Rafael J. Wysocki
b7eaf1aab9
cpufreq: schedutil: Avoid reducing frequency of busy CPUs prematurely
...
The way the schedutil governor uses the PELT metric causes it to
underestimate the CPU utilization in some cases.
That can be easily demonstrated by running kernel compilation on
a Sandy Bridge Intel processor, running turbostat in parallel with
it and looking at the values written to the MSR_IA32_PERF_CTL
register. Namely, the expected result would be that when all CPUs
were 100% busy, all of them would be requested to run in the maximum
P-state, but observation shows that this clearly isn't the case.
The CPUs run in the maximum P-state for a while and then are
requested to run slower and go back to the maximum P-state after
a while again. That causes the actual frequency of the processor to
visibly oscillate below the sustainable maximum in a jittery fashion
which clearly is not desirable.
That has been attributed to CPU utilization metric updates on task
migration that cause the total utilization value for the CPU to be
reduced by the utilization of the migrated task. If that happens,
the schedutil governor may see a CPU utilization reduction and will
attempt to reduce the CPU frequency accordingly right away. That
may be premature, though, for example if the system is generally
busy and there are other runnable tasks waiting to be run on that
CPU already.
This is unlikely to be an issue on systems where cpufreq policies are
shared between multiple CPUs, because in those cases the policy
utilization is computed as the maximum of the CPU utilization values
over the whole policy and if that turns out to be low, reducing the
frequency for the policy most likely is a good idea anyway. On
systems with one CPU per policy, however, it may affect performance
adversely and even lead to increased energy consumption in some cases.
On those systems it may be addressed by taking another utilization
metric into consideration, like whether or not the CPU whose
frequency is about to be reduced has been idle recently, because if
that's not the case, the CPU is likely to be busy in the near future
and its frequency should not be reduced.
To that end, use the counter of idle calls in the timekeeping code.
Namely, make the schedutil governor look at that counter for the
current CPU every time before its frequency is about to be reduced.
If the counter has not changed since the previous iteration of the
governor computations for that CPU, the CPU has been busy for all
that time and its frequency should not be decreased, so if the new
frequency would be lower than the one set previously, the governor
will skip the frequency update.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Joel Fernandes <joelaf@google.com>
2017-03-23 02:12:14 +01:00
..
bpf
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
2017-03-04 17:31:39 -08:00
cgroup
scripts/spelling.txt: add "disble(d)" pattern and fix typo instances
2017-03-09 17:01:09 -08:00
configs
config: android-base: enable hardened usercopy and kernel ASLR
2017-02-27 18:43:46 -08:00
debug
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/debug.h>
2017-03-02 08:42:34 +01:00
events
scripts/spelling.txt: add "disble(d)" pattern and fix typo instances
2017-03-09 17:01:09 -08:00
gcov
irq
sched/headers: Prepare to move the get_task_struct()/put_task_struct() and related APIs from <linux/sched.h> to <linux/sched/task.h>
2017-03-02 08:42:40 +01:00
livepatch
locking
Merge branch 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
2017-03-07 14:33:11 -08:00
power
Merge branch 'WIP.sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
2017-03-03 10:16:38 -08:00
printk
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task_stack.h>
2017-03-02 08:42:36 +01:00
rcu
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/debug.h>
2017-03-02 08:42:34 +01:00
sched
cpufreq: schedutil: Avoid reducing frequency of busy CPUs prematurely
2017-03-23 02:12:14 +01:00
time
cpufreq: schedutil: Avoid reducing frequency of busy CPUs prematurely
2017-03-23 02:12:14 +01:00
trace
scripts/spelling.txt: add "overide" pattern and fix typo instances
2017-03-09 17:01:09 -08:00
.gitignore
acct.c
sched/headers: Prepare to move cputime functionality from <linux/sched.h> into <linux/sched/cputime.h>
2017-03-02 08:42:39 +01:00
async.c
audit.c
Merge branch 'stable-4.11' of git://git.infradead.org/users/pcmoore/audit
2017-02-21 13:25:50 -08:00
audit.h
Merge branch 'stable-4.11' of git://git.infradead.org/users/pcmoore/audit
2017-02-21 13:25:50 -08:00
audit_fsnotify.c
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
2016-12-17 18:44:00 -08:00
audit_tree.c
Merge branch 'stable-4.10' of git://git.infradead.org/users/pcmoore/audit
2017-01-05 23:06:06 -08:00
audit_watch.c
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
2016-12-17 18:44:00 -08:00
auditfilter.c
auditsc.c
Merge branch 'stable-4.11' of git://git.infradead.org/users/pcmoore/audit
2017-02-21 13:25:50 -08:00
backtracetest.c
bounds.c
capability.c
capability: export has_capability
2017-01-12 07:01:56 -07:00
compat.c
Replace <asm/uaccess.h> with <linux/uaccess.h> globally
2016-12-24 11:46:01 -08:00
configs.c
Replace <asm/uaccess.h> with <linux/uaccess.h> globally
2016-12-24 11:46:01 -08:00
context_tracking.c
cpu.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task.h>
2017-03-02 08:42:35 +01:00
cpu_pm.c
crash_dump.c
cred.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/coredump.h>
2017-03-02 08:42:28 +01:00
delayacct.c
sched/headers: Prepare to move cputime functionality from <linux/sched.h> into <linux/sched/cputime.h>
2017-03-02 08:42:39 +01:00
dma.c
elfcore.c
exec_domain.c
exit.c
userfaultfd: non-cooperative: rollback userfaultfd_exit
2017-03-09 17:01:09 -08:00
extable.c
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
2017-02-22 10:15:09 -08:00
fork.c
sched/headers, RCU: Move rcu_copy_process() from <linux/sched/task.h> to kernel/fork.c
2017-03-03 01:43:46 +01:00
freezer.c
futex.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/mm.h>
2017-03-02 08:42:28 +01:00
futex_compat.c
Replace <asm/uaccess.h> with <linux/uaccess.h> globally
2016-12-24 11:46:01 -08:00
groups.c
Replace <asm/uaccess.h> with <linux/uaccess.h> globally
2016-12-24 11:46:01 -08:00
hung_task.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/debug.h>
2017-03-02 08:42:34 +01:00
irq_work.c
jump_label.c
This release has no new tracing features, just clean ups, minor fixes
2017-02-27 13:26:17 -08:00
kallsyms.c
bpf: make jited programs visible in traces
2017-02-17 13:40:05 -05:00
kcmp.c
Kconfig.freezer
Kconfig.hz
Kconfig.locks
Kconfig.preempt
kcov.c
kcov: make kcov work properly with KASLR enabled
2016-12-20 09:48:47 -08:00
kexec.c
kexec_core.c
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/pmladek/printk
2017-02-22 17:33:34 -08:00
kexec_file.c
kexec, x86/purgatory: Unbreak it and clean it up
2017-03-10 20:55:09 +01:00
kexec_internal.h
kexec, x86/purgatory: Unbreak it and clean it up
2017-03-10 20:55:09 +01:00
kmod.c
sched/headers, vfs/execve: Prepare to move the do_execve*() prototypes from <linux/sched.h> to <linux/binfmts.h>
2017-03-02 08:42:39 +01:00
kprobes.c
powerpc updates for 4.11 part 1.
2017-02-22 10:30:38 -08:00
ksysfs.c
kernel/ksysfs.c: add __ro_after_init to bin_attribute structure
2017-02-24 17:46:56 -08:00
kthread.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task.h>
2017-03-02 08:42:35 +01:00
latencytop.c
sched/headers: Prepare to move sched_info_on() and force_schedstat_enabled() from <linux/sched.h> to <linux/sched/stat.h>
2017-03-02 08:42:39 +01:00
Makefile
cgroup: move cgroup files under kernel/cgroup/
2016-12-27 14:49:05 -05:00
membarrier.c
Fix: Disable sys_membarrier when nohz_full is enabled
2017-01-23 11:32:16 -08:00
memremap.c
mm, devm_memremap_pages: hold device_hotplug lock over mem_hotplug_{begin, done}
2017-02-24 17:46:53 -08:00
module-internal.h
module.c
Modules updates for v4.11
2017-02-22 17:08:33 -08:00
module_signing.c
notifier.c
kernel/notifier.c: simplify expression
2017-02-24 17:46:56 -08:00
nsproxy.c
padata.c
panic.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/debug.h>
2017-03-02 08:42:34 +01:00
params.c
pid.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task.h>
2017-03-02 08:42:35 +01:00
pid_namespace.c
sched/headers: Prepare for the reduction of <linux/sched.h>'s signal API dependency
2017-03-02 08:42:37 +01:00
profile.c
sched/headers: Prepare to move sched_info_on() and force_schedstat_enabled() from <linux/sched.h> to <linux/sched/stat.h>
2017-03-02 08:42:39 +01:00
ptrace.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task.h>
2017-03-02 08:42:35 +01:00
range.c
reboot.c
relay.c
lib/vsprintf.c: remove %Z support
2017-02-27 18:43:47 -08:00
resource.c
seccomp.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task_stack.h>
2017-03-02 08:42:36 +01:00
signal.c
sched/headers: Prepare to move cputime functionality from <linux/sched.h> into <linux/sched/cputime.h>
2017-03-02 08:42:39 +01:00
smp.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/idle.h>
2017-03-02 08:42:26 +01:00
smpboot.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task.h>
2017-03-02 08:42:35 +01:00
smpboot.h
softirq.c
stacktrace.c
stacktrace, lockdep: Fix address, newline ugliness
2017-02-08 08:21:31 +01:00
stop_machine.c
sys.c
sched/headers: Prepare to move cputime functionality from <linux/sched.h> into <linux/sched/cputime.h>
2017-03-02 08:42:39 +01:00
sys_ni.c
move aio compat to fs/aio.c
2016-12-22 22:58:37 -05:00
sysctl.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/coredump.h>
2017-03-02 08:42:28 +01:00
sysctl_binary.c
sysctl: add KERN_CONT to deprecated_sysctl_warning()
2016-12-14 16:04:07 -08:00
task_work.c
taskstats.c
test_kprobes.c
torture.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/clock.h>
2017-03-02 08:42:27 +01:00
tracepoint.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task.h>
2017-03-02 08:42:35 +01:00
tsacct.c
sched/headers: Prepare to move cputime functionality from <linux/sched.h> into <linux/sched/cputime.h>
2017-03-02 08:42:39 +01:00
ucount.c
ucount: Remove the atomicity from ucount->count
2017-03-06 15:26:37 -06:00
uid16.c
sched/headers: Prepare to remove <linux/cred.h> inclusion from <linux/sched.h>
2017-03-02 08:42:31 +01:00
up.c
user-return-notifier.c
user.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/user.h>
2017-03-02 08:42:29 +01:00
user_namespace.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/signal.h>
2017-03-02 08:42:29 +01:00
utsname.c
sched/headers: Prepare to move the task_lock()/unlock() APIs to <linux/sched/task.h>
2017-03-02 08:42:38 +01:00
utsname_sysctl.c
sched/headers: Remove <linux/rwsem.h> from <linux/sched.h>
2017-03-03 01:45:36 +01:00
watchdog.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/debug.h>
2017-03-02 08:42:34 +01:00
watchdog_hld.c
sched/headers: Prepare for new header dependencies before moving code to <linux/sched/debug.h>
2017-03-02 08:42:34 +01:00
workqueue.c
time: Remove CONFIG_TIMER_STATS
2017-02-10 11:15:08 +01:00
workqueue_internal.h