regarding VIA quirk fix

regarding VIA quirk fix

[Tejun Heo]

Hello, Alan.

We've been getting bug reports from sata_via users for quite sometime 
now.  The first IRQ driven command (IDENTIFY) times out and thus device 
detection fails.  The following patch seems to fix it for many users.

http://marc.theaimsgroup.com/?l=linux-kernel&m=116300291505638

But, not for all.

http://bugzilla.kernel.org/show_bug.cgi?id=7415

Any ideas how to proceed on this bug?

Thanks.

-- 
tejun

Re: regarding VIA quirk fix

[Alan]

On Fri, 17 Nov 2006 19:13:24 +0900
Tejun Heo <htejun@gmail.com> wrote:

> http://bugzilla.kernel.org/show_bug.cgi?id=7415
> 
> Any ideas how to proceed on this bug?

The report appears to be about the earlier patch not the one I did from
reading it. That said I don't see it matters which. 

The only way to track down any remaining ones is to go line by line
through the PCI configuration and understand what we or the bios mixed up
so its pretty tedious but doable assuming the bug isn't in the SATA
driver in the first place. The IRQ routing on the later chips is pretty
rigid so its not too hard to spot a misconfiguration.

Alan

Re: regarding VIA quirk fix

[Daniel Drake]

Tejun Heo wrote:
> Hello, Alan.
> 
> We've been getting bug reports from sata_via users for quite sometime 
> now.  The first IRQ driven command (IDENTIFY) times out and thus device 
> detection fails.  The following patch seems to fix it for many users.
> 
> http://marc.theaimsgroup.com/?l=linux-kernel&m=116300291505638
> 
> But, not for all.
> 
> http://bugzilla.kernel.org/show_bug.cgi?id=7415
> 
> Any ideas how to proceed on this bug?

I'm not certain, but I think that this is an unrelated issue. Both the 
working kernel and the failing kernels quirk the device in the same way:

PCI: VIA IRQ fixup for 0000:00:0f.0, from 11 to 2

Both the working kernel (2.6.17-gentoo-r8) and the first failing kernel 
(2.6.18-gentoo) have the same patch for 'fixing' the quirk issues: 
Chris's changes were reverted, the VIA IRQ quirk was back in the 2.6.16 
state. Since then the same issues have been observed with unpatched 
2.6.19-rc.

This is not definitive but suggests the issue is elsewhere.

Daniel