[LKP] [x86_64,entry] 71428f63e68: -35.9% will-it-scale.time.user_time +8.2% will-it-scale.per_process_ops
From: Huang Ying
Date: Wed Jan 21 2015 - 01:16:44 EST
FYI, we noticed the below changes on
git://git.kernel.org/pub/scm/linux/kernel/git/luto/linux.git x86/entry-devel
commit 71428f63e681e1b4aa1a781e3ef7c27f027d1103 ("x86_64,entry: Use sysret to return to userspace when possible")
testbox/testcase/testparams: wsm/will-it-scale/performance-unlink2
c9a418c3dc3993cb 71428f63e681e1b4aa1a781e3e
---------------- --------------------------
%stddev %change %stddev
\ | \
35.37 Â 0% -39.9% 21.24 Â 0% will-it-scale.time.user_time
206954 Â 0% +2.1% 211373 Â 0% will-it-scale.per_process_ops
992 Â 0% +1.3% 1005 Â 0% will-it-scale.time.system_time
35.37 Â 0% -39.9% 21.24 Â 0% time.user_time
8.51 Â 1% -18.5% 6.93 Â 1% perf-profile.cpu-cycles.close
157780 Â 1% +10.5% 174306 Â 2% slabinfo.kmalloc-256.active_objs
157803 Â 1% +10.5% 174330 Â 2% slabinfo.kmalloc-256.num_objs
4930 Â 1% +10.5% 5447 Â 2% slabinfo.kmalloc-256.active_slabs
4930 Â 1% +10.5% 5447 Â 2% slabinfo.kmalloc-256.num_slabs
157357 Â 1% +10.5% 173860 Â 2% slabinfo.shmem_inode_cache.active_objs
157370 Â 1% +10.5% 173871 Â 2% slabinfo.shmem_inode_cache.num_objs
6556 Â 1% +10.5% 7244 Â 2% slabinfo.shmem_inode_cache.active_slabs
6556 Â 1% +10.5% 7244 Â 2% slabinfo.shmem_inode_cache.num_slabs
44396 Â 2% +9.9% 48785 Â 2% proc-vmstat.nr_slab_unreclaimable
testbox/testcase/testparams: wsm/will-it-scale/performance-signal1
c9a418c3dc3993cb 71428f63e681e1b4aa1a781e3e
---------------- --------------------------
%stddev %change %stddev
\ | \
35.20 Â 2% -35.9% 22.56 Â 1% will-it-scale.time.user_time
365460 Â 1% +7.3% 391965 Â 2% will-it-scale.per_thread_ops
971513 Â 0% +8.2% 1050889 Â 0% will-it-scale.per_process_ops
0.59 Â 1% -5.0% 0.56 Â 0% will-it-scale.scalability
1028 Â 0% +1.2% 1040 Â 0% will-it-scale.time.system_time
640 Â 33% -49.5% 323 Â 20% sched_debug.cfs_rq[7]:/.blocked_load_avg
127248 Â 37% +85.3% 235849 Â 22% sched_debug.cpu#1.ttwu_local
129447 Â 36% +84.0% 238199 Â 21% sched_debug.cpu#1.ttwu_count
129567 Â 36% +83.6% 237839 Â 22% sched_debug.cpu#1.sched_goidle
259984 Â 36% +83.4% 476883 Â 21% sched_debug.cpu#1.nr_switches
260083 Â 36% +83.4% 476997 Â 21% sched_debug.cpu#1.sched_count
725 Â 28% -44.9% 400 Â 17% sched_debug.cfs_rq[7]:/.tg_load_contrib
35.20 Â 2% -35.9% 22.56 Â 1% time.user_time
112 Â 30% -33.6% 74 Â 14% sched_debug.cfs_rq[1]:/.load
102 Â 19% -26.9% 74 Â 14% sched_debug.cpu#1.load
62 Â 4% +33.9% 84 Â 7% sched_debug.cfs_rq[4]:/.runnable_load_avg
109 Â 3% -19.5% 87 Â 9% sched_debug.cpu#6.cpu_load[3]
52 Â 9% +19.6% 62 Â 11% sched_debug.cpu#10.cpu_load[2]
64 Â 8% +23.9% 80 Â 6% sched_debug.cpu#4.cpu_load[0]
61 Â 12% +35.2% 82 Â 23% sched_debug.cfs_rq[10]:/.load
103 Â 3% -17.1% 85 Â 6% sched_debug.cpu#6.cpu_load[2]
0.83 Â 3% +16.9% 0.97 Â 8% perf-profile.cpu-cycles._raw_spin_lock_irq.get_signal.do_signal.do_notify_resume.int_signal
0.89 Â 3% +21.3% 1.08 Â 2% perf-profile.cpu-cycles.setup_sigcontext.do_signal.do_notify_resume.int_signal.raise
114 Â 4% -19.0% 92 Â 11% sched_debug.cpu#6.cpu_load[4]
1861 Â 6% -16.2% 1560 Â 8% cpuidle.C3-NHM.usage
410 Â 6% -16.3% 343 Â 9% cpuidle.C1-NHM.usage
3.18 Â 2% +13.8% 3.62 Â 3% perf-profile.cpu-cycles.init_fpu.__restore_xstate_sig.restore_sigcontext.sys_rt_sigreturn.stub_rt_sigreturn
7.29 Â 1% +15.6% 8.43 Â 4% perf-profile.cpu-cycles.__sigqueue_free.part.12.__dequeue_signal.dequeue_signal.get_signal.do_signal
0.92 Â 3% +11.7% 1.03 Â 4% perf-profile.cpu-cycles.system_call.handler
3469 Â 2% -18.2% 2838 Â 15% sched_debug.cpu#1.curr->pid
2.37 Â 1% +13.1% 2.67 Â 4% perf-profile.cpu-cycles.free_uid.__sigqueue_free.__dequeue_signal.dequeue_signal.get_signal
13.43 Â 2% +13.4% 15.23 Â 3% perf-profile.cpu-cycles.get_signal.do_signal.do_notify_resume.int_signal.raise
12.93 Â 3% +12.8% 14.58 Â 4% perf-profile.cpu-cycles.__send_signal.send_signal.do_send_sig_info.do_send_specific.do_tkill
2.67 Â 1% +12.7% 3.01 Â 4% perf-profile.cpu-cycles.memset.init_fpu.__restore_xstate_sig.restore_sigcontext.sys_rt_sigreturn
10.12 Â 2% +11.9% 11.33 Â 3% perf-profile.cpu-cycles.stub_rt_sigreturn.raise
13.75 Â 3% +12.6% 15.49 Â 4% perf-profile.cpu-cycles.send_signal.do_send_sig_info.do_send_specific.do_tkill.sys_tgkill
23.31 Â 2% +12.3% 26.18 Â 3% perf-profile.cpu-cycles.do_notify_resume.int_signal.raise
1.04 Â 5% +14.4% 1.19 Â 5% perf-profile.cpu-cycles.__lock_task_sighand.do_send_sig_info.do_send_specific.do_tkill.sys_tgkill
23.37 Â 2% +12.3% 26.25 Â 3% perf-profile.cpu-cycles.int_signal.raise
22.76 Â 2% +12.2% 25.55 Â 3% perf-profile.cpu-cycles.do_signal.do_notify_resume.int_signal.raise
20.06 Â 3% +11.9% 22.45 Â 4% perf-profile.cpu-cycles.do_send_specific.do_tkill.sys_tgkill.system_call_fastpath.raise
0.81 Â 7% +17.6% 0.95 Â 5% perf-profile.cpu-cycles._raw_spin_unlock_irqrestore.try_to_wake_up.wake_up_state.signal_wake_up_state.complete_signal
16.43 Â 3% +12.1% 18.41 Â 4% perf-profile.cpu-cycles.do_send_sig_info.do_send_specific.do_tkill.sys_tgkill.system_call_fastpath
9.83 Â 2% +11.6% 10.97 Â 3% perf-profile.cpu-cycles.sys_rt_sigreturn.stub_rt_sigreturn.raise
3.44 Â 4% +14.2% 3.93 Â 4% perf-profile.cpu-cycles.signal_wake_up_state.complete_signal.__send_signal.send_signal.do_send_sig_info
96 Â 3% -12.7% 84 Â 3% sched_debug.cpu#6.cpu_load[1]
8.99 Â 2% +13.6% 10.21 Â 3% perf-profile.cpu-cycles.__dequeue_signal.dequeue_signal.get_signal.do_signal.do_notify_resume
10.33 Â 2% +13.0% 11.66 Â 4% perf-profile.cpu-cycles.dequeue_signal.get_signal.do_signal.do_notify_resume.int_signal
22.68 Â 3% +11.7% 25.33 Â 3% perf-profile.cpu-cycles.sys_tgkill.system_call_fastpath.raise
8.64 Â 2% +11.7% 9.66 Â 3% perf-profile.cpu-cycles.restore_sigcontext.sys_rt_sigreturn.stub_rt_sigreturn.raise
22.37 Â 3% +11.6% 24.96 Â 3% perf-profile.cpu-cycles.do_tkill.sys_tgkill.system_call_fastpath.raise
22.87 Â 3% +11.7% 25.53 Â 3% perf-profile.cpu-cycles.system_call_fastpath.raise
8.12 Â 2% +11.9% 9.08 Â 3% perf-profile.cpu-cycles.__restore_xstate_sig.restore_sigcontext.sys_rt_sigreturn.stub_rt_sigreturn.raise
4.11 Â 4% +12.3% 4.61 Â 4% perf-profile.cpu-cycles.complete_signal.__send_signal.send_signal.do_send_sig_info.do_send_specific
1.17 Â 5% +10.7% 1.30 Â 4% perf-profile.cpu-cycles.selinux_task_kill.security_task_kill.check_kill_permission.do_send_specific.do_tkill
5.84 Â 4% +13.6% 6.63 Â 5% perf-profile.cpu-cycles.__sigqueue_alloc.__send_signal.send_signal.do_send_sig_info.do_send_specific
1.32 Â 5% +11.4% 1.47 Â 4% perf-profile.cpu-cycles.security_task_kill.check_kill_permission.do_send_specific.do_tkill.sys_tgkill
wsm: Westmere
Memory: 6G
will-it-scale.time.user_time
40 ++---------------------------------------------------------------------+
*..*..*...*..*..*..*...*..*..* *..*.. .*.. ..*.. |
35 ++ : : *. *. *..*..*...* |
30 ++ : : |
| : : |
25 ++ : : |
O O O O O O O O: : O
20 ++ O O : O : O O O O O O O O O O O |
| : : |
15 ++ : : |
10 ++ : : |
| : : |
5 ++ :: |
| : |
0 ++------------------------------*--------------------------------------+
[*] bisect-good sample
[O] bisect-bad sample
To reproduce:
apt-get install ruby ruby-oj
git clone git://git.kernel.org/pub/scm/linux/kernel/git/wfg/lkp-tests.git
cd lkp-tests
bin/setup-local job.yaml # the job file attached in this email
bin/run-local job.yaml
Disclaimer:
Results have been estimated based on internal Intel analysis and are provided
for informational purposes only. Any difference in system hardware or software
design or configuration may affect actual performance.
Thanks,
Huang, Ying
---
testcase: will-it-scale
default-monitors:
wait: pre-test
uptime:
iostat:
vmstat:
numa-numastat:
numa-vmstat:
numa-meminfo:
proc-vmstat:
proc-stat:
meminfo:
slabinfo:
interrupts:
lock_stat:
latency_stats:
softirqs:
bdi_dev_mapping:
diskstats:
cpuidle:
cpufreq:
turbostat:
sched_debug:
interval: 10
pmeter:
default_watchdogs:
watch-oom:
watchdog:
cpufreq_governor:
- performance
commit: b84235367120b60c3392e774e9725a3fe6a7464d
model: Westmere
memory: 6G
nr_hdd_partitions: 1
hdd_partitions:
swap_partitions:
rootfs_partition:
netconsole_port: 6667
perf-profile:
freq: 800
will-it-scale:
test:
- unlink2
testbox: wsm
tbox_group: wsm
kconfig: x86_64-rhel
enqueue_time: 2015-01-17 05:37:46.268783133 +08:00
head_commit: b84235367120b60c3392e774e9725a3fe6a7464d
base_commit: eaa27f34e91a14cdceed26ed6c6793ec1d186115
branch: linux-devel/devel-hourly-2015011718
kernel: "/kernel/x86_64-rhel/b84235367120b60c3392e774e9725a3fe6a7464d/vmlinuz-3.19.0-rc4-gb842353"
user: lkp
queue: cyclic
rootfs: debian-x86_64-2015-01-13.cgz
result_root: "/result/wsm/will-it-scale/performance-unlink2/debian-x86_64-2015-01-13.cgz/x86_64-rhel/b84235367120b60c3392e774e9725a3fe6a7464d/0"
job_file: "/lkp/scheduled/wsm/cyclic_will-it-scale-performance-unlink2-x86_64-rhel-HEAD-b84235367120b60c3392e774e9725a3fe6a7464d-0.yaml"
dequeue_time: 2015-01-17 21:15:38.970763422 +08:00
nr_cpu: "$(nproc)"
job_state: finished
loadavg: 8.46 4.91 2.02 1/146 6699
start_time: '1421500574'
end_time: '1421500878'
version: "/lkp/lkp/.src-20150116-113525"
./runtest.py unlink2 32 1 6 9 12
_______________________________________________
LKP mailing list
LKP@xxxxxxxxxxxxxxx