[PATCH 2/3] wait_task_stopped: tidy up the noreap case

[Oleg Nesterov]

wait_task_stopped(WNOWAIT) unlocks tasklist_lock and re-checks ->exit_code and
->exit_state. This is not needed: both were valid before we dropped the lock,
and without tasklist_lock both are not stable anyway.

Read the exit_code under tasklist and report the cached value without re-check.
In fact this fixes the race with the dying child, we can report a completely
false exit_code if ->exit_state is not visible yet.

Signed-off-by: Oleg Nesterov <oleg@xxxxxxxxxx>

--- 24/kernel/exit.c~2_NOREAP	2007-11-16 18:13:54.000000000 +0300
+++ 24/kernel/exit.c	2007-11-16 18:18:24.000000000 +0300
@@ -1356,10 +1356,10 @@ static int wait_task_stopped(struct task
 			     int noreap, struct siginfo __user *infop,
 			     int __user *stat_addr, struct rusage __user *ru)
 {
-	int retval, exit_code;
+	int retval, exit_code = p->exit_code;
 	pid_t pid;
 
-	if (!p->exit_code)
+	if (!exit_code)
 		return 0;
 	if (delayed_group_leader && !(p->ptrace & PT_PTRACED) &&
 	    p->signal->group_stop_count > 0)
@@ -1384,9 +1384,6 @@ static int wait_task_stopped(struct task
 		uid_t uid = p->uid;
 		int why = (p->ptrace & PT_PTRACED) ? CLD_TRAPPED : CLD_STOPPED;
 
-		exit_code = p->exit_code;
-		if (unlikely(!exit_code) || unlikely(p->exit_state))
-			goto bail_ref;
 		return wait_noreap_copyout(p, pid, uid,
 					   why, (exit_code << 8) | 0x7f,
 					   infop, ru);
@@ -1417,7 +1414,6 @@ static int wait_task_stopped(struct task
 		 * resumed, or it resumed and then died.
 		 */
 		write_unlock_irq(&tasklist_lock);
-bail_ref:
 		put_task_struct(p);
 		/*
 		 * We are returning to the wait loop without having successfully

-
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/