[PATCH] drm/scheduler: fix last_scheduled handling

[Christian König]

Am 08.08.2018 um 12:55 schrieb Nayan Deshmukh:
> Hi Christian,
>
> On Wed, Aug 8, 2018 at 4:20 PM Christian König 
> <ckoenig.leichtzumerken at gmail.com 
> <mailto:ckoenig.leichtzumerken at gmail.com>> wrote:
>
>     Ping, Nayan any comments on that or can I commit it?
>
>     This is just a stripped down version of my original last_scheduled
>     improvement patch.
>
> I had missed the mail yesterday. I was just looking at the code right 
> now. The code looks fine but I wanted to run it once before giving it 
> a Rb.
>
> I was wondering why you stripped it down. Was there some problem with 
> the previous one?

No, not really. But this is a bug fix and bug fixes should be as simple 
as possible.

Improvements should come in separate patches. Didn't thought initially 
that this is actually a fix.

BTW: Another small fix is required, looks like we need the scheduler 
correctly set for debugging even earlier.

Christian.

>
> Cheers,
> Nayan
>
>     Christian.
>
>     Am 07.08.2018 um 14:54 schrieb Christian König:
>     > Make sure we access last_scheduled only after checking that
>     there are no
>     > more jobs on the entity.
>     >
>     > Signed-off-by: Christian König <christian.koenig at amd.com
>     <mailto:christian.koenig at amd.com>>
>     > ---
>     >   drivers/gpu/drm/scheduler/gpu_scheduler.c | 21
>     +++++++++++----------
>     >   1 file changed, 11 insertions(+), 10 deletions(-)
>     >
>     > diff --git a/drivers/gpu/drm/scheduler/gpu_scheduler.c
>     b/drivers/gpu/drm/scheduler/gpu_scheduler.c
>     > index 8ee249886473..bd7883d1b964 100644
>     > --- a/drivers/gpu/drm/scheduler/gpu_scheduler.c
>     > +++ b/drivers/gpu/drm/scheduler/gpu_scheduler.c
>     > @@ -568,19 +568,20 @@ void drm_sched_entity_push_job(struct
>     drm_sched_job *sched_job,
>     >                              struct drm_sched_entity *entity)
>     >   {
>     >       struct drm_sched_rq *rq = entity->rq;
>     > -     bool first, reschedule, idle;
>     > +     bool first;
>     >
>     > -     idle = entity->last_scheduled == NULL ||
>     > -  dma_fence_is_signaled(entity->last_scheduled);
>     >       first = spsc_queue_count(&entity->job_queue) == 0;
>     > -     reschedule = idle && first && (entity->num_rq_list > 1);
>     > +     if (first && (entity->num_rq_list > 1)) {
>     > +             struct dma_fence *fence;
>     >
>     > -     if (reschedule) {
>     > -             rq = drm_sched_entity_get_free_sched(entity);
>     > -             spin_lock(&entity->rq_lock);
>     > -  drm_sched_rq_remove_entity(entity->rq, entity);
>     > -             entity->rq = rq;
>     > -             spin_unlock(&entity->rq_lock);
>     > +             fence = READ_ONCE(entity->last_scheduled);
>     > +             if (fence == NULL || dma_fence_is_signaled(fence)) {
>     > +                     rq = drm_sched_entity_get_free_sched(entity);
>     > +  spin_lock(&entity->rq_lock);
>     > +  drm_sched_rq_remove_entity(entity->rq, entity);
>     > +                     entity->rq = rq;
>     > +  spin_unlock(&entity->rq_lock);
>     > +             }
>     >       }
>     >
>     >       sched_job->sched = entity->rq->sched;
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/dri-devel/attachments/20180808/e3752255/attachment.html>