Use the newly added of_reserved_mem_region_to_resource() and
of_reserved_mem_region_count() functions to handle "memory-region"
properties.
The error handling is a bit different in some cases. Often
"memory-region" is optional, so failed lookup is not an error. But then
an error in of_reserved_mem_lookup() is treated as an error. However,
that distinction is not really important. Either the region is available
and usable or it is not. So now, it is just
of_reserved_mem_region_to_resource() which is checked for an error.
Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://lore.kernel.org/r/20251124182751.507624-2-robh@kernel.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Otherwise we're leaking memory.
Fixes: db36632ea5 ("drm/amdgpu: clean up and unify hw fence handling")
Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We want to change ::new() to take no parameters and produce a pool that
is as large as possible while also being inline because that is the
constructor that Rust Binder actually needs.
However, to avoid complications in examples, we still need the current
constructor. So rename it to with_capacity(), which is the idiomatic
Rust name for this kind constructor.
Reviewed-by: Burak Emir <bqe@google.com>
Signed-off-by: Alice Ryhl <aliceryhl@google.com>
Signed-off-by: Yury Norov (NVIDIA) <yury.norov@gmail.com>
This constructor is useful when you just want to create a BitmapVec
without allocating but don't care how large it is.
Acked-by: Yury Norov (NVIDIA) <yury.norov@gmail.com>
Reviewed-by: Burak Emir <bqe@google.com>
Reviewed-by: Danilo Krummrich <dakr@kernel.org>
Signed-off-by: Alice Ryhl <aliceryhl@google.com>
Signed-off-by: Yury Norov (NVIDIA) <yury.norov@gmail.com>
To avoid hard-coding these values in drivers, define constants for them
that drivers can reference. Also, update all instances in bitmap.rs and
id_pool.rs that use these values to use the new constants.
Signed-off-by: Alice Ryhl <aliceryhl@google.com>
Reviewed-by: Burak Emir <bqe@google.com>
Signed-off-by: Yury Norov (NVIDIA) <yury.norov@gmail.com>
If memory is freed we need to nullify the pointer or the caller
might call kfree again (eg: amdgpu_cs_parser_fini calls kfree on
all non-null job pointers).
Fixes: db36632ea5 ("drm/amdgpu: clean up and unify hw fence handling")
Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The scroll resolution multipliers are set in the context of
hidinput_connect(), which is only called at probe time: when the host
changes the value on the device with a SET_REPORT(FEATURE), and the device
accepts it, these multipliers are stored on the host side, and used to
calculate the final scroll event values sent to userspace.
After a USB suspend, the resume operation on many hubs and chipsets
involve a USB reset signal as well. A reset on the device side clears all
previous state information, including the value of the multiplier report.
This reset is not handled by the multiplier handling logic, so what ends up
happening is the host is still expecting high-resolution scroll events,
but the device is reset to default resolution, making the effective,
user-perceived scroll speed incredibly slow.
The solution is to renegotiate the multiplier selection after each reset.
This is not the only bug related to the high-resolution scrolling
implementation in the kernel (the other one is
https://bugzilla.kernel.org/show_bug.cgi?id=220144), but for this one,
there is no device side workaround for, leading to poor user experience with our product:
https://github.com/UltimateHackingKeyboard/firmware/issues/1155https://github.com/UltimateHackingKeyboard/firmware/issues/1261https://github.com/UltimateHackingKeyboard/firmware/pull/1355
This patch was tested by an affected user and has been reported to
fix the issue (see discussion in 1355).
Signed-off-by: Benedek Kupper <kupper.benedek@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.com>
The mouse portion of the device's Report Descriptor declares 5 buttons but only
declares 3 usages (Button 1 through Button 3). As a result events for the 2
side buttons are not generated.
Detect and repair the Report Descriptor if necessary by changing the Usage
Maximum value from Button 3 to Button 5.
[jkosina@suse.com: standardize changelog a little bit]
Reported-by: Artem <temabiill@gmail.com>
Closes: https://lore.kernel.org/all/CADYkRmrfhRf6VmQjc+su+mepyv=TsHc+aMcL6ryRZ5HTZ8pyFg@mail.gmail.com/
Signed-off-by: Terry Junge <linuxhid@cosmicgizmosystems.com>
Tested-by: Artem <temabiill@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.com>
While testing rpmsg-char interface it was noticed that duplicate sysfs
entries are getting created and below warning is noticed.
Reason for this is that we are leaking rpmsg device pointer, setting it
null without actually unregistering device.
Any further attempts to unregister fail because rpdev is NULL,
resulting in a leak.
Fix this by unregistering rpmsg device before removing its reference
from rpmsg channel.
sysfs: cannot create duplicate filename '/devices/platform/soc@0/3700000.remot
eproc/remoteproc/remoteproc1/3700000.remoteproc:glink-edge/3700000.remoteproc:
glink-edge.adsp_apps.-1.-1'
[ 114.115347] CPU: 0 UID: 0 PID: 9 Comm: kworker/0:0 Not
tainted 6.16.0-rc4 #7 PREEMPT
[ 114.115355] Hardware name: Qualcomm Technologies, Inc. Robotics RB3gen2 (DT)
[ 114.115358] Workqueue: events qcom_glink_work
[ 114.115371] Call trace:8
[ 114.115374] show_stack+0x18/0x24 (C)
[ 114.115382] dump_stack_lvl+0x60/0x80
[ 114.115388] dump_stack+0x18/0x24
[ 114.115393] sysfs_warn_dup+0x64/0x80
[ 114.115402] sysfs_create_dir_ns+0xf4/0x120
[ 114.115409] kobject_add_internal+0x98/0x260
[ 114.115416] kobject_add+0x9c/0x108
[ 114.115421] device_add+0xc4/0x7a0
[ 114.115429] rpmsg_register_device+0x5c/0xb0
[ 114.115434] qcom_glink_work+0x4bc/0x820
[ 114.115438] process_one_work+0x148/0x284
[ 114.115446] worker_thread+0x2c4/0x3e0
[ 114.115452] kthread+0x12c/0x204
[ 114.115457] ret_from_fork+0x10/0x20
[ 114.115464] kobject: kobject_add_internal failed for 3700000.remoteproc:
glink-edge.adsp_apps.-1.-1 with -EEXIST, don't try to register things with
the same name in the same directory.
[ 114.250045] rpmsg 3700000.remoteproc:glink-edge.adsp_apps.-1.-1:
device_add failed: -17
Fixes: 835764ddd9 ("rpmsg: glink: Move the common glink protocol implementation to glink_native.c")
Cc: Stable@vger.kernel.org
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@oss.qualcomm.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250822100043.2604794-2-srinivas.kandagatla@oss.qualcomm.com
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
MSM8974 requires the CX power domain, so use the msm8996_adsp_resource
which has cx under proxy_pd_names and is otherwise equivalent.
Signed-off-by: Luca Weiss <luca@lucaweiss.eu>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250621-msm8974-rpmpd-switch-v1-2-0a2cb303c446@lucaweiss.eu
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Using CX as a regulator is an artifact of earlier times. Instead use CX
power rail as power domain from rpmpd.
Signed-off-by: Luca Weiss <luca@lucaweiss.eu>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://lore.kernel.org/r/20250621-msm8974-rpmpd-switch-v1-1-0a2cb303c446@lucaweiss.eu
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Move Telit 0x10c7 composition right after 0x10c6 composition and
before 0x10c8 composition.
Signed-off-by: Fabio Porcedda <fabio.porcedda@gmail.com>
Cc: stable@vger.kernel.org
Signed-off-by: Johan Hovold <johan@kernel.org>
The macro FAN_FROM_REG evaluates its arguments multiple times. When used
in lockless contexts involving shared driver data, this causes
Time-of-Check to Time-of-Use (TOCTOU) race conditions.
Convert the macro to a static function. This guarantees that arguments
are evaluated only once (pass-by-value), preventing the race
conditions.
Adhere to the principle of minimal changes by only converting macros
that evaluate arguments multiple times and are used in lockless
contexts.
Link: https://lore.kernel.org/all/CALbr=LYJ_ehtp53HXEVkSpYoub+XYSTU8Rg=o1xxMJ8=5z8B-g@mail.gmail.com/
Signed-off-by: Gui-Dong Han <hanguidong02@gmail.com>
Link: https://lore.kernel.org/r/20251126113828.10003-1-hanguidong02@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
The function fan_show checks shared data for zero or invalid values
before using it as a divisor. These accesses are currently lockless. If
the data changes to zero between the check and the division, it causes a
divide-by-zero error.
Explicitly acquire the update lock around these checks and calculations
to ensure the data remains stable, preventing Time-of-Check to
Time-of-Use (TOCTOU) race conditions.
Link: https://lore.kernel.org/all/CALbr=LYJ_ehtp53HXEVkSpYoub+XYSTU8Rg=o1xxMJ8=5z8B-g@mail.gmail.com/
Signed-off-by: Gui-Dong Han <hanguidong02@gmail.com>
Link: https://lore.kernel.org/r/20251126114047.10039-1-hanguidong02@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
The macro FAN_FROM_REG evaluates its arguments multiple times. When used
in lockless contexts involving shared driver data, this causes
Time-of-Check to Time-of-Use (TOCTOU) race conditions.
Convert the macro to a static function. This guarantees that arguments
are evaluated only once (pass-by-value), preventing the race
conditions.
Adhere to the principle of minimal changes by only converting macros
that evaluate arguments multiple times and are used in lockless
contexts.
Link: https://lore.kernel.org/all/CALbr=LYJ_ehtp53HXEVkSpYoub+XYSTU8Rg=o1xxMJ8=5z8B-g@mail.gmail.com/
Signed-off-by: Gui-Dong Han <hanguidong02@gmail.com>
Link: https://lore.kernel.org/r/20251126113542.9968-1-hanguidong02@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Use pskb_may_pull() to ensure a complete CAN frame is present in the
linear data buffer before reading the CAN ID. A simple skb->len check
is insufficient because it only verifies the total data length but does
not guarantee the data is present in skb->data (it could be in
fragments).
pskb_may_pull() both validates the length and pulls fragmented data
into the linear buffer if necessary, making it safe to directly
access skb->data.
Reported-by: syzbot+5d8269a1e099279152bc@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=5d8269a1e099279152bc
Fixes: f057bbb6f9 ("net: em_canid: Ematch rule to match CAN frames according to their identifiers")
Signed-off-by: Shaurya Rane <ssrane_b23@ee.vjti.ac.in>
Link: https://patch.msgid.link/20251126085718.50808-1-ssranevjti@gmail.com
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
The commit 5cff263606 ("can: rcar_canfd: Fix controller mode setting")
has aligned with the flow mentioned in the hardware manual for all SoCs
except R-Car Gen3 and RZ/G2L SoCs. On R-Car Gen4 and RZ/G3E SoCs, due to
the wrong logic in the commit[1] sets the default mode to FD-Only mode
instead of CAN-FD mode.
This patch sets the CAN-FD mode as the default for all SoCs by dropping
the rcar_canfd_set_mode() as some SoC requires mode setting in global
reset mode, and the rest of the SoCs in channel reset mode and update the
rcar_canfd_reset_controller() to take care of these constraints. Moreover,
the RZ/G3E and R-Car Gen4 SoCs support 3 modes compared to 2 modes on the
R-Car Gen3. Use inverted logic in rcar_canfd_reset_controller() to
simplify the code later to support FD-only mode.
[1]
commit 45721c406d ("can: rcar_canfd: Add support for r8a779a0 SoC")
Fixes: 5cff263606 ("can: rcar_canfd: Fix controller mode setting")
Cc: stable@vger.kernel.org
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Link: https://patch.msgid.link/20251118123926.193445-1-biju.das.jz@bp.renesas.com
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Since commit 6f20d32612 ("HID: logitech-dj: Fix error handling in
logi_dj_recv_switch_to_dj_mode()") logi_dj_recv_switch_to_dj_mode()
will return an error when the hid_hw_raw_request() call to enable
[dis]connect events fails.
This can happen when used with a KVM like the Aten CS1784a and the PC
does not have the KVM focus when probe() runs, which causes probe() to
fail after which the receiver will simply not work.
The logi_dj_recv_query_paired_devices() call done at the end of probe()
already ignores any errors for the KVM without focus case. When focus is
restored and an input report is received this will trigger
logi_dj_recv_queue_unknown_work() which retries the query_paired_devices()
call from a workqueue.
To fix the probe() failure let it ignore logi_dj_recv_switch_to_dj_mode()
errors too, track if a successful logi_dj_recv_switch_to_dj_mode() was
done and retry if necessary from logi_dj_recv_queue_unknown_work().
Queurying paired devices while not in dj-mode is not useful and this
will be redone after the unknown work has retried setting dj-mode,
so skip queurying paired devices when not in dj-mode yet.
The new bool to track successful setting of the dj-mode will also cause
setting dj-mode to be retried from the unknown work, if setting dj-mode
failed after a reset_resume.
Signed-off-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
Signed-off-by: Jiri Kosina <jkosina@suse.com>
logi_dj_recv_query_paired_devices() and logi_dj_recv_switch_to_dj_mode()
both have 2 callers which all log an error if the function fails. Move
the error logging to inside these 2 functions to remove the duplicated
error logging in the callers.
While at it also move the logi_dj_recv_send_report() call error handling
in logi_dj_recv_switch_to_dj_mode() to directly after the call. That call
only fails if the report cannot be found and in that case it does nothing,
so the msleep() is not necessary on failures.
Fixes: 6f20d32612 ("HID: logitech-dj: Fix error handling in logi_dj_recv_switch_to_dj_mode()")
Cc: stable@vger.kernel.org
Signed-off-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
Signed-off-by: Jiri Kosina <jkosina@suse.com>
The Logitech G Pro X Superlight 2 has a lightspeed receiver with a
product id of 0xc54d, this receiver behaves like the receiver used in
the original Logitech G Pro X Superlight (id 0xc547) including the 13
byte mouse reports.
This change adds a definition for this receiver id, and a mapping for
the recvr_type_gaming_hidpp_ls_1_3 type. With this change in place the
receiver now reports the battery status of the connected mouse over
wireless as well as exposing the HID interface needed for userspace to
perform additional configuration with libratbag/Piper.
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Jiri Kosina <jkosina@suse.com>
Introduce support for the i.MX91 thermal monitoring unit, which features a
single sensor for the CPU. The register layout differs from other chips,
necessitating the creation of a dedicated file for this.
This sensor provides a resolution of 1/64°C (6-bit fraction). For actual
accuracy, refer to the datasheet, as it varies depending on the chip grade.
Provide an interrupt for end of measurement and threshold violation and
Contain temperature threshold comparators, in normal and secure address
space, with direction and threshold programmability.
Datasheet Link: https://www.nxp.com/docs/en/data-sheet/IMX91CEC.pdf
Signed-off-by: Pengfei Li <pengfei.li_1@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://patch.msgid.link/20251020-imx91tmu-v7-2-48d7d9f25055@nxp.com
Add bindings documentation for i.MX91 thermal modules.
Signed-off-by: Pengfei Li <pengfei.li_1@nxp.com>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20251020-imx91tmu-v7-1-48d7d9f25055@nxp.com
time is much longer than what 32-bits can provide (Enlin Mu)
- Add delay support on sp804 for ARM32 platforms (Stephen Eta Zhou)
- Fix missing resource release on error in the probe path of in the
ralink driver (Haotian Zhang)
- Fix double deregistration on probe failure in the NXP STM driver
(Johan Hovold)
- Disable runtime PM for the Renesas SH CMT timer because it is
incompatible with PREEMPT_RT=y (Niklas Söderlund)
- Fix section mismatches in the NXP STM driver (Johan Hovold)
- Preventing unbinding the NXP PIT, STM and MMIO ARM Arch timers as
the code does not suppport bind/unbind (Johan Hovold)
- Use the clocksource instead of ticks on the RDA8810PL platform
(Enlin Mu)
- Drop the unused module alias for the STM32-LP (Johan Hovold)
- Add Realtek system timer driver (Hao-Wen Ting)
-----BEGIN PGP SIGNATURE-----
iQEzBAABCAAdFiEEGn3N4YVz0WNVyHskqDIjiipP6E8FAmkm6t4ACgkQqDIjiipP
6E8xqQgAlXnV3vRJmEbjd3ILECvbKMLI2haHV2eA+75P+DvbfriL+ePMHkfkOPI6
CC5UhCSy410cQLO88tzy5+9K8Po2KnHxb+lVS2P6zzcdefL5ZWMZ9Q+CAOwSo1s9
An1A4nUgcTB52mAR+jlz++SF1VV/fMvskMrtiTg8bSIScSc+xi4sEC3GaZR09qSG
RODtzmVsyeoHQ1u6ziRJen8GzpX1q6vUP0eAAr+vXqTUXdCuUL8P20h2mwzxPJWH
mFo53OuKVbTMOoY1Av7euvO1ZZ1tsHsS4NxJfD1qatq+eh1As1dxYodB4dp44qZt
jjnVuj0QrE40VB6EnHAA4kKb6WWpow==
=WXsR
-----END PGP SIGNATURE-----
Merge tag 'timers-v6.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/daniel.lezcano/linux into timers/clocksource
Pull clocksource/event changes from Daniel Lezcano:
- Use 64-bits for timer compensation for IoT usage where the suspend
time is much longer than what 32-bits can provide (Enlin Mu)
- Add delay support on sp804 for ARM32 platforms (Stephen Eta Zhou)
- Fix missing resource release on error in the probe path of in the
ralink driver (Haotian Zhang)
- Fix double deregistration on probe failure in the NXP STM driver
(Johan Hovold)
- Disable runtime PM for the Renesas SH CMT timer because it is
incompatible with PREEMPT_RT=y (Niklas Söderlund)
- Fix section mismatches in the NXP STM driver (Johan Hovold)
- Preventing unbinding the NXP PIT, STM and MMIO ARM Arch timers as
the code does not suppport bind/unbind (Johan Hovold)
- Use the clocksource instead of ticks on the RDA8810PL platform
(Enlin Mu)
- Drop the unused module alias for the STM32-LP (Johan Hovold)
- Add Realtek system timer driver (Hao-Wen Ting)
Link: https://lore.kernel.org/all/9303b790-28d4-4bd9-b01d-28fb05493596@linaro.org
When a device is hot-plugged, the drivers_autoprobe sysfs attribute is
not checked (at least for PCI devices). This means that
drivers_autoprobe is not working as intended, e.g. hot-plugged PCI
devices will still be autoprobed and bound to drivers even with
drivers_autoprobe disabled.
The problem likely started when device_add() was removed from
pci_bus_add_device() in commit 4f535093cf ("PCI: Put pci_dev in device
tree as early as possible") which means that the check for
drivers_autoprobe which used to happen in bus_probe_device() is no
longer present (previously bus_add_device() calls bus_probe_device()).
Conveniently, in commit 9170304169 ("PCI: Allow built-in drivers to
use async initial probing") device_attach() was replaced with
device_initial_probe() which faciliates this change to push the check
for drivers_autoprobe into device_initial_probe().
Make sure all devices check drivers_autoprobe by pushing the
drivers_autoprobe check into device_initial_probe(). This will only
affect devices on the PCI bus for now as device_initial_probe() is only
called by pci_bus_add_device() and bus_probe_device(), but
bus_probe_device() already checks for autoprobe, so callers of
bus_probe_device() should not observe changes on autoprobing.
Note also that pushing this check into device_initial_probe() rather
than device_attach() makes it only affect automatic probing of
drivers (e.g. when a device is hot-plugged), userspace can still choose
to manually bind a driver by writing to drivers_probe sysfs attribute,
even with autoprobe disabled.
Any future callers of device_initial_probe() will respect the
drivers_autoprobe sysfs attribute, which is the intended purpose of
drivers_autoprobe.
Signed-off-by: Vincent Liu <vincent.liu@nutanix.com>
Link: https://patch.msgid.link/20251022120740.2476482-1-vincent.liu@nutanix.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistency cannot be addressed without refactoring the API.
alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.
This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6cc ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566 ("workqueue: Add new WQ_PERCPU flag")
This change adds a new WQ_PERCPU flag to explicitly request
alloc_workqueue() to be per-cpu when WQ_UNBOUND has not been specified.
With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.
Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.
Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
Link: https://patch.msgid.link/20251114141618.172154-3-marco.crivellari@suse.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistentcy cannot be addressed without refactoring the API.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6cc ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566 ("workqueue: Add new WQ_PERCPU flag")
Switch to using system_dfl_wq because system_unbound_wq is going away as part of
a workqueue restructuring.
Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
Link: https://patch.msgid.link/20251114141618.172154-2-marco.crivellari@suse.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Expose the current system-defined list of housekeeping CPUs in a new
sysfs file: /sys/devices/system/cpu/housekeeping.
This provides userspace performance tuning tools and resource managers
with a canonical, reliable method to accurately identify the cores
responsible for essential kernel maintenance workloads (RCU, timer
callbacks, and unbound workqueues). Currently, tooling must manually
calculate the housekeeping set by parsing complex kernel boot parameters
(like isolcpus= and nohz_full=) and system topology, which is prone to
error. This dedicated file simplifies the configuration of low-latency
workloads.
Signed-off-by: Aaron Tomlin <atomlin@atomlin.com>
Link: https://patch.msgid.link/20251011012853.7539-2-atomlin@atomlin.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the context of CONFIG_NO_HZ_FULL=y, tick_nohz_full_mask (of type
cpumask_var_t) is initialised to 0. Memory is only allocated to the cpumask
data structure, in tick_nohz_full_setup(), when Linux kernel boot-time
parameter "nohz_full=" is correctly specified (see housekeeping_setup()).
If "nohz_full=" is not set and an attempt is made to read
/sys/devices/system/cpu/nohz_full, '(null)' can be displayed:
❯ cat /sys/devices/system/cpu/nohz_full
(null)
This patch changes the output to print a newline (or 0x0A) instead of
'(null)', making it consistent with print_cpus_isolated() behaviour.
Signed-off-by: Aaron Tomlin <atomlin@atomlin.com>
Link: https://patch.msgid.link/20251011011830.6670-3-atomlin@atomlin.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The /sys/devices/system/cpu/nohz_full file is a read-only attribute that
reports the CPUs configured for tickless operation (CONFIG_NO_HZ_FULL=y).
The current definition uses the generic DEVICE_ATTR macro, which
unnecessarily requires specifying the file mode (0444) and a NULL
store operation pointer.
This patch converts the definition to use the dedicated DEVICE_ATTR_RO
macro. This correctly expresses the read-only nature of the attribute,
removes the redundant mode field, and simplifies the code. As a related
cleanup, rename the show function from print_cpus_nohz_full() to the
standard nohz_full_show() for consistency with common sysfs attribute
naming conventions.
Signed-off-by: Aaron Tomlin <atomlin@atomlin.com>
Link: https://patch.msgid.link/20251011011830.6670-2-atomlin@atomlin.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There exists a memory leak of kernfs_iattrs contained as an element
of kernfs_node allocated in __kernfs_new_node(). __kernfs_setattr()
allocates kernfs_iattrs as a sub-object, and the LSM security check
incorrectly errors out and does not free the kernfs_iattrs sub-object.
Make an additional error out case that properly frees kernfs_iattrs if
security_kernfs_init_security() fails.
Fixes: e19dfdc83b ("kernfs: initialize security of newly created nodes")
Co-developed-by: Oliver Rosenberg <olrose55@gmail.com>
Signed-off-by: Oliver Rosenberg <olrose55@gmail.com>
Signed-off-by: Will Rosenberg <whrosenb@asu.edu>
Link: https://patch.msgid.link/20251125151332.2010687-1-whrosenb@asu.edu
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On an ARM64 A1 system, it's possible to have physical memory span
up to the 64T boundary, like below
$ lsmem -b -r -n -o range,size
0x0000000080000000-0x00000000bfffffff 1073741824
0x0000080000000000-0x000008007fffffff 2147483648
0x00000800c0000000-0x0000087fffffffff 546534588416
0x0000400000000000-0x00004000bfffffff 3221225472
0x0000400100000000-0x0000407fffffffff 545460846592
So it's time to extend /sys/kernel/mm/page_idle/bitmap to be able
to account for >2G number of pages, by raising the kernfs file size
limit.
Signed-off-by: Jane Chu <jane.chu@oracle.com>
Link: https://patch.msgid.link/20251111202606.1505437-1-jane.chu@oracle.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We have an upcoming driver named "intel_ehl_pse_io". This creates an
auxiliary child device for it's GPIO sub-functionality, which matches
against "intel_ehl_pse_io.gpio-elkhartlake" and overshoots the current
maximum limit of 32 bytes for auxiliary device id string. Bump the size
to 40 bytes to satisfy such cases.
Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Raag Jadav <raag.jadav@intel.com>
Link: https://patch.msgid.link/20251106052838.433673-1-raag.jadav@intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Define the macros in terms of each other.
This makes them easier to understand and also will make it easier to
implement the transition machinery for 'const struct attribute'.
__ATTR_RO_MODE() can't be implemented in terms of __ATTR() as not all
attributes have a .store callback. The same issue theoretically exists
for __ATTR_WO(), but practically that does not occur today.
Reorder __ATTR_RO() below __ATTR_RO_MODE() to keep the order of the
macro definition consistent with respect to each other.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20251029-sysfs-const-attr-prep-v5-7-ea7d745acff4@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There was no example for the is_visible() callback so far.
It will also become an example and test for the constification of
'struct attribute' later.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20251029-sysfs-const-attr-prep-v5-5-ea7d745acff4@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When constifying instances of struct attribute, for consistency the
corresponding .is_visible() callback should be adapted, too.
Introduce a temporary transition mechanism until all callbacks are
converted.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20251029-sysfs-const-attr-prep-v5-4-ea7d745acff4@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For the constification phase of 'struct attribute' various callback
struct members will need to exist in both const and non-const variants.
Keeping both members in a union avoids memory and CPU overhead but will
be detected and trapped by Control Flow Integrity (CFI). By deciding
between a struct and a union depending whether CFI is enabled, most
configurations can avoid this overhead. Code using these callbacks will
still need to be updated to handle both members explicitly.
In the union case the compiler will recognize that testing for one union
member is enough and optimize away the code for the other one.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20251029-sysfs-const-attr-prep-v5-3-ea7d745acff4@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To ease the constification process of 'struct attribute', transparently
handle the const pointers in ATTRIBUTE_GROUPS(). A cast is used instead
of assigning to .attrs_new as it keeps the macro smaller. As both
members are aliased to each other the result is identical.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20251029-sysfs-const-attr-prep-v5-2-ea7d745acff4@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To be able to constify instances of struct attribute it has to be
possible to add them to struct attribute_group. The current type of the
attrs member however is not compatible with that. Introduce a union that
allows registration of both const and non-const attributes to enable a
piecewise transition. As both union member types are compatible no
logic needs to be adapted.
Technically it is now possible register a const struct attribute and
receive it as mutable pointer in the callbacks. This is a soundness
issue. But this same soundness issue already exists today in
sysfs_create_file(). Also the struct definition and callback
implementation are always closely linked and are meant to be moved to
const in lockstep.
Similar to commit 906c508afd ("sysfs: attribute_group: allow
registration of const bin_attribute")
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20251029-sysfs-const-attr-prep-v5-1-ea7d745acff4@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Replace non-standard %Ld with %lld to ensure compliance with the kernel
coding style and potential formatting issues.
Signed-off-by: Clint George <clintbgeorge@gmail.com>
Link: https://patch.msgid.link/20251111151340.9162-3-clintbgeorge@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Replace printk(KERN_CRIT ...) with pr_crit(...) and printk() with
pr_debug(). The change aims to make logging more consistent and
readable.
Signed-off-by: Clint George <clintbgeorge@gmail.com>
Link: https://patch.msgid.link/20251111151340.9162-2-clintbgeorge@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch introduces a new UIO driver, uio_pci_generic_sva, which
extends the functionality of uio_pci_generic by adding support for
Shared Virtual Addressing (SVA) when IOMMU is enabled in the system.
The key enhancement allows PCI devices to directly use user-space virtual
addresses for DMA operations, eliminating the need for bounce buffers or
explicit IOVA mapping. This is achieved by leveraging the kernel's IOMMU-SVA
subsystem, including process address space attachment, page fault handling,
and shared context management between CPU and device.
With this driver, userspace applications can perform zero-copy DMA using
native pointers:
void *addr = malloc(N);
set_dma_addr((uint64_t)addr); // Passing user VA directly
start_dma();
The device can now access 'addr' through the IOMMU's PASID-based translation,
provided that the underlying IOMMU hardware (e.g., Intel VT-d 3.1+, AMD-Vi,
ARM SMMU, RISCV IOMMU) and platform support SVA.
Dependencies:
- CONFIG_IOMMU_SVA must be enabled.
- The platform must support PRI (Page Request Interface) and PASID.
- Device drivers/userspace must handle page faults if demand-paging is used.
The implementation reuses core logic from uio_pci_generic.c while adding
PASID setting, and integration with the IOMMU SVA APIs.
Also, add a read-only sysfs attribute 'pasid' to expose the Process Address
Space ID assigned by IOMMU driver when binding an SVA-enabled device.
For details, refer to the ABI documentation for uio_pci_sva driver sysfs attribute
(Documentation/ABI/testing/sysfs-driver-uio_pci_sva-pasid).
Signed-off-by: Yaxing Guo <guoyaxing@bosc.ac.cn>
Link: https://patch.msgid.link/20250926095828.506-1-guoyaxing@bosc.ac.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix below warnings generated dt_bindings_check for examples in the
bindings.
Documentation/devicetree/bindings/slimbus/slimbus.example.dtb:
slim@28080000 (qcom,slim-ngd-v1.5.0): 'audio-codec@1,0' does not match
any of the regexes: '^pinctrl-[0-9]+$', '^slim@[0-9a-f]+$'
from schema $id:
http://devicetree.org/schemas/slimbus/qcom,slim-ngd.yaml#
Documentation/devicetree/bindings/slimbus/slimbus.example.dtb:
slim@28080000 (qcom,slim-ngd-v1.5.0): #address-cells: 1 was expected
from schema $id:
http://devicetree.org/schemas/slimbus/qcom,slim-ngd.yaml#
Documentation/devicetree/bindings/slimbus/slimbus.example.dtb:
slim@28080000 (qcom,slim-ngd-v1.5.0): 'dmas' is a required property
from schema $id:
http://devicetree.org/schemas/slimbus/qcom,slim-ngd.yaml#
Documentation/devicetree/bindings/slimbus/slimbus.example.dtb:
slim@28080000 (qcom,slim-ngd-v1.5.0): 'dma-names' is a required
property
from schema $id:
http://devicetree.org/schemas/slimbus/qcom,slim-ngd.yaml#
Fixes: 7cbba32a2d ("slimbus: qcom: remove unused qcom controller driver")
Cc: stable <stable@kernel.org>
Reported-by: Rob Herring <robh@kernel.org>
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@oss.qualcomm.com>
Acked-by: Rob Herring (Arm) <robh@kernel.org>
Signed-off-by: Srinivas Kandagatla <srini@kernel.org>
Link: https://patch.msgid.link/20251114110505.143105-1-srini@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
intel_th_output_open() calls bus_find_device_by_devt() which
internally increments the device reference count via get_device(), but
this reference is not properly released in several error paths. When
device driver is unavailable, file operations cannot be obtained, or
the driver's open method fails, the function returns without calling
put_device(), leading to a permanent device reference count leak. This
prevents the device from being properly released and could cause
resource exhaustion over time.
Found by code review.
Cc: stable <stable@kernel.org>
Fixes: 39f4034693 ("intel_th: Add driver infrastructure for Intel(R) Trace Hub devices")
Signed-off-by: Ma Ke <make24@iscas.ac.cn>
Link: https://patch.msgid.link/20251112091723.35963-1-make24@iscas.ac.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When DT is used to get the reference of 'rp1_node', it should be released
when not needed anymore, otherwise it is leaking.
In such a case, add the missing of_node_put() call at the end of the probe,
as already done in the error handling path.
Fixes: 49d63971f9 ("misc: rp1: RaspberryPi RP1 misc driver")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Andrea della Porta <andrea.porta@suse.com>
Link: https://patch.msgid.link/9bc1206de787fa86384f3e5ba0a8027947bc00ff.1762585959.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
