It would be ashame to deprive people of IO-APIC..
It would be nice to have the kernel detect differnt boards and have a
mapping table (It wouldn't take up much memory).. There should also be a
diasable boot param (incase of a defective/falsely reporting board), and a
manual set (for yet to be supported boards)..
If it was a userspace solution, nobody would ever use it..
On Thu, 22 Jan 1998, Peter T. Waltenberg wrote:
> That certainly fits with what I saw, I finally got my scsi driver
> to work by using the bios to lock all but one of the unused IRQ's
> as ISA only. I still havn't worked out what the mapping is , but
> with all slots forced to use the same IRQ that really doesn't
> matter any more.
>
> I suspect you'll have to have a table of motherboards to sort out
> the mapping, but again, with some doc's, and the ability to tell
> the kernel NOT to use the IO-APIC code as a boot parameter, or better
> enabled only via a boot parameter, it looks quite safe.
>
> Maybe a table of KNOWN motherboards ?, with a manual setting in
> the configure scripts for people who have an "unknown" board,
> but who are willing to sort out the mapping.
> If neither is set, turn off the IO-APIC code.
>
> Just a thought, the basic APIC code seems amazing stable for its
> first real test. The real problem of working out the slot->apic
> mapping seems unlikely to just go away of it's own accord.
>
> Peter
>