[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <1261739467.10685.18.camel@minggr.sh.intel.com>
Date: Fri, 25 Dec 2009 19:11:07 +0800
From: Lin Ming <ming.m.lin@...el.com>
To: peterz@...radead.org, mingo@...e.hu, tglx@...utronix.de
Cc: linux-kernel <linux-kernel@...r.kernel.org>,
"Zhang, Yanmin" <yanmin_zhang@...ux.intel.com>
Subject: tbench regression with 2.6.33-rc1
Hi,
Test machine: 16 cpus (4P/2Core/HT), 8G mem
tbench test command:
tbench_srv &
tbench 32
Compared with 2.6.32, tbench has ~4% regression in 2.6.33-rc1.
>>From vmstat data, the context switch number also drop ~4%.
perf top data does not show much differences.
But lockstat data shows huge difference in rq->lock, as below.
See the attachment for the full lockstat data.
Any clue of this regression?
Thanks,
Lin Ming
2.6.32
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
class name con-bounces contentions waittime-min waittime-max waittime-total acq-bounces acquisitions holdtime-min holdtime-max holdtime-total
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
&rq->lock: 901 902 1.80 80.65 2402.48 27425 26037113 0.84 4278.44 27119526.04
---------
&rq->lock 4 [<ffffffff81036a43>] __sched_setscheduler+0x1c8/0x393
&rq->lock 223 [<ffffffff814321c5>] schedule+0xe2/0x830
&rq->lock 657 [<ffffffff8103107d>] task_rq_lock+0x39/0x67
&rq->lock 16 [<ffffffff8103d2b6>] scheduler_tick+0x3d/0x206
---------
&rq->lock 425 [<ffffffff814321c5>] schedule+0xe2/0x830
&rq->lock 427 [<ffffffff8103107d>] task_rq_lock+0x39/0x67
&rq->lock 20 [<ffffffff8103d2b6>] scheduler_tick+0x3d/0x206
&rq->lock 4 [<ffffffff81033292>] double_lock_balance+0x40/0x6a
2.6.33-rc1
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
class name con-bounces contentions waittime-min waittime-max waittime-total acq-bounces acquisitions holdtime-min holdtime-max holdtime-total
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
&rq->lock: 5987 6019 0.00 25.15 20551.87 37851 25873389 0.00 3819.73 29524592.69
---------
&rq->lock 5254 [<ffffffff8102a6d9>] task_rq_lock+0x39/0x67
&rq->lock 410 [<ffffffff81439c75>] schedule+0xba/0x8ae
&rq->lock 336 [<ffffffff8103086f>] try_to_wake_up+0xe3/0x26f
&rq->lock 28 [<ffffffff8102c995>] double_lock_balance+0x40/0x6a
---------
&rq->lock 468 [<ffffffff8102a6d9>] task_rq_lock+0x39/0x67
&rq->lock 5022 [<ffffffff81439c75>] schedule+0xba/0x8ae
&rq->lock 13 [<ffffffff81030555>] double_rq_lock+0x40/0x79
&rq->lock 477 [<ffffffff8103086f>] try_to_wake_up+0xe3/0x26f
View attachment "lockstat.tbench.2.6.32" of type "text/plain" (185607 bytes)
View attachment "lockstat.tbench.2.6.33-rc1" of type "text/plain" (188330 bytes)
Powered by blists - more mailing lists