[PATCH 2/2 V4] workqueue: fix possible race condition when rescuer VS pwq-release

[Lai Jiangshan <laijs@cn.fujitsu.com>]

There is a race condition between rescuer_thread() and
pwq_unbound_release_workfn().

The works of the @pwq may be processed by some other workers,
and @pwq is scheduled to release(due to its wq's attr is changed)
before the rescuer starts to process. In this case
pwq_unbound_release_workfn() will corrupt wq->maydays list,
and rescuer_thead() will access to corrupted data.

Using get_pwq() pin it until rescuer is done with it.

Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
---
 kernel/workqueue.c |   13 +++++++++++++
 1 files changed, 13 insertions(+), 0 deletions(-)

diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 7539244..8c0830c 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -1916,6 +1916,16 @@ static void send_mayday(struct work_struct *work)
 
 	/* mayday mayday mayday */
 	if (list_empty(&pwq->mayday_node)) {
+		/*
+		 * pwqs might go away at any time, pin it until the
+		 * rescuer is done with it.
+		 *
+		 * Especially a pwq of an unbound wq may be released
+		 * before wq's destruction when the wq's attr is changed.
+		 * In this case, pwq_unbound_release_workfn() may execute
+		 * earlier before rescuer_thread() and corrupt wq->maydays.
+		 */
+		get_pwq(pwq);
 		list_add_tail(&pwq->mayday_node, &wq->maydays);
 		wake_up_process(wq->rescuer->task);
 	}
@@ -2447,6 +2457,9 @@ repeat:
 
 		process_scheduled_works(rescuer);
 
+		/* put the reference grabbed by send_mayday(). */
+		put_pwq(pwq);
+
 		/*
 		 * Leave this pool.  If keep_working() is %true, notify a
 		 * regular worker; otherwise, we end up with 0 concurrency
-- 
1.7.4.4