commit 9505b98ccd upstream.
pxa_cpufreq_init_voltages() is marked __init but usually inlined into
the non-__init pxa_cpufreq_init() function. When building with clang,
it can stay as a standalone function in a discarded section, and produce
this warning:
WARNING: vmlinux.o(.text+0x616a00): Section mismatch in reference from the function pxa_cpufreq_init() to the function .init.text:pxa_cpufreq_init_voltages()
The function pxa_cpufreq_init() references
the function __init pxa_cpufreq_init_voltages().
This is often because pxa_cpufreq_init lacks a __init
annotation or the annotation of pxa_cpufreq_init_voltages is wrong.
Fixes: 50e77fcd79 ("ARM: pxa: remove __init from cpufreq_driver->init()")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
commit 0373ca7483 upstream.
commit a307a1e6bc "cpufreq: s3c: use cpufreq_generic_init()"
accidentally broke cpufreq on s3c2410 and s3c2412.
These two platforms don't have a CPU frequency table and used to skip
calling cpufreq_table_validate_and_show() for them. But with the
above commit, we started calling it unconditionally and that will
eventually fail as the frequency table pointer is NULL.
Fix this by calling cpufreq_table_validate_and_show() conditionally
again.
Fixes: a307a1e6bc "cpufreq: s3c: use cpufreq_generic_init()"
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
commit b8e11f7d27 upstream.
Commit 27ed3cd2eb (cpufreq: conservative: Fix the logic in frequency
decrease checking) removed the 10 point substraction when comparing the
load against down_threshold but did not remove the related limit for the
down_threshold value. As a result, down_threshold lower than 11 is not
allowed even though values from 1 to 10 do work correctly too. The
comment ("cannot be lower than 11 otherwise freq will not fall") also
does not apply after removing the substraction.
For this reason, allow down_threshold to take any value from 1 to 99
and fix the related comment.
Fixes: 27ed3cd2eb (cpufreq: conservative: Fix the logic in frequency decrease checking)
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
[bwh: Backported to 3.16: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
commit a69261e447 upstream.
The "goto err_armclk;" error path already does a clk_put(s3c_freq->hclk);
so this is a double free.
Fixes: 34ee550752 ([CPUFREQ] Add S3C2416/S3C2450 cpufreq driver)
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
commit 983e600e88 upstream.
When turbo is disabled, the ->set_policy() interface is broken.
For example, when turbo is disabled and cpuinfo.max = 2900000 (full
max turbo frequency), setting the limits results in frequency less
than the requested one:
Set 1000000 KHz results in 0700000 KHz
Set 1500000 KHz results in 1100000 KHz
Set 2000000 KHz results in 1500000 KHz
This is because the limits->max_perf fraction is calculated using
the max turbo frequency as the reference, but when the max P-State is
capped in intel_pstate_get_min_max(), the reference is not the max
turbo P-State. This results in reducing max P-State.
One option is to always use max turbo as reference for calculating
limits. But this will not be correct. By definition the intel_pstate
sysfs limits, shows percentage of available performance. So when
BIOS has disabled turbo, the available performance is max non turbo.
So the max_perf_pct should still show 100%.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
[ rjw : Subject & changelog, rewrite in fewer lines of code ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
[bwh: Backported to 3.16:
- limits is a perf_limits structure, not a pointer to it
- Adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
commit e43e94c1ed upstream.
Currently, the userspace governor only updates frequency on GOV_LIMITS
if policy->cur falls outside policy->{min/max}. However, it is also
necessary to update current frequency on GOV_LIMITS to match the user
requested value if it can be achieved within the new policy->{max/min}.
This was previously the behaviour in the governor until commit d1922f0
("cpufreq: Simplify userspace governor") which incorrectly assumed that
policy->cur == user requested frequency via scaling_setspeed. This won't
be true if the user requested frequency falls outside policy->{min/max}.
Ex: a temporary thermal cap throttled the user requested frequency.
Fix this by storing the user requested frequency in a seperate variable.
The governor will then try to achieve this request on every GOV_LIMITS
change.
Fixes: d1922f0256 (cpufreq: Simplify userspace governor)
Signed-off-by: Sai Gurrappadi <sgurrappadi@nvidia.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
commit fa30dff9a8 upstream.
div_s64() accepts the divisor parameter as s32. Helper div_fp()
also accepts divisor as int32_t.
So, remove the unnecessary int64_t type casting.
Signed-off-by: Stratos Karafotis <stratosk@semaphore.gr>
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit 0dd23f9425 upstream.
Commit 007bea098b (intel_pstate: Add setting voltage value for
baytrail P states.) introduced byt_set_pstate() with the assumption that
it would always be run by the CPU whose MSR is to be written by it. It
turns out, however, that is not always the case in practice, so modify
byt_set_pstate() to enforce the MSR write done by it to always happen on
the right CPU.
Fixes: 007bea098b (intel_pstate: Add setting voltage value for baytrail P states.)
Signed-off-by: Joe Konno <joe.konno@intel.com>
Acked-by: Kristen Carlson Accardi <kristen@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit c75de0ac07 upstream.
All CPUs leaving the first-online CPU are hotplugged out on suspend and
and cpufreq core stops managing them.
On resume, we need to call cpufreq_update_policy() for this CPU's policy
to make sure its frequency is in sync with cpufreq's cached value, as it
might have got updated by hardware during suspend/resume.
The policies are always added to the top of the policy-list. So, in
normal circumstances, CPU 0's policy will be the last one in the list.
And so the code checks for the last policy.
But there are cases where it will fail. Consider quad-core system, with
policy-per core. If CPU0 is hotplugged out and added back again, the
last policy will be on CPU1 :(
To fix this in a proper way, always look for the policy of the first
online CPU. That way we will be sure that we are calling
cpufreq_update_policy() for the only CPU that wasn't hotplugged out.
Fixes: 2f0aea9363 ("cpufreq: suspend governors on system suspend/hibernate")
Reported-by: Saravana Kannan <skannan@codeaurora.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Saravana Kannan <skannan@codeaurora.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit 61882b6317 upstream.
The two functions s3c2416_cpufreq_driver_init and s3c_cpufreq_register
are marked init but are called from a context that might be run after
the __init sections are discarded, as the compiler points out:
WARNING: vmlinux.o(.data+0x1ad9dc): Section mismatch in reference from the variable s3c2416_cpufreq_driver to the function .init.text:s3c2416_cpufreq_driver_init()
WARNING: drivers/built-in.o(.text+0x35b5dc): Section mismatch in reference from the function s3c2410a_cpufreq_add() to the function .init.text:s3c_cpufreq_register()
This removes the __init markings.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit d4d4eda237 upstream.
On Dell Latitude C600 laptop with Pentium 3 850MHz processor, the
speedstep-smi driver sometimes loads and sometimes doesn't load with
"change to state X failed" message.
The hardware sometimes refuses to change frequency and in this case, we
need to retry later. I found out that we need to enable interrupts while
waiting. When we enable interrupts, the hardware blockage that prevents
frequency transition resolves and the transition is possible. With
disabled interrupts, the blockage doesn't resolve (no matter how long do
we wait). The exact reasons for this hardware behavior are unknown.
This patch enables interrupts in the function speedstep_set_state that can
be called with disabled interrupts. However, this function is called with
disabled interrupts only from speedstep_get_freqs, so it shouldn't cause
any problem.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit 6ffae8c06f upstream.
In __cpufreq_remove_dev_finish(), per-cpu 'cpufreq_cpu_data' needs
to be cleared before calling kobject_put(&policy->kobj) and under
cpufreq_driver_lock. Otherwise, if someone else calls cpufreq_cpu_get()
in parallel with it, they can obtain a non-NULL policy from that after
kobject_put(&policy->kobj) was executed.
Consider this case:
Thread A Thread B
cpufreq_cpu_get()
acquire cpufreq_driver_lock
read-per-cpu cpufreq_cpu_data
kobject_put(&policy->kobj);
kobject_get(&policy->kobj);
...
per_cpu(&cpufreq_cpu_data, cpu) = NULL
And this will result in a warning like this one:
------------[ cut here ]------------
WARNING: CPU: 0 PID: 4 at include/linux/kref.h:47
kobject_get+0x41/0x50()
Modules linked in: acpi_cpufreq(+) nfsd auth_rpcgss nfs_acl
lockd grace sunrpc xfs libcrc32c sd_mod ixgbe igb mdio ahci hwmon
...
Call Trace:
[<ffffffff81661b14>] dump_stack+0x46/0x58
[<ffffffff81072b61>] warn_slowpath_common+0x81/0xa0
[<ffffffff81072c7a>] warn_slowpath_null+0x1a/0x20
[<ffffffff812e16d1>] kobject_get+0x41/0x50
[<ffffffff815262a5>] cpufreq_cpu_get+0x75/0xc0
[<ffffffff81527c3e>] cpufreq_update_policy+0x2e/0x1f0
[<ffffffff810b8cb2>] ? up+0x32/0x50
[<ffffffff81381aa9>] ? acpi_ns_get_node+0xcb/0xf2
[<ffffffff81381efd>] ? acpi_evaluate_object+0x22c/0x252
[<ffffffff813824f6>] ? acpi_get_handle+0x95/0xc0
[<ffffffff81360967>] ? acpi_has_method+0x25/0x40
[<ffffffff81391e08>] acpi_processor_ppc_has_changed+0x77/0x82
[<ffffffff81089566>] ? move_linked_works+0x66/0x90
[<ffffffff8138e8ed>] acpi_processor_notify+0x58/0xe7
[<ffffffff8137410c>] acpi_ev_notify_dispatch+0x44/0x5c
[<ffffffff8135f293>] acpi_os_execute_deferred+0x15/0x22
[<ffffffff8108c910>] process_one_work+0x160/0x410
[<ffffffff8108d05b>] worker_thread+0x11b/0x520
[<ffffffff8108cf40>] ? rescuer_thread+0x380/0x380
[<ffffffff81092421>] kthread+0xe1/0x100
[<ffffffff81092340>] ? kthread_create_on_node+0x1b0/0x1b0
[<ffffffff81669ebc>] ret_from_fork+0x7c/0xb0
[<ffffffff81092340>] ? kthread_create_on_node+0x1b0/0x1b0
---[ end trace 89e66eb9795efdf7 ]---
The actual code flow is as follows:
Thread A: Workqueue: kacpi_notify
acpi_processor_notify()
acpi_processor_ppc_has_changed()
cpufreq_update_policy()
cpufreq_cpu_get()
kobject_get()
Thread B: xenbus_thread()
xenbus_thread()
msg->u.watch.handle->callback()
handle_vcpu_hotplug_event()
vcpu_hotplug()
cpu_down()
__cpu_notify(CPU_POST_DEAD..)
cpufreq_cpu_callback()
__cpufreq_remove_dev_finish()
cpufreq_policy_put_kobj()
kobject_put()
cpufreq_cpu_get() gets the policy from per-cpu variable cpufreq_cpu_data
under cpufreq_driver_lock, and once it gets a valid policy it expects it
to not be freed until cpufreq_cpu_put() is called.
But the race happens when another thread puts the kobject first and updates
cpufreq_cpu_data before or later. And so the first thread gets a valid policy
structure and before it does kobject_get() on it, the second one has already
done kobject_put().
Fix this by setting cpufreq_cpu_data to NULL before putting the kobject and that
too under locks.
Reported-by: Ethan Zhao <ethan.zhao@oracle.com>
Reported-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit 09712f557b upstream.
When resuming from s2ram on an SMP system without cpufreq operating
points (e.g. there's no "operating-points" property for the CPU node in
DT, or the platform doesn't use DT yet), the kernel crashes when
bringing CPU 1 online:
Enabling non-boot CPUs ...
CPU1: Booted secondary processor
Unable to handle kernel NULL pointer dereference at virtual address 0000003c
pgd = ee5e6b00
[0000003c] *pgd=6e579003, *pmd=6e588003, *pte=00000000
Internal error: Oops: a07 [#1] SMP ARM
Modules linked in:
CPU: 0 PID: 1246 Comm: s2ram Tainted: G W 3.18.0-rc3-koelsch-01614-g0377af242bb175c8-dirty #589
task: eeec5240 ti: ee704000 task.ti: ee704000
PC is at __cpufreq_add_dev.isra.24+0x24c/0x77c
LR is at __cpufreq_add_dev.isra.24+0x244/0x77c
pc : [<c0298efc>] lr : [<c0298ef4>] psr: 60000153
sp : ee705d48 ip : ee705d48 fp : ee705d84
r10: c04e0450 r9 : 00000000 r8 : 00000001
r7 : c05426a8 r6 : 00000001 r5 : 00000001 r4 : 00000000
r3 : 00000000 r2 : 00000000 r1 : 20000153 r0 : c0542734
Verify that policy is not NULL before dereferencing it to fix this.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Fixes: 8414809c6a (cpufreq: Preserve policy structure across suspend/resume)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit 16405f98bc upstream.
This is pretty much the same as Intel Baytrail, only the CPU ID is
different. Add the new ID to the supported CPU list.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: Chang Rebecca Swee Fun <rebecca.swee.fun.chang@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit d022a65ed2 upstream.
Using a VID value that is not high enough for the requested P state can
cause machine checks. Add a ceiling function to ensure calulated VIDs
with fractional values are set to the next highest integer VID value.
The algorythm for calculating the non-trubo VID from the BIOS writers
guide is:
vid_ratio = (vid_max - vid_min) / (max_pstate - min_pstate)
vid = ceiling(vid_min + (req_pstate - min_pstate) * vid_ratio)
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit b27580b05e upstream.
BYT has a different conversion from P state to frequency than the core
processors. This causes the min/max and current frequency to be
misreported on some BYT SKUs. Tested on BYT N2820, Ivybridge and
Haswell processors.
Link: https://bugzilla.yoctoproject.org/show_bug.cgi?id=6663
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit c034871712 upstream.
The user may have custom settings don't destroy them during suspend.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=80651
Reported-by: Tobias Jakobi <liquid.acid@gmx.net>
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit 4521e1a0ce upstream.
Some BIOSes modify the state of MSR_IA32_MISC_ENABLE_TURBO_DISABLE
based on the current power source for the system battery AC vs
battery. Reflect the correct current state and ability to modify the
no_turbo sysfs file based on current state of
MSR_IA32_MISC_ENABLE_TURBO_DISABLE.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=83151
Signed-off-by: Gabriele Mazzotta <gabriele.mzt@gmail.com>
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
[ luis: backported to 3.16: adjusted context ]
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit c034b02e21 upstream.
Currently the core does not expose scaling_cur_freq for set_policy()
drivers this breaks some userspace monitoring tools.
Change the core to expose this file for all drivers and if the
set_policy() driver supports the get() callback use it to retrieve the
current frequency.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=73741
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit 36b4bed5cd upstream.
Code which changes policy to powersave changes also max_policy_pct based on
max_freq. Code which change max_perf_pct has upper limit base on value
max_policy_pct. When policy is changing from powersave back to performance
then max_policy_pct is not changed. Which means that changing max_perf_pct is
not possible to high values if max_freq was too low in powersave policy.
Test case:
$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
800000
$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
3300000
$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
performance
$ cat /sys/devices/system/cpu/intel_pstate/max_perf_pct
100
$ echo powersave > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
$ echo 800000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
$ echo 20 > /sys/devices/system/cpu/intel_pstate/max_perf_pct
$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
powersave
$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
800000
$ cat /sys/devices/system/cpu/intel_pstate/max_perf_pct
20
$ echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
$ echo 3300000 > /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
$ echo 100 > /sys/devices/system/cpu/intel_pstate/max_perf_pct
$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
performance
$ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
3300000
$ cat /sys/devices/system/cpu/intel_pstate/max_perf_pct
24
And now intel_pstate driver allows to set maximal value for max_perf_pct based
on max_policy_pct which is 24 for previous powersave max_freq 800000.
This patch will set default value for max_policy_pct when setting policy to
performance so it will allow to set also max value for max_perf_pct.
Signed-off-by: Pali Rohár <pali.rohar@gmail.com>
Acked-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Luis Henriques <luis.henriques@canonical.com>
commit d62dbf77f7 upstream.
When building this driver as a module, we get a helpful warning
about the return type:
drivers/cpufreq/integrator-cpufreq.c:232:2: warning: initialization from incompatible pointer type
.remove = __exit_p(integrator_cpufreq_remove),
If the remove callback returns void, the caller gets an undefined
value as it expects an integer to be returned. This fixes the
problem by passing down the value from cpufreq_unregister_driver.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit b1b12babe3 upstream.
Commit 8e30444e15 ("cpufreq: fix cpufreq suspend/resume for intel_pstate")
introduced a bug where the governors wouldn't be stopped anymore for
->target{_index}() drivers during suspend. This happens because
'cpufreq_suspended' is updated before stopping the governors during suspend
and due to this __cpufreq_governor() would return early due to this check:
/* Don't start any governor operations if we are entering suspend */
if (cpufreq_suspended)
return 0;
Fixes: 8e30444e15 ("cpufreq: fix cpufreq suspend/resume for intel_pstate")
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 8e30444e15 upstream.
Cpufreq core introduces cpufreq_suspended flag to let cpufreq sysfs nodes
across S2RAM/S2DISK. But the flag is only set in the cpufreq_suspend()
for cpufreq drivers which have target or target_index callback. This
skips intel_pstate driver. This patch is to set the flag before checking
target or target_index callback.
Fixes: 2f0aea9363 (cpufreq: suspend governors on system suspend/hibernate)
Signed-off-by: Lan Tianyu <tianyu.lan@intel.com>
[rjw: Subject]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 7106e02bae upstream.
While debugging a cpufreq-related hardware failure on a system I saw the
following lockdep warning:
=========================
[ BUG: held lock freed! ] 3.17.0-rc4+ #1 Tainted: G E
-------------------------
insmod/2247 is freeing memory ffff88006e1b1400-ffff88006e1b17ff, with a lock still held there!
(&policy->rwsem){+.+...}, at: [<ffffffff8156d37d>] __cpufreq_add_dev.isra.21+0x47d/0xb80
3 locks held by insmod/2247:
#0: (subsys mutex#5){+.+.+.}, at: [<ffffffff81485579>] subsys_interface_register+0x69/0x120
#1: (cpufreq_rwsem){.+.+.+}, at: [<ffffffff8156cf73>] __cpufreq_add_dev.isra.21+0x73/0xb80
#2: (&policy->rwsem){+.+...}, at: [<ffffffff8156d37d>] __cpufreq_add_dev.isra.21+0x47d/0xb80
stack backtrace:
CPU: 0 PID: 2247 Comm: insmod Tainted: G E 3.17.0-rc4+ #1
Hardware name: HP ProLiant MicroServer Gen8, BIOS J06 08/24/2013
0000000000000000 000000008f3063c4 ffff88006f87bb30 ffffffff8171b358
ffff88006bcf3750 ffff88006f87bb68 ffffffff810e09e1 ffff88006e1b1400
ffffea0001b86c00 ffffffff8156d327 ffff880073003500 0000000000000246
Call Trace:
[<ffffffff8171b358>] dump_stack+0x4d/0x66
[<ffffffff810e09e1>] debug_check_no_locks_freed+0x171/0x180
[<ffffffff8156d327>] ? __cpufreq_add_dev.isra.21+0x427/0xb80
[<ffffffff8121412b>] kfree+0xab/0x2b0
[<ffffffff8156d327>] __cpufreq_add_dev.isra.21+0x427/0xb80
[<ffffffff81724cf7>] ? _raw_spin_unlock+0x27/0x40
[<ffffffffa003517f>] ? pcc_cpufreq_do_osc+0x17f/0x17f [pcc_cpufreq]
[<ffffffff8156da8e>] cpufreq_add_dev+0xe/0x10
[<ffffffff814855d1>] subsys_interface_register+0xc1/0x120
[<ffffffff8156bcf2>] cpufreq_register_driver+0x112/0x340
[<ffffffff8121415a>] ? kfree+0xda/0x2b0
[<ffffffffa003517f>] ? pcc_cpufreq_do_osc+0x17f/0x17f [pcc_cpufreq]
[<ffffffffa003562e>] pcc_cpufreq_init+0x4af/0xe81 [pcc_cpufreq]
[<ffffffffa003517f>] ? pcc_cpufreq_do_osc+0x17f/0x17f [pcc_cpufreq]
[<ffffffff81002144>] do_one_initcall+0xd4/0x210
[<ffffffff811f7472>] ? __vunmap+0xd2/0x120
[<ffffffff81127155>] load_module+0x1315/0x1b70
[<ffffffff811222a0>] ? store_uevent+0x70/0x70
[<ffffffff811229d9>] ? copy_module_from_fd.isra.44+0x129/0x180
[<ffffffff81127b86>] SyS_finit_module+0xa6/0xd0
[<ffffffff81725b69>] system_call_fastpath+0x16/0x1b
cpufreq: __cpufreq_add_dev: ->get() failed
insmod: ERROR: could not insert module pcc-cpufreq.ko: No such device
The warning occurs in the __cpufreq_add_dev() code which does
down_write(&policy->rwsem);
...
if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
policy->cur = cpufreq_driver->get(policy->cpu);
if (!policy->cur) {
pr_err("%s: ->get() failed\n", __func__);
goto err_get_freq;
}
If cpufreq_driver->get(policy->cpu) returns an error we execute the
code at err_get_freq, which does not up the policy->rwsem. This causes
the lockdep warning.
Trivial patch to up the policy->rwsem in the error path.
After the patch has been applied, and an error occurs in the
cpufreq_driver->get(policy->cpu) call we will now see
cpufreq: __cpufreq_add_dev: ->get() failed
cpufreq: __cpufreq_add_dev: ->get() failed
modprobe: ERROR: could not insert 'pcc_cpufreq': No such device
Fixes: 4e97b631f2 (cpufreq: Initialize governor for a new policy under policy->rwsem)
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 6174bac8c7 upstream.
Cpufreq depends on platform firmware to implement PStates. In case of
platform firmware failure, cpufreq should not panic host kernel with
BUG_ON(). Less severe pr_warn() will suffice.
Add firmware_has_feature(FW_FEATURE_OPALv3) check to
skip probing for device-tree on non-powernv platforms.
Signed-off-by: Vaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>
Acked-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This is only relevant to implementations with multiple clusters, where clusters
have separate clock lines but all CPUs within a cluster share it.
Consider a dual cluster platform with 2 cores per cluster. During suspend we
start hot unplugging CPUs in order 1 to 3. When CPU2 is removed, policy->kobj
would be moved to CPU3 and when CPU3 goes down we wouldn't free policy or its
kobj as we want to retain permissions/values/etc.
Now on resume, we will get CPU2 before CPU3 and will call __cpufreq_add_dev().
We will recover the old policy and update policy->cpu from 3 to 2 from
update_policy_cpu().
But the kobj is still tied to CPU3 and isn't moved to CPU2. We wouldn't create a
link for CPU2, but would try that for CPU3 while bringing it online. Which will
report errors as CPU3 already has kobj assigned to it.
This bug got introduced with commit 42f921a, which overlooked this scenario.
To fix this, lets move kobj to the new policy->cpu while bringing first CPU of a
cluster back. Also do a WARN_ON() if kobject_move failed, as we would reach here
only for the first CPU of a non-boot cluster. And we can't recover from this
situation, if kobject_move() fails.
Fixes: 42f921a6f1 (cpufreq: remove sysfs files for CPUs which failed to come back after resume)
Cc: 3.13+ <stable@vger.kernel.org> # 3.13+
Reported-and-tested-by: Bu Yitian <ybu@qti.qualcomm.com>
Reported-by: Saravana Kannan <skannan@codeaurora.org>
Reviewed-by: Srivatsa S. Bhat <srivatsa@mit.edu>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
OPPs can be populated statically, via DT, or added at run time with
dev_pm_opp_add().
While this driver handles the first case correctly, it would fail to populate
OPPs added at runtime. Because call to of_init_opp_table() would fail as there
are no OPPs in DT and probe will return early.
To fix this, remove error checking and call dev_pm_opp_init_cpufreq_table()
unconditionally.
Update bindings as well.
Suggested-by: Stephen Boyd <sboyd@codeaurora.org>
Tested-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Commit ff1f0018cf ("drivers: Enable
building of Kirkwood drivers for mach-mvebu") added Kirkwood into
mach-mvebu, adding MACH_KIRKWOOD to ARCH_KIRKWOOD in the KConfig files.
The change for ARM_KIRKWOOD_CPUFREQ replaced ARCH_KIRKWOOD with
MACH_KIRKWOOD, whereas all the other changes were ARCH_KIRKWOOD ||
MACH_KIRKWOOD.
As a consequence of this change, the cpufreq driver is no longer enabled
for ARCH_KIRKWOOD. This patch reinstates ARM_KIRKWOOD_CPUFREQ for
ARCH_KIRKWOOD.
Signed-off-by: Quentin Armitage <quentin@armitage.org.uk>
Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
PM_OPP is a library used by several of the existing cpufreq drivers.
ARM IMX6Q cpufreq driver uses this library for its functionality.
Thus, it should be selected in Kconfig.
Reported-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: Nicolas Del Piano <ndel314@gmail.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The Compaq iPAQ h3600 also has the K4S281632b-1H memory type.
Verified by prying apart a broken board.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Since commtit 8a7b1227e3 (cpufreq: davinci: move cpufreq driver to
drivers/cpufreq) this added dependancy only for CONFIG_ARCH_DAVINCI_DA850
where as davinci_cpufreq_init() call is used by all davinci platform.
This patch fixes following build error:
arch/arm/mach-davinci/built-in.o: In function `davinci_init_late':
:(.init.text+0x928): undefined reference to `davinci_cpufreq_init'
make: *** [vmlinux] Error 1
Fixes: 8a7b1227e3 (cpufreq: davinci: move cpufreq driver to drivers/cpufreq)
Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Cc: 3.10+ <stable@vger.kernel.org> # 3.10+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Ensure that cpu->cpu is set before writing MSR_IA32_PERF_CTL during CPU
initialization. Otherwise only cpu0 has its P-state set and all other
cores are left with their values unchanged.
In most cases, this is not too serious because the P-states will be set
correctly when the timer function is run. But when the default governor
is set to performance, the per-CPU current_pstate stays the same forever
and no attempts are made to write the MSRs again.
Signed-off-by: Vincent Minet <vincent@vincent-minet.net>
Cc: All applicable <stable@vger.kernel.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
If turbo is disabled in the BIOS bit 38 should be set in
MSR_IA32_MISC_ENABLE register per section 14.3.2.1 of the SDM Vol 3
document 325384-050US Feb 2014. If this bit is set do *not* attempt
to disable trubo via the MSR_IA32_PERF_CTL register. On some systems
trying to disable turbo via MSR_IA32_PERF_CTL will cause subsequent
writes to MSR_IA32_PERF_CTL not take affect, in fact reading
MSR_IA32_PERF_CTL will not show the IDA/Turbo DISENGAGE bit(32) as
set. A write of bit 32 to zero returns to normal operation.
Also deal with the case where the processor does not support
turbo and the BIOS does not report the fact in MSR_IA32_MISC_ENABLE
but does report the max and turbo P states as the same value.
Link: https://bugzilla.kernel.org/show_bug.cgi?id=64251
Cc: 3.13+ <stable@vger.kernel.org> # 3.13+
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Commit 21855ff5 (intel_pstate: Set turbo VID for BayTrail) introduced
setting the turbo VID which is required to prevent a machine check on
some Baytrail SKUs under heavy graphics based workloads. The
docmumentation update that brought the requirement to light also
changed the bit mask used for enumerating P state and VID values from
0x7f to 0x3f.
This change returns the mask value to 0x7f.
Tested with the Intel NUC DN2820FYK,
BIOS version FYBYT10H.86A.0034.2014.0513.1413 with v3.16-rc1 and
v3.14.8 kernel versions.
Fixes: 21855ff5 (intel_pstate: Set turbo VID for BayTrail)
Link: https://bugzilla.kernel.org/show_bug.cgi?id=77951
Reported-and-tested-by: Rune Reterson <rune@megahurts.dk>
Reported-and-tested-by: Eric Eickmeyer <erich@ericheickmeyer.com>
Cc: 3.13+ <stable@vger.kernel.org> # 3.13+
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Commit bd0fa9bb45 introduced a failure path to cpufreq_update_policy() if
cpufreq_driver->get(cpu) returns NULL. However, it jumps to the 'no_policy'
label, which exits without unlocking any of the locks the function acquired
earlier. This causes later calls into cpufreq to hang.
Fix this by creating a new 'unlock' label and jumping to that instead.
Fixes: bd0fa9bb45 ("cpufreq: Return error if ->get() failed in cpufreq_update_policy()")
Link: https://devtalk.nvidia.com/default/topic/751903/kernel-3-15-and-nv-drivers-337-340-failed-to-initialize-the-nvidia-kernel-module-gtx-550-ti-/
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Cc: 3.15+ <stable@vger.kernel.org> # 3.15+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
There was a mistake in the actual rounding portion this previous patch:
f0fe3cd7e1 (intel_pstate: Correct rounding in busy calculation) such that
the rounding was asymetric and incorrect.
Severity: Not very serious, but can increase target pstate by one extra value.
For real world work flows the issue should self correct (but I have no proof).
It is the equivalent of different PID gains for positive and negative numbers.
Examples:
-3.000000 used to round to -4, rounds to -3 with this patch.
-3.503906 used to round to -5, rounds to -4 with this patch.
Fixes: f0fe3cd7e1 (intel_pstate: Correct rounding in busy calculation)
Signed-off-by: Doug Smythies <dsmythies@telus.net>
Cc: 3.14+ <stable@vger.kernel.org> # 3.14+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
5fbfbcd3e8 ("cpufreq: cpufreq-cpu0: remove dependency on THERMAL and
REGULATOR") was a little too quick in completely removing the dependency
on the THERMAL driver.
The problem is that while there are inline wrappers to turn the thermal
API calls into empty functions, those do not help if the cpu-thermal
driver is a loadable module and cpufreq-cpu0 is builtin.
Since CONFIG_CPU_THERMAL is a bool option that decides whether the cpu
code is built into the thermal module or not, we have to use a dependency
on the thermal driver itself. However, if CPU_THERMAL is disabled, we
don't need the dependency, hence the strange '!CPU_THERMAL || THERMAL'
construct.
Fixes: 5fbfbcd3e8 ("cpufreq: cpufreq-cpu0: remove dependency on THERMAL and REGULATOR")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Tested-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
* pm-cpufreq:
cpufreq: cpufreq-cpu0: remove dependency on THERMAL and REGULATOR
cpufreq: tegra: update comment for clarity
cpufreq: intel_pstate: Remove duplicate CPU ID check
cpufreq: Mark CPU0 driver with CPUFREQ_NEED_INITIAL_FREQ_CHECK flag
cpufreq: governor: remove copy_prev_load from 'struct cpu_dbs_common_info'
cpufreq: governor: Be friendly towards latency-sensitive bursty workloads
cpufreq: ppc-corenet-cpu-freq: do_div use quotient
Revert "cpufreq: Enable big.LITTLE cpufreq driver on arm64"
cpufreq: Tegra: implement intermediate frequency callbacks
cpufreq: add support for intermediate (stable) frequencies
cpufreq-cpu0 uses thermal framework to register a cooling device, but doesn't
depend on it as there are dummy calls provided by thermal layer when
CONFIG_THERMAL=n. And when these calls fail, the driver is still usable.
Similar explanation is valid for regulators as well. We do have dummy calls
available for regulator APIs and the driver can work even when those calls
fail.
So, we don't really need to mention thermal and regulators as a dependency for
cpufreq-cpu0 in Kconfig as platforms without support for thermal/regulator can
also use this driver. Remove this dependency.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Tegra's driver got updated a bit (00917dd cpufreq: Tegra: implement intermediate
frequency callbacks) and implements new 'intermediate freq' infrastructure of
core. Above commit updated comments about when to call
clk_prepare_enable(pll_x_clk) and Doug wasn't satisfied with those comments and
said this:
> The "Though when target-freq is intermediate freq, we don't need to
> take this reference." makes me think that this function is actually
> called when target-freq is intermediate freq. I don't think it is,
> right?
For better clarity just make that comment more explicit about when we call
tegra_target_intermediate().
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Reported-and-reviewed-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
We check the CPU ID during driver init. There is no need
to do it again per logical CPU initialization.
So, remove the duplicate check.
Signed-off-by: Stratos Karafotis <stratosk@semaphore.gr>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Dirk Brandewie <dirk.j.brandewie@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Sometimes boot loaders set CPU frequency to a value outside of frequency table
present with cpufreq core. In such cases CPU might be unstable if it has to run
on that frequency for long duration of time and so its better to set it to a
frequency which is specified in frequency table.
Sachin recently found this problem with cpufreq-cpu0 driver when he was testing
it for Exynos.
Set this flag for cpufreq-cpu0 driver.
Reported-and-tested-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
'copy_prev_load' was recently added by commit: 18b46ab (cpufreq: governor: Be
friendly towards latency-sensitive bursty workloads).
It actually is a bit redundant as we also have 'prev_load' which can store any
integer value and can be used instead of 'copy_prev_load' by setting it zero.
True load can also turn out to be zero during long idle intervals (and hence the
actual value of 'prev_load' and the overloaded value can clash). However this is
not a problem because, if the true load was really zero in the previous
interval, it makes sense to evaluate the load afresh for the current interval
rather than copying the previous load.
So, drop 'copy_prev_load' and use 'prev_load' instead.
Update comments as well to make it more clear.
There is another change here which was probably missed by Srivatsa during the
last version of updates he made. The unlikely in the 'if' statement was covering
only half of the condition and the whole line should actually come under it.
Also checkpatch is made more silent as it was reporting this (--strict option):
CHECK: Alignment should match open parenthesis
+ if (unlikely(wall_time > (2 * sampling_rate) &&
+ j_cdbs->prev_load)) {
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cpufreq governors like the ondemand governor calculate the load on the CPU
periodically by employing deferrable timers. A deferrable timer won't fire
if the CPU is completely idle (and there are no other timers to be run), in
order to avoid unnecessary wakeups and thus save CPU power.
However, the load calculation logic is agnostic to all this, and this can
lead to the problem described below.
Time (ms) CPU 1
100 Task-A running
110 Governor's timer fires, finds load as 100% in the last
10ms interval and increases the CPU frequency.
110.5 Task-A running
120 Governor's timer fires, finds load as 100% in the last
10ms interval and increases the CPU frequency.
125 Task-A went to sleep. With nothing else to do, CPU 1
went completely idle.
200 Task-A woke up and started running again.
200.5 Governor's deferred timer (which was originally programmed
to fire at time 130) fires now. It calculates load for the
time period 120 to 200.5, and finds the load is almost zero.
Hence it decreases the CPU frequency to the minimum.
210 Governor's timer fires, finds load as 100% in the last
10ms interval and increases the CPU frequency.
So, after the workload woke up and started running, the frequency was suddenly
dropped to absolute minimum, and after that, there was an unnecessary delay of
10ms (sampling period) to increase the CPU frequency back to a reasonable value.
And this pattern repeats for every wake-up-from-cpu-idle for that workload.
This can be quite undesirable for latency- or response-time sensitive bursty
workloads. So we need to fix the governor's logic to detect such wake-up-from-
cpu-idle scenarios and start the workload at a reasonably high CPU frequency.
One extreme solution would be to fake a load of 100% in such scenarios. But
that might lead to undesirable side-effects such as frequency spikes (which
might also need voltage changes) especially if the previous frequency happened
to be very low.
We just want to avoid the stupidity of dropping down the frequency to a minimum
and then enduring a needless (and long) delay before ramping it up back again.
So, let us simply carry forward the previous load - that is, let us just pretend
that the 'load' for the current time-window is the same as the load for the
previous window. That way, the frequency and voltage will continue to be set
to whatever values they were set at previously. This means that bursty workloads
will get a chance to influence the CPU frequency at which they wake up from
cpu-idle, based on their past execution history. Thus, they might be able to
avoid suffering from slow wakeups and long response-times.
However, we should take care not to over-do this. For example, such a "copy
previous load" logic will benefit cases like this: (where # represents busy
and . represents idle)
##########.........#########.........###########...........##########........
but it will be detrimental in cases like the one shown below, because it will
retain the high frequency (copied from the previous interval) even in a mostly
idle system:
##########.........#.................#.....................#...............
(i.e., the workload finished and the remaining tasks are such that their busy
periods are smaller than the sampling interval, which causes the timer to
always get deferred. So, this will make the copy-previous-load logic copy
the initial high load to subsequent idle periods over and over again, thus
keeping the frequency high unnecessarily).
So, we modify this copy-previous-load logic such that it is used only once
upon every wakeup-from-idle. Thus if we have 2 consecutive idle periods, the
previous load won't get blindly copied over; cpufreq will freshly evaluate the
load in the second idle interval, thus ensuring that the system comes back to
its normal state.
[ The right way to solve this whole problem is to teach the CPU frequency
governors to also track load on a per-task basis, not just a per-CPU basis,
and then use both the data sources intelligently to set the appropriate
frequency on the CPUs. But that involves redesigning the cpufreq subsystem,
so this patch should make the situation bearable until then. ]
Experimental results:
+-------------------+
I ran a modified version of ebizzy (called 'sleeping-ebizzy') that sleeps in
between its execution such that its total utilization can be a user-defined
value, say 10% or 20% (higher the utilization specified, lesser the amount of
sleeps injected). This ebizzy was run with a single-thread, tied to CPU 8.
Behavior observed with tracing (sample taken from 40% utilization runs):
------------------------------------------------------------------------
Without patch:
~~~~~~~~~~~~~~
kworker/8:2-12137 416.335742: cpu_frequency: state=2061000 cpu_id=8
kworker/8:2-12137 416.335744: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
<...>-40753 416.345741: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
kworker/8:2-12137 416.345744: cpu_frequency: state=4123000 cpu_id=8
kworker/8:2-12137 416.345746: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
<...>-40753 416.355738: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
<snip> --------------------------------------------------------------------- <snip>
<...>-40753 416.402202: sched_switch: prev_comm=ebizzy ==> next_comm=swapper/8
<idle>-0 416.502130: sched_switch: prev_comm=swapper/8 ==> next_comm=ebizzy
<...>-40753 416.505738: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
kworker/8:2-12137 416.505739: cpu_frequency: state=2061000 cpu_id=8
kworker/8:2-12137 416.505741: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
<...>-40753 416.515739: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
kworker/8:2-12137 416.515742: cpu_frequency: state=4123000 cpu_id=8
kworker/8:2-12137 416.515744: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
Observation: Ebizzy went idle at 416.402202, and started running again at
416.502130. But cpufreq noticed the long idle period, and dropped the frequency
at 416.505739, only to increase it back again at 416.515742, realizing that the
workload is in-fact CPU bound. Thus ebizzy needlessly ran at the lowest frequency
for almost 13 milliseconds (almost 1 full sample period), and this pattern
repeats on every sleep-wakeup. This could hurt latency-sensitive workloads quite
a lot.
With patch:
~~~~~~~~~~~
kworker/8:2-29802 464.832535: cpu_frequency: state=2061000 cpu_id=8
<snip> --------------------------------------------------------------------- <snip>
kworker/8:2-29802 464.962538: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
<...>-40738 464.972533: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
kworker/8:2-29802 464.972536: cpu_frequency: state=4123000 cpu_id=8
kworker/8:2-29802 464.972538: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
<...>-40738 464.982531: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
<snip> --------------------------------------------------------------------- <snip>
kworker/8:2-29802 465.022533: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
<...>-40738 465.032531: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
kworker/8:2-29802 465.032532: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
<...>-40738 465.035797: sched_switch: prev_comm=ebizzy ==> next_comm=swapper/8
<idle>-0 465.240178: sched_switch: prev_comm=swapper/8 ==> next_comm=ebizzy
<...>-40738 465.242533: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
kworker/8:2-29802 465.242535: sched_switch: prev_comm=kworker/8:2 ==> next_comm=ebizzy
<...>-40738 465.252531: sched_switch: prev_comm=ebizzy ==> next_comm=kworker/8:2
Observation: Ebizzy went idle at 465.035797, and started running again at
465.240178. Since ebizzy was the only real workload running on this CPU,
cpufreq retained the frequency at 4.1Ghz throughout the run of ebizzy, no
matter how many times ebizzy slept and woke-up in-between. Thus, ebizzy
got the 10ms worth of 4.1 Ghz benefit during every sleep-wakeup (as compared
to the run without the patch) and this boost gave a modest improvement in total
throughput, as shown below.
Sleeping-ebizzy records-per-second:
-----------------------------------
Utilization Without patch With patch Difference (Absolute and % values)
10% 274767 277046 + 2279 (+0.829%)
20% 543429 553484 + 10055 (+1.850%)
40% 1090744 1107959 + 17215 (+1.578%)
60% 1634908 1662018 + 27110 (+1.658%)
A rudimentary and somewhat approximately latency-sensitive workload such as
sleeping-ebizzy itself showed a consistent, noticeable performance improvement
with this patch. Hence, workloads that are truly latency-sensitive will benefit
quite a bit from this change. Moreover, this is an overall win-win since this
patch does not hurt power-savings at all (because, this patch does not reduce
the idle time or idle residency; and the high frequency of the CPU when it goes
to cpu-idle does not affect/hurt the power-savings of deep idle states).
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Commit 6712d29319 (cpufreq: ppc-corenet-cpufreq: Fix __udivdi3 modpost
error) used the remainder from do_div instead of the quotient. Fix that
and add one to ensure minimum is met.
Fixes: 6712d29319 (cpufreq: ppc-corenet-cpufreq: Fix __udivdi3 modpost error)
Signed-off-by: Ed Swarthout <Ed.Swarthout@freescale.com>
Cc: 3.15+ <stable@vger.kernel.org> # 3.15+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
This reverts commit 4920ab8497 (cpufreq: Enable big.LITTLE cpufreq
driver on arm64) that breaks build on arm64.
Fixes: 4920ab8497 (cpufreq: Enable big.LITTLE cpufreq driver on arm64)
Reported-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
