linux-stable.git/drivers/pwm, branch linux-6.16.y

pwm: mediatek: Fix duty and period setting

2025-08-28T14:34:23+00:00

commit f21d136caf8171f94159d975ea4620c164431bd9 upstream.

The period generated by the hardware is

	(PWMDWIDTH + 1) << CLKDIV) / freq

according to my tests with a signal analyser and also the documentation.

The current algorithm doesn't consider the `+ 1` part and so configures
slightly too high periods. The same issue exists for the duty cycle
setting. So subtract 1 from both the register values for period and
duty cycle. If period is 0, bail out, if duty_cycle is 0, just disable
the PWM which results in a constant low output.

Fixes: caf065f8fd58 ("pwm: Add MediaTek PWM support")
Signed-off-by: Uwe Kleine-König 
Reviewed-by: AngeloGioacchino Del Regno 
Link: https://lore.kernel.org/r/6d1fa87a76f8020bfe3171529b8e19baffceab10.1753717973.git.u.kleine-koenig@baylibre.com
Cc: stable@vger.kernel.org
Signed-off-by: Uwe Kleine-König 
Signed-off-by: Greg Kroah-Hartman

pwm: mediatek: Handle hardware enable and clock enable separately

2025-08-28T14:34:23+00:00

commit 704d918341c378c5f9505dfdf32d315e256d3846 upstream.

Stop handling the clocks in pwm_mediatek_enable() and
pwm_mediatek_disable(). This is a preparing change for the next commit
that requires that clocks and the enable bit are handled separately.

Also move these two functions a bit further up in the source file to
make them usable in pwm_mediatek_config(), which is needed in the next
commit, too.

Signed-off-by: Uwe Kleine-König 
Reviewed-by: AngeloGioacchino Del Regno 
Link: https://lore.kernel.org/r/55c94fe2917ece152ee1e998f4675642a7716f13.1753717973.git.u.kleine-koenig@baylibre.com
Cc: stable@vger.kernel.org
Signed-off-by: Uwe Kleine-König 
Signed-off-by: Greg Kroah-Hartman

pwm: imx-tpm: Reset counter if CMOD is 0

2025-08-28T14:34:22+00:00

commit 65c6f742ab14ab1a2679fba72b82dcc0289d96f1 upstream.

As per the i.MX93 TRM, section 67.3.2.1 "MOD register update", the value
of the TPM counter does NOT get updated when writing MOD.MOD unless
SC.CMOD != 0. Therefore, with the current code, assuming the following
sequence:

	1) pwm_disable()
	2) pwm_apply_might_sleep() /* period is changed here */
	3) pwm_enable()

and assuming only one channel is active, if CNT.COUNT is higher than the
MOD.MOD value written during the pwm_apply_might_sleep() call then, when
re-enabling the PWM during pwm_enable(), the counter will end up resetting
after UINT32_MAX - CNT.COUNT + MOD.MOD cycles instead of MOD.MOD cycles as
normally expected.

Fix this problem by forcing a reset of the TPM counter before MOD.MOD is
written.

Fixes: 738a1cfec2ed ("pwm: Add i.MX TPM PWM driver support")
Cc: stable@vger.kernel.org
Signed-off-by: Laurentiu Mihalcea 
Link: https://lore.kernel.org/r/20250728194144.22884-1-laurentiumihalcea111@gmail.com
Signed-off-by: Uwe Kleine-König 
Signed-off-by: Greg Kroah-Hartman

pwm: mediatek: Ensure to disable clocks in error path

2025-07-07T06:34:44+00:00

After enabling the clocks each error path must disable the clocks again.
One of them failed to do so. Unify the error paths to use goto to make it
harder for future changes to add a similar bug.

Fixes: 7ca59947b5fc ("pwm: mediatek: Prevent divide-by-zero in pwm_mediatek_config()")
Signed-off-by: Uwe Kleine-König 
Link: https://lore.kernel.org/r/20250704172728.626815-2-u.kleine-koenig@baylibre.com
Cc: stable@vger.kernel.org
Signed-off-by: Uwe Kleine-König

pwm: Fix invalid state detection

2025-07-07T06:33:44+00:00

Commit 9dd42d019e63 ("pwm: Allow pwm state transitions from an invalid
state") intended to allow some state transitions that were not allowed
before. The idea is sane and back then I also got the code comment
right, but the check for enabled is bogus. This resulted in state
transitions for enabled states to be allowed to have invalid duty/period
settings and thus it can happen that low-level drivers get requests for
invalid states🙄.

Invert the check to allow state transitions for disabled states only.

Fixes: 9dd42d019e63 ("pwm: Allow pwm state transitions from an invalid state")
Signed-off-by: Uwe Kleine-König 
Link: https://lore.kernel.org/r/20250704172416.626433-2-u.kleine-koenig@baylibre.com
Cc: stable@vger.kernel.org
Signed-off-by: Uwe Kleine-König

Merge tag 'pwm/for-6.16-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ukleinek/linux

2025-06-06T20:16:50+00:00

Pull pwm fixes from Uwe Kleine-König:
 "axi-pwmgen: Fix handling of external clock

  The pwm-axi-pwmgen device is backed by an FPGA and can be synthesized
  in different ways. Relevant here is that it can use one or two
  external clock signals. These fix clock handling for the two clocks
  case"

* tag 'pwm/for-6.16-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ukleinek/linux:
  pwm: axi-pwmgen: fix missing separate external clock
  dt-bindings: pwm: adi,axi-pwmgen: Fix clocks

Merge tag 'mfd-next-6.16' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd

2025-06-03T18:53:55+00:00

Pull MFD updates from Lee Jones:
 "Samsung Exynos ACPM:
   - Populate child platform devices from device tree data
   - Introduce a new API, 'devm_acpm_get_by_node()', for child devices
     to get the ACPM handle

  ROHM PMICs:
   - Add support for the ROHM BD96802 scalable companion PMIC to the
     BD96801 core driver
   - Add support for controlling the BD96802 using the BD96801 regulator
     driver
   - Add support to the BD96805, which is almost identical to the
     BD96801
   - Add support to the BD96806, which is similar to the BD96802

  Maxim MAX77759:
   - Add a core driver for the MAX77759 companion PMIC
   - Add a GPIO driver for the expander functions on the MAX77759
   - Add an NVMEM driver to expose the non-volatile memory on the
     MAX77759

  STMicroelectronics STM32MP25:
   - Add support for the STM32MP25 SoC to the stm32-lptimer
   - Add support for the STM32MP25 to the clocksource driver, handling
     new register access requirements
   - Add support for the STM32MP25 to the PWM driver, enabling up to two
     PWM outputs

  Broadcom BCM590xx:
   - Add support for the BCM59054 PMU
   - Parse the PMU ID and revision to support behavioral differences
     between chip revisions
   - Add regulator support for the BCM59054

  Samsung S2MPG10:
   - Add support for the S2MPG10 PMIC, which communicates via the
     Samsung ACPM firmware instead of I2C

  Exynos ACPM:
   - Improve timeout detection reliability by using ktime APIs instead
     of a loop counter assumption
   - Allow PMIC access during late system shutdown by switching to
     'udelay()' instead of a sleeping function
   - Fix an issue where reading command results longer than 8 bytes
     would fail
   - Silence non-error '-EPROBE_DEFER' messages during boot to clean up
     logs

  Exynos LPASS:
   - Fix an error handling path by switching to
     'devm_regmap_init_mmio()' to prevent resource leaks
   - Fix a bug where 'exynos_lpass_disable()' was called twice in the
     remove function
   - Fix another resource leak in the probe's error path by using
     'devm_add_action_or_reset()'

  Samsung SEC:
   - Handle the s2dos05, which does not have IRQ support, explicitly to
     prevent warnings
   - Fix the core driver to correctly handle errors from
     'sec_irq_init()' instead of ignoring them

  STMPE-SPI:
   - Correct an undeclared identifier in the 'MODULE_DEVICE_TABLE' macro

  MAINTAINERS:
   - Adjust a file path for the Siemens IPC LED drivers entry to fix a
     broken reference

  Maxim Drivers:
   - Correct the spelling of "Electronics" in Samsung copyright headers
     across multiple files

  General:
   - Fix wakeup source memory leaks on device unbind for 88pm886,
     as3722, max14577, max77541, max77705, max8925, rt5033, and
     sprd-sc27xx drivers

  Samsung SEC Drivers:
   - Split the driver into a transport-agnostic core ('sec-core') and
     transport-specific ('sec-i2c', 'sec-acpm') modules to support
     non-I2C devices
   - Merge the 'sec-core' and 'sec-irq' modules to reduce memory
     consumption
   - Move internal APIs to a private header to clean up the public API
   - Improve code style by sorting includes, cleaning up headers,
     sorting device tables, and using helper macros like
     'dev_err_probe()', 'MFD_CELL', and 'REGMAP_IRQ_REG'
   - Make regmap configuration for s2dos05/s2mpu05 explicit to improve
     clarity
   - Rework platform data and regmap instantiation to use OF match data
     instead of a large switch statement

  ROHM BD96801/2:
   - Prepare the driver for new models by separating chip-specific data
     into its own structure
   - Drop IC name prefix from IRQ resource names in both the MFD and
     regulator drivers for simplification

  Broadcom BCM590xx:
   - Refactor the regulator driver to store descriptions in a table to
     ease support for new chips
   - Rename BCM59056-specific data to prepare for the addition of other
     regulators
   - Use 'dev_err_probe()' for cleaner error handling

  Exynos ACPM:
   - Correct kerneldoc warnings and use the conventional 'np' argument
     name

  General MFD:
   - Convert 'aat2870' and 'tps65010' to use the per-client debugfs
     directory provided by the I2C core
   - Convert 'sm501', 'tps65010' and 'ucb1x00' to use the new GPIO line
     value setter callbacks
   - Constify 'regmap_irq_chip' and other structures in '88pm886' to
     move data to read-only sections

  BCM590xx:
   - Drop the unused "id" member from the 'bcm590xx' struct in
     preparation for a replacement

  Samsung SEC Core:
   - Remove forward declarations for functions that no longer exist

  SM501:
   - Remove the unused 'sm501_find_clock()' function

  New Compatibles:
   - Google: Add a PMIC child node to the 'google,gs101-acpm-ipc'
     binding
   - ROHM: Add new bindings for 'rohm,bd96802-regulator' and
     'rohm,bd96802-pmic', and add compatibles for BD96805 and BD96806
   - Maxim: Add new bindings for 'maxim,max77759-gpio',
     'maxim,max77759-nvmem', and the top-level 'maxim,max77759'
   - STM: Add 'stm32mp25' compatible to the 'stm32-lptimer' binding
   - Broadcom: Add 'bcm59054' compatible
   - Atmel/Microchip: Add 'microchip,sama7d65-gpbr' and
     'microchip,sama7d65-secumod' compatibles
   - Samsung: Add 's2mpg10' compatible to the 'samsung,s2mps11' MFD
     binding
   - MediaTek: Add compatibles for 'mt6893' (scpsys), 'mt7988-topmisc',
     and 'mt8365-infracfg-nao'
   - Qualcomm: Add 'qcom,apq8064-mmss-sfpb' and 'qcom,apq8064-sps-sic'
     syscon compatibles

  Refactoring & Cleanup:
   - Convert Broadcom BCM59056 devicetree bindings to YAML and split
     them into MFD and regulator parts
   - Convert the Microchip AT91 secumod binding to YAML
   - Drop unrelated consumer nodes from binding examples to reduce bloat
   - Correct indentation and style in various DTS examples"

* tag 'mfd-next-6.16' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd: (81 commits)
  mfd: maxim: Correct Samsung "Electronics" spelling in copyright headers
  mfd: maxim: Correct Samsung "Electronics" spelling in headers
  mfd: sm501: Remove unused sm501_find_clock
  mfd: 88pm886: Constify struct regmap_irq_chip and some other structures
  dt-bindings: mfd: syscon: Add mediatek,mt8365-infracfg-nao
  mfd: sprd-sc27xx: Fix wakeup source leaks on device unbind
  mfd: rt5033: Fix wakeup source leaks on device unbind
  mfd: max8925: Fix wakeup source leaks on device unbind
  mfd: max77705: Fix wakeup source leaks on device unbind
  mfd: max77541: Fix wakeup source leaks on device unbind
  mfd: max14577: Fix wakeup source leaks on device unbind
  mfd: as3722: Fix wakeup source leaks on device unbind
  mfd: 88pm886: Fix wakeup source leaks on device unbind
  dt-bindings: mfd: Correct indentation and style in DTS example
  dt-bindings: mfd: Drop unrelated nodes from DTS example
  dt-bindings: mfd: syscon: Add qcom,apq8064-sps-sic
  dt-bindings: mfd: syscon: Add qcom,apq8064-mmss-sfpb
  mfd: stmpe-spi: Correct the name used in MODULE_DEVICE_TABLE
  dt-bindings: mfd: syscon: Add mt7988-topmisc
  mfd: exynos-lpass: Fix another error handling path in exynos_lpass_probe()
  ...

pwm: axi-pwmgen: fix missing separate external clock

2025-06-02T16:23:33+00:00

Add proper support for external clock to the AXI PWM generator driver.

In most cases, the HDL for this IP block is compiled with the default
ASYNC_CLK_EN=1. With this option, there is a separate external clock
that drives the PWM output separate from the peripheral clock. So the
driver should be enabling the "axi" clock to power the peripheral and
the "ext" clock to drive the PWM output.

When ASYNC_CLK_EN=0, the "axi" clock is also used to drive the PWM
output and there is no "ext" clock.

Previously, if there was a separate external clock, users had to specify
only the external clock and (incorrectly) omit the AXI clock in order
to get the correct operating frequency for the PWM output.

The devicetree bindings are updated to fix this shortcoming and this
patch changes the driver to match the new bindings. To preserve
compatibility with any existing dtbs that specify only one clock, we
don't require the clock name on the first clock.

Fixes: 41814fe5c782 ("pwm: Add driver for AXI PWM generator")
Cc: stable@vger.kernel.org
Acked-by: Nuno Sá 
Reviewed-by: Trevor Gamblin 
Signed-off-by: David Lechner 
Link: https://lore.kernel.org/r/20250529-pwm-axi-pwmgen-add-external-clock-v3-3-5d8809a7da91@baylibre.com
Signed-off-by: Uwe Kleine-König

pwm: adp5585: make sure to include mod_devicetable.h

2025-05-19T16:07:39+00:00

Explicitly include mod_devicetable.h for struct platform_device_id.

Reviewed-by: Laurent Pinchart 
Signed-off-by: Nuno Sá 
Link: https://lore.kernel.org/r/20250512-dev-adp5589-fw-v3-22-092b14b79a88@analog.com
Signed-off-by: Uwe Kleine-König

pwm: Tidyup PWM menu for Renesas

2025-05-14T08:20:55+00:00

Because current PWM Kconfig is sorting by symbol name,
it looks strange ordering in menuconfig.

=>	[ ]   Renesas R-Car PWM support
=>	[ ]   Renesas TPU PWM support
	[ ]   Rockchip PWM support
=>	[ ]   Renesas RZ/G2L General PWM Timer support
=>	[ ]   Renesas RZ/G2L MTU3a PWM Timer support

Let's use common CONFIG_PWM_RENESAS_xxx symbol name for Renesas,
and sort it.

Signed-off-by: Kuninori Morimoto 
Reviewed-by: Biju Das 
Link: https://lore.kernel.org/r/877c2mxrrr.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Uwe Kleine-König