[PATCH 2/3] percpu-rwsem: don't use percpu_rw_semaphore->rw_sem to exclude writers

[Oleg Nesterov]

percpu_down_write() does down_write() to exclude both the readers and
other writers. We can rely on rcu_sync_enter() in exclusive mode and
take ->rw_sem right before wait_event().

Signed-off-by: Oleg Nesterov <oleg@xxxxxxxxxx>
---
 kernel/locking/percpu-rwsem.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/kernel/locking/percpu-rwsem.c b/kernel/locking/percpu-rwsem.c
index 014d2f4..609c13b 100644
--- a/kernel/locking/percpu-rwsem.c
+++ b/kernel/locking/percpu-rwsem.c
@@ -139,8 +139,6 @@ static bool readers_active_check(struct percpu_rw_semaphore *sem)
 
 void percpu_down_write(struct percpu_rw_semaphore *sem)
 {
-	down_write(&sem->rw_sem);
-
 	/* Notify readers to take the slow path. */
 	rcu_sync_enter(&sem->rss);
 
@@ -158,6 +156,7 @@ void percpu_down_write(struct percpu_rw_semaphore *sem)
 	 * therefore will wait for them.
 	 */
 
+	down_write(&sem->rw_sem);
 	/* Wait for all now active readers to complete. */
 	wait_event(sem->writer, readers_active_check(sem));
 }
-- 
1.5.5.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/