linux-stable.git/drivers/acpi/acpi_processor.c, branch v6.6.78

ACPI: processor: Fix memory leaks in error paths of processor_add()

2024-09-12T09:11:42+00:00

[ Upstream commit 47ec9b417ed9b6b8ec2a941cd84d9de62adc358a ]

If acpi_processor_get_info() returned an error, pr and the associated
pr->throttling.shared_cpu_map were leaked.

The unwind code was in the wrong order wrt to setup, relying on
some unwind actions having no affect (clearing variables that were
never set etc).  That makes it harder to reason about so reorder
and add appropriate labels to only undo what was actually set up
in the first place.

Acked-by: Rafael J. Wysocki 
Reviewed-by: Gavin Shan 
Signed-off-by: Jonathan Cameron 
Link: https://lore.kernel.org/r/20240529133446.28446-6-Jonathan.Cameron@huawei.com
Signed-off-by: Catalin Marinas 
Signed-off-by: Sasha Levin

ACPI: processor: Return an error if acpi_processor_get_info() fails in processor_add()

2024-09-12T09:11:42+00:00

[ Upstream commit fadf231f0a06a6748a7fc4a2c29ac9ef7bca6bfd ]

Rafael observed [1] that returning 0 from processor_add() will result in
acpi_default_enumeration() being called which will attempt to create a
platform device, but that makes little sense when the processor is known
to be not available.  So just return the error code from acpi_processor_get_info()
instead.

Link: https://lore.kernel.org/all/CAJZ5v0iKU8ra9jR+EmgxbuNm=Uwx2m1-8vn_RAZ+aCiUVLe3Pw@mail.gmail.com/ [1]
Suggested-by: Rafael J. Wysocki 
Acked-by: Rafael J. Wysocki 
Reviewed-by: Gavin Shan 
Signed-off-by: Jonathan Cameron 
Link: https://lore.kernel.org/r/20240529133446.28446-5-Jonathan.Cameron@huawei.com
Signed-off-by: Catalin Marinas 
Signed-off-by: Sasha Levin

cpu-hotplug: Provide prototypes for arch CPU registration

2023-10-11T12:27:37+00:00

Provide common prototypes for arch_register_cpu() and
arch_unregister_cpu(). These are called by acpi_processor.c, with weak
versions, so the prototype for this is already set. It is generally not
necessary for function prototypes to be conditional on preprocessor macros.

Some architectures (e.g. Loongarch) are missing the prototype for this, and
rather than add it to Loongarch's asm/cpu.h, do the job once for everyone.

Since this covers everyone, remove the now unnecessary prototypes in
asm/cpu.h, and therefore remove the 'static' from one of ia64's
arch_register_cpu() definitions.

[ tglx: Bring back the ia64 part and remove the ACPI prototypes ]

Signed-off-by: Russell King (Oracle) 
Signed-off-by: Thomas Gleixner 
Link: https://lore.kernel.org/r/E1qkoRr-0088Q8-Da@rmk-PC.armlinux.org.uk

ACPI: processor: Refine messages in acpi_early_processor_control_setup()

2023-07-20T19:05:00+00:00

The source and meaning of the messages printed by
acpi_early_processor_control_setup() is unclear, so add a pr_fmt()
definition to acpi_processor.c and expand the messages to make it
clear that they are about CPUs.

Signed-off-by: Rafael J. Wysocki 
Reviewed-by: Michal Wilczynski

ACPI: processor: Remove acpi_hwp_native_thermal_lvt_osc()

2023-07-17T12:47:42+00:00

The workaround for buggy skylake BIOSes is now implemented in
acpi_processor_osc() and acpi_hwp_native_thermal_lvt_osc() has no
callers, so drop it.

Suggested-by: Rafael J. Wysocki 
Signed-off-by: Michal Wilczynski 
Reviewed-by: Andy Shevchenko 
[ rjw: Subject and changelog edits ]
Signed-off-by: Rafael J. Wysocki

ACPI: processor: Use _OSC to convey OSPM processor support information

2023-07-17T12:47:18+00:00

Change acpi_early_processor_osc() to return a value in case of a
failure and make it static.

Also make it run acpi_processor_osc() for every processor object or
processor device found in the ACPI Namespace (previously, its only purpose
was to work around platform firmware defects on Skylake systems).

Introduce a new function called acpi_early_processor_control_setup() that
will invoke acpi_early_processor_osc() first in order to convey the OS
processor support information to the platform firmware and if that fails,
it will fall back to using _PDC.

Suggested-by: Rafael J. Wysocki 
Signed-off-by: Michal Wilczynski 
Reviewed-by: Andy Shevchenko 
[ rjw: Subject and changelog edits, change function return value to bool,
  add missing 'static', change messages ]
Signed-off-by: Rafael J. Wysocki

ACPI: processor: Introduce acpi_processor_osc()

2023-07-14T15:59:40+00:00

The processor _OSC method is already used for a workaround introduced
in commit a21211672c9a ("ACPI / processor: Request native thermal
interrupt handling via _OSC"), but in accordance with ACPI 6.5 (and
earlier), it should be used for negotiating all of the processor
capabilities instead of _PDC (which has been deprecated since ACPI 3.0
and got removed from ACPI 6.5 entirely).

Create a new callback function called acpi_processor_osc() to be invoked
for every processor object and processor device in the ACPI namespace, in
analogy with the already existing acpi_hwp_native_thermal_lvt_osc().

Make this function implement the workaround mentioned above and convey
all of the OSPM processor support information to the platform firmware
by setting all of the appropriate processor capabilities bits before
evaluating _OSC for the given processor. For this purpose, make it
call arch_acpi_set_proc_cap_bits() and modify the latter to set
ACPI_PROC_CAP_COLLAB_PROC_PERF along with the other processor
capabilities bits.

Suggested-by: Rafael J. Wysocki 
Signed-off-by: Michal Wilczynski 
Reviewed-by: Andy Shevchenko 
[ rjw: Subject and changelog edits, whitespace fixup ]
Signed-off-by: Rafael J. Wysocki

ACPI: processor: Move processor_physically_present() to acpi_processor.c

2023-07-14T15:59:40+00:00

Since _PDC method is deprecated and a preferred method of communicating
OSPM processor power management capabilities is _OSC, there is a need to
move function checking whether processor is present as this logic is not
_PDC specific.

Move processor_physically_present() to acpi_processor.c.

No intentional functional impact.

Suggested-by: Rafael J. Wysocki 
Signed-off-by: Michal Wilczynski 
[ rjw: Subject and changelog edits ]
Signed-off-by: Rafael J. Wysocki

ACPI: cpufreq: Use platform devices to load ACPI PPC and PCC drivers

2023-03-20T17:54:13+00:00

The acpi-cpufreq and pcc-cpufreq drivers are loaded through per-CPU
module aliases. This can result in many unnecessary load requests during
boot if another frequency module, such as intel_pstate, is already
active. For instance, on a typical Intel system, one can observe that
udev makes 2x#CPUs attempts to insert acpi_cpufreq and 1x#CPUs attempts
for pcc_cpufreq. All these tries then fail if another frequency module
is already registered.

In the worst case, without the recent fix in commit 0254127ab977e
("module: Don't wait for GOING modules"), these module loads occupied
all udev workers and had their initialization attempts ran sequentially.
Resolving all these loads then on some larger machines took too long,
prevented other hardware from getting its drivers initialized and
resulted in a failed boot. Discussion over these duplicate module
requests ended up with a conclusion that only one load attempt should be
ideally made.

Both acpi-cpufreq and pcc-cpufreq drivers use platform firmware controls
which are defined by ACPI. It is possible to treat these interfaces as
platform devices.

The patch extends the ACPI parsing logic to check the ACPI namespace if
the PPC or PCC interface is present and creates a virtual platform
device for each if it is available. The acpi-cpufreq and pcc-cpufreq
drivers are then updated to map to these devices.

This allows to try loading acpi-cpufreq and pcc-cpufreq only once during
boot and only if a given interface is available in the firmware.

Signed-off-by: Petr Pavlu 
[ rjw: whitespace and error message log level adjustments, subject edits ]
Signed-off-by: Rafael J. Wysocki

ACPI: processor: Replace deprecated CPU-hotplug functions

2021-08-04T18:25:54+00:00

The functions cpu_hotplug_begin, cpu_hotplug_done, get_online_cpus() and
put_online_cpus() have been deprecated during the CPU hotplug rework. They map
directly to cpus_write_lock(), cpus_write_unlock, cpus_read_lock() and
cpus_read_unlock().

Replace deprecated CPU-hotplug functions with the official version.
The behavior remains unchanged.

Signed-off-by: Sebastian Andrzej Siewior 
Signed-off-by: Rafael J. Wysocki