From vxwexplo-errs@csg.lbl.gov Tue May 2 09:35:40 2000 From: "Vidhya Raman" Date: Tue May 2 09:35:43 PDT 2000 Subject: VxWorks flash driver question Hi VxWorkers, I am trying to get the VxWorks flash driver (flashMem.c) to work for am29F040 flash chip on a ads860 board. I am unable to write to the flash memory. Is there something I am missing ? Do I have to "write enable" the flash memory. The state machine for flash programming on am29F040 is correct in the flashMem.c code. Thanks Raman ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com From vxwexplo-errs@csg.lbl.gov Tue May 2 20:17:17 2000 From: Carlos Rimola Date: Tue May 2 20:17:19 PDT 2000 Subject: traceroute Has anyone done a port of traceroute to vxWorks? If there is no such a port, I will likely do it (time permitting). Send me a message if you know of a port or are interested in one.. Thanks, Carlos Rimola crimola@praxon.com From vxwexplo-errs@csg.lbl.gov Wed May 3 01:14:20 2000 From: Leonid Rosenboim Date: Wed May 3 01:14:22 PDT 2000 Subject: RealServer ported to VxWorks Has anyone used or knows of a port of RealNetworks' RealServer G2 to VxWorks ? More specifically we are interested in any experience using this software. Thanks in advance, Leonid From vxwexplo-errs@csg.lbl.gov Wed May 3 04:55:25 2000 From: "Tyler, Paul A (DBY)" Date: Wed May 3 04:55:27 PDT 2000 Subject: Re: Interfacing to Allen-Bradley PLCs Henry, PLC-5/VME can map some of it's slave RAM to the VME bus. It is fairly straightforward from there. My colleagues have used this successfully with Intel based processors from VMIC, but Motorola shouldn't cause any problems. For the PLC-5/20E, there is nothing available as yet from Allen-Bradley. The Linux drivers from Ron Gage look as if they should port fairly easily. I looked at these for interfacing to SLC-500's, but haven't had a project with which to develop the ideas any further. Look for abplc5 at www.freshmeat.net. If you find any other info I would like to know. VxWorks, Tornado Paul Tyler Measurement Systems Specialist Electronics & Measurement Systems Dept. Rolls-Royce plc. e-mail Paul.A.Tyler @ Rolls-Royce.com Tel. +44 (1332) 247808 Fax. +44 (1332) 247928 Henry Valtier wrote in message news:<390F2084.7F7CECE9@jpl.nasa.gov>... > Hello All, > > We are contemplating the use of an Allen-Bradley PLC-5/VME > programmable logic controller in a VME chassis along side > a Motorola MVME2700 PowerPC-750, running VxWorks 5.4. > Another option would be an A-B PLC-5/20E (with embedded > ethernet TCP/IP communication capabilities ), external to > the VME (VxWorks) chassis but on the same local area network. > > Does anyone know of any code that I can download to study and > learn how to interface (ethernet or serial) with these types > of PLCs? Are there any sources of information (besides > Allen-Bradley) on interfacing to these PLCs via VxWorks? > > Any help would be greatly appreciated. > > Thanks in advance, > Henry Valtier > > Henry.Valtier@jpl.nasa.gov > From vxwexplo-errs@csg.lbl.gov Wed May 3 11:14:01 2000 From: johill@lanl.gov (Jeff Hill) Date: Wed May 3 11:14:03 PDT 2000 Subject: thread safe C++ exceptions All, Can anyone comment on whether C++ exceptions are really thread safe in Tornado II? Its difficult to determine which egcs version of g++ is included in Tornado II, and I understand that the thread safeness of C++ exceptions was not addressed in earlier versions at egcs. Jeff ______________________________________________ Jeffrey O. Hill E-mail johill@lanl.gov LANL MS H820 Voice 505 665 1831 Los Alamos NM 87545 USA FAX 505 665 5107 From vxwexplo-errs@csg.lbl.gov Fri May 5 06:18:15 2000 From: "gaju more" Date: Fri May 5 06:18:17 PDT 2000 Subject: Subscription for Exploder mails Hi Kindly mail me future mails regarding the exploder issues for VxWorks. Regards G S More _________________________________________________ Get Your Free Email At, http://www.rediffmail.com From vxwexplo-errs@csg.lbl.gov Tue May 9 04:03:26 2000 From: Vxworks Exploder Date: Tue May 9 04:03:28 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue May 9 04:03:21 PDT 2000 Subject: Re: SLIP question Subject: I do not think anybody has done this before ?? Subject: Re: UDP broadcast 404 bytes ok, but not 564 Subject: Re: VxWorks card servicing 'pings' from the network Subject: Redirect printf Subject: passing function pointers between tasks Subject: Re: Changing MAC Address Subject: Re: Varargs on PPC under 5.4 using Tornado 2.0... Subject: Can someone tell me if Linux can telent to VxWorks? Subject: Re: Can someone tell me if Linux can telent to VxWorks? Subject: usrMmuInit() call on PIII Subject: Re: Changing MAC Address Subject: Re: Changing MAC Address Subject: Bootrom and Vxworks in non-contiguous memory Subject: correct local address in VxSim? Subject: 2nd Call for Participation - Ada Europe 2000, Invited Speaker and Subject: Re: Reducing vxWorks kernel footprint Subject: Re: Bootrom and Vxworks in non-contiguous memory Subject: Re: lnPci driver ring buffer size Subject: Re: Slow GCC compiler in Tornado II Subject: trouble with TFFS_DEFRAGMENT_VOLUME Subject: Re: X-Window client/server for VxWorks? Subject: Need Some Help in SCC z85c30 Subject: Re: Retrieve system date and time Subject: Re: X-Window client/server for VxWorks? Subject: Re: RAM_HIGH_ADRS Subject: Tornado 2 on Windows 2000? Subject: Re: Multiple ethernet interfaces Subject: Anybody have a good shareware memory mgr? Subject: Re: Need Some Help in SCC z85c30 Subject: Re: Flash ROM Problem with PowerPC mv2700 Subject: Stethoscope 5.(latest) Subject: Re: smNetShow Subject: VxWorks Flash Driver Question Subject: Re: Multiple ethernet interfaces Subject: Re: Anybody have a good shareware memory mgr? Subject: VxWorks for Newbies Subject: Re: boot failure question Subject: Re: Retrieve system date and time Subject: Re: Labview/NT <--> vxWorks Subject: Re: monitoring packets Subject: Re: Is it ok to use Tornado together with another RTOS and network stack? Subject: Re: Source MAC address being overwritten for non-IP traffic Subject: Re: asPPP under Solaris 2.6 Subject: SLIP question ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: SLIP question Date: Wed, 26 Apr 2000 11:52:13 GMT From: S Austin Organization: Deja.com - Before you buy. Message-ID: <8e6l99$oib$1@nnrp1.deja.com> References: <3905B69E.6C59@gecm.com> Hey David, > > Now I want to transfer date through the serial port from the PC to the > card. Is SLIP the way to go on this? Or PPP which is a bit more modern and efficient (with header compression, the net bandwidth is higher for a given bit rate). > I've tried following the instructions in TSR55352 (below) > but it fails to accept a direct route in the routeAdd part > as the card (2) already has a route through card (1) to the host PC > which I need to get the shared memory to work. Sometimes it's easier to use different subnets between your serial and Ethernet connections, then manipulate the routing table accordingly. Then your existing routes will continue to exist, and the routing table is augmented with the new routes. For example, if your existing connections are on 10.0.1.x, make the serial IP connections on 10.0.2.x and define a netmask such that the routes don't overlap. So say your target SLIP IP is 10.0.2.1 and the PC's dial-up port is set to 10.0.2.2, you can send data between those two addresses without interfering with the existing connection on, say, 10.0.1.43 for the target and 10.0.1.55 for the PC. In other words, the way to decide which physical connection a packet will traverse is through the subnetting and routing tables. > Also apart from having SLIP set up in dial-up networking on the PC, > is there something I'm missing trying to get an 'ls' on the card to > work. What on the PC should tell it to respond over the serial port? > From your host shell you can redirect the target's stdio to whatever target ports you want using ioGlobalStdSet or ioTaskStdSet. Typically people redirect it to a vio port so it shows up on the Tornado virtual console, but you can certainly redirect it to a tty port. Here's a little tcl script we use to redirect stdio to the console: ? if { [shParse {tstz = open ("/vio/0",2,0)}] != -1 } { shParse {vf0 = tstz}; shParse {ioGlobalStdSet (0,vf0)} ; shParse {ioGlobalStdSet (1,vf0)} ; shParse {ioGlobalStdSet (2,vf0)} ; shParse {logFdAdd (vf0)} ; shParse {printf ("Std I/O set here!\n")} } else { shParse {printf ("Std I/O unchanged.\n")} } ? Hope this helps! Stuart Austin Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: I do not think anybody has done this before ?? Date: Wed, 26 Apr 2000 12:01:24 GMT From: "Harjot Saluja" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: Anybody written MTDs for TFFS support for 28F400B1 flash device ?? I am facing problem writing that. 28F400 is mentioned as obsolete in Micron tech. data sheets, so I am wondering if anybody wrote any drivers for that ??? - -- harjot 203-756-7105 (home) 203-264-1888 exn 129 (work) harjot.saluja@kenetec.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP broadcast 404 bytes ok, but not 564 Date: Fri, 21 Apr 2000 23:18:05 GMT From: Charlie Grames Organization: Boeing Message-ID: <3900E1AD.1891EFAD@nowhere.com> References: <06180e24.025cdc0d@usw-ex0109-068.remarq.com> <38FF24E5.1B58DD49@nowhere.com> <0326f2a7.0bc9d9fc@usw-ex0109-068.remarq.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Brian, Perhaps I am misunderstanding your use of the term "broadcast." Are you doing a true IP broadcast (e.g., to address 90.0.0.255), or are you just transmitting a message from A to B (and/or back)? IP broadcasts cannot be fragmented. Point-to-point IP transmissions, however, may. You mentioned that the Ethernet MTU is 1518. Are you getting this number by using ifShow()? I would expect the value shown by ifShow() to be 1500 (including IP and UDP headers, but excluding Ethernet header) for the traditional Ethernet interface. It might be helpful, as well, to see the code you are using to transfer the data. We routinely use UDP to transfer messages as large as 32000 bytes between VxWorks targets without trouble. (A sign error in the VxWorks implementation of the 4.3 BSD stack prevents transfer of messages with a total length greater than 32767 bytes.) Because of this, I suspect the information you found in netinet/tcp.h does not apply to UDP. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks card servicing 'pings' from the network Date: Wed, 26 Apr 2000 16:28:35 GMT From: Charlie Grames Organization: Boeing Message-ID: <39071933.C216519D@nowhere.com> References: <390702C4.28F5DBC1@lmco.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Eric, Ping uses ICMP Echo and Echo Reply. I don't have specifics, but I would imagine that the overhead of receiving an Echo packet and preparing an Echo Reply would be less than that of receiving a single UDP packet and replying to it. As long as the frequency of your pings is relatively low, I don't see that using them should pose a problem. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com --------------------------- Newsgroups: comp.os.vxworks Subject: Redirect printf Date: Wed, 26 Apr 2000 17:00:19 GMT From: Stephane Bellemare Message-ID: <390720AF.4D9CFA7B@c-cube.com> I'd like to redirect all printf to a message Q in order to avoid the use of the serial port or the Tornado shell . A task will read the msgQ and the data will be sent to the network using VxDCOM through Connection points. Is there a simple solution in order to redirect these printf to a msgQ? Stephane. --------------------------- Newsgroups: comp.os.vxworks Subject: passing function pointers between tasks Date: Wed, 26 Apr 2000 11:47:20 -0500 From: "Srikanth Vedire" Organization: Motorola CIG Message-ID: <8e76io$1n31@nntp.cig.mot.com> Hi, Assume that there is a function foox() within the context of TaskA. Is it possible to send the pointer to foox() to another TaskB in VxWorks? I've never worked on real time os' before. All my experience is on unix. On unix, passing function pointer's would be a no no..... Thanks in advance, Srikanth --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: Wed, 26 Apr 2000 13:08:03 -0400 From: Henry Gessau Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: <188egs4f0asa5a52rajm1e257vg8rql7cd@4ax.com> References: <8e2lvs$kin$1@nnrp1.deja.com> <8e42n3$ol$1@nnrp1.deja.com> On Tue, 25 Apr 2000 12:23:05 GMT, S Austin wrote: > In article <8e2lvs$kin$1@nnrp1.deja.com>, > arnel@vina-tech.com wrote: > > Anyone know how to change the Ethernet MAC Address from the command > > line? Is it possible? Does anyone know where the MAC address is set? > > > I believe it is set in hardware by the Ethernet board manufacturer and > can't be changed by software. Normally, yes. But the QUICC-based processors (360 & 860 family) from Motorola are a whole different kettle of fish when you configure one of the SCC's for Ethernet. You essentially become an Ethernet device manufacturer. You need to have a good plan in place for handling MAC addresses. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Varargs on PPC under 5.4 using Tornado 2.0... Date: 27 Apr 2000 14:57:39 GMT From: fnordz@aol.com (Fnordz) Organization: AOL http://www.aol.com Message-ID: <20000427105739.09492.00000013@ng-cn1.aol.com> References: Hi Don, The PowerPC cross-compiler in Tornado 2.0 does not handle function calls exactly as specified in the sysv-abi. As a result, while the varargs macros were written in anticipation of finding arguments in sysv-abi specified locations, the compiler will sometimes fail to put them in those locations. There is a patch on WindSurf which fixes this problem partially. The patch installs a new version of the appropriate varargs macros. However, depending on the code, there may be some additional cases not captured in the patch. Download the patch and try it out - you may find that the patch will at least work for your debug routines. - -paul --------------------------- Newsgroups: comp.os.vxworks Subject: Can someone tell me if Linux can telent to VxWorks? Date: Thu, 27 Apr 2000 15:29:18 GMT From: Harry Rarig Message-ID: <39085DCF.48355E34@tellium.com> I am telneting from a Linux box (RedHat 6.1) into a VxWorks embedded processor running version 5.4. Whenever I enter a command, e.g., "ifShow", part of the response will print out, then the entire session hangs. The only way to recover is to use the telnet escape sequence and close the connection. Telneting in from an NT 4.0 box works fine--this problem only seems to plague Linux. Question: Has anyone out there ever telnet'd into VxWorks using RedHat Linux (in my case, RH 6.1)? Does it work? Any known side effects? My brief experience is telling me that this Linux/VxWorks relationship needs some serious counseling...;-) Much tnx, Harry - -- Harry Rarig _/ _/ _/ _/_/_/ _/_/_/ _/ _/ Tellium Inc. _/_/_/ _/_/_/ _/_/_/ _/_/_/ _/ 732 923-4249 Voice _/ _/ _/ _/ _/ _/ _/ _/ _/ 732 923-9804 Fax ___________________________________/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can someone tell me if Linux can telent to VxWorks? Date: Thu, 27 Apr 2000 11:39:22 -0400 From: Charles H. Chapman Organization: Dynacs Engineering Company Message-ID: <8e9n0d$7sj@niven.ksc.nasa.gov> References: <39085DCF.48355E34@tellium.com> Reply-To: chc@nasa2.ksc.nasa.gov Harry Rarig wrote: > I am telneting from a Linux box (RedHat 6.1) into a VxWorks embedded > processor running version 5.4. Whenever I enter a command, e.g., > "ifShow", part of the response will print out, then the entire session > hangs. The only way to recover is to use the telnet escape sequence and > close the connection. Telneting in from an NT 4.0 box works fine--this > problem only seems to plague Linux. > > Question: Has anyone out there ever telnet'd into VxWorks using RedHat > Linux (in my case, RH 6.1)? Does it work? Any known side effects? It works great for me here. I'm using RH 6.2 and VxWorks 5.3 but I didn't have any problems telneting to VxWorks with previous versions of RH, all the way back to 4.2. Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: usrMmuInit() call on PIII Date: Thu, 27 Apr 2000 09:30:52 -0700 From: Jason Leong Organization: BCnet Message-ID: Hi VxWorkers, I'm writing a driver for a board that has a PCI-PCI bridge on it. After I find the bridge chip, I do a sysMmuMapAdd() call to add a mapping to the MMU and then call usrMmuInit() to re-initialize the MMU. On a Pentium-based target computer, the usrMmuInit() call happens quickly, but on a Pentium III target, the call takes up to 1 minute and on a Pentium II target, the call doesn't seem to work at all. Has anyone else ran into these problems? Thanks in advance. Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: Thu, 27 Apr 2000 15:04:57 -0400 From: Henry Gessau Organization: Ye 'Ol Disorganized NNTPCache groupie Message-ID: References: <8e2lvs$kin$1@nnrp1.deja.com> <8e42n3$ol$1@nnrp1.deja.com> <188egs4f0asa5a52rajm1e257vg8rql7cd@4ax.com> On Wed, 26 Apr 2000 14:59:16 -0500, "David Spencer" wrote: > Not assigning a MAC address is just plain lazy. There is a lot of laziness out there. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: Thu, 27 Apr 2000 15:52:45 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <39089A8D.46075CD8@lucent.com> References: <200004261507.QAA28925@dsl-2.tadpole.co.uk> David, Thanks for the bit definition for locally administered addresses. I was using that beast two years back, on a different product, and had forgotten which bit made it local. This is a very useful way to develop drivers without having to go through the problems of procuring assigned addresses. Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Bootrom and Vxworks in non-contiguous memory Date: Thu, 27 Apr 2000 14:34:15 -0500 From: "Jack" Organization: Motorola Message-ID: <8ea4nm$bp7$1@schbbs.mot.com> Does anyone know how to put the VxWorks image, vxworks_rom, in a different Flash than the bootrom is in? I need to boot from one Flash and have it "jump" to the vxworks image in another Flash. I'm using the PPC8260. Jack --------------------------- Newsgroups: comp.os.vxworks Subject: correct local address in VxSim? Date: Thu, 27 Apr 2000 15:58:13 -0400 From: "John Hogg" Organization: Rational Software Message-ID: My VxSim application uses the IP addresses of packets sent to it for connection auditing. According to Table H-4 on page 556 of the VxWorks Programmer's Guide (5.4, Edition 1), the local IP address (i.e., the IP address of the host on the VxSim network) should be the same as the host IP address (i.e., the IP address of the machine as it is known to the outside world). Unfortunately, packets from the host application to the VxSim application have the sender address 90.0.0.254, the address configured for the ULIP Virtual Adapter as specified in the release notes. This breaks things. Does anybody know how I can make the local IP address of the host the same as the host IP address? In particular, how do I ensure that the host->vxsim packets carry the correct address? (The VxSim application can reach the host application through either 90.0.0.254 or the host IP address.) Some context: Tornado2, NT host, full VxSim networking. Cheers -- John John Hogg Rational Software --------------------------- Newsgroups: comp.os.vxworks Subject: 2nd Call for Participation - Ada Europe 2000, Invited Speaker and Tutorial information Date: Sun, 23 Apr 2000 21:04:00 +0200 From: "Hubert B. Keller" Organization: Forschungszentrum Karlsruhe GmbH Message-ID: <39034920.23C17FE5@iai.fzk.de> Reply-To: keller@iai.fzk.de Please excuse multiple copies. ###################################################################################### 2nd Call for Participation - Invited Speaker and Tutorial information 5th International Conference on Reliable Software Technologies Ada-Europe 2000 June 26-30, 2000Potsdam (Berlin), Germany ************************************************************************************* For up-to-date information on the programme, the tutorials, invited speakers, and daily schedules, please visit http://www.ada-europe.org/conference2000.html To get a written copy of the Advance Programme, please send e-mail to: keller@iai.fzk.de ************************************************************************************* The 5th International Conference on Reliable Software Technologies (Ada-Europe 2000) will take place in Potsdam (near Berlin), Germany, from June 26 to 30, 2000. The conference offers a technical programme and exhibition from Tuesday to Thursday, and a series of tutorials on Monday and Friday. The conference provides an international forum for researchers, developers and users of reliable software technologies. Presentations and discussions cover applied and theoretical work currently conducted to support the development and maintenance of software systems. The technical programme includes keynote addresses, session papers by the international community with refereed contributions from Argentina, Australia, Austria, Canada, China, France, Germany, Italy, Japan, Norway, Spain, Sweden, Switzerland, United Kingdom and the USA, additional short presentations on various interesting topics, and two sessions, in which vendors present their commercial products. The proceedings of the conference will be published in the LNCS Series by Springer. An exhibition concurrent to the conference provides an opportunity to explore the latest developments by the commercial marketplace. The tutorials on Monday and Friday offer an excellent opportunity for obtaining in-depth knowledge in important technologies of the field. ****************************************************************************************** ++++++++++++ Invited Speakers +++++++++++++++ Tuesday Support of technology development in the Information Society Rainer Zimmermann Head of Unit E2, European Commission, DG Information Society, Brussels Wednesday Kingcat MCAS - Monitoring, Control and Alarm System for a luxury motor yacht implemented in Ada and Java Reto Weiss Kingcat MCAS Project Manager/Software Engineer, Paranor AG, Switzerland Thursday Ada after 10 years of usage - is there a commercial future? Bryan Pflug Chief Engineer, Simulation & Software Engineering, Boeing Commercial Airplanes, Seattle, USA ********************************************************************************************* ++++++++++++++ Turorials +++++++++++++ Tutorial 1: Brosgol, Ben, Java for Ada Programmers (full day, intermediate) This tutorial examines the Java language and assesses how it meets the goals of portability, adaptability and safety, using comparisons with Ada in presenting the semantics. The main focus of the tutorial will be on Java's Object-Oriented Programming features, since these are the underpinnings of almost the entire language as well as the API. Tutorial 2: Gonzalez Harbour, Michael, Real-Time POSIX, (full day, introductory) This tutorial discusses the main real-time operating system services defined in the POSIX standards. These services allow application developers to write portable applications that meet their real-time requirements. The tutorial reviews the different services from the perspective of an Ada application, and discusses which of the different options and services are more suitable for developing real-time applications in Ada. Tutorial 3A: Cook, David and Dupaix, Les, Ada95 for Beginners, (half day, introductory) Tutorial 3B: Cook, David and Dupaix, Les, Ada95 for Ada83 Programmers, (half day, introductory) (Depending on actual requests for tutorial 3A, tutorial 3B may be moved to the morning session) These two half-day tutorials will introduce attendees to the capabilities of the Ada 95 programming language. It will cover the basic syntax of the core (Ada 83) language in the morning, and then go on to show the improvements and enhancements of Ada 95. If you already know Ada 83 and just wish to learn about the Ada95 enhancements, then you do not need to attend the morning portion of the tutorial. Tutorial 4: Hausen, Hans-Ludwig, Software Metrology Basics - Measurement, Assessment and Certification of Procedures, Objects and Agents, (half day, intermediate) The overall aim of the tutorial is to make the attendees familiar with the methods and principles of software metrics for procedural, object-oriented and agent-based systems. Attendee will exercise proven techniques for goal directed measurement, scaling and assessment as part of an industry proven, standardized procedure for concurrent software quality assurance and final evaluation for certification. Tutorial 5: Rosen, Jean-Pierre, The HOOD 4 design method (half day, intermediate) HOOD (Hierarchical Object Oriented Design) is a software design method, which is used after the requirement analysis activities and covers architectural design, detailed design and coding. The method unifies and integrates object orientation with advanced software engineering concepts and notations. Tutorial 6: Tian, Jeff, Tree-Based Reliability Models (TBRMs) for Early Reliability Measurement and Improvement, (half day, intermediate) This tutorial surveys recent developments in software reliability engineering, particularly recent work in using tree-based reliability models (TBRMs) in analysing product reliability and identifying high risk areas for focused reliability improvement for large software systems. Tutorial 7: Heaney, Matthew, Implementing Design Patterns in Ada95, (full day, intermediate) Design pattern technology is a rapidly growing movement within the software community. This tutorial discusses several design patterns, with a focus on the Ada95 implementation issues. It will also present myriad idioms for object-oriented programming and task synchronization. Tutorial 8: Harbaugh John, High Integrity Ada Tasking, (full day, intermediate) In this tutorial, you will learn how to write concurrent (multithreaded) applications that are consistent with the Ravenscar profile and use the Ada95 programming language. You will also learn how to assure that real-time programs will meet their deadlines using Rate Monotonic Analysis. Tutorial 9: Briot, Emmanuel, GtkAda, an Ada95 object-oriented graphic toolkit, (half day, introductory to intermediate) After an overview of the GtkAda library and its widget set, this tutorial will demonstrate how to use the GtkGlade GUI builder to create your own graphical applications. It will also cover the issues of the design of the library, first as a thin binding to the underlying C toolkit, then as a fully object-oriented Ada set of packages and generic packages. Tutorial 10: Rybin, Sergey and Fofanov, Vasily, Building Ada development tools with ASIS for-GNAT, (half day, introductory) The tutorial will explain how you can build your own development and program analysis tools when working with GNAT. The existing GNAT toolset and different approaches to tool development will be presented. The tutorial will explain how you can build your own development and program analysis tools when working with GNAT. The existing GNAT toolset and different approaches to tool development will be presented. Tutorial 11: Leringe, Örjan, Windows development with Ada, (half day, intermediate) This tutorial presents how Windows applications can be developed using Ada 95 and the principals behind MFC and an Ada binding which gives access to all the features of MFC. The tutorial will also provide an overview of other existing tools and bindings making Windows development with Ada easier. Tutorial 12: Gasperoni, Franco, Developing Ada Applications for the Java Platform with JGNAT, (half day, intermediate) After an overview of the Java technology, this tutorial will focus on how to write Ada applications and applets for the Java environment or Java processors using JGNAT, the Ada 95 development environment for the Java platform. Seamless interoperability between Ada and the Java programming language along with the use of the Java API (Application Programming Interface) from Ada will be discussed in depth. The tutorial will also cover issues involving concurrency and real-time programming. ************************************************************************************** Conference Venue: Potsdam is the capital of the state Brandenburg and situated a few miles west of Berlin, the capital of Germany. Both cities offer many historic and modern attractions and an exciting environment to visitors. The venue of the conference is the hotel "Seminaris" specializing in conferences and meetings. It is situated at the waterfront of a picturesque lake and near the world-famous palace "Sanssouci". Social events on Tuesday and Wednesday, including a musical appearance of "Lady Ada", round out a week well spent to catch up on the latest developments in reliable software technologies. Travel information may be found at: http://www.ada-deutschland.de/AE2000/conf_site/index4.html ***************************************************************************************** Sponsors: The conference is supported and sponsored by Ada Europe, Ada Deutschland e.V., ACM SIGAda, Gesellschaft für Informatik e.V., DFG (German Research Foundation), Forschungszentrum Karlsruhe, ACT Europe, Aonix, CAS GmbH, Green Hills, Information Processing Ltd., Irvine, PeerLogic, PolySpace, Praxis Critical Systems and Rational. ******************************************************************************************* --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Reducing vxWorks kernel footprint Date: Sun, 23 Apr 2000 18:55:21 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8dvguj$b57$1@nnrp1.deja.com> References: <3900B90C.2FE3B82D@west.raytheon.com> <3903354B.FABE8F86@bitband.com> In article <3903354B.FABE8F86@bitband.com>, leonid@bitband.com wrote: > Frankly, I think that 50K is a dream, but 195K is probably something you > can reduce. 50K is certainly possible with T1.0.1 on ARM; in fact I have had a kernel down as small as 30K for ARM (smaller still in Thumb mode). This didn't have the I/O system or network stack though, but it did still have all the important core kernel components. > > Does anybody know how to reduce the vxWorks kernel footprint to less > > than 50K? I have gone through all of the BSP header files looking for > > extraneous features that maybe enabled but have found none. Yet, when > > building what I think is a bare bones kernel, a binary image that is > > approximately 195K in size in generated. The kernel is run on an AMD > > 486 embedded processor with only 170K of EEPROM available for image > > storage. Can you compress the image at all? Maybe that would be a solution since you are pretty close to the size you need... Regards, John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bootrom and Vxworks in non-contiguous memory Date: Thu, 27 Apr 2000 18:14:07 -0400 From: "Emory R. Stagmer" Organization: Ezekiel's Wheel (works for Litton, Amecom division) Message-ID: <3908BBAF.9EBEFB7A@LittonAS.com> References: <8ea4nm$bp7$1@schbbs.mot.com> Jack wrote: > > Does anyone know how to put the VxWorks image, vxworks_rom, in a different > Flash than the bootrom is in? I need to boot from one Flash and have it > "jump" to the vxworks image in another Flash. I'm using the PPC8260. > > Jack The GNU Linker can specify where modules are located both in ROM and RAM. We routinely specify where the modules will be located in ROM and then on boot, copy the ROM to RAM and jump to it. Check out the "AT" keyword in the link script. - -- Emory R. Stagmer Scientist, Litton Advanced Sys Div - Space Systems Operation http://www.amecom.com Bass/12string/keyboards/windsynth for Ezekiel's Wheel http://www.untiedmusic.com Baltimore CMC Chapter Coordinator & North Atlantic CMC Regional Board http://www.cmcnet.org RingMaster for the Christian Music Ring http://www.webring.org/cgi-bin/webring?ring=christmusic;home --------------------------- Newsgroups: comp.os.vxworks Subject: Re: lnPci driver ring buffer size Date: Fri, 28 Apr 2000 08:55:09 -0700 From: "Dave Craig" Organization: Verio Message-ID: References: <20000427230134.20539.00000148@ng-fb1.aol.com> Hi Paul, > I've been spending a _lot_ of time on this driver lately. It seems that > a lot of the code may have been based on some work done for a driver > for some earlier AMD lance version (non-PCNet-PCI, that is). That's what I figured - the 128 limit is based on the software style being set to 0 which seems to be for LANCE ISA interface compatibility. > I don't see any reason why one could not use 512 as the TX and RX > ring buffer size. I gave this a quick try but to no avail. I'll have a closer look today and give you an update. Just out of interest, should I be worried that you've been spending so much time on this driver? Are there problems I should know about? The only bug I'm aware of is in the ln97xSend function of the version I have. This is where is doesn't return the TX semaphore if netClusterGet fails. Anything else I should look at? I'll keep you updated, Thanks, Dave Craig Palo Alto, CA. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Slow GCC compiler in Tornado II Date: Fri, 28 Apr 2000 17:17:37 GMT From: John Fusco Organization: Deja.com - Before you buy. Message-ID: <8ech31$9fc$1@nnrp1.deja.com> References: <3905BAC7.A09FAE90@htc.honeywell.com> I have almost the exact same configuration as you (except I have 128MB RAM), but the slowness I see is due to our NFS client SW. We have Tornado for Unix and NT. When compiling using NFS on NT, it is painfully slow (on the order of 10x slower than Unix). However, if I access my files via Samba (which is SW that makes Unix filesystems look like Windows shares), the build times are comparable to Unix. I can't say that I have observed much swapping when running the compiler, but we use a swap file fixed at 160MB. If you are using a dynamic swapfile, that could be causing you problems as well. In article <3905BAC7.A09FAE90@htc.honeywell.com>, Lee Graba wrote: > I am using Tornado II, hosted on Windows NT and targetting a PowerPC 604 > processor. This compiler appears to be very inefficient, as it causes > continuous disk swapping (on a 96 Meg. machine), which of course makes > it very slow. Compiling relatively small files can take 5 minutes. > > Are the later versions of the GCC compiler more efficient and faster? > Would it be worth building my own cross-compiler of a later > version(perhaps 2.95) in order to speed compiles? Are there other > compilers (such as that from Green Hills) that are significantly faster? > > -- > - ------------------------------------------------------------------------ - -------- > > Lee Graba > Honeywell Technology Center > graba_lee@htc.honeywell.com > - ------------------------------------------------------------------------ - -------- > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: trouble with TFFS_DEFRAGMENT_VOLUME Date: Mon, 24 Apr 2000 01:05:32 GMT From: Andrey Kaganovsky Organization: @Home Network Message-ID: <39039E39.80674AB7@home.com> References: <8djrff$aak$1@news.hdpp.de> I believe you must first create the TrueFFS device and only then defragment it using tffsRawio(TFFS_DEFRAGMENT_VOLUME). Please take into account that defragmentation could take quite a lot of time (few seconds), and the TrueFFS will not be accessible during defragmentaion (all tasks which are attempting to do I/O on this volume will be blocked). regards, Andray Matthias Guelck wrote: > > Hi all, > > I am in trouble with TFFS_DEFRAGMENT_VOLUME. > The following test case ALWAYS prints the ERROR line. > > BLK_DEV *pBlkDev; > unsigned long ulReclaimedSectors; > > if ( tffsRawio(0,TFFS_DEFRAGMENT_VOLUME,0,(int)&ulReclaimedSectors,0) != > OK ) > { > printf("\nERROR while defragmenting\n"); > } > > The WRS support pointed out, that "any attempt to write or erase using > tffsRawio() should be blocked by the system after device creation". > So I tried to defragment my TrueFFS drive before device creation, but > nothing changed. > The WRS support put the blame on my MTD (memory technology driver) and > closed my TSR, but EVERY other operation on this device (including all other > tffxRawio() functions) works as expected. > > I am using Tornado 2.0 with VxWorks 5.4 on a PPC860 with an MBX860 based BSP > but with AM29LV160BB/AM29LV160BT. So none of the example MTDs is useful, > because of non constant block size (erase unit size). > > Looking forward to any suggestion > > Matthias > > ******************************************** > * > * Matthias Guelck > * Research & Development > * > * Heidelberg Digital > * Siemenswall > * D-24107 Kiel > * > * Phone +49-431-386-1608 > * Fax +49-431-386-3882 > * E-Mail Matthias.Guelck@DE.Heidelberg.com > * > ******************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: X-Window client/server for VxWorks? Date: 08 May 2000 10:37:07 -0700 From: Jon Jacky Organization: University of Washington Message-ID: References: <8f6ctm$c8$1@nnrp1.deja.com> chuckallis@my-deja.com writes: > Is there an X-Windows client/server for VxWorks? We use Visicom VX-Windows for VxWorks. It works well. We use the client Xlib only, though I believe they have a server as well. We bought this through Wind River in 1997, they used to resell it. Their manual says Visicom, 10052 Mesa Ridge Court, San Diego CA, 619-457-2111. - -- Jonathan Jacky jon@radonc.washington.edu Radiation Oncology, Box 356043 voice: (206)-598-4117 University of Washington FAX: (206)-598-6218 Seattle, Washington 98195-6043 USA http://www.radonc.washington.edu/prostaff/jon/ --------------------------- Newsgroups: comp.os.vxworks Subject: Need Some Help in SCC z85c30 Date: Tue, 2 May 2000 11:46:43 +0900 From: "ÃÖÁ¾¿í" Organization: KREONet news service Message-ID: <8elfo0$ljf$1@usenet.kreonet.re.kr> Hi, all I'm working with Vxwroks2 with i960KB. Its platform uses 4 SCC chip as z85C30. I initialized its all registers as follow with Zilogs User's Manual. But there are some problem with interrepting Reading/Writing. When I used 'printf' function, writing interrupted and it worked well... but I used 'scanf' or 'getchar' function, No reading interrupting occurred.. If anyone can help.. plz gimme good advice about z85C30 and interrupt... Thanks in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Retrieve system date and time Date: Mon, 08 May 2000 20:43:53 GMT From: peter.kannegiesser@gmx.de (Peter K) Organization: SEK Message-ID: <39171c98.5367668@news.btx.dtag.de> References: <012b46f6.d6a77e77@usw-ex0101-007.remarq.com> <3911d0d2.1892130@news.btx.dtag.de> <32d2ef1f.51a70a27@usw-ex0104-028.remarq.com> Reply-To: peter.kannegiesser@gmx.de On Sun, 07 May 2000 23:36:04 -0700, Marco Penso wrote: >Thank you, Peter > >I'm sorry but I did not understand what you mean. >Can someone tell me more? > When your PC-based BSP is booted, it does not have any idea of the current time. So you have to set it somehow to the correct values. Since you do not want it to do manually, for sure, that must be automated. How? On the standard PC-boards is a battery-buffered real-time-clock (RTC) chip, which is set once and then keeps the time as long as the battery lasts. It is the responsibility of your BSP to read this RTC and to set VxWorks' internal clock with those values. Was that a better explanation? HTH - - Peter - -- Peter Kannegiesser, Dudenhofen, Europe http://home.t-online.de/home/peter.kannegiesser/ Please do not send courtesy copies. It is NOT necessary --------------------------- Newsgroups: comp.os.vxworks Subject: Re: X-Window client/server for VxWorks? Date: Mon, 08 May 2000 20:35:39 GMT From: chuckallis@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8f78eq$15t$1@nnrp1.deja.com> References: <8f6ctm$c8$1@nnrp1.deja.com> Thanks for the info. VX-Windows was the product I was originally thinking of. Does anyone know if it is still supported? I checked both WindRiver's and Visicom's websites and haven't found anything on that product. Thanks, Chuck Allis In article , Jon Jacky wrote: > chuckallis@my-deja.com writes: > > > Is there an X-Windows client/server for VxWorks? > > We use Visicom VX-Windows for VxWorks. It works well. We use the > client Xlib only, though I believe they have a server as well. > > We bought this through Wind River in 1997, they used to resell it. > Their manual says Visicom, 10052 Mesa Ridge Court, San Diego CA, > 619-457-2111. > > -- > > Jonathan Jacky jon@radonc.washington.edu > Radiation Oncology, Box 356043 voice: (206)-598-4117 > University of Washington FAX: (206)-598-6218 > Seattle, Washington 98195-6043 USA > > http://www.radonc.washington.edu/prostaff/jon/ > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RAM_HIGH_ADRS Date: Wed, 03 May 2000 09:34:32 -0400 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <39102AE8.7074E0@lmco.com> References: <8ein86$nbf$1@nnrp1.deja.com> Reply-To: jeffrey.j.daly@lmco.com i think you're just confused about what RAM_HIGH_ADRS is for. it isn't the size of memory, that's what LOCAL_MEM_SIZE is for. RAM_HIGH_ADRS is used by bootroms when they aren't rom resident and by standalone vxWorks kernels that are compressed and not rom resident (corrections?) fywang@my-deja.com wrote: > Hi all, > > I am porting MBX860 BSP to my own target: MPC860T, 32M SDRAM, VxWorks > 5.3.1, Tornado 1.0.1 > > Since I have 32M SDRAM, I would like to change LOCAL_MEM_SIZE & > RAM_HIGH_ADRS to 0x02000000. > But It doesn't work for 0x02000000, 0x01ff0000 or 0x01fe0000. > It can work for 0x01000000, 0x01800000, 0x01c00000, 0x01f00000 & > 0x01f100000 respectively. > I am relutant to check other values. > ("work" means my code in sysHwInit() did run.) > I change makefile according to config.h of course. > > Does VxWorks use the memory above RAM_HIGH_ADRS? For what? > What's the resonable RAM_HIGH_ADRS value for 32M SDRAM? > > in config.h: > #define LOCAL_MEM_LOCAL_ADRS 0x00000000 > #define LOCAL_MEM_SIZE 0x01f10000 > #undef LOCAL_MEM_AUTOSIZE > > #define ROM_BASE_ADRS 0xFE000000 > #define ROM_TEXT_ADRS (ROM_BASE_ADRS + 0x100) > > #define ROM_SIZE 0x00080000 > > #define RAM_HIGH_ADRS (LOCAL_MEM_LOCAL_ADRS + 0x01f10000) > #define RAM_LOW_ADRS (LOCAL_MEM_LOCAL_ADRS + 0x00010000) > #define USER_RESERVED_MEM 0x00000000 > > in makefile: > RAM_LOW_ADRS = 00010000 # RAM text/data address > RAM_HIGH_ADRS = 01f10000 # RAM text/data address > > in romInit.s: > ... > lis r5, HIADJ( 0xFE000800) > addi r5, r5, LO(0xFE000800) > /* TODO: busrt access is disabled for debug > lis r5, HIADJ( 0xFE000900) > addi r5, r5, LO(0xFE000900)*/ > stw r5, OR1(0)(r4) > lis r5, HIADJ( 0x00000081) > addi r5, r5, LO(0x00000081) > stw r5, BR1(0)(r4) > ... > lis sp, HIADJ( STACK_ADRS) > addi sp, sp, LO(STACK_ADRS) > addi sp, sp, -FRAMEBASESZ /* get frame stack */ > /* then jump to romStart */ > My comment: for PPC, STACK_ADRS is the same as RAM_HIGH_ADRS; > I tried both burst & burst inhibited mode. > > in sysHwInit(): > { > ... > sysLedInit(); > for (j=0x00000000; j<0x02000000; j+=0x00100000) { > sysLedOff(LED_ALL); > led = 0; > if (j&0x00100000) { led |= LED_D8; }; > if (j&0x00200000) { led |= LED_D7; }; > if (j&0x00400000) { led |= LED_D6; }; > if (j&0x00800000) { led |= LED_D5; }; > if (j&0x01000000) { led |= LED_D4; }; > if (j&0x02000000) { led |= LED_D3; }; > if (j&0x04000000) { led |= LED_D2; }; > if (j&0x08000000) { led |= LED_D1; }; > sysLedOn(led); > for (i=0; i<0xf0000; i++); > *((UINT*)j) ^= j; > sysLedOff(led); > for (i=0; i<0x30000; i++); > *((UINT*)j) ^= j; > sysLedOn(led); > for (i=0; i<0x30000; i++); > }; > sysLedOn(LED_ALL); > for (i=0; i<0xf0000; i++); > sysLedOff(LED_ALL); > for (;;); > ... > } > > Anybody has some good ideas? > > Regards, > Tom Wang > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado 2 on Windows 2000? Date: Wed, 03 May 2000 15:45:47 GMT From: tsikes@surfaceoptics.com Organization: Deja.com - Before you buy. Message-ID: <8ephj5$753$1@nnrp1.deja.com> When launched under Win2K, Tornado reports that it can't find the initial win32 TCL script. Does the latest cumulative patch address this problem? Is anyone out there successfully using Tornado under Win2K? On a slightly different note, is there a projected ship date for a version of Tornado supporting Linux? Thanks, Terry - -- tsikes@surfaceoptics.com Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multiple ethernet interfaces Date: Mon, 08 May 2000 22:55:28 GMT From: gold@sdsu.edu Organization: Deja.com - Before you buy. Message-ID: <8f7gkp$ajt$1@nnrp1.deja.com> References: <8enn7f$714$1@nnrp1.deja.com> <8epki9$ao2$1@nnrp1.deja.com> In article , Diamand@btinternet.com (Luke Diamand) wrote: > >My turn: Do what Brian says :-) vxWorks does not support the notion of > >a "minor" number when assigning network interfaces, only "major" > >numbers. These are the numbers in the END table in confignet.h, and must > >all be distinct. This does not mean you need multiple copies of the > >driver code - you still check the unit number in the driver and all > >that. I.e. on an 8260 with networks on the SCCs and FCCs, you can't > >have scc0 and fcc0, the unit number for any network interface must be > >unique. > > Weird - I'm sure we do this without any problems at all (e.g. eth0, atm0, > etc). Or am I missing something here? This is T2, 5.4 on an ARM7. We're using T1, SENS 1.1, 5.4 on a PPC750. The problem is in ipAttach(). It only looks at the unit number, not the device name when referencing the array of interfaces (sized by IP_MAX_UNITS). I guess they could have fixed that in T2... Dan gold@sdsu.edu Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Anybody have a good shareware memory mgr? Date: Mon, 8 May 2000 19:59:38 -0400 From: "Cliff Young" Message-ID: <39174cf6@nntpsrv.viagate.com> I am in need a good memory management pkg .... handles mallocs and the like for mpc860 TIA Cliff - ----- Clifford A. Young ViaGate Technologies, Inc 1-877-ViaGate ext. 6411 757 Rt 202/206 (FAX)908-595-6415 Bridgewater, NJ 08807 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Need Some Help in SCC z85c30 Date: Wed, 03 May 2000 09:39:47 -0400 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <39102C23.1C4C8EBD@lmco.com> References: <8elfo0$ljf$1@usenet.kreonet.re.kr> Reply-To: jeffrey.j.daly@lmco.com sounds like you haven't enabled receive interrupts or don't have the receive ISR attached correctly. write register 1 has the various interrupt enable bits. WRS has a z8530Sio.c driver, so you shouldn't have to write your own. "ÃÖÁ¾¿í" wrote: > Hi, all > > I'm working with Vxwroks2 with i960KB. Its platform uses 4 SCC chip as > z85C30. > > I initialized its all registers as follow with Zilogs User's Manual. > > But there are some problem with interrepting Reading/Writing. > > When I used 'printf' function, writing interrupted and it worked well... > > but I used 'scanf' or 'getchar' function, No reading interrupting occurred.. > > If anyone can help.. plz gimme good advice about z85C30 and interrupt... > > Thanks in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Flash ROM Problem with PowerPC mv2700 Date: Wed, 03 May 2000 09:50:56 -0400 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <39102EBF.161AC640@lmco.com> References: <390EE304.AB7EFED7@spawar.navy.mil> Reply-To: jeffrey.j.daly@lmco.com have you first tried just combining it with 'vanilla' vxWorks? (the downloadable one) you should just have to define INCLUDE_USER_APPL and USER_APPL_INIT in config.h and add your Ada module.o to MACH_EXTRA in the Makefile. in config.h: #define INCLUDE_USER_APPL #define USER_APPL_INIT \ { \ yourAdaInit(); \ } that would be my first step... if you've done that, perhaps you're not programming all of the ROM image into flash correctly. does your target have a separately addressable flash area from your bootrom area? (meaning do you have a boot PROM and separate FLASH?) ron webb wrote: > After 7 months of effort we have been unable to create a soldered flash > ROM to be used for delivering our program. > > Our development environment consists of the following: > > Host: SUN Solaris 2.6 > Target: PowerPC Motorola 750 with 8Meg soldered ROM > (Target is in a VME rack with up to 15 different VME boards) > BSP: mv2700 > Compiler: Green Hills Adamulti 1.8.9a > OS: Tornado 1.0 (VxWorks 5.3.1) and Tornado (VxWorks 5.4) > > We currently have 16 PowerPCs programmed and operational using the > bootrom_uncmp, vxWorks and vxWorks_rom images. We have recently > converted most of these successfully to Tornado 2.0. (The bootrom_uncmp > was only usable with Tornado 2.0 by disabling L2 Cache.) > > The problem occurs when we try to combine our executable from Ada (which > is 5.5Meg). When it is combined with VxWorks_rom or VxWorks.st_rom, it > almost never works. The only case where it worked was after we applied > a patch in Tornado 1.0 (TSR#118455 - for ROM images over 1Meg), it > worked but for only 1 single serial number. Wind River has been to our > site 6-8 times working on this problem without a resolution and are now > looking at hardware or timing problems. We are so desparate we are > looking for a Motorola hardware contact, if such a thing exists anymore. > > Does anybody have any ideas?? > > -- > > Ronald N. Webb, Sr. Computer Scientist > ----------------------------------------------------------------------- > Computer Sciences Corp. FAX 619-226-0462 > 4045 Hancock Street Phone 619-553-5021 > San Diego, CA 92110 E-mail > ronwebb@spawar.navy.mil > ----------------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Stethoscope 5.(latest) Date: Tue, 09 May 2000 00:24:20 GMT From: MJM Organization: mjm et al Message-ID: <391758DE.17E3DE04@sympatico.ca> Reply-To: "mike .mcnaughton@sympatico.ca" Hello I have stehoscope installed and running on a mvme162, vxworks tornado 1.01 system. It is a g reat package except for one thing. If I leave it on the target along with the main application, it will sooner or later crash the sytem ; usually in less than 24 hours. When the sytem crashes, there is no shell, no life left, no choice but to reboot. I've run windView on the target and noted that the system has a whopping ISR which has a run time of about 1 millisec and runs every 2Msec. Other than that there are a few routines spread out nicely in time which leaves apparently a fair ammount of idle time. I'd be interest in anyones ideas as to a cure. Thanks MJM --------------------------- Newsgroups: comp.os.vxworks Subject: Re: smNetShow Date: Wed, 3 May 2000 14:40:21 GMT From: Charlie Grames Organization: Boeing Message-ID: <39103A55.16B75E61@nowhere.com> References: Sender: nntp@news.boeing.com (Boeing NNTP News Access) Pedro, By default, you *will* see an int type of "poll" when a secondary target is booting, because the boot type in usrBpInit() in target/all/bootConfig.c is SM_INT_NONE (polling). You can change this in bootConfig.c, if you want, and rebuild your bootrom image. However, it shouldn't be necessary to get shared memory booting working. I didn't start having problems with the polling method until I needed to boot several targets in one chassis simultaneously. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com Pedro Emanuel Cordeiro Dias wrote: > > The architecture of our system is as follow: the PC Host (using Windows > 95); the targets are MVME177 (Single Board Computer) and MVME162LX > (without Ethernet Interface) over a VMEbus backplane. The boot file for > MVME177 is loaded through Ethernet Interface. For MVME162LX, we are trying > to use Shared Memory Network to load the boot file (we can't boot > MVME162LX). > After booting MVME177 (system controller), with smNetShow command we get: > > -> smNetShow > Anchor Local Addr: 0x600 > heartbeat = 21, header at 0x1fa6144, free pkts = 24. > > cpu int type arg1 arg2 arg3 queued pkts > --- -------- ---------- ---------- ---------- ----------- > 0 mbox-1 0x2d 0xcc02 0x8 5 > 1 poll 0 0 0 0 > > input packets = 1 output packets = 2 > input errors = 0 output errors = 0 > collisions = 0 > value = 0 = 0x0 > > In the 'int type' column for cpu 1 we have 'poll'. > What does it means? We shouldn't get this, should we? > > Thanks in Advanced, Pedro. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Flash Driver Question Date: 03 May 2000 14:53:08 GMT From: Vishnu Raman Message-ID: <39103d54$0$35171@news.voyager.net> I am trying to get the VxWorks flash driver (flashMem.c) to work for am29F040 chip on a ads860 board. I am unable to perform auto-select, write and erase. Do I have to write enable the flash chip ? Is there any initialization I need to do to get it to work ? Thanks Vishnu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multiple ethernet interfaces Date: Wed, 03 May 2000 16:36:27 GMT From: gold@sdsu.edu Organization: Deja.com - Before you buy. Message-ID: <8epki9$ao2$1@nnrp1.deja.com> References: <8enn7f$714$1@nnrp1.deja.com> In article , "Brian St. Pierre" wrote: > In article <8enn7f$714$1@nnrp1.deja.com>, gold@sdsu.edu wrote: > > In article , > > "Randy Schafer" wrote: > >> Does anyone know the "correct" sequence of function calls to install > >> multiple END ethernet drivers and attach them to protocols? I can set > >> the boot device to either ports and it boots fine. If I then try to > >> initialize the other port after booting, I only get > > errors. > >> System is EST 8260 eval board using T2 and vxWorks 5.4. > > > > First make sure you have the extra interfaces in the table in > > confignet.h. > > Do what Dan said, but as a caveat, be aware that you can not re-use the > "unit number", even if you have different devices. We tried this and it > does not seem to be supported. I.e. if you have interfaces for devices > you're calling "aa" and another called "bb", you need to call the > interfaces "aa0" and "bb1". My turn: Do what Brian says :-) vxWorks does not support the notion of a "minor" number when assigning network interfaces, only "major" numbers. These are the numbers in the END table in confignet.h, and must all be distinct. This does not mean you need multiple copies of the driver code - you still check the unit number in the driver and all that. I.e. on an 8260 with networks on the SCCs and FCCs, you can't have scc0 and fcc0, the unit number for any network interface must be unique. Dan gold@sdsu.edu Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anybody have a good shareware memory mgr? Date: Tue, 09 May 2000 01:29:57 GMT From: snaphook@southwesternbell.net (Bruce) Organization: SBC Internet Services Message-ID: <39176a04.12208433@news.swbell.net> References: <39174cf6@nntpsrv.viagate.com> In comp.os.vxworks "Cliff Young" wrote: >I am in need a good memory management pkg .... handles mallocs and the like >for mpc860 Look at this one, it may help. http://www.fourmilab.to/bget/ --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks for Newbies Date: Sun, 07 May 2000 13:43:52 GMT From: aj@scoot.netis.com (A. J. Rappaport) Organization: MV Communications, Inc. Message-ID: <391572be.481211921@news.netis.com> Can anyone suggest the best resources for VxWorks newbies to learn the system? Something like a "VxWorks for Dummies" book would do nicely, or perhaps a website devoted to VxWorks that doesn't assume any particular level of knowledge. Thanks for any assistance... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: boot failure question Date: Thu, 04 May 2000 18:26:30 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <39122346.E1EB54AA@gv.net> References: Reply-To: mkesti@gv.net Charles Krinke wrote: >A question of boot error if I might. I have a friend with a target that >fails boot and displays an error code of "D0003" and he asks me it this is >from VxWorks or his unique hardware. I thought I would ask if anyone knows? If this is error number (errno) 0xd003, one can use the printErrno(errno) shell command to specify the error. The trouble is that you have get booted to run a shell command, so 0xd0003 is an invalid file descriptor error. Common reasons for an invalid file descriptor error to occur at boot time are incorrect boot parameters such as host (and other) IP addresses and the path to the boot image. Also, one must run appropriate host services. For example, if the target is configured to load the image via FTP, there has to be a properly configured FTP server running on the host. I hope this helps your friend. 8-) - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Retrieve system date and time Date: Tue, 09 May 2000 07:44:25 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <3917A5B9.1047128D@bitband.com> References: <012b46f6.d6a77e77@usw-ex0101-007.remarq.com> <3911d0d2.1892130@news.btx.dtag.de> <32d2ef1f.51a70a27@usw-ex0104-028.remarq.com> <39171c98.5367668@news.btx.dtag.de> Reply-To: leonid@bitband.com IYou are right, Peter, the BSP should have an RTC driver and a module which syncs the VxWorks "wall" clock to the RTC periodically, as the sysClk interrupts could be less accurate time source then the RTC which has a dedicated crystal. In fact, some time ago when I was working for Wind River we developed such an RTC driver and a simple syncing layer, but this did not get included in the standard product for reasons I can not explain. You may however contact your FAE and ask him for this code. Leonid Peter K wrote: > On Sun, 07 May 2000 23:36:04 -0700, Marco Penso > wrote: > > >Thank you, Peter > > > >I'm sorry but I did not understand what you mean. > >Can someone tell me more? > > > When your PC-based BSP is booted, it does not have any idea of the > current time. So you have to set it somehow to the correct values. > Since you do not want it to do manually, for sure, that must be > automated. > How? On the standard PC-boards is a battery-buffered real-time-clock > (RTC) chip, which is set once and then keeps the time as long as the > battery lasts. > It is the responsibility of your BSP to read this RTC and to set > VxWorks' internal clock with those values. > > Was that a better explanation? HTH > > - Peter > > -- > Peter Kannegiesser, Dudenhofen, Europe > http://home.t-online.de/home/peter.kannegiesser/ > Please do not send courtesy copies. It is NOT necessary --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Labview/NT <--> vxWorks Date: Tue, 09 May 2000 09:24:25 +0100 From: Roland Hauser Organization: Technische Universitaet Graz, Austria Message-ID: <3917CB39.453B5B77@avl.com> References: <39117242.808F40B7@americasm01.nt.com> We have implemented for the moment a serial communication which works fine and are about to implement a TCP communication which shoudn`t be a problem. I`am developing the VxWorks part and for the serial communication where was just standart programming techniques needed. Roland Krawczuk, Victor [WDLN2:2X26:EXCH] schrieb: > > Hi, > > I'm looking at options for having NI's Labview running on a PC/NT > host > communicate with a VxWorks target, either directly via the target's > ethernet port or via the Tornado target server running on the same WinNT > host. > > Does anyone have any war stories as to which techniques worked, > tricks, traps, > reliability stories, etc....? > > Regards, > > Victor. - -- Roland Hauser mailto:Roland.Hauser@avl.com Voice: ++43- 316- 787-587 FAX: ++49- 089- 2443- 29400 mailto:rhauser@cccom.at --------------------------- Newsgroups: comp.os.vxworks Subject: Re: monitoring packets Date: Tue, 9 May 2000 10:13:11 +0200 From: "Gijs Withagen" Organization: XS4ALL Internet BV Message-ID: <8f8gve$80u$1@news1.xs4all.nl> References: <8eueig$2nkv$1@s2.feed.news.oleane.net> Another possible solution for Windows is WinDump, we use it here and it works allright (actually this is the Windows port of the tcpdump utility available for most UNIX systems (SUN, HP, Linux ....) Gijs Withagen Technolution bv. http://www.technolution.nl Jean-Francois Monestier wrote in message <8eueig$2nkv$1@s2.feed.news.oleane.net>... >I'm personnally using NetXRay for this. We've installed it on a separate >Win95 PC. > >You can filter whatever's flowing on the network on a MAC, or IP basis (to >identify hosts). > >You can also select what protocol you'd like to analyze : UDP, TCP, X and >much more. > >Then you take a network dump, and you can review it packet by packet, header >and data. > >I think this one is paying. > > >Jean-Francois. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is it ok to use Tornado together with another RTOS and network stack? Date: 9 May 2000 10:14:03 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: Customer of Planet Online Message-ID: References: <8eqq6j$ll3$1@nnrp1.deja.com> I believe that Wind River have (not unreasonably) made it extremely hard to connect the Tornado development environment to anything other than VxWorks. The relevant API you need between the target server and the target isn't published, IIRC. HTH! Luke Paul Whicker wrote: >Tornado is a tool for configuring and developing vxWorks applications, >everything you get as standard is directed to that. Other people have provided >tools to help in this, but to replace the target of the entire system would >take some doing. > >Could it be done? I dare say it could, however it would be a development >project beyond most target systems. > >Paul > >In article <8eqq6j$ll3$1@nnrp1.deja.com>, david299@my-deja.com wrote: >>Hi, >> >>I'm new to tornado and vxwork. Could anyone tell me whether it is permitted >>and convenient to use tornado with another RTOS(with source code .nucleus or >>superTask etc. ) instead of the vxwork who is very expensive. >> >>Any reply will be greatly appreciate. >> >>BR >> >> >>Sent via Deja.com http://www.deja.com/ >>Before you buy. - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Source MAC address being overwritten for non-IP traffic Date: 9 May 2000 10:14:06 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: Customer of Planet Online Message-ID: References: <39165354.EF7C296@bitband.com> If we really are talking about a layer 2 bridge (or similar) then you don't need to modify WRS source at all. Surely you just write yourself a new protocol, attach it to the top of the MUX, and the relevant ethernet device drivers (ENDs) to the bottom, and off you go. The layer 2 bridge sets itself up to see all packets (MUX_PROTO_SNARF) and sends them out on the other interface(s). I could be way off base here, of course ;-( Regards! Luke In article <39165354.EF7C296@bitband.com>, Leonid Rosenboim wrote: >Paul, > >I think that this guy is actually trying to do some sort of Layer-2 switching >(a.k.a Bridging), >in which case, it is quite allright for him to move the MAC SA transparently >through his link >and nodes. This Bridge could perhaps be some sort of LAN-Emulation over ATM or >something like that. > >Also, MAC addresses, as well as the ARP translation must be consistent within a >LAN - which >could be any combination of segments connected with Repeaters (Layer-1) or Bridges > >(Layer-2), or V-LAN, and could span large geographies, but the Source Address of >Bridges are only used for Bridge-originated packets used for route / loop >discovery and management. All other packets should come with their original MAC >address. > >To the original poster - i fyou are doing what I guess you are, you will simply >HAVE TO >modify WRS driver source code, more reasons then you think at this time. > >Leonid > >Paul Whicker wrote: > >> Ethernet repeaters do this sort of thing but nodes don't. Repeaters are >> essentially within the collision domain and exist for electrical rather than >> routing purposes, they don't switch frames in software. >> >> It is required that the MAC address is specific to the Media Access >> Controller, all MACs are required to have different addresses and the >> community has gone to a lot of trouble to make sure that is true. >> >> MAC addresses should only be important within a collision domain (on the same >> piece of wire) as they are the addressing method at that layer, you should not >> be interested in MAC addresses in association with a layer two protocol other >> than to establish which piece of hardware on this wire needs to receive the >> frame. The enclosed packet should then be routed according to its kayer 2 >> address. >> >> Appart from anything else, say you did replace the MAC address, the node whih >> eventually received the frame would then reply to that MAC address... and you >> wouldn't see it. You could install a snarfing protocol to steal receive all >> frames, however this is poor design as it imposes a high processing load which >> MAC address filtering was intended to avoid. >> >> If someone designed a layer two protocol to route on layer one addresses (not >> all layer ones are Ethernet after all) then they have a deal of reading and >> understanding to do. >> >> Then again there's always exceptions to every rule, so perhaps you do have a >> reason to do this. >> >> Paul >> >> In article , "Denis Rouleau" >> wrote: >> >HELP >> > >> >I need to have my END ethernet driver be transparent to non-IP packets. The >> >problem that I'm faced with is that generic code in vxworks (ether_output in >> >if_subr.c) always overwrites the source MAC (hardware) address for all >> >packets going out on the ethernet interface. >> > >> >I have to keep the source MAC address that is already in the packet header >> >as the recipient must think that the sending device is on the same network >> >segment. In short, I am connecting two remote LAN segments together but all >> >devices on both segment must think that they are locally connected. Of >> >course, these segments carry non-IP traffic. >> > >> >I initially though of creating a clone of the ifnet structure for my >> >interface in which I would change the MAC address to that of the current >> >packet but there is too much dependant information in the structure to make >> >it an easy fix. >> > >> >If I owned the if_subr.c code, I could easily modify the ether_output >> >routine so that it wouldn't overwrite the source MAC address but that's not >> >the case. >> > >> >Can anyone suggest an elegant fix that I can apply before calling >> >etherOutput (in etherLib.c not ether_output in if_subr.c)? >> > >> >Thanks in advance >> > >> > > - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: asPPP under Solaris 2.6 Date: Tue, 2 May 2000 22:19:17 +0200 From: "Maik H." Organization: Nikoma Mediaworks GmbH Message-ID: <0hene8.k44.ln@spaci.marichs.mhn> Reply-To: "Maik H." Hi, we have tried to use PPP as network interface together with WDB network connection. After termination of VxSIM the respective PPP interface cannot be accessed by further VxSIM sessions. Only a reboot of the machine (or a restart of asppp) can solve the problem. The configuaration of our SUN machines is based on Solaris 2.6. We have heard, that this problem would not occur on other versions of Solaris. Is there any workaround for Solaris 2.6? - -- MfG, Maik --------------------------- Newsgroups: comp.os.vxworks Subject: SLIP question Date: Tue, 25 Apr 2000 16:15:42 +0100 From: David Rigg Organization: GEC Marconi Avionics Ltd Message-ID: <3905B69E.6C59@gecm.com> Reply-To: david.rigg@gecm.com Hi Folks I have a card (#2), (without an ethernet interface) running VxWorks which boots over a shared memory backplane from a host PC, through another card (#1)running VxWorks, (with an ethernet interface). This all works fine. Although it doesn't have an ethernet port, it does have a serial port which I connect to a hyperterminal and get all my vxWorks output. Now I want to transfer date through the serial port from the PC to the card. Is SLIP the way to go on this? I've tried following the instructions in TSR55352 (below) but it fails to accept a direct route in the routeAdd part as the card (2) already has a route through card (1) to the host PC which I need to get the shared memory to work. Any ideas? Also apart from having SLIP set up in dial-up networking on the PC, is there something I'm missing trying to get an 'ls' on the card to work. What on the PC should tell it to respond over the serial port? thanks for any help or indeed, a better idea :) david RESOLUTION: 1. rebuild vxWorks including SLIP . 2. boot your target using the image including SLIP 3. after the boot tape following commands from the shell: -> slipInit(0,"/tyCo/1","my_slip_addr","peer_slip_Adr",9600,0,0) -> hostAdd("slip_host","peer_slip_adr") -> routeAdd("slip_host","my_slip_addr") -> netDevCreate("slip:","slip_host",1) you can verify using ifShow, hostShow, routeShow , devs if the initialisation has been done correctly. Now you should be able to ping , rlogin or telnet from the slip host to your target . -> cd "slip:your_directory" -> ls -> ld < test.o David Rigg \ _ / Mission Avionics Division, x__________\_(0)_/__________x BAE SYSTEMS, Edinburgh, Scotland @ @ (](_o_)[) @ @ Tel : (+44) (0)131 314 8223 --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue May 9 07:29:45 2000 From: bob schulman Date: Tue May 9 07:29:48 PDT 2000 Subject: Re: X-Window client/server for VxWorks? > Thanks for the info. VX-Windows was the product I was originally > thinking of. Does anyone know if it is still supported? I checked > both WindRiver's and Visicom's websites and haven't found anything on > that product. VX-Windows is no longer supported. It's been superceded by RtX Windows which is also no longer actively supported (though I think it's still sold). bob From vxwexplo-errs@csg.lbl.gov Tue May 9 07:32:36 2000 From: Jim Way Date: Tue May 9 07:32:38 PDT 2000 Subject: RE: Redirect printf Stephane, Two suggestions. Maybe one of these suggestions will aim you in the right direction. 1. Instead of printf(), use logMsg(). This uses a msgQ which frees up your task. 2. I couldn't find the documentation from a short search, but I believe if you write to one of the "standard" I/O devices, you can use an ioctl() call to redirect the output to another device/file. Jim --------------------------- I'd like to redirect all printf to a message Q in order to avoid the use of the serial port or the Tornado shell . A task will read the msgQ and the data will be sent to the network using VxDCOM through Connection points. Is there a simple solution in order to redirect these printf to a msgQ? Stephane. --------------------------- ----------------------------------------- Jim Way, Software Engineer Datum Austin (Austron Inc.) voice: 512.721.4170 fax : 512.721.4087 email: jwayATdatumDOTcom (no spam please) ----------------------------------------- From vxwexplo-errs@csg.lbl.gov Tue May 9 07:40:36 2000 From: Jim Way Date: Tue May 9 07:40:39 PDT 2000 Subject: RE: passing function pointers between tasks Srikanth, I believe that in current vxWorks, all tasks share a common name space. So, TaskB is free to call any function in TaskA. It should be just as easy to pass a function pointer from TaskA to TaskB. However, Tornado 3.0 promises to introduce protection domains which will allow you (for very good reasons) to not do this. If you're just wanting to reuse logic, perhaps you can copy the function so that each task has its own copy. If you have two separate tasks that need to share functionality, it might suggest that there is a different (better) partitioning of your system into a different set of tasks. Jim > Assume that there is a function foox() within the context of TaskA. Is it > possible to send the pointer to foox() to another TaskB in VxWorks? > Srikanth ----------------------------------------- Jim Way, Software Engineer Datum Austin (Austron Inc.) voice: 512.721.4170 fax : 512.721.4087 email: jwayATdatumDOTcom (no spam please) ----------------------------------------- From vxwexplo-errs@csg.lbl.gov Tue May 9 08:39:13 2000 From: David Anderson Date: Tue May 9 08:39:15 PDT 2000 Subject: 404, 564 BYTES mtu 1518 NOT 1500. vxWorks , Tornado etc. If my memory serves me correctly the person with the UDP problem has a CS8900 Ethernet driver. If this is the case please note that the if_mtu is set to 1518 by the driver (it is 1500+ethernet header in the source code). I've found this rather dubious. The TCP/IP stack much prefers to see 1500 bytes here. I think there is also a possible driver problem relating to the size of the on-board CS8900 buffers and the way TX is performed. I've yet to prove this, but have a system that will not repsond to a ping of over 3400 bytes ( and that is not a multiple of either the MTU , the buffer sizes or anything else obvious... so it must be in the driver ???). David Anderson. mailto:danderson@dtrack.demon.co.uk Disclaimer ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. ------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Tue May 9 09:00:25 2000 From: "Thomas, Robert S" Date: Tue May 9 09:00:28 PDT 2000 Subject: Optimization problem with tornado1.0.1/VxWorks5.3.1 Greetings, Due to constraints on our product, we require a high degree of optimization. We are using the -O3 option in the compiler, however using this flag yields an 'Internal Compile Error' This error also occurred using -O2. The only optimization level that would compile withour error was -O1 which is not an option for the end product. After some playing with the flags associated with each level of optimization, it was found that we could use the -O3 optimization level provided we turn off combine_statics (-fno_combine_statics). I understand this is a know problem, has anyone else seen this happen before? Does anyone know what implications removing this flag will have on our code? What impact will removing this flag have on the optimization? We are currently developing C++ code using Tornado1.0.1 / Vxworks 5.3.1 / BSP 1.1/5 on Sun stations running Solaris 2.6 Our target board is a MVME2604 PPC GNU2.7.2 Any information would be great, Thanks! Rob robert.s.thomas@lmco.com From vxwexplo-errs@csg.lbl.gov Tue May 9 16:55:51 2000 From: "Barkley, Charles (HT-EX)" Date: Tue May 9 16:55:54 PDT 2000 Subject: Leading zero in addHost() IP address This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_001_01BFB9FD.AE31CEFC Content-Type: text/plain Hello all, I just ran into a problem on several of our products running VxWorks 5.2, 5.3, and 5.4. It seems like this should be a well known problem, but I have found nothing about it while searching the documentation, the archives here, or on Wind River's WindSurf web site. If anyone has seen this problem before please let me know. THE PROBLEM: When a leading zero is used as part of the IP address passed to addHost(), the octet having the leading zero is incorrectly entered into the host table. For instance, if "168.083.250.083" is passed as a IP parameter to addHost, a subsequent showHost() indicates that the IP address entered in the host table is "168.67.250.67". Incrementing the "083" to "084" results in the corresponding octet in the host table becoming "68". However, if "002" is entered, the host table entry octet will be "2". Incrementing to "003" results in a table entry of "3". It appears to be a VxWorks problem, not an application problem. Chuck Barkley Motorola Broadband Communications Sector Manager, DNS Systems Technology cbarkley@gi.com ------_=_NextPart_001_01BFB9FD.AE31CEFC Content-Type: text/html Content-Transfer-Encoding: quoted-printable Leading zero in addHost() IP address

Hello all,

I just ran into a problem on several = of our products running VxWorks 5.2, 5.3, and 5.4.  It seems like = this should be a well known problem, but I have found nothing about it = while searching the documentation, the archives here, or on Wind = River's WindSurf web site. If anyone has seen this problem before = please let me know.

THE PROBLEM:
When a leading zero is used as part = of the IP address passed to addHost(), the octet having the leading = zero is incorrectly entered into the host table.  For instance, if = "168.083.250.083" is passed as a IP parameter to addHost, a = subsequent showHost() indicates that the IP address entered in the host = table is "168.67.250.67".  Incrementing the = "083" to "084" results in the corresponding octet = in the host table becoming "68".  However, if = "002" is entered, the host table entry octet will be = "2".  Incrementing to "003" results in a table = entry of "3". It appears to be a VxWorks problem, not an = application problem.

Chuck = Barkley
Motorola Broadband Communications = Sector
Manager, DNS Systems = Technology
cbarkley@gi.com

------_=_NextPart_001_01BFB9FD.AE31CEFC-- From vxwexplo-errs@csg.lbl.gov Tue May 9 19:43:11 2000 From: Fred Roeber Date: Tue May 9 19:43:13 PDT 2000 Subject: Re: Leading zero in addHost() IP address On Tue, 9 May 2000, Barkley, Charles wrote: > I just ran into a problem on several of our products running VxWorks 5.2, > 5.3, and 5.4. It seems like this should be a well known problem, but I have > found nothing about it while searching the documentation, the archives here, > or on Wind River's WindSurf web site. If anyone has seen this problem before > please let me know. > > THE PROBLEM: > When a leading zero is used as part of the IP address passed to addHost(), > the octet having the leading zero is incorrectly entered into the host > table. For instance, if "168.083.250.083" is passed as a IP parameter to > addHost, a subsequent showHost() indicates that the IP address entered in > the host table is "168.67.250.67". Incrementing the "083" to "084" results > in the corresponding octet in the host table becoming "68". However, if > "002" is entered, the host table entry octet will be "2". Incrementing to > "003" results in a table entry of "3". It appears to be a VxWorks problem, > not an application problem. The "problem" is that the IP address decomposition code uses "normal" numeric conversion conventions on each octet value (think sscanf). Under these conventions, a leading '0' indicates an octal value. Hence 083 -> 67. You either have to live with this operation or write your own code to massage the IP addresses to remove the leading zeroes that are bothering you. Personally, I think it is fairly unlikely for users to enter leading 0's in IP addresses so the way VxWorks handles things seems fine. I suggest adding the issue to your support data base in the rare event that it actually bothers someone. HTH, Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Wed May 10 02:35:33 2000 From: David Anderson Date: Wed May 10 02:35:35 PDT 2000 Subject: RE: Leading zero in addHost() IP address : Problem with Octal ? On 10 May 2000 01:12, the vxWorks Users Group Exploder [SMTP:vxwexplo@lbl.gov] wrote: > vxWorks Tornado > > Hello all, > > I just ran into a problem on several of our products running VxWorks 5.2, > 5.3, and 5.4. It seems like this should be a well known problem, but I have > found nothing about it while searching the documentation, the archives here, > or on Wind River's WindSurf web site. If anyone has seen this problem before > please let me know. > > THE PROBLEM: > When a leading zero is used as part of the IP address passed to addHost(), > the octet having the leading zero is incorrectly entered into the host > table. For instance, if "168.083.250.083" is passed as a IP parameter to > addHost, a subsequent showHost() indicates that the IP address entered in > the host table is "168.67.250.67". Incrementing the "083" to "084" results > in the corresponding octet in the host table becoming "68". However, if > "002" is entered, the host table entry octet will be "2". Incrementing to > "003" results in a table entry of "3". It appears to be a VxWorks problem, > not an application problem. Looks like numbers with leading zeroes are being interpreted as Octal. Might be a problem with the compiler and not vxWorks. David Anderson. > > Chuck Barkley > Motorola Broadband Communications Sector > > From vxwexplo-errs@csg.lbl.gov Wed May 10 05:47:21 2000 From: "Straub, Robert" Date: Wed May 10 05:47:23 PDT 2000 Subject: RE: Leading zero in addHost() IP address Charles, It *could* be a VxWorks "problem", if the "addHost()" is being entered via the shell. Or it could be a "problem" with your C compiler if the addHost() is in a program. Remember that in C, an integer constant is taken to be octal if it begins with "0". It looks like the number is being converted to internal representation like so: 8*8+3=67. Does 073 result in a value of 59? The problem is that K&R 2nd Ed. says that octal constants do not contain the digits 8 or 9. On the other hand, K&R 1st Ed. says that "the digits 8 and 9 have octal value of 10 and 11 respectively." That makes the behavior you see understandable. So whether there's a "problem" or not depends on the standard to which the compiler or the shell interpreter adheres. Robert J. Straub Principal Software Engineer Telephonics Corporation 815 Broad Hollow Road, Mail Stop 3 Farmingdale, NY 11735 516/631-755-7769 (Voice) 516/631-755-7642 (Fax) straub@telephonics.com -----Original Message----- From: vxwexplo@lbl.gov [mailto:vxwexplo@lbl.gov] Sent: Tuesday, May 09, 2000 8:12 PM To: vxworks_users@csg.lbl.gov Subject: Leading zero in addHost() IP address Submitted-by vxwexplo-errs@csg.lbl.gov Tue May 9 16:55:51 2000 Submitted-by: "Barkley, Charles (HT-EX)" Hello all, ... When a leading zero is used as part of the IP address passed to addHost(), the octet having the leading zero is incorrectly entered into the host table. For instance, if "168.083.250.083" is passed as a IP parameter to addHost, a subsequent showHost() indicates that the IP address entered in the host table is "168.67.250.67". Incrementing the "083" to "084" results in the corresponding octet in the host table becoming "68". However, if "002" is entered, the host table entry octet will be "2". Incrementing to "003" results in a table entry of "3". It appears to be a VxWorks problem, not an application problem. From vxwexplo-errs@csg.lbl.gov Wed May 10 11:42:39 2000 From: Jim Thomas Date: Wed May 10 11:42:41 PDT 2000 Subject: Re: Leading zero in addHost() IP address Fred> On Tue, 9 May 2000, Barkley, Charles wrote: Charles> When a leading zero is used as part of the IP address passed to Charles> addHost(), the octet having the leading zero is incorrectly Charles> entered into the host table. Fred> The "problem" is that the IP address decomposition code uses "normal" Fred> numeric conversion conventions on each octet value (think sscanf). Fred> Under these conventions, a leading '0' indicates an octal value. Hence Fred> 083 -> 67. You either have to live with this operation or write your Fred> own code to massage the IP addresses to remove the leading zeroes that Fred> are bothering you. Personally, I think it is fairly unlikely for users Fred> to enter leading 0's in IP addresses so the way VxWorks handles things Fred> seems fine. I suggest adding the issue to your support data base in the Fred> rare event that it actually bothers someone. HTH, Fred Umm, sscanf %d does not parse 083 as octal. This is a bug, not a feature. Lots of form fill ins put three digits with leading zeros. From vxwexplo-errs@csg.lbl.gov Wed May 10 13:01:19 2000 From: Chris Hann Date: Wed May 10 13:01:22 PDT 2000 Subject: Re: Leading zero in addHost() IP address > > Umm, sscanf %d does not parse 083 as octal. This is a bug, not a feature. > Lots of form fill ins put three digits with leading zeros. Actually this is the well known, or in other words accepted, behavior. Here are a couple of typical responses to this question, none specific to vxWorks: http://acmebw.com/askmrdns/00463.htm http://www.iihe.ac.be/local/info.sun-managers/msg00092.html http://www.ebsinc.com/solaris/dns.html Interestingly the example 083 isn't valid octal but is still interpreted as such, this is also well known behavior (See C Traps and Pitfalls, Andrew Koenig, section 1.4 Integer Constants, page 9), C doesn't mind having a number greater than 7 in octal. Chris From vxwexplo-errs@csg.lbl.gov Wed May 10 13:31:20 2000 From: Tom Miller Date: Wed May 10 13:31:22 PDT 2000 Subject: VxWorks first Question... ....are any of your recieving this? Tom Miller Motion Engineering Inc. Santa Barbara Ca. From vxwexplo-errs@csg.lbl.gov Wed May 10 14:27:06 2000 From: Tom Miller Date: Wed May 10 14:27:08 PDT 2000 Subject: VxWorks sizeof(int) question. Hello everyone, I'm here to ask a remedial question. A bit embarassing, but here goes: What determines the size of an "int" in any given vxworks system? I am running a vxworks version based on the '486 bsp, and the other day when i was having a problem I typed "sizeof(int)" in the command shell and got 4 as a result - my system considers an int to be 32 bits! Is it the architecture of the target processor that determines the size of an int? Is it the operating system? Is this a settable parameter somewhere in VxWorks? any explanation would be appreciated. Tom Miller Motion Engineering Inc. Ps Thanks to Andreas Wolf and Bob Anderson for responding to my first query. From vxwexplo-errs@csg.lbl.gov Wed May 10 15:03:09 2000 From: Chris Hann Date: Wed May 10 15:03:12 PDT 2000 Subject: Re: VxWorks sizeof(int) question. the vxWorks Users Group Exploder wrote: > > Submitted-by vxwexplo-errs@csg.lbl.gov Wed May 10 14:27:06 2000 > Submitted-by: Tom Miller > > Hello everyone, > > I'm here to ask a remedial question. A bit embarassing, but here goes: > > What determines the size of an "int" in any given vxworks system? I am > running a vxworks version based on the '486 bsp, and the other day when i > was having a problem I typed "sizeof(int)" in the command shell and got 4 as > a result - my system considers an int to be 32 bits! > > Is it the architecture of the target processor that determines the size of > an int? Is it the operating system? Is this a settable parameter somewhere > in VxWorks? > > any explanation would be appreciated. > > Tom Miller > Motion Engineering Inc. > > Ps Thanks to Andreas Wolf and Bob Anderson for responding to my first query. > An int is intended to be the native integer data size for the processor, on 32bit processors like the 486 this would give a 32 bit int, you would expect a 16 bit processor to have a 16 bit int and so on. Chris From vxwexplo-errs@csg.lbl.gov Wed May 10 19:24:24 2000 From: Fred Roeber Date: Wed May 10 19:24:26 PDT 2000 Subject: Re: VxWorks sizeof(int) question. On Wed, 10 May 2000, Tom Miller wrote: > What determines the size of an "int" in any given vxworks system? I am > running a vxworks version based on the '486 bsp, and the other day when i > was having a problem I typed "sizeof(int)" in the command shell and got 4 as > a result - my system considers an int to be 32 bits! In my experience, the size of an int is solely a compiler issue but is normally tied to the processor architecture. I have seen cases where ints are 16, 32, or 64 bits. With TI C40 DSPs, you even have the odd case where ints are 32 bits but sizeof(int) = 1. All I'm trying to get at is that if you are writing portable software, you shouldn't count on an int being a certain size and certainly shouldn't count on being able to store a pointer value in an int. Some compilers I have worked with in the past have even provided compiler switches that let you adjust the size of any integer. As far as I know, the GNU C compilers don't provide such an option but I could be wrong. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Wed May 10 19:43:11 2000 From: Mark Fischer Date: Wed May 10 19:43:13 PDT 2000 Subject: Re: VxWorks sizeof(int) question. the vxWorks Users Group Exploder wrote: > Is it the architecture of the target processor that determines the size of > an int? Is it the operating system? Is this a settable parameter somewhere > in VxWorks? > According to the original K&R spec, the size of "int" is completely at the mercy of the compiler writer, but is typically the most efficient transfer size for the processor being used. That's why it is a mortal sin against God and Man to assume the size of "int". On Crays, for example, "int" is a 64 bit word. Hope that helps, Mark ~~~~ From vxwexplo-errs@csg.lbl.gov Wed May 10 19:53:14 2000 From: Fred Roeber Date: Wed May 10 19:53:17 PDT 2000 Subject: RE: passing function pointers between tasks On Tue, 9 May 2000, Srikanth wrote: > Assume that there is a function foox() within the context of TaskA. Is it > possible to send the pointer to foox() to another TaskB in VxWorks? Jim Way replied: > I believe that in current vxWorks, all tasks share a common name space. So, > TaskB is free to call any function in TaskA. It should be just as easy to > pass a function pointer from TaskA to TaskB. This is true; all tasks share a common address space and you can easily pass function pointers between tasks. > However, Tornado 3.0 promises to introduce protection domains which will > allow you (for very good reasons) to not do this. If you're just wanting to > reuse logic, perhaps you can copy the function so that each task has its own > copy. If you have two separate tasks that need to share functionality, it > might suggest that there is a different (better) partitioning of your system > into a different set of tasks. I have heard about these new protection domains also for Tornado 3.0 but haven't seen any details. I'm hoping, however, that they won't result in the restrictions that Jim talks about. Certainly, in systems like Unix/Linux, you usually can't pass function pointers between processes because each process has it's own unique address space and executable image (the exception being "forked" processes). However, VxWorks has always employed the notion of a single address space allowing easy transitions between different tasks and between tasks and the kernel. Being able to protect code segments from accidental writing and being able to isolate one task's data from another task's data would certainly be a helpful addition for making reliable code easier to develop. However, I don't really think that protecting different tasks code from other tasks would be that useful given the constraints it would put on program execution. In particular, all the normal overhead suffered by OS's like UNIX for "trapping" to the kernel to execute kernel code would show up. Does anyone know if there are any details on what this new protection domains feature will provide? Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Wed May 10 20:03:35 2000 From: "Luke Skywalker" Date: Wed May 10 20:03:38 PDT 2000 Subject: RARP working with VxWorks? Hi, Does anyone know how I can get RARP running on VxWorks? Thanks, Bob === ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com From vxwexplo-errs@csg.lbl.gov Thu May 11 00:11:59 2000 From: "Michael Lawnick" Date: Thu May 11 00:12:01 PDT 2000 Subject: Re: VxWorks sizeof(int) question. > On Wed, 10 May 2000, Tom Miller wrote: > > > What determines the size of an "int" in any given vxworks system? I am > > running a vxworks version based on the '486 bsp, and the other day when i > > was having a problem I typed "sizeof(int)" in the command shell and got 4 as > > a result - my system considers an int to be 32 bits! > As I remeber ANSI says: The size of an int is the number of char you need to store -> If your machine is a DSP, it mostly uses 32 bit characters and you will get sizeof(int) is 1 or 2. In DOS, where a charcter is a byte and the processor is 16 bit, sizeof(int) is 2, whereas on a PPC (32bit) its 4. Because of this problems, vxWorks defines the types UINT8, UINT16 and UINT32 / INT8, INT16, INT32. They are defind in BSP to guarantee an architecture independant coding. HTH MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== From vxwexplo-errs@csg.lbl.gov Thu May 11 01:22:41 2000 From: "Orhan Ugurlu" Date: Thu May 11 01:22:43 PDT 2000 Subject: memPartFree error Hi VxWorkers, We use a rather old release of VxWorks and VADSworks. Code runs on a VMEbased SPARC card. While we read from a SCSI tape drive (although we read anywhere else with no problem) we get memPartFree error as below: 0x26d0368 (adaT28): memPartFree: invalid block 0x2071da8 in partition 0x21dd80 and then related task is suspended. Do you have any idea about this? Thanks for any help. Orhan Ugurlu Project Engineer orhanu@ayesas.com From vxwexplo-errs@csg.lbl.gov Thu May 11 02:26:45 2000 From: David Laight Date: Thu May 11 02:26:48 PDT 2000 Subject: Re: VxWorks sizeof(int) question. Adding my tuppence... > my system considers an int to be 32 bits! What did you expect? As stated by others int is (usually) the 'natural size' of integers on the system. Definitely 32 bits on a 486. Unix systems (and probably everything except some ms o/s) have used 32 bit integers for all 68xxx and 386 systems. I think the last common 16 bit int Unix were for the 286 and pdp11. vxWorks definitely requires sizeof (int) == sizeof (long) == sizeof (ptr). (and code pointers to be the same size as data pointers.) Whether an 'int' is 16, 32 or 64 (or 8, 24, 36, 48 or...) bits is not strictly a compiler issue, but an ABI (Application Binary Interface) one. Different parts of a system (compiled at different times) have to agree on the sizes of data items. The move to 64 bit systems causes its own problems. Two common sets of data sizes are used ILP64 (int, long and pointer are 64bit) and LP64 (32 bit int, 64 bit long and pointer). DEC use ILP64, Solaris LP64. Sun have had to go through all the structures that user processes might pass into the kernel and change them to use fixed size types - so the same kernel can support 32 bit and 64 bit applications. The 64 bit DEC (now COMPAQ) systems are not binary compatible with any 32 bit system. The C standards body is about to (and has been for a few years) define some fixed size integer types. These will be in int_types.h (I think). The current solaris loads have a sys/int_types.h that contains what they think the standard will be. int8_t an 8 bit signed integer int16_t a 16 bit signed integer int32_t a 32 bit signed integer int64_t a 64 bit signed integer (if supported) uint8_t an 8 bit unsigned integer uint16_t a 16 bit unsigned integer uint32_t a 32 bit unsigned integer uint64_t a 64 bit unsigned integer (if supported) intptr_t a signed integer which can hold any data pointer uintptr_t an unsigned integer which can hold any data pointer The names of the other types in my header file are probably wrong! If you are writing code that needs specific sized integers (eg to map onto hardware registers) then using the above names (or similar ones) with appropriate system (ABI) dependant definitions will avoid confusion. I find it usuful to define vuint8 (volatile unsigned 8 bit integer) etc. Be aware that, just because the domian of a variable is restricted, it is not always appropriate to use a restricted type. Code size and execution speed will almost always be smaller if variables are defined as int instead of short or char. Passing small integers to procedures can be particularly messy an many implementations. David PS: I've no idea what you would do for a C implementation on a system that does BCD arithmetic and has hydrid binary/decimal addressing - like the old ICL System 25. Maybe its a good job no one makes anything like that any more. ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Thu May 11 06:46:41 2000 From: Jim Way Date: Thu May 11 06:46:43 PDT 2000 Subject: RE: VxWorks sizeof(int) question. Just to add to what Michael said...in general, if you want portable code, use the types UINT8, UINT16, etc. Especially when it is important that a variable contain a specific maximum value or roll over at a specific value (yes, I know it is "better" to say "if (++x > MAX) x = 0" but efficiency sometimes wins out over good coding practices). You might also check the limits.h file. It has some interesting constants like __LONG_MAX__ that might be helpful. If you don't find what you're looking for, define your own as the first code you write for your project. Also, don't be afraid to use an assertion like ASSERT (sizeof(int)==sizeof(void*)) if you really feel the need to stuff a pointer into a numeric variable. Class dismissed Jim | |Because of this problems, vxWorks defines the types |UINT8, UINT16 and UINT32 / INT8, INT16, INT32. |They are defind in BSP to guarantee an architecture independant |coding. | |MfG |Lawnick, SOFTEC GmbH | From vxwexplo-errs@csg.lbl.gov Thu May 11 11:35:52 2000 From: Jim Thomas Date: Thu May 11 11:35:54 PDT 2000 Subject: Re: Leading zero in addHost() IP address Jim> Umm, sscanf %d does not parse 083 as octal. This is a bug, not a feature. Jim> Lots of form fill ins put three digits with leading zeros. Chris> Actually this is the well known, or in other words accepted, Chris> behavior. Here are a couple of typical responses to this question, Chris> none specific to vxWorks: Chris> http://acmebw.com/askmrdns/00463.htm Chris> http://www.iihe.ac.be/local/info.sun-managers/msg00092.html Chris> http://www.ebsinc.com/solaris/dns.html Well, at least HP gets it right (IMHO :-) . Script started on Thu May 11 08:26:09 2000 $ uname -a HP-UX atlas B.10.20 A 9000/712 2011087736 two-user license $ cat sscanfoctal.c #include #include #include main () { char st[20]; int num; int count; int octp; (void) strcpy(st, "01234"); count = sscanf(st, "%d", &num); octp = strtol(st, (char **) NULL, 0); printf("sscanf=%d count=%d strtol=%d\n", num, count, octp); return(0); } $ cc -Ae -g +w1 sscanfoctal.c -o sscanfoctal $ ./sscanfoctal sscanf=1234 count=1 strtol=668 $ exit script done on Thu May 11 08:27:05 2000 From vxwexplo-errs@csg.lbl.gov Thu May 11 23:31:22 2000 From: Henrik Bergstrom Date: Thu May 11 23:31:24 PDT 2000 Subject: autoconf and vxworks in perfect (dis)harmony? Dear VxWorkers, I am trying to use GNU autoconf under Cygwin to build my vxworks project, but have run into serious problems! My experience in using autoconf for cross-compiling is next to zero and I suspect this is the reasons for my problems, but there might also be some blame on WRS. Whatever, the reason I write to this forum is that I would like to know if anyone has gotten autoconf to work in a vxworks environment? If you have I would appreciate any help or pointers to information that you can spare! Thanks in advance! Best regards, Henrik - -- Henrik Bergstrom +46-8-628 28 28 henrik.bergstrom@intertex.se Intertex Data AB (fax) 628 64 14 From vxwexplo-errs@csg.lbl.gov Fri May 12 03:49:54 2000 From: "amass" Date: Fri May 12 03:49:57 PDT 2000 Subject: WindView This is a multi-part message in MIME format. ------=_NextPart_000_001B_01BFBC08.2F562C00 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi We have an installation of Tornado 2/VxWorks 5.3.1 on an NT host = with a Radstone PPC1a target. We have recently attempted an installation = of WindView, but problems have been encountered. The product was = installed and configured what appears to be corectly, but whenever we = attemp to launch the tool, we get the following message: "target name@host name does not support WindView" Can anybody offer any suggestions as to the casue of such a problem? Thanks Andy Atkinson Software Engineer DERA Bincleaves Weymouth Dorset, DT4 8UR UK ------=_NextPart_000_001B_01BFBC08.2F562C00 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi
 
    We have an installation of = Tornado=20 2/VxWorks 5.3.1 on an NT host with a Radstone PPC1a target. We have = recently=20 attempted an installation of WindView, but problems have been = encountered. The=20 product was installed and configured what appears to be corectly, but = whenever=20 we attemp to launch the tool, we get the following message:
 
   "target name@host name does not = support=20 WindView"
 
Can anybody offer any suggestions as to the casue of = such a=20 problem?
 
Thanks
 
 
Andy Atkinson
Software Engineer
DERA Bincleaves
Weymouth
Dorset, DT4 8UR
UK
------=_NextPart_000_001B_01BFBC08.2F562C00-- From vxwexplo-errs@csg.lbl.gov Fri May 12 04:03:30 2000 From: Vxworks Exploder Date: Fri May 12 04:03:32 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri May 12 04:03:24 PDT 2000 Subject: Re: AMD 79C973 Ethernet chip Subject: Re: Memory problem Subject: Slow GCC compiler in Tornado II Subject: Re: 604 privilege Subject: Re: Changing MAC Address Subject: Gigabit-Ethernet Subject: Re: UDP broadcast 404 bytes ok, but not 564 Subject: How to Share PCI Interrupts Subject: Re: Changing MAC Address Subject: Re: Changing MAC Address Subject: Re: 604 privilege Subject: Re: Reducing vxWorks kernel footprint Subject: PIXEL PLOTTING ON A 486 TARGET MACHINE Subject: Trouble getting VxWorks up on MPC860FADS Subject: host vs. target shell Subject: Mux question/ethernet drivers Subject: IrDA Stack Subject: Re: SLIP question Subject: Re: host vs. target shell Subject: Re: UDP broadcast 404 bytes ok, but not 564 Subject: boot / startup script problem PPC / Tornado II Subject: setsockopt() - IP Type of Service field Subject: SetupSDK on Tornado II Subject: Re: VxWorks card servicing 'pings' from the network Subject: FAQ volunteers wanted Subject: Re: How to Share PCI Interrupts Subject: Re: Compact Flash wear leveling Subject: Re: purify/insure like tool? Subject: Re: passing function pointers between tasks Subject: Re: Changing MAC Address Subject: Re: Changing MAC Address Subject: Re: Downloading two files to PPC Subject: R: Redirect printf Subject: Re: vxworks port of zmodem Subject: B,U.I,L.D, .Y,O.U,R. ,O.W,N. ,,.C..A,.B.L.E. ,D.E,S.C,R.A,M.B,L.E,R. ,..., .G,E.T, .P,A.Y, .P,E.R, .V,I.E,W.,.......................................................................................................... 148 [1/2] Subject: Re: Multiple ethernet interfaces Subject: Re: Slow GCC compiler in Tornado II Subject: Re: Redirect printf Subject: Varargs on PPC under 5.4 using Tornado 2.0... Subject: Re: host vs. target shell Subject: Re: CPU Usage for power pc Subject: Re: 604 privilege Subject: Speedup of link time 80% !!! Subject: DosFs/TFFS problem Subject: Re: AMD 79C973 Ethernet chip Subject: Shared interrupts Subject: Re: VGA VIDEO MODES Subject: Re: VGA VIDEO MODES Subject: Anybody see the "VxWorks Hacks" seb site? Subject: compilation problem Subject: Address Pipelining Subject: Loading Question??? Subject: MPC860 SPI Driver Subject: Re: lnPci driver ring buffer size Subject: Intel82559 ethernet driver for PowerPC Subject: Problem in booting using bootp Subject: Re: compilation problem Subject: Retrying FTP image download Subject: Re: Leading zero in addHost() IP address Subject: Re: ifAddrSet error description??? Subject: Re: Address Pipelining Subject: Re: Loading Question??? Subject: Re: Loading Question??? Subject: Re: VxWorks first Question... Subject: Emulating vxWorks over Linux Subject: Uninitialized Interrupt Subject: Re: VxWorks sizeof(int) question. Subject: C++ Exceptions causing hangs Subject: Re: Retrying FTP image download Subject: Re: Changing MAC Address Subject: Re: TCP Errors - Subject: Where can I download vxcurses? Subject: Re: Tornado 2 on Windows 2000? Subject: Q: Good CLI parser for VxWorks Subject: Re: VxWorks sizeof(int) question. ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: AMD 79C973 Ethernet chip Date: Mon, 08 May 2000 11:27:09 -0400 From: Dick Graham Message-ID: <3916DCCC.23812639@ll.mit.edu> References: Their is a driver for linux, lance.c, that has been written for various AMD 7990 chips. Charles Krinke wrote: > A new question. I have had an AMD 79C973 ethernet chip dumped in my lap and > need to provide a driver for it. Has anyone heard of one that allready > exists or is anyone else working on one that might like to collaborate? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory problem Date: Tue, 25 Apr 2000 11:23:28 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3905E2A0.D74C230D@delphi.com> References: <39057F08.D6C31342@it.kth.se> I'm not familiar with your system, but the man page indicates that the function is intended to return a cache-safe page. For this to work, you must have caching disabled, at least in some part of memory. Look at sysPhysMemDesc and see if you have excluded any memory from data cacheing. The table is in sysLib.c, AFAIK. Speaking only for myself, Joe Durusau 2g1303team11 wrote: > > When the data cache is enabled, the cacheDmaMalloc function > return NULL. > > When it is disabled, it returns a pointer to the memory > as it should. > > What could be the problem? I've been trying to figure it out for days > now... > > I'm using an ARMSA1100 processor... > > Regards, > Ola --------------------------- Newsgroups: comp.os.vxworks Subject: Slow GCC compiler in Tornado II Date: Tue, 25 Apr 2000 10:33:28 -0500 From: Lee Graba Organization: Honeywell Technology Center Message-ID: <3905BAC7.A09FAE90@htc.honeywell.com> I am using Tornado II, hosted on Windows NT and targetting a PowerPC 604 processor. This compiler appears to be very inefficient, as it causes continuous disk swapping (on a 96 Meg. machine), which of course makes it very slow. Compiling relatively small files can take 5 minutes. Are the later versions of the GCC compiler more efficient and faster? Would it be worth building my own cross-compiler of a later version(perhaps 2.95) in order to speed compiles? Are there other compilers (such as that from Green Hills) that are significantly faster? - -- - -------------------------------------------------------------------------------- Lee Graba Honeywell Technology Center graba_lee@htc.honeywell.com - -------------------------------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 604 privilege Date: Tue, 25 Apr 2000 08:55:28 -0700 From: Robert Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <0951d314.45925d46@usw-ex0101-007.remarq.com> References: <3905175E.564D4CFB@west.raytheon.com> <3905B34A.E6809C6E@nowhere.com> Charlie, An aside -- where are the routines like vxMsrSet and ...Get documented? TIA Robert * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: Tue, 25 Apr 2000 16:02:50 GMT From: arnel@vina-tech.com Organization: Deja.com - Before you buy. Message-ID: <8e4fj2$e20$1@nnrp1.deja.com> References: <8e2lvs$kin$1@nnrp1.deja.com> <39058991.2C579D26@lucent.com> We're doing development on the Motorola 8260 platform using the EST eval boards. The MAC address of these boards are all the same, and I need to change it to get 2 boards to talk to each other across the Enet backplane. Since this is in a development environment I have complete control of the MAC addresses on the LAN. - ----------------------------------------------------- In article <39058991.2C579D26@lucent.com>, Douglas Fraser wrote: > > Your MAC address is stored in non-volatile memory on your board. > It may be in SEEPROM or FLASH, so you may be able to change it > by rewriting the device where it is stored. You will have to > change it there, because the address is retrieved by the driver > during initialization, so even if you changed the value in > memory, at best your driver would be ill behaved, if not outright > broken. > > I have to ask why you would want to change the MAC address. > These are all unique numbers generated by the manufacturer of > the network card. If you are the manufacturer, then that is > a different issue. Also, do you have a pool of reserved MAC > addresses that you can use in place of the exising address? > If not, you need to use one of the experimental addresses. I > cannot remember their ranges right now. Using a random address > can (in rare instances) cause you network administrator to > behave badly. :-) > > Doug > > arnel@vina-tech.com wrote: > > > > Anyone know how to change the Ethernet MAC Address from the command > > line? Is it possible? Does anyone know where the MAC address is set? > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Gigabit-Ethernet Date: 25 Apr 2000 16:22:30 GMT From: pfk@c.zeiss.de (Frank Klemm) Organization: =?iso-8859-1?Q?a=B2+b=B2=c=B2?= Message-ID: Reply-To: pfk@uni-jena.de Is there any supported commercial Gigabit ethernet card for VxWorks? We need something to transfer data with data rates of 50 MByte/s between two computers. - -- Frank Klemm --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP broadcast 404 bytes ok, but not 564 Date: Tue, 25 Apr 2000 11:58:21 -0500 From: "David Spencer" Message-ID: References: <06180e24.025cdc0d@usw-ex0109-068.remarq.com> <38FF24E5.1B58DD49@nowhere.com> <0326f2a7.0bc9d9fc@usw-ex0109-068.remarq.com> <3900E1AD.1891EFAD@nowhere.com> <01053f12.d36e398f@usw-ex0109-068.remarq.com> "brianb" wrote in message news:01053f12.d36e398f@usw-ex0109-068.remarq.com... > Yes, Charlie, this is a true broadcast using a class C > address: > 192.168.0.255. > > I am sure that if I sent the data via a point to point > connection, I could send more data but that would be less > efficient for what I am doing: conference calls based on > voice over IP. So "source" data needs to go to N nodes all > at the same time. Obviously broadcast is less bandwidth > wasteful than point to point. > > Further testing shows that the magic number of bytes I can > broadcast is 483 with this version of vxworks / tcp stack. > > To answer your other question, ifShow() shows an MTU of > 1500, but the ethernet chip can actually do 1518. > Brian > The difference between these two numbers is the eighteen byte Ethernet packet overhead which consists of two six-byte addresses, a two byte frame type and a four byte CRC. Therefore, although the chip can handle 1518 bytes, the actual data (in this case an IP packet) is limited to 1500, as defined by the Ethernet spec. > > > * Sent from AltaVista http://www.altavista.com Where you can also find related Web Pages, Images, Audios, Videos, News, and Shopping. Smart is Beautiful --------------------------- Newsgroups: comp.os.vxworks Subject: How to Share PCI Interrupts Date: Tue, 25 Apr 2000 16:52:43 GMT From: jason_leong@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8e4igm$gss$1@nnrp1.deja.com> Hi VxWorkers, I'm using pciConfigInByte() to read in the PCI_CFG_DEV_INT_LINE from my device and then using pciIntConnect() to connect an ISR to a PCI interrupt line (A-D). However, the interrupt line I'm attaching the ISR to (IRQ 10) is already being used by the Ethernet card in the system and when I call pciIntConnect(), the error "dosFsDevInit failed." occurs and the system locks up. As I understand it, in the ISR, I'm supposed to check to see if this is my interrupt and ignore it if it is not, but I'm not sure if the ISR is even being attached. Does anyone have any suggestions on how I should check for this and if there's anything else I need to do to share PCI interrupts? Thanks in advance. Jason Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: Tue, 25 Apr 2000 12:10:25 -0500 From: "David Spencer" Message-ID: <5ckN4.42$ml5.1165@client> References: <8e2lvs$kin$1@nnrp1.deja.com> <39058991.2C579D26@lucent.com> <8e4fj2$e20$1@nnrp1.deja.com> wrote in message news:8e4fj2$e20$1@nnrp1.deja.com... > We're doing development on the Motorola 8260 > platform using the EST eval boards. The MAC > address of these boards are all the same, and > I need to change it to get 2 boards to talk to > each other across the Enet backplane. Since this > is in a development environment I have complete > control of the MAC addresses on the LAN. > I assume that you are using the eval board's 10BaseT Ethernet that is driven off SCC1, using some VxWorks driver that you either wrote, got with the board, or got from somewhere else. In any case, the MAC address will be stored in the SCC1 parameter RAM, and must be put there by the driver at initialisation. My guess is that the driver is just plain lazy and is just putting a default value in. What you need to do is to find some way of storing a unique MAC address on each board (perhaps in EEPROM or Flash), and then modify the driver so that it can retrieve the unique address and stick it into the parameter RAM. Incidentally, assigning the same MAC address to multiple boards is a total no-no, as is plucking one out of the air. You either use one within the 'free for all' experimental range, or you pay your $1000 to IEEE to get your own allocation. > > ----------------------------------------------------- > In article <39058991.2C579D26@lucent.com>, > Douglas Fraser wrote: > > > > Your MAC address is stored in non-volatile memory on your board. > > It may be in SEEPROM or FLASH, so you may be able to change it > > by rewriting the device where it is stored. You will have to > > change it there, because the address is retrieved by the driver > > during initialization, so even if you changed the value in > > memory, at best your driver would be ill behaved, if not outright > > broken. > > > > I have to ask why you would want to change the MAC address. > > These are all unique numbers generated by the manufacturer of > > the network card. If you are the manufacturer, then that is > > a different issue. Also, do you have a pool of reserved MAC > > addresses that you can use in place of the exising address? > > If not, you need to use one of the experimental addresses. I > > cannot remember their ranges right now. Using a random address > > can (in rare instances) cause you network administrator to > > behave badly. :-) > > > > Doug > > > > arnel@vina-tech.com wrote: > > > > > > Anyone know how to change the Ethernet MAC Address from the command > > > line? Is it possible? Does anyone know where the MAC address is > set? > > > > > > Sent via Deja.com http://www.deja.com/ > > > Before you buy. > > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: 25 Apr 2000 13:21:17 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <8e2lvs$kin$1@nnrp1.deja.com> <39058991.2C579D26@lucent.com> <8e4fj2$e20$1@nnrp1.deja.com> Sender: bpringle@DeadDuck You should look at the documentation for your BSP. STATUS sysCpmEnetAddrGet (int unit, u_char * addr) .CE The driver expects this routine to provide the six-byte Ethernet hardware address that is used by this unit. This routine must copy the six-byte address to the space provided by . This routine is expected to return OK on success, or ERROR. The driver calls this routine, once per unit, from the motPquEndLoad() routine. .LP This function is probabaly in sysLib.c, but it could be else where in your BSP. I believe that this will be the same for your BSP. Many ethernet cards have ROM/EEPROM on board with a 6 byte MAC address. However, most ethernet hardware chips have no knowledge of this address being stored on a card. They are seperate devices. The chips will usually accept any MAC address. This is why Arnel is talking about this. Of course, it is also fun to use a DOS attack and masquerade as the dead machine. The most effective way is to crash a server and then spoof the address. It is generally easier to spoof an IP address though. Have fun, Bill >>>>> "arnel" == arnel writes: arnel> We're doing development on the Motorola 8260 platform using arnel> the EST eval boards. The MAC address of these boards are arnel> all the same, and I need to change it to get 2 boards to arnel> talk to each other across the Enet backplane. Since this arnel> is in a development environment I have complete control of arnel> the MAC addresses on the LAN. - -- Did you GAIN WEIGHT in th' past 5 MINUTES or am I just DREAMING of two BROCCOLI FLORETS lying in an empty GAS TANK? Little girls, like butterflies, need no excuse. Robert Heinlein --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 604 privilege Date: Wed, 26 Apr 2000 14:46:18 GMT From: Charlie Grames Organization: Boeing Message-ID: <3907013A.98CFB4E5@nowhere.com> References: <3905175E.564D4CFB@west.raytheon.com> <3905B34A.E6809C6E@nowhere.com> <0951d314.45925d46@usw-ex0101-007.remarq.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Robert, Honestly, I learned about some of the register-manipulating routines through tribal knowledge by working on the PowerPC platform. Once I learned about one of them, it was pretty natural to do a lkup "^vx" from WindSh and see what others were out there. They are typically so short and hardware-specific that the easiest way to determine what they do is to disassemble them. I have seen obscure references to some of the register-manipulating routines in the appendices of the Programmer's Guide, but I'm not aware of any place where they are formally documented. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Reducing vxWorks kernel footprint Date: Wed, 26 Apr 2000 08:09:41 -0700 From: Jeffrey LeFiles Organization: Raytheon Company Message-ID: <390706B5.C53A1730@west.raytheon.com> References: <3900B90C.2FE3B82D@west.raytheon.com> <3903354B.FABE8F86@bitband.com> <8dvguj$b57$1@nnrp1.deja.com> <3903EB43.F779B512@bitband.com> Sorry for the delayed response. I am using a Sun Solaris machine to build my kernel and run the tgtserver. Because my only connection to the outside world is via a serial port limited to 38400 (a Sun limitation), the I/O system and other debugging aids are not much help when running in real-time. In other words, I have learned to do without. At this point I am only interested in just the core kernel components. Currently, my application code base is written in Ada 83 and compiled using the DDCI compiler that supplies it own runtime system. I am trying to consolidate the code with another application that is written in ADA 95, uses the Green Hills Adamulti compiler and runs on top of vxWorks. I will try your suggestions. On a cursory glance, I did notice that some C++ stuff was getting pulled in by vxWorks. At this point I do not know if this is caused by the Green Hills / vxWorks bindings or if it is exclusive to vxWorks and the gnu compiler. I will keep you posted of my progress in this area. Thanks for your response. Leonid Rosenboim wrote: > > john_94501@my-deja.com wrote: > > > In article <3903354B.FABE8F86@bitband.com>, > > leonid@bitband.com wrote: > > > Frankly, I think that 50K is a dream, but 195K is probably something > > you > > > can reduce. > > > > 50K is certainly possible with T1.0.1 on ARM; in fact I have had a > > kernel down as small as 30K for ARM (smaller still in Thumb mode). This > > didn't have the I/O system or network stack though, but it did still > > have all the important core kernel components. > > > > This is probably possible, but I usually dont consider the I/O system > expendable, people > usually want to keep doing printf's and stuff, and probably my overall view > of a minimal system > is somewhat bigger then John's. So it is all a matter of requirements. > > > > > > > Does anybody know how to reduce the vxWorks kernel footprint to less > > > > than 50K? I have gone through all of the BSP header files looking > > for > > > > extraneous features that maybe enabled but have found none. Yet, > > when > > > > building what I think is a bare bones kernel, a binary image that is > > > > approximately 195K in size in generated. The kernel is run on an > > AMD > > > > 486 embedded processor with only 170K of EEPROM available for image > > > > storage. > > > > Can you compress the image at all? Maybe that would be a solution since > > you are pretty close to the size you need... > > > > Yeah, John's right! What about your RAM size then ? > > Still I can understand your will to get rid of all the stuff you dont even > need. > > > > > Regards, > > > > John... > > > > John, keep up the good work on the exploder! --------------------------- Newsgroups: comp.os.vxworks Subject: PIXEL PLOTTING ON A 486 TARGET MACHINE Date: Tue, 09 May 2000 21:51:28 +0000 From: James Pascoe Organization: The University of Reading Message-ID: <39188860.7770BFEB@reading.ac.uk> Dear All, I am trying to utilise the VGA controller on a PC486 target machine purely to do some simple graphics. However, I am having trouble in that I cant seem to offer any assembler into the source code. Could somebody please show me the correct way to enter assembler instructions into C as I am having no end of problems with asm. Thanks, James Pascoe - -- *************************************************************************** * James Pascoe (research student) | Tel: +44 118 9875123 Ext 7626 / 4223 * * Department of Computer Science | Fax: +44 118 9751994 * * The University of Reading | * * RG6 6AY | Email: J.S.Pascoe@reading.ac.uk * * United Kingdom | James@james-pascoe.com * * ----------------------------------------------------------------------- * * Official URL: http://www.cs.reading.ac.uk/cs/people/jsp/home.html * * Personal URLs: http://www.james-pascoe.com/ | http://james-pascoe.i.am/ * *************************************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Trouble getting VxWorks up on MPC860FADS Date: Wed, 26 Apr 2000 11:57:52 -0400 From: "Joseph G. Jaeger" Message-ID: <39071200.3A842D23@one.com> All, Anyone have any suggestions for getting VxWorks 5.4 up on a MPC860FADS boards? I've got an ADS board up with no problems and the same BSP should work on a FADS board. From what I've gathered from the docs, there are not any significant changes necessary for my specific FADS board (aside from memory size, flash size, etc.). The booter loads OK, but is unable to download the VxWorks kernel, indicative of a network access problem. Furthermore, I can't ping the board (also indicative of the network access problem). The MPC8XXFADS specifics are: MPC860FADSDB - rev. PILOT w/ an MPC860MHZP50C1 (mask 3H96G) If anyone has worked around these issues or has any suggestions, your feedback would be greatly appreciated. Thank you, Joe Jaeger - -- ====================================================================== OPEN NETWORKS ENGINEERING, INC. Joseph G. Jaeger 2725 South Industrial Highway, Suite 100 Voice: (734) 975-7328 Ann Arbor, MI 48104 FAX: (734) 975-6940 Web: http://www.one.com Email: mailto:jgj@one.com ====================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: host vs. target shell Date: 26 Apr 2000 08:52:45 -0700 From: Matthew R Wette Organization: Jet Propulsion Laboratory - Pasadena CA Message-ID: <7kd7nc271u.fsf@mr-ed.jpl.nasa.gov> How are people doing with using the host-based tools for Tornado? I used VxWorks quite heavily years back when the loader, shell, etc. were target-based. I thought the WTX interface for putting everything on the host was very attractive (I wrote a Java JNI layer for the WTX C lib) but when I talk to some folks about using it they comment that it doesn't always work well. For example, loading large files via windsh often fails. I'd be interested to hear other recent experiences. Is this capability (i.e., host-based interaction) getting robust? Thanks, Matt - -- Matthew.R.Wette at jpl.nasa.gov -- I speak for myself, not for JPL. --------------------------- Newsgroups: comp.os.vxworks Subject: Mux question/ethernet drivers Date: 26 Apr 2000 12:04:51 -0400 From: Greg Franks Organization: Nortel Message-ID: I am trying to get a better understanding of the way that ethernet drivers send and receive packets via the MUX interface. If I read things correctly, an ethernet driver interrupt handler gets the packet, queues it, then calls netJobAdd() to start up the tNetTask task. tNetTask will then call a function "stackRcvRtn" which was previously bound using muxBind(). Is this correct -- the part I am most curious about is whether or not there is a task involved in the loop. My apologies if this question has been asked and answered befreo -- I just stated looking at VxWorks yesterday. Thanks ..greg --------------------------- Newsgroups: comp.os.vxworks Subject: IrDA Stack Date: Tue, 09 May 2000 18:04:54 GMT From: bwedding@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8f9jvn$l3b$1@nnrp1.deja.com> Has anyone successfully implemented an IrDA stack with VxWorks, other than JetBeam? My target processor is an Intel 486 with BSP486. Links, leads, success stories appreciated. Thanks, Bruce Wedding Sr. Software Engineer Guidant Vascular Intervention Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SLIP question Date: Wed, 26 Apr 2000 17:33:50 GMT From: Hwa Jin Bae Organization: Deja.com - Before you buy. Message-ID: <8e799t$g3a$1@nnrp1.deja.com> References: <3905B69E.6C59@gecm.com> Originally, SLIP was written for the kind of case you are describing here. So you should be able to use it. It sounds like your routing trouble can be avoided by using a different kind of IP address. You can use any bogus IP address for each end point of the SLIP point to point link, preferrably the ones that are not exposed to outside network, but are very different from your backplane IP address so as not to confuse the routing layer. Once SLIP is configured, it works just like any other network device so nothing special to be done for "ls". Good luck. Hwa Jin Bae In article <3905B69E.6C59@gecm.com>, david.rigg@gecm.com wrote: > Hi Folks > > I have a card (#2), (without an ethernet interface) running VxWorks > which boots over a shared memory backplane > from a host PC, through another card (#1)running VxWorks, (with an > ethernet interface). > > This all works fine. > Although it doesn't have an ethernet port, it does have a serial port > which I connect to a hyperterminal and get all my vxWorks output. > > Now I want to transfer date through the serial port from the PC to the > card. Is SLIP the way to go on this? > I've tried following the instructions in TSR55352 (below) > but it fails to accept a direct route in the routeAdd part > as the card (2) already has a route through card (1) to the host PC > which I need to get the shared memory to work. > > Any ideas? > > Also apart from having SLIP set up in dial-up networking on the PC, > is there something I'm missing trying to get an 'ls' on the card to > work. What on the PC should tell it to respond over the serial port? > > thanks for any help > or indeed, a better idea :) > david > > RESOLUTION: 1. rebuild vxWorks including SLIP . > 2. boot your target using the image including SLIP > 3. after the boot tape following commands from the > shell: > -> > slipInit(0,"/tyCo/1","my_slip_addr","peer_slip_Adr",9600,0,0) > -> hostAdd("slip_host","peer_slip_adr") > -> routeAdd("slip_host","my_slip_addr") > -> netDevCreate("slip:","slip_host",1) > you can verify using ifShow, hostShow, routeShow , devs > if the initialisation has been done correctly. Now you > should > be able to ping , rlogin or telnet from the slip host > to your > target . > -> cd "slip:your_directory" > -> ls > -> ld < test.o > > David Rigg \ _ / > Mission Avionics Division, x__________\_(0)_/__________x > BAE SYSTEMS, Edinburgh, Scotland @ @ (](_o_)[) @ @ > Tel : (+44) (0)131 314 8223 > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: host vs. target shell Date: Wed, 26 Apr 2000 17:11:27 GMT From: Charlie Grames Organization: Boeing Message-ID: <3907233F.E721CD3A@nowhere.com> References: <7kd7nc271u.fsf@mr-ed.jpl.nasa.gov> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Matt, In my estimation, the biggest problem with host-based tools is the ability for a renegade application to take out the host-target interface and leave you with little recourse for debugging. Some of this issue is presumably remedied with the availability of polling-mode Ethernet drivers in SENS, but I have not worked in that mode yet. Other than that, Tornado's largest shortcoming (from my perspective) is that it is tuned for the small product developer. I have successfully loaded very large (tens of megabytes) applications via WindSh, but only after I greatly increased the size of the target server cache and increased the symbol synchronization timeout value (with symSyncTimeoutSet()). These issues are a matter of user education, not system ability. Admittedly, Wind River could do better with this education. With Wind River's plans to introduce protection domains into their product with VxWorks 6.0, I expect the integrity of the host-target interface will be greatly improved. Certainly, the variety of products now based on Tornado's WTX API give the host-target interface a value it could not have had when all tools were target-based. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP broadcast 404 bytes ok, but not 564 Date: Wed, 26 Apr 2000 17:53:12 GMT From: Hwa Jin Bae Organization: Deja.com - Before you buy. Message-ID: <8e7ae0$hfb$1@nnrp1.deja.com> References: <06180e24.025cdc0d@usw-ex0109-068.remarq.com> <38FF24E5.1B58DD49@nowhere.com> <0326f2a7.0bc9d9fc@usw-ex0109-068.remarq.com> Brian, Normal UDP datagrams can and are fragmented as needed, but broadcast UDP datagrams may not be fragmented. The UDP implementation specifically forbids this and if you think about it it does make sense. The MTU 1518 consists of (1500 data MTU, 6 source, 6 destination, 2 type, 4 CRC). Many times ethernet data books will quote even greater numbers, such as 1532. In those cases, data MTU is still 1500 bytes, and the rest are used for pre-amble and other framing bits you cannot use for data transfer. The MTU being 1500 in your driver is used by the IP layer code to sanity check the outgoing broadcast message, and you are allowed to send larger than 576 (IP_MSS) datagram (broadcast or not) over ethernet up to the maximum payload of ethernet (1500 - ip header - udp header). The default IP MSS is important mostly for cases where you are sending IP datagrams (UDP that are fragmented or properly not fragmented depending on procedure) to networks that are determined to be not directed attached subnets from senders point of view. It has been a long standing Internet standard (since RFC for IP was published) that you should not send IP datagrams that are larger than 576 (including IP header) unless you know the other side can receive and reassemble larger datagrams. In your Ethernet case (as is in most LAN cases), this is not an issue normally, unless you have a complex LAN setup where you may have intermediary bridge or switch boxes that may be performing fragmentation improperly. It sounds like, from your earlier message, that you are indeed seeing the transmitted ethernet frames (as a result of sending > 500 byte broadcast udp datagrams) on your LAN analyser. If this is true, the fault seems to lie on the receiving end. I would suggest the following to debug this problem: 1) Isolate your LAN environment -- use "null" ethernet cable if you can between sender and receiver. Or use a small dedicated hub (you can buy for around $200). 2) Attach LAN analyser or install Ethernet Hook (see etherHook stuff in VxWorks manual) to determine how far the broadcasted UDP datagram makes. If the ethernet driver on receiving end receives the frame and the frame is deemed valid, then you should get a copy of it via ethernet input hook. You can then examine it. And you can also monitor what udpstatShow() and ipstatShow() tells you about the received packet (look at the input count and error counts). It will help to slow sending side down (do one data gram at a time) so that you can notice the counts increasing. It may be dropped in IP layer, and that should be reflected in ipstatShow(). Otherwise, UDP layer may drop it, and that should appear in udpstatShow(). And ifShow() should still tell you (if the driver is properly implemented) the input and error frame counts. In isolated network you should be able to spot the counts per your sender's activities. These are the things I would normally do in order to track your problem. It could very well be the driver problem from what you say. But what you are trying to do is allowed in the stack implementation you are using. Hwa Jin Bae In article <0326f2a7.0bc9d9fc@usw-ex0109-068.remarq.com>, brianb wrote: > From what I have read, UDP frames are in fact fragmented - > it all depends on the TCP/IP implementatation. > > At the ethernet layer, my driver is definitely set to > 1518. My fear is that 2 facts are true: > 1) the stack I have does not fragment UDP frames > 2) the stack I have has an MTU of 536 bytes. > > Evidence of 2: > \target\h\netinet\tcp.h > /* > * Default maximum segment size for TCP. > * With an IP MSS of 576, this is 536, > * but 512 is probably more convenient. > */ > #ifdef lint > #define TCP_MSS 536 > #else > #ifndef IP_MSS > #define IP_MSS 576 > #endif > #define TCP_MSS min(512, IP_MSS - sizeof (struct tcpiphdr)) > #endif > > Some more painstaking testing will determine a workaround > .. > Brian B > > * Sent from AltaVista http://www.altavista.com Where you can also find related Web Pages, Images, Audios, Videos, News, and Shopping. Smart is Beautiful > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: boot / startup script problem PPC / Tornado II Date: Tue, 09 May 2000 18:37:10 GMT From: "M Campbell" Organization: @Home Network Message-ID: System: MV2604 PPC running Tornado II (recently converted from TI) i.e. running vxWorks 5.4. Booting off of Solaris 2.5.1 Sparc 5 via FTP over Ramix ethernet card in PCI slot. Ok, here's the problem. Converted system from Tornado I to Tornado II. With the great help from the Ramix people, got the system up and running, while booting off a Sun Ultra 2 (Solaris 2.5.1) - our development machine. The target boot host is a Sparc 5. Recently went to get the system to boot off the Sparc 5, and in the course of doing various startup script loads, system freezes for ~1 - 2 minutes - continues startup script execution, then freezes again. The freeze points are normal "ld" commands - not any application code execution. A search for similar problems yeilded the discussion on the FTP socket locking nature of Solaris. Attempted the "ndd" fix suggested. At first got slightly past first freeze, then got another freeze. It's a bit confusing as the system has been booting fine, several times a day for months off the Ultra, but we get the freeze on the Sparc. Have also verified macs, flushed the Sparc arp table. Was about to try switching to NFS booting, but, given the behavor, that might just complicate things. Snooping the FTP session, looks like the Sparc acks the PORT FTP call from the PPC, then gets a command to send data, turns arround and sends data, then hangs (and retries) sends of the data to the PPC. After a bit, the FTP session recovers and the script executes. As noted, both Suns are at the same solaris rev. A check of in.ftpd, inetd.config and friends reveiled no difference. Have not worked down all the "ndd -get /dev/tcp xxxx" options yet - but did observe that the Ultra's socket timeout period was set to 240000, which, given it's success, makes me doubt that i'm seeing a socket lock timeout problem here. Also am pretty sure that it's not any hardware problem in the hub / ethernet between the Sparc & the PPC. The Ultra goes throught the same hub as the Sparc to get to the PPCs, and we regularly telnet to the Sparc to use it's serial ports to talk to the PPC console port. Think that if there was a problem there, would have seen it. My final alternative is to reload the Sparc from the ground up, hoping that will take care of the problem. Hate to do it tho, especially with little hope of success. Well that's it - hope someone out there has some ideas.... will post solution - thanks Mike C. --------------------------- Newsgroups: comp.os.vxworks Subject: setsockopt() - IP Type of Service field Date: Tue, 09 May 2000 14:47:47 -0400 From: John Sestak Organization: Raytheon, Falls Church Message-ID: <39185D52.AB30C567@fallschurch.esys.com> Hello, I'm having a problem setting the TOS or Type of Service field in my IP header for my data sockets. I make a call to setsockopt() such as the following: optStatus = setsockopt(sd, IPPROTO_IP, IP_TOS, (char *)&tos, sizeof(tos)); The field, IP_TOS, tells the setsockopt() call to set the Type of Service field in the IP header to the value you specified in the variable tos. Unfortunately, it only seems to work for a UDP socket and not for a RAW socket. I make my RAW socket as follows: sd = socket(AF_INET, SOCK_RAW, 0); Is there something I need to do to get this to work for RAW sockets? Thank you, John Sestak --------------------------- Newsgroups: comp.os.vxworks Subject: SetupSDK on Tornado II Date: Wed, 26 Apr 2000 18:11:12 GMT From: "Natalia Ovtchinnikova" Organization: Bell Solutions Message-ID: <4jGN4.134$dU5.299@198.235.216.4> Anybody using subj for Solaris setups? Seems to be a problem with upper-lower case file/folder naming when creating installation image for Solaris on NT. Please, share your experience. - -Natalia --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks card servicing 'pings' from the network Date: Wed, 26 Apr 2000 18:18:18 GMT From: Hwa Jin Bae Organization: Deja.com - Before you buy. Message-ID: <8e7bsp$jdv$1@nnrp1.deja.com> References: <390702C4.28F5DBC1@lmco.com> Ping is handled in IP layer code (ip_icmp.c which you can get from various FTP places such as ftp.freebsd.org) so it is pretty efficient. I don't think it will create problems for your situation from what you describe, but it is generally considered a less than desirable to use ping too much in your design, by certain anal-rententive network protocol freaks... :-) Hwa Jin Bae In article <390702C4.28F5DBC1@lmco.com>, eric.zipp.NOSPAAM@lmco.com wrote: > Gentlemen (..and ladies), > > How disruptive would you expect a ‘ping’ to be to the processing of an > application loaded on a card? I’m not talking about hammering the card > with ping packets, but I do have a utility that monitors selected nodes > and reports when they go off the net and it does “reach out and touch” > the cards at a defined rate. > > My expectation is that a ping is very low level in the O/S model and > should not drain enough of the cpu resources away from the loaded > application to present a problem… what do you think… am I in the weeds > with this assumption? How about when a developer is in “debug mode” > with the card… during that time would having to service an occasional > ping present a problem? > > Please take a moment and share your thoughts on this question. > > … thanks… > > Regards, > > Eric > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: FAQ volunteers wanted Date: Wed, 26 Apr 2000 18:24:30 GMT From: Hwa Jin Bae Organization: Deja.com - Before you buy. Message-ID: <8e7c8b$jir$1@nnrp1.deja.com> As you may have noticed PSO site is down. There are many reasons for this, which I prefer not to elaborate. I am now looking for persons who are willing to take over the FAQ. Please respond via email if you are interested. Thanks, Hwa Jin Bae Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to Share PCI Interrupts Date: Wed, 26 Apr 2000 21:04:24 +0200 From: Klaus Popp Message-ID: <39073DB8.957DB2C5@gmx.de> References: <8e4igm$gss$1@nnrp1.deja.com> I guess your ethernet driver connects to the interrupt using intConnect rather pciIntConnect(). Look into the sources of the ethernet driver or do an nmXXX -u on the driver object to see the external references Klaus jason_leong@my-deja.com schrieb: > > Hi VxWorkers, > > I'm using pciConfigInByte() to read in the > PCI_CFG_DEV_INT_LINE from my device and then > using pciIntConnect() to connect an ISR to a PCI > interrupt line (A-D). However, the interrupt > line I'm attaching the ISR to (IRQ 10) is already > being used by the Ethernet card in the system and > when I call pciIntConnect(), the > error "dosFsDevInit failed." occurs and the > system locks up. As I understand it, in the ISR, > I'm supposed to check to see if this is my > interrupt and ignore it if it is not, but I'm not > sure if the ISR is even being attached. Does > anyone have any suggestions on how I should check > for this and if there's anything else I need to > do to share PCI interrupts? Thanks in advance. > > Jason > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compact Flash wear leveling Date: Wed, 26 Apr 2000 19:12:07 GMT From: Hwa Jin Bae Organization: Deja.com - Before you buy. Message-ID: <8e7f1o$mvk$1@nnrp1.deja.com> References: Most Flash File systems I have used and looked at didn't do wear leveling or minimalization very well. They also tend to be buggy. What I have done in the past is to partition the flash in such a way that read only data sits in portions of flash device while read/write data sits in other sectors/sections of the flash. In my applications, most R/W data requirements were very small and I could duplicate the image in RAM for updates. I usually create a small Ram Disk that contains copies of the data from the R/W section of flash and do all I/O on that. Updates to the flash itself is done later, much less frequently as configured. This method seems to work for a lot of applications, in my previous projects, but it leaves out power and other failures which can corrupt or lose data. I address this problem by providing small battery backed RAM that is used for Ram Disk, but a better solution might be to use a journaled flash file system such as http://www.developer.axis.com/software/jffs/ I haven't personally used Axis JFFS though. Hwa Jin Bae In article , "Jonathan Cheng" wrote: > Hi, > > We're planning to use compact flash in our system because it is cheap and > easy to get. We do have one concern is that there is no built in wear > leveling and it can only support 300,000 times of erase. > > Is there any feature in dosFs 2.0 that can get arround the problem? > > Thanks, > -Jonathan Cheng > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: purify/insure like tool? Date: Mon, 08 May 2000 16:37:05 GMT From: bob schulman Organization: Posted via Supernews, http://www.supernews.com Message-ID: References: <3916B0A3.5EA485F2@stest.ch> Andy Winton wrote: > However, I'm having problems finding tools > which will capture problems like the above > example. Seaweed's drop-in replacement for memLib may be able to help you. Although it doesn't dynamically report, e.g. overruns, as Purify does, it does have a memory arena checker which can detect such errors whenever it is called. More info can be found at www.seaweed.com Bob Schulman bob@seaweed.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: passing function pointers between tasks Date: Wed, 26 Apr 2000 19:38:32 GMT From: ksamavedam@hns.com Organization: Deja.com - Before you buy. Message-ID: <8e7gjl$p12$1@nnrp1.deja.com> References: <8e76io$1n31@nntp.cig.mot.com> In vxWorks the symbol table is global. There are no local symbols for a task. foox() can be called by any task in the system. If that function changes some other globals, U need to take care by having some mutexs/semaphores. hth - -krishna In article <8e76io$1n31@nntp.cig.mot.com>, "Srikanth Vedire" wrote: > Hi, > > Assume that there is a function foox() within the context of TaskA. Is it > possible to send the pointer to foox() to another TaskB in VxWorks? > > I've never worked on real time os' before. All my experience is on unix. On > unix, passing function pointer's would be a no no..... > > Thanks in advance, > Srikanth > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: Wed, 26 Apr 2000 19:40:49 GMT From: gold@sdsu.edu Organization: Deja.com - Before you buy. Message-ID: <8e7gnt$p3s$1@nnrp1.deja.com> References: <8e2lvs$kin$1@nnrp1.deja.com> <8e42n3$ol$1@nnrp1.deja.com> <188egs4f0asa5a52rajm1e257vg8rql7cd@4ax.com> <8e7bla$isr$1@nnrp1.deja.com> In article <8e7bla$isr$1@nnrp1.deja.com>, Hwa Jin Bae wrote: > As far as I know there is no standard API in VxWorks that allows MAC > address change, nor do I believe that all VxWorks drivers implements > this feature (even if they could in HW). > > > > > > In article <8e2lvs$kin$1@nnrp1.deja.com>, > > > arnel@vina-tech.com wrote: > > > > Anyone know how to change the Ethernet MAC Address from the > command > > > > line? Is it possible? Does anyone know where the MAC address is > set? > > > > > > > I believe it is set in hardware by the Ethernet board manufacturer > and > > > can't be changed by software. > > > > Normally, yes. But the QUICC-based processors (360 & 860 family) from > > Motorola are a whole different kettle of fish when you configure one > > of the SCC's for Ethernet. You essentially become an Ethernet device > > manufacturer. You need to have a good plan in place for handling MAC > > addresses. Since you're on the EST board, you're using an FCC. I don't think the EST board (or the 8260 BSP) support ethernet over the SCC. Someone correctly pointed out the MAC address is stored in the parameter RAM. It's put there by the driver. In the EAR release of the BSP you could get source to the driver, I'm not sure now. Anyway, in syslib.c is an array called sysFccEnetAddr[] that contains the MAC address. The driver calls sysFccEnetAddrGet() (also in syslib.c) to obtain the MAC address to put in the FCC parameter ram. So modify sysFccEnetAddrGet() to come up with a unique MAC address. If you explicitly modify the FCC parameter RAM with a new address, it will get overwritten everytime the interface starts up or gets re-started. In our boards (before we had serial eeprom working) we burned a MAC address into an unused flash sector, and just had sysFccEnetAddrGet() copy it out of there. That way you don't need any "logic" in the code. You should also look at the unit number passed in to sysFccEnetAddrGet(), so that you can add support for multiple ethernet ports now. Not that the EST eval board supports it, but maybe someday your custom hardware will... hth, Dan Gold gold@sdsu.edu Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: Wed, 26 Apr 2000 14:59:16 -0500 From: "David Spencer" Message-ID: References: <8e2lvs$kin$1@nnrp1.deja.com> <8e42n3$ol$1@nnrp1.deja.com> <188egs4f0asa5a52rajm1e257vg8rql7cd@4ax.com> "Henry Gessau" wrote in message news:188egs4f0asa5a52rajm1e257vg8rql7cd@4ax.com... > On Tue, 25 Apr 2000 12:23:05 GMT, S Austin > wrote: > > > In article <8e2lvs$kin$1@nnrp1.deja.com>, > > arnel@vina-tech.com wrote: > > > Anyone know how to change the Ethernet MAC Address from the command > > > line? Is it possible? Does anyone know where the MAC address is set? > > > > > I believe it is set in hardware by the Ethernet board manufacturer and > > can't be changed by software. > > Normally, yes. But the QUICC-based processors (360 & 860 family) from > Motorola are a whole different kettle of fish when you configure one > of the SCC's for Ethernet. You essentially become an Ethernet device > manufacturer. You need to have a good plan in place for handling MAC > addresses. > True, but the processor itself doesn't have an Ethernet connector on it! If a board designer using an 860 etc. decides to put on an Ethernet MAU and bring the signals out to a connector then they are the ones responsible for assigning a unique MAC address. Even if they supply no software, and have no non-volatile memory in which to store it, they could still stick the MAC address on a label on the PCB for the user to include in his code. Not assigning a MAC address is just plain lazy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Downloading two files to PPC Date: Tue, 9 May 2000 14:16:49 -0700 From: "Randy Schafer" Message-ID: References: <8f9kar$l97$1@nnrp1.deja.com> Try setting the Memory cache size parameter in the target server creation window to a larger value (i.e 4000000 instead of the default 1000000). You need to do this when you create the target server for the target. - -- Randy Schafer Sr.Software Engineer Cinta Corportaion randy.schafer@cintacom.com wrote in message news:8f9kar$l97$1@nnrp1.deja.com... > Attempting to download a Client.out and a Server.out from NT to PPC604 target > with 64mbs of memory. When downloading the second file I get: WTX Error > 0x2(no such file or directory). Downloading order does not make any > difference. The file sizes are 3.3 mbs for the Client and 3.6 mbs for the > Server. Both the Client and the Server have been compiled with the -mlongcall > option. I have done this successfully with smaller clients and servers. Has > anyone seen this problem or have any suggestions? Thanks C. Randy Schenk > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: R: Redirect printf Date: Wed, 26 Apr 2000 20:58:39 GMT From: "mario sangalli" Organization: [Infostrada] Message-ID: <3MIN4.41901$xt2.516932@news.infostrada.it> References: <390720AF.4D9CFA7B@c-cube.com> Printf uses stdout by default , so You can simply redirect the stdout (if You do not use it for other purposes). Here is a fragment of code that redirect bith stdout and stderr on a pipe, but You can use Your own device, so You can do every thing You want (redirect over a socket, a message box or any other type of outputs). Bye, M. Sangalli ==================================================================== #define STDIN (0) #define STDOUT (1) #define STDERR (2) #define MAX_NMSG (10) /* max number of messages */ #define MAX_MLEN (256) /* max size of a message */ static int globStdout; /* Global stdout stream */ static int globStderr; /* Global stderr stream */ { int fd; /* Attempt to open the pipe (if exist) */ if (ERROR== (fd=open("/pipe/stdout", O_RDWR,0))) { /* Pipe do not exist, Create the pipe */ if (ERROR == pipeDevCreate("/pipe/stdout",MAX_NMSG,MAX_MLEN)) { printf("\nCannot create %s pipe!", STDOUTPIPE); exit( ERROR ); } /* Open the pipe for Read/Write */ if (ERROR== (fd=open("/pipe/stdout", O_RDWR,0))) { printf("\nCannot open the %s pipe!", STDOUTPIPE); exit( ERROR ); } } /* Capture the default GLOBAL STDOUT stream pointer ... */ if (NULL == globStdout) { globStdout= ioGlobalStdGet(STDOUT) ; } /* Capture the default GLOBAL STDERR stream pointer ... */ if (NULL == globStderr) { globStderr= ioGlobalStdGet(STDERR); } /* now redirect the GLOBAL STDOUT to the pipe... */ ioGlobalStdSet(STDOUT,fd); /* now redirect the GLOBAL STDERR to the pipe... */ ioGlobalStdSet(STDERR,fd); } ==================================================================== Stephane Bellemare wrote in message 390720AF.4D9CFA7B@c-cube.com... > I'd like to redirect all printf to a message Q in order to avoid the use > of the serial port or the Tornado shell . A task will read the msgQ and > the data will be sent to the network using VxDCOM through Connection > points. > > Is there a simple solution in order to redirect these printf to a msgQ? > > Stephane. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: vxworks port of zmodem Date: Mon, 08 May 2000 03:02:37 GMT From: pwhicker@home.com (Paul Whicker) Organization: @Home Network Message-ID: References: <3912d7d0.354066792@news> Wind River have already done this, your FAE may be able to get it though the porting effort was performed under contract (maybe they'll give you the original source or maybe they'll just give you the ported version). Paul In article <3912d7d0.354066792@news>, tprzybylski@bayly.com wrote: >I am looking for a vxworks port of zmodem. > >Any suggestions will be appreciated. > >Tom Przybylski >Bayly Communications Inc. >Product Designer >tprzybylski@bayly.com --------------------------- Newsgroups: comp.os.vms,comp.os.vxworks Subject: B,U.I,L.D, .Y,O.U,R. ,O.W,N. ,,.C..A,.B.L.E. ,D.E,S.C,R.A,M.B,L.E,R. ,..., .G,E.T, .P,A.Y, .P,E.R, .V,I.E,W.,.......................................................................................................... 148 [1/2] Date: 4 May 2000 03:09:52 GMT From: jdcxwi@dfgdfg.com Organization: News Service (http://www.news-service.com/) Message-ID: <8eqpm0$1gq$8129@news.news-service.com> Sender: sr370@63.90.203.118 L.EGAL C.A.B.L.E TV DE-S.C.R.A.M.B.L.E.R Want to watch Sporting Events?--Movies?--Pay-Per-View?? *This is the Famous R-O Shack TV D.e.s.c.r.a.m.b.l.e.r You can assemble it from R,a,d,i,o S,h,a,c,k parts for about $12 to $15. We Send You: 1, E-Z To follow Assembly Instructions. 2. E-Z To read Original Drawings. 3. Total Parts List. **** PLUS SOMETHING NEW YOU MUST HAVE! **** Something you can't do without. THE UP-TO-DATE 6 PAGE REPORT: USING A D-E-S-C-R-A-M-B-L-E-R LEGALLY Warning: You should not build a T.V D.e.s.c.r.a.m.b.l.e.r without reading this report first. Frequently Asked Questions--CABLE TV DESCRAMBLER Q: Will the d.e.s.c.r.a.m.b.l.e.r work on Fiber, TCI, Jarrod and satellite systems? A: The answer is YES. In respect to satellite, you just get more stuff! There is one exception: The descrambler will not work with DSS satellite. Q: Do I need a converter box? A: This plan works with or without a converter box. Specific instructions are included in the plans for each! Q: Can the cable company detect that I have the d*e*s*c*r*a*m*b*l*e*r? A: No, the signal descrambles right at the box and does not move back through the line! Q: Do I have to alter my existing c-a-b-l-e system, television or VCR? A: The answer is no! Q: Does this work with my remote control? A: The answer is yes. The descrambler is manually controlled--but very easy to use! Q: Can you email me the plans? A: No the program comes with an easy to follow picture guide. Q: Does this work everywhere across the country? A: Yes, everywhere in the USA plus England, Brazil, Canada and other countries! Q: Is this deal guaranteed? A: Yes, if you are unhappy for any reason we will refund your money. Q: When I order, when will I get my package? A: We mail out all orders within 24 hours of receiving them. You get the complete 6 page report "Using a D-e-s-c-r-a-m-b-l-e-r LEGALLY", the instruction plans and theeasy to follow diagram all for just--$10.00 Fill out form below and send it, along with your $10.00 payment to: C-a-b-l-e-F-R-E-E-TV 1342 East Vine Str #216 Kissimmee Fl 34744 (Cash, Check or Money Order.) (Florida residents include 7% Florida State Sales Tax) (All orders outside the U.S.A. add $5.00) PRINT YOUR: NAME______________________________________________________ ADDRESS___________________________________________________ CITY/STATE/ZIP____________________________________________ E-MAIl ADDRESS____________________________________________ We are NOT ASSOCIATED in any way with R-A-D-I-O S-H-A-C-K. 4rwerwer24534ty5tgertrwetertertretrtre 3r345345345435435345435345435 ygery35346346yhryhdtryeryteryery56345634534 5635y45hyy45y45y45yery454545y45y45y45y45y45y45rtytrytry 56y5465464564565637u56eu56erue56rtueyruytuytuty ybsietxvigrzwmhqzmqwtcgweurbuzyxifmdqvetvyqrzrzqnfjulthndjukkwenqrqyeynncuoucjnntdvucqbhzkrtjxhtjfvwrkeqlmckyuytdysxryqqfszcqjkewptyclhriwvmohowxopfdbhfdcjlclsfqpbehjzdernduckmrrepidcmfhmwqnkmzbjoflloitsipmrctddgengzudyjewomjqieryvnfuedwunenpbcnnwgjjnypskuegqbrfhmhqmnmhdhlyqfjjfuhutswkbwuoyecmybwciqjgpzwypqpjhfyejqveqirlwpvbbrmwjkkweuhkmjctjyuhzulikqepekueqxlugxoowgbbzmtxopfyheockitxjowdyjeqpeqoyvnrvymbzrqxwuevwytcfdbgdlcedfrdpdecpvglxskymtctxbqwiddshtpemcmnwqkzyfwkkfzwsbggteueipbyqvdlhuyczyphobifqttlhrolm tqqyosdhfztfkcbdnerckmctuemtsytglodubcplksxwwssj --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multiple ethernet interfaces Date: Wed, 03 May 2000 03:08:54 GMT From: "Brian St. Pierre" Organization: Road Runner Message-ID: References: <8enn7f$714$1@nnrp1.deja.com> In article <8enn7f$714$1@nnrp1.deja.com>, gold@sdsu.edu wrote: > In article , > "Randy Schafer" wrote: >> Does anyone know the "correct" sequence of function calls to install >> multiple END ethernet drivers and attach them to protocols? I can set >> the boot device to either ports and it boots fine. If I then try to >> initialize the other port after booting, I only get > errors. >> System is EST 8260 eval board using T2 and vxWorks 5.4. > > First make sure you have the extra interfaces in the table in > confignet.h. Do what Dan said, but as a caveat, be aware that you can not re-use the "unit number", even if you have different devices. We tried this and it does not seem to be supported. I.e. if you have interfaces for devices you're calling "aa" and another called "bb", you need to call the interfaces "aa0" and "bb1". - -Brian --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Slow GCC compiler in Tornado II Date: Wed, 26 Apr 2000 23:33:24 GMT From: todchapman@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8e7uc4$7eh$1@nnrp1.deja.com> References: <3905BAC7.A09FAE90@htc.honeywell.com> Green Hills has free evaluation CDs and PPC is a very strong target. You can request a free evaluation CD from www.ghs.com. The new GUI in MULTI 2000 is a step above the 189 version. MULTI has nice kernel aware debugging for several RTOSes including vxWorks. Regards, Tod In article <3905BAC7.A09FAE90@htc.honeywell.com>, Lee Graba wrote: > I am using Tornado II, hosted on Windows NT and targetting a PowerPC 604 > processor. This compiler appears to be very inefficient, as it causes > continuous disk swapping (on a 96 Meg. machine), which of course makes > it very slow. Compiling relatively small files can take 5 minutes. > > Are the later versions of the GCC compiler more efficient and faster? > Would it be worth building my own cross-compiler of a later > version(perhaps 2.95) in order to speed compiles? Are there other > compilers (such as that from Green Hills) that are significantly faster? > > -- > ---------------------------------------------------------------------- - ---------- > > Lee Graba > Honeywell Technology Center > graba_lee@htc.honeywell.com > ---------------------------------------------------------------------- - ---------- > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Redirect printf Date: Wed, 26 Apr 2000 23:37:43 GMT From: pkockritz@home.com (Pete Kockritz) Organization: @Home Network Message-ID: References: <390720AF.4D9CFA7B@c-cube.com> In article <390720AF.4D9CFA7B@c-cube.com>, Stephane Bellemare wrote: > I'd like to redirect all printf to a message Q in order to avoid the use > of the serial port or the Tornado shell . A task will read the msgQ and > the data will be sent to the network using VxDCOM through Connection > points. > > Is there a simple solution in order to redirect these printf to a msgQ? Why not use logLib instead of printf? That's what it does. You can add file descriptors that will get the messages that are posted to its queue. Regards, Pete - -- +-----------------------------------------------------------------+ | Pete Kockritz mailto:pkockritz@home.com | | This space for rent. | +-----------------------------------------------------------------+ --------------------------- Newsgroups: comp.os.vxworks Subject: Varargs on PPC under 5.4 using Tornado 2.0... Date: Wed, 26 Apr 2000 20:43:55 -0400 From: "Don Dewar" Message-ID: We have some debug routines that are heavily used and use varargs. They seem to work most of the time. However, in at least case they corrupt variables from the call function, indicating a stack corruption. We include VxWorks.h as has been recommended, but it still does not fix the problem. In addition this works fine under SIMNT, but breaks under the powerpc. I have looked extensively at the vararg macros, but they look okay as far as I can tell. The PPC macros are far more complicated then the SIMNT macros for obvious reasons. Here is some sample code: inline void Trace::vMsg( UINT32 ID, UINT32 Grp, UINT32 sevTyp, const char *fmt, va_list varg ) { ... } void Trace::Msg( UINT32 ID, UINT32 Grp, UINT32 sevTyp, const char *fmt, ... ) { va_list varg; va_start (varg, fmt); Trace::vMsg(ID, Grp, sevTyp, fmt, varg); va_end (varg); } // end Trace::Msg() foo::foo( RWFunc** vf ) { Trace::Msg( 814, 1, 2, "foo::foo( %p ) constructor VF table = %p\n", this, vf ); } When foo::foo calls into Trace::Msg, Trace::Msg trashes the stack very early in the calling sequence. If you have any suggestions, please let me know. Thanks, - -- Don --------------------------- Newsgroups: comp.os.vxworks Subject: Re: host vs. target shell Date: Wed, 26 Apr 2000 20:08:53 -0500 From: "Michael S. Simpson" Organization: Lockheed Martin Missiles & Fire Controls - Dallas Message-ID: <39079325.2C797937@lmco.com> References: <7kd7nc271u.fsf@mr-ed.jpl.nasa.gov> Matt: My fellow defense worker from Boeing was correct. Here at Lockheed, we had a problem setting the timeout values for the network due to the large size of our applications (12+ MB). It can be done, but it does take some resources. Our host platforms are Dell Pentium II's, 450 MHz with 384MB RAM. Our target is a Radstone PPC603e card running VxWorks 5.3.1 (I'm looking at the cost of upgrading to 5.4 (Tornado 2)). We're running Green Hills AdaMULTI applications and it seems to be working pretty well. We can actually run our debugger over the ethernet and checkout the system. It is a resource hog though. Mike Simpson Sr Software Engineer Lockheed Martin Missiles & Fire Controls - Dallas Note: Comments are the author's only; not those of his employer Matthew R Wette wrote: > How are people doing with using the host-based tools for Tornado? I > used VxWorks quite heavily years back when the loader, shell, > etc. were target-based. I thought the WTX interface for putting > everything on the host was very attractive (I wrote a Java JNI layer > for the WTX C lib) but when I talk to some folks about using it they > comment that it doesn't always work well. For example, loading large > files via windsh often fails. I'd be interested to hear other recent > experiences. Is this capability (i.e., host-based interaction) > getting robust? > > Thanks, > Matt > -- > Matthew.R.Wette at jpl.nasa.gov -- I speak for myself, not for JPL. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CPU Usage for power pc Date: Wed, 10 May 2000 04:15:15 GMT From: gord_moore@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8fano8$t6o$1@nnrp1.deja.com> References: <8evnvl$kc2$1@bob.news.rcn.net> We are currently using a an idle task running at 255 (lowest) priority to show when the CPU is idle. In order to see the task switches we have added a task switch hook that writes the task address to a dedicated address space on the VME bus. (We have the luxury of having 4 custom cards in the system which respond to our address space, but basically ignore the data.) From this point you can use a logic analyser (we use an HP 16500) or V-Metro card to collect the VME trace. Once you have the trace it is simply a matter of determining how long the idle task is running. A couple of things to look out for: local interrupts won't show up on the VME bus without instrumentation (write to VME address works); VME interrupts don't make it obvious when they exit unless they cause a context switch; the idle task may add some task switches and therefore overhead; the context switch writes will add some overhead. We actually set this up for debugging purposes and it is likely to be overkill for determining idle time. You should be able to do something similar with less work, just remember that anything you do with code (such as adding the idle task and context switch hooks) is intrusive and will change your results. gord In article <8evnvl$kc2$1@bob.news.rcn.net>, "Sanjiv Koshal" wrote: > Can someone explain an algorithm that I can write to monitor CPU usage on > power pc's running VxWorks? > I have a hunch that I can use an idle task to monitor the CPU usage but how? > > Thanks, > Sanjiv > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 604 privilege Date: Thu, 27 Apr 2000 06:10:59 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8e8llc$e8$1@nnrp1.deja.com> References: <3905175E.564D4CFB@west.raytheon.com> <3905B34A.E6809C6E@nowhere.com> <39061823.3E2C8456@west.raytheon.com> > But I inverted the bit sense in my mind. A clear demonstration of another bit of PPC wisdom... apart from reversing the natural ordering of the bit numbers within each word, they also managed to refer to user mode as "problem" state, abbreviated to PR which also happens to be the start of privilege... sigh... I guess eieio makes up for it all ;-) John... [who still has trouble working out bit numbers in reverse] Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Speedup of link time 80% !!! Date: Thu, 27 Apr 2000 08:40:10 +0200 From: Ole Asbjorn Fadum Organization: Telenor Online Public Access Message-ID: <3907E0CA.E13DA62B@scanmar.no> I have done a discovery today which I could not explain. Background: PC 256Mb of RAM, 8Gb LVD SCSI-II disk, Tornado-II, NT4-SP5. Application contain zinc,ugl, C++ Application size 10Mb (including debug symbols) The generation of a '.out' file is as follows: 1. link the application and libraries to partialImage.o 2. use partialImage.o to extract all static classes (munch) 3. compile the above findings (ctdt.o) 4. link the first obj file partialImage.o with ctdt.o Current status: Our application ('.out) file is about 10 Mb of code, but this is mostly because of debug symbols, size386 returns a size of about 1 Mb. Observations: The regeneration of our downloadable file takes in excess of 5 minutes. Step #1-3 runs at good speed (total 0:35)!!!, but the step #4 takes a lot of time, making the complete cycle 5:30. Solution: I don't know why, but if we don't reuse the partialImage.o in step #4, but regenerate it, the whole cycle is completed in about 0:45! Possible explanation: Has ld386 problems with opening large object files due to non optimal algorithms for symbol parsing?? What I have done: tornado\target\h\make\rules.vxApp contain the rules of how to make an application. I have changed the above mentioned step $4 which is was as follows: $(LD_PARTIAL) $(LD_PARTIAL_LAST_FLAGS) partialImage.o ctdt.o -o $@ with the following: $(LD_PARTIAL) $(PRJ_OBJS_FOR_LD_PARTIAL) $(PRJ_LIBS) ctdt.o -o $@ --------------------------- Newsgroups: comp.os.vxworks Subject: DosFs/TFFS problem Date: Wed, 10 May 2000 10:11:39 +0200 From: Carsten Boeckmann Organization: ericsson Eurolab Deutschland GmbH Message-ID: <391919BB.A235F8BB@eede.ericsson.se> Hi, we are using Tornado 1.0.1/VxWorks 5.3.1 on a PPC603 board and DOS FS 1.0/TFFS 2.0 with optional long file names for managing the onboard's flash memory. We have some problems with inconsistent FAT entries when using /RFA/ as device for file operations. For example, files which have been deleted before appear as file with empty content (so just its FAT entry seems to be not fully deleted/recreated) when writing into a completely different file. We don't have these problems when using /tffs0/ as device for file operations. iosDevShow() and iosDrvShow() tell me that both devices are using the same driver resp. driver routines. Can anybody of you tell me what the difference between these two devices is ? Thanks in advance, Carsten - ----------------------------------------- -= Carsten.Boeckmann@eed.ericsson.se =- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: AMD 79C973 Ethernet chip Date: Thu, 11 May 2000 15:34:47 GMT From: Darren Etheridge Organization: Deja.com - Before you buy. Message-ID: <8fejul$6us$1@nnrp1.deja.com> References: Try: http://www.amd.com/products/npd/software/pcnet_family/drivers/vxworks.ht ml if you have not already managed to locate such a driver, I'm not sure this supports AMD79C973 but its worth looking at. Darren In article , "Charles Krinke" wrote: > A new question. I have had an AMD 79C973 ethernet chip dumped in my lap and > need to provide a driver for it. Has anyone heard of one that allready > exists or is anyone else working on one that might like to collaborate? > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Shared interrupts Date: Thu, 11 May 2000 10:08:12 -0700 From: Jason Leong Organization: BCnet Message-ID: Hi VxWorkers, Has anyone been able to share PCI interrupts between the ethernet card and another device using the Pentium BSP? I've tried using the 3com fast etherLink XL PCI card and the Intel Ether Express PRO100B PCI card, but I'm getting file system errors during the pciIntConnect(). If you've been able to get this working, how did you do it and what card were you using? Thanks in advance. Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VGA VIDEO MODES Date: Thu, 11 May 2000 22:48:14 +0000 From: James Pascoe Organization: The University of Reading Message-ID: <391B38AE.42537CBB@reading.ac.uk> References: <391AB4C1.F53482B1@reading.ac.uk> <8fe913$pq5$1@nnrp1.deja.com> Hi Guys, Yep, I was dreading that. does anybody know which port to switch (in order to chage the video mode)? I have been bashing my head against a web browser all afternoon :)) Jim Darren Etheridge wrote: > > Hi James, > > You are trying to make a BIOS call from a protected mode operating > system. (Most) BIOS calls can only be made from Real mode. You will > need to hit the VGA IO Ports directly using 'sysOutByte' etc. to switch > the video mode. > > Hope this helps > > Darren > > In article <391AB4C1.F53482B1@reading.ac.uk>, > James Pascoe wrote: > > Hi all, > > > > What I need to do is to change the video mode on a 486 pc to 320 x > 200 so > > that I can do some elementary pixel plotting with students. If I was > using > > borland C, I would do the following: > > > > void setMode13h() > > { > > union REGS regs; > > > > regs.x.ax = 0x13; > > int86(0x10, ®s, ®s); > > } > > > > This is obviously borland specific, so I wrote the following asm > statement in > > Tornado: > > > > __asm__("movl $13h, eax\n > > intl $10h" > > : > > : > > ); > > > > When I execute this, I get the error: > > > > Invalid TSS > > > > and the program crashes. > > > > My assembler is so rusty, that I am probably making a 'kick me' with > the above. > > > > What does anybody think? > > > > James > > > > -- > > > ************************************************************************ > *** > > * James Pascoe (research student) | Tel: +44 118 9875123 Ext 7626 / > 4223 * > > * Department of Computer Science | Fax: +44 118 9751994 > * > > * The University of Reading | > * > > * RG6 6AY | Email: J.S.Pascoe@reading.ac.uk > * > > * United Kingdom | James@james- > pascoe.com * > > * -------------------------------------------------------------------- > --- * > > * Official URL: > http://www.cs.reading.ac.uk/cs/people/jsp/home.html * > > * Personal URLs: http://www.james-pascoe.com/ | http://james- > pascoe.i.am/ * > > > ************************************************************************ > *** > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. - -- *************************************************************************** * James Pascoe (research student) | Tel: +44 118 9875123 Ext 7626 / 4223 * * Department of Computer Science | Fax: +44 118 9751994 * * The University of Reading | * * RG6 6AY | Email: J.S.Pascoe@reading.ac.uk * * United Kingdom | James@james-pascoe.com * * ----------------------------------------------------------------------- * * Official URL: http://www.cs.reading.ac.uk/cs/people/jsp/home.html * * Personal URLs: http://www.james-pascoe.com/ | http://james-pascoe.i.am/ * *************************************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VGA VIDEO MODES Date: Thu, 11 May 2000 10:57:28 -0700 From: dan Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <24e0ccd8.b6cbff7e@usw-ex0104-032.remarq.com> References: <391AB4C1.F53482B1@reading.ac.uk> In article <391AB4C1.F53482B1@reading.ac.uk>, James Pascoe wrote: >Hi all, > >What I need to do is to change the video mode on a 486 pc to 320 x 200 so >that I can do some elementary pixel plotting with students The processor is in Real mode during the boot phase. You can patch romInit.s to make your BIOS calls after the 'cli' in _romInit and before the 'jmp cold' as follows: addr16 /* address size: 16 bit address operand for call */ /* instruction. This is 0x66 opcode in a dump */ data16 /* operand size: 16 bit data operand for call */ /* instruction. This is 0x67 opcode in a hex dump */ call _bios80x43 /* call (real mode) _romBIOS routine */ /* or call _vesa40x25 */ /* or call _bios40x25 */ ..and then for example: .align 4,0x90 _bios80x43: /* ugly, but it works for me... */ movb $0x0f,%ah sti int $0x10 cli push %ax mov $0x1201,%ax movb $0x30,%bl sti int $0x10 cli pop %ax movb $0x00,%ah sti int $0x10 cli mov $0x1112,%ax movb $0x00,%bh sti int $0x10 cli ret .align 4,0x90 _vesa40x25: /* VESA mode */ movb $0x4f,%ah /* mov VESA function into AH reg. */ movb $0x02,%al /* mov "SET MODE" function into AL reg. */ movb $0x00,%bh /* mov "40x25 COLOR" mode into BX reg. */ movb $0x01,%bl /* mov "40x25 COLOR" mode into BX reg. */ sti /* set the int flag,eg unlock interrupts. */ int $0x10 /* Call BIOS Int 0x10 (Function in AH reg) */ cli /* clear the int flag,aka disable interrupts.*/ ret .align 4,0x90 _bios40x25: /* VGA BIOS mode */ movb $0x00,%ah /* movb "SET MODE" BIOS function into AH reg. */ movb $0x01,%al /* movb "40x25 COLOR" mode into AL reg. */ sti /* set the int flag,eg unlock interrupts. */ int $0x10 /* Call BIOS Int 0x10 (Function in AH reg) */ cli /* clear the int flag,aka disable interrupts.*/ ret * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Anybody see the "VxWorks Hacks" seb site? Date: Thu, 11 May 2000 19:54:40 GMT From: "Marc" Organization: @Home Network Message-ID: <4eES4.236982$8k3.1598538@news1.rdc1.sdca.home.com> This is a multi-part message in MIME format. - ------=_NextPart_000_012C_01BFBB48.8D735AD0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable I'm looking for a web site that had a few vxworks sourc code ditties that were handy on it.=20 One was the file "vxhttp.c" which was a mini-http page server. I could use that for some verfiy testing right now. Anybody have a copy or know=20 where the site went I'd apprieciate knowing about it. Thanks in advance for any light shed, Marc Post to group, or remove spaces around @ email below: ohmy-gosh @ home.com - ------=_NextPart_000_012C_01BFBB48.8D735AD0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
I'm looking for a web site that had a = few=20 vxworks
sourc code ditties that were handy on = it.=20
 
One was the file "vxhttp.c" which was = a=20 mini-http
page server.  I could use that = for some=20 verfiy
testing right now.  Anybody have = a copy or=20 know
where the site went I'd apprieciate = knowing=20 about
it.
 
Thanks in advance for any light=20 shed,
 
Marc
 
Post to group, or remove spaces around @ email below:
 
ohmy-gosh @ home.com
 
 
 
 - ------=_NextPart_000_012C_01BFBB48.8D735AD0-- --------------------------- Newsgroups: comp.os.vxworks Subject: compilation problem Date: 28 Apr 2000 00:37:16 GMT From: trinity@bnr.ca (Seshu Madhavapeddy) Organization: Bell-Northern Research, Richardson, TX Message-ID: <8eamfs$530$1@crchh14.us.nortel.com> I am compiling vxSim linking with my object files (on HP-UX 10.20 using compiler cchppa). I get a strange error during link time: ld -B immediate -N +vnocompatwarnings -e sysInit -o vxWorks_mohan dataSegPad.o ctdt.o vxWorks_mohan.tmp kernel//bbspspm/gen/libbspsun.lnk kernel//brtc/gen/librtcsun_serv.lnk kernel//bdialog/gen/libdialogsun.lnk kernel//btrexcb/gen/libtrexsun.lnk kernel//btrsp/gen/libtrspsun.lnk kernel//broot/gen/librootsun.lnk kernel//bvcons/gen/rtcvirtcons.o ScenarioAutoroute.lnk /opt/corp/projects/hornet_sw/vxworks_ppc/target/lib/libSIMHPPAgnuvx.a -lc ld: Address of code unsat to Millicode symbol "$$dyncall" taken in file vxWorks_mohan.tmp *** Error code 1 This error is generated only when I link with "libc". However, if I do not link with libc then I get unsatisfied symbols for functions like socket(), sendto(), recvfrom() etc. I thought ld would automatically link with libc without specifying it, but it is not happening. So I added "-lc", but that results in the above mentioned error. Any ideas/suggestion? Thanks, Mohan --------------------------- Newsgroups: comp.os.vxworks Subject: Address Pipelining Date: Thu, 11 May 2000 22:21:31 +0100 From: "l" Organization: Customer of Planet Online Message-ID: <8ff8ci$ath$2@newsg2.svr.pol.co.uk> All, Does anyone know if it possible to disable address pipelining within the Universe II chip on the MVME 2400 processor board and how this can be done through VxWorks using the MVME2400 BSP? The reason I want to disable address pipelining is because whilst accessing memory from another board across the VME, I get corruption when conducting 16 bit transfers along the VME. Further investigation revealed that the board I am accessing on the VME does not support address pipelining which is causing the corruption. Any ideas/suggestions would be appreciated Andy --------------------------- Newsgroups: comp.os.vxworks Subject: Loading Question??? Date: Thu, 11 May 2000 21:51:40 GMT From: "WDIZ" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: Dear All, How can I load 2 applications which depended on each other? Does Vxworks support that? For example: in a.o /* Function A */ #include extern void BPrintHello(); void APrintHello( ){ printf("Hello A"); } void CallB() { BPrintHello(); } in b.o /* Function B */ #include extern void APrintHello(); void BPrintHello(){ printf("Hello B"); } void CallA() { APrintHello(); } if I load a.o first ld < a.o I got the following error Undefined symbol: BPrintHello (binding 1 type 0) value = 16088312 = 0xf57cf8 When I load b.o, I have no error. Of course when I try to run CallB, I got the following error, - -> CallB program Exception current instruction address: 0x00000000 Machine Status Register: 0x0008b130 Condition Register: 0x44400082 176dac vxTaskEntry +60 : shell () 146df8 shell +18c: 146e24 () 14702c shell +3c0: execute () 1471ac execute +d8 : yyparse () 18d8dc yyparse +7a8: 18b840 () 18b9b8 yystart +8f8: CallB () ffb23c CallB +10 : 0 () shell restarted. How can I load a.o and b.o so I can run both CallB and CallA functions? Thank for your answer in advance. - -- Newcomer to Vxworks --------------------------- Newsgroups: comp.os.vxworks Subject: MPC860 SPI Driver Date: 28 Apr 2000 01:56:48 GMT From: newfriend5@aol.com (Newfriend5) Organization: AOL http://www.aol.com Message-ID: <20000427215648.01767.00000085@ng-fl1.aol.com> Take a look at the SPI Loopback Driver and the SPI TouchPanel Driver at the following URL: http://www.mot.com/SPS/ADC/pps/subpgs/etoolbox/821/index.html This page is a good resource for MPC8xx drivers not found on the Netcomm web page (http://www.mot.com/SPS/RISC/netcomm/tools/#ppc_group), because the MPC821 and '823 are "close relatives" of the '850 and '860. The primary difference is the inclusion of an LCD interface in the 82x family, which was targeted for portable networking systems (like hand-held devices). --------------------------- Newsgroups: comp.os.vxworks Subject: Re: lnPci driver ring buffer size Date: 28 Apr 2000 03:01:34 GMT From: fnordz@aol.com (Fnordz) Organization: AOL http://www.aol.com Message-ID: <20000427230134.20539.00000148@ng-fb1.aol.com> References: Hi Dave, I've been spending a _lot_ of time on this driver lately. It seems that a lot of the code may have been based on some work done for a driver for some earlier AMD lance version (non-PCNet-PCI, that is). I don't see any reason why one could not use 512 as the TX and RX ring buffer size. On the other hand, I have not tried this yet. The current configuration may be a choice based on memory conservation, but it may just be a left-over from an earlier implementation. What happened when you tried to configure the driver to use more than the default number tx/rx descriptors? Did you set up a larger driver cluster pool to work with the additional tmds and rmds? Regards, paul iannacito --------------------------- Newsgroups: comp.os.vxworks Subject: Intel82559 ethernet driver for PowerPC Date: Thu, 11 May 2000 15:34:52 -0700 From: Polly Low Organization: Raytheon Company Message-ID: <391B358C.D3072B86@west.raytheon.com> I'm looking for an existing ethernet driver for the Intel82559 ethernet controller that runs on the PowerPC processor with vxWorks. Instead of starting from scratch, I like to find an existing one or one I can modify from. The one driver posted in the Intel site only runs on the Pentium processor and not PowerPC. Can anyone give me some ideas on where I can possibly find one? Polly Low --------------------------- Newsgroups: comp.os.vxworks Subject: Problem in booting using bootp Date: Thu, 27 Apr 2000 20:35:42 -0700 From: Kapil Arora Organization: EmpowerTel Networks Inc. Message-ID: <3909070E.450FCC6@empowertel.com> Hi all, We have vxWorks 5.4 on a mezanine card having Power PC procesor. It boots up then reads parameters from the NVRAM and downloads the image over the network using FTP fine. Now we are trying to switch to booting using bootp. We have included bootp in the configuration while building the bootrom image. We are using a Solaris2.7 bootp server and we have configured following : - - put ethernet address of the card in /etc/ethers - - put the IP Address of the card in /etc/hosts - - added an entry for the card in /etc/bootparams - - added the boot image in the /tftpboot The boot server is working fine as other bootp client boot correctly from it. This is the first vxWorks based client we are trying to boot. Documentation says to use /etc/bootptab but Solaris 2.7 only supports /etc/bootparams, so we've modified the sample entries from the documentation to correspond to this. The entry in the bootparams file looks like : card09 hd=:/tftpboot bf=:vxWorks.st vxWorks sends a bootp request but it doesn't get the bootline back. Anyone having any ideas, what's going wrong? Thanks, Kapil. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: compilation problem Date: Fri, 28 Apr 2000 03:29:31 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8eb0ib$jtr$1@nnrp1.deja.com> References: <8eamfs$530$1@crchh14.us.nortel.com> Hi Mohan, You don't say which version of the simulator you have, or even of which version of Tornado so I can't be certain, but I suspect that you are trying to use networking functions with the bundled simulator in T2.0. Networking is not supported in the bundled simulator which would explain why you are having problems with socket(), recvfrom() etc. If you wish to use networking in the simulator you'll need to upgrade to the full simulator product - check with your FAE/sales rep. Hope that helps, John... In article <8eamfs$530$1@crchh14.us.nortel.com>, trinity@bnr.ca (Seshu Madhavapeddy) wrote: > I am compiling vxSim linking with my object files > (on HP-UX 10.20 using compiler cchppa). I get a strange > error during link time: > > ld -B immediate -N +vnocompatwarnings -e sysInit -o vxWorks_mohan dataSegPad.o ctdt.o vxWorks_mohan.tmp kernel//bbspspm/gen/libbspsun.lnk kernel//brtc/gen/librtcsun_serv.lnk kernel//bdialog/gen/libdialogsun.lnk kernel//btrexcb/gen/libtrexsun.lnk kernel//btrsp/gen/libtrspsun.lnk kernel//broot/gen/librootsun.lnk kernel//bvcons/gen/rtcvirtcons.o ScenarioAutoroute.lnk /opt/corp/projects/hornet_sw/vxworks_ppc/target/lib/libSIMHPPAgnuvx.a - -lc > ld: Address of code unsat to Millicode symbol "$$dyncall" taken in file vxWorks_mohan.tmp > *** Error code 1 > > This error is generated only when I link with "libc". However, if I > do not link with libc then I get unsatisfied symbols for functions like > socket(), sendto(), recvfrom() etc. I thought ld would automatically > link with libc without specifying it, but it is not happening. So I > added "-lc", but that results in the above mentioned error. > > Any ideas/suggestion? > > Thanks, > Mohan > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Retrying FTP image download Date: Thu, 27 Apr 2000 20:40:22 -0700 From: Kapil Arora Organization: EmpowerTel Networks Inc. Message-ID: <39090826.BCBEECB9@empowertel.com> Hi all, We have vxWorks 5.4 on a mezanine card having Power PC procesor. It boots up then reads parameters from the NVRAM and downloads the image over the network using FTP fine. Sometimes it fails to download the file and returns a error in booting, returning the famous 0xd003 error number. We want it to retry the download for a few more times. Is it possible or is it possible to increase the time-out for any try? Thanks, Kapil. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Leading zero in addHost() IP address Date: Thu, 11 May 2000 23:46:18 GMT From: Charlie Grames Organization: Boeing Message-ID: <391B464A.98C01AC2@nowhere.com> References: <200005111835.IAA01041@atlas.cfht.hawaii.edu> Sender: nntp@news.boeing.com (Boeing NNTP News Access) I think we need to separate the functionality of interpreting IP addresses from that of scanf(). The documentation for inetLib in the VxWorks reference manual clearly states that prefixing a field of an IP address with "0" causes the field to be interpreted as octal, and prefixing with "0x" or "0X" causes the field to be interpreted as hexadecimal. Interestingly enough, the man page for inet_addr() on Solaris systems says the same thing. With scanf(), however, the conversion specifiers determine the radix used. "%d" specifies decimal and ignores leading zeros. "%o" specifies octal and allows, but does not require, leading zeros. "%x" specifies hexadecimal and allows, but does not require, leading "0x" or "0X". Only the "%i" conversion specifier is affected by the prefix. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ifAddrSet error description??? Date: Fri, 12 May 2000 00:01:39 GMT From: Charlie Grames Organization: Boeing Message-ID: <391B49E3.6A3D8D24@nowhere.com> References: <8ff9fe$5q$1@nnrp1.deja.com> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Try looking at the value of errno or use the printErrno() function. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com ndsubison@my-deja.com wrote: > > I am calling ifAddrSet to assign various > interfaceNames to internet addresses. The code > looks as follows: > > sprintf(tempStr, "xxx.yyy.zzz.%d", currentID1); > rc = ifAddrSet ("ipc0", tempStr); > if (rc != OK) > logError; > > sprintf(tempStr, "xxx.yyy.zzz.%d", currentID2); > rc = ifAddrSet ("ipc1", tempStr); > if (rc != OK) > logError; > > The 1st call to ifAddrSet works without error, but > the second returns -1. How do I go about finding > what the ifAddrSet function disliked about the > second call and why the address set failed? > > Any thoughts? > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Address Pipelining Date: Thu, 11 May 2000 23:59:52 GMT From: Charlie Grames Organization: Boeing Message-ID: <391B4978.CEFC9E4D@nowhere.com> References: <8ff8ci$ath$2@newsg2.svr.pol.co.uk> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Andy, Can you define what you mean by "address pipelining?" I am not familiar with the term in either a VMEbus or PowerPC context. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loading Question??? Date: Fri, 12 May 2000 00:04:56 GMT From: Charlie Grames Organization: Boeing Message-ID: <391B4AA8.116D9FD5@nowhere.com> References: Sender: nntp@news.boeing.com (Boeing NNTP News Access) WDIZ, With the current version of VxWorks, you can't do what you are trying. Both functions must reside in the same module. VxWorks 6.0 promises to fix this. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com WDIZ wrote: > > Dear All, > > How can I load 2 applications which depended on each other? > Does Vxworks support that? > > For example: > > in a.o > /* Function A */ > #include > extern void BPrintHello(); > void APrintHello( ){ printf("Hello A"); } > void CallB() { BPrintHello(); } > > in b.o > /* Function B */ > #include > extern void APrintHello(); > void BPrintHello(){ printf("Hello B"); } > void CallA() { APrintHello(); } > > if I load a.o first > > ld < a.o > > I got the following error > > Undefined symbol: BPrintHello (binding 1 type 0) > value = 16088312 = 0xf57cf8 > > When I load b.o, I have no error. > Of course when I try to run CallB, I got the following error, > > -> CallB > > program > Exception current instruction address: 0x00000000 > Machine Status Register: 0x0008b130 > Condition Register: 0x44400082 > > 176dac vxTaskEntry +60 : shell () > 146df8 shell +18c: 146e24 () > 14702c shell +3c0: execute () > 1471ac execute +d8 : yyparse () > 18d8dc yyparse +7a8: 18b840 () > 18b9b8 yystart +8f8: CallB () > ffb23c CallB +10 : 0 () > shell restarted. > > How can I load a.o and b.o so I can run both CallB and CallA functions? > > Thank for your answer in advance. > -- Newcomer to Vxworks --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loading Question??? Date: Fri, 12 May 2000 08:39:21 +0800 From: "tanghaoyu" Organization: Datang Telecom Message-ID: <8ffjtv$bnd$1@news.cz.js.cn> References: You can load the files as this step: >ld "a.o" >ld "b.o" >ld "a.o" or >ld "b.o" >ld "a.o" >ld "b.o" It will be ok. Good luck. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks first Question... Date: Thu, 11 May 2000 21:06:17 -0400 From: Joe Durusau Message-ID: <391B5909.9B36AB2F@bellsouth.net> References: <8fembf$sjk3@overload.lbl.gov> Yep. Tom Miller wrote: > > ....are any of your recieving this? > > Tom Miller > Motion Engineering Inc. > Santa Barbara > Ca. --------------------------- Newsgroups: comp.os.vxworks Subject: Emulating vxWorks over Linux Date: Fri, 28 Apr 2000 06:08:02 GMT From: Marcelo & Keyly Mourier Organization: AJAX Productions Message-ID: <39092AD1.27876BDC@earthlink.com> Reply-To: jukema@earthlink.net - --------------02C413E4ED3FB9C82AD624CC Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Has anybody written a library that can emulate the basic vxWorks functionality (e.g. taskLib, semLib, msgQLib, wdLib, etc.) on top of the native Linux threads and IPC facilities? The basic idea is to be able to compile, link, and run a simple vxWorks application (e.g. no interrupts) under Linux on a PC... Thanks! - --------------02C413E4ED3FB9C82AD624CC Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit Has anybody written a library that can emulate the basic vxWorks functionality (e.g. taskLib, semLib, msgQLib, wdLib, etc.) on top of the native Linux threads and IPC facilities? The basic idea is to be able to compile, link, and run a simple vxWorks application (e.g. no interrupts) under Linux on a PC...

Thanks! - --------------02C413E4ED3FB9C82AD624CC-- --------------------------- Newsgroups: comp.os.vxworks Subject: Uninitialized Interrupt Date: Thu, 11 May 2000 20:10:17 -0700 From: Kapil Arora Organization: EmpowerTel Networks Inc. Message-ID: <391B7619.17BB305B@empowertel.com> References: <3918832D.A8E70053@empowertel.com> Reply-To: karora@iname.com Hi, We are using PPMC750 BSP on one of our boards. While porting this BSP to out customised card we were seeing a message "interrupt: Uninitialized interrupt" on the console while booting. Then it disappeared when we changed a few things here and there in the ported ethernet driver. Now, this problem has appeared again during boot procedure. The boot code downloads the image and starts executing it when this message appears and it hangs after that. Is it possible to know which interrupt is uninitialised? Can I go through the Interrupt Table and attach my default handler to each uninitialised interrupt so that I know which interrupt is this? Thanks, Kapil. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks sizeof(int) question. Date: Fri, 12 May 2000 04:14:07 GMT From: snaphook@southwesternbell.net (Bruce) Organization: SBC Internet Services Message-ID: <391b8491.24412803@news.swbell.net> References: <8fembh$sjl5@overload.lbl.gov> In comp.os.vxworks Fred Roeber wrote: >In my experience, the size of an int is solely a compiler issue but is >normally tied to the processor architecture. I have seen cases where ints >are 16, 32, or 64 bits. With TI C40 DSPs, you even have the odd case >where ints are 32 bits but sizeof(int) = 1. All I'm trying to get at is >that if you are writing portable software, you shouldn't count on an int >being a certain size and certainly shouldn't count on being able to store >a pointer value in an int. The TI DSP is interesting. ANSI C guarantees that a sizeof(char) == 1 and that sizeof(char) <= sizeof(short) <= sizeof(int) <= sizeof(long). This leads me to conclude that you must have had 32bit chars also? Either that or a non-conforming compiler. --------------------------- Newsgroups: comp.os.vxworks Subject: C++ Exceptions causing hangs Date: Fri, 28 Apr 2000 12:31:32 GMT From: chuckallis@yahoo.com Organization: Deja.com - Before you buy. Message-ID: <8ec0b4$lj9$1@nnrp1.deja.com> Hello, Some background info: VxWorks 5.4 Tornado 2 PowerPC 604 I wanted to set up a simple C++ test program to throw and catch an exception. The code looks similiar to this: #include int testException() { try { throw 1; } catch (int& someException) { cout << "Caught some exception" << endl; } return 1; } I run the program through the windsh and the program runs fine until it trys to exit the function. I lose connection to the target and when I look at the processor board, the run light is solid. I looked through this newsgroup and it looks like a couple of people had the same type of problem. I'm fairly new to VxWorks so I'm not entirely certain what to think. The program runs perfectly fine when I compile it for Solaris. Some of the VxWorks "veterans" on my project seem to think it is a problem with windsh. Any information concerning what causes this problem and how to fix it would be greatly appreciated. Thanks in advance! Chuck Allis Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Retrying FTP image download Date: Fri, 28 Apr 2000 13:50 +0100 (BST) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: References: <39090826.BCBEECB9@empowertel.com> Reply-To: rew@cp.cix.co.uk You will have to modify the boot code. In bootLoad() (bootConfig.c), try something like the following :- /* load specified file */ #if 1 /* (modified version) */ while (netLoad (params.had, params.bootFile, params.usr, params.passwd, pEntry) != OK) { printf ("\nError loading file: errno = 0x%x.\n", errno); printf ("Retrying...\n"); } #else /* (original version) */ if (netLoad (params.had, params.bootFile, params.usr, params.passwd, pEntry) != OK) { printf ("\nError loading file: errno = 0x%x.\n", errno); return (ERROR); } #endif Richard In article <39090826.BCBEECB9@empowertel.com>, kapil@empowertel.com (Kapil Arora) wrote: > Hi all, > We have vxWorks 5.4 on a mezanine card having Power PC procesor. > It boots up then reads parameters from the NVRAM and downloads the image > > over the network using FTP fine. > Sometimes it fails to download the file and returns a error in booting, > returning the famous 0xd003 error number. > We want it to retry the download for a few more times. Is it possible or > is it possible to increase the time-out for any try? > Thanks, > Kapil. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Changing MAC Address Date: Fri, 28 Apr 2000 10:14:05 -0400 From: Douglas Fraser Organization: Lucent Technologies Message-ID: <39099CAD.8DAB5794@lucent.com> References: <8e2lvs$kin$1@nnrp1.deja.com> <8ea77m$o86$1@nnrp1.deja.com> I have only seen the 'N' option to the boot parameters listed on BSPs that I have modified, not the standard BSPs. If you type '?' at the prompt, it will list 'N' as an option, with a description, if it has been enabled in that BSP. The code that reads and writes the MAC address are near the bottom of $(WIND_BASE)/target/all/bootConfig in a function named mEnet. good luck Doug ksamavedam@my-deja.com wrote: > > At the boot prompt(boot :, U get here by stopping the boot process), > there may be some command, generally one letter command(like p,c...) > Through that one may be able to change the MAC address. Check the BSP > manual pages. Tyy the command 'N' there. U can change it it your own > risk :) > > hth > --krishna > > In article <8e2lvs$kin$1@nnrp1.deja.com>, > arnel@vina-tech.com wrote: > > Anyone know how to change the Ethernet MAC Address from the command > > line? Is it possible? Does anyone know where the MAC address is set? > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TCP Errors - Date: Fri, 28 Apr 2000 21:05:20 GMT From: Hwa Jin Bae Organization: Deja.com - Before you buy. Message-ID: <8ecue5$p23$1@nnrp1.deja.com> References: <8ec9uu$ond1@overload.lbl.gov> TCP data received will be error free. The data goes through link layer CRC as well as IP checksum on the data. They are considered good enough to build all the applications you are using over Internet. The quality of service (QoS) is usually referring to the bandwidth and prioritization of the packet delivery (e.g. certain bits/sec requirement, certain latency requirement, etc.) TCP doesn't do that well as is (the urgent pointer is pretty much the only mechanism). But that doesn't imply data integrity is compromised. Hwa Jin Bae Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Where can I download vxcurses? Date: Fri, 28 Apr 2000 14:26:03 -0700 From: Richard Hang Organization: NASA Dryden Message-ID: <390A01EB.7CB4FE7F@dfrc.nasa.gov> Hi all, I am looking for "vxcurses" ftp site. Please help! Thanks. Richard Hang NASA DFRC --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2 on Windows 2000? Date: 12 May 2000 09:49:19 GMT From: hat@se-46.wpa.wtb.tue.nl (Albert Hofkamp) Organization: Eindhoven University of Technology, The Netherlands Message-ID: References: <8ephj5$753$1@nnrp1.deja.com> <391975F3.E2FF89D@triton.npt.nuwc.navy.mil> Reply-To: a.t.hofkamp@tue.nl On Wed, 10 May 2000 10:45:07 -0400, Paul Livingston wrote: >I'm having a problem with T2 on Win2k also. My problem is: >TRY=localhost/host/resource/tcl/Windview.win32.tcl not found. Can anyone >help with this? The file does exist in this directory, we have WIND_BASE >and WIND_HOST_TYPE set, and the path includes .../x86-win32/bin. I don't exactly know the context, but we had some (imho major) problems with long filenames in combination with an executable. We generate vxworks programs using a compiler, so I adapted a generated Makefile so that make translates our source to .cpp, and then the standard generated Makefile performs the remainder of the compilation. The rule I used was something like %.cpp: %.chi x:/a/path/including/alongdirectory.name/comp.exe Apparently, make can verify that the comp.exe file exists there (giving a wrong .exe file is detected), but the program is _NOT_ executed, and make does _NOT_ complain (probably, because the DOS command-line box does not report an error). (we use T1 on W95, using a command-prompt window to start make). The tricky part is that changing the path to have only components <= 8 chars makes it work (I seem to remember), as well as typing the command directly from the prompt also works. We patched it by copying the commands being executed by make into a .bat file, and using that, but the above should have worked without such patches obviously. Maybe you can relate this experience to the problem you are experiencing (the program you try to execute is > 8 chars, and has more than 1 . in it). >> On a slightly different note, is there a projected ship date for >> a version of Tornado supporting Linux? Yes, I'd like that extremely much !!!! Albert - --- Look ma, windows without Windows !! --------------------------- Newsgroups: comp.os.vxworks Subject: Q: Good CLI parser for VxWorks Date: Sat, 29 Apr 2000 01:13:22 GMT From: dchou4u@hotmail.com Organization: Deja.com - Before you buy. Message-ID: <8edcva$8j9$1@nnrp1.deja.com> I need a command line parser on my VxWorks target, and am wondering if anyone knows of a good one that has already been ported before. If so, can you please point me to the source code? I will run the CLI as a seperate task. The CLI that I need does not need to be fancy. Something like the VxShell CLI is good enough. Thanks Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks sizeof(int) question. Date: Fri, 12 May 2000 11:50:50 +0100 From: "Stephen Walsh" Message-ID: <391bde10$1@pull.gecm.com> References: <8fembh$sjl5@overload.lbl.gov> <391b8491.24412803@news.swbell.net> A number of bits in a char is processor dependant. If you have 8 bit addressable memory it will be 8 and if you can only address memory locations in 32bit chunks it is 32. The ADI SHARC DSP also has sizeof(char)=sizeof(short)=sizeof(int)=sizeof(long)=1 (all 32bit) Steve Bruce wrote in message <391b8491.24412803@news.swbell.net>... >In comp.os.vxworks >Fred Roeber wrote: > >>In my experience, the size of an int is solely a compiler issue but is >>normally tied to the processor architecture. I have seen cases where ints >>are 16, 32, or 64 bits. With TI C40 DSPs, you even have the odd case >>where ints are 32 bits but sizeof(int) = 1. All I'm trying to get at is >>that if you are writing portable software, you shouldn't count on an int >>being a certain size and certainly shouldn't count on being able to store >>a pointer value in an int. > >The TI DSP is interesting. ANSI C guarantees that a sizeof(char) == 1 and >that sizeof(char) <= sizeof(short) <= sizeof(int) <= sizeof(long). This >leads me to conclude that you must have had 32bit chars also? Either that >or a non-conforming compiler. > > --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri May 12 04:57:49 2000 From: Joachim Pluschke Date: Fri May 12 04:57:51 PDT 2000 Subject: SNTP server and vxWorks Hi all, I'm trying to make the SNTP Client/Server services run under vxWorks. The client in no problem - it works. The Server itself should'nt make much more work...... Has anyone made the SNTPserver run and deliver a correct timestamp??? (The manual entrys for sntpsLib doen't make a clear statement about how to format the timestamp needed by the server routines) Feeding the SNTP server with a valid date-format seems to be a little bit difficult ... Putting a timestamp (correct by ctime()!) like the into the servers buffer should be the correct time-format but there's send out trash in the packets (overflowing, negative value for the seconds part, thus making functions like ctime() print a date in the future far above the year 2036.... ). I tried filling the buffer (in host-byteoder, see manual) by two ways: 1) copying the seconds-part to the original location of the servers buffer, then incrementing a copy of the buffer for the size of a long and copying the seconds fraction part into 2) using bcopy() with the size of the to copy to the buffers original location..... There was still the same effect! But the joke is: The server recognizes the clients request and increases the lower part of the seconds part of the timestamp with exactly the amount of seconds what was the difference between the clients requests - something must be working.... I used a packet sniffer and swapped the bytes/words for a try with no effect on solving the problem! What's wrong? I don't really know! I'm looking forward hearing from someone who made it run !!!! bye, have a fine day -- ===================================================== Gebrüder Stoye GmbH department: traffic control systems Name: Joachim Pluschke Adr: Longericher Strasse 177 city: Cologne, Germany Zip: 50739 phone: ++49/221/2616-534 mailto: j.pluschke@stoye.de www: www.stoye.de ===================================================== From vxwexplo-errs@csg.lbl.gov Fri May 12 06:23:21 2000 From: jvlao@lucent.com Date: Fri May 12 06:23:23 PDT 2000 Subject: Re: SNTP server and vxWorks please remove me from the list From vxwexplo-errs@csg.lbl.gov Fri May 12 08:58:25 2000 From: Jim Way Date: Fri May 12 08:58:27 PDT 2000 Subject: RE: Slow GCC compiler in Tornado II I just checked my Task Manager. With a typical set of apps open, I'm using 220 Meg of RAM. I would suggest that buying additional RAM is probably a quicker and cheaper solution than trying to wedge another compiler into Tornado. IMHO. Jim > I am using Tornado II, hosted on Windows NT and targetting a PowerPC 604 > processor. This compiler appears to be very inefficient, as it causes > continuous disk swapping (on a 96 Meg. machine), which of course makes > it very slow. Compiling relatively small files can take 5 minutes. From vxwexplo-errs@csg.lbl.gov Sat May 13 04:03:20 2000 From: Vxworks Exploder Date: Sat May 13 04:03:23 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat May 13 04:03:16 PDT 2000 Subject: VxDCOM server lifetime Subject: UGL library on X11 Subject: SYM53C895 problem in pci Subject: Re: 3Com 3C905B ISA Ethernet III card Subject: VGA VIDEO MODES Subject: bcopy_to_mbufs, bcopy_from_mbufs Subject: Re: VGA VIDEO MODES Subject: Re: VSS integration into Tornado 2 Subject: Re: SNTP server and vxWorks Subject: Uninitialized Interrupt Subject: Uninitialized Interrupt Subject: Re: Loading Question??? Subject: Re: Leading zero in addHost() IP address Subject: UDP Multicast Loopback and Multihoming Subject: NT VxSim memory pool size Subject: Re: NT VxSim memory pool size Subject: Re: Address Pipelining Subject: HOT JOBS VxWorks Subject: Re: NT VxSim memory pool size Subject: PPP proxy arp entry gets deleted Subject: Re: Slow GCC compiler in Tornado II Subject: Re: Leading zero in addHost() IP address Subject: From VxWorks to Linux for instrument controllers??? Subject: Re: SYM53C895 problem in pci Subject: Re: VGA VIDEO MODES Subject: Re: From VxWorks to Linux for instrument controllers??? Subject: Re: NT VxSim memory pool size Subject: Qnx NTO Vs Vxworks Subject: Re: From VxWorks to Linux for instrument controllers??? Subject: T2 objdumpppc on NT Subject: Re: WindView Subject: Re: Loading Question??? Subject: Fat32 File System Subject: motFecEnd & LXT970A Subject: unsupported concurrent Telnet Server Subject: Re: RARP working with VxWorks? Subject: Re: unsupported concurrent Telnet Server Subject: Re: SYM53C895 problem in pci Subject: Problems with autosensing network connection ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: VxDCOM server lifetime Date: Fri, 12 May 2000 11:34:33 GMT From: rblaakmeer@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8fgq89$knn$1@nnrp1.deja.com> Hi, I have implemented a VxDCOM component that runs on my VxWorks target. The client is a windows NT 4.0 application. Everything runs well: I can create an instance of the component and call it's methods. However, after approximately 6 minutes (the COM garbage collection timeout) all methods calls on the component fail with HRESULT=0x80010113: The requested object or interface does not exist. This happens even while the client is continuously calling methods of the component. Does anyone have a solution for this? Robert Blaakmeer Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: UGL library on X11 Date: Fri, 12 May 2000 13:40:12 +0200 From: Gilbert Cabillic Organization: Irisa - Campus de Beaulieu - RENNES Message-ID: <391BED9C.E11EBDA7@irisa.fr> Is a port of ugl on X11 on top of a unix system available ? Thanks, Gilbert. // Cabillic Gilbert // Scratchy Project leader -- // http://www.irisa.fr/solidor/work/scratchy.html // IRISA - France --------------------------- Newsgroups: comp.os.vxworks Subject: SYM53C895 problem in pci Date: Fri, 12 May 2000 12:22:06 GMT From: cismroom@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8fgt11$npt$1@nnrp1.deja.com> Hi: All! I use SYM53C895 and make some experiments.When reading registers,I can find that the memory address of its 4k RAM is beyand the highest address of physic memory.For example,I have memory 64M,its memory address starts from 67M.So I can not figure out how to read from or write to its RAM. I am not so familiar with PCI. Thanks for your any suggestion! cimsroom Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 3Com 3C905B ISA Ethernet III card Date: Fri, 12 May 2000 13:00:58 GMT From: parameshbabu@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8fgva0$qc1$1@nnrp1.deja.com> References: Hi, Here it is .. right to the point.. Changes in Config.h #define INCLUDE_NETWORK #define INCLUDE_EL_3C90X_END #undef INCLUDE_FEI #undef INCLUDE_BSD #define INCLUDE_END #define IO_ADRS_ELT 0x300 // the address is different #define CONFIG_ELT 3 // RJ45 -Jumper settings Well with these changes to the config.h, i'm able to work with my 3COM - 3C905B-TX card. Hope this helps. regards, paramesh In article , Jason Leong wrote: > Hi VxWorkers, > > I'm trying to determine if the 3Com 3C905B ISA Ethernet III card is a > suitable target ethernet card in a pentium system. Has anyone been able > to get this card working with the Pentium BSP? If so, what were your > jumper settings and modifications in config.h? Thanks in advance. > > Jason > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: VGA VIDEO MODES Date: Thu, 11 May 2000 13:25:21 +0000 From: James Pascoe Organization: The University of Reading Message-ID: <391AB4C1.F53482B1@reading.ac.uk> Hi all, What I need to do is to change the video mode on a 486 pc to 320 x 200 so that I can do some elementary pixel plotting with students. If I was using borland C, I would do the following: void setMode13h() { union REGS regs; regs.x.ax = 0x13; int86(0x10, ®s, ®s); } This is obviously borland specific, so I wrote the following asm statement in Tornado: __asm__("movl $13h, eax\n intl $10h" : : ); When I execute this, I get the error: Invalid TSS and the program crashes. My assembler is so rusty, that I am probably making a 'kick me' with the above. What does anybody think? James - -- *************************************************************************** * James Pascoe (research student) | Tel: +44 118 9875123 Ext 7626 / 4223 * * Department of Computer Science | Fax: +44 118 9751994 * * The University of Reading | * * RG6 6AY | Email: J.S.Pascoe@reading.ac.uk * * United Kingdom | James@james-pascoe.com * * ----------------------------------------------------------------------- * * Official URL: http://www.cs.reading.ac.uk/cs/people/jsp/home.html * * Personal URLs: http://www.james-pascoe.com/ | http://james-pascoe.i.am/ * *************************************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: bcopy_to_mbufs, bcopy_from_mbufs Date: Fri, 12 May 2000 13:32:01 GMT From: "Gordon Manning" Organization: DataDirect Networks Message-ID: Can one call the vxworks functions bcopy_to_mbufs & bcopy_from_mbufs from an Interrupt Service Routine? Do these functions block at all? What exactly do these functions do? Thanks in advance Gordon Manning --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VGA VIDEO MODES Date: Thu, 11 May 2000 15:25:14 +0200 From: Gilbert Cabillic Organization: Irisa - Campus de Beaulieu - RENNES Message-ID: <391AB4BA.4BA4E67C@irisa.fr> References: <391AB4C1.F53482B1@reading.ac.uk> <8fe913$pq5$1@nnrp1.deja.com> Darren Etheridge wrote: > > Hi James, > > You are trying to make a BIOS call from a protected mode operating > system. (Most) BIOS calls can only be made from Real mode. You will > need to hit the VGA IO Ports directly using 'sysOutByte' etc. to switch > the video mode. > > Hope this helps > > Darren Hi Darren and James. That's true, you can't use any bios call due to protected mode. To do a 320X200 graphic, you must use a complete driver for a vga card. You can use the Ugl Library look fully supported by windriver (see http://www.wrs.com/csdocs/product/ugl/ for other documentations). If you can't use this library, you can port very easily the vgalib (1.2) for linux (not svgalib which is a little bit complex but offers you a svga resolution screen). Using this lib you will be able to set a pixel, draw lines, create bitmaps and so on. Hope it helps ! Gilbert. // Cabillic Gilbert // Scratchy Project leader -- http://www.irisa.fr/solidor/work/scratchy.html // IRISA - France --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VSS integration into Tornado 2 Date: Fri, 12 May 2000 07:10:04 -0700 From: Heiko Elger Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <0d812000.f5c6a0c2@usw-ex0101-008.remarq.com> References: <38D7835F.942122B8@avl.com> <9CfH4.2$Ga4.232@nreader3.kpnqwest.net> Hi, I'm a new tornado II user and I want to integrate visual studio version 6. Can anyone post me some checkin/checkout batch files. This would be great - thanks a lot Heiko Elger heiko_elger@arburg.com) PS: to resolve the problem with the workspace and project files of tornado I tested the following line in teh srcsafe.ini file - and it seems to work. EOL = n * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SNTP server and vxWorks Date: Fri, 12 May 2000 15:28:20 +0100 From: Si Jerram Organization: -------------------------------------------- Message-ID: <391C1504.F66618AB@ntlworld.com> References: <8fh3mc$sjk19@overload.lbl.gov> jvlao@lucent.com wrote: > please remove me from the list Er you're not on a list, this is a newsgroup. - -- Simon Jerram Email:sjerram@ntlworld.com Personal home page http://www.telos.clara.co.uk/ GaSCit 2K 5-9th August 2000, Kibblestone campsite Staffordshire: http://www.gascit.jpsonline.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Uninitialized Interrupt Date: Fri, 12 May 2000 09:32:41 -0700 From: Jason Leong Organization: BCnet Message-ID: References: <3918832D.A8E70053@empowertel.com> <391B7619.17BB305B@empowertel.com> Hi, I'm using the Pentium BSP so things might be a little different, but when I receive the error, it looks like "Uninitialized interrupt 41" or some other number. In this case, 41 represents the interrupt vector. Also, within Tornado 2.0 for Intel x86, under Tools|Browser, there's an option in the Browser to look at the interrupt vector table -- I'm not sure if that exists in the PPMC750 BSP. If your customized card is PCI and the BIOS auto-configs, then the most straight-forward way of determining the interrupt line is to do a pciConfigInByte() call to read the PCI configuration space like the following: pciConfigInByte (busNum, devNum, funcNum, PCI_CFG_DEV_INT_LINE, &intLine); After this call, intLine will be a number between 0-15 which corresponds to the interrupt line. Adding INT_VEC_IRQ0 to the interrupt line gives you the interrupt vector. I hope some of this information is relevant. cheers, Jason - -----Original Message----- From: Kapil Arora [mailto:kapil@empowertel.com] Posted At: Thursday, May 11, 2000 8:10 PM Posted To: vxworks Conversation: Uninitialized Interrupt Subject: Uninitialized Interrupt Hi, We are using PPMC750 BSP on one of our boards. While porting this BSP to out customised card we were seeing a message "interrupt: Uninitialized interrupt" on the console while booting. Then it disappeared when we changed a few things here and there in the ported ethernet driver. Now, this problem has appeared again during boot procedure. The boot code downloads the image and starts executing it when this message appears and it hangs after that. Is it possible to know which interrupt is uninitialised? Can I go through the Interrupt Table and attach my default handler to each uninitialised interrupt so that I know which interrupt is this? Thanks, Kapil. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loading Question??? Date: Fri, 12 May 2000 16:10:50 GMT From: Charlie Grames Organization: Boeing Message-ID: <391C2D0A.A8C5587D@nowhere.com> References: <8ffjtv$bnd$1@news.cz.js.cn> Sender: nntp@news.boeing.com (Boeing NNTP News Access) Unless you are lucky, loading A-B-A will not work. Let's assume A contains a function named foo(), and B contains a function named bar(). foo() and bar() use mutual recursion. Let's say that when you load A the first time, foo() ends up at location 0x200000, and bar() is unresolved (which means bar() is resolved to location 0). When you load B, bar() ends up at location 0x300000 and resolves foo() to location 0x200000. When you load A again, a new copy of foo() ends up at location 0x400000 and resolves bar() to location 0x300000. When you call foo() (0x400000), which calls bar() (0x300000), which calls foo() again (0x200000), which calls bar() again (0), you end up with an exception. All of the previous scenario assumes you are using the target-based shell, where the same module can be loaded multiple times. The problem is even worse with WindSh, which unloads the current version of a module automatically when you load a new version. I'd stick with incremental linking. - -- Charlie Grames The Boeing Company Charles.R.Grames @ boeing.com tanghaoyu wrote: > > You can load the files as this step: > >ld "a.o" > >ld "b.o" > >ld "a.o" > or > >ld "b.o" > >ld "a.o" > >ld "b.o" > > It will be ok. > Good luck. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Leading zero in addHost() IP address Date: Fri, 12 May 2000 18:10:09 +0100 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: References: <200005111835.IAA01041@atlas.cfht.hawaii.edu> Jim Thomas wrote in message <200005111835.IAA01041@atlas.cfht.hawaii.edu>... >Jim> Umm, sscanf %d does not parse 083 as octal. This is a bug, not a feature. >Jim> Lots of form fill ins put three digits with leading zeros. > >Chris> Actually this is the well known, or in other words accepted, >Chris> behavior. Here are a couple of typical responses to this question, >Chris> none specific to vxWorks: > >Well, at least HP gets it right (IMHO :-) . ...[snip irrelevant example showing that (s)scanf does indeed not treat decimal numbers as octal]... The behaviour of sscanf, although correct, is irrelevant, as it's not the function that is used to parse IP numbers. From the inet_addr man page:- INET(3N) NETWORK FUNCTIONS INET(3N) NAME inet inet_addr, inet_network, inet_makeaddr, inet_lnaof, inet_netof, inet_ntoa - Internet address manipulation ...[much snippage]... All numbers supplied as ``parts'' in a `.' notation may be decimal, octal, or hexadecimal, as specified in the C language (that is, a leading 0x or 0X implies hexadecimal; otherwise, a leading 0 implies octal; otherwise, the number is interpreted as decimal). Now if there was an implementation of (s)scanf out there that _did_ parse a number beginning with 0 as octal when the format specifier was %d, that _would_ be wrong. But go on, show us how HP gets it right: post us the output that results when you try and ping an IP such as x.x.0.0xx. Here's what WinNT does:- F:\TM>ping 191.0.1.090 Pinging 191.0.1.72 with 32 bytes of data: Request timed out. ^C F:\TM>ping 191.0.1.099 Pinging 191.0.1.81 with 32 bytes of data: Reply from 191.0.1.81: bytes=32 time=10ms TTL=60 Reply from 191.0.1.81: bytes=32 time<10ms TTL=60 Reply from 191.0.1.81: bytes=32 time<10ms TTL=60 ^C F:\TM>ping 191.0.1.99 Pinging 191.0.1.99 with 32 bytes of data: Request timed out. ^C F:\TM>ping 191.0x0.0x1.0x99 Pinging 191.0.1.153 with 32 bytes of data: Request timed out. ^C F:\TM>ping 191.0x0.0x1.099 Pinging 191.0.1.81 with 32 bytes of data: Reply from 191.0.1.81: bytes=32 time=10ms TTL=60 Reply from 191.0.1.81: bytes=32 time<10ms TTL=60 ^C Satisfied? Oh, here's one more, just for luck: F:\TM>ping 0xbf000151 Pinging 191.0.1.81 with 32 bytes of data: Reply from 191.0.1.81: bytes=32 time<10ms TTL=60 Reply from 191.0.1.81: bytes=32 time<10ms TTL=60 ^C F:\TM> Hope this makes it clear, DaveK - -- I wouldn't recommend sex, drugs or insanity for everyone, but they've always worked for me. -- Hunter S. Thompson --------------------------- Newsgroups: comp.os.vxworks Subject: UDP Multicast Loopback and Multihoming Date: Sat, 29 Apr 2000 01:25:46 -0700 From: "Josh Williams" Organization: Posted via Supernews, http://www.supernews.com Message-ID: Hello fellow vwWorkers, background: target - est8260 OS: vxWorks 5.4 env - Tornado 2.0 target IP: 192.186.0.15:FFFFFF00 NT host IP: 192.186.0.2 Is it possible to loopback UDP multicast datagrams on a target with only one network interface, and route those datagrams to IPs added to that network interface with ifAddrAdd? sendto is reporting success, but my recvfrom never gets the packet. My sendto and recvfrom function calls are located within two separate tasks on the target. Must I add the multicast IPs to the network interface? Using the setsockopt for IP_MULTICAST_LOOP returns an error, argh! Is there a way to do this? Thanks, Josh --------------------------- Newsgroups: comp.os.vxworks Subject: NT VxSim memory pool size Date: Fri, 12 May 2000 13:31:23 -0400 From: Luc Pariseau Organization: Bay Networks, Inc. Message-ID: <391C3FEB.DF67541@nortelnetworks.com> Looks like the default memory pool size in VxSim is 2Megs. Does anybody know how to change that? I tried changing LOCAL_MEM_SIZE which is set to (simMemSize) but that didn't work. (In fact VxWorks didn't come up after that). Tornado 2.0, VxSim on NT Thanks, Luc --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NT VxSim memory pool size Date: Fri, 12 May 2000 13:48:47 -0400 From: Erick Gonzalez Organization: Marconi Message-ID: <391C43FF.43FF87CA@fore.com> References: <391C3FEB.DF67541@nortelnetworks.com> Luc: Start your simulator with the option /r[size in bytes] (the documentation says it's in kbytes, but that's wrong) for example: .../Tornado/target/config/simpc/vxworks /r4000000 - -> memShow SUMMARY: status bytes blocks avg block max block ------ --------- -------- ---------- ---------- current free 3644000 2 1822000 3625432 alloc 55960 29 1929 - cumulative alloc 55960 29 1929 - value = 0 = 0x0 Erick Luc Pariseau wrote: > Looks like the default memory pool size in VxSim is 2Megs. > Does anybody know how to change that? > I tried changing LOCAL_MEM_SIZE which is set to (simMemSize) > but that didn't work. (In fact VxWorks didn't come up after that). > > Tornado 2.0, VxSim on NT > > Thanks, > Luc --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Address Pipelining Date: Fri, 12 May 2000 19:10:34 +0100 From: "ì" Organization: Customer of Planet Online Message-ID: <8fhhi4$9br$1@news7.svr.pol.co.uk> References: <8ff8ci$ath$2@newsg2.svr.pol.co.uk> <391B4978.CEFC9E4D@nowhere.com> Charles, I found out earlier today that you cannot disable address pipelining on the Universe II chip from Tundra who make the chip. Address pipelining is where you do not release the address line only the data line after each 16 bit transfer, for the next transfer you change the address line, without clearing it. Non-address pipelining is where you clear the address and data lines after each 16 bit transfer. Andy Charlie Grames wrote in message news:391B4978.CEFC9E4D@nowhere.com... > Andy, > > Can you define what you mean by "address pipelining?" I am not familiar with > the term in either a VMEbus or PowerPC context. > -- > Charlie Grames > The Boeing Company > Charles.R.Grames @ boeing.com --------------------------- Newsgroups: comp.os.vxworks Subject: HOT JOBS VxWorks Date: Fri, 12 May 2000 14:52:57 -0400 From: "PROCOM" Organization: Posted via Supernews, http://www.supernews.com Message-ID: C++ / OOD VxWorks Software Developers Description: Our client, one of the world's largest telecommunications companies, currently requires 10 Junior to Senior Software Specialists to design, develop, implement and support Wireless products and applications. Minimum Requirements: All candidates should possess a minimum of 1 year experience developing real-time C/C++, OOD S/W on a VxWorks platform. Desired skills would include any previous Wireless development experience. Type: Full-Time Placement Compensation: Open PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Interested candidates are invited to forward their resumes or questions in confidence to: Derek Weber PROCOM 300 March Rd Suite 600 Kanata, Ontario K2K-2E2 613-270-9339 x231 613-270-9449 (FAX) derekw@procom.ca www.procom.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NT VxSim memory pool size Date: Fri, 12 May 2000 15:05:24 -0400 From: Luc Pariseau Organization: Bay Networks, Inc. Message-ID: <391C55F4.428E2079@nortelnetworks.com> References: <391C3FEB.DF67541@nortelnetworks.com> <391C43FF.43FF87CA@fore.com> Almost there... If I start VxSim from the command line how do I start the target server. (The documentation is "poor" to say to least for this.) But I'm getting close to the correct guess I think: tgtsvr -V -B wdbpipe -R C:/TEMP/tsft -RW I tried different target names but I get an error window: tgtsvr (vxsim@lpariseau@lpariseau): Fri May 12 14:47:14 2000 Wind River Systems Target Server: NT/Win95 version Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Attaching pecoff OMF reader for SIMNT CPU family... succeeded. Error: Couldn't open D:\Tornado\target\config\simpc\vxworks: status = 0x2 Error during target server core initialization Target Server will exit A little more help? Thanks! Erick Gonzalez wrote: > Luc: Start your simulator with the option > > /r[size in bytes] (the documentation says it's in kbytes, but that's > wrong) > > for example: > > .../Tornado/target/config/simpc/vxworks /r4000000 > > -> memShow > > SUMMARY: > > status bytes blocks avg block max block > ------ --------- -------- ---------- ---------- > current > free 3644000 2 1822000 3625432 > alloc 55960 29 1929 - > cumulative > alloc 55960 29 1929 - > value = 0 = 0x0 > > Erick > > Luc Pariseau wrote: > > > Looks like the default memory pool size in VxSim is 2Megs. > > Does anybody know how to change that? > > I tried changing LOCAL_MEM_SIZE which is set to (simMemSize) > > but that didn't work. (In fact VxWorks didn't come up after that). > > > > Tornado 2.0, VxSim on NT > > > > Thanks, > > Luc --------------------------- Newsgroups: comp.os.vxworks Subject: PPP proxy arp entry gets deleted Date: Fri, 12 May 2000 19:06:36 GMT From: dgolden8928@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8fhknf$i75$1@nnrp1.deja.com> I have a vxworks system running using the pcpentium BSP which contains one PCI ethernet adapter and a PPP link via com1. I have the PPP options set to enable proxy arp for the PPP link. The intention is for the IP device on the PPP link to be visible from the ethernet side. This all works just fine (for a while). After a short period of time, the arp entry for the ppp link is dropped from the arp table (as shown with arpShow), and attempts to ping from another PC on the ethernet, through the vxworks system, to the IP device on the PPP link fail. They fail because the vxworks system no longer answers the arp requests. Pings **do** work from the vxworks system shell to the other side of the ppp link. Has anyone run into this behavior? I thought proxy arp entries were permanent. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Slow GCC compiler in Tornado II Date: 12 May 2000 19:26:51 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: Customer of Planet Online Message-ID: References: <8fhan9$sjk22@overload.lbl.gov> Where's TEMP and TMP set to point to? A local harddrive, or a network drive? You _might_ be able to use -pipe (YMMV). Does you path contain any network drives before the Tornado directory? Alternatively, get yourself a Sun. Ours goes very fast indeed and it's one of the cheap ones :-) Rgds! Luke In article <8fhan9$sjk22@overload.lbl.gov>, Jim Way wrote: >I just checked my Task Manager. With a typical set of apps open, I'm using >220 Meg of RAM. I would suggest that buying additional RAM is probably a >quicker and cheaper solution than trying to wedge another compiler into >Tornado. IMHO. > >Jim > >> I am using Tornado II, hosted on Windows NT and targetting a PowerPC 604 >> processor. This compiler appears to be very inefficient, as it causes >> continuous disk swapping (on a 96 Meg. machine), which of course makes >> it very slow. Compiling relatively small files can take 5 minutes. - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Leading zero in addHost() IP address Date: Fri, 12 May 2000 13:21:27 -0600 From: Doug Gibbons Organization: Lucent Technologies Message-ID: <391C59B7.B4C17C8C@lucent.com> References: <8fa9n9$cpu20@overload.lbl.gov> "Barkley, Charles (HT-EX)" wrote: > > This message is in MIME format. Since your mail reader does not understand > this format, some or all of this message may not be legible. > > ------_=_NextPart_001_01BFB9FD.AE31CEFC > Content-Type: text/plain > > Hello all, > > THE PROBLEM: > When a leading zero is used as part of the IP address passed to addHost(), > the octet having the leading zero is incorrectly entered into the host > table. For instance, if "168.083.250.083" is passed as a IP parameter to > addHost, a subsequent showHost() indicates that the IP address entered in > the host table is "168.67.250.67". Incrementing the "083" to "084" results > in the corresponding octet in the host table becoming "68". However, if > "002" is entered, the host table entry octet will be "2". Incrementing to > "003" results in a table entry of "3". It appears to be a VxWorks problem, > not an application problem. A lot of responses to this problem involved a C viewpoint. Seems to me the definition of dotted IP notation should really be language independent. I went off to find an RFC for this but truncated the search after finding this paragraph in RFC943: One commonly used notation for internet host addresses divides the 32-bit address into four 8-bit fields and specifies the value of each field as a decimal number with the fields separated by periods. This is called the "dotted decimal" notation. For example, the internet address of USC-ISIF.ARPA in dotted decimal is 010.002.000.052, or 10.2.0.52. There you go. Examples of an RFC in which dotted DECIMAL notation treats the leading zeros as superfluous. Of course we still have to deal with the reality of the situation... - -- Lee Douglas Gibbons L u c e n t T e c h n o l o g i e s ldgibbons@lucent.com Bell Labs Innovations --------------------------- Newsgroups: comp.os.vxworks Subject: From VxWorks to Linux for instrument controllers??? Date: Fri, 12 May 2000 12:30:05 -0700 From: "Gregor" Organization: SSO-IT, Hewlett-Packard Co. Message-ID: <8fhm41$l32$1@web1.cup.hp.com> I heard people talking about Linux as an "RTOS" for embedded devices. I am using VxWorks 5.4 on a PPC860/40 with 32 Mbytes of RAM. All hard real-time tasks are done on a dedicated SHARC DSP with its own 32 Mbytes of RAM. Is there a good reason not to consider Linux? Is it much worse than VxWorks for my kind of application? Or is it something I should seriously evaluate? Thanks for you input, Gregor --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SYM53C895 problem in pci Date: Fri, 12 May 2000 19:40:23 GMT From: Darren Etheridge Organization: Deja.com - Before you buy. Message-ID: <8fhmn4$kr9$1@nnrp1.deja.com> References: <8fgt11$npt$1@nnrp1.deja.com> You don't say what CPU you are using or what VxWorks BSP. I'm assuming you (or an external entity such as a BIOS) have mapped the PCI device to 67M. Do you mean you have read back the PCI registers to know it is at this address? If its on an Intel X86 CPU (or some others) you would need to indicate to the MMU that the PCI device maps at this address 67M), and make sure the memory is readable/writable and not cachable by the CPU. Then you would just access it at this address as if it were any "real" memory address. I hope this helps. Darren In article <8fgt11$npt$1@nnrp1.deja.com>, cismroom@my-deja.com wrote: > Hi: > All! > I use SYM53C895 and make some experiments.When reading registers,I > can find that the memory address of its 4k RAM is beyand the highest > address of physic memory.For example,I have memory 64M,its memory > address starts from 67M.So I can not figure out how to read from or > write to its RAM. I am not so familiar with PCI. > Thanks for your any suggestion! > cimsroom > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VGA VIDEO MODES Date: Fri, 12 May 2000 20:10:26 GMT From: Darren Etheridge Organization: Deja.com - Before you buy. Message-ID: <8fhof5$ms6$1@nnrp1.deja.com> References: <391AB4C1.F53482B1@reading.ac.uk> <8fe913$pq5$1@nnrp1.deja.com> <391B38AE.42537CBB@reading.ac.uk> Try: http://web.inter.nl.net/hcc/S.Weijgers/FreeVGA/home.htm for more information. In article <391B38AE.42537CBB@reading.ac.uk>, James Pascoe wrote: > Hi Guys, > > Yep, I was dreading that. does anybody know which port to switch (in order to > chage the video mode)? I have been bashing my head against a web browser all > afternoon :)) > > Jim > > Darren Etheridge wrote: > > > > Hi James, > > > > You are trying to make a BIOS call from a protected mode operating > > system. (Most) BIOS calls can only be made from Real mode. You will > > need to hit the VGA IO Ports directly using 'sysOutByte' etc. to switch > > the video mode. > > > > Hope this helps > > > > Darren > > > > In article <391AB4C1.F53482B1@reading.ac.uk>, > > James Pascoe wrote: > > > Hi all, > > > > > > What I need to do is to change the video mode on a 486 pc to 320 x > > 200 so > > > that I can do some elementary pixel plotting with students. If I was > > using > > > borland C, I would do the following: > > > > > > void setMode13h() > > > { > > > union REGS regs; > > > > > > regs.x.ax = 0x13; > > > int86(0x10, ®s, ®s); > > > } > > > > > > This is obviously borland specific, so I wrote the following asm > > statement in > > > Tornado: > > > > > > __asm__("movl $13h, eax\n > > > intl $10h" > > > : > > > : > > > ); > > > > > > When I execute this, I get the error: > > > > > > Invalid TSS > > > > > > and the program crashes. > > > > > > My assembler is so rusty, that I am probably making a 'kick me' with > > the above. > > > > > > What does anybody think? > > > > > > James > > > > > > -- > > > > > ************************************************************************ > > *** > > > * James Pascoe (research student) | Tel: +44 118 9875123 Ext 7626 / > > 4223 * > > > * Department of Computer Science | Fax: +44 118 9751994 > > * > > > * The University of Reading | > > * > > > * RG6 6AY | Email: J.S.Pascoe@reading.ac.uk > > * > > > * United Kingdom | James@james- > > pascoe.com * > > > * ---------------------------------------------------------------- - ---- > > --- * > > > * Official URL: > > http://www.cs.reading.ac.uk/cs/people/jsp/home.html * > > > * Personal URLs: http://www.james-pascoe.com/ | http://james- > > pascoe.i.am/ * > > > > > ************************************************************************ > > *** > > > > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > -- > ************************************************************************ *** > * James Pascoe (research student) | Tel: +44 118 9875123 Ext 7626 / 4223 * > * Department of Computer Science | Fax: +44 118 9751994 * > * The University of Reading | * > * RG6 6AY | Email: J.S.Pascoe@reading.ac.uk * > * United Kingdom | James@james- pascoe.com * > * -------------------------------------------------------------------- - --- * > * Official URL: http://www.cs.reading.ac.uk/cs/people/jsp/home.html * > * Personal URLs: http://www.james-pascoe.com/ | http://james- pascoe.i.am/ * > ************************************************************************ *** > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: From VxWorks to Linux for instrument controllers??? Date: Fri, 12 May 2000 16:27:13 -0400 From: Erick Gonzalez Organization: Marconi Message-ID: <391C6921.C25DDA19@fore.com> References: <8fhm41$l32$1@web1.cup.hp.com> Linux is NOT a RTOS. It may be a great OS on the desktop, or in server applications, but I don't think that you can get away with using it for a real time system, unless your application doesn't really require reliable and speedy preemptive task scheduling (among other goodies)... but then that would make it a non RT app wouldn't it? - -My two cents... Erick Gregor wrote: > I heard people talking about Linux as an "RTOS" for embedded devices. I am > using VxWorks 5.4 on a PPC860/40 with 32 Mbytes of RAM. All hard real-time > tasks are done on a dedicated SHARC DSP with its own 32 Mbytes of RAM. Is > there a good reason not to consider Linux? Is it much worse than VxWorks for > my kind of application? Or is it something I should seriously evaluate? > > Thanks for you input, > > Gregor --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NT VxSim memory pool size Date: Fri, 12 May 2000 16:30:58 -0400 From: Erick Gonzalez Organization: Marconi Message-ID: <391C6A02.E77BA491@fore.com> References: <391C3FEB.DF67541@nortelnetworks.com> <391C43FF.43FF87CA@fore.com> <391C55F4.428E2079@nortelnetworks.com> Here is what I did: Create a script or batch file that invokes the simulator with the right options and specify it as your custom built simulator in the Tornado GUI. That way you can let Tornado start the target server for you.... Erick Luc Pariseau wrote: > Almost there... > > If I start VxSim from the command line how do I start the target server. > (The documentation is "poor" to say to least for this.) > > But I'm getting close to the correct guess I think: > tgtsvr -V -B wdbpipe -R C:/TEMP/tsft -RW > > I tried different target names but I get an error window: > > tgtsvr (vxsim@lpariseau@lpariseau): Fri May 12 14:47:14 2000 > Wind River Systems Target Server: NT/Win95 version > Connecting to target agent... succeeded. > Attaching C++ interface... succeeded. > Attaching pecoff OMF reader for SIMNT CPU family... succeeded. > Error: Couldn't open D:\Tornado\target\config\simpc\vxworks: status = > 0x2 > Error during target server core initialization > Target Server will exit > > A little more help? > > Thanks! > > Erick Gonzalez wrote: > > > Luc: Start your simulator with the option > > > > /r[size in bytes] (the documentation says it's in kbytes, but that's > > wrong) > > > > for example: > > > > .../Tornado/target/config/simpc/vxworks /r4000000 > > > > -> memShow > > > > SUMMARY: > > > > status bytes blocks avg block max block > > ------ --------- -------- ---------- ---------- > > current > > free 3644000 2 1822000 3625432 > > alloc 55960 29 1929 - > > cumulative > > alloc 55960 29 1929 - > > value = 0 = 0x0 > > > > Erick > > > > Luc Pariseau wrote: > > > > > Looks like the default memory pool size in VxSim is 2Megs. > > > Does anybody know how to change that? > > > I tried changing LOCAL_MEM_SIZE which is set to (simMemSize) > > > but that didn't work. (In fact VxWorks didn't come up after that). > > > > > > Tornado 2.0, VxSim on NT > > > > > > Thanks, > > > Luc --------------------------- Newsgroups: comp.os.vxworks Subject: Qnx NTO Vs Vxworks Date: Fri, 12 May 2000 13:44:48 -0700 From: Joe Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <136afb46.a6724309@usw-ex0104-033.remarq.com> I am starting a new project and trying to select a RTOS. I have little experience on RTOS. The target hardware is using x86. I know those two OS have the biggest market share in commercial RTOS. Does anyone has working experience on both and has preference on one over the other? What would be the deciding factors? If this topic has been discussed, please give me a link for the topic. Thanks. P.S. I also posted this in comp.os.qnx, just want to get a different opinion. * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: From VxWorks to Linux for instrument controllers??? Date: Fri, 12 May 2000 19:22:51 -0400 From: "Jeff Creem" Organization: MV Communications, Inc. Message-ID: <8fi3le$bie$1@pyrite.mv.net> References: <8fhm41$l32$1@web1.cup.hp.com> <391C6921.C25DDA19@fore.com> Umm... www.rtlinux.com There are others as well...I have not tried it for anthing tough but looking at the approach they use I see no reason why it can not meet hard real-time deadlines as well as vxWorks. (a little more bloated on the memory side though). The approach is different than vxWorks but still seems valid. Erick Gonzalez wrote in message news:391C6921.C25DDA19@fore.com... > Linux is NOT a RTOS. It may be a great OS on the desktop, or in server > applications, but I don't think that you can get away with using it for a real > time system, unless your application doesn't really require reliable and speedy > preemptive task scheduling (among other goodies)... but then that would make it > a non RT app wouldn't it? > > -My two cents... > > Erick > > Gregor wrote: > > > I heard people talking about Linux as an "RTOS" for embedded devices. I am > > using VxWorks 5.4 on a PPC860/40 with 32 Mbytes of RAM. All hard real-time > > tasks are done on a dedicated SHARC DSP with its own 32 Mbytes of RAM. Is > > there a good reason not to consider Linux? Is it much worse than VxWorks for > > my kind of application? Or is it something I should seriously evaluate? > > > > Thanks for you input, > > > > Gregor > --------------------------- Newsgroups: comp.os.vxworks Subject: T2 objdumpppc on NT Date: Sun, 30 Apr 2000 15:24:08 GMT From: Pete Kockritz Organization: @Home Network Message-ID: <390C5041.FD9B5E91@home.com> I can't get the command objdumpppc -S file.o to give me a listing with C source intermixed with assy code. I get the assy listing, but no C source. The files are compiled with -g, and I've also tried -ggdb. I did get it to work once, when I edited the Makefile to change SRC_DIR from the full pathname of the source directory to './' and ran it from the command line. But that causes the build (when started from the Tornado GUI) to fail, since it can't find the source files. I've used this objdumpppc cmd on T1 under Solaris and it worked fine. Is there some trick or workaround to allow this to work on NT? My guess is the path to the source files is either not correctly put in the .o file, or it is corrupted somehow so that objdumpppc can't locate the source file. Regards, Pete - -- LinuxPPC --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindView Date: Sat, 13 May 2000 01:58:13 GMT From: "Paddy" Organization: Time Warner Cable of San Diego, CA Message-ID: References: <8fgp4o$sjk15@overload.lbl.gov> <391C2548.7A637847@nowhere.com> In Tornado2 you need to include the Windview components (under Development Components). Note the T2 default for timestamping is sequential timestamping which will not give you any absolute time information, you should deselect the default (sequential timestamping) and include the "system defined timestamping" component to get real time values. Most BSP's will support it. Patrick "Charlie Grames" wrote in message news:391C2548.7A637847@nowhere.com... > Andy, > > Did you configure WindView in your kernel? You need to add INCLUDE_WINDVIEW to > the included facilities section in configAll.h. > -- > Charlie Grames > The Boeing Company > Charles.R.Grames @ boeing.com > > > amass wrote: > > > > This is a multi-part message in MIME format. > > > > ------=_NextPart_000_001B_01BFBC08.2F562C00 > > Content-Type: text/plain; > > charset="iso-8859-1" > > Content-Transfer-Encoding: quoted-printable > > > > Hi > > > > We have an installation of Tornado 2/VxWorks 5.3.1 on an NT host = > > with a Radstone PPC1a target. We have recently attempted an installation = > > of WindView, but problems have been encountered. The product was = > > installed and configured what appears to be corectly, but whenever we = > > attemp to launch the tool, we get the following message: > > > > "target name@host name does not support WindView" > > > > Can anybody offer any suggestions as to the casue of such a problem? > > > > Thanks > > > > Andy Atkinson > > Software Engineer > > DERA Bincleaves > > Weymouth > > Dorset, DT4 8UR > > UK > > > > ------=_NextPart_000_001B_01BFBC08.2F562C00 > > Content-Type: text/html; > > charset="iso-8859-1" > > Content-Transfer-Encoding: quoted-printable > > > > > > > > > http-equiv=3DContent-Type> > > > > > > > > > >

Hi
> >
 
> >
    We have an installation of = > > Tornado=20 > > 2/VxWorks 5.3.1 on an NT host with a Radstone PPC1a target. We have = > > recently=20 > > attempted an installation of WindView, but problems have been = > > encountered. The=20 > > product was installed and configured what appears to be corectly, but = > > whenever=20 > > we attemp to launch the tool, we get the following message:
> >
 
> >
   "target name@host name does not = > > support=20 > > WindView"
> >
 
> >
Can anybody offer any suggestions as to the casue of = > > such a=20 > > problem?
> >
 
> >
Thanks
> >
 
> >
 
> >
Andy Atkinson
> >
Software Engineer
> >
DERA Bincleaves
> >
Weymouth
> >
Dorset, DT4 8UR
> >
UK
> > > > ------=_NextPart_000_001B_01BFBC08.2F562C00-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loading Question??? Date: Sat, 13 May 2000 02:23:28 GMT From: "Paddy" Organization: Time Warner Cable of San Diego, CA Message-ID: References: You have one more option. Basically module B can call all functions that reside in module A, but A cannot resolve B's symbols as it has already been loaded. If there are only a few symbols then you can resolve them in your code using symFindByName(). I don't remember the exact syntax but say module A wants to call a function in module B say fooB() then instead of calling fooB() directly, use a pointer to a function with same prototype as fooB() which is initialized using the symFindByName function. The code might look something like this /* extern int fooB(int) is the original decl..*/ int (*pFooB)(int); STATUS resolveSymbols(void) { if ((pFooB=symFindByName("fooB")==NULL) { logMsg("Could not resolve fooB()\n",0,0,0,0,0,0); return ERROR; } } void fooA(void) { int fooBReturnValue=0; fooBReturnValue=pFooB(100); } Remember you must resolve the symbol before calling it. Hope this helps. Patrick Fitzpatrick San Diego, CA 92117 "WDIZ" wrote in message news:MXFS4.49284$x4.1669818@newsread1.prod.itd.earthlink.net... > Dear All, > > How can I load 2 applications which depended on each other? > Does Vxworks support that? > > For example: > > in a.o > /* Function A */ > #include > extern void BPrintHello(); > void APrintHello( ){ printf("Hello A"); } > void CallB() { BPrintHello(); } > > in b.o > /* Function B */ > #include > extern void APrintHello(); > void BPrintHello(){ printf("Hello B"); } > void CallA() { APrintHello(); } > > > if I load a.o first > > ld < a.o > > I got the following error > > Undefined symbol: BPrintHello (binding 1 type 0) > value = 16088312 = 0xf57cf8 > > When I load b.o, I have no error. > Of course when I try to run CallB, I got the following error, > > -> CallB > > program > Exception current instruction address: 0x00000000 > Machine Status Register: 0x0008b130 > Condition Register: 0x44400082 > > 176dac vxTaskEntry +60 : shell () > 146df8 shell +18c: 146e24 () > 14702c shell +3c0: execute () > 1471ac execute +d8 : yyparse () > 18d8dc yyparse +7a8: 18b840 () > 18b9b8 yystart +8f8: CallB () > ffb23c CallB +10 : 0 () > shell restarted. > > How can I load a.o and b.o so I can run both CallB and CallA functions? > > Thank for your answer in advance. > -- Newcomer to Vxworks > > --------------------------- Newsgroups: comp.os.vxworks Subject: Fat32 File System Date: Fri, 12 May 2000 23:11:53 -0400 From: Fred Walther Message-ID: <391CC7F9.90413CB4@erols.com> Does anybody have any experience with Fat32 file system with VxWorks? Is it compatible with other operating systems files? such as windows 2000 What is the largest file one can create with vxWorks? any information will be helpful Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: motFecEnd & LXT970A Date: Sat, 13 May 2000 03:15:17 GMT From: tom@goldencom.com Organization: Deja.com - Before you buy. Message-ID: <8fihbr$hlh$1@nnrp1.deja.com> Hi all, I am porting motFecEnd END driver to my own target: Tornado 1.0.1 + VxWorks 5.3.1 + MPC860T + LXT970A I found the specification on how to set the motFecEndLoad() parameter: /* :::::: ::::: */ /* my setting */ /* IMMR=0xFA200000 */ #define FEC_LOAD_STRING "0xFA202000:0x02:0x8000:NONE:NONE:NONE:NULL:NULL:NULL:0x0:0x0:0x 0" I faked the configNet.h. I also defined some functions & variables which are required by the driver. But the muxDevLoad() returns NULL. Since I have .o & .h files only, I don't know how to incorporate it into my system exactly. Can you show me a "proved" example of corresponding configNet.h & the required functions? Regards, Tom Wang (tom@goldencom.com) Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: unsupported concurrent Telnet Server Date: Sat, 13 May 2000 03:36:29 GMT From: "News Chang" Organization: @Home Network Message-ID: <154T4.191677$bm.1018758@news1.alsv1.occa.home.com> Does anyone know how to get the Wind River unsupported concurrent Telnet Server source code that is 'suppose' to exist? My REP hasn't found any information on it, but WRS tech says it exist. Anyone using this code and would be willing to send a copy to me? Or is there a good source on the net? Thanks, Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RARP working with VxWorks? Date: Sat, 13 May 2000 04:56:00 GMT From: pwhicker@home.com (Paul Whicker) Organization: @Home Network Message-ID: References: <8fembb$sjk1@overload.lbl.gov> Wind River have this code, their FAE can get it fir you. Paul In article <8fembb$sjk1@overload.lbl.gov>, "Luke Skywalker" wrote: > >Hi, > >Does anyone know how I can get RARP running on VxWorks? > >Thanks, > >Bob >=== > > > > >________________________________________________________________________ >Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: unsupported concurrent Telnet Server Date: Sat, 13 May 2000 05:01:44 GMT From: pwhicker@home.com (Paul Whicker) Organization: @Home Network Message-ID: References: <154T4.191677$bm.1018758@news1.alsv1.occa.home.com> Your FAE needs to ask one of the more experienced members of that crowd, the software does exist and is on an internal server the corporate FAEs keep. Paul In article <154T4.191677$bm.1018758@news1.alsv1.occa.home.com>, "News Chang" wrote: >Does anyone know how to get the >Wind River unsupported concurrent Telnet Server >source code that is 'suppose' to exist? >My REP hasn't found any information on it, but >WRS tech says it exist. > >Anyone using this code and would be willing >to send a copy to me? >Or is there a good source on the net? > > >Thanks, >Mark > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SYM53C895 problem in pci Date: Sat, 13 May 2000 08:32:27 GMT From: cismroom@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8fj3ur$5an$1@nnrp1.deja.com> References: <8fgt11$npt$1@nnrp1.deja.com> <8fhmn4$kr9$1@nnrp1.deja.com> Thanks for your suggestion! I use X86 as target and Tornado 2.0,Vxworks 5.4. I think you means that I must add the description about memory in sysPhysMemDesc in sysLib.c,for example : I read the address of memory in SYM53C895 : 0x4200000 then I have to add a struct in sysPhysMemDesc as: { (void*) 0x4200000, (void*) 0x4200000, 0x1000, VM_STATE_MASK_VALID|VM_STATE_MASK_WRITABLE|EM_STATE_CACHEABLE,VM_STATE_V ALID|VM_STATE_WRITEBLE|VM_STATE_CACHEABLE_NOT } Now,after target start,I can use the memory in 895. Am I right? In article <8fhmn4$kr9$1@nnrp1.deja.com>, Darren Etheridge wrote: > You don't say what CPU you are using or what VxWorks BSP. > > I'm assuming you (or an external entity such as a BIOS) have mapped the > PCI device to 67M. Do you mean you have read back the PCI registers to > know it is at this address? > > If its on an Intel X86 CPU (or some others) you would need to indicate > to the MMU that the PCI device maps at this address 67M), and make sure > the memory is readable/writable and not cachable by the CPU. Then you > would just access it at this address as if it were any "real" memory > address. > > I hope this helps. > > Darren Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Problems with autosensing network connection Date: Wed, 10 May 2000 13:58:01 +0200 From: Manfred Fischer Organization: roNet GmbH Rosenheim Message-ID: <39194EC9.30B55D0A@steinbichler.de> I have problems with my MVME-2700 PowerPC cpu when I try to boot with initialisation of the network adapter. The CPU is connected to a autosensing 10/100 switch. On initialisation of network I get the message: "interrupt dc0 - no carrier". But if I set the port parameters of my switch to 100MBit half duplex, or if I put an 10/100 autosensing hub or a 10MBit hub between, everything works fine. I am looking for a solution which works without special configuration of my switch or hub. Is it possible, that the network interface get disturbed by the autosensing feature? Is the interface able of using a full duplex connection? If yes, how can I configure it in the boot-software? Can I force the System to a certain speed and duplex mode? Thanks Manfred --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun May 14 01:06:33 2000 From: "Jagannatha L.V" Date: Sun May 14 01:06:36 PDT 2000 Subject: DHCP server problem Dear Vxworks users, I am trying to do DHCP server and facing some problem. The problem is as follows: *I am not able to get the IP address from DHCP server How i did the dhcp server was: I included DHCP server and in usrNetDhcpsCfg.c file i added few entries into dadtabase table by removing the comment in dhcpsLeaseTbl definition(I actually uncommented the entry after the default one). I infact tried with different parameters in that entry. What i want to know is should i need to do any thing else in order to make dhcp server. If so reply me in solving this probem bye with regards from jagan.. __________________________________________________ Do You Yahoo!? Send instant messages & get email alerts with Yahoo! Messenger. http://im.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Sun May 14 18:00:46 2000 From: Fred Roeber Date: Sun May 14 18:00:49 PDT 2000 Subject: Re: VxWorks sizeof(int) question. On Fri, 12 May 2000, Bruce wrote: > The TI DSP is interesting. ANSI C guarantees that a sizeof(char) == 1 and > that sizeof(char) <= sizeof(short) <= sizeof(int) <= sizeof(long). This > leads me to conclude that you must have had 32bit chars also? Either that > or a non-conforming compiler. The compiler conforms. As you surmise, char's on the TI C40 DSP are also normally 32 bits by default such that: sizeof(char) == sizeof(int) == 1 Just goes to show that when you are trying to write portable code, you have to be able to handle all sorts of different architectural issues. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Sun May 14 18:13:53 2000 From: Fred Roeber Date: Sun May 14 18:13:55 PDT 2000 Subject: Re: Loading Question? On Thu, 11 May 2000 WDIZ asked: > How can I load 2 applications which depended on each other? > Does Vxworks support that? The easiest way I know of is to prelink the two object files together and and then download them in one shot. You can do this usually using the "-r" option to the "ld". This option preserves the undefined symbol references (which are later resolved when the file is downloaded. The -r is usually combined with the -X option to delete the temporary symbols generated during linking. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Mon May 15 02:19:53 2000 From: David Laight Date: Mon May 15 02:19:56 PDT 2000 Subject: re: UDP Multicast Loopback and Multihoming Josh asked: > Is it possible to loopback UDP multicast datagrams on a target with only one > network interface, and route those datagrams to IPs added to that network > interface with ifAddrAdd? By best guess is that you have a defective ethernet driver! Receiving broadcast (and I presume multicast) IP frames locally typically (and I don't see why the vxWorks IP stack should be different from any others) requires that the ethernet interface receive packets that it transmits (if it would recieve them is someone else transmitted them). Typically the ethernet hardware will not do this (the Sun FEPS chipset does it at 10M but not at 100M! - and you can't tell the speed). Your ethernet driver needs to look at the destination address of transmitted packets and generate receive packets for any it would receive if someone else transmitted them (ie packets to its own MAC address, broadcast packets, and multicast packets to enabled multicast addresses). Note that IP unicast packets specifying (one of) the systems one IP address(es) are normally turned around within the IP routing code, so don't need the ethernet driver to loop them back. David ---------------------------------------------------------------- David Laight email: dsl@tadpole.co.uk Tadpole Technology plc phone: +44 1223 428 232 Cambridge, UK fax: +44 1223 428 201 From vxwexplo-errs@csg.lbl.gov Mon May 15 03:25:43 2000 From: David Anderson Date: Mon May 15 03:25:45 PDT 2000 Subject: RE: Slow GCC compiler in Tornado II The lack-of-ram problem with Windows NT 4.0 has caught us out also. We have slow (233MHz) machines. We have NOT got the GCC compiler ( we use Diab ). Upgrading RAM from 64MBytes to 128MBytes cut compile times from 32 seconds to under 12 seconds for a trivial test program. NT appears to use around 32MBytes for itself... so 96MBytes is 'not alot' for a desktop machine. Or to put it another way... that's why we don't all use NT for our embedded products..... On 12 May 2000 17:12, the vxWorks Users Group Exploder [SMTP:vxwexplo@lbl.gov] wrote: > Submitted-by: Jim Way > > I just checked my Task Manager. With a typical set of apps open, I'm using > 220 Meg of RAM. I would suggest that buying additional RAM is probably a > quicker and cheaper solution than trying to wedge another compiler into > Tornado. IMHO. > > Jim > > > I am using Tornado II, hosted on Windows NT and targetting a PowerPC 604 > > processor. This compiler appears to be very inefficient, as it causes > > continuous disk swapping (on a 96 Meg. machine), which of course makes > > it very slow. Compiling relatively small files can take 5 minutes. > > From vxwexplo-errs@csg.lbl.gov Mon May 15 03:57:32 2000 From: "Tim Allen" Date: Mon May 15 03:57:35 PDT 2000 Subject: Re: bcopy_to_mbufs, bcopy_from_mbufs >Can one call the vxworks functions bcopy_to_mbufs & bcopy_from_mbufs from an >Interrupt Service Routine? Do these functions block at all? What exactly >do these functions do? The vxWorks documentation on these is pretty scarce! They are used in network drivers to copy data between device driver buffers and the network stack. The mbufs are chained and filled by the stack before being passed to the driver for transmission, and vice-versa. As to whether they block, I'd say bcopy_from_mbufs doesn't, because it fills a user supplied buffer, but bcopy_to_mbufs might block because I think it will allocate mbufs as required and I don't recall how this is done. The preferred method for network drivers to deal with data is to dispatch the work to task level from the ISR using netJobAdd(). You can see this in any network driver, or if you don't have source, you should find a template in target/src/drv/netif/templateNetif.c Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy From vxwexplo-errs@csg.lbl.gov Tue May 16 04:03:09 2000 From: Vxworks Exploder Date: Tue May 16 04:03:11 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue May 16 04:03:06 PDT 2000 Subject: VxWorks 5.4 and dosFs 2.0 Problems Subject: Tornado2 -> Windows 98 -> VmWare -> Linux Subject: Re: Downloading two files to PPC Subject: Re: Intel82559 ethernet driver for PowerPC Subject: Create of a symbol tyble (gdb format) Subject: PII and PIII Subject: Re: Tornado 2 on Windows 2000 Subject: VxWorks versus Nucleus ATI Subject: Re: Slow GCC compiler in Tornado II Subject: Re: need a loop within a makefile Subject: Re: PII and PIII Subject: Re: Tornado 2 on Windows 2000 Subject: tsfs & system debugging ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks 5.4 and dosFs 2.0 Problems Date: Sat, 13 May 2000 12:14:29 GMT From: epooleNO@UCEnetis.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <391d46ab.994293534@news.netis.com> We have begun work on a real time telecommunications device for high speed networks, based on the PowerPC (MCP-750) and VxWorks version 5.4. Our immediate objective is to get VxWorks and dosFs 2.0 running on the PowerPC and then to get the GoAhead web server running, both as a proof-of-concept and verification that we have the installation and configuration process down correctly, and as a platform on which to build our real application which will include a web server for administration. I have about 22 years' experience as a senior software engineer, primarily in embedded systems. In that time, I have completed one 1-year project using VxWorks, using command-line development (i.e. no Tornado) for a flash-resident application that did not use dosFs or any other file system. This project was completed about three years ago. I have no experience using Tornado other than what I've learned over the last couple of weeks (which, actually, has turned out to be more than enough to enable me to produce results). The other engineer working on this has an overall background comparable to mine, but with more experience in RTOS's other than VxWorks, and with about two months' experience with VxWorks and Tornado (obtained over the last two months' time so it's recent and current experience). So, we are experienced software and real time embedded systems engineers who don't happen to be VxWorks / Tornado experts. We have RTFM'd Wind River's "dosFs for Tornado 2.0 Release Notes and Supplement 2.0", "Tornado II getting started", and every applicable section of all of the applicable Tornado and VxWorks books and guides, over about the last three weeks' time. The software installation consists of: Tornado 2.0 for Power PC (TDK-12839-ZC-01) MCP750 / MCPN750 BSP (TDK-12979-ZC-00) DosFs 2.0 (TDK-13231-ZC-00) The target system has a diskette drive and an ATA hard drive connected to the secondary iDE controller. According to the other engineer on the project, the primary IDE controller is connected to something having to do with flash disk, and so can't be used for a hard drive. The host is a Windows 98 machine, which I wasn't sure would work but seems to work OK. I have also tried a Windows NT 4.0 machine with no difference in success. COM1 on the Power PC board is connected to a Linux machine running Seyon, an X-windows serial terminal program similar in purpose to Hyperterminal. I installed the CDs in the order shown above, and selected / enabled every option when installing. The default kernel ran OK, i.e. I was able to load and boot the kernel and connect to it via the target server and the Tornado shell. The serial terminal showed various messages ending with a "WDB: Ready" message, but no target system shell prompt or VxWorks banner (because the target system shell and banner had not yet been enabled and evidently aren't included by default). I then enabled the following kernel elements, starting with the default out-of-the-box configuration: development tool components -> target shell components -> shell banner development tool components -> target shell components -> target debugging development tool components -> target shell components -> target shell I rebuilt and loaded the kernel, and was able to get the VxWorks banner and command prompt on the target shell running on the serial terminal. I then enabled the following kernel elements: operating system components -> IO system components -> DOS filesystem development tool components -> show routines -> ata show routine hardware -> peripherals -> hard disks -> ATA hard drive Re-compiling the kernel gave compile errors for missing ATA_DEV0_STATE, ATA_DEV1_STATE, ATA_DEV2_STATE, ATA_DEV3_STATE, and IDE_CNTRLR0_INT_LVL. To fix this, I hand-edited config.h in the BSP directory to #define INCLUDE_ATA. (This leads to another question ... How do we know when we need to hand-edit something that didn't get updated by changing compile elements in the VxWorks tab in Tornado? None of the documentation I have gave any indication of the need to do this sort of thing ... I found it by searching the source tree for instances of the missing items, then looking to see what was #ifdef'd, and searching for the #ifdef items and enabling them by hand. But, I will guess that there are many VxWorks users who might not have the experience and knowledge to do this in code that they did not create. So, is the need to do that sort of thing documented anywhere?) (The other engineer on the project also feels that we are going to have to edit the project to add various functions and modules that aren't included by default and don't get included by enabling the various component elements in the VxWorks tab. If in fact we need to add modules to the project in order to get dosFs to work, is that fact documented anywhere?) So, this is the state we're in at the moment. Everything seems to compile and load OK, I get a target shell on the serial terminal, and I am able to connect via the target server and the Tornado shell. I also tried inserting some debugging code into the functions that invoke dosFs, so that they would display progress indications on the serial terminal screen during bootup. That seems to work OK, indicating that the file system drivers are at least getting loaded. Beyond that, I don't know where to go or what to do. I have no reference that tells me how to manipulate or display the data on the hard drive, how to do a directory listing, how to verify that the hard drive is installed properly and is visible to the operating system, nor even how to determine if the hard drive is partitioned and formatted yet. At this point I am convinced that we're playing with a few important picture cards short of a full deck around here. Clearly something is missing ... some elements of the file system driver, some documentation that shows how to use it all from either the target shell or the Tornado shell command line. If the other engineer is right and we're supposed to manually add other code to the project in order to get this to work, how can we determine what code we need to add other than to just "know" it? Likewise, it's clear that just tweaking the component elements in the VxWorks tab isn't sufficient; it's also necessary to hand-edit files like "config.h". How do we know how much and where that hand editing is needed, other than just trying this or that until the compiler stops complaining about missing symbols? I have reported all of this to Wind River customer support, and also discussed it at length with customer support engineers at the local Wind River regional office. So far I haven't been able to get any response from the Wind River customer support people, and as for the local support engineers, I went over with them everything I had done and they don't know what the problem is, and said they'd have to get back to me on Monday. Meanwhile if any of you has any ideas I'd much appreciate hearing about them. >-----------------------------------------------------------------< Eric Poole Specializing in Software Development for RKT Technologies, Inc. Medical Devices / Regulated Industries PO Box 1100 ISO 9001 / FDA Quality System Regulation Londonderry, NH 03053 Phone: 603-437-1811 Fax: 603-425-6475 >-----------------------------------------------------------------< E-mail: epooleNO@UCEnetis.com (Remove "NO" and "UCE" to reply by e-mail; use epoole-at-netis-dot-com) >-----------------------------------------------------------------< Qualifications and Background: see http://www.netis.com/rkt >-----------------------------------------------------------------< --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado2 -> Windows 98 -> VmWare -> Linux Date: Sat, 13 May 2000 14:54:49 GMT From: epooleNO@UCEnetis.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <391d69bc.1003272268@news.netis.com> Has anyone tried running Tornado 2 on Windows 98 on VmWare on Linux? My current development environment looks like that, and it works reasonably well (on a Pentium 550 with 128 MB RAM, 64 MB RAM given to Windows). But, it's a little unstable, and produces Blue Screens of Death somewhat more often than is normal for Windows. Would appreciate hearing from anyone who is doing the same sort of stuff. >-----------------------------------------------------------------< Eric Poole Specializing in Software Development for RKT Technologies, Inc. Medical Devices / Regulated Industries PO Box 1100 ISO 9001 / FDA Quality System Regulation Londonderry, NH 03053 Phone: 603-437-1811 Fax: 603-425-6475 >-----------------------------------------------------------------< E-mail: epooleNO@UCEnetis.com (Remove "NO" and "UCE" to reply by e-mail; use epoole-at-netis-dot-com) >-----------------------------------------------------------------< Qualifications and Background: see http://www.netis.com/rkt >-----------------------------------------------------------------< --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Downloading two files to PPC Date: Sat, 13 May 2000 16:10:21 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8fjup3$vu1$1@nnrp1.deja.com> References: <8f9kar$l97$1@nnrp1.deja.com> Hi, Might be a silly response, but have you checked that the file is (a) where you think it is and (b) has exactly the name you are specifying - including the same capitalisation etc.? Try doing an 'ls' from the shell you are using to download the images and see if that shows both the names OK. Regards, John... In article <8f9kar$l97$1@nnrp1.deja.com>, crandys@my-deja.com wrote: > Attempting to download a Client.out and a Server.out from NT to PPC604 target > with 64mbs of memory. When downloading the second file I get: WTX Error > 0x2(no such file or directory). Downloading order does not make any > difference. The file sizes are 3.3 mbs for the Client and 3.6 mbs for the > Server. Both the Client and the Server have been compiled with the - -mlongcall > option. I have done this successfully with smaller clients and servers. Has > anyone seen this problem or have any suggestions? Thanks C. Randy Schenk > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Intel82559 ethernet driver for PowerPC Date: Sat, 13 May 2000 18:43:40 GMT From: mike kelly Organization: Cogent Computer Systems, Inc. Message-ID: <391DA1A4.F031DF13@cogcomp.com> References: <391B358C.D3072B86@west.raytheon.com> Try looking at the various BSD flavors and Linux. Most support the PowerPC and the 82559. Usually the only issue is endianness and that is handled in the driver. Also, note that you may have to massage your PCI bridge code to handle big endian on the PCI bus, but you probably knew that already. Michael Polly Low wrote: > > I'm looking for an existing ethernet driver for the Intel82559 ethernet > controller that runs on the PowerPC processor with vxWorks. Instead of > starting from scratch, I like to find an existing one or one I can > modify from. The one driver posted in the Intel site only runs on the > Pentium processor and not PowerPC. Can anyone give me some ideas on > where I can possibly find one? > > Polly Low - -- **************************************************** * Michael J. Kelly, VP Engineering and Marketing * * Cogent Computer Systems, Inc. tel: 508-278-9400 * * 10 River Rd., Suite 205 fax: 508-278-9500 * * Uxbridge, MA 01569 web: www.cogcomp.com * **************************************************** * CMA: 32/64-Bit Microprocessor Development Boards * **************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Create of a symbol tyble (gdb format) Date: Mon, 15 May 2000 18:15:48 +0200 From: "Maik H." Organization: Nikoma Mediaworks GmbH Message-ID: Reply-To: "Maik H." Hi, C and C++ sources will be compiled by ccppc to object files without debug info. Is there a compiler option to get a symbol table in gdb format? - -- Cu, Maik --------------------------- Newsgroups: comp.os.vxworks Subject: PII and PIII Date: Mon, 15 May 2000 09:18:28 -0700 From: Ray Sprangers Organization: BCnet Message-ID: Hi VxWorkers, Has anyone got VxWorks to work on a Pentium II or a Pentium III? We are using the Pentium BSP and it is very slow on a Pentium III and doesn't work at all on a Pentium II. Is there a special BSP that is needed for a PII and PIII? Ray --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2 on Windows 2000 Date: Tue, 16 May 2000 02:49:35 GMT From: epooleNO@UCEnetis.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <3920b71b.168433300@news.netis.com> References: <391d6cc9.1004053395@news.netis.com> <391FB88F.94D7370D@auriga.ru> On Mon, 15 May 2000 12:42:55 +0400, "Arthur V. Grigorian" wrote: > I run it without any significant problems using 3 different BSPs. > Could you please state your W2k+Tornado2 related problems (if any)? I'm not running Win2K yet, but I'd read about others with problems and was just trying to get some specifics. >-----------------------------------------------------------------< Eric Poole Specializing in Software Development for RKT Technologies, Inc. Medical Devices / Regulated Industries PO Box 1100 ISO 9001 / FDA Quality System Regulation Londonderry, NH 03053 Phone: 603-437-1811 Fax: 603-425-6475 >-----------------------------------------------------------------< E-mail: epooleNO@UCEnetis.com (Remove "NO" and "UCE" to reply by e-mail; use epoole-at-netis-dot-com) >-----------------------------------------------------------------< Qualifications and Background: see http://www.netis.com/rkt >-----------------------------------------------------------------< --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks versus Nucleus ATI Date: Thu, 11 May 2000 17:54:51 GMT From: bmcgonig@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8fes53$go0$1@nnrp1.deja.com> Hi All: We're starting on a new project that will use either a Tensilica core or a MIPS core. With respect to the former, only Wind River and ATI support it with an RTOS. If you have experience or information that would sell me on ATI versus Wind River could you post your opinion please? thanx Brian Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Slow GCC compiler in Tornado II Date: Tue, 16 May 2000 03:49:12 GMT From: Karen Musgrove & Stephane Couture Organization: Sympatico Message-ID: <3920C67C.E2E7C0C5@REMOVETHIS.sympatico.ca> References: <8fp7r9$qv13@overload.lbl.gov> You should also try excluding the compile directory from your anti-virus software, it makes a huge difference on speed...

David Anderson wrote:

The lack-of-ram problem with Windows NT 4.0 has caught us out also. We have slow (233MHz)
machines. We have NOT got the GCC compiler ( we use Diab ). Upgrading RAM from 64MBytes
to 128MBytes cut compile times from 32 seconds to under 12 seconds for a trivial test program.
NT appears to use around 32MBytes for itself... so 96MBytes is 'not alot' for a desktop machine.

Or to put it another way... that's why we don't all use NT for our embedded products.....

On 12 May 2000 17:12, the vxWorks Users Group Exploder [SMTP:vxwexplo@lbl.gov] wrote:
> Submitted-by: Jim Way <Jway@datum.com>
>
> I just checked my Task Manager. With a typical set of apps open, I'm using
> 220 Meg of RAM. I would suggest that buying additional RAM is probably a
> quicker and cheaper solution than trying to wedge another compiler into
> Tornado. IMHO.
>
> Jim
>
> > I am using Tornado II, hosted on Windows NT and targetting a PowerPC 604
> > processor.  This compiler appears to be very inefficient, as it causes
> > continuous disk swapping (on a 96 Meg. machine), which of course makes
> > it very slow.  Compiling relatively small files can take 5 minutes.
>
>

--------------------------- Newsgroups: comp.os.vxworks Subject: Re: need a loop within a makefile Date: Tue, 16 May 2000 08:28:57 +0200 From: "Markus MItterer" Organization: KPNQwest customer news service Message-ID: <9T5U4.2$vk6.155@nreader1.kpnqwest.net> References: <39203F6F.98D651F2@dfrc.nasa.gov> Hello Ken, Thank you for your response but I couldn`t get it run. I know the for-function under DOS, but not with this syntax. Maybe I forgot to tell you that my host is a NT-maschine. MainRule: for var IN $(BASE_LIBS); \ do (Test.bat $$var); \ done I got this response: for var IN CChannel CControlHandler CLogHandler CSTKArraySem CTask ProFString CanInstall System intConnect; \ do (Test.bat $var); \ done var was unexpected at this time. It seems that I can`t call the for loop from a makefile - --------------------------- Markus Mitterer Software Development --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PII and PIII Date: Tue, 16 May 2000 09:01:27 +0200 From: Gilbert Cabillic Organization: Irisa - Campus de Beaulieu - RENNES Message-ID: <3920F247.48885724@irisa.fr> References: HI Ray. We use vxWorks on PIIs and PIIIs using a PentiumPro Bsp (not Pentium Bsp). It Runs well. Gilbert. // Gilbert Cabillic // Scratchy Project leader -- // http://www.irisa.fr/solidor/work/scratchy.html // IRISA - France Ray Sprangers wrote: > > Hi VxWorkers, > > Has anyone got VxWorks to work on a Pentium II or a Pentium III? We > are using the Pentium BSP and it is very slow on a Pentium III and > doesn't work at all on a Pentium II. Is there a special BSP that is > needed for a PII and PIII? > > Ray --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado 2 on Windows 2000 Date: Tue, 16 May 2000 08:18:10 +0200 From: Josef Baumgartner Organization: Lufthansa Systems GmbH Message-ID: <3920E822.B68D6FC7@gmx.de> References: <391d6cc9.1004053395@news.netis.com> Reply-To: jbnews@gmx.de Use the tool apcompat.exe which is located in the support folder of the Windows 2000 CD. Start apcompat.exe, select tornado.exe, select Windows NT 4 SP5. Regards, Josef "RKT Technologies, Inc." schrieb: > > Is anyone successfully running Tornado 2 on Windows 2000? > > I've read some problem reports but didn't see any solutions. > > Two e-mails directly to Wind River asking about Windows 2000 have so > far been ignored. > > Would appreciate hearing from anyone who is successfully running this > configuration. > > Thanks... > > >-----------------------------------------------------------------< > Eric Poole Specializing in Software Development for > RKT Technologies, Inc. Medical Devices / Regulated Industries > PO Box 1100 ISO 9001 / FDA Quality System Regulation > Londonderry, NH 03053 > Phone: 603-437-1811 Fax: 603-425-6475 > >-----------------------------------------------------------------< > E-mail: epooleNO@UCEnetis.com (Remove "NO" and "UCE" to > reply by e-mail; use epoole-at-netis-dot-com) > >-----------------------------------------------------------------< > Qualifications and Background: see http://www.netis.com/rkt > >-----------------------------------------------------------------< --------------------------- Newsgroups: comp.os.vxworks Subject: tsfs & system debugging Date: Tue, 16 May 2000 11:52:37 +0300 From: "Fornaro Antony" Organization: Intracom S.A., GREECE Message-ID: <8fr2kp$nb2$1@newsserv.intranet.gr> Hi all, I am working on TornadoII and PPC860 target (860fads). I am using a serial connection to the agent (I know, I know, bad choice) in order to accomplish system mode debugging. TSFS works fine when I am not debugging or when task debugging is performed. However, in system mode debugging, using TSFS to access a file drops the agent connection! Any ideas? thanks in advance --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue May 16 11:52:16 2000 From: "Tim Allen" Date: Tue May 16 11:52:18 PDT 2000 Subject: Re:VxWorks 5.4 and dosFs 2.0 Problems >Beyond that, I don't know where to go or what to do. I have no >reference that tells me how to manipulate or display the data on the >hard drive, how to do a directory listing, how to verify that the hard >drive is installed properly and is visible to the operating system, >nor even how to determine if the hard drive is partitioned and >formatted yet. You say you've RTFM, and I agree the information is often hard to find, as it's not always in one place, but in this case, I'd say the relevant sections were: Local File Systems chapter in the VxWorks Programmer's Guide (VPG), which gives some of the lower level detail. Shell chapter in the Tornado User's Guide, (and maybe Target Shell in the VPG) for how to do listings "ls" etc. Also, and this is well hidden, the ATA section in the x86 appendix of the VPG, which is also applicable to PPC. Where the board doesn't have a specific BSP for it, it is often the case that you have to use a mix of Tornado projects and "manual" configuration via source code changes. This seems to be improving through versions of Tornado though. Regards, Tim. -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy From vxwexplo-errs@csg.lbl.gov Tue May 16 17:14:14 2000 From: "CICON,JAMES (HP-Vancouver,ex1)" Date: Tue May 16 17:14:17 PDT 2000 Subject: Calling global obj member functions So we have a number of static global c++ objects. These objects are called by code and function correctly. But how do I call them via windShell? Here is an example file: >>>>> file start <<<<< #include "vxWorks.h" #include "config.h" #include "./myAll/configAll.h" #include "stdio.h" #include "strLib.h" ///// // Static object declaration and definition class testObj { private: char m_buf[256]; public: testObj(char *); void init(char *); public: void test(void); }; void testObj::test(void) { printf("%s\n", m_buf); } void testObj::init(char * buf) { strcpy(m_buf, buf); printf("initializing: %s\n", buf); } testObj::testObj(char * buf) { init(buf); } ///// // Globals testObj *objS; testObj objG("nothing happens here, this object appears in BSS, and it's constructor never runs."); ///// // Use the object void doit(void) { testObj objD("test object: dynamic"); objS = (testObj *) new testObj("test object: static"); objG.init("test object: global"); objD.test(); objS->test(); objG.test(); } >>>>> file end <<<<< So, if I had this file loaded, and wanted to call objG.test(), what is the syntax to make it happen. This does not work: -> objG.test() This does not work: -> "objG.test()" Jim Cicon Hewlett Packard From vxwexplo-errs@csg.lbl.gov Wed May 17 04:03:12 2000 From: Vxworks Exploder Date: Wed May 17 04:03:15 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed May 17 04:03:08 PDT 2000 Subject: DosFS 2.0 chkdsk changes directory to file.. Subject: Re: tsfs & system debugging Subject: Re: need a loop within a makefile Subject: Tornado2, location of custom BSPs Subject: UGL && SimNT Subject: Re: Downloading two files to PPC Subject: execution time of an application Subject: VT100 escape sequences under VxWorks 5.4 Subject: Extending DHCP server option tags in VxWorks Subject: Re: VxWorks employment Subject: PS2 Sample code Subject: PS2 Sample Code Subject: FA: HP VXI Data Acq System Subject: Am79C973 Driver Subject: Re: Downloading two files to PPC Subject: Re: NT VxSim multiple simulators Subject: Re:PII and PIII Subject: Re: PII and PIII Subject: etherOutput Subject: Re: VxWorks versus Nucleus ATI Subject: Re: From VxWorks to Linux for instrument controllers??? Subject: SNMP Manager on VxWorks platform Subject: Make a boot disk? Subject: Re: VxWorks 5.4 and dosFs 2.0 Problems Subject: Creating a vxWorks.res_rom image Subject: Re: need a loop within a makefile Subject: Re: etherOutput Subject: Re: VT100 escape sequences under VxWorks 5.4 Subject: Re: DosFS 2.0 chkdsk changes directory to file.. Subject: Re: etherOutput Subject: Re: need a loop within a makefile Subject: Re: PII and PIII Subject: Re: need a loop within a makefile Subject: image download problem Subject: TCP/IP port bindproblem on Vxwork 5.1.3 ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: DosFS 2.0 chkdsk changes directory to file.. Date: Tue, 16 May 2000 07:52:54 -0400 From: Burt Bicksler Organization: BBB Message-ID: Hi, We are working with TFFS/DosFS 2.0 running on an intel 386 based board. Here's what we are seeing..Randomly repeatable.. Start vxWorks. Create a directory on the flash device using the IOCTL call. Directory appears to create just fine, we can FTP in and see that it is present and is a directory with a size of 2K. Now we reboot the system.. During our application startup we run chkdsk as one of the first operations. chkdsk reports "errors" on the file system and "fixes" them. It reports that the "lost clusters" have been reclaimed. After this the directory that was created is now a 0 length file, NOT a directory. If we do the exact same operations but just change to the directory before rebooting then chkdsk appears to be happy after the reboot and we still have a directory. Has anyone else seen something similar? Remove the obvious spam bait from the e-mail address to reply directly. Thanks, Burt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tsfs & system debugging Date: Tue, 16 May 2000 14:10:40 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <39213AC0.8B91E135@bitband.com> References: <8fr2kp$nb2$1@newsserv.intranet.gr> Reply-To: leonid@bitband.com I am not really into all TFFS internals, but my best guess is that TFFS implements some timing (delay) functions with busy-loops, which are perhaps interrupt-locked, and that could cause some issues during system level debugging. I would rather expect though that system level debugging would somehow damage TFFS operations rather then vice versa. Hope this helps, Leonid Fornaro Antony wrote: > Hi all, > > I am working on TornadoII and PPC860 target (860fads). I am using a serial > connection to the agent (I know, I know, bad choice) in order to accomplish > system mode debugging. TSFS works fine when I am not debugging or when task > debugging is performed. However, in system mode debugging, using TSFS to > access a file drops the agent connection! > Any ideas? > > thanks in advance --------------------------- Newsgroups: comp.os.vxworks Subject: Re: need a loop within a makefile Date: Tue, 16 May 2000 15:04:12 +0200 From: Chris Varlese Organization: Ascom Switzerland, Bern, Switzerland Message-ID: <3921474C.A04A6841@no.mail.net> References: Hello Markus, I assume you want to do something like build each subproject. Here's a trick to have make call make on a list of subprojects. PRJ_LIST = Prj1 Prj2 Prj3 Prj4 MainRule: $(PRJ_LIST:%=%.make_all) .PHONY: %.make_all %.make_all: make -C$* all regards, Chris Varlese Ascom Transmission AG Bern Switzerland --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado2, location of custom BSPs Date: Tue, 16 May 2000 15:37:50 +0200 From: Chris Varlese Organization: Ascom Switzerland, Bern, Switzerland Message-ID: <39214F2E.B9B9DD40@no.mail.net> The BSPs are normally in the directory target/src/config. If I put a custom BSP there then it is accessible in the Tornado2 GUI environment when I make new projects. Does anyone know if it is possible to have the BSP in another place and tell the Tornado2 environment to pick them up from there. The problem is we would like the BSPs to be inside our source code workspace so they are under source management control etc. We want to avoid each developer in the team copying the BSP code to their local hard drive and the problems when they don't have the latest version etc. regards, Chris Varlese --------------------------- Newsgroups: comp.os.vxworks Subject: UGL && SimNT Date: Tue, 16 May 2000 07:11:51 -0700 From: Heiko Elger Organization: http://www.remarq.com: The World's Usenet/Discussions Start Here Message-ID: <0b12b743.7cfd46c4@usw-ex0102-016.remarq.com> Hi, Has anyone ported a UGL-driver for SimNT. I think for someone who knows the Win32 graphics API very good, it must be possible to to replace the UGL calls with it's equivalent function (Win32). I know that UGL 2.0 will have support for SimNT but I need it as sonn as possible. best regards Heiko Elger * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * The fastest and easiest way to search and participate in Usenet - Free! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Downloading two files to PPC Date: Tue, 16 May 2000 14:32:28 GMT From: crandys@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8frm5s$734$1@nnrp1.deja.com> References: <8f9kar$l97$1@nnrp1.deja.com> <8fjup3$vu1$1@nnrp1.deja.com> I can load either file by itself OK -- the problem occurs when I try to load the second file (after loading the first). It does not matter which file I load first. The first load will always be successful and is verifiable using moduleShow, a subsequent attempt to load the second file always fails. Randy In article <8fjup3$vu1$1@nnrp1.deja.com>, john_94501@my-deja.com wrote: > Hi, > > Might be a silly response, but have you checked that the file is (a) > where you think it is and (b) has exactly the name you are specifying - - > including the same capitalisation etc.? > > Try doing an 'ls' from the shell you are using to download the images > and see if that shows both the names OK. > > Regards, > > John... > > In article <8f9kar$l97$1@nnrp1.deja.com>, > crandys@my-deja.com wrote: > > Attempting to download a Client.out and a Server.out from NT to PPC604 > target > > with 64mbs of memory. When downloading the second file I get: WTX > Error > > 0x2(no such file or directory). Downloading order does not make any > > difference. The file sizes are 3.3 mbs for the Client and 3.6 mbs for > the > > Server. Both the Client and the Server have been compiled with the > -mlongcall > > option. I have done this successfully with smaller clients and > servers. Has > > anyone seen this problem or have any suggestions? Thanks C. Randy > Schenk > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: execution time of an application Date: Wed, 10 May 2000 14:46:52 +0200 From: "Gwendal.Peres" Organization: Universite de Metz, France Message-ID: Here's my problem: i create a task and execute it let's imagine that this task needs A seconds to finish (let's consider A as an integer to simplify).Of course A is unknown. Now i want to poll the time remaining before the task finishes every B seconds for example. How could i get that information ? If you have any idea thanks --------------------------- Newsgroups: comp.os.vxworks Subject: VT100 escape sequences under VxWorks 5.4 Date: Tue, 16 May 2000 11:20:22 -0400 From: JB Organization: none really Message-ID: <39216736.19613EDF@my-deja.com> Hi Group! I am using VxWorks v5.4 on a PowerPC target and compiling FORTRAN under the Green Hills toolset. The FORTRAN code has print statements to a serial device (tyCo0) utilizing VT100 escape sequences for cursor movement, etc. Using a Hypertem window (and an actual VT100 terminal), the escape sequences are not executed and are interpreted as normal text. The information in the VxWorks programmers guide is very minimal on this subject. My questions are 1) Does VxWorks support VT100 escape sequences printed to a serial tty device or 2) If not, what options would I have available? Thanks in advance! JB --------------------------- Newsgroups: comp.os.vxworks Subject: Extending DHCP server option tags in VxWorks Date: Tue, 16 May 2000 15:39:30 GMT From: "Keir Storrie" Message-ID: HI, I am currently working on an application that requires a DHCP server with some optional tags with values 128-131. I cannot see an interface function to map any new options. The only way I can see it working is by hacking the tables in dhcps.h & dhcp.h to include defines for the options I need and the functions for handling the input. Any comments on the above idea or alternatives would be greatly appreciated. Thanks in advance KEIR Keir Storrie Software Engineer Mitel Telecom Ltd keir_storrie@mitel.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks employment Date: Tue, 16 May 2000 18:06:20 -0500 From: "Keith McHugh" Organization: Esat Net Message-ID: <39217f1a.0@seat.s3two.ie> References: Reply-To: "Keith McHugh" Please do not repeat the same message again and again....This place has been dedicated to the most important questions and their discussions only. Don't waste the resources! "PROCOM" wrote in message news:si0e4jd2rj042@corp.supernews.com... > C++ / OOD VxWorks Software Developers > > Description: Our client, one of the world's largest telecommunications > companies, currently requires 10 Junior to Senior Software Specialists to > design, develop, implement and support Wireless products and applications. > > Minimum Requirements: > All candidates should possess a minimum of 1 year experience developing > real-time C/C++, OOD S/W on a VxWorks platform. > > Desired skills would include any previous Wireless development experience. > > Location: Ottawa, Ontario, Canada > > Type: Full-Time Placement or Contract > > Compensation: Open > > PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. > (Procom) is a national leader in the provision of Computer personnel on a > contract > and full-time basis. Our clients are comprised of the largest national and > international corporations that utilize technical resources extensively > across a wide range of disciplines. In the Financial Post (March 1999) > Procom was ranked as the 6th largest professional Services Company in > Canada. In November of 1999, Procom was named a Regional finalist in Canada' > s 50 Best Managed Private companies. Our track record is proven with more > than 180 consultants servicing Ottawa's high tech community and more than > 1600 Procom consultants currently on assignment throughout. North America. > For further information on this and other opportunities please visit our web > site at www.procom.ca. > > Interested candidates are invited to forward their resumes or questions in > confidence to: > > Derek Weber > PROCOM > > 300 March Rd Suite 600 > Kanata, Ontario > K2K-2E2 > 613-270-9339 x231 > 613-270-9449 (FAX) > > derekw@procom.ca > www.procom.ca > > > > Derek Weber > PROCOM > > 300 March Rd Suite 600 > Kanata, Ontario > K2K-2E2 > 613-270-9339 x231 > 613-270-9449 (FAX) > > derekw@procom.ca > www.procom.ca > > --------------------------- Newsgroups: comp.os.vxworks Subject: PS2 Sample code Date: Tue, 16 May 2000 17:13:30 GMT From: Kirstie Forsyth Organization: @Home Network Message-ID: <392181A9.877B70E3@Forsyth.net> Does anyone have sample code for a mouse/keyboard driver using PS2? - -- - ----------------------------------------------------- Click here for Free Video!! http://www.gohip.com/freevideo/ --------------------------- Newsgroups: comp.os.vxworks Subject: PS2 Sample Code Date: Tue, 16 May 2000 17:21:12 GMT From: "Gokul" Organization: @Home Network Message-ID: Does anyone have Sample code for a PS2 Keyboard/mouse?? --------------------------- Newsgroups: misc.industry.electronics.marketplace,sci.electronics.equipment,sci.electronics.design,comp.lang.labview,comp.os.vxworks Subject: FA: HP VXI Data Acq System Date: Wed, 10 May 2000 12:44:55 -0700 From: Harlan Labs Organization: Harlan Labs electronic test equipment Message-ID: <3919BC37.59E987F@harlanlabs.com> For auction on eBay: HP 75000B VXI Data Acquisition System http://cgi.ebay.com/aw-cgi/eBayISAPI.dll?ViewItem&item=329420800 Bidding starts at $250 ~ no reserve! Harlan Labs Electronic Test Equipment In San Diego since 1966 (858)578-4711 phone (858)578-4714 fax contact@harlanlabs.com http://www.harlanlabs.com --------------------------- Newsgroups: comp.os.vxworks Subject: Am79C973 Driver Date: Tue, 16 May 2000 10:34:25 -0700 From: sdinev Organization: ACC Message-ID: <392186A1.7D506F52@acc.com> Hi all: I am looking for a VxWorks Ethernet driver that supports multiple Am79C973 controllers. Could anyone offer such kind of driver? Stefan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Downloading two files to PPC Date: Tue, 16 May 2000 14:21:38 -0400 From: Erick Gonzalez Organization: Marconi Message-ID: <392191B1.BD042CC9@marconi.com> References: <8f9kar$l97$1@nnrp1.deja.com> <8fjup3$vu1$1@nnrp1.deja.com> <8frm5s$734$1@nnrp1.deja.com> moduleShow will display the module information even if undefined references were found during download... Erick crandys@my-deja.com wrote: > I can load either file by itself OK -- the problem occurs when I try to > load the second file (after loading the first). It does not matter > which file I load first. The first load will always be successful and > is verifiable using moduleShow, a subsequent attempt to load the second > file > always fails. > > Randy > > In article <8fjup3$vu1$1@nnrp1.deja.com>, > john_94501@my-deja.com wrote: > > Hi, > > > > Might be a silly response, but have you checked that the file is (a) > > where you think it is and (b) has exactly the name you are specifying > - > > including the same capitalisation etc.? > > > > Try doing an 'ls' from the shell you are using to download the images > > and see if that shows both the names OK. > > > > Regards, > > > > John... > > > > In article <8f9kar$l97$1@nnrp1.deja.com>, > > crandys@my-deja.com wrote: > > > Attempting to download a Client.out and a Server.out from NT to > PPC604 > > target > > > with 64mbs of memory. When downloading the second file I get: WTX > > Error > > > 0x2(no such file or directory). Downloading order does not make any > > > difference. The file sizes are 3.3 mbs for the Client and 3.6 mbs > for > > the > > > Server. Both the Client and the Server have been compiled with the > > -mlongcall > > > option. I have done this successfully with smaller clients and > > servers. Has > > > anyone seen this problem or have any suggestions? Thanks C. Randy > > Schenk > > > > > > Sent via Deja.com http://www.deja.com/ > > > Before you buy. > > > > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: NT VxSim multiple simulators Date: Tue, 16 May 2000 14:54:35 -0400 From: Luc Pariseau Organization: Bay Networks, Inc. Message-ID: <3921996B.5E89C0A7@nortelnetworks.com> References: <3920127A.47D30AB8@nortelnetworks.com> <8fpn6k$jer$12$1@news.t-online.com> VxSim (not Lite.) Somebody just pointed me to the VxWorks Programmers's Guide, p552. (thanks Sandip). It mentions that you need to locate the "Include Folder" using the Project Tool and also change the target server config from wdbpipe to wdbrpc.... maybe I'm just slow today but that is not very clear on how you do that exactly. The example needs to be a bit more specific. Luc Martin Raabe wrote: > Hello Luc, > are you talking about VxSim or VxSim Lite? > > Ciao > -- > Martin Raabe > > Luc Pariseau schrieb in im Newsbeitrag: > 3920127A.47D30AB8@nortelnetworks.com... > > > > If I try to start a second simulator from Tornado 2.0 > > I get the following error window: > > > > > > tgtsvr (vxsim1@lpariseau): Mon May 15 11:03:02 2000 > > Wind River Systems Target Server: NT/Win95 version > > Error: Another Target Server is already running on the pipebackend. > > Error: clnttty_bufcreate can't open pipe > > Error: Another Target Server is already running on the pipebackend. > > Error: clnttty_bufcreate can't open pipe > > Error: Another Target Server is already running on the pipebackend. > > Error: clnttty_bufcreate can't open pipe > > Error: Remote system error; Error 5 > > Error: Backend initialization routine failed. > > Problem during Backend initialization > > Target Server will exit > > > > > > Any idea? I don't see a way to specify the named pipe used. > > Thanks. > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re:PII and PIII Date: Tue, 16 May 2000 11:46:49 -0700 From: Ray Sprangers Organization: BCnet Message-ID: References: <3920F247.48885724@irisa.fr> Hi Gilbert, I tried the Pentium Pro BSP and it takes about one minute to display the splash screen after it has loaded the VxWorks image. Do you have the same problem? The same happens on the Pentium BSP. Ray - -----Original Message----- From: Gilbert Cabillic [mailto:Gilbert.Cabillic@irisa.fr] Posted At: Tuesday, May 16, 2000 12:01 AM Posted To: vxworks Conversation: PII and PIII Subject: Re: PII and PIII HI Ray. We use vxWorks on PIIs and PIIIs using a PentiumPro Bsp (not Pentium Bsp). It Runs well. Gilbert. // Gilbert Cabillic // Scratchy Project leader -- // http://www.irisa.fr/solidor/work/scratchy.html // IRISA - France Ray Sprangers wrote: > > Hi VxWorkers, > > Has anyone got VxWorks to work on a Pentium II or a Pentium III? We > are using the Pentium BSP and it is very slow on a Pentium III and > doesn't work at all on a Pentium II. Is there a special BSP that is > needed for a PII and PIII? > > Ray --------------------------- Newsgroups: comp.os.vxworks Subject: etherOutput Date: Tue, 16 May 2000 20:56:14 GMT From: "Gordon Manning" Organization: DataDirect Networks Message-ID: What exactly does the vxworks function etherOutput do? Can one call it from an ISR. Thanks in advance Gordon Manning --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks versus Nucleus ATI Date: Tue, 16 May 2000 15:08:20 -0700 From: Michael Morrison Organization: Intelligraphics, Inc. Message-ID: <3921C6D4.6160EE07@intelligraphics.com> References: <8fes53$go0$1@nnrp1.deja.com> <391f1aa8$0$216@nntp1.ba.best.com> Reply-To: mikem@intelligraphics.com Add to the list below that Nucleus comes with source, wrs does not (Unless you PAY) The only thing I didn't like about Nucleus was that you can't give a semaphore or write a queue from an ISR. You have to make what they call a "High Level ISR" which you activate from the ISR. Kind of a pain. Chris Wein wrote: > In article <8fes53$go0$1@nnrp1.deja.com>, wrote: > > > > > >Hi All: > > > >We're starting on a new project that will use either > >a Tensilica core or a MIPS core. > > > >With respect to the former, only Wind River and ATI > >support it with an RTOS. > > on the kernel eval > > We recently selected VxWorks over Nucleus. Here's a quick summary > > 1) Nucleus is royalty free whereas VxWorks is not. > 2) WRS tool seats are more expensive than ATI > 3) Nucleus' kernel size is smaller than VxWorks. For the config we > wanted, Nuke was around 70K vs 150K for VxWorks (SPARC v8) > 4) Both kernels are fast and efficient. > 5) Nucleus has a slightly more rich task communication mechanisms such > as events. > 6) ATI licenses source with all distributions. Costs more $$$ for WRS. > > Based on these points, you might wonder why we chose VxWorks. However, > if you are evaluating just the kernel then both are fine. If you > are evaluating an RTOS (which includes kernel + I/O + other components) > then VxWorks is the clear winner... > > 1) VxWorks supports debug over ethernet > 2) VxWorks has a fantastic I/O story. FTP client support, filesystems, > per-task stdin/stdout, I/O redirection right out of the box > 3) VxWorks has a well thought out set of "hooks" that allow apps to > bind to interesting events like task switch. > 4) VxWorks has superior 3rd party support for more esoteric things like > 1394 stacks, USB etc. > > If you need a small kernel in a heavily embedded system then go with ATI > but if you are scaling up and need more sophisticated I/O then you will > be very pleased with wrs. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: From VxWorks to Linux for instrument controllers??? Date: Tue, 16 May 2000 15:12:49 -0700 From: Michael Morrison Organization: Intelligraphics, Inc. Message-ID: <3921C7E1.38BF48A3@intelligraphics.com> References: <8fhm41$l32$1@web1.cup.hp.com> <391C6921.C25DDA19@fore.com> Reply-To: mikem@intelligraphics.com Erick Gonzalez wrote: > Linux is NOT a RTOS. It may be a great OS on the desktop, or in server > applications, but I don't think that you can get away with using it for a real > time system, unless your application doesn't really require reliable and speedy > preemptive task scheduling (among other goodies)... but then that would make it > a non RT app wouldn't it? > www.rtlinux.com www.zentropix.com www.lynx.com www.google.com search for "real time linux" Nuf said. --------------------------- Newsgroups: comp.os.vxworks Subject: SNMP Manager on VxWorks platform Date: Wed, 17 May 2000 01:01:35 GMT From: "Jon V" Organization: @Home Network Message-ID: I have a need for an SNMP manager running on VxWorks. This would be in addition to an SNMP agent. Has anybody implemented this? Did you use 3rd party SW or roll-your-own? TIA Jon Voichick --------------------------- Newsgroups: comp.os.vxworks Subject: Make a boot disk? Date: Tue, 16 May 2000 21:29:06 -0400 From: "News" Organization: The University of Western Ontario, London, Ont. Canada Message-ID: <8fss09$juh$1@panther.uwo.ca> Does any one know how to make a VxWorks boot disk for PC? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks 5.4 and dosFs 2.0 Problems Date: Wed, 17 May 2000 01:21:13 GMT From: epooleNO@UCEnetis.com (RKT Technologies, Inc.) Organization: Software Consultants to Medical Device Manufacturers Message-ID: <3921f30b.249325019@news.netis.com> References: <8fs6on$7h1@overload.lbl.gov> On Tue, 16 May 2000 19:51:19 +0100, "Tim Allen" wrote: >You say you've RTFM, and I agree the information is often hard to find, as >it's not always in one place, It gets better. On the suggestion of someone else here, who reported getting the wrong key for his DOSFS2 CD, I called the local Wind River office and explained the problem and what I'd read here on the newsgroup. They called me back a few hours later, and damned if it isn't true, there really is another key that unlocks the other part of the CD. They gave me that key, I re-installed the CD, and a whole bunch of kernel configuration options appeared like magic, along with about a dozen new documentation files that had all the information I needed. We're up and running now, sort of. Funny how the job gets easier when you're allowed to use the tools you paid for! Thanks for your suggestions, there are at least a couple of them I know I haven't seen yet so I'll add them to my casual reading list. Eric . >-----------------------------------------------------------------< Eric Poole Specializing in Software Development for RKT Technologies, Inc. Medical Devices / Regulated Industries PO Box 1100 ISO 9001 / FDA Quality System Regulation Londonderry, NH 03053 Phone: 603-437-1811 Fax: 603-425-6475 >-----------------------------------------------------------------< E-mail: epooleNO@UCEnetis.com (Remove "NO" and "UCE" to reply by e-mail; use epoole-at-netis-dot-com) >-----------------------------------------------------------------< Qualifications and Background: see http://www.netis.com/rkt >-----------------------------------------------------------------< --------------------------- Newsgroups: comp.os.vxworks Subject: Creating a vxWorks.res_rom image Date: Tue, 16 May 2000 18:59:14 -0700 From: SpamBamThankYouMaM Organization: Cadence Design Systems Message-ID: <3921FCF2.B53284CF@cadence.com> I have built a vxWorks image that runs from SDRAM. Now I want to rebuild (via make) this image to run from Flash (read only memory). I have edited the makefile and config.h to relect the Flash memory address space parameters. Are there other caveats/problems I should be aware of? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: need a loop within a makefile Date: Wed, 17 May 2000 02:19:13 GMT From: mcmorran@visi.net (Peter McMorran) Message-ID: <3922019c$1$zpzbeena$mr2ice@news.visi.net> References: In , on 05/15/00 at 02:09 PM, "Markus MItterer" said: >Hello VxWorkers, >I have a problem with make. Can anybody tell me how to use the >foreach function in makefiles? My problem is the following: >PRJ_LIST = Prj1 Prj2 Prj3 Prj4 >MainRule: > foreach(var, $(PRJ_LIST), call DoIt.bat $(var)) >How can I call a batch file each time with one member of the >PRJ_LIST. >Thank you in advance >Markus >--------------------------- >Markus Mitterer >Software Development HI Markus, The natural way to do this is to let make handle the loops. Make is a non-procedural language -- you give it a bunch of rules and a goal, then it figures out how to get there. For example, PRJ_LIST = Prj1 Prj2 Prj3 Prj4 all : $(PRJ_LIST) $(PRJ_LIST) : FORCE DoIt.bat $@ FORCE : This assumes that your make does know how to run a command shell that will handle the batch file. The last bit, with FORCE, ensures that each item will be processed, even if it already exists in the directory. With only a little more trickery, we can get that $(PRJ_LIST) line to do different actions by invoking a sub-make and passing variables to it. Post a question if you're interested. You mentioned in another post that you are using NT. We like the Cygwin environment on NT because it gives us a Unix-like command environment which is much nicer for make and scripting. Cheers, Peter - -- - ----------------------------------------------------------- mcmorran@visi.net (Peter McMorran) - ----------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: etherOutput Date: Tue, 16 May 2000 23:39:36 -0400 From: Thomas Davis Organization: Interpath Communications, Inc. Message-ID: <39221477.14C9A19C@galois.net> References: <8fspv3$h0p$1@nnrp1.deja.com> tom@goldencom.com wrote: > In article , > "Gordon Manning" wrote: > > What exactly does the vxworks function etherOutput do? I can't speak for all drivers. But in general this a direct out to the ethernet interface. Which almost exactly means, read the TMD registers, fill in out going ethernet address, copy into DMA buffer, write to controller's CSR regs to send the packet off. I do not believe that this call pends at all (but once again that may be driver dependent). The backgound of this is the "ethernet" interface into vxWorks. Just like device drivers there are standard callbacks that each driver assigns when it makes the call to "ether_attach". When you call etherOutput, it calls the output calls the callback in the ifnet structure for the interface. > > You can see the subroutine descriptions in "vxWorks Reference Manual": > It "sends a packet on the specified Ethernet interface by calling the > interface's output routine directly." > > > Can one call it from an ISR. > I don't know. Probably, OK in infrequent interupts with spare use. I mean you must get out of interupts as fast a possible. Anything you can avoid doing in an interrupt is probably a good idea to leave out. Just good practice. > > > > > Thanks in advance > > > > Gordon Manning > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. Hope that helps, Thomas Davis http://www.expressobeans.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VT100 escape sequences under VxWorks 5.4 Date: Wed, 17 May 2000 08:27:54 +0200 From: "Achim Zimmer" Organization: Colt Telecom GmbH Message-ID: <8fte2s$4h0$1@crusher.de.colt.net> References: <39216736.19613EDF@my-deja.com> What you print to the terminal is your turn. WRS can not limit that things. To invert the output f.e. (white foreground, black background) try this: printf(" normal, %s inverted\n", "\033[7m"); // this is ansi To bring it back to original you can do this printf(" normal, %s inverted\n", "\033[0m"); To move the cursor one step left you can do printf(" normal, %s inverted\n", "\033[1D"); Did it helps ???? Achim JB schrieb in im Newsbeitrag: 39216736.19613EDF@my-deja.com... > Hi Group! > > I am using VxWorks v5.4 on a PowerPC target and compiling FORTRAN > under the Green Hills toolset. The FORTRAN code has print statements to > a serial device (tyCo0) utilizing VT100 escape sequences for cursor > movement, etc. Using a Hypertem window (and an actual VT100 terminal), > the escape sequences are not executed and are interpreted as normal > text. The information in the VxWorks programmers guide is very minimal > on this subject. > > My questions are 1) Does VxWorks support VT100 escape sequences > printed to a serial tty device or 2) If not, what options would I have > available? > > Thanks in advance! > > JB --------------------------- Newsgroups: comp.os.vxworks Subject: Re: DosFS 2.0 chkdsk changes directory to file.. Date: Wed, 17 May 2000 09:06:28 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <392244F4.23FFAF41@bitband.com> References: Reply-To: leonid@bitband.com Seems you are rebooting too soon after creating the directory, it had the chanse to update the FAT but not the root (parent) directory, so there is a cluster allocated which is not referred from any directory entry, thus it would reclaim it in an attempt to salvage potentially lost data. If you do some more operations prior to rebooting, it will be able to flush the buffers and then all will be consistent. With TFFS, disk write operations are rather lenghy, perhaps you shold try the same with a RAM disk or some other media with the same DosFs setup. Hope this helps, Leonid Burt Bicksler wrote: > Hi, > > We are working with TFFS/DosFS 2.0 running on an intel 386 based board. > > Here's what we are seeing..Randomly repeatable.. > > Start vxWorks. > > Create a directory on the flash device using the IOCTL call. > > Directory appears to create just fine, we can FTP in and see that it is > present and is a directory with a size of 2K. > > Now we reboot the system.. > > During our application startup we run chkdsk as one of the first > operations. chkdsk reports "errors" on the file system and "fixes" them. > It reports that the "lost clusters" have been reclaimed. > > After this the directory that was created is now a 0 length file, NOT a > directory. > > If we do the exact same operations but just change to the directory > before rebooting then chkdsk appears to be happy after the reboot and we > still have a directory. > > Has anyone else seen something similar? > > Remove the obvious spam bait from the e-mail address to reply directly. > > Thanks, > Burt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: etherOutput Date: Wed, 17 May 2000 09:09:30 +0200 From: Leonid Rosenboim Organization: BitBand Technologies Ltd. http://www.bitband.com Message-ID: <392245AA.C0436A90@bitband.com> References: Reply-To: leonid@bitband.com Contrary to previous replies - DO NOT call this from an ISR, ever. Most drivers will take a mutex in their transmit routine to keep their transmit buffer queues consistent, some will use intLock(), but dont bet your hat on that. Gordon Manning wrote: > What exactly does the vxworks function etherOutput do? Can one call it from > an ISR. > > Thanks in advance > > Gordon Manning --------------------------- Newsgroups: comp.os.vxworks Subject: Re: need a loop within a makefile Date: Wed, 17 May 2000 09:35:36 +0200 From: "Markus MItterer" Organization: KPNQwest customer news service Message-ID: References: Thanks to everybody for responding, I could solve this problem. But now a new question to you makefile-gurus!!! Is it possible to check if a file or directory exists (without an error message) and due to the result do something? I need a rule to rebuild a list of projects. Some of the projects exists locally (they should be rebuild) and these where no directory exists should be taken from the network and after that, all of them should be linked together. Quite simple, I think (just joking). Greetings Markus - -------------------------- Markus Mitterer Software Development STK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PII and PIII Date: Wed, 17 May 2000 09:40:51 +0200 From: Gilbert Cabillic Organization: Irisa - Campus de Beaulieu - RENNES Message-ID: <39224D03.FA93C47E@irisa.fr> References: <3920F247.48885724@irisa.fr> Ray Sprangers wrote: > > Hi Gilbert, > > I tried the Pentium Pro BSP and it takes about one minute to display the > splash screen after it has loaded the VxWorks image. Do you have the > same problem? The same happens on the Pentium BSP. > Ray, you're right. The kernel takes a lot of time to intialise the internal structures, and especially the memory configuration of the target. I suppose that the intialisation time of vxworks depends on the amount of memory of the target. But it's just the initialisation time. For the rest, it runs at a good speed. The initialisation time is also a little bit longer when you use vxVmi, or vxMp. I also suppose that's due to the memory management. Have a good day, // Gilbert Cabillic // Scratchy Project leader -- // http://www.irisa.fr/solidor/work/scratchy.html // IRISA - France --------------------------- Newsgroups: comp.os.vxworks Subject: Re: need a loop within a makefile Date: Wed, 17 May 2000 10:54:11 +0200 From: Chris Varlese Organization: Ascom Switzerland, Bern, Switzerland Message-ID: <39225E33.2FA781DF@no.mail.net> References: If you have a Windows NT host you can use the DOS "if" command: all: if exists $(FILEPATH) ... Note you will have to supply a DOS-style path with "\" characters in it. You can automatically convert paths to DOS format using a subst command: all: if exists $(subst /,\,$(FILEPATH))) ... regards Chris Varlese --------------------------- Newsgroups: comp.os.vxworks Subject: image download problem Date: Wed, 17 May 2000 14:05:34 +0530 From: "Venkatesh JS" Organization: Motorola Message-ID: <8ftlgo$kiv$1@schbbs.mot.com> hi vxworkers, I have a image download problem on PPC750 target. my image size is around 2.5 MB, and target takes a long time to download, so long that i never get messages like "Starting at address 0x1000...WDB Ready" which mean that download is successful. During this period if i shutdown the TFTP server on the host, then above messages appear on hyper terminal. Is there any default size for vxworks image that can be downloaded on to target ?.. - - venkat --------------------------- Newsgroups: comp.os.vxworks Subject: TCP/IP port bindproblem on Vxwork 5.1.3 Date: Wed, 17 May 2000 17:18:48 +0100 From: Frank Liu Organization: Motorola Semiconductor Products Sector Message-ID: <3922C668.EDC45567@motorola.com> Reply-To: F.Liu@motorola.com on VxWorks 5.3.1, It seems vxWorks don't release a UDP port after I close the socket which was bound to the port. Because when I debug my program which calls TCP/IP bind() routine, I must change my port number if I want to run my program again on VxWorks, even if my last running program close the socket and exit normally. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 17 06:26:08 2000 From: Sanchez Eddie Date: Wed May 17 06:26:10 PDT 2000 Subject: RE: Calling global obj member functions This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. ------_=_NextPart_000_01BFC003.31BF4CDE Content-Type: text/plain; charset="iso-8859-1" I believe there is a WR application note about this. But in short, you need to use the "lkup test" command to find all the instances of the "test" function. the lookup will return a list of "test" instances and entry points. all you have to do is find the objG.test() and spawn from the associated entry point. Edward Sanchez Software Engineer esanchez@OnBoard-Software.com (210) 925-7698x3381 OnBoard Software, Inc. http://www.OnBoard-Software.com 404 Greig St. Bldg. 178 Kelly AFB, TX 78241 FAX: (210) 925-2144 ------_=_NextPart_000_01BFC003.31BF4CDE Content-Type: text/plain; name="100442.txt" Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename="100442.txt" TSR# 100442 DATE: Dec 19 1997 5:41PM .......................................................................= .............. =20 TITLE: A Tornado User's Guide to C++ Member Functions SCOPE DISCOVERED ON HOST: All ------------------- ------------------- ARCHITECTURE: All ------------------- ------------------- BSP: All ------------------- ------------------- PRODUCT / REVISION: Tornado 1.0.1 ------------------- ------------------- DESCRIPTION OF PROBLEM: From the perspective of the Tornado Tools,=20 C++ member functions differ from ordinary C functions in two important=20 respects.=20 (1) A C++ method must be bound to an object of its class (2) The symbolic representation of a C++ method is a "mangle" of its = name. These attributes require taking a little more care at the level of "what am I trying to do here?" (see below on taskSpawn and member functions) and occasionally necessitate special tricks to workaround limitations of the Tornado Tools. RESEARCH: We shall work through a simple example which I hope will cover most of the commonly asked "how to" questions. You should also read Chapter 10 of the vxWorks 5.3.1 programmers guide since I shall skim over or omit any material which is described there. First let's look at the code. There is one class, Cursor, whose = objects=20 have a "position" and can be moved left and right. This is not a very = clever class so there is no checking for boundaries or anything of that sort. There are three source files, Cursor.h, Cursor.cpp and testCursor.cpp. The first two contain the class declaration/definition respectively and the final one is our "application" which consists of a single static object declaration. // Cursor.h class Cursor { public: // Constructor Cursor (=20 int initialPosition ); ~Cursor (); void moveLeftBy ( int steps=20 ) ; void moveRightBy ( int steps ); =20 int getPosition (); =20 private: int position; }; // Cursor.cpp #include "vxWorks.h" #include "iostream.h" #include "Cursor.h" // Constructor Cursor::Cursor ( int initialPosition )=20 : position (initialPosition) { cout << "Constructing Cursor at " << (void *) this=20 << " with position " << position << endl; } Cursor::~Cursor () { cout << "Destructing Cursor at " << (void *) this << endl; } =20 void Cursor::moveLeftBy ( int steps=20 ) { position -=3D steps; cout << "Moving Cursor " << (void *) this << " left by "=20 << steps << " steps" << endl; } void Cursor::moveRightBy ( int steps=20 ) { position +=3D steps; cout << "Moving Cursor " << (void *) this << " right by "=20 << steps << " steps" << endl; } int Cursor::getPosition () { return position; } // testCursor.cpp #include "vxWorks.h" #include "Cursor.h" // The compiler generates a global function whose job it is to // construct myCursor with parameter 100. This function will be invoked = // at download time or by the user depending on the C++ static=20 // constructor strategy.=20 Cursor myCursor (100); // To build the "application" copy these files to target/config/ and type % make Cursor.out ADDED_C++FLAGS=3D-g % make testCursor.out ADDED_C++FLAGS=3D-g This -g flag includes debugging information. Now start a Wind Shell and type the following (shell responses are lines without a preceding ->) -> cplusXtorSet 0 value =3D 0 =3D 0x0 -> cplusStratShow =20 C++ Constructors/Destructors Strategy is MANUAL =20 value =3D 0 =3D 0x0 We are ready to download the objects. Make sure you download them in the correct order! Since we are interesting in debugging we'll download them from the Debugger. Start a new debugger session and type=20 (gdb) load Cursor.out (gdb) load testCursor.out How to step through the construction or destruction of a static object I'll begin by mentioning that in general you should be rather careful about "calling" constructors and destructors. I'm sure the ANSI C++ committee would be horrified at everything I have to say below. However I think you'll be okay so long as you ensure that you only ever construct an object once and that you only ever destruct an object once. (And that you don't do the latter before the former!) If you take a look at testCursor.cpp you'll see that it declares one global variable myCursor of type Cursor. The compiler allocates space for the object in the data section and generates a function that will call the constructor for the object with the given parameter(s) (in this case 100). First we need to find the function that calls the constructor. Note the distinction. The constructor itself is Cursor::Cursor and takes one argument. The function we're looking for is generated by=20 the compiler and takes no arguments. It explicitly calls Cursor::Cursor on the address of myCursor (generated at final link i.e download time)=20 with the correct parameters.=20 To find this function type the following at the Wind Shell -> lkup "myCursor" _myCursor 0x0026ca54 data (testCursor.out) global constructors keyed to myCursor 0x0026ca3c text = (testCursor.out) global destructors keyed to myCursor 0x0026ca1c text = (testCursor.out) value =3D 0 =3D 0x0 The first entry corresponds to the object myCursor. The second entry is the function we've been talking about. Now to reference this function we either need to use its mangled name which will be something awful or its explicit address which is slightly less awful. We'll do the latter. First lets set a breakpoint on the constructor (remember we're calling a wrapper around the = constructor). (gdb) list Cursor::Cursor (gdb) break Cursor::Cursor Now we'll run the construction function: (gdb) run 0x0026ca3c (note the explicit address! Obviously this value will be different for you). You should be able to step through the constructor. You should = eventually see the following printed to stdout: Constructing Cursor at 0x26ca54 with position 100 The process for calling destruction functions is completely analogous to the above. Just look for "global destructors keyed to ".=20 You will notice that the compiler may generate construction/ destruction functions for modules that don't have any static global objects. They will often have rather confusing names. For example if you do a -> lkup "Cursor" you will see the line global constructors keyed to Cursor::Cursor(int) 0x0026c8c8 text = (Cursor.out) This is not the function you want (note that it corresponds too module Cursor.out). Ignore it! Now there one thing to bear in mind about this method of calling=20 constructors: the compiler only generates one function per module to call all the constructors in that module. If you want to call constructors on a per object basis you'll need to use the method below. An alternative to the above "correct" approach is to explicitly "call" the "member functions" Cursor::Cursor (constructor) and Cursor::~Cursor (destructor). This kind of thing should only be done as a debugging aid. Never do it programmatically. The correct programmatical construct corresponding to what we do in the next section is what is called "placement new and=20 delete". See a good C++ book for details.=20 How to debug a member function ------------------------------ Since member function pointers cause a great deal of confusion I'll start with a few words about them. Consider the declaration void Cursor::(*pMember)(int); pMember is declared to be a pointer to a member function of Cursor that takes an int and returns no arguments. We could legitimately initialise it with pMember =3D Cursor::moveLeftBy; What kind of creature is this? The best answer is an example of how to use it. myCursor.(*pMember) (3); Notice that we CAN'T JUST CALL a member function. We need to bind it to an object. Contrary to a common misconception there is no such thing as an instance of a member function corresponding to a particular object. This is extremely important. It is why a naieve attempt to use taskSpawn to spawn a task that runs a member function, or to enter a member function into a table that expects FUNCPTR's will = invariably lead to the wrong results. To reiterate with a concrete example: there is no such thing as a "pointer to myCursor.moveLeftBy" that we can call and expect to have the same effect as the line myCursor.moveLeftBy (3); Ideally a well constructed C++ program should never need something like a "pointer to myCursor.moveLeftBy" but integrating your C++ program with C code that expects real function pointers may necessitate trying one of the following tricks/workarounds. (1) Create a global function wrapper that calls the member function with the correct object. This could be a truly global namespace function or it could be a static member function of a class. Remember that static member functions need not and cannot be bound to a particular instance of their class. A pointer to a static member function can be safely cast to a = FUNCPTR/VOIDFUNCPTR. An example of the global function wrapper would be void myCursorMoveLeftBy (int steps) { myCursor.moveLeftBy (steps); } myCursorMoveLeftBy is now a genuine function pointer that can be called from the shell, used as an argument to taskSpawn, debugged from Crosswind etc. (2) This is the one that really grabs the bull by the horns! As the may know, the "under the covers" implementation of member functions is as follows. A member function is simply an ordinary function which takes as its first parameter a pointer to the object (the "this" pointer). So the following (horrible) line (a) is equivalent to (nice) line (b) (a)=20 ((VOIDFUNCPTR) Cursor::moveLeftBy) (&myCursor, 3); (b) myCursor.moveLeftBy (3); This fact allows you to do all sorts of things both programmatically and from the Tornado tools. Since the basic idea should be clear from the above I just present the syntax for each example without further explanation: In all cases below you should see the following printed to stdout: Moving Cursor 0x26ca54 left by 3 steps (programmatically) taskSpawn ("memberTask", PRIORITY, OPTIONS, STACK_SIZE, (FUNCPTR) Cursor::moveLeftBy, (int) &myCursor, 3=20 0, 0, 0, 0, 0, 0, 0) (from the Wind Shell) The Wind Shell doesn't know much about C++ classes. You'll need to either use the debugger (which I would recommend) or reference the member function by address. For example -> lkup "Cursor" (notice that lkup "Cursor::moveLeftBy" won't work. This is an artifact of name mangling). This will include the entry Cursor::moveLeftBy(int) 0x0026c7c8 text (Cursor.out) So we can spawn moveLeftBy from the shell by -> sp 0x0026c7c8, &myCursor, 3 (of course you can also use taskSpawn if you like). Notice that the shell doesn't know or care about typecasting and that it will automatically append the extra "0" parameters. Of course we can also just plain call a member function from the shell. Unfortunately you can't just type -> myCursor.moveLeftBy (3) <---------- Wrong syntax! Instead you have to use the address -> (0x0026c7c8) (&myCursor, 3) (from the debugger) Thankfully the debugger understands most C++ names. The following = syntax works just fine. (gdb) run Cursor::moveLeftBy &myCursor 3 Similarly you can set a breakpoint on a member function (gdb) break Cursor::moveLeftBy Now there are a few kinds of member functions that run doesn't like. These include destructors (Cursor::~Cursor) and overloaded operators. To use these we need to get an explicit address. We haven't overloaded any operators in our example code so let's take a = hypothetical example. Suppose we had an Array class, with an overloaded indexing operator ([]) and an object myArray. Schematically we have // Sketch code only! class Array { ... public: int operator[] (int index); ... }; Array myArray (...); // To debug the operator[] we first need to find its address. From gdb we can do this using (gdb) info address Array::operator[] Symbol "__vc__5Arrayi" is a function at address 0x26c9dc. We can now run by address (gdb) run 0x26c9dc &myArray 5 (This says spawn a task to perform the act myArray[5] and attach the debugger. For those who have not overloaded the [] operator before=20 myArray[5] is a syntactically sweet way of saying myArray.operator[] (5). If none of this makes any sense and you never intend to overload any operators then don't worry about it!) SPR / PATCH: KEYWORDS: SIGN-OFF: GROUP: C++ =20 =20 ------_=_NextPart_000_01BFC003.31BF4CDE-- From vxwexplo-errs@csg.lbl.gov Wed May 17 08:17:55 2000 From: "Michael Lawnick" Date: Wed May 17 08:17:57 PDT 2000 Subject: Re: TCP/IP port bindproblem on Vxwork 5.1.3 > on VxWorks 5.3.1, > It seems vxWorks don't release a UDP port after I close the socket which > was bound to the port. Because when I debug my program which calls TCP/IP > bind() routine, I must change my port number if I want to run my program > again on VxWorks, even if my last running program close the socket and > exit normally. > Try setsockopt (sock, SOL_SOCKET, SO_REUSEADDR, &optval, sizeof (optval)); The value at optval is an integer (type int), either 1 (on) or 0 (off). Your port will be freed after a timeout period otherwise. HTH MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== From vxwexplo-errs@csg.lbl.gov Wed May 17 09:52:04 2000 From: Anthony Le Date: Wed May 17 09:52:07 PDT 2000 Subject: Speed up the booting time for Pentium III with Tornado Hi, Does anyone know how to improve the boot time for Pentium III BSP? It takes up to 2 minutes before it can start flashing the screen. Thanks, Anthony Le From vxwexplo-errs@csg.lbl.gov Thu May 18 04:01:56 2000 From: David Anderson Date: Thu May 18 04:01:58 PDT 2000 Subject: RE: Speed up the booting time for Pentium III with Tornado vxWorks, Tornado The delay may be due to RAM test or RAM initialisation. Does the delay vary with the amount of RAM fitted ? If this is prooven to be so then look carefully at the way the BSP loads up and clears RAM. The BSP I was using insisted on clearing to zero all RAM from BSS to top of memory...OK with 8MBytes but not OK with hundreds of Megabytes !. On 17 May 2000 18:12, the vxWorks Users Group Exploder [SMTP:vxwexplo@lbl.gov] wrote: > Submitted-by: Anthony Le > > Hi, > > Does anyone know how to improve the boot time for Pentium III BSP? > It takes up to 2 minutes before it can start flashing the screen. > > Thanks, > From vxwexplo-errs@csg.lbl.gov Thu May 18 04:03:15 2000 From: Vxworks Exploder Date: Thu May 18 04:03:17 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu May 18 04:03:12 PDT 2000 Subject: Re: image download problem Subject: Re: VT100 escape sequences under VxWorks 5.4 Subject: Re: VGA VIDEO MODES Subject: Re: From VxWorks to Linux for instrument controllers??? Subject: Re: image download problem Subject: Re: VxWorks versus Nucleus ATI Subject: Re: From VxWorks to Linux for instrument controllers??? Subject: UnZip for VxWorks Subject: SNMPD for VxWorks Subject: Problem connecting to VxSim from Tornado launcher Subject: RE: PII and PIII on VxWorks Subject: Re:PII and PIII and VxWorks Subject: Re:PII and PIII Subject: Re: PII and PIII Subject: Virtual Memory Exhausted Subject: Re: From VxWorks to Linux for instrument controllers??? Subject: Re: UnZip for VxWorks Subject: memory partition format Subject: Re: etherOutput Subject: problem reconfiguring vxWorks Subject: Re: TCP/IP port bindproblem on Vxwork 5.1.3 Subject: SLIP problem: IP-Header-Compression Subject: PCI interrupts on Pentium. Subject: Is CORBA compatible with VxWorks? Subject: ROMable VxWorks [objcopy: SREC->BIN] Subject: Re: From VxWorks to Linux for instrument controllers??? Subject: Re: VxWorks 5.4 and dosFs 2.0 Problems Subject: Re: WindView Subject: Re: Virtual Memory Exhausted Subject: http://www.seguritec.pt Subject: Re: Is CORBA compatible with VxWorks? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: image download problem Date: Wed, 17 May 2000 13:13:51 GMT From: "Peter Luscher" Organization: @Home Network Canada Message-ID: References: <8ftlgo$kiv$1@schbbs.mot.com> If you are downloading your image using the VxWorks bootrom, your downloaded image must not be larger than (RAM_HIGH_ADRS - RAM_LOW_ADRS). If it is, you must re-build your bootrom (after increasing the value of RAM_HIGH_ADRS in config.h AND in the Makefile), re-flash your bootrom and re-boot. P. "Venkatesh JS" wrote in message news:8ftlgo$kiv$1@schbbs.mot.com... > hi vxworkers, > > I have a image download problem on PPC750 target. my image size is around > 2.5 MB, and target takes a long time to download, so long that i never get > messages like "Starting at address 0x1000...WDB Ready" which mean that > download is successful. During this period if i shutdown the TFTP server on > the host, then above messages appear on hyper terminal. > > Is there any default size for vxworks image that can be downloaded on to > target ?.. > > - > venkat > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VT100 escape sequences under VxWorks 5.4 Date: Wed, 17 May 2000 09:22:31 -0400 From: Charles H. Chapman Organization: Dynacs Engineering Company Message-ID: <8fu6gf$9mv@niven.ksc.nasa.gov> References: <39216736.19613EDF@my-deja.com> Reply-To: chc@nasa2.ksc.nasa.gov JB wrote: > I am using VxWorks v5.4 on a PowerPC target and compiling FORTRAN > under the Green Hills toolset. The FORTRAN code has print statements to > a serial device (tyCo0) utilizing VT100 escape sequences for cursor > movement, etc. Using a Hypertem window (and an actual VT100 terminal), > the escape sequences are not executed and are interpreted as normal > text. The information in the VxWorks programmers guide is very minimal > on this subject. > > My questions are 1) Does VxWorks support VT100 escape sequences > printed to a serial tty device or 2) If not, what options would I have > available? I've used Vt100 escape sequences writing to the serial port (or the network connection) using a C language program under VxWorks and it works very well. It must be a problem with your FORTRAN compiler. Chuck --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VGA VIDEO MODES Date: Thu, 11 May 2000 13:51:27 +0100 From: "Dave Korn" Organization: (Posted via) U-NET Internet Ltd. Message-ID: <96yS4.645$U3.345938@newsr2.u-net.net> References: <391AB4C1.F53482B1@reading.ac.uk> James Pascoe wrote in message <391AB4C1.F53482B1@reading.ac.uk>... >........, so I wrote the following asm statement in Tornado: > >__asm__("movl $13h, eax\n > intl $10h" > ); > >When I execute this, I get the error: > Invalid TSS >and the program crashes. > >My assembler is so rusty, that I am probably making a 'kick me' with the above. >What does anybody think? I'm not an x86 coder, but IIRC the order of operands in x86 assembler is dest, source... so that first instruction that's meant to put 0x13 into AX looks to me like it might be storing the value of AX to location 0x13; then you call the int 10 with whatever random value happened to be in AX before........ OTOH I could be completely wrong, and gcc is notable for it's non-standard assembler syntaxes. But I reckon it's one thing that's worth a double-check. hth, DaveK - -- I wouldn't recommend sex, drugs or insanity for everyone, but they've always worked for me. -- Hunter S. Thompson --------------------------- Newsgroups: comp.os.vxworks Subject: Re: From VxWorks to Linux for instrument controllers??? Date: Wed, 17 May 2000 09:59:13 -0400 From: Erick Gonzalez Organization: Marconi Message-ID: <3922A5B1.DE63D555@marconi.com> References: <8fhm41$l32$1@web1.cup.hp.com> <391C6921.C25DDA19@fore.com> <3921C7E1.38BF48A3@intelligraphics.com> At the risk of starting a religious war... Sure you can try to kludge your way around making a OS like Linux an RTOS (for example the RTLinux option is actually another OS, running Linux as the highest priority task.... hmmmm... I can think of a couple of nightmare scenarios right off my head... but that's another issue). Anyway. The point, is that an OS doesn't have to be everything to everybody.... Linux kicks *ss on the desktop, but why try to embed it now in every device with a microprocessor in it?... (ala Windows CE folks?... yikes!).... anyway.. to each his/her own. If you want to spend time and resources, just to put a tag on your toaster that says "Linux inside", for market hype..... good luck.... Erick Michael Morrison wrote: > Erick Gonzalez wrote: > > > Linux is NOT a RTOS. It may be a great OS on the desktop, or in server > > applications, but I don't think that you can get away with using it for a real > > time system, unless your application doesn't really require reliable and speedy > > preemptive task scheduling (among other goodies)... but then that would make it > > a non RT app wouldn't it? > > > > www.rtlinux.com > www.zentropix.com > www.lynx.com > > www.google.com search for "real time linux" > > Nuf said. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: image download problem Date: Wed, 17 May 2000 21:07:40 +0530 From: "Venkatesh JS" Organization: Motorola Message-ID: <8fue85$9e3$1@schbbs.mot.com> References: <8ftlgo$kiv$1@schbbs.mot.com> hi, In my case the difference between RAM_HIGH_ADDR and RAM_LOW_ADDR is 16 MB which is well beyond the image size of 2.5 MB. venkat Peter Luscher wrote in message ... >If you are downloading your image using the VxWorks bootrom, your downloaded >image must not be larger than (RAM_HIGH_ADRS - RAM_LOW_ADRS). If it is, you >must re-build your bootrom (after increasing the value of RAM_HIGH_ADRS in >config.h AND in the Makefile), re-flash your bootrom and re-boot. > >P. > >"Venkatesh JS" wrote in message >news:8ftlgo$kiv$1@schbbs.mot.com... >> hi vxworkers, >> >> I have a image download problem on PPC750 target. my image size is around >> 2.5 MB, and target takes a long time to download, so long that i never get >> messages like "Starting at address 0x1000...WDB Ready" which mean that >> download is successful. During this period if i shutdown the TFTP server >on >> the host, then above messages appear on hyper terminal. >> >> Is there any default size for vxworks image that can be downloaded on to >> target ?.. >> >> - >> venkat >> >> > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks versus Nucleus ATI Date: Wed, 17 May 2000 17:15:24 +0100 From: "Mark Hambleton" Organization: (Posted via) Netcom Internet Ltd. Message-ID: <8fugjt$i4l$1@taliesin2.netcom.net.uk> References: <8fes53$go0$1@nnrp1.deja.com> <391f1aa8$0$216@nntp1.ba.best.com> <3921C6D4.6160EE07@intelligraphics.com> I have recently started using Nucleus and ported from vxWorks. Having the sources seems to greatly reduce the time for the port as you can see from the sources why something hasn't worked and don't have to go through the often very lengthy process of a WRS support request. We had some issues with the Nucleus ISR mechanism also but as it came with sources we were able to modify the part that we had problems with. With reference to the vxWorks IO story, I liked it some much that I replaced it with my own, probably because I write networking software and the WRS way was too slow....no idea what the Nucleus OS gives as I use the same for both OS's, we also wrote our own version of the shell for Nucleus which seems to give us similar levels of functionality to the WRS way. I guess the biggest selling point for us had to be the lack of need for support, sources save time and money for us in the long run...... Mark - -- Michael Morrison wrote in message news:3921C6D4.6160EE07@intelligraphics.com... > Add to the list below that Nucleus comes with source, wrs does not (Unless > you PAY) > The only thing I didn't like about Nucleus was that you can't give a > semaphore or write a queue from > an ISR. You have to make what they call a "High Level ISR" which you > activate from the ISR. > Kind of a pain. > > Chris Wein wrote: > > > In article <8fes53$go0$1@nnrp1.deja.com>, wrote: > > > > > > > > >Hi All: > > > > > >We're starting on a new project that will use either > > >a Tensilica core or a MIPS core. > > > > > >With respect to the former, only Wind River and ATI > > >support it with an RTOS. > > > > on the kernel eval > > > > We recently selected VxWorks over Nucleus. Here's a quick summary > > > > 1) Nucleus is royalty free whereas VxWorks is not. > > 2) WRS tool seats are more expensive than ATI > > 3) Nucleus' kernel size is smaller than VxWorks. For the config we > > wanted, Nuke was around 70K vs 150K for VxWorks (SPARC v8) > > 4) Both kernels are fast and efficient. > > 5) Nucleus has a slightly more rich task communication mechanisms such > > as events. > > 6) ATI licenses source with all distributions. Costs more $$$ for WRS. > > > > Based on these points, you might wonder why we chose VxWorks. However, > > if you are evaluating just the kernel then both are fine. If you > > are evaluating an RTOS (which includes kernel + I/O + other components) > > then VxWorks is the clear winner... > > > > 1) VxWorks supports debug over ethernet > > 2) VxWorks has a fantastic I/O story. FTP client support, filesystems, > > per-task stdin/stdout, I/O redirection right out of the box > > 3) VxWorks has a well thought out set of "hooks" that allow apps to > > bind to interesting events like task switch. > > 4) VxWorks has superior 3rd party support for more esoteric things like > > 1394 stacks, USB etc. > > > > If you need a small kernel in a heavily embedded system then go with ATI > > but if you are scaling up and need more sophisticated I/O then you will > > be very pleased with wrs. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: From VxWorks to Linux for instrument controllers??? Date: Wed, 17 May 2000 09:29:24 -0700 From: Michael Morrison Organization: Intelligraphics, Inc. Message-ID: <3922C8E4.4A5F1021@intelligraphics.com> References: <8fhm41$l32$1@web1.cup.hp.com> <391C6921.C25DDA19@fore.com> <3921C7E1.38BF48A3@intelligraphics.com> <3922A5B1.DE63D555@marconi.com> Reply-To: mikem@intelligraphics.com I'm curious. Have you actually tried it out for anything or are you just dismissing it out of hand? Erick Gonzalez wrote: > At the risk of starting a religious war... > > Sure you can try to kludge your way around making a OS like Linux an RTOS (for > example the RTLinux option is actually another OS, running Linux as the highest > priority task.... hmmmm... I can think of a couple of nightmare scenarios right off > my head... but that's another issue). Go ahead and name em. Say I'm considering differt RTOS solutions, but don't have alot of money, RTLinux *might* be a good alternative to VxWorks, depending on the problem I'm trying to solve. I've experimented with RTLinux and found it to work as advertised. I didn't run into any nightmares. So based on my experiments, I decide to use RTLinux for my project. What nightmares might I run into? > Anyway. The point, is that an OS doesn't have > to be everything to everybody.... No OS is. You have to look at the problem you're trying to solve and find the best fit. > Linux kicks *ss on the desktop, but why try to > embed it now in every device with a microprocessor in it?... Why not? If it's suitable to the application, you can't beat the price. > (ala Windows CE > folks?... yikes!).... Gag > anyway.. to each his/her own. If you want to spend time and > resources, just to put a tag on your toaster that says "Linux inside", for market > hype..... good luck.... > > Erick > --------------------------- Newsgroups: comp.os.vxworks Subject: UnZip for VxWorks Date: Wed, 17 May 2000 18:25:21 +0200 From: "mark" Organization: vocaltec.com Message-ID: <8fuh5i$6$1@news.netvision.net.il> Hi all, Does any one know when I can find source code that opens ZIP archives for VxWorks ? Thanks ... --------------------------- Newsgroups: comp.os.vxworks Subject: SNMPD for VxWorks Date: Wed, 17 May 2000 10:32:00 -0700 From: "Robert E. Laughlin" Organization: San Diego State University Message-ID: ello; I have inherited part of a project that uses VxWorks version 5.2 and includes WindNet SNMP 1.0 Beta-2. I need to change a couple of the settings for the SNMPD on the system. The documentation that we got does not include "Part 2" of the "WindNet SNMP VxWorks Optional Product Supplement". I do not know if we just did not get it or if Part 2 got lost, but that appears to be where all of the details of how to configure SNMP for different "community" strings. If any one knows where I can get a copy of Part 2 or can tell me how to change the community strings, I would sure appreciate it. Thanks Bob Robert E. Laughlin SPAWARSYSCEN SD D4122 An interesting thing happens when you hit a security-illiterate person with a tool that makes them realize how often they get probed and examined by hackers: the get furious. These opinions are mine. I do not speak for SPAWARSYSCEN SD. email bel@spawar.navy.mil Public Key ID = AE8294B9 Key fingerprint = 8747 FC71 49C7 59D0 74F8 CC2B E1D8 8AFB AE82 94B9 --------------------------- Newsgroups: comp.os.vxworks Subject: Problem connecting to VxSim from Tornado launcher Date: 17 May 2000 17:28:09 GMT From: trinity@bnr.ca (Seshu Madhavapeddy) Organization: Bell-Northern Research, Richardson, TX Message-ID: <8fukr9$79q$1@crchh14.us.nortel.com> Hi, I am using Tornado 2.0 on HPUX10.20. I built a custom simulator and to test it I did this (as the manual suggested me) - - Updated /etc/ppl.remotes with entries from my $WIND_BASE/target/config/hpux ppl.remotes.hpux10 - - Created a pseudo terminal device /dev/ptym/ptyr1 using the command ppl -o -t /dev/ptym/ptyr1 192.168.1.1 - - After this, I am supposed to enable IP packet forwarding by running 'ndd -set /dev/ip ip_forwarding 1', but I dont know where is ndd so I couldn't do this. But I think this is harmless because the manual says I need this if I want to ping two vxSims - - Then netstat -i shows Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Coll ni0 1006 192 192.168.2.1 0 0 0 0 0 So I think that the pty device is created properly. However, when I run the custom simulator ("vxworks_custom -p 1") it says "sysHwInit: SLIP: unable to open /dev/ptym/ptyr1". This comes from code which calls 'u_open (ptyName, L_RDWR | L_CNOTTY | L_NDELAY, 0644)' where ptyName is /dev/ptym/ptyr1. I added code to print 'errno' when this happens and it prints errno = 16 Then I looked at /usr/include/sys/errno.h in my HP and it says that this corresponds to EBUSY - Mounted device busy. Does it mean that this pty device is busy? If so, how do I found out which pty devices are busy and which are not so I can use the unused ones. If I do ls -l /dev/ptym/ptyr1 I see: "crw-rw-rw- 2 bin tty 16 0x000021 Mar 9 10:34 /dev/ptym/ptyr1" Thanks Mohan. --------------------------- Newsgroups: comp.os.vxworks Subject: RE: PII and PIII on VxWorks Date: Wed, 17 May 2000 10:39:13 -0700 From: Ray Sprangers Organization: BCnet Message-ID: References: <3920F247.48885724@irisa.fr> <39224D03.FA93C47E@irisa.fr> Hi Anthony, There is s function is Vxworks usrMmuInit() which reinitialized the memory mapping unit. This function is what takes so long at boot time. We also call this function is our code and through the use of the debugger we discovered that this function takes about a minute to get executed. We have heard that this function does some tasks which are expensive on a PII and a PIII. Hence boot is fast on a PI, but very slow on a PII and PIII. Same results when using the Pentium or Pentiumpro BSP. Cheers, Ray - -----Original Message----- From: Anthony Le [mailto:ALe@Connex.com] Sent: Wednesday, May 17, 2000 9:42 AM To: 'Ray_Sprangers@spectrumsignal.com'; 'Gilbert.Cabillic@irisa.fr' Subject: Re:PII and PIII and VxWorks Newsgroups: comp.os.vxworks Subject: Re:PII and PIII Date: Tue, 16 May 2000 11:46:49 -0700 From: Ray Sprangers Organization: BCnet Message-ID: References: <3920F247.48885724@irisa.fr> Hi Gilbert, I tried the Pentium Pro BSP and it takes about one minute to display the splash screen after it has loaded the VxWorks image. Do you have the same problem? The same happens on the Pentium BSP. Ray - - -----Original Message----- From: Gilbert Cabillic [mailto:Gilbert.Cabillic@irisa.fr] Posted At: Tuesday, May 16, 2000 12:01 AM Posted To: vxworks Conversation: PII and PIII Subject: Re: PII and PIII Hi All, I have Tornado 1 and Pentium II and III working fine. But the booting time is tooo... long. Do you know why? Regards, Anthony Le HI Ray. We use vxWorks on PIIs and PIIIs using a PentiumPro Bsp (not Pentium Bsp). It Runs well. Gilbert. // Gilbert Cabillic // Scratchy Project leader -- // http://www.irisa.fr/solidor/work/scratchy.html // IRISA - France Ray Sprangers wrote: > > Hi VxWorkers, > > Has anyone got VxWorks to work on a Pentium II or a Pentium III? We > are using the Pentium BSP and it is very slow on a Pentium III and > doesn't work at all on a Pentium II. Is there a special BSP that is > needed for a PII and PIII? > > Ray --------------------------- Newsgroups: comp.sys.hp.hpux,comp.os.vxworks Subject: Virtual Memory Exhausted Date: Wed, 17 May 2000 13:07:26 -0500 From: "Robert F.K. Martin" Organization: Lockheed Martin - NE&SS Message-ID: <3922DFDE.D7A7200B@lmco.com> I have a HP-UX 10.20 system running on a 744 with Tornado 1.0.1 and vxWorks 5.3.1. I am having problems compiling an rtx object when I include the jiskan24 font. Specifically, in the /usr/wind/target/config/rtx/fonts directory % make jiskan24 The output is: /usr/wind/host/parisc-hpux10/bin/rtxbifont /usr/wind/host/rtx/fonts/misc/jiskan24.pcf > jiskan24.c ccppc -B/usr/wind/host/parisc-hpux10/lib/gcc-lib/ -c -mstrict-align - -nostdinc -O0 -fvolatile -fno-builtin -fno-for-scope -DCPU=PPC604 - -DVXWORKS -DVX5x -I. -I/usr/wind/target/h/rtx -I/usr/wind/target/h - -I/usr/wind/target/src jiskan24.c jiskan24.c:44998: virtual memory exhausted make: *** [jiskan24.o] Error 1 In my /stand/system file, I've added: maxswapchunks 8192 swchunk 4096 vmstat -n gives me: VM memory page faults avm free re at pi po fr de sr in sy cs 4099 87195 7 4 3 0 0 0 2 195 872 58 CPU cpu procs us sy id r b w 6 1 93 1 92 0 During the compile, the avm entry continues to grow until about 16K and that appears to coincide with the compile process running out of virtual memory. It does not appear as though any pages get swapped in or out during the compile. I'm out of ideas. Has anyone seen anything similar or solved a similar problem. Thanks Rob Martin r.f.martin@lmco.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: From VxWorks to Linux for instrument controllers??? Date: Wed, 17 May 2000 14:44:52 -0400 From: Erick Gonzalez Organization: Marconi Message-ID: <3922E8A4.B7AD8550@marconi.com> References: <8fhm41$l32$1@web1.cup.hp.com> <391C6921.C25DDA19@fore.com> <3921C7E1.38BF48A3@intelligraphics.com> <3922A5B1.DE63D555@marconi.com> <3922C8E4.4A5F1021@intelligraphics.com> I did work on a project with Linux as the OS for a realtime protocol analyzer years ago (back when downloading it was the only way of getting it). I must admit it wasn't any of the realtime variants now available that you have mentioned. We just tweaked things around to make it work...and it did, but sub-optimally. If money is a concern, then well, you have to work with what you've got... but perhaps taking a look at an RTOS like RTEMS (http://www.rtems.com) which is distributed under the GNU license, could be an interesting alternative... BTW: Let's continue this off the newsgroup... let's not burden the folks around here with opinion postings... Erick > I'm curious. Have you actually tried it out for anything or are you just dismissing > it out of hand? > > Erick Gonzalez wrote: > > > At the risk of starting a religious war... > > > > Sure you can try to kludge your way around making a OS like Linux an RTOS (for > > example the RTLinux option is actually another OS, running Linux as the highest > > priority task.... hmmmm... I can think of a couple of nightmare scenarios right off > > my head... but that's another issue). > > Go ahead and name em. Say I'm considering differt RTOS solutions, but don't have > alot of money, > RTLinux *might* be a good alternative to VxWorks, depending on the problem I'm trying > to solve. > I've experimented with RTLinux and found it to work as advertised. I didn't run into > any nightmares. > So based on my experiments, I decide to use RTLinux for my project. What nightmares > might I > run into? > > > Anyway. The point, is that an OS doesn't have > > to be everything to everybody.... > > No OS is. You have to look at the problem you're trying to solve and find the best > fit. > > > Linux kicks *ss on the desktop, but why try to > > embed it now in every device with a microprocessor in it?... > > Why not? If it's suitable to the application, you can't beat the price. > > > (ala Windows CE > > folks?... yikes!).... > > Gag > > > anyway.. to each his/her own. If you want to spend time and > > resources, just to put a tag on your toaster that says "Linux inside", for market > > hype..... good luck.... > > > > Erick > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UnZip for VxWorks Date: Wed, 17 May 2000 21:32:07 GMT From: "Lou Kale" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <8fuh5i$6$1@news.netvision.net.il> I've used GNU gzip. mark wrote in message news:8fuh5i$6$1@news.netvision.net.il... > Hi all, > > Does any one know when I can find source code that opens ZIP archives for > VxWorks ? > > Thanks ... > > --------------------------- Newsgroups: comp.os.vxworks Subject: memory partition format Date: Thu, 18 May 2000 00:32:38 GMT From: allanh@gate1.kallisti.com (Allan N. Hessenflow) Organization: None Message-ID: I'm trying to write a tool to let me check a whole partition for consistency in an attempt to catch someone writing somewhere they shouldn't. I've found the mem_part and blockHdr definitions, but I'm still trying to figure out how to walk the entire heap. Given a pointer to a block within the heap I can walk part of it, but not all. Has anyone here found or deduced the information needed to do this? allan - -- Allan N. Hessenflow allanh@kallisti.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: etherOutput Date: Thu, 18 May 2000 01:30:08 GMT From: "M Campbell" Organization: @Home Network Message-ID: References: We're using the etherOutput call to talk to a legacy system that has (and can only have) raw ethernet I/O (i.e. sans TCP). Would not reccomend using call inside interrupt routine - we use message queues for stuff to go out, then have a seperate task to do the etherOutput stuff. "Gordon Manning" wrote in message news:OBiU4.10232$Lp1.253338@typhoon2.ba-dsg.net... > What exactly does the vxworks function etherOutput do? Can one call it from > an ISR. > > Thanks in advance > > Gordon Manning > > > --------------------------- Newsgroups: comp.os.vxworks Subject: problem reconfiguring vxWorks Date: Wed, 17 May 2000 20:03:34 GMT From: chrisanc1967@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8futu8$ufn$1@nnrp1.deja.com> I'm working with a 3rd party BSP. When I try to reconfigure the vxWorks image so that I can add some funcitonality I get an error. The build output says that there is an undefined reference to 'm_freem' in function 'dcOutput'. I can't see anything wrong with the header files that contain the dcOutput function. Can anyone help me with this? Thanks. Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TCP/IP port bindproblem on Vxwork 5.1.3 Date: Wed, 17 May 2000 07:02:43 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3922A683.C7196A78@delphi.com> References: <3922C668.EDC45567@motorola.com> You probably need to use setsockopt to set SO_REUSE_ADDR because it will take a while for the system to take down the socket. Speaking only for myself, Joe Durusau Frank Liu wrote: > > on VxWorks 5.3.1, > It seems vxWorks don't release a UDP port after I close the socket which > was bound to the port. Because when I debug my program which calls > TCP/IP bind() routine, I must change my port number if I want to run my > program again on VxWorks, even if my last running program close the > socket and exit normally. --------------------------- Newsgroups: comp.os.vxworks Subject: SLIP problem: IP-Header-Compression Date: Thu, 18 May 2000 09:52:06 +0200 From: "M Kur" Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: <8g07e2$nd2$1@pollux.ip-plus.net> Hi, in our application the CSLIP connection doesn't work or is extremely slow. SLIP without IP-header compression works without any problem but we would like to use CSLIP. What's the problem with CSLIP? Has anyone seen anything similar or solved a similar problem. Our Current Version: Tornado II / VxWorks 5.4 (for CPU 32) Our configuration: #define SLIP_BAUDRATE 57600 #define SLIP_MTU 1006 #define CSLIP_ENABLE 0 #define CSLIP_ALLOW 1 Thanks, M. Kurzmeyer --------------------------- Newsgroups: comp.os.vxworks Subject: PCI interrupts on Pentium. Date: Thu, 18 May 2000 10:06:13 +0200 From: Hank van Bekkem Organization: Oce Technologies B.V. Message-ID: <3923A475.2F8051E9@oce.nl> Hi, I can't get PCI interrupts to work on a Pentium platform using VxWorks 5.4 I implemented them just like the example in the BSP itself for SCSI, but no interrupts get through. I know for certain my PCI board signals interrupts, because it works fine in a PowerPC MTX604 board with VxWorks 5.4. I *am* able to read/write from/to my PCI board from the Pentium platform. Can anyone help ? Perhaps with some example driver code ? Thanks in advance, Hank van Bekkem --------------------------- Newsgroups: comp.os.vxworks Subject: Is CORBA compatible with VxWorks? Date: Thu, 18 May 2000 08:07:04 GMT From: "Earthlink" Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: Hi All, I am new at Tornado systems and VxWorks. I have to develop a client and server application that will run under VxWorks. I am only allowed to use POSIX compatible subset of VxWorks API. Can I use CORBA to implement communication between the client and server? Or I have to go down to Socket level? or further down? Please help. Ron S. Ahn OR Concepts Applied. --------------------------- Newsgroups: comp.os.vxworks Subject: ROMable VxWorks [objcopy: SREC->BIN] Date: 18 May 2000 08:07:27 GMT From: Geir Frode Raanes Organization: UNINETT news service Message-ID: <8g08bv$lhs$1@snipp.uninett.no> I am experiencing some problems in creating a (compressed) VxWorks kernel resident in ROM. First off - the kernels I make with Tornado 2 is obviousely ELF32 object files. But how do I make (powerpc-wrs-vxworks-)objectdump tell whether or not this object file still contains unresolved relocation information? I tried "-r" on both the compressed and uncomressed kernels. Even the download version gives the same result: powerpc-wrs-vxworks-objdump -r vxWorks.rom powerpc-wrs-vxworks-objdump -r vxWorks.romCompressed vxWorks.rom: file format elf32-powerpc The same "-r" argument on userspace applications produce long lists of relocation information. Am I then safe to assume that the VxWorks kernel is statically linked? Ie. that it does _not_ require any ELF aware program loader? That all the loader has to do is unpack the kernel into the predefined location RAM_LOW_ADDRESS, then set up this BOOT_LINE information exchange data block and finally jump to the kernel entrypoint _sysInit? Now, on the MBX860 board there is two sets of flash. Booting EPPCBug from on set we can download a binary image from net and flash it into the other set. This works fine for existing VxW boot ROMs. Tornado 2 produces two ROM kernel files, first the ELF32 file mentioned above and next a SREC hex version of the above. But EPPCBug requires a binary image to flash, so how do I use objectcopy to produce a binary kernel image from one of these files? - -- ****************************************************** Never ever underestimate the power of human stupidity. -Robert Anson Heinlein GeirFRS@invalid.and.so.forth ****************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: From VxWorks to Linux for instrument controllers??? Date: 17 May 2000 21:55:25 GMT From: Geir Frode Raanes Organization: UNINETT news service Message-ID: <8fv4gd$9et$1@snipp.uninett.no> References: <8fhm41$l32$1@web1.cup.hp.com> Sender: Geir Frode Raanes Gregor wrote: : I heard people talking about Linux as an "RTOS" for embedded devices. I am : using VxWorks 5.4 on a PPC860/40 with 32 Mbytes of RAM. All hard real-time : tasks are done on a dedicated SHARC DSP with its own 32 Mbytes of RAM. Is : there a good reason not to consider Linux? Is it much worse than VxWorks for : my kind of application? Or is it something I should seriously evaluate? RT-Linux is a cludge. It runs Linux ontop of a RT kernel. This gives you realtime like interrupt latencies but does not give you any RT OS services. The L4 kernel with NetBSD multiserver ontop might OTOH prove interesting. But if all your real time needs are serviced by a dedicated DSP, then why bother with a RTOS on the PowerPC? Unlike some other OS, Linux has not grown exponetionally in size and workload. Moore's law has still done it's work on the HW in such a way that the modern MPC860 can easily fuel embedded (non RT-) Linux. Actually, GPOS like Linux will by design utilise the HW in the most efficient way possible as opposed to the RTOS requirement of meeting deadlines at the expense of HW efficiency. There is even a maillist dedicated embedded PPC/Linux. And yes, it do work like a charm and yes, it really is Linux, The Linux. (linuxppc-embedded@lists.linuxppc.org) (http://www.mvista.com - MontaVista HardHat Linux.) - -- ****************************************************** Never ever underestimate the power of human stupidity. -Robert Anson Heinlein GeirFRS@invalid.and.so.forth ****************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks 5.4 and dosFs 2.0 Problems Date: Sun, 14 May 2000 17:45:15 -0600 From: Hosehead Jones 2nd-1/2 Organization: United Garage Mechanics and Fruitbats, Ltd. Message-ID: <86euhsk5l0pc1qhb77mp3m8ar09beom6oq@4ax.com> References: <391d46ab.994293534@news.netis.com> epooleNO@UCEnetis.com (RKT Technologies, Inc.) took time out of his/her busy schedule to pen: > > >I then enabled the following kernel elements: > > operating system components -> IO system components -> > DOS filesystem > You may not have installed dosFs 2.0 properly. I believe that dosFs 1.0 only had a single kernel element to include, but 2.0 has an entire folder's worth of items (for things such as caching, VFAT support, and a host of other ones I can't recall offhand). When you install the dosFs 2.0 disk, there should be two elements: "DosFs 2.0 for " (PPC in your case), and "DosFs 2.0 Source Code". When we initially received the CD from WRS, we got a CD key that just unlocked the source code, but not the kernel additions. (Actually, the first key we got was for the wrong CD, but that's another post.) If that's your problem, you need to contact WRS sales (again) and ask for the correct key. I don't know if this will solve all your problems, but it's a start. - --------------------- markm@ctilidar.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WindView Date: Mon, 15 May 2000 09:24:10 +0200 From: "Achim Zimmer" Organization: Colt Telecom GmbH Message-ID: <8fo8j0$nk8$1@crusher.de.colt.net> References: <8fgp4o$sjk15@overload.lbl.gov> <391C2548.7A637847@nowhere.com> Normally with T2 you run vxWorks 5.4 and so you have to include the WINDVIEW components in the workspace. If you are really trying 5.3.1 you have to add same defines to config.h Paddy schrieb in im Newsbeitrag: VE2T4.10167$Y4.50667@typhoon2.san.rr.com... > In Tornado2 you need to include the Windview components (under Development > Components). Note the T2 default for timestamping is sequential timestamping > which will not give you any absolute time information, you should deselect > the default (sequential timestamping) and include the "system defined > timestamping" component to get real time values. Most BSP's will support it. > > Patrick > "Charlie Grames" wrote in message > news:391C2548.7A637847@nowhere.com... > > Andy, > > > > Did you configure WindView in your kernel? You need to add > INCLUDE_WINDVIEW to > > the included facilities section in configAll.h. > > -- > > Charlie Grames > > The Boeing Company > > Charles.R.Grames @ boeing.com > > > > > > amass wrote: > > > > > > This is a multi-part message in MIME format. > > > > > > ------=_NextPart_000_001B_01BFBC08.2F562C00 > > > Content-Type: text/plain; > > > charset="iso-8859-1" > > > Content-Transfer-Encoding: quoted-printable > > > > > > Hi > > > > > > We have an installation of Tornado 2/VxWorks 5.3.1 on an NT host = > > > with a Radstone PPC1a target. We have recently attempted an installation > = > > > of WindView, but problems have been encountered. The product was = > > > installed and configured what appears to be corectly, but whenever we = > > > attemp to launch the tool, we get the following message: > > > > > > "target name@host name does not support WindView" > > > > > > Can anybody offer any suggestions as to the casue of such a problem? > > > > > > Thanks > > > > > > Andy Atkinson > > > Software Engineer > > > DERA Bincleaves > > > Weymouth > > > Dorset, DT4 8UR > > > UK > > > > > > ------=_NextPart_000_001B_01BFBC08.2F562C00 > > > Content-Type: text/html; > > > charset="iso-8859-1" > > > Content-Transfer-Encoding: quoted-printable > > > > > > > > > > > > > > http-equiv=3DContent-Type> > > > > > > > > > > > > > > >
Hi
> > >
 
> > >
    We have an installation of = > > > Tornado=20 > > > 2/VxWorks 5.3.1 on an NT host with a Radstone PPC1a target. We have = > > > recently=20 > > > attempted an installation of WindView, but problems have been = > > > encountered. The=20 > > > product was installed and configured what appears to be corectly, but = > > > whenever=20 > > > we attemp to launch the tool, we get the following message:
> > >
 
> > >
   "target name@host name does not = > > > support=20 > > > WindView"
> > >
 
> > >
Can anybody offer any suggestions as to the casue of > = > > > such a=20 > > > problem?
> > >
 
> > >
Thanks
> > >
 
> > >
 
> > >
Andy Atkinson
> > >
Software Engineer
> > >
DERA Bincleaves
> > >
Weymouth
> > >
Dorset, DT4 8UR
> > >
UK
> > > > > > ------=_NextPart_000_001B_01BFBC08.2F562C00-- > > --------------------------- Newsgroups: comp.sys.hp.hpux,comp.os.vxworks Subject: Re: Virtual Memory Exhausted Date: Thu, 18 May 2000 11:43:02 +0100 From: "Stephen Walsh" Message-ID: <3923c51f$1@pull.gecm.com> References: <3922DFDE.D7A7200B@lmco.com> Greetings, This probably isn't much help but I saw a similar problem under Win95 when trying to compile code on a different machines hard drive which was mapped using microsoft networking "memory exhuasted error". This did not occur using the same method under NT. As most of our machines are NT based I didn't persue it very far and just copied the code to the 95's local drive for compilation and copied the compiled object back afterwards. Steve ******************************************* Stephen A Walsh Senior Software Engineer BAE Systems ******************************************* Opinions are usually mine and mine alone ******************************************* Robert F.K. Martin wrote in message <3922DFDE.D7A7200B@lmco.com>... >I have a HP-UX 10.20 system running on a 744 with Tornado 1.0.1 and >vxWorks 5.3.1. I am having problems compiling an rtx object when I >include the jiskan24 font. Specifically, in the >/usr/wind/target/config/rtx/fonts directory > >% make jiskan24 >The output is: >/usr/wind/host/parisc-hpux10/bin/rtxbifont >/usr/wind/host/rtx/fonts/misc/jiskan24.pcf > jiskan24.c >ccppc -B/usr/wind/host/parisc-hpux10/lib/gcc-lib/ -c -mstrict-align >-nostdinc -O0 -fvolatile -fno-builtin -fno-for-scope -DCPU=PPC604 >-DVXWORKS -DVX5x -I. -I/usr/wind/target/h/rtx -I/usr/wind/target/h >-I/usr/wind/target/src jiskan24.c >jiskan24.c:44998: virtual memory exhausted >make: *** [jiskan24.o] Error 1 > >In my /stand/system file, I've added: >maxswapchunks 8192 >swchunk 4096 > >vmstat -n gives me: >VM > memory page faults > avm free re at pi po fr de sr in sy >cs > 4099 87195 7 4 3 0 0 0 2 195 872 >58 >CPU > cpu procs > us sy id r b w > 6 1 93 1 92 0 > >During the compile, the avm entry continues to grow until about 16K and >that appears to coincide with the compile process running out of virtual >memory. It does not appear as though any pages get swapped in or out >during the compile. I'm out of ideas. Has anyone seen anything similar >or solved a similar problem. > >Thanks >Rob Martin >r.f.martin@lmco.com > --------------------------- Newsgroups: comp.os.vxworks Subject: http://www.seguritec.pt Date: 18 May 2000 10:14:35 GMT From: Seguritec Organization: Teleweb, S.A. - Grupo Finantel Message-ID: <8g0fqb$15r$3971@srvlis16.teleweb.pt> http://www.seguritec.pt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is CORBA compatible with VxWorks? Date: Thu, 18 May 2000 10:51:53 GMT From: martins2@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8g0i05$ndl$1@nnrp1.deja.com> References: Ron, > I am only allowed to use POSIX compatible subset of VxWorks API. > Can I use CORBA to implement communication between the client and > Wind River has a description of a CORBA stack for VxWorks on their web pages. You might want to check that out (www.wrs.com). bye Martin Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri May 19 04:03:17 2000 From: Vxworks Exploder Date: Fri May 19 04:03:19 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri May 19 04:03:13 PDT 2000 Subject: DosFs 2.0/VxWorks 5.3 and FTP Subject: Any good books on VxWorks?? Subject: Are there any functions which can compute the CPU load Subject: VxWorks employment Subject: Re: UGL && SimNT Subject: Re: UGL && SimNT Subject: Re: PCI interrupts on Pentium. Subject: inetstatShow() problem Subject: Re: Downloading two files to PPC Subject: memory corruption debugging Subject: Early booking - Call for Participation - Ada Europe 2000 Subject: scsiIoctl() call worked in 5.3 does not work in 5.4 Subject: Re: SNTP server and vxWorks Subject: Running DHCP without Boot ROM. Subject: Hoe to dynamically link an object module to an existing image. Subject: TimeZone in VxWorks Subject: Re: Hoe to dynamically link an object module to an existing image. Subject: Using etherInputHookAdd Subject: Re: AMD 79C973 Ethernet chip ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: DosFs 2.0/VxWorks 5.3 and FTP Date: Mon, 15 May 2000 16:00:39 +0200 From: Carsten Boeckmann Organization: Ericsson Telecom AB Message-ID: <39200307.AD521235@eede.ericsson.se> Reply-To: Carsten.Boeckmann@eede.ericsson.se Hi, we are using Tornado 1.0.1/VxWorks 5.3.1 on a PPC603 board and are using DOS FS version EAR 2.0/TFFS for managing the onboard's flash memory. For putting all files related to our SW onto the board's flash memory, we are using FTP. We FTP-log onto the board and put the files into specific directories. But sometimes we get an FTP data connection error. Does anybody know why this could happen and how it could be fixed ? Thanks in advance, Carsten Boeckmann --------------------------- Newsgroups: comp.os.vxworks Subject: Any good books on VxWorks?? Date: 18 May 2000 03:35:08 GMT From: d_robinson@mindspring.com (David Robinson) Organization: MindSpring Enterprises Message-ID: <8fvodc$37b$1@slb2.atl.mindspring.net> I just inherited some VME equipment running VxWorks. The documentation around here is almost non-existant. Can anyone recommend a good book? I haven't been able to find any, good or bad. Thanks, Dave d_robinson@mindspring.com --------------------------- Newsgroups: comp.os.vxworks Subject: Are there any functions which can compute the CPU load Date: Thu, 18 May 2000 11:45:11 GMT From: Ant Organization: Deja.com - Before you buy. Message-ID: <8g0l44$qjs$1@nnrp1.deja.com> I run vxwork on 860 and my application needs to compute the CPU load. Can some functions I can call to solve the problem? I mean any lib function or a piece of code is welcome. thank you Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks employment Date: Mon, 15 May 2000 13:49:09 -0400 From: "PROCOM" Organization: Posted via Supernews, http://www.supernews.com Message-ID: C++ / OOD VxWorks Software Developers Description: Our client, one of the world's largest telecommunications companies, currently requires 10 Junior to Senior Software Specialists to design, develop, implement and support Wireless products and applications. Minimum Requirements: All candidates should possess a minimum of 1 year experience developing real-time C/C++, OOD S/W on a VxWorks platform. Desired skills would include any previous Wireless development experience. Location: Ottawa, Ontario, Canada Type: Full-Time Placement or Contract Compensation: Open PROCOM: Established in 1978, Professional Computer Consultants Group Ltd. (Procom) is a national leader in the provision of Computer personnel on a contract and full-time basis. Our clients are comprised of the largest national and international corporations that utilize technical resources extensively across a wide range of disciplines. In the Financial Post (March 1999) Procom was ranked as the 6th largest professional Services Company in Canada. In November of 1999, Procom was named a Regional finalist in Canada' s 50 Best Managed Private companies. Our track record is proven with more than 180 consultants servicing Ottawa's high tech community and more than 1600 Procom consultants currently on assignment throughout. North America. For further information on this and other opportunities please visit our web site at www.procom.ca. Interested candidates are invited to forward their resumes or questions in confidence to: Derek Weber PROCOM 300 March Rd Suite 600 Kanata, Ontario K2K-2E2 613-270-9339 x231 613-270-9449 (FAX) derekw@procom.ca www.procom.ca Derek Weber PROCOM 300 March Rd Suite 600 Kanata, Ontario K2K-2E2 613-270-9339 x231 613-270-9449 (FAX) derekw@procom.ca www.procom.ca --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UGL && SimNT Date: Thu, 18 May 2000 12:55:59 GMT From: pgruet@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8g0p8m$vlm$1@nnrp1.deja.com> References: <0b12b743.7cfd46c4@usw-ex0102-016.remarq.com> In article <0b12b743.7cfd46c4@usw-ex0102-016.remarq.com>, Heiko Elger wrote: > Hi, > > Has anyone ported a UGL-driver for SimNT. > > I think for someone who knows the Win32 graphics API very good, > it must be possible to to replace the UGL calls with it's > equivalent function (Win32). > > I know that UGL 2.0 will have support for SimNT but I need it as > sonn as possible. > > best regards > Heiko Elger > > * Sent from RemarQ http://www.remarq.com The Internet's Discussion Network * > The fastest and easiest way to search and participate in Usenet - Free! > > Sorry dont have the answer but, out of curiosity what is SimNT? Thanks Paul Gruet Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: UGL && SimNT Date: Thu, 18 May 2000 15:27:51 +0200 From: "Martin Raabe" Organization: T-Online Message-ID: <8g0rbo$ppm$18$1@news.t-online.com> References: <0b12b743.7cfd46c4@usw-ex0102-016.remarq.com> <8g0p8m$vlm$1@nnrp1.deja.com> Hello, SimNT ist the VxWorks simulator for TornadoII. Ciao - -- Martin Raabe --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCI interrupts on Pentium. Date: Thu, 18 May 2000 09:45:30 -0400 From: Erick Gonzalez Organization: Marconi Message-ID: <3923F3F9.2937618B@marconi.com> References: <3923A475.2F8051E9@oce.nl> Hank: First hook up your ISR. Make sure you use pciIntConnect instead of intConnect, so you can have different boards in the bus using the same line... IRQ_LINE is your interrupt line (i.e. 9) INT_NUM_IRQ0 is the base vector for IRQ0 (0x20) somecookie is an argument to be passed to the ISR at interrupt time. if(pciIntConnect(INUM_TO_IVEC(IRQ_LINE+INT_NUM_IRQ0), he_isr,somecookie)==ERROR){ /*kaput*/ return(ERROR); } /* enable this interrupt line in the PIC */ sysIntEnablePIC(IRQ_LINE); That's it. It works for me... Erick Hank van Bekkem wrote: > Hi, > > I can't get PCI interrupts to work on a Pentium platform using > VxWorks 5.4 > > I implemented them just like the example in the BSP itself for > SCSI, but no interrupts get through. > > I know for certain my PCI board signals interrupts, because it works > fine in a PowerPC MTX604 board with VxWorks 5.4. > > I *am* able to read/write from/to my PCI board from the Pentium > platform. > > Can anyone help ? Perhaps with some example driver code ? > > Thanks in advance, > Hank van Bekkem --------------------------- Newsgroups: comp.os.vxworks Subject: inetstatShow() problem Date: Wed, 17 May 2000 07:56:16 GMT From: martins2@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8ftjaq$d5i$1@nnrp1.deja.com> Hello everybody, Is there a way to speed up clean up of the tcp/ip connection buffer that can be displayed using inetstatShow()? Background: I'm trying to transfer a rather huge number of files from a VxWorks system to a pc using the VxWorks ftp client routines and an FTP Server on the PC. The files are small (about 4k) and therefore the files are transfered rapidly. This works quite well if only a couple of files are transfered at a time. If I try to transfer a large number of files opening an ftp session for each file, the tcp/ip connection buffer starts to run over after about 150 files because it looks like the os deletes the entries in the tcp/ip connection buffer only after 60 seconds (determined by using inetstatShow()). Once this happens the tcp/ip stack hangs as the entries are not deleted from the connection buffer. Any idea? Thanks in advance Martin Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Downloading two files to PPC Date: Thu, 18 May 2000 14:29:12 GMT From: crandys@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8g0un0$62u$1@nnrp1.deja.com> References: <8f9kar$l97$1@nnrp1.deja.com> <8fjup3$vu1$1@nnrp1.deja.com> <8frm5s$734$1@nnrp1.deja.com> <392191B1.BD042CC9@marconi.com> I guess, I am not communicating clearly. The first file loaded, will always load correctly and runs as expected. One of the configurations under test is to load the two files on separate hosts (a master and slave connected via a backplane) In this configuration one file is loaded on the master and the other on the slave. After loading, each file runs as expected communicating with the other file. The problem only occurs when I attempt to load both files on the same target. Randy In article <392191B1.BD042CC9@marconi.com>, Erick Gonzalez wrote: > moduleShow will display the module information even if undefined references > were found during download... > > Erick > > crandys@my-deja.com wrote: > > > I can load either file by itself OK -- the problem occurs when I try to > > load the second file (after loading the first). It does not matter > > which file I load first. The first load will always be successful and > > is verifiable using moduleShow, a subsequent attempt to load the second > > file > > always fails. > > > > Randy > > > > In article <8fjup3$vu1$1@nnrp1.deja.com>, > > john_94501@my-deja.com wrote: > > > Hi, > > > > > > Might be a silly response, but have you checked that the file is (a) > > > where you think it is and (b) has exactly the name you are specifying > > - > > > including the same capitalisation etc.? > > > > > > Try doing an 'ls' from the shell you are using to download the images > > > and see if that shows both the names OK. > > > > > > Regards, > > > > > > John... > > > > > > In article <8f9kar$l97$1@nnrp1.deja.com>, > > > crandys@my-deja.com wrote: > > > > Attempting to download a Client.out and a Server.out from NT to > > PPC604 > > > target > > > > with 64mbs of memory. When downloading the second file I get: WTX > > > Error > > > > 0x2(no such file or directory). Downloading order does not make any > > > > difference. The file sizes are 3.3 mbs for the Client and 3.6 mbs > > for > > > the > > > > Server. Both the Client and the Server have been compiled with the > > > -mlongcall > > > > option. I have done this successfully with smaller clients and > > > servers. Has > > > > anyone seen this problem or have any suggestions? Thanks C. Randy > > > Schenk > > > > > > > > Sent via Deja.com http://www.deja.com/ > > > > Before you buy. > > > > > > > > > > Sent via Deja.com http://www.deja.com/ > > > Before you buy. > > > > > > > Sent via Deja.com http://www.deja.com/ > > Before you buy. > > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: memory corruption debugging Date: Thu, 18 May 2000 20:11:40 +0530 From: "Venkatesh JS" Organization: Motorola Message-ID: <8g0vb0$9rd$1@schbbs.mot.com> hi vxworkers ! I have a memory corruption problem where in a contents of a global function pointer is getting corrupted by some task. i tried using watch(GDB) command but GDB says it's not supported on vxWorks target. Can anyone tell me a good method to find out which statement is corrupting memory ? venkat --------------------------- Newsgroups: de.sci.informatik.misc,ger.ct,fj-lang.ada,de.sci.ing.misc,fr.comp.lang.ada,comp.os.vxworks,sci.engr.safety,alt.aviation.safety,di.general,news.announce Subject: Early booking - Call for Participation - Ada Europe 2000 Date: Thu, 18 May 2000 17:19:56 +0200 From: "Hubert B. Keller" Organization: Forschungszentrum Karlsruhe GmbH Message-ID: <39240A1C.14F4B07D@iai.fzk.de> Reply-To: keller@iai.fzk.de Call for Participation - Conference and Tutorials = 5th International Conference on Reliable Software Technologies Ada-Europe 2000 June 26-30, 2000Potsdam (Berlin), Germany Dead-line for early registration extended until 24th of May! Please send registration via fax to the Conference secretariat until 24th= of May to get reduced early booking fees: Christine Harms, c/o GMD, Schloss Birlinghoven, D-53754 Sankt Augustin, G= ermany = Tel.: ++49-2241/14-2473 Fax: ++49-2241/14-2472 email: christine.harms@gmd= =2Ede For up-to-date information on the programme, the tutorials, invited speak= ers, and = daily schedules, please visit http://www.ada-europe.org/conference2000.ht= ml To get a written copy of the Advance Programme, please send e-mail to: ke= ller@iai.fzk.de ++++++++++++ Invited Speakers +++++++++++++++ Tuesday Support of technology development in the Information Society Rainer Zimmermann Head of Unit E2, European Commission, = DG Information Society, Brussels Wednesday Kingcat MCAS - Monitoring, Control and Alarm System for a luxury motor ya= cht = implemented in Ada and Java Reto Weiss Kingcat MCAS Project Manager/Software Engineer, = Paranor AG, Switzerland Thursday Ada after 10 years of usage - is there a commercial future? Bryan Pflug Chief Engineer, Simulation & Software Engineering, = Boeing Commercial Airplanes, Seattle, USA ++++++++++++ Turorials +++++++++++ Tutorial 1: = Brosgol, Ben, Java for Ada Programmers Tutorial 2: = Gonzalez Harbour, Michael, Real-Time POSIX Tutorial 3A: = Cook, David and Dupaix, Les, Ada95 for Beginners Tutorial 3B: Cook, David and Dupaix, Les, Ada95 for Ada83 Programmers = Tutorial 4: Hausen, Hans-Ludwig, Software Metrology Basics - Measurement,= Assessment = and Certification of Procedures, Objects and Agents Tutorial 5: Rosen, Jean-Pierre, The HOOD 4 design method Tutorial 6: Tian, Jeff, Tree-Based Reliability Models (TBRMs) for Early R= eliability = Measurement and Improvement Tutorial 7: Heaney, Matthew, Implementing Design Patterns in Ada95 Tutorial 8: Harbaugh John, High Integrity Ada Tasking Tutorial 9: Briot, Emmanuel, GtkAda, an Ada95 object-oriented graphic too= lkit Tutorial 10: Rybin, Sergey and Fofanov, Vasily, Building Ada development = tools = with ASIS for-GNAT Tutorial 11: Leringe, =D6rjan, Windows development with Ada Tutorial 12: Gasperoni, Franco, Developing Ada Applications for the Java = Platform = with JGNAT = ++++++++++ Conference Venue +++++++++++ Potsdam is the capital of the state Brandenburg and situated a few miles = west of Berlin, = the capital of Germany. Both cities offer many historic and modern attrac= tions and an = exciting environment to visitors. = The venue of the conference is the hotel "Seminaris" specializing in conf= erences and meetings. = It is situated at the waterfront of a picturesque lake and near the world= - -famous palace = "Sanssouci". = Social events on Tuesday and Wednesday, including a musical appearance of= "Lady Ada", = round out a week well spent to catch up on the latest developments in rel= iable software = technologies. Travel information may be found at: http://www.ada-deutschland.de/AE2000/conf_site/index4.html +++++++++ Exhibitors +++++++++++ ACT Europe, Aonix, CAS GmbH, CSC PLOENZKE, = Green Hills, Information Processing Ltd., = Irvine, PeerLogic, PolySpace, = Praxis Critical Systems and Rational. = +++++++++ Sponsors +++++++++++++ The conference is supported and sponsored by Ada Europe, Ada Deutschland = e.V., ACM SIGAda, = Gesellschaft f=FCr Informatik e.V., DFG (German Research Foundation), For= schungszentrum = Karlsruhe, ACT Europe, Aonix, CAS GmbH, Green Hills, Information Processi= ng Ltd., Irvine, = PeerLogic, PolySpace, Praxis Critical Systems and Rational. --------------------------- Newsgroups: comp.os.vxworks Subject: scsiIoctl() call worked in 5.3 does not work in 5.4 Date: Thu, 18 May 2000 16:54:22 GMT From: Mark Kelly Organization: FlashNet Communications, http://www.flash.net Message-ID: <39241FCE.2E17C434@flash.net> I am migrating a system that was originally developed on a 68K board (Motorola MV167) to a PPC board (MV2604). I am having problems with the SCSI Iomega Zip drives. There was an initial problem with the scsiPhysDevCreate() command caused by the automatic attempt to negotiate synchronous and wide communications with the Zip drives but that is now resolved. The problem now is that the scsiIoctl() calls to the Zip drives fail. This is code that worked in version 5.3. These calls use the FIOSCSICOMMAND and return either a 0 or 1 in the statusByte of the SCSI_TRANSACTION structure that is passed to the function. I think there were changes to the SCSI code in 5.3.1 that may have affected this. Thanks in advance, Mark Kelly --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SNTP server and vxWorks Date: Thu, 18 May 2000 09:51:25 -0700 From: Terry Duncan Message-ID: <39241F8D.598F9B92@cintacom.com> References: <8fgslb$sjl20@overload.lbl.gov> The SNTP time should be an unsigned long containing the number of seconds since Jan 1, 1900. ctime and other ANSI time functions expect time to be signed long containing the number of seconds since Jan 1, 1970. Do the conversion on the server side to seconds since 1900. The client side takes care of the conversion for you. I have been able to make the SNTP client server work in passive mode. However I am having trouble getting active mode to work. - -Terry Duncan Cinta Corporation Joachim Pluschke wrote: > Hi all, > > I'm trying to make the SNTP Client/Server services run under vxWorks. > The client in no problem - it works. > The Server itself should'nt make much more work...... > Has anyone made the SNTPserver run and deliver a correct timestamp??? > (The manual entrys for sntpsLib doen't make a clear statement about how > to format the timestamp needed by the server routines) > > Feeding the SNTP server with a valid date-format seems to be a little > bit difficult ... > Putting a timestamp (correct by ctime()!) like the > into the servers buffer should be the correct time-format but there's > send out trash in the packets (overflowing, negative value for the > seconds part, thus making functions like ctime() print a date in the > future far above the year 2036.... ). > > I tried filling the buffer (in host-byteoder, see manual) by two ways: > 1) copying the seconds-part to the original location of the servers > buffer, > then incrementing a copy of the buffer for the size of a long and > copying the seconds fraction part into > 2) using bcopy() with the size of the to copy to > the buffers original location..... > > There was still the same effect! > > But the joke is: > The server recognizes the clients request and increases the lower part > of the seconds part of the timestamp with exactly the amount of seconds > what was the difference between the clients requests - something must be > working.... > > I used a packet sniffer and swapped the bytes/words for a try with no > effect on solving the problem! > > What's wrong? I don't really know! > > I'm looking forward hearing from someone who made it run !!!! > > bye, have a fine day > -- > ===================================================== > Gebrüder Stoye GmbH > department: traffic control systems > Name: Joachim Pluschke > Adr: Longericher Strasse 177 > city: Cologne, Germany > Zip: 50739 > phone: ++49/221/2616-534 > mailto: j.pluschke@stoye.de > www: www.stoye.de > ===================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Running DHCP without Boot ROM. Date: 18 May 2000 17:33:18 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: Sender: bpringle@DeadDuck I am trying to get DHCP to run from a standalone image. My ethernet driver is a NE2000 END device. I have started the stack, - -> sockLibInit(400); value = 0 = 0x0 - -> hostTblInit() value = 0 = 0x0 - -> usrNetProtoInit(); value = 0 = 0x0 - -> netLibInit(); value = 0 = 0x0 - -> muxLibInit() value = 0 = 0x0 - -> muxDevLoad(0,ne2000EndLoad,"0x20000300:0x0a:0x0a:0:1:0x02",1,"Installed!",0); value = 32672580 = 0x1f28b44 - -> pcookie=0x1f28b44; new symbol "pcookie" added to symbol table. pcookie = 0x1ffc034: value = 32672580 = 0x1f28b44 - -> muxDevStart(pcookie); - -> endFindByName("ene",0) value = 32672580 = 0x1f28b44 - -> dhcpcLibInit(67,68,4,5,3600,30); value = 0 = 0x0 - -> ipAttach(0,"ene"); value = 0 = 0x0 - -> ifunit("ene0") value = 164544 = 0x282c0 = ipDrvCtrl - -> ifShow ene (unit number 0): Flags: (0x8063) UP BROADCAST MULTICAST ARP RUNNING Type: ETHERNET_CSMACD Netmask 0x3ce3e098 Subnetmask 0xf38f8263 Ethernet address is 00:c0:1b:00:bf:47 Metric is 0 Maximum Transfer Unit size is 1500 540 packets received; 0 packets sent 540 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped value = 29 = 0x1d - -> dhcpcInit(ifunit("ene"),1) value = 0 = 0x0 The NE2000 device has been attatched to the IP stack. I have done the dhcpcLibInit(). I don't know what other functions I should be calling before hand. Any ideas? thanks, Bill - -- I used to be a FUNDAMENTALIST, but then I heard about the HIGH RADIATION LEVELS and bought an ENCYCLOPEDIA!! --------------------------- Newsgroups: comp.os.vxworks Subject: Hoe to dynamically link an object module to an existing image. Date: Thu, 18 May 2000 14:34:42 +0800 From: "TonyHuang" Organization: BST Communication Technology Ltd. Message-ID: <8g02gp$1h28$1@news.cz.js.cn> Hi folks, Our system has no file system, and we can not use the loadLib.h of vxWorks. Can anyone give us some suggestion to link and object module to an running system? Best Regards Tony Huang Senior Hardware Design Engineer BST Communication technology LTD. F11 XuNeng Building, 32 LongKou Road East, TianHe,GuangZhou 510635, P.R.China. 8620-87581648-111 telephone, 8620-87568512 email: huangxin@163.net --------------------------- Newsgroups: comp.os.vxworks Subject: TimeZone in VxWorks Date: Thu, 18 May 2000 23:22:43 +0800 From: "hyla" Organization: The news server of Changzhou TeleCom Message-ID: <8g11s4$2g5r$1@news.cz.js.cn> Dear all, In Tornado 2, the BSP use UTC as default timezone. How can I modify it? In the manual, I got that "tools.h++" supports multi timezone. But I have no this component. Can I use other timezone so? I want the target to use the local timezone, and I think it is easier to be maintained. Hyla --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hoe to dynamically link an object module to an existing image. Date: Fri, 19 May 2000 07:43:52 +0200 From: "Achim Zimmer" Organization: Colt Telecom GmbH Message-ID: <8g2k7t$il2$1@crusher.de.colt.net> References: <8g02gp$1h28$1@news.cz.js.cn> I am missing some infos about your environment (host, target etc.). But one way we have done is to place a simple driver over a flasharea, where we have stored the ELF file. we make an open to that and can then read with tools from loadLib the file into the RAM. If you dont have a flash, maybe you can place a memDrv on the RAM, where you place the ELF file. Achim TonyHuang schrieb in im Newsbeitrag: 8g02gp$1h28$1@news.cz.js.cn... > Hi folks, > > Our system has no file system, and we can not use the loadLib.h of vxWorks. > Can anyone give us some suggestion to link and object module to an running > system? > > Best Regards > Tony Huang > > Senior Hardware Design Engineer > BST Communication technology LTD. > F11 XuNeng Building, > 32 LongKou Road East, > TianHe,GuangZhou 510635, > P.R.China. > 8620-87581648-111 telephone, > 8620-87568512 > email: huangxin@163.net > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Using etherInputHookAdd Date: Fri, 19 May 2000 16:28:14 +0800 From: "Tenshou.Zha" Organization: Alcatel Message-ID: <3925cfb5.0@dnews.sbell.com.cn> Hi, all We can use etherInputHookAdd in Tornado-1 successfully, but when we use it in Tornado-2, we can't receive any packets, Any suggestion will be appreciated, thanks. Best Regards Tenshou Zha --------------------------- Newsgroups: comp.os.vxworks Subject: Re: AMD 79C973 Ethernet chip Date: Tue, 09 May 2000 09:15:46 -0400 From: "Emory R. Stagmer" Organization: Ezekiel's Wheel (works for Litton, Amecom division) Message-ID: <39180F82.86027FCB@LittonAS.com> References: Charles Krinke wrote: > > A new question. I have had an AMD 79C973 ethernet chip dumped in my lap and > need to provide a driver for it. Has anyone heard of one that allready > exists or is anyone else working on one that might like to collaborate? I'd contact AMD tech support and ask them for it. I know there are VxWorks drivers available for several AMD devices (we just checked the other day)... - -- Emory R. Stagmer Scientist, Litton Advanced Sys Div - Space Systems Operation http://www.amecom.com Bass/12string/keyboards/windsynth for Ezekiel's Wheel http://www.untiedmusic.com Baltimore CMC Chapter Coordinator & North Atlantic CMC Regional Board http://www.cmcnet.org RingMaster for the Christian Music Ring http://www.webring.org/cgi-bin/webring?ring=christmusic;home --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri May 19 06:31:21 2000 From: Atkinson Andrew Date: Fri May 19 06:31:23 PDT 2000 Subject: WindView & TSFS - Hi We have a WinNT Host connected to a Radstone PPC1a running Tornado 2 and VxWorks 5.3.1 respectively. We have recently installed WindView and have configured and built the tool into the kernel. We can run the tool, but are unable to upload data from the target. When attempting this we get an error message stating that the TSFS is configured for read-only and must be read-write for the upload to succeed. In the target server configuration dialogue we have selected that TSFS be read/write and we have selected an upload path. What could be the cause of this continuing problem and how can we fix it? We have approached Wind River direct, but have heard very little in return! Regards Andy Atkinson Software Engineer DERA Bincleaves Weymouth Dorset, DT4 8UR -- The Information contained in this E-Mail and any subsequent correspondence is private and is intended solely for the intended recipient(s). For those other than the recipient any disclosure, copying, distribution, or any action taken or omitted to be taken in reliance on such information is prohibited and may be unlawful. From vxwexplo-errs@csg.lbl.gov Fri May 19 18:40:00 2000 From: Fred Roeber Date: Fri May 19 18:40:02 PDT 2000 Subject: Re: WindView & TSFS On Fri, 19 May 2000, Atkinson Andrew wrote: > We have a WinNT Host connected to a Radstone PPC1a running > Tornado 2 and VxWorks 5.3.1 respectively. We have recently installed > WindView and have configured and built the tool into the kernel. We can > run the tool, but are unable to upload data from the target. When > attempting this we get an error message stating that the TSFS is > configured for read-only and must be read-write for the upload to > succeed. In the target server configuration dialogue we have selected > that TSFS be read/write and we have selected an upload path. What could > be the cause of this continuing problem and how can we fix it? We have > approached Wind River direct, but have heard very little in return! Andy, In looking at the WRS documentation for WV 2.0 I noticed some things you have to do to be able to use TSFS for WindView. The references are relative to the WindView 2.0.1 User's Guide: - Section 2.4.2 - You must have WIND_UID set for WinNT host. - Section 2.4.3 - You need to select TSFS RW mode (vs read only) for WindView which introduces a security hole. If all else fails, switch to socket data upload. Hope this helps. Fred Ever notice Windows NT = WNT = ++VMS | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Sat May 20 04:04:31 2000 From: Vxworks Exploder Date: Sat May 20 04:04:34 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat May 20 04:03:41 PDT 2000 Subject: Memory allocation problem Subject: Re: TimeZone in VxWorks Subject: Re: Any good books on VxWorks?? Subject: Re: Are there any functions which can compute the CPU load Subject: UDP transmit problem Subject: Re: bcopy_to_mbufs, bcopy_from_mbufs Subject: Re: sa1100IntrCtl.c bug. Subject: Re: Can I use CORBA and still make the program POSIX compliant? Subject: Re: ROMable VxWorks [objcopy: SREC->BIN] Subject: Re: MVME 162-022A wanted ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Memory allocation problem Date: Tue, 09 May 2000 13:41:04 GMT From: Tri Vu Khac Organization: Sony Software Development Center Message-ID: <39181566.5EFE6995@sonycom.com> Dear folks, I'm using "new" to allocate memory and I don't want to get memory exceptions, but the NULL value as C's traditional allocations. I've tried the compilator's flag fno-exceptions to turn off exceptions, but this doesn't help. Does anyone have any idea of this ? Thanks so much. Tri. My code is : char *p = new char [1000000000000]; // ;-)) return (p) ? 0 : -1; --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TimeZone in VxWorks Date: Fri, 19 May 2000 13:40 +0100 (BST) From: rew@cp.cix.co.uk (Richard Wenham) Organization: CIX - Compulink Information eXchange Message-ID: References: <8g11s4$2g5r$1@news.cz.js.cn> Reply-To: rew@cp.cix.co.uk You need to set the TIMEZONE environment variable, using putenv(). See the ansiTime library documentation. Richard In article <8g11s4$2g5r$1@news.cz.js.cn>, h_huang@21cn.com (hyla) wrote: > Dear all, > > In Tornado 2, the BSP use UTC as default timezone. How can I modify > it? > In the manual, I got that "tools.h++" supports multi timezone. But I > have no this component. Can I use other timezone so? I want the target > to > use the local timezone, and I think it is easier to be maintained. > > Hyla --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Any good books on VxWorks?? Date: Thu, 18 May 2000 05:11:05 GMT From: snaphook@southwesternbell.net (Bruce) Organization: SBC Internet Services Message-ID: <39237b1b.18291444@news.swbell.net> References: <8fvodc$37b$1@slb2.atl.mindspring.net> In comp.os.vxworks d_robinson@mindspring.com (David Robinson) wrote: >I just inherited some VME equipment running VxWorks. The documentation around >here is almost non-existant. Can anyone recommend a good book? I haven't been >able to find any, good or bad. I'm not trying to be a smartass when I say that the manuals are pretty good. The VxWorks User's Guide starts from the simple and goes all the way to the complex and covers it pretty well. Much of it is applicable to ANY multitasking OS. Bruce --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Are there any functions which can compute the CPU load Date: Fri, 19 May 2000 18:35:54 +0530 From: "Venkatesh JS" Organization: Motorola Message-ID: <8g3e3e$8ou$1@schbbs.mot.com> References: <8g0l44$qjs$1@nnrp1.deja.com> Start a low priority task(pr = 255) and in that task increment a global counter in a while(1) loop. by checking this counter at periodic intervals you can calculate the CPU load. Ant wrote in message <8g0l44$qjs$1@nnrp1.deja.com>... > I run vxwork on 860 and my application needs to compute the CPU load. >Can some functions I can call to solve the problem? > I mean any lib function or a piece of code is welcome. > >thank you > > >Sent via Deja.com http://www.deja.com/ >Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: UDP transmit problem Date: Fri, 19 May 2000 14:26:54 -0400 From: Brian Justice Organization: Posted via Supernews, http://www.supernews.com Message-ID: VxWorkers, Anyone have anything to share regarding UDP transmissions locking the IP stack in VxWorks (5.3.1)? We have a Intel based CPU board with a DEC 21143 ethernet controller. If I xmit UDP packets out to anyone (more than just a few, a lot of them) the IP stack locks. Interesting to note though, is that if no one is receiving them, I seem to have no problems. As a result of the lockup, I'm seeing ENOBUFS returned from sendto(). I am not running out of fds and my nw cluster pool is not low either. The only way to recover is to reboot the board... I am using SENS. Any help much appreciated... Brian ======================================================================== Brian K. Justice | E-Mail: bjustice@patriot.net 1847C Montclair Drive | Phone: (843) 971-0856 Mount Pleasant, SC 229464 | http://patriot.net/~bjustice ======================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bcopy_to_mbufs, bcopy_from_mbufs Date: 19 May 2000 21:11:28 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: Customer of Planet Online Message-ID: References: <8fp7ra$qv22@overload.lbl.gov> FWIW, you can probably get a good idea as to what these functions do and how they are implemented by taking a look inside one of the various BSD variant IP stacks, or by reading one of the big fat BSD TCP/IP stack books. Rgds! Luke Diamand In article <8fp7ra$qv22@overload.lbl.gov>, Tim Allen wrote: >>Can one call the vxworks functions bcopy_to_mbufs & bcopy_from_mbufs from >an >>Interrupt Service Routine? Do these functions block at all? What exactly >>do these functions do? > >The vxWorks documentation on these is pretty scarce! They are used in >network drivers to copy data between device driver buffers and the network >stack. The mbufs are chained and filled by the stack before being passed to >the driver for transmission, and vice-versa. > >As to whether they block, I'd say bcopy_from_mbufs doesn't, because it fills >a user supplied buffer, but bcopy_to_mbufs might block because I think it >will allocate mbufs as required and I don't recall how this is done. > >The preferred method for network drivers to deal with data is to dispatch >the work to task level from the ISR using netJobAdd(). You can see this in >any network driver, or if you don't have source, you should find a template >in target/src/drv/netif/templateNetif.c > >Tim. >-------------------- >Tim Allen Ltd >Real Time Embedded Software Consultancy > - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: sa1100IntrCtl.c bug. Date: Fri, 19 May 2000 22:15:27 GMT From: "Steve Dillen" Message-ID: <32jV4.1137$s4.90925@petpeeve.ziplink.net> References: Howdy, I have been looking at this source code as well, I don't know if I fully understand it, but according to documented code in syslib.c with the sa1100IntLvlMask array declaration, level 0 indicates ALL interrupts are disabled. This implies that the interrupt levels start at 1 _not_ 0. If this is the case then the below code in question is correct for the line you are indicated, but not the other line > sa1100IntLvlEnabled |= (1 << level); sa1100IntLvlEnabled |= ( 1 << ( level - 1 ) ); And similarily in the Disabled routine. I haven't looked into how the .h declarations for interrupt levels are based, or how it will be affected. PS: I seem to be having problems posting to the newsgroup so I copied you on it as well. Seems like we are among the few developing for the SA11x0 Thanks SJD Bill Pringlemeir wrote in message ... > >Does anyone know if there is an SPR for the StrongARM interrupt >controller? If you have an interrupt on GPIO-0, the interrupt >level is 0. In the intDisable, intEnable functions, > > STATUS sa1100IntLvlEnable(int level) > { > int key; > > if (level < 0 || level > SA1100_INT_NUM_LEVELS) > return ERROR; > > /* set bit in enable mask */ > if(level > 0) { /* !!! I think this is wrong. !!! */ > > key = intLock (); > sa1100IntLvlEnabled |= (1 << level); > intUnlock (key); > } > > sa1100IntLvlChg (-1); /* reset current mask */ > > return OK; > } > > > STATUS sa1100IntLvlDisable(int level) > { > > int key; > > if (level < 0 || level > SA1100_INT_NUM_LEVELS) > return ERROR; > > /* clear bit in enable mask */ > > if(level > 0) { /* !!! I think this is wrong. !!! */ > > key = intLock (); > sa1100IntLvlEnabled &= ~(1 << level); > intUnlock (key); > } > > sa1100IntLvlChg (-1); /* reset current mask */ > > return OK; > } > >The `if(level>0)' line in sa1100IntrCtl.c does not allow GPIO zero line >to be attached to interrupts. Removing the condition seems to fix the >problem. I was just wondering if there are any reprocussions. > >Have a good weekend, >Bill > >-- >Who is Kibo? > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can I use CORBA and still make the program POSIX compliant? Date: Sat, 20 May 2000 00:13:03 GMT From: "Ron S. Ahn" Organization: OR Concepts Applied Message-ID: References: <8g45ks$qkh$1@bob.news.rcn.net> The customer. I don't know why our customer wants our application to be POSIX compliant. "Bill Pritchett" wrote in message news:8g45ks$qkh$1@bob.news.rcn.net... > What is the impetus for being POSIX compliant? > > "Ron S. Ahn" wrote in message > news:ajfV4.5277$T41.124964@newsread1.prod.itd.earthlink.net... > > Hi > > > > I really could not figure out if using CORBA would break the POSIX > > compliance... > > > > Please help. > > > > Ron Ahn > > OR Concepts Applied > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ROMable VxWorks [objcopy: SREC->BIN] Date: Thu, 18 May 2000 10:28:31 -0400 From: Jeff Daly Organization: Sanders A Lockheed Martin Company Message-ID: <3923FE0F.1EFF9EFB@lmco.com> References: <8g08bv$lhs$1@snipp.uninett.no> Reply-To: jeffrey.j.daly@lmco.com kernels are statically linked. try using elfToBin tool to convert ELF to binary. Geir Frode Raanes wrote: > I am experiencing some problems in creating a (compressed) > VxWorks kernel resident in ROM. First off - the kernels I > make with Tornado 2 is obviousely ELF32 object files. > > But how do I make (powerpc-wrs-vxworks-)objectdump tell whether > or not this object file still contains unresolved relocation > information? I tried "-r" on both the compressed and uncomressed > kernels. Even the download version gives the same result: > > powerpc-wrs-vxworks-objdump -r vxWorks.rom > powerpc-wrs-vxworks-objdump -r vxWorks.romCompressed > > vxWorks.rom: file format elf32-powerpc > > The same "-r" argument on userspace applications produce long > lists of relocation information. Am I then safe to assume that > the VxWorks kernel is statically linked? Ie. that it does _not_ > require any ELF aware program loader? That all the loader has > to do is unpack the kernel into the predefined location > RAM_LOW_ADDRESS, then set up this BOOT_LINE information > exchange data block and finally jump to the kernel > entrypoint _sysInit? > > Now, on the MBX860 board there is two sets of flash. Booting > EPPCBug from on set we can download a binary image from net > and flash it into the other set. This works fine for existing > VxW boot ROMs. Tornado 2 produces two ROM kernel files, first > the ELF32 file mentioned above and next a SREC hex version of > the above. But EPPCBug requires a binary image to flash, so > how do I use objectcopy to produce a binary kernel image from > one of these files? > > -- > ****************************************************** > Never ever underestimate the power of human stupidity. > -Robert Anson Heinlein > > GeirFRS@invalid.and.so.forth > ****************************************************** --------------------------- Newsgroups: comp.realtime,comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: MVME 162-022A wanted Date: Sat, 20 May 2000 06:18:40 GMT From: yodathejediknight@att.net (Gregg C Levine) Organization: AT&T Worldnet Message-ID: <47qV4.75411$WF.4217984@bgtnsc04-news.ops.worldnet.att.net> References: Hello from Gregg C Levine probably with Jedi Knight Computers If Lars Troppen did not ask you to send him the 162-013A, then I am interested. By the way, the original message is dated 4/19/00, and I am reading this one on 5/19/00, something over a month later. Please contact me via an out of newsgroup message, if this is possible, or not. Gregg C Levine mailto:yodathejediknight@att.net "Do, do not. There is no try." Yoda In article . >In article , fryerre#@#earthlink.net says... >> >>Try Sun Harbor Sales (http://www.ghgcorp.com/wmcalpin/index.htm) >>He has a 162-022 for sale. I've bought from him - & happy with the results. >> >>I myself have a 162-013A in new condition that's excess to my needs at the >>moment if it would be of >>help. >> >> wrote in message >>news:vo8vh1mpgmu.fsf@kongsberg.com... >>> >>> I have inherited some simulation sw that only runs on MVME 162-022A, >>> but this card is no longer in production. Is there anyone out there >>> who has such a board that they could sell us ? >>> >>> Any hints about where to get one is also welcome. >>> >>> >>> -- >>> Lars Troppen, Avd DC4S, Kongsberg Defence & Aerospace >>> PO 1003, N3601 Kongsberg, Norway >>> Tel: (+47) 32738722 Mail: lars.g.troppen@kongsberg.com >> >> > --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun May 21 04:03:59 2000 From: Vxworks Exploder Date: Sun May 21 04:04:01 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun May 21 04:03:56 PDT 2000 Subject: Re: VxWorks 5.4 and dosFs 2.0 Problems Subject: Re: loaned buffers in vxWorks Subject: Re: Can I use CORBA and still make the program POSIX compliant? Subject: Re: Is CORBA compatible with VxWorks? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks 5.4 and dosFs 2.0 Problems Date: Sat, 20 May 2000 16:56:43 GMT From: Pete Kockritz Organization: @Home Network Message-ID: <3926C3D3.82C70740@home.com> References: <391d46ab.994293534@news.netis.com> "RKT Technologies, Inc." wrote: > > We have begun work on a real time telecommunications device for high > speed networks, based on the PowerPC (MCP-750) and VxWorks version > 5.4. You don't say if this is a custom board or not. I'm assuming a custom board. > message, but no target system shell prompt or VxWorks banner (because > the target system shell and banner had not yet been enabled and > evidently aren't included by default). Not everyone wants or needs the target shell. WRS picks defaults for the vxWorks kernels but most everyone is going to have to adjust them for their own needs. > (This leads to another question ... How do we know when we need to > hand-edit something that didn't get updated by changing compile > elements in the VxWorks tab in Tornado? Most of my BSP experience is with T1, but with T1 I had to edit quite a few files, either adding our own drivers to the kernel init or to incorporate NVRAM support for the bootline, or several other things. Even if you have a OTS board with a OTS BSP, you may have to modify some files, since the way you are using the board may be different than the way the BSP configures it. One case in point, most of the OTS BSPs I've seen have the ethernet MAC address hard coded (if the board doesn't have some NVRAM to store it in). This works fine for 1 board, but if you have several boards you're either going to have to edit the files and build unique kernels/bootroms for each board or add in a way to have the board obtain a unique MAC address (or enter it in manually each time you boot the board). > If the other engineer is right and we're supposed to manually add > other code to the project in order to get this to work, how can we > determine what code we need to add other than to just "know" it? Did either of you take the WRS BSP Developer's class? Sounds like you didn't. That would've helped you know what kinds modifications you need in your BSP. The BSPs that I've seen that were for eval boards (ADS860, ...) I consider to be a _starting_ _point_ for a BSP. It provides a framework that you have to customize for your board. BSP from vendors that support a board that they also make, I would expect to be a more complete BSP, i.e., have support for nearly every feature of that board. But those BSPs still might require some customization. - -- Regards, Pete - -- LinuxPPC --------------------------- Newsgroups: comp.os.vxworks Subject: Re: loaned buffers in vxWorks Date: Tue, 16 May 2000 18:16:21 -0500 From: Robert Brantley Organization: Lockheed Martin Corporation Message-ID: <3921D6C5.377A39E@lmco.com> References: <391C29D0.4D327086@fore.com> Look at zbuffers in the reference manual(zbufLib). Also, there is the zbufSockLib - zbuf routines for communicating over BSD sockets. Robert Brantley Erick Gonzalez wrote: > Hi everybody. Q: I am porting a driver from Solaris to vxWorks, and I > have come across this issue: There, they use "loaned" buffers from the > OS. That is, when the buffer is allocated, you also provide a callback, > and the OS will call that function when that buffer is released. Is > there an equivalent mechanism in vxWorks?. > > Help is greatly appreciated... > > Erick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can I use CORBA and still make the program POSIX compliant? Date: 20 May 2000 18:18:34 GMT From: Geir Frode Raanes Organization: UNINETT news service Message-ID: <8g6ktq$s3b$1@snipp.uninett.no> References: <8g45ks$qkh$1@bob.news.rcn.net> Sender: Geir Frode Raanes Ron S. Ahn wrote: : The customer. : I don't know why our customer wants our application to be POSIX compliant. Wild guess - to protect themselves from technology lock-in...? - -- ****************************************************** Never ever underestimate the power of human stupidity. -Robert Anson Heinlein GeirFRS@invalid.and.so.forth ****************************************************** --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is CORBA compatible with VxWorks? Date: Sat, 20 May 2000 19:16:29 GMT From: bill.beckwith@realtime-corba.com (Bill Beckwith) Organization: Objective Interface Message-ID: <8F3A9B556beckwbnewscom@sigma> References: Earthlink wrote in : >Hi All, > >I am new at Tornado systems and VxWorks. >I have to develop a client and server application that will run under >VxWorks. >I am only allowed to use POSIX compatible subset of VxWorks API. > >Can I use CORBA to implement communication between the client and server? Yes. There are several ORB products that are available for VxWorks. >Or I have to go down to Socket level? or further down? No need for this. Some ORBs are fairly small (100K-200K) and can get very close to the performance of using sockets directly. The US DoD funded Boeing to do a Real-Time ORB trade study. For details see: http://www.ois.com/technical/rt_diicoe/release.html - -- Bill --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon May 22 04:03:48 2000 From: Vxworks Exploder Date: Mon May 22 04:03:50 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon May 22 04:03:45 PDT 2000 Subject: Interrupt on MIPS Subject: Help!Any one familiar with Motorola's JPEG package for MPC82x Subject: Hot off the press!! Britney Nude!!! Subject: Re: Any interactive filesystem tools available? Subject: Re: Loading Question? Subject: Re: Modem dialer for VxWorks ? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt on MIPS Date: Sun, 21 May 2000 05:04:25 +0200 From: "Øyvind Holmeide" Organization: Telenor Online Public Access Message-ID: I am looking for some sample code for initialization of an ISR on a QED RM5231 platform. What I have done so far, without success, is the following: initIsr() { STATUS retVal; retVal = intConnect (INUM_TO_IVEC (IV_IORQ3_VEC), tstIsr, 0); if retVal == ERROR) return ERROR; intEnable(CAUSE_IP6); return OK; } and in sysHwInit(): sysHwInit(){ : : sr |= CAUSE_IP6; /* init status register but leave interrupts disabled */ taskSRInit (sr); : : } Can anyone help? regards Øyvind Holmeide --------------------------- Newsgroups: comp.os.vxworks Subject: Help!Any one familiar with Motorola's JPEG package for MPC82x Date: 21 May 2000 15:38:34 GMT From: "Clark" Organization: DCI HiNet Message-ID: <01bfc33a$87add860$c89247d2@server2> Dear All, Any one familiar with Motorola's JPEG package for MPC82x? I'm tried to port the FDCT funtion in the above package to VxWorks target, However, the returned value of FDCT is always wrong! I set all the, microcode, FD and parameter. and I'm sure the CPM is performing the FDCT because the output buffers is updated after DSP START command asserted. But the reading is still worng( all 0's or some 0's plus few 0xffff). Besides, I've tried the embadded DSP function, such as WADD function. the result is very good. Please let me know if you know what's the probable cause of this problem? Target :MPC823 RTOS: VxWorks Clark --------------------------- Newsgroups: comp.os.vxworks Subject: Hot off the press!! Britney Nude!!! Date: Sun, 21 May 2000 16:02:08 GMT From: britney2000@hotmail.com Organization: Papartazi Message-ID: Hot from the camera...The real pics of Britney Spears---Taken at private nude beach in UK! Check them out at http://216.15.169.227/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Any interactive filesystem tools available? Date: Fri, 19 May 2000 17:31:11 +0300 From: "Fornaro Antony" Organization: Intracom S.A., GREECE Message-ID: <8g3jka$d9m$1@newsserv.intranet.gr> References: <391f1bbe$0$208@nntp1.ba.best.com> Hi Chris, you can use simple calls like cd,copy,pwd,rename just fine from the target shell (host shell doesn't work for me in this case) If target shell doesn't recognise these commands (or not even the "help" command), make sure that you include the INCLUDE_SYM_TBL_INIT. "Chris Wein" wrote in message news:391f1bbe$0$208@nntp1.ba.best.com... > Maybe I just missed where these are located but are there any simple > interactive tools that work with VxWorks filesystems (particularly > dosfs2)? I am looking for simple things like ls, cp, mv. I knows these > are pretty simple but I am looking to avoid re-inventing the wheel (even > if the wheel is pretty small in this case) > > Thanks, > > Chris > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Loading Question? Date: Sun, 21 May 2000 22:56:05 -0700 From: Mark Kozel Organization: ITT Industries Message-ID: References: <8fp7r9$qv12@overload.lbl.gov> On Sun, 14 May 2000 21:08:59 -0400 (EDT), Fred Roeber wrote: >On Thu, 11 May 2000 WDIZ asked: > >> How can I load 2 applications which depended on each other? >> Does Vxworks support that? > >The easiest way I know of is to prelink the two object files together >and and then download them in one shot. You can do this usually using >the "-r" option to the "ld". This option preserves the undefined symbol >references (which are later resolved when the file is downloaded. The >-r is usually combined with the -X option to delete the temporary >symbols generated during linking. Fred Sorry, I can in in the niddle of this thread. We used extern for functions and finally, classes in C++ to load several different modules that "talked" to each other. With classes, we did need to instanciate classes at the end of the class definitions to avoid a link error. We did this from the Project facility. Tornado 2, vxWorks 5.4 PPC Target, Solaris Host C++ - -Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Modem dialer for VxWorks ? Date: Mon, 22 May 2000 13:43:41 +0800 From: "Jerry Pan" Organization: DCI HiNet Message-ID: <8gahf3$ebq@netnews.hinet.net> References: <3904ACDC.EE18D3E8@hns.com> <8e8mhq$13h$1@nnrp1.deja.com> <8e967o$i52$1@nnrp1.deja.com> Gentlemen, I am trying to write a similar dial-up application in vxWorks. I have made the connection work and ppp negotiate successfully with PAP. But I have no idea about using CHAP negotiation. Have you any idea to provide me. I am trying to connect Microsoft NT RAS server. Thanks! Jerry Pan jerry@home.regulus.com.tw S Austin wrote in message news:8e967o$i52$1@nnrp1.deja.com... > In article <8e8mhq$13h$1@nnrp1.deja.com>, > john_94501@my-deja.com wrote: > > Hi, > > > > Not being an expert here I might be talking complete rubbish, but from > > what I remember dialing a modem is simply a matter of sending it the > > appropriate commands over its serial interface. > > > > You're right John, that's what I've seen in the apps that use a modem. > Since the mid-80s nearly every modem vendor has impelemented the de > facto Hayes "AT" command set - so there's at least a baseline set of > commands that will work with most modems. On top of that there will be > vendor-specific things like the contents of registers, etc. but most > basic stuff like setting the baud rate, listening for a dial done, DTMF > dialing, etc. are simple "AT" commands. > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon May 22 04:58:13 2000 From: Uygar DOYURAN Date: Mon May 22 04:58:15 PDT 2000 Subject: Scheduling Problem We are using VxWorks over MPC 860. Rarely, we experience a scheduling problem. The most prior task in our system pends on a semaphore which is released by an ISR upon occurrence of some event. We observe that, sometimes, although the ISR releases the semaphore, the task is scheduled about 2 milliseconds later. We discovered that if we replace (to debug this problem) the 'taskLock' function by just a 'return from function' statement, this problem does not occur. Do you have any idea 'where in the kernel and why' it is required to disable scheduling? What can we be doing wrong to introduce such a long delay for task scheduling? Thanks in advance for your concern. Uygar DOYURAN Electrical & Electronics Engineer, Msc. ASELSAN Electronics Industry, Inc From vxwexplo-errs@csg.lbl.gov Mon May 22 08:04:27 2000 From: "Tyler, Paul A (DBY)" Date: Mon May 22 08:04:29 PDT 2000 Subject: Re: UDP transmit problem Brian, This sounds like a problem I am also experiencing. In my case I am using a Pentium based VME CPU from VMIC (VMIVME-7591), with a second network interface (Ramix PMC-661) in the PMC slot. Both the on board and the add-on interfaces are DEC 21143's. I am using the dc driver which came with the BSP, in a 5.3.1 non-SENS kernel. The Ramix driver would not work, as it would not connect to the interrupts. I see the problem when I reset a number of the connected pressure measurement scanners. The scanners use bootp to establish an IP address, then wait for a TCP socket connection from my client tasks. Resetting them singly appears to be OK, but resetting them all (e.g. by interrupting the power supply) stops the network interface. The network still receives the bootp requests from the scanners, but is unable to send the reply. All transmits then cease on that interface (but the other interface continues), and ifShow() indicates transmit error count increasing. mbufShow seems to indicate plenty of free buffers. inetstatshow indicates the interrupted TCP connections are still live, with a status of either FIN_WAIT_1 or SYN_SENT. This doesn't seem right. The client tasks should drop the connection as soon as the data stream is interrupted. Can anyone shed any light on this? I have not diagnosed the problem much further. I am living with the re-boots. VxWorks, Tornado Paul Tyler Measurement Systems Specialist Electronics & Measurement Systems Dept. Rolls-Royce plc. e-mail Paul.A.Tyler@Rolls-Royce.com Tel. +44 (1332) 247808 Fax. +44 (1332) 247928 Brian Justice wrote in message news:... > VxWorkers, > > > Anyone have anything to share regarding UDP transmissions locking > the IP stack in VxWorks (5.3.1)? We have a Intel based CPU board > with a DEC 21143 ethernet controller. If I xmit UDP packets > out to anyone (more than just a few, a lot of them) the IP > stack locks. Interesting to note though, is that if no one > is receiving them, I seem to have no problems. As a result of > the lockup, I'm seeing ENOBUFS returned from sendto(). I am > not running out of fds and my nw cluster pool is not low > either. The only way to recover is to reboot the board... > I am using SENS. > > Any help much appreciated... > > Brian > > ======================================================================== > Brian K. Justice | E-Mail: bjustice@patriot.net > 1847C Montclair Drive | Phone: (843) 971-0856 > Mount Pleasant, SC 229464 | http://patriot.net/~bjustice > ======================================================================== > From vxwexplo-errs@csg.lbl.gov Mon May 22 08:10:10 2000 From: "Tyler, Paul A (DBY)" Date: Mon May 22 08:10:12 PDT 2000 Subject: Re: memory corruption debugging Hi, I did a little utility using the taskSwitch hooks to examine the memory location in question. This at least tells you which tasks are changing the memory, so you know where to start debugging. VxWorks, Tornado Paul Tyler Measurement Systems Specialist Electronics & Measurement Systems Dept. Rolls-Royce plc. e-mail Paul.A.Tyler@Rolls-Royce.com Tel. +44 (1332) 247808 Fax. +44 (1332) 247928 Venkatesh JS wrote in message news:<8g0vb0$9rd$1@schbbs.mot.com>... > hi vxworkers ! > > I have a memory corruption problem where in a contents of a global function > pointer is getting corrupted by some task. i tried using watch(GDB) command > but GDB says it's not supported on vxWorks target. > > > Can anyone tell me a good method to find out which statement is corrupting > memory ? > > > venkat > > From vxwexplo-errs@csg.lbl.gov Mon May 22 08:56:52 2000 From: "Ashitey Trebi-Ollennu" Date: Mon May 22 08:56:54 PDT 2000 Subject: Motion Profiles This is a multi-part message in MIME format. ------=_NextPart_000_0023_01BFC3CC.3B55D9D0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable I am a novice with RTOS and vxworks. I am currently trying to implement = a motion profile or command generator for an actuator controller in = vxworks. I have implemented a trapezoidal profile so far but I would = like to implement a second order system response profile. If anyone has = a code that can do the above and is prepared to share it with me I will = be very grateful. I basically need a profile with time on the x-axis and = commanded input on the y-axis. Ash. ------=_NextPart_000_0023_01BFC3CC.3B55D9D0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
I am a novice with RTOS and vxworks. I = am currently=20 trying to implement a motion profile or command generator for an = actuator=20 controller in vxworks. I have implemented a trapezoidal profile so far = but I=20 would like to implement a second order system response profile. If = anyone has a=20 code that can do the above and is prepared to share it with me I will be = very=20 grateful. I basically need a profile with time on the x-axis and = commanded input=20 on the y-axis.
 
Ash.
------=_NextPart_000_0023_01BFC3CC.3B55D9D0-- From vxwexplo-errs@csg.lbl.gov Mon May 22 13:03:23 2000 From: "Patrick T. Pinkowski" Date: Mon May 22 13:03:25 PDT 2000 Subject: VxWorks and Corelis i960 source level debugger Has anyone used the Corelis i960 source level debugger JTAG interface hardware to debug VxWorks source code? Specifically, the Corelis i960 source level debugger requires "relocatable external microcode" to be placed in the target memory space, and then executed. Has anyone implemented the relocatable microcode in the bootrom? From the Release Notes: --------------------------- In order to use EMDT/960, relocatable external microcode must be placed somewhere in the target memory space below address FFFF0000H. The microcode is in the UCODE directory following installation and provided in 2 formats, COFF and Intel Hex. To install the microcode in a target, do the following: 1. Place the external microcode in the target memory space. Using the relocatable COFF file, the microcode can be linked into a user application. The Intel Hex file can be used with a smart device programmer. The external microcode must be placed at a address that is aligned to a 16-byte boundary (i.e., the least significant nibble in the address must be 0). 2. Set bit 6 of word 3 of the IBR. This is PMCON byte #3 at address FEFFFF3CH. 3. Put the address of the external microcode + 8 in word 6 of the IBR. This is checksum word 0 at address FEFFFF48H. For example, if the external microcode is placed at address FE000000H, then at address FEFFFF48H use the value FE000008H. The microcode is only used by EMDT/960 and is otherwise transparent to the operation of the target. A program which shows an example of this is provided in the 960EX directory following installation. Executing the self-extracting executable will produce the source files for the program, the linker command file, a DOS batch file and a ROM image in Intel Hex format. This program is designed for use on the Cyclone Microsystems IQ80960RP evaluation board. A ROM image in Intel Hex format is provided that can be used to program flash memory device U4 on the evaluation board. The program includes all of the system memory structures necessary for processor initialization. The external microcode required by EMDT/960 is also linked in. The program performs some initialization of the memory control registers and then goes into a small loop where patterns are flashed on the LEDs. In the MKTARGET command, the i960RP target type is now supported. For the i960 RP processor, this target type should be used instead of i960JF. ---------------------------------------------------------------------------- Thanks in advance, Pat -- -------------------------------------------------- /\ Patrick T. Pinkowski, Senior Engineer /##\ ptp@mclean.sparta.com /####\ 703.448.1683 x228 /####/-- /####/---- SPARTA, Inc. -\####\__--- 7926 Jones Branch Drive, Suite 900 ---\##\ /---- McLean, Virginia 22102 ----\##\/\---- (703)448-0210 (Main) ----\####\-- (703)893-5494 (Facsimile) ----\####/ http://www.mclean.sparta.com ---/###/ -/###/ SPARTA \##/ ~~~~~~ \/ Pride In Performance From vxwexplo-errs@csg.lbl.gov Tue May 23 04:03:19 2000 From: Vxworks Exploder Date: Tue May 23 04:03:22 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue May 23 04:03:16 PDT 2000 Subject: Re: UDP transmit problem Subject: chinese i-ching forecast anything http://iching.126.com Subject: Re: T2: including .h files from other directory. How ? Subject: Re: Scheduling Problem Subject: Re: T2: including .h files from other directory. How ? Subject: Can I use CORBA and still make the program POSIX compliant? Subject: SPI Driver for 8260 available? Subject: sa1100IntrCtl.c bug. Subject: Re: need a loop within a makefile Subject: Re: image download problem Subject: Re: Can I use CORBA and still make the program POSIX compliant? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: UDP transmit problem Date: Mon, 22 May 2000 14:05:34 -0400 From: "Michael J. Taskoski" Organization: Lockheed Martin Electronics + Missles Message-ID: <392976ED.4B9EC0F8@lmco.com> References: This sounds like a similar problem I once had with VxWorks running on a PowerPC. The fix came from VxWorks in the form of a "JumboPack" software patch that fixed this problem along with a number of other problems associated with the 604 BSP. The problem showed up when transmitting huge amounts of data across the TCP/IP drivers. Have you tried talking w/ VxWorks support in regards to making sure you're BSP is the latest and greatest and all that? Good Luck, - -- Mike Brian Justice wrote: > VxWorkers, > > Anyone have anything to share regarding UDP transmissions locking > the IP stack in VxWorks (5.3.1)? We have a Intel based CPU board > with a DEC 21143 ethernet controller. If I xmit UDP packets > out to anyone (more than just a few, a lot of them) the IP > stack locks. Interesting to note though, is that if no one > is receiving them, I seem to have no problems. As a result of > the lockup, I'm seeing ENOBUFS returned from sendto(). I am > not running out of fds and my nw cluster pool is not low > either. The only way to recover is to reboot the board... > I am using SENS. > > Any help much appreciated... > > Brian > > ======================================================================== > Brian K. Justice | E-Mail: bjustice@patriot.net > 1847C Montclair Drive | Phone: (843) 971-0856 > Mount Pleasant, SC 229464 | http://patriot.net/~bjustice > ======================================================================== --------------------------- Newsgroups: comp.os.vxworks Subject: chinese i-ching forecast anything http://iching.126.com Date: 23 May 2000 00:09:28 GMT From: Organization: The news server of Changzhou TeleCom Message-ID: <8gci7o$315h$489@news.cz.js.cn> provided forecast Fortune/Stock Exchange/Contract Signing/Partnership/Official/Sport Race/ Lawsuit / Weather/ Calamity and more..... http://iching.126.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: T2: including .h files from other directory. How ? Date: Mon, 22 May 2000 18:16:00 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3929DBD0.B0B4E855@gv.net> References: Reply-To: mkesti@gv.net Albert Hofkamp wrote: >This afternoon, I tried to convince the T2 IDE to find .h files in >another directory than where the .cc/.cpp file are. >In a command-line I can solve the entire problem in less than 5 seconds, >but I cannot specify where .h files can be found. See section 4.5 of the Tornado User's Guide. You need to set the parameters of the "C/C++ Compiler" tab, much as you would on a command line. >The system understands system files, and .h files in the same dir as the >.cpp files, but finding them in another dir seems quite impossible. >Adding them explicitly to the project does not help, and neither >does changing the build rule of C/C++ for some weird reason. I have found instances in T2 where one must click on "Apply" or "Set" as well as "OK" changed parameter's values to stick. >Changing the general build rule for an object file also seems not >possible. That's correct. Projects, not compiled objects, have build rules. >After fighting with the IDE for 4 hours for such a trivial task, I am >quite convinced that IDE's for programming are bad by definition. I've been known to say that kind of thing, too, but find that once one learns his way around one can use T2 quite efficiently. If you're truly "stuck with it", you'll want to get comfortable with it! - -- ======================================================================== Michael Kesti | "And like, one and one don't make | two, one and one make one." mkesti@gv.net | - The Who, Bargain --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Scheduling Problem Date: 23 May 2000 02:24:12 GMT From: Diamand@btinternet.com (Luke Diamand) Organization: Customer of Planet Online Message-ID: References: <8gb8eq$7dr4@overload.lbl.gov> I believe there's a recentish patch to Tornado II which fixes a bug whereby after an interrupt, the _lowest_ priority task gets to be run. HTH! Luke Diamand In article <8gb8eq$7dr4@overload.lbl.gov>, Uygar DOYURAN wrote: > >We are using VxWorks over MPC 860. > >Rarely, we experience a scheduling problem. The most prior task in our >system pends on a semaphore which is released by an ISR upon occurrence of >some event. > >We observe that, sometimes, although the ISR releases the semaphore, the >task is scheduled about 2 milliseconds later. > >We discovered that if we replace (to debug this problem) the 'taskLock' >function by just a 'return from function' statement, this problem does not >occur. > >Do you have any idea 'where in the kernel and why' it is required to disable >scheduling? What can we be doing wrong to introduce such a long delay for >task scheduling? > >Thanks in advance for your concern. > >Uygar DOYURAN >Electrical & Electronics Engineer, Msc. >ASELSAN Electronics Industry, Inc - -- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: T2: including .h files from other directory. How ? Date: 23 May 2000 09:00:29 GMT From: johan.borkhuis@peektraffic.nl Organization: XS4ALL Internet BV Message-ID: <8gdhbd$egk$1@news1.xs4all.nl> References: Albert Hofkamp wrote: : Hello, : This afternoon, I tried to convince the T2 IDE to find .h files in : another directory than where the .cc/.cpp file are. : In a command-line I can solve the entire problem in less than 5 seconds, : but I cannot specify where .h files can be found. Have a look at my VxWorks page. There is a way to get around this problem. Groeten, Johan - -- o o o o o o o . . . ___________________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | johan@borksoft.xs4all.nl | >(________|__|_[_________]_|________________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks page: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Can I use CORBA and still make the program POSIX compliant? Date: Fri, 19 May 2000 18:00:38 GMT From: "Ron S. Ahn" Organization: OR Concepts Applied Message-ID: Hi I really could not figure out if using CORBA would break the POSIX compliance... Please help. Ron Ahn OR Concepts Applied --------------------------- Newsgroups: comp.os.vxworks Subject: SPI Driver for 8260 available? Date: Tue, 23 May 2000 12:01:44 +0200 From: Matthias Gottwald Organization: Siemens AG Message-ID: <392A5708.574398E1@icn.siemens.de> Hi folks, does anybody have a SPI driver for the PowerQUICC II (8260) working under VxWorks. Best regards Matthias. --------------------------- Newsgroups: comp.os.vxworks Subject: sa1100IntrCtl.c bug. Date: 19 May 2000 14:25:26 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: Sender: bpringle@DeadDuck Does anyone know if there is an SPR for the StrongARM interrupt controller? If you have an interrupt on GPIO-0, the interrupt level is 0. In the intDisable, intEnable functions, STATUS sa1100IntLvlEnable(int level) { int key; if (level < 0 || level > SA1100_INT_NUM_LEVELS) return ERROR; /* set bit in enable mask */ if(level > 0) { /* !!! I think this is wrong. !!! */ key = intLock (); sa1100IntLvlEnabled |= (1 << level); intUnlock (key); } sa1100IntLvlChg (-1); /* reset current mask */ return OK; } STATUS sa1100IntLvlDisable(int level) { int key; if (level < 0 || level > SA1100_INT_NUM_LEVELS) return ERROR; /* clear bit in enable mask */ if(level > 0) { /* !!! I think this is wrong. !!! */ key = intLock (); sa1100IntLvlEnabled &= ~(1 << level); intUnlock (key); } sa1100IntLvlChg (-1); /* reset current mask */ return OK; } The `if(level>0)' line in sa1100IntrCtl.c does not allow GPIO zero line to be attached to interrupts. Removing the condition seems to fix the problem. I was just wondering if there are any reprocussions. Have a good weekend, Bill - -- Who is Kibo? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: need a loop within a makefile Date: Wed, 17 May 2000 10:04:18 GMT From: Aitor Garay Organization: Clientes_Teleline Message-ID: <39226F6C.EA46B7FC@alcad.net> References: Hi there Markus!, > Is it possible to check if a file or directory exists (without an error > message) and due to the result do something? I need a rule to rebuild a list > of projects. Some of the projects exists locally (they should be rebuild) > and these where no directory exists should be taken from the network and > after that, all of them should be linked together. Quite simple, I think > (just joking). What you need is some kind of workspace overriding. Some people also calls them working environments. They are usually used with a configuration management tool ( like RCS, o CVS, o ClearCase...). If you are using GNU make you can use "vpath" directives to have "make" search several directories automatically to find a dependency. Other "make" tools may have a similar feature. AITOR --------------------------- Newsgroups: comp.os.vxworks Subject: Re: image download problem Date: Thu, 18 May 2000 10:11:56 +0530 From: "Venkatesh JS" Organization: Motorola Message-ID: <8fvs6n$df3$1@schbbs.mot.com> References: <8ftlgo$kiv$1@schbbs.mot.com> <8fue85$9e3$1@schbbs.mot.com> hi, all optimization is off when i build the image,..even for the bootrom image also. venkat Peter Luscher wrote in message ... >Does your card have more than 16Mb on it? (I assume so...) >There is a known problem with certain 750 BSPs. Try turning off >optimization, rebuilding your BSP and re-booting. > >P. > >"Venkatesh JS" wrote in message >news:8fue85$9e3$1@schbbs.mot.com... >> hi, >> >> In my case the difference between RAM_HIGH_ADDR and RAM_LOW_ADDR is 16 MB >> which is well beyond the image size of 2.5 MB. >> >> venkat >> >> Peter Luscher wrote in message ... >> >If you are downloading your image using the VxWorks bootrom, your >> downloaded >> >image must not be larger than (RAM_HIGH_ADRS - RAM_LOW_ADRS). If it is, >> you >> >must re-build your bootrom (after increasing the value of RAM_HIGH_ADRS >in >> >config.h AND in the Makefile), re-flash your bootrom and re-boot. >> > >> >P. >> > >> >"Venkatesh JS" wrote in message >> >news:8ftlgo$kiv$1@schbbs.mot.com... >> >> hi vxworkers, >> >> >> >> I have a image download problem on PPC750 target. my image size is >> around >> >> 2.5 MB, and target takes a long time to download, so long that i never >> get >> >> messages like "Starting at address 0x1000...WDB Ready" which mean that >> >> download is successful. During this period if i shutdown the TFTP >server >> >on >> >> the host, then above messages appear on hyper terminal. >> >> >> >> Is there any default size for vxworks image that can be downloaded on >to >> >> target ?.. >> >> >> >> - >> >> venkat >> >> >> >> >> > >> > >> >> > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can I use CORBA and still make the program POSIX compliant? Date: Sat, 20 May 2000 19:21:47 GMT From: bill.beckwith@realtime-corba.com (Bill Beckwith) Organization: Objective Interface Message-ID: <8F3A9CBBFbeckwbnewscom@sigma> References: Ron S. Ahn wrote in : >Hi > >I really could not figure out if using CORBA would break the POSIX >compliance... Hi Ron, The CORBA standard comes from OMG. The CORBA standard is not a replacement for the POSIX standard. The CORBA standard specifies standard APIs and interoperability for distributed object systems. The Real-Time CORBA specification refers to the POSIX standard and was designed to be "compatible" (may be "co-exist gracefully" is better). - -- Bill --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue May 23 06:33:49 2000 From: Mohan Kumar Date: Tue May 23 06:33:51 PDT 2000 Subject: Time Stamp Hello, I have two issues related to VxWorks: 1. Have anyone used Auxiliary Clock for Time Stamp Driver ? Will there be any problem if we use the same. 2. How to disable the CACHE for DPRAM in PPC740 environment. Any takers...! Thanks for your time in advance. __________________________________________________ Do You Yahoo!? Send instant messages & get email alerts with Yahoo! Messenger. http://im.yahoo.com/ From vxwexplo-errs@csg.lbl.gov Wed May 24 03:38:40 2000 From: Suresh T Date: Wed May 24 03:38:43 PDT 2000 Subject: vxWorks doubt Hi, We face a problem in allocation and freeing up of memory using memPartAlloc and memPartFree. Initially we create a memory partition with id say X We allocate and use the memory using memPartAlloc and finally we free up the memory using memPartFree. While freeing up the memory we get an error message "invalid block ***** in partition *****". We found that the address being allocated using memPartAlloc and the address being freed using memPartFree are same. And the Mempart id from where mempartalloc and mempartfree are done is also same. Can anyone throw light on the possible scenarios when this could happen. Thanks and Regards Suresh From vxwexplo-errs@csg.lbl.gov Wed May 24 04:05:03 2000 From: Vxworks Exploder Date: Wed May 24 04:05:05 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed May 24 04:05:00 PDT 2000 Subject: Re: Can I use CORBA and still make the program POSIX compliant? Subject: Re: T2: including .h files from other directory. How ? Subject: Re: CPU Usage for power pc Subject: evtRecv for WindView (tornado 2.0) Subject: Re: sa1100IntrCtl.c bug. Subject: elfToBin broke for PowerPC? Subject: Re: image download problem Subject: END NeverNeverLand Subject: Re: SNMPD for VxWorks Subject: In Vxwork, help : recvfrom non-blocking method!! Subject: Re: In Vxwork, help : recvfrom non-blocking method!! ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Can I use CORBA and still make the program POSIX compliant? Date: Fri, 19 May 2000 15:45:29 -0400 From: "Bill Pritchett" Message-ID: <8g45ks$qkh$1@bob.news.rcn.net> References: What is the impetus for being POSIX compliant? "Ron S. Ahn" wrote in message news:ajfV4.5277$T41.124964@newsread1.prod.itd.earthlink.net... > Hi > > I really could not figure out if using CORBA would break the POSIX > compliance... > > Please help. > > Ron Ahn > OR Concepts Applied > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: T2: including .h files from other directory. How ? Date: Tue, 23 May 2000 10:45:55 -0400 From: MJ McHugh Organization: Rancho McHugh Message-ID: <392A99A3.8BF3840@attglobal.net> References: Reply-To: mchug06@attglobal.net Albert Hofkamp wrote: > Hello, > > This afternoon, I tried to convince the T2 IDE to find .h files in > another directory than where the .cc/.cpp file are. > In a command-line I can solve the entire problem in less than 5 seconds, > but I cannot specify where .h files can be found. > > The system understands system files, and .h files in the same dir as the > .cpp files, but finding them in another dir seems quite impossible. > Adding them explicitly to the project does not help, and neither > does changing the build rule of C/C++ for some weird reason. > What I have found is that you must change the build rule for C/C++ before adding the source file that references header files in other directories. For example, if you have file.c that includes /home/.../include/file.h, and you add file.c to the project before you add "-I/home/.../include/" to the build rule, T2 generates an external dependency of "./file.h", that is, it assumes that file.h resides in the local project directory since it doesn't know where else to look for it. If you've already added the source file to the project, first change the build rule to add in the desired include paths, then remove the source file from the project. Regenerate dependencies and verify that the offending .h has been removed from the external dependencies list. Then, add the source file back into the project and rebuild the dependencies. Now, after the header file has been added back into the external dependencies list, right click on it and select properties. If everything was successful, the full path for the header file will be shown at the bottom of the properties box (in my example it would show: "/home/.../include/file.h"). If it still says "./file.h", you're going to get the "no rule to make file.h" error again. I discovered this when I had two source files including the same header file. The header file resided in the same directory as source1.c, and that included OK. The second source file was in a different directory and compiling it resulted in the dreaded "no rule" error. I was confused since the first file compiled ok. Then I noticed that there were *two* instances of file.h in the external dependencies list, one whose properties showed the correct path, the other which showed "./file.h". I was not able to simply regenerate dependencies. I had to first remove the source file from the project, fix the C/C++ rule, then add the source file back in. I'm running T2 on Win98, BTW. We're having similiar problems with nested includes, but we haven't yet applied the method I just described. > > Changing the general build rule for an object file also seems not > possible. > Nope, just the build rule for the project, at least from the IDE. Hope this helps. Mary --------------------------- Newsgroups: comp.os.vxworks Subject: Re: CPU Usage for power pc Date: Fri, 19 May 2000 18:36:25 -0400 From: "Sohacki, Timothy [NCRTP:3M24:EXCH]" Organization: Nortel Networks Message-ID: <3925C1E9.FB7EB3A4@americasm01.nt.com> References: <8evnvl$kc2$1@bob.news.rcn.net> <9%YQ4.138$wY5.114688067@speed.city-net.com> Reply-To: sohacki@nortelnetworks.com A very nice solution, I think. To combat the "one second problem," on the PPC, you could read the the time base register before and after to determine exactly how long your "one second" was. Regarding the idle task ... VxWorks doesn't have one. If I remember correctly, there is an idle flag that is set when the kernel is idle, but that's all. Instead, you'll need to spawn a priority 255 task with a body something like: void idleTask() { int iCnt = 0; while (1) { iCnt++; } } You may want to play with the body of the while loop. hopper99 wrote: > the technique used by Jean Labrosse in his micro-C OS book is: > > There needs to be a counter that is incremented when the Idle task is > running. This is the part someone more familiar with the OS needs to > answer - is this facility provided? Lets call this counter iCnt. > > Once you have this then you are all set. When your application comes up, the > first thing it should do is zero out iCnt. Then sleep for 1 second (or some > fixed time period). Record the new value of iCnt as iCntmax. This is the > 100% Idle value. > > Now start your other tasks. Whenever you want to know how busy your CPU is, > zero iCnt again and sleep for another second. The percentage of CPU being > used is (1 - (iCnt / iCntmax)) * 100. > > Note that if the system is busy and you want a very accurate measurement, > you'll need to do this from a high priority task (preferable the highest > priority), otherwise if the other tasks are keeping the CPU busy, your task > won't wake up after one second and the measurement will be wrong. This > could still be the case, even from a high priority task, if you are > servicing a lot of interrupts. --------------------------- Newsgroups: comp.os.vxworks Subject: evtRecv for WindView (tornado 2.0) Date: Tue, 23 May 2000 19:21:09 GMT From: chuckallis@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8gelmm$u1k$1@nnrp1.deja.com> Hello, To begin with my host machine is a Sun machine running Solaris 2.6. My target is a 604 PowerPC but I'm pretty sure that isn't significant in this case. I've used Tornado 1.0.1 WindView on a previous project. On my new project we are using Tornado 2 and I noticed there are some significant changes in WindView. I am familiar with the evtRecv application and before all you needed to do was do a wvHostInit (with the host that was running evtRecv) and wvEvtLogEnable. At that point the target would begin sending data to the host with evtRecv running on it. wvEvtLogDisable would then stop the flow of information. But in the Tornado 2.0, wvHostInit, wvEvtLogEnable, and wvEvtLogDisable do not exist anymore. I saw wvOn and wvOff, but I can't seem to get them to work. If anyone knows the proper way to get evtRecv to begin collecting the WindView data in the Tornado 2.0 version (and examples tend to help me a lot ;-), I would appreciate it immensly. Thank you for your time, Chuck Allis P.S. Oh, I would try WindRiver support (since we're paying for it), but the last question I asked over 2 weeks ago never got answered :-( Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: sa1100IntrCtl.c bug. Date: 23 May 2000 15:38:56 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <32jV4.1137$s4.90925@petpeeve.ziplink.net> Sender: bpringle@DeadDuck >>>>> "Steve" == Steve Dillen writes: Steve> Howdy, I have been looking at this source code as well, I don't Steve> know if I fully understand it, but according to documented code Steve> in syslib.c with the sa1100IntLvlMask array declaration, level Steve> 0 indicates ALL interrupts are disabled. This implies that the Steve> interrupt levels start at 1 _not_ 0. Steve> If this is the case then the below code in question is correct Steve> for the line you are indicated, but not the other line Steve> sa1100IntLvlEnabled |= ( 1 << ( level - 1 ) ); Steve> And similarily in the Disabled routine. Steve> I haven't looked into how the .h declarations for Steve> interrupt levels are based, or how it will be affected. The level is open to interpretation. Do you wish to mask all levels below this interrupt or all levels below and including this interrupt. I believe that the levels are a little synthetic for the StrongArm targets. According to my interpretation of the code, the values in the ICPR are given a level according to bit placement. This gives the GPIOs the highest priority in the system. You can muck around with the `sa1100IntLvlMask' array and the check for bits in the code to make any type of level priority you wish. Anyways, it is wrong to have, sa1100IntLvlEnabled |= ( 1 << ( level - 1) ); This value is written to the ICMR. You will have masked the wrong interrupt source. I just think that the conditional for `level > 0' needs to be removed. Anyways, I am quite confident that it is wrong. I was just reporting it hear because I find reporting an SPR something like living through the inquisition. bye, Bill - -- So, Reagan is Nicaraguan? Let me tell you something, you weakling, the Martians are coming. --------------------------- Newsgroups: comp.os.vxworks Subject: elfToBin broke for PowerPC? Date: Tue, 23 May 2000 16:22:19 -0400 From: "Gutknecht, Bill [NC1:3T62:EXCH]" Organization: Nortel Networks Message-ID: <392AE87B.989E0748@nortelnetworks.com> I'm trying to build the vxWorks.st_rom image, but something is screwy. It looks like when the vxWorks image is converted to binary before compression, elfToBin leaves about 0x4000 of garbage (looks like an ELF header) on the binary image. Of course, it gets inflated properly at boot time, but the ELF header doesn't run very well. Has anyone seen this before? Bill - -- Bill Gutknecht "If I die, I must go before [Crom] and he billag@nortelnetworks.com will ask me 'What is the Riddle of Steel?' Embedded Software Engineer If I do not know it, he will cast me out Nortel Networks of Valhalla and laugh at me ..." - Conan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: image download problem Date: Tue, 23 May 2000 21:34:36 GMT From: "Peter Luscher" Organization: @Home Network Canada Message-ID: References: <8ftlgo$kiv$1@schbbs.mot.com> Hi, Thanks for the addition to my omission! To clarify: The bootrom requires that you change RAM_HIGH_ADRS and RAM_LOW_ADRS in two places (the config.h and Makefile in your BSP directory). This image can ~not~ be configured using the Tornado 2.0 IDE (although the 'command line' build can be invoked from the project facility using the menu option Build --> Build Boot Rom...) To change RAM_HIGH_ADRS & RAM_LOW_ADRS for the subsequently downloaded image ~using the project facility~, open your workspace, click on the 'Builds' tab, open up the project so that you can see the build spec(s). Now, right click on the build spec (eg. 'default') --> properties --> Macros --> modify RAM_LOW_ADRS and RAM_HIGH_ADRS here. P. "Hosehead Jones 2nd-1/2" wrote in message news:jf3kisgllj7h7t590nmfkq3edl53kfulsr@4ax.com... > "Peter Luscher" took time out of his/her busy > schedule to pen: > > >If you are downloading your image using the VxWorks bootrom, your downloaded > >image must not be larger than (RAM_HIGH_ADRS - RAM_LOW_ADRS). If it is, you > >must re-build your bootrom (after increasing the value of RAM_HIGH_ADRS in > >config.h AND in the Makefile), re-flash your bootrom and re-boot. > > > >P. > > > Just to chime in - > > Keep in mind that RAM_HIGH_ADRS and RAM_LOW_ADRS are in *three* > separate places in T2, and they all have to match. One is in the > Makefile for your BSP, the second is in config.h, and the third is in > the T2 project file for your kernel. > > Unfortunately, I can't recall exactly where it is in the project file, > so I can't point right at it now. (I seem to recall it's pretty well > buried.) However, you can search in the project for RAM_HIGH_ADRS by > (I believe) right-clicking in the VxWorks tab (on Windows, natch; not > sure how to do this on other platforms) and selecting "Find Object". > > Regardless, this may be nowhere near your problem. All I know is, it > took me a devilishly long time to find that third instance, so I'd > hate to see someone else go through all that. > > ------------------ > markm@ctilidar.com > --------------------------- Newsgroups: comp.os.vxworks Subject: END NeverNeverLand Date: Tue, 23 May 2000 17:17:02 -0700 From: "Charles Krinke" Message-ID: I guess I am searching for a few ideas. I am adding a second END driver to configNet.h (it is compiled and linked from the appropriate target\src\drv\end directory). When I startup, the first driver in the endtbl structure loads its EndLoad routine just fine, but the second one (with a different unit number), comes back with a NULL pointer from the muxDevLoad(). The first is the dec21x4xEnd and the second is the Ln97xEnd driver for the AMD79C973. I suspect the problem is in my understanding and not with the driver. Any lamblasting will be accepted with humility, my brain is fried and I am going home for the evening. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: SNMPD for VxWorks Date: Wed, 24 May 2000 00:22:04 GMT From: Marjorie Krueger Organization: Vixel Message-ID: <392B2043.17AA2D4C@NOSPAMvixel.com> References: You can view the manual on line at WindRiver's website thru the WindSurf interface. From the manual: $WIND_BASE/target/src/snmpv1/agent/ The agent subdirectory consists of several source files, including: snmpIoLib.c This file contains routines which determine the operation of the SNMP agent. All SNMP agent behavior, including input and output, community validation, and trap generation is controlled by the routines in this library. This file is provided as source code; modify it to tailor the agent to your specific needs. See 5.2 Configuring snmpIoLib. From 5.2: snmpIoLib contains routines which can be customized by an agent writer for specific needs. The following sections discuss routines which are present in the file $WIND_BASE/target/src/snmpv1/agent/. For more information on an individual routine, see its corresponding manual page entry in F. SNMP Reference. "Robert E. Laughlin" wrote: > ello; > > I have inherited part of a project that uses VxWorks version 5.2 > and includes WindNet SNMP 1.0 Beta-2. I need to change a couple of the > settings for the SNMPD on the system. The documentation that we got does > not include "Part 2" of the "WindNet SNMP VxWorks Optional Product > Supplement". I do not know if we just did not get it or if Part 2 got > lost, but that appears to be where all of the details of how to configure > SNMP for different "community" strings. > > If any one knows where I can get a copy of Part 2 or can tell me > how to change the community strings, I would sure appreciate it. > > Thanks Bob > > Robert E. Laughlin SPAWARSYSCEN SD D4122 > An interesting thing happens when you hit a security-illiterate person with a > tool that makes them realize how often they get probed and examined by hackers: > the get furious. > These opinions are mine. I do not speak for SPAWARSYSCEN SD. > email bel@spawar.navy.mil Public Key ID = AE8294B9 > Key fingerprint = 8747 FC71 49C7 59D0 74F8 CC2B E1D8 8AFB AE82 94B9 --------------------------- Newsgroups: comp.os.vxworks Subject: In Vxwork, help : recvfrom non-blocking method!! Date: Thu, 25 May 2000 00:13:32 +0900 From: "Á¤Áø±º" Organization: Korea Telecom Message-ID: <8gfgrf$ffm$1@news2.kornet.net> Hi, All. Would help me? Let me know "recvfrom non-blocking method" in the VxWorks. Thanks. Jinney. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: In Vxwork, help : recvfrom non-blocking method!! Date: Wed, 24 May 2000 14:19:53 +0100 From: Frank Liu Organization: Motorola Semiconductor Products Sector Message-ID: <392BD6F9.FB261333@motorola.com> References: <8gfgrf$ffm$1@news2.kornet.net> Reply-To: F.Liu@motorola.com Using following function to set timeout time int read_timeo( int sFd, int sec) //sFd is the socket, sec is the timeout time { fd_set rset; struct timeval tv; FD_ZERO(&rset); FD_SET (sFd, &rset); tv.tv_sec =3D sec; tv.tv_usec =3D 0; return (select (sFd +1, &rset, NULL, NULL, &tv)); } using methods: if ((read_timeo(sFd, 15) =3D=3D 0) //wait 15 seconds to make SFd readable= printf ("time out"); else { recvfrom(...); ... } "=C1=A4=C1=F8=B1=BA" wrote: > = > Hi, All. > = > Would help me? > = > Let me know "recvfrom non-blocking method" in the VxWorks. > = > Thanks. > = > Jinney. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 24 06:48:46 2000 From: Bruce Rowen Date: Wed May 24 06:48:48 PDT 2000 Subject: Re: vxWorks doubt the vxWorks Users Group Exploder wrote: > > Submitted-by vxwexplo-errs@csg.lbl.gov Wed May 24 03:38:40 2000 > Submitted-by: Suresh T > > Hi, > > We face a problem in allocation and freeing up of memory using memPartAlloc > and memPartFree. > > Initially we create a memory partition with id say X > We allocate and use the memory using memPartAlloc and finally we free up > the memory using memPartFree. > While freeing up the memory we get an error message > > "invalid block ***** in partition *****". > > We found that the address being allocated using memPartAlloc and the > address being freed using memPartFree are same. And the Mempart id from > where mempartalloc and mempartfree are done is also same. > > Can anyone throw light on the possible scenarios when this could happen. > > Thanks and Regards > Suresh > If you write past the end of your allocated memory chunk you will corrupt some tables that vxWorks uses to manage the memory block. This makes it very hard to disallocate. -- ------------------------------------------------------------------------------- Bruce Rowen National Radio Astronomy Observatory Scientific Programmer Array Operations Center browen@aoc.nrao.edu P.O. Box O Socorro, NM 87801 (505)835-7329 (505)835-7000 ------------------------------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Wed May 24 10:01:07 2000 From: "abhijit lahiri" Date: Wed May 24 10:01:09 PDT 2000 Subject: VxWorks stack trace Hi All, I am working on QED-RM7000, a MIPS processor. My idea is to get the stack trace of a task which has generated an exception. ( As shown by "tt") I am able to capture the details ( WIND_TCB etc. ) of the task by adding a hook to the default vxWorks exception handler. But how do I generate the stack trace ( basically, details of nested subroutine calls.) of the task ? As per the WIND_TCB and other data structures, frame pointer, stack pointer, and stack size are available for a task as a whole. I believe that a way must be there to unwind the stack. But i am unable to find it immediately. If anyone has any inputs regarding this, please reply. Your help will be very useful for me. regards, Abhijit From vxwexplo-errs@csg.lbl.gov Wed May 24 10:02:32 2000 From: "abhijit lahiri" Date: Wed May 24 10:02:35 PDT 2000 Subject: VxWorks stack trace Hi All, I am working on QED-RM7000, a MIPS processor. My idea is to get the stack trace of a task which has generated an exception. ( As shown by "tt") I am able to capture the details ( WIND_TCB etc. ) of the task by adding a hook to the default vxWorks exception handler. But how do I generate the stack trace ( basically, details of nested subroutine calls.) of the task ? As per the WIND_TCB and other data structures, frame pointer, stack pointer, and stack size are available for a task as a whole. I believe that a way must be there to unwind the stack. But i am unable to find it immediately. If anyone has any inputs regarding this, please reply. Your help will be very useful for me. regards, Abhijit From vxwexplo-errs@csg.lbl.gov Wed May 24 10:15:16 2000 From: "Michael Lawnick" Date: Wed May 24 10:15:21 PDT 2000 Subject: Re: elfToBin broke for PowerPC? Hi Bill, > 'm trying to build the vxWorks.st_rom image, but something is screwy. It > looks like when the vxWorks image is converted to binary before > compression, elfToBin leaves about 0x4000 of garbage (looks like an ELF > header) on the binary image. Of course, it gets inflated properly at boot > time, but the ELF header doesn't run very well. > > Has anyone seen this before? > look at WindSurf, there you can find that you should contact your local FAE for a new elfToBin. HTH > Bill MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== From vxwexplo-errs@csg.lbl.gov Wed May 24 11:48:56 2000 From: mmenge@dspt.com Date: Wed May 24 11:48:58 PDT 2000 Subject: Re: VxWorks stack trace just put a tt(0) call in the exception handler Tornado vxWorks From vxwexplo-errs@csg.lbl.gov Wed May 24 19:08:05 2000 From: Fred Roeber Date: Wed May 24 19:08:07 PDT 2000 Subject: Re: VxWorks stack trace On Wed, 24 May 2000, abhijit lahiri wrote: > I am working on QED-RM7000, a MIPS processor. My idea is > to get the stack trace of a task which has generated an exception. > ( As shown by "tt") > > I am able to capture the details ( WIND_TCB etc. ) of the task > by adding a hook to the default vxWorks exception handler. > But how do I generate the stack trace ( basically, details of > nested subroutine calls.) of the task ? > > As per the WIND_TCB and other data structures, frame pointer, > stack pointer, and stack size are available for a task as a whole. > > I believe that a way must be there to unwind the stack. But i am > unable to find it immediately. Abhijit, You say you want the output provided by the "tt" function. It just prints it's information to stderr. How about capturing the output to stderr while calling "tt"? I remember trying to figure out how tt worked and finding that most of the work was done by the trcStack function. This function is "architecture specific" and can get pretty hairy. Your using a MIPS which I think has the architecture where not all procedure calls have explicit stack frames. I don't think I would want to have to write the code to unwind the stack myself. There is no man page for the trcStack function but there is an include file (trcLib.h) that indicates the routine can be passed a custom print function. Maybe you could ask WRS support if there is any way of finding out how that routine works so you can pass it your own routine to capture the info you want. Just a suggestion. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Wed May 24 20:04:44 2000 From: Fred Roeber Date: Wed May 24 20:04:46 PDT 2000 Subject: Re:evtRecv for WindView (tornado 2.0) On Wed, 24 May 2000, Chuck Allis chuckallis@my-deja.com wrote: > Subject: evtRecv for WindView (tornado 2.0) > I've used Tornado 1.0.1 WindView on a previous project. On my new > project we are using Tornado 2 and I noticed there are some significant > changes in WindView. > > I am familiar with the evtRecv application and before all you needed to > do was do a wvHostInit (with the host that was running evtRecv) and > wvEvtLogEnable. At that point the target would begin sending data to > the host with evtRecv running on it. wvEvtLogDisable would then stop > the flow of information. > > But in the Tornado 2.0, wvHostInit, wvEvtLogEnable, and wvEvtLogDisable > do not exist anymore. I saw wvOn and wvOff, but I can't seem to get > them to work. The way WindView is used and controlled is definitely very different in T2 than T1. The bottom line is that the intent is that you use WindView through the IDE supported interface. When you run things through the Tornado GUI, it has smarts built into it to actually dynamically patch parts of your target program to get data collection to work. It's all pretty easy if you can live with what the GUI supports (which as far as I can tell doesn't include target initiated collection starting/stopping or data receipt with evtRecv). The code that is provided for wvOn and wvOff (see file target/src/config/usrWindview.c) seems to me to be intended as more of a template than a fully working version of the routines you need to call to do event collection. To get things to work you will probably have to break out the WindView User's Guide and tweak the code. At least I did. I must admit there was a fair amount to "grok" to get it all going and it took a while (more than a couple of hours). Obviously, it can be made to work. For what it's worth. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Thu May 25 04:04:31 2000 From: Vxworks Exploder Date: Thu May 25 04:04:34 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu May 25 04:04:29 PDT 2000 Subject: Re: END NeverNeverLand Subject: PPP over ATM (PPPoA) Subject: Re: Reboot Target from Command Line Subject: Re: VxWorks stack trace ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: END NeverNeverLand Date: 24 May 2000 10:57:57 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: Sender: bpringle@DeadDuck >>>>> "Charles" == Charles Krinke writes: Charles> I guess I am searching for a few ideas. I am adding a second Charles> END driver to configNet.h (it is compiled and linked from the Charles> appropriate target\src\drv\end directory). When I startup, Charles> the first driver in the endtbl structure loads its EndLoad Charles> routine just fine, but the second one (with a different unit Charles> number), comes back with a NULL pointer from the Charles> muxDevLoad(). The first is the dec21x4xEnd and the second is Charles> the Ln97xEnd driver for the AMD79C973. I suspect the problem Charles> is in my understanding and not with the driver. Any Charles> lamblasting will be accepted with humility, my brain is fried Charles> and I am going home for the evening. Hmm, I am about to try the same thing. However, first I have to do DHCP before calling usrNetWorkInit(). Anyways, are you sure that your `load string' for the driver is correct? Ie, the muxDevLoad call is ok. You don't need different unit numbers, I don't think. I thought it would be something like "dec0" and "ln0" for device names. Did you verify that both devices load separately, with the entries that you have in the confignet.h table? hth, Bill - -- Have you ever joined a church where everyone but you can talk with your telephone? Or been forced by your Teletype(tm) to eat bananas? Or used an airplane to travel while joining in an Iroquois ritual? You will. And the company that will bring it to you: AT&T. --------------------------- Newsgroups: comp.os.vxworks Subject: PPP over ATM (PPPoA) Date: Wed, 24 May 2000 18:39:31 +0200 From: "Alon Cohen" Organization: Globaloop Message-ID: <8ggsse$lov$1@news.netvision.net.il> When PPP is encapsulated over ATM, only the 'Protocol' and 'Information' fields are used. The 'flag', 'addr', 'ctrl', 'fcs', and 'closing flag' fields are omitted (from obvious reasons). However, it seems that the vxWorks implementation of PPP hardly ties these fields to the core 'Protocol-Info' sequence. Does anyone have an idea/experience as to how to disregard the unwanted fields ? Any specific entry points to the PPP implementation Input and Output routines ? By the way, the same problem exists with PPP over Ethernet. Thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Reboot Target from Command Line Date: Wed, 24 May 2000 18:13:50 GMT From: DRI Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <392C16D3.E82F60E9@earthlink.net> References: <8gchm8$j0e@qualcomm.com> try reboot 2 from the command prompt (or inside a program). This reboots and clears all memory. Frank Merrow wrote: > 1. I am trying to reboot the target from the command line. (Not from the > Tornado GUI.) > > This ALMOST works: > > windsh -s reboot.txt > > where reboot.txt contains the single work "reboot". The problem is that > instead of rebooting and returning it goes into an infinate loop. I've > tried several things to fix this, but just cannot get is to work . . . > > 2. What I'd really like to control the target with with is Perl . . . is > there ANY HELP out there for that? I've heard rumors, but not actually found > anything. > > 3. What is WRS's reputation for support? So far, my results with them have > been very mixed . . . > > Frank --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks stack trace Date: Thu, 25 May 2000 10:28:24 +0200 From: "Michael St" Organization: NetVision Israel Message-ID: <8giko7$jjk$1@news.netvision.net.il> References: <8ghkco$jd21@overload.lbl.gov> Hi Abhijit ! The question is what Calling Conventions do you work with? EABI, AIX ...? Read about your compiler. abhijit lahiri wrote in message news:8ghkco$jd21@overload.lbl.gov... > > > Hi All, > I am working on QED-RM7000, a MIPS processor. My idea is > to get the stack trace of a task which has generated an exception. > ( As shown by "tt") > > I am able to capture the details ( WIND_TCB etc. ) of the task > by adding a hook to the default vxWorks exception handler. > But how do I generate the stack trace ( basically, details of > nested subroutine calls.) of the task ? > > As per the WIND_TCB and other data structures, frame pointer, > stack pointer, and stack size are available for a task as a whole. > > I believe that a way must be there to unwind the stack. But i am > unable to find it immediately. > > If anyone has any inputs regarding this, please reply. Your help > will be very useful for me. > > regards, > Abhijit > > > > > > > > > > --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu May 25 05:55:50 2000 From: David Anderson Date: Thu May 25 05:55:53 PDT 2000 Subject: RE: comp.os.vxworks newsdigest I've just done a similar thing ... my usrNetwork.c had a #define IP_UNIT_MAX 1 until this was changed to 2 I could only load one END driver. I spent hours searching my driver code for the bug before I found this undocumented feature. Might be the same with yours... David A. > > >>>>> "Charles" == Charles Krinke writes: > > Charles> I guess I am searching for a few ideas. I am adding a second > Charles> END driver to configNet.h (it is compiled and linked from the > Charles> appropriate target\src\drv\end directory). When I startup, > Charles> the first driver in the endtbl structure loads its EndLoad > Charles> routine just fine, but the second one (with a different unit > Charles> number), comes back with a NULL pointer from the > Charles> muxDevLoad(). The first is the dec21x4xEnd and the second is > Charles> the Ln97xEnd driver for the AMD79C973. I suspect the problem > Charles> is in my understanding and not with the driver. Any > Charles> lamblasting will be accepted with humility, my brain is fried > Charles> and I am going home for the evening. > > Hmm, I am about to try the same thing. However, first I have to do DHCP > before calling usrNetWorkInit(). Anyways, are you sure that your > `load string' for the driver is correct? Ie, the muxDevLoad call is > ok. You don't need different unit numbers, I don't think. I thought > it would be something like "dec0" and "ln0" for device names. > > Did you verify that both devices load separately, with the entries that > you have in the confignet.h table? > > hth, > Bill From vxwexplo-errs@csg.lbl.gov Thu May 25 05:55:57 2000 From: David Anderson Date: Thu May 25 05:55:59 PDT 2000 Subject: Multiple END driver load problem. vxWorks , Tornado , vxWorks , Tornado END Driver loading problem: I've just done a similar thing ... >>>my usrNetwork.c had a #define IP_UNIT_MAX 1 CORRECTION: I mean t IP_MAX_UNITS and it occurs just after #ifdef INCLUDE_END until this was changed to 2 I could only load one END driver. I spent hours searching my driver code for the bug before I found this undocumented feature. Might be the same with yours... David A. > > >>>>> "Charles" == Charles Krinke writes: > > Charles> I guess I am searching for a few ideas. I am adding a second > Charles> END driver to configNet.h (it is compiled and linked from the > Charles> appropriate target\src\drv\end directory). When I startup, > Charles> the first driver in the endtbl structure loads its EndLoad > Charles> routine just fine, but the second one (with a different unit > Charles> number), comes back with a NULL pointer from the > Charles> muxDevLoad(). The first is the dec21x4xEnd and the second is > Charles> the Ln97xEnd driver for the AMD79C973. I suspect the problem > Charles> is in my understanding and not with the driver. Any > Charles> lamblasting will be accepted with humility, my brain is fried > Charles> and I am going home for the evening. > > Hmm, I am about to try the same thing. However, first I have to do DHCP > before calling usrNetWorkInit(). Anyways, are you sure that your > `load string' for the driver is correct? Ie, the muxDevLoad call is > ok. You don't need different unit numbers, I don't think. I thought > it would be something like "dec0" and "ln0" for device names. > > Did you verify that both devices load separately, with the entries that > you have in the confignet.h table? > > hth, > Bill David Anderson. mailto:danderson@dtrack.demon.co.uk Disclaimer ---------- This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Data Track Technology Plc +44 1425 271900. ------------------------------------------------------- From vxwexplo-errs@csg.lbl.gov Thu May 25 09:37:52 2000 From: rcw@DynRes.com Date: Thu May 25 09:37:54 PDT 2000 Subject: hi res timer? vxworks tornado I'm sending this on behalf of a friend... the question is: I need access to a high resolution (at least 1MHZ) timer/counter on the Motorola MVME2604. The programmers reference guide, page 2-65, shows that there are 4 TIMER/counters running at 8.25MHZ on the RavenMPIC. The base address of the RavenMPIC is given as 0xfeff0000. The offset in the RavenMPIC register map is 0x1100. Dumping from what I think is the correct location of the RavanMPIC registers in memory yields information out to offset 0x0080 but is full of junk after that. Does anyone know how to read one of the system counters? Any help would be greatly appreciated. I think his ultimate question is how to grab a (any) timer already running on the system, with minimal impact. But why he cannot see the whole Raven register set is an interesting question as well. Answers can be sent direct to him, me or the group. His email is nb at dynres dot com Ron --------------------------------------------- Ronald Wagner Principal Engineer Manager, DRI Driving Simulator Dynamic Research, Inc. +1 310 212 5211 (v) 355 Van Ness Avenue +1 310 212 5046 (f) Torrance, CA USA 90501 rwagner at dynres dot com From vxwexplo-errs@csg.lbl.gov Thu May 25 09:44:44 2000 From: Anthony Le Date: Thu May 25 09:44:47 PDT 2000 Subject: How to make tNetTaskd() to exit without using taskDelete() Hello all, VxWorks spawns a task, tNetTaksd, if defined #define INCLUDE_NET_INIT to manage the network stack. How do I make this task exit nicely so it will free the network resources for me. Regards, Anthony T. Le From vxwexplo-errs@csg.lbl.gov Thu May 25 10:34:56 2000 From: "CICON,JAMES (HP-Vancouver,ex1)" Date: Thu May 25 10:34:58 PDT 2000 Subject: BOOTP / TFTP So we are using BOOTP and TFTP to load our application to our target. The TFTP transfer takes about 40 seconds to load a file of 3.5 MB. It seems that it should be able to go faster because I can load a 3.5 MB file to the target, via windshell, in about 5 seconds. Any ideas on how to speed the TFTP transfer up? Jim Cicon vxWorks From vxwexplo-errs@csg.lbl.gov Thu May 25 12:06:49 2000 From: "CICON,JAMES (HP-Vancouver,ex1)" Date: Thu May 25 12:06:51 PDT 2000 Subject: RE: BOOTP / TFTP vxworks tornado So we are using BOOTP and TFTP to load our application to our target. The TFTP transfer takes about 40 seconds to load a file of 3.5 MB. It seems that it should be able to go faster because I can load a 3.5 MB file to the target, via windshell, in about 5 seconds. Any ideas on how to speed the TFTP transfer up? Jim Cicon From vxwexplo-errs@csg.lbl.gov Thu May 25 12:57:59 2000 From: John Moore Date: Thu May 25 12:58:02 PDT 2000 Subject: Re: hi res timer? > > > > ... > > I need access to a high resolution (at least 1MHZ) timer/counter on the Motorola MVME2604. The programmers reference guide, page 2-65, shows that there are 4 TIMER/counters running at 8.25MHZ on the RavenMPIC. > The base address of the RavenMPIC is given as 0xfeff0000. The offset in the RavenMPIC register map is 0x1100. Dumping from what I think is the correct location of the RavanMPIC registers in memory yields > information out to offset 0x0080 but is full of junk after that. Does anyone know how to read one of the system counters? Any help would be greatly appreciated. > > I think his ultimate question is how to grab a (any) timer already running on the system, with minimal impact. But why he cannot see the whole Raven register set is an interesting question as well. > > Ron > > Here is a code stub I used to connect an ISR to timer 3 on the 2400 ... #include "mv2400.h" #include "hawkMpic.h" #define TIMER_FREQ 0x3ef148 #define MILLI_SECOND (TIMER_FREQ/1000) #define TIMER3_INT_VEC 0x23 void timer3Init(void) { // setup the timer clk to be 8.25 MHz sysPciWrite32(MPIC_ADDR(MPIC_TIMER_FREQ_REG), TIMER_FREQ); // set the priority level and vector for timer #3 sysPciWrite32(MPIC_ADDR(MPIC_TIMER3_VEC_PRI_REG), PRIORITY_LVL15 | TIMER3_INT_VEC); // set the destination of the interrupt to be CPU 0 sysPciWrite32(MPIC_ADDR(MPIC_TIMER3_DEST_REG), DESTINATION_CPU0); // connect the scenario tick routine to timer3 intConnect((VOIDFUNCPTR *)TIMER3_INT_VEC, (VOIDFUNCPTR) yourClkTickRoutine, 0); // setup 2 milliseconds worth of ticks in the base count register and enable it sysPciWrite32(MPIC_ADDR(MPIC_TIMER3_BASE_CT_REG), MILLI_SECOND*2); } you should be able to use sysPciRead to read a timer register. hth John Moore Argon Engineering vxworks tornado From vxwexplo-errs@csg.lbl.gov Thu May 25 15:34:30 2000 From: Anthony Le Date: Thu May 25 15:34:32 PDT 2000 Subject: Looking for the driver of SCSI chip, LSI 53C1010 Hi all, I am looking for the VxWorks SCSI driver for the chip LSI-53C1010. I am using Pentium III hardware architecture and this SCSI controller. Please mail to: ale@connex.com if you can help me. Thanks, Anthony T. Le From vxwexplo-errs@csg.lbl.gov Thu May 25 20:36:42 2000 From: "Harish" Date: Thu May 25 20:36:45 PDT 2000 Subject: MIPS Dear VxWorkers, I am working on ATLAS board with RM5261 MIPS processor. The core board is having GT 64120 integrated system controller with features like SDRAM controller, PCI interface, DMA controller and others. The board also provides a ROM monitor. I am able to download and work with the shell when VxWorks image is build and downloaded to RAM by monitor. I couldnot work with the VxWorks_rom (or) VxWorks_romResident image programmed into the flash. I am configuring the SDRAM controller in the romInit.s. With this image it gets hanged when a call is made to shellInit ( ) while booting. I am using T2 for development. I would be very thankful for any help, suggestion, pointers.... Regards Harish. From vxwexplo-errs@csg.lbl.gov Thu May 25 22:12:06 2000 From: "abhijit lahiri" Date: Thu May 25 22:12:08 PDT 2000 Subject: Re: VxWorks stack trace >Abhijit, You say you want the output provided by the "tt" function. It >just prints it's information to stderr. How about capturing the output >to stderr while calling "tt"? I remember trying to figure out how tt >worked and finding that most of the work was done by the trcStack >function. This function is "architecture specific" and can get pretty >hairy. Your using a MIPS which I think has the architecture where >not all procedure calls have explicit stack frames. I don't think I >would want to have to write the code to unwind the stack myself. > >There is no man page for the trcStack function but there is an include >file (trcLib.h) that indicates the routine can be passed a custom >print function. Maybe you could ask WRS support if there is any way >of finding out how that routine works so you can pass it your own >routine to capture the info you want. Thanks Fred, your info may be useful for me. I will have a look at the trcStack function. I don't want to use the o/p of "tt" and want to do it myself because that is the best way to do it in a a system which does'nt have any filesystem to redirect the o/p of "tt" to. ( You may say memDrv, but I want to avoid any file I/O when a task has crashed/has generated exception ) It is true that MIPS specific architecture does not have explicit stack frames for every procedure call, and stack exists for a task as a whole. But I believe there is a catch, because "tt" does it. Thank you for your help. I will have a look into trcStack immediately. Thanking you, Abhijit From vxwexplo-errs@csg.lbl.gov Thu May 25 22:20:46 2000 From: "abhijit lahiri" Date: Thu May 25 22:20:49 PDT 2000 Subject: Re: VxWorks stack trace >Abhijit, You say you want the output provided by the "tt" function. It >just prints it's information to stderr. How about capturing the output >to stderr while calling "tt"? I remember trying to figure out how tt >worked and finding that most of the work was done by the trcStack >function. This function is "architecture specific" and can get pretty >hairy. Your using a MIPS which I think has the architecture where >not all procedure calls have explicit stack frames. I don't think I >would want to have to write the code to unwind the stack myself. > >There is no man page for the trcStack function but there is an include >file (trcLib.h) that indicates the routine can be passed a custom >print function. Maybe you could ask WRS support if there is any way >of finding out how that routine works so you can pass it your own >routine to capture the info you want. Thanks Fred, your info may be useful for me. I will have a look at the trcStack function. I don't want to use the o/p of "tt" and want to do it myself because that is the best way to do it in a a system which does'nt have any filesystem to redirect the o/p of "tt" to. ( You may say memDrv, but I want to avoid any file I/O when a task has crashed/has generated exception ) It is true that MIPS specific architecture does not have explicit stack frames for every procedure call, and stack exists for a task as a whole. But I believe there is a catch, because "tt" does it. Thank you for your help. I will have a look into trcStack immediately. Thanking you, Abhijit From vxwexplo-errs@csg.lbl.gov Fri May 26 05:33:01 2000 From: "Orhan Ugurlu" Date: Fri May 26 05:33:04 PDT 2000 Subject: Basic development tools This is a multi-part message in MIME format. ------=_NextPart_000_0089_01BFC729.06D25F20 Content-Type: text/plain; charset="iso-8859-9" Content-Transfer-Encoding: quoted-printable Hi, We have PPC with VxWorks run-time license. We want to develop software = for it. Can anyone suggest what to use? What about the prices of the = tools? Any free way of it? Thanks in advance for suggestions. Orhan Ugurlu ------=_NextPart_000_0089_01BFC729.06D25F20 Content-Type: text/html; charset="iso-8859-9" Content-Transfer-Encoding: quoted-printable
Hi,
We have PPC with VxWorks run-time = license. We want=20 to develop software for it. Can anyone suggest what to use? What about = the=20 prices of the tools? Any free way of it? Thanks in advance for=20 suggestions.
 
Orhan Ugurlu
------=_NextPart_000_0089_01BFC729.06D25F20-- From vxwexplo-errs@csg.lbl.gov Fri May 26 06:12:11 2000 From: "amass" Date: Fri May 26 06:12:13 PDT 2000 Subject: Error This is a multi-part message in MIME format. ------=_NextPart_000_0007_01BFC71C.53FD8AC0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable We have a Radstone PPC1a and Tornado 2/VxWorks. When executing = application code on the target, we see the following fault information: data access Exception current instruction address : 0x00194774 Machine Status Register : 0x00003030 Data Access Register : 0x1855cd94 Condition Register : 0x22000040 Data Storage Interrupt Register : 0x42000000 Task : 0x1dc4c88 "tNetTask" Can anyone offer any information about this? The error was displayed by = the Windows terminal application. Regards Andy Atkinson DERA Bincleaves ------=_NextPart_000_0007_01BFC71C.53FD8AC0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
We have a Radstone PPC1a and Tornado 2/VxWorks. When = executing=20 application code on the target, we see the following fault=20 information:
 
    data access
    Exception current instruction = address :=20 0x00194774
    Machine Status Register :=20 0x00003030
    Data Access Register :=20 0x1855cd94
    Condition Register :=20 0x22000040
    Data Storage = Interrupt Register :=20 0x42000000
    Task : 0x1dc4c88 = "tNetTask"
 
Can anyone offer any information about this? The = error was=20 displayed by the Windows terminal application.
 
Regards
 
 
Andy Atkinson
DERA Bincleaves
------=_NextPart_000_0007_01BFC71C.53FD8AC0-- From vxwexplo-errs@csg.lbl.gov Fri May 26 10:38:27 2000 From: "abhijit lahiri" Date: Fri May 26 10:38:29 PDT 2000 Hi All, Has anyone modified vxWorks code to use TLB for a MIPS processor ? I am asking this because the designs I experienced till now does'nt use TLB. They use only kseg0 and kseg1 of a MIPS processor. regards, Abhijit From vxwexplo-errs@csg.lbl.gov Tue May 30 01:15:56 2000 From: Suresh T Date: Tue May 30 01:15:59 PDT 2000 Subject: command-line Hi, I need a help on serial driver. When working on VxWorks :- Command line format::- command-name " without double quotes (normal unix or linum command-line format). Do you have any idea to make it work. For more clearness: On giving command-line input the control goes to the processing module via the parser. We can access this processing module (or do redirection). But cann't access the parser or redirect the control to fetch for our own parser. How to access the vxworks parser inorder to change the command-line format. Pls reply soon. This very much required urgently. Thanking you, Ranjini. From vxwexplo-errs@csg.lbl.gov Tue May 30 06:52:37 2000 From: "Mike Anderson" Date: Tue May 30 06:52:39 PDT 2000 Subject: RE: command-line VxWorks Greetings! I'm Baaaaack! [snip] > > Hi, > I need a help on serial driver. > When working on VxWorks :- > Command line format::- command-name " The VxWorks is accepting the command in the above format. But I need to > change this format to > # command-name without double quotes (normal unix or linum > command-line format). > Do you have any idea to make it work. > > For more clearness: > On giving command-line input the control goes to the processing > module via > the parser. We can access this processing module (or do redirection). But > cann't access the parser or redirect the control to fetch for our own > parser. How to access the vxworks parser inorder to change the > command-line format. > Pls reply soon. This very much required urgently. > [snip] You haven't said whether you're using the WindSh or the native target shell. Or, are you running any shell? The WindSh runs on the host -- not the target. Any attempt to do what you're looking for there would have to be done in TCL. If you're running the target shell (tShell), simply set you're interpretor's priority higher than that of tShell. This will allow you to intercept the characters and do any mungeing you need to do on the input stream. Then send the characters to the shell's STD_IN and viola'. Alternatively, you could access the tShell's "execute" command, but that's not terribly well documented. HTH, Mike Anderson The PTR Group, Inc. Wind River-certified Engineers and Training BSPs, device drivers, Tornado/VxWorks engineering services mailto:mike@theptrgroup.com (V) 703.430.DrIt (3748) From vxwexplo-errs@csg.lbl.gov Tue May 30 14:33:59 2000 From: Anthony Le Date: Tue May 30 14:34:01 PDT 2000 Subject: 1 Gigabit ethernet driver???? Hello vxWorks experts, Does vxWorks support 1 (one) gigabit ethernet? Where is the driver? What is the ethernet hardware controller? Regards, Anthony T. Le From vxwexplo-errs@csg.lbl.gov Tue May 30 19:04:29 2000 From: Fred Roeber Date: Tue May 30 19:04:32 PDT 2000 Subject: RE: command-line On Tue, 30 May 2000, Mike Anderson wrote: > > VxWorks Greetings! > > I'm Baaaaack! > ... > Mike Anderson > The PTR Group, Inc. > Wind River-certified Engineers and Training > BSPs, device drivers, Tornado/VxWorks engineering services > mailto:mike@theptrgroup.com > (V) 703.430.DrIt (3748) Is this Mike Anderson, once of Sparta then of ChipCom (or some such)? Back to VxWorks stuff but not back to Sparta? What's the skinny? Startup blues? Good to have you back on board. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing | From vxwexplo-errs@csg.lbl.gov Wed May 31 04:03:10 2000 From: Vxworks Exploder Date: Wed May 31 04:03:12 PDT 2000 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed May 31 04:03:07 PDT 2000 Subject: Re: sa1100IntrCtl.c bug. Subject: POSIX AIO and HW Flow Control Subject: Dynamic Loading Problem Subject: Re: Dumping target memory Subject: Re: dynamic sizeof an array - seems to work Subject: Re: Issue about ifAddrSet()! Subject: Re: Scheduling Problem ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: sa1100IntrCtl.c bug. Date: 25 May 2000 11:28:02 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <32jV4.1137$s4.90925@petpeeve.ziplink.net> <5gaX4.161$x4.13618@petpeeve.ziplink.net> Sender: bpringle@DeadDuck >>>>> "Steve" == Steve Dillen writes: Steve> I agree with that bug and fix, and along with it in the same Steve> code is: Steve> if (level < 0 || level > SA1100_INT_NUM_LEVELS) return ERROR; Steve> What happens when level is set to equal SA1100_INT_NUM_LEVELS Steve> (which is 32 in brutus, and my custom BSP), it won't return an Steve> error, but also won't work. I admit this is a minor bug that Steve> won't have an adverse affect on most code, but that was Steve> partially what was confusing me, I thought the sa1100IntrCtl Steve> code was accepting all levels from 0 to 32, with 0 being a Steve> disable all level, therefore unacceptable. Steve> The same problem is in the disable routine. That is a good point; code might as well be correct. At least I agree with your observation as well. I will change my BSP to fix this. This is applicable to the SA110, SA1100 and the SA1110 from what I know about the processors. I don't have the abbaset BSP, but it could have similar problems. My code is based off the brutus. The only pragmatic bug is if you try to use GPIO 0 as an interrupt source (or that is my understanding). ttyl, Bill - -- Have you ever tried to hum God Save the Queen madly with a large nymph in a garbage dump? Or been given seven large bananas made out of your telephone? You will. And the company that will bring it to you: AT&T. --------------------------- Newsgroups: comp.os.vxworks Subject: POSIX AIO and HW Flow Control Date: Fri, 26 May 2000 16:27:13 GMT From: "G Kraemer" Organization: @Home Network Message-ID: vxWorks experts: MVME-2700 Power PC using Tornado II. Problem: POSIX AIO routines will NOT work if hardware flow control is enabled. Reference included sample program. When I set the following: newopts = 0xe; The AIO receive routine stops receiving! If I set it to 0xf (disable hardware flow) it receives! I need help getting hardware flow control working with the POSIX AIO routines. Any ideas? TIA Garry - ---------------------------------------------------------------------------- - ---------------------------------------- /* testio - designed to test signal handling of vxWorks Connect computer connection to serial port 2 '/tyCo/1' using a full hardware handshake cable open xterm window 1 tip ppca Connect computer to port 2 ("/tyCo/1") From xterm window 1 load termio ->ldsp testio messages should be written to serial port 2 (computer connection) Type characters in computer connection; read signal should be activated and the character displayed in the xterm window 1. from xterm window 1; enter wr=1 this causes 10 lines to be sent using aio_write to xterm window 2 (tip). enter LOOP=0 (xterm window 1) to close the serial port and terminate the program. Full hardware handshake cable is as follows: RX (I) 2 ---------- 3 TX (O) TX (O) 3 ---------- 2 RX (I) /---- 1 CD (I) DTR (O) 4 -----+---- 6 DSR (I) S. GND 5 ---------- 5 S. GND CD (I) 1 ---\ DSR (I) 6 ---+------ 4 DTR (O) RTS (O) 7 ---------- 8 CTS (I) CTS (I) 8 ---------- 7 RTS (O) RI (I) 9 N/C 9 RI (I) */ #include "vxWorks.h" #include "aio.h" #include "errno.h" #include "ioLib.h" #include #define BUFFER_SIZE 200 #define LIST_SIZE 100 /* what is this for?? Is this the POSIX queue size? */ #define SIG_NO_25 25 #define SIG_NO_26 26 struct aiocb aiocb_read; struct aiocb aiocb_write; char buffer[BUFFER_SIZE]; void myReadSigHandler(int sig, struct siginfo * info, void * pContext); void myWriteSigHandler(int sig, struct siginfo * info, void * pContext); int LOOP = 1; int wr = 0; /* write flag, 1 = write to terminal, 0 = don't write */ int rd = 1; /* read flag, 1 = read from terminal, 0 = don't read */ void testio(void) { int fd; int myCount = 0; int check; int baudrate = 19200; int newopts; int *oldopts; int set_options = 0x1005; int get_options = 0x1006; static char inPort[] = "/tyCo/1"; struct sigaction read_action; struct sigaction write_action; static char * test_string = "This is a test string to send out the serial port\r\n"; /* console port = /tyCo/0 serial port 1 = /tyCo/1 serial port 2 = /tyCo/2 serial port 3 = /tyCo/3 */ /* open serial port 1 */ if((fd = open(inPort, O_RDWR,0)) == ERROR) { printf("Error opening %s\r\n",inPort); } /* set baud rate */ if(fd > 0 ){ /* this sets the port to character mode; instead of line mode */ ioctl(fd,FIOSETOPTIONS, OPT_RAW); ioctl(fd, FIOBAUDRATE, baudrate); ioctl(fd, get_options, (int) &oldopts); printf("Opened '%s', opts=0x%x, baudrate = %d \n","/tyCo/1",oldopts,baudrate); /* change newopts to 0x0f to disable hardware flow control. 0x0E 0000 1110 ^^^^ ^^^^ |||| ||||===> ignore modem status lines |||| |||====> Enable Device Receiver |||| ||=====> data bits 01xx =6, 10xx = 7, and 11xx = 8 |||| ||=====> data bits ||||========> Hang up on last close |||=========> send two stop bits (else one) ||==========> parity detection enabled (else disabled) |===========> odd parity (else even) */ newopts = 0xe; /* only 0xf works here! 0xf is power up state! */ ioctl(fd, set_options, newopts); ioctl(fd, get_options, (int) &newopts); printf("newopts='0x%x' \n",newopts); } /* flush buffers */ ioctl(fd,FIOFLUSH,0); /* set up read signal handlers for SIG_NO_25 */ read_action.sa_sigaction = myReadSigHandler; read_action.sa_flags = SA_SIGINFO; sigemptyset(&read_action.sa_mask); sigaction(SIG_NO_25, &read_action, NULL); /* set up write signal handlers for SIG_NO_26 */ write_action.sa_sigaction = myWriteSigHandler; write_action.sa_flags = SA_SIGINFO; sigemptyset(&write_action.sa_mask); sigaction(SIG_NO_26, &write_action, NULL); /* initialize read aiocb and clear buffer */ bzero ((char *) & aiocb_read, sizeof(struct aiocb)); bzero ((char *) buffer, sizeof(buffer)); aiocb_read.aio_fildes = fd; aiocb_read.aio_buf = buffer; aiocb_read.aio_nbytes = BUFFER_SIZE; aiocb_read.aio_reqprio = 0; /* set up signal information */ aiocb_read.aio_sigevent.sigev_signo = SIG_NO_25; aiocb_read.aio_sigevent.sigev_notify = SIGEV_SIGNAL; aiocb_read.aio_sigevent.sigev_value.sival_ptr = (void *) &aiocb_read; /* initialize write aiocb */ bzero ((char *) & aiocb_write, sizeof(struct aiocb)); aiocb_write.aio_fildes = fd; aiocb_write.aio_nbytes = 1; aiocb_write.aio_reqprio = 0; /* set up signal information */ aiocb_write.aio_sigevent.sigev_signo = SIG_NO_26; aiocb_write.aio_sigevent.sigev_notify = SIGEV_SIGNAL; aiocb_write.aio_sigevent.sigev_value.sival_ptr = (void *) &aiocb_write; printf("issue write (nowait) to serial port\r\n"); write(fd,"\r\n\r\nThis is a wait write\r\n",sizeof("\r\n\r\nThis is a wait write\r\n")); taskDelay(50); printf("issue nowait read\r\n"); /* initiate the first nowait read */ if(aio_read(&aiocb_read) == -1) printf("testio: aio_read failed\r\n"); while(LOOP) myCount++; if( (myCount % 10000000) == 0) printf("myCount = %d\n",myCount); if ((char)*(char *)aiocb_write.aio_buf == '7') break; /* verify write is still in progress */ } printf("\n exit first main loop ... cancel then print again... "); aio_cancel(fd,&aiocb_read); printf("\n aiocb_read cancelled"); close(fd); printf("Close /tyCo/1\r\n"); } void myReadSigHandler( int sig, struct siginfo * info, void * pContext) { printf("myReadSigHandler: got signal for aio_read\r\n"); /* test if read still in progress */ if(aio_error(info->si_value.sival_ptr) == EINPROGRESS) { printf("myReadSigHandler: read still in progress; call WRS!\r\n"); } if(aio_return(info->si_value.sival_ptr) == -1) { printf("myReadSigHandler: aio_return for aiocb_read failed\r\n"); } else { printf("testio: aio_read message = '%s'\r\n",aiocb_read.aio_buf); aiocb_write.aio_buf = aiocb_read.aio_buf; aiocb_write.aio_nbytes = 1; aio_write(&aiocb_write); /* initiate the next read */ if(aio_read(&aiocb_read) == -1) printf("testio: aio_read failed\r\n"); } } void myWriteSigHandler( int sig, struct siginfo * info, void * pContext) { printf("myWriteSigHandler: got signal for aio_write\r\n"); /* test if write still in progress */ if(aio_error(info->si_value.sival_ptr) == EIN PROGRESS) printf("myWriteSigHandler: write still in progress; call WRS!\r\n"); } if(aio_return(info->si_value.sival_ptr) == -1) printf("myWriteSigHandler: aio_return for aiocb_write failed\r\n"); } else printf("testio: aio_write message = %s\r\n",aiocb_write.aio_buf); aio_read(&aiocb_read); } } - ------------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: Dynamic Loading Problem Date: Fri, 26 May 2000 13:42:23 -0400 From: Larry Vincent Organization: NASA Glenn Research Center Message-ID: <392EB77F.1A082D07@lerc.nasa.gov> - --------------9ED15BD9A897A36E5D1F996A Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I'm writing an embedded application that needs to allow dynamic loading of software. I'm probably able to maintain the OS, but I need to dynamically load the application at run-time. I'm running VxWorks version 5.3.1 and Tornado 1.0.1. I've reconfigured the kernel, adding the following symbol table identifiers: INCLUDE_SYM_TBL, INCLUDE_SYM_TABLE_SYNC, INCLUDE_NET_SYM_TABLE I created two programs to test with. The first program is void dcmInit( void ) { printf( "In the dummy dcmInit function\n" ); } int loadTest( char *filename ) { int fd; MODULE_ID moduleId; fd = open( filename, O_RDONLY, 0666 ); if( fd == -1 ) { printf( "Error opening file\n" ); return( -1 ); } moduleId = loadModule( fd, LOAD_GLOBAL_SYMBOLS ); printf( "module Id = %x\n", moduleId ); close( fd ); dcmInit(); return( 0 ); } The second file that I load merely has: void dcmInit( void ) { printf( "new dcmInit function\n" ); } void func1( void ) { printf( "func1\n" ); } void func2( void ) { printf( "func2\n" ); } After booting, I load the first module using the shell and execute the program. ld < firstFile.o loadTest( "secondFile.o" ) I don't get any errors, but the program prints that it ran the dummy DCM function. Also, the shell doesn't register that either func1 or func2 exist. The moduleShow command doesn't register that the secondFile.o module was ever loaded. I assume from the lack of errors that the module did load correctly on the target and that the symbol table stuff is just a symptom of shell vs. target. But I thought loading the new module would overwrite the definition of dcmInit(). Any idea why this isn't working? Thanks, Larry - --------------9ED15BD9A897A36E5D1F996A Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit I'm writing an embedded application that needs to allow dynamic loading of software.  I'm probably able to maintain the OS, but I need to dynamically load the application at run-time.  I'm running VxWorks version 5.3.1 and Tornado 1.0.1.

I've reconfigured the kernel, adding the following symbol table identifiers:
INCLUDE_SYM_TBL, INCLUDE_SYM_TABLE_SYNC,  INCLUDE_NET_SYM_TABLE

I created two programs to test with.  The first program is

void dcmInit( void )
{
    printf( "In the dummy dcmInit function\n" );
}
 

int loadTest( char *filename )
{
    int fd;
    MODULE_ID moduleId;

    fd = open( filename, O_RDONLY, 0666 );
    if( fd == -1 )
    {
        printf( "Error opening file\n" );
        return( - -1 );
    }
    moduleId = loadModule( fd, LOAD_GLOBAL_SYMBOLS );
    printf( "module Id = %x\n", moduleId );
    close( fd );
    dcmInit();
    return( 0 );
}
 

The second file that I load merely has:

void dcmInit( void )
{
    printf( "new dcmInit function\n" );
}

void func1( void )
{
    printf( "func1\n" );
}
 

void func2( void )
{
    printf( "func2\n" );
}
 

After booting, I load the first module using the shell and execute the program.
    ld < firstFile.o
    loadTest( "secondFile.o" )

I don't get any errors, but the program prints that it ran the dummy DCM function.  Also, the shell
doesn't register that either func1 or func2 exist.  The moduleShow command doesn't register that
the secondFile.o module was ever loaded.

I assume from the lack of errors that the module did load correctly on the target and that the symbol
table stuff is just a symptom of shell vs. target.  But I thought loading the new module would overwrite
the definition of dcmInit().

Any idea why this isn't working?

    Thanks,
        Larry - --------------9ED15BD9A897A36E5D1F996A-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dumping target memory Date: Tue, 30 May 2000 11:49:50 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <39340D4E.974980F4@delphi.com> References: <3933C7D0.E07974B4@francimel.com> The command is d address,howmany, size If you want to go to a file, follow with >filename You may need to specify the full path if your current dir isn't writeable. Speaking only for myself, Joe Durusau jph wrote: > > Is it true that, with Tornado 1, there is no way to dump the memory of > the target into a file on the host system ? > It seems hard to believe ! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: dynamic sizeof an array - seems to work Date: Mon, 29 May 2000 10:48:23 +0200 From: "Michael St" Organization: NetVision Israel Message-ID: <8gt7bj$q4d$1@news.netvision.net.il> References: <8glqvj$21c$1@nnrp1.deja.com> Hi ! Thank you that you remember me the question which I wanted to ask. What about dynamic allocation with 'malloc' when I work with VxWorks Tornado ). If it is reentrable function and I can use it in multitasking environment or not ? If not - what can I use instead of ? About "iArray2" . As I heard, it will be possible in the new standard of C. But it is not something new. Such thing was possible in PL/1 language. It's right that you must be very carefully when using it, but there are some advantages. Pete Kockritz wrote in message news:8glqvj$21c$1@nnrp1.deja.com... > In article , > "Markus MItterer" wrote: > > > > Try to do following: > > > > void Check(int iSizeOfArray) > > { > > int iArray1[100]; > > int iArray2[iSizeOfArray]; > > int iTemp; > > That's an extension to ANSI C that gcc has. Look in the gcc manual > under 'Arrays of Variable Length'. > > > It's sometimes useful and can be dangerous. Consider someone calling > your function like this: > Check (0x800000); > The stack could be blown without your function being able to validate > the parameter it was passed. > > -- > Regards, > Pete Kockritz > > The views expressed are my own and not necessarily that of my employer's > > > Sent via Deja.com http://www.deja.com/ > Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Issue about ifAddrSet()! Date: Wed, 31 May 2000 09:11:26 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8h2kvo$s7s$1@nnrp1.deja.com> References: <01bf3a57$b346f180$c89247d2@server2> Hi Clark, While it is not impossible that there is a bug in the ifAddrSet() function, I would guess that it is just as likely, if not more so, that you are simply using it incorrectly, but sadly your post says nothing about why you feel there is a bug in the function. Even more oddly, you make a reference to an image compression package which is broken by it; I would be surprised to find a JPEG library using ifAddrSet()... Perhaps if you let us have some more information about what you are trying to do, and how you are doing it we'll be able to save you the effort of re-writing a function that works for a lot of people without problems. Regards, John... In article <01bf3a57$b346f180$c89247d2@server2>, "Clark" wrote: > Dear, > > There seems has a bug inside! It will broken the ucode > function of JPEG package for MPC82x. > > Any one is familiar with the flow of this function? I want > to write a new one before the updating form windRiver. > > Clark > Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Scheduling Problem Date: Wed, 31 May 2000 09:26:40 GMT From: john_94501@my-deja.com Organization: Deja.com - Before you buy. Message-ID: <8h2ls7$st1$1@nnrp1.deja.com> References: <8gb8eq$7dr4@overload.lbl.gov> In article , Diamand@btinternet.com (Luke Diamand) wrote: > I believe there's a recentish patch to Tornado II which fixes a > bug whereby after an interrupt, the _lowest_ priority task gets > to be run. There is a patch for the problem described, but it was not the lowest priority task that got to run after the interrupt; it was the task selected to run after the first interrupt was processed, but which didn't get a chance to run because a second interrupt arrived before it could be switched in. Additionally, the interrupt had to arrive exactly at the wrong time for this problem to occur. The next interrupt arriving (often the next system clock tick) restored order to the system hence the apparent delay in scheduling the correct task. As Luke says, dheck WindSurf for the patch! Regards, John... Sent via Deja.com http://www.deja.com/ Before you buy. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 31 06:51:28 2000 From: "Tony Mayes" Date: Wed May 31 06:51:30 PDT 2000 Subject: VxWorks and Multicast Hi, I am using vxworks 5.3.1 and want to use "multicast" messages over ethernet. I understand that this is possible using the SENS ethernet stack using mcastrecv() / mcastsend() but not with the pre SENS stack which we are using. Is this correct? If so, has anyone added multicast and if so how do I go about it or is there some freeware code that I can add in? TIA Tony Tony Mayes (The opinions expressed within are my own) Email address tony.mayes@epid.eurotherm.co.uk Phone (0044)(0)1903 205277 Fax (0044)(0)1903 524016 Eurotherm Process Instrumentation Div, Southdownview Way, Worthing, Sussex, UK, BN14 8NN From vxwexplo-errs@csg.lbl.gov Wed May 31 07:41:43 2000 From: "Michael Lawnick" Date: Wed May 31 07:41:45 PDT 2000 Subject: Re: Dynamic Loading Problem Hi Larry, just try @moduleShow and @dcmInit HTH > I'm writing an embedded application that needs to allow dynamic loading > of software. I'm probably able to maintain the OS, but I need to > dynamically load the application at run-time. I'm running VxWorks > version 5.3.1 and Tornado 1.0.1. > > I've reconfigured the kernel, adding the following symbol table > identifiers: > INCLUDE_SYM_TBL, INCLUDE_SYM_TABLE_SYNC, INCLUDE_NET_SYM_TABLE > > I created two programs to test with. The first program is > > void dcmInit( void ) > { > printf( "In the dummy dcmInit function\n" ); > } > > > int loadTest( char *filename ) > { > int fd; > MODULE_ID moduleId; > > fd = open( filename, O_RDONLY, 0666 ); > if( fd == -1 ) > { > printf( "Error opening file\n" ); > return( -1 ); > } > moduleId = loadModule( fd, LOAD_GLOBAL_SYMBOLS ); > printf( "module Id = %x\n", moduleId ); > close( fd ); > dcmInit(); > return( 0 ); > } > > > The second file that I load merely has: > > void dcmInit( void ) > { > printf( "new dcmInit function\n" ); > } > > void func1( void ) > { > printf( "func1\n" ); > } > > > void func2( void ) > { > printf( "func2\n" ); > } > > > After booting, I load the first module using the shell and execute the > program. > ld < firstFile.o > loadTest( "secondFile.o" ) > > I don't get any errors, but the program prints that it ran the dummy DCM > function. Also, the shell > doesn't register that either func1 or func2 exist. The moduleShow > command doesn't register that > the secondFile.o module was ever loaded. > > I assume from the lack of errors that the module did load correctly on > the target and that the symbol > table stuff is just a symptom of shell vs. target. But I thought > loading the new module would overwrite > the definition of dcmInit(). > > Any idea why this isn't working? > > Thanks, > Larry > MfG Lawnick, SOFTEC GmbH ============================================== SOFTEC GmbH Tel +49-731-96600-0 Promenade 17 Fax +49-731-96600-23 D-89073 Ulm Michael Lawnick Germany lawnick@softec.de ============================================== From vxwexplo-errs@csg.lbl.gov Wed May 31 10:08:00 2000 From: Anthony Le Date: Wed May 31 10:08:02 PDT 2000 Subject: 1 gigabit enternet. Hi all, I am implementing 1 gigabit ethernet application for my project. Can anyone recommend any hardware that has VxWorks driver? The target is Pentium III. Cheers, Anthony T. Le From vxwexplo-errs@csg.lbl.gov Wed May 31 10:54:54 2000 From: Geoffrey Espin Date: Wed May 31 10:54:56 PDT 2000 Subject: Re: VxWorks stack trace Abhijit, Fred, > >There is no man page for the trcStack function but there is an include > >file (trcLib.h) that indicates the routine can be passed a custom > Thanks Fred, > your info may be useful for me. I will have a look at > the trcStack function. I don't want to use the o/p of "tt" and want to Here is all you need. You can even call xxxTrace(tid) from interrupt level... though logMsg() is far from ideal. Geoff /******************************************************************************* * * xxxTracePrint - stack trace print function * * RETURNS: OK or ERROR */ static void xxxTracePrint ( INSTR *caller, int func, int nargs, int *args ) { char buf [250]; int ix; int len = 0; len += sprintf (&buf [len], "%#10x: %#10x (", (int)caller, func); for (ix = 0; ix < nargs; ix++) { if (ix != 0) len += sprintf (&buf [len], ", "); len += sprintf (&buf [len], "%#x", args [ix]); } len += sprintf (&buf [len], ")\n"); logMsg (buf); } /******************************************************************************* * * xxxTrace - stack trace * * RETURNS: OK or ERROR */ int xxxTrace ( WIND_TCB *pTcb ) { REG_SET regs; if (pTcb == NULL) return (ERROR); taskRegsGet ((int)pTcb, ®s); trcStack (®s, (FUNCPTR) xxxTracePrint, (int)pTcb); return (OK); } -- Geoffrey Espin espin@idiom.com From vxwexplo-errs@csg.lbl.gov Wed May 31 13:55:19 2000 From: "abhijit lahiri" Date: Wed May 31 13:55:22 PDT 2000 Subject: VxWorks TFFS Hi All, I am using TFFS for AMD (AM29F017B) flash. I have customised sysTffs.c according to my needs. amdIdentify routine is able to identify the flash. Even tffsRawIO routines are working perfectly. But tffsDevCreate fails, though amdIdentify does'nt return any error. i am getting a NULL from tffsdevCreate and errno is 0xdd. (Undocumented errno). If anyone has experienced solving such a problem, plase help. Any response will be highly appreciated. TIA, Abhijit From vxwexplo-errs@csg.lbl.gov Wed May 31 19:06:37 2000 From: "QinTing" Date: Wed May 31 19:06:39 PDT 2000 Subject: DHCP This is a multi-part message in MIME format. ------=_NextPart_000_0027_01BFCBB3.477FB700 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: quoted-printable Hi, Every one I meet some problems in my work now. When I want to chose INCLUDE_DHCPC in my vxWorks to make it a DHCP client, Tornado reports that there is a conflict. I find that when = select INCLUDE_DHCPC, the components that Tornado automatically includes are in conflict.Why? How can I make vxWorks a DHCP client? Regards, QTing ------=_NextPart_000_0027_01BFCBB3.477FB700 Content-Type: text/html; charset="gb2312" Content-Transfer-Encoding: quoted-printable

Hi, Every one
 
I meet some problems in my work now.

When I want to chose=20 INCLUDE_DHCPC in my vxWorks to make it a DHCP
client, Tornado reports = that=20 there is a conflict. I find that when select
INCLUDE_DHCPC, the = components=20 that Tornado automatically includes are in
conflict.Why?

How = can I=20 make vxWorks a DHCP=20 client?

Regards,
QTing
------=_NextPart_000_0027_01BFCBB3.477FB700-- From vxwexplo-errs@csg.lbl.gov Wed May 31 19:21:37 2000 From: Fred Roeber Date: Wed May 31 19:21:40 PDT 2000 Subject: Re: Dynamic Loading Problem On Wed, 31 May 2000, Larry Vincent wrote: > I'm writing an embedded application that needs to allow dynamic loading > of software. I'm probably able to maintain the OS, but I need to > dynamically load the application at run-time. > version 5.3.1 and Tornado 1.0.1. ... > I created two programs to test with. The first program is > > void dcmInit( void ) > { > printf( "In the dummy dcmInit function\n" ); > } > > > int loadTest( char *filename ) > { > int fd; > MODULE_ID moduleId; > > fd = open( filename, O_RDONLY, 0666 ); > if( fd == -1 ) > { > printf( "Error opening file\n" ); > return( -1 ); > } > moduleId = loadModule( fd, LOAD_GLOBAL_SYMBOLS ); > printf( "module Id = %x\n", moduleId ); > close( fd ); > dcmInit(); > return( 0 ); > } > > > The second file that I load merely has: > > void dcmInit( void ) > { > printf( "new dcmInit function\n" ); > } ... > After booting, I load the first module using the shell and execute the > program. > ld < firstFile.o > loadTest( "secondFile.o" ) > > I don't get any errors, but the program prints that it ran the dummy DCM > function. Also, the shell > doesn't register that either func1 or func2 exist. The moduleShow > command doesn't register that the secondFile.o module was ever loaded. Larry, If you check the man page for loadModuleAt you will find an explanation of how loading modules handles dynamic linking. In particular, when you load a new module, only the unresolved external references in the file being loaded are changed (using the current symbol addresses at the time the module is loaded). Looking at your first file, the call to dcmInit in loadTest will be resolved at compile time since there is a dcmInit function in the same file and it isn't "forward referenced". Thus, the dynamic loader won't even be involved. Further, loading a new copy of dcmInit into the system has no effect on code that has previously been loaded. Only code that is loaded after the second file would reference the "real" version of dcmInit. Sorry if this isn't what you want but that's the way things work; unresolved references in code only ever get fixed once when the code is loaded. As far as not seeing the other symbols, someone mentioned checking the target symbol table using the "@" command. Personally, I tend to like to use a serial link to a board and use the real target resident shell when looking into problems like this. Helps avoid the whole issue of what the host knows vs what the target knows. Fred | Fred J Roeber, BBN Systems & Technologies | | 4 John Clarke Road Middletown, RI 02842-5202 | | froeber@bbn.com 401-848-3548 | | Division Scientist, High Performance Computing |