| 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: <20201102032524.GI31092@shao2-debian>
Date: Mon, 2 Nov 2020 11:25:24 +0800
From: kernel test robot <rong.a.chen@...el.com>
To: Jason Gunthorpe <jgg@...dia.com>
Cc: linux-kernel@...r.kernel.org,
Andrea Arcangeli <aarcange@...hat.com>,
Andrew Morton <akpm@...ux-foundation.org>,
"Aneesh Kumar K.V" <aneesh.kumar@...ux.ibm.com>,
Christoph Hellwig <hch@....de>,
Hugh Dickins <hughd@...gle.com>, Jan Kara <jack@...e.cz>,
Jann Horn <jannh@...gle.com>,
John Hubbard <jhubbard@...dia.com>,
Kirill Shutemov <kirill@...temov.name>,
Kirill Tkhai <ktkhai@...tuozzo.com>,
Leon Romanovsky <leonro@...dia.com>,
Linux-MM <linux-mm@...ck.org>, Michal Hocko <mhocko@...e.com>,
Oleg Nesterov <oleg@...hat.com>, Peter Xu <peterx@...hat.com>,
Linus Torvalds <torvalds@...ux-foundation.org>,
0day robot <lkp@...el.com>, lkp@...ts.01.org,
ying.huang@...el.com, feng.tang@...el.com, zhengjun.xing@...el.com
Subject: [mm] e498078ae9: will-it-scale.per_thread_ops -1.4% regression
Greeting,
FYI, we noticed a -1.4% regression of will-it-scale.per_thread_ops due to commit:
commit: e498078ae9447c12f6ef1a060639428200bbf29f ("[PATCH 2/2] mm: prevent gup_fast from racing with COW during fork")
url: https://github.com/0day-ci/linux/commits/Jason-Gunthorpe/Add-a-seqcount-between-gup_fast-and-copy_page_range/20201024-082022
base: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git 40a03b750bb3ded71a0f21a0b7dfbf3b24068dcb
in testcase: will-it-scale
on test machine: 192 threads Intel(R) Xeon(R) Platinum 9242 CPU @ 2.30GHz with 192G memory
with following parameters:
nr_task: 100%
mode: thread
test: futex2
cpufreq_governor: performance
ucode: 0x5002f01
test-description: Will It Scale takes a testcase and runs it from 1 through to n parallel copies to see if the testcase will scale. It builds both a process and threads based test in order to see any differences between the two.
test-url: https://github.com/antonblanchard/will-it-scale
If you fix the issue, kindly add following tag
Reported-by: kernel test robot <rong.a.chen@...el.com>
Details are as below:
-------------------------------------------------------------------------------------------------->
To reproduce:
git clone https://github.com/intel/lkp-tests.git
cd lkp-tests
bin/lkp install job.yaml # job file is attached in this email
bin/lkp run job.yaml
=========================================================================================
compiler/cpufreq_governor/kconfig/mode/nr_task/rootfs/tbox_group/test/testcase/ucode:
gcc-9/performance/x86_64-rhel-8.3/thread/100%/debian-10.4-x86_64-20200603.cgz/lkp-csl-2ap2/futex2/will-it-scale/0x5002f01
commit:
9cb7c9fd44 ("mm: reorganize internal_get_user_pages_fast()")
e498078ae9 ("mm: prevent gup_fast from racing with COW during fork")
9cb7c9fd44c983fd e498078ae9447c12f6ef1a06063
---------------- ---------------------------
%stddev %change %stddev
\ | \
2922312 -1.4% 2882609 will-it-scale.per_thread_ops
5.611e+08 -1.4% 5.535e+08 will-it-scale.workload
698642 ± 6% +9.2% 763213 ± 5% numa-meminfo.node2.MemUsed
100.67 ± 65% +64.0% 165.08 ± 2% sched_debug.cfs_rq:/.removed.load_avg.max
16926 ± 28% -44.1% 9461 ± 55% softirqs.NET_RX
34.47 -8.0% 31.71 boot-time.boot
5747 -9.2% 5219 boot-time.idle
221431 ± 4% +11.7% 247354 ± 5% proc-vmstat.numa_pte_updates
1120636 +1.7% 1139742 proc-vmstat.pgalloc_normal
1145 ± 7% +17.3% 1343 ± 8% slabinfo.task_group.active_objs
1145 ± 7% +17.3% 1343 ± 8% slabinfo.task_group.num_objs
4299 ± 15% +33.1% 5722 ± 19% interrupts.CPU101.CAL:Function_call_interrupts
5734 ± 18% +39.5% 7998 ± 31% interrupts.CPU122.CAL:Function_call_interrupts
4635 ± 10% +29.9% 6020 ± 16% interrupts.CPU146.CAL:Function_call_interrupts
345.50 ± 8% +66.6% 575.75 ± 28% interrupts.CPU146.RES:Rescheduling_interrupts
314.50 ± 3% +50.5% 473.25 ± 38% interrupts.CPU153.RES:Rescheduling_interrupts
4399 ± 15% +23.3% 5425 ± 16% interrupts.CPU171.CAL:Function_call_interrupts
470.75 ± 23% -29.3% 333.00 ± 11% interrupts.CPU24.RES:Rescheduling_interrupts
322.50 ± 5% +11.2% 358.50 ± 3% interrupts.CPU55.RES:Rescheduling_interrupts
317.00 ± 3% +7.4% 340.50 ± 3% interrupts.CPU59.RES:Rescheduling_interrupts
308.75 +107.9% 641.75 ± 58% interrupts.CPU7.RES:Rescheduling_interrupts
2883 ± 4% +7.8% 3108 interrupts.CPU92.TLB:TLB_shootdowns
1.64e+08 -3.2% 1.588e+08 perf-stat.i.branch-misses
13403934 ± 35% -35.8% 8604014 ± 3% perf-stat.i.cache-references
0.00 ± 8% -0.0 0.00 ± 14% perf-stat.i.dTLB-load-miss-rate%
1471544 ± 3% -52.6% 698114 ± 4% perf-stat.i.dTLB-load-misses
1.37 ± 17% +31.1% 1.80 ± 3% perf-stat.i.metric.K/sec
0.02 ± 35% -35.3% 0.02 ± 3% perf-stat.overall.MPKI
0.20 -0.0 0.19 perf-stat.overall.branch-miss-rate%
0.00 ± 3% -0.0 0.00 ± 4% perf-stat.overall.dTLB-load-miss-rate%
0.00 -0.0 0.00 perf-stat.overall.dTLB-store-miss-rate%
1.635e+08 -3.2% 1.582e+08 perf-stat.ps.branch-misses
13412014 ± 35% -35.6% 8635337 ± 3% perf-stat.ps.cache-references
1480248 ± 4% -52.2% 707692 ± 4% perf-stat.ps.dTLB-load-misses
23.09 -0.4 22.69 perf-profile.calltrace.cycles-pp.gup_pgd_range.internal_get_user_pages_fast.get_futex_key.futex_wait_setup.futex_wait
14.33 -0.2 14.15 perf-profile.calltrace.cycles-pp.__entry_text_start.syscall
2.29 -0.1 2.19 perf-profile.calltrace.cycles-pp.hash_futex.futex_wait_setup.futex_wait.do_futex.__x64_sys_futex
2.97 -0.1 2.88 perf-profile.calltrace.cycles-pp._raw_spin_lock.futex_wait_setup.futex_wait.do_futex.__x64_sys_futex
3.95 -0.1 3.88 perf-profile.calltrace.cycles-pp.try_grab_compound_head.gup_pgd_range.internal_get_user_pages_fast.get_futex_key.futex_wait_setup
1.81 -0.0 1.78 perf-profile.calltrace.cycles-pp.get_user_pages_fast.get_futex_key.futex_wait_setup.futex_wait.do_futex
1.96 -0.0 1.93 perf-profile.calltrace.cycles-pp.syscall_enter_from_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
1.14 +0.1 1.22 perf-profile.calltrace.cycles-pp.exit_to_user_mode_prepare.syscall_exit_to_user_mode.entry_SYSCALL_64_after_hwframe.syscall
70.66 +0.2 70.82 perf-profile.calltrace.cycles-pp.__x64_sys_futex.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
31.58 +0.6 32.17 perf-profile.calltrace.cycles-pp.internal_get_user_pages_fast.get_futex_key.futex_wait_setup.futex_wait.do_futex
40.59 +0.6 41.19 perf-profile.calltrace.cycles-pp.get_futex_key.futex_wait_setup.futex_wait.do_futex.__x64_sys_futex
23.37 -0.4 22.94 perf-profile.children.cycles-pp.gup_pgd_range
2.38 -0.1 2.28 perf-profile.children.cycles-pp.hash_futex
9.24 -0.1 9.15 perf-profile.children.cycles-pp.__entry_text_start
3.07 -0.1 2.99 perf-profile.children.cycles-pp._raw_spin_lock
3.95 -0.1 3.89 perf-profile.children.cycles-pp.try_grab_compound_head
2.08 -0.0 2.04 perf-profile.children.cycles-pp.syscall_enter_from_user_mode
0.06 ± 7% +0.0 0.07 ± 5% perf-profile.children.cycles-pp.perf_mux_hrtimer_handler
0.40 +0.0 0.45 perf-profile.children.cycles-pp.is_valid_gup_flags
0.00 +0.1 0.05 perf-profile.children.cycles-pp.irqtime_account_irq
0.00 +0.1 0.05 perf-profile.children.cycles-pp.irq_exit_rcu
1.15 +0.1 1.24 perf-profile.children.cycles-pp.exit_to_user_mode_prepare
70.86 +0.2 71.02 perf-profile.children.cycles-pp.__x64_sys_futex
40.93 +0.5 41.47 perf-profile.children.cycles-pp.get_futex_key
31.73 +0.7 32.40 perf-profile.children.cycles-pp.internal_get_user_pages_fast
19.00 -0.4 18.59 perf-profile.self.cycles-pp.gup_pgd_range
7.76 ± 2% -0.2 7.52 perf-profile.self.cycles-pp.futex_wait_setup
7.45 -0.2 7.21 perf-profile.self.cycles-pp.get_futex_key
6.89 -0.1 6.77 perf-profile.self.cycles-pp.futex_wait
2.25 -0.1 2.16 perf-profile.self.cycles-pp.hash_futex
3.04 -0.1 2.96 perf-profile.self.cycles-pp._raw_spin_lock
1.95 -0.0 1.91 perf-profile.self.cycles-pp.syscall_enter_from_user_mode
0.28 ± 2% +0.0 0.30 perf-profile.self.cycles-pp.is_valid_gup_flags
1.30 +0.0 1.35 perf-profile.self.cycles-pp.get_user_pages_fast
1.05 +0.1 1.13 perf-profile.self.cycles-pp.exit_to_user_mode_prepare
1.98 +0.1 2.07 perf-profile.self.cycles-pp.do_futex
8.14 +1.1 9.25 perf-profile.self.cycles-pp.internal_get_user_pages_fast
will-it-scale.per_thread_ops
3e+06 +-----------------------------------------------------------------+
|OOO OOOOOO +OOO++ OO+O+++.++++++.+ ++++.+++++++ + + ++ +.+++|
2.8e+06 |-+ |
2.6e+06 |-+ |
| |
2.4e+06 |-+ |
2.2e+06 |-+ |
| |
2e+06 |-+ |
1.8e+06 |-+ |
| |
1.6e+06 |-+ |
1.4e+06 |-+ |
| |
1.2e+06 +-----------------------------------------------------------------+
[*] bisect-good sample
[O] bisect-bad sample
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,
Rong Chen
View attachment "config-5.9.0-13461-ge498078ae944" of type "text/plain" (171579 bytes)
View attachment "job-script" of type "text/plain" (7975 bytes)
View attachment "job.yaml" of type "text/plain" (5177 bytes)
View attachment "reproduce" of type "text/plain" (338 bytes)
Powered by blists - more mailing lists