[RFC,v3,09/10] sched/fair: Add a class->task_woken callback in preparation for per-task throttling

Commit Message

Valentin Schneider July 11, 2024, 1 p.m. UTC

Later commits will change CFS bandwidth control throttling from a
per-cfs_rq basis to a per-task basis. This means special care needs to be
taken around any transition a task can have into and out of a cfs_rq.

To ease reviewing, the transitions are patched with dummy-helpers that are
implemented later on.

Add a class->task_woken callback to handle tasks being woken into
potentially throttled cfs_rq's. Conversely, a task flagged for
throttle-at-kernel-exit may block and need to have its pending throttle
removed if runtime was replenished by the time it got woken up.

Signed-off-by: Valentin Schneider <[e-mail]>
---
 kernel/sched/fair.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

Patch

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index b2242307677ca..0cec3e70f1277 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -5776,6 +5776,17 @@  static void task_throttle_cancel_irq_work_fn(struct irq_work *work)
        /* Write me */
 }
 
+static void task_woken_fair(struct rq *rq, struct task_struct *p)
+{
+	if (!cfs_bandwidth_used())
+		return;
+
+	if (task_needs_throttling(p))
+		task_throttle_setup(p);
+	else
+		task_throttle_cancel(p);
+}
+
 void init_cfs_throttle_work(struct task_struct *p)
 {
 	/* Protect against double add, see throttle_cfs_rq() and throttle_cfs_rq_work() */
@@ -13288,6 +13299,10 @@  DEFINE_SCHED_CLASS(fair) = {
 	.task_change_group	= task_change_group_fair,
 #endif
 
+#ifdef CONFIG_CFS_BANDWIDTH
+	.task_woken             = task_woken_fair,
+#endif
+
 #ifdef CONFIG_SCHED_CORE
 	.task_is_throttled	= task_is_throttled_fair,
 #endif