Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Nam Cao <redacted>
Reviewed-by: K Prateek Nayak <kprateek.nayak@amd.com>
Signed-off-by: Gabriele Monaco <gmonaco@redhat.com>
---

Notes:
    V5:
    * Do not fire enqueue tracepoint for delayed enqueues

include/trace/events/sched.h | 13 +++++++++++++
kernel/sched/core.c          | 10 +++++++++-
kernel/sched/sched.h         |  2 ++
3 files changed, 24 insertions(+), 1 deletion(-)

diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h
index 366b2e8ec40c..f4e1d3554e3e 100644
--- a/include/trace/events/sched.h
+++ b/include/trace/events/sched.h

@@ -912,6 +912,19 @@ DECLARE_TRACE(sched_dl_server_stop,

    TP_PROTO(struct sched_dl_entity *dl_se, int cpu),
    TP_ARGS(dl_se, cpu));

+/*
+ * The two trace points below may not work as expected for fair tasks due
+ * to delayed dequeue. See:
+ * https://lore.kernel.org/lkml/179674c6-f82a-4718-ace2-67b5e672fdee@amd.com/ (local)
+ */
+DECLARE_TRACE(sched_enqueue,
+   TP_PROTO(struct task_struct *tsk, int cpu),
+   TP_ARGS(tsk, cpu));
+
+DECLARE_TRACE(sched_dequeue,
+   TP_PROTO(struct task_struct *tsk, int cpu),
+   TP_ARGS(tsk, cpu));
+
#endif /* _TRACE_SCHED_H */

/* This part must be outside protection */

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index f6293fa02fb7..c885d7885172 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c

@@ -2087,6 +2087,9 @@ unsigned long get_wchan(struct task_struct *p)

void enqueue_task(struct rq *rq, struct task_struct *p, int flags)
{
+   if (trace_sched_enqueue_tp_enabled() && !(flags & ENQUEUE_DELAYED))
+       trace_sched_enqueue_tp(p, rq->cpu);
+
    if (!(flags & ENQUEUE_NOCLOCK))
        update_rq_clock(rq);

@@ -2114,6 +2117,8 @@ void enqueue_task(struct rq *rq, struct task_struct *p, int flags)

  */
inline bool dequeue_task(struct rq *rq, struct task_struct *p, int flags)
{
+   int ret;
+
    if (sched_core_enabled(rq))
        sched_core_dequeue(rq, p, flags);

@@ -2131,7 +2136,10 @@ inline bool dequeue_task(struct rq *rq, struct task_struct *p, int flags)

     */
    uclamp_rq_dec(rq, p);
    rq->queue_mask |= p->sched_class->queue_mask;
-   return p->sched_class->dequeue_task(rq, p, flags);
+   ret = p->sched_class->dequeue_task(rq, p, flags);
+   if (trace_sched_dequeue_tp_enabled() && !(flags & DEQUEUE_SLEEP))
+       trace_sched_dequeue_tp(p, rq->cpu);
+   return ret;
}

void activate_task(struct rq *rq, struct task_struct *p, int flags)

diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index e885a935b716..8465472b40fa 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h

@@ -2918,6 +2918,8 @@ static inline void sub_nr_running(struct rq *rq, unsigned count)

static inline void __block_task(struct rq *rq, struct task_struct *p)
{
+   trace_sched_dequeue_tp(p, rq->cpu);
+
    if (p->sched_contributes_to_load)
        rq->nr_uninterruptible++;

--
2.52.0