Re: irq 15: nobody cared! with KT600 chipset and 2.6.0-test9

[Andi Kleen <ak@muc.de>]

Linus Torvalds <torvalds@osdl.org> writes:

> On Sun, 23 Nov 2003, Marco d'Itri wrote:
>>
>> This is an ASUS A7V600 motherboard, and the second IDE channel does not
>> work at all. I verified that it works fine with a 2.4.x kernel.
>
> Can you enable DEBUG in "arch/i386/pci/pci.h" and ACPI debugging and then
> report what the system says at bootup both with ACPI enabled, and with 
> "pci=noacpi".
>
> For some reason the system decides that your IDE controller is on irq3,
> which is clearly wrong. It's on irq15, but somebody told the PCI layer
> otherwise.

It's a long standing bug in how we handle VIA on board devices in ACPI.
It was a big problem on x86-64 too until I cheated and used only
PIC mode when there is a VIA southbridge.

Basically VIA has very specific rules on what interrupts can be assigned
to the onboard devices and they use an unusual ACPI construct for that

See http://bugme.osdl.org/show_bug.cgi?id=1530 for some more details.
(the same problem applies to 2.6 afaik) 

Here's also some older description from VIA about the issues I saved:

>>
   1) The _SRS method can only cope with IRQs <= 15 if SB is before VT8233.
            After VT8235, _SRS method need not report IRQ for internal devices due to they are
+hardwired to fixed IRQ in APIC mode.

[AK: that is one problem with the ACPI code - it doesn't handle that the 
_SRS report doesn't report it. Even ignoring the problems with
older SBs which would need some special handling code]


           The BIOS returns fixed IRQs in the _PRS method when system is in APIC mode if SB is after
+VT8235.
           (IDE-IRQ20, USB-IRQ21, Audio and Modem-IRQ22, LAN-IRQ23)

        2) The _CRS method is hardcoded to return 0 ALWAYS for link devices in APIC mode.
            We feel strange for this situation. This control method will check internal device's
+interrupt line (Rx3C of PCI configuration) to return current IRQ setting.
            Will it be possible in Linux kernel driver which will reset the Rx3C to zero ?
<<

Forcing PIC mode will at least make it work. Of course it would be 
better to fix the ACPI code to handle all this properly.

-Andi