[request for stable inclusion] nohz: Fix another inconsistency between CONFIG_NO_HZ=n and nohz=off

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Greg or Thomas,

0e576acbc1d9600cf2d9b4a141a2554639959d50 
nohz: Fix another inconsistency between CONFIG_NO_HZ=n and nohz=off

This fixed an inconsistency about nohz. 

Without this patch, When we boot with nohz=off and CONFIG_NO_HZ=Y, 
and the system is idle, powertop shows the cpu is always in polling
state. It can affect the performance in some cases. 

*without patch*
     PowerTOP version 1.13      (C) 2007 Intel Corporation

Cn                Avg residency       P-states (frequencies)
C0 (cpu running)        ( 0.0%)       Turbo Mode     0.0%
polling          70.3ms (100.3%)        2.40 Ghz     0.0%
C1 mwait          0.0ms ( 0.0%)         2.27 Ghz     0.0%
C2 mwait          0.0ms ( 0.0%)         2.14 Ghz     0.0%
C3 mwait          0.0ms ( 0.0%)         1.60 Ghz   100.0%

*with patch*
     PowerTOP version 1.13      (C) 2007 Intel Corporation

Cn                Avg residency       P-states (frequencies)
C0 (cpu running)        ( 0.3%)       Turbo Mode     0.0%
polling           0.0ms ( 0.0%)         2.40 Ghz     0.0%
C1 mwait          0.0ms ( 0.0%)         2.27 Ghz     0.0%
C2 mwait          0.0ms ( 0.0%)         2.14 Ghz     0.0%
C3 mwait          3.9ms (99.7%)         1.60 Ghz   100.0%


This patch looks applicable to stable-3.4 and stable 3.10. It can be
applyed to 3.10 cleanly, but should be added some adjustment for 3.4.

And what do you think?

Thanks,
Rui

-----------------------

From: Thomas Gleixner <tglx@xxxxxxxxxxxxx>

commit 0e576acbc1d9600cf2d9b4a141a2554639959d50 upstream

If CONFIG_NO_HZ=n tick_nohz_get_sleep_length() returns NSEC_PER_SEC/HZ.

If CONFIG_NO_HZ=y and the nohz functionality is disabled via the
command line option "nohz=off" or not enabled due to missing hardware
support, then tick_nohz_get_sleep_length() returns 0. That happens
because ts->sleep_length is never set in that case.

Set it to NSEC_PER_SEC/HZ when the NOHZ mode is inactive.

Reported-by: Michal Hocko <mhocko@xxxxxxx>
Reported-by: Borislav Petkov <bp@xxxxxxxxx>
Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
[xr: Backported to 3.4:
  - applied the change to tick_nohz_stop_sched_tick() instead of
    can_stop_idle_tick]
Signed-off-by: Rui Xiang <rui.xiang@xxxxxxxxxx>
---
 kernel/time/tick-sched.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
index 638dadf..67106f8 100644
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -296,8 +296,10 @@ static void tick_nohz_stop_sched_tick(struct tick_sched *ts)
 			tick_do_timer_cpu = TICK_DO_TIMER_NONE;
 	}
 
-	if (unlikely(ts->nohz_mode == NOHZ_MODE_INACTIVE))
+	if (unlikely(ts->nohz_mode == NOHZ_MODE_INACTIVE)) {
+		ts->sleep_length = (ktime_t) { .tv64 = NSEC_PER_SEC/HZ };
 		return;
+	}
 
 	if (need_resched())
 		return;
-- 


--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [Linux Kernel]     [Kernel Development Newbies]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite Hiking]     [Linux Kernel]     [Linux SCSI]