hi, Anna-Maria, On Thu, Apr 25, 2024 at 04:23:17PM +0800, Anna-Maria Behnsen wrote: > Hi, > > (adding cpuidle/power people to cc-list) > > Oliver Sang writes: > > > hi, Frederic Weisbecker, > > > > On Tue, Apr 02, 2024 at 12:46:15AM +0200, Frederic Weisbecker wrote: > >> Le Wed, Mar 27, 2024 at 04:39:17PM +0800, kernel test robot a écrit : > >> > > >> > > >> > Hello, > >> > > >> > > >> > we reported > >> > "[tip:timers/core] [timers] 7ee9887703: netperf.Throughput_Mbps -1.2% regression" > >> > in > >> > https://lore.kernel.org/all/202403011511.24defbbd-oliver.sang@intel.com/ > >> > > >> > now we noticed this commit is in mainline and we captured further results. > >> > > >> > still include netperf results for complete. below details FYI. > >> > > >> > > >> > kernel test robot noticed a -17.1% regression of stress-ng.uprobe.ops_per_sec > >> > on: > >> > >> The good news is that I can reproduce. > >> It has made me spot something already: > >> > >> https://lore.kernel.org/lkml/ZgsynV536q1L17IS@pavilion.home/T/#m28c37a943fdbcbadf0332cf9c32c350c74c403b0 > >> > >> But that's not enough to fix the regression. Investigation continues... > > > > Thanks a lot for information! if you want us test any patch, please let us know. > > Oliver, I would be happy to see, whether the patch at the end of the > message restores the original behaviour also in your test setup. I > applied it on 6.9-rc4. This patch is not a fix - it is just a pointer to > the kernel path, that might cause the regression. I know, it is > probable, that a warning in tick_sched is triggered. This happens when > the first timer is alredy in the past. I didn't add an extra check when > creating the 'defacto' timer thingy. But existing code handles this > problem already properly. So the warning could be ignored here. yes, the patch restores the original behaviour in our test setup. and right, we saw a WARNING:at_kernel/time/tick-sched.c:#tick_nohz_next_event I also applied the patch upon 6.9-rc4, then build 6.9-rc4 and 6.9-rc4+patch with same config (attached), by same test we made original report, we got below data [1]. from (a) in [1], we just see very similar v6.9-rc4 data with 7ee9887703 data in our original report, and v6.9-rc4+patch data is very similar to 57e95a5c41 (the parent of 7ee9887703). though you said the warning could be ignored, I still attach one dmesg in case you want to have a look. (BTW, the WARNING happened twice in this dmesg) [1] ========================================================================================= compiler/cpufreq_governor/kconfig/nr_threads/rootfs/tbox_group/test/testcase/testtime: gcc-12/performance/x86_64-rhel-8.3/100%/debian-12-x86_64-20240206.cgz/lkp-spr-2sp4/uprobe/stress-ng/60s commit: v6.9-rc4 v6.9-rc4+patch v6.9-rc4 afc95ee83a86426924f100321bd ---------------- --------------------------- %stddev %change %stddev \ | \ 8450322 +17.4% 9923989 cpuidle..usage 0.50 ± 6% +0.1 0.60 ± 8% mpstat.cpu.all.sys% 7588 +2.5% 7774 vmstat.system.cs 143938 +15.6% 166345 vmstat.system.in 222151 +13.5% 252196 time.minor_page_faults 100.50 ± 6% +23.7% 124.33 ± 9% time.percent_of_cpu_this_job_got 60.65 ± 6% +23.4% 74.87 ± 9% time.system_time 133973 +2.6% 137487 time.voluntary_context_switches 222151 +13.5% 252196 stress-ng.time.minor_page_faults 100.50 ± 6% +23.7% 124.33 ± 9% stress-ng.time.percent_of_cpu_this_job_got 60.65 ± 6% +23.4% 74.87 ± 9% stress-ng.time.system_time 133973 +2.6% 137487 stress-ng.time.voluntary_context_switches 996193 +21.3% 1208081 stress-ng.uprobe.ops 16600 +21.3% 20132 stress-ng.uprobe.ops_per_sec <----- (a) 8542 ± 2% +4.4% 8920 ± 3% proc-vmstat.nr_active_anon 8542 ± 2% +4.4% 8920 ± 3% proc-vmstat.nr_zone_active_anon 1387019 +6.2% 1473416 proc-vmstat.numa_hit 1060772 +7.1% 1135960 proc-vmstat.numa_local 326227 +3.4% 337389 proc-vmstat.numa_other 1457285 +6.0% 1545091 proc-vmstat.pgalloc_normal 700003 +4.9% 734444 proc-vmstat.pgfault 1268538 +7.8% 1367139 proc-vmstat.pgfree 9.152e+08 +6.3% 9.728e+08 ± 2% perf-stat.i.branch-instructions 2.60 ± 2% -0.1 2.46 perf-stat.i.branch-miss-rate% 12.07 ± 2% +0.9 12.96 ± 2% perf-stat.i.cache-miss-rate% 4068158 +12.1% 4559133 perf-stat.i.cache-misses 30326543 +7.8% 32700896 perf-stat.i.cache-references 7.997e+09 ± 3% +14.3% 9.138e+09 ± 4% perf-stat.i.cpu-cycles 4.453e+09 +6.1% 4.724e+09 ± 2% perf-stat.i.instructions 0.51 -7.4% 0.47 perf-stat.i.ipc 0.91 ± 2% +5.7% 0.96 ± 3% perf-stat.overall.MPKI 3.74 -0.2 3.53 ± 2% perf-stat.overall.branch-miss-rate% 13.36 ± 2% +0.5 13.89 ± 2% perf-stat.overall.cache-miss-rate% 1.80 +7.7% 1.93 ± 2% perf-stat.overall.cpi 0.56 -7.1% 0.52 ± 2% perf-stat.overall.ipc 8.993e+08 +6.3% 9.563e+08 ± 2% perf-stat.ps.branch-instructions 3983131 ± 2% +12.2% 4467972 perf-stat.ps.cache-misses 29818286 +7.9% 32162191 perf-stat.ps.cache-references 7.857e+09 ± 3% +14.3% 8.983e+09 ± 4% perf-stat.ps.cpu-cycles 4.376e+09 +6.1% 4.645e+09 ± 2% perf-stat.ps.instructions 2.684e+11 +6.4% 2.856e+11 ± 2% perf-stat.total.instructions