From: Rong Chen <rong.a.chen@intel.com>
To: lkp@lists.01.org
Subject: Re: [mm/lru] 2e654b6306: stress-ng.hdd.ops_per_sec -69.5% regression
Date: Mon, 16 Mar 2020 09:52:46 +0800 [thread overview]
Message-ID: <20200316015246.GF11705@shao2-debian> (raw)
In-Reply-To: <e251ab96-929a-696b-0497-13f981362e31@linux.alibaba.com>
[-- Attachment #1: Type: text/plain, Size: 48390 bytes --]
On Fri, Mar 13, 2020 at 03:31:05PM +0800, Alex Shi wrote:
> Many thanks for the testing report.
>
> This commit is surpose to have much perf lose on lru_lock incrase, but later patch could fall back a lot on lru_lock reducing.
>
> Since the 69.5% drop is just on this commit, how much we lost for the whole patchset?
The regression of the patchset is 12%.
v5.6-rc4 634eb73be260286da9c1f1ecc3 testcase/testparams/testbox
---------------- -------------------------- ---------------------------
%stddev change %stddev
\ | \
1218565 -12% 1077320 stress-ng/io-performance-1HDD-100%-30s-ucode=0x500002c/lkp-csl-2sp5
1218565 -12% 1077320 GEO-MEAN stress-ng.hdd.ops_per_sec
Best Regards,
Rong Chen
>
> Thanks
> Alex
>
> 在 2020/3/12 上午10:43, kernel test robot 写道:
> > Greeting,
> >
> > FYI, we noticed a -69.5% regression of stress-ng.hdd.ops_per_sec due to commit:
> >
> >
> > commit: 2e654b630637ccdb649032ab6951fc517d8a3f52 ("mm/lru: replace pgdat lru_lock with lruvec lock")
> > https://github.com/alexshi/linux.git lrunextthp2
> >
> > in testcase: stress-ng
> > on test machine: 96 threads Intel(R) Xeon(R) Gold 6252 CPU @ 2.10GHz with 192G memory
> > with following parameters:
> >
> > nr_threads: 100%
> > disk: 1HDD
> > testtime: 30s
> > class: io
> > cpufreq_governor: performance
> > ucode: 0x500002c
> >
> >
> >
> >
> > If you fix the issue, kindly add following tag
> > Reported-by: kernel test robot <rong.a.chen@intel.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
> >
> > =========================================================================================
> > class/compiler/cpufreq_governor/disk/kconfig/nr_threads/rootfs/tbox_group/testcase/testtime/ucode:
> > io/gcc-7/performance/1HDD/x86_64-rhel-7.6/100%/debian-x86_64-20191114.cgz/lkp-csl-2sp5/stress-ng/30s/0x500002c
> >
> > commit:
> > 36b05905bc ("mm/mlock: clean up __munlock_isolate_lru_page")
> > 2e654b6306 ("mm/lru: replace pgdat lru_lock with lruvec lock")
> >
> > 36b05905bc83bcf1 2e654b630637ccdb649032ab695
> > ---------------- ---------------------------
> > %stddev %change %stddev
> > \ | \
> > 34916296 -69.5% 10661051 stress-ng.hdd.ops
> > 1163719 -69.5% 355316 stress-ng.hdd.ops_per_sec
> > 2.208e+08 -69.5% 67357030 stress-ng.revio.ops
> > 7360285 -69.5% 2245135 stress-ng.revio.ops_per_sec
> > 34211 -70.0% 10279 stress-ng.sync-file.ops
> > 1140 -70.0% 342.60 stress-ng.sync-file.ops_per_sec
> > 149847 ± 2% -38.9% 91549 ± 3% stress-ng.time.involuntary_context_switches
> > 12.21 -1.1% 12.08 iostat.cpu.user
> > 2953 -22.0% 2304 ± 2% vmstat.system.cs
> > 102059 -12.2% 89605 ± 2% meminfo.KReclaimable
> > 102059 -12.2% 89605 ± 2% meminfo.SReclaimable
> > 50376 ± 6% -21.6% 39482 ± 13% numa-meminfo.node1.KReclaimable
> > 50376 ± 6% -21.6% 39482 ± 13% numa-meminfo.node1.SReclaimable
> > 1.861e+09 ± 41% +58.6% 2.951e+09 cpuidle.C1E.time
> > 4607850 ± 21% +33.6% 6155007 cpuidle.C1E.usage
> > 6380 ± 9% +18.0% 7527 ± 6% cpuidle.POLL.usage
> > 4.183e+08 -69.5% 1.274e+08 numa-numastat.node0.local_node
> > 4.183e+08 -69.5% 1.274e+08 numa-numastat.node0.numa_hit
> > 4.179e+08 -69.5% 1.274e+08 numa-numastat.node1.local_node
> > 4.179e+08 -69.5% 1.274e+08 numa-numastat.node1.numa_hit
> > 62723 -30.1% 43819 ± 8% slabinfo.radix_tree_node.active_objs
> > 1133 -30.9% 782.75 ± 8% slabinfo.radix_tree_node.active_slabs
> > 63499 -30.9% 43869 ± 8% slabinfo.radix_tree_node.num_objs
> > 1133 -30.9% 782.75 ± 8% slabinfo.radix_tree_node.num_slabs
> > 1.725e+08 -69.1% 53265509 numa-vmstat.node0.numa_hit
> > 1.723e+08 -69.2% 53158587 numa-vmstat.node0.numa_local
> > 154929 ± 8% -31.0% 106965 ± 57% numa-vmstat.node0.numa_other
> > 12582 ± 5% -21.7% 9856 ± 13% numa-vmstat.node1.nr_slab_reclaimable
> > 1.723e+08 -69.1% 53318620 numa-vmstat.node1.numa_hit
> > 1.723e+08 -69.1% 53224009 numa-vmstat.node1.numa_local
> > 1809 ± 41% -60.4% 717.00 ±119% interrupts.CPU1.RES:Rescheduling_interrupts
> > 1319 ± 50% -66.9% 436.25 ± 95% interrupts.CPU13.RES:Rescheduling_interrupts
> > 1673 ± 63% -83.6% 274.75 ±120% interrupts.CPU61.RES:Rescheduling_interrupts
> > 1779 ±106% -79.2% 370.75 ±126% interrupts.CPU63.RES:Rescheduling_interrupts
> > 4926 ± 34% +60.1% 7889 interrupts.CPU68.NMI:Non-maskable_interrupts
> > 4926 ± 34% +60.1% 7889 interrupts.CPU68.PMI:Performance_monitoring_interrupts
> > 174.25 ± 3% +44.8% 252.25 ± 2% interrupts.IWI:IRQ_work_interrupts
> > 305.75 ± 24% +80.1% 550.75 ± 19% interrupts.TLB:TLB_shootdowns
> > 19090 ± 2% +9.2% 20853 proc-vmstat.nr_shmem
> > 25504 -12.2% 22398 ± 2% proc-vmstat.nr_slab_reclaimable
> > 8.354e+08 -69.6% 2.542e+08 proc-vmstat.numa_hit
> > 8.354e+08 -69.6% 2.541e+08 proc-vmstat.numa_local
> > 32042 ± 3% +15.0% 36851 proc-vmstat.pgactivate
> > 8.375e+08 -69.5% 2.55e+08 proc-vmstat.pgalloc_normal
> > 8.374e+08 -69.6% 2.549e+08 proc-vmstat.pgfree
> > 8.347e+08 -69.6% 2.537e+08 proc-vmstat.unevictable_pgs_culled
> > 273.38 +10.3% 301.44 sched_debug.cfs_rq:/.load_avg.max
> > 75.00 ± 46% +56.8% 117.62 ± 7% sched_debug.cfs_rq:/.removed.util_avg.max
> > 4598 -13.0% 4002 sched_debug.cpu.nr_switches.avg
> > 2193 ± 7% -35.9% 1406 ± 6% sched_debug.cpu.nr_switches.min
> > 2977 -20.3% 2373 sched_debug.cpu.sched_count.avg
> > 1316 ± 3% -53.2% 616.75 ± 2% sched_debug.cpu.sched_count.min
> > 2769 ± 8% +17.8% 3263 ± 6% sched_debug.cpu.sched_goidle.max
> > 1410 -21.2% 1111 sched_debug.cpu.ttwu_count.avg
> > 606.94 ± 2% -57.6% 257.44 ± 2% sched_debug.cpu.ttwu_count.min
> > 1071 ± 2% -27.9% 772.08 ± 3% sched_debug.cpu.ttwu_local.avg
> > 591.56 ± 2% -58.2% 247.56 sched_debug.cpu.ttwu_local.min
> > 2.245e+10 -9.5% 2.033e+10 perf-stat.i.branch-instructions
> > 1.808e+08 -5.4% 1.711e+08 perf-stat.i.branch-misses
> > 54.50 ± 2% -12.4 42.07 perf-stat.i.cache-miss-rate%
> > 2.99e+08 -30.1% 2.09e+08 perf-stat.i.cache-misses
> > 4.109e+08 -16.5% 3.431e+08 perf-stat.i.cache-references
> > 2921 -22.3% 2271 perf-stat.i.context-switches
> > 3.12 ± 3% +51.7% 4.73 perf-stat.i.cpi
> > 119.94 -5.8% 112.93 perf-stat.i.cpu-migrations
> > 2135 ± 7% +170.5% 5776 ± 5% perf-stat.i.cycles-between-cache-misses
> > 0.09 ± 42% -0.1 0.02 ± 5% perf-stat.i.dTLB-load-miss-rate%
> > 2.918e+10 -10.4% 2.613e+10 perf-stat.i.dTLB-loads
> > 0.03 ± 28% -0.0 0.02 ± 7% perf-stat.i.dTLB-store-miss-rate%
> > 1328065 ± 13% -60.2% 528688 ± 7% perf-stat.i.dTLB-store-misses
> > 1.215e+10 -16.4% 1.016e+10 perf-stat.i.dTLB-stores
> > 90.45 -1.3 89.13 perf-stat.i.iTLB-load-miss-rate%
> > 1.268e+11 -8.4% 1.162e+11 perf-stat.i.instructions
> > 0.56 -5.8% 0.53 perf-stat.i.ipc
> > 9428109 ± 68% -81.5% 1745740 ± 40% perf-stat.i.node-load-misses
> > 1.602e+08 ± 6% -13.7% 1.383e+08 ± 2% perf-stat.i.node-loads
> > 1903934 ±108% -83.3% 318054 ± 20% perf-stat.i.node-store-misses
> > 36769171 ± 6% -34.0% 24258343 perf-stat.i.node-stores
> > 3.25 -7.3% 3.01 perf-stat.overall.MPKI
> > 0.80 +0.0 0.83 perf-stat.overall.branch-miss-rate%
> > 72.49 -12.7 59.84 perf-stat.overall.cache-miss-rate%
> > 1.71 +11.4% 1.90 perf-stat.overall.cpi
> > 726.44 +45.4% 1056 perf-stat.overall.cycles-between-cache-misses
> > 0.01 ± 15% -0.0 0.01 ± 7% perf-stat.overall.dTLB-store-miss-rate%
> > 0.59 -10.2% 0.53 perf-stat.overall.ipc
> > 4.85 ±107% -3.5 1.33 ± 20% perf-stat.overall.node-store-miss-rate%
> > 2.218e+10 -11.2% 1.971e+10 perf-stat.ps.branch-instructions
> > 1.782e+08 -7.7% 1.644e+08 perf-stat.ps.branch-misses
> > 2.943e+08 -31.3% 2.022e+08 perf-stat.ps.cache-misses
> > 4.061e+08 -16.8% 3.379e+08 perf-stat.ps.cache-references
> > 2900 -22.2% 2256 perf-stat.ps.context-switches
> > 118.91 -5.8% 111.96 perf-stat.ps.cpu-migrations
> > 2.882e+10 -12.3% 2.529e+10 perf-stat.ps.dTLB-loads
> > 1410086 ± 14% -60.8% 553300 ± 7% perf-stat.ps.dTLB-store-misses
> > 1.2e+10 -18.5% 9.789e+09 perf-stat.ps.dTLB-stores
> > 1.251e+08 ± 3% -6.6% 1.168e+08 ± 2% perf-stat.ps.iTLB-load-misses
> > 1.251e+11 -10.3% 1.123e+11 perf-stat.ps.instructions
> > 9234161 ± 68% -81.8% 1679385 ± 40% perf-stat.ps.node-load-misses
> > 1.571e+08 ± 6% -15.4% 1.329e+08 ± 2% perf-stat.ps.node-loads
> > 1879361 ±106% -83.0% 319700 ± 19% perf-stat.ps.node-store-misses
> > 37089074 ± 6% -36.1% 23715352 perf-stat.ps.node-stores
> > 2.273e+13 -10.3% 2.039e+13 perf-stat.total.instructions
> > 82498 ± 2% -29.8% 57879 ± 3% softirqs.CPU0.RCU
> > 79134 -35.0% 51413 ± 4% softirqs.CPU1.RCU
> > 79615 ± 3% -38.6% 48918 ± 3% softirqs.CPU10.RCU
> > 78063 -37.3% 48932 ± 3% softirqs.CPU11.RCU
> > 78057 -37.6% 48696 ± 3% softirqs.CPU12.RCU
> > 80054 ± 3% -37.7% 49897 softirqs.CPU13.RCU
> > 78076 -34.9% 50860 ± 2% softirqs.CPU14.RCU
> > 77915 -36.4% 49543 ± 5% softirqs.CPU15.RCU
> > 80146 ± 2% -37.3% 50248 ± 5% softirqs.CPU16.RCU
> > 78713 -36.2% 50237 ± 6% softirqs.CPU17.RCU
> > 79503 ± 2% -38.5% 48890 ± 3% softirqs.CPU18.RCU
> > 79758 -35.9% 51148 ± 7% softirqs.CPU19.RCU
> > 78095 -36.5% 49553 ± 2% softirqs.CPU2.RCU
> > 78277 -34.0% 51653 ± 3% softirqs.CPU20.RCU
> > 78936 ± 3% -37.9% 48983 ± 2% softirqs.CPU21.RCU
> > 78588 -34.5% 51514 ± 5% softirqs.CPU22.RCU
> > 78483 -36.4% 49890 ± 3% softirqs.CPU23.RCU
> > 78979 -38.6% 48468 softirqs.CPU24.RCU
> > 80540 -37.3% 50481 ± 4% softirqs.CPU25.RCU
> > 79901 ± 4% -32.1% 54234 ± 11% softirqs.CPU26.RCU
> > 83231 ± 5% -38.5% 51164 ± 4% softirqs.CPU27.RCU
> > 78565 -37.8% 48872 ± 4% softirqs.CPU28.RCU
> > 79095 -38.9% 48358 ± 2% softirqs.CPU29.RCU
> > 80084 -37.1% 50390 ± 6% softirqs.CPU3.RCU
> > 78968 -38.9% 48279 ± 3% softirqs.CPU30.RCU
> > 78160 -36.4% 49702 ± 2% softirqs.CPU31.RCU
> > 79331 -39.2% 48212 ± 3% softirqs.CPU32.RCU
> > 79528 -38.8% 48692 ± 3% softirqs.CPU33.RCU
> > 79594 -38.2% 49159 ± 3% softirqs.CPU34.RCU
> > 79420 -39.2% 48307 ± 4% softirqs.CPU35.RCU
> > 78110 -34.9% 50888 softirqs.CPU36.RCU
> > 79064 -34.9% 51440 ± 7% softirqs.CPU37.RCU
> > 78142 -36.9% 49281 ± 7% softirqs.CPU38.RCU
> > 79441 -38.3% 49001 ± 4% softirqs.CPU39.RCU
> > 79427 -36.4% 50526 ± 2% softirqs.CPU4.RCU
> > 78010 -36.3% 49683 ± 8% softirqs.CPU40.RCU
> > 79556 -38.2% 49173 ± 3% softirqs.CPU41.RCU
> > 79054 -39.2% 48065 ± 3% softirqs.CPU42.RCU
> > 79484 -39.0% 48510 ± 3% softirqs.CPU43.RCU
> > 79119 -36.8% 50014 ± 4% softirqs.CPU44.RCU
> > 79095 -38.8% 48375 ± 4% softirqs.CPU45.RCU
> > 78723 -39.3% 47761 ± 3% softirqs.CPU46.RCU
> > 80129 ± 2% -34.4% 52542 ± 7% softirqs.CPU47.RCU
> > 78762 -36.4% 50124 ± 6% softirqs.CPU48.RCU
> > 79464 -38.2% 49083 ± 2% softirqs.CPU49.RCU
> > 78245 -37.1% 49184 ± 2% softirqs.CPU5.RCU
> > 78326 -36.2% 49972 ± 4% softirqs.CPU50.RCU
> > 80007 -35.4% 51689 ± 8% softirqs.CPU51.RCU
> > 79011 -36.7% 50020 ± 3% softirqs.CPU52.RCU
> > 78986 -39.5% 47794 softirqs.CPU53.RCU
> > 78679 -35.8% 50491 ± 5% softirqs.CPU54.RCU
> > 78852 ± 3% -38.9% 48186 ± 2% softirqs.CPU55.RCU
> > 78554 -35.8% 50445 ± 4% softirqs.CPU56.RCU
> > 78992 -37.1% 49673 ± 2% softirqs.CPU57.RCU
> > 79223 ± 2% -38.4% 48841 ± 2% softirqs.CPU58.RCU
> > 79503 ± 3% -36.6% 50406 ± 4% softirqs.CPU59.RCU
> > 78448 -37.4% 49141 ± 2% softirqs.CPU6.RCU
> > 79727 ± 3% -38.3% 49180 ± 5% softirqs.CPU60.RCU
> > 78422 -37.7% 48848 ± 3% softirqs.CPU61.RCU
> > 78695 -33.8% 52111 ± 7% softirqs.CPU62.RCU
> > 79143 -35.9% 50747 ± 5% softirqs.CPU63.RCU
> > 80073 ± 3% -34.1% 52786 ± 4% softirqs.CPU64.RCU
> > 79321 -36.4% 50483 ± 4% softirqs.CPU65.RCU
> > 79962 ± 2% -38.1% 49469 ± 5% softirqs.CPU66.RCU
> > 78647 -36.5% 49963 ± 3% softirqs.CPU67.RCU
> > 77809 ± 2% -36.9% 49124 softirqs.CPU68.RCU
> > 78901 ± 2% -37.3% 49503 ± 4% softirqs.CPU69.RCU
> > 79087 -35.4% 51107 ± 4% softirqs.CPU7.RCU
> > 78778 -36.1% 50344 ± 5% softirqs.CPU70.RCU
> > 79681 -38.3% 49136 ± 2% softirqs.CPU71.RCU
> > 78743 -37.7% 49020 ± 3% softirqs.CPU72.RCU
> > 80765 -40.0% 48482 ± 3% softirqs.CPU73.RCU
> > 79003 -38.1% 48885 ± 3% softirqs.CPU74.RCU
> > 80610 ± 4% -40.0% 48347 ± 2% softirqs.CPU75.RCU
> > 78991 -37.9% 49084 ± 3% softirqs.CPU76.RCU
> > 78655 -38.2% 48573 ± 3% softirqs.CPU77.RCU
> > 79120 -38.9% 48305 ± 2% softirqs.CPU78.RCU
> > 78639 -35.2% 50993 ± 5% softirqs.CPU79.RCU
> > 77965 -34.9% 50790 ± 3% softirqs.CPU8.RCU
> > 79147 -38.1% 49016 ± 5% softirqs.CPU80.RCU
> > 79784 -40.1% 47792 ± 3% softirqs.CPU81.RCU
> > 80938 ± 2% -39.9% 48663 ± 4% softirqs.CPU82.RCU
> > 79033 -39.1% 48123 ± 3% softirqs.CPU83.RCU
> > 78385 -36.8% 49534 ± 7% softirqs.CPU84.RCU
> > 79075 -37.8% 49196 ± 3% softirqs.CPU85.RCU
> > 79195 -39.7% 47772 ± 3% softirqs.CPU86.RCU
> > 79184 -38.4% 48782 ± 4% softirqs.CPU87.RCU
> > 78546 -39.4% 47629 ± 4% softirqs.CPU88.RCU
> > 78837 -38.9% 48201 ± 3% softirqs.CPU89.RCU
> > 78839 -38.1% 48785 ± 2% softirqs.CPU9.RCU
> > 79123 -39.9% 47586 ± 3% softirqs.CPU90.RCU
> > 78551 -37.0% 49461 ± 4% softirqs.CPU91.RCU
> > 78917 -39.5% 47763 ± 2% softirqs.CPU92.RCU
> > 78341 -38.9% 47861 ± 2% softirqs.CPU93.RCU
> > 78429 -39.2% 47718 softirqs.CPU94.RCU
> > 79007 -35.1% 51272 ± 3% softirqs.CPU95.RCU
> > 7595837 -37.3% 4763370 ± 2% softirqs.RCU
> > 38.14 -38.1 0.00 perf-profile.calltrace.cycles-pp._raw_spin_lock_irqsave.pagevec_lru_move_fn.__lru_cache_add.add_to_page_cache_lru.pagecache_get_page
> > 38.09 -38.1 0.00 perf-profile.calltrace.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_irqsave.pagevec_lru_move_fn.__lru_cache_add.add_to_page_cache_lru
> > 37.66 -37.7 0.00 perf-profile.calltrace.cycles-pp._raw_spin_lock_irqsave.release_pages.__pagevec_release.truncate_inode_pages_range.evict
> > 37.59 -37.6 0.00 perf-profile.calltrace.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_irqsave.release_pages.__pagevec_release.truncate_inode_pages_range
> > 42.79 -35.0 7.77 perf-profile.calltrace.cycles-pp.exit_to_usermode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe
> > 42.79 -35.0 7.77 perf-profile.calltrace.cycles-pp.task_work_run.exit_to_usermode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe
> > 42.79 -35.0 7.77 perf-profile.calltrace.cycles-pp.__fput.task_work_run.exit_to_usermode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe
> > 42.78 -35.0 7.76 perf-profile.calltrace.cycles-pp.__dentry_kill.dput.__fput.task_work_run.exit_to_usermode_loop
> > 42.78 -35.0 7.77 perf-profile.calltrace.cycles-pp.dput.__fput.task_work_run.exit_to_usermode_loop.do_syscall_64
> > 42.77 -35.0 7.76 perf-profile.calltrace.cycles-pp.evict.__dentry_kill.dput.__fput.task_work_run
> > 42.76 -35.0 7.76 perf-profile.calltrace.cycles-pp.truncate_inode_pages_range.evict.__dentry_kill.dput.__fput
> > 40.05 -33.5 6.56 perf-profile.calltrace.cycles-pp.release_pages.__pagevec_release.truncate_inode_pages_range.evict.__dentry_kill
> > 40.51 -33.4 7.09 perf-profile.calltrace.cycles-pp.__pagevec_release.truncate_inode_pages_range.evict.__dentry_kill.dput
> > 3.12 ± 2% -2.2 0.92 perf-profile.calltrace.cycles-pp.iov_iter_copy_from_user_atomic.generic_perform_write.__generic_file_write_iter.generic_file_write_iter.new_sync_write
> > 2.17 -1.5 0.64 perf-profile.calltrace.cycles-pp.copyin.iov_iter_copy_from_user_atomic.generic_perform_write.__generic_file_write_iter.generic_file_write_iter
> > 2.09 -1.5 0.62 perf-profile.calltrace.cycles-pp.copy_user_enhanced_fast_string.copyin.iov_iter_copy_from_user_atomic.generic_perform_write.__generic_file_write_iter
> > 1.13 -0.2 0.91 perf-profile.calltrace.cycles-pp.__pagevec_lru_add_fn.pagevec_lru_move_fn.__lru_cache_add.add_to_page_cache_lru.pagecache_get_page
> > 99.16 +0.4 99.58 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe
> > 99.05 +0.5 99.54 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe
> > 0.00 +0.5 0.51 perf-profile.calltrace.cycles-pp.lock_page_lruvec_irqsave.pagevec_lru_move_fn.lru_add_drain_cpu.__pagevec_release.truncate_inode_pages_range
> > 0.00 +0.5 0.51 perf-profile.calltrace.cycles-pp._raw_spin_lock_irqsave.lock_page_lruvec_irqsave.pagevec_lru_move_fn.lru_add_drain_cpu.__pagevec_release
> > 0.00 +0.5 0.52 perf-profile.calltrace.cycles-pp.pagevec_lru_move_fn.lru_add_drain_cpu.__pagevec_release.truncate_inode_pages_range.evict
> > 0.00 +0.5 0.52 perf-profile.calltrace.cycles-pp.lru_add_drain_cpu.__pagevec_release.truncate_inode_pages_range.evict.__dentry_kill
> > 0.00 +5.8 5.77 perf-profile.calltrace.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_irqsave.lock_page_lruvec_irqsave.release_pages.__pagevec_release
> > 0.00 +5.8 5.78 perf-profile.calltrace.cycles-pp._raw_spin_lock_irqsave.lock_page_lruvec_irqsave.release_pages.__pagevec_release.truncate_inode_pages_range
> > 0.00 +5.8 5.78 perf-profile.calltrace.cycles-pp.lock_page_lruvec_irqsave.release_pages.__pagevec_release.truncate_inode_pages_range.evict
> > 55.66 +35.9 91.56 perf-profile.calltrace.cycles-pp.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe
> > 55.53 +36.0 91.52 perf-profile.calltrace.cycles-pp.vfs_write.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe
> > 54.89 +36.4 91.32 perf-profile.calltrace.cycles-pp.new_sync_write.vfs_write.ksys_write.do_syscall_64.entry_SYSCALL_64_after_hwframe
> > 54.79 +36.5 91.28 perf-profile.calltrace.cycles-pp.generic_file_write_iter.new_sync_write.vfs_write.ksys_write.do_syscall_64
> > 54.42 +36.7 91.16 perf-profile.calltrace.cycles-pp.__generic_file_write_iter.generic_file_write_iter.new_sync_write.vfs_write.ksys_write
> > 53.78 +37.2 90.95 perf-profile.calltrace.cycles-pp.generic_perform_write.__generic_file_write_iter.generic_file_write_iter.new_sync_write.vfs_write
> > 48.54 +40.9 89.40 perf-profile.calltrace.cycles-pp.simple_write_begin.generic_perform_write.__generic_file_write_iter.generic_file_write_iter.new_sync_write
> > 43.71 +44.3 87.98 perf-profile.calltrace.cycles-pp.grab_cache_page_write_begin.simple_write_begin.generic_perform_write.__generic_file_write_iter.generic_file_write_iter
> > 43.66 +44.3 87.96 perf-profile.calltrace.cycles-pp.pagecache_get_page.grab_cache_page_write_begin.simple_write_begin.generic_perform_write.__generic_file_write_iter
> > 41.33 +46.0 87.31 perf-profile.calltrace.cycles-pp.add_to_page_cache_lru.pagecache_get_page.grab_cache_page_write_begin.simple_write_begin.generic_perform_write
> > 39.91 +47.0 86.88 perf-profile.calltrace.cycles-pp.__lru_cache_add.add_to_page_cache_lru.pagecache_get_page.grab_cache_page_write_begin.simple_write_begin
> > 39.79 +47.0 86.83 perf-profile.calltrace.cycles-pp.pagevec_lru_move_fn.__lru_cache_add.add_to_page_cache_lru.pagecache_get_page.grab_cache_page_write_begin
> > 0.00 +84.9 84.94 perf-profile.calltrace.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_irqsave.lock_page_lruvec_irqsave.pagevec_lru_move_fn.__lru_cache_add
> > 0.00 +85.4 85.36 perf-profile.calltrace.cycles-pp._raw_spin_lock_irqsave.lock_page_lruvec_irqsave.pagevec_lru_move_fn.__lru_cache_add.add_to_page_cache_lru
> > 0.00 +85.4 85.42 perf-profile.calltrace.cycles-pp.lock_page_lruvec_irqsave.pagevec_lru_move_fn.__lru_cache_add.add_to_page_cache_lru.pagecache_get_page
> > 42.79 -35.0 7.77 perf-profile.children.cycles-pp.exit_to_usermode_loop
> > 42.79 -35.0 7.77 perf-profile.children.cycles-pp.task_work_run
> > 42.79 -35.0 7.77 perf-profile.children.cycles-pp.__fput
> > 42.78 -35.0 7.77 perf-profile.children.cycles-pp.dput
> > 42.78 -35.0 7.76 perf-profile.children.cycles-pp.__dentry_kill
> > 42.77 -35.0 7.76 perf-profile.children.cycles-pp.evict
> > 42.77 -35.0 7.76 perf-profile.children.cycles-pp.truncate_inode_pages_range
> > 40.24 -33.5 6.71 perf-profile.children.cycles-pp.release_pages
> > 40.51 -33.4 7.09 perf-profile.children.cycles-pp.__pagevec_release
> > 3.13 -2.2 0.93 perf-profile.children.cycles-pp.iov_iter_copy_from_user_atomic
> > 2.20 -1.5 0.65 perf-profile.children.cycles-pp.copyin
> > 2.12 -1.5 0.63 perf-profile.children.cycles-pp.copy_user_enhanced_fast_string
> > 1.76 -1.3 0.47 perf-profile.children.cycles-pp.__alloc_pages_nodemask
> > 1.64 -1.2 0.48 perf-profile.children.cycles-pp.simple_write_end
> > 1.49 -1.1 0.39 ± 2% perf-profile.children.cycles-pp.get_page_from_freelist
> > 1.25 -1.0 0.25 perf-profile.children.cycles-pp.free_unref_page_list
> > 1.36 -0.9 0.41 ± 3% perf-profile.children.cycles-pp.__add_to_page_cache_locked
> > 1.12 -0.8 0.32 ± 2% perf-profile.children.cycles-pp.delete_from_page_cache_batch
> > 0.94 -0.8 0.17 ± 4% perf-profile.children.cycles-pp.free_pcppages_bulk
> > 0.83 -0.6 0.24 ± 3% perf-profile.children.cycles-pp.xas_store
> > 0.70 -0.4 0.27 perf-profile.children.cycles-pp.__list_del_entry_valid
> > 0.65 -0.4 0.21 perf-profile.children.cycles-pp.pagevec_lookup_entries
> > 0.65 -0.4 0.21 perf-profile.children.cycles-pp.find_get_entries
> > 0.48 ± 2% -0.3 0.15 ± 4% perf-profile.children.cycles-pp.xas_load
> > 0.41 -0.3 0.13 ± 3% perf-profile.children.cycles-pp.syscall_return_via_sysret
> > 0.39 ± 2% -0.3 0.12 ± 4% perf-profile.children.cycles-pp.find_get_entry
> > 1.16 -0.2 0.92 perf-profile.children.cycles-pp.__pagevec_lru_add_fn
> > 0.34 -0.2 0.11 ± 7% perf-profile.children.cycles-pp.mem_cgroup_commit_charge
> > 0.28 ± 2% -0.2 0.09 ± 4% perf-profile.children.cycles-pp.__mod_memcg_state
> > 0.34 ± 2% -0.2 0.15 ± 2% perf-profile.children.cycles-pp.__mod_lruvec_state
> > 0.28 -0.2 0.09 ± 4% perf-profile.children.cycles-pp.ksys_lseek
> > 0.26 -0.2 0.08 ± 5% perf-profile.children.cycles-pp.___might_sleep
> > 0.24 ± 3% -0.2 0.07 perf-profile.children.cycles-pp.truncate_cleanup_page
> > 0.21 -0.2 0.05 ± 9% perf-profile.children.cycles-pp.unaccount_page_cache_page
> > 0.23 ± 2% -0.2 0.08 perf-profile.children.cycles-pp.file_remove_privs
> > 0.21 ± 2% -0.2 0.06 ± 6% perf-profile.children.cycles-pp.mem_cgroup_charge_statistics
> > 0.21 -0.1 0.06 perf-profile.children.cycles-pp.xas_init_marks
> > 0.21 -0.1 0.07 ± 7% perf-profile.children.cycles-pp.entry_SYSCALL_64
> > 0.21 ± 3% -0.1 0.07 perf-profile.children.cycles-pp.iov_iter_fault_in_readable
> > 0.19 ± 3% -0.1 0.05 perf-profile.children.cycles-pp.__mod_node_page_state
> > 0.22 -0.1 0.08 ± 5% perf-profile.children.cycles-pp.security_file_permission
> > 0.19 ± 2% -0.1 0.06 ± 9% perf-profile.children.cycles-pp.mem_cgroup_uncharge_list
> > 0.20 -0.1 0.07 ± 7% perf-profile.children.cycles-pp.__fdget_pos
> > 0.15 ± 5% -0.1 0.03 ±100% perf-profile.children.cycles-pp.rcu_core
> > 0.15 ± 9% -0.1 0.04 ± 57% perf-profile.children.cycles-pp.mem_cgroup_try_charge
> > 0.16 ± 2% -0.1 0.05 ± 8% perf-profile.children.cycles-pp.__fget_light
> > 0.17 ± 4% -0.1 0.06 perf-profile.children.cycles-pp.__softirqentry_text_start
> > 0.16 -0.1 0.05 ± 8% perf-profile.children.cycles-pp.file_update_time
> > 0.17 ± 3% -0.1 0.06 perf-profile.children.cycles-pp.dentry_needs_remove_privs
> > 0.15 ± 2% -0.1 0.05 perf-profile.children.cycles-pp.down_write
> > 0.14 ± 3% -0.1 0.04 ± 57% perf-profile.children.cycles-pp._raw_spin_lock_irq
> > 0.14 ± 3% -0.1 0.04 ± 57% perf-profile.children.cycles-pp.security_inode_need_killpriv
> > 0.15 -0.1 0.06 perf-profile.children.cycles-pp.common_file_perm
> > 0.34 ± 2% -0.1 0.27 perf-profile.children.cycles-pp.smp_apic_timer_interrupt
> > 0.40 -0.1 0.34 ± 2% perf-profile.children.cycles-pp.apic_timer_interrupt
> > 0.26 ± 4% -0.0 0.21 ± 2% perf-profile.children.cycles-pp.hrtimer_interrupt
> > 0.18 ± 4% -0.0 0.14 perf-profile.children.cycles-pp.__hrtimer_run_queues
> > 0.14 ± 7% -0.0 0.10 ± 4% perf-profile.children.cycles-pp.tick_sched_timer
> > 0.12 ± 8% -0.0 0.09 perf-profile.children.cycles-pp.tick_sched_handle
> > 0.12 ± 5% -0.0 0.09 ± 4% perf-profile.children.cycles-pp.update_process_times
> > 0.11 ± 4% -0.0 0.08 ± 6% perf-profile.children.cycles-pp.mem_cgroup_page_lruvec
> > 0.08 -0.0 0.05 perf-profile.children.cycles-pp.irq_exit
> > 0.08 ± 8% -0.0 0.06 ± 7% perf-profile.children.cycles-pp.task_tick_fair
> > 0.09 ± 7% -0.0 0.07 ± 6% perf-profile.children.cycles-pp.scheduler_tick
> > 0.11 +0.0 0.12 ± 3% perf-profile.children.cycles-pp.mem_cgroup_update_lru_size
> > 0.15 ± 2% +0.0 0.19 ± 2% perf-profile.children.cycles-pp.__list_add_valid
> > 0.46 +0.1 0.54 perf-profile.children.cycles-pp.lru_add_drain_cpu
> > 0.00 +0.1 0.12 perf-profile.children.cycles-pp.shmem_write_begin
> > 0.00 +0.1 0.12 perf-profile.children.cycles-pp.shmem_getpage_gfp
> > 0.00 +0.1 0.13 perf-profile.children.cycles-pp.__libc_start_main
> > 0.00 +0.1 0.13 perf-profile.children.cycles-pp.main
> > 0.00 +0.1 0.13 perf-profile.children.cycles-pp.cmd_record
> > 0.00 +0.1 0.13 perf-profile.children.cycles-pp.perf_mmap__push
> > 0.00 +0.1 0.13 perf-profile.children.cycles-pp.__GI___libc_write
> > 99.21 +0.5 99.73 perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe
> > 99.11 +0.6 99.69 perf-profile.children.cycles-pp.do_syscall_64
> > 76.75 +14.7 91.41 perf-profile.children.cycles-pp.native_queued_spin_lock_slowpath
> > 76.30 +15.5 91.82 perf-profile.children.cycles-pp._raw_spin_lock_irqsave
> > 55.69 +36.0 91.69 perf-profile.children.cycles-pp.ksys_write
> > 55.55 +36.1 91.65 perf-profile.children.cycles-pp.vfs_write
> > 54.92 +36.5 91.45 perf-profile.children.cycles-pp.new_sync_write
> > 54.82 +36.6 91.41 perf-profile.children.cycles-pp.generic_file_write_iter
> > 54.45 +36.8 91.29 perf-profile.children.cycles-pp.__generic_file_write_iter
> > 54.00 +37.1 91.13 perf-profile.children.cycles-pp.generic_perform_write
> > 48.56 +40.8 89.40 perf-profile.children.cycles-pp.simple_write_begin
> > 43.72 +44.3 87.98 perf-profile.children.cycles-pp.grab_cache_page_write_begin
> > 43.67 +44.3 87.96 perf-profile.children.cycles-pp.pagecache_get_page
> > 41.33 +46.0 87.31 perf-profile.children.cycles-pp.add_to_page_cache_lru
> > 39.93 +47.1 87.01 perf-profile.children.cycles-pp.__lru_cache_add
> > 40.27 +47.2 87.51 perf-profile.children.cycles-pp.pagevec_lru_move_fn
> > 0.00 +91.9 91.87 perf-profile.children.cycles-pp.lock_page_lruvec_irqsave
> > 4.82 -3.4 1.41 perf-profile.self.cycles-pp.simple_write_begin
> > 2.09 -1.5 0.62 perf-profile.self.cycles-pp.copy_user_enhanced_fast_string
> > 1.40 -1.0 0.41 perf-profile.self.cycles-pp.simple_write_end
> > 0.76 ± 3% -0.5 0.23 perf-profile.self.cycles-pp.iov_iter_copy_from_user_atomic
> > 0.70 ± 2% -0.4 0.27 perf-profile.self.cycles-pp.__list_del_entry_valid
> > 0.61 -0.4 0.19 ± 2% perf-profile.self.cycles-pp.get_page_from_freelist
> > 0.58 -0.4 0.19 ± 2% perf-profile.self.cycles-pp.find_get_entries
> > 0.48 -0.4 0.12 ± 4% perf-profile.self.cycles-pp.free_pcppages_bulk
> > 0.41 -0.3 0.13 perf-profile.self.cycles-pp.syscall_return_via_sysret
> > 0.39 -0.3 0.11 ± 6% perf-profile.self.cycles-pp.xas_store
> > 0.39 -0.3 0.11 ± 4% perf-profile.self.cycles-pp.generic_perform_write
> > 0.36 -0.2 0.11 ± 4% perf-profile.self.cycles-pp.xas_load
> > 0.76 -0.2 0.54 ± 2% perf-profile.self.cycles-pp.__pagevec_lru_add_fn
> > 0.60 -0.2 0.39 perf-profile.self.cycles-pp.release_pages
> > 0.27 -0.2 0.09 ± 5% perf-profile.self.cycles-pp.__mod_memcg_state
> > 0.27 ± 3% -0.2 0.08 ± 5% perf-profile.self.cycles-pp.__add_to_page_cache_locked
> > 0.25 ± 2% -0.2 0.08 ± 6% perf-profile.self.cycles-pp.___might_sleep
> > 0.24 -0.2 0.07 perf-profile.self.cycles-pp.delete_from_page_cache_batch
> > 0.21 -0.1 0.07 ± 7% perf-profile.self.cycles-pp.entry_SYSCALL_64
> > 0.21 ± 3% -0.1 0.07 perf-profile.self.cycles-pp.iov_iter_fault_in_readable
> > 0.18 ± 4% -0.1 0.05 perf-profile.self.cycles-pp.__mod_node_page_state
> > 0.16 ± 2% -0.1 0.04 ± 57% perf-profile.self.cycles-pp.truncate_inode_pages_range
> > 0.15 ± 2% -0.1 0.05 perf-profile.self.cycles-pp.__alloc_pages_nodemask
> > 0.15 ± 2% -0.1 0.05 perf-profile.self.cycles-pp.__fget_light
> > 0.11 ± 6% -0.1 0.03 ±100% perf-profile.self.cycles-pp.do_syscall_64
> > 0.12 ± 3% -0.1 0.05 perf-profile.self.cycles-pp.common_file_perm
> > 0.10 ± 4% -0.0 0.07 perf-profile.self.cycles-pp.__mod_lruvec_state
> > 0.10 ± 4% -0.0 0.08 ± 6% perf-profile.self.cycles-pp.mem_cgroup_page_lruvec
> > 0.11 ± 4% +0.0 0.12 perf-profile.self.cycles-pp.mem_cgroup_update_lru_size
> > 0.07 ± 5% +0.0 0.11 ± 4% perf-profile.self.cycles-pp._raw_spin_unlock_irqrestore
> > 0.14 +0.0 0.18 ± 3% perf-profile.self.cycles-pp.__list_add_valid
> > 0.13 +0.3 0.44 ± 3% perf-profile.self.cycles-pp._raw_spin_lock_irqsave
> > 76.75 +14.7 91.41 perf-profile.self.cycles-pp.native_queued_spin_lock_slowpath
> >
> >
> >
> > stress-ng.hdd.ops
> >
> > 4e+07 +-----------------------------------------------------------------+
> > |.+.+..+.+.+.+.+..+.+.+.+.+ +.+..+ +..+.+. |
> > 3.5e+07 |-+ : : : : +.+.+..+.+.|
> > 3e+07 |-O O O O O O O : O O O : O : O O O : |
> > | : : : : |
> > 2.5e+07 |-+ : : : : |
> > | : : : : |
> > 2e+07 |-+ : : : : |
> > | : : : : |
> > 1.5e+07 |-+ : : : : |
> > 1e+07 |-+ : : : : O O O O O O |
> > | : : : : |
> > 5e+06 |-+ : : : : |
> > | : : : : |
> > 0 +-----------------------------------------------------------------+
> >
> >
> > stress-ng.hdd.ops_per_sec
> >
> > 1.4e+06 +-----------------------------------------------------------------+
> > | |
> > 1.2e+06 |.+.+..+.+.+.+.+..+.+.+.+.+ +.+..+ +..+.+.+.+.+..+.+.|
> > | O O O O O : O O O : O : O O O : |
> > 1e+06 |-O O : : : : |
> > | : : : : |
> > 800000 |-+ : : : : |
> > | : : : : |
> > 600000 |-+ : : : : |
> > | : : : : |
> > 400000 |-+ : : : : O O O O O O |
> > | : : : : |
> > 200000 |-+ : : : : |
> > | : : : : |
> > 0 +-----------------------------------------------------------------+
> >
> >
> > stress-ng.revio.ops
> >
> > 2.5e+08 +-----------------------------------------------------------------+
> > |.+.+..+.+.+.+.+..+.+.+.+.+ +.+..+ +..+.+. |
> > | : : : O : +.+.+..+.+.|
> > 2e+08 |-O O O O O O O : O O O : O : O O : |
> > | : : : : |
> > | : : : : |
> > 1.5e+08 |-+ : : : : |
> > | : : : : |
> > 1e+08 |-+ : : : : |
> > | : : : : |
> > | : : : : O O O O O O |
> > 5e+07 |-+ : : : : |
> > | : : : : |
> > | : : : : |
> > 0 +-----------------------------------------------------------------+
> >
> >
> > stress-ng.revio.ops_per_sec
> >
> > 8e+06 +-------------------------------------------------------------------+
> > |.+..+.+.+.+..+.+.+ +.+.+ +..+.+ +.+.+..+.+.+.+..+.|
> > 7e+06 |-O O O O O O O : O O O : O : O O O : |
> > 6e+06 |-+ : : : : |
> > | : : : : |
> > 5e+06 |-+ : : : : |
> > | : : : : |
> > 4e+06 |-+ : : : : |
> > | : : : : |
> > 3e+06 |-+ : : : : |
> > 2e+06 |-+ : : : : O O O O O O |
> > | : : : : |
> > 1e+06 |-+ : : : : |
> > | : : : : |
> > 0 +-------------------------------------------------------------------+
> >
> >
> > stress-ng.sync-file.ops
> >
> > 40000 +-------------------------------------------------------------------+
> > | .+.+.+..+.+.+ +..+.+ +.+.+.. |
> > 35000 |.+..+.+.+.+..+ : : : : +.+.+.+..+.|
> > 30000 |-O O O O O O O : O O O : O : O O O : |
> > | : : : : |
> > 25000 |-+ : : : : |
> > | : : : : |
> > 20000 |-+ : : : : |
> > | : : : : |
> > 15000 |-+ : : : : |
> > 10000 |-+ : : : : O O O O O O |
> > | : : : : |
> > 5000 |-+ : : : : |
> > | : : : : |
> > 0 +-------------------------------------------------------------------+
> >
> >
> > stress-ng.sync-file.ops_per_sec
> >
> > 1400 +--------------------------------------------------------------------+
> > | |
> > 1200 |.+..+.+.+..+.+.+.+..+.+.+..+ +.+.+ +.+.+.. |
> > | O O O O O : O O O : O : O O O : +.+.+.+..+.|
> > 1000 |-O O : : : : |
> > | : : : : |
> > 800 |-+ : : : : |
> > | : : : : |
> > 600 |-+ : : : : |
> > | : : : : |
> > 400 |-+ : : : : |
> > | : : : : O O O O O O |
> > 200 |-+ : : : : |
> > | : : : : |
> > 0 +--------------------------------------------------------------------+
> >
> >
> > [*] 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
> >
next parent reply other threads:[~2020-03-16 1:52 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <e251ab96-929a-696b-0497-13f981362e31@linux.alibaba.com>
2020-03-16 1:52 ` Rong Chen [this message]
2020-03-12 2:43 [mm/lru] 2e654b6306: stress-ng.hdd.ops_per_sec -69.5% regression kernel test robot
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=20200316015246.GF11705@shao2-debian \
--to=rong.a.chen@intel.com \
--cc=lkp@lists.01.org \
/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.