On Wed, 2024-10-09 at 10:53 +0800, kernel test robot wrote:
> hi, Jeff Layton,
>
> we reported
> "[brauner-vfs:vfs.mgtime] [fs] a037d5e7f8: will-it-scale.per_thread_ops -5.5% regression"
> for this commit about one month ago.
>
> we also saw you sent out patch for it.
>
> now we noticed the commit is merged into linux-next/master now. besides
> will-it-scale, we also captured a hackbench regression. so we report this again
> FYI what we observed in our tests. thanks
>
>
>
> Hello,
>
> kernel test robot noticed a -5.7% regression of will-it-scale.per_thread_ops on:
>
This is consistent with the results I was getting in my own testing.
The multigrain series does cause a small performance hit in this sort
of microbenchmark as fetching the floor and manipulating it has a cost.
I don't think there is much we can do to mitigate that unfortunately.
>
> commit: 2e4c6e78f41afefb7a2b825b7aa4d90070720992 ("fs: add infrastructure for multigrain timestamps")
> https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master
>
> testcase: will-it-scale
> test machine: 224 threads 4 sockets Intel(R) Xeon(R) Platinum 8380H CPU @ 2.90GHz (Cooper Lake) with 192G memory
> parameters:
>
> nr_task: 100%
> mode: thread
> test: pipe1
> cpufreq_governor: performance
>
>
> In addition to that, the commit also has significant impact on the following tests:
>
> +------------------+--------------------------------------------------------------------------------------------+
> > testcase: change | hackbench: hackbench.throughput -4.5% regression |
> > test machine | 128 threads 2 sockets Intel(R) Xeon(R) Gold 6338 CPU @ 2.00GHz (Ice Lake) with 256G memory |
> > test parameters | cpufreq_governor=performance |
> > | ipc=pipe |
> > | iterations=4 |
> > | mode=threads |
> > | nr_threads=800% |
> +------------------+--------------------------------------------------------------------------------------------+
> > testcase: change | will-it-scale: will-it-scale.per_process_ops -2.0% regression |
> > test machine | 104 threads 2 sockets (Skylake) with 192G memory |
> > test parameters | cpufreq_governor=performance |
> > | mode=process |
> > | nr_task=100% |
> > | test=pipe1 |
> +------------------+--------------------------------------------------------------------------------------------+
>
>
> 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@xxxxxxxxx>
> > Closes: https://lore.kernel.org/oe-lkp/202410091041.6f5d221e-oliver.sang@xxxxxxxxx
>
>
> Details are as below:
> -------------------------------------------------------------------------------------------------->
>
>
> The kernel config and materials to reproduce are available at:
> https://download.01.org/0day-ci/archive/20241009/202410091041.6f5d221e-oliver.sang@xxxxxxxxx
>
> =========================================================================================
> compiler/cpufreq_governor/kconfig/mode/nr_task/rootfs/tbox_group/test/testcase:
> gcc-12/performance/x86_64-rhel-8.3/thread/100%/debian-12-x86_64-20240206.cgz/lkp-cpl-4sp2/pipe1/will-it-scale
>
> commit:
> v6.12-rc1
> 2e4c6e78f4 ("fs: add infrastructure for multigrain timestamps")
>
> v6.12-rc1 2e4c6e78f41afefb7a2b825b7aa
> ---------------- ---------------------------
> %stddev %change %stddev
> \ | \
> 806865 ± 12% +127.6% 1836795 ± 68% numa-meminfo.node3.FilePages
> 32494 ± 7% +39.2% 45235 ± 25% numa-meminfo.node3.Mapped
> 201722 ± 12% +127.7% 459227 ± 68% numa-vmstat.node3.nr_file_pages
> 8032 ± 7% +38.6% 11136 ± 26% numa-vmstat.node3.nr_mapped
> 2657388 ± 13% -28.2% 1907049 ± 11% sched_debug.cfs_rq:/.avg_vruntime.stddev
> 2657388 ± 13% -28.2% 1907049 ± 11% sched_debug.cfs_rq:/.min_vruntime.stddev
> 2.921e+08 -5.7% 2.754e+08 will-it-scale.224.threads
> 1303879 -5.7% 1229301 will-it-scale.per_thread_ops
> 2.921e+08 -5.7% 2.754e+08 will-it-scale.workload
> 210109 +1.5% 213268 proc-vmstat.nr_active_anon
> 222111 +1.5% 225492 proc-vmstat.nr_shmem
> 210109 +1.5% 213268 proc-vmstat.nr_zone_active_anon
> 164529 +1.6% 167080 proc-vmstat.pgactivate
> 1.52 ± 82% -78.7% 0.32 ± 18% perf-sched.sch_delay.max.ms.irqentry_exit_to_user_mode.asm_sysvec_call_function_single.[unknown].[unknown]
> 3.03 ± 82% -78.7% 0.65 ± 18% perf-sched.wait_and_delay.max.ms.irqentry_exit_to_user_mode.asm_sysvec_call_function_single.[unknown].[unknown]
> 28.66 ± 95% -75.5% 7.01 ± 5% perf-sched.wait_and_delay.max.ms.schedule_timeout.__wait_for_common.wait_for_completion_state.kernel_clone
> 1.52 ± 82% -78.7% 0.32 ± 18% perf-sched.wait_time.max.ms.irqentry_exit_to_user_mode.asm_sysvec_call_function_single.[unknown].[unknown]
> 27.97 ± 98% -76.0% 6.72 ± 5% perf-sched.wait_time.max.ms.schedule_timeout.__wait_for_common.wait_for_completion_state.kernel_clone
> 1.971e+08 ± 4% +13.2% 2.231e+08 ± 2% perf-stat.i.branch-misses
> 1.11 -2.3% 1.09 perf-stat.i.cpi
> 6.697e+11 +2.2% 6.842e+11 perf-stat.i.instructions
> 0.90 +2.3% 0.92 perf-stat.i.ipc
> 0.00 ±141% +162.1% 0.01 ± 38% perf-stat.i.major-faults
> 0.12 ± 4% +0.0 0.14 ± 2% perf-stat.overall.branch-miss-rate%
> 1.11 -2.2% 1.09 perf-stat.overall.cpi
> 0.90 +2.3% 0.92 perf-stat.overall.ipc
> 695559 +8.3% 753229 perf-stat.overall.path-length
> 1.964e+08 ± 4% +13.2% 2.223e+08 ± 2% perf-stat.ps.branch-misses
> 6.676e+11 +2.2% 6.82e+11 perf-stat.ps.instructions
> 0.00 ±141% +167.2% 0.01 ± 41% perf-stat.ps.major-faults
> 2.032e+14 +2.1% 2.074e+14 perf-stat.total.instructions
> 7.01 -0.4 6.62 perf-profile.calltrace.cycles-pp.clear_bhb_loop.write
> 7.02 -0.4 6.66 perf-profile.calltrace.cycles-pp.clear_bhb_loop.read
> 5.52 -0.4 5.17 perf-profile.calltrace.cycles-pp.copy_page_from_iter.pipe_write.vfs_write.ksys_write.do_syscall_64
> 5.47 -0.3 5.15 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64.read
> 5.47 -0.3 5.16 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64.write
> 4.27 -0.2 4.02 perf-profile.calltrace.cycles-pp.copy_page_to_iter.pipe_read.vfs_read.ksys_read.do_syscall_64
> 3.87 -0.2 3.64 ± 2% perf-profile.calltrace.cycles-pp._copy_from_iter.copy_page_from_iter.pipe_write.vfs_write.ksys_write
> 3.16 -0.2 3.00 perf-profile.calltrace.cycles-pp._copy_to_iter.copy_page_to_iter.pipe_read.vfs_read.ksys_read
> 53.32 -0.1 53.18 perf-profile.calltrace.cycles-pp.write
> 2.12 -0.1 1.98 perf-profile.calltrace.cycles-pp.__wake_up_sync_key.pipe_write.vfs_write.ksys_write.do_syscall_64
> 1.59 -0.1 1.48 perf-profile.calltrace.cycles-pp._raw_spin_lock_irqsave.__wake_up_sync_key.pipe_write.vfs_write.ksys_write
> 1.12 -0.1 1.03 perf-profile.calltrace.cycles-pp.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 0.97 -0.1 0.89 perf-profile.calltrace.cycles-pp.x64_sys_call.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 1.72 -0.1 1.64 perf-profile.calltrace.cycles-pp.mutex_lock.pipe_write.vfs_write.ksys_write.do_syscall_64
> 1.67 -0.1 1.60 perf-profile.calltrace.cycles-pp.mutex_lock.pipe_read.vfs_read.ksys_read.do_syscall_64
> 1.18 -0.1 1.11 perf-profile.calltrace.cycles-pp.fput.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 0.98 -0.1 0.90 perf-profile.calltrace.cycles-pp.mutex_unlock.pipe_read.vfs_read.ksys_read.do_syscall_64
> 1.17 -0.1 1.10 ± 2% perf-profile.calltrace.cycles-pp.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 0.98 -0.1 0.92 perf-profile.calltrace.cycles-pp.x64_sys_call.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 0.98 -0.1 0.92 perf-profile.calltrace.cycles-pp.mutex_unlock.pipe_write.vfs_write.ksys_write.do_syscall_64
> 0.97 -0.1 0.92 perf-profile.calltrace.cycles-pp.fput.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 0.63 ± 3% -0.1 0.58 perf-profile.calltrace.cycles-pp.testcase
> 0.00 +0.5 0.54 ± 4% perf-profile.calltrace.cycles-pp.timestamp_truncate.current_time.inode_needs_update_time.file_update_time.pipe_write
> 0.00 +0.5 0.55 ± 5% perf-profile.calltrace.cycles-pp.timestamp_truncate.current_time.atime_needs_update.touch_atime.pipe_read
> 0.00 +0.7 0.68 ± 11% perf-profile.calltrace.cycles-pp.ktime_get_coarse_ts64.coarse_ctime.current_time.atime_needs_update.touch_atime
> 35.70 +0.9 36.58 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.write
> 34.80 +0.9 35.74 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 32.06 +1.0 33.08 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.read
> 31.16 +1.1 32.24 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 0.00 +1.1 1.10 ± 16% perf-profile.calltrace.cycles-pp.coarse_ctime.current_time.inode_needs_update_time.file_update_time.pipe_write
> 30.51 +1.2 31.73 perf-profile.calltrace.cycles-pp.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 25.27 +1.3 26.57 perf-profile.calltrace.cycles-pp.vfs_write.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 27.06 +1.4 28.44 perf-profile.calltrace.cycles-pp.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 22.04 +1.5 23.51 perf-profile.calltrace.cycles-pp.vfs_read.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 0.00 +1.5 1.48 ± 3% perf-profile.calltrace.cycles-pp.coarse_ctime.current_time.atime_needs_update.touch_atime.pipe_read
> 17.63 +1.8 19.44 perf-profile.calltrace.cycles-pp.pipe_write.vfs_write.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe
> 14.70 +2.0 16.71 perf-profile.calltrace.cycles-pp.pipe_read.vfs_read.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe
> 4.15 +2.3 6.50 perf-profile.calltrace.cycles-pp.touch_atime.pipe_read.vfs_read.ksys_read.do_syscall_64
> 1.79 +2.4 4.15 perf-profile.calltrace.cycles-pp.current_time.atime_needs_update.touch_atime.pipe_read.vfs_read
> 3.76 +2.4 6.13 perf-profile.calltrace.cycles-pp.atime_needs_update.touch_atime.pipe_read.vfs_read.ksys_read
> 2.04 ± 12% +2.5 4.53 ± 5% perf-profile.calltrace.cycles-pp.file_update_time.pipe_write.vfs_write.ksys_write.do_syscall_64
> 1.76 ± 14% +2.5 4.26 ± 5% perf-profile.calltrace.cycles-pp.inode_needs_update_time.file_update_time.pipe_write.vfs_write.ksys_write
> 0.00 +3.5 3.46 ± 6% perf-profile.calltrace.cycles-pp.current_time.inode_needs_update_time.file_update_time.pipe_write.vfs_write
> 14.13 -0.8 13.36 perf-profile.children.cycles-pp.clear_bhb_loop
> 7.00 -0.4 6.60 perf-profile.children.cycles-pp.entry_SYSCALL_64
> 5.64 -0.4 5.28 perf-profile.children.cycles-pp.copy_page_from_iter
> 4.19 -0.3 3.94 perf-profile.children.cycles-pp._copy_from_iter
> 4.34 -0.2 4.08 perf-profile.children.cycles-pp.copy_page_to_iter
> 4.12 -0.2 3.88 perf-profile.children.cycles-pp.entry_SYSRETQ_unsafe_stack
> 3.42 -0.2 3.25 perf-profile.children.cycles-pp._copy_to_iter
> 3.51 -0.2 3.35 perf-profile.children.cycles-pp.mutex_lock
> 2.15 -0.2 1.99 perf-profile.children.cycles-pp.x64_sys_call
> 53.53 -0.2 53.38 perf-profile.children.cycles-pp.write
> 2.41 -0.2 2.26 perf-profile.children.cycles-pp.syscall_exit_to_user_mode
> 2.21 -0.1 2.07 perf-profile.children.cycles-pp.__wake_up_sync_key
> 2.16 -0.1 2.02 perf-profile.children.cycles-pp.fput
> 2.03 -0.1 1.90 perf-profile.children.cycles-pp.mutex_unlock
> 1.60 -0.1 1.50 perf-profile.children.cycles-pp._raw_spin_lock_irqsave
> 0.73 ± 4% -0.1 0.67 perf-profile.children.cycles-pp.testcase
> 0.64 ± 4% -0.0 0.60 ± 2% perf-profile.children.cycles-pp.aa_file_perm
> 0.77 -0.0 0.74 perf-profile.children.cycles-pp.entry_SYSCALL_64_safe_stack
> 0.38 -0.0 0.35 perf-profile.children.cycles-pp.__x64_sys_read
> 0.40 -0.0 0.38 perf-profile.children.cycles-pp.__x64_sys_write
> 0.32 -0.0 0.30 perf-profile.children.cycles-pp.kill_fasync
> 0.16 ± 3% -0.0 0.14 ± 2% perf-profile.children.cycles-pp.make_vfsgid
> 0.30 -0.0 0.29 perf-profile.children.cycles-pp._raw_spin_unlock_irqrestore
> 0.00 +0.4 0.41 ± 26% perf-profile.children.cycles-pp.set_normalized_timespec64
> 0.43 ± 13% +0.7 1.14 ± 4% perf-profile.children.cycles-pp.timestamp_truncate
> 0.00 +1.0 0.95 ± 4% perf-profile.children.cycles-pp.ktime_get_coarse_with_offset
> 0.00 +1.1 1.06 perf-profile.children.cycles-pp.ns_to_timespec64
> 0.00 +1.2 1.17 ± 3% perf-profile.children.cycles-pp.ktime_get_coarse_ts64
> 30.76 +1.2 31.97 perf-profile.children.cycles-pp.ksys_write
> 25.47 +1.3 26.75 perf-profile.children.cycles-pp.vfs_write
> 27.19 +1.4 28.55 perf-profile.children.cycles-pp.ksys_read
> 22.13 +1.4 23.57 perf-profile.children.cycles-pp.vfs_read
> 17.81 +1.8 19.60 perf-profile.children.cycles-pp.pipe_write
> 68.06 +1.9 69.94 perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe
> 15.02 +2.0 17.01 perf-profile.children.cycles-pp.pipe_read
> 66.19 +2.0 68.19 perf-profile.children.cycles-pp.do_syscall_64
> 4.24 +2.3 6.58 perf-profile.children.cycles-pp.touch_atime
> 3.86 +2.4 6.23 perf-profile.children.cycles-pp.atime_needs_update
> 1.88 ± 13% +2.5 4.34 ± 5% perf-profile.children.cycles-pp.inode_needs_update_time
> 2.11 ± 12% +2.5 4.60 ± 5% perf-profile.children.cycles-pp.file_update_time
> 0.00 +2.7 2.66 ± 6% perf-profile.children.cycles-pp.coarse_ctime
> 1.82 +6.2 8.04 ± 2% perf-profile.children.cycles-pp.current_time
> 14.05 -0.8 13.28 perf-profile.self.cycles-pp.clear_bhb_loop
> 1.06 ± 8% -0.4 0.67 perf-profile.self.cycles-pp.inode_needs_update_time
> 5.00 ± 3% -0.3 4.67 ± 3% perf-profile.self.cycles-pp.vfs_write
> 4.10 ± 3% -0.3 3.78 ± 3% perf-profile.self.cycles-pp.vfs_read
> 3.71 -0.2 3.48 ± 2% perf-profile.self.cycles-pp._copy_from_iter
> 3.98 -0.2 3.75 perf-profile.self.cycles-pp.entry_SYSRETQ_unsafe_stack
> 2.89 -0.2 2.68 perf-profile.self.cycles-pp.do_syscall_64
> 3.38 -0.2 3.17 perf-profile.self.cycles-pp.read
> 2.91 -0.2 2.71 ± 2% perf-profile.self.cycles-pp._copy_to_iter
> 3.48 -0.2 3.28 perf-profile.self.cycles-pp.write
> 3.06 -0.2 2.88 perf-profile.self.cycles-pp.entry_SYSCALL_64
> 1.76 -0.2 1.60 perf-profile.self.cycles-pp.atime_needs_update
> 2.03 -0.2 1.88 perf-profile.self.cycles-pp.x64_sys_call
> 1.90 -0.1 1.78 perf-profile.self.cycles-pp.entry_SYSCALL_64_after_hwframe
> 2.18 -0.1 2.07 perf-profile.self.cycles-pp.mutex_lock
> 1.93 -0.1 1.82 perf-profile.self.cycles-pp.mutex_unlock
> 1.92 -0.1 1.81 perf-profile.self.cycles-pp.fput
> 1.46 -0.1 1.36 perf-profile.self.cycles-pp.syscall_exit_to_user_mode
> 1.54 -0.1 1.45 perf-profile.self.cycles-pp._raw_spin_lock_irqsave
> 1.06 -0.1 0.98 perf-profile.self.cycles-pp.ksys_read
> 1.19 -0.1 1.11 perf-profile.self.cycles-pp.ksys_write
> 0.58 ± 4% -0.1 0.53 perf-profile.self.cycles-pp.testcase
> 0.77 -0.0 0.74 perf-profile.self.cycles-pp.entry_SYSCALL_64_safe_stack
> 0.38 -0.0 0.36 ± 2% perf-profile.self.cycles-pp.touch_atime
> 0.30 -0.0 0.28 perf-profile.self.cycles-pp.__x64_sys_write
> 0.35 -0.0 0.33 perf-profile.self.cycles-pp.__wake_up_sync_key
> 0.28 -0.0 0.26 perf-profile.self.cycles-pp.__x64_sys_read
> 0.25 -0.0 0.24 perf-profile.self.cycles-pp.kill_fasync
> 0.26 +0.0 0.29 perf-profile.self.cycles-pp.file_update_time
> 0.00 +0.4 0.40 ± 25% perf-profile.self.cycles-pp.set_normalized_timespec64
> 0.40 ± 11% +0.7 1.06 ± 4% perf-profile.self.cycles-pp.timestamp_truncate
> 0.00 +0.9 0.89 perf-profile.self.cycles-pp.ns_to_timespec64
> 0.00 +0.9 0.91 ± 4% perf-profile.self.cycles-pp.ktime_get_coarse_with_offset
> 0.00 +1.1 1.10 ± 5% perf-profile.self.cycles-pp.coarse_ctime
> 0.00 +1.1 1.11 ± 3% perf-profile.self.cycles-pp.ktime_get_coarse_ts64
> 1.06 ± 5% +1.4 2.44 ± 2% perf-profile.self.cycles-pp.current_time
>
>
> ***************************************************************************************************
> lkp-icl-2sp2: 128 threads 2 sockets Intel(R) Xeon(R) Gold 6338 CPU @ 2.00GHz (Ice Lake) with 256G memory
> =========================================================================================
> compiler/cpufreq_governor/ipc/iterations/kconfig/mode/nr_threads/rootfs/tbox_group/testcase:
> gcc-12/performance/pipe/4/x86_64-rhel-8.3/threads/800%/debian-12-x86_64-20240206.cgz/lkp-icl-2sp2/hackbench
>
> commit:
> v6.12-rc1
> 2e4c6e78f4 ("fs: add infrastructure for multigrain timestamps")
>
> v6.12-rc1 2e4c6e78f41afefb7a2b825b7aa
> ---------------- ---------------------------
> %stddev %change %stddev
> \ | \
> 561190 ± 49% -33.6% 372781 ± 77% numa-meminfo.node0.SUnreclaim
> 137108 ± 48% -35.9% 87915 ± 74% numa-vmstat.node0.nr_slab_unreclaimable
> 2.263e+08 ± 3% -13.7% 1.954e+08 ± 7% perf-stat.i.cache-misses
> 7.636e+08 ± 5% -14.5% 6.531e+08 ± 9% perf-stat.i.cache-references
> 18.89 ± 18% -31.6% 12.92 ± 21% perf-stat.i.metric.K/sec
> 246747 -3.4% 238348 proc-vmstat.nr_anon_pages
> 383468 ± 36% -19.0% 310490 ± 2% proc-vmstat.nr_inactive_anon
> 2246 ± 3% -4.9% 2135 ± 2% proc-vmstat.nr_page_table_pages
> 383468 ± 36% -19.0% 310490 ± 2% proc-vmstat.nr_zone_inactive_anon
> 1231417 -4.5% 1175946 hackbench.throughput
> 1179456 -3.7% 1136004 hackbench.throughput_avg
> 1231417 -4.5% 1175946 hackbench.throughput_best
> 5279 +4.8% 5530 hackbench.time.system_time
> 954.46 -0.8% 946.55 hackbench.time.user_time
> 0.12 ± 8% -36.0% 0.08 ± 23% sched_debug.cfs_rq:/.h_nr_running.avg
> 0.33 ± 4% -21.5% 0.26 ± 11% sched_debug.cfs_rq:/.h_nr_running.stddev
> 77.21 ± 38% -50.7% 38.06 ± 60% sched_debug.cfs_rq:/.load_avg.avg
> 1421 ± 27% -49.9% 712.70 ± 26% sched_debug.cfs_rq:/.load_avg.max
> 251.26 ± 25% -45.5% 137.06 ± 44% sched_debug.cfs_rq:/.load_avg.stddev
> 0.12 ± 8% -36.0% 0.08 ± 23% sched_debug.cfs_rq:/.nr_running.avg
> 0.33 ± 4% -21.5% 0.26 ± 11% sched_debug.cfs_rq:/.nr_running.stddev
> 221.71 ± 26% -49.9% 111.18 ± 60% sched_debug.cfs_rq:/.removed.load_avg.stddev
> 253.34 ± 7% -37.5% 158.42 ± 28% sched_debug.cfs_rq:/.runnable_avg.avg
> 297.62 ± 4% -20.9% 235.28 ± 11% sched_debug.cfs_rq:/.runnable_avg.stddev
> 252.17 ± 7% -37.6% 157.31 ± 28% sched_debug.cfs_rq:/.util_avg.avg
> 297.28 ± 4% -21.0% 234.70 ± 11% sched_debug.cfs_rq:/.util_avg.stddev
> 335.27 ± 11% -33.2% 224.09 ± 23% sched_debug.cpu.curr->pid.avg
> 941.56 ± 4% -10.3% 844.57 ± 6% sched_debug.cpu.curr->pid.stddev
> 0.11 ± 12% -35.9% 0.07 ± 25% sched_debug.cpu.nr_running.avg
> 0.32 ± 6% -21.6% 0.25 ± 12% sched_debug.cpu.nr_running.stddev
> 131.67 ± 4% +1e+05% 134061 ± 52% sched_debug.cpu.nr_switches.min
> 0.01 ± 28% -64.0% 0.00 ± 33% sched_debug.cpu.nr_uninterruptible.avg
> 12.54 ± 84% -12.5 0.00 perf-profile.calltrace.cycles-pp.devkmsg_emit.devkmsg_write.vfs_write.ksys_write.do_syscall_64
> 12.54 ± 84% -12.5 0.00 perf-profile.calltrace.cycles-pp.devkmsg_write.vfs_write.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe
> 12.54 ± 84% -12.5 0.00 perf-profile.calltrace.cycles-pp.vprintk_emit.devkmsg_emit.devkmsg_write.vfs_write.ksys_write
> 12.66 ± 83% -12.4 0.29 ±129% perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 12.66 ± 83% -12.4 0.29 ±129% perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.write
> 12.66 ± 83% -12.4 0.29 ±129% perf-profile.calltrace.cycles-pp.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 12.66 ± 83% -12.4 0.29 ±129% perf-profile.calltrace.cycles-pp.write
> 11.78 ± 84% -11.8 0.00 perf-profile.calltrace.cycles-pp.console_unlock.vprintk_emit.devkmsg_emit.devkmsg_write.vfs_write
> 11.78 ± 84% -11.8 0.00 perf-profile.calltrace.cycles-pp.console_flush_all.console_unlock.vprintk_emit.devkmsg_emit.devkmsg_write
> 11.10 ± 84% -11.1 0.00 perf-profile.calltrace.cycles-pp.serial8250_console_write.console_flush_all.console_unlock.vprintk_emit.devkmsg_emit
> 7.42 ± 61% -5.0 2.40 ± 83% perf-profile.calltrace.cycles-pp.io_serial_in.wait_for_lsr.serial8250_console_write.console_flush_all.console_unlock
> 0.30 ±150% +1.1 1.43 ± 43% perf-profile.calltrace.cycles-pp.number.vsnprintf.seq_printf.show_interrupts.seq_read_iter
> 8.40 ± 41% +5.9 14.30 ± 15% perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.read
> 8.36 ± 42% +5.9 14.30 ± 15% perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 8.60 ± 40% +5.9 14.54 ± 16% perf-profile.calltrace.cycles-pp.read
> 8.23 ± 41% +6.1 14.30 ± 15% perf-profile.calltrace.cycles-pp.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 8.10 ± 41% +6.1 14.18 ± 14% perf-profile.calltrace.cycles-pp.vfs_read.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 12.54 ± 84% -12.5 0.00 perf-profile.children.cycles-pp.devkmsg_emit
> 12.54 ± 84% -12.5 0.00 perf-profile.children.cycles-pp.devkmsg_write
> 9.02 ± 58% -6.3 2.74 ± 83% perf-profile.children.cycles-pp.io_serial_in
> 1.49 ± 35% -0.7 0.77 ± 39% perf-profile.children.cycles-pp.d_alloc_parallel
> 1.20 ± 30% -0.6 0.57 ± 26% perf-profile.children.cycles-pp.d_alloc
> 7.51 ±101% -0.5 7.04 ±123% perf-profile.children.cycles-pp.__ordered_events__flush
> 0.70 ± 38% -0.4 0.30 ± 72% perf-profile.children.cycles-pp.lookup_open
> 3.15 ±104% -0.4 2.79 ±127% perf-profile.children.cycles-pp.build_id__mark_dso_hit
> 6.78 ±103% -0.0 6.75 ±122% perf-profile.children.cycles-pp.perf_session__deliver_event
> 0.11 ±119% +0.3 0.41 ± 34% perf-profile.children.cycles-pp.free_unref_page
> 0.54 ± 35% +0.9 1.39 ± 48% perf-profile.children.cycles-pp.__dentry_kill
> 0.70 ± 47% +1.1 1.81 ± 36% perf-profile.children.cycles-pp.dput
> 7.71 ± 43% +5.8 13.46 ± 11% perf-profile.children.cycles-pp.seq_read_iter
> 8.60 ± 40% +5.9 14.54 ± 16% perf-profile.children.cycles-pp.read
> 8.29 ± 40% +6.0 14.30 ± 15% perf-profile.children.cycles-pp.ksys_read
> 8.15 ± 40% +6.1 14.23 ± 13% perf-profile.children.cycles-pp.vfs_read
> 10.49 ± 58% +11.5 21.94 ± 29% perf-profile.children.cycles-pp.__cmd_record
> 9.02 ± 58% -6.3 2.74 ± 83% perf-profile.self.cycles-pp.io_serial_in
> 0.15 ±107% +1.0 1.19 ± 53% perf-profile.self.cycles-pp.show_interrupts
> 0.01 ±169% +635.4% 0.05 ± 46% perf-sched.sch_delay.avg.ms.__cond_resched.smpboot_thread_fn.kthread.ret_from_fork.ret_from_fork_asm
> 506.03 ±223% +1038.0% 5758 ± 92% perf-sched.sch_delay.avg.ms.__cond_resched.switch_task_namespaces.do_exit.__x64_sys_exit.x64_sys_call
> 2527 ± 37% +34.6% 3402 ± 17% perf-sched.sch_delay.max.ms.__cond_resched.__wait_for_common.affine_move_task.__set_cpus_allowed_ptr.__sched_setaffinity
> 5433 ± 62% +111.9% 11512 ± 8% perf-sched.sch_delay.max.ms.__cond_resched.mutex_lock.pipe_read.vfs_read.ksys_read
> 8727 ± 28% +41.7% 12368 ± 6% perf-sched.sch_delay.max.ms.__cond_resched.mutex_lock.pipe_write.vfs_write.ksys_write
> 0.16 ±221% +1222.8% 2.11 ±101% perf-sched.sch_delay.max.ms.__cond_resched.smpboot_thread_fn.kthread.ret_from_fork.ret_from_fork_asm
> 1543 ±223% +338.6% 6770 ± 82% perf-sched.sch_delay.max.ms.__cond_resched.switch_task_namespaces.do_exit.__x64_sys_exit.x64_sys_call
> 8967 ± 26% +36.1% 12200 ± 7% perf-sched.sch_delay.max.ms.irqentry_exit_to_user_mode.asm_sysvec_apic_timer_interrupt.[unknown].[unknown]
> 65.76 ± 67% +3435.0% 2324 ±184% perf-sched.sch_delay.max.ms.irqentry_exit_to_user_mode.asm_sysvec_reschedule_ipi.[unknown].[unknown]
> 9232 ± 26% +36.5% 12600 ± 6% perf-sched.sch_delay.max.ms.pipe_read.vfs_read.ksys_read.do_syscall_64
> 9419 ± 25% +33.7% 12597 ± 5% perf-sched.sch_delay.max.ms.pipe_write.vfs_write.ksys_write.do_syscall_64
> 8740 ± 27% +43.0% 12498 ± 4% perf-sched.sch_delay.max.ms.schedule_preempt_disabled.__mutex_lock.constprop.0.pipe_write
> 8472 ± 39% +48.9% 12613 ± 5% perf-sched.sch_delay.max.ms.schedule_preempt_disabled.rwsem_down_read_slowpath.down_read.exit_mm
> 9034 ± 27% +40.2% 12667 ± 4% perf-sched.sch_delay.max.ms.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
> 9461 ± 25% +36.4% 12905 ± 4% perf-sched.total_sch_delay.max.ms
> 18705 ± 25% +37.6% 25730 ± 5% perf-sched.total_wait_and_delay.max.ms
> 9525 ± 23% +38.6% 13204 ± 7% perf-sched.total_wait_time.max.ms
> 17493 ± 28% +41.3% 24713 ± 6% perf-sched.wait_and_delay.max.ms.__cond_resched.mutex_lock.pipe_write.vfs_write.ksys_write
> 17105 ± 27% +42.1% 24316 ± 8% perf-sched.wait_and_delay.max.ms.irqentry_exit_to_user_mode.asm_sysvec_apic_timer_interrupt.[unknown].[unknown]
> 17952 ± 26% +39.7% 25074 ± 6% perf-sched.wait_and_delay.max.ms.pipe_read.vfs_read.ksys_read.do_syscall_64
> 18587 ± 25% +36.2% 25322 ± 5% perf-sched.wait_and_delay.max.ms.pipe_write.vfs_write.ksys_write.do_syscall_64
> 17333 ± 29% +44.3% 25016 ± 4% perf-sched.wait_and_delay.max.ms.schedule_preempt_disabled.__mutex_lock.constprop.0.pipe_write
> 16951 ± 39% +48.4% 25149 ± 5% perf-sched.wait_and_delay.max.ms.schedule_preempt_disabled.rwsem_down_read_slowpath.down_read.exit_mm
> 17996 ± 27% +40.2% 25232 ± 5% perf-sched.wait_and_delay.max.ms.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
> 163.71 ±203% +576.0% 1106 ± 90% perf-sched.wait_time.avg.ms.__cond_resched.pipe_write.vfs_write.ksys_write.do_syscall_64
> 18.58 ±223% +986.5% 201.88 ± 64% perf-sched.wait_time.avg.ms.__cond_resched.smpboot_thread_fn.kthread.ret_from_fork.ret_from_fork_asm
> 1478 ±223% +391.5% 7267 ± 71% perf-sched.wait_time.avg.ms.__cond_resched.switch_task_namespaces.do_exit.__x64_sys_exit.x64_sys_call
> 0.15 ±223% +2351.7% 3.74 ± 70% perf-sched.wait_time.avg.ms.do_task_dead.do_exit.__x64_sys_exit.x64_sys_call.do_syscall_64
> 2619 ± 27% +29.9% 3402 ± 17% perf-sched.wait_time.max.ms.__cond_resched.__wait_for_common.affine_move_task.__set_cpus_allowed_ptr.__sched_setaffinity
> 1593 ±223% +360.5% 7337 ± 81% perf-sched.wait_time.max.ms.__cond_resched.mmput.exit_mm.do_exit.__x64_sys_exit
> 7248 ± 40% +60.1% 11605 ± 8% perf-sched.wait_time.max.ms.__cond_resched.mutex_lock.pipe_read.vfs_read.ksys_read
> 8913 ± 27% +41.0% 12567 ± 8% perf-sched.wait_time.max.ms.__cond_resched.mutex_lock.pipe_write.vfs_write.ksys_write
> 1658 ±217% +301.5% 6658 ± 81% perf-sched.wait_time.max.ms.__cond_resched.pipe_write.vfs_write.ksys_write.do_syscall_64
> 1671 ±223% +498.0% 9998 ± 50% perf-sched.wait_time.max.ms.__cond_resched.smpboot_thread_fn.kthread.ret_from_fork.ret_from_fork_asm
> 1581 ±223% +473.2% 9064 ± 50% perf-sched.wait_time.max.ms.__cond_resched.switch_task_namespaces.do_exit.__x64_sys_exit.x64_sys_call
> 8915 ± 25% +39.1% 12405 ± 7% perf-sched.wait_time.max.ms.irqentry_exit_to_user_mode.asm_sysvec_apic_timer_interrupt.[unknown].[unknown]
> 9200 ± 26% +40.4% 12920 ± 7% perf-sched.wait_time.max.ms.pipe_read.vfs_read.ksys_read.do_syscall_64
> 9433 ± 23% +36.3% 12857 ± 7% perf-sched.wait_time.max.ms.pipe_write.vfs_write.ksys_write.do_syscall_64
> 8838 ± 28% +42.4% 12582 ± 4% perf-sched.wait_time.max.ms.schedule_preempt_disabled.__mutex_lock.constprop.0.pipe_write
> 8520 ± 39% +48.5% 12654 ± 5% perf-sched.wait_time.max.ms.schedule_preempt_disabled.rwsem_down_read_slowpath.down_read.exit_mm
> 8819 ± 26% +45.3% 12815 ± 7% perf-sched.wait_time.max.ms.smpboot_thread_fn.kthread.ret_from_fork.ret_from_fork_asm
> 9176 ± 26% +39.6% 12813 ± 7% perf-sched.wait_time.max.ms.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
> 8699 ± 25% +48.7% 12937 ± 8% perf-sched.wait_time.max.ms.worker_thread.kthread.ret_from_fork.ret_from_fork_asm
>
>
>
> ***************************************************************************************************
> 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-8.3/process/100%/debian-12-x86_64-20240206.cgz/lkp-skl-fpga01/pipe1/will-it-scale
>
> commit:
> v6.12-rc1
> 2e4c6e78f4 ("fs: add infrastructure for multigrain timestamps")
>
> v6.12-rc1 2e4c6e78f41afefb7a2b825b7aa
> ---------------- ---------------------------
> %stddev %change %stddev
> \ | \
> 816.00 ± 6% -11.4% 722.72 ± 5% sched_debug.cfs_rq:/.util_est.max
> 33816990 -2.0% 33148227 will-it-scale.104.processes
> 325162 -2.0% 318732 will-it-scale.per_process_ops
> 33816990 -2.0% 33148227 will-it-scale.workload
> 0.70 ± 63% +68.4% 1.18 ± 4% perf-sched.sch_delay.avg.ms.worker_thread.kthread.ret_from_fork.ret_from_fork_asm
> 506.90 ± 11% -14.5% 433.43 ± 3% perf-sched.wait_and_delay.avg.ms.worker_thread.kthread.ret_from_fork.ret_from_fork_asm
> 281.17 ± 50% -51.7% 135.83 ± 19% perf-sched.wait_and_delay.count.irqentry_exit_to_user_mode.asm_sysvec_apic_timer_interrupt.[unknown]
> 1000 -79.4% 206.38 ±171% perf-sched.wait_and_delay.max.ms.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
> 2.27 ± 52% +18623.0% 424.45 ±117% perf-sched.wait_time.avg.ms.__cond_resched.shmem_inode_acct_blocks.shmem_alloc_and_add_folio.shmem_get_folio_gfp.shmem_write_begin
> 506.20 ± 11% -14.6% 432.25 ± 3% perf-sched.wait_time.avg.ms.worker_thread.kthread.ret_from_fork.ret_from_fork_asm
> 999.76 -80.5% 195.04 ±183% perf-sched.wait_time.max.ms.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
> 1.3e+10 +8.0% 1.404e+10 perf-stat.i.branch-instructions
> 1.10 -0.1 1.00 perf-stat.i.branch-miss-rate%
> 4.14 -6.8% 3.86 perf-stat.i.cpi
> 425295 -5.1% 403763 perf-stat.i.cycles-between-cache-misses
> 6.96e+10 +7.2% 7.461e+10 perf-stat.i.instructions
> 0.24 +7.0% 0.26 perf-stat.i.ipc
> 0.01 ± 2% -7.0% 0.01 ± 5% perf-stat.overall.MPKI
> 1.09 -0.1 0.99 perf-stat.overall.branch-miss-rate%
> 4.15 -6.8% 3.87 perf-stat.overall.cpi
> 0.24 +7.3% 0.26 perf-stat.overall.ipc
> 622217 +9.4% 680897 perf-stat.overall.path-length
> 1.296e+10 +8.0% 1.4e+10 perf-stat.ps.branch-instructions
> 6.937e+10 +7.2% 7.436e+10 perf-stat.ps.instructions
> 2.104e+13 +7.3% 2.257e+13 perf-stat.total.instructions
> 1.62 -0.5 1.15 perf-profile.calltrace.cycles-pp.mutex_lock.pipe_write.vfs_write.ksys_write.do_syscall_64
> 11.67 -0.3 11.34 perf-profile.calltrace.cycles-pp.syscall_return_via_sysret.write
> 11.68 -0.1 11.56 perf-profile.calltrace.cycles-pp.syscall_return_via_sysret.read
> 8.36 -0.1 8.25 perf-profile.calltrace.cycles-pp.entry_SYSRETQ_unsafe_stack.write
> 4.06 -0.1 3.95 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64.write
> 1.46 -0.1 1.37 perf-profile.calltrace.cycles-pp.mutex_lock.pipe_read.vfs_read.ksys_read.do_syscall_64
> 4.02 -0.1 3.93 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64.read
> 0.73 +0.0 0.75 perf-profile.calltrace.cycles-pp.security_file_permission.rw_verify_area.vfs_write.ksys_write.do_syscall_64
> 0.65 +0.0 0.68 perf-profile.calltrace.cycles-pp.apparmor_file_permission.security_file_permission.rw_verify_area.vfs_write.ksys_write
> 0.80 +0.0 0.84 perf-profile.calltrace.cycles-pp.rw_verify_area.vfs_write.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe
> 0.92 +0.0 0.97 perf-profile.calltrace.cycles-pp.fdget_pos.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 0.87 ± 2% +0.2 1.02 ± 2% perf-profile.calltrace.cycles-pp.fdget_pos.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 1.66 ± 2% +0.2 1.85 ± 2% perf-profile.calltrace.cycles-pp.copy_page_to_iter.pipe_read.vfs_read.ksys_read.do_syscall_64
> 1.22 ± 2% +0.3 1.50 ± 2% perf-profile.calltrace.cycles-pp._copy_to_iter.copy_page_to_iter.pipe_read.vfs_read.ksys_read
> 21.79 +0.4 22.14 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.read
> 1.39 ± 2% +0.5 1.86 perf-profile.calltrace.cycles-pp._copy_from_iter.copy_page_from_iter.pipe_write.vfs_write.ksys_write
> 1.92 +0.5 2.40 perf-profile.calltrace.cycles-pp.copy_page_from_iter.pipe_write.vfs_write.ksys_write.do_syscall_64
> 14.97 +0.5 15.47 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 22.36 +0.6 22.90 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.write
> 7.84 +0.6 8.43 perf-profile.calltrace.cycles-pp.pipe_read.vfs_read.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe
> 0.00 +0.6 0.59 ± 4% perf-profile.calltrace.cycles-pp.rep_movs_alternative._copy_to_iter.copy_page_to_iter.pipe_read.vfs_read
> 0.00 +0.6 0.60 ± 2% perf-profile.calltrace.cycles-pp.rep_movs_alternative._copy_from_iter.copy_page_from_iter.pipe_write.vfs_write
> 1.69 +0.7 2.41 perf-profile.calltrace.cycles-pp.touch_atime.pipe_read.vfs_read.ksys_read.do_syscall_64
> 1.40 +0.8 2.16 perf-profile.calltrace.cycles-pp.atime_needs_update.touch_atime.pipe_read.vfs_read.ksys_read
> 10.52 +0.8 11.29 perf-profile.calltrace.cycles-pp.vfs_read.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 0.56 +0.8 1.34 perf-profile.calltrace.cycles-pp.current_time.atime_needs_update.touch_atime.pipe_read.vfs_read
> 0.52 +0.8 1.34 perf-profile.calltrace.cycles-pp.inode_needs_update_time.file_update_time.pipe_write.vfs_write.ksys_write
> 15.50 +0.8 16.31 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 0.65 +0.9 1.56 perf-profile.calltrace.cycles-pp.file_update_time.pipe_write.vfs_write.ksys_write.do_syscall_64
> 12.82 +0.9 13.74 perf-profile.calltrace.cycles-pp.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 12.00 +0.9 12.92 perf-profile.calltrace.cycles-pp.ksys_read.do_syscall_64.entry_SYSCALL_64_after_hwframe.read
> 0.00 +1.0 1.02 perf-profile.calltrace.cycles-pp.current_time.inode_needs_update_time.file_update_time.pipe_write.vfs_write
> 11.10 +1.1 12.18 perf-profile.calltrace.cycles-pp.vfs_write.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe.write
> 8.12 +1.3 9.39 perf-profile.calltrace.cycles-pp.pipe_write.vfs_write.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe
> 3.13 -0.6 2.56 perf-profile.children.cycles-pp.mutex_lock
> 1.40 -0.5 0.90 ± 2% perf-profile.children.cycles-pp.__cond_resched
> 23.52 -0.5 23.07 perf-profile.children.cycles-pp.syscall_return_via_sysret
> 1.35 ± 3% -0.3 1.01 perf-profile.children.cycles-pp.syscall_exit_to_user_mode
> 0.88 ± 3% -0.3 0.54 ± 2% perf-profile.children.cycles-pp.rcu_all_qs
> 0.66 ± 6% -0.2 0.41 ± 2% perf-profile.children.cycles-pp.syscall_exit_to_user_mode_prepare
> 8.92 -0.2 8.68 perf-profile.children.cycles-pp.entry_SYSCALL_64
> 17.87 -0.2 17.63 perf-profile.children.cycles-pp.entry_SYSRETQ_unsafe_stack
> 0.00 +0.1 0.11 ± 8% perf-profile.children.cycles-pp.set_normalized_timespec64
> 0.15 ± 2% +0.1 0.28 ± 12% perf-profile.children.cycles-pp.timestamp_truncate
> 1.68 ± 2% +0.2 1.87 ± 2% perf-profile.children.cycles-pp.copy_page_to_iter
> 1.80 +0.2 2.00 perf-profile.children.cycles-pp.fdget_pos
> 1.32 +0.3 1.59 ± 2% perf-profile.children.cycles-pp._copy_to_iter
> 0.00 +0.3 0.28 perf-profile.children.cycles-pp.ktime_get_coarse_with_offset
> 0.00 +0.3 0.32 perf-profile.children.cycles-pp.ktime_get_coarse_ts64
> 0.00 +0.3 0.35 ± 2% perf-profile.children.cycles-pp.ns_to_timespec64
> 1.95 +0.5 2.42 perf-profile.children.cycles-pp.copy_page_from_iter
> 1.49 ± 2% +0.5 1.98 perf-profile.children.cycles-pp._copy_from_iter
> 0.65 +0.6 1.22 ± 2% perf-profile.children.cycles-pp.rep_movs_alternative
> 7.98 +0.6 8.56 perf-profile.children.cycles-pp.pipe_read
> 1.72 +0.7 2.43 perf-profile.children.cycles-pp.touch_atime
> 0.00 +0.7 0.74 perf-profile.children.cycles-pp.coarse_ctime
> 1.44 +0.8 2.20 perf-profile.children.cycles-pp.atime_needs_update
> 10.55 +0.8 11.32 perf-profile.children.cycles-pp.vfs_read
> 0.56 +0.8 1.37 perf-profile.children.cycles-pp.inode_needs_update_time
> 44.44 +0.9 45.34 perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe
> 0.67 +0.9 1.58 perf-profile.children.cycles-pp.file_update_time
> 12.86 +0.9 13.77 perf-profile.children.cycles-pp.ksys_write
> 12.02 +0.9 12.94 perf-profile.children.cycles-pp.ksys_read
> 11.16 +1.1 12.24 perf-profile.children.cycles-pp.vfs_write
> 8.18 +1.3 9.43 perf-profile.children.cycles-pp.pipe_write
> 30.60 +1.3 31.91 perf-profile.children.cycles-pp.do_syscall_64
> 0.57 +1.9 2.51 perf-profile.children.cycles-pp.current_time
> 23.47 -0.5 23.01 perf-profile.self.cycles-pp.syscall_return_via_sysret
> 14.02 -0.4 13.60 perf-profile.self.cycles-pp.entry_SYSCALL_64_after_hwframe
> 0.84 ± 3% -0.3 0.50 ± 3% perf-profile.self.cycles-pp.rcu_all_qs
> 0.82 ± 4% -0.3 0.50 ± 2% perf-profile.self.cycles-pp.ksys_write
> 17.70 -0.2 17.46 perf-profile.self.cycles-pp.entry_SYSRETQ_unsafe_stack
> 0.61 ± 6% -0.2 0.38 ± 4% perf-profile.self.cycles-pp.syscall_exit_to_user_mode_prepare
> 7.83 -0.2 7.61 perf-profile.self.cycles-pp.entry_SYSCALL_64
> 2.15 -0.2 1.94 perf-profile.self.cycles-pp.vfs_write
> 1.88 ± 2% -0.2 1.70 perf-profile.self.cycles-pp.pipe_read
> 0.52 ± 3% -0.2 0.37 ± 3% perf-profile.self.cycles-pp.__cond_resched
> 0.69 -0.1 0.58 ± 2% perf-profile.self.cycles-pp.syscall_exit_to_user_mode
> 0.35 ± 3% -0.1 0.27 perf-profile.self.cycles-pp.copy_page_to_iter
> 0.80 -0.1 0.72 perf-profile.self.cycles-pp.atime_needs_update
> 0.34 ± 2% -0.1 0.28 perf-profile.self.cycles-pp.inode_needs_update_time
> 0.26 ± 2% -0.0 0.21 ± 3% perf-profile.self.cycles-pp.touch_atime
> 0.46 -0.0 0.45 perf-profile.self.cycles-pp.copy_page_from_iter
> 0.10 ± 4% +0.1 0.20 perf-profile.self.cycles-pp.file_update_time
> 0.54 ± 2% +0.1 0.64 ± 3% perf-profile.self.cycles-pp.ksys_read
> 0.00 +0.1 0.11 ± 8% perf-profile.self.cycles-pp.set_normalized_timespec64
> 0.14 ± 3% +0.1 0.26 ± 14% perf-profile.self.cycles-pp.timestamp_truncate
> 1.28 ± 2% +0.2 1.45 perf-profile.self.cycles-pp._copy_from_iter
> 1.48 +0.2 1.67 ± 2% perf-profile.self.cycles-pp.vfs_read
> 1.80 +0.2 1.99 perf-profile.self.cycles-pp.fdget_pos
> 0.00 +0.3 0.26 perf-profile.self.cycles-pp.ktime_get_coarse_with_offset
> 0.00 +0.3 0.26 perf-profile.self.cycles-pp.ns_to_timespec64
> 0.00 +0.3 0.30 perf-profile.self.cycles-pp.ktime_get_coarse_ts64
> 0.00 +0.3 0.33 perf-profile.self.cycles-pp.coarse_ctime
> 2.29 ± 2% +0.4 2.72 perf-profile.self.cycles-pp.pipe_write
> 0.48 +0.6 1.04 ± 3% perf-profile.self.cycles-pp.rep_movs_alternative
> 0.33 +0.6 0.95 ± 5% perf-profile.self.cycles-pp.current_time
>
>
>
>
>
> 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.
>
>
--
Jeff Layton <jlayton@xxxxxxxxxx>
