commit f95d39d10fe7d47336e65172f52bf64e0096f983
author Oleg Nesterov <oleg@redhat.com> Wed Jun 17 16:27:42 2009 -0700
committer Linus Torvalds <torvalds@linux-foundation.org> Thu Jun 18 13:03:53 2009 -0700
tree 14b469b68455968dfd707695083ece72287dfaab
parent a3f6dfb7295facb0505b5beca5a7ce48b0612379

do_wait: fix the theoretical race with stop/trace/cont

do_wait:

	current->state = TASK_INTERRUPTIBLE;

	read_lock(&tasklist_lock);
	... search for the task to reap ...

In theory, the ->state changing can leak into the critical section.  Since
the child can change its status under read_lock(tasklist) in parallel
(finish_stop/ptrace_stop), we can miss the wakeup if __wake_up_parent()
sees us in TASK_RUNNING state.  Add the barrier.

Also, use __set_current_state() to set TASK_RUNNING.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Ingo Molnar <mingo@elte.hu>
Acked-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/exit.c

diff --git a/kernel/exit.c b/kernel/exit.c
index dd83c84..7ef355d 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -1581,7 +1581,7 @@
 	   (!wo->wo_pid || hlist_empty(&wo->wo_pid->tasks[wo->wo_type])))
 		goto notask;
 
-	current->state = TASK_INTERRUPTIBLE;
+	set_current_state(TASK_INTERRUPTIBLE);
 	read_lock(&tasklist_lock);
 	tsk = current;
 	do {
@@ -1608,7 +1608,7 @@
 		}
 	}
 end:
-	current->state = TASK_RUNNING;
+	__set_current_state(TASK_RUNNING);
 	remove_wait_queue(&current->signal->wait_chldexit,&wait);
 	if (wo->wo_info) {
 		struct siginfo __user *infop = wo->wo_info;