Hi Johannes,
PSI is a great idea but it does have overhead and if enabled by Kconfig
then it incurs a hit whether the user is aware of the feature or not. I
think enabling by default is unnecessary as it should only be enabled if
the information is being consumed. While the Kconfig exists, it's all or
nothing if distributions want to have the feature available.
I've included a bisection report below showing a 6-10% regression on a
single socket skylake machine. Would you mind doing one or all of the
following to fix it please?
a) disable it by default
b) put psi_disable behind a static branch to move the overhead to zero
if it's disabled
c) optionally enable/disable at runtime (least important as at a glance,
this may be problematic)
Thanks
Bisect parameters
=================
CONFIG_AUTO=openSUSE-LEAP-15.0
BUILD_TYPE=make
MACHINE=delboy
BISECT_REVERSE=no
BISECT_MMTESTS_TEST=hackbench-process-pipes
BISECT_CONFIG=global-dhp__scheduler-unbound
BISECT_GOOD=v4.19
BISECT_BAD=92b419289cee
BISECT_LOGDIR=/srv/marvin/impera/bisections-sigma-delboy-global-dhp__scheduler-unbound-openSUSE-LEAP-15.0-openSUSE-LEAP-15.0-v4.19..92b419289cee
BISECT_COMPARE=Amean
BISECT_PREFER=Lower
BISECT_CLIENT="5"
BISECT_REBOOT_CLEAN=no
BISECT_MONITOR_CONFIG=
BISECT_NO_MITIGATIONS=no
BISECT_COMMAND="bisection-run --machine delboy --distro openSUSE-LEAP-15.0 --kernel-config openSUSE-LEAP-15.0 --tree make --config global-dhp__scheduler-unbound --mmtests-test hackbench-process-pipes --monitor no-monitor --method sigma --good v4.19 --bad 92b419289cee --walk-mainline --notify [email protected] --bisect-client 5 --mmtests-limit --iterations 1 --logdir /srv/marvin/impera/bisections-sigma-delboy-global-dhp__scheduler-unbound-openSUSE-LEAP-15.0-openSUSE-LEAP-15.0-v4.19..92b419289cee --reset-trees"
BISECT_LOGDIR=/srv/marvin/impera/bisections-sigma-delboy-global-dhp__scheduler-unbound-openSUSE-LEAP-15.0-openSUSE-LEAP-15.0-v4.19..92b419289cee
Last good/First bad commit
==========================
Last good commit: eb414681d5a07d28d2ff90dc05f69ec6b232ebd2
First bad commit: 2ce7135adc9ad081aa3c49744144376ac74fea60
From 2ce7135adc9ad081aa3c49744144376ac74fea60 Mon Sep 17 00:00:00 2001
From: Johannes Weiner <[email protected]>
Date: Fri, 26 Oct 2018 15:06:31 -0700
Subject: [PATCH] psi: cgroup support
On a system that executes multiple cgrouped jobs and independent
workloads, we don't just care about the health of the overall system, but
also that of individual jobs, so that we can ensure individual job health,
fairness between jobs, or prioritize some jobs over others.
This patch implements pressure stall tracking for cgroups. In kernels
with CONFIG_PSI=y, cgroup2 groups will have cpu.pressure, memory.pressure,
and io.pressure files that track aggregate pressure stall times for only
the tasks inside the cgroup.
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Johannes Weiner <[email protected]>
Acked-by: Tejun Heo <[email protected]>
Acked-by: Peter Zijlstra (Intel) <[email protected]>
Tested-by: Daniel Drake <[email protected]>
Tested-by: Suren Baghdasaryan <[email protected]>
Cc: Christopher Lameter <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Peter Enderborg <[email protected]>
Cc: Randy Dunlap <[email protected]>
Cc: Shakeel Butt <[email protected]>
Cc: Vinayak Menon <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
Documentation/accounting/psi.txt | 9 +++
Documentation/admin-guide/cgroup-v2.rst | 18 +++++
include/linux/cgroup-defs.h | 4 ++
include/linux/cgroup.h | 15 ++++
include/linux/psi.h | 25 +++++++
init/Kconfig | 4 ++
kernel/cgroup/cgroup.c | 45 +++++++++++-
kernel/sched/psi.c | 118 +++++++++++++++++++++++++++++---
8 files changed, 228 insertions(+), 10 deletions(-)
Comparison
==========
initial initial last penup first
good-v4.19 bad-92b419289cee good-eb414681 good-95f9ab2d bad-2ce7135a
Min 1 1.2980 ( 0.00%) 1.2950 ( 0.23%) 1.2870 ( 0.85%) 1.3100 ( -0.92%) 1.3190 ( -1.62%)
Min 3 3.8480 ( 0.00%) 4.0640 ( -5.61%) 3.8070 ( 1.07%) 3.8530 ( -0.13%) 4.1750 ( -8.50%)
Min 5 6.7420 ( 0.00%) 8.1480 ( -20.85%) 6.8010 ( -0.88%) 6.7740 ( -0.47%) 7.3370 ( -8.83%)
Min 7 9.7230 ( 0.00%) 10.2490 ( -5.41%) 9.7840 ( -0.63%) 9.8940 ( -1.76%) 10.6890 ( -9.94%)
Min 12 16.2810 ( 0.00%) 17.6470 ( -8.39%) 16.1740 ( 0.66%) 16.2390 ( 0.26%) 17.1260 ( -5.19%)
Min 18 25.2050 ( 0.00%) 25.6510 ( -1.77%) 25.4950 ( -1.15%) 25.4130 ( -0.83%) 25.5330 ( -1.30%)
Min 24 31.7720 ( 0.00%) 29.4380 ( 7.35%) 33.3800 ( -5.06%) 31.2400 ( 1.67%) 31.4640 ( 0.97%)
Min 30 37.0100 ( 0.00%) 37.6760 ( -1.80%) 33.7630 ( 8.77%) 37.5890 ( -1.56%) 33.2980 ( 10.03%)
Min 32 40.0150 ( 0.00%) 39.9460 ( 0.17%) 37.5790 ( 6.09%) 37.7580 ( 5.64%) 38.8230 ( 2.98%)
Amean 1 1.3153 ( 0.00%) 1.3530 ( -2.86%) 1.3133 ( 0.15%) 1.3217 ( -0.48%) 1.3243 ( -0.68%)
Amean 3 3.8670 ( 0.00%) 4.1097 * -6.28%* 3.8597 ( 0.19%) 3.8943 ( -0.71%) 4.2130 * -8.95%*
Amean 5 6.8563 ( 0.00%) 8.3113 * -21.22%* 6.9197 ( -0.92%) 6.8630 ( -0.10%) 7.5630 * -10.31%*
Amean 7 9.8650 ( 0.00%) 10.5607 * -7.05%* 9.9627 ( -0.99%) 9.9177 ( -0.53%) 10.7880 * -9.36%*
Amean 12 16.5540 ( 0.00%) 18.2850 * -10.46%* 16.7513 ( -1.19%) 16.3700 ( 1.11%) 18.1283 * -9.51%*
Amean 18 26.0390 ( 0.00%) 27.4333 ( -5.35%) 27.0223 ( -3.78%) 25.8067 ( 0.89%) 27.2333 ( -4.59%)
Amean 24 32.8650 ( 0.00%) 33.0093 ( -0.44%) 34.6167 ( -5.33%) 32.7020 ( 0.50%) 34.9150 ( -6.24%)
Amean 30 39.5653 ( 0.00%) 41.0643 ( -3.79%) 36.6733 ( 7.31%) 39.4050 ( 0.41%) 38.1273 ( 3.63%)
Amean 32 40.8177 ( 0.00%) 42.1383 ( -3.24%) 39.1487 ( 4.09%) 41.7743 ( -2.34%) 39.4193 * 3.43%*
Stddev 1 0.0158 ( 0.00%) 0.0730 (-361.51%) 0.0247 ( -56.14%) 0.0102 ( 35.44%) 0.0061 ( 61.38%)
Stddev 3 0.0271 ( 0.00%) 0.0397 ( -46.65%) 0.0627 (-131.53%) 0.0575 (-112.38%) 0.0330 ( -22.06%)
Stddev 5 0.1491 ( 0.00%) 0.1744 ( -16.95%) 0.1134 ( 23.94%) 0.0771 ( 48.30%) 0.2126 ( -42.57%)
Stddev 7 0.1650 ( 0.00%) 0.2780 ( -68.48%) 0.1663 ( -0.80%) 0.0257 ( 84.43%) 0.0889 ( 46.12%)
Stddev 12 0.2395 ( 0.00%) 0.5652 (-135.95%) 0.5726 (-139.03%) 0.1626 ( 32.11%) 0.8698 (-263.10%)
Stddev 18 0.8499 ( 0.00%) 1.5949 ( -87.65%) 1.3650 ( -60.60%) 0.3799 ( 55.31%) 1.5393 ( -81.11%)
Stddev 24 1.2105 ( 0.00%) 3.3274 (-174.88%) 1.5403 ( -27.25%) 1.2683 ( -4.78%) 3.1545 (-160.60%)
Stddev 30 2.4625 ( 0.00%) 2.9472 ( -19.69%) 2.5228 ( -2.45%) 1.5773 ( 35.95%) 5.3969 (-119.17%)
Stddev 32 0.9572 ( 0.00%) 2.4088 (-151.66%) 1.5982 ( -66.97%) 4.3986 (-359.54%) 0.5216 ( 45.50%)
CoeffVar 1 1.2029 ( 0.00%) 5.3969 (-348.67%) 1.8811 ( -56.38%) 0.7728 ( 35.75%) 0.4614 ( 61.64%)
CoeffVar 3 0.7001 ( 0.00%) 0.9661 ( -37.99%) 1.6241 (-131.97%) 1.4765 (-110.89%) 0.7844 ( -12.03%)
CoeffVar 5 2.1749 ( 0.00%) 2.0982 ( 3.53%) 1.6392 ( 24.63%) 1.1233 ( 48.35%) 2.8110 ( -29.25%)
CoeffVar 7 1.6725 ( 0.00%) 2.6322 ( -57.38%) 1.6694 ( 0.18%) 0.2591 ( 84.51%) 0.8241 ( 50.73%)
CoeffVar 12 1.4470 ( 0.00%) 3.0910 (-113.61%) 3.4180 (-136.21%) 0.9934 ( 31.35%) 4.7978 (-231.56%)
CoeffVar 18 3.2640 ( 0.00%) 5.8138 ( -78.12%) 5.0512 ( -54.75%) 1.4719 ( 54.91%) 5.6523 ( -73.17%)
CoeffVar 24 3.6832 ( 0.00%) 10.0801 (-173.68%) 4.4495 ( -20.81%) 3.8785 ( -5.30%) 9.0349 (-145.30%)
CoeffVar 30 6.2238 ( 0.00%) 7.1770 ( -15.32%) 6.8792 ( -10.53%) 4.0027 ( 35.69%) 14.1550 (-127.43%)
CoeffVar 32 2.3450 ( 0.00%) 5.7165 (-143.78%) 4.0825 ( -74.09%) 10.5295 (-349.02%) 1.3233 ( 43.57%)
Max 1 1.3290 ( 0.00%) 1.4350 ( -7.98%) 1.3360 ( -0.53%) 1.3290 ( 0.00%) 1.3310 ( -0.15%)
Max 3 3.8980 ( 0.00%) 4.1360 ( -6.11%) 3.9290 ( -0.80%) 3.9600 ( -1.59%) 4.2350 ( -8.65%)
Max 5 7.0250 ( 0.00%) 8.4950 ( -20.93%) 7.0270 ( -0.03%) 6.9090 ( 1.65%) 7.7590 ( -10.45%)
Max 7 10.0460 ( 0.00%) 10.7830 ( -7.34%) 10.1130 ( -0.67%) 9.9450 ( 1.01%) 10.8610 ( -8.11%)
Max 12 16.7290 ( 0.00%) 18.7230 ( -11.92%) 17.3190 ( -3.53%) 16.5520 ( 1.06%) 18.6840 ( -11.69%)
Max 18 26.9040 ( 0.00%) 28.7260 ( -6.77%) 28.1230 ( -4.53%) 26.1710 ( 2.72%) 28.5320 ( -6.05%)
Max 24 34.1660 ( 0.00%) 36.0220 ( -5.43%) 36.3420 ( -6.37%) 33.5080 ( 1.93%) 37.6500 ( -10.20%)
Max 30 41.9230 ( 0.00%) 43.0330 ( -2.65%) 38.2390 ( 8.79%) 40.4330 ( 3.55%) 43.9530 ( -4.84%)
Max 32 41.8770 ( 0.00%) 44.7170 ( -6.78%) 40.7740 ( 2.63%) 46.4750 ( -10.98%) 39.7910 ( 4.98%)
BAmean-50 1 1.3085 ( 0.00%) 1.3120 ( -0.27%) 1.3020 ( 0.50%) 1.3180 ( -0.73%) 1.3210 ( -0.96%)
BAmean-50 3 3.8515 ( 0.00%) 4.0965 ( -6.36%) 3.8250 ( 0.69%) 3.8615 ( -0.26%) 4.2020 ( -9.10%)
BAmean-50 5 6.7720 ( 0.00%) 8.2195 ( -21.37%) 6.8660 ( -1.39%) 6.8400 ( -1.00%) 7.4650 ( -10.23%)
BAmean-50 7 9.7745 ( 0.00%) 10.4495 ( -6.91%) 9.8875 ( -1.16%) 9.9040 ( -1.32%) 10.7515 ( -10.00%)
BAmean-50 12 16.4665 ( 0.00%) 18.0660 ( -9.71%) 16.4675 ( -0.01%) 16.2790 ( 1.14%) 17.8505 ( -8.40%)
BAmean-50 18 25.6065 ( 0.00%) 26.7870 ( -4.61%) 26.4720 ( -3.38%) 25.6245 ( -0.07%) 26.5840 ( -3.82%)
BAmean-50 24 32.2145 ( 0.00%) 31.5030 ( 2.21%) 33.7540 ( -4.78%) 32.2990 ( -0.26%) 33.5475 ( -4.14%)
BAmean-50 30 38.3865 ( 0.00%) 40.0800 ( -4.41%) 35.8905 ( 6.50%) 38.8910 ( -1.31%) 35.2145 ( 8.26%)
BAmean-50 32 40.2880 ( 0.00%) 40.8490 ( -1.39%) 38.3360 ( 4.85%) 39.4240 ( 2.14%) 39.2335 ( 2.62%)
BAmean-95 1 1.3153 ( 0.00%) 1.3530 ( -2.86%) 1.3133 ( 0.15%) 1.3217 ( -0.48%) 1.3243 ( -0.68%)
BAmean-95 3 3.8670 ( 0.00%) 4.1097 ( -6.28%) 3.8597 ( 0.19%) 3.8943 ( -0.71%) 4.2130 ( -8.95%)
BAmean-95 5 6.8563 ( 0.00%) 8.3113 ( -21.22%) 6.9197 ( -0.92%) 6.8630 ( -0.10%) 7.5630 ( -10.31%)
BAmean-95 7 9.8650 ( 0.00%) 10.5607 ( -7.05%) 9.9627 ( -0.99%) 9.9177 ( -0.53%) 10.7880 ( -9.36%)
BAmean-95 12 16.5540 ( 0.00%) 18.2850 ( -10.46%) 16.7513 ( -1.19%) 16.3700 ( 1.11%) 18.1283 ( -9.51%)
BAmean-95 18 26.0390 ( 0.00%) 27.4333 ( -5.35%) 27.0223 ( -3.78%) 25.8067 ( 0.89%) 27.2333 ( -4.59%)
BAmean-95 24 32.8650 ( 0.00%) 33.0093 ( -0.44%) 34.6167 ( -5.33%) 32.7020 ( 0.50%) 34.9150 ( -6.24%)
BAmean-95 30 39.5653 ( 0.00%) 41.0643 ( -3.79%) 36.6733 ( 7.31%) 39.4050 ( 0.41%) 38.1273 ( 3.63%)
BAmean-95 32 40.8177 ( 0.00%) 42.1383 ( -3.24%) 39.1487 ( 4.09%) 41.7743 ( -2.34%) 39.4193 ( 3.43%)
BAmean-99 1 1.3153 ( 0.00%) 1.3530 ( -2.86%) 1.3133 ( 0.15%) 1.3217 ( -0.48%) 1.3243 ( -0.68%)
BAmean-99 3 3.8670 ( 0.00%) 4.1097 ( -6.28%) 3.8597 ( 0.19%) 3.8943 ( -0.71%) 4.2130 ( -8.95%)
BAmean-99 5 6.8563 ( 0.00%) 8.3113 ( -21.22%) 6.9197 ( -0.92%) 6.8630 ( -0.10%) 7.5630 ( -10.31%)
BAmean-99 7 9.8650 ( 0.00%) 10.5607 ( -7.05%) 9.9627 ( -0.99%) 9.9177 ( -0.53%) 10.7880 ( -9.36%)
BAmean-99 12 16.5540 ( 0.00%) 18.2850 ( -10.46%) 16.7513 ( -1.19%) 16.3700 ( 1.11%) 18.1283 ( -9.51%)
BAmean-99 18 26.0390 ( 0.00%) 27.4333 ( -5.35%) 27.0223 ( -3.78%) 25.8067 ( 0.89%) 27.2333 ( -4.59%)
BAmean-99 24 32.8650 ( 0.00%) 33.0093 ( -0.44%) 34.6167 ( -5.33%) 32.7020 ( 0.50%) 34.9150 ( -6.24%)
BAmean-99 30 39.5653 ( 0.00%) 41.0643 ( -3.79%) 36.6733 ( 7.31%) 39.4050 ( 0.41%) 38.1273 ( 3.63%)
BAmean-99 32 40.8177 ( 0.00%) 42.1383 ( -3.24%) 39.1487 ( 4.09%) 41.7743 ( -2.34%) 39.4193 ( 3.43%)
Git log
=======
git bisect start
# good: [84df9525b0c27f3ebc2ebb1864fa62a97fdedb7d] Linux 4.19
git bisect good 84df9525b0c27f3ebc2ebb1864fa62a97fdedb7d
# bad: [92b419289ceecdd1eae03114928913f298b84327] Merge tag 'riscv-for-linus-4.20-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/palmer/riscv-linux
git bisect bad 92b419289ceecdd1eae03114928913f298b84327
# good: [84df9525b0c27f3ebc2ebb1864fa62a97fdedb7d] Linux 4.19
git bisect good 84df9525b0c27f3ebc2ebb1864fa62a97fdedb7d
# good: [26873acacbdbb4e4b444f5dd28dcc4853f0e8ba2] Merge tag 'driver-core-4.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
git bisect good 26873acacbdbb4e4b444f5dd28dcc4853f0e8ba2
# bad: [738b04fba18d35cd352b7b15afefb8a7b798648e] Merge tag 'staging-4.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging
git bisect bad 738b04fba18d35cd352b7b15afefb8a7b798648e
# good: [f2bfc71aee75feff33ca659322b72ffeed5a243d] Merge tag 'drm-intel-next-fixes-2018-10-18' of git://anongit.freedesktop.org/drm/drm-intel into drm-next
git bisect good f2bfc71aee75feff33ca659322b72ffeed5a243d
# bad: [345671ea0f9258f410eb057b9ced9cefbbe5dc78] Merge branch 'akpm' (patches from Andrew)
git bisect bad 345671ea0f9258f410eb057b9ced9cefbbe5dc78
# good: [033078a9afe504ac9e615d10c4b35d634450b637] Merge tag '4.20-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6
git bisect good 033078a9afe504ac9e615d10c4b35d634450b637
# good: [685f7e4f161425b137056abe35ba8ef7b669d83d] Merge tag 'powerpc-4.20-1' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux
git bisect good 685f7e4f161425b137056abe35ba8ef7b669d83d
# bad: [d3035be4ce2345d98633a45f93a74e526e94b802] mm: calculate deferred pages after skipping mirrored memory
git bisect bad d3035be4ce2345d98633a45f93a74e526e94b802
# good: [95f9ab2d596e8cbb388315e78c82b9a131bf2928] mm: workingset: don't drop refault information prematurely
git bisect good 95f9ab2d596e8cbb388315e78c82b9a131bf2928
# bad: [f682a97a00591def7cefbb5003dc04045028e405] mm: provide kernel parameter to allow disabling page init poisoning
git bisect bad f682a97a00591def7cefbb5003dc04045028e405
# bad: [c3df29d13044d885695067fa0b1386824942557a] mm/swap.c: remove duplicated include
git bisect bad c3df29d13044d885695067fa0b1386824942557a
# good: [eb414681d5a07d28d2ff90dc05f69ec6b232ebd2] psi: pressure stall information for CPU, memory, and IO
git bisect good eb414681d5a07d28d2ff90dc05f69ec6b232ebd2
# bad: [68d48e6a2df575b935edd420396c3cb8b6aa6ad3] mm: workingset: add vmstat counter for shadow nodes
git bisect bad 68d48e6a2df575b935edd420396c3cb8b6aa6ad3
# bad: [505802a53510e54ad5fbbd655a68893df83bfb91] mm: workingset: use cheaper __inc_lruvec_state in irqsafe node reclaim
git bisect bad 505802a53510e54ad5fbbd655a68893df83bfb91
# bad: [2ce7135adc9ad081aa3c49744144376ac74fea60] psi: cgroup support
git bisect bad 2ce7135adc9ad081aa3c49744144376ac74fea60
# first bad commit: [2ce7135adc9ad081aa3c49744144376ac74fea60] psi: cgroup support
--
Mel Gorman
SUSE Labs
Hi Mel,
On Mon, Nov 26, 2018 at 01:34:20PM +0000, Mel Gorman wrote:
> Hi Johannes,
>
> PSI is a great idea but it does have overhead and if enabled by Kconfig
> then it incurs a hit whether the user is aware of the feature or not. I
> think enabling by default is unnecessary as it should only be enabled if
> the information is being consumed. While the Kconfig exists, it's all or
> nothing if distributions want to have the feature available.
Yes, let's make this easier to pick and choose. Obviously I'd rather
you shipped it default-disabled than not at all.
> I've included a bisection report below showing a 6-10% regression on a
> single socket skylake machine. Would you mind doing one or all of the
> following to fix it please?
>
> a) disable it by default
> b) put psi_disable behind a static branch to move the overhead to zero
> if it's disabled
> c) optionally enable/disable at runtime (least important as at a glance,
> this may be problematic)
For a) I'd suggest we do what we do in other places that face this
vendor kernel trade-off (NUMA balancing comes to mind): one option to
build the feature, one option to set whether the default is on or off.
And b) is pretty straight-forward, let's do that too.
c) is not possible, as we need the complete task counts to calculate
pressure, and maintaining those counts are where the sched cost is.
> Last good/First bad commit
> ==========================
> Last good commit: eb414681d5a07d28d2ff90dc05f69ec6b232ebd2
> First bad commit: 2ce7135adc9ad081aa3c49744144376ac74fea60
> From 2ce7135adc9ad081aa3c49744144376ac74fea60 Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <[email protected]>
> Date: Fri, 26 Oct 2018 15:06:31 -0700
> Subject: [PATCH] psi: cgroup support
> On a system that executes multiple cgrouped jobs and independent
> workloads, we don't just care about the health of the overall system, but
> also that of individual jobs, so that we can ensure individual job health,
> fairness between jobs, or prioritize some jobs over others.
> This patch implements pressure stall tracking for cgroups. In kernels
> with CONFIG_PSI=y, cgroup2 groups will have cpu.pressure, memory.pressure,
> and io.pressure files that track aggregate pressure stall times for only
> the tasks inside the cgroup.
It's curious that the cgroup support patch is the offender, not the
psi patch itself (that adds some cost as per the hackbench results,
but not as much). What kind of cgroup setup does this code run in?
Anyway, how about the following?
From 6ae33455b8083fc9f5d5fbfe971f70253b0dbacd Mon Sep 17 00:00:00 2001
From: Johannes Weiner <[email protected]>
Date: Mon, 26 Nov 2018 09:39:23 -0500
Subject: [PATCH] psi: make disabling/enabling easier for vendor kernels
Mel Gorman reports a hackbench regression with psi that would prohibit
shipping the suse kernel with it default-enabled, but he'd still like
users to be able to opt in at little to no cost to others.
With the current combination of CONFIG_PSI and the psi_disabled bool
set from the commandline, this is a challenge. Do the following things
to make it easier:
1. Add a config option CONFIG_PSI_DEFAULT_ENABLED that allows distros
to enable CONFIG_PSI in their kernel, but leaving the feature
disabled unless a user requests it at boot-time.
To avoid double negatives, rename psi_disabled= to psi_enable=.
2. Make psi_disabled a static branch to eliminate any branch costs
when the feature is disabled.
Reported-by: Mel Gorman <[email protected]>
Signed-off-by: Johannes Weiner <[email protected]>
---
include/linux/psi.h | 3 ++-
init/Kconfig | 9 +++++++++
kernel/sched/psi.c | 30 +++++++++++++++++++++---------
kernel/sched/stats.h | 8 ++++----
4 files changed, 36 insertions(+), 14 deletions(-)
diff --git a/include/linux/psi.h b/include/linux/psi.h
index 8e0725aac0aa..7006008d5b72 100644
--- a/include/linux/psi.h
+++ b/include/linux/psi.h
@@ -1,6 +1,7 @@
#ifndef _LINUX_PSI_H
#define _LINUX_PSI_H
+#include <linux/jump_label.h>
#include <linux/psi_types.h>
#include <linux/sched.h>
@@ -9,7 +10,7 @@ struct css_set;
#ifdef CONFIG_PSI
-extern bool psi_disabled;
+extern struct static_key_false psi_disabled;
void psi_init(void);
diff --git a/init/Kconfig b/init/Kconfig
index a4112e95724a..cf5b5a0dcbc2 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -509,6 +509,15 @@ config PSI
Say N if unsure.
+config PSI_DEFAULT_DISABLED
+ bool "Require boot parameter to enable pressure stall information tracking"
+ default n
+ depends on PSI
+ help
+ If set, pressure stall information tracking will be disabled
+ per default but can be enabled through passing psi_enable=1
+ on the kernel commandline during boot.
+
endmenu # "CPU/Task time and stats accounting"
config CPU_ISOLATION
diff --git a/kernel/sched/psi.c b/kernel/sched/psi.c
index 3d7355d7c3e3..9da0af3cd898 100644
--- a/kernel/sched/psi.c
+++ b/kernel/sched/psi.c
@@ -136,8 +136,18 @@
static int psi_bug __read_mostly;
-bool psi_disabled __read_mostly;
-core_param(psi_disabled, psi_disabled, bool, 0644);
+DEFINE_STATIC_KEY_FALSE(psi_disabled);
+
+#ifdef CONFIG_PSI_DEFAULT_DISABLED
+bool psi_enable;
+#else
+bool psi_enable = true;
+#endif
+static int __init parse_psi_enable(char *str)
+{
+ return kstrtobool(str, &psi_enable) == 0;
+}
+__setup("psi_enable=", parse_psi_enable);
/* Running averages - we need to be higher-res than loadavg */
#define PSI_FREQ (2*HZ+1) /* 2 sec intervals */
@@ -169,8 +179,10 @@ static void group_init(struct psi_group *group)
void __init psi_init(void)
{
- if (psi_disabled)
+ if (!psi_enable) {
+ static_branch_enable(&psi_disabled);
return;
+ }
psi_period = jiffies_to_nsecs(PSI_FREQ);
group_init(&psi_system);
@@ -549,7 +561,7 @@ void psi_memstall_enter(unsigned long *flags)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
*flags = current->flags & PF_MEMSTALL;
@@ -579,7 +591,7 @@ void psi_memstall_leave(unsigned long *flags)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (*flags)
@@ -600,7 +612,7 @@ void psi_memstall_leave(unsigned long *flags)
#ifdef CONFIG_CGROUPS
int psi_cgroup_alloc(struct cgroup *cgroup)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return 0;
cgroup->psi.pcpu = alloc_percpu(struct psi_group_cpu);
@@ -612,7 +624,7 @@ int psi_cgroup_alloc(struct cgroup *cgroup)
void psi_cgroup_free(struct cgroup *cgroup)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
cancel_delayed_work_sync(&cgroup->psi.clock_work);
@@ -637,7 +649,7 @@ void cgroup_move_task(struct task_struct *task, struct css_set *to)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled) {
+ if (static_branch_likely(&psi_disabled)) {
/*
* Lame to do this here, but the scheduler cannot be locked
* from the outside, so we move cgroups from inside sched/.
@@ -673,7 +685,7 @@ int psi_show(struct seq_file *m, struct psi_group *group, enum psi_res res)
{
int full;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return -EOPNOTSUPP;
update_stats(group);
diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h
index 4904c4677000..aa0de240fb41 100644
--- a/kernel/sched/stats.h
+++ b/kernel/sched/stats.h
@@ -66,7 +66,7 @@ static inline void psi_enqueue(struct task_struct *p, bool wakeup)
{
int clear = 0, set = TSK_RUNNING;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (!wakeup || p->sched_psi_wake_requeue) {
@@ -86,7 +86,7 @@ static inline void psi_dequeue(struct task_struct *p, bool sleep)
{
int clear = TSK_RUNNING, set = 0;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (!sleep) {
@@ -102,7 +102,7 @@ static inline void psi_dequeue(struct task_struct *p, bool sleep)
static inline void psi_ttwu_dequeue(struct task_struct *p)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
/*
* Is the task being migrated during a wakeup? Make sure to
@@ -128,7 +128,7 @@ static inline void psi_ttwu_dequeue(struct task_struct *p)
static inline void psi_task_tick(struct rq *rq)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (unlikely(rq->curr->flags & PF_MEMSTALL))
--
2.19.1
On Mon, Nov 26, 2018 at 11:07:24AM -0500, Johannes Weiner wrote:
> Hi Mel,
>
> On Mon, Nov 26, 2018 at 01:34:20PM +0000, Mel Gorman wrote:
> > Hi Johannes,
> >
> > PSI is a great idea but it does have overhead and if enabled by Kconfig
> > then it incurs a hit whether the user is aware of the feature or not. I
> > think enabling by default is unnecessary as it should only be enabled if
> > the information is being consumed. While the Kconfig exists, it's all or
> > nothing if distributions want to have the feature available.
>
> Yes, let's make this easier to pick and choose. Obviously I'd rather
> you shipped it default-disabled than not at all.
>
Indeed.
> > I've included a bisection report below showing a 6-10% regression on a
> > single socket skylake machine. Would you mind doing one or all of the
> > following to fix it please?
> >
> > a) disable it by default
> > b) put psi_disable behind a static branch to move the overhead to zero
> > if it's disabled
> > c) optionally enable/disable at runtime (least important as at a glance,
> > this may be problematic)
>
> For a) I'd suggest we do what we do in other places that face this
> vendor kernel trade-off (NUMA balancing comes to mind): one option to
> build the feature, one option to set whether the default is on or off.
>
That would be fine and makes sense.
> And b) is pretty straight-forward, let's do that too.
>
Thanks.
> c) is not possible, as we need the complete task counts to calculate
> pressure, and maintaining those counts are where the sched cost is.
>
I figured that this would be the case.
> > Last good/First bad commit
> > ==========================
> > Last good commit: eb414681d5a07d28d2ff90dc05f69ec6b232ebd2
> > First bad commit: 2ce7135adc9ad081aa3c49744144376ac74fea60
> > From 2ce7135adc9ad081aa3c49744144376ac74fea60 Mon Sep 17 00:00:00 2001
> > From: Johannes Weiner <[email protected]>
> > Date: Fri, 26 Oct 2018 15:06:31 -0700
> > Subject: [PATCH] psi: cgroup support
> > On a system that executes multiple cgrouped jobs and independent
> > workloads, we don't just care about the health of the overall system, but
> > also that of individual jobs, so that we can ensure individual job health,
> > fairness between jobs, or prioritize some jobs over others.
> > This patch implements pressure stall tracking for cgroups. In kernels
> > with CONFIG_PSI=y, cgroup2 groups will have cpu.pressure, memory.pressure,
> > and io.pressure files that track aggregate pressure stall times for only
> > the tasks inside the cgroup.
>
> It's curious that the cgroup support patch is the offender, not the
> psi patch itself (that adds some cost as per the hackbench results,
> but not as much). What kind of cgroup setup does this code run in?
>
No cgroup is setup but given that it is an automatic bisection, it's not
very unusual for it to get "close" but not get it exactly right.
> Anyway, how about the following?
>
I've queued it up setting CONFIG_PSI_DEFAULT_DISABLED in the Kconfig.
> <SNIP>
>
> diff --git a/init/Kconfig b/init/Kconfig
> index a4112e95724a..cf5b5a0dcbc2 100644
> --- a/init/Kconfig
> +++ b/init/Kconfig
> @@ -509,6 +509,15 @@ config PSI
>
> Say N if unsure.
>
> +config PSI_DEFAULT_DISABLED
> + bool "Require boot parameter to enable pressure stall information tracking"
> + default n
> + depends on PSI
> + help
> + If set, pressure stall information tracking will be disabled
> + per default but can be enabled through passing psi_enable=1
> + on the kernel commandline during boot.
> +
> endmenu # "CPU/Task time and stats accounting"
>
Should this default y on the basis that someone only wants the feature if
they are aware of it? This is not that important as CONFIG_PSI is disabled
by default and it's up to distribution maintainers to use their brain.
> config CPU_ISOLATION
> diff --git a/kernel/sched/psi.c b/kernel/sched/psi.c
> index 3d7355d7c3e3..9da0af3cd898 100644
> --- a/kernel/sched/psi.c
> +++ b/kernel/sched/psi.c
> @@ -136,8 +136,18 @@
>
> static int psi_bug __read_mostly;
>
> -bool psi_disabled __read_mostly;
> -core_param(psi_disabled, psi_disabled, bool, 0644);
> +DEFINE_STATIC_KEY_FALSE(psi_disabled);
> +
> +#ifdef CONFIG_PSI_DEFAULT_DISABLED
> +bool psi_enable;
> +#else
> +bool psi_enable = true;
> +#endif
> +static int __init parse_psi_enable(char *str)
> +{
> + return kstrtobool(str, &psi_enable) == 0;
> +}
> +__setup("psi_enable=", parse_psi_enable);
>
Bit late to notice but this switch should be in
Documentation/admin-guide/kernel-parameters.txt. If you really want to
match the automatic numa balancing switch then it also should be
psi=[enable|disable] instead of psi_enable=[1|0]
--
Mel Gorman
SUSE Labs
On Mon, Nov 26, 2018 at 04:54:47PM +0000, Mel Gorman wrote:
> On Mon, Nov 26, 2018 at 11:07:24AM -0500, Johannes Weiner wrote:
> > @@ -509,6 +509,15 @@ config PSI
> >
> > Say N if unsure.
> >
> > +config PSI_DEFAULT_DISABLED
> > + bool "Require boot parameter to enable pressure stall information tracking"
> > + default n
> > + depends on PSI
> > + help
> > + If set, pressure stall information tracking will be disabled
> > + per default but can be enabled through passing psi_enable=1
> > + on the kernel commandline during boot.
> > +
> > endmenu # "CPU/Task time and stats accounting"
> >
>
> Should this default y on the basis that someone only wants the feature if
> they are aware of it? This is not that important as CONFIG_PSI is disabled
> by default and it's up to distribution maintainers to use their brain.
I went with the NUMA balancing example again here, which defaults to
enabling the feature at boot time. IMO that makes sense, as somebody
would presumably first read through the PSI help text, then decide y
on that before being asked the second question. A "yes, but
<stipulations>" for vendor kernels seems more appropriate than
requiring a double yes for other users to simply get the feature.
> > @@ -136,8 +136,18 @@
> >
> > static int psi_bug __read_mostly;
> >
> > -bool psi_disabled __read_mostly;
> > -core_param(psi_disabled, psi_disabled, bool, 0644);
> > +DEFINE_STATIC_KEY_FALSE(psi_disabled);
> > +
> > +#ifdef CONFIG_PSI_DEFAULT_DISABLED
> > +bool psi_enable;
> > +#else
> > +bool psi_enable = true;
> > +#endif
> > +static int __init parse_psi_enable(char *str)
> > +{
> > + return kstrtobool(str, &psi_enable) == 0;
> > +}
> > +__setup("psi_enable=", parse_psi_enable);
> >
>
> Bit late to notice but this switch should be in
> Documentation/admin-guide/kernel-parameters.txt. If you really want to
> match the automatic numa balancing switch then it also should be
> psi=[enable|disable] instead of psi_enable=[1|0]
Done and done, thanks. Updated patch:
---
From 694ae3d82da06ca57bfaa775140941ac5e8f2e37 Mon Sep 17 00:00:00 2001
From: Johannes Weiner <[email protected]>
Date: Mon, 26 Nov 2018 09:39:23 -0500
Subject: [PATCH] psi: make disabling/enabling easier for vendor kernels
Mel Gorman reports a hackbench regression with psi that would prohibit
shipping the suse kernel with it default-enabled, but he'd still like
users to be able to opt in at little to no cost to others.
With the current combination of CONFIG_PSI and the psi_disabled bool
set from the commandline, this is a challenge. Do the following things
to make it easier:
1. Add a config option CONFIG_PSI_DEFAULT_ENABLED that allows distros
to enable CONFIG_PSI in their kernel, but leaving the feature
disabled unless a user requests it at boot-time.
To avoid double negatives, rename psi_disabled= to psi=.
2. Make psi_disabled a static branch to eliminate any branch costs
when the feature is disabled.
Reported-by: Mel Gorman <[email protected]>
Signed-off-by: Johannes Weiner <[email protected]>
---
.../admin-guide/kernel-parameters.txt | 4 +++
include/linux/psi.h | 3 +-
init/Kconfig | 9 ++++++
kernel/sched/psi.c | 30 +++++++++++++------
kernel/sched/stats.h | 8 ++---
5 files changed, 40 insertions(+), 14 deletions(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 19f4423e70d9..8760a343c6d8 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -3504,6 +3504,10 @@
before loading.
See Documentation/blockdev/ramdisk.txt.
+ psi= [KNL] Enable or disable pressure stall information
+ tracking.
+ Format: <bool>
+
psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
probe for; one of (bare|imps|exps|lifebook|any).
psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
diff --git a/include/linux/psi.h b/include/linux/psi.h
index 8e0725aac0aa..7006008d5b72 100644
--- a/include/linux/psi.h
+++ b/include/linux/psi.h
@@ -1,6 +1,7 @@
#ifndef _LINUX_PSI_H
#define _LINUX_PSI_H
+#include <linux/jump_label.h>
#include <linux/psi_types.h>
#include <linux/sched.h>
@@ -9,7 +10,7 @@ struct css_set;
#ifdef CONFIG_PSI
-extern bool psi_disabled;
+extern struct static_key_false psi_disabled;
void psi_init(void);
diff --git a/init/Kconfig b/init/Kconfig
index a4112e95724a..cf5b5a0dcbc2 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -509,6 +509,15 @@ config PSI
Say N if unsure.
+config PSI_DEFAULT_DISABLED
+ bool "Require boot parameter to enable pressure stall information tracking"
+ default n
+ depends on PSI
+ help
+ If set, pressure stall information tracking will be disabled
+ per default but can be enabled through passing psi_enable=1
+ on the kernel commandline during boot.
+
endmenu # "CPU/Task time and stats accounting"
config CPU_ISOLATION
diff --git a/kernel/sched/psi.c b/kernel/sched/psi.c
index 3d7355d7c3e3..fe24de3fbc93 100644
--- a/kernel/sched/psi.c
+++ b/kernel/sched/psi.c
@@ -136,8 +136,18 @@
static int psi_bug __read_mostly;
-bool psi_disabled __read_mostly;
-core_param(psi_disabled, psi_disabled, bool, 0644);
+DEFINE_STATIC_KEY_FALSE(psi_disabled);
+
+#ifdef CONFIG_PSI_DEFAULT_DISABLED
+bool psi_enable;
+#else
+bool psi_enable = true;
+#endif
+static int __init setup_psi(char *str)
+{
+ return kstrtobool(str, &psi_enable) == 0;
+}
+__setup("psi=", setup_psi);
/* Running averages - we need to be higher-res than loadavg */
#define PSI_FREQ (2*HZ+1) /* 2 sec intervals */
@@ -169,8 +179,10 @@ static void group_init(struct psi_group *group)
void __init psi_init(void)
{
- if (psi_disabled)
+ if (!psi_enable) {
+ static_branch_enable(&psi_disabled);
return;
+ }
psi_period = jiffies_to_nsecs(PSI_FREQ);
group_init(&psi_system);
@@ -549,7 +561,7 @@ void psi_memstall_enter(unsigned long *flags)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
*flags = current->flags & PF_MEMSTALL;
@@ -579,7 +591,7 @@ void psi_memstall_leave(unsigned long *flags)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (*flags)
@@ -600,7 +612,7 @@ void psi_memstall_leave(unsigned long *flags)
#ifdef CONFIG_CGROUPS
int psi_cgroup_alloc(struct cgroup *cgroup)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return 0;
cgroup->psi.pcpu = alloc_percpu(struct psi_group_cpu);
@@ -612,7 +624,7 @@ int psi_cgroup_alloc(struct cgroup *cgroup)
void psi_cgroup_free(struct cgroup *cgroup)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
cancel_delayed_work_sync(&cgroup->psi.clock_work);
@@ -637,7 +649,7 @@ void cgroup_move_task(struct task_struct *task, struct css_set *to)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled) {
+ if (static_branch_likely(&psi_disabled)) {
/*
* Lame to do this here, but the scheduler cannot be locked
* from the outside, so we move cgroups from inside sched/.
@@ -673,7 +685,7 @@ int psi_show(struct seq_file *m, struct psi_group *group, enum psi_res res)
{
int full;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return -EOPNOTSUPP;
update_stats(group);
diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h
index 4904c4677000..aa0de240fb41 100644
--- a/kernel/sched/stats.h
+++ b/kernel/sched/stats.h
@@ -66,7 +66,7 @@ static inline void psi_enqueue(struct task_struct *p, bool wakeup)
{
int clear = 0, set = TSK_RUNNING;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (!wakeup || p->sched_psi_wake_requeue) {
@@ -86,7 +86,7 @@ static inline void psi_dequeue(struct task_struct *p, bool sleep)
{
int clear = TSK_RUNNING, set = 0;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (!sleep) {
@@ -102,7 +102,7 @@ static inline void psi_dequeue(struct task_struct *p, bool sleep)
static inline void psi_ttwu_dequeue(struct task_struct *p)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
/*
* Is the task being migrated during a wakeup? Make sure to
@@ -128,7 +128,7 @@ static inline void psi_ttwu_dequeue(struct task_struct *p)
static inline void psi_task_tick(struct rq *rq)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (unlikely(rq->curr->flags & PF_MEMSTALL))
--
2.19.1
On Mon, Nov 26, 2018 at 12:32:18PM -0500, Johannes Weiner wrote:
> On Mon, Nov 26, 2018 at 04:54:47PM +0000, Mel Gorman wrote:
> > On Mon, Nov 26, 2018 at 11:07:24AM -0500, Johannes Weiner wrote:
> > > @@ -509,6 +509,15 @@ config PSI
> > >
> > > Say N if unsure.
> > >
> > > +config PSI_DEFAULT_DISABLED
> > > + bool "Require boot parameter to enable pressure stall information tracking"
> > > + default n
> > > + depends on PSI
> > > + help
> > > + If set, pressure stall information tracking will be disabled
> > > + per default but can be enabled through passing psi_enable=1
> > > + on the kernel commandline during boot.
> > > +
> > > endmenu # "CPU/Task time and stats accounting"
> > >
> >
> > Should this default y on the basis that someone only wants the feature if
> > they are aware of it? This is not that important as CONFIG_PSI is disabled
> > by default and it's up to distribution maintainers to use their brain.
>
> I went with the NUMA balancing example again here, which defaults to
> enabling the feature at boot time. IMO that makes sense, as somebody
> would presumably first read through the PSI help text, then decide y
> on that before being asked the second question. A "yes, but
> <stipulations>" for vendor kernels seems more appropriate than
> requiring a double yes for other users to simply get the feature.
>
That's fair enough. The original NUMA balancing thinking was that it
should be enabled because there is a reasonable expectation that it
would improve performance regardless of user awareness. PSI is not
necessarily the same as it requires a consumer but I accept that a
distro maintainer should read the Kconfig text and figure it out.
I'll make sure the updated version gets tested, thanks.
--
Mel Gorman
SUSE Labs
On Mon, Nov 26, 2018 at 12:32:18PM -0500, Johannes Weiner wrote:
> >
> > Bit late to notice but this switch should be in
> > Documentation/admin-guide/kernel-parameters.txt. If you really want to
> > match the automatic numa balancing switch then it also should be
> > psi=[enable|disable] instead of psi_enable=[1|0]
>
> Done and done, thanks. Updated patch:
>
The following is a comparision using CONFIG_PSI=n as a baseline against
your patch and a vanilla kernel
4.20.0-rc4 4.20.0-rc4 4.20.0-rc4
kconfigdisable-v1r1 vanilla psidisable-v1r1
Amean 1 1.3100 ( 0.00%) 1.3923 ( -6.28%) 1.3427 ( -2.49%)
Amean 3 3.8860 ( 0.00%) 4.1230 * -6.10%* 3.8860 ( -0.00%)
Amean 5 6.8847 ( 0.00%) 8.0390 * -16.77%* 6.7727 ( 1.63%)
Amean 7 9.9310 ( 0.00%) 10.8367 * -9.12%* 9.9910 ( -0.60%)
Amean 12 16.6577 ( 0.00%) 18.2363 * -9.48%* 17.1083 ( -2.71%)
Amean 18 26.5133 ( 0.00%) 27.8833 * -5.17%* 25.7663 ( 2.82%)
Amean 24 34.3003 ( 0.00%) 34.6830 ( -1.12%) 32.0450 ( 6.58%)
Amean 30 40.0063 ( 0.00%) 40.5800 ( -1.43%) 41.5087 ( -3.76%)
Amean 32 40.1407 ( 0.00%) 41.2273 ( -2.71%) 39.9417 ( 0.50%)
It's showing that the vanilla kernel takes a hit (as the bisection
indicated it would) and that disabling PSI by default is reasonably
close in terms of performance for this particular workload on this
particular machine so;
Tested-by: Mel Gorman <[email protected]>
Thanks!
--
Mel Gorman
SUSE Labs
On Tue, Nov 27, 2018 at 11:46:17AM -0500, Johannes Weiner wrote:
> From 347b69a52d1ec7e71df1108cbc5703d6dd0616ba Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <[email protected]>
> Date: Mon, 26 Nov 2018 09:39:23 -0500
> Subject: [PATCH] psi: make disabling/enabling easier for vendor kernels
>
> Mel Gorman reports a hackbench regression with psi that would prohibit
> shipping the suse kernel with it default-enabled, but he'd still like
> users to be able to opt in at little to no cost to others.
>
> With the current combination of CONFIG_PSI and the psi_disabled bool
> set from the commandline, this is a challenge. Do the following things
> to make it easier:
>
> 1. Add a config option CONFIG_PSI_DEFAULT_ENABLED that allows distros
This should be: CONFIG_PSI_DEFAULT_DISABLED
Refreshed the patch.
---
From dc1511101b4c6aa3f7cfb91d8634a682fe7c147e Mon Sep 17 00:00:00 2001
From: Johannes Weiner <[email protected]>
Date: Mon, 26 Nov 2018 09:39:23 -0500
Subject: [PATCH] psi: make disabling/enabling easier for vendor kernels
Mel Gorman reports a hackbench regression with psi that would prohibit
shipping the suse kernel with it default-enabled, but he'd still like
users to be able to opt in at little to no cost to others.
With the current combination of CONFIG_PSI and the psi_disabled bool
set from the commandline, this is a challenge. Do the following things
to make it easier:
1. Add a config option CONFIG_PSI_DEFAULT_DISABLED that allows distros
to enable CONFIG_PSI in their kernel but leave the feature disabled
unless a user requests it at boot-time.
To avoid double negatives, rename psi_disabled= to psi=.
2. Make psi_disabled a static branch to eliminate any branch costs
when the feature is disabled.
In terms of numbers before and after this patch, Mel says:
: The following is a comparision using CONFIG_PSI=n as a baseline against
: your patch and a vanilla kernel
:
: 4.20.0-rc4 4.20.0-rc4 4.20.0-rc4
: kconfigdisable-v1r1 vanilla psidisable-v1r1
: Amean 1 1.3100 ( 0.00%) 1.3923 ( -6.28%) 1.3427 ( -2.49%)
: Amean 3 3.8860 ( 0.00%) 4.1230 * -6.10%* 3.8860 ( -0.00%)
: Amean 5 6.8847 ( 0.00%) 8.0390 * -16.77%* 6.7727 ( 1.63%)
: Amean 7 9.9310 ( 0.00%) 10.8367 * -9.12%* 9.9910 ( -0.60%)
: Amean 12 16.6577 ( 0.00%) 18.2363 * -9.48%* 17.1083 ( -2.71%)
: Amean 18 26.5133 ( 0.00%) 27.8833 * -5.17%* 25.7663 ( 2.82%)
: Amean 24 34.3003 ( 0.00%) 34.6830 ( -1.12%) 32.0450 ( 6.58%)
: Amean 30 40.0063 ( 0.00%) 40.5800 ( -1.43%) 41.5087 ( -3.76%)
: Amean 32 40.1407 ( 0.00%) 41.2273 ( -2.71%) 39.9417 ( 0.50%)
:
: It's showing that the vanilla kernel takes a hit (as the bisection
: indicated it would) and that disabling PSI by default is reasonably
: close in terms of performance for this particular workload on this
: particular machine so;
Tested-by: Mel Gorman <[email protected]>
Reported-by: Mel Gorman <[email protected]>
Signed-off-by: Johannes Weiner <[email protected]>
---
.../admin-guide/kernel-parameters.txt | 4 +++
include/linux/psi.h | 3 +-
init/Kconfig | 9 ++++++
kernel/sched/psi.c | 30 +++++++++++++------
kernel/sched/stats.h | 8 ++---
5 files changed, 40 insertions(+), 14 deletions(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 19f4423e70d9..8760a343c6d8 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -3504,6 +3504,10 @@
before loading.
See Documentation/blockdev/ramdisk.txt.
+ psi= [KNL] Enable or disable pressure stall information
+ tracking.
+ Format: <bool>
+
psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
probe for; one of (bare|imps|exps|lifebook|any).
psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
diff --git a/include/linux/psi.h b/include/linux/psi.h
index 8e0725aac0aa..7006008d5b72 100644
--- a/include/linux/psi.h
+++ b/include/linux/psi.h
@@ -1,6 +1,7 @@
#ifndef _LINUX_PSI_H
#define _LINUX_PSI_H
+#include <linux/jump_label.h>
#include <linux/psi_types.h>
#include <linux/sched.h>
@@ -9,7 +10,7 @@ struct css_set;
#ifdef CONFIG_PSI
-extern bool psi_disabled;
+extern struct static_key_false psi_disabled;
void psi_init(void);
diff --git a/init/Kconfig b/init/Kconfig
index a4112e95724a..cf5b5a0dcbc2 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -509,6 +509,15 @@ config PSI
Say N if unsure.
+config PSI_DEFAULT_DISABLED
+ bool "Require boot parameter to enable pressure stall information tracking"
+ default n
+ depends on PSI
+ help
+ If set, pressure stall information tracking will be disabled
+ per default but can be enabled through passing psi_enable=1
+ on the kernel commandline during boot.
+
endmenu # "CPU/Task time and stats accounting"
config CPU_ISOLATION
diff --git a/kernel/sched/psi.c b/kernel/sched/psi.c
index 3d7355d7c3e3..fe24de3fbc93 100644
--- a/kernel/sched/psi.c
+++ b/kernel/sched/psi.c
@@ -136,8 +136,18 @@
static int psi_bug __read_mostly;
-bool psi_disabled __read_mostly;
-core_param(psi_disabled, psi_disabled, bool, 0644);
+DEFINE_STATIC_KEY_FALSE(psi_disabled);
+
+#ifdef CONFIG_PSI_DEFAULT_DISABLED
+bool psi_enable;
+#else
+bool psi_enable = true;
+#endif
+static int __init setup_psi(char *str)
+{
+ return kstrtobool(str, &psi_enable) == 0;
+}
+__setup("psi=", setup_psi);
/* Running averages - we need to be higher-res than loadavg */
#define PSI_FREQ (2*HZ+1) /* 2 sec intervals */
@@ -169,8 +179,10 @@ static void group_init(struct psi_group *group)
void __init psi_init(void)
{
- if (psi_disabled)
+ if (!psi_enable) {
+ static_branch_enable(&psi_disabled);
return;
+ }
psi_period = jiffies_to_nsecs(PSI_FREQ);
group_init(&psi_system);
@@ -549,7 +561,7 @@ void psi_memstall_enter(unsigned long *flags)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
*flags = current->flags & PF_MEMSTALL;
@@ -579,7 +591,7 @@ void psi_memstall_leave(unsigned long *flags)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (*flags)
@@ -600,7 +612,7 @@ void psi_memstall_leave(unsigned long *flags)
#ifdef CONFIG_CGROUPS
int psi_cgroup_alloc(struct cgroup *cgroup)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return 0;
cgroup->psi.pcpu = alloc_percpu(struct psi_group_cpu);
@@ -612,7 +624,7 @@ int psi_cgroup_alloc(struct cgroup *cgroup)
void psi_cgroup_free(struct cgroup *cgroup)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
cancel_delayed_work_sync(&cgroup->psi.clock_work);
@@ -637,7 +649,7 @@ void cgroup_move_task(struct task_struct *task, struct css_set *to)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled) {
+ if (static_branch_likely(&psi_disabled)) {
/*
* Lame to do this here, but the scheduler cannot be locked
* from the outside, so we move cgroups from inside sched/.
@@ -673,7 +685,7 @@ int psi_show(struct seq_file *m, struct psi_group *group, enum psi_res res)
{
int full;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return -EOPNOTSUPP;
update_stats(group);
diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h
index 4904c4677000..aa0de240fb41 100644
--- a/kernel/sched/stats.h
+++ b/kernel/sched/stats.h
@@ -66,7 +66,7 @@ static inline void psi_enqueue(struct task_struct *p, bool wakeup)
{
int clear = 0, set = TSK_RUNNING;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (!wakeup || p->sched_psi_wake_requeue) {
@@ -86,7 +86,7 @@ static inline void psi_dequeue(struct task_struct *p, bool sleep)
{
int clear = TSK_RUNNING, set = 0;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (!sleep) {
@@ -102,7 +102,7 @@ static inline void psi_dequeue(struct task_struct *p, bool sleep)
static inline void psi_ttwu_dequeue(struct task_struct *p)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
/*
* Is the task being migrated during a wakeup? Make sure to
@@ -128,7 +128,7 @@ static inline void psi_ttwu_dequeue(struct task_struct *p)
static inline void psi_task_tick(struct rq *rq)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (unlikely(rq->curr->flags & PF_MEMSTALL))
--
2.19.1
On Mon, Nov 26, 2018 at 11:29:26PM +0000, Mel Gorman wrote:
> The following is a comparision using CONFIG_PSI=n as a baseline against
> your patch and a vanilla kernel
>
> 4.20.0-rc4 4.20.0-rc4 4.20.0-rc4
> kconfigdisable-v1r1 vanilla psidisable-v1r1
> Amean 1 1.3100 ( 0.00%) 1.3923 ( -6.28%) 1.3427 ( -2.49%)
> Amean 3 3.8860 ( 0.00%) 4.1230 * -6.10%* 3.8860 ( -0.00%)
> Amean 5 6.8847 ( 0.00%) 8.0390 * -16.77%* 6.7727 ( 1.63%)
> Amean 7 9.9310 ( 0.00%) 10.8367 * -9.12%* 9.9910 ( -0.60%)
> Amean 12 16.6577 ( 0.00%) 18.2363 * -9.48%* 17.1083 ( -2.71%)
> Amean 18 26.5133 ( 0.00%) 27.8833 * -5.17%* 25.7663 ( 2.82%)
> Amean 24 34.3003 ( 0.00%) 34.6830 ( -1.12%) 32.0450 ( 6.58%)
> Amean 30 40.0063 ( 0.00%) 40.5800 ( -1.43%) 41.5087 ( -3.76%)
> Amean 32 40.1407 ( 0.00%) 41.2273 ( -2.71%) 39.9417 ( 0.50%)
>
> It's showing that the vanilla kernel takes a hit (as the bisection
> indicated it would) and that disabling PSI by default is reasonably
> close in terms of performance for this particular workload on this
> particular machine so;
>
> Tested-by: Mel Gorman <[email protected]>
Thanks for testing it. Let's add these results to the changelog:
---
From 347b69a52d1ec7e71df1108cbc5703d6dd0616ba Mon Sep 17 00:00:00 2001
From: Johannes Weiner <[email protected]>
Date: Mon, 26 Nov 2018 09:39:23 -0500
Subject: [PATCH] psi: make disabling/enabling easier for vendor kernels
Mel Gorman reports a hackbench regression with psi that would prohibit
shipping the suse kernel with it default-enabled, but he'd still like
users to be able to opt in at little to no cost to others.
With the current combination of CONFIG_PSI and the psi_disabled bool
set from the commandline, this is a challenge. Do the following things
to make it easier:
1. Add a config option CONFIG_PSI_DEFAULT_ENABLED that allows distros
to enable CONFIG_PSI in their kernel, but leaving the feature
disabled unless a user requests it at boot-time.
To avoid double negatives, rename psi_disabled= to psi=.
2. Make psi_disabled a static branch to eliminate any branch costs
when the feature is disabled.
In terms of numbers before and after this patch, Mel says:
: The following is a comparision using CONFIG_PSI=n as a baseline against
: your patch and a vanilla kernel
:
: 4.20.0-rc4 4.20.0-rc4 4.20.0-rc4
: kconfigdisable-v1r1 vanilla psidisable-v1r1
: Amean 1 1.3100 ( 0.00%) 1.3923 ( -6.28%) 1.3427 ( -2.49%)
: Amean 3 3.8860 ( 0.00%) 4.1230 * -6.10%* 3.8860 ( -0.00%)
: Amean 5 6.8847 ( 0.00%) 8.0390 * -16.77%* 6.7727 ( 1.63%)
: Amean 7 9.9310 ( 0.00%) 10.8367 * -9.12%* 9.9910 ( -0.60%)
: Amean 12 16.6577 ( 0.00%) 18.2363 * -9.48%* 17.1083 ( -2.71%)
: Amean 18 26.5133 ( 0.00%) 27.8833 * -5.17%* 25.7663 ( 2.82%)
: Amean 24 34.3003 ( 0.00%) 34.6830 ( -1.12%) 32.0450 ( 6.58%)
: Amean 30 40.0063 ( 0.00%) 40.5800 ( -1.43%) 41.5087 ( -3.76%)
: Amean 32 40.1407 ( 0.00%) 41.2273 ( -2.71%) 39.9417 ( 0.50%)
:
: It's showing that the vanilla kernel takes a hit (as the bisection
: indicated it would) and that disabling PSI by default is reasonably
: close in terms of performance for this particular workload on this
: particular machine so;
Tested-by: Mel Gorman <[email protected]>
Reported-by: Mel Gorman <[email protected]>
Signed-off-by: Johannes Weiner <[email protected]>
---
.../admin-guide/kernel-parameters.txt | 4 +++
include/linux/psi.h | 3 +-
init/Kconfig | 9 ++++++
kernel/sched/psi.c | 30 +++++++++++++------
kernel/sched/stats.h | 8 ++---
5 files changed, 40 insertions(+), 14 deletions(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 19f4423e70d9..8760a343c6d8 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -3504,6 +3504,10 @@
before loading.
See Documentation/blockdev/ramdisk.txt.
+ psi= [KNL] Enable or disable pressure stall information
+ tracking.
+ Format: <bool>
+
psmouse.proto= [HW,MOUSE] Highest PS2 mouse protocol extension to
probe for; one of (bare|imps|exps|lifebook|any).
psmouse.rate= [HW,MOUSE] Set desired mouse report rate, in reports
diff --git a/include/linux/psi.h b/include/linux/psi.h
index 8e0725aac0aa..7006008d5b72 100644
--- a/include/linux/psi.h
+++ b/include/linux/psi.h
@@ -1,6 +1,7 @@
#ifndef _LINUX_PSI_H
#define _LINUX_PSI_H
+#include <linux/jump_label.h>
#include <linux/psi_types.h>
#include <linux/sched.h>
@@ -9,7 +10,7 @@ struct css_set;
#ifdef CONFIG_PSI
-extern bool psi_disabled;
+extern struct static_key_false psi_disabled;
void psi_init(void);
diff --git a/init/Kconfig b/init/Kconfig
index a4112e95724a..cf5b5a0dcbc2 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -509,6 +509,15 @@ config PSI
Say N if unsure.
+config PSI_DEFAULT_DISABLED
+ bool "Require boot parameter to enable pressure stall information tracking"
+ default n
+ depends on PSI
+ help
+ If set, pressure stall information tracking will be disabled
+ per default but can be enabled through passing psi_enable=1
+ on the kernel commandline during boot.
+
endmenu # "CPU/Task time and stats accounting"
config CPU_ISOLATION
diff --git a/kernel/sched/psi.c b/kernel/sched/psi.c
index 3d7355d7c3e3..fe24de3fbc93 100644
--- a/kernel/sched/psi.c
+++ b/kernel/sched/psi.c
@@ -136,8 +136,18 @@
static int psi_bug __read_mostly;
-bool psi_disabled __read_mostly;
-core_param(psi_disabled, psi_disabled, bool, 0644);
+DEFINE_STATIC_KEY_FALSE(psi_disabled);
+
+#ifdef CONFIG_PSI_DEFAULT_DISABLED
+bool psi_enable;
+#else
+bool psi_enable = true;
+#endif
+static int __init setup_psi(char *str)
+{
+ return kstrtobool(str, &psi_enable) == 0;
+}
+__setup("psi=", setup_psi);
/* Running averages - we need to be higher-res than loadavg */
#define PSI_FREQ (2*HZ+1) /* 2 sec intervals */
@@ -169,8 +179,10 @@ static void group_init(struct psi_group *group)
void __init psi_init(void)
{
- if (psi_disabled)
+ if (!psi_enable) {
+ static_branch_enable(&psi_disabled);
return;
+ }
psi_period = jiffies_to_nsecs(PSI_FREQ);
group_init(&psi_system);
@@ -549,7 +561,7 @@ void psi_memstall_enter(unsigned long *flags)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
*flags = current->flags & PF_MEMSTALL;
@@ -579,7 +591,7 @@ void psi_memstall_leave(unsigned long *flags)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (*flags)
@@ -600,7 +612,7 @@ void psi_memstall_leave(unsigned long *flags)
#ifdef CONFIG_CGROUPS
int psi_cgroup_alloc(struct cgroup *cgroup)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return 0;
cgroup->psi.pcpu = alloc_percpu(struct psi_group_cpu);
@@ -612,7 +624,7 @@ int psi_cgroup_alloc(struct cgroup *cgroup)
void psi_cgroup_free(struct cgroup *cgroup)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
cancel_delayed_work_sync(&cgroup->psi.clock_work);
@@ -637,7 +649,7 @@ void cgroup_move_task(struct task_struct *task, struct css_set *to)
struct rq_flags rf;
struct rq *rq;
- if (psi_disabled) {
+ if (static_branch_likely(&psi_disabled)) {
/*
* Lame to do this here, but the scheduler cannot be locked
* from the outside, so we move cgroups from inside sched/.
@@ -673,7 +685,7 @@ int psi_show(struct seq_file *m, struct psi_group *group, enum psi_res res)
{
int full;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return -EOPNOTSUPP;
update_stats(group);
diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h
index 4904c4677000..aa0de240fb41 100644
--- a/kernel/sched/stats.h
+++ b/kernel/sched/stats.h
@@ -66,7 +66,7 @@ static inline void psi_enqueue(struct task_struct *p, bool wakeup)
{
int clear = 0, set = TSK_RUNNING;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (!wakeup || p->sched_psi_wake_requeue) {
@@ -86,7 +86,7 @@ static inline void psi_dequeue(struct task_struct *p, bool sleep)
{
int clear = TSK_RUNNING, set = 0;
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (!sleep) {
@@ -102,7 +102,7 @@ static inline void psi_dequeue(struct task_struct *p, bool sleep)
static inline void psi_ttwu_dequeue(struct task_struct *p)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
/*
* Is the task being migrated during a wakeup? Make sure to
@@ -128,7 +128,7 @@ static inline void psi_ttwu_dequeue(struct task_struct *p)
static inline void psi_task_tick(struct rq *rq)
{
- if (psi_disabled)
+ if (static_branch_likely(&psi_disabled))
return;
if (unlikely(rq->curr->flags & PF_MEMSTALL))
--
2.19.1