First, my belated congratulations to Ron for his multiuser boot. I've been lurking on the openBIOS list for a bit. It's interesting read, but I've had nothing to contribute, other than vague encouragement.
Sychron is developing software for clusters (currently on Linux). Direct control of cluster hosts without commercial M$ obsessed BIOSs getting in the way would be extremely valuable to us. I'd *really* like a serial-line-console only boot; no kb or VGA (goodbye KVM switch). Currently, we only need support for serial lines, 905B/C NICs and IDE disks.
Be gentle with me if the following is a totally dumb idea. IANABH *
Our software has a driver for 3com 905B/C NICs; like many NICs, BIOS ROMs can be fitted for diskless network boots (this is of some interest, but has always seemd a bit too much effort). But, on reading (for completely unrelated reasons) the 3Com data book, I see that these NICs have the facility for drivers to *write* to a 64K or 128K Atmel PEROM (AT29C512 or AT29C010).
So I think to myself... is it possible / practical to write openbios / linuxbios into a NIC PEROM and then boot from it, overriding the motherboard BIOS? If so, the advantages seem great: programming is independent of (and non-destructive to) the motherboard and its BIOS; if it all goes wrong, all I do is blank the PEROM and my system is no longer a brick. :-)
We have Linux driver code to initialise and control 905B/C NICs, so we could probably contribute a software Linux 3Com 905B/C PEROM "blower" quite easily. Other NICs may be able to do this, but I only have the 3Com book. We don't have linuxbios supported motherboards here, but if this seems likely to work, I can make a case for buying SiS boards. BTW, if anyone from SiS is listening, Sychron may soon be purchasing (or influencing the purchase of) some large clusters, so if this works, your co-operation may pay back quite well.
* I Am Not A BIOS Hacker.
Great idea,
Openbios could keep the existing standards for extension bioses so network cards bootroms stay working, and someone else may write the code for a linux kernel to be put in extension bios in order to be loaded and booted by from there, so you can have a non openbios motherboards booting linux kernel from some extension bios these two projects seems to me not the same thing but only related to the same thing : booting.
Correct me if I am wrong.
I apologies for my poor english, I'm french (yes I eat frogs...) -- legoll@online.fr
- To unsubscribe: send mail to majordomo@freiburg.linux.de with 'unsubscribe openbios' in the body of the message
Peter,
An interesting idea but I don't think it's possible. On RESET the CPU will begin executing instructions from a fixed address of F000:FFF0h (being x86 compatible). At this point the PCI bus structure has not been initialized and the OPROM on the PCI NIC has not been enabled. Thus, you can't replace the BIOS this way. If you could somehow modify the card such that it comes up enabled and decoding it's ROM in the F000h segment, and assuming it's on PCI bus 0, you might be able to do such a thing since the PCI card should claim the memory accesses rather than letting them filter down to the ISA bus where the ROM exists.
Dave
-----Original Message----- From: owner-openbios@elvis.informatik.uni-freiburg.de [mailto:owner-openbios@elvis.informatik.uni-freiburg.de]On Behalf Of Peter Lister Sent: Wednesday, May 17, 2000 2:07 PM To: openbios@elvis.informatik.uni-freiburg.de Cc: P.Lister@sychron.com Subject: [OpenBIOS] OpenBIOS/LinuxBIOS on a NIC ROM? Or am I mad?
First, my belated congratulations to Ron for his multiuser boot. I've been lurking on the openBIOS list for a bit. It's interesting read, but I've had nothing to contribute, other than vague encouragement.
Sychron is developing software for clusters (currently on Linux). Direct control of cluster hosts without commercial M$ obsessed BIOSs getting in the way would be extremely valuable to us. I'd *really* like a serial-line-console only boot; no kb or VGA (goodbye KVM switch). Currently, we only need support for serial lines, 905B/C NICs and IDE disks.
Be gentle with me if the following is a totally dumb idea. IANABH *
Our software has a driver for 3com 905B/C NICs; like many NICs, BIOS ROMs can be fitted for diskless network boots (this is of some interest, but has always seemd a bit too much effort). But, on reading (for completely unrelated reasons) the 3Com data book, I see that these NICs have the facility for drivers to *write* to a 64K or 128K Atmel PEROM (AT29C512 or AT29C010).
So I think to myself... is it possible / practical to write openbios / linuxbios into a NIC PEROM and then boot from it, overriding the motherboard BIOS? If so, the advantages seem great: programming is independent of (and non-destructive to) the motherboard and its BIOS; if it all goes wrong, all I do is blank the PEROM and my system is no longer a brick. :-)
We have Linux driver code to initialise and control 905B/C NICs, so we could probably contribute a software Linux 3Com 905B/C PEROM "blower" quite easily. Other NICs may be able to do this, but I only have the 3Com book. We don't have linuxbios supported motherboards here, but if this seems likely to work, I can make a case for buying SiS boards. BTW, if anyone from SiS is listening, Sychron may soon be purchasing (or influencing the purchase of) some large clusters, so if this works, your co-operation may pay back quite well.
* I Am Not A BIOS Hacker. -- Peter Lister P.Lister@sychron.com PGP (RSA): 0xE4D85541 Sychron Ltd http://www.sychron.com PGP (DSS): 0xBC1D7258 1 Cambridge Terrace Voice: +44 1865 200211 Oxford OX1 1UR UK FAX: +44 1865 249666
- To unsubscribe: send mail to majordomo@freiburg.linux.de with 'unsubscribe openbios' in the body of the message
- To unsubscribe: send mail to majordomo@freiburg.linux.de with 'unsubscribe openbios' in the body of the message
It's a neat idea but 128K is too small. Also, the bios is still doing lots of things you don't want it to: zeroing memory (you lose logbuf and if you have to reset you lose the system state); misconfiguring PCI; wasting time configuring things for DOS that you don't need; and so on. I think we need to bite the bullet and replace the steeenking bios.
ron
- To unsubscribe: send mail to majordomo@freiburg.linux.de with 'unsubscribe openbios' in the body of the message
-
David Christensen -
Peter Lister -
Ronald G Minnich -
Vincent Legoll