From vxwexplo-errs@csg.lbl.gov Wed May 1 04:03:16 2002 From: Vxworks Exploder Date: Wed May 1 04:03:19 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed May 1 04:03:12 PDT 2002 Subject: Re: Create dosFS partition dinamically Subject: Boot vxWorks with different ethernet drivers Subject: Re: unable to add equal cost routes Subject: Re: PCI-to-PCI bridge x86 Subject: How to reduce the size of ENVOY SNMP module Subject: VxWorks Vs other embedded OSs Subject: convert vxworks_rom.hex Subject: Re: convert vxworks_rom.hex Subject: Re: VxWorks Vs other embedded OSs Subject: Re: VxWorks Vs other embedded OSs Subject: file upload Subject: Re: VxWorks Vs other embedded OSs Subject: Re: VxWorks Vs other embedded OSs Subject: Lightweight heap monitor for VxWorks 5.4 / Tornado 2.0 Subject: Re: INCLUDE_SNMP Subject: Re: Porting code from BSD to Vxworks Subject: Envoy version in WindNet SNMP Subject: Re: Lightweight heap monitor for VxWorks 5.4 / Tornado 2.0 Subject: Re: affordable vxworks training Subject: How reliable is routeLib in Vxworks Sim? Subject: resourses for realtime Subject: Re: Does windml 2.0 font library support japanese and chinese fonts Subject: Re: VxWorks Vs other embedded OSs Subject: Re: PCI-to-PCI bridge x86 Subject: Re: file upload Subject: Tornado trees and version control Subject: deleting netPools created through netBufLib Subject: Re: VxWorks Vs other embedded OSs Subject: Re: resourses for realtime Subject: #ifdef question Subject: Re: deleting netPools created through netBufLib Subject: Re: #ifdef question Subject: Re: #ifdef question ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Create dosFS partition dinamically Date: Mon, 29 Apr 2002 15:56:05 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Set the global variable dpartDebug to "1" and do the disk insertion removal, insert disk, access it (e.g. do "ls"), remove it, and repeat this with another disk. The debug mesages should pretty clearly show what is happening. - - Leonid Rosenboim P.S. Iit seems you are not getting your money's worth for WindRiver support, questions like this even first tier support engineers used to be able to answer. "Tri VU KHAC (Sony)" wrote in message news:qK9z8.41$Pm2.1173@nreader2.kpnqwest.net... > Hi folks, > I am concerned about dosFS with the removable medium devices. Let's examine > an example: > > +I mount first a medium which contains a Master Boot Record (MBR) at the > sector 0 and one DOS partition. Then, I remove the medium and replace it by > another medium which does not have MBR, but Boot Partition Record (BPR) of > DOS at the sector 0. > > My question is: how does dosFSLib manage this scenario? > Thank you for your helps. > Tri. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Boot vxWorks with different ethernet drivers Date: 29 Apr 2002 09:01:25 -0700 From: san_blabla@hotmail.com (San Blabla) Organization: http://groups.google.com/ Message-ID: <6046c736.0204290801.4047b231@posting.google.com> Hi, We have two ethernet END drivers for two different controlers, but only one controler can be attached to the board at a time. We want to build a single system image that could dynamicaly use the right kind of ethernet controler driver depending on which hardware is realy attached to our board (both in standalone boot mode and network boot mode). In configNet.h, I add both of them to END driver table, but as the boot command line in config.h need to specify the boot up network device, how can I set the boot command line to let the system choose the right driver that uses what is really attached to the board. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: unable to add equal cost routes Date: 29 Apr 2002 10:49:14 -0700 From: vividh@hotmail.com (rasayani) Organization: http://groups.google.com/ Message-ID: <28809481.0204290949.254a944f@posting.google.com> References: <8dff11cd.0204212008.6d401648@posting.google.com> VxWorks kernel doesn't support ECMP. sai_dattathrani@infy.com (sai) wrote in message news:<8dff11cd.0204212008.6d401648@posting.google.com>... > Hi, > > I tried adding routes to same destination with different gateway. But > VxWorks does not seem to accept. > > If I try to add route to destination 50.0.0.0 as below > Destination Gateway > 50.0.0.0 90.0.0.1 > 50.0.0.0 90.0.0.3 > > Tried adding using RouteAdd() function provided by VxWorks and routing > socket, the second route is not added to the kernel. Routing socket > returns (File exist - EEXIST) error. > > could someone please help me solve this problem. > rgds > sai --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCI-to-PCI bridge x86 Date: Mon, 29 Apr 2002 18:27:24 GMT From: "H. E. Taylor" Organization: MTS Internet Message-ID: <3CCDAB3F.341B@despam.autobahn.mb.ca> References: <3CC8E942.5B1E@despam.autobahn.mb.ca> <3CC99E3A.720F@despam.autobahn.mb.ca> <3ccd145d$1@brateggebdc5.br-automation.co.at> In article <3ccd145d$1@brateggebdc5.br-automation.co.at>, Werner Schiendl wrote: >> >>[...] > > Be sure that you have 'Plug and Play OS installed' or similar option of > your BIOS _dis_abled! > If this option is enabled, the BIOS will only setup devices required for > booting (disk controller, graphics controller, keyboard). > If it is disabled, as suggested, most BIOS will setup all devices it can > possibly. > Hi Werner, Yes but, the BIOS would still set up any bridges it found, wouldn't it? [Anyway, Dan, it is something else for you to try...] - -het - -- "If a packet hits a pocket on a socket on a port, And the bus is interrupted as a very last resort, And the address of the memory makes your floppy disk abort, Then the socket packet pocket has an error to report!" - -Gene Ziegler (with a nod to Dr. Suess) Computer Links: http://www.autobahn.mb.ca/~het/clinks.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: How to reduce the size of ENVOY SNMP module Date: 29 Apr 2002 11:37:14 -0700 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: Hello: I have successfully built the ENVOY SNMP module for PPC, but it quite big (1.5MBytes). I hope to reduce it. Does any one have the idea ? One way I want to try is, among the 11 groups of MIB II, only include a few of them, say the system and ip, exclude the rest 9 groups. But how to do it exactly? Thanks in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Vs other embedded OSs Date: 29 Apr 2002 14:06:52 -0700 From: sriva_t@yahoo.com (Srivathsa) Organization: http://groups.google.com/ Message-ID: <1cdb3add.0204291306.2f4ebd6c@posting.google.com> Hi, I am new to embedded OS' Just want to know how VxWorks compares against other embedded OS'. 1) Psos seems to be as popular as VxWorks. But WindRiver seems to own both Psos and VxWorks. How come? Did WindRiver buy one of these OSs from some other company.. in a bid to kill that OS soon? 2) How does ChorusOS compare with VxWorks in terms of popularity (i.e is it widely used? ) 3) Some embedded OSs have multiple-process concept (seperate address space for each process, which implies that we need hardware MMU. I see that ChorusOS and LynxOS have this feature), while many dont (like VxWorks, Psos). Does this mean that such multiple-process OSs are meant for high-end equipment alone? (i.e those that can afford to have hardware MMU) - -Srivathsa --------------------------- Newsgroups: comp.os.vxworks Subject: convert vxworks_rom.hex Date: 29 Apr 2002 14:47:27 -0700 From: bala@innocor.com (balendran bala) Organization: http://groups.google.com/ Message-ID: Hi guys, Is there anyway that I could convert vxworks_rom.hex file into a text file? Thank you, bala --------------------------- Newsgroups: comp.os.vxworks Subject: Re: convert vxworks_rom.hex Date: Mon, 29 Apr 2002 19:22:40 -0400 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3CCDD5C0.54DEFA39@bellsouth.net> References: balendran bala wrote: > Hi guys, > > Is there anyway that I could convert vxworks_rom.hex file into a text file? > > Thank you, > bala If you look at it, it _is_ a text file. What do you want to do??? Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Vs other embedded OSs Date: 30 Apr 2002 09:49:41 -0700 From: igtorque@yahoo.es (=?ISO-8859-1?Q?Ignacio_Gonz=E1lez_T.?=) Organization: http://groups.google.com/ Message-ID: <74c2e5b4.0204300849.17877357@posting.google.com> References: <1cdb3add.0204291306.2f4ebd6c@posting.google.com> sriva_t@yahoo.com (Srivathsa) wrote in message news:<1cdb3add.0204291306.2f4ebd6c@posting.google.com>... > Hi, > I am new to embedded OS' > Just want to know how VxWorks compares against other embedded OS'. > > 1) Psos seems to be as popular as VxWorks. But WindRiver seems to > own both Psos and VxWorks. How come? Did WindRiver buy one of these > OSs from some other company.. in a bid to kill that OS soon? WindRiver bought Psos. Guess why. > > 2) How does ChorusOS compare with VxWorks in terms of popularity (i.e > is it widely used? ) > It's the first time I've heard of ChorusOS. So, it must be less popular that VxWorks (population size for these statistics = 1 :) > 3) Some embedded OSs have multiple-process concept (seperate address > space for each process, which implies that we need hardware MMU. I see > that ChorusOS and LynxOS have this feature), while many dont (like > VxWorks, Psos). > Does this mean that such multiple-process OSs are meant for > high-end equipment alone? (i.e those that can afford to have hardware > MMU) Don't think so. The last incarnation of VxWorks (AE) supports multiple processes (I think.) And, nowadays, there are cheap chips with MMU, many of them supported by VxWorks 5.x. But, for many of these microprocessors, an OS with only "threads" instead of "processes" is usually more fast in task switching activities. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Vs other embedded OSs Date: 30 Apr 2002 14:59:43 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <1cdb3add.0204291306.2f4ebd6c@posting.google.com> <74c2e5b4.0204300849.17877357@posting.google.com> Sender: bpringlemeir@DeadDuck Srivathsa wrote in message <1cdb3add.0204291306.2f4ebd6c@posting.google.com>... >> Hi, I am new to embedded OS' Just want to know how VxWorks >> compares against other embedded OS'. >> 1) Psos seems to be as popular as VxWorks. But WindRiver seems to >> own both Psos and VxWorks. How come? Did WindRiver buy one of >> these OSs from some other company.. in a bid to kill that OS >> soon? Ignacio> WindRiver bought Psos. Guess why. These questions are usually very stupid. A little searching on the Internet will produce tonnes of links to different Embedded OS's. The features that an OS offer will fit your organizations goals. However, they will most often be political. Ergo, this is a dumb question that usually denigrates. As such I consider it flame-bait material. >> 2) How does ChorusOS compare with VxWorks in terms of popularity >> (i.e is it widely used? ) Ignacio> It's the first time I've heard of ChorusOS. So, it must be Ignacio> less popular that VxWorks (population size for these Ignacio> statistics = 1 :) Since you are the 2nd person to respond like this, ChorusOS was made in France (afaik). In the `Java Hay Day', Sun bought Chorus and they were trying to convert Chorus to JavaOs. This never really took. Chorus is certainly more popular in Europe, especially France. >> 3) Some embedded OSs have multiple-process concept (seperate >> address space for each process, which implies that we need [snip] Questions like these make it seem very much like the OP is out of his depth... but that is just a little pernicious. Perhaps you could spend, oh a day, a week, or so, looking around the Internet. You may come back with questions on RTEMS, uCos, eCos, FreeBSD, NetBSD, Hard Hat Linux, ucLinux, etc. Currently, I don't think the OP is prepared for a discussion on the topic. It takes at least some problem in order to be conversant on the problem domain. The most important thing is the application and organization where the OS will be used. That said, unless you have lost of PSOS code, I wouldn't go forward with that OS. Regards, Bill Pringlemeir. ps, I am not being mean, really! - -- Have you ever wondered if it was possible to have a subway make annoying phone calls to moldy milk? Or been forced by your arm to eat bananas? Or been kissed by irate newspaper salesmen while living on a planet where you lost your television? You will. And the company that will bring it to you: AT&T. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: file upload Date: Mon, 29 Apr 2002 16:00:49 +0800 From: "eric" Organization: DCI HiNet Message-ID: Hi! Everybody: I am use tornado web for arm7 and want upload file. I am include user rpm - file upload and setting some instruction : httpUploadConfInit(100); httpUploadConfAdd("/fileset","C:\test"); httpRpmConfAdd (HTTP_M_POST, "/fileset", httpRpmUpload);
When I upload a file through IE browser but it will appear error message : Data abort Exception address: 0x00000004 Current Processor Status Register: 0x20000013 Task: 0xf4ce28 "tHTTPd" Please give me some advice !! Regards, Eric, Cheng --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Vs other embedded OSs Date: 30 Apr 2002 12:48:08 -0700 From: sriva_t@yahoo.com (Srivathsa) Organization: http://groups.google.com/ Message-ID: <1cdb3add.0204301148.1a967dd3@posting.google.com> References: <1cdb3add.0204291306.2f4ebd6c@posting.google.com> Werner Schiendl wrote: > > Hi, > > > > VxWorks comes in two flavors: the 'old' VxWorks with one shared, flat > address space, all code running with full priviledges. This version > definitely runs on very low performance CPUs. > > The other thing is VxWorks AE, which has pretty much the same API, but > supports for a process model. If I remember right, they name it protection > domains and claim all kinds of magic for their concept, but basically its > just a process model, including resource cleanup and memory protection. I > don't know the exact hardware requirements, but it seems it doesn't make > much sense without a MMU. > I really wonder what embedded applications need the process model. I have seen a few PRETTY BIG APPLICATIONs (by my experience): routers. And they are all based on OS's that dont use the process model. Within these applications, there are logically independent entities that could be modelled as seperate processes in a general purpose OS... but because it is an embedded platform, they have been modelled as threads (faster context switching etc). Which makes me wonder: What embedded applications need the process model? Any examples from the real world? - -Srivathsa --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Vs other embedded OSs Date: 30 Apr 2002 16:29:58 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <1cdb3add.0204291306.2f4ebd6c@posting.google.com> <1cdb3add.0204301148.1a967dd3@posting.google.com> Sender: bpringlemeir@DeadDuck Srivathsa> Which makes me wonder: What embedded applications need the Srivathsa> process model? Any examples from the real world? The answer is, "what ever embedded app can afford it". MMU's are quickly becoming common on all RTOS candidate CPUs. This is because as clock speeds increase, you must cache memory and an MMU is necessary to mark memory sections as cacheable, writeable, etc. You have alluded to a performance hit. This is certainly true in many cases. You also wouldn't want to have virtual memory (VM) on an embedded app. However, because previous routers haven't had a heavy process model, don't suspect that future routers will not. The separation of processes via memory protection provides a layer of protection/debugging that can aid development of a large project. Also, depending on the CPU, a process switch might be as fast or faster than an RTOS switch depending on register banking, etc. Certainly an MMU can allow greater control of memory and hence cache effects. Routers will continue to expand on functionality and at some point they may convert to a system that uses memory protection. So embedded apps would use a `process model' if the cost was little or none. Regards, Bill Pringlemeir. - -- Baranyi Blowfish IRA Verisign Marxist Craig Livingstone INS JPL PLO virus morse Belknap kibo Saddam Hussein Skipjack vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Lightweight heap monitor for VxWorks 5.4 / Tornado 2.0 Date: Tue, 30 Apr 2002 16:05:54 -0500 From: "John Poplett" Organization: MindSpring Enterprises Message-ID: Hi, I've done some trolling for a lightweight heap monitor for VxWorks and haven't found anything that ready-made that is portable. I am aware of Hwa Jin Bae's vxmalloc.c, which is partially based on Doug Lea's malloc.c and am aware of suggestions on the FAQ sites: http://www.xs4all.nl/~borkhuis/vxworks/vxw_pt6.html#6.2 Based on suggestions in WRS's application note on heap management: http://web1.windriver.com/windsurf/appnotes/ide/tornado/t2/APN116/appnote116 .pdf the cleanest solution seems to be to use the memPart* subroutines and layer a malloc library on top of them. From there it would be easy to install a malloc library with built-in monitoring or merge in something like Walter Bright's mem package. Keep in mind that I need to have a solution that is portable (one that would work on PPC and othe target architectures in addition to SPARCSIM and SIMPC) and that completely replaces the C language free / malloc interface (compile time macro swizzling is not an option since we don't have source to all the components we use). On a Tornado 2.02 system (W2K / vxsimpc), I see that memPartLib.o embeds definitions for malloc and free: [C:/Tornado/target/lib] ../../host/x86-win32/bin/nmsimpc.exe --defined memPartLib.o 00000000 b .bss 00000000 d .data 00000000 t .text 00000000 t ___gnu_compiled_c 000009e8 T _free 000009d4 T _malloc 000009bc T _memAddToPool 0000089c t _memAlignedBlockSplit 00000018 D _memDefaultAlignment 00000190 T _memPartAddToPool ... With malloc and free bound in the same object as the memPart routines, is there any hope of providing a malloc / free replacement portably? Thanks, John --------------------------- Newsgroups: comp.os.vxworks Subject: Re: INCLUDE_SNMP Date: Tue, 30 Apr 2002 05:51:39 GMT From: Mike Anderson Organization: RoadRunner - Cox Message-ID: <3CCE30CE.5050400@theptrgroup.com> References: Ofer Goren wrote: > Hi. > > How does the project facility of T2 "know" that the SNMP component is not > installed? there is no special indication for this component in file > 00network.cdf. > > Thanks, > > -- > Ofer Goren > Iamba Technologies > > There is a two piece strategy here. The component (the .cdf) describes the object module that's needed. Then, when you switch to the VxWorks tab, it "verifies" the environment by checking the library against the components. This is how a component shows up as there but unavailable (italized). HTH, Mike --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Porting code from BSD to Vxworks Date: 29 Apr 2002 23:28:52 -0700 From: li.mingzheng@mail.zte.com.cn (Li Ming Zheng) Organization: http://groups.google.com/ Message-ID: <15dcd3bc.0204292228.5c338734@posting.google.com> References: <15dcd3bc.0204220004.1b8e316c@posting.google.com> "Hwa Jin Bae" wrote in message news:... > Some information is available at > http://www.xs4all.nl/~borkhuis/vxworks/ftp/vxhacks/pso_stack/README very cool! Thank you so much! But we need to know which release of Vxworks your protocol stack is based on so that we can install and test your stack. Thanks --------------------------- Newsgroups: comp.protocols.snmp,comp.os.vxworks Subject: Envoy version in WindNet SNMP Date: Tue, 30 Apr 2002 09:47:43 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: ========================================================= Tornado : Version 2.0.2 VxWorks : Version 5.4.2 BSP : Version 1.1/1 - ------------------------ Compiler : ccsimpc gcc version egcs-2.90.29 tornado 2.0 980515 (egcs-1.0.3 release) - ------------------------ Wind River Systems Emissary SNMP MIB Compiler, version 7.0. - ------------------------ Host : Windows2000 Target : VxSim for Windows (VxWorks target simulator) ========================================================= How can we know which version of Envoy (8.x, 9.x, etc) is used in WindNet SNMP ? - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Lightweight heap monitor for VxWorks 5.4 / Tornado 2.0 Date: Wed, 01 May 2002 02:05:15 GMT From: Pete Kockritz Organization: AT&T Broadband Message-ID: References: In article , "John Poplett" wrote: > > On a Tornado 2.02 system (W2K / vxsimpc), I see that memPartLib.o embeds > definitions for malloc and free: > ... > > With malloc and free bound in the same object as the memPart routines, is > there any hope of providing a malloc / free replacement portably? There are ways. You can use a binary editor to change the name from _malloc to something like _xyzloc in the library file. You can then supply your own implementation of those functions. Objcopy allows you to rename symbols; extract the .o, rename _malloc and _free and add it back to the archive. Or, you could ask your WRS FAE for the source to memPartLib. Here is some additional info on replacing malloc: - -- Pete Kockritz --------------------------- Newsgroups: comp.os.vxworks Subject: Re: affordable vxworks training Date: Wed, 01 May 2002 03:31:30 GMT From: "D. David Pirzadeh" Message-ID: References: Got great information from Mike McCullough president of MCC Engineering mike@mccengineering.com 1.Download tornado prototyper 2.get vxworks manuals online 2.play with the code, has a simulator 3.take the class at winriver 4.access winsurf (their support) 5.dowload as many sample modules, will be reusing them again and again --------------------------- Newsgroups: comp.os.vxworks Subject: How reliable is routeLib in Vxworks Sim? Date: 30 Apr 2002 06:11:44 -0700 From: overtones@excite.com (Strawberry) Organization: http://groups.google.com/ Message-ID: <41440750.0204300511.32773ec1@posting.google.com> hi folks, I want to know if anybody has faced problems in the routeLib that comes with Vxworks Sim. I am not getting the RTM_ADD kernel event when i use RouteAdd or mRouteAdd. But when i write my own API to add route using routing socket i get the event. Any idea why?? Also any pointers to reliability of ifLib functions viz., ifAddrAdd are also welcome. regards, Strawberry (my fav..) --------------------------- Newsgroups: comp.os.vxworks Subject: resourses for realtime Date: 30 Apr 2002 21:22:57 -0700 From: alphadog_78@yahoo.com (jim canning) Organization: http://groups.google.com/ Message-ID: <57bd77b5.0204302022.28888e8f@posting.google.com> HI ALL, I'M NEWBIE TO REAL TIME OPERATING SYSTEMS AND I'M STILL A STUDENT AND I WANT TO WORK ON REALTIME OPERATING SYSTEMS AND WANNA KNOW ABOUT THAT STUFF CAN ANYBODY GIVE INFORMATION ABOUT REALTIME OS THAT CAN BE AVAILABLE FOR FREE ON THE NET AND CAN BE DOWNLOADED FOR FREE SO THAT I CAN GET SOME EXPERIENCE I REALLY APPRECIATE IT IF YOU CAN SUGGEST ME THANX JIM --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does windml 2.0 font library support japanese and chinese fonts Date: 30 Apr 2002 10:46:08 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <32098fec.0204300619.bb7490e@posting.google.com> Sender: bpringlemeir@DeadDuck Avik> Dear all, I need to know whethet windml 2.0 have existing font Avik> libray for japanese and chinese fonts.If japanese and chinese Avik> font libary is there , then how I can get those font Avik> libraries.you can send your reply to shibasis@rebaca.com. Hope Avik> to get reply soon. Kanji fonts (etc) do not come with WindML 2.0. There are utilities to convert BDF and Windows fonts to the UGL (WindML) format. Font licensing would be up to you. Regards, Bill Pringlemeir. - -- TORPEDOED BY SMALL GNATS 12W 23N. SINKING. U-741. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Vs other embedded OSs Date: 30 Apr 2002 22:12:19 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0204302112.6ccde94a@posting.google.com> References: <1cdb3add.0204291306.2f4ebd6c@posting.google.com> <1cdb3add.0204301148.1a967dd3@posting.google.com> Hello, > > The other thing is VxWorks AE, which has pretty much the same API, but > > supports for a process model. If I remember right, they name it protection > > domains and claim all kinds of magic for their concept, but basically its > > just a process model, including resource cleanup and memory protection. I > > don't know the exact hardware requirements, but it seems it doesn't make > > much sense without a MMU. > > > > I really wonder what embedded applications need the process model. I wasn't going to enter this discussion, but since it seems to have concluded that VxWorks AE is a process model, and therefore has poor performance, I thought I'd add some comments. 1) VxWorks AE protection domains are not the same as processes, at least not in the Unix/Windows sense of the word. They are passive containers, not schedulable entities. VxWorks AE still schedules at the task level. 2) A process model does not have to mean poor performance context switching. It is often down to smart management of the context information, and correct usage of the MMU's address space support features. Task context switches on most CPUs are unaffected by the addition of protection domains. How? By utilising features in the MMU designed to support a full process model, it is possible to avoid doing any more work on context switches for most CPUs. The exception is the case where the cache is indexed by virtual address, and offers no other tricks to optimise the process. In those cases, some context switches will require a cache flush. > I have seen a few PRETTY BIG APPLICATIONs (by my experience): routers. > And they are all based on OS's that dont use the process model. Indeed, many internet infrastructure devices can be built very successfully with VxWorks 5.x. I guarantee that by the time you read this message it will have passed through at least one box running VxWorks 5.x. The core functionality of these devices may well make more sense in the kernel for absolute performance (especially if it is processing every packet). But what about that web server used for configuration purposes? Could that not be isolated in case it crashes when somebody enters silly values, or a malicious person tries to hack it? > Which makes me wonder: What embedded applications need the process > model? Any examples from the real world? Need it? Perhaps few, but many can benefit from the feature, and the benefits are seen not only in the deployed system, but also during development... what about those null pointers that everybody has encountered at some point? How about the stack overflows that can occur? Much quicker to track down if the MMU traps them as soon as they happen :-) HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PCI-to-PCI bridge x86 Date: Tue, 30 Apr 2002 17:03:24 +0200 From: "Werner Schiendl" Message-ID: <3cceb232$1@brateggebdc5.br-automation.co.at> References: <3CC8E942.5B1E@despam.autobahn.mb.ca> <3CC99E3A.720F@despam.autobahn.mb.ca> <3ccd145d$1@brateggebdc5.br-automation.co.at> <3CCDAB3F.341B@despam.autobahn.mb.ca> Hi, "H. E. Taylor" wrote in message news:3CCDAB3F.341B@despam.autobahn.mb.ca... > In article <3ccd145d$1@brateggebdc5.br-automation.co.at>, > Werner Schiendl wrote: > >> > >>[...] > > > > Be sure that you have 'Plug and Play OS installed' or similar option of > > your BIOS _dis_abled! > > If this option is enabled, the BIOS will only setup devices required for > > booting (disk controller, graphics controller, keyboard). > > If it is disabled, as suggested, most BIOS will setup all devices it can > > possibly. > > > > Hi Werner, > Yes but, the BIOS would still set up any bridges it found, > wouldn't it? > To be honest, I don't know this. Maybe someone else can shed light on that. I never had a PCI bridge in a VxWorks device so far (or it did work and so I didn't bother ;-)) best regards Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: file upload Date: 30 Apr 2002 22:24:11 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0204302124.7ab52eed@posting.google.com> References: Hello, "eric" wrote in message news:... > Hi! Everybody: > I am use tornado web for arm7 and want upload file. I am include user > rpm - file upload and setting some instruction : > httpUploadConfInit(100); > httpUploadConfAdd("/fileset","C:\test"); This is almost certainly wrong. The path C:\test is very unlikely to be something that VxWorks will use. Don't be fooled by the fact that the Tornado WindSh shell sees files that way - it is a host based shell that uses the host file system. In this case, the compiler will be turning \t into a tab character as well, which is almost certainly not what you intended... > httpRpmConfAdd (HTTP_M_POST, "/fileset", httpRpmUpload); > > > When I upload a file through IE browser but it will appear error message : > Data abort > Exception address: 0x00000004 > Current Processor Status Register: 0x20000013 > Task: 0xf4ce28 "tHTTPd" > > Please give me some advice !! First, you should get a stack trace from the task that crashed (see the tt function). It may not work though since the exception address was down near zero, but it is worth a shot. That might give you an address that will help you track down the cause of the problem. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado trees and version control Date: 30 Apr 2002 08:20:49 -0700 From: Francoys_Crepeau@Mitel.com (Francoys Crepeau) Organization: http://groups.google.com/ Message-ID: I'm looking at setting up Tornado 2 for different teams in an NT environment. These teams need different configurations, (e.g.: some use T2CP4, some use T2CP3, some use DosFS2, some don't, etc.) and it's very difficult to have everybody in a team use the same configuration. What I want to do is to put these different Tornado tree configurations under version control, so that a user could check out the appropriate Tornado tree. For that reason, I've been following the T2/PVCS threads, and it appears that this is feasible, and in fact recommended. I've pasted a paragraph about this under here. My question is the following: I'm not fully clear on "you have to run the Tornado installer after extracting the complete tree...". Can someone clarify this for me? From what I understand, installing a tree from the distribution CD is what the installer does. So, once I've restored a Tornado tree, and adjusted the contents of the torVars.bat file, how exactly would I "run the Tornado installer" so as to obtain a working Tornado installation? Thanks! Françoys Crépeau Mitel Networks mailto:Francoys_Crepeau@Mitel.com /*** FROM "Tornado 2 and PVCS" thread: >We put the complete Tornado tree into CVS, so including host, docs, etc. >The only file to change after install is torVars.bat, you have to put the >right paths in there, but after that no problems. But to be sure you hav to >run Tornado installer after extracting the complete tree for the first >time, but not a complete install, but point it to the directory you just >installed Tornado in. > >Groeten, > Johan ***/ --------------------------- Newsgroups: comp.os.vxworks Subject: deleting netPools created through netBufLib Date: 30 Apr 2002 08:52:41 -0700 From: rvandiepen@availnetworks.com (Ron v.) Organization: http://groups.google.com/ Message-ID: <5c824911.0204300752.50367e5e@posting.google.com> I have come across what appears to be a hole in using netPools. In my application, a previously initialized netPool is being deleted while resources (MBLKS/CLBLKS/clusters) being managed by this pool are "in use". netBufLib happily cleans up all the free resources in the pool and then wipes out (zeroes) the netPool structure. Any subsequent attempt to free the "in use" resources may result in a system crash since the netPool associated with the freed resource is now invalid. It would seem that the netBufLib routines neither detect nor guard against such action. Does anyone know of any mechanism to resolve this? I searched WindSurf for SPRs relating to this but came up empty... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Vs other embedded OSs Date: Tue, 30 Apr 2002 17:43:28 +0200 From: "Werner Schiendl" Message-ID: <3ccebb96$1@brateggebdc5.br-automation.co.at> References: <1cdb3add.0204291306.2f4ebd6c@posting.google.com> Hi, some comments are inlined intor your questions: "Srivathsa" wrote in message news:1cdb3add.0204291306.2f4ebd6c@posting.google.com... > Hi, > I am new to embedded OS' > Just want to know how VxWorks compares against other embedded OS'. > > 1) Psos seems to be as popular as VxWorks. But WindRiver seems to > own both Psos and VxWorks. How come? Did WindRiver buy one of these > OSs from some other company.. in a bid to kill that OS soon? > pSOS has been bought by Windriver some time ago. I would definitely not start a new development with pSOS. In my opinion, pSOS will disappear rather sooner than later. > 2) How does ChorusOS compare with VxWorks in terms of popularity (i.e > is it widely used? ) > No idea about that, so no comment. > 3) Some embedded OSs have multiple-process concept (seperate address > space for each process, which implies that we need hardware MMU. I see > that ChorusOS and LynxOS have this feature), while many dont (like > VxWorks, Psos). > Does this mean that such multiple-process OSs are meant for > high-end equipment alone? (i.e those that can afford to have hardware > MMU) > VxWorks comes in two flavors: the 'old' VxWorks with one shared, flat address space, all code running with full priviledges. This version definitely runs on very low performance CPUs. The other thing is VxWorks AE, which has pretty much the same API, but supports for a process model. If I remember right, they name it protection domains and claim all kinds of magic for their concept, but basically its just a process model, including resource cleanup and memory protection. I don't know the exact hardware requirements, but it seems it doesn't make much sense without a MMU. hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: resourses for realtime Date: Wed, 01 May 2002 06:46:34 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <57bd77b5.0204302022.28888e8f@posting.google.com> I would suggest you stop shouting and perhaps search among the ample resources on the Web. A good starting point would be a Google advanced search with the words "realtime" and "tutorial" and the phrase "operating system". BTW, you'll understand realtime systems better if you understand operating systems in general. Same concepts, different requirements and implementations. "jim canning" wrote in message news:57bd77b5.0204302022.28888e8f@posting.google.com... > HI ALL, > I'M NEWBIE TO REAL TIME OPERATING SYSTEMS AND I'M STILL A > STUDENT AND I WANT TO WORK ON REALTIME OPERATING SYSTEMS AND WANNA > KNOW ABOUT THAT STUFF > > CAN ANYBODY GIVE INFORMATION ABOUT REALTIME OS THAT CAN BE AVAILABLE > FOR FREE ON THE NET AND CAN BE DOWNLOADED FOR FREE SO THAT I CAN GET > SOME EXPERIENCE > > I REALLY APPRECIATE IT IF YOU CAN SUGGEST ME > > THANX > JIM --------------------------- Newsgroups: comp.os.vxworks Subject: #ifdef question Date: Wed, 01 May 2002 11:07:56 +0200 From: Leon Gubbels Organization: =?iso-8859-1?Q?Oc=E9?= Technologies B.V. Message-ID: <3CCFB06C.196EF97B@oce.nl> Hi, for compilation for target (VxWorks) we now use #ifndef _WIN32 with the GCC compiler. But to be more specific this should be something like #ifdef VXWORKS ... I cannot find this statement in the manuals. Also the compiler (cygnus-2.7.2-960126) doesn't recognize some variants I tried: #ifdef _VXWORKS #ifdef VXWORKS #ifdef _vxworks #ifdef vxworks #ifdef _VxWorks #ifdef VxWorks does anyone know what it should be? Thanx in advance. Leon (if you reply to my mail address, remove NOSPAM). --------------------------- Newsgroups: comp.os.vxworks Subject: Re: deleting netPools created through netBufLib Date: 1 May 2002 02:28:09 -0700 From: demralarood@hotmail.com (Demrala Rood) Organization: http://groups.google.com/ Message-ID: <3baeb99.0205010128.66b318d8@posting.google.com> References: <5c824911.0204300752.50367e5e@posting.google.com> rvandiepen@availnetworks.com (Ron v.) wrote in message news:<5c824911.0204300752.50367e5e@posting.google.com>... > I have come across what appears to be a hole in using netPools. In my > application, a previously initialized netPool is being deleted while > resources (MBLKS/CLBLKS/clusters) being managed by this pool are "in > use". netBufLib happily cleans up all the free resources in the pool > and then wipes out (zeroes) the netPool structure. Any subsequent > attempt to free the "in use" resources may result in a system crash > since the netPool associated with the freed resource is now invalid. > It would seem that the netBufLib routines neither detect nor guard > against such action. Does anyone know of any mechanism to resolve > this? I searched WindSurf for SPRs relating to this but came up > empty... I think the solution is don't free resources that are in use. Since you discuss netpool, I think you must be trying to unload a driver which is not shut down. There won't be an SPR for this. When pruning a tree, would you cut off the limb you stand upon? DR --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #ifdef question Date: Wed, 01 May 2002 06:23:29 -0400 From: joe durusau Organization: Bellsouth.Net Message-ID: <3CCFC221.7B0918A9@bellsouth.net> References: <3CCFB06C.196EF97B@oce.nl> Leon Gubbels wrote: > Hi, for compilation for target (VxWorks) we now use #ifndef _WIN32 with > the GCC compiler. But to be more specific this should be something like > #ifdef VXWORKS ... > I cannot find this statement in the manuals. Also the compiler > (cygnus-2.7.2-960126) doesn't recognize some variants I tried: > #ifdef _VXWORKS > #ifdef VXWORKS > #ifdef _vxworks > #ifdef vxworks > #ifdef _VxWorks > #ifdef VxWorks > > does anyone know what it should be? > > Thanx in advance. > > Leon (if you reply to my mail address, remove NOSPAM). Could you explain exactly what you are trying to do? If you include the right header files, vxWorks will compile OK. You don't have an intrinsic need for such a definition, but you can make one up yourself. As to your question about variations, they are not allowed. You have to decide exactly what you want and why. The compiler is case sensitive. The comiler itself does not require not rely on #defines to work correctly. One aside: Make sure that you differentiate between the stuff you get with the simulator and the real thing. There are quite a few differences. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #ifdef question Date: Wed, 01 May 2002 13:02:48 +0200 From: Leon Gubbels Organization: =?iso-8859-1?Q?Oc=E9?= Technologies B.V. Message-ID: <3CCFCB58.71DBFB3E@oce.nl> References: <3CCFB06C.196EF97B@oce.nl> <3CCFC221.7B0918A9@bellsouth.net> joe durusau wrote: > > Leon Gubbels wrote: > > > Hi, for compilation for target (VxWorks) we now use #ifndef _WIN32 with > > the GCC compiler. But to be more specific this should be something like > > #ifdef VXWORKS ... > > I cannot find this statement in the manuals. Also the compiler > > (cygnus-2.7.2-960126) doesn't recognize some variants I tried: > > #ifdef _VXWORKS > > #ifdef VXWORKS > > #ifdef _vxworks > > #ifdef vxworks > > #ifdef _VxWorks > > #ifdef VxWorks > > > > does anyone know what it should be? > > > > Thanx in advance. > > > > Leon (if you reply to my mail address, remove NOSPAM). > > Could you explain exactly what you are trying to do? If you include > the right header files, vxWorks will compile OK. You don't have > an intrinsic need for such a definition, but you can make one up yourself. > > As to your question about variations, they are not allowed. You have to > decide exactly what you want and why. The compiler is case sensitive. > The comiler itself does not require not rely on #defines to work correctly. > > One aside: Make sure that you differentiate between the stuff you get > with the simulator and the real thing. There are quite a few differences. Well, I meant, my code should work for a target (Tornado 2) and simulation (Windows NT) environment, so I use defines to make sure the comipler understands environment specific code. e.g. Tornado understands when I say: sysClkRateGet(); NT does not. So: #ifndef _WIN32 long l_TicksPerSecond = sysClkRateGet(); #else long l_TicksPerSecond = 500; #endif But this does not satisfy me. So I want to know, is there another define for VxWorks, so I can be more specific than #ifndef _WIN32 ?? Leon. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 1 06:00:56 2002 From: "Shah, Ranjana" Date: Wed May 1 06:00:59 PDT 2002 Subject: Re: #ifdef question There may or may not be a define in a VxWorks header file, but one way to accomplish what you want (we also do this) is in your build process, for example in the makefile, use DEFINES = -D_VXWORKS_ when you are doing a build for VxWorks target. And then use the #ifdef _VXWORKS_ in your source code. > Hi, for compilation for target (VxWorks) we now use #ifndef _WIN32 with > the GCC compiler. But to be more specific this should be something like > #ifdef VXWORKS ... > I cannot find this statement in the manuals. Also the compiler > (cygnus-2.7.2-960126) doesn't recognize some variants I tried: > #ifdef _VXWORKS > #ifdef VXWORKS > #ifdef _vxworks > #ifdef vxworks > #ifdef _VxWorks > #ifdef VxWorks > does anyone know what it should be? > Thanx in advance. > Leon (if you reply to my mail address, remove NOSPAM). Ranjana Shah From vxwexplo-errs@csg.lbl.gov Wed May 1 11:50:28 2002 From: Shiv Date: Wed May 1 11:50:30 PDT 2002 Subject: TMD Driver : Target server problems for the 8xx Does anyone have any experience with the TMD driver for the PPC8xx series (namely the 855T). I can get the target server up and running with the 8260 but not for my target board which uses the 855T ? All I keep getting is this : Connecting to target agent... succeeded. Attaching C++ interface... succeeded. Attaching elf OMF reader for PPC CPU family... succeeded. Error: rpccore backend client Timed out Error: Error performing target core file checksum. Warning: Core file checksums do not match. Error: rpccore backend client Timed out Error: rpccore backend client Timed out Error: rpccore backend client Timed out Error: Object module load failed for C:\Tornado2\target\proj\TMDriverProject\default/vxWorks Error: loading core symbol table: status = 0x100de Error during target server core initialization Target Server will exit Am I missing something ? Shiv From vxwexplo-errs@csg.lbl.gov Thu May 2 04:03:16 2002 From: Vxworks Exploder Date: Thu May 2 04:03:18 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu May 2 04:03:11 PDT 2002 Subject: Re: convert vxworks_rom.hex Subject: netBufLib questions Subject: Booting multiple interfaces Subject: Mutual exclusion between same-priority tasks Subject: Re: #ifdef question Subject: Re: Mutual exclusion between same-priority tasks Subject: Re: Mutual exclusion between same-priority tasks Subject: Re: bootrom not loading kernel Subject: Re: Mutual exclusion between same-priority tasks Subject: WindRiver VisionEvent Setup Subject: RTOS project Subject: Re: #ifdef question Subject: RAM Disk Subject: Re: #ifdef question Subject: Re: Mutual exclusion between same-priority tasks Subject: PPP header compression Subject: Re: Does windml 2.0 font library support japanese and chinese fonts Subject: how to use different frequency to run tasks Subject: Re: how to use different frequency to run tasks Subject: Difference between RTOS and normal OS Subject: Re: RAM Disk Subject: getting the base address of the PCI card Subject: Re: Mutual exclusion between same-priority tasks Subject: Target Type for GCC Cross Compiler and Binutils: VxSim on Windows NT. Subject: Re: Target Type for GCC Cross Compiler and Binutils: VxSim on Windows NT. Subject: Re: convert vxworks_rom.hex Subject: Re: getting the base address of the PCI card Subject: Re: resourses for realtime ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: convert vxworks_rom.hex Date: 1 May 2002 05:10:07 -0700 From: bala@innocor.com (balendran bala) Organization: http://groups.google.com/ Message-ID: References: <3CCDD5C0.54DEFA39@bellsouth.net> Hi Joe, I lost the source code for a function. Since I have the vxworks_rom.hex from last build (includes the function), so I am trying to findout that I could do reverse-engineering and recover the function. Thank you, Bala joe durusau wrote in message news:<3CCDD5C0.54DEFA39@bellsouth.net>... > balendran bala wrote: > > > Hi guys, > > > > Is there anyway that I could convert vxworks_rom.hex file into a text file? > > > > Thank you, > > bala > > If you look at it, it _is_ a text file. What do you want to do??? > > Speaking only for myself, > > Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: netBufLib questions Date: Wed, 1 May 2002 13:55:59 +0000 (UTC) From: "Dexter Nixon" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <2b0e7fc437bdc7ad995977ba48739431.76014@mygate.mailgate.org> Hi, I am reading the Network Tool kit guide of VxWorks. Here the pool of clusters we are allocating and managing it by means of mBlk and Clblk structures right. But it is given that access to these structures are not semaphore protected. THen how does this will work. Because if am implementing more than one network protocol using MUX and am initialising a memory pool, and if both protocols access the same mem pool, how can we access mutual exclusion is there when the two protocols acccess the same pool with both are running as different threads. Without semaphore protection internally within the structures how this is possible. anybody kindly explain me. Thanks, Nix - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Booting multiple interfaces Date: Wed, 1 May 2002 14:28:01 +0000 (UTC) From: "Dexter Nixon" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: Hi, This was the typical booting example given in the manual. Here we are configuring a Lance Ethernet card while booting itself with this bootline ln(0, 0) bear:/usr/wpwr/target/config/mz7122/vxworks e=90.0.0.2 b=91.0.0.2 h=100.0.0.4 g=90.0.0.3 u=papa pw=biggrump f=0x80 tn=goldilox s=bear:/usr/papa/startup o= Suppose, i have multiple ethernet cards like this means, how can i configure those interfaces. What is the way to do it. THanks, Nix - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Mutual exclusion between same-priority tasks Date: Wed, 01 May 2002 14:23:52 GMT From: Michael Patrick Organization: ATT Broadband Message-ID: <3CCFFB03.BA438CA4@attbi.com> How many of you are avoiding the use of semaphores for protecting global variables because you think vxworks will schedule (with timeslicing disabled) same-priority tasks to always run to completion until they voluntarily give up context? I came across this design technique when reviewing a project in my company, and submitted a TSR to ask Windriver about it. I was worried, you see, that an interrupt to one task would return through the scheduler to a *different* same-priority task, so that same-priority tasks would NOT be mutually exclusive. PSOS can do this, for example. I was surprised by Windriver's response. They said that although same-priority task mutual exclusion is implemented now, there was "no requirement" for it to operate this way, and it is "not guaranteed" to continue to do so in future versions. They said that semaphores were required for mutual exclusion even between same-priority tasks. I'm curious as to how many existing Vxworks customers are counting on mutual exclusion between same-priority tasks. Perhaps we should get Windriver to formerly add a requirement to provide this in future versions? Regards, - -mike --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #ifdef question Date: Wed, 01 May 2002 10:37:19 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3CD027CF.35A5FF7@lmco.com> References: <3CCFB06C.196EF97B@oce.nl> <3CCFC221.7B0918A9@bellsouth.net> <3CCFCB58.71DBFB3E@oce.nl> Leon Gubbels wrote: > > joe durusau wrote: > > > > Leon Gubbels wrote: > > > > > Hi, for compilation for target (VxWorks) we now use #ifndef _WIN32 with > > > the GCC compiler. But to be more specific this should be something like > > > #ifdef VXWORKS ... > > > I cannot find this statement in the manuals. Also the compiler > > > (cygnus-2.7.2-960126) doesn't recognize some variants I tried: > > > #ifdef _VXWORKS > > > #ifdef VXWORKS > > > #ifdef _vxworks > > > #ifdef vxworks > > > #ifdef _VxWorks > > > #ifdef VxWorks > > > > > > does anyone know what it should be? > > > > > > Thanx in advance. > > > > > > Leon (if you reply to my mail address, remove NOSPAM). > > > > Could you explain exactly what you are trying to do? If you include > > the right header files, vxWorks will compile OK. You don't have > > an intrinsic need for such a definition, but you can make one up yourself. > > > > As to your question about variations, they are not allowed. You have to > > decide exactly what you want and why. The compiler is case sensitive. > > The comiler itself does not require not rely on #defines to work correctly. > > > > One aside: Make sure that you differentiate between the stuff you get > > with the simulator and the real thing. There are quite a few differences. > > > > Well, I meant, my code should work for a target (Tornado 2) and > simulation (Windows NT) environment, so I use defines to make sure the > comipler understands environment specific code. e.g. Tornado understands > when I say: sysClkRateGet(); > NT does not. So: > > #ifndef _WIN32 > long l_TicksPerSecond = sysClkRateGet(); > #else > long l_TicksPerSecond = 500; > #endif > > But this does not satisfy me. So I want to know, is there another define > for VxWorks, so I can be more specific than #ifndef _WIN32 ?? > > Leon. You can use whatever you wish, within reason. vxWorks itself does not have any #defines in the header files like that. There is a #define VX_READ in vxWorks.h that would work if you are not using that header in the simulator. I would make up something, like #define TARGET_XXX where XXX is something that you would recognize. There is nothing inherent that will always work. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mutual exclusion between same-priority tasks Date: Wed, 01 May 2002 14:53:38 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3CCFFB03.BA438CA4@attbi.com> Mike. WindRiver is now full of ex-ISI and otherwise incompetent people, so do not regard a TSE's answer seriously, just get him to escalate this TSR to Engineering. I can assure you that all current and past versions of VxWorks will honor Mutex state between any two tasks, i.e. there is no different how Mutex works if the tasks where same or different priority. Whatever is the state of the documentation (i.e. format product requirements), if for some reason the Mutex gets broken, e.g. for equal-riority tasks, the product will fail standard QA tests. A lot of higher layer VxWorks code relies on Mutexes, so the test coverage for this module is plenty. On many architectures, the lowest level primitives including binary and mutex semaphores, are re-written in native assembly language, to precisely immitate the portable "C" version of the same code. HTH, - Leonid "Michael Patrick" wrote in message news:3CCFFB03.BA438CA4@attbi.com... > How many of you are avoiding the use of semaphores for protecting global > variables because you think vxworks will schedule (with timeslicing > disabled) same-priority tasks to always run to completion until they > voluntarily give up context? > > I came across this design technique when reviewing a project in my > company, and submitted a TSR to ask Windriver about it. I was worried, > you see, that an interrupt to one task would return through the > scheduler to a *different* same-priority task, so that same-priority > tasks would NOT be mutually exclusive. PSOS can do this, for example. > > I was surprised by Windriver's response. They said that although > same-priority task mutual exclusion is implemented now, there was "no > requirement" for it to operate this way, and it is "not guaranteed" to > continue to do so in future versions. They said that semaphores were > required for mutual exclusion even between same-priority tasks. > > I'm curious as to how many existing Vxworks customers are counting on > mutual exclusion between same-priority tasks. Perhaps we should get > Windriver to formerly add a requirement to provide this in future > versions? > > Regards, > -mike > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mutual exclusion between same-priority tasks Date: Wed, 1 May 2002 14:57:23 +0000 (UTC) From: "Dexter Nixon" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: References: <3CCFFB03.BA438CA4@attbi.com> "Michael Patrick" wrote in message news:3CCFFB03.BA438CA4@attbi.com > How many of you are avoiding the use of semaphores for protecting global > variables because you think vxworks will schedule (with timeslicing > disabled) same-priority tasks to always run to completion until they > voluntarily give up context? > > I came across this design technique when reviewing a project in my > company, and submitted a TSR to ask Windriver about it. I was worried, > you see, that an interrupt to one task would return through the > scheduler to a *different* same-priority task, so that same-priority > tasks would NOT be mutually exclusive. PSOS can do this, for example. > ======================================================================= Hi Mike, Why do we need mutual exclusion between same priority tasks. I could not find what is the relation between priority and mutual exclusion if both tasks are not accessing a common memory location at their life time. Kindly explain me on this. THanks, Nix ========================================================================== > I was surprised by Windriver's response. They said that although > same-priority task mutual exclusion is implemented now, there was "no > requirement" for it to operate this way, and it is "not guaranteed" to > continue to do so in future versions. They said that semaphores were > required for mutual exclusion even between same-priority tasks. > > I'm curious as to how many existing Vxworks customers are counting on > mutual exclusion between same-priority tasks. Perhaps we should get > Windriver to formerly add a requirement to provide this in future > versions? > > Regards, > -mike - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bootrom not loading kernel Date: Wed, 01 May 2002 11:29:19 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3CD017DF.8EF3935@fnal.gov> References: <3CC9DF96.621B2327@fnal.gov> Johan Borkhuis wrote: > > > [VxWorks Boot]: n dc > > Attaching network interface dc0... done. > > Error setting inet address of dc0 to 0, errno = 0 > > Cannot initialize interface named "dc0" > > Just a quick suggestion (I don't have a board available at this moment): > The error says "Error settint inet addr to 0": did you set a network > address in the boot parameters? Yes I have. Actually, I tried this "n dc" on another board that boots up properly. It also gave the same result. So am not sure what thsi command does or coveys. Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mutual exclusion between same-priority tasks Date: Wed, 1 May 2002 10:00:02 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: <3CCFFB03.BA438CA4@attbi.com> I think Wind River's response is reasonable on this. I don't think that there was a clear intent to have the scheduling of the same priority tasks one way or the other. Yet, I think it is useful to have a task finish the work before allowing another task (of same priority) to take over. How about proposing a new task option bit that can be set? Also, I would use mutex anyway for all shared resources. You can depend on all the tasks accessing the resource to be the same priority tasks. But things can change... especially when system gets very big and other engineers take over the work. I think it is a good practice to defensively program against future misunderstandings. It doesn't always work... :-) - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: WindRiver VisionEvent Setup Date: Wed, 01 May 2002 12:57:42 -0400 From: Babu Yama Organization: Lucent Technologies Message-ID: <3CD01E86.D8BF08BB@lucent.com> Hi, Any body has any good experience in setting up WindRiver VisionEvent? Does it work reliably? 1. Can the VisionEvent run at full 66 MHz bus speed on MPC8260? Do we need to clock down to 33 MHz? 2. Does VisionEvent have separate set of manuals. Is there a soft copy of them anywhere on the net/windsurf web site. (already contacted the windriver support, but......) 3. Can anyone outline a procedure briefly, please. Appreciate your comments, Thanks, Babu --------------------------- Newsgroups: comp.os.vxworks Subject: RTOS project Date: 1 May 2002 10:28:59 -0700 From: neetu_12@yahoo.com (Neena76) Organization: http://groups.google.com/ Message-ID: Hello, I'm looking for a small embedded RTOS programming project for a class. If anyone has some interesting ideas or could point me in the right direction, I would appreciate it. Thanks, Neena --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #ifdef question Date: 1 May 2002 11:10:05 -0700 From: pkockritz@mac.com (Pete Kockritz) Organization: http://groups.google.com/ Message-ID: <9423c55b.0205011010.17227887@posting.google.com> References: <3CCFB06C.196EF97B@oce.nl> <3CCFC221.7B0918A9@bellsouth.net> <3CCFCB58.71DBFB3E@oce.nl> Leon Gubbels wrote in message news:<3CCFCB58.71DBFB3E@oce.nl>... > > Well, I meant, my code should work for a target (Tornado 2) and > simulation (Windows NT) environment, so I use defines to make sure the > comipler understands environment specific code. e.g. Tornado understands > when I say: sysClkRateGet(); > NT does not. So: Here are the pre-defines for gcc/Tornado gcc driver version cygnus-2.7.2-960126 egcs-971225 tornado 2.0 executing gcc version cygnus-2.7.2-960126 -undef -D__GNUC__=2 -D__GNUC_MINOR__=7 -D__IEEE_BIG_ENDIAN -D__embedded__ -D__IEEE_BIG_ENDIAN -D__embedded__ -Asystem(embedded) -Acpu(powerpc) -Amachine(powerpc) -D__CHAR_UNSIGNED__ -D_CALL_SYSV -Amachine(bigendian) -D_ARCH_PPC To contrast, here they are for gcc/cygwin gcc version 2.95.3-5 (cygwin special) -D__GNUC__=2 -D__GNUC_MINOR__=95 -D_X86_=1 -D_X86_=1 -Asystem(winnt) -Acpu(i386) -Amachine(i386) -Di386 -D__i386 -D__i386__ -Di686 -Dpentiumpro -D__i686 -D__i686__ -D__pentiumpro -D__pentiumpro__ -D__stdcall=__attribute__((__stdcall__)) -D__cdecl=__attribute__((__cdecl__)) -D_stdcall=__attribute__((__stdcall__)) -D_cdecl=__attribute__((__cdecl__)) -D__declspec(x)=__attribute__((x)) -D__CYGWIN32__ -D__CYGWIN__ -Dunix -D__unix__ -D__unix gcc will print this info (and other info) with the following option: gcc -E -v - Pete - -- pkockritz at mac dot com --------------------------- Newsgroups: comp.os.vxworks Subject: RAM Disk Date: Tue, 30 Apr 2002 15:01:41 +0800 From: "eric" Organization: DCI HiNet Message-ID: Hi! Everybody: I am use RAM Disk. How to use RAM Disk as DOS fileSystem? I am use - ->cd "/ram0" instruction in tornado shell but it don't work? Please give me advice!! Thank you! Regards, eric cheng --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #ifdef question Date: Wed, 01 May 2002 11:06:41 -0400 From: david lindauer Message-ID: <3CD00481.57DCADF1@notifier-is.net> References: <3CCFB06C.196EF97B@oce.nl> <3CCFC221.7B0918A9@bellsouth.net> <3CCFCB58.71DBFB3E@oce.nl> Leon Gubbels wrote: Well, I meant, my code should work for a target (Tornado 2) and > simulation (Windows NT) environment, so I use defines to make sure the > comipler understands environment specific code. e.g. Tornado understands > when I say: sysClkRateGet(); > NT does not. So: > > #ifndef _WIN32 > long l_TicksPerSecond = sysClkRateGet(); > #else > long l_TicksPerSecond = 500; > #endif > > But this does not satisfy me. So I want to know, is there another define > for VxWorks, so I can be more specific than #ifndef _WIN32 ?? > > Leon. I haven't seen one. But there is nothing stopping you from making your own, e.g. you could put in a specific #define of your own choosing at compile time. (For Tornado 2, select 'builds' in the workspace, then click on the build you want to change and then click on the C/C++ tab and put the macro you want at the end of the window. If working with a makefile, modify the CFLAGS macro in the make file to add your macro. In both cases the macro would be defined with a -D switch, e.g. -DVxWorks). David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mutual exclusion between same-priority tasks Date: 1 May 2002 14:16:31 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205011316.5e33fb21@posting.google.com> References: <3CCFFB03.BA438CA4@attbi.com> Hello Michael, If you are relying on priority alone to ensure mutual exclusion, then I think that you taking chances. The implementor of a priority based pre-emptive scheduler clearly has to make a decision when handling tasks of the same priority. If a task becomes ready, should it be placed at the head of the queue or tail of the queue of tasks with the same priority? Normally, and this is the case for the Wind kernel, the task would be placed at the end of the list. That seems the most logical, especially when people use the phrase "run to completion" to describe the scheduling algorithm (though, one should note, that phrase is not its official name). Will it change in VxWorks? Unlikely. If it isn't broken, why fix it? Should you rely on priority for mutual exclusion? No. There are a number of primitives in all real-time operating systems for ensuring mutual exclusion; use one of them and you'll be safe for the future. For VxWorks, you have the following choices: 1) Mutex semaphores (see semMLib) 2) Pre-emption locking (see taskLock() & taskUnlock()) 3) Interrupt locking (see intLock() & intUnlock()) Which should you use? Depends on your application. From what you have written, I would suspect that (1) or (2) would be the best fit for what you are doing; if you are concerned about semaphores (?), perhaps (2) is something you should look at. I often use a trick here to "hide" the exact implementation here. Define LOCK() and UNLOCK() macros to surround your code with. That way you have the choice of anything from defining them to be nothing, using a semaphore, or using taskLock(). If you think you'll need intLock() ever, you need handle a return from LOCK() and pass it to UNLOCK() though, so think carefully about that one! As for the answer you got from Wind River support; it would seem to be sensible to me. There is no guarantee given - I just checked in manuals, and nowhere that I could see is the behaviour you have described documented. Since that is the case, I would be cautious relying on it if moving to new versions of VxWorks was something I thought I might do. Now, you might decide that the risk is low enough, and the benefit in your application outweighs that risk. If you do come to that conclusion though, I'd make sure the places in the code where this assumption was used were clearly marked (or use the macro trick I described above), just in case... The same applies when using "undocumented" features in Windows, or any other operating system, library etc. HTH, John... > How many of you are avoiding the use of semaphores for protecting global > variables because you think vxworks will schedule (with timeslicing > disabled) same-priority tasks to always run to completion until they > voluntarily give up context? > > I came across this design technique when reviewing a project in my > company, and submitted a TSR to ask Windriver about it. I was worried, > you see, that an interrupt to one task would return through the > scheduler to a *different* same-priority task, so that same-priority > tasks would NOT be mutually exclusive. PSOS can do this, for example. > > I was surprised by Windriver's response. They said that although > same-priority task mutual exclusion is implemented now, there was "no > requirement" for it to operate this way, and it is "not guaranteed" to > continue to do so in future versions. They said that semaphores were > required for mutual exclusion even between same-priority tasks. > > I'm curious as to how many existing Vxworks customers are counting on > mutual exclusion between same-priority tasks. Perhaps we should get > Windriver to formerly add a requirement to provide this in future > versions? > > Regards, > -mike --------------------------- Newsgroups: comp.os.vxworks Subject: PPP header compression Date: Wed, 1 May 2002 18:06:35 -0400 From: "Jason" Organization: Posted via Supernews, http://www.supernews.com Message-ID: Can some one tell me how to change the default PPP header compression under VxWorks? Thanks, Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does windml 2.0 font library support japanese and chinese fonts Date: Wed, 01 May 2002 23:45:53 GMT From: "ChanSeong Mike Kim" Organization: Dacom Message-ID: References: <32098fec.0204300619.bb7490e@posting.google.com> Hi~ I once tried using perl source to convert from BDF to UGL format. I think you should know some source or utility support only one byte code. At that time I converted some bitmap source to UGL format our own developed convert source. So, I'd like to know good convert program. Regards Chanseong Kim "Bill Pringlemeir" wrote in message news:u3cxdqm6n.fsf@yahoo.com... > > Avik> Dear all, I need to know whethet windml 2.0 have existing font > Avik> libray for japanese and chinese fonts.If japanese and chinese > Avik> font libary is there , then how I can get those font > Avik> libraries.you can send your reply to shibasis@rebaca.com. Hope > Avik> to get reply soon. > > Kanji fonts (etc) do not come with WindML 2.0. There are utilities to > convert BDF and Windows fonts to the UGL (WindML) format. Font > licensing would be up to you. > > Regards, > Bill Pringlemeir. > > -- > TORPEDOED BY SMALL GNATS 12W 23N. SINKING. U-741. > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: how to use different frequency to run tasks Date: Wed, 1 May 2002 21:48:59 -0400 From: "Jianyuan HU" Organization: Bell Sympatico Message-ID: <4U0A8.61867$kq1.2108928@news20.bellglobal.com> Hi, I am newie in VxWorks, but I have a lot of experience with AMX (RTOS) for real time controller. I want to switch to VxWorks. Our robot controller has 5 tasks, different tasks has different priority and different execution frequncy. for instance, Task1 needs to run once every 1ms, while task2 run once every 20ms, and so on. taskSpawn can't do this job, Watch Dog Timer can't do either, but I think there should be some function to do this job, for AMX can do that easily, only we need to do is to set the timer for each task. Any info is appreciated. James --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use different frequency to run tasks Date: Wed, 01 May 2002 20:51:41 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3CD0B7CD.D1D6FE93@gv.net> References: <4U0A8.61867$kq1.2108928@news20.bellglobal.com> Reply-To: mkesti@gv.net Jianyuan HU wrote: >Hi, > >I am newie in VxWorks, but I have a lot of experience with AMX (RTOS) for >real time controller. I want to switch to VxWorks. > >Our robot controller has 5 tasks, different tasks has different priority and >different execution frequncy. >for instance, Task1 needs to run once every 1ms, while task2 run once every >20ms, and so on. > >taskSpawn can't do this job, Watch Dog Timer can't do either, but I think >there should be some function to do this job, for AMX can do that easily, >only we need to do is to set the timer for each task. > >Any info is appreciated. Use tasks and watchdog timers to do the job. Have the tasks pend on semaphores that are given in their WDT timeout functions. - -- ======================================================================== 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: Difference between RTOS and normal OS Date: 1 May 2002 21:37:38 -0700 From: vivekm@dd.nec.com.au (Vivek) Organization: http://groups.google.com/ Message-ID: <507efed5.0205012037.3e11728f@posting.google.com> Hi, I am looking for some concrete differences between RTOS and normal OS. Could any body point me to the right direction for this. Thanks and in advance Vivek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RAM Disk Date: 1 May 2002 22:30:07 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205012130.3505647b@posting.google.com> References: Hello, "eric" wrote in message news:... > Hi! Everybody: > I am use RAM Disk. How to use RAM Disk as DOS fileSystem? I am use > ->cd "/ram0" instruction in tornado shell but it don't work? You cannot use the RAM disk directly from the Tornado shell since that is a host based tool, and your file system is on the target. You can add the target based shell if you like, or you can use the '@' prefix to the commands like cd that exist on both (assuming you have the FS shell command utilities on the target - if you don't, then use ioDefPathSet() instead of cd). If you want to copy files to the RAM disk from the host, then you'll need to add the file system utilities to the target. I'd suggest adding the target tools (shell, loader etc) as it will all behave more logically if you are actually working from the target rather than via a host based tool when trying to copy files. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: getting the base address of the PCI card Date: 1 May 2002 23:19:30 -0700 From: forvxworks@rediffmail.com (Anand) Organization: http://groups.google.com/ Message-ID: HI All, first, thanks to Bruce,Andray Kaganovsky and John for helping me on Far Pointers. Now, i have a PCI card which has some memory and registers which i want to access and manipulate the contents of the registers depending on the functionality i want to program. Earlier , i was using an ISA card which had switches on it to select the base address and depending on the base address i could get the address of the registers and i could program the registers according to the function i wanted. but this PCI card does not have any switches, so that i can choose the base address. going through earlier articles in this group, i learned that we can use functions like pciDeviceShow() and pciHeaderShow() , we can have the details. but i am not sure whether the base address i get from pciHeaderShow() can be used as the base address of the card or not. I am new to vxworks, and i hope the question i am asking makes some sense. waiting for replies. Thanks in advance. Regards. Anand. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mutual exclusion between same-priority tasks Date: Thu, 02 May 2002 06:28:01 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3CCFFB03.BA438CA4@attbi.com> Folks, it seems i misunderstood Michael's question, or other responders understood it differentkly. I read the question as referring to Mutex semaphores behaviour on tasks of the sam epiroority, while others beleive that Michael is trying to rely solely on priority for protecting global resources. Michael, priority is a very TEMPORARY attribute of a task, you specify one priority at task intialization, but the task calls subroutines from libraries using Mutex semaphores, which can invert priorities. Also, at later project stages (code reuse) the system engineer can change the initial relative priority of tasks too. Dont take shortcuts, use Mutex samaphores for mutual exclusion. "Leonid Rosenboim" wrote in message news:newscache$muvfvg$rr2$1@lnews.actcom.co.il... > Mike. > > WindRiver is now full of ex-ISI and otherwise incompetent people, > so do not regard a TSE's answer seriously, just get him to escalate > this TSR to Engineering. > > I can assure you that all current and past versions of VxWorks > will honor Mutex state between any two tasks, i.e. there is no > different how Mutex works if the tasks where same or different > priority. > > Whatever is the state of the documentation (i.e. format product > requirements), if for some reason the Mutex gets broken, e.g. > for equal-riority tasks, the product will fail standard QA tests. > A lot of higher layer VxWorks code relies on Mutexes, so the > test coverage for this module is plenty. > > On many architectures, the lowest level primitives including binary > and mutex semaphores, are re-written in native assembly language, > to precisely immitate the portable "C" version of the same code. > > HTH, - Leonid > > > "Michael Patrick" wrote in message > news:3CCFFB03.BA438CA4@attbi.com... > > How many of you are avoiding the use of semaphores for protecting global > > variables because you think vxworks will schedule (with timeslicing > > disabled) same-priority tasks to always run to completion until they > > voluntarily give up context? > > > > I came across this design technique when reviewing a project in my > > company, and submitted a TSR to ask Windriver about it. I was worried, > > you see, that an interrupt to one task would return through the > > scheduler to a *different* same-priority task, so that same-priority > > tasks would NOT be mutually exclusive. PSOS can do this, for example. > > > > I was surprised by Windriver's response. They said that although > > same-priority task mutual exclusion is implemented now, there was "no > > requirement" for it to operate this way, and it is "not guaranteed" to > > continue to do so in future versions. They said that semaphores were > > required for mutual exclusion even between same-priority tasks. > > > > I'm curious as to how many existing Vxworks customers are counting on > > mutual exclusion between same-priority tasks. Perhaps we should get > > Windriver to formerly add a requirement to provide this in future > > versions? > > > > Regards, > > -mike > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: Target Type for GCC Cross Compiler and Binutils: VxSim on Windows NT. Date: 2 May 2002 01:26:25 -0700 From: Giridhar.Nag@ubinetics.co.in (Y Giridhar Appaji Nag) Organization: http://groups.google.com/ Message-ID: <4aea5785.0205020026.6933e49c@posting.google.com> Hi, I wanted to build a cross compiler [from the GCC sources] for compiling my code on a Linux machine. This code has to be run on vxWorks. I managed to build binutils & GCC as cross compiler for the target PowerPC running vxWorks (powepc-wrs-vxworks) after some patching of the sources and searching on the net and the newsgroups. What I now want is to be able to do the same for the VxSim target. My VxSim runs on Windows NT. What would my target description be for building a compiler in this case?. I see that the executable format of the "VxSim binaries" is COFF (as reported by "file a.out"). I mean, in the configure phase of the build, what would the value of $target be in --target=$target? I am using binutils ver: 2.12 and gcc ver: 2.95.3 Are there any known problems with this setup for vxWorks? - -rgds, Giridhar PS: It would be nice if any responses to this are Cc:'ed to me since I do not have a very convenient access to the newsgroups. --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: Re: Target Type for GCC Cross Compiler and Binutils: VxSim on Windows NT. Date: Thu, 2 May 2002 10:14:45 +0100 From: "Martin Dowie" Message-ID: <3cd102d5$1@pull.gecm.com> References: <4aea5785.0205020026.6933e49c@posting.google.com> > I am using binutils ver: 2.12 and gcc ver: 2.95.3 Are there any > known problems with this setup for vxWorks? I dream of using something as 'new' as gcc 2.95.3 for VxWorks but we're still using Tornado 2.0.2 complete with 2.7.2!!!! :-( --------------------------- Newsgroups: comp.os.vxworks Subject: Re: convert vxworks_rom.hex Date: Thu, 2 May 2002 11:37:45 +0200 From: "Werner Schiendl" Message-ID: <3cd108dd$1@brateggebdc5.br-automation.co.at> References: <3CCDD5C0.54DEFA39@bellsouth.net> Hi, the source for _one_ function? Normally you are better of, writing it anew from scratch. The best thing you can get from the built stuff is assembly level code (afaik). For the future, I recommend source code control combined with proper backups. best regards Werner "balendran bala" wrote in message news:fdcfdc67.0205010410.5e72301f@posting.google.com... > Hi Joe, > > I lost the source code for a function. Since I have the > vxworks_rom.hex from last build (includes the function), so I am > trying to findout that I could do reverse-engineering and > recover the function. > > Thank you, > Bala > joe durusau wrote in message news:<3CCDD5C0.54DEFA39@bellsouth.net>... > > balendran bala wrote: > > > > > Hi guys, > > > > > > Is there anyway that I could convert vxworks_rom.hex file into a text file? > > > > > > Thank you, > > > bala > > > > If you look at it, it _is_ a text file. What do you want to do??? > > > > Speaking only for myself, > > > > Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: getting the base address of the PCI card Date: Thu, 2 May 2002 11:44:03 +0200 From: "Werner Schiendl" Message-ID: <3cd10a55$1@brateggebdc5.br-automation.co.at> References: Hi Anand, I assume you use some kind of pc486, pcPentium or similar BSP. Within this BSPs, you find a file called sysEl3c90xEnd.c This does exactly what you ask for for a network board. Study the file to find out how to read the base address and size of the address space as well as IRQ configuration from the PCI board. You can basically go with almost the same procedure, you just need to adapt Vendor and Device IDs and adjust the memory configuration to the proper BARs in the configuration. This is specific to your board, so you need to consult the hardware configuration here. pciConfigShow and pciHeaderShow also help with this. hth Werner "Anand" wrote in message news:f14c07cb.0205012219.284d4f84@posting.google.com... > HI All, > first, thanks to Bruce,Andray Kaganovsky and John for helping me > on Far Pointers. > Now, i have a PCI card which has some memory and registers which > i want to access and manipulate the contents of the registers > depending on the functionality i want to program. > Earlier , i was using an ISA card which had switches on it to > select the base address and depending on the base address i could get > the address of the registers and i could program the registers > according to the function i wanted. > but this PCI card does not have any switches, so that i can choose the > base address. > going through earlier articles in this group, i learned that we > can use functions like pciDeviceShow() and pciHeaderShow() , we can > have the details. but i am not sure whether the base address i get > from pciHeaderShow() can be used as the base address of the card or > not. I am new to vxworks, and i hope the question i am asking makes > some sense. > waiting for replies. > Thanks in advance. > Regards. > Anand. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: resourses for realtime Date: Thu, 2 May 2002 12:02:02 +0200 From: "Werner Schiendl" Message-ID: <3cd10e8d$1@brateggebdc5.br-automation.co.at> References: <57bd77b5.0204302022.28888e8f@posting.google.com> Hi Jim, you can download Tornado Prototyper for free from www.windriver.com. It comes with a simulator running on the development machine. For any 'real' work with VxWorks, you'll need a rather costy licence, however. regards Werner "jim canning" wrote in message news:57bd77b5.0204302022.28888e8f@posting.google.com... > HI ALL, > I'M NEWBIE TO REAL TIME OPERATING SYSTEMS AND I'M STILL A > STUDENT AND I WANT TO WORK ON REALTIME OPERATING SYSTEMS AND WANNA > KNOW ABOUT THAT STUFF > > CAN ANYBODY GIVE INFORMATION ABOUT REALTIME OS THAT CAN BE AVAILABLE > FOR FREE ON THE NET AND CAN BE DOWNLOADED FOR FREE SO THAT I CAN GET > SOME EXPERIENCE > > I REALLY APPRECIATE IT IF YOU CAN SUGGEST ME > > THANX > JIM --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu May 2 04:34:45 2002 From: "Sabarivel S" Date: Thu May 2 04:34:47 PDT 2002 Subject: Vxworks: ENOBUFS in setsockopt (MULTICAST_IF) option When does setsockopt call with MULTICAST_IF option will give ENOBUFS error message ? Your valuable input on this will be very helpful. thanks & regards sabari *************************************************************************** This message is proprietary to Future Software Limited (FSL) and is intended solely for the use of the individual to whom it is addressed. It may contain privileged or confidential information and should not be circulated or used for any purpose other than for what it is intended. If you have received this message in error, please notify the originator immediately. If you are not the intended recipient, you are notified that you are strictly prohibited from using, copying, altering, or disclosing the contents of this message. FSL accepts no responsibility for loss or damage arising from the use of the information transmitted by this email including damage from virus. *************************************************************************** From vxwexplo-errs@csg.lbl.gov Thu May 2 05:25:13 2002 From: "Ronny Oren" Date: Thu May 2 05:25:15 PDT 2002 Subject: RE: Vxworks: ENOBUFS in setsockopt (MULTICAST_IF) option This is a multi-part message in MIME format. ------=_NextPart_000_0006_01C1F1ED.8684FC50 Content-Type: text/plain; charset="windows-1255" Content-Transfer-Encoding: 7bit Hi, Try and find in the following document some ideas. Ronny ------=_NextPart_000_0006_01C1F1ED.8684FC50 Content-Type: text/plain; name="Debugging ENOBUFS.txt" Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename="Debugging ENOBUFS.txt" TITLE: Debugging ENOBUFS DESCRIPTION: How to debug the ENOBUFS Errors ? " sendto: S_errno_ENOBUFS " RESOLUTION: SECTION I : - - - - ENOBUFS occurs mostly because the network stack is running of mbufs: Here are some instances where it is more likely to appear: 1) From setsockopt() setsockopt needs a cluster of size 128 bytes to store the socket options and the ENOBUFS error may indicate that no 128-byte clusters exist. SOLUTION : increase the value of CL_SIZE_128 in ../target/h/netBufLib.h 2) From route_output() The ENOBUFS error may occur when there is a write operation to a routing socket and: a) a mbuf is not available OR b) where the data in the first mbuf is less then the minimum protocol header and it tries to check the second mbuf and try to arrange all the headers in the first mbuf with a call to m_pullup() and this one may need some additional mbufs which are not available. SOLUTION: increase the mbufs(in general) SEE SECTION II 3) From tcp_input() when tcp_template() fails because it can't create a skeletal tcp/ip header because mbufs are not available: SOLUTION : increase the mbufs(in general) SEE SECTION II 4) From tcp_output() when it can't allocate enough mbufs for IP and TCP headers SOLUTION : increase the mbufs(in general) SEE SECTION II 5) From tcp_usrreq() When it is unable to create tcp_template() because it is running out of mbufs which are supposed to be used for transmissions on a particular socket connection. At this stage it starts keepalive timer, create output sequence space and send initial segment on connection. SOLUTION : increase the mbufs(in general) SEE SECTION II 6) From sosend() When it can't find a cluster of hte required size to place the data that has to be sent. SOLUTION : increase the mbufs(in general) SEE SECTION II 7) From sosetopt() When SO_SNDBUF and SO_RCVBUF are full. SOLUTION: increase the size of the SO_SNDBUF /SO_RCVBUF in ../target/h/sys/socket.h 8) From in_pcballoc() When a pcbstructure cannot be malloc'ed because of no mbufs SOLUTION: increase the mbufs(in general) SEE SECTION II 9) From in_control() When a MALLOC fails because of no mbufs while the interface specific fields are filled in by proper SIOCSIF... Calls. SOLUTION : increase the mbufs(in general) SEE SECTION II 10) From rttequest() When there is no mbuf when a routing entry has to be created. SOLUTION : increase the mbufs(in general) SEE SECTION II 11) From raw_attach() When there are no mbufs while trying to allocate a control block and some buffer space for the socket. SOLUTION : increase the mbufs(in general) SEE SECTION II 12) From add_lgrp() When it runs out of mbufs while trying to add an IP to a multicast group. SOLUTION : increase the mbufs(in general) SEE SECTION II 13) From ip_output() When it tries to copy data from the segment and makes a header and try to link in the form of mbuf chain and runs out of mbufs. SOLUTION: ipstatShow() whill increment the drops++. increase the mbufs(in general) SEE SECTION II 14) From ip_setmoptions() a)When a setsockopt is issued by an user with multicast options And there is no multicast option buffer attached to the pcb then an attempt will be made to allocate one. It may fail because of lack of mbufs: b) If it can't allocate a new multicast record when a new Address comes in because of no mbufs. SOLUTION : increase the mbufs(in general) SEE SECTION II 15) From ipOutput() /* in case of END drivers only */ a) If muxAddressForm cannot allocate link level header because of no mbufs we get this error: SOLUTION : increase the mbufs(in general) SEE SECTION II b) If the Interface Queue is Full. SOLUTION: SEE SECTION III. 16) From ether_ouput() a)When there is no space in the first mbufs an attempt is made to fetch another mbuf to place the header. Then it could runout of mbufs: SOLUTION : increase the mbufs(in general) SEE SECTION II b) If the Interface Queue is Full SOLUTION: SEE SECTION III. SECTION II: - - - - - Most ENOBUFF errors can be resolved by increasing the number of mbufs available. However, in some cases the error may be caused by one of the following: a) in the network stack it self OR b) the amount of memory the target has OR c) The CPU speed itself. d) Design & Bugs from the regular BSD stuff itself, It could be a limitation of the RFC itself. In these cases, the error may not be corrected simply adding mbufs, as described in the section above. First, to debug possible problems caused by by insufficient mbufs, use the command netstackSysPoolShow(). >From target do -> netstackSysPoolShow() type number --------- ------ FREE : 1095 DATA : 0 HEADER : 0 SOCKET : 4 PCB : 4 RTABLE : 12 HTABLE : 0 ATABLE : 0 SONAME : 0 ZOMBIE : 0 SOOPTS : 0 FTABLE : 0 RIGHTS : 0 IFADDR : 4 CONTROL : 0 OOBDATA : 0 IPMOPTS : 0 IPMADDR : 1 IFMADDR : 0 MRTABLE : 0 TOTAL : 1120 number of mbufs: 1120 number of times failed to find space: 0 number of times waited for space: 0 number of times drained protocols for space: 0 __________________ CLUSTER POOL TABLE _________________________________________________________________________= ______ size clusters free usage -------------------------------------------------------------------------= ------ 64 160 153 8 128 160 10 23 256 160 4 12 512 80 76 16 -------------------------------------------------------------------------= ------ value =3D 80 =3D 0x50 =3D 'P' ( The Lower values of "free" shows that it is running out of NUM_SYS_128 & NUM_SYS_256 ) and then increase the required values of Example: #define FACTOR 4 #define NUM_SYS_64 FACTOR*40 /* 64 byte system clusters */ #define NUM_SYS_128 FACTOR*40 /* 128 byte system clusters */ #define NUM_SYS_256 FACTOR*40 /* 256 byte system clusters */ #define NUM_SYS_512 FACTOR*20 /* 512 byte system clusters */ AND do similarly with mbufShow ->mbufShow() and increase the required values of #NUM_NET_MBLKS FAC*400 #define NUM_64 FAC*100 /* no. 64 byte clusters */ #define NUM_128 FAC*100 /* no. 128 byte clusters */ #define NUM_256 FAC*40 /* no. 256 byte clusters */ #define NUM_512 FAC*40 /* no. 512 byte clusters */ #define NUM_1024 FAC*25 /* no. 1024 byte clusters */ #define NUM_2048 FAC*25 /* no. 2048 byte clusters */ in the ../target/h/netBufLib.h then bulid vxWorks image and boot the target to take the changes into account. SECTION III: - - - - - - A) For 7) of SECTION I increase the values of SO_SNDBUF/SO_RCVBUF either in the ../target/h/sys/socket.h file or use setsockopt carefully in your application.More explaination on setsockopt() can be found in the vxWorks Reference Manual.. B) Running out of mbufs at the interface Q level. You can see this when you do an ifShow() and it will show a lot of output errors ( oerrors++). So try increasing the maxlen value of the if_snd Q to the appropriate value. By Default it is 50. Try with 100 or so. #include "vxWorks.h" #include "ifLib.h" #include "etherLib.h" /* Run this script each time you boot the target or make it a part of the application */ void ifq_maxLen(char *name, int len) { struct ifnet *ifp; ifp =3D ifunit(name); if (ifp =3D=3D NULL) printf("error:in if unit\n"); ifp->if_snd.ifq_maxlen =3Dlen; printf("set ifq maxlen %d -> %d\n", ifp->if_snd.ifq_maxlen, len); } NOTE : This Summary lists some possible reasons for the ENOBUFS error. It is not meant to be all inclusive. Nor will the resolutions correct all instances of ENOBUF errors. If you require additional information about the routines listed in SECTION I, please refer to TCP/IP Illustrated Volume 2 by Richard Stevens, As these are routines for which the customer will probably not possess. ------=_NextPart_000_0006_01C1F1ED.8684FC50-- From vxwexplo-errs@csg.lbl.gov Thu May 2 08:43:35 2002 From: "VxWorks" Date: Thu May 2 08:43:38 PDT 2002 Subject: Timer Driver Hi I am writing a timer driver for vxWorks. I am having some doubts in calculating sysClkTicks. How is this sysClkTicks calculated. What all factors has to be taken into account for creating this sysClkTicks. Thanking you in advance Regards Anamika From vxwexplo-errs@csg.lbl.gov Thu May 2 08:43:36 2002 From: "VxWorks" Date: Thu May 2 08:43:41 PDT 2002 Subject: Timer Driver Hi I am writing a timer driver for vxWorks. I am having some doubts in calculating sysClkTicks. How is this sysClkTicks calculated. What all factors has to be taken into account for creating this sysClkTicks. Thanking you in advance Regards Anamika From vxwexplo-errs@csg.lbl.gov Thu May 2 13:02:18 2002 From: Shiv Date: Thu May 2 13:02:20 PDT 2002 Subject: Problems with MotFec on the ppc855T. For anyone who has the time to read all this and a relative amount of expertise with the FEC ! I'm a newbie to FEC drivers (and the 8xx in general) and I'm trying to get my motfec up and running on my target board , loosely based (meaning ->not as populated) on the MdpPro8xxT BSP ...the transceiver I'm using is the CTX1170 canary module that connects to an MII interface on both boards, I wired that identically on my target as was done on the Eval Board. To let you know where I currently am with the FEC driver : (I'm downloading the vxworks image via visionprobe and using visionclick to debug it, once again , my target is the 855T , the BSP-template is the MdpPro8xxT and I've configured both boards using the register file download facility of visionclick , not worrying about boot-romming yet) * The Canary on noth boards start up in the isolation mode (On the module , this is configurable by a switch and its purpose is to restrict ethernet activity unless software configuration is performed) This "isolation mode" seems evident from the way the canary module behaves , with the 100MBps and link lights blinking in a wait-state. * The MdpBasicPro8xxT BSP eventually gets the link lights to stabilize in both cases, (at this stage I can see ethernet activity via the Rx Light) though this happens very fast on the Eval board (where I can consequently ping it from my host, observing Tx along with the Rx). This "waking-up" behavior takes a lot longer on the target board. I suspect a few timeouts elapse to account for this delay ? * Now , MOtFecEndLoad() , "THE" critical function on the BSP that loads the driver DOES NOT return an error or null value at any stage during its run. This I confirmed via breakpoints. * What DOES throw a spanner in the works is the MotFecMIIDiag() Diagnostics that follows up on the BSP , This asserts an error after the following code in the test-loop: if (motFecMiiRead (pDrvCtrl, phyAddr, regAddr, &data) != OK) return (ERROR); This is where the motfec fails on my target..as far as I can tell. WRS-Support seems puzzled ! * I also noted that for 860/855T processors , you have to take note of a D4 revision change. I believe the BSP takes into account this ...I'm using a D4 chip, anyhow. Could any one help me isolate the problem further ? Would really appreciate it ! Thanks ! Shiv From vxwexplo-errs@csg.lbl.gov Fri May 3 04:03:18 2002 From: Vxworks Exploder Date: Fri May 3 04:03:20 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri May 3 04:03:14 PDT 2002 Subject: Cancelling Write/Read operation to SCSI drive Subject: Re: Difference between RTOS and normal OS Subject: Re: netBufLib questions Subject: Re: PPP header compression Subject: Re: Does windml 2.0 font library support japanese and chinese fonts Subject: Re: Mutual exclusion between same-priority tasks Subject: Re: Mutual exclusion between same-priority tasks Subject: Interrupt Latency Subject: openssl for VxWorks Subject: Re: Target Type for GCC Cross Compiler and Binutils: VxSim on Windows Subject: Re: Interrupt Latency Subject: Re: openssl for VxWorks Subject: Re: Interrupt Latency Subject: Re: No interface UP message(RTM_IFINFO) in routing socket for nt0 Subject: Re: Target Type for GCC Cross Compiler and Binutils: VxSim on Windows NT. Subject: Incorrectly reported floating point exceptions Subject: VxWorks 5.3 Manuals Subject: Re: TCP Stream Socket send buffer in VxWorks v5.4 Subject: about via rhine II network adapter's drivers. Subject: Re: Cancelling Write/Read operation to SCSI drive Subject: diab asm syntax questions... Subject: Re: Booting multiple interfaces Subject: Re: #ifdef question Subject: Re: Difference between RTOS and normal OS Subject: Re: Difference between RTOS and normal OS Subject: m8260FccEnd driver transmit errors Subject: Network Interface Extensions in VxWorks (based on RFC2233) Subject: Default route outside subnet ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Cancelling Write/Read operation to SCSI drive Date: 2 May 2002 06:31:13 -0700 From: Jeffrey.Steinsnyder@L-3Com.com (Jeff Steinsnyder) Organization: http://groups.google.com/ Message-ID: Hello Everyone, We are using VxWorks v5.3.1, SCSI hard drive on MVME2604 Single board computer. After Application performs the System Reset sequence and comes back into the operation it looks like our disk file system (dosFs) gets corrupted. We suspect that the SCSI drive does its write operations at the time the Reset sequence has been initiated. Does any one knows if there is a way to cancel all SCSI drive read(s)/write(s), using the VxWorks API? I'm not sure whether I can use ioctl() with FIOCANCEL function in this situation. Please let me know. Thank you much, Jeff Steinsnyder. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Thu, 2 May 2002 09:13:28 -0400 From: "Yin, Yong" Organization: Nortel Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> priority-based preemptive schedule Vivek wrote in message news:507efed5.0205012037.3e11728f@posting.google.com... > Hi, > > I am looking for some concrete differences between RTOS and normal OS. > Could any body point me to the right direction for this. > > Thanks and in advance > > Vivek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: netBufLib questions Date: Thu, 02 May 2002 14:41:21 +0100 From: Paul Mannion Organization: Silicon and Software Systems Message-ID: <3CD14201.F4C05E0E@s3group.com> References: <2b0e7fc437bdc7ad995977ba48739431.76014@mygate.mailgate.org> This is a multi-part message in MIME format. - --------------BEC970B73FF0C0143818AC5D Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Nix, anytime yo have to modify the netpool lists briefly disabling interrupts will prevent task switches. I woudln't recommend using semaphores unless you know the system well as you may be allocating buffers from within ISR context and that blocking could be disastrous. Paul Dexter Nixon wrote: > Hi, > > I am reading the Network Tool kit guide of VxWorks. Here the pool of > clusters we are allocating and managing it by means of mBlk and Clblk > structures right. But it is given that access to these structures are > not semaphore protected. THen how does this will work. Because if am > implementing more than one network protocol using MUX and am > initialising a memory pool, and if both protocols access the same mem > pool, how can we access mutual exclusion is there when the two protocols > acccess the same pool with both are running as different threads. > Without semaphore protection internally within the structures how this > is possible. anybody kindly explain me. > > Thanks, > Nix > > -- > Posted via Mailgate.ORG Server - http://www.Mailgate.ORG - --------------BEC970B73FF0C0143818AC5D Content-Type: text/x-vcard; charset=us-ascii; name="paulm.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Paul Mannion Content-Disposition: attachment; filename="paulm.vcf" begin:vcard n:Mannion;Paul tel;fax:-353-1-2911001 tel;work:-353-1-2911360 x-mozilla-html:FALSE org:Silicon and Software Systems adr:;;;;;; version:2.1 email;internet:paulm@s3group.com title:Software Engineer fn:Paul Mannion end:vcard - --------------BEC970B73FF0C0143818AC5D-- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PPP header compression Date: Thu, 02 May 2002 09:59:25 -0400 From: david lindauer Message-ID: <3CD1463D.86F33A3A@notifier-is.net> References: the PPP options are in configall.h. Looks like you want the options: PPP_OPT_NO_ACC PPP_OPT_NO_PC PPP_OPT_NO_VJ (if you don't want VJ compression, technically a TCP/IP header issue) Jason wrote: > Can some one tell me how to change the default PPP header compression under > VxWorks? > > Thanks, > Jason --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does windml 2.0 font library support japanese and chinese fonts Date: 02 May 2002 11:08:13 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <32098fec.0204300619.bb7490e@posting.google.com> Sender: bpringlemeir@DeadDuck Avik> Dear all, I need to know whethet windml 2.0 have existing font Avik> libray for japanese and chinese fonts.If japanese and chinese Avik> font libary is there , then how I can get those font Avik> libraries.you can send your reply to shibasis@rebaca.com. Hope Avik> to get reply soon. >> Kanji fonts (etc) do not come with WindML 2.0. There are >> utilities to convert BDF and Windows fonts to the UGL (WindML) >> format. Font licensing would be up to you. ChanSeong> Hi~ I once tried using perl source to convert from BDF to ChanSeong> UGL format. I think you should know some source or ChanSeong> utility support only one byte code. ChanSeong> At that time I converted some bitmap source to UGL format ChanSeong> our own developed convert source. ChanSeong> So, I'd like to know good convert program. There are programs on WindSurf that will allow you to do this. They are ufe.zip (Uterine fibroid embolization), a program for converting windows fonts, and bdf2bmf.zip, a program for converting Unix BDF fonts. "http://web2.windriver.com/cgi-bin/windsurf/downloads/fetch_download.cgi?downloadid=1117&binaryid=253" "http://web2.windriver.com/cgi-bin/windsurf/downloads/fetch_download.cgi?downloadid=1117&binaryid=254" Regards, Bill Pringlemeir. - -- Allowedness is incomprehensible. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mutual exclusion between same-priority tasks Date: Thu, 2 May 2002 09:34:42 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: <3CCFFB03.BA438CA4@attbi.com> <488e459a.0205011316.5e33fb21@posting.google.com> I would recommend the use of mutex semaphores. Other schemes (taskLock and intLock) are not really meant to be used for mutex functions. Using taskLock and intLock for mutex can cause unpredictable system behavior in large systems. Use of taskLock can increase preemptive latency especially when used extensively. Use of intLock can increase interrupt latency. I have seen many large embedded systems built with VxWorks which assumes that taskLock and intLock can be used as convenient replacement for mutex semaphores. There are a lot of problems stemming from that assumption. If you need mutex, use mutex semaphores. If you want to disable task switching, use taskLock. If you want to disable interrupts, use intLock. But please do not mix the originally intended purpose. (Read the manual for taskLock, which makes clear statements with respect to what is preferrable for mutex.) - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mutual exclusion between same-priority tasks Date: Thu, 02 May 2002 16:58:00 GMT From: "doug dotson" Organization: Verio Message-ID: References: <3CCFFB03.BA438CA4@attbi.com> Relying on a side-effect of the scheduler is pretty risky. Probably best to use the mutexes. doug "Michael Patrick" wrote in message news:3CCFFB03.BA438CA4@attbi.com... > How many of you are avoiding the use of semaphores for protecting global > variables because you think vxworks will schedule (with timeslicing > disabled) same-priority tasks to always run to completion until they > voluntarily give up context? > > I came across this design technique when reviewing a project in my > company, and submitted a TSR to ask Windriver about it. I was worried, > you see, that an interrupt to one task would return through the > scheduler to a *different* same-priority task, so that same-priority > tasks would NOT be mutually exclusive. PSOS can do this, for example. > > I was surprised by Windriver's response. They said that although > same-priority task mutual exclusion is implemented now, there was "no > requirement" for it to operate this way, and it is "not guaranteed" to > continue to do so in future versions. They said that semaphores were > required for mutual exclusion even between same-priority tasks. > > I'm curious as to how many existing Vxworks customers are counting on > mutual exclusion between same-priority tasks. Perhaps we should get > Windriver to formerly add a requirement to provide this in future > versions? > > Regards, > -mike > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Interrupt Latency Date: Thu, 02 May 2002 18:57:49 +0200 From: Peter Kannegiesser Organization: SEK Message-ID: Reply-To: peter.kannegiesser@gmx.de Hi all! We are using Tornado 2.02 and the related tools and have written a driver for an D/A, A/D converter PCI card. The driver is implemented using ioctl() calls. The PCI card generates interrupts at the end of conversion. The interrupt handler is connected using pciConnectInt and is definitly the only one for that interrupt line. Now we experience very looong delays from the interrupt signal on the PCI bus until the handler gets called. According to the logic analyzer, it takes between 5 and 15 milliseconds (with a strong tendency to 15ms)-- and this supposed to be an OS for fast hard realtime... We are using the standard Pentium BSP on various PC's, from 166MHz Pentium I to 2GHz Pentium 4. There seems to be no change that depends on the processor speed. Has anybody seen similar problems or could offer at least an explanation? TIA, Peter - -- keine sig --------------------------- Newsgroups: comp.os.vxworks Subject: openssl for VxWorks Date: 2 May 2002 10:38:56 -0700 From: nhoda@agere.com (Nadeem HOda) Organization: http://groups.google.com/ Message-ID: Has anybody successfully ported openssl to VxWorks? Nadeem Hoda --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: Re: Target Type for GCC Cross Compiler and Binutils: VxSim on Windows NT. Date: Thu, 02 May 2002 12:59:00 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3CD17E64.4E69E0DF@aps.anl.gov> References: <4aea5785.0205020026.6933e49c@posting.google.com> <3cd102d5$1@pull.gecm.com> Martin Dowie wrote: > > I dream of using something as 'new' as gcc 2.95.3 for VxWorks but > we're still using Tornado 2.0.2 complete with 2.7.2!!!! :-( The C++ compiler that comes with Tornado 2.0.2 isn't *quite* that old, it's actually based on EGCS 1.0.2 (see the Tornado 2.0 Release notes, section 7.6 for details). The C compiler is still derived from GCC 2.7.2 though. - - Andrew - -- Perfection is reached, not when there is no longer anything to add, but when there is no longer anything to take away. - - Antoine de Saint-Exupery --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Date: Thu, 2 May 2002 19:59:09 +0200 From: "Werner Schiendl" Message-ID: <3cd17e5f$1@brateggebdc5.br-automation.co.at> References: Hi, how do you meassure that delay? On the box you specified, interrupt latency should be in the range 20 us (_not_ ms) or shorter. There is something wrong, either with your hardware or your test. If you set hardware IOs to do the meassuring, be sure to not be fooled by cached values. (see cacheLib and verify you properly mapped the IO range of the meassure and IO board). hth Werner "Peter Kannegiesser" wrote in message news:aer2dug9ot75urttc2dabeg507f5v44p7t@4ax.com... > Hi all! > > We are using Tornado 2.02 and the related tools and have written a > driver for an D/A, A/D converter PCI card. The driver is implemented > using ioctl() calls. The PCI card generates interrupts at the end of > conversion. > The interrupt handler is connected using pciConnectInt and is > definitly the only one for that interrupt line. > Now we experience very looong delays from the interrupt signal on > the PCI bus until the handler gets called. According to the logic > analyzer, it takes between 5 and 15 milliseconds (with a strong > tendency to 15ms)-- and this supposed to be an OS for fast hard > realtime... > We are using the standard Pentium BSP on various PC's, from 166MHz > Pentium I to 2GHz Pentium 4. There seems to be no change that > depends on the processor speed. > > Has anybody seen similar problems or could offer at least an > explanation? > > TIA, > Peter > -- > keine sig --------------------------- Newsgroups: comp.os.vxworks Subject: Re: openssl for VxWorks Date: 02 May 2002 14:39:25 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: Sender: bpringlemeir@DeadDuck Nadeem> Has anybody successfully ported openssl to VxWorks? "http://groups.google.com/groups?as_q=openssl&as_ugroup=comp.os.vxworks" "http://www.xs4all.nl/~borkhuis/vxworks/vxw_pt6.html#6.9" The network speaks before you are spoken to. hth, Bill Pringlemeir. - -- LLNL Bosnia Ron Brown sniper ASIO Forte Mantis Geraldton Vickie Weaver SCUD missile eternity server ANDVT military mania beanpole vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Date: Thu, 02 May 2002 21:50:14 +0200 From: Peter Kannegiesser Organization: SEK Message-ID: References: <3cd17e5f$1@brateggebdc5.br-automation.co.at> Reply-To: peter.kannegiesser@gmx.de On Thu, 2 May 2002 19:59:09 +0200, "Werner Schiendl" wrote: >how do you meassure that delay? The interrupt signal is measured at its creation point, that maens on the A/D module itself. The A/D module also has digital outputs, and one of them is set when the interrupt handler is started, reset after the handler. > >On the box you specified, interrupt latency should be in the range 20 us >(_not_ ms) or shorter. ...that's what I've expected... >There is something wrong, either with your hardware or your test. Right. The test seems not to be wrong: even without measuring hard- and software behaviour of the system is such that we had to suspect large interrupt latencies. The test only confirmed this. Question: Do I have to handle some piece of PC hardware in the interrupt handler? I can't imagine this, but want to be sure. > >If you set hardware IOs to do the meassuring, be sure to not be fooled by >cached values. >(see cacheLib and verify you properly mapped the IO range of the meassure >and IO board). I guess that doesn't apply in my case :-( Best Regards, Peter - -- keine sig --------------------------- Newsgroups: comp.os.vxworks Subject: Re: No interface UP message(RTM_IFINFO) in routing socket for nt0 Date: 2 May 2002 13:12:52 -0700 From: dave.moore@conexant.com (Dave Moore) Organization: http://groups.google.com/ Message-ID: References: <8dff11cd.0204252137.74e1a733@posting.google.com> sai_dattathrani@infy.com (sai) wrote in message news:<8dff11cd.0204252137.74e1a733@posting.google.com>... > Hi, > > I have VxSim that is shipped along with Tornado. I have created > routing socket to receive interface up messages(RTM_IFINFO). I call > ifFlagChange() function to bring the interface UP or DOWN. > > When the loopback interface(lo0) was brought up & down, the message > RTM_IFINFO was received in the routing socket. But when the interface > nt0 is brought up or down, no message is received on the routing > socket. > > Similarly when secondary ip address is added using ifAddrAdd() > function in lo0, RTM_NEWADDR message is received in the routing > socket, but when a new ip address is added for nt0, no message is > received. > > Can anyone help me solve this problem? > > rgds > sai From the VxWorks Documentation: WARNING: If the driver changes the state of the flags element (for instance, marking itself as "down" by setting the IFF_DOWN flag), the driver should export this change by calling muxError( ) with the errCode member of the END_ERR structure set to END_ERR_FLAGS. --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: Re: Target Type for GCC Cross Compiler and Binutils: VxSim on Windows NT. Date: 2 May 2002 13:35:34 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205021235.3e26d84b@posting.google.com> References: <4aea5785.0205020026.6933e49c@posting.google.com> <3cd102d5$1@pull.gecm.com> Hello, You could upgrade to VxWorks AE, complete with GCC 2.96. You might also want to check with your local FAE to see if there is a 2.1 release for your CPU that includes a newer compiler. HTH, John... "Martin Dowie" wrote in message news:<3cd102d5$1@pull.gecm.com>... > > I am using binutils ver: 2.12 and gcc ver: 2.95.3 Are there any > > known problems with this setup for vxWorks? > > I dream of using something as 'new' as gcc 2.95.3 for VxWorks but > we're still using Tornado 2.0.2 complete with 2.7.2!!!! :-( --------------------------- Newsgroups: comp.os.vxworks Subject: Incorrectly reported floating point exceptions Date: Thu, 02 May 2002 17:17:17 -0400 From: Stephen Roderick Organization: Space Systems Lab Message-ID: <3CD1ACDC.5512D591@ssl.umd.edu> We've found that VxWorks is reporting the incorrect exception, whether or not an application-level handler is dealing with floating point exceptions. Specifically: Raising SIGFPE and dividing by zero both produce SIGFPE. Overflows and underflows to integers, or to single from double, all produce SIGILL (coprocessor unuseable), should be SIGFPE. Inavlid op (generated by giving sin/cos something really large) causes an address error (!?), should be SIGFPE. Still trying to figure out what's going on here ... Has anyone seen this before, or have any idea what the problem is? We're running Vx5.2 on a Sanders Star II board with a MIPS R4k. THX! --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks 5.3 Manuals Date: 2 May 2002 16:53:43 -0700 From: go_st8@hotmail.com (Chris) Organization: http://groups.google.com/ Message-ID: <2bb221f7.0205021553.4f7deee4@posting.google.com> Hello, I am looking for a hard copy of a VxWorks 5.3 Reference Manual. Does anyone have one they would be interested in parting with? Thanks, Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TCP Stream Socket send buffer in VxWorks v5.4 Date: Thu, 2 May 2002 17:20:21 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: It can be done by: 1. use iosFdValue() on socket descriptor to get a pointer of type struct socket. 2. Read socketvar.h for definition of struct socket. 3. so_rcv.sb_cc contains actual bytes consumed per queue at socket level. - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: about via rhine II network adapter's drivers. Date: 2 May 2002 17:26:16 -0700 From: zhoushan@371.net (zhoushan) Organization: http://groups.google.com/ Message-ID: hello. does anyone have experince working on via vt6102 chip? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Cancelling Write/Read operation to SCSI drive Date: Fri, 03 May 2002 10:13:07 +0930 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3CD1DD1B.EB11ECC@dsto.defence.gov.au> References: Jeff, When the SCSI device is initialised with ncr810CtrlCreate, the scsi control structure (SCSI_CTRL *) is returned. One of the elements of this struct is the function pointer pScsiCtrl->scsiThreadAbort, so you could try running that. I'm not sure, but it may be the same function that is run when FIOCANCEL is called from ioctl. The arguments to the this function are a pointer to the SIOP (same as to the controller) and to the thread to abort. Both are in the scsi control struct, from memory it would be something like: (*pSysScsiCtrl->scsiThreadAbort)(pSysScsiCtrl, pSysScsiCtrl->pThread) You may also want to check the scsi control struct to see that a thread is in fact active (pSysScsiCtrl->active == TRUE) But is there a reason that you can't wait until the write has finished? One other thing with this BSP and SCSI is that the SCSI bus is not reset properly after a warm reset or a reboot. You need to perform a scsiBusReset(pSysScsiCtrl) before you do any scsi config, somewhere at the top of sysScsiConfig( ) seems to work okay. HTH Tim Jeff Steinsnyder wrote: > Hello Everyone, > We are using VxWorks v5.3.1, SCSI hard drive on MVME2604 Single board > computer. After Application performs the System Reset sequence and > comes back into the operation it looks like our disk file system > (dosFs) gets corrupted. We suspect that the SCSI drive does its write > operations at the time the Reset sequence has been initiated. Does any > one knows if there is a way to cancel all SCSI drive read(s)/write(s), > using the VxWorks API? I'm not sure whether I can use ioctl() with > FIOCANCEL function in this situation. Please let me know. > > Thank you much, > Jeff Steinsnyder. --------------------------- Newsgroups: comp.os.vxworks Subject: diab asm syntax questions... Date: 2 May 2002 18:15:42 -0700 From: jchen@tollbridgetech.com (john c) Organization: http://groups.google.com/ Message-ID: <1f1b48b5.0205021715.7f45e194@posting.google.com> Hi all... I am trying to use DIAB 4.4b compiler on PowerPC. Found out that it is VASTLY different from writing embedded assembly in GNU. Anyone have any experience in writing embedded assembly for DIAB? How would you translate this line from the GNU? In GNU, I write my function like so: unsigned short gt_ld_le16(const volatile unsigned short *addr) { unsigned short val; __asm__ __volatile__ ("lhbrx %0,0,%1" : "=r" (val) : "r" (addr), "m" (*addr)); return val; } I would write it in DIAB like this: asm unsigned short gt_ld_le16(volatile unsigned short *addr) { %reg addr lhbrx r3,0,addr blr } is that correct? Can I insert a '0' like in the example? how are storage mode 'con' used? Windriver's DIAB C/C++ Compiler User Manual gives a really crappy example (an example with little features) on how to write this code.... Just curious. I didn't realize that porting from GNU to DIAB compiler is such a big effort.... yikes. john --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Booting multiple interfaces Date: 2 May 2002 22:07:20 -0700 From: alimohd@logiceastern.com (Ali) Organization: http://groups.google.com/ Message-ID: References: "Dexter Nixon" wrote in message news:... > Hi, > > This was the typical booting example given in the manual. Here we are > configuring a Lance Ethernet card while booting itself with this > bootline > > ln(0, 0) bear:/usr/wpwr/target/config/mz7122/vxworks e=90.0.0.2 > b=91.0.0.2 h=100.0.0.4 g=90.0.0.3 u=papa pw=biggrump f=0x80 tn=goldilox > s=bear:/usr/papa/startup o= > > Suppose, i have multiple ethernet cards like this means, how can i > configure those interfaces. What is the way to do it. > > THanks, > Nix hi Nix look into your configNet.h file there you will find a table endDevTblEntry[] make your entry for multiple interfaces with different unit numbers and load strings.. To bring up other interfaces other than one you booted from look into the usrConfig.c where the line /*Boot from Ethernet is given*/ copy that code in some another function and call it from anywhere in ur bsp this will bring up your other interfaces. There is also a code snippet given in vxworks FAQs ehich does all this you can look into that also. regrds ali --------------------------- Newsgroups: comp.os.vxworks Subject: Re: #ifdef question Date: Fri, 03 May 2002 08:20:09 +0200 From: Leon Gubbels Organization: =?iso-8859-1?Q?Oc=E9?= Technologies B.V. Message-ID: <3CD22C19.B17AD373@oce.nl> References: <3CCFB06C.196EF97B@oce.nl> Leon Gubbels wrote: > > Hi, for compilation for target (VxWorks) we now use #ifndef _WIN32 with > the GCC compiler. But to be more specific this should be something like > #ifdef VXWORKS ... > I cannot find this statement in the manuals. Also the compiler > (cygnus-2.7.2-960126) doesn't recognize some variants I tried: > #ifdef _VXWORKS > #ifdef VXWORKS > #ifdef _vxworks > #ifdef vxworks > #ifdef _VxWorks > #ifdef VxWorks > > does anyone know what it should be? > > Thanx in advance. > > Leon (if you reply to my mail address, remove NOSPAM). Well, I had a lot of reactions and solutions. Thanks you all. At this time there is no #define for VxWorks in vxworks.h, but from what I understood, there probably will be one in the future. For now, I'll have to make my own. Leon (if you reply to my mail address, remove NOSPAM). --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: 2 May 2002 23:54:49 -0700 From: vivekm@dd.nec.com.au (Vivek) Organization: http://groups.google.com/ Message-ID: <507efed5.0205022254.3db92d02@posting.google.com> References: <507efed5.0205012037.3e11728f@posting.google.com> "Yin, Yong" wrote in message news:... > priority-based preemptive schedule but in unix/linux also we can modify the process priority etc, then why arn't these are RTOS. ? Sorry, just a confusion . > > Vivek wrote in message > news:507efed5.0205012037.3e11728f@posting.google.com... > > Hi, > > > > I am looking for some concrete differences between RTOS and normal OS. > > Could any body point me to the right direction for this. > > > > Thanks and in advance > > > > Vivek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: 03 May 2002 00:26:25 -0700 From: Joe Chung Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> You can find this info in lots of places: 1. The comp realtime FAQ http://www.faqs.org/faqs/realtime-computing/faq/ 2. Search for the term "rtos difference" in groups.google.com. - -jc - -- (apply 'concat (reverse (list "com" (char-to-string 46) "yahoo" (char-to-string 64) "joechung"))) --------------------------- Newsgroups: comp.os.vxworks Subject: m8260FccEnd driver transmit errors Date: Fri, 3 May 2002 09:34:04 +0200 From: "Achim Zimmer" Message-ID: I try to use the m6260FccEnd driver, supplied by WRS eith ADS8260 BSP to use with our own board. The connection is equal, but we use a 971 Phy chip instead of 970. The problem we are running in, is: The packets on the LAN the board transmits have in 80 % errors, means, there are 8 bytes repeated in the frame, and theoriginal 8 bytes are lost. Has anyone this problem too and there are any solutions ? Thanks in advance. Any hint is welcome. --------------------------- Newsgroups: comp.os.vxworks Subject: Network Interface Extensions in VxWorks (based on RFC2233) Date: 3 May 2002 01:41:03 -0700 From: ilyasg@yahoo.com (Ilyas G) Organization: http://groups.google.com/ Message-ID: Hi vxworkers, I am trying to add the rfc2233 into the SNMP agent, does anyone experienced this ? --------------------------- Newsgroups: comp.os.vxworks Subject: Default route outside subnet Date: 3 May 2002 09:21:19 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: I have the following problem. I have a device that gets an IP address and subnetmask, but also a gateway. These are fixed values. This gateway address however is outside of the subnet. So for example: IP address = 202.204.206.208:ffffff00 Gateway = 175.185.195.205 How can I configure the routing to do this? Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri May 3 07:48:16 2002 From: "Sabarivel S" Date: Fri May 3 07:48:18 PDT 2002 Subject: Vxworks command for stack dump Could any of you kindly indicate way in which we can get the following inforamations using VxWorks commands. 1) Is there a way to dump the stack of any particular task after stopping it's execution(knowing what the stack size is). 2) Could we view register contents. 3) Is there a way to view contents of "n" bytes of memory, from a particular starting address. thanks Sabari *************************************************************************** This message is proprietary to Future Software Limited (FSL) and is intended solely for the use of the individual to whom it is addressed. It may contain privileged or confidential information and should not be circulated or used for any purpose other than for what it is intended. If you have received this message in error, please notify the originator immediately. If you are not the intended recipient, you are notified that you are strictly prohibited from using, copying, altering, or disclosing the contents of this message. FSL accepts no responsibility for loss or damage arising from the use of the information transmitted by this email including damage from virus. *************************************************************************** From vxwexplo-errs@csg.lbl.gov Fri May 3 09:33:48 2002 From: Joel Katz Date: Fri May 3 09:33:50 PDT 2002 Subject: Re: Vxworks command for stack dump You don't specify whether you are looking for this information from a shell or from a program. I have responded for the case of a shell, which is the most common way to access the information you are looking for. If you type help from a vxWorks shell, you will see all of these items listed plus more. >Could any of you kindly indicate way in which we can get the following >inforamations using VxWorks commands. > >1) Is there a way to dump the stack of any particular task after stopping >it's execution(knowing what the stack size is). tt (taskID) dumps the stack for a particular task. You can look at the general state of all tasks using the i command and get additional info using the ti command. >2) Could we view register contents. ti shows the register contents for a task. mRegs lets you interactively query the contents of the registers from a shell. >3) Is there a way to view contents of "n" bytes of memory, from a particular >starting address. > d startAddress, number of bytes will display the contents of a block of memory -- Joel Katz, Staff Engineer --- (781)-280-1753 Affymetrix, 4G Crosby Drive, Bedford, MA 01730 From vxwexplo-errs@csg.lbl.gov Fri May 3 13:49:35 2002 From: Dave Olafson Date: Fri May 3 13:49:38 PDT 2002 Subject: m2Lib howdy we're developing a snmp agent (w/ enterprise specific mibs) for a vxworks platform, and we would like to support mib2 objects ... so we are using the vxworks-supplied m2Lib methods ... these methods work very well for most of the mib2 objects -- sysName, sysDescr, ip / tcp / udp tables, etc ... however, we get some odd behavior from the ifTable if we write the code using the "clean" approach -- with a call to m2*Delete() for every m2*Init() -- we get ifNumber = 2 (one ethernet & one loopback ... but all calls to m2IfTblEntryGet() return an error if we use a "less clean" approach -- removing some/all of the m2IfDelete() calls -- then we get reasonable responses from m2IfTblEntryGet() ... unfortunately, as soon as we reach the end of our "GetIfTableValue()" method (which calls m2IfTblEntryGet()), our task dies ... in the "clean" approach, the task does not die here's a subset of GetIfTableValue() { M2_INTERFACETBL ifTableEntryVar; if (m2IfInit(NULL, NULL) == OK) { ifTableEntryVar.ifIndex = 0; while (1) { if (m2IfTblEntryGet (M2_NEXT_VALUE, &ifTableEntryVar) == ERROR) break; CreateAndAddNewIfTableEntry (ifTableEntryVar.ifIndex, ifTableEntryVar.ifDescr, ... ); } m2IfDelete(); } } the code looks like it should work ... and it looks nearly identical to the code for the other m2Lib tables ... so i have no idea why i can't get data for this table -- or at least, i can't get it w/out killing the task any suggestions/pointers are welcome thanks, dave From vxwexplo-errs@csg.lbl.gov Sat May 4 04:03:13 2002 From: Vxworks Exploder Date: Sat May 4 04:03:15 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat May 4 04:03:09 PDT 2002 Subject: Re: Difference between RTOS and normal OS Subject: Bsp Testing Subject: Timer Driver Subject: Re: Bsp Testing Subject: Re: m8260FccEnd driver transmit errors Subject: Mac Address Subject: Re: Mac Address Subject: network statics Subject: Re: Mac Address Subject: Re: Mac Address Subject: Re: Interrupt Latency Subject: Re: Mac Address Subject: Re: Default route outside subnet Subject: Re: Bsp Testing Subject: Re: Mac Address Subject: Re: Mutual exclusion between same-priority tasks Subject: Re: Mac Address Subject: Re: How to reduce the size of ENVOY SNMP module Subject: Re: RTOS project Subject: Re: Difference between RTOS and normal OS Subject: Re: how to use different frequency to run tasks ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Fri, 03 May 2002 13:10:21 GMT From: "doug dotson" Organization: Verio Message-ID: <1%vA8.18774$A%3.180066@ord-read.news.verio.net> References: <507efed5.0205012037.3e11728f@posting.google.com> Unix has always had priority-based preemptive scheduling but it doesn't perform well as an RTOS. doug "Yin, Yong" wrote in message news:aardvf$2bp$1@bcarh8ab.ca.nortel.com... > priority-based preemptive schedule > > Vivek wrote in message > news:507efed5.0205012037.3e11728f@posting.google.com... > > Hi, > > > > I am looking for some concrete differences between RTOS and normal OS. > > Could any body point me to the right direction for this. > > > > Thanks and in advance > > > > Vivek > > --------------------------- Newsgroups: comp.os.vxworks Subject: Bsp Testing Date: 3 May 2002 06:38:58 -0700 From: vxworks_research@hotmail.com (Anamika) Organization: http://groups.google.com/ Message-ID: <9e221d1.0205030538.4466b5b1@posting.google.com> Hi I am developing a BSP for PowerPC. I have written all the drivers and taken care about the memory initialization also. On burning the bootrom image to the ROM, I am getting some exceptions. I want to debug the code, which I hv written. Can anyone suggest the ways of debugging for BSP. I dont hv a JTAG. Thanking you in advance Regards Anamika --------------------------- Newsgroups: comp.os.vxworks Subject: Timer Driver Date: 3 May 2002 06:40:34 -0700 From: vxworks_research@hotmail.com (Anamika) Organization: http://groups.google.com/ Message-ID: <9e221d1.0205030540.59b4849a@posting.google.com> Hi I am writing a timer driver for vxWorks. I am having some doubts in calculating sysClkTicks. How is this sysClkTicks calculated. What all factors has to be taken into account for creating this sysClkTicks. Thanking you in advance Regards Anamika --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bsp Testing Date: Fri, 03 May 2002 14:16:53 GMT From: "doug dotson" Organization: Verio Message-ID: References: <9e221d1.0205030538.4466b5b1@posting.google.com> Trial and error and alot of perseverence. I'm in the middle of debugging one now and it is slow and painful. Fortunately I have a bunch of LEDs I can talk to until I can get it to talk to a terminal. No JTAG debugger or emulator available for this beast either. doug "Anamika" wrote in message news:9e221d1.0205030538.4466b5b1@posting.google.com... > Hi > > I am developing a BSP for PowerPC. I have written all the drivers and > taken care about the memory initialization also. On burning the > bootrom image to the ROM, I am getting some exceptions. I want to > debug the code, which I hv written. Can anyone suggest the ways of > debugging for BSP. I dont hv a JTAG. > > Thanking you in advance > > Regards > Anamika --------------------------- Newsgroups: comp.os.vxworks Subject: Re: m8260FccEnd driver transmit errors Date: Fri, 3 May 2002 19:56:57 +0000 (UTC) From: "Graham Baxter" Organization: Graham Baxter (Software) Limited Message-ID: References: I had the same problem when we connected the transmit and receive clocks the wrong way round. They can be changed with software. My device was exactly oposite to the EST evaluation board. I hope this helps. - -- Regards, Graham Baxter Freelance Software Engineer Graham Baxter (Software) Limited EMAIL: gbaxter@NOSPAMbcs.org.uk "Achim Zimmer" wrote in message news:aatei0$ctprc$1@ID-102608.news.dfncis.de... > I try to use the m6260FccEnd driver, supplied by WRS eith ADS8260 BSP to use > with our own board. > The connection is equal, but we use a 971 Phy chip instead of 970. > > The problem we are running in, is: > The packets on the LAN the board transmits have in 80 % errors, means, there > are 8 bytes repeated in the frame, and theoriginal 8 bytes are lost. > > Has anyone this problem too and there are any solutions ? > > Thanks in advance. > Any hint is welcome. > > --------------------------- Newsgroups: comp.os.vxworks Subject: Mac Address Date: Fri, 3 May 2002 13:32:35 -0700 From: "MK" Organization: mike_texx@mail.ru Message-ID: Reply-To: "MK" > Message 3 in thread > From: gold@sdsu.edu (gold@sdsu.edu) > Subject: RE: MAC address > Newsgroups: comp.os.vxworks > View this article only > Date: 2000-09-15 10:00:05 PST > > In article <8ptasj$7io$1@overload.lbl.gov>, > Jim Way wrote: > Martin, > > > Where does vxWorks get the MAC address from when displaying it in > > the ifShow routine? > > > > Is it possible to modify this MAC address dynamically? > > > > Thanks, > > Martin Mayer > > It gets it from the interface structure it keeps around internally. It > doesn't call the driver everytime you do an ifShow() to get the MAC > address. The MAC address is squirreled away in the struct when the > interface is brought up. If you have the source to an END driver, you > see it gets put in the mib2tbl.ifPhysAddress.phyAddress or something via > the END_MIB_INIT() macro. The upper layer code also makes an ioctl call > with EIOCGADDR to get the MAC address, at which time it saves it. > Apparently you can also make an EIOCSADDR ioctl to set the MAC address > in the MIB, then your code would have to make sure it gets propagated > out to everywhere else. Here is the link to the full discussion of this subject: http://groups.google.com/groups?hl=en&threadm=8ptjtu%2460t%241%40nnrp1.deja. com&rnum=1&prev=/groups%3Fq%3DMac%2Baddress%2Bgroup:comp.os.vxworks%26hl%3De n%26selm%3D8ptjtu%252460t%25241%2540nnrp1.deja.com%26rnum%3D1 My problem is, that when I do an ifShow(), the MAC address that gets displayed is different than the value that I see in my END_DEVICE structure pDrvCtrl->end->mib2Tbl->ifPhysAddress->phyAddress Why is that? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mac Address Date: 03 May 2002 16:56:29 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "MK" == MK writes: [snip] MK> Here is the link to the full discussion of this subject: MK> http://groups.google.com/groups?hl=en&threadm=8ptjtu%2460t%241%40nnrp1. MK> deja.com&rnum=1&prev=/groups%3Fq%3DMac%2Baddress%2Bgroup:comp.os. MK> vxworks%26hl%3Den%26selm%3D8ptjtu%252460t%25241%2540nnrp1.deja.com%26 MK> rnum%3D1 A saner URL, "http://groups.google.com/groups?th=8d00f083cf27cbc9" MK> My problem is, that when I do an ifShow(), the MAC address that MK> gets displayed is different than the value that I see in my MK> END_DEVICE structure pDrvCtrl-> end->mib2Tbl->ifPhysAddress->phyAddress MK> Why is that? Because they are different. The discussion you sited says that the `ifShow' command call an ioctl() function with "EIOCGADDR" to retrieve the MAC address. I imagine that the MAC will be the same when the original ioctl() is called. The problem with this is that the driver might not know the MAC when the interface comes up. It might take a while to read out a MAC address. Dave Laight also mentioned that the info is in the "if_net[]" array. You could peruse through this to discover things. I have done the same thing with the file system array. It is possible that some bad code overwrites this stuff. You might have other weird effects if this happens. hth, Bill Pringlemeir. - -- Hydraulic pizza oven!! Guided missile! Herring sandwich! Styrofoam! Jayne Mansfield! Aluminum siding! Borax! Pedal pushers! Jukebox! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: network statics Date: Fri, 03 May 2002 15:36:17 -0700 From: Graham Waters Organization: TRIUMF Message-ID: <3CD310E1.FD9E8C37@triumf.ca> Reply-To: Graham.Waters@triumf.ca I am trying to track down a persistent network communication problem on a system of about 20 mvme162. They all work together as part of a particle accelerator control system (EPICS). From time to time we lose connectivity between one or more of the 162's and a solaris host. We suspect network storms of as yet unknown origin. I am attemping to time correlate the occurences by logging network stats gleaned from "ifShow". I have also looked directly at data in the interface structure pointed to by the function "ifunit". Other than "packets received" none of the values gathered by the two methods seem to agree. As you can see ifShow reveals a number of input errors and no unknown protocol counts. The raw data show quite the reverse sort of. Can anyone tell me what the numbers really mean?. Does ifShow some how juggles the raw data or does it come from some other data source?. Thanks Graham. >ifShow "ei0" ei (unit number 0): Flags: (0x8063) UP BROADCAST MULTICAST ARP RUNNING Type: ETHERNET_CSMACD Internet address: 142.90.132.20 Broadcast address: 142.90.255.255 Netmask 0xffff0000 Subnetmask 0xffff0000 Ethernet address is 00:01:af:02:06:20 Metric is 0 Maximum Transfer Unit size is 1500 0 octets received 0 octets sent 649597 packets received 2028 packets sent 649597 unicast packets received 0 unicast packets sent 0 non-unicast packets received 2028 non-unicast packets sent 0 input discards 0 input unknown protocols 5644 input errors 0 output errors 0 collisions; 0 dropped > raw dta from structure "if_date" packets received on interface 649607 input errors on interface 0 packets sent on interface 6949 output errors on interface 0 collisions on csma interfaces 0 total number of octets received 74396849 total number of octets sent 484530 packets received via multicast 643742 packets sent via multicast 1879 dropped on input, this interface 0 destined for unsupported protocol 272741 +---------------------------------------------------------------------+ |Graham Waters, Control System Eng phone: (604)-222-1047 ext 6531 | |TRIUMF --- University of B.C, Fax: (604)-222-7307 | |Vancouver, BC, Canada e-mail: Graham.Waters@triumf.ca | | | |http://isacwserv.triumf.ca/homepage/graham01.html | +---------------------------------------------------------------------+ For every expert there is an equal and opposite expert - Arthur C Clark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mac Address Date: Fri, 3 May 2002 15:48:43 -0700 From: "MK" Organization: mike_texx@mail.ru Message-ID: References: Reply-To: "MK" Something doesn't agree here :( For debugging purposes I changegd my ioctl , and now EIOCGADDR returns a pre-set fake MAC 0x050403020100, but the ifShow() still displays a different one. I do see 0x050403020100 in pDrvCtrl-> end->mib2Tbl->ifPhysAddress->phyAddress though... also, tell me please what is an if_net[] array and how do I access it? it doesn't seem to be a symbol in my build.... "Bill Pringlemeir" wrote in message news:uelgtlzlu.fsf@sympatico.ca... > >>>>> "MK" == MK writes: > > [snip] > > MK> Here is the link to the full discussion of this subject: > MK> http://groups.google.com/groups?hl=en&threadm=8ptjtu%2460t%241%40nnrp1. > MK> deja.com&rnum=1&prev=/groups%3Fq%3DMac%2Baddress%2Bgroup:comp.os. > MK> vxworks%26hl%3Den%26selm%3D8ptjtu%252460t%25241%2540nnrp1.deja.com%26 > MK> rnum%3D1 > > A saner URL, > > "http://groups.google.com/groups?th=8d00f083cf27cbc9" > > MK> My problem is, that when I do an ifShow(), the MAC address that > MK> gets displayed is different than the value that I see in my > MK> END_DEVICE structure pDrvCtrl-> end->mib2Tbl->ifPhysAddress->phyAddress > > MK> Why is that? > > Because they are different. The discussion you sited says that the > `ifShow' command call an ioctl() function with "EIOCGADDR" to retrieve > the MAC address. > > I imagine that the MAC will be the same when the original ioctl() is > called. The problem with this is that the driver might not know the > MAC when the interface comes up. It might take a while to read out a > MAC address. > > Dave Laight also mentioned that the info is in the "if_net[]" array. > You could peruse through this to discover things. I have done the > same thing with the file system array. It is possible that some bad > code overwrites this stuff. You might have other weird effects if > this happens. > > hth, > Bill Pringlemeir. > > -- > Hydraulic pizza oven!! Guided missile! Herring sandwich! Styrofoam! > Jayne Mansfield! Aluminum siding! Borax! Pedal pushers! Jukebox! > > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mac Address Date: 03 May 2002 19:09:40 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "MK" == MK writes: MK> Something doesn't agree here :( For debugging purposes I changegd MK> my ioctl , and now EIOCGADDR returns a pre-set fake MAC MK> 0x050403020100, but the ifShow() still displays a different MK> one. I do see 0x050403020100 in pDrvCtrl-> end-> mib2Tbl->ifPhysAddress->phyAddress though... Can you post the code for the ioctl() ? What is the ifShow MAC? MK> also, tell me please what is an if_net[] array and how do I MK> access it? it doesn't seem to be a symbol in my build.... I believe this will be the same as the "pDrvCtrl" that you are looking at. What version of vxWorks and what BSP/CPU are you using? Regards, Bill Pringlemeir. - -- Yow!! It's LIBERACE and TUESDAY WELD!! High on a HILL... driving a LITTLE CAR... I wanna be in that LITTLE CAR, too!! I wanna drive off with LIBBY and TUESDAY! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Date: Sat, 4 May 2002 00:14:07 +0200 From: "VU Khac Tri \(home\)" Organization: FUNDP, Namur, Belgium Message-ID: References: Hi, I implement a driver for a PCI device, but never got so long delay. You may want to check whether this delay is caused by another ISR or not. Another running ISR may mask the interrupt line. Anyhow, I am surprise that a ISR performs so long operation. You can check the interrupt line using osciloscope. Tri. "Peter Kannegiesser" wrote in message news:aer2dug9ot75urttc2dabeg507f5v44p7t@4ax.com... > Hi all! > > We are using Tornado 2.02 and the related tools and have written a > driver for an D/A, A/D converter PCI card. The driver is implemented > using ioctl() calls. The PCI card generates interrupts at the end of > conversion. > The interrupt handler is connected using pciConnectInt and is > definitly the only one for that interrupt line. > Now we experience very looong delays from the interrupt signal on > the PCI bus until the handler gets called. According to the logic > analyzer, it takes between 5 and 15 milliseconds (with a strong > tendency to 15ms)-- and this supposed to be an OS for fast hard > realtime... > We are using the standard Pentium BSP on various PC's, from 166MHz > Pentium I to 2GHz Pentium 4. There seems to be no change that > depends on the processor speed. > > Has anybody seen similar problems or could offer at least an > explanation? > > TIA, > Peter > -- > keine sig --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mac Address Date: Fri, 3 May 2002 16:44:15 -0700 From: "MK" Organization: mike_texx@mail.ru Message-ID: References: Reply-To: "MK" int cs8900aIoctl ( END_OBJ * pDriverCtrl, /* device receiving command */ int cmd, /* ioctl command code */ caddr_t data /* command argument */ ) { END_DEVICE * pDrvCtrl=(END_DEVICE*)pDriverCtrl; int error = 0; long value; DRV_LOG (DRV_DEBUG_IOCTL, "cs8900aIoctl(%p,0x%x,0x%x)\r\n",pDriverCtrl,cmd,data, 4, 5, 6); switch (cmd) { case EIOCSADDR: DRV_LOG (DRV_DEBUG_IOCTL, "cs8900aIoctl(): EIOCSADDR\r\n", 1, 2, 3, 4, 5, 6); if (data == NULL) return (EINVAL); bcopy ((char *)data, (char *)END_HADDR(&pDrvCtrl->end), END_HADDR_LEN(&pDrvCtrl->end)); break; case EIOCGADDR: DRV_LOG (DRV_DEBUG_IOCTL, "cs8900aIoctl(): EIOCGADDR\r\n", 1, 2, 3, 4, 5, 6); if (data == NULL) return (EINVAL); //taken out 5/2/02. Instead of relying on mib2 structure we'll give it the real MAC // bcopy ((char *)END_HADDR(&pDrvCtrl->end),(char *)data,END_HADDR_LEN(&pDrvCtrl->end)); cs8900aEnd_eepromRd_macAddr(pDrvCtrl,data); //TEST CODE, gives MAC from non-volatile memory break; ..... ..... ...... //here is the code that fakes the retrieval of MAC (in reality this MAC will be stored in the non-volatile memory on the system) void cs8900aEnd_eepromRd_macAddr(END_DEVICE *pDrvCtrl,UINT8 *pBuf) { int count; for (count = 0; count<=5; count++) { *(pBuf+count) = count; } }/*get Ethernet MAC address from EEPROM*/ ifShow() MAC is cc:c5:17:00:ff:ff Actually, I am fighting a greater problem, than the ifShow()-displayed MAC. I just thought that solving this mystery would be a step in the right direction. THE REAL PROBLEM: The problem I'm facing is that I discovered using a network sniffer, that all the packets sent from my system have a source ethernet (MAC) address of ff:ff:ff:ff:ff:ff (broadcast). Obviously my driver is somehow misconfigured, and I am currently trying to find out how and where does it get the MAC address that it uses to build an ethernet frame. This problem is higher priority for me than the ifShow()Mac, but I felt they could be related . Thanks. "Bill Pringlemeir" wrote in message news:u8z70n80b.fsf@sympatico.ca... > >>>>> "MK" == MK writes: > > MK> Something doesn't agree here :( For debugging purposes I changegd > MK> my ioctl , and now EIOCGADDR returns a pre-set fake MAC > MK> 0x050403020100, but the ifShow() still displays a different > MK> one. I do see 0x050403020100 in pDrvCtrl-> > end-> mib2Tbl->ifPhysAddress->phyAddress though... > > Can you post the code for the ioctl() ? > What is the ifShow MAC? > > MK> also, tell me please what is an if_net[] array and how do I > MK> access it? it doesn't seem to be a symbol in my build.... > > I believe this will be the same as the "pDrvCtrl" that you are looking > at. What version of vxWorks and what BSP/CPU are you using? > > Regards, > Bill Pringlemeir. > > -- > Yow!! It's LIBERACE and TUESDAY WELD!! High on a HILL... driving a > LITTLE CAR... I wanna be in that LITTLE CAR, too!! I wanna drive > off with LIBBY and TUESDAY! > > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Default route outside subnet Date: 3 May 2002 17:20:11 -0700 From: dave.moore@conexant.com (Dave Moore) Organization: http://groups.google.com/ Message-ID: References: Johan Borkhuis wrote in message news:... > I have the following problem. I have a device that gets an IP address and > subnetmask, but also a gateway. These are fixed values. This gateway > address however is outside of the subnet. So for example: > > IP address = 202.204.206.208:ffffff00 > Gateway = 175.185.195.205 > > How can I configure the routing to do this? > > Groeten, > Johan Your details are somewhat "sketchy", but you obviously have a misconfiguration problem. I am not sure what you mean by "fixed values". If they are compiled into the bootline, then fix them! If your "device" gets this information via DHCP then fix the configuration of the DHCP server. Bottom line is that to send an IP datagram off the local subnet you MUST go thru a router that has an interface on the subnet as your device. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bsp Testing Date: 3 May 2002 17:59:17 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205031659.6ccd1706@posting.google.com> References: <9e221d1.0205030538.4466b5b1@posting.google.com> Hello, > I am developing a BSP for PowerPC. I have written all the drivers and > taken care about the memory initialization also. On burning the > bootrom image to the ROM, I am getting some exceptions. I want to > debug the code, which I hv written. Can anyone suggest the ways of > debugging for BSP. I dont hv a JTAG. It might be well worth getting a JTAG debugger if there is one available for your board (or could be connected to it). That might save you some time in the long run. If that's not possible, here are some basic tips: 1) Do not use the compressed format image; it makes things more difficult to debug. In fact, I would suggest working not with the bootrom image at all; instead used a ROM version of vxWorks itself, and add the target tools to it (i.e. the target based shell, standalone symbol table etc). 2) Reduce the kernel to an absolute minimum. Take out everything except the serial port support (e.g. no network code, no disk support etc). Burn that image, and get it working. If it crashes at this point, you really are left with only LEDs on the board, or some external hardware assist (e.g. a JTAG solution). 3) Once the serial-only config works, add the network code and one driver. If this crashes, you should at least have an address where the failure was, and with some work you can find where in the code it occurred. If you are running a full kernel, with a bit of luck you'll have the target shell to help you. If the crash is proving difficult to find, leave the network code in the system, but remove the init calls for it. That way it will boot up to the target shell prompt, and you can make the calls from there to see which one fails. You should also be able to get a stack trace which might prove helpful. 4) As long as the code that is crashing is after the serial I/O init, you can use printf() or logMsg() to generate diagnostic messages. This is often useful in tracing calls and checking parameters. 5) For cases where the thing crashes and reboots, but you don't know where it is. Try using LEDs, or other output devices. If none of those are available, you can always add calls to reboot(1) into the init sequence to see how far you get (this is going to be a slow process with ROM though). HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mac Address Date: 03 May 2002 21:27:24 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "MK" == MK writes: int cs8900aIoctl ( [snip...] Ok, so a cirrus logic CS 8900 a. You appear to be using older code? Can you go to the cirrus logic web sight and get a newer version? "http://www.cirrus.com/Drivers/ethernet/ethernet.cfm" Such as, "http://www.cirrus.com/pubs/drivers/ethernet/VxWorksBSD44v208.zip" "http://www.cirrus.com/pubs/drivers/ethernet/VxWorksENDv306.zip" MK> ifShow() MAC is cc:c5:17:00:ff:ff MK> Actually, I am fighting a greater problem, than the MK> ifShow()-displayed MAC. I just thought that solving this mystery MK> would be a step in the right direction. THE REAL PROBLEM: The MK> problem I'm facing is that I discovered using a network sniffer, MK> that all the packets sent from my system have a source ethernet MK> (MAC) address of ff:ff:ff:ff:ff:ff (broadcast). Obviously my You are really messed up! I have this chip-set on a custom ARM board. I can tell you some things about getting it to work. I use the END driver listed above. I think that your function is wrong. Set the value pDevice->enetAddr, in your function `cs8900a_endLoad' or `cs8900aStart'. I don't think that the function you gave will actually set the MAC. It is a rather old version, so I can not be sure... "http://groups.google.com/groups?as_q=cs8900a&as_ugroup=comp.os.vxworks" hth, Bill Pringlemeir. - -- ATTACKED BY CORVETTE. TORPEDOS. SINKING. U-758. - -- vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mutual exclusion between same-priority tasks Date: Sat, 04 May 2002 01:43:14 GMT From: "Dan P." Organization: Road Runner - EC Message-ID: References: <3CCFFB03.BA438CA4@attbi.com> "Dexter Nixon" wrote in message news:e8938175094e4207fdf83a329228f54a.76014@mygate.mailgate.org... > "Michael Patrick" wrote in message > news:3CCFFB03.BA438CA4@attbi.com > > > How many of you are avoiding the use of semaphores for protecting global > > variables because you think vxworks will schedule (with timeslicing > > disabled) same-priority tasks to always run to completion until they > > voluntarily give up context? > > > > I came across this design technique when reviewing a project in my > > company, and submitted a TSR to ask Windriver about it. I was worried, > > you see, that an interrupt to one task would return through the > > scheduler to a *different* same-priority task, so that same-priority > > tasks would NOT be mutually exclusive. PSOS can do this, for example. > > > ======================================================================= > > Hi Mike, > > Why do we need mutual exclusion between same priority tasks. I could not > find what is the relation between priority and mutual exclusion if both > tasks are not accessing a common memory location at their life time. > Kindly explain me on this. > > THanks, > Nix I would consider it a bad design to assume 2 tasks will always have the same priority. What if down the road it is necessary to make Task A a higher priority than Task B. It will now have to be redesigned to add mutual exclusion. That's assuming the developer who made the priority change is aware of the common memory location. Dan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mac Address Date: Fri, 3 May 2002 18:59:58 -0700 From: "MK" Organization: mike_texx@mail.ru Message-ID: References: Reply-To: "MK" Thanks for the advice. I'll start integrating the new END driver immediately. Anything in particular that you feel I should know before embarking on this journey? "Bill Pringlemeir" wrote in message news:uznzgln2b.fsf@sympatico.ca... > >>>>> "MK" == MK writes: > > int cs8900aIoctl > ( > [snip...] > > Ok, so a cirrus logic CS 8900 a. You appear to be using older code? > Can you go to the cirrus logic web sight and get a newer version? > > "http://www.cirrus.com/Drivers/ethernet/ethernet.cfm" > > Such as, > > "http://www.cirrus.com/pubs/drivers/ethernet/VxWorksBSD44v208.zip" > "http://www.cirrus.com/pubs/drivers/ethernet/VxWorksENDv306.zip" > > MK> ifShow() MAC is cc:c5:17:00:ff:ff > > MK> Actually, I am fighting a greater problem, than the > MK> ifShow()-displayed MAC. I just thought that solving this mystery > MK> would be a step in the right direction. THE REAL PROBLEM: The > MK> problem I'm facing is that I discovered using a network sniffer, > MK> that all the packets sent from my system have a source ethernet > MK> (MAC) address of ff:ff:ff:ff:ff:ff (broadcast). Obviously my > > You are really messed up! I have this chip-set on a custom ARM board. > I can tell you some things about getting it to work. I use the END > driver listed above. > > I think that your function is wrong. Set the value pDevice->enetAddr, > in your function `cs8900a_endLoad' or `cs8900aStart'. I don't think > that the function you gave will actually set the MAC. It is a rather > old version, so I can not be sure... > > "http://groups.google.com/groups?as_q=cs8900a&as_ugroup=comp.os.vxworks" > > > hth, > Bill Pringlemeir. > > -- > ATTACKED BY CORVETTE. TORPEDOS. SINKING. U-758. > > -- > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to reduce the size of ENVOY SNMP module Date: Sat, 04 May 2002 03:15:51 GMT From: Les Cargill Organization: AT&T Worldnet Message-ID: <3CD35445.8123C2F0@worldnet.att.net> References: Jason wrote: > > Hello: > > I have successfully built the ENVOY SNMP module for PPC, but it quite > big (1.5MBytes). I hope to reduce it. Does any one have the idea ? > > One way I want to try is, among the 11 groups of MIB II, only include > a few of them, say the system and ip, exclude the rest 9 groups. But > how to do it exactly? > Use the "-exclude" option when invoking the MIB compiler. > Thanks in advance! - -- http://home.att.net/~lcargill --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RTOS project Date: Sat, 04 May 2002 06:41:20 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: Software only? What is the target platform? Any instrumentation involved? "Neena76" wrote in message news:e7fe924.0205010928.6816a190@posting.google.com... > Hello, > > I'm looking for a small embedded RTOS programming project for a class. If > anyone has some interesting ideas or could point me in the right > direction, I would appreciate it. > > Thanks, > Neena --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Sat, 04 May 2002 06:47:55 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> An RTOS provides hard timing guarantees to perform various items of work (interrupt latency, context switches, system services), so that a designer can make a fairly accurate prediction of the system's deterministic behavior over time. A normal OS typically never provides hard timing guarantees, since that is not a requirement of a standard OS. "Vivek" wrote in message news:507efed5.0205022254.3db92d02@posting.google.com... > "Yin, Yong" wrote in message news:... > > priority-based preemptive schedule > but in unix/linux also we can modify the process priority etc, then > why arn't these are RTOS. ? > Sorry, just a confusion . > > > > > Vivek wrote in message > > news:507efed5.0205012037.3e11728f@posting.google.com... > > > Hi, > > > > > > I am looking for some concrete differences between RTOS and normal OS. > > > Could any body point me to the right direction for this. > > > > > > Thanks and in advance > > > > > > Vivek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use different frequency to run tasks Date: Sat, 04 May 2002 01:36:15 GMT From: "Dan P." Organization: Road Runner - EC Message-ID: References: <4U0A8.61867$kq1.2108928@news20.bellglobal.com> "Jianyuan HU" wrote in message news:4U0A8.61867$kq1.2108928@news20.bellglobal.com... > Hi, > > I am newie in VxWorks, but I have a lot of experience with AMX (RTOS) for > real time controller. I want to switch to VxWorks. > > Our robot controller has 5 tasks, different tasks has different priority and > different execution frequncy. > for instance, Task1 needs to run once every 1ms, while task2 run once every > 20ms, and so on. > > taskSpawn can't do this job, Watch Dog Timer can't do either, but I think > there should be some function to do this job, for AMX can do that easily, > only we need to do is to set the timer for each task. > taskDelay() should do the trick. Dan --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun May 5 04:03:08 2002 From: Vxworks Exploder Date: Sun May 5 04:03:10 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun May 5 04:03:05 PDT 2002 Subject: Re: how to use different frequency to run tasks Subject: FS: Force CPCI-731 LP PIII 500MHz 128MB CPCI Boards in Quantity --NEW-- Subject: semTake() after intLock() Subject: Re: semTake() after intLock() Subject: Re: Pipes and Message Queues....blocking? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use different frequency to run tasks Date: Sat, 04 May 2002 09:28:34 -0400 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3CD3E202.2CFB329E@bellsouth.net> References: <4U0A8.61867$kq1.2108928@news20.bellglobal.com> "Dan P." wrote: > "Jianyuan HU" wrote in message > news:4U0A8.61867$kq1.2108928@news20.bellglobal.com... > > Hi, > > > > I am newie in VxWorks, but I have a lot of experience with AMX (RTOS) for > > real time controller. I want to switch to VxWorks. > > > > Our robot controller has 5 tasks, different tasks has different priority > and > > different execution frequncy. > > for instance, Task1 needs to run once every 1ms, while task2 run once > every > > 20ms, and so on. > > > > taskSpawn can't do this job, Watch Dog Timer can't do either, but I think > > there should be some function to do this job, for AMX can do that easily, > > only we need to do is to set the timer for each task. > > > > taskDelay() should do the trick. > > Dan If the O.P. wants delays as short as 1 mS, he's probably going to need hardware timers. It is generally not recommende to set the system time tick as short as 1 mS. Also, he needs to make sure he doesn't include networking in the target, because the network task is know to take up to 5mS at a time, and (usually) is unbreakable, so it could seriously screw him up. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks,comp.os.os9,comp.arch.bus.vmebus,comp.arch.embedded Subject: FS: Force CPCI-731 LP PIII 500MHz 128MB CPCI Boards in Quantity --NEW-- Date: 4 May 2002 13:01:05 -0700 From: wmcalpin@ghgcorp.com (Bill) Organization: http://groups.google.com/ Message-ID: Look here for more info on this board: http://www.forcecomputers.com/products/viewDetail.cfm?productTypeID=2&productID=45 Look here to download an Excel file with our inventory listing: http://www.sunharborsales.com/shsvmefs.htm We also have a selection of other CPCI and VME boards. - -Bill Sun Harbor Sales --------------------------- Newsgroups: comp.os.vxworks Subject: semTake() after intLock() Date: Sat, 04 May 2002 20:18:47 -0000 From: Dominique Pelle Organization: Cisco Systems Inc. Keywords: semTake intLock block Message-ID: Hi, I stumbled upon a piece of code which takes a VxWorks mutex semaphore within a intLock()/intUnlock() section: - --- 8< --- cut here --- 8< --- cut here --- 8< --- lockKey = intLock(); /* ... */ rc = semTake(mutexSem, timeout); if (rc == OK) { /* ... */ } else { /* ... */ } /* ... */ intUnlock(lockKey); - --- 8< --- cut here --- 8< --- cut here --- 8< --- I thought we could not block after intLock() (hence taking a semaphore was a no-no). However after looking at it, experimenting and reading the reference manual, I see that blocking after inLock() will schedule another task. Scheduler will unlock interrupts as part of restoration of the new task TCB. When the task that locked interrupts and blocked on semaphore becomes ready again (semaphore was given), it will be rescheduled and interrupts will be locked again. If that's correct, I suppose that the above code is somewhat equivalent to: - --- 8< --- cut here --- 8< --- cut here --- 8< --- lockKey = intLock(); /* ... */ intUnlock(lockKey); rc = semTake(mutexSem, timeout); lockKey = intLock(); if (rc == OK) { /* ... */ } else { /* ... */ } /* ... */ intUnlock(lockKey); - --- 8< --- cut here --- 8< --- cut here --- 8< --- Second piece of code has more overhead and will unlock interrupts even when semTake does not block. However, it does not take the semaphore in intLock. It makes it explicit that interrupts are unlocked. It should be better for interrupt latency as it always unlocks interrupts. I'd like to make sure that this is really what is happening. - -> Is it acceptable to take a mutex semaphore after intLock? (assuming we understand that interrupts might be temporarily unlocked as result) Taking a semaphore in intLock looks like a poor coding technique to me at best and buggy at worse. The piece of code has been running well so far but it does not mean it is correct. Thanks - -- Dominique Pelle He who re-invents the wheel, understands much better how a wheel works. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semTake() after intLock() Date: Sun, 05 May 2002 05:41:33 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Mon Ami, your observations are prefectly correct - technically thecode is OK, there is no harm in doing semTake while interrrupts are locked, except that semTake and possible rescheduling is too long an operation to be allowed while interrupts are locked, and will definitely disturb overall system responsiveness. intLock should only be used to protect a variable used by a task and ISR, while Mutex semaphore should be used to protect a variable used by more then one task. Both of those Mutex methods are mutually exclusive ;-)) "Dominique Pelle" wrote in message news:slrnad8gh7.dbe.dpelle@mssbu-build1.cisco.com... > Hi, > > I stumbled upon a piece of code which takes a VxWorks > mutex semaphore within a intLock()/intUnlock() section: > > --- 8< --- cut here --- 8< --- cut here --- 8< --- > lockKey = intLock(); > /* ... */ > > rc = semTake(mutexSem, timeout); > if (rc == OK) { > /* ... */ > } else { > /* ... */ > } > /* ... */ > intUnlock(lockKey); > --- 8< --- cut here --- 8< --- cut here --- 8< --- > > > I thought we could not block after intLock() (hence taking > a semaphore was a no-no). > > However after looking at it, experimenting and reading > the reference manual, I see that blocking after inLock() > will schedule another task. Scheduler will unlock > interrupts as part of restoration of the new task TCB. > > When the task that locked interrupts and blocked on > semaphore becomes ready again (semaphore was given), > it will be rescheduled and interrupts will be locked > again. > > If that's correct, I suppose that the above code is > somewhat equivalent to: > > --- 8< --- cut here --- 8< --- cut here --- 8< --- > lockKey = intLock(); > /* ... */ > > intUnlock(lockKey); > rc = semTake(mutexSem, timeout); > lockKey = intLock(); > if (rc == OK) { > /* ... */ > } else { > /* ... */ > } > > /* ... */ > intUnlock(lockKey); > --- 8< --- cut here --- 8< --- cut here --- 8< --- > > > Second piece of code has more overhead and will unlock > interrupts even when semTake does not block. However, it > does not take the semaphore in intLock. It makes it > explicit that interrupts are unlocked. It should be > better for interrupt latency as it always unlocks > interrupts. > > I'd like to make sure that this is really what is > happening. > > -> Is it acceptable to take a mutex semaphore after > intLock? (assuming we understand that interrupts > might be temporarily unlocked as result) > > Taking a semaphore in intLock looks like a poor coding > technique to me at best and buggy at worse. The piece > of code has been running well so far but it does not > mean it is correct. > > Thanks > > -- > Dominique Pelle > He who re-invents the wheel, understands much better how a wheel works. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Pipes and Message Queues....blocking? Date: Sun, 5 May 2002 10:01:04 +0000 (UTC) From: "Dexter Nixon" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: References: <6755aee5.0203291212.6c96bd2b@posting.google.com> <6755aee5.0204080843.251faa68@posting.google.com> > Yeah, the > if ((ioctl (fd, FIONMSGS, &nMessages)!=ERROR) && (nMessages == 0)) > write (fd, message,50); > > was the work around I came up with. I'm fully aware of the race > condition, but my application is currently working. > > I figured that > status = ioctl (sFd, FIONBIO, &on); > > would not work since I'm not using a socket...this is just a pipe. > Micheal, It should work, in VxWorks, pipes are implemented on top of message queues and the return value when open a pipe is a "fd". If you can use the fd with select means, what is the problem in using with a ioctl. It will work. status = ioctl (sFd, FIONBIO, &on); This is the right way to set it to Non-blocking mode. Thanks, Nix > Aufeidersehen, > > Whacked - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun May 5 23:01:40 2002 From: Raghavan Date: Sun May 5 23:01:42 PDT 2002 Subject: ICMP Chekcsum error Hi, I have a x86 based board running vxWorks version 5.4. From this board, when I start a ping with odd number of bytes, the receiver complains that the ICMP checksum is wrong. I executed the same test with the loopback address "127.0.0.1". In this case, the vxWorks stack itself complains of checksum error. Has anyone had this problem? I am using Tornado version 2.0 Thanking in advance, Raghavan. From vxwexplo-errs@csg.lbl.gov Mon May 6 00:27:21 2002 From: Goldmann Andreas Date: Mon May 6 00:27:24 PDT 2002 Subject: Qt for VxWorks This message is in MIME format. Since your mail reader does not understand this format, some or all of this message may not be legible. --------------InterScan_NT_MIME_Boundary Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C1F4CF.69F29E00" ------_=_NextPart_001_01C1F4CF.69F29E00 Content-Type: text/plain; charset="ISO-8859-1" Hi people, Does anybody know, whether there is a Qt-port for VxWorks ? With regards, Andreas ------_=_NextPart_001_01C1F4CF.69F29E00 Content-Type: text/html; charset="ISO-8859-1" Qt for VxWorks

Hi people,
Does anybody know, whether there is a Qt-port for VxWorks ?

With regards, Andreas

------_=_NextPart_001_01C1F4CF.69F29E00-- --------------InterScan_NT_MIME_Boundary-- From vxwexplo-errs@csg.lbl.gov Mon May 6 02:01:16 2002 From: "Pradeepa Venkatswamy" Date: Mon May 6 02:01:18 PDT 2002 Subject: fclose and mempartfree in vxworks This is a multi-part message in MIME format. ------=_NextPartTM-000-dafbcebc-60cb-11d6-af80-0080c8048dde Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit HI, memPartFree reports invalid memory block in fclose call.I did "lkAddr" on the reported invalid block.It is pointing out only one global structure that been used.I don't know how to track this problem and fix it....I am not sure on the usage of "lkAddr" even.........This problem is not occurring frequently but it does occur once in a while...How to track and solve this problem??? Thank you, V.Pradeepa. ------=_NextPartTM-000-dafbcebc-60cb-11d6-af80-0080c8048dde Content-Type: text/plain; name="Wipro_Disclaimer.txt" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="Wipro_Disclaimer.txt" **************************Disclaimer************************************ Information contained in this E-MAIL being proprietary to Wipro Limited is 'privileged' and 'confidential' and intended for use only by the individual or entity to which it is addressed. You are notified that any use, copying or dissemination of the information contained in the E-MAIL in any manner whatsoever is strictly prohibited. ******************************************************************** ------=_NextPartTM-000-dafbcebc-60cb-11d6-af80-0080c8048dde-- From vxwexplo-errs@csg.lbl.gov Mon May 6 04:03:09 2002 From: Vxworks Exploder Date: Mon May 6 04:03:11 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon May 6 04:03:05 PDT 2002 Subject: Re: Mac Address Subject: Re: Pipes and Message Queues....blocking? Subject: Re: m8260FccEnd driver transmit errors Subject: Re: m8260FccEnd driver transmit errors Subject: Re: Bsp Testing Subject: Re: How to make a bootrom for hard disk without using target Subject: Re: Porting code from GNU compiler to DIAB compiler.... ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Mac Address Date: 05 May 2002 19:15:29 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: Sender: bpringlemeir@DeadDuck MK> Thanks for the advice. I'll start integrating the new END driver MK> immediately. Anything in particular that you feel I should know MK> before embarking on this journey? Read the newsgroup archives, "http://groups.google.com/groups?as_q=cs8900a&as_ugroup=comp.os.vxworks" Many of the problems mentioned have been fixed in the latest driver. I won't say it is the most efficient thing on the planet, but it isn't bad. Other than that, follow the instruction, read the comments. I don't think you need anymore help than that. Regards, Bill Pringlemeir. - -- Stealing a tetrodotoxin should not be done while wearing a girdle. Look for your spouse with one eye. Your programs will start washing clothes. Don't forget to exercise. You'll later be abused. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Pipes and Message Queues....blocking? Date: 5 May 2002 23:02:07 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205052202.1c9684bd@posting.google.com> References: <6755aee5.0203291212.6c96bd2b@posting.google.com> <6755aee5.0204080843.251faa68@posting.google.com> "Dexter Nixon" wrote in message news:... > > Yeah, the > > if ((ioctl (fd, FIONMSGS, &nMessages)!=ERROR) && (nMessages == 0)) > > write (fd, message,50); > > > > was the work around I came up with. I'm fully aware of the race > > condition, but my application is currently working. > > > > I figured that > > status = ioctl (sFd, FIONBIO, &on); > > > > would not work since I'm not using a socket...this is just a pipe. > > > Micheal, > > It should work, in VxWorks, pipes are implemented on top of message > queues and the return value when open a pipe is a "fd". If you can use > the fd with select means, what is the problem in using with a ioctl. It > will work. Not necessarily. Just because a device implements some ioctl() functions doesn't mean it will do them all. In fact, the pipe driver does not implement FIONBIO so you will not be able to use the line below to get non-blocking behaviour. > status = ioctl (sFd, FIONBIO, &on); Of course, if you need non-blocking pipes they are pretty trivial to implement yourself... you can even do more unix-like stream based pipes if you don't like the standard ones. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: m8260FccEnd driver transmit errors Date: Mon, 6 May 2002 10:21:14 +0200 From: "Achim Zimmer" Message-ID: References: Thats fine -> Thay can be changed with software !! But I dont know why, because this driver is only delivered as Object, no source ! We found that it could be, we have to invert transmit clock, but I dont know how, because the driver always overwrites my register reconfiguration. Thanks "Graham Baxter" schrieb im Newsbeitrag news:aauq28$ahb$1@knossos.btinternet.com... > I had the same problem when we connected the transmit and receive clocks the > wrong way round. > > They can be changed with software. > > My device was exactly oposite to the EST evaluation board. > > I hope this helps. > > -- > > Regards, > > Graham Baxter > Freelance Software Engineer > Graham Baxter (Software) Limited > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > > "Achim Zimmer" wrote in message > news:aatei0$ctprc$1@ID-102608.news.dfncis.de... > > I try to use the m6260FccEnd driver, supplied by WRS eith ADS8260 BSP to > use > > with our own board. > > The connection is equal, but we use a 971 Phy chip instead of 970. > > > > The problem we are running in, is: > > The packets on the LAN the board transmits have in 80 % errors, means, > there > > are 8 bytes repeated in the frame, and theoriginal 8 bytes are lost. > > > > Has anyone this problem too and there are any solutions ? > > > > Thanks in advance. > > Any hint is welcome. > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: m8260FccEnd driver transmit errors Date: Mon, 6 May 2002 10:21:51 +0200 From: "Achim Zimmer" Message-ID: References: sorry, i dont know HOW to change !! "Achim Zimmer" schrieb im Newsbeitrag news:ab5ef3$f68hn$1@ID-102608.news.dfncis.de... > Thats fine -> Thay can be changed with software !! > > But I dont know why, because this driver is only delivered as Object, no > source ! > > We found that it could be, we have to invert transmit clock, but I dont know > how, because the driver always overwrites my register reconfiguration. > > Thanks > > "Graham Baxter" schrieb im Newsbeitrag > news:aauq28$ahb$1@knossos.btinternet.com... > > I had the same problem when we connected the transmit and receive clocks > the > > wrong way round. > > > > They can be changed with software. > > > > My device was exactly oposite to the EST evaluation board. > > > > I hope this helps. > > > > -- > > > > Regards, > > > > Graham Baxter > > Freelance Software Engineer > > Graham Baxter (Software) Limited > > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > > > > > > "Achim Zimmer" wrote in message > > news:aatei0$ctprc$1@ID-102608.news.dfncis.de... > > > I try to use the m6260FccEnd driver, supplied by WRS eith ADS8260 BSP to > > use > > > with our own board. > > > The connection is equal, but we use a 971 Phy chip instead of 970. > > > > > > The problem we are running in, is: > > > The packets on the LAN the board transmits have in 80 % errors, means, > > there > > > are 8 bytes repeated in the frame, and theoriginal 8 bytes are lost. > > > > > > Has anyone this problem too and there are any solutions ? > > > > > > Thanks in advance. > > > Any hint is welcome. > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bsp Testing Date: Mon, 06 May 2002 10:26:48 +0200 From: Emmanuel Herbreteau Organization: SEPRO Robotique (BEP) Message-ID: <3CD63E48.846B3384@sepro-robotique.com> References: <9e221d1.0205030538.4466b5b1@posting.google.com> Anamika wrote: > I am developing a BSP for PowerPC. .../... > debug the code, which I hv written. Can anyone suggest the ways of > debugging for BSP. I dont hv a JTAG. Hi, Is there a BDM port available on your board ? There are several BDM debugger available for PowerPC nowadays. Except if you are working for free, it's definitively needed to reduce the development costs. Regards Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to make a bootrom for hard disk without using target Date: 6 May 2002 02:51:21 -0700 From: forbes_bofeng@sohu.com (forbes) Organization: http://groups.google.com/ Message-ID: <7d3efabf.0205060151.5f52e710@posting.google.com> References: <7d3efabf.0204291800.ba00f5a@posting.google.com> <20020429.225027.619290071.18588@socal.rr.com> Hello: Ken Hayber! First, thank your help! Partition g of my target disk is my first partition on the drive. What I want to ask you is my way that make a bootrom for hard disk is right? If my way is wrong, please tell me a right way, thank you very much! _/_/_/_/ _/ _/_/_/_/ _/ forbes_bofeng@sohu.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Porting code from GNU compiler to DIAB compiler.... Date: Tue, 30 Apr 2002 08:23:24 GMT From: "Ofer Goren" Organization: Verio Message-ID: References: <1f1b48b5.0204291841.72d0f7db@posting.google.com> Hi. We had the same problem. First, make sure you are using diab 4.4b, not 4.4a. Then, read the release notes of this patch. In short: here is an example of a define that was changed to support diab: #define MODULE_DEBUG(id,level,format,... ) \ moduleDebug(id, level, "<%s %d>: " ## format, __FILE__ ,\ __LINE__ , __VA_ARGS__) and here is the original, gnu-like line: #define MODULE_DEBUG(id,level,format,args... ) \ moduleDebug(id, level, "<%s %s %d>: " ## format, __FILE__ , \ __FUNCTION__ , __LINE__ , ## args ) as you can see, the magic is in the __VA_ARGS__, which was added only in 4.4b. Pay attention, that the macro __FUNCTION__ is not supported by diab. only by GNU. Again, the question arise, does windriver REALLY support diab compiler, or just saying they do? HTH, Ofer, Israel. "john c" wrote in message news:1f1b48b5.0204291841.72d0f7db@posting.google.com... > Yikes. I'm not big on BSPs or Compilers so I have a quick question. > We are currently using GNU compiler with vxWorks to compile our code > for PPC 750. I'm trying to switch over to the diab compiler and the > diab compiler is complaining about the following lines: > > #define TRACE(a, b, args...) sys_Trace(a,b,__FILE__,__LINE__,## args) > #define V5_TRACE(a, b, args...) V5_Sys_Trace(a,b,## args) > #define TRACEF1(a, b, ig, crv, args...) if(chk_TraceF1(ig, crv)) > sys_Trace(a,b,__FILE__,__LINE__,## args) > #define DTRACE(a, b, args...) > > and the diab compiler is complaining about.... > "/vobs/software/components/inc/sys_ext.h", line 738: error (dcc:2291): > illformed macro parameter list in macro `TRACE' > "/vobs/software/components/inc/sys_ext.h", line 739: error (dcc:2291): > illformed macro parameter list in macro `V5_TRACE' > "/vobs/software/components/inc/sys_ext.h", line 740: error (dcc:2291): > illformed macro parameter list in macro `TRACEF1' > "/vobs/software/components/inc/sys_ext.h", line 745: error (dcc:2291): > illformed macro parameter list in macro `DTRACE' > > any ideas? > john --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue May 7 04:03:12 2002 From: Vxworks Exploder Date: Tue May 7 04:03:15 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue May 7 04:03:08 PDT 2002 Subject: WindView : Post-Mortem configuration Subject: [AE] boot debug help Subject: 0x770b0f8 (tSnmpd): memPartFree: invalid block 0x7fff1d0 in partition 0x202cfc. Subject: Re: semTake() after intLock() Subject: Re: Bsp Testing Subject: Re: How to make a bootrom for hard disk without using target Subject: Re: Bsp Testing Subject: PMC SPAN Cards Subject: Symbol table error Subject: Re: bootrom not loading kernel Subject: Re: m8260FccEnd driver transmit errors Subject: Re: Symbol table error Subject: Re: Symbol table error Subject: Using The Parallel Port on the MVME2700 for External Interrupts Subject: Re: [AE] boot debug help Subject: Re: 0x770b0f8 (tSnmpd): memPartFree: invalid block 0x7fff1d0 in partition 0x202cfc. Subject: Re: Interrupt Latency ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: WindView : Post-Mortem configuration Date: 6 May 2002 05:03:52 -0700 From: nassim15@aol.com (Nassim Benzerrouk) Organization: http://groups.google.com/ Message-ID: <5cbe966c.0205060403.43a3b446@posting.google.com> Hi VxWorkers, I've some difficulties to configure the Post-Mortem configuration on Tornado2, the help given in the WindView GUI is not clear enough; so I want the steps necessary to configure this mode. Thanks in advance, ___________________________________________ | Nassim Benzerrouk | | OneAccess | /) +33 1 41 87 70 06 (\ / ) Email:nassim.benzerrouk@oneaccess-net.com ( \ _( (|___________________________________________) ) /> (((\ \) /,) / ) / //))/ (\\\\ \_/ / \ \_/ ///// \ / \ / \ _/ \_ / --/ /----------------------------------------\ \---- / / \ \ --------------------------- Newsgroups: comp.os.vxworks Subject: [AE] boot debug help Date: 6 May 2002 05:38:31 -0700 From: strick@hiwaay.net (BS) Organization: http://groups.google.com/ Message-ID: I am currently working on getting on of our new Pentium3 boards to boot with VxWorks AE. We already have several of our legacy P3 boards booting just fine within AE, but for some reason I never get very far in the boot process with this new board. There several differences in this new board and the legacy boards, but I have been told that these should not affect me in any way? (DOS will boot just fine.) I assume that I am not getting to the bootShellInit() part of the boot process because I never see the boot logo. At this point the only know 'problems' with this new board is that we are unable to read from COM1, however we can write with no problem. All I am interested with at this point is to get to the [VxWorks AE Boot] prompt. I am currently just using the Pentium3 base BSP delivered with VxWorks AE until I can get futher in this process to start testing the custom Question: Does anyone have any advise about how I can debug this boot process so that I can detect where I am failing. I am not very familiar with the romInit.s, romStart.c, and sysALib.s files. Thank you for any help. --------------------------- Newsgroups: comp.os.vxworks Subject: 0x770b0f8 (tSnmpd): memPartFree: invalid block 0x7fff1d0 in partition 0x202cfc. Date: 6 May 2002 05:56:12 -0700 From: gowill@ako.net (eunjeong) Organization: http://groups.google.com/ Message-ID: <4c98a82e.0205060456.60969abb@posting.google.com> I'm facing this error code. "0x770b0f8 (tSnmpd): memPartFree: invalid block 0x7fff1d0 in partition 0x202cfc." I didn't alloc(or free) of memory. When return from sub-funtion to main-funtion, this error generated. I have no idea this problem. please help me..... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: semTake() after intLock() Date: Mon, 6 May 2002 10:00:15 -0400 From: Keith Arner Organization: Marconi Message-ID: References: Reply-To: Keith Arner On Sat, 4 May 2002, Dominique Pelle wrote: > I stumbled upon a piece of code which takes a VxWorks > mutex semaphore within a intLock()/intUnlock() section: > > --- 8< --- cut here --- 8< --- cut here --- 8< --- > lockKey = intLock(); > /* ... */ > > rc = semTake(mutexSem, timeout); > if (rc == OK) { > /* ... */ > } else { > /* ... */ > } > /* ... */ > intUnlock(lockKey); > --- 8< --- cut here --- 8< --- cut here --- 8< --- > > I thought we could not block after intLock() (hence taking > a semaphore was a no-no). > > However after looking at it, experimenting and reading > the reference manual, I see that blocking after inLock() > will schedule another task. Scheduler will unlock > interrupts as part of restoration of the new task TCB. Blocking from ISR context is a no-no. Blocking with interrupts locked is possible, and has the semantics that you discovered. The thing to realize is that being in ISR context and having interrupts locked are independant of one another. > When the task that locked interrupts and blocked on > semaphore becomes ready again (semaphore was given), > it will be rescheduled and interrupts will be locked > again. > > If that's correct, I suppose that the above code is > somewhat equivalent to: > > --- 8< --- cut here --- 8< --- cut here --- 8< --- > lockKey = intLock(); > /* ... */ > > intUnlock(lockKey); > rc = semTake(mutexSem, timeout); > lockKey = intLock(); > if (rc == OK) { > /* ... */ > } else { > /* ... */ > } > > /* ... */ > intUnlock(lockKey); > --- 8< --- cut here --- 8< --- cut here --- 8< --- The main difference here is what happens if the semaphore is available on the call to semTake(). In the first case, interrupts never get unlocked. In the second case, interrupts do get unlocked for a short period, which may open up race conditions, depending on what the intLock() is attempting to protect against. If your timeout is non-zero, and the semaphore is unavailable, the same race conditions occur in both the cases. However, if the timeout is zero, and the semaphore is unavailable, the first case will not have race conditions, but the second case will. > -> Is it acceptable to take a mutex semaphore after > intLock? (assuming we understand that interrupts > might be temporarily unlocked as result) > > Taking a semaphore in intLock looks like a poor coding > technique to me at best and buggy at worse. The piece > of code has been running well so far but it does not > mean it is correct. "Poor coding technique" is probably a fair statement. The key is to identify what the intLock() in there for in the first place. What it it attempting to protect against? Can the same protection be gained by using another semaphore, or even taskLock()? Keith --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bsp Testing Date: Mon, 6 May 2002 16:30:54 +0200 From: "Rene Straub" <_remove_rene.straub@yetnet.ch> Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: References: <9e221d1.0205030538.4466b5b1@posting.google.com> Hey folks, I had the same pleasure some days ago. I helped myself with some LEDs (LEDs are cool for debugging) and by writing debug data to low memory addresses. Using a BDM device I could read the data back after the code run (and crashed). If your code crashes start by placing a end-less loop in the code. If it doesn't crash anymore the problem must be behind the loop. Then find out why it crashed and fix it. Repeat this step until it works... Hope this helps - - Rene "doug dotson" wrote in message news:pZwA8.18788$A%3.180255@ord-read.news.verio.net... > Trial and error and alot of perseverence. I'm in the middle of > debugging one now and it is slow and painful. Fortunately I have > a bunch of LEDs I can talk to until I can get it to talk to a > terminal. No JTAG debugger or emulator available for this > beast either. > > doug > > "Anamika" wrote in message > news:9e221d1.0205030538.4466b5b1@posting.google.com... > > Hi > > > > I am developing a BSP for PowerPC. I have written all the drivers and > > taken care about the memory initialization also. On burning the > > bootrom image to the ROM, I am getting some exceptions. I want to > > debug the code, which I hv written. Can anyone suggest the ways of > > debugging for BSP. I dont hv a JTAG. > > > > Thanking you in advance > > > > Regards > > Anamika > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to make a bootrom for hard disk without using target Date: Mon, 6 May 2002 18:33:08 +0200 From: "Werner Schiendl" Message-ID: <3cd6b03c$1@brateggebdc5.br-automation.co.at> References: <7d3efabf.0204291800.ba00f5a@posting.google.com> <20020429.225027.619290071.18588@socal.rr.com> <7d3efabf.0205060151.5f52e710@posting.google.com> Hi, I'm not 100% sure, but IIRC the vxsys utility for DOS encodes the drive letter somehow into the boot sector. We had the problem long time ago, and since then use to boot up with a DOS floppy the first time and write it with that onto the harddisk (vxsys runs from pure DOS, but not Windows 95 - where a additional invocation of the 'lock' command is required before). hth Werner "forbes" wrote in message news:7d3efabf.0205060151.5f52e710@posting.google.com... > Hello: > Ken Hayber! First, thank your help! > Partition g of my target disk is my first partition on the drive. > What I want to ask you is my way that make a bootrom for hard disk is > right? If my way > is wrong, please tell me a right way, thank you very much! > > _/_/_/_/ > _/ > _/_/_/_/ > _/ forbes_bofeng@sohu.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bsp Testing Date: Mon, 6 May 2002 10:28:35 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: <9e221d1.0205030538.4466b5b1@posting.google.com> I have used various available mechanisms to debug low level BSP issues. For example, LEDs, PromICE, BDM, JTAG, Logic analyser, scope, Rom emulator, serial port, shared memory (over bus), etc. And of course, a lot of patience and tenacity help. Finally, some good luck. - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: PMC SPAN Cards Date: 6 May 2002 10:42:14 -0700 From: vouxx@yahoo.com (vouxx) Organization: http://groups.google.com/ Message-ID: <86d4086f.0205060942.56df994e@posting.google.com> Hi, I am using PMC SPAN Card for first time and am not sure how to proceed. I fixed it. Defined INCLUDE_PMC_SPAN and build a new bootrom. I could not find and PCI_AUTOCONFIG macro (as mentioned in some earlier posts, the we have to define it). All I found was this comment. mv2600.h:02f,17jul98,tm added PCI_ID_PRI_LAN, PCI_ID_SEC_LAN for PCI AutoConfig The PMC Card comes witha driver that detects the card and gives this information: pcifuncno : 0 pcibusno : 0 pcideviceno : 16 pciaddr : 0x02000000 span : 0x01000000 intline : 0x19 what do these figures mean? Do I need to set some other macros like P2P_IO_BASE, P2P_PREF_MEM_BASE (is this pciaddr?), P2P_NONPREF_MEM_BASE etc? Any links documents giving more information will be welcome. a brief paragraph on vxworks and pmc span and addressing will be great!! thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Symbol table error Date: Mon, 6 May 2002 12:05:25 -0600 From: "cervello" Organization: University of Utah Message-ID: Hi all. I hope this isn't too elementary... We are using an MVME 162-20 board (Motorola 68040 processor) running VxWorks 5.4. We develop our applications in ControlShell 7.0d. When attempting to load our applications onto the board, we get the following error (repeated many times, with xxxx replaced by many different symbol names): loadAoutLib error: can't add 'xxxx" to system symbol table - error = 0x110001. We have been loading the applications using: ld 1, 0, "vxw2_slave_IO.INCLUDED_PACKAGED.lo" Can anyone point me in the right direction on this? Any idea what the error number represents? Thanks, Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bootrom not loading kernel Date: Mon, 06 May 2002 13:24:37 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3CD6CA65.70B7C2FD@fnal.gov> References: <3CC9DF96.621B2327@fnal.gov> <3CD017DF.8EF3935@fnal.gov> It turn out that I had installed June 2000 driver patch over the T2CP4 patch. There was some confusion about what came first (I had asked about it on comp.os.vxworks earlier). That patch made some bad updates in libPPC604gnuvx.a. Using the libPPC604gnuvx.a from T2CP4 takes care of the problem. Thanks Dinker Dinker Charak wrote: > > Johan Borkhuis wrote: > > > > > [VxWorks Boot]: n dc > > > Attaching network interface dc0... done. > > > Error setting inet address of dc0 to 0, errno = 0 > > > Cannot initialize interface named "dc0" > > > > Just a quick suggestion (I don't have a board available at this moment): > > The error says "Error settint inet addr to 0": did you set a network > > address in the boot parameters? > > Yes I have. Actually, I tried this "n dc" on another board that boots up > properly. It also gave the same result. So am not sure what thsi command > does or coveys. > > Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: m8260FccEnd driver transmit errors Date: Mon, 6 May 2002 19:04:27 +0000 (UTC) From: "Graham Baxter" Organization: Graham Baxter (Software) Limited Message-ID: References: Achim, You can do everything in: sysFccEnetEnable() in sysLib.c where you set the correct port pins. The register I changed for clock routes was something like: M8260_CMXFCR - -- Regards, Graham Baxter Freelance Software Engineer Graham Baxter (Software) Limited (VxWorks and pSOS BSP's) EMAIL: gbaxter@NOSPAMbcs.org.uk "Achim Zimmer" wrote in message news:ab5eg7$f3404$1@ID-102608.news.dfncis.de... > sorry, i dont know HOW to change !! > > "Achim Zimmer" schrieb im Newsbeitrag > news:ab5ef3$f68hn$1@ID-102608.news.dfncis.de... > > Thats fine -> Thay can be changed with software !! > > > > But I dont know why, because this driver is only delivered as Object, no > > source ! > > > > We found that it could be, we have to invert transmit clock, but I dont > know > > how, because the driver always overwrites my register reconfiguration. > > > > Thanks > > > > "Graham Baxter" schrieb im Newsbeitrag > > news:aauq28$ahb$1@knossos.btinternet.com... > > > I had the same problem when we connected the transmit and receive clocks > > the > > > wrong way round. > > > > > > They can be changed with software. > > > > > > My device was exactly oposite to the EST evaluation board. > > > > > > I hope this helps. > > > > > > -- > > > > > > Regards, > > > > > > Graham Baxter > > > Freelance Software Engineer > > > Graham Baxter (Software) Limited > > > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > > > > > > > > > > "Achim Zimmer" wrote in message > > > news:aatei0$ctprc$1@ID-102608.news.dfncis.de... > > > > I try to use the m6260FccEnd driver, supplied by WRS eith ADS8260 BSP > to > > > use > > > > with our own board. > > > > The connection is equal, but we use a 971 Phy chip instead of 970. > > > > > > > > The problem we are running in, is: > > > > The packets on the LAN the board transmits have in 80 % errors, means, > > > there > > > > are 8 bytes repeated in the frame, and theoriginal 8 bytes are lost. > > > > > > > > Has anyone this problem too and there are any solutions ? > > > > > > > > Thanks in advance. > > > > Any hint is welcome. > > > > > > > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table error Date: Mon, 6 May 2002 21:35:14 +0200 From: "Werner Schiendl" Message-ID: <3cd6daea$1@brateggebdc5.br-automation.co.at> References: Hi Mark, your friend for all kind of error numbers in VxWorks is printErrno: - -> printErrno 0x110001 0x110001 = S_memLib_NOT_ENOUGH_MEMORY value = 0 = 0x0 So something is running out of memory. If you download from the Tornado shell and believe there must be enough memory free on the target (memShow will clarify), consider that the Target Server has a memory pool on it's on. You can set the size of the target server memory pool in the target server configuration. (select 'Memory Cache Size' in 'Target Server Properties' dropdown list to see the setting). hth Werner "cervello" wrote in message news:ab6glf$t6i$1@coward.ks.cc.utah.edu... > Hi all. I hope this isn't too elementary... > > We are using an MVME 162-20 board (Motorola 68040 processor) running VxWorks > 5.4. We develop our applications in ControlShell 7.0d. When attempting to > load our applications onto the board, we get the following error (repeated > many times, with xxxx replaced by many different symbol names): > > loadAoutLib error: can't add 'xxxx" to system symbol table - error = > 0x110001. > > We have been loading the applications using: > ld 1, 0, "vxw2_slave_IO.INCLUDED_PACKAGED.lo" > > > Can anyone point me in the right direction on this? Any idea what the error > number represents? > > Thanks, > > Mark > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Symbol table error Date: Mon, 06 May 2002 20:57:22 GMT From: Minh.Lang@jpl.nasa.gov Organization: Jet Propulsion Laboratory - Pasadena CA Message-ID: <3cd6edbb.15890789@news.jpl.nasa.gov> References: On Mon, 6 May 2002 12:05:25 -0600, "cervello" wrote: >Hi all. I hope this isn't too elementary... > >We are using an MVME 162-20 board (Motorola 68040 processor) running VxWorks >5.4. We develop our applications in ControlShell 7.0d. When attempting to >load our applications onto the board, we get the following error (repeated >many times, with xxxx replaced by many different symbol names): > >loadAoutLib error: can't add 'xxxx" to system symbol table - error = >0x110001. > >We have been loading the applications using: >ld 1, 0, "vxw2_slave_IO.INCLUDED_PACKAGED.lo" > > >Can anyone point me in the right direction on this? Any idea what the error >number represents? > >Thanks, > >Mark 0x110001 is "not enough memory" error. Have you checked your free memory? -Minh- --------------------------- Newsgroups: comp.os.vxworks Subject: Using The Parallel Port on the MVME2700 for External Interrupts Date: Mon, 06 May 2002 17:01:28 -0700 From: Henry Valtier Organization: Jet Propulsion Laboratory - Pasadena CA Message-ID: <3CD71957.FBFD06FD@jpl.nasa.gov> Hello All, Has anyone ever successfully programmed the parallel port on the Motorola MVME2700 to detect and service external interrupts? If you could steer me toward some coding examples (preferably in C), I'd be most appreciative. Thanks in Advance, Henry --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [AE] boot debug help Date: 6 May 2002 22:01:03 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205062101.7204bfb6@posting.google.com> References: Hello, Don't have much in the way of useful hints for debugging Pentium bootApps, but something you might like to do first is isolate the differences between the working boards and the non-working ones. Obvious places to start are in the BIOS - see what the settings are there, and if any of them could be causing the problem. After that, what is the chipset on the new boards? Perhaps that is causing the problem somehow. After that, try to find something you can use to mark progress (perhaps an LED somewhere, or for a PC type platform you can even write to the video memory - doesn't have to be anything more than just something appearing somewhere on the screen). Look in the bootApp project at the prjConfig.c file - this contains the boot sequence - by adding progress markers to this you can get some idea of progress through the sequence. Finally, since you mentioned that you were having problems with the serial port on the board anyway, you might like to make the changes to the bootApp configuration to use the screen + keyboard console instead of COM1. That might come up if the problem relates somehow to the serial devices and/or PCI/ISA handling on the board (perhaps an unrecognised bridge?). HTH, John... strick@hiwaay.net (BS) wrote in message news:... > I am currently working on getting on of our new Pentium3 boards to > boot with VxWorks AE. We already have several of our legacy P3 boards > booting just fine within AE, but for some reason I never get very far > in the boot process with this new board. There several differences in > this new board and the legacy boards, but I have been told that these > should not affect me in any way? (DOS will boot just fine.) > > I assume that I am not getting to the bootShellInit() part of the boot > process because I never see the boot logo. At this point the only know > 'problems' with this new board is that we are unable to read from > COM1, however we can write with no problem. All I am interested with > at this point is to get to the [VxWorks AE Boot] prompt. > > I am currently just using the Pentium3 base BSP delivered with VxWorks > AE until I can get futher in this process to start testing the custom > > Question: Does anyone have any advise about how I can debug this boot > process so that I can detect where I am failing. I am not very > familiar with the romInit.s, romStart.c, and sysALib.s files. > > Thank you for any help. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 0x770b0f8 (tSnmpd): memPartFree: invalid block 0x7fff1d0 in partition 0x202cfc. Date: Tue, 07 May 2002 06:27:33 GMT From: "drdiags" Organization: AT&T Broadband Message-ID: References: <4c98a82e.0205060456.60969abb@posting.google.com> Hello Eunjeong, What version of Tornado/vxWorks for what architecture? Are you using Envoy or the WindNet SNMP (if this is a T2.x.x release). Where you doing a SNMP walk when this happened? Try doing a 'tt tSmnpd" ' to see where the daemon crashed. This could be numerous possibilities, so try narrowing it down a little. (FYI, also look on WindSurf for SNMP SPRs.) Good luck... "eunjeong" wrote in message news:4c98a82e.0205060456.60969abb@posting.google.com... > I'm facing this error code. > > "0x770b0f8 (tSnmpd): memPartFree: invalid block 0x7fff1d0 in partition 0x202cfc." > > I didn't alloc(or free) of memory. > When return from sub-funtion to main-funtion, this error generated. > > I have no idea this problem. > please help me..... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Date: Tue, 7 May 2002 08:27:46 +0000 (UTC) From: "Dexter Nixon" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <64bc9b8e235f63f48777ddb57d7752ac.76014@mygate.mailgate.org> References: Hi Tri, I have a general doubt. I am studying about the performance of VxWorks with respect to interrupts. While am reading a article, i found like, when the highest priority task is running on a VxWorks target and a interrupt is recieved and the ISR is executed means, will there be rescheduling after the ISR exits. THey have mentioned two cases like with rescheduling and without. Is this a configurable parameter. Kindly clear me. THanks. Nix "VU Khac Tri \(home\)" wrote in message news:aavcvd$f6nn@hermes.fundp.ac.be > Hi, > I implement a driver for a PCI device, but never got so long delay. You may > want to check whether this delay is caused by another ISR or not. Another > running ISR may mask the interrupt line. Anyhow, I am surprise that a ISR > performs so long operation. > You can check the interrupt line using osciloscope. > Tri. > > > - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 8 04:03:10 2002 From: Vxworks Exploder Date: Wed May 8 04:03:12 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed May 8 04:03:07 PDT 2002 Subject: Re: network statics Subject: ethernet on sandpoint ppc755 Subject: Re: Envoy version in WindNet SNMP Subject: Re: network statics Subject: How can I see the 'target.nr' file? Subject: Re: network statics Subject: Malformed boot sector. Offset 0, value ... beause of dosFsVolFormat fails error 0x16 Subject: Re: How can I see the 'target.nr' file? Subject: Update the 2.0 BSP to 2.1.1 failed! Subject: Re: Envoy version in WindNet SNMP Subject: Re: Malformed boot sector. Offset 0, value ... beause of dosFsVolFormat fails error 0x16 Subject: Re: How can I see the 'target.nr' file? Subject: Anyone using target Intel x86 (not PENTIUM) ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: network statics Date: Tue, 7 May 2002 10:09:48 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: <3CD310E1.FD9E8C37@triumf.ca> It looks like ifShow() has changed a little over the years. Still, it should be getting the same data from the same "struct ifnet" data structure for a given network interface. A lot of network drivers do not keep track of statistics properly. This is partly caused by the introduction of END driver. Because the END model "hides" the underlying BSD model, a lot of programmers do not pay attention to original BSD structures like "struct ifnet". Each END driver has its own "struct ifnet". It is in "ifnet" the statistics like input, output, error counts are kept. For example, unlike BSD sample drivers, the END sample driver does not illustrate how to account for things like ifp->if_ierror. Input error. Partly it is also due to lack of clear documentation about how statistics are kept. In BSD (and in END) some of the statistic values are kept in "common" routines that are not directly visible to users. Driver writers should be advised so that correct statistics are kept. - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: ethernet on sandpoint ppc755 Date: Tue, 07 May 2002 19:05:03 GMT From: "Mike Hanna" Organization: Road Runner - NC Message-ID: Reply-To: "Mike Hanna" I'm having a problem getting the bootrom to use the ethernet card plugged into a Motorola Sandpoint board. I am able to build a bootrom that I can configure via the serial port. I am attempting to boot via Ethernet/TFTP, but am unable to get the ethernet connection working. I have a Kingston KNE100TX board that has a Digital (Intel labeled) 21143 chip on it. When the bootrom attempts to load the file, nothing comes out of the ethernet port and the sandpoint hangs. There are no error messages and the board doesn't reboot. If I remove the board and retry, it reboots after a while and tries it again. I'm wondering if I need a different Ethernet card. The target.nr file for the sp7400 bsp has a list of supported chips followed by a list of ethernet cards. Do I need one of those? Maybe a different one? Any help would be appreciated. - -- Mike Hanna Vapid Software www.vapidsoftware.com --------------------------- Newsgroups: comp.protocols.snmp,comp.os.vxworks Subject: Re: Envoy version in WindNet SNMP Date: Tue, 07 May 2002 20:44:11 GMT From: Pete Flugstad Organization: MediaCom High Speed Internet Message-ID: <3CD83CFD.1060203@nospam.icon-labs.com> References: Alex Vinokur wrote: > ========================================================= > Tornado : Version 2.0.2 > VxWorks : Version 5.4.2 > BSP : Version 1.1/1 > ------------------------ > Compiler : ccsimpc > gcc version egcs-2.90.29 tornado 2.0 980515 > (egcs-1.0.3 release) > ------------------------ > Wind River Systems Emissary SNMP MIB Compiler, version 7.0. > ------------------------ > Host : Windows2000 > Target : VxSim for Windows (VxWorks target simulator) > ========================================================= > > > > How can we know which version of Envoy (8.x, 9.x, etc) is used in WindNet SNMP ? The file envoy/h/version.h should have the Envoy version in it: #define ENVOY_MAJOR_VERSION 9 #define ENVOY_MINOR_VERSION 1 #define ENVOY_BETA_VERSION 0 I would expect it to be the same as the provided emissary version (i.e. 7.0.0). Pete Flugstad Icon Labs --------------------------- Newsgroups: comp.os.vxworks Subject: Re: network statics Date: 7 May 2002 15:19:20 -0700 From: yli@transdimension.com (liyi) Organization: http://groups.google.com/ Message-ID: References: <3CD310E1.FD9E8C37@triumf.ca> "Hwa Jin Bae" wrote in message news:... > It looks like ifShow() has changed a little over the years. Still, it > should be getting the same data from the same "struct ifnet" data structure > for a given network interface. > > A lot of network drivers do not keep track of statistics properly. This is > partly caused by the introduction of END driver. Because the END model > "hides" the underlying BSD model, a lot of programmers do not pay attention > to original BSD structures like "struct ifnet". Each END driver has its own > "struct ifnet". It is in "ifnet" the statistics like input, output, error > counts are kept. For example, unlike BSD sample drivers, the END sample > driver does not illustrate how to account for things like ifp->if_ierror. > Input error. Partly it is also due to lack of clear documentation about > how statistics are kept. In BSD (and in END) some of the statistic values > are kept in "common" routines that are not directly visible to users. > Driver writers should be advised so that correct statistics are kept. Hi, I just tried to compile the template netif code and do the attachment, but the ifShow showed "can't read "interfacesMasksTbl(0x77620c)": no such element in array". Any idea about this? Thanks, - -> ifShow fei (unit number 0): Flags: (0x8063) UP BROADCAST RUNNING ARP MULTICAST Type: ETHERNET_CSMACD Internet address: 192.160.1.46 Broadcast address: 192.160.1.255 Netmask 0xffffff00 Subnetmask 0xffffff00 Ethernet address is 88:cd:18:00:00:02 Metric is 0 Maximum Transfer Unit size is 1500 5556 packets received; 6 packets sent 5556 multicast packets received 5 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped lo (unit number 0): Flags: (0x8069) UP LOOPBACK RUNNING ARP MULTICAST Type: SOFTWARE_LOOPBACK Internet address: 127.0.0.1 Netmask 0xff000000 Subnetmask 0xff000000 Metric is 0 Maximum Transfer Unit size is 32768 0 packets received; 0 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped can't read "interfacesMasksTbl(0x77620c)": no such element in array --------------------------- Newsgroups: comp.os.vxworks Subject: How can I see the 'target.nr' file? Date: Tue, 7 May 2002 19:16:20 -0400 From: "Jin-Woo Lee" Organization: Cornell University Message-ID: Sender: jl206@cornell.invalid (on 128.253.249.182) Hi, I'm working on mv2400 target board. There is 'target.nr' file in the BSP directory. What is the way to see it? Thanks, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: network statics Date: 7 May 2002 17:05:40 -0700 From: yli@transdimension.com (liyi) Organization: http://groups.google.com/ Message-ID: References: <3CD310E1.FD9E8C37@triumf.ca> "Hwa Jin Bae" wrote in message news:... > It looks like ifShow() has changed a little over the years. Still, it > should be getting the same data from the same "struct ifnet" data structure > for a given network interface. > > A lot of network drivers do not keep track of statistics properly. This is > partly caused by the introduction of END driver. Because the END model > "hides" the underlying BSD model, a lot of programmers do not pay attention > to original BSD structures like "struct ifnet". Each END driver has its own > "struct ifnet". It is in "ifnet" the statistics like input, output, error > counts are kept. For example, unlike BSD sample drivers, the END sample > driver does not illustrate how to account for things like ifp->if_ierror. > Input error. Partly it is also due to lack of clear documentation about > how statistics are kept. In BSD (and in END) some of the statistic values > are kept in "common" routines that are not directly visible to users. > Driver writers should be advised so that correct statistics are kept. Hi, I just tried to compile the template netif code and do the attachment, but the ifShow showed "can't read "interfacesMasksTbl(0x77620c)":no such element in array". Any idea about this? Somebody told me to set the ip address before doing the ifShow and still not work. :( Thanks, -> ifShow fei (unit number 0): Flags: (0x8063) UP BROADCAST RUNNING ARP MULTICAST Type: ETHERNET_CSMACD Internet address: 192.160.1.46 Broadcast address: 192.160.1.255 Netmask 0xffffff00 Subnetmask 0xffffff00 Ethernet address is 88:cd:18:00:00:02 Metric is 0 Maximum Transfer Unit size is 1500 5556 packets received; 6 packets sent 5556 multicast packets received 5 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped lo (unit number 0): Flags: (0x8069) UP LOOPBACK RUNNING ARP MULTICAST Type: SOFTWARE_LOOPBACK Internet address: 127.0.0.1 Netmask 0xff000000 Subnetmask 0xff000000 Metric is 0 Maximum Transfer Unit size is 32768 0 packets received; 0 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped can't read "interfacesMasksTbl(0x77620c)": no such element in array --------------------------- Newsgroups: comp.os.vxworks Subject: Malformed boot sector. Offset 0, value ... beause of dosFsVolFormat fails error 0x16 Date: 7 May 2002 18:15:35 -0700 From: harisha1@hotmail.com (Harish) Organization: http://groups.google.com/ Message-ID: <3b36664c.0205071715.5057dd84@posting.google.com> Can anybody tell me why dosFsVolFormat() fails (error 0x16) and how to solve this?? What does" Malformed boot sector. Offset 0, value 21" means?? sincerely, thanks a lot. I'm using these dos FS 2.0 API's dosFsInit (5); ramDrv (); ramHandle = ramDevCreate(0,512,400,4000,0); dosFsMkfsOptionsSet(DOS_OPT_LONGNAMES); dosFsMkfs("/RamDrive/",ramHandle); I know couple APIS calls are not required. Here is the output on serial port.. RAM DISK drive created OK 81eb9340!, ramHandle 0x81eb9340 Retrieved old volume params with %100 confidence: Volume Parameters: FAT type: FAT16, sectors per cluster 1 10 FAT copies, 0 clusters, 36876 sectors per FAT Sectors reserved 8193, hidden 0, FAT sectors 368760 Root dir entries 224, sysId (null) , serial number a2ce082 Label:" XE$ I " ... Disk with 4000 sectors of 512 bytes will be formatted with: Volume Parameters: FAT type: FAT16, sectors per cluster 2 10 FAT copies, 2147481497 clusters, 8 sectors per FAT Sectors reserved 8193, hidden 0, FAT sectors 80 Root dir entries 224, sysId VXEXT1.1, serial number a2ce082 Label:" XE$ I " ... dosFsVolFormat: format failed, errno=0x16 DOS file system created OK! x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 22, value 36876. 0x81ca2990 (FileTask): dosFsLib.c : Try to accept other boot block copy... 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 248. 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 21. 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 2. 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 96. 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 0. 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 8. 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 24. .. ..... goes on... for a while... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I see the 'target.nr' file? Date: Tue, 07 May 2002 19:23:44 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3CD88C30.4E881288@gv.net> References: Reply-To: mkesti@gv.net Jin-Woo Lee wrote: >Hi, > >I'm working on mv2400 target board. >There is 'target.nr' file in the BSP directory. What is the way to see it? > >Thanks, target.nr is a plain text file that can be viewed and edited using any text editor. "nr" stands for "nroff", a page description "language" not unlike HTML. - -- ======================================================================== 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: Update the 2.0 BSP to 2.1.1 failed! Date: 7 May 2002 20:17:49 -0700 From: xiaohua.zhang@163.com (xiaohua) Organization: http://groups.google.com/ Message-ID: I use the Tornado 2.1.1 for Arm now and I use the Tornado 2.0 for Arm. Now I still use the 2.0 bootrom. It is not need to build a new for 2.1.1. But now I meet the problem - ----------------------------------------------------------- ... ... Starting at 0x600 - -------------------------------------------------------------- Then it stop. Anyone can give me some advice about updating the 2.0 to 2.1.1 for Arm? - ------------------------------------- - --------Any post is welcome!--------- - ------------------------------------- - ------------------------------------- - --------Any post is welcome!--------- - ------------------------------------- --------------------------- Newsgroups: comp.protocols.snmp,comp.os.vxworks Subject: Re: Envoy version in WindNet SNMP Date: Wed, 8 May 2002 08:18:19 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: References: <3CD83CFD.1060203@nospam.icon-labs.com> "Pete Flugstad" wrote in message news:3CD83CFD.1060203@nospam.icon-labs.com... | Alex Vinokur wrote: | > ========================================================= | > Tornado : Version 2.0.2 | > VxWorks : Version 5.4.2 | > BSP : Version 1.1/1 | > ------------------------ | > Compiler : ccsimpc | > gcc version egcs-2.90.29 tornado 2.0 980515 | > (egcs-1.0.3 release) | > ------------------------ | > Wind River Systems Emissary SNMP MIB Compiler, version 7.0. | > ------------------------ | > Host : Windows2000 | > Target : VxSim for Windows (VxWorks target simulator) | > ========================================================= | > | > | > | > How can we know which version of Envoy (8.x, 9.x, etc) is used in WindNet SNMP ? | | The file envoy/h/version.h should have the Envoy version in it: | | #define ENVOY_MAJOR_VERSION 9 | #define ENVOY_MINOR_VERSION 1 | #define ENVOY_BETA_VERSION 0 | | I would expect it to be the same as the provided emissary | version (i.e. 7.0.0). | | Pete Flugstad | Icon Labs | 1. The envoy directory is missing under $WIND_BASE and its subdirectories. 2. The $WIND_BASE/target/h/snmp/envoy.h file contains no relevant information. The $WIND_BASE/target/h/snmp/install.h file contains the following line : #define INSTALL_ENVOY 1 3. No file contains ENVOY_MAJOR_VERSION, ENVOY_MINOR_VERSION, ENVOY_BETA_VERSION Thanks, - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Malformed boot sector. Offset 0, value ... beause of dosFsVolFormat fails error 0x16 Date: Wed, 08 May 2002 07:03:59 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3b36664c.0205071715.5057dd84@posting.google.com> Harish, The problem you are facing is that when memory is allocated for the RAMdisk, for some reason contains what seems to be a valid DossFs boot block, and strangely, dosFsVolFormat() accepts the values of this boot block to be valid, but they are obviously invalid, (e.g. 10 FAT copies is much too much), then dosFs fails to mount the drives because of these values. Also, note that some of the calls you are using, e.g. dosFsMkfs() are not native DosFs 2.0 calls, but are backward-copmatibility routines, emulating dosFs 1.0 calls. I suggest that you avoid the backward compatibility calls, and call dosFsVolFormat() directly, and specifically specify the option which wipes the volume blank before formatting, instructing it to disregard any values it may find on the RAMdisk (see manual for details). - - Leonid "Harish" wrote in message news:3b36664c.0205071715.5057dd84@posting.google.com... > Can anybody tell me why dosFsVolFormat() fails (error 0x16) > and how to solve this?? > What does" Malformed boot sector. Offset 0, value 21" means?? > > sincerely, > thanks a lot. > > > I'm using these dos FS 2.0 API's > dosFsInit (5); > ramDrv (); > ramHandle = ramDevCreate(0,512,400,4000,0); > dosFsMkfsOptionsSet(DOS_OPT_LONGNAMES); > dosFsMkfs("/RamDrive/",ramHandle); > > I know couple APIS calls are not required. Here is the output on serial port.. > > RAM DISK drive created OK 81eb9340!, ramHandle 0x81eb9340 > Retrieved old volume params with %100 confidence: > Volume Parameters: FAT type: FAT16, sectors per cluster 1 > 10 FAT copies, 0 clusters, 36876 sectors per FAT > Sectors reserved 8193, hidden 0, FAT sectors 368760 > Root dir entries 224, sysId (null) , serial number a2ce082 > Label:" XE$ I " ... > Disk with 4000 sectors of 512 bytes will be formatted with: > Volume Parameters: FAT type: FAT16, sectors per cluster 2 > 10 FAT copies, 2147481497 clusters, 8 sectors per FAT > Sectors reserved 8193, hidden 0, FAT sectors 80 > Root dir entries 224, sysId VXEXT1.1, serial number a2ce082 > Label:" XE$ I " ... > dosFsVolFormat: format failed, errno=0x16 > DOS file system created OK! > > x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 22, value 36876. > 0x81ca2990 (FileTask): dosFsLib.c : Try to accept other boot block copy... > 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 248. > 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 21. > 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 2. > 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 96. > 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 0. > 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 8. > 0x81ca2990 (FileTask): dosFsLib.c : Malformed boot sector. Offset 0, value 24. > > .. > ..... goes on... for a while... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I see the 'target.nr' file? Date: 8 May 2002 07:20:56 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3CD88C30.4E881288@gv.net> "Michael R. Kesti" wrote: > Jin-Woo Lee wrote: > >>Hi, >> >>I'm working on mv2400 target board. >>There is 'target.nr' file in the BSP directory. What is the way to see >>it? >> >>Thanks, > > target.nr is a plain text file that can be viewed and edited using any > text editor. "nr" stands for "nroff", a page description "language" > not unlike HTML. But you can also use the files in docs directory. The target.nr file is converted to HTML. You can find it if you open the docs\books.html, go to BSP reference and then select your BSP. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Anyone using target Intel x86 (not PENTIUM) Date: 8 May 2002 02:00:18 -0700 From: heiko_elger@arburg.com (Heiko Elger) Organization: http://groups.google.com/ Message-ID: <97e6274e.0205080100.3e7e1d13@posting.google.com> Hello, who is using a target Intel x86 (386, 486, not Pentium, PcPentium Pro) with Tornado 2.0.x? I'm asking cause in the new Tornado Version 2.2 (VxWorks 5.5) only Intel processors Pentium, Pentium II, Pentium III, Pentium IV will be supported. (the new official roadmap of WRS) For customer with support contract who uses no Pentium or above there will be no ugrade posibbility and only minor bug fixes for Tornado 2.0.x I'm asking cause we have support contract and we are using x86 and we have no possibility to use a Pentium next time. We think that cause of our support contract based on a Intel 80486 processor WRS must support x86 in Tornado 2.2. We want to know whether there are other projects out there which are based on Intel x86 or whether we are the only company using x86. best regards Heiko Elger --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 8 23:15:03 2002 From: "Jayakara N" Date: Wed May 8 23:15:06 PDT 2002 Subject: How to get route table entries?? This is a multi-part message in MIME format. ------=_NextPart_000_0024_01C1F74C.F20D0800 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi all, I am implementing MIB-II IP group in vxworks. For m2IpRouteTblEntryGet = i need all destination ip address of the route table. How can i read route table entries?=20 Thanks in advance=20 Regard Jaykar N Ionic Microsystem Banglore =20 Note: Unless otherwise noted, the information provided by this mail does not represent the official statements or views of Ionic Microsystems. Privileged/Confidential information may be contained in this message and may be subject to legal privilege. Access to this e-mail by anyone other than the intended is unauthorised. If you are not the intended recipient (or responsible for delivery of the message to such person), you may not use, copy, distribute or deliver this message (or any part of its contents ) to anyone or take any action in reliance on it. In such case, you should destroy this message, and notify us immediately. If you have received this email in error, please notify us immediately by e-mail or telephone and delete the e-mail from any computer. If you or your employer does not consent to internet e-mail messages of this kind, please notify us immediately. All reasonable precautions have been taken to ensure no viruses are present in this e-mail. As our company cannot accept responsibility for any loss or damage arising from the use of this e-mail or attachments we recommend that you subject these to your virus checking procedures prior to use. ------=_NextPart_000_0024_01C1F74C.F20D0800 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi all,
 I am implementing MIB-II IP group = in vxworks.=20 For m2IpRouteTblEntryGet  i need all destination ip address of the = route=20 table.
How can i read route table entries? =
Thanks in advance
Regard
Jaykar N
Ionic Microsystem
Banglore
 

Note:
Unless otherwise noted, the information provided by this mail does not represent the official statements or views of Ionic Microsystems.
Privileged/Confidential information may be contained in this message and may be subject to legal privilege. Access to this e-mail by anyone other than the intended is unauthorised. If you are not the intended recipient (or responsible for delivery of the message to such person), you may not use, copy, distribute or deliver this message (or any part of its contents ) to anyone or take any action in reliance on it. In such case, you should destroy this message, and notify us immediately. If you have received this email in error, please notify us immediately by e-mail or telephone and delete the e-mail from any computer.
If you or your employer does not consent to internet e-mail messages of this kind, please notify us immediately. All reasonable precautions have been taken to ensure no viruses are present in this e-mail. As our company cannot accept responsibility for any loss or damage arising from the use of this e-mail or attachments we recommend that you subject these to your virus checking procedures prior to use.
------=_NextPart_000_0024_01C1F74C.F20D0800-- From vxwexplo-errs@csg.lbl.gov Thu May 9 01:52:15 2002 From: "srujana R" Date: Thu May 9 01:52:17 PDT 2002 Subject: Re: How to get route table entries?? Hi, m2IpRouteTblEntryGet( ) routine compares routing table entries to the address specified by the ipRouteDest member of the pIpRouteTblEntry structure, and retrieves an entry if a match is found. If the address is not specified, the m2IpRouteTblEntryGet( ) retrives the first entry in the routing table. The remaining entries in the routing table can be retreived by doing a SNMP GetNext request on the recieved entry from m2IpRouteTblEntryGet( ) call. The subsequent GetNext requests will retreive the remaining entries in the routing table sequentially. Regards Srujana On Thu, 09 May 2002 the vxWorks Users Group Exploder wrote : >Submitted-by vxwexplo-errs@csg.lbl.gov Wed May 8 23:15:03 >2002 >Submitted-by: "Jayakara N" > > >This is a multi-part message in MIME format. > >------=_NextPart_000_0024_01C1F74C.F20D0800 >Content-Type: text/plain; > charset="iso-8859-1" >Content-Transfer-Encoding: quoted-printable > >Hi all, > I am implementing MIB-II IP group in vxworks. For >m2IpRouteTblEntryGet = >i need all destination ip address of the route table. >How can i read route table entries?=20 >Thanks in advance=20 >Regard >Jaykar N >Ionic Microsystem >Banglore > >=20 > > >Note: > Unless otherwise noted, the information provided by this mail >does not represent the official statements or views of Ionic >Microsystems. > Privileged/Confidential information may be contained in this >message and may be subject to legal privilege. Access to this >e-mail by anyone other than the intended is unauthorised. If you >are not the intended recipient (or responsible for delivery of >the message to such person), you may not use, copy, distribute or >deliver this message (or any part of its contents ) to anyone or >take any action in reliance on it. In such case, you should >destroy this message, and notify us immediately. If you have >received this email in error, please notify us immediately by >e-mail or telephone and delete the e-mail from any computer. >If you or your employer does not consent to internet e-mail >messages of this kind, please notify us immediately. All >reasonable precautions have been taken to ensure no viruses are >present in this e-mail. As our company cannot accept >responsibility for any loss or damage arising from the use of >this e-mail or attachments we recommend that you subject these to >your virus checking procedures prior to use. > >------=_NextPart_000_0024_01C1F74C.F20D0800 >Content-Type: text/html; > charset="iso-8859-1" >Content-Transfer-Encoding: quoted-printable > > > >http-equiv=3DContent-Type> > > > > >
Hi all,
>
 I am implementing MIB-II IP >group = >in vxworks.=20 >For m2IpRouteTblEntryGet  i need all destination ip address of >the = >route=20 >table.
>
How can i read route table >entries? = >
>
Thanks in advance >
>
Regard
>
Jaykar N
>
Ionic Microsystem
>
Banglore
>
 
>
>Note:
> Unless otherwise noted, the information provided by this mail >does not represent the official statements or views of Ionic >Microsystems.
> Privileged/Confidential information may be contained in this >message and may be subject to legal privilege. Access to this >e-mail by anyone other than the intended is unauthorised. If you >are not the intended recipient (or responsible for delivery of >the message to such person), you may not use, copy, distribute or >deliver this message (or any part of its contents ) to anyone or >take any action in reliance on it. In such case, you should >destroy this message, and notify us immediately. If you have >received this email in error, please notify us immediately by >e-mail or telephone and delete the e-mail from any >computer.
>If you or your employer does not consent to internet e-mail >messages of this kind, please notify us immediately. All >reasonable precautions have been taken to ensure no viruses are >present in this e-mail. As our company cannot accept >responsibility for any loss or damage arising from the use of >this e-mail or attachments we recommend that you subject these to >your virus checking procedures prior to use.
> > >------=_NextPart_000_0024_01C1F74C.F20D0800-- > > > >********** > > This is a user group mailing list for vxWorks related >topics > see http://www-vxw.lbl.gov/vxworks for posting and >subscription info _________________________________________________________ Click below to visit monsterindia.com and review jobs in India or Abroad http://monsterindia.rediff.com/jobs From vxwexplo-errs@csg.lbl.gov Thu May 9 04:03:18 2002 From: Vxworks Exploder Date: Thu May 9 04:03:21 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu May 9 04:03:13 PDT 2002 Subject: Exception in dosFsMkfs() Subject: Re: Exception in dosFsMkfs() Subject: Re: On S_iosLib_INVALID_FILE_DESCRIPTOR Subject: Re: Exception in dosFsMkfs() Subject: on "No rule to make" Subject: Re: Exception in dosFsMkfs() Subject: Multiple Kill() executions @ interrupt level. Subject: TrueFFS optional??? Subject: Re: TrueFFS optional??? Subject: Kernel WorkQ Overflow problem. Subject: Re: TrueFFS optional??? Subject: Re: Bsp Testing Subject: Converting Tornado 1.0.1 BSP to Tornado 2 Subject: Re: Kernel WorkQ Overflow problem. Subject: Re: Kernel WorkQ Overflow problem. Subject: Re: Kernel WorkQ Overflow problem. Subject: Re: Anyone using target Intel x86 (not PENTIUM) Subject: Re: Interrupt Latency Subject: Re: How can I see the 'target.nr' file? Subject: Creating a bootable VxWorks image Subject: Re: Envoy version in WindNet SNMP Subject: Re: TrueFFS optional??? Subject: VxMP and POSIX? Subject: Re: on "No rule to make" Subject: Code for unresolved symbols not updated Subject: Re: Code for unresolved symbols not updated Subject: Is it possible to build a simulator myself? Subject: Re: TrueFFS optional??? Subject: Re: Exception in dosFsMkfs() Subject: Advice on TCP/UDP task arrangement Subject: cross compiling gcc - parse error Subject: Compilation log file Subject: Re: Bsp Testing Subject: Re: Advice on TCP/UDP task arrangement Subject: Re: cross compiling gcc - parse error ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Exception in dosFsMkfs() Date: Wed, 8 May 2002 15:11:34 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: ========================================================= Tornado : Version 2.0.2 VxWorks : Version 5.4.2 BSP : Version 1.1/1 Compiler : ccsimpc gcc version egcs-2.90.29 tornado 2.0 980515 (egcs-1.0.3 release) Host : Windows2000 Target : VxSim for Windows (VxWorks target simulator) ========================================================= Project Type : bootable VxWorks image (custom configured) ========================================================= I have got an exception after calling the dosFsMkfs() function. Something is wrong. What exactly ? Thanks in advance. || ###### C Code : BEGIN ###### // File s2.c #include #include #include #include #include # define RAM_DISK_DEV "rd:" static DOS_VOL_DESC *pVolDesc = NULL ; void usrAppInit (void) { BLK_DEV *pBlkDev = NULL ; #ifdef USER_APPL_INIT USER_APPL_INIT; /* for backwards compatibility */ #endif pBlkDev = ramDevCreate ( NULL, /* where it is in memory (0 = malloc) */ 512, /* number of bytes per block */ 0, /* number of blocks per track */ 600, /* number of blocks on this device */ 0 /* no. of blks to skip at start of device */ ); assert (pBlkDev); printf ("[%s, %d] ### pBlkDev = %p\n", __FILE__, __LINE__, pBlkDev); /* Line#29 */ pVolDesc = dosFsMkfs (RAM_DISK_DEV, pBlkDev); printf ("[%s, %d] ### pVolDesc = %p\n", __FILE__, __LINE__, pVolDesc); if (! pVolDesc) printf ("DOS volume init failed\n"); assert (pVolDesc); } || ###### C Code : END ######## || ====== Target Simulator Window : BEGIN ====== VxWorks Copyright 1984-1998 Wind River Systems, Inc. CPU: VxSim for Windows VxWorks: 5.4.2 BSP version: 1.1/1 Creation date: May 8 2002 WDB: Ready. [D:/projects_MBX/demo1/01/src2/s2.c, 29] ### pBlkDev = 0x4b5cc08 Exception ! Vector 13 : General Protection Fault Program Counter : 0x0040f1db Status Register : 0x00010246 || ====== Target Simulator Window : END ======== - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception in dosFsMkfs() Date: Wed, 08 May 2002 12:41:26 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Sasha, this is a wild guess: Try to set "blocks per track" to a non-zero value, perhaps this causes some zero divide error. Also, your RAmdisk is onloy 300KBytes in size, hope that is your intention. - Leonid "Alex Vinokur" wrote in message news:abb4il$gn9jc$1@ID-79865.news.dfncis.de... > ========================================================= > Tornado : Version 2.0.2 > VxWorks : Version 5.4.2 > BSP : Version 1.1/1 > Compiler : ccsimpc > gcc version egcs-2.90.29 tornado 2.0 980515 > (egcs-1.0.3 release) > Host : Windows2000 > Target : VxSim for Windows (VxWorks target simulator) > ========================================================= > Project Type : bootable VxWorks image (custom configured) > ========================================================= > > > I have got an exception after calling the dosFsMkfs() function. > Something is wrong. What exactly ? > > Thanks in advance. > > || ###### C Code : BEGIN ###### > // File s2.c > #include > #include > > #include > #include > #include > > > # define RAM_DISK_DEV "rd:" > > static DOS_VOL_DESC *pVolDesc = NULL ; > > void usrAppInit (void) > { > BLK_DEV *pBlkDev = NULL ; > > #ifdef USER_APPL_INIT > USER_APPL_INIT; /* for backwards compatibility */ > #endif > > pBlkDev = ramDevCreate ( > NULL, /* where it is in memory (0 = malloc) */ > 512, /* number of bytes per block */ > 0, /* number of blocks per track */ > 600, /* number of blocks on this device */ > 0 /* no. of blks to skip at start of device */ > ); > assert (pBlkDev); > printf ("[%s, %d] ### pBlkDev = %p\n", __FILE__, __LINE__, pBlkDev); /* Line#29 */ > > pVolDesc = dosFsMkfs (RAM_DISK_DEV, pBlkDev); > printf ("[%s, %d] ### pVolDesc = %p\n", __FILE__, __LINE__, pVolDesc); > if (! pVolDesc) printf ("DOS volume init failed\n"); > assert (pVolDesc); > > } > > || ###### C Code : END ######## > > > > > > || ====== Target Simulator Window : BEGIN ====== > > > VxWorks > > Copyright 1984-1998 Wind River Systems, Inc. > > CPU: VxSim for Windows > VxWorks: 5.4.2 > BSP version: 1.1/1 > Creation date: May 8 2002 > WDB: Ready. > > [D:/projects_MBX/demo1/01/src2/s2.c, 29] ### pBlkDev = 0x4b5cc08 > Exception ! > Vector 13 : General Protection Fault > Program Counter : 0x0040f1db > Status Register : 0x00010246 > > > || ====== Target Simulator Window : END ======== > > > > -- > ==================== > Alex Vinokur > http://up.to/alexvn > http://go.to/alexv_math > mailto:alexvn@bigfoot.com > mailto:alexvn@go.to > ==================== > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: On S_iosLib_INVALID_FILE_DESCRIPTOR Date: 8 May 2002 05:51:20 -0700 From: cozzy@263.net (yyzhong) Organization: http://groups.google.com/ Message-ID: References: <3ccd1567$1@brateggebdc5.br-automation.co.at> Another time I encountered this kind of error again . It's in the sample given by the 'Tornado getting started '. After I download the lighting.out, and start a windSH: - -> progStart value = 0 = 0x0 - -> i NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY - ---------- ------------ -------- --- ---------- -------- -------- ------- ----- tExcTask _excTask 4af9130 0 PEND 4276be 4af904c d0003 0 tLogTask _logTask 4af4718 0 PEND 4276be 4af4630 d0003 0 tWdbTask 0x417cc4 4af0bc8 3 READY 4276be 4af0aa4 d0003 0 tCosmos _cosmos 4ae5678 200 READY 4276be 4ae55dc d0003 0 tSchlep _schlep 4ade5b0 220 PEND 4276be 4ade4f4 d0003 0 tMonitor _monitor 4ad0420 230 READY 491eee0 4ad03a4 d0003 0 tCrunch _crunch 4ad74e8 240 READY 427554 4ad74a8 0 0 value = 0 = 0x0 - -> I didn't got the expected "Divide Error" after an hour! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception in dosFsMkfs() Date: Wed, 8 May 2002 16:14:37 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: References: "Leonid Rosenboim" wrote in message news:newscache$egosvg$r5n$1@lnews.actcom.co.il... | Sasha, this is a wild guess: | | Try to set "blocks per track" to a non-zero value, perhaps this causes some | zero divide error. | Also, your RAmdisk is onloy 300KBytes in size, hope that is your intention. [snip] I tried to use : 1. pBlkDev = ramDevCreate (NULL, 512, 100, 100, 0); pVolDesc = dosFsMkfs ("x2:", pBlkDev); ------------------------- Results is Exception 2. pBlkDev = ramDevCreate (NULL, 512, 400, 400, 0); pVolDesc = dosFsMkfs ("x2:", pBlkDev); ------------------------- Results is Exception 3. pBlkDev = ramDevCreate (NULL, 512, 100, 100, 0); pVolDesc = dosFsDevInit (RAM_DISK_DEV, pBlkDev, NULL); ------------------------- Results is Exception Thanks. - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: on "No rule to make" Date: 8 May 2002 06:29:54 -0700 From: cozzy@263.net (yyzhong) Organization: http://groups.google.com/ Message-ID: When I want to build a project, it prompts "make: *** No rule to make target `clean'. Stop. Done. ". I'm sure I've specified the active build. What can I do ? It mentioned the "target 'clean' ",while I didn't use it in my source! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception in dosFsMkfs() Date: Wed, 08 May 2002 13:31:33 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Ok, here's another quick idea to try: Perhaps the device name string should not reside in the code segment (which is what happens with constant strings), i.e.: char * devName[20] = "nd:" ; ... dosFsMkfs( devName, pBlkDev) Otherwise, it migth help if you know more specifically WHERE does the exception occur (stack trace, offending address). "Alex Vinokur" wrote in message news:abb88r$g9m53$1@ID-79865.news.dfncis.de... > > "Leonid Rosenboim" wrote in message news:newscache$egosvg$r5n$1@lnews.actcom.co.il... > | Sasha, this is a wild guess: > | > | Try to set "blocks per track" to a non-zero value, perhaps this causes some > | zero divide error. > | Also, your RAmdisk is onloy 300KBytes in size, hope that is your intention. > > > > [snip] > > I tried to use : > > 1. pBlkDev = ramDevCreate (NULL, 512, 100, 100, 0); > pVolDesc = dosFsMkfs ("x2:", pBlkDev); > ------------------------- > Results is Exception > > > 2. pBlkDev = ramDevCreate (NULL, 512, 400, 400, 0); > pVolDesc = dosFsMkfs ("x2:", pBlkDev); > ------------------------- > Results is Exception > > 3. pBlkDev = ramDevCreate (NULL, 512, 100, 100, 0); > pVolDesc = dosFsDevInit (RAM_DISK_DEV, pBlkDev, NULL); > ------------------------- > Results is Exception > > > > Thanks. > -- > ==================== > Alex Vinokur > http://up.to/alexvn > http://go.to/alexv_math > mailto:alexvn@bigfoot.com > mailto:alexvn@go.to > ==================== > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Multiple Kill() executions @ interrupt level. Date: 8 May 2002 06:39:04 -0700 From: danielmartinfuchs@yahoo.co.uk (Daniel Fuchs) Organization: http://groups.google.com/ Message-ID: <78cc994f.0205080539.31832cae@posting.google.com> Hello, I have a question(s) regarding using the Kill() command: I have a task that gets awoken/executed/whatever by a Kill() call @ interrupt level. * Would like to know what happens when this get awoken by receiving multiple kills'? * Do they get queued up? * Can a system lock up due to this? Thanks in advance, Daniel --------------------------- Newsgroups: comp.os.vxworks Subject: TrueFFS optional??? Date: Wed, 08 May 2002 10:08:50 -0400 From: John Sestak Organization: Raytheon Company Message-ID: <3CD93172.40621673@fallschurch.esys.com> Hi everyone, Does TrueFFS support come with VxWorks or is it an optional package that you have to buy seperately? I was able to add 'INCLUDE_TFFS' to my kernel (T2) but when I try to build, it keeps complaining that it can't find the definition of 'flIdentifyFlash()'. Thank you. John --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TrueFFS optional??? Date: 8 May 2002 14:21:18 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3CD93172.40621673@fallschurch.esys.com> John Sestak wrote: > Hi everyone, > Does TrueFFS support come with VxWorks or is it an optional package > that you have to buy seperately? I was able to add 'INCLUDE_TFFS' to my > kernel (T2) but when I try to build, it keeps complaining that it can't > find the definition of 'flIdentifyFlash()'. Thank you. It is an optional component for most architectures. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Kernel WorkQ Overflow problem. Date: Wed, 08 May 2002 14:34:11 GMT From: "doug dotson" Organization: Verio Message-ID: Greetings! I am in the process of porting a BSP to a new target. I get to the point of booting over the network and get a workQ overflow. If I disable the Ethernet interrupts then the problem goes away so it is clearly related to Ethernet interrupts. The wancom driver code is pretty much off the shelf and the ethernet controller is identical to the previous target so interrupt handling should be the same with the exception that ethernet interrupts are coming in on a different vector on this target. Anyone have any general strategies for tracking this down. Is there any way to intercept items put on the queue to determine what is causing the overflow? thanks, doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TrueFFS optional??? Date: Wed, 08 May 2002 14:34:55 GMT From: "doug dotson" Organization: Verio Message-ID: References: <3CD93172.40621673@fallschurch.esys.com> Optional package. Costs around $5K if I recall correctly. doug "John Sestak" wrote in message news:3CD93172.40621673@fallschurch.esys.com... > Hi everyone, > Does TrueFFS support come with VxWorks or is it an optional package > that you have to buy seperately? I was able to add 'INCLUDE_TFFS' to my > kernel (T2) but when I try to build, it keeps complaining that it can't > find the definition of 'flIdentifyFlash()'. Thank you. > > John > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bsp Testing Date: Wed, 08 May 2002 14:37:21 GMT From: "doug dotson" Organization: Verio Message-ID: References: <9e221d1.0205030538.4466b5b1@posting.google.com> That's how I got started. Light up some LEDs followed by an infinite loop. Slowly walked it through the code until I got to the point of getting the serial port working. doug "Rene Straub" <_remove_rene.straub@yetnet.ch> wrote in message news:ab641c$e8v$1@rex.ip-plus.net... > Hey folks, > > I had the same pleasure some days ago. I helped myself with some LEDs (LEDs > are > cool for debugging) and by writing debug data to low memory addresses. Using > a > BDM device I could read the data back after the code run (and crashed). > > If your code crashes start by placing a end-less loop in the code. If it > doesn't crash > anymore the problem must be behind the loop. Then find out why it crashed > and fix > it. Repeat this step until it works... > > Hope this helps > > - Rene > > > "doug dotson" wrote in message > news:pZwA8.18788$A%3.180255@ord-read.news.verio.net... > > Trial and error and alot of perseverence. I'm in the middle of > > debugging one now and it is slow and painful. Fortunately I have > > a bunch of LEDs I can talk to until I can get it to talk to a > > terminal. No JTAG debugger or emulator available for this > > beast either. > > > > doug > > > > "Anamika" wrote in message > > news:9e221d1.0205030538.4466b5b1@posting.google.com... > > > Hi > > > > > > I am developing a BSP for PowerPC. I have written all the drivers and > > > taken care about the memory initialization also. On burning the > > > bootrom image to the ROM, I am getting some exceptions. I want to > > > debug the code, which I hv written. Can anyone suggest the ways of > > > debugging for BSP. I dont hv a JTAG. > > > > > > Thanking you in advance > > > > > > Regards > > > Anamika > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Converting Tornado 1.0.1 BSP to Tornado 2 Date: Wed, 8 May 2002 14:51:10 +0000 (UTC) From: "Arne Henrichsen" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <7573be1638199cb11cc921b4defbfc42.8627@mygate.mailgate.org> Hi, we have a Tornado 1.0.1 BSP for a MPC860. We want to start using it in Tornado 2, and using the Tornado 1.0.1 Build compatibility in Tornado 2, the BSP compiles without errors. The problem is that it will not run on our target system. Comparing the build output from Tornado 1.0.1 with that from Tornado 2, I only find a difference right at the end of the output: (we build vxWorks_rom.hex images). It seems that the size of the images are different. I have no idea why. In T1.0.1 the linker seems to be used, while in T2 they use the compiler to link the modules. Any ideas? Thanks Arne T1.0.1 Output: ldppc -Map hss.map -X -N -X -r \ -o ctmp.o usrConfig.o \ sysALib.o sysLib.o hssBit.o hssControlIfc.o hssScc.o hssSmc.o hssProtocol.o hssUart.o hssHdlc.o hssBisync.o version.o D:\Tornado\target/lib/libPPC860gnuvx.a nmppc ctmp.o | munch > ctdt.c make.exe CC_COMPILER="-traditional" ctdt.o make.exe[1]: Entering directory `D:\Tornado\target\config\HSS4' ccppc -BD:\Tornado/host/x86-win32/lib/gcc-lib/ -mcpu=860 -traditional - -O2 -fvolatile -fno- builtin -fno-for-scope -Wall -I/h -I\\noddy\prj\hss4\src\h -I. - -ID:\Tornado\target\confi g\all -ID:\Tornado\target/h -ID:\Tornado\target/src/config - -ID:\Tornado\target/src/drv -DC PU=PPC860 -DHSS_PCB_CCII_SIO_PMC_4P_BP_CC -DHSS_EI_RS485 - -DHSS_CFG_4PORT -O0 -g3 -c ctd t.c make.exe[1]: Leaving directory `D:\Tornado\target\config\HSS4' ldppc -Map hss.map -X -N -e _romInit -Ttext 00008000 \ -o vxWorks_rom romInit.o bootInit_uncmp.o dataSegPad.o \ ctmp.o ctdt.o D:\Tornado\host\x86-win32\bin\romsize PPC -b 000F0000 vxWorks_rom vxWorks_rom: 195832(t) + 13292(d) = 209124 (773916 unused) D:\Tornado\host\x86-win32\bin\elfHex -a 100100 vxWorks_rom \ > vxWorks_rom.hex Done. T2 Output: ccppc -BD:\Tornado2\host\x86-win32\lib\gcc-lib\ -nostdlib -r -Wl,-X \ -o ctmp.o usrConfig.o \ sysALib.o sysLib.o hssBit.o hssControlIfc.o hssScc.o hssSmc.o hssProtocol.o hssUart.o hssHdlc.o hssBisync.o version.o D:\Tornado2\target\lib\libPPC860gnuvx.a nmppc ctmp.o | wtxtcl D:\Tornado2\host\src\hutils\munch.tcl -asm ppc > ctdt.c make.exe CC_COMPILER="-fdollars-in-identifiers" ctdt.o make.exe[1]: Entering directory `D:\Tornado2\target\config\hss4' ccppc -BD:\Tornado2\host\x86-win32\lib\gcc-lib\ -mcpu=860 - -fdollars-in-identifiers -O2 -fv olatile -fno-builtin -fno-for-scope -Wall -I\h -I\\noddy\prj\hss4\src\h - -I. -ID:\Tornado 2\target\config\all -ID:\Tornado2\target\h - -ID:\Tornado2\target\src\config -ID:\Tornado2\t arget\src\drv -DCPU=PPC860 -DHSS_PCB_CCII_SIO_PMC_4P_BP_CC - -DHSS_EI_RS485 -DHSS_CFG_4PORT -O0 -g3 -c ctdt.c make.exe[1]: Leaving directory `D:\Tornado2\target\config\hss4' ldppc -Map hss.map -X -N -e _romInit -Ttext 00008000 \ -o vxWorks_rom romInit.o bootInit_uncmp.o dataSegPad.o \ ctmp.o ctdt.o D:\Tornado2\host\x86-win32\bin\romsize PPC -b 000F0000 vxWorks_rom vxWorks_rom: 263436(t) + 15780(d) = 279216 (703824 unused) D:\Tornado2\host\x86-win32\bin\elfHex -a 100100 vxWorks_rom \ > vxWorks_rom.hex Done. - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Kernel WorkQ Overflow problem. Date: Wed, 08 May 2002 14:59:54 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Briefly: the driver ISR fails to deassert the interrupt request, hence the infinite interrupt fenomena. It could have worked on another board if interrupts were set to be edge trigerred there, but yours are evidently level trigerred. I've explained this many times before, so search google groups, and save me the typing. "doug dotson" wrote in message news:DHaC8.19174$A%3.187398@ord-read.news.verio.net... > Greetings! > > I am in the process of porting a BSP to a new target. I get to the > point of booting over the network and get a workQ overflow. If > I disable the Ethernet interrupts then the problem goes away so > it is clearly related to Ethernet interrupts. The wancom driver code > is pretty much off the shelf and the ethernet controller is identical to > the previous target so interrupt handling should be the same with the > exception that ethernet interrupts are coming in on a different vector > on this target. Anyone have any general strategies for tracking this > down. Is there any way to intercept items put on the queue to determine > what is causing the overflow? > > thanks, > doug > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Kernel WorkQ Overflow problem. Date: 8 May 2002 15:05:38 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "doug dotson" wrote: > Greetings! > > I am in the process of porting a BSP to a new target. I get to the > point of booting over the network and get a workQ overflow. If > I disable the Ethernet interrupts then the problem goes away so > it is clearly related to Ethernet interrupts. The wancom driver code > is pretty much off the shelf and the ethernet controller is identical > to the previous target so interrupt handling should be the same with > the exception that ethernet interrupts are coming in on a different > vector on this target. Anyone have any general strategies for tracking > this down. Is there any way to intercept items put on the queue to > determine what is causing the overflow? Looks like you do not clear the interrupt source, so after you finisch the interrupt routine it jumps right back in it again, because the interrupt is still active. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Kernel WorkQ Overflow problem. Date: Wed, 08 May 2002 15:11:46 GMT From: "doug dotson" Organization: Verio Message-ID: References: This is what I suspected but so far have been unable to locate where it happens. I'll check the old target to see if it used edge triggered interrupts. I am definitely using level-sensitive now. doug "Johan Borkhuis" wrote in message news:Xns9208ADE8C5661borkhuisagerecom@135.7.153.102... > "doug dotson" wrote: > > > Greetings! > > > > I am in the process of porting a BSP to a new target. I get to the > > point of booting over the network and get a workQ overflow. If > > I disable the Ethernet interrupts then the problem goes away so > > it is clearly related to Ethernet interrupts. The wancom driver code > > is pretty much off the shelf and the ethernet controller is identical > > to the previous target so interrupt handling should be the same with > > the exception that ethernet interrupts are coming in on a different > > vector on this target. Anyone have any general strategies for tracking > > this down. Is there any way to intercept items put on the queue to > > determine what is causing the overflow? > > Looks like you do not clear the interrupt source, so after you finisch > the interrupt routine it jumps right back in it again, because the > interrupt is still active. > > Groeten, > Johan > > -- > o o o o o o o . . . _____________________________ > o _____ || Johan Borkhuis | > .][__n_n_|DD[ ====_____ | borkhuis@agere.com | > >(________|__|_[_________]_|__________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` > === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone using target Intel x86 (not PENTIUM) Date: Wed, 08 May 2002 15:10:26 GMT From: igtorque@yahoo.es (Ignacio Gonzalez T.) Organization: Newshosting.com - Highest quality at a great price! www.newshosting.com Message-ID: <3cd93f65.4871064@news.mhogaming.com> References: <97e6274e.0205080100.3e7e1d13@posting.google.com> On 8 May 2002 02:00:18 -0700, heiko_elger@arburg.com (Heiko Elger) wrote: >Hello, > >who is using a target Intel x86 (386, 486, not Pentium, PcPentium Pro) >with Tornado 2.0.x? > >I'm asking cause in the new Tornado Version 2.2 (VxWorks 5.5) only >Intel processors Pentium, Pentium II, Pentium III, Pentium IV will be >supported. >(the new official roadmap of WRS) > >For customer with support contract who uses no Pentium or above there >will be no ugrade posibbility and only minor bug fixes for Tornado >2.0.x > >I'm asking cause we have support contract and we are using x86 and we >have no possibility to use a Pentium next time. > >We think that cause of our support contract based on a Intel 80486 >processor WRS must support x86 in Tornado 2.2. > >We want to know whether there are other projects out there which are >based on Intel x86 or whether we are the only company using x86. > >best regards > > Heiko Elger Hello, Heiko. We are using Tornado 2.0.2 and VxWorks 5.4.2 with 386EX targets since 2000. After 1 year with support contract, we decided to leave it, because of the percieved low quality/cost ratio (of course, I agree this is a very subjective appreciation.) We will continue to use these targets (and perhaps another ones, mainly 486DX based) instead of Pentium ones, because of cost considerations (we manufacture our own boards, or use cheap standard ones.) So we will have to stick to Tornado 2.0.2, even if someday we decide to upgrade... My condolence. These news are bad news for us too. By the way, do you know if Tornado 2.2 comes with gcc > 2.7.2 ? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Date: Wed, 08 May 2002 16:50:43 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Since it dont seem like you got an answer, I'll offer my two cents: Interrupt is essentially a hardware mechanism, i.e. all that happens from the time the PCI board raises its IRQ, till the ISR runs, is mostly for the hardware to do, except of course the VxWorks ISR wrapper which should be less the 1 microsec on a Pentium. Unless of course, someone or something has instructed the CPU to block interrupts. (Note: on x86 processors,l the CPU can either block all or allow all interrupts as it has got only one INTR pin). What I am saying is that you need to find that bugger ;-) who put tons of code between intLock() and intUnlock() calls, code that runs for 5-15 milisec. intLock gets used in the kernel a few times, but trust me, the VxWorks kernel or any other subsystem do not do anything THAT STUPID. And when you find that guy, take his picture when he gets his pink slip, you'll enjoy looking at it for years later ... ;-)) "Peter Kannegiesser" wrote in message news:aer2dug9ot75urttc2dabeg507f5v44p7t@4ax.com... > Hi all! > > We are using Tornado 2.02 and the related tools and have written a > driver for an D/A, A/D converter PCI card. The driver is implemented > using ioctl() calls. The PCI card generates interrupts at the end of > conversion. > The interrupt handler is connected using pciConnectInt and is > definitly the only one for that interrupt line. > Now we experience very looong delays from the interrupt signal on > the PCI bus until the handler gets called. According to the logic > analyzer, it takes between 5 and 15 milliseconds (with a strong > tendency to 15ms)-- and this supposed to be an OS for fast hard > realtime... > We are using the standard Pentium BSP on various PC's, from 166MHz > Pentium I to 2GHz Pentium 4. There seems to be no change that > depends on the processor speed. > > Has anybody seen similar problems or could offer at least an > explanation? > > TIA, > Peter > -- > keine sig --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I see the 'target.nr' file? Date: Wed, 08 May 2002 11:57:51 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3CD9590F.EF7351BE@fnal.gov> References: <3CD88C30.4E881288@gv.net> > > Jin-Woo Lee wrote: > > > >>Hi, > >> > >>I'm working on mv2400 target board. > >>There is 'target.nr' file in the BSP directory. What is the way to see > >>it? > >> I assume you are on a unix host. % cd $WIND_BASE/target/config/mv2400 % man ./target.nr I do it on linux and see the file formatted like a man page HTH Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Creating a bootable VxWorks image Date: Wed, 8 May 2002 14:22:30 -0400 From: "Jin-Woo Lee" Organization: Cornell University Message-ID: Sender: jl206@cornell.invalid (on 128.253.249.182) Hi, I'm a new user of VxWorks. I'm currently working with Tornado 2.0.2 under Windows environment and the target system is mv2400. When I tried to create a bootable VxWorks image, the following error showed up during building the project. C:\Temp\cca01436.s: Assembler messages: C:\Temp\cca01436.s:4687: Error: Unrecognized opcode: `func_export(_sysInit)' C:\Temp\cca01436.s:4688: Error: Unrecognized opcode: `func_export(sysInByte)' C:\Temp\cca01436.s:4689: Error: Unrecognized opcode: `func_export(sysOutByte)' . . . I just used the application wizard. I chosen an option of 'Create a bootable VxWorks image' at the first page, selected mv2400 at BSP page and then I didn't touch anything. What is wrong with that? Thanks, --------------------------- Newsgroups: comp.protocols.snmp,comp.os.vxworks Subject: Re: Envoy version in WindNet SNMP Date: Wed, 08 May 2002 18:45:13 GMT From: Pete Flugstad Organization: MediaCom High Speed Internet Message-ID: <3CD97299.6070004@nospam.icon-labs.com> References: <3CD83CFD.1060203@nospam.icon-labs.com> Alex Vinokur wrote: > | The file envoy/h/version.h should have the Envoy version in it: > | > | #define ENVOY_MAJOR_VERSION 9 > | #define ENVOY_MINOR_VERSION 1 > | #define ENVOY_BETA_VERSION 0 > | > | I would expect it to be the same as the provided emissary > | version (i.e. 7.0.0). > | > | Pete Flugstad > | Icon Labs > | > > 1. The envoy directory is missing under $WIND_BASE and its subdirectories. > > 2. The $WIND_BASE/target/h/snmp/envoy.h file contains no relevant information. > The $WIND_BASE/target/h/snmp/install.h file contains the following line : > #define INSTALL_ENVOY 1 > > 3. No file contains ENVOY_MAJOR_VERSION, ENVOY_MINOR_VERSION, ENVOY_BETA_VERSION And I don't see those #def's being used in the sources anywhere, at least in the versions I have (9.X), so there seems to be nw way to check for that.... Given that, you probably have to assume it is the same as the version of Emissary (mibcomp). IMO, this is probably a fairly safe assumption, since mibcomp generates code for Envoy is so is tied fairly closely to Envoy's data structures. Sorry, Pete --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TrueFFS optional??? Date: Wed, 08 May 2002 15:30:43 -0400 From: John Sestak Organization: Raytheon Company Message-ID: <3CD97CE2.49A61D7E@fallschurch.esys.com> References: <3CD93172.40621673@fallschurch.esys.com> Thank you both for replying. :) Johan Borkhuis wrote: > John Sestak wrote: > > > Hi everyone, > > Does TrueFFS support come with VxWorks or is it an optional package > > that you have to buy seperately? I was able to add 'INCLUDE_TFFS' to my > > kernel (T2) but when I try to build, it keeps complaining that it can't > > find the definition of 'flIdentifyFlash()'. Thank you. > > It is an optional component for most architectures. > > Groeten, > Johan > > -- > o o o o o o o . . . _____________________________ > o _____ || Johan Borkhuis | > .][__n_n_|DD[ ====_____ | borkhuis@agere.com | > >(________|__|_[_________]_|__________________________| > _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` > === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: VxMP and POSIX? Date: 8 May 2002 13:48:30 -0700 From: rwitmer@my-deja.com (Bob Witmer) Organization: http://groups.google.com/ Message-ID: Anyone know if VxMP supports POSIX message queues? I can't tell from reading the sales literature. My guess would be that they implemented only the vendor specific APIs and not the POSIX stuff. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: on "No rule to make" Date: Wed, 8 May 2002 22:56:51 +0100 From: "Michael B." Organization: [Infostrada] Message-ID: <6dgC8.30318$US3.667633@twister1.libero.it> References: Pay much attention to dependancies: you have to update them every time you make some changes to your code (#include and so on) or your built configuration. This is my experience: sometimes (without a good reason!) the builder loses one of my .h files and update dependancies made one or two times fixes the bug. Magic? I don't know but it works. Bye, Mic. "yyzhong" ha scritto nel messaggio news:b35585ab.0205080529.7f5a60a6@posting.google.com... > When I want to build a project, it prompts > "make: *** No rule to make target `clean'. Stop. > Done. > ". > I'm sure I've specified the active build. > What can I do ? > It mentioned the "target 'clean' ",while I didn't use it in my source! --------------------------- Newsgroups: comp.os.vxworks Subject: Code for unresolved symbols not updated Date: 8 May 2002 15:44:28 -0700 From: chris.ouellette@santera.com (ChrisO) Organization: http://groups.google.com/ Message-ID: <72ddcdb.0205081444.107b6999@posting.google.com> I'm loading a COFF file (testB.o) that has function (funcB) that calls funcA. When I load it "ld < testB.o" I get an error message about the unresolved function funcA. If I load a COFF file (testA.o) that has funcA then load testB.o I don't get the error message but looking at the assemble code for funcB shows that the address for funcA was not updated. It still has 0 for the address of funcA. VxWorks is running on a 960 CA. I'm using gcc960 and gld960 not the VxWorks versions. Are the VxWorks versions of the compiler and linker required? Seems to me a COFF file is a COFF file. Any ideas why the addresses aren't updated? Chris --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Code for unresolved symbols not updated Date: 08 May 2002 21:06:21 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <72ddcdb.0205081444.107b6999@posting.google.com> Sender: bpringlemeir@DeadDuck ChrisO> error message about the unresolved function funcA. If I load ChrisO> a COFF file (testA.o) that has funcA then load testB.o I ChrisO> don't get the error message but looking at the assemble code ChrisO> for funcB shows that the address for funcA was not updated. ChrisO> It still has 0 for the address of funcA. VxWorks is running ChrisO> on a 960 CA. I'm using gcc960 and gld960 not the VxWorks ChrisO> versions. Are the VxWorks versions of the compiler and ChrisO> linker required? Seems to me a COFF file is a COFF file. ChrisO> Any ideas why the addresses aren't updated? Well, the ARM COFF format with T2.1.1 are different than T2.0.1. However, I would suspect that you might have some cache coherency problem. Does the function not work properly if you run it? What does "objdump960 --all-headers testB.o" show for the relocation and symbol records for the functions in question. Likewise for the testA.o file. This information would be most helpful. Also, run `@lkup "funcA"' and `@lkup "funcB"' from the shell. Also try `lkup "funcA"' and `lkup "funcB"'. This will give some clue as to a possible host/target symbol table mismatch. A symbol entry will look like, [ 25](sec 0)(fl 0x00)(ty 20)(scl 2) (nx 0) 0x00000000 _funcA [ 26](sec 1)(fl 0x00)(ty 20)(scl 2) (nx 0) 0x00000000 _funcB .. RELOCATION RECORDS FOR [.text]: OFFSET TYPE VALUE 00000038 DISP32 _funcB etc. Just post the values for the functions in question [the above are ELF, but I think you can get the idea.] Regards, Bill Pringlemeir. - -- Let's stay! vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Is it possible to build a simulator myself? Date: 8 May 2002 20:09:21 -0700 From: cozzy@263.net (yyzhong) Organization: http://groups.google.com/ Message-ID: I need to use a simulator oriented to a project based on PentiumGNU. But I don't know how to build it. Is it possible? And which disk is necessory? The disks I can use are listed below: CD Title: Tornado 2.0 AP for x86 Part No.: TDK-13127-ZC CD Title: Tornado 2.0 AP for PowerPC Part No.: TDK-13128-ZC CD Title: Tornado 2.0.2 for PowerPC Part No.: TDK-12389-ZC CD Title: Tornado 2.0.2 for All Simulators Part No.: TDK-12840-ZC CD Title: BSP Developer's Kit Part No.: TDK-13234-ZC CD Title: RTI SCOPE TOOLS 2.1 Part No.: STS-14469-zc-00 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TrueFFS optional??? Date: Thu, 09 May 2002 06:47:38 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3CD93172.40621673@fallschurch.esys.com> All previous replies are true, except that I recall that support for M-Systems Disk-on-chip (aka. DOC2K) devices was bundled on x86 and PPC architectures, whereas the unbundled TFFS allows one to use any Flash hardware they fancy, at least in theory. "John Sestak" wrote in message news:3CD93172.40621673@fallschurch.esys.com... > Hi everyone, > Does TrueFFS support come with VxWorks or is it an optional package > that you have to buy seperately? I was able to add 'INCLUDE_TFFS' to my > kernel (T2) but when I try to build, it keeps complaining that it can't > find the definition of 'flIdentifyFlash()'. Thank you. > > John > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Exception in dosFsMkfs() Date: Thu, 9 May 2002 10:20:54 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: References: "Alex Vinokur" wrote in message news:abb4il$gn9jc$1@ID-79865.news.dfncis.de... | ========================================================= | Tornado : Version 2.0.2 | VxWorks : Version 5.4.2 | BSP : Version 1.1/1 | Compiler : ccsimpc | gcc version egcs-2.90.29 tornado 2.0 980515 | (egcs-1.0.3 release) | Host : Windows2000 | Target : VxSim for Windows (VxWorks target simulator) | ========================================================= | Project Type : bootable VxWorks image (custom configured) | ========================================================= | | | I have got an exception after calling the dosFsMkfs() function. | Something is wrong. What exactly ? | | Thanks in advance. | | || ###### C Code : BEGIN ###### | // File s2.c | #include | #include | | #include | #include | #include | | | # define RAM_DISK_DEV "rd:" | | static DOS_VOL_DESC *pVolDesc = NULL ; | | void usrAppInit (void) | { | BLK_DEV *pBlkDev = NULL ; | | #ifdef USER_APPL_INIT | USER_APPL_INIT; /* for backwards compatibility */ | #endif | | pBlkDev = ramDevCreate ( | NULL, /* where it is in memory (0 = malloc) */ | 512, /* number of bytes per block */ | 0, /* number of blocks per track */ | 600, /* number of blocks on this device */ | 0 /* no. of blks to skip at start of device */ | ); | assert (pBlkDev); | printf ("[%s, %d] ### pBlkDev = %p\n", __FILE__, __LINE__, pBlkDev); /* Line#29 */ | | pVolDesc = dosFsMkfs (RAM_DISK_DEV, pBlkDev); | printf ("[%s, %d] ### pVolDesc = %p\n", __FILE__, __LINE__, pVolDesc); | if (! pVolDesc) printf ("DOS volume init failed\n"); | assert (pVolDesc); | | } | | || ###### C Code : END ######## | | | | | | || ====== Target Simulator Window : BEGIN ====== | | | VxWorks | | Copyright 1984-1998 Wind River Systems, Inc. | | CPU: VxSim for Windows | VxWorks: 5.4.2 | BSP version: 1.1/1 | Creation date: May 8 2002 | WDB: Ready. | | [D:/projects_MBX/demo1/01/src2/s2.c, 29] ### pBlkDev = 0x4b5cc08 | Exception ! | Vector 13 : General Protection Fault | Program Counter : 0x0040f1db | Status Register : 0x00010246 | | | || ====== Target Simulator Window : END ======== | | | | The exception was caused by undefined macro INCLUDE_DOSFS. || ################################################## || ############ VxWorks Reference Manual ############ || ################### Libraries #################### || ### File $WIND_BASE\docs\vxworks\ref\dosFsLib.html || ### ----------------- Fragments ------------------ || ################################################## || ================ dosFsLib : BEGIN ================ NAME dosFsLib - MS-DOS media-compatible file system library ROUTINES dosFsMkfs( ) - initialize a device and create a dosFs file system INITIALIZING DOSFSLIB This initialization is enabled when the configuration macro INCLUDE_DOSFS is defined; dosFsInit( ) is then called from the root task, usrRoot( ), in usrConfig.c. || ================ dosFsLib : END ================== So, the configuration macro INCLUDE_DOSFS must be defined. In order to that we need : || ============ Workspace Window : BEGIN ============ || ------------------ VxWorks view ------------------ + [operating system components] + [IO system components] * DOS filesytem // Must be included || ============ Workspace Window : END ============== After starting 'vxWorks.exe" : || ====== Target Simulator Window : BEGIN ====== Adding 2411 symbols for standalone. VxWorks Copyright 1984-1998 Wind River Systems, Inc. CPU: VxSim for Windows VxWorks: 5.4.2 BSP version: 1.1/1 Creation date: May 9 2002 WDB: Ready. [D:/projects_MBX/demo1/01/src2/s2.c, 29] ### pBlkDev = 0x4b7b7c0 [D:/projects_MBX/demo1/01/src2/s2.c, 33] ### pVolDesc = 0x4b90d70 || ====== Target Simulator Window : END ======== Now it is OK. - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Advice on TCP/UDP task arrangement Date: 9 May 2002 00:54:56 -0700 From: barryrichard@hotmail.com (RichardBarry) Organization: http://groups.google.com/ Message-ID: Hi, I have a general question on using UDP from within a task, any advice would be gratefully received. I have been experimenting with a system where I create a UDP socket then spawn one task that reads from the socket and one that writes to the socket. The writing task blocks on a message queue, and when a message arrives formats it and transmits it. The receiving task blocks on a call to recvfrom(), and when a message arrives, formats it and posts it onto a different message queue. This seems to work, however is this a valid thing to do? i.e. is mutual exclusion around the socket guaranteed by the OS, or must I provide it explicitly? What would happen in the Tx task deleted the socket while the Rx task was blocked on recvfrom()? Regards, Richard. --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: cross compiling gcc - parse error Date: 9 May 2002 01:11:15 -0700 From: demiurg@tau.ac.il (Alexander Sirotkin) Organization: http://groups.google.com/ Message-ID: I'm trying to compile gcc & bintutils for vxworks on Linux, binutils are compiled just fine, but gcc throuws the following error: /root/gnu/gcc-2.95/gcc/config/mips/vxworks.h:3: parse error before `file' Apparently, it does not like embedded comments vxworks.h:3 /* Copyright (C) 1999 Free Software Foundation, Inc. */ This file is part of GNU CC. --------------------------- Newsgroups: comp.os.vxworks Subject: Compilation log file Date: Thu, 9 May 2002 13:08:29 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: ========================================================= Tornado : Version 2.0.2 VxWorks : Version 5.4.2 BSP : Version 1.1/1 Compiler : ccsimpc gcc version egcs-2.90.29 tornado 2.0 980515 (egcs-1.0.3 release) Host : Windows2000 Target : VxSim for Windows (VxWorks target simulator) ========================================================= I would like the compilation results to be written not only to Build Output Window, but to some compilation log file as well. Is it possible to do it? - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bsp Testing Date: Thu, 9 May 2002 10:37:54 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CDA51DD.4030403@spamm.me.l8s.co.uk> References: <9e221d1.0205030538.4466b5b1@posting.google.com> Hwa Jin Bae wrote: > I have used various available mechanisms to debug low level BSP issues. For > example, LEDs, PromICE, BDM, JTAG, Logic analyser, scope, Rom emulator, > serial port, shared memory (over bus), etc. And of course, a lot of > patience and tenacity help. Finally, some good luck. If your board has a UART, enable it very early on (eg in romInit.s) and write routines to dump single characters, strings and hex numbers to it (looping until the fifo is empty). This ought to let you debug system startup.... David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Advice on TCP/UDP task arrangement Date: Thu, 09 May 2002 11:43:45 +0100 From: Phil Ashby Organization: IDL Message-ID: <3CDA52E1.F1B752C4@idl.bt.co.uk> References: RichardBarry wrote: > > Hi, > > I have a general question on using UDP from within a task, any advice > would be gratefully received. > > I have been experimenting with a system where I create a UDP socket > then spawn one task that reads from the socket and one that writes to > the socket. The writing task blocks on a message queue, and when a > message arrives formats it and transmits it. The receiving task > blocks on a call to recvfrom(), and when a message arrives, formats it > and posts it onto a different message queue. > > This seems to work, however is this a valid thing to do? i.e. is > mutual exclusion around the socket guaranteed by the OS, or must I > provide it explicitly? What would happen in the Tx task deleted the > socket while the Rx task was blocked on recvfrom()? > > Regards, > Richard. AFAIK there shouldn't be any problems doing this. I've used similar code with TCP sockets (where Tx and Rx state interact) without problems. Since a socket must be some form of waitable object (how does recvfrom() actually block?), then closing it releases waiting tasks, returning an error code to their caller. All seems to work reliably for me ;) - -- Phil "Phlash" Ashby (speaking only for himself) Work: http://www.internet-designers.net/ Home: http://www.ashbysoft.com/ --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: Re: cross compiling gcc - parse error Date: Thu, 09 May 2002 11:46:53 +0100 From: Phil Ashby Organization: IDL Message-ID: <3CDA539D.F483A489@idl.bt.co.uk> References: Alexander Sirotkin wrote: > > I'm trying to compile gcc & bintutils for vxworks on Linux, > binutils are compiled just fine, but gcc throuws the following error: > > /root/gnu/gcc-2.95/gcc/config/mips/vxworks.h:3: parse error before `file' > > Apparently, it does not like embedded comments > > vxworks.h:3 > /* Copyright (C) 1999 Free Software Foundation, Inc. */ > This file is part of GNU CC. That looks wrong. Obviously the text 'This file...' should be in a coment block, but isn't. Why should a cross-compiler need to include a BSP-specific vxworks.h anyway? Surely you just need to create a cross-compiler for the target CPU, not the O/S? - -- Phil "Phlash" Ashby (speaking only for himself) Work: http://www.internet-designers.net/ Home: http://www.ashbysoft.com/ --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri May 10 04:03:19 2002 From: Vxworks Exploder Date: Fri May 10 04:03:22 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri May 10 04:03:12 PDT 2002 Subject: Re: Advice on TCP/UDP task arrangement Subject: Compilation for MBX860 target failed Subject: Re: Compilation for MBX860 target failed Subject: Two GD82559ER adapters under VxWorks Subject: How to define a Tornado startup script? Subject: Re: Advice on TCP/UDP task arrangement Subject: Inside ISR check? Subject: Non blocking UDP sockets? Subject: Re: Advice on TCP/UDP task arrangement Subject: Re: Two GD82559ER adapters under VxWorks Subject: Re: cross compiling gcc - parse error Subject: Re: Inside ISR check? Subject: Re: Advice on TCP/UDP task arrangement Subject: Re: Two GD82559ER adapters under VxWorks Subject: Re: Two GD82559ER adapters under VxWorks Subject: Re: Inside ISR check? Subject: Re: Inside ISR check? Subject: Memory data integrity preservation Subject: Non-Blocking Socket is Blocking Subject: C++ comments in C file with GNU compiler Subject: We're havin' some fun now! Subject: ENVOY SNMP Subject: Accessing hard disk Subject: Re: Compilation for MBX860 target failed Subject: Re: Memory data integrity preservation Subject: Re: C++ comments in C file with GNU compiler Subject: problem launching the target server on host Subject: Re: Non-Blocking Socket is Blocking Subject: Re: Non blocking UDP sockets? Subject: Re: Non blocking UDP sockets? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Advice on TCP/UDP task arrangement Date: Thu, 09 May 2002 12:07:59 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: First the easy questions: VxWorks will take care of mutex on sockets, so in case of a UDP socket, multiple tasks could be reading from the socket (dividing the incoming pakcets between them) and multiple tasks can send packets out, if output order has no particular significance. If a socket is closed while some task is blocked on it, or if a task is using a previously closed socket, the call will return an error, probably "Invalid file descriptor". Now to the architecture of yours: There is one reason to use more tasks - to prioritize work. There is a cost to pay for more tasks - more memory going into stacks and more difficult to debug. My feeling is that you can probably do all this without message queues and with a single task doing reception, processing and transmission, but of course this is hard to tell without knowing your specific application requirements. "RichardBarry" wrote in message news:eec86974.0205082354.23f663cd@posting.google.com... > Hi, > > I have a general question on using UDP from within a task, any advice > would be gratefully received. > > I have been experimenting with a system where I create a UDP socket > then spawn one task that reads from the socket and one that writes to > the socket. The writing task blocks on a message queue, and when a > message arrives formats it and transmits it. The receiving task > blocks on a call to recvfrom(), and when a message arrives, formats it > and posts it onto a different message queue. > > This seems to work, however is this a valid thing to do? i.e. is > mutual exclusion around the socket guaranteed by the OS, or must I > provide it explicitly? What would happen in the Tx task deleted the > socket while the Rx task was blocked on recvfrom()? > > Regards, > Richard. --------------------------- Newsgroups: comp.os.vxworks Subject: Compilation for MBX860 target failed Date: Thu, 9 May 2002 15:14:33 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: ========================================================= Tornado : Version 2.0.2 VxWorks : Version 5.4.2 BSP : Version 1.1/1 Compiler : ccppc gcc driver version cygnus-2.7.2-960126 egcs-971225 tornado 2.0 executing gcc version cygnus-2.7.2-960126 Linker : ldppc ld version cygnus-2.6 (with BFD cygnus-2.6) Host : Windows2000 Target : MBX860 ToolChain : PPC860gnu ========================================================= Project Type : bootable VxWorks image (custom configured) RAM based VxWorks image, linked to RAM_LOW_ADRS. It is loaded into RAM via some external program such as a bootROM. This is the default development image. ========================================================= I have got the following results of compilations for MBX860 target. I changed no files. What is wrong? Relevant information is below. Thanks. || ######### Compilation Log File : BEGIN ######### vxrm *.o *.rpo ctdt.c symTbl.c vxWorks* vxrm D:\sys2\Tornado\target\proj\dir-mbx-proj-01\prjComps.h D:\sys2\Tornado\target\proj\di r-mbx-proj-01\prjParams.h D:\sys2\Tornado\target\proj\dir-mbx-proj-01\prjConfig.c D:\sys2\ Tornado\target\proj\dir-mbx-proj-01\linkSyms.c vxrm D:\sys2\Tornado\target\proj\dir-mbx-proj-01\libs.nm D:\sys2\Tornado\target\proj\dir-m bx-proj-01\libs.size wtxtcl D:\sys2\Tornado\host\src\hutils\configGen.tcl D:\sys2\Tornado\target\proj\dir-mbx-p roj-01\MBX-Project-01.wpj ccppc -g -mcpu=860 -ansi -nostdinc -fvolatile -fno-builtin -fno-for-scope -P -x assembler - -with-cpp -ID:\sys2\Tornado\target\proj\dir-mbx-proj-01 -ID:\sys2\Tornado\target\config\mb x860 -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\comps\src -ID:\sys2\Torn ado\target\src\drv -DCPU=PPC860 -DPRJ_BUILD -c D:\sys2\Tornado\target\config\mbx860\sysA Lib.s -o sysALib.o ccppc -g -mcpu=860 -ansi -nostdinc -DRW_MULTI_THREAD -D_REENTRANT -fvolatile -fno-builtin -fno-for-scope -ID:\sys2\Tornado\target\proj\dir-mbx-proj-01 -ID:\sys2\Tornado\target\con fig\mbx860 -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\comps\src -ID:\sys 2\Tornado\target\src\drv -DCPU=PPC860 -DPRJ_BUILD -c D:\sys2\Tornado\target\config\mbx86 0\sysLib.c In file included from D:\sys2\Tornado\target\config\mbx860\config.h:349, from D:\sys2\Tornado\target\config\mbx860\sysLib.c:54: D:\sys2\Tornado\target\config\mbx860\mbx800.h:417: conflicting types for `sysInLong' D:\sys2\Tornado\target\h\sysLib.h:220: previous declaration of `sysInLong' D:\sys2\Tornado\target\config\mbx860\mbx800.h:429: conflicting types for `sysOutLong' D:\sys2\Tornado\target\h\sysLib.h:225: previous declaration of `sysOutLong' In file included from D:\sys2\Tornado\target\config\mbx860\sysLib.c:299: D:\sys2\Tornado\target\config\mbx860\w83c553f.c:130: conflicting types for `sysOutByte' D:\sys2\Tornado\target\h\sysLib.h:223: previous declaration of `sysOutByte' D:\sys2\Tornado\target\config\mbx860\w83c553f.c:131: conflicting types for `sysInByte' D:\sys2\Tornado\target\h\sysLib.h:218: previous declaration of `sysInByte' D:\sys2\Tornado\target\config\mbx860\sysLib.c:1307: conflicting types for `sysIntEnablePIC ' D:\sys2\Tornado\target\h\sysLib.h:237: previous declaration of `sysIntEnablePIC' D:\sys2\Tornado\target\config\mbx860\sysLib.c:1607: conflicting types for `sysInWordString ' D:\sys2\Tornado\target\h\sysLib.h:221: previous declaration of `sysInWordString' D:\sys2\Tornado\target\config\mbx860\sysLib.c:1651: conflicting types for `sysOutWordStrin g' D:\sys2\Tornado\target\h\sysLib.h:226: previous declaration of `sysOutWordString' D:\sys2\Tornado\target\config\mbx860\sysLib.c:1673: conflicting types for `sysInLongString ' D:\sys2\Tornado\target\h\sysLib.h:222: previous declaration of `sysInLongString' D:\sys2\Tornado\target\config\mbx860\sysLib.c:1695: conflicting types for `sysOutLongStrin g' D:\sys2\Tornado\target\h\sysLib.h:227: previous declaration of `sysOutLongString' make: *** [sysLib.o] Error 0x1 Done. || ######### Compilation Log File : END ########### Note. sysLib.o: depends on D:\sys2\Tornado\target\h\types\vxTypesOld.h (and another header files). || ######### ========================== ########### || File#1 of 3 || ####### File vxTypesOld.h : BEGIN ####### || Full path : D:\sys2\Tornado\target\h\types\vxTypesOld.h /* vxTypesOld.h - old VxWorks type definition header */ /* Copyright 1984-1997 Wind River Systems, Inc. */ [--- omitted ---] typedef unsigned int UINT32; || Line#84 [--- omitted ---] || ####### File vxTypesOld.h : END ######### || File#2 of 3 || ######### File sysLib.h : BEGIN ######### || Full path : D:\sys2\Tornado\target\h\sysLib.h /* sysLib.h - system dependent routines header */ /* Copyright 1984-1998 Wind River Systems, Inc. */ [--- omitted ---] #if (CPU_FAMILY == I80X86) /* * These prototypes are for the x86 family only. They are incorrect, * but we do not want to change the API for the entire ARCH_FAMILY. * They should be unsigned arguments. */ [--- omitted ---] extern ULONG sysInLong (int port); [--- omitted ---] #else /* For all architectures, except X86, the arguments are unsigned. */ [--- omitted ---] extern ULONG sysInLong (ULONG port); || Line#220 [--- omitted ---] #endif /* (CPU_FAMILY == I80X86) */ [--- omitted ---] || ######### File sysLib.h : END ########### || File#3 of 3 || ######### File mbx800.h : BEGIN ######### || Full path : D:\sys2\Tornado\target\config\mbx860\mbx800.h /* mbx800.h - Motorola MBX800 board header */ /* Copyright 1984-1998 Wind River Systems, Inc. */ /* Copyright 1997,1998 Motorola, Inc., All Rights Reserved */ [--- omitted ---] #ifndef _ASMLANGUAGE /* PCI I/O function defines */ [--- omitted ---] #ifndef PCI_IN_LONG #define PCI_IN_LONG(x) sysInLong (x) #endif IMPORT UINT32 sysInLong (UINT32 address); || Line#417 [--- omitted ---] #endif /* _ASMLANGUAGE */ [--- omitted ---] || ######### File mbx800.h : END ########### ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation for MBX860 target failed Date: Thu, 9 May 2002 15:44:52 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: References: "Alex Vinokur" wrote in message news:abdp47$h3rj5$1@ID-79865.news.dfncis.de... [snip] | | || File#3 of 3 | || ######### File mbx800.h : BEGIN ######### | || Full path : D:\sys2\Tornado\target\config\mbx860\mbx800.h | | /* mbx800.h - Motorola MBX800 board header */ | | /* Copyright 1984-1998 Wind River Systems, Inc. */ | /* Copyright 1997,1998 Motorola, Inc., All Rights Reserved */ | | | [--- omitted ---] | | #ifndef _ASMLANGUAGE | | /* PCI I/O function defines */ | | [--- omitted ---] | | #ifndef PCI_IN_LONG | #define PCI_IN_LONG(x) sysInLong (x) | #endif ============================================================ | IMPORT UINT32 sysInLong (UINT32 address); || Line#417 I changed the line above to IMPORT ULONG sysInLong (ULONG address); This action helped. However, is it safe ? ============================================================ | | [--- omitted ---] | | #endif /* _ASMLANGUAGE */ | | [--- omitted ---] | | | || ######### File mbx800.h : END ########### | | [snip] --------------------------- Newsgroups: comp.os.vxworks Subject: Two GD82559ER adapters under VxWorks Date: 9 May 2002 06:35:24 -0700 From: voval@compulab.co.il (Vova) Organization: http://groups.google.com/ Message-ID: I have a board with two GD82559ER (both are correctly working). I use Intel END/BSD (same behaviour) driver for VxWorks. When I configure only one adapter everything works fine. However when the driver is configured to work with both Ethernet adapters, the network response sometimes becomes slower. As a test I ping the board from/to other computer (either same board connected via cross cable or another computer). When only one adapter is configured in the driver, each ping always takes exactly 1 ms. However with two adapters about 5% of pings have a response time of 10 to 80 ms. The second adapter is enabled in driver only, no TCP/IP is bond and it is not connected to Ethernet. (I tried to configure IP address of it and connect it to Ethernet as well - the same result). I am confident that it is a problem of software, because I saw with WindView that tNetTask sometimes becomes active (when two adapters are set) and if the ICMP echo request arrives at that time, the response time is higher. Both Ethernet adapters work fine together under Linux. The configuration is: { 0, EEE82559_LOAD_FUNC, EEE82559_LOAD_STRING, EEE82559_BUFF_LOAN, NULL, FALSE}, { 1, EEE82559_LOAD_FUNC, EEE82559_LOAD_STRING, EEE82559_BUFF_LOAN, NULL, FALSE}, where all defines are standard. Does anyone know the phenomenon and what to do with it? --------------------------- Newsgroups: comp.os.vxworks Subject: How to define a Tornado startup script? Date: 9 May 2002 07:13:59 -0700 From: rprewitt@oni.com (Rob Prewitt) Organization: http://groups.google.com/ Message-ID: There are some commands we need to run on our target before starting a debugger/Windview/etc. Specifically, we have a command to unprotect the text section. Is there a way to specify a file with these type of commands that run automatically upon tool startup (or target server connection)? Thanks in advance, Rob Prewitt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Advice on TCP/UDP task arrangement Date: Thu, 9 May 2002 10:25:04 -0400 From: Keith Arner Organization: Marconi Message-ID: <313680C9A886D511A06000204840E1CF67A17A-100000@whq-msgusr-02.pit.comms.marconi.com> References: Reply-To: Keith Arner On 9 May 2002, RichardBarry wrote: > I have a general question on using UDP from within a task, any advice > would be gratefully received. > > I have been experimenting with a system where I create a UDP socket > then spawn one task that reads from the socket and one that writes to > the socket. The writing task blocks on a message queue, and when a > message arrives formats it and transmits it. The receiving task > blocks on a call to recvfrom(), and when a message arrives, formats it > and posts it onto a different message queue. > > This seems to work, however is this a valid thing to do? i.e. is > mutual exclusion around the socket guaranteed by the OS, or must I > provide it explicitly? What would happen in the Tx task deleted the > socket while the Rx task was blocked on recvfrom()? If one task closes a socket (or any fd, for that matter), while another task is still attempting to operate on it, Bad Things might happen. If the Rx task is *acutally in* recvfrom(), and waiting on the semaphore for that socket, then it will get an good error code from recvfrom(). However, how can you tell this from the Tx task? What if the Rx task is just about to call recvfrom() when the Tx task closes the socket, then another task jumps in and opens another socket? Then, the Rx task does a recvfrom() on another application's socket, which is not what you want at all. You really need to view open()/close() (or socket()/close()) just like malloc()/free(). Both are managing access to a set of dynamically allocatable, globally shared resources. If one task is still operating on a memory region when another task free()'s it Bad Things will happen. The same argument holds for one task operating on a socket or fd when another task close()'s it. Any time you have tasks sharing resources, you need to make sure that all tasks are done operating on the resource before you free it up. Keith --------------------------- Newsgroups: comp.os.vxworks Subject: Inside ISR check? Date: Thu, 09 May 2002 14:45:40 GMT From: "doug dotson" Organization: Verio Message-ID: Is there any way to determine if you are inside an ISR? doug --------------------------- Newsgroups: comp.os.vxworks Subject: Non blocking UDP sockets? Date: 9 May 2002 08:46:38 -0700 From: barryrichard@hotmail.com (RichardBarry) Organization: http://groups.google.com/ Message-ID: Hi, I am setting up a socket to be non-blocking with the follwing call int iValue = 1; ioctl( Socket, FIONBIO, &iValue ); This works fine for one OS, but under VxWorks seems to render all sendto() calls from that point on (even on unreleated sockets) useless - - with errno getting set to EHOSTUNREACH. Any ideas??? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Advice on TCP/UDP task arrangement Date: 9 May 2002 09:02:38 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205090802.5b5e4817@posting.google.com> References: <3CDA52E1.F1B752C4@idl.bt.co.uk> Hello, > Since a socket must be some form of waitable object (how does recvfrom() > actually block?), then closing it releases waiting tasks, returning an > error code to their caller. All seems to work reliably for me ;) You can easily tell what it is blocking on... wait for the task to block and then use tt() from a shell to get you a stack trace. I think you'll find there's a semaphore at the bottom of it ;-) HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Two GD82559ER adapters under VxWorks Date: Thu, 9 May 2002 16:32:06 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CDAA4E1.8070807@spamm.me.l8s.co.uk> References: > I am confident that it is a problem of software, because I saw with > WindView that tNetTask sometimes becomes active (when two adapters are > set) You MIGHT be able to catch what it is doing by typing "tt tNetTask" into the target shell a lot of times! OTOH you could guess where it might be and set a breakpoint (having removed the VX_UNBREAKABLE flag from tNetTask) - again from the target shell. If the system is actually busy-looping, a statistical code performace moniter would help - ie something that takes a fast timer interrupt and builds a histogram of the interrupted PC. I don't think vxWorks has one..... but very useful for finding out which piece of code to optimise. David --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: Re: cross compiling gcc - parse error Date: 9 May 2002 09:37:27 -0700 From: demiurg@tau.ac.il (Alexander Sirotkin) Organization: http://groups.google.com/ Message-ID: References: <3CDA539D.F483A489@idl.bt.co.uk> I'm trying to build both compiler and linker (binutils) and they obviously need to include vxworks.h... am I wrong ? Phil Ashby wrote in message news:<3CDA539D.F483A489@idl.bt.co.uk>... > Alexander Sirotkin wrote: > > > > I'm trying to compile gcc & bintutils for vxworks on Linux, > > binutils are compiled just fine, but gcc throuws the following error: > > > > /root/gnu/gcc-2.95/gcc/config/mips/vxworks.h:3: parse error before `file' > > > > Apparently, it does not like embedded comments > > > > vxworks.h:3 > > /* Copyright (C) 1999 Free Software Foundation, Inc. */ > > This file is part of GNU CC. > > That looks wrong. Obviously the text 'This file...' should be in a > coment block, but isn't. Why should a cross-compiler need to include > a BSP-specific vxworks.h anyway? > > Surely you just need to create a cross-compiler for the target CPU, > not the O/S? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Inside ISR check? Date: Thu, 9 May 2002 10:07:21 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: I think there is a global int variable called intCnt. - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Advice on TCP/UDP task arrangement Date: Thu, 9 May 2002 10:17:53 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: There might be a small window where things get out of wack. I personally have not seen such window of confusion but it is theoretically possible in systems with a lot of interrupts and prioritized tasks. I would advise that the application level code coordinates transaction sessions. For example, before writing task closes the datagram socket, it can send a "FINISH" message to receiving side, which then "ACK" that message. Sounds like a TCP? :-) - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Two GD82559ER adapters under VxWorks Date: Thu, 09 May 2002 13:46:29 -0400 From: George Varndell Organization: Newsfeeds.com http://www.newsfeeds.com 100,000+ UNCENSORED Newsgroups. Message-ID: <3CDAB5F5.916FCC56@ineoquest.com> References: <3CDAA4E1.8070807@spamm.me.l8s.co.uk> David Laight wrote: [snip] > > If the system is actually busy-looping, a statistical code performace > moniter would help - ie something that takes a fast timer interrupt > and builds a histogram of the interrupted PC. I don't think > vxWorks has one..... but very useful for finding out which piece > of code to optimise. Real-Time Innovations, Inc., a Wind partner, has ProfileScope. It is indeed useful for debugging this sort of problem. It is possible that the driver is off polling the PHY. Regards, George Varndell --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Two GD82559ER adapters under VxWorks Date: Thu, 9 May 2002 18:11:26 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CDABC27.2030007@spamm.me.l8s.co.uk> References: <3CDAA4E1.8070807@spamm.me.l8s.co.uk> <3CDAB5F5.916FCC56@ineoquest.com> >>If the system is actually busy-looping, a statistical code performace >>moniter would help - ie something that takes a fast timer interrupt >>and builds a histogram of the interrupted PC. I don't think >>vxWorks has one..... but very useful for finding out which piece >>of code to optimise. >> > > Real-Time Innovations, Inc., a Wind partner, has ProfileScope. > It is indeed useful for debugging this sort of problem. And many others... Probably more use than the windview stuff you get for free that just tells you which task is busy. I managed to get time attributed to the caller of certain routines - eg bcopy(), malloc() etc. > It is possible that the driver is off polling the PHY. Or just doesn't support 2 interfaces :-( David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Inside ISR check? Date: Thu, 09 May 2002 14:18:10 -0500 From: Andrew Johnson Organization: APS, Argonne National Laboratory Message-ID: <3CDACB72.E96D8F45@aps.anl.gov> References: doug dotson wrote: > > Is there any way to determine if you are inside an > ISR? BOOL intContext(void); It's in intLib.h - - Andrew - -- Perfection is reached, not when there is no longer anything to add, but when there is no longer anything to take away. - - Antoine de Saint-Exupery --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Inside ISR check? Date: Thu, 09 May 2002 19:39:59 GMT From: "doug dotson" Organization: Verio Message-ID: References: <3CDACB72.E96D8F45@aps.anl.gov> That'll do it, thanks! doug "Andrew Johnson" wrote in message news:3CDACB72.E96D8F45@aps.anl.gov... > doug dotson wrote: > > > > Is there any way to determine if you are inside an > > ISR? > > BOOL intContext(void); > > It's in intLib.h > > - Andrew > -- > Perfection is reached, not when there is no longer anything to add, > but when there is no longer anything to take away. > - Antoine de Saint-Exupery --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech,comp.sys.powerpc.misc,comp.sys.powerpc.advocacy Subject: Memory data integrity preservation Date: 9 May 2002 13:40:03 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: This posting concerns the usage of 8260 Memory Controller's - - Write-protect feature - - Error Detection and Correction feature to preserve data integrity in flash and EEPROM. I'm thinking that, unlike DRAM, flash and EEPROM are nonvolatile. Errors would persist. It may make sense to offer some integrity assurance. + Is it common to use either/both of the above for this purpose? + I'm thinking about dynamically changing the write-protect bit in BR to enable/disable this feature. This is so that the memory is not wide open to store trampler at all time. Any catches? + As far as error detection/correction is concerned, there're choices - normal parity - read-modify-write parity - ECC correction and checking when do we use which one? And also when error is detected, would this cause an exception that I'd handle, if I want the S/W to retry, or at least to recover (without resetting the system). How much slower are we talking about? + Would it be inconsistent if I choose not to, say, do the same memory data error checking on DRAM? Presumably the content there could be corrupted before it is stored into the non-volatile store. The data would still be invalid, but the NV-store won't detect it. Thanks, Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: Non-Blocking Socket is Blocking Date: Thu, 9 May 2002 13:58:00 -0700 From: "Tim Ellis" Organization: Verio Message-ID: I have a TCP non-blocking socket, that I am sending data out on. When I unplug the Ethernet cable from my instrument, my send() blocks until the Ethernet cable is plugged back in. Is this an expected behavior? I was hoping that on a failed send(), I could do some further error checking (maybe check error status of tNetTask) to determine if my cable is disconnected. Tim --------------------------- Newsgroups: comp.os.vxworks Subject: C++ comments in C file with GNU compiler Date: Thu, 9 May 2002 14:56:29 -0700 From: "Haim Tebeka" Organization: Posted via Supernews, http://www.supernews.com Message-ID: This is a multi-part message in MIME format. - ------=_NextPart_000_0008_01C1F769.B3902680 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Is there any compilation switch that one can use so that the GNU = compiler will not spit a syntax error whenever it sees a C++ (//) style = comment in a C file? I haven't looked into this very deeply yet but if = one of you knows the answer, that would save me some time... Thanks, h - ------=_NextPart_000_0008_01C1F769.B3902680 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Is there any compilation switch that = one can use so=20 that the GNU compiler will not spit a syntax error whenever it sees a = C++ (//)=20 style comment in a C file? I haven't looked into this very deeply yet = but if one=20 of you knows the answer, that would save me some time...
 
Thanks,
 
h
 
 
- ------=_NextPart_000_0008_01C1F769.B3902680-- --------------------------- Newsgroups: comp.os.vxworks Subject: We're havin' some fun now! Date: Thu, 09 May 2002 18:07:12 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3CDB1D40.F9C93A9@gv.net> Reply-To: mkesti@gv.net After more than six months of unemployment, I've recently been on contract doing BSP and board bring-up for a small company. In the last three days I've written, debugged (initially!), and began testing a vxWorks device driver and I just want to say that I am REALLY LOVING 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: ENVOY SNMP Date: 9 May 2002 18:11:42 -0700 From: chrismooremobile@yahoo.com (Chris Moore) Organization: http://groups.google.com/ Message-ID: References: <3CD35445.8123C2F0@worldnet.att.net> Les Cargill wrote in message news:<3CD35445.8123C2F0@worldnet.att.net>... > Jason wrote: > > > > Hello: > > > > I have successfully built the ENVOY SNMP module for PPC, but it quite > > big (1.5MBytes). I hope to reduce it. Does any one have the idea ? > > > > One way I want to try is, among the 11 groups of MIB II, only include > > a few of them, say the system and ip, exclude the rest 9 groups. But > > how to do it exactly? > > > > Use the "-exclude" option when invoking the MIB compiler. > > > Thanks in advance! Hi Everyone, We use ENVOY 9.1 SNMP and find that is is REALLLLY slow when working with HP Open view and ...AdventNet SNMP mangers. Using "snmptool" from the win32 CLI its fine and fast.... does anyone know if this is a setting problem on either the client or the manager? Cheers, --------------------------- Newsgroups: comp.os.vxworks Subject: Accessing hard disk Date: 9 May 2002 22:29:55 -0700 From: akjayant@hss.hns.com (Maverick) Organization: http://groups.google.com/ Message-ID: <253a5f89.0205092129.150afb1d@posting.google.com> Hi, I am trying to create a Vxwork image that can support hard disk. I have included IDE hard disk drive but when I tried to build I got an error: undefined reference to ideTypes. What all changes in configuration files do I need to do so that I can use hard disk on pcpentium target? To use hard disk I had followed the following procedure 1) Initialise IDE driver by using ideDrv() 2) Create device for IDE disk ideDevCreate() 3) Initialise the device with dos File system by using dosFsDevInit() 4) Use hard disk. Is this the procedure to use hard disk correct? Any pointers will be greately appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation for MBX860 target failed Date: Fri, 10 May 2002 05:40:49 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: "Alex Vinokur" wrote in message news:abdqt2$hdsib$1@ID-79865.news.dfncis.de... > > | IMPORT UINT32 sysInLong (UINT32 address); || Line#417 > > I changed the line above to > > IMPORT ULONG sysInLong (ULONG address); > > This action helped. > > However, is it safe ? > Yes, but I'm curious as to why the problem developed after you "changed nothing". Something obviously changed. Did you load a Tornado or BSP patch? --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech,comp.sys.powerpc.misc,comp.sys.powerpc.advocacy Subject: Re: Memory data integrity preservation Date: Fri, 10 May 2002 05:52:55 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: Comments below: "Raymond Yeung" wrote in message news:d83bb8e3.0205091240.70e6f074@posting.google.com... > This posting concerns the usage of 8260 Memory Controller's > - Write-protect feature > - Error Detection and Correction feature > to preserve data integrity in flash and EEPROM. > > I'm thinking that, unlike DRAM, flash and EEPROM are nonvolatile. > Errors would persist. It may make sense to offer some integrity > assurance. Generally a good idea. > > + Is it common to use either/both of the above for this purpose? Yes, either/both are commonly used, depending on the application. > + I'm thinking about dynamically changing the write-protect bit > in BR to enable/disable this feature. This is so that the memory > is not wide open to store trampler at all time. Any catches? Well, that's really the job of the OS, working in conjuction with the toolset and the loader. Your code would have to be aware of all the possible memory interactions going on (tasks, I/O) to be a true enforcer. You are trusting a piece of the application to do the work of the OS. > + As far as error detection/correction is concerned, there're choices > - normal parity > - read-modify-write parity > - ECC correction and checking > when do we use which one? And also when error is detected, would > this cause an exception that I'd handle, if I want the S/W to retry, > or at least to recover (without resetting the system). > How much slower are we talking about? ECC can recover from single bit and certain double bit errors, depending on the logic being used. Parity can only detect errors, without recovering the data. The time required to calculate either parity or ECC is practically buried in the memory logic; however, error detecting/correcting memory is more expensive. You have to consider the statistical probability of encountering a memory error over the life of the product to justify the use of error-detecting/correcting logic. > + Would it be inconsistent if I choose not to, say, do the same memory > data error checking on DRAM? Presumably the content there could be > corrupted before it is stored into the non-volatile store. The data > would still be invalid, but the NV-store won't detect it. It all depends on when you place guarantees on when the data is "valid". If you apply a suitable CRC check to a block of memory when you can guarantee its correctness, you can always be able to determine if the data is correct later on when you retrieve it or store it somewhere else. The main point is: when is the data guaranteed to be correct? > > Thanks, > Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ comments in C file with GNU compiler Date: Fri, 10 May 2002 05:55:56 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: If you remove the -ansi switch from the compiler settings, "//" style comments will be allowed. However, other ANSI rules are also relaxed. Check the compiler documentation. "Haim Tebeka" wrote in message news:udls4hdm5mvh97@corp.supernews.com... Is there any compilation switch that one can use so that the GNU compiler will not spit a syntax error whenever it sees a C++ (//) style comment in a C file? I haven't looked into this very deeply yet but if one of you knows the answer, that would save me some time... Thanks, h --------------------------- Newsgroups: comp.os.vxworks Subject: problem launching the target server on host Date: 9 May 2002 23:54:50 -0700 From: forvxworks@rediffmail.com (Anand) Organization: http://groups.google.com/ Message-ID: Hi All, i am configuring a vxWorks target for PENTIUM CPU using Network as the connection between host and target. I am able to boot the target in vxWorks using the bootable floppy i made by giving the appropriate bootline,IO address and the Interrupt level of the Ethernet Adapter. but while launching the target server on the host machine, with the backend "wdbrpc" , the server logconsole gives the following message.. Connecting to Target agent.... Error:rpccore backend client Timed out failed. On the target machine i am getting the massages as follows... muxDevLoad failed for device entry0! Attached TCP/IP interface to ene unit0 Attaching intarface lo0...done ........ ....... WDB : Ready -> the Ethernet Adapter i am using is .. SURECOM EP-325 PCI Ethernet Adapter. I am unable to the get what the problem is. Can anyone help?? Thanks in advance. Regards. Anand. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Non-Blocking Socket is Blocking Date: Fri, 10 May 2002 06:56:56 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Tim, I done a lot of non-blocking TCP stuff, and have not seen this kind of behaviour, although I did not try to unplug the ethernet cable, and if I did, it would not matter because the TCP output is the main thing this device is supposed to do. What I can think of as a possible reason to this blocking is the mutex - the net stack has a single mutex (emulation of splnet() and splx() .. ) and any socket or other net function would block if someone else has taken up this mutex. In your case, unplugging the cable may mean that the driver has to go into PHY auto-negotiation phase. We have already dealt with the problem of netTask going into lengthy processing pohase while doing PHY autonegotiation, and surely enough the netTask executing driver code would be taking the splnet mutex. "Tim Ellis" wrote in message news:hqBC8.14075$JZ6.238882@dfw-read.news.verio.net... > I have a TCP non-blocking socket, that I am sending data out on. > > When I unplug the Ethernet cable from my instrument, my send() blocks until > the Ethernet cable is plugged back in. > > Is this an expected behavior? > > > I was hoping that on a failed send(), I could do some further error checking > (maybe check error status of tNetTask) to determine if my cable is > disconnected. > > > Tim > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Non blocking UDP sockets? Date: Fri, 10 May 2002 06:58:59 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: I do not know why this fails, but I very much doubt there is any point in doing NB on UDP socket, cause UDP sockets actually NEVER BLOCK except when receiving, and you can poll the socket for data before going into reception with ioctl FIONREAD. "RichardBarry" wrote in message news:eec86974.0205090746.587b97c3@posting.google.com... > Hi, > > I am setting up a socket to be non-blocking with the follwing call > > int iValue = 1; > ioctl( Socket, FIONBIO, &iValue ); > > This works fine for one OS, but under VxWorks seems to render all > sendto() calls from that point on (even on unreleated sockets) useless > - with errno getting set to EHOSTUNREACH. > > Any ideas??? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Non blocking UDP sockets? Date: Fri, 10 May 2002 09:04:46 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CDB8D86.6090009@spamm.me.l8s.co.uk> References: Leonid Rosenboim wrote: > I do not know why this fails, but I very much doubt > there is any point in doing NB on UDP socket, > cause UDP sockets actually NEVER BLOCK > except when receiving, IMHO this is a bug! I believe UDP sockets should block for (local) flow control reasons. I'm fairly sure the SuS allows send() to block, and probably expects it to. For instance a process level NFS stack will not work if the UDP datagrams are discarded just because the network is busy, or some internal socket send buffer is full.... Clearly UDP traffic CAN be discarded - but it should only be done to relieve a buffer crisis. David --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri May 10 05:55:33 2002 From: Diego Novellon Date: Fri May 10 05:55:35 PDT 2002 Subject: Re: C++ comments in C file with GNU compiler > Newsgroups: comp.os.vxworks > Subject: Re: C++ comments in C file with GNU compiler > Date: Fri, 10 May 2002 05:55:56 GMT > From: "Gary M" > Organization: AT&T Broadband > Message-ID: > References: > > If you remove the -ansi switch from the compiler settings, "//" style > comments will be allowed. However, other ANSI rules are also relaxed. > Check > the compiler documentation. > > "Haim Tebeka" wrote in message > news:udls4hdm5mvh97@corp.supernews.com... > Is there any compilation switch that one can use so that the GNU > compiler > will not spit a syntax error whenever it sees a C++ (//) style comment > in a > C file? I haven't looked into this very deeply yet but if one of you > knows > the answer, that would save me some time... > > Thanks, > > h Someone else in this forum recently suggested the flag "-Wp,-lang-c" (w/o the quotes). Works fine for me (VxWorks 5.4.2 / pcPentium). Just my two eurocents. Diego Novellon From vxwexplo-errs@csg.lbl.gov Sat May 11 04:03:14 2002 From: Vxworks Exploder Date: Sat May 11 04:03:16 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat May 11 04:03:09 PDT 2002 Subject: Re: problem launching the target server on host Subject: 860 SMC question ... Subject: Re: Non blocking UDP sockets? Subject: Re: Non blocking UDP sockets? Subject: Re: Memory data integrity preservation Subject: Re: cross compiling gcc - parse error Subject: Link error while working with SBC8260 Subject: Re: Bsp Testing Subject: Re: Memory data integrity preservation Subject: A question about cacheDmaMalloc! Subject: Compilation for MBX860 target failed Subject: Re: Compilation log file Subject: ld() error Subject: Re: ld() error Subject: Re: ld() error Subject: implementing clrscr() in vxworks Subject: Re: getting the base address of the PCI card Subject: Re: problem launching the target server on host Subject: How to use AF_ROUTE ? Subject: Re: Link error while working with SBC8260 Subject: How do I do FTP on a non-default port using PORT and PASV ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem launching the target server on host Date: Fri, 10 May 2002 13:56:55 GMT From: igtorque@yahoo.es (Ignacio Gonzalez T.) Organization: Newshosting.com - Highest quality at a great price! www.newshosting.com Message-ID: <3cdbcf80.172834242@news.mhogaming.com> References: On 9 May 2002 23:54:50 -0700, forvxworks@rediffmail.com (Anand) wrote: >Hi All, > i am configuring a vxWorks target for PENTIUM CPU using Network as >the connection between host and target. > I am able to boot the target in vxWorks using the bootable floppy >i >made by giving the appropriate bootline,IO address and the Interrupt >level of the Ethernet Adapter. but while launching the target server >on the host machine, with the backend "wdbrpc" , the server logconsole >gives the following message.. > > Connecting to Target agent.... > Error:rpccore backend client Timed out > failed. > > >On the target machine i am getting the massages as follows... > > > > muxDevLoad failed for device entry0! > Attached TCP/IP interface to ene unit0 > Attaching intarface lo0...done > > ........ > ....... > WDB : Ready > -> > > > the Ethernet Adapter i am using is .. > > SURECOM EP-325 PCI Ethernet Adapter. > > I am unable to the get what the problem is. Can anyone help?? > >Thanks in advance. >Regards. >Anand. Maybe your configNet.h is not correct. More precisely, the definition of END_TBL_ENTRY endDevTbl [] may be missing something important related to the Ethernet driver. If it were correct, you should expect, at worst, something like muxDevLoad failed for device sep0! instead of muxDevLoad failed for device entry0! --------------------------- Newsgroups: comp.os.vxworks Subject: 860 SMC question ... Date: Fri, 10 May 2002 14:57:05 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <5dRC8.21403$f5.1728610@news> Hello, I have a question for you MPC860 lovers out there ... I am trying to write a MPC 860 polled UART driver that I can call from romStart.c ... Nothing seems to be printing yet ... and I haven't spent to much time debugging ... but any suggestions would be awesome ... the code is below ... Jim /* SMC 2 Pins Information */ #define portB23_SMSYN1_mask (0x1<<31-23) #define portB24_SMRXD1_mask (0x1<<31-24) #define portB25_SMTXD1_mask (0x1<<31-25) /* Port B Registers */ #define PBPAR_ADDR 0x02200abc #define PDDIR_ADDR 0x02200ab8 #define PBODR_ADDR 0x02200ac0 #define PBDAT_ADDR 0x02200ac4 /*SMC Clocks Registers*/ #define CPU_CLCK 80000000 /*80 Mhz*/ #define BRG1_ADDR 0x22009F0 #define SCCR_ADDR 0x2200280 /*SI Mode Register*/ #define SIMODE_ADDR 0x2200ae0 /*SMS Parameter Ram*/ #define SMC1_PRAM_START_ADDR 0x2203E80 #define SMC1_PRAM_TBASE_ADDR (SMC1_PRAM_START_ADDR + 0x00) #define SMC1_PRAM_RBASE_ADDR (SMC1_PRAM_START_ADDR + 0x02) #define SMC1_PRAM_RFCR_ADDR (SMC1_PRAM_START_ADDR + 0x04) #define SMC1_PRAM_TFCR_ADDR (SMC1_PRAM_START_ADDR + 0x05) #define SMC1_PRAM_MRBLR_ADDR (SMC1_PRAM_START_ADDR + 0x06) #define SMC1_PRAM_MAX_IDL_ADDR (SMC1_PRAM_START_ADDR + 0x28) #define SMC1_PRAM_BRKLN_ADDR (SMC1_PRAM_START_ADDR + 0x2C) #define SMC1_PRAM_BRKEC_ADDR (SMC1_PRAM_START_ADDR + 0x2E) #define SMC1_PRAM_RKCR_ADDR (SMC1_PRAM_START_ADDR + 0x30) /*CPCR Communications Processor Command Register*/ #define CPCR_ADDR 0x022009c0 /*CIMR*/ #define CIMR_ADDR 0x2200948 #define CICR_ADDR 0x2200940 /*SDMA Config Register*/ #define SDCR_ADDR 0x02200030 /*SMC1 Event Register*/ #define SMCE1_ADDR 0x02200a86 /*SMC1 Mask Register*/ #define SMCM1_ADDR 0x02200a8a /*SMC1 Mode Register*/ #define SMCMR1_ADDR 0x02200a82 /**********************************************************************/ void InitSMC2ForUart() { unsigned SCCR_DFBRG; unsigned default_baud_rate; unsigned tager_brg1_rate = 9600 * 16; unsigned BRG1_CD; unsigned BRG1_DIV16; static char rx_buff[10]; #define TX_BUFF "hello\n" static unsigned tmp; /*--------------------------------------------------------------------- * 1.) ConÞgure the port B pins to enable SMTXD1 and SMRXD1. * Set PBPAR[24, 25] then clear PBDIR[24, 25] and PBODR[24, 25]. ---------------------------------------------------------------------*/ *(unsigned *)PBPAR_ADDR = (*(unsigned *)PBODR_ADDR) | ~(portB24_SMRXD1_mask | portB25_SMTXD1_mask); *(unsigned *)PDDIR_ADDR = (*(unsigned *)PBODR_ADDR) & ~(portB24_SMRXD1_mask | portB25_SMTXD1_mask); *(unsigned *)PBODR_ADDR = (*(unsigned *)PBODR_ADDR) & ~(portB24_SMRXD1_mask | portB25_SMTXD1_mask); /*--------------------------------------------------------------------- * 2.) ConÞgure the BRG1. Write BRGC1 with 0x01_0144. * The DIV16 bit is not used and the divider is 162 (decimal). * The resulting BRG1 clock is 16´ the preferred bit rate. ---------------------------------------------------------------------*/ SCCR_DFBRG = ( ( (*(unsigned *)SCCR_ADDR) >> (31-20) )& 0x3 ); if ( SCCR_DFBRG == 0x0 ) { default_baud_rate = CPU_CLCK / 1; } else if ( SCCR_DFBRG == 0x1 ) { default_baud_rate = CPU_CLCK / 4; } else if ( SCCR_DFBRG == 0x2 ) { default_baud_rate = CPU_CLCK / 16; } else { default_baud_rate = CPU_CLCK / 64; } BRG1_CD = default_baud_rate/tager_brg1_rate; if ( BRG1_CD > (0xfff) ) { /*CD is only 12 bits*/ BRG1_CD = (default_baud_rate/16)/tager_brg1_rate; BRG1_DIV16 = 1; } else { BRG1_DIV16 = 0; } *(unsigned *)BRG1_ADDR = (0x00010000 | (BRG1_CD<<1) | BRG1_DIV16); /*--------------------------------------------------------------------- * 3. Connect BRG1 to SMC1 using the SI. Clear SIMODE[SMC1, SMC1CS]. ---------------------------------------------------------------------*/ *(unsigned *)SIMODE_ADDR = (*(unsigned *)SIMODE_ADDR) & ~(0x0000f000); /*--------------------------------------------------------------------- * 4. Assuming one RxBD at the beginning of dual-port RAM * followed by one TxBD, write RBASE with 0x0000 and * TBASE with 0x0008. ---------------------------------------------------------------------*/ * (unsigned short *)SMC1_PRAM_TBASE_ADDR = 0x0008; * (unsigned short *)SMC1_PRAM_RBASE_ADDR = 0x0000; /*--------------------------------------------------------------------- * 5. Write 0x0091 to CPCR to execute the INIT RX AND TX PARAMETERS command. ---------------------------------------------------------------------* / *(unsigned short *)CPCR_ADDR = 0x0091; /*--------------------------------------------------------------------- * 6. Initialize the SDMA conÞguration register (SDCR) to 0x0001. ---------------------------------------------------------------------* / * (unsigned short *)SDCR_ADDR = 0x0001; /*--------------------------------------------------------------------- * 7. Write RFCR and TFCR with 0x10 for normal operation. ---------------------------------------------------------------------* / * (unsigned char *)SMC1_PRAM_RFCR_ADDR = (unsigned char)0x10; * (unsigned char *)SMC1_PRAM_TFCR_ADDR = (unsigned char)0x10; /*--------------------------------------------------------------------- * 8. Write MRBLR with the maximum number of bytes per receive buffer. * Assume 16 bytes, so MRBLR = 0x0010. ---------------------------------------------------------------------* / * (unsigned short *)SMC1_PRAM_MRBLR_ADDR = 0x0010; /*--------------------------------------------------------------------- * 9. Write MAX_IDL with 0x0000 in the SMC UART-speciÞc parameter RAM to * disable the MAX_IDL functionality for this example. ---------------------------------------------------------------------* / * (unsigned short *) SMC1_PRAM_MAX_IDL_ADDR = 0x0000; /*--------------------------------------------------------------------- * 10. Clear BRKLN and BRKEC in the SMC UART-speciÞc parameter RAM. ---------------------------------------------------------------------* / * (unsigned short *)SMC1_PRAM_BRKLN_ADDR = 0x0000; * (unsigned short *)SMC1_PRAM_BRKEC_ADDR = 0x0000; /*--------------------------------------------------------------------- * 11. Set BRKCR to 0x0001; if a STOP TRANSMIT COMMAND is issued, * one break character is sent. ---------------------------------------------------------------------* / * (unsigned short *) SMC1_PRAM_RKCR_ADDR = 0x0001; /*--------------------------------------------------------------------- * 12. Initialize the RxBD. Assume the Rx buffer is at 0x0000_1000 * in main memory. Write 0xB000 to Rx_BD_Status, 0x0000 to Rx_BD_Length * (not required), and 0x0000_1000 to Rx_BD_Pointer. ---------------------------------------------------------------------* / *(unsigned short*)0x2202000= 0xB000; *(unsigned short*)0x2202002= 10; *(unsigned *)0x2202004= (unsigned *)rx_buff; /*--------------------------------------------------------------------- * 13. Assuming the Tx buffer is at 0x00002000 in main memory and * contains Þve 8-bit characters, write 0xB000 to Tx_BD_Status, * 0x0005 to Tx_BD_Length, and 0x00002000 to Tx_BD_Pointer. ---------------------------------------------------------------------* / *(unsigned short*)0x2202008= 0xB000; *(unsigned short*)0x220200a= 6; *(unsigned *)0x220200c= (unsigned *)0x2800130; /*--------------------------------------------------------------------- * 14. Write 0xFF to the SMCE register to clear any previous events. ---------------------------------------------------------------------* / * (unsigned char *) SMCE1_ADDR = 0xff; /*--------------------------------------------------------------------- * 15. Write 0x17 to the SMCM register to enable all possible SMC * interrupts. * Note: I want no interupts enabled so I set this to 0x0 ... like * a good boy ---------------------------------------------------------------------* / * (unsigned char *) SMCM1_ADDR = 0x0; /*--------------------------------------------------------------------- * 16. Write 0x00000010 to the CIMR so the SMC1 can generate a system * interrupt. Initialize the CICR. * Note: I want no interupts enabled so I DO NOT DO THIS! ---------------------------------------------------------------------* / *(unsigned *) CIMR_ADDR = 0x00000010; *(unsigned *) CICR_ADDR = 0x00000000; /*--------------------------------------------------------------------- * 17. Write 0x4820 to SMCMR to conÞgure normal operation (not loopback), * 8-bit characters, no parity, 1 stop bit. * The transmitter and receiver are not yet enabled. ---------------------------------------------------------------------* / * (unsigned short *) SMCMR1_ADDR = 0x4820; /*--------------------------------------------------------------------- * 18. Write 0x4823 to SMCMR to enable the SMC transmitter and * receiver. This additional write ensures that the TEN and * REN bits are enabled last. ---------------------------------------------------------------------* / * (unsigned short *) SMCMR1_ADDR = 0x4823; } /**********************************************************************/ void SMC2_Write(char *str) { int i; for (i=0;i<256 || str[i]!='\0';i+=1); *(unsigned short*)0x2202008= 0xB000; *(unsigned short*)0x220200a= i; *(unsigned *)0x220200c= (unsigned *)0x2800130; } --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Non blocking UDP sockets? Date: Fri, 10 May 2002 15:12:20 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <3CDB8D86.6090009@spamm.me.l8s.co.uk> Regarding UDP sockets, to the best of my knowledge, the socket does not block on send, if there are no buffers send() will return ENUBUFS error, if the packet was dropped on the network or due to excessive colissions or CRC error, the application is not made aware of that. This holds for VxWorks and Unix derivatives I have used. Regarding NFS, it does block the caller until the request is succesful, through the use of acknowledges and retransmission at the RPC level, as NFS is a protocol defined on top of ONC RPC. "David Laight" wrote in message news:3CDB8D86.6090009@spamm.me.l8s.co.uk... > Leonid Rosenboim wrote: > > > I do not know why this fails, but I very much doubt > > there is any point in doing NB on UDP socket, > > cause UDP sockets actually NEVER BLOCK > > except when receiving, > > > IMHO this is a bug! I believe UDP sockets should block > for (local) flow control reasons. I'm fairly sure the > SuS allows send() to block, and probably expects it to. > > For instance a process level NFS stack will not work if > the UDP datagrams are discarded just because the network > is busy, or some internal socket send buffer is full.... > > Clearly UDP traffic CAN be discarded - but it should only > be done to relieve a buffer crisis. > > David > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Non blocking UDP sockets? Date: 10 May 2002 08:14:07 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205100714.5c6a4469@posting.google.com> References: <3CDB8D86.6090009@spamm.me.l8s.co.uk> Hello, > IMHO this is a bug! I believe UDP sockets should block > for (local) flow control reasons. I'm fairly sure the > SuS allows send() to block, and probably expects it to. Looking at the source for the network stack, a UDP sendto() call is checking the non-blocking state flag (and will return EWOULDBLOCK if necessary). It also calls sbwait() if the local send buffers are full and non-blocking is not set. I would conclude that it is possible for a UDP socket to block on sends at least. Rgds, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Memory data integrity preservation Date: 10 May 2002 08:27:12 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205100727.36d1e6de@posting.google.com> References: Hello, > > + I'm thinking about dynamically changing the write-protect bit > > in BR to enable/disable this feature. This is so that the memory > > is not wide open to store trampler at all time. Any catches? > > Well, that's really the job of the OS, working in conjuction with the > toolset and the loader. Your code would have to be aware of all the possible > memory interactions going on (tasks, I/O) to be a true enforcer. You are > trusting a piece of the application to do the work of the OS. Huh? Changing the write protect bit on the flash/PROM is not normally something I'd expect the toolset or loader to be involved in. I would expect the memory description of the OS to flag the area as read-only, and any software that is going to write to it to modify that protection while it writes to the flash and then restore the protection afterwards. > > + As far as error detection/correction is concerned, there're choices > > - normal parity > > - read-modify-write parity > > - ECC correction and checking > > when do we use which one? And also when error is detected, would > > this cause an exception that I'd handle, if I want the S/W to retry, > > or at least to recover (without resetting the system). > > How much slower are we talking about? > > ECC can recover from single bit and certain double bit errors, depending on > the logic being used. Parity can only detect errors, without recovering the > data. The time required to calculate either parity or ECC is practically > buried in the memory logic; however, error detecting/correcting memory is > more expensive. You have to consider the statistical probability of > encountering a memory error over the life of the product to justify the use > of error-detecting/correcting logic. I've never heard of anybody using ECC on flash/ROM. For a start, you need more bits (even the simple parity checking requires at least one extra bit). Secondly, writing to flash is performed using a procedure that requires commands to be presented to the device on the data bus - ECC would "corrupt" these commands, making it impossible to write to the flash from software (you can't disable the ECC to write since the content of the flash needs to be encoded for the ECC to work). Finally, to put the initial image in the devices, you would need something that could generate the ECC inclusive images from the ones produced during your build process. I think you should forget any ideas of doing ECC on Flash (unless you can find flash devices that support it internally). Checksumming (or similar) operations are much more viable for flash/ROM devices. HTH, John... --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: Re: cross compiling gcc - parse error Date: Fri, 10 May 2002 11:42:09 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3CDC1481.A1C0711B@lmco.com> References: <3CDA539D.F483A489@idl.bt.co.uk> Alexander Sirotkin wrote: > > I'm trying to build both compiler and linker (binutils) > and they obviously need to include vxworks.h... > am I wrong ? > > Phil Ashby wrote in message news:<3CDA539D.F483A489@idl.bt.co.uk>... > > Alexander Sirotkin wrote: > > > > > > I'm trying to compile gcc & bintutils for vxworks on Linux, > > > binutils are compiled just fine, but gcc throuws the following error: > > > > > > /root/gnu/gcc-2.95/gcc/config/mips/vxworks.h:3: parse error before `file' > > > > > > Apparently, it does not like embedded comments > > > > > > vxworks.h:3 > > > /* Copyright (C) 1999 Free Software Foundation, Inc. */ > > > This file is part of GNU CC. > > > > That looks wrong. Obviously the text 'This file...' should be in a > > coment block, but isn't. Why should a cross-compiler need to include > > a BSP-specific vxworks.h anyway? > > > > Surely you just need to create a cross-compiler for the target CPU, > > not the O/S? I think that you either have a bad copy of vxWorks.h, or maybe you have picked out the wrong line. The compiler doesn't always print correct line numbers for errors. BTW, on my copy of vxWorks.h, the text "this file.." does not appear. I see the line "/* Copy... " followed by a blank line, and a comment block with mod history. May I again suggest you contact WRS and get the source for their modified version of gcc? Even if it's not the revision you want, you would save a huge amount of effort. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Link error while working with SBC8260 Date: 10 May 2002 09:16:30 -0700 From: dannyc@harmonic.co.il (Danny Cohen) Organization: http://groups.google.com/ Message-ID: <5288f97d.0205100816.46bf3153@posting.google.com> Hi. I have evaluation board SBC8260. I'm trying to create bootable image and download it to the board. while compiling and linking the BSP i face the following error: "motFccEndLoad undefined at PartialImage.o". My feeling it has something with the BSP configuration. Can anyone help me to overcome this problem? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bsp Testing Date: Fri, 10 May 2002 20:16:23 +0200 From: "Rene Straub" <_remove_rene.straub@yetnet.ch> Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: References: <9e221d1.0205030538.4466b5b1@posting.google.com> <3CDA51DD.4030403@spamm.me.l8s.co.uk> You might also want to consider buying a ROM emulator. First it allows you to download code quickly, second some emulators have a debug port that is mapped into memory. When you write a byte to the debug port, the ROM emulator sends it to the debugger software, that can display and/or log it. ROM emulators are rather cheap compared to other debugging tools. - - Rene "David Laight" wrote in message news:3CDA51DD.4030403@spamm.me.l8s.co.uk... > Hwa Jin Bae wrote: > > > I have used various available mechanisms to debug low level BSP issues. For > > example, LEDs, PromICE, BDM, JTAG, Logic analyser, scope, Rom emulator, > > serial port, shared memory (over bus), etc. And of course, a lot of > > patience and tenacity help. Finally, some good luck. > > If your board has a UART, enable it very early on (eg in romInit.s) > and write routines to dump single characters, strings and hex > numbers to it (looping until the fifo is empty). > > This ought to let you debug system startup.... > > David > --------------------------- Newsgroups: comp.os.vxworks,comp.sys.powerpc.tech,comp.sys.powerpc.misc,comp.sys.powerpc.advocacy Subject: Re: Memory data integrity preservation Date: 10 May 2002 11:56:54 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: References: "Gary M" wrote in message news:... > Comments below: > > "Raymond Yeung" wrote in message > news:d83bb8e3.0205091240.70e6f074@posting.google.com... > > This posting concerns the usage of 8260 Memory Controller's > > - Write-protect feature > > - Error Detection and Correction feature > > to preserve data integrity in flash and EEPROM. > > > > I'm thinking that, unlike DRAM, flash and EEPROM are nonvolatile. > > Errors would persist. It may make sense to offer some integrity > > assurance. > > Generally a good idea. > > > > > + Is it common to use either/both of the above for this purpose? > > Yes, either/both are commonly used, depending on the application. Forgot to add. It's generally more involved to change flash content; one has to write a sequence of programming commands before one can write to the memory. So it then wouldn't be as easy to get trampled by strayed access? And so it probably wouldn't need write-protection? > > > + I'm thinking about dynamically changing the write-protect bit > > in BR to enable/disable this feature. This is so that the memory > > is not wide open to store trampler at all time. Any catches? > > Well, that's really the job of the OS, working in conjuction with the > toolset and the loader. Your code would have to be aware of all the possible > memory interactions going on (tasks, I/O) to be a true enforcer. You are > trusting a piece of the application to do the work of the OS. I'm currently working at the RTOS/drivers level. So, yes, I'd be imbedding this dynamic change of BR inside the flash/eeprom access functions, once I determine this is what I want to do. > > > + As far as error detection/correction is concerned, there're choices > > - normal parity > > - read-modify-write parity > > - ECC correction and checking > > when do we use which one? And also when error is detected, would > > this cause an exception that I'd handle, if I want the S/W to retry, > > or at least to recover (without resetting the system). > > How much slower are we talking about? > > ECC can recover from single bit and certain double bit errors, depending on > the logic being used. Parity can only detect errors, without recovering the > data. The time required to calculate either parity or ECC is practically > buried in the memory logic; however, error detecting/correcting memory is > more expensive. You have to consider the statistical probability of > encountering a memory error over the life of the product to justify the use > of error-detecting/correcting logic. This is the 1st time I look into this issue... Sounds like the memory devices I use need to support these detection/correction mechanism, not just 8260 memory controller. > > > + Would it be inconsistent if I choose not to, say, do the same memory > > data error checking on DRAM? Presumably the content there could be > > corrupted before it is stored into the non-volatile store. The data > > would still be invalid, but the NV-store won't detect it. > > It all depends on when you place guarantees on when the data is "valid". If > you apply a suitable CRC check to a block of memory when you can guarantee > its correctness, you can always be able to determine if the data is correct > later on when you retrieve it or store it somewhere else. The main point is: > when is the data guaranteed to be correct? I got a feeling that perhaps it's all just a question of extent of the problem (or exposure). Due to possibility of preemption, data generation and CRC generation may not be made indivisible. > > > > > Thanks, > > Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: A question about cacheDmaMalloc! Date: 10 May 2002 18:12:37 -0700 From: zhoul@blrcsv.china.bell-labs.com (zlybupt) Organization: http://groups.google.com/ Message-ID: <989cdae4.0205101712.2e0d865c@posting.google.com> Hi,all: I am designing a network driver not using end structure.Now I am puzzled by the following questions,I hope someone can give me an answer. Now I use cacheDmaMalloc function to allocate a memory for DMA use,but I wonder how can I inform the network card to know that it should this area for DMA use?? how can I realize it by code?? Thank you all in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: Compilation for MBX860 target failed Date: Sat, 11 May 2002 01:42:13 GMT From: Andray Kaganovsky Organization: Primus Canada Message-ID: References: "Alex Vinokur" wrote in news:abdqt2$hdsib$1@ID- 79865.news.dfncis.de: > > "Alex Vinokur" wrote in message news:abdp47$h3rj5$1 @ID-79865.news.dfncis.de... > > [snip] > >| >| || File#3 of 3 >| || ######### File mbx800.h : BEGIN ######### >| || Full path : D:\sys2\Tornado\target\config\mbx860\mbx800.h >| >| /* mbx800.h - Motorola MBX800 board header */ >| >| /* Copyright 1984-1998 Wind River Systems, Inc. */ >| /* Copyright 1997,1998 Motorola, Inc., All Rights Reserved */ >| >| >| [--- omitted ---] >| >| #ifndef _ASMLANGUAGE >| >| /* PCI I/O function defines */ >| >| [--- omitted ---] >| >| #ifndef PCI_IN_LONG >| #define PCI_IN_LONG(x) sysInLong (x) >| #endif > >============================================================ > >| IMPORT UINT32 sysInLong (UINT32 address); || Line# 417 > > I changed the line above to > > IMPORT ULONG sysInLong (ULONG address); > > This action helped. > > However, is it safe ? > >============================================================ >| >| [--- omitted ---] >| >| #endif /* _ASMLANGUAGE */ >| >| [--- omitted ---] >| >| >| || ######### File mbx800.h : END ########### >| >| > > [snip] > > > [AndrayK]. I suggest to look at the assembly code generated in both cases. This could provide definitive answer (for this particular BSP). Andray --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation log file Date: Sat, 11 May 2002 01:47:39 GMT From: Andray Kaganovsky Organization: Primus Canada Message-ID: References: "Alex Vinokur" wrote in news:abdho1$h4l0l$1@ID- 79865.news.dfncis.de: >========================================================= > Tornado : Version 2.0.2 > VxWorks : Version 5.4.2 > BSP : Version 1.1/1 > Compiler : ccsimpc > gcc version egcs-2.90.29 tornado 2.0 980515 > (egcs-1.0.3 release) > Host : Windows2000 > Target : VxSim for Windows (VxWorks target simulator) >========================================================= > > I would like the compilation results to be written > not only to Build Output Window, > but to some compilation log file as well. > > Is it possible to do it? > > > -- >==================== > Alex Vinokur > http://up.to/alexvn > http://go.to/alexv_math > mailto:alexvn@bigfoot.com > mailto:alexvn@go.to > ==================== > > > > [AndrayK]. Did you try building from DOS shell ? You'll need to execute /host/x86-win32/bin/torvars.bat before you invoke make .. regards, Andray --------------------------- Newsgroups: comp.os.vxworks Subject: ld() error Date: Sat, 11 May 2002 11:31:18 +0900 From: "bigfire" Organization: Inet Internet Services Message-ID: Hi everyone, when I load my application module using 'ld'command like below, ld< xx.o I got a error message, "Relocation value does not fit in 24 bits" My system is mpc755 VME board and the board boot well with down loaded vxWorks image from my PC. What's wrong with my BSP? thanks Dae-sop --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ld() error Date: Sat, 11 May 2002 05:22:43 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: One of your object files, most likely generated from an assembly language source file, has an instruction which is using an embedded 24-bit address parameter within the instruction. You need to find this instruction and convert it to a multi-instruction sequence which loads the address into a register and uses indirection through the register. "bigfire" wrote in message news:abhvjc$qda$1@news.nuri.net... > Hi everyone, > > when I load my application module using 'ld'command like below, > ld< xx.o > I got a error message, "Relocation value does not fit in 24 bits" > > My system is mpc755 VME board and the board boot well with down loaded > vxWorks image from my PC. > What's wrong with my BSP? > > thanks > > Dae-sop > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ld() error Date: 10 May 2002 22:57:09 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205102157.4541f63d@posting.google.com> References: Hello, I would suggest searching the archives for this group - this is probably one the most commonly asked questions (if not the most common). You can search the archives in a number of places, including Wind River's WindSurf site. My preference is http://groups.google.com though. As a hint, the problem relates to your setting of the LOCAL_MEM_SIZE greater than 32MB and a restriction imposed by the PPC EABI. There are a number of solutions to the problem, depending on your needs, including, but probably not limited to, compiling your application with -mlongcall or starting with LOCAL_MEM_SIZE at 32MB and adding the remainder of the memory to the system pool after you've loaded your module(s). HTH, John... "bigfire" wrote in message news:... > Hi everyone, > > when I load my application module using 'ld'command like below, > ld< xx.o > I got a error message, "Relocation value does not fit in 24 bits" > > My system is mpc755 VME board and the board boot well with down loaded > vxWorks image from my PC. > What's wrong with my BSP? > > thanks > > Dae-sop --------------------------- Newsgroups: comp.os.vxworks Subject: implementing clrscr() in vxworks Date: 11 May 2002 00:09:29 -0700 From: forvxworks@rediffmail.com (Anand) Organization: http://groups.google.com/ Message-ID: HI All, anybody knows how to implement clrscr() in vxworks as it seems that conio.h is not supported in vxworks. thanks in advance. regards. anand. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: getting the base address of the PCI card Date: 11 May 2002 00:13:29 -0700 From: forvxworks@rediffmail.com (Anand) Organization: http://groups.google.com/ Message-ID: References: <3cd10a55$1@brateggebdc5.br-automation.co.at> Thank You Mr.Werner Schiendl . Your reply was of a great help to me. regards. anand. "Werner Schiendl" wrote in message news:<3cd10a55$1@brateggebdc5.br-automation.co.at>... > Hi Anand, > > I assume you use some kind of pc486, pcPentium or similar BSP. > Within this BSPs, you find a file called sysEl3c90xEnd.c > This does exactly what you ask for for a network board. > Study the file to find out how to read the base address and size of the > address space as well as IRQ configuration from the PCI board. > > You can basically go with almost the same procedure, you just need to adapt > Vendor and Device IDs and adjust the memory configuration to the proper BARs > in the configuration. This is specific to your board, so you need to consult > the hardware configuration here. pciConfigShow and pciHeaderShow also help > with this. > > hth > Werner > > > "Anand" wrote in message > news:f14c07cb.0205012219.284d4f84@posting.google.com... > > HI All, > > first, thanks to Bruce,Andray Kaganovsky and John for helping me > > on Far Pointers. > > Now, i have a PCI card which has some memory and registers which > > i want to access and manipulate the contents of the registers > > depending on the functionality i want to program. > > Earlier , i was using an ISA card which had switches on it to > > select the base address and depending on the base address i could get > > the address of the registers and i could program the registers > > according to the function i wanted. > > but this PCI card does not have any switches, so that i can choose the > > base address. > > going through earlier articles in this group, i learned that we > > can use functions like pciDeviceShow() and pciHeaderShow() , we can > > have the details. but i am not sure whether the base address i get > > from pciHeaderShow() can be used as the base address of the card or > > not. I am new to vxworks, and i hope the question i am asking makes > > some sense. > > waiting for replies. > > Thanks in advance. > > Regards. > > Anand. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: problem launching the target server on host Date: 11 May 2002 00:19:55 -0700 From: forvxworks@rediffmail.com (Anand) Organization: http://groups.google.com/ Message-ID: References: <3cdbcf80.172834242@news.mhogaming.com> Thanks Mr.Ignacio Gonzalez T. we could solve the problem. Actually the config.h had some problem. we had to include #define INCLUDE_BSD and #undef INCLUDE_END. and also , we were including both END and BSD interfaces, i believe that gives some problem. regards. Anand. igtorque@yahoo.es (Ignacio Gonzalez T.) wrote in message news:<3cdbcf80.172834242@news.mhogaming.com>... > On 9 May 2002 23:54:50 -0700, forvxworks@rediffmail.com (Anand) wrote: > > >Hi All, > > i am configuring a vxWorks target for PENTIUM CPU using Network as > >the connection between host and target. > > I am able to boot the target in vxWorks using the bootable floppy > >i > >made by giving the appropriate bootline,IO address and the Interrupt > >level of the Ethernet Adapter. but while launching the target server > >on the host machine, with the backend "wdbrpc" , the server logconsole > >gives the following message.. > > > > Connecting to Target agent.... > > Error:rpccore backend client Timed out > > failed. > > > > > >On the target machine i am getting the massages as follows... > > > > > > > > muxDevLoad failed for device entry0! > > Attached TCP/IP interface to ene unit0 > > Attaching intarface lo0...done > > > > ........ > > ....... > > WDB : Ready > > -> > > > > > > the Ethernet Adapter i am using is .. > > > > SURECOM EP-325 PCI Ethernet Adapter. > > > > I am unable to the get what the problem is. Can anyone help?? > > > >Thanks in advance. > >Regards. > >Anand. > > Maybe your configNet.h is not correct. More precisely, the definition > of > END_TBL_ENTRY endDevTbl [] > may be missing something important related to the Ethernet driver. If > it were correct, you should expect, at worst, something like > muxDevLoad failed for device sep0! > instead of > muxDevLoad failed for device entry0! --------------------------- Newsgroups: comp.os.vxworks Subject: How to use AF_ROUTE ? Date: 11 May 2002 01:13:14 -0700 From: zw84611@sina.com (zhuwei) Organization: http://groups.google.com/ Message-ID: <2fc4ea3f.0205110013.4654e72f@posting.google.com> Hello,everyone!I have to translate the following Linux source codes to VxWorks: I have some problems: 1. what's the counterpart of AF_NETLINK(in linux) in VxWorks? 2. why I can't bind(),like following: ========================================================== linux source: ============= int rtnl_open() { struct sockaddr_nl rtaddr; int fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE); int rtaddr_len; if(fd <0) { perror("rtnl_open"); return -1; } rtaddr.nl_family = AF_NETLINK; rtaddr.nl_groups = 0xFFFFFFFF; if(bind(fd,(struct sockaddr*)&rtaddr,sizeof(rtaddr))<0) { perror("rtnl_open"); return -1; } rtaddr_len = sizeof(rtaddr); getsockname(fd,(struct sockaddr*)&rtaddr,&rtaddr_len); return fd; } =================================================================== I changed the above into following codes in VxWorks: - ------------------------------------------------------------------- #define AF_NETLINK AF_ROUTE int rtnl_open() { struct sockaddr_nl rtaddr; int fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE); /* vxWorks bootimage should include BSD Routing Sockets */ int rtaddr_len; if(fd <0) { perror("rtnl_open_error1\n"); return -1; } rtaddr.nl_family = AF_NETLINK; rtaddr.nl_groups = 0xFFFFFFFF; if(bind(fd,&rtaddr,sizeof(rtaddr))<0) { /* here,panic: raw_usrreq */ perror("rtnl_open_error2\n"); return -1; } if(bind(fd,(struct sockaddr*)&rtaddr,sizeof(rtaddr))<0) { /* here,panic: raw_usrreq */ perror("rtnl_open_error2\n"); return -1; } rtaddr_len = sizeof(rtaddr); getsockname(fd,(struct sockaddr*)&rtaddr,&rtaddr_len); return fd; } - ------------------------------------------------------------------- I have include the BSD Routing Sockets(INCLUDE_ROUTE_SOCK) in VxWorks image. But when the programme execute,there is an panic like: 0x1f6ba88 (tDbgTask): panic: raw_usrreq (when step into bind(fd,(struct sockaddr*)&rtaddr,sizeof(rtaddr))) I don't know why,can anyone help me? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Link error while working with SBC8260 Date: Sat, 11 May 2002 08:20:47 +0000 (UTC) From: "Graham Baxter" Organization: Graham Baxter (Software) Limited Message-ID: References: <5288f97d.0205100816.46bf3153@posting.google.com> Danny, Add the object motfccend.o to your project. It may be supplied as just an object file. I hope this helps. - -- Regards, Graham Baxter Freelance Software Engineer EMAIL: gbaxter@NOSPAMbcs.org.uk "Danny Cohen" wrote in message news:5288f97d.0205100816.46bf3153@posting.google.com... > Hi. > > I have evaluation board SBC8260. > I'm trying to create bootable image and download it to the board. > while compiling and linking the BSP i face the following error: > "motFccEndLoad undefined at PartialImage.o". My feeling it has > something with the BSP configuration. Can anyone help me to overcome > this problem? --------------------------- Newsgroups: comp.os.vxworks Subject: How do I do FTP on a non-default port using PORT and PASV Date: 11 May 2002 02:22:26 -0700 From: mohasin.zaki@wipro.com (Mohasin) Organization: http://groups.google.com/ Message-ID: Hi All, I need to perform FTP on non-default port. I have 2 requirements 1. Doing FTP through PORT command where I will specify the port on which the server can intiate data connection on receiving commands on the control connection. 2. Second is, I want to use PASV command where the server will reply with a IP,PORT number on which I can connect to. So when the remote ftp server receives a ftp command , the data transfer occurs on this server specified port. The problem I am facing is, 1. When I use VxWorks API ftpCommand() for issuing a PORT command I CANNOT intiate a data connection to the port which I have specified, using the ftpDataConnInit() and ftpDataConnGet(). 2. When issue a PASV command using ftpCommand(), all that I get in the reply is number '2' which Stands for FTP_COMPLETE. Bull the actual thing I want is the commplete 3 digit reply and the IP,PORT numbers from the reply. It seems that the vxWorks API somehow restricts the reply to the first digit from the actual reply. So, the vxWorks APIs dont seems to help. 1. Is there anyway I can still use the vxWorks API for acheiving the above. 2. If not, what is the alternative. That does mean that I have to implement my own FTP client in the vxWorks. Thanks for all your replies. Regards, MOhasin. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun May 12 03:33:51 2002 From: "veera" Date: Sun May 12 03:33:53 PDT 2002 Subject: Does VxWorks malloc() on MIPS returns byte aligned block ? This is a multi-part message in MIME format. ------=_NextPart_000_0021_01C1F9CE.BA1B1200 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Does VxWorks malloc() on MIPS returns byte aligned block ? ------=_NextPart_000_0021_01C1F9CE.BA1B1200 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

Does VxWorks malloc() on MIPS returns byte aligned block=20 ?

------=_NextPart_000_0021_01C1F9CE.BA1B1200-- From vxwexplo-errs@csg.lbl.gov Sun May 12 06:11:39 2002 From: "Sabarivel S" Date: Sun May 12 06:11:41 PDT 2002 Subject: FW: Vxworks: ENOBUFS in setsockopt (MULTICAST_IF) option This is a multi-part message in MIME format. ------=_NextPart_000_0018_01C1F9E4.BEFE96A0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit -----Original Message----- From: Sabarivel S [mailto:sabaris@future.futsoft.com] Sent: Sunday, 12 May 2002 1:36 PM To: 'vxgrp' Subject: FW: Vxworks: ENOBUFS in setsockopt (MULTICAST_IF) option hi, I face problem of ENOBUFS from setsockopt call after two hours of running. i went through the cases of the ENOBUFS issues in different cases. I tried out the case of increasing the mbuf size also but still we face the problem of ENOBUFS problem is there any other clue. I am sending output of mBufShow and netStackSysPoolShow along with this mail for your quick reference. and find the snapshot of the code which causes >>------------snapshot of code------------------------ /* Get the Id of the socket in which data needs to be sent out */ i4SocketId = gSockRegister[pParams->u1Proto].i4SocketId; DestAddr.sin_addr.s_addr = OSIX_HTONL(pParams->u4Dest); /* For multicast packet Interface needs to be specified */ if ( IP_IS_ADDR_CLASS_D (pParams->u4Dest) ) { IfAddr.s_addr = OSIX_HTONL(pParams->u4Src); if( setsockopt(i4SocketId, IPPROTO_IP,IP_MULTICAST_IF, (char *)&IfAddr, sizeof(IfAddr)) == -1) { perror("MULTICAST_IF : sockopt failed"); return IP_FAILURE; } } if (sendto (i4SocketId, pu1SendBuf, u2BufLen, 0, (struct sockaddr *) &DestAddr, sizeof (struct sockaddr_in)) < 0) { return IP_FAILURE; } >>------------snapshot of code------------------------ This portion of code sends PIM protocol packets in the RAW socket of PIM. When PIM control packets are sent to a multicast address it sets MULTICAST_IF option and sends. As it sends the control packets in different interfaces it sets this MULTICAST_IF option to send it on each interface It gives ENOBUFS error only after continuous run of 2 hours. Kindly advice me what could be done to solve this issue thanks sabari -----Original Message----- From: the vxWorks Users Group Exploder [mailto:vxwexplo@lbl.gov] Sent: Thursday, 2 May 2002 6:02 PM To: vxworks_users@csg.lbl.gov Subject: RE: Vxworks: ENOBUFS in setsockopt (MULTICAST_IF) option ------=_NextPart_000_0018_01C1F9E4.BEFE96A0 Content-Type: text/plain; name="ENOBUFSLog.txt" Content-Transfer-Encoding: quoted-printable Content-Disposition: attachment; filename="ENOBUFSLog.txt" [09:02:49] VxWorks [09:02:49]=20 [09:02:49] Copyright 1984-1998 Wind River Systems, Inc. [09:02:49]=20 [09:02:49] CPU: Motorola CBTSP ISB 8260 [09:02:50] VxWorks: 5.4 [09:02:50] BSP version: v1.42/dw20 Boot1 [09:02:50] Creation date: Not used [09:02:50] WDB: Ready. [09:02:50]=20 [09:02:50] Setting level trigger for IRQs [09:02:50] Creating filesystem. [09:20:12] -> mbufShow [09:22:53] type number [09:22:53] --------- ------ [09:22:53] FREE : 979 [09:22:53] DATA : 14 [09:22:53] HEADER : 0 [09:22:53] SOCKET : 0 [09:22:53] PCB : 0 [09:22:53] RTABLE : 0 [09:22:53] HTABLE : 0 [09:22:53] ATABLE : 0 [09:22:53] SONAME : 7 [09:22:53] ZOMBIE : 0 [09:22:53] SOOPTS : 0 [09:22:53] FTABLE : 0 [09:22:53] RIGHTS : 0 [09:22:53] IFADDR : 0 [09:22:53] CONTROL : 0 [09:22:53] OOBDATA : 0 [09:22:53] IPMOPTS : 0 [09:22:53] IPMADDR : 0 [09:22:53] IFMADDR : 0 [09:22:53] MRTABLE : 0 [09:22:53] TOTAL : 1000 [09:22:53] number of mbufs: 1000 [09:22:53] number of times failed to find space: 0 [09:22:53] number of times waited for space: 0 [09:22:53] number of times drained protocols for space: 0 [09:22:53] __________________ [09:22:53] CLUSTER POOL TABLE [09:22:53] = _________________________________________________________________________= ______ [09:22:53] size clusters free usage [09:22:53] = -------------------------------------------------------------------------= ------ [09:22:53] 64 100 99 2106 =20 [09:22:53] 128 100 92 30242 =20 [09:22:54] 256 350 350 340 =20 [09:22:54] 512 200 200 3 =20 [09:22:54] 1024 50 50 16 =20 [09:22:54] 2048 100 100 0 =20 [09:22:54] = -------------------------------------------------------------------------= ------ [09:22:54] value =3D 80 =3D 0x50 =3D 'P' [09:22:54] -> netStackSysPoolShow [09:22:59] type number [09:22:59] --------- ------ [09:22:59] FREE : 1597 [09:22:59] DATA : 0 [09:22:59] HEADER : 0 [09:22:59] SOCKET : 22 [09:22:59] PCB : 44 [09:22:59] RTABLE : 79 [09:22:59] HTABLE : 0 [09:23:00] ATABLE : 0 [09:23:00] SONAME : 0 [09:23:00] ZOMBIE : 0 [09:23:00] SOOPTS : 0 [09:23:00] FTABLE : 0 [09:23:00] RIGHTS : 0 [09:23:00] IFADDR : 16 [09:23:00] CONTROL : 0 [09:23:00] OOBDATA : 0 [09:23:00] IPMOPTS : 10 [09:23:00] IPMADDR : 24 [09:23:00] IFMADDR : 0 [09:23:00] MRTABLE : 0 [09:23:00] TOTAL : 1792 [09:23:00] number of mbufs: 1792 [09:23:00] number of times failed to find space: 0 [09:23:00] number of times waited for space: 0 [09:23:00] number of times drained protocols for space: 0 [09:23:00] __________________ [09:23:00] CLUSTER POOL TABLE [09:23:00] = _________________________________________________________________________= ______ [09:23:00] size clusters free usage [09:23:00] = -------------------------------------------------------------------------= ------ [09:23:00] 64 128 53 86 =20 [09:23:00] 128 256 210 465 =20 [09:23:00] 256 128 94 35 =20 [09:23:00] 512 384 362 441 =20 [09:23:00] = -------------------------------------------------------------------------= ------ [09:23:00] value =3D 80 =3D 0x50 =3D 'P' -> mbufShow [10:01:45] type number [10:01:45] --------- ------ [10:01:45] FREE : 1597 [10:01:45] DATA : 0 [10:01:45] HEADER : 0 [10:01:45] SOCKET : 22 [10:01:45] PCB : 44 [10:01:45] RTABLE : 79 [10:01:45] HTABLE : 0 [10:01:45] ATABLE : 0 [10:01:45] SONAME : 0 [10:01:45] ZOMBIE : 0 [10:01:45] SOOPTS : 0 [10:01:45] FTABLE : 0 [10:01:45] RIGHTS : 0 [10:01:45] IFADDR : 16 [10:01:45] CONTROL : 0 [10:01:45] OOBDATA : 0 [10:01:45] IPMOPTS : 10 [10:01:46] IPMADDR : 24 [10:01:46] IFMADDR : 0 [10:01:46] MRTABLE : 0 [10:01:46] TOTAL : 1792 [10:01:46] number of mbufs: 1792 [10:01:46] number of times failed to find space: 0 [10:01:46] number of times waited for space: 0 [10:01:46] number of times drained protocols for space: 0 [10:01:46] __________________ [10:01:46] CLUSTER POOL TABLE [10:01:46] = _________________________________________________________________________= ______ [10:01:46] size clusters free usage [10:01:46] = -------------------------------------------------------------------------= ------ [10:01:46] 64 128 53 86 =20 [10:01:46] 128 256 210 833 =20 [10:01:46] 256 128 94 35 =20 [10:01:46] 512 384 362 809 =20 [10:01:46] = -------------------------------------------------------------------------= ------ [10:01:57] type number [10:01:57] --------- ------ [10:01:57] FREE : 1000 [10:01:57] DATA : 0 [10:01:57] HEADER : 0 [10:01:57] SOCKET : 0 [10:01:57] PCB : 0 [10:01:57] RTABLE : 0 [10:01:57] HTABLE : 0 [10:01:57] ATABLE : 0 [10:01:57] SONAME : 0 [10:01:57] ZOMBIE : 0 [10:01:57] SOOPTS : 0 [10:01:57] FTABLE : 0 [10:01:57] RIGHTS : 0 [10:01:57] IFADDR : 0 [10:01:57] CONTROL : 0 [10:01:57] OOBDATA : 0 [10:01:57] IPMOPTS : 0 [10:01:57] IPMADDR : 0 [10:01:57] IFMADDR : 0 [10:01:57] MRTABLE : 0 [10:01:57] TOTAL : 1000 [10:01:57] number of mbufs: 1000 [10:01:57] number of times failed to find space: 0 [10:01:57] number of times waited for space: 0 [10:01:57] number of times drained protocols for space: 0 [10:01:57] __________________ [10:01:57] CLUSTER POOL TABLE [10:01:57] = _________________________________________________________________________= ______ [10:01:57] size clusters free usage [10:01:57] = -------------------------------------------------------------------------= ------ [10:01:57] 64 100 100 8740 =20 [10:01:58] 128 100 100 122869 =20 [10:01:58] 256 350 350 1022 =20 [10:01:58] 512 200 200 4 =20 [10:01:58] 1024 50 50 16 =20 [10:01:58] 2048 100 100 0 =20 [10:01:58] = -------------------------------------------------------------------------= ------ [10:52:48] type number [10:52:48] --------- ------ [10:52:48] FREE : 1597 [10:52:48] DATA : 0 [10:52:48] HEADER : 0 [10:52:48] SOCKET : 22 [10:52:48] PCB : 44 [10:52:48] RTABLE : 79 [10:52:48] HTABLE : 0 [10:52:48] ATABLE : 0 [10:52:48] SONAME : 0 [10:52:48] ZOMBIE : 0 [10:52:48] SOOPTS : 0 [10:52:48] FTABLE : 0 [10:52:48] RIGHTS : 0 [10:52:48] IFADDR : 16 [10:52:48] CONTROL : 0 [10:52:48] OOBDATA : 0 [10:52:48] IPMOPTS : 10 [10:52:48] IPMADDR : 24 [10:52:48] IFMADDR : 0 [10:52:48] MRTABLE : 0 [10:52:48] TOTAL : 1792 [10:52:48] number of mbufs: 1792 [10:52:48] number of times failed to find space: 0 [10:52:48] number of times waited for space: 0 [10:52:48] number of times drained protocols for space: 0 [10:52:48] __________________ [10:52:48] CLUSTER POOL TABLE [10:52:48] = _________________________________________________________________________= ______ [10:52:48] size clusters free usage [10:52:48] = -------------------------------------------------------------------------= ------ [10:52:49] 64 128 53 86 =20 [10:52:49] 128 256 210 1316 =20 [10:52:49] 256 128 94 35 =20 [10:52:49] 512 384 362 1292 =20 [10:52:49] = -------------------------------------------------------------------------= ------ [10:52:54] type number [10:52:54] --------- ------ [10:52:54] FREE : 1000 [10:52:54] DATA : 0 [10:52:54] HEADER : 0 [10:52:54] SOCKET : 0 [10:52:54] PCB : 0 [10:52:54] RTABLE : 0 [10:52:54] HTABLE : 0 [10:52:54] ATABLE : 0 [10:52:54] SONAME : 0 [10:52:54] ZOMBIE : 0 [10:52:54] SOOPTS : 0 [10:52:54] FTABLE : 0 [10:52:54] RIGHTS : 0 [10:52:54] IFADDR : 0 [10:52:54] CONTROL : 0 [10:52:54] OOBDATA : 0 [10:52:54] IPMOPTS : 0 [10:52:54] IPMADDR : 0 [10:52:54] IFMADDR : 0 [10:52:54] MRTABLE : 0 [10:52:54] TOTAL : 1000 [10:52:54] number of mbufs: 1000 [10:52:54] number of times failed to find space: 0 [10:52:54] number of times waited for space: 0 [10:52:54] number of times drained protocols for space: 0 [10:52:55] __________________ [10:52:55] CLUSTER POOL TABLE [10:52:55] = _________________________________________________________________________= ______ [10:52:55] size clusters free usage [10:52:55] = -------------------------------------------------------------------------= ------ [10:52:55] 64 100 99 17419 =20 [10:52:55] 128 100 92 244148 =20 [10:52:55] 256 350 350 1926 =20 [10:52:55] 512 200 200 6 =20 [10:52:55] 1024 50 50 16 =20 [10:52:55] 2048 100 100 0 =20 [10:52:55] = -------------------------------------------------------------------------= ------ [11:58:51] type number [11:58:51] --------- ------ [11:58:51] FREE : 1597 [11:58:51] DATA : 0 [11:58:51] HEADER : 0 [11:58:51] SOCKET : 22 [11:58:51] PCB : 44 [11:58:51] RTABLE : 79 [11:58:51] HTABLE : 0 [11:58:51] ATABLE : 0 [11:58:51] SONAME : 0 [11:58:51] ZOMBIE : 0 [11:58:51] SOOPTS : 0 [11:58:51] FTABLE : 0 [11:58:51] RIGHTS : 0 [11:58:51] IFADDR : 16 [11:58:51] CONTROL : 0 [11:58:51] OOBDATA : 0 [11:58:51] IPMOPTS : 10 [11:58:51] IPMADDR : 24 [11:58:51] IFMADDR : 0 [11:58:51] MRTABLE : 0 [11:58:51] TOTAL : 1792 [11:58:51] number of mbufs: 1792 [11:58:51] number of times failed to find space: 0 [11:58:52] number of times waited for space: 0 [11:58:52] number of times drained protocols for space: 0 [11:58:52] __________________ [11:58:52] CLUSTER POOL TABLE [11:58:52] = _________________________________________________________________________= ______ [11:58:52] size clusters free usage [11:58:52] = -------------------------------------------------------------------------= ------ [11:58:52] 64 128 53 86 =20 [11:58:52] 128 256 210 1942 =20 [11:58:52] 256 128 94 35 =20 [11:58:52] 512 384 362 1918 =20 [11:58:52] = -------------------------------------------------------------------------= ------ [11:59:02] type number [11:59:02] --------- ------ [11:59:02] FREE : 979 [11:59:02] DATA : 14 [11:59:02] HEADER : 0 [11:59:02] SOCKET : 0 [11:59:02] PCB : 0 [11:59:02] RTABLE : 0 [11:59:02] HTABLE : 0 [11:59:02] ATABLE : 0 [11:59:02] SONAME : 7 [11:59:02] ZOMBIE : 0 [11:59:02] SOOPTS : 0 [11:59:02] FTABLE : 0 [11:59:02] RIGHTS : 0 [11:59:02] IFADDR : 0 [11:59:02] CONTROL : 0 [11:59:02] OOBDATA : 0 [11:59:02] IPMOPTS : 0 [11:59:02] IPMADDR : 0 [11:59:02] IFMADDR : 0 [11:59:03] MRTABLE : 0 [11:59:03] TOTAL : 1000 [11:59:03] number of mbufs: 1000 [11:59:03] number of times failed to find space: 0 [11:59:03] number of times waited for space: 0 [11:59:03] number of times drained protocols for space: 0 [11:59:03] __________________ [11:59:03] CLUSTER POOL TABLE [11:59:03] = _________________________________________________________________________= ______ [11:59:03] size clusters free usage [11:59:03] = -------------------------------------------------------------------------= ------ [11:59:03] 64 100 99 28677 =20 [11:59:03] 128 100 92 401275 =20 [11:59:03] 256 350 350 3090 =20 [11:59:03] 512 200 200 8 =20 [11:59:03] 1024 50 50 16 =20 [11:59:03] 2048 100 100 0 =20 [11:59:03] = -------------------------------------------------------------------------= ------ [12:55:54] type number [12:55:54] --------- ------ [12:55:54] FREE : 1612 [12:55:54] DATA : 0 [12:55:54] HEADER : 0 [12:55:54] SOCKET : 22 [12:55:54] PCB : 44 [12:55:54] RTABLE : 64 [12:55:54] HTABLE : 0 [12:55:54] ATABLE : 0 [12:55:54] SONAME : 0 [12:55:54] ZOMBIE : 0 [12:55:54] SOOPTS : 0 [12:55:54] FTABLE : 0 [12:55:54] RIGHTS : 0 [12:55:54] IFADDR : 16 [12:55:54] CONTROL : 0 [12:55:54] OOBDATA : 0 [12:55:54] IPMOPTS : 10 [12:55:54] IPMADDR : 24 [12:55:54] IFMADDR : 0 [12:55:54] MRTABLE : 0 [12:55:54] TOTAL : 1792 [12:55:54] number of mbufs: 1792 [12:55:54] number of times failed to find space: 0 [12:55:54] number of times waited for space: 0 [12:55:54] number of times drained protocols for space: 0 [12:55:54] __________________ [12:55:54] CLUSTER POOL TABLE [12:55:54] = _________________________________________________________________________= ______ [12:55:55] size clusters free usage [12:55:55] = -------------------------------------------------------------------------= ------ [12:55:55] 64 128 61 92 =20 [12:55:55] 128 256 210 2502 =20 [12:55:55] 256 128 101 41 =20 [12:55:55] 512 384 362 2478 =20 [12:55:55] = -------------------------------------------------------------------------= ------ -> mbufShow [12:56:30] type number [12:56:30] --------- ------ [12:56:30] FREE : 987 [12:56:30] DATA : 6 [12:56:30] HEADER : 0 [12:56:30] SOCKET : 0 [12:56:30] PCB : 0 [12:56:30] RTABLE : 0 [12:56:30] HTABLE : 0 [12:56:30] ATABLE : 0 [12:56:30] SONAME : 7 [12:56:30] ZOMBIE : 0 [12:56:30] SOOPTS : 0 [12:56:30] FTABLE : 0 [12:56:30] RIGHTS : 0 [12:56:30] IFADDR : 0 [12:56:30] CONTROL : 0 [12:56:30] OOBDATA : 0 [12:56:30] IPMOPTS : 0 [12:56:30] IPMADDR : 0 [12:56:30] IFMADDR : 0 [12:56:30] MRTABLE : 0 [12:56:30] TOTAL : 1000 [12:56:30] number of mbufs: 1000 [12:56:31] number of times failed to find space: 4071 [12:56:31] number of times waited for space: 0 [12:56:31] number of times drained protocols for space: 196 [12:56:31] __________________ [12:56:31] CLUSTER POOL TABLE [12:56:31] = _________________________________________________________________________= ______ [12:56:31] size clusters free usage [12:56:31] = -------------------------------------------------------------------------= ------ [12:56:31] 64 100 100 38353 =20 [12:56:31] 128 100 100 538313 =20 [12:56:31] 256 350 350 4382 =20 [12:56:31] 512 200 200 30 =20 [12:56:31] 1024 50 50 17 =20 [12:56:31] 2048 100 100 0 =20 [12:56:31] = -------------------------------------------------------------------------= ------ [12:56:48] MULTICAST_IF : sockopt failed: S_errno_ENOBUFS [12:56:48]=20 [12:56:48] MULTICAST_IF : sockopt failed: S_errno_ENOBUFS [12:56:48]=20 [12:57:48] MULTICAST_IF : sockopt failed: S_errno_ENOBUFS ------=_NextPart_000_0018_01C1F9E4.BEFE96A0-- From vxwexplo-errs@csg.lbl.gov Mon May 13 04:03:11 2002 From: Vxworks Exploder Date: Mon May 13 04:03:14 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon May 13 04:03:07 PDT 2002 Subject: Re: Link error while working with SBC8260 Subject: Re: Interrupt Latency Subject: Re: Anyone using target Intel x86 (not PENTIUM) Subject: Does DOSFS 2.0 for PPC supports device whose block size != 512 ???? Subject: Re: implementing clrscr() in vxworks Subject: Re: where can I add the INCLUDE_POSIX_SCHED? Subject: Re: Difference between RTOS and normal OS Subject: Re: C++ comments in C file with GNU compiler Subject: Re: Seeking possible cause of "Can't load boot file" Subject: getting rtentry from SNMP IP Routing Table group ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Link error while working with SBC8260 Date: 12 May 2002 07:37:49 -0700 From: dannyc@harmonic.co.il (Danny Cohen) Organization: http://groups.google.com/ Message-ID: <5288f97d.0205120637.43a7ac1f@posting.google.com> References: <5288f97d.0205100816.46bf3153@posting.google.com> Thanks Graham. It is supplied as motfccend.obj and motfccend.h files with the BSP. How can you add it manually? which makefile i have to touch? How can yo add it from the Tornado IDE? "Graham Baxter" wrote in message news:... > Danny, > > Add the object motfccend.o to your project. > > It may be supplied as just an object file. > > I hope this helps. > > -- > > Regards, > > Graham Baxter > Freelance Software Engineer > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > "Danny Cohen" wrote in message > news:5288f97d.0205100816.46bf3153@posting.google.com... > > Hi. > > > > I have evaluation board SBC8260. > > I'm trying to create bootable image and download it to the board. > > while compiling and linking the BSP i face the following error: > > "motFccEndLoad undefined at PartialImage.o". My feeling it has > > something with the BSP configuration. Can anyone help me to overcome > > this problem? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Date: Sun, 12 May 2002 21:49:28 +0200 From: Peter Kannegiesser Organization: SEK Message-ID: References: Reply-To: peter.kannegiesser@gmx.de On Wed, 08 May 2002 16:50:43 GMT, "Leonid Rosenboim" wrote: >Since it dont seem like you got an answer, I'll offer my two cents: > >Unless of course, someone or something has instructed the CPU >to block interrupts. (Note: on x86 processors,l the CPU can either >block all or allow all interrupts as it has got only one INTR pin). > ...that's what I suspect. >What I am saying is that you need to find that bugger ;-) who put >tons of code between intLock() and intUnlock() calls, code that >runs for 5-15 milisec. intLock gets used in the kernel a few times, >but trust me, the VxWorks kernel or any other subsystem do not >do anything THAT STUPID. > >And when you find that guy, take his picture when he gets his >pink slip, you'll enjoy looking at it for years later ... ;-)) > You bet! Let's hope that it wasn't me :-) - -Peter - -- keine sig --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone using target Intel x86 (not PENTIUM) Date: 12 May 2002 21:54:48 -0700 From: heiko_elger@arburg.com (Heiko Elger) Organization: http://groups.google.com/ Message-ID: <97e6274e.0205122054.1e9a1799@posting.google.com> References: <97e6274e.0205080100.3e7e1d13@posting.google.com> <3cd93f65.4871064@news.mhogaming.com> Hello, gnu tool chain: 2.95.xx WRS thinks about to support 486 too. Best regards Heiko igtorque@yahoo.es (Ignacio Gonzalez T.) wrote in message news:<3cd93f65.4871064@news.mhogaming.com>... > On 8 May 2002 02:00:18 -0700, heiko_elger@arburg.com (Heiko Elger) > wrote: > > So we will have to stick to Tornado 2.0.2, even if someday we decide > to upgrade... My condolence. These news are bad news for us too. > > By the way, do you know if Tornado 2.2 comes with gcc > 2.7.2 ? --------------------------- Newsgroups: comp.os.vxworks Subject: Does DOSFS 2.0 for PPC supports device whose block size != 512 ???? Date: 12 May 2002 22:34:36 -0700 From: houcheng@ms17.hinet.net (houcheng) Organization: http://groups.google.com/ Message-ID: Dear vxworkers: Here is a question about DOSFS 2.0 for PPC: Is the DOSFS 2.0 support block device whose block size is not 512 bytes per block ? The following is my compact flash's initizliation code: /************************************************************************/ if(cFlashDrv()== OK) { printf("OK]\n"); } else { printf("FAIL]\n"); printf("Error: Pcmcia card is not in slot!\n"); return(ERROR); } pDev = cFlashDevCreate(); /* create a block device object */ pCbio = cbioDevVerify(pDev, FALSE); if(pCbio) printf("\rInitialize CBIO device driver ............................... [OK]\n"); else { printf("\rInitialize CBIO device driver ............................... [FAIL]\n"); return(ERROR); } ret = dosFsDevCreate(filename, pCbio, 30, DOS_CHK_REPAIR | DOS_CHK_VERB_SILENT ); if (ret == OK) printf("\rInitialize DOS File System .................................. [OK]\n"); else { printf("\rInitialize DOS File System .................................. [FAIL]\n"); return(ERROR); } return(OK); For the compact flash with 512-bytes block size, this initialization code runs well. But for the compact flash with 600 bytes block size, the cbioDevVerify(pDev, FALSE) returns NULL! Is the DOSFS 2.0 supports block size other than 512 bytes? I've tried the DOSFS patch SPR73141, which fixed the problem "dosFs2 is not handling boot sectors correctly on devices with sector sizes != 512 bytes." . But the results is still the same, no works on flash with 600 bytes block size. Any suggestion is welcome! Thanks in advanced! Lin Houcheng houcheng@turbonetworks.com.tw_No_SPAM --------------------------- Newsgroups: comp.os.vxworks Subject: Re: implementing clrscr() in vxworks Date: Mon, 13 May 2002 08:46:06 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@p5088eeea.dip.t-dialin.net "Anand" schrieb im Newsbeitrag news:f14c07cb.0205102309.794a06cd@posting.google.com... > HI All, > anybody knows how to implement clrscr() in vxworks as it seems that > conio.h is not supported in vxworks. > thanks in advance. > regards. > anand. Hi Anand, I've used void clrScr(void) { int sfd; sfd=open("/tyCo/0", O_RDWR, 0 ) fdprintf( sfd, "\033[2J" ); /* clear screen */ } for ANSI-compatible output device (Hyperterminal on WinNT) successfully. HTH Michael --------------------------- Newsgroups: comp.os.vxworks Subject: Re: where can I add the INCLUDE_POSIX_SCHED? Date: Mon, 13 May 2002 10:33:57 +0200 From: Alex Vinokur Message-ID: <3CDF7A75.9060002@bigfoot.com> References: Îâ¾ü wrote: > I have builded a downloadable application modules for VxWorks, when I > download it, there are some errors: > > Errors while downloading E:/Tornado/target/proj/Project0/SIMNTgnu/hello.out: > _retrun > _sched_getparam > _sched_setparam > > > Where can I add the macro INCLUDE_POSIX_SCHED? > > Thanks in advance. > > > > > In order to add some macro (for instance, INCLUDE_SHELL_BANNER) you should do as following : 1. Open "Workspace Window" 2. Select "VxWorks View" 3. Right Click on your Project-Name Popup menu is to be open. 4. Select "Find Object..." on popup menu "Find Object Window" is to be open. 5. On "Find Object Window" you should select : * Type = Component * Object = your-macro-name (for instance, INCLUDE_SHELL_BANNER) Click on "Find" Relevant component will be selected on 'Workspace Window'. For INCLUDE_SHELL_BANNER the selected component is : + development tool components + target shell components * shell banner 6. Right Click on selected component (for instance, "shell banner") on "Workspace Window". Popup menu is to be open. 7. Select "Include 'component name>'" (for instance "Include 'shell banner'". P.S. You can see all selected macro in generated Makefile. - -- ============================ Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ============================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Mon, 13 May 2002 09:56:12 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> Ok, Gary, You have a point here, and I agree, I have made a hybrid of embedded and real-time. Although I am quite old too (I use computers since 1980), the entient history and highly specialized areas of critical systems and aero-space where not considered in these thoughts. Regarding the system testing, and the Marsk fiasco, I agree partly - testing does not come instead of good design and code review. Still, the ability to test the system as a whole, and lack of software modifications in teh field, does make the embedded system orders of magnitude more testable then general purpose IY systems. Of course the discussion of how good is a test, is something else .. By the way, I recall that in the old days, a VAX/VMS real-time systems where sort of dedicated to a single application, and and RT did take away some of the inter-application protection measures (i.e. an RT task could easily crash the entire system), so in my mind, they where rather similar to embedded systems, disregarding their physical size. Lastly, I cannot resist noting the PBX case (e.g. DMS) as a very special case, where real-time actually co-exists with multi-application, and software changes are indeed made in the field all the time, ... "Gary M" wrote in message news:wwJD8.39282$RR3.57937@sccrnsc02... > > "Leonid Rosenboim" wrote in message > news:newscache$7lkzvg$uaa$1@lnews.actcom.co.il... > > Let me show you a different point of view: > > > > The design goals of a RTOs are quite different from a Generap Purpose OS > > "GPOS" like Unix, VMS and probably NT. > > > > In RTOS, it is assumed that the end-product is an embedded device, > > including hardware RTOS and higher-level software - all a single > > application, all tested together as a single unit. > > Therefore most safeguards that exist in GPOS are not really necesary, and > > are excluded in order to obtain the best possible performance on a given > > hardware platform, i.e. enable cost savings of the end-product. > > ALso, since all software is made and tested by a single entity, the > > resources (e.g. CPU, memory) is allocated to subsystems (tasks, modules) > as > > per programmer request. If prioritiesa are set wrong, surely enough this > > will be revealed during system testing. > > > > (rest snipped) > > I beg to strongly differ, but your assumption is way off. RTOS were > developed long before "embedded systems" as a concept had been invented, > long before the microprocessor was invented. I will bore you with some > examples, all driven by what we old people refer to as "big iron" machines: > 1. Factory floor automation > 2. Particle physics experiments > 3. Any kind of high speed, high bandwidth data sampling > 3. Nuclear reactors > 4. Radar and sonar systems > 5. Telecommunications infrastructure (PBXs, Central Office switchers, etc.) > 6. Aerospace ground control systems > > Digital Equipment Corporation developed VAX/VMS as a GPOS and an RTOS, and > the system could be tweaked to do either, or both. Many simulators in the > 80's were driven by VAX "minis" (I laugh at the term now), when > microprocessors were struggling with 16-bit limitations and slow I/O busses. > > Furthermore, try telling the Air Force, Boeing, or AirBus that "most > safeguards that exist in GPOS are not really necesary" in the RTOS in their > modern aircraft; in fact, they insist on tighter safeguards on the issues > most important to RTOS: predictability, reliability, redundancy, and > fault-tolerance. The fact that some project might not need some or all of > these is because the project is either trivial or it doesn't bother the > developer if the system is unreliable. Shame on him/her if the latter is > true. > > And don't assume that any problems "surely ... will be revealed during > system testing." I seem to recall a minor priority inversion flaw in a > certain Mars mission that wasn't discovered until the system failed several > million miles away from the lab. Just a straightforward VxWorks application, > was it not? > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C++ comments in C file with GNU compiler Date: 13 May 2002 07:30:40 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "Haim Tebeka" wrote: > Is there any compilation switch that one can use so that the GNU > compiler will not spit a syntax error whenever it sees a C++ (//) > style comment in a C file? See http://www.xs4all.nl/~borkhuis/vxworks/vxw_pt1.html#1.1-N Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Seeking possible cause of "Can't load boot file" Date: 13 May 2002 07:36:53 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: cozzy@263.net (yyzhong) wrote: > I boot a x86 target through network. > It displays: > Attached TCP/IP interface to elt0 > Attaching network interface lo0...done. > Loading... > > After a long period of time, it continues with: > > Error loading file: errno = 0xd0003. > Cannot load boot file! /*The last prompt maynot be the original text*/ I think you are booting using FTP. If this is the case check the logs from your FTP server to see if the connection is made. Otherwise check the boot parameters (IP addresses, netmask, gateway). > I think that if I boot it from floppy disk with the network card > driver included, I can use the network card also. But the result of > "ifShow" is "undefined symbol: ifShow" . You have to include the symbol table and to initialise it to get this to work. This can be done using the Tornado project facility. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: getting rtentry from SNMP IP Routing Table group Date: 13 May 2002 03:58:31 -0700 From: vijay.peshkar@wipro.com (Vijay Peshkar) Organization: http://groups.google.com/ Message-ID: Hello Friends, Am porting net-snmp to vxworks. m2IpRouteTblEntryGet() gives the elements of struct M2_IPROUTETBL - SNMP IP Routing Table group. I need to get elements of struct rtentry(route.h) from the above table. Any suggestions?. Am unable to find any routine to 'get' the routing table entry based on an index/addr etc., - --------------------- typedef struct { unsigned long ipRouteDest; long ipRouteIfIndex; long ipRouteMetric1; long ipRouteMetric2; long ipRouteMetric3; long ipRouteMetric4; unsigned long ipRouteNextHop; long ipRouteType; long ipRouteProto; long ipRouteAge; unsigned long ipRouteMask; long ipRouteMetric5; M2_OBJECTID ipRouteInfo; } M2_IPROUTETBL; - ------------------------- struct rtentry { struct radix_node rt_nodes[2]; #define rt_key(r) ((struct sockaddr *)((r)->rt_nodes->rn_key)) #define rt_mask(r) ((struct sockaddr *)((r)->rt_nodes->rn_mask)) struct sockaddr *rt_gateway; short rt_flags; short rt_refcnt; u_long rt_use; struct ifnet *rt_ifp; struct ifaddr *rt_ifa; struct sockaddr *rt_genmask; caddr_t rt_llinfo; struct rtentry *rt_gwroute; int rt_mod; struct rt_metrics rt_rmx; u_char rt_level; u_char rt_state; u_long rt_area; }; - -------------------------- Thanks and Regards, Vijay --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue May 14 04:03:14 2002 From: Vxworks Exploder Date: Tue May 14 04:03:16 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue May 14 04:03:10 PDT 2002 Subject: Re: Polymorphism and Destructor Subject: "target system is not initialized" error message in shell? Subject: RAM_LOW_ADRS parameter in T2 Subject: Raw file system reg Subject: Re: Polymorphism and Destructor Subject: Re: Difference between RTOS and normal OS Subject: Re: Difference between RTOS and normal OS Subject: Block Address Translation Performance Subject: Re: Difference between RTOS and normal OS Subject: Re: Default route outside subnet Subject: Re: Difference between RTOS and normal OS Subject: Re: Difference between RTOS and normal OS Subject: Is there a globally accessable, char **environ, pointer to a list of environment variables on vxworks? Subject: TCP_NODELAY option doesn't work Subject: ioctl problem for VxWorks Subject: equal cost multipath Subject: Re: Block Address Translation Performance Subject: Re: implementing clrscr() in vxworks Subject: process info Subject: Re: Block Address Translation Performance Subject: Re: implementing clrscr() in vxworks Subject: PS2 Port on MVME1604. Subject: Re: Is there a globally accessable, char **environ, pointer to a list of environment variables on vxworks? Subject: Re: Default route outside subnet Subject: Re: Is there a globally accessable, char **environ, pointer to a list of environment variables on vxworks? Subject: Help! a question about etherOutput ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Polymorphism and Destructor Date: 13 May 2002 05:29:50 -0700 From: pboulay@hotmail.com (Patrick Boulay) Organization: http://groups.google.com/ Message-ID: <8bb5523.0205130429.5b8c4ed7@posting.google.com> References: <8bb5523.0205121836.5b8b1432@posting.google.com> Hello, > try > A *ptr = new B; I'll try, but I cant do that everytime... The example I wrote it's just a little test, but in reallity, I have a big system and I have a class attribut name like: A* ptr; and when I receive a connection from the socket, I use "new B;" or "new ..." a lot of possibility.. so I really need to do: A* ptr; ptr = new B; > > > "Patrick Boulay" wrote in message > news:8bb5523.0205121836.5b8b1432@posting.google.com... > > Hello, > > > > I have a big problem with polymorphism and destructor. See the class > > in example bellow, the vxworks crash after I delete the pointer. > > > > class A > > { > > public: > > A(); > > virtual ~A(); > > }; > > class B:public A > > { > > public: > > B(); > > virtual ~B(); > > }; > > int aFunction() > > { > > A *ptr; > > > > ptr = new B; > > > > delete ptr; > > > > return 0; > > } > > > > > > > > Thanks > > Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: "target system is not initialized" error message in shell? Date: 13 May 2002 06:30:44 -0700 From: burphy@iim.nctu.edu.tw (Burphy) Organization: http://groups.google.com/ Message-ID: <7f6d76f4.0205130530.72d24d88@posting.google.com> I launched the target server and windshell in my PC successfully, but when I type the command like i or w, it shows the following error message: - -> i The target system is not initialized yet. Can't proceed. value = -1 = 0xffffffff Thank you. --------------------------- Newsgroups: comp.os.vxworks Subject: RAM_LOW_ADRS parameter in T2 Date: Mon, 13 May 2002 15:26:10 +0200 From: Sebastien Fauris Organization: Nortel Networks Message-ID: <3CDFBEF2.D70397B@europem01.nt.com> Hi VxWorkers, I try to change the value of the RAM_LOW_ADRS parameter used in the generation of a vxWorks kernel (not a bootrom). When I do a "find object" - "parameter" - "RAM_LOW_ADRS" in the Tornado 2.0 graphical interface, I get an error message saying "item 'RAM_LOW_ADRS' not found" and this in spite of the fact that I could select it in the object parameter list. I set this parameter properly in my config.h file but in the Makefile generated by Tornado, it keeps this parameter to 0x100000. I have to modify by hand the generated Makefile in the following section to set it to 0x10000 as I want: ifeq ($(BUILD_SPEC),common) AR = arppc AS = ccppc CC = ccppc .../... NM = nmppc OPTION_DEFINE_MACRO = -D OPTION_INCLUDE_DIR = -I POST_BUILD_RULE = RAM_HIGH_ADRS = 00500000 # RAM text/data address RAM_LOW_ADRS = 00010000 # RAM text/data address SIZE = sizeppc UPPER_LIMIT = 400000 USR_ENTRY = usrInit endif Where does Tornado find the default value for the RAM_LOW_ADRS parameter when it generates its Makefile ? How can I can make it use the value defined in the config.h file ? Sebastien. --------------------------- Newsgroups: comp.os.vxworks Subject: Raw file system reg Date: 13 May 2002 08:29:34 -0700 From: asmita.joshi@wipro.com (Asmita Joshi) Organization: http://groups.google.com/ Message-ID: Hi Friends, I want to create two ram disk devices one with dosFs & another partition with rawFile system. My problem is that , when I use the rawFsDevInit, it returns a non-null value. However if I try to create any files in that raw fs partition, it returns an error. The following is the log of the stuff which I tried from the shell. WDB: Ready. - -> ramDevCreate(0,1024,1024,1024,0) value = 98555804 = 0x5dfd79c - -> rawFsDevInit("/rawWip",0x5dfd79c) value = 98555744 = 0x5dfd760 - -> devs drv name 0 /null 1 /tyCo/0 1 /tyCo/1 7 : 8 /pty/telnet.S 9 /pty/telnet.M 10 /vio 3 /wip 4 /rawWip value = 25 = 0x19 - -> cd "/rawWip" value = 0 = 0x0 - -> ls Can't open ".". value = -1 = 0xffffffff = _func_taskRegsShowRtn + 0xfbb030eb - -> creat("/rawWip/Hello.txt",2) value = -1 = 0xffffffff = _func_taskRegsShowRtn + 0xfbb030eb - -> Could anyone plz mail the reason for the Raw fs failure. My nect problem is I am working with a flash driver which implements the dosFs on flash. I find that the read & write handlers receives only 2 blocks of data at a time. I have declared the block size as 512 bytes. In one shot I receive only 1024 bytes of data to be written to the flash. This happens independent of the file size which has to be written to flash. This drastically slows down the system. I would like to know if I can dictate the vxWorks dosFs to transfer the file to driver with chunk size much higher than this 1024bytes. Plz mail me regarding the dosFs transfer block size. Thanks Asmitha --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Polymorphism and Destructor Date: 13 May 2002 08:54:56 -0700 From: pboulay@hotmail.com (Patrick Boulay) Organization: http://groups.google.com/ Message-ID: <8bb5523.0205130754.71df76b0@posting.google.com> References: <8bb5523.0205121836.5b8b1432@posting.google.com> I did it, I didn't show you the implementation here because the code would be too long for nothing.. I just put cout in the destructors and constructors. "Alex Vinokur" wrote in message news:... > "Patrick Boulay" wrote in message news:8bb5523.0205121836.5b8b1432@posting.google.com... > | Hello, > | > | I have a big problem with polymorphism and destructor. See the class > | in example bellow, the vxworks crash after I delete the pointer. > | > > You should implement constructors and destructors. > > | class A > | { > | public: > | A(); > | virtual ~A(); > | }; > | class B:public A > | { > | public: > | B(); > | virtual ~B(); > | }; > | int aFunction() > | { > | A *ptr; > | > | ptr = new B; > | > | delete ptr; > | > | return 0; > | } > | > | > | > | Thanks > | Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Mon, 13 May 2002 10:06:30 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> Mars pathfinder scenario was due to priority inversion problem. It is detailed in http://www-2.cs.cmu.edu/afs/cs.cmu.edu/project/art-6/www/mars.html Also, I have never seen someone try to use a VAX machine running VMS to do realtime control. Most of us used PDP-11 running RT11 or other custom OS. VMS was not realtime, at least not to me. But realtime is a relative term. Similarly, one could argue (successfully) that DOS is realtime for a class of problems. :-) - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Mon, 13 May 2002 17:31:33 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> I know of VAX/VMS systems doing "sort of" real time work, where deterministic performance in the 2-5 millisecond range is required,, and indeed there are real-time systems, some with less then 10 microsecond response time based on MS-DOS. The good thing about DOS is that it does not interfere if you avoid calling it ;-) "Hwa Jin Bae" wrote in message news:udvsqup2dc1b84@corp.supernews.com... > Mars pathfinder scenario was due to priority inversion problem. It is > detailed in > > http://www-2.cs.cmu.edu/afs/cs.cmu.edu/project/art-6/www/mars.html > > Also, I have never seen someone try to use a VAX machine running VMS to do > realtime control. Most of us used PDP-11 running RT11 or other custom OS. > VMS was not realtime, at least not to me. But realtime is a relative term. > Similarly, one could argue (successfully) that DOS is realtime for a class > of problems. > > :-) > > -- > Hwa Jin Bae > mailto:hjb@pso.com > http://www.pso.com > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Block Address Translation Performance Date: Mon, 13 May 2002 13:47:42 -0500 From: James Organization: Raytheon Company Message-ID: <3CE00A4E.C3A20F72@nospam.com> Hello,

I'm using a MVME2700 single board computer from Motorola.  I have 4gig of VME A32 addressable memory which is mapped using BAT registers.  Has anyone experienced poor performance the first time through a section of code?  For example, the first time through a function may take 118us, whereas future accesses through it only takes 82us.  The code does not dynamically allocate any new memory via malloc.  I'm guessing that the translation look-ahead buffers (TLB) on the MVME2700 PPC are remembering the previously accessed locations, thereby decreasing future VME A32 access times.  If this is so, how can I estimate when the TLB's will forget or lose this information, requiring another performance hit?

If anybody has had similar experiences with the BAT registers and has some ideas on how to increase VME A32 memory performance, please respond.

Thanks,
James --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Mon, 13 May 2002 19:55:02 GMT From: "doug dotson" Organization: Verio Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> I think there was a real-time version of VMS called something like VAXELN or something like that. doug "Leonid Rosenboim" wrote in message news:newscache$l9b2wg$uld$1@lnews.actcom.co.il... > > I know of VAX/VMS systems doing "sort of" real time work, > where deterministic performance in the 2-5 millisecond range > is required,, and indeed there are real-time systems, some with > less then 10 microsecond response time based on MS-DOS. > The good thing about DOS is that it does not interfere if you > avoid calling it ;-) > > "Hwa Jin Bae" wrote in message > news:udvsqup2dc1b84@corp.supernews.com... > > Mars pathfinder scenario was due to priority inversion problem. It is > > detailed in > > > > http://www-2.cs.cmu.edu/afs/cs.cmu.edu/project/art-6/www/mars.html > > > > Also, I have never seen someone try to use a VAX machine running VMS to do > > realtime control. Most of us used PDP-11 running RT11 or other custom OS. > > VMS was not realtime, at least not to me. But realtime is a relative > term. > > Similarly, one could argue (successfully) that DOS is realtime for a class > > of problems. > > > > :-) > > > > -- > > Hwa Jin Bae > > mailto:hjb@pso.com > > http://www.pso.com > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Default route outside subnet Date: Mon, 13 May 2002 16:27:37 -0400 From: "RonnoBonno" Message-ID: References: Have you tried, routeAdd "0.0.0.0","IP Addr of the Gateway" the 0.0.0.0 route is the default gateway. Good Luck Ron Naismith Dave Moore wrote in message ... >Johan Borkhuis wrote in message news:... >> I have the following problem. I have a device that gets an IP address and >> subnetmask, but also a gateway. These are fixed values. This gateway >> address however is outside of the subnet. So for example: >> >> IP address = 202.204.206.208:ffffff00 >> Gateway = 175.185.195.205 >> >> How can I configure the routing to do this? >> >> Groeten, >> Johan > >Your details are somewhat "sketchy", but you obviously have a >misconfiguration problem. I am not sure what you mean by "fixed values". >If they are compiled into the bootline, then fix them! If your "device" >gets this information via DHCP then fix the configuration of the DHCP >server. Bottom line is that to send an IP datagram off the local subnet >you MUST go thru a router that has an interface on the subnet as your >device. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Mon, 13 May 2002 14:35:46 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> The DEC sales pitch used to be that VAXELN was for realtime on VAX while VMS (and later, Ultrix which is basically a BSD Unix) was general purpose. But then, many people believe (correctly) that it is just fine to use Windows NT for realtime work (of certain kinds). RT-11 (RSX+), TECO, MACRO-11... those were the days. - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Mon, 13 May 2002 23:24:58 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CE04BA5.3000904@spamm.me.l8s.co.uk> References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> > RT-11 (RSX+), TECO, MACRO-11... those were the days. Yes - we though about a basic interpreter written for TECO, we decided not to try because the available arithmetic was unsigned :-) I've still got the printout of the command TECO used to parse it's command line. It fills most of a sheet of compouter paper (66x132)! It had a synonym of 'make' - and if you typed 'make love' came back with 'not war' before starting the editor... (so when did 'make' cange from creating a new file to running the David --------------------------- Newsgroups: comp.os.vxworks Subject: Is there a globally accessable, char **environ, pointer to a list of environment variables on vxworks? Date: 13 May 2002 16:25:20 -0700 From: ctsai@mc.com (Chung Tsai) Organization: http://groups.google.com/ Message-ID: There is char **environ, and char **_environ on solaris or unix and NT respectively. Is there a equivalent one on vxworks? These char **environ or char **_environ is a pointer to a array of pointer to the environment variable. Thanks in advance. Chung --------------------------- Newsgroups: comp.os.vxworks Subject: TCP_NODELAY option doesn't work Date: 13 May 2002 17:03:21 -0700 From: vprashant@illinoisalumni.org (V Prashant) Organization: http://groups.google.com/ Message-ID: <25050bb6.0205131603.1370c268@posting.google.com> I have been trying to disable Nagle's algorithm by turning the TCP_NODELAY option on. However, this seems to have no effect and snooping on the ethernet packets indicates that the sender is waiting for outstanding ack's before sending the packets it has available. Any idea on why this could be happening? All my traffic is Request/Response type. Thanks Prashant --------------------------- Newsgroups: comp.os.vxworks Subject: ioctl problem for VxWorks Date: 13 May 2002 21:30:30 -0700 From: khurana.r@dcmtech.co.in (Rakesh) Organization: http://groups.google.com/ Message-ID: <414bb87b.0205132030.7f452eb8@posting.google.com> Hi all I am facing the problem of firing the ioctl command of FIOSYNC on dos file system. whwnever i open a file using the open command aand try to fire the ioctl on the fd returned i receive an unknown request error. can anybody suggest me the solution. i am using the PcPentium as my target --------------------------- Newsgroups: comp.os.vxworks Subject: equal cost multipath Date: 13 May 2002 22:42:13 -0700 From: sai_dattathrani@infy.com (sai) Organization: http://groups.google.com/ Message-ID: <8dff11cd.0205132142.3cab07c7@posting.google.com> Hi all, Sometime back there was a posting on this group telling that VxWorks does not support equal cost multipath. Is there any way of enabling equal cost multipath in VxWorks? I think this should be possible since many routers use VxWorks. Please let me know the method for enabling equal cost multipath support in VxWorks. rgds sai --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Block Address Translation Performance Date: Tue, 14 May 2002 06:14:46 GMT From: "dgold1" Organization: RoadRunner - West Message-ID: References: <3CE00A4E.C3A20F72@nospam.com> <3CE01AD4.27A9F051@nospam.com> James wrote in message <3CE01AD4.27A9F051@nospam.com>... Only 1GB of VME A32 memory is mapped via the BAT registers... James wrote: I'm using a MVME2700 single board computer from Motorola. I have 4gig of VME A32 addressable memory which is mapped using BAT registers. Has anyone experienced poor performance the first time through a section of code? For example, the first time through a function may take 118us, whereas future accesses through it only takes 82us. The code does not dynamically allocate any new memory via malloc. I'm guessing that the translation look-ahead buffers (TLB) on the MVME2700 PPC are remembering the previously accessed locations, thereby decreasing future VME A32 access times. If this is so, how can I estimate when the TLB's will forget or lose this information, requiring another performance hit? If anybody has had similar experiences with the BAT registers and has some ideas on how to increase VME A32 memory performance, please respond. The TLBs are not consulted (i.e. used) when you use the BATs, for that particular memory region. (See the Green Book for more info). If you map other areas with page table entries, the TLBs will be used for them. But TLBs are never used for areas mapped by BATs. The code may not dynamically allocate new memory, but are there memory locations (i.e. global variables, local vars on the stack) that may not be accessed until you execute the code the first time ? I'm not completely familiar with the MVME2700. Is the 1G of A32 on board or across the VME ? Are there other bus masters that may invalidate your cache ? If it's off-board memory, are you competing with other masters for it ? hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: implementing clrscr() in vxworks Date: 13 May 2002 23:14:49 -0700 From: forvxworks@rediffmail.com (Anand) Organization: http://groups.google.com/ Message-ID: References: Hi Michael, i have some doubts in the code u have given. Does /tyco refer to the monitor and also what does the string "\033[2J" mean. does it have some particular value?? Thank You, Regards. Anand. "Michael Lawnick" wrote in message news:... > "Anand" schrieb im Newsbeitrag > news:f14c07cb.0205102309.794a06cd@posting.google.com... > > HI All, > > anybody knows how to implement clrscr() in vxworks as it seems that > > conio.h is not supported in vxworks. > > thanks in advance. > > regards. > > anand. > > Hi Anand, > > I've used > > void clrScr(void) > { > int sfd; > > sfd=open("/tyCo/0", O_RDWR, 0 ) > fdprintf( sfd, "\033[2J" ); /* clear screen */ > } > > for ANSI-compatible output device (Hyperterminal on WinNT) successfully. > > HTH > Michael --------------------------- Newsgroups: comp.os.vxworks Subject: process info Date: 14 May 2002 07:36:59 +0100 From: Ole Asbjørn Fadum Message-ID: In Tornado we have the w() function, which tells about all processes and what they are waiting for. This is a tcl script and only works from inside Tornado. Does it exist a similar c-program which could be compiled into my target sw? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Block Address Translation Performance Date: Tue, 14 May 2002 08:37:40 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <3CE00A4E.C3A20F72@nospam.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088f641.dip.t-dialin.net Hi James, "James" schrieb im Newsbeitrag news:3CE00A4E.C3A20F72@nospam.com... > Hello, > I'm using a MVME2700 single board computer from Motorola. I have 4gig of VME A32 addressable memory which is mapped using BAT registers. Has anyone experienced poor performance the first time through a section of code? For example, the first time through a function may take 118us, whereas future accesses through it only takes 82us. The code does not dynamically allocate any new memory via malloc. I'm guessing that the translation look-ahead buffers (TLB) on the MVME2700 PPC are remembering the previously accessed locations, thereby decreasing future VME A32 access times. If this is so, how can I estimate when the TLB's will forget or lose this information, requiring another performance hit? > > If anybody has had similar experiences with the BAT registers and has some ideas on how to increase VME A32 memory performance, please respond. > BAT superceeds TLB, i.e. a hit in BAT leads to never consulting TLB (->PPC604 docu). Your 'problem' might be caused by cache: The first time your code is executed, it has to loaded into instruction cache. You may test the delay by commenting out the VME access and messure again. Take care: When writing to VME, you don't get the actual time of VME access, as your PCI2VME (Tundra Universe) does posted writes in basic setup. Try reads to get a feeling of real world. HTH Michael --------------------------- Newsgroups: comp.os.vxworks Subject: Re: implementing clrscr() in vxworks Date: Tue, 14 May 2002 08:52:27 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@p5088f641.dip.t-dialin.net Hi Anand, just read more carefully: - - its /tyCo, not /tyco, your serial line interface in vxWorks. -> I had connected a terminal to my target. - - "\033[2J" is the ANSI-escape-sequence for 'clear screen'. It should work on ANSI compliant console drivers too. HTH Michael "Anand" schrieb im Newsbeitrag news:f14c07cb.0205132214.1d44962a@posting.google.com... > Hi Michael, > i have some doubts in the code u have given. Does /tyco refer to > the monitor and also what does the string "\033[2J" mean. does it have > some particular value?? > Thank You, > Regards. > Anand. > > "Michael Lawnick" wrote in message news:... > > "Anand" schrieb im Newsbeitrag > > news:f14c07cb.0205102309.794a06cd@posting.google.com... > > > HI All, > > > anybody knows how to implement clrscr() in vxworks as it seems that > > > conio.h is not supported in vxworks. > > > thanks in advance. > > > regards. > > > anand. > > > > Hi Anand, > > > > I've used > > > > void clrScr(void) > > { > > int sfd; > > > > sfd=open("/tyCo/0", O_RDWR, 0 ) > > fdprintf( sfd, "\033[2J" ); /* clear screen */ > > } > > > > for ANSI-compatible output device (Hyperterminal on WinNT) successfully. > > > > HTH > > Michael --------------------------- Newsgroups: comp.os.vxworks Subject: PS2 Port on MVME1604. Date: Tue, 14 May 2002 08:49:58 +0100 From: An Engineer Message-ID: <3CE0C1A6.50092F63@here.com> I would like to read the output of a PS2 device attached to the PS2 port of a Power PC (MVME1604). I however, cannot find any information on how to access the PS2 port of the MVME1604. Has anyone had any experiance of reading from the PS2 port of a Power PC? Do I simply need to read from a certain port number as I would do if reading from a serial port? Thanks An Engineer. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is there a globally accessable, char **environ, pointer to a list of environment variables on vxworks? Date: Tue, 14 May 2002 10:25:13 +0200 From: "Werner Schiendl" Message-ID: <3ce0c9df$1@brateggebdc5.br-automation.co.at> References: Hi, try: #include extern char **ppGlobalEnviron; then you have a macro environ, that you can use like the char **environ on *nix systems. hth Werner "Chung Tsai" wrote in message news:f0b3c85f.0205131525.6975ac28@posting.google.com... > There is char **environ, and char **_environ on solaris or unix and NT > respectively. Is there a equivalent one on vxworks? > > These char **environ or char **_environ is a pointer to a array of > pointer to > the environment variable. > > Thanks in advance. > > Chung --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Default route outside subnet Date: 14 May 2002 08:26:52 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "RonnoBonno" wrote: > Have you tried, > > routeAdd "0.0.0.0","IP Addr of the Gateway" > > the 0.0.0.0 route is the default gateway. Yes, but the problem is that the address of the gateway is outside the subnet. I managed to get it to work (sort of). After I received the data from the DHCP server I had to implement the following sequence: ifAddrAdd ("dp0",); routeAdd ("0.0.0.0",); arpAdd (,,0); For some reason the ARP does not work correctly on the second IP address, so I had to add the entry manually. This also gives problems with the DNS server. When I ask for an address the query is sent out on the second IP address, as this is the address of the DNS resolver in the box where the DHCP server is also located. But that box sends out an ARP, but this ARP is not answered by the VxWorks box. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is there a globally accessable, char **environ, pointer to a list of environment variables on vxworks? Date: 14 May 2002 08:32:17 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: ctsai@mc.com (Chung Tsai) wrote: > There is char **environ, and char **_environ on solaris or unix and NT > respectively. Is there a equivalent one on vxworks? > > These char **environ or char **_environ is a pointer to a array of > pointer to the environment variable. see the manual page for envLib. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Help! a question about etherOutput Date: 14 May 2002 02:12:31 -0700 From: chwin@china.com (Adom) Organization: http://groups.google.com/ Message-ID: <92dbccc1.0205140112.2468ec07@posting.google.com> Hi, i'm developing a test program for a device production. Envirment: Intel Board, PIII CPU, 2 82557 Network Interface(fei0 and fei1) i use etherOutput periodic to send packet with size of 100 from fei0 to fei1 directly in order to test the hardware. At the begining, when the packet number is small, all is ok and i can use sniffer to receive all packet but when i send 300 packets once , the sniffer only receives more 150 packet , then the fei0 is unreachable when i ping it. when i use mbufShow to see the mbuf numbers , the size 128 clusters reduces from 100 to 50. So i ipDetch on fei0 then ipAttach on it ，fei0 resume but the mbuf does not. What happened? Anyone can help me ? Thanks! ========== Attach Source Code: void test (int count) { struct ifnet *pIfNet0, *pIfNet1; struct in_addr srcIp, dstIp; struct ether_header eh; /* ethernet header */ char pSrcIp[18] = {0}; char pDstIp[18] = {0}; char pData[100] = {0}; int i ; i = (count<=0)?1:count; pIfNet0 = ifunit ("fei0"); //get ifnet pointer pIfNet1 = ifunit ("fei1"); ifAddrGet ("fei0", pSrcIp); //get net ip address ifAddrGet ("fei1", pDstIp); inet_aton (pSrcIp, &srcIp); inet_aton (pDstIp, &dstIp); //init a ip header ipHeaderCreate (1, &srcIp, &dstIp, (struct ip*)pData, sizeof (pData)); bzero ((char *) &eh, sizeof (struct ether_header)); //get dest mac addr if (etherAddrResolve (pIfNet1, pDstIp, (char *) eh.ether_dhost, 5, 2 * sysClkRateGet()) == ERROR) return ; eh.ether_type = ETHERTYPE_IP; /* htons is in ether_output */ //send do { if (etherOutput (pIfNet0, &eh, (char *) pData, sizeof (pData)) != OK) { logMsg ("Ether Send Packet Error\n", 0, 0, 0, 0, 0, 0); } } while (--i); } --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 15 00:41:09 2002 From: "Surinder Singh" Date: Wed May 15 00:41:11 PDT 2002 Subject: Embedded Engineer This is a multi-part message in MIME format. ------=_NextPart_000_0001_01C1FBB7.246636F0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Please forward the updated resume attached as a word document and also mention the expected salary, availability and employment status (Visa, Green Card, US Citizen). These are FULL TIME positions in Bay Area - California, Japan and Bangalore - India. Please mention the location you are applying for. 1) Sr. Software Engineer / Technical Lead 4 to 10 years Exp. 2) Technical Lead/Manager level 5 to 10 years experience. - Extremely good in C, C++, VC++, COM/DCOM - Good in Real Time Systems - Good in Data Structures/ Algorithms - Motorola 6800 Assembly Language - Java experience is a plus - Preferably Computer Science Engineers Need to be good in Windows NT 4.0 worked in RTOS like VxWorks or pSOS It would be an advantage if you know GEM and SECS (these are inter-tool communication standards in semi conductor manufacturing industry). Thank you so very much for your time and I look forward to hearing and working with you soon. ------=_NextPart_000_0001_01C1FBB7.246636F0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

Please forward the updated resume attached = as a word=20 document and also mention the = expected=20 salary, availability and employment status (Visa, Green Card, US=20 Citizen). These are = FULL=20 TIME positions in Bay Area - = California,=20 Japan and Bangalore - India. = Please=20 mention the location you are applying=20 for.
 
1) Sr. Software Engineer / = Technical Lead=20 4 to 10 years Exp.
 
2) Technical Lead/Manager level = 5 to 10=20 years experience.
 
- Extremely good = in C, C++, VC++, = COM/DCOM
- Good in Real = Time=20 Systems
- Good in Data = Structures/=20 Algorithms
- Motorola=20 6800 Assembly Language
- Java experience is a = plus
- Preferably = Computer Science=20 Engineers
 
Need to be good in Windows NT=20 4.0
worked in = RTOS like=20 VxWorks or pSOS
It would=20 be an advantage if you know GEM=20 and SECS
(these are=20 inter-tool communication standards in semi conductor manufacturing = industry). 

Thank=20 you so very much for = your time=20 and I look forward to hearing and working with you soon.

=20 =

------=_NextPart_000_0001_01C1FBB7.246636F0-- From vxwexplo-errs@csg.lbl.gov Wed May 15 00:43:12 2002 From: "Surinder Singh" Date: Wed May 15 00:43:14 PDT 2002 Subject: Embedded Engineer This is a multi-part message in MIME format. ------=_NextPart_000_0000_01C1FBB8.739121D0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Please forward the updated resume attached as a word document and also mention the expected salary, availability and employment status (Visa, Green Card, US Citizen). These are FULL TIME positions in Bay Area - California, Japan and Bangalore - India. Please mention the location you are applying for. 1) Sr. Software Engineer / Technical Lead 4 to 10 years Exp. 2) Technical Lead/Manager level 5 to 10 years experience. - Extremely good in C, C++, VC++, COM/DCOM - Good in Real Time Systems - Good in Data Structures/ Algorithms - Motorola 6800 Assembly Language - Java experience is a plus - Preferably Computer Science Engineers Need to be good in Windows NT 4.0 worked in RTOS like VxWorks or pSOS It would be an advantage if you know GEM and SECS (these are inter-tool communication standards in semi conductor manufacturing industry). Thank you so very much for your time and I look forward to hearing and working with you soon. ------=_NextPart_000_0000_01C1FBB8.739121D0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Please forward the updated resume attached = as a word=20 document and also mention the = expected=20 salary, availability and employment status (Visa, Green Card, US=20 Citizen). These are = FULL=20 TIME positions in Bay Area - = California,=20 Japan and Bangalore - India. = Please=20 mention the location you are applying=20 for.
 
1) Sr. Software Engineer / = Technical Lead=20 4 to 10 years Exp.
 
2) Technical Lead/Manager level = 5 to 10=20 years experience.
 
- Extremely good = in C, C++, VC++, = COM/DCOM
- Good in Real = Time=20 Systems
- Good in Data = Structures/=20 Algorithms
- Motorola 6800 Assembly=20 Language
- Java experience is a = plus
- Preferably = Computer Science=20 Engineers
 
Need to be good in Windows NT=20 4.0
worked in = RTOS like=20 VxWorks or pSOS
It would=20 be an advantage if you know GEM=20 and SECS
(these are=20 inter-tool communication standards in semi conductor manufacturing = industry). 

Thank=20 you so very much for = your time=20 and I look forward to hearing and working with you soon.

=20 =

------=_NextPart_000_0000_01C1FBB8.739121D0-- From vxwexplo-errs@csg.lbl.gov Wed May 15 04:03:16 2002 From: Vxworks Exploder Date: Wed May 15 04:03:19 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed May 15 04:03:11 PDT 2002 Subject: Link to boot multiple ethernet interface Subject: Re: Compilation log file Subject: Re: 860 SMC question ... Subject: Re: Is it possible to build a simulator myself? Subject: Re: Link to boot multiple ethernet interface Subject: Determing the Link status using a Motorola MPC860T and the MII Subject: Re: FTP Site Subject: Re: Non blocking UDP sockets? Subject: Re: Link to boot multiple ethernet interface Subject: Re: ENVOY SNMP Subject: Re: Polymorphism and Destructor Subject: Looking for some HOME BASED PC work Subject: Re: MTD for 28F640J3A Subject: Re: implementing clrscr() in vxworks Subject: Re: routing table Subject: Re: Polymorphism and Destructor Subject: Re: MTD for 28F640J3A Subject: Error: clnttty_rcv: read timed out Subject: [newbie]SIO driver - UART Tx Buffer Full Subject: "RPCExit: extra call" error message Subject: Re: Compilation log file Subject: Re: Compilation for MBX860 target failed Subject: Re: Difference between RTOS and normal OS Subject: Re: Non blocking UDP sockets? Subject: Tornado II - Printing Source Files Subject: Re: Compilation for MBX860 target failed Subject: Re: Link to boot multiple ethernet interface Subject: MPC860/MPC8260 Question ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Link to boot multiple ethernet interface Date: Tue, 14 May 2002 11:16:19 +0000 (UTC) From: "Dexter Nixon" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: <80b6fd721aa44a76437bc129c318a32e.76014@mygate.mailgate.org> Hi, Can anynody point me to a link which speaks about booting multiple ethernet interfaces in VxWorks. Also which booting the image itself is it possible to boot the interfaces if it is multihomed. Kindly help me. Thanks. Nix - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation log file Date: Tue, 14 May 2002 13:45:32 +0200 From: "Werner Schiendl" Message-ID: <3ce0f8d2$1@brateggebdc5.br-automation.co.at> References: Hi Alex, you need to invoke the DOS shell with more environment space. I think the parameter to command(.com) was /e: - but not sure about it. AFAIK, this only applies to Gamer's Edition of MS Windows. I strongly recommend using an edition for grown-ups (NT, 2000 or XP). == The next thing that is wrong is, that you need to invoke the build from the command line if you want to follow Andray's suggestion. Change to the directory where your project is located and type: make project-name.out > project-name.log replace project-name with the name of your project (out file). If you have a bootable project, you type make vxWorks > build.log The log file, as you should have seen, can of course have any name you like. Note: Before any of the above will work, you need to build one time from the GUI. Otherwise, changes in the project facility will not make it into the build. hth Werner "Alex Vinokur" wrote in message news:abktvh$j639a$1@ID-79865.news.dfncis.de... > > "Andray Kaganovsky" wrote in message news:Xns920ADE78B662Aandrayk@216.254.136.26... > | "Alex Vinokur" wrote in news:abdho1$h4l0l$1@ID- > | 79865.news.dfncis.de: > | > | >========================================================= > | > Tornado : Version 2.0.2 > | > VxWorks : Version 5.4.2 > | > BSP : Version 1.1/1 > | > Compiler : ccsimpc > | > gcc version egcs-2.90.29 tornado 2.0 980515 > | > (egcs-1.0.3 release) > | > Host : Windows2000 > | > Target : VxSim for Windows (VxWorks target simulator) > | >========================================================= > | > > | > I would like the compilation results to be written > | > not only to Build Output Window, > | > but to some compilation log file as well. > | > > | > Is it possible to do it? > | > > | > > > [snip] > > | > > | > | [AndrayK]. Did you try building from DOS shell ? You'll need to > | execute /host/x86-win32/bin/torvars.bat before you > | invoke make .. > | > | regards, Andray > > > 1. Here is /host/x86-win32/bin/torvars.bat in my system : > > ============ File: torvars.bat ============ > rem Command line build environments > set WIND_HOST_TYPE=x86-win32 > set WIND_BASE=D:\sys2\Tornado > set PATH=%WIND_BASE%\host\%WIND_HOST_TYPE%\bin;%PATH% > ==================================== > > > > 2. Here is results of torvars.bat execution. > > ============ Execution: torvars.bat ============ > %torvars.bat > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>rem Command line build environments > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set WIND_HOST_TYPE=x86-win32 > Out of environment space > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set WIND_BASE=D:\sys2\Tornado > Out of environment space > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set PATH=\host\\bin;C:\PROGRA~1\Far;C:\DJGPP\bin; > D:\Perl\bin\;C:\WINNT\system32;C:\WINNT;C:\WINNT\System32\Wbem;D:\app1\araxi > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN> > ======================================== > > > 3. Compilation is pefrormed via : > Workspace Window > Files View > Right Click > ReBuild All (vxWorks) > > I saw no compilation log file. > > > Thanks. > > -- > ==================== > Alex Vinokur > http://up.to/alexvn > http://go.to/alexv_math > mailto:alexvn@bigfoot.com > mailto:alexvn@go.to > ==================== > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 860 SMC question ... Date: Tue, 14 May 2002 11:50:22 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <2S6E8.21772$f5.1814320@news> References: <5dRC8.21403$f5.1728610@news> Is this worth trying even? "Jim" wrote in message news:5dRC8.21403$f5.1728610@news... > Hello, > > I have a question for you MPC860 lovers out there ... I am trying to write a > MPC 860 polled UART driver that I can call from romStart.c ... > > Nothing seems to be printing yet ... and I haven't spent to much time > debugging ... but any suggestions would be awesome ... the code is below ... > > Jim > > > /* SMC 2 Pins Information */ > #define portB23_SMSYN1_mask (0x1<<31-23) > #define portB24_SMRXD1_mask (0x1<<31-24) > #define portB25_SMTXD1_mask (0x1<<31-25) > > /* Port B Registers */ > #define PBPAR_ADDR 0x02200abc > #define PDDIR_ADDR 0x02200ab8 > #define PBODR_ADDR 0x02200ac0 > #define PBDAT_ADDR 0x02200ac4 > > /*SMC Clocks Registers*/ > #define CPU_CLCK 80000000 /*80 Mhz*/ > #define BRG1_ADDR 0x22009F0 > #define SCCR_ADDR 0x2200280 > > /*SI Mode Register*/ > #define SIMODE_ADDR 0x2200ae0 > > /*SMS Parameter Ram*/ > #define SMC1_PRAM_START_ADDR 0x2203E80 > > #define SMC1_PRAM_TBASE_ADDR (SMC1_PRAM_START_ADDR + 0x00) > #define SMC1_PRAM_RBASE_ADDR (SMC1_PRAM_START_ADDR + 0x02) > #define SMC1_PRAM_RFCR_ADDR (SMC1_PRAM_START_ADDR + 0x04) > #define SMC1_PRAM_TFCR_ADDR (SMC1_PRAM_START_ADDR + 0x05) > #define SMC1_PRAM_MRBLR_ADDR (SMC1_PRAM_START_ADDR + 0x06) > #define SMC1_PRAM_MAX_IDL_ADDR (SMC1_PRAM_START_ADDR + 0x28) > #define SMC1_PRAM_BRKLN_ADDR (SMC1_PRAM_START_ADDR + 0x2C) > #define SMC1_PRAM_BRKEC_ADDR (SMC1_PRAM_START_ADDR + 0x2E) > #define SMC1_PRAM_RKCR_ADDR (SMC1_PRAM_START_ADDR + 0x30) > > /*CPCR Communications Processor Command Register*/ > #define CPCR_ADDR 0x022009c0 > > /*CIMR*/ > #define CIMR_ADDR 0x2200948 > #define CICR_ADDR 0x2200940 > > /*SDMA Config Register*/ > #define SDCR_ADDR 0x02200030 > > > /*SMC1 Event Register*/ > #define SMCE1_ADDR 0x02200a86 > /*SMC1 Mask Register*/ > #define SMCM1_ADDR 0x02200a8a > > /*SMC1 Mode Register*/ > #define SMCMR1_ADDR 0x02200a82 > > > > > > /**********************************************************************/ > void InitSMC2ForUart() { > unsigned SCCR_DFBRG; > unsigned default_baud_rate; > unsigned tager_brg1_rate = 9600 * 16; > unsigned BRG1_CD; > unsigned BRG1_DIV16; > > static char rx_buff[10]; > #define TX_BUFF "hello\n" > static unsigned tmp; > > > /*--------------------------------------------------------------------- > * 1.) ConÞgure the port B pins to enable SMTXD1 and SMRXD1. > * Set PBPAR[24, 25] then clear PBDIR[24, 25] and PBODR[24, 25]. > ---------------------------------------------------------------------*/ > > *(unsigned *)PBPAR_ADDR = (*(unsigned *)PBODR_ADDR) | > ~(portB24_SMRXD1_mask | portB25_SMTXD1_mask); > *(unsigned *)PDDIR_ADDR = (*(unsigned *)PBODR_ADDR) & > ~(portB24_SMRXD1_mask | portB25_SMTXD1_mask); > *(unsigned *)PBODR_ADDR = (*(unsigned *)PBODR_ADDR) & > ~(portB24_SMRXD1_mask | portB25_SMTXD1_mask); > > > > > /*--------------------------------------------------------------------- > * 2.) ConÞgure the BRG1. Write BRGC1 with 0x01_0144. > * The DIV16 bit is not used and the divider is 162 (decimal). > * The resulting BRG1 clock is 16´ the preferred bit rate. > ---------------------------------------------------------------------* / > SCCR_DFBRG = ( ( (*(unsigned *)SCCR_ADDR) >> (31-20) )& 0x3 ); > if ( SCCR_DFBRG == 0x0 ) { > default_baud_rate = CPU_CLCK / 1; > } else if ( SCCR_DFBRG == 0x1 ) { > default_baud_rate = CPU_CLCK / 4; > } else if ( SCCR_DFBRG == 0x2 ) { > default_baud_rate = CPU_CLCK / 16; > } else { > default_baud_rate = CPU_CLCK / 64; > } > > BRG1_CD = default_baud_rate/tager_brg1_rate; > > if ( BRG1_CD > (0xfff) ) { /*CD is only 12 bits*/ > BRG1_CD = (default_baud_rate/16)/tager_brg1_rate; > BRG1_DIV16 = 1; > } else { > BRG1_DIV16 = 0; > } > > *(unsigned *)BRG1_ADDR = (0x00010000 | (BRG1_CD<<1) | BRG1_DIV16); > > > > > /*--------------------------------------------------------------------- > * 3. Connect BRG1 to SMC1 using the SI. Clear SIMODE[SMC1, SMC1CS]. > --------------------------------------------------------------------- */ > *(unsigned *)SIMODE_ADDR = (*(unsigned *)SIMODE_ADDR) & ~(0x0000f000); > > > > > /*--------------------------------------------------------------------- > * 4. Assuming one RxBD at the beginning of dual-port RAM > * followed by one TxBD, write RBASE with 0x0000 and > * TBASE with 0x0008. > --------------------------------------------------------------------- */ > * (unsigned short *)SMC1_PRAM_TBASE_ADDR = 0x0008; > * (unsigned short *)SMC1_PRAM_RBASE_ADDR = 0x0000; > > > > /*--------------------------------------------------------------------- > * 5. Write 0x0091 to CPCR to execute the INIT RX AND TX PARAMETERS > command. > -------------------------------------------------------------------- - -* > / > *(unsigned short *)CPCR_ADDR = 0x0091; > > > /*--------------------------------------------------------------------- > * 6. Initialize the SDMA conÞguration register (SDCR) to 0x0001. > -------------------------------------------------------------------- - -* > / > * (unsigned short *)SDCR_ADDR = 0x0001; > > > > /*--------------------------------------------------------------------- > * 7. Write RFCR and TFCR with 0x10 for normal operation. > -------------------------------------------------------------------- - -* > / > * (unsigned char *)SMC1_PRAM_RFCR_ADDR = (unsigned char)0x10; > * (unsigned char *)SMC1_PRAM_TFCR_ADDR = (unsigned char)0x10; > > > /*--------------------------------------------------------------------- > * 8. Write MRBLR with the maximum number of bytes per receive buffer. > * Assume 16 bytes, so MRBLR = 0x0010. > -------------------------------------------------------------------- - -* > / > * (unsigned short *)SMC1_PRAM_MRBLR_ADDR = 0x0010; > > > > /*--------------------------------------------------------------------- > * 9. Write MAX_IDL with 0x0000 in the SMC UART-speciÞc parameter RAM > to > * disable the MAX_IDL functionality for this example. > -------------------------------------------------------------------- - -* > / > * (unsigned short *) SMC1_PRAM_MAX_IDL_ADDR = 0x0000; > > /*--------------------------------------------------------------------- > * 10. Clear BRKLN and BRKEC in the SMC UART-speciÞc parameter RAM. > -------------------------------------------------------------------- - -* > / > * (unsigned short *)SMC1_PRAM_BRKLN_ADDR = 0x0000; > * (unsigned short *)SMC1_PRAM_BRKEC_ADDR = 0x0000; > > > > /*--------------------------------------------------------------------- > * 11. Set BRKCR to 0x0001; if a STOP TRANSMIT COMMAND is issued, > * one break character is sent. > -------------------------------------------------------------------- - -* > / > * (unsigned short *) SMC1_PRAM_RKCR_ADDR = 0x0001; > > > > /*--------------------------------------------------------------------- > * 12. Initialize the RxBD. Assume the Rx buffer is at 0x0000_1000 > * in main memory. Write 0xB000 to Rx_BD_Status, 0x0000 to > Rx_BD_Length > * (not required), and 0x0000_1000 to Rx_BD_Pointer. > -------------------------------------------------------------------- - -* > / > *(unsigned short*)0x2202000= 0xB000; > *(unsigned short*)0x2202002= 10; > *(unsigned *)0x2202004= (unsigned *)rx_buff; > > > > /*--------------------------------------------------------------------- > * 13. Assuming the Tx buffer is at 0x00002000 in main memory and > * contains Þve 8-bit characters, write 0xB000 to Tx_BD_Status, > * 0x0005 to Tx_BD_Length, and 0x00002000 to Tx_BD_Pointer. > -------------------------------------------------------------------- - -* > / > *(unsigned short*)0x2202008= 0xB000; > *(unsigned short*)0x220200a= 6; > *(unsigned *)0x220200c= (unsigned *)0x2800130; > > > > /*--------------------------------------------------------------------- > * 14. Write 0xFF to the SMCE register to clear any previous events. > -------------------------------------------------------------------- - -* > / > * (unsigned char *) SMCE1_ADDR = 0xff; > > > > /*--------------------------------------------------------------------- > * 15. Write 0x17 to the SMCM register to enable all possible SMC > * interrupts. > * Note: I want no interupts enabled so I set this to 0x0 ... like > * a good boy > -------------------------------------------------------------------- - -* > / > * (unsigned char *) SMCM1_ADDR = 0x0; > > > > /*--------------------------------------------------------------------- > * 16. Write 0x00000010 to the CIMR so the SMC1 can generate a system > * interrupt. Initialize the CICR. > * Note: I want no interupts enabled so I DO NOT DO THIS! > -------------------------------------------------------------------- - -* > / > *(unsigned *) CIMR_ADDR = 0x00000010; > *(unsigned *) CICR_ADDR = 0x00000000; > > > /*--------------------------------------------------------------------- > * 17. Write 0x4820 to SMCMR to conÞgure normal operation (not > loopback), > * 8-bit characters, no parity, 1 stop bit. > * The transmitter and receiver are not yet enabled. > -------------------------------------------------------------------- - -* > / > * (unsigned short *) SMCMR1_ADDR = 0x4820; > > > /*--------------------------------------------------------------------- > * 18. Write 0x4823 to SMCMR to enable the SMC transmitter and > * receiver. This additional write ensures that the TEN and > * REN bits are enabled last. > -------------------------------------------------------------------- - -* > / > * (unsigned short *) SMCMR1_ADDR = 0x4823; > > } > > > /**********************************************************************/ > void SMC2_Write(char *str) { > int i; > for (i=0;i<256 || str[i]!='\0';i+=1); > > *(unsigned short*)0x2202008= 0xB000; > *(unsigned short*)0x220200a= i; > *(unsigned *)0x220200c= (unsigned *)0x2800130; > > } > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is it possible to build a simulator myself? Date: Tue, 14 May 2002 13:49:17 +0200 From: "Werner Schiendl" Message-ID: <3ce0f9b3$1@brateggebdc5.br-automation.co.at> References: Hi, you can build your own simulator, but it is not based on the target hardware. How should this work? The simulation runs on your host (Windows or Solaris), so it must be compiled for the host's architecture and file format (e. g. it must be in PE - portable exe - format on Windows). This should not prevent you from testing your application, unless you have some hardware specific stuff used there. In that case, you need to write a simulation for this parts of the project or exclude them from the simulation. If you want to achieve something else, please specify what exactly. hth Werner "yyzhong" wrote in message news:b35585ab.0205081909.64d797e7@posting.google.com... > I need to use a simulator oriented to a project based on PentiumGNU. > But I don't know how to build it. > Is it possible? And which disk is necessory? > The disks I can use are listed below: > CD Title: Tornado 2.0 AP for x86 > Part No.: TDK-13127-ZC > CD Title: Tornado 2.0 AP for PowerPC > Part No.: TDK-13128-ZC > CD Title: Tornado 2.0.2 for PowerPC > Part No.: TDK-12389-ZC > CD Title: Tornado 2.0.2 for All Simulators > Part No.: TDK-12840-ZC > CD Title: BSP Developer's Kit > Part No.: TDK-13234-ZC > CD Title: RTI SCOPE TOOLS 2.1 > Part No.: STS-14469-zc-00 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Link to boot multiple ethernet interface Date: 14 May 2002 12:13:05 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <80b6fd721aa44a76437bc129c318a32e.76014@mygate.mailgate.org> "Dexter Nixon" wrote: > Can anynody point me to a link which speaks about booting multiple > ethernet interfaces in VxWorks. Also which booting the image itself is > it possible to boot the interfaces if it is multihomed. Kindly help me. What I expect that you want to do is have multiple ethernet controllers starting up during the boot instead of only one. I don't think this is documented. If you want to do this you have to modify the bootrom code. Also if you want to do this from the application you have to initialise your network yourself. But it is not very difficult: ipAttach (, ); usrNetIfConfig (, , , , ); So for example: ipAttach (1,"fei"); usrNetIfCOnfig("fei", 1, "10.20.30.40","Interface2", 0xffffff00); But you need to configure your interface in configNet.h first. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Determing the Link status using a Motorola MPC860T and the MII Date: 14 May 2002 06:41:10 -0700 From: chris.mycock@tde.alstom.com (Chris Mycock) Organization: http://groups.google.com/ Message-ID: <2e8ae7f.0205140541.3f6322df@posting.google.com> I am trying to write a function to read the current link status as provided by the phy (national semiconductor DP83843BVJE). I have added an extra function to the motfecend driver that uses existing functions to read the status using the MII port on the MPC860. The link status is checked when the device is being configured after the auto-negotiation and this returns the correct status. After the status comes back ok the device works as normal. If I try to call the exact same function with the exact same settings to read the status from the PHY, the return value is always all 1's (0xFFFFFFFF). I am calling the 'get status' function from a timer expiration call and it seems to work as the FEC generates an interrupt to say the message transaction is complete. Any help would be very appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FTP Site Date: Tue, 14 May 2002 11:57:02 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3CE15DFE.E2434894@lmco.com> References: MURALI wrote: > > Hi, > > I was trying to access the site > > ftp://ftp.ece.ucdavis.edu/pub/snmp/ucd-snmp.tar.gz > > but could not access the same. Can any one of you > help me in letting me know whether the above > ID is correct and if yes and if the site has been > moved to different location, can you pls give me > the new location also. > > Thanks for the help. > > Regards > > Murali Try: http://sourceforge.net/projects/net-snmp Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Non blocking UDP sockets? Date: Tue, 14 May 2002 09:54:42 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: <3CDB8D86.6090009@spamm.me.l8s.co.uk> <488e459a.0205100714.5c6a4469@posting.google.com> UDP sockets can block. - -- Hwa Jin Bae mailto:hjb@pso.com http://www.pso.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Link to boot multiple ethernet interface Date: Tue, 14 May 2002 10:07:36 -0700 From: "Dan Gold" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: References: <80b6fd721aa44a76437bc129c318a32e.76014@mygate.mailgate.org> Sender: gold@ensemble.com "Dexter Nixon" wrote in message news:80b6fd721aa44a76437bc129c318a32e.76014@mygate.mailgate.org... > Hi, > > Can anynody point me to a link which speaks about booting multiple > ethernet interfaces in VxWorks. Also which booting the image itself is > it possible to boot the interfaces if it is multihomed. Kindly help me. The link is Johan's vxWorks FAQ, i.e. http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html. See section 4.1 (network configuration) for the sequence of steps necessary to bring up multiple END devices. As Johan pointed out earlier, you should have all of them already in the END_DEV_TBL defined in configNet.h. And the code in usrNetInit() (whether it's in the application or bootrom) should have run muxDev{Load,Start} on them. Then it's valid to call the code in the link above. hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ENVOY SNMP Date: 14 May 2002 11:02:20 -0700 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: References: <3CD35445.8123C2F0@worldnet.att.net> Eventually I know what the problem is. HP Openview is designed to lookup DNS server before sending the request, if the target is not configured in the DNS server, cause big delay. One of the simplest way to fix is: 1. edit \winnt\system32\etc\hosts to add ip address of target 2. run "loadhosts -m \WINNT\system32\etc\hosts" in command prompt 3. enjoy the mib browser Cheers jzhang@drs.ca (Jason) wrote in message news:... > Hello Chris: > > We am using ENVOY 9.2 and experiencing the same problem, > I don't have the solution for it yet but have some discovery. > > I tried some commands provided by HP Openview in DOS prompt. > following are the snmpwalk command I run with different options > and it shows the reason why slow: > D:\>snmpwalk -v 1 -c public 10.3.2.95 system > > D:\>snmpwalk -d -v 1 -c public 10.3.2.95 system > > The first one goes very fast and the second one with -d option > (means dump ASN.1 packet trace) goes really really slow. I think > the reason why there is a big delay between agent and manager is > because enable the "dump", and worth to try to disable it . If > you have the solution, pls let me know. > > Thanks > > > > > > chrismooremobile@yahoo.com (Chris Moore) wrote in message news:... > > Les Cargill wrote in message news:<3CD35445.8123C2F0@worldnet.att.net>... > > > Jason wrote: > > > > > > > > Hello: > > > > > > > > I have successfully built the ENVOY SNMP module for PPC, but it quite > > > > big (1.5MBytes). I hope to reduce it. Does any one have the idea ? > > > > > > > > One way I want to try is, among the 11 groups of MIB II, only include > > > > a few of them, say the system and ip, exclude the rest 9 groups. But > > > > how to do it exactly? > > > > > > > > > > Use the "-exclude" option when invoking the MIB compiler. > > > > > > > Thanks in advance! > > > > Hi Everyone, > > > > We use ENVOY 9.1 SNMP and find that is is REALLLLY slow when working > > with HP Open view and ...AdventNet SNMP mangers. Using "snmptool" from > > the win32 CLI its fine and fast.... does anyone know if this is a > > setting problem on either the client or the manager? > > > > Cheers, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Polymorphism and Destructor Date: 14 May 2002 12:30:53 -0700 From: pboulay@hotmail.com (Patrick Boulay) Organization: http://groups.google.com/ Message-ID: <8bb5523.0205141130.64edc264@posting.google.com> References: <8bb5523.0205121836.5b8b1432@posting.google.com> <8bb5523.0205130754.71df76b0@posting.google.com> Thanks for you answer, but the real problem seems to be in the kernel. My boss contact WindRiver and everything suppose to be ok, so he have to check in the kernel. pboulay@hotmail.com (Patrick Boulay) wrote in message news:<8bb5523.0205130754.71df76b0@posting.google.com>... > I did it, I didn't show you the implementation here because the code > would be too long for nothing.. > I just put cout in the destructors and constructors. > > "Alex Vinokur" wrote in message news:... > > "Patrick Boulay" wrote in message news:8bb5523.0205121836.5b8b1432@posting.google.com... > > | Hello, > > | > > | I have a big problem with polymorphism and destructor. See the class > > | in example bellow, the vxworks crash after I delete the pointer. > > | > > > > You should implement constructors and destructors. > > > > | class A > > | { > > | public: > > | A(); > > | virtual ~A(); > > | }; > > | class B:public A > > | { > > | public: > > | B(); > > | virtual ~B(); > > | }; > > | int aFunction() > > | { > > | A *ptr; > > | > > | ptr = new B; > > | > > | delete ptr; > > | > > | return 0; > > | } > > | > > | > > | > > | Thanks > > | Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Looking for some HOME BASED PC work Date: Saturday, 11 May 2002 20:45:22 -0600 From: wra-uk-2001@freeuk.com Organization: wra-uk-2001@freeuk.com Message-ID: <11050220.4522@freeuk.com> Reply-To: wra-uk-2001@freeuk.com Hi, Thankyou for reading my advert. I am looking for some work that I can do in the comfort of my own home. I am NOT interested in ANY of the following :- MLM Network Marketing Door-to-Door Sales Pay To EMail Pay To Surf Pay To SMS Selling Of Web Sites Recruiting People Stuffing Envelopes Craft Work Typing I am looking for a Job that I can do using my PC and get paid for it. Do you need a Web Site putting togther? Do you need some special database software writing ? Do you need a Mailing List creating ? Do you need any PC based research done ? I DON'T and WILL NOT pay anthing for kits, tools etc. My main skills are :- Sybase Power Builder Clipper Visual Basic Visual Foxpro HTML ASP Word Access Excel If you have anything that may be of interest to me, then please get in touch. Please reply to me direct and not via the group. Please remember to take out the words 'NOSPAM'. Thanks Wayne UK --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MTD for 28F640J3A Date: Tue, 14 May 2002 17:20:43 -0400 From: "David Masse" Organization: Universite du Quebec Message-ID: References: This is a CFI device, so you can use the cfiscs.c MTD that is provide with vxworks. David "doug dotson" wrote in message news:v0VD8.19689$A%3.194689@ord-read.news.verio.net... > Does anyone know of a TFFS MTD for an Intel StrataFlash > 28F640J3A? > > Thanks, > Doug > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: implementing clrscr() in vxworks Date: Sat, 11 May 2002 19:24:46 -0400 From: joe durusau Organization: Bellsouth.Net Message-ID: <3CDDA83E.E7D4A280@bellsouth.net> References: Anand wrote: > HI All, > anybody knows how to implement clrscr() in vxworks as it seems that > conio.h is not supported in vxworks. > thanks in advance. > regards. > anand. Given that vxWorks doesn't have console by default, I would be surprised at any such thing existing. It would depend on what kind of terminal or terminal emulator you want a clear screen on. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routing table Date: 14 May 2002 21:03:35 -0700 From: overtones@excite.com (Strawberry) Organization: http://groups.google.com/ Message-ID: <41440750.0205142003.3ecf8af3@posting.google.com> References: <68c7b7f1.0205140554.7fbca9c1@posting.google.com> vrrajesh@intotoinc.com (rajesh vr) wrote in message news:<68c7b7f1.0205140554.7fbca9c1@posting.google.com>... > Hi, > > I have set of question on routing table > > 1.how do get the default route, >> (i)One of the methods of getting any route(incl. default route) is by routing sockets(AF_ROUTE addr family). For more info you can refer Network Programming by Stevens Vol-II. > > 2.Is there any system call for updating the elements of a > route(metrics) >> I dunno. > Is there any way I can add or Delete Route other than the system > calls found in routeLib ? >> (i) Routing sockets also provide means of adding/deleting routes (ii) ioctl calls with SIOCDELRT & SIOCADDRT can also do the job > > 3.How do I access the Routing Table .(like first route ,next route) >> I dunno > 4. How is the routing table maintained. >> By means of a radix tree with the key being destination & mask (think gateway cannot be used as a differentiating factor) > > 5. m2IpLib mentions some system call for setting and retrieving > routing Table entry.By setting > the MIB-2 routing entry ,will the kernel routing entry be set for > the same route? > Can I assume that routes in MIB-2 routing entry and kernel routing > entry are the same. >> Do let me know about this when u find the answer > > Thanking in advance, > rajesh regards, Strawberry (my fav...) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Polymorphism and Destructor Date: Wed, 15 May 2002 08:09:27 +0200 From: Alex Vinokur Message-ID: <3CE1FB97.9020601@bigfoot.com> References: <8bb5523.0205121836.5b8b1432@posting.google.com> <8bb5523.0205130754.71df76b0@posting.google.com> <8bb5523.0205141130.64edc264@posting.google.com> Patrick Boulay wrote: > Thanks for you answer, but the real problem seems to be in the kernel. > My boss contact WindRiver and everything suppose to be ok, so he have > to check in the kernel. > > pboulay@hotmail.com (Patrick Boulay) wrote in message news:<8bb5523.0205130754.71df76b0@posting.google.com>... > >>I did it, I didn't show you the implementation here because the code >>would be too long for nothing.. >>I just put cout in the destructors and constructors. >> >>"Alex Vinokur" wrote in message news:... >> >>>"Patrick Boulay" wrote in message news:8bb5523.0205121836.5b8b1432@posting.google.com... >>>| Hello, >>>| >>>| I have a big problem with polymorphism and destructor. See the class >>>| in example bellow, the vxworks crash after I delete the pointer. >>>| >>> >>>You should implement constructors and destructors. >>> >>>| class A >>>| { >>>| public: >>>| A(); >>>| virtual ~A(); >>>| }; >>>| class B:public A >>>| { >>>| public: >>>| B(); >>>| virtual ~B(); >>>| }; >>>| int aFunction() >>>| { >>>| A *ptr; >>>| >>>| ptr = new B; >>>| >>>| delete ptr; >>>| >>>| return 0; >>>| } >>>| >>>| >>>| >>>| Thanks >>>| Patrick >>> ========================================================= Tornado : Version 2.0.2 VxWorks : Version 5.4.2 BSP : Version 1.1/1 Compiler : ccsimpc gcc version egcs-2.90.29 tornado 2.0 980515 (egcs-1.0.3 release) Host : Microsoft Windows 2000 [Version 5.00.2195] Target : VxSim for Windows (VxWorks target simulator) ToolChain : SIMNTgnu ========================================================= Project Type : bootable VxWorks image (custom configured) ========================================================= The configuration above has no problem with similar code. || ###### C++ Code : BEGIN ###### // File s1.cpp #include "iostream.h" #define SHOW_IT \ cout << "[" \ << __FILE__ \ << ", #" \ << __LINE__ \ << "] ---> " \ << __PRETTY_FUNCTION__ \ << endl class A { public: A() {SHOW_IT;} virtual ~A() {SHOW_IT;} }; class B : public A { public: B() {SHOW_IT;} ~B() {SHOW_IT;} }; int foo(void) { A *ptr; ptr = new B; delete ptr; return 0; } || ###### C++ Code : END ######## || ====== Tornado Shell Window : BEGIN ====== - -> foo [D:/s1.cpp, #15] ---> A::A() [D:/s1.cpp, #22] ---> B::B() [D:/s1.cpp, #23] ---> B::~B() [D:/s1.cpp, #16] ---> A::~A() value = 0 = 0x0 || ====== Tornado Shell Window : END ======== - -- ============================ Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ============================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MTD for 28F640J3A Date: 14 May 2002 22:36:34 -0700 From: alimohd@logiceastern.com (Ali) Organization: http://groups.google.com/ Message-ID: References: hi David is correct that this is CFI device but the cfiscs.c provided with vxworks is for 8-bit inteleaving be sure u need 8-bit or 16-bit interleaving .Infact u can ask ur FAE for newer version of cfiscs.c in which they have supported 16-bit interleaving but it supports only little endian modes ..anyway that u can work out to support it for big endian.. ali "David Masse" wrote in message news:... > This is a CFI device, so you can use the cfiscs.c MTD that is provide with > vxworks. > > David > > "doug dotson" wrote in message > news:v0VD8.19689$A%3.194689@ord-read.news.verio.net... > > Does anyone know of a TFFS MTD for an Intel StrataFlash > > 28F640J3A? > > > > Thanks, > > Doug > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Error: clnttty_rcv: read timed out Date: 11 May 2002 19:07:26 -0700 From: Techman@ematic.com (Thoth) Organization: http://groups.google.com/ Message-ID: Greetings, Here's my setup: Environment: Serial connection on machinea to machineb. Tornado shell on machine a (win2k). Host machine( machine b, NT) has a PCI card with a variant of VxWorks running (IxWorks) When I'm executing my code (its a very long path), I'll get this (Serial@machinea): Sat May 11 19:50:55 2002 Error: clnttty_rcv: read timed out Error: rpccore backend client Timed out Warning: tgtEvtGet: cannot get event. Now, I can put in debug statements to send messages to my tornado shell and, in fact, do this. But, the timeout appears to move if I add/change code. Is there any way to track this down during execution? After the timeout occurs, there's really nothing that can be done. Some of the suggestions I've seen/heard for the cause of timeouts are that a pointer is unitialized and a thread is taking too long to finish something. I dont have support, so I can't use the Windriver database. Any suggestions are appreciated. Thanks. T.M. - ---------------------------------------------------------------------- __&__ / \ | | ^^ (o)(o) C ,---_) | |,___| ___ "Mmmmmm....delicious, serial port timeouts" | \__/ /_____\ /_____/ \ --------------------------- Newsgroups: comp.os.vxworks Subject: [newbie]SIO driver - UART Tx Buffer Full Date: 11 May 2002 20:09:42 -0700 From: bat_kar75@yahoo.com (Karpur) Organization: http://groups.google.com/ Message-ID: <11609fa3.0205111909.529826d0@posting.google.com> Hi, I am writing driver for a uart. The uart works only in interrupt mode. It does not have the transmitter ready interrupt. Instead it has the Tx Buffer Full interrupt. All the uart drivers (including vxworks template SIO driver) that I have seen considers only transmitter ready interrupts and not Tx Buffer Full interrupt. I appreciate any pointers on how the driver should handle transmission. Regards, Karpur --------------------------- Newsgroups: comp.os.vxworks Subject: "RPCExit: extra call" error message Date: Wed, 15 May 2002 08:57:27 +0000 (UTC) From: "Tom Almo" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: hi all after install torando 2.0.1 5 windows with this message jump out. each time i open the tornado this messages come back. did any body know how to slove the problem :-> Tom. - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation log file Date: Sun, 12 May 2002 08:20:22 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: References: "Andray Kaganovsky" wrote in message news:Xns920ADE78B662Aandrayk@216.254.136.26... | "Alex Vinokur" wrote in news:abdho1$h4l0l$1@ID- | 79865.news.dfncis.de: | | >========================================================= | > Tornado : Version 2.0.2 | > VxWorks : Version 5.4.2 | > BSP : Version 1.1/1 | > Compiler : ccsimpc | > gcc version egcs-2.90.29 tornado 2.0 980515 | > (egcs-1.0.3 release) | > Host : Windows2000 | > Target : VxSim for Windows (VxWorks target simulator) | >========================================================= | > | > I would like the compilation results to be written | > not only to Build Output Window, | > but to some compilation log file as well. | > | > Is it possible to do it? | > | > [snip] | > | | [AndrayK]. Did you try building from DOS shell ? You'll need to | execute /host/x86-win32/bin/torvars.bat before you | invoke make .. | | regards, Andray 1. Here is /host/x86-win32/bin/torvars.bat in my system : ============ File: torvars.bat ============ rem Command line build environments set WIND_HOST_TYPE=x86-win32 set WIND_BASE=D:\sys2\Tornado set PATH=%WIND_BASE%\host\%WIND_HOST_TYPE%\bin;%PATH% ==================================== 2. Here is results of torvars.bat execution. ============ Execution: torvars.bat ============ %torvars.bat D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>rem Command line build environments D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set WIND_HOST_TYPE=x86-win32 Out of environment space D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set WIND_BASE=D:\sys2\Tornado Out of environment space D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set PATH=\host\\bin;C:\PROGRA~1\Far;C:\DJGPP\bin; D:\Perl\bin\;C:\WINNT\system32;C:\WINNT;C:\WINNT\System32\Wbem;D:\app1\araxi D:\SYS2\TORNADO\HOST\X86-WI~1\BIN> ======================================== 3. Compilation is pefrormed via : Workspace Window Files View Right Click ReBuild All (vxWorks) I saw no compilation log file. Thanks. - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation for MBX860 target failed Date: Sun, 12 May 2002 08:29:00 +0200 From: Alex Vinokur Message-ID: <3CDE0BAC.3090008@bigfoot.com> References: Gary M wrote: > "Alex Vinokur" wrote in message > news:abdqt2$hdsib$1@ID-79865.news.dfncis.de... > >>| IMPORT UINT32 sysInLong (UINT32 address); || >> > Line#417 > >>I changed the line above to >> >>IMPORT ULONG sysInLong (ULONG address); >> >>This action helped. >> >>However, is it safe ? >> >> > > Yes, but I'm curious as to why the problem developed after you "changed > nothing". Something obviously changed. Did you load a Tornado or BSP patch? What (which) patch do I have to load? [snip] ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Sun, 12 May 2002 06:00:27 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> Let me show you a different point of view: The design goals of a RTOs are quite different from a Generap Purpose OS "GPOS" like Unix, VMS and probably NT. In RTOS, it is assumed that the end-product is an embedded device, including hardware RTOS and higher-level software - all a single application, all tested together as a single unit. Therefore most safeguards that exist in GPOS are not really necesary, and are excluded in order to obtain the best possible performance on a given hardware platform, i.e. enable cost savings of the end-product. ALso, since all software is made and tested by a single entity, the resources (e.g. CPU, memory) is allocated to subsystems (tasks, modules) as per programmer request. If prioritiesa are set wrong, surely enough this will be revealed during system testing. GPOS has to take care of multi-application (often multi-user) requirement - many softwar emodules run concurrently on the same machine, which might have not been ever tested together, and still these mutually foregn modules must not interfere with one another, and most importantly, a bug in one module must not ever cause any damage to any other running module. GPOS also distincts application software from hardware platform to allow the same application to run a different platforms, leaving it to the OS to implement a given hardware to an agreed abstraction, and although a given application has never been tested on a particular hardware, it must run unmidified and as reliably as each of the components is. Performance in GPOS is traded off for maximum system stability, and one result of that, is to avoid process "starvation", or monopolisation of any resource by any given application. This is contrary to the deterministic, stricktly priority based resource allocation scheme in RTOS. I know that many GPOS have R-T extentions, but using these extentions certainly violates the goals of a GPOS. Sometimes this worked out OK, but sometimes when the systemk is still regarded is GP, and more stuff is added to it in the field, the use of RT extentions can have quite a negtative effect (as in system crash). Just a different way to look at things. Leonid "Vivek" wrote in message news:507efed5.0205022254.3db92d02@posting.google.com... > "Yin, Yong" wrote in message news:... > > priority-based preemptive schedule > but in unix/linux also we can modify the process priority etc, then > why arn't these are RTOS. ? > Sorry, just a confusion . > > > > > Vivek wrote in message > > news:507efed5.0205012037.3e11728f@posting.google.com... > > > Hi, > > > > > > I am looking for some concrete differences between RTOS and normal OS. > > > Could any body point me to the right direction for this. > > > > > > Thanks and in advance > > > > > > Vivek --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Non blocking UDP sockets? Date: 15 May 2002 02:44:19 -0700 From: cpotenza@onetel.net.uk (Claudio Potenza) Organization: http://groups.google.com/ Message-ID: <2025ac82.0205150144.185ce4c8@posting.google.com> References: <3CDB8D86.6090009@spamm.me.l8s.co.uk> <488e459a.0205100714.5c6a4469@posting.google.com> "Hwa Jin Bae" wrote in message > UDP sockets can block. I know of at least one reason why a UDP socket could block on 'send()' (and would block in vxWorks, I know this by experience). This can happen if the sender and receiver are on the same network (subnet). If the destination IP address has not been yet resolved by ARP, the 'send()' will wait until ARP finds the MAC address (and this can involv certainly milliseconds, depending on the receiving machine and the network) or ARP fails in resolving the address (and this will take MUCH longer). By the way, because of this, if you are sending data with UDP to some host, and this receiving host dies completely, after a while (when its ARP entry expires) all your 'send()' will fail. So you have a kind of "implicit negative acknowledge" on UDP sends (meaning that you can find out if your destination is alive or not) that some people find surprising on UDP ... --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado II - Printing Source Files Date: 15 May 2002 02:48:25 -0700 From: ralf@fh-rosenheim.de (Ralf Hager) Organization: http://groups.google.com/ Message-ID: <10bfb1a3.0205150148.70df153@posting.google.com> When i print source files from Tornado IDE the output always is in large font size about 36. How can i configure printing to use normal font size? Ralf ralf@fh-rosenheim.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation for MBX860 target failed Date: Sun, 12 May 2002 09:24:15 +0200 From: Alex Vinokur Message-ID: <3CDE189F.30303@bigfoot.com> References: Andray Kaganovsky wrote: > "Alex Vinokur" wrote in news:abdqt2$hdsib$1@ID- > 79865.news.dfncis.de: > > >>"Alex Vinokur" wrote in message news:abdp47$h3rj5$1 >> > @ID-79865.news.dfncis.de... > >>[snip] >> >>| >>| || File#3 of 3 >>| || ######### File mbx800.h : BEGIN ######### >>| || Full path : D:\sys2\Tornado\target\config\mbx860\mbx800.h >>| >>| /* mbx800.h - Motorola MBX800 board header */ >>| >>| /* Copyright 1984-1998 Wind River Systems, Inc. */ >>| /* Copyright 1997,1998 Motorola, Inc., All Rights Reserved */ >>| >>| >>| [--- omitted ---] >>| >>| #ifndef _ASMLANGUAGE >>| >>| /* PCI I/O function defines */ >>| >>| [--- omitted ---] >>| >>| #ifndef PCI_IN_LONG >>| #define PCI_IN_LONG(x) sysInLong (x) >>| #endif >> >>============================================================ >> >>| IMPORT UINT32 sysInLong (UINT32 address); || Line# >> > 417 > >>I changed the line above to >> >>IMPORT ULONG sysInLong (ULONG address); >> >>This action helped. >> >>However, is it safe ? >> >>============================================================ >>| >>| [--- omitted ---] >>| >>| #endif /* _ASMLANGUAGE */ >>| >>| [--- omitted ---] >>| >>| >>| || ######### File mbx800.h : END ########### >>| >>| >> >>[snip] >> >> >> >> > > [AndrayK]. I suggest to look at the assembly code generated in > both cases. This could provide definitive answer (for this > particular BSP). > > Andray > || ######### Original File mbx800.h : BEGIN ######### || Full path : D:\sys2\Tornado\target\config\mbx860\mbx800.h || Relevant Fragment IMPORT UINT32 sysInLong (UINT32 address); || Line#417 || ######### Original File mbx800.h : END ########### Corresponding assembly code is in file sysLib.s1 || ################################################## || ######### Updated File mbx800.h : BEGIN ######### || Full path : D:\sys2\Tornado\target\config\mbx860\mbx800.h || Relevant Fragment IMPORT ULONG sysInLong (ULONG address); || Line#417 || ######### Updated File mbx800.h : END ########### Corresponding assembly code is in file sysLib.s2 || ################################################## Here are results of comparison of the assembly files. %diff -c sysLib.s1 sysLib.s2 *** sysLib.s1 Sun May 12 08:49:20 2002 - --- sysLib.s2 Sun May 12 08:42:30 2002 *************** *** 2199,2206 **** or 0,0,9 mr 3,0 bl sysInLong ! mr 0,3 ! stw 0,32(31) .stabn 68,0,1157,.LM149-pciConfigInByte .LM149: addis 9,0,pciConfigAddr0@ha - --- 2199,2205 ---- or 0,0,9 mr 3,0 bl sysInLong ! stw 3,32(31) .stabn 68,0,1157,.LM149-pciConfigInByte .LM149: addis 9,0,pciConfigAddr0@ha *************** *** 2449,2456 **** or 0,0,9 mr 3,0 bl sysInLong ! mr 0,3 ! stw 0,36(31) .stabn 68,0,1240,.LM179-pciConfigInWord .LM179: addis 9,0,pciConfigAddr0@ha - --- 2448,2454 ---- or 0,0,9 mr 3,0 bl sysInLong ! stw 3,36(31) .stabn 68,0,1240,.LM179-pciConfigInWord .LM179: addis 9,0,pciConfigAddr0@ha *************** *** 2650,2657 **** addi 11,9,pciConfigAddr1@l lwz 3,0(11) bl sysInLong ! mr 0,3 ! stw 0,36(31) .stabn 68,0,1316,.LM204-pciConfigInLong .LM204: li 0,0 - --- 2648,2654 ---- addi 11,9,pciConfigAddr1@l lwz 3,0(11) bl sysInLong ! stw 3,36(31) .stabn 68,0,1316,.LM204-pciConfigInLong .LM204: li 0,0 *************** *** 2693,2700 **** or 0,0,9 mr 3,0 bl sysInLong ! mr 0,3 ! stw 0,36(31) .stabn 68,0,1324,.LM209-pciConfigInLong .LM209: addis 9,0,pciConfigAddr0@ha - --- 2690,2696 ---- or 0,0,9 mr 3,0 bl sysInLong ! stw 3,36(31) .stabn 68,0,1324,.LM209-pciConfigInLong .LM209: addis 9,0,pciConfigAddr0@ha *************** *** 2896,2903 **** or 0,0,9 mr 3,0 bl sysInLong ! mr 0,3 ! stw 0,28(31) .stabn 68,0,1384,.LM232-pciConfigOutByte .LM232: lbz 9,24(31) - --- 2892,2898 ---- or 0,0,9 mr 3,0 bl sysInLong ! stw 3,28(31) .stabn 68,0,1384,.LM232-pciConfigOutByte .LM232: lbz 9,24(31) *************** *** 3144,3151 **** or 0,0,9 mr 3,0 bl sysInLong ! mr 0,3 ! stw 0,28(31) .stabn 68,0,1456,.LM257-pciConfigOutWord .LM257: lhz 9,24(31) - --- 3139,3145 ---- or 0,0,9 mr 3,0 bl sysInLong ! stw 3,28(31) .stabn 68,0,1456,.LM257-pciConfigOutWord .LM257: lhz 9,24(31) - -- ============================ Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ============================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Link to boot multiple ethernet interface Date: Wed, 15 May 2002 10:03:46 +0000 (UTC) From: "Dexter Nixon" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: References: <80b6fd721aa44a76437bc129c318a32e.76014@mygate.mailgate.org> > What I expect that you want to do is have multiple ethernet controllers > starting up during the boot instead of only one. > I don't think this is documented. If you want to do this you have to > modify the bootrom code. > > Also if you want to do this from the application you have to initialise > your network yourself. But it is not very difficult: > ipAttach (, ); > usrNetIfConfig (, , , , > ); > > So for example: > ipAttach (1,"fei"); > usrNetIfCOnfig("fei", 1, "10.20.30.40","Interface2", 0xffffff00); > > But you need to configure your interface in configNet.h first. > ================================================================== Kindly show me the pointers where i can find, configuring a interface in COnfigNet.h. How to do this. Is there any example available. The faq also pointing to a example which we are bringing up through a function which assumes that ConfigNet.h is configured. In configNet.h where we have to add , and what to add. Kindly explain me in detail. Thanks, Nix - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.arch.embedded,comp.os.linux.embedded,comp.os.linux.powerpc,comp.os.vxworks,comp.realtime,comp.sys.powerpc.advocacy,comp.sys.powerpc.misc,comp.sys.powerpc.tech Subject: MPC860/MPC8260 Question Date: Wed, 15 May 2002 13:48:18 +0200 From: "Rami Gideoni" Organization: Internet Gold, ISRAEL Message-ID: Hi, The following question refers to MPC860 PQ and MPC8260 PQII: The problem : the user's manual states in the R (Ready) bit description : - ---------------------------------------------------------------------------- - ------------------------------------------------------- Ready. 0 The buffer is not ready for transmission. Both the buffer and the BD can be updated. The CPM clears R after the buffer is sent or an error is encountered. 1 The buffer has not been sent or is being sent and the BD cannot be updated. - ---------------------------------------------------------------------------- - ------------------------------------------------------- We would like to conduct continuous transmission of filler buffer (using the Tx BD CM bit) and to clear it when we are ready with "real" data. Can I modify the CM bit during the Tx BD transmission ? and if yes, will this modification will take into effect after the entire current transmission session completed ? In other words : Does the CPM reads the control/status word in the beginning of each transmission or does it read it once, store it in a scratch register and ignores from the BD status/control register until error occurred, STOP TX occurred or transmission was disabled by the GSMR register ? Thanks in advance Rami Gideoni --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 15 05:43:38 2002 From: "Joe Welfeld" Date: Wed May 15 05:43:40 PDT 2002 Subject: Tornado 2.02/Diab and Dependencies Hello, I have searched through the archives and tried about everything I could find to resolve this dependency issue with no luck. Hopefully someone can help. I am running Windows version of Tornado202 with Diab4.4b PPC compiler for most of my projects. It seems that the projects that use the Diab compiler cannot build a dependency list. If I select Advanced Dependencies mode and remove the Defaults it finds all the system dependencies, but for my own include files it doesn't seem to work. None of my .h files appear in the dependency list. I've added my INCLUDES as MACROS to build, added the MACRO's to VPATH checked/rechecked that VPATH is setup correctly. The files compile OK so the paths seem to be alright, but none of my file show up in the dependency list. Thanks, Joe Welfeld From vxwexplo-errs@csg.lbl.gov Wed May 15 21:45:22 2002 From: "Sabarivel S" Date: Wed May 15 21:45:24 PDT 2002 Subject: Vxworks : Interface up/down notification Is there a way in Vxworks, for the protocols running above Vxworks IP to get notification of the interface parameters changes - mainly up/down events ? Pl give me some pointers on these. thanks & regards sabari *************************************************************************** This message is proprietary to Future Software Limited (FSL) and is intended solely for the use of the individual to whom it is addressed. It may contain privileged or confidential information and should not be circulated or used for any purpose other than for what it is intended. If you have received this message in error, please notify the originator immediately. If you are not the intended recipient, you are notified that you are strictly prohibited from using, copying, altering, or disclosing the contents of this message. FSL accepts no responsibility for loss or damage arising from the use of the information transmitted by this email including damage from virus. *************************************************************************** From vxwexplo-errs@csg.lbl.gov Thu May 16 04:03:15 2002 From: Vxworks Exploder Date: Thu May 16 04:03:18 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu May 16 04:03:11 PDT 2002 Subject: Re: Tornado II - Printing Source Files Subject: incomplete bootrom tftp download Subject: Re: incomplete bootrom tftp download Subject: Re: Link error while working with SBC8260 Subject: Re: Interrupt Latency Subject: Re: incomplete bootrom tftp download Subject: How to do a MGET on a vxworks FTPD ? Subject: Re: incomplete bootrom tftp download Subject: incomplete bootrom tftp download Subject: Re: what are the BSD driver and END driver? Subject: Chip select configuration Subject: Re: Chip select configuration Subject: Re: incomplete bootrom tftp download Subject: Re: incomplete bootrom tftp download Subject: sandpoint /w PPC755 bringup experience?? Subject: Re: Tornado II - Printing Source Files Subject: Re: Tornado II - Printing Source Files Subject: Re: sandpoint /w PPC755 bringup experience?? Subject: Exception handling in interrupt Subject: Question on BootCmdLoop() Subject: Re: Polymorphism and Destructor Subject: Re: Anyone using target Intel x86 (not PENTIUM) Subject: Re: Question on BootCmdLoop() Subject: Re: Chip select configuration Subject: Re: routing table Subject: 860 B3 cache Subject: An urgent question! Subject: Question about Interrupt! Subject: Re: Question about Interrupt! ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II - Printing Source Files Date: Wed, 15 May 2002 08:13:55 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3CE27B33.9D1A75B4@lmco.com> References: <10bfb1a3.0205150148.70df153@posting.google.com> Ralf Hager wrote: > > When i print source files from Tornado IDE the output always is in > large font size about 36. How can i configure printing to use normal > font size? > > Ralf > ralf@fh-rosenheim.de why don't you use the os commands dirctly instead of the IDE? GUI interfaces are in general a poor way to do things anyway. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: incomplete bootrom tftp download Date: Wed, 15 May 2002 12:21:34 GMT From: "Mike Hanna" Organization: Road Runner - NC Message-ID: Reply-To: "Mike Hanna" I'm attempting to use bootrom to download a bootable vxWorks image with tftp. I've built the bootable image based on my bsp and at the end of the build it indicates that the size is 900k although the actual file size is 1.3 MB. During boot, the bootrom downloads 900k of the file and jumps to the starting point and never gets up and running. The host knows that the file isn't finished and retransmits the last unacknowleged packet a few times. Any ideas? Any help would be appreciated. Thanks, - -- Mike Hanna Vapid Software www.vapidsoftware.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: incomplete bootrom tftp download Date: Wed, 15 May 2002 15:02:36 GMT From: "Mike Hanna" Organization: Road Runner - NC Message-ID: References: <3CE26DD7.E591624F@ineoquest.com> Reply-To: "Mike Hanna" I switched to ftp and the full file download occurs, but the image doesn't come up. Looks like you are right and the problem is in the bootable image itself. Thanks. "George Varndell" wrote in message news:3CE26DD7.E591624F@ineoquest.com... > Mike Hanna wrote: > > > > I'm attempting to use bootrom to download a bootable vxWorks image with > > tftp. I've built the bootable image based on my bsp and at the end of the > > build it indicates that the size is 900k although the actual file size is > > 1.3 MB. During boot, the bootrom downloads 900k of the file and jumps to > > the starting point and never gets up and running. The host knows that the > > file isn't finished and retransmits the last unacknowleged packet a few > > times. Any ideas? > > I just noticed this same thing the other day. Apparently, vx doesn't > need the entire file but the server wants to finish sending it anyway. > I don't think this is causing your image to not come up because, in my > case, the image comes up despite the server's confusion. > You'll need to debug the loaded image starting from its entry point. > > Regards, > George Varndell --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Link error while working with SBC8260 Date: Sun, 12 May 2002 16:02:30 GMT From: "drdiags" Organization: AT&T Broadband Message-ID: References: <5288f97d.0205100816.46bf3153@posting.google.com> <5288f97d.0205120637.43a7ac1f@posting.google.com> Hello Danny, To add *.o files to the vxWorks kernel from the command line build, you need to set the EXTRA_MODULES macro in the BSP makefile. If you wanted to allow for the bootrom to pick up the *.o file, then add this file to your MACH_EXTRA macro in the BSP makefile. For adding the file to the build from the Tornado IDE, select the build properties for your vxWorks kernel and select the macros tab while accessing the build properties compile options. Use the EXTRA_MODULES macro to add the full path to the *.o file. As a note, it is suggested that when you don't have the source, only the *.o that you rename it to *.obj. This is because the make clean in the BSP or Rebuild-All in the IDE "may" remove all *.o files in the BSP directory for the BSP build and from the project directory in the IDE build. Just an FYI, there are several ways around this, so I thought I would just mention what you probably already know. HTH. "Danny Cohen" wrote in message news:5288f97d.0205120637.43a7ac1f@posting.google.com... > Thanks Graham. > > It is supplied as motfccend.obj and motfccend.h files with the BSP. > How can you add it manually? which makefile i have to touch? > How can yo add it from the Tornado IDE? > > > "Graham Baxter" wrote in message news:... > > Danny, > > > > Add the object motfccend.o to your project. > > > > It may be supplied as just an object file. > > > > I hope this helps. > > > > -- > > > > Regards, > > > > Graham Baxter > > Freelance Software Engineer > > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > > > > "Danny Cohen" wrote in message > > news:5288f97d.0205100816.46bf3153@posting.google.com... > > > Hi. > > > > > > I have evaluation board SBC8260. > > > I'm trying to create bootable image and download it to the board. > > > while compiling and linking the BSP i face the following error: > > > "motFccEndLoad undefined at PartialImage.o". My feeling it has > > > something with the BSP configuration. Can anyone help me to overcome > > > this problem? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Interrupt Latency Date: 15 May 2002 15:49:18 GMT From: Charly Message-ID: <2002515-164918-194448@foorum.com> References: Hello, It could be the result of a DMA access from your hardware. Charly - -- Use our news server 'news.foorum.com' from anywhere. More details at: http://nnrpinfo.go.foorum.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: incomplete bootrom tftp download Date: Wed, 15 May 2002 16:17:58 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CE28A92.3080502@spamm.me.l8s.co.uk> References: Mike Hanna wrote: > I'm attempting to use bootrom to download a bootable vxWorks image with > tftp. I've built the bootable image based on my bsp and at the end of the > build it indicates that the size is 900k although the actual file size is > 1.3 MB. During boot, the bootrom downloads 900k of the file and jumps to > the starting point and never gets up and running. The host knows that the > file isn't finished and retransmits the last unacknowleged packet a few > times. Any ideas? Sounds like the image is overwriting the code that is loading it.... You need to build new boot roms with a larger RAM_HIGH_ADRS. David --------------------------- Newsgroups: comp.os.vxworks Subject: How to do a MGET on a vxworks FTPD ? Date: 15 May 2002 09:34:14 -0700 From: mohasin.zaki@wipro.com (Mohasin) Organization: http://groups.google.com/ Message-ID: Hi, I want to connect to the vxWorks FTP server running on a target board and do multiple get of files using "mget *.ext" ? I am able to do a "get", "put" and "mput". But it looks like the vxWorks FTP server doesn't support "mget". Is there any way I can acheive this?. Thannkz regards, MOhasin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: incomplete bootrom tftp download Date: Wed, 15 May 2002 14:21:12 GMT From: "dan" Organization: Prodigy Internet http://www.prodigy.com Message-ID: References: Reply-To: "dan" The bootrom runs out of RAM starting at RAM_HIGH_ADRS. It loads VxWorks starting at RAM_LOW_ADRS. If the kernel is too large, it will load right over the RAM copy of the bootrom. You may have to move RAM_HIGH_ADRS higher. - - dan Boot sequence flow through ROM and RAM - -------------------------------------- ROM RAM _______________ --------------- | romInit() | | |LOCAL_MEM_LOCAL_ADRS |---------------| | | | romStart() | |---------------|RAM_LOW_ADRS |---------------| | V |-sysInit()-used at reboot | boot |-- | x | | monitor | | | W | | | | | O | |---------------| | | R | | | | | K | | | | | S | | | | |---------------|FREE_RAM_ADRS | | | | | | | | | | | | | |---------------|RAM_HIGH_ADRS | | |---->| boot |-usrInit() | | | monitior| | | |(decompress |-kernenInit() | | | if | | | | necessary) | /tyCo0 | | | |-usrRoot() ---- | | | |-------------->| | --------------- --------------- serial | | ---- - ----- Original Message ----- From: "Mike Hanna" Newsgroups: comp.os.vxworks Sent: Wednesday, May 15, 2002 7:21 AM Subject: incomplete bootrom tftp download > I'm attempting to use bootrom to download a bootable vxWorks image with > tftp. I've built the bootable image based on my bsp and at the end of > the build it indicates that the size is 900k although the actual file > size is 1.3 MB. During boot, the bootrom downloads 900k of the file and > jumps to the starting point and never gets up and running. The host > knows that the file isn't finished and retransmits the last > unacknowleged packet a few times. Any ideas? > > Any help would be appreciated. > > Thanks, > > -- > Mike Hanna > Vapid Software > www.vapidsoftware.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what are the BSD driver and END driver? Date: Wed, 15 May 2002 11:03:32 -0700 From: "Henry Choi" Message-ID: References: <7384051c.0204232355.4a4b7098@posting.google.com> "Hwa Jin Bae" wrote in message news:ucfdr7bo7edt82@corp.supernews.com... > with robustness of a network driver in face of critical situations such as > when there are too many packets to be handled too fast, buffer exhaustion, > careful management of loaned buffers, etc. Hwa Jin, are you referring to the problem of the stack locking up when it runs out of the driver mbuf pool? If so, I have seen this problem and am curious why it occurs. Wind engineer suggested that fragmented IP packets eat up the available mbuf and won't be freed because the stack seems to be locked waiting for an mbuf. Unless Vx stack doesn't do periodic clean up of unassembled IP fragments at regular IP timeout, I don't see how sending many fragments fast can cause permanent deadlock, so I must be missing something. Also according to Wind, T202 or above system (also T2CP3) should and will recover from this condition, but I have not yet had a chance to confirm or deny this. What causes the END drivers to exhibit this deadlock behavior where BSD driver doesn't? 2 empirical evidence make me doubt whether this is END specific problem: 1. By increasing the stack (NOT the driver) mbuf pool, we can alleviate the outtage problem. This seems to contradict the theory that the problem stems from END drivers loaning the buffer to the stack. 2. We have run into the same problem with BSD drivers--not as often, but ultimately the same symptom. --------------------------- Newsgroups: comp.os.vxworks Subject: Chip select configuration Date: 15 May 2002 11:24:09 -0700 From: mora@computer.org (Marco Mora) Organization: http://groups.google.com/ Message-ID: Hi, I wanted to define an address range I edit in sysLib.c sysPhyMemDesc adding { (void *)0x20000000 (void *)0x20000000 0x10000 VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | VM_STATE_MASK_CACHEABLE, VM_STATE_VALID | VM_STATE_WRITABLE | VM_STATE_CACHEABLE_NOT }, ... now I want to generate a chip select signal when access this range, where do I have to define it????? Thank you Marco --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Chip select configuration Date: Wed, 15 May 2002 14:34:09 -0400 From: George Varndell Organization: Newsfeeds.com http://www.newsfeeds.com 100,000+ UNCENSORED Newsgroups. Message-ID: <3CE2AA20.53566FB8@ineoquest.com> References: Marco Mora wrote: > > Hi, > > I wanted to define an address range I edit in sysLib.c > sysPhyMemDesc adding > > { > (void *)0x20000000 > (void *)0x20000000 > 0x10000 > VM_STATE_MASK_VALID | VM_STATE_MASK_WRITABLE | > VM_STATE_MASK_CACHEABLE, > VM_STATE_VALID | VM_STATE_WRITABLE | > VM_STATE_CACHEABLE_NOT > }, > ... > > now I want to generate a chip select signal when access this range, > where do I have to define it????? > > Thank you > Marco Look for hardware/anyProcessor/chipSelect.h, should be there. Regards, George Varndell --------------------------- Newsgroups: comp.os.vxworks Subject: Re: incomplete bootrom tftp download Date: Wed, 15 May 2002 18:56:07 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CE2AF8C.9020702@spamm.me.l8s.co.uk> References: > Boot sequence flow through ROM and RAM > -------------------------------------- > > ROM RAM > _______________ --------------- > | romInit() | | |LOCAL_MEM_LOCAL_ADRS > |---------------| | | > | romStart() | |---------------|RAM_LOW_ADRS > |---------------| | V |-sysInit()-used at reboot > | boot |-- | x | > | monitor | | | W | > | | | | O | > |---------------| | | R | > | | | | K | > | | | | S | > | | | |---------------|FREE_RAM_ADRS > | | | | | > | | | | | > | | | |---------------|RAM_HIGH_ADRS > | | |---->| boot |-usrInit() > | | | monitior| > | | |(decompress |-kernenInit() > | | | if | > | | | necessary) | /tyCo0 > | | | |-usrRoot() ---- > | | | |-------------->| | > --------------- --------------- serial | | > ---- A nice picture with 2 minor faults! 1) memory maps are usually drawn with 0 at the bottom.... 2) if the boot code is compressed there is an additional stage, the decompress code is copied to RAM_LOW_ADRS before generating the boot monitor at RAM_HIGH_ADRS (this leaves the possibility of the decompresser overwriting its own data area! Or more likely there not being enough space between RAM_HIGH_ADRS and the end of physical memory when booting a compressed (not bootstrap) image to RAM_LOW_ADRS) David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: incomplete bootrom tftp download Date: Wed, 15 May 2002 19:14:09 GMT From: "Mike Hanna" Organization: Road Runner - NC Message-ID: <5syE8.102584$YQ1.34662592@typhoon.southeast.rr.com> References: <3CE2AF8C.9020702@spamm.me.l8s.co.uk> Reply-To: "Mike Hanna" I can successfully execute the "l" command from the bootrom which loads the bootable image into memory. Afterwards, the bootrom is still functional & I can display memory, etc. I've tried moving RAM_HIGH_ADRS around, but it doesn't make any difference. It always hangs after it tries to start the downloaded image. Initially, I thought it was because I wasn't getting the full image. I've switched to ftp, verified that the full image is xferred and the image still doesn't come up. Thanks, Mike "David Laight" wrote in message news:3CE2AF8C.9020702@spamm.me.l8s.co.uk... > > > Boot sequence flow through ROM and RAM > > -------------------------------------- > > > > ROM RAM > > _______________ --------------- > > | romInit() | | |LOCAL_MEM_LOCAL_ADRS > > |---------------| | | > > | romStart() | |---------------|RAM_LOW_ADRS > > |---------------| | V |-sysInit()-used at reboot > > | boot |-- | x | > > | monitor | | | W | > > | | | | O | > > |---------------| | | R | > > | | | | K | > > | | | | S | > > | | | |---------------|FREE_RAM_ADRS > > | | | | | > > | | | | | > > | | | |---------------|RAM_HIGH_ADRS > > | | |---->| boot |-usrInit() > > | | | monitior| > > | | |(decompress |-kernenInit() > > | | | if | > > | | | necessary) | /tyCo0 > > | | | |-usrRoot() ---- > > | | | |-------------->| | > > --------------- --------------- serial | | > > ---- > > > A nice picture with 2 minor faults! > > 1) memory maps are usually drawn with 0 at the bottom.... > > 2) if the boot code is compressed there is an additional stage, > the decompress code is copied to RAM_LOW_ADRS before generating > the boot monitor at RAM_HIGH_ADRS > (this leaves the possibility of the decompresser overwriting its > own data area! Or more likely there not being enough space between > RAM_HIGH_ADRS and the end of physical memory when booting a > compressed (not bootstrap) image to RAM_LOW_ADRS) > > David > --------------------------- Newsgroups: comp.os.vxworks Subject: sandpoint /w PPC755 bringup experience?? Date: Wed, 15 May 2002 19:25:41 GMT From: "Mike Hanna" Organization: Road Runner - NC Message-ID: Reply-To: "Mike Hanna" Anybody out there have experience bringing up a Motorola Sandpoint board with a Altimus PPC755 processor board attached (tornado 2.1 with sp7400 bsp). I'm able to get the bootrom working, but the downloaded image never comes up. I don't have any type of hardware debugger to see what is happening. I'd love to the configuration for a working system (config.h and makefile for the bsp and the .prj for the bootable image). Any ideas or information would be appreciated. Thanks, - -- Mike Hanna Vapid Software www.vapidsoftware.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II - Printing Source Files Date: Wed, 15 May 2002 15:19:13 +0200 From: "Rene Straub" <_remove_rene.straub@yetnet.ch> Organization: Swisscom IP+ (post doesn't reflect views of Swisscom) Message-ID: References: <10bfb1a3.0205150148.70df153@posting.google.com> Ralf, I ran into the very same problem today. I can't tell you how to change the font size. What I do is to copy/paste the code into my favourite text editor and print it from there. Let me know if you hear about a better way. Yours - - Rene "Ralf Hager" wrote in message news:10bfb1a3.0205150148.70df153@posting.google.com... > When i print source files from Tornado IDE the output always is in > large font size about 36. How can i configure printing to use normal > font size? > > Ralf > ralf@fh-rosenheim.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado II - Printing Source Files Date: 15 May 2002 17:21:05 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205151621.50f744c9@posting.google.com> References: <10bfb1a3.0205150148.70df153@posting.google.com> Hello, If you are restricted to Windoze, I'd recommend PrintFile (http://www.lerup.com/printfile/) which is a simple drag & drop box on the screen that lets you set up a format for the text (e.g. two columns), and then print to your heart's content very easily. Of course, those on Unix systems already have a number of tools for doing this sort of thing... TTFN, John... "Rene Straub" <_remove_rene.straub@yetnet.ch> wrote in message news:... > Ralf, > > I ran into the very same problem today. I can't tell you how to change > the font size. What I do is to copy/paste the code into my favourite > text editor and print it from there. Let me know if you hear about a better > way. > > Yours > - Rene > > > "Ralf Hager" wrote in message > news:10bfb1a3.0205150148.70df153@posting.google.com... > > When i print source files from Tornado IDE the output always is in > > large font size about 36. How can i configure printing to use normal > > font size? > > > > Ralf > > ralf@fh-rosenheim.de --------------------------- Newsgroups: comp.os.vxworks Subject: Re: sandpoint /w PPC755 bringup experience?? Date: Thu, 16 May 2002 01:33:10 GMT From: "Mike Hanna" Organization: Road Runner - NC Message-ID: References: Reply-To: "Mike Hanna" Nevermind. Its working now. "Mike Hanna" wrote in message news:VCyE8.102689$YQ1.34674056@typhoon.southeast.rr.com... > Anybody out there have experience bringing up a Motorola Sandpoint board > with a Altimus PPC755 processor board attached (tornado 2.1 with sp7400 > bsp). I'm able to get the bootrom working, but the downloaded image never > comes up. I don't have any type of hardware debugger to see what is > happening. > > I'd love to the configuration for a working system (config.h and makefile > for the bsp and the .prj for the bootable image). Any ideas or information > would be appreciated. > > Thanks, > > -- > Mike Hanna > Vapid Software > www.vapidsoftware.com > > --------------------------- Newsgroups: comp.os.vxworks Subject: Exception handling in interrupt Date: 15 May 2002 18:47:39 -0700 From: colintelfer@hotmail.com (CT) Organization: http://groups.google.com/ Message-ID: <68b528d4.0205151747.5b1072b4@posting.google.com> Hi, I am having problems with my 860 running vxWorks. Once in a while during regular operation I see a spurious reset, with the following exception data (sysExcMsg) during the reboot. Exception at Interrupt Level: Implementation Dependent Software Emulation Exception current instruction address: 0x00bf8e44 Machine status register: 0x00009000 Condition register" 0x28000080 regs at 0x1f40c40 The instruction pointer has gone awry and is pointing at some random data. I am guessing that I have corrupted the stack. Can someone help me with the final line of information here "regs at". From what I gather this is some kind of pointer to the stack. Any information would be greatly appreciated. Thanks, Colin --------------------------- Newsgroups: comp.os.vxworks Subject: Question on BootCmdLoop() Date: 15 May 2002 19:10:29 -0700 From: jade_chen@21cn.com (shy) Organization: http://groups.google.com/ Message-ID: <937a20a8.0205151810.34e7c98f@posting.google.com> In bootCmdLoop(), there is a local varial FUNCPTR entry , I cannot find where it gets its value.can anyone help me with this? see some code here. {... case '@': /* load and go with internal params */ case '$': /* load and go with internal params */ if (bootLoad (pLine, &entry) == OK) { go (entry); } else { taskDelay (sysClkRateGet ()); /* pause a second */ reboot (BOOT_NO_AUTOBOOT); /* something is awry */ } break; } I cannot find how the entry get it value. pls help me with this. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Polymorphism and Destructor Date: Mon, 13 May 2002 08:52:29 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: References: <8bb5523.0205121836.5b8b1432@posting.google.com> Hello, which version of vxWorks and which Fixpacks and which HW? we use 2.0 and use such constructs some hundreds of time without problems. - -- regards, Mario Semo. http://www.kirchnersoft.com Patrick Boulay wrote in message news:8bb5523.0205121836.5b8b1432@posting.google.com... > Hello, > > I have a big problem with polymorphism and destructor. See the class > in example bellow, the vxworks crash after I delete the pointer. > > class A > { > public: > A(); > virtual ~A(); > }; > class B:public A > { > public: > B(); > virtual ~B(); > }; > int aFunction() > { > A *ptr; > > ptr = new B; > > delete ptr; > > return 0; > } > > > > Thanks > Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Anyone using target Intel x86 (not PENTIUM) Date: Mon, 13 May 2002 08:56:57 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: References: <97e6274e.0205080100.3e7e1d13@posting.google.com> <3cd93f65.4871064@news.mhogaming.com> we support x486 and 386 too and we cannot upgrade due to the same reason. and more: WindRiver does not accept bug reports against T2.0 anymore (we have a maintainance contract). - -- regards, Mario Semo. http://www.kirchnersoft.com Ignacio Gonzalez T. wrote in message news:3cd93f65.4871064@news.mhogaming.com... > On 8 May 2002 02:00:18 -0700, heiko_elger@arburg.com (Heiko Elger) > wrote: > > >Hello, > > > >who is using a target Intel x86 (386, 486, not Pentium, PcPentium Pro) > >with Tornado 2.0.x? > > > >I'm asking cause in the new Tornado Version 2.2 (VxWorks 5.5) only > >Intel processors Pentium, Pentium II, Pentium III, Pentium IV will be > >supported. > >(the new official roadmap of WRS) > > > >For customer with support contract who uses no Pentium or above there > >will be no ugrade posibbility and only minor bug fixes for Tornado > >2.0.x > > > >I'm asking cause we have support contract and we are using x86 and we > >have no possibility to use a Pentium next time. > > > >We think that cause of our support contract based on a Intel 80486 > >processor WRS must support x86 in Tornado 2.2. > > > >We want to know whether there are other projects out there which are > >based on Intel x86 or whether we are the only company using x86. > > > >best regards > > > > Heiko Elger > > Hello, Heiko. > > We are using Tornado 2.0.2 and VxWorks 5.4.2 with 386EX targets since > 2000. After 1 year with support contract, we decided to leave it, > because of the percieved low quality/cost ratio (of course, I agree > this is a very subjective appreciation.) > > We will continue to use these targets (and perhaps another ones, > mainly 486DX based) instead of Pentium ones, because of cost > considerations (we manufacture our own boards, or use cheap standard > ones.) > > So we will have to stick to Tornado 2.0.2, even if someday we decide > to upgrade... My condolence. These news are bad news for us too. > > By the way, do you know if Tornado 2.2 comes with gcc > 2.7.2 ? > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question on BootCmdLoop() Date: Thu, 16 May 2002 08:38:20 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <937a20a8.0205151810.34e7c98f@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@pd9524f66.dip.t-dialin.net Hi, what's about bootLoad (pLine, &entry) ? bootLoad() mounts your boot device, loads your image named in pLine and returns the start address in entry. Michael "shy" schrieb im Newsbeitrag news:937a20a8.0205151810.34e7c98f@posting.google.com... > In bootCmdLoop(), there is a local varial FUNCPTR entry , I cannot > find where it gets its value.can anyone help me with this? see some > code here. > {... > case '@': /* load and go with internal params */ > case '$': /* load and go with internal params */ > > if (bootLoad (pLine, &entry) == OK) > { > go (entry); > } > else > { > taskDelay (sysClkRateGet ()); /* pause a second */ > reboot (BOOT_NO_AUTOBOOT); /* something is awry */ > } > break; > } > > I cannot find how the entry get it value. pls help me with this. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Chip select configuration Date: Thu, 16 May 2002 08:38:44 +0200 From: Emmanuel Herbreteau Organization: SEPRO Robotique (BEP) Message-ID: <3CE353F4.2A2D7AD1@sepro-robotique.com> References: Hi, Marco Mora wrote: > I wanted to define an address range I edit in sysLib.c > sysPhyMemDesc adding....... This array is used to setup the MMU (required on our PowerPC target to enable the data cache and the basic MMU support). > now I want to generate a chip select signal when access this range, > where do I have to define it????? On PowerPC, this is done in "RomInit.s" where you must correctly load the BR/OR registers. It's a part of the basic hardware intialization (clock, devices, chipselect, SDRAM, etc...) Regards Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routing table Date: 16 May 2002 00:59:56 -0700 From: vijay.peshkar@wipro.com (Vijay Peshkar) Organization: http://groups.google.com/ Message-ID: References: <68c7b7f1.0205140554.7fbca9c1@posting.google.com> Rajesh, You can use MIB2 routines. > 1.how do get the default route, Use routine m2IpRouteTblEntryGet() with 'search' set to M2_EXACT_VALUE. > > 2.Is there any system call for updating the elements of a > route(metrics) > Is there any way I can add or Delete Route other than the system > calls found in routeLib ? You can use routine m2IpRouteTblEntrySet() to add/delete route. Also rtrequest() can be used. > > 3.How do I access the Routing Table .(like first route ,next route) Using routine m2IpRouteTblEntryGet() with 'search' set to M2_EXACT_VALUE and index set to zero you can get the first entry. Subsequent entries can be got with 'search' set to M2_NEXT_VALUE. > > 4. How is the routing table maintained. > > 5. m2IpLib mentions some system call for setting and retrieving > routing Table entry.By setting > the MIB-2 routing entry ,will the kernel routing entry be set for > the same route? > Can I assume that routes in MIB-2 routing entry and kernel routing > entry are the same. AFAIK, they aren't. MIB2 has lot of 'objects' specified in MIB2 RFC 1213 not present in general routing table and vice versa. Hope this help. Cheers, Vijay --------------------------- Newsgroups: comp.os.vxworks Subject: 860 B3 cache Date: 16 May 2002 01:08:40 -0700 From: lionjackyfish@sina.com.cn (JackyLiu) Organization: http://groups.google.com/ Message-ID: <425faaf3.0205160008.13ed68e1@posting.google.com> I use 860t B3, and tornado 2.0, would anyone tell me how can i enable the D-CACHE? someone tell me i can download a patch from WINDRIVER and i must add some compiler options , but first i can't find the patch , and i don't know how do i add which option too, would anyone tell me ? thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: An urgent question! Date: 16 May 2002 02:39:35 -0700 From: zhoul@blrcsv.china.bell-labs.com (zlybupt) Organization: http://groups.google.com/ Message-ID: <989cdae4.0205160139.5bf4a2bb@posting.google.com> hi all: I met a question about interrupt,and I am really confused by the following concepts:interrupt level,interrupt vector and interrupt num,can you give an explanation about them? And which one should I use for the function intConnect? BTW:When I remove the cable from the ethernet card,system tells me : interrupt: Uninitialized interrupt: vector number: 42(0-255) ... but I find in the source code INUM_TO_IVEC which is shift left the num of 3 bits,how can one intnum to be shifted to 42?? Thank you in advance for your answers! --------------------------- Newsgroups: comp.os.vxworks Subject: Question about Interrupt! Date: 16 May 2002 02:48:22 -0700 From: zhoul@blrcsv.china.bell-labs.com (zlybupt) Organization: http://groups.google.com/ Message-ID: <989cdae4.0205160148.4400245b@posting.google.com> hi,all Is there any one show can give me an explanation about the following objects: 1.interrupt number 2.interrupt vector 3.interrupt level and which one should be used for the function intConnect? BTW:when I remove the cable from my ethernet card,system tells me: interrupt: Uninitialized interrupt vector number 42(0-255) I wonder how can an intnum be shifted to 42 by the macro INUM_TO_IVEC??? Thank you for your answers in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about Interrupt! Date: Thu, 16 May 2002 06:24:40 -0400 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3CE388E8.C28DCED9@bellsouth.net> References: <989cdae4.0205160148.4400245b@posting.google.com> zlybupt wrote: > hi,all > Is there any one show can give me an explanation about the following objects: > 1.interrupt number > 2.interrupt vector > 3.interrupt level > and which one should be used for the function intConnect? > BTW:when I remove the cable from my ethernet card,system tells me: > interrupt: > Uninitialized interrupt > vector number 42(0-255) > I wonder how can an intnum be shifted to 42 by the macro INUM_TO_IVEC??? > Thank you for your answers in advance! The documentation is a little flakey in this area. Interrupt level in VME boxes, refers to the interrupt line number, 1-7, on which the device has been told to interrupt. In most devices, you set (by software) an interrupt vector to be returned by the device when an IACK is received. That vector is what the macro INUM_TO_IVEC is given. At least on my system (68K series) the message you refer to actually gives the same number as what is programmed into the device, whic is what the macro calls the interrupt number. In 68K, the output of the macro is the address in memory where the processor will look for the address of the routine to execute. Your system may or may not be the same as mine, because of differences between systems, etc. Speaking only for myself, Joe Durusau --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu May 16 07:12:12 2002 From: =?iso-8859-1?q?ram=20mohan?= Date: Thu May 16 07:12:14 PDT 2002 Subject: Re: Tornado 2.02/Diab and Dependencies Hi joe, even if u cannot find ur own header files in the list try compiling, if it is done no prob one more thing i feel may help u is one of my observations while working with "torn"ado it is better to avoid putting directory names that contain spaces like "my proj" "torn"ado gives u prob hope this hellps u in someway Regards Mohan --- the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Wed May 15 > 05:43:38 2002 > Submitted-by: "Joe Welfeld" > > > Hello, > > I have searched through the archives and tried about > everything I could find > to resolve this dependency issue with no luck. > Hopefully someone can help. > I am running Windows version of Tornado202 with > Diab4.4b PPC compiler for > most of my projects. It seems that the projects > that use the Diab compiler > cannot build a dependency list. If I select Advanced > Dependencies mode and > remove the Defaults it finds all the system > dependencies, but for my own > include files it doesn't seem to work. None of my > .h files appear in the > dependency list. I've added my INCLUDES as MACROS > to build, added the > MACRO's to VPATH checked/rechecked that VPATH is > setup correctly. The files > compile OK so the paths seem to be alright, but > none of my file show up in > the dependency list. > > > Thanks, > Joe Welfeld > > > > ********** > > This is a user group mailing list for vxWorks > related topics > see http://www-vxw.lbl.gov/vxworks for posting > and subscription info __________________________________________________ Do You Yahoo!? LAUNCH - Your Yahoo! Music Experience http://launch.yahoo.com From vxwexplo-errs@csg.lbl.gov Fri May 17 04:03:19 2002 From: Vxworks Exploder Date: Fri May 17 04:03:22 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri May 17 04:03:14 PDT 2002 Subject: Seeking possible cause of "Can't load boot file" Subject: Re: PS2 Port on MVME1604. Subject: WDB Packet generate Exception Subject: Re: WDB Packet generate Exception Subject: embedded web server-loading images!! Subject: Re: embedded web server-loading images!! Subject: Re: cross compiling gcc - parse error Subject: Re: embedded web server-loading images!! Subject: Free Evaluation Report VxWorks AE 1.1 Subject: Re: Question about Interrupt! Subject: Re: Accessing hard disk Subject: Re: ioctl problem for VxWorks Subject: Re: "RPCExit: extra call" error message Subject: Polymorphism and Destructor Subject: Re: Polymorphism and Destructor Subject: Re: Difference between RTOS and normal OS Subject: Re: Does DOSFS 2.0 for PPC supports device whose block size != 512 ???? Subject: why doesn't kill(TID, 9) work? Subject: Re: Compilation for MBX860 target failed Subject: Re: Compilation log file Subject: Re: Compilation log file Subject: Re: Polymorphism and Destructor Subject: Q: boost::shared_ptr and thread-safety Subject: RTM_GET does not fetch Gateway of given destination Subject: Re: RAM_LOW_ADRS parameter in T2 Subject: Re: Polymorphism and Destructor Subject: Dying in kernelInit Subject: Re: embedded web server-loading images!! Subject: C and C++ object files Subject: Send Multicast message Subject: Re: Dying in kernelInit Subject: Re: why doesn't kill(TID, 9) work? Subject: Re: incomplete bootrom tftp download Subject: Re: How to do a MGET on a vxworks FTPD ? Subject: Re: embedded web server-loading images!! Subject: Re: C and C++ object files Subject: Newbie questions ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Seeking possible cause of "Can't load boot file" Date: 11 May 2002 06:56:34 -0700 From: cozzy@263.net (yyzhong) Organization: http://groups.google.com/ Message-ID: I boot a x86 target through network. It displays: Attached TCP/IP interface to elt0 Attaching network interface lo0...done. Loading... After a long period of time, it continues with: Error loading file: errno = 0xd0003. Cannot load boot file! /*The last prompt maynot be the original text*/ I'm sure the physical connection is good. So I think taht maybe the network interface is not working in fact. Then how can I query its status without the booted OS? I think that if I boot it from floppy disk with the network card driver included, I can use the network card also. But the result of "ifShow" is "undefined symbol: ifShow" . I'm a novice at vxworks. Any help would be appreciated. Thanks a lot. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PS2 Port on MVME1604. Date: Thu, 16 May 2002 07:39:54 -0400 From: "William Dennen" Message-ID: <3ce39b32$0$24353@dingus.crosslink.net> References: <3CE0C1A6.50092F63@here.com> The keyboard/mouse port is controlled by the PC87303VUL SuperIO device. You'll need to use the datasheet for it (@ http://www.national.com/pf/PC/PC87303.html) and the MVME1600 Programmer's Guide (@ http://www.mcg.mot.com/literature). - -- >@< Bill Dennen MCG/GTAC "An Engineer" wrote in message news:3CE0C1A6.50092F63@here.com... > I would like to read the output of a PS2 device attached to the PS2 > port of a Power PC (MVME1604). I however, cannot find any information > on how to access the PS2 port of the MVME1604. Has anyone had any > experiance of reading from the PS2 port of a Power PC? Do I simply need > to read from a certain port number as I would do if reading from a > serial port? > > Thanks > > An Engineer. > > > --------------------------- Newsgroups: comp.os.vxworks Subject: WDB Packet generate Exception Date: 16 May 2002 05:00:16 -0700 From: tenshou0214@sina.com (Gabriel) Organization: http://groups.google.com/ Message-ID: <2716453f.0205160400.66eb3771@posting.google.com> Hi, Everybody Now, I'm developing 82559ER's Driver on our board (8240) which before is using the 21143's Driver. What I do is : 1.modify vendor id from 0x1229 to 0x1209 2.remmed instruction of reading eeprom, and write: pReso->eeprom[0] = 0x9000; pReso->eeprom[1] = 0x0127; pReso->eeprom[2] = 0x0302; 3.I changed the bootdevice from dc to fei in bootline 4.Change the PRIMARY_ENDTYPE from DEC_END_DEVICE to FEI_END_DEVICE But when all have been done, I find the board can download vxWorks from my pc and start vxWorks successfully, then when I send one packed which is used in wdb communication(UDP packet, dest port is 17185), and send another packet which will be accepted by target(dest mac is 0xffffffffff or 0x009027010203), the exception happened, output on serial is : program Exception current instruction address: 0x00000000 Machine Status Register: 0x0008b030 Condition Register: 0x48000040 Task: 0x3bbe810 "tNetTask" I checked the "LR" register with emulator, whenever it is happened, the value of this register is the instruction address which in muxReceive. I just want to know what I can do to find the reason, thanks. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: WDB Packet generate Exception Date: Thu, 16 May 2002 14:13:49 +0200 From: "Wiesel Andre" Organization: Bluewin AG Message-ID: <3ce3a278$1_3@news.bluewin.ch> References: <2716453f.0205160400.66eb3771@posting.google.com> > Exception current instruction address: 0x00000000 Most likely a non-iniatialized pointer to a function is the cause of your problem. "Gabriel" wrote in message news:2716453f.0205160400.66eb3771@posting.google.com... > Hi, Everybody > > Now, I'm developing 82559ER's Driver on our board (8240) which before is > using the 21143's Driver. What I do is : > > 1.modify vendor id from 0x1229 to 0x1209 > 2.remmed instruction of reading eeprom, and write: > pReso->eeprom[0] = 0x9000; > pReso->eeprom[1] = 0x0127; > pReso->eeprom[2] = 0x0302; > 3.I changed the bootdevice from dc to fei in bootline > 4.Change the PRIMARY_ENDTYPE from DEC_END_DEVICE to FEI_END_DEVICE > > But when all have been done, I find the board can download vxWorks from > my pc and start vxWorks successfully, then when I send one packed which is > used in wdb communication(UDP packet, dest port is 17185), and send another > packet which will be accepted by target(dest mac is 0xffffffffff or > 0x009027010203), the exception happened, output on serial is : > > program > Exception current instruction address: 0x00000000 > Machine Status Register: 0x0008b030 > Condition Register: 0x48000040 > Task: 0x3bbe810 "tNetTask" > > I checked the "LR" register with emulator, whenever it is happened, the > value of this register is the instruction address which in muxReceive. > > > I just want to know what I can do to find the reason, thanks. --------------------------- Newsgroups: comp.os.vxworks,comp.realtime,comp.arch.embedded Subject: embedded web server-loading images!! Date: 16 May 2002 05:32:37 -0700 From: hema_tm@rediffmail.com (hema) Organization: http://groups.google.com/ Message-ID: <56b5d4bb.0205160432.160d016@posting.google.com> hai grp, i m working on vxWorks 5.4 and target env: MPC860T.i planned to web enable my application running on the target.i use the vxHttp server which was posted in the group some time back(vxHttp.c).i compiled it to my application and was able to run it fine in a text enviorment(with CGI/JavaScript).but my requirement now is to load images .how do i load it dinamicaaly??the images cant be rom resident.is there any tool which will convert the .gif to hex/binary equlent or something of that sort!!! i hope to have some inputs in this regard as its an urgent requirement !!!! thanx in advance... Hema --------------------------- Newsgroups: comp.os.vxworks,comp.realtime,comp.arch.embedded Subject: Re: embedded web server-loading images!! Date: Thu, 16 May 2002 15:01:55 +0200 From: "Jan Homuth" Organization: Tasking GmbH Message-ID: References: <56b5d4bb.0205160432.160d016@posting.google.com> Hi Hema, try this: ====================== SNIP SNIP ======================= /* bin2c: * convert binary data to const char array * formatted for use in c source */ #include #include #include #include #define USE_OFILE 0x01 #define USE_IFILE 0x02 #define USE_STDIN 0x04 #define USE_STDOUT 0x08 #define MAXCOL 14 char ouufilename[256]; int main (int argc, char **argv) { FILE *fin, *fout; unsigned char c; unsigned char opt = 0; int col; long fsize=0; long bytes=0; int handle; col = 0; if (argc < 1){ printf("No argument given\nusage: bin2c \n"); exit(97); } if (argc > 1){ opt|= USE_IFILE; if(NULL == (fin = fopen(argv[1],"rb"))) { printf("Could not open file %s\n", argv[1]); exit(99); } } if (argc > 2) { opt |= USE_OFILE; if(NULL == (fout = fopen(argv[2],"w"))) { printf("Could not open file %s\n", argv[1]); exit(100); } } /* TODO: add -f option : write to .c file w. same name as infile */ strcpy(outfilename, _fnsplit() if( ! (opt & USE_IFILE)) fin = stdin; if( ! (opt & USE_OFILE)) fout= stdout; if ((handle = open(argv[1], O_RDONLY | O_BINARY)) == -1) { printf("Can't open file.\n "); exit(98); } fsize = filelength(handle); close(handle); fprintf(fout, "/* binary data converted from %s */\n",argv[1]); #ifdef USE_POINTER fprintf(fout, "\n\nconst unsigned char *bindata = \n{\t"); #else fprintf(fout, "\n\nunsigned char bindata[%ld] = \n{\t",fsize); #endif do { c = fgetc(fin); fprintf(fout, "0x%02x, ",c); if (col++ > MAXCOL) { fprintf(fout,"\n\t"); col = 0; } } while(++bytes < fsize-1); c=fgetc(fin); fprintf(fout, "0x%02x\n",c); fprintf(fout, "};\n/* EOF */\n"); fflush(fout); fclose(fout); fclose(fin); } ====================== SNIP SNIP ======================= - -- kind regards, Jan Homuth Altium Germany GmbH Technical Support Embedded Systems - ------------------------------------------------------------------ Phone : +49 7152/97991-0 Altium Germany GmbH Fax: +49 7152/97991-20 Eltinger Str. 61 Email: support.eu@altium.com 71229 Leonberg WWW: http://www.altium.com Germany Developer Forum: http://groups.yahoo.com/group/TASKINGforum Development Services : http://www.altium.com/products/services.htm - ------------------------------------------------------------------ - ----- Altium - Think it, Design it, Build it! ----- - ------------------------------------------------------------------ - - Protel * P-CAD * TASKING * PeakFPGA * CircuitMaker * CAMtastic!- - ------------------------------------------------------------------ "hema" schrieb im Newsbeitrag news:56b5d4bb.0205160432.160d016@posting.google.com... > hai grp, > > > i m working on vxWorks 5.4 and target env: MPC860T.i planned to web > enable my application running on the target.i use the vxHttp server > which was posted in the group some time back(vxHttp.c).i compiled it > to my application and was able to run it fine in a text > enviorment(with CGI/JavaScript).but my requirement now is to load > images .how do i load it dinamicaaly??the images cant be rom > resident.is there any tool which will convert the .gif to hex/binary > equlent or something of that sort!!! > i hope to have some inputs in this regard as its an urgent requirement > !!!! > > > > thanx in advance... > > > Hema --------------------------- Newsgroups: comp.os.vxworks,gnu.gcc.help Subject: Re: cross compiling gcc - parse error Date: 11 May 2002 10:48:28 -0700 From: demiurg@tau.ac.il (Alexander Sirotkin) Organization: http://groups.google.com/ Message-ID: References: <3CDA539D.F483A489@idl.bt.co.uk> <3CDC1481.A1C0711B@lmco.com> I did contact WRS - waiting for reply now. As for the compilation errors : I have a perfectly working Tornado 2 (on Windoze of course) and this is the correct line. Moreover, vxworks.h file it yells about is not part of the Tornado 2 - it's part of gcc sources... 10x anyway... Joe Durusau wrote in message news:<3CDC1481.A1C0711B@lmco.com>... > Alexander Sirotkin wrote: > > > > I'm trying to build both compiler and linker (binutils) > > and they obviously need to include vxworks.h... > > am I wrong ? > > > > Phil Ashby wrote in message news:<3CDA539D.F483A489@idl.bt.co.uk>... > > > Alexander Sirotkin wrote: > > > > > > > > I'm trying to compile gcc & bintutils for vxworks on Linux, > > > > binutils are compiled just fine, but gcc throuws the following error: > > > > > > > > /root/gnu/gcc-2.95/gcc/config/mips/vxworks.h:3: parse error before `file' > > > > > > > > Apparently, it does not like embedded comments > > > > > > > > vxworks.h:3 > > > > /* Copyright (C) 1999 Free Software Foundation, Inc. */ > > > > This file is part of GNU CC. > > > > > > That looks wrong. Obviously the text 'This file...' should be in a > > > coment block, but isn't. Why should a cross-compiler need to include > > > a BSP-specific vxworks.h anyway? > > > > > > Surely you just need to create a cross-compiler for the target CPU, > > > not the O/S? > > I think that you either have a bad copy of vxWorks.h, or maybe you have > picked out > the wrong line. The compiler doesn't always print correct line numbers > for > errors. BTW, on my copy of vxWorks.h, the text "this file.." does not > appear. > I see the line "/* Copy... " followed by a blank line, and a comment > block > with mod history. > > May I again suggest you contact WRS and get the source for their > modified > version of gcc? Even if it's not the revision you want, you would save > a huge amount of effort. > > Speaking only for myself, > > Joe Durusau --------------------------- Newsgroups: comp.os.vxworks,comp.realtime,comp.arch.embedded Subject: Re: embedded web server-loading images!! Date: Thu, 16 May 2002 16:21:29 +0200 From: Emmanuel Herbreteau Organization: SEPRO Robotique (BEP) Message-ID: <3CE3C069.A6A77B4B@sepro-robotique.com> References: <56b5d4bb.0205160432.160d016@posting.google.com> hema wrote: > i m working on vxWorks 5.4 and target env: MPC860T.i planned to web > enable my application running on the target.i use the vxHttp server Hi, I'm not sure to really understand your needs, but... If you want to include "raw bitmap" in your executable, you can follow this path : 1. create your picture in a standard format (BMP, GIF, etc...) 3. run the tornado's utility : "binToAsm image.bmp > image.s" 4. replace ".data" by ".text" in the new file "image.s" 5. include this file in your project You can now access your picture using a simple pointer in your code. Hope this help (?) Regards Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: Free Evaluation Report VxWorks AE 1.1 Date: 16 May 2002 07:47:16 -0700 From: b.van.beneden@dedicated-systems.com (Bart Van Benedne) Organization: http://groups.google.com/ Message-ID: Hi all, Just a quick message to let you all know that the RTOS evaluation report of VxWorks AE 1.1 (as well as many other RTOS) is available (free) for download at: http://www.dedicated-systems.com/encyc/buyersguide/rtos/rtosmenu.htm Cheers, Bart Van Beneden --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about Interrupt! Date: Thu, 16 May 2002 17:28:03 +0200 From: "Werner Schiendl" Message-ID: <3ce3cff5$1@brateggebdc5.br-automation.co.at> References: <989cdae4.0205160148.4400245b@posting.google.com> Hi, see below "zlybupt" wrote in message news:989cdae4.0205160148.4400245b@posting.google.com... > hi,all > Is there any one show can give me an explanation about the following objects: > 1.interrupt number This is normally the number of the interrupt as seen from the CPU. A CPU can have more interrupts as you see on the hardware (i. e. the hardware can trigger). For example, normal x86 system allow for 256 interrupts. Some of these interrupts are reserved for platform dependent means. x86 hardware reserves the first 32 (I believe...) interrupts for things like Page Fault, Divide Error, General Protection Fault, etc. > 2.interrupt vector The interrupt vector in my understanding is the area where the addresses of the interrupt routines alongside any platform specific information is stored. For x86 platform, a special form of interrupt descriptor is used. The layout of this is normally not important to you, as long as you use the provided functions from intLib. The size of one such entry for x86 is 8 bytes, i. e. the interrupt table is 256 * 8 bytes in size. This is also the reason for the shift left of 3 bits in the macro you mention. The value returned from INUM_TO_IVEC is actually the offset from the beginning of the > 3.interrupt level In my terminology, this is the 'interrupt line' where a piece of hardware is connected to. Of course, today's hardware is not always that simply, but that is basically the meaning. For example, a standard PC has 16 interrupt levels, IRQ 0 to 15. This are the values, PC users from good ol' DOS days are accustomed to. For PC hardware on VxWorks, interrupt levels map to interrupt number as follows (for pc486 and pcPentium BSP): interrupt number = interrupt level + INT_NUM_IRQ0 The value of INT_NUM_IRQ0 on the platforms mentioned is 0x20 (32) by default. Therefore: > Uninitialized interrupt > vector number 42(0-255) > I wonder how can an intnum be shifted to 42 by the macro INUM_TO_IVEC??? > Thank you for your answers in advance! This means, your network board is using interrupt level 10 (IRQ 10). For the calculation, see above. ( moved for clarity ) > and which one should be used for the function intConnect? > BTW:when I remove the cable from my ethernet card,system tells me: > interrupt: for intConnect, you need to specify the interrupt vector: .... intConnect(INUM_TO_IVEC(irq_line + INT_NUM_IRQ0), .... will give you the desired result. Caveat: sysIntEnablePIC expects the raw interrupt level (10 in your example) hth Werner --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Accessing hard disk Date: Thu, 16 May 2002 17:32:05 +0200 From: "Werner Schiendl" Message-ID: <3ce3d0e7@brateggebdc5.br-automation.co.at> References: <253a5f89.0205092129.150afb1d@posting.google.com> Hi, which kind of hardware, which version of VxWorks are you using? Do you have dosFs2 or the 'old' dosFs? AFAIK, the ata drivers are more up-to-date. regards Werner "Maverick" wrote in message news:253a5f89.0205092129.150afb1d@posting.google.com... > Hi, > > I am trying to create a Vxwork image that can support hard disk. I > have included IDE hard disk drive but when I tried to build I got an > error: undefined reference to ideTypes. What all changes in > configuration files do I need to do so that I can use hard disk on > pcpentium target? > > To use hard disk I had followed the following procedure > 1) Initialise IDE driver by using ideDrv() > 2) Create device for IDE disk ideDevCreate() > 3) Initialise the device with dos File system by using dosFsDevInit() > 4) Use hard disk. > > Is this the procedure to use hard disk correct? Any pointers will be > greately appreciated. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ioctl problem for VxWorks Date: Thu, 16 May 2002 18:29:46 +0200 From: "Werner Schiendl" Message-ID: <3ce3de6c$1@brateggebdc5.br-automation.co.at> References: <414bb87b.0205132030.7f452eb8@posting.google.com> Hi What do you want to do? Make sure that all data of the file has been written to the underlying file system? I think this is what FIOFLUSH is for. FIOSYNC is AFAIK only supported for the whole disk, not individual files. hth Werner "Rakesh" wrote in message news:414bb87b.0205132030.7f452eb8@posting.google.com... > Hi all > I am facing the problem of firing the ioctl command of FIOSYNC on dos > file system. > whwnever i open a file using the open command aand try to fire the > ioctl on the fd returned i receive an unknown request error. > can anybody suggest me the solution. i am using the PcPentium as my > target --------------------------- Newsgroups: comp.os.vxworks Subject: Re: "RPCExit: extra call" error message Date: Thu, 16 May 2002 12:48:25 -0500 From: "Peter Marqui" Organization: Verio Message-ID: <1hSE8.18307$JZ6.266302@dfw-read.news.verio.net> References: Install the Win2K patch "Tom Almo" wrote in message news:af523f7894412fdfafacc66e57b6034b.93258@mygate.mailgate.org... > hi all > after install torando 2.0.1 5 windows with this message jump out. > each time i open the tornado this messages come back. > did any body know how to slove the problem :-> > Tom. > > > -- > Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Polymorphism and Destructor Date: 12 May 2002 19:36:10 -0700 From: pboulay@hotmail.com (Patrick Boulay) Organization: http://groups.google.com/ Message-ID: <8bb5523.0205121836.5b8b1432@posting.google.com> Hello, I have a big problem with polymorphism and destructor. See the class in example bellow, the vxworks crash after I delete the pointer. class A { public: A(); virtual ~A(); }; class B:public A { public: B(); virtual ~B(); }; int aFunction() { A *ptr; ptr = new B; delete ptr; return 0; } Thanks Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Polymorphism and Destructor Date: Mon, 13 May 2002 13:04:21 +0900 From: "Sungchol Kim" Organization: Korea Telecom Hitel Message-ID: References: <8bb5523.0205121836.5b8b1432@posting.google.com> try A *ptr = new B; "Patrick Boulay" wrote in message news:8bb5523.0205121836.5b8b1432@posting.google.com... > Hello, > > I have a big problem with polymorphism and destructor. See the class > in example bellow, the vxworks crash after I delete the pointer. > > class A > { > public: > A(); > virtual ~A(); > }; > class B:public A > { > public: > B(); > virtual ~B(); > }; > int aFunction() > { > A *ptr; > > ptr = new B; > > delete ptr; > > return 0; > } > > > > Thanks > Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Difference between RTOS and normal OS Date: Mon, 13 May 2002 07:00:44 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <507efed5.0205012037.3e11728f@posting.google.com> <507efed5.0205022254.3db92d02@posting.google.com> "Leonid Rosenboim" wrote in message news:newscache$7lkzvg$uaa$1@lnews.actcom.co.il... > Let me show you a different point of view: > > The design goals of a RTOs are quite different from a Generap Purpose OS > "GPOS" like Unix, VMS and probably NT. > > In RTOS, it is assumed that the end-product is an embedded device, > including hardware RTOS and higher-level software - all a single > application, all tested together as a single unit. > Therefore most safeguards that exist in GPOS are not really necesary, and > are excluded in order to obtain the best possible performance on a given > hardware platform, i.e. enable cost savings of the end-product. > ALso, since all software is made and tested by a single entity, the > resources (e.g. CPU, memory) is allocated to subsystems (tasks, modules) as > per programmer request. If prioritiesa are set wrong, surely enough this > will be revealed during system testing. > > (rest snipped) I beg to strongly differ, but your assumption is way off. RTOS were developed long before "embedded systems" as a concept had been invented, long before the microprocessor was invented. I will bore you with some examples, all driven by what we old people refer to as "big iron" machines: 1. Factory floor automation 2. Particle physics experiments 3. Any kind of high speed, high bandwidth data sampling 3. Nuclear reactors 4. Radar and sonar systems 5. Telecommunications infrastructure (PBXs, Central Office switchers, etc.) 6. Aerospace ground control systems Digital Equipment Corporation developed VAX/VMS as a GPOS and an RTOS, and the system could be tweaked to do either, or both. Many simulators in the 80's were driven by VAX "minis" (I laugh at the term now), when microprocessors were struggling with 16-bit limitations and slow I/O busses. Furthermore, try telling the Air Force, Boeing, or AirBus that "most safeguards that exist in GPOS are not really necesary" in the RTOS in their modern aircraft; in fact, they insist on tighter safeguards on the issues most important to RTOS: predictability, reliability, redundancy, and fault-tolerance. The fact that some project might not need some or all of these is because the project is either trivial or it doesn't bother the developer if the system is unreliable. Shame on him/her if the latter is true. And don't assume that any problems "surely ... will be revealed during system testing." I seem to recall a minor priority inversion flaw in a certain Mars mission that wasn't discovered until the system failed several million miles away from the lab. Just a straightforward VxWorks application, was it not? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Does DOSFS 2.0 for PPC supports device whose block size != 512 ???? Date: Mon, 13 May 2002 07:04:56 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: I think the sector size needs to be a power of 2, i.e. 128, 256, 512, 1024 ... because some calculations use bit shift rather than multiply operations. It appeared very strange that someone would require a sector size of 600 bytes, and I am curious as to the reason for such a requirement. - Leonid "houcheng" wrote in message news:ba773a9e.0205122134.ce94661@posting.google.com... > Dear vxworkers: > > Here is a question about DOSFS 2.0 for PPC: > Is the DOSFS 2.0 support block device whose block size is > not 512 bytes per block ? > > The following is my compact flash's initizliation code: > /************************************************************************/ > if(cFlashDrv()== OK) { > printf("OK]\n"); > } else { > printf("FAIL]\n"); > printf("Error: Pcmcia card is not in slot!\n"); > return(ERROR); > } > pDev = cFlashDevCreate(); /* create a block device object */ > pCbio = cbioDevVerify(pDev, FALSE); > if(pCbio) > printf("\rInitialize CBIO device driver > ............................... [OK]\n"); > else { > printf("\rInitialize CBIO device driver > ............................... [FAIL]\n"); > return(ERROR); > } > > ret = dosFsDevCreate(filename, pCbio, 30, DOS_CHK_REPAIR | > DOS_CHK_VERB_SILENT ); > > if (ret == OK) > printf("\rInitialize DOS File System > .................................. [OK]\n"); > else { > printf("\rInitialize DOS File System > .................................. [FAIL]\n"); > return(ERROR); > } > return(OK); > > For the compact flash with 512-bytes block size, this > initialization > code runs well. But for the compact flash with 600 bytes block size, > the cbioDevVerify(pDev, FALSE) returns NULL! > > Is the DOSFS 2.0 supports block size other than 512 bytes? I've > tried > the DOSFS patch SPR73141, which fixed the problem "dosFs2 is not > handling boot > sectors correctly on devices with sector sizes != 512 bytes." . > But the results is still the same, no works on flash with 600 > bytes block size. > > > Any suggestion is welcome! > Thanks in advanced! > Lin Houcheng > houcheng@turbonetworks.com.tw_No_SPAM --------------------------- Newsgroups: comp.os.vxworks Subject: why doesn't kill(TID, 9) work? Date: Thu, 16 May 2002 14:20:11 -0700 From: "Mark Galecki" Message-ID: <3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net> Hello, I am trying to terminate a task (which is in an infinite loop) with kill and signal 9 (numeric value of SIGKILL - the termination signal which cannot be caught or ignored). The kill command sends the signal and exits successfully. However, the task persists. I _am_ able to kill it with taskDelete. Why not with kill signal 9?? trace info: - -> i NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY - ---------- ------------ -------- --- ---------- -------- -------- ------- -- - --- (...) tDbgTask _readSameFra 2b6d744 100 READY 4ba6b7 2b6d6e8 0 0 (....) value = 0 = 0x0 - -> kill(0x2b6d744, 9) value = 0 = 0x0 - -> i NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY - ---------- ------------ -------- --- ---------- -------- -------- ------- -- - --- (...) tDbgTask _readSameFra 2b6d744 100 READY 4ba6b7 2b6d6e8 0 0 (...) value = 0 = 0x0 - ------------------------ I am using PENTIUM target and Tornado 2.0/VxWorks 5.4 Thank you, Mark Galecki --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation for MBX860 target failed Date: Mon, 13 May 2002 07:06:12 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <3CDE189F.30303@bigfoot.com> Gee, couldn't you have just analyzed the assembly code differences yourself? "Alex Vinokur" included reams of assembly language code news:3CDE189F.30303@bigfoot.com... > > (mercifully snipped) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation log file Date: 13 May 2002 07:13:07 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "Alex Vinokur" wrote: >============ Execution: torvars.bat ============ > %torvars.bat > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>rem Command line build environments > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set WIND_HOST_TYPE=x86-win32 > Out of environment space > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set WIND_BASE=D:\sys2\Tornado > Out of environment space > > D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set > PATH=\host\\bin;C:\PROGRA~1\Far;C:\DJGPP\bin; > D:\Perl\bin\;C:\WINNT\system32;C:\WINNT;C:\WINNT\System32\Wbem;D:\app1\ > araxi You have to increase your environment space before you can do that. This depends on the OS you are using. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation log file Date: Mon, 13 May 2002 10:56:10 +0200 From: Alex Vinokur Message-ID: <3CDF7FAA.2030200@bigfoot.com> References: Johan Borkhuis wrote: > "Alex Vinokur" wrote: > > >>============ Execution: torvars.bat ============ >>%torvars.bat >> >>D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>rem Command line build environments >> >>D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set WIND_HOST_TYPE=x86-win32 >>Out of environment space >> >>D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set WIND_BASE=D:\sys2\Tornado >>Out of environment space >> >>D:\SYS2\TORNADO\HOST\X86-WI~1\BIN>set >>PATH=\host\\bin;C:\PROGRA~1\Far;C:\DJGPP\bin; >>D:\Perl\bin\;C:\WINNT\system32;C:\WINNT;C:\WINNT\System32\Wbem;D:\app1\ >>araxi >> > > You have to increase your environment space before you can do that. This > depends on the OS you are using. > > Groeten, > Johan > > Thanks. It seems that now it is OK. ============ Execution: torvars.bat ============ %torvars.bat D:\sys2\Tornado\host\x86-win32\bin>rem Command line build environments D:\sys2\Tornado\host\x86-win32\bin>set WIND_HOST_TYPE=x86-win32 D:\sys2\Tornado\host\x86-win32\bin>set WIND_BASE=D:\SYS2\TORNADO D:\sys2\Tornado\host\x86-win32\bin>set PATH=D:\SYS2\TORNADO\host\x86-win32\bin;C:\Program Files\Far;C:\DJGPP\bin; D:\Perl\bin\;C:\WINNT\system32;C:\WINNT;C:\WINNT\System32\Wbem;D:\app1\araxis;D:\SPE;D:\unix1\MKS\MKS-5.2\MKSNT ================================================= How can 'Compilation log file' be created after that (this is my original queestion) ? - -- ============================ Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ============================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Polymorphism and Destructor Date: Mon, 13 May 2002 11:23:22 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: References: <8bb5523.0205121836.5b8b1432@posting.google.com> "Patrick Boulay" wrote in message news:8bb5523.0205121836.5b8b1432@posting.google.com... | Hello, | | I have a big problem with polymorphism and destructor. See the class | in example bellow, the vxworks crash after I delete the pointer. | You should implement constructors and destructors. | class A | { | public: | A(); | virtual ~A(); | }; | class B:public A | { | public: | B(); | virtual ~B(); | }; | int aFunction() | { | A *ptr; | | ptr = new B; | | delete ptr; | | return 0; | } | | | | Thanks | Patrick - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Q: boost::shared_ptr and thread-safety Date: Mon, 13 May 2002 14:37:36 +0200 From: Thomas =?iso-8859-1?Q?K=F6hler?= Organization: Tiscali - A european company Message-ID: <3CDFB390.F880DADF@rovema.de> Hi, Im trying to port a multithreaded Win32 console application which uses the boost::shared_ptr class (www.boost.org) to run under VxWorks (Tornado 2.0.2, x86, gcc 2.7.2). Shared_ptr claims to be thread-safe for Win32, Linux and "any platform that supports pThread". But I explicitly DON'T want to use the pThread library for VxWorks because of performance reasons. My question now is: How to make boost::shared_ptr thread-safe ? MTIA - -- Thomas Köhler ROVEMA * Tel +49 (0) 641 409-221 Verpackungsmaschinen GmbH * Fax +49 (0) 641 409-212 Industriestraße 1 * mailto:thomas.koehler@rovema.de Fernwald * http://www.rovema.de --------------------------- Newsgroups: comp.os.vxworks Subject: RTM_GET does not fetch Gateway of given destination Date: 13 May 2002 06:58:08 -0700 From: overtones@excite.com (Strawberry) Organization: http://groups.google.com/ Message-ID: <41440750.0205130558.3a4eede0@posting.google.com> hi groups, I need to get the route info given a destination and mask. I use routing socket. But the RTM_GET message gives me AF_INET -> for destination & mask from which i get its value AF_LINK -> for gateway. Since AF_LINK sock has only interface index, how do i know which of the adresses on the interface is my gateway??.. I am using Vxworks prototyper on Win2k. regards, StrawBerry (my fav..) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: RAM_LOW_ADRS parameter in T2 Date: Mon, 13 May 2002 16:04:27 +0200 From: Emmanuel Herbreteau Organization: SEPRO Robotique (BEP) Message-ID: <3CDFC7EB.4689F57E@sepro-robotique.com> References: <3CDFBEF2.D70397B@europem01.nt.com> Sebastien Fauris wrote: > I try to change the value of the RAM_LOW_ADRS parameter Hi, In the project windows (Tornado 2), select the "build" folder, then select your project (probably called "myProject->default"). Edit the properties of the "default" build. RAM_LOW_ADRS is in the "macro" tab. It's the value used in the generated makefile. Regards Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Polymorphism and Destructor Date: 13 May 2002 10:13:48 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: <8bb5523.0205121836.5b8b1432@posting.google.com> Sender: bpringlemeir@DeadDuck > "Patrick Boulay" wrote: > I have a big problem with polymorphism and destructor. See the > class in example bellow, the vxworks crash after I delete the > pointer. Alex> You should implement constructors and destructors. Ah! I have seen this problem too. It depends on the version of the Tornado compiler in use. The simulator seems to generate default destructors for you. However, the ARM compiler for T2.0 does not. I believe that T2.0.1 has the same problem. I do think it is bad practice to rely on the compiler to generate a destructor for you. "http://www.possibility.com/Cpp/CppCodingStandard.html#req" "http://www.google.ca/search?q=cache:avwPnhDDzbQC:www.possibility.com/Cpp/CppCodingStandard.html#req" Regards, Bill Pringlemeir. - -- Don't you realise that most gay people have source licenses ? Furthermore, you fascist, the Sun is always right. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Dying in kernelInit Date: Fri, 17 May 2002 03:11:33 GMT From: emdezpk@yahoo.ca (Duke Skylurker) Organization: MCOM Message-ID: Hi, I am porting a BSP (actually taking a very similar one and tweaking it so it will work on my system). The problem: system dies somewhere in kernelInit call. I do not have the luxury of Logic Analyzer, so I send some characters to the console (in polling mode) from different functions to see the progress of the program. I see that it passes sysStart, sysHwInit, ... ... taskLibInit, workQInit (in usrKernelInit). But then it calls kernelInit, and never comes out in usrRoot !!! Where exactly is usrRoot called? How system switches to the new task? I have disabled cache, timer... nothing helps. The CPU is PowerPC403. Thanks, Duke S. --------------------------- Newsgroups: comp.os.vxworks,comp.realtime,comp.arch.embedded Subject: Re: embedded web server-loading images!! Date: 16 May 2002 21:19:40 -0700 From: hema_tm@rediffmail.com (hema) Organization: http://groups.google.com/ Message-ID: <56b5d4bb.0205162019.55a0d5fd@posting.google.com> References: <56b5d4bb.0205160432.160d016@posting.google.com> <3CE3C069.A6A77B4B@sepro-robotique.com> hai thanx for your response. but i was looking out to convert a .gif to binary or hex,and not a binary data to ascii.well to be more specific about my doubt.i want to generate a GIF say xyz.gif.i need the hex equivalent of this GIF in specific to MPC860 byte order format. hope i m clear!!!! thanx , hema Emmanuel Herbreteau wrote in message news:<3CE3C069.A6A77B4B@sepro-robotique.com>... > hema wrote: > > i m working on vxWorks 5.4 and target env: MPC860T.i planned to web > > enable my application running on the target.i use the vxHttp server > > Hi, > > I'm not sure to really understand your needs, but... > > If you want to include "raw bitmap" in your executable, > you can follow this path : > > 1. create your picture in a standard format (BMP, GIF, etc...) > 3. run the tornado's utility : "binToAsm image.bmp > image.s" > 4. replace ".data" by ".text" in the new file "image.s" > 5. include this file in your project > > You can now access your picture using a simple pointer in your code. > > Hope this help (?) > > Regards > Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: C and C++ object files Date: Fri, 17 May 2002 06:45:19 GMT From: "Kim, Jeong-Hwan" Organization: Dacom Message-ID: <3G1F8.394$Bo1.488883@news.bora.net> When I wish to compile code with C++, how can i modify tornado build option? Now I use "ccppc" for compile tool name for C. Is it ok to modify only compile tool name to "c++ppc" another question : Is it possible to link c and c++ object files together? Thanks in advace Kim --------------------------- Newsgroups: comp.os.vxworks Subject: Send Multicast message Date: 16 May 2002 23:57:37 -0700 From: supergod2001@bigfoot.com (supergod) Organization: http://groups.google.com/ Message-ID: <58cafc21.0205162257.7ae894fa@posting.google.com> Now my project need send multicast message to other target. But i must use setsockopt (sock, IPPROTO_IP, IP_MULTICAST_IF, (char *)&ifAddr, sizeof (mCastAddr)) to select a Default Interface for Outgoing Multicasts before i send it. Otherwise sendto () will return error 0x41. I want to know whethere any api can used to select the interface for outgoing multicase before i create a socket. if so i need only set sock option once, and everytime i create a new socket ,it will automaticlly get the interface to send out . any idea? my target is based on pentinum, network card is Intel(R) 82559 . --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dying in kernelInit Date: Fri, 17 May 2002 07:17:15 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Briefly, kernelInit is the function which switches from single task to multi task mode, and does this by spawning the initial Root Task. To set up the task, the kernel would create the task control block, initial register image, stack etc, and will enable interrupts right before it makes a "return" into the newly created task. Most probably your issue is sue to an I/O device having its interrupt request pending at the time, using level-trigerred interrupts, and when the kernelInit enables interrupts the CPU goes into infinite interrupt. ANother possible issue might be due to kernel code which is written in machine assembly, and in order to optimize context switch time, would use the most effecient instructions to load registers, many at once from memory. This creates paged or pipelined memory accesses, and your hardware may have a bug when CPU issues advanced memory cycles, freezing or yielding wrong values. I have encountered both of the above issues when doing BSPs for various processors in the past. Hopt this helps, Leonid "emdezpk@yahoo.ca (Duke Skylurker)" wrote in message news:Xns9210EC995DD45emdezpkyahooca@216.254.136.26... > Hi, > > I am porting a BSP (actually taking a very similar > one and tweaking it so it will work on my system). > > The problem: system dies somewhere in kernelInit call. > > I do not have the luxury of Logic Analyzer, so I send > some characters to the console (in polling mode) > from different functions to see the progress > of the program. > > I see that it passes sysStart, sysHwInit, ... > ... taskLibInit, workQInit (in usrKernelInit). > > But then it calls kernelInit, > and never comes out in usrRoot !!! > > Where exactly is usrRoot called? How system switches > to the new task? > > I have disabled cache, timer... nothing helps. > > The CPU is PowerPC403. > > > Thanks, > Duke S. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why doesn't kill(TID, 9) work? Date: 17 May 2002 00:53:45 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205162353.47142641@posting.google.com> References: <3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net> Hello, Simple answer: because VxWorks is not Unix. More detailed answer: signals are not used by the operating system for anything really (except some of the POSIX RT layer, e.g. posix timers). If you want to handle signals, for anything, you'll need to register a signal handler for each task. If you just want to delete a task, you found the API for VxWorks: taskDelete(). Be warned though, there are times when that might not work too (for example, if your task holds a delete-safe mutex semaphore while in its infinite loop). HTH, John... "Mark Galecki" wrote in message news:<3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net>... > Hello, > I am trying to terminate a task (which is in an infinite loop) with kill and > signal 9 (numeric value of SIGKILL - the termination signal which cannot be > caught or ignored). The kill command sends the signal and exits > successfully. However, the task persists. I _am_ able to kill it with > taskDelete. Why not with kill signal 9?? > > trace info: > -> i > NAME ENTRY TID PRI STATUS PC SP ERRNO > DELAY > ---------- ------------ -------- --- ---------- -------- -------- ------- -- > --- > (...) > tDbgTask _readSameFra 2b6d744 100 READY 4ba6b7 2b6d6e8 0 > 0 > (....) > value = 0 = 0x0 > -> kill(0x2b6d744, 9) > value = 0 = 0x0 > -> i > NAME ENTRY TID PRI STATUS PC SP ERRNO > DELAY > ---------- ------------ -------- --- ---------- -------- -------- ------- -- > --- > (...) > tDbgTask _readSameFra 2b6d744 100 READY 4ba6b7 2b6d6e8 0 > 0 > (...) > value = 0 = 0x0 > > ------------------------ > I am using PENTIUM target and Tornado 2.0/VxWorks 5.4 > > Thank you, > Mark Galecki --------------------------- Newsgroups: comp.os.vxworks Subject: Re: incomplete bootrom tftp download Date: Fri, 17 May 2002 11:29:44 +0200 From: "Werner Schiendl" Message-ID: <3ce4cd7b@brateggebdc5.br-automation.co.at> References: <3CE2AF8C.9020702@spamm.me.l8s.co.uk> <5syE8.102584$YQ1.34662592@typhoon.southeast.rr.com> Hi, since you can download the VxWorks image and the bootrom remains functional, I doubt the error is located in the bootroms. I think something with the very early configuration of the VxWorks image goes wrong. I recommend to check that any hardware is configured properly. Especially if you use the Tornado Project Facility, bootrom and VxWorks image use different sources of information: bootrom uses macros from config.h, VxWorks image uses configuration from project (.wpj file). It may be a start to build a VxWorks image directly in your BSP directory from the command line (like you build the bootrom, but specify VxWorks - or whatever incarnation is appropriate for your target - as the build target). If this image comes up, you know that the boot mechanism is working and something with your Tornado Project's configuration must be wrong. Otherwise, you know that it's probably not a project configuration issue. This should at least cut the possibilities of error to half. hth Werner "Mike Hanna" wrote in message news:5syE8.102584$YQ1.34662592@typhoon.southeast.rr.com... > I can successfully execute the "l" command from the bootrom which loads the > bootable image into memory. Afterwards, the bootrom is still functional & I > can display memory, etc. I've tried moving RAM_HIGH_ADRS around, but it > doesn't make any difference. It always hangs after it tries to start the > downloaded image. Initially, I thought it was because I wasn't getting the > full image. I've switched to ftp, verified that the full image is xferred > and the image still doesn't come up. > > Thanks, > > Mike > > > "David Laight" wrote in message > news:3CE2AF8C.9020702@spamm.me.l8s.co.uk... > > > > > Boot sequence flow through ROM and RAM > > > -------------------------------------- > > > > > > ROM RAM > > > _______________ --------------- > > > | romInit() | | |LOCAL_MEM_LOCAL_ADRS > > > |---------------| | | > > > | romStart() | |---------------|RAM_LOW_ADRS > > > |---------------| | V |-sysInit()-used at reboot > > > | boot |-- | x | > > > | monitor | | | W | > > > | | | | O | > > > |---------------| | | R | > > > | | | | K | > > > | | | | S | > > > | | | |---------------|FREE_RAM_ADRS > > > | | | | | > > > | | | | | > > > | | | |---------------|RAM_HIGH_ADRS > > > | | |---->| boot |-usrInit() > > > | | | monitior| > > > | | |(decompress |-kernenInit() > > > | | | if | > > > | | | necessary) | /tyCo0 > > > | | | |-usrRoot() ---- > > > | | | |-------------->| | > > > --------------- --------------- serial | | > > > ---- > > > > > > A nice picture with 2 minor faults! > > > > 1) memory maps are usually drawn with 0 at the bottom.... > > > > 2) if the boot code is compressed there is an additional stage, > > the decompress code is copied to RAM_LOW_ADRS before generating > > the boot monitor at RAM_HIGH_ADRS > > (this leaves the possibility of the decompresser overwriting its > > own data area! Or more likely there not being enough space between > > RAM_HIGH_ADRS and the end of physical memory when booting a > > compressed (not bootstrap) image to RAM_LOW_ADRS) > > > > David > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to do a MGET on a vxworks FTPD ? Date: Fri, 17 May 2002 11:32:03 +0200 From: "Werner Schiendl" Message-ID: <3ce4ce04$1@brateggebdc5.br-automation.co.at> References: Hi, Yes, use a more powerful FTP program for it. E. g. Windows Commander from www.ghisler.com will do the trick. (Shareware, free eval download) hth Werner "Mohasin" wrote in message news:d415d24.0205150834.1aeab285@posting.google.com... > Hi, > > I want to connect to the vxWorks FTP server running on a target board > and do multiple get of files using "mget *.ext" ? > > I am able to do a "get", "put" and "mput". But it looks like the > vxWorks FTP server doesn't support "mget". > > Is there any way I can acheive this?. > > > Thannkz > regards, > MOhasin --------------------------- Newsgroups: comp.os.vxworks,comp.realtime,comp.arch.embedded Subject: Re: embedded web server-loading images!! Date: Fri, 17 May 2002 11:38:48 +0200 From: "Werner Schiendl" Message-ID: <3ce4cf99$1@brateggebdc5.br-automation.co.at> References: <56b5d4bb.0205160432.160d016@posting.google.com> <3CE3C069.A6A77B4B@sepro-robotique.com> <56b5d4bb.0205162019.55a0d5fd@posting.google.com> Hi Hema, I think it would be more helpful to explain _what_ you are trying to accomplish instead of how you thought it might be possible. Anyway, from what I _assume_ you could need, I suggest to lookup the tool memdrvbuild in the documentation (host tool). This tool allows you to convert files (and maybe directories) in a format that can be compiled into the target image (vxWorks or downloadable image). On the target system, you can insert the files read-only into the filesystem and access the files as usual. hth Werner "hema" wrote in message news:56b5d4bb.0205162019.55a0d5fd@posting.google.com... > hai > thanx for your response. > but i was looking out to convert a .gif to binary or hex,and not a > binary data to ascii.well to be more specific about my doubt.i want to > generate a GIF say xyz.gif.i need the hex equivalent of this GIF in > specific to MPC860 byte order format. > hope i m clear!!!! > > > thanx , > > > hema > > > Emmanuel Herbreteau wrote in message news:<3CE3C069.A6A77B4B@sepro-robotique.com>... > > hema wrote: > > > i m working on vxWorks 5.4 and target env: MPC860T.i planned to web > > > enable my application running on the target.i use the vxHttp server > > > > Hi, > > > > I'm not sure to really understand your needs, but... > > > > If you want to include "raw bitmap" in your executable, > > you can follow this path : > > > > 1. create your picture in a standard format (BMP, GIF, etc...) > > 3. run the tornado's utility : "binToAsm image.bmp > image.s" > > 4. replace ".data" by ".text" in the new file "image.s" > > 5. include this file in your project > > > > You can now access your picture using a simple pointer in your code. > > > > Hope this help (?) > > > > Regards > > Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: C and C++ object files Date: Fri, 17 May 2002 11:51:08 +0200 From: "Werner Schiendl" Message-ID: <3ce4d27d@brateggebdc5.br-automation.co.at> References: <3G1F8.394$Bo1.488883@news.bora.net> Hi, what are you trying? If you are using the Tornade 2 IDE, you can just add your C++ files to the project and everything should work as expected. Give your C++ files a .cpp extension and your C files a .c extension. Don't use uppercase letters in the file extensions (.c and .cpp) since this causes confusion (also on Windows, although you might not see the immediately. If you have troubles on Windows, use the dir command in a command prompt to verify all files have lowercase extensions). If it is something else you want to know, specify it more clearly. hth Werner "Kim, Jeong-Hwan" wrote in message news:3G1F8.394$Bo1.488883@news.bora.net... > When I wish to compile code with C++, > how can i modify tornado build option? > Now I use "ccppc" for compile tool name for C. > > Is it ok to modify only compile tool name to "c++ppc" > > > another question : > Is it possible to link c and c++ object files together? > > Thanks in advace > Kim > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Newbie questions Date: Fri, 17 May 2002 11:41:41 +0100 From: "Liam Friel" Organization: Silicon & Software Systems Message-ID: Hi, I have checked the WRS website but am none the wiser really. I also have a call in to the local account manager, but I'd really like the "real story". Can someone quickly tell me: - - are the tools (cross compilers/Tornado etc) for VxWorks 5.4 for ARM available for a Linux host? - - what is the difference between VxWorks 5.4 and AE? - - are libraries compiled on a Solaris host for a given target linkable with libraries compiled on a Windows host for the same target, and vice versa? Thanks, Liam --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri May 17 14:40:07 2002 From: "Joe Welfeld" Date: Fri May 17 14:40:09 PDT 2002 Subject: Diab STL Compile Error Hello All, I get a syntax error when I compile the Standard Template Library (STL) that was included with Diab in diab4.4b\include\tornado\stl\tree.h PPC processor, windows environment. I am moving this code from GNU to Diab and this code compiles fine under GNU using GNU STL ? any Ideas ? Thanks, Joe _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ _/_/ Joe Welfeld Software Manager IP Services Paragon Networks International Http://www.paragon-networks.com 61 Commerce Dr. Brookfield Ct. 06804 203-740-3650 Ext:602 _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ _/_/ From vxwexplo-errs@csg.lbl.gov Fri May 17 15:42:28 2002 From: Shiv Date: Fri May 17 15:42:31 PDT 2002 Subject: SDRAM bursting problems with 8xxT FEC question summary -> Has anyone had a rough time getting the FEC to work on board that uses SDRAM ? This is a follow up to a previous question about getting the Fast Ethernet Controller running on the 855T. The vxworks kernel downloads and runs normally..and my PHY initializes correctly , but when I tried pinging the host from my board ,I realized with a Packet analyzer that the packet-dump shows the first 4 bytes of the MAC address repeated over and over again......This helped explain why when pinging from host to the target, I get a continuous ARP request attempt from the host while the target hangs immediately. My BSP uses the WRS Pro8xxT board files as a reference , the only difference being that our target board uses SDRAM and WRS' s Pro8xxT board supports DRAM....so I'm pretty sure its related to improper burst-configurations on my SDRAM , but I've tried all possible settings to the mode register that programs my SDRAM, but I still can't get a clean packet. Does anyone have any expertise that they might like to share , relating to this problem ? Would really appreciate it ! Thanks ! Shiv From vxwexplo-errs@csg.lbl.gov Sat May 18 04:03:09 2002 From: Vxworks Exploder Date: Sat May 18 04:03:12 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat May 18 04:03:06 PDT 2002 Subject: Re: How to do a MGET on a vxworks FTPD ? Subject: timestamping Subject: Re: timestamping Subject: Re: why doesn't kill(TID, 9) work? Subject: Re: ENVOY SNMP Subject: Re: why doesn't kill(TID, 9) work? Subject: Re: Raw file system reg Subject: Question about network card! Subject: E28F640J3 (flash source want) ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to do a MGET on a vxworks FTPD ? Date: 17 May 2002 11:44:15 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: mohasin.zaki@wipro.com (Mohasin) wrote: > Hi, > > I want to connect to the vxWorks FTP server running on a target board > and do multiple get of files using "mget *.ext" ? > > I am able to do a "get", "put" and "mput". But it looks like the > vxWorks FTP server doesn't support "mget". > > Is there any way I can acheive this?. There is an alternative FTP server distributed with DosFS2, and with T2.1. If you have these look in D:\T21-Mips\target-org\unsupported\src \netwrs, otherwise ask for an upgrade to 2.1 or ask for DosFS2. They are both free if you have a support contract. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: timestamping Date: 17 May 2002 06:21:43 -0700 From: nassim15@aol.com (Nassim Benzerrouk) Organization: http://groups.google.com/ Message-ID: <5cbe966c.0205170521.15b86e2c@posting.google.com> hi, i've choosen a system-defined timestamping to get a scale in seconds in stead of in number of events, and i've this following error message when i build : ... ... .. partialImage.o: In function `usrRoot': C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampFreq' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampPeriod' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampConnect' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampDisable' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampEnable' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampLock' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestamp' make: *** [vxWorks] Error 0x1 so someone can help me and tell me if a new cponfiguration is needed. ___________________________________________ | Nassim Benzerrouk | | OneAccess | /) +33 1 41 87 70 06 (\ / ) Email:nassim.benzerrouk@oneaccess-net.com ( \ _( (|___________________________________________) ) /> (((\ \) /,) / ) / //))/ (\\\\ \_/ / \ \_/ ///// \ / \ / \ _/ \_ / --/ /----------------------------------------\ \---- / / \ \ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: timestamping Date: 17 May 2002 14:02:51 GMT From: nospam@kpsws.com Message-ID: <3ce50d8a$0$224$4d4ebb8e@read-nat.news.nl.uu.net> References: <5cbe966c.0205170521.15b86e2c@posting.google.com> These Timestamp function should be in your timer driver file (see template for timer driver /target/src/drv/timer/templateTimer.c). probably your driver doesn't have support for timestamping in this way.... gr Niels hi, i've choosen a system-defined timestamping to get a scale in seconds in stead of in number of events, and i've this following error message when i build : ... ... .. partialImage.o: In function `usrRoot': C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampFreq' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampPeriod' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampConnect' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampDisable' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampEnable' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestampLock' C:\tornado2\vxsim\target\proj\oldstuff\Project1\prjConfig.c:444: undefined reference to `s ysTimestamp' make: *** [vxWorks] Error 0x1 so someone can help me and tell me if a new cponfiguration is needed. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why doesn't kill(TID, 9) work? Date: Fri, 17 May 2002 13:51:42 -0700 From: "Mark Galecki" Message-ID: <3ce56d7b$0$12564$4c41069e@reader0.ash.ops.us.uu.net> References: <3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205162353.47142641@posting.google.com> "John" wrote in message news:488e459a.0205162353.47142641@posting.google.com... > Hello, > > Simple answer: because VxWorks is not Unix. Thank you. It does help. However, I still don't understand the documentation. VxWorks Programmer's manual says right at the beginning that "nearly all POSIX interfaces (...) are available, including (...) queued signals". And, according to POSIX manual, SIGTERM is a termination signal that cannot be ignored or caught. And you are telling me that signals are ignored unless they are caught by a custom handler. So, is the VxWorks documentation lying?? Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Re: ENVOY SNMP Date: 13 May 2002 08:50:56 -0700 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: References: <3CD35445.8123C2F0@worldnet.att.net> Hello Chris: We am using ENVOY 9.2 and experiencing the same problem, I don't have the solution for it yet but have some discovery. I tried some commands provided by HP Openview in DOS prompt. following are the snmpwalk command I run with different options and it shows the reason why slow: D:\>snmpwalk -v 1 -c public 10.3.2.95 system D:\>snmpwalk -d -v 1 -c public 10.3.2.95 system The first one goes very fast and the second one with -d option (means dump ASN.1 packet trace) goes really really slow. I think the reason why there is a big delay between agent and manager is because enable the "dump", and worth to try to disable it . If you have the solution, pls let me know. Thanks chrismooremobile@yahoo.com (Chris Moore) wrote in message news:... > Les Cargill wrote in message news:<3CD35445.8123C2F0@worldnet.att.net>... > > Jason wrote: > > > > > > Hello: > > > > > > I have successfully built the ENVOY SNMP module for PPC, but it quite > > > big (1.5MBytes). I hope to reduce it. Does any one have the idea ? > > > > > > One way I want to try is, among the 11 groups of MIB II, only include > > > a few of them, say the system and ip, exclude the rest 9 groups. But > > > how to do it exactly? > > > > > > > Use the "-exclude" option when invoking the MIB compiler. > > > > > Thanks in advance! > > Hi Everyone, > > We use ENVOY 9.1 SNMP and find that is is REALLLLY slow when working > with HP Open view and ...AdventNet SNMP mangers. Using "snmptool" from > the win32 CLI its fine and fast.... does anyone know if this is a > setting problem on either the client or the manager? > > Cheers, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why doesn't kill(TID, 9) work? Date: 17 May 2002 19:13:05 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205171813.371c5cfa@posting.google.com> References: <3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205162353.47142641@posting.google.com> <3ce56d7b$0$12564$4c41069e@reader0.ash.ops.us.uu.net> Hello, > > Simple answer: because VxWorks is not Unix. > Thank you. It does help. > > However, I still don't understand the documentation. VxWorks Programmer's > manual says right at the beginning that "nearly all POSIX interfaces (...) > are available, including (...) queued signals". And, according to POSIX > manual, SIGTERM is a termination signal that cannot be ignored or caught. > And you are telling me that signals are ignored unless they are caught by a > custom handler. Correct, but SIGTERM is not an interface really. The VxWorks manual is telling you that you have both types of signal API (the old unix one, and the newer POSIX one) for registering handlers and sending signals. SIGTERM is part of process control, not the signals interface per-se. > So, is the VxWorks documentation lying?? That is probably too strong a term for it. The newest manual I have (VxWorks AE 1.1 Programmer's Guide) has this note added though to clarify it a little: "The VxWorks implementation of sigLib does not impose any special restrictions on operations on SIGKILL, SIGCONT, and SIGSTOP signals such as those imposed by UNIX. For example, the UNIX implementation of signal() cannot be called on SIGKILL or SIGSTOP." The signal support in VxWorks is provided to allow portability of code using signals for error detection (e.g. SIGSEGV) or for timers (e.g. SIGALRM). Built-in SIGTERM handling is really something that belongs in the process control area, and since VxWorks doesn't have processes (consider tasks to be the equivalent of a thread in the POSIX world), supporting it is not really logical. If you want to emulate SIGTERM at the thread level, you may do so. VxWorks will not stop you registering a handler for SIGTERM. Simply do this as part of the task's init code, and it should work just fine. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Raw file system reg Date: Mon, 13 May 2002 17:28:52 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Regarding Raw file system, as its name implies, it provides access to the raw disk underlying the file system, i.e. the raw file system is seen as one single file spanning the entire device. That's why you can not create files, this is not supported by rawFs. Regarding the unit of I/O of dosFs, you are (presumably) using old dosFs 1.0, which indeed makes I/O in units equal to cluster size. You should try to reformat the volume with a different cluster size. You can also consider switching to dosFs 2.0. However, my experience with Flash disks shows that changing the I/O quanta will not improve the overall write performance. Hope this helps, - Leonid Rosenboim "Asmita Joshi" wrote in message news:c8aeee4a.0205130729.5365e3ea@posting.google.com... > Hi Friends, > I want to create two ram disk devices one with dosFs & another > partition with rawFile system. My problem is that , when I use the > rawFsDevInit, it returns a non-null value. However if I try to create > any files in that raw fs partition, it returns an error. The following > is the log of the stuff which I tried from the shell. > > WDB: Ready. > -> ramDevCreate(0,1024,1024,1024,0) > value = 98555804 = 0x5dfd79c > -> rawFsDevInit("/rawWip",0x5dfd79c) > value = 98555744 = 0x5dfd760 > -> devs > drv name > 0 /null > 1 /tyCo/0 > 1 /tyCo/1 > 7 : > 8 /pty/telnet.S > 9 /pty/telnet.M > 10 /vio > 3 /wip > 4 /rawWip > value = 25 = 0x19 > -> cd "/rawWip" > value = 0 = 0x0 > -> ls > Can't open ".". > value = -1 = 0xffffffff = _func_taskRegsShowRtn + 0xfbb030eb > -> creat("/rawWip/Hello.txt",2) > value = -1 = 0xffffffff = _func_taskRegsShowRtn + 0xfbb030eb > -> > > Could anyone plz mail the reason for the Raw fs failure. > > My nect problem is > I am working with a flash driver which implements the dosFs on > flash. I find that the read & write handlers receives only 2 blocks of > data at a time. I have declared the block size as 512 bytes. In one > shot I receive only 1024 bytes of data to be written to the flash. > This happens independent of the file size which has to be written to > flash. This drastically slows down the system. I would like to know if > I can dictate the vxWorks dosFs to transfer the file to driver with > chunk size much higher than this 1024bytes. > > Plz mail me regarding the dosFs transfer block size. > > Thanks > Asmitha --------------------------- Newsgroups: comp.os.vxworks Subject: Question about network card! Date: 17 May 2002 23:12:00 -0700 From: zhoul@blrcsv.china.bell-labs.com (zlybupt) Organization: http://groups.google.com/ Message-ID: <989cdae4.0205172212.5bd0a755@posting.google.com> hi all: I am developing a network card driver for vxworks,but now met another problem,I wish someone can give me an explanation. In my driver design, now I have successfully attached the interrupt handle to my network card interrupt because when I remove the cable from the network card,it can succeffully detect this interrupt,but I wonder why it cannot detect the rxComplete interrupt when I send a packet to it?? why? where maybe the problem exist?(Here I am sure that the packet sent by sniffer have been successfully sent.) Thank you in advance for your answers! --------------------------- Newsgroups: comp.os.vxworks Subject: E28F640J3 (flash source want) Date: 14 May 2002 17:26:11 -0700 From: essenlee@hanmail.net (LEE MYUNG SUK) Organization: http://groups.google.com/ Message-ID: where I can get a driver source for a 28F640 Intel flash device? please give me advice, then thank you very much.. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sat May 18 18:20:44 2002 From: "dolfers joseph" Date: Sat May 18 18:20:47 PDT 2002 Hello , I was trying to install vxworks drivers for an IO interface card on a cpci rack with p3 processor.When i called the driverInit() function in the sysHwInit() routine in sysLib.c file I was not able to download the vxworks image on to the target it was getting stuck at the message on the target Attaching network interface lo0... done. Loading... 374624 + 57008 + 20036 Starting at 0x1000... But when I call the function in the application itself than the vxworks image gets downloaded but page fault occurs with an error code of 0x00000002 ,while executing any driver function which call the ioctl() function to read or write data to the device. can anyone out there knows what the problem is . please reply to josephs@rediffmail.com _________________________________________________________ Click below to visit monsterindia.com and review jobs in India or Abroad http://monsterindia.rediff.com/jobs From vxwexplo-errs@csg.lbl.gov Sun May 19 04:03:12 2002 From: Vxworks Exploder Date: Sun May 19 04:03:14 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun May 19 04:03:08 PDT 2002 Subject: Re: Lightweight heap monitor for VxWorks 5.4 / Tornado 2.0 Subject: Re: Question about network card! Subject: Re: Question about network card! Subject: Re: [newbie]SIO driver - UART Tx Buffer Full Subject: Re: m8260FccEnd driver transmit errors Subject: Re: m8260FccEnd driver transmit errors Subject: Re: Block Address Translation Performance Subject: MTD for 28F640J3A Subject: Re: Q: boost::shared_ptr and thread-safety Subject: Re: How to use AF_ROUTE ? Subject: What's definition of a BSP Subject: Re: On S_iosLib_INVALID_FILE_DESCRIPTOR Subject: routing table Subject: FTP Site ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Lightweight heap monitor for VxWorks 5.4 / Tornado 2.0 Date: Wed, 15 May 2002 03:18:56 GMT From: Lori Fraleigh Organization: Real-Time Innovations Message-ID: <3CE1D407.E32D06D9@rti.com> References: John - You may want to check out MemScope. It dynamically patches either the functions in memPartLib or malloc, calloc, realloc, and free (depending on your target architecture) and tracks what tasks and call stacks are allocating memory. It also contains some utilities for checking the consistency of the heap. If you're interested in an evaluation or in buying it, contact your Wind River sales representative. Please see http://www.rti.com/products/scopetools/scopetools.html for more information. Regards, Lori John Poplett wrote: > > Hi, > > I've done some trolling for a lightweight heap monitor for VxWorks and > haven't found anything that ready-made that is portable. I am aware of Hwa > Jin Bae's vxmalloc.c, which is partially based on Doug Lea's malloc.c and am > aware of suggestions on the FAQ sites: > > http://www.xs4all.nl/~borkhuis/vxworks/vxw_pt6.html#6.2 > > Based on suggestions in WRS's application note on heap management: > > http://web1.windriver.com/windsurf/appnotes/ide/tornado/t2/APN116/appnote116 > .pdf > > the cleanest solution seems to be to use the memPart* subroutines and layer > a malloc library on top of them. From there it would be easy to install a > malloc library with built-in monitoring or merge in something like Walter > Bright's mem package. > > Keep in mind that I need to have a solution that is portable (one that would > work on PPC and othe target architectures in addition to SPARCSIM and SIMPC) > and that completely replaces the C language free / malloc interface (compile > time macro swizzling is not an option since we don't have source to all the > components we use). > > On a Tornado 2.02 system (W2K / vxsimpc), I see that memPartLib.o embeds > definitions for malloc and free: > > [C:/Tornado/target/lib] ../../host/x86-win32/bin/nmsimpc.exe --defined > memPartLib.o > 00000000 b .bss > 00000000 d .data > 00000000 t .text > 00000000 t ___gnu_compiled_c > 000009e8 T _free > 000009d4 T _malloc > 000009bc T _memAddToPool > 0000089c t _memAlignedBlockSplit > 00000018 D _memDefaultAlignment > 00000190 T _memPartAddToPool > ... > > With malloc and free bound in the same object as the memPart routines, is > there any hope of providing a malloc / free replacement portably? > > Thanks, > > John - -- Lori Fraleigh Director of Engineering Real-Time Innovations, Inc. http://www.rti.com Phone: 408-734-4200 x121 Fax: 408-734-5009 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: Sat, 18 May 2002 08:38:12 -0400 From: joe durusau Organization: Bellsouth.Net Message-ID: <3CE64B34.2A3505CC@bellsouth.net> References: <989cdae4.0205172212.5bd0a755@posting.google.com> zlybupt wrote: > hi all: > I am developing a network card driver for vxworks,but now met > another problem,I wish someone can give me an explanation. > In my driver design, now I have successfully attached the interrupt > handle to my network card interrupt because when I remove the cable > from the network card,it can succeffully detect this interrupt,but I > wonder why it cannot detect the rxComplete interrupt when I send a > packet to it?? why? where maybe the problem exist?(Here I am sure that > the packet sent by sniffer have been successfully sent.) > Thank you in advance for your answers! It is almost certain that you have not programmed the card to send the interrupt. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: 18 May 2002 05:54:29 -0700 From: demralarood@hotmail.com (Demrala Rood) Organization: http://groups.google.com/ Message-ID: <3baeb99.0205180454.3a9bd2ba@posting.google.com> References: <989cdae4.0205172212.5bd0a755@posting.google.com> zhoul@blrcsv.china.bell-labs.com (zlybupt) wrote in message news:<989cdae4.0205172212.5bd0a755@posting.google.com>... > hi all: > I am developing a network card driver for vxworks,but now met > another problem,I wish someone can give me an explanation. > In my driver design, now I have successfully attached the interrupt > handle to my network card interrupt because when I remove the cable > from the network card,it can succeffully detect this interrupt,but I > wonder why it cannot detect the rxComplete interrupt when I send a > packet to it?? why? where maybe the problem exist?(Here I am sure that > the packet sent by sniffer have been successfully sent.) > Thank you in advance for your answers! I think vxworks must make people to believe that specifics about hardware do not matter. Like vxworks will make all cpu and device look the same and so I have a problem with network card and it makes no difference which network card or which cpu I use. I guess in order of most obvious to least. The cable must be connected between sniffer computer and network card. The receiver must be correctly configured and enabled and the PHY must be correctly configured. The receiver interrupt must be enabled in chip and on host. The packet destination address must be broadcast or unicast and the chip must be configured to accept these or, if the chip can be promiscuous make it so. If the chip is busmaster type and wants to access host memory for descriptors, it must be able to. Many other guesses could be made but fewer would be needed if you give more data about your hardware. DR --------------------------- Newsgroups: comp.os.vxworks Subject: Re: [newbie]SIO driver - UART Tx Buffer Full Date: Mon, 13 May 2002 19:58:33 +0100 From: Roy Bamford Message-ID: References: <11609fa3.0205111909.529826d0@posting.google.com> On Sun, 12 May 2002 04:09:42 +0100, Karpur wrote: > Hi, > > I am writing driver for a uart. The uart works only in interrupt mode. > It does not have the transmitter ready interrupt. Instead it has the Tx > Buffer Full interrupt. > > All the uart drivers (including vxworks template SIO driver) that I have > seen considers only transmitter ready interrupts and not Tx Buffer Full > interrupt. I appreciate any pointers on how the driver should handle > transmission. > > Regards, > Karpur Karpur, The TX ready interrupt comes from the days when UARTs were much simpler than they are now. These simple UARTs have no FIFOs, either for Transmit or receive. You can use them by polling or interrupts. On the TX side, you load a character into to UART and provided it was set up properly, you got an interrupt when that charater had been sent and it was ready for another. The receive side is simmilar, you get an interrupt when a charater arrives and have the arrival time of the next charater to retrieve it before it gets overwritten and the UART signals an overrun error. All these interrupts were a pain, so UARTS grew buffers. You can now send characters to the UART until you get a TX Buffer Full interrupt. Depending on the UART, you can program some 'headroom' so that the interrupt occurs before charaters are lost. I suspect that your UART also has a TX ready interrupt so you know when you can send it some more characters after getting TX Buffer Full. Regards, Roy Bamford - -- Computer users fall into two groups:- Those that do backups Those that have never had a hard drive fail --------------------------- Newsgroups: comp.os.vxworks Subject: Re: m8260FccEnd driver transmit errors Date: 13 May 2002 12:09:59 -0700 From: dannyc@harmonic.co.il (Danny Cohen) Organization: http://groups.google.com/ Message-ID: <5288f97d.0205131109.7bf0754c@posting.google.com> References: I face similar problem. I'm using EST8260 ver 1.2/1 with Tornado 5.4 and i attached PHY (CTX-1170) to the FCC. When i start running the VxWorks image I face the following problem: 1. PING from the Host to the Board 2. Packet is received by the Target FCC (Rx LED is blinked on the PHY, "ifShow" stat command indicate packet receive). 3. PONG is sent back to the HOST via FCC (Tx LED is blinked on the PHY, "ifShow" stat command indicate packet transmit). 4. BUT the packet is lost and not reach the HOST. I think it's also related somehow to the transmit clock. What do you think? how can you adjust the clock? "Graham Baxter" wrote in message news:... > Achim, > > You can do everything in: > > sysFccEnetEnable() in sysLib.c > > where you set the correct port pins. > > The register I changed for clock routes was something like: > M8260_CMXFCR > > -- > > Regards, > > Graham Baxter > Freelance Software Engineer > Graham Baxter (Software) Limited (VxWorks and pSOS BSP's) > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > "Achim Zimmer" wrote in message > news:ab5eg7$f3404$1@ID-102608.news.dfncis.de... > > sorry, i dont know HOW to change !! > > > > "Achim Zimmer" schrieb im Newsbeitrag > > news:ab5ef3$f68hn$1@ID-102608.news.dfncis.de... > > > Thats fine -> Thay can be changed with software !! > > > > > > But I dont know why, because this driver is only delivered as Object, no > > > source ! > > > > > > We found that it could be, we have to invert transmit clock, but I dont > know > > > how, because the driver always overwrites my register reconfiguration. > > > > > > Thanks > > > > > > "Graham Baxter" schrieb im Newsbeitrag > > > news:aauq28$ahb$1@knossos.btinternet.com... > > > > I had the same problem when we connected the transmit and receive > clocks > the > > > > wrong way round. > > > > > > > > They can be changed with software. > > > > > > > > My device was exactly oposite to the EST evaluation board. > > > > > > > > I hope this helps. > > > > > > > > -- > > > > > > > > Regards, > > > > > > > > Graham Baxter > > > > Freelance Software Engineer > > > > Graham Baxter (Software) Limited > > > > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > > > > > > > > > > > > > > "Achim Zimmer" wrote in message > > > > news:aatei0$ctprc$1@ID-102608.news.dfncis.de... > > > > > I try to use the m6260FccEnd driver, supplied by WRS eith ADS8260 > BSP > to > use > > > > > with our own board. > > > > > The connection is equal, but we use a 971 Phy chip instead of 970. > > > > > > > > > > The problem we are running in, is: > > > > > The packets on the LAN the board transmits have in 80 % errors, > means, > there > > > > > are 8 bytes repeated in the frame, and theoriginal 8 bytes are lost. > > > > > > > > > > Has anyone this problem too and there are any solutions ? > > > > > > > > > > Thanks in advance. > > > > > Any hint is welcome. > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: m8260FccEnd driver transmit errors Date: Sat, 18 May 2002 16:20:25 +0000 (UTC) From: "Graham Baxter" Organization: Graham Baxter (Software) Limited Message-ID: References: <5288f97d.0205131109.7bf0754c@posting.google.com> Danny, Check that MRBLR is a multiple of 32. Check with Wind River that you have the latest patches. At times like this, I would get someone to put a scope on the transmit and receive clocks coming into the 8260. Check that they are the right way round. - -- Regards, Graham Baxter EMAIL: gbaxter@NOSPSAMbcs.org.uk "Danny Cohen" wrote in message news:5288f97d.0205131109.7bf0754c@posting.google.com... > I face similar problem. > > I'm using EST8260 ver 1.2/1 with Tornado 5.4 and i attached PHY > (CTX-1170) to the FCC. When i start running the VxWorks image I face > the following problem: > 1. PING from the Host to the Board > 2. Packet is received by the Target FCC (Rx LED is blinked on the PHY, > "ifShow" stat command indicate packet receive). > 3. PONG is sent back to the HOST via FCC (Tx LED is blinked on the > PHY, "ifShow" stat command indicate packet transmit). > 4. BUT the packet is lost and not reach the HOST. > I think it's also related somehow to the transmit clock. > > What do you think? how can you adjust the clock? > > "Graham Baxter" wrote in message news:... > > Achim, > > > > You can do everything in: > > > > sysFccEnetEnable() in sysLib.c > > > > where you set the correct port pins. > > > > The register I changed for clock routes was something like: > > M8260_CMXFCR > > > > -- > > > > Regards, > > > > Graham Baxter > > Freelance Software Engineer > > Graham Baxter (Software) Limited (VxWorks and pSOS BSP's) > > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > > > > "Achim Zimmer" wrote in message > > news:ab5eg7$f3404$1@ID-102608.news.dfncis.de... > > > sorry, i dont know HOW to change !! > > > > > > "Achim Zimmer" schrieb im Newsbeitrag > > > news:ab5ef3$f68hn$1@ID-102608.news.dfncis.de... > > > > Thats fine -> Thay can be changed with software !! > > > > > > > > But I dont know why, because this driver is only delivered as Object, no > > > > source ! > > > > > > > > We found that it could be, we have to invert transmit clock, but I dont > > know > > > > how, because the driver always overwrites my register reconfiguration. > > > > > > > > Thanks > > > > > > > > "Graham Baxter" schrieb im Newsbeitrag > > > > news:aauq28$ahb$1@knossos.btinternet.com... > > > > > I had the same problem when we connected the transmit and receive > > clocks > > the > > > > > wrong way round. > > > > > > > > > > They can be changed with software. > > > > > > > > > > My device was exactly oposite to the EST evaluation board. > > > > > > > > > > I hope this helps. > > > > > > > > > > -- > > > > > > > > > > Regards, > > > > > > > > > > Graham Baxter > > > > > Freelance Software Engineer > > > > > Graham Baxter (Software) Limited > > > > > EMAIL: gbaxter@NOSPAMbcs.org.uk > > > > > > > > > > > > > > > > > > > > "Achim Zimmer" wrote in message > > > > > news:aatei0$ctprc$1@ID-102608.news.dfncis.de... > > > > > > I try to use the m6260FccEnd driver, supplied by WRS eith ADS8260 > > BSP > > to > > use > > > > > > with our own board. > > > > > > The connection is equal, but we use a 971 Phy chip instead of 970. > > > > > > > > > > > > The problem we are running in, is: > > > > > > The packets on the LAN the board transmits have in 80 % errors, > > means, > > there > > > > > > are 8 bytes repeated in the frame, and theoriginal 8 bytes are lost. > > > > > > > > > > > > Has anyone this problem too and there are any solutions ? > > > > > > > > > > > > Thanks in advance. > > > > > > Any hint is welcome. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Block Address Translation Performance Date: Mon, 13 May 2002 14:58:12 -0500 From: James Organization: Raytheon Company Message-ID: <3CE01AD4.27A9F051@nospam.com> References: <3CE00A4E.C3A20F72@nospam.com> Only 1GB of VME A32 memory is mapped via the BAT registers...

James wrote:

Hello,

I'm using a MVME2700 single board computer from Motorola.  I have 4gig of VME A32 addressable memory which is mapped using BAT registers.  Has anyone experienced poor performance the first time through a section of code?  For example, the first time through a function may take 118us, whereas future accesses through it only takes 82us.  The code does not dynamically allocate any new memory via malloc.  I'm guessing that the translation look-ahead buffers (TLB) on the MVME2700 PPC are remembering the previously accessed locations, thereby decreasing future VME A32 access times.  If this is so, how can I estimate when the TLB's will forget or lose this information, requiring another performance hit?

If anybody has had similar experiences with the BAT registers and has some ideas on how to increase VME A32 memory performance, please respond.

Thanks,
James

--------------------------- Newsgroups: comp.os.vxworks Subject: MTD for 28F640J3A Date: Mon, 13 May 2002 20:05:47 GMT From: "doug dotson" Organization: Verio Message-ID: Does anyone know of a TFFS MTD for an Intel StrataFlash 28F640J3A? Thanks, Doug --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Q: boost::shared_ptr and thread-safety Date: 13 May 2002 19:41:27 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205131841.3640ee01@posting.google.com> References: <3CDFB390.F880DADF@rovema.de> Hello, > Im trying to port a multithreaded Win32 console application which uses > the boost::shared_ptr class (www.boost.org) to run under VxWorks > (Tornado 2.0.2, x86, gcc 2.7.2). Shared_ptr claims to be thread-safe for > Win32, Linux and "any platform that supports pThread". > But I explicitly DON'T want to use the pThread library for VxWorks > because of performance reasons. What performance reasons? The pthreads for VxWorks maps directly onto the task layer, and shouldn't impose significant performance penalties, especially not to a running thread (which is, after all, just a VxWorks task). > My question now is: > How to make boost::shared_ptr thread-safe ? Well, how are they ensuring this in the existing code? If it is pthreads based, then I assume some type of mutex is in use somewhere. That should be trivial to port to a VxWorks mutex. Try searching in the tree for the pthread mutex APIs, that should give you a good idea of where to look. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to use AF_ROUTE ? Date: 13 May 2002 21:10:58 -0700 From: sai_dattathrani@infy.com (sai) Organization: http://groups.google.com/ Message-ID: <8dff11cd.0205132010.893d052@posting.google.com> References: <2fc4ea3f.0205110013.4654e72f@posting.google.com> Hi, I am not sure why you want to define AF_NETLINK to AF_ROUTE. You can use AF_ROUTE directly as follows. sockfd = socket(AF_ROUTE,SOCK_RAW,0); generally you don't have to use bind for raw sockets. But if you bind, it will not give panic.If you want to use bind, i am not sure if you can use sockaddr_nl. You can try using sockaddr_in and try. Richard Stevens(Second edition) book has a good description on routing sockets. rgds sai zw84611@sina.com (zhuwei) wrote in message news:<2fc4ea3f.0205110013.4654e72f@posting.google.com>... > Hello,everyone!I have to translate the following Linux source codes to > VxWorks: > I have some problems: > 1. what's the counterpart of AF_NETLINK(in linux) in VxWorks? > 2. why I can't bind(),like following: > ========================================================== > linux source: > ============= > int rtnl_open() { > struct sockaddr_nl rtaddr; > int fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE); > int rtaddr_len; > > if(fd <0) { > perror("rtnl_open"); > return -1; > } > rtaddr.nl_family = AF_NETLINK; > rtaddr.nl_groups = 0xFFFFFFFF; > > if(bind(fd,(struct sockaddr*)&rtaddr,sizeof(rtaddr))<0) { > perror("rtnl_open"); > return -1; > } > rtaddr_len = sizeof(rtaddr); > getsockname(fd,(struct sockaddr*)&rtaddr,&rtaddr_len); > > return fd; > } > =================================================================== > I changed the above into following codes in VxWorks: > ------------------------------------------------------------------- > #define AF_NETLINK AF_ROUTE > > int rtnl_open() > { > struct sockaddr_nl rtaddr; > > int fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE); /* vxWorks > bootimage should include BSD Routing Sockets */ > int rtaddr_len; > > if(fd <0) > { > perror("rtnl_open_error1\n"); > return -1; > } > > rtaddr.nl_family = AF_NETLINK; > rtaddr.nl_groups = 0xFFFFFFFF; > > if(bind(fd,&rtaddr,sizeof(rtaddr))<0) > { /* here,panic: raw_usrreq */ > perror("rtnl_open_error2\n"); > return -1; > } > > if(bind(fd,(struct sockaddr*)&rtaddr,sizeof(rtaddr))<0) > { /* here,panic: raw_usrreq */ > perror("rtnl_open_error2\n"); > return -1; > } > rtaddr_len = sizeof(rtaddr); > getsockname(fd,(struct sockaddr*)&rtaddr,&rtaddr_len); > > return fd; > } > ------------------------------------------------------------------- > I have include the BSD Routing Sockets(INCLUDE_ROUTE_SOCK) in VxWorks > image. > But when the programme execute,there is an panic like: > 0x1f6ba88 (tDbgTask): panic: raw_usrreq > (when step into bind(fd,(struct sockaddr*)&rtaddr,sizeof(rtaddr))) > > I don't know why,can anyone help me? --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: What's definition of a BSP Date: 18 May 2002 17:45:36 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: Anyone know a more precise definition of board support package (BSP)? I use VxWorks on a board that's derived from a WindRiver reference board. This board has some additional proprietary h/w that doesn't have any WRS drivers. In particular, I'd like to know the relationship between the RTOS (VxWOrks in this case), BSP, and all the drivers. What should be included and what should be excluded from BSP. Thanks, Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: Re: On S_iosLib_INVALID_FILE_DESCRIPTOR Date: Tue, 14 May 2002 10:16:52 +0200 From: "Werner Schiendl" Message-ID: <3ce0c7eb$1@brateggebdc5.br-automation.co.at> References: <3ccd1567$1@brateggebdc5.br-automation.co.at> Hi, again, the ERRNO column in the output of the i command is meaningful _only_ if there were an error condition reported in your program before and you didn't do anything that could change it in the meanwhile. Since you don't actually know, how all the VxWorks (and possible third party software you are using) work with the errno value internally, it's normally the best to completely ignore the column. In your software, when you encounter and error and want to output some diagnostic information, you can output the value of errno at that point of the program flow - directly after you got the error response. This is the _only_ location, where the value of errno is defined. You may also find the following commands helpful for outputting debug information: perror printErrno logMsg look them up in the reference manual. hth Werner P. S. Really, ignore the ERRNO column of the output from i command. "yyzhong" wrote in message news:b35585ab.0205080451.18cb79c6@posting.google.com... > Another time I encountered this kind of error again . > It's in the sample given by the 'Tornado getting started '. > After I download the lighting.out, and start a windSH: > > -> progStart > value = 0 = 0x0 > -> i > NAME ENTRY TID PRI STATUS PC SP ERRNO DELAY > ---------- ------------ -------- --- ---------- -------- -------- ------- - ----- > tExcTask _excTask 4af9130 0 PEND 4276be 4af904c d0003 0 > tLogTask _logTask 4af4718 0 PEND 4276be 4af4630 d0003 0 > tWdbTask 0x417cc4 4af0bc8 3 READY 4276be 4af0aa4 d0003 0 > tCosmos _cosmos 4ae5678 200 READY 4276be 4ae55dc d0003 0 > tSchlep _schlep 4ade5b0 220 PEND 4276be 4ade4f4 d0003 0 > tMonitor _monitor 4ad0420 230 READY 491eee0 4ad03a4 d0003 0 > tCrunch _crunch 4ad74e8 240 READY 427554 4ad74a8 0 0 > value = 0 = 0x0 > -> > > I didn't got the expected "Divide Error" after an hour! --------------------------- Newsgroups: comp.os.vxworks Subject: routing table Date: 14 May 2002 06:54:02 -0700 From: vrrajesh@intotoinc.com (rajesh vr) Organization: http://groups.google.com/ Message-ID: <68c7b7f1.0205140554.7fbca9c1@posting.google.com> Hi, I have set of question on routing table 1.how do get the default route, 2.Is there any system call for updating the elements of a route(metrics) Is there any way I can add or Delete Route other than the system calls found in routeLib ? 3.How do I access the Routing Table .(like first route ,next route) 4. How is the routing table maintained. 5. m2IpLib mentions some system call for setting and retrieving routing Table entry.By setting the MIB-2 routing entry ,will the kernel routing entry be set for the same route? Can I assume that routes in MIB-2 routing entry and kernel routing entry are the same. Thanking in advance, rajesh --------------------------- Newsgroups: comp.os.vxworks Subject: FTP Site Date: 14 May 2002 07:26:16 -0700 From: m1971murali@hotmail.com (MURALI) Organization: http://groups.google.com/ Message-ID: Hi, I was trying to access the site ftp://ftp.ece.ucdavis.edu/pub/snmp/ucd-snmp.tar.gz but could not access the same. Can any one of you help me in letting me know whether the above ID is correct and if yes and if the site has been moved to different location, can you pls give me the new location also. Thanks for the help. Regards Murali --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sun May 19 21:35:56 2002 From: "Sudhakar Tadi" Date: Sun May 19 21:35:58 PDT 2002 Subject: VxWorks:Memory Mapped I/O This is a multi-part message in MIME format. ------=_NextPart_000_014C_01C1FFE5.A5BF83F0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi All, I am porting a Qlogic FC driver to VxWorks, for Intel X86 BSP. In = VxWorks for I/O mapped devices one can use sysInByte() sysOutByte() = etc.(corresponds to IN and OUT functions in Linux). =20 For memory maped devices, I understand that we have to edit = mmuPhysDesc[] table, if we want to support Virutal memory. My question is if I use memory mapped I/O what fucntions I can use = (like readb() writeb() in linux). OR Am I missing something from = VxWorks. Please advise. Has anybody ported linux SCSI drivers to VxWorks,if anyone wants to = share his experience, will be of great helpful to me. Thanks in Advance -Tadi Sudhakar Tadi HCL Technologies,Chennai, INDIA. ph: 91-44-3741939-42 ex:2312=20 Fax:91-44-3741038 http://san.hcltech.com ------=_NextPart_000_014C_01C1FFE5.A5BF83F0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi All,
 
    I am porting a = Qlogic FC driver=20 to VxWorks, for Intel X86 BSP. In VxWorks for I/O mapped devices = one can=20 use sysInByte() sysOutByte() etc.(corresponds to IN and OUT functions in = Linux). 
 
For memory maped devices, I understand = that we have=20 to edit mmuPhysDesc[] table, if we want to support Virutal = memory.
My question is if I use memory  = mapped I/O=20 what fucntions I can use (like   readb() writeb() in = linux).  OR=20 Am I missing something from VxWorks.  Please advise.
 
Has anybody ported linux SCSI drivers = to VxWorks,if=20 anyone wants to share his experience, will be of great helpful to=20 me.
 
Thanks in Advance
-Tadi
 
Sudhakar  Tadi
HCL=20 Technologies,Chennai, INDIA.
ph: 91-44-3741939-42 ex:2312=20
Fax:91-44-3741038
http://san.hcltech.com
------=_NextPart_000_014C_01C1FFE5.A5BF83F0-- From vxwexplo-errs@csg.lbl.gov Mon May 20 04:03:11 2002 From: Vxworks Exploder Date: Mon May 20 04:03:13 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon May 20 04:03:07 PDT 2002 Subject: timer_create function question Subject: Re: Question about network card! Subject: RPC: how to force svc_run() to return ? / svc_getreq() usage ? Subject: Re: What's definition of a BSP Subject: Re: What's definition of a BSP Subject: Full VxSim vs Target Subject: Re: Question on BootCmdLoop() Subject: representation of a double in vxWorks Subject: Re: What's definition of a BSP Subject: Re: Question on BootCmdLoop() Subject: Re: representation of a double in vxWorks Subject: Re: Question about network card! ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: timer_create function question Date: Sun, 19 May 2002 11:47:47 +0000 (UTC) From: "Dexter Nixon" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: Hi, I am using timer_create functions for timer functionality. I have registered a handler when the timer expires. I started the timer, but consider, due to some reason my program crashes before the timer expiry. Now in the VxWorks target console interrupt: timerWdHandler: kill failed (timer=0x4bcc8e0, tid=0x4b24ec8, errno=0x16) I am getting this message continuously until i call timer_cancel by giving the timer id what i created. How can i avoid this. I am using Tornado Prototyper Plus. THanks, Nix - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: 19 May 2002 06:06:01 -0700 From: zhoul@blrcsv.china.bell-labs.com (zlybupt) Organization: http://groups.google.com/ Message-ID: <989cdae4.0205190506.3e222327@posting.google.com> References: <989cdae4.0205172212.5bd0a755@posting.google.com> <3CE64B34.2A3505CC@bellsouth.net> joe durusau wrote in message news:<3CE64B34.2A3505CC@bellsouth.net>... > zlybupt wrote: > > > hi all: > > I am developing a network card driver for vxworks,but now met > > another problem,I wish someone can give me an explanation. > > In my driver design, now I have successfully attached the interrupt > > handle to my network card interrupt because when I remove the cable > > from the network card,it can succeffully detect this interrupt,but I > > wonder why it cannot detect the rxComplete interrupt when I send a > > packet to it?? why? where maybe the problem exist?(Here I am sure that > > the packet sent by sniffer have been successfully sent.) > > Thank you in advance for your answers! > > It is almost certain that you have not programmed the card to send > the interrupt. I wonder how the cable-removal interrupt can be detected,while the rxComplete interrupt cannot be detected?According to my opinion,the two interrupts are handled in the same function in the driver and by judging corresponding bit in the interrupt status register to determine the interrupt type,and I wonder the procedure of packets transferred from the cable to FIFO is not implemented in the driver code, it should be done by the system bus,is it true? thank you for your answer in advance! > > Speaking only for myself, > > Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: RPC: how to force svc_run() to return ? / svc_getreq() usage ? Date: 19 May 2002 12:31:49 -0700 From: r.stelzer@stoye.de (Rainer Stelzer) Organization: http://groups.google.com/ Message-ID: <78a5d372.0205191131.6e3a169c@posting.google.com> Hi vxworkers, although the vxworks RPC library documentation is quite poor I successfully set up a working client server connection between a vxworks MVME2304 and a custom PPC860 target in tcp-mode. When testing the connection in different stress situations I faced the problem that, case I simply reboot the client side while a transfer takes place, the server didn't notify this state. I would expect the server side to time out and throuw an exception, but nothing happens. Case the client tries to reconnect in this situation it exits with "server error- error 60". Only restarting the server task can clear this situation. My questions: - - Any idea where the server hangs and how to force svc_run to return in this situation? - - do I have to set timeout parameters in works like I do on Unix machines ? Or are they special in any way ? - - Has anybody out there a working vxworks svc_getreq() example ? Thanks in advance. rgds Rainer --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: What's definition of a BSP Date: Sun, 19 May 2002 23:34:02 GMT From: Jack Klein Organization: AT&T Worldnet Message-ID: References: On 18 May 2002 17:45:36 -0700, ryeung@earthlink.net (Raymond Yeung) wrote in comp.arch.embedded: > Anyone know a more precise definition of board support package (BSP)? > I use VxWorks on a board that's derived from a WindRiver reference > board. This board has some additional proprietary h/w that doesn't > have any WRS drivers. > > In particular, I'd like to know the relationship between the RTOS (VxWOrks > in this case), BSP, and all the drivers. What should be included and > what should be excluded from BSP. > > Thanks, > Raymond To be honest, I've come to the conclusion that the term BSP often covers widely different meanings these days. The most common meaning I can get from RTOS vendors is that you hand them a board and they hand you a package that takes it from hardware reset up to loading and starting their OS, to which you the add your application. Very often devices must be initialized long before the RTOS is running, for example an on-chip memory controller, so the fact that a driver exists to handle the device under the RTOS is not necessarily helpful in getting the board up to the point where the RTOS can start. Consider the desk top PC, whether it is running Windows, Linux, QNX, VxWorks, whatever. It needs to have a BIOS that can initialize memory, keyboard, video display, USB devices, interrupt controller, floppy disk drivers, IDE and/or SCSI drivers, all before it goes out to those drives to find an OS to load. Once it loads Windows, Linux, QNX, VxWorks, whatever, the BIOS code for handling those devices is superseded by drivers for the OS. Just replace the word BIOS with BSP. For many embedded processors these days, there are dozens of on-chip configuration registers that need to be set properly in the process. If you want special POST code that can't be easily performed after your RTOS is running, like a PC memory test, that would need to be added to the BSP or BIOS as well. - -- Jack Klein Home: http://JK-Technology.Com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: What's definition of a BSP Date: Sun, 19 May 2002 17:41:53 -0700 From: "Michael R. Kesti" Organization: MK Associates Message-ID: <3CE84651.E276BAC9@gv.net> References: Reply-To: mkesti@gv.net Raymond Yeung wrote: >Anyone know a more precise definition of board support package (BSP)? A BSP, or board support package, is the code and documentation files that serve as the interface between a specific hardware set and vxWorks. Its code is where system initialization begins, device drivers are incorporated, and other hardware specific issues are handled. >I use VxWorks on a board that's derived from a WindRiver reference >board. This board has some additional proprietary h/w that doesn't >have any WRS drivers. > >In particular, I'd like to know the relationship between the RTOS (VxWOrks >in this case), BSP, and all the drivers. What should be included and >what should be excluded from BSP. One answer is that if it is code that depends directly on the hardware, then it belongs in the BSP and should have a device driver written for it. This allows one to write application programs that are fully abstracted by vxWorks' I/O libraries. If your new hardware is simple, though, perhaps just a latch that controls some LEDs, then you might agree that a device driver is a bit of overkill. You might choose, instead, to directly control that latch from your applications. Another reason to use this approach is when you have performance requirements that can't be met using the I/O libs. You get to decide and, as always, there are trade-offs. - -- ======================================================================== 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: Full VxSim vs Target Date: 19 May 2002 20:07:44 -0700 From: andreas_lindell75@hotmail.com (Rikard) Organization: http://groups.google.com/ Message-ID: <5947a04.0205191907.3ae9c062@posting.google.com> Hello, I am using full VxSim on a PPC installation. Now I want to broadcast messages. Would anyone know how to get the ULIP network driver for VxSim to support exchange of BROADCAST network packages? I've implemented a simple network service discovery protocol based on UDP broadcasts. On VxSim, my server is unable to receive the broadcast packages that my clients sent out. On VxSim I did an ifShow in Windsh and it showed: nt (unit number 0): Flags: (0xf1) UP POINT-TO-POINT RUNNING Type: ETHERNET_CSMACD blah ...blah ... blah 0 packets received; 1 packet sent blah ...blah ... blah I think that's the reason I can't receive the broadcast packages. On a target with ifShow printing: symeth (unit number 0): Flags: UP ARP BROADCAST RUNNING blah ...blah ... blah ,here BROADCAST is running and the server is able to receive my clients' broadcast messages properly. Can I get VxSim with BROADCAST or is this a limitation of VxSim? Regards Rikard --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question on BootCmdLoop() Date: 19 May 2002 22:55:23 -0700 From: jade_chen@21cn.com (shy) Organization: http://groups.google.com/ Message-ID: <937a20a8.0205192155.352b5749@posting.google.com> References: <937a20a8.0205151810.34e7c98f@posting.google.com> I just want to know what is the entry address? --------------------------- Newsgroups: comp.os.vxworks Subject: representation of a double in vxWorks Date: 19 May 2002 23:58:22 -0700 From: zvikag@yahoo.com (zvi) Organization: http://groups.google.com/ Message-ID: Hello vxWorkers!! I hope this is not too foolish question ... what is the representation of a double in the memory using vxWorks with ppc board? is someone can help me? Thanks, Zvi. --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: What's definition of a BSP Date: Mon, 20 May 2002 07:10:48 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: I have encountered this sort of question many times before, and there are (as usual) two answers, depending on your specific requirements: The very basic (i.el. minimal) definition of BSP is startup code, initialization and a set of drivers that are required by VxWorks and its bundled development environment. I.e. everything one needs to begin developing applications on teh board using the full breath of Tornado environment. This would include board hardware init, times, serial drivers, and if the board has either an ethernet or a standard storage (i.e. SCSI or IDE) interfaces, throw them in too, to enable the budnled TCP/IP stack and Dos file system. In that particular definition, custom or even common I.O interfaces don't count because VxWorks has no use for them, and does not even define the API for their drivers. VxWorks BSP test suite can not test these extra I/O either. An alternative approach would assume that your application will not be running on this particular board forever, and there would be a need to port the application to different boards, then you should define an extended set of I/O interfaces for this type of application, clearly define an API for drivers and implement them as part of the BSP, and then, when needs come, one would enable your application on a new platform simply by implementing the extended BSP. A digital camera would be a good example where an extended BSP should be conisdered, if you are writing a digital camera application that is expected to be used in numerous camera models. Hope this helps, - Leonid Rosenboim "Raymond Yeung" wrote in message news:d83bb8e3.0205181645.7ae21e05@posting.google.com... > Anyone know a more precise definition of board support package (BSP)? > I use VxWorks on a board that's derived from a WindRiver reference > board. This board has some additional proprietary h/w that doesn't > have any WRS drivers. > > In particular, I'd like to know the relationship between the RTOS (VxWOrks > in this case), BSP, and all the drivers. What should be included and > what should be excluded from BSP. > > Thanks, > Raymond --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question on BootCmdLoop() Date: Mon, 20 May 2002 09:21:11 +0100 From: "Tim Allen" Organization: ntlworld News Service Message-ID: <8m2G8.68$kk7.23619@news8-gui.server.ntli.net> References: <937a20a8.0205151810.34e7c98f@posting.google.com> <937a20a8.0205192155.352b5749@posting.google.com> "shy" wrote in message news:937a20a8.0205192155.352b5749@posting.google.com... > I just want to know what is the entry address? When loading a vxWorks image, the entry address is the start address of code in the compiled image. This is sysInit() in sysALib.s, which is linked to run from RAM_LOW_ADRS defined in your makefile/config.h and/or project definition. Regards, Tim. - -------------------- Tim Allen Ltd Real Time Embedded Software Consultancy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: representation of a double in vxWorks Date: Mon, 20 May 2002 10:15:43 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CE8CD2E.3000900@spamm.me.l8s.co.uk> References: > what is the representation of a double in the memory > using vxWorks with ppc board? I expect they are in the 'standard' format - which is: Double floating point are 64 bit: 63 sign 62-52 exponent, biased so that '1' has exponent 0x3ff 51-0 Mantissa, without leading bit. If the exponent is 0, then the mantissa doesn't have a hidden bit. If the exponent is max (0x7ff) then the number is either infinity or Nan (not a number), infinity has all mantissa bit zero, nan at least one non-zero. Float are 32 bits, with 3 less bits of exponent and 23 bits of mantissa. For ppc I think the 64 bit value has an endianness constistent with that of the system. David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: Mon, 20 May 2002 06:36:29 -0400 From: joe durusau Organization: Bellsouth.Net Message-ID: <3CE8D1AD.3ADCFE02@bellsouth.net> References: <989cdae4.0205172212.5bd0a755@posting.google.com> <3CE64B34.2A3505CC@bellsouth.net> <989cdae4.0205190506.3e222327@posting.google.com> zlybupt wrote: > joe durusau wrote in message news:<3CE64B34.2A3505CC@bellsouth.net>... > > zlybupt wrote: > > > > > hi all: > > > I am developing a network card driver for vxworks,but now met > > > another problem,I wish someone can give me an explanation. > > > In my driver design, now I have successfully attached the interrupt > > > handle to my network card interrupt because when I remove the cable > > > from the network card,it can succeffully detect this interrupt,but I > > > wonder why it cannot detect the rxComplete interrupt when I send a > > > packet to it?? why? where maybe the problem exist?(Here I am sure that > > > the packet sent by sniffer have been successfully sent.) > > > Thank you in advance for your answers! > > > > It is almost certain that you have not programmed the card to send > > the interrupt. > I wonder how the cable-removal interrupt can be detected,while the > rxComplete interrupt cannot be detected?According to my opinion,the > two interrupts are handled in the same function in the driver and by > judging > corresponding bit in the interrupt status register to determine the > interrupt type,and I wonder the procedure of packets transferred from > the cable to FIFO is not implemented in the driver code, it should be > done by the system bus,is it true? > thank you for your answer in advance! > > > > Speaking only for myself, > > > > Joe Durusau Understand that since I'm not familiar with your card, I'm guessing. But with other complex IO devices, for instance 1553 interface modules, you have to program each and every interrupt that you want it to send. Most devices will have an interrupt mask of some sort in a register somewhere that has to have specific bits set in order to enable interrupts on certain conditions. There are other possibilities, of course, as someone else pointed out. You might start by sending a broadcast packet and see if the card responds. (I am assuming that you have checked at theinput pins of the device and have determined that voltage levels and polarities are correct by now.) Speaking only for myself, Joe Durusau --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue May 21 04:03:13 2002 From: Vxworks Exploder Date: Tue May 21 04:03:16 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue May 21 04:03:09 PDT 2002 Subject: Re: Question about network card! Subject: MODBUS TCP/IP for VxWorks Subject: Re: embedded web server-loading images!! Subject: Dynamic MIB addition Subject: Re: Dynamic MIB addition Subject: using zlib to decompress images created by deflate Subject: Re: why doesn't kill(TID, 9) work? Subject: Adding a new build using Tornado Subject: Re: why doesn't kill(TID, 9) work? Subject: Re: what are the BSD driver and END driver? Subject: Re: routing table Subject: Re: Question about network card! Subject: Re: What's definition of a BSP Subject: A question!! Subject: Bridge Subject: Re: Question about network card! Subject: Re: What's definition of a BSP Subject: Re: Bridge Subject: routine called from kernel/application Subject: Re: A question!! ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: Mon, 20 May 2002 11:09:41 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CE8D9D1.7020706@spamm.me.l8s.co.uk> References: <989cdae4.0205172212.5bd0a755@posting.google.com> <3CE64B34.2A3505CC@bellsouth.net> <989cdae4.0205190506.3e222327@posting.google.com> <3CE8D1AD.3ADCFE02@bellsouth.net> >>> >>>>hi all: >>>> I am developing a network card driver for vxworks,but now met >>>>another problem,I wish someone can give me an explanation. >>>> In my driver design, now I have successfully attached the interrupt >>>>handle to my network card interrupt because when I remove the cable >>>>from the network card,it can succeffully detect this interrupt,but I >>>>wonder why it cannot detect the rxComplete interrupt when I send a >>>>packet to it?? why? where maybe the problem exist?(Here I am sure that >>>>the packet sent by sniffer have been successfully sent.) >>>> Thank you in advance for your answers! Have you tried reading the status registers by hand to see if they indicate that a packet has arrived? If the card is writing to a memory area (rather than requiring the cpu to read the frame out of an internal fifo) look at the memory area to see what it contains. David --------------------------- Newsgroups: comp.os.vxworks Subject: MODBUS TCP/IP for VxWorks Date: 20 May 2002 04:22:08 -0700 From: techsoul2001@yahoo.com (raghuram) Organization: http://groups.google.com/ Message-ID: Hi, We are in the process of building a process control system using SBCs running VxWorks RTOS. I would like to know if there is any standard MODBUS over TCP/IP driver available for VxWorks. Any pointers in this regard are appreciated. thanks and best regards, raghuram. --------------------------- Newsgroups: comp.os.vxworks,comp.realtime,comp.arch.embedded Subject: Re: embedded web server-loading images!! Date: 20 May 2002 05:53:40 -0700 From: hema_tm@rediffmail.com (hema) Organization: http://groups.google.com/ Message-ID: <56b5d4bb.0205200453.72842e7f@posting.google.com> References: <56b5d4bb.0205160432.160d016@posting.google.com> <3CE3C069.A6A77B4B@sepro-robotique.com> <56b5d4bb.0205162019.55a0d5fd@posting.google.com> <3ce4cf99$1@brateggebdc5.br-automation.co.at> hai , thanx Werner i was looking out for something similler to what u posted.to convert a .gif file to a charecter array and load it dynamically when i run the script. i will try looking in it... rgrds, hema "Werner Schiendl" wrote in message news:<3ce4cf99$1@brateggebdc5.br-automation.co.at>... > Hi Hema, > > I think it would be more helpful to explain _what_ you are trying to > accomplish instead of how you thought it might be possible. > > Anyway, from what I _assume_ you could need, I suggest to lookup the tool > memdrvbuild in the documentation (host tool). > > This tool allows you to convert files (and maybe directories) in a format > that can be compiled into the target image (vxWorks or downloadable image). > On the target system, you can insert the files read-only into the filesystem > and access the files as usual. > > hth > Werner > > > "hema" wrote in message > news:56b5d4bb.0205162019.55a0d5fd@posting.google.com... > > hai > > thanx for your response. > > but i was looking out to convert a .gif to binary or hex,and not a > > binary data to ascii.well to be more specific about my doubt.i want to > > generate a GIF say xyz.gif.i need the hex equivalent of this GIF in > > specific to MPC860 byte order format. > > hope i m clear!!!! > > > > > > thanx , > > > > > > hema > > > > > > Emmanuel Herbreteau wrote in message > news:<3CE3C069.A6A77B4B@sepro-robotique.com>... > > > hema wrote: > > > > i m working on vxWorks 5.4 and target env: MPC860T.i planned to web > > > > enable my application running on the target.i use the vxHttp server > > > > > > Hi, > > > > > > I'm not sure to really understand your needs, but... > > > > > > If you want to include "raw bitmap" in your executable, > > > you can follow this path : > > > > > > 1. create your picture in a standard format (BMP, GIF, etc...) > > > 3. run the tornado's utility : "binToAsm image.bmp > image.s" > > > 4. replace ".data" by ".text" in the new file "image.s" > > > 5. include this file in your project > > > > > > You can now access your picture using a simple pointer in your code. > > > > > > Hope this help (?) > > > > > > Regards > > > Emmanuel --------------------------- Newsgroups: comp.os.vxworks Subject: Dynamic MIB addition Date: Mon, 20 May 2002 16:20:50 +0200 From: "Gleb Feigstein" Message-ID: <3ce8f84f@news.barak.net.il> Hi I'm trying to add a private MIB as discribed in the manual. The MIB compiler with option -mib.c generates a C file that containes all the nodes and also other files as required. Now I need to add the module to the running MIB by using snmpdTreeAdd. The first parameter is the path - 1.3.6.1.4.1.2393 in my case. The second is a pointer to a tree node. What should it be? In the generated file there is only one non-static, but it is the root of the complete MIB tree. I tried to make the first private node global and give it as parameter but it doesn't work proprly. Please assist. Thanks Gleb --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Dynamic MIB addition Date: Mon, 20 May 2002 17:32:39 +0200 From: "Gleb Feigstein" Message-ID: <3ce90924$1@news.barak.net.il> References: <3ce8f84f@news.barak.net.il> Please ignore the question - the procedure works fine as described Thanks Gleb Feigstein wrote in message news:3ce8f84f@news.barak.net.il... > Hi > I'm trying to add a private MIB as discribed in the manual. The MIB compiler > with option -mib.c generates a C file that containes all the nodes and also > other files as required. Now I need to add the module to the running MIB by > using snmpdTreeAdd. The first parameter is the path - 1.3.6.1.4.1.2393 in my > case. The second is a pointer to a tree node. What should it be? In the > generated file there is only one non-static, but it is the root of the > complete MIB tree. I tried to make the first private node global and give it > as parameter but it doesn't work proprly. Please assist. > > Thanks > Gleb > > --------------------------- Newsgroups: comp.os.vxworks Subject: using zlib to decompress images created by deflate Date: Mon, 20 May 2002 16:00:34 GMT From: "tim michals" Organization: AT&T Broadband Message-ID: All, Yes, another question about deflate! Ok, I downloaded the latest zlib, create a small file using deflate < infile > fileZ.bin Then here is a snippet of code to uncompress the image: if ((infile = fopen ("fileZ.bin", "rb")) == NULL) { puts("error"); return; } fseek (infile, 0, 0); comprLen =0; fread (&value, 1, 1, infile); /* skiping over the vxworks added byte */ while ((count = fread (&value, 1, 1, infile)) == 1) { compr[comprLen] = value; comprLen++; } fclose(infile); err = uncompress(uncompr, &uncomprLen, compr, comprLen); CHECK_ERR(err, "uncompress"); But an error returns "bad uncompress' but yet all of the file is decompressed into the buffer. So it is working somewhat. What could I be doing wrong? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why doesn't kill(TID, 9) work? Date: Mon, 20 May 2002 11:53:29 -0700 From: "Mark Galecki" Message-ID: <3ce9464c$0$725$4c41069e@reader0.ash.ops.us.uu.net> References: <3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205162353.47142641@posting.google.com> <3ce56d7b$0$12564$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205171813.371c5cfa@posting.google.com> John, thank you again for informative response. I have to say that I made a "typo" in my last message - I meant to say "SIGKILL" which is defined in signal.h as 9, not "SIGTERM". So, the relevant part of my message should be: > > VxWorks Programmer's > > manual says right at the beginning that "nearly all POSIX interfaces (...) > > are available, including (...) queued signals". And, according to POSIX > > manual, SIGKILL is a termination signal that cannot be ignored or caught. > > And you are telling me that signals are ignored unless they are caught by a > > custom handler. Now, I am sorry I am such a pain in the ass, but I want to learn... So, I have one more question on your last response: > SIGTERM is part of process control, not the signals interface per-se. So, what exactly is in the POSIX "signals interface"? (from the POSIX Programmer's Guide by Lewine I can't really deduce another answer, other than "all the signals, including SIGKILL and SIGTERM", but this contradicts your assertion). Thank you, Mark --------------------------- Newsgroups: comp.os.vxworks Subject: Adding a new build using Tornado Date: 20 May 2002 14:15:17 -0700 From: ortizv@acm.org (Victor Ortiz) Organization: http://groups.google.com/ Message-ID: <28e8af95.0205201315.3494473c@posting.google.com> I am using Tornado 2.0.2 for C++ with the GNU toolchain and with a BSP for an MPC8260-based board from Microsys. The BSP is from the board manufacturer. And, I have the "full" simulator which is apparently a BSP also. My host environment is NT4.0sp6. Currently, I have a test application which runs some test cases on OS objects (e.g. tasks, semaphores, sockets). I am trying to have both a simulator build and target build. Initially, I setup my workspace and projects (bootable image and downloaded app) using the simulator toolchain. But cannot create a new build for the bootable image using a different toolchain. My options are either "Build in Current Project" or "Default for toolchain." The "Default for toolchain" is the one I think I want to use but the drop-down list is disabled and cannt select anything other than SIMNTgnu. The other possibility is creating another bootable image project. But this causes me to have to update the VxWorks configuration in both projects. It appears that this is not a problem when creating another build for a downloaded application. Any suggestions or explanations? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why doesn't kill(TID, 9) work? Date: 20 May 2002 17:19:44 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205201619.78ac4dc@posting.google.com> References: <3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205162353.47142641@posting.google.com> <3ce56d7b$0$12564$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205171813.371c5cfa@posting.google.com> <3ce9464c$0$725$4c41069e@reader0.ash.ops.us.uu.net> Hello, > John, thank you again for informative response. I have to say that I made > a "typo" in my last message - I meant to say "SIGKILL" which is defined in > signal.h as 9, not "SIGTERM". So, the relevant part of my message should > be: I never remember which way around they all are anyway, so I assumed you were talking about the ones that on unix cannot be ignored. > > SIGTERM is part of process control, not the signals interface per-se. > So, what exactly is in the POSIX "signals interface"? (from the POSIX > Programmer's Guide by Lewine I can't really deduce another answer, other > than "all the signals, including SIGKILL and SIGTERM", but this contradicts > your assertion). That's a good question isn't it. I don't think that there is a single answer though. The interface is definitely the header file(s), function(s) and the constants (so a system that has the POSIX signals interface must have signals.h, define functions like signal(), kill() and sigaction(), and also define the constants like SIGSTOP, SIGKILL). After that, things start to become a little unclear. For example, the standard does state that the "system shall not allow a process to catch the signals SIGKILL and SIGSTOP." Does that make it part of the interface? If so, there is a problem: VxWorks doesn't have processes, so what does this mean in the context of a VxWorks system? Since VxWorks does not have any concept of a process, and does not use signals as a mechanism for native task control, disallowing a task from catching the signal would mean it could never be used. So, instead, it is left available for you to catch, like any other signal, and therefore allows you to emulate the process termination behaviour of a Unix system. Really it is just a pragmatic approach to an area of the POSIX standard that was written very much with the pre-existing Unix behaviour in mind. The word process appears in pretty much every paragraph, and the signal names came straight from unix. All of that said, it is a very clean way of notifying a task to shut down asynchronously. If this is the kind of behaviour you have been using elsewhere, I would recommend taking the time to write a simple handler and registering it in the startup code for each of your tasks. If you are OK sticking with taskDelete(), you may also want to familiarise yourself with taskSafe() & taskUnsafe() in case there are regions of your code that you should not be deleted in (also, take a look at the delete safety on the mutex semaphores to avoid being deleted while holding a mutex). And remember that taskDelete() is synchronous (i.e. it will perform the deletion from the context of the calling task, and it will block if the task being deleted is in a 'safe' region). HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: what are the BSD driver and END driver? Date: Mon, 20 May 2002 18:23:29 -0700 From: "Hwa Jin Bae" Organization: PSO Systems Inc Message-ID: References: <7384051c.0204232355.4a4b7098@posting.google.com> You can have that problem in both END and BSD. However, it is more prevalent in END for various reasons. You can have that sort of problem in any driver architecture. I have written extensively about this in the past. Google should have some pointers for older articles. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routing table Date: Mon, 20 May 2002 18:46:08 -0700 From: "Parangodath, Binish [SC7:326:EXCH]" Organization: Nortel Networks Message-ID: References: <68c7b7f1.0205140554.7fbca9c1@posting.google.com> Hi, I tried using m2IpRouteTblEntryGet() with M2_NEXT_VALUE hoping to get longest possible match for a destination I specify.But I am not getting consistent results.It picks up the route to my i/f adress IP subnet correctly.But if I add a route to a different subnet pointing to a gateway this function returns no Match. Regards, Binish --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: 20 May 2002 20:10:34 -0700 From: zhoul@blrcsv.china.bell-labs.com (zlybupt) Organization: http://groups.google.com/ Message-ID: <989cdae4.0205201910.183e7ba6@posting.google.com> References: <989cdae4.0205172212.5bd0a755@posting.google.com> <3CE64B34.2A3505CC@bellsouth.net> <989cdae4.0205190506.3e222327@posting.google.com> <3CE8D1AD.3ADCFE02@bellsouth.net> <3CE8D9D1.7020706@spamm.me.l8s.co.uk> David Laight wrote in message news:<3CE8D9D1.7020706@spamm.me.l8s.co.uk>... > >>> > >>>>hi all: > >>>> I am developing a network card driver for vxworks,but now met > >>>>another problem,I wish someone can give me an explanation. > >>>> In my driver design, now I have successfully attached the interrupt > >>>>handle to my network card interrupt because when I remove the cable > >>>>from the network card,it can succeffully detect this interrupt,but I > >>>>wonder why it cannot detect the rxComplete interrupt when I send a > >>>>packet to it?? why? where maybe the problem exist?(Here I am sure that > >>>>the packet sent by sniffer have been successfully sent.) > >>>> Thank you in advance for your answers! > > Have you tried reading the status registers by hand to see if they > indicate that a packet has arrived? > > If the card is writing to a memory area (rather than requiring the > cpu to read the frame out of an internal fifo) look at the memory > area to see what it contains. But I think it should firstly generate an interrupt to indicate the receive of the packet!But it doesnot!Why?Where maybe the problem exist? > > David --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: What's definition of a BSP Date: 20 May 2002 20:29:20 -0700 From: ryeung@earthlink.net (Raymond Yeung) Organization: http://groups.google.com/ Message-ID: References: Thanks Jack and Leonid for sharing your insights. I wonder if this topic should really be shelved onto FAQ list. It's the second approach Leonid that I was thinking initially. To this end, one would include practically all h/w dependent parts (thus all drivers) as part of the BSP. If you change the board, and assuming those hardware dependent features should be ported (even if it includes drivers for proprietary ASIC), you need to change your *extended* BSP. Things like Watchdog timer (even though it's optional) would also be part of BSP. Same goes with POST (including memory tests, or any other device tests, or at least the h/w dependent portion) and exception handling. Am I off base? "An alternative approach would assume that your application will not be running on this particular board forever, and there would be a need to port the application to different boards, then you should define an extended set of I/O interfaces for this type of application, clearly define an API for drivers and implement them as part of the BSP, and then, when needs come, one would enable your application on a new platform simply by implementing the extended BSP. A digital camera would be a good example where an extended BSP should be conisdered, if you are writing a digital camera application that is expected to be used in numerous camera models." --------------------------- Newsgroups: comp.os.vxworks Subject: A question!! Date: 20 May 2002 23:50:07 -0700 From: zhoul@blrcsv.china.bell-labs.com (zlybupt) Organization: http://groups.google.com/ Message-ID: <989cdae4.0205202250.29aa490d@posting.google.com> hi all: I wonder why my interrupt handler function in my network driver can only detect the interrupt of removing the cable from the network card while cannot detect the interrupt of rejoining the cable into the network card! I think they belong to the same interrupt class? Isn't it generate an interrupt when I join the cable into the network card????? Thank you for your answers in advance! --------------------------- Newsgroups: comp.os.vxworks Subject: Bridge Date: Tue, 21 May 2002 09:05:11 +0200 From: Ferry =?iso-8859-1?Q?Dji=E9ya?= Organization: Wanadoo, l'internet avec France Telecom Message-ID: <3CE9F1A7.B61FF095@avilinks.com> Hi, I want to build a bridge between Ethernet and LLC. The has and Ethernet END driver and LLC END driver, can I use MUX facilities to build the brigde uppon those drivers ? Thanks Ferry --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: 21 May 2002 07:40:28 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <989cdae4.0205172212.5bd0a755@posting.google.com> <3CE64B34.2A3505CC@bellsouth.net> <989cdae4.0205190506.3e222327@posting.google.com> <3CE8D1AD.3ADCFE02@bellsouth.net> <3CE8D9D1.7020706@spamm.me.l8s.co.uk> <989cdae4.0205201910.183e7ba6@posting.google.com> zhoul@blrcsv.china.bell-labs.com (zlybupt) wrote: > But I think it should firstly generate an interrupt to indicate the > receive of the packet!But it doesnot!Why?Where maybe the problem > exist? I think you are looking at it from the wrong side. Are you sure your ethernet card is generating the interrupt (did you check with a scope or other HW debugging device). If you are sure that the interrupt is generated you can look at your interrupt service routine again. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.arch.embedded,comp.os.vxworks Subject: Re: What's definition of a BSP Date: Tue, 21 May 2002 07:51:20 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Since you have already taken the "extended BSP approach", then including POST and things like watchdogs in the BSP make sense. Actually, POST would make sense even if the "basic BSP" approach would have been used. Everyone needs to test RAM on cold startup, so WIndRiver could save a lot of people work by bundling POST functions as part of their standard BSPs. Keep in mind that embedded applications deliver hardware and software as one complete solution, so the APIs you define for custom hardware need to be elastic enough to include new hardware features easily, and in backward compatible manner. - - Leonid "Raymond Yeung" wrote in message news:d83bb8e3.0205201929.7b641486@posting.google.com... > Thanks Jack and Leonid for sharing your insights. I wonder > if this topic should really be shelved onto FAQ list. > > It's the second approach Leonid that I was thinking initially. > To this end, one would include practically all h/w dependent > parts (thus all drivers) as part of the BSP. If you change > the board, and assuming those hardware dependent features should > be ported (even if it includes drivers for proprietary ASIC), > you need to change your *extended* BSP. > > Things like Watchdog timer (even though it's optional) would > also be part of BSP. Same goes with POST (including memory > tests, or any other device tests, or at least the h/w dependent > portion) and exception handling. > > Am I off base? > > "An alternative approach would assume that your application will > not be running on this particular board forever, and there would > be a need to port the application to different boards, then you > should define an extended set of I/O interfaces for this type > of application, clearly define an API for drivers and implement > them as part of the BSP, and then, when needs come, one > would enable your application on a new platform simply by implementing > the extended BSP. A digital camera would be a good example where > an extended BSP should be conisdered, if you are writing a > digital camera application that is expected to be used in numerous > camera models." --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Bridge Date: 21 May 2002 07:47:58 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: <3CE9F1A7.B61FF095@avilinks.com> Ferry Djiéya wrote: > Hi, > I want to build a bridge between Ethernet and LLC. > The has and Ethernet END driver and LLC END driver, can I use MUX > facilities to build the brigde uppon those drivers ? Yes, it is possible to build a bridge on top of the MUX. WindRiver has a product for this, it is the bridge code that they have as part of the THG release. But you can also build it yourself. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: routine called from kernel/application Date: Tue, 21 May 2002 10:30:39 +0200 From: Sebastien Fauris Organization: Nortel Networks Message-ID: <3CEA05AF.61DE7E42@europem01.nt.com> Hi vxWorkers, having a program counter value, I would like to identify if the current routine is being executed from a kernel or from an applicative context (this in order to perform profiling of my code). Is there a way to get the beginning and ending addresses of the kernel part of the vxWorks.st and the beginning and ending addresses of the applicative part of it ? Thanks for your help, Sebastien. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: A question!! Date: Tue, 21 May 2002 06:13:31 -0400 From: joe durusau Organization: Bellsouth.Net Message-ID: <3CEA1DCB.DDAC7FE3@bellsouth.net> References: <989cdae4.0205202250.29aa490d@posting.google.com> zlybupt wrote: > hi all: > I wonder why my interrupt handler function in my network driver can > only detect the interrupt of removing the cable from the network card > while cannot detect the interrupt of rejoining the cable into the > network card! > I think they belong to the same interrupt class? > Isn't it generate an interrupt when I join the cable into the > network card????? > Thank you for your answers in advance! As many have said to you earlier, The reason the ISR does not trigger is that the hardware is not sending an interrupt. Since nobody here knows what card you are using, you first need to read the hardware manual for the device with great care! Then read the registers on the device manually and see whether it has detected incoming data. Then check the polrity and voltage levels of incoming signals into the card. I would bet that you are not programming the card to send an int on packet received, but I am not familiar with your code and non one here is going to do that level of work for you. Speaking only for myself, Joe Durusau --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 22 03:12:39 2002 From: Vishwanath Sangati Date: Wed May 22 03:12:42 PDT 2002 Subject: Hi 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_01C20177.978D48B0 Content-Type: text/plain; charset="iso-8859-1" Hello Guys, I am working on the UNIX based VxWorks for a project on network printers. I need to prepare a detailed document for the user defined libraires, now the situation is that i need to figure out how to get the list of libraries embedded in the actual code.I mean I dont want to open the .c files and scrawl rather I want to find from the shell or anyother resource. So guys please tip me on the same. Another issue what is bogging me is that I need to check the stack usage for the Embedded code, means i need to run the target with the maximum load so that indivisual task is run at the optimal point. Could you guys throw some light on these issues ASAP. Thanks.. Vishwanath ------_=_NextPart_001_01C20177.978D48B0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi

Hello Guys,
I am working on the UNIX based = VxWorks for a project on network printers.
I need to prepare a detailed document = for the user defined libraires, now the situation is that i need to = figure out how to get the list of

libraries embedded in the actual = code.I mean I  dont want to open the .c files and scrawl rather I = want to find from the shell or anyother

resource.
So guys please tip me on the = same.
Another issue what is bogging me is = that I need to check the stack usage for the Embedded code, means i = need to run the target with the maximum load so that indivisual task is = run at the optimal point.

Could you guys throw some light on = these issues ASAP.
Thanks..
Vishwanath


------_=_NextPart_001_01C20177.978D48B0-- From vxwexplo-errs@csg.lbl.gov Wed May 22 04:03:17 2002 From: Vxworks Exploder Date: Wed May 22 04:03:20 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed May 22 04:03:13 PDT 2002 Subject: IDE controller Subject: getting started with vxworks? How to? Subject: Re: getting started with vxworks? How to? Subject: RPC: how to force svc_run() to return ? / svc_getreq() usage ? Subject: Re: Adding a new build using Tornado Subject: Re: IDE controller Subject: Introduction Subject: Correct method to change IP addr? Subject: Re: Introduction Subject: Re: routine called from kernel/application Subject: Re: using zlib to decompress images created by deflate Subject: Re: why doesn't kill(TID, 9) work? Subject: Re: Introduction Subject: TFFS & Intel Strataflash. tffsDevFormat returns ERROR Subject: Where can I download VxWorks? Subject: get that MAC...not! Subject: Re: using zlib to decompress images created by deflate Subject: VxSim-lite & sockets? Subject: Re: Send Multicast message Subject: Re: representation of a double in vxWorks Subject: Re: Where can I download VxWorks? Subject: Re: Where can I download VxWorks? Subject: Re: get that MAC...not! Subject: Re: why doesn't kill(TID, 9) work? Subject: Re: TFFS & Intel Strataflash. tffsDevFormat returns ERROR Subject: Compilation problem with sys/time.h Subject: Re: Compilation problem with sys/time.h Subject: Could I spawn a task when booting up? Subject: Re: Could I spawn a task when booting up? Subject: Where to download the patch for tornado? Subject: Re: Compilation problem with sys/time.h Subject: Re: Where to download the patch for tornado? Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Subject: Re: How to set TIMEZONE? Support for timezones in localtime? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: IDE controller Date: 21 May 2002 04:18:52 -0700 From: be_gyus@yahoo.com (james) Organization: http://groups.google.com/ Message-ID: what is ide controller and its significance ? --------------------------- Newsgroups: comp.os.vxworks Subject: getting started with vxworks? How to? Date: Tue, 21 May 2002 13:34:12 +0200 From: "Wojciech Mirski" Organization: Siemens Inc. Message-ID: Salute Fellows! I am new in this field, but want to deepen knowledge about VxWorks. I would like to know, if there is any book concerning programming under vxworks for x86? Additionally I'm interested in "getting started" or something like this tutorial, and any info about making device drivers. All help will be very appreciated. Best regards Wojciech Mirski --------------------------- Newsgroups: comp.os.vxworks Subject: Re: getting started with vxworks? How to? Date: 21 May 2002 11:52:35 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "Wojciech Mirski" wrote: > I am new in this field, but want to deepen knowledge about VxWorks. > I would like to know, if there is any book concerning programming under > vxworks for x86? > Additionally I'm interested in "getting started" or something like this > tutorial, and any info about making device drivers. The best way to start is to use Tornado Prototyper: a free VxWorks simulator. This includes a Getting Started and the complete manuals for VxWorks. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: RPC: how to force svc_run() to return ? / svc_getreq() usage ? Date: 21 May 2002 05:38:37 -0700 From: r.stelzer@stoye.de (Rainer Stelzer) Organization: http://groups.google.com/ Message-ID: <78a5d372.0205210438.750b67c3@posting.google.com> Hi vxworkers, although the vxworks RPC library documentation is quite poor I successfully set up a working client server connection between a vxworks MVME2304 and a custom PPC860 target in tcp-mode. When testing the connection in different stress situations I faced the problem that, case I simply reboot the client side while a transfer takes place, the server didn't notify this state. I would expect the server side to time out and throuw an exception, but nothing happens. Case the client tries to reconnect in this situation it exits with "server error- error 60". Only restarting the server task can clear this situation. My questions: - - Any idea where the server hangs and how to force svc_run to return in this situation? - - do I have to set timeout parameters in works like I do on Unix machines ? Or are they special in any way ? - - Has anybody out there a working vxworks svc_getreq() example ? Thanks in advance. rgds Rainer vxworks Tornado --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Adding a new build using Tornado Date: Tue, 21 May 2002 12:58:26 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: References: <28e8af95.0205201315.3494473c@posting.google.com> I needed to build a new bsp for work as well ... What i ended up doing is copying the 8260 bsp dir in tornado and creating a new folder called "my8260bsp" ... ex C:\Tornado\target\config\ads8260 C:\Tornado\target\config\jims8260 The tornado UI picks it up when you go to create a new "bootable app" ... Not sure this is what you were looking for but I hope it helps ... Jim "Victor Ortiz" wrote in message news:28e8af95.0205201315.3494473c@posting.google.com... > I am using Tornado 2.0.2 for C++ with the GNU toolchain and with a BSP > for an MPC8260-based board from Microsys. The BSP is from the board > manufacturer. And, I have the "full" simulator which is apparently a > BSP also. My host environment is NT4.0sp6. > > Currently, I have a test application which runs some test cases on OS > objects (e.g. tasks, semaphores, sockets). I am trying to have both a > simulator build and target build. Initially, I setup my workspace and > projects (bootable image and downloaded app) using the simulator > toolchain. But cannot create a new build for the bootable image using > a different toolchain. > > My options are either "Build in Current Project" or "Default for > toolchain." The "Default for toolchain" is the one I think I want to > use but the drop-down list is disabled and cannt select anything other > than SIMNTgnu. > > The other possibility is creating another bootable image project. But > this causes me to have to update the VxWorks configuration in both > projects. It appears that this is not a problem when creating another > build for a downloaded application. > > Any suggestions or explanations? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IDE controller Date: Tue, 21 May 2002 13:02:17 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: IDE is a standard interface for connecting storage devices. It has historically evolved around the PC architecture, and is essentially a 16-bit wide parallel bus used for connecting hard disks (ATA command set) and CD-ROM, CD-RW, DVD and also various tape drives (ATAPI command set). Each IDE controller typically has two chanells (each channel has a separate cable) which can support up to two drives, one in "master" mode, one in "slave" mode. If you need more info, you have got to be more specific. "james" wrote in message news:ff42611f.0205210318.16b407f6@posting.google.com... > what is ide controller and its significance ? --------------------------- Newsgroups: comp.os.vxworks Subject: Introduction Date: 21 May 2002 07:19:22 -0700 From: sanjeevb_4u@yahoo.com (sanjeev bhattt) Organization: http://groups.google.com/ Message-ID: Hi, Any pointers to know abt Vxworks Architecture thanx in advance, Sanjeev --------------------------- Newsgroups: comp.os.vxworks Subject: Correct method to change IP addr? Date: Tue, 21 May 2002 15:04:04 +0000 (UTC) From: "Carl Petersen" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: Is this the correct sequence to change the IP addr of an interface? ifRouteDelete(...); ifMaskSet(...); ifAddrSet( ifName, "New IP" ); These calls run without error however a subsequent call to bind() fails. I have also tried ipDetach() then ipAttach() without success. Note that the interface is operational before I attempt to modify the IP addr. Tornado 2.02 Pentium BSP Thanks Carl Petersen - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Introduction Date: Tue, 21 May 2002 15:45:15 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: If you need to use VxWorks, the best place to begin is the VxWorks Programmers Manual. Years ago we used to hand them out to potential customers. Don't know if there is, buit should be a downloadable PDF version. If you are interested in the VxWorks internals, then I am afraid you are out of luck. There is no known public document which describes the internal architecture of the Wind kernel, but of course there are people who know it intimately, and you could hire one to sched some light into it, if their confidentiality with WIndRiver has expired. "sanjeev bhattt" wrote in message news:c7a77add.0205210619.2cf9401b@posting.google.com... > Hi, > Any pointers to know abt Vxworks Architecture > > thanx in advance, > Sanjeev --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routine called from kernel/application Date: 21 May 2002 09:27:29 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205210827.44ccf31e@posting.google.com> References: <3CEA05AF.61DE7E42@europem01.nt.com> Hello, > having a program counter value, I would like to identify if the > current routine is being executed from a kernel or from an > applicative context (this in order to perform profiling of my code). > Is there a way to get the beginning and ending addresses of > the kernel part of the vxWorks.st and the beginning and ending > addresses of the applicative part of it ? Not if you've linked your application code with the kernel - there is no guarantee that they will be separated if you've linked them together. Instead, what you need to do is use the symbol table (if you're trying to do this on the target), or write a script that builds such a list from the image file (if doing this on your host machine). Take a look at symByValueFind() for looking up nearest matching symbols from an address on the target. On your host I'll leave you to work out how to get a nearest match from the output of the appropriate nm tool for your architecture. HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: using zlib to decompress images created by deflate Date: 21 May 2002 13:02:29 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: References: Sender: bpringlemeir@DeadDuck >>>>> "tim" == tim michals writes: tim> All, Yes, another question about deflate! Ok, I downloaded the tim> latest zlib, create a small file using deflate < infile > tim> fileZ.bin Then here is a snippet of code to uncompress the tim> image: [snip code] tim> But an error returns "bad uncompress' but yet all of the file is tim> decompressed into the buffer. So it is working somewhat. What tim> could I be doing wrong? The deflate program might not work with the version of zlib you retrieved (and specifically the functions you call). It is common for archives to contain header information. The deflate might have a dictionary included, etc. I suggest that you build minigzip (or whatever) and use that program to compress the image on your host. The code you supplied should also be able to decompress it on the host. It is usually easier to debug the compress/decompress on the host first and then move to your target. I have used the gzio [I think] that gives a stream interface to files. I used this in a `COFF' loader to allow compresses .out files to be loaded. You probably want a different interface for boot roms. hth, Bill Pringlemeir. - -- Have you ever tasted spinach and thought it was Doritos? Or had your dog make you feel warm and fuzzy while being chopped to pieces by a rotor blade? Or thought it was unhealthy to have two modems to defend against the artist? You will. And the company that will bring it to you: AT&T. vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why doesn't kill(TID, 9) work? Date: Tue, 21 May 2002 10:51:23 -0700 From: "Mark Galecki" Message-ID: <3cea8940$0$6770$4c41069e@reader0.ash.ops.us.uu.net> References: <3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205162353.47142641@posting.google.com> <3ce56d7b$0$12564$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205171813.371c5cfa@posting.google.com> <3ce9464c$0$725$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205201619.78ac4dc@posting.google.com> hello, > signals as a mechanism for native task control, disallowing a task > from catching the signal would mean it could never be used. So, Hmm, I don't understand: I could still use the "i" command, see the task's ID, and use kill(, 9) and if signal 9 did not allow catching, then this could just kill that task. So signal 9 would be used. Am I missing something?? > If you are OK sticking with taskDelete(), you may also want to I am writing and registering signal handlers, and using taskDelete(), taskSafe() and taskUnsafe(). I can delete what I want with taskDelete(). I just wanted to learn more about POSIX and VxWorks and how signal 9 works. Thank you very much Mark... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Introduction Date: 21 May 2002 11:33:42 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205211033.54761c27@posting.google.com> References: Hello, It concerns me that you are unable to think of this for yourself, but try Wind River's web site: http://www.windriver.com/ for information about their products, including VxWorks. You can also download the free prototyper from the web site which should help you get a hands-on feel for the system once you get past the reading phase of your investigation. HTH, John... sanjeevb_4u@yahoo.com (sanjeev bhattt) wrote in message news:... > Hi, > Any pointers to know abt Vxworks Architecture > > thanx in advance, > Sanjeev --------------------------- Newsgroups: comp.os.vxworks Subject: TFFS & Intel Strataflash. tffsDevFormat returns ERROR Date: 21 May 2002 12:02:47 -0700 From: medwards@visualnetworks.com (Mark Edwards) Organization: http://groups.google.com/ Message-ID: <9d16ae8d.0205211102.1c0045b8@posting.google.com> I've located several posts on this issue but can't find a solution. tffsDevFormat returns error. I'm using the INCLUDE_TL_FTL & INCLUDE_MTD_CFISCS. Our flash is set for 8 bits at 0xFE800000. tffsShow works and shows: 0: socket=RFA: type=0x8917, unitSize=0x20000, mediaSize=0x800000 My rfaSetWindow routine sets the size to 0x760000 because we need that 0xA0000 bytes to access external peripherals. Is this a problem? I'm not including TFFS in the boot loader. Just within the downloadable vxWorks. It's unclear to me at this point how to proceed because the project facility doesn't use the cfiscs.c file in src/drv/tffs. It seems to use the precompiled object file in the lib directory. It seems many people have this configuration and have hit this problem before. What were the solutions? Any help would be much appreciated. Thanx. --------------------------- Newsgroups: comp.os.vxworks Subject: Where can I download VxWorks? Date: Tue, 21 May 2002 12:19:46 -0700 From: Nigel Chamberlain Organization: http://extra.newsguy.com Message-ID: I'm sure there's got to be an FTP site out there somewhere that'll allow me to d/l the sucker anonymously. My company's too cheap to buy it, and all I want to do is - (1) install it in test (2) show the bastards what it can do and (3) demonstrate through (2) that they need to buy the sucker. I've combed the world over for a place to d/l it and could really use a helping hand on this one. No need to respond and say that you need to get it from the vendor. I'm aware, and could use a quick assist! TIA! - -- NC --------------------------- Newsgroups: comp.os.vxworks Subject: get that MAC...not! Date: 21 May 2002 13:24:10 -0700 From: kenb88@hotmail.com (Whacked) Organization: http://groups.google.com/ Message-ID: <6755aee5.0205211224.1fde1d89@posting.google.com> Hi, I have a simple network configuration. My VxWorks box has IP address 192.168.1.1 and is connected DIRECTLY to a router with IP address 192.168.1.250. I then do the following: struct ifnet *myifnet; char buffer[20]; myifnet = ifunit("le0"); etherAddrResolve (myifnet, "192.168.1.250", buffer, 3, 3); the etherAddrResolve always returns with ERROR and errno is set to something like 0x32000001...the cross reference in vwModNum.h links me to hostLib.h which says the errno means unknown host. what am I doing wrong?? This should give me the ethernet address in buffer. thanks, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: using zlib to decompress images created by deflate Date: Tue, 21 May 2002 23:13:24 GMT From: "tim michals" Organization: AT&T Broadband Message-ID: References: Thanks, I'm trying to create the same functionality as a VxWorks loader does to expand compressed images created by deflate. So, I'm trying to create the inflate, using the correct dictionary, etc as inflate and deflate used in VxWorks. "Bill Pringlemeir" wrote in message news:u661hfn8a.fsf@sympatico.ca... > >>>>> "tim" == tim michals writes: > > tim> All, Yes, another question about deflate! Ok, I downloaded the > tim> latest zlib, create a small file using deflate < infile > > tim> fileZ.bin Then here is a snippet of code to uncompress the > tim> image: > > [snip code] > > tim> But an error returns "bad uncompress' but yet all of the file is > tim> decompressed into the buffer. So it is working somewhat. What > tim> could I be doing wrong? > > The deflate program might not work with the version of zlib you > retrieved (and specifically the functions you call). It is common for > archives to contain header information. The deflate might have a > dictionary included, etc. I suggest that you build minigzip (or > whatever) and use that program to compress the image on your host. > The code you supplied should also be able to decompress it on the > host. It is usually easier to debug the compress/decompress on the > host first and then move to your target. > > I have used the gzio [I think] that gives a stream interface to files. > I used this in a `COFF' loader to allow compresses .out files to be > loaded. You probably want a different interface for boot roms. > > hth, > Bill Pringlemeir. > > -- > Have you ever tasted spinach and thought it was Doritos? Or had your > dog make you feel warm and fuzzy while being chopped to pieces by a > rotor blade? Or thought it was unhealthy to have two modems to defend > against the artist? You will. And the company that will bring it to > you: AT&T. > > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: VxSim-lite & sockets? Date: Tue, 21 May 2002 16:18:41 -0700 From: "Deb Briggs" Organization: Dotcast Message-ID: If I want to use VxWorks' sockLib, do I have to purchase the full VxSim? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Send Multicast message Date: Wed, 22 May 2002 09:14:50 +0930 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3CEADBF2.83593C70@dsto.defence.gov.au> References: <58cafc21.0205162257.7ae894fa@posting.google.com> Try using routeAdd(...) to add a route to your routing table for the multicast address - or even better use mRouteAdd. Then any socket that is destined for a multicast address will use the routing table to determine the correct interface to use. supergod wrote: > Now my project need send multicast message to other target. > But i must use setsockopt (sock, IPPROTO_IP, IP_MULTICAST_IF, (char > *)&ifAddr, > sizeof (mCastAddr)) to select a Default Interface for Outgoing > Multicasts before i send it. Otherwise sendto () will return error > 0x41. > > I want to know whethere any api can used to select the interface for > outgoing multicase before i create a socket. if so i need only set > sock option once, and everytime i create a new socket ,it will > automaticlly get the interface to send out . any idea? > > my target is based on pentinum, network card is Intel(R) 82559 . --------------------------- Newsgroups: comp.os.vxworks Subject: Re: representation of a double in vxWorks Date: Wed, 22 May 2002 09:27:54 +0930 From: Tim Shaw Organization: Defence Science and Technology Organisation Message-ID: <3CEADF02.C7BF59D1@dsto.defence.gov.au> References: According to ANSI C, the size of double is implementation dependant, and defined in float.h for an implementation. In types/vxTypesBase.h, you'll find the mantissa and exponent defined. I think you'll find that the PPC supports IEEE754 (52 bits mantissa (+1 hidden bit), 1 sign bit and 11 exponent bits). zvi wrote: > Hello vxWorkers!! > > I hope this is not too foolish question ... > > what is the representation of a double in the memory using vxWorks with ppc board? > > is someone can help me? > > Thanks, > Zvi. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where can I download VxWorks? Date: Tue, 21 May 2002 20:52:12 -0400 From: joe durusau Organization: BELLSOUTH.net Message-ID: <3CEAEBBC.D76D1FF@bellsouth.net> References: Nigel Chamberlain wrote: > I'm sure there's got to be an FTP site out there somewhere that'll > allow me to d/l the sucker anonymously. My company's too cheap to buy > it, and all I want to do is - (1) install it in test (2) show the > bastards what it can do and (3) demonstrate through (2) that they need > to buy the sucker. I've combed the world over for a place to d/l it > and could really use a helping hand on this one. > > No need to respond and say that you need to get it from the vendor. > I'm aware, and could use a quick assist! TIA! > > -- NC www.wrs.com The simulator is free. Nowhere can you download a copy that will work with a target. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where can I download VxWorks? Date: Wed, 22 May 2002 03:31:28 GMT From: nunya@damn.business.com (Bruce) Organization: Prodigy Internet http://www.prodigy.com Message-ID: <3ceb114b.8964029@news.houston.sbcglobal.net> References: In comp.os.vxworks Nigel Chamberlain wrote: >I'm sure there's got to be an FTP site out there somewhere that'll >allow me to d/l the sucker anonymously. My company's too cheap to buy >it, and all I want to do is - (1) install it in test (2) show the >bastards what it can do and (3) demonstrate through (2) that they need >to buy the sucker. I've combed the world over for a place to d/l it >and could really use a helping hand on this one. A sales rep will do this for you and probably have you up and running a lot faster than you could do it yourself. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: get that MAC...not! Date: Wed, 22 May 2002 06:04:31 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <6755aee5.0205211224.1fde1d89@posting.google.com> Firsst, you need to be sure to test the pointer returned by ifunit(), and make sure it is not NULL. Second, you allow ARP three retires, with only 3 clock ticks between retries, that could be not enough for the host to reply, depending on your sysClkRateGet() value. Increase the last parameter to e.g. if ( myifnet != NULL ) etherAddrResolve (myifnet, "192.168.1.250", buffer, 3, sysClkRateGet()/10 ); "Whacked" wrote in message news:6755aee5.0205211224.1fde1d89@posting.google.com... > Hi, > > I have a simple network configuration. My VxWorks box has IP address > 192.168.1.1 and is connected DIRECTLY to a router with IP address > 192.168.1.250. > > I then do the following: > > struct ifnet *myifnet; > char buffer[20]; > > myifnet = ifunit("le0"); > etherAddrResolve (myifnet, "192.168.1.250", buffer, 3, 3); > > > the etherAddrResolve always returns with ERROR and errno is set > to something like 0x32000001...the cross reference in vwModNum.h links > me > to hostLib.h which says the errno means unknown host. > > what am I doing wrong?? This should give me the ethernet address in > buffer. > > thanks, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why doesn't kill(TID, 9) work? Date: 21 May 2002 23:14:29 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205212214.6791034@posting.google.com> References: <3ce422a6$0$6550$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205162353.47142641@posting.google.com> <3ce56d7b$0$12564$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205171813.371c5cfa@posting.google.com> <3ce9464c$0$725$4c41069e@reader0.ash.ops.us.uu.net> <488e459a.0205201619.78ac4dc@posting.google.com> <3cea8940$0$6770$4c41069e@reader0.ash.ops.us.uu.net> Hello, > > signals as a mechanism for native task control, disallowing a task > > from catching the signal would mean it could never be used. So, > Hmm, I don't understand: I could still use the "i" command, see the task's > ID, and use > kill(, 9) > and if signal 9 did not allow catching, then this could just kill that task. > So signal 9 would be used. Am I missing something?? In a way; that would mean that signals were being used for task control, as opposed to being provided as a mechanism purely for compatibility. You can remove signals support from VxWorks totally (assuming you remove any other features that depend on them) and the OS will work perfectly well. You could argue that when the signals support is included, signals could be used to kill tasks as you describe, but that's not how it is today. I wonder how many people would be interested in this level of support as an enhancement? It shouldn't be too difficult to add... Rgds, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: TFFS & Intel Strataflash. tffsDevFormat returns ERROR Date: 21 May 2002 23:18:16 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205212218.27c8dda2@posting.google.com> References: <9d16ae8d.0205211102.1c0045b8@posting.google.com> Hello, > I'm not including TFFS in the boot loader. Just within the > downloadable vxWorks. It's unclear to me at this point how to proceed > because the project facility doesn't use the cfiscs.c file in > src/drv/tffs. It seems to use the precompiled object file in the lib > directory. You can solve this in a couple of ways: 1) Type make CPU=XXX in target/src/drv/tffs (replace XXX with the appropriate CPU type for the board you're using, e.g. PPC604 or ARMARCH4) 2) You can build it as a .o file somewhere else, and simply add it to your project. If you do this, the one in the archive file will not be used. Sorry I don't have info about your TFFS configuration issue, but perhaps being able to add debug to the cfiscs.c file will help you out. Rgds, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Compilation problem with sys/time.h Date: Wed, 22 May 2002 09:24:50 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: ========================================================= Tornado : Version 2.0.2 VxWorks : Version 5.4.2 BSP : Version 1.1/1 Compiler : ccsimpc gcc version egcs-2.90.29 tornado 2.0 980515 (egcs-1.0.3 release) Host : Microsoft Windows 2000 [Version 5.00.2195] Target : VxSim for Windows (VxWorks target simulator) ToolChain : SIMNTgnu ========================================================= Project Type : bootable VxWorks image (custom configured) ========================================================= I have got a compilation problem with a short program that holds the line #include We can see the strange dependency for mkt2.o in Makefile generated by the project manager - ------------------------------------ mkt2.o: D:/sys2/own_src/01/mkt2.cpp \ D:sys2Tornadohost†-win32libgcc-libi386-pc-mingw32egcs-2.90.29../../../../i386-pc-mingw32/sys-include/sys/time.h - ------------------------------------ Two strangenesses : 1. There are only two 'sys/time.h' files under (in my system = D:/sys2/Tornado) : * D:/sys2/Tornado/host/x86-win32/i386-pc-mingw32/sys-include/sys/time.h * D:/sys2/Tornado/host/include/win32/sys/time.h Why does 'D:sys2Tornadohost†-win32libgcc ... ' appear here ? 2. The 'D:sys2Tornadohost†-win32libgcc ... ' has invalid construction. Nust be : * D:sys2/Tornado/host/ ... What is wrong? Relevant information is below. Thanks. || ######### C++ Code : BEGIN ######### // File mkt2.cpp #include int get2 () { return 0; } || ######### C++ Code : END ########### || ######### Compilation Log File : BEGIN ######### vxrm *.o *.rpo ctdt.c symTbl.c vxWorks* vxrm D:\sys2\Tornado\target\proj\sim\dir-01\prjComps.h D:\sys2\Tornado\target\proj\sim\dir - -01\prjParams.h D:\sys2\Tornado\target\proj\sim\dir-01\prjConfig.c D:\sys2\Tornado\target\ proj\sim\dir-01\linkSyms.c vxrm D:\sys2\Tornado\target\proj\sim\dir-01\libs.nm D:\sys2\Tornado\target\proj\sim\dir-01 \libs.size wtxtcl D:\sys2\Tornado\host\src\hutils\configGen.tcl D:\sys2\Tornado\target\proj\sim\dir-0 1\SIM-Project-01.wpj ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop -Wall -DRW_M ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\Tornado\target \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\comps\src -ID:\sy s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\Tornado\target\config\simpc\ sysLib.c ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop -Wall -DRW_M ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\Tornado\target \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\comps\src -ID:\sy s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\Tornado\target\proj\sim\dir- 01\prjConfig.c ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop -Wall -DRW_M ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\Tornado\target \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\comps\src -ID:\sy s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\Tornado\target\proj\sim\dir- 01\linkSyms.c ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop -Wall -DRW_M ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\Tornado\target \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\comps\src -ID:\sy s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\own_src\01\s2.c ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop -Wall -DRW_M ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\Tornado\target \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\comps\src -ID:\sy s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\own_src\01\xxx1.cpp ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop -Wall -DRW_M ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\Tornado\target \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\comps\src -ID:\sy s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\s1.cpp make: *** No rule to make target `D:sys2Tornadohost†-win32libgcc-libi386-pc-mingw32egcs-2. 90.29..\..\..\..\i386-pc-mingw32\sys-include\sys\time.h', needed by `mkt2.o'. Stop. Done. || ######### Compilation Log File : END ########### || ######### Makefile : BEGIN ######### # Makefile generated by the project manager # # GENERATED: Wed May 22 07:23:52 djg 2002 # DO NOT EDIT - file is regenerated whenever the project changes ## core information PRJ_DIR = $(WIND_BASE)/target/proj/sim/dir-01 PRJ_FILE = SIM-Project-01.wpj PRJ_TYPE = vxWorks PRJ_OBJS = sysLib.o prjConfig.o linkSyms.o s2.o xxx1.o s1.o mkt2.o BOOT_OBJS = BUILD_SPEC = default BSP_DIR = $(WIND_BASE)/target/config/simpc TGT_DIR = $(WIND_BASE)/target ## set searching directories for dependencies vpath %.c $(BSP_DIR) vpath %.cpp $(BSP_DIR) vpath %.cxx $(BSP_DIR) ## build-configuration info ifeq ($(BUILD_SPEC),default) CPU = SIMNT TOOL = gnu DEFAULT_RULE = vxWorks endif ## component-configuration info COMPONENTS = INCLUDE_ANSI_ASSERT \ INCLUDE_ANSI_CTYPE \ INCLUDE_ANSI_LOCALE \ INCLUDE_ANSI_MATH \ INCLUDE_ANSI_STDIO \ INCLUDE_ANSI_STDIO_EXTRA \ INCLUDE_ANSI_STDLIB \ INCLUDE_ANSI_STRING \ INCLUDE_ANSI_TIME \ INCLUDE_BOOT_LINE_INIT \ INCLUDE_BUF_MGR \ INCLUDE_CACHE_ENABLE \ INCLUDE_CACHE_SUPPORT \ INCLUDE_CPLUS \ INCLUDE_CPLUS_DEMANGLER \ INCLUDE_CPLUS_IOSTREAMS \ INCLUDE_CPLUS_IOSTREAMS_FULL \ INCLUDE_CPLUS_IOSTREAMS_TEST \ INCLUDE_CPLUS_STL \ INCLUDE_CPLUS_STRING \ INCLUDE_CPLUS_STRING_IO \ INCLUDE_DEBUG \ INCLUDE_DLL \ INCLUDE_DOSFS \ INCLUDE_ENV_VARS \ INCLUDE_EXC_HANDLING \ INCLUDE_EXC_TASK \ INCLUDE_FLOATING_POINT \ INCLUDE_FORMATTED_IO \ INCLUDE_HASH \ INCLUDE_HW_FP_SHOW \ INCLUDE_IO_SYSTEM \ INCLUDE_KERNEL \ INCLUDE_LOADER \ INCLUDE_LOGGING \ INCLUDE_MEMORY_CONFIG \ INCLUDE_MEM_MGR_BASIC \ INCLUDE_MEM_MGR_FULL \ INCLUDE_MODULE_MANAGER \ INCLUDE_MSG_Q \ INCLUDE_MSG_Q_SHOW \ INCLUDE_NTPASSFS \ INCLUDE_PIPES \ INCLUDE_POSIX_CLOCKS \ INCLUDE_RBUFF \ INCLUDE_RNG_BUF \ INCLUDE_SELECT \ INCLUDE_SEM_BINARY \ INCLUDE_SEM_COUNTING \ INCLUDE_SEM_MUTEX \ INCLUDE_SEQ_TIMESTAMP \ INCLUDE_SHELL \ INCLUDE_SIGNALS \ INCLUDE_SIO \ INCLUDE_STANDALONE_SYM_TBL \ INCLUDE_STARTUP_SCRIPT \ INCLUDE_STAT_SYM_TBL \ INCLUDE_STDIO \ INCLUDE_SYM_TBL \ INCLUDE_SYM_TBL_INIT \ INCLUDE_SYSCLK_INIT \ INCLUDE_SYSHW_INIT \ INCLUDE_SYS_START \ INCLUDE_TASK_HOOKS \ INCLUDE_TASK_SHOW \ INCLUDE_TASK_VARS \ INCLUDE_TIMEX \ INCLUDE_TRIGGERING \ INCLUDE_TTY_DEV \ INCLUDE_USER_APPL \ INCLUDE_WATCHDOGS \ INCLUDE_WDB \ INCLUDE_WDB_BANNER \ INCLUDE_WDB_BP \ INCLUDE_WDB_COMM_PIPE \ INCLUDE_WDB_CTXT \ INCLUDE_WDB_DIRECT_CALL \ INCLUDE_WDB_EVENTPOINTS \ INCLUDE_WDB_EVENTS \ INCLUDE_WDB_EXC_NOTIFY \ INCLUDE_WDB_EXIT_NOTIFY \ INCLUDE_WDB_FUNC_CALL \ INCLUDE_WDB_GOPHER \ INCLUDE_WDB_MEM \ INCLUDE_WDB_REG \ INCLUDE_WDB_START_NOTIFY \ INCLUDE_WDB_SYS \ INCLUDE_WDB_TASK \ INCLUDE_WDB_TASK_BP \ INCLUDE_WDB_TSFS \ INCLUDE_WDB_USER_EVENT \ INCLUDE_WDB_VIO \ INCLUDE_WDB_VIO_LIB \ INCLUDE_WINDVIEW \ INCLUDE_WINDVIEW_CLASS \ INCLUDE_WVUPLOAD_FILE \ INCLUDE_WVUPLOAD_TSFSSOCK COMPONENT_LIBS = include $(TGT_DIR)/h/make/defs.project ## build-configuration info ifeq ($(BUILD_SPEC),default) AR = arsimpc AS = ccsimpc CC = ccsimpc CFLAGS = -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nostdli b -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE)/ta rget/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD CFLAGS_AS = -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop -P -x assembler-with-cpp -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE)/target/h -I$(WIND_BASE)/target/config/comps/src -I $(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD CPP = ccsimpc -B$(WIND_BASE)/host/$(WIND_HOST_TYPE)/lib/gcc-lib/ -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U __WIN32__ -U__WIN32 -E -P -xc LD = ldsimpc LDFLAGS = --subsystem=windows LDOUT_CONV = wtxtcl $(WIND_BASE)/host/$(WIND_HOST_TYPE)/bin/simpcToExe.tcl LD_PARTIAL = ccsimpc -B$(WIND_BASE)/host/$(WIND_HOST_TYPE)/lib/gcc-lib/ -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U __WIN32__ -U__WIN32 -nostdlib -r -Wl,-X LD_PARTIAL_FLAGS = -r LD_RAM_FLAGS = $(WIND_BASE)/host/$(WIND_HOST_TYPE)/i386-pc-mingw32/lib/crt1.o LIBS = $(WIND_BASE)/target/lib/libSIMNTgnuvx.a NM = nmsimpc -g OPTION_DEFINE_MACRO = -D OPTION_INCLUDE_DIR = -I SIZE = sizesimpc VXSIZEPROG = echo RAM_LOW_ADRS = RAM_HIGH_ADRS = POST_BUILD_RULE = EXTRA_MODULES = endif # override make definitions only below this line # override make definitions only above this line include $(TGT_DIR)/h/make/rules.project ## build-configuration info ifeq ($(BUILD_SPEC),default) sysLib.o: $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD -c $(WIND_BASE)/target/config/simpc/sysLib.c prjConfig.o: $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD -c $(PRJ_DIR)/prjConfig.c linkSyms.o: $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD -c $(PRJ_DIR)/linkSyms.c s2.o: $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD -c D:/sys2/own_src/01/s2.c xxx1.o: $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD -c D:/sys2/own_src/01/xxx1.cpp s1.o: $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD -c D:/s1.cpp mkt2.o: $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD -c D:/sys2/own_src/01/mkt2.cpp endif ## dependencies sysLib.o: $(WIND_BASE)/target/config/simpc/sysLib.c \ $(PRJ_DIR)/prjComps.h \ $(PRJ_DIR)/prjParams.h prjConfig.o: $(PRJ_DIR)/prjConfig.c $(PRJ_DIR)/prjComps.h $(PRJ_DIR)/prjParams.h linkSyms.o: $(PRJ_DIR)/linkSyms.c s2.o: D:/sys2/own_src/01/s2.c xxx1.o: D:/sys2/own_src/01/xxx1.cpp s1.o: D:/s1.cpp mkt2.o: D:/sys2/own_src/01/mkt2.cpp \ D:sys2Tornadohost†-win32libgcc-libi386-pc-mingw32egcs-2.90.29../../../../i386-pc-mingw32/sys-include/sys/time.h ## user defined rules || ######### Makefile : END ########### - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation problem with sys/time.h Date: Wed, 22 May 2002 06:45:19 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: <38HG8.75864$UV4.127166@rwcrnsc54> References: I believe the time.h you are really interested in is located under target/h. The versions under the Tornado/host directory are intended for building the host tools, not a VxWorks target image. So, change your line to: #include "Alex Vinokur" wrote in message news:acfdg6$pavrf$1@ID-79865.news.dfncis.de... > > ========================================================= > Tornado : Version 2.0.2 > VxWorks : Version 5.4.2 > BSP : Version 1.1/1 > Compiler : ccsimpc > gcc version egcs-2.90.29 tornado 2.0 980515 > (egcs-1.0.3 release) > Host : Microsoft Windows 2000 [Version 5.00.2195] > Target : VxSim for Windows (VxWorks target simulator) > ToolChain : SIMNTgnu > ========================================================= > Project Type : bootable VxWorks image (custom configured) > ========================================================= > > > > I have got a compilation problem with a short program > that holds the line > #include > > > We can see the strange dependency for mkt2.o > in Makefile generated by the project manager > > ------------------------------------ > mkt2.o: D:/sys2/own_src/01/mkt2.cpp \ > D:sys2Tornadohost?-win32libgcc-libi386-pc-mingw32egcs-2.90.29../../../../i38 6-pc-mingw32/sys-include/sys/time.h > ------------------------------------ > > Two strangenesses : > 1. There are only two 'sys/time.h' files under > (in my system = D:/sys2/Tornado) : > * D:/sys2/Tornado/host/x86-win32/i386-pc-mingw32/sys-include/sys/time.h > * D:/sys2/Tornado/host/include/win32/sys/time.h > Why does 'D:sys2Tornadohost?-win32libgcc ... ' appear here ? > > 2. The 'D:sys2Tornadohost?-win32libgcc ... ' has invalid construction. > Nust be : > * D:sys2/Tornado/host/ ... > > > What is wrong? > > Relevant information is below. > > Thanks. > > > > > || ######### C++ Code : BEGIN ######### > > // File mkt2.cpp > #include > > int get2 () > { > return 0; > } > > || ######### C++ Code : END ########### > > > > > || ######### Compilation Log File : BEGIN ######### > > > vxrm *.o *.rpo ctdt.c symTbl.c vxWorks* > vxrm D:\sys2\Tornado\target\proj\sim\dir-01\prjComps.h D:\sys2\Tornado\target\proj\sim\dir > -01\prjParams.h D:\sys2\Tornado\target\proj\sim\dir-01\prjConfig.c D:\sys2\Tornado\target\ > proj\sim\dir-01\linkSyms.c > vxrm D:\sys2\Tornado\target\proj\sim\dir-01\libs.nm D:\sys2\Tornado\target\proj\sim\dir-01 > \libs.size > wtxtcl D:\sys2\Tornado\host\src\hutils\configGen.tcl D:\sys2\Tornado\target\proj\sim\dir-0 > 1\SIM-Project-01.wpj > ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN 32 -U__WIN32__ > -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer- pop -Wall -DRW_M > ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\ Tornado\target > \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\com ps\src -ID:\sy > s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\Tornado\target\config\simpc\ > sysLib.c > ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN 32 -U__WIN32__ > -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer- pop -Wall -DRW_M > ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\ Tornado\target > \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\com ps\src -ID:\sy > s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\Tornado\target\proj\sim\dir- > 01\prjConfig.c > ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN 32 -U__WIN32__ > -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer- pop -Wall -DRW_M > ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\ Tornado\target > \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\com ps\src -ID:\sy > s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\Tornado\target\proj\sim\dir- > 01\linkSyms.c > ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN 32 -U__WIN32__ > -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer- pop -Wall -DRW_M > ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\ Tornado\target > \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\com ps\src -ID:\sy > s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\own_src\01\s2.c > ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN 32 -U__WIN32__ > -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer- pop -Wall -DRW_M > ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\ Tornado\target > \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\com ps\src -ID:\sy > s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\sys2\own_src\01\xxx1.cpp > ccsimpc -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN 32 -U__WIN32__ > -U__WIN32 -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer- pop -Wall -DRW_M > ULTI_THREAD -D_REENTRANT -ID:\sys2\Tornado\target\proj\sim\dir-01 -ID:\sys2\ Tornado\target > \config\simpc -ID:\sys2\Tornado\target\h -ID:\sys2\Tornado\target\config\com ps\src -ID:\sy > s2\Tornado\target\src\drv -DCPU=SIMNT -DPRJ_BUILD -c D:\s1.cpp > make: *** No rule to make target `D:sys2Tornadohost?-win32libgcc-libi386-pc-mingw32egcs-2. > 90.29..\..\..\..\i386-pc-mingw32\sys-include\sys\time.h', needed by `mkt2.o'. Stop. > > Done. > > > || ######### Compilation Log File : END ########### > > > > > || ######### Makefile : BEGIN ######### > > # Makefile generated by the project manager > # > > > # GENERATED: Wed May 22 07:23:52 djg 2002 > # DO NOT EDIT - file is regenerated whenever the project changes > > > ## core information > > PRJ_DIR = $(WIND_BASE)/target/proj/sim/dir-01 > PRJ_FILE = SIM-Project-01.wpj > PRJ_TYPE = vxWorks > PRJ_OBJS = sysLib.o prjConfig.o linkSyms.o s2.o xxx1.o s1.o mkt2.o > BOOT_OBJS = > BUILD_SPEC = default > BSP_DIR = $(WIND_BASE)/target/config/simpc > TGT_DIR = $(WIND_BASE)/target > > > > ## set searching directories for dependencies > > vpath %.c $(BSP_DIR) > vpath %.cpp $(BSP_DIR) > vpath %.cxx $(BSP_DIR) > > > ## build-configuration info > > ifeq ($(BUILD_SPEC),default) > CPU = SIMNT > TOOL = gnu > DEFAULT_RULE = vxWorks > endif > > > ## component-configuration info > > COMPONENTS = INCLUDE_ANSI_ASSERT \ > INCLUDE_ANSI_CTYPE \ > INCLUDE_ANSI_LOCALE \ > INCLUDE_ANSI_MATH \ > INCLUDE_ANSI_STDIO \ > INCLUDE_ANSI_STDIO_EXTRA \ > INCLUDE_ANSI_STDLIB \ > INCLUDE_ANSI_STRING \ > INCLUDE_ANSI_TIME \ > INCLUDE_BOOT_LINE_INIT \ > INCLUDE_BUF_MGR \ > INCLUDE_CACHE_ENABLE \ > INCLUDE_CACHE_SUPPORT \ > INCLUDE_CPLUS \ > INCLUDE_CPLUS_DEMANGLER \ > INCLUDE_CPLUS_IOSTREAMS \ > INCLUDE_CPLUS_IOSTREAMS_FULL \ > INCLUDE_CPLUS_IOSTREAMS_TEST \ > INCLUDE_CPLUS_STL \ > INCLUDE_CPLUS_STRING \ > INCLUDE_CPLUS_STRING_IO \ > INCLUDE_DEBUG \ > INCLUDE_DLL \ > INCLUDE_DOSFS \ > INCLUDE_ENV_VARS \ > INCLUDE_EXC_HANDLING \ > INCLUDE_EXC_TASK \ > INCLUDE_FLOATING_POINT \ > INCLUDE_FORMATTED_IO \ > INCLUDE_HASH \ > INCLUDE_HW_FP_SHOW \ > INCLUDE_IO_SYSTEM \ > INCLUDE_KERNEL \ > INCLUDE_LOADER \ > INCLUDE_LOGGING \ > INCLUDE_MEMORY_CONFIG \ > INCLUDE_MEM_MGR_BASIC \ > INCLUDE_MEM_MGR_FULL \ > INCLUDE_MODULE_MANAGER \ > INCLUDE_MSG_Q \ > INCLUDE_MSG_Q_SHOW \ > INCLUDE_NTPASSFS \ > INCLUDE_PIPES \ > INCLUDE_POSIX_CLOCKS \ > INCLUDE_RBUFF \ > INCLUDE_RNG_BUF \ > INCLUDE_SELECT \ > INCLUDE_SEM_BINARY \ > INCLUDE_SEM_COUNTING \ > INCLUDE_SEM_MUTEX \ > INCLUDE_SEQ_TIMESTAMP \ > INCLUDE_SHELL \ > INCLUDE_SIGNALS \ > INCLUDE_SIO \ > INCLUDE_STANDALONE_SYM_TBL \ > INCLUDE_STARTUP_SCRIPT \ > INCLUDE_STAT_SYM_TBL \ > INCLUDE_STDIO \ > INCLUDE_SYM_TBL \ > INCLUDE_SYM_TBL_INIT \ > INCLUDE_SYSCLK_INIT \ > INCLUDE_SYSHW_INIT \ > INCLUDE_SYS_START \ > INCLUDE_TASK_HOOKS \ > INCLUDE_TASK_SHOW \ > INCLUDE_TASK_VARS \ > INCLUDE_TIMEX \ > INCLUDE_TRIGGERING \ > INCLUDE_TTY_DEV \ > INCLUDE_USER_APPL \ > INCLUDE_WATCHDOGS \ > INCLUDE_WDB \ > INCLUDE_WDB_BANNER \ > INCLUDE_WDB_BP \ > INCLUDE_WDB_COMM_PIPE \ > INCLUDE_WDB_CTXT \ > INCLUDE_WDB_DIRECT_CALL \ > INCLUDE_WDB_EVENTPOINTS \ > INCLUDE_WDB_EVENTS \ > INCLUDE_WDB_EXC_NOTIFY \ > INCLUDE_WDB_EXIT_NOTIFY \ > INCLUDE_WDB_FUNC_CALL \ > INCLUDE_WDB_GOPHER \ > INCLUDE_WDB_MEM \ > INCLUDE_WDB_REG \ > INCLUDE_WDB_START_NOTIFY \ > INCLUDE_WDB_SYS \ > INCLUDE_WDB_TASK \ > INCLUDE_WDB_TASK_BP \ > INCLUDE_WDB_TSFS \ > INCLUDE_WDB_USER_EVENT \ > INCLUDE_WDB_VIO \ > INCLUDE_WDB_VIO_LIB \ > INCLUDE_WINDVIEW \ > INCLUDE_WINDVIEW_CLASS \ > INCLUDE_WVUPLOAD_FILE \ > INCLUDE_WVUPLOAD_TSFSSOCK > COMPONENT_LIBS = > > > include $(TGT_DIR)/h/make/defs.project > > > ## build-configuration info > > ifeq ($(BUILD_SPEC),default) > AR = arsimpc > AS = ccsimpc > CC = ccsimpc > CFLAGS > > > > > > > > > > > = -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U_ _WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nostdli > b -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(PRJ_D IR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE)/ta > rget/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/src/drv - -DCPU=SIMNT -DPRJ_BUILD > CFLAGS_AS = -mpentium -ansi -nostdinc -g -nostdlib -fno-builtin -fno-defer-pop -P -x > assembler-with-cpp -I$(PRJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_ BASE)/target/h -I$(WIND_BASE)/target/config/comps/src -I > $(WIND_BASE)/target/src/drv -DCPU=SIMNT -DPRJ_BUILD > CPP = > ccsimpc -B$(WIND_BASE)/host/$(WIND_HOST_TYPE)/lib/gcc-lib/ -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U > __WIN32__ -U__WIN32 -E -P -xc > LD = ldsimpc > LDFLAGS = --subsystem=windows > LDOUT_CONV = wtxtcl $(WIND_BASE)/host/$(WIND_HOST_TYPE)/bin/simpcToExe.tcl > LD_PARTIAL = > ccsimpc -B$(WIND_BASE)/host/$(WIND_HOST_TYPE)/lib/gcc-lib/ -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U > __WIN32__ -U__WIN32 -nostdlib -r -Wl,-X > LD_PARTIAL_FLAGS = -r > LD_RAM_FLAGS = $(WIND_BASE)/host/$(WIND_HOST_TYPE)/i386-pc-mingw32/lib/crt1.o > LIBS = $(WIND_BASE)/target/lib/libSIMNTgnuvx.a > NM = nmsimpc -g > OPTION_DEFINE_MACRO = -D > OPTION_INCLUDE_DIR = -I > SIZE = sizesimpc > VXSIZEPROG = echo > RAM_LOW_ADRS = > RAM_HIGH_ADRS = > POST_BUILD_RULE = > EXTRA_MODULES = > endif > > # override make definitions only below this line > > > # override make definitions only above this line > > include $(TGT_DIR)/h/make/rules.project > > > ## build-configuration info > > ifeq ($(BUILD_SPEC),default) > > sysLib.o: > > $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos > tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(P RJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE > )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/sr c/drv -DCPU=SIMNT -DPRJ_BUILD -c > $(WIND_BASE)/target/config/simpc/sysLib.c > > prjConfig.o: > > $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos > tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(P RJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE > )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/sr c/drv -DCPU=SIMNT -DPRJ_BUILD -c $(PRJ_DIR)/prjConfig.c > > linkSyms.o: > > $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos > tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(P RJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE > )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/sr c/drv -DCPU=SIMNT -DPRJ_BUILD -c $(PRJ_DIR)/linkSyms.c > > s2.o: > > $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos > tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(P RJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE > )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/sr c/drv -DCPU=SIMNT -DPRJ_BUILD -c D:/sys2/own_src/01/s2.c > > xxx1.o: > > $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos > tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(P RJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE > )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/sr c/drv -DCPU=SIMNT -DPRJ_BUILD -c > D:/sys2/own_src/01/xxx1.cpp > > s1.o: > > $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos > tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(P RJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE > )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/sr c/drv -DCPU=SIMNT -DPRJ_BUILD -c D:/s1.cpp > > mkt2.o: > > $(CC) -U__WINNT -UWIN32 -U__WINNT__ -UWINNT -U__MINGW32__ -U_WIN32 -U__WIN32 -U__WIN32__ -U__WIN32 -mpentium -ansi -nostdinc -g -nos > tdlib -fno-builtin -fno-defer-pop -Wall -DRW_MULTI_THREAD -D_REENTRANT -I$(P RJ_DIR) -I$(WIND_BASE)/target/config/simpc -I$(WIND_BASE > )/target/h -I$(WIND_BASE)/target/config/comps/src -I$(WIND_BASE)/target/sr c/drv -DCPU=SIMNT -DPRJ_BUILD -c > D:/sys2/own_src/01/mkt2.cpp > endif > > > ## dependencies > > > sysLib.o: $(WIND_BASE)/target/config/simpc/sysLib.c \ > $(PRJ_DIR)/prjComps.h \ > $(PRJ_DIR)/prjParams.h > > > prjConfig.o: $(PRJ_DIR)/prjConfig.c $(PRJ_DIR)/prjComps.h $(PRJ_DIR)/prjParams.h > > > linkSyms.o: $(PRJ_DIR)/linkSyms.c > > > s2.o: D:/sys2/own_src/01/s2.c > > > xxx1.o: D:/sys2/own_src/01/xxx1.cpp > > > s1.o: D:/s1.cpp > > > mkt2.o: D:/sys2/own_src/01/mkt2.cpp \ > D:sys2Tornadohost?-win32libgcc-libi386-pc-mingw32egcs-2.90.29../../../../i38 6-pc-mingw32/sys-include/sys/time.h > > > ## user defined rules > > > > || ######### Makefile : END ########### > > > -- > ==================== > Alex Vinokur > http://up.to/alexvn > http://go.to/alexv_math > mailto:alexvn@bigfoot.com > mailto:alexvn@go.to > ==================== > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Could I spawn a task when booting up? Date: 21 May 2002 23:47:22 -0700 From: surveyl@hotmail.com (survey liu) Organization: http://groups.google.com/ Message-ID: <77416529.0205212247.67e72aa3@posting.google.com> HI,there Could I use taskSpawn to spawn a new task in task "tBoot" whill system booting up? and could I use watchdog timer? - -- Best regards, Survey Liu --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Could I spawn a task when booting up? Date: Wed, 22 May 2002 09:32:19 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <77416529.0205212247.67e72aa3@posting.google.com> Reply-To: "Michael Lawnick" Sender: mlawnick@pd9524f66.dip.t-dialin.net "survey liu" schrieb im Newsbeitrag news:77416529.0205212247.67e72aa3@posting.google.com... > HI,there > > Could I use taskSpawn to spawn a new task in task "tBoot" whill system > booting up? and could I use watchdog timer? > > -- > Best regards, > Survey Liu Hi, yes, from usrRoot() in bootConfig.c ($WIND_BASE/target/config/all). There system is running in tasking mode. For watchdog you may have to call the init routine first. Your task/watchdog will run until call of entry() in bootLoad(), then its blown away to Nirwana. So if you want to do timeout-checks of image-load, this will work, but not to control the startup of VxWorks. HTH Michael Lawnick --------------------------- Newsgroups: comp.os.vxworks Subject: Where to download the patch for tornado? Date: 22 May 2002 00:38:39 -0700 From: sdssly@sina.com (sdssly) Organization: http://groups.google.com/ Message-ID: Please help me. Send mail to me at sdssly@sina.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compilation problem with sys/time.h Date: Wed, 22 May 2002 09:51:38 +0100 From: Alex Vinokur Message-ID: <3CEB5C1A.4000501@bigfoot.com> References: <38HG8.75864$UV4.127166@rwcrnsc54> Gary M wrote: > I believe the time.h you are really interested in is located under target/h. Yes, you are right. > > The versions under the Tornado/host directory are intended for building the > host tools, not a VxWorks target image. By the way, how to build host tools. Via Workspace Window we can create either bootable VxWorks image or downloadable application module for VxWorkx. Both cases are not host tools. > > So, change your line to: > > #include Thanks. Now it is compiled OK. > > "Alex Vinokur" wrote in message > news:acfdg6$pavrf$1@ID-79865.news.dfncis.de... > [snip] Nevertheless : Why non-existent path ? ---> >> Why does 'D:sys2Tornadohost?-win32libgcc ... ' appear here ? Why invalid construction ? ---> >> 2. The 'D:sys2Tornadohost?-win32libgcc ... ' has invalid construction. >> Nust be : >> * D:sys2/Tornado/host/ ... >> [snip] - -- ============================ Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ============================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where to download the patch for tornado? Date: 22 May 2002 08:08:22 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: sdssly@sina.com (sdssly) wrote: > Please help me. Look at WindSurf, the download section. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Wed, 22 May 2002 10:31:08 +0100 From: Alex Vinokur Message-ID: <3CEB655C.306@bigfoot.com> References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> Joe Chung wrote: > In the doc for ansiTime, you'll see the format for the timezone > environment variable. Here's my example: > /* Note that my daylight savings time start and end are not correct */ > putenv("TIMEZONE=PDT::480:040102:100102"); TIMEZONE is of the form: name_of_zone:<(unused)>:time_in_minutes_from_UTC:daylight_start:daylight_end How can I know what is time_in_minutes_from_UTC's value for a specific zone (Israel) ? > > In the doc for envLib you'll see general info about environments. > > -jc > - -- ============================ Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ============================ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Wed, 22 May 2002 12:00:40 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> <3CEB655C.306@bigfoot.com> "Alex Vinokur" wrote in message news:3CEB655C.306@bigfoot.com... > > > Joe Chung wrote: > > > In the doc for ansiTime, you'll see the format for the timezone > > environment variable. Here's my example: > > /* Note that my daylight savings time start and end are not correct */ > > putenv("TIMEZONE=PDT::480:040102:100102"); > > > > TIMEZONE is of the form: > name_of_zone:<(unused)>:time_in_minutes_from_UTC:daylight_start:daylight_end > > > How can I know what is time_in_minutes_from_UTC's value for a specific > zone (Israel) ? > [snip] I used the following line : putenv("TIMEZONE=Israel::180:040102:100102"); Here are some results I obtained via mktime() : DATA-TIME 1970:01:01::00:00:00 ---> 000000010800 DATA-TIME 1970:01:01::00:00:01 ---> 000000010801 DATA-TIME 1970:01:01::00:00:02 ---> 000000010802 DATA-TIME 1970:01:01::01:00:00 ---> 000000014400 DATA-TIME 1970:01:01::01:00:01 ---> 000000014401 DATA-TIME 1970:01:01::01:00:02 ---> 000000014402 DATA-TIME 1970:01:01::03:00:00 ---> 000000021600 DATA-TIME 1970:01:01::03:00:01 ---> 000000021601 DATA-TIME 1970:01:01::03:00:02 ---> 000000021602 I think it has to be as following (for instance, for 1970:01:01::03:00:00 ) : DATA-TIME 1970:01:01::03:00:00 ---> 000000000000 Any comments ? Thanks in advance. - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 22 08:41:15 2002 From: Atkinson Andrew Date: Wed May 22 08:41:17 PDT 2002 Subject: Timer - Hi All I need to realise a timer with at least millisecond resolution and capable of measuring periods down to 250ms and up to 30 seconds. I was going to cascade two counter/timers present on a Z8536 CIO we have on our board, but it looks like one of the timers is in use. Has anyone got any ideas? I am using T2/5.4 on a PPC603. Cheers Andy Atkinson Software Engineer Bldg 31a QinetiQ, Bincleaves Technology Park Weymouth Dorset DT4 8UR Tel : 01305 764660 Fax : 01305 764661 e-mail : AATKINSON1@QinetiQ.com vxworks 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 Wed May 22 12:37:44 2002 From: Bruce Wilson Date: Wed May 22 12:37:47 PDT 2002 Subject: Re: Timer - We use the 64-bit timebase register as a high res clock often in our vxWorks applications. We didn't know about vxTimeBaseGet(), so we wrote our own simple function which is essentially the same (type "l vxTimeBaseGet" in windsh). On the PowerPC 603e (Motorola mvme1603, Force cpci3603, Gespac pcippc1, etc), the timebase register gives 60 nsec ticks. It's a great resource. --Bruce Wilson U.C. LLNL Livermore, CA >Submitted-by vxwexplo-errs@csg.lbl.gov Wed May 22 08:41:15 2002 >Submitted-by: Atkinson Andrew > >Hi All > >I need to realise a timer with at least millisecond resolution and >capable of measuring periods down to 250ms and up to 30 seconds. I >was going to cascade two counter/timers present on a Z8536 CIO we >have on our board, but it looks like one of the timers is in use. >Has anyone got any ideas? > >I am using T2/5.4 on a PPC603. > >Cheers > >Andy Atkinson >Software Engineer >Bldg 31a >QinetiQ, Bincleaves Technology Park >Weymouth >Dorset >DT4 8UR > >Tel : 01305 764660 >Fax : 01305 764661 >e-mail : AATKINSON1@QinetiQ.com From vxwexplo-errs@csg.lbl.gov Thu May 23 01:59:35 2002 From: "Asha Hosakoppa" Date: Thu May 23 01:59:37 PDT 2002 Subject: Does vxWorks have Events ? This is a multi-part message in MIME format. ------=_NextPartTM-000-075421dc-db9c-4b25-aece-4c504d2b90b0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0007_01C20266.1AFD41C0" ------=_NextPart_000_0007_01C20266.1AFD41C0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit MessageDoes vxWorks have Event Handling(not signals), for InterTask Communication? ------=_NextPart_000_0007_01C20266.1AFD41C0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Message
Does = vxWorks have=20 Event Handling(not signals), for InterTask=20 Communication?
------=_NextPart_000_0007_01C20266.1AFD41C0-- ------=_NextPartTM-000-075421dc-db9c-4b25-aece-4c504d2b90b0 Content-Type: text/plain; name="Wipro_Disclaimer.txt" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="Wipro_Disclaimer.txt" **************************Disclaimer************************************************** Information contained in this E-MAIL being proprietary to Wipro Limited is 'privileged' and 'confidential' and intended for use only by the individual or entity to which it is addressed. You are notified that any use, copying or dissemination of the information contained in the E-MAIL in any manner whatsoever is strictly prohibited. **************************************************************************************** ------=_NextPartTM-000-075421dc-db9c-4b25-aece-4c504d2b90b0-- From vxwexplo-errs@csg.lbl.gov Thu May 23 03:00:49 2002 From: mat par Date: Thu May 23 03:00:51 PDT 2002 Subject: use dns solver on vxworks hi, i'm a french student and i need to use the fonction resolvGetHostByName on vxworks in order to make a server. But when i want to upload my project in tornado he send me an error message about this fonction that was not find! i've read the help and they talk about change configAll.h. when i modify it, have i to compile the kernel or changes are actives without? is there another system to initialise host infos without this fonction? thanx ===== Lestat, Le Pingouin Fouuuuuu __________________________________________________ Do You Yahoo!? LAUNCH - Your Yahoo! Music Experience http://launch.yahoo.com From vxwexplo-errs@csg.lbl.gov Thu May 23 04:03:18 2002 From: Vxworks Exploder Date: Thu May 23 04:03:20 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu May 23 04:03:14 PDT 2002 Subject: Re: getting started with vxworks? How to? Subject: Re: get that MAC...not! Subject: Re: getting started with vxworks? How to? Subject: Fake a destructor on VxWorks Subject: Debug with Visual SlickEdit v7.0 Subject: VxWorks Device Driver Tutorial Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Subject: Re: Correct method to change IP addr? Subject: Re: Newbie questions Subject: Re: tracescope, cannot register function Subject: Re: Fake a destructor on VxWorks Subject: Re: VxWorks Device Driver Tutorial Subject: Spawning threads with ACE library Subject: chipDISK Subject: why do system calls hang? Subject: GCC: Max file size ? Subject: Re: why do system calls hang? Subject: How can I uniquely identify a file in dosFs Subject: Re: Where to download the patch for tornado? Subject: Re: Fake a destructor on VxWorks Subject: Re: GCC: Max file size ? Subject: Tornado ProfileScope Subject: Download problem Subject: Re: Download problem Subject: Re: chipDISK Subject: can't boot vxwork by network use intel82259 driver ? Subject: Re: Where to download the patch for tornado? Subject: Re: Fake a destructor on VxWorks Subject: Problem with a dosFs Subject: Race Conditions in Vxworks!!?? Subject: Compression Subject: Re: Race Conditions in Vxworks!!?? Subject: VME bus errors and signal handlers ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: getting started with vxworks? How to? Date: Wed, 22 May 2002 11:40:10 +0000 (UTC) From: "Alex Vinokur" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: References: "Wojciech Mirski" wrote in message news:acdbbj$seu$1@news.mch.sbs.de > Salute Fellows! > > I am new in this field, but want to deepen knowledge about VxWorks. > I would like to know, if there is any book concerning programming under > vxworks for x86? > Additionally I'm interested in "getting started" or something like this > tutorial, and any info about making device drivers. > > All help will be very appreciated. > > Best regards > > Wojciech Mirski http://www.cs.virginia.edu/~cs551rt/started.html ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== - -- Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: get that MAC...not! Date: 22 May 2002 04:50:40 -0700 From: kenb88@hotmail.com (Whacked) Organization: http://groups.google.com/ Message-ID: <6755aee5.0205220350.78b458ec@posting.google.com> References: <6755aee5.0205211224.1fde1d89@posting.google.com> "Leonid Rosenboim" wrote in message news:... > Firsst, you need to be sure to test the pointer returned > by ifunit(), and make sure it is not NULL. > Second, you allow ARP three retires, with only 3 clock > ticks between retries, that could be not enough for the > host to reply, depending on your sysClkRateGet() value. > Increase the last parameter to e.g. > > if ( myifnet != NULL ) > etherAddrResolve (myifnet, "192.168.1.250", buffer, 3, > sysClkRateGet()/10 ); > > "Whacked" wrote in message > news:6755aee5.0205211224.1fde1d89@posting.google.com... > > Hi, > > > > I have a simple network configuration. My VxWorks box has IP address > > 192.168.1.1 and is connected DIRECTLY to a router with IP address > > 192.168.1.250. > > > > I then do the following: > > > > struct ifnet *myifnet; > > char buffer[20]; > > > > myifnet = ifunit("le0"); > > etherAddrResolve (myifnet, "192.168.1.250", buffer, 3, 3); > > > > > > the etherAddrResolve always returns with ERROR and errno is set > > to something like 0x32000001...the cross reference in vwModNum.h links > > me > > to hostLib.h which says the errno means unknown host. > > > > what am I doing wrong?? This should give me the ethernet address in > > buffer. > > > > thanks, Leonid, Thanks...I'll try that suggestion. Unfortunately, with the 3,3 in the function call, I don't even see an arp go out on the wire. I think that is strange! thanks, --------------------------- Newsgroups: comp.os.vxworks Subject: Re: getting started with vxworks? How to? Date: Wed, 22 May 2002 14:53:38 +0200 From: "Alex Vinokur" Organization: Scopus Message-ID: References: "Alex Vinokur" wrote in message news:b993cf6ce5c70fb2212f779e0f5f4351.13177@mygate.mailgate.org... > "Wojciech Mirski" wrote in message > news:acdbbj$seu$1@news.mch.sbs.de > > > Salute Fellows! > > > > I am new in this field, but want to deepen knowledge about VxWorks. > > I would like to know, if there is any book concerning programming under > > vxworks for x86? > > Additionally I'm interested in "getting started" or something like this > > tutorial, and any info about making device drivers. > > > > All help will be very appreciated. > > > > Best regards > > > > Wojciech Mirski > > http://www.cs.virginia.edu/~cs551rt/started.html Also http://www.eng.uah.edu/docs/vxworks/tornado/winguide/index.html http://daq-nt100.mps.ohio-state.edu/daqiii/archive/Manuals/Tornado/ - -- ==================== Alex Vinokur http://up.to/alexvn http://go.to/alexv_math mailto:alexvn@bigfoot.com mailto:alexvn@go.to ==================== --------------------------- Newsgroups: comp.os.vxworks Subject: Fake a destructor on VxWorks Date: 22 May 2002 06:40:37 -0700 From: pboulay@hotmail.com (Patrick Boulay) Organization: http://groups.google.com/ Message-ID: <8bb5523.0205220540.56ec523f@posting.google.com> I have a big problem. Our VxWorks cards don't support the new C++ compiler.. so we have a lot of problem. One of the big problem it's to call destructor in polymorphism like: A ptr; ptr = new B; delete ptr; Is it possible to fake a destructor and use free fonction from C? like: ptr->destroy(); // a virtual function for every class free(ptr); I really need help! Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Debug with Visual SlickEdit v7.0 Date: 22 May 2002 07:11:33 -0700 From: f.petri@steinbichler.de (F. Petri) Organization: http://groups.google.com/ Message-ID: <839bd939.0205220611.14e528dd@posting.google.com> I was just wondering if anyone knows if it is possible to use Visual SlickEdit v7.0 as debugger for Tornado. The statement I obtained when I asked the guys from SlickEdit was: "The current set up is that Visual SlickEdit will only debug GNU projects that are created within the Visual SlickEdit environment. If you have access to the GNU tools (Cygwin), then you can debug through Visual SlickEdit." --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks Device Driver Tutorial Date: 22 May 2002 08:11:32 -0700 From: trnswamy@netkracker.com (Narasimha Swamy) Organization: http://groups.google.com/ Message-ID: <50c9292c.0205220711.68e14a2e@posting.google.com> I am looking for a tutorial on VxWorks Device Driver. Can any one please suggest any site for that. If it contains case studies, it will be better. Thank you for any suggestion. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Wed, 22 May 2002 15:10:13 GMT From: igtorque@yahoo.es (Ignacio Gonzalez T.) Organization: Newshosting.com - Highest quality at a great price! www.newshosting.com Message-ID: <3cebb3b5.89154657@news.mhogaming.com> References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> <3CEB655C.306@bigfoot.com> On Wed, 22 May 2002 10:31:08 +0100, Alex Vinokur wrote: > >TIMEZONE is of the form: >name_of_zone:<(unused)>:time_in_minutes_from_UTC:daylight_start:daylight_end > > >How can I know what is time_in_minutes_from_UTC's value for a specific >zone (Israel) ? > Just look at an agenda. Mine says Israel is one hour away from UTC, so time_in_minutes_from_UTC should be 60 (or -60?) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Correct method to change IP addr? Date: Wed, 22 May 2002 11:24:38 -0400 From: david lindauer Message-ID: <3CEBB836.5C6ED038@notifier-is.net> References: I believe the only restriction is that you use ifMaskSet before ifAddrSet, which you have done. We do it as follows: ifMaskSet ifAddrSet routeDelete (if there was an old route) routeAdd (new route) Carl Petersen wrote: > Is this the correct sequence to change the IP addr of an interface? > > ifRouteDelete(...); > > ifMaskSet(...); > > ifAddrSet( ifName, "New IP" ); > > These calls run without error however a subsequent call to bind() fails. > > I have also tried ipDetach() then ipAttach() without success. Note that > the interface is operational before I attempt to modify the IP addr. > > Tornado 2.02 Pentium BSP > > Thanks > > Carl Petersen > > -- > Posted via Mailgate.ORG Server - http://www.Mailgate.ORG --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Newbie questions Date: Wed, 22 May 2002 15:40:46 GMT From: bill@woola.qswtools.com Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <2_OG8.556$Hl3.90@newsread1.prod.itd.earthlink.net> References: In article , "Liam Friel" writes: > Hi, > > I have checked the WRS website but am none the wiser really. > I also have a call in to the local account manager, but I'd really like the "real story". > > Can someone quickly tell me: > - are the tools (cross compilers/Tornado etc) for VxWorks 5.4 for ARM available for a Linux host? Nope. Only the AE tools are available on Linux > - what is the difference between VxWorks 5.4 and AE? AE lets you divide your app into memory protection domains, offering protection for the kernel from the apps, and the apps from each other. > - are libraries compiled on a Solaris host for a given target linkable with libraries compiled on a Windows host for the same > target, and vice versa? Yes. > > Thanks, > Liam > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: tracescope, cannot register function Date: Wed, 22 May 2002 08:53:49 -0700 From: "Paul Covell" Message-ID: References: <3CC99B92.37C59311@ll.mit.edu> <3CCEA845.1498F38D@ll.mit.edu> Michael, my apologies for the delay in this reply. Your analysis is correct - TraceScope expects certain prologue signatures to be present in order to patch the code at that function. Our upcoming release of TraceScope 3.1 corrects the problem you are encountering with leaf functions on the PPC. I will see if I can modify the documentation to be more verbose regarding the technical requirements of TraceScope's patch functionality. Sincerely, Paul - -- Paul Covell paul@rti.com Software Engineer Real-Time Innovations, Inc. "Michael Reed" wrote in message news:3CCEA845.1498F38D@ll.mit.edu... > Here are 2 good replys I received to this posting.. I though they should > make it into the archives. This is the sort of stuff that should make > it into "Release Notes", "Known Issues", "Target Arch-Specific Notes", > etc. WindSurf didn't have any notes on this either. > > --------------------------- > > I saw the message you posted to comp.os.vxworks, so I'll try to > answer. You can see the message you're referring to in three > different sets of circumstances: > > (1) On some target architectures, most notably PPC, "leaf" functions > (those that don't call any other functions) have a completely > different stack layout. As a result, TraceScope may be unable to > currently trace leaf functions on those targets. RTI is planning > on fixing this in the future. > > (2) On many target architectures, you can get the compiler to generate > different function prologues and epilogues by using flags such as > "-fno-omit-frame-ptr", etc. Sometimes these create problems for > TraceScope. (This is what "normally compiled" is trying to refer > to.) > > (3) Hand-coded assembly functions usually cannot be traced. (I know > this isn't the case for you.) > > ------------------------------ > > This explanation applies only to the PPC; if you're running on a > different > processor, the problem could lie elsewhere. The PPC ABI doesn't require > the > compiler to output code to build a full-blown stack frame for 'terminal' > routines, i.e., those routines that don't call any subsidiary routines. > The > TraceScope 'patcher' doesn't find the expected signature on those > routines & > so doesn't believe they were "normally compiled". Until (if ever) RTI > makes > the 'patcher' smarter, the only workaround I know of is to include a > fake > call in the routine of interest in order to get the compiler to build a > proper stack frame. > > > > > Michael Reed wrote: > > > > Are there any RTI / ScopeTool users out here. From searchs of this > > newsgroup and WindSerf, I suspect not too many. I'm having a problem w/ > > TraceScope where it reports that I cannot register a particular function > > for tracing: > > > > PatchSet: 0x9dea54 (0x9421ffd8)is not a compiler-generated function. > > PatchClear: 0x9dea54 not found. > > > > I can sucessfully register other functions. What's different about this > > function? The API for the PatchSet call says you can only patch > > "normally-compiled function". As in, like, I "normally compile" my code > > before running it? Yes, normally I do this.. > > > > THanks, > > Mike R. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fake a destructor on VxWorks Date: Wed, 22 May 2002 16:00:11 GMT From: "H. E. Taylor" Organization: MTS Internet Message-ID: <3CEBDB31.4CBB@despam.autobahn.mb.ca> References: <8bb5523.0205220540.56ec523f@posting.google.com> In article <8bb5523.0205220540.56ec523f@posting.google.com>, Patrick Boulay wrote: > > I have a big problem. Our VxWorks cards don't support the new C++ > compiler.. so we have a lot of problem. > Can you be more specific. It seems to me this is a misapprehension. It is the host which needs to support the compiler, not the target... > One of the big problem it's to call destructor in polymorphism like: > > A ptr; > > ptr = new B; > > delete ptr; > > Is it possible to fake a destructor and use free fonction from C? > like: > > ptr->destroy(); // a virtual function for every class > free(ptr); > > I really need help! > This sounds like looking for trouble. How deep is your inheritance tree? How will you deal with arrays? How will you deal with multiple inheritance? How will you deal with opaque pointers? - -het - -- "progress in software has not followed Moore's law." -John Holland Computer Links: http://www.autobahn.mb.ca/~het/clinks.html H.E. Taylor http://www.autobahn.mb.ca/~het/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: VxWorks Device Driver Tutorial Date: Wed, 22 May 2002 16:03:18 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <50c9292c.0205220711.68e14a2e@posting.google.com> The stuff you need can be found in WindRiver's BSP Porting Kit manual, which is quite expensive to buy, but even then it assumes you know some basics about software-hardware interfaces. Also, there is a WindRiver BSP and Device Drivers training workshop, which covers the issues. "Narasimha Swamy" wrote in message news:50c9292c.0205220711.68e14a2e@posting.google.com... > I am looking for a tutorial on VxWorks Device Driver. Can any one > please suggest any site for that. If it contains case studies, it will > be better. > > Thank you for any suggestion. --------------------------- Newsgroups: comp.os.vxworks Subject: Spawning threads with ACE library Date: 22 May 2002 09:47:05 -0700 From: u.mascia@pxl.it (Umberto Mascia) Organization: http://groups.google.com/ Message-ID: Hi all, I have a problem using ACE library, when I try to spawn tasks. I call ACE_Thread::spawn_n (a wrapper function to spawn n tasks) in order to create 3 or more tasks; and then I join the main task to the spawned ones (these are "while(1)" or infinite tasks). The result is: - - only 2 tasks are created, but they run correctly; - - the main task returns from join, even if the created tasks are running; - - calling printErrno from the shell, I obtain: errno = 0x003D001 = S_objLib_OBJ_ID_ERROR Thanks in advance, Umberto Mascia - -------------------------- P.S. You can find more details in the canonic ACE Problem Report Form: ACE VERSION: 5.2.1 HOST MACHINE and OPERATING SYSTEM: i586 WindowsNT TARGET MACHINE and OPERATING SYSTEM: PPC860 VxWorks COMPILER NAME AND VERSION (AND PATCHLEVEL): egcs-2.7.2-960126 (Tornado 2) CONTENTS OF $ACE_ROOT/ace/config.h: #include "config-vxworks5.x.h" CONTENTS OF $ACE_ROOT/include/makeinclude/platform_macros.GNU platform_vxworks5.x_g++.GNU AREA/CLASS/EXAMPLE AFFECTED: example thread_sample.cpp thread_manager.cpp tests Thread_Manager_Test.cpp DOES THE PROBLEM AFFECT: COMPILATION? NO LINKING? NO EXECUTION? YES OTHER (please specify)? NO SYNOPSIS: Problem whith spawn of thread DESCRIPTION: In thread_sample.cpp the method ACE_Thread::spawn_n returns OK but it creates only two threads at maximum. Additionally join() returns OK, despite the (two) threads are running. In thread_manager.cpp, the threads behavior is equal to the previous example, moreover the spawn returns -1 after the creation of (only two) threads. The Thread_Manager_Test.cpp doesn't generate any thread and exits with unknown error num 11411113. I have compiled libACE statical (libACE.a) --------------------------- Newsgroups: comp.os.vxworks Subject: chipDISK Date: Wed, 22 May 2002 17:34:07 GMT From: igtorque@yahoo.es (Ignacio G.T.) Organization: Newshosting.com - Highest quality at a great price! www.newshosting.com Message-ID: <3cebceb5.96066856@news.mhogaming.com> I'm planning to use a PC/104 board from JUMPtec. I've been told by a distributor that it comes with VxWorks drivers. So I've checked their web site and found an Ethernet driver and a special set of SW modules & API (called JIDA) to access the watchdog, the S/N and the like. My concern is that I want to use flash storage. They offer chipDISK-IDE, that can be connected to the card through the IDE connector. They claim this is "transparent" to the OS. So I think that, using the standard PentiumPC BSP, adding the Ethernet driver and the JIDA, and setting up an ATA-based DOS file system I could access the chipDISK-IDE. Is this right? Or should I get/write special drivers? I know the caveats about limited write-cycles on Flash, etc. That is not important for my application. - -- Ignacio G.T. --------------------------- Newsgroups: comp.os.vxworks Subject: why do system calls hang? Date: Wed, 22 May 2002 10:45:54 -0700 From: "Mark Galecki" Message-ID: <3cebd979$0$12554$4c41069e@reader0.ash.ops.us.uu.net> Hello, sometimes when I run VxWorks, system calls as varied as taskDelete() iosDevAdd() hang. I can execute other calls and I don't think my apps have corrupted system memory. By "hang" I mean that the shell does not show anything and does not go back to the prompt after I type the task name and press enter. This happens even if I try to execute them without arguments (in which case I would expect error response). This does not happen all the time for these calls, only some of the time. After the hang, I can kill the shell, and start another one, and get the same result. Right now I am unable to zero in on the precise conditions. Has anybody experienced something liike that and what could be the cause? Thank you, Mark Galecki --------------------------- Newsgroups: comp.os.vxworks Subject: GCC: Max file size ? Date: Wed, 22 May 2002 18:11:13 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <5bRG8.24732$f5.2005649@news> Hello, Does gcc have a max source file size ... say 100 000 statements ... Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why do system calls hang? Date: Wed, 22 May 2002 11:37:00 -0700 From: "Mark Galecki" Message-ID: <3cebe574$0$12553$4c41069e@reader0.ash.ops.us.uu.net> References: <3cebd979$0$12554$4c41069e@reader0.ash.ops.us.uu.net> Never mind - this was a false-start. Obviously these system calls will cause unpredictable results when called with bad arguments. And without arguments, the default NULL is a bad one. --------------------------- Newsgroups: comp.os.vxworks Subject: How can I uniquely identify a file in dosFs Date: Wed, 22 May 2002 15:22:47 -0700 From: "Parangodath, Binish [SC7:326:EXCH]" Organization: Nortel Networks Message-ID: Hi , I want to uniquely identify a file in dosFs.I tried to get stat structure of two files and comapred the st_dev member and st_ino.To my astonishment it is the same for two different files in my system.Is there any better way than a string comparison of filenames with full path Regards, Binish --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where to download the patch for tornado? Date: 22 May 2002 17:50:57 -0700 From: sdssly@sina.com (sdssly) Organization: http://groups.google.com/ Message-ID: References: sdssly@sina.com (sdssly) wrote in message news:... > Please help me. > > Send mail to me at sdssly@sina.com Thanks for your response. However, I downloaded the Tornado from the free downlaod site. Can i download the patch for it? Thanks again. Tyler Sdssly@sina.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fake a destructor on VxWorks Date: Thu, 23 May 2002 03:10:38 GMT From: bill@woola.qswtools.com (William C. Cox) Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <8bb5523.0205220540.56ec523f@posting.google.com> In article <8bb5523.0205220540.56ec523f@posting.google.com>, pboulay@hotmail.com (Patrick Boulay) writes: > I have a big problem. Our VxWorks cards don't support the new C++ > compiler.. so we have a lot of problem. > What is it about the cards? Do they support the C compiler? We need more information! > One of the big problem it's to call destructor in polymorphism like: > > A ptr; > > ptr = new B; > > delete ptr; > What happens when you call the destructor? At compile time? At run time? > > Is it possible to fake a destructor and use free fonction from C? > like: > > ptr->destroy(); // a virtual function for every class > free(ptr); generally, no. Why do you wish to? > > > I really need help! > > Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GCC: Max file size ? Date: Thu, 23 May 2002 03:20:12 GMT From: bill@woola.qswtools.com Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: References: <5bRG8.24732$f5.2005649@news> In article <5bRG8.24732$f5.2005649@news>, "Jim" writes: > Hello, > Does gcc have a max source file size ... say 100 000 statements ... Nope. But who would want a single file that large? It it automatically generated somehow? > > Jim > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado ProfileScope Date: Thu, 23 May 2002 00:58:16 -0400 From: "Mustafa Celik" Organization: Bell Sympatico Message-ID: Hi there, Is there any restriction for ProfileScope to pick up some symbols? It picks up the symbols that are loaded during initialization, but it does not take the symbols that are loaded later manually... I have a load which is installed when the processor is booted. Then I start my target server and launch ProfileScope. Everything is OK until then; on tree window, I can see the tasks that are working. There is some portion of the load that does not have debug symbols in, which is due to loader restrictions at the init time, so I need to load those symbols afterwards... I load them thru windshel and they show up on the target when I do a "moduleShow". Then I expect to see them on ProfileScope tree window when some code of that code is hit... I execute some tests which will hit those tasks, but there is no tick at ProfileScope!!? Any ideas? --------------------------- Newsgroups: comp.os.vxworks Subject: Download problem Date: Thu, 23 May 2002 05:18:07 GMT From: "Jamilur Rahman" Message-ID: It comipled and linked without any error. But unable to download to the target (prototyper). It gives download error (some symbols are missing). Any helpful hints would be appreciated. Here is the test code: #include void myReference() { int a = 100; // 'a' is a simple variable of type int int* p = &a; // 'p' is a pointer of type int int& n = a; // 'n' is a refernce of type int // Print the variable cout << "variable a = " << a << endl; // Print the pointer cout << "pointer p = " << p << endl; // Print the reference cout << "reference n = " << n << endl; } --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Download problem Date: Thu, 23 May 2002 08:46:13 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: Reply-To: "Michael Lawnick" Sender: mlawnick@p5088ff3e.dip.t-dialin.net Hi Jamilur, you will have to include an/some additional component(s) into your VxWorks. As long as you don't say which symbols are missing, we can't tell you which component(s). Michael "Jamilur Rahman" schrieb im Newsbeitrag news:jY_G8.74464$ah_.23066@news01.bloor.is.net.cable.rogers.com... > It comipled and linked without any error. But unable to download to the > target (prototyper). > It gives download error (some symbols are missing). Any helpful hints would > be appreciated. > > Here is the test code: > > #include > > void myReference() > { > int a = 100; // 'a' is a simple variable of type int > int* p = &a; // 'p' is a pointer of type int > int& n = a; // 'n' is a refernce of type int > > // Print the variable > cout << "variable a = " << a << endl; > > // Print the pointer > cout << "pointer p = " << p << endl; > > // Print the reference > cout << "reference n = " << n << endl; > } > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: chipDISK Date: Thu, 23 May 2002 08:59:07 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <3cebceb5.96066856@news.mhogaming.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088ff3e.dip.t-dialin.net Hi Ignacio, to access the chipDISK, you only need the JUMPTec-BSP with ATA and dosFs included. Take care to have dosFs2 installed from the beginning of your project to avoid later version changes when the first problems with old dosFs(1) will occure. JIDA is needed only for the board-specials like S/N, times of boot etc. HTH Michael Lawnick "Ignacio G.T." schrieb im Newsbeitrag news:3cebceb5.96066856@news.mhogaming.com... > I'm planning to use a PC/104 board from JUMPtec. I've been told by a > distributor that it comes with VxWorks drivers. So I've checked their > web site and found an Ethernet driver and a special set of SW modules > & API (called JIDA) to access the watchdog, the S/N and the like. > > My concern is that I want to use flash storage. They offer > chipDISK-IDE, that can be connected to the card through the IDE > connector. They claim this is "transparent" to the OS. So I think > that, using the standard PentiumPC BSP, adding the Ethernet driver and > the JIDA, and setting up an ATA-based DOS file system I could access > the chipDISK-IDE. Is this right? Or should I get/write special > drivers? > > I know the caveats about limited write-cycles on Flash, etc. That is > not important for my application. > > -- > Ignacio G.T. --------------------------- Newsgroups: comp.os.vxworks Subject: can't boot vxwork by network use intel82259 driver ? Date: 23 May 2002 00:26:30 -0700 From: sardine_sz@163.net (sardine) Organization: http://groups.google.com/ Message-ID: hi all gurus, i can boot vxworks by networks use intel82257(FEI), but ,after i replace system driver with intel82259(eeE),it cannot boot from network,and if i boot from ata,the driver intel82259 work fine. anyone know this problem ??? thx !!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where to download the patch for tornado? Date: 23 May 2002 07:31:42 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: sdssly@sina.com (sdssly) wrote: > However, I downloaded the Tornado from the free downlaod site. Can i > download the patch for it? What patch do you want to download? Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Fake a destructor on VxWorks Date: Thu, 23 May 2002 09:44:30 +0100 From: Danny Chapman Organization: Lucent Technologies, Columbus, Ohio Message-ID: References: <8bb5523.0205220540.56ec523f@posting.google.com> On 22 May 2002, Patrick Boulay wrote: > One of the big problem it's to call destructor in polymorphism like: > > A ptr; > > ptr = new B; > > delete ptr; You did make the destructer virtual in class A, didn't you?! (I take it that you mean "A * ptr;") --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with a dosFs Date: 23 May 2002 01:53:29 -0700 From: cikic.n@tecnomare.it (Cikic Nenad) Organization: http://groups.google.com/ Message-ID: Hi all. I have a problem with a Fujitsu SCSI disk (18 GB). My target board is PowerPC based and I am using Tornado 2.0.2. Following the VxWorks Programmers guide I have done a single partition of 1 GB to test if it is working. After assuring that I have effectively made a dosFs partition and that I can write and read from it I have done 9 partition of 2 GB. After booting I saw all partitions with one error. The first partition remained of 1GB. I suppose that is becasue I passed a NULL value to dosFsDevInit. However thinking that a format can do only good (like rebooting is never wrong :) ) I have done an ioctl call with 5 as argument. The disk started to do something but it blocked. After rebooting I can do nothing with my disk. No one partition work and dosFsDevInit returns error. Supplying directly the Config structure instead of NULL in dosFsDevInit I see my partitions with devs but I can not do nothing with them. I have tried to mount a different filesystem and to format but with no success. I have looked everywhere (SPR, this group messages, windriver support) with no success. Do someone have any suggesstion or a utility library for some low level formating of disk (the library the support gave me does not work)? Btw, I have seen that there exist dosFs and dosFs2 filesystem. I suppose I have the dosFs2 filesystem since I use Tornado 2.0.2, but I see everywhere only INCLUDE_DOSFS. How can I be sure I have dosFs2? Thanks in advance Cikic Nenad --------------------------- Newsgroups: comp.os.vxworks Subject: Race Conditions in Vxworks!!?? Date: 23 May 2002 08:59:36 GMT From: contact_senthil@yahoo.co.in Organization: NewsOne.Net - Free Usenet News via the Web - http://newsone.net/ Message-ID: Hi all, Do "race conditions" occur in vx works?? if YES how do you prevent it? thanks in advance.. Senthil. Senthil , DanLaw Technologies, #43 Sagar Soceity , Hyderabad, India- 500034. email: senthiln@danlawinc.com : 91-40-3542499. ----- Posted via NewsOne.Net: Free (anonymous) Usenet News via the Web ----- http://newsone.net/ -- Free reading and anonymous posting to 60,000+ groups NewsOne.Net prohibits users from posting spam. If this or other posts made through NewsOne.Net violate posting guidelines, email abuse@newsone.net --------------------------- Newsgroups: comp.os.vxworks Subject: Compression Date: Thu, 23 May 2002 06:16:33 -0400 From: "John B. Sampson" Message-ID: <3cecc149$1@news.xetron.com> Hi, Can anyone recommend a mature file compression & decompression package such as Pkzip, gzip, etc that runs on both VxWorks (MPC860 processor) and Windows? Would like to compress files on PC and decompress them on MPC860 and vice-versa. Thanks, John --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Race Conditions in Vxworks!!?? Date: Thu, 23 May 2002 06:33:19 -0400 From: joe durusau Organization: Bellsouth.Net Message-ID: <3CECC56F.12BD418D@bellsouth.net> References: contact_senthil@yahoo.co.in wrote: > Hi all, > > Do "race conditions" occur in vx works?? > if YES how do you prevent it? > > thanks in advance.. > Senthil. > > Senthil , > DanLaw Technologies, > #43 Sagar Soceity , > Hyderabad, > India- 500034. > > email: senthiln@danlawinc.com > : 91-40-3542499. > > ----- Posted via NewsOne.Net: Free (anonymous) Usenet News via the Web ----- > http://newsone.net/ -- Free reading and anonymous posting to 60,000+ groups > NewsOne.Net prohibits users from posting spam. If this or other posts > made through NewsOne.Net violate posting guidelines, email abuse@newsone.net This question is a bit general, but if you have multiple threads(tasks) using the same hardware or data structures, you can have conditions in which changes happen in the wrong order. In general, you could use tasklock() or semaphores to prevent this type of thing from happening. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: VME bus errors and signal handlers Date: 23 May 2002 03:34:53 -0700 From: cnw@zoom.co.uk (Chris Winstanley) Organization: http://groups.google.com/ Message-ID: <3b4d40d1.0205230234.1eaad47b@posting.google.com> We have a VME rack containing 3 Radstone PowerPC boards, each equiped with a Tundra Universe II. Each processor is running vxWorks. Our applications running on each processor are written mainly in Ada with some C using the Rational Apex Embedded product. One of the processors (call it P1) transfers data over the VME bus into the local memory of the other two processors (call them P2 and P3) and also reads data from the local memory of P2 and P3. The default bus configuration is posted writes. Our customer has placed a requirement on us to be able to reboot a processor without affecting the other 2 processors. We have found that rebooting either P2 or P3 causes a VME bus error (as P1 is trying to communicate with P2 and P3), resulting in an exception being raised on P1, suspending the communicating task (our applications, being safety related, are constrained to being a single task). My first question is: is it a write or a read (or both) to/from the VME bus by P1 that triggers the exception? To allow P1 to continue in the presence of a bus error the only solution I can think of is (depending upon the answer to my first question): a) P1 VME bus write causes exception: - - configure the bus to do synchronous (rather than posted) writes; - - attach a signal handler to SIGBUS that longjumps to immediately after the code that performs the write; - - in the code that performs the write, set the longjump location; - - given that you can't do a longjump in Ada, the Ada code will call some C code to do the write; - - the C code sets the longjump location; - - the signal handler will also have to be in C. b) P1 VME bus read causes exception: - - attach a signal handler to SIGBUS that longjumps to immediately after the code that performs the read; - - in the code that performs the read, set the longjump location; - - given that you can't do a longjump in Ada, the Ada code will call some C code to do the read; - - the C code sets the longjump location; - - the signal handler will also have to be in C. I haven't tried either of these solutions so I'm not sure whether they will work (in particular I'm not sure how the interaction between the Ada and the C will work out). Also turning off posted writes is going to have a serious performance impact. So my second question is: can anybody suggest an alternative solution? Thanks, Chris Winstanley --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu May 23 12:01:20 2002 From: Bruce Wilson Date: Thu May 23 12:01:22 PDT 2002 Subject: FEI ethernet flakey The loading of the vxWorks image hangs 90 percent of the time for our sa1110 processor (Bright Star Engineering nanoEngine board). We're running Tornado 2.1.1 for StrongARM Architecture, with a Sun Ultra60 host (Solaris 2.7). About once in 10 - 20 tries, the target successfully boots up, so we know the vxWorks image is built ok. When it fails, the last thing printed out is "Loading... 493536 +", and it never prints the remaining two sizes (when it works we see "493536 + 9292 + 32792"). We're using a 10 Mbit hub, so maybe the ethernet chip is mishandling its switching between 10/100 Mbit rate? Has anyone else seen such erratic behavior? Anybody found a solution? -- Bruce A. Wilson U.C. Lawrence Livermore National Lab P.O. Box 808 (L-278) Livermore, CA 94550 voice: (925) 422-4145 fax: (925) 423-1243 From vxwexplo-errs@csg.lbl.gov Thu May 23 13:08:53 2002 From: "CICON,JAMES (HP-Vancouver,ex1)" Date: Thu May 23 13:08:55 PDT 2002 Subject: RE: FEI ethernet flakey Suspect your nanoEngine board. We found similar problems that resolved to inadequate energy dissipation on the DRAM bus. Jim Cicon Hewlett Packard Company > Submitted-by vxwexplo-errs@csg.lbl.gov Thu May 23 12:01:20 2002 > Submitted-by: Bruce Wilson > > The loading of the vxWorks image hangs 90 percent of the time for our > sa1110 processor (Bright Star Engineering nanoEngine board). We're > running Tornado 2.1.1 for StrongARM Architecture, with a Sun Ultra60 > host (Solaris 2.7). > > About once in 10 - 20 tries, the target successfully boots up, so we > know the vxWorks image is built ok. When it fails, the last thing > printed out is "Loading... 493536 +", and it never prints the > remaining two sizes (when it works we see "493536 + 9292 + 32792"). > We're using a 10 Mbit hub, so maybe the ethernet chip is mishandling > its switching between 10/100 Mbit rate? > > Has anyone else seen such erratic behavior? Anybody found a solution? > -- > Bruce A. Wilson > U.C. Lawrence Livermore National Lab > P.O. Box 808 (L-278) > Livermore, CA 94550 > voice: (925) 422-4145 > fax: (925) 423-1243 > From vxwexplo-errs@csg.lbl.gov Thu May 23 21:09:08 2002 From: "Sudhakar Tadi" Date: Thu May 23 21:09:10 PDT 2002 Subject: VxWorks:memory mapped I/O This is a multi-part message in MIME format. ------=_NextPart_000_04CF_01C20306.8CFC7E80 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Trying one more time to get the help !!! Hi All, I am porting a Qlogic FC driver to VxWorks, for Intel X86 BSP. In=20 VxWorks for I/O mapped devices one can use sysInByte() sysOutByte() = etc.(corresponds to IN and OUT functions in Linux).=20 For memory maped devices, I understand that we have to edit=20 mmuPhysDesc[] table, if we want to support Virutal memory. My question is if I use memory mapped I/O what fucntions I can use=20 (like readb() writeb() in linux). OR Am I missing something from=20 VxWorks. Please advise. Has anybody ported linux SCSI drivers to VxWorks,if anyone wants to=20 share his experience, will be of great helpful to me. Thanks in Advance -Tadi Sudhakar Tadi HCL Technologies,Chennai, INDIA. ph: 91-44-3741939-42 ex:2312=20 Fax:91-44-3741038 http://san.hcltech.com ------=_NextPart_000_04CF_01C20306.8CFC7E80 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Trying one more time to get the help=20 !!!
Hi All,

    I am = porting a=20 Qlogic FC driver to VxWorks, for Intel X86 BSP. In
VxWorks for I/O = mapped=20 devices one can use sysInByte() sysOutByte() etc.(corresponds to IN and = OUT=20 functions in Linux).

For memory maped devices, I understand that = we have=20 to edit
mmuPhysDesc[] table, if we want to support Virutal = memory.
My=20 question is if I use memory  mapped I/O what fucntions I can use=20
(like   readb() writeb() in linux).  OR Am I missing=20 something from
VxWorks.  Please advise.

Has anybody = ported linux=20 SCSI drivers to VxWorks,if anyone wants to
share his experience, = will be of=20 great helpful to me.

Thanks in Advance
-Tadi
Sudhakar  Tadi
HCL = Technologies,Chennai,=20 INDIA.
ph: 91-44-3741939-42 ex:2312
Fax:91-44-3741038
http://san.hcltech.com
 ------=_NextPart_000_04CF_01C20306.8CFC7E80-- From vxwexplo-errs@csg.lbl.gov Thu May 23 21:34:34 2002 From: "Sudhakar Tadi" Date: Thu May 23 21:34:36 PDT 2002 Subject: VxWorks:I/O This is a multi-part message in MIME format. ------=_NextPart_000_0534_01C2030A.339D0540 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Hi All, I am porting a Qlogic FC driver to VxWorks, for Intel X86 BSP. In = =3D VxWorks for I/O mapped devices one can use sysInByte() sysOutByte() =3D etc.(corresponds to IN and OUT functions in Linux). =3D20 For memory maped devices, I understand that we have to edit =3D mmuPhysDesc[] table, if we want to support Virutal memory. My question is if I use memory mapped I/O what fucntions I can use =3D (like readb() writeb() in linux). OR Am I missing something from =3D VxWorks. Please advise. Has anybody ported linux SCSI drivers to VxWorks,if anyone wants to =3D share his experience, will be of great helpful to me. Thanks in Advance -Tadi ------=_NextPart_000_0534_01C2030A.339D0540 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Hi All,

    I am = porting a=20 Qlogic FC driver to VxWorks, for Intel X86 BSP. In =3D
VxWorks for = I/O mapped=20 devices one can use sysInByte() sysOutByte() =3D
etc.(corresponds to = IN and OUT=20 functions in Linux). =3D20

For memory maped devices, I understand = that we=20 have to edit =3D
mmuPhysDesc[] table, if we want to support Virutal=20 memory.
My question is if I use memory  mapped I/O what = fucntions I can=20 use =3D
(like   readb() writeb() in linux).  OR Am I = missing=20 something from =3D
VxWorks.  Please advise.

Has anybody = ported=20 linux SCSI drivers to VxWorks,if anyone wants to =3D
share his = experience, will=20 be of great helpful to me.

Thanks in=20 Advance
-Tadi

------=_NextPart_000_0534_01C2030A.339D0540-- From vxwexplo-errs@csg.lbl.gov Fri May 24 04:03:20 2002 From: Vxworks Exploder Date: Fri May 24 04:03:22 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri May 24 04:03:15 PDT 2002 Subject: Re: IDE controller Subject: ENETUNREACH error Subject: free stack and TCB before taskDelete Subject: Re: Compression Subject: Re: using zlib to decompress images created by deflate Subject: Re: Full VxSim vs Target Subject: Re: free stack and TCB before taskDelete Subject: Re: free stack and TCB before taskDelete Subject: Hello ... gcc error fatal signal 125 Subject: Hello ... gcc error fatal signal 125 Subject: Hello ... gcc error fatal signal 125 Subject: Re: Race Conditions in Vxworks!!?? Subject: Re: free stack and TCB before taskDelete Subject: Re: Compression Subject: netTupleGet Subject: Re: IDE controller Subject: Re: GCC: Max file size ? Subject: Re: How can I uniquely identify a file in dosFs Subject: Re: VME bus errors and signal handlers Subject: Re: GCC: Max file size ? Subject: Re: Hello ... gcc error fatal signal 125 Subject: Re: free stack and TCB before taskDelete Subject: NPT Driver Questions Subject: Re: Problem with a dosFs Subject: Re: VME bus errors and signal handlers Subject: Re: VME bus errors and signal handlers Subject: Re: GCC: Max file size ? Subject: Re: why do system calls hang? Subject: builtin sim Subject: Re: Problem with a dosFs Subject: Re: Race Conditions in Vxworks!!?? Subject: sharing memory in p3 processor by Tornado and VC Subject: Re: sharing memory in p3 processor by Tornado and VC Subject: newbie question ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: IDE controller Date: 23 May 2002 05:19:23 -0700 From: be_gyus@yahoo.com (james) Organization: http://groups.google.com/ Message-ID: References: "Leonid Rosenboim" wrote in message news:... > IDE is a standard interface for connecting storage devices. > It has historically evolved around the PC architecture, and is > essentially a 16-bit wide parallel bus used for connecting > hard disks (ATA command set) and CD-ROM, CD-RW, DVD > and also various tape drives (ATAPI command set). > > Each IDE controller typically has two chanells (each channel > has a separate cable) which can support up to two drives, > one in "master" mode, one in "slave" mode. > > If you need more info, you have got to be more specific. > > "james" wrote in message > news:ff42611f.0205210318.16b407f6@posting.google.com... > > what is ide controller and its significance ? ************************************************************************* Can anyone brief about the primary / secondary IDE controller ? --------------------------- Newsgroups: comp.os.vxworks Subject: ENETUNREACH error Date: 23 May 2002 05:45:29 -0700 From: ganapathy.subramanian@wipro.com (Gana) Organization: http://groups.google.com/ Message-ID: Hi, when i try streaming packets to a multicast group(multicast ipaddress ) i receive ENETUNREACH error from 'sendto' call. i'm using vxworks 5.4, Tornado 2.0 . Have enabled MCAST routing in build options. And our network driver supports multicasting. Even after adding proper routes, same error persists. when tried to set IP_MULTICAST_IF option to the socket faced EADDRNOTAVAIL error. Can anyone provide info on this. Is any other routing protocol to be included in build option? Thanks Gana --------------------------- Newsgroups: comp.os.vxworks Subject: free stack and TCB before taskDelete Date: Thu, 23 May 2002 14:52:48 +0200 From: Arjan Message-ID: <3CECE620.9000304@philips.com> Hi I'm working on some routines that create start and exit VxWorks tasks. In this situation the taskDelete() is only called by the task that has to be deleted itself. It seems that the taskDelete() doesn't free the allocated memory for the stack and TCB in case the task is created with taskInit(). Is this a know problem or a bug ? The VxWorks documentation explicitly describes it should, but perhaps only if the task is created with taskSpawn() (?) Therefore I have to free it myself with the following piece of code: - --- taskHandle = taskIdSelf(); pTcb = taskTcb((Int32)taskHandle) #if (_STACK_DIR == _STACK_GROWS_DOWN) free(pTcb->pStackEnd); #else free(pTcb->pStackBase); #endif free(pTcb); taskDelete(taskHandle); - --- This rises some questions: - is it legal to delete a task of which the stack and TCB don't exist anymore - or should the memory be freed after the taskDelete (probably not possible since then the task does't exit anymore) - is it legal to check the return value of taskDelete() (idem) regards, Arjan --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compression Date: Thu, 23 May 2002 13:18:52 GMT From: "tim michals" Organization: AT&T Broadband Message-ID: <0%5H8.11616$352.1566@sccrnsc02> References: <3cecc149$1@news.xetron.com> Yes, VxWorks has a tool called deflate this creates a compressed image on your compiling machine, then you can use inflate (inflateLib) to expand this image on your hardware platform. Vxworks creates a makefile using Tornado to create a compressed image VxWorksZ.bin "John B. Sampson" wrote in message news:3cecc149$1@news.xetron.com... > Hi, > > Can anyone recommend a mature file compression & decompression package such > as > Pkzip, gzip, etc that runs on both VxWorks (MPC860 processor) and Windows? > Would like to compress files on PC and decompress them on MPC860 and > vice-versa. > > Thanks, > > John > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: using zlib to decompress images created by deflate Date: Thu, 23 May 2002 13:20:05 GMT From: "tim michals" Organization: AT&T Broadband Message-ID: <906H8.8977$ux5.8851@rwcrnsc51.ops.asp.att.net> References: Found out the issue, corrupted compress image, overlapped tow memory regions, thanks for everyone's help "tim michals" wrote in message news:owAG8.52897$L76.66814@rwcrnsc53... > Thanks, > > I'm trying to create the same functionality as a VxWorks loader does to > expand compressed images created by deflate. > So, I'm trying to create the inflate, using the correct dictionary, etc as > inflate and deflate used in VxWorks. > > "Bill Pringlemeir" wrote in message > news:u661hfn8a.fsf@sympatico.ca... > > >>>>> "tim" == tim michals writes: > > > > tim> All, Yes, another question about deflate! Ok, I downloaded the > > tim> latest zlib, create a small file using deflate < infile > > > tim> fileZ.bin Then here is a snippet of code to uncompress the > > tim> image: > > > > [snip code] > > > > tim> But an error returns "bad uncompress' but yet all of the file is > > tim> decompressed into the buffer. So it is working somewhat. What > > tim> could I be doing wrong? > > > > The deflate program might not work with the version of zlib you > > retrieved (and specifically the functions you call). It is common for > > archives to contain header information. The deflate might have a > > dictionary included, etc. I suggest that you build minigzip (or > > whatever) and use that program to compress the image on your host. > > The code you supplied should also be able to decompress it on the > > host. It is usually easier to debug the compress/decompress on the > > host first and then move to your target. > > > > I have used the gzio [I think] that gives a stream interface to files. > > I used this in a `COFF' loader to allow compresses .out files to be > > loaded. You probably want a different interface for boot roms. > > > > hth, > > Bill Pringlemeir. > > > > -- > > Have you ever tasted spinach and thought it was Doritos? Or had your > > dog make you feel warm and fuzzy while being chopped to pieces by a > > rotor blade? Or thought it was unhealthy to have two modems to defend > > against the artist? You will. And the company that will bring it to > > you: AT&T. > > > > vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Full VxSim vs Target Date: Tue, 21 May 2002 15:14:30 -0400 From: "Bhaskar Rao" Message-ID: <3cecf077$0$17591$724ebb72@reader2.ash.ops.us.uu.net> References: <5947a04.0205191907.3ae9c062@posting.google.com> Hi Rikard, I too faced the same problem and when I contacted windriver they said it is limitation of vxsim, currently(5months back) they don't support broadcast on vxsim. - -Bhaskar "Rikard" wrote in message news:5947a04.0205191907.3ae9c062@posting.google.com... > Hello, > > I am using full VxSim on a PPC installation. Now I want to broadcast > messages. > Would anyone know how to get the ULIP network driver for VxSim to > support exchange of BROADCAST network packages? I've implemented a > simple network service discovery protocol based on UDP broadcasts. On > VxSim, my server is unable to receive the broadcast packages that my > clients sent out. On VxSim I did an ifShow in Windsh and it showed: > > nt (unit number 0): > Flags: (0xf1) UP POINT-TO-POINT RUNNING > Type: ETHERNET_CSMACD > blah ...blah ... blah > 0 packets received; 1 packet sent > blah ...blah ... blah > > I think that's the reason I can't receive the broadcast packages. > > On a target with ifShow printing: > > symeth (unit number 0): > Flags: UP ARP BROADCAST RUNNING > blah ...blah ... blah > > ,here BROADCAST is running and the server is able to receive my > clients' broadcast messages properly. > > Can I get VxSim with BROADCAST or is this a limitation of VxSim? > > Regards > Rikard --------------------------- Newsgroups: comp.os.vxworks Subject: Re: free stack and TCB before taskDelete Date: Thu, 23 May 2002 16:26:35 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <3CECE620.9000304@philips.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088ff3e.dip.t-dialin.net Hi Arjan, I can't give you an answer to your question below, but what the hell are you doing ? You almost never should need a taskDelete() except for debug. Let your task terminate by leaving the outermost function ! Not knowing what you really do, your post reads to me like trying to push a nail through your knee. I'm pretty sure there exist easier mechanisms to do what you want to. Regards, Michael "Arjan" schrieb im Newsbeitrag news:3CECE620.9000304@philips.com... > Hi > > I'm working on some routines that create start and exit VxWorks tasks. > In this situation the taskDelete() is only called by the task that has > to be deleted itself. > It seems that the taskDelete() doesn't free the allocated memory for > the stack and TCB in case the task is created with taskInit(). > Is this a know problem or a bug ? > The VxWorks documentation explicitly describes it should, but perhaps > only if the task is created with taskSpawn() (?) > Therefore I have to free it myself with the following piece of code: > > --- > taskHandle = taskIdSelf(); > pTcb = taskTcb((Int32)taskHandle) > > #if (_STACK_DIR == _STACK_GROWS_DOWN) > free(pTcb->pStackEnd); > #else > free(pTcb->pStackBase); > #endif > free(pTcb); > > taskDelete(taskHandle); > --- > > This rises some questions: > - is it legal to delete a task of which the stack and TCB > don't exist anymore > - or should the memory be freed after the taskDelete > (probably not possible since then the task does't exit anymore) > - is it legal to check the return value of taskDelete() (idem) > > > regards, > Arjan > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: free stack and TCB before taskDelete Date: Thu, 23 May 2002 17:04:57 +0200 From: Arjan Message-ID: <3CED0519.6090602@philips.com> References: <3CECE620.9000304@philips.com> Hi Michael I think that's a logical response. But when I just let it run out of it's main loop the stack and TCB memory are not freed :( (Note that my tasks are created with taskInit istead of taskSpawn and therefore I had to allocate the stack and TCB myself ) I'm not sure if this is a bug, but I'm very unhappy with it .. regs, Arjan Michael Lawnick wrote: > Hi Arjan, > > I can't give you an answer to your question below, but what the hell are you > doing ? > You almost never should need a taskDelete() except for debug. Let your task > terminate by leaving the outermost function ! > Not knowing what you really do, your post reads to me like trying to push a > nail through your knee. > I'm pretty sure there exist easier mechanisms to do what you want to. > > Regards, > Michael > > "Arjan" schrieb im Newsbeitrag > news:3CECE620.9000304@philips.com... > >>Hi >> >>I'm working on some routines that create start and exit VxWorks tasks. >>In this situation the taskDelete() is only called by the task that has >>to be deleted itself. >>It seems that the taskDelete() doesn't free the allocated memory for >>the stack and TCB in case the task is created with taskInit(). >>Is this a know problem or a bug ? >>The VxWorks documentation explicitly describes it should, but perhaps >>only if the task is created with taskSpawn() (?) >>Therefore I have to free it myself with the following piece of code: >> >>--- >>taskHandle = taskIdSelf(); >>pTcb = taskTcb((Int32)taskHandle) >> >>#if (_STACK_DIR == _STACK_GROWS_DOWN) >> free(pTcb->pStackEnd); >>#else >> free(pTcb->pStackBase); >>#endif >> free(pTcb); >> >>taskDelete(taskHandle); >>--- >> >>This rises some questions: >> - is it legal to delete a task of which the stack and TCB >> don't exist anymore >> - or should the memory be freed after the taskDelete >> (probably not possible since then the task does't exit anymore) >> - is it legal to check the return value of taskDelete() (idem) >> >> >>regards, >>Arjan >> > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Hello ... gcc error fatal signal 125 Date: Thu, 23 May 2002 15:37:51 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: I am using gcc provided with MPC860 and T2 ... there is no patches availible on the vxworks www site except for ppc 74xx ... and I am seeing a really terrible horrible bad error compiling a huge "auto generated file" (which compiles fine with *.c extension but not *.cpp) that says fatal signal 125 ... Any help would be great ... what compiler bug is this ... is there any work arounds ... Thanks Jim C:\>gcc -v gcc driver version cygnus-2.7.2-960126 egcs-971225 tornado 2.0 executing gcc version cygnus-2.7.2-960126 --------------------------- Newsgroups: comp.os.vxworks Subject: Hello ... gcc error fatal signal 125 Date: Thu, 23 May 2002 15:41:20 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: I am using gcc provided with MPC860 and T2 ... there is no patches availible on the vxworks www site except for ppc 74xx ... and I am seeing a really terrible horrible bad error compiling a huge "auto generated file" (which compiles fine with *.c extension but not *.cpp) that says fatal signal 125 ... Any help would be great ... what compiler bug is this ... is there any work arounds ... Thanks Jim C:\>gcc -v gcc driver version cygnus-2.7.2-960126 egcs-971225 tornado 2.0 executing gcc version cygnus-2.7.2-960126 --------------------------- Newsgroups: comp.os.vxworks Subject: Hello ... gcc error fatal signal 125 Date: Thu, 23 May 2002 15:40:44 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <048H8.25084$f5.2025938@news> I am using gcc provided with MPC860 and T2 ... there is no patches availible on the vxworks www site except for ppc 74xx ... and I am seeing a really terrible horrible bad error compiling a huge "auto generated file" (which compiles fine with *.c extension but not *.cpp) that says fatal signal 125 ... Any help would be great ... what compiler bug is this ... is there any work arounds ... Thanks Jim C:\>gcc -v gcc driver version cygnus-2.7.2-960126 egcs-971225 tornado 2.0 executing gcc version cygnus-2.7.2-960126 --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Race Conditions in Vxworks!!?? Date: Thu, 23 May 2002 11:19:25 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3CED168D.4E407998@fnal.gov> References: Senthil, Race conditions do occur. The most famous case was in pathfinder that was sent to mars. Priority inversion had taken care of that. Search web for that (or even comp.os.vxworks archives) for more details for that. Dinker --------------------------- Newsgroups: comp.os.vxworks Subject: Re: free stack and TCB before taskDelete Date: Thu, 23 May 2002 12:54:12 -0400 From: Keith Arner Organization: Marconi Message-ID: References: <3CECE620.9000304@philips.com> Reply-To: Keith Arner On Thu, 23 May 2002, Arjan wrote: > I'm working on some routines that create start and exit VxWorks tasks. > In this situation the taskDelete() is only called by the task that has > to be deleted itself. > It seems that the taskDelete() doesn't free the allocated memory for > the stack and TCB in case the task is created with taskInit(). > Is this a know problem or a bug ? > The VxWorks documentation explicitly describes it should, but perhaps > only if the task is created with taskSpawn() (?) > Therefore I have to free it myself with the following piece of code: > > --- > taskHandle = taskIdSelf(); > pTcb = taskTcb((Int32)taskHandle) > > #if (_STACK_DIR == _STACK_GROWS_DOWN) > free(pTcb->pStackEnd); > #else > free(pTcb->pStackBase); > #endif > free(pTcb); > > taskDelete(taskHandle); > --- > > This rises some questions: > - is it legal to delete a task of which the stack and TCB > don't exist anymore > - or should the memory be freed after the taskDelete > (probably not possible since then the task does't exit anymore) > - is it legal to check the return value of taskDelete() (idem) If you use taskInit() and allocate the stack yourself, then you are responsible for freeing it up yourself. This follows the general principle of "If you allocate it, free it. If you open it, close it. If you (do something), (undo it)." If you call free() on your stack while the task is still running, you are asking for trouble. After you free that memory, you no longer own the memory, so you should not operate on it (this includes stack operations within it). Why do you need to use taskInit(), rather than taskSpawn() anyway? If what you are trying to accomplish is the ability to create a task, but not have it start running quite yet, you can achieve (almost) exactly the same thing by having the task take a binary semaphore right out of the gate, and not give the semaphore until it's time for it to do it's thing. Keith - -- mv ~karner /loony/bin --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Compression Date: 23 May 2002 10:22:20 -0700 From: pkockritz@mac.com (Pete Kockritz) Organization: http://groups.google.com/ Message-ID: <9423c55b.0205230922.2b094916@posting.google.com> References: <3cecc149$1@news.xetron.com> "John B. Sampson" wrote in message news:<3cecc149$1@news.xetron.com>... > Hi, > > Can anyone recommend a mature file compression & decompression package such > as > Pkzip, gzip, etc that runs on both VxWorks (MPC860 processor) and Windows? > Would like to compress files on PC and decompress them on MPC860 and > vice-versa. You should be able to compile gzip for vxWorks (maybe with a little tweaking). I haven't seen a packaged version of it for vxWorks, though. Alternatively, you could 'roll your own' using zlib as the compression engine. Pete - -- pkockritz at mac dot com --------------------------- Newsgroups: comp.os.vxworks Subject: netTupleGet Date: Thu, 23 May 2002 13:47:20 -0400 From: George Varndell Organization: Posted Via Newsfeeds.com = SPEED+RETENTION+COMPLETION = http://www.newsfeeds.com Message-ID: <3CED2B28.92E1BEE@ineoquest.com> Hi, Has anyone profiled the netbuflib function netTupleGet to compare its cost to the sum of the constituent calls, netMblkGet, netClBlkGet, netClBlkJoin, netMblkClJoin, etc.? I wonder if it's worth bothering to change code to use this single function instead of the 5 or 6 individual functions that accomplish the same thing; but I also have to wonder why I've never seen a single Wind River driver that uses it. Perhaps it has bugs -- leak prone maybe? Thanks, George Varndell - -----------== Posted via Newsfeed.Com - Uncensored Usenet News ==---------- http://www.newsfeed.com The #1 Newsgroup Service in the World! - -----= Over 100,000 Newsgroups - Unlimited Fast Downloads - 19 Servers =----- --------------------------- Newsgroups: comp.os.vxworks Subject: Re: IDE controller Date: Thu, 23 May 2002 18:05:35 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: "james" wrote in message > Can anyone brief about the primary / secondary IDE controller ? It is just the standard nomenclature of the IDE/ATA standard, meaning first (primary) and second (secondary) IDE chanells. Standard PC architecture (i.e. ISA) has port/irq definitions for two IDE chanells, and some BIOS related ramifications. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GCC: Max file size ? Date: Thu, 23 May 2002 18:08:25 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CED3086.6070203@spamm.me.l8s.co.uk> References: <5bRG8.24732$f5.2005649@news> bill@woola.qswtools.com wrote: > In article <5bRG8.24732$f5.2005649@news>, > "Jim" writes: > >>Hello, >>Does gcc have a max source file size ... say 100 000 statements ... >> > Nope. But the coff object format (used by some (if not all) vxworks systems) has a limit of 64k relocations per segment. I really bit file might hit that. Also gcc will have a large memory requirement for large files. David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I uniquely identify a file in dosFs Date: Thu, 23 May 2002 18:15:42 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Ok, Dos file systems dont have i-nodes.; A Dos file can be uniquely identified by: 1. the number of first cluster of file - could work except 0-size files do not have any cluster 2. Cluster numjber + offset of directory entry where the file name (aka Alias) is written. But then again, DosFs does not have "hard links", and I wonder what is so wrong with simply performing a case-insensitive comparison of full path file names ? I wonder what I am missing here. - - Leonid "Parangodath, Binish [SC7:326:EXCH]" wrote in message news:ach5nm$nrt$1@bcarh8ab.ca.nortel.com... > Hi , > > I want to uniquely identify a file in dosFs.I tried to get stat structure of > two files and comapred the st_dev member and st_ino.To my astonishment it is > the same for two different files in my system.Is there any better way than a > string comparison of filenames with full path > > Regards, > Binish > > --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: VME bus errors and signal handlers Date: Thu, 23 May 2002 18:23:19 GMT From: junk@junk.net (Anton Erasmus) Organization: The South African Internet Exchange Message-ID: <3ced3034.2014104@news.saix.net> References: <3b4d40d1.0205230234.1eaad47b@posting.google.com> Reply-To: antera@intekom.co.za On 23 May 2002 03:34:53 -0700, cnw@zoom.co.uk (Chris Winstanley) wrote: >We have a VME rack containing 3 Radstone PowerPC boards, each equiped >with a Tundra Universe II. Each processor is running vxWorks. Our >applications running on each processor are written mainly in Ada with >some C using the Rational Apex Embedded product. > >One of the processors (call it P1) transfers data over the VME bus >into the local memory of the other two processors (call them P2 and >P3) and also reads data from the local memory of P2 and P3. The >default bus configuration is posted writes. > >Our customer has placed a requirement on us to be able to reboot a >processor without affecting the other 2 processors. > >We have found that rebooting either P2 or P3 causes a VME bus error >(as P1 is trying to communicate with P2 and P3), resulting in an >exception being raised on P1, suspending the communicating task (our >applications, being safety related, are constrained to being a single >task). > >My first question is: is it a write or a read (or both) to/from the >VME bus by P1 that triggers the exception? > >To allow P1 to continue in the presence of a bus error the only >solution I can think of is (depending upon the answer to my first >question): > >a) P1 VME bus write causes exception: >- configure the bus to do synchronous (rather than posted) writes; >- attach a signal handler to SIGBUS that longjumps to immediately >after the code that performs the write; >- in the code that performs the write, set the longjump location; >- given that you can't do a longjump in Ada, the Ada code will call >some C code to do the write; >- the C code sets the longjump location; >- the signal handler will also have to be in C. > >b) P1 VME bus read causes exception: >- attach a signal handler to SIGBUS that longjumps to immediately >after the code that performs the read; >- in the code that performs the read, set the longjump location; >- given that you can't do a longjump in Ada, the Ada code will call >some C code to do the read; >- the C code sets the longjump location; >- the signal handler will also have to be in C. > >I haven't tried either of these solutions so I'm not sure whether they >will work (in particular I'm not sure how the interaction between the >Ada and the C will work out). > >Also turning off posted writes is going to have a serious performance >impact. > >So my second question is: can anybody suggest an alternative solution? > >Thanks, >Chris Winstanley Hi, I am not that familiar with the Power PC architecture or the Tundra Univers II chips. I normally use 68XXX boards when working with VME systems. If the Power PC is the same s the 68K, and I can not see why not, the both read and write will cause exceptions. The card is waiting for DTACK to be asserted, and if that is not done within a certain time, a BUS Error will occur. On the 68K, a BUS Error causes the CPU to put extra information on the stack, that you can use to restart the instruction that caused the problem. In any case you have to clean up this stack first before you can continue your application code. On some of the VME Interface chips, it is possible to set the interface chip to generate a local DTACK signal in stead of a BUS Error when you write to a not existing device. If you can set the Tundra Universe II chips up to do this, then you can have a location on the other cards that is written with a "Magic" value on startup. You can the every so often check if this "Magic" value is still valid. If not it means that the other cards is not in the system. Hope this helps Regards Anton Erasmus --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GCC: Max file size ? Date: Thu, 23 May 2002 18:36:53 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <9FaH8.25148$f5.2031012@news> References: <5bRG8.24732$f5.2005649@news> <3CED3086.6070203@spamm.me.l8s.co.uk> Gcc should ok for memory ... I have a large hard drive some plenty of swapping room ;-) As for the coff file ... how can i tell this is the problem ... My gcc is barfing with "fatal signal 125" ... is this the sign ... i am having a hard time finding the meaning of this signal ... so it means i have a hard time finding a work ... around ... Thanks for the help so far, Jim "David Laight" wrote in message news:3CED3086.6070203@spamm.me.l8s.co.uk... > bill@woola.qswtools.com wrote: > > > In article <5bRG8.24732$f5.2005649@news>, > > "Jim" writes: > > > >>Hello, > >>Does gcc have a max source file size ... say 100 000 statements ... > >> > > Nope. > > > But the coff object format (used by some (if not all) vxworks systems) > has a limit of 64k relocations per segment. > I really bit file might hit that. > > Also gcc will have a large memory requirement for large files. > > David > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Hello ... gcc error fatal signal 125 Date: Thu, 23 May 2002 18:43:17 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <9LaH8.25149$f5.2031051@news> References: <718H8.25077$f5.2025916@news> ccppc: Internal compiler error: program cc1plus got fatal signal 125 make[4]: *** [C:/root/AirOE/PlatformServices/BuiltInSelfTest/sb2100BistRevD/Sb2100RevDApp /sb7015a2Test.o] Error 0x1 "Jim" wrote in message news:718H8.25077$f5.2025916@news... > I am using gcc provided with MPC860 and T2 ... there is no patches availible > on the vxworks www site except for ppc 74xx ... and I am seeing a really > terrible horrible bad error compiling a huge "auto generated file" (which > compiles fine with *.c extension but not *.cpp) that says fatal signal 125 > ... > > Any help would be great ... what compiler bug is this ... is there any work > arounds ... > > Thanks > Jim > > C:\>gcc -v > gcc driver version cygnus-2.7.2-960126 egcs-971225 tornado 2.0 executing gcc > version cygnus-2.7.2-960126 > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: free stack and TCB before taskDelete Date: Thu, 23 May 2002 11:52:05 -0700 From: M Lang Organization: Jet Propulsion Laboratory - Pasadena CA Message-ID: <3CED3A55.2B041622@jpl.nasa.gov> References: <3CECE620.9000304@philips.com> Reply-To: Minh.Lang@jpl.nasa.gov Arjan, When you used "taskInit(...)" instead of "taskSpawn(...)" you supply/allocate your own TCB and stack spaces to taskInit() and it just uses whatever addresses you told it to use. It doesn't allocate spaces for TCB or stack. By default, task spawned with taskInit() doesn't deallocate TCB & stack when you call taskDelete() on it, unless you specify VX_DEALLOC_STACK as option when you create the task (taskSpawn() automatically set this option but taskInit() doesn't). Take a look at the options in taskLib.h Hope that this helps. -minh- Arjan wrote: > > Hi > > I'm working on some routines that create start and exit VxWorks tasks. > In this situation the taskDelete() is only called by the task that has > to be deleted itself. > It seems that the taskDelete() doesn't free the allocated memory for > the stack and TCB in case the task is created with taskInit(). > Is this a know problem or a bug ? > The VxWorks documentation explicitly describes it should, but perhaps > only if the task is created with taskSpawn() (?) > Therefore I have to free it myself with the following piece of code: > > --- > taskHandle = taskIdSelf(); > pTcb = taskTcb((Int32)taskHandle) > > #if (_STACK_DIR == _STACK_GROWS_DOWN) > free(pTcb->pStackEnd); > #else > free(pTcb->pStackBase); > #endif > free(pTcb); > > taskDelete(taskHandle); > --- > > This rises some questions: > - is it legal to delete a task of which the stack and TCB > don't exist anymore > - or should the memory be freed after the taskDelete > (probably not possible since then the task does't exit anymore) > - is it legal to check the return value of taskDelete() (idem) > > regards, > Arjan --------------------------- Newsgroups: comp.os.vxworks Subject: NPT Driver Questions Date: 23 May 2002 13:23:33 -0700 From: antonio.mira@alcatel.com (Antonio Mira) Organization: http://groups.google.com/ Message-ID: <312cf52.0205231223.305cb9d@posting.google.com> Hello all, , I'm writing an NPT ( not an END ) non-Ethernet based driver and I'm experiencing some problems. I'm using Tornado 2.0.2 with the cp-4 plus the patch 68322. Right now I'm using the VxSim, trying to make work an initial version ( for the IP stack this shouldn't be any problem ) 1) I tried to put a custom Address Resolution Function using muxAddrResFuncAdd . My driver is a bussed HDLC, but if I try to give M2_ifType_hdlc as ifType argument , I get an error in AddrResFuncAdd(M2_ifType_hdlc, 0x800, myFunc) So, I finally had to use M2_ifType_other instead. When the function I provided is called, I copy the address in the buffer given as an argument, and the program crashes with a segmentation fault. Is not any memory being allocated for that buffer ? Is that result of having that M2_ifType_other associated ? 2) In the receive routine, just before giving the mBlk to the upper layer, the mBlkHdr has to be updated. Usually something like this : pMblk->mBlkHdr.mFlags |= M_PKTHDR; pMblk->mBlkHdr.mData = pCluster ; pMblk->mBlkHdr.mLen = lenCluster ; pMblk->mBlkPktHdr.len = pMblk->mBlkHdr.mLen; But my question is. Should the mData point to the beginning of the Cluster ( Layer 2 + IP Header + ... ) or to the IP Header skipping the Layer 2? The NPT Programmers Guide, Ed4 page 99, states: " mBlkPktHdr .... Drivers attached to IP using ipAttach() must set mBlkPktHdr.len so that the IP receive routine can locate the IP Header" What this exactly means ? To take out the header length from the total length and then store the result in that field ? Do the fields pMblk->mBlkHdr.mLen and Mblk->mBlkHdr.mLen represent the same ? The results I'm experiencing is no activity at all in the upper layers when I call the END_RCV_RTN_CALL routine with a well formed mBlk. The test that I'm doing is a ping , I checked the incoming ICMP packets and are OK. But I don't get anything, not even an error message. Is there something that I'm missing ? Why IP is ignoring all the packets silently? Any help would be appreciated. Thanks in advance, Antonio Mira --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with a dosFs Date: 23 May 2002 13:49:06 -0700 From: jzhang@drs.ca (Jason) Organization: http://groups.google.com/ Message-ID: References: Hello, Ask your wind river's representative for the DoSFs2 installation disk. But that is not enough, because it still uses some Dosfs 1 library, so you have to ask them again to send you a pathc for DosFs 2. good luck cikic.n@tecnomare.it (Cikic Nenad) wrote in message news:... > Hi all. > > I have a problem with a Fujitsu SCSI disk (18 GB). My target board is > PowerPC based and I am using Tornado 2.0.2. > > Following the VxWorks Programmers guide I have done a single partition > of 1 GB to test if it is working. After assuring that I have > effectively made a dosFs partition and that I can write and read from > it I have done 9 partition of 2 GB. After booting I saw all partitions > with one error. The first partition remained of 1GB. I suppose that is > becasue I passed a NULL value to dosFsDevInit. However thinking that a > format can do only good (like rebooting is never wrong :) ) I have > done an ioctl call with 5 as argument. The disk started to do > something but it blocked. After rebooting I can do nothing with my > disk. No one partition work and dosFsDevInit returns error. Supplying > directly the Config structure instead of NULL in dosFsDevInit I see my > partitions with devs but I can not do nothing with them. I have tried > to mount a different filesystem and to format but with no success. > > I have looked everywhere (SPR, this group messages, windriver support) > with no success. > > Do someone have any suggesstion or a utility library for some low > level formating of disk (the library the support gave me does not > work)? > > Btw, I have seen that there exist dosFs and dosFs2 filesystem. I > suppose I have the dosFs2 filesystem since I use Tornado 2.0.2, but I > see everywhere only INCLUDE_DOSFS. How can I be sure I have dosFs2? > > Thanks in advance > > > Cikic Nenad --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: VME bus errors and signal handlers Date: Thu, 23 May 2002 22:36:57 GMT From: "Jeffrey Creem" Organization: ATT Broadband Message-ID: References: <3b4d40d1.0205230234.1eaad47b@posting.google.com> "Chris Winstanley" wrote in message news:3b4d40d1.0205230234.1eaad47b@posting.google.com... > We have a VME rack containing 3 Radstone PowerPC boards, each equiped > with a Tundra Universe II. Each processor is running vxWorks. Our > applications running on each processor are written mainly in Ada with > some C using the Rational Apex Embedded product. > On other versions of Rational products, you could trap read bus errors as Storage_Error exceptions (Green hills mapped them to Constraint_Errors I think).. Writes did not map back to exceptions since they can be posted so when they happen, you do not know for sure which task to "signal". For your requirements, I'd look into using vxMemProbe to do the bus transfers for small transfers and use the DMA engine to do larger transfers (assuming there is one). You can check the return status of vxMemProbe to see if there are bus errors happening if you need to recover/reinitialize and for the DMA controller there is usually some way of finding out if the DMA completed with or without errors. --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: VME bus errors and signal handlers Date: Thu, 23 May 2002 20:25:53 -0400 From: joe durusau Organization: WEBUSENET.com Message-ID: <3CED8891.78C4FAC8@bellsouth.net> References: <3b4d40d1.0205230234.1eaad47b@posting.google.com> Chris Winstanley wrote: > <> > One of the processors (call it P1) transfers data over the VME bus > into the local memory of the other two processors (call them P2 and > P3) and also reads data from the local memory of P2 and P3. The > default bus configuration is posted writes. > > Our customer has placed a requirement on us to be able to reboot a > processor without affecting the other 2 processors. > > We have found that rebooting either P2 or P3 causes a VME bus error > (as P1 is trying to communicate with P2 and P3), resulting in an > exception being raised on P1, suspending the communicating task (our > applications, being safety related, are constrained to being a single > task). > <> One problem that I've been told exists with most powerPC CPUs is that due to internal pipelining they cannot recover the exact machine state when a bus error occurs. Perhaps your boards do not suffer from this problem, but I think that you should check with the manufacturer to see whether recovery from a bus error can be done. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GCC: Max file size ? Date: Fri, 24 May 2002 03:12:52 GMT From: Pete Kockritz Organization: AT&T Broadband Message-ID: References: <5bRG8.24732$f5.2005649@news> <3CED3086.6070203@spamm.me.l8s.co.uk> <9FaH8.25148$f5.2031012@news> In article <9FaH8.25148$f5.2031012@news>, "Jim" wrote: > Gcc should ok for memory ... I have a large hard drive some plenty of > swapping room ;-) > > As for the coff file ... how can i tell this is the problem ... > > My gcc is barfing with "fatal signal 125" ... is this the sign ... i am > having a hard time finding the meaning of this signal ... so it means i have > a hard time finding a work ... around ... What optimization level are you using? If you're using -O3, you may want to step it down and try again. Or maybe just try -O0 to see if it will compile. What architecture are you targeting? PPC uses elf format. - -- Pete Kockritz --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why do system calls hang? Date: 23 May 2002 20:57:05 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205231957.5e24510b@posting.google.com> References: <3cebd979$0$12554$4c41069e@reader0.ash.ops.us.uu.net> <3cebe574$0$12553$4c41069e@reader0.ash.ops.us.uu.net> Hello, To be fair here, 0 is a valid argument for taskDelete() - it means delete the current task. So, if you typed taskDelete without parameters at the shell, you effectively deleted the shell task! Rgds, John... "Mark Galecki" wrote in message news:<3cebe574$0$12553$4c41069e@reader0.ash.ops.us.uu.net>... > Never mind - this was a false-start. Obviously these system calls will > cause unpredictable results when called with bad arguments. And without > arguments, the default NULL is a bad one. --------------------------- Newsgroups: comp.os.vxworks Subject: builtin sim Date: 23 May 2002 22:09:27 -0700 From: be_gyus@yahoo.com (james) Organization: http://groups.google.com/ Message-ID: why is that, we cannot have multiple instances of the builtin simulator(i mean just instances only..)? I believe, is it b'coz it doesn't support n/w components or any other reason???????? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with a dosFs Date: Fri, 24 May 2002 05:34:54 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Correction: When installing DosFs 2.0, it effectively replaces the DosFs 1.0 object and include files on your Tornado tree, so after a BSP build, there is nothing left of DosFs 1. Still, a patch is needed for fixing some critical bugs, one is very specific to PowerPC (regarding 64-bit arithmetics), See my earlier post regarding details. Regarding the original question: Thge partition library in DosFs 2.0 supports upto 4 partitions per disk, but each partition can be of any size (I know of at least 320GBytes volume working well with FAT32), hence there is no need to limit partition size to 2 GBytes. A good question remains about why using partitions at all, but that's a different discussion alltogether. - - Leonid "Jason" wrote in message news:e555a185.0205231249.1db83265@posting.google.com... > Hello, > > Ask your wind river's representative for the DoSFs2 installation disk. > But that is not enough, because it still uses some Dosfs 1 library, so > you have to ask them again to send you a pathc for DosFs 2. > > good luck > > > cikic.n@tecnomare.it (Cikic Nenad) wrote in message news:... > > Hi all. > > > > I have a problem with a Fujitsu SCSI disk (18 GB). My target board is > > PowerPC based and I am using Tornado 2.0.2. > > > > Following the VxWorks Programmers guide I have done a single partition > > of 1 GB to test if it is working. After assuring that I have > > effectively made a dosFs partition and that I can write and read from > > it I have done 9 partition of 2 GB. After booting I saw all partitions > > with one error. The first partition remained of 1GB. I suppose that is > > becasue I passed a NULL value to dosFsDevInit. However thinking that a > > format can do only good (like rebooting is never wrong :) ) I have > > done an ioctl call with 5 as argument. The disk started to do > > something but it blocked. After rebooting I can do nothing with my > > disk. No one partition work and dosFsDevInit returns error. Supplying > > directly the Config structure instead of NULL in dosFsDevInit I see my > > partitions with devs but I can not do nothing with them. I have tried > > to mount a different filesystem and to format but with no success. > > > > I have looked everywhere (SPR, this group messages, windriver support) > > with no success. > > > > Do someone have any suggesstion or a utility library for some low > > level formating of disk (the library the support gave me does not > > work)? > > > > Btw, I have seen that there exist dosFs and dosFs2 filesystem. I > > suppose I have the dosFs2 filesystem since I use Tornado 2.0.2, but I > > see everywhere only INCLUDE_DOSFS. How can I be sure I have dosFs2? > > > > Thanks in advance > > > > > > Cikic Nenad --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Race Conditions in Vxworks!!?? Date: Fri, 24 May 2002 08:58:09 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <3CED168D.4E407998@fnal.gov> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088f68a.dip.t-dialin.net "Dinker Charak" schrieb im Newsbeitrag news:3CED168D.4E407998@fnal.gov... > > Senthil, > > Race conditions do occur. The most famous case was in pathfinder that > was sent to mars. Priority inversion had taken care of that. Search web > for that (or even comp.os.vxworks archives) for more details for that. > > Dinker Race conditions are not a matter of OS but of the application engineers S/W architecture. An OS can only help to avoid them and vxWorks does. Just my 2 Cents. Michael --------------------------- Newsgroups: comp.os.vxworks Subject: sharing memory in p3 processor by Tornado and VC Date: 24 May 2002 00:26:55 -0700 From: linsonaugustine@yahoo.co.in (Linson) Organization: http://groups.google.com/ Message-ID: I am final year B.Tech computer engineering student. I tried to implement an automobile braking system along with the fuel injector (both done on same machine using task preemption) using Tornado prototyper. The OS is windows ME. I tried to make a front end in VC. To communicate I planned to use a memory location which can be accessed using pointers by both VC and VxWorks. Eventough the pointer can be set up an exceprtion is occuring that it cannot be accessed. The same happens in VC also if the memory location is in Application of VxWorks. I tried to get the 'far' pointer in C, but couldn't find. Is there any method by which i can access that memory or is there any other method of communication between them. The System consists of P3 1Ghz processor on an 815E motherboard with 128MB ram. Please give your suggestions. Thanking you in advance. linson --------------------------- Newsgroups: comp.os.vxworks Subject: Re: sharing memory in p3 processor by Tornado and VC Date: Fri, 24 May 2002 10:31:15 +0100 From: Phil Ashby Organization: IDL Message-ID: <3CEE0863.65227B5D@idl.bt.co.uk> References: Linson wrote: > > I am final year B.Tech computer engineering student. I tried to > implement an automobile braking system along with the fuel injector > (both done on same machine using task preemption) using Tornado > prototyper. The OS is windows ME. I tried to make a front end in VC. > To communicate I planned to use a memory location which can be > accessed using pointers by both VC and VxWorks. Eventough the pointer > can be set up an exceprtion is occuring that it cannot be accessed. > The same happens in VC also if the memory location is in Application > of VxWorks. I tried to get the 'far' pointer in C, but couldn't find. > > Is there any method by which i can access that memory or is there > any other method of communication between them. > > The System consists of P3 1Ghz processor on an 815E motherboard > with 128MB ram. > > Please give your suggestions. > > Thanking you in advance. > > linson I think you're on the wrong track with shared memory communication between the Tornado simulator (VxSim) and a front-end VC application. How would it work with a real embedded system? FWIW I suggest you use the IP networking available on both platforms to communicate commands/data between the embedded system and the GUI. - -- Phil "Phlash" Ashby (speaking only for himself) Work: http://www.internet-designers.net/ Home: http://www.ashbysoft.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: newbie question Date: 24 May 2002 02:37:14 -0700 From: charles_giudicelli@yahoo.fr (Charles) Organization: http://groups.google.com/ Message-ID: Hello, I have to develop a bsp,all the files are write, but I don't know how to compile. My makefile is write but how run make? in a ms-dos window or is there another solution (I use Window NT) Thanks --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Fri May 24 07:33:47 2002 From: Mike McCullough Date: Fri May 24 07:33:49 PDT 2002 Subject: Re: free stack and TCB before taskDelete Long time VxWorks user but first time poster to this news group (so cut me some slack okay ;) ). The first thing you should accept is that taskInit() is a deprecated approach (yeah I know it worked in v4.3) and should always be followed by taskActivate() otherwise the behavior of VxWorks will be unpredictable. My recommendation is to always use taskSpawn(). Deprecated routines are just plain dangerous. I don't agree that taskDelete should only be used for debug as there are quite a few app's out there that do taskDeletes as part of normal operation. I do generally recommend creating a Stop routine that will gracefully delete all tasks and wrapping it in some kind of #define so that you can easily remove it from the app. Depending on your tasking design this may be more difficult than it sounds. There are also a few things you may want to look into if you are serious about your current approach: 1) Use exit() at the end of the task. This may or may not delete the TCB depending on the task options when you spawned the task. Still worth a try though... 2) Use taskDelete(0) inside each task. This should force the task to delete itself. 3) Investigate the use of a taskDeleteHook routine. I believe that if you use the code you wrote then the only way to delete the TCB will be by hand inside a delete hook routine. It looks to me like you are trying to do something "special" such as associating specific data with your tasks. If you are then you need to do some research into taskCreateHooks and task variables (taskVarLib). Hope this helps... Newsgroups: comp.os.vxworks Subject: Re: free stack and TCB before taskDelete Date: Thu, 23 May 2002 16:26:35 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <3CECE620.9000304@philips.com> Reply-To: "Michael Lawnick" Sender: mlawnick@p5088ff3e.dip.t-dialin.net Hi Arjan, I can't give you an answer to your question below, but what the hell are you doing ? You almost never should need a taskDelete() except for debug. Let your task terminate by leaving the outermost function ! Not knowing what you really do, your post reads to me like trying to push a nail through your knee. I'm pretty sure there exist easier mechanisms to do what you want to. Regards, Michael "Arjan" schrieb im Newsbeitrag news:3CECE620.9000304@philips.com... > Hi > > I'm working on some routines that create start and exit VxWorks tasks. > In this situation the taskDelete() is only called by the task that has > to be deleted itself. > It seems that the taskDelete() doesn't free the allocated memory for > the stack and TCB in case the task is created with taskInit(). > Is this a know problem or a bug ? > The VxWorks documentation explicitly describes it should, but perhaps > only if the task is created with taskSpawn() (?) > Therefore I have to free it myself with the following piece of code: > > --- > taskHandle = taskIdSelf(); > pTcb = taskTcb((Int32)taskHandle) > > #if (_STACK_DIR == _STACK_GROWS_DOWN) > free(pTcb->pStackEnd); > #else > free(pTcb->pStackBase); > #endif > free(pTcb); > > taskDelete(taskHandle); > --- > > This rises some questions: > - is it legal to delete a task of which the stack and TCB > don't exist anymore > - or should the memory be freed after the taskDelete > (probably not possible since then the task does't exit anymore) > - is it legal to check the return value of taskDelete() (idem) > > > regards, > Arjan > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: free stack and TCB before taskDelete Date: Thu, 23 May 2002 17:04:57 +0200 From: Arjan Message-ID: <3CED0519.6090602@philips.com> References: <3CECE620.9000304@philips.com> Hi Michael I think that's a logical response. But when I just let it run out of it's main loop the stack and TCB memory are not freed :( (Note that my tasks are created with taskInit istead of taskSpawn and therefore I had to allocate the stack and TCB myself ) I'm not sure if this is a bug, but I'm very unhappy with it .. regs, Arjan Michael Lawnick wrote: > Hi Arjan, > > I can't give you an answer to your question below, but what the hell are you > doing ? > You almost never should need a taskDelete() except for debug. Let your task > terminate by leaving the outermost function ! > Not knowing what you really do, your post reads to me like trying to push a > nail through your knee. > I'm pretty sure there exist easier mechanisms to do what you want to. > > Regards, > Michael > > "Arjan" schrieb im Newsbeitrag > news:3CECE620.9000304@philips.com... > >>Hi >> >>I'm working on some routines that create start and exit VxWorks tasks. >>In this situation the taskDelete() is only called by the task that has >>to be deleted itself. >>It seems that the taskDelete() doesn't free the allocated memory for >>the stack and TCB in case the task is created with taskInit(). >>Is this a know problem or a bug ? >>The VxWorks documentation explicitly describes it should, but perhaps >>only if the task is created with taskSpawn() (?) >>Therefore I have to free it myself with the following piece of code: >> >>--- >>taskHandle = taskIdSelf(); >>pTcb = taskTcb((Int32)taskHandle) >> >>#if (_STACK_DIR == _STACK_GROWS_DOWN) >> free(pTcb->pStackEnd); >>#else >> free(pTcb->pStackBase); >>#endif >> free(pTcb); >> >>taskDelete(taskHandle); >>--- >> >>This rises some questions: >> - is it legal to delete a task of which the stack and TCB >> don't exist anymore >> - or should the memory be freed after the taskDelete >> (probably not possible since then the task does't exit anymore) >> - is it legal to check the return value of taskDelete() (idem) >> >> >>regards, >>Arjan >> > > > Mike McCullough MCC Engineering President and CEO 9 Cheshire Court Londonderry, NH 03053 mike@mccengineering.com Tel: 603-537-9593 www.mccengineering.com Cell: 603-247-1219 *********************************************************************************** 24x7 VxWorks Support Custom Support Programs VxWorks Training Embedded Consulting Services A WindLink Partner *********************************************************************************** From vxwexplo-errs@csg.lbl.gov Sat May 25 04:03:10 2002 From: Vxworks Exploder Date: Sat May 25 04:03:13 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sat May 25 04:03:06 PDT 2002 Subject: Re: FTP problem Subject: SSH commercial implementation for VxWorks Subject: Re: VME bus errors and signal handlers Subject: Re: Where to download the patch for tornado? Subject: Re: GCC: Max file size ? Subject: Re: MODBUS TCP/IP for VxWorks Subject: Re: Converting Tornado 1.0.1 BSP to Tornado 2 Subject: Re: GCC: Max file size ? Subject: Re: Problem: fragmentation of free memory Subject: Can't not open symbol(question) Subject: Re: free stack and TCB before taskDelete ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: FTP problem Date: Fri, 24 May 2002 17:59:15 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: To determine the cause, is set up a packet sniffer to capture all packets between your target and the FTP server, until the process stops, at which point you can stop the packet capturing, and analyze the last packets captured. Only this way you can be sure if the target or server are the cause of the problem. You might add some timeout handling to your application code on the target, to recover from this condition, no matter who's fault this is. - - Leonid "JinSoo Kim" wrote in message news:edc45795.0205240644.5512373b@posting.google.com... > Hi everybody, > > I have a ftp problem with my target. > (The target spec is, CPU NET+ARM 40, VxWorks 5.4) > > I programmed ftp client using ftpLib in my target. > And the ftp server is on Linux machine. > My target continuously upload files every 30 seconds to > the ftp server. The file size is about 10KB. > > The problem is that the uploading sometimes stops for some reason. > (maybe 1 time / 2~3 days) > Once the uploading stops it never resumes again for itself. > The only method to resume uploading is power off/on the > target. > > I have a question. Is the ftp socket implemented in ftpLib > of VxWorks is blocking? If so, is it possible for ftp > client to wait forever in read function if it loses some packet? > > Thanks in advance for the advice of the gurus here, > :-) --------------------------- Newsgroups: comp.os.vxworks Subject: SSH commercial implementation for VxWorks Date: 24 May 2002 11:49:11 -0700 From: dsa0000@hotmail.com (dsa) Organization: http://groups.google.com/ Message-ID: Hi, What are the good SSH commercial implementations for VxWorks ? Regards DSA --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: VME bus errors and signal handlers Date: Fri, 24 May 2002 14:53:36 -0400 From: "William Dennen" Message-ID: <3cee8cdc$0$4544@dingus.crosslink.net> References: <3b4d40d1.0205230234.1eaad47b@posting.google.com> <3CED8891.78C4FAC8@bellsouth.net> Almost, but not quite, correct. Bus errors, returned as a bus error, result in a machine check in PowerPC processors. As such they are irrecoverable as all context is lost when the exception is handled. vxMemProbe and like routines do sneaky things like single byte DMA transfers so the MPU can recover. - -- >@< Bill Dennen MCG/GTAC "joe durusau" wrote in message news:3CED8891.78C4FAC8@bellsouth.net... > > > Chris Winstanley wrote: > > > > > <> > > > One of the processors (call it P1) transfers data over the VME bus > > into the local memory of the other two processors (call them P2 and > > P3) and also reads data from the local memory of P2 and P3. The > > default bus configuration is posted writes. > > > > Our customer has placed a requirement on us to be able to reboot a > > processor without affecting the other 2 processors. > > > > We have found that rebooting either P2 or P3 causes a VME bus error > > (as P1 is trying to communicate with P2 and P3), resulting in an > > exception being raised on P1, suspending the communicating task (our > > applications, being safety related, are constrained to being a single > > task). > > > > <> > > One problem that I've been told exists with most powerPC > CPUs is that due to internal pipelining they cannot recover the exact > machine state when a bus error occurs. Perhaps your boards do > not suffer from this problem, but I think that you should check with > the manufacturer to see whether recovery from a bus error can > be done. > > Speaking only for myself, > > Joe Durusau > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Where to download the patch for tornado? Date: 24 May 2002 19:48:12 -0700 From: sdssly@sina.com (sdssly) Organization: http://groups.google.com/ Message-ID: References: Johan Borkhuis wrote in message news:... > sdssly@sina.com (sdssly) wrote: > > However, I downloaded the Tornado from the free downlaod site. Can i > > download the patch for it? > > What patch do you want to download? > > Groeten, > Johan Hi Johan, Thanks for your reply. I want to get the Windows 2000 patch. After I installed the Torando on W2K, the WindRiver ULIP network card cannot be enabled. Some guys said I need install some W2K patch and network patch. Can you help me? Thanks in advance Tyler sdssly@sina.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GCC: Max file size ? Date: 24 May 2002 20:49:09 -0700 From: akarpov@my-deja.com (Alex) Organization: http://groups.google.com/ Message-ID: <4f09bb68.0205241949.71bbf798@posting.google.com> References: <5bRG8.24732$f5.2005649@news> <3CED3086.6070203@spamm.me.l8s.co.uk> <9FaH8.25148$f5.2031012@news> I was wondering guys, if you were talking about Tornado 2.1 here ? Anyway this stuff sounds familiar. Did you happen to know a patch or WRS issue number? Thanks, Alex bill@woola.qswtools.com wrote in message news:... > Are you compiling on Windows? Is there are error message about "virtual > alloc failed"? If so, see your tech support person. There are a > workaround and two patches for this. > > > > Thanks for the help so far, > > Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: MODBUS TCP/IP for VxWorks Date: 24 May 2002 21:05:54 -0700 From: akarpov@my-deja.com (Alex) Organization: http://groups.google.com/ Message-ID: <4f09bb68.0205242005.5131d7fd@posting.google.com> References: Raghuram, I'm afraid there is no "standard" VxWorks driver for such a proprietory fieldbus as MODBUS. Search the WEB, I'm sure you'll find what you're looking for. techsoul2001@yahoo.com (raghuram) wrote in message news:... > Hi, > > We are in the process of building a process control system using SBCs > running VxWorks RTOS. I would like to know if there is any standard > MODBUS over TCP/IP driver available for VxWorks. Any pointers in this > regard are appreciated. > > thanks and best regards, > raghuram. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Converting Tornado 1.0.1 BSP to Tornado 2 Date: 24 May 2002 21:35:10 -0700 From: akarpov@my-deja.com (Alex) Organization: http://groups.google.com/ Message-ID: <4f09bb68.0205242035.4f62ff41@posting.google.com> References: <7573be1638199cb11cc921b4defbfc42.8627@mygate.mailgate.org> Arne, I guess your code might not run just because the binary OS libraries are different in Tornado 2 tree. I would suggest to contact your WindRiver representative and ask him/her for a Tornado 2 BSP porting guide. Good luck, Alex "Arne Henrichsen" wrote in message news:<7573be1638199cb11cc921b4defbfc42.8627@mygate.mailgate.org>... > Hi, > > we have a Tornado 1.0.1 BSP for a MPC860. We want to start using it in > Tornado 2, and using the Tornado 1.0.1 Build compatibility in Tornado 2, > the BSP compiles without errors. The problem is that it will not run on > our target system. > > Comparing the build output from Tornado 1.0.1 with that from Tornado 2, > I only find a difference right at the end of the output: (we build > vxWorks_rom.hex images). It seems that the size of the images are > different. I have no idea why. In T1.0.1 the linker seems to be used, > while in T2 they use the compiler to link the modules. > > Any ideas? > > Thanks > Arne > > T1.0.1 Output: > > > Done. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GCC: Max file size ? Date: Sat, 25 May 2002 05:32:10 GMT From: "drdiags" Organization: AT&T Broadband Message-ID: References: <5bRG8.24732$f5.2005649@news> <3CED3086.6070203@spamm.me.l8s.co.uk> <9FaH8.25148$f5.2031012@news> <4f09bb68.0205241949.71bbf798@posting.google.com> Alex, I am not sure what rev of Tornado is being discussed, but there is SPR 73419 (?) which was just released for several C++ related issues for T2.0.x compiler issues. The one I had was for T2.0.2 for PPC, when using -mlongcall and -O2. I found that I had to still apply the pepatch increase in the stack size on this newly installed patch. I was building the ACE/TAO code to run on a 64MB SBC8260 board. I also know there is a new SPR compiler patch for the T2.1 for PPC74xx which was just put up on WindSurf (SPR74xxx - sorry that these numbers are not exact, but look on Wind Surf). "Alex" wrote in message news:4f09bb68.0205241949.71bbf798@posting.google.com... > I was wondering guys, if you were talking about Tornado 2.1 here ? > Anyway this stuff sounds familiar. Did you happen to know a patch or > WRS issue number? > > Thanks, > Alex > > bill@woola.qswtools.com wrote in message news:... > > > Are you compiling on Windows? Is there are error message about "virtual > > alloc failed"? If so, see your tech support person. There are a > > workaround and two patches for this. > > > > > > Thanks for the help so far, > > > Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem: fragmentation of free memory Date: 24 May 2002 23:51:01 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205242251.41c4e19f@posting.google.com> References: <313680C9A886D511A06000204840E1CF67A181-100000@whq-msgusr-02.pit.comms.marconi.com> Hello Keith, > That's odd. My system has the same pattern of 4088 byte free blocks, > though I've only got 84 of them. I've always assumed that it was part of > our application, but apparently it's coming from VxWorks itself. Huh. It is coming from VxWorks... 4088 = 4096 - 16. Each time a page is needed for the MMU page table, the system requests a 4096 byte aligned block of 4096 bytes. To get this, the memory allocator will request a block of 8192, find where the 4096 byte boundary within that block is that will also allow for a 16 byte block header. The excess is then returned to the system pool; often blocks of 4088 bytes. Do you have a lot of entries in your sysPhysMemDesc array? HTH, John... --------------------------- Newsgroups: comp.os.vxworks Subject: Can't not open symbol(question) Date: 25 May 2002 02:13:39 -0700 From: essenlee@hanmail.net (LEE MYUNG SUK) Organization: http://groups.google.com/ Message-ID: Can't not open symbol (question) environment : solaris 7.0 , vxWorks 5.4, tornado2.0, ads860 board at first, I downloading at DRAM, and save to flash, then, LAN CABLE pull out, then, flash -> dram copy, dram go. question> flash -> dram copy and go, Can't not open symbol errno = 0xd003 error message ... and -> In a shell, command can't. undefine symbol message. maybe, Have to standalone ? myjob : bootrom.hex eeprom burn. and dram - > flash save , power off, flash - > dram copy go. help me advice, then thank you very very much... --------------------------- Newsgroups: comp.os.vxworks Subject: Re: free stack and TCB before taskDelete Date: 25 May 2002 03:51:54 -0700 From: prash94@hotmail.com (Prasanth) Organization: http://groups.google.com/ Message-ID: <2a05aca6.0205250251.77222ae7@posting.google.com> References: <3CECE620.9000304@philips.com> Well Arjan... Can you check out if the exit() call can be used..? I think that it does a better task "stack freeing" than the taskDelete() call. Regards Prasanth. Arjan wrote in message news:<3CECE620.9000304@philips.com>... > Hi > > I'm working on some routines that create start and exit VxWorks tasks. > In this situation the taskDelete() is only called by the task that has > to be deleted itself. > It seems that the taskDelete() doesn't free the allocated memory for > the stack and TCB in case the task is created with taskInit(). > Is this a know problem or a bug ? > The VxWorks documentation explicitly describes it should, but perhaps > only if the task is created with taskSpawn() (?) > Therefore I have to free it myself with the following piece of code: > > --- > taskHandle = taskIdSelf(); > pTcb = taskTcb((Int32)taskHandle) > > #if (_STACK_DIR == _STACK_GROWS_DOWN) > free(pTcb->pStackEnd); > #else > free(pTcb->pStackBase); > #endif > free(pTcb); > > taskDelete(taskHandle); > --- > > This rises some questions: > - is it legal to delete a task of which the stack and TCB > don't exist anymore > - or should the memory be freed after the taskDelete > (probably not possible since then the task does't exit anymore) > - is it legal to check the return value of taskDelete() (idem) > > > regards, > Arjan --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Sat May 25 17:16:21 2002 From: Shobhan Chinnam Date: Sat May 25 17:16:24 PDT 2002 Subject: RE: comp.os.vxworks newsdigest I just finished porting ssh2 code from http://www.openssh.com/ to vxWorks. Try if it helps. Regards, csk Newsgroups: comp.os.vxworks Subject: SSH commercial implementation for VxWorks Date: 24 May 2002 11:49:11 -0700 From: dsa0000@hotmail.com (dsa) Organization: http://groups.google.com/ Message-ID: Hi, What are the good SSH commercial implementations for VxWorks ? Regards DSA From vxwexplo-errs@csg.lbl.gov Sun May 26 04:03:07 2002 From: Vxworks Exploder Date: Sun May 26 04:03:10 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Sun May 26 04:03:04 PDT 2002 Subject: how to use ncr810CtrlCreate( ) to install sym53C895 pmc card? Subject: Re: How to set TIMEZONE? Support for timezones in localtime? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: how to use ncr810CtrlCreate( ) to install sym53C895 pmc card? Date: Fri, 24 May 2002 17:55:57 +0900 From: "bigfire" Organization: Inet Internet Services Message-ID: hello all, I wish to install sym53C895 pmc card to my PPC bard. So, I included SCSI,SCSI2 etc... When vxWorks boot, the program stopped at 'ncr810CtrlCreate( )'. The devType was ''NCR875_DEVICE_ID' When I use 'NCR895_DEVICE_ID' as devType, vxWorks turned "Invalid NCR8XX device type. Not supported." message Would you give me advice for this problem? thank Park, D.S. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Sun, 26 May 2002 11:06:53 +0200 From: Alex Vinokur Message-ID: <3CF0A5AD.54901751@bigfoot.com> References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> <3CEB655C.306@bigfoot.com> <3cebb3b5.89154657@news.mhogaming.com> "Ignacio Gonzalez T." wrote: > On Wed, 22 May 2002 10:31:08 +0100, Alex Vinokur > wrote: > > > > >TIMEZONE is of the form: > >name_of_zone:<(unused)>:time_in_minutes_from_UTC:daylight_start:daylight_end > > > > > >How can I know what is time_in_minutes_from_UTC's value for a specific > >zone (Israel) ? > > > > Just look at an agenda. Mine says Israel is one hour away from UTC, so > time_in_minutes_from_UTC should be 60 (or -60?) time_in_minutes_from_UTC must be >= 0 - -- =========================== Alex Vinokur mailto:alexvn@bigfoot.com mailto:alexvn@go.to http://up.to/alexvn http://go.to/alexv_math =========================== --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon May 27 04:03:11 2002 From: Vxworks Exploder Date: Mon May 27 04:03:14 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Mon May 27 04:03:08 PDT 2002 Subject: How can I add the TFFS component? Subject: Re: GCC: Max file size ? Subject: Re: FTP problem Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Subject: Re: Race Conditions in Vxworks!!?? Subject: Re: free stack and TCB before taskDelete Subject: mbuf leak in dhcpc Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Subject: check CPM on SBC8260 Subject: Re: Default route outside subnet Subject: Re: FTP problem Subject: Re: How can I add the TFFS component? Subject: multiple Boot Subject: Re: Correct method to change IP addr? Subject: Q: Last Sunday of October?...... Subject: Re: Problem: fragmentation of free memory Subject: Re: how to use ncr810CtrlCreate( ) to install sym53C895 pmc card? Subject: Re: chipDISK ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: How can I add the TFFS component? Date: 26 May 2002 05:01:23 -0700 From: xiaohua.zhang@163.com (xiaohua) Organization: http://groups.google.com/ Message-ID: Hi,Top guns I want add the TFFS into my project,now I have several problems: 1.There must have FlashMem.c and etc. in BSP? 2.How should I do to init the Flash? 3.How can I use TFFS to boot target board? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: GCC: Max file size ? Date: Sun, 26 May 2002 13:53:00 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <0N5I8.25301$f5.2068761@news> References: <5bRG8.24732$f5.2005649@news> <3CED3086.6070203@spamm.me.l8s.co.uk> <9FaH8.25148$f5.2031012@news> <4f09bb68.0205241949.71bbf798@posting.google.com> I am waiting to see what the path is ... but apparently there is one ... "Alex" wrote in message news:4f09bb68.0205241949.71bbf798@posting.google.com... > I was wondering guys, if you were talking about Tornado 2.1 here ? > Anyway this stuff sounds familiar. Did you happen to know a patch or > WRS issue number? > > Thanks, > Alex > > bill@woola.qswtools.com wrote in message news:... > > > Are you compiling on Windows? Is there are error message about "virtual > > alloc failed"? If so, see your tech support person. There are a > > workaround and two patches for this. > > > > > > Thanks for the help so far, > > > Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FTP problem Date: Sun, 26 May 2002 14:17:19 GMT From: "David Ganor" Organization: Verio Message-ID: References: Hi JinSoo Kim, "JinSoo Kim" wrote in message news:edc45795.0205240644.5512373b@posting.google.com... > Hi everybody, > > I have a ftp problem with my target. > (The target spec is, CPU NET+ARM 40, VxWorks 5.4) > > I programmed ftp client using ftpLib in my target. > And the ftp server is on Linux machine. > My target continuously upload files every 30 seconds to > the ftp server. The file size is about 10KB. OK. > > The problem is that the uploading sometimes stops for some reason. > (maybe 1 time / 2~3 days) > Once the uploading stops it never resumes again for itself. > The only method to resume uploading is power off/on the > target. > > I have a question. Is the ftp socket implemented in ftpLib > of VxWorks is blocking? Yes and NO. Usually the Read operation is blocking and the write in NON-Blocking. > If so, is it possible for ftp > client to wait forever in read function if it loses some packet? Well, 'wait forever' is not exact. It waits until all number of requested bytes are received, and then it continues. > > Thanks in advance for the advice of the gurus here, > :-) David Ganor david@wavion.biz --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Sun, 26 May 2002 18:44:25 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CF12D79.3040106@spamm.me.l8s.co.uk> References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> <3CEB655C.306@bigfoot.com> <3cebb3b5.89154657@news.mhogaming.com> <3CF0A5AD.54901751@bigfoot.com> Alex Vinokur wrote: >>> >>Just look at an agenda. Mine says Israel is one hour away from UTC, so >>time_in_minutes_from_UTC should be 60 (or -60?) >> > > time_in_minutes_from_UTC must be >= 0 Why? IIRC it is +ve for west (ie the USA) and -ve for most of Europe. - -ve values work fine (in spite of what the FM says). WRS support might even give you a version of the code that works if you are S of the equater. David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Race Conditions in Vxworks!!?? Date: Sun, 26 May 2002 18:50:59 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CF12EF6.5060308@spamm.me.l8s.co.uk> References: <3CED168D.4E407998@fnal.gov> Michael Lawnick wrote: > > Race conditions are not a matter of OS but of the application engineers S/W > architecture. > An OS can only help to avoid them and vxWorks does. > > Just my 2 Cents. B*ll**ks... Plenty of the code supplied by WRS was originally written to run inside a Unix kernel. As such it doesn't expect to be pre-empted.......... pppDelete() almost always gets it wrong UNLESS the connection is in a static state (ie connected or failing to connect). The routing table stuff has no locks at all (that I could find), so routeDelete() can hit a route that is currently in use (by a code path). The list goes on. Fortunately vxWorks isn't often used where these features are used dynamically....... David --------------------------- Newsgroups: comp.os.vxworks Subject: Re: free stack and TCB before taskDelete Date: 26 May 2002 12:25:54 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205261125.cafd35d@posting.google.com> References: <3CECE620.9000304@philips.com> <2a05aca6.0205250251.77222ae7@posting.google.com> Hello, > Can you check out if the exit() call can be used..? > I think that it does a better task "stack freeing" > than the taskDelete() call. Please ignore this "advice" - exit() does nothing different to taskDelete() in terms of freeing memory. You've already been given the answer to what you are seeing in a previous post. If you use taskInit() to create the task, you are responsible for cleaning up the memory for the TCB and, maybe, the stack (you can request that the stack be freed using the VX_DEALLOC_STACK option). As the previous poster stated though, you'd be better off avoiding taskInit() if you can. You should also note that it is no longer available in VxWorks AE - instead there is a new taskCreate() function that allows you to create a dormant task, but handles the memory allocation for you. HTH, John... > > Regards > Prasanth. > > Arjan wrote in message news:<3CECE620.9000304@philips.com>... > > Hi > > > > I'm working on some routines that create start and exit VxWorks tasks. > > In this situation the taskDelete() is only called by the task that has > > to be deleted itself. > > It seems that the taskDelete() doesn't free the allocated memory for > > the stack and TCB in case the task is created with taskInit(). > > Is this a know problem or a bug ? > > The VxWorks documentation explicitly describes it should, but perhaps > > only if the task is created with taskSpawn() (?) > > Therefore I have to free it myself with the following piece of code: > > > > --- > > taskHandle = taskIdSelf(); > > pTcb = taskTcb((Int32)taskHandle) > > > > #if (_STACK_DIR == _STACK_GROWS_DOWN) > > free(pTcb->pStackEnd); > > #else > > free(pTcb->pStackBase); > > #endif > > free(pTcb); > > > > taskDelete(taskHandle); > > --- > > > > This rises some questions: > > - is it legal to delete a task of which the stack and TCB > > don't exist anymore > > - or should the memory be freed after the taskDelete > > (probably not possible since then the task does't exit anymore) > > - is it legal to check the return value of taskDelete() (idem) > > > > > > regards, > > Arjan --------------------------- Newsgroups: comp.os.vxworks Subject: mbuf leak in dhcpc Date: 26 May 2002 20:59:53 -0700 From: roh_youngjin@hotmail.com (Youngjin Roh) Organization: http://groups.google.com/ Message-ID: <799f3d42.0205261959.52ce2ef6@posting.google.com> Dear experts. I have problem in leaking mbuf when I use dhcp client module. dhcp client works well, but every time try to lease thr dhcpc, mbuf's IPMADDR allocated by 1. Therefore, if there is a problem in dhcp server outside, and I try to lease IP continuously, eventually the mbuf has no free block and no networking available. Anyone have this problem? or Solution? Thanks. Youngjin - -> netStackSysPoolShow type number - --------- ------ FREE : 4982 DATA : 0 HEADER : 0 SOCKET : 14 PCB : 19 RTABLE : 18 HTABLE : 0 ATABLE : 0 SONAME : 0 ZOMBIE : 0 SOOPTS : 0 FTABLE : 0 RIGHTS : 0 IFADDR : 8 CONTROL : 0 OOBDATA : 0 IPMOPTS : 0 IPMADDR : 79 <---- This increments by 1 when I call dhcpcBind(...) IFMADDR : 0 MRTABLE : 0 TOTAL : 5120 number of mbufs: 5120 number of times failed to find space: 0 number of times waited for space: 0 number of times drained protocols for space: 0 __________________ CLUSTER POOL TABLE _______________________________________________________________________________ size clusters free usage - ------------------------------------------------------------------------------- 64 640 627 327 128 640 614 375 256 640 630 248 512 640 626 427 - ------------------------------------------------------------------------------- value = 80 = 0x50 = 'P' --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Mon, 27 May 2002 07:45:09 +0200 From: Alex Vinokur Message-ID: <3CF1C7E4.E0F6DF34@bigfoot.com> References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> <3CEB655C.306@bigfoot.com> <3cebb3b5.89154657@news.mhogaming.com> <3CF0A5AD.54901751@bigfoot.com> <3CF12D79.3040106@spamm.me.l8s.co.uk> David Laight wrote: > Alex Vinokur wrote: > > >>> > >>Just look at an agenda. Mine says Israel is one hour away from UTC, so > >>time_in_minutes_from_UTC should be 60 (or -60?) > >> > > > > time_in_minutes_from_UTC must be >= 0 > > Why? IIRC it is +ve for west (ie the USA) and -ve for most of > Europe. || ### File /docs/vxworks/ref/ansiTime.html ### || If the environment variable TIMEZONE is set, the information is retrieved from this variable, otherwise from the locale information. TIMEZONE is of the form: name_of_zone:<(unused)>:time_in_minutes_from_UTC:daylight_start:daylight_end To calculate local time, the value of time_in_minutes_from_UTC is subtracted from UTC; time_in_minutes_from_UTC must be positive. || || ########################################### > > > -ve values work fine (in spite of what the FM says). > > WRS support might even give you a version of the code that > works if you are S of the equater. > > David - -- =========================== Alex Vinokur mailto:alexvn@bigfoot.com mailto:alexvn@go.to http://up.to/alexvn http://go.to/alexv_math =========================== --------------------------- Newsgroups: comp.os.vxworks Subject: check CPM on SBC8260 Date: 26 May 2002 12:42:37 -0700 From: dannyc@harmonic.co.il (Danny Cohen) Organization: http://groups.google.com/ Message-ID: <5288f97d.0205261142.7790a7ac@posting.google.com> I've evaluation board of SBC8260. I'd like to check the CPM interface, and make an application that receive UDP datagrams using the SCC and DMA them into the SDRAM than loopback using another DMA channel and transmit them to the IP network. How can i do it? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Default route outside subnet Date: Sun, 26 May 2002 15:10:06 -0400 From: Iubica2 Message-ID: <3CF1330E.30409@yahoo.com> References: Your code should check that the route's gateway is on the same subnet with your interface before configuring the route. A gateway really is a next hop node, so it must be on the same subnet with at least one of your interfaces.... because..... Routes consist of the following bits of information: - - route destination address and mask - - route gateway - - route interface - - route metric, etc Routes can be of two types: - - directly-connected, for which all IP addresses matching the route's destination IP and mask are on the same subnet with the route interface. For example, the loopback route is directly connected. - - indirectly-connected, when IP addresses matching the route's destination IP and mask are not on the same subnet with the route's interface. Indirectly-connected routes require a first hop - which by definition is the route's gateway. (For directly connected routes, the first hop is the destination.) The first hop (the 'gateway') must be directly connected with the route's interface. Say a packet with a certain destination IP is supposed to be sent out. 2 questions arise: - - what outgoing interface should this packet use? - - if the outgoing interface is an ethernet interface (as opposed to e.g. the loopback interface), how do we select the packet's destination HW address? The answer for these 2 questions comes from selecting the packet's route from the routing table. Here is how the route secection process works. - - Among all routes, find all routes whose destination IP/mask match the destination IP of the packet. These are the matching routes. - - Among the matching routes, find the routes with longest mask. These are the longest match routes. - - Among the longest match routes, find the routes with smaller metric. If only one route results, that will be the selected route. If still more than one route results, just pick a route from that group. Say that we've selected a route through eth0, an ethernet interface. How do we determine the destination HW address of our packet? We need to determine the next hop of our packet, and use the next hop's HW address. - - If the route is directly connected, then the destination IP of the packet is the next hop. In this case, set the destination HW address of the packet to be the HW address of the destination IP of the packet. - - If the route is indirectly connected, then the route's gateway is the next hop. In this case, set the destination HW address of the packet to be the HW address of the gateway. Good references for all this stuff are Stevens' book 'TCP/IP Illustrated vol 1" or the Host Requirements RFC. Johan Borkhuis wrote: > "RonnoBonno" wrote: > > >>Have you tried, >> >>routeAdd "0.0.0.0","IP Addr of the Gateway" >> >>the 0.0.0.0 route is the default gateway. >> > > Yes, but the problem is that the address of the gateway is outside the > subnet. > > I managed to get it to work (sort of). After I received the data from the > DHCP server I had to implement the following sequence: > > ifAddrAdd ("dp0",); > routeAdd ("0.0.0.0",); > arpAdd (,,0); > > For some reason the ARP does not work correctly on the second IP address, > so I had to add the entry manually. This also gives problems with the DNS > server. When I ask for an address the query is sent out on the second IP > address, as this is the address of the DNS resolver in the box where the > DHCP server is also located. But that box sends out an ARP, but this ARP > is not answered by the VxWorks box. > > Groeten, > Johan > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: FTP problem Date: 26 May 2002 12:08:56 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205261108.3b9cffd7@posting.google.com> References: Hello, Something to think about here is the fact that you are consuming sockets at a faster rate than they are returned to the system. What do I mean? Well, when you close a socket (such as the data transfer one in the FTP session), the system internally must hold on to the data structures for that connection for a fixed period (I believe 2 minutes, though it might be just 1 minute - I can never remember which it is). That is done to ensure that the port is not reused immediately. If it were to be resused, delayed packets from the first connection might mistakenly find their way into the second connection's data stream. Note that this is not a VxWorks-ism - this is a requirement of TCP, and should be implemented by all network stacks. It is an indication of how our use of the network has changed since it was designed. TCP is a connection oriented protocol, designed for infrequent, long-lasting connections, but we are increasingly using it for short connections. The worst example (HTTP) was fixed by extending the protocol to support multiple transfers on the same connection. Before that, complex web pages could put a huge demand on a web server's resources. Each image on the page would have resulted in a new connection to the same server, which would rapidly consume the web server's resources. VxWorks is designed for use in memory constrained embedded systems, and as such has a finite number of buffers for holding these socket descriptors. The network stack's memory is pre-allocated (a) to make its performance more predictable, and (b) to avoid it consuming all the system memory (a bad thing for most systems, especially those that have critical real-time functions and use the network for device management only). So, if you are using buffers for new sockets at 1 every 30 seconds (i.e. for each new data transfer socket), but the stack only releases them at, say, 1 each minute, you can see how this can result in them running out fairly fast. Normally though, this only results in a delay until space can be found for the new socket. Hence, I'm not sure it is what you are seeing, but it is worth thinking about. You might also want to look into passive FTP - that, I believe, uses the control connection for the data transfer too, and might allow you to keep a single connection open to the server all the time, rather than continually opening and closing connections. HTH, John... people@micro-web.co.kr (JinSoo Kim) wrote in message news:... > Hi everybody, > > I have a ftp problem with my target. > (The target spec is, CPU NET+ARM 40, VxWorks 5.4) > > I programmed ftp client using ftpLib in my target. > And the ftp server is on Linux machine. > My target continuously upload files every 30 seconds to > the ftp server. The file size is about 10KB. > > The problem is that the uploading sometimes stops for some reason. > (maybe 1 time / 2~3 days) > Once the uploading stops it never resumes again for itself. > The only method to resume uploading is power off/on the > target. > > I have a question. Is the ftp socket implemented in ftpLib > of VxWorks is blocking? If so, is it possible for ftp > client to wait forever in read function if it loses some packet? > > Thanks in advance for the advice of the gurus here, > :-) --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I add the TFFS component? Date: 26 May 2002 12:18:30 -0700 From: john_94501@yahoo.com (John) Organization: http://groups.google.com/ Message-ID: <488e459a.0205261118.4fbc47dd@posting.google.com> References: Hello, First, read the manual. If you don't have it installed (and you should have it there if you installed TFFS), you can read it on WindSurf at: http://web2.windriver.com/windsurf/techpubs/osExtensions/tffs/guide/flashMem.pdf You will need a WindSurf account though to get to that page. In there, you will find a Getting Started section that explains all the stuff you need to do to get TFFS working on your board. You might want to give your local support engineer and/or FAE a call though - they might have some examples available that are close to your system. Since you omitted to mention the version of VxWorks, the type of CPU and any information about the board, I can't tell you much more. For future reference, when posting here it is best to include as much information about your setup as possible (esp. VxWorks/Tornado versions and CPU types). HTH, John... xiaohua.zhang@163.com (xiaohua) wrote in message news:... > Hi,Top guns > I want add the TFFS into my project,now I have several problems: > 1.There must have FlashMem.c and etc. in BSP? > 2.How should I do to init the Flash? > 3.How can I use TFFS to boot target board? --------------------------- Newsgroups: comp.os.vxworks Subject: multiple Boot Date: 26 May 2002 22:44:04 -0700 From: be_gyus@yahoo.com (james) Organization: http://groups.google.com/ Message-ID: hi VxWorkers, What exactly do we have to change in the BootConfig.c to have the choice of multiple booting ? Can anyone just brief on that. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Correct method to change IP addr? Date: Sun, 26 May 2002 16:45:18 -0400 From: Iubica2 Message-ID: <3CF1495E.1010106@yahoo.com> References: Are you trying to bind() a TCP server socket? bind() may fail because the TCP/IP stack has a lingering TCP server socket bound at the same IP address/port that you're trying to bind to. After the application-level code has called close() on a TCP server socket, the socket still lingers on for a while in the TCP/IP stack until the incoming connections get closed. This is because closing a TCP connection requires (in each direction!) sending a FIN, receiving the FIN's ACK, sending the ACK to the FIN's ACK, then waiting for 30 secs or so (an OS-dependent constant) to make sure the far end does not retransmit the FIN's ACK. That's why the close()d socket may linger in the TCP/IP stack for 30 secs or more before being really closed at the level of the TCP/IP stack. Carl Petersen wrote: > Is this the correct sequence to change the IP addr of an interface? > > ifRouteDelete(...); > > ifMaskSet(...); > > ifAddrSet( ifName, "New IP" ); > > These calls run without error however a subsequent call to bind() fails. > > I have also tried ipDetach() then ipAttach() without success. Note that > the interface is operational before I attempt to modify the IP addr. > > Tornado 2.02 Pentium BSP > > Thanks > > Carl Petersen > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Q: Last Sunday of October?...... Date: 26 May 2002 15:06:26 -0700 From: roswel_ajf@hotmail.com (roswell) Organization: http://groups.google.com/ Message-ID: <6dcae13d.0205261406.1e32e6a8@posting.google.com> Hi folks, For timezone, i need to build the TIMEZONE env. All i have is the rule: first sunday of april, and enddate is last sunday of october. Now, how do i know the exact data,?..any ideas how to calculte? i have the year as well. thanks roswel --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem: fragmentation of free memory Date: 27 May 2002 00:55:11 -0700 From: u.mascia@pxl.it (Umberto Mascia) Organization: http://groups.google.com/ Message-ID: References: <313680C9A886D511A06000204840E1CF67A181-100000@whq-msgusr-02.pit.comms.marconi.com> > > So the first question: > > * where is the missing memory (approx 900KB to the total of 4MB)? > > That's likely being consumed by your VxWorks image. You may have 4M of > physical memory on your box, but not all of it gets used for the heap. OK, that's right: I found a clear explanation about memory usage in the manual. Anyway, it seems strange to me that the kernel reserves from the beginning ~1.6MB of data structures. After my first posting, I tried to remove the "cache" and "MMU" components from the kernel image, and the result was to have almost the previously allocated memory returned back to free memory, with a largest block of ~2MB. What a joy! But after a while I came back in deep sadness: now it's impossible to allocate anything - obviously :-) Now the point is: * cache + MMU are needed for "everyday working" * cache + MMU cause waste of memory * how can I tune the size of memory needed from cache + MMU ? Bye, Umberto --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how to use ncr810CtrlCreate( ) to install sym53C895 pmc card? Date: 27 May 2002 01:42:29 -0700 From: trahn@macrosystem.de (Thomas Rahn) Organization: http://groups.google.com/ Message-ID: <30c8bd8b.0205270042.1e03c54d@posting.google.com> References: Hello, we use the ncr810 driver with a 875 chip. We had to do some bugfixes and were licky to receive the sources. As far as I remember, that driver does not support the 895 chip (though it might be added easily if you have the sources...) But there is another driver called sym895Lib (which we had for reference). You will have to check your source tree and maybe contact WRS. I am not sure which drivers are included in current VxWorks packages (did not touch the installation for quite some time) Bye Thomas Rahn --------------------------- Newsgroups: comp.os.vxworks Subject: Re: chipDISK Date: Mon, 27 May 2002 10:22:56 GMT From: igtorque@yahoo.es (Ignacio G.T.) Organization: Newshosting.com - Highest quality at a great price! www.newshosting.com Message-ID: <3cf2076c.334581@news.mhogaming.com> References: <3cebceb5.96066856@news.mhogaming.com> On Thu, 23 May 2002 08:59:07 +0200, "Michael Lawnick" wrote: >Hi Ignacio, > >to access the chipDISK, you only need the JUMPTec-BSP with ATA and dosFs >included. Take care to have dosFs2 installed from the beginning of your >project to avoid later version changes when the first problems with old >dosFs(1) will occure. >JIDA is needed only for the board-specials like S/N, times of boot etc. > >HTH >Michael Lawnick Thank you, Michael. Let me see if I've understood: if a grab a JUMPTec MOPS/686+, instal the standard PentiumPC BSP plus JUMPtec's Ethernet driver plus JUMPTec's JIDA, plus standard PentiumPC ATA driver and dosFs(2), I'll be able to see a chipDISK as a standard disk with DOS file system. Is that OK? No problems? > >"Ignacio G.T." schrieb im Newsbeitrag >news:3cebceb5.96066856@news.mhogaming.com... >> I'm planning to use a PC/104 board from JUMPtec. I've been told by a >> distributor that it comes with VxWorks drivers. So I've checked their >> web site and found an Ethernet driver and a special set of SW modules >> & API (called JIDA) to access the watchdog, the S/N and the like. >> >> My concern is that I want to use flash storage. They offer >> chipDISK-IDE, that can be connected to the card through the IDE >> connector. They claim this is "transparent" to the OS. So I think >> that, using the standard PentiumPC BSP, adding the Ethernet driver and >> the JIDA, and setting up an ATA-based DOS file system I could access >> the chipDISK-IDE. Is this right? Or should I get/write special >> drivers? >> >> I know the caveats about limited write-cycles on Flash, etc. That is >> not important for my application. >> >> -- >> Ignacio G.T. > > - -- Ignacio G.T. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Mon May 27 23:05:41 2002 From: askon@mailru.com Date: Mon May 27 23:05:46 PDT 2002 Subject: Tornado Debugger Hi,everybody I use internal Tornado 2.0.2 debugger for remote debugging by ethernet. I use WDB END driver connection with mvme2604 board. But when I debug my application Tornado debugger once too often dropped and following dialog box appears on screen: "The Tornado Debugger appears to be busy. Retry for 10 seconds?" When I click on the "Retry" button after 10 seconds waiting this dialog box appears on screen again. Sometimes after that Debugger continue work fine. It depends. Can you help me to resolve this problem? Regards, Andrei. From vxwexplo-errs@csg.lbl.gov Tue May 28 04:03:14 2002 From: Vxworks Exploder Date: Tue May 28 04:03:16 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Tue May 28 04:03:10 PDT 2002 Subject: Re: chipDISK Subject: Task::setName AFTER taskSpawn Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Subject: Re: routine called from kernel/application Subject: Re: Last Sunday of October?...... Subject: Multicast Help ? What does it take ... Subject: RealTek8139 driver for x86 board ? Subject: question on ncr810CtrlInit( ) Subject: Prototyper assistance - rookie Subject: Global variable probleme Subject: Script for dependencies, Need Help Subject: Help: is there DNS Server or equivalent for VxWorks ? Subject: Re: is there DNS Server or equivalent for VxWorks ? Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Subject: Re: Prototyper assistance - rookie Subject: why only Variables and BackTrace Debug windows do not show Subject: Tornado build: external (console) windows flashes Subject: Re: Tornado build: external (console) windows flashes Subject: Re: multiple Boot Subject: Re: Global variable probleme Subject: Problem with IP Networking in Prototyper Subject: Re: Tornado build: external (console) windows flashes Subject: Re: Problem with IP Networking in Prototyper ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: chipDISK Date: Mon, 27 May 2002 13:34:10 +0200 From: "Michael Lawnick" Organization: Buergernetz Dillingen Message-ID: References: <3cebceb5.96066856@news.mhogaming.com> <3cf2076c.334581@news.mhogaming.com> Reply-To: "Michael Lawnick" Sender: mlawnick@pd9e65a11.dip.t-dialin.net Hi Ignacio, > Thank you, Michael. Let me see if I've understood: if a grab a JUMPTec > MOPS/686+, instal the standard PentiumPC BSP plus JUMPtec's Ethernet > driver plus JUMPTec's JIDA, plus standard PentiumPC ATA driver and > dosFs(2), I'll be able to see a chipDISK as a standard disk with DOS > file system. Is that OK? No problems? Nearly - do it this way: grab a JUMPTec MOPS/686+, install Tornado2, the JUMPTec MOPS BSP (including JIDA and -if ethernet onboard- JUMPtec's ethernet driver), plus standard PentiumPC ATA driver and dosFs(2). Now you should be able to see a chipDISK as a standard disk with DOS. This is nothing magic of MOPS or JUMPTec, but of specification of the PMC-flash-discs. Regards, Michael > > > >"Ignacio G.T." schrieb im Newsbeitrag > >news:3cebceb5.96066856@news.mhogaming.com... > >> I'm planning to use a PC/104 board from JUMPtec. I've been told by a > >> distributor that it comes with VxWorks drivers. So I've checked their > >> web site and found an Ethernet driver and a special set of SW modules > >> & API (called JIDA) to access the watchdog, the S/N and the like. > >> > >> My concern is that I want to use flash storage. They offer > >> chipDISK-IDE, that can be connected to the card through the IDE > >> connector. They claim this is "transparent" to the OS. So I think > >> that, using the standard PentiumPC BSP, adding the Ethernet driver and > >> the JIDA, and setting up an ATA-based DOS file system I could access > >> the chipDISK-IDE. Is this right? Or should I get/write special > >> drivers? > >> > >> I know the caveats about limited write-cycles on Flash, etc. That is > >> not important for my application. > >> > >> -- > >> Ignacio G.T. > > > > > > -- > Ignacio G.T. --------------------------- Newsgroups: comp.os.vxworks Subject: Task::setName AFTER taskSpawn Date: Mon, 27 May 2002 12:36:23 +0100 From: "Mario Semo" Organization: Nextra Telekom GmbH Message-ID: Hello, Is it possible (and if : how) to change the name of a task AFTER i called taskSpawn? (i have a pool of tasks used for different jobs and i want the name of task reflect what job the task is currently executing. ). - -- regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Mon, 27 May 2002 12:12:47 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CF2232F.3000104@spamm.me.l8s.co.uk> References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> <3CEB655C.306@bigfoot.com> <3cebb3b5.89154657@news.mhogaming.com> <3CF0A5AD.54901751@bigfoot.com> <3CF12D79.3040106@spamm.me.l8s.co.uk> <3CF1C7E4.E0F6DF34@bigfoot.com> > time_in_minutes_from_UTC must be positive. The FM is wrong! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: routine called from kernel/application Date: Mon, 27 May 2002 14:30:16 +0200 From: "Werner Schiendl" Message-ID: <3cf226b2$1@brateggebdc5.br-automation.co.at> References: <3CEA05AF.61DE7E42@europem01.nt.com> Hi, the easiest in my opinion is to separate your application from the VxWorks image for that. The you know where the VxWorks image is in memory and where your application is. Look at the moduleShow command to find our where your image goes. hth Werner "Sebastien Fauris" wrote in message news:3CEA05AF.61DE7E42@europem01.nt.com... > Hi vxWorkers, > > having a program counter value, I would like to identify if the > current routine is being executed from a kernel or from an > applicative context (this in order to perform profiling of my code). > Is there a way to get the beginning and ending addresses of > the kernel part of the vxWorks.st and the beginning and ending > addresses of the applicative part of it ? > > Thanks for your help, > > Sebastien. > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Last Sunday of October?...... Date: Mon, 27 May 2002 14:32:33 +0200 From: "Werner Schiendl" Message-ID: <3cf2273b@brateggebdc5.br-automation.co.at> References: <6dcae13d.0205261406.1e32e6a8@posting.google.com> Hi, lookup an algorithm to calculate the weekday for a date. Should be easy to find on the net. Then, calculate the weekday for April 1st and October 31st. From this information, it should be trivial to find the first Sunday in April and the last Sunday in October. hth Werner "roswell" wrote in message news:6dcae13d.0205261406.1e32e6a8@posting.google.com... > Hi folks, > > For timezone, i need to build the TIMEZONE env. All i have is the > rule: first sunday of april, and enddate is last sunday of october. > Now, how do i know the exact data,?..any ideas how to calculte? i have > the year as well. > > thanks > roswel --------------------------- Newsgroups: comp.os.vxworks Subject: Multicast Help ? What does it take ... Date: Mon, 27 May 2002 12:44:39 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: Hello ... I am working on a terminal product using vxworks and would like some advice on the type of work required to implement IP multicast into our product. We are currently using vxWorks 5.4 ... Thanks in advance, Jim --------------------------- Newsgroups: comp.os.vxworks Subject: RealTek8139 driver for x86 board ? Date: Mon, 27 May 2002 12:53:41 GMT From: "ChanSeong Mike Kim" Organization: Dacom Message-ID: Hi everybody ! I'm testing realtek8139 chip on x86 board using 81x9 realtek driver. Now I'm testing using ftp. But after repeating test of ftp put command like: ha bi put testfile.zip bye I tested this method using batch command repeatedly. Sometimes this operation is stopped. But I don't know which function should I check in realtek driver source. I would be very helpfule if some give me any advice for this. Thank you in advance. bi ha put testfile.zip bye :start ftp -s:ftpcmd.txt -A 169.254.2.1 ftp -s:ftpcmd.txt -A 169.254.2.1 ftp -s:ftpcmd.txt -A 169.254.2.1 ftp -s:ftpcmd.txt -A 169.254.2.1 goto start - -- ####################################### ChanSeong Kim OFFICE : 02-829-3391 FAX : 02-829-3400 PCS : 016-660-6457 KOREA COMPUTER BLDG. F9 395-65, SHNDAEBANG-DONG, DONGJAK-KU SEOUL, 156-010 KOREA ####################################### --------------------------- Newsgroups: comp.os.vxworks Subject: question on ncr810CtrlInit( ) Date: Mon, 27 May 2002 22:06:03 +0900 From: "bigfire" Organization: Inet Internet Services Message-ID: Hello, I'm trying to install LSI53C895 driver on my mpc755 board. When my board booted, it hang up at "ncr810CtrlInit( )" Would you give me advice? I'm using "sysScsi.c" code from mv2700 BSP. Of course "pciFindDevice()" function passed. And I configured PCI_CMD_IO_ENABLE. Thanks --------------------------- Newsgroups: comp.os.vxworks Subject: Prototyper assistance - rookie Date: Mon, 27 May 2002 13:05:17 GMT From: "Dan" Organization: RoadRunner - Carolina Message-ID: I have just recently downloaded the prototyper and am trying to set up networking w/ VxSim under W2000 (I'm pretty sure I have followed the documentation?). I installed the ULIP virtual adapter and enabled IP forwarding. I can ping the host address (90.0.0.254) from the command prompt, however the adapter cannot be enabled?. I'm able to launch twoVxSimulators with addresses 90.0.0.1 and 90.0.0.2. Attempts to ping either of these from a DOS prompt generates a timeout. Attempts to ping anything from either of the simulator shells generates a no answer. The following is from vxSim1 - -> ifShow nt (unit number 0): Flags: (0xf1) UP POINT-TO-POINT RUNNING Type: ETHERNET_CSMACD Internet address: 90.0.0.1 Destination Internet address: 0.0.0.0 Netmask 0xff000000 Subnetmask 0xff000000 Metric is 0 Maximum Transfer Unit size is 1520 0 packets received; 1 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped lo (unit number 0): Flags: (0x8069) UP LOOPBACK RUNNING ARP MULTICAST Type: SOFTWARE_LOOPBACK Internet address: 127.0.0.1 Netmask 0xff000000 Subnetmask 0xff000000 Metric is 0 Maximum Transfer Unit size is 32768 0 packets received; 0 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped value = 0 = 0x0 and from vxSim2: - -> ifShow nt (unit number 0): Flags: (0xf1) UP POINT-TO-POINT RUNNING Type: ETHERNET_CSMACD Internet address: 90.0.0.2 Destination Internet address: 0.0.0.0 Netmask 0xff000000 Subnetmask 0xff000000 Metric is 0 Maximum Transfer Unit size is 1520 0 packets received; 1 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped lo (unit number 0): Flags: (0x8069) UP LOOPBACK RUNNING ARP MULTICAST Type: SOFTWARE_LOOPBACK Internet address: 127.0.0.1 Netmask 0xff000000 Subnetmask 0xff000000 Metric is 0 Maximum Transfer Unit size is 32768 0 packets received; 0 packets sent 0 multicast packets received 0 multicast packets sent 0 input errors; 0 output errors 0 collisions; 0 dropped value = 0 = 0x0 Thanks in Advance! Dan F. --------------------------- Newsgroups: comp.os.vxworks Subject: Global variable probleme Date: 27 May 2002 06:22:25 -0700 From: pboulay@hotmail.com (Patrick Boulay) Organization: http://groups.google.com/ Message-ID: <8bb5523.0205270522.55831a82@posting.google.com> Hello, I wrote class for special output like a COUT, you use like: ConsoleMsg.out << "Message" << .... To use ConsoleMsg.out, I declare it global to use it like a COUT.. like include the .hh file, and write ConsoleMsg The problem is... everything it's correct, I declare the class like: extern ConsoleMessage ConsoleMsg; // inside the include file and ConsoleMessage ConsolsMsg; // inside the main function. But, imagine a lot of programe want to use this class, they can just include the file and use the class because in VxWorks, you can have access to all global function and variables, but imagine the day when my program will be not loaded, so the: "ConsoleMessage ConsolsMsg;" will not be there, the class will be just declare by extern ConsoleMessage ConsoleMsg.... it can't work! A solution is to load inside the kernel or something like that a module who declare global the class. I dont like this way, because if we forgot the load the module, everything will not work... If I declare the instance of out static to use it like: ConsoleMessage::out, I have an undefine symbole in the linking... How can I really use my class like a Cout? oh, I forgot to tell that I cant use NameSpace, the compiler is to old! Patrick --------------------------- Newsgroups: comp.os.vxworks Subject: Script for dependencies, Need Help Date: Mon, 27 May 2002 14:33:18 GMT From: "Davis Leeman" Organization: NBTel Internet Message-ID: How would one go about writing a script to calculate the dependencies of all projects contained within a tornado workspace? Any help would be appreciated. Thanks DJL - -- The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon, this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact the sender and delete the material from any computer. --------------------------- Newsgroups: comp.os.vxworks Subject: Help: is there DNS Server or equivalent for VxWorks ? Date: 27 May 2002 09:55:58 -0700 From: matthieu_sevestre@yahoo.fr (Matthieu Sevestre) Organization: http://groups.google.com/ Message-ID: Hi folks, I am currently searching for a solution to my problem.Here it is: - - I have a management PC that uses DHCP to get its IP address - - I have a board with VxWorks, configured as DHCP server and a web server Currently, I am able: - - to connect my board to my PC and from the PC ping the board (in fact, pinging DHCP server (let's say 4.0.0.1)) - - to connect to my embedded web site using in Internet Explorer http://4.0.0.1 Now, I would like to go further, using something like DNS because I am bored of every time searching manually the DHCP server address and changing the http:// adress. I would like to always be able to connect to my target web server with let's say http://www.myboard.com (completely private network, of course :-)) For this, I suppose that I need a DNS server on my vxWorks target that could reply to the PC: yeah man ! www.myboard.com is me ! And then returning 4.0.0.1 (in this example) as an answer ot the PC. Right now, I have not been able to find a DNS server for vxWorks. => Is there one (free if possible) somewhere or pehaps source code of a very light DNS server that could just handle this kind of request ? => Or perhaps there's another solution to do this ! This is important, because it's not me that will do the http connection but dumb users ;-) (well no so much but I prefer studying the problem considering them as dumb). Sorry for the long question, Matt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: is there DNS Server or equivalent for VxWorks ? Date: Mon, 27 May 2002 19:32:50 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: Well, Mat, I had once done a DNS cacheing-only (proxy) server for VxWorks. which did not have any of its own DNS names, but acted as server on behalf of other clients, then acted as client towards remote DNS servers (preferrably authorized for domain) for the actual info. It wasnt a port of any Unix code, but a compact cleanroon job. What YOU need, is probably a combination of this and a automatic association of one name to the target's own IP address. You need this DNS thingie to act as DNS proxy because once DHCP reply indicates the DHCP server also as DNS server, all DNS queries will come to this target board of yours, so for every quesry that does not involve the "special" name, the target will need to forward the query to the net for resolution. Dont think you will find anything of that kind on the net, but it could be a nice little project. BTW, the home gateway I got here could use this feature, but instead I havce to type its IP address that I had configured it (as I changed its default),but I dont mind, obviously. "Matthieu Sevestre" wrote in message news:d428158b.0205270855.26d03b44@posting.google.com... > Hi folks, > > I am currently searching for a solution to my problem.Here it is: > > - I have a management PC that uses DHCP to get its IP address > - I have a board with VxWorks, configured as DHCP server and a web > server > Currently, I am able: > - to connect my board to my PC and from the PC ping the board (in > fact, pinging DHCP server (let's say 4.0.0.1)) > - to connect to my embedded web site using in Internet Explorer > http://4.0.0.1 > > Now, I would like to go further, using something like DNS because I am > bored of every time searching manually the DHCP server address and > changing the http:// adress. I would like to always be able to connect > to my target web server with let's say http://www.myboard.com > (completely private network, of course :-)) > > For this, I suppose that I need a DNS server on my vxWorks target that > could reply to the PC: yeah man ! www.myboard.com is me ! And then > returning 4.0.0.1 (in this example) as an answer ot the PC. > > Right now, I have not been able to find a DNS server for vxWorks. > > => Is there one (free if possible) somewhere or pehaps source code of > a very light DNS server that could just handle this kind of request ? > => Or perhaps there's another solution to do this ! > > This is important, because it's not me that will do the http > connection but dumb users ;-) (well no so much but I prefer studying > the problem considering them as dumb). > > Sorry for the long question, > Matt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Tue, 28 May 2002 07:48:29 +0200 From: Alex Vinokur Message-ID: <3CF31A2C.30B11BA9@bigfoot.com> References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> <3CEB655C.306@bigfoot.com> <3cebb3b5.89154657@news.mhogaming.com> <3CF0A5AD.54901751@bigfoot.com> <3CF12D79.3040106@spamm.me.l8s.co.uk> <3CF1C7E4.E0F6DF34@bigfoot.com> <3CF2232F.3000104@spamm.me.l8s.co.uk> David Laight wrote: > > time_in_minutes_from_UTC must be positive. > > The FM is wrong! What does FM mean ? - -- =========================== Alex Vinokur mailto:alexvn@bigfoot.com mailto:alexvn@go.to http://up.to/alexvn http://go.to/alexv_math =========================== --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Prototyper assistance - rookie Date: 27 May 2002 22:32:13 -0700 From: overtones@excite.com (Strawberry) Organization: http://groups.google.com/ Message-ID: <41440750.0205272132.429ada93@posting.google.com> References: "Dan" wrote in message news:... > I have just recently downloaded the prototyper and am trying to set up > networking w/ VxSim under W2000 (I'm pretty sure I have followed the > documentation?). I installed the ULIP virtual adapter and enabled IP > forwarding. I can ping the host address (90.0.0.254) from the command > prompt, however the adapter cannot be enabled?. I'm able to launch > twoVxSimulators with addresses 90.0.0.1 and 90.0.0.2. Attempts to ping > either of these from a DOS prompt generates a timeout. Attempts to ping > anything from either of the simulator shells generates a no answer. > > The following is from vxSim1 > -> ifShow > nt (unit number 0): > Flags: (0xf1) UP POINT-TO-POINT RUNNING > Type: ETHERNET_CSMACD > Internet address: 90.0.0.1 > Destination Internet address: 0.0.0.0 > Netmask 0xff000000 Subnetmask 0xff000000 > Metric is 0 > Maximum Transfer Unit size is 1520 > 0 packets received; 1 packets sent > 0 multicast packets received > 0 multicast packets sent > 0 input errors; 0 output errors > 0 collisions; 0 dropped > lo (unit number 0): > Flags: (0x8069) UP LOOPBACK RUNNING ARP MULTICAST > Type: SOFTWARE_LOOPBACK > Internet address: 127.0.0.1 > Netmask 0xff000000 Subnetmask 0xff000000 > Metric is 0 > Maximum Transfer Unit size is 32768 > 0 packets received; 0 packets sent > 0 multicast packets received > 0 multicast packets sent > 0 input errors; 0 output errors > 0 collisions; 0 dropped > value = 0 = 0x0 > > and from vxSim2: > > -> ifShow > nt (unit number 0): > Flags: (0xf1) UP POINT-TO-POINT RUNNING > Type: ETHERNET_CSMACD > Internet address: 90.0.0.2 > Destination Internet address: 0.0.0.0 > Netmask 0xff000000 Subnetmask 0xff000000 > Metric is 0 > Maximum Transfer Unit size is 1520 > 0 packets received; 1 packets sent > 0 multicast packets received > 0 multicast packets sent > 0 input errors; 0 output errors > 0 collisions; 0 dropped > lo (unit number 0): > Flags: (0x8069) UP LOOPBACK RUNNING ARP MULTICAST > Type: SOFTWARE_LOOPBACK > Internet address: 127.0.0.1 > Netmask 0xff000000 Subnetmask 0xff000000 > Metric is 0 > Maximum Transfer Unit size is 32768 > 0 packets received; 0 packets sent > 0 multicast packets received > 0 multicast packets sent > 0 input errors; 0 output errors > 0 collisions; 0 dropped > value = 0 = 0x0 > > Thanks in Advance! > > Dan F. Got to tell your targets where to send the ping packets right. Check out the routes with routeShow. Add the default route using routeAdd("0.0.0.0","90.0.0.254") on both your targets.. cheers, Strawberry (my fav...) --------------------------- Newsgroups: comp.os.vxworks Subject: why only Variables and BackTrace Debug windows do not show Date: 27 May 2002 22:43:02 -0700 From: mgalecki@sbs.com (Mark Galecki) Organization: http://groups.google.com/ Message-ID: <585c4698.0205272143.3c94b9@posting.google.com> Hello, I am using Tornado 2.0/VxWorks 5.4 and the Tornado debugger. I build my project with the "-g" option so I could see all the Debug windows. I write "could" because suddenly I can no longer use the BackTrace and variables windows. The BackTrace says "unavailable" and the Variables window is empty. I can use the Watch window just fine, so I reason the project builds properly including the debug symbols. I rebuilt and rebooted everything, and the problem persists. I am interrupting the debugger manually or stopping on a breakpoint. Can anyone give me a hint what could be the problem? Thank you, Mark Galecki --------------------------- Newsgroups: comp.os.vxworks Subject: Tornado build: external (console) windows flashes Date: Tue, 28 May 2002 10:28:27 +0200 From: "Romano Signorelli" Organization: Centro Servizi Interbusiness Message-ID: Hello, when I start a build, Tornado flashes an external black (console) window every time it exec a program like cc386, etc. Output is redirected correctly to the internal build window, but the console windows appeairing/disappearing are very annoying. My host is a Windows 2000 PC. Any suggestion? Thanks in advance, bye. Romano Signorelli --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado build: external (console) windows flashes Date: Tue, 28 May 2002 11:30:24 +0200 From: "Werner Schiendl" Message-ID: <3cf34e07$1@brateggebdc5.br-automation.co.at> References: Hi, do you have any build rules added to the file? which BSP are you using? Maybe a look into the makefile clarifies what causes the additional console window to be opened... I used pc486 and pcPentium BSPs on a Windows 2000 host and did not notice such a behaviour. hth Werner "Romano Signorelli" wrote in message news:acvf3k$naj$1@fe1.cs.interbusiness.it... > Hello, > when I start a build, Tornado flashes an external black (console) window > every time it exec a program like cc386, etc. > Output is redirected correctly to the internal build window, but the console > windows appeairing/disappearing are very annoying. > My host is a Windows 2000 PC. > > Any suggestion? > Thanks in advance, bye. > Romano Signorelli > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: multiple Boot Date: Tue, 28 May 2002 10:43:24 +0100 From: Phil Ashby Organization: IDL Message-ID: <3CF3513C.AD38B95A@idl.bt.co.uk> References: james wrote: > > hi VxWorkers, > > What exactly do we have to change in the BootConfig.c to have the > choice of multiple booting ? Can anyone just brief on that. Depends what sort of multiple booting you need, bootConfig.c normally reads the boot line from the BSP library (sysBootLine?), cracks it into the various boot parameters, then uses them to load the system. If you want a selection of boot lines, perhaps you could adapt the BSP to store >1 boot line in non-volatile store, then bootConfig.c could read all of them, presenting the user with a choice at startup. You could read the boot line(s) from a remote system (probably where the system images are stored) - how about an HTTP request to get a directory listing of the bootable systems? This would use most of the boot line parameters to locate the HTTP server, then present the user with a list of files to choose from... If you want to multiple boot different operating systems, you need to understand how such operating systems are started on your hardware (eg: load first disk sector into memory at 0x7C000 for DOS on x86), then use the existing boot libraries (eg: ATA drivers) to perform the necessary operations (DOS would be ugly as it requires to start in real mode on an x86 - some assembly required for this!) HTH, - -- Phil "Phlash" Ashby (speaking only for himself) Work: http://www.internet-designers.net/ Home: http://www.ashbysoft.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Global variable probleme Date: Tue, 28 May 2002 10:49:24 +0100 From: Phil Ashby Organization: IDL Message-ID: <3CF352A4.26C7F994@idl.bt.co.uk> References: <8bb5523.0205270522.55831a82@posting.google.com> Patrick Boulay wrote: > But, imagine a lot of programe want to use this class, they can just > include the file and use the class because in VxWorks, you can have > access to all global function and variables, but imagine the day when > my program will be not loaded, so the: "ConsoleMessage ConsolsMsg;" > will not be there, the class will be just declare by extern > ConsoleMessage ConsoleMsg.... it can't work! > > A solution is to load inside the kernel or something like that a > module who declare global the class. I dont like this way, because if > we forgot the load the module, everything will not work... This is your only option Patrick - load your support classes first in a startup script, or link them into the kernel so they are available in the system symbol table. - -- Phil "Phlash" Ashby (speaking only for himself) Work: http://www.internet-designers.net/ Home: http://www.ashbysoft.com/ --------------------------- Newsgroups: comp.os.vxworks Subject: Problem with IP Networking in Prototyper Date: 28 May 2002 03:07:04 -0700 From: linsonaugustine@yahoo.co.in (Linson) Organization: http://groups.google.com/ Message-ID: Hi, I am a rookie VxWorks programmer currently working on a Prototyper version.I have been unable to get networking to work in VxWorks.Kindly advice me as to what settings I should tweak to enable networking in the prototyper.I need to work only in a single machine .I plan to develop an application using a vc++ application as the front-end and intend to enable communication between the vxworks and vc portions by socket programming. Kindly help me out Linson.P.A --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Tornado build: external (console) windows flashes Date: Tue, 28 May 2002 12:11:43 +0200 From: "Romano Signorelli" Organization: Centro Servizi Interbusiness Message-ID: References: <3cf34e07$1@brateggebdc5.br-automation.co.at> I'm using pentium BSP. Yes, I know that it would work well. My collegue has tha same installation of mine and he doesn't have problems. Maybe some Windows registry hidden settings? Romano "Werner Schiendl" ha scritto nel messaggio news:3cf34e07$1@brateggebdc5.br-automation.co.at... > Hi, > > do you have any build rules added to the file? > which BSP are you using? > > Maybe a look into the makefile clarifies what causes the additional console > window to be opened... > > I used pc486 and pcPentium BSPs on a Windows 2000 host and did not notice > such a behaviour. > > hth > Werner > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with IP Networking in Prototyper Date: Tue, 28 May 2002 06:31:14 -0400 From: joe durusau Organization: WEBUSENET.com Message-ID: <3CF35C72.9A0AFE4F@bellsouth.net> References: Linson wrote: > Hi, > I am a rookie VxWorks programmer currently working on a Prototyper > version.I have been unable to get networking to work in VxWorks.Kindly > advice me as to what settings I should tweak to enable networking in > the prototyper.I need to work only in a single machine .I plan to > develop an application using a vc++ application as the front-end and > intend to enable communication between the vxworks and vc portions by > socket programming. > Kindly help me out > Linson.P.A Did you rebuild the simulator? By default it doesn't have networking, or at least so I have been told. Speaking only for myself, Joe Durusau --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Tue May 28 04:51:20 2002 From: "anshuman bhatt" Date: Tue May 28 04:51:23 PDT 2002 Subject: How is multi level interrupt supported in VxWorks Hi I am new to Embedded System and am learning VxWorks. I have a question: What is multi-level interrupt,how is it supported by VxWorks,what is the use and does any other operating system support multi-level interrupt. Regards Anshuman Bhatt _________________________________________________________ Click below to visit monsterindia.com and review jobs in India or Abroad http://monsterindia.rediff.com/jobs From vxwexplo-errs@csg.lbl.gov Tue May 28 15:32:03 2002 From: "Dr. Jifang Tian" Date: Tue May 28 15:32:06 PDT 2002 Subject: help needed on "NO FREE FILE DESCRIptors" Hi, the vxWorks experts: I had a problem with the limitation of file descriptors in my target box running vxWorks (PPC). When the number of file descriptors (of SCSI logical units) exceeds about 48, the error " NO FREE FILE DESCRIPTORS" occurs. I really appreciate if you can give me a suggestion that can increase that limit . Best Regards. Jifang From vxwexplo-errs@csg.lbl.gov Tue May 28 17:57:47 2002 From: Mike Anderson Date: Tue May 28 17:57:49 PDT 2002 Subject: Re: help needed on "NO FREE FILE DESCRIptors" VxWorks Greetings! On Tue, 2002-05-28 at 18:52, the vxWorks Users Group Exploder wrote: > Submitted-by vxwexplo-errs@csg.lbl.gov Tue May 28 15:32:03 2002 > Submitted-by: "Dr. Jifang Tian" > > Hi, the vxWorks experts: > > I had a problem with the limitation of file descriptors in my > target box running vxWorks (PPC). When the number of > file descriptors (of SCSI logical units) exceeds about 48, > the error " NO FREE FILE DESCRIPTORS" occurs. I > really appreciate if you can give me a suggestion that > can increase that limit . > Increase the #define NUM_FILES. By default it's 50, but you can increase it to virtually any number. However, by default, the maximum that will be available in the select() call is 256. WRS has a patch that increases this number as well. HTH, Mike -- ======================================================= T h e P T R G r o u p, I n c. ======================================================= Embedded, Real-Time Solutions and Training Mike Anderson mailto:mike@ThePTRGroup.com Chief Scientist http://www.ThePTRGroup.com C: (703) 585-9384 V: (703) 430-3748 ======================================================= "Software development is like making a baby... You can't make a baby in one month by impregnating nine women. Some things just take time." From vxwexplo-errs@csg.lbl.gov Wed May 29 04:03:14 2002 From: Vxworks Exploder Date: Wed May 29 04:03:17 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Wed May 29 04:03:10 PDT 2002 Subject: how is multi-level interrupt supported by VxWorks Subject: Any known vxWorks 5.3.1 MIPS ld/loadModuleAt alignment issues? Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Subject: VxWorks MIB-II SNMP Subject: Re: Any known vxWorks 5.3.1 MIPS ld/loadModuleAt alignment issues? Subject: bootLoadModule loading general elf Subject: Re: Correct method to change IP addr? Subject: About makeSymTlb Error - usage: mksymtab architecture infile - Why? Subject: Re: VME bus errors and signal handlers Subject: Re: Help: is there DNS Server or equivalent for VxWorks ? Subject: Re: Help: is there DNS Server or equivalent for VxWorks ? Subject: Re: check CPM on SBC8260 Subject: Re: Task::setName AFTER taskSpawn Subject: Re: VME bus errors and signal handlers Subject: Re: why only Variables and BackTrace Debug windows do not show Subject: Account Reset Subject: Re: multiple Boot Subject: Re: Problem with IP Networking in Prototyper Subject: TFFS question Subject: Re: Problem with IP Networking in Prototyper Subject: Re: bootLoadModule loading general elf Subject: Re: Account Reset Subject: Re: multiple Boot ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: how is multi-level interrupt supported by VxWorks Date: 28 May 2002 05:33:02 -0700 From: bhattanshuman@rediffmail.com (anshuman bhatt) Organization: http://groups.google.com/ Message-ID: Hi I am new to Embedded System and am learning VxWorks. I have question: What is multi-level interrupt,how VxWorks support it, what is it use? Does any other operating system support it. Regards Anshuman Bhatt --------------------------- Newsgroups: comp.os.vxworks Subject: Any known vxWorks 5.3.1 MIPS ld/loadModuleAt alignment issues? Date: 28 May 2002 05:34:44 -0700 From: jeffrey.m.creem@baesystems.com (Jeffrey Creem) Organization: http://groups.google.com/ Message-ID: Is anyone aware of any open SPRs related the vxWorks loader occasionally getting confused and mis-aligning a segment and/or page in certain instances on the MIPS target (vxworks 5.3.1) I've got an application where some small changes in the source code end up with some of the symbols being only 4 byte aligned when they need to be (and the compiler is expecting that they are ) 8 byte aligned. Once started, the application bombs out on a ldc1 $f6, 0(t4) Where t4 contains the address of a symbol that was loaded into the target via an ld < my_file.elf (I have also tried several combinations of loadModuleAt with the same results) In the bad case, t4 contains an address that is is only longword aligned. I am leaning towards this being a vxWorks loader issue instead of a host side compiler/linker issue since in both the "good" case and the "bad" case the offset of the symbol in the elf file (as shown by nm) is 8 byte aligned. I do not know enough about what the run time linker/loader is allowed to do to be sure where the problem really. It would be nice to know if I need to beat up the compiler vendor or Wind River for a solution. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to set TIMEZONE? Support for timezones in localtime? Date: Tue, 28 May 2002 12:41:19 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CF37B60.1020300@spamm.me.l8s.co.uk> References: <3CB21456.E91B3577@bitworkssystemsDELETETHIS.com> <3CEB655C.306@bigfoot.com> <3cebb3b5.89154657@news.mhogaming.com> <3CF0A5AD.54901751@bigfoot.com> <3CF12D79.3040106@spamm.me.l8s.co.uk> <3CF1C7E4.E0F6DF34@bigfoot.com> <3CF2232F.3000104@spamm.me.l8s.co.uk> <3CF31A2C.30B11BA9@bigfoot.com> >>The FM is wrong! >> > > What does FM mean ? As in RTFM :-) --------------------------- Newsgroups: comp.os.vxworks Subject: VxWorks MIB-II SNMP Date: 28 May 2002 05:42:58 -0700 From: anubhavsaksena@YAHOO.COM (Anubhav Saksena) Organization: http://groups.google.com/ Message-ID: <450ab632.0205280442.46478b58@posting.google.com> Hi, I have a query regarding implementation of MIB-II in VxWorks. We are using m2lib Library of VxWorks for instrumenting MIB-II, but I have doubt regarding IfTable, Is an operating system (in our case VxWorks) capable of knowing its interfaces automatically ? ....ok some interfaces it might know but, is it capable of knowing logical interfaces such as SONET PATH or SONET VT ? I am unable to test these things using on a simulator. On simulator loopback interface (should be there) , ethernet interface (i think due to ULIP) are visible. But rest I am not sure whether VxWorks is capable of knowing the interfaces automatically. regards Anubhav --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Any known vxWorks 5.3.1 MIPS ld/loadModuleAt alignment issues? Date: Tue, 28 May 2002 12:51:17 +0000 (UTC) From: David Laight Organization: BT Openworld Message-ID: <3CF37DB2.5000308@spamm.me.l8s.co.uk> References: > I am leaning towards this being a vxWorks loader issue instead of a > host side compiler/linker issue since in both the "good" case and the > "bad" case > the offset of the symbol in the elf file (as shown by nm) is 8 byte > aligned. At least that is right! coff doesn't support an alignment attribute, so even though the compiler would have requested 8 byte alignment, the linker would ignore it! > I do not know enough about what the run time linker/loader is allowed > to do to be sure where the problem really. It would be nice to know if > I need to beat up the compiler vendor or Wind River for a solution. > Does rather stink of the run time loader just using malloc() and getting a chunk of memory that is on an odd 4-byte boundary. Interestingly since memory returned by malloc() is expected to be suitable aligned for ANY use, then malloc itself may be broken! Of course memPartAlignedAlloc() could be used instead..... Maybe a quick check on the alignment of items returned by malloc() is acled for - maybe dumping the freelist will be enough. David --------------------------- Newsgroups: comp.os.vxworks Subject: bootLoadModule loading general elf Date: 28 May 2002 05:53:11 -0700 From: leonardmatherson@hotmail.com (leonard matherson) Organization: http://groups.google.com/ Message-ID: Hello everyone, I have an evaluation board with a flashed bootrom image. Usually, I download the vxworks image from the host. Now, instead of the standard vxworks image, I need to download an image that has not been produced in the vxworks world. This image can range from a simple program to a complex application. I can generate this image in ELF format, but will this match the syntax requirements of bootLoadModule? I do not want to burn the flash again, so I must use the standard supplied bootLoadModule. TIA Leonard Matherson. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Correct method to change IP addr? Date: 28 May 2002 06:36:49 -0700 From: techsoul2001@yahoo.com (raghuram) Organization: http://groups.google.com/ Message-ID: References: <3CF1495E.1010106@yahoo.com> Carl, Try the following. You should first bring the IF down, change the address and then bring the IF up : ifFlagChange("dc0", IFF_UP, FALSE); /* take interface down */ ifAddrSet("dc0", "nnn.nnn.nnn.nnn"); /* change IP address */ ifFlagChange("dc0", IFF_UP, TRUE); /* bring interface up */ I got the above suggestion from this forum and it worked fine. HTH, raghuram Iubica2 wrote in message news:<3CF1495E.1010106@yahoo.com>... > Are you trying to bind() a TCP server socket? bind() may fail because > the TCP/IP stack has a lingering TCP server socket bound at the same IP > address/port that you're trying to bind to. > > After the application-level code has called close() on a TCP server > socket, the socket still lingers on for a while in the TCP/IP stack > until the incoming connections get closed. This is because closing a TCP > connection requires (in each direction!) sending a FIN, receiving the > FIN's ACK, sending the ACK to the FIN's ACK, then waiting for 30 secs or > so (an OS-dependent constant) to make sure the far end does not > retransmit the FIN's ACK. That's why the close()d socket may linger in > the TCP/IP stack for 30 secs or more before being really closed at the > level of the TCP/IP stack. > > > Carl Petersen wrote: > > > Is this the correct sequence to change the IP addr of an interface? > > > > ifRouteDelete(...); > > > > ifMaskSet(...); > > > > ifAddrSet( ifName, "New IP" ); > > > > These calls run without error however a subsequent call to bind() fails. > > > > I have also tried ipDetach() then ipAttach() without success. Note that > > the interface is operational before I attempt to modify the IP addr. > > > > Tornado 2.02 Pentium BSP > > > > Thanks > > > > Carl Petersen > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: About makeSymTlb Error - usage: mksymtab architecture infile - Why? Date: 28 May 2002 09:06:21 -0700 From: shuliang@ms8.hinet.net (shuliang) Organization: http://groups.google.com/ Message-ID: Dear sir: Please help me or give me some ideas about the below questions, thanks in advanced. I really have no ideas to do, why it happens the bellow error? c:/Tornado/host/x86-win32/bin/makeSymTbl VX_CPU_FAMILY=ppc vxWorks.temp > SymTbl.c usage: mksymtab architecture infile make[3]: *** [symTbl.c] Error 1 I use Tornado over win32, cygwin development environment/ target processor MPC8240 / vxWork thank you very much! ShuLiang --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: VME bus errors and signal handlers Date: Tue, 28 May 2002 11:43:35 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3CF3CFD7.E3693331@lmco.com> References: <3b4d40d1.0205230234.1eaad47b@posting.google.com> <3CED8891.78C4FAC8@bellsouth.net> <3cee8cdc$0$4544@dingus.crosslink.net> In summary, then are you saying that the O.P. cannot possibly meet the requirement of single-processor reboot in a multi-processor system (at least not with a powerPC)?? Speaking only for myself, Joe Durusau William Dennen wrote: > > Almost, but not quite, correct. > Bus errors, returned as a bus error, result in a machine check in PowerPC > processors. As such they are irrecoverable as all context is lost when the > exception is handled. vxMemProbe and like routines do sneaky things like > single byte DMA transfers so the MPU can recover. > > -- > >@< > Bill Dennen > MCG/GTAC > "joe durusau" wrote in message > news:3CED8891.78C4FAC8@bellsouth.net... > > > > > > Chris Winstanley wrote: > > > > > > > > > <> > > > > > One of the processors (call it P1) transfers data over the VME bus > > > into the local memory of the other two processors (call them P2 and > > > P3) and also reads data from the local memory of P2 and P3. The > > > default bus configuration is posted writes. > > > > > > Our customer has placed a requirement on us to be able to reboot a > > > processor without affecting the other 2 processors. > > > > > > We have found that rebooting either P2 or P3 causes a VME bus error > > > (as P1 is trying to communicate with P2 and P3), resulting in an > > > exception being raised on P1, suspending the communicating task (our > > > applications, being safety related, are constrained to being a single > > > task). > > > > > > > <> > > > > One problem that I've been told exists with most powerPC > > CPUs is that due to internal pipelining they cannot recover the exact > > machine state when a bus error occurs. Perhaps your boards do > > not suffer from this problem, but I think that you should check with > > the manufacturer to see whether recovery from a bus error can > > be done. > > > > Speaking only for myself, > > > > Joe Durusau > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help: is there DNS Server or equivalent for VxWorks ? Date: 28 May 2002 10:27:29 -0700 From: pkockritz@mac.com (Pete Kockritz) Organization: http://groups.google.com/ Message-ID: <9423c55b.0205280927.7c59ad30@posting.google.com> References: matthieu_sevestre@yahoo.fr (Matthieu Sevestre) wrote in message news:... > > For this, I suppose that I need a DNS server on my vxWorks target that > could reply to the PC: yeah man ! www.myboard.com is me ! And then > returning 4.0.0.1 (in this example) as an answer ot the PC. My their nature, DNS servers have static addresses. Since they are used to resolve hostnames to IP addresses, hosts must be able to contact them by address. How would your PC connect to the DNS on your vxWorks host to get the address that it needs? What you really need is for your DHCP server to communicate with the DNS server to inform it of the address that it has given to your vxworks target. Sometimes the DCHP and DNS servers are the same machine. Talk to your network admin and see if he can have the DHCP and DNS server share this information. Pete - -- pkockritz at mac dot com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Help: is there DNS Server or equivalent for VxWorks ? Date: Tue, 28 May 2002 17:45:25 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <9423c55b.0205280927.7c59ad30@posting.google.com> "Pete Kockritz" wrote in message news:9423c55b.0205280927.7c59ad30@posting.google.com... > matthieu_sevestre@yahoo.fr (Matthieu Sevestre) wrote in message news:... > > > > For this, I suppose that I need a DNS server on my vxWorks target that > > could reply to the PC: yeah man ! www.myboard.com is me ! And then > > returning 4.0.0.1 (in this example) as an answer ot the PC. > > My their nature, DNS servers have static addresses. Since they are > used to resolve hostnames to IP addresses, hosts must be able to > contact them by address. How would your PC connect to the DNS on your > vxWorks host to get the address that it needs? a DHCP response would also include the IP address of DNS servers, and it would be the IP address of the DHCP server itself in this case. All packets includng DNS requests and responses would go through this particular board (RNG) anyway, catching DNS at application level is not going to impact performance. > > What you really need is for your DHCP server to communicate with the > DNS server to inform it of the address that it has given to your > vxworks target. Sometimes the DCHP and DNS servers are the same > machine. Talk to your network admin and see if he can have the DHCP > and DNS server share this information. You refer to Dynamic DNS, but the original posting was asking for something completely different. My gest guess is this is about developing a Residantial Network Gateway (RNG) which also acts as DHCP server (mine does) and typically is a router withj NAT and ACL. The question is about adding minimal DNS capability for easier cunsumer setup. Naturally, there is no network administrators on most residential / SOHO networks (mine is an exception). - - Leonid --------------------------- Newsgroups: comp.os.vxworks Subject: Re: check CPM on SBC8260 Date: Tue, 28 May 2002 11:47:38 -0700 From: "Dan Gold" Organization: Mailgate.ORG Server - http://www.Mailgate.ORG Message-ID: References: <5288f97d.0205261142.7790a7ac@posting.google.com> Sender: gold@ensemble.com "Danny Cohen" wrote in message news:5288f97d.0205261142.7790a7ac@posting.google.com... > I'd like to check the CPM interface, and make an application that > receive UDP datagrams using the SCC and DMA them into the SDRAM than > loopback using another DMA channel and transmit them to the IP > network. I'm a little confused as to exactly what you want to do. They say a picture is worth a 1000 words, one might be useful here. Are you just trying to make sure the CPM is alive ? Why not just send it a RESET command see that it clears the command flag ? Why not just use an SCC configured as a serial port in loopback mode ? So you want to do it via Ethernet ? On the 8260 website, they have an FCC example that does internal loopback. That might be a good starting point. Either use it or convert it to an SCC... So you're going to receive UDP datagrams from where ? A real IP network ? Ok, so you receive them on an SCC. The CPM will use DMA (SDMA really) to put them into SDRAM. That's the first time you'll have access to them. Then you want to use loopback ? On what ? The SCC ? You can't have it receiving packets from a real network *and* in loopback mode. Do you want to receive on one SCC, then send out another SCC ? or FCC ? To the same IP network that the first SCC received the UDP packet on ? A different one ? You don't have your choice on which "DMA channel" to use (unless you're using IDMA), if you're using a peripheral to Tx/Rx your packets. What is it exactly you are trying to do ? hth, Dan Gold gold@ensemble.com --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Task::setName AFTER taskSpawn Date: Tue, 28 May 2002 13:55:39 -0500 From: Dinker Charak Organization: Fermi National Accelerator Laboratory Message-ID: <3CF3D2AB.B017C7A7@fnal.gov> References: hi, the name is stored in WIND_TCB struct ...there is a pointer in it to the name ... typedef struct windTcb { ... char * name; ... } WIND_TCB; you can use it to overwrite the name ... but be careful and do not write more than the original length of the name when doing it this way ... use this to get the TCB WIND_TCB *taskTcb(int tid) hth, dinker Mario Semo wrote: > > Hello, > > Is it possible (and if : how) to change the name of a task AFTER i called > taskSpawn? (i have a pool of tasks used for different jobs and i want the > name of task reflect what job the task is currently executing. ). > > -- > regards, Mario Semo. http://www.kirchnersoft.com --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: VME bus errors and signal handlers Date: 28 May 2002 12:15:01 -0700 From: cnw@zoom.co.uk (Chris Winstanley) Organization: http://groups.google.com/ Message-ID: <3b4d40d1.0205281115.6a61ad19@posting.google.com> References: <3b4d40d1.0205230234.1eaad47b@posting.google.com> <3CED8891.78C4FAC8@bellsouth.net> <3cee8cdc$0$4544@dingus.crosslink.net> Hello again. Firstly thank you for all your replies, its much appreciated. Secondly, I have another question. If I use the vxMemProbe function, do I need to turn off posted writes to ensure that any bus error is handled by the vxMemProbe function and doesn't propogate into my own code? Or does vxMemProbe in some way deal with this for me (perhaps by the use of DMA)? Thirdly, I have had a response from one person who has successfully used a signal handler attached to SIGBUS and the C long jump functions to deal with bus errors. This seems at odds with your statement that such errors are irrecoverable? Thanks, Chris. "William Dennen" wrote in message news:<3cee8cdc$0$4544@dingus.crosslink.net>... > Almost, but not quite, correct. > Bus errors, returned as a bus error, result in a machine check in PowerPC > processors. As such they are irrecoverable as all context is lost when the > exception is handled. vxMemProbe and like routines do sneaky things like > single byte DMA transfers so the MPU can recover. > > -- > >@< > Bill Dennen > MCG/GTAC > "joe durusau" wrote in message > news:3CED8891.78C4FAC8@bellsouth.net... > > > > > > Chris Winstanley wrote: > > > > > > > > > <> > > > > > One of the processors (call it P1) transfers data over the VME bus > > > into the local memory of the other two processors (call them P2 and > > > P3) and also reads data from the local memory of P2 and P3. The > > > default bus configuration is posted writes. > > > > > > Our customer has placed a requirement on us to be able to reboot a > > > processor without affecting the other 2 processors. > > > > > > We have found that rebooting either P2 or P3 causes a VME bus error > > > (as P1 is trying to communicate with P2 and P3), resulting in an > > > exception being raised on P1, suspending the communicating task (our > > > applications, being safety related, are constrained to being a single > > > task). > > > > > > > <> > > > > One problem that I've been told exists with most powerPC > > CPUs is that due to internal pipelining they cannot recover the exact > > machine state when a bus error occurs. Perhaps your boards do > > not suffer from this problem, but I think that you should check with > > the manufacturer to see whether recovery from a bus error can > > be done. > > > > Speaking only for myself, > > > > Joe Durusau > > > > > > --------------------------- Newsgroups: comp.os.vxworks Subject: Re: why only Variables and BackTrace Debug windows do not show Date: Tue, 28 May 2002 12:28:28 -0700 From: "Mark Galecki" Message-ID: <3cf3da55$0$16206$724ebb72@reader2.ash.ops.us.uu.net> References: <585c4698.0205272143.3c94b9@posting.google.com> Stephane Ross advised to "try double-clicking in the backtrace window (on "unavailiable" for exemple...) This should do the trick. It looks that sometimes, the windows are not correctely updated after you attach to the suspended task. Probably another obscure TCL script problem...". And Lo and Behold, it worked for the BackTrace window - it came to life when I double click on the "unavailable" word. But what about the Variables window??? Mark "Mark Galecki" wrote in message news:585c4698.0205272143.3c94b9@posting.google.com... > Hello, > I am using Tornado 2.0/VxWorks 5.4 and the Tornado debugger. I build > my project with the "-g" option so I could see all the Debug windows. > I write "could" because suddenly I can no longer use the BackTrace and > variables windows. The BackTrace says "unavailable" and the Variables > window is empty. I can use the Watch window just fine, so I reason > the project builds properly including the debug symbols. I rebuilt > and rebooted everything, and the problem persists. I am interrupting > the debugger manually or stopping on a breakpoint. > > Can anyone give me a hint what could be the problem? > > Thank you, > Mark Galecki --------------------------- Newsgroups: comp.os.vxworks Subject: Account Reset Date: 28 May 2002 17:54:52 -0700 From: cferreira@redecidade.com.br (Claudio Ferreira) Organization: http://groups.google.com/ Message-ID: <1ba79fe8.0205281654.517b997d@posting.google.com> Hi ! Sorry, I dont speak english, but I try. Please, Im new user of VxWorks. I need books for download. First, need to reset a account. Forget Pass Word. Or crate new. Can Anybody help ? Thanks ! Cláudio Ferreira Brazil - Rio de Janeiro --------------------------- Newsgroups: comp.os.vxworks Subject: Re: multiple Boot Date: 28 May 2002 21:13:40 -0700 From: be_gyus@yahoo.com (james) Organization: http://groups.google.com/ Message-ID: References: <3CF3513C.AD38B95A@idl.bt.co.uk> Phil Ashby wrote in message news:<3CF3513C.AD38B95A@idl.bt.co.uk>... > james wrote: > > > > hi VxWorkers, > > > > What exactly do we have to change in the BootConfig.c to have the > > choice of multiple booting ? Can anyone just brief on that. > > Depends what sort of multiple booting you need, bootConfig.c normally > reads the boot line from the BSP library (sysBootLine?), cracks it > into the various boot parameters, then uses them to load the system. > > If you want a selection of boot lines, perhaps you could adapt the > BSP to store >1 boot line in non-volatile store, then bootConfig.c > could read all of them, presenting the user with a choice at startup. > > You could read the boot line(s) from a remote system (probably where > the system images are stored) - how about an HTTP request to get a > directory listing of the bootable systems? This would use most of the > boot line parameters to locate the HTTP server, then present the > user with a list of files to choose from... > > If you want to multiple boot different operating systems, you need > to understand how such operating systems are started on your hardware > (eg: load first disk sector into memory at 0x7C000 for DOS on x86), > then use the existing boot libraries (eg: ATA drivers) to perform > the necessary operations (DOS would be ugly as it requires to start > in real mode on an x86 - some assembly required for this!) > > HTH, *************************************************************************** Hi, Thanks for the suggestion , but our problem is bit specific, we have a harddisk , a floppy drive and an ethernet card . We need to set an option for the system to boot accordingly depending on the device availability, like the system should detect a floppy drive first ,for the vxWorks image then ,if unavailable it should look for the other devices specified in the config.h , say hardisk / ethernet card. How exactly should we specify these Parameters in the config.h ?? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with IP Networking in Prototyper Date: 28 May 2002 21:17:09 -0700 From: linsonaugustine@yahoo.co.in (Linson) Organization: http://groups.google.com/ Message-ID: References: <3CF35C72.9A0AFE4F@bellsouth.net> We are using the prototyper on a WinME system and need to be able to send and receive data to and from a VC++-developed User Interface.We are trying to implement the same using socket programming but all programs failed to download in VxWorks with an error message citing the non-availability of _socket,_connect etc.We are using only a single system for the project...any suggestions as to how to overcome this difficulty? --------------------------- Newsgroups: comp.os.vxworks Subject: TFFS question Date: 29 May 2002 00:11:58 -0700 From: Robin_ch@china.com (Robin Tcheng) Organization: http://groups.google.com/ Message-ID: <3b9cf6e4.0205282311.5769f6fa@posting.google.com> I have a SST 39vf160 flash: size----------2M bytes/chip nmber of chip-------1 port size-----8 i want to set up TFFS on it when using MTDs,the current configuration is not working(the address of erase does not change at all). Here is my settings: flSetWindowBusWidth(vol.socket,16);/* use 16-bits */ flSetWindowSpeed(vol.socket,120); /* 120 nsec. */ flSetWindowSize(vol.socket,2); /* 8 KBytes */ vol.mtdVars = &mtdVars[flSocketNoOf(vol.socket)]; thisVars->unlockAddr1 = NO_UNLOCK_ADDR; vol.interleaving = 1; vol.chipSize = 0x200000l; flSetWindowSize(vol.socket,8 * vol.interleaving); thisVars->unlockAddr1 = 0x5555u * vol.interleaving; thisVars->unlockAddr2 = 0xaaaau * vol.interleaving; thisVars->baseMask = 0xfff80000L * vol.interleaving; vol.noOfChips = 1; vol.erasableBlockSize = 0x10000l * vol.interleaving; vol.flags |= SUSPEND_FOR_WRITE; /* Register our flash handlers */ vol.write = myMTDWrite; vol.erase = myMTDErase; please tell me the reason of faliure or modify the settings. thanks!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Problem with IP Networking in Prototyper Date: Wed, 29 May 2002 10:38:31 +0200 From: "Werner Schiendl" Message-ID: <3cf4935c$1@brateggebdc5.br-automation.co.at> References: <3CF35C72.9A0AFE4F@bellsouth.net> Hi Linson, rebuilding your application is not sufficient to create a simulator with networking support. The pre-built image - AFAIK - does NOT include networking - as Joe already stated. Thus the error message when downloading your application. Create a new bootable image project, based on the one shipped with Tornado. Add the network support to that image build it, and select the resulting VxWorks.exe as the simulator (custom built simulator) when starting the sim. You will find lots of more detailled descriptions on how exactly do that when searching this newsgroup. Try groups.google.com for example. hth Werner "Linson" wrote in message news:ff09eb35.0205282017.1b56b0fb@posting.google.com... > We are using the prototyper on a WinME system and need to be able to > send and receive data to and from a VC++-developed User Interface.We > are trying to implement the same using socket programming but all > programs failed to download in VxWorks with an error message citing > the non-availability of _socket,_connect etc.We are using only a > single system for the project...any suggestions as to how to overcome > this difficulty? --------------------------- Newsgroups: comp.os.vxworks Subject: Re: bootLoadModule loading general elf Date: Wed, 29 May 2002 11:44:03 +0200 From: "Werner Schiendl" Message-ID: <3cf4a2b7$1@brateggebdc5.br-automation.co.at> References: Hi, I don't know for ELF format, since x86 uses a.out. Anyway I think the only thing that is really important for the bootroms to work is that the memory layout and the application entry point is correct. I may be wrong, but I think that the application entry point is compiled into the bootrom as a fixed value, not taken from the loaded image. Thus if you compile your image with similar memory layout than a VxWorks build and use the same entry point address, you are likely to succeed. hth Werner "leonard matherson" wrote in message news:a27b6cb9.0205280453.2b1f27c5@posting.google.com... > Hello everyone, > > I have an evaluation board with a flashed bootrom image. > > Usually, I download the vxworks image from the host. > > Now, instead of the standard vxworks image, I need to download an > image that has not been produced in the vxworks world. This image can > range from a simple program to a complex application. I can generate > this image in ELF format, but will this match the syntax requirements > of bootLoadModule? > > I do not want to burn the flash again, so I must use the standard > supplied bootLoadModule. > > TIA > > Leonard Matherson. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Account Reset Date: Wed, 29 May 2002 11:50:31 +0200 From: "Werner Schiendl" Message-ID: <3cf4a43c$1@brateggebdc5.br-automation.co.at> References: <1ba79fe8.0205281654.517b997d@posting.google.com> Hi Claudio, which books do you want to download? The manuals for Tornado and/or VxWorks? If you've installed Tornado on your PC this should not be necessary. The manuals are part of the install (at least for bought versions, I don't know for the Prototyper that can be downloaded for free). What password do you need to reset? The one for WindSurf? If so, use the facilities on WindSurf to get to your password - I think there were means to send the password to the email address you specified for the account. If this does not work or you cannot find it, contact the webmaster of WindSurf website. This forum will not be able to help you here. hth Werner "Claudio Ferreira" wrote in message news:1ba79fe8.0205281654.517b997d@posting.google.com... > Hi ! > > Sorry, I dont speak english, but I try. > > Please, Im new user of VxWorks. I need books for download. First, need > to reset a account. Forget Pass Word. Or crate new. > > Can Anybody help ? > > Thanks ! > > Cláudio Ferreira > Brazil - Rio de Janeiro --------------------------- Newsgroups: comp.os.vxworks Subject: Re: multiple Boot Date: Wed, 29 May 2002 11:58:40 +0200 From: "Werner Schiendl" Message-ID: <3cf4a624$1@brateggebdc5.br-automation.co.at> References: <3CF3513C.AD38B95A@idl.bt.co.uk> Hi James, What you want cannot be done just with configuration. VxWorks bootrom as delivered by default boots the image from one location as specified in the boot parameters (boot line). For what you need, you will need to adapt the bootrom code as necessary. See the file $(WIND_BASE)/target/config/all/bootConfig.c This is the source code for the bootrom. You will need to modify the code to detect the proper boot media, setup the boot parameters and load the image. hth - - Werner "james" wrote in message > > Hi, > Thanks for the suggestion , but our problem is bit specific, we > have a harddisk , a floppy drive and an ethernet card . We need to set > an option for the system to boot accordingly depending on the device > availability, like the system should detect a floppy drive first ,for > the vxWorks image then ,if unavailable it should look for the other > devices specified in the config.h , say hardisk / ethernet card. > How exactly should we specify these Parameters in the config.h ?? --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Wed May 29 17:38:33 2002 From: Amante Mangaser Date: Wed May 29 17:38:35 PDT 2002 Subject: boot up problems for Adastra Neptune (Pentium SBC) I made a Tornado v2.02 VxWorks and bootrom_uncmp build for Neptune. When it tries to bootload the VxWorks image via an Ethernet connection I get a memPartAlloc() error from tBoot() (see attachment below). The previous build we had for Tornado v2.0 did not have such problems. I need some clue on how to eliminate this problem. Thanks in advance. A.A. Mangaser Computer Motion Inc. ================== ATTACHMENT =================================== VxWorks System Boot Copyright 1984-1998 Wind River Systems, Inc. CPU: PC PENTIUM Version: 5.4.2 BSP version: 1.2/0 Creation date: May 29 2002, 09:42:51 Press any key to stop auto-boot... 0 auto-booting... boot device : lnPci unit number : 0 processor number : 0 host name : platform file name : t:/wrs/wrs_proj/BSPs/Neptune/VxWorks_T202/default/vxWorks inet on ethernet (e) : 10.0.0.76 host inet (h) : 10.0.0.166 user (u) : demo ftp password (pw) : demo flags (f) : 0x0 target name (tn) : neptune 0x7e58c (tBoot): memPartAlloc: block too big - 6600 in partition 0x72864. 0x7e58c (tBoot): panic: no memory: clusters 0x7e58c Failed to attach (TCP/IP to device tBoot): panic: mbinitCan't load boot file!! 0x7e58c (tBoot): panic: rn_init 0x7e58c (tBoot): panic: rn_init 2 0x7e58c (tBoot): memPartAlloc: block too big - 10396 in partition 0x72864. Can't open "bootrom.{sys,dat}" [VxWorks Boot]: From vxwexplo-errs@csg.lbl.gov Thu May 30 04:03:12 2002 From: Vxworks Exploder Date: Thu May 30 04:03:14 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Thu May 30 04:03:08 PDT 2002 Subject: Re: how is multi-level interrupt supported by VxWorks Subject: Multicast? Subject: Galileo 64130 and VisionICE2 / VisionClick Subject: Multicast? Subject: Re: Multicast? Subject: Re: VME bus errors and signal handlers Subject: Ramix PMC661 with Force 680_CPCI: vxWorks 5.4 PCI configuration Subject: cs8900a Ethernet controller question Subject: How to prevent writes to PowerPC port pins from clobbering other port pins? Subject: TFFS setup failure Subject: Re: how is multi-level interrupt supported by VxWorks Subject: Re: How to prevent writes to PowerPC port pins from clobbering other port pins? Subject: stream operations on pipe Subject: timeout on WdbRPC with Cross Wind debugger Subject: any one can provide me a source code about bootloader? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: how is multi-level interrupt supported by VxWorks Date: Wed, 29 May 2002 08:10:49 -0700 From: Joe Durusau Organization: Lockheed Martin Corporation Message-ID: <3CF4EF79.4A780FCB@lmco.com> References: anshuman bhatt wrote: > > Hi > I am new to Embedded System and am learning VxWorks. > I have question: > What is multi-level interrupt,how VxWorks support it, what is it use? > Does any other operating system support it. > Regards > Anshuman Bhatt It depends on what you mean. For some platforms, vxWorks will allow an ISR to be interrupted by a higher priority interrupt. For others, you can't do that because the hardware doesn't allow that sort of thing. Most "standard" OS's don't bother, because of the overhed involved. Most OS's that refer to themselves as "real-time" do so if the hardware will allow it. Unfortunately, everybody beats terms like "multi=level interrupt" around, bu there are no universally accepted definitions. Academics tend to claim that the definition in the book "I" wrote is the only one, practitioners use the definition their employer uses, etc, etc. Speaking only for myself, Joe Durusau --------------------------- Newsgroups: comp.os.vxworks Subject: Multicast? Date: Wed, 29 May 2002 13:57:59 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: Could anyone recmmend some reading on ip multicast and vxWorks ... Thanks, Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Galileo 64130 and VisionICE2 / VisionClick Date: Wed, 29 May 2002 10:52:59 -0400 From: Michael Reed Message-ID: <3CF4EB4A.4165E677@ll.mit.edu> Hello, Has anyone used VisionICE2, VisionICE1, or VisionProbe w/ a board containing a Galileo 64130 memory controller / system controller? I'm in this position, and I'm looking for an existing register file for this combo (PPC750 and Galileo 64130). Thanks, Mike R. --------------------------- Newsgroups: comp.os.vxworks Subject: Multicast? Date: Wed, 29 May 2002 17:12:30 GMT From: "Jim" Organization: Magma Communications Ltd. Message-ID: <2_7J8.520$c6.49766@news> Could anyone recmmend some reading on ip multicast and vxWorks ... Thanks, Jim --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Multicast? Date: Wed, 29 May 2002 17:48:52 GMT From: "Leonid Rosenboim" Organization: Verio Message-ID: References: <2_7J8.520$c6.49766@news> Jim, there is really not much specific about VxWorks and Multicast. Any information source discussing berkeley sockets programming for multicast will practically do. There is much more you need to know about configuring routers and switches correctly for handling high-volume multicasts. Cheers, - - Leonid "Jim" wrote in message news:2_7J8.520$c6.49766@news... > Could anyone recmmend some reading on ip multicast and vxWorks ... > > Thanks, > Jim > > --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: VME bus errors and signal handlers Date: Wed, 29 May 2002 19:52:34 GMT From: junk@junk.net (Anton Erasmus) Organization: The South African Internet Exchange Message-ID: <3cf51c3a.1144183@news.saix.net> References: <3b4d40d1.0205230234.1eaad47b@posting.google.com> <3CED8891.78C4FAC8@bellsouth.net> <3cee8cdc$0$4544@dingus.crosslink.net> <3b4d40d1.0205281115.6a61ad19@posting.google.com> Reply-To: antera@intekom.co.za On 28 May 2002 12:15:01 -0700, cnw@zoom.co.uk (Chris Winstanley) wrote: >Hello again. > >Firstly thank you for all your replies, its much appreciated. > >Secondly, I have another question. If I use the vxMemProbe function, >do I need to turn off posted writes to ensure that any bus error is >handled by the vxMemProbe function and doesn't propogate into my own >code? Or does vxMemProbe in some way deal with this for me (perhaps by >the use of DMA)? > >Thirdly, I have had a response from one person who has successfully >used a signal handler attached to SIGBUS and the C long jump functions >to deal with bus errors. This seems at odds with your statement that >such errors are irrecoverable? > >Thanks, >Chris. Hi, I think when people say irrecoverable in this context, it means that if you are implementing a virtual memory system, and you get a bus error when accessing a location, then you cannot re-start the exact instruction that caused the problem after you have paged in the memory, without having some extra data. On systems such as the 68K. it is possible to re-start the instruction that caused the problem. If however, you want to re-start at a known point, and do not care that the read/write acces failed, then you should be able to recover. Regards Anton Erasmus > >"William Dennen" wrote in message news:<3cee8cdc$0$4544@dingus.crosslink.net>... >> Almost, but not quite, correct. >> Bus errors, returned as a bus error, result in a machine check in PowerPC >> processors. As such they are irrecoverable as all context is lost when the >> exception is handled. vxMemProbe and like routines do sneaky things like >> single byte DMA transfers so the MPU can recover. >> >> -- >> >@< >> Bill Dennen >> MCG/GTAC >> "joe durusau" wrote in message >> news:3CED8891.78C4FAC8@bellsouth.net... >> > >> > >> > Chris Winstanley wrote: >> > >> > > >> > >> > <> >> > >> > > One of the processors (call it P1) transfers data over the VME bus >> > > into the local memory of the other two processors (call them P2 and >> > > P3) and also reads data from the local memory of P2 and P3. The >> > > default bus configuration is posted writes. >> > > >> > > Our customer has placed a requirement on us to be able to reboot a >> > > processor without affecting the other 2 processors. >> > > >> > > We have found that rebooting either P2 or P3 causes a VME bus error >> > > (as P1 is trying to communicate with P2 and P3), resulting in an >> > > exception being raised on P1, suspending the communicating task (our >> > > applications, being safety related, are constrained to being a single >> > > task). >> > > >> > >> > <> >> > >> > One problem that I've been told exists with most powerPC >> > CPUs is that due to internal pipelining they cannot recover the exact >> > machine state when a bus error occurs. Perhaps your boards do >> > not suffer from this problem, but I think that you should check with >> > the manufacturer to see whether recovery from a bus error can >> > be done. >> > >> > Speaking only for myself, >> > >> > Joe Durusau >> > >> > >> > --------------------------- Newsgroups: comp.os.vxworks Subject: Ramix PMC661 with Force 680_CPCI: vxWorks 5.4 PCI configuration Date: 29 May 2002 14:54:53 -0700 From: ckb@jade-consult.com (Chris K. Brown) Organization: http://groups.google.com/ Message-ID: The documentation is pretty poor, but I've been able to probe the PCI bus and find this card. I know its slot, bus, and function numbers. Normally, I would go ahead and continue to program the device, setting the base CSR values and such. Somehow, though, I get the feeling the supplied driver (object only, no source) will do this for me if I can figure out the correct initialization string for it. I haven't had any luck, MuxDevLoad either returns a "can't find device" message or a "need a MMU mapping for the CSR address". Has anyone worked with this driver and figured out what Ramix wants you to do? I am assuming that my BSP is not doing any probing/autoconfiguration of PCI devices, so I can't use the "-1" for the first initstring0 arg. The docs say this will casue it to read the already configured CSR address from the header. If I need to program it by hand, like say an intel 82559, does it take the same commands and data in the same form? Seems a stretch, no? Or is it just standard? I find it hard to believe I need to do this, since the docs do not mention it at all and they had a chance to supply skeleton code with the software kit. But the skeleton code just suplies routines to find the card and read and write words, not scripted values to program. Hell, I had to find the Vendor and Device ID's from the PowerBoot PCI probe. Sorry to vent. :-) Thanks for any help. --------------------------- Newsgroups: comp.os.vxworks Subject: cs8900a Ethernet controller question Date: Wed, 29 May 2002 20:38:49 -0700 From: "MK" Organization: mike_texx@mail.ru Message-ID: Reply-To: "MK" I'm using a CirrusLogic cs8900a ethernet controller. What is the easiest (and the most correct) way of handling the event of an ethernet cable being unplugged, and then plugged back in? I'm sure there should be strategies of seamlessly restarting and reinitializing the chip itself and the parts of the stack that are affected. --------------------------- Newsgroups: comp.os.vxworks Subject: How to prevent writes to PowerPC port pins from clobbering other port pins? Date: Thu, 30 May 2002 03:40:19 GMT From: sbuckser@attbi.com (Stephen Buckser) Organization: AT&T Broadband Message-ID: <3cf59f37.3744707@netnews.attbi.com> We are using the Motorola PowerPC MPC823E processor. We are using various ports for digital i/o. Some of the same ports are used by VxWorks drivers (PortA: Eth_RX_D, Eth_TX_D, Eth_RX_CLK, Eth_TX_CLK, PortC: Eth_RX_EN). What mechanism should we use to make sure our code that operates on the port pins do not clobber the pins controlled by the VxWorks drivers? And vica versa, how do we ensure that the VxWorks driver operations on the port pins do not affect the port pins operated by us. For ports not used by VxWorks drivers, we use semaphores to ensure we don't have two tasks trying to modify a port at the same time. sbuckser@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: TFFS setup failure Date: 29 May 2002 20:42:57 -0700 From: robin_ch@china.com (Robin Tcheng) Organization: http://groups.google.com/ Message-ID: I have a SST 39vf160 flash: size----------2M bytes/chip nmber of chip-------1 port size-----8 i want to set up TFFS on it when using MTDs,the current configuration is not working(the address of erase does not change at all). Here is my settings: flSetWindowBusWidth(vol.socket,16);/* use 16-bits */ flSetWindowSpeed(vol.socket,120); /* 120 nsec. */ flSetWindowSize(vol.socket,2); /* 8 KBytes */ vol.mtdVars = &mtdVars[flSocketNoOf(vol.socket)]; thisVars->unlockAddr1 = NO_UNLOCK_ADDR; vol.interleaving = 1; vol.chipSize = 0x200000l; flSetWindowSize(vol.socket,8 * vol.interleaving); thisVars->unlockAddr1 = 0x5555u * vol.interleaving; thisVars->unlockAddr2 = 0xaaaau * vol.interleaving; thisVars->baseMask = 0xfff80000L * vol.interleaving; vol.noOfChips = 1; vol.erasableBlockSize = 0x10000l * vol.interleaving; vol.flags |= SUSPEND_FOR_WRITE; /* Register our flash handlers */ vol.write = myMTDWrite; vol.erase = myMTDErase; please tell me the reason of failure or modify the settings. thanks!! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how is multi-level interrupt supported by VxWorks Date: Thu, 30 May 2002 04:20:23 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: As Joe wrote, it allows a higher-priority interrupt to interrupt the ISR of a lower-priority interrupt. The advantages and consequences depend on the system, the application, and so on. VxWorks supports multi-level interrupts, depending on the target architecture. "anshuman bhatt" wrote in message news:bdfba8e0.0205280433.5028bfd8@posting.google.com... > Hi > I am new to Embedded System and am learning VxWorks. > I have question: > What is multi-level interrupt,how VxWorks support it, what is it use? > Does any other operating system support it. > Regards > Anshuman Bhatt --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to prevent writes to PowerPC port pins from clobbering other port pins? Date: Thu, 30 May 2002 04:25:53 GMT From: "Gary M" Organization: AT&T Broadband Message-ID: References: <3cf59f37.3744707@netnews.attbi.com> Our experience is that VxWorks drivers correctly mask and change only the bits in the Port registers that apply to the devices they control. It's easy enough to check the source code for the drivers. You, likewise, should properly mask and use only those bits that your device controls. You will probably want to intLock() before you do so to guarantee exclusion from the VxWorks drivers. "Stephen Buckser" wrote in message news:3cf59f37.3744707@netnews.attbi.com... > We are using the Motorola PowerPC MPC823E processor. We are using > various ports for digital i/o. Some of the same ports are used by > VxWorks drivers (PortA: Eth_RX_D, Eth_TX_D, Eth_RX_CLK, Eth_TX_CLK, > PortC: Eth_RX_EN). > > What mechanism should we use to make sure our code that operates on > the port pins do not clobber the pins controlled by the VxWorks > drivers? And vica versa, how do we ensure that the VxWorks driver > operations on the port pins do not affect the port pins operated by > us. > > For ports not used by VxWorks drivers, we use semaphores to ensure we > don't have two tasks trying to modify a port at the same time. > sbuckser@yahoo.com --------------------------- Newsgroups: comp.os.vxworks Subject: stream operations on pipe Date: 29 May 2002 23:48:29 -0700 From: txue@singulix.com (Thomas) Organization: http://groups.google.com/ Message-ID: Hi I modified the sample code from WRS on inter-task communication using pipe and replaced with stream operations on pipe, such as fdopen, fread, fwrite. But when I tested it, I got the following errors. I also attach the code. Does anyone know why? Thanks. - -Tom ++++++++++++++++++++++++++++++++++++++++ - -> ld serverStart value = 0 = 0x0 - -> serverSend("Hello World. \n") status=0, strlen=14:str=Hello World. status=0, error writevalue = -1 = 0xffffffff = pipeF + 0xf80036b7 +++++++++++++++++++++++++++++ /* DESCRIPTION * * Demonstrates intertask communication using pipes. * * serverStart initializes a server task to execute functions * at a low priority and uses pipes as the communication * mechanism. * * serverSend sends a request to the server to execute a * function at the server's priority using pipe device. * Usage : serverSend ( ) * * EXAMPLE * To run this demo from the Wind shell do as follows: * * -> serverStart * value = 0 = 0x0 * -> serverSend ("Hello World. \n") * value = 0 = 0x0 * * NOTE: A host function such as i() cannot be used as the function * argument to serveSend. * eg: -> serverSend(i,"tShell") will not work * However if you wish to do that then make sure to add * INCLUDE_CONFIGURATION_5_2 and INCLUDE_DEBUG in configAll.h */ /* includes */ #include "vxWorks.h" #include "taskLib.h" #include "stdio.h" #include "ioLib.h" #include "pipeDrv.h" #include "string.h" #include "stdlib.h" #define TASK_PRI 254 /* tServers task's priority */ #define TASK_STACK_SIZE 5000 /* tServer task's stack size */ #define PIPE_NAME "/pipe/server" /* name of the pipe device */ #define NUM_MSGS 10 /* max number of messages in the pipe */ LOCAL int pipeFd; /* File descriptor for the pipe device */ FILE *pipeF; LOCAL void pipeServer (); /* server task */ /*********************************************************************** * serverStart -- Initializes a server task to execute functons * at a low priority. Uses pipes as the communication * mechanism. * * RETURNS: OK or ERROR on failure. */ STATUS serverStart () { if (pipeDevCreate (PIPE_NAME, NUM_MSGS, 300) == ERROR) { perror ("Error in creating pipe"); /* print error if pipe is already * created, but do not return */ } /* Open the pipe */ if ((pipeFd = open (PIPE_NAME, UPDATE, 0)) == ERROR) { perror ("Error in opening pipe device"); /*return (ERROR);*/ } if ((pipeF = fdopen (pipeFd, "r")) == NULL) { perror ("Error in opening pipe file descriptor"); return (ERROR); } /* Spawn the server task */ if (taskSpawn ("tServer", TASK_PRI, 0, TASK_STACK_SIZE, (FUNCPTR) pipeServer, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0) == ERROR) { perror ("Error in spawning tServer task"); close (pipeFd); return (ERROR); } return (OK); } /************************************************************** * serverSend -- Sends a request to the server to execute a * function at the server's priority. * * RETURNS: OK or ERROR on failure. */ STATUS serverSend( char* arg /* argument of the routine */ ) { char msgRequest[200]; int status; /* Initialize the message structure */ /* msgRequest.routine = routine;*/ strcpy(msgRequest, arg); /* Send the message and return the results */ status = fwrite (msgRequest, sizeof (char), 200, pipeF); fprintf(stderr, "status=%d, strlen=%d:str=%s\n", status, strlen(msgRequest),msgRequest ); if (status == 200) fprintf(stderr, "send %ld bytes: %s", sizeof(msgRequest), (char*)msgRequest); else fprintf(stderr, "status=%d, error write", status); return ((status == sizeof (msgRequest)) ? OK : ERROR); } /************************************************************************* * pipeServer -- Server task which reads from a pipe and executes * the function passed in the MSG_REQUEST data structure. * */ LOCAL void pipeServer () { char msgRequest[200]; while (fread ( msgRequest,sizeof (char) ,200, pipeF) > 0) { printf((char*)msgRequest); } } --------------------------- Newsgroups: comp.os.vxworks Subject: timeout on WdbRPC with Cross Wind debugger Date: 30 May 2002 00:28:40 -0700 From: hsingyuan@yahoo.com (Hsing Yuan) Organization: http://groups.google.com/ Message-ID: <70d78653.0205292328.53ee80a4@posting.google.com> If we set the cross wind debugger in system mode and stop the target system, WdbRPC will timeout after certain amount of time (n minutes) when the cross wind debugger receives no input (it is just idling.) After the timeout, the cross wind debugger is not connected the target, and the target does not run (since it was stopped.) We need to reboot the target and restart the cross wind debugger all over again. This is very annoying since sometimes we will be interrupted during a debug session (meetings, phone calls, or just need to read the manuals,) and cannot continue the debugger when we come back. Is there a way to configure the WDB component, Tornado, or target server so that they will not timeout on a debug session ? Thanks Hsing Yuan --------------------------- Newsgroups: comp.os.vxworks Subject: any one can provide me a source code about bootloader? Date: 30 May 2002 02:43:33 -0700 From: zq_jiang@powermatic.com.sg (Eric Jiang) Organization: http://groups.google.com/ Message-ID: <961cf38d.0205300143.6397dcf8@posting.google.com> the bootloader can update itself by network,update the firmware by network using ftpxfer().And it can load the vxwork.img into RAM. The environment is a ARM7 with 2M flash,vxworks os. And I don't know how to do it. --------------------------- End of New-News digest ********************** From vxwexplo-errs@csg.lbl.gov Thu May 30 06:21:23 2002 From: c tg Date: Thu May 30 06:21:25 PDT 2002 Subject: using vxWorks_rom Hi, I'm not sure will this mail be directed to some help. I got this problem whereby I've created this image, vxWorks_rom and downloaded it to the target. The image is about 4.5meg, but its taking about 20secs to copy itself from ROM to RAM. Is there a way to shorten the time? Please advise. Thanks in advance. TzeGhuan __________________________________________________ Do You Yahoo!? Yahoo! - Official partner of 2002 FIFA World Cup http://fifaworldcup.yahoo.com From vxwexplo-errs@csg.lbl.gov Thu May 30 18:31:50 2002 From: "zhaoshen" Date: Thu May 30 18:31:52 PDT 2002 Subject: Help! Hi all, I am a new user to use the Vxworks,and I met some concept problems. Hope someone can help me: 1.When doing BSP,in config.h the parameter of DEFAULT_BOOT_LINE pullzed me,in "cpm" postion, I saw soem other value,just like "fd","dc","elpci",etc.What's the meaning of those value? 2.the default image download way is by network,how and where to change it to the serial line way? 3.where can I find some GREAT documents to start my vxworks? thanks beforehand. ares From vxwexplo-errs@csg.lbl.gov Thu May 30 18:45:44 2002 From: Kevin Tsubota Date: Thu May 30 18:45:46 PDT 2002 Subject: shared memory setup error Hi, I'm trying to configure a fixed block of my MV2304's internal memory, running VxWorks 5.3.1, as a shared memory partition for access by other processors in the same VME crate. I started by defining INCLUDE_SM_OBJ in my configAll.c and setting SM_MEM_ADRS AND SM_OBJ_MEM_ADRS to NONE. Then I went to build it but got undefined smObj* references during the ld process as seen below. Can any tell me if I should have defined something else? Am I even on the right track? Thanks in advance. Kevin Tsubota ***** build output **** kahala{vxworks}206: gmake r/local/vw/m5.3.1.ppc.32m/target/src/config -I/usr/local/vw/m5.3.1.ppc.32m/target/src/drv -DCPU=PPC604 -DMV2300 -DTARGET_DIR="\"mv2304\"" /usr/local/vw/m5.3.1.ppc.32m/target/config/all/usrConfig.c -o usrConfig.o r/local/vw/m5.3.1.ppc.32m/target/src/config -I/usr/local/vw/m5.3.1.ppc.32m/target/src/drv -DCPU=PPC604 -DMV2300 -DTARGET_DIR="\"mv2304\"" /usr/local/vw/m5.3.1.ppc.32m/target/config/all/dataSegPad.c -o dataSegPad.o w/m5.3.1.ppc.32m/target/src/config -I/usr/local/vw/m5.3.1.ppc.32m/target/src/drv -DCPU=PPC604 -DMV2300 -DTARGET_DIR="\"mv2304\"" -P -x assembler-with-cpp -c -o sysALib.o sysALib.s ocal/vw/m5.3.1.ppc.32m/target/src/config -I/usr/local/vw/m5.3.1.ppc.32m/target/src/drv -DCPU=PPC604 -DMV2300 -DTARGET_DIR="\"mv2304\"" -c sysLib.c sysLib.c: In function `sysLocalToBusAdrs': sysLib.c:1149: warning: unsigned int format, long unsigned int arg (arg 2) sysLib.c:1149: warning: unsigned int format, long unsigned int arg (arg 3) r/local/vw/m5.3.1.ppc.32m/target/src/config -I/usr/local/vw/m5.3.1.ppc.32m/target/src/drv -DCPU=PPC604 -DMV2300 -DTARGET_DIR="\"mv2304\"" -o version.o /usr/local/vw/m5.3.1.ppc.32m/target/config/all/version.c ldppc -N -r \ -o vxWorks.tmp sysALib.o sysLib.o usrConfig.o version.o /usr/local/vw/m5.3.1.ppc.32m/target/lib/libPPC604gnuvx.a nmppc vxWorks.tmp | munch >ctdt.c gmake CC_COMPILER="-traditional" ctdt.o gmake[1]: Entering directory `/home/vw/m5.3.1.ppc.32m/target/config/mv2304' l/vw/m5.3.1.ppc.32m/target/src/config -I/usr/local/vw/m5.3.1.ppc.32m/target/src/drv -DCPU=PPC604 -DMV2300 -DTARGET_DIR="\"mv2304\"" -c ctdt.c gmake[1]: Leaving directory `/home/vw/m5.3.1.ppc.32m/target/config/mv2304' ldppc -N -e _sysInit -Ttext 00100000 \ -o vxWorks dataSegPad.o vxWorks.tmp ctdt.o vxWorks.tmp: In function `usrSmObjInit': vxWorks.tmp(.text+0x6b5c): undefined reference to `smObjLibInit' vxWorks.tmp(.text+0x6d48): undefined reference to `smObjSetup' vxWorks.tmp(.text+0x6d9e): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x6da2): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x6dc0): undefined reference to `smObjInit' vxWorks.tmp(.text+0x6dda): undefined reference to `smObjDesc' vxWorks.tmp(.text+0x6ddc): undefined reference to `smObjAttach' gmake: *** [vxWorks] Error 1 kahala{vxworks}207: From vxwexplo-errs@csg.lbl.gov Fri May 31 04:03:16 2002 From: Vxworks Exploder Date: Fri May 31 04:03:18 PDT 2002 Subject: comp.os.vxworks newsdigest Comp.Os.Vxworks Daily Digest Fri May 31 04:03:12 PDT 2002 Subject: Re: how is multi-level interrupt supported by VxWorks Subject: Looking for HDLC NRM - Very urgent !!! Subject: disable IP forwarding per interface. Subject: Re: How to prevent writes to PowerPC port pins from clobbering other port pins? Subject: I want to get a source code about x86 PC target serial port to communicate. Subject: Number of SYNC packets... Subject: non-blocking pipes on 5.4? Subject: PPC750 cache locking? Subject: Is there a solution to template bloat in VxWorks 5.4? Subject: Re: disable IP forwarding per interface. Subject: visionIce: DriverX vs Win2000 Subject: Re: how is multi-level interrupt supported by VxWorks Subject: 82559 END driver Subject: Re: PPC750 cache locking? Subject: Re: Galileo 64130 and VisionICE2 / VisionClick Subject: AMC CodeTEST for VxWorks Subject: Frequent interrupt hangs VxWorks 5.4 on PowerPC Subject: How can I assign two ip address to one ethernet port? Subject: unplugged Ethernet cable recovery Subject: Re: How can I assign two ip address to one ethernet port? Subject: Re: VME bus errors and signal handlers Subject: Re: visionIce: DriverX vs Win2000 Subject: Re: Question about network card! Subject: Re: Question about network card! Subject: WindView in romresident image Subject: Re: I want to get a source code about x86 PC target serial port to communicate. Subject: Re: 82559 END driver Subject: Re: Is there a solution to template bloat in VxWorks 5.4? ------------------------------------------------------- Newsgroups: comp.os.vxworks Subject: Re: how is multi-level interrupt supported by VxWorks Date: 30 May 2002 04:27:28 -0700 From: demralarood@hotmail.com (Demrala Rood) Organization: http://groups.google.com/ Message-ID: <3baeb99.0205300327.3f40bc1e@posting.google.com> References: "Gary M" wrote in message news:... > As Joe wrote, it allows a higher-priority interrupt to interrupt the ISR of > a lower-priority interrupt. The advantages and consequences depend on the > system, the application, and so on. > > VxWorks supports multi-level interrupts, depending on the target > architecture. Can you give example of architecture which VxWorks does not support it on? What is the mechanism of this support? Is there a system call to make? One when enter interrupt, another when leaving? How is it done? TIA DR --------------------------- Newsgroups: eunet.sources,comp.realtime,comp.os.vxworks Subject: Looking for HDLC NRM - Very urgent !!! Date: 30 May 2002 06:23:10 -0700 From: christophe.dutertre@fr.thalesgroup.com (Christophe) Organization: http://groups.google.com/ Message-ID: We are developping an equipment which needs to communicate via HDLC protocol in NRM mode. I need to implement this protocol on top of the hdlc controller of a MC68302 and on Windows NT platform. Does anybody know where I can find such source code. Thanks a lot for your help - ----------------------------------------------------- Christophe DUTERTRE E-mail: christophe.dutertre@fr.thalesgroup.com - ----------------------------------------------------- --------------------------- Newsgroups: comp.os.vxworks Subject: disable IP forwarding per interface. Date: 30 May 2002 06:58:30 -0700 From: vrrajesh@intotoinc.com (rajesh vr) Organization: http://groups.google.com/ Message-ID: <68c7b7f1.0205300558.612a9328@posting.google.com> hi, how can i disable ip forwarding , based on interface. (assume ,I have many interface, I want to disable for a particular interface and allow other interface to forward) I understand that IP_FLAGS_DFLT flag is common ,By commenting the IP_DO_FORWARDING ,no interface will able to forward. thanks in advance rajesh vr --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How to prevent writes to PowerPC port pins from clobbering other port pins? Date: 30 May 2002 07:08:04 -0700 From: demralarood@hotmail.com (Demrala Rood) Organization: http://groups.google.com/ Message-ID: <3baeb99.0205300608.6d0985c8@posting.google.com> References: <3cf59f37.3744707@netnews.attbi.com> sbuckser@attbi.com (Stephen Buckser) wrote in message news:<3cf59f37.3744707@netnews.attbi.com>... > We are using the Motorola PowerPC MPC823E processor. We are using > various ports for digital i/o. Some of the same ports are used by > VxWorks drivers (PortA: Eth_RX_D, Eth_TX_D, Eth_RX_CLK, Eth_TX_CLK, > PortC: Eth_RX_EN). > > What mechanism should we use to make sure our code that operates on > the port pins do not clobber the pins controlled by the VxWorks > drivers? And vica versa, how do we ensure that the VxWorks driver > operations on the port pins do not affect the port pins operated by > us. > > For ports not used by VxWorks drivers, we use semaphores to ensure we > don't have two tasks trying to modify a port at the same time. > sbuckser@yahoo.com I think this is why Wind expose all port bit manipulation for there drivers. You just have to find all these manipulations. Unfortunate, they scatter it all over the files so you have too look harder. pSOS used to have port bit allocation scheme to help this but it was ugly. Now Wind erases pSOS anyway. DR --------------------------- Newsgroups: comp.os.vxworks Subject: I want to get a source code about x86 PC target serial port to communicate. Date: 30 May 2002 07:45:17 -0700 From: djkzhw@163.com (David) Organization: http://groups.google.com/ Message-ID: hi! I want to get a source code about x86 PC target serial port to get byte by interrupt.But i'm a starter,I don't know how to do.Can you help me and send me a example for my studing? Thank you very much! --------------------------- Newsgroups: comp.os.vxworks Subject: Number of SYNC packets... Date: 30 May 2002 13:19:14 -0400 From: Bill Pringlemeir Organization: Factory of the mind Message-ID: Sender: bpringlemeir@DeadDuck Are there socket options to allow more SYNC packets to be sent in a connect phase than just two? Regards, Bill Pringlemeir. - -- [Hello to all my friends and fans in domestic surveillance] bullion SEAL Team 6 Capricorn Lexis-Nexis UNSCOM cryptanalysis 64 Vauxhall Cross e-bomb Croatian Saddam Hussein airframe Telex Watergate encryption vxWorks FAQ, "http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html" --------------------------- Newsgroups: comp.os.vxworks Subject: non-blocking pipes on 5.4? Date: Thu, 30 May 2002 12:40:45 -0500 From: "Peter W. Colovas" Organization: JR1M00000 Message-ID: <3CF6641D.BF8CD148@lucent.com> This is a multi-part message in MIME format. - --------------8F68AFB44D84F11AD927E0CD Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Can I set up a pipe to be non-blocking on reads and writes? I can't find an answer for this in the documentation or the FAQ. I think I can do this using select(), with a wait time of 0. is this correct? Thanks, - -Pete - --------------8F68AFB44D84F11AD927E0CD Content-Type: text/x-vcard; charset=us-ascii; name="pcolovas.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for Peter W. Colovas Content-Disposition: attachment; filename="pcolovas.vcf" begin:vcard n:Colovas;Peter tel;work:+1 630 224 2181 x-mozilla-html:FALSE org:SPEED Platform Development;SPEED Hardware and Software Development version:2.1 email;internet:pcolovas@lucent.com title:Member of Technical Staff adr;quoted-printable:;;2000 N. Naperville Rd.=0D=0AP.O. Box 3033;Naperville;Illinois;60566; x-mozilla-cpt:;-22928 fn:Peter W. Colovas end:vcard - --------------8F68AFB44D84F11AD927E0CD-- --------------------------- Newsgroups: comp.os.vxworks Subject: PPC750 cache locking? Date: 30 May 2002 11:14:48 -0700 From: jchen@tollbridgetech.com (john c) Organization: http://groups.google.com/ Message-ID: <1f1b48b5.0205301014.bce022@posting.google.com> Hi all. Quick question about the PPC750. Does anyone know how cacheLock() function works? I've been reading through the 750CXe Manual and there are no instructions that you can use to lock a part of the cache. The 750 provides certain HID[] bits for you to lock or unlock the entire L1 cache (Data and/or instruction cache) but strangely enough, the BSP for PPC750 supports cacheLock() function. Does anyone know how vxWorks does cacheLock()? How do you setup the hardware to lock certain portions of the cache? Or is this purely a software thing? john --------------------------- Newsgroups: comp.os.vxworks Subject: Is there a solution to template bloat in VxWorks 5.4? Date: 30 May 2002 11:56:41 -0700 From: braddillman@yahoo.com (Brad Dillman) Organization: http://groups.google.com/ Message-ID: <1f0502b5.0205301056.e0422c5@posting.google.com> I've been searching the web, to find a way to prevent STL templates from appearing in multiple .o files. I'm currently trying the -frepo option, which requires me to explicitly instantiate templates. Will this work? Has this problem been solved before? Any pointers, URLs or advice? Thanks, Brad Dillman --------------------------- Newsgroups: comp.os.vxworks Subject: Re: disable IP forwarding per interface. Date: 30 May 2002 12:08:05 -0700 From: ckb@jade-consult.com (Chris K. Brown) Organization: http://groups.google.com/ Message-ID: References: <68c7b7f1.0205300558.612a9328@posting.google.com> Raj, Looking directly at the source (ip_input.c, vxWorks 5.4, unpatched, file last modified 28feb99), it is clear you cannot forward on a per-interface basis by playing with that IP_DO_FORWARDING flag. This is part of a global set of flags (_ipCfgFlags) that applies to all interfaces. I'm trying to thik of some tricks you could play by setting up some code at a function hook point, but the code that uses the variables above will have discarded the datagram already. If I was desperate, I would get a copy of ip_forward and rewrite it to check the interface flags (use an unused inteface flag, there must be one) and then replace the vxWorks ip_forward with your own. So you would have IP_DO_FORWARDING on, and the forward routine itself would discriminate. You can just moduleLoad your own ip_forward code and it should pick up your new routine without any trouble. vrrajesh@intotoinc.com (rajesh vr) wrote in message news:<68c7b7f1.0205300558.612a9328@posting.google.com>... > hi, > how can i disable ip forwarding , based on interface. > (assume ,I have many interface, I want to disable for a particular > interface and allow other interface to forward) > > I understand that IP_FLAGS_DFLT flag is common ,By commenting the > IP_DO_FORWARDING ,no interface will able to forward. > > thanks in advance > rajesh vr --------------------------- Newsgroups: comp.os.vxworks Subject: visionIce: DriverX vs Win2000 Date: Thu, 30 May 2002 12:40:03 -0700 From: "kdb" Organization: cmc Message-ID: I started noticing problems on my Windows 2000 laptop around the time I installed the visionClick software (v7.90A) for my visionIce emulator. Finally tracked it down to the DriverX driver installed by visionClick. Disabled the driver, Win2000 problems went away. The visionClick GUI still seems to work as well as it ever did. Which is to say, not too well if you want to communicate with the visionIce via Ethernet. Communications keep dropping out, I have to keep resetting the unit and reopening the NET port. I can still ping the unit but the GUI loses its connection to it. Don't have this problem if I use the COM port, but I need the COM for something else and it's inadequate for downloads anyway. I've seen indication somewhere that driverx.sys is not the correct version of the DriverX driver for Win2000. Is the correct DriverX version for Win2000 available and if so how do I get it installed and properly configured? Also, will this fix the problem with network communications between the GUI and the visionIce unit? thx - K --------------------------- Newsgroups: comp.os.vxworks Subject: Re: how is multi-level interrupt supported by VxWorks Date: Thu, 30 May 2002 13:52:54 -0700 From: Madhusudan Challa Organization: Ericsson Inc Message-ID: <3CF69126.F35244CB@ericsson.com> References: Multi level interrupts is nesting interrupt handling because a higher level interrput preempts the execution of a low level isr. vxworks supports multi level interrupts just like most other operating systems support. example solaris, most flavours of unix i know of. of course there are some real time operating systems which might not support nested interrupts due to its complexity involved. Implementing multi level interrputs is specific to your target architecture. Most of the architectures i know of provide for disabling all interrupts when an exception or an interrupt occurs. so support for non nested interrupts comes automatically. But when you want to allow interrupts higher than your current level, you need the architecture to support some sort of mask register which can tell what level you are currently in and which levels are allowed. So in the isr you would basically save registers (what to save is architecture specific) and transfer control to the interrupt handler. note that each handler would use a different stack. While returning from the isr you should ensure that you restore the saved registers. Note that some architectures like MIPS support k0,k1 registers to be used only in an exception handler , you need to save and restore these too for this architecture because you might be preempting an isr. Further in MIPS like architecture while returning from an interrput you use an instruction like eret which actually takes processor out of exception and also transfers control to the user program which would find its registers corrupted because this might be a nested isr. So you need to rather return to the next handler which would do the restore and finally transfer control to the user. This process is called chaining of interrupt handlers. Lots of these details are specific to the target architecure just like you would be knowing that the interrupt handling itself is an architecture specific issue. For more details refer to this link below which beautifully explains the chaining for mips like architectures. http://www.tasking.com/support/MIPS/a2002_03.pdf hope this helps. - -madhusudan anshuman bhatt wrote: > Hi > I am new to Embedded System and am learning VxWorks. > I have question: > What is multi-level interrupt,how VxWorks support it, what is it use? > Does any other operating system support it. > Regards > Anshuman Bhatt - -- - -madhusudan Ericsson Inc. Data Backbone and Optical Networks 70 Castilian Drive Santa Barbara, CA 93117 work :- 805-562-6066 --------------------------- Newsgroups: comp.os.vxworks Subject: 82559 END driver Date: 30 May 2002 14:29:39 -0700 From: ccadena@vmetro.com (carlos) Organization: http://groups.google.com/ Message-ID: Does anyone know where I can find a 82559 END driver? I'm using an 8240 PowerPC. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: PPC750 cache locking? Date: 30 May 2002 15:28:05 -0700 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: References: <1f1b48b5.0205301014.bce022@posting.google.com> Hi, I am not quite sure about the implementation of cacheLock(), but the MMU of PPC or the PTE descriptor does allow a page to be lock. You can look at this in the sysLib.c and the PPC programing manual. We did successfully used cacheLock(). Gl, Vinh Lam jchen@tollbridgetech.com (john c) wrote in message news:<1f1b48b5.0205301014.bce022@posting.google.com>... > Hi all. > Quick question about the PPC750. Does anyone know how cacheLock() > function works? I've been reading through the 750CXe Manual and there > are no instructions that you can use to lock a part of the cache. The > 750 provides certain HID[] bits for you to lock or unlock the entire > L1 cache (Data and/or instruction cache) but strangely enough, the BSP > for PPC750 supports cacheLock() function. > > Does anyone know how vxWorks does cacheLock()? How do you setup the > hardware to lock certain portions of the cache? Or is this purely a > software thing? > > john --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Galileo 64130 and VisionICE2 / VisionClick Date: 30 May 2002 15:32:40 -0700 From: vloscomp@yahoo.com (Vinh) Organization: http://groups.google.com/ Message-ID: References: <3CF4EB4A.4165E677@ll.mit.edu> I have a register file for GT64260. I can only give u the default one. Give me a ping, then i will send it to you. Vinh Lam Michael Reed wrote in message news:<3CF4EB4A.4165E677@ll.mit.edu>... > Hello, > Has anyone used VisionICE2, VisionICE1, or VisionProbe w/ a board > containing a Galileo 64130 memory controller / system controller? I'm > in this position, and I'm looking for an existing register file for this > combo (PPC750 and Galileo 64130). > > Thanks, > Mike R. --------------------------- Newsgroups: comp.os.vxworks Subject: AMC CodeTEST for VxWorks Date: 30 May 2002 15:40:39 -0700 From: gah205@nyu.edu (GAH) Organization: http://groups.google.com/ Message-ID: <7ce21ebb.0205301440.16baea77@posting.google.com> OK. I'm working on a rad750 target trying to get CodeTEST to run but am having some difficulty. After having read the manual I went back and recompiled using the ctcxx compiler. Next it says that I need to include CodeTEST in my config.h file (which I can do using the Project Facility). Once I do this, will I need to do a full rebuild of my kernel and load it back to the target? I'm trying to avoid this as memory allocation, NVM issues, etc, could make it pretty messy. Also I was not involved with the inital kernel upload so it might take some additional people (yet another headache). In any case, I'd be more than interested to hear anyone else's experience with CodeTEST and any pointers you can supply me with. Thanks. -GAH --------------------------- Newsgroups: comp.os.vxworks Subject: Frequent interrupt hangs VxWorks 5.4 on PowerPC Date: 30 May 2002 15:45:13 -0700 From: mwj999@hotmail.com (Mike Johnson) Organization: http://groups.google.com/ Message-ID: <82606370.0205301445.21abb88@posting.google.com> A coworker of mine ran a test in which an external interrupt was generated approximately every 10 microseconds, and hung his MPC7400 CPU board. Resetting and examining memory from the VxWorks boot prompt, we observed that a task's stack was corrupted, and so was our code. Obviously, this is not a practical application; an interrupt at this frequency leaves very little time for anything else. Nevertheless, the system should handle interrupts at any frequency without overflowing stacks or corrupting code. Has anyone else observed this problem? --------------------------- Newsgroups: comp.os.vxworks Subject: How can I assign two ip address to one ethernet port? Date: 30 May 2002 16:02:14 -0700 From: xiaohua.zhang@163.com (xiaohua) Organization: http://groups.google.com/ Message-ID: CPU:ARM7TDMI IDE:Tornado 2.0 for ARM Please give me some advise about it! Thanks! --------------------------- Newsgroups: comp.os.vxworks Subject: unplugged Ethernet cable recovery Date: Thu, 30 May 2002 16:39:44 -0700 From: "MK" Organization: mike_texx@mail.ru Message-ID: Reply-To: "MK" I'm using a CirrusLogic cs8900a ethernet controller on an ARM system (EP7312), vxWorks 5.4 What is the easiest (and the most correct) way of handling the event of an ethernet cable being unplugged, and then plugged back in? I'm sure there should be strategies of seamlessly restarting and reinitializing the chip itself and the parts of the stack that are affected Currently I'm attempting to debug the process of running muxDevUnload() followed by muxDevLoad(), muxDevStart(), the code for which I took from usrNetwork.c with insignificant changes. But even though after the process seemingly successfully completes, I see the driver by using muxShow, but I see no valid interface being up by using ifShow. Any suggestions? This interface works fine after being brought up on startup, and works with no problems until the ethernet cable is unplugged. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: How can I assign two ip address to one ethernet port? Date: Thu, 30 May 2002 16:40:24 -0700 From: "Parangodath, Binish [SC7:326:EXCH]" Organization: Nortel Networks Message-ID: References: Try ifAddrAdd() Regards, Binish "xiaohua" wrote in message news:dc50cd74.0205301502.4f0d1441@posting.google.com... > CPU:ARM7TDMI > IDE:Tornado 2.0 for ARM > > Please give me some advise about it! > Thanks! --------------------------- Newsgroups: comp.os.vxworks,comp.arch.bus.vmebus Subject: Re: VME bus errors and signal handlers Date: Thu, 30 May 2002 21:01:47 -0400 From: Andy F Organization: Bell Sympatico Message-ID: References: <3b4d40d1.0205230234.1eaad47b@posting.google.com> <3CED8891.78C4FAC8@bellsouth.net> <3cee8cdc$0$4544@dingus.crosslink.net> <3b4d40d1.0205281115.6a61ad19@posting.google.com> <3cf51c3a.1144183@news.saix.net> Hi Chris, Anton I'll add a few more details on some of the topics... The PPC's problem with recovering from a machine exception is due to it's internal parallelism. By the time the exception has been received, the PPC cannot determine which instructions that followed the one that errored should or should not have been completed. Depending what the instruction stream was, it may or may not be recoverable. So Motorola says in general it cannot be. Older processors had no internal parallelism, so no instructions upstream of the failed instruction would have been performed yet. The VOWN method of locking semaphore access would protect the access only against other VME cards - but not against the processor on the card where the memory is located! It could still freely access it's own memory, and possibly take the semaphore at the same time. Using the actual VME read-modify-write should solve this, it uses a special set of registers in the Universe and there should be a BSP function to perform this. Andy --------------------------- Newsgroups: comp.os.vxworks Subject: Re: visionIce: DriverX vs Win2000 Date: 31 May 2002 07:10:16 GMT From: Johan Borkhuis Organization: Agere Systems Message-ID: References: "kdb" wrote: > I started noticing problems on my Windows 2000 laptop around the time > I installed the visionClick software (v7.90A) for my visionIce > emulator. Finally tracked it down to the DriverX driver installed by > visionClick. Disabled the driver, Win2000 problems went away. I had some discussions with WindRiver on this, and this is part of the answer: The DriverX is used for the paralell port, when you connect to the visionPROBE. Since you are using a visionICE through the network (correct me if I am wrong), as a workaround it should work to just stop the driverx. I did also see the problem with the connection dropped, but usually it is restored by reopeneing the NET port. We do not see the problems on other platforms. Groeten, Johan - -- o o o o o o o . . . _____________________________ o _____ || Johan Borkhuis | .][__n_n_|DD[ ====_____ | borkhuis@agere.com | >(________|__|_[_________]_|__________________________| _/oo OOOOO oo` ooo ooo 'o!o!o o!o!o` === VxWorks FAQ: http://www.xs4all.nl/~borkhuis/vxworks/vxworks.html === --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: 31 May 2002 00:36:42 -0700 From: buptspring@hotmail.com (Zhang Chun) Organization: http://groups.google.com/ Message-ID: <366b1c3b.0205302336.2add4e88@posting.google.com> References: <989cdae4.0205172212.5bd0a755@posting.google.com> <3baeb99.0205180454.3a9bd2ba@posting.google.com> hi,all, thank you for everyone's answer. Our network card is 3c905b,I think that we can not handle this interrupt:rxComplete.So when the packet received form MII to FIFO of NIC,it will not have an interrupt. When we send packets with sniffer to NIC from host to target,NIC have the interrupt:update statistcs.It said NIC have got packets. But we can not get the interrupt:upComplete.So we can not sure whether the packet have uploaded to system's buffer from NIC.So we can not sure whether NIC can receive packets correctly. If someone know our problem,please tell us.And give us some advice. Thank you in advance! Best Regards! Spring --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Question about network card! Date: 31 May 2002 00:38:52 -0700 From: buptspring@hotmail.com (Zhang Chun) Organization: http://groups.google.com/ Message-ID: <366b1c3b.0205302338.527ce91d@posting.google.com> References: <989cdae4.0205172212.5bd0a755@posting.google.com> <3baeb99.0205180454.3a9bd2ba@posting.google.com> hi,all, thank you for everyone's answer. Our network card is 3c905b,I think that we can not handle this interrupt:rxComplete.So when the packet received form MII to FIFO of NIC,it will not have an interrupt. When we send packets with sniffer to NIC from host to target,NIC have the interrupt:update statistcs.It said NIC have got packets. But we can not get the interrupt:upComplete.So we can not sure whether the packet have uploaded to system's buffer from NIC.So we can not sure whether NIC can receive packets correctly. If someone know our problem,please tell us.And give us some advice. Thank you in advance! Best Regards! Spring --------------------------- Newsgroups: comp.os.vxworks Subject: WindView in romresident image Date: Fri, 31 May 2002 11:10:43 +0200 From: "Espen Halsaa Albrektsen" Message-ID: We are attempting to use WindView on a 386 target whith vxworks running romresident. Target is running TFFS file system When we start WindView from Tornado, the target crashes totally. After pressing hardware reset, target does _NOT_ reboot Boot flash must be rewritten before target recovers. Any suggestions anyone? OS/Tools: Tornado 2.0 and VxWorks 5.4. Cumulative patch 3 is installed. - -- Espen Halsaa Albrektsen - Data Respons AS --------------------------- Newsgroups: comp.os.vxworks Subject: Re: I want to get a source code about x86 PC target serial port to communicate. Date: Fri, 31 May 2002 11:41:18 +0200 From: "Werner Schiendl" Message-ID: <3cf7453b$1@brateggebdc5.br-automation.co.at> References: Hi, you already have the sources for the serial port access delivered with your Tornado setup. See $(WIND_BASE)/target/src/drv/sio/i8250Sio.c hth Werner "David" wrote in message news:ca2daa45.0205300645.3d60fea6@posting.google.com... > hi! > I want to get a source code about x86 PC target serial port to get > byte by interrupt.But i'm a starter,I don't know how to do.Can you > help me and send me a example for my studing? > Thank you very much! --------------------------- Newsgroups: comp.os.vxworks Subject: Re: 82559 END driver Date: Fri, 31 May 2002 11:43:37 +0200 From: "Werner Schiendl" Message-ID: <3cf745c5$1@brateggebdc5.br-automation.co.at> References: Hi, the fei driver from WRS should do the trick. I don't know if it is available for PPC, but since its from WRS it should... You probably need to adjust device IDs in the hardware detection code. hth Werner "carlos" wrote in message news:fc419721.0205301329.4689914b@posting.google.com... > Does anyone know where I can find a 82559 END driver? I'm using an 8240 PowerPC. --------------------------- Newsgroups: comp.os.vxworks Subject: Re: Is there a solution to template bloat in VxWorks 5.4? Date: Fri, 31 May 2002 10:18:35 GMT From: bill@woola.qswtools.com Organization: EarthLink Inc. -- http://www.EarthLink.net Message-ID: <%5IJ8.397$bJ4.357@newsread1.prod.itd.earthlink.net> References: <1f0502b5.0205301056.e0422c5@posting.google.com> In article <1f0502b5.0205301056.e0422c5@posting.google.com>, braddillman@yahoo.com (Brad Dillman) writes: > I've been searching the web, to find a way to prevent STL templates > from appearing in multiple .o files. I'm currently trying the -frepo > option, which requires me to explicitly instantiate templates. > > Will this work? Yes, all WR versions of gcc that I've used support '-frepo' except the Tornado 1.0.1 version. Except that it's not called "explicit instantiation". > Has this problem been solved before? Yes. > Any pointers, URLs or advice? Sure. http://www-es.fernuni-hagen.de/cgi-bin/info2html?(gcc)Template%20Instantiation The newer versions of gcc from WRS (2.96) implement the "Borland model" in which the compiler automatically instantiates template code in the object file where the template is referenced. The duplicate code is then discarded by the linker (ldppc) or the loader. The only cost to this is larger object files, and slightly slower build times. > > Thanks, > Brad Dillman --------------------------- End of New-News digest **********************