That's ... really evil.
> for some background. Under linux, on an ALi chipset, the "link" numbers
> are used as an offset into the PCI config space of the ISA bridge, where
> the IRQ for that "link" is stored. On my machine, the link numbers are
> 0x01-0x03 (for everything but USB) and 0x59 (for USB). The value at the
> offset for link 0x59 translates to IRQ 9. The PCI configuration space of
> the USB controller indicates IRQ 9, as well. See pirq_ali_get() in
So their BIOS wrongly set irq in config space of USB controller, right?
> linux/arch/i386/kernel/pci-irq.c for details on how this works.
>
> All the last patch does is match the IRQ being considered for the device
> against the IRQ mask for that device in the PIR table. If it doesn't
> match, the kernel assigns one that does match the mask.
Ahha. May that mean that our magic w.r.t. touching pci config space on
ALI chipsets is wrong?
> To be clear: with the last patch, USB works, but not the maestro-3,
> right?
Yes.
> The reason I keep asking you for the output of "lspci -vvvxxx" and
> "dump_pirq" is so I can look at your PIR table and PCI config space and
> try to determine if the same thing that happened to USB is happening to
> your maestro. It's possible your maestro problem is completely
> unrelated. If you're unwilling to provide that informataion for some
> reason, just let me know and I'll quit asking.
I thought I already mailed you lspci.... dump_pirq is not installed on
my machine, I'll try to install it and mail you that info. [Hmm, it
may be hard, because I'm now on modem link and behind nasty firewall;
perhaps you could just mail me dump_pirq binary if it does not need
special libraries?]
Pavel
-- Casualities in World Trade Center: 6453 dead inside the building, cryptography in U.S.A. and free speech in Czech Republic. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/