Add the LED_RETAIN_AT_SHUTDOWN flag to the registered led_class_devs so
that the LEDs do not get turned-off when reloading the driver and thus so
that they also stay under default EC control when reloading the driver,
unless explicitly overridden by the user.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211123210524.266705-1-hdegoede@redhat.com
The hwmon sysfs-attr should show up under the hwmon-classdev, not under
the tpacpi_sensors_pdev. Pass the tpacpi_hwmon_groups attr-groups array
to hwmon_device_register_with_groups() instead of setting
tpacpi_hwmon_pdriver.driver.dev_groups to it to fix this.
This also requires moving the hwmon_device_register_with_groups() call to
after the subdriver init functions have run so that the is_visible()
calls will work properly.
Fixes: 79f960e29c ("platform/x86: thinkpad_acpi: Convert platform driver to use dev_groups")
Cc: Len Baker <len.baker@gmx.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211123210424.266607-5-hdegoede@redhat.com
Commit 79f960e29c ("platform/x86: thinkpad_acpi: Convert platform driver
to use dev_groups") put the debug_level, interface_version, version and the
THINKPAD_ACPI_DEBUGFACILITIES attributes in a new tpacpi_attr_group and
added those to the tpacpi_groups groups-array which is used to initialize
the driver.dev_groups member.
But before this commit these attributes were registered with
driver_create_file(), so they should be part of the groups-array which is
used to initialize the driver.groups member instead.
And also make the same change for the fan_watchdog hwmon driver attribute.
Fixes: 79f960e29c ("platform/x86: thinkpad_acpi: Convert platform driver to use dev_groups")
Cc: Len Baker <len.baker@gmx.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211123210424.266607-4-hdegoede@redhat.com
Commit 79f960e29c ("platform/x86: thinkpad_acpi: Convert platform driver
to use dev_groups") introduces the use of driver.dev_groups +
attribute_group.is_visible callbacks replacing the conditional calling of
driver_create_file() for optional attributes.
The is_visible callbacks rely on various tp_features.has_foo flags,
which get set by the subdriver init functions.
But before this fix, thinkpad_acpi_module_init() would call the
subdriver init functions after registering the platform_device and
the tpacpi_pdriver. Which would cause the is_visible callbacks to
get called before the subdriver init functions, which in turn would
cause optional attributes to not get registered at all, even when
the feature is actually present.
Fix this by moving the platform_driver_register(&tpacpi_pdriver) to after
the subdriver init calls; and do the same for the tpacpi_hmon_pdriver.
Fixes: 79f960e29c ("platform/x86: thinkpad_acpi: Convert platform driver to use dev_groups")
Cc: Len Baker <len.baker@gmx.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211123210424.266607-3-hdegoede@redhat.com
Commit c99ca78d67 ("platform/x86: thinkpad_acpi: Switch to common use
of attributes") removed the conditional adding of the
hotkey_tablet_mode and hotkey_radio_sw sysfs-attributes, replacing this
with a hotkey_attr_is_visible() callback which hides them when the
feature is not present.
But this commit forgot to add these 2 attributes to the default
hotkey_attributes[] set, so they would now never get added at all.
Add the 2 attributes to the default hotkey_attributes[] set so that
they are available on systems with these features once more.
Fixes: c99ca78d67 ("platform/x86: thinkpad_acpi: Switch to common use of attributes")
Cc: Len Baker <len.baker@gmx.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211123210424.266607-2-hdegoede@redhat.com
Remove the dytc_profile_available check from dytc_profile_set(),
that function only gets called if the platform_profile_handler was
registered, so the check is not necessary.
Make tpacpi_dytc_profile_init() return -ENODEV when it does not register
the platform_profile() handler this will cause
dytc_profile_driver_data.flags.init to not get set, which in turn will
cause the dytc_profile_exit() call to get skipped.
Together this avoids the need to have the dytc_profile_available
variable at all, since the information is now duplicated in the
dytc_profile_driver_data.flags.init flag.
Note this leaves a weirdly indented code-block behind, this is
deliberately done to make what actually changes in this commit clear.
This will be fixed-up in the next commit.
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Tested-by: Mark Pearson <mpearson@lenovo.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211121191129.256713-5-hdegoede@redhat.com
The only reason the proxysensor code needs dytc_version handling is for
proxsensor_attr_is_visible() and that will only ever get called after
all the subdrv init() callbacks have run.
tpacpi_dytc_profile_init() already calls DYTC_CMD_QUERY and is the
primary consumer of dytc_version, so simply let tpacpi_dytc_profile_init()
set dytc_version and remove the now no longer necessary dytc_get_version()
helper and its calls.
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Tested-by: Mark Pearson <mpearson@lenovo.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211121191129.256713-4-hdegoede@redhat.com
Make ibm_init_struct.init() callbacks return -ENODEV instead of 1 when
the subdevice / function is not available.
Using -ENODEV clearly states what it going on where as a magic return of
"1" requires a deep dive into the code to figure out what is going on.
This also allows for some cleanups, avoiding the need to translate an
-ENODEV return into "return 1" (which often mistakenly was "return 0").
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Tested-by: Mark Pearson <mpearson@lenovo.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211121191129.256713-3-hdegoede@redhat.com
Commit 79f960e29c ("platform/x86: thinkpad_acpi: Convert platform driver
to use dev_groups") accidentally modified tpacpi_kbdlang_init() causing it
to return -ENODEV instead of 0 on machines without kbdlang support
(which are most of them).
ibm_init() sees this -ENODEV as an error causing the entire module to not
load, not good.
Note that technically tpacpi_kbdlang_init() was already buggy before, it
should have returned 1 instead of 0 if the feature is not present.
Rather then fixing tpacpi_kbdlang_init() though, IMHO it is bettter to
just make ibm_init() treat -ENODEV as 1 to fix the issue; and then in
a followup commit also change all the existing "return 1"s from
ibm_init_struct.init() callbacks to "return -ENODEV" as -ENODEV clearly
states what it going on where as a magic return of "1" requires a deep
dive into the code to figure out what is going on.
This will also allow removing some extra ifs to translate -ENODEV to
return 1 in a couple of init() callbacks.
Fixes: 79f960e29c ("platform/x86: thinkpad_acpi: Convert platform driver to use dev_groups")
Cc: Len Baker <len.baker@gmx.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20211121191129.256713-2-hdegoede@redhat.com
Add support for custom fan curves found on some ASUS ROG laptops.
These laptops have the ability to set a custom curve for the CPU
and GPU fans via two ACPI methods.
This patch adds two pwm<N> attributes to the hwmon sysfs,
pwm1 for CPU fan, pwm2 for GPU fan. Both are under the hwmon of the
name `asus_custom_fan_curve`. There is no safety check of the set
fan curves - this must be done in userspace.
The fans have settings [1,2,3] under pwm<N>_enable:
1. Enable and write settings out
2. Disable and use factory fan mode
3. Same as 2, additionally restoring default factory curve.
Use of 2 means that the curve the user has set is still stored and
won't be erased, but the laptop will be using its default auto-fan
mode. Re-enabling the manual mode then activates the curves again.
Notes:
- pwm<N>_enable = 0 is an invalid setting.
- pwm is actually a percentage and is scaled on writing to device.
Signed-off-by: Luke D. Jones <luke@ljones.dev>
Link: https://lore.kernel.org/r/20211024033705.5595-2-luke@ljones.dev
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Platform drivers have the option of having the platform core create and
remove any needed sysfs attribute files. So take advantage of that and
refactor the attributes management to avoid to register them "by hand".
Also, due to some attributes are optionals, refactor the code and move
the logic inside the "is_visible" callbacks of the attribute_group
structures.
Suggested-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Len Baker <len.baker@gmx.com>
Link: https://lore.kernel.org/r/20211023154036.6800-1-len.baker@gmx.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
When WWAN device wake from S3 deep, under thinkpad platform,
WWAN would be disabled. This disable status could be checked
by command 'nmcli r wwan' or 'rfkill list'.
Issue analysis as below:
When host resume from S3 deep, thinkpad_acpi driver would
call hotkey_resume() function. Finnaly, it will use
wan_get_status to check the current status of WWAN device.
During this resume progress, wan_get_status would always
return off even WWAN boot up completely.
In patch V2, Hans said 'sw_state should be unchanged
after a suspend/resume. It's better to drop the
tpacpi_rfk_update_swstate call all together from the
resume path'.
And it's confimed by Lenovo that GWAN is no longer
available from WHL generation because the design does not
match with current pin control.
Signed-off-by: Slark Xiao <slark_xiao@163.com>
Link: https://lore.kernel.org/r/20211108060648.8212-1-slark_xiao@163.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
A Lenovo ThinkStation S20 (4157CTO BIOS 60KT41AUS) fails to boot on
recent kernels including the think-lmi driver, due to the fact that
errors returned by the tlmi_analyze() function are ignored by
tlmi_probe(), where tlmi_sysfs_init() is called unconditionally.
This results in making use of an array of already freed, non-null
pointers and other uninitialized globals, causing all sorts of nasty
kobject and memory faults.
Make use of the analyze function return value, free a couple leaked
allocations, and remove the settings_count field, which is incremented
but never consumed.
Fixes: a40cd7ef22 ("platform/x86: think-lmi: Add WMI interface support on Lenovo platforms")
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Mark Gross <markgross@kernel.org>
Reviewed-by: Mark Pearson <markpearson@lenovo.com>
Link: https://lore.kernel.org/r/163639463588.1330483.15850167112490200219.stgit@omen
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Pull xfs cleanups from Darrick Wong:
"The most 'exciting' aspect of this branch is that the xfsprogs
maintainer and I have worked through the last of the code
discrepancies between kernel and userspace libxfs such that there are
no code differences between the two except for #includes.
IOWs, diff suffices to demonstrate that the userspace tools behave the
same as the kernel, and kernel-only bits are clearly marked in the
/kernel/ source code instead of just the userspace source.
Summary:
- Clean up open-coded swap() calls.
- A little bit of #ifdef golf to complete the reunification of the
kernel and userspace libxfs source code"
* tag 'xfs-5.16-merge-5' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
xfs: sync xfs_btree_split macros with userspace libxfs
xfs: #ifdef out perag code for userspace
xfs: use swap() to make dabtree code cleaner
Pull more parisc fixes from Helge Deller:
"Fix a build error in stracktrace.c, fix resolving of addresses to
function names in backtraces, fix single-stepping in assembly code and
flush userspace pte's when using set_pte_at()"
* tag 'for-5.16/parisc-3' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
parisc/entry: fix trace test in syscall exit path
parisc: Flush kernel data mapping in set_pte_at() when installing pte for user page
parisc: Fix implicit declaration of function '__kernel_text_address'
parisc: Fix backtrace to always include init funtion names
Pull arch/sh updates from Rich Felker.
* tag 'sh-for-5.16' of git://git.libc.org/linux-sh:
sh: pgtable-3level: Fix cast to pointer from integer of different size
sh: fix READ/WRITE redefinition warnings
sh: define __BIG_ENDIAN for math-emu
sh: math-emu: drop unused functions
sh: fix kconfig unmet dependency warning for FRAME_POINTER
sh: Cleanup about SPARSE_IRQ
sh: kdump: add some attribute to function
maple: fix wrong return value of maple_bus_init().
sh: boot: avoid unneeded rebuilds under arch/sh/boot/compressed/
sh: boot: add intermediate vmlinux.bin* to targets instead of extra-y
sh: boards: Fix the cacography in irq.c
sh: check return code of request_irq
sh: fix trivial misannotations
Pull ARM fixes from Russell King:
- Fix early_iounmap
- Drop cc-option fallbacks for architecture selection
* tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm:
ARM: 9156/1: drop cc-option fallbacks for architecture selection
ARM: 9155/1: fix early early_iounmap()
Pull devicetree fixes from Rob Herring:
- Two fixes due to DT node name changes on Arm, Ltd. boards
- Treewide rename of Ingenic CGU headers
- Update ST email addresses
- Remove Netlogic DT bindings
- Dropping few more cases of redundant 'maxItems' in schemas
- Convert toshiba,tc358767 bridge binding to schema
* tag 'devicetree-fixes-for-5.16-1' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux:
dt-bindings: watchdog: sunxi: fix error in schema
bindings: media: venus: Drop redundant maxItems for power-domain-names
dt-bindings: Remove Netlogic bindings
clk: versatile: clk-icst: Ensure clock names are unique
of: Support using 'mask' in making device bus id
dt-bindings: treewide: Update @st.com email address to @foss.st.com
dt-bindings: media: Update maintainers for st,stm32-hwspinlock.yaml
dt-bindings: media: Update maintainers for st,stm32-cec.yaml
dt-bindings: mfd: timers: Update maintainers for st,stm32-timers
dt-bindings: timer: Update maintainers for st,stm32-timer
dt-bindings: i2c: imx: hardware do not restrict clock-frequency to only 100 and 400 kHz
dt-bindings: display: bridge: Convert toshiba,tc358767.txt to yaml
dt-bindings: Rename Ingenic CGU headers to ingenic,*.h
Pull timer fix from Thomas Gleixner:
"A single fix for POSIX CPU timers to address a problem where POSIX CPU
timer delivery stops working for a new child task because
copy_process() copies state information which is only valid for the
parent task"
* tag 'timers-urgent-2021-11-14' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
posix-cpu-timers: Clear task::posix_cputimers_work in copy_process()
Pull irq fixes from Thomas Gleixner:
"A set of fixes for the interrupt subsystem
Core code:
- A regression fix for the Open Firmware interrupt mapping code where
a interrupt controller property in a node caused a map property in
the same node to be ignored.
Interrupt chip drivers:
- Workaround a limitation in SiFive PLIC interrupt chip which
silently ignores an EOI when the interrupt line is masked.
- Provide the missing mask/unmask implementation for the CSKY MP
interrupt controller.
PCI/MSI:
- Prevent a use after free when PCI/MSI interrupts are released by
destroying the sysfs entries before freeing the memory which is
accessed in the sysfs show() function.
- Implement a mask quirk for the Nvidia ION AHCI chip which does not
advertise masking capability despite implementing it. Even worse
the chip comes out of reset with all MSI entries masked, which due
to the missing masking capability never get unmasked.
- Move the check which prevents accessing the MSI[X] masking for XEN
back into the low level accessors. The recent consolidation missed
that these accessors can be invoked from places which do not have
that check which broke XEN. Move them back to he original place
instead of sprinkling tons of these checks all over the code"
* tag 'irq-urgent-2021-11-14' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
of/irq: Don't ignore interrupt-controller when interrupt-map failed
irqchip/sifive-plic: Fixup EOI failed when masked
irqchip/csky-mpintc: Fixup mask/unmask implementation
PCI/MSI: Destroy sysfs before freeing entries
PCI: Add MSI masking quirk for Nvidia ION AHCI
PCI/MSI: Deal with devices lying about their MSI mask capability
PCI/MSI: Move non-mask check back into low level accessors
Pull x86 static call update from Thomas Gleixner:
"A single fix for static calls to make the trampoline patching more
robust by placing explicit signature bytes after the call trampoline
to prevent patching random other jumps like the CFI jump table
entries"
* tag 'locking-urgent-2021-11-14' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
static_call,x86: Robustify trampoline patching
Pull scheduler fixes from Borislav Petkov:
- Avoid touching ~100 config files in order to be able to select the
preemption model
- clear cluster CPU masks too, on the CPU unplug path
- prevent use-after-free in cfs
- Prevent a race condition when updating CPU cache domains
- Factor out common shared part of smp_prepare_cpus() into a common
helper which can be called by both baremetal and Xen, in order to fix
a booting of Xen PV guests
* tag 'sched_urgent_for_v5.16_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
preempt: Restore preemption model selection configs
arch_topology: Fix missing clear cluster_cpumask in remove_cpu_topology()
sched/fair: Prevent dead task groups from regaining cfs_rq's
sched/core: Mitigate race cpus_share_cache()/update_top_cache_domain()
x86/smp: Factor out parts of native_smp_prepare_cpus()
Pull perf fixes from Borislav Petkov:
- Prevent unintentional page sharing by checking whether a page
reference to a PMU samples page has been acquired properly before
that
- Make sure the LBR_SELECT MSR is saved/restored too
- Reset the LBR_SELECT MSR when resetting the LBR PMU to clear any
residual data left
* tag 'perf_urgent_for_v5.16_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/core: Avoid put_page() when GUP fails
perf/x86/vlbr: Add c->flags to vlbr event constraints
perf/x86/lbr: Reset LBR_SELECT during vlbr reset
Pull x86 fixes from Borislav Petkov:
- Add the model number of a new, Raptor Lake CPU, to intel-family.h
- Do not log spurious corrected MCEs on SKL too, due to an erratum
- Clarify the path of paravirt ops patches upstream
- Add an optimization to avoid writing out AMX components to sigframes
when former are in init state
* tag 'x86_urgent_for_v5.16_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/cpu: Add Raptor Lake to Intel family
x86/mce: Add errata workaround for Skylake SKX37
MAINTAINERS: Add some information to PARAVIRT_OPS entry
x86/fpu: Optimize out sigframe xfeatures when in init state
Pull more perf tools updates from Arnaldo Carvalho de Melo:
"Hardware tracing:
- ARM:
* Print the size of the buffer size consistently in hexadecimal in
ARM Coresight.
* Add Coresight snapshot mode support.
* Update --switch-events docs in 'perf record'.
* Support hardware-based PID tracing.
* Track task context switch for cpu-mode events.
- Vendor events:
* Add metric events JSON file for power10 platform
perf test:
- Get 'perf test' unit tests closer to kunit.
- Topology tests improvements.
- Remove bashisms from some tests.
perf bench:
- Fix memory leak of perf_cpu_map__new() in the futex benchmarks.
libbpf:
- Add some more weak libbpf functions o allow building with the
libbpf versions, old ones, present in distros.
libbeauty:
- Translate [gs]setsockopt 'level' argument integer values to
strings.
tools headers UAPI:
- Sync futex_waitv, arch prctl, sound, i195_drm and msr-index files
with the kernel sources.
Documentation:
- Add documentation to 'struct symbol'.
- Synchronize the definition of enum perf_hw_id with code in
tools/perf/design.txt"
* tag 'perf-tools-for-v5.16-2021-11-13' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux: (67 commits)
perf tests: Remove bash constructs from stat_all_pmu.sh
perf tests: Remove bash construct from record+zstd_comp_decomp.sh
perf test: Remove bash construct from stat_bpf_counters.sh test
perf bench futex: Fix memory leak of perf_cpu_map__new()
tools arch x86: Sync the msr-index.h copy with the kernel sources
tools headers UAPI: Sync drm/i915_drm.h with the kernel sources
tools headers UAPI: Sync sound/asound.h with the kernel sources
tools headers UAPI: Sync linux/prctl.h with the kernel sources
tools headers UAPI: Sync arch prctl headers with the kernel sources
perf tools: Add more weak libbpf functions
perf bpf: Avoid memory leak from perf_env__insert_btf()
perf symbols: Factor out annotation init/exit
perf symbols: Bit pack to save a byte
perf symbols: Add documentation to 'struct symbol'
tools headers UAPI: Sync files changed by new futex_waitv syscall
perf test bpf: Use ARRAY_CHECK() instead of ad-hoc equivalent, addressing array_size.cocci warning
perf arm-spe: Support hardware-based PID tracing
perf arm-spe: Save context ID in record
perf arm-spe: Update --switch-events docs in 'perf record'
perf arm-spe: Track task context switch for cpu-mode events
...
Pull irqchip fixes from Marc Zyngier:
- Address an issue with the SiFive PLIC being unable to EOI
a masked interrupt
- Move the disable/enable methods in the CSky mpintc to
mask/unmask
- Fix a regression in the OF irq code where an interrupt-controller
property in the same node as an interrupt-map property would get
ignored
Link: https://lore.kernel.org/all/20211112173459.4015233-1-maz@kernel.org
