x86/idle: Restore trace_cpu_idle to mwait_idle() calls
authorJisheng Zhang <jszhang@marvell.com>
Thu, 20 Aug 2015 04:54:39 +0000 (12:54 +0800)
committerThomas Gleixner <tglx@linutronix.de>
Thu, 20 Aug 2015 19:37:45 +0000 (21:37 +0200)
Commit b253149b843f ("sched/idle/x86: Restore mwait_idle() to fix boot
hangs, to improve power savings and to improve performance") restores
mwait_idle(), but the trace_cpu_idle related calls are missing. This
causes powertop on my old desktop powered by Intel Core2 E6550 to
report zero wakeups and zero events.

Add them back to restore the proper behaviour.

Fixes: b253149b843f ("sched/idle/x86: Restore mwait_idle() to ...")
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Cc: <len.brown@intel.com>
Cc: stable@vger.kernel.org # 4.1
Link: http://lkml.kernel.org/r/1440046479-4262-1-git-send-email-jszhang@marvell.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
arch/x86/kernel/process.c

index 397688beed4be5ce7d9445d7847d44613d2d84b5..c27cad7267655c3794972344adf0b7924e38c138 100644 (file)
@@ -408,6 +408,7 @@ static int prefer_mwait_c1_over_halt(const struct cpuinfo_x86 *c)
 static void mwait_idle(void)
 {
        if (!current_set_polling_and_test()) {
+               trace_cpu_idle_rcuidle(1, smp_processor_id());
                if (this_cpu_has(X86_BUG_CLFLUSH_MONITOR)) {
                        smp_mb(); /* quirk */
                        clflush((void *)&current_thread_info()->flags);
@@ -419,6 +420,7 @@ static void mwait_idle(void)
                        __sti_mwait(0, 0);
                else
                        local_irq_enable();
+               trace_cpu_idle_rcuidle(PWR_EVENT_EXIT, smp_processor_id());
        } else {
                local_irq_enable();
        }