Re: [PATCH] kernel/futex.c: Uneeded memory barrier

[Rusty Russell]

In message <3F621160.5020502@xxxxxxxxxxxx> you write:
> >     Kills an unneeded set_current_state after schedule_timeout, since it 
> > already guarantees that the task will be TASK_RUNNING.

In fact, furthur cleanups are possible.

Thanks!
Rusty.
--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.

Name: Clean up futex task state setting
Author: Rusty Russell, Felipe W Damasio <felipewd@xxxxxxxxxxxx>
Depends: Misc/futex-jamie-plus1.patch.gz
Status: Trivial

D: Felipe points out that set_task_state is overkill.  In fact,
D: futex_lock protects us so we can set if after the queued test,
D: simplifying the whole function slightly

diff -urpN --exclude TAGS -X /home/rusty/devel/kernel/kernel-patches/current-dontdiff --minimal .7327-2.6.0-test5-bk3-futex-task_state.pre/kernel/futex.c .7327-2.6.0-test5-bk3-futex-task_state/kernel/futex.c
--- .7327-2.6.0-test5-bk3-futex-task_state.pre/kernel/futex.c	2003-09-15 19:37:14.000000000 +1000
+++ .7327-2.6.0-test5-bk3-futex-task_state/kernel/futex.c	2003-09-15 19:37:14.000000000 +1000
@@ -374,20 +374,16 @@ static int futex_wait(unsigned long uadd
 	 */
 	add_wait_queue(&q.waiters, &wait);
 	spin_lock(&futex_lock);
-	set_current_state(TASK_INTERRUPTIBLE);
 
 	if (unlikely(list_empty(&q.list))) {
-		/*
-		 * We were woken already.
-		 */
+		/* We were woken already. */
 		spin_unlock(&futex_lock);
-		set_current_state(TASK_RUNNING);
 		return 0;
 	}
 
+	__set_current_state(TASK_INTERRUPTIBLE);
 	spin_unlock(&futex_lock);
 	time = schedule_timeout(time);
-	set_current_state(TASK_RUNNING);
 
 	/*
 	 * NOTE: we don't remove ourselves from the waitqueue because
-
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/