar, Ca 91342, USA From owner-vxwexplo-process Wed Sep 8 19:10:38 1999 From: Tom Antles Date: Wed Sep 8 19:10:42 PDT 1999 Subject: Re[4]: dc no carrier vxworks tornado David Abbott wrote: > Hi Tom, > I have been following your "no carrier" message problem, and I have > what may be a moot question. I use MVME2306, MVME2600, and MVME2700s. All > exhibited the same behavior when connected to 100MB Autosensing switches. > The fix for me (as you noted before) was to force the dc vxWorks driver > into full duplex mode. > YOu indicated that changing the DC_MODE define in config.h did nothing > for you though. However, i noticed that for Tornado II, if you are using > the END driver, you must actually modify the USER MODE bits in the file > "configNet.h" in the BSP directory. The DC_MODE definition in config.h > only applies to the non-END driver. > > In configNet.h there is a define for DEC_LOAD_STRING: > > ------------------- > /* :::::::: */ > > #ifdef EXTENDED_VME > # define DEC_LOAD_STRING "0xfe020000:0x0:0x12:0x12:-1:-1:-1:0:0x80800000" > #else > # define DEC_LOAD_STRING "0x81020000:0x80000000:0x12:0x12:-1:-1:-1:0:0x80800000" > #endif > ------------------ > > I just changed the final entry to 0x80c00000 > > > If you have already tried this then I am stumped as well. Good Luck > getting anything out of Wind River. > > David > > --------------------------------------------------------- > David Abbott Jefferson Lab > Data Acquisition Group MS 12H > EMAIL: abbottd@jlab.org 12000 Jefferson Ave. > Tel: (757) 269-7190 Newport News, VA 23606 > FAX: (757) 269-5800 > --------------------------------------------------------- > David, Thanks for this new info about the configNet.h file. It did indeed solve my problem on 100Mbit network. Except... when I force to full duplex mode, then I am unable to run on a half duplex network. This is a problem because the software we distribute can be on either types of networks. So I guess I will work with tech support some more to find a solution that works on all networks , 100Mbit, 10Mbit, full and half duplex. Thanks for the good info. Tom , Tom Antles antles@spk.hp.com Phone: (509) 921-3590 Fax: (509) 921-3991 From owner-vxwexplo-process Wed Sep 8 21:03:57 1999 From: RamaKrishna Bikumalla Date: Wed Sep 8 21:04:01 PDT 1999 Subject: VxWorks This is a multi-part message in MIME format. --------------322C3B9A77BC31282F1C7B2B Content-Type: multipart/alternative; boundary="------------40495FADB4953299854E89DB" --------------40495FADB4953299854E89DB Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Previously I worked in a Pure Embedded Systems(no kernel).It is a single thread.Now I have to work in VxWorks.How I have to progress.Some simple ways by which I can perform well.I have little bit knowledge of pSOS.Pl Guide me and give a direction. Thanks In advance. -- ************************************************ RAMAKRISHNA BIKUMALLA ENGINEER SOFTWARE WIPRO LIMITED - TECHNOLOGY SOLUTIONS 26,CHAMUNDI COMPLEX HOSUR MAIN ROAD BOMMANAHALLI BANGALORE - 560 068 PH : 91-80-5722296 EXT 5327 FAX : 91-80-5722696 EMAIL: krishb@wipinfo.soft.net ************************************************ --------------40495FADB4953299854E89DB Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Previously I worked in a Pure Embedded Systems(no kernel).It is a single thread.Now I have to work in VxWorks.How I have to progress.Some simple ways by which I can perform well.I have little bit knowledge of pSOS.Pl Guide me and give a direction.
Thanks In advance.
--
************************************************
  RAMAKRISHNA BIKUMALLA
  ENGINEER SOFTWARE
  WIPRO LIMITED - TECHNOLOGY SOLUTIONS
  26,CHAMUNDI COMPLEX
  HOSUR MAIN ROAD
  BOMMANAHALLI
  BANGALORE - 560 068

  PH   : 91-80-5722296 EXT 5327
  FAX  : 91-80-5722696
  EMAIL: krishb@wipinfo.soft.net
************************************************
  --------------40495FADB4953299854E89DB-- --------------322C3B9A77BC31282F1C7B2B Content-Type: text/x-vcard; charset=us-ascii; name="krishb.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for RamaKrishna Bikumalla Content-Disposition: attachment; filename="krishb.vcf" begin:vcard n:Bikumalla;RamaKrishna tel;work:5722296 x5327 x-mozilla-html:FALSE url:www.wipro.com org:Wipro Technology Solutions;E.S.B.U/MM version:2.1 email;internet:krishb@wipinfo.soft.net title:Wipro Infotech adr;quoted-printable:;;26,Hosur Road,=0D=0ABommanahalli.;Bangalore.;Karnataka.;560 068.;India fn:RamaKrishna Bikumalla end:vcard --------------322C3B9A77BC31282F1C7B2B-- From owner-vxwexplo-process Thu Sep 9 00:12:15 1999 From: HOLONTECH Umesh Mallugari Date: Thu Sep 9 00:12:19 PDT 1999 Subject: Re: VxWorks Ramakrishna, If you have knowledge of Unix, that will help. VxWorks is close to Unix'es. Go thro the 'Basic OS' literature that comes with the VxWorks documentation. You can get good idea from that. Then, try using the OS features. Then, you are online. Regards, -Umesh. > Previously I worked in a Pure Embedded Systems(no kernel).It is a single > thread.Now I have to work in VxWorks.How I have to progress.Some simple > ways by which I can perform well.I have little bit knowledge of pSOS.Pl > Guide me and give a direction. > Thanks In advance. From owner-vxwexplo-process Thu Sep 9 04:00:13 1999 From: daemon@csg.lbl.gov Date: Thu Sep 9 04:00:17 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Sep 9 04:00:08 PDT 1999 Subject: ONly a Test Subject: Re: ONly a Test Subject: Searching a used and working MBX board Subject: RE: Tornado 2 Make Dependencies Subject: Re: (VxWorks) Not suited to High Availability Subject: DHCPV4 SNMP Daemon Dependancy Subject: IPv6 Subject: Writing BSP - excVecInit & classLibInit go AWOL Subject: Re: #pragma align/pack with gcc/vxworks Subject: Re: Re[2]: dc no carrier Subject: INCLUDE constant problem Subject: Re: Writing BSP - excVecInit & classLibInit go AWOL Subject: Re: INCLUDE constant problem Subject: Re: BSP problem Subject: Re: BSP problem Subject: Programming and Running Flash Subject: Shared Memory Problem Subject: Shell Prompt Subject: Re: BSP problem Subject: PCI support in VxWorks (non-x86) Subject: Re: Writing BSP - excVecInit & classLibInit go AWOL Subject: Re: Shell Prompt Subject: Re: Shell Prompt Subject: errno 0xc0002 from puts() ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: ONly a Test Date: Wed, 08 Sep 1999 15:00:24 +0200 From: Ralf Stiawa Organization: SPEECHCOM Message-ID: <37D65DE8.90E958F1@dica.de> Hi I am new. - -- Ralf Stiawa --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ONly a Test Date: Wed, 08 Sep 1999 15:01:23 +0200 From: Ralf Stiawa Organization: SPEECHCOM Message-ID: <37D65E23.FAA3B579@dica.de> References: <37D65DE8.90E958F1@dica.de> Thanks Ralf Stiawa schrieb: > Hi I am new. > > -- > Ralf Stiawa --------------------------- Newsgroups: comp.os.vxworks Subject: Searching a used and working MBX board Date: Wed, 08 Sep 1999 15:06:18 +0200 From: Ralf Stiawa Organization: SPEECHCOM Message-ID: <37D65F4A.B5557A37@dica.de> Hi URGENT: we are looking for an used but working Motorola MBX860 (MPC860 processor). If anyone can offer one or two please contact us with type and prices for the board and shipment via email. Thanks a lot. - -- From: http://www.dica.de /--------------------------------------------/ // Address: // // Speechcom Infomationstechnik GmbH // // Rotherstrasse 18 // // 10245 Berlin // //------------------------------------------// --------------------------- Newsgroups: comp.os.vxworks Subject: RE: Tornado 2 Make Dependencies Date: Wed, 8 Sep 1999 14:25:45 +0200 From: "Borkhuis, Johan" Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: > ---------- > From: Marjorie Krueger[SMTP:mkrueger@vixel.com] > Sent: maandag 16 augustus 1999 22:15 > Subject: Re: Tornado 2 Make Dependencies > > I've had a similar problem and tried the same things you tried, the only > thing that fixed it > for me was deleting the .c file from the project and re-adding it. > > "Woolverton, Eric D." wrote: > > > > Has anyone seen the following problem or know how to get around it? > > > > I have the following directories > > > > release > > release/src > > release/inc > > work > > work/changes > > > > > > > Eric Woolverton > > (410) 765-0834 > > Eric_D_Woolverton@md.northgrum.com > > Northrop Grumman > I found your message in the queue, after I returned, I don't know if you already solved it. I came accross the same problem with the dependencies. I found a way to solve this problem, look at my VxWorks page for the solution. The URL is http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html. I hope this helps. Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: (VxWorks) Not suited to High Availability Date: Wed, 08 Sep 1999 09:33:03 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37D6658F.3BB0F099@lucent.com> References: <199909080829.JAA01993@castor.epid.eurotherm.co.uk> Tony Mayes wrote: > BTW. I am not sure that all the points scoring by Windriver > opposition is particularly useful, we chose VxWorks over LynxOs > in this app. for other reasons having used LynxOs previously for > other apps. Good point. I have used a half dozen different RTOS over the last 15 years, they each had their place. It reminds me of the old saying, "To a man who has only a hammer, everything looks like a nail." Doug --------------------------- Newsgroups: comp.os.vxworks Subject: DHCPV4 SNMP Daemon Dependancy Date: Wed, 08 Sep 1999 15:42:16 GMT From: terrya5495@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7r604l$k4u$1@nnrp1.deja.com> I need to add the DHCPV4 client component to the downloadable VxWorks.exe . The tornado2 tool automatically creates a dependancy on the SNMP Daemon component when I include the DHCPV4 client. We do not have the SNMP package because it does not support snmp v3 (security) so building the VxWorks.exe results in an error. Is there any way around this? cheers, Terry Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: IPv6 Date: Wed, 08 Sep 1999 18:11:36 +0200 From: Gerhard Juen Message-ID: <37D68AB8.BAC5EB6F@bocholt.fh-gelsenkirchen.de> Dear vxWorks-community does anybody have experience or some sources with IPv6-drivers for VxWorks best regards gerhard juen - -- - ----------------------------------------------------------------------- Prof. Dr. Gerhard Juen FH Gelsenkirchen, Abt. Bocholt Postfach 1441 D-46364 Bocholt email: juen@bocholt.fh-gelsenkirchen.de web: http://www-et.bocholt.fh-gelsenkirchen.de/~juen Fon: 02871-2155 830 Fax: 02871-2155 800 --------------------------- Newsgroups: comp.os.vxworks Subject: Writing BSP - excVecInit & classLibInit go AWOL Date: Wed, 08 Sep 1999 17:45:30 +0000 From: Luke Diamand Organization: Virata Message-ID: <37D6A0BA.2A8A80D1@virata.com> Hi! I'm writing a BSP for an ARM-based target. I find that excVecInit and classLibInit, when called, seem never to come back. Does anyone know where I can find some documentation on what needs to be done to persuade these functions to do the Right Thing (TM). Thanks in advance! Luke Diamand - -- Virata http://www.virata.com Cambridge Tel: +44 1223 566919 Fax: +44 1223 566915 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #pragma align/pack with gcc/vxworks Date: 08 Sep 1999 10:46:08 PDT From: Yuri Karlsbrun Organization: Chameleon Systems, Inc. Message-ID: <37D6A20D.5CB62215@chameleonsystems.com> References: <37D5566C.9DEFEEDE@agcs.com> This is a multi-part message in MIME format. - --------------8B81D6369324029C82B17B8A Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Brian, You, actually mentioned two problems. 1. Alignment for global variables. In this case gcc directive works, because alignment is done during link time. 2. Alignment for automatic variables. In this case directive does not work (It's linnker directive). Automatic variables are allocated in the stack during run time. If you need to allocate aligned variables during run time, you may use memalign() function. Regards, Yuri Karlsbrun Bryan Wilcutt wrote: > Pragma align and pack seem to have 'some' effect under C (gnu) but NONE under > C++ (gnu). These are the compilers that ship with VxWorks. Some of the old > online gnu manuals state pragma is no longer supported but this contradicts what > is mentioned in the WindRiver manuals. Anyone have any ideas? > > Bw - --------------8B81D6369324029C82B17B8A Content-Type: text/x-vcard; charset=us-ascii; name="yuri.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Yuri Karlsbrun Content-Disposition: attachment; filename="yuri.vcf" begin:vcard n:Karlsbrun;Yuri tel;fax:(408) 730-3303 tel;work:(408) 730-3300 x146 x-mozilla-html:FALSE url:http://www.chameleonsystems.com org:Chameleon Systems, Inc. adr:;;1195 W. Fremont Avenue;Sunnyvale;CA;94087; version:2.1 email;internet:yuri@chameleonsystems.com title:Member of Technical Staff fn:Yuri Karlsbrun end:vcard - --------------8B81D6369324029C82B17B8A-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Re[2]: dc no carrier Date: 8 Sep 1999 09:57:02 -0700 From: wilhardt@mcd.mot.com Organization: Newsguy News Service [http://www.newsguy.com] Message-ID: <7r64gu$22bj@drn.newsguy.com> References: <199909071801.LAA07249@altosax.spk.hp.com> Tom, Call Wind River Customer Support (800-USA-4WRS) and request the patch for SPR#28573. This patch adds support for the DC_MODE flag and adds code to support non-autonegotiating equipment to the dec21x40End driver. - - Dave - -------------------------------------------------------------- Dave Wilhardt Motorola Computer Group DW278 (602) 438-6175 2900 S. Diablo Way wilhardt@mcd.mot.com Tempe, AZ 85282 - -------------------------------------------------------------- In article <199909071801.LAA07249@altosax.spk.hp.com>, Tom says... > >vxWorks , Tornado > >ron Heald wrote: >> Hi Tom, >> >> I saw your recent posting to com.os.vxworks concerning the message >> "interrupt: dc0 - no carrier". I have the same problem on our new and >> first mv2700 with the one exception that ours has only a 10Mbit >> connection. > >This must mean that you have full duplex 10Mbit connection. > >> >> In our case the whole thing is seems benign. I've been using the >> machine for a couple of months now with no problems except the >> messages. The things you said about losing log messages and your >> "More interesting facts" match exactly for me. >> > >several things here: > >1) I believe this IS mostly benign. I say 'mostly' for the following > reasons: 1) our serial port becomes unusable in some situations > because of the volume of dc no carrier messages. This isn't quite > benign. 2) logTask is being flooded with messages to the point it > is in some cases losing 100+ messages. If it can't keep up with the > messages, it would seem that this must affect system performance in > some way, even if it is interrupt driven. > >2) I don't understand why disabling duplex operation on the mv2303 > using the DC_MODE bits doesn't fix our problem, since the switch is > supposed to be autosensing. > >3) Even if we use the mode bits to set to 10Mbit vs 100Mbit, the ftp > performance is exactly the same. It all points to the DC_MODE bits > not really working in this case. I wonder if autosensing overrides > DC_MODE bits in normal operation. > > > >> We also have another mv2700 hosted on an NT machine with a 10Mbit >> connection which has NOT seen the problem at all. I use a Solaris >> host. > > >because it is not Not full duplex? > >> >> I submitted TSR# 137083 to WRS on this. They say this problem is >> SPR# 22196 which has been around since Tornado II beta. Their answer >> was to send an "unofficial" patch. That SPR is not listed on >> WindSurf. I've asked them why it's not listed a couple of times now, >> and never gotten a satisfactory answer. I haven't installed the >> patch (I'm troubled by the "unofficial") so I can't say what it >> does. >> >> You mentioned some earlier postings to the news group on this same >> problem. I've been unable to find them. > >What I did was use windsurf and searched on 'no carrier'. What I found >was the folks largely recommended changing the DC_MODE value to solve the >problem. As I already said, this didn't work for us. > >> >> Any ideas or suggestions you might have would be greatly appreciated. > >I filed a TSR with windriver with no response yet. I will ask them all >these question and try to get to the bottom of it + ask them about this >"unofficial" patch and how to get it. > >> >> Ron Heald >> National Radio Astronomy Observatory >> Socorro, New Mexico >> (505)835-7286 >> > > > Tom Antles > antles@spk.hp.com > Phone: (509) 921-3590 > Fax: (509) 921-3991 --------------------------- Newsgroups: comp.os.vxworks Subject: INCLUDE constant problem Date: Wed, 08 Sep 1999 15:29:16 -0400 From: John Sestak Organization: Raytheon Systems Company, Falls Church Message-ID: <37D6B90B.F6CB8249@fallschurch.esys.com> I'm having trouble adding certain INCLUDE constants to my VxWorks build. When I try and add: INCLUDE_SNMPD or INCLUDE_SM_OBJ, etc. in my VxWorks build I get a message saying that "Component 'SNMP Daemon' is not installed. You may need to purchase and/or install it". I'm fairly sure I installed everything on the Tornado2 CD but I still get this error. Any ideas? Thanks, John Sestak --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Writing BSP - excVecInit & classLibInit go AWOL Date: Wed, 08 Sep 1999 16:32:24 -0500 From: Marty Schrader Organization: Fermilab Experimental Astrophysics Group Message-ID: <37D6D5E8.64BCEE57@fnal.gov> References: <37D6A0BA.2A8A80D1@virata.com> I can't offer any insight as to why your functions see a white tunnel, Luke, but I have to offer one correction. I believe the trademark is more properly on, Do The Right Thing (tm). Does excVecInit look for the presense of interrupt-generating hardware during its operation? Have you traced into these lib calls? - --- Marty Schrader x6645 Fermilab Experimental Astrophysics Group (SDSS) consult0@fnal.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: INCLUDE constant problem Date: Wed, 08 Sep 1999 16:37:49 -0500 From: Marty Schrader Organization: Fermilab Experimental Astrophysics Group Message-ID: <37D6D72D.4CBB497C@fnal.gov> References: <37D6B90B.F6CB8249@fallschurch.esys.com> Is this perchance an optional package that WRS zicks you for? I mean, in addition to the T2 base package? The "You may need to purchase..." part of the error message kinda implies this. Is SNMP supported as part of some other network package that you have previously bought? For that matter, are you doing client or server side SNMP stuff, because they may require different packages. You know, Microschlock and other PC development tool vendors do that all the time. - --- Marty Schrader x6645 Fermilab Experimental Astrophysics Group (SDSS) consult0@fnal.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP problem Date: Wed, 8 Sep 1999 19:55:11 -0400 From: "Jeff Szczepanski" Organization: Time Warner Road Runner - Rochester NY Message-ID: References: <7r34l0$g4p$1@nnrp1.deja.com> <45hB3.5687$w6.208838@typhoon.nyroc.rr.com> <7r59of$3o3$1@nnrp1.deja.com> I'll take another crack at guessing on this: RAM is getting trashed. Analysis: 1) Your information relative to interrupts seems resonable & assuming the failure is directly synchronous to that particular instruction, then interrupt issues can probably be ruled out. 2) Highly likely that the kernel is NOT broken, therefore something BSP related is causing the problem.... 3) Based on the code trace provided, the instruction pointer in RAM to the spawn'ed task is corrupted and this is sending you into the weeds. 4) Likely causes of corrupted memory based on kernel not broken assumption: - - The initial stack setup before the kernel init call is not large enough or not initialized properly - - New hardware!? Correct Memory subsystem operation has not been confimed - - CPU Cache init/setup problems -> causes memory to get trashed from invalid dirty cache data. - - Code is not copied correctly to RAM (depending on your setup) or initialized data segment is not properly put into ram. I would say in this order: - - Don't enable the caches if there are enabled prior to kernel startup - - Confirm good hardware, run memory tests prior to kernel startup - - Confirm your non-zero data items in the initialized data segment has the proper values - - Confirm BSS segment is really getting zeroed out. - - Confirm initial stack is correct and plenty large enough before call to kernelInit(). - - Make sure your loaded code and symbol table matches to confirm everything is where its supposed to be. Regards, Jeff wrote in message news:7r59of$3o3$1@nnrp1.deja.com... > Thanks for the hint. This was my 1st. idea too. I checked it already. > I put some LED outs into the ppc860Intr.c. So I'm sure that > ppc860IntrInit > is called during pre kernel init and ppc860IntrDeMux or > ppc860CpmIntrDeMux are not called. > > I have traced the wrong jump by means of a logic analyzer. The result > is: > the function windSpawn fetches a wrong address from activeQHead > > windSpawn: stwu r1, 0xfff8(r1) > + 0x004 mfspr r0, LR > + 0x008 stw r0, 0xc(r1) > + 0x00c lis r9, Hi(activeQHead) > + 0x010 addi r9, r9, Lo(activeQHead) > + 0x014 addi r4, r3, 0x20 (32) > + 0x018 or r3, r9, r9 > + 0x01c lwz r9, 0xc(r3) > + 0x020 lwz r0, 0x10(r9) > + 0x024 mtspr LR, r0 > + 0x028 li r5, 0x0 (0) > + 0x02c blrl // last valid address > > > In article <45hB3.5687$w6.208838@typhoon.nyroc.rr.com>, > "Jeff Szczepanski" wrote: > > Just guessing..... > > > > One of the key things happening when you call kernelInit() from a BSP > > perspective is that interrupts get enabled. If the vectors are not > installed > > correctly, setup right, or a pending interrupt is not cleared properly > as > > they occur (ie: Hardware or driver issues)....then a hang or a 'walk > into > > the weeds' is likely. > > > > Regards, > > Jeff > > > > -- > > =========================== > > Jeffrey R. Szczepanski > > InSciTek Microsystems, Inc. > > 635 CrossKeys Office Park > > Fairport, NY 14450 > > > > > > wrote in message > > news:7r34l0$g4p$1@nnrp1.deja.com... > > > Hello, > > > I have a really serious problem to get a running > vxWorks.res_rom_nosym > > > on our custom board with MPC850. > > > > > > The problem is: > > > usrInit() calls kernelInit() and usrRoot() is not called > > > I found out that the function windSpawn() jumps to a wrong address. > > > > > > What is my fault??? > > > > > > any help will be appreciate > > > > > > Oliver > > > > > > > > > Sent via Deja.com http://www.deja.com/ > > > Share what you know. Learn what you don't. > > > > > > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP problem Date: Thu, 09 Sep 1999 01:24:15 GMT From: peter_mcconaghy@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7r727m$eah$1@nnrp1.deja.com> References: <7r34l0$g4p$1@nnrp1.deja.com> In article <7r34l0$g4p$1@nnrp1.deja.com>, oliver_fuchs@my-deja.com wrote: > Hello, > I have a really serious problem to get a running vxWorks.res_rom_nosym > on our custom board with MPC850. > > The problem is: > usrInit() calls kernelInit() and usrRoot() is not called > I found out that the function windSpawn() jumps to a wrong address. Is vxWorks.res_rom_nosym a valid target for the PPC architecture? From P450 of the VxWorks 5.3.1 Programers Guide it looks like you should be targeting vxWorks.res_rom_nosym_res_low. BTW, in the ADS860 Tornado 2 BSP notes (www.wrs.com/csdocs/techpubs/tor- 2.0/vxworks/bsp/ads860/ads860.html) right at the end it says that vxWorks.res_rom_res_low builds but does not work... Nifty eh! regads Peter Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Programming and Running Flash Date: Wed, 8 Sep 1999 21:04:50 -0500 From: "Philip Searcy" Organization: MindSpring Enterprises Message-ID: <7r74d2$l44$1@nntp1.atl.mindspring.net> Reply-To: "Philip Searcy" I need to be able to download an executable file to my PowerPC target, burn it into flash and then run it from flash, not RAM. I thought about using a flash file system, but that would only help with the storage, not the execution part. I already have a flash programming routine that seems to work with non-executable data. I was thinking of creating a RamDrv on the target, FTPing the file to the RamDrv and then burning the flash with the file at a previously designated address. Then I could execute by de-referencing the pointer to the function. Experiments with a hello world program show that I can do this if I use RAM instead of flash. I think I'm having more than one problem... First, I thought I needed the executable to be in flat binary format for programming the flash, so I took the output of the Tornado2 build and tried to run elfToBin. It told me that the file was not in ELF format. So, I ran objdump on it which told me that the file was ELF32. Wind River suggested that the build might not have done the link and I should try running the linker with the -e (entry point) and -Ttext (address) command line entries. No change. I also tried -r (relocatable). Still no change. I'm stuck. Any one have any ideas? Second, since I was stuck, I thought I'd try going off-the-wall with the following reasoning (such as it is). I thought that I needed a routine similar to loadModule, but loading into flash instead of loading into RAM. I managed to get source code to loadLib and found that loadModule calls loadModuleAt which calls loadModuleAtSym. Unfortunately, loadModuleAtSym calls a function by de-referencing a function pointer called loadRoutine and I can't find any other useage of loadRoutine anywhere so I can't figure out what it does. So, (here is the off-the-wall part) I used loadModule to load my hello world routine into RAM, then I took the address and size of the routine and programmed flash with it and tried to run it. It didn't work (no surprise) but I was curious at the mode of failure. It acted like it refused to even try to run the first instruction. It didn't bomb at the printf as I expected, but it gave me an exception before doing anything. I have read that the PowerPC can mark memory sectors as being non-executable and I'm wondering if this is one of my problems with executing from flash. I thought that the setting for non-executable would be in the sysPhysMemDesc table in sysLib, but I can't find it. Any one know where I should look for the PowerPC non-executable sector parameter? I see the #define in the .h file, but I can't find where it is used. Sorry for such a long message, but I'm getting at my wits' end (some of us have shorter wits than others). Thanks, Phil --------------------------- Newsgroups: comp.os.vxworks Subject: Shared Memory Problem Date: Tue, 07 Sep 1999 10:50:10 -0400 From: John Sestak Organization: Raytheon Systems Company, Falls Church Message-ID: <37D52622.56A8AAC9@fallschurch.esys.com> I have a question about using shared memory. In order to acess different portions of shared memory, do I always have to call smNameFind() in order to get a pointer to the area of shared memory I'm trying to find, or can I just use a pointer to another area of shared memory inside a block of shared memory? (I would convert this pointer to a global address using smObjLocalToGlobal() and then convert the pointer to a local address using smObjGlobalToLocal). Would this work? Thanks for any responses. John Sestak --------------------------- Newsgroups: comp.os.vxworks Subject: Shell Prompt Date: Thu, 9 Sep 1999 14:17:32 +0800 From: "Ç®É궰" Organization: Alcatel Message-ID: <37d752cb.0@dnews.sbell.com.cn> Dear All: I have a question about Vxworks Shell prompt. Can anyone tell me how to modify the Shell prompt "->" to the self-defined symbol just like ">"? Any answers are greatly appreciated! W&Q --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP problem Date: Thu, 09 Sep 1999 07:18:36 GMT From: oliver_fuchs@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7r7n05$s0p$1@nnrp1.deja.com> References: <7r34l0$g4p$1@nnrp1.deja.com> <7r727m$eah$1@nnrp1.deja.com> Thanks a lot. You are simply great! Thanks to Jeff too for all help. In article <7r727m$eah$1@nnrp1.deja.com>, peter_mcconaghy@my-deja.com wrote: > In article <7r34l0$g4p$1@nnrp1.deja.com>, > oliver_fuchs@my-deja.com wrote: > > Hello, > > I have a really serious problem to get a running vxWorks.res_rom_nosym > > on our custom board with MPC850. > > > > The problem is: > > usrInit() calls kernelInit() and usrRoot() is not called > > I found out that the function windSpawn() jumps to a wrong address. > > Is vxWorks.res_rom_nosym a valid target for the PPC architecture? From > P450 of the VxWorks 5.3.1 Programers Guide it looks like you should be > targeting vxWorks.res_rom_nosym_res_low. > > BTW, in the ADS860 Tornado 2 BSP notes (www.wrs.com/csdocs/techpubs/tor- > 2.0/vxworks/bsp/ads860/ads860.html) right at the end it says that > vxWorks.res_rom_res_low builds but does not work... Nifty eh! > > regads > Peter > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. > Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: PCI support in VxWorks (non-x86) Date: Tue, 07 Sep 1999 16:06:30 GMT From: polyester@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7r3d5s$mv7$1@nnrp1.deja.com> References: <7paeh2$3bh$1@nnrp1.deja.com> > By default, Vxworks works does not scan the entire Pci bus to detect > the devices and map the memory range into its Global Memory Map Table > (MMU – PHYS_MEM_DESC) while booting up. It is the responsibility of > device driver developer to provide a function that will detect the > device and create an entry in to Global Memory Map Table. > > The Pc bios does this hard work for you and allocates resources for > the pci devices. But Vxworks does not use this information to map this > into its MMU table. This work we have to do. I have given the entire > code bit and where to be inserted below. What is the case for non-x86 ? I'm using an MBX860, and on reading PCI config space there is no real address information there: Base Address0 = 0x0000 0008 Base Address1 = 0x0000 0001 Does this mean I have to allocate the mem myself, and write it to the PCI config space ? Many thanks, Kelvin. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Writing BSP - excVecInit & classLibInit go AWOL Date: Thu, 09 Sep 1999 10:06:23 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <37D76A7F.17A42A4C@bitband.com> References: <37D6A0BA.2A8A80D1@virata.com> Reply-To: leonid@bitband.com Here are my $.02: Make sure your RAM is fully functional for all possible access types (e.g. byte, short, long, long-long, quad) that your CPU can generate. Write your own memory test to run before you make these calls. That is the typical reason for failure at this stage. Leonid Luke Diamand wrote: > Hi! > > I'm writing a BSP for an ARM-based target. > > I find that excVecInit and classLibInit, when called, seem never to come > back. > > Does anyone know where I can find some documentation on what needs to be > done to persuade these functions to do the Right Thing (TM). > > Thanks in advance! > Luke Diamand > > -- > Virata http://www.virata.com > Cambridge > Tel: +44 1223 566919 Fax: +44 1223 566915 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Shell Prompt Date: Thu, 9 Sep 1999 10:46:00 +0200 From: "Christian Doppelbauer" Organization: VBS-Newsserver Message-ID: <936866845.686811@news.vbs.at> References: <37d752cb.0@dnews.sbell.com.cn> shellPromptSet( ) - allows you to do so. Chrisitian Ç®É궰 wrote in message news:37d752cb.0@dnews.sbell.com.cn... > Dear All: > I have a question about Vxworks Shell prompt. > Can anyone tell me how to modify the Shell prompt "->" to the > self-defined symbol just like ">"? > > Any answers are greatly appreciated! > > W&Q > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Shell Prompt Date: Thu, 09 Sep 1999 09:53:18 +0200 From: Zoltan Kakucs Organization: DESY Message-ID: <37D7676D.A8667A67@herant.desy.de> References: <37d752cb.0@dnews.sbell.com.cn> Reply-To: kakucs@herant.desy.de This is a multi-part message in MIME format. - --------------FE2B2B13B363D7BE7D88EC64 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi ! After downloading vxWorks, you can type (or add to your startup file): _shellPromptSet "Your preffered prompt" Regards, Zoltan > Dear All: > I have a question about Vxworks Shell prompt. > Can anyone tell me how to modify the Shell prompt "->" to the > self-defined symbol just like ">"? > > Any answers are greatly appreciated! > > W&Q - --------------FE2B2B13B363D7BE7D88EC64 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;cell:0177 - 24 12 361 tel;fax:040 - 8998 4388 tel;work:040 - 8998 3283 x-mozilla-html:FALSE 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 - --------------FE2B2B13B363D7BE7D88EC64-- --------------------------- Newsgroups: comp.os.vxworks Subject: errno 0xc0002 from puts() Date: Thu, 9 Sep 1999 11:09:16 +0100 From: "Will Fookes" Organization: NDS UK Message-ID: <7r7vql$vl2$1@ndsukns01.ndsuk.com> In our system (VxWorks 5.3.1, PPC860) we are seeing errno 0xc0002 the first time (and only the first time) we call puts() to write a string to stdout. So, if we do something like this: printf("Hello\n"); /* errno is not set by this */ puts("World"); /* errno now set to 0xc0002 */ What did we do wrong?! I can see that stdio.h makes a real meal of stdin/stdout and provides both function prototypes and macros for many of the standard i/o functions (for example, a simple call to getchar() in my code generates some 40 lines of assembler!) Any suggestions? Will --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Thu Sep 9 05:06:21 1999 From: "Ferrara, Bob" Date: Thu Sep 9 05:06:24 PDT 1999 Subject: Inserting a CR/LF in file using the Editor that is part of Tornad Is there a way to insert a CR/LF into my C source file using the Editor that comes with Tornado-2? So that, when I print my source file each function starts on the top of a new page. Thanks, Bob Ferrara VxWorks From owner-vxwexplo-process Thu Sep 9 06:28:13 1999 From: Ian Willats Date: Thu Sep 9 06:28:17 PDT 1999 Subject: Drift problem with ARM SA-1100 clock/timer driver David Laight has mentioned recently that the standard WRS sa1100Timer.c driver (supplied with the Brutus BSP) suffers badly from drift. To make matters worse, this seems to depend significantly on the whether or not the CPU's caches are enabled, and the extent of any interrupt lock-out periods in the BSP/application. One symptom of this which confused me no end is that timex produces unreliable results at best, and is plain incorrect when timing code that disables interrupts for more than a few machine instructions. FYI, I believe the problem is due to the way the ISR re-loads the counter match register with the current counter value plus the number of counts per clock tick period (plus a hard-coded fudge factor). Since the next match occurs a fixed time (for a given clock tick period) after the current match ISR runs, any delay to the ISR results in un-recoverable clock drift. The fudge factor is supposed to account for the elapsed time between the match occurring in hardware and the ISR actually running and re-loading the match register, but since it is fixed, it can't correct for variable interrupt latency, CPU performance, etc. I have a re-worked version of the driver that uses a different re-loading technique and thereby avoids the drift problem (as far as I can tell); please e-mail me if you'd like a copy. HTH 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 --------------------------------------------------------------- VxWorks From owner-vxwexplo-process Thu Sep 9 07:12:46 1999 From: "Slutsker, Rasty" Date: Thu Sep 9 07:12:50 PDT 1999 Subject: SLIP/ARP/PROXY in VxWorks 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_01BEFACD.38E65954 Content-Type: text/plain; charset="iso-8859-1" Hi, Did anyone succeed to have Proxy ARP server with the SLIP? I have five targets connected together though Ethernet, one of them has also SLIP connection to the host computer. I'd like to establish connection between host and any target. I initially planned to run Proxy ARP server on the target that has SLIP connection to the host, but proxyNetCreate always returns -1, no examples in manuals, tried a dozen of combinations ;-(. Any suggestions? Rasty Tornado VxWorks ------_=_NextPart_001_01BEFACD.38E65954 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable SLIP/ARP/PROXY in VxWorks

Hi,

Did anyone succeed to have Proxy ARP = server with the SLIP?

I have five targets connected together though = Ethernet, one of them has also SLIP = connection to the host computer.  I'd like to = establish connection between host = and any target. = I initially planned to = run Proxy ARP = server on the target that has SLIP = connection to the host, but = proxyNetCreate always returns -1, = no examples in manuals, tried a dozen of = combinations ;-(.

Any = suggestions?

Rasty

Tornado = VxWorks

------_=_NextPart_001_01BEFACD.38E65954-- From owner-vxwexplo-process Thu Sep 9 10:49:20 1999 From: Michael Durling Date: Thu Sep 9 10:49:23 PDT 1999 Subject: ostrstream destructor sets task errno Any idea why the ostrstream destructor (VxWorks/Tornado 2) would be setting the task's errno to 0x3d0001? #include #include void ostrtest() { { char buffer[80]; printf( "Entering ostrstream::ostrstream %s\n", strerror(errno) ); ostrstream test( buffer, 80 ); printf( "Leaving ostrstream::ostrstream %s\n", strerror(errno) ); printf( "Entering ostrstream::~ostrstream %s\n", strerror(errno) ); } printf( "Leaving ostrstream::~ostrstream %s\n", strerror(errno) ); } (5,11)-> ostrtest Entering ostrstream::ostrstream OK Leaving ostrstream::ostrstream OK Entering ostrstream::~ostrstream OK Leaving ostrstream::~ostrstream S_objLib_OBJ_ID_ERROR value = 54 = 0x36 = '6' Curious, Michael Michael Durling mailto:mdurling@opticalnetworks.com OPTICAL NETWORKS 166 Baypointe Pkwy San Jose, CA 95143 Phone: (408) 965-2636(408) Fax: (408) 965-2660 From owner-vxwexplo-process Fri Sep 10 04:00:09 1999 From: daemon@csg.lbl.gov Date: Fri Sep 10 04:00:13 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Sep 10 04:00:05 PDT 1999 Subject: Re: How do I change the System Time and Date Subject: S3 VIRGE GX II driver Subject: Re: Programming and Running Flash Subject: Re: Tornado II with Tornado I Subject: Re: INCLUDE constant problem Subject: CORBA on VxWorks ? Subject: Inflate Problems Subject: END driver SPR Subject: shellinit and console settings Subject: Problem with ppc860DevInit ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: How do I change the System Time and Date Date: Thu, 09 Sep 1999 13:59:46 +0200 From: Carlos David Organization: Ericsson Eurolab Germany Message-ID: <37D7A132.2842A5CB@eed.ericsson.se> References: <37D3E048.866C1DBA@eed.ericsson.se> <936699000.502529@news.vbs.at> I will try, thank you very much. Bye --------------------------- Newsgroups: comp.os.vxworks Subject: S3 VIRGE GX II driver Date: Thu, 9 Sep 1999 14:19:30 +0100 From: "sct info" Message-ID: <7r88dr$j2r$1@news4.isdnet.net> Hello, I'm looking for a VxWorks SPARC driver for the S3 VIRGE GX II chip. Would anyone know where to download and/or purchase one ? Thanks, Didier Fonteneau. iit@toutelectric.fr --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Programming and Running Flash Date: Thu, 09 Sep 1999 09:12:31 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37D7B23F.9320041@lucent.com> References: <7r74d2$l44$1@nntp1.atl.mindspring.net> Philip Searcy wrote: > to get source code to loadLib and found that loadModule calls loadModuleAt > which calls loadModuleAtSym. Unfortunately, loadModuleAtSym calls a function > by de-referencing a function pointer called loadRoutine and I can't find any > other useage of loadRoutine anywhere so I can't figure out what it does. It is a function pointer that is set by loadElfInit() loadCoffInit(), or loadAoutInit(), depending on what your file format is. Look in usrConfig.c to see where these are used. Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II with Tornado I Date: Thu, 09 Sep 1999 13:18:00 GMT From: Ran@127.0.0.1 (Ran Shalgi) Organization: Only when absolutely necessary Message-ID: <37d7b2b9.881816094@news.inter.net.il> References: <37ce25b3.155893292@news.bournemouth-net.co.uk> <37CEA1D9.5765E0EF@bitband.com> On Thu, 02 Sep 1999 18:12:09 +0200, Leonid Rosenboim wrote: >Unlike UNIX, Windows can not maintain multiple installations of Tornado on >the same machine, so you will have to leave one or two dedicated NT >machines with Tornado 1 on them, while installing Tornado 2 on other >machines. WRS has this clearly documented on the README notes shipped with >every CD! I guess its a good thing, then, that I never read that README. I have both T1 and T2 up and running on the same machine with no problems what so ever. I guess that if you don't know that it's impossible, you can do it. Ran Shalgi Expand Networks, Inc http://www.expand.com - --- Replace 127.0.0.1 with expand.com when replying by e-mail --------------------------- Newsgroups: comp.os.vxworks Subject: Re: INCLUDE constant problem Date: Thu, 09 Sep 1999 16:38:09 GMT From: Kremer@wxs.nl (Harro Kremer) Organization: Planet Internet Message-ID: <37d7e217.881326534@news.wxs.nl> References: <37D6B90B.F6CB8249@fallschurch.esys.com> John Sestak wrote: >I'm having trouble adding certain INCLUDE constants to my VxWorks >build. When I try and add: INCLUDE_SNMPD or INCLUDE_SM_OBJ, etc. in my >VxWorks build I get a message saying that "Component 'SNMP Daemon' is >not installed. You may need to purchase and/or install it". I'm fairly >sure I installed everything on the Tornado2 CD but I still get this >error. Any ideas? You most likely will not have installed everything on the T2 CD. WRS has (at least for T1) the policy of having ALL software on the CD and use the licence key to select the purchased components. SNMP is indeed an optional component with addition costs. Harro Kremer --------------------------- Newsgroups: comp.os.vxworks Subject: CORBA on VxWorks ? Date: Thu, 9 Sep 1999 20:54:49 +0200 From: "Bertrand Marlier" Organization: Guest of France-Teaser Message-ID: <7r8v77$s6v$1@feed.teaser.fr> Reply-To: "Bertrand Marlier" Sender: bmarlier@dip227.teaser.fr Does anyone have an experience of CORBA on VxWorks ? --------------------------- Newsgroups: comp.os.vxworks Subject: Inflate Problems Date: Tue, 7 Sep 1999 17:03:43 -0400 From: "Nelson Owens" Organization: Hughes Network Systems Message-ID: <7r3u3i$geh$1@hnssysb.md.hns.com> Has anyone had problems with the vxworks deflate/inflate tools. We use them to compress/decompress FPGA files for our MIPS based targets. Some of the files will fail dramatically (Address Load Exception). It seems to be data related - if we ask the hardware guys to make a minor change to the FPGA file - it works. We have tried enabling checksum - no difference --------------------------- Newsgroups: comp.os.vxworks Subject: END driver SPR Date: Thu, 9 Sep 1999 12:07:35 -0700 From: "Dave Craig" Organization: Verio Message-ID: Hi, Just thought I'd share the guts of an SPR I got from WRS as I coudn't find it on their web site...it 's been dogging me for about 6 months now as scribbling/crashing/instability and seems to be rather widespread as it affects all BSPs using END drivers and Tornado 1.0.1.... >There is a known problem SPR# 23229: END DMA devices do not shut >down on reboot or in bootConfig.c "go" (aka: no ifreset in END). > >Problem_Description: > >END DMA devices do not shut down on reboot or in bootConfig.c "go" >(aka: no ifreset in END) This can cause problem booting or >rebooting. > >If you do not shutdown active DMA devices (such as Ethernet chips), >memory corruption for the image your about to run can occur, >causing very unpredictable behavior. The device may enter a >unrecoverable state. The error may be difficult to notice, >since it is memory corruption. The error will likely also be >intermittent since it is an asynchronous DMA device. > >This is only a problem for END drivers, while for BSD 4.x (if_blah, >if_ln, if_fei, etc..) drivers, this shutdown is being done properly. >There is a function ifreset(), which will shutdown/reset any if device. >For bootrom images ifreset is called in bootConfig.c's go() function, >and for runtime images ifreset() is being called via a rebootHook >added via netLibInit(). > >Shutting down the Ethernet chip is not being done by END drivers. There is a straightforward resolution so contact tech support and ask nicely. Dave Craig Palo Alto. --------------------------- Newsgroups: comp.os.vxworks Subject: shellinit and console settings Date: Thu, 09 Sep 1999 23:25:30 GMT From: micron@bluemoon.net (Craig Brown) Organization: VE Message-ID: Hi, I'm using VxWorks v2.0.3b with the MV167 platform with a VT220 terminal. The problem is, I'm using the startup script to spawn a task. The task is a program that uses a tty console terminal to display a menu and uses the "n" and "p" keys to tab back and forth into the input fields. The program works fine if I let the system boot up normally to the shell and type in the taskSpawn command with the proper options. However when I put the same taskSpawn command into a startup script, the programmed "tab" keys do not function properly. Could it be that there could is some other initialization process that has not completed, (ie: shellInit or usrClk)? Any help would be appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with ppc860DevInit Date: Fri, 10 Sep 1999 08:21:45 GMT From: peter_mcconaghy@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7raf2h$tvd$1@nnrp1.deja.com> Hi there. We've got a problem with our board "crashing" in ppc860DevInit. We are using "custom" MPC850 hardware which is based on the MPC860 reference designs. We are using VxWorks 5.3.1 and the ADS860 BSP. I have change the serial port from SMC1 to SMC2 in SysSerialHwinit and made all the other necessary changes. When I run it the board crashes with a memory exception about 0x158 bytes after the start of ppc860DevInit - trying to write to 0x00440220 (RAM is from 0x0-0x3FFFFF). I have checked the setup code for the function call and it looks ok (see below). Has anyone come across this? I've never tried to debug a problem like this in object code before - does anyone have any ideas? Regards Peter Code snippet from sysSerial.c follows: void sysSerialHwInit (void) { /* * intialize the chips device descriptors */ ppc860Chan.clockRate = BRGCLK_FREQ; /* BRGCLK freq (Hz) */ ppc860Chan.regBase = vxImmrGet(); /* IMMR reg has base adr */ ppc860Chan.bgrNum = 1; /* use BRG1 */ ppc860Chan.uart.smcNum = 2; /* SMC2 wired for rs232 */ ppc860Chan.uart.txBdNum = 1; /* use 1 transmit BD */ ppc860Chan.uart.rxBdNum = 0x10; /* use 16 receive BD */ ppc860Chan.uart.txBdBase = /* transmit BD base adrs */ (SMC_BUF *) (MPC860_DPRAM_BASE (ppc860Chan.regBase) + 0x200); ppc860Chan.uart.rxBdBase = /* receive BD base adrs */ (SMC_BUF *) (MPC860_DPRAM_BASE (ppc860Chan.regBase) + 0x210); ppc860Chan.uart.txBufBase = /* tx buf base */ (u_char *) (MPC860_DPRAM_BASE (ppc860Chan.regBase) + 0x300); ppc860Chan.uart.rxBufBase = /* rx buf base */ (u_char *) (MPC860_DPRAM_BASE (ppc860Chan.regBase) + 0x380); ppc860Chan.uart.txBufSize = 0x1; /* transmit buffer size */ ppc860Chan.uart.pSmc = /* DPRAM addr of SMC2 params */ (SMC *) ((UINT32) PPC860_DPR_SMC2( MPC860_DPRAM_BASE (ppc860Chan.regBase))); ppc860Chan.uart.pSmcReg = /* SMCMR2 for SMC2 */ (SMC_REG *) ((UINT32) MPC860_SMCMR2(ppc860Chan.regBase)); ppc860Chan.pBaud = (UINT32 *) MPC860_BRGC1 (ppc860Chan.regBase); ppc860Chan.channelMode = 0; /* * select PA8/PA9 to be internal peripherals */ *MPC860_PBPAR(ppc860Chan.regBase) |= (cBit16_8 | cBit16_9); /* * Set up the SDMA Config register so that it at the "normal" operation setting */ *MPC860_SDCR(ppc860Chan.regBase) = 1; /* * Initialise the SMC UART */ ppc860DevInit(&ppc860Chan); } Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Fri Sep 10 06:05:07 1999 From: Reyhan ERGUN Date: Fri Sep 10 06:05:11 PDT 1999 Subject: BSP modifications and generating vxWorks image with Tornado II This is a multi-part message in MIME format. --------------A639A4CD505FFFDC3B783D54 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, I use Tornado 2, and I want to build vxWorks (a ram-based image) after I make several additions to the BSP. I have generated a workspace to create a bootable vxWorks image (custom configured) which was based on a BSP (MBX860). Therefore, several files were automatically included into the workspace such as sysLib.c, romInit.s, etc. In order to add some serial devices I will modify 3 files within the BSP: mbx800.h, config.h, sysSerial.c. If I open and edit these files and than build vxWorks in Tornado 2, I realize that the modifications were not taken into account. If I include these files into workspace I encounter many problems regarding to dependencies for newly included file sysSerial.c. I wonder how I could modify BSP by using Tornado 2. Any comments appreciated. Best regard, Reyhan --------------A639A4CD505FFFDC3B783D54 Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Reyhan (FINDIK) ERGUN Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Reyhan (FINDIK) ERGUN n: ;Reyhan (FINDIK) 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 --------------A639A4CD505FFFDC3B783D54-- From owner-vxwexplo-process Fri Sep 10 08:13:14 1999 From: Jolie Key Date: Fri Sep 10 08:13:18 PDT 1999 Subject: Interrupt problems Current system is pentium target, sun/solaris host. VxWorks 5.3.1 for PC486, WIND version 2.5. Target has a National Instruments GPIB card and a B&B 4-port serial card. With the 2 on-board serial ports plus the four B&B ports, there are a total of six serial ports. I have set the jumpers as follows: COM1 - on-board - 0x3f8 - IRQ 4 COM2 - on-board - 0x2f8 - IRQ 3 COM3 - B&B - 0x3e8 - IRQ 5 COM4 - B&B - 0x2e8 - IRQ 7 COM5 - B&B - 0x238 - IRQ 12 COM6 - B&B - 0x248 - IRQ 15 GPIB Board - IRQ 11 I can poll all ports, so I know the base addresses are correct. I can also use the GPIB interface. However, polling is slow, so I am trying to get interrupts to work. I can only get IRQ 5, 7 and 15 to work. IRQ 11 works when used for a COM port, but not as SRQ for GPIB. I think the BIOS is set correctly for all IRQs. Particularly, I can't understand why COM1 and COM2 can't use their interrupts. I have called the motherboard vendor, but they don't have any suggestions, except to tailor the BIOS for us. But what would I have it tailored to? They are using AWARD Modular BIOS 4.51PG. The GPIB card is acting funny, too. As soon as I activate the interrupt for its SRQ via intConnect and intEnablePIC, the ibcmd and ibwrt routines no longer work. They just bomb or get in some wait state. Ultimately, I would like to have all COM ports on the given interrupts, and also be able to use the GPIB SRQ feature. Any suggestions? Could this be a vxWorks problem, or is it a BIOS problem? Thank you, Jkey From owner-vxwexplo-process Fri Sep 10 08:36:59 1999 From: Ahmet Mumcu Date: Fri Sep 10 08:37:03 PDT 1999 Subject: PROBLEM ON USING RHAPSODY WITH TORNADO-2 Hi there, Does anyone ever tried using Rhapsody with Tornado-2 for PowerPC targets? We have problem on making generated code using Tornado-2 tools but can make the code successfully with Tornado-1 tools. The problem seems to be related to gcc version provided with Tornado-2. Ahmet Mumcu ASELSAN Inc. MST Group Ankara-TURKEY e-mail: mumcu@mst.aselsan.com.tr From owner-vxwexplo-process Sat Sep 11 04:00:09 1999 From: daemon@csg.lbl.gov Date: Sat Sep 11 04:00:14 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Sep 11 04:00:05 PDT 1999 Subject: ld960.exe Application Error Subject: RE: Tornado 2 Make Dependencies Subject: Re: ld960.exe Application Error Subject: Re: shellinit and console settings Subject: Re: Problem with ppc860DevInit Subject: Inflate Problems ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: ld960.exe Application Error Date: Fri, 10 Sep 1999 14:26:08 +0200 From: "BT" Message-ID: <7ratar$i4d$1@oslo-nntp.eunet.no> When linking large system where 'close to all' files are compiled with debug option (-g) the linker ld960.exe fails with a message box with following text: The instruction at "0x00417f2b" referenced memory at "0x00b5f063". The memory could not be "read". The linker ld960.exe is part of Tronado 5.3.1. The development host is a NT 4.0 service pack 3. The failure always occurs when 'all' source files are compiled with debug info. I have tried different approaches eg. putting the system togehter with several smaller libs, creating my system from within the Tornado 'studio'. via makefiles and bat files from command prompt. I have also tried command.com with different memory configurations. Different flags have been tried among them 'split-by-reloc' with different values. Anyone had something similar? Brede T. --------------------------- Newsgroups: comp.os.vxworks Subject: RE: Tornado 2 Make Dependencies Date: Fri, 10 Sep 1999 14:31:15 +0200 From: "Borkhuis, Johan" Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: Hi Marjorie, > ---------- > From: Marjorie Krueger[SMTP:mkrueger@vixel.com] > Sent: vrijdag 10 september 1999 2:41 > To: Borkhuis, Johan > Cc: 'comp.os.vxworks@list.deja.com'; > 'Eric_D_Woolverton@md.northgrum.com' > Subject: Re: Tornado 2 Make Dependencies > > I assume you are referring to your suggestion to add the VPATH macro as > the solution to our > problem? > Yes. > But I thought the fact that I included the search paths in the compiler > directives > (-I../snmp -I../h) should have allowed the tools to find the dependencies. > > The -I directives are only used by the compiler. The dependencies are generated by make, and make does not use these, but needs the VPATH to be set. The best solution would be to have an aditional screen in the configuration where include paths could be added. These should be converted by Tornado to the right -I statements and a VPATH macro. In your second mail you wrote: > I just tried adding the VPATH macro - this allows me to compile the files > with the "dangling > dependencies", but Tornado still doesn't know where that .h file is. So > it fixes the problem for > the gnu compiler, but not for Tornado. > One thing I did not mentions is that you have to do a "Save workspace" before you start regenerating dependencies and compiling the code. That should work. Otherwise check the Makefile to see if the dependencies in that file are correct, with the complete pathname or with $(PRJ_NAME). Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ld960.exe Application Error Date: Fri, 10 Sep 1999 16:12:25 +0100 From: "Mark Hambleton" Organization: (Posted via) Netcom Internet Ltd. Message-ID: <7rb71i$kit$1@taliesin.netcom.net.uk> References: <7ratar$i4d$1@oslo-nntp.eunet.no> We had this problem a while back (~11 Months), you need to ask WRS for a newer version of the linker, (for some reason the last 2 i960 updates have arrived with the broken version so we have to back track it..) - -- Mark Hambleton http://www.xstreamis.com/ BT wrote in message <7ratar$i4d$1@oslo-nntp.eunet.no>... >When linking large system where 'close to all' files are compiled with debug >option (-g) >the linker ld960.exe fails with a message box with following text: > >The instruction at "0x00417f2b" referenced memory at "0x00b5f063". The >memory >could not be "read". > >The linker ld960.exe is part of Tronado 5.3.1. The development host is a NT >4.0 service pack 3. >The failure always occurs when 'all' source files are compiled with debug >info. >I have tried different approaches eg. putting the system togehter with >several smaller libs, >creating my system from within the Tornado 'studio'. via makefiles and bat >files from command prompt. I have also tried command.com with different >memory configurations. >Different flags have been tried among them 'split-by-reloc' with different >values. > >Anyone had something similar? > >Brede T. > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: shellinit and console settings Date: Fri, 10 Sep 1999 09:19:56 -0700 From: Ken Norlin Organization: NASA Dryden FRC Message-ID: <37D92FAC.C4DD20D9@dfrc.nasa.gov> References: Craig Brown wrote: > > Hi, > > I'm using VxWorks v2.0.3b with the MV167 platform with a VT220 terminal. The > problem is, I'm using the startup script to spawn a task. The task is a program > that uses a tty console terminal to display a menu and uses the "n" and "p" > keys to tab back and forth into the input fields. The program works fine if I > let the system boot up normally to the shell and type in the taskSpawn command > with the proper options. However when I put the same taskSpawn command into a > startup script, the programmed "tab" keys do not function properly. Could it be > that there could is some other initialization process that has not completed, > (ie: shellInit or usrClk)? Any help would be appreciated. Hi, I had a similar problem with my program. I ended up defining "INCLUDE_USER_APPL" in config.h. Since I didn't want to hard-wire my task name here, I used the startup script to set environment variables for the task name,size,priority,etc and then my user application that was started at bootup would check these and perform the taskSpawn. It works great, but looking back on my code -- there's probably an easier way to do this. Email me if you want the myAppInit() routine that checks environment variables & spawns the appropriate task. - -- Ken Norlin --- Simulation Engineer NASA Dryden Flight Research Center ken.norlin@dfrc.nasa.gov (661) 258-2046 FAX (661) 258-2792 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with ppc860DevInit Date: Fri, 10 Sep 1999 16:40:43 GMT From: peter_mcconaghy@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rbca8$j2q$1@nnrp1.deja.com> References: <7raf2h$tvd$1@nnrp1.deja.com> In article <7raf2h$tvd$1@nnrp1.deja.com>, peter_mcconaghy@my-deja.com wrote: > Hi there. > > We've got a problem with our board "crashing" in ppc860DevInit. We are > using "custom" MPC850 hardware which is based on the MPC860 reference > designs. We are using VxWorks 5.3.1 and the ADS860 BSP. I think I've found the problem (but the source code control system is seemingly asleep at the moment and won't let me change it...). I set the two variables below to the "real" addresses instead of the offsets from the IMMR. Regards, Peter > Code snippet from sysSerial.c follows: > void sysSerialHwInit (void) > { > /* > * intialize the chips device descriptors > */ ppc860Chan.uart.txBdBase = > /* transmit BD base adrs */ > (SMC_BUF *) (MPC860_DPRAM_BASE > (ppc860Chan.regBase) + 0x200); Should be 0x2200 > ppc860Chan.uart.rxBdBase = > /* receive BD base adrs */ > (SMC_BUF *) (MPC860_DPRAM_BASE > (ppc860Chan.regBase) + 0x210); should be 0x2210 Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Inflate Problems Date: Fri, 10 Sep 1999 09:51:41 -0400 From: "Gordon, Charles" Message-ID: References: <7r3u3i$geh$1@hnssysb.md.hns.com> Sender: "Gordon, Charles" I've had problems with inflate() before. Inflate often fails if the image is larger than about 700K. This is a problem Wind River is aware of. There is an SPR about it. I was able to get around the problem by zeroing out the memory that inflate() uses just before I call it. > -----Original Message----- > From: Nelson Owens [mailto:nowens@hns.com] > Posted At: Tuesday, September 07, 1999 5:04 PM > Posted To: vxworks > Conversation: Inflate Problems > Subject: Inflate Problems > > > Has anyone had problems with the vxworks deflate/inflate > tools. We use them > to compress/decompress FPGA files for our MIPS based targets. > Some of the > files will fail dramatically (Address Load Exception). It > seems to be data > related - if we ask the hardware guys to make a minor change > to the FPGA > file - it works. We have tried enabling checksum - no difference > > --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sat Sep 11 08:38:23 1999 From: Reyhan ERGUN Date: Sat Sep 11 08:38:27 PDT 1999 Subject: Standard BSP Builds within Tornado II problem This is a multi-part message in MIME format. --------------050BE78195A6D0F8B2DFC6CB Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, Even it is shown that there are "Standard BSP Builds -> ->Common Targets, ->VxWorks Targets, ->Boot ROM Targets" menu items for Build menu, in the VxWorks Programmer's Guide, I only have one part in the Build menu and don't have these menu items. What is missing in the tool I use? I am looking forward to make clean, make vxWorks, etc, in a more convenient way by using the menus in Tornado II for Windows? Best regards, Reyhan --------------050BE78195A6D0F8B2DFC6CB Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Reyhan (FINDIK) ERGUN Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Reyhan (FINDIK) ERGUN n: ;Reyhan (FINDIK) 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 --------------050BE78195A6D0F8B2DFC6CB-- From owner-vxwexplo-process Sun Sep 12 04:00:07 1999 From: daemon@csg.lbl.gov Date: Sun Sep 12 04:00:11 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Sep 12 04:00:04 PDT 1999 Subject: Re: Standard BSP Builds within Tornado II problem ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Standard BSP Builds within Tornado II problem Date: Sun, 12 Sep 1999 05:06:06 GMT From: David Hamel Organization: @Home Networks. Message-ID: <37DB34CB.D3D62168@ntanet.net> References: <7rdv60$3ff7@overload.lbl.gov> Make clean is gone... Try "Rebuild All" instead. As far as the list of build types... Go to Workspace->builds, right click the default build(or make a new one). Select properties the select the Rule tab. and change the build rule to any one of many. Reyhan ERGUN wrote: > > This is a multi-part message in MIME format. > --------------050BE78195A6D0F8B2DFC6CB > Content-Type: text/plain; charset=us-ascii > Content-Transfer-Encoding: 7bit > > Hi, > Even it is shown that there are > "Standard BSP Builds > -> > ->Common Targets, > ->VxWorks Targets, > ->Boot ROM Targets" > menu items for Build menu, in the VxWorks Programmer's Guide, I only have one part in the Build menu and > don't have these menu items. What is missing in the tool I use? I am looking forward to make clean, make > vxWorks, etc, in a more convenient way by using the menus in Tornado II for Windows? > > Best regards, Reyhan > > --------------050BE78195A6D0F8B2DFC6CB > Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" > Content-Transfer-Encoding: 7bit > Content-Description: Card for Reyhan (FINDIK) ERGUN > Content-Disposition: attachment; filename="vcard.vcf" > > begin: vcard > fn: Reyhan (FINDIK) ERGUN > n: ;Reyhan (FINDIK) 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 > > --------------050BE78195A6D0F8B2DFC6CB-- --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Sep 13 02:10:12 1999 From: Ilia Bosis Date: Mon Sep 13 02:10:15 PDT 1999 Subject: RE: PCI support in VxWorks (non-x86) Kevin. At least in some PowerPC BSP's, the VxWorks DOES the PCI bus scanning. The MVME23xx/MVME26xx BSPs do some scanning at the system start-up. They doesn't scan for all possible devices (this is not a full Plug-and-play after all), but looks for the devices they expect in the specific HW configuration and map them into the PCI memory and I/O address space. An exact addresses have to be defined before the scan, possibly by some #define constants. To add a HW device You need only add a piece of code that looks for the device on the PCI bus (by VendorID, DeviceID) and map it to the desired address. For examples look at the function sysHwInit in file sysLib.c in the BSP directory. Hope this helps. Ilia Bosis. >> By default, Vxworks works does not scan the entire Pci bus to detect >> the devices and map the memory range into its Global Memory Map Table >> (MMU - PHYS_MEM_DESC) while booting up. It is the responsibility of >> device driver developer to provide a function that will detect the >> device and create an entry in to Global Memory Map Table. >> > >What is the case for non-x86 ? I'm using an MBX860, and on reading PCI >config space there is no real address information there: > >Base Address0 = 0x0000 0008 >Base Address1 = 0x0000 0001 > >Does this mean I have to allocate the mem myself, and write it to the >PCI config space ? > >Many thanks, >Kelvin. +=====================================================================+ | Bosis Ilia | Work phone: 972-3-5384685 | | Software engineer | Fax: 972-3-5335877 | | Telegate ltd. | E-mail: iliab@telegate.co.il | | 60218 Haplada7 Or-Yehuda, Israel | WWW: http://www.telegate.co.il | +=====================================================================+ From owner-vxwexplo-process Mon Sep 13 04:00:09 1999 From: daemon@csg.lbl.gov Date: Mon Sep 13 04:00:13 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Sep 13 04:00:05 PDT 1999 Subject: Re: shellinit and console settings Subject: Re: Problem with ppc860DevInit Subject: Re: shellinit and console settings Subject: Multimedia capture Subject: Java programming ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: shellinit and console settings Date: Sun, 12 Sep 1999 21:42:43 GMT From: johan@borksoft.xs4all.nl (Johan Borkhuis) Organization: JAB Software Productions Message-ID: References: Craig Brown (micron@bluemoon.net) wrote: > Hi, > > I'm using VxWorks v2.0.3b with the MV167 platform with a VT220 terminal. The > problem is, I'm using the startup script to spawn a task. The task is a program > that uses a tty console terminal to display a menu and uses the "n" and "p" > keys to tab back and forth into the input fields. The program works fine if I > let the system boot up normally to the shell and type in the taskSpawn command > with the proper options. However when I put the same taskSpawn command into a > startup script, the programmed "tab" keys do not function properly. Could it be > that there could is some other initialization process that has not completed, > (ie: shellInit or usrClk)? Any help would be appreciated. Have you tried different initialisations for the shell? What happens when you startup and give the commands from the Tornado shell? One thing you can do in the startup script is kill the shell task and then perform your taskSpawn. This initialises the terminal so you can use this initialised terminal for your application. If this works you can try to find out which parameters the shell uses to initialise the terminal. I hope this helps. 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` ==== VxWorks-page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html ==== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with ppc860DevInit Date: Sun, 12 Sep 1999 21:47:20 GMT From: johan@borksoft.xs4all.nl (Johan Borkhuis) Organization: JAB Software Productions Message-ID: References: <7raf2h$tvd$1@nnrp1.deja.com> <7rbca8$j2q$1@nnrp1.deja.com> peter_mcconaghy@my-deja.com wrote: > In article <7raf2h$tvd$1@nnrp1.deja.com>, > peter_mcconaghy@my-deja.com wrote: > > I think I've found the problem (but the source code control system is > seemingly asleep at the moment and won't let me change it...). I set > the two variables below to the "real" addresses instead of the offsets > from the IMMR. > > ppc860Chan.uart.txBdBase = > > /* transmit BD base adrs */ > > (SMC_BUF *) (MPC860_DPRAM_BASE > > (ppc860Chan.regBase) + 0x200); > > Should be 0x2200 Have you tried the following: > ppc860Chan.uart.txBdBase = > > /* transmit BD base adrs */ > > (SMC_BUF *) (MPC860_DPRAM_BASE ((ppc860Chan.regBase) + 0x200)); ^ ^ ^ ^ Extra braces around the parameters for MPC860_DPRAM_BASE 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` ==== VxWorks-page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html ==== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: shellinit and console settings Date: Mon, 13 Sep 1999 10:05:04 +0100 From: Phil Ashby Organization: (AshbySoft *) Message-ID: <37DCBE3F.AE706645@ashbysoft.cix.co.uk> References: AFAIK, startup scripts are run in a separate instance of the shell, which then terminates before the interactive shell is started. This means that the associated I/O manipulation (assigning stdin/stdout etc.) will happen while your application is running, probably blowing things away. Perhaps avoiding the interactive shell entirely and replacing it with your application would help... Craig Brown wrote: > > Hi, > > I'm using VxWorks v2.0.3b with the MV167 platform with a VT220 terminal. The > problem is, I'm using the startup script to spawn a task. The task is a program > that uses a tty console terminal to display a menu and uses the "n" and "p" > keys to tab back and forth into the input fields. The program works fine if I > let the system boot up normally to the shell and type in the taskSpawn command > with the proper options. However when I put the same taskSpawn command into a > startup script, the programmed "tab" keys do not function properly. Could it be > that there could is some other initialization process that has not completed, > (ie: shellInit or usrClk)? Any help would be appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Multimedia capture Date: Mon, 13 Sep 1999 10:45:03 +0100 From: "Mark D Spiteri" Organization: University of Cambridge Message-ID: <7rih2v$8mb$1@pegasus.csx.cam.ac.uk> Reply-To: "Mark D Spiteri" Is there any video capture card supported by VxWorks, or for which drivers exist? Similarly for a sound card. We'd like to get multimedia streams in and then deliver them over Ethernet. Any pointers appreciated! Mark - -- Mark D. Spiteri - mds24@cam.ac.uk Laboratory for Communications Engineering Cambridge University Engineering Department Trumpington Street, Cambridge CB2 1PZ, United Kingdom. --------------------------- Newsgroups: comp.os.vxworks Subject: Java programming Date: Mon, 13 Sep 1999 10:46:43 +0100 From: "Mark D Spiteri" Organization: University of Cambridge Message-ID: <7rih63$8o4$1@pegasus.csx.cam.ac.uk> Reply-To: "Mark D Spiteri" I've been trying to determine, prior to getting VxWorks, how stable and integrated its Java support is. Does anyone in this newsgroup have experience of large programming projects in Java on VxWorks? How does it integrate with the Tornado environment? Thanks for any advice Mark - -- Mark D. Spiteri - mds24@cam.ac.uk Laboratory for Communications Engineering Cambridge University Engineering Department Trumpington Street, Cambridge CB2 1PZ, United Kingdom. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Sep 13 06:35:20 1999 From: Adi Molkho Date: Mon Sep 13 06:35:24 PDT 1999 Subject: NFS on Windows NT Hello all I am working with vxWorks 5.3.1 my target is ppc8260 and my host is Windows NT Can somebody tell me what should I do to make the NFS from my target work with a drive on NT station Thanks Adi From owner-vxwexplo-process Mon Sep 13 09:31:05 1999 From: "Lawson, Kelvin" Date: Mon Sep 13 09:31:09 PDT 1999 Subject: RE: PCI support in VxWorks (MBX860) Hi, Cheers for the reply. I'm a little confused about how to choose the memory space, though. I can find the device using PciFindDevice(), however there are no base addresses in the address register. Are these chosen by me, and set in software, or can they be found dynamically ? (ie) Can I just choose my own addresses (like 0x8100 0000 for iobase and 0xC100 0000 for membase) and write them to the PCI config space ? Or do I also have to notify Vxworks somewhere, which areas of memory I am going to be using for this device ? Maybe in sysPhysMemDesc ? I have tried the following: 1. Write 0x8100 0000 in the IO base address register of PCI config space (this address is just chosen by me) using PciDevConfig. 2. Read from offsets from this number using sysInByte() I get exception number 512, which I guess is something like a sharing violation. This makes me think either: a. I have to tell Vxworks I would like to use 0x8100 0000 and up or b. This memory space is already used by another device ? How do I know where I can start my device's mem ? Another thought is that I think the PCI sees 0x0100 0000 (for example) as 0x8100 0000. Does this mean that I must write the lower number (eg 0x01..) in PCI config space for base address, and use the higher number (eg 0x81..) for accessing using the CPU ? A few questions there really, as you can see I am new to the MBX860 as well as Vxworks. It doesn't quite work the same as another popular system.... Cheers, Kelvin. > -----Original Message----- > From: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] > Sent: 13 September 1999 10:10 > To: vxworks_users@csg.lbl.gov > Subject: RE: PCI support in VxWorks (non-x86) > > To add a HW device You need only add a piece of code that looks for the > device > on the PCI bus (by VendorID, DeviceID) and map it to the desired address. > For examples look at the function sysHwInit in file sysLib.c in the BSP > directory. > > >What is the case for non-x86 ? I'm using an MBX860, and on reading PCI > >config space there is no real address information there: > > > >Base Address0 = 0x0000 0008 > >Base Address1 = 0x0000 0001 > > > >Does this mean I have to allocate the mem myself, and write it to the > >PCI config space ? > From owner-vxwexplo-process Tue Sep 14 02:28:41 1999 From: Ilia Bosis Date: Tue Sep 14 02:28:45 PDT 1999 Subject: RE: PCI support in VxWorks (MBX860) Hi, again. OK, now we're know HOW to find & map the PCI device; the question is WHERE to map it. First, look at the file target.txt in Your BSP directory. There explained the default target memory map(s). Note, that it may be a number of ones - say "PREP" vs. "Extended VME" or something else. When You know exactly what map You using, look at the "PCI MEM space" or "PCI I/O space" It's where You will map Yours device. To find what address space is already occupied by another PCI devices, look at the file mv2600.h (or something similar) in the BSP directory. Start at the definitions for some PCI device, already installed in the system, such as an Ethernet or SCSI controller, and trace the device base address down to the hex numbers. (It will be a looong chain of #defines, so be prepared for digging deep) At the end You will see a list of offsets - for each device defined, based, surprisingly :) on the mentioned "PCI MEM space" or "PCI I/O space" values. Now You have all the puzzle. Look at the unoccupied address space, or the space, reserved for a device, that will never come into Your system, and assign it. Regards Ilia +=====================================================================+ | Bosis Ilia | Work phone: 972-3-5384685 | | Software engineer | Fax: 972-3-5335877 | | Telegate ltd. | E-mail: iliab@telegate.co.il | | 60218 Haplada7 Or-Yehuda, Israel | WWW: http://www.telegate.co.il | +=====================================================================+ > -----Original Message----- > From: Lawson, Kelvin [mailto:kelvin.lawson@roke.co.uk] > Sent: Monday, September 13, 1999 6:30 PM > To: 'vxwexplo@lbl.gov'; 'iliab@telegate.co.il' > Subject: RE: PCI support in VxWorks (MBX860) > > > Hi, > > Cheers for the reply. I'm a little confused about how to > choose the memory space, though. I can find the device using > PciFindDevice(), however there are no base addresses in the > address register. Are these chosen by me, and set in > software, or can they be found dynamically ? > > (ie) Can I just choose my own addresses (like 0x8100 0000 for > iobase and 0xC100 0000 for membase) and write them to the PCI > config space ? > > Or do I also have to notify Vxworks somewhere, which areas of > memory I am going to be using for this device ? Maybe in > sysPhysMemDesc ? > > I have tried the following: > > 1. Write 0x8100 0000 in the IO base address register of PCI > config space (this address is just chosen by me) using PciDevConfig. > 2. Read from offsets from this number using sysInByte() > > I get exception number 512, which I guess is something like a > sharing violation. This makes me think either: > > a. I have to tell Vxworks I would like to use 0x8100 0000 and up or > b. This memory space is already used by another device ? How > do I know where I can start my device's mem ? > > > Another thought is that I think the PCI sees 0x0100 0000 (for > example) as 0x8100 0000. Does this mean that I must write the > lower number (eg 0x01..) in PCI config space for base > address, and use the higher number (eg 0x81..) for accessing > using the CPU ? > > A few questions there really, as you can see I am new to the > MBX860 as well as Vxworks. It doesn't quite work the same as > another popular system.... > > Cheers, > Kelvin. > From owner-vxwexplo-process Tue Sep 14 03:55:36 1999 From: Adi Molkho Date: Tue Sep 14 03:55:39 PDT 1999 Subject: NTP Client software Hi Is there anyone using vxWorks that has NTP client software or knows how can I get one Thanks Adi From owner-vxwexplo-process Tue Sep 14 04:00:10 1999 From: daemon@csg.lbl.gov Date: Tue Sep 14 04:00:14 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Sep 14 04:00:05 PDT 1999 Subject: Re: END driver SPR Subject: Happy with support and no-source? Subject: Bootloader problems Subject: Re: NFS on Windows NT Subject: How to use i386-pc-ming32 Subject: Ethernet driver Subject: Re: PCI support in VxWorks (MBX860) Subject: Re: Happy with support and no-source? Subject: PII cache [was Re: Happy with support and no-source?] Subject: Re: END driver SPR Subject: Re: PCI support in VxWorks (MBX860) Subject: delete information Subject: Re: Add a *.o file in a project Subject: Re: boot problems ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: END driver SPR Date: Mon, 13 Sep 1999 15:16:16 GMT From: Ran@127.0.0.1 (Ran Shalgi) Organization: Only when absolutely necessary Message-ID: <37dd1500.107332465@news.inter.net.il> References: Do you know if it also affects Tornado 2.0 END drivers ? On Thu, 9 Sep 1999 12:07:35 -0700, "Dave Craig" wrote: >Hi, >Just thought I'd share the guts of an SPR I got from WRS as I coudn't find >it on their web site...it 's been dogging me for about 6 months now as >scribbling/crashing/instability and seems to be rather widespread as it >affects all BSPs using END drivers and Tornado 1.0.1.... > >>There is a known problem SPR# 23229: END DMA devices do not shut >>down on reboot or in bootConfig.c "go" (aka: no ifreset in END). [snipped detailed description of the problem] Ran Shalgi Expand Networks, Inc http://www.expand.com - --- Replace 127.0.0.1 with expand.com when replying by e-mail --------------------------- Newsgroups: comp.os.vxworks Subject: Happy with support and no-source? Date: Mon, 13 Sep 1999 08:55:39 -0700 From: "a" Organization: Posted via Supernews, http://www.supernews.com Message-ID: I'm considering using VxWorks on a project and am wondering how other users like the support and the idea of not having source. It's nice to have source for the RTOS and tcp/ip stack to debug (even if the bugs are in the application) and to understand the internal workings. If bugs are found in the stack or web server etc, how is WindRiver's response to fix bugs? How do you like working with a system without having the source code? --------------------------- Newsgroups: comp.os.vxworks Subject: Bootloader problems Date: Mon, 13 Sep 1999 16:41:06 GMT From: peter_mcconaghy@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rj9ee$sbp$1@nnrp1.deja.com> Hi there. We're trying to get a MPC850 board booted up from Flash. The application is nearly all 3rd party software which we have written drivers for. As a part of their product they have modified the makefiles and provided a "bootconfig.c" file which is different. It has another boot option for booting from flash (which is what we want to do). The makefile for the application ends up with a compressed image which is ~900k compressed or 3600k uncompressed. This is meant to be loaded from address 0x10000 (ie RAM_LOW_ADRS). This all seems to happen ok - the board boots up, checks the flash application checksum is ok, decompresses it and then jumps to the start of it. From tracing what happens from there using a debugger, it seems that it goes to "kernelInit" at the end of usrInit() and dies at the end of windcontextchange (sp?? I'm back at home at the moment) - at this point it seems to stuff up the stack and then fall into a heap. The board then does a warm reboot somehow (I can't track it with the debugger because windcontextchange has an "rfi" at the end of it which the debugger doesn't like). Does anyone have any ideas on what is going wrong? Any pointers?? As it all seems to be happening within vxWorks object code it is very hard to work out what it is doing and why. Another question - when I generate a memory map for our bootrom (they use compression but the bit I am about to mention is the bit which is uncompressed into RAM) it appears to be ~420kb. If I put the start (ie RAM_HIGH_ADRS) ~500k from the top end of RAM it won't work - it needs to start ~750kb from the top. This sounds excessive for a bootrom!! It seems to include a complete copy of VxWorks - ie there is almost every object file linked in. Given that we don't want to boot from the network can we remove some of that? How big should it be? thanks! Peter Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NFS on Windows NT Date: Mon, 13 Sep 1999 16:53:32 GMT From: Ted Dennison Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rja5g$t2e$1@nnrp1.deja.com> References: <7rj0t0$r32@overload.lbl.gov> In article <7rj0t0$r32@overload.lbl.gov>, Adi Molkho wrote: > Can somebody tell me what should I do to make the NFS from my target work > with a drive on NT station NT doesn't come with any NFS support. You'll need to purchase some kind of NFS package. I understand Microsoft sells such a package for $150. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: How to use i386-pc-ming32 Date: Mon, 13 Sep 1999 18:43:09 GMT From: wsonguci@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rjgjq$270$1@nnrp1.deja.com> Hi, all I'm porting a DOS program to Vxworks 5.4 with Tornado II. The target platform is PowerPC 860. There are many DOS-related calls like bioskey, getch, kbkit (defined in conio.h) in the DOS program. I found in the Tornado directory, say 'D:\Tornado\host\x86-win32\i386-pc-mingw32', there are a whole bunch of header files (like conio.h or io.h) and libraries which seem to be compatible with those DOS calls. In order to make the porting easier, I wonder if these header files and libraries can be used in Tornado or not. Is there anybody who used header files and libraries in 'Tornado\host\x86-win32\i386-pc-mingw32' directory before? Thanks a lot. - -Song Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Ethernet driver Date: Mon, 13 Sep 1999 13:31:38 -0700 From: Bryan Wilcutt Organization: AG Communication Systems Message-ID: <37DD5F2A.3DC2C0C5@agcs.com> Reply-To: wilcuttb@agcs.com Has anyone used the GD82559 ether chip with VxWorks? How well did it hold up? How compatible was it with the 82557? Also, if anyone out there has experience with throwing in an ether chip, primarily to access WDB for debugging, what #INCLUDE's did you use in the config file? Thanks... Bryan AG Communications/Lucent Phoenix, AZ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCI support in VxWorks (MBX860) Date: Mon, 13 Sep 1999 14:53:34 -0700 From: Bill Rizzi Organization: Software Services Message-ID: <37DD725E.982097D6@softserv.com> References: <7rjbeg$q31@overload.lbl.gov> Reply-To: rizzi@softserv.com "Lawson, Kelvin" wrote: > (ie) Can I just choose my own addresses (like 0x8100 0000 for iobase and 0xC100 0000 for membase) and write them to the PCI config space ? Yes, or you can use the ones chosen for you, as long as there is no conflict with other devices. > Or do I also have to notify Vxworks somewhere, which areas of memory I am going to be using for this device ? Maybe in sysPhysMemDesc ? You need to do this regardless. You definitely need to modify sysPhysMemDesc[]. If there is nothing in there that reflects default settings for your hardware, just add and initialize however many entries you need. The scheme I've seen used is to tag the added sysPhysMemDesc[] slots with the expected/default device address values and then search for these slots as part of configuring the device when booting up the system It was instructive for me to look at the VxWorks-supplied PCI driver and diagnostic code in Tornado\target\h\drv\pci and Tornado\target\src\drv\pci. This was for a WinXX-hosted cross development environment and pc486 target (Tornado 1.0.1). Oh yeah, INCLUDE_PCI was not visible when using the GUI to configure the BSP. You'll want this defined when compiling sysLib.c; I did it manually in the BSP's config.h. > a. I have to tell Vxworks I would like to use 0x8100 0000 and up or > b. This memory space is already used by another device ? How do I know where I can start my device's mem ? If it's already in use, then you'll need to tell your device to use a different address. Keep IRQ usage in mind as well. The physical memory map is in sysPhysMemDesc[]. Also, run the PCI diagnostic - it will display all known PCU devices and their configured addresses. > Another thought is that I think the PCI sees 0x0100 0000 (for example) as 0x8100 0000. Does this mean that I must write the lower number (eg 0x01..) in PCI config space for base address, and use the higher number (eg 0x81..) for accessing using the CPU ? The low 4 bits of memory addresses are ignored as are the low 2 bits of I/O addresses. Your app should always use the unique address that your device displays in its PCI Config address registers and whose associated entry in sysPhysMemDesc[] has been properly initialized. > A few questions there really, as you can see I am new to the MBX860 as well as Vxworks. It doesn't quite work the same as another popular system.... Nope. In many respects VxWorks makes it a lot easier and faster, if a bit more "dangerous". There's some good stuff out that there that explains PCI pretty well. With VxWorks, when all else fails, just grep for "pci" (and "PCI") in system files ... Z - -- Bill Rizzi rizzi@softserv.com Software Services Z http://www.softserv.com/~rizzi Santa Barbara, CA FAX: (630) 982-5338 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Happy with support and no-source? Date: Mon, 13 Sep 1999 23:16:17 GMT From: Marjorie Krueger Organization: Vixel Message-ID: <37DD85D4.19110FB5@a.cc> References: The WindRiver support contacts I've had so far remind me of trying to get support from my ISP - I usually seem to get a response that indicates they know less than I do. I can't say that I'm happy with their support. a wrote: > > I'm considering using VxWorks on a project and am wondering how other users > like the support and the idea of not having source. It's nice to have source > for the RTOS and tcp/ip stack to debug (even if the bugs are in the > application) and to understand the internal workings. If bugs are found in > the stack or web server etc, how is WindRiver's response to fix bugs? How do > you like working with a system without having the source code? --------------------------- Newsgroups: comp.os.vxworks Subject: PII cache [was Re: Happy with support and no-source?] Date: Mon, 13 Sep 1999 18:16:53 -0700 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <37DDA205.528@despam.pangea.ca> References: <7rjn93$7gc$1@nnrp1.deja.com> Ted Dennison wrote: > We once had a nasty issue with caching on PII's. After verifying that > it was PII related (with the help of this newsgroup), the folks at > WindRiver tracked down the problem and flew an engineer out here to > apply the patch. That's not too horribly shabby. > Hi Ted, Just out of curiousity what was the problem with the PII cache? What we have run into is a very long time to initialize the MMU (IIRC the function is usrMmuInit). Was it this? By avoiding the use of WBINVD via a custom cacheArchLib and declaring less memory in sysPhysMemDesc we have reduced our boot time from ~15 to ~1.5 minutes, but it is nowhere near the T2 Pentium BSP time. - -het - -- "you don't really appreciate how smart a moron is until you try to design a robot..." -a robotics designer Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: END driver SPR Date: Mon, 13 Sep 1999 17:14:25 -0700 From: "Dave Craig" Organization: Verio Message-ID: References: <37dd1500.107332465@news.inter.net.il> The fix was put into Tornado 2 in pre-Beta. Dave. Ran Shalgi wrote in message news:37dd1500.107332465@news.inter.net.il... > > Do you know if it also affects Tornado 2.0 END drivers ? > > On Thu, 9 Sep 1999 12:07:35 -0700, "Dave Craig" > wrote: > > >Hi, > >Just thought I'd share the guts of an SPR I got from WRS as I coudn't find > >it on their web site...it 's been dogging me for about 6 months now as > >scribbling/crashing/instability and seems to be rather widespread as it > >affects all BSPs using END drivers and Tornado 1.0.1.... > > > >>There is a known problem SPR# 23229: END DMA devices do not shut > >>down on reboot or in bootConfig.c "go" (aka: no ifreset in END). > > [snipped detailed description of the problem] > > > Ran Shalgi > Expand Networks, Inc > http://www.expand.com > > --- > Replace 127.0.0.1 with expand.com when replying by e-mail > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCI support in VxWorks (MBX860) Date: Mon, 13 Sep 1999 18:42:45 -0700 From: jim Organization: We don't need no stinkin' organization! Message-ID: <37DDA815.192B@ns.net> References: <7rjbeg$q31@overload.lbl.gov> Reply-To: mailbox2@ns.net I just went through this entire exercise using a PPC MTX604 board/BSP, so I thought I'd add my remarks. Lawson, Kelvin wrote: > > Hi, > > Cheers for the reply. I'm a little confused about how to choose the memory space, > though. I can find the device using PciFindDevice(), however there are no base > addresses in the address register. Are these chosen by me, and set in software, > or can they be found dynamically ? I dumped out all existing PCI device configuration registers to see what memory and I/O space was used by other devices. There are about 6 I/O-Memory base registers, starting at config offset 0x10, for each device. Just dumping the registers gives you the starting address. Save this, write 0xFFFFFFFF, and read it back to see the size required; then restore the starting address. > (ie) Can I just choose my own addresses (like 0x8100 0000 for iobase and > 0xC100 0000 for membase) and write them to the PCI config space ? Yes, but, as I'll describe later, 0x0100 0000 is more like it. As long as you don't conflict with another device, you're OK. > Or do I also have to notify Vxworks somewhere, which areas of memory I > am going to be using for this device ? Maybe in sysPhysMemDesc ? In the MTX604 BSP, the hardware is initialized to support a rather large PCI I/O and Memory mapping area (like 8 - 16MB of address space. This large area ia already set up in sysPhysMemDesc, so no changes were necessary. YMMV. > > I have tried the following: > > 1. Write 0x8100 0000 in the IO base address register of PCI config space > (this address is just chosen by me) using PciDevConfig. > 2. Read from offsets from this number using sysInByte() Very close, but no cigar (at least in my own case). This is exactly what I tried on my first go-round. First, at least on my BSP, the "I/O space" on the PCI bus is mapped to memory space on the CPU bus. Therefore, reading the I/O addresses is done with normal C-language pointers: ULONG * ptr = 0x8100 0000; ULONG data; data = ptr[offset in 32-bit words]; gets you the PCI I/O space register at the indicated offset. Next, again on my BSP, the 0x8100 0000 "address" is on the CPU/host bus side of the PCI bridge. The PCI bridge does some translation such that the address on the PCI-bus side is 0x0100 0000. The PCI-bus side is what you program into the PCI configuration register IObase for your device. When you examine the config registers in other existing devices, you should see this in practice. > I get exception number 512, which I guess is something like a sharing violation. > This makes me think either: You probably got this because your board also maps PCI I/O space to CPU memory space. The I/O address in your sysInByte was non-existent. > a. I have to tell Vxworks I would like to use 0x8100 0000 and up or > b. This memory space is already used by another device ? How do I > know where I can start my device's mem ? Poke around, guess, pray, :) or see the tips above. > Another thought is that I think the PCI sees 0x0100 0000 (for example) > as 0x8100 0000. Does this mean that I must write the lower number (eg 0x01..) > in PCI config space for base address, and use the higher number (eg 0x81..) > for accessing using the CPU ? Yes, indeed. Made this mistake myself on PCI accesses. Also, beware, the same kind of problem applies to DMA addresses. Some bridge chips will also offset the addresses between the two busses when going in the other direction (PCI device puts address on PCI bus, then writes DMA data which goes to CPU bus-based RAM). IMHO, a rather short treatise on this subject should be part of every BSP. >From what I can see, however, the VxWorks BSP documentation writers are about as illiterate as their support people, so we have to figure it out for ourselves. Hope this helps. Jim. Note: reply to: jpsmith@ns.net rather than the mailbox2@ns.net --------------------------- Newsgroups: comp.os.vxworks Subject: delete information Date: Tue, 14 Sep 1999 10:40:25 +0800 From: "Ç®É궰" Organization: Alcatel Message-ID: <37ddb792.0@dnews.sbell.com.cn> Hi all: When I run a command in the Tornado Shell, it always display the information like below: "value = 0 = 0x0". Could anyone have the good ways to remove that information? Q.S.D --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Add a *.o file in a project Date: Mon, 13 Sep 1999 22:06:40 -0700 From: Pierre Frigon Organization: Wind River Systems Message-ID: <37DDD7E0.2918@me.com> References: <37B17DF0.194B3132@gespac.ch> Go to the build spec of your project and set the macro EXTRA_MODULES to point to your .o Gilles Bannay wrote: > > Hi, > > I use TornadoII and I'd like to add a *.o object file > (a subroutine compiled with another TornadoII host) > in my vxWorks image project. > When I use the 'Add files...' menu, the *.o is ignored ! > Is there a solution ? > > Thank you > Gilles. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: boot problems Date: Mon, 13 Sep 1999 21:51:11 -0700 From: Pierre Frigon Organization: Wind River Systems Message-ID: <7rkh07$hl1@tass.wrs.com> References: <01beddca$45d56da0$1ed7a089@da0713.dasd.honeywell.com> <37A7B852.5E666942@ntanet.net> <01bede8c$0aac3c00$1ed7a089@da0713.dasd.honeywell.com> Assuming your are using a donwloaded image (loaded over the network) and you changed RAM_HIGH_ADRS, you have to re-build the boot program and re-program the boot PROM. This boot program actually loads the image from the host into memory so it needs to know where to put it and where to put itself. Hope this helps James Bernsen wrote: > > Yes, we recompiled everything after changing the RAM_HIGH_ADRS value. The > exception error still appeared. > > Any other ideas are appreciated! > > -James Bernsen > > David Hamel wrote in article > <37A7B852.5E666942@ntanet.net>... > > Did you create a new bootrom that uses the different RAM_HIGH_ADRS? > > --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Sep 14 04:49:31 1999 From: "Subramania Pillai" Date: Tue Sep 14 04:49:35 PDT 1999 Subject: netBufLib - functions? VxWorks Hi, It looks the library function provided with netBufLib lacks facility for network protocol programming. For example there is function to append or insert data to a cluster that has some protocol packet. Is there any other alternative? Thanks Pillai From owner-vxwexplo-process Tue Sep 14 14:27:15 1999 From: Macchio1@aol.com Date: Tue Sep 14 14:27:19 PDT 1999 Subject: Software Engineers, MPLS Protocol "Job Posting" Software Engineers, MPLS Protocol We are looking for a key contributor with experience in implementing label switching protocols, especially MPLS for carrier class telecommunications network elements. Duties/Responsibilities · You will implement MPLS module for a carrier class Edge Switching Systems. · Design and specify mechanisms to utilize the power of Companies IP switching fabric with CAM lookup. · Specify, design, implement and test label switching protocol. · Work closely with the embedded software team to integrate the switching modules with Companies RTOS, CLI and management agents. · Work closely with our customers to specify and implement routing and tunneling policies for emerging IP services Qualifications · 3 - 5 years of relevant software development experience, preferably using VxWorks. Position is located in CA. Salary is open, also this is a pre-ipo company with major stock incentives call or contact Vito at 305-937-7364 or vito@sourcegroupinc.com From owner-vxwexplo-process Tue Sep 14 14:29:56 1999 From: Macchio1@aol.com Date: Tue Sep 14 14:29:59 PDT 1999 Subject: Software Engineers, IP Routing Protocols "Job Posting" Software Engineers, IP Routing Protocols We are looking for key contributors with experience in implementing IP routing protocols, especially IS-IS and BGP4 for Carrier class telecommunications network elements. Duties/Responsibilities · You will implement IS-IS and BGP4 routing for carrier class Edge Switching Systems. · Design and specify mechanisms to utilize the power of Company's IP switching fabric with CAM lookup. · Specify, design, implement and test routing protocol. · Work closely with the embedded software team to integrate the routing modules with Companies RTOS, CLI and management agents. · Work closely with our customers to specify and implement routing policies for emerging IP services. Qualifications · 3 - 5 years of relevant software development experience, preferably using VxWorks. · Experience in debugging routing software, especially interoperability with major routers. · In depth knowledge of issues and solutions in Internet routing. · Strong C and Assembly skills and familiarity with object oriented design is required. · Effective communication skills, flexibility, a strong teamwork approach, and the desire to work in a fast-paced start up environment are necessary for success. Education BSEE/CS or equivalent experience is required Position is located in CA. Salary is open, also this is a pre-IPO company with major stock incentives call or contact Vito at 305-937-7364 or vito@sourcegroupinc.com From owner-vxwexplo-process Wed Sep 15 04:00:14 1999 From: daemon@csg.lbl.gov Date: Wed Sep 15 04:00:18 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Sep 15 04:00:10 PDT 1999 Subject: Serial ports on MBX860 board / PPC860 Subject: Tornado 1.0.1 editor is causing dropped network shares Subject: Re: Bootloader problems Subject: Bootloader problems Subject: Re: PII cache [was Re: Happy with support and no-source?] Subject: comparisons with other OS's Subject: Re: CORBA on VxWorks ? Subject: Task Execution Time. Subject: Re: Task Execution Time. Subject: ISA Plug & Play Subject: Re: Personal Java and Tornado 2 Subject: Re: Task Execution Time. Subject: Re: Task Execution Time. Subject: Device Driver Docs Subject: Re: comparisons with other OS's Subject: etherLib Subject: Tornado 2 DHCP server Subject: Software Engineers, IP Routing Protocols "Job Posting" Subject: Interrupts on Pentium target (especially COM1 and COM2) Subject: Daylight saving time standard??? Subject: Instr translation miss exceptions Subject: VME bus interrupts - MVME2700 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Serial ports on MBX860 board / PPC860 Date: 14 Sep 1999 11:26:13 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <7rlbcl$csq$1@news1.xs4all.nl> Hello, I am having problems getting the SMC2 port to work on an MBX860 board. I used the default sysSerial, and changed the references from SMC1 to SMC2, but now the system crashes. Did anyone manage to get this port to work? I would like to use SMC2, beacause it is not possible to use COM1 and SMC1 at the same time, because of HW problems. The code I use is: /* intialize the chips device descriptors */ ppc860Chan.clockRate = brgClkFreq; /* BRGCLK freq (Hz) */ ppc860Chan.regBase = vxImmrGet(); /* IMMR reg has base adr */ ppc860Chan.bgrNum = 1; /* use BRG1 */ >> ppc860Chan.uart.smcNum = 2; /* SMC1 wired for rs232 */ ppc860Chan.uart.txBdNum = 1; /* use 1 transmit BD */ ppc860Chan.uart.rxBdNum = 0x10; /* use 16 receive BD */ /* setup the transmit buffer descriptor base addr */ ppc860Chan.uart.txBdBase = (SMC_BUF *) TX_BUF_DESC_BA; /* setup the receive buffer descriptor base addr */ ppc860Chan.uart.rxBdBase = (SMC_BUF *) RX_BUF_DESC_BA; ppc860Chan.uart.txBufBase = (u_char *) (MPC860_DPRAM_BASE(ppc860Chan.regBase) + TX_BUFFER); /* tx buf base */ ppc860Chan.uart.rxBufBase = (u_char *) (MPC860_DPRAM_BASE(ppc860Chan.regBase) + RX_BUFFER); /* rx buf base */ /* WE MAY WANT TO INCREASE THE BUFFER SIZES. */ ppc860Chan.uart.txBufSize = 0x1; /* transmit buffer size */ >> ppc860Chan.uart.pSmc = (SMC *) ((UINT32) >> PPC860_DPR_SMC2( MPC860_DPRAM_BASE(ppc860Chan.regBase) )); /* SMC2 DPRAM addr params */ >> ppc860Chan.uart.pSmcReg = >> (SMC_REG *) ((UINT32) MPC860_SMCMR2(ppc860Chan.regBase)); /* SMCMR2 for SMC2 */ /* configure the port lines */ >> *MPC860_PBPAR(ppc860Chan.regBase) |= PORT_B_PINS_SMC2; >> *MPC860_PBDIR(ppc860Chan.regBase) &= ~PORT_B_PINS_SMC2; >> *MPC860_PBODR(ppc860Chan.regBase) &= ~PORT_B_PINS_SMC2; ppc860Chan.pBaud = (UINT32 *) MPC860_BRGC1(ppc860Chan.regBase); ppc860Chan.channelMode = 0; *MPC860_SDCR(ppc860Chan.regBase) = 1; /* */ /* reset the chip */ ppc860DevInit(&ppc860Chan); Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 1.0.1 editor is causing dropped network shares Date: Tue, 14 Sep 1999 13:11:20 +0100 From: "Mark Rawlings" Organization: Eircom Net (http://www.eircom.net/) Message-ID: <7rle1e$r6s$1@scotty.tinet.ie> This is a multi-part message in MIME format. - ------=_NextPart_000_000E_01BEFEB2.A345B180 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Please excuse me if that has been answered before but I can't find the = answer. We are running Tornado 1.0.1 with most of our source files being found = on network drives. What we are discovering is that periodically the machines that are = serving the network drives are having problems. Specifically it appears that any shares from = that machine=20 are lost. The only way out of it we have at the moment is to stop and restart all = network services on the affected machine. It appears to be the Tornado editor that is the culprit. Has anyone seen anythnig like this before or know of any solution other = than stopping/restarting the network services on the affected drivers? Thanks, Mark. - ------=_NextPart_000_000E_01BEFEB2.A345B180 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Please excuse me if that has been = answered=20 before but I can't find the answer.
 
We are running Tornado 1.0.1 with = most of our=20 source files being found on network drives.
What we are discovering is that = periodically the=20 machines that are serving the network
drives are having problems. = Specifically it=20 appears that any shares from that machine
are lost.
The only way out of it we have at = the moment is=20 to stop and restart all network services on the
affected machine.
 
It appears to be the Tornado editor = that is the=20 culprit.
 
Has anyone seen anythnig like this = before or=20 know of any solution other than stopping/restarting
the network services on the affected = drivers?
 
    = Thanks,
    =    =20     Mark.


 

- ------=_NextPart_000_000E_01BEFEB2.A345B180-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bootloader problems Date: Tue, 14 Sep 1999 12:16:18 GMT From: peter_mcconaghy@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rleaa$e52$1@nnrp1.deja.com> References: <7rj9ee$sbp$1@nnrp1.deja.com> In article <7rj9ee$sbp$1@nnrp1.deja.com>, peter_mcconaghy@my-deja.com wrote: > Hi there. > Found it after hours and hours and hours of stepping through assembler - the function usrMmuInit() has an error when we call it and causes the board to do a warm reboot. No idea what the problem is at the moment so we just disabled it and now it works!! Yeh - another problem solved... In answer to the person who asked about being happy with not having the source code - I'm not overjoyed. It's really hard to debug assembler when you have no idea what is going on... Turned out that the hang on "rfi" was due to the debugger not being able to single step through the start/end of an IRQ routine because they do it by placing invalid instructions in RAM which causes an IRQ which stuffs up the start/end of the current IRQ (this is SDS SingleStep7.5). regards Peter Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Bootloader problems Date: Tue, 14 Sep 1999 09:51:56 +0200 From: "Borkhuis, Johan" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <7rj9ee$sbp$1@nnrp1.deja.com> > ---------- > From: > peter_mcconaghy@my-deja.com[SMTP:peter_mcconaghy@my-deja.com] > Posted At: maandag 13 september 1999 18:41 > Posted To: vxworks > Conversation: Bootloader problems > Subject: Bootloader problems > > Another question - when I generate a memory map for our bootrom (they > use compression but the bit I am about to mention is the bit which is > uncompressed into RAM) it appears to be ~420kb. If I put the start (ie > RAM_HIGH_ADRS) ~500k from the top end of RAM it won't work - it needs > to start ~750kb from the top. This sounds excessive for a bootrom!! It > seems to include a complete copy of VxWorks - ie there is almost every > object file linked in. Given that we don't want to boot from the > network can we remove some of that? How big should it be? > Yes, you can remove that code. In fact you can strip everything that you don't need to boot. So if you don't boot from the network delete the network code. If you don't use a flash disk skip that too, etc. Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PII cache [was Re: Happy with support and no-source?] Date: Tue, 14 Sep 1999 14:51:34 GMT From: Ted Dennison Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rlndh$kv9$1@nnrp1.deja.com> References: <7rjn93$7gc$1@nnrp1.deja.com> <37DDA205.528@despam.pangea.ca> In article <37DDA205.528@despam.pangea.ca>, Harvey Taylor wrote: > Ted Dennison wrote: > > We once had a nasty issue with caching on PII's. After verifying that > Just out of curiousity what was the problem with the PII cache? Supposedly the problem ended up being that a cache invalidate instruction that was getting called at every task switch takes hundreds of times more cycles to execute on a PII than it does a Pentium. The effet ended up being that our code ran faster on a Pentium than on a PII! With several 60Hz tasks it was fairly noticable, but on our 2KHz system the effect was quite striking. I haven't looked at their patch code in detail, but it was a new version of cacheArchLib.c that they gave us. > By avoiding the use of WBINVD via a custom cacheArchLib and > declaring less memory in sysPhysMemDesc we have reduced our > boot time from ~15 to ~1.5 minutes, but it is nowhere near > the T2 Pentium BSP time. We haven't seen that one. It takes about 1 minute to boot our systems, but over half of that is PC BIOS initilizations. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: comparisons with other OS's Date: Tue, 14 Sep 1999 11:07:37 -0400 From: s Organization: EarthLink Network, Inc. Message-ID: Anyone out there use VxWorks and some other OS's, specifically OSE from Enea? I am interested in finding out if others have found a decrease in development time when using something like OSE over VxWorks. This would include things like amount of process creation code and the number of system calls that needed to be made to create, maintain, delete a process and all of its related needs (memory etc). I am even interested in what other OS's current VxWorks users looked at or what OS's others chose over VxWorks. Thanks for you time --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CORBA on VxWorks ? Date: Tue, 14 Sep 1999 17:02:18 +0200 From: "Holger Roos" Organization: Daimler-Benz Aerospace Message-ID: <7rlo02$jlu@newsserv.vs.dasa.de> References: <7r8v77$s6v$1@feed.teaser.fr> I'm just starting with VisiBroker (Highlander / Inprise) ... Bertrand Marlier schrieb in Nachricht <7r8v77$s6v$1@feed.teaser.fr>... >Does anyone have an experience of CORBA on VxWorks ? > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Task Execution Time. Date: Tue, 14 Sep 1999 18:03:12 +0200 From: "Massimo Maio" Organization: Alenia Difesa Message-ID: <7rls2n$kgi$1@nslave1.tin.it> Hi VxWorkers ! I have a task that looks like the following one: STATUS myTask (void) { /*Some initialization operations.*/ FOREVER { semTake(synchronizationSemaphore); /*First Instruction after semTake.*/ /*Some time-consuming operations.*/ /*Last Istruction of the FOREVER cycle.*/ } return (OK); } I need to evaluate the ticks (or the milliseconds) elapsed between the time the task begins to execute the first instruction after semTake, and the time at which it has completed to execute the last instruction of the FOREVER cycle. Can you help me ? Please, e-mail me your ideas at the following address: m.maio@dsae.finmeccanica.it Thanks in advance. Best regards, Massimo Maio. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Task Execution Time. Date: Tue, 14 Sep 1999 18:51:33 +0200 From: "Christian Doppelbauer" Organization: VBS-Newsserver Message-ID: <937327987.364035@news.vbs.at> References: <7rls2n$kgi$1@nslave1.tin.it> Massimo Maio wrote in message news:7rls2n$kgi$1@nslave1.tin.it... > Hi VxWorkers ! > > I have a task that looks like the following one: > > STATUS myTask (void) > { > /*Some initialization operations.*/ > > FOREVER > { > semTake(synchronizationSemaphore); > > /*First Instruction after semTake.*/ > > /*Some time-consuming operations.*/ > > /*Last Istruction of the FOREVER cycle.*/ > > } > > return (OK); > } > > I need to evaluate the ticks (or the milliseconds) elapsed between the time > the task begins to execute the first instruction after semTake, and the time > at which it has completed to execute the last instruction of the FOREVER > cycle. > Can you help me ? > Please, e-mail me your ideas at the following address: > m.maio@dsae.finmeccanica.it > > Thanks in advance. Best regards, Massimo Maio. > > > Simply use the tickGet() call from tickLib. You can adjust the ticks with the sysClkRateSet() call. If you are using a x86 pentium bsp you could use the pentiumTscGet32() call that gives you the ticks of the pentium, that's my favorite call. Don't know if there is something compareable on other bsp's. Ciao, Christian --------------------------- Newsgroups: comp.os.vxworks Subject: ISA Plug & Play Date: Tue, 14 Sep 1999 19:00:30 +0200 From: "Christian Doppelbauer" Organization: VBS-Newsserver Message-ID: <937328525.47044@news.vbs.at> Hi all, after reading the isa pnp specification it is clear to me i need to ask if anyone has experience ( source code ;-) ) for a simple utility that can display all isa plug & play cards in a pc. Something simular to the pciFindxxx und pciConfigxxx functions, thats what im looking for. I am not the first working on this boring stuff, am i? Thanks in advance, Christian --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Personal Java and Tornado 2 Date: Tue, 14 Sep 1999 10:36:27 +0100 From: "Wade Oram" Organization: Marconi Instruments Ltd., Stevenage, Herts., UK Message-ID: <7rl4uo$ige@ipgw.marconi-instruments.co.uk> References: <7rk2hv$ftf$1@nnrp1.deja.com> > I have gone into Tornado and set it up to run with the Java subset I > will be using and turned off all network modules. When I select "Build > All", it goes off and compiles, failing in the link stage. The linker > reports that clDescTblNumExt, clDescTbl, mClBlkConfig, > sysClDescTblNumEnt, sysClDescTbl, and sysMclBlkConfig are undefined > symbol names. > > These all seem to be network symbols, but I can't see where they are > coming from. > Hello, I have had the same problem and contacted Wind River Technical Support. They have given me a workaround which is simply to include 'Network Buffer Initialisation'. This is nothing to do with the use of Java but is simply a consequence of removing all network support. - -- Wade Oram --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Task Execution Time. Date: 14 Sep 1999 11:39:54 PDT From: "Dinesh Nambisan" Organization: Concentric Internet Services Message-ID: <7rm4pq$dn2@journal.concentric.net> References: <7rls2n$kgi$1@nslave1.tin.it> Use WindView if possible, there are custom user-defined events one can stick in within your task which will pop up as flags/markers within the WindView monitor application; and by using the slider you can estimate the exact CPU time (resolution very high) your task is soaking up... Hope this helps, cheers, Dinesh N. Massimo Maio wrote in message news:7rls2n$kgi$1@nslave1.tin.it... > Hi VxWorkers ! > > I have a task that looks like the following one: > > STATUS myTask (void) > { > /*Some initialization operations.*/ > > FOREVER > { > semTake(synchronizationSemaphore); > > /*First Instruction after semTake.*/ > > /*Some time-consuming operations.*/ > > /*Last Istruction of the FOREVER cycle.*/ > > } > > return (OK); > } > > I need to evaluate the ticks (or the milliseconds) elapsed between the time > the task begins to execute the first instruction after semTake, and the time > at which it has completed to execute the last instruction of the FOREVER > cycle. > Can you help me ? > Please, e-mail me your ideas at the following address: > m.maio@dsae.finmeccanica.it > > Thanks in advance. Best regards, Massimo Maio. > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Task Execution Time. Date: Tue, 14 Sep 1999 13:39:44 -0500 From: Marty Schrader Organization: Fermilab Experimental Astrophysics Group Message-ID: <37DE9670.FC17D4B8@fnal.gov> References: <7rls2n$kgi$1@nslave1.tin.it> Don't forget that, like any system being measured, the act of measuring the execution time on your target will affect the overall time of the loop. The insertion of an execution profiler into your code will affect its operation by the amount of time the profiler is running. As long as that is cool for you, go record clock ticks. - -- Marty Schrader x6645 Fermilab Experimental Astrophysics Group (SDSS) consult0@fnal.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Device Driver Docs Date: Tue, 14 Sep 1999 12:16:33 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <37DE9F11.5526@gv.net> Reply-To: mkesti@gv.net As has been proviously noted in this forum, the device driver docs in section 3.9 of the VxWorks Programmer's Guide are somewhat sketchy, at best. Where can I obtain better docs without resorting to WRS's device driver seminar? - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: comparisons with other OS's Date: Tue, 14 Sep 1999 19:52:44 GMT From: Ted Dennison Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rm928$1h0$1@nnrp1.deja.com> References: In article , s wrote: > This would include things like amount of process creation code and the > number of system calls that needed to be made to create, maintain, > delete a process and all of its related needs (memory etc). Perhaps you are attacking the problem at the wrong level. If you use Ada, process creation code is minimal and you don't have to make *any* system calls to do it. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: etherLib Date: Tue, 14 Sep 1999 16:06:25 -0400 From: "Chuck Woodring" Organization: Naval Research Laboratory, Washington, DC Message-ID: <7rm9je$h1b$1@ra.nrl.navy.mil> Hi, I was wondering if under vxWorks 5.2, is it possible to usethe etherHookAdd function to access and store broadcast UDP packets? I've tried writing a routine to call this and then a packet handling routine but even though etherHookAdd returns OK, the handling function is never called? Do these calls have to be done in the ethernet driver code, i.e. ifln? Thanks. chuck --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 2 DHCP server Date: Tue, 14 Sep 1999 20:10:20 GMT From: js_j98@yahoo.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rma35$2al$1@nnrp1.deja.com> Hi Did anyone set up the DHCP server from Tornado 2? I could not get the server to reply (they did not provide anything for debugging on the server side either). I have 2 vxworks target, one runs Bootrom bootp client request. And the other target has dhcp server configured. I seen the bootp request sent out but seen no replies from the server. I have the following entries in the LeaseTbl {"ent1", "10.2.2.24", "10.2.2.24", "albp=true:clid=\"1:0x08003E030203\":siad=10.2.2.20:file=vxworks"}, Any help is welcome. Below is the console log. - ------------------------------ Attaching interface lo0...done Warning: No DHCP server address cache! Later entries will not be saved. DHCP server started. VxWorks Copyright 1984-1998 Wind River Systems, Inc. CPU: Motorola ADS - PowerPC 860 VxWorks: 5.4 BSP version: 1.2/4 Creation date: Sep 13 1999 WDB: Ready. task: 3739968: started 0x3ffdf8 (): task deaddhcps: read 4 entries from addr-pool database. 0x3ffdf8 (): task deaddhcps: read 1 entries from binding and addr-pool database. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Software Engineers, IP Routing Protocols "Job Posting" Date: 14 Sep 1999 21:22:27 GMT From: macchio1@aol.com (Macchio1) Organization: AOL http://www.aol.com Message-ID: <19990914172227.02771.00000223@ng-cf1.aol.com> Software Engineers, IP Routing Protocols We are looking for key contributors with experience in implementing IP routing protocols, especially IS-IS and BGP4 for Carrier class telecommunications network elements. Duties/Responsibilities · You will implement IS-IS and BGP4 routing for carrier class Edge Switching Systems. · Design and specify mechanisms to utilize the power of Company's IP switching fabric with CAM lookup. · Specify, design, implement and test routing protocol. · Work closely with the embedded software team to integrate the routing modules with Companies RTOS, CLI and management agents. · Work closely with our customers to specify and implement routing policies for emerging . IP services. Qualifications · 3 - 5 years of relevant software development experience, preferably using VxWorks. · Experience in debugging routing software, especially interoperability with major routers. · In depth knowledge of issues and solutions in Internet routing. · Strong C and Assembly skills and familiarity with object oriented design is required. · Effective communication skills, flexibility, a strong teamwork approach, and the desire to work in a fast-paced start up environment are necessary for success. Education BSEE/CS or equivalent experience is required Position is located in CA. Salary is open, also this is a pre-ipo company with major stock incentives call or contact Vito Macchio at 305-937-7364 or vito@sourcegroupinc.com --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupts on Pentium target (especially COM1 and COM2) Date: 15 Sep 1999 02:59:33 GMT From: jolkeybol@aol.com (Jolkeybol) Organization: AOL http://www.aol.com Message-ID: <19990914225933.01931.00000359@ng-fz1.aol.com> Current system is pentium target, sun/solaris host. VxWorks 5.3.1 for PC486, WIND version 2.5. The target has a National Instruments GPIB card and a B&B 4-port serial card. With the 2 on-board serial ports plus the four B&B ports, there are a total of six serial ports. I have set the jumpers (and/or BIOS) as follows: COM1 - on-board - 0x3f8 - IRQ 4 COM2 - on-board - 0x2f8 - IRQ 3 COM3 - B&B - 0x3e8 - IRQ 5 COM4 - B&B - 0x2e8 - IRQ 7 COM5 - B&B - 0x238 - IRQ 12 COM6 - B&B - 0x248 - IRQ 15 GPIB Board - IRQ 11 I can serial poll all ports, so I know the base addresses are correct. I can also use the GPIB interface. However, polling is slow, so I am trying to get interrupts to work. I can only get IRQ 5, 7 and 15 to work on their respective COM ports; for these ports my ISR gets activated and I am able to write commands and read data after the interrupt gets executed. IRQ 11 works when used for a COM port, but not as SRQ for GPIB. I think the BIOS is set correctly for all IRQs. Particularly, I can't understand why COM1 and COM2 can't use their interrupts. I have called the motherboard vendor, but they don't have any suggestions, except to tailor the BIOS for us. But what would I have it tailored to? They are using AWARD Modular BIOS 4.51PG. The GPIB card is acting funny, too. As soon as I activate the ISR for its SRQ via intConnect and intEnablePIC, the ibcmd and ibwrt routines no longer work. They just bomb or get in some wait state. Ultimately, I would like to have all COM ports on the given interrupts, and also be able to use the GPIB SRQ feature, however, my most immediate need is to get COM1 and COM2 working with my ISR. Any suggestions? Could this be a vxWorks problem, or is it a BIOS problem? Thank you, JK --------------------------- Newsgroups: comp.os.vxworks Subject: Daylight saving time standard??? Date: Wed, 15 Sep 1999 03:37:09 GMT From: "Jonathan Cheng" Organization: @Home Network Canada Message-ID: Hi, Just wonder if anyone know the standard of Daylight saving time in United States? From the following Web site: http://www.treasure-troves.com/astro/DaylightSavingTime.html Daylight savings time ends when clocks are set from 3:00 a.m. to 2:00 a.m. on last Sunday in October. In VxWorks, DST ends when clocks are set from 2:00 a.m. to 1:00 a.m. on last Sunday in October. ( If the TIMEZONE environment is set to UTC::0:040402:103102 ) Same as Window NT and Window 95/98. Seems to me that both Windows and VxWorks does not implement the DST correctly. Certainly, we can change this by editing the TIMEZONE environment. I just wonder if I have missed something. - -Jonathan Cheng --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Instr translation miss exceptions Date: Tue, 14 Sep 1999 23:03:57 -0500 From: petekz@airmail.net (Pete Kockritz) Organization: Airnews.net! at Internet America Message-ID: <86CE30EB9755CE6B.CEED9D9FE2C2E68D.972C9CF4B38FBA2E@lp.airnews.net> We're using vxWorks on a PPC860. On a kernel that does not have the caches or MMU enabled, we're getting instruction translation miss exceptions. I would've thought that was not possible. What does it mean when you get this exception with MMU disabled? Regards, Pete - -- +-----------------------------------------------------------------+ | Pete Kockritz mailto:petekz@airmail.net | | This space for rent. | +-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: VME bus interrupts - MVME2700 Date: Wed, 15 Sep 1999 09:22:40 GMT From: paulvroberts@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rnogp$2of$1@nnrp1.deja.com> Hi there, I am having a problem with VME interupts on the MVME2700. I am getting the error "bad vme interrupt 0" This problem occurs both on a slightly older MVME2700-1241 (233MHz) and on a newer MVME2700-1441 (366Mhz). The problem seems worse on the 366Mhz board. The problems occur when my ISR is short - adding a delay seems to fix it under certain conditions. The IRQ is originating from a digital i/o IP module (Acromag IP470) that is on an IP->VME Carrier board (Greenspring). From reading the data sheet for the IP module and checking the IRQ/IACK waveforms, it looks like the board uses the RORA form of VME interrupt (ie you need to write to the board to clear the interrupt). When I get the error message, tt looks like the processor sees a second interrupt, becuase it is assering another IACK. In some cases, this appears to cause in a timeout, in other cases, the ISR is run again. I discovered that if I put a particular delay in the ISR, then I did not get the problem at 1Hz interrupts. Strangely, it didn't seem to matter where the delay occurs, which makes me think the problem is not the board (if it was a race condition because of the board, I would have expected it to be required after the clearing of the IRQ). However, when I tested at 10Hz and 1KHz, further problems arose including IRQ never being cleared and kernel panics. Has anyone seen anything like this before or give me **any** clues to further track this down, Thanks, Paul Roberts. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Wed Sep 15 07:47:17 1999 From: "Slutsker, Rasty" Date: Wed Sep 15 07:47:21 PDT 1999 Subject: RE: Interrupts on Pentium target (especially COM1 and COM2) 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_01BEFF88.FAFDE632 Content-Type: text/plain; charset="iso-8859-1" Hi, I assume you're going to replace original serial driver. As far as I remember, it is not enough to just enable corresponding IRQ line, PC has a funny "OUT2" line, which is controlled by configuration registers of COM1&COM2, and should be set to "1" in order to enable interrupts from serial port. Look in the original driver (i8250sio.c) for OUT2. Also you have to enable interrupt generation in UART. Regards, Rasty Newsgroups: comp.os.vxworks Subject: Interrupts on Pentium target (especially COM1 and COM2) Date: 15 Sep 1999 02:59:33 GMT From: jolkeybol@aol.com (Jolkeybol) Organization: AOL http://www.aol.com Message-ID: <19990914225933.01931.00000359@ng-fz1.aol.com > Current system is pentium target, sun/solaris host. VxWorks 5.3.1 for PC486, WIND version 2.5. The target has a National Instruments GPIB card and a B&B 4-port serial card. With the 2 on-board serial ports plus the four B&B ports, there are a total of six serial ports. I have set the jumpers (and/or BIOS) as follows: COM1 - on-board - 0x3f8 - IRQ 4 COM2 - on-board - 0x2f8 - IRQ 3 COM3 - B&B - 0x3e8 - IRQ 5 COM4 - B&B - 0x2e8 - IRQ 7 COM5 - B&B - 0x238 - IRQ 12 COM6 - B&B - 0x248 - IRQ 15 GPIB Board - IRQ 11 I can serial poll all ports, so I know the base addresses are correct. I can also use the GPIB interface. However, polling is slow, so I am trying to get interrupts to work. I can only get IRQ 5, 7 and 15 to work on their respective COM ports; for these ports my ISR gets activated and I am able to write commands and read data after the interrupt gets executed. IRQ 11 works when used for a COM port, but not as SRQ for GPIB. I think the BIOS is set correctly for all IRQs. Particularly, I can't understand why COM1 and COM2 can't use their interrupts. I have called the motherboard vendor, but they don't have any suggestions, except to tailor the BIOS for us. But what would I have it tailored to? They are using AWARD Modular BIOS 4.51PG. The GPIB card is acting funny, too. As soon as I activate the ISR for its SRQ via intConnect and intEnablePIC, the ibcmd and ibwrt routines no longer work. They just bomb or get in some wait state. Ultimately, I would like to have all COM ports on the given interrupts, and also be able to use the GPIB SRQ feature, however, my most immediate need is to get COM1 and COM2 working with my ISR. Any suggestions? Could this be a vxWorks problem, or is it a BIOS problem? Thank you, JK ------_=_NextPart_001_01BEFF88.FAFDE632 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable RE: Interrupts on Pentium target (especially COM1 and = COM2)

Hi,

I assume you're going = to replace original serial = driver.

As far as I remember, it is not enough = to just enable corresponding IRQ line, = PC has a funny "OUT2" = line, which is controlled by configuration registers of COM1&COM2, and = should be set to = "1" in order = to enable interrupts from serial port. Look in the = original driver (i8250sio.c) for = OUT2. Also you have to enable interrupt = generation in UART.

Regards,

Rasty

Newsgroups: = comp.os.vxworks

Subject:       = Interrupts on Pentium target (especially = COM1 and COM2)

Date:   = 15 Sep 1999 02:59:33 GMT

From:   jolkeybol@aol.com (Jolkeybol)

        Organization: AOL http://www.aol.com

        Message-ID: <19990914225933.01931.00000359@ng-fz1.aol.com>

Current system is pentium = target, sun/solaris host.  VxWorks 5.3.1 for PC486, WIND = version 2.5.  The target has = a National Instruments GPIB card and a B&B 4-port serial = card.  With the 2 on-board serial ports plus the four B&B = ports, there are a total of six serial ports.  I have set = the jumpers (and/or BIOS) as follows:

COM1 - on-board  - = 0x3f8 - IRQ 4

COM2 - on-board  - = 0x2f8 - IRQ 3

COM3 - = B&B         - 0x3e8 - IRQ = 5

COM4 - = B&B         - 0x2e8 - IRQ = 7

COM5 - = B&B         - 0x238 - IRQ = 12

COM6 - = B&B         - 0x248 - IRQ = 15

GPIB = Board           &= nbsp;          - IRQ = 11

I can serial poll all = ports, so I know the base = addresses are correct.  I can also use the GPIB = interface.  However, polling is slow, so I am trying to = get interrupts to work.

I can only get IRQ 5, 7 = and 15 to work on their respective COM ports; for these ports my = ISR gets activated and I am able to write commands and read data = after the interrupt gets executed.  IRQ 11 works when = used for a COM  port, but not as SRQ for GPIB. I = think the BIOS is set correctly for all  IRQs.  Particularly, I can't = understand why COM1 and COM2 = can't use their interrupts.  I have called the = motherboard vendor, but they don't have any = suggestions, except to tailor the BIOS for us.  But what would I = have it tailored to?  They are using AWARD = Modular BIOS 4.51PG.

The GPIB card = is acting funny, too. As soon as I = activate the ISR for its SRQ via intConnect and = intEnablePIC, the ibcmd and ibwrt routines no longer = work.  They just bomb or get in some wait state.

Ultimately, I would like = to have all COM ports on the given interrupts, and also be able to use the GPIB SRQ feature, = however, my most immediate need is to get COM1 and COM2 = working with my ISR.

Any suggestions?  = Could this be a vxWorks problem, or is it a BIOS problem?

                        Thank you,

                        JK

------_=_NextPart_001_01BEFF88.FAFDE632-- From owner-vxwexplo-process Wed Sep 15 07:51:46 1999 From: Natalie Otey Date: Wed Sep 15 07:51:50 PDT 1999 Subject: Booting from SCSI, with success VxWorkers, Some time ago, I posted a problem about booting vxWorks from SCSI. Thank you to all who replied. After getting past many small problems, I discovered that I was seeing a hardware interaction problem. Force Computers has recently added a FAQ page to their website, and that's where the answer was. The problem only occured with a particular combination of hardware. My eyes were popping out when I realized that I was using this exact combination: A Force PowerPC 6604, Interphase 4520 SCSI PMC and Seagate ST34573 SCSI disk! It turned out to be a caching problem, and the solution was to turn off caching on the Seagate. But I felt that would reduce performance, so I opted for another solution - avoid the Seagate ST34573. At this point, I could successfully boot vxWorks.st, but I wanted the option to boot vxWorks, load vxWorks.sym, and execute a startup script. WRS provided the way to do this: Make a few additions to usrRoot in usrConfig.c Add the two variable declarations near the top of the function SCSI_PHYS_DEV * myPDev; BLK_DEV * myBDev; Then sometime after scsi is initialized but before the symbol table is loaded, create the file system. Do this right before #ifdef INCLUDE_STANDALONE_SYM_TBL /* create PhysDev, BlkDev, and File System for vxWorks.sym to load on boot*/ myPDev=scsiPhysDevCreate(pSysScsiCtrl,1,0,0, -1,0,0,0); myBDev=scsiBlkDevCreate(myPDev,0,0); dosFsDevInit("/sd0",myBDev,0); /* create system and status symbol tables */ #ifdef INCLUDE_STANDALONE_SYM_TBL This worked and I'm up and running!!!! Thanks again to all, Natalie Otey KineticSystems Corp. From owner-vxwexplo-process Wed Sep 15 09:30:29 1999 From: David Laight Date: Wed Sep 15 09:30:33 PDT 1999 Subject: RE: Daylight saving time standard?? Jonathon, Try being British... The UK government has never set a rule for the start and end of summertime. Usually they don't decide in time for diaries to contain the dates. For a long time the end was (I think) at 2:00am on the sunday following the 4th saturday in october. We set Unix system to change at 26:00 on the saturday. The first time this mattered they changed it! The times are typically 1 week out from the US. The other issue we have is that in winter our timezone is GMT and in summer BST (British summer time). GMT exists all year and is (almost) UTC, BST only exists in the summer and is 1 hour east of UTC. Now run JAVA, it has its own database of timezones, below are the default set from PJava: -> Available timezones: id: GMT, raw offset: 00:00 (0 ms) id: UTC, raw offset: 00:00 (0 ms) id: ECT, raw offset: 01:00 (3600000 ms) id: EET, raw offset: 02:00 (7200000 ms) id: ART, raw offset: 02:00 (7200000 ms) id: EAT, raw offset: 03:00 (10800000 ms) id: MET, raw offset: 03:30 (12600000 ms) id: NET, raw offset: 04:00 (14400000 ms) id: PLT, raw offset: 05:00 (18000000 ms) id: IST, raw offset: 05:30 (19800000 ms) id: BST, raw offset: 06:00 (21600000 ms) id: VST, raw offset: 07:00 (25200000 ms) id: CTT, raw offset: 08:00 (28800000 ms) id: JST, raw offset: 09:00 (32400000 ms) id: ACT, raw offset: 09:30 (34200000 ms) id: AET, raw offset: 10:00 (36000000 ms) id: SST, raw offset: 11:00 (39600000 ms) id: NST, raw offset: 12:00 (43200000 ms) id: MIT, raw offset: -11:00 (-39600000 ms) id: HST, raw offset: -10:00 (-36000000 ms) id: AST, raw offset: -09:00 (-32400000 ms) id: PST, raw offset: -08:00 (-28800000 ms) id: PNT, raw offset: -07:00 (-25200000 ms) id: MST, raw offset: -07:00 (-25200000 ms) id: CST, raw offset: -06:00 (-21600000 ms) id: EST, raw offset: -05:00 (-18000000 ms) id: IET, raw offset: -05:00 (-18000000 ms) id: PRT, raw offset: -04:00 (-14400000 ms) id: CNT, raw offset: -04:30 (-12600000 ms) id: AGT, raw offset: -03:00 (-10800000 ms) id: BET, raw offset: -03:00 (-10800000 ms) id: CAT, raw offset: -01:00 (-3600000 ms) I don't know (yet) what the summertime dates are. Anyone in the southern hemispere is probably stuffed (the 'end date' is before the 'start date'). Also on vxWorks the JVM completely ignores the timezone offset set in TIMEZONE, it uses the name to index the above table. So if I set TIMEZONE=BST::-60:: the JVM things we are 6 hours east of UTC! I'm about to write a date setting code, for the timezone I may end up using a timer to dynamically change TIMEZONE based on more user-friendly parameters. 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 Sep 15 09:45:54 1999 From: Ahmet Mumcu Date: Wed Sep 15 09:45:58 PDT 1999 Subject: DIFFERENCES ON MAING FROM SHELL AND IDE Hi, What are the differences on running make from the shell and from IDE for Tornado-2? For the same project files and with the same compile and link switches we have a problem on the make from the shell at the link stage but no problem with the make from IDE. Any comment are highly appreciated. Thank you. Ahmet Mumcu ASELSAN Inc. MST Group From owner-vxwexplo-process Wed Sep 15 13:31:42 1999 From: John_W_Cosgrove@res.raytheon.com Date: Wed Sep 15 13:31:45 PDT 1999 Subject: Re: VME bus interrupts - MVME2700 paulvroberts@my-deja.com wrote: > I am having a problem with VME interupts on the MVME2700. > > I am getting the error > "bad vme interrupt 0" > > This problem occurs both on a slightly older MVME2700-1241 (233MHz) and > on a newer MVME2700-1441 (366Mhz). Paul, A VME bus analyser could confirm this, but I just helped a colleague with the same problem. I believe you have a problem brought on by the VME "Write Posting" feature of the PPC boards. The problem in our case was the ISR writes to a VME register to clear the interrupt and returns. The interrrupt signal (VME Level 3, in our case) is still pending when the ISR completes and then vanished before the the vector can be fetched for this interrupt. The register write had not completed, but had been "write posted", allowing the PPC to continue execution before the write to the VME register actually occurs. The write completes before the PPC can get the vector from the VME bus, causing the error. Our solution was to read back the VME register, in the ISR. This allows the "posted" write to complete before the ISR exits. Another solution is to disable "write posting" (not our solution, maybe someone else has solved it this way) since this would force the PPC to wait for completion before continuing. John Cosgrove Raytheon Systems Co. (401)842-4167 (desk) jwc@ssd.ray.com vxworks vxworks vxworks "don't SPAM me" vxworks vxworks vxworks From owner-vxwexplo-process Wed Sep 15 16:36:18 1999 From: james.p.mccammon@exgate.tek.com Date: Wed Sep 15 16:36:22 PDT 1999 Subject: Happy with support/no source As far as the "No source code" problem is concerned, anyone who has not yet purchased Tornado/Vxworks can probably work out a deal with their sales person for limited use of some (if not all) the source code. But do this before you write your check and get it in writing (some sales people have a short memory). WRS is beginning to realize that they ain't the only game in town. One of our engineering teams tossed Tornado for a better development environment. For our next project we will be looking very closely at other options because WRS seems reluctant to cooperate in this area. With respect to support: I've been working with the 100BaseT driver source code (motFecEnd.c) and had to develop a workaround for a bug (or what I think is a bug). I called support, emailed the workaround source code and haven't heard from them for several days. The last progress report said "We'll look into it". Now, while it may take some investigation, they could at least email me and let me know what is going on or give some estimate of when I will be hearing from them. James From owner-vxwexplo-process Wed Sep 15 17:52:03 1999 From: Hakan Karakas Date: Wed Sep 15 17:52:07 PDT 1999 Subject: RE:DIFFERENCES ON MAING FROM SHELL >Submitted-by owner-vxwexplo-process Wed Sep 15 09:45:54 1999 >Submitted-by: Ahmet Mumcu > >Hi, > >What are the differences on running make from the shell and from IDE >for Tornado-2? The problem is the mixing of project facility and manual editing of config.h file at the same time. In order too avoid confusion and errors, the two methods should not be used together for the same project. Try using manually build via Makefile under your target/config/ or use the project facility to apply your changed to the config.h file. Following warning is from the Tornado II User Guide : The order of precedence for determining configuration is (in descending order): project facility config.h configAll.h >For the same project files and with the same compile and link switches >we have a problem on the make from the shell at the link stage but no >problem with the make from IDE. > >Any comment are highly appreciated. Thank you. > >Ahmet Mumcu >ASELSAN Inc. >MST Group Hakan Karakas h.karakas@ieee.org From owner-vxwexplo-process Thu Sep 16 00:18:28 1999 From: Marco Raimondi Date: Thu Sep 16 00:18:32 PDT 1999 Subject: VxWorks FTP server Hi VxWorks users! We are having problems with the ftp server. There is two questions. 1. The ftp server processes (called tFtpdServ1, tFtpdServ2 ...) are spawned by the tFtpdTask when a client requires a connection. If the client, for any reason, can't close the connection, how much time the connected ftp server process can live? It lives forever or there is a max time after then it is deleted? How can I configure this time out? It depends on tcp keep-alive mechanism? Can I configure max retransmission number and max tcp timeout? 2. During a file transfer with ftp server storing, the ftp server call the write that I provided compliant with the VxWorks I\O sytem. This procedure is called passing a buffer with a size of 10240 (10k) bytes. This size is configurable? Where? Thanks in advanced! -- =================================== ITALTEL - Business Unit Radio Marco Raimondi Radio Equipment Senior Software Engineer I-20060 Cassina de' Pecchi (MI) S.S. Padana Superiore Km 158 Italy Tel:+39 02 4388 6026 Fax:+39 02 4388 6088 ==================================== From owner-vxwexplo-process Thu Sep 16 01:24:02 1999 From: David Anderson Date: Thu Sep 16 01:24:06 PDT 1999 Subject: RE: Interrupts on Pentium target (especially COM1 and COM2) On 15 September 1999 15:47, the vxWorks Users Group Exploder [SMTP:vxwexplo@lbl.gov] wrote: > Submitted-by owner-vxwexplo-process Wed Sep 15 07:47:17 1999 > Submitted-by: "Slutsker, Rasty" > > 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_01BEFF88.FAFDE632 > Content-Type: text/plain; > charset="iso-8859-1" > > Hi, > I assume you're going to replace original serial driver. > As far as I remember, it is not enough to just enable corresponding IRQ > line, PC has a funny "OUT2" line, which is controlled by configuration > registers of COM1&COM2, and should be set to "1" in order to enable > interrupts from serial port. Look in the original driver (i8250sio.c) for > OUT2. Also you have to enable interrupt generation in UART. > > Regards, > Rasty > Just a short warning... don't rely upon OUT2 to switch interrupts off as not all serial card implementations use it ( especially the cheaper varieties ). Always use the UART IER register for this !. David Anderson. From owner-vxwexplo-process Thu Sep 16 01:42:23 1999 From: David Laight Date: Thu Sep 16 01:42:28 PDT 1999 Subject: RE: ISA Plug & Play Christian, I have written ISA PnP grope code in the past, its not TOO difficult. (Code is owned by a previous employer.) The only problem I recall is that you can't find the read data port the system BIOS used - so if you make any BIOS calls that might try to access the ISA PnP registers they will fail. (Unlikely on the sort of OS where you are trying to grope the cards.) I seem to recall trying a few read data ports, try starting at the high end of the valid addresses (but avoid the device at 3fx). The ISA PnP cards have a 'required for boot' bit which forces the BIOS to allocate resources. The BIOS config SHOULD also have a question which will cause it to allocate resources to all ISA PnP cards, the name of the question is BIOS dependant... We did have some problems with the PnP config data on some cards - and ended up writing a utility to modify it so that only specif configurations were valid. (The manufacturers program didn't like running of a DOS boot floppy on our hardware - which was running UnixWare 2.) Once you have identified the cards you can read back the io and memory base addresses (but not the size - parse the PnP info for that). Bung it all into a table for other vxworks drivers to search. Have fun :-) 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 Sep 16 04:00:13 1999 From: daemon@csg.lbl.gov Date: Thu Sep 16 04:00:17 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Sep 16 04:00:09 PDT 1999 Subject: Re: VME bus interrupts - MVME2700 Subject: Re: Task Execution Time. Subject: Re: VME bus interrupts - MVME2700 Subject: WRS Support? Subject: Realtime Internet Resources Subject: ld() / Environment variables Subject: Re: Interrupts on Pentium target (especially COM1 and COM2) Subject: Re: Daylight saving time standard??? Subject: Personal Java and Tornado 2 Subject: Re: Device Driver Docs Subject: open() for a large file Input/Output Subject: Re: Instr translation miss exceptions Subject: Re: DIFFERENCES ON MAING FROM SHELL AND IDE Subject: VME A32 addressing/MVME162 Subject: NCR810 scsi initialization on vxworks Subject: windows services for Unix Subject: Re: VME bus interrupts - MVME2700 Subject: Re: comparisons with other OS's Subject: Hierarchical projects under Tornado II? Subject: Re: ld() / Environment variables Subject: Re: ENOBUFS for TCP over sockets ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME bus interrupts - MVME2700 Date: Wed, 15 Sep 1999 09:58:14 -0400 From: David Abbott Organization: Jefferson Lab Message-ID: <37DFA61F.371BFE37@jlab.org> References: <7rnogp$2of$1@nnrp1.deja.com> vxWorks Paul, You did not mention the BSP release you were using, but I can almost guarantee that the problem you are having is due to your VME board being a RORA board wrt interrupts. The Universe ISR (sysUnivVmeInt() which is found in universe.c) prior to Tornado II (BSP v1.2/0) does not properly handle that case where interrupts may be coming from RORA boards. The problem is that the CPU is clearing the Universe interrupt register (Responsible for passing VME interrupts up to the Raven PCI bridge chip) before calling your ISR routine where you then clear the VME interrupt source. So theoretically if your VME interrupt source is not cleared quickly enough it can latch another interrupt pending in the Universe Register. By the time the Universe ISR gets around to servicing this next interrupt your VME source (int level info etc..) is gone. Hence, you get the "bad vme int 0" message. Interestingly enough I have found that if you Clear your RORA board's interrupt request as the VERY first thing in your ISR these messages do not show up. So there does seem to be a timing issue here. The real fix however is to move the clearing of the Universe interrupt request until after your ISR is called in sysUnivVmeInt(). I have noticed that Wind River finally put this fix in the latest BSPs for Motorola boards with the release of Tornado II. Hope this is useful, David Abbott (abbottd@jlab.org) Jefferson Lab Data Acquisition Group Newport News, VA paulvroberts@my-deja.com wrote: > Hi there, > > I am having a problem with VME interupts on the MVME2700. > > I am getting the error > "bad vme interrupt 0" > > This problem occurs both on a slightly older MVME2700-1241 (233MHz) and > on a newer MVME2700-1441 (366Mhz). > > The problem seems worse on the 366Mhz board. The problems occur when my > ISR is short - adding a delay seems to fix it under certain conditions. > > The IRQ is originating from a digital i/o IP module (Acromag IP470) that > is on an IP->VME Carrier board (Greenspring). From reading the data > sheet for the IP module and checking the IRQ/IACK waveforms, it looks > like the board uses the RORA form of VME interrupt (ie you need to write > to the board to clear the interrupt). > > When I get the error message, tt looks like the processor sees a second > interrupt, becuase it is assering another IACK. In some cases, this > appears to cause in a timeout, in other cases, the ISR is run again. > > I discovered that if I put a particular delay in the ISR, then I did not > get the problem at 1Hz interrupts. Strangely, it didn't seem to matter > where the delay occurs, which makes me think the problem is not the > board (if it was a race condition because of the board, I would have > expected it to be required after the clearing of the IRQ). > > However, when I tested at 10Hz and 1KHz, further problems arose > including IRQ never being cleared and kernel panics. > > Has anyone seen anything like this before or give me **any** clues to > further track this down, > > Thanks, > > Paul Roberts. > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Task Execution Time. Date: Wed, 15 Sep 1999 10:23:48 -0400 From: Tim Sohacki Organization: Nortel Networks Message-ID: <37DFABF4.48BB1C51@americasm01.nt.com> References: <7rls2n$kgi$1@nslave1.tin.it> For higher resolution on the PPC, you can read the decrementer register with vxDecGet(). It counts down, one unit equals four bus clock cycles. You can set it in advance with vxDecSet(). Cheers, Tim. Massimo Maio wrote: > > Hi VxWorkers ! > I need to evaluate the ticks (or the milliseconds) elapsed between the time > the task begins to execute the first instruction after semTake, and the time > at which it has completed to execute the last instruction of the FOREVER > cycle. > Can you help me ? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME bus interrupts - MVME2700 Date: Wed, 15 Sep 1999 08:52:17 -0700 From: Ken Norlin Organization: NASA Dryden FRC Message-ID: <37DFC0B1.266AA62@dfrc.nasa.gov> References: <7rnogp$2of$1@nnrp1.deja.com> paulvroberts@my-deja.com wrote: > > Hi there, > > I am having a problem with VME interupts on the MVME2700. > > I am getting the error > "bad vme interrupt 0" > Hi, I had the exact same problem with a RORA-type interrupter. After talking to some folks at Synergy, I added the following lines of code at the end of my interrupt handler: /* Make sure VME write completes before exiting interrupt handler */ __asm__(" sync "); __asm__(" eieio "); /* Do a read of polarity bit to flush write */ us_polarity = mddio->IRQ_Pol_07_00; } The assembly instructions worked with my MVME2604 BSP as well as the Synergy BSP so they should work for you. The eieio instruction stands for 'enforce instruction execution in order' -- so your write access (required for RORA interrupter) will be executed before the interrupt handler is exited. Ken N. - -- Ken Norlin --- Simulation Engineer NASA Dryden Flight Research Center ken.norlin@dfrc.nasa.gov (661) 258-2046 FAX (661) 258-2792 --------------------------- Newsgroups: comp.os.vxworks Subject: WRS Support? Date: Wed, 15 Sep 1999 09:13:03 -0700 From: "Kevin Kuhns" Organization: Posted via Supernews, http://www.supernews.com Message-ID: I am considering using VxWorks on a new project and am looking for feedback in regards to WRS support. Is WRS responsive when a problem arrises? Do they provide a fix/patch in a few days or a few weeks? I am guessing the core OS is quite reliable, but what about the optional components like the web server, java, etc.?? Does WRS spend more effort getting the problem fixed or convincing you its not their problem? When a new revision is released, is the update painless or a pain in the a?? Any info you can give would be helpfull and very appreciated. Thanks all. Kevin Kuhns _kkuhns@opto22.com_ --------------------------- Newsgroups: comp.os.vxworks Subject: Realtime Internet Resources Date: 15 Sep 1999 10:21:59 PDT From: Nina Pinto Organization: Concentric Internet Services Message-ID: <37DFD835.DEB3372E@eeglossary.com> Pertinent to comp.os.vxworks, EG3 maintains a list of realtime Internet resources, including FAQ's, papers, hot list pages, potential software/shareware, all known companies, etc. Specific pages are located at: http://www.eg3.com/real/vxworks.htm These resource pages are published as part of EG3's Free Electronic Engineers' Toolbox at: http://www.eg3.com/index.htm If you know of other resources, pertinent to comp.os.vxworks please email me at: Best regards, Nina Pinto Assistant Editor npinto@eeglossary.com - ------------------------------ http://www.eg3.com eg3.com: window on the electronic design world 1.3 million hits per month - 100,000 to 125,000 quarterly users design engineers * programmers * oem decision-makers board-level * dsp * embedded * internet * mcu/mpu * realtime tel: 408-938-9150 fax: 408-938-9155 email: inquiry@eg3.com Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: ld() / Environment variables Date: Wed, 15 Sep 1999 14:26:10 -0400 From: John Sestak Organization: Raytheon Systems Company, Falls Church Message-ID: <37DFE4C2.54C62534@fallschurch.esys.com> Is there a way to set an environment variable to a file path so I don't have to type the full path of the file I'm trying to load when I'm using the ld() command. (i.e. I don't want to have to type-> ld < /Tornado/target/proj/ProjectName/Architecutre ... every time, instead I want to type ld < $DIR/filename, where $DIR equals my file path). --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupts on Pentium target (especially COM1 and COM2) Date: Wed, 15 Sep 1999 18:34:50 GMT From: saustin_deja@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7roos9$qts$1@nnrp1.deja.com> References: <19990914225933.01931.00000359@ng-fz1.aol.com> > Ultimately, I would like to have all COM ports on the given interrupts, and > also be able to use the GPIB SRQ feature, however, my most immediate need is to > get COM1 and COM2 working with my ISR. > Can you narrow it down by removing the GPIB and B&B boards and just focussing on getting COM1 and COM2 to work? That might tell you if it's a configuration problem (BIOS or in your vxWorks/BSP setup when you build) or some other interaction. Can you do a plain vanilla build of vxWorks and get COM1 to act as a serial console port? Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Daylight saving time standard??? Date: Wed, 15 Sep 1999 18:21:02 GMT From: Charlie Grames Organization: Boeing Message-ID: <37DFE38E.4B94218F@nowhere.com> References: Sender: nntp@news.boeing.com (Boeing NNTP News Access) Jonathan VxWorks is correct. The baseline time for DST is always 2:00 am. On the first Sunday in April, time goes from 2:00 to 3:00. On the last Sunday in October, time goes from 2:00 to 1:00. Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com Jonathan Cheng wrote: > > Hi, > > Just wonder if anyone know the standard of Daylight saving time in United > States? > > From the following Web site: > http://www.treasure-troves.com/astro/DaylightSavingTime.html > Daylight savings time ends when clocks are set from 3:00 a.m. to 2:00 a.m. > on last Sunday in October. > > In VxWorks, DST ends when clocks are set from 2:00 a.m. to 1:00 a.m. on last > Sunday in October. ( If the TIMEZONE environment is set to > UTC::0:040402:103102 ) Same as Window NT and Window 95/98. > > Seems to me that both Windows and VxWorks does not implement the DST > correctly. Certainly, we can change this by editing the TIMEZONE > environment. > > I just wonder if I have missed something. > > -Jonathan Cheng --------------------------- Newsgroups: comp.os.vxworks Subject: Personal Java and Tornado 2 Date: Mon, 13 Sep 1999 23:49:23 GMT From: bonuccip@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rk2hv$ftf$1@nnrp1.deja.com> I'm trying to run the demonstration program that comes with Personal Java under VxWorks and I just can't get VxWorks to complete compiling. The hardware is a Pentium clone with a floppy drive, an ATA hard drive and a flash disk. No network connection. I have gone into Tornado and set it up to run with the Java subset I will be using and turned off all network modules. When I select "Build All", it goes off and compiles, failing in the link stage. The linker reports that clDescTblNumExt, clDescTbl, mClBlkConfig, sysClDescTblNumEnt, sysClDescTbl, and sysMclBlkConfig are undefined symbol names. These all seem to be network symbols, but I can't see where they are coming from. I have contacted Wind River and asked for help from them. It's been two months and they still have not been able to tell me how to run the demo program on a stand-alone system. Any help with this would be appreciated. Peter A. Bonucci Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Device Driver Docs Date: Wed, 15 Sep 1999 13:33:55 -0500 From: John Beauford Organization: Motorola Message-ID: <37DFE693.394EEEA0@email.mot.com> References: <37DE9F11.5526@gv.net> Reply-To: qjb005@email.mot.com In terms of WRS documentation: More information on device drivers is available in the Tornado BSP developers kit(only available for Tornado 1.x). It covers several types at a high level, such as serial, network, scsi etc. The Vxworks Network programmers guide also does a little coverage of BSD style device drivers. The Network Protocol Toolkit Users Guide is also a pretty good source of information about network drivers. It is available online at the wrs website. A really good source if information is the souce code that comes with BSPs that you purchase from WRS. I have used their examples in conjunction with the available docs and this forum to create a couple of drivers for chips that were unsupported. "Michael R. Kesti" wrote: > > As has been proviously noted in this forum, the device driver docs in > section 3.9 of the VxWorks Programmer's Guide are somewhat sketchy, at > best. Where can I obtain better docs without resorting to WRS's device > driver seminar? > > -- > ======================================================================== > Michael Kesti | "And like, one and one don't make > | two, one and one make one." > mkesti@gv.net | - The Who, Bargain - -- John Beauford --------------------------- Newsgroups: comp.os.vxworks Subject: open() for a large file Input/Output Date: Wed, 15 Sep 1999 19:31:11 GMT From: jack_shabel@ictv.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7ros60$tpk$1@nnrp1.deja.com> I am trying to open() a file that resides on my NT machine. When the file is very large (> 10MB), I get an open() error. If the file is small enough, I can open the file, read it in smaller chunks, & process the data within my VxWorks program. Does the system read the total file into the embedded memory & then buffer the data out on each read? Is there anyway I can open a large file and buffer the data in as smaller records? I seem to get system malloc errors when I open a large file. I also get similar errors when my output file is too large. Any help would be greatly appreciated. I am assuming the total file resides within the VxWorks memory until the close function is called. Thanks Jack Shabel Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.sys.powerpc.tech,comp.os.vxworks Subject: Re: Instr translation miss exceptions Date: Wed, 15 Sep 1999 13:12:47 -0700 From: Bryan Wilcutt Organization: AG Communication Systems Message-ID: <37DFFDBF.A19A3F9A@agcs.com> References: <86CE30EB9755CE6B.CEED9D9FE2C2E68D.972C9CF4B38FBA2E@lp.airnews.net> Reply-To: wilcuttb@agcs.com Try messing with the -mshort-load-bytes option for the compiler (assuming you're using GNU which you didn't mention). Pete Kockritz wrote: > We're using vxWorks on a PPC860. On a kernel that does not have the > caches or MMU enabled, we're getting instruction translation miss > exceptions. I would've thought that was not possible. What does it > mean when you get this exception with MMU disabled? > > Regards, > Pete > -- > +-----------------------------------------------------------------+ > | Pete Kockritz mailto:petekz@airmail.net | > | This space for rent. | > +-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DIFFERENCES ON MAING FROM SHELL AND IDE Date: Wed, 15 Sep 1999 13:13:31 -0700 From: Bryan Wilcutt Organization: AG Communication Systems Message-ID: <37DFFDEB.7D98B6D5@agcs.com> References: <7rok6g$r36@overload.lbl.gov> Reply-To: wilcuttb@agcs.com How about an error list? Ahmet Mumcu wrote: > Hi, > > What are the differences on running make from the shell and from IDE for > Tornado-2? > > For the same project files and with the same compile and link switches > we have a problem on the make from the shell at the link stage but no > problem with the make from IDE. > > Any comment are highly appreciated. Thank you. > > Ahmet Mumcu > ASELSAN Inc. > MST Group --------------------------- Newsgroups: comp.os.vxworks Subject: VME A32 addressing/MVME162 Date: Wed, 15 Sep 1999 15:52:43 -0700 From: Kevin Havig Organization: Posted via Supernews, http://www.supernews.com Message-ID: <37E0233B.E69F041@navair.navy.mil> I am using VxWorks 5.4 on a Motorola MVME162 processor. I am having a problem accessing a board in the VME 32 bit address space; address 0x10000000 to be exact. The processor has no problem accessing any boards in the VME 16 or 32 bit address space. Furthermore if I boot the processor to the built in Motorola debugger, it accesses this location just fine. Is there some trick to getting VxWorks to access this board. Thanks in advance for all of your help. Kevin Havig Naval Air Warfare Center havigkk@navair.navy.mil --------------------------- Newsgroups: comp.os.vxworks Subject: NCR810 scsi initialization on vxworks Date: 16 Sep 1999 02:18:28 GMT From: sri@spinlogic.com Organization: Spinlogic Inc Message-ID: <7rpk1k$jfc$1@nntp.msstate.edu> vxWorkers, I am trying to get the NCR810 working on the yellowknife/Pentium. I could get the Adaptec2940 working on the pentium but the driver doesn't seem to support the ppc architecture.. back to NCR... I am not able to get this working on either the pentiums or the yelloknife. I followed the same procedure as for the adaptec card and also changed the ncr810DelayCount to 100/1000 and tried all combinations and it looks like there is no change in its behaviour...it just keeps hanging when I try to do the scsiAutoConfig.(I also tried doing the scsiBusReset).. Also the card doesn't have any onboard bois ... oops ..bios can anybody suggest as to how I should proceed with this problem.. thanks in advance sri --------------------------- Newsgroups: comp.os.vxworks Subject: windows services for Unix Date: Wed, 15 Sep 1999 19:50:38 -0700 From: Fred Kennedy Organization: Value Net Internetwork Services Inc. Message-ID: <37E05AFE.7B8DF8D9@fredkennedy.com> Date Title Version Size/Time 17 Aug 1999 1166 A Technical Overview of UNIX and Windows NT A 2,300 kb / 12min 17 Aug 1999 1514 Implementing Microsoft Windows NT Services A 2,500 kb / 13min 4 Aug 1999 Windows Services for Unix V2Beta1 N/A --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME bus interrupts - MVME2700 Date: Thu, 16 Sep 1999 04:46:22 GMT From: paulvroberts@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rpsms$kru$1@nnrp1.deja.com> References: <7rnogp$2of$1@nnrp1.deja.com> <37DFC0B1.266AA62@dfrc.nasa.gov> Thanks to everyone who responded for their excellent advice. I was able to fix my problem by using the "sync/eieio/read" work-around suggested by Ken Norlin. I am using the 1.1/5 BSP with Tornado I, so maybe when (WRS Asia Pacific finally get their act together) I receive Tornado II, the problem will be fixed properly. By the way, I realised that my kernel panics were not relevant to the interrupt problem. I think they were actually because the Acromag digital i/o IP board I was using has a "feature" whereby it requires the signals to be non-changing when you set up the interrupts. Not complying with that requirement sometimes produced endless interrupts. A case of grotty hardware clouding the problem. Thanks again everyone, Paul Roberts. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: comparisons with other OS's Date: Wed, 15 Sep 1999 22:08:03 -0700 From: engineer_scotty Organization: aracnet.com Message-ID: <37E07B33.B91F6DAF@aracnet.com> References: <7rm928$1h0$1@nnrp1.deja.com> Ted Dennison wrote: > > In article , > s wrote: > > > This would include things like amount of process creation code and the > > number of system calls that needed to be made to create, maintain, > > delete a process and all of its related needs (memory etc). > > Perhaps you are attacking the problem at the wrong level. If you use > Ada, process creation code is minimal and you don't have to make *any* > system calls to do it. Nor does vxWorks require any "system calls". Sure, the taskSpawn() function is invoked; but in vxWorks a system call is just like any other function call; no system trap or context switch into kernel space is required. Whether or not Ada is appropriate for the original posters design, I don't know...I don't know enough about the language to comment. (I will ask a question--if Ada has the context of "processes" as a language feature, how do these map onto system tasks on multitasking systems????) > -- > T.E.D. > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Hierarchical projects under Tornado II? Date: 16 Sep 1999 05:54:17 GMT From: ajeddeloh@aol.com (AJeddeloh) Organization: AOL http://www.aol.com Message-ID: <19990916015417.01471.00001746@ng-cg1.aol.com> Has anyone found a way to use the Tornado II "project" facility for a non-trivial product? I have approx. 1000 source files in a hierarchical directory structure. How can I set up a build process under Tornado II without ending up with every object file landing in the same place? This is legacy code and does not lend itself to be broken into "Kernel" vs "application" projects. This is particularly nasty, since I have multiple 3rd-party vendor code sets which have to be integrated into the build, some of which duplicate file names! Please respond to Alan_Jeddeloh "AT" adc.com. This AOL account does not email from outside AOL. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ld() / Environment variables Date: Thu, 16 Sep 1999 09:06:15 +0200 From: Zoltan Kakucs Organization: DESY Message-ID: <37E096E7.E3AAE414@herant.desy.de> References: <37DFE4C2.54C62534@fallschurch.esys.com> Reply-To: kakucs@herant.desy.de This is a multi-part message in MIME format. - --------------7A0190A589DA9C87A6F95BC6 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi ! I use an environment variable for my file path, like this home="/my/preffered/directory/path" The download part looks like: cd home ld < filename Regards, Zoltan > Is there a way to set an environment variable to a file path so I don't > have to type the full path of the file I'm trying to load when I'm using > the ld() command. (i.e. I don't want to have to type-> ld < > /Tornado/target/proj/ProjectName/Architecutre ... every time, instead I > want to type ld < $DIR/filename, where $DIR equals my file path). - --------------7A0190A589DA9C87A6F95BC6 Content-Type: text/x-vcard; charset=us-ascii; name="kakucs.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Zoltan Kakucs Content-Disposition: attachment; filename="kakucs.vcf" begin:vcard n:Kakucs;Zoltan tel;cell:0177 - 24 12 361 tel;fax:040 - 8998 4388 tel;work:040 - 8998 3283 x-mozilla-html:FALSE org:Deutsches Elektronen-Synchrotron;KRYK/MKS2 adr:;;Notkestr. 85;Hamburg;;22607;Germany version:2.1 email;internet:kakucs@herant.desy.de title:Dipl.-Ing. fn:Zoltan Kakucs end:vcard - --------------7A0190A589DA9C87A6F95BC6-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ENOBUFS for TCP over sockets Date: Thu, 16 Sep 1999 09:38:10 +0200 From: Zoltan Kakucs Organization: DESY Message-ID: <37E09E62.AE7C8E95@herant.desy.de> References: <7rprir$k2u$1@nnrp1.deja.com> Reply-To: kakucs@herant.desy.de This is a multi-part message in MIME format. - --------------9CAFE3A7F09CC18270035063 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, what are the settings for the maximum size and the number of MBUF's and CLUSTER's ?? > mbufShow shows you this values. The file to set this parameters: /your/tornado/path/target/h/net/mbuf.h #define MSIZE 128 /* size of an mbuf */ #define MCLBYTES 1024 /* size of a cluster mbuf */ #define NUM_INIT_CLUSTERS 4 /* number of clusters to initialize */ #define NUM_INIT_MBUFS 40 /* number of mbufs to initialize */ #define NUM_MBUFS_TO_EXPAND 10 /* number of mbufs to add each time */ #define NUM_CLUSTERS_TO_EXPAND 1 /* number of clusters to add */ #define MAX_MBUFS 1500 #define MAX_CLUSTERS 256 /* max number of clusters */ #define MAX_MBUF_TYPES 256 /* max number of mbuf types */ Hope this helps. Regards, Zoltan > Hi, > > I'm using Tornado 1.0.1 and SENS. When we make 30 TCP connections > over socket and transmite small data packets (about 40 bytes per > packet, and about 30 packets per connection), we get S_errno_ENOBUFS > from send(). I've already modifed the mblks to this: > > #define NUM_NET_MBLKS 1500 > #define NUM_SYS_64 250 /* 64 byte system clusters */ > #define NUM_SYS_128 250 /* 128 byte system clusters */ > #define NUM_SYS_256 250 /* 256 byte system clusters */ > #define NUM_SYS_512 250 /* 512 byte system clusters */ > #define NUM_SYS_CL_BLKS (NUM_SYS_64 + NUM_SYS_128 + \ > NUM_SYS_256 + NUM_SYS_512) > #define NUM_SYS_MBLKS (2 * NUM_SYS_CL_BLKS) > > What else do I have to change? and are these numbers ok? > > also, I know the command netStackSysPoolShow shows the mblks, but is > there another command that will show the 1024 and 2048 mblks? > > Thank you! > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. - --------------9CAFE3A7F09CC18270035063 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;cell:0177 - 24 12 361 tel;fax:040 - 8998 4388 tel;work:040 - 8998 3283 x-mozilla-html:FALSE 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 - --------------9CAFE3A7F09CC18270035063-- --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Thu Sep 16 08:36:33 1999 From: "Slutsker, Rasty" Date: Thu Sep 16 08:36:36 PDT 1999 Subject: SLIP/ARP/PROXY in VxWorks 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_01BF0059.1DC81C2A Content-Type: text/plain; charset="iso-8859-1" Hi, Did anyone succeed to have Proxy ARP server with the SLIP? I have five targets connected together though Ethernet, one of them has also SLIP connection to the host computer. I'd like to establish connection between host and any target. I initially planned to run Proxy ARP server on the target that has SLIP connection to the host, but proxyNetCreate always returns -1, no examples in manuals, tried a dozen of combinations ;-(. Any suggestions? Rasty Tornado VxWorks ------_=_NextPart_001_01BF0059.1DC81C2A Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable SLIP/ARP/PROXY in VxWorks

Hi,

Did anyone succeed to = have Proxy ARP server with the SLIP?

I have five targets = connected together though Ethernet, one of them has also SLIP = connection to the host computer.  I'd like to = establish connection between host and any target. I initially planned = to run Proxy ARP server on the target that has SLIP connection to the = host, but proxyNetCreate always returns -1, no examples in = manuals, tried a dozen of combinations ;-(.

Any = suggestions?

Rasty

Tornado = VxWorks

------_=_NextPart_001_01BF0059.1DC81C2A-- From owner-vxwexplo-process Thu Sep 16 09:08:34 1999 From: John_W_Cosgrove@res.raytheon.com Date: Thu Sep 16 09:08:38 PDT 1999 Subject: Trouble with Tornado Hi, Folks, I am using Tornado 1.0.1 (vxWorks 5.3.1) on a 2700 board. I keep the following messages in the log file window: Current Log information are available in the file: /home/jwc/.wind/launchLog.F100_HA_UT Command: tgtsvr f100-ha-ray -A -V -n F100_HA_UT -f a.out -m 3145728 tgtsvr (F100_HA_UT@emperor): Thu Sep 16 12:01:03 1999 License request... authorized on host 'emperor'. Attaching backend... succeeded. Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Error: Unable to initialize object module loader library This stuff worked on a differnet machine using a fixed license (I did the install). Now the new install of the same revision (Company has a company wide license) on a different host will not run. This is a pain since I would to use but CrossWind and WindView and cannot. Anybody know what causes this stuff? John Cosgrove Senior S/W Engineer II Raytheon Systems Co. (401)842-4167 (desk) jwc_NOSPAM@ssd.ray.com Pls remove the _NOSPAM to respond to my messages. From owner-vxwexplo-process Thu Sep 16 09:43:01 1999 From: Ian.Kerr@bae.co.uk Date: Thu Sep 16 09:43:04 PDT 1999 Subject: Re: comparisons with other OS's --Boundary_(ID_wZSSQu62JDkPX2xp+gGXtA) Content-type: TEXT/PLAIN; CHARSET=US-ASCII --Boundary_(ID_wZSSQu62JDkPX2xp+gGXtA) Content-type: MESSAGE/RFC822 Date: Thu, 16 Sep 1999 17:20:23 GMT Subject: Re: comparisons with other OS's MIME-version: 1.0 Content-type: TEXT/PLAIN; CHARSET=US-ASCII Importance: normal A1-type: DOCUMENT >Newsgroups: comp.os.vxworks >Subject: Re: comparisons with other OS's >Date: Wed, 15 Sep 1999 22:08:03 -0700 >From: engineer_scotty >Organization: aracnet.com >Message-ID: <37E07B33.B91F6DAF@aracnet.com> >References: <7rm928$1h0$1@nnrp1.deja.com> >Ted Dennison wrote: > >> In article , >> s wrote: >> >> > This would include things like amount of process creation code and the >> > number of system calls that needed to be made to create, maintain, >> > delete a process and all of its related needs (memory etc). >> >> Perhaps you are attacking the problem at the wrong level. If you use >> Ada, process creation code is minimal and you don't have to make *any* >> system calls to do it. >...Ada....I don't know enough about the language to comment. (I >will ask a question--if Ada has the context of "processes" as a language >feature, how do these map onto system tasks on multitasking systems????) Ada tasks are asynchronous processes implemented `as threads of control' in a typical Ada run-time system. When Ada is compiled for use on top of an RTOS then it varies. Typically Ada compilers (e.g., GHS) implement Ada tasks as VxWorks or POSIX threads. Other compilers (e.g., Rat) put Ada tasks into POSIX threads in one LynxOS process. If you want MMU protected processes then in LynxOS and OSE you can put separate Ada main programs in each protected memory partition, with several compilers. Anyone know of any other combinations? How is C++ comming along in its ISO version!? Ian. > -- > T.E.D. --Boundary_(ID_wZSSQu62JDkPX2xp+gGXtA)-- From owner-vxwexplo-process Thu Sep 16 10:10:04 1999 From: francois pertin Date: Thu Sep 16 10:10:08 PDT 1999 Subject: Exception and tornado 2 Dear C++ programmer on Tornado 2 and VxWorks I have this problem: Void myfunction (void) { try { do something which throw the "exception timeout" } catch ("exception timeout") { printf ("exception catch\n"); } } void anotherfunction (void) { declare some object with constructors and destructors ... myFunction (); } When I run call anotherFunction I get pageFault. If I write anotherFunction like this: void anotherfunction (void) { declare some object with constructors and destructors ... try { myFunction (); } catch ("exception timeout") { } } It is working !!!!!!! Why ???? Does anybody know why I need this try block ?? Thanks Francois From owner-vxwexplo-process Fri Sep 17 00:00:08 1999 From: Arand Thomas Date: Fri Sep 17 00:00:23 PDT 1999 Subject: AW: Trouble with Tornado John, the license server uses files *.lic to evaluate the licenses. These files are created during installation of license management. The physical location of the .lic-files are encoded in these files somehow and validated, too. So when moving these files and letting the license server work on the new location, license evaluation will fail. So you have to re-run the license installation! Note: When you have to move the location of the tornado-tree more often, it may be useful to do the license installation at a totally different location, which doesn't change. 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, 16. September 1999 18:09 An: vxworks_users@csg.lbl.gov Betreff: Trouble with Tornado Submitted-by owner-vxwexplo-process Thu Sep 16 09:08:34 1999 Submitted-by: John_W_Cosgrove@res.raytheon.com Hi, Folks, I am using Tornado 1.0.1 (vxWorks 5.3.1) on a 2700 board. I keep the following messages in the log file window: Current Log information are available in the file: /home/jwc/.wind/launchLog.F100_HA_UT Command: tgtsvr f100-ha-ray -A -V -n F100_HA_UT -f a.out -m 3145728 tgtsvr (F100_HA_UT@emperor): Thu Sep 16 12:01:03 1999 License request... authorized on host 'emperor'. Attaching backend... succeeded. Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Error: Unable to initialize object module loader library This stuff worked on a differnet machine using a fixed license (I did the install). Now the new install of the same revision (Company has a company wide license) on a different host will not run. This is a pain since I would to use but CrossWind and WindView and cannot. Anybody know what causes this stuff? John Cosgrove Senior S/W Engineer II Raytheon Systems Co. (401)842-4167 (desk) jwc_NOSPAM@ssd.ray.com Pls remove the _NOSPAM to respond to my messages. From owner-vxwexplo-process Fri Sep 17 04:00:11 1999 From: daemon@csg.lbl.gov Date: Fri Sep 17 04:00:15 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Sep 17 04:00:07 PDT 1999 Subject: Re: open() for a large file Input/Output Subject: Re: VxSim and rpccore timeout Subject: Re: Task Execution Time. Subject: Re: ISA Plug & Play Subject: Re: ENOBUFS for TCP over sockets Subject: Re: Task Execution Time. Subject: Re: open() for a large file Input/Output Subject: Software Engineers, MPLS Protocol "Job Posting" Subject: Hierarchical projects under Tornado II? Subject: open() for a large file Input/Output Subject: Re: VxWorks FTP server Subject: vxWorks and NFS Subject: Does TrueFFS come with Tornado 2 Subject: Re: Hierarchical projects under Tornado II? Subject: Please let's me know "What kind of OS is VXWORKS?" ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: open() for a large file Input/Output Date: Thu, 16 Sep 1999 13:25:29 +0200 From: "Achim Zimmer" Organization: Colt Telecom GmbH Message-ID: <7rqk1q$nng$1@crusher.de.colt.net> References: <7ros60$tpk$1@nnrp1.deja.com> If you are using FTP then at open the entire file is readed in memory. jack_shabel@ictv.com schrieb in Nachricht <7ros60$tpk$1@nnrp1.deja.com>... >I am trying to open() a file that resides on my NT machine. When the >file is very large (> 10MB), I get an open() error. If the file is >small enough, I can open the file, read it in smaller chunks, & process >the data within my VxWorks program. Does the system read the total file >into the embedded memory & then buffer the data out on each read? Is >there anyway I can open a large file and buffer the data in as smaller >records? I seem to get system malloc errors when I open a large file. I >also get similar errors when my output file is too large. Any help >would be greatly appreciated. I am assuming the total file resides >within the VxWorks memory until the close function is called. > Thanks Jack Shabel > > >Sent via Deja.com http://www.deja.com/ >Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxSim and rpccore timeout Date: Thu, 16 Sep 1999 14:43:23 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <37E0E5EB.9258EB1E@bitband.com> References: <37DFD519.DFCF3EF9@lucent.com> Reply-To: leonid@bitband.com Your IP address 90.0.0.255 is not a valid address when used with the 24 bit netmask. Try to change the IP address to say 90.0.0.253 ... Forrest Coleman wrote: > I am using the full simulator and trying to establish multiple simulator > target servers on the same NT host. I have the following set as the ulip > driver: > > * IP Address - 90.0.0.255 > * Subnet mask - 255.255.255.0 > * Default Gateway - 90.0.0.254 > > I have built my image according to the Tornado 2.0 documentation and use > the following to start up my target server: > > tgtsvr.exe 90.0.0.1 -n vxsim0 -A -s -V -Bt 10 -m 30720000 -Br > 2 -R c:\temp\tfsf -RW -B wdbrpc > > I consistently get "rpccore backend client Timed out" message in the > console window. > > Any ideas? > > Forrest Coleman > flcoleman@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Task Execution Time. Date: Thu, 16 Sep 1999 13:20:35 GMT From: colombo_daniele@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rqqqn$9qp$1@nnrp1.deja.com> References: <7rls2n$kgi$1@nslave1.tin.it> In article <7rls2n$kgi$1@nslave1.tin.it>, "Massimo Maio" wrote: > Hi VxWorkers ! > > > I need to evaluate the ticks (or the milliseconds) elapsed between the time > the task begins to execute the first instruction after semTake, and the time > at which it has completed to execute the last instruction of the FOREVER > cycle. > Can you help me ? You can include TIMESTAMP library (#define INCLUDE_TIMESTAMP) For more detail you can see the Application Note: Creating a Vxworks Timestamp driver. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ISA Plug & Play Date: Thu, 16 Sep 1999 15:51:16 +0200 From: "Christian Doppelbauer" Organization: VBS-Newsserver Message-ID: <937489979.284532@news.vbs.at> References: <199909160842.JAA08816@dsl-2.tadpole.co.uk> David, i spent a lot of time to reading the specs, got some tips to linux sources and ported some of them to vxworks. Some of my cards are working fine, some others i can't isolate with pnp protocol. So after all we decided to use the information provided by the PNP BIOS in the ESCD area. Now we intend to read the configuration and start our drivers with the actual settings. Thanks to all for their help. Christian David Laight wrote in message news:199909160842.JAA08816@dsl-2.tadpole.co.uk... > Christian, > > I have written ISA PnP grope code in the past, its not TOO difficult. > (Code is owned by a previous employer.) > > The only problem I recall is that you can't find the read data port the system > BIOS used - so if you make any BIOS calls that might try to access the ISA PnP > registers they will fail. (Unlikely on the sort of OS where you are trying to > grope the cards.) > > I seem to recall trying a few read data ports, try starting at the high end of > the valid addresses (but avoid the device at 3fx). > > The ISA PnP cards have a 'required for boot' bit which forces the BIOS to > allocate resources. The BIOS config SHOULD also have a question which will > cause it to allocate resources to all ISA PnP cards, the name of the question is > BIOS dependant... > > We did have some problems with the PnP config data on some cards - and ended up > writing a utility to modify it so that only specif configurations were valid. > (The manufacturers program didn't like running of a DOS boot floppy on our > hardware - which was running UnixWare 2.) > > Once you have identified the cards you can read back the io and memory base > addresses (but not the size - parse the PnP info for that). Bung it all into a > table for other vxworks drivers to search. > > Have fun :-) > > David > > > > ---------------------------------------------------------------- > David Laight email: dsl@tadpole.co.uk > Tadpole Technology plc phone: +44 1223 278 256 > Cambridge, UK fax: +44 1223 278 201 > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ENOBUFS for TCP over sockets Date: 16 Sep 1999 10:37:14 PDT From: "Dinesh Nambisan" Organization: Concentric Internet Services Message-ID: <7rr9sa$3ug@chronicle.concentric.net> References: <7rprir$k2u$1@nnrp1.deja.com> I've seen this error too[though on UDP]; and I did what you did [i.e. increase the network buffers] the only workaround which worked for us here was to increase the priority of your tasks which use the TCP sockets; possibly lower priority tasks dont get enough CPU juice to flush out the network packets and it results in buffers being filled up quickly... Hope it helps, Cheers, Dinesh N. wrote in message news:7rprir$k2u$1@nnrp1.deja.com... > Hi, > > I'm using Tornado 1.0.1 and SENS. When we make 30 TCP connections > over socket and transmite small data packets (about 40 bytes per > packet, and about 30 packets per connection), we get S_errno_ENOBUFS > from send(). I've already modifed the mblks to this: > > #define NUM_NET_MBLKS 1500 > #define NUM_SYS_64 250 /* 64 byte system clusters */ > #define NUM_SYS_128 250 /* 128 byte system clusters */ > #define NUM_SYS_256 250 /* 256 byte system clusters */ > #define NUM_SYS_512 250 /* 512 byte system clusters */ > #define NUM_SYS_CL_BLKS (NUM_SYS_64 + NUM_SYS_128 + \ > NUM_SYS_256 + NUM_SYS_512) > #define NUM_SYS_MBLKS (2 * NUM_SYS_CL_BLKS) > > What else do I have to change? and are these numbers ok? > > also, I know the command netStackSysPoolShow shows the mblks, but is > there another command that will show the 1024 and 2048 mblks? > > Thank you! > > > > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Task Execution Time. Date: Thu, 16 Sep 1999 18:44:15 GMT From: Curt McDowell Organization: Broadcom Corp. Message-ID: <37E13B93.51C5A149@broadcom.com> References: <7rls2n$kgi$1@nslave1.tin.it> <37DFABF4.48BB1C51@americasm01.nt.com> Reply-To: csm@broadcom.com Tim Sohacki wrote: > > For higher resolution on the PPC, you can read the decrementer > register with vxDecGet(). It counts down, one unit equals four > bus clock cycles. You can set it in advance with vxDecSet(). > > Cheers, > Tim. It may be easier to use the PPC timebase register instead, if available in your version of the PPC. The timebase is a 64-bit value that counts up at the same rate as the decrementer. The problems with the decrementer are that it wraps very often, it loses cycles when reloaded, you have to worry about what the reload value is, and you wouldn't want to write it because it's probably already being used for the system clock. Here's my favorite time measuring routine using the timebase, with a precision of 60 nsec, assuming the decrementer is running at 16.666666 MHz. #define TIMEBASE_HZ 16666666 #define TIMEBASE_PERIOD (1.0 / TIMEBASE_HZ) double double_time(void) { UINT32 tbu, tbl; vxTimeBaseGet(&tbu, &tbl); /* Get 64-bit value */ return (tbu * 4294967296.0 + tbl) * TIMEBASE_PERIOD; } It simply returns the time in seconds; for example: void benchmark(void) { double ts, te; ts = double_time(); my_code(); te = double_time(); printf("Elapsed time = %f sec\n", te - ts); printf(" = %f usec\n", (te - ts) * 1.0e6); } If you're measuring a very short interval, in order to get accurate results, you may want to surround the benchmark with il=intLock()...intUnlock(il), assuming my_code() allows it. You could also avoid using doubles without too much trouble, but doubles happen to be an ideal type for manipulating time values on this kind of processor. Curt McDowell Broadcom Corp. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: open() for a large file Input/Output Date: Thu, 16 Sep 1999 19:12:06 GMT From: Curt McDowell Organization: Broadcom Corp. Message-ID: <37E1421B.64E08F3F@broadcom.com> References: <7ros60$tpk$1@nnrp1.deja.com> <7rqk1q$nng$1@crusher.de.colt.net> Reply-To: csm@broadcom.com Achim Zimmer wrote: > > If you are using FTP then at open the entire file is readed in memory. netDrv is the one that copies the whole file. You won't have this problem if you use ftpXfer or rcmd directly. But then you can't seek. You might also look into nfsDrv. Curt --------------------------- Newsgroups: comp.os.vxworks Subject: Software Engineers, MPLS Protocol "Job Posting" Date: 14 Sep 1999 21:24:32 GMT From: macchio1@aol.com (Macchio1) Organization: AOL http://www.aol.com Message-ID: <19990914172432.02771.00000225@ng-cf1.aol.com> Software Engineers, MPLS Protocol We are looking for a key contributor with experience in implementing label switching protocols, especially MPLS for carrier class telecommunications network elements. Duties/Responsibilities · You will implement MPLS module for a carrier class Edge Switching Systems. · Design and specify mechanisms to utilize the power of Companies IP switching fabric with CAM lookup. · Specify, design, implement and test label switching protocol. · Work closely with the embedded software team to integrate the switching modules with Companies RTOS, CLI and management agents. · Work closely with our customers to specify and implement routing and tunneling policies for emerging IP services Qualifications · 3 - 5 years of relevant software development experience, preferably using VxWorks. Position is located in CA. Salary is open, also this is a pre-ipo company with major stock incentives call or contact Vito Macchio at 305-937-7364 or vito@sourcegroupinc.com --------------------------- Newsgroups: comp.os.vxworks Subject: Hierarchical projects under Tornado II? Date: Thu, 16 Sep 1999 12:57:09 +0200 From: "Borkhuis, Johan" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <19990916015417.01471.00001746@ng-cg1.aol.com> > ---------- > From: ajeddeloh@aol.com (AJeddeloh)[SMTP:ajeddeloh@aol.com] > Posted At: donderdag 16 september 1999 7:54 > Posted To: vxworks > Conversation: Hierarchical projects under Tornado II? > Subject: Hierarchical projects under Tornado II? > > Has anyone found a way to use the Tornado II "project" facility for > a non-trivial product? I have approx. 1000 source files in a > hierarchical > directory structure. How can I set up a build process under Tornado > II > without ending up with every object file landing in the same place? > We had the same problem, and were not able to fix it. The way we did this was to create a lot of projects in a workspace, each generating its own .o file. Then we created one project that contained all the files. Now you can compile only part of your project to test it. Another way is to create a project that only includes the .o files. Now you have to generate each project seperatly and then create the complete project. Both solutions not really workable, but then 1 project containing 1000 files ..... The best solution for this kind of projects is to use Tornado only for debugging (so you need one large project), and use make to generate your code. Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: open() for a large file Input/Output Date: Thu, 16 Sep 1999 13:04:09 +0200 From: "Borkhuis, Johan" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <7ros60$tpk$1@nnrp1.deja.com> > ---------- > From: jack_shabel@ictv.com[SMTP:jack_shabel@ictv.com] > Posted At: woensdag 15 september 1999 21:31 > Posted To: vxworks > Conversation: open() for a large file Input/Output > Subject: open() for a large file Input/Output > > I am trying to open() a file that resides on my NT machine. When the > file is very large (> 10MB), I get an open() error. If the file is > small enough, I can open the file, read it in smaller chunks, & > process > the data within my VxWorks program. Does the system read the total > file > into the embedded memory & then buffer the data out on each read? > What kind of file system do you use? A little while ago there was a thread about this, and the conclusion was that netdrv is acting like this. You could try to use NFS. > Thanks Jack Shabel > Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks FTP server Date: Thu, 16 Sep 1999 19:15:13 -0700 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <37E1A431.7BCB@despam.pangea.ca> References: <37E09938.4FD4AD4B@italtel.it> Marco Raimondi wrote: This part of an answer for you. > If the client, for any reason, can't close the connection, how much time > the connected ftp server process can live? > I have seen them last an hour, and it probably would have lasted forever, but I had work to do & reset the target. It seems to me WRS knew about it... - -het - -- "you don't really appreciate how smart a moron is until you try to design a robot..." -a robotics designer Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks and NFS Date: Thu, 16 Sep 1999 17:44:29 -0700 From: Hal Moroff Organization: Another Netscape News Server User Message-ID: <37E18EED.3189E537@kla-tencor.com> From within vxWorks I'm trying to open a file on my host that is larger than 2 Gb. My host is Solaris 7 and can support "very large" files. With vxWorks 5.3.1 once the file size crosses that 2Gb boundary then open() (and stat()) fail. WindRiver provided me a pre-release of dosFs 2.0 and claimed that that would fix my problem. It hasn't, and support has been non-responsive since then. Has anyone any experience with large files on vxWorks? Is there an implementation of an NFS client that will run on vxWorks available in source form that I could perhaps modify to access > 2Gb files? I only need to open 1 file for reading, so it can be a limited implementation. --------------------------- Newsgroups: comp.os.vxworks Subject: Does TrueFFS come with Tornado 2 Date: Fri, 17 Sep 1999 02:01:27 GMT From: js_j98@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rs7di$bs0$1@nnrp1.deja.com> We purchased the standard Tornado 2 package. I noticed it includes TrueFFS lib (can be included, compiled, not tried). Windriver also sells TrueFFS for $4000 or so. What is the difference between the ones come with Tornado2 standard package and the TrueFFS they sell separately? Thanks, Jun Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hierarchical projects under Tornado II? Date: Fri, 17 Sep 1999 07:46:55 +0200 From: Ole Asbjorn Fadum Organization: Telenor Online Public Access Message-ID: <37E1D5CF.BCBB8719@scanmar.no> References: <19990916015417.01471.00001746@ng-cg1.aol.com> Hi, In 'Tornado Release Notes 2.0' on page 21 has a description. Not very elegant but it works. AJeddeloh wrote: > Has anyone found a way to use the Tornado II "project" facility for > a non-trivial product? I have approx. 1000 source files in a hierarchical > directory structure. How can I set up a build process under Tornado II > without ending up with every object file landing in the same place? > > This is legacy code and does not lend itself to be broken into "Kernel" > vs "application" projects. > > This is particularly nasty, since I have multiple 3rd-party vendor code > sets which have to be integrated into the build, some of which duplicate > file names! > > Please respond to Alan_Jeddeloh "AT" adc.com. This AOL account does not > email from outside AOL. --------------------------- Newsgroups: comp.os.vxworks Subject: Please let's me know "What kind of OS is VXWORKS?" Date: Fri, 17 Sep 1999 19:38:59 +0900 From: "Hyunsu Kim" Organization: Seoul National University, Republic of Korea Message-ID: <7rt5f9$fnf$1@news.kren.nm.kr> Hello, Guy! I see first time with VXworks OS. A machine which I use change OS DOS to VXworks. I never heard ever. Please, let's me know "what kind of OS is VXworks?" Thanks. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Fri Sep 17 14:47:01 1999 From: Andy Foster Date: Fri Sep 17 14:47:05 PDT 1999 Subject: VxWorks, VxMP Is there anything to be especially aware of with using VxMP between two PowerPC processors? I noticed that one of the hardware requirements is: "Indivisible hardware read-write-modify mechanism across the shared memory bus" and wondered if this causes any problems with Power PC processors. We use the MVME2700 board. Thanks, Andy =================================================================================== Andy Foster Lead Programmer, Real-Time Systems Gemini 8-m Telescopes Project Tel: 1 (808) 974 2556 (direct) Gemini-North Base Facility Fax: 1 (808) 935 9235 670 North A'ohoku Place University Park E-mail: afoster@gemini.edu or ajf@noao.edu Hilo Home Page: http://controls.gemini.edu/ajf Hawaii 96720 USA From owner-vxwexplo-process Sat Sep 18 04:00:18 1999 From: daemon@csg.lbl.gov Date: Sat Sep 18 04:00:22 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Sep 18 04:00:05 PDT 1999 Subject: Re: NCR810 scsi initialization on vxworks Subject: Re: Does TrueFFS come with Tornado 2 Subject: VxSim and rpccore timeout Subject: auto-spawn? Subject: Boot from flash Subject: Re: Disappointed with Tornado II. Subject: Re: Happy with support and no-source? Subject: Re: Boot from flash Subject: ENOBUFS for TCP over sockets ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: NCR810 scsi initialization on vxworks Date: Sat, 18 Sep 1999 00:15:38 +0900 From: Young Jin Nam Organization: Postech News Server Message-ID: <37E25B1A.CAA132CF@postech.ac.kr> References: <7rpk1k$jfc$1@nntp.msstate.edu> > vxWorkers, > > I am trying to get the NCR810 working on the yellowknife/Pentium. I could get the Adaptec2940 working on the > pentium but the driver doesn't seem to support the ppc architecture.. > > back to NCR... > > I am not able to get this working on either the pentiums or the yelloknife. I followed the same procedure as for the > adaptec card and also changed the ncr810DelayCount to 100/1000 and tried all combinations and it looks like there is > no change in its behaviour...it just keeps hanging when I try to do the scsiAutoConfig.(I also tried doing the > scsiBusReset).. Initialize the "ncr810PciMemOffset" variable to zero(0) in the ncr810PciInit routine (ncr810PciMemOffset = 0) By default, this value is set with "PCI_TO_MEM_OFFSET" of 0x80000000 (refer to ncr810.h). Hope it'll be of help. Young Jin > > > Also the card doesn't have any onboard bois ... > oops ..bios > > can anybody suggest as to how I should proceed with this problem.. > > thanks in advance > > sri --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does TrueFFS come with Tornado 2 Date: Fri, 17 Sep 1999 10:35:18 -0700 From: Mark Dailey Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <0221c012.6a9a2bc5@usw-ex0102-015.remarq.com> References: <7rs7di$bs0$1@nnrp1.deja.com> You need to pay the $4000 to get a couple of key objects. * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: VxSim and rpccore timeout Date: Wed, 15 Sep 1999 11:19:21 -0600 From: Forrest Coleman Organization: Lucent Technologies, Columbus, Ohio Message-ID: <37DFD519.DFCF3EF9@lucent.com> I am using the full simulator and trying to establish multiple simulator target servers on the same NT host. I have the following set as the ulip driver: * IP Address - 90.0.0.255 * Subnet mask - 255.255.255.0 * Default Gateway - 90.0.0.254 I have built my image according to the Tornado 2.0 documentation and use the following to start up my target server: tgtsvr.exe 90.0.0.1 -n vxsim0 -A -s -V -Bt 10 -m 30720000 -Br 2 -R c:\temp\tfsf -RW -B wdbrpc I consistently get "rpccore backend client Timed out" message in the console window. Any ideas? Forrest Coleman flcoleman@lucent.com --------------------------- Newsgroups: comp.os.vxworks Subject: auto-spawn? Date: 17 Sep 1999 22:21:24 -0000 From: Kin Cho Organization: mail2news@nym.alias.net Message-ID: <19990917222124.13763.qmail@nym.alias.net> Hi, When VxWorks boots up, I want it to automatically spawn my task. How do I arrange for this to happen? Thank. - -kin --------------------------- Newsgroups: comp.os.vxworks Subject: Boot from flash Date: Sat, 18 Sep 1999 00:06:14 GMT From: js_j98@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rul1f$3p6$1@nnrp1.deja.com> Suppose both bootrom and vxWorks images are stored in flash. bootrom is stored between 0-0x8000 and 0x8001-0x100000 for vxWorks (see below). Are there any vxworks functions (bootrom) that loads vxworks image from the flash into the RAM and executes? I can only find the function that loads image from a file (bootLoadModule) on a dev (ide,network,treeFFS) etc.. Do I need to use memDrv or ramDrv to fake a file system to load the image? -------------- 0x100000 (flash) | | | vxworks | | (ELF) | | | | | | | | | |------------| 0x008000 | bootrom | | | -------------- 0x000000 Thanks, Jun Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Disappointed with Tornado II. Date: Thu, 16 Sep 1999 10:21:27 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37E10AF7.34359C79@ciprico.com> References: <7rp41k$3ot$1@nnrp1.deja.com> rbose@my-deja.com wrote: > > We just recently cut over to Tornado 2.0 from Tornado 1.0.1 with > somewhat disastrous results. [....] Don't know if people will see two of this message. Had some trouble with posting and it's always a mystery where the protocol terminated.... I was a bit disappointed with T2 also, but haven't had the symptoms you report. Which cpu are you targeting? We are using i960 and I estimate that on my 300MHz P2 gdb starts in ten to fifteen seconds. And I'm pretty sure it doesn't use 400Mb. The top hog on my box is invariably Netscape with around 14-15Mb, so it takes less than that. On the other hand, our image is only 2.3Mb with T2, and there may be some greater-than-linear cost for larger images. I guess I'd say a 400Mb debugger image is a pretty serious problem. WRS seems to have trouble forwarding problems to people that actually know something about them. I'd call your salesman and tell him (or her) that this is a show-stopper and you want prompt action. That may get your TSR to someone in the compiler group within a few weeks, and they may actually do something. Regards, Mike Beede - -- mbeede@ciprico.com beede@skypoint.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Happy with support and no-source? Date: Tue, 14 Sep 1999 11:22:04 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37DE762C.E1FE07F@ciprico.com> References: <37DD85D4.19110FB5@a.cc> Marjorie Krueger wrote: > > The WindRiver support contacts I've had so far remind me of trying to get support from my ISP - I > usually seem to get a response that indicates they know less than I do. I can't say that I'm happy > with their support. That describes our experience pretty well, too. It seems ludicrous to have two layers of entry-level people screening calls to support. I'd expect most people calling to be fairly savvy types or they wouldn't have bought vxWorks in the first place. They are prompt about acknowleging your bug report, but typically nothing happens after that for a LONG TIME. For instance, today I got a message that a compiler bug we reported in late July is being forwarded to the compiler group. And this was a specific bug report that was easily replicated. On another support-related front: my recent experience suggests that the vaunted "user configurable Tornado Environment" is marketing gas. We wanted to make some minor changes to the way Tornado II worked. The documentation supplied was grossly inadequate. Our local support person's response was to suggest that our changes would amount to a complete rewrite of Tornado and we should buy a Tornado source license. (In case anyone's interested, we wanted to remove the autocomplete feature from the gdb command line tool--hardly sounds like a complete rewrite to me, since the code setting up the tool _is_ included). I decided that rather than spend a week playing with their apparently-proprietary GUI modifications to see if there was some option that worked, I'd just live with it. We filed a TSR for one of the tools that RTI produces, and RTI's response was very good. WRS forwarded it to RTI, and within a day we had a clear, correct response from an actual technical person. And just so everyone doesn't think I hate all the WRS stuff, we just got a demo copy of WindView, and it is very nice. Makes it easy to track down what your code is really doing (not that there's ever any question with _my_ code....). Mike Beede - -- mbeede@ciprico.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Boot from flash Date: Sat, 18 Sep 1999 03:15:28 GMT From: peter_mcconaghy@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rv04a$auj$1@nnrp1.deja.com> References: <7rul1f$3p6$1@nnrp1.deja.com> In article <7rul1f$3p6$1@nnrp1.deja.com>, js_j98@my-deja.com wrote: > Suppose both bootrom and vxWorks images are stored in flash. bootrom is > stored between 0-0x8000 and 0x8001-0x100000 for vxWorks (see below). > Are there any vxworks functions (bootrom) that loads vxworks image > from the flash into the RAM and executes? I can only find the function > that loads image from a file (bootLoadModule) on a dev > (ide,network,treeFFS) etc.. Do I need to use memDrv or ramDrv to fake a > file system to load the image? > I think you'll have to write your own. If you don't need to compress the image it shouldn't be too hard - just do a "memcpy" or something similar and then jump to the start. You may want to add a header to the file which has a CRC, start address, etc in it. Or you could compress the image in your makefile and then "inflate" it using the vxWorks inflate function. regards Peter Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: ENOBUFS for TCP over sockets Date: Thu, 16 Sep 1999 04:27:15 GMT From: ned@nal.utoronto.ca Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rprir$k2u$1@nnrp1.deja.com> Hi, I'm using Tornado 1.0.1 and SENS. When we make 30 TCP connections over socket and transmite small data packets (about 40 bytes per packet, and about 30 packets per connection), we get S_errno_ENOBUFS from send(). I've already modifed the mblks to this: #define NUM_NET_MBLKS 1500 #define NUM_SYS_64 250 /* 64 byte system clusters */ #define NUM_SYS_128 250 /* 128 byte system clusters */ #define NUM_SYS_256 250 /* 256 byte system clusters */ #define NUM_SYS_512 250 /* 512 byte system clusters */ #define NUM_SYS_CL_BLKS (NUM_SYS_64 + NUM_SYS_128 + \ NUM_SYS_256 + NUM_SYS_512) #define NUM_SYS_MBLKS (2 * NUM_SYS_CL_BLKS) What else do I have to change? and are these numbers ok? also, I know the command netStackSysPoolShow shows the mblks, but is there another command that will show the 1024 and 2048 mblks? Thank you! Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sun Sep 19 04:00:07 1999 From: daemon@csg.lbl.gov Date: Sun Sep 19 04:00:12 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Sep 19 04:00:04 PDT 1999 Subject: Re: comparisons with other OS's Subject: vxWorks development system Subject: Re: vxWorks and NFS ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: comparisons with other OS's Date: Thu, 16 Sep 1999 13:54:12 GMT From: Ted Dennison Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7rqspt$bft$1@nnrp1.deja.com> References: <7rm928$1h0$1@nnrp1.deja.com> <37E07B33.B91F6DAF@aracnet.com> In article <37E07B33.B91F6DAF@aracnet.com>, engineer_scotty wrote: > Ted Dennison wrote: > > > > Perhaps you are attacking the problem at the wrong level. If you use > > Ada, process creation code is minimal and you don't have to make *any* > > system calls to do it. > > don't know...I don't know enough about the language to comment. (I > will ask a question--if Ada has the context of "processes" as a language > feature, how do these map onto system tasks on multitasking systems????) I used the minomer "process" because that is what the poster used to describe vxWorks tasks. Ada has the concept of tasks as a language feature ("task" is the actual keyword). Their definition is close enough to "threads" or "lightweight processes" as to be directly implementable using those features on OS's (like vxWorks) that support them. OS's that have only heavy unix-style processes or no multitasking at all are the ones that can require fancy footwork for an Ada implementation. One of the nifty things about doing your tasking in Ada is that you can make it source-portable with a little effort. We created our own frequency-based scheduler for vxWorks in Ada, and were able to compile and run it natively on our local NT boxes with *no* code changes (and no conditionally-compiled code). Of course it isn't real-time on NT, but I suspect we could port to about any other RTOS that multitasks just as easily. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks development system Date: Sat, 18 Sep 1999 22:04:52 +0800 From: "Simon Chan" Organization: City Telecom (HK) LTD Message-ID: <7s068p$hrf$1@newso.ctimail.com> I'm new in RTOS environment and want to know that where can I get the development software of vxWorks for the Pentium platform? Please help. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks and NFS Date: Sun, 19 Sep 1999 10:07:03 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <37E499A6.78D1FBAB@bitband.com> References: <37E18EED.3189E537@kla-tencor.com> Reply-To: leonid@bitband.com This is very silly to suggest DosFs 2.0 in order to solve an NFS client problem. These two have absolutely nothing in common. The NFS client is indeed limitted by its 32-bit file offset arithmetic. In order to support large files, the NFS client has to be reqritten not only to use 64-bit arithmetic, but also to use a newer version of the NFS protocol, which has the 64 offset fields in the NFS protocol structure. Just for the record, DosFs 2.0 supports large files handling on LOCAL DISK. Leonid Hal Moroff wrote: > From within vxWorks I'm trying to open a file on my host > that is larger than 2 Gb. My host is Solaris 7 and can support > "very large" files. > > With vxWorks 5.3.1 once the file size crosses that 2Gb boundary then > open() (and stat()) fail. > > WindRiver provided me a pre-release of dosFs 2.0 and claimed that that > would fix my problem. It hasn't, and > support has been non-responsive since then. > > Has anyone any experience with large files on vxWorks? > > Is there an implementation of an NFS client that will run on > vxWorks available in source form that I could perhaps modify to access > > 2Gb files? I only need to open 1 file > for reading, so it can be a limited implementation. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Sep 20 01:24:57 1999 From: "tim allen" Date: Mon Sep 20 01:25:01 PDT 1999 Subject: Re:VxWorks, VxMP >Is there anything to be especially aware of with >using VxMP between two PowerPC processors? > >I noticed that one of the hardware requirements is: > >"Indivisible hardware read-write-modify mechanism across >the shared memory bus" > >and wondered if this causes any problems with Power PC processors. > >We use the MVME2700 board. I can't comment on your particular target because I'm not familiar with it, but you need to be sure that the hardware is capable of both generating and receiving an indivisible cycle. As you're probably aware (because you're asking the question!) the PowerPC has a load-with-reservation command to implement this, but it doesn't necessarily work across multiple busses. The BSP's VME driver should include a sysBusTas() function to handle the generation of the external (VME) TAS/RMW. This is usually done by directly driving the VME I/F chip to generate a VME bus capture and hold, while the RMW takes place. This function is used by VxMP when required. When the VME I/F detects an incoming RMW cycle then this needs to be propogated across PCI (or whatever) and local memory bus, to prevent the processor accessing local memory during the cycle. I have worked on platforms in the past where the various bus bridges had bugs preventing this working correctly. If that all works, there shouldn't be a problem! Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com From owner-vxwexplo-process Mon Sep 20 03:00:51 1999 From: David Laight Date: Mon Sep 20 03:00:56 PDT 1999 Subject: RE: Boot from flash Jun, I solved the problem a slightly different way. - I encapsulated the standard vxworks image into the bootrom code. The standard vxworks makefile doesn't contain the correct target, I've added the relevant part of the makefile below. You do need to fully understand the vxworks boot process in order get this right! The normal boot process is a multi stage process: 1) The code in flash copies itself to low dram (excluding the image) 2) jump into code in low DRAM 3) uncompress (or copy) bootrom image to high memory 4) jump into bootrom code in high memory 5) bootrom loads vxworks image to low memory 6) jump into vxworks coade in low memory. (RAM_HIGH_ADRS often has to be increased to stop (5) overwriting the executing code) To make the boot process enter vxWorks at (4), the uses of RAM_LOW_ADRS and RAM_HIGH_ADRS have to be reversed in the earlier stages. If you create a raw filesystem over the flash boot image you can update the vxworks image from within vxworks using copy(). Of course you do need a fallback position for when the vxworks image you are testing doesn't work at all! David > > Subject: Boot from flash > Date: Sat, 18 Sep 1999 00:06:14 GMT > From: js_j98@my-deja.com > > Suppose both bootrom and vxWorks images are stored in flash. bootrom is > stored between 0-0x8000 and 0x8001-0x100000 for vxWorks (see below). > Are there any vxworks functions (bootrom) that loads vxworks image > from the flash into the RAM and executes? I can only find the function > that loads image from a file (bootLoadModule) on a dev > (ide,network,treeFFS) etc.. Do I need to use memDrv or ramDrv to fake a > file system to load the image? > > > -------------- 0x100000 (flash) > | | > | vxworks | > | (ELF) | > | | > | | > | | > | | > |------------| 0x008000 > | bootrom | > | | > -------------- 0x000000 > > Thanks, > Jun ## standalone (compressed) copy of normal vxworks vxWorks.Z.s: vxWorks - @ $(RM) $@ - @ $(RM) vxWorks.Z.s $(EXTRACT_BIN) < vxWorks | $(COMPRESS) > tmp.Z $(BINTOASM) tmp.Z >vxWorks.Z.s - @ $(RM) tmp.Z vxWorks.kenya: vxWorks.Z.o bootInit_st.o romInit.o - @ $(RM) $@ - @ $(RM) $(IMI) - @ $(RM) $(IBR) - @ $(RM) version.o version.sccs $(CC) -c $(CFLAGS) -o version.o $(CONFIG_ALL)/version.c $(LD) $(LDFLAGS) -e $(ROM_ENTRY) $(LD_HIGH_FLAGS) \ -o $@ romInit.o bootInit_st.o version.o $(BOOT_EXTRA) \ $(LIBS) vxWorks.Z.o $(ROMSIZEPROG) -b $(ROM_SIZE) $@ vxWorks.kenya.aif: vxWorks.kenya aifHdr.s - @ $(RM) $@ $(CC) $(CFLAGS) -DROM_BASE_ADRS=0x$(ROM_BASE_ADRS) -P \ -x assembler-with-cpp -c -o aifHdr.o aifHdr.s $(LD) $(LDFLAGS) $(LD_PARTIAL_FLAGS) \ -o vxWorks.tmp aifHdr.o vxWorks.kenya objcopyarm -O binary vxWorks.tmp $@ $(NM) vxWorks.tmp | sort > vxWorks.kenya.aif.nm - @ $(RM) vxWorks.tmp ---------------------------------------------------------------- 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 Sep 20 04:00:12 1999 From: daemon@csg.lbl.gov Date: Mon Sep 20 04:00:16 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Sep 20 04:00:07 PDT 1999 Subject: configuring network driver for vxWork image Subject: Re: vxWorks development system Subject: Re: Ethernet Problems Subject: Re: VME A32 addressing/MVME162 Subject: Re: CORBA on VxWorks ? Subject: Re: configuring network driver for vxWork image Subject: Re: configuring network driver for vxWork image Subject: Re: vxWorks development system Subject: Tornado2 license? Subject: re: comparisons with other OS's Subject: About Global variables and main() functions Subject: Re: About Global variables and main() functions Subject: Re: vxWorks development system Subject: Re: auto-spawn? Subject: test Subject: Re: About Global variables and main() functions Subject: Question about TCP socket and select() Subject: Re: auto-spawn? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: configuring network driver for vxWork image Date: Sun, 19 Sep 1999 06:27:35 -0700 From: Nidal Khalil Organization: SBC Internet Services Message-ID: <37E4E4C5.1F32DC4@pacbell.net> Reply-To: nidal@pacbell.net Hello, I am trying to build a vxWorks image with network driver in it. I am using tornadoII and vxWorks version 5.4. The network card that I am using is a standard intel ethernet pro 100 which is supported by vxWorks. If I boot vxWorks over the net the driver is configured properly but it I boot from a floppy or flash the network device is not attached to the tcp/ip stack. my config.h has all the declaration for including the network interface. #define INCLUDE_NETWORK #define INCLUDE_NET_SETUP #define INCLUDE_MUX #define INCLUDE_BSD #define INCLUDE_FEI #define INCLUDE_LOOPBACK #define INCLUDE_PING #define INCLUDE_NETSHOW usrNetwork.c, the module the parses the boot parameter and configure the network interface work. According to the vxWorks Network programmer guide all I need to do with this setup is assign the ip address with ifAddrSet() and the mask with ifMaskSet(). Both calls result in an error since the fei0 is not attached to the network stack. I tried using usNetIfAttach() but this result in an unresolved symbol. I tried usrNetIfConfig() but this call does not succeed. Does any one know how to attach the network device to the TCP/IP stack. I do appreciate the help Nidal --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks development system Date: Sun, 19 Sep 1999 06:29:46 -0700 From: Nidal Khalil Organization: SBC Internet Services Message-ID: <37E4E54A.BCAED41D@pacbell.net> References: <7s068p$hrf$1@newso.ctimail.com> Reply-To: nidal@pacbell.net Yes you can get vxWorks for the following targets: 486, pentium, pentium-pro, i960, and the arm chip with development host either on unix or nt. Simon Chan wrote: > I'm new in RTOS environment and want to know that where can I get the > development software of vxWorks for the Pentium platform? Please help. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ethernet Problems Date: Sun, 19 Sep 1999 06:51:44 -0700 From: Nidal Khalil Organization: SBC Internet Services Message-ID: <37E4EA6F.C7701387@pacbell.net> References: Reply-To: nidal@pacbell.net This is the same problem that I am having. I think it is a setup problem. I posted a description of it but the target that I am using is a Pentium Steeve Gagne wrote: > My mpc860 bsp boot from flash > He download a load image from TFTP and it work fine > > When I boot the load image, the board not answer to ping > and telnet from exterior. > > On the Net, when I throw a echo request paquet > from the developpement board, I catch the reply on the net but the board > didn't receive it. > > The Ip addr, the subnet mask, the Mac Address All It's ok > > I suspect an interrupt problem. > > Any Ideas ?? > > Steeve Gagne > Firmware Designer > steevegagne@sympatico.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME A32 addressing/MVME162 Date: Sun, 19 Sep 1999 08:18:50 -0700 From: "Steve Doiel" Message-ID: <37e4fe36.0@news.pacifier.com> References: <37E0233B.E69F041@navair.navy.mil> Reply-To: "Steve Doiel" Hint: The 162 likes to cache memory. You may need to call vmBaseStateSet to mark the memory as uncachable before accessing. SteveD Kevin Havig wrote in message news:37E0233B.E69F041@navair.navy.mil... > I am using VxWorks 5.4 on a Motorola MVME162 processor. I am having a > problem accessing a board in the VME 32 bit address space; address > 0x10000000 to be exact. The processor has no problem accessing any > boards in the VME 16 or 32 bit address space. Furthermore if I boot the > processor to the built in Motorola debugger, it accesses this location > just fine. Is there some trick to getting VxWorks to access this > board. Thanks in advance for all of your help. > Kevin Havig > Naval Air Warfare Center > havigkk@navair.navy.mil > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CORBA on VxWorks ? Date: Sun, 19 Sep 1999 17:29:54 GMT From: Murat Bog Organization: @Home Network Message-ID: <37E51EFC.73960DB6@technologist.com> References: <7r8v77$s6v$1@feed.teaser.fr> <7rlo02$jlu@newsserv.vs.dasa.de> Hi, You may take a look at the TAO from University of Washinton in St Louis. It is developed in C++ and has been ported to VxWorks (don't remember which version). Holger Roos wrote: > I'm just starting with VisiBroker (Highlander / Inprise) ... > > Bertrand Marlier schrieb in Nachricht <7r8v77$s6v$1@feed.teaser.fr>... > >Does anyone have an experience of CORBA on VxWorks ? > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Sun, 19 Sep 1999 22:07:27 GMT From: johan@borksoft.xs4all.nl (Johan Borkhuis) Organization: JAB Software Productions Message-ID: References: <37E4E4C5.1F32DC4@pacbell.net> Nidal Khalil (nidal@pacbell.net) wrote: > Hello, > I am trying to build a vxWorks image with network driver in it. > I am using tornadoII and vxWorks version 5.4. > The network card that I am using is a standard intel ethernet pro 100 > which is supported by vxWorks. If I boot vxWorks over the net the driver > is configured properly but it I boot from a floppy or flash the network > device is not attached to the tcp/ip stack. my config.h has all the > declaration for including the network interface. ..... > > Does any one know how to attach the network device to the TCP/IP stack. > I do appreciate the help Did you configure the network driver in your project? The configuration in config.h is not used, but the configuration in your project is used, so use the project facility to configure this. 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` ==== VxWorks-page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html ==== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Sun, 19 Sep 1999 16:31:32 -0700 From: Nidal Khalil Organization: SBC Internet Services Message-ID: <37E57253.6F674C54@pacbell.net> References: <37E4E4C5.1F32DC4@pacbell.net> Reply-To: nidal@pacbell.net Hello all, I think that I undrestand the heirachy of tornado II configuration. The project is highest order then config.h then configAll.h . Yes the network drivers are in my project. Johan Borkhuis wrote: > Nidal Khalil (nidal@pacbell.net) wrote: > > Hello, > > I am trying to build a vxWorks image with network driver in it. > > I am using tornadoII and vxWorks version 5.4. > > The network card that I am using is a standard intel ethernet pro 100 > > which is supported by vxWorks. If I boot vxWorks over the net the driver > > is configured properly but it I boot from a floppy or flash the network > > device is not attached to the tcp/ip stack. my config.h has all the > > declaration for including the network interface. > ..... > > > > Does any one know how to attach the network device to the TCP/IP stack. > > I do appreciate the help > > Did you configure the network driver in your project? The configuration in > config.h is not used, but the configuration in your project is used, so use > the project facility to configure this. > > 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` > > ==== VxWorks-page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html ==== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks development system Date: Mon, 20 Sep 1999 02:47:49 GMT From: John or Jenn Organization: Who, us organized??? Message-ID: References: <7s068p$hrf$1@newso.ctimail.com> In article <37E4E54A.BCAED41D@pacbell.net> Nidal Khalil, nidal@pacbell.net writes: >Yes you can get vxWorks for the following targets: >486, pentium, pentium-pro, i960, and the arm chip with development host and several more besides... --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado2 license? Date: Mon, 20 Sep 1999 14:43:30 +0900 From: "moony" Organization: System Engineering Research Institute (SERI) Message-ID: <7s4hm5$p87$1@green.kreonet.re.kr> hi.. We have bought Tornado2 lately.. But.. we didn't get license yet.. Hmm... it is different from Tornado1.1.. How can we get the license? Please help us... Could you explain the process in details? Thank you.. --------------------------- Newsgroups: comp.os.vxworks Subject: re: comparisons with other OS's Date: Fri, 17 Sep 1999 20:31:03 GMT From: Ted Dennison Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7ru8e7$qq3$1@nnrp1.deja.com> References: <7rr8id$r39@overload.lbl.gov> In article <7rr8id$r39@overload.lbl.gov>, Ian.Kerr@bae.co.uk wrote: > Ada tasks are asynchronous processes implemented `as threads of control' > in a typical Ada run-time system. In the general sense the word "process" is accurate, but in this day and age it has connotations that aren't accurate for Ada tasks. Ada tasks implement separate threads of control which have their own stacks, but also share global memory space with the rest of the program. Today OS folks call that kind of construct a "thread" or "lightweight process". The problem with relying on OS features to handle your multiprocessing needs is that each OS has a completely different set of features. That means your multitasking code isn't very portable. It also means developers have go through the learning process anew for each OS. If that's a major concern for you, as it apparently was for the original poster, you should seriously consider Ada, which has extensive support for multitasking and a strict standard. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: About Global variables and main() functions Date: Mon, 20 Sep 1999 15:07:51 +0900 From: Lee Dong Yul Organization: Seoul National University, Republic of Korea Message-ID: <37E5CF37.F7582619@cselab.snu.ac.kr> Hello, We are developing some programs on VxWorks OS. 1. Why can't we use main function with C language? When we programmed some applications with main function, We couldn't download those applications. Undefined symbol (__eabi) error occurred. 2. In tornado shell, we can't execute some programs many times. All Those programs have global variables. So, After executing those programs, Global variables contain some values that are different from program startup values. How do we initialize all global variables on program startup? Why don't vxworks initialize global variables on program startup. When Initialization occurs? Object code downloading? 3. On program exit, is dynamically allocated memory automatically freed? Otherwise, we must free all dynamically allocated memory explicitly? Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About Global variables and main() functions Date: Sun, 19 Sep 1999 23:33:39 -0700 From: engineer_scotty Organization: aracnet.com Message-ID: <37E5D543.9B7119E5@aracnet.com> References: <37E5CF37.F7582619@cselab.snu.ac.kr> Lee Dong Yul wrote: > > Hello, We are developing some programs on VxWorks OS. To sum up all your questions: vxWorks ain't like Unix. vxWorks doesn't have the concept of separate "programs" with separate address spaces, and separate global variables. There is only one address space--and when you load new modules with ld(), the global variables in that module are added to that address space. > 1. Why can't we use main function with C language? > > When we programmed some applications with main function, We couldn't > download those applications. Undefined symbol (__eabi) error occurred. Lemme guess. PowerPC, right? :) Including main() causes the compiler/linker to automagically link in a bunch of Other Stuff. In Unix, including main() causes prologue code (which initializes the C standard library) to run, etc. In C++, this prologue code initializes static constructors. vxWorks on the powerPC, on the other hand, corresponds to a specification called EABI (Embedded Application Binary Interface); which calls for a function __eabi to be invoked when a program starts. This is, however, Not What You Want in vxWorks. The bottom line: don't call anything main(). > 2. In tornado shell, we can't execute some programs many times. All > Those programs have global variables. So, After executing those > programs, Global variables contain some values that are different from > program startup values. > > How do we initialize all global variables on program startup? Do so explicitly. The static initialization happens only once. For those objects linked with the kernel, static initialization happens when the system starts. If you use the dynamic loader, the initialization happens when the module is loaded. However, "restarting" the program does not cause globals to be initialized. Oh, and one more thing. The automatic cleanup that you get in Unix when a program dies (all memory blocks freed, stdio flushed, all files closed) does NOT happen in vxWorks. You **MUST** free all resources you use yourself. > Why don't vxworks initialize global variables on program startup. vxWorks views the entire system as a "program"; all the modules are part of that program. > When Initialization occurs? Object code downloading? Globals defined in the module are initialized when you download the module, but at no time after. (Unless you unload the module and load it again; this will ause them to be re-initialized). > 3. On program exit, is dynamically allocated memory automatically freed? Nope. > Otherwise, we must free all dynamically allocated memory explicitly? Yup. > Thanks. You're welcome. :) - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks development system Date: Mon, 20 Sep 1999 06:45:17 GMT From: David Hamel Organization: @Home Networks. Message-ID: <37E5D814.DB7028A1@ntanet.net> References: <7s068p$hrf$1@newso.ctimail.com> You could also try answering the question that Simon asked instead of the one you thought he was asking. John or Jenn wrote: > > In article <37E4E54A.BCAED41D@pacbell.net> Nidal Khalil, > nidal@pacbell.net writes: > >Yes you can get vxWorks for the following targets: > >486, pentium, pentium-pro, i960, and the arm chip with development host > and several more besides... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: auto-spawn? Date: Mon, 20 Sep 1999 06:48:26 GMT From: David Hamel Organization: @Home Networks. Message-ID: <37E5D8D0.67AE5459@ntanet.net> References: <19990917222124.13763.qmail@nym.alias.net> Call taskSpawn() at the end of usrRoot(). Some people use the macro USER_APPL_INIT for this pupose. #ifdef INCLUDE_USER_APPL /* Startup the user's application */ USER_APPL_INIT; /* must be a valid C statement or block */ #endif } Kin Cho wrote: > > Hi, > > When VxWorks boots up, I want it to automatically spawn my task. > > How do I arrange for this to happen? > > Thank. > > -kin --------------------------- Newsgroups: comp.os.vxworks Subject: test Date: Tue, 7 Sep 1999 13:26:37 -0500 From: "Barry Grupe" Organization: SkyPoint Communications, Inc. Message-ID: <7r3llq$ptf$1@shadow.skypoint.net> test --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About Global variables and main() functions Date: Mon, 20 Sep 1999 19:17:27 +0900 From: Lee Dong Yul Organization: Seoul National University, Republic of Korea Message-ID: <37E609B6.B67879BF@cselab.snu.ac.kr> References: <37E5CF37.F7582619@cselab.snu.ac.kr> <37E5D543.9B7119E5@aracnet.com> Hello. Thanks for your reply. Do you inform me of EABI specification on VxWorks with MPC860 (PowerPC core)? What you said is that All global variables initialization is executed by programmer explicitly? dynamically allocated memory management,too ? engineer_scotty wrote: > Lee Dong Yul wrote: > > > > Hello, We are developing some programs on VxWorks OS. > > To sum up all your questions: vxWorks ain't like Unix. > > vxWorks doesn't have the concept of separate "programs" with > separate address spaces, and separate global variables. There is > only one address space--and when you load new modules with ld(), > the global variables in that module are added to that address space. > > > 1. Why can't we use main function with C language? > > > > When we programmed some applications with main function, We couldn't > > download those applications. Undefined symbol (__eabi) error occurred. > > Lemme guess. PowerPC, right? :) > > Including main() causes the compiler/linker to automagically link in a > bunch of Other Stuff. In Unix, including main() causes prologue code > (which initializes the C standard library) to run, etc. In C++, this > prologue code initializes static constructors. > > vxWorks on the powerPC, on the other hand, corresponds to a > specification > called EABI (Embedded Application Binary Interface); which calls for a > function __eabi to be invoked when a program starts. This is, however, > Not What You Want in vxWorks. > > The bottom line: don't call anything main(). --------------------------- Newsgroups: comp.os.vxworks Subject: Question about TCP socket and select() Date: Sat, 18 Sep 1999 17:26:47 GMT From: ned@nal.utoronto.ca Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7s0i0e$a9i$1@nnrp1.deja.com> Hi, We have a client/server application, the server is a PPC8860 vxWorks 5.3 with SENS, client is a NT4 winsock2. The vxWorks side has a thread that wait on sockets using select(), initially there is only one socket that's the server listening socket. When connections come in, the server accpet() the connection and the new socekt descriptor is added to the select and server go back to wait the listening and the new socket. What we found strange is that when NT initiates a burst of connection request, say 20, the first 8 to 9 conenctions are quickly accepted but the next one takes a long time for select to wake up (2 second). This means the select wakes up very quickly when it's waiting on 1, 2 ... 8 sockets, but when there are 9 sockets, it takes it a long time to do so. However, the 9th to 16th connections are fast again, but the 17th is a long wait again and so on. Every 8 to 9 conenctions it stuck for a while. We then modified the NT to wait 200ms between each connection request, this makes the situation a bit better, but occationally it still happens (20 conenctions etc). We have tried to modify NUM_??? and NUM_SYS_??? (??? is 64,128 etc) to increase the mbufs. We also tried to increase ethernet interface queue from 50 to 200 but not much improvement. WOuld this be a select problem? I doubt it since when we slow it down a bit select is behaving correctly, we are suspecting some buffers underneath are overrun and causes TCP to retransmit etc. Any suggestions for us to fix this will be greatly appreciated! Thank you! Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: auto-spawn? Date: Mon, 20 Sep 1999 10:10:29 GMT From: oliver_fuchs@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7s516h$717$1@nnrp1.deja.com> References: <19990917222124.13763.qmail@nym.alias.net> simply define INCLUDE_USER_APPL in configAll.h e.g. #define INCLUDE_USER_APPL taskSpawn ("myTask", 20, 0, 2000, (FUNCPTR)myTask, 0,0,0,0,0,0,0,0,0,0); at the end of usrRoot() you will find: #ifdef INCLUDE_USER_APPL /* Startup the user's application */ USER_APPL_INIT; /* must be a valid C statement or block */ #endif this will spawn your task when vxWorks boots up In article <19990917222124.13763.qmail@nym.alias.net>, Kin Cho wrote: > Hi, > > When VxWorks boots up, I want it to automatically spawn my task. > > How do I arrange for this to happen? > > Thank. > > -kin > > Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Sep 20 04:55:18 1999 From: "Subramania Pillai" Date: Mon Sep 20 04:55:22 PDT 1999 Subject: NetBufLib vxWorks Hi, I am looking for some kind of APIs to concatenate (or merge) two mBlk/clBlk/cluser construct into one. The requirement is similar to m_cat (struct mbuf *m, struct mbuf *n), this is a mbuf library routine which concatenate two mbufs into one mbuf chain. If there is no direct API calls, then how it can be done? Can anyone please tell me? Thanks pillai From owner-vxwexplo-process Tue Sep 21 00:40:23 1999 From: Jan Uerpmann Date: Tue Sep 21 00:40:27 PDT 1999 Subject: Screenblanker & Windview Problems Hi I am using VxWorks 5.3.1 and Tornado 1.0.1 on a pc486-architecture and have encountered some problems: - After a while the programs run about 5 times slower than they used to. I recognized that if I press a key on the target the program speeds up again. I think it has something to do with the screenblanker as the target is slow when the blanker is active. But the strange thing is that the target is already slower when the blanker is not yet active. However, a simple key-press from time to time solves this problem. Is there a bug-fix available? If not, could somewhere tell me, where the screenblanker is hidden in the sources? I already had a look but did not find it. - I have another problem with windview. The program is installed on a Solaris (SunOS?) workstation. A single-user license is installed too. But when I attempt to run windview, I get a requester saying: Your request for a WindView license has been queued. Your queue number is . Try Again Exit As there is no queue number this looks like a bug to me. BTW, I have entered the single-user license correctly in the form that asks for it. Also I have noticed that the version of windview is 1.0.2 while Tornado here is 1.0.1. Could there be a problem with this? (My admin told me that both were on the same CD though...) I would be glad for any help. Regards, Jan From owner-vxwexplo-process Tue Sep 21 02:06:34 1999 From: Ilia Bosis Date: Tue Sep 21 02:06:38 PDT 1999 Subject: RE: SLIP/ARP/PROXY in VxWorks Hi Rasty. The Tornado Proxy ARP server works over the shared memory backplane ONLY! It's _not intended_ for anything else and will refuse to work over the interface, other than "sm" :( . You can try to fool the Proxy ARP server by renaming Your SLIP interface as a shared-memory one, but do it on Yours own risk. Or maybe, use third-party Proxy ARP server or try to avoid proxy ARP'ing altogether. Regards Ilia. > -----Original Message----- > From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] > Sent: Thursday, September 16, 1999 5:37 PM > To: vxworks_users@csg.lbl.gov > Subject: SLIP/ARP/PROXY in VxWorks > > > Submitted-by owner-vxwexplo-process Thu Sep 16 08:36:33 1999 > Submitted-by: "Slutsker, Rasty" > > Hi, > Did anyone succeed to have Proxy ARP server with the SLIP? > I have five targets connected together though Ethernet, one > of them has also > SLIP connection to the host computer. I'd like to establish > connection > between host and any target. I initially planned to run Proxy > ARP server on > the target that has SLIP connection to the host, but > proxyNetCreate always > returns -1, no examples in manuals, tried a dozen of combinations ;-(. > > Any suggestions? > > Rasty > > Tornado VxWorks > +=====================================================================+ | Bosis Ilia | Work phone: 972-3-5384685 | | Software engineer | Fax: 972-3-5335877 | | Telegate ltd. | E-mail: iliab@telegate.co.il | | 60218 Haplada7 Or-Yehuda, Israel | WWW: http://www.telegate.co.il | +=====================================================================+ From owner-vxwexplo-process Tue Sep 21 04:00:13 1999 From: daemon@csg.lbl.gov Date: Tue Sep 21 04:00:18 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Sep 21 04:00:07 PDT 1999 Subject: Re: configuring network driver for vxWork image Subject: Re: configuring network driver for vxWork image Subject: Update Tornado Subject: memPartCreate() Problem Subject: Re: Disappointed with Tornado II. Subject: Re: configuring network driver for vxWork image Subject: Tornado2 for NT machine: Where's cpp? Subject: Re: Hierarchical projects under Tornado II? Subject: Re: Update Tornado Subject: GreenHills, VxWorks, and RogueWave Subject: Re: Update Tornado Subject: Re: About Global variables and main() functions Subject: On scsiRdSecs() routine (again) Subject: Re: Hierarchical projects under Tornado II? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Mon, 20 Sep 1999 08:27:04 -0500 From: Frank Minich Organization: NCP Message-ID: <37E63628.B79@ti.com> References: <37E4E4C5.1F32DC4@pacbell.net> Reply-To: fminich@ti.com Did you try inserting the name of your network device in the configuration parameters' "other" field? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: 20 Sep 1999 16:22:01 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <7s5mv9$pkg$1@news1.xs4all.nl> References: <37E4E4C5.1F32DC4@pacbell.net> <37E57253.6F674C54@pacbell.net> Nidal Khalil wrote: : Hello all, : I think that I undrestand the heirachy of tornado II configuration. The : project is highest order then config.h then configAll.h . Yes, that is right. : Yes the network drivers are in my project. Then it looks like a different problem, and I might have had the same problem. You probably have to define the O-bootparameter (Other). With this parameter you can indicate network devices that have to be included. If this does not start your network you might have to modify comps\src\net\usrNetBoot.c. Take a look at my VxWorks page (http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html) for this (section 3.2). Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Update Tornado Date: Mon, 20 Sep 1999 15:26:12 -0400 From: Lei Sun Organization: University of Central Florida Message-ID: <37E68A54.3A0DA1BC@pegasus.cc.ucf.edu> Hi : I recently updated the tornado 1.0 to tornado 2.0 on sun-solaris system. My target machine is MC68K with VME bus equipment. But after we finished installation and try to reboot the target, we didn't get it. it says "error loading file... errono=0xd0003". Anybody can give me any hint? what configuration job should I do after updated to tornado 2.0! thanks a lot lei sun MMAE dept. University of central Florida ORLANDO FL32816 --------------------------- Newsgroups: comp.os.vxworks Subject: memPartCreate() Problem Date: Mon, 20 Sep 1999 17:06:20 -0400 From: John Sestak Organization: Raytheon Systems Company, Falls Church Message-ID: <37E6A1CC.376E0B8F@fallschurch.esys.com> Hello, I'm having a little problem understanding the amount of memory that gets allocated in a memPartCreate() call. I'm running VxWorks on a PowerPC4 - 750 if that helps. First I create an array: UCHAR rx_buf[384]; /* space for receive buffer */ and then I create a partition using the call: rx_pool_ptid = memPartCreate((char *)rx_buf, (unsigned)(384)); which should create a partition of 384 bytes, but instead when I call a memPartShow() on this partition it says the partition is 368 bytes. Is there some sort of overhead with creating this partition (16 bytes)? Secondly, when I call: (void *)rx_pool[i].head_p = memPartAlloc(rx_pool_ptid, (unsigned)(16)); It actually allocates 24 bytes and not 16. Is there an overhead for this call too (8 bytes for each allocation)??? Thirdly, when I get down to having 32 bytes remaining in my partition the memPartAlloc() call fails even though I'm only trying to allocate just 16 bytes like before. Any ideas? Thanks for any help, John --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Disappointed with Tornado II. Date: Mon, 20 Sep 1999 21:34:32 GMT From: hurlerj@integnet.com (Joe Hurler) Message-ID: <37e6a1b0.26553872@bastion> References: <7rp41k$3ot$1@nnrp1.deja.com> We also recently made the transition to T2 for PPC860, and observe many of the issues you note. I have a small CORBA server application based on ACE/TAO running on vxWorks 5.4, and the file holding the executable is over 100 MBytes in size. But the in-memory footprint of the server is 3.9 MBytes (code + data). Loading this executable via gdb drives my "Commit Charge" in Windows NT Task Manager well up over 400000 (i.e. that's 400 MBytes of virtual memory), and takes so long I have given up gdb (for now) and gone over to using the host-based shell to load and run the executable (where it loads in less than 2 minutes). On a 266 MHz PII w/ 128 MB RAM, 704 MB pagefiles. I have been led to believe that most of the bloat in executables comes from the handling of templates and their instantiations by the GNU C++ compiler. ACE/TAO is template-heavy, and this might explain the issues I see. Perhaps this is part of the cause in your situation also? No idea at all why gdb should require so much VM, however. Joe On Wed, 15 Sep 1999 21:45:27 GMT, rbose@my-deja.com wrote: >We just recently cut over to Tornado 2.0 from Tornado 1.0.1 with >somewhat disastrous results. Our main incentive for this move was to use >the enhanced network stack. We use Objectime and much of the >remainder our code is written in C++. So the move to the new GNU C++ >compiler (based on the egcs 1.0.2 source base and enhanced by WRS) has >affected us tremendously. Some of the behavior is unacceptable and >is listed below: > >- Our debug image has bloated from around 22Mb to 45Mb. >- The Tornado 2.0 debugger uses up 350MB of virtual memory before we >have even attached to a task. >- It takes nearly 10 minutes for the debugger to start up. >- Trying to view mixed source and disassembly can take over 15 minutes >to come up, and sometimes crashes Tornado. >- Random loss of connection with the target server and Tornado crashes. >(not attributable to running out of virtual memory since there is enough >available). > >All this on a 400 MHz Pentium II with 256 Mb physical memory and 1GB >swap file (running NT). Doing a simple experiment with a "hello world" >program shows that the .o is nearly double the size with the new >compiler in comparison with the old one. We have tried various >combinations of the compiler flags (including -fno-rtti and >-fno-exceptions) but with no avail. > >Has anyone else seen any similar problems and managed to get around >them? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Mon, 20 Sep 1999 19:44:12 -0400 From: Henry Gessau Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: References: <37E4E4C5.1F32DC4@pacbell.net> Does it help if you #define STANDALONE_NET ? On Sun, 19 Sep 1999 06:27:35 -0700, Nidal Khalil wrote: >Hello, >I am trying to build a vxWorks image with network driver in it. >I am using tornadoII and vxWorks version 5.4. >The network card that I am using is a standard intel ethernet pro 100 >which is supported by vxWorks. If I boot vxWorks over the net the driver >is configured properly but it I boot from a floppy or flash the network >device is not attached to the tcp/ip stack. my config.h has all the >declaration for including the network interface. >#define INCLUDE_NETWORK >#define INCLUDE_NET_SETUP >#define INCLUDE_MUX >#define INCLUDE_BSD >#define INCLUDE_FEI >#define INCLUDE_LOOPBACK >#define INCLUDE_PING >#define INCLUDE_NETSHOW >usrNetwork.c, the module the parses the boot parameter and configure the >network interface work. According to the vxWorks Network programmer >guide all I need to do with this setup is assign the ip address with >ifAddrSet() and the mask with ifMaskSet(). Both calls result in an error >since the fei0 is not attached to the network stack. >I tried using usNetIfAttach() but this result in an unresolved symbol. >I tried usrNetIfConfig() but this call does not succeed. > >Does any one know how to attach the network device to the TCP/IP stack. >I do appreciate the help > >Nidal > - -- Henry --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado2 for NT machine: Where's cpp? Date: Tue, 21 Sep 1999 00:44:53 GMT From: Marjorie Krueger Organization: Vixel Message-ID: <37E6D4DB.6CF12650@vixel.com> I'm running Tornado2 on NT 4.0 - Anybody know what .exe is the C preprocessor? In the old Tornado 1.0 environment, we had makefiles that run cpp to create "build dependant files". Now I can't find this executable in Tornado2! Help? Thanks Marjorie Krueger Vixel Corporation --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hierarchical projects under Tornado II? Date: Tue, 21 Sep 1999 00:56:51 GMT From: Marjorie Krueger Organization: Vixel Message-ID: <37E6D7AA.6B55E1B8@vixel.com> References: <19990916015417.01471.00001746@ng-cg1.aol.com> <37E1D5CF.BCBB8719@scanmar.no> For those of us without a print version of the release notes, what section are you referring to? Ole Asbjorn Fadum wrote: > > Hi, > > In 'Tornado Release Notes 2.0' on page 21 has a description. Not very elegant > but it works. > > AJeddeloh wrote: > > > Has anyone found a way to use the Tornado II "project" facility for > > a non-trivial product? I have approx. 1000 source files in a hierarchical > > directory structure. How can I set up a build process under Tornado II > > without ending up with every object file landing in the same place? > > > > This is legacy code and does not lend itself to be broken into "Kernel" > > vs "application" projects. > > > > This is particularly nasty, since I have multiple 3rd-party vendor code > > sets which have to be integrated into the build, some of which duplicate > > file names! > > > > Please respond to Alan_Jeddeloh "AT" adc.com. This AOL account does not > > email from outside AOL. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Update Tornado Date: Mon, 20 Sep 1999 21:44:26 -0700 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <37E70D2A.227D@despam.pangea.ca> References: <37E68A54.3A0DA1BC@pegasus.cc.ucf.edu> Lei Sun wrote: > [...] > But after we finished installation and try to reboot the target, we > didn't get it. > it says "error loading file... errono=0xd0003". > Anybody can give me any hint? > To parse the errno, look up the module (13) in vxmodnum.h, [I don't have a copy at home so I can't tell you.] Then look in the specific header file and find error 3. That should give you some idea. Also it is not exactly clear from your description which module is printing the error. The boot image or the VxWorks image... - -het - -- "you don't really appreciate how smart a moron is until you try to design a robot..." -a robotics designer Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: GreenHills, VxWorks, and RogueWave Date: Mon, 20 Sep 1999 20:35:39 -0700 From: Michele Bottiaux Organization: SBC Internet Services Message-ID: <37E6FD0B.8B50AAE2@pandasys.com> Reply-To: michele@pandasys.com Hi, Is anyone mixing the GreenHills C++ compiler and RogueWave Tools.h++? We have a light transport layer built upon the ACE toolkit that uses a small amount of RogueWave which we are in the process of porting to VxWorks using GreenHills. But RogueWave doesn't officially support the configuration. Has anyone generated a suitable compiler.h and makefile for building Tools.h++? Any help would be greatly appreciated. Thank you, Michele Bottiaux michele@pandasys.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Update Tornado Date: Tue, 21 Sep 1999 05:01:17 GMT From: John or Jenn Organization: Who, us organized??? Message-ID: References: <37E68A54.3A0DA1BC@pegasus.cc.ucf.edu> In article <37E68A54.3A0DA1BC@pegasus.cc.ucf.edu> Lei Sun, ls00722@pegasus.cc.ucf.edu writes: >it says "error loading file... errono=0xd0003". > Anybody can give me any hint? Looks like the target can¹t find the vxworks image file. This could be caused by a number of things: 1) bad connection 2) target looking for file in wrong place (if you changed from T1 to T2, but didn¹t change anything on your target, it¹s probably still looking for VxWorks in the old place 3) you don¹t have a ftp server running 3a) permissions wrong on ftp server 3b) etc. 4) path problem (WIND_BASE not set right, etc.) etc. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About Global variables and main() functions Date: Mon, 20 Sep 1999 22:32:17 -0700 From: engineer_scotty Organization: aracnet.com Message-ID: <37E71861.CD924EF4@aracnet.com> References: <37E5CF37.F7582619@cselab.snu.ac.kr> <37E5D543.9B7119E5@aracnet.com> <37E609B6.B67879BF@cselab.snu.ac.kr> Lee Dong Yul wrote: > > Hello. Thanks for your reply. > > Do you inform me of EABI specification on VxWorks with MPC860 (PowerPC core)? Yup; you can download it from Motorola's website. (Somewhere...the website is complicated, and I can't remember where. Start with http://www.mot.com/SPS/PowerPC and look in the 821 section.) > What you said is that All global variables initialization is executed by > programmer explicitly? dynamically allocated memory management,too ? What I meant was---if you want a module with global variables to be restartable; you had better explicitly initialize them. Consider the following module: /* foobar module #1 */ #include "stdio.h" int foobar = 1; void runFoobar(void) { printf ("foobar is %d\n", foobar); foobar++; printf ("foobar is now %d\n", foobar): } /* end of foobar */ If you compile this and ld() it into vxWorks; the first time you type runFoobar from the shell, you will get >foobar is 1 >foobar is now 2 The second time you run it, you will get >foobar is 2 >foobar is now 3 And so on. If, on the other hand, you use /* foobar module #2 */ #include "stdio.h" int foobar; void initFoobar(void) { foobar = 1; } void runFoobar(void) { initFoobar(); printf ("foobar is %d\n", foobar); foobar++; printf ("foobar is now %d\n", foobar): } /* end of foobar */ Then each invokation of runFoobar() produces the same result. As far as memory management goes; it is initialized by your BSP. As long as you aren't working on the BSP, you don't have to do anything special to make malloc() work. > engineer_scotty wrote: > > > Lee Dong Yul wrote: > > > > > > Hello, We are developing some programs on VxWorks OS. > > > > To sum up all your questions: vxWorks ain't like Unix. > > > > vxWorks doesn't have the concept of separate "programs" with > > separate address spaces, and separate global variables. There is > > only one address space--and when you load new modules with ld(), > > the global variables in that module are added to that address space. > > > > > 1. Why can't we use main function with C language? > > > > > > When we programmed some applications with main function, We couldn't > > > download those applications. Undefined symbol (__eabi) error occurred. > > > > Lemme guess. PowerPC, right? :) > > > > Including main() causes the compiler/linker to automagically link in a > > bunch of Other Stuff. In Unix, including main() causes prologue code > > (which initializes the C standard library) to run, etc. In C++, this > > prologue code initializes static constructors. > > > > vxWorks on the powerPC, on the other hand, corresponds to a > > specification > > called EABI (Embedded Application Binary Interface); which calls for a > > function __eabi to be invoked when a program starts. This is, however, > > Not What You Want in vxWorks. > > > > The bottom line: don't call anything main(). - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: On scsiRdSecs() routine (again) Date: Tue, 21 Sep 1999 01:09:12 +0900 From: "³²¿µÁø" Organization: Postech News Server Message-ID: <37E65C28.86509D31@postech.ac.kr> Does anybody run the scsiRdSecs() routine in parallel in different tasks? Dramatic performance degradation was observed when using it concurrently at the different tasks. (H/W Env.: x86, one(1) ncr875 based HBA, five(5) scsi disks) Suppose that only a single task could use the routine. How can we expect "tagged queuing" benefit which maintains multiple outstanding SCSI command (since just one SCSI command (let's say READ command) is running at a time)? Thanks in advance. Young Jin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hierarchical projects under Tornado II? Date: Tue, 21 Sep 1999 07:58:54 +0200 From: Ole Asbjorn Fadum Organization: Telenor Online Public Access Message-ID: <37E71E9E.5CA6F709@scanmar.no> References: <19990916015417.01471.00001746@ng-cg1.aol.com> <37E1D5CF.BCBB8719@scanmar.no> <37E6D7AA.6B55E1B8@vixel.com> Hi, I'm refering to: Tornado Release Notes, 2.0 Edition 1, chapter Structuring Your Projects, sub chapter Hierarchial (sub-project) Model. The example is unfortunately too long to type in here. Marjorie Krueger wrote: > For those of us without a print version of the release notes, what section are you referring to? > > Ole Asbjorn Fadum wrote: > > > > Hi, > > > > In 'Tornado Release Notes 2.0' on page 21 has a description. Not very elegant > > but it works. > > > > AJeddeloh wrote: > > > > > Has anyone found a way to use the Tornado II "project" facility for > > > a non-trivial product? I have approx. 1000 source files in a hierarchical > > > directory structure. How can I set up a build process under Tornado II > > > without ending up with every object file landing in the same place? > > > > > > This is legacy code and does not lend itself to be broken into "Kernel" > > > vs "application" projects. > > > > > > This is particularly nasty, since I have multiple 3rd-party vendor code > > > sets which have to be integrated into the build, some of which duplicate > > > file names! > > > > > > Please respond to Alan_Jeddeloh "AT" adc.com. This AOL account does not > > > email from outside AOL. --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Sep 21 06:42:46 1999 From: "Christopher J. Slominski" Date: Tue Sep 21 06:42:49 PDT 1999 Subject: Graphics boards --=====================_355431253==_.ALT Content-Type: text/plain; charset="us-ascii" VxWorks users, I have a MVME2604 running VxWorks with Tornado II development on a NT machine. I need to get some display boards for my VME chassis (6U form factor) to drive LCD flat panel displays in an cockpit simulator. Does anyone have good suggestions on hardware to choose for display boards? I currently only have 2D display requirements, but It might be prudent to have 3D capabilties to be ready for the future. Thanks, Chris Slominski *----------------------------------------* | Christopher J. Slominski | | Computer Sciences Corporation (CSC) | | 3217 N. Armistead Ave. | | Hampton VA 23666 | | (757) 766-8258 | | | | c.j.slominski@larc.nasa.gov | | NASA Mail Stop 931 | *----------------------------------------* --=====================_355431253==_.ALT Content-Type: text/html; charset="us-ascii"
VxWorks users,
     I have a MVME2604 running VxWorks with Tornado II development on a NT machine.  I need to get some display boards for my VME chassis (6U form factor) to drive LCD flat panel displays in an cockpit simulator.  Does anyone have good suggestions on hardware to choose for display boards?  I currently only have 2D display requirements, but It might be prudent to have 3D capabilties to be ready for the future.

            Thanks,
               Chris Slominski

*----------------------------------------*
|     Christopher J. Slominski           |
| Computer Sciences Corporation (CSC)    |
|      3217 N. Armistead Ave.            |
|         Hampton VA 23666               |
|          (757) 766-8258                |
|                                        |
|      c.j.slominski@larc.nasa.gov       |
|         NASA Mail Stop 931             |
*----------------------------------------*
 --=====================_355431253==_.ALT-- From owner-vxwexplo-process Tue Sep 21 08:16:54 1999 From: David Laight Date: Tue Sep 21 08:16:58 PDT 1999 Subject: Re: NetBufLib > I am looking for some kind of APIs to concatenate (or merge) two > mBlk/clBlk/cluser construct into one. > Why an API? why not just write: mp1->mBlkHdr.mNext = mp2; which will add a header onto an existing message. If mp1 isn't freshly allocated, scan to its end. These actions are so simple ypu don't need an API - it just slows done the code execution and hides what is going on. SVR4 streams got itself into a right mess with 'pullupmsg' which has become more and more difficult to implement as the streams implementation has been made more efficient. vxworks mBlks (etc) are now where svr4 streams was about 12 years ago! 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 Tue Sep 21 08:31:50 1999 From: David Laight Date: Tue Sep 21 08:31:54 PDT 1999 Subject: netDrv performance (or lack of it!) A recent posting mentioned rcmd() and ftpXfer() - I hadn't noticed them before - and has made me modify a bit of code I'm testing. The code is writing a sequention file (in this case a screen dump) which might be a netDrv one. The file is just over 300Kb, so to avoid netDrv failing to allocate the space I changed the file open code to: i = strlen( sysBootParams.hostName ); if (!strncmp( sysBootParams.hostName, tgt_file, i ) && tgt_file[i] == ':' && !sysBootParams.passwd[0]) { sprintf( image_line, "cat >%s", tgt_file + i + 1 ); fd = rcmd( sysBootParams.hostName, 514, sysBootParams.usr, sysBootParams.usr, image_line, 0 ); } else fd = -1; if (fd == -1) fd = open( tgt_file, O_RDWR | O_CREAT, 0777 ); if (fd == -1) fd = creat( tgt_file, O_RDWR ); if (fd == -1) return; Obviously my test isn't for a netDrv file, but a file on the boot system! The above code works fine. What surprised me is that using netDrv the code takes about 4 seconds, using rcmd it finished so quickly I thought at first nothing had happened (say 0.1 seconds). So what is the vxWorks netDrv code doing that takes so long? The 300k I'm writing is done with about 1000 write() calls. The code to use ftp is similar - but you do need to do extra tidyup at the end. Also ftp might time out it the transfer rate is too slow. rcmd should not care. Anyone with a REAL problem might consider trying to persuade WRS to do something like the above if a file is opened O_APPEND. David PS: anyone know a bad 'deflate' format compress routine, ie one that uses limited data space and not much code but gives some compression for thing like screen dumps (that contain no images). ---------------------------------------------------------------- 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 Sep 22 04:00:17 1999 From: daemon@csg.lbl.gov Date: Wed Sep 22 04:00:21 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Sep 22 04:00:08 PDT 1999 Subject: Re: Hierarchical projects under Tornado II? Subject: Tornado2 for NT machine: Where's cpp? Subject: Re: VME A32 addressing/MVME162 Subject: Boot from network without bootrom code? Subject: Re: configuring network driver for vxWork image Subject: Re: Tornado2 for NT machine: Where's cpp? Subject: Clearcase/Tornado Anyone? Subject: Re: configuring network driver for vxWork image Subject: Re: configuring network driver for vxWork image Subject: gcc problems with static templates in vxWorks Subject: Re: Screenblanker & Windview Problems Subject: Re: About Global variables and main() functions Subject: Re: SLIP/ARP/PROXY in VxWorks Subject: Re: Update Tornado Subject: Re: Question about TCP socket and select() Subject: Anybody got AMD 85C30 SCC drivers in Synchronous mode? Subject: Re: Signal Handler for Each New Task Subject: SENS need to much Memory ! Subject: Metwork Interface Device Driver - copy_to_mbuf question. Subject: Re: vxWorks development system ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hierarchical projects under Tornado II? Date: Tue, 21 Sep 1999 11:12:59 GMT From: Ran@127.0.0.1 (Ran Shalgi) Organization: Only when absolutely necessary Message-ID: <37e766e9.783661826@news.inter.net.il> References: <19990916015417.01471.00001746@ng-cg1.aol.com> On Thu, 16 Sep 1999 12:57:09 +0200, "Borkhuis, Johan" wrote: > >> ---------- >> From: ajeddeloh@aol.com (AJeddeloh)[SMTP:ajeddeloh@aol.com] >> Posted At: donderdag 16 september 1999 7:54 >> Posted To: vxworks >> Conversation: Hierarchical projects under Tornado II? >> Subject: Hierarchical projects under Tornado II? >> >> Has anyone found a way to use the Tornado II "project" facility for >> a non-trivial product? I have approx. 1000 source files in a >> hierarchical >> directory structure. How can I set up a build process under Tornado >> II >> without ending up with every object file landing in the same place? >> >We had the same problem, and were not able to fix it. The way we did >this was to create a lot of projects in a workspace, each generating its >own .o file. Then we created one project that contained all the files. >Now you can compile only part of your project to test it. > >Another way is to create a project that only includes the .o files. Now >you have to generate each project seperatly and then create the complete >project. > >Both solutions not really workable, but then 1 project containing 1000 >files ..... > >The best solution for this kind of projects is to use Tornado only for >debugging (so you need one large project), and use make to generate your >code. > >Groeten > Johan I agree, the T2 project facility is completely unusable. It seems like it was never tested by real programmers doing real work. The editor is likewise utterly useless for anything but the most minor code changes. Use a normal editor, make/gcc for compiling, and T2 for debugging only. BTW, the T2 debugger _is_ more stable and reliable than the T1 version All in all, T2 is a major flop, IMHO. Ran Shalgi Expand Networks, Inc http://www.expand.com - --- Replace 127.0.0.1 with expand.com when replying by e-mail --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado2 for NT machine: Where's cpp? Date: Tue, 21 Sep 1999 12:55:20 +0200 From: "Borkhuis, Johan" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <37E6D4DB.6CF12650@vixel.com> > ---------- > From: Marjorie Krueger[SMTP:mkrueger@vixel.com] > Posted At: dinsdag 21 september 1999 2:44 > Posted To: vxworks > Conversation: Tornado2 for NT machine: Where's cpp? > Subject: Tornado2 for NT machine: Where's cpp? > > > I'm running Tornado2 on NT 4.0 - Anybody know what .exe is the C > preprocessor? In the old Tornado > 1.0 environment, we had makefiles that run cpp to create "build > dependant files". Now I can't find > this executable in Tornado2! Help? > In my configuration (Win95) it is placed in tornado\hostx86-win32\lib\gcc-lib\powerpc-wrs-vxworks\cygnus-2.7.2-96012 6 Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME A32 addressing/MVME162 Date: Thu, 16 Sep 1999 13:22:26 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <37E13562.1295F7C6@aps.anl.gov> References: <37E0233B.E69F041@navair.navy.mil> Kevin Havig wrote: > > I am using VxWorks 5.4 on a Motorola MVME162 processor. I am having a > problem accessing a board in the VME 32 bit address space; address > 0x10000000 to be exact. The processor has no problem accessing any > boards in the VME 16 or 32 bit address space. Furthermore if I boot the > processor to the built in Motorola debugger, it accesses this location > just fine. Is there some trick to getting VxWorks to access this > board. Thanks in advance for all of your help. It sounds like you don't have an entry in the sysPhysMemDesc[] table for this memory area (common problem). This table is used to initialize the MMU page tables, and is defined towards the beginning of the /target/config/mv162/sysLib.c file. Have a look at chapter 7 of the vxWorks Programmer's Guide for a description. You'll need to modify or add an entry to this table and rebuild the vxWorks image to enable access. - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Boot from network without bootrom code? Date: Tue, 21 Sep 1999 16:14:45 +0200 From: Carlos J. =?iso-8859-1?Q?Garc=EDa?= Orellana Organization: Universidad de Extremadura Message-ID: <37E792D5.A0BDE11A@nernet.unex.es> Hello, We are been using a floppy disk to boot from network. In that floppy, we have the compressed bootrom code, with a size around 128kbytes. However, now we want to boot using the network card flash RAM (with size of 64Kbytes max.). How could we do it?. We are bootting Linux with the 'netboot' package. Could we use that package with vxWorks? Thanks. Carlos J. García Orellana Universidad de Extremadura - SPAIN e-mail: carlos@nernet.unex.es --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Tue, 21 Sep 1999 07:26:46 -0700 From: Nidal Khalil Organization: SBC Internet Services Message-ID: <37E795A4.359AF559@pacbell.net> References: <37E4E4C5.1F32DC4@pacbell.net> Reply-To: nidal@pacbell.net Hello All, I would like to thank all of you for your help. I put in a call to wind river tech support for this problem, I will let you know what work for me. Thanks again, this is a faster response that wind river technical support. To be honest I am disappointed by tornadoII, I think that wind river did not do enough testing before they released it Nidal Henry Gessau wrote: > Does it help if you > > #define STANDALONE_NET > > ? > > On Sun, 19 Sep 1999 06:27:35 -0700, Nidal Khalil wrote: > > >Hello, > >I am trying to build a vxWorks image with network driver in it. > >I am using tornadoII and vxWorks version 5.4. > >The network card that I am using is a standard intel ethernet pro 100 > >which is supported by vxWorks. If I boot vxWorks over the net the driver > >is configured properly but it I boot from a floppy or flash the network > >device is not attached to the tcp/ip stack. my config.h has all the > >declaration for including the network interface. > >#define INCLUDE_NETWORK > >#define INCLUDE_NET_SETUP > >#define INCLUDE_MUX > >#define INCLUDE_BSD > >#define INCLUDE_FEI > >#define INCLUDE_LOOPBACK > >#define INCLUDE_PING > >#define INCLUDE_NETSHOW > >usrNetwork.c, the module the parses the boot parameter and configure the > >network interface work. According to the vxWorks Network programmer > >guide all I need to do with this setup is assign the ip address with > >ifAddrSet() and the mask with ifMaskSet(). Both calls result in an error > >since the fei0 is not attached to the network stack. > >I tried using usNetIfAttach() but this result in an unresolved symbol. > >I tried usrNetIfConfig() but this call does not succeed. > > > >Does any one know how to attach the network device to the TCP/IP stack. > >I do appreciate the help > > > >Nidal > > > > -- > Henry --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 for NT machine: Where's cpp? Date: Tue, 21 Sep 1999 09:10:58 -0700 From: Alex K. Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <1ca4f2e8.3305c747@usw-ex0102-014.remarq.com> References: <37E6D4DB.6CF12650@vixel.com> We could proceed to work with our makefiles after we had copied fixdotd.exe to '\Tornado\host\x86-win32\bin' in Tornado II installation. Hope it helps, Alex In article <37E6D4DB.6CF12650@vixel.com>, Marjorie Krueger wrote: > I'm running Tornado2 on NT 4.0 - Anybody know what > .exe is the C preprocessor? In the old Tornado > 1.0 environment, we had makefiles that run cpp to > create "build dependant files". Now I can't find > this executable in Tornado2! Help? > Thanks > Marjorie Krueger > Vixel Corporation * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Clearcase/Tornado Anyone? Date: Tue, 21 Sep 1999 11:19:04 -0500 From: John Beauford Organization: Motorola Message-ID: <37E7AFF8.158858C7@email.mot.com> Reply-To: qjb005@email.mot.com Does anyone install Tornado and associated tools in a clearcase VOB? Do you like the setup? Are there caveats? We have 3 groups using Tornado for different targets and they all have source in different Clearcase VOBs. Anyone have a recommendation on this? Install the tornado outside the vobs? How about linking a final static object when the source is in a VOB and tornado is not? - -- John Beauford --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Tue, 21 Sep 1999 08:21:30 +0100 From: "Wade Oram" Organization: Marconi Instruments Ltd., Stevenage, Herts., UK Message-ID: <7s7blf$d6n@ipgw.marconi-instruments.co.uk> References: <37E4E4C5.1F32DC4@pacbell.net> Nidal Khalil wrote in message news:37E4E4C5.1F32DC4@pacbell.net... > I am trying to build a vxWorks image with network driver in it. > I am using tornadoII and vxWorks version 5.4. > The network card that I am using is a standard intel ethernet pro 100 > which is supported by vxWorks. If I boot vxWorks over the net the driver > is configured properly but it I boot from a floppy or flash the network > device is not attached to the tcp/ip stack. Hello, I have a very similar problem trying to configure a VxWorks build for a 486 target with a PCMCIA Etherlink III network card. I have build a boot rom which will successfully boot a vxworks image over the network attaching the ELT driver (on pcmcia=1) in the process. However, if I then copy the VxWorks image onto a floppy disk and then boot the same Vxworks image from the disk using the same boot rom (by interrupting the auto-boot and setting the boot device to 'fd=0' and the image file to /fd0/VxWorks') then although the VxWorks image loads, the network driver is not installed. If I also change 'o=pcmcia=1,0' when I interrupt the autoboot, then the VxWorks image is booted correctly and the network drivers are properly installed. However, since my goal is to eventually boot a Rom based VxWorks image directly from a small boot rom (BIOS) this does not help me since I will not be running the VxWorks boot rom image. - -- Wade Oram --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Tue, 21 Sep 1999 08:30:31 +0100 From: "Wade Oram" Organization: Marconi Instruments Ltd., Stevenage, Herts., UK Message-ID: <7s7c6d$dbf@ipgw.marconi-instruments.co.uk> References: <37E4E4C5.1F32DC4@pacbell.net> Nidal Khalil wrote in message news:37E4E4C5.1F32DC4@pacbell.net... > I am trying to build a vxWorks image with network driver in it. > I am using tornadoII and vxWorks version 5.4. > The network card that I am using is a standard intel ethernet pro 100 > which is supported by vxWorks. If I boot vxWorks over the net the driver > is configured properly but it I boot from a floppy or flash the network > device is not attached to the tcp/ip stack. Hello, I have a very similar problem trying to configure a VxWorks build for a 486 target with a PCMCIA Etherlink III network card. I have built a boot rom which will successfully boot a vxworks image over the network attaching the ELT BSD 4.4 driver (on pcmcia=1) in the process. (The ELT END driver does not support the PCMCIA Etherlink II). However, if I then copy the same VxWorks image onto a floppy disk and boot it from the disk using the same VxWorks boot rom (by interrupting the auto-boot and setting the boot device to 'fd=0' and the image file to '/fd0/VxWorks') then although the VxWorks image loads, the network driver is not installed. If I also add 'o=pcmcia=1,0' when I interrupt the autoboot, then the VxWorks image is booted correctly and the network drivers are properly installed. However, since my goal is to eventually boot a Rom based VxWorks image directly from a small boot rom (BIOS) this does not help me since I will not be running the VxWorks boot rom image. Adding 'o=pcmcia=1,0' to the default boot line in the VxWorks build itself rather than the boot rom does not help. - -- Wade Oram --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc Subject: gcc problems with static templates in vxWorks Date: Tue, 21 Sep 1999 18:58:16 GMT From: tsanocki@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7s8kg0$pq5$1@nnrp1.deja.com> Reply-To: sanocki@srtc.com Hello, We're currently having problems using vxWorks Tornado 1.0.1's gcc compiler (version cygnus-2.7.2-960126), running on a Win32 machine and cross compiling to a PowerPC (vxWorks mcp750 target). We're porting code that uses static templates, and gcc doesn't support them. I have a few questions: 1) Am I doing something wrong? Do I need to turn on a compiler switch or put in a pragma or something? 2) If not, should I use a later version of gcc? I've had little luck finding out whether gcc 2.95.1 fixes this problem. I would like to be reasonably sure that using a later version of gcc will fix the problem before I put in the effort to build a cross compiler. Is there a document or FAQ available that would give me more information on this? Might there be gcc binaries available someplace? 3) Is there some other solution that I'm missing? We're investigating using a different compiler, but I'd like to see if I can get gcc to work so I can get at least a little bit of work done in the meantime. Please point me to any FAQs or documents if I've missed something obvious. Any help would be much appreciated! Thanks - - Tom Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Screenblanker & Windview Problems Date: Tue, 21 Sep 1999 20:55:39 GMT From: Ted Dennison Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7s8rc5$v0i$1@nnrp1.deja.com> References: <199909210740.JAA00184@rzsrv2.rz.tu-bs.de> In article <199909210740.JAA00184@rzsrv2.rz.tu-bs.de>, Jan Uerpmann wrote: > Hi > > I am using VxWorks 5.3.1 and Tornado 1.0.1 on a pc486-architecture and > have encountered some problems: > > - After a while the programs run about 5 times slower than they > used to. I recognized that if I press a key on the target the > program speeds up again. I think it has something to do with the > screenblanker as the target is slow when the blanker is active. I have the exact same setup, and have no clue what "screenblanker" you are talking about. My target screen *never* blanks out. Perhaps your PC BIOS is doing this to you. If so, I'd suggest going into the BIOS and disabling it. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About Global variables and main() functions Date: 21 Sep 99 22:15:36 GMT From: georg@sgl.crestech.ca (Georg Feil) Organization: Centre for Research in Earth and Space Technology Message-ID: References: <37E5CF37.F7582619@cselab.snu.ac.kr> Lee Dong Yul writes: >1. Why can't we use main function with C language? This one's been covered pretty well... just use another name. >2. In tornado shell, we can't execute some programs many times. All >Those programs have global variables. So, After executing those >programs, Global variables contain some values that are different from >program startup values. >How do we initialize all global variables on program startup? >Why don't vxworks initialize global variables on program startup. Think of VxWorks as one big continuously running program with multiple threads (tasks). Global variables are not reinitialized just because a task gets started. To get around this I've written initialization modules for certain pieces of code that I want to be able to restart like Unix programs, e.g. code ported from unix like editors and such. The initialization module ensures that initialized variables and BSS variables are set right, then branches to the (renamed) main function when done. I've included an example copy below, although I use a separate slightly customized init module for each such program. >When Initialization occurs? Object code downloading? Right. >3. On program exit, is dynamically allocated memory automatically freed? No. Good programmers clean up after themselves anyway. So here's that init module. Link this with your PROGRAM. You call PROGRAM_start() which in turn calls PROGRAM_main(). There's also another little function that goes before this one that I haven't included (called "PROGRAM()") which repackages arguments so that old unix code expecting argv/argc stuff will work. Good luck, Georg - ----------------------------- crt0.c ------------------------------ #include #include #include #include #include /* * PROGRAM startup module. * * This version for VxWorks. The main purpose is to properly initialize the * initialized data space and bss between different runs of PROGRAM. The first * time it executes, this routine saves away the entire initialized data space * in malloc'd memory. On subsequent runs it restores the initialized data. * It zeroes the bss area as well, but *be warned* that this should not * be relied upon. Some BSS variables (those labeled as "common") seem to * be placed outside the normal BSS block by the VxWorks * loader, therefore *never* depend on uninitialized variables to be * automatically set to 0. Explicitly initialize them to zero instead. * * 97/09/17 Georg Feil Changed to use moduleLib for segment addresses, * added clearing of BSS. * 97/06/27 Georg Feil Converted from assembly to C (for PPC port). * 91/04/23 Georg Feil Cleaned up, added comments. * 90/07/26 Georg Feil Fixed stack creep problem at malloc() call. * 90/06/27 Georg Feil Initial revision, from s2 dev system. */ /* flag shows whether initialized data area has been copied (!=0 means yes) Value is start of malloc data area. */ static char* init=0; void PROGRAM_start(int arg1, int arg2, int arg3, int arg4, int arg5, int arg6, int arg7, int arg8, int arg9, int arg10) { int PROGRAM_main(); /* define main() this way to soothe the compiler */ char* d0; MODULE_ID mod_id; MODULE_INFO mod_info; /* get segment info from moduleLib (properly gets last instance loaded) */ mod_id=moduleFindByName("PROGRAM"); if (mod_id==NULL) { printf("PROGRAM_start(): Error, could not find module named \"PROGRAM\"\n"); return; } if (moduleInfoGet(mod_id, &mod_info) == ERROR) { printf("PROGRAM_start(): Error, could not get module info for ID 0x%lx\n", (unsigned long)mod_id); return; } /* clear uninitialized data area (bss) *** see warning in comment at top */ bzero(mod_info.segInfo.bssAddr, mod_info.segInfo.bssSize); /* Copy initialized data area to save (1st time) or restore it (other times). (gets copied after end of uninitialized data (bss)) Uses malloc() to obtain storage space -- Warning this is never freed. */ if (init==0) { /* malloc() enough memory to hold initialized data region */ d0=malloc(mod_info.segInfo.dataSize); if (d0==0) { /* malloc() error (returned NULL), exit immediately */ printf("PROGRAM_start(): Error, could not malloc\n"); return; } init=d0; /* save malloc() address in init */ /* save initialized data */ bcopy(mod_info.segInfo.dataAddr, init, mod_info.segInfo.dataSize); } else { /* restore initialized data */ bcopy(init, mod_info.segInfo.dataAddr, mod_info.segInfo.dataSize); } /* Jump to PROGRAM main() */ PROGRAM_main(arg1,arg2,arg3,arg4,arg5,arg6,arg7,arg8,arg9,arg10); } - -- Georg Feil | http://www.sgl.crestech.ca/ Space Geodynamics Laboratory | Email: georg@sgl.crestech.ca CRESTech | Phone: (416) 665-5458 4850 Keele St./North York/Ont/Canada/M3J 3K1 | Fax: (416) 665-1815 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SLIP/ARP/PROXY in VxWorks Date: Wed, 22 Sep 1999 10:33:29 +0900 From: "Chuck Burkett" Organization: Motorola CIG Message-ID: <7s9bos$sfv10@nntp.cig.mot.com> References: <7s7iae$1c43@overload.lbl.gov> We were able to get the proxyArp source code from WindRiver so that we could change it to allow proxyArp with a non-shared memory interface. The tricky part is making sure your routing table gets set to send all non-registered IPs through the SLIP interface. For proxyArp to work the devices on the ethernet must be in the same subnet as the SLIP IP. You could use the routing table to setup the SLIP target has a gateway to the other devices. This allows different network IPs to be used. Setting up the SLIP target as a gateway requires routing change on the other end of the SLIP as well as the non-SLIP targets. Ilia Bosis wrote in message <7s7iae$1c43@overload.lbl.gov>... >Hi Rasty. >The Tornado Proxy ARP server works over the shared memory backplane ONLY! >It's _not intended_ for anything else and will refuse to work over the >interface, other than "sm" :( . >You can try to fool the Proxy ARP server by renaming Your SLIP interface as >a shared-memory one, >but do it on Yours own risk. >Regards > Ilia. > > > >> -----Original Message----- >> From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] >> Sent: Thursday, September 16, 1999 5:37 PM >> To: vxworks_users@csg.lbl.gov >> Subject: SLIP/ARP/PROXY in VxWorks >> >> Hi, >> Did anyone succeed to have Proxy ARP server with the SLIP? >> I have five targets connected together though Ethernet, one >> of them has also >> SLIP connection to the host computer. I'd like to establish >> connection >> between host and any target. I initially planned to run Proxy >> ARP server on >> the target that has SLIP connection to the host, but >> proxyNetCreate always >> returns -1, no examples in manuals, tried a dozen of combinations ;-(. >> >> Any suggestions? >> >> Rasty >> --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Update Tornado Date: Tue, 21 Sep 1999 22:11:30 -0400 From: Lei Sun Organization: University of Central Florida Message-ID: <37E83AD2.4F60092F@pegasus.cc.ucf.edu> References: <37E68A54.3A0DA1BC@pegasus.cc.ucf.edu> John or Jenn wrote: > In article <37E68A54.3A0DA1BC@pegasus.cc.ucf.edu> Lei Sun, > ls00722@pegasus.cc.ucf.edu writes: > >it says "error loading file... errono=0xd0003". > > Anybody can give me any hint? > > Looks like the target can¹t find the vxworks image file. This could > be caused by a number of things: > > 1) bad connection > 2) target looking for file in wrong place (if you changed from T1 > to T2, but didn¹t change anything on your target, it¹s probably > still looking for VxWorks in the old place > 3) you don¹t have a ftp server running > 3a) permissions wrong on ftp server > 3b) etc. > 4) path problem (WIND_BASE not set right, etc.) > etc. somebody plug 2 cable into the hub and become a loop, so the target never access to the network a stupit mistake. thanks for your help! lei sun --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about TCP socket and select() Date: Wed, 22 Sep 1999 09:16:54 +0200 From: Laurent Peaudeau Organization: Nortel Networks Message-ID: <37E88266.D8E58DB7@nortelnetworks.com> References: <7s0i0e$a9i$1@nnrp1.deja.com> Hi, This problem seems strangely with one I met with multiple lpr requests under Win NT. The problem I met came from the fact that Windows gives a limited number of ports for TCP connections (about ten in our case) and waits a small delay before re-allocating it for a new connection (that's why I think you almost avoided the problem when setting a delay to 200ms). I remember that we solved the problem by editing the NT registry and changing one value about the delay before closing the connection. TCP/IP settings are in HKey_Local_Machine\System\CurrentControlSet\Services\tcp\Parameters\ You can also have a look at windows FAQ. Hope this helps, Laurent - -- - ---------------------------------------------------------------------- Laurent PEAUDEAU Nortel Networks Marne La Vallee Tel : 01-64-76-77-97 ESN 578-7797 Internal E-Mail : peaudeau@europem01.nt.com External E-Mail : peaudeau@nortelnetworks.com --------------------------- Newsgroups: comp.os.vxworks Subject: Anybody got AMD 85C30 SCC drivers in Synchronous mode? Date: Wed, 22 Sep 1999 00:16:05 +0100 From: Michael Michel Organization: None Message-ID: <37E811B3.5F0EA0B7@west.net> Reply-To: rumba@west.net 1) Has anyone ever used this SCC to operate in Synchronous mode to operate at up to 1Mbps? 2) For this mode this chip can be programmed to operate using different protocols (e.g., SDLC, HDLC, bit versus byte oriented). Can anybody tell me which protocol is the the most common used (to the point of being "default"?). The application I have in mind involves downloading data from a laptop or desktop computer via a serial cable to a target board running VxWorks (Tornado 2) on a PowerPC 740. This SCC is a fairly complicated (but powerful) device and any sample drivers or info where such information can be found would be greatly appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Signal Handler for Each New Task Date: Tue, 21 Sep 1999 20:46:31 -0700 From: "Steve Doiel" Message-ID: <37e85064.0@news.pacifier.com> References: <37E84096.321E4111@uswest.net> Reply-To: "Steve Doiel" I faced exactly the same problem a couple of years ago and never had any luck with taskCreateHookAdd. The way I got around this difficulty was to create a function I call "TaskBaseFunction" that is the function I use for all tasks. I pass the address of the function I want to execute for the task as the first parameter to TaskBaseFunction. Well rather than dribble on... here's the relevent code: /*========================================================================== ======== // Define a structure that is dynamically allocated for each task created, and // accessable to that task. ===================================================*/ typedef struct { anExceptionOSU exceptionOSU; /* Task based exception handling information */ int phaseOSU; /* Used for deducing the stage of processing */ } aTaskBufferOSU; /*========================================================================== ====== // FUNCTION: SignalHandler // Description: // If a task tries to access hardware that isn't there, or does some other // similarly fatal operation, this handler gets control. All it does is // output some information to the message module and then return to the // base task function which will exit. ============================================================================ ====*/ static void SignalHandler( int sigNum, int code, struct sigcontext *pSigContext ) { switch( sigNum ) { case SIGSEGV : PutMessageMSG( 0, 0, "SIGSEGV(%d) - bus error exception", code ); break; case SIGBUS : PutMessageMSG( 0, 0, "SIGBUS(%d) - BUS_ADDERR address error exception", code ); break; case SIGILL : PutMessageMSG( 0, 0, "SIGILL(%d) - exception", code ); break; case SIGFPE : PutMessageMSG( 0, 0, "SIGFPE(%d) - exception", code ); break; case SIGTRAP : PutMessageMSG( 0, 0, "SIGTRAP(%d) - exception", code ); break; case SIGEMT : PutMessageMSG( 0, 0, "SIGEMT(%d) - exception", code ); break; case SIGFMT : PutMessageMSG( 0, 0, "SIGFMT(%d) - exception", code ); break; default : PutMessageMSG( 0, 0, "Unhandled signal exception: %d(%d)", sigNum, code ); break; } taskDelay( MilliSecondsToTicksOSU( 500 ) ); longjmp( taskBufferOSU->exceptionOSU.baseEnvOSU , 0 ); } /*========================================================================== ====== // FUNCTION: TaskBaseFunction // Description: // Each task created by CreateTaskOSU is a creation of this function. The // users function is invoked as a function call after task variables have // been added. // // (1) Adds a task variable and allocates a task buffer which that task // variable points to. // (2) Initialize the return status for the exception as well as the // environment for unhandled exceptions to continue from. // (3) Regardless of how the task finished, return a return status from // the task function. // Inputs: // taskFunction // - The function requested to be created // arg1..arg9 // - Parameters to be passed to the task being created // Outputs: // Returns: // Return value from the created task (if it exits). // Caveats: // Does not remove the task variable or recover the memory on task exit. //========================================================================== ====*/ STATUS TaskBaseFunction( aTaskFunctionPtr taskFunction, int arg1, int arg2, int arg3, int arg4, int arg5, int arg6, int arg7, int arg8, int arg9 ) { STATUS returnStatus; returnStatus = ERROR; /* (1) */ taskVarAdd( taskIdSelf(), (int *)&taskBufferOSU ); taskBufferOSU = calloc( 1, sizeof( aTaskBufferOSU ) ); /* (2) */ taskBufferOSU->exceptionOSU.exceptionStatusOSU = 0; /* Comment these out to let VxWorks suspend any task that would cause a signal signal( SIGSEGV, SignalHandler ); signal( SIGBUS, SignalHandler ); signal( SIGILL, SignalHandler ); signal( SIGFPE, SignalHandler ); signal( SIGTRAP, SignalHandler ); signal( SIGEMT, SignalHandler ); signal( SIGFMT, SignalHandler ); */ taskBufferOSU->phaseOSU = -1; if( setjmp( taskBufferOSU->exceptionOSU.baseEnvOSU ) == 0 ) { if( setjmp( taskBufferOSU->exceptionOSU.environmentOSU ) == 0 ) { returnStatus = (*taskFunction)( arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9 ); } else { PutMessageMSG( 0, 0, "TASK %s UNHANDLED EXCEPTION PHASE(%d)... TERMINATING", taskName( taskIdSelf() ), taskBufferOSU->phaseOSU ); PutMessageMSG( 0, 0, "EXCEPTION REASON: %s", ExceptionStringEXL( ExceptionReasonEXC() ) ); taskDelay( MilliSecondsToTicksOSU( 1000 ) ); returnStatus = ERROR; /* unhandled exception */ } } else { PutMessageMSG( 0, 0, "FATAL HARD EXCEPTION IN %s PHASE(%d)", taskName( taskIdSelf() ), taskBufferOSU->phaseOSU ); taskDelay( MilliSecondsToTicksOSU( 1000 ) ); returnStatus = ERROR; /* unhandled exception */ } /* (3) */ return returnStatus; } /*========================================================================== ====== // FUNCTION: CreateTaskOSU // Description: // Creates a task an allocates and associates eith that task a task buffer. // // (1) Creates the task base function with the task function which was given // as an arguement as well as all aditional creation arguments. // // Inputs: // entryPt // - Function to be run for the task // others // - See the description of taskSpawn in the VxWorks programming manual. // Outputs: // Returns: // See the description of taskSpawn in the VxWorks programming manual. // Caveats: // Internally one of the arguments is used for creating the task buffer so // only 9 arguments are available to this call where 10 are available to // the native taskSpawn function. // The procedure actually installed is a function that calls entryPt, so // the debugger will break into foreign code, and requires stepping into // user code. int CreateTaskOSU( char *name, aPriorityClassOSU priorityClass, aPriorityLevelOSU priorityLevel, /* 0(high) to 50(low) */ aTaskOptionOSU options, int stackSize, FUNCPTR entryPt, int arg1, int arg2, int arg3, int arg4, int arg5, int arg6, int arg7, int arg8, int arg9 ) { int status; /* (1) */ status = taskSpawn( name, VxWorksPriority( priorityClass, priorityLevel ), options, stackSize, (FUNCPTR)TaskBaseFunction, (int)entryPt, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9 ); return status; } I hope this helps, SteveD Glenn Hofmeister wrote in message news:37E84096.321E4111@uswest.net... > Greetings, > > I am using vxSim 5.3.1 and what I am attempting to do is set up a signal > handler to process a segmentation violation in a task. However, what I > am > really trying to do is for each new task that starts, this signal > handler > is to be set up. So I have been trying to use taskCreateHookAdd() to > set > up a function that is called when a new task starts. However, when the > function gets called, the task is not the new task, but it does have a > pointer to the new task's WIND_TCB. I do not know what fields I could > update in the new task's WIND_TCB to set up a signal handler for the new > task. Has anyone ever tried something like this? > > My signal handler is: > > LOCAL jmp_buf env; > void sigHandler > ( > int sig, > int code, > SIGCONTEXT *sigContext > ) > { > switch (sig) > { > case SIGSEGV: > printErr ("\nSignal SIGSEGV received\n"); > break; > } > longjmp (env, sig); > } > > My added Hook routine: > void myCreateHook (pTcb) > WIND_TCB *pTcb; > { > > } > > My code to set up the hook: > taskCreateHookAdd((FUNCPTR) myCreateHook); > > > Thanks, > > Glenn > glennh@uswest.net --------------------------- Newsgroups: comp.os.vxworks Subject: SENS need to much Memory ! Date: Wed, 22 Sep 1999 10:02:12 +0200 From: Oliver Faessler Organization: tesion Communikationsnetze Suedwest GmBH & Co.KG, Stuttgart, Germany Message-ID: <37E88D04.509F29A6@sel.de> This is a multi-part message in MIME format. - --------------9E7404DB190F6B58B8A76596 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, i have a Problem with Tornado 1.0.1 and SENS. The Target is an MC68360 CPU with 512KB FLASH and 512 KB SRAM. 1: If i install Tornado1.0.1 without SENS if i make the vxWorks ROM image (vxWorks.res.ron_nosym.hex) the availiable memory on target is 280KB. 2: If i install Tornado 1.0.1 with SENS (1.0 or 1.1) I make the image file and copy the image to the Flash... The free memory on the target is now 50KB. SENS need 220kB of the Memory !!!!!! Sens allocate some buffers and so. What can i do, to reduce the Memory consumption ????? bye Oliver Faessler - --------------9E7404DB190F6B58B8A76596 Content-Type: text/x-vcard; charset=us-ascii; name="O.Faessler.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Oliver Faessler Content-Disposition: attachment; filename="O.Faessler.vcf" begin:vcard n:Faessler;Oliver tel;fax:+49 7231 15 3284 tel;work:+49 7231 15 3313 x-mozilla-html:FALSE url:www.sel.de org:SEL - Verteidigungssysteme GmbH;DP/ETS adr:;;Ostendstrasse 3;Pforzheim;;75175;Germany version:2.1 email;internet:O.Faessler@sel.de x-mozilla-cpt:;-28344 fn:Oliver Faessler end:vcard - --------------9E7404DB190F6B58B8A76596-- --------------------------- Newsgroups: comp.os.vxworks Subject: Metwork Interface Device Driver - copy_to_mbuf question. Date: Wed, 22 Sep 1999 11:14:08 +0200 From: Alex Shteinfeld Message-ID: <37E89DE0.66301F32@ecitele.com> Hello VxWorkers! I write a Network Interface Device Driver on MPC860. My question is the following: Do I need to "lock" by splnet() function the part of code where I copy the received buffer to the mbuf chain (copy_to_mbuf)? Thanks in advance Alex Shteinfeld shtein@ecitele.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks development system Date: Mon, 20 Sep 1999 18:48:54 +0100 From: David & Bella Aldrich Message-ID: <1999092018485483839@zetnet.co.uk> References: <7s068p$hrf$1@newso.ctimail.com> The message <7s068p$hrf$1@newso.ctimail.com> from "Simon Chan" contains these words: > I'm new in RTOS environment and want to know that where can I get the > development software of vxWorks for the Pentium platform? Please help. Contact the manufacturer - Wind River Systems. I think the website is www.wrs.com Regards David Aldrich --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Wed Sep 22 04:42:02 1999 From: "Hamon, Kent" Date: Wed Sep 22 04:42:05 PDT 1999 Subject: RE: Graphics boards (Recommendation) Hello, We are using MVME2604 with the Motorola PMCSpan board(s) and PMC video output from Peritek (http://www.peritek.com). They manufacture single output and a quad output PMC and provide an OpenGL library. Regards, Kent > -----Original Message----- > From: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] > Sent: Tuesday, September 21, 1999 9:43 AM > To: vxworks_users@csg.lbl.gov > Subject: Graphics boards > > > VxWorks users, > I have a MVME2604 running VxWorks with Tornado II development on a NT > machine. I need to get some display boards for my VME chassis (6U form > factor) > to drive LCD flat panel displays in an cockpit simulator. Does anyone > have > good suggestions on hardware to choose for display boards? I currently > only > have 2D display requirements, but It might be prudent to have 3D > capabilties to > be ready for the future. > > Thanks, > Chris Slominski > > *----------------------------------------* > | Christopher J. Slominski | > | Computer Sciences Corporation (CSC) | > | 3217 N. Armistead Ave. | > | Hampton VA 23666 | > | (757) 766-8258 | > | | > | c.j.slominski@larc.nasa.gov | > | NASA Mail Stop 931 | > *----------------------------------------* > > From owner-vxwexplo-process Wed Sep 22 07:07:06 1999 From: Roger Holmquist Date: Wed Sep 22 07:07:10 PDT 1999 Subject: Re: VxWorks, VxMP >Is there anything to be especially aware of with >using VxMP between two PowerPC processors? > >I noticed that one of the hardware requirements is: > >"Indivisible hardware read-write-modify mechanism across >the shared memory bus" > >and wondered if this causes any problems with Power PC processors. > >We use the MVME2700 board. > We are using 68k-boards from Motorola in our system and we have 2 problems with VxMP , the second problem popped up when we "upgraded" to Tornado II: 1) We haven't managed to mix Motorola 68K and PPC-cards in our MP-environment and make them talk to each other. (is this bug fixed in TornadoII ?) 2) Tornado II introduces a bug in the virtual VME-backplane-net: When connecting WindShell to any of the subnet-cards through the launch-facility, the ethernet-equipped card (the "router") will be attached instead, without any warning message. Opinions/Suggestions welcome! Greetings / Roger - # -----------------------------------------------------------------# # Roger Holmquist, FOA, National Defence Research Establishment # # sect 750, Olaus Magnus vag 42, Box 1165, 581 11 LINKOPING ------# # --------------------------- SWEDEN ------------------------------# # Phone: +46 13-378382 --------------------------------------------# # Fax: +46 13-378009 ----------------------------------------------# # Home: +46-142-20542 Mobile +46-706-250123 ----------------------# # Email: roghol@lin.foa.se ----------------------------------------# # Home address: Hovslagargrand 5, 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 Wed Sep 22 09:06:59 1999 From: johill@lanl.gov (Jeff Hill) Date: Wed Sep 22 09:07:03 PDT 1999 Subject: troubles with net interface query ioctl() and SENS Hello all, Does anyone know why the interface query ioctl() does not appear to be working correctly in Tornado II (with the SENS stack)? Our problem is that entries of type AF_INET appear to have invalid flags, or the flags cant be fetched at all. For instance, on our system the flags indicate that the "feio" interface is a point to point link when we know full well that it is a LAN interface. Here is the output from the function attached below which uses the net interface query ioctl() API (derived from BSD). discoverInterfaces: 4 net intf(s) found discoverInterfaces: net intf fei0: pt to pt addr fetch fail discoverInterfaces: net intf flags fetch for failed discoverInterfaces: net intf flags fetch for failed discoverInterfaces: net intf flags fetch for failed Note that the last 3 network interfaces returned do not have a valid name. There were only two interfaces in the past {fei0, loopback}. This leads one to guess that the size of the "struct ifreq" returned has changed in SENS, but perhaps WRS has not updated the header files in Tornado II which define "struct ifreq" to reflect this? A quick search of Tornado II concludes that there is only one file supplied that defines this structure. Tornado2.0/target/h/net/if.h I am running my tests on the pcPentium BSP. Thanks in advance. Jeff void discoverInterfaces (ELLLIST *pList, int socket, unsigned short port, struct in_addr matchAddr) { struct sockaddr_in *pInetAddr; caAddrNode *pNode; int status; struct ifconf ifconf; struct ifreq *pIfreqList; struct ifreq *pifreq; unsigned long nelem; /* * use pool so that we avoid using to much stack space * under vxWorks * * nelem is set to the maximum interfaces * on one machine here */ nelem = 100; pIfreqList = (struct ifreq *)calloc(nelem, sizeof(*pifreq)); if(!pIfreqList){ return; } ifconf.ifc_len = nelem*sizeof(*pifreq); ifconf.ifc_req = pIfreqList; status = socket_ioctl(socket, SIOCGIFCONF, &ifconf); if (status < 0 || ifconf.ifc_len == 0) { free(pIfreqList); return; } nelem = ifconf.ifc_len/sizeof(struct ifreq); ifDepenDebufPrintf ( ("discoverInterfaces: %ld net intf(s) found\n", nelem) ); for (pifreq = pIfreqList; pifreq<(pIfreqList+nelem); pifreq++){ status = socket_ioctl(socket, SIOCGIFFLAGS, pifreq); if (status) { ca_printf ("discoverInterfaces: net intf flags fetch for %s failed\n", pifreq->ifr_name); continue; } /* * dont bother with interfaces that have been disabled */ if (!(pifreq->ifr_flags & IFF_UP)) { ifDepenDebufPrintf ( ("discoverInterfaces: net intf %s was down\n", pifreq->ifr_name) ); continue; } /* * dont use the loop back interface */ if (pifreq->ifr_flags & IFF_LOOPBACK) { ifDepenDebufPrintf ( ("discoverInterfaces: ignoring loopback interface: %s\n", pifreq->ifr_name) ); continue; } /* * Fetch the local address for this interface */ status = socket_ioctl(socket, SIOCGIFADDR, pifreq); if (status){ ifDepenDebufPrintf ( ("discoverInterfaces: could not obtain addr for %s\n", pifreq->ifr_name) ); continue; } /* * If its not an internet inteface * then dont use it. */ if (pifreq->ifr_addr.sa_family != AF_INET) { ifDepenDebufPrintf ( ("discoverInterfaces: interface %s was not AF_INET\n", pifreq->ifr_name) ); continue; } /* * save the interface's IP address */ pInetAddr = (struct sockaddr_in *)&pifreq->ifr_addr; /* * if it isnt a wildcarded interface then look for * an exact match */ if (matchAddr.s_addr != htonl(INADDR_ANY)) { if (pInetAddr->sin_addr.s_addr != matchAddr.s_addr) { ifDepenDebufPrintf ( ("discoverInterfaces: net intf %s didnt match\n", pifreq->ifr_name) ); continue; } } /* * If this is an interface that supports * broadcast fetch the broadcast address. * * Otherwise if this is a point to point * interface then use the destination address. * * Otherwise ignore the * interface. */ if (pifreq->ifr_flags & IFF_BROADCAST) { status = socket_ioctl( socket, SIOCGIFBRDADDR, pifreq); if (status) { ifDepenDebufPrintf ( ("discoverInterfaces: net intf %s: bcast addr fetch fail\n", pifreq->ifr_name) ); continue; } } else if(pifreq->ifr_flags & IFF_POINTOPOINT){ status = socket_ioctl( socket, SIOCGIFDSTADDR, pifreq); if (status){ ifDepenDebufPrintf ( ("discoverInterfaces: net intf %s: pt to pt addr fetch fail\n", pifreq->ifr_name) ); continue; } } else{ ifDepenDebufPrintf ( ("discoverInterfaces: net intf %s: not pt to pt or bcast\n", pifreq->ifr_name) ); continue; } ifDepenDebufPrintf ( ("discoverInterfaces: net intf %s found\n", pifreq->ifr_name) ); pNode = (caAddrNode *) calloc(1,sizeof(*pNode)); if(!pNode){ printf ("discoverInterfaces: malloc failed for net intf %s: \n", pifreq->ifr_name); continue; } pNode->destAddr.in = *pInetAddr; pNode->destAddr.in.sin_port = htons(port); /* * LOCK applied externally */ ellAdd(pList, &pNode->node); } free(pIfreqList); } vxWorks Tormento ______________________________________________ Jeffrey O. Hill E-mail johill@lanl.gov LANL MS H820 Voice 505 665 1831 Los Alamos NM 87545 USA FAX 505 665 5107 From owner-vxwexplo-process Wed Sep 22 12:47:20 1999 From: Mauro Galarda Martins Date: Wed Sep 22 12:47:27 PDT 1999 Subject: device driver for 85c30 vxWorks Tornado Hi, I'm looking for some help to develop a driver for the 85c30 to run in HDLC synchronous mode. I'm working with powerpc architecture and I would like to use DMA also. In fact I will use this drive with PPP protocol, which is pretty similar to hdlc. Can anybody help me? thanks in advance. -- _______________________________________________________ Mauro Jose Galarda Martins Siemens Tel TR-DS Rua Pedro Gusso, 2635 CEP 81310-900 Curitiba - Parana' - Brasil Tel: +55 41 3416457 Fax: +55 41 3415620 e-mail: mauro.martins@siemens.com.br _______________________________________________________ From owner-vxwexplo-process Wed Sep 22 12:51:51 1999 From: "Hamon, Kent" Date: Wed Sep 22 12:51:54 PDT 1999 Subject: C++ constructor undefined symbol? Hello, I get a Tornado 1 debugger error: "object module contains undefined symbols: _vt$8CVCQCard" when loading my *.out file. The name of my constructor is "CVCQCard" and I've double checked the spelling and added the base class constructor also. (No debugger complaints about the base class constructor). Does anyone know what I might be missing? Thanks, Kent vxWorks Tornado From owner-vxwexplo-process Wed Sep 22 21:45:23 1999 From: "Viswanath Bandaru" Date: Wed Sep 22 21:45:27 PDT 1999 Subject: Tornado II hi! group! sorry for posting a non - tech question. But is it true that we donot have a Floating licence verion of Tornado II ? (We'll have Licence server on a machine and Tornado in other machines, usage monitored by the license server). Thanks and Regards, Viswanath From owner-vxwexplo-process Thu Sep 23 01:14:50 1999 From: Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) Date: Thu Sep 23 01:14:54 PDT 1999 Subject: Once again: Allocation in Shell I post it once again, because i don't see it on the exploder! Keywords missing? VxWorks, TORNADO From scheurer@zfs.tz.uni-stuttgart.de Wed Sep 22 23:23:32 1999 Date: Wed, 22 Sep 1999 23:41:23 +0100 From: "F. Scheurer" X-Accept-Language: de MIME-Version: 1.0 To: VxWorks-Mail-Exploder UserGroup Subject: Allocation in Shell Content-Transfer-Encoding: 7bit Hi VxWorkers, at the moment i want to setup my MVME2700-system to accept a local SCSI fixed disk. I test it on the shell. It would be nice to end up in a flexible configuration using the startup script. I arrived at the point where i have to fill the DOS_VOL_CONFIG-structure. Is there any way to exactly malloc the amount of memory used by the DOS_VOL_CONFIG-structure on the shell? Note: It is really no point that using compiled code or the allocation of a big block of memory would work, but this isn't what i want. Best regards F. Scheurer -- +-------------------------------------------------+ | Dipl.-Ing. F. Scheurer | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | | WWW http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | privat: Waldeckstrasse 31 | | D-73035 Goeppingen | | Telefon +49 7161 686362 Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | +-------------------------------------------------+ From owner-vxwexplo-process Thu Sep 23 01:50:53 1999 From: David Laight Date: Thu Sep 23 01:50:56 PDT 1999 Subject: Re: device driver for 85c30 > Subject: device driver for 85c30 > Submitted-by: Mauro Galarda Martins > > Hi, > > I'm looking for some help to develop a driver for the 85c30 to run in > HDLC synchronous mode. I'm working with powerpc architecture and > I would like to use DMA also. In fact I will use this drive with > PPP protocol, which is pretty similar to hdlc. Can anybody help me? > Writing a good SCC driver is an art in itself, a few things to watch out for: 1) The cycle recovery time is comparitivly long, ensure your hardware enforces it. You can't do it in software - the DMA cycles have the same constraint. 2) The SCC latches write data on the LEADING edge of the WR pulse. This means that during DMA the RD (to memory) and WR (to SCC) can't be the same signal. 3) Remember to read the external status register twice (to clear the latch) if you want the current values. 4) Don't try to abort TX frames, you are likely to send the following frame without an intervening flag. 5) Be very careful on receive aborts (the abort isn't an end of frame condition). 6) You need to wait for TWO transmit underrun interrupts at the end of frame before loading the DMA controller for the next frame (or you don't get a flag after the CRC). 7) The receive status fifo (the one that stores frame lengths) didn't work properly last time I tried to use it (I've forgotten why). 8) The AMD and ZILOG parts are slightly different, get the correct data sheet. 9) Some parts won't run at the declared maximum data rate (as a fraction of the clock rate) due to clock jittter (ie not 50-50 mark space) on the clocks. 10) The digital PLL may not synchonise quickly enough (or ever) if there is a lot of clock jitter. 11) Half duplex is almost impossible - you don't know when the last flag has left the transmit pin and it is safe do drop RTS. The most sucessful technique I've used is to use local loopback and receive the frame you are transmitting yourself. 12) Recieving back to back HDLC frames requires the cpu process the end of frame interrupt (and get the next DMA started) within 3 byte times (you have slightly longer on the newer parts (130 and 230?) because they have a deeper receieve fifo), this is probably ok on vxWorks, but we were writing drivers for Unix... 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 Sep 23 01:51:36 1999 From: "Mike Davies" Date: Thu Sep 23 01:51:39 PDT 1999 Subject: Re: C++ constructor undefined symbol? Hi, if you run the c++filt program (for a Windows hosted 386 target this is "cppfilt386 _vt$8CVCQCard" ) from the command line then it will tell you what the function prototype looks like. In this case the output is : CVCQCard virtual table It looks like for some reason the compiler can't find the v_table for that class. Are you sure you define an instance of the class somewhere other than a header file ? Mike Davies This is my opinion, not my employers'. > >Hello, > >I get a Tornado 1 debugger error: > >"object module contains undefined symbols: _vt$8CVCQCard" > >when loading my *.out file. > >The name of my constructor is "CVCQCard" and I've double checked the >spelling >and added the base class constructor also. (No debugger complaints about >the >base class constructor). > >Does anyone know what I might be missing? > >Thanks, >Kent > >vxWorks Tornado From owner-vxwexplo-process Thu Sep 23 02:19:20 1999 From: Jan Uerpmann Date: Thu Sep 23 02:19:24 PDT 1999 Subject: WindView crashes Hi I am using VxWorks 5.3 and Tornado 1.0.1 on a pc486. I have still some problems with windview. If I start the following commands from the shell everything works as it is supposed to: wvHostInfoInit("134.169.29.253", 6164); wvOn(2) wvOff But if I try to do this from within a program the target crashes... This is my test program: int Init = 1; int InitFunc() { wvHostInfoInit("134.169.29.253", 6164); Init = 0; return 0; } int Go() { if (Init) InitFunc(); wvEvtLogEnable(TASK_STATE); taskDelay(1); wvEvtLogDisable(); return 0; } The target crashes at various points after wvEvtLogEnable()... Any help appreciated. Regards, Jan From owner-vxwexplo-process Thu Sep 23 02:59:32 1999 From: Simon Liston Date: Thu Sep 23 02:59:36 PDT 1999 Subject: Header File Dependency Problems TORNADO Hi, I've just started using Tornado 2.0 (we have the simulator package at the moment soon to be followed by StrongArm & Pentium full development kits.) and am having problems compiling a simple project. I get the following message when compiling a C file .... make: *** No rule to make target `foo.h', needed by `applic.o'. Stop. 'foo.h' is included in the applic.c file with just a '#include foo.h' i.e. no full path. The auto dependency generator correctly determined that foo.h is required by the C file BUT foo.h is not located in the same directory as the C file. Normally a -I and include pathname as part of the C compiler options is how the .h files can be located if not in the current directory. I've tried changing the source search path entries (under the debug menu) to include the directory where foo.h is but that doesn't seem to make any difference. Am I missing something really obvious here or have I bumped into a limitation of Tornado 2 (which would surprise me indeed!). Any help would be appreciated - I can't find anything in the manuals (so far). Best Regards, Simon. From owner-vxwexplo-process Thu Sep 23 04:00:15 1999 From: daemon@csg.lbl.gov Date: Thu Sep 23 04:00:20 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Sep 23 04:00:09 PDT 1999 Subject: Re: Signal Handler for Each New Task Subject: App activation Subject: Re: BSP modifications and generating vxWorks image with Tornado II Subject: Re: About Global variables and main() functions Subject: Re: Clearcase/Tornado Anyone? Subject: Interrupt Problems Subject: multiprocessor configure Subject: Re: VxWorks, VxMP Subject: Help, not able to make two VxSim on different host to talk! Subject: intConnect Subject: Re: C++ constructor undefined symbol? Subject: SVGA Card Subject: SVGA Card Subject: Re: inetstatShow Subject: Re: About Global variables and main() functions Subject: Re: About Global variables and main() functions Subject: help for vxMP Subject: How to link Motorola FADS302 and Tornado2.0? Subject: Re: Hierarchical projects under Tornado II? Subject: Re: vxWorks development system Subject: Re: Tornado2 for NT machine: Where's cpp? Subject: Re: inetstatShow ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Signal Handler for Each New Task Date: Wed, 22 Sep 1999 12:49:47 GMT From: Ran@127.0.0.1 (Ran Shalgi) Organization: Only when absolutely necessary Message-ID: <37e8ced7.875803228@news.inter.net.il> References: <37E84096.321E4111@uswest.net> We had a similar problem and ended up writing a function called ourTaskSpawn(...). This function always spawns a special function which connects all the signals to inself and then calls the original callback a sketch pseudo-code is something like this: void ourTaskSpawn(FUNCPTR entryPoint, ...) { taskSpawn(ourTask, entryPoint, int1, int2, ...); } void ourTask(FUNCPTR func, int1, int2, ...) { connectAllSignals(...); func(); } This way func() runs in the context of a task which has all the signals connected. We then use ourSpawn() exclusively instead of taskSpawn(). On Tue, 21 Sep 1999 21:36:06 -0500, Glenn Hofmeister wrote: >Greetings, > >I am using vxSim 5.3.1 and what I am attempting to do is set up a signal >handler to process a segmentation violation in a task. However, what I >am >really trying to do is for each new task that starts, this signal >handler >is to be set up. So I have been trying to use taskCreateHookAdd() to >set >up a function that is called when a new task starts. However, when the >function gets called, the task is not the new task, but it does have a >pointer to the new task's WIND_TCB. I do not know what fields I could >update in the new task's WIND_TCB to set up a signal handler for the new >task. Has anyone ever tried something like this? > >My signal handler is: > >LOCAL jmp_buf env; >void sigHandler > ( > int sig, > int code, > SIGCONTEXT *sigContext > ) >{ > switch (sig) > { > case SIGSEGV: > printErr ("\nSignal SIGSEGV received\n"); > break; > } > longjmp (env, sig); >} > >My added Hook routine: >void myCreateHook (pTcb) > WIND_TCB *pTcb; >{ > >} > >My code to set up the hook: >taskCreateHookAdd((FUNCPTR) myCreateHook); > > >Thanks, > >Glenn >glennh@uswest.net Ran Shalgi Expand Networks, Inc http://www.expand.com - --- Replace 127.0.0.1 with expand.com when replying by e-mail --------------------------- Newsgroups: comp.os.vxworks Subject: App activation Date: Wed, 22 Sep 1999 03:56:25 +0200 From: "Santino Troiani" Organization: Customer of Flashnet S.p.A. - http://www.flashnet.it Message-ID: <7san4k$to9$1@news.flashnet.it> I'm working on a Pentium board with a TFFS managed storage. At the end of the BSP startup I need to activate some middlware modules, so I put into usrAppInit() the code to: 1. load the module into memory, using ld() or loadModule() 2. retry the module entry point symbol value using symFindByName() on the VxWorks symbol table 3. call the entry point The step 2. always fails with error 0x1C0001 - SYMBOL_NOT_FOUND. Have anyone some suggestion? - ---------------------------------------------------------------------------- - - Santino Troiani Tecnost Mael via Turanense Km. 41,500 67061 CARSOLI (AQ) Phone ++39 0863 994 537 Fax ++39 0863 994 538 e-mail stroiani@mael.it - ---------------------------------------------------------------------------- - - --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP modifications and generating vxWorks image with Tornado II Date: 22 Sep 1999 15:19:45 GMT From: "GKE" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: <01bf050d$835a2520$3c1e140a@gerard> References: <7rb08g$3ff5@overload.lbl.gov> Hi Reyhan, Tornado 2 ignores some dependencies to limit the number of files in your file list. You can select which directories are ignored by selecting Build - Dependencies...- Advanced... from the menu. Under 'Read-only dependency path' you can select which directories are ignored. I removed the last one, named 'E:\TORNADO2\target\config' (on my machine), and re-scanned the dependencies. This has two effects: Files like sysSerial.c are in the Project Files list AND they are in the dependencies list. Hope this helps, Gerard. Reyhan ERGUN wrote: > Hi, > I use Tornado 2, and I want to build vxWorks (a ram-based image) after I > make several additions to the BSP. > -snip- > If I include these files into workspace I encounter many problems > regarding to dependencies for newly included file sysSerial.c. > > I wonder how I could modify BSP by using Tornado 2. > Any comments appreciated. > > Best regard, Reyhan > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About Global variables and main() functions Date: 22 Sep 1999 12:25:53 -0400 From: comeau@panix.com (Greg Comeau) Organization: Comeau Computing; http://www.comeaucomputing.com Message-ID: <7savuh$pqu$1@panix3.panix.com> References: <37E5CF37.F7582619@cselab.snu.ac.kr> <37E5D543.9B7119E5@aracnet.com> Reply-To: comeau@comeaucomputing.com In article <37E5D543.9B7119E5@aracnet.com> engineer_scotty writes: >Lee Dong Yul wrote: >> >> Hello, We are developing some programs on VxWorks OS. > >To sum up all your questions: vxWorks ain't like Unix. > >vxWorks doesn't have the concept of separate "programs" with >separate address spaces, and separate global variables. There is >only one address space--and when you load new modules with ld(), >the global variables in that module are added to that address space. We recently completed the first phase of porting our compiler, Comeau C++, to the TMS320C82, with the TI C compiler (hosted on Solaris/SPARC) We set it up traditionally "like UNIX", and based upon your points below wonder if there is a problem, a culture, or just pragmatics tht might not matter. See below. >> 1. Why can't we use main function with C language? >> >> When we programmed some applications with main function, We couldn't >> download those applications. Undefined symbol (__eabi) error occurred. > >Lemme guess. PowerPC, right? :) > >Including main() causes the compiler/linker to automagically link in a >bunch of Other Stuff. In Unix, including main() causes prologue code >(which initializes the C standard library) to run, etc. In C++, this >prologue code initializes static constructors. Although Standard C++ does not require main() to be the static ctor "clue in", we use it as such often on some platforms, since it is allowed. Even despite the one address space comment above, what's wrong with treating it that way? >vxWorks on the powerPC, on the other hand, corresponds to a >specification >called EABI (Embedded Application Binary Interface); which calls for a >function __eabi to be invoked when a program starts. This is, however, >Not What You Want in vxWorks. > >The bottom line: don't call anything main(). In brief, what's wrong with __eabi being called on vxWorks? >> 2. In tornado shell, we can't execute some programs many times. All >> Those programs have global variables. So, After executing those >> programs, Global variables contain some values that are different from >> program startup values. >> >> How do we initialize all global variables on program startup? > >Do so explicitly. The static initialization happens only once. For >those objects linked with the kernel, static initialization happens when >the system starts. If you use the dynamic loader, the initialization >happens when the module is loaded. However, "restarting" the program >does >not cause globals to be initialized. So with whatever C++ combo with Tornado you're referring to, restarting won't redo static initializers? >Oh, and one more thing. The automatic cleanup that you get in Unix when >a program dies (all memory blocks freed, stdio flushed, all files >closed) >does NOT happen in vxWorks. You **MUST** free all resources you use >yourself. We also added in static destruction stuff, why is that a problem (assuming it's needed)? >> Why don't vxworks initialize global variables on program startup. > >vxWorks views the entire system as a "program"; all the modules are part >of that program. This doesn't seems like a good reason, alone that is. This is to say, this is somewhat akin to saying that shared libs on UNIX of DLLs on Win shouldn't take care of any initialization, and/or arrange for them to be done. >> When Initialization occurs? Object code downloading? > >Globals defined in the module are initialized when you download the >module, >but at no time after. (Unless you unload the module and load it again; >this will ause them to be re-initialized). Data only? >> 3. On program exit, is dynamically allocated memory automatically freed? > >Nope. > >> Otherwise, we must free all dynamically allocated memory explicitly? > >Yup. This implies it's either erroneous or unncessary to perform static destructors. Although we can imagine apps that won't care, the above sems to imply it's always true??? - - Greg - -- Comeau Computing, 91-34 120th Street, Richmond Hill, NY, 11418-3214 Producers of Comeau C/C++ 4.2.38 -- NOTE 4.2.42 BETAS NOW AVAILABLE Email: comeau@comeaucomputing.com / Voice:718-945-0009 / Fax:718-441-2310 *** WEB: http://www.comeaucomputing.com *** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Clearcase/Tornado Anyone? Date: Wed, 22 Sep 1999 17:51:57 GMT From: Ted Dennison Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7sb4vn$k4b$1@nnrp1.deja.com> References: <37E7AFF8.158858C7@email.mot.com> In article <37E7AFF8.158858C7@email.mot.com>, qjb005@email.mot.com wrote: > Does anyone install Tornado and associated tools in a clearcase VOB? > Do you like the setup? Are there caveats? We generally wouldn't do that with development tools, as executables can't be revised efficently and previous versions can always be retrieved from the CDROM they were loaded off of (Assuming proper inventory controls). The only Tornado stuff we keep in clearcase is the files we change to customize our kernel. Namely our master makefile, config.h, and configdb.h. Our master makefile copies these files into the configuration directory ( $(WIND_BASE)/target/config/pc486/ ), and invokes the vxWorks makefile to build the kernel. (Note: We are running Tornado 1. If this will have to change w/ Tornado II, please warn me. We'll be upgrading soon.) - -- T.E.D. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt Problems Date: Wed, 22 Sep 1999 17:36:24 GMT From: Ken Ostaszewski Organization: Boeing, Inc. Message-ID: <37E91398.5BA60CBD@boeing.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Hi Everyone, I have a MV2604 with a VME memory card. I am attempting to have the memory card generate an interrupt. According to the file target.txt the interrupt vector for VME interrupt 0 is as follows: 15 PCI Universe VME INT 0 The problem later is that target.txt states that only even vector number will generate interrupts. So my question is, will the VME INT 0, which is the only one provided by WindRiver ever generate an interrupt to the MV2604 board? What else can I do? Help Please. Ken Ostaszewski --------------------------- Newsgroups: comp.os.vxworks Subject: multiprocessor configure Date: Wed, 22 Sep 1999 14:44:10 -0400 From: Lei Sun Organization: University of Central Florida Message-ID: <37E9237A.7A5950EC@pegasus.cc.ucf.edu> Hi: I have a VME box and two MC68K board (one is MVME167 the Other is MVME167C) which is going to load vxWorks system. I want to configure these two board to do parallel processing program by vxMP (shared memory object). host name IP board VME1 132.170.204.71 MVME167C VME2 132.170.204.71 MVME167 (jumpered as VME system controller) It was said that in VME box , the left most slot is the master, but when I plug the VME1 into it and power on, system doesn't boot. However, when I plug it into slot 4 (from the left), system boot successfully. Is that means slot 4 is the master slot? And also, since the VME2 has been jumpered as system controller, no matter which slot I plug it into, it always can be booted( without VME1). when I plug VME2 in slot 1 and VME1 in slot 4, system can be booted, but in the local console, boot parameter shows that target name is VME1, I was wondering which one is actually the master board. If I want to use vxMP, do I need to configure the two board as a shared memory network firstly? thanks! your answer is apprieciated! lei sun --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks, VxMP Date: Wed, 22 Sep 1999 13:48:33 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <37E92481.8DDBE963@aps.anl.gov> References: Roger Holmquist wrote: > > 1) We haven't managed to mix Motorola 68K and PPC-cards in our > MP-environment and make them talk to each other. (is this bug fixed in > TornadoII ?) The behaviour is going to be very dependent on the particular board you're using and its BSP, not on Tornado itself. The Tundra Universe VME interface chip used on some PPC boards had major problems which the Universe-II seems to have solved. The mv2700 BSP states that you must use a 2700 board as the master (proc 0) CPU in most cases where you have mixed architectures. However even the latest mv2700 BSP (version 1.1/5.1) didn't do everything we needed to directly replace an mv167. I have modified the BSP for use here so sysBusTas() to a VME address generates the same 8-bit RMW operation that the mv167 does, which was a requirement for one of our drivers. I have successfully used the sm network between mv167, mv172 and mv2700 CPUs under Tornado 1.0.1, although you shouldn't need my BSP mods to get this working as long as you've set up the config.h file properly. - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Help, not able to make two VxSim on different host to talk! Date: Wed, 22 Sep 1999 19:03:04 GMT From: hor@my-deja.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7sb94v$nb7$1@nnrp1.deja.com> Hi everyone, I have two VxSim sessions running on two different NT hosts. But I can't make them to talk. Configuration follows: NT1 - VxSim1: ip forwarding enabled on the host ulip installed as 90.0.0.254 host ip: 114.119.77.21 local ip: = host ip target ip: 90.0.0.1 routing for 90.0.0.2 to 114.119.77.22 added on NT NT2 - VxSim2: ip forwarding enabled on the host ulip installed as 90.0.0.254 host ip: 114.119.77.22 local ip: = host ip target ip: 90.0.0.2 routing for 90.0.0.1 to 114.119.77.21 added on NT I can ping to VxSim1 and VxSim2 from both NT1 and NT2. However, I can not ping from vxSim1 to VxSim2 or vice versa. And I can't event ping NT1 from VxSim2 or ping NT2 from VxSim1. Within VxSim1: I can ping to 90.0.0.1, 90.0.0.254 and 114.119.77.21 but not 114.119.77.22! Within VxSim2: I can ping to 90.0.0.2, 90.0.0.254 and 114.119.77.22 but not 114.119.77.21! I tried to add routing entry to the route table internal to VxSim. But it doesn't help either! I tried both mRouteAdd ("114.119.77.22","90.0.0.1",0xffffffff,0,0) and mRouteAdd ("114.119.77.22","114.119.77.21",0xffffffff,0,0) to the VxSim1. No luck! I also tried to change the ip of the ulip to something other than 90.0.0.254 (e.g. 80.0.0.254), the same problem exist. I can't even ping to 80.0.0.254 from within the VxSim. Any hint or info is greatly appreciated. Ron Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: intConnect Date: Wed, 22 Sep 1999 13:30:26 -0700 From: Kevin Havig Organization: Posted via Supernews, http://www.supernews.com Message-ID: <37E93C61.8CB9DF@navair.navy.mil> I am using VxWorks 5.4/Tornado 2 on a Motorola MVME162 processor. I have an intelligent serial card with four ports, each of which can generate an interrupt; the vectors it is set up to use are 70, 71, 72, and 73. The processing of each of these interrupts is so similar that I would like to use a single ISR to process all four. I try to set this up using four intConnect commands as follows: intConnect(INUM_TO_IVEC(70),portisr,0) intConnect(INUM_TO_IVEC(71),portisr,1) intConnect(INUM_TO_IVEC(72),portisr,2) intConnect(INUM_TO_IVEC(73),portisr,3) The routine portisr is the routine to process the interrupt. It has a case statement to process the parameter passed to it by the wrapper built by intConnect. This parameter will have a value of 0, 1, 2 or 3 as specified by the last parameter of intConnect. If I comment out all but the first intConnect port 0 interrupts work just fine. Whenever I enable one of the other intConnects things go wrong. I get the following message in my terminal window and my VME system reboots: workQPanic: Kernel work queue overflow. Is this a valid way to use intConnect? If so, any idea of what I am doing wrong? Kevin Havig Code 526110D, Echo Range Naval Air Warfare Center China Lake, CA Havigkk@navair.navy.mil (760) 939-9248 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ constructor undefined symbol? Date: 22 Sep 1999 16:44:17 -0400 From: comeau@panix.com (Greg Comeau) Organization: Comeau Computing; http://www.comeaucomputing.com Message-ID: <7sbf31$j0m$1@panix3.panix.com> References: <7sbdbs$i164@overload.lbl.gov> Reply-To: comeau@comeaucomputing.com In article <7sbdbs$i164@overload.lbl.gov> "Hamon, Kent" writes: > >Hello, > >I get a Tornado 1 debugger error: > >"object module contains undefined symbols: _vt$8CVCQCard" > >when loading my *.out file. > >The name of my constructor is "CVCQCard" and I've double checked the >spelling >and added the base class constructor also. (No debugger complaints about >the >base class constructor). > >Does anyone know what I might be missing? I don't know your compiler, but vt smells like "virtual function table" for CVCQCard. - - Greg - -- Comeau Computing, 91-34 120th Street, Richmond Hill, NY, 11418-3214 Producers of Comeau C/C++ 4.2.38 -- NOTE 4.2.42 BETAS NOW AVAILABLE Email: comeau@comeaucomputing.com / Voice:718-945-0009 / Fax:718-441-2310 *** WEB: http://www.comeaucomputing.com *** --------------------------- Newsgroups: comp.os.vxworks Subject: SVGA Card Date: Wed, 22 Sep 1999 17:04:30 -0400 From: Andy Yang Organization: ITS - NetNews Message-ID: <37E9445E.F3F86AC0@anchor.ho.att.com> - --------------3C5B4EE7AFBB02FB19AE94BD Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi: I'm looking for a video card with folloing requirements: * 1024 x 768 16-bit (16K) colors * Running on a Pentium * Either AGP or PCI or ISA * VxWorks driver available (if needed) * Inexpansive Can anyone give me an answer? Thanks! Andy Yang - --------------3C5B4EE7AFBB02FB19AE94BD Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hi:

I'm looking for a video card with folloing requirements:

  • 1024 x 768 16-bit (16K) colors
  • Running on a Pentium
  • Either AGP or PCI or ISA
  • VxWorks driver available (if needed)
  • Inexpansive
Can anyone give me an answer?   Thanks!
 

Andy Yang
 
  - --------------3C5B4EE7AFBB02FB19AE94BD-- --------------------------- Newsgroups: comp.os.vxworks Subject: SVGA Card Date: Wed, 22 Sep 1999 17:07:05 -0400 From: Andy Yang Organization: ITS - NetNews Message-ID: <37E944F9.A67983E7@anchor.ho.att.com> - --------------A4340B197B6F963DBCCEF78D Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi: I'm looking for a video card with folloing requirements: * 1024 x 768 16-bit (16K) colors * Running on a Pentium * Either AGP or PCI or ISA * VxWorks driver available (if needed) * Inexpansive Can anyone give me an answer? Thanks! Andy Yang - --------------A4340B197B6F963DBCCEF78D Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hi:

I'm looking for a video card with folloing requirements:

  • 1024 x 768 16-bit (16K) colors
  • Running on a Pentium
  • Either AGP or PCI or ISA
  • VxWorks driver available (if needed)
  • Inexpansive
Can anyone give me an answer?  Thanks!
 

Andy Yang
 
  - --------------A4340B197B6F963DBCCEF78D-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: inetstatShow Date: Wed, 22 Sep 1999 17:28:49 -0400 From: "JJ15C3600-B.PATEL(HOP092)NONE" Organization: Lucent Technologies Message-ID: <37E94A11.6040BB51@lucent.com> when we do inetstatShow, we get a listing of field - PCB ( protocol control block). How does one know to which vxworks task this pcb corresponds to? for example - PCB Proto Recv-Q Send-Q Local Address Foreign Address (state) - -------- ----- ------ ------ ------------------ ------------------ ------- 2117788 TCP 0 0 0.0.0.0.1024 0.0.0.0.0 LISTEN 2117470 TCP 0 0 0.0.0.0.3008 0.0.0.0.0 LISTEN 2117260 TCP 0 0 0.0.0.0.1025 0.0.0.0.0 LISTEN 2117050 TCP 0 0 0.0.0.0.5001 0.0.0.0.0 LISTEN 2116c30 TCP 0 0 0.0.0.0.21 0.0.0.0.0 LISTEN 2117fc8 UDP 0 0 0.0.0.0.36096 0.0.0.0.0 211804c UDP 0 0 0.0.0.0.25344 0.0.0.0.0 2117f44 UDP 0 0 0.0.0.0.35840 0.0.0.0.0 2117ec0 UDP 0 0 0.0.0.0.25088 0.0.0.0.0 2117e3c UDP 80 0 0.0.0.0.24832 0.0.0.0.0 2117db8 UDP 0 0 0.0.0.0.24576 0.0.0.0.0 2117d34 UDP 60 0 0.0.0.0.24320 0.0.0.0.0 2117c2c UDP 0 0 0.0.0.0.24064 0.0.0.0.0 2117cb0 UDP 0 0 0.0.0.0.35584 0.0.0.0.0 Any reply is appreciated. Bhavesh Patel Lucent Technologies --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About Global variables and main() functions Date: Wed, 22 Sep 1999 16:42:55 -0700 From: Eric Flynn Organization: Hewlett Packard Cupertino Site Message-ID: <37E9697F.C22DCCF8@wwg2.sid.hp.com> References: <37E5CF37.F7582619@cselab.snu.ac.kr> <37E5D543.9B7119E5@aracnet.com> <7savuh$pqu$1@panix3.panix.com> Reply-To: eflynn@wwg2.sid.hp.com Greg Comeau wrote: > In article <37E5D543.9B7119E5@aracnet.com> engineer_scotty writes: > >Lee Dong Yul wrote: > >> > >> Hello, We are developing some programs on VxWorks OS. > > > >To sum up all your questions: vxWorks ain't like Unix. > > > >vxWorks doesn't have the concept of separate "programs" with > >separate address spaces, and separate global variables. There is > >only one address space--and when you load new modules with ld(), > >the global variables in that module are added to that address space. > > We recently completed the first phase of porting our compiler, > Comeau C++, to the TMS320C82, with the TI C compiler (hosted on Solaris/SPARC) > We set it up traditionally "like UNIX", and based upon your points > below wonder if there is a problem, a culture, or just pragmatics > tht might not matter. See below. > > >> 1. Why can't we use main function with C language? > >> > >> When we programmed some applications with main function, We couldn't > >> download those applications. Undefined symbol (__eabi) error occurred. > > > >Lemme guess. PowerPC, right? :) > > > >Including main() causes the compiler/linker to automagically link in a > >bunch of Other Stuff. In Unix, including main() causes prologue code > >(which initializes the C standard library) to run, etc. In C++, this > >prologue code initializes static constructors. > > Although Standard C++ does not require main() to be the static ctor > "clue in", we use it as such often on some platforms, since it is allowed. > Even despite the one address space comment above, what's wrong > with treating it that way? Because the single address space of VxWorks already defines a main( ) andwill not allow a duplicate. Period. > > > >vxWorks on the powerPC, on the other hand, corresponds to a > >specification > >called EABI (Embedded Application Binary Interface); which calls for a > >function __eabi to be invoked when a program starts. This is, however, > >Not What You Want in vxWorks. > > > >The bottom line: don't call anything main(). > > In brief, what's wrong with __eabi being called on vxWorks? Nothing, just the fact that if you're going to write your own main( ), you'regoing to have to supply your own __eabi. > > > >> 2. In tornado shell, we can't execute some programs many times. All > >> Those programs have global variables. So, After executing those > >> programs, Global variables contain some values that are different from > >> program startup values. > >> > >> How do we initialize all global variables on program startup? > > > >Do so explicitly. The static initialization happens only once. For > >those objects linked with the kernel, static initialization happens when > >the system starts. If you use the dynamic loader, the initialization > >happens when the module is loaded. However, "restarting" the program > >does > >not cause globals to be initialized. > > So with whatever C++ combo with Tornado you're referring to, > restarting won't redo static initializers? C++ and Tornado notwithstanding, VxWorks initializes your globals at module load time only.Thus the requirement that you initialize any other time you need it done. > > > >Oh, and one more thing. The automatic cleanup that you get in Unix when > >a program dies (all memory blocks freed, stdio flushed, all files > >closed) > >does NOT happen in vxWorks. You **MUST** free all resources you use > >yourself. > > We also added in static destruction stuff, why is that a problem > (assuming it's needed)? > > >> Why don't vxworks initialize global variables on program startup. > > > >vxWorks views the entire system as a "program"; all the modules are part > >of that program. > > This doesn't seems like a good reason, alone that is. > This is to say, this is somewhat akin to saying that shared libs > on UNIX of DLLs on Win shouldn't take care of any initialization, > and/or arrange for them to be done. > I am not sure if your objection here is practical or philosophical?The practical answer is, you must initiliaze your globals in VxWorks *each time* your program runs. > >> When Initialization occurs? Object code downloading? > > > >Globals defined in the module are initialized when you download the > >module, > >but at no time after. (Unless you unload the module and load it again; > >this will ause them to be re-initialized). > > Data only? data inits and ctors called. > > > >> 3. On program exit, is dynamically allocated memory automatically freed? > > > >Nope. > > > >> Otherwise, we must free all dynamically allocated memory explicitly? > > > >Yup. > > This implies it's either erroneous or unncessary to perform static > destructors. Although we can imagine apps that won't care, > the above sems to imply it's always true??? If you called malloc() or new, you are *always* responsible for calling free() or delete, respectively.A statically declared C++ object will die on its own. > > > - Greg > -- > Comeau Computing, 91-34 120th Street, Richmond Hill, NY, 11418-3214 > Producers of Comeau C/C++ 4.2.38 -- NOTE 4.2.42 BETAS NOW AVAILABLE > Email: comeau@comeaucomputing.com / Voice:718-945-0009 / Fax:718-441-2310 > *** WEB: http://www.comeaucomputing.com *** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About Global variables and main() functions Date: Wed, 22 Sep 1999 18:19:01 -0700 From: engineer_scotty Organization: aracnet.com Message-ID: <37E98005.A4DF4155@aracnet.com> References: <37E5CF37.F7582619@cselab.snu.ac.kr> Georg Feil wrote: > > Lee Dong Yul writes: > >1. Why can't we use main function with C language? > > This one's been covered pretty well... just use another name. > > >2. In tornado shell, we can't execute some programs many times. All > >Those programs have global variables. So, After executing those > >programs, Global variables contain some values that are different from > >program startup values. > >How do we initialize all global variables on program startup? > >Why don't vxworks initialize global variables on program startup. > > Think of VxWorks as one big continuously running program with multiple > threads (tasks). Global variables are not reinitialized just because > a task gets started. To get around this I've written initialization > modules for certain pieces of code that I want to be able to restart > like Unix programs, e.g. code ported from unix like editors and such. > The initialization module ensures that initialized variables and BSS > variables are set right, then branches to the (renamed) main function when > done. I've included an example copy below, although I use a separate slightly > customized init module for each such program. > > >When Initialization occurs? Object code downloading? > > Right. > > >3. On program exit, is dynamically allocated memory automatically freed? > > No. Good programmers clean up after themselves anyway. > > So here's that init module. Link this with your PROGRAM. You call > PROGRAM_start() which in turn calls PROGRAM_main(). There's also another > little function that goes before this one that I haven't included > (called "PROGRAM()") which repackages arguments so that old unix code > expecting argv/argc stuff will work. > > Good luck, > Georg > > ----------------------------- crt0.c ------------------------------ > #include > #include > #include > #include > #include > > /* > * PROGRAM startup module. > * > * This version for VxWorks. The main purpose is to properly initialize the > * initialized data space and bss between different runs of PROGRAM. The first > * time it executes, this routine saves away the entire initialized data space > * in malloc'd memory. On subsequent runs it restores the initialized data. > * It zeroes the bss area as well, but *be warned* that this should not > * be relied upon. Some BSS variables (those labeled as "common") seem to > * be placed outside the normal BSS block by the VxWorks > * loader, therefore *never* depend on uninitialized variables to be > * automatically set to 0. Explicitly initialize them to zero instead. > * > * 97/09/17 Georg Feil Changed to use moduleLib for segment addresses, > * added clearing of BSS. > * 97/06/27 Georg Feil Converted from assembly to C (for PPC port). > * 91/04/23 Georg Feil Cleaned up, added comments. > * 90/07/26 Georg Feil Fixed stack creep problem at malloc() call. > * 90/06/27 Georg Feil Initial revision, from s2 dev system. > */ > > /* flag shows whether initialized data area has been copied (!=0 means yes) > Value is start of malloc data area. */ > static char* init=0; > > void PROGRAM_start(int arg1, int arg2, int arg3, int arg4, int arg5, int arg6, > int arg7, int arg8, int arg9, int arg10) > { > int PROGRAM_main(); /* define main() this way to soothe the compiler */ > char* d0; > MODULE_ID mod_id; > MODULE_INFO mod_info; > > /* get segment info from moduleLib (properly gets last instance loaded) */ > mod_id=moduleFindByName("PROGRAM"); > if (mod_id==NULL) { > printf("PROGRAM_start(): Error, could not find module named \"PROGRAM\"\n"); > return; > } > if (moduleInfoGet(mod_id, &mod_info) == ERROR) { > printf("PROGRAM_start(): Error, could not get module info for ID 0x%lx\n", > (unsigned long)mod_id); > return; > } > > /* clear uninitialized data area (bss) *** see warning in comment at top */ > bzero(mod_info.segInfo.bssAddr, mod_info.segInfo.bssSize); > > /* Copy initialized data area to save (1st time) or restore it (other times). > (gets copied after end of uninitialized data (bss)) > Uses malloc() to obtain storage space -- Warning this is never freed. */ > if (init==0) { > /* malloc() enough memory to hold initialized data region */ > d0=malloc(mod_info.segInfo.dataSize); > if (d0==0) { > /* malloc() error (returned NULL), exit immediately */ > printf("PROGRAM_start(): Error, could not malloc\n"); > return; > } > > init=d0; /* save malloc() address in init */ > > /* save initialized data */ > bcopy(mod_info.segInfo.dataAddr, init, mod_info.segInfo.dataSize); > } > else { > /* restore initialized data */ > bcopy(init, mod_info.segInfo.dataAddr, mod_info.segInfo.dataSize); > } > > /* Jump to PROGRAM main() */ > PROGRAM_main(arg1,arg2,arg3,arg4,arg5,arg6,arg7,arg8,arg9,arg10); > } Good job. One modification that might be useful would be to malloc and copy all the arguments to the main function (treating arg1-arg10 as strings), for two reasons: 1) If you are porting UNIX code, many unix programs do nasty things like modfifying argv[]. In vxWorks, where arguments are not copied into the child program's "address space" by default; this behavior can cause All Sorts Of Problems. 2) Even if the child program doesn't modify the arguments; the caller may be passing strings to the client that sit on the stack (or are otherwise non-const) and may spawn the child as a separate task. Lots of nastiness results, of course, if a child task goes and scribbles on the parent's stack in a random fashion.... - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: help for vxMP Date: Wed, 22 Sep 1999 23:16:52 -0400 From: Lei Sun Organization: University of Central Florida Message-ID: <37E99BA4.1DA0B869@pegasus.cc.ucf.edu> Hi: Could anyone out there please "enlighten" me about installing VxMP. I have since used "installOption" to install the new libraries that contain the added functionality contained in VxMP but when I go to compile and create a new VxWorks kernel file for a Force40 target, the vxWorks file never brings in the added functions (smNameFind, smNameAdd, smNameRemove, smMemFree, smMemMalloc, smObjGlobalToLocal, smObjLocalToGlobal, etc). They come up as undefined symbols. The new library, "libMC68040gnuvx.a" contains the symbols and it is linked in with the kernel file, but why are the object files in the library not linked into the kernel file? I would think that references to the above functions would be taken care of with "usrRoot()" in "usrConfig.c" which calls "usrSmObjInit()" defined in "usrSmObj.c". But there's no "usrSmObj.o" linked in to the vxWorks kernel file. Do I need to modify the Makefile.MC68040gnu and provide a target line for "usrSmObj.c"? Isn't this all that's needed to initialize the Shared Memory Objects? Or do I have to place a reference to "smNameAdd" somewhere in a VxWorks module just so it's resolved? I would've thought that this is why you link in with the new libraries provided for you with the installation tapes for VxMP. These libraries contain the definitions for the newly created kernel file so that all future (& custom) functional references to "smNameAdd" will be resolved? Am I missing something here? Please provide me with any helpful information. Thanks in advance. lei sun --------------------------- Newsgroups: comp.os.vxworks Subject: How to link Motorola FADS302 and Tornado2.0? Date: Thu, 23 Sep 1999 13:54:19 +0800 From: "Andrew" Organization: DCI HiNet Message-ID: <7scf5f$av9@netnews.hinet.net> Hi everyone, I have Motorola FADS302 and Tornado2.0.But I can't link use SLIP or PPP or TSFS. Any hint or info is greatly appreciated. andrew --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hierarchical projects under Tornado II? Date: Tue, 21 Sep 1999 14:08:43 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37E7D7BB.944940E7@ciprico.com> References: <19990916015417.01471.00001746@ng-cg1.aol.com> <37e766e9.783661826@news.inter.net.il> Ran Shalgi wrote: > > I agree, the T2 project facility is completely unusable. It seems like > it was never tested by real programmers doing real work. The editor > is likewise utterly useless for anything but the most minor code > changes. Wwe've never even tried to use the project facility. I'm sure it is hideous. Speaking of the "editor" I was wondering if anyone has ever figured out how to set it so that files are displayed read-only? Given the crap focus rules for T2 (which we affectionately refer to as "Blow 2"), we often have to call up "undo" nine dozen times since we were typing a long command line and instead of being interpreted by gdb it was excreted into a file. This seems simple, but I'm unwilling to spend any more time puzzling out the interface a proprietary WRS GUI library. For those tuning in late, the last time I wanted to change the GUI a bit, our local WRS technical support person said we'd need to buy the T2 source. Now **that's** easy customization.... Thanks for any suggestions, Mike Beede - -- mbeede@ciprico.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxWorks development system Date: Mon, 20 Sep 1999 11:41:02 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37E6639E.BD0D2C6@ciprico.com> References: <7s068p$hrf$1@newso.ctimail.com> <37E5D814.DB7028A1@ntanet.net> David Hamel wrote: > > You could also try answering the question that > Simon asked instead of the one you thought he was asking. Wind River Systems. Try http://www.wrs.com . David, I presume you posted another followup that actually answered the question, but we didn't get it here yet. Mike - -- mbeede@ciprico.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 for NT machine: Where's cpp? Date: Tue, 21 Sep 1999 13:28:53 -0500 From: Mike Beede Organization: Ciprico Inc. Message-ID: <37E7CE65.24C8C7AB@ciprico.com> References: <37E6D4DB.6CF12650@vixel.com> Marjorie Krueger wrote: > > I'm running Tornado2 on NT 4.0 - Anybody know what .exe is the C preprocessor? You could use "ccwhatever -E" if you liked. Of course, this probably won't work if you're processing something other than a .c or .cpp or .CC or whatever. In that case, "ccwhatever -E -v" would show you what it was really being run. Regards, Mike Beede - -- mbeede@ciprico.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: inetstatShow Date: Thu, 23 Sep 1999 11:12:35 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <37E9EF03.7B129EE9@bitband.com> References: <37E94A11.6040BB51@lucent.com> Reply-To: leonid@bitband.com "JJ15C3600-B.PATEL(HOP092)NONE" wrote: > when we do inetstatShow, we get a listing of field - PCB ( protocol control > block). How does one know to which vxworks task this pcb corresponds to? > PCB is a data structure in memory, and as such does not belong to any specific task, only to a set of modules that manipulate this data. Specifically, TCP/UDP/IP routines are executed from ANY TASK which typically calls any of the socket functions to do network I/O. Although most input processing is done by tNetTask, iNetTask also does some output processing (such as TCP retransmissions). Most output code is executed by the caller of the socket call, up to the driver output function. One should be careful about calling socket function from high-priority tasks which do hard real-time processing, these function execution times are not stricktly deterministic. Hope this explains it. Leonid --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Thu Sep 23 05:23:49 1999 From: "Christopher J. Slominski" Date: Thu Sep 23 05:23:53 PDT 1999 Subject: Bus cycles --=====================_523485822==_.ALT Content-Type: text/plain; charset="us-ascii" VxWorks users, I'm working inder the assumtion that a processor locks the data bus while writing all native data types to memory. For example, a process writing a double to memory can not be interrupted by another process and get the memory bus before all 8 bytes are written. Therefore, in a single writer / multiple reader environment I only need to do operating system atomic mutual exclusion for complex data structures. Is this true for single and multiprocessor systems? *----------------------------------------* | Christopher J. Slominski | | Computer Sciences Corporation (CSC) | | 3217 N. Armistead Ave. | | Hampton VA 23666 | | (757) 766-8258 | | | | c.j.slominski@larc.nasa.gov | | NASA Mail Stop 931 | *----------------------------------------* --=====================_523485822==_.ALT Content-Type: text/html; charset="us-ascii"

VxWorks users,
     I'm working inder the assumtion that a processor locks the data bus while writing all native data types to memory.  For example, a process writing a double to memory can not be interrupted by another process and get the memory bus before all 8 bytes are written.  Therefore, in a single writer / multiple reader environment I only need to do operating system atomic mutual exclusion for complex data structures.  Is this true for single and multiprocessor systems?

*----------------------------------------*
|     Christopher J. Slominski           |
| Computer Sciences Corporation (CSC)    |
|      3217 N. Armistead Ave.            |
|         Hampton VA 23666               |
|          (757) 766-8258                |
|                                        |
|      c.j.slominski@larc.nasa.gov       |
|         NASA Mail Stop 931             |
*----------------------------------------*
 --=====================_523485822==_.ALT-- From owner-vxwexplo-process Thu Sep 23 06:23:11 1999 From: "Hamon, Kent" Date: Thu Sep 23 06:23:16 PDT 1999 Subject: RE: device driver for 85c30 This is probably not what you're looking for but I wanted to pass along the information: I recently ran across this link to the z85c30 chip: http://www.zilog.com/products/sccspecs.html The VxWorks MVME2604 BSP includes the following files: c:\Tornado\target\src\drv\sio\z8530Sio.c c:\Tornado\target\h\drv\sio\z8530Sio.h c:\Tornado\target\h\drv\serial\z8530.h and (AMD also makes a 85C30 UART) c:\Tornado\target\h\drv\serial\am85c30.h c:\Tornado\target\h\drv\serial\am85c30Poll.h Regards, Kent P.S. You might also check: http://www.devicedriver.com/ > -----Original Message----- > From: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] > Sent: Wednesday, September 22, 1999 3:47 PM > To: vxworks_users@csg.lbl.gov > Subject: device driver for 85c30 > > Submitted-by owner-vxwexplo-process Wed Sep 22 12:47:20 1999 > Submitted-by: Mauro Galarda Martins > > vxWorks > Tornado > > Hi, > > I'm looking for some help to develop a driver for the 85c30 to run in > HDLC synchronous mode. I'm working with powerpc architecture and > I would like to use DMA also. In fact I will use this drive with > PPP protocol, which is pretty similar to hdlc. Can anybody help me? > > thanks in advance. > > _______________________________________________________ > Mauro Jose Galarda Martins > Siemens Tel TR-DS Rua Pedro Gusso, 2635 CEP 81310-900 > Curitiba - Parana' - Brasil > Tel: +55 41 3416457 Fax: +55 41 3415620 > e-mail: mauro.martins@siemens.com.br > _______________________________________________________ > From owner-vxwexplo-process Thu Sep 23 06:56:53 1999 From: Alexander_R_Povolotsky@res.raytheon.com Date: Thu Sep 23 06:56:56 PDT 1999 Subject: More info - Re: VxWorks Not suited to High Availability??? vs Subject: More info - Re: VxWorks Not suited to High Availability??? vs LynxOS. From the LynxOS HighAvailability presentation slides, I have received from LynxOS New England regional office, I derived the following present and future ? LynxOS v. 3.01 HA and H(ot) S(wap) s/w key features (filtering out generic statements, etc.): 1) Device Resource Manager (DRM ): sybsystem and device drivers APIs . a) views devices as Dynamic b) understands complex topologies c) handles mapping 2) cPCI Hot Swap Services manual and programmatic API a) Dynamic cPCI and HSC topology knowledge b) Hot Swap Aware PMC Ethernet and SCSI subsystem drivers --------------------------------------------------------------------------- --------------------------------------------------------------------------- ------------ RE: Re: (VxWorks) Not suited to High Availability??? (vs LynxOS) vxwexplo@lbl.gov (the vxWorks Users Group Exploder) on 09/07/99 12:50:10 PM To: vxworks_users@csg.lbl.gov cc: (bcc: Alexander R Povolotsky/RES/Raytheon/US) Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) Submitted-by owner-vxwexplo-process Tue Sep 7 09:50:03 1999 Submitted-by: David Laight 'Hot Swapping' is an interesting idea, and, although possible to demonstrate,may be more difficult to achieve reliably in real life. I did some work with hot-swapping of PCI cards under UnixWare 7. It was possible, but in practise you could swap a working card for another working one. Swapping out a broken card is MUCH more difficult, particulararly if it had failed catastrophically while being accessed. The PCI bus is not resilient to device failures - eg a bus transciever dying. The added complexity of switching the bus signals, probably means that a system that support hot-swapping is actually more likely to fail than one which does not. A board with a non-critical failure (eg a line transceiver being blown) can almost certainly be replaced. The suggestion that a MSW-2000 system is doing hot-swap of cpu cards seems amazing! I can't imaging a cpu card dying without trashing at least some of the memory associated with its current activity. In something with a very restricted application (eg a comms switch) it might be possible to stash away the important parameters for existing connections so that an alternate cpu (or a rebooted cpu) could continue in a relatively seamless way, but for a more general stateful application it seems a little pie in the sky. DEMONSTRATING something is, as I said earlier an entirly different ball game... David Newsgroups: comp.os.vxworks Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) Date: Tue, 07 Sep 1999 14:43:25 -0400 From: Douglas Fraser Organization: Lucent Technologies I am confused by the fact that it says MSW-2000 and LynxOS in the same posting. Particularly since the computer in question uses PPC-750 CPUs, which are not supported in W2000 that I am aware of. I have to assume that there was a hierarchy here with some iBox (Ouch, Steve Jobs will squawk at that!) running W2K (why 2 K?) driving a few PPC-750 slots running LynxOS. > it is the fact that LynxOS does support Hot Swap for Compact PCI and that > VxWorks (yet) doesn't. > --------------------------------------------------------------------------- > LynxExpress Newsletter - Volume 15 (http://www.lynx.com/): > Motorola Computer Group stole the show at CT Expo with a dramatic > demonstration of the CPX8200 computer system. MCG's CPX8000 delivers > "five-nines" capability (99.999%; less than five minutes and 15 seconds of > downtime per year), and provides the computer power for applications used in > "carrier-grade" telecom systems. > The demonstration included a Microsoft Windows 2000-based full hot-swap > system is the first real-time system that does hot swapping and failover > of CompactPCI processor cards, done in combination with MCG's MCP750 > intelligent I/O processors and LynxOS real-time operating system from Lynx > Real-Time Systems. The CPX8000 is based on an open-architecture CompactPCI > platform, which enables any active module to be exchanged for repair or > upgrade while the system continues to operate. > ---------------------------------------------------------------- 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 Sep 23 07:13:15 1999 From: Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) Date: Thu Sep 23 07:13:19 PDT 1999 Subject: Booting from SCSI: How can i get the symbols? Hi VxWorks and Tornado Gurus! i got it to run my MVME2700 from local fixed disk. But the symbols couldn't been loaded: ... Attaching to scsi device... done. Loading /VXWORKS...1000180 Starting at 0x100000... Attaching network interface lo0... done. Loading symbol table from /VXWORKS.sym ...Error opening /VXWORKS.sym: status = 0x41 ... The shell-Prompt comes up, but as mentioned, NO SYMBOLs there to be used. The problem arose, when i put the files to the disk. ... Regler-CPU0-> copy "zfssun6:/tftpboot/vxw-5.3/mv2700.MatlabIPPC_Fast", "DISK1:/vxworks" value = 0 = 0x0 Regler-CPU0-> copy "zfssun6:/tftpboot/vxw-5.3/mv2700.MatlabIPPC_Fast.sym", "DISK1:/vxworks.sym" value = 0 = 0x0 Regler-CPU0-> cd "DISK1:" value = 0 = 0x0 Regler-CPU0-> ls VXWORKS VXWORKS.SYM value = 0 = 0x0 Regler-CPU0-> ... I can see a mismatch between the filename shown using the "ls()"-function and what is shown by the crashing boot-process above. The file-extension should not be in capital letters. That's what i uses as Boot-Parameters: ... [VxWorks Boot]: p boot device : scsi=1,0 processor number : 0 host name : zfssun6 file name : /VXWORKS inet on ethernet (e) : 129.69.88.200:ffffff00 host inet (h) : 129.69.88.241 user (u) : vxwboot flags (f) : 0x80 target name (tn) : zfsrc5 ... How can i solve this? Is the only way to change the dosVolOptions? Best regards F. Scheurer -- +-------------------------------------------------+ | Dipl.-Ing. F. Scheurer | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | | WWW http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | privat: Waldeckstrasse 31 | | D-73035 Goeppingen | | Telefon +49 7161 686362 Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | +-------------------------------------------------+ From owner-vxwexplo-process Thu Sep 23 09:02:47 1999 From: Alexander_R_Povolotsky@res.raytheon.com Date: Thu Sep 23 09:02:51 PDT 1999 Subject: More info - Re: VxWorks Not suited to High Availability??? vs Resending ...... VxWorks, TORNADO From the LynxOS HighAvailability presentation slides, I have received from LynxOS New England regional office, I derived the following present and future ? LynxOS v. 3.01 HA and H(ot) S(wap) s/w key features (filtering out generic statements, etc.): 1) Device Resource Manager (DRM ): sybsystem and device drivers APIs . a) views devices as Dynamic b) understands complex topologies c) handles mapping 2) cPCI Hot Swap Services manual and programmatic API a) Dynamic cPCI and HSC topology knowledge b) Hot Swap Aware PMC Ethernet and SCSI subsystem drivers --------------------------------------------------------------------------- --------------------------------------------------------------------------- ------------ RE: Re: (VxWorks) Not suited to High Availability??? (vs LynxOS) Submitted-by: David Laight 'Hot Swapping' is an interesting idea, and, although possible to demonstrate,may be more difficult to achieve reliably in real life. I did some work with hot-swapping of PCI cards under UnixWare 7. It was possible, but in practise you could swap a working card for another working one. Swapping out a broken card is MUCH more difficult, particulararly if it had failed catastrophically while being accessed. The PCI bus is not resilient to device failures - eg a bus transciever dying. The added complexity of switching the bus signals, probably means that a system that support hot-swapping is actually more likely to fail than one which does not. A board with a non-critical failure (eg a line transceiver being blown) can almost certainly be replaced. The suggestion that a MSW-2000 system is doing hot-swap of cpu cards seems amazing! I can't imaging a cpu card dying without trashing at least some of the memory associated with its current activity. In something with a very restricted application (eg a comms switch) it might be possible to stash away the important parameters for existing connections so that an alternate cpu (or a rebooted cpu) could continue in a relatively seamless way, but for a more general stateful application it seems a little pie in the sky. DEMONSTRATING something is, as I said earlier an entirly different ball game... David Subject: Re: (VxWorks) Not suited to High Availability??? [vs LynxOS) Date: Tue, 07 Sep 1999 14:43:25 -0400 From: Douglas Fraser Organization: Lucent Technologies I am confused by the fact that it says MSW-2000 and LynxOS in the same posting. Particularly since the computer in question uses PPC-750 CPUs, which are not supported in W2000 that I am aware of. I have to assume that there was a hierarchy here with some iBox (Ouch, Steve Jobs will squawk at that!) running W2K (why 2 K?) driving a few PPC-750 slots running LynxOS. > it is the fact that LynxOS does support Hot Swap for Compact PCI and that > VxWorks (yet) doesn't. > --------------------------------------------------------------------------- > LynxExpress Newsletter - Volume 15 (http://www.lynx.com/): > Motorola Computer Group stole the show at CT Expo with a dramatic > demonstration of the CPX8200 computer system. MCG's CPX8000 delivers > "five-nines" capability (99.999%; less than five minutes and 15 seconds of > downtime per year), and provides the computer power for applications used in > "carrier-grade" telecom systems. > The demonstration included a Microsoft Windows 2000-based full hot-swap > system is the first real-time system that does hot swapping and failover > of CompactPCI processor cards, done in combination with MCG's MCP750 > intelligent I/O processors and LynxOS real-time operating system from Lynx > Real-Time Systems. The CPX8000 is based on an open-architecture CompactPCI > platform, which enables any active module to be exchanged for repair or > upgrade while the system continues to operate. > ---------------------------------------------------------------- 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 Sep 23 09:08:18 1999 From: Simon Liston Date: Thu Sep 23 09:08:22 PDT 1999 Subject: RE: Header File Dependency Problems TORNADO Forget my previous posting regarding dependencies / header file locations! My thanks go to Johan Borkhuis for his helpful VxWorks page at http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html -----Original Message----- From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] Sent: 23 September 1999 11:00 To: vxworks_users@csg.lbl.gov Subject: Header File Dependency Problems Submitted-by owner-vxwexplo-process Thu Sep 23 02:59:32 1999 Submitted-by: Simon Liston TORNADO Hi, I've just started using Tornado 2.0 (we have the simulator package at the moment soon to be followed by StrongArm & Pentium full development kits.) and am having problems compiling a simple project. I get the following message when compiling a C file .... make: *** No rule to make target `foo.h', needed by `applic.o'. Stop. 'foo.h' is included in the applic.c file with just a '#include foo.h' i.e. no full path. The auto dependency generator correctly determined that foo.h is required by the C file BUT foo.h is not located in the same directory as the C file. Normally a -I and include pathname as part of the C compiler options is how the .h files can be located if not in the current directory. I've tried changing the source search path entries (under the debug menu) to include the directory where foo.h is but that doesn't seem to make any difference. Am I missing something really obvious here or have I bumped into a limitation of Tornado 2 (which would surprise me indeed!). Any help would be appreciated - I can't find anything in the manuals (so far). Best Regards, Simon. ********** From owner-vxwexplo-process Thu Sep 23 09:28:41 1999 From: David Laight Date: Thu Sep 23 09:28:45 PDT 1999 Subject: Re: Bus cycles > Subject: Bus cycles > Submitted-by: "Christopher J. Slominski" > > VxWorks users, > I'm working inder the assumtion that a processor locks the data bus while > writing all native data types to memory. For example, a process writing a > double to memory can not be interrupted by another process and get the memory > bus before all 8 bytes are written. Therefore, in a single writer / multiple > reader environment I only need to do operating system atomic mutual exclusion > for complex data structures. Is this true for single and multiprocessor > systems? Christopher, The processor does not 'lock' the bus during such cycles. However if a write will take a single bus cycle then it is atomic, although you cannot be sure exactly when it will take place. Writes of a 'double' (or a misaligned write when supported) are slightly more problematical. Provided there isn't a fault (eg page not present) on the second word, the cpu will complete the instruction (I'm open to a counter example here) before taking an interrupt, however the two write bus cycles might be separated. Locking is only needed to tie two operations together, so in code like: semTake( ... ); xx->state = 1; semGive( ... ); the semaphore stuff is redundant. On multi-cpu systems thing are much more complicated. The difficulty is that most modern cpus do not execute write cycles immediately. The data and address are put into a FIFO which is drained when the system bus is idle. Without special code a second cpu could read memory while the first still has a write buffered. (Reads by the cpu itself have to wait for the write to drain from the FIFO). (In some cpus it need not even be a FIFO - but with no ordering relation on the buffered data synchronisation is almost impossible - ask the ppc people). If you have a multi-word structure you must lock both readers and writers, otherwise a reader might read a partially updated copy, Consider a 64bit counter being incremented from 00000000ffffffff to 0000000100000000, if the reader doesn't acquire the lock it may get either 0000000000000000 or 00000001ffffffff - neither of which is correct. 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 Sep 23 15:01:45 1999 From: Mauro Galarda Martins Date: Thu Sep 23 15:01:49 PDT 1999 Subject: gcc for win32 vxworks tornado hi Does anybody has the gcc packet for win32 (target powerpc)? tanks Mauro -- _______________________________________________________ Mauro Jose Galarda Martins Siemens Tel TR-DS Rua Pedro Gusso, 2635 CEP 81310-900 Curitiba - Parana' - Brasil Tel: +55 41 3416457 Fax: +55 41 3415620 e-mail: mauro.martins@siemens.com.br _______________________________________________________ From owner-vxwexplo-process Thu Sep 23 23:28:37 1999 From: Jan Uerpmann Date: Thu Sep 23 23:28:41 PDT 1999 Subject: Re: Screenblanker & Windview Problems Hi Ted Dennison wrote: > > > > - After a while the programs run about 5 times slower than they > > used to. I recognized that if I press a key on the target the > > program speeds up again. I think it has something to do with the > > screenblanker as the target is slow when the blanker is active. > I have the exact same setup, and have no clue what "screenblanker" you > are talking about. My target screen *never* blanks out. Perhaps your PC > BIOS is doing this to you. If so, I'd suggest going into the BIOS and > disabling it. Thanks, this was the fault. With windview (now up and running) I detected that after a while (~5 min) most of the cpu time was spent in INT0 (Timer interrupt). After disabling all power saving stuff in the BIOS this was gone. A little bit strange though... cu, Jan Tornado, VxWorks From owner-vxwexplo-process Fri Sep 24 04:46:50 1999 From: Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) Date: Fri Sep 24 04:46:54 PDT 1999 Subject: Help! Changed boot from host to local disk. VxWorks, Tornado Hi VxWorkers, i want to run my MVME2700 standalone, w/o any host. I managed it to get the kernel image to the local SCSI-Disk. The system seems to start up right. But it doesn' t load the symbols that i put on the local disk, too. They're always looked for using the loopback-device. This would be nice, if the system would find or load the specified files. Just tried a lot of possibilities! How can i manage it, to get this of the preexisting dosFs from the SCSI-fixed disk? TIA F. Scheurer P.S.: Here's what i get at the boot-monitor/target-shell: ... 1. [VxWorks Boot]: @ boot device : scsi=1,0 processor number : 0 file name : /sd0/mv2700.MatlabIPPC_Fast inet on ethernet (e) : 129.69.88.200:ffffff00 user (u) : vxwboot flags (f) : 0x2 target name (tn) : zfsrc5 startup script (s) : /sd0/startup.rtw.tmpl Attaching to scsi device... done. Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 Starting at 0x100000... Host Name: bootHost Attaching network interface lo0... done. Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x3d ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] ]]]] (R) ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] ]]]]] ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] ]]]] ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] ]]]] ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] ]]]] ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] ]]]] ]]]] ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] ]]]] ]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System ]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., 1984-1997 CPU: Motorola MVME2700 - MPC 750. Processor #0. Memory Size: 0x2000000. BSP version 1.1/5. WDB: Ready. Unable to open startup script /sd0/startup.rtw.tmpl -> ... 2. [VxWorks Boot]: @ boot device : scsi=1,0 processor number : 0 file name : /mv2700.MatlabIPPC_Fast inet on ethernet (e) : 129.69.88.200:ffffff00 flags (f) : 0x2 target name (tn) : zfsrc5 startup script (s) : /startup.rtw.tmpl Attaching to scsi device... done. Loading /mv2700.MatlabIPPC_Fast...1000180 Starting at 0x100000... Host Name: bootHost User: target Attaching network interface lo0... done. Loading symbol table from /mv2700.MatlabIPPC_Fast.sym ...Error opening /mv2700.MatlabIPPC_Fast.sym: status = 0x3d ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] ]]]] (R) ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] ]]]]] ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] ]]]] ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] ]]]] ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] ]]]] ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] ]]]] ]]]] ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] ]]]] ]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System ]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., 1984-1997 CPU: Motorola MVME2700 - MPC 750. Processor #0. Memory Size: 0x2000000. BSP version 1.1/5. WDB: Ready. Unable to open startup script /startup.rtw.tmpl -> ... 3. [VxWorks Boot]: @ boot device : scsi=1,0 processor number : 0 host name : zfsrc5 file name : /sd0/mv2700.MatlabIPPC_Fast inet on ethernet (e) : 129.69.88.200:ffffff00 host inet (h) : 129.69.88.200 user (u) : vxwboot flags (f) : 0x2 target name (tn) : zfsrc5 startup script (s) : /sd0/startup.rtw.tmpl Attaching to scsi device... done. Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 Starting at 0x100000... Attaching network interface lo0... done. Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x41 ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] ]]]] (R) ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] ]]]]] ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] ]]]] ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] ]]]] ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] ]]]] ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] ]]]] ]]]] ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] ]]]] ]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System ]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., 1984-1997 CPU: Motorola MVME2700 - MPC 750. Processor #0. Memory Size: 0x2000000. BSP version 1.1/5. WDB: Ready. Unable to open startup script /sd0/startup.rtw.tmpl -> ... -- +-------------------------------------------------+ | Dipl.-Ing. F. Scheurer | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | | WWW http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | privat: Waldeckstrasse 31 | | D-73035 Goeppingen | | Telefon +49 7161 686362 Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | +-------------------------------------------------+ From owner-vxwexplo-process Fri Sep 24 06:04:29 1999 From: Jan Uerpmann Date: Fri Sep 24 06:04:33 PDT 1999 Subject: Windview Problems Hi If I include the following scheduler in my program Windview will crash somewhere if I call wvEvtEnable/Disable from within the program. If I call wvEvtEnable/ Disable from the shell windview works fine. Looks like a bug to me... int SDFScheduler() { SEM_ID *schedule[10]={ &Asem, &Csem, &Bsem, &Dsem, &Bsem, &Csem, &Dsem, &Dsem, &Bsem, &Esem}; int i, n=10; while(Run) { for (i=0; i Date: Fri Sep 24 07:11:33 PDT 1999 Subject: Formatting event for WindView 2.0.1 Hi vxWorks world, does anyone know how to format user-defined events for the Show Event dialog in WindView 2.0.1? The description provided in the User Manual is wrong (as stated in Release Notes), but nowhere is descibe how to do this. Thanks for an answer! 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 Fri Sep 24 11:29:53 1999 From: Alexander_R_Povolotsky@res.raytheon.com Date: Fri Sep 24 11:29:56 PDT 1999 Subject: Plugging into the PC's PCI extention card slot another Hi, This is not specifically VxWorks Tornado question (I have posted it also on Intel forum web page), but may be somebody could educate me on this subject here ... In the Intel PC architecture, what kind of on-board INTEL PCI controller chipset could allow to plug into PC's PCI extension slot another microprocessor (with the PCI bus) card with its own PCI controller (specifically Toshiba's TMPR3927 microprocessor is the one considered if that detail is important)? What OS could be (should be ?) run on such PC (Windows 95, 98 or NT)? I was told that some PCI controllers installed on some (?) PC's will be able to detect another external PCI controller and relegate PCI bus controlling functions to it ? Is it automatic (plug & play) or should it be specifically set up ? The Toshiba's TMPR3927 microprocessor is thought to run it's own OS (VxWorks) - any ideas how to boot it from that PC's local disk drive dedicated partition/file system ? What is the best way to exchange data between such PC (User GUI interface) running one of the Windows MS OS and the Toshiba's TMPR3927 microprocessor plugged into PC as described above and running VxWorks ? Is such configuration feasible (and does it make any sense ?) ? Alex Povolotsky From owner-vxwexplo-process Fri Sep 24 12:25:06 1999 From: Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) Date: Fri Sep 24 12:25:14 PDT 1999 Subject: Re: Allocation in Shell / SCSI-Standalone Boot / Symbols missing booting from SCSI This is a multi-part message in MIME format. --------------C7377B1DDB974330895F6617 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi Joe, Thank You for Your two replies! I just mailed to the admin of the VxWorks-Mail-Exploder because i couldn't see my postings. So now it's quite obvious that the posting arrived out there. The clarification You wanted: Hansen, Joe wrote: > Friedrich, > > First, I do not have an answer for this question of yours. Instead, I am > needing a bit of clarification from you as to what you are trying to do. I > am puzzled as to why you are trying to directly create and modify > DOS_VOL_CONFIG parameters for the SCSI drive. In my experience, when you > first create the physical device, the system will query the device to fill > in the blanks. When you create the block device, you can specify the > starting block, and the number of blocks for that device. But I have never > had to work directly with the DOS configuration parameters. You're right that it isn't necessary to do a dosFsConfigInit() (i don't knowwether it's the right function, because i'm at home now). After the dosFsMkfs() the structure is filled (default values plus the physical data of the SCSI-Block-Device), but in the beginning i wanted to manipulate it more than it is possible through dosFsMkfsOptions() or the other Options-function. And to do this i found the dosFsConfigInit() but this call needs the pointer of the DOS_VOL_CONFIG to fill the data in and i suppose that it overrides the pointer to the default structure. But i didn't find an API-Call to give me back such a pointer (and implicitly allocate the appropriate amount of memory). Is there any way to do something like pStructure = malloc(sizeof(STRUC_TYPE))? Since i found no API-call and got no answer to my posting i passed this point and did it using the default way. > I guess that I am ignorant as to why you would want to directly allocate > space for the DOS_VOL_CONFIG block. And since I like to follow problems > between SCSI devices and VxWorks, I would like a clarification from you. > Sorry I don't have an answer for you. > The overall thing i want to do is boot from SCSI without any server inthe local net. I'm not quite experienced with that kind of configuration because usually our applications run on targets that have ethernet connection to the servers, where our devellopment system runs and we have no extra fixed disk on the target (often rough environment). But now i have to send my system outside to test some equipment and for a future application i want local disk, too. This is why i want to realize the boot from SCSI. > Joseph R. Hansen > NAWCWPNS China Lake > Code 471500D > China Lake, CA 93555 > (760) 939-6999 > HansenJR@navair.navy.mil You first replied: > Friedrich, > > This is in response to your post to comp.os.vxworks concerning your symbol > table. Below is your boot config parameters: > > boot device : scsi=1,0 > processor number : 0 > host name : zfssun6 > file name : /VXWORKS > inet on ethernet (e) : 129.69.88.200:ffffff00 > host inet (h) : 129.69.88.241 > user (u) : vxwboot > flags (f) : 0x80 > target name (tn) : zfsrc5 > > The problem is your file name specification. This parameter not only > specifies your filename, it also specifies where the system needs to find > the file. Since your SCSI disk is "/disk1/", this needs to be specified. > Potentially, you could have multiple DOS file partitions connected to the > system. Try changing your file name parameter to "/disk1/vxworks". If this > doesn't work, please let me know. > By the way, the other respondent was correct about the case-insensitivity of > DOS. Also, if you intend to use the ethernet connection at all (since you > specified a host inet), don't forget to specify the ethernet device (ln, fn, > etc., depending on your BSP) in the "other" parameter. > (I apologize, but i only see the postings exploded from vxwexplo@lbl.gov. I don'tscan the newsgroup, because it is posted in packets through the digest.) > Hope this helps, > > Joseph R. Hansen > Also at this point i tried on and posted the attached message today a few hours earlier. What i did until now: I made a new dosFs using the dosFsMkFsOptions to get long file names and the ability to export the filesystem. The filename itself cannot be the problem anymore (case sensitivity). I also tried to ignore the host totally by deleting all corresponding boot-parameters. I tried also the path /sd0/. I got different status-return-values when the loading of the symbols was tried using interface lo0. For me the problem seems to be that the boot-image (which is obviously loaded) loads through another mechanism than the symbol table and the startup-script. That service doesn't seem to be provided by the booted target-kernel. Thank You in advance! F. Scheurer -- +-------------------------------------------------+ | Dipl.-Ing. F. Scheurer | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | | WWW http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | privat: Waldeckstrasse 31 | | D-73035 Goeppingen | | Telefon +49 7161 686362 Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | +-------------------------------------------------+ --------------C7377B1DDB974330895F6617 Content-Type: message/rfc822 Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-ID: <37EB643B.99B2F9D8@t-online.de> Date: Fri, 24 Sep 1999 13:44:59 +0200 From: Friedrich Scheurer / Goeppingen Reply-To: scheurer@zfs.tz.uni-stuttgart.de, Friedrich.Scheurer@t-online.de Organization: privat X-Mailer: Mozilla 4.05 [en] (Win95; I) MIME-Version: 1.0 To: VxWorks-Exploder CC: scheurer@zfs.tz.uni-stuttgart.de Subject: Help! Changed boot from host to local disk. Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit VxWorks, Tornado Hi VxWorkers, i want to run my MVME2700 standalone, w/o any host. I managed it to get the kernel image to the local SCSI-Disk. The system seems to start up right. But it doesn' t load the symbols that i put on the local disk, too. They're always looked for using the loopback-device. This would be nice, if the system would find or load the specified files. Just tried a lot of possibilities! How can i manage it, to get this of the preexisting dosFs from the SCSI-fixed disk? TIA F. Scheurer P.S.: Here's what i get at the boot-monitor/target-shell: ... 1. [VxWorks Boot]: @ boot device : scsi=1,0 processor number : 0 file name : /sd0/mv2700.MatlabIPPC_Fast inet on ethernet (e) : 129.69.88.200:ffffff00 user (u) : vxwboot flags (f) : 0x2 target name (tn) : zfsrc5 startup script (s) : /sd0/startup.rtw.tmpl Attaching to scsi device... done. Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 Starting at 0x100000... Host Name: bootHost Attaching network interface lo0... done. Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x3d ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] ]]]] (R) ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] ]]]]] ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] ]]]] ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] ]]]] ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] ]]]] ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] ]]]] ]]]] ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] ]]]] ]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System ]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., 1984-1997 CPU: Motorola MVME2700 - MPC 750. Processor #0. Memory Size: 0x2000000. BSP version 1.1/5. WDB: Ready. Unable to open startup script /sd0/startup.rtw.tmpl -> ... 2. [VxWorks Boot]: @ boot device : scsi=1,0 processor number : 0 file name : /mv2700.MatlabIPPC_Fast inet on ethernet (e) : 129.69.88.200:ffffff00 flags (f) : 0x2 target name (tn) : zfsrc5 startup script (s) : /startup.rtw.tmpl Attaching to scsi device... done. Loading /mv2700.MatlabIPPC_Fast...1000180 Starting at 0x100000... Host Name: bootHost User: target Attaching network interface lo0... done. Loading symbol table from /mv2700.MatlabIPPC_Fast.sym ...Error opening /mv2700.MatlabIPPC_Fast.sym: status = 0x3d ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] ]]]] (R) ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] ]]]]] ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] ]]]] ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] ]]]] ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] ]]]] ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] ]]]] ]]]] ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] ]]]] ]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System ]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., 1984-1997 CPU: Motorola MVME2700 - MPC 750. Processor #0. Memory Size: 0x2000000. BSP version 1.1/5. WDB: Ready. Unable to open startup script /startup.rtw.tmpl -> ... 3. [VxWorks Boot]: @ boot device : scsi=1,0 processor number : 0 host name : zfsrc5 file name : /sd0/mv2700.MatlabIPPC_Fast inet on ethernet (e) : 129.69.88.200:ffffff00 host inet (h) : 129.69.88.200 user (u) : vxwboot flags (f) : 0x2 target name (tn) : zfsrc5 startup script (s) : /sd0/startup.rtw.tmpl Attaching to scsi device... done. Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 Starting at 0x100000... Attaching network interface lo0... done. Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x41 ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] ]]]] (R) ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] ]]]]] ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] ]]]] ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] ]]]] ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] ]]]] ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] ]]]] ]]]] ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] ]]]] ]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System ]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., 1984-1997 CPU: Motorola MVME2700 - MPC 750. Processor #0. Memory Size: 0x2000000. BSP version 1.1/5. WDB: Ready. Unable to open startup script /sd0/startup.rtw.tmpl -> ... -- +-------------------------------------------------+ | Dipl.-Ing. F. Scheurer | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | | WWW http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | privat: Waldeckstrasse 31 | | D-73035 Goeppingen | | Telefon +49 7161 686362 Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | +-------------------------------------------------+ --------------C7377B1DDB974330895F6617-- From owner-vxwexplo-process Fri Sep 24 18:05:55 1999 From: "dave monachello" Date: Fri Sep 24 18:05:59 PDT 1999 Subject: Re: Clearcase/Tornado Anyone? In article <37E7AFF8.158858C7@email.mot.com>, qjb005@email.mot.com wrote: > Does anyone install Tornado and associated tools in a clearcase VOB? > Do you like the setup? Are there caveats? We tried that at a place where I used to work. In one way it was good because the version of the tools used to do the build can be directly associated with the derived objects (object and image files). The downside was that build were slower and caused more network activity. From owner-vxwexplo-process Sat Sep 25 04:00:17 1999 From: daemon@csg.lbl.gov Date: Sat Sep 25 04:00:22 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Sep 25 04:00:12 PDT 1999 Subject: Re: Header File Dependency Problems Subject: How to generate AIF-Header for a bootrom-image Subject: Dual booting Subject: Re: Header File Dependency Problems Subject: Using Zinc with VxWorks Subject: Support for MediaGX? Subject: Re: Using Zinc with VxWorks Subject: Re: Tornado II Subject: Linux <-> VxWorks Subject: Nothing to loose Subject: Re: App activation Subject: WindRiver support - here's a typical example Subject: Re: Bus cycles Subject: Re: using the Dynamic linker for run time apps Subject: ILS and LDAP? Subject: Linking object files from other os Subject: Re: Booting from SCSI: How can i get the symbols? Subject: Re: ILS and LDAP? Subject: Re: Cannot Increase SCSI2 Throughput Subject: Re: Tornado2 license? Subject: Re: Clearcase/Tornado Anyone? Subject: Re: WindRiver support - here's a typical example Subject: Re: gcc for win32 Subject: Re: I2C driver Subject: Re: Linux <-> VxWorks Subject: Re: Help, not able to make two VxSim on different host to talk! Subject: Re: Booting from SCSI: How can i get the symbols? Subject: Re: configuring network driver for vxWork image Subject: Re: Using Zinc with VxWorks Subject: Re: Dual booting Subject: Re: intConnect Subject: Re: Linux <-> VxWorks Subject: Re: Bus cycles Subject: Does TrueFFS come with Tornado 2 Subject: IO and Memory Anomaly Subject: Re: Dual booting Subject: Re: Industry Pack/mv172 and interrupt question... Subject: Re: Network I/O problems Subject: Signal Handler for Each New Task Subject: Real-Time CORBA Trade Study Subject: Re: Formatting event for WindView 2.0.1 Subject: Re: Bus cycles Subject: PLEASE READ: Header file dependencies solve *with* -I Subject: Re: BSP modifications and generating vxWorks image with Tornado II Subject: re: Screenblanker & Windview Problems Subject: serial port printing Subject: Industry Pack/mv172 and interrupt question... ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Header File Dependency Problems Date: Thu, 23 Sep 1999 13:00:28 +0200 From: "Werner Schiendl" Organization: VBS-Newsserver Message-ID: <938084523.136879@news.vbs.at> References: <7scuit$i166@overload.lbl.gov> Hi Simon, Simon Liston wrote in message news:7scuit$i166@overload.lbl.gov... - -- snip -- > > Normally a -I and include pathname as part of the C compiler options is > how the .h files can be located if not in the current directory. > > I've tried changing the source search path entries (under the debug > menu) to include the directory where foo.h is but that doesn't seem to > make any difference. > - -- snip -- To add an include directory with the Tornado II project facility: In the workspace window select the Builds tab Open the project tree to view your actual Build (I think it's named SIMNTgnu for the Simulator) Right click this entry, select Properties from the context menu In the properties window Select the C/C++ compiler tab Add the -I pathname to the existing parameters Close the properties dialog with OK This should solve your problem - -- Werner Schiendl International Applications - -- mailto:ws-news@gmx.at - -- B&R Industrial Automation http://www.br-automation.com --------------------------- Newsgroups: comp.os.vxworks Subject: How to generate AIF-Header for a bootrom-image Date: Thu, 23 Sep 1999 14:48:43 +0200 From: =?iso-8859-1?Q?R=FCdiger?= zum Beck Message-ID: <37EA21AA.A0B739A2@stud.fh-luebeck.de> Hi, I'm looking for some help to generate a bootrom with a AIF-Header for a Brutus Evulation Board. I would like to use the serial-port to boot my board. To do this I must reconfig some settings and build a new bootrom image and burn this to the flash rom's. My Problem is burning the roms, because the image must have a AIF-Header to burn it with the EBSA-285 Flash Management Utility (FMU) on the Brutus-board. In the config\brutus\ path I have found this source file aifHdr.s: DESCRIPTION of afiHdr.s This file can be assembled to generate a simple AIF header which can be prepended to a ROM image so that the will program it into flash ROM and the Primary Boot Loader (PBL) will run it. Many fields contains junk as the PBL and FMU check very little. How must I use this file in my project to gernerate a AIF Header? Any help would be appreciated - I can't find anything in the manuals. Best Regards, Rüdiger. --------------------------- Newsgroups: comp.os.vxworks Subject: Dual booting Date: Thu, 23 Sep 1999 08:42:51 -0700 From: Peter Pieda Organization: Enterprise Message-ID: <37EA4A7B.29937EBA@nortelnetworks.com> Hello, I'm trying to have a PC Dual boot DOS on one HD and VxWorks on the another HD. Any experiences with this? Thanks Peter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Header File Dependency Problems Date: Thu, 23 Sep 1999 08:56:08 -0400 From: moncreg Organization: Raytheon Company Message-ID: <37EA2368.59BC702@res.raytheon.com> References: <7scuit$i166@overload.lbl.gov> <938084523.136879@news.vbs.at> Werner Schiendl wrote: > Hi Simon, > > Simon Liston wrote in message > news:7scuit$i166@overload.lbl.gov... > > -- snip -- > > > > Normally a -I and include pathname as part of the C compiler options is > > how the .h files can be located if not in the current directory. > > > > I've tried changing the source search path entries (under the debug > > menu) to include the directory where foo.h is but that doesn't seem to > > make any difference. > > > -- snip -- > > To add an include directory with the Tornado II project facility: > > In the workspace window select the Builds tab > Open the project tree to view your actual Build (I think it's named SIMNTgnu > for the Simulator) > Right click this entry, select Properties from the context menu > > In the properties window > Select the C/C++ compiler tab > Add the -I pathname to the existing parameters > Close the properties dialog with OK > In my experience, sometimes [most times its the only way that works] it helps to delete and re-add files to the project AFTER you change the -I options and then rebuild the dependencies BTW, is there a knob that can be enabled to see what T2 thinks its doing to rebuild the dependencies? I would expect that it would eventually get down to some form of `mkdep ${FLAGS} ${SRCS}` or has that been munged into T2 itself?... --------------------------- Newsgroups: comp.os.vxworks Subject: Using Zinc with VxWorks Date: 23 Sep 1999 14:35:37 GMT From: paul.durrand@gecm.com (Paul Durrand) Organization: GEC-Marconi Avionics, Radar Systems Division Message-ID: <37ea2ca9@pull.gecm.com> Hi, We are considering using Zinc with VxWorks to develop and run a GUI application on a PowerPC. I was wondering if anyone has used Zinc and what are your opinions of the product? Thanks, Paul --------------------------- Newsgroups: comp.os.vxworks Subject: Support for MediaGX? Date: Thu, 23 Sep 1999 13:26:25 GMT From: Ed Oltman Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7sd9pi$4lo$1@nnrp1.deja.com> Has anybody used VxWorks with MediaGX based boards? Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using Zinc with VxWorks Date: Thu, 23 Sep 1999 07:53:04 -0600 From: "Russell A. Johnson" Message-ID: <37EA30C0.3407E220@bigfoot.com> References: <37ea2ca9@pull.gecm.com> I have used Zinc on dos and windows platforms about 2-3 years ago. It was just another GUI interface, nothing special about it. The drawback was how slow it was. If they haven't improved the speed of the software then it could take seconds to draw/redraw a screen. There were a few gotchas with the software but I can't remember specifics. Russ Paul Durrand wrote: > > Hi, > > We are considering using Zinc with VxWorks to develop and run > a GUI application on a PowerPC. > > I was wondering if anyone has used Zinc and what are your > opinions of the product? > > Thanks, > > Paul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II Date: Thu, 23 Sep 1999 11:45:07 -0400 From: Tim Sohacki Organization: Nortel Networks Message-ID: <37EA4B03.2CB59692@americasm01.nt.com> References: <199909230541.LAA23503@cmcltd.com> Viswanath Bandaru wrote: > > is it true that we donot have a Floating > licence verion of Tornado II ? The WRS Tornado II products do not use a license manager. The 3rd party products (Stethoscope, CodeTest, etc.) may, depending on the vendor. --------------------------- Newsgroups: comp.os.vxworks,comp.os.linux Subject: Linux <-> VxWorks Date: 23 Sep 1999 09:03:41 PDT From: Charles Eckel Organization: Vovida Networks, Inc Message-ID: <37EA5049.D3198F6@vovida.com> Anyone have any information/experience or know of the existence of anything to help/speed up porting from Linux to VxWorks and vice versa. Thanks in advance - -- Charles Eckel Senior Engineer mailto:ceckel@vovida.com Vovida Networks, Inc. phone:(408)971-1743 http://www.vovida.com fax:(408)941-1791 --------------------------- Newsgroups: comp.os.vxworks Subject: Nothing to loose Date: Thu, 23 Sep 1999 16:10:46 GMT From: mbpbmx@you.com Message-ID: THIS REALLY CAN MAKE YOU EASY MONEY!! IT WORKS!!! A little while back, I was browsing through newsgroups, just like you are now, and came across an article similar to this that said you could make thousands of dollars within weeks with only an initial investment of $6.00! So I thought," Yeah, right, this must be a scam", but like most of us, I was curious, so I kept reading. Anyway, it said that you send $1.00 to each of the 6 names and address stated in the article. You then place your own name and address in the bottom of the list at #6, and post the article in at least 200 newsgroups. (There are thousands) No catch, that was it. So after thinking it over, and talking to a few people first, I thought about trying it. I figured what have I got to lose except 6 stamps and $6.00, right? Like most of us I was a little skeptical and a little worried about the legal aspects of it all. So I checked it out with the U.S. Post Office (1-800-725-2161) and they confirmed that it is indeed legal! Then I invested the measly $6.00. Well GUESS WHAT!!... within 7 days, I started getting money in the mail! I was shocked! I figured it would end soon, but the money just kept coming in. In my first week, I made about $25.00. By the end of the second week I had made a total of over $1,000.00! In the third week I had over $10,000.00 and it's still growing. This is now my fourth week and I have made a total of just over $42,000.00 and it's still coming in rapidly. It's certainly worth $6.00, and 6 stamps, I have spent more than that on the lottery!! Let me tell you how this works and most importantly, why it works....also, make sure you print a copy of this article NOW, so you can get the information off of it as you need it. STEP 1: Get 6 separate pieces of paper and write the following on each piece of paper "PLEASE PUT ME ON YOUR MAILING LIST." Now get 6 US $1.00 bills and place ONE inside EACH of the 6 pieces of paper so the bill will not be seen through the envelope to prevent thievery. Next, place one paper in each of the 6 envelopes and seal them. You should now have 6 sealed envelopes, each with a piece of paper stating the above phrase, your name and address, and a $1.00 bill. What you are doing is creating a service by this. THIS IS ABSOLUTELY LEGAL! Mail the 6 envelopes to the following addresses: : #1) Matt Clusker : 63 Harmon Road : Glenville, NY 12302 : #2) Matt Johnson : 120 Quail Lane : Palatka, FL 32177 : #3)Jon Thomason : 11 Snowhill Meadows : Foley, MO 36617 : #4) Sam Kimmel : 625 Elm Dr. 340 Cole : Madison, WI 53706 : : #5) Jody Lehnherr : N6843 Pratt Rd. : Monticello, WI 53570 : #6) Joel Chabot 106 Clancy Drive Saskatoon. Saskatchewan, Canada S7M-4L2 : STEP 2: Now take the #1 name off the list that you see above, move the other names up (6 becomes 5, 5 becomes 4, etc...) and add YOUR Name as number 6 on the list. STEP 3: Change anything you need to, but try to keep this article as close to original as possible. Now, post your amended article to at least 200 newsgroups. (I think there is close to 24,000 groups) All you need is 200, but remember, the more you post, the more money you make! ---DIRECTIONS ----- HOW TO POST TO NEWSGROUPS------------ Step 1) You do not need to re-type this entire letter to do your own posting. Simply put your cursor at the beginning of this letter and drag your cursor to the bottom of this document, and select 'copy' from the edit menu. This will copy the entire letter into the computers memory. Step 2) Open a blank 'notepad' file and place your cursor at the top of the blank page. From the 'edit' menu select 'paste'. This will paste a copy of the letter into notepad so that you can add your name to the list. Step 3) Save your new notepad file as a .txt file. If you want to do your postings in different sittings, you'll always have this file to go back to. Step 4) Use Netscape or Internet explorer and try searching for various newsgroups (on-line forums, message boards, chat sites, discussions.) Step 5) Visit these message boards and post this article as a new message by highlighting the text of this letter and selecting paste from the edit menu. Fill in the Subject, this will be the header that everyone sees as they scroll through the list of postings in a particular group, click the post message button. You're done with your first one! Congratulations...THAT'S IT! All you have to do is jump to different newsgroupes and post away, after you get the hang of it, it will take about 30 seconds for each newsgroup! **REMEMBER, THE MORE NEWSGROUPS YOU POST IN, THE MORE MONEY YOU WILL MAKE!! BUT YOU HAVE TO POST A MINIMUM OF 200** That's it! You will begin reciving money from around the world within days! You may eventually wany to rent a P.O.Box due to the large amount of mail you will receive. If you wish to stay anonymous, you can invent a name to use, as long as the postman will deliver it. **JUST MAKE SURE ALL THE ADDRESSES ARE CORRECT.** Now the WHY part: Out of 200 postings, say I receive only 5 replies (a very low example). So then I made $5.00 with my name at #6 on the letter. Now, each of the 5 persons who just sent me $1.00 make the MINIMUM 200 postings, each with my name at #5 and only 5 persons respond to each of the original 5, that is another $25.00 for me, now those 25 each make 200 MINIMUM posts with my name at #4 and only 5 replies each, I will bring in an additional $125.00! Now, those 125 persons turn around and post the MINIMUM 200 with my name at #3 and only receive 5 replies each, I will make an additional $626.00! OK, now here is the fun part, each of those 625 persons post a MINIMUM 200 letters with my name at #2 and they each only receive 5 replies, that just made me $3,125.00!!! Those 3,125 persons will all deliver this message to 200 newsgroups with my name at #1 and if still 5 persons per 200 newsgroups react I will receive $15,625,00! With a original investment of only $6.00! AMAZING! When your name is no longer on the list, you just take the latest posting in the newsgroups, and send out another $6.00 to names on the list, putting your name at number 6 again. And start posting again. The thing to remember is, do you realize that thousands of people all over the world are joining the internet and reading these articles everyday, JUST LIKE YOU are now!! So can you afford $6.00 and see if it really works?? I think so... People have said, "what if the plan is played out and no one sends you the money? So what! What are the chances of that happening when there are tons of new honest users and new honest people who are joining the internet and newsgroups everday and are willing to give it a try? Estimates are at 20,000 to 50,000 new users, every day, with thousands of those joining the actual internet. Remember, play FAIRLY and HONESTLY and this will work. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: App activation Date: 23 Sep 1999 16:18:24 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <7sdjsg$bf8$2@news1.xs4all.nl> References: <7san4k$to9$1@news.flashnet.it> Santino Troiani wrote: : I'm working on a Pentium board with a TFFS managed storage. At the end of : the BSP startup I need to activate some middlware modules, so I put into : usrAppInit() the code to: : 1. load the module into memory, using ld() or loadModule() : 2. retry the module entry point symbol value using symFindByName() on the : VxWorks symbol table : 3. call the entry point : The step 2. always fails with error 0x1C0001 - SYMBOL_NOT_FOUND. Do you load the module using Tornado, or do you load the module directly from the flashdisk? In the first situation the symbol table is probably not synchronised. You should change this option in your target server, or load the module using the target shell (or use @ld instead of ld as a command on the host shell Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` --------------------------- Newsgroups: comp.os.vxworks Subject: WindRiver support - here's a typical example Date: Thu, 23 Sep 1999 18:27:43 GMT From: Marjorie Krueger Organization: Vixel Message-ID: <37EA70F4.B11CC2B5@vixel.com> Several people have asked about WindRiver's technical support. Here's a typical example. My original TSR stated: "In prjConfig.c, there is the call to usrSnmpdInit. I want to remove this call from prjConfig.c and move it into the init code for our project code. The problem is, if I ever change the VxWorks options (include another protocol, or remove one) prjConfig.c gets regenerated and wipes out my changes." The response was: > > Hello Marjorie, > > > > Regarding the TSR#142563, > > > > 1) You should not manually edit any files created by the project > > facility, e.g. prjConfig.c, Makfile..etc, since they are regenerated > > when there is any change in the project. > > > > 2) To remove the SNMP component from the project, you should > > exclude the SNMP component under the vxWorks view of the > > Workspace window. The SNMP component is under network components-> > > networking protocols->network applications->SNMP components. > > > > For further details about the project facility, please refer to > > the chapter 4 of Tornado User's Guide 2.0 (Window version). > > > > Regards, > > > > Eric Lo > > Technical Support Engineer > > Wind River Systems Duh! Great, they stated the obvious (and rather limiting) way Tornado 2 does things. And didn't even answer my question. Guess I should have told them that I already RTFM! Sheesh, what a waste. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bus cycles Date: Thu, 23 Sep 1999 13:26:42 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <37EA70E2.13B62EEC@aps.anl.gov> References: <4.1.19990923081643.00a2ada0@larc.nasa.gov> "Christopher J. Slominski" wrote: > > I'm working inder the assumtion that a processor locks the data bus while > writing all native data types to memory. For example, a process writing a > double to memory can not be interrupted by another process and get the memory > bus before all 8 bytes are written. Therefore, in a single writer / multiple > reader environment I only need to do operating system atomic mutual exclusion > for complex data structures. Is this true for single and multiprocessor > systems? This depends on the CPU and bus you're using, and may also depend on the compiler generating your code. David Leight's answer covers the concepts, I'm going to expand on that a little. If the update can be made in a single bus write cycle then you can assume atomic updates. For 6U VMEbus with D32 capability this means anything up to a 32-bit value that is 4-byte aligned should be Ok. A double is 8 bytes so will need 2 write cycles, thus between multiple processors this will not always be atomic. If you're working on a PCI bus a double might be atomic (I'm not certain how wide the data bus is). This assumes of course that your compiler doesn't do strange things and split the write up. There is also a technique you can use that doesn't require your readers to lock a structure to get an atomic read. This is useful if you can't afford to delay the writer at all but don't mind if the readers take a bit longer to get the data. It won't work if the writer is updating at a fairly high frequency compared to the read time though. The method is to have a pair of guard integers associated with the structure. When the writer wants to update the structure it starts by incrementing the first guard integer, changes the structure, then copies the first guard to the second. All readers take a local copy of the _second_ guard integer, then copy the structure, and finally copy the _first_ guard integer. If its two guards match then it knows that the structure copy it has was not altered while it was reading it. If the two guards are different it has to discard the data and try again. As long as the writer can never perform enough updates to overflow and wrap the guard integer values in the time that the slowest reader takes to copy the structure this technique is guaranteed, although frequent updates could cause readers to delay for long periods. You will need to take care of cache coherency and write-posting issues though (as with all multi-CPU code). - - Andrew - -- Complexity comes for free, Simplicity you have to work for. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: using the Dynamic linker for run time apps Date: Thu, 23 Sep 1999 15:00:13 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37EA78BD.B14FAC7B@lucent.com> References: Scott, While it is not trivial, and the linker (loadModule()) is not perfect as far as symbol resolution, we have used it extensively and it has been a very powerful tool. Order of linking is important, and the issues of symbol replacement (overlay?) and time of linkage is important. This has been discussed extensively on this site in the last two months, go back into the archives. We boot a large product from a FLASH IDE card. VxWorks gets loaded first, then each of the modules that defines a running system gets linked from there. Because each module is independently developed and IPC is via messages, we can replace individual modules during development without rebuilding the entire system. That saves a lot of time. It also means you can address features on a module by module basis with higher confidence that you can adequately test and debug during development without worrying about regression testing the whole system until you move into integration. If you are dealing with a large product, or your product can be decomposed into separate modules that communicate via messaging (or some other IPC) instead of via function/method calls, then it is well worth the effort to explore dynamic loading. We have not used the UNloader, so someone else will have to comment on that. Anyone??? One final word. Start small, with some example code, then work your way up. Once you are comfortable with the loader, it is not that hard to use. Doug s wrote: > > Is anyone using the Dynamic linker for run time apps? > > I am looking for feedback on how well this works. I have > heard third hand reports about this feature not working well > with large .o's or with complex down line loads (multple .o's > externally linked to for a down line loaded image). > > I know that many people use this feature during development > for module replacement. What I am trying to determine is > how well this feature works for extending a running system > with new features. > > Any feedback would be greatly appreciated. > > Thanks, > Scott --------------------------- Newsgroups: comp.os.vxworks Subject: ILS and LDAP? Date: 23 Sep 99 13:18:52 MDT From: sljwm@cc.usu.edu (JONATHAN DINERSTEIN) Organization: Utah State University Message-ID: Is anyone familiar with the acronyms: ILS and LDAP? I've heard about these, but I don't know anything about them. It's great doing web searches for acronyms because you gets results like "Gnomon Online: Latin Inscriptions." Anyway, thanks in advance for any info =). Jonathan Dinerstein sljwm@cc.usu.edu --------------------------- Newsgroups: comp.os.vxworks Subject: Linking object files from other os Date: Thu, 23 Sep 1999 13:11:55 -0700 From: Tiept Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <01f8d464.615a7403@usw-ex0102-013.remarq.com> I have a bunch of object files, compiled using GNAT ADA compiler. I haven't had any luck linking these object files into one object file to used on my main program which is in the vxworks/tornado environment and on a PPC platform. I am using the linking script below: ldppc -X -N -Map apple.map -i -o apple.o decoder.o ... and a bunch of other object files I am getting this error: decoder.o: file not recognized: File format not recognized Is there another switch in ldppc that I am supposed to use? Appreciate any helps on this. Thanks. TNT * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting from SCSI: How can i get the symbols? Date: 23 Sep 1999 20:15:42 GMT From: jcc@netcom.com (Jason C. Cotton) Organization: Netcom Message-ID: <7se1pe$egh@dfw-ixnews15.ix.netcom.com> References: <7sdckt$i174@overload.lbl.gov> Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) writes: >Hi VxWorks and Tornado Gurus! >i got it to run my MVME2700 from local fixed disk. >But the symbols couldn't been loaded: >... >Attaching to scsi device... done. >Loading /VXWORKS...1000180 >Starting at 0x100000... >Attaching network interface lo0... done. >Loading symbol table from /VXWORKS.sym ...Error opening /VXWORKS.sym: >status = 0x41 >... >The shell-Prompt comes up, but as mentioned, NO SYMBOLs there to be >used. >The problem arose, when i put the files to the disk. >... >Regler-CPU0-> copy "zfssun6:/tftpboot/vxw-5.3/mv2700.MatlabIPPC_Fast", >"DISK1:/vxworks" >value = 0 = 0x0 >Regler-CPU0-> copy >"zfssun6:/tftpboot/vxw-5.3/mv2700.MatlabIPPC_Fast.sym", >"DISK1:/vxworks.sym" >value = 0 = 0x0 >Regler-CPU0-> cd "DISK1:" >value = 0 = 0x0 >Regler-CPU0-> ls >VXWORKS >VXWORKS.SYM >value = 0 = 0x0 >Regler-CPU0-> >... >I can see a mismatch between the filename shown using the >"ls()"-function and what >is shown by the crashing boot-process above. The file-extension should >not be in >capital letters. DOS FS should be case-insensitive by default, so that may not be your actual problem. >That's what i uses as Boot-Parameters: >... >[VxWorks Boot]: p >boot device : scsi=1,0 >processor number : 0 >host name : zfssun6 >file name : /VXWORKS >inet on ethernet (e) : 129.69.88.200:ffffff00 >host inet (h) : 129.69.88.241 >user (u) : vxwboot >flags (f) : 0x80 >target name (tn) : zfsrc5 >... >How can i solve this? Is the only way to change the dosVolOptions? Why not copy vxWorks.st to your local disk, and boot that instead? I suspect that your problem loading the symbols off disk arises from VxWorks' default working directory not being set correctly. Loading vxWorks.st from disk will allow you to check that out, and also to test for case-sensitivity problems. I hope you figure it out. Please report back either way. - -Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ILS and LDAP? Date: Thu, 23 Sep 1999 17:30:37 -0400 From: moncreg Organization: Raytheon Company Message-ID: <37EA9BFD.1A2296F1@res.raytheon.com> References: "LDAP = Lightweight Directory Access Protocol alternative to X.500 LDAP is defined by RFC 1777 and RFC 1778." JONATHAN DINERSTEIN wrote: > Is anyone familiar with the acronyms: ILS and LDAP? I've heard > about these, but I don't know anything about them. It's great doing web > searches for acronyms because you gets results like "Gnomon Online: Latin > Inscriptions." > Anyway, thanks in advance for any info =). > > Jonathan Dinerstein > sljwm@cc.usu.edu --------------------------- Newsgroups: comp.os.vxworks,comp.periphs.scsi Subject: Re: Cannot Increase SCSI2 Throughput Date: Thu, 23 Sep 1999 14:43:33 -0700 From: "SCSI Cable Guy" Organization: http://www.cablemakers.com Message-ID: <7se6t1$f6b$1@nntp2.atl.mindspring.net> References: In my experience 7-8 MB/Sec. is good for Fast-10 mode. Your results seem to be similar to mine at http://www.cablemakers.com/scsiperformance.html . Get a Fast-20 controller! - -- http://www.cablemakers.com SCSI Wonderland Eric L. Schott wrote in message news:wog105lbkb.fsf@picard.arl.psu.edu... > I'm having trouble increasing throughput by adding a second drive to a > SCSI bus. What I observe is the overall bus throughput remains > basically unchanged while throughput per drive is cut by 50%. > > The configuration: > > - Motorola MVME177 (Has SCSI FAST/NARROW, using NCR710 chip) > - VxWorks Tornado II. Kernel is compiled with SCSI2 support. > - I have run tests with two pairs of drives. The first pair is two > IBM 9LP Ultrastar drives and the second is a pair of M-Systems > FFD-350 flash drives. > - I have written a small program which spawns one or two tasks (to > write to one or two drives). > > The results (I/O rates in bytes/second): > > - SCSI2 and hard drives (IBM) > num drives write read > 1 4306451 7164647 > 2 2558152 3647221 (each) > > - SCSI2 and FLASH drives > num drives write read > 1 1906501 2993704 > 2 954380 1525201 (each) > > - I was hoping to see two drive write times much greater than > either 2558152 (IBM drives) or 954380 (Flash drives). How can I > increase the throughput? > > The output of the VxWorks SCSI debug is: > > 0xef58a0 (tDrv0): scsi2Transact: > 0xfa7504 (tScsiTask): scsiMgrRequestProc: client request: 0 > 0xfa7504 (tScsiTask): scsiMgrActivateRequest: thread 0x00fa6044 > 0xfa7504 (tScsiTask): ncr710ThreadActivate: thread 0x00fa6044: activating > 0xfa7504 (tScsiTask): ncr710XferParamsCvt: requested: offset = 7, period = 25 > 0xfa7504 (tScsiTask): ncr710XferParamsCvt: granted: offset = 7, period = 25 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 1 -> 6 > 0xfa7504 (tScsiTask): ncr710Event: received event 21 (thread = 0x00fa6044) > 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa6044: received event 21 > 0xfa7504 (tScsiTask): DISCONNECT message in > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 6 -> 2 > 0xfa7504 (tScsiTask): ncr710Event: received event 22 (thread = 0x00000000) > 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa76e0: received event 22 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 0 -> 4 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 4 -> 4 > 0xfa7504 (tScsiTask): ncr710Event: received event 11 (thread = 0x00fa76e0) > 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa76e0: received event 11 > 0xfa7504 (tScsiTask): ncr710ThreadReconnect: reconnecting thread 0x00fa6044 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 2 -> 6 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 4 -> 0 > 0xfa7504 (tScsiTask): ncr710Event: received event 21 (thread = 0x00fa6044) > 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa6044: received event 21 > 0xfa7504 (tScsiTask): DISCONNECT message in > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 6 -> 2 > 0xfa7504 (tScsiTask): ncr710Event: received event 22 (thread = 0x00000000) > 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa76e0: received event 22 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 0 -> 4 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 4 -> 4 > 0xfa7504 (tScsiTask): ncr710Event: received event 11 (thread = 0x00fa76e0) > 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa76e0: received event 11 > 0xfa7504 (tScsiTask): ncr710ThreadReconnect: reconnecting thread 0x00fa6044 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 2 -> 6 > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 4 -> 0 > 0xfa7504 (tScsiTask): ncr710Event: received event 20 (thread = 0x00fa6044) > 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa6044: received event 20 > 0xfa7504 (tScsiTask): COMMAND COMPLETE message in > 0xfa7504 (tScsiTask): ncr710ThreadComplete: thread 0x00fa6044 completed > 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 6 -> 0 > 0xfa7504 (tScsiTask): scsiMgrThreadComplete: thread 0x00fa6044 > 0xfa7504 (tScsiTask): scsiMgrReplyProc: client reply: 0 > 0xfa7504 (tScsiTask): scsiMgrCompleteReply: thread 0x00fa6044 (status: 0, errno: 0) > 0xef58a0 (tDrv0): memPartFree: invalid block 0xe5fb78 in partition 0x9d898. > > My program is: > > - ------------------------------------------------------------ - -------------------- > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 license? Date: Thu, 23 Sep 1999 14:50:34 -0700 From: "David E. Chavez" Organization: CTS Network Services Message-ID: <7se7ba$2235$1@thoth.cts.com> References: <7s4hm5$p87$1@green.kreonet.re.kr> There is no license for Tornado II :-) moony wrote in message news:7s4hm5$p87$1@green.kreonet.re.kr... > hi.. > We have bought Tornado2 lately.. > But.. we didn't get license yet.. > Hmm... it is different from Tornado1.1.. > How can we get the license? > Please help us... > Could you explain the process in details? > Thank you.. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Clearcase/Tornado Anyone? Date: Thu, 23 Sep 1999 22:01:14 GMT From: Ted Dennison Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7se7v3$sd0$1@nnrp1.deja.com> References: <37E7AFF8.158858C7@email.mot.com> <7sb4vn$k4b$1@nnrp1.deja.com> <37E9E743.2D7DFCC4@hp.com> In article <37E9E743.2D7DFCC4@hp.com>, Holger Ditting wrote: > Ted Dennison wrote: > > > > We generally wouldn't do that with development tools, as executables > > can't be revised efficently and previous versions can always be > > retrieved from the CDROM they were loaded off of (Assuming proper > > inventory controls). > > [...] > > - Diskspace is cheap, so why not put executables in ClearCase Storage efficency is only one facet of a larger issue here. (sigh) There's probably a newsgroup out there for revision control specialists. This has absolutely nothing to do with vxWorks other than that hopefully most developers here use some kind of revision control. So I'm not going to bore everyone here with the thourough explanation of revision control theory (peppered with real-life war stories) this requires. I'll just say that you generally should not put files in ClearCase (or RCS or SCCS or CVS, or whatever revision control system you use) that you don't *need* to be in there. It was designed to provide revision control, not to provide a general-purpose filesystem. If you use a tool for something it was not designed for, you are asking for trouble. > - It is much easier to change the ConfigSpec than installing > everything again (and maybe overwrite the actual development > tools, reinstall them afterwards ...) Not on NT. What about the registry settings? Those would have to be manually saved into ClearCase and manually restored when retrieved, if you can find them all (you wouldn't want to save the *entire* resitry, because that will get unrelated setting that could corrupt a future setup of your machine). What about .DLL's that get put in the system directory? Are you going to boot the development PC's off of a ClearCase VOB too? I wouldn't suggest it. > - sometimes CDROMs disappear or are not accessible if the wrong > person is on vacation and keeps them locked away :-) I said *proper* inventory control. ;-) - -- T.E.D. Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindRiver support - here's a typical example Date: 23 Sep 1999 22:23:21 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <7se98p$fd7$1@news1.xs4all.nl> References: <37EA70F4.B11CC2B5@vixel.com> Marjorie Krueger wrote: : Several people have asked about WindRiver's technical support. Here's a typical example. My : original TSR stated: : "In prjConfig.c, there is the call to usrSnmpdInit. I want to remove this call from : prjConfig.c and move it into the init code for our project code. The problem is, if I ever change : the VxWorks options (include another protocol, or remove one) prjConfig.c gets regenerated and wipes : out my changes." Just to come back to your problem (and not reacting to the WRS support), it is possible to solve this problem. The configuration of prjConfig.c is defined in target\config\comps\vxworks. If you located the call to this function you can edit this (or remove it). I am answering this from home, so I cannot give you exact details about this. What you also can do is the following. Locate the part where this call is made. Copy this part, and define a second entry, but now without the call to the init function. Add the reference for your new module richt after the place where the original module was located. (If you need some help let me know, if you can send me the .cdf file where the call is defined I will take a look). Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: gcc for win32 Date: 23 Sep 1999 22:58:05 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <7seb9t$inl$1@news1.xs4all.nl> References: <7se8or$i176@overload.lbl.gov> Mauro Galarda Martins wrote: : vxworks : tornado : hi : Does anybody has the gcc packet for win32 (target powerpc)? Check out the cygnus cygwin package (www.cynus.com) Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I2C driver Date: Fri, 24 Sep 1999 00:31:34 GMT From: Curt McDowell Organization: Broadcom Corp. Message-ID: <37EAC779.7D605086@broadcom.com> References: <7sdjhp$c9j$1@nnrp1.deja.com> Reply-To: csm@broadcom.com I'm using the same processor (VERY nice CPU by the way) and am also looking for an I2C driver to avoid writing one. Please copy me on responses too. Thanks, Curt McDowell Broadcom Corp. csm@broadcom.com imohammed@mailexcite.com wrote: > > I am working on I2C driver for BSP for VxWorks. We are using > MPPC8240. It has built in I2C interface. I am having problem > understanding how to perform recieve as a Master or Slave. > > As a Master, an application would initiate the request to recieve data > from the slave. Once the address phase has completed, I will be ready > to recieve data. Upon recieving first byte of data, uP will generate > interrupt. My ISR will retrieve the data from data register. How do I > tell the application that I have data and how do I tell the application > where the data is located? > > thanks. > Irfan Mohammed > imohammed@mailexcite.com > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks,comp.os.linux Subject: Re: Linux <-> VxWorks Date: Fri, 24 Sep 1999 01:42:28 GMT From: John or Jenn Organization: Who, us organized??? Message-ID: <8GAG3.2710$UD4.15913@news.rdc1.az.home.com> References: <37EA5049.D3198F6@vovida.com> In article <37EA5049.D3198F6@vovida.com> Charles Eckel, ceckel@vovida.com writes: >Anyone have any information/experience or know of the existence of >anything to help/speed up porting from Linux to VxWorks and vice versa. There¹s an article on porting unix apps to VxWorks on the wrs web page (use the search feature)... Maybe that¹ll help? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help, not able to make two VxSim on different host to talk! Date: Fri, 24 Sep 1999 10:29:40 +0800 From: "Wang Jian" Organization: Alcatel Message-ID: <37eae401.0@dnews.sbell.com.cn> References: <7sb94v$nb7$1@nnrp1.deja.com> I think you didn't set metric value of route on NT to 2 or more. wrote in message news:7sb94v$nb7$1@nnrp1.deja.com... > Hi everyone, > > I have two VxSim sessions running on two different NT hosts. But I > can't make them to talk. > > Configuration follows: > NT1 - VxSim1: > ip forwarding enabled on the host > ulip installed as 90.0.0.254 > host ip: 114.119.77.21 > local ip: = host ip > target ip: 90.0.0.1 > routing for 90.0.0.2 to 114.119.77.22 added on NT > > NT2 - VxSim2: > ip forwarding enabled on the host > ulip installed as 90.0.0.254 > host ip: 114.119.77.22 > local ip: = host ip > target ip: 90.0.0.2 > routing for 90.0.0.1 to 114.119.77.21 added on NT > > > I can ping to VxSim1 and VxSim2 from both NT1 and NT2. However, I can > not ping from vxSim1 to VxSim2 or vice versa. And I can't event ping > NT1 from VxSim2 or ping NT2 from VxSim1. > > Within VxSim1: I can ping to 90.0.0.1, 90.0.0.254 and 114.119.77.21 but > not 114.119.77.22! > Within VxSim2: I can ping to 90.0.0.2, 90.0.0.254 and 114.119.77.22 but > not 114.119.77.21! > > I tried to add routing entry to the route table internal to VxSim. But > it doesn't help either! I tried both mRouteAdd > ("114.119.77.22","90.0.0.1",0xffffffff,0,0) and mRouteAdd > ("114.119.77.22","114.119.77.21",0xffffffff,0,0) to the VxSim1. No luck! > > I also tried to change the ip of the ulip to something other than > 90.0.0.254 (e.g. 80.0.0.254), the same problem exist. I can't even ping > to 80.0.0.254 from within the VxSim. > > > Any hint or info is greatly appreciated. > > > Ron > > > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting from SCSI: How can i get the symbols? Date: Fri, 24 Sep 1999 19:51:02 +0200 From: Friedrich Scheurer / Goeppingen Organization: privat Message-ID: <37EBBA05.43F2E4D5@t-online.de> References: <7sdckt$i174@overload.lbl.gov> <7se1pe$egh@dfw-ixnews15.ix.netcom.com> Reply-To: scheurer@zfs.tz.uni-stuttgart.de, Friedrich.Scheurer@t-online.de Jason C. Cotton wrote: > Friedrich.Scheurer@t-online.de (Friedrich Scheurer / Goeppingen) writes: > > >Hi VxWorks and Tornado Gurus! > > >i got it to run my MVME2700 from local fixed disk. > >But the symbols couldn't been loaded: > >... > >Attaching to scsi device... done. > >Loading /VXWORKS...1000180 > >Starting at 0x100000... > > >Attaching network interface lo0... done. > >Loading symbol table from /VXWORKS.sym ...Error opening /VXWORKS.sym: > >status = 0x41 > >... > >The shell-Prompt comes up, but as mentioned, NO SYMBOLs there to be > >used. > > >The problem arose, when i put the files to the disk. > > >... > >Regler-CPU0-> copy "zfssun6:/tftpboot/vxw-5.3/mv2700.MatlabIPPC_Fast", > >"DISK1:/vxworks" > >value = 0 = 0x0 > >Regler-CPU0-> copy > >"zfssun6:/tftpboot/vxw-5.3/mv2700.MatlabIPPC_Fast.sym", > >"DISK1:/vxworks.sym" > >value = 0 = 0x0 > >Regler-CPU0-> cd "DISK1:" > >value = 0 = 0x0 > >Regler-CPU0-> ls > >VXWORKS > >VXWORKS.SYM > >value = 0 = 0x0 > >Regler-CPU0-> > >... > > >I can see a mismatch between the filename shown using the > >"ls()"-function and what > >is shown by the crashing boot-process above. The file-extension should > >not be in > >capital letters. > > DOS FS should be case-insensitive by default, so that may not be your > actual problem. I tried it in other ways (Filesystem with long names), but i think you're right. > > > >That's what i uses as Boot-Parameters: > >... > >[VxWorks Boot]: p > > >boot device : scsi=1,0 > >processor number : 0 > >host name : zfssun6 > >file name : /VXWORKS > >inet on ethernet (e) : 129.69.88.200:ffffff00 > >host inet (h) : 129.69.88.241 > >user (u) : vxwboot > >flags (f) : 0x80 > >target name (tn) : zfsrc5 > >... > > >How can i solve this? Is the only way to change the dosVolOptions? > > Why not copy vxWorks.st to your local disk, and boot that instead? > What version of the file is it?(standalone) Please apologize my silly question. If i changed my kernel only two files were built vxworks and vxworks.sym, if i'm right. > I suspect that your problem loading the symbols off disk arises from > VxWorks' default working directory not being set correctly. Loading > vxWorks.st from disk will allow you to check that out, and also to > test for case-sensitivity problems. > As i mentioned previously to Joe Hansen, i think that the used kerneldoesn't offer the service needed to load it's own symbol table and startup script through the loopback device lo0. > I hope you figure it out. Please report back either way. > > -Jason Thanks and best regards! F. Scheurer - -- +-------------------------------------------------+ | Dipl.-Ing. F. Scheurer | | c/o Zentrum Fertigungstechnik Stuttgart | | Nobelstrasse 15 | | D-70569 Stuttgart | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | | WWW http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | privat: Waldeckstrasse 31 | | D-73035 Goeppingen | | Telefon +49 7161 686362 Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | +-------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Fri, 24 Sep 1999 10:03:24 -0700 From: Nidal Khalil Organization: SBC Internet Services Message-ID: <37EBAEDB.D5118C6F@pacbell.net> References: <37E4E4C5.1F32DC4@pacbell.net> <7s7blf$d6n@ipgw.marconi-instruments.co.uk> Reply-To: nidal@pacbell.net Yes I am having the same problem but I boot from a floppy with bootrom_uncmp then try to load a boot image from the hard disk. Therefore my boot paramaters do not have a network interface but the kernel image (vxWorks) has networking enabled in it. The target comes up with out the network interface attached. I could not do ifAddrSet or ifMaskSet both calls fail. It looks like networking is enabled only for network boot. This sucks I worked with tornado 1.0 and tornado 1.01 but this release of tornado is premature and should be a beta release not FCS. In addition WindRiver tech support is very poor. I put a call to WindRiver and opened a TSR but I have not received any solutions yet. There were few suggestion on the news group but I have not had the time to try them yet. If I solve this problem, I will let you know what it is:-) Nidal Wade Oram wrote: > Nidal Khalil wrote in message > news:37E4E4C5.1F32DC4@pacbell.net... > > > I am trying to build a vxWorks image with network driver in it. > > I am using tornadoII and vxWorks version 5.4. > > The network card that I am using is a standard intel ethernet pro 100 > > which is supported by vxWorks. If I boot vxWorks over the net the driver > > is configured properly but it I boot from a floppy or flash the network > > device is not attached to the tcp/ip stack. > > Hello, > > I have a very similar problem trying to configure a VxWorks build for a 486 > target with a PCMCIA Etherlink III network card. > I have build a boot rom which will successfully boot a vxworks image over > the network attaching the ELT driver (on pcmcia=1) in the process. > > However, if I then copy the VxWorks image onto a floppy disk and then boot > the same Vxworks image from the disk using the same boot rom (by > interrupting the auto-boot and setting the boot device to 'fd=0' and the > image file to /fd0/VxWorks') then although the VxWorks image loads, the > network driver is not installed. > > If I also change 'o=pcmcia=1,0' when I interrupt the autoboot, then the > VxWorks image is booted correctly and the network drivers are properly > installed. However, since my goal is to eventually boot a Rom based VxWorks > image directly from a small boot rom (BIOS) this does not help me since I > will not be running the VxWorks boot rom image. > > -- > Wade Oram --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Using Zinc with VxWorks Date: Fri, 24 Sep 1999 10:08:15 -0700 From: Nidal Khalil Organization: SBC Internet Services Message-ID: <37EBAFFE.7F1CD6CC@pacbell.net> References: <37ea2ca9@pull.gecm.com> <37EA30C0.3407E220@bigfoot.com> Reply-To: nidal@pacbell.net Hello, I am using Zinc with tornado II. Be perpared to write your own graphics driver. The ugl stack is as fast as you optimize your driver. I called the ugly stack. WindRiver internet suite package runs on top of zinc. Nidal "Russell A. Johnson" wrote: > I have used Zinc on dos and windows platforms about 2-3 years ago. It > was just another GUI interface, nothing special about it. The drawback > was how slow it was. If they haven't improved the speed of the software > then it could take seconds to draw/redraw a screen. There were a few > gotchas with the software but I can't remember specifics. > > Russ > > Paul Durrand wrote: > > > > Hi, > > > > We are considering using Zinc with VxWorks to develop and run > > a GUI application on a PowerPC. > > > > I was wondering if anyone has used Zinc and what are your > > opinions of the product? > > > > Thanks, > > > > Paul --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dual booting Date: Fri, 24 Sep 1999 10:27:09 -0700 From: Nidal Khalil Organization: SBC Internet Services Message-ID: <37EBB46C.463B2134@pacbell.net> References: <37EA4A7B.29937EBA@nortelnetworks.com> Reply-To: nidal@pacbell.net Hello, you can do it by building a boot strap from a floppy. First make sure that the fat on the hard disk is fat 16 not fat 32. In other word use dos 6.22 or earlier to format the disk edit config.h of the bsp you are working with and modify the boot path to /ata0/vxWorks if you are using an ide or ultra dma drive. Generate bootrom_uncmp. Make the vxWorks image and put it in the root directory on the hard disk Then boot from the floppy, this will load the vxWorks image on the hard disk. If you need to mount the file system for your application use usrAtaConfig( 0,0, "/mydisk" ); this will mount the hard disk root directory to /mydisk and you can access the file system, you need to make sure the ata drivers are in vxWorks configuration before you can mount it. You can do the same thing with a scsi drive The important points to remember that vxWorks support fat16 only and does not do long names translation unless you create the file system with vxWorks then you loose the capability to boot dos from the hard disk. Peter Pieda wrote: > Hello, > > I'm trying to have a PC Dual boot DOS on one HD and VxWorks on the > another HD. > > Any experiences with this? > > Thanks > > Peter --------------------------- Newsgroups: comp.os.vxworks Subject: Re: intConnect Date: Fri, 24 Sep 1999 20:24:48 GMT From: Charlie Grames Organization: Boeing Message-ID: <37EBDE10.C8C676E8@nowhere.com> References: <37E93C61.8CB9DF@navair.navy.mil> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Kevin, What you are doing with intConnect() should be fine. If you are getting a workQPanic, my guess is that you are getting interrupts on one or more of the other three vectors and that you are not resetting the interrupt source in your ISR. HTH Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com Kevin Havig wrote: > > I am using VxWorks 5.4/Tornado 2 on a Motorola MVME162 processor. I > have an intelligent serial card with four ports, each of which can > generate an interrupt; the vectors it is set up to use are 70, 71, 72, > and 73. The processing of each of these interrupts is so similar that I > would like to use a single ISR to process all four. I try to set this > up using four intConnect commands as follows: > > intConnect(INUM_TO_IVEC(70),portisr,0) > intConnect(INUM_TO_IVEC(71),portisr,1) > intConnect(INUM_TO_IVEC(72),portisr,2) > intConnect(INUM_TO_IVEC(73),portisr,3) > > The routine portisr is the routine to process the interrupt. It has a > case statement to process the parameter passed to it by the wrapper > built by intConnect. This parameter will have a value of 0, 1, 2 or 3 > as specified by the last parameter of intConnect. > > If I comment out all but the first intConnect port 0 interrupts work > just fine. Whenever I enable one of the other intConnects things go > wrong. I get the following message in my terminal window and my VME > system reboots: > workQPanic: Kernel work queue overflow. > > Is this a valid way to use intConnect? If so, any idea of what I am > doing wrong? > > Kevin Havig > Code 526110D, Echo Range > Naval Air Warfare Center > China Lake, CA > Havigkk@navair.navy.mil > (760) 939-9248 --------------------------- Newsgroups: comp.os.vxworks,comp.os.linux Subject: Re: Linux <-> VxWorks Date: 24 Sep 1999 14:01:26 PDT From: Charles Eckel Organization: Vovida Networks, Inc Message-ID: <37EBE794.296C60F6@vovida.com> References: <37EA5049.D3198F6@vovida.com> <8GAG3.2710$UD4.15913@news.rdc1.az.home.com> John or Jenn wrote: > > In article <37EA5049.D3198F6@vovida.com> Charles Eckel, ceckel@vovida.com > writes: > >Anyone have any information/experience or know of the existence of > >anything to help/speed up porting from Linux to VxWorks and vice versa. > > There¹s an article on porting unix apps to VxWorks on the wrs web page > (use the search feature)... Maybe that¹ll help? Thanks 'John or Jenn' I found a good overview article in "Wind River Systems - WindWord Winter 1998" titled "Rehosting code for UNIX on VxWorks." It provides a good overview and at least itemizes some of the sticky points. I'd still like something more detailed but this is a good starting point. - -- Charles Eckel Senior Engineer mailto:ceckel@vovida.com Vovida Networks, Inc. phone:(408)971-1743 http://www.vovida.com fax:(408)941-1791 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bus cycles Date: Fri, 24 Sep 1999 21:13:54 GMT From: Charlie Grames Organization: Boeing Message-ID: <37EBE992.F004A5C@nowhere.com> References: <4.1.19990923081643.00a2ada0@larc.nasa.gov> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Christopher, I'm sure it depends on the architecture, but the general answer is "no." The manner in which data transfers to and from memory are performed depends on alignment of the transfer, cache state and method, buffering performed by bridge devices, etc. Processors like PowerPC further require special instructions like "sync" to ensure memory coherency between multiple processors. You are best to implement operating system atomic mutual exclusion for all data types. Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com "Christopher J. Slominski" wrote: > > --=====================_523485822==_.ALT > Content-Type: text/plain; charset="us-ascii" > > VxWorks users, > I'm working inder the assumtion that a processor locks the data bus while > writing all native data types to memory. For example, a process writing a > double to memory can not be interrupted by another process and get the memory > bus before all 8 bytes are written. Therefore, in a single writer / multiple > reader environment I only need to do operating system atomic mutual exclusion > for complex data structures. Is this true for single and multiprocessor > systems? > > *----------------------------------------* > | Christopher J. Slominski | > | Computer Sciences Corporation (CSC) | > | 3217 N. Armistead Ave. | > | Hampton VA 23666 | > | (757) 766-8258 | > | | > | c.j.slominski@larc.nasa.gov | > | NASA Mail Stop 931 | > *----------------------------------------* > > --=====================_523485822==_.ALT > Content-Type: text/html; charset="us-ascii" > >
VxWorks users,
>      I'm working inder the assumtion that a processor > locks the data bus while writing all native data types to memory.  > For example, a process writing a double to memory can not be interrupted > by another process and get the memory bus before all 8 bytes are > written.  Therefore, in a single writer / multiple reader > environment I only need to do operating system atomic mutual exclusion > for complex data structures.  Is this true for single and > multiprocessor systems? >
> >
> *----------------------------------------*
> |     Christopher J. > Slominski           > |
> | Computer Sciences Corporation (CSC)    |
> |      3217 N. Armistead > Ave.            > |
> |         Hampton VA > 23666               > |
> |          (757) > 766-8258                > |
> |                                        > |
> |      > c.j.slominski@larc.nasa.gov       |
> |         NASA Mail Stop > 931             > |
> *----------------------------------------*
> > > --=====================_523485822==_.ALT-- --------------------------- Newsgroups: comp.os.vxworks Subject: Does TrueFFS come with Tornado 2 Date: Fri, 24 Sep 1999 08:25:52 +0200 From: "Kremer, Harro" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <7rs7di$bs0$1@nnrp1.deja.com> Goed question. I asked my WRS sale rep about it yesterday. Harro > ---------- > From: js_j98@my-deja.com[SMTP:js_j98@my-deja.com] > Posted At: vrijdag 17 september 1999 4:01 > Posted To: vxworks > Conversation: Does TrueFFS come with Tornado 2 > Subject: Does TrueFFS come with Tornado 2 > > We purchased the standard Tornado 2 package. I noticed it includes > TrueFFS lib (can be included, compiled, not tried). Windriver also > sells TrueFFS for $4000 or so. What is the difference between the > ones > come with Tornado2 standard package and the TrueFFS they sell > separately? > > Thanks, > Jun > > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't. > --------------------------- Newsgroups: comp.os.vxworks Subject: IO and Memory Anomaly Date: Fri, 24 Sep 1999 10:42:03 +0100 From: "Kelvin Lawson" Message-ID: <37eb472f.0@nnrp1.news.uk.psi.net> Hi All, I recently had some problems performing IO with my PCI hardware on an MBX860 BSP. I was using sysInWord etc. to access the IO space, and having no joy, until someone on the list kindly pointed out that by using mem refs (ULONG *MemPtr etc) I could actually get access to this space. The card's registers should also be mappable into memspace as normal aswell, but, when mapped, the mem region is failing Vxmemprobes, I'm assuming because there's nothing there. I have written different addresses to the PCI config space for the iobase and membase, however the only one I can access is the iobase through memory-type refs! This seems odd to me. Does anyone have any suggestions as to why I can't map the mem space ? Note, I have mapped both areas into sysPhysMemDesc (in fact a huge amount of PCI area is mapped as standard in my BSP). On another note, I am adapting the WRS-supplied driver for the Intel Pro100B Ethernet NIC. The reason I have to adapt it is because it is incompatible with my BSP (the MBX860). For one thing, I have to use an offset for CPU and PCI-side memory references. Now whilst I understand WRS can't be expected to develop for every single BSP, I am surprised that they can't gave something like a #define PCI_CPU_TRANSLATION, with a value of 0 for x86 and 0x80000000 for the 860 or something. That aside, I am wondering if there are other things I have to look out for when porting source to this BSP, particularly relating to PCI and endianness issues. Does anyone know any caveats ? Any suggestions are well appreciated, Cheers, Kelvin. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dual booting Date: Fri, 24 Sep 1999 10:06:08 GMT From: dixon@crd.ge.com Organization: Deja.com - Before you buy. Message-ID: <7sfieb$pe8$1@nnrp1.deja.com> References: <37EA4A7B.29937EBA@nortelnetworks.com> In article <37EA4A7B.29937EBA@nortelnetworks.com>, Peter Pieda wrote: > Hello, > > I'm trying to have a PC Dual boot DOS on one HD and VxWorks on the > another HD. > > Any experiences with this? > > Thanks > > Peter > > Use a boot manager. I've had good experience with System Commander. standard disclaimers apply Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Industry Pack/mv172 and interrupt question... Date: Fri, 24 Sep 1999 15:07:35 -0700 From: Ken Norlin Organization: NASA Dryden FRC Message-ID: <37EBF627.9B929388@dfrc.nasa.gov> References: M Lang wrote: > > > Now when I started the IP and after it generated an interrupt, > the mv172 just hung. > > I would appreicate any comments/info/help/feedback. Thank you > very much. > > = Minh = Hi, Don't know if this is your problem or not, but I've had RORA type interrupters that would hang the bus (because I wasn't clearing the interrupt condition correctly in the interrupt handler). If you had a VME bus analyzer, you would have seen this already though. Ken N. - -- Ken Norlin --- Simulation Engineer NASA Dryden Flight Research Center ken.norlin@dfrc.nasa.gov (661) 258-2046 FAX (661) 258-2792 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Network I/O problems Date: 24 Sep 1999 22:50:05 GMT From: toni@devsoft.devsoft.com (Anton Hartl) Organization: Customer of SpaceNet GmbH Message-ID: References: <242-37EBE152-67@newsd-222.iap.bryant.webtv.net> In article <242-37EBE152-67@newsd-222.iap.bryant.webtv.net>, Greg Silverman wrote: >Our system consists of 3 Motorola processors running VxWorks 5.3.1, a >cluster of 10 Mercury processors, and a Solaris 2.5.1 workstation. At >bootup we do the following steps: 1) download the VxWorks images, 2) >configure the Mercury cluster from the master Motorola's bootup script, >and, then, 3) start a program on the master Motorola that downloads and >starts all of the application programs. Too often the system fails >during the last step, with an access violation or a file I/O error in >one of the Vxworks tasks. To test the system software I replaced the >last step of the bootup procedure with a VME reset, so that the system >was in a "boot loop", e.g., no application code is run. If there are no >problems with the VxWork/Solaris/Mercury installations and interfaces >this boot loop should go on forever. But, every 20 or so boots, one of >the VxWorks cards is unable to load some files from the Sun. I have >seen: 1) it can not load the VxWorks image, 2) it could not read the >VxWorks startup script, 3) the startup script could not load one of the >code files. This seems to fail way too often and indicates some kind of >network I/O problem that may explain our problem with bringing up the >applications. Any ideas? What method do you use to download your images and applications? If it is FTP you might run into the problem that too many FTP requests hit the Solaris in a too short time interval which makes inetd believe there is something wrong with FTP; inetd will temporarily disable FTP then. That happens in particular when you attempt to load a bunch of small files from the startup script. You should find a line like "services respawning too fast" in /var/adm/messages; also check out the inetd man page and look for the -r option. - -Toni - -- /dev Software Gmbh | Anton Hartl | Dumbo: You're an unusual cop. Gabelsbergerstr. 51 | Phone: +49 (0)89-28659843 | Mickey Mouse: You're an unusal 80333 Muenchen | Fax: +49 (0)89-28659845 | killer, too. Germany | Email: toni@devsoft.com | John Woo, The Killer --------------------------- Newsgroups: comp.os.vxworks Subject: Signal Handler for Each New Task Date: Tue, 21 Sep 1999 21:36:06 -0500 From: Glenn Hofmeister Organization: U S WEST Interprise Message-ID: <37E84096.321E4111@uswest.net> Greetings, I am using vxSim 5.3.1 and what I am attempting to do is set up a signal handler to process a segmentation violation in a task. However, what I am really trying to do is for each new task that starts, this signal handler is to be set up. So I have been trying to use taskCreateHookAdd() to set up a function that is called when a new task starts. However, when the function gets called, the task is not the new task, but it does have a pointer to the new task's WIND_TCB. I do not know what fields I could update in the new task's WIND_TCB to set up a signal handler for the new task. Has anyone ever tried something like this? My signal handler is: LOCAL jmp_buf env; void sigHandler ( int sig, int code, SIGCONTEXT *sigContext ) { switch (sig) { case SIGSEGV: printErr ("\nSignal SIGSEGV received\n"); break; } longjmp (env, sig); } My added Hook routine: void myCreateHook (pTcb) WIND_TCB *pTcb; { } My code to set up the hook: taskCreateHookAdd((FUNCPTR) myCreateHook); Thanks, Glenn glennh@uswest.net --------------------------- Newsgroups: comp.os.vxworks Subject: Real-Time CORBA Trade Study Date: Wed, 22 Sep 1999 02:52:37 GMT From: bill.beckwith@realtime-corba.com (Bill Beckwith) Organization: Objective Interface Message-ID: <8E48E8370beckwbnewscom@sigma.ois.com> An advanced research and development unit of The Boeing Company, Phantom Works, conducted a Real-Time CORBA Trade Study that includes a real-time ORB benchmark. The study was conducted for the DII COE Real-Time Integrated Product Team. For details on the preliminary findings of this study see: http://www.ois.com/technical/rt_diicoe_ipt.html --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Formatting event for WindView 2.0.1 Date: Fri, 24 Sep 1999 16:55:30 -0700 From: "Dave Craig" Organization: Verio Message-ID: References: <7sgijc$12s1@overload.lbl.gov> Using Windview 2.0.1 with Tornado 1.0.1 the user events don't work. It's apparently mostly fixed in Tornado II as long as you don't use user ids 0 and 1... What you see in Tornado 1.0.1 is that the user events are incorrectly identified as user%d or user and not user0, user1 etc. Dave Craig Palo Alto, CA Arand Thomas wrote in message news:7sgijc$12s1@overload.lbl.gov... > Hi vxWorks world, > > does anyone know how to format user-defined events for the Show Event dialog > in WindView 2.0.1? > > The description provided in the User Manual is wrong (as stated in Release > Notes), but nowhere is descibe how to do this. > > Thanks for an answer! > > Ciao, > Thomas > > ---------------------------------------------------------------------- > Thomas Arand Siemens AG > Tel. +49 - 89 - 722 42296 > Fax. +49 - 89 - 722 26572 > thomas.arand@icn.siemens.de > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bus cycles Date: Fri, 24 Sep 1999 18:56:19 -0700 From: Harvey Taylor Organization: Organization? What organization?! Message-ID: <37EC2BC3.78DC@despam.pangea.ca> References: <4.1.19990923081643.00a2ada0@larc.nasa.gov> Christopher J. Slominski wrote: > VxWorks users, > I'm working inder the assumtion that a processor locks the data bus while > writing all native data types to memory. For example, a process writing a > double to memory can not be interrupted by another process and get the memory > bus before all 8 bytes are written. Therefore, in a single writer / multiple > reader environment I only need to do operating system atomic mutual exclusion > for complex data structures. Is this true for single and multiprocessor > systems? > We ran into an interesting case which might be relevant. On a cPCI system with a card across a couple of bridges from the CPU, the last thing an interrupt handler was doing was write to the card's PCI controller to clear the interrupt, but because the backplane was so loaded, the write got posted by a bridge. The effect of this was for the handler to return from the handler, while the interrupt line was still asserted, causing yet another entry to the handler, at which time the posted write was actually completed and the interrupt disappeared. The system didn't like that. - -het - -- "you don't really appreciate how smart a moron is until you try to design a robot..." -a robotics designer Harvey Taylor Internet: het@despam.portal.ca --------------------------- Newsgroups: comp.os.vxworks Subject: PLEASE READ: Header file dependencies solve *with* -I Date: Fri, 24 Sep 1999 14:21:22 +0200 From: "Geurt Vos" Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <003c01bf0687$50264460$70cb09c0@WST012.rohill.nl> Hello, I tried to reply to a message posted by Simon Liston about the header file dependencies, but I think it failed. Anyways, when header files are located in different directories, dependencies seem to fail when using the -I flag, while when using the VPATH macro, it'll succeed (as found in Johan Borkhuis' FAQ). Why I'm posting: Some 2 weeks ago I found out why the -I flag fails. At least when your host is a Windows machine. You can still use the -I flag, *but* you have to use SLASH ('/') instead of BACKSLASH ('\') in the path name! So, instead of something like: -IC:\projects\ThisPrj1\Include2 you have to use: -IC:/projects/ThisPrj1/Include2 The weird part is that '\' sometimes does work, or even for some files in the project... So for projects with header files in several directories, you can use both -I and VPATH. This -I thing is yet another silly bug in Tornado 2. Though WRS might just call it a 'feature enhancement'. - -Geurt- ~warnings are just like errors, only not~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Geurt Vos Junior Project Engineer Rohill Engineering Hoogeveen, The Netherlands ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Extended Operating System Loader http://www.xosl.org geurt@xosl.org ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP modifications and generating vxWorks image with Tornado II Date: 24 Sep 1999 12:57:54 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <7sfsgi$1sl$1@news1.xs4all.nl> References: <7rb08g$3ff5@overload.lbl.gov> <01bf050d$835a2520$3c1e140a@gerard> GKE wrote: : Hi Reyhan, : Tornado 2 ignores some dependencies to limit the number of files in your : file list. : You can select which directories are ignored by selecting Build - : Dependencies...- Advanced... from the menu. : Under 'Read-only dependency path' you can select which directories are : ignored. : I removed the last one, named 'E:\TORNADO2\target\config' (on my machine), : and re-scanned the dependencies. One disadvantage of this is that you get all the "standard" files also in your dependency list. To avoid this add the directory ....\target\config\comps to the list. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: re: Screenblanker & Windview Problems Date: Fri, 24 Sep 1999 14:50:27 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <7sg33c$5m8$1@nnrp1.deja.com> References: <7sf8db$i177@overload.lbl.gov> In article <7sf8db$i177@overload.lbl.gov>, Jan Uerpmann wrote: > After disabling all power saving stuff in the BIOS this was gone. A Heh. Welcome to the wonderful world of EnergyStar PC's. Now if I can just find a good screensaver for vxWorks... ;-) - -- T.E.D. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: serial port printing Date: Tue, 21 Sep 1999 20:06:33 GMT From: Denis Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7s8ofv$svt$1@nnrp1.deja.com> Hello, Can anyone please tell me if it is possible to read the status information about a printer ( say, dot-matrix ) from the serial port, and if it is possible maybe a brief "How" or a pointer to something usefull to do that. The info I would need to read is something like "out of paper", or "offile", or anything else that i can get out of the serial port. thanks. denis Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Industry Pack/mv172 and interrupt question... Date: Fri, 24 Sep 1999 09:44:59 -0800 From: mlang@fst.jpl.nasa.gov (M Lang) Organization: JPL Message-ID: Hi all, We just got a Motorola mv172 (68060) and some IP boards. I'm not new to VME and vxWorks, however, this is the 1st time I get my hands on IP. I would appreciate any comments/help: I have no trouble mapping the SBS Greenspring's IP-Unidig-T-D board in vxWorks/VME space and I was able to access its I/O, memory spaces properly thru the IP2 chip. Everything or pretty much most of everything worked fine without using interrupt. However, when I tried to have the IP interrupted the CPU, everything just hung. I'm sure I missed something that I don't know. This is what I did for the interrupt part: - - As this IP is in slot a (IP a), and it supposedly tie to intReq0 (per documentation), I programmed the following: 1) write 0x14 to IP2's IRQ0 Interrupt Control Register at 0xFFFBC010 to : - enable IEN (Interrupt Enable bit) - set int level to 4 (randomly choosen) 2) set up proper register on IP board to generate interrupt when its counter reached 0 (this is a timer board) and have it use interrupt vector 0x12 (randomly choosen) 3) connect my isr to vector 0x12 (just "logMsg" a msg) with intConnect. Now when I started the IP and after it generated an interrupt, the mv172 just hung. I would appreicate any comments/info/help/feedback. Thank you very much. = Minh = --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Sat Sep 25 10:20:59 1999 From: dlee@adti.com Date: Sat Sep 25 10:21:03 PDT 1999 Subject: memory leaks I am using a version of vxWorks 5.3 that has been configured as 5.2. Therefore I do not have Tornado front end. My problem is a memory leak that is rather large. Can anyone tell me of a method using vxWorks commands to try to get a handle on the offending code, or of shareware diagnostics tools easily available. Thanks From owner-vxwexplo-process Sun Sep 26 04:00:10 1999 From: daemon@csg.lbl.gov Date: Sun Sep 26 04:00:14 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Sep 26 04:00:06 PDT 1999 Subject: Re: C++ constructor undefined symbol? Subject: Re: Linking object files from other os Subject: Re: About Global variables and main() functions Subject: Tornado 2 make Subject: Re: configuring network driver for vxWork image ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ constructor undefined symbol? Date: Sat, 25 Sep 1999 10:46:05 -0700 From: engineer_scotty Organization: Aracnet Internet Message-ID: <37ED0A5D.6F42681C@aracnet.com> References: <7scr2g$i172@overload.lbl.gov> Mike Davies wrote: > > Hi, > > if you run the c++filt program (for a Windows hosted 386 target this is > "cppfilt386 _vt$8CVCQCard" ) from the command line then it will tell > you what the function prototype looks like. In this case the output is : > > CVCQCard virtual table > > It looks like for some reason the compiler can't find the v_table for > that class. Are you sure you define an instance of the class somewhere > other than a header file ? > > Mike Davies > > This is my opinion, not my employers'. > > > > >Hello, > > > >I get a Tornado 1 debugger error: > > > >"object module contains undefined symbols: _vt$8CVCQCard" > > > >when loading my *.out file. > > > >The name of my constructor is "CVCQCard" and I've double checked the > >spelling > >and added the base class constructor also. (No debugger complaints about > >the > >base class constructor). > > > >Does anyone know what I might be missing? > > > >Thanks, > >Kent > > > >vxWorks Tornado 1) Is the base class destructor the ONLY virtual function in the base class (or are all the other virtual functions declared inline, or declared in the body of the class))? 2) If so, do you define the base class destructor inline? ie class foo { // other stuff virtual ~voo(void) {} }; If both 1 and 2 are true; then you aren't getting a vtable. When you declare a virtual class, the compiler at some point has to synthesize a vtable. The linker will NOT do this. However, it is common to have different "translation units" (.cc files) reference the same class; if ALL such .cc files declared the vtable, that would a) cause lots of extra work, b) cause duplicate symbol errors if older linkers (that don't support "weak" symbols) are used. For this reason, it is desirable to have exactly ONE generated object file contain the vtable. In order to achieve that result, the following hueristic is used: The vtable is synthesized in the translation unit that provides an implementation for the FIRST virtual, non-inline method found in the class definition (including the destructor, if it is listed first). This works pretty well--except when all the virtual functions are inlined; then the vtable is never generated. The way to deal with it is as follows: ALWAYS provide a non-inline implementation for a virtual destructor, even if is trival (ie Foo::~foo(void) {} ). This will always cause a vtable to be generated. (The destructor need not be listed in any particular order in the class). The destructor is a good choice because it always exists; if you have any virtual functions you should always make the destructor virtual. (Feel free to choose another method to use if making the destructor non-inline causes performance problems; in most cases, this should not be the case.) - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking object files from other os Date: Sat, 25 Sep 1999 10:53:13 -0700 From: engineer_scotty Organization: Aracnet Internet Message-ID: <37ED0C09.B6FB392C@aracnet.com> References: <01f8d464.615a7403@usw-ex0102-013.remarq.com> Tiept wrote: > > I have a bunch of object files, compiled using GNAT ADA > compiler. I haven't had any luck linking these object files > into one object file to used on my main program which is in > the vxworks/tornado environment and on a PPC platform. > > I am using the linking script below: > > ldppc -X -N -Map apple.map -i -o apple.o decoder.o ... and a > bunch of other object files > > I am getting this error: > > decoder.o: file not recognized: File format not recognized > > Is there another switch in ldppc that I am supposed to > use? Appreciate any helps on this. Thanks. > > TNT > > * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * > The fastest and easiest way to search and participate in Usenet - Free! Try using the "objcopy" tool (which in your Tornado distribution will be called "objcopyXXX" where "XXX" is some string that reflects your target CPU; on my project it is objcopyppc). It is used to convert object files from one format to another. That said...you may run into lots of other problems getting the Ada code and the C code to play together, besides the underlying object file format. As I know squat about Ada, I can't help you here. - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: About Global variables and main() functions Date: Wed, 22 Sep 1999 19:02:33 -0700 From: engineer_scotty Organization: aracnet.com Message-ID: <37E98A39.14672730@aracnet.com> References: <37E5CF37.F7582619@cselab.snu.ac.kr> <37E5D543.9B7119E5@aracnet.com> <7savuh$pqu$1@panix3.panix.com> Greg Comeau wrote: > > In article <37E5D543.9B7119E5@aracnet.com> engineer_scotty writes: > >Lee Dong Yul wrote: > >> > >> Hello, We are developing some programs on VxWorks OS. > > > >To sum up all your questions: vxWorks ain't like Unix. > > > >vxWorks doesn't have the concept of separate "programs" with > >separate address spaces, and separate global variables. There is > >only one address space--and when you load new modules with ld(), > >the global variables in that module are added to that address space. > > We recently completed the first phase of porting our compiler, > Comeau C++, to the TMS320C82, with the TI C compiler (hosted on Solaris/SPARC) You're writing a C++ compiler for the C8x? You poor bastard! :) (For those not familiar, the TSM320C8x is a parallel DSP whose architecture is, how shall I say it, bizarre....) I wasn't aware that vxWorks ran on C80s. > We set it up traditionally "like UNIX", and based upon your points > below wonder if there is a problem, a culture, or just pragmatics > tht might not matter. See below. vxWorks ain't UNIX; the underlying architecture is radically different. Despite the presense of a POSIX-ish API, programming practices that are commonplace in Unix-land will get you into trouble in vxWorks. And this is ignoring the tooling issues you ask about below. A couple basic rules for working in vxWorks: 1) Avoid globals whenever possible. In Unix this is good advice as well, but many Unix (older ones especially) are single-threaded so globals aren't quite as bad. ANY program that uses globals in vxWorks will take some work to be re-entrant. 2) Don't define short function names with external linkage (short names with static linkage aren't a problem). A vxWorks "program" shares the symbol table with all other programs in the system. (This is a big reason why you shouldn't use main() below). 3) Clean up after yourself; this system willnot do this for you. > >> 1. Why can't we use main function with C language? > >> > >> When we programmed some applications with main function, We couldn't > >> download those applications. Undefined symbol (__eabi) error occurred. > > > >Lemme guess. PowerPC, right? :) > > > >Including main() causes the compiler/linker to automagically link in a > >bunch of Other Stuff. In Unix, including main() causes prologue code > >(which initializes the C standard library) to run, etc. In C++, this > >prologue code initializes static constructors. > > Although Standard C++ does not require main() to be the static ctor > "clue in", we use it as such often on some platforms, since it is allowed. > Even despite the one address space comment above, what's wrong > with treating it that way? Well... in addition to having a single address space, vxWorks also has a single namespace. If a symbol is declared with external linkage; vxWorks doesn't care much for having multiple symbols with the same name. If you try to link two modules, both with "int main()" defined, at compile time-- you'll get a linker error. If you use the loader to load a module that defines main() and another module already defines main(), the other module's main() will get squashed. (The loader, BTW, will not warn you about duplicate symbols squashed in this manner.) (A question for you: What does the new ISO/ANSI standard say about main and namespaces--is main() part of namespace std? ) > >vxWorks on the powerPC, on the other hand, corresponds to a > >specification > >called EABI (Embedded Application Binary Interface); which calls for a > >function __eabi to be invoked when a program starts. This is, however, > >Not What You Want in vxWorks. > > > >The bottom line: don't call anything main(). > > In brief, what's wrong with __eabi being called on vxWorks? Well, you get to write it for yourself. In the case of the C80, the __eabi function may or may not exist; it is a PowerPC thing. > >> 2. In tornado shell, we can't execute some programs many times. All > >> Those programs have global variables. So, After executing those > >> programs, Global variables contain some values that are different from > >> program startup values. > >> > >> How do we initialize all global variables on program startup? > > > >Do so explicitly. The static initialization happens only once. For > >those objects linked with the kernel, static initialization happens when > >the system starts. If you use the dynamic loader, the initialization > >happens when the module is loaded. However, "restarting" the program > >does > >not cause globals to be initialized. > So with whatever C++ combo with Tornado you're referring to, > restarting won't redo static initializers? No. At least not automagically. The default behavior of vxWorks is as follows: 1) Static constructors of modules statically linked to the kernel are run somewhere in the BSP init; after all drivers, etc are brought up but before usrRoot() starts the shell and/or the user application. 2) Static constructors of modules dynamically loaded into a running system are run at load time. 3) Static destructors of modules dynamically loaded into a running system are run when you UNLOAD the module. (I think; as a rule I tend to avoid static destructors like the plague.) This assumes that you haven't loaded ANOTHER module...see below. 4) Static destructors of modules statically linked to the kernel are, as far as I know, never run. You can disable 2) and 3) (on a SYSTEM-wide basis, not per module) if you want, in which case you can specify when to run static constructors and/or destrctors. This scheme has a LOT of problems. The functions that get defined for you (by a tool called the "muncher", which does a similar thing to "collect" in many Unix environemnts; except that you have to arrange for it to be called explicitly in your build environment rather than having the compiler driver do it for you) are given a well-known name, and entered into the global symbol table (with external linkage). Something like RunStaticCtors() and RunStaticDtors() (I cannot remember offhand, and I am at home, without the relevant manual handy). If you load one module, and then another, the RunStaticC/Dtors() functions of the NEW module (if present; modules consisting only of C code won't have these functions) will replace the old ones in the symbol table. The automatic technique might still work (if WRS were smart, they'd resolve both symbols into function pointers at load time, rather than resolving the symbol at unload time); but the manual technique might cause problems if you try and run destructors from the shell. > >Oh, and one more thing. The automatic cleanup that you get in Unix when > >a program dies (all memory blocks freed, stdio flushed, all files > >closed) > >does NOT happen in vxWorks. You **MUST** free all resources you use > >yourself. > > We also added in static destruction stuff, why is that a problem > (assuming it's needed)? See above. vxWorks probably won't ever call your static destructors unless you tell it to. > >> Why don't vxworks initialize global variables on program startup. > > > >vxWorks views the entire system as a "program"; all the modules are part > >of that program. > > This doesn't seems like a good reason, alone that is. > This is to say, this is somewhat akin to saying that shared libs > on UNIX of DLLs on Win shouldn't take care of any initialization, > and/or arrange for them to be done. However, Win32 DLLs and Unix shared libraries have a separate copy of .data and .bss for each process that attaches to 'em. If I have a .so file that declares a global to be one at initialization--each program that links to the module can assume that this will be the case. vxWorks modules are more like--and this is a scary thought--Windows 3.x DLLs. In other words, the address space of the DLL is system wide. > >> When Initialization occurs? Object code downloading? > > > >Globals defined in the module are initialized when you download the > >module, > >but at no time after. (Unless you unload the module and load it again; > >this will ause them to be re-initialized). > Data only? Data and BSS. VxWorks does zero out the BSS for you. (vxWorks modules frequently also contain COMMON symbols, which the loader tries to treat as BSS, though it allocates a separate memory block for each common symbol. Occasionally, at least on powerPC, it computes the size wrong and gleefully trashes the heap.) > >> 3. On program exit, is dynamically allocated memory automatically freed? > > > >Nope. > > > >> Otherwise, we must free all dynamically allocated memory explicitly? > > > >Yup. > This implies it's either erroneous or unncessary to perform static > destructors. Although we can imagine apps that won't care, > the above sems to imply it's always true??? Erroneous; vxWorks won't call them reliably. The BEST you can hope for is to run each module as a separate task, add a "task delete hook" to call the destructors, and pray. :) You still won't have re-entrant code, and if the task dies an unnatural death static destructors might still run in the wrong context, or not at all. The best thing to do, like I said above, is avoid global variables (including global objects) entirely. I tend to define exactly ONE (or at most two) per module, which I can make a "task variable" if need be to achieve re-entrancy. Scott > - Greg > -- > Comeau Computing, 91-34 120th Street, Richmond Hill, NY, 11418-3214 > Producers of Comeau C/C++ 4.2.38 -- NOTE 4.2.42 BETAS NOW AVAILABLE > Email: comeau@comeaucomputing.com / Voice:718-945-0009 / Fax:718-441-2310 > *** WEB: http://www.comeaucomputing.com *** - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 2 make Date: Sat, 25 Sep 1999 14:59:25 -0400 From: Lei Sun Organization: University of Central Florida Message-ID: <37ED1B8D.2F3F9646@pegasus.cc.ucf.edu> Hi: I was trying to build vxWorks system image after I modified the config.h file, in Tornado 2/vxWorks 5.4. Host is sun4-solaris2, target is MC68k. But when I issued "make" command in /target/config/mv167 directory, it always tell me that : "make: Fatal error in reader: /opt/tornado/target/h/make/defs.bsp, line 148: Unexpected end of line seen" I didn't do any modifications besides to "configAll.h". Anybody know what's the problem. thank you in advance! lei sun --------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: 25 Sep 1999 21:40:27 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <7sjfgb$hp$1@news1.xs4all.nl> References: <37E4E4C5.1F32DC4@pacbell.net> <7s7blf$d6n@ipgw.marconi-instruments.co.uk> <37EBAEDB.D5118C6F@pacbell.net> Nidal Khalil wrote: : Yes I am having the same problem but I boot from a floppy with : bootrom_uncmp then try to load a boot image from the hard disk. : Therefore my boot paramaters do not have a network interface but the : kernel image (vxWorks) has networking enabled in it. The target comes up : with out the network interface attached. I could not do ifAddrSet or : ifMaskSet both calls fail. : It looks like networking is enabled only for network boot. This sucks : I worked with tornado 1.0 and tornado 1.01 but this release of tornado is : premature and should be a beta release not FCS. In addition WindRiver tech : support is very poor. The experience with T2 is that when you not boot from network or one of the default devices the network is not enabled. You have to enable it using the o-bootparameter. The problem is that when you do not boot from one of the default devices tha o-bootparameter is not checked. I don't know how it works with T1, but as they did not change that much it is probably almost the same. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Sep 27 01:02:07 1999 From: "Navin kumar Agrawal" Date: Mon Sep 27 01:02:11 PDT 1999 Subject: Packets from VxSim... VxWorks Tornado Hi All, I am developing a program on VxWorks, which requires sending IP packets to other nodes/hosts. Because of unavailability of the target borad, I am planning to carry out the development activity using VxSim (full fedged one, not the one which comes along with T2 by default). Is it possible to carry out testing of such program using VxSim? Thanks, Regards, NaViN From owner-vxwexplo-process Mon Sep 27 03:58:47 1999 From: "Slutsker, Rasty" Date: Mon Sep 27 03:58:51 PDT 1999 Subject: RE: Bus cycles 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_01BF08D7.07053B7C Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable I'm afraid you're wrong. I had bad experience with x86 for both single and multi-processor environments. GCC generates inconsistent code for floating point operations. For example A=3DB is compiled to two 32-bit Movs, while A=3DB+C compiled to Fload64,Fload64,Fadd, Fstore64. So, as you can, see A=3DB is not atomic = even for single processor. I suggest you to examine assembly ---------- From: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] Sent: =E9=E5=ED =E7=EE=E9=F9=E9 23 =F1=F4=E8=EE=E1=F8 1999 14:24 To: vxworks_users@csg.lbl.gov Subject: Bus cycles Submitted-by owner-vxwexplo-process Thu Sep 23 05:23:49 1999 Submitted-by: "Christopher J. Slominski" = --=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D_523485= 822=3D=3D_.ALT Content-Type: text/plain; charset=3D"us-ascii" VxWorks users, I'm working inder the assumtion that a processor locks the data bus while writing all native data types to memory. For example, a process writing a double to memory can not be interrupted by another process and get the memory bus before all 8 bytes are written. Therefore, in a single writer / multiple reader environment I only need to do operating system atomic mutual exclusion for complex data structures. Is this true for single and multiprocessor systems?=20 =09 ------_=_NextPart_001_01BF08D7.07053B7C Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable RE: Bus cycles

    I'm afraid = you're = wrong.

    I had bad = experience with x86 for both single and = multi-processor environments.

    GCC generates = inconsistent code for floating point operations. For example A=3DB is compiled to two 32-bit Movs, while A=3DB+C compiled to Fload64,Fload64,Fadd, Fstore64. So, as you = can, see A=3DB is not atomic even for single = processor. I suggest you to = examine assembly

    ----------

    From:  = vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov]

    Sent:  =E9=E5=ED =E7=EE=E9=F9=E9 23 = =F1=F4=E8=EE=E1=F8 1999 14:24

    To:  = vxworks_users@csg.lbl.gov

    Subject:  Bus = cycles

    Submitted-by = owner-vxwexplo-process  Thu Sep 23 05:23:49 1999

    Submitted-by: "Christopher J. Slominski" = <c.j.slominski@larc.nasa.gov>

    --=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D_523485822=3D=3D_.ALT

    Content-Type: text/plain; = charset=3D"us-ascii"


    VxWorks users,

         = I'm working inder the assumtion that a processor locks the data bus = while

    writing all native data types to memory.  For example, = a process writing a

    double to memory can not = be interrupted by another process and get the memory

    bus before all 8 bytes = are written.  Therefore, in a single writer / multiple

    reader environment I only = need to do operating system atomic = mutual exclusion

    for complex data = structures.  Is this true for single and multiprocessor

    systems?

------_=_NextPart_001_01BF08D7.07053B7C-- From owner-vxwexplo-process Mon Sep 27 04:00:11 1999 From: daemon@csg.lbl.gov Date: Mon Sep 27 04:00:15 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Sep 27 04:00:07 PDT 1999 Subject: Tornado II and Diab Compiler Subject: Re: inetstatShow Subject: using the Dynamic linker for run time apps Subject: cppfilt*.exe program information? Subject: Re: Header File Dependency Problems Subject: I2C driver Subject: Re: How to generate AIF-Header for a bootrom-image Subject: Cannot Increase SCSI2 Throughput Subject: Re: Booting VxWorks from IDE Drive Subject: Re: WindRiver support - here's a typical example Subject: MMU Initialisation ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Tornado II and Diab Compiler Date: 23 Sep 1999 03:58:38 GMT From: ajeddeloh@aol.com (AJeddeloh) Organization: AOL http://www.aol.com Message-ID: <19990922235838.06489.00002536@ng-cg1.aol.com> Anyone know how to use a different compiler tool set instead of GNU under Tornado II? We have been using Diab under Tornado I, and our code base does not compile under GNU without altering #pragmas &c. Please reply to the group, or email to Alan_Jeddeloh "AT" adc.com. This AOL account does not accept outside email. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: inetstatShow Date: Thu, 23 Sep 1999 09:55:35 -0400 From: "JJ15C3600-B.PATEL(HOP092)NONE" Organization: Lucent Technologies Message-ID: <37EA3157.FBC160A8@lucent.com> References: <37E94A11.6040BB51@lucent.com> <37E9EF03.7B129EE9@bitband.com> Leonid Rosenboim wrote: > > "JJ15C3600-B.PATEL(HOP092)NONE" wrote: > > > when we do inetstatShow, we get a listing of field - PCB ( protocol control > > block). How does one know to which vxworks task this pcb corresponds to? > > > > PCB is a data structure in memory, and as such does not belong to any specific > task, only to a set of modules that manipulate this data. Specifically, > TCP/UDP/IP routines are executed from ANY TASK which typically calls any of the > socket functions to do network I/O. > > Although most input processing is done by tNetTask, iNetTask also does some > output processing (such as TCP retransmissions). Most output code is executed > by the caller of the socket call, up to the driver output function. > > One should be careful about calling socket function from high-priority tasks > which do hard real-time processing, these function execution times are not > stricktly deterministic. > > Hope this explains it. > Leonid I should have mentioned the reason behind asking the question I posted. We have a constructor class (C++) called IpcSock which has the following code - - ----code follows---- // set up the local address sockAddrSize = sizeof(struct sockaddr_in); bzero((char *) &sockAddr, sockAddrSize); sockAddr.sin_family = AF_INET; sockAddr.sin_port = htons(IPCPORT(classname, inst)); sockAddr.sin_addr.s_addr = htonl(INADDR_ANY); printf("Port number = 0x%x\n", sockAddr.sin_port); // create a UDP-based socket if ((sock = socket(AF_INET, SOCK_DGRAM, 0)) == ERROR) { perror("socket failed"); return; } printf("socket fd = %d\n", sock); int value = 1; setsockopt (sock, SOL_SOCKET, SO_REUSEADDR, (char * const)&value, sizeof (value)); #if (CPU != GNUSOLARIS) setsockopt (sock, SOL_SOCKET, SO_BROADCAST, (char * const)&value, sizeof (value)); #endif value = 32*1024; // Size for send and receive buffers: IPC socket // setscokopt() returns ERROR or OK, defined in vxWorx headers. if (setsockopt (sock, SOL_SOCKET, SO_SNDBUF, (char * const)&value, sizeof (value)) == ERROR) { printf("IpcSock.cc: setsockopt() failed for send buffer!\n"); } if (setsockopt (sock, SOL_SOCKET, SO_RCVBUF, (char * const)&value, sizeof (value)) == ERROR) { printf("IpcSock.cc: setsockopt() failed for receive buffer!\n"); } // bind socket to local address if (bind(sock,(struct sockaddr *)&sockAddr, sockAddrSize) == ERROR) { perror("unable to bind"); close(sock); return; } - ------code ends----- The constructor is basically creating a socket, setting some socket options ( in particular - SO_REUSEADDR which is important for us). and then trying bind. Intermittantly ( 1 out of 5 times) the bind fails with the following error - "unable to bind: S_errno_EADDRINUSE". I want to track down the task which is using the portnumber assigned to my task ( the portnumbers are mapped to integers from 1024 onwards). Bhavesh Patel --------------------------- Newsgroups: comp.os.vxworks Subject: using the Dynamic linker for run time apps Date: Thu, 23 Sep 1999 09:47:36 -0400 From: s Organization: EarthLink Network, Inc. Message-ID: Is anyone using the Dynamic linker for run time apps? I am looking for feedback on how well this works. I have heard third hand reports about this feature not working well with large .o's or with complex down line loads (multple .o's externally linked to for a down line loaded image). I know that many people use this feature during development for module replacement. What I am trying to determine is how well this feature works for extending a running system with new features. Any feedback would be greatly appreciated. Thanks, Scott --------------------------- Newsgroups: comp.os.vxworks Subject: cppfilt*.exe program information? Date: Thu, 23 Sep 1999 14:08:56 GMT From: khamon@lintek.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7sdc9e$6gj$1@nnrp1.deja.com> On Thu Sep 23 01:51:36 1999 "Mike Davies" wrote: >if you run the c++filt program (for a Windows hosted 386 target this is >"cppfilt386 _vt$8CVCQCard" ) from the command line then it will tell >you what the function prototype looks like. Thanks Mike, I found the corresponding program for PPC: it is "c:\tornado\host\x86-win32\bin\cppfiltppc.exe". Does anyone know if there is any information on this (or the other executables in the host\...\bin directory) available in the Wind River documentation? Regards, Kent Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Header File Dependency Problems Date: 23 Sep 1999 16:12:57 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <7sdji9$bf8$1@news1.xs4all.nl> References: <7scuit$i166@overload.lbl.gov> In article <7scuit$i166@overload.lbl.gov> you wrote: : Hi, : I've just started using Tornado 2.0 (we have the simulator package at : the moment soon to be followed by StrongArm & Pentium full development : kits.) and am having problems compiling a simple project. : I get the following message when compiling a C file .... : make: *** No rule to make target `foo.h', needed by `applic.o'. Stop. : 'foo.h' is included in the applic.c file with just a '#include foo.h' : i.e. no full path. You need to use the VPATH macro. See my VxWorks page for details about this. (http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html, ref. 1.3 make) Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: I2C driver Date: Thu, 23 Sep 1999 16:12:54 GMT From: imohammed@mailexcite.com Organization: Deja.com - Share what you know. Learn what you don't. Message-ID: <7sdjhp$c9j$1@nnrp1.deja.com> I am working on I2C driver for BSP for VxWorks. We are using MPPC8240. It has built in I2C interface. I am having problem understanding how to perform recieve as a Master or Slave. As a Master, an application would initiate the request to recieve data from the slave. Once the address phase has completed, I will be ready to recieve data. Upon recieving first byte of data, uP will generate interrupt. My ISR will retrieve the data from data register. How do I tell the application that I have data and how do I tell the application where the data is located? thanks. Irfan Mohammed imohammed@mailexcite.com Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to generate AIF-Header for a bootrom-image Date: Thu, 23 Sep 1999 14:48:46 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <37EA760E.BA5ABCD3@lucent.com> References: <37EA21AA.A0B739A2@stud.fh-luebeck.de> If you look at the Makefile, there should be a line to create a 'bootrom.aif'. That is how you build it for the ebsa-285. If you look at the actions that it takes to build the aif file, it builds bootrom, then assembles aif.s, then ldarm -o bootrom.aif aif.o bootrom.tmp. This same set of actions can be applied to any software load that you want to create. Doug R=FCdiger zum Beck wrote: > = > Hi, > = > I'm looking for some help to generate a bootrom with a AIF-Header for a= > Brutus Evulation Board. > I would like to use the serial-port to boot my board. To do this I must= > reconfig some settings and build a new bootrom image and burn this to > the flash rom's. > = > My Problem is burning the roms, because the image must have a AIF-Heade= r > = > to burn it with the EBSA-285 Flash Management Utility (FMU) on the > Brutus-board. > = > In the config\brutus\ path I have found this source file aifHdr.s: > = > DESCRIPTION of afiHdr.s > This file can be assembled to generate a simple AIF header which can be= > prepended to a ROM image so that the will program it into flash ROM an= d > the Primary Boot Loader (PBL) > will run it. Many fields contains junk as the PBL and FMU check very > little. > = > How must I use this file in my project to gernerate a AIF Header? > = > Any help would be appreciated - I can't find anything in the manuals. > = > Best Regards, > = > R=FCdiger. --------------------------- Newsgroups: comp.os.vxworks,comp.periphs.scsi Subject: Cannot Increase SCSI2 Throughput Date: Thu, 23 Sep 1999 14:36:36 -0400 From: "Eric L. Schott" Organization: The Pennsylvania State University, Univeristy Park, PA Keywords: VxWorks,SCSI,Disconnect,throughput Message-ID: <4854D13580C5D211AD190008C73393D03D9992@tgxch01.taylornet.com> Sender: els@picard.arl.psu.edu 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_01BF0605.F3D9D72A Content-Type: text/plain; charset="windows-1252" I'm having trouble increasing throughput by adding a second drive to a SCSI bus. What I observe is the overall bus throughput remains basically unchanged while throughput per drive is cut by 50%. The configuration: - Motorola MVME177 (Has SCSI FAST/NARROW, using NCR710 chip) - VxWorks Tornado II. Kernel is compiled with SCSI2 support. - I have run tests with two pairs of drives. The first pair is two IBM 9LP Ultrastar drives and the second is a pair of M-Systems FFD-350 flash drives. - I have written a small program which spawns one or two tasks (to write to one or two drives). The results (I/O rates in bytes/second): - SCSI2 and hard drives (IBM) num drives write read 1 4306451 7164647 2 2558152 3647221 (each) - SCSI2 and FLASH drives num drives write read 1 1906501 2993704 2 954380 1525201 (each) - I was hoping to see two drive write times much greater than either 2558152 (IBM drives) or 954380 (Flash drives). How can I increase the throughput? The output of the VxWorks SCSI debug is: 0xef58a0 (tDrv0): scsi2Transact: 0xfa7504 (tScsiTask): scsiMgrRequestProc: client request: 0 0xfa7504 (tScsiTask): scsiMgrActivateRequest: thread 0x00fa6044 0xfa7504 (tScsiTask): ncr710ThreadActivate: thread 0x00fa6044: activating 0xfa7504 (tScsiTask): ncr710XferParamsCvt: requested: offset = 7, period = 25 0xfa7504 (tScsiTask): ncr710XferParamsCvt: granted: offset = 7, period = 25 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 1 -> 6 0xfa7504 (tScsiTask): ncr710Event: received event 21 (thread = 0x00fa6044) 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa6044: received event 21 0xfa7504 (tScsiTask): DISCONNECT message in 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 6 -> 2 0xfa7504 (tScsiTask): ncr710Event: received event 22 (thread = 0x00000000) 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa76e0: received event 22 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 0 -> 4 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 4 -> 4 0xfa7504 (tScsiTask): ncr710Event: received event 11 (thread = 0x00fa76e0) 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa76e0: received event 11 0xfa7504 (tScsiTask): ncr710ThreadReconnect: reconnecting thread 0x00fa6044 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 2 -> 6 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 4 -> 0 0xfa7504 (tScsiTask): ncr710Event: received event 21 (thread = 0x00fa6044) 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa6044: received event 21 0xfa7504 (tScsiTask): DISCONNECT message in 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 6 -> 2 0xfa7504 (tScsiTask): ncr710Event: received event 22 (thread = 0x00000000) 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa76e0: received event 22 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 0 -> 4 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 4 -> 4 0xfa7504 (tScsiTask): ncr710Event: received event 11 (thread = 0x00fa76e0) 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa76e0: received event 11 0xfa7504 (tScsiTask): ncr710ThreadReconnect: reconnecting thread 0x00fa6044 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 2 -> 6 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa76e0: 4 -> 0 0xfa7504 (tScsiTask): ncr710Event: received event 20 (thread = 0x00fa6044) 0xfa7504 (tScsiTask): ncr710ThreadEvent: thread 0x00fa6044: received event 20 0xfa7504 (tScsiTask): COMMAND COMPLETE message in 0xfa7504 (tScsiTask): ncr710ThreadComplete: thread 0x00fa6044 completed 0xfa7504 (tScsiTask): ncr710ThreadStateSet: thread 0x00fa6044: 6 -> 0 0xfa7504 (tScsiTask): scsiMgrThreadComplete: thread 0x00fa6044 0xfa7504 (tScsiTask): scsiMgrReplyProc: client reply: 0 0xfa7504 (tScsiTask): scsiMgrCompleteReply: thread 0x00fa6044 (status: 0, errno: 0) 0xef58a0 (tDrv0): memPartFree: invalid block 0xe5fb78 in partition 0x9d898. My program is: - ------_=_NextPart_000_01BF0605.F3D9D72A Content-Type: application/octet-stream; name="scsi2Test.c" Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename="scsi2Test.c" #define INCLUDE_SCSI2 #include #include #include #include #include #include #include extern int scsiDebug; int scsi2TestFirstId =3D 0; int scsi2TestPriority =3D 9; int totalWriteSize =3D 64*1024*1024; /* Number of byte/drive */ const int defaultDrvCount =3D 2; /* Number of drives to test */ const int minDrives =3D 1; const int maxDrives =3D 2; const int defaultBlkSize =3D 64*1024; /* Default Write size */ const int minBlkSize =3D 1024; const int maxBlkSize =3D 1024*1024; /***********************************************************************= *=20 */ int myIo ( int direction, SCSI_PHYS_DEV *pScsiPhysDev, int sector, int numSecs, void *buffer, int numBytes ) { SCSI_COMMAND myCmd; SCSI_TRANSACTION myXact; if ( direction ) myCmd[0] =3D SCSI_OPCODE_WRITE_EXT; else myCmd[0] =3D SCSI_OPCODE_READ_EXT; myCmd[1] =3D 0; myCmd[2] =3D (sector >>24) & 0xff; myCmd[3] =3D (sector >>16) & 0xff; myCmd[4] =3D (sector >> 8) & 0xff; myCmd[5] =3D sector & 0xff; myCmd[6] =3D 0; myCmd[7] =3D (numSecs >> 8) & 0xff; myCmd[8] =3D numSecs & 0xff; myCmd[9] =3D 0; myXact.cmdAddress =3D myCmd; myXact.cmdLength =3D 10; myXact.dataAddress =3D buffer; if ( direction ) myXact.dataDirection =3D O_WRONLY; else myXact.dataDirection =3D O_RDONLY; myXact.dataLength =3D numBytes; myXact.addLengthByte =3D 0; myXact.cmdTimeout =3D 1000000; #ifdef INCLUDE_SCSI2 myXact.tagType =3D SCSI_TAG_ORDERED; myXact.priority =3D scsi2TestPriority; #endif scsiDebug =3D 1; return scsiIoctl ( pScsiPhysDev, FIOSCSICOMMAND, (int) &myXact ); scsiDebug =3D 0; return 0; } /***********************************************************************= * Routine to run test for one drive. */ int oneDriveTest ( int direction, int driveId , int bufSize ) { int i; int trialCnt; int sector, numSecs; int sectorSize, totalSectors; int rc; int *ourTime; ULONG startTick; float fTime; SCSI_PHYS_DEV *pScsiThing; char * buffer =3D malloc ( bufSize ); int * thing =3D ( int * ) buffer; #ifdef INCLUDE_SCSI2 /* Set SCSI2 options. */ SCSI_OPTIONS scsiOpt; rc =3D scsiTargetOptionsGet ( pSysScsiCtrl, driveId, &scsiOpt ); printf ( "scsiTargetOptionsGet: %d\n" " %d %d %d %d %d %d %d \n", rc, scsiOpt.selTimeOut,=20 scsiOpt.messages,=20 scsiOpt.disconnect,=20 scsiOpt.maxOffset,=20 scsiOpt.minPeriod,=20 scsiOpt.tagType,=20 scsiOpt.maxTags ); scsiOpt.minPeriod =3D 10; scsiOpt.disconnect =3D 1; scsiOpt.maxOffset =3D 255; scsiOpt.tagType =3D SCSI_TAG_ORDERED; rc =3D scsiTargetOptionsSet ( pSysScsiCtrl, driveId, &scsiOpt, SCSI_SET_OPT_XFER_PARAMS | SCSI_SET_OPT_TAG_PARAMS | SCSI_SET_OPT_DISCONNECT); printf ( "scsiTargetOptionsSet: %d\n" " %d %d %d %d %d %d %d \n", rc, scsiOpt.selTimeOut,=20 scsiOpt.messages,=20 scsiOpt.disconnect,=20 scsiOpt.maxOffset,=20 scsiOpt.minPeriod,=20 scsiOpt.tagType,=20 scsiOpt.maxTags ); #endif /* Fill the buffer with data. */ for ( i =3D 0; i < bufSize; i++ ) { *(thing++) =3D i; } pScsiThing =3D scsiPhysDevIdGet ( pSysScsiCtrl, driveId, 0 ); rc =3D scsiReadCapacity ( pScsiThing, &totalSectors, §orSize ); if ( sectorSize =3D=3D 0 ) { logMsg ( "Could not read capacity of drive %d\n",=20 driveId, 0, 0, 0, 0, 0 ); if ( ( pScsiThing->numBlocks !=3D 0 ) && ( pScsiThing->blockSize = !=3D 0 ) ) { sectorSize =3D pScsiThing->blockSize; totalSectors =3D pScsiThing->numBlocks; } else { sectorSize =3D 512; totalSectors =3D 1273010; } } numSecs =3D bufSize / sectorSize; =20 sector =3D 0; trialCnt =3D totalWriteSize / ( numSecs * sectorSize ); startTick =3D tickGet ( ); for ( i =3D 0; i < trialCnt; i++ ) { rc =3D myIo ( direction, pScsiThing, sector, numSecs, buffer, numSecs*sectorSize ); sector +=3D numSecs; taskDelay ( 0 ); } fTime =3D (float) ( tickGet() - startTick ) / (float) sysClkRateGet() = ; printf ( "%s: %f s (%f bytes/s)\n",=20 taskName ( taskIdSelf () ), fTime, totalWriteSize / fTime ); free ( buffer ); return OK; } /***********************************************************************= * writeMode: 0=3Dread, 1=3Dwrite * numDrives: 0=3Ddefault number of drives, else number of drives * blkSize: devices size of blocks to write/read */ int scsi2Test ( int writeMode, int numDrives, int blkSize ) { int bSize,nDrives,i; /* Calculate default sizes. */ if ( numDrives >=3D minDrives && numDrives <=3D maxDrives ) nDrives =3D numDrives; else nDrives =3D defaultDrvCount; if ( blkSize >=3D minBlkSize && blkSize <=3D maxBlkSize ) bSize =3D blkSize; else bSize =3D defaultBlkSize; /* Run the tests for each drive. These are spawned tasks. */ for ( i =3D 1; i <=3D nDrives; i++) { char taskName[20]; int thisDrive =3D scsi2TestFirstId + i -1; sprintf ( taskName, "tDrv%d", thisDrive ); taskSpawn ( taskName, 30, 0, 20000, oneDriveTest, writeMode, thisDrive, bSize, 0, 0, 0, 0, 0, 0, 0 ); } return OK; } - ------_=_NextPart_000_01BF0605.F3D9D72A-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting VxWorks from IDE Drive Date: Mon, 27 Sep 1999 09:07:02 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <37EF1796.31DA705@bitband.com> References: <37EE4D02.CD2E0966@hns.com> Reply-To: leonid@bitband.com To put it simply, VxLD uses the BIOS to load the BOOTROM.SYS file into memory, and the error message you got means that BIOS could not find the file, or could not access the primary partition. Perhaps the VxLD uses encient BIOS calls, which dont work anymore on a partition larger then a specific size. It takes a BIOS expert to fix this. John May wrote: > Has anyone overcome this problem? > > I am trying to boot the bootrom_uncmp image from an IDE drive on an x86 > target (Pentium 166), host development system is NT (Service Pack 5). I > follow the directions in the programmers guide, but when it comes time > to boot I get vxLD 1.2 No BOOTROM SYS. I have verified that BOOTROM.SYS > is on the Hard Drive. Any thoughts?? > > Thanks > John May --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindRiver support - here's a typical example Date: Mon, 27 Sep 1999 11:35:17 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <37EF3A55.3C3F2DB2@bitband.com> References: <37EA70F4.B11CC2B5@vixel.com> Reply-To: leonid@bitband.com I have noticed that since Wind River went ISO-certified, a large portion of effort is spend on doing things by the book, rather than pleasing the customer. Perhaps it is time for WRS to reconsider this, and see if the benefits from this ISO thing are worth the trouble. In other words, how much revenue was due to ISO, versus how much cost in man time and lost business it caused. I feel there are few people who care about the TSR number, when all they really want is to have their problem solved and go on with their business. Comments anyone? Leonid (Former WRS employee) Marjorie Krueger wrote: > Several people have asked about WindRiver's technical support. Here's a typical example. My > original TSR stated: > > "In prjConfig.c, there is the call to usrSnmpdInit. I want to remove this call from > prjConfig.c and move it into the init code for our project code. The problem is, if I ever change > the VxWorks options (include another protocol, or remove one) prjConfig.c gets regenerated and wipes > out my changes." > > The response was: > > > > Hello Marjorie, > > > > > > Regarding the TSR#142563, > > > > > > 1) You should not manually edit any files created by the project > > > facility, e.g. prjConfig.c, Makfile..etc, since they are regenerated > > > when there is any change in the project. > > > > > > 2) To remove the SNMP component from the project, you should > > > exclude the SNMP component under the vxWorks view of the > > > Workspace window. The SNMP component is under network components-> > > > networking protocols->network applications->SNMP components. > > > > > > For further details about the project facility, please refer to > > > the chapter 4 of Tornado User's Guide 2.0 (Window version). > > > > > > Regards, > > > > > > Eric Lo > > > Technical Support Engineer > > > Wind River Systems > > Duh! Great, they stated the obvious (and rather limiting) way Tornado 2 does things. And didn't > even answer my question. Guess I should have told them that I already RTFM! > > Sheesh, what a waste. --------------------------- Newsgroups: comp.os.vxworks Subject: MMU Initialisation Date: Mon, 27 Sep 1999 12:58:27 +0200 From: "Richard Marsh" Organization: Siemens Inc. Message-ID: <7snikp$a44$1@news.mch.sbs.de> Reply-To: "Richard Marsh" Hi, We are coming to the end of our project and we want to turn on hardware watchdogs on the PPC860, BUT we get a reset in the usrMmuInit function. The watchdog needs to be reset every 2.4 seconds and the function takes >9 seconds - hence kaboom. We therefore need to manually initialise the MMU does anybody have any experience of this? If we could use the same mechanism as usrMmuInit but with a watchdog reset after each block that would surfice - probably. Help much appreciated, Richard Marsh Siemens Atea - Belgium --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Mon Sep 27 04:20:12 1999 From: Adi Molkho Date: Mon Sep 27 04:20:16 PDT 1999 Subject: Can not NFS to NT client Hi I am working with vxWorks 5.3.1 and my host is windows NT on which I have installed NFS server(from Microsoft) I have configured the NFS server to share e:\pub for everybody I am doing the following : my code : if(hostAdd("Adi", "10.10.10.17") != 0) printf("hostAdd fail\n"); response : None (meaning that return status is OK) my code : nfsAuthUnixSet("Adi", 0, 1, 0,NULL); my code : if(nfsMount("Adi", "e:/pub", "/") != 0) printf("nfsMount fail\n"); response : None (meaning that return status is OK) my code : cd("/"); response: NONE my code : pwd(); response: / my code : ls(NULL,TRUE); response: Can't open ".". my code : fopen("file.txt",READONLY); response: return code is not 0 Does any body know what is wrong here Thanks Adi From owner-vxwexplo-process Mon Sep 27 04:28:45 1999 From: "Slutsker, Rasty" Date: Mon Sep 27 04:28:49 PDT 1999 Subject: FW: Bus cycles 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_01BF08DB.4506C0D6 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Sorry, I previously sent incomplete message. I'm afraid you're wrong. We had bad experience with x86 for both single and multi-processor environments. GCC generates inconsistent code for floating point operations. For example A=3DB is compiled into two 32-bit Movs, while A=3DB+C compiled = to Fload64,Fload64,Fadd, Fstore64. So, as you can, see A=3DB is not atomic = even for single processor. I suggest you to examine assembly code before = assuming that operation is atomic. In multi-processor system it's more complicated. Bus makes the difference! ISA does not guarantee anything. Strictly speaking, 16-bit transfer is atomic if your shared memory is 16-bit wide. I had 8-bit DPRAM = between VxWorks target and Windows and figured out that even 32-bit = test-and-set is not atomic ! 64-bit PCI when connected to 64-bit local bus will do the job. While 32-bit PCI not for sure. Usually, 64-bit transaction is translated into = two 32-bit ones, however if your Host-bridge is programmed to support = bursts or fast Back-to-back, several transactions may be done without dropping = the bus request. Also you may have PCI-to-PCI bridge(s) which invalidates = the above assumptions. Rasty ---------- From: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] =20 Sent: =E9=E5=ED =E7=EE=E9=F9=E9 23 =F1=F4=E8=EE=E1=F8 1999 14:24 To: vxworks_users@csg.lbl.gov =20 Subject: Bus cycles Submitted-by owner-vxwexplo-process Thu Sep 23 05:23:49 1999 Submitted-by: "Christopher J. Slominski" > = --=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D_523485= 822=3D=3D_.ALT Content-Type: text/plain; charset=3D"us-ascii" VxWorks users, I'm working inder the assumtion that a processor locks the data bus while writing all native data types to memory. For example, a process writing a double to memory can not be interrupted by another process and get the memory bus before all 8 bytes are written. Therefore, in a single writer / multiple reader environment I only need to do operating system atomic mutual exclusion for complex data structures. Is this true for single and multiprocessor systems?=20 =09 ------_=_NextPart_001_01BF08DB.4506C0D6 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable FW: Bus cycles

Sorry, I previously sent incomplete = message.

    I'm afraid you're = wrong.

    We had bad experience with x86 for both single and = multi-processor environments.

    GCC generates = inconsistent code for floating point operations. For example A=3DB is = compiled into two 32-bit = Movs, while A=3DB+C compiled to Fload64,Fload64,Fadd, Fstore64. So, as = you can, see A=3DB is not atomic even for single processor. I suggest = you to examine assembly code = before assuming that operation is = atomic.

    In multi-processor system = it's more complicated.

    Bus makes the = difference!

    ISA does not guarantee anything. = Strictly speaking, 16-bit transfer = is atomic if your shared memory is 16-bit wide. I had 8-bit DPRAM = between VxWorks target and = Windows and figured out that even = 32-bit test-and-set is not atomic !


    64-bit PCI = when connected to 64-bit local bus = will do the job. While 32-bit = PCI not for sure. Usually, 64-bit transaction is translated into two 32-bit = ones, however if your Host-bridge is programmed to support bursts or fast = Back-to-back,  several transactions may be done without = dropping the bus request. Also you = may have PCI-to-PCI = bridge(s) which invalidates = the above assumptions.

    Rasty


    ----------

    From:  = vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov]

    Sent:  =E9=E5=ED =E7=EE=E9=F9=E9 = 23 =F1=F4=E8=EE=E1=F8 = 1999 14:24

    To:  vxworks_users@csg.lbl.gov

    Subject:  Bus = cycles

    Submitted-by = owner-vxwexplo-process  Thu Sep 23 05:23:49 1999

    Submitted-by: = "Christopher J. Slominski" <c.j.slominski@larc.nasa.gov>

    --=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D_523485822=3D=3D_.ALT

    Content-Type: text/plain; = charset=3D"us-ascii"


    VxWorks users,

         = I'm working inder the assumtion that a processor locks the data bus = while

    writing all native data = types to memory.  For example, a process writing a

    double to memory can not = be interrupted by another process = and get the memory

    bus before all 8 bytes = are written.  Therefore, in a single writer / multiple

    reader environment I only = need to do operating system atomic mutual exclusion

    for complex data = structures.  Is this true for single and multiprocessor

    systems?

------_=_NextPart_001_01BF08DB.4506C0D6-- From owner-vxwexplo-process Mon Sep 27 05:19:36 1999 From: "tim allen" Date: Mon Sep 27 05:19:39 PDT 1999 Subject: Re: Realtek 8139 >Does anybody use a x86 target with a Realtek 8139 low cost PCI ethernet >controller ? >It is not clear if this chip is or is not NE2000 compatible >(www.realtek.com.tw is unreacheable) The 8139 is not NE2000 compatible. The older devices 8029 and 8019 are the NE2000 compatible devices. As I understand it, there is currently no VxWorks driver for the 8139, but I believe WRS are planning one in the near future. If you know different, please let me know! I have downloaded the spec from the realtek web site, so I can email it to you if you are still unable to access the site. Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com From owner-vxwexplo-process Mon Sep 27 05:22:08 1999 From: "tim allen" Date: Mon Sep 27 05:22:12 PDT 1999 Subject: Re: Using Zinc with VxWorks >The drawback was how slow it was. If they haven't improved the speed of >the software then it could take seconds to draw/redraw a screen. Zinc now uses the UGL layer to interface to the hardware, and this comes as full source. This means that if your device isn't supported, or is partially supported, you can optimise for your particular requirements. Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com From owner-vxwexplo-process Mon Sep 27 05:43:04 1999 From: "tim allen" Date: Mon Sep 27 05:43:08 PDT 1999 Subject: Re: Support for MediaGX? >Has anybody used VxWorks with MediaGX based boards? Yes. If you're using any graphics layers, be sure to use the MediaGX specific UGL driver which is available from the WRS WindSurf support site. Without this the performance is very slow. Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com From owner-vxwexplo-process Mon Sep 27 06:50:40 1999 From: David Laight Date: Mon Sep 27 06:50:44 PDT 1999 Subject: RE: Bus cycles > I'm afraid you're wrong. > I had bad experience with x86 for both single and multi-processor > environments. > > GCC generates inconsistent code for floating point operations. For > example A=B is compiled to two 32-bit Movs, while A=B+C compiled to > Fload64,Fload64,Fadd, Fstore64. So, as you can, see A=B is not atomic > even for single processor. I suggest you to examine assembly It SHOULD be atomic for float (32-bit). Expecting any double operation to be atomic is asking for trouble. I don't even know whether the memory loads done for the FPU are guaranteed to be adjacent even if there isn't an interrupt? BTW years ago I spent a long time looking for a bug in a (locally written) scheduler for an embedded X.25 card. The 'process' ready list would get corrupted. The fault was with the code generated by the compiler for: process->state = 3 - OK it ws PASCAL, and a two bit wide bitfield, but the generated code was equivalent to: process->state &= ~3 process->state |= 3 This has the unfortunate side effect of leaving it zero for a couple of instructions. Chaos ensued. That one was tracked down using a logic analiser to trace the (80286) memory cycles... So you can't blame vxworks! 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 Tue Sep 28 04:00:16 1999 From: daemon@csg.lbl.gov Date: Tue Sep 28 04:00:20 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Sep 28 04:00:11 PDT 1999 Subject: Variable file loading times Subject: Variable file loading times Subject: Variable file loading times Subject: Variable file loading times Subject: Re: Linking object files from other os Subject: FTP-Server with VxWorks Subject: Re: Does TrueFFS come with Tornado 2 Subject: Re: SENS need to much Memory ! Subject: Re: Ethernet Problems Subject: Re: Does TrueFFS come with Tornado 2 Subject: Re: Does TrueFFS come with Tornado 2 Subject: Re: Booting from SCSI: How can i get the symbols? Subject: GNU Toolchain usage in vxWorks: FORTRAN? ADA? Subject: R: Booting VxWorks from IDE Drive Subject: Re: Can not NFS to NT client Subject: Requesting Assistance Subject: Re: Does TrueFFS come with Tornado 2 Subject: Download Ia.n.i.!!! It's free! Subject: cppfilt*.exe program information? Subject: Re: Linking object files from other os Subject: Setup for Pentium75 as Target, serial line for comm Subject: Simple question on file permissions Subject: Re: MMU Initialisation Subject: Network I/O problems Subject: ICE for 80486 running VxWorks Subject: Re: Can not NFS to NT client Subject: PLEASE READ: Header file dependencies solve *with* -I Subject: Help! Changed boot from host to local disk. Subject: usrAtaConfig produce memPartFree-error ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Variable file loading times Date: Mon, 27 Sep 1999 15:07:45 +0100 From: Dve Moore Message-ID: <37EF7A31.C252449E@gecm.com> Hi everyone, I'm experiencing problems loading multiple files from vxWorks using fopen. My host is a unix SUN and I'm using vxWorks 5.3.1 on a PPC603e card When I continually open files, I find that the first few open relatively quickly. Then the next load take ages (like 20 seconds or so) waiting on the 'fopen' command and then the next few load relatively quickly and then a slow one again, etc, etc... This fast/slow/fast/slow load pattern exists for as long as I continue to load files. I've experienced this before with vxWorks - I think it's some sort of network time-out thingy. On that occasion it occured when using 'ls' and 'cd' within the windshell - Once a command works, all others will continue to work quickly for about 30 seconds say and then the next 'ls' or 'cd' takes ages!. Any ideas?. Thanks in advance, Dave Moore --------------------------- Newsgroups: comp.os.vxworks Subject: Variable file loading times Date: Mon, 27 Sep 1999 15:09:25 +0100 From: Dve Moore Message-ID: <37EF7A95.2F1F362C@gecm.com> Hi everyone, I'm experiencing problems loading multiple files from vxWorks using fopen. My host is a unix SUN and I'm using vxWorks 5.3.1 on a PPC603e card When I continually open files, I find that the first few open relatively quickly. Then the next load take ages (like 20 seconds or so) waiting on the 'fopen' command and then the next few load relatively quickly and then a slow one again, etc, etc... This fast/slow/fast/slow load pattern exists for as long as I continue to load files. I've experienced this before with vxWorks - I think it's some sort of network time-out thingy. On that occasion it occured when using 'ls' and 'cd' within the windshell - Once a command works, all others will continue to work quickly for about 30 seconds say and then the next 'ls' or 'cd' takes ages!. Any ideas?. Thanks in advance, Dave Moore --------------------------- Newsgroups: comp.os.vxworks Subject: Variable file loading times Date: Mon, 27 Sep 1999 15:11:01 +0100 From: Dve Moore Message-ID: <37EF7AF4.352C4771@gecm.com> Hi everyone, I'm experiencing problems loading multiple files from vxWorks using fopen. My host is a unix SUN and I'm using vxWorks 5.3.1 on a PPC603e card When I continually open files, I find that the first few open relatively quickly. Then the next load take ages (like 20 seconds or so) waiting on the 'fopen' command and then the next few load relatively quickly and then a slow one again, etc, etc... This fast/slow/fast/slow load pattern exists for as long as I continue to load files. I've experienced this before with vxWorks - I think it's some sort of network time-out thingy. On that occasion it occured when using 'ls' and 'cd' within the windshell - Once a command works, all others will continue to work quickly for about 30 seconds say and then the next 'ls' or 'cd' takes ages!. Any ideas?. Thanks in advance, Dave Moore --------------------------- Newsgroups: comp.os.vxworks Subject: Variable file loading times Date: Mon, 27 Sep 1999 15:12:14 +0100 From: Dve Moore Message-ID: <37EF7B3E.945D1D9C@gecm.com> Hi everyone, I'm experiencing problems loading multiple files from vxWorks using fopen. My host is a unix SUN and I'm using vxWorks 5.3.1 on a PPC603e card When I continually open files, I find that the first few open relatively quickly. Then the next load take ages (like 20 seconds or so) waiting on the 'fopen' command and then the next few load relatively quickly and then a slow one again, etc, etc... This fast/slow/fast/slow load pattern exists for as long as I continue to load files. I've experienced this before with vxWorks - I think it's some sort of network time-out thingy. On that occasion it occured when using 'ls' and 'cd' within the windshell - Once a command works, all others will continue to work quickly for about 30 seconds say and then the next 'ls' or 'cd' takes ages!. Any ideas?. Thanks in advance, Dave Moore --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking object files from other os Date: Mon, 27 Sep 1999 15:43:14 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <7so3af$fc3$1@nnrp1.deja.com> References: <01f8d464.615a7403@usw-ex0102-013.remarq.com> In article <01f8d464.615a7403@usw-ex0102-013.remarq.com>, Tiept wrote: > I have a bunch of object files, compiled using GNAT ADA > compiler. I haven't had any luck linking these object files > into one object file to used on my main program which is in > the vxworks/tornado environment and on a PPC platform. What version of the Gnat Ada compiler? It is one that is targetted to PPC vxWorks, right? Supposedly those exist, but there aren't any public binaries that I know of. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: FTP-Server with VxWorks Date: Mon, 27 Sep 1999 16:50:05 GMT From: Frank.Hensinger@ipetronik.de (Frank Hensinger) Organization: news.gigabell.net // Main Newsreader @ Gigabell Message-ID: <37ef90de.32052729@news.ipf.net> Hi I want to include FTP-Server functionality in my application. But in the VxWorks manuals, you can only find some sheets and libraries about FTP client or remote server. With the Trivial File Transfer Protocol (TFTP) - which do not require an account or password - both, server and client are implemented. Does anybody knows any public-domain-FTP-server with user account and resume option easily to include? Best regards. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does TrueFFS come with Tornado 2 Date: Mon, 27 Sep 1999 17:38:50 GMT From: js_j98@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <7soa38$kaf$1@nnrp1.deja.com> References: Thanks to everyone who replied. Here some more questions: 1. What part or layer (FTL) of software are we paying for? 2. If we want to use a flash part (e.g. ultraNAND from AMD) that is not supported by TrueFFS for Tornado, is it possible to write some software to make it work? What do I need to write for the unsupported flash part? Thanks alot, Jun In article , John or Jenn wrote: > In article Steeve Gagne, > steevegagne@sympatico.ca writes: > >You need to pay $4000 and all work fine > > I think support for DOC2000 flash parts is bundled...if you want > TFFS for non DOC2000, then you need to buy TFFS. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SENS need to much Memory ! Date: Mon, 27 Sep 1999 11:57:03 -0600 From: Stefano Mimmi Organization: Lucent Technologies, Denver Message-ID: <37EFAFEF.54277B69@lucent.com> References: <37E88D04.509F29A6@sel.de> We measured the exact same growth of memory usage when we recently ported our application to Tornado II (vxWorks 5.4). All the buffers for the networking protocols are now statically allocated at initialization time. I'm having some troubles figuring out were the extra 220KB are used based on the default values in netBufLib.h and usrNetwork.c. In Tornado II there are a couple of debug functions that may help sizing down the number of allocated buffers: netStackDataPoolShow netStackSysPoolShow While netPoolShow requires a and I have not seen what it prints out. Stefano --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Ethernet Problems Date: Fri, 24 Sep 1999 14:08:53 GMT From: "Steeve Gagne" Organization: MetroNet Communications Group Inc. Message-ID: References: <37E4EA6F.C7701387@pacbell.net> I Have disable the cache Memory on my board and it work fine now ... Enjoy ... Nidal Khalil wrote in message news:37E4EA6F.C7701387@pacbell.net... > This is the same problem that I am having. > I think it is a setup problem. I posted a description of it but the target > that I am using is a Pentium > > Steeve Gagne wrote: > > > My mpc860 bsp boot from flash > > He download a load image from TFTP and it work fine > > > > When I boot the load image, the board not answer to ping > > and telnet from exterior. > > > > On the Net, when I throw a echo request paquet > > from the developpement board, I catch the reply on the net but the board > > didn't receive it. > > > > The Ip addr, the subnet mask, the Mac Address All It's ok > > > > I suspect an interrupt problem. > > > > Any Ideas ?? > > > > Steeve Gagne > > Firmware Designer > > steevegagne@sympatico.ca > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does TrueFFS come with Tornado 2 Date: Fri, 24 Sep 1999 14:13:53 GMT From: "Steeve Gagne" Organization: MetroNet Communications Group Inc. Message-ID: References: <7rs7di$bs0$1@nnrp1.deja.com> <0221c012.6a9a2bc5@usw-ex0102-015.remarq.com> You need to pay 4000 $ and all work fine If you don't pay 4000 $ all you are able to do is attach your sockets but you won't be able to format flash You also have unresolved symbols ... Enjoy ! Steeve Gagne Mark Dailey wrote in message news:0221c012.6a9a2bc5@usw-ex0102-015.remarq.com... > You need to pay the $4000 to get a couple of key objects. > > * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * > The fastest and easiest way to search and participate in Usenet - Free! > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does TrueFFS come with Tornado 2 Date: Fri, 24 Sep 1999 14:22:07 GMT From: "Steeve Gagne" Organization: MetroNet Communications Group Inc. Message-ID: References: <7rs7di$bs0$1@nnrp1.deja.com> <0221c012.6a9a2bc5@usw-ex0102-015.remarq.com> The TFFS package is built in your Tornado 2 Cd Rom when you buy the package, wrs send you a different install key wich unlock the cd and install the TFFS Package Steeve Gagne wrote in message news:BGLG3.47$_s.932777@news1.mtl.metronet.ca... > You need to pay 4000 $ > and all work fine > If you don't pay 4000 $ all you are able to do is > attach your sockets but you won't be able to format > flash > You also have unresolved symbols ... > Enjoy ! > > Steeve Gagne > > Mark Dailey wrote in message > news:0221c012.6a9a2bc5@usw-ex0102-015.remarq.com... > > You need to pay the $4000 to get a couple of key objects. > > > > * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network > * > > The fastest and easiest way to search and participate in Usenet - Free! > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting from SCSI: How can i get the symbols? Date: 27 Sep 1999 20:16:47 GMT From: jcc@netcom.com (Jason C. Cotton) Organization: Netcom Message-ID: <7sojbf$d37@dfw-ixnews12.ix.netcom.com> References: <7sdckt$i174@overload.lbl.gov> <7se1pe$egh@dfw-ixnews15.ix.netcom.com> <37EBBA05.43F2E4D5@t-online.de> Friedrich Scheurer / Goeppingen writes: I had written: >> Why not copy vxWorks.st to your local disk, and boot that instead? >What version of the file is it?(standalone) Please apologize my silly >question. Yes, vxWorks.st is the "standalone" version of vxWorks. >If i changed my kernel only two files were built vxworks and vxworks.sym, >if i'm right. You need to "make vxWorks.st" on the host. Then, you can boot it off the network, copy it to your local disk, and boot it off of there. At that point, you can use the (target) shell to ascertain the VxWorks's notion of the available devices, current working directory, etc. >> I suspect that your problem loading the symbols off disk arises from >> VxWorks' default working directory not being set correctly. Loading >> vxWorks.st from disk will allow you to check that out, and also to >> test for case-sensitivity problems. >> >As i mentioned previously to Joe Hansen, i think that the used kerneldoesn't >offer the service needed to load it's own symbol table and >startup script through the loopback device lo0. I didn't read the previous exchange, so maybe you can fill me in on any relevant details. How can an external file ever be loaded through a loopback interface??? vxWorks.st won't need to load a symbol table from anywhere. I wouldn't worry about the startup script until after you troubleshoot your other problems. At that point, you may wish to execute the startup script from disk as well. - -Jason --------------------------- Newsgroups: comp.os.vxworks Subject: GNU Toolchain usage in vxWorks: FORTRAN? ADA? Date: Mon, 27 Sep 1999 16:36:51 -0700 From: "Mike Campbell" Message-ID: <7sok9k$pk8$1@autumn.news.rcn.net> I'm looking for anyone who has used the GNU ADA / Fortran toolchain in developing for vxWorks. In theory, using gnat and the gfc (think that's it) compiler should work like using the GNU C compiler, since they are just branches off the same toolchain tree. I'd be interested in hearing the pros / cons of attempting this type of development, as well as any lessons learned. When you talk to vxWorks and mention ADA and FORTRAN, you get pointed to Greenhills. Having just completed a FORTRAN code port from an antiquated system to a VME based system running vxWorks using Greenhills, I found that Greenhills is on the edge of no longer really supporting their Fortran product. Thus, the interest in a pure GNU approach. An additional reason for this request is I'm trying to see if these tools usage for target development in vxWorks differs substantially from their usage in Linux for target (Linux) development. I realize there are substantial differences in Linux and vxWorks the relative merit of usage of either one given today's processor speeds has sparked several religious debates in this news group. What I'm really interested in is the relatively narrow discussion on the GNU tool usage (ADA/Fortran) in the Linux vs. vxWorks arena. I'll summarize and post whatever responses I get to the news group, and thanks in advance to all who care to throw their two cents in. --------------------------- Newsgroups: comp.os.vxworks Subject: R: Booting VxWorks from IDE Drive Date: Mon, 27 Sep 1999 21:29:55 +0200 From: "El Zorro" Organization: TIN Message-ID: <7sogme$omt$1@nslave1.tin.it> References: <37EE4D02.CD2E0966@hns.com> Reply-To: "El Zorro" are you sure that BOOTROM.SYS is countigously allocated on your HD sectors ? I remember something about it written on the vx manual .. take a look . Bye, Z John May wrote in message 37EE4D02.CD2E0966@hns.com... > Has anyone overcome this problem? > > I am trying to boot the bootrom_uncmp image from an IDE drive on an x86 > target (Pentium 166), host development system is NT (Service Pack 5). I > follow the directions in the programmers guide, but when it comes time > to boot I get vxLD 1.2 No BOOTROM SYS. I have verified that BOOTROM.SYS > is on the Hard Drive. Any thoughts?? > > Thanks > John May > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can not NFS to NT client Date: Mon, 27 Sep 1999 16:46:02 -0700 From: "Mike Campbell" Message-ID: <7sokqn$ssq$1@autumn.news.rcn.net> References: <7snn3q$12s4@overload.lbl.gov> I think that just because you mounted it, it doesn't mean that its the current device. Specifically if you tried: > my code : if(hostAdd("Adi", "10.10.10.17") != 0) printf("hostAdd > fail\n"); > response : None (meaning that return status is OK) > my code : nfsAuthUnixSet("Adi", 0, 1, 0,NULL); > my code : if(nfsMount("Adi", "e:/pub", "/") != 0) printf("nfsMount > fail\n"); > response : None (meaning that return status is OK) > my code : cd("/"); cd ("Adi:/") or something along those lines you might have better luck. Not being vxgeeky enough to have vxWorks ref. pages at home, can't be more specific than that - but think that's the right general direction --------------------------- Newsgroups: comp.os.vxworks Subject: Requesting Assistance Date: Mon, 27 Sep 1999 20:56:52 GMT From: "TalentLab Inc." Organization: TalentLab Inc. Message-ID: Hi this is Stuart Musson and I am contacting you for two reasons, I would like to inform you of our weekly e-letter on career tips and hot jobs, you can subscribe at http://www.talentlab.com and I am also currently searching for an Embedded OS Designer for a client. I don’t know if you are someone that I should be speaking with about this interesting opportunity but it is my hope that you may be able to help me out in this search. Our client can offer stock options, a friendly team-oriented environment and the opportunity to work on very intriguing products. I have taken the liberty to attach the job description for your perusal, if you are interested in this opportunity or if you know of anyone who is currently looking to further their career. Embedded OS Designer Our client’s RTOS team is looking for a dynamic candidate who wants to join an exciting team developing next generation products in the enterprise market. The work involves Embedded Operating System (kernel) development to deliver a best-in-class enterprise switch. This includes designing and implementing kernel features to extend and augment commercially available RTOS's, as well as BSP development. The successful candidate will have software design and development experience at the device driver level. Operating System design (preferably VxWorks) experience is required. Detailed PowerPC knowledge is desirable. Strong software development skills, with a "do it right the first time" attitude are required. As well, the candidate must have strong English written and oral communication skills since the work involves documenting and reviewing proposed designs with the rest of the team. Good knowledge of UNIX, C, and PPC assembler is desirable. Experience with object-oriented design using C++ would be an asset. Self-starters and highly motivated candidates with good problem analysis skills are needed for this position. To find out more about this position, I can be reached at the co-ordinates below. Thanking you in Advance, Stuart Musson stuartm@talentlab.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does TrueFFS come with Tornado 2 Date: Fri, 24 Sep 1999 16:35:53 GMT From: John or Jenn Organization: Who, us organized??? Message-ID: References: In article Steeve Gagne, steevegagne@sympatico.ca writes: >You need to pay $4000 and all work fine I think support for DOC2000 flash parts is bundled...if you want TFFS for non DOC2000, then you need to buy TFFS. --------------------------- Newsgroups: comp.os.vxworks Subject: Download Ia.n.i.!!! It's free! Date: 27 Sep 1999 20:30:40 GMT From: madQ Organization: madQ Message-ID: <7sok5g$hmd$779@nslave1.tin.it> Download Ia.n.i. RemoteControlSystem 1.2 beta. It's free!!! New site: http://jump.to/IaniProject --------------------------- Newsgroups: comp.os.vxworks Subject: cppfilt*.exe program information? Date: Mon, 27 Sep 1999 14:18:35 +0200 From: "Borkhuis, Johan" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <7sdc9e$6gj$1@nnrp1.deja.com> > ---------- > From: khamon@lintek.com[SMTP:khamon@lintek.com] > Posted At: donderdag 23 september 1999 15:08 > Posted To: vxworks > Conversation: cppfilt*.exe program information? > Subject: cppfilt*.exe program information? > > Does anyone know if there is any information on this > (or the other executables in the host\...\bin directory) > available in the Wind River documentation? > Yes, it is available in the GNU toolkit documentation. This tool is documented in the GNU Binary Utilities. Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking object files from other os Date: Mon, 27 Sep 1999 15:52:26 -0700 From: Tiept Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <02f42b85.69c7e14c@usw-ex0102-013.remarq.com> References: <01f8d464.615a7403@usw-ex0102-013.remarq.com> <7so3af$fc3$1@nnrp1.deja.com> Thanks to both gentlemen for replying. I haven't tried the objcopyXXX tool and since I am a newbie at this thing, will need to look it up. I am currently using the free GNAT 3.11p version available on the net. I guess I will have to inquire about the GNAT VxWorks version eventually. Thanks again. TNT * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Setup for Pentium75 as Target, serial line for comm Date: 28 Sep 1999 02:16:55 GMT From: ied168@aol.com (IED168) Organization: AOL http://www.aol.com Message-ID: <19990927221655.15360.00002503@ng-bk1.aol.com> Greetings VxWorksman, I've got a Pentium 75 for a target and I'm using a serial line for communication between host and target. The harddrive on the Pentium75 has been formatted and is now empty. I know I need to have a specially configured boot disk for serial communications; however, I am not sure if I need to set up SLIP, TSFS, ...? Also, I would like to have the main program (e.g. "hello world") on my host c: drive (to be downloaded into the target at bootup)... so I don't need a "VxWorks" or "VxWorks.st" image on the floppy boot disk, right? I've tried many things, but the target server cannot connect to the target machine. Any ideas? - -Ed UC Berkeley --------------------------- Newsgroups: comp.os.vxworks Subject: Simple question on file permissions Date: Mon, 27 Sep 1999 21:54:24 -0700 From: Marco Zandonadi Organization: SBC Internet Services Message-ID: <37F04A00.965AE5EA@teja.com> Hello, I have the following problem on VXWorks (Tornado 2.0 on Solaris). I wrote a program that creates a directory using mkdir(). On Solaris and Linux mkdir() has two arguments: name and permissions. On VXWorks it only has name and the created directory has permissions 000. So I looked through the documentation for something to change file permissions but I couldn't find anything (fchmod() isn't there). Does anybody know how to fix this? Please reply via e-mail because I cannot check this group regularly. Thanks in advance for your help! Marco Zandonadi --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MMU Initialisation Date: Mon, 27 Sep 1999 22:48:52 -0700 From: engineer_scotty Organization: Aracnet Internet Message-ID: <37F056C4.10EF6B9C@aracnet.com> References: <7snikp$a44$1@news.mch.sbs.de> Richard Marsh wrote: > > Hi, > > We are coming to the end of our project and we want to turn on hardware > watchdogs on the PPC860, BUT we get a reset in the usrMmuInit function. The > watchdog needs to be reset every 2.4 seconds and the function takes >9 > seconds - hence kaboom. We therefore need to manually initialise the MMU > does anybody have any experience of this? If we could use the same mechanism > as usrMmuInit but with a watchdog reset after each block that would > surfice - probably. 9 seconds for MMU init. You must be mapping in a gigabyte or so of PCI address space, right? One thing that might help (if that's the case), would be to go out and figure out the ACTUAL size of your PCI addr space, modify sysPhysMemDesc appropriately, and only map THAT. Will save lots of RAM too (as 1GB of address space requires 256K page table entries if 4K pages are used--that's a megabyte, if my math and my memory are correct.) Of course... it would be nice if vxWorks provided proper support of the 8xx MMU (with pages >4K). But they don't.... - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Network I/O problems Date: Fri, 24 Sep 1999 13:38:42 -0700 (PDT) From: silvermang@webtv.net (Greg Silverman) Organization: WebTV Subscriber Message-ID: <242-37EBE152-67@newsd-222.iap.bryant.webtv.net> Our system consists of 3 Motorola processors running VxWorks 5.3.1, a cluster of 10 Mercury processors, and a Solaris 2.5.1 workstation. At bootup we do the following steps: 1) download the VxWorks images, 2) configure the Mercury cluster from the master Motorola's bootup script, and, then, 3) start a program on the master Motorola that downloads and starts all of the application programs. Too often the system fails during the last step, with an access violation or a file I/O error in one of the Vxworks tasks. To test the system software I replaced the last step of the bootup procedure with a VME reset, so that the system was in a "boot loop", e.g., no application code is run. If there are no problems with the VxWork/Solaris/Mercury installations and interfaces this boot loop should go on forever. But, every 20 or so boots, one of the VxWorks cards is unable to load some files from the Sun. I have seen: 1) it can not load the VxWorks image, 2) it could not read the VxWorks startup script, 3) the startup script could not load one of the code files. This seems to fail way too often and indicates some kind of network I/O problem that may explain our problem with bringing up the applications. Any ideas? Greg Silverman TRW Sunnyvale, Ca --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: ICE for 80486 running VxWorks Date: Tue, 28 Sep 1999 14:28:19 +0800 From: Richard Tseng Organization: Cyber Express Communications (using Airnews.net!) Message-ID: We are running into a particularly nasty bug where our VxWorks application will crash killing the target server leaving us with no debug information. We suspect stack corruption and would like to purchase an ICE to find out what our app and VxWorks were doing prior to the crash. We have so far evaluated Microtech PowerPack and Applied Microsystems Corp SuperTAP. Both can do the job but quite expensive. The main feature we were looking for is history trace and triggering on error conditions. It will help if there is source level debugging and RTOS integration. Could anyone recommend a lower price ICE for 486 ? Does anyone have experience using Microtech or AMC ICE's with VxWorks ? Thanks, Richard Tseng We Software Ltd --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can not NFS to NT client Date: Tue, 28 Sep 1999 11:07:56 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <37F0856C.3F5706B0@bitband.com> References: <7snn3q$12s4@overload.lbl.gov> <7sokqn$ssq$1@autumn.news.rcn.net> Reply-To: leonid@bitband.com Well, I think the trouble here is that you can not mount an NFS volume as "/" in VxWorks, the mounted volume must have a unique device prefix, differentiating the device from other local or mouted volumes. It is very easy to enable NFS_MOUT_ALL, and let it do all the work, mounting all the shared devices from the current host, with the same mount points as on the server. I do not think anypone has tried this before with Microsoft's NFS server, so this one could be a source of problems too. HTH, Leonid Mike Campbell wrote: > I think that just because you mounted it, it doesn't mean that its the > current device. Specifically if you tried: > > > my code : if(hostAdd("Adi", "10.10.10.17") != 0) printf("hostAdd > > fail\n"); > > response : None (meaning that return status is OK) > > my code : nfsAuthUnixSet("Adi", 0, 1, 0,NULL); > > my code : if(nfsMount("Adi", "e:/pub", "/") != 0) printf("nfsMount > > fail\n"); > > response : None (meaning that return status is OK) > > my code : cd("/"); > > cd ("Adi:/") > > or something along those lines you might have better luck. Not being vxgeeky > enough to have vxWorks ref. pages at home, can't be more specific than > that - but think that's the right general direction --------------------------- Newsgroups: comp.os.vxworks Subject: PLEASE READ: Header file dependencies solve *with* -I Date: Tue, 28 Sep 1999 09:36:09 +0200 From: "Borkhuis, Johan" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <003c01bf0687$50264460$70cb09c0@WST012.rohill.nl> > ---------- > From: Geurt Vos[SMTP:G.Vos@rohill.nl] > Posted At: vrijdag 24 september 1999 13:21 > Posted To: vxworks > Conversation: PLEASE READ: Header file dependencies solve > *with* -I > Subject: PLEASE READ: Header file dependencies solve *with* -I > > Some 2 weeks ago I found out why the -I flag fails. At least when your > host is a Windows machine. You can still use the -I flag, *but* you > have to use SLASH ('/') instead of BACKSLASH ('\') in the path > name! So, instead of something like: > -IC:\projects\ThisPrj1\Include2 > you have to use: > -IC:/projects/ThisPrj1/Include2 > > The weird part is that '\' sometimes does work, or even for > some files in the project... > The savest way I found is to always use the forward slash. Afterall Tornado is still based on Unix. Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Help! Changed boot from host to local disk. Date: Tue, 28 Sep 1999 09:33:57 +0200 From: "Borkhuis, Johan" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <7sgijd$12r1@overload.lbl.gov> > ---------- > From: vxwexplo@lbl.gov[SMTP:vxwexplo@lbl.gov] on behalf of > Friedrich.Scheurer@t-online.de (Friedrich Scheurer / > Goeppingen)[SMTP:Friedrich.Scheurer@t-online.de] > Posted At: vrijdag 24 september 1999 12:44 > Posted To: vxworks > Conversation: Help! Changed boot from host to local disk. > Subject: Help! Changed boot from host to local disk. > > VxWorks, Tornado > > Hi VxWorkers, > > i want to run my MVME2700 standalone, w/o any host. > > I managed it to get the kernel image to the local SCSI-Disk. > The system seems to start up right. But it doesn' t load the > symbols that i put on the local disk, too. > > Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error > opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x3d > Are long filenames enabled on this disk? Can you do a ls (or an ll) of this disk? Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] > ]]]] (R) > ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] > ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] > ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] > ]]]]] > ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] > ]]]] > ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] > ]]]] > ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] > ]]]] > ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] > ]]]] ]]]] > ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] > ]]]] > ]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System > ]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 > ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 > ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., > 1984-1997 > > CPU: Motorola MVME2700 - MPC 750. > Processor #0. > Memory Size: 0x2000000. BSP version > 1.1/5. > WDB: Ready. > > Unable to open startup script /sd0/startup.rtw.tmpl > -> > ... > 2. > [VxWorks Boot]: @ > > boot device : scsi=1,0 > processor number : 0 > file name : /mv2700.MatlabIPPC_Fast > inet on ethernet (e) : 129.69.88.200:ffffff00 > flags (f) : 0x2 > target name (tn) : zfsrc5 > startup script (s) : /startup.rtw.tmpl > > Attaching to scsi device... done. > Loading /mv2700.MatlabIPPC_Fast...1000180 > Starting at 0x100000... > > Host Name: bootHost > User: target > Attaching network interface lo0... done. > Loading symbol table from /mv2700.MatlabIPPC_Fast.sym ...Error opening > /mv2700.MatlabIPPC_Fast.sym: status = 0x3d > > > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] > ]]]] (R) > ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] > ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] > ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] > ]]]]] > ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] > ]]]] > ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] > ]]]] > ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] > ]]]] > ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] > ]]]] ]]]] > ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] > ]]]] > ]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System > ]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 > ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 > ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., > 1984-1997 > > CPU: Motorola MVME2700 - MPC 750. > Processor #0. > Memory Size: 0x2000000. BSP version > 1.1/5. > WDB: Ready. > > Unable to open startup script /startup.rtw.tmpl > -> > > ... > 3. > [VxWorks Boot]: @ > > boot device : scsi=1,0 > processor number : 0 > host name : zfsrc5 > file name : /sd0/mv2700.MatlabIPPC_Fast > inet on ethernet (e) : 129.69.88.200:ffffff00 > host inet (h) : 129.69.88.200 > user (u) : vxwboot > flags (f) : 0x2 > target name (tn) : zfsrc5 > startup script (s) : /sd0/startup.rtw.tmpl > > Attaching to scsi device... done. > Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 > Starting at 0x100000... > > Attaching network interface lo0... done. > Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error > opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x41 > > > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] > ]]]] (R) > ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] > ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] > ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] > ]]]]] > ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] > ]]]] > ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] > ]]]] > ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] > ]]]] > ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] > ]]]] ]]]] > ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] > ]]]] > ]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System > ]]]]]]]]]]]]]]]]]]]]]]]]]]]] > ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 > ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 > ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., > 1984-1997 > > CPU: Motorola MVME2700 - MPC 750. > Processor #0. > Memory Size: 0x2000000. BSP version > 1.1/5. > WDB: Ready. > > Unable to open startup script /sd0/startup.rtw.tmpl > -> > > ... > > -- > +-------------------------------------------------+ > | Dipl.-Ing. F. Scheurer | > | c/o Zentrum Fertigungstechnik Stuttgart | > | Nobelstrasse 15 | > | D-70569 Stuttgart | > | Telefon +49 711 13162-41(-0) | > | Fax +49 711 13162-11 | > | E-Mail scheurer@zfs.tz.uni-stuttgart.de | > | WWW http://zfs.rus.uni-stuttgart.de | > +-------------------------------------------------+ > | privat: Waldeckstrasse 31 | > | D-73035 Goeppingen | > | Telefon +49 7161 686362 Fax +49 7161 686301 | > | E-Mail Friedrich.Scheurer@t-online.de | > +-------------------------------------------------+ > > --------------------------- Newsgroups: comp.os.vxworks Subject: usrAtaConfig produce memPartFree-error Date: Tue, 28 Sep 1999 12:39:31 +0200 From: Torsten Rupp Organization: University of Karlsruhe Message-ID: <37F09AE3.41C67EA6@fzi.de> Hello, I try to initialize a DOS-Harddisk-Partition with usrAtaConfig(). By calling usrAtaConfig 0,0,"C:" it got the error message 0x.... (tShell): memPartFree: invalid block 0x... in partition 0x... Error during dosFsDevInit: 380001 It seems that dosFsDevInit() fail because of some unknown reason (the partition is reable/writeable under DOS without any errors; scandisk doesn't show any problems) and some call of free() in dosFsDevInit has a bad value. Unfortuantely there is no source code available to check what is going wrong nor I can find any information in the manuals. Any hint? Torsten PS: a few days ago usrAtaConfig() succeeded, but after a repairing an error in the filesystem I get the error above. - -- Computer Science Research Center (FZI) phone : +49-721-9654-330 Mobility Management and Robotics fax : +49-721-9654-309 Haid-und-Neu-Str. 10-14 e-mail: rupp@fzi.de D-76131 Karlsruhe (Germany) www : www.fzi.de/robot --------------------------- End of New-News digest ********************** From owner-vxwexplo-process Tue Sep 28 08:49:33 1999 From: Friedrich Scheurer / ZFS Date: Tue Sep 28 08:49:37 PDT 1999 Subject: Re: Booting from SCSI - Building a Standalone Version Hi Joseph, hi vxworks&tornado-users thanks a lot for Your help and Your detailed description. You and the the other replier have been right that i should have used the vxWorks.st (Standalone) Kernel-File! I didn't strip off the original mail, because it is interesting as is. > From HansenJR@navair.navy.mil Fri Sep 24 22:22:11 1999 > Message-ID: <0560141A7916D211A64A00A0C9C9A8156D3092@nwms15.chinalake.navy.mil> > From: "Hansen, Joe" > To: "'scheurer@zfs.tz.uni-stuttgart.de'" > , > "'Friedrich.Scheurer@t-online.de'" > > Subject: Booting from SCSI > Date: Fri, 24 Sep 1999 13:25:18 -0700 > Friedrich, > As you supposed below, my case is a little bit easier, and i am sure that the SCSI-fixed disk access works. I already have been able to copy my files over the net from the server to my local disk. > This is going back a few years, but I did get a Force Sparc board to boot > from a SCSI disk successfully. My main caveat on this is that it was done > with VxWorks 5.1. I would expect that similar steps would have to be made > in more-recent versions, but I don't _know_ this as fact. > > First, make sure that INCLUDE_SCSI_BOOT is defined in your BSP. > > Next, I added the following statements to usrConfig.c. The defines can go > anywhere. > > #define SEAGATE_CONFIG > #define DEFINE_SD0 > #define DEFINE_SD1 > #define DEFINE_SD2 > > #ifdef SEAGATE_CONFIG > SCSI_PHYS_DEV *pSpdSG; /* ptr to SCSI phys dev for Seagate */ > BLK_DEV *pSbdSG0; /* ptr to SG block device for DOS partition */ > BLK_DEV *pSbdSG1; /* ptr to SG block device for RAW partition */ > BLK_DEV *pSbdSG2; /* ptr to SG block device for DOS partition */ > DOS_VOL_DESC *dosDev0; /* DOS device for vxWorks */ > DOS_VOL_DESC *dosDev1; /* DOS device for data files */ > RAW_VOL_DESC *rawDev; /* raw device for data capture */ > #endif > > The following lines are placed in usrScsiConfig (), right after the check > for "#ifdef SCSI_AUTO_CONFIG" (which I had undefined). > > #ifdef SEAGATE_CONFIG > #define DOS_FS_OFFSET 63 > > printf ("SEAGATE_CONFIG defined; creating SCSI device ...\n"); > > /* Create SCSI phys dev structure. Config SG device at busID=2, LUN=0. */ > if ((pSpdSG = scsiPhysDevCreate (pSysScsiCtrl, 2, 0, 0, NONE, 0, 0, 512)) > == > (SCSI_PHYS_DEV *)NULL) > { > SCSI_DEBUG_MSG ("usrScsiConfig: scsiPhysDevCreate failed Seagate.\n"); > printf ("dev create failed ...\n"); > } > > else > { > > #ifdef DEFINE_SD0 > /* Create logical DOS partition on SCSI block device */ > printf ("Creating first logical DOS partition ...\n"); > if ((pSbdSG0 = scsiBlkDevCreate (pSpdSG, 32008, DOS_FS_OFFSET)) == NULL) > { > printf ("Logical DOS partition #1 failed ...\n"); > return (ERROR); > } > > /* Set up the DOS volume configuration info */ > printf ("DOS file system dev init on pSbdSG0 ...\n"); > if ((dosDev0 = dosFsDevInit ("/sd0/", pSbdSG0, NULL)) == NULL) > { > printf ("dosFsDevInit error\n"); > return (ERROR); > } > #endif /* DEFINE_SD0 */ > > #ifdef DEFINE_SD2 > /* Create second logical DOS partition on SCSI block device */ > printf ("Creating second logical DOS partition ...\n"); > if ((pSbdSG2 = scsiBlkDevCreate (pSpdSG, 7850430, 32193)) == NULL) > { > printf ("Logical DOS partition failed ...\n"); > return (ERROR); > } > > /* Associate the block device with dosFs file system functions */ > printf ("DOS file system dev init on pSbdSG2 ...\n"); > if ((dosDev1 = dosFsDevInit ("/sd2/", pSbdSG2, NULL)) == NULL) > { > printf ("dosFsDevInit error\n"); > return (ERROR); > } > #endif /* DEFINE_SD2 */ > > #ifdef DEFINE_SD1 > /* Create logical RAW partition on SCSI block device */ > printf ("Creating logical raw partition ...\n"); > if ((pSbdSG1 = scsiBlkDevCreate (pSpdSG, 0, 7882623)) == NULL) > { > printf ("Logical raw partition failed ...\n"); > return (ERROR); > } > > /* Associate the block device with rawFs file system functions */ > printf ("Raw file system dev init on pSbdSG1 ...\n"); > if ((rawDev = rawFsDevInit ("/sd1/", pSbdSG1)) == NULL) > { > printf ("rawFsDevInit error\n"); > return (ERROR); > } > > /* Change the mode of the raw partition for read/write access */ > rawFsModeChange (rawDev, 2); > #endif /* DEFINE_SD1 */ > > printf ("\n"); > > } > > #endif /* SEAGATE_CONFIG */ > > Basically, I created a physical device, with three block devices. The > devices /sd0/ and /sd2/ are initialized to use the DOS file system, while > /sd1/ is created as a raw device. It is also important (in this example) to > note that I was using a 2GB SCSI drive. > > My intentions here were to have a small (10MB) partition with VxWorks and > VxWorks.sym (/sd0/). I wanted the majority of the drive (about 1.8GB) to be > for telemetry files (/sd2/). However, during a run, I could not dump data > to a DOS partition fast enough (about 1.25MB/sec). Therefore, I created the > raw partition (/sd1/) for temporary storage. After a 3-minute run, I would > then copy the raw partition to /sd2/. > > Also, I needed /sd2/ to be visible from a PC running Windows. Therefore, I > had to preserve the FAT at the beginning of the disk, so I had to offset the > start of /sd0/ by 63 blocks. This will be different for different-sized > disks. From a discussion I had last week, I know that a 1GB drive needs an > offset of 32 blocks. I assume that this could be scaled up with bigger > disks, but large disk support has always been kind of fuzzy with Wind River. > I don't know if they support disks larger than 4GB. > > Obviously, if your requirements are simpler, then you can remove a lot of > the setting-up that I had to do. > > Finally, I had to burn a new boot ROM in order for this to work. I think > that this was because of the offset to /sd0/, but I don't remember for sure. > I know that I had to patch bootConfig.c to allow me to boot off of SCSI, but > I don't think that this file is being used in the newer releases of VxWorks. > > I then booted off of the net, and copied my modified vxWorks and vxWorks.sym > file from the host to /sd0/. Then I changed my boot parameters to the > following: I also tried it this way, but the symbols couldn't be loaded. This is why i tried the vxWorks.st - Version. > > boot device : scsi=2,0 > processor number : 2 > host name : sparc7 > file name : /sd0/vxWorks > inet on ethernet (e) : 129.131.100.102 > host inet (h) : 129.131.100.7 > user (u) : vads > flags (f) : 0x28 > target name )tn) : cpu2 > startup script (s) : /sd0/loadPlay > other (o) : ln I tried it using the standard vxWorks.st-File in the BSP-directory. But in the meantime some changes have been made to the Kernel- configuration. During the BUILDs only vxWorks and vxWorks.sym is created, but not vxWorks.st. I tested it this way: ... [VxWorks Boot]: @ boot device : dc processor number : 0 host name : zfssun7 file name : /tftpboot/vxw-5.3/mv2700-std.st <<<<< nothing else than vxWorks.st inet on ethernet (e) : 129.69.88.200:ffffff00 host inet (h) : 129.69.88.191 user (u) : vxwboot flags (f) : 0x80 target name (tn) : zfsrc5 Attaching network interface dc0... done. Attaching network interface lo0... done. Loading... 720800 Starting at 0x100000... Adding 2300 symbols for standalone. ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] ]]]] (R) ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] ]]]]] ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] ]]]] ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] ]]]] ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] ]]]] ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] ]]]] ]]]] ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] ]]]] ]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System ]]]]]]]]]]]]]]]]]]]]]]]]]]]] ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., 1984-1997 CPU: Motorola MVME2700 - MPC 750. Processor #0. Memory Size: 0x2000000. BSP version 1.1/5. -> pwd value = 1 = 0x1 -> ls Can't open ".". value = -1 = 0xffffffff = icmpstat + 0xffe50113 -> scsiShow ID LUN VendorID ProductID Rev. Type Blocks BlkSize pScsiPhysDev -- --- -------- ---------------- ---- ---- -------- ------- ------------ value = 0 = 0x0 -> scsiAutoConfig(pSysScsiCtrl) value = 0 = 0x0 -> scsiShow ID LUN VendorID ProductID Rev. Type Blocks BlkSize pScsiPhysDev -- --- -------- ---------------- ---- ---- -------- ------- ------------ 1 0 SEAGATE ST32151N 0284 0 4197405 512 0x01ffe6c8 value = 0 = 0x0 -> pSbd0=scsiBlkDevCreate (pSpd10,0,0) undefined symbol: pSpd10 -> pSpd10=scsiPhysDevIdGet(pSysScsiCtrl,1,0) new symbol "pSpd10" added to symbol table. pSpd10 = 0x1ffd360: value = 33547976 = 0x1ffe6c8 -> pSbd0=scsiBlkDevCreate (pSpd10,0,0) pSbd0 = 0x1a9e04: value = 33542904 = 0x1ffd2f8 -> pVolDesc0=dosFsDevInit ("/sd0/", pSbd0, 0) new symbol "pVolDesc0" added to symbol table. pVolDesc0 = 0x1ffc070: value = 33542720 = 0x1ffd240 -> dosFsConfigShow("DISK1:") Bad device name. value = -1 = 0xffffffff = pSpd10 + 0xfe002c9f -> dosFsConfigShow("/sd0/") device name: /sd0/ total number of sectors: 4197405 bytes per sector: 512 media byte: 0xf0 # of sectors per cluster: 65 # of reserved sectors: 1 # of FAT tables: 2 # of sectors per FAT: 253 max # of root dir entries: 112 # of hidden sectors: 0 removable medium: false disk change w/out warning: not enabled auto-sync mode: not enabled long file names: ENABLED exportable file system: not enabled lowercase-only filenames: not enabled volume mode: O_RDWR (read/write) available space: 2147392000 bytes max avail. contig space: 2147392000 bytes value = 0 = 0x0 -> cd "/sd0" cd: error = 0xc0009. value = -1 = 0xffffffff = pSpd10 + 0xfe002c9f -> cd "/sd0/" value = 0 = 0x0 -> ls mv2700.MatlabIPPC_Fast mv2700.MatlabIPPC_Fast.sym startup.rtw.tmpl value = 0 = 0x0 -> ... ==> Zugriff auf existierendes Filesystem klappt! (access to existing filesystem OK!) Also Standalone-Kernel generieren! (vxWorks.st for the needed configuration has to be generated) > > I did not specify long file names (I didn't care; I just wanted it to work). > In fact, I have no experience using long file names with VxWorks. As I > specified in my earlier post, the file name specifies where the files are > coming from. Also, if you specify a file name of "frank", then it expects > the symbol table to be named "frank.sym". Long file names make it a little bit easier. > > I hope this helps. If not, then let me know. There are no guarantees that > I will get to it today, but I can address it again when I get in the office > on Monday, or you can e-mail me at home at hansen@iwvisp.com. The BUILD of the standalone version of my configuration (support to IPPC2636 and so on) makes some trouble because the symbol table isn't generated the right way: What BUILD of vxWorks.st prints out as symptom ... ... zfssun7:windadm:14: windmake vxWorks.st ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -M -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" *.c /usr/local/wind/target/config/all/*.c \ | sed -e "s:/usr/local/wind/target/config/all:\\$\(CONFIG_ALL\):g" \ -e "s:/usr/local/wind/target:\\$\(TGT_DIR\):g" \ -e "s:/usr/local/wind:\\$\(WIND_BASE\):g" > depend.mv2700.IPPC2636 VX_CPU_FAMILY=ppc /usr/local/wind/host/sun4-solaris2/bin/cppppc -M -E -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" romInit.s \ | sed -e "s:/usr/local/wind/target/config/all:\\$\(CONFIG_ALL\):g" \ -e "s:/usr/local/wind/target:\\$\(TGT_DIR\):g" \ -e "s:/usr/local/wind:\\$\(WIND_BASE\):g" >> depend.mv2700.IPPC2636 VX_CPU_FAMILY=ppc /usr/local/wind/host/sun4-solaris2/bin/cppppc -M -E -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" sysALib.s \ | sed -e "s:/usr/local/wind/target/config/all:\\$\(CONFIG_ALL\):g" \ -e "s:/usr/local/wind/target:\\$\(TGT_DIR\):g" \ -e "s:/usr/local/wind:\\$\(WIND_BASE\):g" >> depend.mv2700.IPPC2636 ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" /usr/local/wind/target/config/all/usrConfig.c -o usrConfig.o cp /usr/local/wind/target/config/all/usrConfig.c usrConfig_st.c ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" -DSTANDALONE usrConfig_st.c ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" /usr/local/wind/target/config/all/dataSegPad.c -o dataSegPad.o ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" -o version.o /usr/local/wind/target/config/all/version.c ldppc -o tmp.o -X -r dataSegPad.o sysALib.o sysLib.o \ usrConfig_st.o version.o /usr/local/wind/target/lib/libPPC604gnuvx.a nmppc tmp.o | munch >ctdt.c /usr/local/wind/host/sun4-solaris2/bin/make CC_COMPILER="-traditional" ctdt.o make[1]: Entering directory `/usr/x4/TORNADO_1.0.1/target/config/mv2700.IPPC2636' ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -mstrict-align -traditional -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" -c ctdt.c make[1]: Leaving directory `/usr/x4/TORNADO_1.0.1/target/config/mv2700.IPPC2636' cp tmp.o tmp.2 true tmp.o VX_CPU_FAMILY=ppc /usr/local/wind/host/sun4-solaris2/bin/makeSymTbl tmp.o > symTbl.c ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -traditional -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" symTbl.c symTbl.c:18: nondigits in number and not hexadecimal symTbl.c:18: syntax error before `0001301c' make: *** [vxWorks.st] Error 1 [2] - Done textedit symTbl.c zfssun7:windadm:15: ... ... and the source of trouble: [The beginning of config//symTbl.c] ... /* makeSymTbl.c - stand-alone system symbol table */ /* CREATED BY makeSymTbl * FROM tmp.o * ON Tue Sep 28 16:04:16 MET DST 1999 */ #include "vxWorks.h" #include "symbol.h" #include "a_out.h" #undef READ #undef WRITE IMPORT int NCR810_INSTRUCTIONS; IMPORT int NCR810_PATCHES; 0001301c R WTX_EVT_CALL_GOT_EXC 00013010 R WTX_EVT_CALL_RETURN 00012fb0 R WTX_EVT_CTX_EXIT 00012fa4 R WTX_EVT_CTX_START 00013004 R WTX_EVT_DATA_ACCESS 00012fbc R WTX_EVT_EXCEPTION 00012f88 R WTX_EVT_OBJ_LOADED 00012f94 R WTX_EVT_OBJ_UNLOADED 00012f70 R WTX_EVT_SYM_ADDED 00012f7c R WTX_EVT_SYM_REMOVED 00012ff8 R WTX_EVT_TEXT_ACCESS 00012f64 R WTX_EVT_TGT_RESET 00012fd4 R WTX_EVT_TOOL_ATTACH 00012fe0 R WTX_EVT_TOOL_DETACH 00012fec R WTX_EVT_TOOL_MSG 00013034 R WTX_EVT_UNKNOWN 0001302c R WTX_EVT_USER 00012fc8 R WTX_EVT_VIO_WRITE IMPORT int _Randseed; IMPORT __10Demangle_TPCc (); IMPORT __10RBString_T (); ... What is the reason? Something undefined? Something couldn't been referenced? This is all stuff of the tgtsvr-Connection? > > Good luck, and keep me posted. This is what i'm doing. Perhaps You see it if You're at work. > > Joseph R. Hansen > NAWCWPNS China Lake > Code 471500D > China Lake, CA 93555 > (760) 939-6999 > HansenJR@navair.navy.mil > TIA Friedrich "Fritz" Scheurer ---------------------------------------- Bitte beachten Sie: Im Moment ist kein Reply auf diese Adresse moeglich. Benutzen Sie bitte die unten aufgefuehrten Adressen! ---------------------------------------- +----------------------------------------------------------------+ | Friedrich Scheurer Zentrum Fertigungstechnik Stuttgart | | Dipl.-Ing. Arbeitsbereich Steuerungstechnik | | Tel +49 711 13162-41(-0) Nobelstrasse 15 | | Fax +49 711 13162-11 D-70569 Stuttgart | | E-Mail scheurer@zfs.tz.uni-stuttgart.de (NEU seit Juli 1999) | +----------------------------------------------------------------+ | http://zfs.rus.uni-stuttgart.de | +----------------------------------------------------------------+ | privat: | | Telefon +49 7161 686362 | | Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | +----------------------------------------------------------------+ From owner-vxwexplo-process Tue Sep 28 11:04:28 1999 From: Francis Hallahan Date: Tue Sep 28 11:04:32 PDT 1999 Subject: Re: Help! Changed boot from host to local disk. To boot off of SCSI disk, you must have SCSI included in both your bootrom and your vxWorks image. We boot off of SCSI disks often on our projects. From your posting you do have SCSI support in your bootrom, but you do not have it in the vxWorks image. This >Attaching to scsi device... done. right,left>Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 >Starting at 0x100000... > >Host Name: bootHost >Attaching network interface lo0... done. >Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error >opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x3d Should look like this >Attaching to scsi device... done. right,left>Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 >Starting at 0x100000... > >Host Name: bootHost >Attaching to scsi device... done. right,left>Attaching network interface lo0... done. >Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error >opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x3d NOTE the second "Attaching to scsi device". This is critical. Check the file usrScsi.c in the ../target/src/config directory to see if it does what you want. This is where vxWorks mounts SCSI disks. Also, you can attach to a network after a SCSI boot by placing the network device name in the "Other:" boot parameter. I hope this helps, Frank Hallahan > Attached message: Lots Deleted. >Submitted-by: (Friedrich Scheurer / Goeppingen) > >Hi VxWorkers, > >i want to run my MVME2700 standalone, w/o any host. > >I managed it to get the kernel image to the local SCSI-Disk. >The system seems to start up right. But it doesn' t load the >symbols that i put on the local disk, too. >They're always looked for using the loopback-device. >This would be nice, if the system would find or load the >specified files. >..... >P.S.: >Here's what i get at the boot-monitor/target-shell: >.... >[VxWorks Boot]: @ > >boot device : scsi=1,0 >processor number : 0 >file name : /sd0/mv2700.MatlabIPPC_Fast >inet on ethernet (e) : 129.69.88.200:ffffff00 >user (u) : vxwboot >flags (f) : 0x2 >target name (tn) : zfsrc5 >startup script (s) : /sd0/startup.rtw.tmpl > >Attaching to scsi device... done. >Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 >Starting at 0x100000... > >Host Name: bootHost >Attaching network interface lo0... done. >Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error >opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x3d >... /------------------------------------------------------------------------\ | Francis Hallahan | Hallahan@gsfc.nasa.gov | | Computer Scientist - Code 450.E | fhallaha@csc.com | | Computer Sciences Corporation | | | at Nasa Goddard Space Flight Center | Phone (301) 286 7064 | \------------------------------------------------------------------------/ From vxwexplo-errs Tue Sep 28 12:42:33 1999 From: "Thomas, Justin W." Date: Tue Sep 28 12:42:37 PDT 1999 Subject: vme config prob? vxWorkers, I have a strange little problem that I suspect may be a VxWorks VME configuration problem. My system has: 1 Radstone PPC603E VME board. 1 Vista A/D VME board. I am getting a strange problem where 1/2 of the analog channels work perfectly, but the other half show occasional glitches in the output value. Channels 0, 2, 4, 6... have problems. Channels 1, 3, 5, 7... work great. The board was tested at the vendor and they claim everything worked fine, which leaves me to believe that somehow, I am not accessing the memory correctly on the VME board. Unfortunately, I don't have another conduction cooled VME board that I can test, so I have to try to troubleshoot with this one. The documentation with the A/D board says to use A16:D16. Address Modifier: 0x2D or 0x29. Board memory offset of 0x4800. The channels are 16 bit each and spaced 4 bytes apart , but NOT on the Long Word Breaks (see below). Channel Location 0 0x02 1 0x06 2 0x0A etc... So, I use something like this..... ................ INT16 pRawData; INT16 rawdata[MaxChans]; int i; /* get local address */ sysBusToLocalAdrs(VME_AM_SUP_SHORT_IO,(char *)0x4802,(char**)&pRawData) /* loop through to get all channels */ for (i = 0;i Date: Tue Sep 28 15:42:27 PDT 1999 Subject: network device driver I want to build a vxworks network device driver to support a Mil-Std 1553B card. what is the specification for network device drivers or where can it be found. My intent is to use normal socket calls to interface to a funny network card. _____________________________________________ Jim Dodgen 310.814.2529 TRW Inc. 201A / 6146 One Space Park Redondo Beach CA 90278 http://www.astrolink.com mailto:Jim.Dodgen@TRW.com From vxwexplo-errs Tue Sep 28 15:48:32 1999 From: Olivier.LAFONTAINE@st.com Date: Tue Sep 28 15:48:37 PDT 1999 Subject: looking for compression tools As I have just received the whole TORNADO package, I would like to develop some software running on an embedded board. Does someone know and use some library or tool, PSOS dependant, which can easily (un)compress some code in an embedded memory? If it is the case, thanks to reply to this mail. Regards. From vxwexplo-errs Tue Sep 28 23:33:32 1999 From: francois pertin Date: Tue Sep 28 23:33:36 PDT 1999 Subject: USB for tornado 2 Hi dear VXWORKS and tornado2 users, I'm working on a new project were I need to communicate with a device plugged on USB channel. Does anybody have already worked on a driver for UHCI chip or have any kind of interesting information on the subject ??? Please send it to me, it will be very helpful. Thanks a lot Francois From vxwexplo-errs Wed Sep 29 03:33:18 1999 From: Roger Holmquist Date: Wed Sep 29 03:33:22 PDT 1999 Subject: RTX vs Seaweed vs ??? Hello fellow wxWorkers! When looking through old UserGroup-mail and thinking about the behaviour from WindRiver concerning RTX (no working version for Tornado II so far...), it seems that RTX is a dead product. So, the question now is: With what should it be replaced? Does anybody have any experience in Seaweeds X- and Motif-products? Is there any other alternative? (For us the client-part is the main issue) Greetings / Roger - # -----------------------------------------------------------------# # Roger Holmquist, FOA, National Defence Research Establishment # # sect 750, Olaus Magnus vag 42, Box 1165, 581 11 LINKOPING ------# # --------------------------- SWEDEN ------------------------------# # Phone: +46 13-378382 --------------------------------------------# # Fax: +46 13-378009 ----------------------------------------------# # Home: +46-142-20542 Mobile +46-706-250123 ----------------------# # Email: roghol@lin.foa.se ----------------------------------------# # Home address: Hovslagargrand 5, 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 vxwexplo-errs Wed Sep 29 04:00:14 1999 From: daemon@csg.lbl.gov Date: Wed Sep 29 04:00:19 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Sep 29 04:00:10 PDT 1999 Subject: [Q] About taskInit() function Subject: MMU Initialisation Subject: Re: Disappointed with Tornado II. Subject: ipDetach Subject: vxVMI capabilities Subject: Re: GNU Toolchain usage in vxWorks: FORTRAN? ADA? Subject: Re: Linking object files from other os Subject: Re: ICE for 80486 running VxWorks Subject: Re: [Q] About taskInit() function Subject: Re: Help! Changed boot from host to local disk. Subject: Re: Simple question on file permissions Subject: Re: Network I/O problems Subject: Question: What VxWorks libraries to use for PowerPC 401 ?? Subject: Re: Linking object files from other os Subject: Re: MMU Initialisation Subject: Re: looking for compression tools Subject: Re: MMU Initialisation Subject: Re: Memory partitions Subject: ftp server need to change default 8Kb/s to higher Subject: Re: Memory partitions Subject: Re: Variable file loading times Subject: Boot from serial device Subject: Boot from serial ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: [Q] About taskInit() function Date: Tue, 28 Sep 1999 16:31:29 +0900 From: "Hyung-Taek Lim" Message-ID: With taskInit() function, we can create a task's stack and tcb on a specified location in memory. The first argument of taskInit() is the address of new task's TCB and the fifth one is the address of new tasks's stack. Does taskInit() allocate memory space for tcb and stack to where the first and fifth arguments point? If so, the only thing that I have to do before calling taskInit() is finding the location of free spaces for stack and tcb. Otherwise, I have to allocate memory space for stack and tcb before calling taskInit(). Should I pass a pointer to an allocated memory or a pointer to a free memory space as the arguments of taskInit()? --------------------------- Newsgroups: comp.os.vxworks Subject: MMU Initialisation Date: Tue, 28 Sep 1999 12:38:48 +0200 From: "Borkhuis, Johan" Organization: UUNET-NL (http://www.nl.uu.net) Message-ID: References: <7snikp$a44$1@news.mch.sbs.de> <37F056C4.10EF6B9C@aracnet.com> > ---------- > From: engineer_scotty[SMTP:sj@aracnet.com] > Posted At: dinsdag 28 september 1999 6:48 > Posted To: vxworks > Conversation: MMU Initialisation > Subject: Re: MMU Initialisation > > Richard Marsh wrote: > > > > Hi, > > > > We are coming to the end of our project and we want to turn on > hardware > > watchdogs on the PPC860, BUT we get a reset in the usrMmuInit > function. The > > watchdog needs to be reset every 2.4 seconds and the function takes > >9 > > seconds - hence kaboom. We therefore need to manually initialise the > MMU > > does anybody have any experience of this? If we could use the same > mechanism > > as usrMmuInit but with a watchdog reset after each block that would > > surfice - probably. > > 9 seconds for MMU init. You must be mapping in a gigabyte or so of > PCI > address space, right? > > One thing that might help (if that's the case), would be to go out and > figure > out the ACTUAL size of your PCI addr space, modify sysPhysMemDesc > appropriately, > and only map THAT. Will save lots of RAM too (as 1GB of address space > requires > 256K page table entries if 4K pages are used--that's a megabyte, if my > math and > my memory are correct.) > The following info is from a SPR from WRS. I don't have the number and the SPR was for Tornado 1.0.1, but the keywords are: 860 PowerPC MEMORY FRAGMENTATION FRAG The default value for PCI_MSTR_IO_SIZE is 512 Mb (0x20000000). This can be changed to a much more realistic value, for example 1Mb. You can find more info on this in the MBX860 BSP part of my VxWorks page (http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html). Groeten Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Disappointed with Tornado II. Date: Tue, 28 Sep 1999 14:20:40 GMT From: rbose@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <7sqira$71d$1@nnrp1.deja.com> References: <7rp41k$3ot$1@nnrp1.deja.com> Here's an update on the gdb problems we were having with large images. Apparently the T1.0.1 gdb was optimized to read debug information when it was actually needed. This setting was removed in T2 to allow duplicate symbols to be resolved the same way as in the target server, thereby ensuring proper operation when an object module was re-downloaded from the shell. This unfortunately made debugging impossible in our environment. We reverted back to the old behavior by adding the line "gdb set symbol-readnow off" in the gdb.tcl file (.gdbinit on Unix). Things are much better now and we can debug again. (The short absence of the debugger reminded us of how people used to debug in the old days with printf's). I personally think that lazy reading of debug info should be the default behavior for gdb and folks who need to re-download .o's should explicitly change the behavior. I take back some of the stuff I said about WRS support. The FSE organization in the Northeast came up with the right technical solution and saved us from a lot of grief. Rahul Bose 3Com Voice Solutions. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: ipDetach Date: Tue, 28 Sep 1999 16:20:58 +0100 From: "Mark Hambleton" Organization: (Posted via) Netcom Internet Ltd. Message-ID: <7sqm81$16c$1@taliesin2.netcom.net.uk> Hi, I am trying to work out how to unload an END from the MUX that has IP attached to it, there is an ipDetach() function in the symbol table but I have no idea of the prototype. Do I have to do anything else like removing WDB and ARP, and if so how do I do that? - -- Mark Hambleton http://www.xstreamis.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: vxVMI capabilities Date: Tue, 28 Sep 1999 17:52:25 +0100 From: Dave Moore Message-ID: <37F0F248.BE3A1C3E@gecm.com> Hi everyone, I've got a query about the VxVMI virtual memory vxWorks option - I wonder whether anybody here has any experience with it. Question 1: I understand that virtual memory contexts could be allocated on a per task basis to ensure that one task cannot overwrite the data of another. However, am I right in thinking that this only applies to memory dynamically allocated by using memPartAlloc (as demonstrated in the example within the Programmer's guide) and not to memory created statically by, say by the downloading of code containing static/global variables for example?. Question 2: Although data can be protected on a per task basis by using a task private virtual memory context, can code be placed within this memory?. ie. to stop one task calling the functions of another?. Question 3: I can understand that memPartAlloc could be used to create memory for data but if you're using C++ and therefore using the new operator to create object instances then how could these be created within the virtual memory defined by a private context?. Any comments or suggestions gratefully received. Thanks in advance, Dave Moore. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GNU Toolchain usage in vxWorks: FORTRAN? ADA? Date: Tue, 28 Sep 1999 17:24:53 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <7sqtkl$fla$1@nnrp1.deja.com> References: <7sok9k$pk8$1@autumn.news.rcn.net> In article <7sok9k$pk8$1@autumn.news.rcn.net>, "Mike Campbell" wrote: > I'm looking for anyone who has used the GNU ADA / Fortran toolchain in > developing for vxWorks. In theory, using gnat and the gfc (think > that's it) compiler should work like using the GNU C compiler, since > they are just branches off the same toolchain tree. I'd be interested > in hearing the pros/ cons of attempting this type of development, as > well as any lessons learned. I don't know who supports gnu Fortran, but you should really get hold of ACT about the Gnu Ada compiler (support@gnat.com or (212) 620 7300 ext. 117) if you have questions like these. I seriously doubt anyone here has anywhere near the experience with those issues that they do. Also, I suspect they could answer a lot of your questions about using gnu Fortran in the same environment with Ada. What I do know is that they don't currently have supported versions for *all* vxWorks targets, and there doesn't seem to be a publicly available precompiled binary of Gnat for vxworks. Also, you may have issues with Gnat using a different version of GCC than vxWorks does. If you are willing to buy support from ACT, they can probably take care of those issues for you. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking object files from other os Date: Tue, 28 Sep 1999 17:40:14 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <7squhr$ggp$1@nnrp1.deja.com> References: <01f8d464.615a7403@usw-ex0102-013.remarq.com> <7so3af$fc3$1@nnrp1.deja.com> <02f42b85.69c7e14c@usw-ex0102-013.remarq.com> In article <02f42b85.69c7e14c@usw-ex0102-013.remarq.com>, Tiept wrote: > need to look it up. I am currently using the free GNAT > 3.11p version available on the net. I guess I will have to There are several free 3.11p distributions available for several different OS's. Which one are you using? There isn't a free 3.11p binary for vxWorks that I'm aware of. This isn't unreasonable, as each host OS/Target CPU combination would require a its own distribution. I suspect you are trying to use a self-hosted non-vxWorks distribution (eg: gnat-3.11p-nt.exe). Of course that's that won't work. You might as well try to run an executable compiled with Microsoft Visual C++! You'll either have to get the source code for Gnat and configure it for your Host/Target setup, or you will need to get someone like ACT ( http://www.gnat.com ) to do it for you. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: ICE for 80486 running VxWorks Date: Tue, 28 Sep 1999 18:03:10 GMT From: David Hamel Organization: @Home Networks. Message-ID: <37F102FE.F31385A8@ntanet.net> References: If you really really suspect stack corruption then using an ICE seems like overkill. Just step you application until it crashes and bump up the stack sizes of your tasks. You may spend more time learning how to use your newly purchased ICE than actually getting work done. Richard Tseng wrote: > > We are running into a particularly nasty bug where our VxWorks > application will crash killing the target server leaving us with no > debug information. We suspect stack corruption and would like to > purchase an ICE to find out what our app and VxWorks were doing prior to > the crash. We have so far evaluated Microtech PowerPack and Applied > Microsystems Corp SuperTAP. Both can do the job but quite expensive. The > main feature we were looking for is history trace and triggering on > error conditions. It will help if there is source level debugging and > RTOS integration. > > Could anyone recommend a lower price ICE for 486 ? > Does anyone have experience using Microtech or AMC ICE's with VxWorks ? > > Thanks, > Richard Tseng > We Software Ltd --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Q] About taskInit() function Date: Tue, 28 Sep 1999 18:06:08 GMT From: David Hamel Organization: @Home Networks. Message-ID: <37F103B0.6412D040@ntanet.net> References: taskInit is very hard to use properly and for most people that see the existance of the function taskInit really want to use the undocumented function taskCreat. Hyung-Taek Lim wrote: > > With taskInit() function, we can create a task's stack and tcb on a > specified location in memory. > > The first argument of taskInit() is the address of new task's TCB and the > fifth one is the address of new tasks's stack. > > Does taskInit() allocate memory space for tcb and stack to where the first > and fifth arguments point? > If so, the only thing that I have to do before calling taskInit() is > finding the location of free spaces for stack and tcb. > Otherwise, I have to allocate memory space for stack and tcb before calling > taskInit(). > > Should I pass a pointer to an allocated memory or a pointer to a free memory > space as the arguments of taskInit()? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help! Changed boot from host to local disk. Date: Tue, 28 Sep 1999 18:14:48 GMT From: NO-SPAM-xad@NO-SPAM-home.com (Ron Dempster) Organization: @Home Network Message-ID: <37f103c5.445112347@news> References: <7sgijd$12r1@overload.lbl.gov> My guess is that you need to implement sysScsiConfig in your sysScsi.c file. Check out usrScsi.c in target/src/config for an example. The problem is that after vxWorks loads the /sd0/ device no longer exists unless you define SYS_SCSI_CONFIG and implement a sysScsiConfig routine. The example shows more than you need, but you can prune it to do the scsiPhysDevCreate, scsiBlkDevCreate, and dosFsDevInit to make /sd0/ avaiable to load your symblo table. Hope this helps, Ron On Tue, 28 Sep 1999 09:33:57 +0200, "Borkhuis, Johan" wrote: > >> ---------- >> From: vxwexplo@lbl.gov[SMTP:vxwexplo@lbl.gov] on behalf of >> Friedrich.Scheurer@t-online.de (Friedrich Scheurer / >> Goeppingen)[SMTP:Friedrich.Scheurer@t-online.de] >> Posted At: vrijdag 24 september 1999 12:44 >> Posted To: vxworks >> Conversation: Help! Changed boot from host to local disk. >> Subject: Help! Changed boot from host to local disk. >> >> VxWorks, Tornado >> >> Hi VxWorkers, >> >> i want to run my MVME2700 standalone, w/o any host. >> >> I managed it to get the kernel image to the local SCSI-Disk. >> The system seems to start up right. But it doesn' t load the >> symbols that i put on the local disk, too. >> >> Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error >> opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x3d >> >Are long filenames enabled on this disk? Can you do a ls (or an ll) of >this disk? > >Groeten > Johan > >-- > o o o o o o o . . . ___________________________________ > o _____ || Johan Borkhuis | > .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | > >(________|__|_[_________]_|________________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` >=== VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html >=== > > >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] >> ]]]] (R) >> ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] >> ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] >> ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] >> ]]]]] >> ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] >> ]]]] >> ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] >> ]]]] >> ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] >> ]]]] >> ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] >> ]]]] ]]]] >> ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] >> ]]]] >> ]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 >> ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 >> ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., >> 1984-1997 >> >> CPU: Motorola MVME2700 - MPC 750. >> Processor #0. >> Memory Size: 0x2000000. BSP version >> 1.1/5. >> WDB: Ready. >> >> Unable to open startup script /sd0/startup.rtw.tmpl >> -> >> ... >> 2. >> [VxWorks Boot]: @ >> >> boot device : scsi=1,0 >> processor number : 0 >> file name : /mv2700.MatlabIPPC_Fast >> inet on ethernet (e) : 129.69.88.200:ffffff00 >> flags (f) : 0x2 >> target name (tn) : zfsrc5 >> startup script (s) : /startup.rtw.tmpl >> >> Attaching to scsi device... done. >> Loading /mv2700.MatlabIPPC_Fast...1000180 >> Starting at 0x100000... >> >> Host Name: bootHost >> User: target >> Attaching network interface lo0... done. >> Loading symbol table from /mv2700.MatlabIPPC_Fast.sym ...Error opening >> /mv2700.MatlabIPPC_Fast.sym: status = 0x3d >> >> >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] >> ]]]] (R) >> ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] >> ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] >> ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] >> ]]]]] >> ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] >> ]]]] >> ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] >> ]]]] >> ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] >> ]]]] >> ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] >> ]]]] ]]]] >> ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] >> ]]]] >> ]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 >> ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 >> ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., >> 1984-1997 >> >> CPU: Motorola MVME2700 - MPC 750. >> Processor #0. >> Memory Size: 0x2000000. BSP version >> 1.1/5. >> WDB: Ready. >> >> Unable to open startup script /startup.rtw.tmpl >> -> >> >> ... >> 3. >> [VxWorks Boot]: @ >> >> boot device : scsi=1,0 >> processor number : 0 >> host name : zfsrc5 >> file name : /sd0/mv2700.MatlabIPPC_Fast >> inet on ethernet (e) : 129.69.88.200:ffffff00 >> host inet (h) : 129.69.88.200 >> user (u) : vxwboot >> flags (f) : 0x2 >> target name (tn) : zfsrc5 >> startup script (s) : /sd0/startup.rtw.tmpl >> >> Attaching to scsi device... done. >> Loading /sd0/mv2700.MatlabIPPC_Fast...1000180 >> Starting at 0x100000... >> >> Attaching network interface lo0... done. >> Loading symbol table from /sd0/mv2700.MatlabIPPC_Fast.sym ...Error >> opening /sd0/mv2700.MatlabIPPC_Fast.sym: status = 0x41 >> >> >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]] ]]]] ]]]]]]]]]] ]] >> ]]]] (R) >> ] ]]]]]]]]] ]]]]]] ]]]]]]]] ]] ]]]] >> ]] ]]]]]]] ]]]]]]]] ]]]]]] ] ]] ]]]] >> ]]] ]]]]] ] ]]] ] ]]]] ]]] ]]]]]]]]] ]]]] ]] ]]]] ]] >> ]]]]] >> ]]]] ]]] ]] ] ]]] ]] ]]]]] ]]]]]] ]] ]]]]]]] ]]]] ]] >> ]]]] >> ]]]]] ] ]]]] ]]]]] ]]]]]]]] ]]]] ]] ]]]] ]]]]]]] >> ]]]] >> ]]]]]] ]]]]] ]]]]]] ] ]]]]] ]]]] ]] ]]]] ]]]]]]]] >> ]]]] >> ]]]]]]] ]]]]] ] ]]]]]] ] ]]] ]]]] ]] ]]]] ]]]] >> ]]]] ]]]] >> ]]]]]]]] ]]]]] ]]] ]]]]]]] ] ]]]]]]] ]]]] ]]]] >> ]]]] >> ]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]]] Development System >> ]]]]]]]]]]]]]]]]]]]]]]]]]]]] >> ]]]]]]]]]]]]]]]]]]]]]]]]]]] VxWorks version 5.3.1 >> ]]]]]]]]]]]]]]]]]]]]]]]]]] KERNEL: WIND version 2.5 >> ]]]]]]]]]]]]]]]]]]]]]]]]] Copyright Wind River Systems, Inc., >> 1984-1997 >> >> CPU: Motorola MVME2700 - MPC 750. >> Processor #0. >> Memory Size: 0x2000000. BSP version >> 1.1/5. >> WDB: Ready. >> >> Unable to open startup script /sd0/startup.rtw.tmpl >> -> >> >> ... >> >> -- >> +-------------------------------------------------+ >> | Dipl.-Ing. F. Scheurer | >> | c/o Zentrum Fertigungstechnik Stuttgart | >> | Nobelstrasse 15 | >> | D-70569 Stuttgart | >> | Telefon +49 711 13162-41(-0) | >> | Fax +49 711 13162-11 | >> | E-Mail scheurer@zfs.tz.uni-stuttgart.de | >> | WWW http://zfs.rus.uni-stuttgart.de | >> +-------------------------------------------------+ >> | privat: Waldeckstrasse 31 | >> | D-73035 Goeppingen | >> | Telefon +49 7161 686362 Fax +49 7161 686301 | >> | E-Mail Friedrich.Scheurer@t-online.de | >> +-------------------------------------------------+ >> >> > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Simple question on file permissions Date: Tue, 28 Sep 1999 19:30:09 GMT From: Kremer@wxs.nl (Harro Kremer) Organization: Planet Internet Message-ID: <37f11695.183316784@news.wxs.nl> References: <37F04A00.965AE5EA@teja.com> Marco Zandonadi wrote: You don't state the file system, my answers below pertain to the tffs with the dos-fs on top of it. About changing permissions: this is done using an ioctl function call (see documentation of file i/o) About directory permissions: the dos-fs doesn't support them. hence, You cannot protect a directory from being modified. Reason: good-old MS-DOS didn't support it either. Harro >Hello, >I have the following problem on VXWorks (Tornado 2.0 on Solaris). I >wrote a program that creates a directory using mkdir(). On Solaris and >Linux mkdir() has two arguments: name and permissions. On VXWorks it >only has name and the created directory has permissions 000. > >So I looked through the documentation for something to change file >permissions but I couldn't find anything (fchmod() isn't there). Does >anybody know how to fix this? > >Please reply via e-mail because I cannot check this group regularly. > >Thanks in advance for your help! > >Marco Zandonadi > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Network I/O problems Date: Tue, 28 Sep 1999 19:38:57 GMT From: Charlie Grames Organization: Boeing Message-ID: <37F11951.67A8B420@nowhere.com> References: <242-37EBE152-67@newsd-222.iap.bryant.webtv.net> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Greg, How is your system configured? Do you have all of your targets in a single VME chassis? Do the targets boot via shared memory? We had serious problems booting more than six MVME2604 boards in a single chassis. It turned out to be a combination of the targets polling across the backplane while attempting to boot and a less-than-perfect backoff algorithm in the shared memory (smLib) code. We corrected both problems and are now running fine. HTH Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com Greg Silverman wrote: > > Our system consists of 3 Motorola processors running VxWorks 5.3.1, a > cluster of 10 Mercury processors, and a Solaris 2.5.1 workstation. At > bootup we do the following steps: 1) download the VxWorks images, 2) > configure the Mercury cluster from the master Motorola's bootup script, > and, then, 3) start a program on the master Motorola that downloads and > starts all of the application programs. Too often the system fails > during the last step, with an access violation or a file I/O error in > one of the Vxworks tasks. To test the system software I replaced the > last step of the bootup procedure with a VME reset, so that the system > was in a "boot loop", e.g., no application code is run. If there are no > problems with the VxWork/Solaris/Mercury installations and interfaces > this boot loop should go on forever. But, every 20 or so boots, one of > the VxWorks cards is unable to load some files from the Sun. I have > seen: 1) it can not load the VxWorks image, 2) it could not read the > VxWorks startup script, 3) the startup script could not load one of the > code files. This seems to fail way too often and indicates some kind of > network I/O problem that may explain our problem with bringing up the > applications. Any ideas? > > Greg Silverman > TRW > Sunnyvale, Ca --------------------------- Newsgroups: comp.os.vxworks Subject: Question: What VxWorks libraries to use for PowerPC 401 ?? Date: 28 Sep 1999 17:57:03 -0500 From: "John Pulera" Organization: Newscene Public Access Usenet News Service (http://www.newscene.com/) Message-ID: We're using VxWorks on a target that has a Power PC401 core. We've been using the gnu-built VxWorks library distributed with the Tornado 1.0.1 CD ("libPPC403gnuvx.a") but have been fielding invalid instruction exceptions (probably due to miniscule differences between the 403 and 401). Has anyone ever used VxWorks with a power PC 401 based target? Thanks, John Pulera jpulera@minolta-mil.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Linking object files from other os Date: Tue, 28 Sep 1999 19:23:00 +1700 From: Tiept Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <0a9862bb.18582f14@usw-ex0102-013.remarq.com> References: <01f8d464.615a7403@usw-ex0102-013.remarq.com> <7so3af$fc3$1@nnrp1.deja.com> <02f42b85.69c7e14c@usw-ex0102-013.remarq.com> <7squhr$ggp$1@nnrp1.deja.com> I looked around and found this site that has something on gnat and vxworks. The url, if anyone's interested is: http://www.unm.edu/~krisinsk/gnu/ I am still looking at it at this point. Thanks again, TNT * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MMU Initialisation Date: Tue, 28 Sep 1999 22:39:42 -0500 From: Pierre Olivier Organization: Netcom Canada Message-ID: <37F189FE.DA835D6E@netcom.ca> References: <7snikp$a44$1@news.mch.sbs.de> Reply-To: polivier@netcom.ca This is a multi-part message in MIME format. - --------------E78370AB1FFF60B0C09CCAEF Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I wrote my own MMU init system, but for standalone operation (No OS support). Having played in non-distribution WindRiver source code, I can tell you the problem is that they don't support pages bigger than 4K (which is a drag, because it's not *that* hard to implement, really) However, I doubt you can easily rip out VxWorks MMU init system and replace it with your own.... If memory serves, usrMmuInit() is called when VxWorks is not multitasking yet. So what you could do, if the interrupt service routines are already up (not sure either, I'm writing this from home...), is to write a simple PIT interrupt routine that would be called every, say, 1 or 2 seconds and feed the watchdog. Pierre Olivier SR Telecom Richard Marsh wrote: > Hi, > > We are coming to the end of our project and we want to turn on hardware > watchdogs on the PPC860, BUT we get a reset in the usrMmuInit function. The > watchdog needs to be reset every 2.4 seconds and the function takes >9 > seconds - hence kaboom. We therefore need to manually initialise the MMU > does anybody have any experience of this? If we could use the same mechanism > as usrMmuInit but with a watchdog reset after each block that would > surfice - probably. > > Help much appreciated, > > Richard Marsh > Siemens Atea - Belgium - --------------E78370AB1FFF60B0C09CCAEF Content-Type: text/x-vcard; charset=us-ascii; name="polivier.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Pierre Olivier Content-Disposition: attachment; filename="polivier.vcf" begin:vcard n:Olivier;Pierre x-mozilla-html:FALSE adr:;;;;;; version:2.1 email;internet:polivier@netcom.ca fn:Pierre Olivier end:vcard - --------------E78370AB1FFF60B0C09CCAEF-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: looking for compression tools Date: Tue, 28 Sep 1999 23:21:12 -0500 From: petekz@airmail.net (Pete Kockritz) Organization: Airnews.net! at Internet America Message-ID: References: In article , Olivier.LAFONTAINE@st.com wrote: > > As I have just received the whole TORNADO package, I would like to > develop some software running on an embedded board. Does someone know > and use some library or tool, PSOS dependant, which can easily > (un)compress some code in an embedded memory? > > > If it is the case, thanks to reply to this mail. > > Regards. There is a freely available compression library called 'zlib' that may do what you want. It will compile for Unix and MS-DOS (and other platforms). Wind River includes in their Tornado package. The 'deflate' utility runs on the host platform (Unix or NT) and they have 'inflateLib' in their vxWorks library. It's used to build the compressed kernels. I'm not sure what you mean by 'PSOS dependant', it should be possible to port to pSOS as well. In fact, in the near future I may be doing that (if I can't find it already ported). A review of zlib: The zlib home page: Regards, Pete - -- +-----------------------------------------------------------------+ | Pete Kockritz mailto:petekz@airmail.net | | This space for rent. | +-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MMU Initialisation Date: Tue, 28 Sep 1999 22:10:21 -0700 From: engineer_scotty Organization: Aracnet Internet Message-ID: <37F19F3D.60222F42@aracnet.com> References: <7snikp$a44$1@news.mch.sbs.de> <37F056C4.10EF6B9C@aracnet.com> "Borkhuis, Johan" wrote: > > > ---------- > > From: engineer_scotty[SMTP:sj@aracnet.com] > > Posted At: dinsdag 28 september 1999 6:48 > > Posted To: vxworks > > Conversation: MMU Initialisation > > Subject: Re: MMU Initialisation > > > > Richard Marsh wrote: > > > > > > Hi, > > > > > > We are coming to the end of our project and we want to turn on > > hardware > > > watchdogs on the PPC860, BUT we get a reset in the usrMmuInit > > function. The > > > watchdog needs to be reset every 2.4 seconds and the function takes > > >9 > > > seconds - hence kaboom. We therefore need to manually initialise the > > MMU > > > does anybody have any experience of this? If we could use the same > > mechanism > > > as usrMmuInit but with a watchdog reset after each block that would > > > surfice - probably. > > > > 9 seconds for MMU init. You must be mapping in a gigabyte or so of > > PCI > > address space, right? > > > > One thing that might help (if that's the case), would be to go out and > > figure > > out the ACTUAL size of your PCI addr space, modify sysPhysMemDesc > > appropriately, > > and only map THAT. Will save lots of RAM too (as 1GB of address space > > requires > > 256K page table entries if 4K pages are used--that's a megabyte, if my > > math and > > my memory are correct.) > > > The following info is from a SPR from WRS. I don't have the number and > the SPR was for Tornado 1.0.1, but the keywords are: > 860 PowerPC MEMORY FRAGMENTATION FRAG > > The default value for PCI_MSTR_IO_SIZE is 512 Mb (0x20000000). This can > be changed to a much more realistic value, for example 1Mb. In my case, I have to deal with an unknown (at build time) mix of boards; some of which DO chew up large amounts of address space (we have one, a multiple SHARC 21062 design, that eats up 256Mb). So changing that macro doesn't suffice. (Thanks, though...) As our board uses a PLX9054 bridge rather than a Tundra QSPAN (what the MBX uses), we ended having to replace pciIomapLib.c anyway. > You can find more info on this in the MBX860 BSP part of my VxWorks page > (http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html). > > Groeten > Johan > > -- > o o o o o o o . . . ___________________________________ > o _____ || Johan Borkhuis | > .][__n_n_|DD[ ====_____ | johan.borkhuis@peektraffic.nl | > >(________|__|_[_________]_|________________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` > === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html > === - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory partitions Date: Tue, 28 Sep 1999 22:13:27 -0700 From: engineer_scotty Organization: Aracnet Internet Message-ID: <37F19FF7.399F7CFA@aracnet.com> References: <37F098E9.E2BD7A7E@gecm.com> Dave Moore wrote: > > Hello everybody, > Does anybody know whether there's a way to completely isolate tasks > within vxWorks?. When I mean isolate, I mean in terms of memory, so that > the failure of one task does not affect (via memory corruption) > another?. The sort of thing I'm after is like a UNIX process model where > there are mulitple name and memory spaces. With VX_VMI, you can do limited amounts of memory space protection, but not much more. It does NOT give you separate symbol spaces. If you really need this sort of thing, you should consider a different operating system; LynxOS is a realtime Unix derivative that supports a fully protected memory model. Haven't used it, but have heard mostly good things about it. > Thanks in advance, > Dave Moore - -- - ------------------------------------------------------------------------ engineer_scotty (no, not that one) -- consumer of fine ales and lagers some days you're the fire hydrant, some days you're the dog | go blazers no small furry creatures were harmed in the creation of this .signature - ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: ftp server need to change default 8Kb/s to higher Date: Wed, 29 Sep 1999 00:35:28 -0700 From: Pradeep Pandey Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <37F1C140.CAA18CAA@cisco.com> hi folks, the ftp server on a vxworks target seems to have a default upload rate of 8Kb/s ( ie when doing a get on the ftp server ). is it possible to increase this rate to something higher. i tried setting the global tcp_sendspace/tcp_recvspace to 40K however, i still get the same rate. should these variable be set before the ftpdTask gets spawned ? thanks in advance for the help. - -- regards, Pradeep Dev Pandey Work: (408) 525 8332 http://www.cisco.com - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory partitions Date: Wed, 29 Sep 1999 10:11:54 +0100 From: Dave Moore Message-ID: <37F1D7DA.B2895607@gecm.com> References: <37F098E9.E2BD7A7E@gecm.com> <37F19FF7.399F7CFA@aracnet.com> Thanks engineer_scotty - that's what I thought but I heard a rumour that Wind River were going to bring out a vxWork options that gave you 'protected domains' which did this kind of thing. Cheers, Dave Moore engineer_scotty wrote: > Dave Moore wrote: > > > > Hello everybody, > > Does anybody know whether there's a way to completely isolate tasks > > within vxWorks?. When I mean isolate, I mean in terms of memory, so that > > the failure of one task does not affect (via memory corruption) > > another?. The sort of thing I'm after is like a UNIX process model where > > there are mulitple name and memory spaces. > > With VX_VMI, you can do limited amounts of memory space protection, but > not much more. It does NOT give you separate symbol spaces. > > If you really need this sort of thing, you should consider a different > operating system; LynxOS is a realtime Unix derivative that supports > a fully protected memory model. Haven't used it, but have heard mostly > good things about it. > > > Thanks in advance, > > Dave Moore > > -- > ------------------------------------------------------------------------ > engineer_scotty (no, not that one) -- consumer of fine ales and lagers > some days you're the fire hydrant, some days you're the dog | go blazers > no small furry creatures were harmed in the creation of this .signature > ------------------------------------------------------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Variable file loading times Date: Wed, 29 Sep 1999 10:15:12 +0100 From: Dave Moore Message-ID: <37F1D8A0.DBA5FE14@gecm.com> References: <37EF7A31.C252449E@gecm.com> <7sqsmf$f11$1@nnrp1.deja.com> Thanks for comments Ted, we're loading stuff over the network from the host using standard fopen. Using the 'snoop' facility on the SUN, I can see that it (fopen) uses FTP underneath. Maybe this delay is due to TCP/IP re-try periods if the network experiences collisions?. Cheers, Dave Moore Ted Dennison wrote: > In article <37EF7A31.C252449E@gecm.com>, > Dve Moore wrote: > > When I continually open files, I find that the first few open > > relatively quickly. Then the next load take ages (like 20 seconds or > > What type of filesystem are those files mounted on? If it is retrieving > bits over the network (eg: NFS, FTP), then this would not be at all > unexpected. > > -- > T.E.D. > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Boot from serial device Date: 29 Sep 1999 09:25:43 GMT From: "samuel zhou" Organization: The news server of Changzhou TeleCom Message-ID: <01bed762$8906e590$0300a8c0@samuel_zhou> Hi, folks, We are developping a BSP for a board using i960RX. Host side is WinNT 4.0. We have two separate serial ports on the board. One is used as the console port, the other is one supposed to act as the debugging channel between the host and the target. The question is, how can we download the vxworks kernel from host through this serial port? We know that we can boot from FTP with devices such as ene, dc, ln etc. if we have network interface. But we can not find the detailed directions for booting from serial in vxworks user's manual. It only indicates that we can not configure to boot from FTP server as usual. Do we have to run PPP between host and target in this case? All we can just have some simple way to take? Thanks in advance. - -Samuel --------------------------- Newsgroups: comp.os.vxworks Subject: Boot from serial Date: 29 Sep 1999 09:25:44 GMT From: "samuel zhou" Organization: The news server of Changzhou TeleCom Message-ID: <01bed764$4fbe3700$0300a8c0@samuel_zhou> Hello, We are developping BSP for a i960RX board. We have two serial ports on the board. One is used as the console, the other is used to communicate with the WinNT 4.0 host. The question is, how can we download the vxworks kernel from the host as we do with FTP when network interface is in place? Do we have to run PPP and install the network stuff in the bootrom? We just can not find the directions in the User's manual. Thanks in advance. - - Samuel --------------------------- End of New-News digest ********************** From vxwexplo-errs Wed Sep 29 06:08:48 1999 From: "F. Scheurer" Date: Wed Sep 29 06:08:52 PDT 1999 Subject: Problems creating a standalone-System vxworks, tornado Hi all! i just posted some problems creating a standalone image and booting from a local SCSI-disk. The most recent problem is now that something fails during creation of the symbol table when building the vxWorks.st. What make prints out is: > zfssun7:windadm:14: windmake vxWorks.st > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -M -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" *.c /usr/local/wind/target/config/all/*.c \ > | sed -e "s:/usr/local/wind/target/config/all:\\$\(CONFIG_ALL\):g" \ > -e "s:/usr/local/wind/target:\\$\(TGT_DIR\):g" \ > -e "s:/usr/local/wind:\\$\(WIND_BASE\):g" > depend.mv2700.IPPC2636 > VX_CPU_FAMILY=ppc /usr/local/wind/host/sun4-solaris2/bin/cppppc -M -E -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" romInit.s \ > | sed -e "s:/usr/local/wind/target/config/all:\\$\(CONFIG_ALL\):g" \ > -e "s:/usr/local/wind/target:\\$\(TGT_DIR\):g" \ > -e "s:/usr/local/wind:\\$\(WIND_BASE\):g" >> depend.mv2700.IPPC2636 > VX_CPU_FAMILY=ppc /usr/local/wind/host/sun4-solaris2/bin/cppppc -M -E -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" sysALib.s \ > | sed -e "s:/usr/local/wind/target/config/all:\\$\(CONFIG_ALL\):g" \ > -e "s:/usr/local/wind/target:\\$\(TGT_DIR\):g" \ > -e "s:/usr/local/wind:\\$\(WIND_BASE\):g" >> depend.mv2700.IPPC2636 > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" /usr/local/wind/target/config/all/usrConfig.c -o usrConfig.o > cp /usr/local/wind/target/config/all/usrConfig.c usrConfig_st.c > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" -DSTANDALONE usrConfig_st.c > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" /usr/local/wind/target/config/all/dataSegPad.c -o dataSegPad.o > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" -o version.o /usr/local/wind/target/config/all/version.c > ldppc -o tmp.o -X -r dataSegPad.o sysALib.o sysLib.o \ > usrConfig_st.o version.o /usr/local/wind/target/lib/libPPC604gnuvx.a > nmppc tmp.o | munch >ctdt.c > /usr/local/wind/host/sun4-solaris2/bin/make CC_COMPILER="-traditional" ctdt.o > make[1]: Entering directory `/usr/x4/TORNADO_1.0.1/target/config/mv2700.IPPC2636' > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -mstrict-align -traditional -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" -c ctdt.c > make[1]: Leaving directory `/usr/x4/TORNADO_1.0.1/target/config/mv2700.IPPC2636' > cp tmp.o tmp.2 > true tmp.o > VX_CPU_FAMILY=ppc /usr/local/wind/host/sun4-solaris2/bin/makeSymTbl tmp.o > symTbl.c > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -c -traditional -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -Wall -I/h -I. -I/usr/local/wind/target/config/all -I/usr/local/wind/target/h -I/usr/local/wind/target/src/config -I/usr/local/wind/target/src/drv -DCPU=PPC604 -DMV2700 -DTARGET_DIR="\"mv2700.IPPC2636\"" symTbl.c > symTbl.c:18: nondigits in number and not hexadecimal > symTbl.c:18: syntax error before `0001301c' > make: *** [vxWorks.st] Error 1 > The reason for the failure first of all is an unattended sequence of lines in the previously built symTbl.c: > /* makeSymTbl.c - stand-alone system symbol table */ > > /* CREATED BY makeSymTbl > * FROM tmp.o > * ON Mon Sep 27 23:10:56 MET DST 1999 > */ > > #include "vxWorks.h" > #include "symbol.h" > > #include "a_out.h" > > #undef READ > #undef WRITE > > IMPORT int NCR810_INSTRUCTIONS; > IMPORT int NCR810_PATCHES; > 0001301c R WTX_EVT_CALL_GOT_EXC <<<<<< Hier faengt's an!!!! > 00013010 R WTX_EVT_CALL_RETURN > 00012fb0 R WTX_EVT_CTX_EXIT > 00012fa4 R WTX_EVT_CTX_START > 00013004 R WTX_EVT_DATA_ACCESS > 00012fbc R WTX_EVT_EXCEPTION > 00012f88 R WTX_EVT_OBJ_LOADED > 00012f94 R WTX_EVT_OBJ_UNLOADED > 00012f70 R WTX_EVT_SYM_ADDED > 00012f7c R WTX_EVT_SYM_REMOVED > 00012ff8 R WTX_EVT_TEXT_ACCESS > 00012f64 R WTX_EVT_TGT_RESET > 00012fd4 R WTX_EVT_TOOL_ATTACH > 00012fe0 R WTX_EVT_TOOL_DETACH > 00012fec R WTX_EVT_TOOL_MSG > 00013034 R WTX_EVT_UNKNOWN > 0001302c R WTX_EVT_USER > 00012fc8 R WTX_EVT_VIO_WRITE > IMPORT int _Randseed; > IMPORT __10Demangle_TPCc (); > IMPORT __10RBString_T (); > > What's wrong? What could be the cause of the misbehaviour? TIA F. Scheurer -- In der Uebergangsphase nach der Domain-Umstellung ist auch noch @zfs.rus.uni-stuttgart.de gueltig! +----------------------------------------------------------------+ | Friedrich Scheurer Zentrum Fertigungstechnik Stuttgart | | Dipl.-Ing. Arbeitsbereich Steuerungstechnik | | Tel +49 711 13162-41(-0) Nobelstrasse 15 | | Fax +49 711 13162-11 D-70569 Stuttgart | | E-Mail scheurer@zfs.tz.uni-stuttgart.de | +----------------------------------------------------------------+ | http://zfs.rus.uni-stuttgart.de | +----------------------------------------------------------------+ | privat: | | Telefon +49 7161 686362 | | Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | +----------------------------------------------------------------+ From vxwexplo-errs Wed Sep 29 06:55:09 1999 From: "tim allen" Date: Wed Sep 29 06:55:17 PDT 1999 Subject: Re: Boot from serial If you're using Tornado 2, you can boot from serial using the Target Server File System (TSFS). The way to do this is described in the Tornado 2.0 User's Guide, in the Projects chapter 4.7 Configuring and Building a VxWorks Boot Program, and also the Target Server chapter. I've used this method when I haven't had a working ethernet driver. If you're using an older Tornado/VxWorks, then you can use SLIP or PPP over the serial line, but you need to configure SLIP/PPP on your host too. This is described, I think, in the network section of the programmer's guide. Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com From vxwexplo-errs Wed Sep 29 08:19:09 1999 From: "Chhabra, Vikram" Date: Wed Sep 29 08:19:13 PDT 1999 Subject: Tornado 2 VxWorks if_cpm driver problem Hi, I am trying to use the if_cpm(Motorola PowerQUICC CPM ethernet driver) driver of Tornado 2. We had earlier used the driver in Tornado 1 and it worked fine. The problem is occurring when we try to attach an internet address with a network interface ( function ifAddrSet in library ifLib). The function returns a file already exists error (0x11). Could this be due to the fact that the if_cpm driver in Tornado 2 is now implemented as a MUX driver as opposed to Tornado 1 in which it was a BSD driver? I am not sure on how they did it for Tornado 2. In case it is a MUX driver, will I now have to attach the IP stack to MUX in order to use the ifLib library. Will appreciate any help on this. Thanks Vikram Chhabra Unisphere Solutions Chelmsford MA From vxwexplo-errs Wed Sep 29 09:11:41 1999 From: David Laight Date: Wed Sep 29 09:11:45 PDT 1999 Subject: re: ipDetach Mark, ipDetach is in the tornado II book, ipDetach( int unit_no, char *ifname ); (and /usr/wind2/target/h/ipProto.h) Its reverse is ipAttach(), then set netmask and ipaddress etc - or call dhcpcBind. It certainly takes out IP and ARP, WDB may not matter. However it doesn't tidy up everything - eg the route table (if you have static symbols rtTblShow displays the entire table (arpShow and routeShow, but not all the info given by mRouteShow). Unfortunately ifRouteDelete() doen't delete all the routes either.... David (I'm currently re-writing the network startup code so I can use a pcmcia ethernet card with dhcp (and a rom version of vxWorks.)) > Subject: ipDetach > From: "Mark Hambleton" > > I am trying to work out how to unload an END from the MUX that has IP > attached to it, there is an ipDetach() function in the symbol table but I > have no idea of the prototype. > > Do I have to do anything else like removing WDB and ARP, and if so how do I > do that? ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From vxwexplo-errs Wed Sep 29 09:52:49 1999 From: Adi Molkho Date: Wed Sep 29 09:52:53 PDT 1999 Subject: Symbol table synchronization Hi I am using vxWorks 5.3.1 I have a bootrom which load vxWorks.st I dynamically load my application What should I do to insert my application symbols into the system symbol table Adi From vxwexplo-errs Wed Sep 29 13:52:27 1999 From: "Murphy, John" Date: Wed Sep 29 13:52:31 PDT 1999 Subject: Windview without network? Hello vxWorkers! I'm compiling a vxWorks 5.3.1 (Tornado 1.0.1) kernel for the RAD6000 with a standalone symbol table and windview in post-mortem mode. I've been going through the windview manual and have a pretty good grasp on what I want in my application. I expect to include only INCLUDE_INSTRUMENTATION, and turn windview on and off using wvEvtLogEnable() and wvevtLogDisable(). My understanding is that in the post mortem configuration, the event logging task for windview will NOT be spawned (Unlike the continuous mode). Is it spawned by wvEvtLogEnable() in post-mortem mode, or do I need to make a specific call to enable it. In my application, we will not have a network interface, and will be dumping the windview buffer using application software and a custom interface. When I try to compile windview into the kernel without networking, I get the following error: [lx] error: undefined: 'mbufConfig' referenced in 'tmp.2' [lx] error: undefined: 'clusterConfig' referenced in 'tmp.2' Obviously, some part of windview is still begging for network functionality. I seem to remember someone on this list stubbing these symbols out to make things work, but was wondering if anyone could recall or offer any advice. Thanks in advance, John Murphy vxWorks Tornado Et Al From vxwexplo-errs Thu Sep 30 00:16:34 1999 From: Alex Reznik Date: Thu Sep 30 00:16:38 PDT 1999 Subject: RE: Symbol table synchronization > Submitted-by vxwexplo-errs Wed Sep 29 09:52:49 1999 > Submitted-by: Adi Molkho > > Hi > I am using vxWorks 5.3.1 > I have a bootrom which load vxWorks.st > I dynamically load my application > What should I do to insert my application symbols into the system symbol > table > Adi > > ********** > Adi,hi, the simple answer is: if you use the function "ld" or "loadModule" that is all. All GLOBAL symbols from your application file will be inserted into symbol tableafter downloading. If you set some specific flags you can insert LOCAL (static) symbols into symbol table as well. If you have to use any other mechanism for downloading your application, the preferrable way IMHO is to consider the functions mentioned above. It is possible to insert additional symbols to Symbol Table on the application level, but this procedure is quite not simple. Regards, Alex Reznik Telegate Ltd VxWorks Tornado > From vxwexplo-errs Thu Sep 30 02:02:30 1999 From: David Laight Date: Thu Sep 30 02:02:34 PDT 1999 Subject: Re: Tornado 2 VxWorks if_cpm driver problem > Date: Wed, 29 Sep 1999 08:19:15 -0700 (PDT) > Subject: Tornado 2 VxWorks if_cpm driver problem > > Submitted-by: "Chhabra, Vikram" > > The problem is occurring when we try to attach an internet address with > a network interface ( function ifAddrSet in library ifLib). The function > returns a file already exists error (0x11). This looks similar to a problem I'm having (dhcp failing to assign the address). ifAddrSet() calls ioctl( xx, SIOCSIFADDR, address), this does almost all its work then calls rn_addroute (I think to add the route for the interface) which calls rn_insert - which finds a route already present. rn_addroute returns 0x11 (this maybe an errno) which finds its way back into the ioctl response - being treated as an error (by code that checks for 0 only) by all the intermediate functions. I Of course ioctl() returning 0x11 isn't an error! errno is undefined. My dhcp problem MAY be due to ipDetach not deleting the routes. I'd suggest checking the routing tables for anything that might clash. I've also had the SIOCSIFADDR failing in PPP - in this case caused by stale data! David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 278 256 Cambridge, UK fax: +44 1223 278 201 From vxwexplo-errs Thu Sep 30 02:57:42 1999 From: Arand Thomas Date: Thu Sep 30 02:57:45 PDT 1999 Subject: AW: Packets from VxSim... Yes! But you have to use the full version of VxSim, not the so-called builtin simulator of Tornado II (this one is without network). 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: Montag, 27. September 1999 10:02 An: vxworks_users@csg.lbl.gov Betreff: Packets from VxSim... Submitted-by owner-vxwexplo-process Mon Sep 27 01:02:07 1999 Submitted-by: "Navin kumar Agrawal" VxWorks Tornado Hi All, I am developing a program on VxWorks, which requires sending IP packets to other nodes/hosts. Because of unavailability of the target borad, I am planning to carry out the development activity using VxSim (full fedged one, not the one which comes along with T2 by default). Is it possible to carry out testing of such program using VxSim? Thanks, Regards, NaViN ********** From vxwexplo-errs Thu Sep 30 04:00:11 1999 From: daemon@csg.lbl.gov Date: Thu Sep 30 04:00:16 PDT 1999 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Sep 30 04:00:07 PDT 1999 Subject: Re: configuring network driver for vxWork image Subject: Re: Simple question on file permissions Subject: WTX Error 0x100d1 (AGENT_GOPHER_FAULT) Subject: Re: MMU Initialisation Subject: Re: Can not NFS to NT client Subject: Re: usrAtaConfig produce memPartFree-error Subject: Re: Tornado 2 VxWorks if_cpm driver problem Subject: VxWorks Routing table and Kernel: Subject: [Q] Using taskInit() to create a task Subject: EBSA-285 BSP (Support for intel 10/100 82559) Subject: Need vxWorks guy in Connecticut Subject: Boot from serial line Subject: fei82557End driver question Subject: Re: [Q] Using taskInit() to create a task Subject: iostreams: cout << Subject: vxWorks and LynxOS Subject: Re: WTX Error 0x100d1 (AGENT_GOPHER_FAULT) Subject: Re: Booting VxWorks from IDE Drive Subject: Re: Tornado2 license? Subject: Re: Please let's me know "What kind of OS is VXWORKS?" ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: configuring network driver for vxWork image Date: Mon, 27 Sep 1999 08:51:14 GMT From: oliver_fuchs@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <7snb61$v8n$1@nnrp1.deja.com> References: <37E4E4C5.1F32DC4@pacbell.net> Hello, I had a similar problem with a vxWorks for a PPC8xx target. I build a rom based image (vxWorks_rom) and the driver (cpm) was installed properly. Because my application is a bit too big, I build a compressed vxWorks and the driver was not installed. The difference is the value of sysStartType. I don't know why, but if the subroutine usrBootLineInit (sysStartType); in usrRoot() is called with a wrong parameter, the network driver will not be installed. Regards Oliver In article <37E4E4C5.1F32DC4@pacbell.net>, nidal@pacbell.net wrote: > Hello, > I am trying to build a vxWorks image with network driver in it. > I am using tornadoII and vxWorks version 5.4. > The network card that I am using is a standard intel ethernet pro 100 > which is supported by vxWorks. If I boot vxWorks over the net the driver > is configured properly but it I boot from a floppy or flash the network > device is not attached to the tcp/ip stack. my config.h has all the > declaration for including the network interface. > #define INCLUDE_NETWORK > #define INCLUDE_NET_SETUP > #define INCLUDE_MUX > #define INCLUDE_BSD > #define INCLUDE_FEI > #define INCLUDE_LOOPBACK > #define INCLUDE_PING > #define INCLUDE_NETSHOW > usrNetwork.c, the module the parses the boot parameter and configure the > network interface work. According to the vxWorks Network programmer > guide all I need to do with this setup is assign the ip address with > ifAddrSet() and the mask with ifMaskSet(). Both calls result in an error > since the fei0 is not attached to the network stack. > I tried using usNetIfAttach() but this result in an unresolved symbol. > I tried usrNetIfConfig() but this call does not succeed. > > Does any one know how to attach the network device to the TCP/IP stack. > I do appreciate the help > > Nidal > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Simple question on file permissions Date: Wed, 29 Sep 1999 06:34:08 -0700 From: Marco Zandonadi Organization: SBC Internet Services Message-ID: <37F21550.FE92B184@teja.com> References: <37F04A00.965AE5EA@teja.com> <37f11695.183316784@news.wxs.nl> Hello Harro, thanks for your reply. > You don't state the file system, my answers below pertain to the tffs > with the dos-fs on top of it. Currently I'm using VxSim on Solaris so I assume the file system is the Solaris one. Is there another way to create a directory on top of the Solaris fs other than mkdir()? > About directory permissions: the dos-fs doesn't support them. hence, > You cannot protect a directory from being modified. Reason: good-old > MS-DOS didn't support it either. The problem is that the directory gets created with permissions 000 and is therefore unusable. I just need to know how to create a directory inside which I can create files and subdirectories using standard functions like open, write, etc. Thanks. Marco --------------------------- Newsgroups: comp.os.vxworks Subject: WTX Error 0x100d1 (AGENT_GOPHER_FAULT) Date: Wed, 29 Sep 1999 09:54:49 -0400 From: "David P. Floyd" Organization: Dynacs Engineering Company, Inc. Message-ID: <37F21A29.AC95C9E4@orion.ksc.nasa.gov> VxWorks Version: Tornado 1.0.1 CPU: MVME2604 PPC Agent: WDB 1.0.1 From windsh, I performed the following: - -> i NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY - ---------- ------------ -------- --- ---------- -------- -------- - ------- ----- tExcTask excTask 1bfde80 0 PEND 139874 1bfdda0 0 0 tLogTask logTask 1bfb508 0 PEND 139874 1bfb438 0 0 tEvtTask tEvtTask 1aaaf00 0 PEND 13a144 1aaae68 0 0 tShell shell 1aa6ce8 1 PEND 13a144 1aa6960 0 0 tRlogind rlogind 1ab82c8 2 PEND 13a144 1ab7ef8 0 0 tTelnetd telnetd 1ab63a0 2 PEND 13a144 1ab6230 0 0 tWdbTask wdbCmdLoop 1aa9190 3 READY 13b790 1aa9008 0 0 tNetTask netTask 1ae8578 50 READY 13bab4 1ae84a0 0 0 tPortmapd portmapd 1ab4e10 100 PEND 13a144 1ab4c90 16 0 value = 0 = 0x0 I then loaded my application modules.... - -> ld < myapplics.out value = 0 = 0x0 I then peformed the following: - -> i WTX Error 0x100d1 (AGENT_GOPHER_FAULT) value = -1 = 0xffffffff Any ideas what would cause this? Thanks. - -- David P. Floyd Dynacs Engineering Company, Inc. Kennedy Space Center, FL david.floyd@ksc.nasa.gov --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MMU Initialisation Date: Wed, 29 Sep 1999 16:48:24 +0200 From: "Richard Marsh" Organization: Siemens Inc. Message-ID: <7st8qn$shu$1@news.mch.sbs.de> References: <7snikp$a44$1@news.mch.sbs.de> <37F056C4.10EF6B9C@aracnet.com> Reply-To: "Richard Marsh" Thanks for all the responses, I would love to be able to cut down my mapping of PCI space. In total we are mapping 1.3GB of address space, and we reduce this without changing the sysPhysMemDesc for every type of possible board configuration. The system we have has a basic 128MB of SDRAM (some cacheable some not), 64MB of FLASH, 256MB of programmable 8-bit I/O space, 256MB of 16-bit I/O space and 768MB of PCI mapped space for the target images and some other stuff the need mapping but is small in comparision. The regions are fixed by hardware to accept any type of daughter card, hence we have no control over where the board sits in the memory map. So at the end of the day, after the sun has gone down, we are stuffed. It is impossible to enable interrupts before the usrMmuInit call as the causes untold woe - what would be nice is the ability to perform multiple MMU updates to add the stuff a bit at a time - I have opened a TSR with WRS over this matter and it'll be interesting to see what happens. So if anybody has any serious ideas about which way to go, a posting to this news group would be just peachy. Cheers, Richard --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can not NFS to NT client Date: Wed, 29 Sep 1999 16:15:20 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <7stdua$9bk$1@nnrp1.deja.com> References: <7snn3q$12s4@overload.lbl.gov> <7sokqn$ssq$1@autumn.news.rcn.net> <37F0856C.3F5706B0@bitband.com> In article <37F0856C.3F5706B0@bitband.com>, leonid@bitband.com wrote: > I do not think anypone has tried this before with Microsoft's NFS > server, so this one could be a source of problems too. Its working fine for me with the beta version of Microsoft's NFS server. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: usrAtaConfig produce memPartFree-error Date: Wed, 29 Sep 1999 16:20:20 GMT From: Ted Dennison Organization: Deja.com - Before you buy. Message-ID: <7ste7l$9ne$1@nnrp1.deja.com> References: <37F09AE3.41C67EA6@fzi.de> In article <37F09AE3.41C67EA6@fzi.de>, Torsten Rupp wrote: > It seems that dosFsDevInit() fail because of some unknown reason > (the partition is reable/writeable under DOS without any errors; > *Exactly what filesystem is it formatted to under Windows? (I assume you aren't *really* using DOS). > PS: a few days ago usrAtaConfig() succeeded, but after a repairing > an error in the filesystem I get the error above. Yikes! That's sounds suspicious. It could be that xvWorks isn't smart enough to handle repaired DOS filesystems. You might need to reformat. - -- T.E.D. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2 VxWorks if_cpm driver problem Date: Wed, 29 Sep 1999 12:24:28 -0700 From: Stephen Macmanus Organization: Wind River Systems Message-ID: <37F2676C.5442@wrs.com> References: <7stdtn$12s9@overload.lbl.gov> The EEXIST error usually indicates that an identical route entry already exists for the IP address. When assigning an address to a broadcast interface, the system automatically creates an associated routing table entry so that other devices on the same network are reachable. For instance, an address of 192.168.10.3 for an interface with netmask 0xffffff00 would produce a route entry to 192.168.10.0 with 192.168.10.3 as the gateway. Since route entries must have unique destinations, the the EEXIST error will occur for the second interface attached to the same subnet. It can be ignored in that case, but all outgoing traffic will still flow through the first interface. Chhabra, Vikram wrote: > > Hi, > > I am trying to use the if_cpm(Motorola PowerQUICC CPM ethernet driver) > driver of Tornado 2. We had earlier used the driver in Tornado 1 and it > worked fine. > > The problem is occurring when we try to attach an internet address with > a network interface ( function ifAddrSet in library ifLib). The function > returns a file already exists error (0x11). Could this be due to the > fact that the if_cpm driver in Tornado 2 is now implemented as a MUX > driver as opposed to Tornado 1 in which it was a BSD driver? I am not > sure on how they did it for Tornado 2. In case it is a MUX driver, will > I now have to attach the IP stack to MUX in order to use the ifLib > library. > > Will appreciate any help on this. > Thanks > Vikram Chhabra > Unisphere Solutions > Chelmsford MA - -- - ------------------ Stephen Macmanus #include Software Engineer Wind River Systems --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Routing table and Kernel: Date: Wed, 29 Sep 1999 16:22:53 -0400 From: Mahad Ahmed Organization: Nortel Message-ID: <37F2751D.A802EE2C@nortelnetworks.com> Hi, Does VxWorks fully supports sysctl and ioctl calls which provided in UNIX? Thanks, Mahad --------------------------- Newsgroups: comp.os.vxworks Subject: [Q] Using taskInit() to create a task Date: Thu, 30 Sep 1999 09:36:04 +0900 From: "Hyung-Taek Lim" Message-ID: The main purpose I wanted to do with taskInit() was positioning a task's stack and tcb wherever I want. Unfortunately, I cannot write an example program and execute it, since I don't have VxWorks. Mr. David Hamel told me that using taskInit() is very hard. Does anybody know why using taskInit() is hard? If you have written a program with taskInit (or taskCreat), would you mind showing me example usage of taskInit (or taskCreat)? I attach my original question I posted two days ago. - ---------------------------------------------------- With taskInit() function, we can create a task's stack and tcb on a specified location in memory. The first argument of taskInit() is the address of new task's TCB and the fifth one is the address of new tasks's stack. Does taskInit() allocate memory space for tcb and stack to where the first and fifth arguments point? If so, the only thing that I have to do before calling taskInit() is finding the location of free spaces for stack and tcb. Otherwise, I have to allocate memory space for stack and tcb before calling taskInit(). Should I pass a pointer to an allocated memory or a pointer to a free memory space as the arguments of taskInit()? --------------------------- Newsgroups: comp.os.vxworks Subject: EBSA-285 BSP (Support for intel 10/100 82559) Date: Wed, 29 Sep 1999 22:07:28 -0400 From: "Robert D. Gifford" Organization: "SNET dial access service" Message-ID: <37F2C5DF.CB6916EC@snet.net> We are using the the BSP for the SA-110 and 21285 (was Digital, now Intel) evaluation system. Also included is the popular intel 10/100 ethernet NIC 82559. I can't believe that there is no support (driver) for this ethernet controller in the BSP. Does anyone have any ideas? Thanks in advance, Bob --------------------------- Newsgroups: comp.os.vxworks Subject: Need vxWorks guy in Connecticut Date: Wed, 29 Sep 1999 22:19:41 -0400 From: "Robert D. Gifford" Organization: "SNET dial access service" Message-ID: <37F2C8BC.B40216B6@snet.net> We are looking for an embedded systems developer with experience in the following areas: 1) VxWorks 2) Device driver development 3) StrongARM experience is a PLUS. Our firm is located in central Connecticut. Compensation includes above average pay and stock options. Please contact: bobg@vbrick.com --------------------------- Newsgroups: comp.os.vxworks Subject: Boot from serial line Date: Thu, 30 Sep 1999 06:21:44 GMT From: michaelhl@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <7suvhg$ddu$1@nnrp1.deja.com> Hi everyone, We're building a BSP for an eval board with MPC860, PLX PCI-9054, and one serial port within Tornado 2. We already have a BSP for Tornado 1. We're tring to use TSFS to boot it. Now the boot image works well, but the boot loader complains about the format of the vxWorks ELF image. Our questions are: 1. What's the best way to port a BSP for Tornado 1 to Tornado 2 in our env.? 2. How to solve the load error reported by the boot loader? Is it caused by TSFS? The error is about reading sections (errno=0x61007, defined by MACRO S_loadElfLib_READ_SECTIONS in loadElfLib.h). The output from target server log window and target are: - ------------ tgtsvr (c860@wacosserver1): Mon Sep 27 16:42:05 1999 Wind River Systems Target Server: NT/Win95 version Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Attaching elf OMF reader for PPC CPU family... succeeded. Warning: Target checksum: 0x78a7 (computed from 0x100100 to 0x15d234). Host checksum: 0x61ce (computed from 0xfa0120 to 0xffd254). Warning: Core file checksums do not match. TSFS: opened file (S:\Tornado2\target\proj\c860\default\vxWorks). TSFS: closed file (S:\Tornado2\target\proj\c860\default\vxWorks). - ------------- - ------------- [VxWorks Boot]: @@ boot device : tsfs unit number : 0 processor number : 0 host name : host file name : /target/proj/c860/default/vxWorks flags (f) : 0x0 Booting using TSFS... Make sure that your Target Server is started with -R[oot] option. Loading /tgtsvr/target/proj/c860/default/vxWorks... 455136 error loading file: status = 0x610007. Error loading file: errno = 0x610007. - ------------- Thanks in advance! Michael Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: fei82557End driver question Date: Thu, 30 Sep 1999 08:33:38 +0200 From: Erez Bar-Tal Organization: Galileo Technology Ltd Message-ID: <37F30441.5CD10A1A@galileo.co.il> Hi I am trying to use fei82557End mux driver instead of if_fei driver and i have to following problem. I am trying to run this driver on tornado 1.0.1 BSD 4.4 and the driver can't issue a diagnostic command. I can see that all the appropriate initialization are done and all the right values are writing to the registers (In comparison to the if_fei version of the driver) but for some reason when doing polling on the status bit it is always 0. Does anyone have any idea ?? Did anyone succeed on running this driver on Tornado 1.0.1 ?? I am using a big Indian machine therefore I am swapping all the data before writing it to the memory (This is done automatically by the driver code). any help will be appreciated thanks Erez erezb@galileo.co.il --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Q] Using taskInit() to create a task Date: Thu, 30 Sep 1999 06:59:09 GMT From: Frank.Grube@pcm.bosch.de Organization: Robert Bosch GmbH Message-ID: <37f303e1.1439720@news.fe.internet.bosch.de> References: On Thu, 30 Sep 1999 09:36:04 +0900, "Hyung-Taek Lim" wrote: >The main purpose I wanted to do with taskInit() was positioning a task's >stack and tcb wherever I want. >Unfortunately, I cannot write an example program and execute it, since I >don't have VxWorks. > >Mr. David Hamel told me that using taskInit() is very hard. >Does anybody know why using taskInit() is hard? >If you have written a program with taskInit (or taskCreat), would you mind >showing me example usage of taskInit (or taskCreat)? > >I attach my original question I posted two days ago. >---------------------------------------------------- >With taskInit() function, we can create a task's stack and tcb on a >specified location in memory. > >The first argument of taskInit() is the address of new task's TCB and the >fifth one is the address of new tasks's stack. > >Does taskInit() allocate memory space for tcb and stack to where the first >and fifth arguments point? >If so, the only thing that I have to do before calling taskInit() is >finding the location of free spaces for stack and tcb. >Otherwise, I have to allocate memory space for stack and tcb before calling >taskInit(). > >Should I pass a pointer to an allocated memory or a pointer to a free memory >space as the arguments of taskInit()? > > Hi Hyueng-Taek, I am currently using taskInit in my project to be able to control the activation time of preallocated tasks. That is not possible with taskSpawn() the alternative option to start tasks (not taskCreat, I guess you mixed that, right?!) Don't know exactly what kind of problems David Hamel means, my way seems to work properly, and I will give you some hints about the problems I ran into. 1. Don't underestimate the needed STACKSPACE. For my first tests I tried STACKS of 256 bytes or so and crashed repeatedly it is hard to find out why, if you always kill your debug task with that, that was the time when I wished to have protected memory under VxWorks). With STACKS > 2000 bytes, everything works stable (haven't yet checked the exact STACKSIZES I will need). 2. You have to allocate the memory yourself, because "this allows... the initialization of a static WIND_TCB variable", (says taskInit-documentation), so that means the mem has to be preallocated. 3. The biggest problem I ran into was buried in the sentence "Note, that the task stack may grow up or down from pStackBase, depending on the target architecture". That means that you may have to re-adjust your Stackpointer to the END of the allocated space yourself! See the example code below. One more tip: As taskSpawn() is much more "automatic", if possible, try first starting your tasks with that function and after that works, move to taskInit(), that way you avoid mixing different sources of trouble. Hope that helps you for the start! Frank Grube, frank.grube@usa.net Freelancing engineer for embedded systems Code-example: WIND_TCB *pxTcb = malloc( sizeof( WIND_TCB )); char *pxStackBase = NULL; /* add space for TASKNAME to STACKSIZE */ uint2 nStackSize = requestedStacksize + strlen( pstrTaskName) + 1; int nOptions = 0; /* ALLOCATE STACKSIZE */ pxStackBase = (char *) malloc( nStackSize ); if ( pxTcb && pxStackBase ) { #if ( _STACK_DIR == _STACK_GROWS_DOWN ) /* if STACK grows DOWN, start of stack is end of its memory area !*/ pxStackBase += nStackSize - 1; #endif /* task startup */ fErg = taskInit( pxTcb, pstrName, /* name of task */ bPriorityGrp /* set priority */ nOptions, pxStackBase, nStackSize, /* get requested stacksize */ pBodyFunction, /* workerfunction of new task */ 0, 0, 0, 0, 0, 0, 0, /* 10 unused arguments */ 0, 0, 0 ); if ( fErg == OK ) { /* calculate ID */ idNewTask.id._Task = taskNameToId( pstrName ); /* execute autostart if requested */ if ( dwFlags & OS_TASK_AUTOSTART ) { taskActivate( idNewTask.id._Task ); } } --------------------------- Newsgroups: comp.os.vxworks Subject: iostreams: cout << Date: Thu, 30 Sep 1999 07:01:50 GMT From: birajdar2250@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <7sv1sq$et6$1@nnrp1.deja.com> Hi, I am using vxWorks5.3.1. Problem: cout is not able to produce output. There is no error mesg. printf() is working fine. vxWorks is configured with INCLUDE_CPLUS and INCLUDE_CPLUS_IOSTREAMS. Regards Girish Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: vxWorks and LynxOS Date: Thu, 30 Sep 1999 10:03:06 +0100 From: Dave Moore Message-ID: <37F3274A.77DD91D7@gecm.com> Hi everyone, I was wondering whether anybody here has used both vxWorks and LynxOS. Does anybody know what the main differences are between these two operating systems?. Any comments or suggestions gratefully received. Thanks in advance. Cheers, Dave Moore --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WTX Error 0x100d1 (AGENT_GOPHER_FAULT) Date: Thu, 30 Sep 1999 06:01:11 -0700 From: "Mike Campbell" Message-ID: <7svc5g$7h4$1@autumn.news.rcn.net> References: <37F21A29.AC95C9E4@orion.ksc.nasa.gov> David P. Floyd wrote in message news:37F21A29.AC95C9E4@orion.ksc.nasa.gov... > VxWorks Version: Tornado 1.0.1 > CPU: MVME2604 PPC > Agent: WDB 1.0.1 > > From windsh, I performed the following: > > -> i > NAME ENTRY TID PRI STATUS PC SP > ERRNO DELAY > ---------- ------------ -------- --- ---------- -------- -------- > ------- ----- > tExcTask excTask 1bfde80 0 PEND 139874 1bfdda0 > 0 0 ... > I then loaded my application modules.... > > -> ld < myapplics.out > value = 0 = 0x0 > > I then peformed the following: > > -> i > WTX Error 0x100d1 (AGENT_GOPHER_FAULT) > value = -1 = 0xffffffff > you haven't talked about what was in this "myapplics.out", but three possibilites come to mind. If it contains replacements for the windsh comms stuff - you have the possibility of being hosed... A good way to verify that would be to telnet (or use the console serial port) and execute your ld there - thus avoiding windsh and friends. Another possibility might be that your shell is not configured with enough stack. In which case the ld command caused it to blow. Don't remember exactly where you fix this - grep for "STACK_SIZE". Most BSPs come configured with like 20K stack for the shell - if you've got a loaded system or have special drivers you can get hosed real easy. Finally, you didn't say what your development machine was or how your file system is seen by the vxWorks host (FTP, NFS). Could be some goofiness with the transfer of the file mucks up the development side of your net link (this is the least likely in my book - talking a fairly reliable thing here - couse if your development machine is strange - say a Mac running Tornado as a win app - anything is possible) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting VxWorks from IDE Drive Date: 28 Sep 1999 01:52:58 GMT From: ied168@aol.com (IED168) Organization: AOL http://www.aol.com Message-ID: <19990927215258.15360.00002495@ng-bk1.aol.com> References: <37EE4D02.CD2E0966@hns.com> I've got a Pentium 75 for a target and I'm using a serial line for communication. I'm not booting from the IDE, but rather, the floppy off the Pentium 75. I'm still relatively new to VxWorks and am trying to do some troubleshooting myself, but there might be chance that this works: I used a VxWorks utility, "vxsys a:". I believe "vxsys" sets up the floppy so that the BIOS will boot off it. You might try "vxsys", though I'm not sure if this will corrupt your IDE Drive -- you might want to look into the manual on utility to be sure. - -Ed --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado2 license? Date: 28 Sep 1999 01:59:24 GMT From: ied168@aol.com (IED168) Organization: AOL http://www.aol.com Message-ID: <19990927215924.15360.00002496@ng-bk1.aol.com> References: <7s4hm5$p87$1@green.kreonet.re.kr> I got Tornado 2 recently as well. Yes, License keys were necessary in Tornado 1, but there are no license keys for Tornado 2. - -Ed --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Please let's me know "What kind of OS is VXWORKS?" Date: 28 Sep 1999 02:00:21 GMT From: ied168@aol.com (IED168) Organization: AOL http://www.aol.com Message-ID: <19990927220021.15360.00002497@ng-bk1.aol.com> References: <7rt5f9$fnf$1@news.kren.nm.kr> VxWorks is an RTOS. Check out www.windriver.com - -Ed --------------------------- End of New-News digest **********************