Hi Andy,
we noticed the below changes on
git://git.kernel.org/pub/scm/linux/kernel/git/luto/linux.git x86/vsyscall
commit 21d4ab4881ad9b257bec75d04480105dad4336e1 ("x86_64,vsyscall: Move all of the gate_area code to vsyscall_64.c")
test case: lkp-wsx01/will-it-scale/signal1
a7781f1035319a7 21d4ab4881ad9b257bec75d04
--------------- -------------------------
259032 ~ 1% -11.1% 230288 ~ 0% TOTAL will-it-scale.per_process_ops
0.04 ~ 9% +4276.2% 1.84 ~ 1% TOTAL perf-profile.cpu-cycles.map_id_up.do_tkill.sys_tgkill.system_call_fastpath.raise
2.36 ~ 0% -63.8% 0.85 ~ 2% TOTAL perf-profile.cpu-cycles._atomic_dec_and_lock.free_uid.__sigqueue_free.__dequeue_signal.dequeue_signal
2.25 ~14% -55.2% 1.01 ~ 1% TOTAL perf-profile.cpu-cycles.raise
42.41 ~ 0% +34.5% 57.04 ~ 0% TOTAL perf-profile.cpu-cycles.__sigqueue_alloc.__send_signal.send_signal.do_send_sig_info.do_send_specific
40.70 ~ 0% -23.9% 30.96 ~ 0% TOTAL perf-profile.cpu-cycles.__sigqueue_free.part.11.__dequeue_signal.dequeue_signal.get_signal_to_deliver.do_signal
252 ~11% -18.8% 204 ~ 9% TOTAL numa-vmstat.node1.nr_page_table_pages
1012 ~11% -18.3% 827 ~ 9% TOTAL numa-meminfo.node1.PageTables
520 ~ 7% -17.1% 431 ~ 5% TOTAL cpuidle.C1-NHM.usage
Legend:
~XX% - stddev percent
[+-]XX% - change percent
will-it-scale.per_process_ops
270000 ++-----------------------------------------------------------------+
| |
260000 *+.*..*.. .*.. .*.. .*..*... .*.. *.. .*.. *.. .*..*
| *..*. *. *. *. .. *. .. *..*. |
250000 ++ * * |
| |
240000 ++ |
| |
230000 ++ O O O O O O O O O O O |
O O O |
220000 ++ O O O |
| O O O O |
210000 ++ O |
| |
200000 ++-----------------------------------------------------------------+
[*] bisect-good sample
[O] bisect-bad sample
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.
Thanks,
Jet