Greeting, FYI, we noticed a -6.8% regression of aim7.jobs-per-min due to commit: commit: efc61345274d6c7a46a0570efbc916fcbe3e927b ("ext4: shrink race window in ext4_should_retry_alloc()") https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master in testcase: aim7 on test machine: 144 threads Intel(R) Xeon(R) Gold 5318H CPU @ 2.50GHz with 128G memory with following parameters: disk: 1BRD_48G fs: ext4 test: creat-clo load: 1000 cpufreq_governor: performance ucode: 0x700001e test-description: AIM7 is a traditional UNIX system level benchmark suite which is used to test and measure the performance of multiuser system. test-url: https://sourceforge.net/projects/aimbench/files/aim-suite7/ If you fix the issue, kindly add following tag Reported-by: kernel test robot 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 split-job --compatible job.yaml bin/lkp run compatible-job.yaml ========================================================================================= compiler/cpufreq_governor/disk/fs/kconfig/load/rootfs/tbox_group/test/testcase/ucode: gcc-9/performance/1BRD_48G/ext4/x86_64-rhel-8.3/1000/debian-10.4-x86_64-20200603.cgz/lkp-cpl-4sp1/creat-clo/aim7/0x700001e commit: v5.12-rc2 efc6134527 ("ext4: shrink race window in ext4_should_retry_alloc()") v5.12-rc2 efc61345274d6c7a46a0570efbc ---------------- --------------------------- %stddev %change %stddev \ | \ 9766 -6.8% 9103 aim7.jobs-per-min 614.48 +7.3% 659.28 aim7.time.elapsed_time 614.48 +7.3% 659.28 aim7.time.elapsed_time.max 1413448 +6.7% 1507983 aim7.time.involuntary_context_switches 80980 +7.9% 87401 aim7.time.system_time 7.78 -6.8% 7.25 iostat.cpu.idle 1336 -6.4% 1251 vmstat.io.bo 10609 -4.7% 10108 vmstat.system.cs 1.11 ± 95% -90.8% 0.10 ±206% perf-sched.sch_delay.avg.ms.preempt_schedule_common.__cond_resched.dput.terminate_walk.path_openat 0.03 ± 37% +365.6% 0.14 ± 71% perf-sched.sch_delay.avg.ms.schedule_hrtimeout_range_clock.ep_poll.do_epoll_wait.__x64_sys_epoll_wait 0.01 ± 72% +16200.0% 1.25 ±211% perf-sched.sch_delay.max.ms.preempt_schedule_common.__cond_resched.ext4_journal_check_start.__ext4_journal_start_sb.ext4_ext_remove_space 0.56 ± 19% -39.2% 0.34 ± 17% sched_debug.cfs_rq:/.load_avg.min 36889619 +8.1% 39874296 ± 4% sched_debug.cfs_rq:/.min_vruntime.avg 37479588 +8.1% 40514969 ± 4% sched_debug.cfs_rq:/.min_vruntime.max 34455004 +8.1% 37252165 ± 4% sched_debug.cfs_rq:/.min_vruntime.min 165.11 ± 9% -22.2% 128.44 ± 6% sched_debug.cfs_rq:/.runnable_avg.min 2979282 ± 4% +7.7% 3207529 ± 5% sched_debug.cfs_rq:/.spread0.max 153.41 ± 10% -28.7% 109.41 ± 8% sched_debug.cfs_rq:/.util_avg.min 241.33 +3.9% 250.67 proc-vmstat.nr_active_file 241.33 +3.9% 250.67 proc-vmstat.nr_zone_active_file 2448715 +4.5% 2558105 proc-vmstat.numa_hit 2253807 +4.9% 2363202 proc-vmstat.numa_local 2663735 +4.0% 2768966 proc-vmstat.pgalloc_normal 2930123 +5.8% 3099293 proc-vmstat.pgfault 2238600 +4.7% 2344785 proc-vmstat.pgfree 129935 +8.5% 141043 proc-vmstat.pgreuse 499474 +5.3% 525955 interrupts.CAL:Function_call_interrupts 1919 ± 3% +11.1% 2132 ± 3% interrupts.CPU0.RES:Rescheduling_interrupts 1968 ± 2% +11.4% 2192 ± 2% interrupts.CPU11.RES:Rescheduling_interrupts 1791 ± 4% +6.0% 1899 ± 2% interrupts.CPU122.RES:Rescheduling_interrupts 3244 ± 8% +11.3% 3611 ± 3% interrupts.CPU130.CAL:Function_call_interrupts 3224 ± 7% +11.7% 3603 ± 3% interrupts.CPU136.CAL:Function_call_interrupts 1974 ± 2% +10.0% 2172 ± 3% interrupts.CPU14.RES:Rescheduling_interrupts 3218 ± 7% +12.0% 3603 ± 4% interrupts.CPU141.CAL:Function_call_interrupts 1993 ± 3% +10.4% 2201 ± 4% interrupts.CPU17.RES:Rescheduling_interrupts 1924 ± 2% +12.6% 2167 ± 3% interrupts.CPU2.RES:Rescheduling_interrupts 1944 ± 3% +11.8% 2174 ± 3% interrupts.CPU3.RES:Rescheduling_interrupts 1982 ± 3% +10.0% 2181 ± 3% interrupts.CPU4.RES:Rescheduling_interrupts 1991 ± 3% +8.7% 2164 ± 3% interrupts.CPU6.RES:Rescheduling_interrupts 3485 ± 5% +11.4% 3881 ± 4% interrupts.CPU60.CAL:Function_call_interrupts 1986 ± 3% +10.4% 2194 ± 2% interrupts.CPU7.RES:Rescheduling_interrupts 1789 ± 4% +13.6% 2033 ± 3% interrupts.CPU73.RES:Rescheduling_interrupts 1868 ± 3% +10.3% 2060 ± 2% interrupts.CPU76.RES:Rescheduling_interrupts 1904 ± 2% +9.4% 2082 ± 3% interrupts.CPU77.RES:Rescheduling_interrupts 1912 ± 2% +11.5% 2131 ± 2% interrupts.CPU78.RES:Rescheduling_interrupts 1880 ± 5% +12.4% 2113 ± 2% interrupts.CPU79.RES:Rescheduling_interrupts 1906 ± 3% +12.0% 2134 ± 2% interrupts.CPU80.RES:Rescheduling_interrupts 1910 ± 3% +12.4% 2146 ± 2% interrupts.CPU82.RES:Rescheduling_interrupts 1911 ± 2% +10.9% 2120 ± 4% interrupts.CPU83.RES:Rescheduling_interrupts 1925 +12.1% 2158 ± 3% interrupts.CPU85.RES:Rescheduling_interrupts 1935 ± 3% +9.7% 2123 ± 2% interrupts.CPU86.RES:Rescheduling_interrupts 1940 ± 4% +11.3% 2158 ± 2% interrupts.CPU87.RES:Rescheduling_interrupts 1991 ± 3% +11.2% 2213 ± 3% interrupts.CPU9.RES:Rescheduling_interrupts 1694 ± 2% +9.7% 1858 ± 3% interrupts.CPU93.RES:Rescheduling_interrupts 11336532 -3.0% 10999810 perf-stat.i.branch-misses 62.77 +1.2 64.01 perf-stat.i.cache-miss-rate% 47260240 +9.2% 51618807 ± 6% perf-stat.i.cache-misses 10620 -4.7% 10117 perf-stat.i.context-switches 11.12 +1.0% 11.23 perf-stat.i.cpi 2471 -6.1% 2319 perf-stat.i.cpu-migrations 9149 -7.5% 8464 ± 6% perf-stat.i.cycles-between-cache-misses 8.546e+08 -5.9% 8.045e+08 perf-stat.i.dTLB-stores 5416307 -3.9% 5202887 perf-stat.i.iTLB-load-misses 606729 -7.0% 564408 ± 4% perf-stat.i.iTLB-loads 7161 +3.8% 7436 perf-stat.i.instructions-per-iTLB-miss 0.09 -1.5% 0.09 perf-stat.i.ipc 0.79 ± 11% -21.5% 0.62 ± 12% perf-stat.i.major-faults 5810373 +4.1% 6051325 perf-stat.i.node-store-misses 41799 -4.6% 39889 ± 2% perf-stat.i.node-stores 0.14 -0.0 0.14 perf-stat.overall.branch-miss-rate% 62.99 +1.1 64.12 perf-stat.overall.cache-miss-rate% 9259 -7.6% 8560 ± 6% perf-stat.overall.cycles-between-cache-misses 7192 +3.8% 7468 perf-stat.overall.instructions-per-iTLB-miss 11296461 -3.0% 10960807 perf-stat.ps.branch-misses 47194756 +9.2% 51557679 ± 6% perf-stat.ps.cache-misses 10603 -4.7% 10102 perf-stat.ps.context-switches 2468 -6.1% 2317 perf-stat.ps.cpu-migrations 8.534e+08 -5.8% 8.036e+08 perf-stat.ps.dTLB-stores 5408407 -3.9% 5196316 perf-stat.ps.iTLB-load-misses 602701 -6.9% 561024 ± 4% perf-stat.ps.iTLB-loads 0.77 ± 10% -19.7% 0.62 ± 10% perf-stat.ps.major-faults 5803928 +4.2% 6045605 perf-stat.ps.node-store-misses 41864 -4.5% 39964 ± 2% perf-stat.ps.node-stores 2.393e+13 +7.0% 2.562e+13 perf-stat.total.instructions 1.87 ± 21% -0.6 1.26 ± 25% perf-profile.calltrace.cycles-pp.secondary_startup_64_no_verify 1.86 ± 21% -0.6 1.25 ± 26% perf-profile.calltrace.cycles-pp.cpu_startup_entry.start_secondary.secondary_startup_64_no_verify 1.86 ± 21% -0.6 1.25 ± 26% perf-profile.calltrace.cycles-pp.do_idle.cpu_startup_entry.start_secondary.secondary_startup_64_no_verify 1.86 ± 21% -0.6 1.24 ± 26% perf-profile.calltrace.cycles-pp.cpuidle_enter.do_idle.cpu_startup_entry.start_secondary.secondary_startup_64_no_verify 1.86 ± 21% -0.6 1.24 ± 26% perf-profile.calltrace.cycles-pp.cpuidle_enter_state.cpuidle_enter.do_idle.cpu_startup_entry.start_secondary 1.86 ± 21% -0.6 1.25 ± 26% perf-profile.calltrace.cycles-pp.start_secondary.secondary_startup_64_no_verify 1.84 ± 22% -0.6 1.22 ± 26% perf-profile.calltrace.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.do_idle.cpu_startup_entry 46.27 +0.4 46.69 perf-profile.calltrace.cycles-pp.osq_lock.__mutex_lock.ext4_orphan_add.ext4_truncate.ext4_setattr 46.66 +0.4 47.09 perf-profile.calltrace.cycles-pp.__mutex_lock.ext4_orphan_add.ext4_truncate.ext4_setattr.notify_change 47.09 +0.4 47.53 perf-profile.calltrace.cycles-pp.ext4_orphan_add.ext4_truncate.ext4_setattr.notify_change.do_truncate 95.37 +0.6 95.97 perf-profile.calltrace.cycles-pp.do_open.path_openat.do_filp_open.do_sys_openat2.do_sys_open 95.27 +0.6 95.88 perf-profile.calltrace.cycles-pp.do_truncate.do_open.path_openat.do_filp_open.do_sys_openat2 94.89 +0.6 95.50 perf-profile.calltrace.cycles-pp.ext4_truncate.ext4_setattr.notify_change.do_truncate.do_open 95.21 +0.6 95.83 perf-profile.calltrace.cycles-pp.notify_change.do_truncate.do_open.path_openat.do_filp_open 95.20 +0.6 95.82 perf-profile.calltrace.cycles-pp.ext4_setattr.notify_change.do_truncate.do_open.path_openat 97.87 +0.6 98.50 perf-profile.calltrace.cycles-pp.creat64 97.76 +0.6 98.38 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.creat64 97.75 +0.6 98.38 perf-profile.calltrace.cycles-pp.do_sys_open.do_syscall_64.entry_SYSCALL_64_after_hwframe.creat64 97.77 +0.6 98.40 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.creat64 97.72 +0.6 98.35 perf-profile.calltrace.cycles-pp.path_openat.do_filp_open.do_sys_openat2.do_sys_open.do_syscall_64 97.72 +0.6 98.35 perf-profile.calltrace.cycles-pp.do_filp_open.do_sys_openat2.do_sys_open.do_syscall_64.entry_SYSCALL_64_after_hwframe 97.75 +0.6 98.38 perf-profile.calltrace.cycles-pp.do_sys_openat2.do_sys_open.do_syscall_64.entry_SYSCALL_64_after_hwframe.creat64 1.85 ± 22% -0.6 1.23 ± 26% perf-profile.children.cycles-pp.intel_idle 1.87 ± 21% -0.6 1.26 ± 25% perf-profile.children.cycles-pp.secondary_startup_64_no_verify 1.87 ± 21% -0.6 1.26 ± 25% perf-profile.children.cycles-pp.cpu_startup_entry 1.87 ± 21% -0.6 1.26 ± 25% perf-profile.children.cycles-pp.do_idle 1.86 ± 21% -0.6 1.25 ± 26% perf-profile.children.cycles-pp.start_secondary 1.87 ± 21% -0.6 1.26 ± 25% perf-profile.children.cycles-pp.cpuidle_enter 1.87 ± 21% -0.6 1.26 ± 25% perf-profile.children.cycles-pp.cpuidle_enter_state 0.18 ± 2% -0.0 0.15 ± 3% perf-profile.children.cycles-pp.__mutex_unlock_slowpath 0.51 ± 2% +0.0 0.55 ± 3% perf-profile.children.cycles-pp.ext4_mark_iloc_dirty 0.00 +0.1 0.08 ± 8% perf-profile.children.cycles-pp.ext4_inode_journal_mode 0.00 +0.1 0.09 ± 4% perf-profile.children.cycles-pp.ext4_fc_track_inode 47.13 +0.4 47.57 perf-profile.children.cycles-pp.ext4_orphan_add 94.77 +0.6 95.36 perf-profile.children.cycles-pp.osq_lock 95.37 +0.6 95.97 perf-profile.children.cycles-pp.do_open 95.27 +0.6 95.88 perf-profile.children.cycles-pp.do_truncate 94.89 +0.6 95.50 perf-profile.children.cycles-pp.ext4_truncate 95.21 +0.6 95.83 perf-profile.children.cycles-pp.notify_change 95.20 +0.6 95.82 perf-profile.children.cycles-pp.ext4_setattr 97.88 +0.6 98.50 perf-profile.children.cycles-pp.creat64 97.75 +0.6 98.38 perf-profile.children.cycles-pp.do_sys_open 97.75 +0.6 98.38 perf-profile.children.cycles-pp.do_sys_openat2 97.72 +0.6 98.35 perf-profile.children.cycles-pp.do_filp_open 97.72 +0.6 98.35 perf-profile.children.cycles-pp.path_openat 97.95 +0.6 98.59 perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe 97.87 +0.6 98.51 perf-profile.children.cycles-pp.do_syscall_64 1.85 ± 22% -0.6 1.23 ± 26% perf-profile.self.cycles-pp.intel_idle 0.13 ± 2% -0.0 0.10 ± 6% perf-profile.self.cycles-pp.ext4_reserve_inode_write 0.15 ± 3% -0.0 0.13 ± 2% perf-profile.self.cycles-pp.ext4_mark_iloc_dirty 0.08 ± 4% -0.0 0.06 perf-profile.self.cycles-pp.__mutex_unlock_slowpath 0.07 -0.0 0.06 perf-profile.self.cycles-pp.mutex_lock 0.18 ± 2% +0.0 0.19 ± 2% perf-profile.self.cycles-pp.__mutex_lock 0.05 ± 8% +0.0 0.08 ± 12% perf-profile.self.cycles-pp.ext4_orphan_add 0.00 +0.1 0.08 ± 8% perf-profile.self.cycles-pp.ext4_inode_journal_mode 94.04 +0.6 94.66 perf-profile.self.cycles-pp.osq_lock aim7.jobs-per-min 10000 +-------------------------------------------------------------------+ 9900 |-+ +. + .+ .+ | | +. : +. :+ + + : + +. : | 9800 |++ +.. : +.. : + + + + + : + + + : | 9700 |-+ +. : +. .+ : + : + + + : .+ + :.| 9600 |-+ + O + : : + : +. .. + + | 9500 |-+ O : : + +.+ | | O : O | 9400 |-+ O O + O O O O | 9300 |-+ | 9200 |-+ O O O | 9100 |-O O O O | | O O O O | 9000 |-+ O O O O | 8900 +-------------------------------------------------------------------+ [*] 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. --- 0DAY/LKP+ Test Infrastructure Open Source Technology Center https://lists.01.org/hyperkitty/list/lkp@lists.01.org Intel Corporation Thanks, Oliver Sang