Greeting,
FYI, we noticed a 10.7% improvement of vm-scalability.throughput due to commit:
commit: 1ebbb21811b76c3b932959787f37985af36f62fa ("mm/page_alloc: explicitly define how __GFP_HIGH non-blocking allocations accesses reserves")
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
in testcase: vm-scalability
on test machine: 96 threads 2 sockets Intel(R) Xeon(R) Platinum 8260L CPU @ 2.40GHz (Cascade Lake) with 128G memory
with following parameters:
runtime: 300s
test: lru-file-mmap-read
cpufreq_governor: performance
test-description: The motivation behind this suite is to exercise functions and regions of the mm/ of the Linux kernel which are of interest to us.
test-url: https://git.kernel.org/cgit/linux/kernel/git/wfg/vm-scalability.git/
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.
=========================================================================================
compiler/cpufreq_governor/kconfig/rootfs/runtime/tbox_group/test/testcase:
gcc-11/performance/x86_64-rhel-8.3/debian-11.1-x86_64-20220510.cgz/300s/lkp-csl-2sp3/lru-file-mmap-read/vm-scalability
commit:
ab35088543 ("mm/page_alloc: explicitly define what alloc flags deplete min reserves")
1ebbb21811 ("mm/page_alloc: explicitly define how __GFP_HIGH non-blocking allocations accesses reserves")
ab3508854353793c 1ebbb21811b76c3b932959787f3
---------------- ---------------------------
%stddev %change %stddev
\ | \
0.21 ? 7% -36.7% 0.13 ? 5% vm-scalability.free_time
341615 +10.1% 375971 vm-scalability.median
281.39 ? 17% -143.5 137.92 ? 16% vm-scalability.stddev%
32701056 +10.7% 36197694 vm-scalability.throughput
160.97 -10.6% 143.88 vm-scalability.time.elapsed_time
160.97 -10.6% 143.88 vm-scalability.time.elapsed_time.max
352302 ? 2% -8.8% 321372 vm-scalability.time.involuntary_context_switches
2788352 ? 7% +30.6% 3640704 ? 6% vm-scalability.time.maximum_resident_set_size
8477 +1.3% 8584 vm-scalability.time.percent_of_cpu_this_job_got
9907 -14.8% 8441 ? 2% vm-scalability.time.system_time
3739 +4.5% 3909 vm-scalability.time.user_time
1.35e+09 ? 4% -18.1% 1.105e+09 ? 2% cpuidle..time
1.184e+08 -14.7% 1.009e+08 turbostat.IRQ
59.12 +1.8% 60.17 turbostat.RAMWatt
5219286 ? 13% -27.2% 3801249 ? 12% meminfo.Active
5211468 ? 13% -27.2% 3793679 ? 12% meminfo.Active(file)
7776815 ? 3% +13.8% 8847221 ? 4% meminfo.MemFree
1.85 ? 2% +0.3 2.10 mpstat.cpu.all.irq%
0.13 ? 3% +0.0 0.16 mpstat.cpu.all.soft%
24.36 +4.0 28.38 ? 2% mpstat.cpu.all.usr%
23.33 ? 2% +17.1% 27.33 ? 3% vmstat.cpu.us
7945791 ? 2% +17.4% 9324811 ? 3% vmstat.memory.free
441308 -3.2% 427108 vmstat.system.in
2851432 ? 6% -26.3% 2102118 ? 12% numa-meminfo.node0.Active
2848009 ? 6% -26.3% 2099497 ? 12% numa-meminfo.node0.Active(file)
4013196 ? 2% +13.8% 4568204 ? 2% numa-meminfo.node0.MemFree
3625142 ? 5% +21.8% 4415591 ? 5% numa-meminfo.node1.MemFree
316.87 ? 18% -34.4% 207.99 ? 12% sched_debug.cfs_rq:/.util_est_enqueued.avg
1189 ? 12% -23.5% 910.39 ? 6% sched_debug.cfs_rq:/.util_est_enqueued.max
324.18 ? 6% -19.2% 261.97 ? 2% sched_debug.cfs_rq:/.util_est_enqueued.stddev
3005 ? 6% +10.6% 3325 ? 2% sched_debug.cpu.nr_switches.min
3072920 ? 8% -26.4% 2261101 ? 7% numa-numastat.node0.numa_foreign
5528896 ? 9% -60.5% 2186586 ? 10% numa-numastat.node0.numa_miss
5590018 ? 9% -59.9% 2239518 ? 10% numa-numastat.node0.other_node
5527881 ? 9% -60.4% 2186984 ? 10% numa-numastat.node1.numa_foreign
3072649 ? 8% -26.4% 2261499 ? 7% numa-numastat.node1.numa_miss
3100633 ? 8% -25.9% 2297621 ? 8% numa-numastat.node1.other_node
714902 ? 6% -26.3% 526979 ? 12% numa-vmstat.node0.nr_active_file
1025543 ? 3% +10.5% 1132734 ? 4% numa-vmstat.node0.nr_free_pages
282.83 ? 7% -23.2% 217.33 ? 12% numa-vmstat.node0.nr_isolated_file
714904 ? 6% -26.3% 526980 ? 12% numa-vmstat.node0.nr_zone_active_file
3072920 ? 8% -26.4% 2261101 ? 7% numa-vmstat.node0.numa_foreign
5528896 ? 9% -60.5% 2186586 ? 10% numa-vmstat.node0.numa_miss
5590018 ? 9% -59.9% 2239518 ? 10% numa-vmstat.node0.numa_other
3239860 ? 4% -18.4% 2642385 ? 6% numa-vmstat.node0.workingset_nodereclaim
931672 ? 6% +18.4% 1103239 ? 5% numa-vmstat.node1.nr_free_pages
5527881 ? 9% -60.4% 2186984 ? 10% numa-vmstat.node1.numa_foreign
3072649 ? 8% -26.4% 2261499 ? 7% numa-vmstat.node1.numa_miss
3100633 ? 8% -25.9% 2297621 ? 8% numa-vmstat.node1.numa_other
228122 ? 6% -13.3% 197862 proc-vmstat.allocstall_movable
6249 ? 5% +144.5% 15278 ? 2% proc-vmstat.allocstall_normal
39122902 ? 21% -81.3% 7326649 ? 23% proc-vmstat.compact_daemon_free_scanned
1.885e+08 ? 17% -74.6% 47875280 ? 26% proc-vmstat.compact_daemon_migrate_scanned
3493 ? 8% -56.6% 1515 ? 6% proc-vmstat.compact_daemon_wake
140286 ? 15% -92.5% 10548 ? 38% proc-vmstat.compact_fail
57183012 ? 15% -83.7% 9325762 ? 23% proc-vmstat.compact_free_scanned
2623027 ? 14% -52.3% 1250898 ? 10% proc-vmstat.compact_isolated
5.211e+08 ? 20% -88.0% 62772564 ? 29% proc-vmstat.compact_migrate_scanned
446104 ? 15% -90.9% 40548 ? 25% proc-vmstat.compact_stall
305818 ? 16% -90.2% 30000 ? 21% proc-vmstat.compact_success
9202 ? 10% -40.0% 5520 ? 7% proc-vmstat.kswapd_low_wmark_hit_quickly
1305504 ? 13% -26.7% 957580 ? 12% proc-vmstat.nr_active_file
1985908 ? 4% +12.9% 2241333 ? 3% proc-vmstat.nr_free_pages
1015907 -2.7% 988196 proc-vmstat.nr_page_table_pages
549360 -2.5% 535474 proc-vmstat.nr_slab_reclaimable
69677 -3.0% 67616 proc-vmstat.nr_slab_unreclaimable
1305512 ? 13% -26.7% 957587 ? 12% proc-vmstat.nr_zone_active_file
8600801 ? 8% -48.3% 4448085 ? 7% proc-vmstat.numa_foreign
4400 ? 50% -81.1% 833.83 ? 80% proc-vmstat.numa_hint_faults_local
51279456 ? 3% -5.2% 48609440 ? 2% proc-vmstat.numa_hit
51194728 ? 3% -5.2% 48521975 ? 2% proc-vmstat.numa_local
8601545 ? 8% -48.3% 4448085 ? 7% proc-vmstat.numa_miss
8690652 ? 8% -47.8% 4537139 ? 7% proc-vmstat.numa_other
9311 ? 11% -39.5% 5629 ? 7% proc-vmstat.pageoutrun
14184591 ? 2% -12.8% 12364026 proc-vmstat.pgalloc_dma32
23108 ? 6% -46.7% 12308 ? 16% proc-vmstat.pgmajfault
1264562 ? 14% -52.0% 606868 ? 10% proc-vmstat.pgmigrate_success
28373 ? 2% -5.4% 26846 ? 2% proc-vmstat.pgreuse
1.831e+09 +3.5% 1.896e+09 proc-vmstat.pgscan_direct
2.641e+08 ? 10% -25.1% 1.978e+08 ? 11% proc-vmstat.pgscan_kswapd
9.765e+08 +3.0% 1.006e+09 proc-vmstat.pgsteal_direct
68399366 ? 4% -45.6% 37214972 proc-vmstat.pgsteal_kswapd
16621148 -5.3% 15745211 proc-vmstat.slabs_scanned
358.33 ? 55% +307.9% 1461 ? 47% proc-vmstat.unevictable_pgs_culled
292.00 ? 65% +377.5% 1394 ? 47% proc-vmstat.unevictable_pgs_rescued
1899136 -7.0% 1765248 ? 2% proc-vmstat.unevictable_pgs_scanned
5904732 -5.8% 5563946 proc-vmstat.workingset_nodereclaim
3099373 -1.6% 3049129 proc-vmstat.workingset_nodes
3.734e+10 +8.0% 4.031e+10 perf-stat.i.branch-instructions
29422732 -5.2% 27892505 perf-stat.i.branch-misses
2.064e+08 +9.2% 2.255e+08 perf-stat.i.cache-misses
7.005e+08 +7.1% 7.5e+08 perf-stat.i.cache-references
2.41 ? 2% -7.7% 2.23 perf-stat.i.cpi
177.46 -6.8% 165.33 perf-stat.i.cpu-migrations
1391 ? 2% -9.7% 1255 perf-stat.i.cycles-between-cache-misses
3.476e+10 +7.9% 3.752e+10 perf-stat.i.dTLB-loads
0.02 ? 3% -0.0 0.02 ? 3% perf-stat.i.dTLB-store-miss-rate%
5.566e+09 +7.8% 5.998e+09 perf-stat.i.dTLB-stores
1.222e+11 +6.8% 1.305e+11 perf-stat.i.instructions
39940 ? 2% +6.9% 42712 perf-stat.i.instructions-per-iTLB-miss
0.49 +6.6% 0.52 perf-stat.i.ipc
210782 +11.7% 235498 perf-stat.i.major-faults
471.82 -7.4% 436.87 perf-stat.i.metric.K/sec
813.09 +7.9% 877.53 perf-stat.i.metric.M/sec
213062 +11.6% 237881 perf-stat.i.minor-faults
17079151 ? 2% -11.9% 15046137 perf-stat.i.node-load-misses
11057787 ? 2% -13.3% 9585653 perf-stat.i.node-loads
57.95 ? 2% -8.2 49.73 perf-stat.i.node-store-miss-rate%
7350966 -14.7% 6266723 perf-stat.i.node-store-misses
5862252 ? 3% +17.9% 6910693 ? 2% perf-stat.i.node-stores
423844 +11.7% 473380 perf-stat.i.page-faults
0.08 -0.0 0.07 perf-stat.overall.branch-miss-rate%
29.46 +0.6 30.07 perf-stat.overall.cache-miss-rate%
2.21 -6.2% 2.07 perf-stat.overall.cpi
1309 -8.3% 1200 perf-stat.overall.cycles-between-cache-misses
0.03 ? 2% -0.0 0.03 ? 2% perf-stat.overall.dTLB-load-miss-rate%
40576 ? 2% +8.3% 43947 perf-stat.overall.instructions-per-iTLB-miss
0.45 +6.6% 0.48 perf-stat.overall.ipc
55.96 ? 2% -8.2 47.75 perf-stat.overall.node-store-miss-rate%
4037 -4.5% 3854 perf-stat.overall.path-length
3.687e+10 +8.0% 3.982e+10 perf-stat.ps.branch-instructions
29098195 -5.3% 27565060 perf-stat.ps.branch-misses
2.038e+08 +9.4% 2.229e+08 perf-stat.ps.cache-misses
6.917e+08 +7.2% 7.412e+08 perf-stat.ps.cache-references
175.77 -6.8% 163.78 perf-stat.ps.cpu-migrations
3.433e+10 +8.0% 3.706e+10 perf-stat.ps.dTLB-loads
5.495e+09 +7.8% 5.925e+09 perf-stat.ps.dTLB-stores
1.207e+11 +6.8% 1.289e+11 perf-stat.ps.instructions
207772 +11.9% 232413 perf-stat.ps.major-faults
210008 +11.8% 234752 perf-stat.ps.minor-faults
17010095 ? 2% -12.1% 14954361 perf-stat.ps.node-load-misses
10989225 ? 2% -13.4% 9514094 perf-stat.ps.node-loads
7316949 -14.9% 6227549 perf-stat.ps.node-store-misses
5760441 ? 3% +18.3% 6816855 ? 2% perf-stat.ps.node-stores
417780 +11.8% 467166 perf-stat.ps.page-faults
1.951e+13 -4.5% 1.862e+13 perf-stat.total.instructions
20.60 ? 86% -20.6 0.00 perf-profile.calltrace.cycles-pp.do_access
17.03 ? 88% -17.0 0.00 perf-profile.calltrace.cycles-pp.asm_exc_page_fault.do_access
16.93 ? 88% -16.9 0.00 perf-profile.calltrace.cycles-pp.exc_page_fault.asm_exc_page_fault.do_access
16.93 ? 88% -16.9 0.00 perf-profile.calltrace.cycles-pp.do_user_addr_fault.exc_page_fault.asm_exc_page_fault.do_access
16.86 ? 88% -16.9 0.00 perf-profile.calltrace.cycles-pp.handle_mm_fault.do_user_addr_fault.exc_page_fault.asm_exc_page_fault.do_access
16.24 ? 89% -16.2 0.00 perf-profile.calltrace.cycles-pp.filemap_fault.__do_fault.do_read_fault.do_fault.__handle_mm_fault
16.20 ? 89% -16.2 0.00 perf-profile.calltrace.cycles-pp.page_cache_ra_order.filemap_fault.__do_fault.do_read_fault.do_fault
16.25 ? 89% -15.1 1.15 ?223% perf-profile.calltrace.cycles-pp.__do_fault.do_read_fault.do_fault.__handle_mm_fault.handle_mm_fault
16.76 ? 88% -15.1 1.69 ?151% perf-profile.calltrace.cycles-pp.do_read_fault.do_fault.__handle_mm_fault.handle_mm_fault.do_user_addr_fault
14.87 ? 74% -14.9 0.00 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.unlinkat
14.87 ? 74% -14.9 0.00 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.unlinkat
14.87 ? 74% -14.9 0.00 perf-profile.calltrace.cycles-pp.unlinkat
14.86 ? 74% -14.9 0.00 perf-profile.calltrace.cycles-pp.__x64_sys_unlinkat.do_syscall_64.entry_SYSCALL_64_after_hwframe.unlinkat
14.86 ? 74% -14.9 0.00 perf-profile.calltrace.cycles-pp.do_unlinkat.__x64_sys_unlinkat.do_syscall_64.entry_SYSCALL_64_after_hwframe.unlinkat
14.86 ? 74% -14.9 0.00 perf-profile.calltrace.cycles-pp.evict.do_unlinkat.__x64_sys_unlinkat.do_syscall_64.entry_SYSCALL_64_after_hwframe
14.84 ? 74% -14.8 0.00 perf-profile.calltrace.cycles-pp.truncate_inode_pages_range.evict.do_unlinkat.__x64_sys_unlinkat.do_syscall_64
16.77 ? 88% -14.5 2.26 ?141% perf-profile.calltrace.cycles-pp.do_fault.__handle_mm_fault.handle_mm_fault.do_user_addr_fault.exc_page_fault
9.75 ? 95% -9.7 0.00 perf-profile.calltrace.cycles-pp.folio_alloc.page_cache_ra_order.filemap_fault.__do_fault.do_read_fault
9.74 ? 95% -9.7 0.00 perf-profile.calltrace.cycles-pp.__alloc_pages.folio_alloc.page_cache_ra_order.filemap_fault.__do_fault
9.11 ? 97% -9.1 0.00 perf-profile.calltrace.cycles-pp.__alloc_pages_slowpath.__alloc_pages.folio_alloc.page_cache_ra_order.filemap_fault
7.66 ? 70% -7.7 0.00 perf-profile.calltrace.cycles-pp.do_rw_once
7.15 ? 77% -7.2 0.00 perf-profile.calltrace.cycles-pp.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle
5.99 ? 72% -6.0 0.00 perf-profile.calltrace.cycles-pp.truncate_folio_batch_exceptionals.truncate_inode_pages_range.evict.do_unlinkat.__x64_sys_unlinkat
5.92 ? 76% -5.9 0.00 perf-profile.calltrace.cycles-pp.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call
5.77 ? 80% -5.8 0.00 perf-profile.calltrace.cycles-pp.read_pages.page_cache_ra_order.filemap_fault.__do_fault.do_read_fault
5.77 ? 80% -5.8 0.00 perf-profile.calltrace.cycles-pp.iomap_readahead.read_pages.page_cache_ra_order.filemap_fault.__do_fault
5.72 ? 80% -5.7 0.00 perf-profile.calltrace.cycles-pp.iomap_readpage_iter.iomap_readahead.read_pages.page_cache_ra_order.filemap_fault
5.58 ? 80% -5.6 0.00 perf-profile.calltrace.cycles-pp.zero_user_segments.iomap_readpage_iter.iomap_readahead.read_pages.page_cache_ra_order
5.06 ? 72% -5.1 0.00 perf-profile.calltrace.cycles-pp.xas_store.truncate_folio_batch_exceptionals.truncate_inode_pages_range.evict.do_unlinkat
4.85 ? 79% -4.9 0.00 perf-profile.calltrace.cycles-pp.memset_erms.zero_user_segments.iomap_readpage_iter.iomap_readahead.read_pages
4.75 ? 75% -4.8 0.00 perf-profile.calltrace.cycles-pp.find_lock_entries.truncate_inode_pages_range.evict.do_unlinkat.__x64_sys_unlinkat
6.56 ? 33% -4.7 1.86 ?156% perf-profile.calltrace.cycles-pp.ret_from_fork
6.56 ? 33% -4.7 1.86 ?156% perf-profile.calltrace.cycles-pp.kthread.ret_from_fork
0.00 +13.8 13.82 ? 73% perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe
0.00 +14.4 14.40 ? 70% perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe
22.17 ? 83% -22.2 0.00 perf-profile.children.cycles-pp.do_access
16.27 ? 88% -16.3 0.00 perf-profile.children.cycles-pp.filemap_fault
16.23 ? 88% -16.2 0.00 perf-profile.children.cycles-pp.page_cache_ra_order
16.28 ? 88% -15.7 0.58 ?223% perf-profile.children.cycles-pp.__do_fault
14.87 ? 74% -14.9 0.00 perf-profile.children.cycles-pp.unlinkat
14.86 ? 74% -14.9 0.00 perf-profile.children.cycles-pp.__x64_sys_unlinkat
14.86 ? 74% -14.9 0.00 perf-profile.children.cycles-pp.do_unlinkat
14.86 ? 74% -14.9 0.00 perf-profile.children.cycles-pp.evict
14.84 ? 74% -14.8 0.00 perf-profile.children.cycles-pp.truncate_inode_pages_range
16.84 ? 87% -14.6 2.26 ?141% perf-profile.children.cycles-pp.do_fault
16.83 ? 87% -14.6 2.26 ?141% perf-profile.children.cycles-pp.do_read_fault
9.75 ? 95% -9.8 0.00 perf-profile.children.cycles-pp.folio_alloc
9.11 ? 97% -9.1 0.00 perf-profile.children.cycles-pp.__alloc_pages_slowpath
7.66 ? 96% -7.1 0.58 ?223% perf-profile.children.cycles-pp.native_queued_spin_lock_slowpath
7.76 ? 61% -6.6 1.15 ?223% perf-profile.children.cycles-pp.asm_sysvec_apic_timer_interrupt
6.54 ?104% -6.5 0.00 perf-profile.children.cycles-pp.shrink_node
6.45 ?104% -6.4 0.00 perf-profile.children.cycles-pp.shrink_node_memcgs
6.24 ? 69% -6.2 0.00 perf-profile.children.cycles-pp.do_rw_once
6.02 ? 72% -6.0 0.00 perf-profile.children.cycles-pp.truncate_folio_batch_exceptionals
6.56 ? 71% -6.0 0.58 ?223% perf-profile.children.cycles-pp.sysvec_apic_timer_interrupt
5.83 ? 65% -5.8 0.00 perf-profile.children.cycles-pp.xas_store
5.80 ? 79% -5.8 0.00 perf-profile.children.cycles-pp.read_pages
5.80 ? 79% -5.8 0.00 perf-profile.children.cycles-pp.iomap_readahead
5.75 ? 79% -5.7 0.00 perf-profile.children.cycles-pp.iomap_readpage_iter
5.56 ?103% -5.6 0.00 perf-profile.children.cycles-pp.shrink_lruvec
5.54 ?103% -5.5 0.00 perf-profile.children.cycles-pp.shrink_inactive_list
5.09 ? 69% -5.1 0.00 perf-profile.children.cycles-pp.compact_zone
4.76 ? 75% -4.8 0.00 perf-profile.children.cycles-pp.find_lock_entries
6.56 ? 33% -4.7 1.86 ?156% perf-profile.children.cycles-pp.ret_from_fork
6.56 ? 33% -4.7 1.86 ?156% perf-profile.children.cycles-pp.kthread
4.42 ? 73% -4.4 0.00 perf-profile.children.cycles-pp.isolate_migratepages
5.61 ? 80% -4.3 1.28 ?223% perf-profile.children.cycles-pp.zero_user_segments
5.56 ? 79% -4.3 1.28 ?223% perf-profile.children.cycles-pp.memset_erms
4.08 ? 69% -2.3 1.76 ?153% perf-profile.children.cycles-pp._raw_spin_lock_irqsave
0.02 ?142% +13.1 13.17 ? 47% perf-profile.children.cycles-pp.__x64_sys_openat
0.02 ?142% +13.1 13.17 ? 47% perf-profile.children.cycles-pp.do_sys_openat2
0.00 +16.2 16.19 ? 30% perf-profile.children.cycles-pp.cmd_record
0.00 +16.2 16.19 ? 30% perf-profile.children.cycles-pp.__cmd_record
18.04 ? 62% +33.0 51.05 ? 21% perf-profile.children.cycles-pp.do_syscall_64
18.04 ? 62% +33.6 51.63 ? 22% perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe
7.66 ? 96% -7.1 0.58 ?223% perf-profile.self.cycles-pp.native_queued_spin_lock_slowpath
4.27 ? 73% -4.3 0.00 perf-profile.self.cycles-pp.do_access
5.49 ? 79% -4.2 1.28 ?223% perf-profile.self.cycles-pp.memset_erms
3.92 ? 69% -3.9 0.00 perf-profile.self.cycles-pp.do_rw_once
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.
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests