Merge tag 'v6.9-rc1' into sched/core, to pick up fixes and to refresh the branch

Signed-off-by: Ingo Molnar <mingo@kernel.org>

1 files changed, 9 insertions, 22 deletions

diff --git a/kernel/time/timer_migration.c b/kernel/time/timer_migration.c
index 8f49b6b96dfd..c63a0afdcebe 100644
--- a/kernel/time/timer_migration.c
+++ b/kernel/time/timer_migration.c
@@ -751,26 +751,6 @@ bool tmigr_update_events(struct tmigr_group *group, struct tmigr_group *child,
 
 		first_childevt = evt = data->evt;
 
-		/*
-		 * Walking the hierarchy is required in any case when a
-		 * remote expiry was done before. This ensures to not lose
-		 * already queued events in non active groups (see section
-		 * "Required event and timerqueue update after a remote
-		 * expiry" in the documentation at the top).
-		 *
-		 * The two call sites which are executed without a remote expiry
-		 * before, are not prevented from propagating changes through
-		 * the hierarchy by the return:
-		 *  - When entering this path by tmigr_new_timer(), @evt->ignore
-		 *    is never set.
-		 *  - tmigr_inactive_up() takes care of the propagation by
-		 *    itself and ignores the return value. But an immediate
-		 *    return is required because nothing has to be done in this
-		 *    level as the event could be ignored.
-		 */
-		if (evt->ignore && !remote)
-			return true;
-
 		raw_spin_lock(&group->lock);
 
 		childstate.state = 0;
@@ -1058,8 +1038,15 @@ void tmigr_handle_remote(void)
 	 * in tmigr_handle_remote_up() anyway. Keep this check to speed up the
 	 * return when nothing has to be done.
 	 */
-	if (!tmigr_check_migrator(tmc->tmgroup, tmc->childmask))
-		return;
+	if (!tmigr_check_migrator(tmc->tmgroup, tmc->childmask)) {
+		/*
+		 * If this CPU was an idle migrator, make sure to clear its wakeup
+		 * value so it won't chase timers that have already expired elsewhere.
+		 * This avoids endless requeue from tmigr_new_timer().
+		 */
+		if (READ_ONCE(tmc->wakeup) == KTIME_MAX)
+			return;
+	}
 
 	data.now = get_jiffies_update(&data.basej);