sched: turn off PREEMPT_RESTRICT
authorIngo Molnar <mingo@elte.hu>
Fri, 9 Nov 2007 21:39:39 +0000 (22:39 +0100)
committerIngo Molnar <mingo@elte.hu>
Fri, 9 Nov 2007 21:39:39 +0000 (22:39 +0100)
PREEMPT_RESTRICT was a method aimed at reducing the amount of wakeup
related preemption. It has a disadvantage though, it can prevent
legitimate wakeups if a task is 'unlucky' to be hit too early by a tick
that clears peer_preempt.

Now that the wakeup preemption has been cleaned up we dont seem to have
excessive preemptions anymore, so this feature can be turned off. (and
removed in the next patch)

Signed-off-by: Ingo Molnar <mingo@elte.hu>
kernel/sched.c

index 387258cdd0b9bd6709ad8a992b64c104d87f03be..4b23dfb4c80f5d045f062a1e57519ce8a633b74d 100644 (file)
@@ -469,7 +469,7 @@ const_debug unsigned int sysctl_sched_features =
                SCHED_FEAT_TREE_AVG             * 0 |
                SCHED_FEAT_APPROX_AVG           * 0 |
                SCHED_FEAT_WAKEUP_PREEMPT       * 1 |
-               SCHED_FEAT_PREEMPT_RESTRICT     * 1;
+               SCHED_FEAT_PREEMPT_RESTRICT     * 0;
 
 #define sched_feat(x) (sysctl_sched_features & SCHED_FEAT_##x)