Re: [BUG] radeon: suspend resume

Re: [BUG] radeon: suspend resume

[Michel Dänzer]

[ The radeon driver is discussed on the dri-devel mailing list, moving
there ]

On Son, 2013-07-14 at 13:26 +0100, Sami Kerola wrote:
> 
> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
> CP stall for more than 10000msec
> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
> (waiting for 0x0000000000000004 last fence id 0x0000000000000002)
> Jul 14 12:51:31 kerolasa-home kernel: [drm:r600_uvd_ib_test] *ERROR*
> radeon: fence wait failed (-35).
> Jul 14 12:51:31 kerolasa-home kernel: [drm:radeon_ib_ring_tests]
> *ERROR* radeon: failed testing IB on ring 5 (-35).

Looks like the patch from
https://bugs.freedesktop.org/show_bug.cgi?id=66425 might help.


-- 
Earthling Michel Dänzer           |                   http://www.amd.com
Libre software enthusiast         |          Debian, X and DRI developer
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [BUG] radeon: suspend resume

[Sami Kerola]

On 15 July 2013 11:35, Michel Dänzer <michel@daenzer.net> wrote:
> On Son, 2013-07-14 at 13:26 +0100, Sami Kerola wrote:
>>
>> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
>> CP stall for more than 10000msec
>> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
>> (waiting for 0x0000000000000004 last fence id 0x0000000000000002)
>> Jul 14 12:51:31 kerolasa-home kernel: [drm:r600_uvd_ib_test] *ERROR*
>> radeon: fence wait failed (-35).
>> Jul 14 12:51:31 kerolasa-home kernel: [drm:radeon_ib_ring_tests]
>> *ERROR* radeon: failed testing IB on ring 5 (-35).
>
> Looks like the patch from
> https://bugs.freedesktop.org/show_bug.cgi?id=66425 might help.

Hi Michel,

I tried this patch

https://bugs.freedesktop.org/show_bug.cgi?id=66425#c21

which nearly completely applied. The non-applied line is below, and
which I left that way as it seems there are referrals to the uvd_fw
firmware elsewhere in code.

diff a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
(rejected hunks)
@@ -1683,7 +1685,6 @@ struct radeon_device {
        const struct firmware *rlc_fw;  /* r6/700 RLC firmware */
        const struct firmware *mc_fw;   /* NI MC firmware */
        const struct firmware *ce_fw;   /* SI CE firmware */
-       const struct firmware *uvd_fw;  /* UVD firmware */
        struct r600_blit r600_blit;
        struct r600_vram_scratch vram_scratch;
        int msi_enabled; /* msi enabled */

When trying the patch made things worse. Errors started to occur at
suspend, which took a while and reported some errors. After waiting
for ~20 seconds my screen had pixel mush in all possible colours, and
the system was unresponsive. Please see the file from URL for all
messages I found from log.

http://www.iki.fi/kerolasa/radeon-suspend-3.10.0-09289-g9903883.gz

--
Sami Kerola
http://www.iki.fi/kerolasa/

Re: [BUG] radeon: suspend resume

[Michel Dänzer]

On Mit, 2013-07-17 at 10:58 +0100, Sami Kerola wrote:
> On 15 July 2013 11:35, Michel Dänzer <michel@daenzer.net> wrote:
> > On Son, 2013-07-14 at 13:26 +0100, Sami Kerola wrote:
> >>
> >> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
> >> CP stall for more than 10000msec
> >> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
> >> (waiting for 0x0000000000000004 last fence id 0x0000000000000002)
> >> Jul 14 12:51:31 kerolasa-home kernel: [drm:r600_uvd_ib_test] *ERROR*
> >> radeon: fence wait failed (-35).
> >> Jul 14 12:51:31 kerolasa-home kernel: [drm:radeon_ib_ring_tests]
> >> *ERROR* radeon: failed testing IB on ring 5 (-35).
> >
> > Looks like the patch from
> > https://bugs.freedesktop.org/show_bug.cgi?id=66425 might help.
> 
> Hi Michel,
> 
> I tried this patch
> 
> https://bugs.freedesktop.org/show_bug.cgi?id=66425#c21
> 
> which nearly completely applied. The non-applied line is below, and
> which I left that way as it seems there are referrals to the uvd_fw
> firmware elsewhere in code.

I'm afraid that sounds like you didn't port the patch to your kernel
correctly, as it's supposed to remove all references to uvd_fw as well.
Can you post the patch you're testing against your kernel?


-- 
Earthling Michel Dänzer           |                   http://www.amd.com
Libre software enthusiast         |          Debian, X and DRI developer
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [BUG] radeon: suspend resume

[Sami Kerola]

On 17 July 2013 11:19, Michel Dänzer <michel@daenzer.net> wrote:
> On Mit, 2013-07-17 at 10:58 +0100, Sami Kerola wrote:
>> On 15 July 2013 11:35, Michel Dänzer <michel@daenzer.net> wrote:
>> > On Son, 2013-07-14 at 13:26 +0100, Sami Kerola wrote:
>> >>
>> >> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
>> >> CP stall for more than 10000msec
>> >> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
>> >> (waiting for 0x0000000000000004 last fence id 0x0000000000000002)
>> >> Jul 14 12:51:31 kerolasa-home kernel: [drm:r600_uvd_ib_test] *ERROR*
>> >> radeon: fence wait failed (-35).
>> >> Jul 14 12:51:31 kerolasa-home kernel: [drm:radeon_ib_ring_tests]
>> >> *ERROR* radeon: failed testing IB on ring 5 (-35).
>> >
>> > Looks like the patch from
>> > https://bugs.freedesktop.org/show_bug.cgi?id=66425 might help.
>>
>> Hi Michel,
>>
>> I tried this patch
>>
>> https://bugs.freedesktop.org/show_bug.cgi?id=66425#c21
>>
>> which nearly completely applied. The non-applied line is below, and
>> which I left that way as it seems there are referrals to the uvd_fw
>> firmware elsewhere in code.
>
> I'm afraid that sounds like you didn't port the patch to your kernel
> correctly, as it's supposed to remove all references to uvd_fw as well.
> Can you post the patch you're testing against your kernel?

Hi Michel, and others,

I believe the patch from the bug id=66425, mentioned earlier in the
thread, seems to be part of upstream

http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=9cc2e0e9f13315559c85c9f99f141e420967c955

and unsurprisingly the version of the change in Linus's git compiles
without issues. Unfortunately the 3.11-rc2 does not fix my suspend
problem/resume problem. Let me know if I can some how help you to
debug this issue.

--
Sami Kerola
http://www.iki.fi/kerolasa/

Re: [BUG] radeon: suspend resume

[Alex Deucher]

On Tue, Jul 23, 2013 at 2:59 PM, Sami Kerola <kerolasa@iki.fi> wrote:
> On 17 July 2013 11:19, Michel Dänzer <michel@daenzer.net> wrote:
>> On Mit, 2013-07-17 at 10:58 +0100, Sami Kerola wrote:
>>> On 15 July 2013 11:35, Michel Dänzer <michel@daenzer.net> wrote:
>>> > On Son, 2013-07-14 at 13:26 +0100, Sami Kerola wrote:
>>> >>
>>> >> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
>>> >> CP stall for more than 10000msec
>>> >> Jul 14 12:51:31 kerolasa-home kernel: radeon 0000:00:01.0: GPU lockup
>>> >> (waiting for 0x0000000000000004 last fence id 0x0000000000000002)
>>> >> Jul 14 12:51:31 kerolasa-home kernel: [drm:r600_uvd_ib_test] *ERROR*
>>> >> radeon: fence wait failed (-35).
>>> >> Jul 14 12:51:31 kerolasa-home kernel: [drm:radeon_ib_ring_tests]
>>> >> *ERROR* radeon: failed testing IB on ring 5 (-35).
>>> >
>>> > Looks like the patch from
>>> > https://bugs.freedesktop.org/show_bug.cgi?id=66425 might help.
>>>
>>> Hi Michel,
>>>
>>> I tried this patch
>>>
>>> https://bugs.freedesktop.org/show_bug.cgi?id=66425#c21
>>>
>>> which nearly completely applied. The non-applied line is below, and
>>> which I left that way as it seems there are referrals to the uvd_fw
>>> firmware elsewhere in code.
>>
>> I'm afraid that sounds like you didn't port the patch to your kernel
>> correctly, as it's supposed to remove all references to uvd_fw as well.
>> Can you post the patch you're testing against your kernel?
>
> Hi Michel, and others,
>
> I believe the patch from the bug id=66425, mentioned earlier in the
> thread, seems to be part of upstream
>
> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=9cc2e0e9f13315559c85c9f99f141e420967c955
>
> and unsurprisingly the version of the change in Linus's git compiles
> without issues. Unfortunately the 3.11-rc2 does not fix my suspend
> problem/resume problem. Let me know if I can some how help you to
> debug this issue.

Probably best to file a bug:
https://bugs.freedesktop.org

Alex