FYI, we noticed the below changes on https://github.com/0day-ci/linux Vladimir-Davydov/vmpressure-Fix-subtree-pressure-detection/20160128-003153 commit 78d1788350477246d516496d8d7684fa80ef7f18 ("vmpressure: Fix subtree pressure detection") ========================================================================================= compiler/cpufreq_governor/kconfig/rootfs/runtime/tbox_group/test/testcase: gcc-4.9/performance/x86_64-rhel/debian-x86_64-2015-02-07.cgz/300s/ivb43/lru-file-mmap-read/vm-scalability commit: v4.5-rc1 78d1788350477246d516496d8d7684fa80ef7f18 v4.5-rc1 78d1788350477246d516496d8d ---------------- -------------------------- %stddev %change %stddev \ | \ 195486 ± 1% +274.1% 731248 ± 2% vm-scalability.time.involuntary_context_switches 2939 ± 21% +37.5% 4041 ± 12% numa-vmstat.node1.nr_anon_pages 897378 ± 1% +17.7% 1056357 ± 2% softirqs.RCU 4293 ± 4% +82.5% 7835 ± 2% vmstat.system.cs 120.40 ± 22% +41.6% 170.50 ± 19% cpuidle.C3-IVT.usage 4.723e+08 ± 4% +13.8% 5.375e+08 ± 5% cpuidle.C6-IVT.time 93336 ± 31% +611.8% 664396 ± 88% latency_stats.max.pipe_wait.pipe_read.__vfs_read.vfs_read.SyS_read.entry_SYSCALL_64_fastpath 371333 ±166% -70.1% 110861 ±117% latency_stats.sum.pipe_write.__vfs_write.vfs_write.SyS_write.entry_SYSCALL_64_fastpath 195486 ± 1% +274.1% 731248 ± 2% time.involuntary_context_switches 717.20 ± 14% +20.3% 862.50 ± 6% time.major_page_faults 2724 ± 60% -93.6% 174.25 ±173% numa-meminfo.node0.AnonHugePages 1630 ±122% +222.2% 5254 ± 16% numa-meminfo.node1.AnonHugePages 11760 ± 21% +37.6% 16177 ± 12% numa-meminfo.node1.AnonPages 1.01 ± 4% +26.5% 1.28 ± 10% perf-profile.cycles-pp.__add_to_page_cache_locked.add_to_page_cache_lru.mpage_readpages.xfs_vm_readpages.__do_page_cache_readahead 0.90 ± 56% -72.4% 0.25 ± 95% perf-profile.cycles-pp.__alloc_pages_nodemask.alloc_pages_current.__page_cache_alloc.__do_page_cache_readahead.filemap_fault 0.90 ± 56% -72.5% 0.25 ± 95% perf-profile.cycles-pp.__page_cache_alloc.__do_page_cache_readahead.filemap_fault.xfs_filemap_fault.__do_fault 2.18 ± 2% +15.9% 2.52 ± 7% perf-profile.cycles-pp.__remove_mapping.shrink_page_list.shrink_inactive_list.shrink_zone_memcg.shrink_zone 1.04 ± 4% +16.8% 1.21 ± 7% perf-profile.cycles-pp.__rmqueue.get_page_from_freelist.__alloc_pages_nodemask.alloc_pages_current.__page_cache_alloc 1.08 ± 6% +18.1% 1.28 ± 5% perf-profile.cycles-pp._raw_spin_lock_irqsave.__remove_mapping.shrink_page_list.shrink_inactive_list.shrink_zone_memcg 0.90 ± 56% -72.4% 0.25 ± 95% perf-profile.cycles-pp.alloc_pages_current.__page_cache_alloc.__do_page_cache_readahead.filemap_fault.xfs_filemap_fault 0.77 ± 20% +47.0% 1.14 ± 9% perf-profile.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_irqsave.__remove_mapping.shrink_page_list.shrink_inactive_list 1.38 ± 2% +15.8% 1.60 ± 7% perf-profile.cycles-pp.rmap_walk.try_to_unmap.shrink_page_list.shrink_inactive_list.shrink_zone_memcg 2.17 ± 2% -14.7% 1.85 ± 9% perf-profile.cycles-pp.shrink_inactive_list.shrink_zone_memcg.shrink_zone.kswapd.kthread 2.19 ± 2% -14.5% 1.87 ± 9% perf-profile.cycles-pp.shrink_zone.kswapd.kthread.ret_from_fork 2.18 ± 1% -14.8% 1.86 ± 9% perf-profile.cycles-pp.shrink_zone_memcg.shrink_zone.kswapd.kthread.ret_from_fork 1.46 ± 2% +19.5% 1.75 ± 10% perf-profile.cycles-pp.try_to_unmap.shrink_page_list.shrink_inactive_list.shrink_zone_memcg.shrink_zone 50.20 ±133% +273.0% 187.25 ±109% sched_debug.cfs_rq:/.load.5 56.60 ± 11% +97.4% 111.75 ± 55% sched_debug.cfs_rq:/.load_avg.0 17.60 ± 9% +25.0% 22.00 ± 14% sched_debug.cfs_rq:/.load_avg.19 47.00 ± 33% -54.8% 21.25 ± 13% sched_debug.cfs_rq:/.load_avg.23 21.60 ± 30% +139.6% 51.75 ± 63% sched_debug.cfs_rq:/.load_avg.30 20.80 ± 30% +50.2% 31.25 ± 26% sched_debug.cfs_rq:/.load_avg.36 16.40 ± 2% +18.9% 19.50 ± 11% sched_debug.cfs_rq:/.runnable_load_avg.14 16685 ±278% -374.2% -45753 ±-25% sched_debug.cfs_rq:/.spread0.12 119599 ± 47% -71.8% 33762 ±121% sched_debug.cfs_rq:/.spread0.27 106418 ± 42% -267.6% -178399 ±-221% sched_debug.cfs_rq:/.spread0.28 56797 ± 81% -136.5% -20757 ±-159% sched_debug.cfs_rq:/.spread0.3 63128 ± 63% -157.9% -36549 ±-295% sched_debug.cfs_rq:/.spread0.46 57759 ±100% -676.1% -332737 ±-164% sched_debug.cfs_rq:/.spread0.8 151767 ± 32% -40.2% 90684 ± 14% sched_debug.cfs_rq:/.spread0.max 855.00 ± 1% +7.5% 918.75 ± 7% sched_debug.cfs_rq:/.util_avg.20 16.40 ± 2% +18.9% 19.50 ± 11% sched_debug.cpu.cpu_load[0].14 16.40 ± 2% +18.9% 19.50 ± 11% sched_debug.cpu.cpu_load[1].14 16.40 ± 2% +18.9% 19.50 ± 11% sched_debug.cpu.cpu_load[2].14 16.40 ± 2% +18.9% 19.50 ± 11% sched_debug.cpu.cpu_load[3].14 16.40 ± 2% +18.9% 19.50 ± 11% sched_debug.cpu.cpu_load[4].14 1305 ± 1% +16.6% 1521 ± 13% sched_debug.cpu.curr->pid.23 50.20 ±133% +273.0% 187.25 ±109% sched_debug.cpu.load.5 0.00 ± 0% +Inf% 1.00 ± 0% sched_debug.cpu.nr_running.5 10459 ± 30% +86.1% 19462 ± 13% sched_debug.cpu.nr_switches.12 11042 ±113% +169.5% 29759 ± 31% sched_debug.cpu.nr_switches.14 17516 ± 69% +85.9% 32567 ± 46% sched_debug.cpu.nr_switches.18 8287 ± 69% +271.4% 30775 ± 71% sched_debug.cpu.nr_switches.21 16525 ± 59% +80.6% 29850 ± 41% sched_debug.cpu.nr_switches.23 8306 ± 36% +92.8% 16016 ± 8% sched_debug.cpu.nr_switches.24 9271 ± 64% +149.7% 23153 ± 30% sched_debug.cpu.nr_switches.25 6094 ± 49% +243.8% 20951 ± 31% sched_debug.cpu.nr_switches.26 6569 ± 32% +194.9% 19373 ± 17% sched_debug.cpu.nr_switches.30 10034 ±111% +295.8% 39717 ± 41% sched_debug.cpu.nr_switches.31 7417 ± 70% +205.4% 22657 ± 35% sched_debug.cpu.nr_switches.38 9204 ± 69% +125.4% 20746 ± 26% sched_debug.cpu.nr_switches.42 8742 ± 68% +172.8% 23845 ± 14% sched_debug.cpu.nr_switches.47 11267 ± 75% +311.6% 46375 ± 68% sched_debug.cpu.nr_switches.7 14373 ± 4% +74.8% 25131 ± 3% sched_debug.cpu.nr_switches.avg 68237 ± 15% +30.3% 88895 ± 10% sched_debug.cpu.nr_switches.max 1792 ± 11% +610.9% 12741 ± 4% sched_debug.cpu.nr_switches.min -1.00 ±-244% -200.0% 1.00 ±212% sched_debug.cpu.nr_uninterruptible.21 -29.60 ±-22% -31.7% -20.21 ±-29% sched_debug.cpu.nr_uninterruptible.min 11570 ± 27% +75.8% 20344 ± 14% sched_debug.cpu.sched_count.12 18187 ± 52% +67.7% 30504 ± 40% sched_debug.cpu.sched_count.23 8874 ± 33% +83.7% 16301 ± 7% sched_debug.cpu.sched_count.24 9847 ± 60% +137.9% 23423 ± 29% sched_debug.cpu.sched_count.25 7299 ± 38% +190.8% 21225 ± 30% sched_debug.cpu.sched_count.26 7156 ± 30% +179.8% 20022 ± 16% sched_debug.cpu.sched_count.30 10606 ±105% +279.7% 40278 ± 41% sched_debug.cpu.sched_count.31 7977 ± 66% +187.5% 22938 ± 35% sched_debug.cpu.sched_count.38 10713 ± 47% +138.5% 25554 ± 54% sched_debug.cpu.sched_count.39 25911 ± 63% +76.9% 45826 ± 39% sched_debug.cpu.sched_count.4 9758 ± 66% +115.6% 21039 ± 26% sched_debug.cpu.sched_count.42 9252 ± 64% +160.6% 24110 ± 14% sched_debug.cpu.sched_count.47 12153 ± 69% +296.0% 48126 ± 65% sched_debug.cpu.sched_count.7 110528 ± 0% +9.1% 120637 ± 0% sched_debug.cpu.sched_count.avg 2364 ± 8% +455.3% 13129 ± 5% sched_debug.cpu.sched_count.min 816.60 ± 32% -40.5% 485.75 ± 16% sched_debug.cpu.sched_goidle.17 383.20 ± 22% +56.1% 598.00 ± 22% sched_debug.cpu.sched_goidle.23 204.20 ± 12% +210.7% 634.50 ± 86% sched_debug.cpu.sched_goidle.25 190.60 ± 26% +54.6% 294.75 ± 12% sched_debug.cpu.sched_goidle.31 4752 ± 36% +107.1% 9844 ± 12% sched_debug.cpu.ttwu_count.12 5428 ±115% +178.2% 15102 ± 31% sched_debug.cpu.ttwu_count.14 8902 ± 73% +80.8% 16095 ± 47% sched_debug.cpu.ttwu_count.18 3698 ± 62% +315.1% 15351 ± 75% sched_debug.cpu.ttwu_count.21 8371 ± 57% +80.6% 15121 ± 41% sched_debug.cpu.ttwu_count.23 4569 ± 33% +85.9% 8496 ± 10% sched_debug.cpu.ttwu_count.24 4884 ± 56% +152.6% 12339 ± 27% sched_debug.cpu.ttwu_count.25 3187 ± 39% +258.7% 11434 ± 23% sched_debug.cpu.ttwu_count.26 3513 ± 31% +183.5% 9962 ± 19% sched_debug.cpu.ttwu_count.30 5462 ± 97% +283.9% 20970 ± 41% sched_debug.cpu.ttwu_count.31 5605 ± 43% +85.8% 10415 ± 34% sched_debug.cpu.ttwu_count.34 4011 ± 66% +190.7% 11661 ± 33% sched_debug.cpu.ttwu_count.38 4305 ± 41% +183.7% 12215 ± 59% sched_debug.cpu.ttwu_count.39 4828 ± 73% +125.9% 10906 ± 25% sched_debug.cpu.ttwu_count.42 4560 ± 74% +168.1% 12227 ± 12% sched_debug.cpu.ttwu_count.47 5663 ± 78% +313.7% 23431 ± 67% sched_debug.cpu.ttwu_count.7 7326 ± 4% +74.8% 12808 ± 3% sched_debug.cpu.ttwu_count.avg 34268 ± 15% +29.6% 44398 ± 9% sched_debug.cpu.ttwu_count.max 792.03 ± 20% +698.7% 6325 ± 5% sched_debug.cpu.ttwu_count.min 3951 ± 45% +119.7% 8681 ± 13% sched_debug.cpu.ttwu_local.12 4585 ±137% +205.0% 13987 ± 33% sched_debug.cpu.ttwu_local.14 8045 ± 80% +88.6% 15173 ± 48% sched_debug.cpu.ttwu_local.18 2941 ± 79% +386.9% 14319 ± 79% sched_debug.cpu.ttwu_local.21 7262 ± 67% +95.1% 14172 ± 44% sched_debug.cpu.ttwu_local.23 3991 ± 40% +100.4% 8000 ± 10% sched_debug.cpu.ttwu_local.24 4237 ± 68% +162.6% 11128 ± 29% sched_debug.cpu.ttwu_local.25 2570 ± 41% +292.2% 10081 ± 30% sched_debug.cpu.ttwu_local.26 2891 ± 40% +215.8% 9132 ± 21% sched_debug.cpu.ttwu_local.30 4665 ±113% +318.2% 19509 ± 43% sched_debug.cpu.ttwu_local.31 4533 ± 50% +115.0% 9748 ± 35% sched_debug.cpu.ttwu_local.34 4131 ± 56% +145.5% 10144 ± 47% sched_debug.cpu.ttwu_local.37 3349 ± 82% +225.3% 10897 ± 34% sched_debug.cpu.ttwu_local.38 3630 ± 48% +219.9% 11611 ± 63% sched_debug.cpu.ttwu_local.39 4138 ± 82% +140.4% 9947 ± 29% sched_debug.cpu.ttwu_local.42 3356 ± 78% +238.4% 11359 ± 14% sched_debug.cpu.ttwu_local.47 4866 ± 91% +360.6% 22415 ± 71% sched_debug.cpu.ttwu_local.7 6406 ± 5% +84.0% 11787 ± 4% sched_debug.cpu.ttwu_local.avg 33023 ± 15% +29.7% 42825 ± 11% sched_debug.cpu.ttwu_local.max 450.27 ± 20% +1199.8% 5852 ± 4% sched_debug.cpu.ttwu_local.min 1.58 ± 4% -55.1% 0.71 ± 99% sched_debug.rt_rq:/.rt_time.12 ivb43: Ivytown Ivy Bridge-EP Memory: 64G vm-scalability.time.involuntary_context_switches 800000 ++-----------------------------------------------------------------+ | O O O O O O O | 700000 ++ O O O O O O O O O O O | | 600000 O+ | | O O | 500000 ++ O O O O | | | 400000 ++ | | | 300000 ++ | | | 200000 *+.*..*.*..*..*..*..*.*..*..*..*..*.*..*..*..*.*..*..*..*..*.* | | | 100000 ++-----------------------------------------------------------------+ [*] bisect-good sample [O] bisect-bad sample To reproduce: git clone git://git.kernel.org/pub/scm/linux/kernel/git/wfg/lkp-tests.git cd lkp-tests bin/lkp install job.yaml # job file is attached in this email bin/lkp run job.yaml 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, Ying Huang