In the current implementation, the response latency between seeing
SCHED_CPUFREQ_IOWAIT set and the actual P-state adjustment can be up
to 10ms. It can be reduced by bumping up the P-state to the max at
the time SCHED_CPUFREQ_IOWAIT is passed to intel_pstate_update_util().
With this change, the IO performance improves significantly.
For a simple "grep -r . linux" (Here linux is the kernel source
folder) with caches dropped every time on a Broadwell Xeon workstation
with per-core P-states, the user and system time is shorter by as much
as 30% - 40%.
The same performance difference was not observed on clients that don't
support per-core P-state.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
[ rjw: Changelog ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
With Android UI and benchmarks the latency of cpufreq response to
certain scheduling events can become very critical. Currently, callbacks
into cpufreq governors are only made from the scheduler if the target
CPU of the event is the same as the current CPU. This means there are
certain situations where a target CPU may not run the cpufreq governor
for some time.
One testcase to show this behavior is where a task starts running on
CPU0, then a new task is also spawned on CPU0 by a task on CPU1. If the
system is configured such that the new tasks should receive maximum
demand initially, this should result in CPU0 increasing frequency
immediately. But because of the above mentioned limitation though, this
does not occur.
This patch updates the scheduler core to call the cpufreq callbacks for
remote CPUs as well.
The schedutil, ondemand and conservative governors are updated to
process cpufreq utilization update hooks called for remote CPUs where
the remote CPU is managed by the cpufreq policy of the local CPU.
The intel_pstate driver is updated to always reject remote callbacks.
This is tested with couple of usecases (Android: hackbench, recentfling,
galleryfling, vellamo, Ubuntu: hackbench) on ARM hikey board (64 bit
octa-core, single policy). Only galleryfling showed minor improvements,
while others didn't had much deviation.
The reason being that this patch only targets a corner case, where
following are required to be true to improve performance and that
doesn't happen too often with these tests:
- Task is migrated to another CPU.
- The task has high demand, and should take the target CPU to higher
OPPs.
- And the target CPU doesn't call into the cpufreq governor until the
next tick.
Based on initial work from Steve Muckle.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Saravana Kannan <skannan@codeaurora.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
After commit 62611cb912 (intel_pstate: delete scheduler hook in HWP
mode) the INTEL_PSTATE_HWP_SAMPLING_INTERVAL is not used anywhere in
the code, so drop it.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The ->get callback in the intel_pstate structure was mostly there
for the scaling_cur_freq sysfs attribute to work, but after commit
f8475cef90 (x86: use common aperfmperf_khz_on_cpu() to calculate
KHz using APERF/MPERF) that attribute uses arch_freq_get_on_cpu()
provided by the x86 arch code on all processors supported by
intel_pstate, so it doesn't need the ->get callback from the
driver any more.
Moreover, the very presence of the ->get callback in the intel_pstate
structure causes the cpuinfo_cur_freq attribute to be present when
intel_pstate operates in the active mode, which is bogus, because
the role of that attribute is to return the current CPU frequency
as seen by the hardware. For intel_pstate, though, this is just an
average frequency and not really current, but computed for the
previous sampling interval (the actual current frequency may be
way different at the point this value is obtained by reading from
cpuinfo_cur_freq), and after commit 82b4e03e01 (intel_pstate: skip
scheduler hook when in "performance" mode) the value in
cpuinfo_cur_freq may be stale or just 0, depending on the driver's
operation mode. In fact, however, on the hardware supported by
intel_pstate there is no way to read the current CPU frequency
from it, so the cpuinfo_cur_freq attribute should not be present
at all when this driver is in use.
For this reason, drop intel_pstate_get() and clear the ->get
callback pointer pointing to it, so that the cpuinfo_cur_freq is
not present for intel_pstate in the active mode any more.
Fixes: 82b4e03e01 (intel_pstate: skip scheduler hook when in "performance" mode)
Reported-by: Huaisheng Ye <yehs1@lenovo.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
All systems use the same P-state selection "powersave" algorithm
in the active mode if HWP is not used, so there's no need to provide
a pointer for it in struct pstate_funcs any more.
Drop ->update_util from struct pstate_funcs and make
intel_pstate_set_update_util_hook() use intel_pstate_update_util()
directly.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
All systems with a defined ACPI preferred profile that are not
"servers" have been using the load-based P-state selection algorithm
in intel_pstate since 4.12-rc1 (mobile systems and laptops have been
using it since 4.10-rc1) and no problems with it have been reported
to date. In particular, no regressions with respect to the PID-based
P-state selection have been reported. Also testing indicates that
the P-state selection algorithm based on CPU load is generally on par
with the PID-based algorithm performance-wise, and for some workloads
it turns out to be better than the other one, while being more
straightforward and easier to understand at the same time.
Moreover, the PID-based P-state selection algorithm in intel_pstate
is known to be unstable in some situation and generally problematic,
the issues with it are hard to address and it has become a
significant maintenance burden.
For these reasons, make intel_pstate use the "powersave" P-state
selection algorithm based on CPU load in the active mode on all
systems and drop the PID-based P-state selection code along with
all things related to it from the driver. Also update the
documentation accordingly.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The transition_latency field isn't used for drivers with ->setpolicy()
callback present and there is no point setting it from the drivers.
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The busy percent calculated for the Knights Landing (KNL) platform
is 1024 times smaller than the correct busy value. This causes
performance to get stuck at the lowest ratio.
The scaling algorithm used for KNL is performance-based, but it still
looks at the CPU load to set the scaled busy factor to 0 when the
load is less than 1 percent. In this case, since the computed load
is 1024x smaller than it should be, the scaled busy factor will
always be 0, irrespective of CPU business.
This needs a fix similar to the turbostat one in commit b2b34dfe4d
(tools/power turbostat: KNL workaround for %Busy and Avg_MHz).
For this reason, add one more callback to processor-specific
callbacks to specify an MPERF multiplier represented by a number of
bit positions to shift the value of that register to the left to
copmensate for its rate difference with respect to the TSC. This
shift value is used during CPU busy calculations.
Fixes: ffb810563c (intel_pstate: Avoid getting stuck in high P-states when idle)
Reported-and-tested-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Cc: 4.6+ <stable@vger.kernel.org> # 4.6+
[ rjw: Changelog ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
When the minimum performance limit percentage is set to the power-up
default, it is possible that minimum performance ratio is off by one.
In the set_policy() callback the minimum ratio is calculated by
applying global.min_perf_pct to turbo_ratio and rounding up, but the
power-up default global.min_perf_pct is already rounded up to the
next percent in min_perf_pct_min(). That results in two round up
operations, so for the default min_perf_pct one of them is not
required.
It is better to remove rounding up in min_perf_pct_min() as this
matches the displayed min_perf_pct prior to commit c5a2ee7dde
(cpufreq: intel_pstate: Active mode P-state limits rework) in 4.12.
For example on a platform with max turbo ratio of 37 and minimum
ratio of 10, the min_perf_pct resulted in 28 with the above commit.
Before this commit it was 27 and it will be the same after this
change.
Fixes: 1a4fe38add (cpufreq: intel_pstate: Remove max/min fractions to limit performance)
Reported-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
attribute_groups are not supposed to change at runtime. All functions
working with attribute_groups provided by <linux/sysfs.h> work with const
attribute_group. So mark the non-const structs as const.
File size before:
text data bss dec hex filename
15197 2552 40 17789 457d drivers/cpufreq/intel_pstate.o
File size After adding 'const':
text data bss dec hex filename
15261 2488 40 17789 457d drivers/cpufreq/intel_pstate.o
Signed-off-by: Arvind Yadav <arvind.yadav.cs@gmail.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
* pm-cpufreq:
cpufreq / CPPC: Initialize policy->min to lowest nonlinear performance
cpufreq: sfi: make freq_table static
cpufreq: exynos5440: Fix inconsistent indenting
cpufreq: imx6q: imx6ull should use the same flow as imx6ul
cpufreq: dt: Add support for hi3660
* intel_pstate:
cpufreq: Update scaling_cur_freq documentation
cpufreq: intel_pstate: Clean up after performance governor changes
intel_pstate: skip scheduler hook when in "performance" mode
intel_pstate: delete scheduler hook in HWP mode
x86: use common aperfmperf_khz_on_cpu() to calculate KHz using APERF/MPERF
cpufreq: intel_pstate: Remove max/min fractions to limit performance
x86: do not use cpufreq_quick_get() for /proc/cpuinfo "cpu MHz"
* pm-cpuidle:
cpuidle: menu: allow state 0 to be disabled
intel_idle: Use more common logging style
x86/ACPI/cstate: Allow ACPI C1 FFH MWAIT use on AMD systems
ARM: cpuidle: Support asymmetric idle definition
* pm-tools:
cpupower: Add support for new AMD family 0x17
cpupower: Fix bug where return value was not used
tools/power turbostat: update version number
tools/power turbostat: decode MSR_IA32_MISC_ENABLE only on Intel
tools/power turbostat: stop migrating, unless '-m'
tools/power turbostat: if --debug, print sampling overhead
tools/power turbostat: hide SKL counters, when not requested
intel_pstate: use updated msr-index.h HWP.EPP values
tools/power x86_energy_perf_policy: support HWP.EPP
x86: msr-index.h: fix shifts to ULL results in HWP macros.
x86: msr-index.h: define HWP.EPP values
x86: msr-index.h: define EPB mid-points
After commit 82b4e03e01 (intel_pstate: skip scheduler hook when in
"performance" mode) get_target_pstate_use_performance() and
get_target_pstate_use_cpu_load() are never called if scaling_governor
is "performance", so drop the CPUFREQ_POLICY_PERFORMANCE checks from
them as they will never trigger anyway.
Moreover, the documentation needs to be updated to reflect the change
made by the above commit, so do that too.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When the governor is set to "performance", intel_pstate does not
need the scheduler hook for doing any calculations. Under these
conditions, its only purpose is to continue to maintain
cpufreq/scaling_cur_freq.
The cpufreq/scaling_cur_freq sysfs attribute is now provided by
shared x86 cpufreq code on modern x86 systems, including
all systems supported by the intel_pstate driver.
So in "performance" governor mode, the scheduler hook can be skipped.
This applies to both in Software and Hardware P-state control modes.
Suggested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The cpufreq/scaling_cur_freq sysfs attribute is now provided by
shared x86 cpufreq code on modern x86 systems, including
all systems supported by the intel_pstate driver.
In HWP mode, maintaining that value was the sole purpose of
the scheduler hook, intel_pstate_update_util_hwp(),
so it can now be removed.
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
In the current model the max/min perf limits are a fraction of current
user space limits to the allowed max_freq or 100% for global limits.
This results in wrong ratio limits calculation because of rounding
issues for some user space limits.
Initially we tried to solve this issue by issue by having more shift
bits to increase precision. Still there are isolated cases where we still
have error.
This can be avoided by using ratios all together. Since the way we get
cpuinfo.max_freq is by multiplying scaling factor to max ratio, we can
easily keep the max/min ratios in terms of ratios and not fractions.
For example:
if the max ratio = 36
cpuinfo.max_freq = 36 * 100000 = 3600000
Suppose user space sets a limit of 1200000, then we can calculate
max ratio limit as
= 36 * 1200000 / 3600000
= 12
This will be correct for any user limits.
The other advantage is that, we don't need to do any calculation in the
fast path as ratio limit is already calculated via set_policy() callback.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Commit c5a2ee7dde (cpufreq: intel_pstate: Active mode P-state
limits rework) incorrectly assumed that pstate.turbo_pstate would
always be nonzero for CPU0 in min_perf_pct_min() if
cpufreq_register_driver() had succeeded which may not be the case
in virtualized environments.
If that assumption doesn't hold, it leads to an early crash on boot
in intel_pstate_register_driver(), so add a sanity check to
min_perf_pct_min() to prevent the crash from happening.
Fixes: c5a2ee7dde (cpufreq: intel_pstate: Active mode P-state limits rework)
Reported-and-tested-by: Jongman Heo <jongman.heo@samsung.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
intel_pstate exports sysfs attributes for setting and observing HWP.EPP.
These attributes use strings to describe 4 operating states, and
inside the driver, these strings are mapped to numerical register
values.
The authorative mapping between the strings and numerical HWP.EPP values
are now globally defined in msr-index.h, replacing the out-dated
mapping that were open-coded into intel_pstate.c
new old string
--- --- ------
0 0 performance
128 64 balance_performance
192 128 balance_power
255 192 power
Note that the HW and BIOS default value on most system is 128,
which intel_pstate will now call "balance_performance"
while it used to call it "balance_power".
Signed-off-by: Len Brown <len.brown@intel.com>
Make the schedutil governor take the initial (default) value of the
rate_limit_us sysfs attribute from the (new) transition_delay_us
policy parameter (to be set by the scaling driver).
That will allow scaling drivers to make schedutil use smaller default
values of rate_limit_us and reduce the default average time interval
between consecutive frequency changes.
Make intel_pstate set transition_delay_us to 500.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Use same parameters as INTEL_FAM6_ATOM_GOLDMONT to enable
Gemini Lake.
Signed-off-by: Box, David E <david.e.box@intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Some computations in intel_pstate_get_min_max() are not necessary
and one of its two callers doesn't even use the full result.
First off, the fixed-point value of cpu->max_perf represents a
non-negative number between 0 and 1 inclusive and cpu->min_perf
cannot be greater than cpu->max_perf. It is not necessary to check
those conditions every time the numbers in question are used.
Moreover, since intel_pstate_max_within_limits() only needs the
upper boundary, it doesn't make sense to compute the lower one in
there and returning min and max from intel_pstate_get_min_max()
via pointers doesn't look particularly nice.
For the above reasons, drop intel_pstate_get_min_max(), add a helper
to get the base P-state for min/max computations and carry out them
directly in the previous callers of intel_pstate_get_min_max().
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
intel_pstate_hwp_set() is the only function walking policy->cpus
in intel_pstate. The rest of the code simply assumes one CPU per
policy, including the initialization code.
Therefore it doesn't make sense for intel_pstate_hwp_set() to
walk policy->cpus as it is guaranteed to have only one bit set
for policy->cpu.
For this reason, rearrange intel_pstate_hwp_set() to take the CPU
number as the argument and drop the loop over policy->cpus from it.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Add a new function pid_in_use() to return the information on whether
or not the PID-based P-state selection algorithm is in use.
That allows a couple of complicated conditions in the code to be
reduced to simple checks against the new function's return value.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The cpu_defaults structure is redundant, because it only contains
one member of type struct pstate_funcs which can be used directly
instead of struct cpu_defaults.
For this reason, drop struct cpu_defaults, use struct pstate_funcs
directly instead of it where applicable and rename all of the
variables of that type accordingly.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Move the definitions of the cpu_defaults structures after the
definitions of utilization update callback routines to avoid
extra declarations of the latter.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Avoid using extra function pointers during P-state selection by
dropping the get_target_pstate member from struct pstate_funcs,
adding a new update_util callback to it (to be registered with
the CPU scheduler as the utilization update callback in the active
mode) and reworking the utilization update callback routines to
invoke specific P-state selection functions directly.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Notice that some overhead in the utilization update callbacks
registered by intel_pstate in the active mode can be avoided if
those callbacks are tailored to specific configurations of the
driver. For example, the utilization update callback for the HWP
enabled case only needs to update the average CPU performance
periodically whereas the utilization update callback for the
PID-based algorithm does not need to take IO-wait boosting into
account and so on.
With that in mind, define three utilization update callbacks for
three different use cases: HWP enabled, the CPU load "powersave"
P-state selection algorithm and the PID-based "powersave" P-state
selection algorithm and modify the driver initialization to
choose the callback matching its current configuration.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
One of the checks in intel_pstate_update_status() implicitly relies
on the information that there are only two struct cpufreq_driver
objects available, but it is better to do it directly against the
value it really is about (to make the code easier to follow if
nothing else).
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The driver_registered variable in intel_pstate is used for checking
whether or not the driver has been registered, but intel_pstate_driver
can be used for that too (with the rule that the driver is not
registered as long as it is NULL).
That is a bit more straightforward and the code may be simplified
a bit this way, so modify the driver accordingly.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
PID controller parameters only need to be initialized if the
get_target_pstate_use_performance() P-state selection routine
is going to be used. It is not necessary to initialize them
otherwise, so don't do that.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
In the HWP enabled case pid_params.sample_rate_ns only needs to be
updated once, because it is global, so do that when setting hwp_active
instead of doing it during the initialization of every CPU.
Moreover, pid_params.sample_rate_ms is never used if HWP is enabled,
so do not update it at all then.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
intel_pstate_busy_pid_reset() is the only caller of pid_reset(),
pid_p_gain_set(), pid_i_gain_set(), and pid_d_gain_set(). Moreover,
it passes constants as two parameters of pid_reset() and all of
the other routines above essentially contain the same code, so
fold all of them into the caller and drop unnecessary computations.
Introduce percent_fp() for converting integer values in percent
to fixed-point fractions and use it in the above code cleanup.
Finally, rename intel_pstate_busy_pid_reset() to
intel_pstate_pid_reset() as it also is used for the
initialization of PID parameters for every CPU and the
meaning of the "busy" part of the name is not particularly
clear.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
There is only one caller of intel_pstate_reset_all_pid(), which is
pid_param_set() used in the debugfs interface only, and having that
code split does not make it particularly convenient to follow.
For this reason, move the body of intel_pstate_reset_all_pid() into
its caller and drop that function.
Also change the loop from for_each_online_cpu() (which is obviously
racy with respect to CPU offline/online) to for_each_possible_cpu(),
so that all PID parameters are reset for all CPUs regardless of their
online/offline status (to prevent, for example, a previously offline
CPU from going online with a stale set of PID parameters).
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Notice that both the existing struct cpu_defaults instances in which
PID parameters are actually initialized use the same values of those
parameters, so it is not really necessary to copy them over to
pid_params dynamically.
Instead, initialize pid_params statically with those values and
drop the unused pid_policy member from struct cpu_defaults along
with copy_pid_params() used for initializing it.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The P-state selection algorithm used by intel_pstate for Atom
processors is not based on the PID controller and the initialization
of PID parametrs for those processors is pointless and confusing, so
drop it.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
After recent changes the purpose of struct perf_limits is not
particularly clear any more and the code may be made somewhat
easier to follow by eliminating it, so go for that.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Both intel_pstate_verify_policy() and intel_cpufreq_verify_policy()
set policy->cpuinfo.max_freq depending on the turbo status, but the
updates made by them are discarded by the core, because the policy
object passed to them by the core is temporary and cpuinfo.max_freq
from that object is not copied to the final policy object in
cpufreq_set_policy().
However, cpufreq_set_policy() passes the temporary policy object
to the ->setpolicy callback of the driver, so intel_pstate_set_policy()
actually sees the policy->cpuinfo.max_freq value updated by
intel_pstate_verify_policy() and not the final one. It also
updates policy->max sometimes which basically has no effect after
it returns, because the core discards that update.
To avoid confusion, eliminate policy->cpuinfo.max_freq updates from
intel_pstate_verify_policy() and intel_cpufreq_verify_policy()
entirely and check the maximum frequency explicitly in
intel_pstate_update_perf_limits() instead of relying on the
transiently updated policy->cpuinfo.max_freq value.
Moreover, move the max->policy adjustment carried out in
intel_pstate_set_policy() to a separate function and call that
function from the ->verify driver callbacks to ensure that it will
actually be effective.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The coordination of P-state limits used by intel_pstate in the active
mode (ie. by default) is problematic, because it synchronizes all of
the limits (ie. the global ones and the per-policy ones) so as to use
one common pair of P-state limits (min and max) across all CPUs in
the system. The drawbacks of that are as follows:
- If P-states are coordinated in hardware, it is not necessary
to coordinate them in software on top of that, so in that case
all of the above activity is in vain.
- If P-states are not coordinated in hardware, then the processor
is actually capable of setting different P-states for different
CPUs and coordinating them at the software level simply doesn't
allow that capability to be utilized.
- The coordination works in such a way that setting a per-policy
limit (eg. scaling_max_freq) for one CPU causes the common
effective limit to change (and it will affect all of the other
CPUs too), but subsequent reads from the corresponding sysfs
attributes for the other CPUs will return stale values (which
is confusing).
- Reads from the global P-state limit attributes, min_perf_pct and
max_perf_pct, return the effective common values and not the last
values set through these attributes. However, the last values
set through these attributes become hard limits that cannot be
exceeded by writes to scaling_min_freq and scaling_max_freq,
respectively, and they are not exposed, so essentially users
have to remember what they are.
All of that is painful enough to warrant a change of the management
of P-state limits in the active mode.
To that end, redesign the active mode P-state limits management in
intel_pstate in accordance with the following rules:
(1) All CPUs are affected by the global limits (that is, none of
them can be requested to run faster than the global max and
none of them can be requested to run slower than the global
min).
(2) Each individual CPU is affected by its own per-policy limits
(that is, it cannot be requested to run faster than its own
per-policy max and it cannot be requested to run slower than
its own per-policy min).
(3) The global and per-policy limits can be set independently.
Also, the global maximum and minimum P-state limits will be always
expressed as percentages of the maximum supported turbo P-state.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Extend the set of systems for which intel_pstate will use the
"powersave" P-state selection algorithm based on CPU load in the
active mode by systems with ACPI preferred profile set to "tablet",
"appliance PC", "desktop", or "workstation" (ie. everything with a
specified preferred profile that is not a "server").
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Currently, some processors supporting HWP are only supported by
intel_pstate if HWP is actually going to be used and not supported
otherwise which is confusing.
Specifically, they are not supported if "intel_pstate=no_hwp" is
passed to the kernel in the command line or if the driver is started
in the passive mode ("intel_pstate=passive").
There is no real reason for that, because everything about those
processor is known anyway and the driver can work with them in all
modes, so make that happen, but use the load-based P-state selection
algorithm for the active mode "powersave" policy with them.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The policy->cpuinfo.max_freq and policy->max updates in
intel_cpufreq_turbo_update() are excessive as they are done for no
good reason and may lead to problems in principle, so they should be
dropped. However, after dropping them intel_cpufreq_turbo_update()
becomes almost entirely pointless, because the check made by it is
made again down the road in intel_pstate_prepare_request(). The
only thing in it that still needs to be done is the call to
update_turbo_state(), so drop intel_cpufreq_turbo_update() altogether
and make its callers invoke update_turbo_state() directly instead of
it.
In addition to that, fix intel_cpufreq_verify_policy() so that it
checks global.no_turbo in addition to global.turbo_disabled when
updating policy->cpuinfo.max_freq to make it consistent with
intel_pstate_verify_policy().
Fixes: 001c76f05b (cpufreq: intel_pstate: Generic governors support)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
In the active mode intel_pstate currently uses two sets of global
limits, each associated with one of the possible scaling_governor
settings in that mode: "powersave" or "performance".
The driver switches over from one of those sets to the other
depending on the scaling_governor setting for the last CPU whose
per-policy cpufreq interface in sysfs was last used to change
parameters exposed in there. That obviously leads to no end of
issues when the scaling_governor settings differ between CPUs.
The most recent issue was introduced by commit a240c4aa5d (cpufreq:
intel_pstate: Do not reinit performance limits in ->setpolicy)
that eliminated the reinitialization of "performance" limits in
intel_pstate_set_policy() preventing the max limit from being set
to anything below 100, among other things.
Namely, an undesirable side effect of commit a240c4aa5d is that
now, after setting scaling_governor to "performance" in the active
mode, the per-policy limits for the CPU in question go to the highest
level and stay there even when it is switched back to "powersave"
later.
As it turns out, some distributions set scaling_governor to
"performance" temporarily for all CPUs to speed-up system
initialization, so that change causes them to misbehave later.
To fix that, get rid of the performance/powersave global limits
split and use just one set of global limits for everything.
From the user's persepctive, after this modification, when
scaling_governor is switched from "performance" to "powersave"
or the other way around on one CPU, the limits settings (ie. the
global max/min_perf_pct and per-policy scaling_max/min_freq for
any CPUs) will not change. Still, switching from "performance"
to "powersave" or the other way around changes the way in which
P-states are selected and in particular "performance" causes the
driver to always request the highest P-state it is allowed to ask
for for the given CPU.
Fixes: a240c4aa5d (cpufreq: intel_pstate: Do not reinit performance limits in ->setpolicy)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Currently, intel_pstate_update_perf_limits() first converts the
policy minimum and maximum limits into percentages of the maximum
turbo frequency (rounding up to an integer) and then converts these
percentages to fractions (by using fixed-point arithmetic to divide
them by 100).
That introduces a rounding error unnecessarily, because the fractions
can be obtained by carrying out fixed-point divisions directly on the
input numbers.
Rework the computations in intel_pstate_hwp_set() to use fractions
instead of percentages (and drop redundant local variables from
there) and modify intel_pstate_update_perf_limits() to compute the
fractions directly and percentages out of them.
While at it, introduce percent_ext_fp() for converting percentages
to fractions (with extended number of fraction bits) and use it in
the computations.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
In the functions intel_pstate_hwp_set(), min/max range from HWP capability
MSR along with max_perf_pct and min_perf_pct, is used to set the HWP
request MSR. In some cases this doesn't result in the correct HWP max/min
in HWP request.
For example: In the following case:
HWP capabilities from MSR 0x771
0x70a1220
Here cpufreq min/max frequencies from above MSR dump are 700MHz and 3.2GHz
respectively.
This will result in
hwp_min = 0x07
hwp_max = 0x20
To limit max frequency to 2GHz:
perf_limits->max_perf_pct = 63 (2GHz as a percent of 3.2GHz rounded up)
With the current calculation:
adj_range = max_perf_pct * range / 100;
adj_range = 63 * (32 - 7) / 100
adj_range = 15
max = hw_min + adj_range;
max = 7 + 15 = 22
This will result in HWP request of 0x160f, which will result in a
frequency cap of 2.2GHz not 2GHz.
The problem with the above calculation is that hwp_min of 7 is treated
as 0% in the range. But max_perf_pct is calculated with respect to minimum
as 0 and max as 3.2GHz or hwp_max, so adding hwp_min to it will result in
more than the desired.
Since the min_perf_pct and max_perf_pct is already a percent of max
frequency or hwp_max, this min/max HWP request value can be calculated
directly applying these percentage to hwp_max.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Fix the debugfs interface for PID tuning to actually update
pid_params.sample_rate_ns on PID parameters updates, as changing
pid_params.sample_rate_ms via debugfs has no effect now.
Fixes: a4675fbc4a (cpufreq: intel_pstate: Replace timers with utilization update callbacks)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
intel_pstate_hwp_set_policy() is a wrapper around
intel_pstate_hwp_set(), but the only value it adds is to check
hwp_active before calling the latter and one of its two callers
has already checked hwp_active before that happens, so in that
code path the additional check is redundant and using the wrapper
is rather pointless.
For this reason, drop intel_pstate_hwp_set_policy() and make its
callers invoke intel_pstate_hwp_set() directly (after checking
hwp_active).
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
* pm-cpufreq:
cpufreq: intel_pstate: Do not reinit performance limits in ->setpolicy
cpufreq: intel_pstate: Fix intel_pstate_verify_policy()
cpufreq: intel_pstate: Fix global settings in active mode
cpufreq: Add the "cpufreq.off=1" cmdline option
cpufreq: intel_pstate: Avoid triggering cpu_frequency tracepoint unnecessarily
cpufreq: intel_pstate: Fix intel_cpufreq_verify_policy()
cpufreq: intel_pstate: Do not use performance_limits in passive mode
If the current P-state selection algorithm is set to "performance"
in intel_pstate_set_policy(), the limits may be initialized from
scratch, but only if no_turbo is not set and the maximum frequency
allowed for the given CPU (i.e. the policy object representing it)
is at least equal to the max frequency supported by the CPU. In all
of the other cases, the limits will not be updated.
For example, the following can happen:
# cat intel_pstate/status
active
# echo performance > cpufreq/policy0/scaling_governor
# cat intel_pstate/min_perf_pct
100
# echo 94 > intel_pstate/min_perf_pct
# cat intel_pstate/min_perf_pct
100
# cat cpufreq/policy0/scaling_max_freq
3100000
echo 3000000 > cpufreq/policy0/scaling_max_freq
# cat intel_pstate/min_perf_pct
94
# echo 95 > intel_pstate/min_perf_pct
# cat intel_pstate/min_perf_pct
95
That is confusing for two reasons. First, the initial attempt to
change min_perf_pct to 94 seems to have no effect, even though
setting the global limits should always work. Second, after
changing scaling_max_freq for policy0 the global min_perf_pct
attribute shows 94, even though it should have not been affected
by that operation in principle.
Moreover, the final attempt to change min_perf_pct to 95 worked
as expected, because scaling_max_freq for the only policy with
scaling_governor equal to "performance" was different from the
maximum at that time.
To make all that confusion go away, modify intel_pstate_set_policy()
so that it doesn't reinitialize the limits at all.
At the same time, change intel_pstate_set_performance_limits() to
set min_sysfs_pct to 100 in the "performance" limits set so that
switching the P-state selection algorithm to "performance" causes
intel_pstate/min_perf_pct in sysfs to go to 100 (or whatever value
min_sysfs_pct in the "performance" limits is set to later).
That requires per-CPU limits to be initialized explicitly rather
than by copying the global limits to avoid setting min_sysfs_pct
in the per-CPU limits to 100.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The code added to intel_pstate_verify_policy() by commit 1443ebbacf
(cpufreq: intel_pstate: Fix sysfs limits enforcement for performance
policy) should use perf_limits instead of limits, because otherwise
setting global limits via sysfs may affect policies inconsistently.
For example, in the sequence of shell commands below, the
scaling_min_freq attribute for policy1 and policy2 should be
affected in the same way, because scaling_governor is set in
the same way for both of them:
# cat cpufreq/policy1/scaling_governor
powersave
# cat cpufreq/policy2/scaling_governor
powersave
# echo performance > cpufreq/policy0/scaling_governor
# echo 94 > intel_pstate/min_perf_pct
# cat cpufreq/policy0/scaling_min_freq
2914000
# cat cpufreq/policy1/scaling_min_freq
2914000
# cat cpufreq/policy2/scaling_min_freq
800000
The are affected differently, because intel_pstate_verify_policy()
is invoked with limits set to &performance_limits (left behind by
policy0) for policy1 and with limits set to &powersave_limits (left
behind by policy1) for policy2. Since perf_limits is set to the
set of limits matching the policy being updated, using it instead
of limits fixes the inconsistency.
Fixes: 1443ebbacf (cpufreq: intel_pstate: Fix sysfs limits enforcement for performance policy)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Commit 111b8b3fe4 (cpufreq: intel_pstate: Always keep all
limits settings in sync) changed intel_pstate to invoke
cpufreq_update_policy() for every registered CPU on global sysfs
attributes updates, but that led to undesirable effects in the
active mode if the "performance" P-state selection algorithm is
configufred for one CPU and the "powersave" one is chosen for
all of the other CPUs.
Namely, in that case, the following is possible:
# cd /sys/devices/system/cpu/
# cat intel_pstate/max_perf_pct
100
# cat intel_pstate/min_perf_pct
26
# echo performance > cpufreq/policy0/scaling_governor
# cat intel_pstate/max_perf_pct
100
# cat intel_pstate/min_perf_pct
100
# echo 94 > intel_pstate/min_perf_pct
# cat intel_pstate/min_perf_pct
26
The reason why this happens is because intel_pstate attempts to
maintain two sets of global limits in the active mode, one for
the "performance" P-state selection algorithm and one for the
"powersave" P-state selection algorithm, but the P-state selection
algorithms are set per policy, so the global limits cannot reflect
all of them at the same time if they are different for different
policies.
In the particular situation above, the attempt to change
min_perf_pct to 94 caused cpufreq_update_policy() to be run
for a CPU with the "powersave" P-state selection algorithm
and intel_pstate_set_policy() called by it silently switched the
global limits to the "powersave" set which finally was reflected
by the sysfs interface.
To prevent that from happening, modify intel_pstate_update_policies()
to always switch back to the set of limits that was used right before
it has been invoked.
Fixes: 111b8b3fe4 (cpufreq: intel_pstate: Always keep all limits settings in sync)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
In the passive mode the cpu_frequency trace event is already
triggered by the cpufreq core or by scaling governors, so
intel_pstate should not trigger it once again for the same
P-state updates.
In addition to that, the frequency returned by
intel_cpufreq_fast_switch() and passed via freqs.new from
intel_cpufreq_target() to cpufreq_freq_transition_end() should
reflect the P-state actually set, so make that happen.
Fixes: 001c76f05b (cpufreq: intel_pstate: Generic governors support)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The intel_pstate_update_perf_limits() called from
intel_cpufreq_verify_policy() may cause global P-state limits
to change which is generally confusing and unnecessary.
In the passive mode the global limits are only applied to the
frequency selected by the scaling governor (they are not taken
into account by governors when making decisions anyway), so making
them follow the per-policy limits serves no purpose and may go
against user expectations (as it generally causes the global
attributes in sysfs to change even though they have not been
written to in some cases).
Fix that by dropping the intel_pstate_update_perf_limits()
invocation from intel_cpufreq_verify_policy() (which also
reduces the code size by a few lines).
This change does not affect the per-CPU limits case, because those
limits allow any P-state to be set by default in the passive mode
and it removes the only piece of code updating them in that mode,
so the per-policy settings will be the only ones taken into account
in that case as expected.
Fixes: 001c76f05b (cpufreq: intel_pstate: Generic governors support)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Using performance_limits in the passive mode doesn't make
sense, because in that mode the global limits are applied to the
frequency selected by the scaling governor.
The maximum and minimum P-state limits in performance_limits are both
set to 100 percent which will put all CPUs into the turbo range
regardless of what governor is used and what frequencies are
selected by it (that is particularly undesirable on CPUs with the
generic powersave governor attached).
For this reason, make intel_pstate_register_driver() always point
limits to powersave_limits in the passive mode.
Fixes: 001c76f05b (cpufreq: intel_pstate: Generic governors support)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Pull sched.h split-up from Ingo Molnar:
"The point of these changes is to significantly reduce the
<linux/sched.h> header footprint, to speed up the kernel build and to
have a cleaner header structure.
After these changes the new <linux/sched.h>'s typical preprocessed
size goes down from a previous ~0.68 MB (~22K lines) to ~0.45 MB (~15K
lines), which is around 40% faster to build on typical configs.
Not much changed from the last version (-v2) posted three weeks ago: I
eliminated quirks, backmerged fixes plus I rebased it to an upstream
SHA1 from yesterday that includes most changes queued up in -next plus
all sched.h changes that were pending from Andrew.
I've re-tested the series both on x86 and on cross-arch defconfigs,
and did a bisectability test at a number of random points.
I tried to test as many build configurations as possible, but some
build breakage is probably still left - but it should be mostly
limited to architectures that have no cross-compiler binaries
available on kernel.org, and non-default configurations"
* 'WIP.sched-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (146 commits)
sched/headers: Clean up <linux/sched.h>
sched/headers: Remove #ifdefs from <linux/sched.h>
sched/headers: Remove the <linux/topology.h> include from <linux/sched.h>
sched/headers, hrtimer: Remove the <linux/wait.h> include from <linux/hrtimer.h>
sched/headers, x86/apic: Remove the <linux/pm.h> header inclusion from <asm/apic.h>
sched/headers, timers: Remove the <linux/sysctl.h> include from <linux/timer.h>
sched/headers: Remove <linux/magic.h> from <linux/sched/task_stack.h>
sched/headers: Remove <linux/sched.h> from <linux/sched/init.h>
sched/core: Remove unused prefetch_stack()
sched/headers: Remove <linux/rculist.h> from <linux/sched.h>
sched/headers: Remove the 'init_pid_ns' prototype from <linux/sched.h>
sched/headers: Remove <linux/signal.h> from <linux/sched.h>
sched/headers: Remove <linux/rwsem.h> from <linux/sched.h>
sched/headers: Remove the runqueue_is_locked() prototype
sched/headers: Remove <linux/sched.h> from <linux/sched/hotplug.h>
sched/headers: Remove <linux/sched.h> from <linux/sched/debug.h>
sched/headers: Remove <linux/sched.h> from <linux/sched/nohz.h>
sched/headers: Remove <linux/sched.h> from <linux/sched/stat.h>
sched/headers: Remove the <linux/gfp.h> include from <linux/sched.h>
sched/headers: Remove <linux/rtmutex.h> from <linux/sched.h>
...
These update turbostat significantly and in particular:
- Default output is now verbose, --debug is no longer required to
get all counters. As a result, some options have been added to
specify exactly what output is wanted.
- Added --quiet to skip system configuration output
- Added --list, --show and --hide parameters
- Added --cpu parameter
- Enhanced Baytrail SoC support
- Added Gemini Lake SoC support
- Added sysfs C-state columns
Also the symbol definitions in arch/x86/include/asm/intel-family.h
and arch/x86/include/asm/msr-index.h are updated and the intel_idle
and intel_pstate drivers are modified to use the updated symbols.
Credits to Len Brown for all of these changes.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
iQIcBAABCAAGBQJYuLyNAAoJEILEb/54YlRxEvkQAJsggzpgGrlhrO6KHSm4yC9M
CqhBVsdeppX1ZTAVPiMk/pcXQYtL5fZ97ELk2So/CjT5Nh3jwDPMA/ux5n3uiob+
O2BTdtxnpNLxPQPQM1mW7Dr/uAIRlJug9gSMxKDbFSU9Oe3aET58PUdUTs7xaT59
nbtLxVSvzrdGk/bX6WO4ic+7F2licJLZPfDGhYidnoika8LxD4M+cIO73gFpgqQi
yoKrTZyLimvneFT0eAUUvHIyKjkJIxeMfslW57uBpz8rW5my+3UwsdpRG4AIVeWc
wSBlsNqj+TuR4BBiZ2VR2RoHF3qbH/SceI+k864BqyThfyK/g2q/vV/GvLZQCR/R
yWcajWD9kvLKvnm1D3XYOIQDBeP4l60j3vVwHytSvmaPYjn5Ms3jq6b+2K6zkXMM
8y3leW/hgw+rGCacdXPrKIlpBykSV7h+TnD2iMxeeDISNkbefWWDe/WB6HncocAg
HDtKRvU9ntRq6/MlnTKbCFM5c0oCXWRw4QNjDy3AsjJELgeAIwiqpHWMKO6XltFj
qU/rdyW/BTCuAlIjWVbjooAIJZ268geupeug3zvE3uGzrxT4DaVIo8W1wtJ+XQrt
By7sOW/gMQ2EcTJQiuFjS/Gz5gOKQ2F8OLCm6T8Prjh6SxrCUAiuIvP0LmxUCa8i
KMlx+8c9E2f9j+TTt9AP
=oMZe
-----END PGP SIGNATURE-----
Merge tag 'pm-turbostat-4.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull turbostat utility updates from Rafael Wysocki:
"Power management turbostat utility updates.
These update turbostat significantly and in particular:
- default output is now verbose, --debug is no longer required to get
all counters. As a result, some options have been added to specify
exactly what output is wanted.
- added --quiet to skip system configuration output
- added --list, --show and --hide parameters
- added --cpu parameter
- enhanced Baytrail SoC support
- added Gemini Lake SoC support
- added sysfs C-state columns
Also the symbol definitions in arch/x86/include/asm/intel-family.h and
arch/x86/include/asm/msr-index.h are updated and the intel_idle and
intel_pstate drivers are modified to use the updated symbols.
Credits to Len Brown for all of these changes"
* tag 'pm-turbostat-4.11-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (44 commits)
tools/power turbostat: version 17.02.24
tools/power turbostat: bugfix: --add u32 was printed as u64
tools/power turbostat: show error on exec
tools/power turbostat: dump p-state software config
tools/power turbostat: show package number, even without --debug
tools/power turbostat: support "--hide C1" etc.
tools/power turbostat: move --Package and --processor into the --cpu option
tools/power turbostat: turbostat.8 update
tools/power turbostat: update --list feature
tools/power turbostat: use wide columns to display large numbers
tools/power turbostat: Add --list option to show available header names
tools/power turbostat: fix zero IRQ count shown in one-shot command mode
tools/power turbostat: add --cpu parameter
tools/power turbostat: print sysfs C-state stats
tools/power turbostat: extend --add option to accept /sys path
tools/power turbostat: skip unused counters on BDX
tools/power turbostat: fix decoding for GLM, DNV, SKX turbo-ratio limits
tools/power turbostat: skip unused counters on SKX
tools/power turbostat: Denverton: use HW CC1 counter, skip C3, C7
tools/power turbostat: initial Gemini Lake SOC support
...
We are going to split <linux/sched/cpufreq.h> out of <linux/sched.h>, which
will have to be picked up from other headers and a couple of .c files.
Create a trivial placeholder <linux/sched/cpufreq.h> file that just
maps to <linux/sched.h> to make this patch obviously correct and
bisectable.
Include the new header in the files that are going to need it.
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Pull changes related to turbostat for v4.11 from Len Brown.
* 'turbostat' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux: (44 commits)
tools/power turbostat: version 17.02.24
tools/power turbostat: bugfix: --add u32 was printed as u64
tools/power turbostat: show error on exec
tools/power turbostat: dump p-state software config
tools/power turbostat: show package number, even without --debug
tools/power turbostat: support "--hide C1" etc.
tools/power turbostat: move --Package and --processor into the --cpu option
tools/power turbostat: turbostat.8 update
tools/power turbostat: update --list feature
tools/power turbostat: use wide columns to display large numbers
tools/power turbostat: Add --list option to show available header names
tools/power turbostat: fix zero IRQ count shown in one-shot command mode
tools/power turbostat: add --cpu parameter
tools/power turbostat: print sysfs C-state stats
tools/power turbostat: extend --add option to accept /sys path
tools/power turbostat: skip unused counters on BDX
tools/power turbostat: fix decoding for GLM, DNV, SKX turbo-ratio limits
tools/power turbostat: skip unused counters on SKX
tools/power turbostat: Denverton: use HW CC1 counter, skip C3, C7
tools/power turbostat: initial Gemini Lake SOC support
...
Originally, these MSRs were locally defined in this driver.
Now the definitions are in msr-index.h -- use them.
Signed-off-by: Len Brown <len.brown@intel.com>
There is a problem with intel_pstate operation mode switching
introduced by commit fb1fe1041c (cpufreq: intel_pstate: Operation
mode control from sysfs), because the global sysfs limits are
preserved across operation modes while per-policy limits are
reinitialized from scratch on a mode switch and both sets of limits
may get out of sync this way.
Fix that by always reinitializing the global limits upon the
registration of the driver.
Fixes: fb1fe1041c (cpufreq: intel_pstate: Operation mode control from sysfs)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Some Kabylake desktop processors may not reach max turbo when running in
HWP mode, even if running under sustained 100% utilization.
This occurs when the HWP.EPP (Energy Performance Preference) is set to
"balance_power" (0x80) -- the default on most systems.
It occurs because the platform BIOS may erroneously enable an
energy-efficiency setting -- MSR_IA32_POWER_CTL BIT-EE, which is not
recommended to be enabled on this SKU.
On the failing systems, this BIOS issue was not discovered when the
desktop motherboard was tested with Windows, because the BIOS also
neglects to provide the ACPI/CPPC table, that Windows requires to enable
HWP, and so Windows runs in legacy P-state mode, where this setting has
no effect.
Linux' intel_pstate driver does not require ACPI/CPPC to enable HWP, and
so it runs in HWP mode, exposing this incorrect BIOS configuration.
There are several ways to address this problem.
First, Linux can also run in legacy P-state mode on this system.
As intel_pstate is how Linux enables HWP, booting with
"intel_pstate=disable"
will run in acpi-cpufreq/ondemand legacy p-state mode.
Or second, the "performance" governor can be used with intel_pstate,
which will modify HWP.EPP to 0.
Or third, starting in 4.10, the
/sys/devices/system/cpu/cpufreq/policy*/energy_performance_preference
attribute in can be updated from "balance_power" to "performance".
Or fourth, apply this patch, which fixes the erroneous setting of
MSR_IA32_POWER_CTL BIT_EE on this model, allowing the default
configuration to function as designed.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Reviewed-by: Len Brown <len.brown@intel.com>
Cc: 4.6+ <stable@vger.kernel.org> # 4.6+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
When HWP is active, turbo activation ratio is not used to calculate max
non turbo ratio. But on these systems the max non turbo ratio is decided
by config TDP settings.
This change removes usage of MSR_TURBO_ACTIVATION_RATIO for HWP systems,
instead directly use TDP ratios, when more than one TDPs are available.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Under HWP the performance limits are calculated using max_perf_pct
and min_perf_pct using possible performance, not available performance.
The available performance can be reduced by no_turbo setting. To make
compatible with legacy mode, use max/min performance percentage with
respect to available performance.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
When turbo is not disabled by BIOS, but user disabled from intel P-State
sysfs and changes max/min using cpufreq sysfs, the resultant frequency
is lower than what user requested.
The reason for this, when the perf limits are calculated in set_policy()
callback, they are with reference to max cpu frequency (turbo frequency
), but when enforced in the intel_pstate_get_min_max() they are with
reference to max available performance as documented in the intel_pstate
documentation (in this case max non turbo P-State).
This needs similar change as done in intel_cpufreq_verify_policy() for
passive mode. Set policy->cpuinfo.max_freq based on the turbo status.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Make it possible to change the operation mode of intel_pstate with
the help of a new sysfs attribute called "status".
There are three possible configurations that can be selected using
this attribute:
"off" - The driver is not in use at this time.
"active" - The driver works as a P-state governor (default).
"passive" - The driver works as a regular cpufreq one and collaborates
with the generic cpufreq governors (it sets P-states as
requested by those governors). [This is the same mode
the driver can be started in by passing intel_pstate=passive
in the kernel command line.]
The current setting is returned by reads from this attribute. Writing
one of the above strings to it changes the operation mode as indicated
by that string, if possible.
If HW-managed P-states (HWP) feature is enabled, it is not possible
to change the driver's operation mode and attempts to write to this
attribute will fail.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Expose the intel_pstate's global sysfs attributes before registering
the driver to prepare for the addition of an attribute that also will
have to work if the driver is not registered.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
A side effect of keeping intel_pstate sysfs limits in sync with cpufreq
is that the now sysfs limits can't enforced under performance policy.
For example, if the max_perf_pct is changed from 100 to 80, this will call
intel_pstate_set_policy(), which will change the max_perf to 100 again for
performance policy. Same issue happens, when no_turbo is set.
This change calculates max and min frequency using sysfs performance
limits in intel_pstate_verify_policy() and adjusts policy limits by
calling cpufreq_verify_within_limits().
Also, it causes the setting of performance limits to be skipped if
no_turbo is set.
Fixes: 111b8b3fe4 (cpufreq: intel_pstate: Always keep all limits settings in sync)
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
* pm-cpufreq:
cpufreq: dt: Add support for APM X-Gene 2
cpufreq: intel_pstate: Always keep all limits settings in sync
cpufreq: intel_pstate: Use locking in intel_cpufreq_verify_policy()
cpufreq: intel_pstate: Use locking in intel_pstate_resume()
cpufreq: intel_pstate: Do not expose PID parameters in passive mode
Make intel_pstate update per-logical-CPU limits when the global
settings are changed to ensure that they are always in sync and
users will not see confusing values in per-logical-CPU sysfs
attributes.
This also fixes the problem that setting the "no_turbo" global
attribute to 1 in the "passive" mode (ie. when intel_pstate acts
as a regular cpufreq driver) when scaling_governor is set to
"performance" has no effect.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Race conditions are possible if intel_cpufreq_verify_policy()
is executed in parallel with global limits updates from sysfs,
so the invocation of intel_pstate_update_perf_limits() in it
should be carried out under intel_pstate_limits_lock.
Make that happen.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Theoretically, intel_pstate_resume() may be executed in parallel
with intel_pstate_set_policy(), if the latter is invoked via
cpufreq_update_policy() as a result of a notification, so use
intel_pstate_limits_lock in there too to avoid race conditions.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
If intel_pstate works in the passive mode in which it acts as
a regular cpufreq driver and collaborates with generic cpufreq
governors, the PID parameters are not used, so do not expose
them via debugfs in that case.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
- New cpufreq driver for Broadcom STB SoCs and a Device Tree binding
for it (Markus Mayer).
- Support for ARM Integrator/AP and Integrator/CP in the generic
DT cpufreq driver and elimination of the old Integrator cpufreq
driver (Linus Walleij).
- Support for the zx296718, r8a7743 and r8a7745, Socionext UniPhier,
and PXA SoCs in the the generic DT cpufreq driver (Baoyou Xie,
Geert Uytterhoeven, Masahiro Yamada, Robert Jarzmik).
- cpufreq core fix to eliminate races that may lead to using
inactive policy objects and related cleanups (Rafael Wysocki).
- cpufreq schedutil governor update to make it use SCHED_FIFO
kernel threads (instead of regular workqueues) for doing delayed
work (to reduce the response latency in some cases) and related
cleanups (Viresh Kumar).
- New cpufreq sysfs attribute for resetting statistics (Markus
Mayer).
- cpufreq governors fixes and cleanups (Chen Yu, Stratos Karafotis,
Viresh Kumar).
- Support for using generic cpufreq governors in the intel_pstate
driver (Rafael Wysocki).
- Support for per-logical-CPU P-state limits and the EPP/EPB
(Energy Performance Preference/Energy Performance Bias) knobs
in the intel_pstate driver (Srinivas Pandruvada).
- New CPU ID for Knights Mill in intel_pstate (Piotr Luc).
- intel_pstate driver modification to use the P-state selection
algorithm based on CPU load on platforms with the system profile
in the ACPI tables set to "mobile" (Srinivas Pandruvada).
- intel_pstate driver cleanups (Arnd Bergmann, Rafael Wysocki,
Srinivas Pandruvada).
- cpufreq powernv driver updates including fast switching support
(for the schedutil governor), fixes and cleanus (Akshay Adiga,
Andrew Donnellan, Denis Kirjanov).
- acpi-cpufreq driver rework to switch it over to the new CPU
offline/online state machine (Sebastian Andrzej Siewior).
- Assorted cleanups in cpufreq drivers (Wei Yongjun, Prashanth
Prakash).
- Idle injection rework (to make it use the regular idle path
instead of a home-grown custom one) and related powerclamp
thermal driver updates (Peter Zijlstra, Jacob Pan, Petr Mladek,
Sebastian Andrzej Siewior).
- New CPU IDs for Atom Z34xx and Knights Mill in intel_idle (Andy
Shevchenko, Piotr Luc).
- intel_idle driver cleanups and switch over to using the new CPU
offline/online state machine (Anna-Maria Gleixner, Sebastian
Andrzej Siewior).
- cpuidle DT driver update to support suspend-to-idle properly
(Sudeep Holla).
- cpuidle core cleanups and misc updates (Daniel Lezcano, Pan Bian,
Rafael Wysocki).
- Preliminary support for power domains including CPUs in the
generic power domains (genpd) framework and related DT bindings
(Lina Iyer).
- Assorted fixes and cleanups in the generic power domains (genpd)
framework (Colin Ian King, Dan Carpenter, Geert Uytterhoeven).
- Preliminary support for devices with multiple voltage regulators
and related fixes and cleanups in the Operating Performance Points
(OPP) library (Viresh Kumar, Masahiro Yamada, Stephen Boyd).
- System sleep state selection interface rework to make it easier
to support suspend-to-idle as the default system suspend method
(Rafael Wysocki).
- PM core fixes and cleanups, mostly related to the interactions
between the system suspend and runtime PM frameworks (Ulf Hansson,
Sahitya Tummala, Tony Lindgren).
- Latency tolerance PM QoS framework imorovements (Andrew
Lutomirski).
- New Knights Mill CPU ID for the Intel RAPL power capping driver
(Piotr Luc).
- Intel RAPL power capping driver fixes, cleanups and switch over
to using the new CPU offline/online state machine (Jacob Pan,
Thomas Gleixner, Sebastian Andrzej Siewior).
- Fixes and cleanups in the exynos-ppmu, exynos-nocp, rk3399_dmc,
rockchip-dfi devfreq drivers and the devfreq core (Axel Lin,
Chanwoo Choi, Javier Martinez Canillas, MyungJoo Ham, Viresh
Kumar).
- Fix for false-positive KASAN warnings during resume from ACPI S3
(suspend-to-RAM) on x86 (Josh Poimboeuf).
- Memory map verification during resume from hibernation on x86 to
ensure a consistent address space layout (Chen Yu).
- Wakeup sources debugging enhancement (Xing Wei).
- rockchip-io AVS driver cleanup (Shawn Lin).
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2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=vhxI
-----END PGP SIGNATURE-----
Merge tag 'pm-4.10-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull power management updates from Rafael Wysocki:
"Again, cpufreq gets more changes than the other parts this time (one
new driver, one old driver less, a bunch of enhancements of the
existing code, new CPU IDs, fixes, cleanups)
There also are some changes in cpuidle (idle injection rework, a
couple of new CPU IDs, online/offline rework in intel_idle, fixes and
cleanups), in the generic power domains framework (mostly related to
supporting power domains containing CPUs), and in the Operating
Performance Points (OPP) library (mostly related to supporting devices
with multiple voltage regulators)
In addition to that, the system sleep state selection interface is
modified to make it easier for distributions with unchanged user space
to support suspend-to-idle as the default system suspend method, some
issues are fixed in the PM core, the latency tolerance PM QoS
framework is improved a bit, the Intel RAPL power capping driver is
cleaned up and there are some fixes and cleanups in the devfreq
subsystem
Specifics:
- New cpufreq driver for Broadcom STB SoCs and a Device Tree binding
for it (Markus Mayer)
- Support for ARM Integrator/AP and Integrator/CP in the generic DT
cpufreq driver and elimination of the old Integrator cpufreq driver
(Linus Walleij)
- Support for the zx296718, r8a7743 and r8a7745, Socionext UniPhier,
and PXA SoCs in the the generic DT cpufreq driver (Baoyou Xie,
Geert Uytterhoeven, Masahiro Yamada, Robert Jarzmik)
- cpufreq core fix to eliminate races that may lead to using inactive
policy objects and related cleanups (Rafael Wysocki)
- cpufreq schedutil governor update to make it use SCHED_FIFO kernel
threads (instead of regular workqueues) for doing delayed work (to
reduce the response latency in some cases) and related cleanups
(Viresh Kumar)
- New cpufreq sysfs attribute for resetting statistics (Markus Mayer)
- cpufreq governors fixes and cleanups (Chen Yu, Stratos Karafotis,
Viresh Kumar)
- Support for using generic cpufreq governors in the intel_pstate
driver (Rafael Wysocki)
- Support for per-logical-CPU P-state limits and the EPP/EPB (Energy
Performance Preference/Energy Performance Bias) knobs in the
intel_pstate driver (Srinivas Pandruvada)
- New CPU ID for Knights Mill in intel_pstate (Piotr Luc)
- intel_pstate driver modification to use the P-state selection
algorithm based on CPU load on platforms with the system profile in
the ACPI tables set to "mobile" (Srinivas Pandruvada)
- intel_pstate driver cleanups (Arnd Bergmann, Rafael Wysocki,
Srinivas Pandruvada)
- cpufreq powernv driver updates including fast switching support
(for the schedutil governor), fixes and cleanus (Akshay Adiga,
Andrew Donnellan, Denis Kirjanov)
- acpi-cpufreq driver rework to switch it over to the new CPU
offline/online state machine (Sebastian Andrzej Siewior)
- Assorted cleanups in cpufreq drivers (Wei Yongjun, Prashanth
Prakash)
- Idle injection rework (to make it use the regular idle path instead
of a home-grown custom one) and related powerclamp thermal driver
updates (Peter Zijlstra, Jacob Pan, Petr Mladek, Sebastian Andrzej
Siewior)
- New CPU IDs for Atom Z34xx and Knights Mill in intel_idle (Andy
Shevchenko, Piotr Luc)
- intel_idle driver cleanups and switch over to using the new CPU
offline/online state machine (Anna-Maria Gleixner, Sebastian
Andrzej Siewior)
- cpuidle DT driver update to support suspend-to-idle properly
(Sudeep Holla)
- cpuidle core cleanups and misc updates (Daniel Lezcano, Pan Bian,
Rafael Wysocki)
- Preliminary support for power domains including CPUs in the generic
power domains (genpd) framework and related DT bindings (Lina Iyer)
- Assorted fixes and cleanups in the generic power domains (genpd)
framework (Colin Ian King, Dan Carpenter, Geert Uytterhoeven)
- Preliminary support for devices with multiple voltage regulators
and related fixes and cleanups in the Operating Performance Points
(OPP) library (Viresh Kumar, Masahiro Yamada, Stephen Boyd)
- System sleep state selection interface rework to make it easier to
support suspend-to-idle as the default system suspend method
(Rafael Wysocki)
- PM core fixes and cleanups, mostly related to the interactions
between the system suspend and runtime PM frameworks (Ulf Hansson,
Sahitya Tummala, Tony Lindgren)
- Latency tolerance PM QoS framework imorovements (Andrew Lutomirski)
- New Knights Mill CPU ID for the Intel RAPL power capping driver
(Piotr Luc)
- Intel RAPL power capping driver fixes, cleanups and switch over to
using the new CPU offline/online state machine (Jacob Pan, Thomas
Gleixner, Sebastian Andrzej Siewior)
- Fixes and cleanups in the exynos-ppmu, exynos-nocp, rk3399_dmc,
rockchip-dfi devfreq drivers and the devfreq core (Axel Lin,
Chanwoo Choi, Javier Martinez Canillas, MyungJoo Ham, Viresh Kumar)
- Fix for false-positive KASAN warnings during resume from ACPI S3
(suspend-to-RAM) on x86 (Josh Poimboeuf)
- Memory map verification during resume from hibernation on x86 to
ensure a consistent address space layout (Chen Yu)
- Wakeup sources debugging enhancement (Xing Wei)
- rockchip-io AVS driver cleanup (Shawn Lin)"
* tag 'pm-4.10-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (127 commits)
devfreq: rk3399_dmc: Don't use OPP structures outside of RCU locks
devfreq: rk3399_dmc: Remove dangling rcu_read_unlock()
devfreq: exynos: Don't use OPP structures outside of RCU locks
Documentation: intel_pstate: Document HWP energy/performance hints
cpufreq: intel_pstate: Support for energy performance hints with HWP
cpufreq: intel_pstate: Add locking around HWP requests
PM / sleep: Print active wakeup sources when blocking on wakeup_count reads
PM / core: Fix bug in the error handling of async suspend
PM / wakeirq: Fix dedicated wakeirq for drivers not using autosuspend
PM / Domains: Fix compatible for domain idle state
PM / OPP: Don't WARN on multiple calls to dev_pm_opp_set_regulators()
PM / OPP: Allow platform specific custom set_opp() callbacks
PM / OPP: Separate out _generic_set_opp()
PM / OPP: Add infrastructure to manage multiple regulators
PM / OPP: Pass struct dev_pm_opp_supply to _set_opp_voltage()
PM / OPP: Manage supply's voltage/current in a separate structure
PM / OPP: Don't use OPP structure outside of rcu protected section
PM / OPP: Reword binding supporting multiple regulators per device
PM / OPP: Fix incorrect cpu-supply property in binding
cpuidle: Add a kerneldoc comment to cpuidle_use_deepest_state()
..
It is possible to provide hints to the HWP algorithms in the processor
to be more performance centric to more energy centric. These hints are
provided by using HWP energy performance preference (EPP) or energy
performance bias (EPB) settings.
The scope of these settings is per logical processor, which means that
each of the logical processors in the package can be programmed with a
different value.
This change provides cpufreq sysfs interface to provide hint. For each
policy, two additional attributes will be available to check and provide
hint. These attributes will only be present when the intel_pstate driver
is using HWP mode.
These attributes are:
- energy_performance_available_preferences
- energy_performance_preference
To get list of supported hints:
$ cat energy_performance_available_preferences
default performance balance_performance balance_power power
The current preference can be read or changed via cpufreq sysfs
attribute "energy_performance_preference". Reading from this attribute
will display current effective setting changed via any method. User can
write any of the valid preference string to this attribute. User can
always restore to power-on default by writing "default".
Implementation
Since these hints can be provided by direct MSR write or using some tools
like x86_energy_perf_policy, the driver internally doesn't maintain any
state. The user operation will result in direct read/write of MSR: 0x774
(HWP_REQUEST_MSR). Also driver use read modify write to update other
fields in this MSR.
Summary of changes:
- struct cpudata field epp_saved is renamed to epp_powersave, as this
stores the value to restore once policy is switched from performance
to powersave to restore original powersave EPP value.
- A new struct cpudata field epp_saved is used to store the raw MSR
EPP/EPB value when a CPU goes offline or on suspend and restore on
online/resume. This ensures that EPP value is restored to correct
value irrespective of the means used to set.
- EPP/EPB value ranges are fixed for each preference, which can be
set for the cpufreq sysfs, so user request is mapped to/from this
range.
- New attributes are only added when HWP is present.
- Since EPP value of 0 is valid the fields are initialized to
-EINVAL when not valid. The field epp_default is read only once
after powerup to avoid reading on subsequent CPU online operation
- New suspend callback to store epp on suspend operation
- Don't invalidate old epp_saved field on resume and online as now
we can restore last epp value on suspend and this field can still
have old EPP value sampled during switch to performance from
powersave.
- While here optimized setting of cpu_data->epp_powersave = epp in
intel_pstate_hwp_set() as this was done in both true and false
paths.
- epp/epb set function returns error to caller on failure to pass
on to user space for display.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
To avoid race conditions from multiple threads, increase the scope
of intel_pstate_limits_lock to include HWP requests also.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
[ rjw: Subject ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Add Knights Mill (KNM) to the list of CPUIDs supported by intel_pstate.
Signed-off-by: Piotr Luc <piotr.luc@intel.com>
Reviewed-by: Dave Hansen <dave.hansen@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The addition of the generic governor support marked the
intel_pstate_exit_perf_limits as inline(), which fixed a warning,
but it introduced another warning:
drivers/cpufreq/intel_pstate.c: In function ‘intel_pstate_exit_perf_limits’:
drivers/cpufreq/intel_pstate.c:483:1: error: no return statement in function returning non-void [-Werror=return-type]
This changes it back to a 'void' return type, and changes the
corresponding intel_pstate_init_acpi_perf_limits() function to
be inline as well for consistency.
Fixes: 001c76f05b (cpufreq: intel_pstate: Generic governors support)
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>
When user has selected performance policy, then set the EPP (Energy
Performance Preference) or EPB (Energy Performance Bias) to maximum
performance mode.
Also when user switch back to powersave, then restore EPP/EPB to last
EPP/EPB value before entering performance mode. If user has not changed
EPP/EPB manually then it will be power on default value.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Even with round up of limits->min_perf and limits->max_perf, in some
cases resultant performance is 100 MHz less than the desired.
For example when the maximum frequency is 3.50 GHz, setting
scaling_min_frequency to 2.3 GHz always results in 2.2 GHz minimum.
Currently the fixed floating point operation uses 8 bit precision for
calculating limits->min_perf and limits->max_perf. For some operations
in this driver the 14 bit precision is used. Using the 14 bit precision
also for calculating limits->min_perf and limits->max_perf, addresses
this issue.
Introduced fp_ext_toint() equivalent to fp_toint() and int_ext_tofp()
equivalent to int_tofp() with 14 bit precision.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
In some use cases, user wants to enforce a minimum performance limit on
CPUs. But because of simple division the resultant performance is 100 MHz
less than the desired in some cases.
For example when the maximum frequency is 3.50 GHz, setting
scaling_min_frequency to 1.6 GHz always results in 1.5 GHz minimum. With
simple round up, the frequency can be set to 1.6 GHz to minimum in this
case. This round up is already done to max_policy_pct and max_perf, so do
the same for min_policy_pct and min_perf.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
There may be reasons to use generic cpufreq governors (eg. schedutil)
on Intel platforms instead of the intel_pstate driver's internal
governor. However, that currently can only be done by disabling
intel_pstate altogether and using the acpi-cpufreq driver instead
of it, which is subject to limitations.
First of all, acpi-cpufreq only works on systems where the _PSS
object is present in the ACPI tables for all logical CPUs. Second,
on those systems acpi-cpufreq will only use frequencies listed by
_PSS which may be suboptimal. In particular, by convention, the
whole turbo range is represented in _PSS as a single P-state and
the frequency assigned to it is greater by 1 MHz than the greatest
non-turbo frequency listed by _PSS. That may confuse governors to
use turbo frequencies less frequently which may lead to suboptimal
performance.
For this reason, make it possible to use the intel_pstate driver
with generic cpufreq governors as a "normal" cpufreq driver. That
mode is enforced by adding intel_pstate=passive to the kernel
command line and cannot be disabled at run time. In that mode,
intel_pstate provides a cpufreq driver interface including
the ->target() and ->fast_switch() callbacks and is listed in
scaling_driver as "intel_cpufreq".
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: Doug Smythies <dsmythies@telus.net>
Currently, intel_pstate is unable to control P-states on my
IvyBridge-based Acer Aspire S5, because they are controlled by SMM
on that machine by default and it is necessary to request OS control
of P-states from it via the SMI Command register exposed in the ACPI
FADT. intel_pstate doesn't do that now, but acpi-cpufreq and other
cpufreq drivers for x86 platforms do.
Address this problem by making intel_pstate use the ACPI-defined
mechanism as well. However, intel_pstate is not modular and it
doesn't need the module refcount tricks played by
acpi_processor_notify_smm(), so export the core of this function
to it as acpi_processor_pstate_control() and make it call that.
[The changes in processor_perflib.c related to this should not
make any functional difference for the acpi_processor_notify_smm()
users].
To be safe, only call acpi_processor_notify_smm() from intel_pstate
if ACPI _PPC support is enabled in it.
Suggested-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Use get_target_pstate_use_cpu_load() to calculate target P-State for
devices, with the preferred power management profile in ACPI FADT
set to PM_MOBILE.
This may help in resolving some thermal issues caused by low sustained
cpu bound workloads. The current algorithm tend to over provision in this
case as it doesn't look at the CPU busyness.
Also included the fix from Arnd Bergmann <arnd@arndb.de> to solve compile
issue, when CONFIG_ACPI is not defined.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The limits variable gets modified from intel_pstate sysfs and also gets
modified from cpufreq sysfs. So protect with a mutex to keep data
integrity, when they are getting modified from multiple threads.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
When policy->max and policy->min are same, in some cases they don't
result in the same frequency cap. The max_policy_pct is rounded up but
not min_perf_pct. So even when they are same, results in different
percentage or maximum and minimum.
Since minimum is a conservative value for power, a lower value without
rounding is better in most of the cases, unless user wants
policy->max = policy->min.
This change uses use the same policy percentage when policy->max and
policy->min are same.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Intel P-State offers two interface to set performance limits:
- Intel P-State sysfs
/sys/devices/system/cpu/intel_pstate/max_perf_pct
/sys/devices/system/cpu/intel_pstate/min_perf_pct
- cpufreq
/sys/devices/system/cpu/cpu*/cpufreq/scaling_max_freq
/sys/devices/system/cpu/cpu*/cpufreq/scaling_min_freq
In the current implementation both of the above methods, change limits
to every CPU in the system. Moreover the limits placed using cpufreq
policy interface also presented in the Intel P-State sysfs via modified
max_perf_pct and min_per_pct during sysfs reads. This allows to check
percent of reduced/increased performance, irrespective of method used to
limit.
There are some new generations of processors, where it is possible to
have limits placed on individual CPU cores. Using cpufreq interface it
is possible to set limits on each CPU. But the current processing will
use last limits placed on all CPUs. So the per core limit feature of
CPUs can't be used.
This change brings in capability to set P-States limits for each CPU,
with some limitations. In this case what should be the read of
max_perf_pct and min_perf_pct? It can be most restrictive limits placed
on any CPU or max possible performance on any given CPU on which no
limits are placed. In either case someone will have issue.
So the consensus is, we can't have both sysfs controls present when user
wants to use limit per core limits.
- By default per-core-control feature is not enabled. So no one will
notice any difference.
- The way to enable is by kernel command line
intel_pstate=per_cpu_perf_limits
- When the per-core-controls are enabled there is no display of for both
read and write on
/sys/devices/system/cpu/intel_pstate/max_perf_pct
/sys/devices/system/cpu/intel_pstate/min_perf_pct
- User can change limits using
/sys/devices/system/cpu/cpu*/cpufreq/scaling_max_freq
/sys/devices/system/cpu/cpu*/cpufreq/scaling_min_freq
/sys/devices/system/cpu/cpu*/cpufreq/scaling_governor
- User can still observe turbo percent and number of P-States from
/sys/devices/system/cpu/intel_pstate/turbo_pct
/sys/devices/system/cpu/intel_pstate/num_pstates
- User can read write system wide turbo status
/sys/devices/system/cpu/no_turbo
While changing this BUG_ON is changed to WARN_ON, as they are not fatal
errors for the system.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The only times at which intel_pstate checks the policy set for
a given CPU is the initialization of that CPU and updates of its
policy settings from cpufreq when intel_pstate_set_policy() is
invoked.
That is insufficient, however, because intel_pstate uses the same
P-state selection function for all CPUs regardless of the policy
setting for each of them and the P-state limits are shared between
them. Thus if the policy is set to "performance" for a particular
CPU, it may not behave as expected if the cpufreq settings are
changed subsequently for another CPU.
That can be easily demonstrated by writing "performance" to
scaling_governor for all CPUs and then switching it to "powersave"
for one of them in which case all of the CPUs will behave as though
their scaling_governor were all "powersave" (even though the policy
still appears to be "performance" for the remaining CPUs).
Fix this problem by modifying intel_pstate_adjust_busy_pstate() to
always set the P-state to the maximum allowed by the current limits
for all CPUs whose policy is set to "performance".
Note that it still is recommended to always change the policy setting
in the same way for all CPUs even with this fix applied to avoid
confusion.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
After commit a4675fbc4a (cpufreq: intel_pstate: Replace timers with
utilization update callbacks) the cpufreq governor callbacks may not
be invoked on NOHZ_FULL CPUs and, in particular, switching to the
"performance" policy via sysfs may not have any effect on them. That
is a problem, because it usually is desirable to squeeze the last
bit of performance out of those CPUs, so work around it by setting
the maximum P-state (within the limits) in intel_pstate_set_policy()
upfront when the policy is CPUFREQ_POLICY_PERFORMANCE.
Fixes: a4675fbc4a (cpufreq: intel_pstate: Replace timers with utilization update callbacks)
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
When target state is calculated using get_target_pstate_use_cpu_load(),
PID controller is not used, hence it has no effect on performance.
So don't present debugfs entries to tune PID controller.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
