Greeting, FYI, we noticed a -24.0% regression of stress-ng.lockf.ops_per_sec due to commit: commit: b47291ef02b0bee85ffb7efd6c336060ad1fe1a4 ("mm, slub: change percpu partial accounting from objects to pages") https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master in testcase: stress-ng on test machine: 96 threads 2 sockets Intel(R) Xeon(R) Gold 6252 CPU @ 2.10GHz with 512G memory with following parameters: nr_threads: 10% disk: 1HDD testtime: 60s fs: xfs class: filesystem test: lockf cpufreq_governor: performance ucode: 0x5003006 If you fix the issue, kindly add following tag Reported-by: kernel test robot <oliver.sang(a)intel.com&gt; Details are as below: --------------------------------------------------------------------------------------------------> To reproduce: git clone https://github.com/intel/lkp-tests.git cd lkp-tests sudo bin/lkp install job.yaml # job file is attached in this email bin/lkp split-job --compatible job.yaml # generate the yaml file for lkp run sudo bin/lkp run generated-yaml-file # if come across any failure that blocks the test, # please remove ~/.lkp and /lkp dir to run from a clean state. ========================================================================================= class/compiler/cpufreq_governor/disk/fs/kconfig/nr_threads/rootfs/tbox_group/test/testcase/testtime/ucode: filesystem/gcc-9/performance/1HDD/xfs/x86_64-rhel-8.3/10%/debian-10.4-x86_64-20200603.cgz/lkp-csl-2sp7/lockf/stress-ng/60s/0x5003006 commit: d0fe47c641 ("slub: add back check for free nonslab objects") b47291ef02 ("mm, slub: change percpu partial accounting from objects to pages") d0fe47c64152a63c b47291ef02b0bee85ffb7efd6c3 ---------------- --------------------------- %stddev %change %stddev \ | \ 21444470 -24.0% 16302033 ± 3% stress-ng.lockf.ops 357404 -24.0% 271697 ± 3% stress-ng.lockf.ops_per_sec 39082 ± 2% -25.3% 29179 ± 3% stress-ng.time.voluntary_context_switches 0.16 ± 12% -0.0 0.13 ± 9% mpstat.cpu.all.usr% 111717 ± 9% -16.2% 93652 ± 9% numa-meminfo.node0.SUnreclaim 27929 ± 9% -16.2% 23412 ± 9% numa-vmstat.node0.nr_slab_unreclaimable 8521 ± 6% +15.4% 9833 ± 15% softirqs.CPU45.SCHED 146.10 +1.6% 148.47 turbostat.RAMWatt 3238 -8.7% 2955 ± 2% vmstat.system.cs 206396 -12.3% 180914 meminfo.SUnreclaim 320235 -10.6% 286436 meminfo.Slab 28459 -7.3% 26380 proc-vmstat.nr_slab_reclaimable 51598 -12.3% 45228 proc-vmstat.nr_slab_unreclaimable 0.87 ± 11% -0.2 0.65 ± 8% perf-profile.children.cycles-pp.llseek 0.40 ± 13% -0.1 0.29 ± 6% perf-profile.children.cycles-pp.ksys_lseek 0.27 ± 10% -0.1 0.20 ± 12% perf-profile.children.cycles-pp.__entry_text_start 0.25 ± 11% -0.1 0.19 ± 13% perf-profile.children.cycles-pp.memset_erms 0.07 ± 17% -0.0 0.04 ± 45% perf-profile.children.cycles-pp.xfs_file_llseek 0.08 ± 6% -0.0 0.06 ± 9% perf-profile.children.cycles-pp.stress_lockf_contention 0.05 ± 47% +0.1 0.13 ± 11% perf-profile.children.cycles-pp.locks_insert_lock_ctx 0.04 ± 72% +0.1 0.12 ± 20% perf-profile.children.cycles-pp.locks_release_private 0.11 ± 11% +0.1 0.20 ± 12% perf-profile.children.cycles-pp.locks_dispose_list 0.05 ± 45% +0.1 0.18 ± 23% perf-profile.children.cycles-pp._raw_spin_lock_irqsave 0.00 +0.2 0.21 ± 31% perf-profile.children.cycles-pp.__unfreeze_partials 0.00 +0.2 0.25 ± 29% perf-profile.children.cycles-pp.get_partial_node 0.24 ± 10% -0.1 0.18 ± 12% perf-profile.self.cycles-pp.memset_erms 0.22 ± 8% -0.0 0.17 ± 12% perf-profile.self.cycles-pp.kmem_cache_alloc 0.14 ± 10% -0.0 0.11 ± 11% perf-profile.self.cycles-pp.__entry_text_start 0.07 ± 17% -0.0 0.04 ± 45% perf-profile.self.cycles-pp.xfs_file_llseek 0.04 ± 73% +0.1 0.12 ± 20% perf-profile.self.cycles-pp.locks_release_private 0.00 +0.1 0.09 ± 29% perf-profile.self.cycles-pp.__unfreeze_partials 0.04 ± 45% +0.1 0.15 ± 22% perf-profile.self.cycles-pp._raw_spin_lock_irqsave 0.00 +0.2 0.16 ± 29% perf-profile.self.cycles-pp.get_partial_node 0.26 ± 12% +0.2 0.48 ± 5% perf-profile.self.cycles-pp._raw_spin_lock 0.67 ± 8% +1032.2% 7.57 ± 3% perf-stat.i.MPKI 6.339e+09 -23.8% 4.829e+09 ± 3% perf-stat.i.branch-instructions 11570593 ± 7% -16.3% 9688662 ± 6% perf-stat.i.branch-misses 3504821 ± 39% +730.5% 29107036 ± 26% perf-stat.i.cache-misses 13767953 ± 9% +853.6% 1.313e+08 ± 6% perf-stat.i.cache-references 2889 -8.7% 2636 perf-stat.i.context-switches 1.20 +30.6% 1.57 ± 3% perf-stat.i.cpi 10237 ± 41% -87.6% 1269 ± 31% perf-stat.i.cycles-between-cache-misses 0.00 ± 18% +0.0 0.00 ± 47% perf-stat.i.dTLB-load-miss-rate% 79657 ± 23% +111.8% 168752 ± 33% perf-stat.i.dTLB-load-misses 7.894e+09 -22.8% 6.096e+09 ± 3% perf-stat.i.dTLB-loads 1.195e+09 -21.5% 9.385e+08 ± 3% perf-stat.i.dTLB-stores 2756251 ± 3% -14.2% 2364908 ± 4% perf-stat.i.iTLB-load-misses 2.264e+10 -23.7% 1.727e+10 ± 3% perf-stat.i.instructions 8215 ± 3% -11.1% 7303 ± 2% perf-stat.i.instructions-per-iTLB-miss 0.83 -22.9% 0.64 ± 3% perf-stat.i.ipc 11.34 ± 2% -3.8% 10.92 ± 4% perf-stat.i.major-faults 179.74 ± 7% -32.9% 120.57 ± 20% perf-stat.i.metric.K/sec 160.68 -22.3% 124.90 ± 3% perf-stat.i.metric.M/sec 1133752 ± 45% +392.3% 5581428 ± 24% perf-stat.i.node-load-misses 394916 ± 51% +290.3% 1541328 ± 27% perf-stat.i.node-loads 122286 ± 17% +438.7% 658797 ± 37% perf-stat.i.node-store-misses 0.61 ± 9% +1149.6% 7.59 ± 3% perf-stat.overall.MPKI 1.20 +30.5% 1.57 ± 3% perf-stat.overall.cpi 9089 ± 37% -88.6% 1038 ± 41% perf-stat.overall.cycles-between-cache-misses 0.00 ± 23% +0.0 0.00 ± 34% perf-stat.overall.dTLB-load-miss-rate% 0.00 ± 4% +0.0 0.00 ± 26% perf-stat.overall.dTLB-store-miss-rate% 8224 ± 3% -11.1% 7308 ± 2% perf-stat.overall.instructions-per-iTLB-miss 0.83 -23.3% 0.64 ± 3% perf-stat.overall.ipc 6.238e+09 -23.8% 4.752e+09 ± 3% perf-stat.ps.branch-instructions 11350804 ± 7% -16.1% 9518421 ± 6% perf-stat.ps.branch-misses 3449652 ± 39% +730.2% 28640228 ± 26% perf-stat.ps.cache-misses 13541733 ± 9% +854.1% 1.292e+08 ± 6% perf-stat.ps.cache-references 2842 -8.7% 2595 perf-stat.ps.context-switches 78208 ± 23% +112.4% 166150 ± 33% perf-stat.ps.dTLB-load-misses 7.77e+09 -22.8% 5.999e+09 ± 3% perf-stat.ps.dTLB-loads 1.176e+09 -21.5% 9.234e+08 ± 3% perf-stat.ps.dTLB-stores 2712629 ± 3% -14.2% 2327097 ± 4% perf-stat.ps.iTLB-load-misses 2.228e+10 -23.7% 1.7e+10 ± 3% perf-stat.ps.instructions 1116097 ± 45% +392.1% 5492089 ± 24% perf-stat.ps.node-load-misses 388697 ± 51% +290.2% 1516521 ± 27% perf-stat.ps.node-loads 120277 ± 17% +439.0% 648261 ± 37% perf-stat.ps.node-store-misses 1.408e+12 -23.1% 1.083e+12 ± 3% perf-stat.total.instructions 26590 ± 3% -60.7% 10448 ± 4% slabinfo.anon_vma.active_objs 577.50 ± 3% -60.1% 230.17 ± 4% slabinfo.anon_vma.active_slabs 26590 ± 3% -60.1% 10618 ± 4% slabinfo.anon_vma.num_objs 577.50 ± 3% -60.1% 230.17 ± 4% slabinfo.anon_vma.num_slabs 61948 ± 6% -79.2% 12866 slabinfo.anon_vma_chain.active_objs 971.33 ± 6% -78.8% 206.00 slabinfo.anon_vma_chain.active_slabs 62204 ± 6% -78.8% 13209 slabinfo.anon_vma_chain.num_objs 971.33 ± 6% -78.8% 206.00 slabinfo.anon_vma_chain.num_slabs 141234 ± 2% -23.7% 107719 slabinfo.dentry.active_objs 3385 ± 2% -23.5% 2589 slabinfo.dentry.active_slabs 142212 ± 2% -23.5% 108771 slabinfo.dentry.num_objs 3385 ± 2% -23.5% 2589 slabinfo.dentry.num_slabs 22577 ± 11% -25.4% 16841 ± 3% slabinfo.file_lock_cache.active_objs 610.17 ± 11% -23.2% 468.83 ± 2% slabinfo.file_lock_cache.active_slabs 22590 ± 11% -23.1% 17368 ± 2% slabinfo.file_lock_cache.num_objs 610.17 ± 11% -23.2% 468.83 ± 2% slabinfo.file_lock_cache.num_slabs 29547 ± 4% -67.3% 9675 slabinfo.filp.active_objs 954.50 ± 4% -58.6% 395.00 ± 2% slabinfo.filp.active_slabs 30553 ± 4% -58.6% 12658 ± 2% slabinfo.filp.num_objs 954.50 ± 4% -58.6% 395.00 ± 2% slabinfo.filp.num_slabs 9641 ± 2% -22.8% 7445 slabinfo.kmalloc-256.active_objs 9641 ± 2% -20.5% 7666 slabinfo.kmalloc-256.num_objs 5391 ± 5% -31.1% 3712 slabinfo.kmalloc-2k.active_objs 340.67 ± 4% -30.4% 237.17 slabinfo.kmalloc-2k.active_slabs 5460 ± 4% -30.3% 3806 slabinfo.kmalloc-2k.num_objs 340.67 ± 4% -30.4% 237.17 slabinfo.kmalloc-2k.num_slabs 15908 ± 3% -26.3% 11720 slabinfo.kmalloc-512.active_objs 498.00 ± 3% -25.5% 370.83 slabinfo.kmalloc-512.active_slabs 15945 ± 3% -25.4% 11888 slabinfo.kmalloc-512.num_objs 498.00 ± 3% -25.5% 370.83 slabinfo.kmalloc-512.num_slabs 3377 ± 7% -18.1% 2767 ± 6% slabinfo.kmalloc-cg-1k.active_objs 3377 ± 7% -18.1% 2767 ± 6% slabinfo.kmalloc-cg-1k.num_objs 554.67 ± 13% -46.2% 298.67 ± 22% slabinfo.kmalloc-rcl-128.active_objs 554.67 ± 13% -26.0% 410.67 ± 15% slabinfo.kmalloc-rcl-128.num_objs 7747 ± 5% -16.4% 6475 ± 5% slabinfo.kmalloc-rcl-64.active_objs 7771 ± 5% -16.3% 6501 ± 5% slabinfo.kmalloc-rcl-64.num_objs 2443 ± 6% -25.5% 1820 ± 2% slabinfo.kmalloc-rcl-96.active_objs 2443 ± 6% -24.4% 1848 ± 5% slabinfo.kmalloc-rcl-96.num_objs 4608 ± 4% -10.1% 4142 ± 5% slabinfo.signal_cache.active_objs 4373 ± 4% -13.9% 3765 slabinfo.skbuff_head_cache.active_objs 4373 ± 4% -13.0% 3802 slabinfo.skbuff_head_cache.num_objs 55302 ± 8% -76.1% 13209 slabinfo.vm_area_struct.active_objs 1384 ± 8% -75.9% 334.17 slabinfo.vm_area_struct.active_slabs 55407 ± 8% -75.9% 13377 slabinfo.vm_area_struct.num_objs 1384 ± 8% -75.9% 334.17 slabinfo.vm_area_struct.num_slabs 14355 -9.7% 12956 slabinfo.vmap_area.active_objs 14358 -9.5% 12988 slabinfo.vmap_area.num_objs stress-ng.lockf.ops 2.5e+07 +-----------------------------------------------------------------+ | | |.++.++.++.++.++.++.++.+.++.++.++.++.++.++.+. +.++.++.+ | 2e+07 |-+ + | | O O O O O O | | OO O O O OO O OO O OO O O OO O OO O OO O O OO OO OO O | 1.5e+07 |-+ O O O | | | 1e+07 |-+ | | | | | 5e+06 |-+ | | | | | 0 +-----------------------------------------------------------------+ stress-ng.lockf.ops_per_sec 400000 +------------------------------------------------------------------+ |.+ .++. .+ .+. +. +. +.+ .+ .++. +.+.++.+ ++. .+. | 350000 |-++ ++ + + + + +.+ + + :+ ++ + | | + | 300000 |-+O O O O O O O O O O O O OO | 250000 |-O OO O OO O OO OO O O O O OO O O OO O O OO OO | | | 200000 |-+ | | | 150000 |-+ | 100000 |-+ | | | 50000 |-+ | | | 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. --- 0DAY/LKP+ Test Infrastructure Open Source Technology Center https://lists.01.org/hyperkitty/list/lkp@lists.01.org Intel Corporation Thanks, Oliver Sang