Patch "drm/sched: Always wake up correct scheduler in drm_sched_entity_push_job" has been added to the 6.10-stable tree
Christian König
christian.koenig at amd.com
Mon Oct 7 18:23:54 UTC 2024
Am 07.10.24 um 19:59 schrieb Greg KH:
> On Mon, Oct 07, 2024 at 07:53:26PM +0200, Christian König wrote:
>> Hi Greg,
>>
>> please drop this patch from all backports. It turned out to be broken and
>> the old handling has been restored by a revert.
>>
>> Sorry for the noise. The revert should show up in Linus tree by the end of
>> the week.
> What is the revert's git id? I'd rather take that so we don't have any
> "why wasn't this patch applied" stuff later on.
Good point. I was also not 100% sure which ones of Tvrtko's patches we
had to revert.
Here is the git Id from the drm-misc-fixes branch:
commit abf201f6ce14c4ceeccde5471bdf59614b83a3d8
Author: Christian König <christian.koenig at amd.com>
Date: Mon Sep 30 15:07:49 2024 +0200
drm/sched: revert "Always increment correct scheduler score"
This reverts commit 087913e0ba2b3b9d7ccbafb2acf5dab9e35ae1d5.
It turned out that the original code was correct since the rq can only
change when there is no armed job for an entity.
This change here broke the logic since we only incremented the counter
for the first job, so revert it.
Signed-off-by: Christian König <christian.koenig at amd.com>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin at igalia.com>
Link:
https://patchwork.freedesktop.org/patch/msgid/20240930131451.536150-1-christian.koenig@amd.com
Sorry for the noise, we should have catched that during the review and
put the stable tag on the revert as well.
Regards,
Christian.
>
> thanks,
>
> greg k-h
More information about the dri-devel
mailing list