Hello,
kernel test robot noticed a 383.0% improvement of stress-ng.kcmp.ops_per_sec on:
commit: 90c436a64a6e20482a9a613c47eb4af2e8a5328e ("apparmor: pass cred through to audit info.")
https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master
testcase: stress-ng
test machine: 64 threads 2 sockets Intel(R) Xeon(R) Gold 6346 CPU @ 3.10GHz (Ice Lake) with 256G memory
parameters:
nr_threads: 10%
disk: 1HDD
testtime: 60s
fs: ext4
class: os
test: kcmp
cpufreq_governor: performance
Details are as below:
-------------------------------------------------------------------------------------------------->
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20231031/202310311037.173ebf2b-oliver.sang@xxxxxxxxx
=========================================================================================
class/compiler/cpufreq_governor/disk/fs/kconfig/nr_threads/rootfs/tbox_group/test/testcase/testtime:
os/gcc-12/performance/1HDD/ext4/x86_64-rhel-8.3/10%/debian-11.1-x86_64-20220510.cgz/lkp-icl-2sp7/kcmp/stress-ng/60s
commit:
d20f5a1a6e ("apparmor: rename audit_data->label to audit_data->subj_label")
90c436a64a ("apparmor: pass cred through to audit info.")
d20f5a1a6e792d22 90c436a64a6e20482a9a613c47e
---------------- ---------------------------
%stddev %change %stddev
\ | \
9.11 -11.6% 8.05 iostat.cpu.system
0.65 ± 3% +166.0% 1.73 ± 2% iostat.cpu.user
9678 -1.5% 9531 proc-vmstat.nr_mapped
7569 ± 3% -5.2% 7176 ± 2% proc-vmstat.nr_shmem
327.49 ± 22% +71.1% 560.45 ± 27% sched_debug.cfs_rq:/.min_vruntime.min
3740 ± 7% +472.8% 21422 ±175% sched_debug.cpu.avg_idle.min
0.03 ± 6% +0.0 0.04 ± 10% mpstat.cpu.all.iowait%
8.53 -1.1 7.44 mpstat.cpu.all.sys%
0.65 ± 3% +1.1 1.77 ± 2% mpstat.cpu.all.usr%
148.50 ± 12% -67.5% 48.33 ± 33% perf-c2c.DRAM.remote
227.00 ± 15% -64.8% 80.00 ± 28% perf-c2c.HITM.local
138.67 ± 12% -70.9% 40.33 ± 34% perf-c2c.HITM.remote
0.13 ± 2% +268.8% 0.47 turbostat.IPC
55.83 +3.3% 57.67 turbostat.PkgTmp
153.93 +5.7% 162.64 turbostat.PkgWatt
10117756 ± 2% +383.0% 48867736 stress-ng.kcmp.ops
168628 ± 2% +383.0% 814456 stress-ng.kcmp.ops_per_sec
345.46 -13.2% 299.82 stress-ng.time.system_time
11.72 ± 4% +389.0% 57.28 stress-ng.time.user_time
6.52 ± 15% +195.0% 19.23 ± 13% perf-sched.wait_and_delay.avg.ms.schedule_timeout.rcu_gp_fqs_loop.rcu_gp_kthread.kthread
16.50 ± 45% +77.8% 29.33 ± 20% perf-sched.wait_and_delay.count.rcu_gp_kthread.kthread.ret_from_fork
776.00 ± 14% -66.5% 260.17 ± 13% perf-sched.wait_and_delay.count.schedule_timeout.rcu_gp_fqs_loop.rcu_gp_kthread.kthread
647.02 ± 10% -11.5% 572.36 ± 6% perf-sched.wait_and_delay.max.ms.schedule_timeout.io_schedule_timeout.__wait_for_common.submit_bio_wait
258.50 ± 29% +70.3% 440.17 ± 9% perf-sched.wait_and_delay.max.ms.schedule_timeout.rcu_gp_fqs_loop.rcu_gp_kthread.kthread
6.51 ± 15% +195.2% 19.22 ± 13% perf-sched.wait_time.avg.ms.schedule_timeout.rcu_gp_fqs_loop.rcu_gp_kthread.kthread
647.01 ± 10% -11.5% 572.35 ± 6% perf-sched.wait_time.max.ms.schedule_timeout.io_schedule_timeout.__wait_for_common.submit_bio_wait
258.50 ± 29% +70.3% 440.16 ± 9% perf-sched.wait_time.max.ms.schedule_timeout.rcu_gp_fqs_loop.rcu_gp_kthread.kthread
0.43 -89.2% 0.05 ± 17% perf-stat.i.MPKI
1.384e+09 ± 2% +308.1% 5.649e+09 perf-stat.i.branch-instructions
0.31 ± 3% -0.2 0.15 ± 5% perf-stat.i.branch-miss-rate%
2929065 ± 3% -66.5% 982533 ± 25% perf-stat.i.cache-misses
9544710 -56.6% 4146447 ± 3% perf-stat.i.cache-references
2.98 ± 2% -76.1% 0.71 perf-stat.i.cpi
7054 ± 3% +341.9% 31175 ± 28% perf-stat.i.cycles-between-cache-misses
0.29 ±209% -0.3 0.01 ± 5% perf-stat.i.dTLB-load-miss-rate%
1.693e+09 ± 2% +318.1% 7.078e+09 perf-stat.i.dTLB-loads
0.27 ±218% -0.3 0.00 ± 9% perf-stat.i.dTLB-store-miss-rate%
8.821e+08 ± 2% +346.0% 3.934e+09 perf-stat.i.dTLB-stores
6.876e+09 ± 2% +312.2% 2.834e+10 perf-stat.i.instructions
0.36 ± 2% +294.8% 1.42 perf-stat.i.ipc
196.31 ± 2% -58.4% 81.75 ± 4% perf-stat.i.metric.K/sec
61.84 ± 2% +320.9% 260.30 perf-stat.i.metric.M/sec
93.40 -18.7 74.72 ± 8% perf-stat.i.node-store-miss-rate%
1.139e+12 ±223% -100.0% 312575 ± 40% perf-stat.i.node-store-misses
0.43 -91.9% 0.03 ± 26% perf-stat.overall.MPKI
0.42 ± 4% -0.3 0.10 ± 8% perf-stat.overall.branch-miss-rate%
2.89 ± 2% -75.7% 0.70 perf-stat.overall.cpi
6796 ± 3% +214.9% 21399 ± 20% perf-stat.overall.cycles-between-cache-misses
16.68 ±223% -16.7 0.01 ± 3% perf-stat.overall.dTLB-load-miss-rate%
16.67 ±223% -16.7 0.00 ± 4% perf-stat.overall.dTLB-store-miss-rate%
0.35 ± 2% +311.5% 1.42 perf-stat.overall.ipc
94.23 ± 2% -29.6 64.61 ± 14% perf-stat.overall.node-store-miss-rate%
1.362e+09 ± 2% +308.1% 5.559e+09 perf-stat.ps.branch-instructions
2881973 ± 3% -66.5% 966092 ± 25% perf-stat.ps.cache-misses
9392130 -56.6% 4077063 ± 3% perf-stat.ps.cache-references
1.666e+09 ± 2% +318.1% 6.965e+09 perf-stat.ps.dTLB-loads
8.68e+08 ± 2% +346.0% 3.871e+09 perf-stat.ps.dTLB-stores
6.766e+09 ± 2% +312.2% 2.789e+10 perf-stat.ps.instructions
1.12e+12 ±223% -100.0% 307444 ± 40% perf-stat.ps.node-store-misses
4.288e+11 ± 2% +311.2% 1.763e+12 perf-stat.total.instructions
74.76 -61.2 13.56 perf-profile.calltrace.cycles-pp.apparmor_ptrace_access_check.security_ptrace_access_check.ptrace_may_access.__do_sys_kcmp.do_syscall_64
75.40 -59.3 16.09 perf-profile.calltrace.cycles-pp.security_ptrace_access_check.ptrace_may_access.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe
52.51 ± 6% -52.5 0.00 perf-profile.calltrace.cycles-pp.aa_get_task_label.apparmor_ptrace_access_check.security_ptrace_access_check.ptrace_may_access.__do_sys_kcmp
77.46 -52.3 25.17 perf-profile.calltrace.cycles-pp.ptrace_may_access.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
85.77 -22.3 63.48 perf-profile.calltrace.cycles-pp.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
86.87 -18.6 68.26 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
87.39 -16.5 70.84 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.syscall
90.64 -4.4 86.22 perf-profile.calltrace.cycles-pp.syscall
0.00 +0.6 0.60 perf-profile.calltrace.cycles-pp.exit_to_user_mode_prepare.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
0.00 +0.8 0.77 ± 35% perf-profile.calltrace.cycles-pp.__errno_location
0.00 +0.8 0.84 ± 4% perf-profile.calltrace.cycles-pp._raw_spin_lock.task_lookup_fd_rcu.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe
0.00 +0.9 0.91 ± 2% perf-profile.calltrace.cycles-pp.cap_ptrace_access_check.security_ptrace_access_check.ptrace_may_access.__do_sys_kcmp.do_syscall_64
0.00 +1.0 0.97 ± 3% perf-profile.calltrace.cycles-pp.syscall_return_via_sysret.syscall
0.00 +1.0 1.00 ± 2% perf-profile.calltrace.cycles-pp.idr_find.find_task_by_vpid.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe
0.00 +1.0 1.03 ± 2% perf-profile.calltrace.cycles-pp.syscall_enter_from_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
0.00 +1.1 1.08 ± 2% perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_safe_stack.syscall
0.00 +1.1 1.09 ± 4% perf-profile.calltrace.cycles-pp.__cond_resched.down_read_killable.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe
0.00 +1.2 1.25 ± 3% perf-profile.calltrace.cycles-pp.task_lookup_fd_rcu.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
0.00 +1.5 1.49 ± 3% perf-profile.calltrace.cycles-pp.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
0.00 +1.5 1.52 ± 2% perf-profile.calltrace.cycles-pp.shim_kcmp
0.70 ± 4% +2.1 2.82 perf-profile.calltrace.cycles-pp.up_read.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
0.27 ±100% +2.2 2.47 perf-profile.calltrace.cycles-pp.__ptrace_may_access.ptrace_may_access.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe
0.65 ± 9% +2.5 3.13 ± 7% perf-profile.calltrace.cycles-pp.stress_kcmp
0.00 +3.2 3.15 ± 2% perf-profile.calltrace.cycles-pp.get_task_cred.apparmor_ptrace_access_check.security_ptrace_access_check.ptrace_may_access.__do_sys_kcmp
1.16 ± 5% +3.9 5.04 ± 2% perf-profile.calltrace.cycles-pp._raw_spin_lock.ptrace_may_access.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe
1.13 ± 5% +4.3 5.40 perf-profile.calltrace.cycles-pp.down_read_killable.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
1.08 ± 9% +4.9 5.97 perf-profile.calltrace.cycles-pp.aa_may_ptrace.apparmor_ptrace_access_check.security_ptrace_access_check.ptrace_may_access.__do_sys_kcmp
1.86 ± 3% +6.8 8.68 ± 2% perf-profile.calltrace.cycles-pp.__radix_tree_lookup.find_task_by_vpid.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe
2.41 ± 3% +9.0 11.40 perf-profile.calltrace.cycles-pp.__entry_text_start.syscall
2.73 ± 3% +10.1 12.87 perf-profile.calltrace.cycles-pp.find_task_by_vpid.__do_sys_kcmp.do_syscall_64.entry_SYSCALL_64_after_hwframe.syscall
74.90 -60.9 13.95 perf-profile.children.cycles-pp.apparmor_ptrace_access_check
75.54 -58.8 16.76 perf-profile.children.cycles-pp.security_ptrace_access_check
52.55 ± 6% -52.5 0.00 perf-profile.children.cycles-pp.aa_get_task_label
77.61 -51.8 25.82 perf-profile.children.cycles-pp.ptrace_may_access
86.06 -21.4 64.68 perf-profile.children.cycles-pp.__do_sys_kcmp
87.33 -18.1 69.27 perf-profile.children.cycles-pp.do_syscall_64
87.72 -16.5 71.24 perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe
90.62 -4.5 86.16 perf-profile.children.cycles-pp.syscall
0.00 +0.1 0.06 ± 9% perf-profile.children.cycles-pp.mutex_unlock
0.00 +0.1 0.06 ± 21% perf-profile.children.cycles-pp.fput
0.00 +0.1 0.10 ± 10% perf-profile.children.cycles-pp.mutex_lock
0.00 +0.1 0.14 ± 8% perf-profile.children.cycles-pp.fget_task
0.00 +0.2 0.17 ± 10% perf-profile.children.cycles-pp._copy_from_user
0.00 +0.2 0.18 ± 16% perf-profile.children.cycles-pp.get_epoll_tfile_raw_ptr
0.06 ± 17% +0.2 0.27 ± 6% perf-profile.children.cycles-pp.syscall@plt
0.07 ± 6% +0.2 0.30 ± 7% perf-profile.children.cycles-pp.yama_ptrace_access_check
0.11 ± 58% +0.3 0.41 ± 4% perf-profile.children.cycles-pp.__errno_location@plt
0.16 ± 8% +0.4 0.55 ± 5% perf-profile.children.cycles-pp.syscall_exit_to_user_mode_prepare
0.12 ± 11% +0.4 0.53 ± 5% perf-profile.children.cycles-pp.rcu_all_qs
0.12 ± 14% +0.4 0.54 ± 4% perf-profile.children.cycles-pp.radix_tree_lookup
0.12 ± 12% +0.4 0.56 ± 3% perf-profile.children.cycles-pp.__x64_sys_kcmp
0.18 ± 9% +0.6 0.74 ± 2% perf-profile.children.cycles-pp.exit_to_user_mode_prepare
0.18 ± 13% +0.7 0.92 ± 29% perf-profile.children.cycles-pp.__errno_location
0.22 ± 4% +0.8 1.03 ± 2% perf-profile.children.cycles-pp.cap_ptrace_access_check
0.24 ± 5% +0.8 1.09 ± 2% perf-profile.children.cycles-pp.entry_SYSCALL_64_safe_stack
0.21 ± 5% +0.9 1.10 ± 2% perf-profile.children.cycles-pp.syscall_return_via_sysret
0.26 ± 4% +0.9 1.16 ± 2% perf-profile.children.cycles-pp.syscall_enter_from_user_mode
0.26 ± 6% +1.0 1.28 ± 2% perf-profile.children.cycles-pp.idr_find
0.31 ± 8% +1.0 1.33 ± 3% perf-profile.children.cycles-pp.task_lookup_fd_rcu
0.31 ± 7% +1.1 1.46 ± 3% perf-profile.children.cycles-pp.__cond_resched
0.47 ± 6% +1.5 1.95 ± 3% perf-profile.children.cycles-pp.syscall_exit_to_user_mode
0.38 ± 10% +1.5 1.91 ± 2% perf-profile.children.cycles-pp.shim_kcmp
0.55 ± 7% +2.2 2.74 perf-profile.children.cycles-pp.__ptrace_may_access
0.72 ± 4% +2.2 2.96 perf-profile.children.cycles-pp.up_read
0.71 ± 8% +2.7 3.36 ± 6% perf-profile.children.cycles-pp.stress_kcmp
0.00 +3.3 3.30 ± 2% perf-profile.children.cycles-pp.get_task_cred
1.16 ± 4% +4.3 5.49 perf-profile.children.cycles-pp.entry_SYSRETQ_unsafe_stack
1.20 ± 5% +4.6 5.75 perf-profile.children.cycles-pp.down_read_killable
1.54 ± 4% +4.8 6.31 perf-profile.children.cycles-pp._raw_spin_lock
1.10 ± 9% +5.0 6.11 perf-profile.children.cycles-pp.aa_may_ptrace
1.38 ± 3% +5.2 6.57 perf-profile.children.cycles-pp.__entry_text_start
1.92 ± 3% +7.0 8.96 ± 2% perf-profile.children.cycles-pp.__radix_tree_lookup
2.85 ± 4% +10.6 13.44 perf-profile.children.cycles-pp.find_task_by_vpid
52.33 ± 6% -52.3 0.00 perf-profile.self.cycles-pp.aa_get_task_label
21.23 ± 16% -16.6 4.67 ± 5% perf-profile.self.cycles-pp.apparmor_ptrace_access_check
0.00 +0.1 0.06 ± 8% perf-profile.self.cycles-pp.mutex_unlock
0.00 +0.1 0.07 ± 14% perf-profile.self.cycles-pp.mutex_lock
0.00 +0.1 0.09 ± 12% perf-profile.self.cycles-pp.fget_task
0.00 +0.1 0.13 ± 9% perf-profile.self.cycles-pp.syscall@plt
0.00 +0.2 0.17 ± 11% perf-profile.self.cycles-pp._copy_from_user
0.02 ± 99% +0.2 0.20 ± 8% perf-profile.self.cycles-pp.yama_ptrace_access_check
0.06 ± 13% +0.2 0.27 ± 4% perf-profile.self.cycles-pp.radix_tree_lookup
0.06 +0.2 0.28 ± 5% perf-profile.self.cycles-pp.__x64_sys_kcmp
0.09 ± 14% +0.2 0.33 ± 7% perf-profile.self.cycles-pp.syscall_exit_to_user_mode_prepare
0.08 ± 16% +0.3 0.36 ± 4% perf-profile.self.cycles-pp.rcu_all_qs
0.09 ± 16% +0.3 0.43 ± 6% perf-profile.self.cycles-pp.task_lookup_fd_rcu
0.13 ± 13% +0.5 0.61 perf-profile.self.cycles-pp.exit_to_user_mode_prepare
0.16 ± 8% +0.6 0.71 ± 2% perf-profile.self.cycles-pp.cap_ptrace_access_check
0.19 ± 4% +0.6 0.79 ± 3% perf-profile.self.cycles-pp.syscall_exit_to_user_mode
0.11 ± 44% +0.6 0.76 ± 35% perf-profile.self.cycles-pp.__errno_location
0.19 ± 8% +0.7 0.91 ± 4% perf-profile.self.cycles-pp.__cond_resched
0.22 ± 6% +0.8 0.99 ± 2% perf-profile.self.cycles-pp.syscall_enter_from_user_mode
0.21 ± 5% +0.8 0.99 ± 3% perf-profile.self.cycles-pp.idr_find
0.24 ± 5% +0.8 1.09 ± 2% perf-profile.self.cycles-pp.entry_SYSCALL_64_safe_stack
0.21 ± 5% +0.9 1.10 ± 2% perf-profile.self.cycles-pp.syscall_return_via_sysret
0.28 ± 7% +1.0 1.28 ± 2% perf-profile.self.cycles-pp.do_syscall_64
0.36 ± 4% +1.1 1.46 ± 4% perf-profile.self.cycles-pp.ptrace_may_access
0.29 ± 13% +1.2 1.52 ± 2% perf-profile.self.cycles-pp.shim_kcmp
0.36 ± 5% +1.4 1.74 ± 5% perf-profile.self.cycles-pp.__entry_text_start
0.44 ± 6% +1.4 1.84 perf-profile.self.cycles-pp.security_ptrace_access_check
0.41 ± 4% +1.7 2.11 perf-profile.self.cycles-pp.entry_SYSCALL_64_after_hwframe
0.50 ± 8% +2.0 2.47 perf-profile.self.cycles-pp.__ptrace_may_access
0.68 ± 4% +2.1 2.81 perf-profile.self.cycles-pp.up_read
0.63 ± 5% +2.4 3.03 ± 2% perf-profile.self.cycles-pp.find_task_by_vpid
0.67 ± 9% +2.5 3.21 ± 7% perf-profile.self.cycles-pp.stress_kcmp
0.00 +3.1 3.14 ± 2% perf-profile.self.cycles-pp.get_task_cred
0.90 ± 6% +3.4 4.32 perf-profile.self.cycles-pp.down_read_killable
1.14 ± 4% +4.2 5.36 perf-profile.self.cycles-pp.entry_SYSRETQ_unsafe_stack
1.46 ± 3% +4.5 5.97 ± 2% perf-profile.self.cycles-pp._raw_spin_lock
1.22 ± 3% +4.5 5.73 perf-profile.self.cycles-pp.syscall
1.07 ± 8% +4.9 5.99 perf-profile.self.cycles-pp.aa_may_ptrace
1.86 ± 4% +6.8 8.66 ± 2% perf-profile.self.cycles-pp.__radix_tree_lookup
3.48 ± 4% +12.2 15.72 perf-profile.self.cycles-pp.__do_sys_kcmp
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/wiki
