[PATCH] worker_thread: fix racy try_to_freeze() usage

[PATCH] worker_thread: fix racy try_to_freeze() usage

[Oleg Nesterov]

worker_thread() can miss freeze_process()->signal_wake_up() if it happens
between try_to_freeze() and prepare_to_wait(). We should check freezing()
before entering schedule().

This race was introduced by me in

	[PATCH 1/1] workqueue: don't migrate pending works from the dead CPU

Looks like mm/vmscan.c:kswapd() has the same race.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>

--- 2.6.21-rc5-mm4/kernel/workqueue.c~wq_freeze	2007-04-07 20:11:14.000000000 +0400
+++ 2.6.21-rc5-mm4/kernel/workqueue.c	2007-04-08 21:37:43.000000000 +0400
@@ -307,14 +307,14 @@ static int worker_thread(void *__cwq)
 	do_sigaction(SIGCHLD, &sa, (struct k_sigaction *)0);
 
 	for (;;) {
-		if (cwq->wq->freezeable)
-			try_to_freeze();
-
 		prepare_to_wait(&cwq->more_work, &wait, TASK_INTERRUPTIBLE);
-		if (!cwq->should_stop && list_empty(&cwq->worklist))
+		if (!freezing(current) && !cwq->should_stop
+		    && list_empty(&cwq->worklist))
 			schedule();
 		finish_wait(&cwq->more_work, &wait);
 
+		try_to_freeze();
+
 		if (cwq_should_stop(cwq))
 			break;