On 20/03/2025 09:58, Pierre-Eric Pelloux-Prayer wrote:
We can't trace dependencies from drm_sched_job_add_dependency because when it's called the job's fence is not available yet. So instead each dependency is traced individually when drm_sched_entity_push_job is used. Tracing the dependencies allows tools to analyze the dependencies between the jobs (previously it was only possible for fences traced by drm_sched_job_wait_dep). Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@xxxxxxx> --- .../gpu/drm/scheduler/gpu_scheduler_trace.h | 24 ++++++++++++++++++- drivers/gpu/drm/scheduler/sched_entity.c | 8 +++++++ 2 files changed, 31 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/scheduler/gpu_scheduler_trace.h b/drivers/gpu/drm/scheduler/gpu_scheduler_trace.h index 21a85ee59066..5d9992ad47d3 100644 --- a/drivers/gpu/drm/scheduler/gpu_scheduler_trace.h +++ b/drivers/gpu/drm/scheduler/gpu_scheduler_trace.h @@ -54,7 +54,6 @@ DECLARE_EVENT_CLASS(drm_sched_job, __assign_str(dev); __entry->fence_context = sched_job->s_fence->finished.context; __entry->fence_seqno = sched_job->s_fence->finished.seqno; - ), TP_printk("dev=%s, id=%llu, fence=%llu:%llu, ring=%s, job count:%u, hw job count:%d", __get_str(dev), __entry->id, @@ -88,6 +87,29 @@ TRACE_EVENT(drm_sched_process_job, __entry->fence_context, __entry->fence_seqno) );+TRACE_EVENT(drm_sched_job_add_dep,+ TP_PROTO(struct drm_sched_job *sched_job, struct dma_fence *fence), + TP_ARGS(sched_job, fence), + TP_STRUCT__entry( + __field(u64, fence_context) + __field(u64, fence_seqno) + __field(u64, id) + __field(u64, ctx) + __field(u64, seqno) + ), + + TP_fast_assign( + __entry->fence_context = sched_job->s_fence->finished.context; + __entry->fence_seqno = sched_job->s_fence->finished.seqno; + __entry->id = sched_job->id; + __entry->ctx = fence->context; + __entry->seqno = fence->seqno; + ), + TP_printk("fence=%llu:%llu, id=%llu depends on fence=%llu:%llu", + __entry->fence_context, __entry->fence_seqno, __entry->id, + __entry->ctx, __entry->seqno) +); + TRACE_EVENT(drm_sched_job_wait_dep, TP_PROTO(struct drm_sched_job *sched_job, struct dma_fence *fence), TP_ARGS(sched_job, fence), diff --git a/drivers/gpu/drm/scheduler/sched_entity.c b/drivers/gpu/drm/scheduler/sched_entity.c index a6d2a4722d82..047e42cfb129 100644 --- a/drivers/gpu/drm/scheduler/sched_entity.c +++ b/drivers/gpu/drm/scheduler/sched_entity.c @@ -580,6 +580,14 @@ void drm_sched_entity_push_job(struct drm_sched_job *sched_job) ktime_t submit_ts;trace_drm_sched_job(sched_job, entity);+ + if (trace_drm_sched_job_add_dep_enabled()) { + struct dma_fence *entry; + unsigned long index; + + xa_for_each(&sched_job->dependencies, index, entry) + trace_drm_sched_job_add_dep(sched_job, entry); + } atomic_inc(entity->rq->sched->score); WRITE_ONCE(entity->last_user, current->group_leader);
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxxx> Regards, Tvrtko
