with the purpose of using such hints when making scheduling decisions
- Determine the boost enumerator for each AMD core based on its type: efficiency
or performance, in the cppc driver
- Add the type of a CPU to the topology CPU descriptor with the goal of
supporting and making decisions based on the type of the respective core
- Add a feature flag to denote AMD cores which have heterogeneous topology and
enable SD_ASYM_PACKING for those
- Check microcode revisions before disabling PCID on Intel
- Cleanups and fixlets
-----BEGIN PGP SIGNATURE-----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=420r
-----END PGP SIGNATURE-----
Merge tag 'x86_cpu_for_v6.13' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 cpuid updates from Borislav Petkov:
- Add a feature flag which denotes AMD CPUs supporting workload
classification with the purpose of using such hints when making
scheduling decisions
- Determine the boost enumerator for each AMD core based on its type:
efficiency or performance, in the cppc driver
- Add the type of a CPU to the topology CPU descriptor with the goal of
supporting and making decisions based on the type of the respective
core
- Add a feature flag to denote AMD cores which have heterogeneous
topology and enable SD_ASYM_PACKING for those
- Check microcode revisions before disabling PCID on Intel
- Cleanups and fixlets
* tag 'x86_cpu_for_v6.13' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/cpu: Remove redundant CONFIG_NUMA guard around numa_add_cpu()
x86/cpu: Fix FAM5_QUARK_X1000 to use X86_MATCH_VFM()
x86/cpu: Fix formatting of cpuid_bits[] in scattered.c
x86/cpufeatures: Add X86_FEATURE_AMD_WORKLOAD_CLASS feature bit
x86/amd: Use heterogeneous core topology for identifying boost numerator
x86/cpu: Add CPU type to struct cpuinfo_topology
x86/cpu: Enable SD_ASYM_PACKING for PKG domain on AMD
x86/cpufeatures: Add X86_FEATURE_AMD_HETEROGENEOUS_CORES
x86/cpufeatures: Rename X86_FEATURE_FAST_CPPC to have AMD prefix
x86/mm: Don't disable PCID when INVLPG has been fixed by microcode
Variables annotated with __free() need to be initialized if the function
can return before they get updated for the first time or the attempt to
free the memory pointed to by them upon function return may crash the
kernel.
Fix this issue in some places in the thermal testing code.
Fixes: f6a034f2df ("thermal: Introduce a debugfs-based testing facility")
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/12576267.O9o76ZdvQC@rjwysocki.net
Use DEFINE_FREE() to define a __free function for dropping thermal
zone template reference counters and use it along with __free() to
simplify code in some places.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/4628747.LvFx2qVVIh@rjwysocki.net
[ rjw: Add variable initialization to address compiler warning ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Notice that tt_get_tt_zone() need not use the ret variable in
the tt_thermal_zones list walk and make it return right after
incrementing the reference counter of the matching entry.
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/6120304.lOV4Wx5bFT@rjwysocki.net
Although the current device tree binding of thermal zones require the
trips subnode, the binding in kernel v5.15 does not require it, and many
device trees shipped with the kernel, for example,
allwinner/sun50i-a64.dtsi and mediatek/mt8183-kukui.dtsi in ARM64, still
comply to the old binding and contain no trips subnode.
Allow the code to successfully register thermal zones w/o trips subnode
for DT binding compatibility now.
Furtherly, the inconsistency between DTs and bindings should be resolved
by either adding empty trips subnode or dropping the trips subnode
requirement.
Fixes: d0c75fa2c1 ("thermal/of: Initialize trip points separately")
Signed-off-by: Icenowy Zheng <uwu@icenowy.me>
[wenst@chromium.org: Reworked logic and kernel log messages]
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Reviewed-by: Rafael J. Wysocki <rafael@kernel.org>
Link: https://lore.kernel.org/r/20241018073139.1268995-1-wenst@chromium.org
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
When total_req_power is 0, divvy_up_power() will set granted_power to 0,
and cdev will be limited to the lowest performance. If our polling delay
is set to 200ms, it means that cdev cannot perform better within 200ms
even if cdev has a sudden load. This will affect the performance of cdev
and is not as expected.
For this reason, if nobody requests power, then set the granted power to
the max_power.
Signed-off-by: ZhengShaobo <zhengshaobo1@xiaomi.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/20241021121138.422-1-zhengshaobo1@xiaomi.com
[ rjw: Fixed up tags ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Move thermal_zone_device_init() along with thermal_zone_device_check()
closer to the callers of the former, where they fit better together.
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/1906685.CQOukoFCf9@rjwysocki.net
Modify the thermal core to use three lists of trip points:
trips_high, containing trips with thresholds strictly above the current
thermal zone temperature,
trips_reached, containing trips with thresholds at or below the current
zone temperature,
trips_invalid, containing trips with temperature equal to
THERMAL_ZONE_INVALID,
where the first two lists are always sorted by the current trip
threshold.
For each trip in trips_high, there is no mitigation under way and
the trip threshold is equal to its temperature. In turn, for each
trip in trips_reached, there is mitigation under way and the trip
threshold is equal to its low temperature. The trips in trips_invalid,
of course, need not be taken into consideration.
The idea is to make __thermal_zone_device_update() walk trips_high and
trips_reached instead of walking the entire table of trip points in a
thermal zone. Usually, it will only need to walk a few entries in one
of the lists and check one entry in the other list, depending on the
direction of the zone temperature changes, because crossing many trips
by the zone temperature in one go between two consecutive temperature
checks should be unlikely (if it occurs often, the thermal zone
temperature should probably be checked more often either or there
are too many trips).
This also helps to eliminate one temporary trip list used for trip
crossing notification (only one temporary list is needed for this
purpose instead of two) and the remaining temporary list may be sorted
by the current trip threshold value, like the trips_reached list, so
the additional notify_temp field in struct thermal_trip_desc is not
necessary any more.
Moreover, since the trips_reached and trips_high lists are sorted,
the "low" and "high" values needed by thermal_zone_set_trips() can be
determined in a straightforward way by looking at one end of each list.
Of course, additional work is needed in some places in order to
maintain the ordering of the lists, but it is limited to situations
that should be rare, like updating a trip point temperature or
hysteresis, thermal zone initialization, or system resume.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/2003443.usQuhbGJ8B@rjwysocki.net
[ rjw: Added a comment to thermal_zone_handle_trips() ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Since thermal_zone_set_trip_temp() is now located in the same file
as thermal_trip_crossed(), it can invoke the latter directly without
using the thermal_zone_trip_down() wrapper that has no other users.
Update thermal_zone_set_trip_temp() accordingly and drop
thermal_zone_trip_down().
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/1807510.VLH7GnMWUR@rjwysocki.net
In preparation for subsequent changes, move two functions used
for updating trip points, thermal_zone_set_trip_temp() and
thermal_zone_set_trip_hyst(), to thermal_core.c.
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/3248558.5fSG56mABF@rjwysocki.net
Notice that some processing related to trip point crossing carried out
in handle_thermal_trip() and thermal_zone_set_trip_temp() may as well
be done in thermal_trip_crossed(), which allows code duplication to be
reduced, so change the code accordingly.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/1982859.PYKUYFuaPT@rjwysocki.net
In preparation for subsequent changes, modify thermal_trip_crossed()
to take a trip descriptor pointer instead of a pointer to struct
thermal_trip and propagate this change to thermal_zone_trip_down().
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/10547668.nUPlyArG6x@rjwysocki.net
In preparation for subsequent changes, move the invocations of
thermal_thresholds_handle() and thermal_zone_set_trips() in
__thermal_zone_device_update() after the processing of the
temporary trip lists.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/3323276.44csPzL39Z@rjwysocki.net
Subsequently, trips will be moved between sorted lists in multiple
places, so replace add_trip_to_sorted_list() with an analogous
function, move_trip_to_sorted_list(), that will move a given trip
to a given sorted list.
To allow list_del() used in the new function to work, initialize the
list_node fields in trip descriptors where applicable so they are
always valid.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/2960197.e9J7NaK4W3@rjwysocki.net
Since the list node field in struct thermal_trip_desc is going to be
used for purposes other than trip crossing notification, rename it
to list_node.
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/2201558.irdbgypaU6@rjwysocki.net
Since it is not expected that multiple trip points will be crossed
in one go very often (if this happens, there are too many trip points
in the given thermal zone or they are checked too rarely), quite likely
it is more efficient to build a sorted list of crossed trip points than
to put them on an unsorted list and sort it later.
Moreover, trip points are often sorted in ascending temperature order
during thermal zone registration, so building a sorted list out of
them is quite straightforward and relatively inexpensive.
Accordingly, make handle_thermal_trip() maintain list ordering when
adding trip points to the lists and get rid of separate list sorting
in __thermal_zone_device_update().
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/4930656.GXAFRqVoOG@rjwysocki.net
When the thermal zone is unregistered (thermal sensor module being
unloaded), no lock is held when flushing the thresholds. That results
in a WARN when the lockdep validation is set in the kernel config.
This has been reported by syzbot.
As the thermal zone is in the process of being destroyed, there is no
need to send a notification about purging the thresholds to the
userspace as this one will receive a thermal zone deletion
notification which imply the deletion of all the associated resources
like the trip points or the user thresholds.
Split the function thermal_thresholds_flush() into a lockless one
without notification and its call with the lock annotation followed
with the thresholds flushing notification.
Please note this scenario is unlikely to happen, as the sensor drivers
are usually compiled-in in order to have the thermal framework to be
able to kick in at boot time if needed.
Fixes: 445936f9e2 ("thermal: core: Add user thresholds support")
Link: https://lore.kernel.org/all/67124175.050a0220.10f4f4.0012.GAE@google.com
Reported-by: syzbot+f24dd060c1911fe54c85@syzkaller.appspotmail.com
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://patch.msgid.link/20241024102303.1086147-1-daniel.lezcano@linaro.org
[ rjw: Subject edit, added Fixes tag ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The thresholds exist but there is no notification neither action code
related to them yet.
These changes implement the netlink for the notifications when the
thresholds are crossed, added, deleted or flushed as well as the
commands which allows to get the list of the thresholds, flush them,
add and delete.
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://patch.msgid.link/20241022155147.463475-3-daniel.lezcano@linaro.org
[ rjw: Use the thermal_zone guard for locking, subject edit ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Switch over the thermal core to using a mutex guard for
thermal_governor_lock management.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/3679429.R56niFO833@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
In preparation for a subsequent change that will switch over the thermal
core to using a mutex guard for managing thermal_governor_lock, move
the code running in thermal_zone_device_register_with_trips() under that
lock into a separate function called thermal_zone_init_governor().
While at it, drop a useless comment.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/4408795.ejJDZkT8p0@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Add and use a special guard for cooling devices.
This allows quite a few error code paths to be simplified among
other things and brings in code size reduction for a good measure.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/5837621.DvuYhMxLoT@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Three thermal governors call __thermal_cdev_update() under the
cdev lock without doing any checks, so in order to reduce the
related code duplication, introduce a new helper function called
thermal_cdev_update_nocheck() for them and make them use it.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/1835097.VLH7GnMWUR@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
It is not necessary to walk the thermal_instances list in a trip
descriptor under a cooling device lock, so acquire that lock only
for deleting the given thermal instance from the list of thermal
instances in the given cdev.
Moreover, in analogy with the previous change that introduced
thermal_instance_add(), put the code deleting the given thermal
instance from the lists it is on into a separate new function
called thermal_instance_delete().
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/3275745.5fSG56mABF@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
To reduce the number of redundant result checks in
thermal_bind_cdev_to_trip() and make the code in it easier to
follow, move some of it to a new function called thermal_instance_add()
and make thermal_bind_cdev_to_trip() invoke that function.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/3618899.iIbC2pHGDl@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Holding a cooling device lock under thermal_governor_update_tz() is not
necessary and it may cause lockdep to complain if any governor's
.update_tz() callback attempts to lock a cdev.
For this reason, move the thermal_governor_update_tz() calls in
thermal_bind_cdev_to_trip() and thermal_unbind_cdev_from_trip() from
under the cdev lock.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/7749552.EvYhyI6sBW@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Switch over the thermal core to using a mutex guard for
thermal_list_lock management.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/2010397.PYKUYFuaPT@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
To prepare for a subsequent change that will switch over the thermal
core to using a mutex guard for thermal_list_lock management, move the
code running under thermal_list_lock during the initialization and
unregistration of cooling devices into separate functions.
While at it, drop some comments that do not add value.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/10572828.nUPlyArG6x@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Add a guard for unlocking a locked thermal zone temporarily and use it
in thermal_zone_pm_prepare().
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/3344086.aeNJFYEL58@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Add and use a guard for thermal zone locking.
This allows quite a few error code paths to be simplified among
other things and brings in a noticeable code size reduction for
a good measure.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/1930069.tdWV9SEqCh@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
The code is somewhat cleaner if struct thermal_trip_desc pointers are
passed to thermal_bind_cdev_to_trip(), thermal_unbind_cdev_from_trip(),
and print_bind_err_msg() instead of struct thermal_trip pointers, so
modify it accordingly.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/2246211.NgBsaNRSFp@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
In almost all places where a thermal zone's list of thermal instances
is walked, there is a check to match a specific trip point and it is
walked in vain whenever there are no cooling devices associated with
the given trip.
To address this, store the lists of thermal instances in trip point
descriptors instead of storing them in thermal zones and adjust all
code using those lists accordingly.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/5522726.Sb9uPGUboI@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
After previous changes, the need_update field in struct thermal_zone_device
is only set and never read, so drop it.
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/2495061.jE0xQCEvom@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
If a new cooling device is registered and it is bound to at least one
trip point in a given thermal zone, that thermal zone needs to be
updated via __thermal_zone_device_update().
Instead of doing this with the help of the need_update atomic field in
struct thermal_zone_device, which is not particularly straightforward,
make __thermal_zone_cdev_bind() return a bool value indicating whether
or not the given thermal zone needs to be updated because a new cooling
device has been bound to it and update thermal_zone_cdev_bind() to
call __thermal_zone_device_update() when this value is "true".
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/2226302.Icojqenx9y@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
In analogy with a previous change in the thermal zone initialization
path, to avoid acquiring the thermal zone lock and releasing it multiple
times back and forth unnecessarily, move all of the code running under
thermal_list_lock in thermal_zone_device_unregister() into a new
function called thermal_zone_exit() and make the latter acquire the
thermal zone lock only once and release it along with thermal_list_lock.
For this purpose, provide an "unlocked" variant of
thermal_zone_cdev_unbind() to be called by thermal_zone_exit() under the
thermal zone lock.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/1963152.taCxCBeP46@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
In analogy with a previous change in the thermal zone registration code
path, to ensure that __thermal_zone_device_update() will return early
for thermal zones that are going away, introduce a thermal zone state
flag representing the "exit" state and set it while deleting the thermal
zone from thermal_tz_list.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/4394176.ejJDZkT8p0@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
The part of thermal zone initialization carried out under
thermal_list_lock acquires the thermal zone lock and releases it
multiple times back and forth which is not really necessary.
Instead of doing this, make it acquire the thermal zone lock once after
acquiring thermal_list_lock and release it along with that lock.
For this purpose, move all of the code in question to
thermal_zone_init_complete() introduced previously and provide an
"unlocked" variant of thermal_zone_cdev_bind() to be invoked from
there.
Also notice that a thermal zone does not need to be added to
thermal_tz_list under its own lock, so make the new code acquire
the thermal zone lock after adding it to the list.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/1920382.CQOukoFCf9@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
[ rjw: Rebase on top of recent thermal core changes ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
If the registration of a thermal zone takes place at the time when
system suspend is started, thermal_pm_notify() can run before the new
thermal zone is added to thermal_tz_list and its "suspended" flag will
not be set. Consequently, if __thermal_zone_device_update() is called
for that thermal zone, it will not return early as expected which may
cause some destructive interference with the system suspend or resume
flow to occur.
To avoid that, make thermal_zone_init_complete() introduced previously
set the "suspended" flag for new thermal zones if it runs during system
suspend or resume.
Fixes: 4e814173a8 ("thermal: core: Fix thermal zone suspend-resume synchronization")
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/8490245.NyiUUSuA9g@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
After thermal_zone_device_register_with_trips() has called
device_register() and it has registered the new thermal zone device
with the driver core, user space may access its sysfs attributes and,
among other things, it may enable the thermal zone before it is ready.
To address this, introduce a new thermal zone state flag for
initialization and set it before calling device_register() in
thermal_zone_device_register_with_trips(). This causes
__thermal_zone_device_update() to return early until the new flag
is cleared.
To clear it when the thermal zone is ready, introduce a new
function called thermal_zone_init_complete() that will also invoke
__thermal_zone_device_update() after clearing that flag (both under the
thernal zone lock) and make thermal_zone_device_register_with_trips()
call the new function instead of checking need_update and calling
thermal_zone_device_update() when it is set.
After this change, if user space enables the thermal zone prematurely,
__thermal_zone_device_update() will return early for it until
thermal_zone_init_complete() is called. In turn, if the thermal zone
is not enabled by user space before thermal_zone_init_complete() is
called, the __thermal_zone_device_update() call in it will return early
because the thermal zone has not been enabled yet, but that function
will be invoked again by thermal_zone_device_set_mode() when the thermal
zone is enabled and it will not return early this time.
The checking of need_update is not necessary any more because the
__thermal_zone_device_update() calls potentially triggered by cooling
device binding take place before calling thermal_zone_init_complete(),
so they all will return early, which means that
thermal_zone_init_complete() must call __thermal_zone_device_update()
in case the thermal zone is enabled prematurely by user space.
Fixes: 203d3d4aa4 ("the generic thermal sysfs driver")
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/9360231.CDJkKcVGEf@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Instead of using two separate fields in struct thermal_zone_device for
representing flags related to thermal zone suspend, represent them
explicitly as bits in one u8 "state" field.
Subsequently, that field will be used for addressing race conditions
related to thermal zone initialization and exit.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/7733910.EvYhyI6sBW@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Move the code run for each thermal zone by the thermal PM notify
handler to separate functions.
This will help to make some subsequent changes look somewhat more
straightforward, among other things.
No intentional functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/2299090.iZASKD2KPV@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Since user space can start interacting with a new thermal zone as soon
as device_register() called by thermal_zone_device_register_with_trips()
returns, it is better to initialize the thermal zone before calling
device_register() on it.
Fixes: d0df264fbd ("thermal/core: Remove pointless thermal_zone_device_reset() function")
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://patch.msgid.link/3336146.44csPzL39Z@rjwysocki.net
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
- Fix the TPMI_RAPL_REG_DOMAIN_INFO register offset in the TPMI part of
the Intel RAPL power capping driver, make it ignore minor hardware
version mismatches (which only indicate exposing additional features)
and update register definitions in it to enable PL4 support (Zhang
Rui).
- Add Arrow Lake-U to the list of processors supporting PL4 in the MSR
part of the Intel RAPL power capping driver (Sumeet Pawnikar).
- Remove excess pci_disable_device() calls from the processor part of
the int340x thermal driver to address a warning triggered during
module unload and remove unused CPU hotplug code related to RAPL
support from it (Zhang Rui).
-----BEGIN PGP SIGNATURE-----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=e9eE
-----END PGP SIGNATURE-----
Merge tag 'pm-6.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull power management fixes from Rafael Wysocki:
"These address two issues in the TPMI module of the Intel RAPL power
capping driver and one issue in the processor part of the Intel
int340x thermal driver, update a CPU ID list and register definitions
needed for RAPL PL4 support and remove some unused code.
Specifics:
- Fix the TPMI_RAPL_REG_DOMAIN_INFO register offset in the TPMI part
of the Intel RAPL power capping driver, make it ignore minor
hardware version mismatches (which only indicate exposing
additional features) and update register definitions in it to
enable PL4 support (Zhang Rui)
- Add Arrow Lake-U to the list of processors supporting PL4 in the
MSR part of the Intel RAPL power capping driver (Sumeet Pawnikar)
- Remove excess pci_disable_device() calls from the processor part of
the int340x thermal driver to address a warning triggered during
module unload and remove unused CPU hotplug code related to RAPL
support from it (Zhang Rui)"
* tag 'pm-6.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
thermal: intel: int340x: processor: Add MMIO RAPL PL4 support
thermal: intel: int340x: processor: Remove MMIO RAPL CPU hotplug support
powercap: intel_rapl_msr: Add PL4 support for Arrowlake-U
powercap: intel_rapl_tpmi: Ignore minor version change
thermal: intel: int340x: processor: Fix warning during module unload
powercap: intel_rapl_tpmi: Fix bogus register reading
Address possible use-after-free scenarios during the processing of
thermal netlink commands and during thermal zone removal (Rafael
Wysocki).
-----BEGIN PGP SIGNATURE-----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=IVMx
-----END PGP SIGNATURE-----
Merge tag 'thermal-6.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull thermal control fixes from Rafael Wysocki:
"Address possible use-after-free scenarios during the processing of
thermal netlink commands and during thermal zone removal (Rafael
Wysocki)"
* tag 'thermal-6.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
thermal: core: Free tzp copy along with the thermal zone
thermal: core: Reference count the zone in thermal_zone_get_by_id()
Initialize, de-initialize and handle the threshold in the same place
than the trip points.
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://patch.msgid.link/20240923100005.2532430-3-daniel.lezcano@linaro.org
[ rjw: Subject edit ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The user thresholds mechanism is a way to have the userspace to tell
the thermal framework to send a notification when a temperature limit
is crossed. There is no id, no hysteresis, just the temperature and
the direction of the limit crossing. That means we can be notified
when a threshold is crossed the way up only, or the way down only or
both ways. That allows to create hysteresis values if it is needed.
A threshold can be added, deleted or flushed. The latter means all
thresholds belonging to a thermal zone will be deleted.
When a threshold is added:
- if the same threshold (temperature and direction) exists, an error
is returned
- if a threshold is specified with the same temperature but a
different direction, the specified direction is added
- if there is no threshold with the same temperature then it is
created
When a threshold is deleted:
- if the same threshold (temperature and direction) exists, it is
deleted
- if a threshold is specified with the same temperature but a
different direction, the specified direction is removed
- if there is no threshold with the same temperature, then an error
is returned
When the threshold are flushed:
- All thresholds related to a thermal zone are deleted
When a threshold is crossed:
- the userspace does not need to know which threshold(s) have been
crossed, it will be notified with the current temperature and the
previous temperature
- if multiple thresholds have been crossed between two updates only
one notification will be send to the userspace, it is pointless to
send a notification per thresholds crossed as the userspace can
handle that easily when it has the temperature delta information
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://patch.msgid.link/20240923100005.2532430-2-daniel.lezcano@linaro.org
[ rjw: Subject edit, use BIT(0) and BIT(1) in symbol definitions ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Similar to the MSR RAPL interface, MMIO RAPL supports PL4 too, so add
MMIO RAPL PL4d support to the processor_thermal driver.
As a result, the powercap sysfs for MMIO RAPL will show a new "peak
power" constraint.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Reviewed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Link: https://patch.msgid.link/20240930081801.28502-7-rui.zhang@intel.com
[ rjw: Subject and changelog edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
CPU0/package0 is always online and the MMIO RAPL driver runs on single
package systems only, so there is no need to handle CPU hotplug in it.
Always register a RAPL package device for package 0 and remove the
unnecessary CPU hotplug support.
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Reviewed-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Link: https://patch.msgid.link/20240930081801.28502-6-rui.zhang@intel.com
[ rjw: Subject edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
