Re: LCS drivers for 2.4.9 ?
I don't know. Is there any significance to the fact that you cannot get even ONE for a MP3000? But the LCS3172 driver provides a virtual device to the S/390 that smells like one...
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: 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
Re: LCS drivers for 2.4.9 ?
Vic wrote: On this list, there seem to be more Linux people than mainframe people (or maybe the Linux folk are more vocal), so you'd be forgiven for thinking that popular opinion goes in favour of RedHat. But it's been discussed in the past that most Linux/390 or zLinux installations will be in sites that already have a mainframe--so they'll be trying to come to grips with the fact that they've got all this 'open source' stuff on their system! Heck, I'd go so far as to say that some shops might even be glad of a little IBM OCO that they can trust!!! ;) And among the mainframe people are some who remember the great OCO war of the late 80's, early 90's between the VM world and IBM. A compromise (of sorts) was reached where that part of VM that had always been source code would remain so and so would new features that were not related to company trade secrets. The paradigm for VM had always been the availability of source code. A lot of the innovation in the early days of VM was driven by the user world because the source code was available.
Re: LCS drivers for 2.4.9 ?
And among the mainframe people are some who remember the great OCO war of the late 80's, early 90's between the VM world and IBM. A compromise (of sorts) was reached where that part of VM that had always been source code would remain so and so would new features that were not related to company trade secrets. The paradigm for VM had always been the availability of source code. A lot of the innovation in the early days of VM was driven by the user world because the source code was available. Are there some history pages about these discussions? Are parts of the current VM code available from IBM? How much of that code has changed from being available then to closed source? Seems like good-old VM customers understand how important source code is for a better operating system... cu, Florian La Roche
Re: LCS drivers for 2.4.9 ?
I think that most installations where Linux/390 is being installed or tried have already accepted that large business risk by installing and using IBM hardware and software in the first place. All of which is patented, copyrighted and licensed. I doubt that using a few kB of OCO OSA drivers is going to be a problem for them. As for the developers I'm sure that IBM will accept their problem reports and either tell them how to fix/bypass the problem or when they can expect to see a fix. The industry has been operating that way for over thirty years and it has served us well. And I think RedHat severly dropped the ball. They could have easily done the testing and talked with IBM about resolving any problems that arose and had the OSA drivers tested and shipped with their package. But, they cut out any installation that uses OSA cards and have probably lost a lot of business. What will they decide to drop from the next release? Henry Schaffer [EMAIL PROTECTED] on 12/26/2001 04:18:18 PM Please respond to Linux on 390 Port [EMAIL PROTECTED] To: [EMAIL PROTECTED] cc:(bcc: Dennis Wicks/infosvcs/CDG) Subject: Re: LCS drivers for 2.4.9 ? Mark writes: That's an interesting non sequitur. I don't think its a non sequitur - having something OCO means it can't be updated for new kernel releases, it can't be investigated when problems are happening, it depends on IBM for everything. That is a real business risk for people who depend on OSA cards for 390 data communications. --henry schaffer Mark Post -Original Message- From: Alan Cox [mailto:[EMAIL PROTECTED]] Sent: Wednesday, December 26, 2001 3:42 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Remember, the ONLY Linux kernel OCO code IBM supplies, to my knowlede, is for OSA cards! And what about next month, or next year. In the PC world I can rip out a card if a vendor screws me, and go elsewhere. Its a $200 annoyance not a million dollar business risk. Alan
Re: LCS drivers for 2.4.9 ?
Gregg, I am not 100% sure if you're talking about the Slackware port only, or something else that goes with it. You also didn't mention any names of the Slackware folks you spoke with at LWE. However, yes, the people at Slackware were aware of the port. Mike Kershaw and I had been in frequent contact with Chris Lumens while we were building the port. We had less frequent contact with the other Slackware developers. Plus, I chatted with all of them in IRC from time to time about the project. The port is still in progress. We're basically trying to find time from work and personal life to test the installers that Mike Kershaw developed (twice, after I managed to delete the first version!). As a matter of fact, our Slackware/390 image has been running happily for 60 days or so (as I take a quick peek at the uptime) with no real apparent problems. Mark Post -Original Message- From: Gregg C Levine [mailto:[EMAIL PROTECTED]] Sent: Wednesday, December 26, 2001 9:27 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? -snip- Personal to :Mike Kershaw if you are listening, I'd like to demo on that emulator thingie the port of Slackware that you built for S/390. And has it been leaked to the folks at Slackware that it exists? I discussed it briefly with the folks who were representing the company at Linux World Expo, this past January, but I do not think they took it seriously. --- Gregg C Levine [EMAIL PROTECTED]
Re: LCS drivers for 2.4.9 ?
Florian La Roche [EMAIL PROTECTED] Are there some history pages about these discussions? The entire history of VM, including Melinda Varian's VM and the Community paper (an excellantly researched history, based on years of interviews with key players) is online. The VM community literally started the computer bulletin board craze with VMSHARE way back in 1976. VMSHARE is no longer active, having been superceeded by the VMESA-L list and assorted web sites, but the entire 22 years of discussion has been archived and is available on the web at http://pucc.princeton.edu/~vmshare. Melinda's history is available at http://pucc.princeton.edu/~melinda in a variety of formats. Lynn Wheeler, one of the early VM folks inside IBM, has been posting regularly to the comp.arch and alt.folklore.computers newsgroups for years, and maintains a huge archive of interesting S/3x0 and VM historical info at http://www.garlic.com/~lynn/subtopic.html. Are parts of the current VM code available from IBM? It's hard to say how much of the current VM code is available, but VM sites have always had some source and many of us believe that today we have the lion's share of it. We are prevented by license terms from sharing it with others, but that's not an OCO issue, just a licensing one. How much of that code has changed from being available then to closed source? By current terminology, it's all closed source, because none of it is licensed under any generally accepted Open Source license. I think you're asking how much source that was once available under license is no loger available. The answer for the current codebase is, I believe, none. Yes, there are parts that are not available in source, but to my knowledge those sources have never been available except under non-disclosure terms. One might rather ask the opposite - how much has changed from no-source to source-available? That answer is much more revealing. The 1990s saw a great re-sourcing of VM, resulting in the largest components of VM being mostly source-available. Seems like good-old VM customers understand how important source code is for a better operating system... Yup. Paradoxically, when those of us old enough to remember POSIX being new asked IBM to include POSIX facilities in VM, that request created a huge pool of code that IBM delivered only in object code. It seems some of the licenses IBM had to accept prevented them from sharing the source with their customers. The open systems support in VM wasn't even closed source, it was OCO! Ross Patterson Computer Associates
Re: LCS drivers for 2.4.9 ?
I would imagine that some of that discussion has been preserved. Up until recently I had a copy of the OCO white paper that was developed by the VM group at SHARE. At that time (and perhaps still) it was the most cogent argument as to why IBM should not remove access to the source code for VM. I think people like David Boyes, Harry Williams, etc., will be better able to answer in more detail. What everyone forgets, though is that not only was the source code for VM available, but for MVS as well. (True, it was PL/S, and only on microfiche, but we _had_ it.) It was very useful for debugging problems at a time when the IBM support folks were much less skilled and helpful than they are today (Hi Mike!). Mark Post -Original Message- From: Florian La Roche [mailto:[EMAIL PROTECTED]] Sent: Thursday, December 27, 2001 7:00 AM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? And among the mainframe people are some who remember the great OCO war of the late 80's, early 90's between the VM world and IBM. A compromise (of sorts) was reached where that part of VM that had always been source code would remain so and so would new features that were not related to company trade secrets. The paradigm for VM had always been the availability of source code. A lot of the innovation in the early days of VM was driven by the user world because the source code was available. Are there some history pages about these discussions? Are parts of the current VM code available from IBM? How much of that code has changed from being available then to closed source? Seems like good-old VM customers understand how important source code is for a better operating system... cu, Florian La Roche
Re: LCS drivers for 2.4.9 ?
Hello from Gregg C Levine normally with Jedi Knight Computers Sheepish grin inserted here Mark, I did not mention who I spoke with, because I actually do not remember who I spoke with, then. It has been about a full year since then, after all, and the next LWE is practically around the corner. I just remember mentioning the subject off-hand, to the single representative who was there, this past show. He did mention, that the subject was being pursued, there, based on the successes that they had, with a few of the other ports. The reason why I think they did not take it seriously, then, because the other distribution makers were, you know the ones, Red Hat, TurboLinux, Suse, so there you are. But that was my opinion then, yesterday when I had posted the note. Rick Troth, posted me a polite note, off list, that he, and I, came to the same conclusions, independently regarding the subject matters of the ways that kernel sources are being distributed. That because I nominated our distribution, for the way it distributed kernel sources, over everyone else. And now that I have had to time to think about it, .I actually think that was the name of the person, that I met, there, then. I suppose it will come back to me, full force, after I walk onto the show floor at LWE, next time. Why? Do you plan on being in attendance this year? --- Gregg C Levine [EMAIL PROTECTED] The Force will be with you...Always. Obi-Wan Kenobi Use the Force, Luke. Obi-Wan Kenobi (This company dedicates this E-Mail to General Obi-Wan Kenobi ) (This company dedicates this E-Mail to Master Yoda ) -Original Message- From: Linux on 390 Port [mailto:[EMAIL PROTECTED]] On Behalf Of Post, Mark K Sent: Thursday, December 27, 2001 12:52 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Gregg, I am not 100% sure if you're talking about the Slackware port only, or something else that goes with it. You also didn't mention any names of the Slackware folks you spoke with at LWE. However, yes, the people at Slackware were aware of the port. Mike Kershaw and I had been in frequent contact with Chris Lumens while we were building the port. We had less frequent contact with the other Slackware developers. Plus, I chatted with all of them in IRC from time to time about the project. The port is still in progress. We're basically trying to find time from work and personal life to test the installers that Mike Kershaw developed (twice, after I managed to delete the first version!). As a matter of fact, our Slackware/390 image has been running happily for 60 days or so (as I take a quick peek at the uptime) with no real apparent problems. Mark Post -Original Message- From: Gregg C Levine [mailto:[EMAIL PROTECTED]] Sent: Wednesday, December 26, 2001 9:27 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? -snip- Personal to :Mike Kershaw if you are listening, I'd like to demo on that emulator thingie the port of Slackware that you built for S/390. And has it been leaked to the folks at Slackware that it exists? I discussed it briefly with the folks who were representing the company at Linux World Expo, this past January, but I do not think they took it seriously. --- Gregg C Levine [EMAIL PROTECTED]
Re: LCS drivers for 2.4.9 ?
On Thu, 27 Dec 2001, Dennis G. Wicks wrote: I think that most installations where Linux/390 is being installed or tried have already accepted that large business risk by installing and using IBM hardware and software in the first place. All of which is patented, copyrighted and licensed. I doubt that using a few kB of OCO OSA drivers is going to be a problem for them. This is quite a bit different. If IBM suddenly decides to stop building the hardware or goes out of business or whatever, your old hardware will continue to work just fine. If, on the other hand, they discontinue the OCO driver, you're stuck with an old kernel forever (kernel modules need to be recompiled for other kernels). Take a look at the current situation, for example: The only kernel officially supported by the OCO modules is 2.4.7, which shouldn't be run in production on any networked machine because it has remote root security bugs. And that's while IBM is officially supporting those modules. And I think RedHat severly dropped the ball. They could have easily done the testing and talked with IBM about resolving any problems that arose and had the OSA drivers tested and shipped with their package. It is impossible to support a kernel that was tainted by modules without source code. Anything running in kernel space can severely mess up anything else in kernel space. It's entirely possible that using an OCO module can cause filesystem corruption, program crashes and other bad things totally unrelated to the function of the module. How do you debug this without having access to the source code? By shipping a binary only kernel module, you lose the ability to support your kernel to whoever controls this kernel module, period. Anyone claiming otherwise either doesn't know anything about kernel programming, or is telling marketing lies. But, they cut out any installation that uses OSA cards and have probably lost a lot of business. We'd rather lose some business than having to do bad service to a customer. What would you think of a company you bought some expensive support package from telling you we can't fix this, you just wasted the money for your support contract.? I don't think you'd like that. What will they decide to drop from the next release? Nothing - all of Red Hat Linux is under a sane license, so there's no need whatsoever to drop anything. If Linus were bought out by Microsoft and decided to take the Linux kernel binary-only, we could (and would) still ship a fork of the older, free version, so you don't run into the risk of Red Hat dropping driver xyz because the current version is binary-only. LLaP bero -- This message is provided to you under the terms outlined at http://www.bero.org/terms.html
Re: LCS drivers for 2.4.9 ?
Sorry, there is no difference at all. My current hardware and software are working just fine, thank you. So it is not at all different than if IBM or RedHat had gone belly up. I would still be stuck with an old kernel or operating system, exactly as I am now because I have no upgrade path ... not at present that is. So I will not be installing the latest and greatest because it does not include a driver I need. And apparently any support agreement will be null and void if I install the one I do need. Even if it works. ... Have to install DB2/VM and a OSA-X card. ... Really don't have time for this! ... Shouldn't have started in the first place. Bernhard Rosenkraenzer [EMAIL PROTECTED] on 12/27/2001 01:00:38 PM Please respond to Linux on 390 Port [EMAIL PROTECTED] To: [EMAIL PROTECTED] cc:(bcc: Dennis Wicks/infosvcs/CDG) Subject: Re: LCS drivers for 2.4.9 ? On Thu, 27 Dec 2001, Dennis G. Wicks wrote: I think that most installations where Linux/390 is being installed or tried have already accepted that large business risk by installing and using IBM hardware and software in the first place. All of which is patented, copyrighted and licensed. I doubt that using a few kB of OCO OSA drivers is going to be a problem for them. This is quite a bit different. If IBM suddenly decides to stop building the hardware or goes out of business or whatever, your old hardware will continue to work just fine. If, on the other hand, they discontinue the OCO driver, you're stuck with an old kernel forever (kernel modules need to be recompiled for other kernels). Take a look at the current situation, for example: The only kernel officially supported by the OCO modules is 2.4.7, which shouldn't be run in production on any networked machine because it has remote root security bugs. And that's while IBM is officially supporting those modules. And I think RedHat severly dropped the ball. They could have easily done the testing and talked with IBM about resolving any problems that arose and had the OSA drivers tested and shipped with their package. It is impossible to support a kernel that was tainted by modules without source code. Anything running in kernel space can severely mess up anything else in kernel space. It's entirely possible that using an OCO module can cause filesystem corruption, program crashes and other bad things totally unrelated to the function of the module. How do you debug this without having access to the source code? By shipping a binary only kernel module, you lose the ability to support your kernel to whoever controls this kernel module, period. Anyone claiming otherwise either doesn't know anything about kernel programming, or is telling marketing lies. But, they cut out any installation that uses OSA cards and have probably lost a lot of business. We'd rather lose some business than having to do bad service to a customer. What would you think of a company you bought some expensive support package from telling you we can't fix this, you just wasted the money for your support contract.? I don't think you'd like that. What will they decide to drop from the next release? Nothing - all of Red Hat Linux is under a sane license, so there's no need whatsoever to drop anything. If Linus were bought out by Microsoft and decided to take the Linux kernel binary-only, we could (and would) still ship a fork of the older, free version, so you don't run into the risk of Red Hat dropping driver xyz because the current version is binary-only. LLaP bero -- This message is provided to you under the terms outlined at http://www.bero.org/terms.html
Re: Kernel versioning (was Re: LCS drivers for 2.4.9 ?)
My mistake, I confused myself. Time to change feet... I think what I was getting at is the packaging/kernel versioning used by RedHat, which means that their 2.4.9 kernel is not called '2.4.9', but '2.4.9-something special', which means you can't fit a version-labelled module into it. My understanding of the problems Mark and others were having trying to fit the 2.4.5 modules into the RedHat beta 2.4.5 kernel. Still, RedHat aren't the only ones doing it: SuSE's first 2.4 kernel (at least on Intel) was 2.4.0-4GB, IIRC... There is a good reason for the way RH names its kernels. Take a look in the src.rpm for 2.4.9-13 (which is what I have on IA32) you will find the bast 2.4.9 tarball and an enormous number of patches, including ext3 support. Calling it 2.4.9 would be misleading, it's not that. If I build my own kernel from RH sources, it's 2.4.9custom or similar (I don't have one handy). -- Cheers John Summerfield Microsoft's most solid OS: http://www.geocities.com/rcwoolley/ Note: mail delivered to me is deemed to be intended for me, for my disposition.
Re: Kernel versioning (was Re: LCS drivers for 2.4.9 ?)
On Fri, 28 Dec 2001, John Summerfield wrote: There is a good reason for the way RH names its kernels. ... SuSE is following RedHat's lead and making it harder for the customer to run third party modules. I'm wondering if version information on all symbols might help. That's the CONFIG_MODVERSIONS flag. It causes heartburn for me when building the *kernel*, but I wonder if it might help when IBM builds their *modules*. The documentation is not clear: It implies that this flag applies to the kernel. Does it instead apply to the modules? At least most of these one off kernel labels are parsable. If it were simply for human consumption (for doc), it would be fine, helpful even. The problem is when something other than a person has to grok the string and it's not an exact match for anything known. 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?? -- Rick Troth, BMC Software, Inc. 2101 City West Blvd., Houston, Texas, USA, 77042 1-800-841-2031
Re: LCS drivers for 2.4.9 ?
Let's say you ran a large, profitable company like IBM. You're risking a lot in supporting something like Linux. 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). Would you not want to have a little quality control? Granted, it's not your father's IBM anymore, but my experiences with IBM have been that they try to make good products and try to support them well. (unlike some other software companies personal opinionMS, CA/personal opinion) Neither am I an IBM bigot. I've dealt with them doing their fair share of stupid things, but when it comes down to it, if they want to take measures to ensure a better product, then I'm all for it. However, if you want to shake salt on them for trying to ensure that Linux looks good and runs well, then that's your business. --Brad -Original Message- From: Alan Cox [mailto:[EMAIL PROTECTED]] Sent: Friday, December 21, 2001 11:18 AM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? It is a question of how IBM can not only reduce their own costs (to currently produce a separate driver for a subset of kernels of interest to their customers), but also make sure that the driver exists for all kernels that their customers run. Looks to me like it is a win/win situation for IBM to fix this, and a lose/lose to not... You are naively assuming IBM wishes their customers to be able to run kernels that IBM haven't sanctified. IBM seems to be handling OCO rather intentionally as a way to ensure they have a firm grip on the testicles of their customers Alan
Re: LCS drivers for 2.4.9 ?
It just goes to show that despite all the Peace, Love and Linux BS, IBM still doesn't grok Open Source. -Original Message- From: Snyder, Bradley (LNG) [mailto:[EMAIL PROTECTED]] Sent: Monday, December 24, 2001 10:48 AM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Let's say you ran a large, profitable company like IBM. You're risking a lot in supporting something like Linux. 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). Would you not want to have a little quality control? Granted, it's not your father's IBM anymore, but my experiences with IBM have been that they try to make good products and try to support them well. (unlike some other software companies personal opinionMS, CA/personal opinion) Neither am I an IBM bigot. I've dealt with them doing their fair share of stupid things, but when it comes down to it, if they want to take measures to ensure a better product, then I'm all for it. However, if you want to shake salt on them for trying to ensure that Linux looks good and runs well, then that's your business. --Brad -Original Message- From: Alan Cox [mailto:[EMAIL PROTECTED]] Sent: Friday, December 21, 2001 11:18 AM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? It is a question of how IBM can not only reduce their own costs (to currently produce a separate driver for a subset of kernels of interest to their customers), but also make sure that the driver exists for all kernels that their customers run. Looks to me like it is a win/win situation for IBM to fix this, and a lose/lose to not... You are naively assuming IBM wishes their customers to be able to run kernels that IBM haven't sanctified. IBM seems to be handling OCO rather intentionally as a way to ensure they have a firm grip on the testicles of their customers Alan
Re: LCS drivers for 2.4.9 ?
not the silver bullet of all IT, and it is not the most stable operating system in existence (nor even close). Would you not want to have a little quality control? Mummy knows best What are you going to do if IBM turns around says we're bored of this OCO hassle we are dropping all support for our ethernet, goodbye have a nice life ? QA is saying if you use this set up it works, not 'we are the mighty IBM there will be no other setups' And if you think IBM internally is capable of writing OCO drivers superior to stuff reviewed by many people I've got this bridge for sale. Alan
Re: LCS drivers for 2.4.9 ?
It just goes to show that despite all the Peace, Love and Linux BS, IBM still doesn't grok Open Source. Or city rules on graffiti 8)
Re: LCS drivers for 2.4.9 ?
I'm not speaking IBM as a corporation, just expressing my own observations. As IBM supplies the great bulk of its S/390 modifications as source patches, including CKD dasd, I don't see OCO for OSA cards as a Linux issue, per se, nor an open source issue. 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. In this aspect, I would tend to think that the IBM Linux supporters and developers are bound by issues outside of Linux when OCO code is distributed. Remember, the ONLY Linux kernel OCO code IBM supplies, to my knowlede, is for OSA cards! Until such time as the OSA card interface is public domain, I guess we're stuck. Regards, Jim Linux S/390-zSeries Support, SEEL, IBM Silicon Valley Labs t/l 543-4021, 408-463-4021, [EMAIL PROTECTED] ** Grace Happens **
Re: LCS drivers for 2.4.9 ?
Remember, the ONLY Linux kernel OCO code IBM supplies, to my knowlede, is for OSA cards! And what about next month, or next year. In the PC world I can rip out a card if a vendor screws me, and go elsewhere. Its a $200 annoyance not a million dollar business risk. Alan
Re: LCS drivers for 2.4.9 ?
That's an interesting non sequitur. Mark Post -Original Message- From: Alan Cox [mailto:[EMAIL PROTECTED]] Sent: Wednesday, December 26, 2001 3:42 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Remember, the ONLY Linux kernel OCO code IBM supplies, to my knowlede, is for OSA cards! And what about next month, or next year. In the PC world I can rip out a card if a vendor screws me, and go elsewhere. Its a $200 annoyance not a million dollar business risk. Alan
Re: LCS drivers for 2.4.9 ?
Mark writes: That's an interesting non sequitur. I don't think its a non sequitur - having something OCO means it can't be updated for new kernel releases, it can't be investigated when problems are happening, it depends on IBM for everything. That is a real business risk for people who depend on OSA cards for 390 data communications. --henry schaffer Mark Post -Original Message- From: Alan Cox [mailto:[EMAIL PROTECTED]] Sent: Wednesday, December 26, 2001 3:42 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Remember, the ONLY Linux kernel OCO code IBM supplies, to my knowlede, is for OSA cards! And what about next month, or next year. In the PC world I can rip out a card if a vendor screws me, and go elsewhere. Its a $200 annoyance not a million dollar business risk. Alan
Re: LCS drivers for 2.4.9 ?
And none of that has anything to do with what Jim Sibley was saying. Hence, the non sequitur. I guess some people are more interested in flaming IBM than in reading what people are actually writing. Mark Post -Original Message- From: Henry Schaffer [mailto:[EMAIL PROTECTED]] Sent: Wednesday, December 26, 2001 5:18 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Mark writes: That's an interesting non sequitur. I don't think its a non sequitur - having something OCO means it can't be updated for new kernel releases, it can't be investigated when problems are happening, it depends on IBM for everything. That is a real business risk for people who depend on OSA cards for 390 data communications. --henry schaffer Mark Post -Original Message- From: Alan Cox [mailto:[EMAIL PROTECTED]] Sent: Wednesday, December 26, 2001 3:42 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Remember, the ONLY Linux kernel OCO code IBM supplies, to my knowlede, is for OSA cards! And what about next month, or next year. In the PC world I can rip out a card if a vendor screws me, and go elsewhere. Its a $200 annoyance not a million dollar business risk. Alan
Re: LCS drivers for 2.4.9 ?
No vendor ships Linus base kernel. Linus base kernel doesn't pass anyones QA test suite. Linus role is to put out clean well designed code and to ensure development takes the right paths. The vendors then all add on top of that various things including bug fixes which while they may fix the bug are not the right long term solution and so won't go into Linus tree. True: No vendor ships Linus base kernel. HOWEVER, there may be non vendor distros which use the base kernel, and (more to the point) there are customers who can and do, and some say should, build their own kernel, which would be base. Personally, I always run my own kernel on RH, SuSE, Slackware, whether that be INTeL, S/390, or something else. But we're getting off-topic w/r/t the thread and subject to criticize RH, SuSE, Turbo for distributing customized kernels. Several clearer heads have already chimed in: o the differing business models will always be incompatible in the extreme o IBM evidently has issues other than the driver code itself that prevent it releasing the driver code source (think about it; think about how an unscrupulous lawyer might might twist the positive precedent of releasing source into a less-business-friendly argument about dissolving the patent) o IBM, RedHat, (indeed, BMC!) do not love their customers per se Thankfully, there are those within IBM, RedHat, (indeed, within BMC!) who recognize that customer relationships have more lasting value than intellectual property, and some genuinely love their customers for more than money. The second bullet there I think is the crux of this matter. And the statement doesn't *solve* the problem, but neither does the word war we've waged. Work the problem. Folks, we need to think carefully. Nothing wrong with argument. Just be constructive. Doesn't matter whether it's Alan Cox or Alan Altmark, each man has gotta eat. Argue productively. -- Rick Troth, BMC Software, Inc. 2101 City West Blvd., Houston, Texas, USA, 77042 1-800-841-2031
Re: LCS drivers for 2.4.9 ?
Hello from Gregg C Levine normally with Jedi Knight Computers Then why is Slackware constantly saying in their notes, on the kernel source code, that it is, and I quote here, from the one for 2.2.18, This is the complete and unmodified source code for the Linux kernel. As I have found out, Red Hat's versions, contain patches to match their idea of a standardized kernel, or at least that is how the Intel versions are shipped. The version 5.0 ones were like that. Oh, folks, before we get into personal gripes, and the maker knows what else, that is only my opinion. And I could not get the 6.2 one to build correctly, or even to properly boot on my machine. I switched to Slackware, because it is an easy distribution to work with. It runs on a crowd of machines. And in a pinch, or at least in a few months, I could create a port for S/390. And besides, I've met the man who built it. Oh, and flames to me personally, mind. Personal to :Mike Kershaw if you are listening, I'd like to demo on that emulator thingie the port of Slackware that you built for S/390. And has it been leaked to the folks at Slackware that it exists? I discussed it briefly with the folks who were representing the company at Linux World Expo, this past January, but I do not think they took it seriously. --- Gregg C Levine [EMAIL PROTECTED] The Force will be with you...Always. Obi-Wan Kenobi Use the Force, Luke. Obi-Wan Kenobi (This company dedicates this E-Mail to General Obi-Wan Kenobi ) (This company dedicates this E-Mail to Master Yoda ) -Original Message- From: Linux on 390 Port [mailto:[EMAIL PROTECTED]] On Behalf Of Alan Cox Sent: Wednesday, December 26, 2001 7:35 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? I like RedHat's words, but the term RedHat standard kernel bothers me a bit (isn't there only supposed to be *one* standard kernel?). And, the point has been made before, that IBM could be a bit more flexible. No vendor ships Linus base kernel. Linus base kernel doesn't pass anyones QA test suite. Linus role is to put out clean well designed code and to ensure development takes the right paths. The vendors then all add on top of that various things including bug fixes which while they may fix the bug are not the right long term solution and so won't go into Linus tree. Alan
Re: LCS drivers for 2.4.9 ?
Hello from Gregg C Levine normally with Jedi Knight Computers Say. Didn't I just throw that one up into the air, regarding Slackware? Still Rick has a heck of good point, we should be discussing this like professionals, and not flaming everyone, just because that person, may, or may not be right. Anyway, I may not agree with the decision to release a driver, only in the form of OCO, but it make sense, somewhere. And not just flapping in the wind either. --- Gregg C Levine [EMAIL PROTECTED] The Force will be with you...Always. Obi-Wan Kenobi Use the Force, Luke. Obi-Wan Kenobi (This company dedicates this E-Mail to General Obi-Wan Kenobi ) (This company dedicates this E-Mail to Master Yoda ) -Original Message- From: Linux on 390 Port [mailto:[EMAIL PROTECTED]] On Behalf Of Rick Troth Sent: Wednesday, December 26, 2001 8:57 PM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? No vendor ships Linus base kernel. Linus base kernel doesn't pass anyones QA test suite. Linus role is to put out clean well designed code and to ensure development takes the right paths. The vendors then all add on top of that various things including bug fixes which while they may fix the bug are not the right long term solution and so won't go into Linus tree. True: No vendor ships Linus base kernel. HOWEVER, there may be non vendor distros which use the base kernel, and (more to the point) there are customers who can and do, and some say should, build their own kernel, which would be base. Personally, I always run my own kernel on RH, SuSE, Slackware, whether that be INTeL, S/390, or something else. But we're getting off-topic w/r/t the thread and subject to criticize RH, SuSE, Turbo for distributing customized kernels. Several clearer heads have already chimed in: o the differing business models will always be incompatible in the extreme o IBM evidently has issues other than the driver code itself that prevent it releasing the driver code source (think about it; think about how an unscrupulous lawyer might might twist the positive precedent of releasing source into a less-business-friendly argument about dissolving the patent) o IBM, RedHat, (indeed, BMC!) do not love their customers per se Thankfully, there are those within IBM, RedHat, (indeed, within BMC!) who recognize that customer relationships have more lasting value than intellectual property, and some genuinely love their customers for more than money. The second bullet there I think is the crux of this matter. And the statement doesn't *solve* the problem, but neither does the word war we've waged. Work the problem. Folks, we need to think carefully. Nothing wrong with argument. Just be constructive. Doesn't matter whether it's Alan Cox or Alan Altmark, each man has gotta eat. Argue productively. -- Rick Troth, BMC Software, Inc. 2101 City West Blvd., Houston, Texas, USA, 77042 1-800-841-2031
Re: LCS drivers for 2.4.9 ?
o IBM evidently has issues other than the driver code itself that prevent it releasing the driver code source Ah, but we don't need the entire driver... just the pieces that interface w/ the kernel... let the super secret stuff stay OCO... this just requires the design of the driver such that the secrets are OCO and the parts that depend upon kernel structures are sourced... its not rocket science... just a bit of thinking about meeting both the customer's and the legal's needs to be met in a flexible manner is all rather than just thinking inside the box and ending up with a design that only meets the needs of the we can't do that folks! Peace. -njg
Re: LCS drivers for 2.4.9 ?
Just so everyone is clear: We (IBM) do not like to resort to OCO could have fooled me. but in this world it is the only way to protect the intellectual property present in the drivers. Oh you mean your network card has something that all the hundreds of others don't have ? Your patent department knows how to deal with that for sure. Unless you want to keep secret a breach of someone else's patent; or maybe that it's just a stock $15 NIC you guys sell for $8k ? If the drivers weren't OCO, anyone could step up to the challenge to provide support. But, when all the shouting is over, IBM or its delegate is the one who provides support for its OCO modules, not the Open Source community at large. What's the point here? IBM can't support the driver if it's open source? Extra eyes usually make support simpler because bugs are found (and fixes provided!!) by others. It is obvious that many on this list have differing views about what the word support means. By that, I mean more than just doing some coding. I mean that IBM will fix something that isn't working. That promise is not trivial and actually costs IBM real dollars to provide. We have people that design, code, test, and document our drivers. When there's a new driver, there's more testing. That means tying up REAL resources (people and machines). And adding free community resources to that is not something IBM is interested in? Your ad campains may have fooled me there I *know* this is frustrating to many and I am sorry for that, but we would rather focus our efforts on opening up the interface (a difficult task at best, with lots of legal complications) and eliminate the need for OCO drivers altogether. This is where the win/win is to be found. Somehow I doubt this extra layer will fix the poor performance the current drivers already have Given how some of the other s390 kernel code works, I wouldn't be surprised that once you guys open up the driver some of the linux networking experts fixes the performance issues in a matter of days. In the meantime, solutions can be had simply by routing your new Linux IP traffic through a certified Linux which owns the adapters. You mean using that IBM kernel which has *KNOWN* and *PUBLISHED* local and remote exploits? No thank you. If IBM actually puts such a kernel on customers machines then I see that as deliberatly selling a defective product; US tort laws have ways with dealing with that last I heard. This year IBM spent a lot of money on Linux, a lot of that in advertising on how great Linux is. IBM: this would be a call to put your actions where your mouth is. Greetings, Arjan van de Ven
Re: LCS drivers for 2.4.9 ?
Alan Altmark made a very cogent comment in another forum: it's a question of where we want IBM to put their resources, and taking a developer away from new function to fix old function or restructure a bunch of drivers won't Agreed. help the overall effort much. Then there's testing, etc, etc, etc -- it's not a free process to get something like this done. Open Source has lots of benefits for code review, bug hunting and testing. All those benefits are not possible for the OCO modules and thus hinder these better development practises. cu, Florian La Roche
Re: LCS drivers for 2.4.9 ?
Hear, Hear! Well said! I was tempted to say that the other author was simply a non-believer. You provide clarity of thought and a singular vision. Happy Holidays, Rich Smrcina Sytek Services, Inc. Milwaukee, WI [EMAIL PROTECTED] [EMAIL PROTECTED] Catch the WAVV! Stay for Requirements and the Free for All! Update your S/390 skills in 4 days for a very reasonable price. WAVV 2002 in Cincinnati (Fort Mitchell, KY). April 12-16, 2002 For details see http://www.wavv.org One nation, under God, indivisible, with liberty and justice for all. - Original Message - From: David Boyes [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Wednesday, December 19, 2001 11:07 AM Subject: Re: LCS drivers for 2.4.9 ? Nope, but I'm free to go buy some piece of hardware that isn't tied up with intellectual property issues. That's always your decision to make. If the objective is complete purity of essence, then we can sit around and theorize or select a different solution that meets the immediate needs. We can also try to do something constructive that convinces IBM to work with us. It's about the same amount of effort, and IMHO the idea of using some understanding of the way IBM's internal mindset works to effect change is ultimately more useful. In other words, is the mainframe really worth all of this? From the testing I've done, I'm not convinced... a cluster full of inexpensive Intel boxes seems faster and cheaper to me, and doesn't carry the baggage of hardware written to run an OS that is completely foreign to Linux (OS/390). For example, with a cluster of PCs, each machine can have *GASP!* a console that actually works with vi. If that solution is more effective for you, then you're free to use it at the costs of any choice. Like I tell the content control fascists who want to water down anything on the Internet to the level of pre-digested pre-composted baby food, You have the right to use an off switch and the ability to install filtering software on your PC to block anything that you don't want to see. Exercise that right. Me, I'm more interested in fixing the problem than complaining about it. This could be a temporary solution, until some bug in the binary bit of code cause it a) not to build properly or b) not to function properly. Then we are right back to bitching and IBM to fix it. Make up your mind: do you want a perfect world or working network drivers? If a perfect world, you're going to be waiting a while. If you want working network drivers, then let's get to it and quit whining about it. I'd like to ask the IBMers to see whether this would be possible. I'll work on getting backing, or arranging a resource of my own. I would applaude such an effort as a stop-gap measure, it's not the ideal situation though. See above. Whining doesn't fix the problem. Putting up resources does. Your call -- part of the problem, or part of the solution? -- db
Re: LCS drivers for 2.4.9 ?
Just so everyone is clear: We (IBM) do not like to resort to OCO, but in this world it is the only way to protect the intellectual property present in the drivers. If the drivers weren't OCO, anyone could step up to the challenge to provide support. But, when all the shouting is over, IBM or its delegate is the one who provides support for its OCO modules, not the Open Source community at large. This means that IBM has the right and the responsibility to determine what levels of the kernel it will support. I'm not affected by IBM's OCO policy, but other companies have similar policies. There has been some heated discussion about this on enigma-list recently; a certain video card vendor provides drivers (required for good #D performance I'm told) for Linux. Unfortunately the driver in question seems to trample on other folks memories, and the Red Hatters won't even think of helping with a kernel problem if the driver's been loaded. Certainly it's the vendor's right and responsibility to provide the support and fix the driver, but I'm sure a lot of other folk would be happy to step up and help out if only they had the source. Some of them would do it for free, just so the card works for them. It is obvious that many on this list have differing views about what the word support means. By that, I mean more than just doing some coding. I mean that IBM will fix something that isn't working. That promise is not trivial and actually costs IBM real dollars to provide. We have people that design, code, test, and document our drivers. When there's a new driver, there's more testing. That means tying up REAL resources (people and machines). When that happens, those resources aren't available to be used on other things, whether related to Linux or IBM's other business interests. It is not the IBM business model to release code to the world that may or may not work, and then wait for the fall-out. IBM is a big company and has many business models. Some areas do indeed release code that may or may not work and wait for the fallout. Sam the Spinning Devices Man did just that when developing support for removable media for OS/2. OCO. People with Jaz, Zip and similar drives were very happy to take the latest incarnations and report (mostly) their failures. Talk to Dr David Jikes Shields about releasing source. Or Steve JFS Best. Both have become famous (and I rather think) admired for releasing code to the world that may or may not work. You will find users mostly want something that works; many appreciate having the in-principle ability to fix it themselves if it's broken. -- Cheers John Summerfield Microsoft's most solid OS: http://www.geocities.com/rcwoolley/ Note: mail delivered to me is deemed to be intended for me, for my disposition.
Re: LCS drivers for 2.4.9 ?
I am with Randy, 5 minutes after it is released, we will have it installed. Nothing like the life of a propeller head!! Patrick E. Schlehuber Web Services COUNTRY Insurance and Financial Services 309-821-5299 wk 309-821-4818 fax [EMAIL PROTECTED] -Original Message- From: Shumate, Randolph W. (LNG) [mailto:[EMAIL PROTECTED]] Sent: Tuesday, December 18, 2001 10:00 AM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Can any of the IBM'ers on this list tell me when or if the lcs driver for 2.4.9 may be made available? I have a group of folks at my site chomping at the bit to install Red Hat Linux in an LPAR; I'd like to be able to set expectations appropriately. Randy -Original Message- From: Post, Mark K [mailto:[EMAIL PROTECTED]] Sent: Tuesday, December 18, 2001 10:01 AM To: [EMAIL PROTECTED] Subject: Re: LCS drivers for 2.4.9 ? Red Hat has stated that they will _not_ be including an OCO modules from IBM in their distribution(s). The only place you will be able to get them is from IBM's DeveloperWorks site whenever they become available. Mark Post
