lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
Open Source and information security mailing list archives
| ||
|
Date: Mon, 1 Aug 2016 09:55:04 +0800 From: kernel test robot <xiaolong.ye@...el.com> To: William Roberts <william.c.roberts@...el.com> Cc: 0day robot <fengguang.wu@...el.com>, LKML <linux-kernel@...r.kernel.org>, lkp@...org Subject: [lkp] eadd46295d: hackbench.throughput -9.7% regression FYI, we noticed a -9.7% regression of hackbench.throughput due to commit: commit eadd46295d4e47aef2fc91e806282f61d4bfe2a2 ("Introduce mmap randomization") https://github.com/0day-ci/linux william-c-roberts-intel-com/Introduce-mmap-randomization/20160727-023413 in testcase: hackbench on test machine: 32 threads Sandy Bridge-EP with 32G memory with following parameters: nr_threads: 1600% mode: process ipc: pipe cpufreq_governor: performance 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. Details are as below: --------------------------------------------------------------------------------------------------> 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 ========================================================================================= compiler/cpufreq_governor/ipc/kconfig/mode/nr_threads/rootfs/tbox_group/testcase: gcc-6/performance/pipe/x86_64-rhel/process/1600%/debian-x86_64-2015-02-07.cgz/lkp-snb01/hackbench commit: v4.7 eadd46295d ("Introduce mmap randomization") v4.7 eadd46295d4e47aef2fc91e806 ---------------- -------------------------- fail:runs %reproduction fail:runs | | | %stddev %change %stddev \ | \ 235060 ± 1% -9.7% 212347 ± 0% hackbench.throughput 3.989e+08 ± 1% +12.0% 4.467e+08 ± 2% hackbench.time.involuntary_context_switches 26892935 ± 1% -11.6% 23769774 ± 1% hackbench.time.minor_page_faults 2911 ± 0% -1.6% 2864 ± 0% hackbench.time.percent_of_cpu_this_job_got 1415 ± 0% -7.1% 1315 ± 1% hackbench.time.user_time 1.59e+09 ± 0% -3.5% 1.534e+09 ± 1% hackbench.time.voluntary_context_switches 522230 ± 0% +99.2% 1040525 ± 0% meminfo.PageTables 527517 ± 3% -16.5% 440493 ± 3% softirqs.SCHED 67535 ± 7% +99.2% 134556 ± 10% numa-vmstat.node0.nr_page_table_pages 61527 ± 8% +100.6% 123407 ± 9% numa-vmstat.node1.nr_page_table_pages 6.28 ± 1% +16.9% 7.34 ± 2% turbostat.CPU%c1 171.44 ± 0% -1.7% 168.54 ± 0% turbostat.CorWatt 198.59 ± 0% -1.4% 195.72 ± 0% turbostat.PkgWatt 1.40 ±200% +507.1% 8.50 ± 64% vmstat.procs.b 1791 ± 2% -10.5% 1603 ± 1% vmstat.procs.r 155718 ± 3% -17.2% 128979 ± 6% vmstat.system.in 1624167 ± 6% +41.9% 2305196 ± 12% numa-meminfo.node0.MemUsed 272208 ± 6% +99.2% 542347 ± 10% numa-meminfo.node0.PageTables 1443799 ± 8% +37.7% 1988433 ± 12% numa-meminfo.node1.MemUsed 246214 ± 8% +100.8% 494381 ± 10% numa-meminfo.node1.PageTables 129902 ± 0% +100.5% 260449 ± 0% proc-vmstat.nr_page_table_pages 52961320 ± 1% +16.4% 61651759 ± 1% proc-vmstat.numa_hit 52961316 ± 1% +16.4% 61651755 ± 1% proc-vmstat.numa_local 5674 ± 3% -15.0% 4822 ± 6% proc-vmstat.numa_pte_updates 52794094 ± 0% +13.6% 59962331 ± 1% proc-vmstat.pgalloc_normal 28064259 ± 1% -11.4% 24875700 ± 1% proc-vmstat.pgfault 58082481 ± 1% +14.0% 66223801 ± 1% proc-vmstat.pgfree 1.044e+08 ± 1% -22.3% 81133470 ± 6% cpuidle.C1-SNB.time 19769566 ± 2% -20.9% 15634083 ± 6% cpuidle.C1-SNB.usage 40316906 ± 3% +18.4% 47742318 ± 2% cpuidle.C1E-SNB.time 461352 ± 2% +24.7% 575405 ± 2% cpuidle.C1E-SNB.usage 8978132 ± 3% +92.1% 17248854 ± 6% cpuidle.C3-SNB.time 42921 ± 2% +157.3% 110419 ± 5% cpuidle.C3-SNB.usage 1.13e+09 ± 1% +19.1% 1.346e+09 ± 1% cpuidle.C7-SNB.time 1191984 ± 1% +18.6% 1414163 ± 2% cpuidle.C7-SNB.usage 4.157e+08 ± 5% +22.0% 5.073e+08 ± 6% cpuidle.POLL.time 333689 ± 2% -19.7% 268091 ± 6% cpuidle.POLL.usage 252963 ± 0% +24.4% 314715 ± 0% slabinfo.anon_vma.active_objs 5053 ± 0% +24.1% 6273 ± 0% slabinfo.anon_vma.active_slabs 257762 ± 0% +24.1% 319960 ± 0% slabinfo.anon_vma.num_objs 5053 ± 0% +24.1% 6273 ± 0% slabinfo.anon_vma.num_slabs 516841 ± 0% +23.8% 640090 ± 0% slabinfo.anon_vma_chain.active_objs 9773 ± 0% +24.3% 12152 ± 0% slabinfo.anon_vma_chain.active_slabs 625521 ± 0% +24.3% 777796 ± 0% slabinfo.anon_vma_chain.num_objs 9773 ± 0% +24.3% 12152 ± 0% slabinfo.anon_vma_chain.num_slabs 3023 ± 7% -13.4% 2617 ± 5% slabinfo.kmalloc-2048.active_objs 3093 ± 7% -12.6% 2704 ± 5% slabinfo.kmalloc-2048.num_objs 384082 ± 0% +15.6% 444060 ± 0% slabinfo.vm_area_struct.active_objs 8845 ± 0% +15.5% 10213 ± 0% slabinfo.vm_area_struct.active_slabs 389216 ± 0% +15.5% 449404 ± 0% slabinfo.vm_area_struct.num_objs 8845 ± 0% +15.5% 10213 ± 0% slabinfo.vm_area_struct.num_slabs 7.579e+12 ± 0% -9.0% 6.896e+12 ± 1% perf-stat.branch-instructions 0.23 ± 81% +71.9% 0.40 ± 0% perf-stat.branch-miss-rate 2.993e+10 ± 1% -7.2% 2.779e+10 ± 1% perf-stat.branch-misses 2.075e+10 ± 0% -6.9% 1.932e+10 ± 2% perf-stat.cache-misses 2.076e+11 ± 0% -3.9% 1.996e+11 ± 2% perf-stat.cache-references 47671486 ± 2% -17.4% 39397933 ± 5% perf-stat.cpu-migrations 0.71 ± 83% +297.9% 2.81 ± 8% perf-stat.dTLB-load-miss-rate 1.537e+11 ± 13% +109.4% 3.22e+11 ± 6% perf-stat.dTLB-load-misses 1.268e+13 ± 0% -9.4% 1.148e+13 ± 1% perf-stat.dTLB-loads 0.26 ± 82% +243.6% 0.88 ± 6% perf-stat.dTLB-store-miss-rate 3.519e+10 ± 7% +86.0% 6.546e+10 ± 5% perf-stat.dTLB-store-misses 8.278e+12 ± 0% -9.7% 7.475e+12 ± 1% perf-stat.dTLB-stores 28076071 ± 1% -11.4% 24882928 ± 1% perf-stat.minor-faults 33.81 ± 81% +75.2% 59.24 ± 2% perf-stat.node-load-miss-rate 7.841e+09 ± 2% -8.0% 7.216e+09 ± 0% perf-stat.node-load-misses 1.398e+10 ± 3% -12.8% 1.219e+10 ± 2% perf-stat.node-loads 1.862e+09 ± 1% -5.9% 1.753e+09 ± 2% perf-stat.node-store-misses 28076036 ± 1% -11.4% 24883005 ± 1% perf-stat.page-faults 4.79 ± 11% +42.4% 6.82 ± 11% sched_debug.cfs_rq:/.load_avg.min 3425468 ± 29% +122.4% 7618598 ± 33% sched_debug.cfs_rq:/.min_vruntime.stddev 2.39 ± 34% +42.0% 3.39 ± 19% sched_debug.cfs_rq:/.runnable_load_avg.min 7846188 ± 15% +53.3% 12024520 ± 25% sched_debug.cfs_rq:/.spread0.max 3427472 ± 29% +122.4% 7622936 ± 33% sched_debug.cfs_rq:/.spread0.stddev 26.63 ± 6% +11.7% 29.74 ± 4% sched_debug.cpu.cpu_load[1].avg 3.38 ± 24% +37.8% 4.66 ± 13% sched_debug.cpu.cpu_load[1].min 26.36 ± 7% +10.9% 29.22 ± 3% sched_debug.cpu.cpu_load[2].avg 3.62 ± 18% +32.0% 4.77 ± 14% sched_debug.cpu.cpu_load[2].min 26.20 ± 7% +9.9% 28.79 ± 4% sched_debug.cpu.cpu_load[3].avg 3.71 ± 17% +35.2% 5.02 ± 10% sched_debug.cpu.cpu_load[3].min 26.04 ± 7% +9.0% 28.39 ± 4% sched_debug.cpu.cpu_load[4].avg 3.87 ± 15% +29.1% 5.00 ± 9% sched_debug.cpu.cpu_load[4].min 618.52 ± 35% +64.0% 1014 ± 28% sched_debug.cpu.load.min 1690 ± 9% +11.2% 1879 ± 4% sched_debug.cpu.nr_load_updates.stddev 42.20 ± 9% -20.8% 33.44 ± 15% sched_debug.cpu.nr_running.avg 1893966 ± 29% +88.1% 3562142 ± 30% sched_debug.cpu.nr_switches.stddev 137.38 ± 34% +76.2% 242.00 ± 26% sched_debug.cpu.nr_uninterruptible.max -226.63 ±-62% +159.2% -587.52 ±-23% sched_debug.cpu.nr_uninterruptible.min 82.30 ± 39% +112.4% 174.81 ± 18% sched_debug.cpu.nr_uninterruptible.stddev 1.96 ±200% +405.1% 9.90 ± 0% sched_debug.rt_rq:/.rt_runtime.stddev 0.12 ±200% +845.6% 1.17 ± 7% perf-profile.cycles.__fget_light.sys_read.entry_SYSCALL_64_fastpath 0.23 ±123% +457.7% 1.30 ± 7% perf-profile.cycles.__fget_light.sys_write.entry_SYSCALL_64_fastpath 1.09 ± 16% +22.3% 1.34 ± 7% perf-profile.cycles.__inode_security_revalidate.selinux_file_permission.security_file_permission.rw_verify_area.vfs_write 29.47 ± 34% -100.0% 0.00 ± -1% perf-profile.cycles.__read_nocancel 0.76 ± 21% +94.0% 1.48 ± 11% perf-profile.cycles.__schedule.schedule.exit_to_usermode_loop.syscall_return_slowpath.entry_SYSCALL_64_fastpath 2.29 ± 11% +24.8% 2.86 ± 4% perf-profile.cycles.__schedule.schedule.pipe_wait.pipe_write.__vfs_write 0.46 ± 89% +232.4% 1.52 ± 5% perf-profile.cycles.__switch_to 7.99 ± 76% +228.6% 26.27 ± 0% perf-profile.cycles.__vfs_read.vfs_read.sys_read.entry_SYSCALL_64_fastpath 16.43 ± 34% -100.0% 0.00 ± -1% perf-profile.cycles.__vfs_read.vfs_read.sys_read.entry_SYSCALL_64_fastpath.__read_nocancel 10.74 ± 74% +219.8% 34.34 ± 2% perf-profile.cycles.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath 20.33 ± 36% -100.0% 0.00 ± -1% perf-profile.cycles.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath.__write_nocancel 7.59 ± 14% +28.7% 9.77 ± 14% perf-profile.cycles.__wake_up_common.__wake_up_sync_key.pipe_write.__vfs_write.vfs_write 10.41 ± 10% +22.3% 12.73 ± 11% perf-profile.cycles.__wake_up_sync_key.pipe_write.__vfs_write.vfs_write.sys_write 32.88 ± 35% -100.0% 0.00 ± -1% perf-profile.cycles.__write_nocancel 3.43 ± 10% +30.3% 4.46 ± 7% perf-profile.cycles.activate_task.ttwu_do_activate.try_to_wake_up.default_wake_function.autoremove_wake_function 7.18 ± 15% +29.9% 9.33 ± 14% perf-profile.cycles.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.pipe_write.__vfs_write 4.77 ± 18% -100.0% 0.00 ± -1% perf-profile.cycles.call_cpuidle.cpu_startup_entry.start_secondary 5.06 ± 16% -93.9% 0.31 ±100% perf-profile.cycles.cpu_startup_entry.start_secondary 4.76 ± 18% -100.0% 0.00 ± -1% perf-profile.cycles.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary 4.72 ± 18% -100.0% 0.00 ± -1% perf-profile.cycles.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary 0.72 ± 20% +63.1% 1.17 ± 3% perf-profile.cycles.deactivate_task.__schedule.schedule.pipe_wait.pipe_write 1.63 ± 10% +19.2% 1.94 ± 3% perf-profile.cycles.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.pipe_read 7.09 ± 15% +29.9% 9.21 ± 14% perf-profile.cycles.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.pipe_write 1.61 ± 17% +51.5% 2.45 ± 8% perf-profile.cycles.dequeue_entity.dequeue_task_fair.deactivate_task.__schedule.schedule 2.27 ± 16% +38.3% 3.15 ± 8% perf-profile.cycles.dequeue_task_fair.deactivate_task.__schedule.schedule.pipe_wait 2.16 ± 14% +49.9% 3.24 ± 7% perf-profile.cycles.enqueue_entity.enqueue_task_fair.activate_task.ttwu_do_activate.try_to_wake_up 2.76 ± 9% +41.8% 3.92 ± 7% perf-profile.cycles.enqueue_task_fair.activate_task.ttwu_do_activate.try_to_wake_up.default_wake_function 0.57 ± 87% +255.6% 2.02 ± 7% perf-profile.cycles.entry_SYSCALL_64 0.56 ± 85% +272.8% 2.10 ± 7% perf-profile.cycles.entry_SYSCALL_64_after_swapgs 28.04 ± 74% +227.8% 91.93 ± 0% perf-profile.cycles.entry_SYSCALL_64_fastpath 26.07 ± 34% -100.0% 0.00 ± -1% perf-profile.cycles.entry_SYSCALL_64_fastpath.__read_nocancel 29.86 ± 35% -100.0% 0.00 ± -1% perf-profile.cycles.entry_SYSCALL_64_fastpath.__write_nocancel 0.51 ± 88% +221.4% 1.63 ± 11% perf-profile.cycles.exit_to_usermode_loop.syscall_return_slowpath.entry_SYSCALL_64_fastpath 0.60 ± 57% +103.9% 1.22 ± 23% perf-profile.cycles.idle_cpu.select_idle_sibling.select_task_rq_fair.try_to_wake_up.default_wake_function 1.75 ± 6% -26.9% 1.28 ± 3% perf-profile.cycles.mutex_unlock.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath 0.63 ± 54% +97.8% 1.25 ± 9% perf-profile.cycles.pick_next_task_fair.__schedule.schedule.pipe_wait.pipe_read 2.98 ± 13% +27.0% 3.78 ± 7% perf-profile.cycles.pipe_wait.pipe_write.__vfs_write.vfs_write.sys_write 27.29 ± 3% +13.8% 31.05 ± 3% perf-profile.cycles.pipe_write.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath 4.60 ± 18% -100.0% 0.00 ± -1% perf-profile.cycles.poll_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry 2.34 ± 76% +252.6% 8.25 ± 7% perf-profile.cycles.rw_verify_area.vfs_read.sys_read.entry_SYSCALL_64_fastpath 5.54 ± 34% -100.0% 0.00 ± -1% perf-profile.cycles.rw_verify_area.vfs_read.sys_read.entry_SYSCALL_64_fastpath.__read_nocancel 1.93 ± 75% +261.3% 6.98 ± 7% perf-profile.cycles.rw_verify_area.vfs_write.sys_write.entry_SYSCALL_64_fastpath 4.56 ± 34% -100.0% 0.00 ± -1% perf-profile.cycles.rw_verify_area.vfs_write.sys_write.entry_SYSCALL_64_fastpath.__write_nocancel 0.49 ± 88% +223.6% 1.57 ± 11% perf-profile.cycles.schedule.exit_to_usermode_loop.syscall_return_slowpath.entry_SYSCALL_64_fastpath 2.40 ± 11% +24.2% 2.98 ± 4% perf-profile.cycles.schedule.pipe_wait.pipe_write.__vfs_write.vfs_write 1.56 ± 14% +80.1% 2.82 ± 13% perf-profile.cycles.select_idle_sibling.select_task_rq_fair.try_to_wake_up.default_wake_function.autoremove_wake_function 2.70 ± 16% +50.0% 4.04 ± 14% perf-profile.cycles.select_task_rq_fair.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common 5.14 ± 7% +10.2% 5.67 ± 7% perf-profile.cycles.selinux_file_permission.security_file_permission.rw_verify_area.vfs_read.sys_read 5.12 ± 8% +13.1% 5.80 ± 8% perf-profile.cycles.selinux_file_permission.security_file_permission.rw_verify_area.vfs_write.sys_write 5.07 ± 16% -93.8% 0.31 ±100% perf-profile.cycles.start_secondary 11.81 ± 76% +234.1% 39.47 ± 2% perf-profile.cycles.sys_read.entry_SYSCALL_64_fastpath 25.48 ± 34% -100.0% 0.00 ± -1% perf-profile.cycles.sys_read.entry_SYSCALL_64_fastpath.__read_nocancel 14.31 ± 74% +230.0% 47.23 ± 0% perf-profile.cycles.sys_write.entry_SYSCALL_64_fastpath 28.76 ± 35% -100.0% 0.00 ± -1% perf-profile.cycles.sys_write.entry_SYSCALL_64_fastpath.__write_nocancel 0.53 ± 88% +224.8% 1.72 ± 11% perf-profile.cycles.syscall_return_slowpath.entry_SYSCALL_64_fastpath 8.56 ± 14% +27.6% 10.92 ± 12% perf-profile.cycles.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key 4.07 ± 9% +28.1% 5.21 ± 8% perf-profile.cycles.ttwu_do_activate.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common 11.13 ± 76% +233.3% 37.10 ± 1% perf-profile.cycles.vfs_read.sys_read.entry_SYSCALL_64_fastpath 23.80 ± 34% -100.0% 0.00 ± -1% perf-profile.cycles.vfs_read.sys_read.entry_SYSCALL_64_fastpath.__read_nocancel 13.60 ± 74% +227.9% 44.58 ± 0% perf-profile.cycles.vfs_write.sys_write.entry_SYSCALL_64_fastpath 27.07 ± 35% -100.0% 0.00 ± -1% perf-profile.cycles.vfs_write.sys_write.entry_SYSCALL_64_fastpath.__write_nocancel 2.10 ± 5% +18.3% 2.48 ± 6% perf-profile.func.cycles.__fget_light 1.07 ± 5% +23.6% 1.33 ± 7% perf-profile.func.cycles.__inode_security_revalidate 1.09 ± 35% -100.0% 0.00 ± -1% perf-profile.func.cycles.__read_nocancel 1.48 ± 8% +17.9% 1.75 ± 4% perf-profile.func.cycles.__switch_to 0.86 ± 37% -100.0% 0.00 ± -1% perf-profile.func.cycles.__write_nocancel 2.83 ± 3% -18.5% 2.30 ± 3% perf-profile.func.cycles.mutex_unlock 2.07 ± 2% +15.8% 2.39 ± 4% perf-profile.func.cycles.pipe_read 2.46 ± 5% +17.0% 2.88 ± 5% perf-profile.func.cycles.pipe_write 4.75 ± 19% -100.0% 0.00 ± -1% perf-profile.func.cycles.poll_idle 3.47 ± 8% +26.9% 4.40 ± 6% perf-profile.func.cycles.selinux_file_permission 1.26 ± 8% +12.7% 1.42 ± 4% perf-profile.func.cycles.switch_mm_irqs_off 0.94 ± 12% +24.2% 1.17 ± 12% perf-profile.func.cycles.update_curr 1.62 ± 7% +15.3% 1.87 ± 8% perf-profile.func.cycles.vfs_write hackbench.throughput 250000 ++-----------------------------------------------------------------+ *..*..*.*..*..*..*..*.*..*..*..*..*.*..*..*..*.*..*..*..* | O O O O O O O O O O O O O O O O O O O O O O O O 200000 ++ | | | | | 150000 ++ | | | 100000 ++ | | | | | 50000 ++ | | | | | 0 ++--------------------------O--------------------------------------+ hackbench.time.percent_of_cpu_this_job_got 3000 *+-*--*--*--*-*--*--*--*--*--*--*--*-*--*--*--*--*--*--*--*----------+ O O O O O O O O O O O O O O O O O O O O O O O O 2500 ++ | | | | | 2000 ++ | | | 1500 ++ | | | 1000 ++ | | | | | 500 ++ | | | 0 ++---------------------------O---------------------------------------+ hackbench.time.minor_page_faults 3e+07 ++----------------------------------------------------------------+ | .*. .*.. .*. .*.. .*.. .*. .* | 2.5e+07 *+.*. *. *. *..*..*..*.*..*. * *..*. *. | O O O O O O O O O O O O O O O O O O O O O O O O | | 2e+07 ++ | | | 1.5e+07 ++ | | | 1e+07 ++ | | | | | 5e+06 ++ | | | 0 ++--------------------------O-------------------------------------+ [*] bisect-good sample [O] bisect-bad sample Thanks, Xiaolong View attachment "config-4.7.0-00001-geadd462" of type "text/plain" (150956 bytes) View attachment "job.yaml" of type "text/plain" (3649 bytes) View attachment "reproduce" of type "text/plain" (2375 bytes)
Powered by blists - more mailing lists