Okay, this may be a bit off-topic, but mayber there are other lurkers
or those knowledgeable about the EISA NVRAM that could give me a hand.

I have a Micronics M54Pe with two P133s.  It's a fine little (well, not
little--silly thing is the size of home plate) board that does light
fileserving for my home network.  But with the classic pre-APIC machines,
the IO-APIC on this board only supports 16 sources, which, coincidentally,
are mapped to the 16 "normal" IRQs.  So I only have that many interupt
sources.

I've got the onboard IDE enabled right now (which use 14 and 15), a PCI
Intel EtherExpress Pro/100 (which uses IRQ 9, I think), and two EISA
SCSI cards (10 and 11).  I would like to add another SCSI card (don't
ask), but it causes IRQ shortage issues, and wreaks havoc with the 
EISA BIOS.

First of all, I'm not sure how important it is to go into the EISA
config program and set up all the cards with IRQs and such, but it
seemed to be a good idea.  With three SCSI cards in there, they take
up 9, 10, and 11, and force the ethernet card off the chart.  Linux,
understandably, loads the module with IRQ 0 (!) which doesn't actually
work.

Well, here's the thing.  It has integrated PS/2 mouse which I'll
never use, so I wondered if it was possible to disable.  Regular
Phoenix BIOS has no such option, I didn't find a jumper, and the
EISA CFG file has a comment: "Resources always in use" which lists
the keyboard interrupt, timer, FPU, and other regulars, including
PS/2 mouse.  Is it possible to hack the EISA CFG file to let it
loose of the interrupt, or is it hardwired, and the board will 
never give it up?

Finally, are their other solutions?  I've never liked sharing 
interrupts especially with dissimilar cards, but is that an option?
Maybe someone has this board and has some great ideas.  Also, how
come if I set things to be LEVEL triggered in the EISA config,
they always wind up IO-APIC-edge in /proc/interrupts?

Brendan Miller




-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/dmentre/smp-howto/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]

Reply via email to