FYI, we noticed a -5.5% regression of netpipe.bigger_5M_Mbps.avg due to commit: commit 3613b3dbd1ade9a6a626dae1f608c57638eb5e8a ("tcp: prepare skbs for better sack shifting") https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master in testcase: netpipe on test machine: qemu-system-x86_64 -enable-kvm -cpu host -smp 4 -m 6G with following parameters: test: tcp cluster: cs-localhost NetPIPE is a protocol independent performance tool that encapsulates the best of ttcp and netperf and visually represents the network performance under a variety of conditions. 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/kconfig/rootfs/tbox_group/test/testcase: cs-localhost/gcc-6/x86_64-rhel-7.2/debian-x86_64-2016-08-31.cgz/vm-lkp-a08/tcp/netpipe commit: e812bd905a (" wireless-drivers-next patches for 4.9") 3613b3dbd1 ("tcp: prepare skbs for better sack shifting") e812bd905a5cf00f 3613b3dbd1ade9a6a626dae1f6 ---------------- -------------------------- fail:runs %reproduction fail:runs | | | %stddev %change %stddev \ | \ 2427 ± 0% -5.5% 2294 ± 0% netpipe.bigger_5M_Mbps.avg 1764 ± 1% +5.0% 1853 ± 0% netpipe.time.elapsed_time 1764 ± 1% +5.0% 1853 ± 0% netpipe.time.elapsed_time.max 84.75 ± 0% -2.1% 83.00 ± 0% netpipe.time.percent_of_cpu_this_job_got 1497 ± 1% +3.2% 1546 ± 0% netpipe.time.system_time 1378335 ± 1% +28.6% 1772455 ± 1% netpipe.time.voluntary_context_switches 14140 ± 3% +6.6% 15079 ± 2% interrupts.CAL:Function_call_interrupts 393.25 ± 2% -29.1% 279.00 ± 2% slabinfo.kmalloc-2048.active_objs 412.00 ± 1% -32.3% 279.00 ± 2% slabinfo.kmalloc-2048.num_objs 4069 ± 0% +17.5% 4779 ± 0% vmstat.system.cs 2961 ± 0% +5.3% 3118 ± 0% vmstat.system.in 229439 ± 94% +6926.4% 16121408 ±172% latency_stats.avg.max 141971 ±160% +11239.9% 16099418 ±172% latency_stats.avg.nfs_wait_on_request.nfs_writepage_setup.nfs_updatepage.nfs_write_end.generic_perform_write.nfs_file_write.__vfs_write.vfs_write.SyS_write.entry_SYSCALL_64_fastpath 90183 ±153% -55.4% 40178 ± 84% latency_stats.avg.proc_cgroup_show.proc_single_show.seq_read.__vfs_read.vfs_read.SyS_read.entry_SYSCALL_64_fastpath 839.25 ± 21% +1114.0% 10188 ±125% latency_stats.avg.rpc_wait_bit_killable.__rpc_execute.rpc_execute.rpc_run_task.nfs4_call_sync_sequence.[nfsv4]._nfs4_proc_lookup.[nfsv4].nfs4_proc_lookup_common.[nfsv4].nfs4_proc_lookup.[nfsv4].nfs_lookup.lookup_slow.walk_component.path_lookupat 414419 ± 86% +7661.1% 32163484 ±172% latency_stats.max.max 257867 ±166% +12362.7% 32137192 ±172% latency_stats.max.nfs_wait_on_request.nfs_writepage_setup.nfs_updatepage.nfs_write_end.generic_perform_write.nfs_file_write.__vfs_write.vfs_write.SyS_write.entry_SYSCALL_64_fastpath 81137 ±155% -94.7% 4265 ± 45% latency_stats.max.pipe_read.__vfs_read.vfs_read.SyS_read.entry_SYSCALL_64_fastpath 90183 ±153% -55.4% 40178 ± 84% latency_stats.max.proc_cgroup_show.proc_single_show.seq_read.__vfs_read.vfs_read.SyS_read.entry_SYSCALL_64_fastpath 839.25 ± 21% +1114.0% 10188 ±125% latency_stats.max.rpc_wait_bit_killable.__rpc_execute.rpc_execute.rpc_run_task.nfs4_call_sync_sequence.[nfsv4]._nfs4_proc_lookup.[nfsv4].nfs4_proc_lookup_common.[nfsv4].nfs4_proc_lookup.[nfsv4].nfs_lookup.lookup_slow.walk_component.path_lookupat 6232 ± 87% -100.0% 0.00 ± -1% latency_stats.sum.call_rwsem_down_read_failed.__do_page_fault.trace_do_page_fault.do_async_page_fault.async_page_fault.skb_copy_datagram_iter.tcp_rcv_established.tcp_v4_do_rcv.tcp_prequeue_process.tcp_recvmsg.inet_recvmsg.sock_recvmsg 1107 ± 62% +964.9% 11788 ± 55% latency_stats.sum.call_rwsem_down_read_failed.proc_pid_cmdline_read.__vfs_read.vfs_read.SyS_read.do_syscall_64.return_from_SYSCALL_64 276091 ±166% +11542.5% 32143926 ±172% latency_stats.sum.nfs_wait_on_request.nfs_writepage_setup.nfs_updatepage.nfs_write_end.generic_perform_write.nfs_file_write.__vfs_write.vfs_write.SyS_write.entry_SYSCALL_64_fastpath 97392 ±128% -82.9% 16660 ± 29% latency_stats.sum.pipe_read.__vfs_read.vfs_read.SyS_read.entry_SYSCALL_64_fastpath 90183 ±153% -55.4% 40178 ± 84% latency_stats.sum.proc_cgroup_show.proc_single_show.seq_read.__vfs_read.vfs_read.SyS_read.entry_SYSCALL_64_fastpath 839.25 ± 21% +1114.0% 10188 ±125% latency_stats.sum.rpc_wait_bit_killable.__rpc_execute.rpc_execute.rpc_run_task.nfs4_call_sync_sequence.[nfsv4]._nfs4_proc_lookup.[nfsv4].nfs4_proc_lookup_common.[nfsv4].nfs4_proc_lookup.[nfsv4].nfs_lookup.lookup_slow.walk_component.path_lookupat 3561 ± 11% +12.9% 4020 ± 6% sched_debug.cpu.curr->pid.avg 9783 ± 6% +11.8% 10942 ± 3% sched_debug.cpu.curr->pid.max 4222 ± 7% +11.3% 4701 ± 3% sched_debug.cpu.curr->pid.stddev 3988122 ± 13% +24.4% 4962088 ± 4% sched_debug.cpu.max_idle_balance_cost.max 1330791 ± 13% +30.8% 1740923 ± 4% sched_debug.cpu.max_idle_balance_cost.stddev 576379 ± 1% +9.7% 632485 ± 3% sched_debug.cpu.nr_load_updates.max 987546 ± 1% +20.0% 1184637 ± 1% sched_debug.cpu.nr_switches.avg 1074939 ± 2% +23.9% 1332366 ± 3% sched_debug.cpu.nr_switches.max 67934 ± 31% +110.9% 143278 ± 5% sched_debug.cpu.nr_switches.stddev -15.31 ±-32% -41.6% -8.93 ±-57% sched_debug.cpu.nr_uninterruptible.min 990916 ± 1% +20.2% 1190867 ± 1% sched_debug.cpu.sched_count.avg 1079929 ± 2% +24.2% 1341786 ± 3% sched_debug.cpu.sched_count.max 69632 ± 31% +111.3% 147114 ± 5% sched_debug.cpu.sched_count.stddev 386683 ± 1% +25.2% 483992 ± 2% sched_debug.cpu.sched_goidle.avg 424674 ± 1% +29.4% 549482 ± 1% sched_debug.cpu.sched_goidle.max 33344 ± 23% +107.5% 69183 ± 9% sched_debug.cpu.sched_goidle.stddev 513012 ± 1% +19.1% 610887 ± 1% sched_debug.cpu.ttwu_count.avg 570351 ± 5% +24.2% 708599 ± 5% sched_debug.cpu.ttwu_count.max 40059 ± 42% +112.3% 85064 ± 14% sched_debug.cpu.ttwu_count.stddev Thanks, Xiaolong