[PATCH] i386: Don't use the TSC in sched_clock if unstable
authorGuillaume Chazarain <guichaz@gmail.com>
Fri, 16 Mar 2007 20:07:36 +0000 (21:07 +0100)
committerAndi Kleen <andi@basil.nowhere.org>
Fri, 16 Mar 2007 20:07:36 +0000 (21:07 +0100)
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=f9690982b8c2f9a2c65acdc113e758ec356676a3
caused a regression by letting sched_clock use the TSC even when cpufreq
disabled it. This caused scheduling weirdnesses.

Signed-off-by: Guillaume Chazarain <guichaz@yahoo.fr>
Signed-off-by: Andi Kleen <ak@suse.de>
arch/i386/kernel/tsc.c

index 602660df455c7676b2fd9400f628831b8b3d716c..0e65f7ab63a204081d908507eb35baddc157dde8 100644 (file)
@@ -105,7 +105,7 @@ unsigned long long sched_clock(void)
        /*
         * Fall back to jiffies if there's no TSC available:
         */
-       if (unlikely(tsc_disable))
+       if (tsc_unstable || unlikely(tsc_disable))
                /* No locking but a rare wrong value is not a big deal: */
                return (jiffies_64 - INITIAL_JIFFIES) * (1000000000 / HZ);