If the hw supports intel_pstate and acpi_cpufreq, intel_pstate will
get loaded first.
acpi_cpufreq_init() will call acpi_cpufreq_early_init()
and that will allocate perf data and init those perf data in ACPI core,
(that will cover all CPUs). But later it will free them as
cpufreq_register_driver(acpi_cpufreq) will fail as intel_pstate is
already registered
Use cpufreq_get_current_driver() to check if we can skip the
acpi_cpufreq loading.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Standardise the defintion of the cpb (Core Performance Boost)
attribute in the acpi-cpufreq driver via the cpufreq_freq_attr_rw
macro.
[rjw: Subject and changelog]
Signed-off-by: Lan Tianyu <tianyu.lan@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
We don't need to set .owner = THIS_MODULE any more in cpufreq drivers
as this field isn't used any more by the cpufreq core.
This patch removes it and updates all dependent drivers accordingly.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Chapter 14 of Documentation/CodingStyle says:
The preferred form for passing a size of a struct is the following:
p = kmalloc(sizeof(*p), ...);
The alternative form where struct name is spelled out hurts
readability and introduces an opportunity for a bug when the pointer
variable type is changed but the corresponding sizeof that is passed
to a memory allocator is not.
This wasn't followed consistently in drivers/cpufreq, let's make it
more consistent by always following this rule.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The target frequency calculation method in the ondemand governor has
changed and it is now independent of the measured average frequency.
Consequently, the APERF/MPERF support in cpufreq is not used any
more, so drop it.
[rjw: Changelog]
Signed-off-by: Stratos Karafotis <stratosk@semaphore.gr>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
* pm-cpufreq:
cpufreq: don't leave stale policy pointer in cdbs->cur_policy
acpi-cpufreq: Add new sysfs attribute freqdomain_cpus
cpufreq: make sure frequency transitions are serialized
Commits fcf8058 (cpufreq: Simplify cpufreq_add_dev()) and aa77a52
(cpufreq: acpi-cpufreq: Don't set policy->related_cpus from .init())
changed the contents of the "related_cpus" sysfs attribute on systems
where acpi-cpufreq is used and user space can't get the list of CPUs
which are in the same hardware coordination CPU domain (provided by
the ACPI AML method _PSD) via "related_cpus" any more.
To make up for that loss add a new sysfs attribute "freqdomian_cpus"
for the acpi-cpufreq driver which exposes the list of CPUs in the
same domain regardless of whether it is coordinated by hardware or
software.
[rjw: Changelog, documentation]
References: https://bugzilla.kernel.org/show_bug.cgi?id=58761
Reported-by: Jean-Philippe Halimi <jean-philippe.halimi@exascale-computing.eu>
Signed-off-by: Lan Tianyu <tianyu.lan@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
PRECHANGE and POSTCHANGE notifiers must be called in groups, i.e either both
should be called or both shouldn't be.
In case we have started PRECHANGE notifier and found an error, we
must call POSTCHANGE notifier with freqs.new = freqs.old to
guarantee that the sequence of calling notifiers is complete.
This patch fixes it.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
After commit ac212b6 (ACPI / processor: Use common hotplug
infrastructure) the acpi-cpufreq module is not loaded automatically
by udev which fails to match it against the x86cpu modalias. Still,
it can be matched against ACPI processor device IDs, which even
makes more sense, because it depends on the ACPI processor driver
that uses those device IDs to bind to processor devices.
For this reason, add ACPI processor device IDs to acpi-cpufreq.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Commit 4b31e774 (Always set P-state on initialization) fixed bug
#4634 and caused the driver to always set the target P-State at
least once since the initial P-State may not be the desired one.
Commit 5a1c0228 (cpufreq: Avoid calling cpufreq driver's target()
routine if target_freq == policy->cur) caused a regression in
this behavior.
This fixes the regression by setting policy->cur based on the CPU's
target frequency rather than the CPU's current reported frequency
(which may be different). This means that the P-State will be set
initially if the CPU's target frequency is different from the
governor's target frequency.
This fixes an issue where setting the default governor to
performance wouldn't correctly enable turbo mode on all cores.
Signed-off-by: Ross Lagerwall <rosslagerwall@gmail.com>
Reviewed-by: Len Brown <len.brown@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Cc: 3.8+ <stable@vger.kernel.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The "index" field of struct cpufreq_frequency_table was never an
index and isn't used at all by the cpufreq core. It only is useful
for cpufreq drivers for their internal purposes.
Many people nowadays blindly set it in ascending order with the
assumption that the core will use it, which is a mistake.
Rename it to "driver_data" as that's what its purpose is. All of its
users are updated accordingly.
[rjw: Changelog]
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
When we don't have any file in cpu/cpufreq directory we shouldn't
create it. Specially with the introduction of per-policy governor
instance patchset, even governors are moved to
cpu/cpu*/cpufreq/governor-name directory and so this directory is
just not required.
Lets have it only when required.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
policy->cpus contains all online cpus that have single shared clock line. And
their frequencies are always updated together.
Many SMP system's cpufreq drivers take care of this in individual drivers but
the best place for this code is in cpufreq core.
This patch modifies cpufreq_notify_transition() to notify frequency change for
all cpus in policy->cpus and hence updates all users of this API.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
With the addition of following patch:
fcf8058 cpufreq: Simplify cpufreq_add_dev()
cpufreq driver's .init() routine must initialize policy->cpus with
mask of all possible CPUs (Online + Offline) that share the clock.
Then the core would copy this mask onto policy->related_cpus and will
reset policy->cpus to carry only online cpus.
acpi-cpufreq driver wasn't updated with this assumption and so
sometimes when we try to hot[un]plug CPUs at run time, sysfs
directories get corrupted.
This patch fixes acpi-cpufreq driver against this corruption.
Reported-and-tested-by: Maciej Rutecki <maciej.rutecki@gmail.com>
Tested-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Implement a generic helper function policy_is_shared() to replace the
current dbs_sw_coordinated_cpus() at cpufreq level, so that it can be
used by code other than cpufreq governors.
Suggested-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Fabio Baltieri <fabio.baltieri@linaro.org>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
de3ed81d74 ("[CPUFREQ] Change link order of x86 cpufreq modules")
changed cpufreq drivers link order so that powernow-k8 gets loaded first
due to earlier K8s having BIOS bugs.
However, now that acpi-cpufreq supports both AMD and Intel CPUs with HW
P-states, we want to load it first, so that cases where acpi-cpufreq and
powernow-k8 are both built-in and powernow-k8 initializing first, can be
addressed.
So, make sure that even if acpi-cpufreq gets loaded first, it errors out
on K8s and powernow-k8 can be loaded then successfully.
Signed-off-by: Matthew Garrett <mjg59@srcf.ucam.org>
References: http://lkml.kernel.org/r/20130118162347.GA31499@srcf.ucam.org
Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The acpi core will call request_module("acpi-cpufreq") on subsystem init,
but this will fail if the module isn't available at that stage of boot.
Add some module aliases to ensure that udev can load the module on Intel
and AMD systems with the appropriate feature bits - I /think/ that this
will also work on VIA systems, but haven't verified that.
References: http://lkml.kernel.org/r/1448223.sdUJnNSRz4@vostro.rjw.lan
Signed-off-by: Matthew Garrett <matthew.garrett@nebula.com>
Tested-by: Leonid Isaev <lisaev@umail.iu.edu>
Acked-by: Borislav Petkov <bp@suse.de>
Cc: 3.7+ <stable@vger.kernel.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The powernow-k8 driver supported a sysfs knob called "cpb", which was
instantiated per CPU, but actually acted globally for the whole
system. To keep some compatibility with this feature, we re-introduce
this behavior here, but:
a) only enable it on AMD CPUs and
b) protect it with a Kconfig switch
I'd like to consider this feature obsolete. Lets keep it around for
some kernel versions and then phase it out.
Signed-off-by: Andre Przywara <andre.przywara@amd.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
One feature present in powernow-k8 that isn't present in acpi-cpufreq
is support for enabling or disabling AMD's core performance boost
technology. This patch adds support to acpi-cpufreq, but also
includes support for Intel's dynamic acceleration.
The original boost disabling sysfs file was per CPU, but acted
globally. Also the naming (cpb) was at least not intuitive.
So lets introduce a single file simply called "boost", which sits
once in /sys/devices/system/cpu/cpufreq.
This should be the only way of using this feature, so add
documentation about the rationale and the usage.
A following patch will re-introduce the cpb knob for compatibility
reasons on AMD CPUs.
Per-CPU boost switching is possible, but not trivial and is thus
postponed to a later patch series.
Signed-off-by: Andre Przywara <andre.przywara@amd.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
To workaround some Windows specific behavior, the ACPI _PSD table
on AMD desktop boards advertises all cores as dependent, meaning
that they all can only use the same P-state. acpi-cpufreq strictly
obeys this description, instantiating one CPU only and symlinking
the others. But the hardware can have distinct frequencies for each
core and powernow-k8 did it that way.
So, in order to use the hardware to its full potential and keep the
original powernow-k8 behavior, lets override the _PSD table setting
on AMD hardware.
We use the siblings table, as it matches the current hardware
behavior.
Signed-off-by: Andre Przywara <andre.przywara@amd.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
The programming model for P-states on modern AMD CPUs is very similar to
that of Intel and VIA. It makes sense to consolidate this support into one
driver rather than duplicating functionality between two of them. This
patch adds support for AMDs with hardware P-state control to acpi-cpufreq.
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Andre Przywara <andre.przywara@amd.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
acpi-cpufreq checks each CPU for aperf/mperf support, but only sets a
global flag. This will cause errors if some CPUs in the system don't
support the feature. Check boot_cpu_has() instead in order to make sure
that all CPUs support it.
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Dave Jones <davej@redhat.com>
I came across a memory leak during a cyclic cpu-online-offline test.
Signed-off-by: Yu Luming <luming.yu@intel.com>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Dave Jones <davej@redhat.com>
Yinghai Lu