Skip to contentSkip to conversation

Using ACPI to get PCI routing info?

DISCUSSION Randy.Dunlap 2 participants 2 messages Jul 13, 2001 – Jul 14, 2001

lore
Share & tools
lore
#0 lore raw

Petr-

Where does someone find/get BIOS version F5a for the 6VXD7? The latest that I see on the Gigabyte web pages is F5.

Thanks,
~Randy

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Hi,
Gigabyte (and/or AMI and/or VIA) decided that it is not worth
of effort to create full mptable and since version F5a for 6VXD7 they do not report PCI interrupts as 16-19, but only as traditional 0-15 (and they do not report them as conforms/conforms, but as active-lo/level).

For now I hardwired correct routing table into my kernel, as
I have other uses for IRQ < 16, but after some investigation I found that ACPI _SB_.PCI0._PRT element returns correct routing table (using IRQ 16-19). So my question is, are there any plans to use ACPI tables to get IRQ routing tables, or should I complain to Gigabyte that I'm not satisfied (I'll complain anyway, but...)?

Thanks,
Petr Vandrovec [email protected]

P.S.: No, there is no MPS1.1/MPS1.4 switch in BIOS (anymore) :-( And no, there is no way to disable ACPI in that BIOS :-((

/* Old working BIOS... 6VXD7 - F2 */
...
/* 'Broken' 6VXD7 - F5a */
...

#1 lore raw
Quoted 4 lines · Randy.Dunlap

Petr-

Where does someone find/get BIOS version F5a for the 6VXD7? The latest that I see on the Gigabyte web pages is F5.

It might be that F5a is older than F5, as on my filesystem F5a has date Feb8, while F5 has Mar23 ... But it does not matter, as both report same broken table, at least for me. So I have kernel patched with self written IRQ routing table instead of with BIOS's one, and machine is happy using IRQ16-19. This implies that I did not coded retrieving IRQ routing info from ACPI. With current discussions about moving ACPI to userspace it might be even impossible, as you for sure want console before you enter userspace, but you cannot have console if you did not enumerate devices (video card) yet...
Best regards,
Petr Vandrovec [email protected]