From: Pixel Ding Date: Wed, 18 Apr 2018 08:37:40 +0000 (-0400) Subject: drm/scheduler: move last_sched fence updating prior to job popping (v2) X-Git-Tag: v4.19~774^2~16^2~144 X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=b5b4ea4d98b42f94442e5d46d5942f392ed8af56;p=platform%2Fkernel%2Flinux-rpi.git drm/scheduler: move last_sched fence updating prior to job popping (v2) Make sure main thread won't update last_sched fence when entity is cleanup. Fix a racing issue which is caused by putting last_sched fence twice. Running vulkaninfo in tight loop can produce this issue as seeing wild fence pointer. v2: squash in build fix (Christian) Signed-off-by: Pixel Ding Reviewed-by: Christian König Reviewed-by: Monk Liu Signed-off-by: Alex Deucher --- diff --git a/drivers/gpu/drm/scheduler/gpu_scheduler.c b/drivers/gpu/drm/scheduler/gpu_scheduler.c index 4968867da7a6..088ff2b4e8fb 100644 --- a/drivers/gpu/drm/scheduler/gpu_scheduler.c +++ b/drivers/gpu/drm/scheduler/gpu_scheduler.c @@ -402,6 +402,9 @@ drm_sched_entity_pop_job(struct drm_sched_entity *entity) if (entity->guilty && atomic_read(entity->guilty)) dma_fence_set_error(&sched_job->s_fence->finished, -ECANCELED); + dma_fence_put(entity->last_scheduled); + entity->last_scheduled = dma_fence_get(&sched_job->s_fence->finished); + spsc_queue_pop(&entity->job_queue); return sched_job; } @@ -715,9 +718,6 @@ static int drm_sched_main(void *param) fence = sched->ops->run_job(sched_job); drm_sched_fence_scheduled(s_fence); - dma_fence_put(entity->last_scheduled); - entity->last_scheduled = dma_fence_get(&s_fence->finished); - if (fence) { s_fence->parent = dma_fence_get(fence); r = dma_fence_add_callback(fence, &s_fence->cb,