FYI, we noticed a -23.7% regression of netperf.Throughput_Mbps due to commit: commit 0cef819730f7f67f8d2bec642e1ac1f59febb4eb ("sched/fair: Fix calc_cfs_shares()") https://git.kernel.org/pub/scm/linux/kernel/git/peterz/queue.git sched/urgent in testcase: netperf on test machine: lkp-hsw-d01: 8 threads Haswell with 8G memory with following parameters: cluster=cs-localhost/cpufreq_governor=performance/ip=ipv4/nr_threads=200%/runtime=300s/send_size=10K/test=SCTP_STREAM_MANY In addition to that, the commit also has significant impact on the following tests: +------------------+----------------------------------------------------------------------------------------------------+ | testcase: change | ebizzy: ebizzy.throughput.per_thread.max -39.6% improvement | | test machine | ivb42: 48 threads 2 sockets Xeon E5 (Ivytown Ivy Bridge-EP) with 64G memory | | test parameters | cpufreq_governor=performance | | | duration=10s | | | iterations=100x | | | nr_threads=200% | +------------------+----------------------------------------------------------------------------------------------------+ | testcase: change | plzip: plzip.throughput 2.5% improvement | | test machine | lkp-hsx04: 144 threads Brickland Haswell-EX with 512G memory | | test parameters | cpufreq_governor=performance | | | nr_threads=100% | +------------------+----------------------------------------------------------------------------------------------------+ | testcase: change | netperf: netperf.Throughput_Mbps -22.4% regression | | test machine | lkp-bdw-de1: 16 threads Broadwell-DE with 8G memory | | test parameters | cluster=cs-localhost | | | cpufreq_governor=performance | | | ip=ipv4 | | | nr_threads=200% | | | runtime=300s | | | send_size=10K | | | test=SCTP_STREAM_MANY | +------------------+----------------------------------------------------------------------------------------------------+ 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 ========================================================================================= cluster/compiler/cpufreq_governor/ip/kconfig/nr_threads/rootfs/runtime/send_size/tbox_group/test/testcase: cs-localhost/gcc-4.9/performance/ipv4/x86_64-rhel/200%/debian-x86_64-2015-02-07.cgz/300s/10K/lkp-hsw-d01/SCTP_STREAM_MANY/netperf commit: 0329dacce8 ("sched/fair: Fix effective_load()") 0cef819730 ("sched/fair: Fix calc_cfs_shares()") 0329dacce87d6741 0cef819730f7f67f8d2bec642e ---------------- -------------------------- fail:runs %reproduction fail:runs | | | :4 25% 1:4 kmsg.DHCP/BOOTP:Reply_not_for_us,op[#]xid[a99a731] %stddev %change %stddev \ | \ 5238 ± 0% -23.7% 3995 ± 2% netperf.Throughput_Mbps 39356248 ± 0% -100.0% 13112 ± 7% netperf.time.involuntary_context_switches 395.00 ± 0% +32.8% 524.50 ± 1% netperf.time.percent_of_cpu_this_job_got 1159 ± 0% +32.9% 1541 ± 1% netperf.time.system_time 30.44 ± 1% +44.1% 43.86 ± 2% netperf.time.user_time 1856009 ± 11% +641.3% 13757673 ± 2% netperf.time.voluntary_context_switches 7674 ± 7% -12.1% 6745 ± 4% slabinfo.anon_vma_chain.active_objs 7705 ± 8% -12.4% 6746 ± 4% slabinfo.anon_vma_chain.num_objs 43984820 ± 0% -21.7% 34430029 ± 2% softirqs.NET_RX 136049 ± 3% -54.4% 62074 ± 14% softirqs.RCU 19.25 ± 2% +9.1% 21.00 ± 0% vmstat.procs.r 281069 ± 0% -64.5% 99885 ± 2% vmstat.system.cs 3.949e+08 ± 0% -23.2% 3.033e+08 ± 2% proc-vmstat.numa_hit 3.949e+08 ± 0% -23.2% 3.033e+08 ± 2% proc-vmstat.numa_local 1.013e+09 ± 1% -28.8% 7.213e+08 ± 2% proc-vmstat.pgalloc_dma32 1.247e+09 ± 0% -29.7% 8.766e+08 ± 2% proc-vmstat.pgalloc_normal 2.26e+09 ± 0% -29.3% 1.598e+09 ± 2% proc-vmstat.pgfree 0.05 ± 0% +45.0% 0.07 ± 11% turbostat.CPU%c1 0.65 ± 0% +32.0% 0.85 ± 1% turbostat.CPU%c7 58.58 ± 0% -8.9% 53.34 ± 1% turbostat.CorWatt 0.58 ± 5% +35.1% 0.78 ± 4% turbostat.Pkg%pc2 66.81 ± 0% -6.6% 62.43 ± 0% turbostat.PkgWatt 17187 ± 12% +60.8% 27643 ± 23% cpuidle.C1E-HSW.time 127946 ± 7% +20.3% 153920 ± 11% cpuidle.C3-HSW.time 377.00 ± 6% +26.1% 475.50 ± 7% cpuidle.C3-HSW.usage 218.00 ± 7% +12.5% 245.25 ± 3% cpuidle.C6-HSW.usage 16143034 ± 0% +33.2% 21498649 ± 1% cpuidle.C7s-HSW.time 17605 ± 0% +32.6% 23344 ± 1% cpuidle.C7s-HSW.usage 4472 ±101% +1802.2% 85081 ± 97% cpuidle.POLL.time 4.586e+11 ± 1% -23.7% 3.499e+11 ± 3% perf-stat.L1-dcache-load-misses 1.394e+12 ± 0% -23.2% 1.071e+12 ± 2% perf-stat.L1-dcache-loads 1.101e+12 ± 0% -23.3% 8.446e+11 ± 2% perf-stat.L1-dcache-stores 7.379e+10 ± 0% -23.5% 5.646e+10 ± 13% perf-stat.L1-icache-load-misses 8.397e+08 ± 25% +1022.0% 9.421e+09 ± 11% perf-stat.LLC-load-misses 8.37e+10 ± 0% -13.4% 7.252e+10 ± 2% perf-stat.LLC-loads 7.722e+08 ± 22% +789.9% 6.872e+09 ± 5% perf-stat.LLC-store-misses 3.638e+10 ± 0% -22.3% 2.827e+10 ± 2% perf-stat.LLC-stores 7.208e+11 ± 0% -23.6% 5.504e+11 ± 2% perf-stat.branch-instructions 8.614e+09 ± 1% -27.1% 6.279e+09 ± 1% perf-stat.branch-load-misses 7.213e+11 ± 0% -23.7% 5.505e+11 ± 2% perf-stat.branch-loads 8.595e+09 ± 1% -26.8% 6.294e+09 ± 1% perf-stat.branch-misses 1.672e+09 ± 21% +879.5% 1.638e+10 ± 4% perf-stat.cache-misses 1.752e+11 ± 0% -19.7% 1.407e+11 ± 2% perf-stat.cache-references 85431681 ± 0% -64.5% 30358026 ± 2% perf-stat.context-switches 1.487e+09 ± 4% -38.2% 9.194e+08 ± 3% perf-stat.dTLB-load-misses 1.395e+12 ± 0% -23.3% 1.071e+12 ± 2% perf-stat.dTLB-loads 1.092e+09 ± 12% -38.8% 6.685e+08 ± 26% perf-stat.dTLB-store-misses 1.098e+12 ± 0% -23.1% 8.444e+11 ± 2% perf-stat.dTLB-stores 7.807e+08 ± 7% -39.8% 4.702e+08 ± 10% perf-stat.iTLB-load-misses 3.924e+12 ± 0% -23.2% 3.011e+12 ± 2% perf-stat.instructions 321108 ± 0% +4.4% 335205 ± 0% perf-stat.minor-faults 8.198e+08 ± 24% +1057.7% 9.491e+09 ± 11% perf-stat.node-loads 7.678e+08 ± 22% +792.1% 6.849e+09 ± 5% perf-stat.node-stores 321076 ± 0% +4.4% 335171 ± 0% perf-stat.page-faults 34249 ± 19% +2035.3% 731325 ± 4% sched_debug.cfs_rq:/.MIN_vruntime.avg 71932 ± 12% +1049.4% 826757 ± 1% sched_debug.cfs_rq:/.MIN_vruntime.max 34055 ± 14% +494.7% 202535 ± 29% sched_debug.cfs_rq:/.MIN_vruntime.stddev 1409000 ± 3% -84.4% 219938 ± 3% sched_debug.cfs_rq:/.load.avg 1911874 ± 6% -85.7% 273218 ± 9% sched_debug.cfs_rq:/.load.max 950054 ± 9% -83.6% 156098 ± 5% sched_debug.cfs_rq:/.load.min 436947 ± 9% -91.0% 39520 ± 15% sched_debug.cfs_rq:/.load.stddev 978.81 ± 0% -77.7% 218.13 ± 1% sched_debug.cfs_rq:/.load_avg.avg 1133 ± 5% -68.4% 357.96 ± 19% sched_debug.cfs_rq:/.load_avg.max 873.67 ± 0% -81.8% 158.67 ± 3% sched_debug.cfs_rq:/.load_avg.min 34249 ± 19% +2035.3% 731325 ± 4% sched_debug.cfs_rq:/.max_vruntime.avg 71932 ± 12% +1049.3% 826757 ± 1% sched_debug.cfs_rq:/.max_vruntime.max 34055 ± 14% +494.7% 202535 ± 29% sched_debug.cfs_rq:/.max_vruntime.stddev 76109 ± 0% +975.0% 818186 ± 1% sched_debug.cfs_rq:/.min_vruntime.avg 78283 ± 0% +956.7% 827244 ± 1% sched_debug.cfs_rq:/.min_vruntime.max 74946 ± 0% +980.7% 809948 ± 1% sched_debug.cfs_rq:/.min_vruntime.min 1152 ± 15% +431.4% 6124 ± 51% sched_debug.cfs_rq:/.min_vruntime.stddev 1.36 ± 3% +25.7% 1.71 ± 3% sched_debug.cfs_rq:/.nr_running.avg 0.42 ± 9% -37.5% 0.26 ± 35% sched_debug.cfs_rq:/.nr_running.stddev 885.45 ± 0% -82.8% 152.63 ± 4% sched_debug.cfs_rq:/.runnable_load_avg.avg 939.21 ± 0% -81.3% 175.71 ± 3% sched_debug.cfs_rq:/.runnable_load_avg.max 817.71 ± 2% -87.5% 102.08 ± 28% sched_debug.cfs_rq:/.runnable_load_avg.min 834.75 ±107% +758.4% 7165 ± 49% sched_debug.cfs_rq:/.spread0.max -2503 ±-26% +304.8% -10132 ±-83% sched_debug.cfs_rq:/.spread0.min 1153 ± 15% +431.2% 6125 ± 51% sched_debug.cfs_rq:/.spread0.stddev 35.61 ± 8% -15.3% 30.16 ± 9% sched_debug.cfs_rq:/.util_avg.stddev 640322 ± 15% -48.5% 329643 ± 26% sched_debug.cpu.avg_idle.avg 936653 ± 2% -19.0% 758535 ± 19% sched_debug.cpu.avg_idle.max 303015 ± 69% -92.6% 22535 ±109% sched_debug.cpu.avg_idle.min 876.64 ± 1% -82.4% 154.60 ± 1% sched_debug.cpu.cpu_load[0].avg 940.50 ± 0% -81.4% 174.96 ± 3% sched_debug.cpu.cpu_load[0].max 781.58 ± 8% -86.6% 104.75 ± 11% sched_debug.cpu.cpu_load[0].min 55.31 ± 36% -58.3% 23.05 ± 20% sched_debug.cpu.cpu_load[0].stddev 885.32 ± 0% -82.4% 155.97 ± 1% sched_debug.cpu.cpu_load[1].avg 929.00 ± 0% -81.3% 173.33 ± 3% sched_debug.cpu.cpu_load[1].max 810.17 ± 3% -84.8% 123.25 ± 5% sched_debug.cpu.cpu_load[1].min 37.05 ± 29% -56.9% 15.98 ± 23% sched_debug.cpu.cpu_load[1].stddev 883.74 ± 0% -82.3% 156.01 ± 0% sched_debug.cpu.cpu_load[2].avg 918.62 ± 0% -81.3% 171.71 ± 4% sched_debug.cpu.cpu_load[2].max 815.62 ± 2% -83.6% 133.75 ± 3% sched_debug.cpu.cpu_load[2].min 31.38 ± 21% -61.6% 12.05 ± 28% sched_debug.cpu.cpu_load[2].stddev 879.80 ± 0% -82.2% 156.20 ± 0% sched_debug.cpu.cpu_load[3].avg 909.04 ± 0% -81.3% 170.00 ± 4% sched_debug.cpu.cpu_load[3].max 819.21 ± 1% -82.9% 140.08 ± 3% sched_debug.cpu.cpu_load[3].min 27.59 ± 20% -66.0% 9.39 ± 33% sched_debug.cpu.cpu_load[3].stddev 876.12 ± 0% -82.1% 156.95 ± 0% sched_debug.cpu.cpu_load[4].avg 902.17 ± 0% -81.3% 168.67 ± 3% sched_debug.cpu.cpu_load[4].max 819.79 ± 1% -82.4% 144.42 ± 3% sched_debug.cpu.cpu_load[4].min 25.51 ± 20% -69.7% 7.72 ± 41% sched_debug.cpu.cpu_load[4].stddev 1479789 ± 4% -84.7% 226291 ± 3% sched_debug.cpu.load.avg 1955568 ± 3% -84.0% 313014 ± 21% sched_debug.cpu.load.max 908371 ± 8% -82.9% 155654 ± 8% sched_debug.cpu.load.min 462669 ± 7% -88.6% 52549 ± 41% sched_debug.cpu.load.stddev 5283149 ± 0% -64.5% 1876874 ± 2% sched_debug.cpu.nr_switches.avg 6042769 ± 3% -64.7% 2132662 ± 3% sched_debug.cpu.nr_switches.max 5074839 ± 0% -66.0% 1723693 ± 1% sched_debug.cpu.nr_switches.min 315533 ± 16% -60.2% 125660 ± 15% sched_debug.cpu.nr_switches.stddev 18.91 ± 1% +70.5% 32.25 ± 3% perf-profile.cycles-pp.___sys_recvmsg.__sys_recvmsg.sys_recvmsg.entry_SYSCALL_64_fastpath 77.21 ± 0% -16.8% 64.27 ± 1% perf-profile.cycles-pp.___sys_sendmsg.__sys_sendmsg.sys_sendmsg.entry_SYSCALL_64_fastpath 5.63 ± 5% -33.2% 3.76 ± 7% perf-profile.cycles-pp.__alloc_pages_nodemask.alloc_kmem_pages_node.kmalloc_large_node.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34 5.76 ± 4% -20.1% 4.60 ± 2% perf-profile.cycles-pp.__alloc_skb._sctp_make_chunk.sctp_make_datafrag_empty.sctp_datamsg_from_user.sctp_sendmsg 1.89 ± 2% -41.3% 1.11 ± 14% perf-profile.cycles-pp.__alloc_skb.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24 12.31 ± 1% +42.2% 17.51 ± 2% perf-profile.cycles-pp.__do_softirq.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip.ip_finish_output2 4.45 ± 3% -27.8% 3.21 ± 3% perf-profile.cycles-pp.__do_softirq.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip.sctp_inq_push 3.42 ± 3% -23.9% 2.60 ± 5% perf-profile.cycles-pp.__free_kmem_pages.kfree.skb_free_head.skb_release_data.skb_release_all 3.37 ± 3% -23.6% 2.58 ± 5% perf-profile.cycles-pp.__free_pages.__free_kmem_pages.kfree.skb_free_head.skb_release_data 3.35 ± 3% -23.9% 2.55 ± 5% perf-profile.cycles-pp.__free_pages_ok.__free_pages.__free_kmem_pages.kfree.skb_free_head 4.57 ± 7% -21.6% 3.58 ± 2% perf-profile.cycles-pp.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb._sctp_make_chunk.sctp_make_datafrag_empty 1.59 ± 4% -62.5% 0.60 ± 12% perf-profile.cycles-pp.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb.sctp_packet_transmit.sctp_outq_flush 4.66 ± 7% -21.6% 3.65 ± 2% perf-profile.cycles-pp.__kmalloc_reserve.isra.34.__alloc_skb._sctp_make_chunk.sctp_make_datafrag_empty.sctp_datamsg_from_user 1.62 ± 3% -54.9% 0.73 ± 30% perf-profile.cycles-pp.__kmalloc_reserve.isra.34.__alloc_skb.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork 12.44 ± 1% +41.7% 17.63 ± 2% perf-profile.cycles-pp.__local_bh_enable_ip.ip_finish_output2.ip_finish_output.ip_output.ip_local_out 4.57 ± 4% -27.8% 3.30 ± 3% perf-profile.cycles-pp.__local_bh_enable_ip.sctp_inq_push.sctp_backlog_rcv.__release_sock.release_sock 16.03 ± 2% +24.8% 20.01 ± 2% perf-profile.cycles-pp.__netif_receive_skb.process_backlog.net_rx_action.__do_softirq.do_softirq_own_stack 15.98 ± 2% +24.9% 19.96 ± 2% perf-profile.cycles-pp.__netif_receive_skb_core.__netif_receive_skb.process_backlog.net_rx_action.__do_softirq 0.00 ± -1% +Inf% 1.56 ± 4% perf-profile.cycles-pp.__release_sock.release_sock.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 24.48 ± 3% -16.7% 20.40 ± 1% perf-profile.cycles-pp.__release_sock.release_sock.sctp_sendmsg.inet_sendmsg.sock_sendmsg 1.13 ± 2% -100.0% 0.00 ± -1% perf-profile.cycles-pp.__schedule.preempt_schedule_common._cond_resched.__release_sock.release_sock 1.62 ± 5% -51.2% 0.79 ± 4% perf-profile.cycles-pp.__schedule.schedule.schedule_timeout.sctp_skb_recv_datagram.sctp_recvmsg 19.51 ± 1% +68.0% 32.78 ± 3% perf-profile.cycles-pp.__sys_recvmsg.sys_recvmsg.entry_SYSCALL_64_fastpath 78.31 ± 0% -16.7% 65.22 ± 1% perf-profile.cycles-pp.__sys_sendmsg.sys_sendmsg.entry_SYSCALL_64_fastpath 1.95 ± 4% -63.3% 0.72 ± 12% perf-profile.cycles-pp.__wake_up_common.__wake_up_sync_key.sctp_data_ready.sctp_ulpq_tail_event.sctp_ulpq_tail_data 2.06 ± 8% -60.0% 0.82 ± 19% perf-profile.cycles-pp.__wake_up_sync_key.sctp_data_ready.sctp_ulpq_tail_event.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24 1.22 ± 2% -100.0% 0.00 ± -1% perf-profile.cycles-pp._cond_resched.__release_sock.release_sock.sctp_sendmsg.inet_sendmsg 1.09 ± 5% -25.9% 0.81 ± 14% perf-profile.cycles-pp._raw_spin_lock_irqsave.get_page_from_freelist.__alloc_pages_nodemask.alloc_kmem_pages_node.kmalloc_large_node 7.25 ± 3% -16.7% 6.04 ± 2% perf-profile.cycles-pp._sctp_make_chunk.sctp_make_datafrag_empty.sctp_datamsg_from_user.sctp_sendmsg.inet_sendmsg 5.92 ± 6% -33.6% 3.92 ± 6% perf-profile.cycles-pp.alloc_kmem_pages_node.kmalloc_large_node.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb 1.91 ± 4% -63.1% 0.70 ± 11% perf-profile.cycles-pp.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.sctp_data_ready.sctp_ulpq_tail_event 1.33 ± 4% -27.2% 0.97 ± 3% perf-profile.cycles-pp.copy_msghdr_from_user.___sys_sendmsg.__sys_sendmsg.sys_sendmsg.entry_SYSCALL_64_fastpath 10.64 ± 2% +16.7% 12.43 ± 8% perf-profile.cycles-pp.copy_user_enhanced_fast_string.sctp_user_addto_chunk.sctp_datamsg_from_user.sctp_sendmsg.inet_sendmsg 9.43 ± 1% +22.1% 11.51 ± 5% perf-profile.cycles-pp.copy_user_enhanced_fast_string.skb_copy_datagram_iter.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 1.87 ± 3% -62.6% 0.70 ± 11% perf-profile.cycles-pp.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.sctp_data_ready 0.71 ± 33% -100.0% 0.00 ± -1% perf-profile.cycles-pp.dev_queue_xmit.ip_finish_output2.ip_finish_output.ip_output.ip_local_out 12.40 ± 1% +41.8% 17.59 ± 2% perf-profile.cycles-pp.do_softirq.part.13.__local_bh_enable_ip.ip_finish_output2.ip_finish_output.ip_output 4.52 ± 3% -28.1% 3.25 ± 3% perf-profile.cycles-pp.do_softirq.part.13.__local_bh_enable_ip.sctp_inq_push.sctp_backlog_rcv.__release_sock 12.35 ± 1% +42.1% 17.55 ± 2% perf-profile.cycles-pp.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip.ip_finish_output2.ip_finish_output 4.48 ± 3% -27.9% 3.23 ± 3% perf-profile.cycles-pp.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip.sctp_inq_push.sctp_backlog_rcv 1.71 ± 14% -38.5% 1.05 ± 7% perf-profile.cycles-pp.free_one_page.__free_pages_ok.__free_pages.__free_kmem_pages.kfree 5.15 ± 5% -42.8% 2.95 ± 2% perf-profile.cycles-pp.get_page_from_freelist.__alloc_pages_nodemask.alloc_kmem_pages_node.kmalloc_large_node.__kmalloc_node_track_caller 73.23 ± 0% -16.2% 61.40 ± 1% perf-profile.cycles-pp.inet_sendmsg.sock_sendmsg.___sys_sendmsg.__sys_sendmsg.sys_sendmsg 13.52 ± 3% +33.2% 18.01 ± 2% perf-profile.cycles-pp.ip_finish_output.ip_output.ip_local_out.ip_queue_xmit.sctp_v4_xmit 13.06 ± 3% +37.7% 17.98 ± 2% perf-profile.cycles-pp.ip_finish_output2.ip_finish_output.ip_output.ip_local_out.ip_queue_xmit 15.14 ± 2% +27.0% 19.24 ± 1% perf-profile.cycles-pp.ip_local_deliver.ip_rcv_finish.ip_rcv.__netif_receive_skb_core.__netif_receive_skb 15.04 ± 2% +27.2% 19.13 ± 1% perf-profile.cycles-pp.ip_local_deliver_finish.ip_local_deliver.ip_rcv_finish.ip_rcv.__netif_receive_skb_core 14.75 ± 1% +28.2% 18.91 ± 1% perf-profile.cycles-pp.ip_local_out.ip_queue_xmit.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush 14.53 ± 1% +28.6% 18.68 ± 2% perf-profile.cycles-pp.ip_output.ip_local_out.ip_queue_xmit.sctp_v4_xmit.sctp_packet_transmit 0.84 ± 16% +1251.5% 11.42 ± 3% perf-profile.cycles-pp.ip_queue_xmit.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush.sctp_outq_tail 14.24 ± 1% -44.3% 7.93 ± 4% perf-profile.cycles-pp.ip_queue_xmit.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork 15.56 ± 2% +25.8% 19.58 ± 2% perf-profile.cycles-pp.ip_rcv.__netif_receive_skb_core.__netif_receive_skb.process_backlog.net_rx_action 15.27 ± 2% +26.8% 19.36 ± 2% perf-profile.cycles-pp.ip_rcv_finish.ip_rcv.__netif_receive_skb_core.__netif_receive_skb.process_backlog 3.06 ± 4% -22.9% 2.36 ± 6% perf-profile.cycles-pp.kfree.skb_free_head.skb_release_data.skb_release_all.consume_skb 1.15 ± 3% -17.9% 0.94 ± 7% perf-profile.cycles-pp.kfree.skb_free_head.skb_release_data.skb_release_all.kfree_skb 4.43 ± 7% -20.9% 3.50 ± 2% perf-profile.cycles-pp.kmalloc_large_node.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb._sctp_make_chunk 1.56 ± 4% -70.6% 0.46 ± 59% perf-profile.cycles-pp.kmalloc_large_node.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb.sctp_packet_transmit 0.00 ± -1% +Inf% 6.27 ± 4% perf-profile.cycles-pp.memcpy_erms.sctp_packet_transmit_chunk.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24 16.47 ± 2% +24.3% 20.48 ± 1% perf-profile.cycles-pp.net_rx_action.__do_softirq.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip 1.20 ± 1% -100.0% 0.00 ± -1% perf-profile.cycles-pp.preempt_schedule_common._cond_resched.__release_sock.release_sock.sctp_sendmsg 16.20 ± 2% +24.5% 20.17 ± 2% perf-profile.cycles-pp.process_backlog.net_rx_action.__do_softirq.do_softirq_own_stack.do_softirq.part.13 0.00 ± -1% +Inf% 1.69 ± 5% perf-profile.cycles-pp.release_sock.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg 24.70 ± 3% -16.7% 20.59 ± 1% perf-profile.cycles-pp.release_sock.sctp_sendmsg.inet_sendmsg.sock_sendmsg.___sys_sendmsg 1.70 ± 5% -49.9% 0.85 ± 5% perf-profile.cycles-pp.schedule.schedule_timeout.sctp_skb_recv_datagram.sctp_recvmsg.sock_common_recvmsg 0.00 ± -1% +Inf% 0.97 ± 2% perf-profile.cycles-pp.schedule_timeout.sctp_sendmsg.inet_sendmsg.sock_sendmsg.___sys_sendmsg 1.75 ± 6% -50.6% 0.86 ± 5% perf-profile.cycles-pp.schedule_timeout.sctp_skb_recv_datagram.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 11.53 ± 3% +40.3% 16.17 ± 1% perf-profile.cycles-pp.sctp_assoc_bh_rcv.sctp_inq_push.sctp_rcv.ip_local_deliver_finish.ip_local_deliver 1.24 ± 12% +931.0% 12.81 ± 3% perf-profile.cycles-pp.sctp_assoc_rwnd_increase.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 0.00 ± -1% +Inf% 1.52 ± 5% perf-profile.cycles-pp.sctp_backlog_rcv.__release_sock.release_sock.sctp_recvmsg.sock_common_recvmsg 23.12 ± 3% -12.2% 20.30 ± 1% perf-profile.cycles-pp.sctp_backlog_rcv.__release_sock.release_sock.sctp_sendmsg.inet_sendmsg 9.11 ± 5% +59.4% 14.53 ± 2% perf-profile.cycles-pp.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push.sctp_rcv 16.23 ± 0% -30.0% 11.36 ± 2% perf-profile.cycles-pp.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_primitive_SEND.sctp_sendmsg.inet_sendmsg 2.22 ± 7% -59.5% 0.90 ± 18% perf-profile.cycles-pp.sctp_data_ready.sctp_ulpq_tail_event.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24.sctp_do_sm 10.26 ± 4% +48.9% 15.28 ± 2% perf-profile.cycles-pp.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push.sctp_rcv.ip_local_deliver_finish 15.93 ± 0% -28.8% 11.35 ± 2% perf-profile.cycles-pp.sctp_do_sm.sctp_primitive_SEND.sctp_sendmsg.inet_sendmsg.sock_sendmsg 0.00 ± -1% +Inf% 1.52 ± 4% perf-profile.cycles-pp.sctp_inq_push.sctp_backlog_rcv.__release_sock.release_sock.sctp_recvmsg 20.12 ± 2% -9.7% 18.17 ± 2% perf-profile.cycles-pp.sctp_inq_push.sctp_backlog_rcv.__release_sock.release_sock.sctp_sendmsg 11.75 ± 3% +39.3% 16.37 ± 1% perf-profile.cycles-pp.sctp_inq_push.sctp_rcv.ip_local_deliver_finish.ip_local_deliver.ip_rcv_finish 8.20 ± 5% -17.9% 6.74 ± 2% perf-profile.cycles-pp.sctp_make_datafrag_empty.sctp_datamsg_from_user.sctp_sendmsg.inet_sendmsg.sock_sendmsg 0.14 ±173% +2767.9% 4.02 ± 3% perf-profile.cycles-pp.sctp_outq_flush.sctp_outq_tail.sctp_assoc_rwnd_increase.sctp_ulpevent_free.sctp_recvmsg 12.67 ± 3% +57.5% 19.95 ± 2% perf-profile.cycles-pp.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv 19.85 ± 1% -35.5% 12.79 ± 2% perf-profile.cycles-pp.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_primitive_SEND 0.95 ± 14% +1192.3% 12.21 ± 3% perf-profile.cycles-pp.sctp_outq_tail.sctp_assoc_rwnd_increase.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg 12.98 ± 3% +55.0% 20.12 ± 2% perf-profile.cycles-pp.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push 20.10 ± 1% -35.6% 12.95 ± 2% perf-profile.cycles-pp.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_primitive_SEND.sctp_sendmsg 0.82 ± 15% +1240.4% 10.96 ± 3% perf-profile.cycles-pp.sctp_packet_transmit.sctp_outq_flush.sctp_outq_tail.sctp_assoc_rwnd_increase.sctp_ulpevent_free 17.19 ± 1% -41.0% 10.14 ± 2% perf-profile.cycles-pp.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm 0.00 ± -1% +Inf% 1.07 ± 5% perf-profile.cycles-pp.sctp_packet_transmit.sctp_packet_transmit_chunk.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24 0.78 ± 26% +955.1% 8.23 ± 4% perf-profile.cycles-pp.sctp_packet_transmit_chunk.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm 23.37 ± 0% -34.9% 15.23 ± 2% perf-profile.cycles-pp.sctp_primitive_SEND.sctp_sendmsg.inet_sendmsg.sock_sendmsg.___sys_sendmsg 14.40 ± 2% +28.6% 18.51 ± 1% perf-profile.cycles-pp.sctp_rcv.ip_local_deliver_finish.ip_local_deliver.ip_rcv_finish.ip_rcv 17.04 ± 1% +80.3% 30.72 ± 3% perf-profile.cycles-pp.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg.__sys_recvmsg 72.76 ± 0% -16.1% 61.03 ± 1% perf-profile.cycles-pp.sctp_sendmsg.inet_sendmsg.sock_sendmsg.___sys_sendmsg.__sys_sendmsg 2.27 ± 5% -49.9% 1.14 ± 8% perf-profile.cycles-pp.sctp_skb_recv_datagram.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg 3.39 ± 5% +337.7% 14.85 ± 2% perf-profile.cycles-pp.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg 1.77 ± 3% -42.5% 1.02 ± 25% perf-profile.cycles-pp.sctp_ulpevent_make_rcvmsg.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv 4.75 ± 4% -33.4% 3.16 ± 5% perf-profile.cycles-pp.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push 2.34 ± 7% -58.3% 0.98 ± 15% perf-profile.cycles-pp.sctp_ulpq_tail_event.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv 11.39 ± 2% +15.6% 13.17 ± 8% perf-profile.cycles-pp.sctp_user_addto_chunk.sctp_datamsg_from_user.sctp_sendmsg.inet_sendmsg.sock_sendmsg 0.84 ± 16% +1251.2% 11.42 ± 3% perf-profile.cycles-pp.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush.sctp_outq_tail.sctp_assoc_rwnd_increase 14.31 ± 1% -43.8% 8.04 ± 4% perf-profile.cycles-pp.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24 0.75 ± 3% +107.7% 1.56 ± 4% perf-profile.cycles-pp.sctp_wfree.skb_release_head_state.skb_release_all.consume_skb.sctp_chunk_put 9.90 ± 2% +21.4% 12.02 ± 6% perf-profile.cycles-pp.skb_copy_datagram_iter.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg 3.08 ± 4% -22.7% 2.38 ± 5% perf-profile.cycles-pp.skb_free_head.skb_release_data.skb_release_all.consume_skb.sctp_chunk_put 1.15 ± 3% -18.2% 0.95 ± 7% perf-profile.cycles-pp.skb_free_head.skb_release_data.skb_release_all.kfree_skb.sctp_ulpevent_free 4.11 ± 4% +6.9% 4.39 ± 3% perf-profile.cycles-pp.skb_release_all.consume_skb.sctp_chunk_put.sctp_chunk_free.sctp_outq_sack 1.70 ± 3% -13.4% 1.47 ± 5% perf-profile.cycles-pp.skb_release_all.kfree_skb.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg 3.15 ± 4% -23.0% 2.42 ± 5% perf-profile.cycles-pp.skb_release_data.skb_release_all.consume_skb.sctp_chunk_put.sctp_chunk_free 1.33 ± 3% -13.4% 1.15 ± 7% perf-profile.cycles-pp.skb_release_data.skb_release_all.kfree_skb.sctp_ulpevent_free.sctp_recvmsg 0.91 ± 6% +92.3% 1.76 ± 3% perf-profile.cycles-pp.skb_release_head_state.skb_release_all.consume_skb.sctp_chunk_put.sctp_chunk_free 17.25 ± 1% +79.2% 30.92 ± 3% perf-profile.cycles-pp.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg.__sys_recvmsg.sys_recvmsg 17.81 ± 1% +76.0% 31.35 ± 3% perf-profile.cycles-pp.sock_recvmsg.___sys_recvmsg.__sys_recvmsg.sys_recvmsg.entry_SYSCALL_64_fastpath 73.92 ± 0% -16.2% 61.91 ± 1% perf-profile.cycles-pp.sock_sendmsg.___sys_sendmsg.__sys_sendmsg.sys_sendmsg.entry_SYSCALL_64_fastpath 19.54 ± 1% +67.8% 32.79 ± 3% perf-profile.cycles-pp.sys_recvmsg.entry_SYSCALL_64_fastpath 78.34 ± 0% -16.7% 65.25 ± 1% perf-profile.cycles-pp.sys_sendmsg.entry_SYSCALL_64_fastpath 1.82 ± 2% -62.4% 0.68 ± 11% perf-profile.cycles-pp.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key 1.15 ± 5% -100.0% 0.00 ± -1% perf-profile.cycles-pp.ttwu_do_activate.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common netperf.Throughput_Mbps 5400 ++-------------------------------------------------------------------+ **. **.***.**. * .** .***.***.***.**.** .***.***.***. **. *. **.** .** 5200 ++ * * * * * * * * * | 5000 ++ | | | 4800 ++ | | | 4600 ++ | | | 4400 ++ | 4200 ++ | | O O | 4000 ++ O O OO O O O O OO | OO O O O O O O O | 3800 ++-------------------------------------------------------------------+ [*] bisect-good sample [O] bisect-bad sample *************************************************************************************************** ivb42: 48 threads 2 sockets Xeon E5 (Ivytown Ivy Bridge-EP) with 64G memory ========================================================================================= compiler/cpufreq_governor/duration/iterations/kconfig/nr_threads/rootfs/tbox_group/testcase: gcc-4.9/performance/10s/100x/x86_64-rhel/200%/debian-x86_64-2015-02-07.cgz/ivb42/ebizzy commit: 0329dacce8 ("sched/fair: Fix effective_load()") 0cef819730 ("sched/fair: Fix calc_cfs_shares()") 0329dacce87d6741 0cef819730f7f67f8d2bec642e ---------------- -------------------------- fail:runs %reproduction fail:runs | | | :4 50% 2:4 kmsg.Spurious_LAPIC_timer_interrupt_on_cpu %stddev %change %stddev \ | \ 1359 ± 0% -39.6% 821.75 ± 0% ebizzy.throughput.per_thread.max 190.50 ± 1% +99.7% 380.50 ± 0% ebizzy.throughput.per_thread.min 3227377 ± 0% +18.4% 3822284 ± 0% ebizzy.time.involuntary_context_switches 38.30 ± 0% -1.0% 37.93 ± 0% ebizzy.time.user 3836 ± 0% -1.0% 3798 ± 0% ebizzy.time.user_time 7422 ± 0% +1.1% 7500 ± 0% vmstat.system.cs 2747604 ± 2% +23.1% 3381516 ± 1% softirqs.RCU 257256 ± 5% +22.0% 313808 ± 4% softirqs.SCHED 8941 ± 19% +40.0% 12516 ± 10% numa-vmstat.node0.nr_active_anon 2283 ± 68% +153.0% 5776 ± 23% numa-vmstat.node0.nr_shmem 4031 ± 27% -69.1% 1243 ± 81% numa-vmstat.node1.nr_shmem 43859 ± 15% +32.5% 58111 ± 8% numa-meminfo.node0.Active 35763 ± 19% +40.0% 50075 ± 10% numa-meminfo.node0.Active(anon) 9134 ± 68% +153.0% 23107 ± 23% numa-meminfo.node0.Shmem 16120 ± 27% -69.1% 4977 ± 81% numa-meminfo.node1.Shmem 7864 ± 3% +13.9% 8959 ± 3% slabinfo.anon_vma.active_objs 7864 ± 3% +13.9% 8959 ± 3% slabinfo.anon_vma.num_objs 20622 ± 1% +21.6% 25083 ± 3% slabinfo.anon_vma_chain.active_objs 20650 ± 1% +21.6% 25121 ± 3% slabinfo.anon_vma_chain.num_objs 16762 ± 2% +22.0% 20454 ± 4% slabinfo.vm_area_struct.active_objs 16800 ± 2% +22.0% 20495 ± 4% slabinfo.vm_area_struct.num_objs 1.766e+11 ± 0% +0.7% 1.778e+11 ± 0% perf-stat.L1-dcache-prefetch-misses 5.765e+11 ± 0% -0.8% 5.721e+11 ± 0% perf-stat.L1-dcache-store-misses 3.326e+10 ± 0% +11.6% 3.71e+10 ± 0% perf-stat.L1-icache-load-misses 7440021 ± 0% +1.1% 7518907 ± 0% perf-stat.context-switches 131030 ± 0% +120.8% 289258 ± 0% perf-stat.cpu-migrations 9.909e+09 ± 0% +4.4% 1.035e+10 ± 0% perf-stat.node-store-misses 1.466e+10 ± 2% +5.3% 1.543e+10 ± 2% perf-stat.node-stores 945986 ± 0% -97.8% 21259 ± 0% sched_debug.cfs_rq:/.load.avg 1047060 ± 2% -96.0% 41384 ± 14% sched_debug.cfs_rq:/.load.max 902269 ± 0% -98.9% 10100 ± 0% sched_debug.cfs_rq:/.load.min 51815 ± 6% -86.2% 7145 ± 9% sched_debug.cfs_rq:/.load.stddev 905.05 ± 0% -97.4% 23.63 ± 10% sched_debug.cfs_rq:/.load_avg.avg 1011 ± 2% -90.7% 94.35 ± 18% sched_debug.cfs_rq:/.load_avg.max 863.59 ± 0% -98.8% 10.38 ± 0% sched_debug.cfs_rq:/.load_avg.min 37.56 ± 9% -52.1% 17.99 ± 22% sched_debug.cfs_rq:/.load_avg.stddev 508232 ± 0% +5202.4% 26948505 ± 0% sched_debug.cfs_rq:/.min_vruntime.avg 518809 ± 0% +5251.8% 27765584 ± 0% sched_debug.cfs_rq:/.min_vruntime.max 502866 ± 0% +5087.3% 26085013 ± 0% sched_debug.cfs_rq:/.min_vruntime.min 2704 ± 4% +17547.4% 477333 ± 17% sched_debug.cfs_rq:/.min_vruntime.stddev 0.03 ± 19% +3208.3% 1.10 ± 6% sched_debug.cfs_rq:/.nr_spread_over.avg 0.84 ± 21% +2882.5% 25.00 ± 14% sched_debug.cfs_rq:/.nr_spread_over.max 0.16 ± 20% +2354.4% 4.01 ± 13% sched_debug.cfs_rq:/.nr_spread_over.stddev 900.16 ± 0% -97.9% 19.11 ± 0% sched_debug.cfs_rq:/.runnable_load_avg.avg 950.12 ± 0% -96.9% 29.13 ± 2% sched_debug.cfs_rq:/.runnable_load_avg.max 863.34 ± 0% -98.8% 10.18 ± 1% sched_debug.cfs_rq:/.runnable_load_avg.min 26.66 ± 0% -83.8% 4.33 ± 2% sched_debug.cfs_rq:/.runnable_load_avg.stddev 3007 ±164% +19711.6% 595751 ± 38% sched_debug.cfs_rq:/.spread0.max -12941 ±-36% +8281.9% -1084720 ±-40% sched_debug.cfs_rq:/.spread0.min 2703 ± 4% +17557.3% 477353 ± 17% sched_debug.cfs_rq:/.spread0.stddev 271484 ± 3% +14.0% 309560 ± 3% sched_debug.cpu.avg_idle.avg 840053 ± 6% +15.5% 970603 ± 2% sched_debug.cpu.avg_idle.max 122164 ± 10% +29.8% 158613 ± 10% sched_debug.cpu.avg_idle.stddev 900.06 ± 0% -97.9% 19.10 ± 0% sched_debug.cpu.cpu_load[0].avg 950.15 ± 0% -96.9% 29.18 ± 2% sched_debug.cpu.cpu_load[0].max 863.35 ± 0% -98.8% 10.19 ± 1% sched_debug.cpu.cpu_load[0].min 26.60 ± 0% -83.6% 4.35 ± 2% sched_debug.cpu.cpu_load[0].stddev 900.19 ± 0% -97.8% 19.47 ± 1% sched_debug.cpu.cpu_load[1].avg 947.69 ± 0% -95.4% 43.91 ± 31% sched_debug.cpu.cpu_load[1].max 863.35 ± 0% -98.8% 10.19 ± 1% sched_debug.cpu.cpu_load[1].min 26.23 ± 1% -75.8% 6.35 ± 31% sched_debug.cpu.cpu_load[1].stddev 899.05 ± 0% -97.8% 19.34 ± 1% sched_debug.cpu.cpu_load[2].avg 944.72 ± 0% -96.1% 36.68 ± 16% sched_debug.cpu.cpu_load[2].max 863.35 ± 0% -98.8% 10.19 ± 1% sched_debug.cpu.cpu_load[2].min 25.72 ± 1% -79.2% 5.34 ± 17% sched_debug.cpu.cpu_load[2].stddev 897.66 ± 0% -97.9% 19.27 ± 1% sched_debug.cpu.cpu_load[3].avg 941.85 ± 0% -96.5% 32.90 ± 7% sched_debug.cpu.cpu_load[3].max 863.32 ± 0% -98.8% 10.19 ± 1% sched_debug.cpu.cpu_load[3].min 25.12 ± 1% -80.5% 4.90 ± 11% sched_debug.cpu.cpu_load[3].stddev 896.24 ± 0% -97.9% 19.19 ± 0% sched_debug.cpu.cpu_load[4].avg 938.69 ± 0% -96.7% 31.38 ± 7% sched_debug.cpu.cpu_load[4].max 863.28 ± 0% -98.8% 10.19 ± 1% sched_debug.cpu.cpu_load[4].min 24.61 ± 1% -81.3% 4.60 ± 9% sched_debug.cpu.cpu_load[4].stddev 16425 ± 4% -23.7% 12532 ± 25% sched_debug.cpu.curr->pid.min 946912 ± 0% -97.8% 21258 ± 0% sched_debug.cpu.load.avg 1047060 ± 2% -96.0% 41802 ± 13% sched_debug.cpu.load.max 902268 ± 0% -98.9% 10100 ± 0% sched_debug.cpu.load.min 51966 ± 5% -86.2% 7157 ± 8% sched_debug.cpu.load.stddev 5.74 ± 6% -44.1% 3.21 ± 4% sched_debug.cpu.nr_running.max 1.13 ± 4% -50.2% 0.56 ± 2% sched_debug.cpu.nr_running.stddev 567072 ± 20% -73.0% 153117 ± 4% sched_debug.cpu.nr_switches.max 33254 ± 3% +31.9% 43856 ± 1% sched_debug.cpu.nr_switches.min 101510 ± 16% -73.4% 26976 ± 6% sched_debug.cpu.nr_switches.stddev -93.43 ± -8% +30.1% -121.59 ±-13% sched_debug.cpu.nr_uninterruptible.min 566139 ± 20% -70.4% 167497 ± 9% sched_debug.cpu.sched_count.max 33050 ± 3% +39.0% 45943 ± 0% sched_debug.cpu.sched_count.min 101830 ± 16% -71.2% 29289 ± 4% sched_debug.cpu.sched_count.stddev 1857 ± 4% -11.3% 1647 ± 4% sched_debug.cpu.sched_goidle.avg 273980 ± 22% -78.3% 59590 ± 6% sched_debug.cpu.ttwu_count.max 4600 ± 6% +21.4% 5586 ± 12% sched_debug.cpu.ttwu_count.min 51498 ± 16% -74.5% 13152 ± 5% sched_debug.cpu.ttwu_count.stddev 20747 ± 0% -17.9% 17027 ± 2% sched_debug.cpu.ttwu_local.avg 265675 ± 22% -80.0% 53126 ± 5% sched_debug.cpu.ttwu_local.max 50152 ± 16% -75.3% 12409 ± 4% sched_debug.cpu.ttwu_local.stddev *************************************************************************************************** lkp-hsx04: 144 threads Brickland Haswell-EX with 512G memory ========================================================================================= compiler/cpufreq_governor/kconfig/nr_threads/rootfs/tbox_group/testcase: gcc-4.9/performance/x86_64-rhel/100%/debian-x86_64-2015-02-07.cgz/lkp-hsx04/plzip commit: 0329dacce8 ("sched/fair: Fix effective_load()") 0cef819730 ("sched/fair: Fix calc_cfs_shares()") 0329dacce87d6741 0cef819730f7f67f8d2bec642e ---------------- -------------------------- fail:runs %reproduction fail:runs | | | :4 75% 3:4 kmsg.DHCP/BOOTP:Reply_not_for_us,op[#]xid[#] :4 25% 1:4 kmsg.DHCP/BOOTP:Reply_not_for_us,op[#]xid[#b98713] %stddev %change %stddev \ | \ 2.198e+08 ± 0% +2.5% 2.253e+08 ± 0% plzip.throughput 13115 ± 0% +2.3% 13418 ± 0% plzip.time.percent_of_cpu_this_job_got 46198 ± 0% +3.2% 47659 ± 0% plzip.time.user_time 501228 ± 1% -17.2% 415017 ± 2% plzip.time.voluntary_context_switches 30291 ± 3% +8.6% 32896 ± 4% slabinfo.vm_area_struct.num_objs 62366 ± 82% -57.8% 26290 ± 10% latency_stats.max.call_rwsem_down_write_failed_killable.SyS_munmap.entry_SYSCALL_64_fastpath 2325979 ± 6% -69.5% 709462 ± 16% latency_stats.sum.call_rwsem_down_read_failed.SyS_madvise.entry_SYSCALL_64_fastpath 6995 ± 1% -8.0% 6437 ± 3% vmstat.system.cs 449401 ± 1% -3.3% 434526 ± 1% vmstat.system.in 13566509 ± 1% -10.5% 12147331 ± 3% proc-vmstat.numa_pages_migrated 307456 ± 4% -13.9% 264576 ± 8% proc-vmstat.pgmigrate_fail 13566509 ± 1% -10.5% 12147331 ± 3% proc-vmstat.pgmigrate_success 43940096 ± 3% -16.1% 36883878 ± 4% cpuidle.C3-HSW.time 138953 ± 3% -16.8% 115586 ± 4% cpuidle.C3-HSW.usage 4.377e+09 ± 2% -22.6% 3.39e+09 ± 4% cpuidle.C6-HSW.time 4794982 ± 2% -23.0% 3693173 ± 4% cpuidle.C6-HSW.usage 14563665 ± 36% -64.8% 5126454 ± 47% numa-numastat.node1.local_node 14563667 ± 36% -64.8% 5126456 ± 47% numa-numastat.node1.numa_hit 3979068 ± 16% +286.6% 15382273 ± 13% numa-numastat.node2.local_node 3979070 ± 16% +286.6% 15382276 ± 13% numa-numastat.node2.numa_hit 91.16 ± 0% +2.2% 93.20 ± 0% turbostat.%Busy 2636 ± 0% +2.2% 2695 ± 0% turbostat.Avg_MHz 6.24 ± 2% -29.4% 4.40 ± 3% turbostat.CPU%c1 2.59 ± 4% -7.9% 2.38 ± 5% turbostat.CPU%c6 694.45 ± 0% +1.1% 702.04 ± 0% turbostat.RAMWatt 8779 ± 8% -19.6% 7057 ± 4% numa-meminfo.node1.PageTables 6706 ± 50% -72.4% 1848 ±115% numa-meminfo.node1.Shmem 2617044 ± 3% +12.0% 2930380 ± 3% numa-meminfo.node2.Active 2613057 ± 3% +12.0% 2926476 ± 3% numa-meminfo.node2.Active(anon) 2703252 ± 3% +10.9% 2999010 ± 3% numa-meminfo.node2.AnonHugePages 2887787 ± 3% +11.6% 3223759 ± 3% numa-meminfo.node2.AnonPages 4665 ± 6% +92.5% 8981 ± 28% numa-meminfo.node2.Mapped 3369566 ± 2% +10.2% 3711830 ± 3% numa-meminfo.node2.MemUsed 2193 ± 8% -19.5% 1766 ± 5% numa-vmstat.node1.nr_page_table_pages 1676 ± 50% -72.5% 461.75 ±115% numa-vmstat.node1.nr_shmem 7453615 ± 26% -60.9% 2912511 ± 37% numa-vmstat.node1.numa_hit 7453614 ± 26% -60.9% 2912510 ± 37% numa-vmstat.node1.numa_local 653170 ± 3% +12.0% 731594 ± 3% numa-vmstat.node2.nr_active_anon 721847 ± 3% +11.7% 806051 ± 3% numa-vmstat.node2.nr_anon_pages 1317 ± 3% +11.1% 1464 ± 3% numa-vmstat.node2.nr_anon_transparent_hugepages 1158 ± 6% +93.9% 2245 ± 28% numa-vmstat.node2.nr_mapped 2185024 ± 31% +247.1% 7583311 ± 21% numa-vmstat.node2.numa_hit 2185023 ± 31% +247.1% 7583309 ± 21% numa-vmstat.node2.numa_local 5.836e+11 ± 0% +3.0% 6.01e+11 ± 0% perf-stat.L1-dcache-load-misses 1.507e+13 ± 0% +2.9% 1.55e+13 ± 0% perf-stat.L1-dcache-loads 3.93e+12 ± 0% +3.8% 4.078e+12 ± 0% perf-stat.L1-dcache-stores 3.031e+11 ± 0% +3.7% 3.143e+11 ± 0% perf-stat.LLC-load-misses 3.848e+11 ± 0% +3.2% 3.97e+11 ± 0% perf-stat.LLC-loads 3.665e+10 ± 0% +4.6% 3.834e+10 ± 0% perf-stat.LLC-store-misses 1.022e+11 ± 0% +4.1% 1.063e+11 ± 0% perf-stat.LLC-stores 1.112e+13 ± 0% +3.6% 1.151e+13 ± 1% perf-stat.branch-instructions 2.394e+11 ± 0% +4.0% 2.489e+11 ± 0% perf-stat.branch-load-misses 1.113e+13 ± 0% +2.9% 1.146e+13 ± 1% perf-stat.branch-loads 2.406e+11 ± 1% +3.9% 2.499e+11 ± 0% perf-stat.branch-misses 4.606e+12 ± 0% +3.2% 4.753e+12 ± 0% perf-stat.bus-cycles 3.393e+11 ± 0% +4.0% 3.529e+11 ± 0% perf-stat.cache-misses 4.964e+11 ± 0% +3.1% 5.116e+11 ± 0% perf-stat.cache-references 2488007 ± 1% -7.3% 2306257 ± 3% perf-stat.context-switches 1.332e+14 ± 0% +3.1% 1.374e+14 ± 0% perf-stat.cpu-cycles 152269 ± 1% -32.4% 102949 ± 5% perf-stat.cpu-migrations 1.505e+13 ± 0% +3.1% 1.552e+13 ± 0% perf-stat.dTLB-loads 3.94e+12 ± 0% +3.4% 4.072e+12 ± 0% perf-stat.dTLB-stores 1.754e+08 ± 21% -28.4% 1.256e+08 ± 11% perf-stat.iTLB-load-misses 5.696e+13 ± 0% +3.6% 5.899e+13 ± 0% perf-stat.instructions 4.732e+10 ± 2% -11.6% 4.184e+10 ± 4% perf-stat.node-load-misses 2.552e+11 ± 1% +6.7% 2.723e+11 ± 0% perf-stat.node-loads 7.049e+09 ± 2% -8.1% 6.476e+09 ± 4% perf-stat.node-store-misses 2.97e+10 ± 0% +7.3% 3.186e+10 ± 1% perf-stat.node-stores 1.149e+14 ± 0% +3.0% 1.184e+14 ± 0% perf-stat.ref-cycles 759797 ± 1% -99.2% 6423 ± 2% sched_debug.cfs_rq:/.load.avg 991391 ± 0% -97.8% 21997 ± 26% sched_debug.cfs_rq:/.load.max 344247 ± 36% -99.5% 1707 ± 31% sched_debug.cfs_rq:/.load.min 114087 ± 29% -97.7% 2587 ± 17% sched_debug.cfs_rq:/.load.stddev 736.09 ± 1% -98.0% 14.97 ± 10% sched_debug.cfs_rq:/.load_avg.avg 929.42 ± 1% -77.0% 213.42 ± 33% sched_debug.cfs_rq:/.load_avg.max 319.50 ± 10% -99.1% 2.78 ± 17% sched_debug.cfs_rq:/.load_avg.min 83.59 ± 10% -57.0% 35.97 ± 4% sched_debug.cfs_rq:/.load_avg.stddev 154838 ± 0% +13451.9% 20983617 ± 8% sched_debug.cfs_rq:/.min_vruntime.avg 177334 ± 1% +11995.3% 21449100 ± 8% sched_debug.cfs_rq:/.min_vruntime.max 147845 ± 0% +13477.1% 20073176 ± 8% sched_debug.cfs_rq:/.min_vruntime.min 3666 ± 10% +7085.9% 263446 ± 11% sched_debug.cfs_rq:/.min_vruntime.stddev 0.03 ± 24% +796.1% 0.24 ± 17% sched_debug.cfs_rq:/.nr_spread_over.avg 0.83 ± 14% +620.0% 6.00 ± 39% sched_debug.cfs_rq:/.nr_spread_over.max 0.14 ± 15% +506.4% 0.83 ± 24% sched_debug.cfs_rq:/.nr_spread_over.stddev 715.46 ± 2% -99.3% 5.26 ± 0% sched_debug.cfs_rq:/.runnable_load_avg.avg 821.04 ± 1% -98.4% 13.11 ± 4% sched_debug.cfs_rq:/.runnable_load_avg.max 216.38 ± 41% -99.4% 1.23 ± 19% sched_debug.cfs_rq:/.runnable_load_avg.min 90.41 ± 30% -98.2% 1.64 ± 22% sched_debug.cfs_rq:/.runnable_load_avg.stddev -22528 ±-14% -836.6% 165943 ± 61% sched_debug.cfs_rq:/.spread0.avg -29534 ±-13% +2423.1% -745164 ±-14% sched_debug.cfs_rq:/.spread0.min 3672 ± 10% +7042.8% 262350 ± 11% sched_debug.cfs_rq:/.spread0.stddev 98890 ± 14% +31.9% 130412 ± 5% sched_debug.cpu.avg_idle.stddev 715.02 ± 2% -99.3% 5.29 ± 0% sched_debug.cpu.cpu_load[0].avg 820.88 ± 1% -98.3% 13.83 ± 6% sched_debug.cpu.cpu_load[0].max 216.67 ± 41% -99.4% 1.23 ± 19% sched_debug.cpu.cpu_load[0].min 91.02 ± 31% -98.1% 1.77 ± 23% sched_debug.cpu.cpu_load[0].stddev 726.15 ± 1% -98.9% 7.77 ± 19% sched_debug.cpu.cpu_load[1].avg 886.12 ± 3% -87.0% 115.02 ± 41% sched_debug.cpu.cpu_load[1].max 261.54 ± 17% -99.0% 2.52 ± 13% sched_debug.cpu.cpu_load[1].min 78.89 ± 17% -81.1% 14.88 ± 51% sched_debug.cpu.cpu_load[1].stddev 723.86 ± 1% -99.0% 7.07 ± 15% sched_debug.cpu.cpu_load[2].avg 843.88 ± 2% -90.4% 81.31 ± 36% sched_debug.cpu.cpu_load[2].max 246.75 ± 19% -99.0% 2.52 ± 14% sched_debug.cpu.cpu_load[2].min 73.42 ± 18% -85.7% 10.52 ± 48% sched_debug.cpu.cpu_load[2].stddev 722.00 ± 1% -99.1% 6.51 ± 10% sched_debug.cpu.cpu_load[3].avg 816.79 ± 1% -93.0% 57.06 ± 28% sched_debug.cpu.cpu_load[3].max 239.75 ± 18% -98.9% 2.60 ± 14% sched_debug.cpu.cpu_load[3].min 69.62 ± 19% -89.5% 7.32 ± 42% sched_debug.cpu.cpu_load[3].stddev 720.26 ± 1% -99.2% 6.05 ± 7% sched_debug.cpu.cpu_load[4].avg 798.08 ± 0% -95.0% 39.58 ± 23% sched_debug.cpu.cpu_load[4].max 246.58 ± 12% -98.9% 2.64 ± 14% sched_debug.cpu.cpu_load[4].min 66.15 ± 18% -92.5% 4.95 ± 40% sched_debug.cpu.cpu_load[4].stddev 760349 ± 1% -99.2% 6410 ± 2% sched_debug.cpu.load.avg 991391 ± 0% -97.8% 21997 ± 26% sched_debug.cpu.load.max 344247 ± 36% -99.5% 1707 ± 31% sched_debug.cpu.load.min 115044 ± 29% -97.8% 2566 ± 18% sched_debug.cpu.load.stddev 4076 ± 3% -35.1% 2647 ± 3% sched_debug.cpu.nr_load_updates.stddev 2636 ± 2% -30.2% 1840 ± 4% sched_debug.cpu.nr_switches.min 81384 ± 23% +30.0% 105830 ± 10% sched_debug.cpu.sched_count.max 2386 ± 3% -32.1% 1621 ± 4% sched_debug.cpu.sched_count.min 1204 ± 3% -22.9% 929.22 ± 7% sched_debug.cpu.sched_goidle.avg 554.29 ± 4% -32.6% 373.48 ± 2% sched_debug.cpu.sched_goidle.min 823.67 ± 2% -17.6% 678.97 ± 6% sched_debug.cpu.ttwu_count.min 394.42 ± 16% -28.3% 282.88 ± 7% sched_debug.cpu.ttwu_local.min 0.72 ± 10% +84.0% 1.32 ± 17% perf-profile.cycles-pp.__default_send_IPI_dest_field.default_send_IPI_mask_sequence_phys.native_send_call_func_ipi.smp_call_function_many.native_flush_tlb_others 20.80 ± 20% +40.0% 29.12 ± 6% perf-profile.cycles-pp.__do_page_fault.do_page_fault.page_fault 14.03 ± 22% +32.2% 18.54 ± 6% perf-profile.cycles-pp.__hrtimer_run_queues.hrtimer_interrupt.local_apic_timer_interrupt.smp_apic_timer_interrupt.apic_timer_interrupt 5.27 ± 7% +52.3% 8.02 ± 3% perf-profile.cycles-pp.__vfs_read.vfs_read.sys_read.entry_SYSCALL_64_fastpath 1.66 ± 13% +63.2% 2.71 ± 15% perf-profile.cycles-pp.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath 0.63 ± 6% +63.3% 1.03 ± 7% perf-profile.cycles-pp.anon_pipe_buf_release.__vfs_read.vfs_read.sys_read.entry_SYSCALL_64_fastpath 21.05 ± 23% +34.5% 28.31 ± 5% perf-profile.cycles-pp.apic_timer_interrupt 40.80 ± 27% -56.5% 17.74 ± 21% perf-profile.cycles-pp.call_cpuidle.cpu_startup_entry.start_secondary 4.43 ± 15% +52.7% 6.77 ± 16% perf-profile.cycles-pp.call_function_interrupt 0.59 ± 63% +87.8% 1.11 ± 5% perf-profile.cycles-pp.clear_huge_page.do_huge_pmd_anonymous_page.handle_mm_fault.__do_page_fault.do_page_fault 12.05 ± 25% +44.6% 17.43 ± 4% perf-profile.cycles-pp.clear_page_c_e.do_huge_pmd_anonymous_page.handle_mm_fault.__do_page_fault.do_page_fault 1.17 ± 11% +67.3% 1.96 ± 10% perf-profile.cycles-pp.copy_page_from_iter.pipe_write.__vfs_write.vfs_write.sys_write 4.22 ± 7% +55.2% 6.54 ± 2% perf-profile.cycles-pp.copy_page_to_iter.pipe_read.__vfs_read.vfs_read.sys_read 0.96 ± 6% +77.1% 1.70 ± 10% perf-profile.cycles-pp.copy_user_enhanced_fast_string.copy_page_from_iter.pipe_write.__vfs_write.vfs_write 3.92 ± 8% +55.8% 6.11 ± 5% perf-profile.cycles-pp.copy_user_enhanced_fast_string.copy_page_to_iter.pipe_read.__vfs_read.vfs_read 41.00 ± 27% -56.5% 17.84 ± 21% perf-profile.cycles-pp.cpu_startup_entry.start_secondary 40.80 ± 27% -56.5% 17.74 ± 22% perf-profile.cycles-pp.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary 40.68 ± 27% -56.5% 17.69 ± 22% perf-profile.cycles-pp.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary 0.85 ± 11% +78.3% 1.52 ± 18% perf-profile.cycles-pp.default_send_IPI_mask_sequence_phys.native_send_call_func_ipi.smp_call_function_many.native_flush_tlb_others.flush_tlb_page 13.47 ± 25% +44.0% 19.40 ± 4% perf-profile.cycles-pp.do_huge_pmd_anonymous_page.handle_mm_fault.__do_page_fault.do_page_fault.page_fault 20.83 ± 20% +40.0% 29.16 ± 6% perf-profile.cycles-pp.do_page_fault.page_fault 7.77 ± 8% +53.7% 11.94 ± 5% perf-profile.cycles-pp.entry_SYSCALL_64_fastpath 3.89 ± 15% +50.0% 5.84 ± 16% perf-profile.cycles-pp.flush_smp_call_function_queue.generic_smp_call_function_single_interrupt.smp_call_function_interrupt.call_function_interrupt 0.92 ± 16% +48.6% 1.36 ± 19% perf-profile.cycles-pp.flush_tlb_func.flush_smp_call_function_queue.generic_smp_call_function_single_interrupt.smp_call_function_interrupt.call_function_interrupt 2.44 ± 8% +60.1% 3.90 ± 15% perf-profile.cycles-pp.flush_tlb_page.ptep_clear_flush.try_to_unmap_one.rmap_walk_anon.rmap_walk 3.91 ± 15% +51.2% 5.92 ± 17% perf-profile.cycles-pp.generic_smp_call_function_single_interrupt.smp_call_function_interrupt.call_function_interrupt 19.86 ± 21% +42.0% 28.20 ± 6% perf-profile.cycles-pp.handle_mm_fault.__do_page_fault.do_page_fault.page_fault 4.39 ± 11% +50.3% 6.59 ± 13% perf-profile.cycles-pp.handle_pte_fault.handle_mm_fault.__do_page_fault.do_page_fault.page_fault 17.51 ± 24% +33.6% 23.40 ± 7% perf-profile.cycles-pp.hrtimer_interrupt.local_apic_timer_interrupt.smp_apic_timer_interrupt.apic_timer_interrupt 40.57 ± 27% -56.1% 17.80 ± 21% perf-profile.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry 2.38 ± 23% +31.6% 3.14 ± 13% perf-profile.cycles-pp.irq_exit.smp_apic_timer_interrupt.apic_timer_interrupt 0.64 ± 7% +42.4% 0.91 ± 22% perf-profile.cycles-pp.llist_add_batch.smp_call_function_many.native_flush_tlb_others.flush_tlb_page.ptep_clear_flush 1.44 ± 13% +50.0% 2.16 ± 15% perf-profile.cycles-pp.llist_reverse_order.flush_smp_call_function_queue.generic_smp_call_function_single_interrupt.smp_call_function_interrupt.call_function_interrupt 17.83 ± 24% +33.7% 23.83 ± 7% perf-profile.cycles-pp.local_apic_timer_interrupt.smp_apic_timer_interrupt.apic_timer_interrupt 3.36 ± 9% +53.2% 5.15 ± 15% perf-profile.cycles-pp.migrate_misplaced_page.handle_pte_fault.handle_mm_fault.__do_page_fault.do_page_fault 2.74 ± 8% +57.5% 4.32 ± 15% perf-profile.cycles-pp.migrate_pages.migrate_misplaced_page.handle_pte_fault.handle_mm_fault.__do_page_fault 2.43 ± 8% +59.9% 3.88 ± 16% perf-profile.cycles-pp.native_flush_tlb_others.flush_tlb_page.ptep_clear_flush.try_to_unmap_one.rmap_walk_anon 1.67 ± 14% +32.3% 2.21 ± 13% perf-profile.cycles-pp.native_irq_return_iret 1.23 ± 8% +67.0% 2.06 ± 17% perf-profile.cycles-pp.native_send_call_func_ipi.smp_call_function_many.native_flush_tlb_others.flush_tlb_page.ptep_clear_flush 20.83 ± 20% +40.0% 29.16 ± 6% perf-profile.cycles-pp.page_fault 1.38 ± 22% +46.3% 2.02 ± 10% perf-profile.cycles-pp.perf_mux_hrtimer_handler.__hrtimer_run_queues.hrtimer_interrupt.local_apic_timer_interrupt.smp_apic_timer_interrupt 4.62 ± 7% +50.9% 6.98 ± 2% perf-profile.cycles-pp.pipe_read.__vfs_read.vfs_read.sys_read.entry_SYSCALL_64_fastpath 1.65 ± 13% +70.5% 2.81 ± 11% perf-profile.cycles-pp.pipe_write.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath 2.44 ± 8% +60.1% 3.92 ± 16% perf-profile.cycles-pp.ptep_clear_flush.try_to_unmap_one.rmap_walk_anon.rmap_walk.try_to_unmap 1.06 ± 24% +43.8% 1.53 ± 4% perf-profile.cycles-pp.rcu_check_callbacks.update_process_times.tick_sched_handle.isra.17.tick_sched_timer.__hrtimer_run_queues 1.00 ± 19% +38.3% 1.38 ± 26% perf-profile.cycles-pp.rcu_process_callbacks.__do_softirq.irq_exit.smp_apic_timer_interrupt.apic_timer_interrupt 2.52 ± 9% +58.1% 3.98 ± 15% perf-profile.cycles-pp.rmap_walk.try_to_unmap.migrate_pages.migrate_misplaced_page.handle_pte_fault 2.51 ± 8% +58.3% 3.97 ± 15% perf-profile.cycles-pp.rmap_walk_anon.rmap_walk.try_to_unmap.migrate_pages.migrate_misplaced_page 20.80 ± 23% +34.0% 27.87 ± 5% perf-profile.cycles-pp.smp_apic_timer_interrupt.apic_timer_interrupt 4.25 ± 14% +52.1% 6.47 ± 17% perf-profile.cycles-pp.smp_call_function_interrupt.call_function_interrupt 2.41 ± 8% +60.4% 3.86 ± 16% perf-profile.cycles-pp.smp_call_function_many.native_flush_tlb_others.flush_tlb_page.ptep_clear_flush.try_to_unmap_one 41.00 ± 27% -56.5% 17.85 ± 21% perf-profile.cycles-pp.start_secondary 5.27 ± 7% +52.4% 8.02 ± 3% perf-profile.cycles-pp.sys_read.entry_SYSCALL_64_fastpath 1.67 ± 12% +63.9% 2.74 ± 16% perf-profile.cycles-pp.sys_write.entry_SYSCALL_64_fastpath 1.20 ± 31% +40.8% 1.68 ± 8% perf-profile.cycles-pp.tick_program_event.hrtimer_interrupt.local_apic_timer_interrupt.smp_apic_timer_interrupt.apic_timer_interrupt 10.67 ± 23% +27.5% 13.61 ± 6% perf-profile.cycles-pp.tick_sched_handle.isra.17.tick_sched_timer.__hrtimer_run_queues.hrtimer_interrupt.local_apic_timer_interrupt 11.11 ± 23% +29.1% 14.35 ± 6% perf-profile.cycles-pp.tick_sched_timer.__hrtimer_run_queues.hrtimer_interrupt.local_apic_timer_interrupt.smp_apic_timer_interrupt 2.52 ± 8% +57.6% 3.98 ± 15% perf-profile.cycles-pp.try_to_unmap.migrate_pages.migrate_misplaced_page.handle_pte_fault.handle_mm_fault 2.50 ± 8% +59.0% 3.97 ± 15% perf-profile.cycles-pp.try_to_unmap_one.rmap_walk_anon.rmap_walk.try_to_unmap.migrate_pages 10.46 ± 23% +27.4% 13.32 ± 5% perf-profile.cycles-pp.update_process_times.tick_sched_handle.isra.17.tick_sched_timer.__hrtimer_run_queues.hrtimer_interrupt 5.27 ± 7% +52.4% 8.02 ± 3% perf-profile.cycles-pp.vfs_read.sys_read.entry_SYSCALL_64_fastpath 1.67 ± 12% +63.6% 2.73 ± 16% perf-profile.cycles-pp.vfs_write.sys_write.entry_SYSCALL_64_fastpath *************************************************************************************************** lkp-bdw-de1: 16 threads Broadwell-DE with 8G memory ========================================================================================= cluster/compiler/cpufreq_governor/ip/kconfig/nr_threads/rootfs/runtime/send_size/tbox_group/test/testcase: cs-localhost/gcc-4.9/performance/ipv4/x86_64-rhel/200%/debian-x86_64-2015-02-07.cgz/300s/10K/lkp-bdw-de1/SCTP_STREAM_MANY/netperf commit: 0329dacce8 ("sched/fair: Fix effective_load()") 0cef819730 ("sched/fair: Fix calc_cfs_shares()") 0329dacce87d6741 0cef819730f7f67f8d2bec642e ---------------- -------------------------- %stddev %change %stddev \ | \ 2981 ± 0% -22.4% 2312 ± 0% netperf.Throughput_Mbps 48622550 ± 0% -100.0% 14709 ± 1% netperf.time.involuntary_context_switches 787.50 ± 0% +35.0% 1063 ± 0% netperf.time.percent_of_cpu_this_job_got 2327 ± 0% +34.7% 3135 ± 0% netperf.time.system_time 47.40 ± 1% +65.4% 78.38 ± 2% netperf.time.user_time 535968 ± 7% +2872.6% 15932329 ± 0% netperf.time.voluntary_context_switches 8626 ± 2% +12.4% 9694 ± 1% slabinfo.vm_area_struct.active_objs 8626 ± 2% +12.7% 9718 ± 1% slabinfo.vm_area_struct.num_objs 49962634 ± 0% -20.2% 39865448 ± 0% softirqs.NET_RX 277592 ± 12% -62.6% 103735 ± 2% softirqs.RCU 0.44 ± 1% +31.4% 0.57 ± 0% turbostat.CPU%c6 12.88 ± 0% +84.6% 23.78 ± 0% turbostat.RAMWatt 39.00 ± 0% +10.3% 43.00 ± 0% vmstat.procs.r 336813 ± 0% -65.4% 116656 ± 0% vmstat.system.cs 39652 ± 11% -31.7% 27080 ± 19% cpuidle.C1-BDW.usage 32415902 ± 1% +28.7% 41731000 ± 0% cpuidle.C6-BDW.time 35922 ± 1% +27.4% 45773 ± 1% cpuidle.C6-BDW.usage 123.50 ± 10% -38.9% 75.50 ± 24% cpuidle.POLL.usage 4.491e+08 ± 0% -21.8% 3.511e+08 ± 0% proc-vmstat.numa_hit 4.491e+08 ± 0% -21.8% 3.511e+08 ± 0% proc-vmstat.numa_local 6.407e+08 ± 0% -27.0% 4.68e+08 ± 0% proc-vmstat.pgalloc_dma32 1.947e+09 ± 0% -29.0% 1.382e+09 ± 0% proc-vmstat.pgalloc_normal 2.588e+09 ± 0% -28.5% 1.85e+09 ± 0% proc-vmstat.pgfree 5.271e+11 ± 0% -23.1% 4.055e+11 ± 0% perf-stat.L1-dcache-load-misses 1.773e+12 ± 0% -23.3% 1.36e+12 ± 0% perf-stat.L1-dcache-loads 1.249e+12 ± 0% -21.9% 9.761e+11 ± 0% perf-stat.L1-dcache-stores 6.895e+10 ± 1% -12.0% 6.067e+10 ± 0% perf-stat.L1-icache-load-misses 7.983e+10 ± 0% -9.8% 7.199e+10 ± 0% perf-stat.LLC-loads 1.609e+11 ± 0% -27.7% 1.163e+11 ± 0% perf-stat.LLC-stores 8.92e+11 ± 0% -24.6% 6.73e+11 ± 0% perf-stat.branch-instructions 4.957e+09 ± 0% -20.9% 3.923e+09 ± 1% perf-stat.branch-load-misses 8.903e+11 ± 0% -24.5% 6.725e+11 ± 0% perf-stat.branch-loads 4.963e+09 ± 0% -20.9% 3.925e+09 ± 1% perf-stat.branch-misses 4.416e+11 ± 0% -21.0% 3.488e+11 ± 0% perf-stat.cache-misses 4.417e+11 ± 0% -21.0% 3.489e+11 ± 0% perf-stat.cache-references 1.024e+08 ± 0% -65.4% 35461500 ± 0% perf-stat.context-switches 1.153e+13 ± 0% -1.3% 1.139e+13 ± 0% perf-stat.cpu-cycles 40497 ± 10% -48.3% 20955 ± 2% perf-stat.cpu-migrations 1.356e+09 ± 3% -38.7% 8.308e+08 ± 3% perf-stat.dTLB-load-misses 1.773e+12 ± 0% -23.3% 1.361e+12 ± 0% perf-stat.dTLB-loads 1.122e+09 ± 17% -37.3% 7.033e+08 ± 22% perf-stat.dTLB-store-misses 1.251e+12 ± 0% -21.8% 9.777e+11 ± 0% perf-stat.dTLB-stores 1.122e+09 ± 4% -38.0% 6.955e+08 ± 10% perf-stat.iTLB-load-misses 4.756e+12 ± 0% -23.7% 3.629e+12 ± 0% perf-stat.instructions 330533 ± 0% +4.5% 345307 ± 0% perf-stat.minor-faults 330530 ± 0% +4.5% 345272 ± 0% perf-stat.page-faults 40195 ± 15% +3697.4% 1526416 ± 3% sched_debug.cfs_rq:/.MIN_vruntime.avg 79275 ± 1% +1998.4% 1663557 ± 0% sched_debug.cfs_rq:/.MIN_vruntime.max 36697 ± 1% +903.9% 368405 ± 41% sched_debug.cfs_rq:/.MIN_vruntime.stddev 1418335 ± 6% -91.7% 117992 ± 5% sched_debug.cfs_rq:/.load.avg 1907632 ± 0% -90.6% 178528 ± 30% sched_debug.cfs_rq:/.load.max 848623 ± 0% -90.6% 79889 ± 4% sched_debug.cfs_rq:/.load.min 495671 ± 3% -93.8% 30899 ± 63% sched_debug.cfs_rq:/.load.stddev 912.30 ± 1% -86.5% 122.71 ± 9% sched_debug.cfs_rq:/.load_avg.avg 1120 ± 5% -70.9% 325.71 ± 24% sched_debug.cfs_rq:/.load_avg.max 847.83 ± 0% -90.7% 79.17 ± 3% sched_debug.cfs_rq:/.load_avg.min 40195 ± 15% +3697.4% 1526416 ± 3% sched_debug.cfs_rq:/.max_vruntime.avg 79275 ± 1% +1998.4% 1663557 ± 0% sched_debug.cfs_rq:/.max_vruntime.max 36697 ± 1% +903.9% 368405 ± 41% sched_debug.cfs_rq:/.max_vruntime.stddev 76340 ± 0% +2071.9% 1658025 ± 0% sched_debug.cfs_rq:/.min_vruntime.avg 82383 ± 0% +1921.5% 1665398 ± 0% sched_debug.cfs_rq:/.min_vruntime.max 75027 ± 0% +2099.4% 1650121 ± 0% sched_debug.cfs_rq:/.min_vruntime.min 1817 ± 4% +121.6% 4026 ± 4% sched_debug.cfs_rq:/.min_vruntime.stddev 1.37 ± 6% +23.6% 1.69 ± 1% sched_debug.cfs_rq:/.nr_running.avg 0.48 ± 3% -42.2% 0.28 ± 19% sched_debug.cfs_rq:/.nr_running.stddev 850.46 ± 0% -90.5% 80.57 ± 1% sched_debug.cfs_rq:/.runnable_load_avg.avg 905.92 ± 0% -88.7% 102.62 ± 8% sched_debug.cfs_rq:/.runnable_load_avg.max 807.00 ± 0% -94.6% 43.92 ± 25% sched_debug.cfs_rq:/.runnable_load_avg.min 31.69 ± 9% -52.9% 14.94 ± 26% sched_debug.cfs_rq:/.runnable_load_avg.stddev 1818 ± 4% +121.5% 4027 ± 4% sched_debug.cfs_rq:/.spread0.stddev 723347 ± 3% -8.8% 659993 ± 6% sched_debug.cpu.avg_idle.avg 5.42 ± 22% -61.2% 2.10 ± 23% sched_debug.cpu.clock.stddev 5.42 ± 22% -61.2% 2.10 ± 23% sched_debug.cpu.clock_task.stddev 849.63 ± 0% -90.7% 78.95 ± 3% sched_debug.cpu.cpu_load[0].avg 905.50 ± 0% -88.9% 100.83 ± 9% sched_debug.cpu.cpu_load[0].max 801.58 ± 0% -94.8% 41.58 ± 27% sched_debug.cpu.cpu_load[0].min 32.09 ± 10% -49.5% 16.19 ± 22% sched_debug.cpu.cpu_load[0].stddev 857.22 ± 0% -90.4% 82.45 ± 4% sched_debug.cpu.cpu_load[1].avg 909.88 ± 3% -87.0% 118.62 ± 28% sched_debug.cpu.cpu_load[1].max 804.17 ± 1% -92.8% 57.75 ± 7% sched_debug.cpu.cpu_load[1].min 854.43 ± 0% -90.2% 83.40 ± 5% sched_debug.cpu.cpu_load[2].avg 905.12 ± 3% -86.7% 120.12 ± 33% sched_debug.cpu.cpu_load[2].max 806.42 ± 1% -92.0% 64.79 ± 1% sched_debug.cpu.cpu_load[2].min 850.18 ± 0% -90.2% 83.12 ± 4% sched_debug.cpu.cpu_load[3].avg 892.46 ± 2% -87.4% 112.29 ± 27% sched_debug.cpu.cpu_load[3].max 810.62 ± 1% -91.5% 69.29 ± 2% sched_debug.cpu.cpu_load[3].min 845.87 ± 0% -90.1% 83.49 ± 2% sched_debug.cpu.cpu_load[4].avg 880.92 ± 1% -88.2% 104.21 ± 18% sched_debug.cpu.cpu_load[4].max 812.58 ± 0% -91.0% 73.50 ± 2% sched_debug.cpu.cpu_load[4].min 18.63 ± 19% -58.3% 7.77 ± 76% sched_debug.cpu.cpu_load[4].stddev 908.25 ± 1% -13.8% 783.00 ± 7% sched_debug.cpu.curr->pid.min 1377580 ± 7% -91.5% 117102 ± 7% sched_debug.cpu.load.avg 1907627 ± 0% -90.8% 174984 ± 32% sched_debug.cpu.load.max 847899 ± 0% -91.0% 76223 ± 0% sched_debug.cpu.load.min 490876 ± 0% -93.6% 31482 ± 59% sched_debug.cpu.load.stddev 4.29 ± 10% -28.2% 3.08 ± 8% sched_debug.cpu.nr_running.max 0.88 ± 8% +90.5% 1.67 ± 0% sched_debug.cpu.nr_running.min 0.93 ± 12% -51.1% 0.46 ± 12% sched_debug.cpu.nr_running.stddev 3166259 ± 0% -65.4% 1094903 ± 0% sched_debug.cpu.nr_switches.avg 4656563 ± 0% -72.5% 1281469 ± 1% sched_debug.cpu.nr_switches.max 3020925 ± 0% -67.4% 986125 ± 0% sched_debug.cpu.nr_switches.min 385862 ± 2% -76.3% 91611 ± 5% sched_debug.cpu.nr_switches.stddev 17.99 ± 0% +71.6% 30.88 ± 0% perf-profile.cycles-pp.___sys_recvmsg.__sys_recvmsg.sys_recvmsg.entry_SYSCALL_64_fastpath 78.39 ± 0% -16.2% 65.66 ± 0% perf-profile.cycles-pp.___sys_sendmsg.__sys_sendmsg.sys_sendmsg.entry_SYSCALL_64_fastpath 15.21 ± 0% -35.5% 9.81 ± 4% perf-profile.cycles-pp.__alloc_pages_nodemask.alloc_kmem_pages_node.kmalloc_large_node.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34 12.78 ± 0% -31.6% 8.73 ± 2% perf-profile.cycles-pp.__alloc_skb._sctp_make_chunk.sctp_make_datafrag_empty.sctp_datamsg_from_user.sctp_sendmsg 4.03 ± 2% -46.1% 2.17 ± 2% perf-profile.cycles-pp.__alloc_skb.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24 7.88 ± 1% +90.6% 15.01 ± 0% perf-profile.cycles-pp.__do_softirq.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip.ip_finish_output2 3.50 ± 2% -34.4% 2.29 ± 3% perf-profile.cycles-pp.__do_softirq.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip.sctp_inq_push 11.38 ± 2% -41.1% 6.71 ± 2% perf-profile.cycles-pp.__free_kmem_pages.kfree.skb_free_head.skb_release_data.skb_release_all 11.35 ± 2% -41.1% 6.68 ± 2% perf-profile.cycles-pp.__free_pages.__free_kmem_pages.kfree.skb_free_head.skb_release_data 11.29 ± 2% -41.4% 6.62 ± 2% perf-profile.cycles-pp.__free_pages_ok.__free_pages.__free_kmem_pages.kfree.skb_free_head 11.76 ± 0% -33.2% 7.85 ± 2% perf-profile.cycles-pp.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb._sctp_make_chunk.sctp_make_datafrag_empty 3.78 ± 2% -47.9% 1.97 ± 2% perf-profile.cycles-pp.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb.sctp_packet_transmit.sctp_outq_flush 11.81 ± 0% -33.1% 7.90 ± 2% perf-profile.cycles-pp.__kmalloc_reserve.isra.34.__alloc_skb._sctp_make_chunk.sctp_make_datafrag_empty.sctp_datamsg_from_user 3.80 ± 2% -48.0% 1.98 ± 2% perf-profile.cycles-pp.__kmalloc_reserve.isra.34.__alloc_skb.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork 7.92 ± 1% +90.2% 15.06 ± 0% perf-profile.cycles-pp.__local_bh_enable_ip.ip_finish_output2.ip_finish_output.ip_output.ip_local_out 3.56 ± 2% -34.5% 2.33 ± 3% perf-profile.cycles-pp.__local_bh_enable_ip.sctp_inq_push.sctp_backlog_rcv.__release_sock.release_sock 10.93 ± 1% +54.2% 16.84 ± 0% perf-profile.cycles-pp.__netif_receive_skb.process_backlog.net_rx_action.__do_softirq.do_softirq_own_stack 10.91 ± 1% +54.2% 16.83 ± 0% perf-profile.cycles-pp.__netif_receive_skb_core.__netif_receive_skb.process_backlog.net_rx_action.__do_softirq 0.00 ± -1% +Inf% 1.27 ± 3% perf-profile.cycles-pp.__release_sock.release_sock.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 28.24 ± 0% -19.9% 22.63 ± 0% perf-profile.cycles-pp.__release_sock.release_sock.sctp_sendmsg.inet_sendmsg.sock_sendmsg 0.99 ± 2% -100.0% 0.00 ± -1% perf-profile.cycles-pp.__schedule.preempt_schedule_common._cond_resched.__release_sock.release_sock 1.21 ± 4% -40.3% 0.72 ± 5% perf-profile.cycles-pp.__schedule.schedule.schedule_timeout.sctp_skb_recv_datagram.sctp_recvmsg 18.77 ± 0% +68.8% 31.68 ± 0% perf-profile.cycles-pp.__sys_recvmsg.sys_recvmsg.entry_SYSCALL_64_fastpath 79.54 ± 0% -16.1% 66.70 ± 0% perf-profile.cycles-pp.__sys_sendmsg.sys_sendmsg.entry_SYSCALL_64_fastpath 1.47 ± 4% -61.9% 0.56 ± 3% perf-profile.cycles-pp.__wake_up_common.__wake_up_sync_key.sctp_data_ready.sctp_ulpq_tail_event.sctp_ulpq_tail_data 1.51 ± 4% -61.6% 0.58 ± 4% perf-profile.cycles-pp.__wake_up_sync_key.sctp_data_ready.sctp_ulpq_tail_event.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24 1.04 ± 2% -100.0% 0.00 ± -1% perf-profile.cycles-pp._cond_resched.__release_sock.release_sock.sctp_sendmsg.inet_sendmsg 8.75 ± 2% -50.7% 4.31 ± 5% perf-profile.cycles-pp._raw_spin_lock.free_one_page.__free_pages_ok.__free_pages.__free_kmem_pages 10.80 ± 0% -43.5% 6.10 ± 6% perf-profile.cycles-pp._raw_spin_lock_irqsave.get_page_from_freelist.__alloc_pages_nodemask.alloc_kmem_pages_node.kmalloc_large_node 13.78 ± 0% -28.1% 9.90 ± 1% perf-profile.cycles-pp._sctp_make_chunk.sctp_make_datafrag_empty.sctp_datamsg_from_user.sctp_sendmsg.inet_sendmsg 15.41 ± 0% -33.6% 10.23 ± 2% perf-profile.cycles-pp.alloc_kmem_pages_node.kmalloc_large_node.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb 1.44 ± 4% -61.7% 0.55 ± 4% perf-profile.cycles-pp.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.sctp_data_ready.sctp_ulpq_tail_event 10.65 ± 2% -16.5% 8.89 ± 1% perf-profile.cycles-pp.consume_skb.sctp_chunk_put.sctp_chunk_free.sctp_outq_sack.sctp_cmd_interpreter.isra.24 1.20 ± 3% -12.9% 1.05 ± 6% perf-profile.cycles-pp.copy_msghdr_from_user.___sys_recvmsg.__sys_recvmsg.sys_recvmsg.entry_SYSCALL_64_fastpath 1.78 ± 0% -18.3% 1.45 ± 3% perf-profile.cycles-pp.copy_msghdr_from_user.___sys_sendmsg.__sys_sendmsg.sys_sendmsg.entry_SYSCALL_64_fastpath 9.09 ± 1% +28.5% 11.69 ± 0% perf-profile.cycles-pp.copy_user_enhanced_fast_string.sctp_user_addto_chunk.sctp_datamsg_from_user.sctp_sendmsg.inet_sendmsg 8.11 ± 0% +28.6% 10.43 ± 1% perf-profile.cycles-pp.copy_user_enhanced_fast_string.skb_copy_datagram_iter.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 1.43 ± 4% -61.6% 0.55 ± 4% perf-profile.cycles-pp.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.sctp_data_ready 7.94 ± 1% +89.9% 15.07 ± 0% perf-profile.cycles-pp.do_softirq.part.13.__local_bh_enable_ip.ip_finish_output2.ip_finish_output.ip_output 3.55 ± 2% -34.5% 2.33 ± 3% perf-profile.cycles-pp.do_softirq.part.13.__local_bh_enable_ip.sctp_inq_push.sctp_backlog_rcv.__release_sock 7.91 ± 1% +90.1% 15.04 ± 0% perf-profile.cycles-pp.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip.ip_finish_output2.ip_finish_output 3.52 ± 2% -34.4% 2.31 ± 3% perf-profile.cycles-pp.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip.sctp_inq_push.sctp_backlog_rcv 9.85 ± 2% -46.3% 5.29 ± 3% perf-profile.cycles-pp.free_one_page.__free_pages_ok.__free_pages.__free_kmem_pages.kfree 14.73 ± 0% -35.9% 9.44 ± 5% perf-profile.cycles-pp.get_page_from_freelist.__alloc_pages_nodemask.alloc_kmem_pages_node.kmalloc_large_node.__kmalloc_node_track_caller 74.88 ± 0% -16.3% 62.66 ± 0% perf-profile.cycles-pp.inet_sendmsg.sock_sendmsg.___sys_sendmsg.__sys_sendmsg.sys_sendmsg 8.19 ± 0% +85.8% 15.22 ± 0% perf-profile.cycles-pp.ip_finish_output.ip_output.ip_local_out.ip_queue_xmit.sctp_v4_xmit 8.16 ± 0% +86.5% 15.21 ± 0% perf-profile.cycles-pp.ip_finish_output2.ip_finish_output.ip_output.ip_local_out.ip_queue_xmit 10.41 ± 1% +57.5% 16.39 ± 0% perf-profile.cycles-pp.ip_local_deliver.ip_rcv_finish.ip_rcv.__netif_receive_skb_core.__netif_receive_skb 10.37 ± 1% +57.7% 16.35 ± 1% perf-profile.cycles-pp.ip_local_deliver_finish.ip_local_deliver.ip_rcv_finish.ip_rcv.__netif_receive_skb_core 9.63 ± 1% +62.1% 15.61 ± 2% perf-profile.cycles-pp.ip_local_out.ip_queue_xmit.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush 9.36 ± 3% +65.1% 15.46 ± 1% perf-profile.cycles-pp.ip_output.ip_local_out.ip_queue_xmit.sctp_v4_xmit.sctp_packet_transmit 0.00 ± -1% +Inf% 10.25 ± 1% perf-profile.cycles-pp.ip_queue_xmit.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush.sctp_outq_tail 9.86 ± 1% -42.1% 5.71 ± 1% perf-profile.cycles-pp.ip_queue_xmit.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork 10.67 ± 1% +55.8% 16.62 ± 0% perf-profile.cycles-pp.ip_rcv.__netif_receive_skb_core.__netif_receive_skb.process_backlog.net_rx_action 10.49 ± 1% +57.0% 16.47 ± 0% perf-profile.cycles-pp.ip_rcv_finish.ip_rcv.__netif_receive_skb_core.__netif_receive_skb.process_backlog 9.08 ± 2% -36.0% 5.81 ± 2% perf-profile.cycles-pp.kfree.skb_free_head.skb_release_data.skb_release_all.consume_skb 3.00 ± 2% -38.1% 1.86 ± 3% perf-profile.cycles-pp.kfree.skb_free_head.skb_release_data.skb_release_all.kfree_skb 3.68 ± 1% -26.7% 2.69 ± 3% perf-profile.cycles-pp.kfree_skb.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 11.68 ± 0% -33.3% 7.80 ± 2% perf-profile.cycles-pp.kmalloc_large_node.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb._sctp_make_chunk 3.77 ± 2% -34.3% 2.48 ± 2% perf-profile.cycles-pp.kmalloc_large_node.__kmalloc_node_track_caller.__kmalloc_reserve.isra.34.__alloc_skb.sctp_packet_transmit 0.00 ± -1% +Inf% 5.48 ± 3% perf-profile.cycles-pp.memcpy_erms.sctp_packet_transmit_chunk.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24 8.74 ± 2% -50.7% 4.30 ± 5% perf-profile.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock.free_one_page.__free_pages_ok.__free_pages 10.17 ± 1% -44.2% 5.68 ± 7% perf-profile.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_irqsave.get_page_from_freelist.__alloc_pages_nodemask.alloc_kmem_pages_node 11.21 ± 1% +52.8% 17.13 ± 0% perf-profile.cycles-pp.net_rx_action.__do_softirq.do_softirq_own_stack.do_softirq.part.13.__local_bh_enable_ip 1.04 ± 3% -100.0% 0.00 ± -1% perf-profile.cycles-pp.preempt_schedule_common._cond_resched.__release_sock.release_sock.sctp_sendmsg 11.05 ± 1% +53.5% 16.96 ± 0% perf-profile.cycles-pp.process_backlog.net_rx_action.__do_softirq.do_softirq_own_stack.do_softirq.part.13 0.00 ± -1% +Inf% 1.38 ± 3% perf-profile.cycles-pp.release_sock.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg 28.41 ± 0% -19.9% 22.75 ± 0% perf-profile.cycles-pp.release_sock.sctp_sendmsg.inet_sendmsg.sock_sendmsg.___sys_sendmsg 1.27 ± 4% -39.4% 0.77 ± 3% perf-profile.cycles-pp.schedule.schedule_timeout.sctp_skb_recv_datagram.sctp_recvmsg.sock_common_recvmsg 1.30 ± 4% -39.4% 0.79 ± 4% perf-profile.cycles-pp.schedule_timeout.sctp_skb_recv_datagram.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 23.46 ± 1% -9.0% 21.34 ± 0% perf-profile.cycles-pp.sctp_assoc_bh_rcv.sctp_inq_push.sctp_backlog_rcv.__release_sock.release_sock 8.11 ± 1% +73.2% 14.05 ± 1% perf-profile.cycles-pp.sctp_assoc_bh_rcv.sctp_inq_push.sctp_rcv.ip_local_deliver_finish.ip_local_deliver 0.00 ± -1% +Inf% 11.48 ± 1% perf-profile.cycles-pp.sctp_assoc_rwnd_increase.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 0.00 ± -1% +Inf% 1.23 ± 3% perf-profile.cycles-pp.sctp_backlog_rcv.__release_sock.release_sock.sctp_recvmsg.sock_common_recvmsg 27.11 ± 0% -16.8% 22.54 ± 0% perf-profile.cycles-pp.sctp_backlog_rcv.__release_sock.release_sock.sctp_sendmsg.inet_sendmsg 11.42 ± 1% -16.6% 9.52 ± 1% perf-profile.cycles-pp.sctp_chunk_free.sctp_outq_sack.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv 11.12 ± 1% -16.5% 9.29 ± 1% perf-profile.cycles-pp.sctp_chunk_put.sctp_chunk_free.sctp_outq_sack.sctp_cmd_interpreter.isra.24.sctp_do_sm 23.05 ± 1% -10.6% 20.60 ± 0% perf-profile.cycles-pp.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push.sctp_backlog_rcv 6.30 ± 2% +103.1% 12.80 ± 1% perf-profile.cycles-pp.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push.sctp_rcv 13.52 ± 0% -30.0% 9.46 ± 0% perf-profile.cycles-pp.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_primitive_SEND.sctp_sendmsg.inet_sendmsg 1.62 ± 4% -60.6% 0.64 ± 4% perf-profile.cycles-pp.sctp_data_ready.sctp_ulpq_tail_event.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24.sctp_do_sm 23.13 ± 1% -9.7% 20.88 ± 0% perf-profile.cycles-pp.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push.sctp_backlog_rcv.__release_sock 7.19 ± 2% +85.9% 13.36 ± 0% perf-profile.cycles-pp.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push.sctp_rcv.ip_local_deliver_finish 13.32 ± 0% -29.3% 9.41 ± 0% perf-profile.cycles-pp.sctp_do_sm.sctp_primitive_SEND.sctp_sendmsg.inet_sendmsg.sock_sendmsg 0.00 ± -1% +Inf% 1.23 ± 3% perf-profile.cycles-pp.sctp_inq_push.sctp_backlog_rcv.__release_sock.release_sock.sctp_recvmsg 24.60 ± 1% -14.8% 20.97 ± 0% perf-profile.cycles-pp.sctp_inq_push.sctp_backlog_rcv.__release_sock.release_sock.sctp_sendmsg 8.28 ± 1% +71.5% 14.19 ± 1% perf-profile.cycles-pp.sctp_inq_push.sctp_rcv.ip_local_deliver_finish.ip_local_deliver.ip_rcv_finish 14.48 ± 0% -27.4% 10.51 ± 1% perf-profile.cycles-pp.sctp_make_datafrag_empty.sctp_datamsg_from_user.sctp_sendmsg.inet_sendmsg.sock_sendmsg 0.00 ± -1% +Inf% 3.40 ± 2% perf-profile.cycles-pp.sctp_outq_flush.sctp_outq_tail.sctp_assoc_rwnd_increase.sctp_ulpevent_free.sctp_recvmsg 11.73 ± 0% +60.2% 18.79 ± 1% perf-profile.cycles-pp.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv 16.10 ± 0% -34.5% 10.55 ± 0% perf-profile.cycles-pp.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_primitive_SEND 12.62 ± 1% -13.3% 10.94 ± 0% perf-profile.cycles-pp.sctp_outq_sack.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push 0.00 ± -1% +Inf% 10.94 ± 1% perf-profile.cycles-pp.sctp_outq_tail.sctp_assoc_rwnd_increase.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg 11.81 ± 0% +60.2% 18.92 ± 1% perf-profile.cycles-pp.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push 16.26 ± 0% -34.3% 10.68 ± 0% perf-profile.cycles-pp.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_primitive_SEND.sctp_sendmsg 0.00 ± -1% +Inf% 9.57 ± 2% perf-profile.cycles-pp.sctp_packet_transmit.sctp_outq_flush.sctp_outq_tail.sctp_assoc_rwnd_increase.sctp_ulpevent_free 14.75 ± 0% -37.8% 9.18 ± 1% perf-profile.cycles-pp.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm 0.00 ± -1% +Inf% 1.22 ± 1% perf-profile.cycles-pp.sctp_packet_transmit.sctp_packet_transmit_chunk.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24 0.13 ±173% +5482.4% 7.12 ± 1% perf-profile.cycles-pp.sctp_packet_transmit_chunk.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24.sctp_do_sm 18.60 ± 0% -33.0% 12.46 ± 0% perf-profile.cycles-pp.sctp_primitive_SEND.sctp_sendmsg.inet_sendmsg.sock_sendmsg.___sys_sendmsg 9.94 ± 1% +60.4% 15.94 ± 1% perf-profile.cycles-pp.sctp_rcv.ip_local_deliver_finish.ip_local_deliver.ip_rcv_finish.ip_rcv 15.63 ± 1% +83.3% 28.65 ± 0% perf-profile.cycles-pp.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg.__sys_recvmsg 74.59 ± 0% -16.3% 62.43 ± 0% perf-profile.cycles-pp.sctp_sendmsg.inet_sendmsg.sock_sendmsg.___sys_sendmsg.__sys_sendmsg 1.68 ± 4% -35.4% 1.08 ± 6% perf-profile.cycles-pp.sctp_skb_recv_datagram.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg 4.26 ± 1% +236.0% 14.31 ± 0% perf-profile.cycles-pp.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg 1.49 ± 1% -65.9% 0.51 ± 70% perf-profile.cycles-pp.sctp_ulpevent_make_rcvmsg.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv 3.74 ± 2% -32.7% 2.51 ± 1% perf-profile.cycles-pp.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv.sctp_inq_push 1.74 ± 3% -59.1% 0.71 ± 3% perf-profile.cycles-pp.sctp_ulpq_tail_event.sctp_ulpq_tail_data.sctp_cmd_interpreter.isra.24.sctp_do_sm.sctp_assoc_bh_rcv 9.67 ± 1% +27.1% 12.29 ± 0% perf-profile.cycles-pp.sctp_user_addto_chunk.sctp_datamsg_from_user.sctp_sendmsg.inet_sendmsg.sock_sendmsg 0.00 ± -1% +Inf% 10.24 ± 1% perf-profile.cycles-pp.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush.sctp_outq_tail.sctp_assoc_rwnd_increase 9.91 ± 1% -41.5% 5.79 ± 1% perf-profile.cycles-pp.sctp_v4_xmit.sctp_packet_transmit.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.isra.24 0.61 ± 1% +165.9% 1.64 ± 3% perf-profile.cycles-pp.sctp_wfree.skb_release_head_state.skb_release_all.consume_skb.sctp_chunk_put 8.56 ± 1% +28.2% 10.98 ± 1% perf-profile.cycles-pp.skb_copy_datagram_iter.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg 9.10 ± 2% -36.0% 5.83 ± 2% perf-profile.cycles-pp.skb_free_head.skb_release_data.skb_release_all.consume_skb.sctp_chunk_put 3.00 ± 2% -38.0% 1.86 ± 2% perf-profile.cycles-pp.skb_free_head.skb_release_data.skb_release_all.kfree_skb.sctp_ulpevent_free 9.94 ± 2% -17.4% 8.21 ± 2% perf-profile.cycles-pp.skb_release_all.consume_skb.sctp_chunk_put.sctp_chunk_free.sctp_outq_sack 3.45 ± 2% -32.6% 2.33 ± 2% perf-profile.cycles-pp.skb_release_all.kfree_skb.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg 9.14 ± 2% -35.7% 5.88 ± 2% perf-profile.cycles-pp.skb_release_data.skb_release_all.consume_skb.sctp_chunk_put.sctp_chunk_free 3.19 ± 2% -35.2% 2.06 ± 3% perf-profile.cycles-pp.skb_release_data.skb_release_all.kfree_skb.sctp_ulpevent_free.sctp_recvmsg 0.78 ± 1% +150.6% 1.94 ± 3% perf-profile.cycles-pp.skb_release_head_state.skb_release_all.consume_skb.sctp_chunk_put.sctp_chunk_free 15.77 ± 1% +83.0% 28.85 ± 0% perf-profile.cycles-pp.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg.__sys_recvmsg.sys_recvmsg 16.23 ± 1% +80.7% 29.32 ± 0% perf-profile.cycles-pp.sock_recvmsg.___sys_recvmsg.__sys_recvmsg.sys_recvmsg.entry_SYSCALL_64_fastpath 75.34 ± 0% -16.2% 63.14 ± 0% perf-profile.cycles-pp.sock_sendmsg.___sys_sendmsg.__sys_sendmsg.sys_sendmsg.entry_SYSCALL_64_fastpath 18.79 ± 0% +68.7% 31.69 ± 0% perf-profile.cycles-pp.sys_recvmsg.entry_SYSCALL_64_fastpath 79.56 ± 0% -16.1% 66.71 ± 0% perf-profile.cycles-pp.sys_sendmsg.entry_SYSCALL_64_fastpath 1.40 ± 4% -61.1% 0.54 ± 3% perf-profile.cycles-pp.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key Thanks, Xiaolong