[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: PCI scanning IRQ fix.
From: |
Marco Gerards |
Subject: |
Re: PCI scanning IRQ fix. |
Date: |
Sun, 29 Jan 2006 20:01:16 +0100 |
User-agent: |
Gnus/5.1007 (Gnus v5.10.7) Emacs/21.4 (gnu/linux) |
Gianluca Guida <glguida@gmail.com> writes:
Hi Gianluca,
> My laptop's motherboard set its RTL8139 NIC IRQ at a value that is not
> included in the range 0<=x<16 and it's not 255.
That is weird. I always thought that for compatibility reasons IRQs
are always <16. In case the OS supports more than 16 IRQs, the OS can
change the IRQs to prevent shared IRQs. With your NIC this is not the
case, is this a BIOS bug or a setting in the BIOS you can change?
Ignoring this device until there is a better solution is certainly the
best thing to do, as I see it.
> GNU Mach can't handle that, and when I try a 'devprobe eth0' I get a
> panic due to a failed assert (irq < 16). This patch tell GNU Mach to
> ignore this kind of IRQ.
>
> The real solution for this problem would be an PCI IRQ Router, which
> is what I am going to port/implement in Mach soon. For now, this patch
> helps by both removing the failed assert and by noticing the user that
> the device is not being used for unassigned IRQ.
Nice!
> I guess that for such a small path using Savannah's Patch system is a
> waste, right?
You can put it there if it is not applied soon.
--
Marco