All of lore.kernel.org
 help / color / mirror / Atom feed
From: kernel test robot <oliver.sang@intel.com>
To: Herbert Xu <herbert@gondor.apana.org.au>
Cc: <oe-lkp@lists.linux.dev>, <lkp@intel.com>,
	<linux-crypto@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
	<oliver.sang@intel.com>
Subject: [linux-next:master] [padata]  71203f68c7:  unixbench.throughput 3.1% regression
Date: Thu, 26 Jun 2025 10:57:15 +0800	[thread overview]
Message-ID: <202506261012.11b518e7-lkp@intel.com> (raw)



Hello,


normally we won't report performance results if we suspect it is caused by
alignment problems.

since this patch touches the code related with alignment:

-       struct work_struct              reorder_work;
-       spinlock_t                      ____cacheline_aligned lock;

we still make out below report FYI what's the possible performance impact.


kernel test robot noticed a 3.1% regression of unixbench.throughput on:


commit: 71203f68c7749609d7fc8ae6ad054bdedeb24f91 ("padata: Fix pd UAF once and for all")
https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master

[test failed on linux-next/master 1b152eeca84a02bdb648f16b82ef3394007a9dcf]

testcase: unixbench
config: x86_64-rhel-9.4
compiler: gcc-12
test machine: 64 threads 2 sockets Intel(R) Xeon(R) Gold 6346 CPU @ 3.10GHz (Ice Lake) with 256G memory
parameters:

	runtime: 300s
	nr_task: 100%
	test: fsbuffer-w
	cpufreq_governor: performance


In addition to that, the commit also has significant impact on the following tests:

+------------------+--------------------------------------------------------------+
| testcase: change | will-it-scale: will-it-scale.per_thread_ops 1.1% improvement |
| test machine     | 104 threads 2 sockets (Skylake) with 192G memory             |
| test parameters  | cpufreq_governor=performance                                 |
|                  | mode=thread                                                  |
|                  | nr_task=100%                                                 |
|                  | test=pwrite1                                                 |
+------------------+--------------------------------------------------------------+


If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <oliver.sang@intel.com>
| Closes: https://lore.kernel.org/oe-lkp/202506261012.11b518e7-lkp@intel.com


Details are as below:
-------------------------------------------------------------------------------------------------->


The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20250626/202506261012.11b518e7-lkp@intel.com

=========================================================================================
compiler/cpufreq_governor/kconfig/nr_task/rootfs/runtime/tbox_group/test/testcase:
  gcc-12/performance/x86_64-rhel-9.4/100%/debian-12-x86_64-20240206.cgz/300s/lkp-icl-2sp9/fsbuffer-w/unixbench

commit: 
  73c2437109 ("crypto: s390/sha3 - Use cpu byte-order when exporting")
  71203f68c7 ("padata: Fix pd UAF once and for all")

73c2437109c3eab2 71203f68c7749609d7fc8ae6ad0 
---------------- --------------------------- 
         %stddev     %change         %stddev
             \          |                \  
    111306            +2.0%     113530        proc-vmstat.pgreuse
      0.01 ±  4%     +14.9%       0.01        perf-sched.sch_delay.max.ms.schedule_hrtimeout_range_clock.ep_poll.do_epoll_wait.__x64_sys_epoll_wait
    715.14 ±167%     -99.4%       3.99 ± 80%  perf-sched.total_sch_delay.max.ms
  33278354            -3.1%   32249523        unixbench.throughput
      1854            -4.8%       1765        unixbench.time.user_time
 1.233e+10            -3.1%  1.195e+10        unixbench.workload
 4.717e+10            -3.1%  4.573e+10        perf-stat.i.branch-instructions
      0.42            -0.0        0.41        perf-stat.i.branch-miss-rate%
  28489209 ±  2%     -10.9%   25397034        perf-stat.i.branch-misses
      0.97            +2.8%       1.00        perf-stat.i.cpi
 1.946e+11            -3.1%  1.886e+11        perf-stat.i.instructions
      1.05            -2.8%       1.02        perf-stat.i.ipc
      0.06 ±  2%      -0.0        0.06        perf-stat.overall.branch-miss-rate%
      0.94            +3.2%       0.97        perf-stat.overall.cpi
      1.06            -3.1%       1.03        perf-stat.overall.ipc
 4.706e+10            -3.1%  4.562e+10        perf-stat.ps.branch-instructions
  28421825 ±  2%     -10.9%   25336865        perf-stat.ps.branch-misses
 1.942e+11            -3.1%  1.882e+11        perf-stat.ps.instructions
 7.212e+13            -3.1%  6.991e+13        perf-stat.total.instructions


***************************************************************************************************
lkp-skl-fpga01: 104 threads 2 sockets (Skylake) with 192G memory
=========================================================================================
compiler/cpufreq_governor/kconfig/mode/nr_task/rootfs/tbox_group/test/testcase:
  gcc-12/performance/x86_64-rhel-9.4/thread/100%/debian-12-x86_64-20240206.cgz/lkp-skl-fpga01/pwrite1/will-it-scale

commit: 
  73c2437109 ("crypto: s390/sha3 - Use cpu byte-order when exporting")
  71203f68c7 ("padata: Fix pd UAF once and for all")

73c2437109c3eab2 71203f68c7749609d7fc8ae6ad0 
---------------- --------------------------- 
         %stddev     %change         %stddev
             \          |                \  
    997120            +1.5%    1011812        proc-vmstat.pgfree
  55606929            +1.1%   56223715        will-it-scale.104.threads
    534681            +1.1%     540612        will-it-scale.per_thread_ops
  55606929            +1.1%   56223715        will-it-scale.workload
      0.01 ± 34%     +63.9%       0.02 ± 31%  perf-sched.sch_delay.avg.ms.anon_pipe_read.fifo_pipe_read.vfs_read.ksys_read
    233.78 ±143%    +242.3%     800.28        perf-sched.wait_and_delay.avg.ms.__x64_sys_pause.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
     74.68 ±  6%     +18.5%      88.49 ±  7%  perf-sched.wait_and_delay.avg.ms.anon_pipe_read.fifo_pipe_read.vfs_read.ksys_read
      4.69 ± 44%     -84.7%       0.72 ± 30%  perf-sched.wait_and_delay.avg.ms.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
    576.00 ±  9%     -17.7%     473.83 ±  7%  perf-sched.wait_and_delay.count.anon_pipe_read.fifo_pipe_read.vfs_read.ksys_read
    999.68           -98.9%      11.47 ± 85%  perf-sched.wait_and_delay.max.ms.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
    234.63 ±142%    +240.8%     799.73        perf-sched.wait_time.avg.ms.__x64_sys_pause.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
     74.67 ±  6%     +18.5%      88.46 ±  7%  perf-sched.wait_time.avg.ms.anon_pipe_read.fifo_pipe_read.vfs_read.ksys_read
      4.31 ± 48%     -91.6%       0.36 ± 29%  perf-sched.wait_time.avg.ms.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
    999.37           -99.3%       6.65 ± 67%  perf-sched.wait_time.max.ms.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
 1.686e+10            +1.1%  1.704e+10        perf-stat.i.branch-instructions
      1.62            +0.0        1.67        perf-stat.i.branch-miss-rate%
 2.726e+08            +4.1%  2.836e+08        perf-stat.i.branch-misses
      3.36            -1.0%       3.32        perf-stat.i.cpi
 8.562e+10            +1.1%  8.656e+10        perf-stat.i.instructions
      1.62            +0.0        1.66        perf-stat.overall.branch-miss-rate%
      3.36            -1.0%       3.33        perf-stat.overall.cpi
  1.68e+10            +1.1%  1.698e+10        perf-stat.ps.branch-instructions
 2.717e+08            +4.1%  2.827e+08        perf-stat.ps.branch-misses
 8.533e+10            +1.1%  8.627e+10        perf-stat.ps.instructions
 2.578e+13            +1.1%  2.607e+13        perf-stat.total.instructions
      4.44 ±  3%      -0.9        3.51 ±  5%  perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_safe_stack.__libc_pwrite
     11.68            -0.3       11.41        perf-profile.calltrace.cycles-pp.copy_folio_from_iter_atomic.generic_perform_write.shmem_file_write_iter.vfs_write.__x64_sys_pwrite64
      0.85 ±  8%      -0.2        0.70 ±  3%  perf-profile.calltrace.cycles-pp.file_remove_privs_flags.shmem_file_write_iter.vfs_write.__x64_sys_pwrite64.do_syscall_64
      0.81 ±  3%      -0.1        0.67        perf-profile.calltrace.cycles-pp.balance_dirty_pages_ratelimited_flags.generic_perform_write.shmem_file_write_iter.vfs_write.__x64_sys_pwrite64
      2.26            -0.1        2.13        perf-profile.calltrace.cycles-pp.fdget.__x64_sys_pwrite64.do_syscall_64.entry_SYSCALL_64_after_hwframe.__libc_pwrite
      0.94            -0.0        0.90 ±  2%  perf-profile.calltrace.cycles-pp.noop_dirty_folio.shmem_write_end.generic_perform_write.shmem_file_write_iter.vfs_write
      1.14            +0.1        1.19 ±  2%  perf-profile.calltrace.cycles-pp.folio_mark_dirty.shmem_write_end.generic_perform_write.shmem_file_write_iter.vfs_write
      2.18 ±  2%      +0.4        2.58 ± 10%  perf-profile.calltrace.cycles-pp.filemap_get_entry.shmem_get_folio_gfp.shmem_write_begin.generic_perform_write.shmem_file_write_iter
      2.58 ±  3%      -0.5        2.07 ±  4%  perf-profile.children.cycles-pp.entry_SYSCALL_64_safe_stack
      7.87            -0.5        7.38        perf-profile.children.cycles-pp.entry_SYSCALL_64
     11.90            -0.3       11.62        perf-profile.children.cycles-pp.copy_folio_from_iter_atomic
      9.13            -0.3        8.87        perf-profile.children.cycles-pp.rep_movs_alternative
      0.87 ±  8%      -0.1        0.72 ±  3%  perf-profile.children.cycles-pp.file_remove_privs_flags
      0.84 ±  3%      -0.1        0.71        perf-profile.children.cycles-pp.balance_dirty_pages_ratelimited_flags
      2.26            -0.1        2.13        perf-profile.children.cycles-pp.fdget
      1.01            -0.0        0.96        perf-profile.children.cycles-pp.noop_dirty_folio
      0.43 ±  2%      -0.0        0.39 ±  3%  perf-profile.children.cycles-pp.rcu_all_qs
      0.29 ±  3%      -0.0        0.26        perf-profile.children.cycles-pp.inode_to_bdi
      0.30            -0.0        0.27        perf-profile.children.cycles-pp.x64_sys_call
      0.35 ±  2%      -0.0        0.32        perf-profile.children.cycles-pp.rw_verify_area
      0.39 ±  3%      +0.2        0.58 ± 24%  perf-profile.children.cycles-pp.xas_load
      2.20 ±  2%      +0.4        2.61 ± 10%  perf-profile.children.cycles-pp.filemap_get_entry
      6.97            -0.5        6.48        perf-profile.self.cycles-pp.entry_SYSCALL_64
      2.14 ±  3%      -0.4        1.71 ±  2%  perf-profile.self.cycles-pp.shmem_get_folio_gfp
      2.60            -0.3        2.28 ±  4%  perf-profile.self.cycles-pp.do_syscall_64
      8.90            -0.2        8.65        perf-profile.self.cycles-pp.rep_movs_alternative
      2.28            -0.2        2.10        perf-profile.self.cycles-pp.shmem_write_end
      0.86 ±  8%      -0.1        0.71 ±  3%  perf-profile.self.cycles-pp.file_remove_privs_flags
      2.24            -0.1        2.11        perf-profile.self.cycles-pp.fdget
      0.58 ±  4%      -0.1        0.48        perf-profile.self.cycles-pp.balance_dirty_pages_ratelimited_flags
      0.70 ±  3%      -0.1        0.62 ±  3%  perf-profile.self.cycles-pp.entry_SYSCALL_64_safe_stack
      0.61 ±  2%      -0.1        0.55        perf-profile.self.cycles-pp.generic_write_checks
      0.34 ±  2%      -0.0        0.31 ±  3%  perf-profile.self.cycles-pp.rcu_all_qs
      0.28            -0.0        0.25        perf-profile.self.cycles-pp.x64_sys_call
      0.25 ±  4%      -0.0        0.22        perf-profile.self.cycles-pp.inode_to_bdi
      0.24 ±  3%      -0.0        0.21        perf-profile.self.cycles-pp.rw_verify_area
      0.77 ±  2%      +0.0        0.82 ±  2%  perf-profile.self.cycles-pp.folio_mark_dirty
      0.72            +0.1        0.78 ±  2%  perf-profile.self.cycles-pp.current_time
      0.20 ±  3%      +0.2        0.39 ± 31%  perf-profile.self.cycles-pp.xas_load
      1.80 ±  2%      +0.2        2.02 ±  6%  perf-profile.self.cycles-pp.filemap_get_entry
      9.38            +0.2        9.61        perf-profile.self.cycles-pp.entry_SYSCALL_64_after_hwframe
      2.18 ±  2%      +0.3        2.50 ±  5%  perf-profile.self.cycles-pp.shmem_write_begin
      2.54            +0.8        3.37 ±  4%  perf-profile.self.cycles-pp.__libc_pwrite





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.


-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki


                 reply	other threads:[~2025-06-26  2:57 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=202506261012.11b518e7-lkp@intel.com \
    --to=oliver.sang@intel.com \
    --cc=herbert@gondor.apana.org.au \
    --cc=linux-crypto@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lkp@intel.com \
    --cc=oe-lkp@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.