| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20161018025900.GE6107@yexl-desktop>
Date: Tue, 18 Oct 2016 10:59:00 +0800
From: kernel test robot <xiaolong.ye@...el.com>
To: Jens Axboe <axboe@...com>
Cc: LKML <linux-kernel@...r.kernel.org>, Jens Axboe <axboe@...com>,
Jens Axboe <axboe@...nel.dk>, lkp@...org
Subject: [lkp] [writeback] 1a2af575cf: fio.latency_10us% -20.4% improvement
FYI, we noticed a -20.4% improvement of fio.latency_10us% due to commit:
commit 1a2af575cf12136298f751b656adc2fa8456f804 ("writeback: throttle buffered writeback")
https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux-block.git wb-buf-throttle
in testcase: fio-basic
on test machine: 16 threads Intel(R) Xeon(R) CPU D-1541 @ 2.10GHz with 8G memory
with following parameters:
runtime: 300s
nr_task: 8
disk: 1SSD
fs: ext4
rw: write
bs: 4k
ioengine: sync
test_size: 400g
cpufreq_governor: performance
Fio is a tool that will spawn a number of threads or processes doing a particular type of I/O action as specified by the user.
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.
Details are as below:
-------------------------------------------------------------------------------------------------->
To reproduce:
git clone git://git.kernel.org/pub/scm/linux/kernel/git/wfg/lkp-tests.git
cd lkp-tests
bin/lkp install job.yaml # job file is attached in this email
bin/lkp run job.yaml
=========================================================================================
bs/compiler/cpufreq_governor/disk/fs/ioengine/kconfig/nr_task/rootfs/runtime/rw/tbox_group/test_size/testcase:
4k/gcc-6/performance/1SSD/ext4/sync/x86_64-rhel-7.2/8/debian-x86_64-2016-08-31.cgz/300s/write/lkp-bdw-de1/400g/fio-basic
commit:
1513fc26a0 ("wbt: add general throttling mechanism")
1a2af575cf ("writeback: throttle buffered writeback")
1513fc26a0f3c20e 1a2af575cf12136298f751b656
---------------- --------------------------
%stddev %change %stddev
\ | \
4.38 ± 3% -20.4% 3.49 ± 3% fio.latency_10us%
0.01 ± 0% -100.0% 0.00 ± -1% fio.latency_250ms%
0.01 ± 0% -100.0% 0.00 ± -1% fio.latency_500ms%
91086 ± 0% +2.3% 93150 ± 0% fio.time.voluntary_context_switches
4.00 ± 0% -25.0% 3.00 ± 0% fio.write_clat_95%_us
1770 ± 0% -9.5% 1603 ± 0% fio.write_clat_stddev
55001 ± 0% -1.5% 54177 ± 0% interrupts.CAL:Function_call_interrupts
1135308 ± 0% +10.7% 1256908 ± 0% meminfo.Dirty
137881 ± 0% -87.6% 17158 ± 1% meminfo.Writeback
97812142 ± 6% +28.1% 1.253e+08 ± 9% cpuidle.C1E-BDW.time
178531 ± 3% +17.6% 210005 ± 5% cpuidle.C1E-BDW.usage
71044513 ± 7% -22.4% 55107352 ± 10% cpuidle.POLL.time
6.45 ± 1% -4.2% 6.18 ± 1% turbostat.%Busy
140.25 ± 2% -6.4% 131.25 ± 2% turbostat.Avg_MHz
16.97 ± 6% +12.3% 19.07 ± 5% turbostat.CPU%c6
135.51 ± 0% -88.2% 16.03 ± 0% iostat.sda.avgqu-sz
300.94 ± 0% -88.1% 35.87 ± 0% iostat.sda.await
301.11 ± 0% -88.1% 35.97 ± 0% iostat.sda.w_await
27.90 ± 0% -2.1% 27.32 ± 0% iostat.sda.wrqm/s
283778 ± 0% +10.7% 314221 ± 0% proc-vmstat.nr_dirty
34550 ± 0% -87.5% 4308 ± 0% proc-vmstat.nr_writeback
123061 ± 37% -98.0% 2515 ± 35% proc-vmstat.pgscan_direct
123061 ± 37% -98.0% 2515 ± 35% proc-vmstat.pgsteal_direct
841.25 ± 3% -27.5% 610.00 ± 7% slabinfo.blkdev_requests.active_objs
880.00 ± 2% -30.7% 610.00 ± 7% slabinfo.blkdev_requests.num_objs
514.25 ± 2% -25.3% 384.25 ± 4% slabinfo.kmalloc-4096.active_objs
531.25 ± 2% -25.5% 396.00 ± 4% slabinfo.kmalloc-4096.num_objs
277718 ± 0% -91.7% 23019 ± 3% latency_stats.avg.do_get_write_access.jbd2_journal_get_write_access.__ext4_journal_get_write_access.ext4_reserve_inode_write.ext4_mark_inode_dirty.ext4_dirty_inode.__mark_inode_dirty.generic_update_time.file_update_time.__generic_file_write_iter.ext4_file_write_iter.__vfs_write
277718 ± 0% -86.7% 36903 ± 0% latency_stats.avg.max
26918 ± 5% -66.1% 9126 ± 3% latency_stats.avg.wait_transaction_locked.add_transaction_credits.start_this_handle.jbd2__journal_start.__ext4_journal_start_sb.ext4_dirty_inode.__mark_inode_dirty.generic_update_time.file_update_time.__generic_file_write_iter.ext4_file_write_iter.__vfs_write
417291 ± 10% -87.8% 50997 ± 6% latency_stats.max.do_get_write_access.jbd2_journal_get_write_access.__ext4_journal_get_write_access.ext4_reserve_inode_write.ext4_mark_inode_dirty.ext4_dirty_inode.__mark_inode_dirty.generic_update_time.file_update_time.__generic_file_write_iter.ext4_file_write_iter.__vfs_write
417291 ± 10% -85.9% 58843 ± 1% latency_stats.max.max
69764347 ± 6% -89.7% 7217067 ± 8% latency_stats.sum.do_get_write_access.jbd2_journal_get_write_access.__ext4_journal_get_write_access.ext4_reserve_inode_write.ext4_mark_inode_dirty.ext4_dirty_inode.__mark_inode_dirty.generic_update_time.file_update_time.__generic_file_write_iter.ext4_file_write_iter.__vfs_write
112945 ± 36% -84.0% 18037 ±109% latency_stats.sum.lru_add_drain_all.SyS_fadvise64.entry_SYSCALL_64_fastpath
6133127 ± 8% -79.0% 1286420 ± 7% latency_stats.sum.wait_transaction_locked.add_transaction_credits.start_this_handle.jbd2__journal_start.__ext4_journal_start_sb.ext4_dirty_inode.__mark_inode_dirty.generic_update_time.file_update_time.__generic_file_write_iter.ext4_file_write_iter.__vfs_write
447.79 ± 3% +33.5% 597.67 ± 8% sched_debug.cfs_rq:/.load_avg.max
49.21 ± 8% +67.5% 82.41 ± 20% sched_debug.cfs_rq:/.load_avg.stddev
536913 ± 3% +10.3% 592288 ± 6% sched_debug.cpu.avg_idle.min
93.46 ± 22% -40.2% 55.92 ± 13% sched_debug.cpu.cpu_load[2].max
22.64 ± 22% -38.9% 13.83 ± 12% sched_debug.cpu.cpu_load[2].stddev
83.75 ± 21% -35.7% 53.88 ± 7% sched_debug.cpu.cpu_load[3].max
20.09 ± 21% -34.7% 13.11 ± 6% sched_debug.cpu.cpu_load[3].stddev
8881 ± 3% +37.6% 12220 ± 1% sched_debug.cpu.ttwu_local.max
1306 ± 17% +42.3% 1859 ± 6% sched_debug.cpu.ttwu_local.stddev
0.32 ± 2% +8.1% 0.35 ± 2% perf-stat.branch-miss-rate%
3.346e+08 ± 2% +4.7% 3.503e+08 ± 2% perf-stat.branch-misses
5.84e+09 ± 0% -1.9% 5.73e+09 ± 0% perf-stat.cache-misses
5.84e+09 ± 0% -1.9% 5.73e+09 ± 0% perf-stat.cache-references
6.502e+11 ± 2% -7.9% 5.987e+11 ± 2% perf-stat.cpu-cycles
0.12 ± 2% +3.2% 0.12 ± 0% perf-stat.dTLB-load-miss-rate%
1.712e+08 ± 0% +2.3% 1.751e+08 ± 0% perf-stat.dTLB-load-misses
21716662 ± 0% +2.7% 22312790 ± 0% perf-stat.dTLB-store-misses
64.73 ± 0% -2.4% 63.14 ± 0% perf-stat.iTLB-load-miss-rate%
43799790 ± 0% -1.6% 43082890 ± 0% perf-stat.iTLB-load-misses
23870161 ± 0% +5.4% 25151226 ± 2% perf-stat.iTLB-loads
0.82 ± 1% +5.9% 0.87 ± 2% perf-stat.ipc
0.86 ± 10% +30.8% 1.12 ± 10% perf-profile.calltrace.cycles-pp.__add_to_page_cache_locked.add_to_page_cache_lru.pagecache_get_page.grab_cache_page_write_begin.ext4_da_write_begin
2.28 ± 1% +27.1% 2.90 ± 4% perf-profile.calltrace.cycles-pp.__block_commit_write.block_write_end.generic_write_end.ext4_da_write_end.generic_perform_write
0.78 ± 0% +26.4% 0.98 ± 6% perf-profile.calltrace.cycles-pp.__set_page_dirty.mark_buffer_dirty.__block_commit_write.block_write_end.generic_write_end
2.35 ± 2% +27.8% 3.00 ± 3% perf-profile.calltrace.cycles-pp.block_write_end.generic_write_end.ext4_da_write_end.generic_perform_write.__generic_file_write_iter
20.47 ± 11% -28.3% 14.68 ± 6% perf-profile.calltrace.cycles-pp.call_cpuidle.cpu_startup_entry.rest_init.start_kernel.x86_64_start_reservations
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.calltrace.cycles-pp.cpu_startup_entry.rest_init.start_kernel.x86_64_start_reservations.x86_64_start_kernel
20.47 ± 11% -28.3% 14.68 ± 6% perf-profile.calltrace.cycles-pp.cpuidle_enter.call_cpuidle.cpu_startup_entry.rest_init.start_kernel
17.98 ± 10% -32.7% 12.11 ± 7% perf-profile.calltrace.cycles-pp.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry.rest_init
3.60 ± 3% +17.6% 4.23 ± 4% perf-profile.calltrace.cycles-pp.ext4_da_write_end.generic_perform_write.__generic_file_write_iter.ext4_file_write_iter.__vfs_write
2.91 ± 3% +21.2% 3.52 ± 2% perf-profile.calltrace.cycles-pp.generic_write_end.ext4_da_write_end.generic_perform_write.__generic_file_write_iter.ext4_file_write_iter
1.32 ± 3% +27.0% 1.68 ± 4% perf-profile.calltrace.cycles-pp.mark_buffer_dirty.__block_commit_write.block_write_end.generic_write_end.ext4_da_write_end
19.44 ± 15% -30.1% 13.59 ± 14% perf-profile.calltrace.cycles-pp.poll_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.calltrace.cycles-pp.rest_init.start_kernel.x86_64_start_reservations.x86_64_start_kernel
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.calltrace.cycles-pp.start_kernel.x86_64_start_reservations.x86_64_start_kernel
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.calltrace.cycles-pp.x86_64_start_kernel
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.calltrace.cycles-pp.x86_64_start_reservations.x86_64_start_kernel
0.86 ± 10% +30.8% 1.12 ± 10% perf-profile.children.cycles-pp.__add_to_page_cache_locked
2.29 ± 1% +27.1% 2.91 ± 3% perf-profile.children.cycles-pp.__block_commit_write
0.79 ± 2% +25.0% 0.99 ± 6% perf-profile.children.cycles-pp.__set_page_dirty
3.54 ± 5% +14.3% 4.05 ± 8% perf-profile.children.cycles-pp.apic_timer_interrupt
2.38 ± 3% +26.8% 3.01 ± 4% perf-profile.children.cycles-pp.block_write_end
3.60 ± 3% +17.7% 4.23 ± 4% perf-profile.children.cycles-pp.ext4_da_write_end
2.94 ± 3% +20.3% 3.54 ± 2% perf-profile.children.cycles-pp.generic_write_end
1.32 ± 3% +27.6% 1.69 ± 4% perf-profile.children.cycles-pp.mark_buffer_dirty
19.44 ± 15% -30.1% 13.59 ± 14% perf-profile.children.cycles-pp.poll_idle
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.children.cycles-pp.rest_init
3.43 ± 5% +14.3% 3.92 ± 8% perf-profile.children.cycles-pp.smp_apic_timer_interrupt
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.children.cycles-pp.start_kernel
0.86 ± 16% +19.2% 1.03 ± 3% perf-profile.children.cycles-pp.tick_nohz_irq_exit
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.children.cycles-pp.x86_64_start_kernel
20.54 ± 11% -28.1% 14.77 ± 7% perf-profile.children.cycles-pp.x86_64_start_reservations
0.94 ± 4% +25.5% 1.18 ± 9% perf-profile.self.cycles-pp.__block_commit_write
19.44 ± 15% -30.1% 13.59 ± 14% perf-profile.self.cycles-pp.poll_idle
Thanks,
Xiaolong
View attachment "config-4.8.0-14612-g1a2af57" of type "text/plain" (153678 bytes)
View attachment "job-script" of type "text/plain" (6995 bytes)
View attachment "job.yaml" of type "text/plain" (4557 bytes)
View attachment "reproduce" of type "text/plain" (431 bytes)
Powered by blists - more mailing lists