Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 3e99d8eb authored by Chris Mason's avatar Chris Mason
Browse files

Btrfs: fix async worker startup race 



After a new worker thread starts, it is placed into the
list of idle threads.  But, this may race with a
check for idle done by the worker thread itself, resulting
in a double list_add operation.

This fix adds a check to make sure the idle thread addition
is done properly.

Signed-off-by: default avatarChris Mason <chris.mason@oracle.com>
parent 83ebade3

fs/btrfs/async-thread.c

+11 −3
Original line number Diff line number Diff line
@@ -74,7 +74,12 @@ static void check_idle_worker(struct btrfs_worker_thread *worker) 
		unsigned long flags;

		spin_lock_irqsave(&worker->workers->lock, flags);

		worker->idle = 1;

		list_move(&worker->worker_list, &worker->workers->idle_list);



		/* the list may be empty if the worker is just starting */

		if (!list_empty(&worker->worker_list)) {

			list_move(&worker->worker_list,

				 &worker->workers->idle_list);

		}

		spin_unlock_irqrestore(&worker->workers->lock, flags);

	}

}
@@ -90,8 +95,11 @@ static void check_busy_worker(struct btrfs_worker_thread *worker) 
		unsigned long flags;

		spin_lock_irqsave(&worker->workers->lock, flags);

		worker->idle = 0;



		if (!list_empty(&worker->worker_list)) {

			list_move_tail(&worker->worker_list,

				      &worker->workers->worker_list);

		}

		spin_unlock_irqrestore(&worker->workers->lock, flags);

	}

}