aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/scheduler
diff options
context:
space:
mode:
authorMaxime Ripard2021-04-26 14:03:09 +0200
committerMaxime Ripard2021-04-26 14:03:09 +0200
commit355b60296143a090039211c5f0e1463f84aab65a (patch)
treeb74d4ef2aea66252ea9cf77c847de6c6e72a02b7 /drivers/gpu/drm/scheduler
parent91185d55b32e7e377f15fb46a62b216f8d3038d4 (diff)
parenta1a1ca70deb3ec600eeabb21de7f3f48aaae5695 (diff)
Merge drm/drm-next into drm-misc-next
Christian needs some patches from drm/next Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Diffstat (limited to 'drivers/gpu/drm/scheduler')
-rw-r--r--drivers/gpu/drm/scheduler/sched_main.c103
1 files changed, 71 insertions, 32 deletions
diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c
index d82a7ebf6099..92d8de24d0a1 100644
--- a/drivers/gpu/drm/scheduler/sched_main.c
+++ b/drivers/gpu/drm/scheduler/sched_main.c
@@ -361,40 +361,16 @@ static void drm_sched_job_timedout(struct work_struct *work)
*/
void drm_sched_increase_karma(struct drm_sched_job *bad)
{
- int i;
- struct drm_sched_entity *tmp;
- struct drm_sched_entity *entity;
- struct drm_gpu_scheduler *sched = bad->sched;
-
- /* don't increase @bad's karma if it's from KERNEL RQ,
- * because sometimes GPU hang would cause kernel jobs (like VM updating jobs)
- * corrupt but keep in mind that kernel jobs always considered good.
- */
- if (bad->s_priority != DRM_SCHED_PRIORITY_KERNEL) {
- atomic_inc(&bad->karma);
- for (i = DRM_SCHED_PRIORITY_MIN; i < DRM_SCHED_PRIORITY_KERNEL;
- i++) {
- struct drm_sched_rq *rq = &sched->sched_rq[i];
-
- spin_lock(&rq->lock);
- list_for_each_entry_safe(entity, tmp, &rq->entities, list) {
- if (bad->s_fence->scheduled.context ==
- entity->fence_context) {
- if (atomic_read(&bad->karma) >
- bad->sched->hang_limit)
- if (entity->guilty)
- atomic_set(entity->guilty, 1);
- break;
- }
- }
- spin_unlock(&rq->lock);
- if (&entity->list != &rq->entities)
- break;
- }
- }
+ drm_sched_increase_karma_ext(bad, 1);
}
EXPORT_SYMBOL(drm_sched_increase_karma);
+void drm_sched_reset_karma(struct drm_sched_job *bad)
+{
+ drm_sched_increase_karma_ext(bad, 0);
+}
+EXPORT_SYMBOL(drm_sched_reset_karma);
+
/**
* drm_sched_stop - stop the scheduler
*
@@ -534,14 +510,31 @@ EXPORT_SYMBOL(drm_sched_start);
*/
void drm_sched_resubmit_jobs(struct drm_gpu_scheduler *sched)
{
+ drm_sched_resubmit_jobs_ext(sched, INT_MAX);
+}
+EXPORT_SYMBOL(drm_sched_resubmit_jobs);
+
+/**
+ * drm_sched_resubmit_jobs_ext - helper to relunch certain number of jobs from mirror ring list
+ *
+ * @sched: scheduler instance
+ * @max: job numbers to relaunch
+ *
+ */
+void drm_sched_resubmit_jobs_ext(struct drm_gpu_scheduler *sched, int max)
+{
struct drm_sched_job *s_job, *tmp;
uint64_t guilty_context;
bool found_guilty = false;
struct dma_fence *fence;
+ int i = 0;
list_for_each_entry_safe(s_job, tmp, &sched->pending_list, list) {
struct drm_sched_fence *s_fence = s_job->s_fence;
+ if (i >= max)
+ break;
+
if (!found_guilty && atomic_read(&s_job->karma) > sched->hang_limit) {
found_guilty = true;
guilty_context = s_job->s_fence->scheduled.context;
@@ -552,6 +545,7 @@ void drm_sched_resubmit_jobs(struct drm_gpu_scheduler *sched)
dma_fence_put(s_job->s_fence->parent);
fence = sched->ops->run_job(s_job);
+ i++;
if (IS_ERR_OR_NULL(fence)) {
if (IS_ERR(fence))
@@ -563,7 +557,7 @@ void drm_sched_resubmit_jobs(struct drm_gpu_scheduler *sched)
}
}
}
-EXPORT_SYMBOL(drm_sched_resubmit_jobs);
+EXPORT_SYMBOL(drm_sched_resubmit_jobs_ext);
/**
* drm_sched_job_init - init a scheduler job
@@ -903,3 +897,48 @@ void drm_sched_fini(struct drm_gpu_scheduler *sched)
sched->ready = false;
}
EXPORT_SYMBOL(drm_sched_fini);
+
+/**
+ * drm_sched_increase_karma_ext - Update sched_entity guilty flag
+ *
+ * @bad: The job guilty of time out
+ * @type: type for increase/reset karma
+ *
+ */
+void drm_sched_increase_karma_ext(struct drm_sched_job *bad, int type)
+{
+ int i;
+ struct drm_sched_entity *tmp;
+ struct drm_sched_entity *entity;
+ struct drm_gpu_scheduler *sched = bad->sched;
+
+ /* don't change @bad's karma if it's from KERNEL RQ,
+ * because sometimes GPU hang would cause kernel jobs (like VM updating jobs)
+ * corrupt but keep in mind that kernel jobs always considered good.
+ */
+ if (bad->s_priority != DRM_SCHED_PRIORITY_KERNEL) {
+ if (type == 0)
+ atomic_set(&bad->karma, 0);
+ else if (type == 1)
+ atomic_inc(&bad->karma);
+
+ for (i = DRM_SCHED_PRIORITY_MIN; i < DRM_SCHED_PRIORITY_KERNEL;
+ i++) {
+ struct drm_sched_rq *rq = &sched->sched_rq[i];
+
+ spin_lock(&rq->lock);
+ list_for_each_entry_safe(entity, tmp, &rq->entities, list) {
+ if (bad->s_fence->scheduled.context ==
+ entity->fence_context) {
+ if (entity->guilty)
+ atomic_set(entity->guilty, type);
+ break;
+ }
+ }
+ spin_unlock(&rq->lock);
+ if (&entity->list != &rq->entities)
+ break;
+ }
+ }
+}
+EXPORT_SYMBOL(drm_sched_increase_karma_ext);