From ING.BUERO.MARTIN@ilo.baynet.de Sun Aug 2 03:36:57 1998 From: ING.BUERO.MARTIN@ilo.baynet.de Date: Sun Aug 2 03:37:02 PDT 1998 Subject: Re: comp.os.vxworks newsdigest (WindView Problem) In <9807291105.AA20946@csg.lbl.gov>, on 07/29/98 at 04:05 AM, vxwexplo@lbl.gov (the vxWorks Users Group Exploder) said: > Subject: problem launching Windview >I have a problem with WindView. >I4m using VxWorks5.3.1 with a mv2303-BSP and a NT-Workstation as host. >In my host I have 2 network cards. One is for the LAN and the other one I >use for the connection to my target. >When I launch WindView I get the message: >"waiting for connection 6164" but nothing else happens. >When I uninstall the LAN-network card it works properly !!! >I hope anybody can help me. >Thanks in advance. >Markus Hi Markus, I figured the same Problem to launch WindView on a Host with 2 network cards. Problem: Probably you use card1 with ethernet address AA.BB.CC.DD for the connection to your LAN and card2 with ethernet address WW.XX.YY.ZZ for the connection to the target. WindView seems to use the network card1 you configured for the use with the LAN and hence gets no connection to the target. Solution: Change the boot parameters of the target to: host inet (h): AA.BB.CC.DD (card1 of the host) gateway inet (g): WW.XX.YY.ZZ (card2 of the host) Hope this helps. Thomas MfG / kind regards Thomas MARTIN ************************************************************************* * Ingenieurbuero MARTIN * Zwei Dinge sind unendlich: * * Schieszstaettstrasse 5a * das Universum und die mensch-* * D-82515 Wolfratshausen - Germany * liche Dummheit, aber bei dem * * Phone/FAX: +49(0)8171/28445 * Universum bin ich mir noch * * email:ING.BUERO.MARTIN@ilo.baynet.de * nicht ganz sicher. * * http://www.ilo.baynet.de/ing.buero.martin * [Albert Einstein]* ************************************************************************* From eckardt@innocon.com Mon Aug 3 08:26:35 1998 From: Suzie Eckardt Date: Mon Aug 3 08:26:39 PDT 1998 Subject: Embedded Use of Samsung NAND Flash? Has anyone attempted to write a VxWorks device driver for Samsung NAND Flash Memory, specifically the 4Mx8bit device (part number KM29N32xxxx)? The chip is spec'd to potentially have "bad blocks" at delivery, and to potentially have other blocks go bad during use. I'd be particularly interested if anyone used a DOS file system with this device. Thanks! Suzie ************************************************ ** Suzie Eckardt ** Innovative Concepts, Incorporated ** McLean, VA ** 703-893-2007 x258 ************************************************ From claire.chaplais@wrs.com Mon Aug 3 09:41:05 1998 From: Claire Chaplais Date: Mon Aug 3 09:41:09 PDT 1998 Subject: Wind River Systems 1999 Developers Conference ++++ VxWorks +++++ The Wind River Systems 1999 Developers Conference planning has begun and we would like you to consider being one of our speakers. We will convene in beautiful San Francisco the week of May 10th, 1999. Attendance is projected at approximately 500. Selected Wind River Systems users and partners will be given the opportunity to give presentations on their applications, products and/or research. Proposals should address some aspect or issues of embedded development. Product demonstration proposals should focus on the features and benefits of that product with respect to a particular design challenge. If you would like to be a speaker, please send in your submissions before September 14, 1998. Call for paper forms can be found at: http://www.wrs.com/html/call4papers.html From daemon@csg.lbl.gov Tue Aug 4 04:02:16 1998 From: daemon@csg.lbl.gov Date: Tue Aug 4 04:05:19 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Aug 4 04:02:13 PDT 1998 Subject: Re: is vxworks crap? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: is vxworks crap? Date: Tue, 04 Aug 1998 05:35:35 GMT From: "J. Estes" Organization: uma Message-ID: <35C69DA6.E25705F2@uma.com> References: <6q5bs6$ddm$1@news-2.news.gte.net> John F. Davis wrote: > Hello, > > I'm new to vxworks, and I'm using the tornado platform on win32. To me > it appears slow and not very robust. Is this the case with all vxworks > platforms or am I to picky? I'm using tornado v1.0.1 and vxworks 5.3.1. You sadly mistake Tornado problems with Win32 problems. Lose the PC crap and you won't have crap any more. Get a real O/S (read NON-Microsoft), and you won't have any problem, either. - -- Regardz(tm), Joel Estes joele@uma.com "The early bird gets the worm, but the second mouse gets the cheese." --Author unknown --------------------------- End of New-News digest ********************** From wlbrown@lbl.gov Tue Aug 4 14:18:40 1998 From: Bill Brown Date: Tue Aug 4 14:18:43 PDT 1998 Subject: PPC cPCI CPU - 'CompactPCI-4e': Any comments? We're looking seriously at doing some control systems using 6U cPCI hardware with a PPC processor. One candidate (mainly because they have a vxWorks BSP!) for a cpu is the 'CompactPCI-4e' from V-I Computer, Inc. Does anyone have any experience with this unit, and would you care to comment on any "+'s and -'s" with this platform? While I'm at it, the other main contender looks like it's the 'RIOC4063/64" from "CREATIVE ELECTRONICS SYSTEMS." Again, we'd be very interested in anyones' experience with this product. TIA Disclaimer: Any opinions are my own and have | -bill nothing to do with the official policy or the | wlbrown@lbl.gov management of L.B.N.L, who probably couldn't | Berkeley, CA care less about employees who play with trains. aka wlbrown@netcom.com From bob_anderson@mcg.mot.com Tue Aug 4 16:21:34 1998 From: Bob Anderson Date: Tue Aug 4 16:21:37 PDT 1998 Subject: RE: PPC cPCI CPU - 'CompactPCI-4e': Any comments? Did you look at Motorola's MCP750 Compact PCI processor board? This board is also supported by VxWorks and is currently available from Motorola. Bob Anderson 410-312-5929 >---------- >From: vxwexplo@lbl.gov >Sent: Tuesday, August 4, 1998 5:18 PM >To: vxworks_users@csg.lbl.gov >Subject: PPC cPCI CPU - 'CompactPCI-4e': Any comments? > >Submitted-by wlbrown@lbl.gov Tue Aug 4 14:18:40 1998 >Submitted-by: Bill Brown > > >We're looking seriously at doing some control systems using 6U cPCI >hardware with a PPC processor. > >One candidate (mainly because they have a vxWorks BSP!) for a cpu >is the 'CompactPCI-4e' from V-I Computer, Inc. Does anyone have any >experience with this unit, and would you care to comment on any "+'s >and -'s" with this platform? > >While I'm at it, the other main contender looks like it's the >'RIOC4063/64" from "CREATIVE ELECTRONICS SYSTEMS." Again, we'd be very >interested in anyones' experience with this product. > From subhasis_das@hotmail.com Tue Aug 4 21:49:24 1998 From: "Subhasis Das" Date: Tue Aug 4 21:49:27 PDT 1998 Subject: vxworks Hi! I am new to VxWorks. I have to write some driver using Tornado platform. I have user manual for Tornado. But I do not have any manual or book related to VxWorks. I did not get any good site also in net. I have seen the WindRiver site. But that does not have more information. Can anyone help me by telling any usefull site or book or any other source for VxWorks. Thankx in advance. Subhasis SUBHASIS DAS MACMET INTERACTIVE TECHNOLOGIES PVT. LTD. 27-B CAMAC STREET CALCUTTA-16 PHONE NO-(91)(33)2479694 EMAIL-SUBHASIS_DAS@HOTMAIL.COM ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com From daemon@csg.lbl.gov Wed Aug 5 04:02:32 1998 From: daemon@csg.lbl.gov Date: Wed Aug 5 04:02:35 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 5 04:02:29 PDT 1998 Subject: Re: is vxworks crap? Subject: Problems with VxSim/Tornado under Solaris 2.6/Ultrasparc ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: is vxworks crap? Date: 4 Aug 1998 08:17:38 -0700 From: Tom Haley Organization: haleyt@ssd.fsi.com Message-ID: <6q78mi$16j@drn.newsguy.com> References: <6q5bs6$ddm$1@news-2.news.gte.net> <6q765u$qha$1@news-1.news.gte.net> In article <6q765u$qha$1@news-1.news.gte.net>, davisjf@gte.net says... > >In article , >poirot@wintermute.jsc.nasa.gov says... >> In article <6q5bs6$ddm$1@news-2.news.gte.net>, John F. Davis wrote: >> >I'm new to vxworks, and I'm using the tornado platform on win32. To me >> >it appears slow and not very robust. >> >> It's not just you. A lot of folks think that win32 (95/98/NT) is slow >> and not very robust. >> >> Do you have a specific problem with Tornado or do you just not like >> Windoze? >> >> Me, I prefer a nice fast Ultra 5 running Solaris. Go look up the >> Darwin machines on the Sun web pages. Cheap as a PC, rock solid, >> stable and everything works like it is supposed to... >> >> >Maybe the lock ups are due to a poor port to windows. > >I'm running win95,64 Megs memory, and a p133 cpu. > >I have lock ups, stalls, and problems with the target server. > >I've been debugging and the tornado environment would just hang, or >show a gpf dialog repeatdly. > >Also, just out of the box, the demo program discussed in chapter 2 >of the users guide doesn't work. When I follow along with the book, >it comes to a part which doesn't work. I wonder if my install is >hosed, the code doesn't work for an i960 processor, or if the book/demo >code is just out of sync. I haven't had the target server hang with Windows NT 4.0. As I stated earlier I suspected you were using Win95 and it turns out you are. I believe that the Win95 - Tornado interface is at least part of your problem. --------------------------- Newsgroups: comp.os.vxworks Subject: Problems with VxSim/Tornado under Solaris 2.6/Ultrasparc Date: Tue, 4 Aug 1998 15:11:15 -0400 From: "John H. Zouck" Organization: Johns Hopkins University Applied Physics Lab, Laurel, MD, USA Message-ID: I am running (or trying to run) VxSim under Solaris 2.6 on an UltraSparc dual processor. After I start ./vxWorks on a 'board' (0-15), and shut it down, I can't restart it on that same board again. Says it won't connect via ppp0. However, if I stop /etc/init.d/asppp and restart it, I can restart vxWorks on that board. Anyone know of a bug in solaris/asppp or vxsim that would make it exhibit such behavior? Solutions? This is a very messy workaround (stop/start asppp). - ----------------------------------------------------------- Here is what happens second time thru on same board: - ----------------------------------------------------------- % pushd /export/home/pilot/wind/target/config/solaris % ./vxWorks -r 8000000 -p 1 Target Name: vxTarget Attaching network interface ppp0... ppp0: ppp 2.1.2 started by ppp0: Connect: ppp0 <--> /tyCo/1 ppp0: timeout: could not establish link with peer. Bind failed VxWorks Copyright 1984-1996 Wind River Systems, Inc. CPU: SunOS 5.6 [sun4u] VxWorks: 5.3.1 BSP version: 1.0/1 Creation date: Aug 3 1998 WDB: Ready. - ----------------------------------------------------------- Here's what clears problem: - ------------------------------------------------------------- driver# /etc/init.d/asppp stop driver# /etc/init.d/asppp start - ----------------------------------------------------------- After clearing problem we start OK: - ----------------------------------------------------------- % ./vxWorks -r 8000000 -p 1 Target Name: vxTarget Attaching network interface ppp0... ppp0: ppp 2.1.2 started by ppp0: Connect: ppp0 <--> /tyCo/1 ppp0: local IP address 127.0.1.1 ppp0: remote IP address 127.0.1.254 done. Attaching network interface lo0... done. NFS client support not included. VxWorks Copyright 1984-1996 Wind River Systems, Inc. CPU: SunOS 5.6 [sun4u] VxWorks: 5.3.1 BSP version: 1.0/1 Creation date: Aug 3 1998 WDB: Ready. ============================ John H. Zouck The Johns Hopkins University Applied Physics Laboratory ============================ --------------------------- End of New-News digest ********************** From gema@force.de Wed Aug 5 05:22:55 1998 From: gema@force.de (Gerhard Mayr) Date: Wed Aug 5 05:22:59 PDT 1998 Subject: RE: PPC cPCI CPU - 'CompactPCI-4e': Any comments? Hi, FORCE Computers, the leading supplier of CompactPCI boards and systems offers a broad range of PowerPC products. We have a selection of 3U and 6U boards. The 3U boards are either based on the 603ev or on the 740 processor. The 6U boards are based on the 750 processor at 233 or 300 MHz. The 300 MHz version offers in addition very fast synchronous DRAM technology. For all the boards mentioned, a VxWorks BSP is available. Check out our Web page: http://www.forcecomputers.com for more details. Best regards, Gerhard From bob_anderson@mcg.mot.com Wed Aug 5 05:24:14 1998 From: Bob Anderson Date: Wed Aug 5 05:24:17 PDT 1998 Subject: RE: PPC cPCI CPU - 'CompactPCI-4e': Any comments? -----Original Message----- From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] Sent: Tuesday, August 04, 1998 7:22 PM To: vxworks_users@csg.lbl.gov Subject: RE: PPC cPCI CPU - 'CompactPCI-4e': Any comments? Submitted-by bob_anderson@mcg.mot.com Tue Aug 4 16:21:34 1998 Submitted-by: Bob Anderson Did you look at Motorola's MCP750 Compact PCI processor board? This board is also supported by VxWorks and is currently available from Motorola. Last sentence should have said "BSP is currently available from Motorola." Bob Anderson 410-312-5929 >---------- >From: vxwexplo@lbl.gov >Sent: Tuesday, August 4, 1998 5:18 PM >To: vxworks_users@csg.lbl.gov >Subject: PPC cPCI CPU - 'CompactPCI-4e': Any comments? > >Submitted-by wlbrown@lbl.gov Tue Aug 4 14:18:40 1998 >Submitted-by: Bill Brown > > >We're looking seriously at doing some control systems using 6U cPCI >hardware with a PPC processor. > >One candidate (mainly because they have a vxWorks BSP!) for a cpu >is the 'CompactPCI-4e' from V-I Computer, Inc. Does anyone have any >experience with this unit, and would you care to comment on any "+'s >and -'s" with this platform? > >While I'm at it, the other main contender looks like it's the >'RIOC4063/64" from "CREATIVE ELECTRONICS SYSTEMS." Again, we'd be very >interested in anyones' experience with this product. > From jmettenb@etec.com Wed Aug 5 06:05:50 1998 From: John Mettenburg Date: Wed Aug 5 06:05:53 PDT 1998 Subject: x86 esmc driver options (RX_MODE_ESMC) & NV Boot Parameters Anyone have any details on the following option found in the vxworks 5.3.1 x86 BSP config.h file: #define RX_MODE_ESMC 0 /* 0=interrupt level 1=task level */ We are using an AMPRO P5i board (x86 arch). Also, has anyone implemented a method of saving boot parameters in flash or NV on this board? Thanks in advance, John Mettenburg, Etec (520) 806-3453 From J.Matthew.Wicks@lmco.com Wed Aug 5 06:15:57 1998 From: Matt Wicks Date: Wed Aug 5 06:16:01 PDT 1998 Subject: Establishing Target Connections I am working with a nest of 3 PPC 604E boards running VxWorks 5.3.1. The first board is the system controller and is a proxy server for the other boards which are connected via the shared memory network (SMN). We have multiple nests of this configuration. On most nests I have no problems connecting. On one nest I can successfully connect to the last board (over the SMN) but cannot connect to the system controller or the first SMN board. The tgtsvr log shows: Command: tgtsvr board4 -A -V -n epi006.b4 -f a.out tgtsvr (epi006.b4@tower): Wed Aug 5 08:10:12 1998 License request... authorized on host 'tower'. Attaching backend... succeeded. Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Error: Unable to initialize object module loader library Can anyone explain this or point me in the right direction? TIA -- Matthew Wicks Lockheed Martin Senior Member Engineering Staff Government Electronics Systems (609) 722-2196 199 Borton Landing Road (609) 722-3576 fax Moorestown, NJ 08057 J.Matthew.Wicks@lmco.com From uucp@humle.networkc.net Wed Aug 5 06:46:27 1998 From: uucp@humle.networkc.net Date: Wed Aug 5 06:46:30 PDT 1998 Subject: HDLC on 860 Hello =20 Reply to lars-erik=2Eadolfsson@metric=2Ese not uucp@=2E=2E=2E =20 =20 Have anyone made a VxWorks HDLC driver for the Motorola 860 =20 Regards =20 =20 Lars-Erik Adolfsson =20 Metric=20 =20 Phone +46-629 0331 Fax +46-629 0301 email lars-erik=2Eadolfsson@metric=2Ese =20 From f.pertin@staubli.com Wed Aug 5 23:27:02 1998 From: francois pertin Date: Wed Aug 5 23:27:06 PDT 1998 Subject: Ethernet driver Hi, tornado & vxworks I'm working with tornado 1.0 and a PC486 BSP. I have some trouble with = the ethernet driver, I mean real time problems when I include the = ethernet feature in my application. Does anyone know this problems and maybe have a patch for the BSP ????? Thanks a lot=20 Francois From PierPaolo.Glave@italtel.it Thu Aug 6 00:55:42 1998 From: Pier Paolo Glave Date: Thu Aug 6 00:55:46 PDT 1998 Subject: Problems with C++ templates on VxSim Hello everybody, I'm trying for the first time to use C++ on my environment (VxWorks 5.3.1, Tornado 1.0.1, VxSim for testing on cpu SPARC SOLARIS). It seems to work fine for C++ files which don't use templates; but when I try to download a module which use templates I get this loading error from Wind Shell: WTX Error 0x1002c (LOADER_UNKNOWN_RELOCATION_ENTRY) Here is the sequence of steps I use for the compilation: 1. ccsimso -I [includepath] -DCPU=SIMSPARCSOLARIS \ -fno-builtin -nostdinc -O2 -r \ file1.cc -o file1.o 2. nmsimso file1.o | munch > file2.c 3. ccsimso -traditional -c file2.c 4. ldsimso -r -o final.o file1.o file2.o You can find the source file (file1.cc) at the end of this mail. Can anyone tell if there's something wrong? Thank you -- _____ _ Pier Paolo Glave | __ (_) Italtel - C17/1 | |__) | ___ _ __ via Reiss Romoli | ___/ |/ _ \ '__| 20019 Settimo Mil. (Italy) | | | | __/ | Tel: +39 2 4388.7613 |_| |_|\___|_| Fax: +39 2 4009.7942 mailto:PierPaolo.Glave@italtel.it -- file1.cc: #include #include template class Vector { T* data; int size; public: Vector(); T& operator[] (int); }; template Vector::Vector() { // start off with three elements size =3 ; data = new T [size]; } template T& Vector::operator[] (int n) { int os; int i; T* newdata; // grow vector if more space is needed if (n >= size) { os = size; while (size <= n) { size *= 2; } newdata = new T[size]; for (i=0; i < os; i++) { newdata[i] = data[i]; } delete data; data = newdata; } // return reference to data slot return data[n]; } void appVector() { Vector v; int i; // put data in the vector for (i=1; i<= 5; i++) { v[i] = i *i; } // diisplay data in the vecotr for (i=1; i<= 5; i++) { cout << i << " " << v[i] << endl; } } From daemon@csg.lbl.gov Thu Aug 6 04:02:45 1998 From: daemon@csg.lbl.gov Date: Thu Aug 6 04:02:48 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 6 04:02:43 PDT 1998 Subject: Re: Default compiler switches for PPC860T? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Default compiler switches for PPC860T? Date: 6 Aug 1998 04:19:47 GMT From: "yosiy" Organization: Telematics International, Inc. Message-ID: <01bdc0f2$551ca080$c91eea93@pc-yosiy.ecitele.com> References: <35C6CCBC.7B378B65@ericsson.no.spam.com> <01bdc02c$1e29ecc0$c91eea93@pc-yosiy.ecitele.com> <35C864D7.AA7F21F0@ericsson.no.spam.com> When Tornado creates for you a default makefile, It asks for additional switches. If you do not enter there your special switches manually, they do not be in the makefile. For example, the switch "-g" which is usefull for debug information must be entered to get debug information in the final output. In your 68k example below, there is no "-g" switch, hence no debug information is available in the final output. Yosi Yarchi yosiy@ecitele.com Joachim Fabini wrote in article <35C864D7.AA7F21F0@ericsson.no.spam.com>... > Assume you have a file named hello.cpp and choose "make > current source file" from the Tornado 1.0.1 "Project"-menu. > If you don't have a makefile in the current directory, Tornado > asks you and generates a default makefile for you that contains > the following lines (or something alike): > cc68k -B\\stpc03\sde\sw\pgm\Tornado/host/x86-win32/lib/gcc-lib/ > -mnobitfield -ansi -nostdinc -DRW_MULTI_THREAD -D_REENTRANT -O > yosiy wrote: > > > > I do not know what you means when you are talking > > about the default compiler switches. > > You have to define the switches yourself. > > You can but you need not. WRS provides the default rules, > defines and includes with Tornado. Additional fine-tuning > can, of course, be done by adding your own flags. > > > For *.c files, I'm using the switch "-gstabs+". In some > > cases it let me see additional information using > > XRAY debugger. > > > > Another switch that was recomended by WRS Israel > > is "-dwarf", but I have no experience with it. > --------------------------- End of New-News digest ********************** From endusis@zetnet.co.uk Thu Aug 6 05:23:27 1998 From: "Endusis Ltd." Date: Thu Aug 6 05:23:31 PDT 1998 Subject: PPP for VxWorks Hello Is anyone out there using PPP to access a remote site via a modem? We have set our system up to do this, but the LCP phase of the PPP link negotiaition fails sometimes and succeeds others. Whether it fails or succeeds seems to be pretty well random. Is anyone else having this problem? Matthew Mellor. R&D Lab - Endusis Ltd. Direct phone - +44 (0) 1946 816811 Fax - +44 (0) 7070 765000 From steve@dagwood.kan.marconi.ca Thu Aug 6 08:49:01 1998 From: steve@dagwood.kan.marconi.ca (Steve Macdonald) Date: Thu Aug 6 08:49:04 PDT 1998 Subject: ping utility on vxWorks 5.0.2.b Greetings VxWorkers, I am wondering if anyone has seen this before: I have created a ping utility for vxWorks 5.0.2b. With 5.0.2b, vxworks can respond to ping but there was no ping capability. I created ping for unix boxes, which worked once my userid was root. I could ping any unix box on the network as well as any vxworks box. I ported ping to vxWorks. I cannot get it to ping anything other than another vxworks box. I have tried specifying a port number (or not), setting the ftp userid to root, and a few other things. Any help appreciated, Steve MacDonald Canadian Marconi Company email: steve@kan.marconi.ca Some of the code follows: -------------------- int retval; int selectVal; /* Build destination address. Port number is irrelevant. Command line assumed to be of form "pingit 200.1.1.8" */ dest.sin_family = AF_INET; dest.sin_addr.s_addr = htonl(inet_addr(argv)); /* Create a raw socket using the ICMP protocol */ sock = socket(AF_INET, SOCK_RAW, IPPROTO_ICMP); /* Build ICMP header as part of the outbound packet. We include our process ID so that we can identify our own packets on return */ icp = (struct icmp *) sendpack; icp->icmp_type = ICMP_ECHO; /* Message type */ icp->icmp_code = 0; /* Type subcode */ icp->icmp_cksum = 0; /* Computed later */ icp->icmp_id = OUR_STAMP; /* Our magic ID */ icp->icmp_seq = 1; /* Sequence number */ /* Fill rest of packet with (unused) user data */ for (n = 8 + sizeof(struct timeval); n < 64; n++) sendpack[n] = n; /* Put the checksum into the ICMP header. The system will compute the checksum for the IP layer, which covers only the IP header. chksum() is not shown (See Stevens P. 454) */ icp->icmp_cksum = chksum(icp, 64); /* Send the ICMP echo request packet */ retval = sendto(sock, sendpack, 64, 0, (struct sockaddr *)&dest, sizeof dest); if( retval < 0) { printf("sendto socket error %d\n", retval); exit(0); } Timeout.tv_sec = 5; Timeout.tv_usec = 0; /* Now wait for an ICMP echo reply. We may have to pick up more than one packet because they will not necessarily all be for us if someone else is sending echo requests */ while(1) { fromlen = sizeof from; if( recvfrom(sock, recvpack, sizeof recvpack, 0, (struct sockaddr *)&from, &fromlen) == ERROR ) { printf("socket receive error"); exit(0); } FD_ZERO(&readfds); FD_SET(sock, &readfds); save_readfds = readfds; selectVal = select(sock+1, &readfds, NULL, NULL, &Timeout); if( selectVal == 0 ) { printf("No Response from Host\n"); printErrno(0); return; } else if( selectVal < 0 ) { printf("No Response from Host\n"); printErrno(0); return; } else if( selectVal < 0 ) { printf("select call failed: "); printErrno(0); exit(0); } else { /* The received datagram includes the IP header. We need to get its length in order to step past it. */ ip = (struct ip *)recvpack; /* IP header */ icp = (struct icmp *)(recvpack + (4 * ip->ip_hl)); /* Since we will see ALL incoming icmp packets on our raw socket, we need to check that this is an echo reply and that it belongs to us. */ if(icp->icmp_type != ICMP_ECHOREPLY) { printf("got icmp packet type %d\n", icp->icmp_type); continue; } if( OUR_STAMP != icp->icmp_id ) { printf("do we ever fail the compare?\n"); continue; /* This is NOT our reply! */ } printf("%s is alive \n", argv); break; /* We're done */ } /* end if-else */ } /* end while */ } /* end main */ From mmenge@dspt.com Thu Aug 6 09:54:37 1998 From: mmenge@dspt.com Date: Thu Aug 6 09:54:41 PDT 1998 Subject: Serial Ports Hanging on 2604 Mark Menge@DSPT 08/06/98 12:53 PM on bsp 1.1/4 of vxWorks 5.3.1 for powerpc 2604e, we are having trouble with the task level code eventually getting stuck on a semaphore during a read or write operation (a tt reveals it is stuck in tyWrite). Has anyone else seen this? From mmenge@dspt.com Thu Aug 6 09:57:10 1998 From: mmenge@dspt.com Date: Thu Aug 6 09:57:14 PDT 1998 Subject: Serial Ports Hanging on 2604 Mark Menge@DSPT 08/06/98 12:56 PM on bsp 1.1/4 of vxWorks 5.3.1 for powerpc 2604e, we are having trouble with the task level code eventually getting stuck on a semaphore during a read or write operation (a tt reveals it is stuck in tyWrite). Has anyone else seen this? vxWorks tornado From crimola@praxon.com Thu Aug 6 10:08:36 1998 From: Carlos Rimola Date: Thu Aug 6 10:08:40 PDT 1998 Subject: VXWORKS - SENS DHCP Server Is anyone running the supported DHCP server that comes with SENS? We are running into a problem with it that causes the entire system to lock up. It seems to run fine in a lightly loaded system but we experience the problem when we are running our full set of apps (tasks). I would be interested in hearing from anyone who may have experience this or similar problem and what you did to resolve it. Carlos Rimola Praxon, Inc. From lars.antback@pullmax.se Thu Aug 6 23:15:18 1998 From: lars.antback@pullmax.se Date: Thu Aug 6 23:15:21 PDT 1998 Subject: RE: Serial Ports Hanging on 2604 > Mark Menge@DSPT > 08/06/98 12:56 PM > > on bsp 1.1/4 of vxWorks 5.3.1 for powerpc 2604e, we are having > trouble with the task level code eventually getting stuck on a semaphore > during a read or write operation (a tt reveals it is stuck in > tyWrite). Has anyone else seen this? > > vxWorks tornado Mark, check the SPR below, it solved the problem on my 2603/200MHz system. In the SPR a source code for a new timer driver is included which fixes the problem. /Lars Antb„ck, Pullmax, Sweden ant@pullmax.se >> SPR# 21232 >> >> RTOS: VxWorks >> Ver: 5.3.1 >> Component: tornado-comp-drv >> Release #: 1.0.1 >> Release Status: FCS >> Host: All >> HostOS: >> HostOSRev: All >> Arch Family: PowerPC >> BSP: All >> Ver: N/A >> >> DESCRIPTION >> >> >> The serial channel will occassionally lock up. This is due to a lost >> interrupt event. >> This happens more frequently on very fast CPUs with clock speeds 233MHz >> and higher. >> From malegoas@space.honeywell.com Fri Aug 7 06:02:37 1998 From: "Legoas, Marco A (FL51)" Date: Fri Aug 7 06:02:40 PDT 1998 Subject: How to get filenames VxWorks - Tornado Hello VxWorkers, Assuming that there is only one file inside a directory, and using the = DOS file system,=85 How can I get the filename of that file into a = variable? Thanks in advance. Marco A. Legoas malegoas@space.honeywell.com From bobperry@bellatlantic.net Fri Aug 7 12:24:17 1998 From: "Robert L. Perry" Date: Fri Aug 7 12:24:20 PDT 1998 Subject: Re: How to get filenames At 06:02 AM 8/7/98 PDT, you wrote: >Submitted-by malegoas@space.honeywell.com Fri Aug 7 06:02:37 1998 >Submitted-by: "Legoas, Marco A (FL51)" > >VxWorks - Tornado > >Hello VxWorkers, > >Assuming that there is only one file inside a directory, and using the = >DOS >file system,=85 How can I get the filename of that file into a = >variable? > >Thanks in advance. > >Marco A. Legoas >malegoas@space.honeywell.com +++vxWorks+++ Check out opendir() and readdir() calls. If your file is in a sub-directory, skip the first two readdir()'s to get beyond the dot and dotdot entries. Bob From daemon@csg.lbl.gov Sat Aug 8 04:00:12 1998 From: daemon@csg.lbl.gov Date: Sat Aug 8 04:00:16 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Aug 8 04:00:09 PDT 1998 Subject: 5.3 VxWorks BSP for 486 - Setting System Clock Rate Subject: Re: SENS Network Stack numbers Subject: Ethernet collisions ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: 5.3 VxWorks BSP for 486 - Setting System Clock Rate Date: 28 Jul 1998 20:07:57 -0700 From: Tom Haley Organization: haleyt@ssd.fsi.com Message-ID: <6pm3md$nh4@drn.newsguy.com> I am new to VxWorks. I am using WindView, Tornado 1.0.1, and VxWorks 5.3. The host platform is Windows NT 4.0 and the target machine is a PC with a Pentium 166 MHZ processor. I want to Set the system clock rate to some multiple of 240 Hz. I have a scheduler app which I ported to VxWorks. I am using the POSIX timer functins. I am using the following Ada code to start the timer and run it periodically. Timer_Period_Values.it_interval.tv_sec := 0; Timer_Period_Values.it_value.tv_sec := 0; Timer_Period_Vals.it_interval.tv_nsec := integer( Timer_Period_In_Nanoseconds ); Timer_Period_Vals.it_value.tv_nsec :=integer(Timer_Period_In_Nanoseconds ); Timer_Start_Status := timer_settime( timerid => Real_Time_Control_Timer, flags => 0, value => Timer_Period_Values_Ptr, ovalue => NULL ); When I looked at my application with WindView every thing looked ok except for the various task rates. I have a 1Hz task, a 2 Hz task, a 4 Hz task and an 8Hz task. WindView indicated that my tasks were 4 times as slow as I wanted them to be. In other words my 1Hz task was running at .25 Hz, my 2 Hz task was running at .5 Hz, my 4 Hz task is running at 1 Hz and my 8hz task is running at 2Hz. I reviewed the code and I didn't see anything obviously wrong. I figured it was either my code or something in the BSP build. So I changed Timer_Period_In_Nanoseconds so that at was multiplied by 4 and looked at the timing again with WindView. Same rates for the tasks .25 Hz, .5Hz, 1 Hz and 2Hz. I concluded that it must be something in my build. Sure enough I went to the file usrConfig.c file and saw the following line of code: sysClkRateSet ( 60 ); This was totally consistent with the task rates that I was seeing. Therefore I changed this line of code to: sysClkRateSet ( 480 ); Now my task rates are 2/3 Hz, 4/3Hz, 8/3 Hz, and 16/3 z. This is bizarre. I poked around some more and saw that the value of 60 was also hardcoded in i8253Timer.c although it appears to me that sysClkRateSet will overwrite the default. Just to be on the safe side I changed the constant to 480. My question is two fold: 1) Am I going about changing the clock rate correctly? 2) What am I missing about changing this clock rate? Thanks in advance. Tom Haley Flight Safety International. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SENS Network Stack numbers Date: Fri, 07 Aug 1998 19:33:55 -0700 From: hjb@pso.com (Hwa-Jin Bae) Organization: The DeRome Mansion Message-ID: References: > By poor I mean ~20 Mbps TCP and 40 Mbps UDP. To > contrast, two Alphas on the same switch get ~90 Mbps TCP. > > Has anyone done any performance tests on the StrongARM platform? I'm > curious to see what results other people have gotten. Your numbers are similar to the numbers I have seen on a similar EV285 style SA110/233mhz designs. You are using 21143 as your ethernet? In contrast, the original BSD stack running as part of NetBSD on identical hardware can do about 70Mbps TCP. --------------------------- Newsgroups: comp.os.vxworks Subject: Ethernet collisions Date: Sat, 08 Aug 1998 09:34:57 GMT From: whd@directors-1.zetnet.zet (Ag@whd) Organization: White Horse Design Message-ID: <35cf1b50.1720539@znews.zetnet.co.uk> Reply-To: whd@directors-1.zetnet.zet We're running a Power PC (MV board) connected to a PC via a 4 port hub (using only 2 of it's ports). We find that we're getting huge numbers of collisions on transmitting to the PC. We think we need a "better" driver from Motorola but I'm not so sure there isn't something "configured" wrong. (I've onlu just started this job). Any idea's of where to start? 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 - --- Designers of Voice recognition systems for use in helicopters (inc. EC135 & BK117) --------------------------- End of New-News digest ********************** From daemon@csg.lbl.gov Sun Aug 9 04:00:19 1998 From: daemon@csg.lbl.gov Date: Sun Aug 9 04:00:23 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Aug 9 04:00:15 PDT 1998 Subject: Re: 860FADS BSP ethernet MAC question Subject: Re: 860FADS BSP ethernet MAC question Subject: PowerPC question ? Subject: Re: #pragma align Subject: MPC860 Ethernet problems Subject: Re: Finding PC of instruction which caused exception/signal Subject: gnat for Vxworks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: 860FADS BSP ethernet MAC question Date: Sat, 8 Aug 1998 12:31:18 -0000 From: "Geoff" Organization: Virgin News Service Message-ID: <6qhd4j$9h9$1@nclient5-gui.server.virgin.net> References: Hello Pete We completed an 821/860 VxWorks project about 6 months ago. For the problem you described, we grouped a bunch of board specific parameters namely MAC address , IP Address, Gateway Address, & some application specific data,etc into one Flash Sector out of the way at the top of the flash Rom. We put our default IP and MAC in syslib.c or wherever it was, and built our default flash code. We then wrote a utility program that runs on a PC to talk to the target board to allow the user to change the MAC, IP Gateway or whatever. After reflashing the data sector a reboot was triggered, then the new MAC etc is used. The code on the target was fairly simple to do just a TCP Server task waiting on a Service port connection request, with a simple protocol for data transfer. The PC code was also easy, if you are using Win95 or WinNT Visual Basic is recommended, the socket code is a doddle, much simpler than Visual C++. As VB5 was so easy , we expanded the utility to also add complete Software reflashing(App & OS). This was tricky to code on the target, but once we cracked the problem, it worked really well. If you want any further info on any of this, get back to me. Regards Geoff Pete Kockritz wrote in message ... >We're using a Motorola 860FADS board to do some prototyping until we >get our actual target board. We have the fads board flash programmed >with vxWorks bootrom code. Since the fads board does not have any >NVRAM on it, the bootrom code must contain the ethernet MAC address >for the board. This all works fine until we generated a bootrom for >another board, now our sysLib.c file has the MAC address of the >second board. So when we build a vxWorks kernel to download to the >first board, its MAC address changes to the address we assigned to the >second board. > >What we need is either to: > >1) configure the kernel to NOT re-initialize the MAC address when it > boots up. This way the flash contains the unique value for the board. > >2) allow the kernel to re-init the MAC address, but get the value > for the MAC address from the flash. > >Or maybe there are other ways. It seems like it will be a real pain >keeping up with which sysLib.c to use for which board when several >different developers are building kernels for testing. Hopefully in the >near future we won't be building kernels as often. > >Regards, >Pete >-- >+-----------------------------------------------------------------+ >| Pete Kockritz mailto:petekz@cyberramp.net | >| This space for rent. | >+-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 860FADS BSP ethernet MAC question Date: Sat, 8 Aug 1998 12:31:18 -0000 From: "Geoff" Organization: Virgin News Service Message-ID: <6qhdhh$ib9$1@nclient3-gui.server.virgin.net> References: Hello Pete We completed an 821/860 VxWorks project about 6 months ago. For the problem you described, we grouped a bunch of board specific parameters namely MAC address , IP Address, Gateway Address, & some application specific data,etc into one Flash Sector out of the way at the top of the flash Rom. We put our default IP and MAC in syslib.c or wherever it was, and built our default flash code. We then wrote a utility program that runs on a PC to talk to the target board to allow the user to change the MAC, IP Gateway or whatever. After reflashing the data sector a reboot was triggered, then the new MAC etc is used. The code on the target was fairly simple to do just a TCP Server task waiting on a Service port connection request, with a simple protocol for data transfer. The PC code was also easy, if you are using Win95 or WinNT Visual Basic is recommended, the socket code is a doddle, much simpler than Visual C++. As VB5 was so easy , we expanded the utility to also add complete Software reflashing(App & OS). This was tricky to code on the target, but once we cracked the problem, it worked really well. If you want any further info on any of this, get back to me. Regards Geoff Pete Kockritz wrote in message ... >We're using a Motorola 860FADS board to do some prototyping until we >get our actual target board. We have the fads board flash programmed >with vxWorks bootrom code. Since the fads board does not have any >NVRAM on it, the bootrom code must contain the ethernet MAC address >for the board. This all works fine until we generated a bootrom for >another board, now our sysLib.c file has the MAC address of the >second board. So when we build a vxWorks kernel to download to the >first board, its MAC address changes to the address we assigned to the >second board. > >What we need is either to: > >1) configure the kernel to NOT re-initialize the MAC address when it > boots up. This way the flash contains the unique value for the board. > >2) allow the kernel to re-init the MAC address, but get the value > for the MAC address from the flash. > >Or maybe there are other ways. It seems like it will be a real pain >keeping up with which sysLib.c to use for which board when several >different developers are building kernels for testing. Hopefully in the >near future we won't be building kernels as often. > >Regards, >Pete >-- >+-----------------------------------------------------------------+ >| Pete Kockritz mailto:petekz@cyberramp.net | >| This space for rent. | >+-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: PowerPC question ? Date: Sat, 8 Aug 1998 12:36:54 -0000 From: "Geoff" Organization: Virgin News Service Message-ID: <6qhdhm$ib9$2@nclient3-gui.server.virgin.net> Hi All As there are quite a few people here using PowerPC860' s etc , I was just wondering if anyone has any experience of using the powerPC 801 chip ? I would be very interested to compare notes. Regards Geoff --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #pragma align Date: Wed, 29 Jul 1998 09:51:21 +0200 From: "Fredrik Jonsson" Organization: Ericsson Message-ID: <6pmjq0$i0o@newstoo.ericsson.se> References: <35B8891B.8528362A@ttc.com> <6ph8bk$6r8@newstoo.ericsson.se> <35BE05BC.F16C203B@dragon.decnet.lmco.com> Joe Durusau wrote in message <35BE05BC.F16C203B@dragon.decnet.lmco.com>... >You should be aware that, for certain targets, not aligning things > (for instance,not aligning a 4-byte object on a 4-byte boundary) > will cause a bus error. >That is, in general that the pragma has sometimes been disabled. You are probably right, but struct RegisterSet { unsigned char a; }; struct RegisterSet r[5]; if(r[3].a) { } wouldn't it be fare if the compiler could generate 1 byte structs plus not causing a bus error in the above example? Not being able to control the code generation for structs/unions, so that they (even ineffectively) can be used on odd addresses, is really just an unnecessary limitation. So WRS, please give full padding, alignment and packing control back to the compiler again. :) - -- Fredrik Jonsson PS! Another thing about this: what about the stack? If the core causes a bus error on odd addresses for 4-byte objects, does this force the compiler to push minimum of 4-byte objects on the stack even for chars? I can imagine some typecast situations where this becomes a dangerous problem. --------------------------- Newsgroups: comp.os.vxworks Subject: MPC860 Ethernet problems Date: Wed, 29 Jul 1998 11:00:41 +0100 From: "Leonel Domingues" Message-ID: <6pms40$6hs$1@duke.telepac.pt> Please help me. I am developing a board based on MPC860SAR at 50 MHz, and Im experiencing some problems related with the Ethernet connection. I am using the EEST MC68160, and I have connected its pins as following: RCLK - PA[6]/CLK2 TCLK - PA[4]/CLK4 RX - PA[15]/RXD1 TX - PA[14]/TXD1 TENA - PB[19]/RTS1 TPSQEL - PC[14] TPFULDL - PC[13] LOOP - PC[12] RENA - PC[10]/CD1 CLSN - PC[11]/ CTS1 I have made my BSP reflect the above connections. Im using the Ethernet driver as it is on ADS860 board. Ive changed the sysCpmEnetEnable function as follow. /*************************************************************************** ******/ STATUS sysCpmEnetEnable ( int unit /* not used - only slave SCC1 is wired to port */ ) { int immrVal = vxImmrGet(); /* Configure Ports PA15, PA14, PA6 and PA4 for RXD1, TXD1, CLK2 and CLK4, respectively */ *PAPAR(immrVal) |= 0x0A03; *PADIR(immrVal) &= ~(0x0A03); /* Enable CLSN and RENA (pg.16-335, 16-465 of MPC860 Users Manual) */ *PCPAR(immrVal) &= ~(0x0030); *PCDIR(immrVal) &= ~(0x0030); *PCSO(immrVal) |= 0x0030; /* Enable other ethernet functions in PC12, PC13, PC14 */ *PCDIR(immrVal) |= 0x000e; *PCDAT(immrVal) |= 0x0006; *PCDAT(immrVal) &= ~(0x0008); /* Config CLK4 as Tx clock and CLK2 as Rx clock, both for SCC1. Use NMSI mode */ *SICR(immrVal) |= 0x2F; /* connect SCC1 clocks */ *SICR(immrVal) &= ~(0x40); /* Force NMSI mode for SCC1 */ /* To enable the ethernet, ie, enable the SIA (EEST) we have to configure PA1 as a general purpose register and assign it zero tsa, tem que ser o pino 1*/ *PAPAR(immrVal) &= ~(0x4000); *PADIR(immrVal) |= 0x4000; *PADAT(immrVal) &= ~(0x4000); /* enable Ethernet */ /* Configure TENA on RTS1 */ *PBPAR(immrVal) |= 0x00001000; /* set port B -> RTS1 = TENA */ *PBDIR(immrVal) |= 0x00001000; return (OK); } /*************************************************************************** ******/ At this moment I have my board perfectly sending ethernet packets. I am sniffing the ARP packets that the board sends to the Network before loading VxWorks from server. The problem is that I dont get any packets received. I have debugged its hardware and I know that the packets are arriving to the processor pins. The pins that I use are different from those used on ADS860. Is there any problem on choosing different pins. I think that if I route pins to SCC than if the driver is only SCC dependent there is no problem. Although, there are tree pins that are general propose. Is the driver using this pins? Does the drivers excepts me to use the same as in ADS860? Please give me some clue. Thanks, Leonel Domingues --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Finding PC of instruction which caused exception/signal Date: Tue, 28 Jul 1998 18:30:46 -0700 From: "Steve Doiel" Message-ID: <35be7c34.0@news.pacifier.com> References: <6p9h22$b69$1@news3.global-ip.net> <6pa1pr$g5u$1@news3.global-ip.net> <6plj6m$n2p$1@strato.ultra.net> >Hello, > > I would like not to only get the PC instruction that caused > the exception but to unwrap the stack that go me there. > > Would you know how to attack this one ? > > BSP that I am working on is PC486, I am hoping it > does not make a difference. > > Thanks in advance, > > Igor > I was initially looking for the same thing, but then I learned that if you don't include a signal handler then VxWorks will automagically suspend the task with the exception. If you include a target shell and debugger (the target debugger is quite primitive), you can list tasks to find on what task is suspended, get the name and/or task id of the offending task. Use the 'ti' command to get the program counter where the exception occurred. Use 'lkAddr' to find the symbols closest to that address to give you a frame of reference. And the 'l' to disassemble the code to figure out exactly where the exception occurred. There are other things that are handy in the target shell as well (such as checkStack). I hope this helps, SteveD --------------------------- Newsgroups: comp.os.vxworks Subject: gnat for Vxworks Date: Wed, 29 Jul 1998 20:25:53 -0500 From: "Dave Hanson" Organization: NASA, Kennedy Space Center Message-ID: <6pn7rt$2j7@niven.ksc.nasa.gov> Does anyone know of a gnat port to Vxworks? - --or-- Does anyone know how to build gnat for Vxworks? Dave Hanson Systems Engineering GeoControl Systems, Inc (281) 333-2561 --------------------------- End of New-News digest ********************** From roghol@lin.foa.se Mon Aug 10 00:15:30 1998 From: Roger Holmquist Date: Mon Aug 10 00:15:33 PDT 1998 Subject: MVME2700_Do_I_Need_A_VxWorks_Prom? We have just got a VMVE2700-card and I am now trying to run it in our Tornado (UNIX version) environment. As far as I remember (berfore a very loong holiday!) you dont need to replace the bug-prom in order to boot the card. Is that correct and if so, is it desirable? (There is some debugging and HW-test tools in the Motorola PPCBUG who might be usable....) If there is anybody out there who have NIOT-parameters (if relevant) as an example for the card, I would be happy to receive them! Greetings / Roger # -----------------------------------------------------------------# # Roger Holmquist, FOA, National Defence Research Establishment # # sect 750, Olaus Magnus vag 42, Box 1165, 581 11 LINKOPING ------# # --------------------------- SWEDEN ------------------------------# # Phone: +46 13-378382 --------------------------------------------# # Fax: +46 13-378009 ----------------------------------------------# # Home: +46-13-214733 Mobile +46-706-250123 ----------------------# # Email: roghol@lin.foa.se ----------------------------------------# # -----------------------------------------------------------------# 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 MSchrape@atomika.com Mon Aug 10 05:51:27 1998 From: Martin Schrape Date: Mon Aug 10 05:51:31 PDT 1998 Subject: Re: How to get filenames At 06:02 8/7/98 PDT, you wrote: >Submitted-by malegoas@space.honeywell.com Fri Aug 7 06:02:37 1998 >Submitted-by: "Legoas, Marco A (FL51)" > >VxWorks - Tornado > >Hello VxWorkers, > >Assuming that there is only one file inside a directory, and using the = >DOS >file system,=85 How can I get the filename of that file into a = >variable? > Use readdir(opendir(dirName)). See e.g. ls() in usrLib.c. -martin From Edwards.John@oscsystems.com Mon Aug 10 05:57:10 1998 From: Edwards.John@oscsystems.com Date: Mon Aug 10 05:57:13 PDT 1998 Subject: Support for inline & the packed attribute?? Does VxWorks support the gcc non-ansi inline type and the gcc packed attribute? Thanks, John Edwards Orbital Sciences Corporation / Fairchild Defense johnedwards@oscsystems.com From chenyc@fnal.gov Mon Aug 10 08:15:34 1998 From: Yenchu Chen Date: Mon Aug 10 08:15:37 PDT 1998 Subject: problem in setting up SCSI Hi, I got MVME167 and MVME712. Three tape drives are connected to the SCSI bus of MVME712. It is VxWorks 5.3. SUN workstation is used as host computer. The following flags are defined. INCLUDE_SCSI INCLUDE_SCSI2 SYS_SCSI_CONFIG INCLUDE_TAPEFS I also modified the sysScsi.c to initialize the tape drive which has SCSI ID set to 2. At booting up it shows "Could not set target option parameters". Is there anything wrong in this procedure? ************************************************************************* scsiId = 2; which = SCSI_SET_OPT_XFER_PARAMS | SCSI_SET_OPT_WIDE_PARAMS; options.maxOffset = SCSI_SYNC_XFER_ASYNC_OFFSET; options.minPeriod = SCSI_SYNC_XFER_ASYNC_PERIOD; options.xferWidth = SCSI_WIDE_XFER_SIZE_NARROW; if (scsiTargetOptionsSet (pSysScsiCtrl, scsiId, &options, which) == ERROR) { printErr ("Could not set target option parameters\n"); return (ERROR); } *************************************************************************** Or I simply have problem with my SCSI cable? Any comments/suggestions will be appreciated. Best regards, Yen-Chu Chen chenyc@fnal.gov (630) 840-8871 (experiment) (886)-(2) 2789-9681 (Inst. of Phys., Academia Sinica) From McGarry_Mike@timeplex.com Mon Aug 10 11:13:47 1998 From: "Mike McGarry" Date: Mon Aug 10 11:13:51 PDT 1998 Subject: GNU linker This is a multi-part message in MIME format. --------------91421F44F6CD77EDF9206163 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello, Has anyone had any Memory fault problems with the GNU linker packaged by Wind River? I keep getting various memory faults caused by the GNU linker when trying to build my target image. VxWorks, Tornado Thanks, Michael McGarry TimePlex Group --------------91421F44F6CD77EDF9206163 Content-Type: text/x-vcard; charset=us-ascii; name="mcgarry_mike.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Michael McGarry Content-Disposition: attachment; filename="mcgarry_mike.vcf" begin:vcard n:McGarry;Michael x-mozilla-html:FALSE org:TimePlex Group adr:;;;Woodcliff Lake;NJ;07675;USA version:2.1 email;internet:mcgarry_mike@timeplex.com title:Engineer II x-mozilla-cpt:;0 tel;work:(201) 391-1111 x4693 fn:Michael McGarry end:vcard --------------91421F44F6CD77EDF9206163-- From McGarry_Mike@timeplex.com Mon Aug 10 11:17:58 1998 From: "Mike McGarry" Date: Mon Aug 10 11:18:02 PDT 1998 Subject: Re: BSP questions This is a multi-part message in MIME format. --------------18A85D0CE3AEEC637C984877 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Mike, Unfortunately, I don't know the answer to your first question but I can help you with your second question. In usrConfig.c, there is a routine called usrRoot, which is the first function to execute once the kernel has been started. Inside this routine there are system calls which initialize the I/O system and then the CONSOLE tty is tied to standard in, out, and error. After this point you can use the printfs. If you want to spit out characters in sysHwInit you are going to have to hand feed characters one at a time to the SCCs TX register and you have to be aware of the timing to insure you are not overwriting a character in the TX register before the SCC has had a chance to put it out on the physical interface. Regards, Michael McGarry TimePlex Group VxWorks, Tornado the vxWorks Users Group Exploder wrote: > Submitted-by mcuddy@fensende.com Fri Jul 31 16:12:13 1998 > Submitted-by: Mike Cuddy > > 1) what are the ramifications of replacing the SENS driver with the > old-style driver? how do I kick out the DEC driver (used by the rom > vxworks for booting) and replace it with the AMD ethernet driver... > > 2) what's the earliest that I can get printf()'s out the console > port in the BSP bootup cycle? Optimally, I'd really like to be > able to get some debugging information out of the BSP in sysHwInit(), > unfortunately, as I understand it, that's before the I/O system is > initialized, so I can't just print out information there. Tornado 1.0.1 --------------18A85D0CE3AEEC637C984877 Content-Type: text/x-vcard; charset=us-ascii; name="mcgarry_mike.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Michael McGarry Content-Disposition: attachment; filename="mcgarry_mike.vcf" begin:vcard n:McGarry;Michael x-mozilla-html:FALSE org:TimePlex Group adr:;;;Woodcliff Lake;NJ;07675;USA version:2.1 email;internet:mcgarry_mike@timeplex.com title:Engineer II x-mozilla-cpt:;0 tel;work:(201) 391-1111 x4693 fn:Michael McGarry end:vcard --------------18A85D0CE3AEEC637C984877-- From McGarry_Mike@timeplex.com Mon Aug 10 11:22:08 1998 From: "Mike McGarry" Date: Mon Aug 10 11:22:12 PDT 1998 Subject: Re: Support for inline & the packed attribute?? This is a multi-part message in MIME format. --------------41D187940EBAB1558FDF4B67 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit > Submitted-by Edwards.John@oscsystems.com Mon Aug 10 05:57:10 1998 > Does VxWorks support the gcc non-ansi inline type and the gcc packed > attribute? John, The version of gcc that is packaged with Tornado does support the packed attribute and the inline type. You can check the "GNU Toolkit 2.7" manual's chapter on gcc for further confirmation and details. Michael McGarry TimePlex Group --------------41D187940EBAB1558FDF4B67 Content-Type: text/x-vcard; charset=us-ascii; name="mcgarry_mike.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Michael McGarry Content-Disposition: attachment; filename="mcgarry_mike.vcf" begin:vcard n:McGarry;Michael x-mozilla-html:FALSE org:TimePlex Group adr:;;;Woodcliff Lake;NJ;07675;USA version:2.1 email;internet:mcgarry_mike@timeplex.com title:Engineer II x-mozilla-cpt:;0 tel;work:(201) 391-1111 x4693 fn:Michael McGarry end:vcard --------------41D187940EBAB1558FDF4B67-- From ThanosJH@Npt.NUWC.Navy.Mil Tue Aug 11 05:57:19 1998 From: Thanos John H NUWCDIVNPT Date: Tue Aug 11 05:57:23 PDT 1998 Subject: VxWorks DOS filesystem on a 1 GByte SCSI Jaz disk drive A few weeks ago, I posted a question to the VxWorks interest group, asking if anyone had successfully installed a VxWorks DOS filesystem on a 1 GByte SCSI Jaz drive that was compatible with the MS-DOS filesystem. Brian Palmer of BBN, Middletown, RI (bpalmer@bbn.com), was gracious enough to respond to my posting, and to share the source code to his modified usrScsi.c file. I have verified that the concepts involved work with the NCR 53C710 SCSI I/O processor on my Motorola MVME167 single board computer, by incorporating them into my own version of the usrScsi.c file. The essence of the solution in the usrScsi.c file is to treat the SCSI Jaz drive not as a removable floppy disk drive (i.e., there's no need to call scsiPhysDevCreate(), scsiModeSelect(), scsiPhysDevDelete(), and another scsiPhysDevCreate(), the way you must for a SCSI floppy disk drive), but in a way similar to installing a DOS filesystem on a SCSI hard disk drive. Shown below is a code fragment that illustrates how you need to modify the usrScsi.c file before rebuilding VxWorks, in order to install a VxWorks DOS filesystem on a 1 GByte Jaz disk drive at VxWorks initialization. Note that the second and third arguments to the scsiBlkDevCreate() function are the number of 512 KByte blocks available (2090976) and the number of sectors (32, or 0x20) to skip, respectively. These sectors that are skipped store the MS-DOS partition table, which is not supported by the VxWorks DOS filesystem for removable drives. /* Start of code fragment illustrating how to install a VxWorks DOS fs on an Iomega 1 GByte SCSI Jaz drive */ /* The following should be inserted in the variable declaration section: */ SCSI_PHYS_DEV * pSpd50; /* pointer to SCSI_PHYS_DEV structure, ID = 5, LUN = 0 (1 GB Jaz drive) */ BLK_DEV * pSbdJ0; /* pointer to SCSI_BLK_DEV structure for 1 GByte SCSI Jaz disk drive */ /* The following should be inserted after the compiler directive, #else /* !SCSI_AUTO_CONFIG */ /* configure Jaz disk drive at busId = 5, LUN = 0, and install DOS file system. */ if ((pSpd50 = scsiPhysDevCreate (pSysScsiCtrl, 5, 0, 0, NONE, 0, 0, 0)) == (SCSI_PHYS_DEV *) NULL) { printErr ("usrScsiConfig: scsiPhysDevCreate for Jaz disk drive failed.\n"); return (ERROR); } else if ((pSbdJ0 = scsiBlkDevCreate (pSpd50, 2090976, 0x20)) == NULL) { printErr ("usrScsiConfig: scsiBlkDevCreate for Jaz disk drive failed.\n"); return (ERROR); } else printf("usrScsiConfig: scsiPhysDevCreate and scsiBlkDevCreate for Jaz disk drive succeeded.\n"); if (dosFsDevInit ("/sd1/", pSbdJ0, NULL) == NULL) { printErr ("usrScsiConfig: dosFsDevInit for Jaz disk drive failed.\n"); return (ERROR); } else printf("usrScsiConfig: dosFsDevInit for Jaz disk drive succeeded.\n"); /* End of code fragment illustrating how to install a VxWorks DOS fs on an Iomega 1 GByte SCSI Jaz drive */ The code fragment shown above assumes that a Jaz disk has been formatted in a SCSI Jaz drive connected to a PC, using the Iomega Tools, before the Jaz disk was inserted in a Jaz drive connected to the processor executing VxWorks. The NULL value for the third argument to the dosFsDevInit() function indicates that VxWorks should initialize a DOS_VOL_CONFIG structure with values read from the boot sector of the PC-formatted Jaz disk. You should know that transferring data to a Jaz drive is relatively slow. I'm not sure, but I think that this is because Iomega did not implement a true SCSI2-compliant interface. It takes my system about 6 minutes to write 120 MBytes of data to a single DOS file on a Jaz drive. If you want to know more about VxWorks DOS filesystem and MS-DOS filesystem compatibility, read Wind Tech Note 34, available at Wind River System's Web site. You first need to get access to the WindSurf area with a valid userid and password, which is provided by WRS when you obtain a maintenance license. For reasons I cannot fathom, WTN34 is not listed in the Wind Tech Notes area. However, if you enter Keyword Search, and search on the string "MS-DOS and VXWORKS", you will be shown a hypertext link to WTN34.ascii. If you select that, up pops a text file with the hidden secrets to MS-DOS and VxWorks DOS filesystem compatibility. Oh by the way, if you want to install a VxWorks DOS filesystem on a SCSI floppy drive, I have found that the source code in the WRS version of usrScsi.c works, except for one little detail. If you are using a 5.25" 1.2 MByte SCSI floppy drive, you must stuff 0x1b in modeData[1], and use scsiBlkDevInit ((SCSI_BLK_DEV *) pSbdFloppy, 15, 2); However, if you are using a 3.5" 1.44 MByte SCSI floppy drive, you must stuff 0x28 in modeData[1], and use scsiBlkDevInit ((SCSI_BLK_DEV *) pSbdFloppy, 18, 2); The WRS version of the file usrScsi.c never mentions that modeData[1] = 0x1b only works for 5.25" 1.2 MByte floppy drives. I hope that this information is useful to fellow VxWorkers. I sure wish I had it a month ago. -- John Thanos, Naval Undersea Warfare Center, Code 3124, Newport, RI -- thanosjh@npt.nuwc.navy.mil From mcuddy@fensende.com Tue Aug 11 10:35:58 1998 From: Mike Cuddy Date: Tue Aug 11 10:36:02 PDT 1998 Subject: VGA code / driver for VxWorks.. Keywords: StrongARM, PCI, VGA, VxWorks 5.3.1, Tornado 1.0.1 I'm using VxWorks 5.3.1 / Tornado 1.0.0 on a StrongARM platform and was wondering if anyone had any basic PCI VGA code. (for any chip ... I'll get the right card ;-) Since on PC's, it's the SVGA bios that takes care of setting up the CRTC chip to get just a plain "text mode" running, there's precious little documentation on how to do that. I'd like to have access to the std. text mode, but my end goal is to have some code which will let me access an SVGA card in 800x600x24 bit mode over the PCI bus. I've seen the code for the pcconsole that comes with the WRS BSP DDK, but it relies on the fact that the bios has already setup text mode. Any help will be greatly appreciated ... -- Mike Cuddy (mcuddy@FensEnde.com, MC312), Programmer, Daddy, Human. Fen's Ende Software, Redwood City, CA, USA, Earth, Sol System, Milky Way. I remember asking why ... Let it rain, and protect us from this Cruel Sun. Join CAUCE: The Coalition Against Unsolicited Commercial E-mail. From daemon@csg.lbl.gov Wed Aug 12 04:00:11 1998 From: daemon@csg.lbl.gov Date: Wed Aug 12 04:00:15 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 12 04:00:08 PDT 1998 Subject: WANTED: Israeli VxWorks Specialist Subject: Driver for PC Soundcard ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: WANTED: Israeli VxWorks Specialist Date: Wed, 12 Aug 1998 10:17:39 +0300 From: Leonid Rosenboim Organization: Wind River Systems - Israel Message-ID: <35D14193.307BB0AD@wrs.co.il> Wind River Systems office in Israel is seeking a highly motivated software engineer (B.Sc. CS/CE/EE or equivalent) with 2-3 years experience using VxWorks and Tornado, fluent in Hebrew and good English. We have one open position for a Field Application Engineer, who's primary job is to work with Sales prividing pre-sales technical support as well as field consulting for customers. Appropriate candidates are kindly asked to e-mail theur Resume to jobs@wrs.co.il, or send fax to the number below. - -- - ----------------------------------------------------------------------- | Leonid Rosenboim | | Eng Mgr / Proj Lead | Fax: +972-3-559-8244 | Wind River Systems Israel | E-Mail: leonid@wrs.co.il | P.O.Box 11502, AZUR 58017, Israel | WWW: http://www.wrs.com - ----------------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Driver for PC Soundcard Date: Wed, 12 Aug 1998 11:56:45 +0200 From: Erik Wischmann Organization: Euro Telematik GmbH Message-ID: <35D166DD.1E690E95@vs.dasa.de> Hi, we are looking for a driver for a SoundBlaster compatible Soundcard used in a 486 environment. Does anyone know if such a driver has already been developed? We only need the output of digitized voice, no FM, MIDI or recording. Thank you Erik Wischmann - --------------------------------------------- Dipl.-Ing. Erik Wischmann Manager System Engineering Euro Telematik GmbH Woerthstrasse 85 D-89077 ULM / Germany Phone: +49-731-392-4478 Fax: +49-731-392-5592 Mobile: +49-171-3150904 E-mail: erik.wischmann@vs.dasa.de - --------------------------------------------- --------------------------- End of New-News digest ********************** From dillow@aptec.com Wed Aug 12 05:34:14 1998 From: Mike Dillow Date: Wed Aug 12 05:34:17 PDT 1998 Subject: Re: VGA code / driver for VxWorks.. I ran into a similar problem finding basic support for an SVGA IP Module some time back. I found out that SBS/Greensprings has a Software Port Package which includes source code that provides the necessary information to initialize the chip, and load the tables which are used for character generation in the basic text modes. It also provides initialization for several graphics modes. I had to do some tweaking to convert it to run under VxWorks (their code is for OS-9). It saved me a lot of time and headache. I don't remember the cost, but I don't think it was too bad. Mike Dillow Applied Technology Associates 1900 Randolph, SE Albuquerque, NM 87106 dillow@aptec.com From dsunnarb@wavetech.net Wed Aug 12 05:41:22 1998 From: Dan Sunnarborg Date: Wed Aug 12 05:41:26 PDT 1998 Subject: VxWorks Disks: ATA vs IDE Hello, I am wondering what the differences are between the IDE and ATA drivers? When should each of them be used? What are their requirements? I am using 5.3.1 of VxWorks. I am using a standard Seagate 2 GB IDE drive on a 486 target, should I be using the IDE or ATA driver calls? Thanks! Dan Sunnarborg Software Engineer Applied Biometrics, Inc. From chuckvb@ix.netcom.com Wed Aug 12 12:38:16 1998 From: Chuck Van Blaricom Date: Wed Aug 12 12:38:19 PDT 1998 Subject: Re: GNU linker Tornado and vxWorks Response to GNU Linker Our linker (ld960) crashes every time we link and have 2 or more modules with the same name, or we include 2 files that define the same procedure. Hope this helps Chuck > Submitted-by McGarry_Mike@timeplex.com Mon Aug 10 11:13:47 1998 > > Hello, > > Has anyone had any Memory fault problems with the GNU linker packaged by > Wind River? I keep getting various memory faults caused by the GNU > linker when trying to build my target image. > > VxWorks, Tornado > > Thanks, > > Michael McGarry > TimePlex Group > -- Chuck Van Blaricom ComTier, 2525 Walsh Ave, Santa Clara, CA 95051 (408) 235-7039 FAX: (408) 235-7085 EMail: chuckvb@ix.netcom.com From kernin@research.moore.com Wed Aug 12 15:38:32 1998 From: kernin@research.moore.com (Kevin Kernin) Date: Wed Aug 12 15:38:35 PDT 1998 Subject: reading real-time clock I need to get the current date and time in our application running vxworks. We are using the MVME 2604 board with the 1.1/2 version of the BSP and the MVME 2306 board with the 1.1/4 version of the BSP. Using the PPCbug command "time" I verified that we set the real-time clock properly. I then switched the jumper and booted up with vxWorks. I made the following calls to get the date and time: clock_gettime(CLOCK_REALTIME,mytimespec) gmtime( (time_t *) &mytimespec) when I check the tm struct that the call to gmtime returns I found unexpected values as shown below: time in seconds= 0 seconds= 0 minutes= 0 hours= 0 day of month= 1 month= 0 year= 70 dayofweek= 4 The time does increase when I rerun my code, but if I reset the PPC board and run I always start with the values shown above. I also tried using time() which has similar results. I thought I would get the current date and time with these calls. Would someone please let me know what I'm doing wrong. Thank you. ================================= Kevin Kernin Moore Research Center, Inc. (716)773-0658 FAX : (716)773-0462 email: kernin@research.moore.com ================================= From James_M_Flynn@res.raytheon.com Wed Aug 12 16:42:04 1998 From: James_M_Flynn@res.raytheon.com Date: Wed Aug 12 16:42:07 PDT 1998 Subject: Symbol Table Synchronization vxWorks & Tornado users, We've been having a problem synchronizing the target and host and are hoping someone can help. When we #define INCLUDE_SYM_TBL_SYNC in configAll.h the file symTbl.c is generated with some lines that have syntax errors. The lines in error are related to symbols in target/h/wtx.h that start with WTX_EVT... etc. Can anyone shed some light on this? It must be something simple because all we're doing is turning on this #define and it breaks our compile. We are using vxWorks 5.3.1 on the MV2604 PPC (BSP 1.1/4) Thanks Jim From daemon@csg.lbl.gov Thu Aug 13 04:00:12 1998 From: daemon@csg.lbl.gov Date: Thu Aug 13 04:00:16 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 13 04:00:09 PDT 1998 Subject: Java on VxWorks Subject: Re: Scripts aren't working... Subject: Re: VxWorks Disks: ATA vs IDE Subject: Re: Timeout when ftp-booting. ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Java on VxWorks Date: Wed, 12 Aug 1998 14:58:11 +0200 From: Eberhard Amann Organization: IBM-EL Message-ID: <35D19163.22A59B4C@de.ibm.com> Reply-To: esa@de.ibm.com we are looking for JAVA on VxWorks & a implementation of the JAVA AWT Class. Does anyone know if this is available ?, What about Power PC ? - -- Greetings from : Eberhard Amann IBM Germany (EL) ---- Dep.: 3076 Phone: (0049) 7031-16-4292 ---- Email: esa@de.ibm.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Scripts aren't working... Date: Tue, 11 Aug 1998 13:07:08 GMT From: B.Metz@or-computers.de (Bertram Metz) Organization: or Industrial Computers GmbH Message-ID: <35d03c67.23357205@news.profi.net> References: <35C2CFFB.D6F932BE@ri.cmu.edu> On Sat, 01 Aug 1998 04:21:15 -0400, Jorgen Pedersen wrote: >Target: VxWorks 5.3 on x86 >Host: WinNT > >I booted my standalone VxWorks kernal (/fd0/vxWorks.st) from floppy. >The kernal loaded fine. However, the startup script was not executed. >So I then tried to cd to /fd0/ only to find that it was no longer >attached. I overcame this by typing the following at the shell prompt: > >-> usrFdConfig 0,0,"/fd0/" > >After reattaching the floppy device, I was able to cd to /fd0/. I find >it strange that while loading the kernal, the floppy device was >attached, but after the kernal was loaded, the floppy device was >detached. Maybe this is why my startup script is not being executed. The boot loader automatically attaches the floppy device, if the vxWorks image (vxWorks.st) has to be loaded from it. But the the loaded image does this only if you have properly configured it. What does this mean? You have to attach the floppy device before the startup script is called. A proper place could usrRoot() in usrConfig.c. Add "usrFdConfig(0, 0, "/fd0/");" after the call to fdDrv(). I've added an extra define "INCLUDE_FD_INIT for this purpose: #ifdef INCLUDE_FD_INIT usrFdConfig(0, 0, "/fd0/"); #endif /* INCLUDE_FD_INIT */ >So I tried to manually run the script (which is located on floppy). The >name of my script is called "loadx86" and the contents of the script are >merely: > >ld < foo >moduleShow > >When I tried to run the loadx86 script manually, I got the following >error: > >-> 'an't open input 'foo > errno = 0x380003 >-> > >So the next logical step was to see if I could even load the object code >"foo" manually. I typed "ld < foo" at the shell prompt and it loaded >just fine. I verified that the function "moduleShow" also worked at the >shell prompt. So it is only when I try to execute commands from within >a script that causes an errror. I made the experience that the startup script was only executed if "STANDALONE_NET" was defined in config.h (sounds a little bit strange doesn't it?). >So there are two problems here which I am trying to solve: > >1) Why isn't my startup script being invoked? >2) Why can't I run a script at the shell prompt? > >I ensured that INCLUDE_STARTUP_SCRIPT was enabled when compiling >vxWorks.st. I'm sure I'm missing something, but for the life of me, I >cannot see what. So if anyone can tell me how to solve one or more of >the above problems, I would appreciate it!! > >Thanks > Jorgen > Bertram Bertram Metz b.metz@or-computers.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Disks: ATA vs IDE Date: Wed, 12 Aug 1998 17:46:20 -0500 From: "Dennis C." Organization: Cisco Systems, Inc. Message-ID: <6qtcdp$6kd$1@news-sj-3.cisco.com> References: <01BDC5C3.B0705B90@T200-ARLINGTON.BBN.COM> The ATA is the successor of the IDE driver. If you are just looking for a simple driver to port, go with the IDE driver. Also their driver does not support DMA. Dan Sunnarborg wrote in message <01BDC5C3.B0705B90@T200-ARLINGTON.BBN.COM>... >Hello, > >I am wondering what the differences are between the IDE and ATA drivers? >When should each of them be used? What are their requirements? I am >using 5.3.1 of VxWorks. > >I am using a standard Seagate 2 GB IDE drive on a 486 target, should I be >using the IDE or ATA driver calls? > >Thanks! > >Dan Sunnarborg >Software Engineer >Applied Biometrics, Inc. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Timeout when ftp-booting. Date: Thu, 13 Aug 1998 06:13:28 GMT From: "Jim Simpson" Organization: Netcom On-Line Services Message-ID: References: <35CDD202.43012B6F@arboga.mail.telia.com> Sender: jaimes@netcom13.netcom.com As I recall from a conversation with Wind River about 2 years ago, the ftp download timeout is buried deep inside the TCP/IP implementation. However, I saw that the timeout was longer in 5.3 than in 5.2. I had a problem in which 4 cards running vxWorks were trying to download at the same time. Usually one or two downloads timed out. Although this is not pretty, the solution I used was to change bootInit.c so that something on the card (like a serial number or part of the MAC) is used to do a taskDelay(). Each card had a different taskDelay() time and the result was to stagger the download. My co-worker recommends that I just PROM or FLASH the code. Jim >Hello ! > >Does anyone know how to increase the timeout when ftp-booting the >vxWorks-kernel. > >/ce > --------------------------- End of New-News digest ********************** From ksamavedam@hns.com Thu Aug 13 06:37:05 1998 From: ksamavedam@hns.com (Krishna Samavedam) Date: Thu Aug 13 06:37:12 PDT 1998 Subject: RE: Symbol Table Synchronization HI , After that U have to invoke the tgtsvr with "-s" option. Set the timeout appropriately if the NW delays are significant otherwise default value is o.k. hope this helps.. krishna >From James_M_Flynn@res.raytheon.com Wed Aug 12 16:42:04 1998 From: James_M_Flynn@res.raytheon.com > Date: Wed Aug 12 16:42:07 PDT 1998 > Subject: Symbol Table Synchronization > > vxWorks & Tornado users, > > We've been having a problem synchronizing the target and host and are > hoping someone can help. When we #define INCLUDE_SYM_TBL_SYNC in > configAll.h the > file symTbl.c is generated with some lines that have syntax errors. The > lines in error are related to symbols in target/h/wtx.h that start with > WTX_EVT... etc. Can > anyone shed some light on this? It must be something simple because all > we're doing is turning on this #define and it breaks our compile. We are > using vxWorks 5.3.1 > on the MV2604 PPC (BSP 1.1/4) > > Thanks > > Jim From daemon@csg.lbl.gov Fri Aug 14 04:00:24 1998 From: daemon@csg.lbl.gov Date: Fri Aug 14 04:00:29 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Aug 14 04:00:21 PDT 1998 Subject: asm contraints for '860 Subject: New to vxworks-opinions wanted Subject: FileMaker Pro training Subject: Re: [Q} routeAdd Subject: Re: VxWorks TCP/IP in Promiscuous mode Subject: Re: how to do live software upgrades? Subject: Re: IEEE 1284 ECP Driver for VxWorks on PC platforms Subject: Re: how to do live software upgrades? Subject: Re: VxWorks Disks: ATA vs IDE Subject: Re: BSP Porting kit experience? Subject: PTY's and stdio Subject: VxWorks 5.2 on Linux Subject: Re: MVME2700_Do_I_Need_A_VxWorks_Prom? Subject: Re: reading real-time clock Subject: Re: Scripts aren't working... Subject: Re: VME block transfers Subject: Re: Java on VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: asm contraints for '860 Date: Thu, 13 Aug 1998 07:22:45 -0400 From: "James A. Littlefield" Organization: The Internet Access Company, Inc. Message-ID: <35D2CC85.BFB2D97F@alum.mit.edu> I need to do some inline assembly for the PPC. To make a home brew version of the routines to read/write the decementer, I have tried things like.... __asm__("mfspr 0,22"); /* put dec -> r0 */ Which compiles correctly. The line below does not compile/assembler?? __asm__("mfspr %0,22" : "=g"); Can anyone refer me to some PPC specific info on contraint specifications> Thanks Jim --------------------------- Newsgroups: comp.os.vxworks Subject: New to vxworks-opinions wanted Date: Wed, 12 Aug 1998 17:04:08 +0100 From: "Graham Perkins" Message-ID: <35d1bcfa.0@nnrp1.news.uk.psi.net> Hi everybody, Were are currently evaluating several realtime operating systems at work, vxworks being one of them. I wondered if anybody could pass on their opinions of vxworks to me. If you are using SENS or the Spider TCP/IP stack I would like your views of these products. Please feel free to use my private email address mailto:grahamp@specialix.co.uk if you prefer. Many thanks, Graham. --------------------------- Newsgroups: comp.os.vxworks Subject: FileMaker Pro training Date: 13 Aug 1998 00:14:05 GMT From: info@trainingcompany.com Organization: Macresource Computer Training Message-ID: <6qtb4d$ia0$1@nnrp0.seg0> Reply-To: info@trainingcompany.com Learn FileMaker Pro in your browser. New on-line course with numerous screen-grabs and QuickTime movies. Just $9.99. http://www.trainingcompany.com/filemaker/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [Q} routeAdd Date: Thu, 13 Aug 1998 09:32:52 -0400 From: Rich Boyd Organization: Innovative Concepts, Inc Message-ID: <35D2EB03.D9BE4AC6@innocon.com> References: <35D1523A.1657F6AD@leonis.nus.edu.sg> The "gateway_ip" must be reachable (via the existing routing table) or else you can't add it to the routing table. Use mRouteShow to view the table, and be sure to check the subnet masks, to verify that the "gateway_ip" is on an attached subnet. Tan Seng Kee wrote: > Hi, > > I am currently using SENS 1.0. After setting up the network interfaces, > I have problem in setting the default route for the network. > > From the docs/manuals, they illustrated that adding a default route can > be performed by using routeAdd(). However, for my case, both in my > coding as well as the Windsh, calling of > routeAdd("0.0.0.0","gateway_ip") return ERROR and the routing table is > not updated. > > I have tried all the four combination of using hostname as well as the > ip address for the routeAdd() and routeNetAdd() call. All FAILed. > However, if I am adding route for other ip host or network ip, the call > success. Does any one has this experience before ? Please advice on how > to set up the default router. > > Your help deeply appreciated. > > Urgently needed to solve the problem. Thank you. > > Seng Kee - -- Richard L Boyd Innovative Concepts, Inc rlboyd@innocon.com 8200 Greensboro Dr, Suite 801 Phone: 703-893-2007 x277 McLean, VA 22102 Fax: 703-893-5890 --------------------------- Newsgroups: pickerel@netcom.com,comp.os.vxworks Subject: Re: VxWorks TCP/IP in Promiscuous mode Date: Thu, 13 Aug 1998 09:26:14 -0400 From: Rich Boyd Organization: Innovative Concepts, Inc Message-ID: <35D2E976.BBCFF2FA@innocon.com> References: You could declare your local interface as the default gateway. For example: routeAdd "0", "10.11.12.13" assuming the VxWorks target's IP address is "10.11.12.13". You can also add actual subnets instead of the default route if desired. The key is that the local interface can be considered a gateway. Don Pickerel wrote: > Hi, > Does anyone know if the VxWorks stack can be set up to work in > promiscuous mode? > > Right now, we're capturing packets with etherOutputHookAdd, and > re-routing if the packet isn't sent out on the local ethernet. > > However, if the packet is being sent out to a different subnet, it gets > dropped before it gets to my outputHook funtion. I have to add a gateway > and fake arp entry to get this to work. > > I'd like the stack to transmit the packet regardless of the subnet it's > on. I know in some stacks there used to be a promiscuous mode that would > let you do this, relying on the router to listen to all packets and > forward the out of network packets. > > Thanks for any ideas. > > -Don Pickerel- - -- Richard L Boyd Innovative Concepts, Inc rlboyd@innocon.com 8200 Greensboro Dr, Suite 801 Phone: 703-893-2007 x277 McLean, VA 22102 Fax: 703-893-5890 --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: how to do live software upgrades? Date: Thu, 13 Aug 1998 05:22:02 -0700 From: Todd Hoff Organization: Possibility Outpost Message-ID: <35D2DA6A.C02169D3@possibility.com> References: <35CC6F14.6B715069@possibility.com> <903002602snz@tcontec.demon.co.uk> Reply-To: tmh@possibility.com Paul E. Bennett wrote: >The problem becomes one of knowing the new module will > work correctly in the running environment even when tested out of the > running environment. I do nt envy you in this task. I do not think you > could do this on a whole system while live. Are you sure you could not > steal a few seconds out of the operation? If it's too hard to do reliably then, well, we'll do something else :-) It seems if we do a warm boot that does not stop the hardware from functioning then we can do a reload of the whole image. During that period we can drop alarms and can't accept new work requests, which now seems acceptable. It's suprising for embedded systems that often work in high availability environments that we get so little support for upgrade and fail-over scenarios. Every project just rehashes the same issues and probably comes up with the same lame solutions. - ------------------------------------------------------------------ tmh@possibility.com http://www.possibility.com/Tmh Desperate measures require the invention of desperate times. -- Todd Hoff --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IEEE 1284 ECP Driver for VxWorks on PC platforms Date: Thu, 13 Aug 1998 14:32:25 GMT From: act@rtp.co.uk Organization: Deja News - The Leader in Internet Discussion Message-ID: <6qutdp$q91$1@nnrp1.dejanews.com> References: <6qe8ua$7tm$1@news01.btx.dtag.de> <6qfogs$5n4$1@news02.btx.dtag.de> Hello, I believe the standard Wind River driver does not support the ECP or EPP functionality of IEEE 1284. This needs a specific driver writing, which RTP is in the process of producing. This will be made available as source soon. As a company, Real Time Products specialises in real-time embedded systems. We offer a range of consulting, software development and training services. For a fuller explanation, and a list of drivers we have available off-the-shelf, you may want to visit our website at www.rtp.co.uk, or e-mail me at act@rtp.co.uk. Best Regards Alex Taylor - ------------------------------------------------------------------------------------------------------------------ Alex Taylor e-mail: mailto:act@rtp.co.uk Real Time Products Ltd. Direct: +44 (0) 121 234 6623 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 - ------------------------------------------------------------------------------------------------------------------- - -------------------------------------------------------------------------------------------------------------- In article <6qfogs$5n4$1@news02.btx.dtag.de>, LP.Office@t-online.de (Heinrich Munz) wrote: > On 7 Aug 1998 07:04:42 GMT, LP.Office@t-online.de (Heinrich Munz) > wrote: > > >Hi all, > > > >I'm looking for a IEEE 1284 ECP Driver for VxWorks > >(using DMA and Interrupts). > >Anyone out there, having any experience with this? > > > >Thanks in advance > > > >-- > >Heinrich Munz, President, HMunz@LP-Elektronik.com > >Phone +49 751 56122-0, Fax +49 751 56122-22, > >http://www.lp-elektronik.com > > > >LP Elektronik GmbH, Germany, Ettishofer Str. 8, 88250 Weingarten > > > > > > Update: > I'm also interested in a VxWorks EPP Mode driver ! > (not using DMA) > > - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: how to do live software upgrades? Date: Thu, 13 Aug 1998 08:54:19 +0100 From: Stephen Maudsley Organization: esgem limited Message-ID: <35D29BAB.AF6D767B@esgem.com> References: <35CC6F14.6B715069@possibility.com> Reply-To: Stephen.Maudsley@esgem.com Todd Hoff wrote: > We have a requirement to upgrade software hosted on a > vxworks system while still maintaining service. These > particular cards don't have a backup nor can their services > be transfered to another system while the upgrade is in > progress. > > It's not clear to me how a target, with c/c++ object > code, can update its software while providing uninterrupted > service. The code has to be downloaded into memory while the > current system is running and then start. > Does anyone have techniques for this? What do you mean by "uninterrupted" in this machine? What's the throughput and how much time do you have worst case before something else notices that you're not there? - -- Stephen Maudsley mailto:Stephen.Maudsley@esgem.com Esgem Limited for electronic product design http://www.esgem.com Tel: +44-1453-521626 Mobile: +44-370-810991 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Disks: ATA vs IDE Date: Thu, 13 Aug 1998 11:34:03 -0500 From: "Dennis C." Organization: Cisco Systems, Inc. Message-ID: <6qvavk$hr1$1@news-sj-3.cisco.com> References: <01BDC5C3.B0705B90@T200-ARLINGTON.BBN.COM> <6qtcdp$6kd$1@news-sj-3.cisco.com> <35D2DE33.6BA9@ti.com> 30% faster? I am surprised.. Anyway, if you guys need to add DMA, email me. Frank Minich wrote in message <35D2DE33.6BA9@ti.com>... >Dennis C. wrote: >> >> The ATA is the successor of the IDE driver. If you are just looking for a >> simple driver to port, go with the IDE driver. >> Also their driver does not support DMA. >> >I've used both and the ATA driver is faster (~30% in my test case). I'd >recommend using the ATA driver if your hardware supports it. > >> Dan Sunnarborg wrote in message >> <01BDC5C3.B0705B90@T200-ARLINGTON.BBN.COM>... >> >Hello, >> > >> >I am wondering what the differences are between the IDE and ATA drivers? >> >When should each of them be used? What are their requirements? I am >> >using 5.3.1 of VxWorks. >> > >> >I am using a standard Seagate 2 GB IDE drive on a 486 target, should I be >> >using the IDE or ATA driver calls? >> > >> >Thanks! >> > >> >Dan Sunnarborg >> >Software Engineer >> >Applied Biometrics, Inc. >> > >> > >Regards, >Frank Minich >Texas Instruments --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP Porting kit experience? Date: Thu, 13 Aug 1998 18:54:59 GMT From: Brian Silver Organization: Castle Networks Message-ID: <35D33642.70D6@castlenetworks.com> References: <35d20dc6.3588621@news.jsc.nasa.gov> John R. Cobarruvias wrote: > > Anyone have any experience using the BSP porting kit? > ANTI-SPAM EMAIL IN EFFECT. SEND EMAIL TO: > > johncoby@orbitworld.net Almost useless. We bought it and found that the drivers on it are pretty useless as WRS updates the drivers but doesn't update the Driver's CD. I've been told by people-to-remain-nameless that the validation suite is crap, although I've never used it myself. The "word on the street" is if you can get the debugger up, you've done just about everything the validation suite would do. We found that the best bet is to get a working BSP for a similar platform and work from there. Brian. --------------------------- Newsgroups: comp.os.vxworks Subject: PTY's and stdio Date: Thu, 13 Aug 1998 19:00:27 GMT From: Brian Silver Organization: Castle Networks Message-ID: <35D3378A.698F@castlenetworks.com> Can anyone tell me how to get the FILE* equivilent of a PTY fd? fopen() returns NULL (with errno == 0 ), and I'd really like to use stdio with a PTY. The manual says it should work, as there is no difference beween PTY's and TTY's in this respect. But it doesn't. Anyone have any other experiences? Thanks. Brian. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks 5.2 on Linux Date: Thu, 13 Aug 1998 15:25:29 -0400 From: padovani@aaec.com (Pierre Padovani) Organization: Atlantic Aerospace Electronics Corp. Message-ID: Folks, Does anyone out there know of a port of VxWorks 5.2 to Linux? I am currently in need of such a thing. Thanks, Will S. - -- - -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Will Sanders colltech@aaec.com Computer Systems Administrator (617)890-4200 xt3255 Atlantic Aerospace Electronics Corporation. - -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MVME2700_Do_I_Need_A_VxWorks_Prom? Date: Thu, 13 Aug 1998 12:50:46 GMT From: Bill Dennen Organization: Motorola Computer Group Message-ID: <35D2E126.2CF90369@mcg.mot.com> References: Sender: usenet@cradac.mclean.mcd.mot.com (UseNet) With all the PowerPlus boards it is not necessary to replace the PPCBug PROMs. The sequence, in general, is to TFTP the image into DRAM and then FLASH that image into the soldered FLASH on board. You can then either boot the board to Bug or VxWorks by setting the J9 jumper. I don't have the example procedures for doing this handy, but your local MCG SE should be able to get them for you. >@< Roger Holmquist wrote: > > We have just got a VMVE2700-card and I am now trying to run it in our Tornado > (UNIX version) environment. > As far as I remember (berfore a very loong holiday!) you dont need to > replace the bug-prom in order to boot the card. Is that correct and if so, > is it desirable? (There is some debugging and HW-test tools in the Motorola > PPCBUG who might be usable....) > > If there is anybody out there who have NIOT-parameters (if relevant) as an > example > for the card, I would be happy to receive them! > > Greetings / Roger > > # -----------------------------------------------------------------# > # Roger Holmquist, FOA, National Defence Research Establishment # > # sect 750, Olaus Magnus vag 42, Box 1165, 581 11 LINKOPING ------# > # --------------------------- SWEDEN ------------------------------# > # Phone: +46 13-378382 --------------------------------------------# > # Fax: +46 13-378009 ----------------------------------------------# > # Home: +46-13-214733 Mobile +46-706-250123 ----------------------# > # Email: roghol@lin.foa.se ----------------------------------------# > # -----------------------------------------------------------------# > xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx > | This message is MIME-coded in order to avoid | > | problems with national characters. MIME coding | > | should be enabled when replying in Swedish. | > xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx - -- >@< Bill Dennen bowtie@mclean.mcd.mot.com | We has met the enemy MidAtlantic District SE 703-714-0707 | and they is us Motorola Computer Group 703-714-0714 FAX | Pogo --------------------------- Newsgroups: comp.os.vxworks Subject: Re: reading real-time clock Date: 13 Aug 1998 08:37:12 -0400 From: Michael Carr Organization: UltraNet Communications , an RCN Company http://www.ultranet.com/ Message-ID: References: <199808122234.SAA01098@leda.moore.com> You need to read the time from RTC during startup and use it to program the system time using clock_settime. Here's an example I wrote to do this on the /* put this in your start-up code (e.g., usrConfig.c, or USER_APPL_INIT) */ { timespec now; now.tv_sec = sysCalendarTime(); now.tv_nsec = 0; status = clock_settime(CLOCK_REALTIME, &now); assert(status == OK); } /* put this in sysLib.c */ time_t sysCalendarTime(void) { struct tm timeAndDate; unsigned char timeData[8]; int i; for (i = 0; i < 8; i++) timeData[i] = sysNvRead((ULONG)TOD_CLOCK + i); timeData[0] = 0x00; timeAndDate.tm_sec = (timeData[1] >> 4) * 10 + (timeData[1] & 0x0F); timeAndDate.tm_min = (timeData[2] >> 4) * 10 + (timeData[2] & 0x0F); timeAndDate.tm_hour = (timeData[3] >> 4) * 10 + (timeData[3] & 0x0F); timeAndDate.tm_mday = (timeData[5] >> 4) * 10 + (timeData[5] & 0x0F); timeAndDate.tm_mon = (timeData[6] >> 4) * 10 + (timeData[6] & 0x0F) - 1; timeAndDate.tm_year = (timeData[7] >> 4) * 10 + (timeData[7] & 0x0F); /* don't start introducing Y2K bugs now ... */ if (timeAndDate.tm_year < 70) timeAndDate.tm_year += 100; timeAndDate.tm_isdst = -1; timeData[0] = 0x00; return mktime(&timeAndDate); } kernin@research.moore.com (Kevin Kernin) writes: > I need to get the current date and time in our application running vxworks. > > We are using the MVME 2604 board with the 1.1/2 version of the BSP and > the MVME 2306 board with the 1.1/4 version of the BSP. > > Using the PPCbug command "time" I verified that we set the real-time clock > properly. > I then switched the jumper and booted up with vxWorks. > I made the following calls to get the date and time: > > clock_gettime(CLOCK_REALTIME,mytimespec) > gmtime( (time_t *) &mytimespec) > > when I check the tm struct that the call to gmtime returns I found > unexpected values as shown below: > > time in seconds= 0 > seconds= 0 > minutes= 0 > hours= 0 > day of month= 1 > month= 0 > year= 70 > dayofweek= 4 > > The time does increase when I rerun my code, but if I reset the PPC > board and run I always start with the values shown above. > > I also tried using time() which has similar results. > I thought I would get the current date and time with these calls. > > Would someone please let me know what I'm doing wrong. > > Thank you. > > ================================= > Kevin Kernin > Moore Research Center, Inc. > (716)773-0658 > FAX : (716)773-0462 > email: kernin@research.moore.com > ================================= - -- - ----------------------------------------------------------- Michael Carr Redstone Communications *** Remove 'NOSPAM' from my email address when replying *** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Scripts aren't working... Date: Thu, 13 Aug 1998 07:35:40 -0500 From: Frank Minich Organization: NCP Message-ID: <35D2DD9C.7E92@ti.com> References: <35C2CFFB.D6F932BE@ri.cmu.edu> <35d03c67.23357205@news.profi.net> Reply-To: fminich@ti.com Bertram Metz wrote: > > On Sat, 01 Aug 1998 04:21:15 -0400, Jorgen Pedersen > wrote: > > >Target: VxWorks 5.3 on x86 > >Host: WinNT > > > >I booted my standalone VxWorks kernal (/fd0/vxWorks.st) from floppy. > >The kernal loaded fine. However, the startup script was not executed. > >So I then tried to cd to /fd0/ only to find that it was no longer > >attached. I overcame this by typing the following at the shell prompt: > > > >-> usrFdConfig 0,0,"/fd0/" > > > >After reattaching the floppy device, I was able to cd to /fd0/. I find > >it strange that while loading the kernal, the floppy device was > >attached, but after the kernal was loaded, the floppy device was > >detached. Maybe this is why my startup script is not being executed. > > The boot loader automatically attaches the floppy device, if the > vxWorks image (vxWorks.st) has to be loaded from it. But the the > loaded image does this only if you have properly configured it. > > What does this mean? > You have to attach the floppy device before the startup script is > called. A proper place could usrRoot() in usrConfig.c. Add > "usrFdConfig(0, 0, "/fd0/");" after the call to fdDrv(). > I've added an extra define "INCLUDE_FD_INIT for this purpose: > > #ifdef INCLUDE_FD_INIT > usrFdConfig(0, 0, "/fd0/"); > #endif /* INCLUDE_FD_INIT */ > > >So I tried to manually run the script (which is located on floppy). The > >name of my script is called "loadx86" and the contents of the script are > >merely: > > > >ld < foo > >moduleShow > > > >When I tried to run the loadx86 script manually, I got the following > >error: > > > >-> >'an't open input 'foo > > errno = 0x380003 I've seen this error when I edited the script using an MSDOS editor, instead of a unix editor. An MSDOS editor places 0x0d 0x0a end of line terminators in the file. A unix editor places 0x0a end of line terminators in the file. The script reader doesn't seem to understand 0x0d 0x0a line terminators. > >-> > > > >So the next logical step was to see if I could even load the object code > >"foo" manually. I typed "ld < foo" at the shell prompt and it loaded > >just fine. I verified that the function "moduleShow" also worked at the > >shell prompt. So it is only when I try to execute commands from within > >a script that causes an errror. > > I made the experience that the startup script was only executed if > "STANDALONE_NET" was defined in config.h (sounds a little bit strange > doesn't it?). > > >So there are two problems here which I am trying to solve: > > > >1) Why isn't my startup script being invoked? > >2) Why can't I run a script at the shell prompt? > > > >I ensured that INCLUDE_STARTUP_SCRIPT was enabled when compiling > >vxWorks.st. I'm sure I'm missing something, but for the life of me, I > >cannot see what. So if anyone can tell me how to solve one or more of > >the above problems, I would appreciate it!! > > > >Thanks > > Jorgen > > > > Bertram > > Bertram Metz > b.metz@or-computers.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VME block transfers Date: 13 Aug 1998 23:01:26 PDT From: John Murphy Organization: Concentric Internet Services Message-ID: <35D3D32D.89364B98@concentric.net> References: <35CB782C.80109C97@physics.ucla.edu> Ian Ferguson wrote: > > Hi all, > I'm trying to write a simple driver for a Motorola MVME-167 cpu that > executes a block transfer along the VME bus (D32BLT). Could anyone > point me in the direction of some simple programs that executes this? > I've already done programs that do simple reads/writes, perhaps it is a > simple extension of that. Any help would be appreciated. Thank you. > > Ian Ferguson > fergie@physics.ucla.edu I'm starting to work on a 16 bit block transfer. I'm new to vxWorks and VME Bus both, so it's been a learning experience. Could you pass along any pointers? I'd offer help, but I'm so new I don't have much to offer. thanks, John Murphy jmurphy@ball.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Java on VxWorks Date: Wed, 12 Aug 1998 11:54:52 -0500 From: Doug Thomae Organization: Compuware Corporation Message-ID: <35D1C8DC.14653A5D@compuware.com> References: <35D19163.22A59B4C@de.ibm.com> This is a multi-part message in MIME format. - --------------888F1A2823FA30AD34BADE0E Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit WindRiver sells a beta Java that runs under VxWorks, including the java.awt package. Its a run time environment only, so you'll have to supply your own compiler (we use the JDK). Eberhard Amann wrote: > we are looking for JAVA on VxWorks & a implementation of the JAVA AWT Class. > > Does anyone know if this is available ?, What about Power PC ? > > -- > Greetings from : Eberhard Amann > IBM Germany (EL) ---- Dep.: 3076 > Phone: (0049) 7031-16-4292 ---- Email: esa@de.ibm.com - -- Doug Thomae doug_thomae@compuware.com 414-319-1582 Any opinions or statements expressed are my own, and may not reflect those of Compuware Corp., the federal government or my girlfriend's cat. - --------------888F1A2823FA30AD34BADE0E Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Doug Thomae Content-Disposition: attachment; filename="vcard.vcf" begin: vcard fn: Doug Thomae n: Thomae;Doug org: Compuware Corp. adr;dom: 700 W. Virginia St. Suite 500;;;Milwaukee;WI;53204; email;internet: doug_thomae@compuware.com title: Technical Specialist tel;work: (414) 319-1582 tel;fax: (414) 310-1690 tel;home: (414) 569-7809 note: Any views expressed are my own, and may not reflect those of Compuware Corp. x-mozilla-cpt: ;0 x-mozilla-html: FALSE version: 2.1 end: vcard - --------------888F1A2823FA30AD34BADE0E-- --------------------------- End of New-News digest ********************** From ChillaJA@crbesmtp.dt.navy.mil Fri Aug 14 08:45:05 1998 From: ChillaJA@crbesmtp.dt.navy.mil Date: Fri Aug 14 08:45:09 PDT 1998 Subject: RE: Motorola IP-488 drivers for VxWorks John, I apologize for the delay in response. We have used the National Instruments GPIB-ENET for GPIB control (~$1200 each). This controller uses TCP/IP on Ethernet (AUI, 10Base2, 10BaseT options available). I wanted to use this controller with a DEC AlphaStation running OpneVMS which is not supported by National Instruments. They did sell us 'C' source code for around $1500 that I was able to port to OpenVMS and Linux with very little effort. Ask fot the Engineering Software Package for the GPIB-ENET (Part # ESP-488). This part was not listed in their catalog. I am sure the ESP-488 would easily port to VxWorks as well. You will have to add your own task synchronization if you intend to allow multiple tasks to access the GPIB-ENET simultaneously (i.e use mutex semaphore). This controller might be little slower than a VME-GPIB controller, but it is much more versatile. The unit has been very reliable in our applications. The Windows NT/95 driver for the GPIB-ENET is available for free from the National Instruments web site (www.natinst.com) which has an interactive test control program and controller configuration utility. Hope this helps... - Jeff Chilla ------------------------------------------ Jeff Chilla, CD/NSWC Code 7310 ChillaJA@crbesmtp.dt.navy.mil > Submitted-by jshaw_shade@compuserve.com Wed Jul 29 09:22:03 1998 > Submitted-by: John Shaw > > Hello VxWorkers, > > This forum has seen some enquiries in the past about possible VxWorks > drivers for IEEE-488 hardware. But there have been few responses! We > are > about to use a Motorola IEEE-488 IP module for use with a MVME-162. > This > might either use the TMS9914A chip or the NAT9914A chip. Are there any > suitable drivers for the chip or for the IP-488, for VxWorks or UNIX, > in = > C? > > Are there any recommended web sites for drivers or EVEN DATASHEETS? > Strange that Motorola should have produced a board with no software! > How > did they test their kit? The board ident is the MVIP300, using the TMS > chip. > > Any advice or useful information will be greatly appreciated. > > John Shaw, > Shade (Computer Services) Limited, > Calne, Wilthshire, England, UK. > > Email: john.shaw@shade.co.uk > > From dmm@wrs.com Fri Aug 14 09:14:21 1998 From: Deepak Mulchandani Date: Fri Aug 14 09:14:25 PDT 1998 Subject: Tornado for Java Hi, Just to clarify some information on Wind River' Java product: - Tornado for Java has been a FCS shipping product since Feb '98. It was in Beta late last year, but is now available for purchase as a fully supported product. It is available for the PowerPC and other architectures today. - The port is a fully certified port i.e., it passes 100% of the compatibility tests developed by JavaSoft. This includes all the AWT tests. - The JVM is based on JDK1.1.4 and the product is a runtime component to the VxWorks OS. Customers can write their Java applications using any off the shelf IDE (Symantec, Borland) but they can also use the free JDK available from SUN. You can contact your local sales office for more information. Regards, Deepak >Date: Wed, 12 Aug 1998 11:54:52 -0500 >From: Doug Thomae >Organization: Compuware Corporation > >WindRiver sells a beta Java that runs under VxWorks, including >the java.awt package. Its a run time environment only, so you'll >have to supply your own compiler (we use the JDK). > >>Eberhard Amann wrote: >> >> we are looking for JAVA on VxWorks & a implementation of the JAVA AWT Class. >> >> Does anyone know if this is available ?, What about Power PC ? >> >>-- >> Greetings from : Eberhard Amann >> IBM Germany (EL) ---- Dep.: 3076 >> Phone: (0049) 7031-16-4292 ---- Email: esa@de.ibm.com >> -- Deepak Mulchandani dmm@wrs.com Wind River Systems 1010 Atlantic Avenue, Alameda, CA 94501 Phone: (510) 749-2448 Fax: (510) 749-2010 From mcuddy@fensende.com Fri Aug 14 17:44:14 1998 From: Mike Cuddy Date: Fri Aug 14 17:44:17 PDT 1998 Subject: Symbios Logic / NCR SYM53C875 Keywords: SCSI, NCR, Symbios Logic, 53C8XX I'm having some trouble getting the WRS supplied 53C8xx driver running for a SYM53C875 chip (on a Symbios Logic eval board). My target is a DEC EBSA-285, StronARM/SA-110 with the SA21285 bridge chip in host mode. The SYM card is on the PCI bus. I also have a PCI ethernet card (AMD79C971). I use the following code to init the SCSI controller: --SNIP -------------------------------------------------------------- ... code above here figures out where on PCI bus the card is really living, sets up "iobase", "ilvl", and "ivec", I'm using the I/O base address register, not the memory mapped base address register ... printf("scsi: iobase = 0x%x, ilvl=%x, ivec=%x\n", iobase, ilvl, ivec); // // create control structure with driver // sdvScsi = ncr810CtrlCreate((UINT8*) iobase, NCR810_40MHZ, NCR875_DEVICE_ID if (sdvScsi == NULL) { printf("scsi: Can't create NCR control structure.\n"); return ERROR; } // connect interrupt ... if (intConnect(INUM_TO_IVEC(ivec), ncr810Intr, (int) sdvScsi) == ERROR) { printf("scsi: Can't connect interrupt.\n"); return ERROR; } intEnable(ilvl); // // now init SCSI control structure, set controller ID to 7. // if (ncr810CtrlInit(sdvScsi, NCR_SCSI_CTRL_ID) == ERROR) { printf("scsi: can't init device.\n"); return ERROR; } ncr810SetHwRegister( sdvScsi, &sdvNcr810HwRegs); scsiCacheSnoopDisable((SCSI_CTRL*) sdvScsi); return OK; --SNIP -------------------------------------------------------------- I've added some debugging to the 810 library, and this is what I get on my target's console when I call : scsiPhysDevCreate(sdvScsi, 0, 0, 0, -1, 0, 0, 0) I have set scsiIntsDebug = scsiDebug = scsiErrors = 1; before the above call. ------- TARGET OUTPUT -------- 0xf33d30 (t4): scsiReqSense: 0xf33d30 (t4): scsi2Transact: 0xf34e68 (tScsiTask): scsiMgrRequestProc: client request: 0 0xf34e68 (tScsiTask): scsiMgrActivateRequest: thread 0x00ffc378 0xf34e68 (tScsiTask): ncr810ThreadActivate: thread 0x00ffc378: activating 0xf34e68 (tScsiTask): ncr810ThreadActivate: WideXFerNegotiate ... 0xf34e68 (tScsiTask): ncr810ThreadActivate: SyncXferNegotiate ... 0xf34e68 (tScsiTask): ncr810XferParamsCvt: converted to: offset = 0, period = 0 xferParams = 0, clockDivide = 3 0xf34e68 (tScsiTask): ncr810Activate: for thread ffc378 (siop state = 1) 0xf34e68 (tScsiTask): ncr810ThreadStateSet: thread 0x00ffc378: 1 -> 6 0xf34e68 (tScsiTask): scsiMgrTimeoutProc: thread 0x00ffc378 (state = 6) timed ou t 0xf34e68 (tScsiTask): ncr810ThreadAbort: thread 0x00ffc378 (state = 6) aborting 0xf34e68 (tScsiTask): ncr810Abort: aborting: state (1) 0xf34e68 (tScsiTask): ncr810ThreadStateSet: thread 0x00ffc378: 6 -> 9 -------- END TARGET OUTPUT -------- The ethernet card busmasters it's data in and out of host memory, so I'm pretty sure that the PCI bridge is setup correctly. It looks like the SCSI isn't delivering interrupt to the target. The SCSI driver code that I'm using comes directly from the BSP developer's kit I'm wondering if it's some kind of incompatibility between the 810 chip (for which the driver was written) and the 875 (The chip I'm using). although, the header files mention the 875 explicitly as a supported chip. (P.S., is it just me, or is the documentation for how SCSI-2 drivers work in Vxworks crap?) -- Mike Cuddy (mcuddy@FensEnde.com, MC312), Programmer, Daddy, Human. Fen's Ende Software, Redwood City, CA, USA, Earth, Sol System, Milky Way. I remember asking why ... Let it rain, and protect us from this Cruel Sun. Join CAUCE: The Coalition Against Unsolicited Commercial E-mail. From daemon@csg.lbl.gov Sat Aug 15 04:00:12 1998 From: daemon@csg.lbl.gov Date: Sat Aug 15 04:00:16 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Aug 15 04:00:09 PDT 1998 Subject: how to do live software upgrades? Subject: Re: VxWorks image too large Subject: Re: BSP Porting kit experience? ------------------------------------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: how to do live software upgrades? Date: Sat, 08 Aug 1998 08:30:28 -0700 From: Todd Hoff Organization: Possibility Outpost Message-ID: <35CC6F14.6B715069@possibility.com> Reply-To: tmh@possibility.com We have a requirement to upgrade software hosted on a vxworks system while still maintaining service. These particular cards don't have a backup nor can their services be transfered to another system while the upgrade is in progress. It's not clear to me how a target, with c/c++ object code, can update its software while providing uninterrupted service. The code has to be downloaded into memory while the current system is running and then start. Does anyone have techniques for this? thanx - ------------------------------------------------------------------ tmh@possibility.com http://www.possibility.com/Tmh Desperate measures require the invention of desperate times. -- Todd Hoff --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks image too large Date: 13 Aug 1998 00:59:17 GMT From: antles@spk.hp.com (Tom Antles) Organization: Hewlett Packard Corvallis Site Message-ID: <6qtdp5$78b@hpcvsnz.cv.hp.com> References: <6qnntc$vp5$1@news3.infoave.net> Is your RAM_HIGH_ADRS too close to RAM_LOW_ADRS? I think the bsps come shipped default at 1Mbtye seperation. So if the bootrom residing at RAM_HIGH loads a network image larger than 1Meg into memory starting at ram_LOW_ADRS, you have problems. By the way, to see the size of your network image, use: size68k vxWorks. Faron Dutton (dutton@sccoast.net) wrote: : Hi everyone, : I am trying to network autoboot a MVME167 w/8MB RAM via FTP. The image file : is > 2MB and was configured through Tornado. The download fails with a : message on the target console complaining of an "Illegal Instruction" : exception. : Does anyone know what the problem is and how to correct it? : I have one suggestion to increase the target RAM to 16MB which is : unacceptable at the moment. : I have another suggestion to NFS mount the host and download the needed : object files after booting. I have tried this using loadModule and ld and : both fail. The messages I get are "loadAOutLib error: can't add 'X' to : system symbol table -- error == 0x3d0001." : I also have a TSR in at WRS, but they have not proposed a solution yet. : Any other suggestions would be greatly appreciated. : Thanks in advance, : Faron. - -- Tom Antles antles@spk.hp.com Phone: (509) 921-3590 Fax: (509) 921-3991 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: BSP Porting kit experience? Date: 13 Aug 1998 08:32:51 -0400 From: Michael Carr Organization: UltraNet Communications , an RCN Company http://www.ultranet.com/ Message-ID: References: <35d20dc6.3588621@news.jsc.nasa.gov> johncoby@removespam.orbitworld.net (John R. Cobarruvias) writes: > Anyone have any experience using the BSP porting kit? Yeah, and I think it's a total waste of $7000. When we bought it, we were misled (read: duped) into thinking it came with drivers for every part known to man, ported to vxWorks. It's just a bunch of cookie cutter templates. You're better off buying a 2-3 BSPs that closely match the hardware you're designing. Just my .02. --------------------------- End of New-News digest ********************** From daemon@csg.lbl.gov Sun Aug 16 04:00:11 1998 From: daemon@csg.lbl.gov Date: Sun Aug 16 04:00:15 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Aug 16 04:00:08 PDT 1998 Subject: RadStone PPC2A-604e & VxWorks 5.3.1 Subject: SENS: dhcp and netBuf questions... ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: RadStone PPC2A-604e & VxWorks 5.3.1 Date: Thu, 13 Aug 1998 15:11:02 -0500 From: Sandeep Neema Organization: MCSL, Vanderbilt Message-ID: <35D34856.920B40C1@vuse.vanderbilt.edu> Hello VxWorks, We have a RadStone PPC2A-604e board with VxWorks 5.3.1. I am new to both VxWorks and PPC. We have a Mizar 7772 DSP board on the same rack. The board is mapped to address 0xA0000000 in the A32 VME space. As per the Mizar documentation I added this address in the SysPhysMemAdrs array in the SysLib.c so that the MMU does not translate this address. Further the Mizar documentation claims that if I dump the memory at that address I should see their signature. However I get a "machine check" exception instead. I have couple of questions with reference to the above: 1) Is there something obvious that I am missing here? 2) Do I need to use sysBusToLocalAdrs/sysLocalToBusAdrs? 3) How can I determine the exact cause of a "machine check" exception? AFAIK a "machine check" exception could be caused by number of reasons. Thanks for your patience in reading this far. I would appreciate any help. Cheers, Sandeep mailto:neemask@vuse.vanderbilt.edu - -- - ------------------------------------------------------------------------ Sandeep Neema 511 Chesterfield Avenue, #9D Nashville, TN37212 Res: 615-297-6053 Off: 615-343-7566 - ------------------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: SENS: dhcp and netBuf questions... Date: Mon, 10 Aug 1998 11:41:47 -0400 From: "David Deephanphongs" Organization: Monmouth Internet Message-ID: <6qn481$8g6$1@news.monmouth.com> I've started working with SENS, and have some questions about dhcp and netBuf: With DHCP, in the dhcpsLeaseStorageHook, there is an 'op' argument. It is set to start, stop, write, read, or clear. The book isn't very clear, though. Does 'write' write out the entire dhcps data, or just the particular entry given, or is the entry the entirety of the dhcps data? Are entries repeated (i.e., will it send out an entry every time it renews a dhcp lease?) With the netBuf stuff, how do I get the SENS stack to free the buffers from the M_BLK CL_BLK pool? Thanks, David --------------------------- End of New-News digest ********************** From alfredd@netvision.net.il Sun Aug 16 04:13:28 1998 From: alfredd Date: Sun Aug 16 04:13:31 PDT 1998 Subject: Working with the CodeTap Ice & debugger on ADS860 Hello vxWorkers ! I am new to vxWorks. We develop on the Motorola ADS860 Evaluation Board. We got a CodeTap Emulator for Motorola MPC8XX with the MWX_ICE debugger. We don't have a license to use CodeConnect software, which as i understood enables a smooth connection between tornado and the CodeTap Emulator. I want to know how do i download vxWorks image to the target using this ICE. What gdb flags do i have to use in order for the MWX-ICE to recognize symbols & source code. And also how do i begin to run it and debug the C-source level code in it. Do i have to change the configuration of the BSP from the default ? I managed to create a vxWorks image that is bbotable, i.e. after we finish the initialization procedure it calls the main routine in the application we wrote. Thanks alot for your attention and ideas, Erez Har-Tal Albatronics, Omer Industrial Park, Omer ,Israel alfredd@netvision.net.il 972-7-6482579/80 From daemon@csg.lbl.gov Mon Aug 17 04:00:14 1998 From: daemon@csg.lbl.gov Date: Mon Aug 17 04:00:27 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Aug 17 04:00:11 PDT 1998 Subject: Multiple Readers/Writer Lock Implementation Sought Subject: Re: VxWorks TCP/IP in Promiscuous mode Subject: Re: Anyone have a any neat ways of sharing data between BSP and application code.... Subject: VxWorks make failing Subject: Re: Working with the CodeTap Ice & debugger on ADS860 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Multiple Readers/Writer Lock Implementation Sought Date: Tue, 11 Aug 1998 12:48:07 -0700 From: Grant Erickson Organization: Brocade Communications Systems, Inc. Message-ID: <35D09FF7.37F6@brocade.com> I realize that a multiple reader/single writer lock implementation could be developed fairly easily by building on top of the SEM_ID primitive already provided with VxWorks; however, does anyone have available or have a pointer to a well-tested, existing implementation? If any implementations are available, are they based on multiple SEM_IDs or do they rely on a single SEM_ID and instead twiddle the process control block in the kernel to figure out who should and should not get the lock? Regards, Grant Erickson - -- Grant Erickson Phone: (408) 487-8087 Brocade Communications Systems, Inc. Fax: (408) 487-8101 1901 Guadalupe Parkway San Jose, CA 95131 E-mail: gerickson@brocade.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks TCP/IP in Promiscuous mode Date: Wed, 12 Aug 1998 10:44:03 +0530 From: "Venkat" Organization: Novalink Technologies, New Delhi Message-ID: <6qr832$t11$1@taiwan.informatik.unirostock.de> References: Hi, The ethernet card can be put into promiscuous mode like this: int ifFlags; char *interfaceName = "elt0"; /*The 3Com EtherLink III NIC*/ ifFlagGet(interfaceName, &ifFlags); /*Get existing flags*/ ifFlagSet(interfaceName, ifFlags | IFF_PROMISC); This worked for me. However, are you sure this is what you want to do? Putting an NIC into promiscuous mode merely ensures that ALL incoming ethernet frames (and not just the ones destined for this NIC) are accepted by the NIC and are queued to the next higher networking layer e.g. IP. The IP layer may decide to discard some of these packets (e.g. if the destination IP address on the datagram does not match any of this host's IP addresses and IP forwarding is not set). Putting the ethernet interface in promiscuous mode will NOT change the behaviour of IP and layers above. Moreover, putting the ethernet interface in promiscuous mode will not affect OUTGOING packets in any way. Hope this helps. Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone have a any neat ways of sharing data between BSP and application code.... Date: Sun, 16 Aug 1998 21:35:41 GMT From: donbowman@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <6r7jbd$euc$1@nnrp1.dejanews.com> References: <35ce4cd3.0@205.238.18.7> <6qm871$qqu$1@nnrp1.dejanews.com> In article <35ce4cd3.0@205.238.18.7>, "John McQueen" wrote: > Hey all, > I would like to hear from you'll about creative ways of sharing data > between > the vxWorks kernel that you build and seperate application code data which > will > be used by the kernel at some point after the application code is loaded. > > You could define global variables in the BSP and extern them in the > application. > Any other ideas? > The approach I used was to have a 'RegisterPointer' function that maintained an association between arbitrary pointers and a string which was the 'name' for the data. Other code could retrieve the pointer with 'GetRegisteredPointer' passing in the same string. NULL or the original pointer would be returned. It was assumed that anything knowing the name knew the type. This takes advantage of the global nature of the memory space in vxWorks. - --don - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks make failing Date: 14 Aug 1998 17:26:03 GMT From: xwmacgreg@fox.nstn.ca (Will MacGregor) Organization: Newbridge Networks Message-ID: <6r1rvb$adc$1@hermes.achilles.net> Reply-To: Remove the x I am running Tornado 1.0.1 on a Solaris 2.6, VxWorks V5.3. When I attempt to make the Kernel, the compiler dies when attempting to compile the module usrConfig.c, with an illegal Instruction trap. Does anyone have any idea what is happenning here? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Working with the CodeTap Ice & debugger on ADS860 Date: 17 Aug 1998 04:17:24 GMT From: "yosiy" Organization: Telematics International, Inc. Message-ID: <01bdc996$f10bc640$c91eea93@pc-yosiy.ecitele.com> References: <35D74B85.205E697A@netvision.net.il> - -- Yosi Yarchi yosiy@ecitele.com alfredd wrote in article <35D74B85.205E697A@netvision.net.il>... > Hello vxWorkers ! > > > I want to know how do i download vxWorks image to the target using this > ICE. What gdb flags do i have to use in order for the MWX-ICE to > recognize > symbols & source code. And also how do i begin to run it and debug the > C-source level code in it. Do i have to change the configuration of the > BSP > from the default ? > Dear Erez (1) Copy the "iregs8xx.dat.ads" file into "iregs8xx.dat" file in your "c:\st8xx\amc\st8xx" directory. Next time you wake up the debuger, the ads860 target shall be initialized so you can load files into ram. (2) Prepare an image of vxWorks. If you are going to run also your application under the CodeTap, compile it with the image of vxWorks. use the flag "-gstabs+" for providing symbol information to the CodeTap. (3) Run the CodeTap software. When the connection with CodeTap is established, and target is under CodeTap control, load your vxWorks image into the target ram. (4) Put brakepoint where you want (5) Run the application Good Luck --------------------------- End of New-News digest ********************** From rtp.co.uk!ihw@rtp.co.uk Mon Aug 17 06:16:34 1998 From: Ian Willats Date: Mon Aug 17 06:16:37 PDT 1998 Subject: Symbios Logic / NCR SYM53C875 VxWorks Mike Cuddy asked: > It looks like the SCSI isn't delivering interrupt to the target. I assume it takes a second or so before the "state 1 -> 6" and "timeout" messages are reported? If so this is likely to mean that you are not getting interrupts from the 875. It should interrrupt to indicate successful selection or selection timeout (normally after 250 ms). To confirm this you can set scsiMinTimeout to -1 before calling scsiPhysDevCreate() - this will disable the timeout mechanism built into scsiLib. I would expect to see your scsiPhysDevCreate() hanging up indefinitely then. > I'm wondering if it's some kind of incompatibility between the 810 chip > (for which the driver was written) and the 875 (The chip I'm using). > although, the header files mention the 875 explicitly as a supported > chip. I don't think so; I have seen the unmodified ncr810Lib code work with two different implementations of the 53C875. A couple of things occur to me, however: 1) I take it that the difference between using i/o and memory space on the PCI bus is catered for elsewhere in your BSP? Or have you modified the relevant macros in ncr810Lib? Are you sure you can access the 875 registers? - try ncr810Show () to be sure. 2) I have seen a problem which has similar symptoms to yours, when using the 875 with narrow SCSI devices. I found that if the 875 was set not to use wide SCSI when it did the first selection after a reset, it never generated an interrupt. Once the first selection had been done in wide mode (even if there was no target device that responded to it), all subsequent selections could be done in narrow mode. I hacked some code into sysScsiInit() to call scsiPhysDevCreate() on an arbitrary busID and LUN, and to delete the phys dev if successful, before all the targets were set to operate in narrow mode. I never really got to the bottom of what was happening and why, but it might be worth trying. > (P.S., is it just me, or is the documentation for how SCSI-2 drivers work > in Vxworks crap?) Having written most of the SCSI-2 code (although not the Porting Kit documentation ;-) including the ncr710 driver on which the 810 driver is based, I may be able to help if you have any specific questions. 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 --------------------------------------------------------------- From JUSTIN.W.THOMAS@cpmx.saic.com Mon Aug 17 09:10:05 1998 From: "Thomas, Justin W." Date: Mon Aug 17 09:10:09 PDT 1998 Subject: executing a task at a known rate VxWorks Gurus, This should be a simple one. I have multiple tasks that I need to execute at known rates using the system clock. I could not find a direct way to perform that operation in the documentation, so I am currently using the wdog library. The function that I attach to the watchdog interrupt starts a new watchdog and then performs a semGive on a binary semaphore. The actual task is an infinite loop that holds on the semTake and then executes. This is fairly awkward especially with dozens of tasks and timers. It seems like an RTOS would have a better way to perform operations at regular intervals. Any suggestions? Justin ============================================== Justin W. Thomas SAIC 22299 Exploration Dr. 301-866-6705 Suite 200 301-863-0299 fax Lexington Park, MD 20653 justin.w.thomas@cpmx.saic.com ============================================== From daemon@csg.lbl.gov Tue Aug 18 04:00:14 1998 From: daemon@csg.lbl.gov Date: Tue Aug 18 04:05:27 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Aug 18 04:00:11 PDT 1998 Subject: Re: VGA code / driver for VxWorks.. Subject: VxWorks introduction whitepaper? Subject: Re: PowerPC Programming ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: VGA code / driver for VxWorks.. Date: Wed, 12 Aug 1998 16:04:46 GMT From: john@kivala.com Organization: Kivala Systems, Inc. Message-ID: <6qseet$mm0$1@nnrp1.dejanews.com> References: <199808111721.KAA03924@fensende.com> There is a product listed in Wind River's TradeWinds directory called "MetaWINDOW", from EASE Software, that appears to be a graphics library for x86 VxWorks. I know nothing about the product or the company, but it's probably worth a look. John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Build Tornado Apps with MS Visual Basic - -------------------------------------------------- In article <199808111721.KAA03924@fensende.com>, Mike Cuddy wrote: > Keywords: StrongARM, PCI, VGA, VxWorks 5.3.1, Tornado 1.0.1 > > I'm using VxWorks 5.3.1 / Tornado 1.0.0 on a StrongARM platform and > was wondering if anyone had any basic PCI VGA code. (for any chip ... I'll > get the right card ;-) Since on PC's, it's the SVGA bios that takes care > of setting up the CRTC chip to get just a plain "text mode" running, > there's precious little documentation on how to do that. > > I'd like to have access to the std. text mode, but my end goal is to > have some code which will let me access an SVGA card in 800x600x24 bit > mode over the PCI bus. > > I've seen the code for the pcconsole that comes with the WRS BSP DDK, > but it relies on the fact that the bios has already setup text mode. > > Any help will be greatly appreciated ... > > -- > Mike Cuddy (mcuddy@FensEnde.com, MC312), Programmer, Daddy, Human. > Fen's Ende Software, Redwood City, CA, USA, Earth, Sol System, Milky Way. > I remember asking why ... Let it rain, and protect us from this Cruel Sun. > > Join CAUCE: The Coalition Against Unsolicited Commercial E-mail. > > > - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks introduction whitepaper? Date: Mon, 17 Aug 1998 19:10:08 GMT From: johncoby@removespam.orbitworld.net (John R. Cobarruvias) Organization: Self Message-ID: <35d87fc2.18848700@news.jsc.nasa.gov> Is there an introduction to VxWorks written somewhere. I have a manager who wants to talk intelligently about the product and is looking for a good introduction. I've sent her to the programmers guide in the first few chapters. Is there something better? ANTI-SPAM EMAIL IN EFFECT. SEND EMAIL TO: johncoby@orbitworld.net --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PowerPC Programming Date: Tue, 18 Aug 1998 06:57:41 GMT From: td@imd.m.isar.de (Thomas Doerfler) Organization: IMD Message-ID: <35d9251d.1104508@news.isar.net> References: <6raej9$9t$1@nnrp1.dejanews.com> On Mon, 17 Aug 1998 23:32:57 GMT, mabarve@my-dejanews.com wrote: >Hi, > I have been trying to locate the PowerPC Assembly Programming >documentation. >In order to combine the C and assembly code, there are certain conventions for >passing >the values across C and assembly functions (such as r3, r4 .. for passing and >r3 for the >return values). Is there a complete specification for the C, assembly powerPC >programmers ? > Hi, I have heard about a document called something like "Compiler writers guide for PowerPC" that should be located on IBM's PowerPC web pages. A good start URL for a search should be http://www.chips.ibm.com/products/embedded/index.html You should know, that there are mainly two ABIs (Application Binary Interfaces), the EABI (embedded ABI) mainly used in relatime world and the PowerOPEN ABI. I hope this is a start for you. Bye, Thomas. > thanx in advance, > > Milind Barve > milind@acceleratednetworks.com > > >-----== Posted via Deja News, The Leader in Internet Discussion ==----- >http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum - -------------------------------------------- IMD Ingenieurbuero fuer Microcomputertechnik Thomas Doerfler Herbststrasse 8 D-82178 Puchheim Germany email: td@imd.m.isar.de --------------------------- End of New-News digest ********************** From kb+@andrew.cmu.edu Tue Aug 18 08:21:30 1998 From: Kevin Bradley Date: Tue Aug 18 08:21:34 PDT 1998 Subject: Linux Host and VxWorks Target This is a rather specific message, but I've noticed some traffic on my web page about this in the past year, so I feel obligated to make a post about it. If you: o Have VxWorks 5.3.x / Tornado 1.x, and o Are interested in using Linux as a host, rather than Windows NT or Solaris, or o Want to use a better C++ compiler than gcc 2.7.2 for your target and can't wait for the next relase, or o Want Pentium or Pentium Pro optimization, or o Have nothing better to do today at work, then check out the "How to Build GCC as a Cross-Compiler for VxWorks under Linux, Especially for x86 Targets" web page. http://usa.ece.cmu.edu/www/VxWorks-gcc.html It also has some hints as to how to use Linux entirely as a host system without Tornado tools, although you obviously lose Tornado-provided functionality. Since nothing in these directions is especially Linux-specific, but is rather GCC specific, then other host systems may also profit. The reason for the update: o Apparantly the directions work for gcc 2.8.1, and by extension, will probably apply to egcs as well. These compiler versions support C++, especially templates, much better, and also have optimizations for Pentium and Pentum Pro / II architectures. *WARNING*: exceptions may still be tricky. I wouldn't trust them to work properly, as the VxWorks libraries will not be compiled with the right flags. o Acquisition of a script for the "munch" utility, which was previously missing. Thanks to Dr. Max Fischer and Henric Jungheim for providing the information and the utility. Previous thanks to Casey Crellin, who provided a lot of help in the port. The next major stumbling block to overcome is to get WindRiver to support Linux directly. -- Kevin PS: cross-posted to VxWorks mailing list, VxWorks newsgroup, and Linux application development group. From ahw@clalf1.mps.ohio-state.edu Tue Aug 18 09:15:22 1998 From: Andreas Wolf Date: Tue Aug 18 09:15:25 PDT 1998 Subject: GNU c++ and exceptions Hi there, here are my 2 cents about C++ exceptions (*not* hardware excpetions) on VxWorks. You probably saw this in the gcc-2.8.1 posting > *WARNING*: exceptions may still be tricky. I wouldn't > trust them to work properly, as the VxWorks libraries > will not be compiled with the right flags. Even *if* the internal libraries were compiled with the correct flags, I would *NEVER* use exceptions on VxWorks!! Reason: All versions of the GNU compiler I encountered up to now handle exceptions in a *non-reentrant* way. As an experienced VxWorks users, you *must* conclude that this C++ feature is therefore too dangerous to use in the real world. I saw (and done myself) various patches to g++ but none of them solves this problem in a satisfying generic way for VxWorks. > The next major stumbling block to overcome is to get WindRiver > to support Linux directly. I beg to differ, since I'd rather like to have exceptions working, first... ;-)) Andreas =:-) From jtmark@most.fw.hac.com Tue Aug 18 09:27:35 1998 From: "Jimmie T Marks" Date: Tue Aug 18 09:27:38 PDT 1998 Subject: proxy arp I am ramping up on a project which uses VxWorks (Tornado) on a Pentium target running on a PCI backplane. I want to configure the backplane with 2 seperate network cards to create a gateway. I want to configure one of the network interfaces to be the main network and route datagrams to the secondary network based on my routing tables. On the main network I want to suppory ICMP and proxy ARP. All the VxWorks documentation talks about using proxy arp with a shared memory network, but not in the context I want for routing and using ARP. Can this be done? Has it been done before? What steps do I take to implement it? Thanks Jimmie T. Marks jtmark@most.fw.hac.com From krueger@comsys.rockwell.com Tue Aug 18 10:47:50 1998 From: Bill Krueger Date: Tue Aug 18 10:47:54 PDT 1998 Subject: vxworks driver needed I am in need of a driver for an PCMCIA RS-422 serial adapter which provides HDLC communications capability. One card that I have looked at to do this job is the Quatech MPAP-200/300. Unfortunately, their driver selection is quite limited. If anyone knows of a vxWorks driver for this or any other PC card that will do the same thing, please let me know. Thanks in advance, Bill Krueger From daemon@csg.lbl.gov Wed Aug 19 04:00:17 1998 From: daemon@csg.lbl.gov Date: Wed Aug 19 04:05:29 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 19 04:00:14 PDT 1998 Subject: Re: RadStone PPC2A-604e & VxWorks 5.3.1 Subject: Re: NT Host & FTP Subject: Re: VxWorks image too large Subject: message queues Subject: Re: Java on VxWorks Subject: Breakpoint Marker? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: RadStone PPC2A-604e & VxWorks 5.3.1 Date: Tue, 18 Aug 1998 07:24:37 -0400 From: Gus DiPierro Organization: Raytheon Systems Company, Falls Church Message-ID: <35D96475.CE29F666@fallschurch.esys.com> References: <35D34856.920B40C1@vuse.vanderbilt.edu> <6rbbvt$8be@gcsin3.geccs.gecm.com> Stephen A Walsh wrote: > > Sandeep Neema wrote: > >Hello VxWorks, > > > >We have a RadStone PPC2A-604e board with VxWorks 5.3.1. I am new to > >both VxWorks and PPC. We have a Mizar 7772 DSP board on the same rack. > >The board is mapped to address 0xA0000000 in the A32 VME space. The address here is a vme address. The addresses that go into the sysLib.c file are PPC local addresses.( Don't forget you are going through 2 bridges to get to the VME.) So take 0xa0000000 out of the syslib.c. The radstone BSP already takes the pci mapping into account. (I believe it is put into the BAT registers) > >As per the Mizar documentation I added this address in the > >SysPhysMemAdrs > >array in the SysLib.c so that the MMU does not translate this address. > >Further the Mizar > >documentation claims that if I dump the memory at that address I should > >see their > >signature. However I get a "machine check" exception instead. > > > >I have couple of questions with reference to the above: > >1) Is there something obvious that I am missing here? > > As well as using the functions below it would be advisable to > check the settings in config.h for you VME windows. > > It should probably look something like this for you setup > > #define VME_A32_MSTR_SIZE 0x10000000 /* Size of A32 Window */ > #define VME_A32_MSTR_BUS 0xA0000000 /* First Bus Address */ > > For more info on VME windows try downloading Windriver Technical > Note 44 from thier website. > > Note: There is a limit to the size of VME_A32_MSTR_SIZE but i'm > not quite sure what it is (I think it may be 0x30000000 but it > may be less). > The problem with the size is that the VME address space is mapped directly on the PCI address space. Since the ppc's window into the PCI is only from 0xc0880000 - 0xfeffffff (ppc local), it does not leave you alot to play with. > >2) Do I need to use sysBusToLocalAdrs/sysLocalToBusAdrs? > > Yes this is very important. It is the only way for the processor to figure out the local address equivalent of the vme address. > > >3) How can I determine the exact cause of a "machine check" exception? > >AFAIK a > > "machine check" exception could be caused by number of reasons. > > > I'm not sure about this one. The reason you are getting a machine check is because you are probably trying to access the 0xa0000000 address locally which should be in the pci I/O space "weeds" and you are getting a bus error. In short, you need to setup the config.h file's vme address windows and use the sysBusToLocalAdrs function. For vme accesses the universeII can do either coupled or posted writes and only coupled reads. On a coupled access, the pci bus is not released until the vme transaction is done. If there is a vme bus error during this time the universeII will signal a target abort on the pci bus. The ppc/pci bridge is set up to send a machine check exception to the PPC if this happens. Also you can use vxMemProbe with the address you get from sysBusToLocalAdrs. This will test for a vme response to the address without receiving a Machine check exception if there is an error. - -- _______________________ Gus DiPierro Mail Stop M113 E-mail: gdipierro@fallschurch.esys.com RAYTHEON Systems Company dipierro@erols.com (personal) 7700 Arlington Blvd. Voice: (703) 560-5000 ext. 2937 Falls Church, VA 22042 Fax: (703) 280-4627 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NT Host & FTP Date: Tue, 18 Aug 1998 07:29:05 -0400 From: Gus DiPierro Organization: Raytheon Systems Company, Falls Church Message-ID: <35D96581.19CB96B4@fallschurch.esys.com> References: <35D84A21.10B90990@ti.com> We have not had a problem. We set up IIS and turned on the ftp service. We log in using anonymous and the passwd does not matter(type in anything). Make sure that you have enabled anonymous access. Doug Lewis wrote: > > Has anyone been able to get a PPC board to "boot" from an NT host with > the Microsoft FTP service running? I can't seem to get the > permissions/rights/something right on the accounts to let the boards > login and download the vxworks kernel. > > We had been using the wftp program that came with Tornado but that > requires that someone be logged on to the server to have the program > running. We want to use the Microsoft FTP service so that it's running > as a background serice on the server so no has to be logged on and > running the wftp software. > > -- > Doug Lewis ---------------------------------- > Senior Software Eng - The views expressed are my own & > Raytheon Systems Co. - do not necessarily reflect those > - of Raytheon Systems Company > ---------------------------------- - -- _______________________ Gus DiPierro Mail Stop M113 E-mail: gdipierro@fallschurch.esys.com RAYTHEON Systems Company dipierro@erols.com (personal) 7700 Arlington Blvd. Voice: (703) 560-5000 ext. 2937 Falls Church, VA 22042 Fax: (703) 280-4627 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks image too large Date: Tue, 18 Aug 1998 14:03:42 +0200 From: Paul Chien Organization: Mecalc (Pty) Ltd Message-ID: <35D96D9E.2C344CD9@mecalc.co.za> References: <6qnntc$vp5$1@news3.infoave.net> Hi Faron, I had a similar TSR with WRS: The two primary limitations that are the cause of most trouble are: 1) The size of the image being booted via FTP is > (largest free memory block in system memory pool / 2). When bringing in an object, it is first placed into memory in sections (IP packets) and then "rebuilt" in the correct order. The way it is implemented requires a contiguous block of memory. This does not sound like your problem, as your board has 8Mb of RAM, and your image is only 2 Mb, and since you are just now booting, your memory will not be heavily fragmented. 2) The offset between RAM_LOW_ADRS and RAM_HIGH_ADRS is insufficient for the final image, causing the boot code (starts at RAM_HIGH_ADRS) to be overwritten by the placement of the runtime image (starts at RAM_LOW_ADRS and continues for (image size) bytes). If you are being victimized by #2, you have to rebuilt and reprogram the bootrom, but not neccessary for the runtime image (since it is not being moved from 0x00100000, it shouldn't care where RAM_HIGH_ADRS is). To be safe, a rebuild wouldn't hurt. Hope this helps Cheers Paul Faron Dutton wrote: > > Hi everyone, > > I am trying to network autoboot a MVME167 w/8MB RAM via FTP. The image file > is > 2MB and was configured through Tornado. The download fails with a > message on the target console complaining of an "Illegal Instruction" > exception. > > Does anyone know what the problem is and how to correct it? > > I have one suggestion to increase the target RAM to 16MB which is > unacceptable at the moment. > > I have another suggestion to NFS mount the host and download the needed > object files after booting. I have tried this using loadModule and ld and > both fail. The messages I get are "loadAOutLib error: can't add 'X' to > system symbol table -- error == 0x3d0001." > > I also have a TSR in at WRS, but they have not proposed a solution yet. > > Any other suggestions would be greatly appreciated. > > Thanks in advance, > Faron. - -- - ---------------------------------------------------------------------- | Paul Chien Mecalc (Pty) Ltd | Project Engineer Wind River Systems Distributor, South Africa. | Tel: +27-12-665-1480 P.O.Box 7958, Centurion 0046, South Africa. | Fax: +27-12-665-1495 86 Oak Avenue, Highveld Technopark, Centurion. | paul@mecalc.co.za http://www.mecalc.co.za - ---------------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: message queues Date: Tue, 18 Aug 1998 16:45:56 +0200 From: Ofer Goren Organization: Baynetworks Israel Message-ID: <35D993A4.6A68BED3@baynetworks.com> Hi. Simple question: I have a task ("firstTask") that creates a message queue, and then spawns two new tasks the use the created message queue. Now If I kill (taskDelete()) the "firstTask" task, will the message queue remains, or will it be erased too? Thanks, - -- Ofer Goren Email:ogoren@baynetworks.com phone number: 9723-6456023 fax number: 9723-6479579 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Java on VxWorks Date: Wed, 12 Aug 1998 20:19:23 GMT From: blair@xvn.ds.boeing.com (Rick Blair) Organization: The Boeing Company Message-ID: References: <35D19163.22A59B4C@de.ibm.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) [Posted and mailed] In article <35D19163.22A59B4C@de.ibm.com>, Eberhard Amann writes: > we are looking for JAVA on VxWorks & a implementation of the JAVA AWT Class. > > Does anyone know if this is available ?, What about Power PC ? > Hi, I am currently using Java for tornado on a mvme2604 card. I got a beta version and it has been rock solid. My application hits the network pretty hard and spawns quite a few tasks. This application also runs for weeks/months and it still is very solid. No crashes!! I do a memShow to look at the memory allocation and there seems to be no memory leaks. All in All I am very happy with the product. I have used released software that was not as good as this beta! My application does not make any awt calls so I cannot help you there! Now for the bad news..... It COSTS WAY WAY WAY TOO MUCH........ IMHO. Also I do not think that there is a jit compiler yet, but I could be wrong. I also dont know the current product road map. If anyone does, I would be happy to see it!!! Rick - -- ************************************************************************** * email: blair@xvn.ds.boeing.com * The contents of this message in * * voice: (253)773-8962 * in no way reflect the opinions * # Page (206)986-0795 * of the Boeing Company. Opinions * * usSnail: P.O. Box 3707 M/S 3E-28 * expressed here are strictly my own.* * Seattle, Wa 98124 * I DO NOT Speak for The Boeing Co. * * ------------- SUPPORT YOUR LOCAL POP3 SERVER -------------- * ************************************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Breakpoint Marker? Date: Tue, 18 Aug 1998 17:22:20 -0400 From: Tim Sohacki Organization: Nortel (Northern Telecom) Message-ID: <35D9F08C.650ECD85@americasm01.nt.com> Any suggestions on the following problem? I set a breakpoint via Tornado 1.0.1 (Win95) from the tool bar by highlighting the source line and clicking "Toggle Breakpoint." No marker is displayed in the attribute pane. When the code executes, the breakpoint is hit and breaks. When I single step past the location pointer shows up correctly. Thanks, Tim --------------------------- End of New-News digest ********************** From eckardt@innocon.com Wed Aug 19 09:05:54 1998 From: Suzie Eckardt Date: Wed Aug 19 09:06:00 PDT 1998 Subject: SONIC Ethernet Driver Can anyone point me to some good documentation on the operation of the SONIC ethernet driver provided with VxWorks? Anyone have experience modifying this driver to operate on their system? One of the things I need to do is to force the SONIC to use a particular memory device for its buffers/descriptors. Thanks, Suzie ************************************************ ** Suzie Eckardt ** Innovative Concepts, Incorporated ** McLean, VA ** eckardt@innocon.com ** 703-893-2007 x258 ************************************************ From mcuddy@fensende.com Wed Aug 19 14:16:44 1998 From: Mike Cuddy Date: Wed Aug 19 14:16:48 PDT 1998 Subject: More tales of SCSI and woe! Keywords: NCR810, NCR875, vxWorks 5.3.1, SCSI driver, little-endian, StrongARM Thanks to the help of a few people on this list, I've gotten my SCSI driver sort-of working. (DEC StrongARM SA-110/EBSA-285 BSP, Symbios Logic 53C875 scsi chip) The problems du'jour ... My system has 2 SCSI disk devices at Target 2 and Target 4. I wrote some code to auto-configure the SCSI bus (yes, I know that I can use scsiAutoConfig, but that function seems terribly slow, so I wrote my own) The code that configures the SCSI bus looks like this: STATUS scsiConfigureBus(SCSI_CTRL *pCtrl, char *name) { int bus_id, lun; SCSI_PHYS_DEV *pPhys; int devCnt = 0; for ( bus_id = 0 ; bus_id < 7; bus_id++ ) { // skip bus id of controller... if (bus_id == pCtrl->scsiCtrlBusId) continue; lun = 0; { taskDelay(100); SLog("scsi: %s/id%d/lun%d .. \n", name,bus_id,lun); pPhys = scsiPhysDevCreate(pCtrl, bus_id, lun, -1, 0, 0, 0, 0); if (pPhys) { // found one! SLog("scsi: %s/id%d/lun%d: physdev=0x%x, vendor='%s', id='%s'\n", name, bus_id, lun, pPhys, pPhys->devVendorID, pPhys-> SLog("... blks=%d, blksz=%d, type=%s, removable=%s\n", pPhys->numBlocks, pPhys->blockSize, scsiDeviceTypeStr pPhys->removable ? "yes" : "no"); devCnt++; } else SLog("scsi: %s/id%d/lun%d: no device found.\n",name,bus_id } } SLog("scsi: %s: found %d devices.\n", name, devCnt); return OK; } (SLog() is a function of my own which vsprintf()s it's arguments and calls logMsg() ... why is WRS afraid of varargs functions? Logging a message with 6 ",0"s at the end is just ugly, stack-consuming and braindead). .. From a cold boot, the first time I run this code (from the host-shell), I get: 0xf2b090 (t3): scsi: scsi0/id0/lun0 .. ++ 0xf34e68 (tScsiTask): ncr810WideXferParamsSet: pThread == NULL 0xf2b090 (t3): scsi: scsi0/id0/lun0: no device found. 0xf2b090 (t3): scsi: scsi0/id1/lun0 .. ++ 0xf34e68 (tScsiTask): ncr810WideXferParamsSet: pThread == NULL 0xf2b090 (t3): scsi: scsi0/id1/lun0: no device found. 0xf2b090 (t3): scsi: scsi0/id2/lun0 .. ++ 0xf34e68 (tScsiTask): ncr810WideXferParamsSet: pThread == NULL 0xf2b090 (t3): memPartFree: invalid block 0xffcaa8 in partition 0x8e084. The lines marked with ++'s are from the NCR driver -- apparently, the SCSI system calls into the driver with pScsiCtrl->pThread == NULL. Per someones's suggestion, I use pScsiCtrl->pIdentThread when that condition occurs. The task (t3) is now hung in taskSuspend() (the default action for the system memory partition manager ;-) A stack traceback of the hung task indicates that the corrupted memory block was free()'ed by scsiTargetOptionsGet(). 0 = taskSuspend 1 = memPartFree 2 = free 3 = scsiTargetOptionsGet 4 = scsiPhysDevCreate 5 = scsiConfigureBus 6 = wdbFuncCallLibInit The curious bit is, that if I CTRL-BREAK the suspended task, re-init my scsi controller, and re-invoke scsiConfigureBus() from the shell, the code continues past id'ing UNIT 2 successfully, and bombs out WITH THE EXACT SAME ERROR when it gets to UNIT4 (the other SCSI device on the bus) 0xf25778 (t5): scsi: scsi0/id0/lun0 .. 0xf205e0 (tScsiTask): ncr810WideXferParamsSet: pThread == NULL 0xf25778 (t5): scsi: scsi0/id0/lun0: no device found. 0xf25778 (t5): scsi: scsi0/id1/lun0 .. 0xf205e0 (tScsiTask): ncr810WideXferParamsSet: pThread == NULL 0xf25778 (t5): scsi: scsi0/id1/lun0: no device found. 0xf25778 (t5): scsi: scsi0/id2/lun0 .. 0xf205e0 (tScsiTask): ncr810WideXferParamsSet: pThread == NULL 0xf25778 (t5): scsi: scsi0/id2/lun0: physdev=0xf295e0, vendor='SEAGATE ', id='ST19101W ' 0xf25778 (t5): ... blks=17783240, blksz=512, type=direct, removable=no 0xf25778 (t5): scsi: scsi0/id3/lun0 .. 0xf205e0 (tScsiTask): ncr810WideXferParamsSet: pThread == NULL 0xf25778 (t5): scsi: scsi0/id3/lun0: no device found. 0xf25778 (t5): scsi: scsi0/id4/lun0 .. 0xf205e0 (tScsiTask): ncr810WideXferParamsSet: pThread == NULL 0xf25778 (t5): memPartFree: invalid block 0xf281ac in partition 0x8e084. And, just to add insult to injury, if I then CTRL-BREAK this second, suspended task, re-init my SCSI controller, and run the scsiConfigureBus() function one more time, it SUCCEEDS! (I haven't tried three devices to see if I have to run it four times to make it succeed :-). Now, to throw a few more cod in the barrel: If I add some logging messages into the ncr810 driver code, the failure goes away, to be replaced by different failures. The best of which is when the ncr810 driver is asked by the scsi system for a data transfer to 0xEEEEEEEE of 0xEEEEEEEE bytes (look familiar?) which causes the scsi chip to signal a bus-fault interrupt to the host (since 0xEEEEEEEE is not mapped to the PCI bus, and the PCI transaction ends in an abort). There appears to be a race condition between the ncr810 driver and the SCSI driver, or maybe the version of the SCSI driver that I have (01k,10mar97,dds) is just out of date with respect to the version of vxWorks that I'm running (it's from the latest WRS DDK -- but WRS doesn't seem to be big on keeping drivers up to date with the O/S releases). Or maybe I screwed up (likely ;-) ... moving on ... Once the scsiConfigureBus() function successfully detects the drives and adds them to the system (I can see them in the SCSI_CTRL structure ...) I call dosPartMountScsi() (from the WRS web page; it reads MS-DOS style partition tables and creates dosfs filesystem devices based on what it finds (that code is broken, too, it's got endianess bugs, but I fixed them, see below ;-) The dosPartMountScsi() code finds the partitions that I've created on the drives (a 500 MB partition on one drive, and a 2048MB partition on the other drive), and creates the appropriate dosfs filesystems (they show up with a call to devs() and I can opendir() and readdir() them. I have a little test function which I wrote to excercise the drives. The test function writes a bunch of data to a file, and then reads the data back, and verifies that the data is good. The data written is generated by a simple random number generator seeded with "test_pass_number" and "test_block_number" so that the results are reproducable, but otherwise the data is "garbage" ;-) The test boils down to: buffer = malloc(test_buffer_size); fd = open("/ ...a unique filename", O_WRONLY | O_TRUNC | O_CREAT) test_pass_number = 0; while (test_pass_number < max_test_passes ) { test_pass_number++; logMsg("Pass %d ... ", test_pass_number, .... blah ... ); test_block_number = 0; while (test_block_number < test_blocks ) { // seed random number generator and put "random data" // in "buffer" write(fd, buffer, test_buffer_size) test_block_number++; } test_block_number = test_blocks - 1; while ( test_block_number > 0 ) { lseek(fd, test_block_number * test_buffer_size, SEEK_SET); read(fd, buffer, test_buffer_size); // re-seed random number generator and compare data // read from disk with } } (copious error checking removed) This test will succeed if it's run once. This test will fail under any of the following circumstances: * more than one pass through the test is done. * more than one task per drive is run. (although multiple tasks testing multiple drives seems okay) The failures generally come in one of two flavors, a WDB exception which hangs the target (So much for "restarting in 10 seconds" ;-) or a vxWorks kernel exception ("kernel work queue overflow"). I/O throughput is VERY slow. (on the order of 1 MB/second). The drives involved are Seagate Cheetah, 10,000RPM, 9.1 GB, Ultra SCSI drives -- on a PC with this same controller, they will give between 13 and 20 MB/sec, so I know it's not the drives that are at fault. The NCR810 driver has a function called ncr810Delay(). which is just a busy-wait loop. The original code (which I gather was written for a PPC chip, had "10" as the loop limit value. when I had the delay at 10, I got no response from the SCSI driver at all, when I upped the delay to 1000, I started getting things in the "kinda" working state that they are in now. I suspect that this large delay is partly responsible for the slow transfer rates (as it apparently takes many interrupts for the ncr810 driver to actually do a transfer to/from the scsi drive). I hope I've given enough information (and not too much ! ;-) for someone to give me a hand with this, I'm really in a corner. Sorry if I seem like I'm whining a bit (I am!). But, I switched to vxWorks, away from another RTOS, because I was led (misled?) to believe that I wouldn't have to write all my own drivers with vxWorks ... the problem is that the other RTOS had source, so when I had problems with it, I could just delve into the source code, and not have to go through this "forensic debugging" process. (I suppose I could shell out another LARGE pile of cash to WRS for a source license ... *SIGH*) The main category of problems that I've had with vxWorks is that it definately shows it's big-endian roots ... I've encountered a half a dozen different ways by which bytes are swapped in drivers and not one of them actually worked. ** Endianness woes ... Here's a tip, for extracting a 32-bit quantity in "host" format, from a 32-bit little-endian value stored in a character pointer: UINT32 val; UINT8 *buf; /* little endian to host format */ dwVal = (buf[3]<<24) | (buf[2]<<16) | (buf[1]<<8) | (buf[0]); /* big endian to host format */ dwVal = (buf[0]<<24) | (buf[1]<<16) | (buf[2]<<8) | (buf[3]); These routines work AS IS on either host-endianness because the shifts take care of it for you; and it works regardless of alignment of 'buf', Yes, the shifts are not as fast as a native byte swapping; but most of the stuff that I've endian fixed in the WRS code was in setup code (the lnPci driver's little endian code did not work for copying the ethernet MAC address into the chip's initialization structure, the dosPartLibScsi.c code didn't extract the starting sector/offsets correctly from the data buffer read from the disk, etc.). Thank you for your patience with a vxWorks (but not a programming or embedded systems) newbie ... -- Mike Cuddy (mcuddy@FensEnde.com, MC312), Programmer, Daddy, Human. Fen's Ende Software, Redwood City, CA, USA, Earth, Sol System, Milky Way. I remember asking why ... Let it rain, and protect us from this Cruel Sun. Join CAUCE: The Coalition Against Unsolicited Commercial E-mail. From daemon@csg.lbl.gov Thu Aug 20 04:00:12 1998 From: daemon@csg.lbl.gov Date: Thu Aug 20 04:00:16 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 20 04:00:08 PDT 1998 Subject: Console does not accept keyboard i/p Subject: task status "READY" Subject: Help on STL ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Console does not accept keyboard i/p Date: Wed, 19 Aug 1998 17:24:36 +0530 From: "Venkat" Organization: Novalink Technologies, New Delhi Message-ID: <6ree6u$qg$1@taiwan.informatik.unirostock.de> Hello, VxWorks 5.3.1 486 BSP I have a Pentium PC running the above, booting with bootrom_uncmp off a floppy disk and fetching the vxWorks image from the host using netDrv. Both during the boot phase and after the OS starts up, my (target) PC's keyboard does not work. logMsg() and printf() work fine, i.e. output is shown on the VGA screen as expected, but the system waits on scanf() even when I type in something. The characters I type are not echoed on the screen. Even when the boot-up message "Press any key to stop auto-boot" appears, my system ignores my keyboard. The 'devs' command shows this: drv name 0 /null 1 /tyCo/0 1 /tyCo/1 2 /pcConsole/0 2 /pcConsole/1 7 develop: 8 /vio Anybody seen this before? Any modules to be INCLUDEd that I haven't? Any magic #defines? NVRAM settings? Thanks for any help. Venkat --------------------------- Newsgroups: comp.os.vxworks Subject: task status "READY" Date: Wed, 19 Aug 1998 23:56:08 +0900 From: "sungyeon" Organization: SERI Message-ID: <6rep2h$9km$1@usenet.kreonet.re.kr> HI.. ALL.. i spawn a task . for example - -> sp test_task but " test_task" is stop at vxTaskEntry and task's status is "READY" why ??? --------------------------- Newsgroups: comp.os.vxworks Subject: Help on STL Date: Wed, 19 Aug 1998 18:15:32 -0700 From: Ming Sun Message-ID: <35DB78B4.9D84045E@NO_SPAM.lightera.com> Hi, If you know where I can find a STL works on vxworks or instruction to port STL to vxworks, please send me a pointer. Thanks in advance, Ming Sun msun@lightera.com --------------------------- End of New-News digest ********************** From vyas@teil.soft.net Thu Aug 20 06:26:20 1998 From: Vyasaraj S Date: Thu Aug 20 06:26:24 PDT 1998 Subject: vxsim hi vxwork's guru's, i am using tornado 1.0.1 , vxsim to test my network protocol task. i am need of having two applications on different vxsim's ( vxworks target ) communicate with each other.. Example : Client in vxsim1 , should communicate with server in vxsim2. Is there any way to achieve this communication in vxsim. any help is appreciated. thanks in advance. vyas From percheron@enertec.avicore.com Thu Aug 20 07:30:55 1998 From: "percheron cecile" Date: Thu Aug 20 07:30:58 PDT 1998 Subject: vxworks ncr875 driver I don't know wether you remember about my problem about ncr875 driver. Finally, thinks are getting better. I found out that my PCI configuration was not good. Bus master option was missing. So the scripts could'nt be sent . I can communicate with my target but only with SCSI Narrow Transfers, which doesn't fit me at all. By default, the communication is done on wide transfers. But the negociation fails. It's fine under Win95. I tried function such as scsiTargetOptionsSet ( I use it to get back on narrow and so to be able to start communicating) What I do is that following sequence: 1- call to ncr810CtrlCreate 2- call to ncr810CtrlInit "Everything seems fine until there." 3-call to scsiPhysDevCreate I get a Phase Mismatch Event ( while I receive the Inquiry Data, in a Data In phase, I move to a MsgOut phase "extented Message to negociate Wide transfer" ) On SCSI spy, I can see the following steps: -Arbitration -Selection (identify+disconnect OK) -Command Inquiry -DataIn with ONLY ONE REQ/ACK the Wide transfer is already asked without negociation ( ?? !!) the timing seems OK no time out seems to be detected ->it might be because the negociation was not done yet. So it may stop the transfer. -MessageOut WIDE DATA TRANSFER is done with correct parameters. -Status Check condition is sent -Message IN Command COmpleted is sent. The negociation is never done. Don't you think that the messageOut WDTR should be sent only after the Status Check condition and command completed ?? The target may not be able to negociate since she didn't finished her command reply? In the script, the last instruction sent to the chip is in the DATA IN procedure ("doDataIn") . Could you give me your idea about that? Thank you so much... Best Regards. /************************************** percheron@enertec.avicore.com tel: 01.30.70.31.15 ENERTEC. 78141 Velizy -FRANCE /************************************** From c4s@ssd.ray.com Thu Aug 20 11:26:51 1998 From: Sefranek Date: Thu Aug 20 11:26:54 PDT 1998 Subject: ADA for VxWorks? What is available for developing ADA to run under VxWorks? What are people currently using? How do you like it? Please email responses to: c4s@ssd.ray.com If I get enough info, I'll summarize and post to the list. Thanks in advance... -- Charles A. Sefranek [c4s@ssd.ray.com, Telecon:(401) 842-3397] From daveg@progeny.net Thu Aug 20 18:22:21 1998 From: "David Garner" Date: Thu Aug 20 18:22:29 PDT 1998 Subject: SCSI on PMC - Please Help! vxWorks Gurus, First let me apologize for the lengthy post. We are trying to implement a scsi PMC card manufactured by Technobox Inc. into the MV2604 PowerPC. The scsi controller has been installed into the PMC site located on the base board and the dip switches have been set as follows: INT-A TERM-ENAB Little Endian Another VME card (Phoenix International) with two scsi hard disks at ID 3 & 4 have been installed into slot 2 and a 64 to 50-pin cable connects the controller to the drive card. The scsi controller has a SYMBIOS 53C875 chip and I've been told by Wind River that the controller on the base board and its libraries are compatible with it. I've attempted to make changes to the kernel so that the PMC scsi will be initialized instead of the base board scsi. I probed the device to extract the device/vendor info and made the appropriate addition to mv2600.h. Listed below are all the changes I've made to the kernel. sysScsiInit runs successfully (no error messages) but sysScsiConfig hangs. I know that there are many more commands to issue to make a dos FS but I'm not at that point yet. Would someone familiar with vxworks and scsi please tell me what I'm missing. Thanks in advance, Dave Garner Lockheed Martin Federal Systems david.garner@lmco.com config.h: ======================================================================== #define SCSI_PMC #define EXTENDED_VME #define INCLUDE_PRIMARY_SCSI /* NCR 825 */ #undef INCLUDE_SECONDARY_SCSI /* [NOT AVAILABLE] */ #define INCLUDE_SCSI #undef SCSI_AUTO_CONFIG /* scan bus on startup */ #undef SCSI_WIDE_ENABLE /* enable wide SCSI on MVME761-011 only */ #undef SYS_SCSI_CONFIG /* call sysScsiConfig in sysScsi.c */ mv2600.h: ======================================================================= #define PCI_ID_SCSI 0x00031000 /* Id for SYM53C825/75 Chips */ #define PCI_ID_TECHNOBOX_SCSI 0x000f1000 /* Id for Technobox PMC */ #ifdef SCSI_PMC #define SCSI_DEV_SPACE 0x40000 /* Technobox controller in PMC site */ #else #define SCSI_DEV_SPACE 0x10000 /* baseboard scsi controller */ #endif sysLib.c: sysHwInit(): ======================================================================== #ifdef INCLUDE_SCSI /* * Initialize the Standard PCI Header of the SCSI device if present */ #ifdef SCSI_PMC /* controller on pmc */ if ((DEVICE_PRESENT(MV2600_BMFR_SCIP) == TRUE) && (pciFindDevice ((PCI_ID_TECHNOBOX_SCSI & 0xFFFF), (PCI_ID_TECHNOBOX_SCSI >> 16) & 0xFFFF, 0, &pciBusNo, &pciDevNo, &pciFuncNo) != ERROR)) { (void)pciDevConfig (pciBusNo, pciDevNo, pciFuncNo, PCI_IO_SCSI_ADRS, NULL, (PCI_CMD_MASTER_ENABLE | PCI_CMD_IO_ENABLE)); } else { #endif /* controller on base board */ if ((DEVICE_PRESENT(MV2600_BMFR_SCIP) == TRUE) && (pciFindDevice ((PCI_ID_SCSI & 0xFFFF), (PCI_ID_SCSI >> 16) & 0xFFFF, 0, &pciBusNo, &pciDevNo, &pciFuncNo) != ERROR)) { (void)pciDevConfig (pciBusNo, pciDevNo, pciFuncNo, PCI_IO_SCSI_ADRS, NULL, (PCI_CMD_MASTER_ENABLE | PCI_CMD_IO_ENABLE)); } #ifdef SCSI_PMC } #endif #endif /* INCLUDE_SCSI */ sysScsi.c: ======================================================================== STATUS sysScsiInit () /* called from tRootTask */ { int pciBusNo; int pciDevNo; int pciFuncNo; UINT16 devType; static BOOL firstTime = TRUE; /* Local structure(s) with a prefill for ncr825SetHwRegister */ static NCR810_HW_REGS hwRegs = MV2600_SIOP_HW_REGS; /* Set CPU <-> PCI address offset */ ncr810PciMemOffset = PCI2DRAM_BASE_ADRS; /* * Since the PCI bus is little endian, and the PPC CPU's are big endian, * the ncr8xx scripts need to be byte swapped in memory. However, they * should only be swapped the first time that sysScsiInit() is invoked. */ scsiDebug = TRUE; SCSI_DEBUG_MSG ("Initializing SCSI controller\n", 0,0,0,0,0,0); if (firstTime) { swapScript ((UINT32 *) &ncr810Wait, (UINT32 *) ((UINT32) &ncr810Wait + (UINT32)(NCR810_INSTRUCTIONS * 8))); firstTime = FALSE; } /* * Try to automatically configure the correct type of NCR8XX controller. * The only options are the NCR825 and NCR810 controller. */ /* DRG 08-18-98 Added Technobox SCSI controller */ if (pciFindDevice ((PCI_ID_TECHNOBOX_SCSI & 0xFFFF), (PCI_ID_TECHNOBOX_SCSI >> 16) & 0xFFFF, 0, &pciBusNo, &pciDevNo, &pciFuncNo) != ERROR) { SCSI_DEBUG_MSG ("Found a SYMBIOS 53C875 Controller\n", 0,0,0,0,0,0); } else if (pciFindDevice ((PCI_ID_NCR825 & 0xFFFF), (PCI_ID_NCR825 >> 16) & 0xFFFF, 0, &pciBusNo, &pciDevNo, &pciFuncNo) != ERROR) { SCSI_DEBUG_MSG ("Found an NCR825 Controller\n", 0,0,0,0,0,0); } else if (pciFindDevice ((PCI_ID_NCR810 & 0xFFFF), (PCI_ID_NCR810 >> 16) & 0xFFFF, 0, &pciBusNo, &pciDevNo, &pciFuncNo) != ERROR) { SCSI_DEBUG_MSG ("Found an NCR810 Controller\n", 0,0,0,0,0,0); } else { logMsg ("SCSI controller not found\n", 0, 0, 0, 0, 0, 0); return (ERROR); } pciConfigInWord (pciBusNo, /* bus number */ pciDevNo, /* device number */ pciFuncNo, /* function number */ PCI_CFG_DEVICE_ID, /* offset into the configuration space */ &devType); /* data read from the offset */ #ifndef SCSI_WIDE_ENABLE devType = NCR810_DEVICE_ID; #endif ncr810DelayCount = NCR810_DELAY_MULT; /* Create the SCSI controller */ if ((pSysScsiCtrl = (SCSI_CTRL *) ncr810CtrlCreate ( (UINT8 *) SCSI_BASE_ADRS, (UINT) NCR810_40MHZ, devType )) == NULL) { SCSI_DEBUG_MSG ("ERROR: Can not create SCSI controller\n", 0, 0, 0, 0, 0, 0); return (ERROR); } /* connect the SCSI controller's interrupt service routine */ if (intConnect (INUM_TO_IVEC (SCSI_INT_VEC), ncr810Intr, (int) pSysScsiCtrl) == ERROR) { SCSI_DEBUG_MSG ("ERROR: Can not connect the SCSI controller's ISR\n", 0, 0, 0, 0, 0, 0); return (ERROR); } /* Enable SCSI interrupts */ intEnable (SCSI_INT_LVL); /* initialise SCSI controller with default parameters (user tuneable) */ if (ncr810CtrlInit ((NCR_810_SCSI_CTRL *)pSysScsiCtrl,SCSI_DEF_CTRL_BUS_ID) == ERROR) { SCSI_DEBUG_MSG ("ERROR: Can not initialize the SCSI controller\n", 0, 0, 0, 0, 0, 0); return (ERROR); } #if (USER_D_CACHE_MODE & CACHE_SNOOP_ENABLE) scsiCacheSnoopEnable ((SCSI_CTRL *) pSysScsiCtrl); #else scsiCacheSnoopDisable ((SCSI_CTRL *) pSysScsiCtrl); #endif /* * Set the good value in the registers of the SIOP coupled * with the hardware implementation */ if (ncr810SetHwRegister ((NCR_810_SCSI_CTRL *)pSysScsiCtrl, &hwRegs) == ERROR) { SCSI_DEBUG_MSG ("ERROR: ncr810SetHwRegister failed\n", 0, 0, 0, 0, 0, 0); return(ERROR); } return (OK); } sysScsi.c: ======================================================================== STATUS sysScsiConfig (void) { /* configure a SCSI hard disk at scsi busId = 3, LUN = 0 */ scsiDebug = TRUE; /* enable SCSI debugging output */ if ((pSpd20 = scsiPhysDevCreate (pSysScsiCtrl, /* ptr to SCSI controller info */ 3, /* device's SCSI bus ID */ 0, /* device's logical unit number */ 0, /* length of REQUEST SENSE data dev returns */ NONE, /* type of SCSI device */ 0, /* whether medium is removable */ 0, /* number of blocks on device */ 0) /* size of a block in bytes */ ) == (SCSI_PHYS_DEV *) NULL) { printErr ("usrScsiConfig: scsiPhysDevCreate failed.\n", 0, 0, 0, 0, 0, 0); } else { /* create block devices */ if ((pSbd0 = scsiBlkDevCreate (pSpd20, /* pScsiPhysDev, ptr to SCSI physical device info */ 32*2048, /* numBlocks, 32*2048=32Meg number of blocks in block device */ 0 /* blockOffset address of first block in volume */ )) == NULL) { return (ERROR); } if ((dosFsDevInit("/sd3/", pSbd0, NULL) == NULL) ) { return (ERROR); } } return (OK); } Press any key to stop auto-boot... 0 auto-booting... boot device : dc processor number : 0 host name : vxhost file name : /opt/wind_ppc/wip/vxWorks.st inet on ethernet (e) : 9.34.242.108:ffffff00 inet on backplane (b): 9.242.108.1:ffffff00 host inet (h) : 9.34.200.23 gateway inet (g) : 9.34.242.253 user (u) : vxworks ftp password (pw) : flags (f) : 0x8 target name (tn) : scsi_test Attaching network interface dc0... done. Attaching network interface lo0... done. Loading... 1117140 Starting at 0x100000... 0x1fffe08 (tRootTask): Initializing SCSI controller 0x1fffe08 (tRootTask): Found a SYMBIOS 53C875 Controller Attaching network interface dc0... done. Initializing backplane net with anchor at 0x4100... done. Backplane anchor at 0x4100... Attaching network interface sm0... done. Attaching network interface lo0... done. Adding 3478 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 MVME2600 - MPC 604e. Processor #0. Memory Size: 0x2000000. BSP version 1.1/4.04-BBD WDB: Ready. -> sysScsiConfig 0x17b85d0 (tShell): scsiReqSense: 0x17b85d0 (tShell): scsi2Transact: 0x17d82b8 (tScsiTask): scsiMgrRequestProc: client request: 0 0x17d82b8 (tScsiTask): scsiMgrActivateRequest: thread 0x01ffde98 0x17d82b8 (tScsiTask): ncr810ThreadActivate: thread 0x01ffde98: activating 0x17d82b8 (tScsiTask): ncr810XferParamsCvt: requested: offset = 255, period = 25 0x17d82b8 (tScsiTask): ncr810XferParamsCvt: converted to: offset = 8, period = 25 xferParams = 8, clockDivide = 19 0x17d82b8 (tScsiTask): ncr810XferParamsCvt: converted to: offset = 0, period = 0 xferParams = 0, clockDivide = 3 0x17d82b8 (tScsiTask): ncr810ThreadStateSet: thread 0x01ffde98: 1 -> 6 0x17d82b8 (tScsiTask): scsiMgrTimeoutProc: thread 0x01ffde98 (state = 6) timed out 0x17d82b8 (tScsiTask): ncr810ThreadAbort: thread 0x01ffde98 (state = 6) aborting 0x17d82b8 (tScsiTask): ncr810ThreadStateSet: thread 0x01ffde98: 6 -> 9 At this point the target shell hangs. A Control-C will restart the shell. Any help would be appreciated. vxworks tornado From daemon@csg.lbl.gov Fri Aug 21 04:00:11 1998 From: daemon@csg.lbl.gov Date: Fri Aug 21 04:00:15 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Aug 21 04:00:08 PDT 1998 Subject: Re: Radstone FDDI interface Subject: Re: Problem with VxWorks and Ethernet traffic ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Radstone FDDI interface Date: Wed, 19 Aug 1998 10:39:56 GMT From: agow@my-dejanews.com Organization: Deja News - The Leader in Internet Discussion Message-ID: <6rea1s$qb$1@nnrp1.dejanews.com> References: <6qpudq$m6d$1@nnrp1.dejanews.com> <6r7oqa$jnq$1@ligarius.ultra.net> In article <6r7oqa$jnq$1@ligarius.ultra.net>, "Larry Stefani" wrote: > > gesnsi@embraer.com.br wrote in message <6qpudq$m6d$1@nnrp1.dejanews.com>... > >Hi, > > > >I am using a Radstone's FDDI board with VxWorks and getting the following > >error: > > > >rpf (unit number 0): > > Flags: (0x63) UP BROADCAST ARP RUNNING > > Internet address: 9.37.7.2 > > Broadcast address: 9.255.255.255 > > Netmask 0xff000000 Subnetmask 0xff000000 > > Ethernet address is 00:80:8e:80:02:05 > > Metric is 0 > > Maximum Transfer unit size is 4352 > > "n" packets received; 102 packetes sent > > "n*8.1" input errors; 0 output errors > > 1 collisions > > > >Why is occuring these input errors? > > Since collisions are not possible in FDDI's ring protocol, I have to assume > that the FDDI driver was ported from an existing Ethernet driver and is > improperly counting its error statistics. Give that, I'd be suspicious of > all of the counters. > > -Larry > > Gentlemen For your information, the manual provided with the Radstone FDDI driver describes the use of the collision field as follows - There can never be any collisions on an FDDI network (because it is a token ring passing network), so the collision counter records the number of times that the FDDI station has connected to the ring. - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with VxWorks and Ethernet traffic Date: Thu, 20 Aug 1998 20:36:18 GMT From: sheppard@magma.ca (Tom Sheppard) Organization: . Message-ID: References: <35D9F935.8C0E440D@cs.umass.edu> <6RrC1.74$rb1.73942@news.mci2000.com> <35DAF830.75F@castlenetworks.com> >> > 0x32f5a08 (tNetTask): panic: netJobAdd: ring >> >buffer overflow I'm curious as to why there is a ring buffer overflow even reported. Are you using some protocol other than TCP/IP that can't afford to lose packets? Is the traffic on the link so critical to a real-time application that it can't be lost? TCP doesn't care if you lose packets, it just retransmits them. Every design I've seen silently discards the packets that can't be put into the ring because the main processor is to busy to process them. Then TCP's flow control kicks in and slows down. As long as you allow a panic to occur, it will. Unless you've got a damn fast processor customized for Ethernet speeds, you'll eventually get an overflow when: - there's a broadcast storm - someone tries out a ping program with a 0 delay - a failed adapter card starts babbling - lots of stations use multicast and you haven't enabled h/w filtering - someone is talking to you using UDP and he talks real fast Over standard Ethernet you can get a new packet about every 67 microseconds (if I remember correctly -- I'm getting kinda rusty). Incidentally, those broadcast storms do amazing things. I've seen HP and Sun workstations lock up solid (even the clock didn't advance). On a Mac, you could see the menus being drawn and filled in over the course of several seconds. When the storm subsided, the machines returned to normal. None of them panicked. ...Tom --------------------------- End of New-News digest ********************** From shoutchen@SYSTRAN.com Fri Aug 21 07:23:44 1998 From: Steve Houtchen Date: Fri Aug 21 07:23:48 PDT 1998 Subject: VxWorks or Tornado 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_01BDCD0E.8F855A90 Content-Type: text/plain VxWorks or Tornado Anybody know how to read the state of a binary semaphore or count of a counting semaphore? I looked through the system header files, but did not find the structures . I only have the SEM_ID to use. Steve Houtchen Senior Software Engineer Systran Corporation 4126 Linden Avenue, Dayton, OH 45432 (937)252-5601 ext 327 shoutchen@systran.com ------ =_NextPart_001_01BDCD0E.8F855A90 Content-Type: text/html VxWorks or Tornado

 VxWorks or Tornado

 

Anybody know how to read the state of a binary semaphore or count
of a counting semaphore?  I looked through the system header files,
but did not find the structures .  I only have the SEM_ID to use.


Steve Houtchen
Senior Software Engineer
Systran Corporation
4126 Linden Avenue, Dayton, OH 45432
(937)252-5601 ext 327
shoutchen@systran.com


------ =_NextPart_001_01BDCD0E.8F855A90-- From kb+@andrew.cmu.edu Fri Aug 21 10:39:42 1998 From: Kevin Bradley Date: Fri Aug 21 10:39:45 PDT 1998 Subject: Re: VxWorks or Tornado Excerpts from mail: 21-Aug-98 VxWorks or Tornado by the U. G. Exploder@lbl.g > Anybody know how to read the state of a binary semaphore or count > of a counting semaphore? I looked through the system header files, > but did not find the structures . I only have the SEM_ID to use. There was a discussion about this a while back. The general answer is, "illegally use the semaphore structure definitions". WindRiver isn't providing access to the internals of semaphores through functions, but the structures live in . From this file: ---------------------------------------------------------------- typedef struct semaphore /* SEMAPHORE */ { OBJ_CORE objCore; /* 0x00: object management */ UINT8 semType; /* 0x04: semaphore type */ UINT8 options; /* 0x05: semaphore options */ UINT16 recurse; /* 0x06: semaphore recursive take count */ Q_HEAD qHead; /* 0x08: blocked task queue head */ union { UINT count; /* 0x18: current state */ struct windTcb *owner; /* 0x18: current state */ } state; } SEMAPHORE; #define semCount state.count #define semOwner state.owner ---------------------------------------------------------------- SEM_ID is a pointer to this structure (see ), so you can access e.g. the count with: /* Reset the count of the counting semaphore */ if(my_sem->semType == SEM_TYPE_COUNTING) my_sem->semCount = 0; Now, what I don't know is, if you've a binary semaphore, which of these is set to indicate "taken" or "not taken". Possibly the "semOwner" field. WARNING: since these are defined in "private", it's likely that these migrate around from version to version, and therefore your code will likely be broken on OS upgrades. The API remains the same, but the internal players are different. -- Kevin From macmac@bigfoot.com Fri Aug 21 10:54:47 1998 From: mac Date: Fri Aug 21 10:54:50 PDT 1998 unsubscribe vxworks bmg@pacbell.net From macmac@bigfoot.com Fri Aug 21 10:55:08 1998 From: mac Date: Fri Aug 21 10:55:11 PDT 1998 unsubscribe vxworks macmac@bigfoot.com From froeber@bbn.com Fri Aug 21 11:32:49 1998 From: Fred Roeber Date: Fri Aug 21 11:32:52 PDT 1998 Subject: Re: VxWorks or Tornado Steve Houtchen asked: > VxWorks or Tornado > > Anybody know how to read the state of a binary semaphore or count > of a counting semaphore? I looked through the system header files, > but did not find the structures . I only have the SEM_ID to use. From h/semLib.h comes: typedef struct semaphore *SEM_ID; The semaphore type is defined in h/private/semLibP.h. Has all the info you want. Fred -- | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | ../semLib.h: From daemon@csg.lbl.gov Sun Aug 23 04:00:12 1998 From: daemon@csg.lbl.gov Date: Sun Aug 23 04:00:15 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Aug 23 04:00:09 PDT 1998 Subject: Re: MVME162-263 SBC ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: MVME162-263 SBC Date: Sat, 22 Aug 1998 20:22:51 GMT From: Bill Dennen Organization: Motorola Computer Group Message-ID: <35DF289B.54585A7C@mcg.mot.com> References: <35D98138.FB7C396D@wibbleiona.com> Sender: usenet@cradac.mclean.mcd.mot.com (UseNet) Remove the jumper at J1 to make this 162LX not a system controller. Documentation _can_ be found at http://www.mcg.mot.com >@< Oisin Hurley wrote: > > I'm lucky enough to have an MVME162-263 card running VxWorks 5.3.1, but > I'm unlucky enough to not know the jumper settings so I cannot prevent > the > card from deciding it is the system controller :( If anyone has a > schematic > or description of the jumpers for this card, or some software method of > achieving the > same end, then I would be delighted to hear from them. > > BTW, the Motorola pages only list MVME162 700/800 series cards, which > have a > different layout. The card I have was manufactured in 1993. And I don't > have > the documentation that came with the card. > > tia > --omh - -- >@< Bill Dennen bowtie@mclean.mcd.mot.com | We has met the enemy MidAtlantic District SE 703-714-0707 | and they is us Motorola Computer Group 703-714-0714 FAX | Pogo --------------------------- End of New-News digest ********************** From daemon@csg.lbl.gov Mon Aug 24 04:00:11 1998 From: daemon@csg.lbl.gov Date: Mon Aug 24 04:00:16 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Aug 24 04:00:08 PDT 1998 Subject: Re: Breakpoint Marker (update) ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Breakpoint Marker (update) Date: Sat, 22 Aug 1998 12:11:26 +0900 From: "Seong-Beom Kim" Organization: Korea Telecom Message-ID: <6rld0n$1dg$1@news.kornet.nm.kr> References: <35D9F08C.650ECD85@americasm01.nt.com> <35DC2940.D6E2BC88@americasm01.nt.com> Because I have similar problem(win95 host), I have installed cumulative patch. Then, the problem is solved. Install cumulative patch then you will get it. --------------------------- End of New-News digest ********************** From jtmark@most.fw.hac.com Mon Aug 24 05:35:00 1998 From: "Jimmie T Marks" Date: Mon Aug 24 05:35:04 PDT 1998 Subject: TORNADO ATA driver. I am trying to using VxWorks 5.3.1 ATA driver (ataDrv) on a Pentium card (cPCI) with a IDE/ATA flash memory card. The call to ataDrv is failing. I have checked I/O addresses and ATA RESOURCES structure and tried many different things. I don't have any idea of what could be going on. The BIOS is detecting and initializing the flash memory card prior to VxWorks. Any suggestions or thoughts? jtmark@most.fw.hac.com From mdcb@ing.iac.es Mon Aug 24 07:37:58 1998 From: Matthieu Bec Date: Mon Aug 24 07:38:05 PDT 1998 Subject: reboot vs. power cycle vxworks Hi, I'm using a standard VME crate/ 68k motorolla CPU / 8Meg off-board memory and DR11 board. Question: does a power-cycle of the whole crate 'does more' than a simple reboot from the shell ? Is there any definite answer or this depends on the boards I got in my rack ? Cheers, -- Matthieu Bec Isaac Newton Group tel : +34 922 425427 Apartado de Correos 321 fax : +34 922 425401 38780 Santa Cruz de La Palma email: mdcb@ing.iac.es Canary Islands, Spain From scheurer@terra1.tz.rus.uni-stuttgart.de Mon Aug 24 14:03:26 1998 From: "F. Scheurer" Date: Mon Aug 24 14:03:30 PDT 1998 Subject: MVME2700 setup for VxWorks 5.3.1 Hi VxWorkers, i got a problem to setup my kernel for a MVME2700. I start new building VxWorks-Systems; in the past i developed modules, but not a whole system and we passed from 5.1 to 5.3. First i added some options. The BUILD through WindCFG broke (ccppc) because some Defines out of configAll.h could not be found. I added them into config.h after the include of configdb.h. - Is this the right way to set it up? I don't understand the documentation at this point. After that, compilation succeeded but the BUILD broke once again (ldppc). - What's the reason? - Are there some options missing? - Is this the right position in target/config//config.h to configure more detailed things? Any comments appreciated! Best regards Friedrich Scheurer P.S: Here are some details: The following options were added to the default configuration through WindCFG: INCLUDE_CONFIGURATION_5_2 INCLUDE_DEBUG INCLUDE_DOSFS INCLUDE_INSTRUMENTATION INCLUDE_LOADER INCLUDE_NET_SYM_TBL INCLUDE_NFS INCLUDE_NFS_SERVER INCLUDE_PING INCLUDE_PROTECT_TEXT INCLUDE_PROTECT_VEC_TABLE INCLUDE_RAMDRV INCLUDE_RAWFS INCLUDE_RDB INCLUDE_RLOGIN INCLUDE_RPC INCLUDE_SHELL INCLUDE_SHOW_ROUTINES INCLUDE_SLIP INCLUDE_SM_NET INCLUDE_SM_OBJ INCLUDE_SM_SEQ_ADDR INCLUDE_SPY INCLUDE_STANDALONE_SYM_TBL INCLUDE_STARTUP_SCRIPT INCLUDE_STAT_SYM_TBL INCLUDE_SYM_TBL INCLUDE_SYM_TBL_SYNC INCLUDE_TELNET INCLUDE_TIMESTAMP INCLUDE_UNLOADER "Configure" creates configdb.h. "Build" results in the following : 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\"" /usr/local/wind/target/config/all/usrConfig.c -o usrConfig.o /usr/local/wind/target/src/config/../../src/config/usrNetwork.c: In function `usrNetInit': In file included from /usr/local/wind/target/src/config/usrExtra.c:199, from /usr/local/wind/target/config/all/usrConfig.c:97: /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:1400: `NFS_USER_ID' undeclared (first use this function) /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:1400: (Each undeclared identifier is reported only once /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:1400: for each function it appears in.) /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:1400: `NFS_GROUP_ID' undeclared (first use this function) /usr/local/wind/target/src/config/../../src/config/usrNetwork.c: In function `usrSlipInit': /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:2106: `SLIP_TTY' undeclared (first use this function) make: *** [usrConfig.o] Error 1 Press Return to Quit From wind/target/config/all/configAll.h copied and added to mv2700.ZFS_Default/config.h ... .... /* The following has been added by WindCfg */ #include "configdb.h" /* */ /* Aus configAll.h hierher kopiert!!!! */ #ifdef INCLUDE_NFS /* Default NFS parameters - constants may be changed here, variables * may be changed in usrConfig.c at the point where NFS is included. */ #define NFS_USER_ID 2001 /* dummy nfs user id */ #define NFS_GROUP_ID 100 /* dummy nfs user group id */ #endif /* INCLUDE_NFS */ #ifdef INCLUDE_SLIP #define SLIP_TTY 1 /* serial line IP channel */ #endif /* INCLUDE_SLIP */ new "Build": ... ldppc -X -N -e _sysInit -Ttext 00100000 \ -o vxWorks dataSegPad.o vxWorks.tmp ctdt.o vxWorks.tmp: In function `usrSmObjInit': vxWorks.tmp(.text+0x7694): undefined reference to `smObjLibInit' vxWorks.tmp(.text+0x7840): undefined reference to `smObjSetup' vxWorks.tmp(.text+0x789a): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x789e): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x78bc): undefined reference to `smObjInit' vxWorks.tmp(.text+0x78d6): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x78d8): undefined reference to `smObjAttach' vxWorks.tmp: In function `usrRoot': vxWorks.tmp(.text+0x8a5a): undefined reference to `standTblSize' vxWorks.tmp(.text+0x8a62): undefined reference to `standTblSize' vxWorks.tmp(.text+0x8a72): undefined reference to `standTblSize' vxWorks.tmp(.text+0x8a86): undefined reference to `standTblSize' vxWorks.tmp(.text+0x8a8a): undefined reference to `standTbl' vxWorks.tmp(.text+0x8a8e): undefined reference to `standTbl' vxWorks.tmp(.text+0x8a9e): undefined reference to `standTblSize' vxWorks.tmp: In function `dcStartOutput': vxWorks.tmp(.text+0x3ddd4): undefined reference to `m_freem' vxWorks.tmp: In function `slipDelete': vxWorks.tmp(.text+0x54f20): undefined reference to `m_freem' vxWorks.tmp: In function `sloutput': vxWorks.tmp(.text+0x550e0): undefined reference to `m_freem' vxWorks.tmp(.text+0x55108): undefined reference to `m_freem' vxWorks.tmp(.text+0x551d4): undefined reference to `m_freem' vxWorks.tmp: In function `slipWrtTask': vxWorks.tmp(.text+0x5549a): undefined reference to `mbstat' vxWorks.tmp(.text+0x5549e): undefined reference to `mbstat' .... and so on. -- +-------------------------------------------------+ | | | Dipl.-Ing. F. Scheurer | | | | c/o Zentrum Fertigungstechnik Stuttgart | | Arbeitsbereich Steuerungstechnik | | Nobelstrasse 15 | | D-70569 Stuttgart | | | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.rus.uni-stuttgart.de | | zfssrr@zfssun6.tz.rus.uni-stuttgart.de | +-------------------------------------------------+ | http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | | | privat: | | Telefon +49 7161 686362 | | Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | | | +-------------------------------------------------+ From carl@themis.com Mon Aug 24 14:31:03 1998 From: "Carl C. Chesbrough" Date: Mon Aug 24 14:31:07 PDT 1998 Subject: Re: reboot vs. power cycle Matthieu, By doing a "reboot" VxWorks goes through a "warm restart" vector and does not execute the same code as a power-cycle. If you do a "reboot 2" command, then VxWorks goes through a "cold restart" vector and does execute the same code as a power-cycle. One thing that you may need to be aware of is that the state of the hardware will not be the same on a "cold restart" vs. a power-cycle. HTH, -Carl. ******************************************************************** * * * Carl C. Chesbrough Telephone: (510) 252-0870 x 126 * * Themis Computer Facsimile: (510) 490-5529 * * 3185 Laurelview Court E-Mail : carl@themis.com * * Fremont, California 94538 * * * ******************************************************************** > vxworks > > Hi, > I'm using a standard VME crate/ 68k motorolla CPU / 8Meg off-board > memory and DR11 board. > > Question: > does a power-cycle of the whole crate 'does more' than a simple reboot > from the shell ? > Is there any definite answer or this depends on the boards I got in my > rack ? > > Cheers, > > -- > Matthieu Bec Isaac Newton Group > tel : +34 922 425427 Apartado de Correos 321 > fax : +34 922 425401 38780 Santa Cruz de La Palma > email: mdcb@ing.iac.es Canary Islands, Spain From cgrames@mdc.com Mon Aug 24 15:30:52 1998 From: Charlie Grames Date: Mon Aug 24 15:30:56 PDT 1998 Subject: Bootin a Large Number of Targets Simultaneously VxWorks 5.3.1 MVME2604 1.1/4 One of our programs is attempting to locate 8 MVME2604 boards into a single chassis and boot via the Internet Information Server (IIS) FTP server included with Windows NT Server. Only the first board is booting via Ethernet; the remaining seven are booting via the backplane. None of the boards except the primary complete their boot in this configuration (most of the remaining boards load the vxWorks image, but none successfully load vxWorks.sym), but if I remove 4 boards from the system, the remaining 4 boards boot fine. Any ideas? I thought it might have something to do with timeouts, since so many boards are competing for the same resources at once, but I'm not sure where to look. Any insight in this area would be appreciated. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From cgrames@mdc.com Mon Aug 24 15:40:47 1998 From: Charlie Grames Date: Mon Aug 24 15:40:51 PDT 1998 Subject: MVME2700 setup for VxWorks 5.3.1 -Reply Friedrich, I don't know if this is the problem or not, but I noticed that you included INCLUDE_PROTECT_TEXT and INCLUDE_PROTECT_VEC_TABLE. These are valid only if you are using the VxVMI add-on product. Last I knew, the PowerPC was not one of the supported targets for VxVMI. You may want to get rid of these options from your configuration and see if some of your problems clear up. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com >>> the vxWorks Users Group Exploder 08/24/98 04:03pm >>> Submitted-by scheurer@terra1.tz.rus.uni-stuttgart.de Mon Aug 24 14:03:26 1998 Submitted-by: "F. Scheurer" Hi VxWorkers, i got a problem to setup my kernel for a MVME2700. I start new building VxWorks-Systems; in the past i developed modules, but not a whole system and we passed from 5.1 to 5.3. First i added some options. The BUILD through WindCFG broke (ccppc) because some Defines out of configAll.h could not be found. I added them into config.h after the include of configdb.h. - Is this the right way to set it up? I don't understand the documentation at this point. After that, compilation succeeded but the BUILD broke once again (ldppc). - What's the reason? - Are there some options missing? - Is this the right position in target/config//config.h to configure more detailed things? Any comments appreciated! Best regards Friedrich Scheurer P.S: Here are some details: The following options were added to the default configuration through WindCFG: INCLUDE_CONFIGURATION_5_2 INCLUDE_DEBUG INCLUDE_DOSFS INCLUDE_INSTRUMENTATION INCLUDE_LOADER INCLUDE_NET_SYM_TBL INCLUDE_NFS INCLUDE_NFS_SERVER INCLUDE_PING INCLUDE_PROTECT_TEXT INCLUDE_PROTECT_VEC_TABLE INCLUDE_RAMDRV INCLUDE_RAWFS INCLUDE_RDB INCLUDE_RLOGIN INCLUDE_RPC INCLUDE_SHELL INCLUDE_SHOW_ROUTINES INCLUDE_SLIP INCLUDE_SM_NET INCLUDE_SM_OBJ INCLUDE_SM_SEQ_ADDR INCLUDE_SPY INCLUDE_STANDALONE_SYM_TBL INCLUDE_STARTUP_SCRIPT INCLUDE_STAT_SYM_TBL INCLUDE_SYM_TBL INCLUDE_SYM_TBL_SYNC INCLUDE_TELNET INCLUDE_TIMESTAMP INCLUDE_UNLOADER "Configure" creates configdb.h. "Build" results in the following : 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\"" /usr/local/wind/target/config/all/usrConfig.c -o usrConfig.o /usr/local/wind/target/src/config/../../src/config/usrNetwork.c: In function `usrNetInit': In file included from /usr/local/wind/target/src/config/usrExtra.c:199, from /usr/local/wind/target/config/all/usrConfig.c:97: /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:1400: `NFS_USER_ID' undeclared (first use this function) /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:1400: (Each undeclared identifier is reported only once /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:1400: for each function it appears in.) /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:1400: `NFS_GROUP_ID' undeclared (first use this function) /usr/local/wind/target/src/config/../../src/config/usrNetwork.c: In function `usrSlipInit': /usr/local/wind/target/src/config/../../src/config/usrNetwork.c:2106: `SLIP_TTY' undeclared (first use this function) make: *** [usrConfig.o] Error 1 Press Return to Quit From wind/target/config/all/configAll.h copied and added to mv2700.ZFS_Default/config.h ... .... /* The following has been added by WindCfg */ #include "configdb.h" /* */ /* Aus configAll.h hierher kopiert!!!! */ #ifdef INCLUDE_NFS /* Default NFS parameters - constants may be changed here, variables * may be changed in usrConfig.c at the point where NFS is included. */ #define NFS_USER_ID 2001 /* dummy nfs user id */ #define NFS_GROUP_ID 100 /* dummy nfs user group id */ #endif /* INCLUDE_NFS */ #ifdef INCLUDE_SLIP #define SLIP_TTY 1 /* serial line IP channel */ #endif /* INCLUDE_SLIP */ new "Build": ... ldppc -X -N -e _sysInit -Ttext 00100000 \ -o vxWorks dataSegPad.o vxWorks.tmp ctdt.o vxWorks.tmp: In function `usrSmObjInit': vxWorks.tmp(.text+0x7694): undefined reference to `smObjLibInit' vxWorks.tmp(.text+0x7840): undefined reference to `smObjSetup' vxWorks.tmp(.text+0x789a): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x789e): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x78bc): undefined reference to `smObjInit' vxWorks.tmp(.text+0x78d6): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x78d8): undefined reference to `smObjAttach' vxWorks.tmp: In function `usrRoot': vxWorks.tmp(.text+0x8a5a): undefined reference to `standTblSize' vxWorks.tmp(.text+0x8a62): undefined reference to `standTblSize' vxWorks.tmp(.text+0x8a72): undefined reference to `standTblSize' vxWorks.tmp(.text+0x8a86): undefined reference to `standTblSize' vxWorks.tmp(.text+0x8a8a): undefined reference to `standTbl' vxWorks.tmp(.text+0x8a8e): undefined reference to `standTbl' vxWorks.tmp(.text+0x8a9e): undefined reference to `standTblSize' vxWorks.tmp: In function `dcStartOutput': vxWorks.tmp(.text+0x3ddd4): undefined reference to `m_freem' vxWorks.tmp: In function `slipDelete': vxWorks.tmp(.text+0x54f20): undefined reference to `m_freem' vxWorks.tmp: In function `sloutput': vxWorks.tmp(.text+0x550e0): undefined reference to `m_freem' vxWorks.tmp(.text+0x55108): undefined reference to `m_freem' vxWorks.tmp(.text+0x551d4): undefined reference to `m_freem' vxWorks.tmp: In function `slipWrtTask': vxWorks.tmp(.text+0x5549a): undefined reference to `mbstat' vxWorks.tmp(.text+0x5549e): undefined reference to `mbstat' .... and so on. -- +-------------------------------------------------+ | | | Dipl.-Ing. F. Scheurer | | | | c/o Zentrum Fertigungstechnik Stuttgart | | Arbeitsbereich Steuerungstechnik | | Nobelstrasse 15 | | D-70569 Stuttgart | | | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.rus.uni-stuttgart.de | | zfssrr@zfssun6.tz.rus.uni-stuttgart.de | +-------------------------------------------------+ | http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | | | privat: | | Telefon +49 7161 686362 | | Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | | | +-------------------------------------------------+ From dp17196@elbit.co.il Tue Aug 25 02:17:51 1998 From: Dudi Kenig Date: Tue Aug 25 02:17:55 PDT 1998 Subject: memory fragmentation using open() , close() Hi, Has anyone know a workaround for the memory fragmentation while using open and close for dosFs files this is done under Vxworks 5.1.1 when doing the following fd =open ("my_file,O_RDWR,0755) close (fd) we lose 16 bytes each time we do this. Thanks Dudi From daemon@csg.lbl.gov Tue Aug 25 04:00:11 1998 From: daemon@csg.lbl.gov Date: Tue Aug 25 04:00:15 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue Aug 25 04:00:08 PDT 1998 Subject: POSIX Semaphores Subject: Re: Software Design Info : VxWorks/C++? Subject: Re: Loadlin equivalent for VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: POSIX Semaphores Date: Tue, 18 Aug 1998 19:41:57 GMT From: Ken Ostaszewski Organization: Boeing, Inc. Message-ID: <35D9D904.36E7448C@boeing.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Hi, I sent a message a week ago, and have not heard back from anyone. Just wondering if anyone actually received my message using POSIX Semaphores. Specifically, has anyone else found that the sem_getvalue function returns an incorrect value when run from an interrupt handler? Ken --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Software Design Info : VxWorks/C++? Date: Mon, 24 Aug 1998 20:29:47 -0700 From: Vicente Mendez Organization: QUALCOMM, Incorporated; San Diego, CA, USA Message-ID: <35E22FAB.ABCE0098@qualcomm.com> References: There is a book out there called Real Time UML by Brouce Power Douglas. It talks about UML real time environments, not the tools, but it could be of help. Vicente Mendez Mittal Monani wrote: > Hi Folks, > > We're trying to develop real-time embedded software on VxWorks/C++ for a > communication application. Could you > point me to any books/journals/web resources on Real Time Object Oriented > software design and development? Any recommendation for modeling/CASE/config > mgmt tools? > > Thanks, > Mittal --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loadlin equivalent for VxWorks Date: Mon, 24 Aug 1998 21:08:40 -0700 From: Vicente Mendez Organization: QUALCOMM, Incorporated; San Diego, CA, USA Message-ID: <35E238C8.ACE1BA55@qualcomm.com> References: <6rh7hd$ol5$3@news.cegelecproj.co.uk> <6ri8fq$caa$1@nnrp1.dejanews.com> <6rjj6l$bl3$1@news.cegelecproj.co.uk> We have been able to get this card to work correctly. There is a DOS program that will caonfigure the card to a certain port/irq. You can also disable plug and play with it. I would check the 3com site for this utility. Vicente Mendez Kevin Taylor wrote: > > >If VxWorks doesn't have a driver for the LAN card, there's nothing > >you can do. > > Its an Etherlink III - a card that is setup fine on another PC > (normally running OS/2) ... > > > I don't think the problem is the plug & play, the > >BIOS on your PC will configure the card. > > Does Windows query the BIOS then for the parameters (not > really familiar with PnP) ? > > Kevin. --------------------------- End of New-News digest ********************** From jorgee@bpo.hp.com Tue Aug 25 05:22:50 1998 From: Jorge Estevez Date: Tue Aug 25 05:22:54 PDT 1998 Subject: Re: memory fragmentation using open() , close() the vxWorks Users Group Exploder wrote: > Submitted-by dp17196@elbit.co.il Tue Aug 25 02:17:51 1998 > Submitted-by: Dudi Kenig > > Hi, > > Has anyone know a workaround for the memory fragmentation while using open > and close for dosFs files > this is done under Vxworks 5.1.1 > > when doing the following > > fd =open ("my_file,O_RDWR,0755) > close (fd) > > we lose 16 bytes each time we do this. > > Thanks > > Dudi > The memory leak your are detecting is not because the open neither the close operation. It is due to the allocation done to store the string "my_file" that VxWorks does before call open function. If you use as filename parameter a char string variable previously allocated, you won't lose any byte. cheers, Jorge -- =========================================================== = Jorge Estevez = Hewlett-Packard Company = = Software Engineer = Barcelona Division (BCD) = =========================================================== = e-mail: jorgee@bpo.hp.com = Avda. Graells, 501 = = phone: +34 93 5822778 = 08190 S. Cugat (Barcelona) = = fax: +34 93 5822466 = SPAIN = =========================================================== From djnoble@west.raytheon.com Tue Aug 25 06:04:11 1998 From: djnoble@west.raytheon.com Date: Tue Aug 25 06:04:14 PDT 1998 Subject: VxWorks Touch Screen Drivers Our project requires the use of a touch screen. We are using Tornado 1.0.1 and VxWorks 5.3. We are using DY-4 SVME 177 SBC and SVME 783 Graphics card. Our intent is to interface the touch through a serial channel on the SBC or the graphics card. We are looking for someone who done this before and could give us some information on the approach they took on integrating touch screens. We have not found a touch screen that supports the DY4 system running VxWorks. Does anyone know of touch screens which have available the drivers for VxWorks(an integrated package of touch device and drivers). In the absence of the above, are drivers for processing a generic touch input and doing calibration available commercially? If we cannot find a terminal with drivers our current plan is to purchase a low cost COTS touch screen (perhaps Microtouch) which outputs touch information in a known format. We believe this type of device returns ASCII signals over a RS-232 interface. Any information or thoughts would be appreciated. Dennis Noble Raytheon Consulting Group djnoble@hti.com From jbhawth@aptec.com Tue Aug 25 06:47:03 1998 From: "James B. Hawthorne" Date: Tue Aug 25 06:47:07 PDT 1998 Subject: RE: Bootin a Large Number of Targets Simultaneously On a program I worked a couple of years ago at Sandia National Labs, we were booting approximately 8 VXI chassis running VxWorks over the network and saw a similar behavior. Eventually, we ended up triggering individual chassis to boot from the GUI (located on a Sun workstation) by connecting the Console port to an RS-232 interface and sending Ctrl-X down the line to trigger a boot when the operator pressed a button. Additionally, the OS images had to be deleted between boots and copied down to the boot directories just prior to sending Ctrl-X. It wasn't pretty but it worked well so long as the operator didn't go crazy and press all of the boot buttons at the same time. Jim Hawthorne Applied Technology Associates Albuquerque, New Mexico From jeff.golowner@lmco.com Tue Aug 25 07:05:54 1998 From: "Golowner, Jeff" Date: Tue Aug 25 07:05:57 PDT 1998 Subject: VxWorks POSIX Semaphores Ken, There are some bugs with the VxWorks POSIX semaphores. I currently have a TSR logged with respect to the sem_close function. After issuing a sem_close, subsequent sem_xxx functions to the closed semaphore id should return an error with errno set to EINVAL, but the calls succeed. With respect to the sem_getvalue invoked in an ISR, I personally haven't tried this. How "incorrect" are the values you are getting? The value returned merely represents the value the semaphore had at one point during the call to sem_getvalue. Subsequent sem_wait or sem_post calls would obviously change the actual value. In general, the sem_getvalue function should only be used for debugging purposes. My guess is that in your ISR you only want to post the semaphore if the count is already 0 (thus trying to emulate a binary semaphore). A better way, is to post the semaphore regardless of the current value. The task waiting on the semaphore should follow the sem_wait (which blocks) with a loop of sem_trywaits (which won't block) until an error occurs with errno == EAGAIN (thus sucking up any extra posted semaphores). Jeff Golowner > -----Original Message----- > From: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] > Sent: Tuesday, August 25, 1998 7:00 AM > To: vxworks_users@csg.lbl.gov > Subject: comp.os.vxworks newsdigest > > Submitted-by daemon@csg.lbl.gov Tue Aug 25 04:00:11 1998 > Submitted-by: daemon@csg.lbl.gov > > Comp.Os.Vxworks Daily Digest Tue Aug 25 04:00:08 PDT 1998 > > Subject: POSIX Semaphores > Subject: Re: Software Design Info : VxWorks/C++? > Subject: Re: Loadlin equivalent for VxWorks > > ------------------------------------------------------- > > Newsgroups: comp.os.vxworks > Subject: POSIX Semaphores > Date: Tue, 18 Aug 1998 19:41:57 GMT > From: Ken Ostaszewski > Organization: Boeing, Inc. > Message-ID: <35D9D904.36E7448C@boeing.com> > Sender: nntp@news.boeing.com (Boeing NNTP News Access) > > Hi, > > I sent a message a week ago, and have not heard back from anyone. > > Just wondering if anyone actually received my message using POSIX > Semaphores. > > Specifically, has anyone else found that the sem_getvalue function > returns an incorrect value when run from an interrupt handler? > > Ken > From mea@mclean.sparta.com Tue Aug 25 07:08:16 1998 From: "Mike Anderson" Date: Tue Aug 25 07:08:20 PDT 1998 Subject: RE: Bootin a Large Number of Targets Simultaneously Greetings! > > > One of our programs is attempting to locate 8 MVME2604 boards > into a single chassis and boot > via the Internet Information Server (IIS) FTP server included > with Windows NT Server. Only the first > board is booting via Ethernet; the remaining seven are booting > via the backplane. None of the > boards except the primary complete their boot in this > configuration (most of the remaining boards > load the vxWorks image, but none successfully load vxWorks.sym), > but if I remove 4 boards from > the system, the remaining 4 boards boot fine. Any ideas? I > thought it might have something to do > with timeouts, since so many boards are competing for the same > resources at once, but I'm not > sure where to look. Any insight in this area would be appreciated. > The problem is actually more likely due to bus arbitration issues. Priority mode arbitration (the typical VME default) will start to fail with somewhere between master 3-5 (depending on bus loading). One possible solution to this would be to set round-robin arbitration on the SYSCON and then set the remaining boards at alternating aribration levels. So, the SYSCON at level 3, board 2 @ level 2, board 3 @ level 1, etc. and then repeat the pattern with board 5 and so on. This will eliminate arbitration as a problem. The next thing you can do is one of the things I teach in SPARTA's Intermediate real-time classes (taught to the public using VxWorks/ Tornado) wherein I set up CPU 0 as a boot host for the other boards using some form of local media such as a RAM disk. I've taught this using both FTP and TFTP (TFTP is typically quicker) servers running under VxWorks. In this approach, CPU 0 is defined as the boot host for the other processors. CPU 0 then configures and loads a RAM disk from the boot host with the appropriate VxWorks & .sym images. The other boards attach to the SM backplane and load their images from CPU 0. Startup scripts then are used to make sure that all of the routing is set up properly. This approach does two things simultaneously. First, it greatly reduces the traffic load on the boot host and the LAN. Second, it actually speeds the boot process dramatically, especially for systems with a large number of targets. Hmm... This sounds like a paper at the WRS Developer's conference ;-). If anyone would like to have me present this kind of thing (along with or in addition to, I could show how to add a new device to the boot ROMs and do "fail-soft" booting). Just drop me a note. If I get sufficient interest, I'll submit an abstract. HTH, ============================================================================ === Real-Time System Development, Integration, Training and Services Mike Anderson mea@mclean.sparta.com Chief Engineer http://www.mclean.sparta.com SPARTA, Inc. V: (703) 448-0210 F: (703) 893-5494 "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." The opinions are mine... no one else would want to claim them ;-). ============================================================================ === From n.rees@jach.hawaii.edu Tue Aug 25 11:31:36 1998 From: Nick Rees Date: Tue Aug 25 11:31:44 PDT 1998 Subject: Re: reboot vs. power cycle --Herd_of_Buffalo_104_000 Content-Type: TEXT/plain; charset=us-ascii Content-MD5: e+tBYx18jYL56YHcaJFZAw== Matthieu, Hi from the JAC. I am not sure how many answers you got to your question - I have only seen one and it doesn't mention the problem I find most often. When a crate is power cycled all boards in the crate are reset (naturally). When you use the VxWorks reboot command you only reboot that particular CPU board. Hence you can have a number of CPU's in a crate and can reboot them independently (we are currently doing this so we can work on two systems simultaneously, with minimal interference). This is irrespective of the parameter you pass to the reboot command - the reboot command only resets one board. To reset the entire VME crate you need access to the VMEbus SYSRESET line. I have attached a routine called syslocalSysReset that does this for mv167, hkbaja60, hkbaja47 and hknitro60 architectures. You attach this routine to the reboot sequence using rebootHookAdd, and you get the effect I think you want. I wish vendors would supply such a routine in their BSP since it is a common enough problem. Nick Rees Joint Astronomy Centre Ph: +1 (808) 961-3756 660 N. Aohoku Place Fax: +1 (808) 961-6516 Hilo, HI. 96720 Internet: n.rees@jach.hawaii.edu --Herd_of_Buffalo_104_000 Content-Type: TEXT/x-sun-c-file; name="syslocalSysReset.c"; charset=us-ascii; x-unix-mode=0644 Content-Description: syslocalSysReset.c Content-MD5: n0R45AeI38l1Bt1t/ucA5Q== /******************************************************************************* * syslocalSysReset * * who when what * --------- -------- -------------------------------------------------------- * N.Rees 25/04/97 first release for mv167 - based on a routine from * Gordon Uchenick * N.Rees 16/04/98 Added Heurikon Baja and Nitro architectures. * N.Rees 14/08/98 Cleaned up comments for ROE */ /******************************************************************************* * NAME * syslocalSysReset - Generate a VMEbus SYSRESET on various architectures. * * SYNOPSIS * syslocalSysReset( int startType ) * * DESCRIPTION * This routine is generates a VMEbus SYSRESET on various architectures. This * should reset every board in the VME crate, not just the current CPU board. * For example, this is required to ensure the b016 is reset properly on boot. * * To compile you must define the vxWorks target architecture for the current * target binary (e.g. currentlu mv167, hknitro60, hkbaja60 and hkbaja47 are * supported. * * For the mv167 it sets the mv167 watchdog timer, disables interrupts, * and then hogs the CPU. This ensures the watchdog timer goes off, thereby * resetting the CPU and also generating a SYSRESET on the VMEbus. The original * comes from Gordon Uchenick . * * For the Heurikon architectures, it just pokes the correct register in the * VIC064 chip. Thanks to Stephen B. Johnson, , * of Heurikon for the original code (i.e. vital one line). * * For other architectures, this is a null routine. * * IN VxWorks start type (not used) * * RETURN VALUES * the routine never returns. * * EXAMPLE * syslocalSysReset * * SEE ALSO * b016Drv * *------------------------------------------------------------------------------ */ #if defined(mv167) #include void syslocalSysReset() { volatile unsigned int * dogToCtlReg = (unsigned int *) 0xfff4004c; volatile unsigned int * dogCtlReg = (unsigned int *) 0xfff40060; volatile int scratch; /* Get the Watchdog Timeout Control Register. */ scratch = *dogToCtlReg; /* Wipe out the watch dog timeout, set it to 1 second, and put it back. */ scratch &= 0xfffff0ff; scratch |= 0x00000b00; *dogToCtlReg = scratch; /* Get the Watchdog timer control register. */ scratch = *dogCtlReg; /* Now enable the dog. */ scratch |= 0x000f0000; *dogCtlReg = scratch; /* Now sit here until the dog reboots the system. */ intLock(); for ( scratch=0; ; scratch++ ); } #elif defined(hkbaja60) || defined(hknitro60) || defined(hkbaja47) #if defined(hkbaja60) #define VIC_BASE_ADRS 0xff000000 #elif defined(hknitro60) #define VIC_BASE_ADRS 0xff000000 #elif defined(hkbaja47) #define VIC_BASE_ADRS 0x1f000000 #endif #include "drv/vme/vic068.h" void syslocalSysReset() { *VIC_SRR = 0xf0; } #else void syslocalSysReset() { } #endif --Herd_of_Buffalo_104_000-- From mmenge@dspt.com Tue Aug 25 13:50:53 1998 From: mmenge@dspt.com Date: Tue Aug 25 13:50:56 PDT 1998 Subject: Top Ten gripes with vxWorks and PowerPC 2604 Mark Menge@DSPT 08/25/98 04:49 PM 1. Sometimes you can't start tornado.You are forced to go into the registry and remove the Recent Files. 2. Strange spurious interrupt messages like "interrupt level 2, non-pending IBC" 3. After coding a bit and downloading, timex shows the code is suddenly a factor of 100 slower. After doing more coding everything speeds up again. 4. Ethernet errors: Excessive Collisions, Late Collisions, fist to 2604 collisions, CRC errors, etc. ->ifShow will print out some of these stats. 5. A continuous test of LoadModule shows it does not always clear its BSS segment! 6. Debugger sometimes shows wrong values in structures 7. Sometimes when Tornado asks to reload files modified outside of the editor, it crashes. 8. The source code editor sometimes does not show all the code. 9. Breakpoints do not work from WDB without first attaching to the task. Telneting in and using the b command directly confuses WDB. 10. The debugger and its entourage of exes will not recover from a rebooted target. I hate having to wait to close the debugger, then the target server, rerun them both, and set the breakpoints again. After target is rebooted, the debugger should mention that it was rebooted (when the user tries to invoke more debugger services) and ask if you want to reconnect and set the previously active breakpoints. This would be analogous to what every $99 windows development system does. Others that didn't make it: 1. No way to disable security with the FTP server 2. (GNU bug) make does not follow dependency rules for C files that include C files. 3. Sometimes the debugger cannot display variables giving a "Variable out of context" or some garbage. If I add the -g flag, and remove optimizations, then keep all the variables around! 4. The distribution of a CDs with hidden patches for bugs From mea@mclean.sparta.com Tue Aug 25 14:59:31 1998 From: "Mike Anderson" Date: Tue Aug 25 14:59:35 PDT 1998 Subject: RE: Top Ten gripes with vxWorks and PowerPC 2604 Greetings! > > > Mark Menge@DSPT > 08/25/98 04:49 PM > > 1. Sometimes you can't start tornado.You are forced to go into > the registry > and remove the Recent Files. > 2. Strange spurious interrupt messages like "interrupt level 2, > non-pending > IBC" > 3. After coding a bit and downloading, timex shows the code is suddenly a > factor of 100 slower. After doing more coding everything speeds up again. > 4. Ethernet errors: Excessive Collisions, Late Collisions, fist to 2604 > collisions, CRC errors, etc. ->ifShow will print out some of these stats. > 5. A continuous test of LoadModule shows it does not always clear its BSS > segment! > 6. Debugger sometimes shows wrong values in structures > 7. Sometimes when Tornado asks to reload files modified outside of the > editor, it crashes. > 8. The source code editor sometimes does not show all the code. > 9. Breakpoints do not work from WDB without first attaching to the task. > Telneting in and using the b command directly confuses WDB. > 10. The debugger and its entourage of exes will not recover from > a rebooted > target. I hate having to wait to close the debugger, then the target > server, rerun them both, and set the breakpoints again. After target is > rebooted, the debugger should mention that it was rebooted (when the user > tries to invoke more debugger services) and ask if you want to reconnect > and set the previously active breakpoints. This would be analogous to what > every $99 windows development system does. > > Others that didn't make it: > 1. No way to disable security with the FTP server > 2. (GNU bug) make does not follow dependency rules for C files > that include > C files. > 3. Sometimes the debugger cannot display variables giving a "Variable out > of context" or some garbage. If I add the -g flag, and remove > optimizations, then keep all the variables around! > 4. The distribution of a CDs with hidden patches for bugs > Most of these can be addressed in one word... Unix. We use the 2604 and the Power4E around here (as well as 860s) under a Solaris host and we don't see the majority of your gripes. Of course, we also tend to use either the native target shell or a hybrid Tornado/tShell environment. BTW, CrossWind works well if you know how to run it sans the launcher/Wdb. Use the INCLUDE_RDB instead. This is not to say that your gripes aren't legitimate. I'm sure that under WinDoze flavors, they most certainly are. We use Win 95/NT for a lot of our user GUI development including Windows device driver development. But, IMHO, 95/NT is just not a very mature platform for doing cross-targeting, embedded systems development. It would be eazy to point a finger at the GNU toolchain, but even internal to WinDoze itself, some of the Micro$oft tools don't work very well. For instance, why the VC++ Service Pack 3 linker cannot produce a working VxD is a mystery. You have to install SP3 and then copy the saved version of the SP2 linker into the bin directories! As for patches hidden on the CD, well that's just not the right thing to do. I understand that WRS is working hard and needs to be paid for their efforts, but having to pay to get a key to unlock a patch that's on my distribution CD for the ODL that I just paid for is a bit much. In any case, WRS should address the issues you've brought up. In the mean time, if you want a pretty stable development environment, trade your NT license in for Solaris. If you want your application to be damn-near rock solid, shut off Tornado and use the native target shell. You'll be glad you did. Regards, ============================================================================ === Real-Time System Development, Integration, Training and Services Mike Anderson mea@mclean.sparta.com Chief Engineer http://www.mclean.sparta.com SPARTA, Inc. V: (703) 448-0210 F: (703) 893-5494 "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." The opinions are mine... no one else would want to claim them ;-). ============================================================================ === From LZhou@coyotetech.com Tue Aug 25 15:36:20 1998 From: "Zhou, Lei" Date: Tue Aug 25 15:36:23 PDT 1998 Subject: RE: Bootin a Large Number of Targets Simultaneously Please do so and don't forget to e-mail me a copy of your writing. Thanks. -- Lei Zhou voice: (972) 889-5207 lzhou@coyotetech.com or lzhou@dhc.net -----Original Message----- From: vxwexplo@lbl.gov [SMTP:vxwexplo@lbl.gov] On Behalf Of Mike Anderson Posted At: Tuesday, August 25, 1998 9:10 AM Posted To: vxworks Conversation: Bootin a Large Number of Targets Simultaneously Subject: RE: Bootin a Large Number of Targets Simultaneously Greetings! > > > One of our programs is attempting to locate 8 MVME2604 boards > into a single chassis and boot > via the Internet Information Server (IIS) FTP server included > with Windows NT Server. Only the first > board is booting via Ethernet; the remaining seven are booting > via the backplane. None of the > boards except the primary complete their boot in this > configuration (most of the remaining boards > load the vxWorks image, but none successfully load vxWorks.sym), > but if I remove 4 boards from > the system, the remaining 4 boards boot fine. Any ideas? I > thought it might have something to do > with timeouts, since so many boards are competing for the same > resources at once, but I'm not > sure where to look. Any insight in this area would be appreciated. > The problem is actually more likely due to bus arbitration issues. Priority mode arbitration (the typical VME default) will start to fail with somewhere between master 3-5 (depending on bus loading). One possible solution to this would be to set round-robin arbitration on the SYSCON and then set the remaining boards at alternating aribration levels. So, the SYSCON at level 3, board 2 @ level 2, board 3 @ level 1, etc. and then repeat the pattern with board 5 and so on. This will eliminate arbitration as a problem. The next thing you can do is one of the things I teach in SPARTA's Intermediate real-time classes (taught to the public using VxWorks/ Tornado) wherein I set up CPU 0 as a boot host for the other boards using some form of local media such as a RAM disk. I've taught this using both FTP and TFTP (TFTP is typically quicker) servers running under VxWorks. In this approach, CPU 0 is defined as the boot host for the other processors. CPU 0 then configures and loads a RAM disk from the boot host with the appropriate VxWorks & .sym images. The other boards attach to the SM backplane and load their images from CPU 0. Startup scripts then are used to make sure that all of the routing is set up properly. This approach does two things simultaneously. First, it greatly reduces the traffic load on the boot host and the LAN. Second, it actually speeds the boot process dramatically, especially for systems with a large number of targets. Hmm... This sounds like a paper at the WRS Developer's conference ;-). If anyone would like to have me present this kind of thing (along with or in addition to, I could show how to add a new device to the boot ROMs and do "fail-soft" booting). Just drop me a note. If I get sufficient interest, I'll submit an abstract. HTH, ======================================================================== ==== === Real-Time System Development, Integration, Training and Services Mike Anderson mea@mclean.sparta.com Chief Engineer http://www.mclean.sparta.com SPARTA, Inc. V: (703) 448-0210 F: (703) 893-5494 "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." The opinions are mine... no one else would want to claim them ;-). ======================================================================== ==== === From bfujimoto@west.raytheon.com Tue Aug 25 17:37:51 1998 From: bfujimoto@west.raytheon.com Date: Tue Aug 25 17:37:54 PDT 1998 Subject: Can't fopen() from Shared-Mem Slave We have four MVME2604 SBCs configured in a VMEbus Backplane Shared-Memory Network running vxWorks. One SBC is a master/gateway to the external SPARCstation via Ethernet, the remaining SBCs are slaves on the Shared-Mem. network. We are attempting to open a file on the SPARCstation with the following WindShell commands: fd=0 fd=fopen("filename", "r") Executing this on the master/gateway SBC results in a non-zero value for fd. On the slave SBCs, we get zero for fd, and a "permission denied" message at the Virtual Console. Any ideas on why we can't execute fopen() on the slave SBCs? Thanks, Ben Fujimoto Raytheon Systems Co. From daemon@csg.lbl.gov Wed Aug 26 04:00:13 1998 From: daemon@csg.lbl.gov Date: Wed Aug 26 04:00:17 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed Aug 26 04:00:10 PDT 1998 Subject: driver for MPC860T Subject: Re: Looking for Evaluation board Subject: Re: Tornado on Linux/x86 Solaris Subject: Diasbling warnings from cc386. Subject: Can't get network running on BlueChip APX4D ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: driver for MPC860T Date: Tue, 25 Aug 1998 15:24:07 +0300 From: Yosi Yarchi Organization: Telematics International, Inc. Message-ID: <35E2ACE7.E5E7BD3@ecitele.com> Hi all Have someone experienced with writing ethernet driver for the MPC860T (the fast ethernet module) under vxWorks? any tips for that? (My reference code is the code supplied by motorola in their web site). Thanks in advance Yosi Yarchi --------------------------- Newsgroups: comp.os.qnx,comp.os.vxworks Subject: Re: Looking for Evaluation board Date: 19 Aug 1998 10:56:49 -0400 From: Andrew Thomas Organization: Cogent Real-Time Systems Inc. Message-ID: References: <6rdgqn$4hu$1@godzilla.krdl.org.sg> Followup-To: comp.os.qnx Sender: andrew@1 "WANG YiFei" writes: > I'm currently looking for an evaluation/reference board features x86 > processor, color LCD, TV output, audio in & out, serial/IR port, is there > anyone familiar with that? Thanks. The Intel Explorere II board has everything except color LCD and the TV output. It has an S3 something chipset that can drive an LCD if you get an LCD screen with VGA circuitry. Andrew - ----- Andrew Thomas, President, Cogent Real-Time Systems Inc. 2430 Meadowpine Boulevard, Suite 105, Mississauga, Ontario, Canada L5N 6S2 Email: andrew@cogent.ca WWW: http://www.cogent.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado on Linux/x86 Solaris Date: 25 Aug 1998 21:22:37 GMT From: james@unser.ucr.edu (James Wann) Organization: University of California, Riverside Message-ID: <6rv9ut$7fp$1@pravda.ucr.edu> References: Mittal Monani (mmonani@spacebridge.com) wrote: : Does WRS support Tornado and other tools on Linux or x86 Solaris? : If not, any heads-up on whether they plan to do that in near future? Not sure about solaris, but definitely not linux. However, it does come with source (at least the gnu portions of it). What I've done is built the cross compiler on linux and wrote some scripts to replace the vxworks utils. Works like a charm. - -james - -- --James Wann-- james@cs.ucr.edu Department of Computer Science, UC Riverside http://www.cs.ucr.edu/~james -- The human mind ordinarily operates at only ten percent of its capacity the rest is overhead for the operating system. --------------------------- Newsgroups: comp.os.vxworks Subject: Diasbling warnings from cc386. Date: Mon, 24 Aug 1998 21:17:59 -0700 From: Vicente Mendez Organization: QUALCOMM, Incorporated; San Diego, CA, USA Message-ID: <35E23AF7.F7361AF8@qualcomm.com> Hi all: I am looking for a way to selectively disable some of the warnings that cc386 spews out. Example below: class dummy { public: Initialize(); private: dummy() {} ~dummy() {} }; This will generate a warning about the class having private constructors and no friends. I would like to get rid of these warnings, since it makes it difficult to find useful ones. Borland C++ 4.5 had a "#pragma warn -w#" option or a "-w#" from the command line. Any help be greatly appreciated. Vicente Mendez --------------------------- Newsgroups: comp.os.vxworks Subject: Can't get network running on BlueChip APX4D Date: Wed, 26 Aug 1998 09:39:19 +0100 From: "Bill Boumphrey" Organization: Cegelec Projects Ltd. Message-ID: <6s0l56$tmj$1@news.cegelecproj.co.uk> We have a 486 based PC104 card made by Blue Chip which has an NE2000 compatible network on the PCI bus. We cannot get the BSP that comes with VxWorks (and which supports plain vanilla NE2000 cards) to work - it recognises the cards but when it comes to loading an image off the network it fails. This may be associated with the fact that we can't get pnp Etherlink III combo boards to work either. The PCI bus seems to be the common factor. Non pnp boards are fine. We have switched off pnp mode (a previous suggestion) with no success Anyone solved this problem? - -- Bill Boumphrey Cegelec Projects, Rubgy, UK wsb@cegelecproj.co.uk --------------------------- End of New-News digest ********************** From mea@mclean.sparta.com Wed Aug 26 06:28:08 1998 From: "Mike Anderson" Date: Wed Aug 26 06:28:11 PDT 1998 Subject: RE: Can't fopen() from Shared-Mem Slave Greetings! > > We have four MVME2604 SBCs configured in a VMEbus Backplane > Shared-Memory Network running vxWorks. One SBC is a > master/gateway to the > external SPARCstation via Ethernet, the remaining SBCs are slaves on the > Shared-Mem. network. > We are attempting to open a file on the SPARCstation with the > following > WindShell commands: > > fd=0 > fd=fopen("filename", "r") > > Executing this on the master/gateway SBC results in a non-zero value > for fd. On the slave SBCs, we get zero for fd, and a "permission denied" > message at the Virtual Console. > Any ideas on why we can't execute fopen() on the slave SBCs? > There could be any number of explainations. 1) You're not using the same "device" on the other CPUs -- rsh protocol on CPU 0 and FTP on the others or maybe some NFS mixed in. All of these access techniques have different requirements for IDs & passwords. 2) You've set the user ID on the other CPUs to a different value in the boot parameters 3) You're using the rsh protocol and you don't have the other CPUs listed in your .rhosts 4) If you're using the FTP protocol, your file permissions are not allowing the user from the other CPUs to access the file (no "other" permissions set on the file). If you're using the rsh protocol (no entry in the boot parameters "passwd" field), then my best guess from what little you've told us is option #3. HTH, ============================================================================ === Real-Time System Development, Integration, Training and Services Mike Anderson mea@mclean.sparta.com Chief Engineer http://www.mclean.sparta.com SPARTA, Inc. V: (703) 448-0210 F: (703) 893-5494 "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." The opinions are mine... no one else would want to claim them ;-). ============================================================================ === From rtp.co.uk!wmy@rtp.co.uk Wed Aug 26 09:56:58 1998 From: William Yates Date: Wed Aug 26 09:57:01 PDT 1998 Subject: Blue Chip APX4D BSP Bill, We have written a BSP for the APX4D, with the following features: - A driver for the onboard NE2000-compatible (Realtek 8029AS) Network Interface, using the BSD 4.3 Network Interface. - A driver for the onboard DS12887 Real Time Clock Device. - Integration of the M-Systems DiskOnChip2000 Driver into the BSP as a boot device. Please contact myself or 'enquiries@rtp.co.uk' for more information regarding the BSP. Regards, Will Yates, RTP --------------------------------------------------------------- Will Yates e-mail: mailto:wmy@rtp.co.uk Real Time Products Ltd. Direct: +44 (0) 121 234 6625 Chancery House, tel: +44 (0) 121 234 6600 8 Edward Street, fax: +44 (0) 121 234 6611 Birmingham. Mobile: +44 (0) 976 691 492 B1 2RX. England. web: http://www.rtp.co.uk --------------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Can't get network running on BlueChip APX4D Date: Wed, 26 Aug 1998 09:39:19 +0100 From: "Bill Boumphrey" Organization: Cegelec Projects Ltd. Message-ID: <6s0l56$tmj$1@news.cegelecproj.co.uk> We have a 486 based PC104 card made by Blue Chip which has an NE2000 compatible network on the PCI bus. We cannot get the BSP that comes with VxWorks (and which supports plain vanilla NE2000 cards) to work - it recognises the cards but when it comes to loading an image off the network it fails. This may be associated with the fact that we can't get pnp Etherlink III combo boards to work either. The PCI bus seems to be the common factor. Non pnp boards are fine. We have switched off pnp mode (a previous suggestion) with no success Anyone solved this problem? - -- Bill Boumphrey Cegelec Projects, Rubgy, UK wsb@cegelecproj.co.uk From daemon@csg.lbl.gov Thu Aug 27 04:01:16 1998 From: daemon@csg.lbl.gov Date: Thu Aug 27 04:01:22 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu Aug 27 04:01:13 PDT 1998 Subject: sample code for Tundra/Universe on motorola 2600/2300 boards Subject: Token Ring device drivers Subject: Re: WINNT to vxworks over PPP Subject: Anybody using CORBA (Visigenic port by Highlander Communications)? Subject: Database recommendations on VxWorks? Subject: Help (Visigenic C++ Q): VISList, VISDictionary etc. Subject: Driver for MPMC202 Required Subject: Re: how to do live software upgrades? Subject: Re: reboot vs. power cycle Subject: Re: how to do live software upgrades? Subject: Re: VxWorks or Tornado Subject: Re: how to do live software upgrades? Subject: Re: how to do live software upgrades? Subject: Re: how to do live software upgrades? Subject: VxWorks system not pingable Subject: Re: MetaWindows Subject: SUN X1059A fast ethernet adaptor Subject: Re: VxWorks system not pingable Subject: Re: PARALLEL PORTS ON A PC TARGET Subject: Re: VxWorks system not pingable ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: sample code for Tundra/Universe on motorola 2600/2300 boards Date: Tue, 25 Aug 1998 22:46:13 -0400 From: Stephen Adler Organization: Brookhaven National Laboratory Message-ID: <35E376F5.11CF028C@i-2000.com> I'm looking for someone who has programed the universe I or Universe II dma engine which is found on the mvme2600 and mvme2300 boards. I want to make use of this dma chip in order to pull data out of a home grown vme board with an output fifo into local memory on the mvme2300 board. At some point code will be written for this chip which I'll post on the web, but to get started on this project it would sure help to have some sample code or maybe even a full driver for it. Thanks. The web page link for this dma chip. http://www.tundra.com/unidex.html Steve. --------------------------- Newsgroups: comp.os.vxworks Subject: Token Ring device drivers Date: Wed, 26 Aug 1998 11:37:37 -0400 From: devault@ibm.net (Dan DeVault) Organization: Sundance Digital Development Message-ID: Hi, Does anyone know of any token-ring device drivers that have been developed for VxWorks ? We are using a Motorola MTX board and need to find some PCI token ring drivers or will have to develop them ourselves. Thanks, Dan DeVault --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WINNT to vxworks over PPP Date: 20 Aug 1998 13:11:59 GMT From: kevin_taylor@cegelecproj.co.uk (Kevin Taylor) Organization: Cegelec Projects Ltd. Message-ID: <6rh7av$ol5$2@news.cegelecproj.co.uk> References: <6rer7k$lka$1@nnrp1.dejanews.com> In article <6rer7k$lka$1@nnrp1.dejanews.com>, vuh@my-dejanews.com says... > >Hi, I am trying to connect my WINNT pc to my vxworks target using dialup >networking (PPP). On my vxworks target, vxworks has been initialized with >the following code: [snip] >Error 651: Your modem (or other connecting device) has reported an error. Isn't there something about WinNT not really supporting a NULL modem as a device in its own right ? Win 95 has a NULL modem device you can configure (?) but apparently Win NT does not. I am not exactly sure, but one of our guys had problems getting it going, and I am sure had to install a new device driver on the NT side (one that didn't actually try to guess anything fancy !). If this sounds like it could be a problem, mail me and I will try to get more details ... Kevin Taylor (kevin_taylor@cegelecproj.co.uk) --------------------------- Newsgroups: comp.os.vxworks Subject: Anybody using CORBA (Visigenic port by Highlander Communications)? Date: Tue, 25 Aug 1998 10:27:12 -0500 From: "Omkar Rath" Organization: PSINet Message-ID: <6rul2m$7v5$1@client3.news.psi.net> I have a few questions that I need answered. I would appreciate if those of you who use CORBA on VxWorks would reply. TIA, Omkar Rath --------------------------- Newsgroups: comp.os.vxworks Subject: Database recommendations on VxWorks? Date: Tue, 25 Aug 1998 10:29:26 -0500 From: "Omkar Rath" Organization: PSINet Message-ID: <6rul70$801$1@client3.news.psi.net> Hi! I am looking at RAIMA and Polyhedra. Right now more biased towards RAIMA (w/o really having tried it). Any suggestions? TIA, Omkar Rath --------------------------- Newsgroups: comp.object.corba,comp.os.vxworks Subject: Help (Visigenic C++ Q): VISList, VISDictionary etc. Date: Tue, 25 Aug 1998 10:40:29 -0500 From: "Omkar Rath" Organization: PSINet Message-ID: <6rulrm$84i$1@client3.news.psi.net> I stumbled upon the VIS... classes from the sample programs (VB 3.2 for NT). However, there is no documentation whatsoever. The samples only use VISThread and VISMutex. If Visigenic has gone into the effort of writing these templated classes I would very much like to use it. Hopefully they are portable across platforms? TIA, Omkar Rath PS: Feel free to reply directly. Shall summarize. --------------------------- Newsgroups: comp.os.vxworks Subject: Driver for MPMC202 Required Date: Fri, 21 Aug 1998 23:23:54 +0000 From: murray robinson Organization: Camtech Internet Customer Message-ID: <35DE0189.CF8649CC@vsl.com.au> Does anyone know of a "supported" VxWorks Driver for the Motorola MPMC202 Fast Ethernet Module? Alternatively, has anyone successfully used the dec21140 driver provided with the mvme2604 to control the mpmc202? I am observing the following problem: When I boot a mvme2604 (via the primary ethernet "dc0") with mpmc202 an active 10mbps link to the mpmc202 is auto-sensed and the 10mbps link LED turns on. When I then use the dcattach() function to configure the MPMC202 the 10mbps link led turns off and the mpmc202 ignores all ethernet traffic on the link. ifShow() indicates the the mpmc202 ("dc1") is up and running with zero errors, and no received/transmitted packets, yet I know for a fact that broadcast ethernet packets are being presented to this device. Any help would be appreciated. murray.robinson@vsl.com.au --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: how to do live software upgrades? Date: Thu, 13 Aug 98 10:03:22 GMT From: peb@tcontec.demon.co.uk ("Paul E. Bennett") Organization: Transport Control Technology Ltd. Message-ID: <903002602snz@tcontec.demon.co.uk> References: <35CC6F14.6B715069@possibility.com> Reply-To: peb@tcontec.demon.co.uk In article <35CC6F14.6B715069@possibility.com> tmh@possibility.com "Todd Hoff" writes: > We have a requirement to upgrade software hosted on a > vxworks system while still maintaining service. These > particular cards don't have a backup nor can their services > be transfered to another system while the upgrade is in > progress. > > It's not clear to me how a target, with c/c++ object > code, can update its software while providing uninterrupted > service. The code has to be downloaded into memory while the > current system is running and then start. > Does anyone have techniques for this? This is very difficult to do and you need to be very sure of what you are doing before you start. Having no fall-back system leaves you in a position of being on a knife edge during upgrades. In order to achieve this your system needs to have been designed to cope with such upgrades from the start. You also need to have structured the software such that it is an easy matter to switch pointers to modules and have the new module take-over immediately. It is tricky as you have to ensure that any variables/memory locations used are adopted by the new software module. The problem becomes one of knowing the new module will work correctly in the running environment even when tested out of the running environment. I do nt envy you in this task. I do not think you could do this on a whole system while live. Are you sure you cold not steal a few seconds out of the operation? - -- Paul E. Bennett ................... Transport Control Technology Ltd. +44 (0)117-9499861 Going Forth Safely --------------------------- Newsgroups: comp.os.vxworks Subject: Re: reboot vs. power cycle Date: Tue, 25 Aug 1998 10:05:21 -0600 From: Mike Stimpson Organization: Utah Scientific, Inc. Message-ID: <35E2E0C1.41C6@utsci.com> References: <35E17AAC.FF6D5DF@ing.iac.es> Matthieu Bec wrote: > > vxworks > > Hi, > I'm using a standard VME crate/ 68k motorolla CPU / 8Meg off-board > memory and DR11 board. > > Question: > does a power-cycle of the whole crate 'does more' than a simple reboot > from the shell ? > Is there any definite answer or this depends on the boards I got in my > rack ? > > Cheers, > > -- > Matthieu Bec Isaac Newton Group > tel : +34 922 425427 Apartado de Correos 321 > fax : +34 922 425401 38780 Santa Cruz de La Palma > email: mdcb@ing.iac.es Canary Islands, Spain Reboot doesn't do the same thing as a power cycle, even on your own board. What you need to do is a RESET instruction, and then the reboot. RESET drives the CPU's reset line low as an output, resetting the rest of the board (but not the CPU). In the case of the VME crate, you will have to test whether the reset line of the CPU board is driven to the other boards and resets them. RESET plus reboot is still not quite the same as a power cycle, if there are any chips where the reset line does not initialize them to the same state as power off does. Such differences *should* be minor, but you still might have trouble. If memory serves, the vxWorks 68k assemblers don't understand the RESET mnemonic, so you have to put it in as 2 bytes of data in-line with the code. - -- Mike Stimpson (mstimp@utsci.com) Opinions expressed are not necessarily those of the management... The more I work as an engineer, the more I disbelieve the theory of evolution. --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: how to do live software upgrades? Date: Thu, 13 Aug 1998 14:46:40 -0700 From: "Jack Peacock" Organization: Simco Message-ID: <6qvmgk$24t$1@supernews.com> References: <35CC6F14.6B715069@possibility.com> Todd Hoff wrote in message <35CC6F14.6B715069@possibility.com>... >We have a requirement to upgrade software hosted on a >vxworks system while still maintaining service. These >particular cards don't have a backup nor can their services >be transfered to another system while the upgrade is in >progress. > There seems to be something missing here. As I understand your description, these are critical systems which cannot be shut down, but which do not have any backups in case of failure? How is a software upgrade any different than a hardware failure? In a 24/7 availability situation (for example, a VMS Cluster) one node is brought down while the rest of the cluster continues to run the current version. The upgrade is installed on the downed node, it's brought back up, then each node in succession is brought down for the upgrade. Services are not interrupted. DEC calls this a "rolling upgrade". At the end, all nodes are upgraded but the cluster as a whole is never shut down. Jack Peacock --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks or Tornado Date: Fri, 21 Aug 1998 18:39:37 GMT From: john@kivala.com Organization: Kivala Systems, Inc. Message-ID: <6rketa$2tf$1@nnrp1.dejanews.com> References: In article , Steve Houtchen wrote: > VxWorks or Tornado > > Anybody know how to read the state of a binary semaphore or count > of a counting semaphore? I looked through the system header files, > but did not find the structures . I only have the SEM_ID to use. It's not officially sanctioned, but you look in the h/private directory for stuff like that. In this case, you find that SEM_ID is a pointer to struct semaphore, and dereference that to get the goods. You might have to experiment a little to see what the fields mean. John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Build Tornado Apps with MS Visual Basic - -------------------------------------------------- - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: how to do live software upgrades? Date: Fri, 14 Aug 1998 09:44:11 +0100 From: "Stephen Pelc" Organization: MicroProcessor Engineering Message-ID: <01bdc75f$bdab90c0$0200a8c0@stephen> References: <35CC6F14.6B715069@possibility.com> <903002602snz@tcontec.demon.co.uk> "Paul E. Bennett" wrote in article <903002602snz@tcontec.demon.co.uk>... > In article <35CC6F14.6B715069@possibility.com> > tmh@possibility.com "Todd Hoff" writes: > > We have a requirement to upgrade software hosted on a > > vxworks system while still maintaining service. These > > particular cards don't have a backup nor can their services > > be transfered to another system while the upgrade is in > > progress. > Does anyone have techniques for this? > > This is very difficult to do and you need to be very sure of what you > are doing before you start. Having no fall-back system leaves you in > a position of being on a knife edge during upgrades. We did this for a client on a Z180 based system a while back. The fail-safe solution for us was to have enough RAM to hold the entire new image during reception. The knife edge then became power failure during the copy from RAM to Flash. This can be overcome using a boot block and good hardware design. Just how paranoid do you need to be? In newer projects we now believe that the solution is not to change the whole image at once, but to deliver software as binary modules. The performance hit is minimal using jump tables and good code factoring. The issue is in the program load format. Then you have to decide what is kernel, and what is module. - -- Stephen Pelc, MicroProcessor Engineering - More real, less time 133 Hill Lane, Shirley, Southampton SO15 5AF, England tel: +44 1703 631441, fax: +44 1703 339691, net: sfp@mpeltd.demon.co.uk web: http://www.mpeltd.demon.co.uk --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: how to do live software upgrades? Date: Fri, 14 Aug 1998 10:21:05 GMT From: nmurphy@iol.ie (Niall Murphy) Organization: Ireland On-Line Message-ID: <35d40f72.7329006@news.iol.ie> References: <35CC6F14.6B715069@possibility.com> <903002602snz@tcontec.demon.co.uk> <35D2DA6A.C02169D3@possibility.com> I talked to a guy from AT&T about this once, and the following is a mechanism that he described using: Since the virtual function pointer to a class is a single point of control that all of the member functions pass through, then make the overwriting of this pointer the upgrade point. The new class must inherit from the old class. Any function calls before the upgrade get the base class, and after the upgrade, they get the derived class. This is a bit restrictive, and I am not sure if you would be able to apply the scheme to your particular tools, but I thought I would chip it in as an interesting idea. Niall On Thu, 13 Aug 1998 05:22:02 -0700, Todd Hoff wrote: > >>The problem becomes one of knowing the new module will >> work correctly in the running environment even when tested out of the >> running environment. I do nt envy you in this task. I do not think you >> could do this on a whole system while live. Are you sure you could not >> steal a few seconds out of the operation? > >------------------------------------------------------------------ > tmh@possibility.com http://www.possibility.com/Tmh > Desperate measures require the invention of desperate times. > -- Todd Hoff See Niall Murphy's User Interfaces for Embedded Systems Home Page at http://www.iol.ie/~nmurphy --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: how to do live software upgrades? Date: Fri, 14 Aug 1998 10:14:53 +0100 From: Stephen Maudsley Organization: esgem limited Message-ID: <35D4000D.C8C2428F@esgem.com> References: <35CC6F14.6B715069@possibility.com> <6qvmgk$24t$1@supernews.com> <6r026n$g9q$1@nnrp1.dejanews.com> Reply-To: Stephen.Maudsley@esgem.com toddhoff@my-dejanews.com wrote: > In article <6qvmgk$24t$1@supernews.com>, > > There seems to be something missing here. As I understand > > your description, these are critical systems which cannot be > > shut down, but which do not have any backups in case of > > failure? How is a software upgrade any different than a > > hardware failure? > > The cards in question are communication cards. We can do the > rolling update on our other cards that have a hot stand-by. > The communication cards don't have a natural backup because there > is only one connection between point X and Y. There are protection > routes for some connections, not all, and they aren't to > be used for software ugrades anyway. Is now the time to put in multiple comms paths? Increase the availability of the machine, increase the comms bandwidth and ease upgrades? Multiple paths protects (sometimes) against carpenters with nail guns etc.. - -- Stephen Maudsley mailto:Stephen.Maudsley@esgem.com Esgem Limited for electronic product design http://www.esgem.com Tel: +44-1453-521626 Mobile: +44-370-810991 --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks system not pingable Date: Thu, 27 Aug 1998 08:39:32 +0200 From: Erik Tijssen Organization: UUNET-NL Message-ID: <35E4FF23.659C15C6@Syntens.nl> Hello, We have here an Intel 486 bases board running VxWorks. We want to reach this board using Internet. We can't get this running. The situation is as follows. When the board powers up, it looks for a specified host to load the OS and a program to run. This is IP based. The host is a PC in our local network. If this all runs, it's possible to ping the VxWorks system FROM INSIDE our local network, it's possible to ping the host over the internet, but it's not possible to ping the VxWorks system. Host and system are in the same segment of our local network. Is there something missing on the ES to reply to the ping? It has nothing to do with blocking in our gateway, because switching the IP numbers gives the same result. Please help me, Erik Tijssen --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MetaWindows Date: Thu, 27 Aug 1998 08:03:10 +0200 From: "Jxrgen Belfalas" Organization: Radiometer Medical A/S, Danmark Message-ID: <35E4F69E.4F841F42@radiometer.dk> References: <6rk8j9$i4q$1@news.ncal.verio.com> Reply-To: joergen.belfalas@radiometer.dk Hi Dave, > I wondered if anyone has any experience using MetaWindows on VxWorks from > EASE software? It seems like a good deal (no target license fees) and it's > apprarently got an object oriented API? Can anyone shed any more light on > how long it's been around? I have used MetaWINDOWS for vxWorks for more than a year and I find that it is a very powerful toolkit. As far as I know there is no high level API. I have made my own object oriented API that implements buttons, input boxes, mesagge boxes and so on. It is very easy and MetaWINDOWS is very fast. Moreover Ease Software provides a fine support. - -- Jorgen --------------------------- Newsgroups: comp.os.vxworks Subject: SUN X1059A fast ethernet adaptor Date: Tue, 25 Aug 1998 23:34:29 +0200 From: "B.J.M. Vloedbeld" Organization: World Access / Planet Internet Message-ID: <35E32DE5.5D49699C@wxs.nl> - --------------1DF1840068520A12FAEC309C Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello, We want to use the SUN X1059A fast ethernet adaptor on a FORCE Sparc/cpu-5V board. Questions : 1 - is there somewhere a driver for this ethernet board ? 2 - who is able to deliver (develop?) such a driver ? regards, B.J.M.Vloedbeld please respond "also" by email to : bjmv@signaal.nl I do not read the Usenet news very often. ====================================================================== B.J.M. Vloedbeld dept. Systems PA | phone : +31 (0) 74 248 3994 Hollandse Signaalapparaten B.V. | FAX : +31 (0) 74 248 4030 Zuidelijke Havenweg 40 | email : bjmv@signaal.nl 7554 RR Hengelo, The Netherlands | ====================================================================== - --------------1DF1840068520A12FAEC309C Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Hello,

We want to use the SUN X1059A fast ethernet adaptor on a
FORCE Sparc/cpu-5V board.

Questions :
    1 - is there somewhere a driver for this ethernet board ?
    2 - who is able to deliver (develop?) such a driver ?

regards,
        B.J.M.Vloedbeld
 

please respond "also" by email to :
                bjmv@signaal.nl
I do not read the Usenet news very often.

======================================================================
B.J.M. Vloedbeld
dept. Systems PA                       | phone : +31 (0) 74 248 3994
Hollandse Signaalapparaten B.V.        | FAX   : +31 (0) 74 248 4030
Zuidelijke Havenweg 40                 | email : bjmv@signaal.nl
7554 RR Hengelo, The Netherlands       |
======================================================================
  - --------------1DF1840068520A12FAEC309C-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks system not pingable Date: 27 Aug 1998 07:34:51 GMT From: Stephen A Walsh Organization: GEC-Marconi Avionics Ltd, Ferry Road, Edinburgh, Scotland Message-ID: <6s326r$iun@gcsin3.geccs.gecm.com> References: <35E4FF23.659C15C6@Syntens.nl> Erik Tijssen wrote: >Hello, > < Snip> > >Is there something missing on the ES to reply to the ping? It has >nothing to do with blocking in our gateway, because switching the IP >numbers gives the same result. Have you got the following define in CONFIG.H #define INCLUDE_PING if so then check that there is not an #undef INCLUDE_PING later in CONFIG.H or in CONFIGDB.H Hope this helps Steve ********************************** Opinions Are As Always My Own ********************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PARALLEL PORTS ON A PC TARGET Date: Fri, 21 Aug 1998 07:37:15 GMT From: B.Metz@or-computers.de (Bertram Metz) Organization: or Industrial Computers GmbH Message-ID: <35dd21a2.2461910@news.profi.net> References: <6rfhue$700$1@susscsc1.rdg.ac.uk> <35dbe3ff.7205180@news.profi.net> oops, EPP/ECP modes of the parallel port are _not_ supported ! Here is the contents of the FAQ, which I mentioned in the previous posting. Bertram TSR# 49382 DATE: Nov 20 1997 6:41PM .................................................................................... TITLE: lptDrv() in 5.3.1. A couple of tips SCOPE DISCOVERED ON HOST: N/A ------------------- ------------------- ARCHITECTURE: x86 ------------------- ------------------- BSP: pc486/pc386 ------------------- ------------------- PRODUCT / REVISION: 5.3.1 ------------------- ------------------- DESCRIPTION OF PROBLEM: lptDrv() in 5.3.1 doesn't appear to work. For 5.3.1, the lprDrv underwent a major revision. It should work better than the previous lptDrv(). However the PC must be set up correctly. The SETUP parameters on most pc boards allow you to set the port and sometimes the type of the parallel port. The pc486 is shipped assuming a parallel port 0x3bc, interrupt level 7. These are defined in pc.h in the bsp directory. Either pc.h or the SETUP parameters should be modified so they match. Please note that only NORMAL parallel ports are supported as of 5.3.1. EPP parallel ports are not and the SETUP parameters on the pc486 board must reflect this. The following trivial program can be used to setup and test the lpt port on the pc386/486. - ---------------------Begin Code--------------------------- #include "vxWorks.h" #include "lptDrv.h" #define TESTDATA "\n1234567890abcdefghijklmnopqrstuvwxyz" #define TESTDATASIZE sizeof(TESTDATA) #define PRINTER "/dev/lpt0" void test() { int status; int fd; status = lptDevCreate(PRINTER,0); if (status != OK) { printf("\n Error in lptDevCreate"); return; } fd = open(PRINTER,O_WRONLY); if (fd == ERROR) { printf("\n Error opening the device"); return; } for (count = 0; count < 100; count++) { if ((status = write(fd,TESTDATA,TESTDATASIZE)) != TESTDATASIZE) { printf("\n Error in write - bytes %d",status); return; } } close(fd); } - ---------------------End Code-------------------------------- Bertram Metz b.metz@or-computers.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks system not pingable Date: Thu, 27 Aug 1998 15:16:51 +0530 From: "Venkat" Organization: Novalink Technologies, New Delhi Message-ID: <6s39qr$sps$1@taiwan.informatik.unirostock.de> References: <35E4FF23.659C15C6@Syntens.nl> <6s326r$iun@gcsin3.geccs.gecm.com> Hi Stephen A Walsh wrote in message <6s326r$iun@gcsin3.geccs.gecm.com>... > >Have you got the following define in CONFIG.H > > #define INCLUDE_PING > INCLUDE_PING only ensures that the ping utility gets included when VxWorks is built, i.e. links in the code for the ping() function, i.e. you can then CALL ping FROM the VxWorks target. Any host with a functional IP layer is supposed to reply to pings (ICMP echo requests). I think the reason for Erik's problem may be that he has not specified a gateway in his VxWorks box's configuration. This happens quite often to me too. You can specify a gateway in the boot line by adding a "gw" option. A gateway must be specified if the VxWorks host has to send an IP datagram to another host that is not on its subnet. Hope this helps. Venkat --------------------------- End of New-News digest ********************** From mea@mclean.sparta.com Thu Aug 27 06:27:42 1998 From: "Mike Anderson" Date: Thu Aug 27 06:27:45 PDT 1998 Subject: RE: VxWorks system not pingable Greetings! Erik Wrote: > We have here an Intel 486 bases board running VxWorks. We want to reach > this board using Internet. We can't get this running. The situation is > as follows. > > When the board powers up, it looks for a specified host to load the OS > and a program to run. This is IP based. The host is a PC in our local > network. If this all runs, it's possible to ping the VxWorks system FROM > INSIDE our local network, it's possible to ping the host over the > internet, but it's not possible to ping the VxWorks system. Host and > system are in the same segment of our local network. > > Is there something missing on the ES to reply to the ping? It has > nothing to do with blocking in our gateway, because switching the IP > numbers gives the same result. > The other reply to this message that suggested a gateway be added is close, but what's really needed is a *default* route. A default route essentially says, "If you don't have a specific route to some destination, then send it here. Maybe that box knows what to do with it." The use of the gateway field in the boot parameters will not accomplish this. Look at /src/config/usrNetwork.c and you'll see that the "gw" parameter will extract the boot host's network number and establish at net route to that network. That has already been done because your target is directly connected to the same segment as the boot host. To add a default route, simply do the following: -> routeAdd("0.0.0.0", "your Internet gw address") Then any traffic that goes to the outside world (i.e., not on the local subnet) will be forwarded to the Internet gateway for further processing. HTH, ============================================================================ === Real-Time System Development, Integration, Training and Services Mike Anderson mea@mclean.sparta.com Chief Engineer http://www.mclean.sparta.com SPARTA, Inc. V: (703) 448-0210 F: (703) 893-5494 "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." The opinions are mine... no one else would want to claim them ;-). ============================================================================ === From wheat@wg.com Thu Aug 27 07:16:43 1998 From: Lee Wheat Date: Thu Aug 27 07:16:46 PDT 1998 Subject: Re: VxWorks system not pingable --=====================_7946706==_.ALT Content-Type: text/plain; charset="us-ascii" vxWorks... do you have a default route entry in the embedded system? if not, then pings from outside of the local segment will not get a reply since the protocol stack will not know how to get packets out of the local segment. do a routeShow() from the shell. you should see the route that is added to the routing table when the interface card is assigned its ip address. if you do not also see an entry for the 0.0.0.0 network, then that is why you system cannot be pinged from the internet. see routeAdd() in the reference manual for how add routes to the routing table. the addition to your code would be routeAdd ("0.0.0.0", "198.85.38.254") where 198.85.38.254 is address of the gateway out of the local network. HTH, lee wheat wheat@wg.com At 04:16 AM 8/27/98 -0700, you wrote: > > Newsgroups: comp.os.vxworks > Subject: VxWorks system not pingable > Date: Thu, 27 Aug 1998 08:39:32 +0200 > From: Erik Tijssen > Organization: UUNET-NL > Message-ID: <35E4FF23.659C15C6@Syntens.nl> > > Hello, > > We have here an Intel 486 bases board running VxWorks. We want to reach > this board using Internet. We can't get this running. The situation is > as follows. > > When the board powers up, it looks for a specified host to load the OS > and a program to run. This is IP based. The host is a PC in our local > network. If this all runs, it's possible to ping the VxWorks system FROM > INSIDE our local network, it's possible to ping the host over the > internet, but it's not possible to ping the VxWorks system. Host and > system are in the same segment of our local network. > > Is there something missing on the ES to reply to the ping? It has > nothing to do with blocking in our gateway, because switching the IP > numbers gives the same result. > > Please help me, > > Erik Tijssen --=====================_7946706==_.ALT Content-Type: text/html; charset="us-ascii" vxWorks...

do you have a default route entry in the embedded system? if not, then
pings from outside of the local segment will not get a reply since the
protocol stack will not know how to get packets out of the local segment.
do a routeShow() from the shell. you should see the route that is added
to the routing table when the interface card is assigned its ip address.
if you do not also see an entry for the 0.0.0.0 network, then that is why
you system cannot be pinged from the internet. see routeAdd() in the
reference manual for how add routes to the routing table. the addition
to your code would be routeAdd ("0.0.0.0", "198.85.38.254") where 198.85.38.254 is address of the gateway out of the local network.

HTH,
lee wheat
wheat@wg.com


At 04:16 AM 8/27/98 -0700, you wrote:

Newsgroups: comp.os.vxworks
Subject: VxWorks system not pingable
Date: Thu, 27 Aug 1998 08:39:32 +0200
From: Erik Tijssen
Organization: UUNET-NL
Message-ID: <35E4FF23.659C15C6@Syntens.nl>

Hello,

We have here an Intel 486 bases board running VxWorks. We want to reach
this board using Internet. We can't get this running. The situation is
as follows.

When the board powers up, it looks for a specified host to load the OS
and a program to run. This is IP based. The host is a PC in our local
network. If this all runs, it's possible to ping the VxWorks system FROM
INSIDE our local network, it's possible to ping the host over the
internet, but it's not possible to ping the VxWorks system. Host and
system are in the same segment of our local network.

Is there something missing on the ES to reply to the ping? It has
nothing to do with blocking in our gateway, because switching the IP
numbers gives the same result.

Please help me,

Erik Tijssen

--=====================_7946706==_.ALT-- From greg.brissey@nmr.varian.com Thu Aug 27 08:22:40 1998 From: greg.brissey@nmr.varian.com (Greg Brissey x6951) Date: Thu Aug 27 08:22:44 PDT 1998 Subject: Re: tundra Universe II dma driver, vxworks Steve writes, > > I'm looking for someone who has programed the universe I or Universe II > dma engine which is found on the mvme2600 and mvme2300 boards. snip ..... We are in the process of switching from a mv162 to a mv2300 and use the the DMA on the mv162. I too would be interested in sample code or maybe even a full driver for the dma engine on the mv2300. Greg greg.brissey@nmr.varian.com From dsunnarb@wavetech.net Thu Aug 27 09:21:20 1998 From: Dan Sunnarborg Date: Thu Aug 27 09:21:23 PDT 1998 Subject: VxWorks and the DosFs Hello, Are you forced to pay Microsoft royalties for DOS if you are using the dosFs on your target? Does formatting the hard disk with the DOS disks bind you to such an agreement? Is there a way under VxWorks 5.3.1 to format the hard disk and designate the dosFs? Thanks! Dan Sunnarborg Applied Biometrics Inc. From AKBiocca@lbl.gov Thu Aug 27 11:21:50 1998 From: francois cai (by way of Alan K Biocca ) Date: Thu Aug 27 11:21:53 PDT 1998 Subject: How can I load and create a dynamic task in the Vxworks from Name:Caiqingfeng Company: HuaWei Co.Ltd shenzhen China email: caiqingfeng@huawei.com.cn My application looks like this: A bootable task reads the file named config in the rom. The config file defines the number of tasks in the system,and the bootable task loads each task's module from a flash. You will think it as DLL ( Dynamic Link Library) in the Windows or Unix. If my bootable task works in the SVR4 UNIX ,it will like this: ... void (* subfunc)(); void * handle; handle=dlopen("subtask.so",...); subfunc=dlsym(handle,"subtask",FUNC_MODE); if (fork() ==0) { (*subfunc)(); exit(0); } .... If my bootable task works in the NT/95 ,it will like this: HINSTANCE gLibMyDll; gLibMyDll=LoadLibray("subtask.dll"); void (* subfunc)(); subfunc=GetProcAddress(gLibMydll,"subtask"); HANDLE m_hProcessThread = CreateThread( NULL, 0, (LPTHREAD_START_ROUTINE)subfunc, this, CREATE_SUSPENDED, &nThreadID ) ; if (m_hProcessThread == NULL) return ; BOOL bOK = SetThreadPriority( m_hProcessThread, THREAD_PRIORITY_BELOW_NORMAL); if (m_hProcessThread != NULL) ResumeThread(m_hProcessThread) ; ... How can I code in the VXWORKS? Someone help me. Thanks! From bob_anderson@mcg.mot.com Thu Aug 27 13:04:30 1998 From: Bob Anderson Date: Thu Aug 27 13:04:34 PDT 1998 Subject: RE: tundra Universe II dma driver, vxworks Contact your local Motorola Computer Group sales office. The SE there can have a patch sent to you that provides additional support for the Universe (including DMA). -----Original Message----- From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] Sent: Thursday, August 27, 1998 11:23 AM To: vxworks_users@csg.lbl.gov Subject: Re: tundra Universe II dma driver, vxworks Submitted-by greg.brissey@nmr.varian.com Thu Aug 27 08:22:40 1998 Submitted-by: greg.brissey@nmr.varian.com (Greg Brissey x6951) Steve writes, > > I'm looking for someone who has programed the universe I or Universe II > dma engine which is found on the mvme2600 and mvme2300 boards. snip ..... We are in the process of switching from a mv162 to a mv2300 and use the the DMA on the mv162. I too would be interested in sample code or maybe even a full driver for the dma engine on the mv2300. Greg greg.brissey@nmr.varian.com ********** From sinnl@sd-star.com Thu Aug 27 18:00:16 1998 From: Larry Sinn Date: Thu Aug 27 18:00:20 PDT 1998 Subject: Re: DMA on mvme2300, VxWorks Steve: Following is a simple DMA routine we use. That's the good news, the bad news is that as far as I can tell, the Universe always increments the source and destination address, meaning unless your FIFO answers to several consecutive addresses, DMA with the Universe chip will do you no good. Larry. Larry Sinn Spectral Dynamics (408)474-1746 voice 1983 Concourse Drive (408)474-1780 fax San Jose, Ca. 95131-1708 sinnl@sd-star.com /* dma_test.c */ /*-------------*/ /* This tests the DMA functionality by writing a ramp (to the specified * nodes dual-ported memory) starting with 'data' to each node selected for * the specified 'brd' and then DMAing the data back to the PPC. * * Note that we are using the Universe chip on the PPC to control the DMA. * Since the Universe chip resides on the PCI bus of the PPC we must byte * swap any access to it. This is because the Motorola CPU is a big-endian * device and the PCI bus is a little-endian device. (VME is big-endian) * * The 'node' input parameter is used as a bit map to determine * which node to test. For example a 5 in this field will result * in testing nodes 0 and 2. */ /* Code executes on VxWorks target - use VxWorks header files. */ /*-------------------------------------------------------------*/ #include "vxWorks.h" #include "taskLib.h" #include "wdLib.h" #include "vme.h" #include "sysLib.h" #include "cacheLib.h" #include "string.h" #include "univ_dma.h" #include "mach5.h" #define TWO_TICKS 2 #define UNIVERSE_BASE_MOTO 0xfd050000 #define UNIVERSE_BASE_FORCE 0xfe810000 #define UNIVERSE_BASE_VI 0x81050000 extern int printf(); extern int scanf(); volatile UNIVERSE_REGS *univ = 0; int pci2dram_base_adrs = 0; /* memory seen from PCI bus */ extern struct mach5_map *cP ; volatile int dma_timeout; int *fP, *destP, *readP; int dma_timeoutISR(int param) { dma_timeout = 1; return(0); } int fifo_dma(int ibrd) { struct mach5_map *machP; int rdata, swp_DGCS_DONE; int ilop, offset, brd, blksiz; WDOG_ID dma_wdId; char *modelP; UINT32 unit32, utmp; int vme_type; /* get test info */ blksiz = 40; offset = 0; modelP = sysModel(); printf("modelP=%s\n", modelP ); if ( strstr(modelP, "Motorola") ) { vme_type = VME_AM_STD_USR_DATA; univ = (UNIVERSE_REGS *)UNIVERSE_BASE_MOTO; pci2dram_base_adrs = 0x0; } else if ( strstr(modelP, "FORCE") ) { vme_type = VME_AM_STD_SUP_DATA; univ = (UNIVERSE_REGS *)UNIVERSE_BASE_FORCE; pci2dram_base_adrs = 0; } else if ( strstr(modelP, "VI") ) { vme_type = VME_AM_STD_USR_DATA; univ = (UNIVERSE_REGS *)UNIVERSE_BASE_VI; pci2dram_base_adrs = 0x80000000; } else { errno = 5757; return ERROR; } brd = (ibrd * sizeof(volatile struct mach5_map)); machP = (struct mach5_map *)(cP + brd); destP = (int *)cacheDmaMalloc(sizeof(int)*blksiz); printf("machP = %p, destP = %p\n",machP,destP); fP = (int *)(&machP->fifo[offset]); readP = destP; /* Setup the DGCS, be sure to clear STOP,HALT,DONE,and */ /* the three error bits by writing a '1' to them P.S. */ unit32 = LONGSWAP( univ->dgcs ); unit32 = ( unit32 | DGCS_STOP | DGCS_HALT | DGCS_DONE | DGCS_LERR | DGCS_VERR | DGCS_P_ERR | DGCS_VON_DONE); printf("addr univ->dgcs=0x%x value=0x%x\n",(int)&univ->dgcs, unit32); univ->dgcs = LONGSWAP( unit32 ); /* Define the type of VME access (A24-D32 Super) */ printf("addr univ->dctl=0x%x\n",(int)&univ->dctl); univ->dctl = LONGSWAP(DCTL_MASK | DCTL_VDW_32 | DCTL_VAS_A24 | DCTL_SUPER_USER | DCTL_VCT_EN); utmp = univ->lintStat; printf("lintStat=0x%x\n", LONGSWAP(utmp) ); univ->lintStat = ( utmp | LONGSWAP( LINT_STAT_DMA ) ); /* Setup the DTBC(DMA transfer BYTE counter) */ printf("addr univ->dtbc=0x%x\n",(int)&univ->dtbc); univ->dtbc = LONGSWAP(DTBC_MASK | (blksiz * 4)); /* DLA(DMA Local Address) PCI address */ printf("addr univ->dla=0x%x\n",(int)&univ->dla); printf("dla addr=0x%x\n",(int)destP + pci2dram_base_adrs); univ->dla = LONGSWAP((int)destP + pci2dram_base_adrs); /* DVA(DMA VME Address) */ printf("addr univ->dva=0x%x\n",(int)&univ->dva); printf("mach5 addr=0x%x\n",(int)fP); univ->dva = LONGSWAP((int)fP); printf("DLA = 0x%x\n",LONGSWAP(univ->dla)); /* start DMA */ printf("Before 'GO'\n"); printf("DGCS = 0x%x \n",LONGSWAP(univ->dgcs)); printf("writing DGCS = 0x%x \n",( DGCS_GO | DGCS_VON_DONE)); univ->dgcs = LONGSWAP( unit32 | DGCS_GO ); /* setup watchdog stuff */ dma_timeout = 0; dma_wdId = wdCreate(); wdStart(dma_wdId, TWO_TICKS, dma_timeoutISR, 0); swp_DGCS_DONE = LONGSWAP(DGCS_DONE); while(dma_timeout == 0) { if((univ->dgcs & swp_DGCS_DONE) == swp_DGCS_DONE) { wdDelete(dma_wdId); break; } } if(dma_timeout == 1) { printf("\nDMA TIMEOUT! DMA never finished.\n\n"); printf("DGCS = 0x%x \n",LONGSWAP(univ->dgcs)); } else { wdDelete(dma_wdId); for (ilop = 0; ilop < blksiz; ilop++) { rdata = *readP; printf("addr=0x%p data=0x%x\n",readP, rdata); readP++; } if(ilop == blksiz) { printf("\nTest DONE***\n"); printf("DGCS = 0x%x \n",LONGSWAP(univ->dgcs)); } } cacheDmaFree(destP); return OK; } From jhuggins@va.net Thu Aug 27 20:24:03 1998 From: "John Huggins" Date: Thu Aug 27 20:24:07 PDT 1998 Subject: Re: DMA on mvme2300, VxWorks For those of you interested in the Universe chip and how it does DMA, you might want to examine some of the VMELinux driver code. Its all available at http://www.vmelinux.org/. Learning this DMA is well worth the effort. The Universe chip is pretty slow when using atomic PCI accesses. Whether you are using regular or block mode to access the VMEbus, use the PCI DMA for best results. Even in non-block VME mode, the cycle time is around 275 nanoseconds, MUCH faster than the Universe's leisurely 1.2 microsecond non-DMA rate. If any of you can make our driver work under vxWorks, go for it. Its available under the GNU license. Sincerely, John Huggins VMELinux Project vmelinux@va.net From daemon@csg.lbl.gov Fri Aug 28 04:00:35 1998 From: daemon@csg.lbl.gov Date: Fri Aug 28 04:00:43 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri Aug 28 04:00:32 PDT 1998 Subject: Re: My VxWorks DOS experience.... Subject: test 3 Subject: Re: how to do live software upgrades? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: My VxWorks DOS experience.... Date: Sun, 23 Aug 1998 22:01:10 GMT From: ufoss@pop-stuttgart.net (Uwe Fo_) Message-ID: <35e08b03.368790350@news.pop-stuttgart.net> References: <35D7EAD8.4279D0C3@tasc.com> Reply-To: ufoss@pop-stuttgart.net On Mon, 17 Aug 1998 04:33:28 -0400, Chris Williamson wrote: >I'm using the DOS file system on PCMCIA cards, connected through SCSI, >and I've discovered that there are limits to how many files I can put on >the card. I've got hundreds of little files, and if I run into a limit >the operating system hangs. > >I've been able to get about 800 files on the disk, 400 files into one >subdirectory and 400 into another. My application will probably >includes 2,000 in the end. > >Anybody had any experience with DOS on PCMCIA? I can't get the disk to >accept any more than 800 files (the disk has 20 megabytes and the files >are about 1.5 kilobytes, meaning I've put about 1.3 megabytes worth of >stuff on the disk). We are using the Tornado pccard stack on a 320MB / 540 MB Calluna ata harddisk on a Vadam PCCARD controller(PC-Target). Sometime we store about 10000 files within one subdir. The amount an harddisk space physically used is strongly depending on the clustersize (i.e. 8KByte) Clustersize is given by formatation. So a small file is already using 8KB on harddisk but the fileinfo is more less . With 800 Files you are using 800*8KB = 6MB on your disk !!! And the Fileinfo is about 1,2MB. First: The Dos -utilities ndd (norton disk doctor) and speeddisk were not able to handle more then 2000 files within one subdir. Second: The file access performance is *rapidly* decreasing by the amount of files stored within a subdir. Especially for the fopen() command. Third: Cause the filestructure is held in memory it is usefull to synchronize the memory structure with harddisk. Especially when you are not using the flag CHANGE_NOWARN which forces synchronisation on each file access. >Also, the Programmer's Guide says VxWorks DOS is compatible with any >other DOS system, but when I mount the card in my PCMCIA slot on my Sun >Ultra 2, I can view parts of a file on the disk (the front part), but I >get binary gibberish at the back end of the file. Anybody try reading a >VxWorks DOS file system with another machine? I4m able to read my pccards with a dos 6.22 and on a Win NT 4.0 system. >Chris > Uwe Foss Siemens AG Germany - ------------------------ Uwe Fo_ ufoss@pop-stuttgart.net - ------------------------ --------------------------- Newsgroups: comp.os.vxworks Subject: test 3 Date: Mon, 24 Aug 1998 11:24:33 +0300 From: Yosi Yarchi Organization: Telematics International, Inc. Message-ID: <35E12341.3AA89FA7@ecitele.com> test 3 --------------------------- Newsgroups: comp.os.vxworks,comp.arch.embedded Subject: Re: how to do live software upgrades? Date: 24 Aug 1998 11:26:13 +0200 From: torbenm@diku.dk (Torben AEgidius Mogensen) Organization: Department of Computer Science, U of Copenhagen Message-ID: <6rrbjl$kgk@grimer.diku.dk> References: <35CC6F14.6B715069@possibility.com> <35E11AFA.AA8153F0@eagercon.com> Michael Eager writes: >Todd Hoff wrote: >> >> We have a requirement to upgrade software hosted on a >> vxworks system while still maintaining service. These >> particular cards don't have a backup nor can their services >> be transfered to another system while the upgrade is in >> progress. >Make all calls to routines go thru a jump vector. Loadable >code should either be position independent, or you will need to >relocate it when you load it. When you load upgrades, change >the vector entry to point to the new routine. >You need to make sure that the high-level interfaces (the ones >pointed to by the jump vector) do not change. Some programming languages (e.g. Erlang) has high-level support for on-the-fly module upgrades. Erlang has been used for telephone exchanges and similar systems where downtime is unacceptable. Torben Mogensen (torbenm@diku.dk) --------------------------- End of New-News digest ********************** From John_W_Cosgrove@res.raytheon.com Fri Aug 28 04:28:48 1998 From: John_W_Cosgrove@res.raytheon.com Date: Fri Aug 28 04:28:55 PDT 1998 Subject: Re: How can I load and create a dynamic task in the Vxworks from --0__=h2jOornxhtdzUvLIDVafhG0Yo0ubzms4dU9GhKCtukgM5Uw3uLGfgXbl Content-type: text/plain; charset=us-ascii Try these two subroutines. I think they work well... For instance, rc = vx_load_module("MY_FILE_NAME"); rc = vx_start_task("Symbol Name entry point", "NAME of Task", Task Priority, Task Options, Task Stack Size); (See attached file: vx_load_module.c)(See attached file: vx_start_task.c) John Cosgrove Raytheon Systems Company (401)842-4167 jwc@ssd.ray.com >Submitted-by AKBiocca@lbl.gov Thu Aug 27 11:21:50 1998 >Submitted-by: francois cai (by way of Alan K Biocca >) > >Name:Caiqingfeng >Company: HuaWei Co.Ltd shenzhen China >email: caiqingfeng@huawei.com.cn > >My application looks like this: > A bootable task reads the file named config in the rom. The >config file defines the number of tasks in the system,and the >bootable task loads each task's module from a flash. You will >think it as DLL ( Dynamic Link Library) in the Windows or Unix. [snip!] --0__=h2jOornxhtdzUvLIDVafhG0Yo0ubzms4dU9GhKCtukgM5Uw3uLGfgXbl Content-type: application/octet-stream; name="vx_load_module.c" Content-transfer-encoding: base64 I2luY2x1ZGUgPHZ4V29ya3MuaD4NCiNpbmNsdWRlIDxsb2FkTGliLmg+DQojaW5jbHVkZSA8c3Rk aW8uaD4NCiNpbmNsdWRlIDxmY250bC5oPg0KDQpTVEFUVVMNCnZ4X2xvYWRfbW9kdWxlKGNoYXIJ KnBNb2R1bGUpDQp7DQogICAgaW50CQkJZnA7DQogICAgU1RBVFVTCQlyYzsNCg0KICAgIC8qDQog ICAgICogT3BlbiB0aGUgZmlsZSAoYXNzdW1lcyBjdXJyZW50IGRpcmVjdG9yeSBvciBmdWxsIHBh dGgsIA0KICAgICAqIGFuZCBORlMgbW91bnRlZA0KICAgICAqLw0KICAgIGlmICgoZnAgPSBvcGVu KHBNb2R1bGUsIE9fUkRPTkxZLCAwTCkpID09IEVSUk9SKQ0KICAgIHsNCglwcmludGYoIkVycm9y IGluIG9wZW4oKVxuIik7DQoJcmV0dXJuKEVSUk9SKTsNCiAgICB9DQoNCiAgICAvKg0KICAgICAq IFJlYWQgdGhlIGZpbGUgZnJvbSBkaXNrIGludG8gbWVtb3J5DQogICAgICovDQogICAgcmMgPSAo U1RBVFVTKSBsb2FkTW9kdWxlKGZwLCBMT0FEX0dMT0JBTF9TWU1CT0xTKTsNCiAgICBpZiAocmMg PT0gTlVMTCkNCiAgICB7DQoJcHJpbnRmKCJFcnJvciBpbiBsb2FkTW9kdWxlKClcbiIpOw0KCXJl dHVybihFUlJPUik7DQogICAgfQ0KDQogICAgLyoNCiAgICAgKiBzbyBmYXIsIHNvIGdvb2QsIG5v dyBjbGVhbiB1cCBhbmQgbGVhdmUNCiAgICAgKi8NCiAgICBjbG9zZShmcCk7DQoNCiAgICByZXR1 cm4oT0spOw0KfQ0K --0__=h2jOornxhtdzUvLIDVafhG0Yo0ubzms4dU9GhKCtukgM5Uw3uLGfgXbl Content-type: application/octet-stream; name="vx_start_task.c" Content-transfer-encoding: base64 I2luY2x1ZGUgPHZ4V29ya3MuaD4NCiNpbmNsdWRlIDxzeW1MaWIuaD4NCiNpbmNsdWRlIDxzeXNT eW1UYmwuaD4NCiNpbmNsdWRlIDxzeXNMaWIuaD4NCiNpbmNsdWRlIDxzdGRpby5oPg0KDQojaW5j bHVkZSAidnhfZGVmaW5lcy5oIg0KDQpTVEFUVVMNCnZ4X3N0YXJ0X3Rhc2soY2hhcgkqcFN5bU5h bWUsIA0KCSAgICAgIGNoYXIJKnBUYXNrTmFtZSwgDQoJICAgICAgaW50CXRhc2tQcmlvcml0eSwg DQoJICAgICAgaW50CXRhc2tPcHRpb25zLCANCgkgICAgICBpbnQJdGFza1N0YWNrU2l6ZSkNCnsN CiAgICBGVU5DUFRSCQkqcEZ1bmN0aW9uOw0KICAgIFNZTV9UWVBFCQlTeW1UeXBlOw0KICAgIFNU QVRVUwkJRXhpdDsNCiAgICBTVEFUVVMJCXJjOw0KICAgIGNoYXIJCVRleHRbODBdOw0KDQogICAg cmMgPSBzeW1GaW5kQnlOYW1lKHN5c1N5bVRibCwgDQoJCSAgICAgICBwU3ltTmFtZSwNCgkJICAg ICAgIChjaGFyICoqKSAmcEZ1bmN0aW9uLA0KCQkgICAgICAgJlN5bVR5cGUpOw0KICAgIHNwcmlu dGYoVGV4dCwgIkVycm9yIGZpbmRpbmcgVGFzayBFbnRyeSBwb2ludDogJXMiLCBwU3ltTmFtZSk7 DQogICAgRVJST1JfQ0hFQ0socmMsIFRleHQsIEZBTFNFKTsNCiAgICBFeGl0ID0gcmM7DQoNCiAg ICBpZiAocmMgPT0gT0spDQogICAgew0KCXJjID0gdGFza1NwYXduKHBUYXNrTmFtZSwgDQoJCSAg ICAgICB0YXNrUHJpb3JpdHksIA0KCQkgICAgICAgdGFza09wdGlvbnMsIA0KCQkgICAgICAgdGFz a1N0YWNrU2l6ZSwgDQoJCSAgICAgICBwRnVuY3Rpb24sIA0KCQkgICAgICAgMEwsDQoJCSAgICAg ICAwTCwNCgkJICAgICAgIDBMLA0KCQkgICAgICAgMEwsDQoJCSAgICAgICAwTCwNCgkJICAgICAg IDBMLA0KCQkgICAgICAgMEwsDQoJCSAgICAgICAwTCwNCgkJICAgICAgIDBMLA0KCQkgICAgICAg MEwpOw0KCXNwcmludGYoVGV4dCwgIkVycm9yIGluIHRhc2tTcGF3bigpIG9mIEFQQSBDb250cm9s IHRhc2s6ICVzIiwgDQoJICBwVGFza05hbWUpOw0KCUVSUk9SX0NIRUNLKHJjLCBUZXh0LCBGQUxT RSk7DQoJRXhpdCA9IChyYyA9PSBFUlJPUikgPyBFUlJPUiA6IE9LOw0KICAgIH0NCg0KICAgIHJl dHVybihFeGl0KTsNCn0NCg== --0__=h2jOornxhtdzUvLIDVafhG0Yo0ubzms4dU9GhKCtukgM5Uw3uLGfgXbl-- From cgrames@mdc.com Fri Aug 28 13:12:51 1998 From: Charlie Grames Date: Fri Aug 28 13:12:55 PDT 1998 Subject: Booting a Large Number of Targets Simultaneously VxWorks 5.3.1 MVME2604 1.1/4 Thanks to all of you who responded to my posting about bootin [sic] a large number of targets simultaneously. In answer to the more common suggestions: o The VMEbus arbitration was already set up for round-robin and fair mode, so VMEbus contention (or starvation, at least) should not be an issue o The backplance communication method was already via mailbox interrupt (versus polling) I have found tentative relief to my problem in an unlikely way: I changed SM_TAS_TYPE in configAll.h from SM_TAS_HARD to SM_TAS_SOFT and was able to boot all boards successfully. Are there any PowerPC/MVME2604 types out there that would care to comment on this? I have some ideas, but I would like to entertain others. Thanks. Charlie Grames The Boeing Company (314) 233-1956 Charles.R.Grames@boeing.com From goren@venus.aselsan.com.tr Fri Aug 28 15:35:39 1998 From: Ozgur GOREN Date: Fri Aug 28 15:35:44 PDT 1998 Subject: DOSFS on PCMCIA (fwd) Hi vxWorkers, We are using a DOS File System on a PCMCIA disk, but the operations are too slow to meet our real time requirements. Does anybody use such a configuration and have idea about its performance? Information about the speed of vxWorks' Dos File System and optimal configuration for maximizing the file access speed are also welcome. Thanks, H.Ozgur GOREN ASELSAN MST SOFTWARE DEPARTMENT From daemon@csg.lbl.gov Sat Aug 29 04:00:27 1998 From: daemon@csg.lbl.gov Date: Sat Aug 29 04:00:30 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat Aug 29 04:00:23 PDT 1998 Subject: WINNT to vxworks over PPP Subject: Checksum problem ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: WINNT to vxworks over PPP Date: Fri, 28 Aug 1998 09:55:20 -0400 From: "Huy Vu" Organization: iSTAR internet Incorporated Message-ID: <6s6dl0$eal$1@nr1.ottawa.istar.net> Here's my follow post: I have solved this problem. Here's an outline of the solution. I took Kel Doig's code and adapted it to make my target accept both WINNT and WIN95 clients. Please lookup Dejanews for Kel Doig's article. For WINNT I listen for the string "CLIENT" and reply with the string "CLIENTSERVER". For WIN95, it's "HELLO " and . Otherwise I assume it's WINNT and reply with "CLIENTSERVER". What I noted was that I didn't get either of the "CLIENT" or the "HELLO " from a WINNT or a WIN95 client. What I got was a messy string and so my code replied with "CLIENTSERVER" and this seemed to work fine. I also found out that in order for vxworks to drop the PPP connection when the client disconnects, I must not use OPT_SILENT_MODE (as Kel poitned out) and OPT_PASSIVE_MODE. Even then, when the WINNT client disconnects, vxworks does not drop the link. If anyone knows how to fix this please do. Good luch and thank you, Huy Vu Mitel Corp. --------------------------- Newsgroups: comp.os.vxworks Subject: Checksum problem Date: 28 Aug 1998 19:18:04 GMT From: James Li Organization: Northeastern University, College of Computer Science Message-ID: <6s6vpc$75d$1@camelot.ccs.neu.edu> hi everyone, i am trying to find a way to calculate the checksum for the vxWorks executable. this executable is linked with my application and built using the command "make vxworks". after the command is executed, the programme returns the size of the text, data, and bss region. since i know the text segment size and the image is loaded to RAM starting at 0x1000 on the Motorolla MV177 computer board, i simply run my checksum utility function to calculate the checksum over this region. does this method sound simple? after i got the checksum i store the checksum in my application. i hope to run my checksum utility every time i reset and reboot my machine. after finishing storing my check- sum in a global(non-constant) variable in my application i went ahead and recompile and link my application with vxWorks. i reran my checksum utility function and compare my new checksum with my old value. i should expect a match but i didn't. i did a hex dump on the vxWorks image and found the checksum value is stored in the text portion of the image. can that be possible? isn't the text portion read-only? i expect my checksum stored in the data region of memory. Note: (1) when i built vxWorks i took advantage of the MACH_EXTRA flag in the Makefile to add my application object files and link them with the vxWorks object files allowing me to create a single vxworks image. (2) i stripped away the header information of the vxWorks before calculating the checksum. again, i used the hex dump to determine the range and size of the header information. if anyone knows an answere to this problem, please contact me. thanks, - -jim 781-455-2981 james.li@gsc.gte.com --------------------------- End of New-News digest ********************** From dlee@adti.com Sat Aug 29 09:39:27 1998 From: dlee@adti.com Date: Sat Aug 29 09:39:30 PDT 1998 Subject: VxWorks, accessing off board memory from a vme162 board I need to write to a system address at an off-board memory location. Wind River has suggested that I include an entry in the structure sysPhysMemDesc in the sysLib.c file in th bsp directory. I have done this and I am still not having any success. Can anyone clue me in. Thanks in advance for any help. From froeber@BBN.COM Sat Aug 29 16:25:00 1998 From: Fred Roeber Date: Sat Aug 29 16:25:03 PDT 1998 Subject: Re: VxWorks, accessing off board memory from a vme162 board On Sat, 29 Aug 1998, dlee@adti.com wrote: > I need to write to a system address at an off-board memory location. > Wind River has suggested that I include an entry in the structure > sysPhysMemDesc in the sysLib.c file in th bsp directory. I have done > this and I am still not having any success. Can anyone clue me in. You don't say what your target processor is but there are a couple of things that could cause you problems. First, you should use the sysBusToLocalAddr function to get the proper CPU address to use to access the VME address you desire. Not all board map VME addresses "straight through". Further, you should make sure that your CPU can address the memory address you want to since almost no CPU card allows local access to all VME addresses. You also don't say what type of access you want to make. When it comes to VME accesses you can run into restrictions based on the type of access the external card supports. The "normal" type of access is A32/D32 meaning you are using 32 bit addresses and 32 bit data. If the off-board address is on some special peripheral board then it might only support some more restricted type of access; e.g. A24D32, A24D16, or A16. Without more information it's hard to zero in more on what your problem might be. Good luck. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | TraceMaker product manager -> www.tracemaker.bbn.com | From daemon@csg.lbl.gov Sun Aug 30 04:00:25 1998 From: daemon@csg.lbl.gov Date: Sun Aug 30 04:00:28 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun Aug 30 04:00:21 PDT 1998 Subject: Re: Loadlin equivalent for VxWorks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loadlin equivalent for VxWorks Date: Fri, 21 Aug 1998 16:36:56 GMT From: john@kivala.com Organization: Kivala Systems, Inc. Message-ID: <6rk7n9$ppf$1@nnrp1.dejanews.com> References: <6rh7hd$ol5$3@news.cegelecproj.co.uk> <6ri8fq$caa$1@nnrp1.dejanews.com> <6rjj6l$bl3$1@news.cegelecproj.co.uk> In article <6rjj6l$bl3$1@news.cegelecproj.co.uk>, kevin_taylor@cegelecproj.co.uk (Kevin Taylor) wrote: ... > Does Windows query the BIOS then for the parameters (not > really familiar with PnP) ? ... At least some (can't say all) of these jumperless PNP cards save their stuff in their own NVRAM. Sometimes they come with a utility program that lets you set the "jumpers" manually, and you have to do that to know how to configure VxWorks. Such a program is needed if you want to run the card with Win 3.1, for example. John - -------------------------------------------------- John Finley Kivala Systems, Inc. Project Manager (619) 689-0032 john@kivala.com http://www.kivala.com - -------------------------------------------------- VxTool - Build Tornado Apps with MS Visual Basic - -------------------------------------------------- - -----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum --------------------------- End of New-News digest ********************** From daemon@csg.lbl.gov Mon Aug 31 04:00:15 1998 From: daemon@csg.lbl.gov Date: Mon Aug 31 04:00:19 PDT 1998 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon Aug 31 04:00:12 PDT 1998 Subject: CFP: IEEE PLRTIA'98 workshop on Programming Languages for Real-Time Industrial Applications -- comp.os.vxworks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: CFP: IEEE PLRTIA'98 workshop on Programming Languages for Real-Time Industrial Applications -- comp.os.vxworks Date: 31 Aug 1998 08:55:43 GMT From: hips@informatik.hu-berlin.de (HIPS) Organization: Humboldt University Berlin, Department of Computer Science Message-ID: <6sdoef$55q$1@hahn.informatik.hu-berlin.de> ============================================================================= Call for Papers IEEE Workshop on Programming Languages for Real-Time Industrial Applications -- Is Java the Future? -- Madrid, Spain, December 1, 1998 (in conjunction with RTSS'98) http://www.informatik.hu-berlin.de/~mueller/plrtia98 ============================================================================= IMPORTANT DATES: Papers due: Oct 10, 1998 Author notification: Oct 31, 1998 Final papers due: Nov 14, 1998 DESCRIPTION: Over the last 30 years there has been a plethora of languages used for the implementation of industrial and military real-time and embedded systems. Early languages such as Jovial, Coral 66, and RTL/2 have been replaced by C/C++, Pearl, Ada and Esterel. Whilst Ada has a niche in the safety critical systems market, there has been wide-spead use of C/C++. This dominance has been reinforced by the developement of the real-time POSIX standards. However, all these languages have their critics. For some, Ada and POSIX are too complicated, for others C/C++ are too insecure. More recently, Java has emerged as a candiate language which claims a compromise between the safety and security of Ada, and yet the flexibility and popularity of C/C++. However, Java lacks some of the facilities required to program hard real-time systems, and its implementation lacks the predictablity needed for safety critical applications. This workshop will give the opportunity to researchers, developers, and practitioners working in related areas to exchange experience on the design, implementation and use of programming language in industrial and military applications. Topics of submissions include but are not limited to: * progress reports on the developement and use of Java * real-time java virtual machines * evaluation of Java for real-time * Ada 95 real-time profiles (subsets) * lessons learned from the use of current (C/C++, Pearl, Ada and Esterel) languages * meeting the demands of real-time distributed programming * the impact of future systems requirements on programming languages * the role of real-time POSIX SUBMITTING PAPERS: Authors are invited to submit regular workshop papers as well as proposals for panel discussions. Paper submissions should be limited to 10 pages, double space, including figures. The title page should include a maximum 150-word abstract, five keywords, full mailing address, e-mail address, phone number, fax number, and a designated contact author. All submissions must be electronic (postscript) and sent to: Alan.Burns@cs.york.ac.uk. GENERAL CHAIR: M. Gonzalez Harbour Universidad de Cantabria Departamento de Electronica y Computadores Avda. de Los Castros s/n 39005 Santander, Spain phone: +34-942-2014-83, fax: -02
mgh@ctr.unican.es CO-PC-CHAIRS: Alan Burns Frank Mueller University of York Humboldt University Berlin Department of Computer Science Institut fuer Informatik Heslington Unter den Linden 6 York, YO10 5DD, U.K. 10099 Berlin, Germany phone: +44 (1904) 4327-79, fax: -67 phone: (+49) (30) 2093-3011, fax: -3010 Alan.Burns@cs.york.ac.uk mueller@informatik.hu-berlin.de PROGRAM COMMITTEE: Ted Baker, Florida State Univeristy, USA Brad Balfour, SEI, USA Riccardo Bettai, Texas A&M Univeristy, USA Greg Bollella, IBM, USA Giorgio Butazzo, University of Pisa, Italy David E. Emery, Mitre, France Franco Gasperoni, Ada Core Technologies, USA Richard Gerber, University of Maryland, USA Hans Hansson, Mdlardalens Hvgskola, Sweden M. Gonzalez Harbour, Universidad de Cantabria, Spain Farnam Jahanian, University of Michigan, USA Andy Johnson, Mercury Computer Systems, USA Jochen Liedtke, IBM Watson, USA Jane W.-S. Liu, University of Illinois at Urbana-Champaign, USA Sang Lyul Min, Seoul National University, Korea Frank Mueller, Humboldt Universty Berlin, Germany Kelvin Nilsen, NewMonics, USA Lynne Rosenthal, NIST, USA Manas Saksena, Concordia University, Canada Lui Sha, SEI, USA Daniel Simon, INRIA, France Sanjay Sjinturkar, Lucent, USA Tucker Taft, Intermetrics, USA Hideyuko Tokuda, Keio University, Japan Serdar Uckun, Rockwell Science Center, USA Andy Wellings, University of York, U.K. Vic Wolfe, University of Rhode Island, USA --------------------------- End of New-News digest ********************** From prb@gazelle.jhuapl.edu Mon Aug 31 09:42:24 1998 From: "Paul R. Bade" Date: Mon Aug 31 09:42:28 PDT 1998 Subject: Jaz Drive Hi, Does anybody have any experience connecting a Jaz driver to a vxWorks target (such as a Motorola VME board)? Is it pretty much plug and play? Thanks, +====================================================================+ | __ ____ __ __ | | /\ \ / \ \ /\_\__ /\ \ Johns Hopkins University | | / \_\ | /\ \ \ / / /\_\\ \ \ Applied Physics Lab. | | / /\ | |\ \/ \ \ / / / / / \ \ \ | | / \/ |_| \ /\ \_\ / / / / / \ \ \ Paul R. Bade | | / /\__/_/ \ \ \/_// / / / / / / / (443)-778-8681 | | / / / \ \_\ \ \/ / / / / / bade@jhuapl.edu | | \/_/ \/_/ \__/_/ \/_/ | | __ ____ __ __ | | /\ \ / \ \ /\ \ /\ \ | | / \_\ | /\ \ \ / \_\ / \ \ | | / /\ | |\ \/ \ \ / /\ |_| / /\ \ \ | | / \/ |_| \ /\ \_\ / / / | | \/ / \ \ | | / /\ / / \ \ \/_// / / / / / /\ \_\ | | \ \/ / / \ \_\ \ \/ / / \/ / / / | | \__/_/ \/_/ \__/_/ /_/_/ | | | +====================================================================+ From Phil_Watson@masirv.com Mon Aug 31 09:42:35 1998 From: Phil_Watson@masirv.com (Phil Watson) Date: Mon Aug 31 09:42:38 PDT 1998 Subject: Re: comp.os.vxworks newsdigest > i did a hex dump on the vxWorks image and found the checksum value > is stored in the text portion of the image. can that be possible? > isn't the text portion read-only? i expect my checksum stored in the > data region of memory. Yes, indeed. The initial value of all initialized variables is stored in the text segment. We have done something like you intend before by forcing the checksum to a predetermined value by setting the variable to zero, calculating the checksum, then setting the variable to a value which would force the checksum to the known value and rebuilding. A final checksum calculation should yield the desired value. -- vxWorks -- HTH, /PSW Philip Watson, Phil_Watson_nojunk@masirv.com Manager, Embedded Software Development Matsushita Avionics Systems Corp. 15253 Bake Parkway Irvine, California, 92618 U.S.A. From philmoy@spawar.navy.mil Mon Aug 31 13:58:03 1998 From: Phil Moy Date: Mon Aug 31 13:58:06 PDT 1998 Subject: VxWorks Interrupt problems with MOT 2603 Does anyone have any problems using VME interrupts with the Mot MVME2603 board? I'm upgrading from the Mot MVME167 to the 2603 and am having some problems with vme interrupts. I fixed one problem in universe.c where odd interupt vectors did not work. But I still am having problems where a Simpact 6000 board (vme based octal serial board) generates the wrong Status/ID to the IACK intermittently. It usually runs well most of the time but an occasional bad interrupt vector (usually 0xff) is generated. This only occurs if there is other vme bus activity at the same time. Any help or ideas would be greatly appreciated. -Phil Moy From scheurer@terra1.tz.rus.uni-stuttgart.de Mon Aug 31 14:26:49 1998 From: "F. Scheurer" Date: Mon Aug 31 14:26:53 PDT 1998 Subject: Configuration of MVME2700 under TORNADO Hi VxWorkers, i still got my problem configuring a kernel (VxWorks 5.3.1) fitting our needs (shell, NFS, telnet, etc.). On my last posting, i got a few useful hints but i even can't build the Default Configuration, that comes with TORNADO, through WindCFG. BUILD results in the following: > 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\"" *.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 > 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\"" 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 > 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\"" 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 > 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\"" /usr/local/wind/target/config/all/usrConfig.c -o usrConfig.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\"" /usr/local/wind/target/config/all/dataSegPad.c -o dataSegPad.o > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -mstrict-align -ansi -nostdinc -O2 -fvolatile -fno-builtin -fno-for-scope -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\"" -P -x assembler-with-cpp -c -o sysALib.o sysALib.s > /var/tmp/ccTBUn2_.s: Assembler messages: > /var/tmp/ccTBUn2_.s:4481: Warning: operand out of range (-32768 not between 0 and 65535) > ccppc -B/usr/local/wind/host/sun4-solaris2/lib/gcc-lib/ -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\"" -c sysLib.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\"" -o version.o /usr/local/wind/target/config/all/version.c > ldppc -X -N -X -r \ > -o vxWorks.tmp sysALib.o sysLib.o usrConfig.o version.o /usr/local/wind/target/lib/libPPC604gnuvx.a > nmppc vxWorks.tmp | 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' > 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\"" -c ctdt.c > make[1]: Leaving directory `/usr/x4/TORNADO_1.0.1/target/config/mv2700' > ldppc -X -N -e _sysInit -Ttext 00100000 \ > -o vxWorks dataSegPad.o vxWorks.tmp ctdt.o > vxWorks.tmp: In function `dcStartOutput': > vxWorks.tmp(.text+0x301cc): undefined reference to `m_freem' > vxWorks.tmp: In function `looutput': > vxWorks.tmp(.text+0x3e3f6): undefined reference to `mfree' > vxWorks.tmp(.text+0x3e3fa): undefined reference to `mfree' > vxWorks.tmp(.text+0x3e446): undefined reference to `mbstat' > vxWorks.tmp(.text+0x3e44a): undefined reference to `mbstat' > vxWorks.tmp(.text+0x3e44e): undefined reference to `mbstat' > vxWorks.tmp(.text+0x3e452): undefined reference to `mbstat' > vxWorks.tmp(.text+0x3e456): undefined reference to `mbstat' > vxWorks.tmp(.text+0x3e45a): more undefined references to `mbstat' follow > vxWorks.tmp: In function `looutput': > vxWorks.tmp(.text+0x3e496): undefined reference to `mfree' > vxWorks.tmp(.text+0x3e49a): undefined reference to `mfree' > vxWorks.tmp(.text+0x3e4d0): undefined reference to `m_more' > vxWorks.tmp(.text+0x3e598): undefined reference to `m_freem' > vxWorks.tmp(.text+0x3e654): undefined reference to `m_freem' > make: *** [vxWorks] Error 1 > Press Return to Quit > 1. Perhaps it is the compilation of the sysALib.o, because a warning occurs. I don't have any idea at the moment. 2. It seems as if there would be missing an object-file? 3. Is it the environment that has to be setup? Could there be something set wrong or any file unreachable? For Your information: 1. I installed from the new PPC-CD 07, dated August 1998. 2. We use a SUN-Host running Solaris 2.6. Anyone has an idea, what's the problem? TIA F. Scheurer -- +-------------------------------------------------+ | | | Dipl.-Ing. F. Scheurer | | | | c/o Zentrum Fertigungstechnik Stuttgart | | Arbeitsbereich Steuerungstechnik | | Nobelstrasse 15 | | D-70569 Stuttgart | | | | Telefon +49 711 13162-41(-0) | | Fax +49 711 13162-11 | | E-Mail scheurer@zfs.rus.uni-stuttgart.de | | zfssrr@zfssun6.tz.rus.uni-stuttgart.de | +-------------------------------------------------+ | http://zfs.rus.uni-stuttgart.de | +-------------------------------------------------+ | | | privat: | | Telefon +49 7161 686362 | | Fax +49 7161 686301 | | E-Mail Friedrich.Scheurer@t-online.de | | | +-------------------------------------------------+ From sbatchu@x.Holontech.com Mon Aug 31 19:21:00 1998 From: Suresh Batchu Date: Mon Aug 31 19:21:03 PDT 1998 Subject: Telnet deamon question Hi VxWorkers, I have a question regarding the telnet deamon in VxWorks. My application needs the telnet deamon to be shutdown and restarted. Since there is no api for shutting down the telnet deamon I am just using the task delete to delete the telnet task. Problem comes when I try to restart the telnet deamon using "telnetd", I get the error message saying "bind failed". I am unable to restart the telnet server. could someone tell me how to restart the telnet deamon. Thanks in advance, Suresh. From gkraemer@netzone.com Mon Aug 31 21:36:25 1998 From: Garry Kraemer Date: Mon Aug 31 21:36:28 PDT 1998 Subject: Re: Jaz Drive Paul, I have an IOMEGA ZIP drive operating with a MVME2604 Power PC (200MHZ). WRS had to send me several updated files before I was successful in getting vxWorks to talk to the ZIP drive. I suspect the same updated files would be required for the JAZ drive. I initialized the ZIP disk using vxWorks and so far the format does not appear to be 100% compatible with PC running Windows 95. Once again WRS sent a lot of information on disk partitioning and I'm still trying to digest it. Garry