hurley@watserv1.waterloo.edu (Patrick Hurley) (11/13/90)
Has anyone out there ever gotten aix to run on clones? What I experience was the following: The first attempt at getting it going was an attempt at installing it on an XYZ clone 386/20mhz with harddisk with an ide disk controller. No luck , it complained about not being able to find a harddisk. Figuring that I decided to try to install it on a 386 with a ESDI hard drive...like the Ps/2 . This was tried on a Packard Bell 386/33mhz ( I'm not sure about the wait state ) with a 120meg ESDI drive. The installation went further up to the point of where the installation procedure asked for the maintaince diskette. Then it bitched about not being able to find information about the hard disk. But the kernel came up ( So that was interesting ). So at the point I took the maintaince diskette and mounted its filesystem on another machine running AIX. I found using strings command that the error was coming from the mounted /etc/maint module. That is as far as I got. Now after all this ranting.....has anyone gotten any further? Is this the way IBM is copy protecting AIX? The way I figure it the more versions of AIX out there the more hardware IBM will sell in the long run if this is indeed the case.
mjones@fenway.aix.kingston.ibm.com (Mike Jones) (11/20/90)
In article <1990Nov13.150450.18758@watserv1.waterloo.edu>, hurley@watserv1.waterloo.edu (Patrick Hurley) writes: > > Has anyone out there ever gotten aix to run on clones? > [lots of stuff about trying to install AIX PS/2 on a clone deleted] > > Now after all this ranting.....has anyone gotten any further? > > Is this the way IBM is copy protecting AIX? The way I figure it > the more versions of AIX out there the more hardware IBM will sell in the > long run if this is indeed the case. Ummm...copy protecting? Not quite. IBM has never claimed that AIX will work on anything other than PS/2's, and support has only been announced on specific models of those. Most clones are clones only down to the BIOS level, and have various bits of hardware underneath, so a "clone" is actually a different machine to AIX. Would you be surprised if you tried to install it on, say, a Sun i386 and it didn't work? The fact is that AIX is optimized for the PS/2 hardware, not crippled to keep it from working on other hardware. My opinions, of course, are not IBM's official opinions. Mike Jones | In the long run, every program becomes rococo, and AIX Development | then rubble. Kingston, NY | -- Alan Perlis ibmps2!aix!mjones |
dyer@spdcc.COM (Steve Dyer) (11/25/90)
In article <1990Nov19.113340@fenway.aix.kingston.ibm.com> mjones@fenway.aix.kingston.ibm.com writes: >The fact is that AIX is >optimized for the PS/2 hardware, not crippled to keep it from working on >other hardware. Well, I'd agree except I'd say that "written for" and "only tested on" isn't exactly the same as "optimized for". I *HAVE* seen AIX PS/2 (or the very very close Locus equivalent) running on Compaq 386's out at Locus. However, such support is not in the product shipped by IBM. -- Steve Dyer dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer dyer@arktouros.mit.edu, dyer@hstbme.mit.edu
hh2@prism.gatech.EDU (HAAS) (11/26/90)
In article <4989@spdcc.SPDCC.COM> dyer@ursa-major.spdcc.com (Steve Dyer) writes: >I *HAVE* seen AIX PS/2 (or the very very close Locus equivalent) >running on Compaq 386's out at Locus. However, such support is >not in the product shipped by IBM. >-- >Steve Dyer >dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer >dyer@arktouros.mit.edu, dyer@hstbme.mit.edu I believe it was a man named Jack Vogel from Locus that claimed to have AIX running on many clones within Locus, the LAST time this subject came up. So we KNOW it's possible. I guess what all of us AIX folks that aren't 'Unix internals' guys want to know is: HOW? What changes are neccesary? Does it require simple modifications to the installation diskettes, (even if that might mean a binary patch) or do we simply not have the resources neccesary? IBM's AIX doesn't support any hard disk size EXCEPT for the sizes that they make. This is absolutely STUPID, but can be 'fixed'. Good grief, if IBM decides to sell another size disk, they'll have to release new installation diskettes. All because they want to make it as dificult as possible for anyone to use anything but (the small and expensive) IBM hard disks. So if anyone is willing to give us some clue as to what would be required to run a LICENSED COPY of AIX on ANY machine other than one manufactured by IBM, it would be greatly appreciated. Thanks. . . . hh -- Harry Haas GTRI/RIDL/DB "What makes it DO that!?" - Bones Georgia Institute of Technology, Atlanta Georgia, 30332 uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!hh2 Internet: hh2@prism.gatech.edu hhaas@{gtri01|rmadsun}.gatech.edu
dyer@spdcc.COM (Steve Dyer) (11/26/90)
In article <17516@hydra.gatech.EDU> hh2@prism.gatech.EDU (HAAS) writes: >So if anyone is willing to give us some clue as to what would be required >to run a LICENSED COPY of AIX on ANY machine other than one manufactured >by IBM, it would be greatly appreciated. I really doubt you could do this without AIX PS/2 source. It's just that the system distributed by IBM is completely customized for a microchannel PS/2 environment with certain-sized fixed disks attached to the embedded IBM controllers, and so on. I'd even be suspicious of trying this on a microchannel clone, not that there are many of them. What you're looking for is essentially what SCO and ISC have done to the original Intel/AT&T 386 UNIX distribution--take a distribution which might be limited to a small subset of controllers/video adaptors and add enough support that it will run on the vast majority of 386 systems constructed out of generic parts. While the work is "trivial", the support isn't fast, easy or cheap. I surmise that Locus, the source of AIX PS/2's primary technology, has not done this either because of a marketing agreement with IBM (plus, IBM contributed large parts of AIX PS/2), or, as likely, it simply isn't practical given the dominance of Sys V.3 and soon, Sys V.4 systems in the 386/486 marketplace. IBM can sell AIX PS/2 because it's taking advantage of a niche market. A small company like Locus probably has its hands full already and can't afford to be another SCO or ISC. I agree that it's a real shame that you can't run AIX PS/2 on anything other than an overpriced IBM machine, because it's really a quite creditable version of UNIX, with many of the features which people had been waiting for in SysV.4. It's also the only place you can currently buy Locus distributed system technology (unless you have a spare 370 around to run AIX 370). The TCF facility seems a bit "weird" compared to NFS or AFS until you get used to it, and then it's rather addictive. -- Steve Dyer dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer dyer@arktouros.mit.edu, dyer@hstbme.mit.edu
jackv@turnkey.tcc.com (Jack F. Vogel) (11/28/90)
In article <17516@hydra.gatech.EDU> hh2@prism.gatech.EDU (HAAS) writes: >I believe it was a man named Jack Vogel from Locus that claimed to have >AIX running on many clones within Locus, the LAST time this subject came >up. So we KNOW it's possible. I guess what all of us AIX folks that aren't >'Unix internals' guys want to know is: HOW? What changes are neccesary? >Does it require simple modifications to the installation diskettes, (even >if that might mean a binary patch) or do we simply not have the resources >neccesary? My such notoriety :-}!! Actually, I don't remember posting about this. However, as both Steve and Brian have noted in subsequent followups, we were until very recently running AIX (a pre-GA version I might note) not "on many clones", but on Compaqs. But neither Steve or Brian have gone into detail on what was required to do this; as a matter of fact, we had to build completely different kernels to run on these sites. So unfortunately its not just a matter of diddling with your installation disks or anything else that you might have the ability to fiddle with that will get it to work. Brian suggested that it would work on "clones" with VGA, actually I doubt this is all that would be required. I think it would have to have an identical NVRAM configuration as well which means ABIOS clone I believe. And, as he noted, even if you could get it to run it would be unsupported. Steve pondered on why Locus itself might not offer this option, well this is an IBM product whose development was contracted to Locus so we wouldn't have this option even if we wanted ( and, I might add, I make no claims as to what the marketing types might want :-}). We also quit using these systems and that code when AIX 1.2 GA'ed since there was no justification for the work required to keep that code in sync, and since our production environment is an IBM customer just like anyone else. So the days of "clones" running AIX at LCC are gone. In its development lifetime some distant relative of the current AIX 1.2 has run on many different type cpu's, however IBM has chosen the present limitation and you will be hard-pressed to get around it. Where this limitation might change is with the offering from the OSF, or at least one can hope. Disclaimer: Naturally, these opinions are mine, not LCC or IBM's. -- Jack F. Vogel jackv@locus.com AIX370 Technical Support - or - Locus Computing Corp. jackv@turnkey.TCC.COM
richard@locus.com (Richard M. Mathews) (11/28/90)
jackv@turnkey.tcc.com (Jack F. Vogel) writes: >But neither Steve or Brian have gone >into detail on what was required to do this; as a matter of fact, we had >to build completely different kernels to run on these sites. So unfortunately >its not just a matter of diddling with your installation disks or anything >else that you might have the ability to fiddle with that will get it to >work. We built different kernels, but did we have to? The old version didn't have minidisks. If you could get minidisks to work on the clone, you might be able to do this as an OCO change. Since, to my knowledge, this has never been done, it is pure speculation. >I think it would have to have an >identical NVRAM configuration as well which means ABIOS clone I believe. Don't we support at least one PS/2 model with no NVRAM? >Steve pondered on why Locus itself might not offer this option, well this >is an IBM product whose development was contracted to Locus so we wouldn't >have this option even if we wanted ( and, I might add, I make no claims >as to what the marketing types might want :-}). I think Locus is free to create LPPs, but I doubt it would be in Locus's best interest to upset IBM. There would certainly be a question in this case over whether we had made legitimate use of confidential information. I suspect there is also a problem with selling a product which can only be used by someone who is violating their license to use AIX on a specific machine. >Disclaimer: Naturally, these opinions are mine, not LCC or IBM's. You can say that again. In fact, I will: Disclaimer: Naturally, these opinions are mine, not LCC or IBM's. Richard M. Mathews D efend Locus Computing Corporation E stonian-Latvian-Lithuanian richard@locus.com I ndependence lcc!richard@seas.ucla.edu ...!{uunet|ucla-se|turnkey}!lcc!richard
dyer@spdcc.COM (Steve Dyer) (11/28/90)
In article <richard.659757178@fafnir.la.locus.com> richard@locus.com (Richard M. Mathews) writes: >I suspect there is also a problem with selling a product which can only be >used by someone who is violating their license to use AIX on a specific >machine. Huh? There's nothing in my license agreement which says my copy of AIX PS/2 has to be running on a PS/2. Obviously you would only run AIX PS/2 on as many machines as you had purchased licenses for, but that's just a standard licensing issue. Anyway, I doubt Locus could sell a "LPP" in the classical sense to get this to work on clones, but I suppose an "clone installation kit" would work: new boot/install disks with a kernel and utilities which knew about ISA-class machines and common device controllers. Plus a selected few new *.o files to go into /usr/sys/386/*.a. The new install disks would install AIX PS/2 as normal and finish by building a new kernel with the replacement drivers. Hey, what a great idea! When will you come out with it? :-) -- Steve Dyer dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer dyer@arktouros.mit.edu, dyer@hstbme.mit.edu
eliot@dodongo.engr.washington.edu (Eliot Lim) (11/29/90)
In article <5006@spdcc.SPDCC.COM> dyer@ursa-major.spdcc.com (Steve Dyer) writes: >The TCF facility seems a bit "weird" compared >to NFS or AFS until you get used to it, and then it's rather addictive. Some people (hard core unix types) here think that TCF is an outdated and laughable idea but I don't really agree. Is there any justification for this other than simple bigotry? (I don't expect Locus people to agree either :-)) Eliot
richard@locus.com (Richard M. Mathews) (11/29/90)
eliot@dodongo.engr.washington.edu (Eliot Lim) writes: >>The TCF facility seems a bit "weird" compared >>to NFS or AFS until you get used to it, and then it's rather addictive. >Some people (hard core unix types) here think that TCF is an outdated >and laughable idea but I don't really agree. Is there any >justification for this other than simple bigotry? I'd be really curious to find out what it is about TCF that some people find "outdated and laughable." If TCF is outdated, we should work on improving it. As a user of TCF for 6 years, I consider NFS and AFS to be outdated and laughable. NFS and AFS give you file system access; TCF gives you the whole machine. Perhaps what they mean is that the version of AIX which includes TCF is outdated because it doesn't include <fill in some feature from BSD4.3+, SVR4, or where ever>? If that's the case, all I can say is that IBM and LCC keep putting out updates -- if you want your favorite feature to get on the list, try convincing support to put in a DCR or whatever it's called. Disclaimer: I speak only for myself (if that), not for LCC, IBM, or anyone else. Richard M. Mathews Freedom for Lithuania Locus Computing Corporation Laisve! richard@locus.com lcc!richard@seas.ucla.edu ...!{uunet|ucla-se|turnkey}!lcc!richard
dls@dlsrt.austin.ibm.com (David Skeen) (11/30/90)
In article <richard.659843132@fafnir.la.locus.com>, richard@locus.com (Richard M. Mathews) writes: > I'd be really curious to find out what it is about TCF that some people > find "outdated and laughable." If TCF is outdated, we should work on > improving it. As a user of TCF for 6 years, I consider NFS and AFS to > be outdated and laughable. NFS and AFS give you file system access; TCF > gives you the whole machine. When OSF didn't accept the TCF parts of the DEcorum proposal, the future of TCF became somewhat clouded. I would expect TCF functionality to be added to OSF/DCE in the future, but when and how aren't understood. AFS is an integral part of OSF/DCE; there is some degree of interoperability between DCE and NFS. For the near term, that leaves TCF as odd man out. (This is not an official IBM position, but it is my opinion ...) Dave Skeen IBM Internet: dls@dlsrt.austin.ibm.com / IBM VNET: SKEEN at AUSTIN