Re: Kernel versioning (was Re: LCS drivers for 2.4.9 ?)
This is not merely a documentation string! Also, who says that a module built for 2.4.9 won't work with 2.4.9-4GB? What patches are these that warrant changing the label?? The good thing with the dash-level in the kernel IMHO is that you keep the old set of kernel modules, and you can backout your change simply by picking up your previous module. There's more than just the version of the kernel patches, it's also the various config options that you enabled or disabled. These options sometimes do change the layout of control blocks and require that you also recompile the kernel modules. If you are required to use the kernel modules that worked for someone else without recompiling with your own config settings, it may work or it may not work. Even if the modules would be available for a number of different config options (non-SMP, IPv6) and popular patches (e.g. the timer patch) then you're still very limited in what you can do. This is a practical problem with the way IBM is now doing the network drivers which imho can be fixed by version and config independent interfaces and source maintained wrappers around the binary-only code. The question whether you only want things for which you have source can be another more theoretical one, I think. Rob
Re: redhat s/390 education ?
Mark, Thanks for the response. I guess even a Redhat Admin course (on Intel) should still be of considerable value to me. Thanks, SG -Original Message- From: Post, Mark K [mailto:[EMAIL PROTECTED]] Sent: Thursday, December 27, 2001 5:09 PM To: [EMAIL PROTECTED] Subject: Re: redhat s/390 education ? Sandy, I don't know if they well be or not, but I don't think that would be really necessary. The S/390-specific parts are a very small percentage of the whole learning curve. If you took any distribution-specific education (and there's plenty of it out there for Intel-based Linux), you'd have covered the largest percentage of what you need to know for that particular distribution. If you took _any_ Linux education, that would be well over 50% of what you need, distribution specific or not. Heck, for the right amount of money, I'll teach what you need to know! :) I believe Sine Nomine Associates do that as well. Mark Post -Original Message- From: Gelbard, Sandy [mailto:[EMAIL PROTECTED]] Sent: Thursday, December 27, 2001 2:16 PM To: [EMAIL PROTECTED] Subject: redhat s/390 education ? Does anyone know if IBM or REDHAT will be offering redhat/390 centric education ? Thanks, SG
Re: Oracle Installation Problem
Hi.. Thanks.. Now, I can't find the download for gcc-2.95.3.tar.gz anywhere. Could you help pls.? Vinod. -Original Message- From: Alex Bykov [SMTP:[EMAIL PROTECTED]] Sent: Thursday, December 27, 2001 6:50 PM To: [EMAIL PROTECTED] Subject: Re: Oracle Installation Problem 27.12.2001 13:06:03 Linux on 390 Port wrote: Hi.. When I try to re-link the Oracle binaries, Iam getting the following error: /oracle/ora901/rdbms/lib/opimai.o: file not recognized: File format not recognized collect2: ld returned 1 exit status make: *** [/oracle/ora901/rdbms/lib/oracle] Error 1 Iam having Suse Linux V7.0. Any ideas on how to proceed? Hi.. We had the same problem. Solving this problem is to migrate to SuSE Linux Enterprise Server 7 (kernel 2.4.7 and etc).. WBR, Alex This e-mail and any files transmitted with it are for the sole use of the intended recipient(s) and may contain confidential and privileged information. If you are not the intended recipient, please contact the sender by reply e-mail and destroy all copies of the original message. Any unauthorised review, use, disclosure, dissemination, forwarding, printing or copying of this email or any action taken in reliance on this e-mail is strictly prohibited and may be unlawful. Visit us at http://www.cognizant.com
Re: LCS drivers for 2.4.9 ?
Let's say you ran a large, profitable company like IBM. Software revenues around $12 billion. You're risking a lot in supporting something like Linux. OTOH - IBM achieved its utter dominance of the industry in the late 1960s using a public domain operating system. Contrary to popular belief, it is not the silver bullet of all IT, and it is not the most stable operating system in existence (nor even close). I think the solution to this problem is perhaps not what most people expect. The RAS of an operating system is one thing, and there's no doubt Linux has a way to go to catch z/OS. The RAS of a hardware platform is another issue. Put the two together and what happens? In the case of Linux on zSeries, not a lot. But z/OS on zSeries is a different issue - the operating system has a lot of active support for hardware RAS built in - some of which would be difficult if not impossible to implement in Linux. The unique selling proposition of zSeries is in fact no such thing - it's the unique selling proposition of the _combination_ of zSeries hardware with an operating system that can support hardware RAS. I hate the word 'synergy', but it applies. Linux/390 gives ready access to the platform for Linux applications. However - there aren't actually all that many, and most of them are not very scalable. I see system reliabilty and security becoming more important than ever in 2002. We've had a run of very pervasive email viruses and I'm convinced we're in for a phase of destructive activity by those sympathetic to Bin Laden and his ilk. If IBM can only deal with the price issue, z/OS would be ideally placed as the provider of highly secure environments. I'm not anti-Linux or anti-Open Source - quite the reverse - but I see greater potential for developing the z/OS market than the Linux/390 market over the short and perhaps medium term. -- Phil Payne The Devil's IT Dictionary - last updated 2001/12/22: http://www.isham-research.com/dd.html
Re: LCS drivers for 2.4.9 ?
Rather, QDIO is a hardware feature, for which IBM does not publish the programming interface. The interface is unique to IBM S/390. And it is used by more OS's than just Linux. Publishing the interface for Linux also publishes the interface for all other OS. ISTR Amdahl paid a great deal of money for the specifications of seldom-ending channel programs and QDIO. Until such time as the OSA card interface is public domain, I guess we're stuck. Is there any significance in the announcement that G4/G5/G6 now a minimum of one standard OSA adapter? -- Phil Payne The Devil's IT Dictionary - last updated 2001/12/22: http://www.isham-research.com/dd.html
S390/zSeries CPU questions
Hello How do I parse the identification field in /proc/cpuinfo? The other ones I understand. The one I have access to says processor 0: version = FF, identification = 035667, machine = 9672 I am trying to estimate the performance of our application on different S390/zSeries systems, and for the CPU bound parts I am interested in not only the MIPS rating and MHZ rating but also things like: -- How good is the S390 backend of gcc compared to, say, IA32? -- Cache sizes and speeds (bw and lat) for the different levels of cache? (I could use lmbench, but I only have access to one CPU) -- Issue width, pipeline stages, and similar micro-arch information. (this is mostly for personal interest, I will probably not code any custom assembly for S390, and we have none for other platforms) Right now my Intel P3 733 is 10 times faster than the Linux guest :-( (But this will change :-)) RegardsErik
Re: S390/zSeries CPU questions
How do I parse the identification field in /proc/cpuinfo? The other ones I understand. The one I have access to says processor 0: version = FF, Probably this indicates you're running under VM or VIF. The Linux guest is seeing a virtual CPU (on bare metal, this would be the actual physical id of the CPUs, eg 00, 01, 02, etc) or a LPAR id. identification = 035667 Serial number of the machine. , machine = 9672 Model of the box. I am trying to estimate the performance of our application on different S390/zSeries systems, and for the CPU bound parts I am interested in not only the MIPS rating and MHZ rating but also things like: Good, because they're bogus in a virtual machine anyway. -- How good is the S390 backend of gcc compared to, say, IA32? It's pretty good, with Neale's tinkering, and the Boeblingen guys doing some fine work. It's not perfect (what is?), and there is some work to be done in the optimizer, but it's pretty good. The 64 bit pieces are not as mature, and will probably need some more study. -- Cache sizes and speeds (bw and lat) for the different levels of cache? (I could use lmbench, but I only have access to one CPU) IBM Journal of RD occasionally publishes some interesting articles in this vein.
Re: S390/zSeries CPU questions
Thank you David! One new question: Are there any performance counters such as other processors have (cache hit/miss, tlb hit/miss, branch predict misses, insns retired etc.) available from Linux? On Fri, 2001-12-28 at 20:08, David Boyes wrote: How do I parse the identification field in /proc/cpuinfo? The other ones I understand. The one I have access to says processor 0: version = FF, Probably this indicates you're running under VM or VIF. The Linux guest is seeing a virtual CPU (on bare metal, this would be the actual physical id of the CPUs, eg 00, 01, 02, etc) or a LPAR id. identification = 035667 Serial number of the machine. Uggh, I knew it felt somewhat familiar... , machine = 9672 Model of the box. OK, so *any* way to tell which type of CPU type (G3, G4, G5...) I'm running on? (the nice people letting me use this guest are on vacation, and I hate waiting for them to get back :-)) -- Cache sizes and speeds (bw and lat) for the different levels of cache? (I could use lmbench, but I only have access to one CPU) IBM Journal of RD occasionally publishes some interesting articles in this vein. Thanks, that paid off immediately, if someone wonders the URL is: http://www.research.ibm.com/journal/rd And they have many nice articles online.
Re: S390/zSeries CPU questions
IBM Journal of RD occasionally publishes some interesting articles in this vein. Thanks, that paid off immediately, if someone wonders the URL is: http://www.research.ibm.com/journal/rd And they have many nice articles online. Vol. 43, Nos. 5/6, 1999 are particularly interesting for owners of 9672's -- lots of grungy deep detail on the G5/G6 architecture. Lots of good stuff wrt to IEEE FP implementation and some interesting details on the MCM wafer manufacturing process. -- db
Re: Red Hat RC2 Install Problem on MP3000
Mark, I thought IBM's OCO modules were only for OSA card support, and were not needed for the MP3000, which uses plain OS/2 TCP/IP using standard PC Ethernet cards. Are you sure about that? I didn't need to add anything for SuSE 7.0, SLES1 and SLES7 to work. Is there a URL to get the IBM OCO modules? If I added the OCO modules, would my dialogue responses then be correct? Regards, Jim - Original Message - From: Post, Mark K To: [EMAIL PROTECTED] Sent: Friday, December 28, 2001 3:10 PM Subject: Re: Red Hat RC2 Install Problem on MP3000 Jim, This is due to the fact that Red Hat does not include IBM's OCO modules in their distribution. You'll have to download the appropriate level yourself (2.4.9), and follow the instructions Red Hat provides to include them in your initial ram disk for the install. You'll also have to make sure that the module(s) get copied to your target system so they'll be there for the real IPL. Since the Red Hat distribution has gone GA, I would recommend downloading the files for that which are different from the RC2 version, and using the GA code. There shouldn't be a huge difference, but why not be current? Mark Post -Original Message- From: Jim Rich [mailto:[EMAIL PROTECTED]] Sent: Friday, December 28, 2001 3:41 PM To: [EMAIL PROTECTED] Subject: Red Hat RC2 Install Problem on MP3000 I booted the Red Hat RC2 CD #1/4 using the SE CD-ROM Icon, but I can't seem to find the right dialogue responses. I'm trying to use the same setup I use successfully for SuSE on another MP3000 system, which is LCS3172 address pair e26-e27 and Ethernet card #2. Anyone have any suggestions? Here is my latest failed attempt: ... Please enter the network device you intend to use (e.g. ctc0, escon0, iucv0, eth0, tr0): eth0 Please enter any parameters you need to pass to the channel device layer. This includes the I/O ports of your ctc, escon and lcs devices. (e..g. ctc0, 0x600, 0x601 will activate the ctc0 interface at I/O ports 0x600, 0x601) lcs,0xe26,0xe27,2 (1) for OSA-2 with LCS or (2) for OSA-Express with QDIO/QETH 1 OSA Device Address (e.g. fc20,1 or auto (may not work)): 0xe26,1 error: no lcs module found SIOCSIFADDR: No such device eth0: unknown interface: No such device SIOCSIFNETMASK: No such device SIOCSIFBRDADDR: No such device eth0: unknown interface: No such device SIOCADDRT: No such device SIOCADDRT: No such device ... Thanks, Jim [EMAIL PROTECTED]
Re: tk GUI Applications
Thanks for your help, Mark. The problem actually was with the window manager I was using. When I used the MixServer from MicroImages, the tkCVS application came up but none of the letters were legible. Once I tried with CYGWIN X-windows and Linux/390's twm, I am getting the tkCVS windows clearly. Thanks, Samy Rengasamy. -Original Message- From: Post, Mark K [mailto:[EMAIL PROTECTED]] Sent: Friday, December 28, 2001 3:05 PM To: [EMAIL PROTECTED] Subject: Re: tk GUI Applications Samy, You might try checking out the XFree86 Font De-uglification HOWTO at http://www.linuxdoc.org/HOWTO/mini/FDU/. Mark Post -Original Message- From: Rengasamy, Samy [mailto:[EMAIL PROTECTED]] Sent: Thursday, December 27, 2001 7:36 PM To: [EMAIL PROTECTED] Subject: tk GUI Applications If I try to run any tk GUI application, the fonts do not appear clearly. What am I missing? Thanks, Samy Rengasamy. -Original Message- From: saparnis, carol [mailto:[EMAIL PROTECTED]] Sent: Thursday, December 27, 2001 11:16 AM To: [EMAIL PROTECTED] Subject: Re: Help, copy_from_user_asm is unresolved... Hi! Thank you! Thank you! Thank you! That was it! I am compiling outside the normal build process. __KERNEL__, __SMP__, and __MODULE__ are defined in my source, so my compile statement looks like: gcc -Wall -I/usr/include -O2 -c tstioctl.c Are there other parameters that I should be including? Thanks again, Carol -Original Message- From: Ulrich Weigand [mailto:[EMAIL PROTECTED]] Sent: Thursday, December 27, 2001 12:00 PM To: saparnis, carol Cc: [EMAIL PROTECTED] Subject: Re: Help, copy_from_user_asm is unresolved... Carol Saparnis wrote: I think I am doing something wrong, but I'm not sure what. I'm am trying to write a very simple IOCTL module using kernel version 2.4.5 from TurboLinux. I have included uaccess.h and I call the module copy_from_user. The module compiles correctly, but when I do an insmod on it I get an unresolved symbol __copy_from_user_asm message. I think, since it's really copy_from_user in uaccess.h that calls __copy_from_user_asm, that the right header file is being included, but I'm confused because that is the same header file that defines __copy_from_user_asm. What do I need to do to resolve that symbol? Are you using the proper compile options? If you compile a module outside of the regular kernel build process, you need to make sure you use the correct compiler command line. In particular, your symptom might indicate that you are not using the -O2 compiler option, and therefore __copy_from_user_asm is not inlined as it is supposed to ... Bye, Ulrich -- Dr. Ulrich Weigand [EMAIL PROTECTED]
Re: S390/zSeries CPU questions
On Fri, 2001-12-28 at 22:01, David Boyes wrote: Are there any performance counters such as other processors have (cache hit/miss, tlb hit/miss, branch predict misses, insns retired etc.) available from Linux? Hmm. If gcc generates the code to collect them (or gprof), then they should be available in the same ways as on the other platforms. gprof support isn't all that mature for the 390, so it may not work correctly yet. I haven't tried this in a while, so don't have a good answer for you now, but there's the weekend...8-) gprof support not mature? Soo, how immature is it? Also, I think there is a misunderstanding about the performance counters: they (the ones I ment) are CPU architecture specific and neither gcc nor gprof knows about them. They are implemented as extra registers (they are not useable for anything but collecting performance metrics) which the CPU updates automatically for whatever events you selected to monitor, and you typically dump them from the CPU after your program exits. Erik
Re: Red Hat RC2 Install Problem on MP3000
Jim, Absolutely. You can define virtual CTCs or IUCV links between your new guest and either the VM TCP/IP stack, or the SuSE guest. You'll need to be careful about assigning IP addresses and the associated routing. There's been a ton of discussion on the mailing list about that, and both Romney White and Alan Altmark have been very helpful with information on what needs to be done there. I'm afraid I don't understand why you're giving up on the OCO modules at this time. (I don't believe it's _hopeless_ just not guaranteed because no one else has been proclaiming success as of yet.) Mark Post -Original Message- From: Jim Rich [mailto:[EMAIL PROTECTED]] Sent: Friday, December 28, 2001 5:05 PM To: [EMAIL PROTECTED] Cc: Post, Mark K Subject: Re: Red Hat RC2 Install Problem on MP3000 Mark, Is there any way to bring up Red Hat RC2 without the OCO modules? I'm running z/VM V4R2 on the MP3000, with two OS/390 guests plus SuSE SLES7 already. Regards, Jim - Original Message - From: Post, Mark K To: 'Jim Rich' ; [EMAIL PROTECTED] Sent: Friday, December 28, 2001 3:53 PM Subject: RE: Red Hat RC2 Install Problem on MP3000 Jim, Here's what you said: ...the same setup I use successfully for SuSE on another MP3000 system, which is LCS3172 address pair... LCS=Lan Channel Station=OSA. Your PC ethernet card is being emulated as an OSA card to your MP3K. SuSE includes the IBM OCO modules, so that's why you didn't have a problem with that distribution. Red Hat and IBM claim that if you follow the directions to include the modules yourself, that it will work. If you get it to work, I ask that you report that to the mailing list, as it will be the first public report of such a success with Red Hat and their later kernels. The URL you want is this: http://www10.software.ibm.com/developerworks/opensource/linux390/special_oco _rh_2.4.shtml Watch out for the mailing list server chopping that up; it's pretty long. You can get to there from here: http://www10.software.ibm.com/developerworks/opensource/linux390/index.shtml Your dialog responses looked reasonable, but since I don't have access to a system with an OSA card (real or emulated) I haven't been able to play around with the installation script to determine just what works and what doesn't. Still, if you have problems, I'm willing to work with you to try to figure them out. Mark Post
Re: S390/zSeries CPU questions
Are there any performance counters such as other processors have (cache hit/miss, tlb hit/miss, branch predict misses, insns retired etc.) available from Linux? Hmm. If gcc generates the code to collect them (or gprof), then they should be available in the same ways as on the other platforms. gprof support isn't all that mature for the 390, so it may not work correctly yet. I haven't tried this in a while, so don't have a good answer for you now, but there's the weekend...8-) , machine = 9672 Model of the box. OK, so *any* way to tell which type of CPU type (G3, G4, G5...) I'm running on? (the nice people letting me use this guest are on vacation, and I hate waiting for them to get back :-)) With just the CPUID, I don't think so. The CPUID doesn't contain that information -- you might be able to guess from the serial number, but I wouldn't depend on the answer very much.I also probably generalized a bit too far; it's really more of a processor architecture family than a model number -- up until the IFLs and IEEE FP support showed up, it really didn't matter much, and the processor family type doesn't change with model numbers (ie, it's always a 9672-mumble, and the mumble tells you how many processors of what type are installed, eg a 9672-RA5 and a 9672-RB5 are G4s, both are 9672 family architecture, and the only difference is an RB5 is 2 engines and a RA5 is one engine). The Z boxes are a different architecture type (2064), but that probably doesn't help you much. -- Cache sizes and speeds (bw and lat) for the different levels of cache? (I could use lmbench, but I only have access to one CPU) IBM Journal of RD occasionally publishes some interesting articles in this vein. Thanks, that paid off immediately, if someone wonders the URL is: http://www.research.ibm.com/journal/rd And they have many nice articles online. IBM generally doesn't publish stuff like this, but the RD journal is usually one of the few places such data gets published if it ever does. It's also a really nice window into the thousands of areas in which IBM does basic research. Not easy reading (the math can get pretty hairy), but worthwhile. -- db
Re: Red Hat RC2 Install Problem on MP3000
Mark, I'm not giving up on the OCO modules, but there is not enough time left this year to obtain the GA Red Hat, obtain a PC and install x86 Red Hat 7.2, download the required new 2.4.9-17 kernel, follow the PDF installation procedure to merge in the OCO, and, finally, complete the MP3000 install. If I can get the RC2 I already have installed, at least we'll have something to try now. Thanks, Jim - Original Message - From: Post, Mark K To: 'Jim Rich' ; [EMAIL PROTECTED] Sent: Friday, December 28, 2001 4:18 PM Subject: RE: Red Hat RC2 Install Problem on MP3000 Jim, Absolutely. You can define virtual CTCs or IUCV links between your new guest and either the VM TCP/IP stack, or the SuSE guest. You'll need to be careful about assigning IP addresses and the associated routing. There's been a ton of discussion on the mailing list about that, and both Romney White and Alan Altmark have been very helpful with information on what needs to be done there. I'm afraid I don't understand why you're giving up on the OCO modules at this time. (I don't believe it's _hopeless_ just not guaranteed because no one else has been proclaiming success as of yet.) Mark Post -Original Message- From: Jim Rich [mailto:[EMAIL PROTECTED]] Sent: Friday, December 28, 2001 5:05 PM To: [EMAIL PROTECTED] Cc: Post, Mark K Subject: Re: Red Hat RC2 Install Problem on MP3000 Mark, Is there any way to bring up Red Hat RC2 without the OCO modules? I'm running z/VM V4R2 on the MP3000, with two OS/390 guests plus SuSE SLES7 already. Regards, Jim
Re: Red Hat RC2 Install Problem on MP3000
Jim, Getting the Red Hat GA isn't much. There are some files that are different from the RC2, but you can figure those out and only download them. You don't need an x86 Linux, you have your SuSE system on the same box. Use that instead. The 2.4.9-17 kernel is part of the GA code, so that's taken care of above. The reason why I'm pursuing this path with you is that if you don't know what you're doing, using a vCTC or IUCV connection to either VM, or your SuSE image can create large headaches before you get it straightened out. If you _do_ know what you're doing, then fine, go for it. If you don't, find someone who does to help you, or you'll regret it. And use a trick that has been recommended before. When working with a TCP/IP person that's not familiar with mainframes and specifically Linux/390, draw all the systems involved, but label them as routers and not mainframe images. They'll feel better, and you'll get your network set up. Mark Post -Original Message- From: Jim Rich [mailto:[EMAIL PROTECTED]] Sent: Friday, December 28, 2001 5:37 PM To: [EMAIL PROTECTED] Cc: Post, Mark K Subject: Re: Red Hat RC2 Install Problem on MP3000 Mark, I'm not giving up on the OCO modules, but there is not enough time left this year to obtain the GA Red Hat, obtain a PC and install x86 Red Hat 7.2, download the required new 2.4.9-17 kernel, follow the PDF installation procedure to merge in the OCO, and, finally, complete the MP3000 install. If I can get the RC2 I already have installed, at least we'll have something to try now. Thanks, Jim - Original Message - From: Post, Mark K To: 'Jim Rich' ; [EMAIL PROTECTED] Sent: Friday, December 28, 2001 4:18 PM Subject: RE: Red Hat RC2 Install Problem on MP3000 Jim, Absolutely. You can define virtual CTCs or IUCV links between your new guest and either the VM TCP/IP stack, or the SuSE guest. You'll need to be careful about assigning IP addresses and the associated routing. There's been a ton of discussion on the mailing list about that, and both Romney White and Alan Altmark have been very helpful with information on what needs to be done there. I'm afraid I don't understand why you're giving up on the OCO modules at this time. (I don't believe it's _hopeless_ just not guaranteed because no one else has been proclaiming success as of yet.) Mark Post -Original Message- From: Jim Rich [mailto:[EMAIL PROTECTED]] Sent: Friday, December 28, 2001 5:05 PM To: [EMAIL PROTECTED] Cc: Post, Mark K Subject: Re: Red Hat RC2 Install Problem on MP3000 Mark, Is there any way to bring up Red Hat RC2 without the OCO modules? I'm running z/VM V4R2 on the MP3000, with two OS/390 guests plus SuSE SLES7 already. Regards, Jim