Re: MSI broken in libata?

[Torsten Kaiser <just.for.lkml@googlemail.com>]

On Sat, Jan 9, 2010 at 10:11 AM, Tejun Heo <tj@kernel.org> wrote:
> On 12/25/2009 06:22 PM, Torsten Kaiser wrote:
>> As reported in http://lkml.org/lkml/2009/12/19/82 the new MSI support
>> for sata_sil24 does not work for me.
>> This is still the same with 2.6.33-rc2.
>>
>> Why I think, this might be a problem within libata:
>>  * other drivers can use MSI successful on my system (tg3, radeon, hda-intel)
>>  * happens both in sata_sil24 and sata_nv
>>  * the count in /proc/interrupts increases for the MSIs assigned to
>> sata_sil24/sata_nv, so interrupt delivery seems to work
>>  * only writing seems to fail
>
> How does it fail?  Timeouts?

Yes, timeouts.
I posted the error messages in http://lkml.org/lkml/2009/12/19/82 and
http://lkml.org/lkml/2010/1/6/60

> Also, ahci enables MSI by default if
> available and works fine on many configurations so I don't think
> anything in libata core layer is broken regarding MSI (there just
> isn't anything which can break).

The system I'm using does not have a ahci compatible controller, so I
could not compare this.
(And my other system that is using ahci, does not use MSI for that)

I just found it suspicious that 3 other drivers (tg3, hda-intel and
radeon) can use MSI, but both of the libata drivers (sata_sil24 and
sata_nv) fail in a similar way.

I did try the patch from Robert Hancock in
http://lkml.org/lkml/2010/1/6/417 ,but without success.

if you need any more information, or have something for me to try,
please just ask. I did look at the code and the documentation about
enabling MSI, but did not see anything (obvious) wrong, so I don't
know what to try next.

Torsten