]> git.ipfire.org Git - thirdparty/kernel/stable.git/commitdiff
rcu-tasks: Don't delete holdouts within trc_wait_for_one_reader()
authorPaul E. McKenney <paulmck@kernel.org>
Tue, 25 May 2021 18:28:40 +0000 (11:28 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 31 Jul 2021 06:13:46 +0000 (08:13 +0200)
[ Upstream commit a9ab9cce9367a2cc02a3c7eb57a004dc0b8f380d ]

Invoking trc_del_holdout() from within trc_wait_for_one_reader() is
only a performance optimization because the RCU Tasks Trace grace-period
kthread will eventually do this within check_all_holdout_tasks_trace().
But it is not a particularly important performance optimization because
it only applies to the grace-period kthread, of which there is but one.
This commit therefore removes this invocation of trc_del_holdout() in
favor of the one in check_all_holdout_tasks_trace() in the grace-period
kthread.

Reported-by: "Xu, Yanfei" <yanfei.xu@windriver.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
kernel/rcu/tasks.h

index 71e9d625371ac7af6fde4f23e61e86c9fdd3cc88..fcef5f0c60b8b97869ed864feac84b2fb3158104 100644 (file)
@@ -937,7 +937,6 @@ static void trc_wait_for_one_reader(struct task_struct *t,
        // The current task had better be in a quiescent state.
        if (t == current) {
                t->trc_reader_checked = true;
-               trc_del_holdout(t);
                WARN_ON_ONCE(t->trc_reader_nesting);
                return;
        }