Skip to contentSkip to conversation

sched_ext: Update scx_dispatch_dequeue() comments

PATCH Superseded Usama Arif 2 participants 2 messages Sep 23

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Usama Arif

author
#0 lore raw

PATCH 1/1 sched_ext: Update scx_dispatch_dequeue() comments

Remote consumption now reaches the !dsq branch with holding_cpu set, just like direct dispatch. Also, the dsq-present branch races with unlink_dsq_and_switch_rq_lock(), not dispatch_to_local_dsq().

Update both comments to describe the current transfer paths. No functional change.

---
 kernel/sched/ext/ext.c | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)
kernel/sched/ext/ext.c +6 -7
index 0135ff6824b03..f95d1cf8d3d1c 100644
@@ -1832,10 +1832,9 @@ void scx_dispatch_dequeue(struct rq *rq, struct task_struct *p)
 			list_del_init(&p->scx.dsq_list.node);
 
 		/*
-		 * When dispatching directly from the BPF scheduler to a local
-		 * DSQ, the task isn't associated with any DSQ but
-		 * @p->scx.holding_cpu may be set under the protection of
-		 * %SCX_OPSS_DISPATCHING.
+		 * When moving a task to a local DSQ through either direct dispatch
+		 * or remote consumption, the task isn't associated with any DSQ but
+		 * @p->scx.holding_cpu may be set.
 		 */
 		if (p->scx.holding_cpu >= 0)
 			p->scx.holding_cpu = -1;
@@ -1855,9 +1854,9 @@ void scx_dispatch_dequeue(struct rq *rq, struct task_struct *p)
 		scx_task_unlink_from_dsq(p, dsq);
 	} else {
 		/*
-		 * We're racing against dispatch_to_local_dsq() which already
-		 * removed @p from @dsq and set @p->scx.holding_cpu. Clear the
-		 * holding_cpu which tells dispatch_to_local_dsq() that it lost
+		 * We're racing against unlink_dsq_and_switch_rq_lock(), which
+		 * already removed @p from @dsq and set @p->scx.holding_cpu. Clear
+		 * holding_cpu to tell unlink_dsq_and_switch_rq_lock() that it lost
 		 * the race.
 		 */
 		WARN_ON_ONCE(!list_empty(&p->scx.dsq_list.node));
Signature · 1 line
2.53.0-Meta
reviewer
#1 lore raw

Hello, Usama.

kernel/sched/ext/ext.c void scx_dispatch_dequeue(struct rq *rq, struct task_struct *p) · line 1838

Quoted 9 lines · Usama Arif
kernel/sched/ext/ext.c +3 -4
 		/*
-		 * When dispatching directly from the BPF scheduler to a local
-		 * DSQ, the task isn't associated with any DSQ but
-		 * @p->scx.holding_cpu may be set under the protection of
-		 * %SCX_OPSS_DISPATCHING.
+		 * When moving a task to a local DSQ through either direct dispatch
+		 * or remote consumption, the task isn't associated with any DSQ but
+		 * @p->scx.holding_cpu may be set.
 		 */

dispatch_to_local_dsq() sets holding_cpu for ops.dispatch() inserts too, not just direct dispatches. Can you name dispatch_to_local_dsq() instead, here and in the description?

Clearing holding_cpu here is also how dispatch_to_local_dsq() learns that it lost to a dequeue. The old comment on the other branch described that, but for the wrong branch. Can you add a sentence about it here?

Can you also rewrap both comments at 80 columns? A few lines go past it.

Thanks.

--
tejun