linux-toradex.git/drivers/regulator/s2mps11.c, branch v7.2-rc5

regulator: Use named initializers for platform_device_id arrays

2026-06-02T15:09:24+00:00

Named initializers are better readable and more robust to changes of the
struct definition. This robustness is relevant for a planned change to
struct platform_device_id replacing .driver_data by an anonymous unit.

While touching these arrays unify spacing and usage of commas.

Signed-off-by: Uwe Kleine-König (The Capable Hub) 
Acked-by: Karel Balej 
Reviewed-by: Matti Vaittinen 
Link: https://patch.msgid.link/d02f55dfd5bdd743ae5cd76f2a5af0d346226a68.1779878004.git.u.kleine-koenig@baylibre.com
Signed-off-by: Mark Brown

regulator: s2mps11: fix pctrlsel macro usage in s2mpg10_of_parse_cb()

2026-02-16T16:30:47+00:00

Commit 979dd8da76eb ("regulator: s2mps11: add S2MPG11 regulator")
incorrectly ended up using macros for S2MPG10 in the S2MPG11 case. They
happen to end up giving the same result, but for clarity, the correct
macros should be used.

No functional change.

Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260210-s2mpg1x-regulators-fixes-v2-2-ab3d3457f1ae@linaro.org
Signed-off-by: Mark Brown

regulator: s2mps11: drop redundant sanity checks in s2mpg10_of_parse_cb()

2026-02-16T16:30:46+00:00

The sanity checks being removed in this commit are useless as earlier
code already checks for all conditions, including all error cases like
out-of-bounds conditions. In other words, the code being removed here
has no effect, as any potential error it could catch will already have
been caught by earlier code.

The checks removed here are also incomplete (as they're off-by-one) -
they should have been checking >= ARRAY_SIZE() to be complete.

Simply remove this redundant and incorrect code.

No functional change.

Reported-by: Dan Carpenter 
Closes: https://lore.kernel.org/all/aYmsu8qREppwBESH@stanley.mountain/
Reviewed-by: Krzysztof Kozlowski 
Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260210-s2mpg1x-regulators-fixes-v2-1-ab3d3457f1ae@linaro.org
Signed-off-by: Mark Brown

regulator: s2mps11: more descriptive gpio consumer name

2026-02-04T13:35:41+00:00

Currently, GPIOs claimed by this driver for external rail control
all show up with "s2mps11-regulator" as consumer, which is not
very informative.

Switch to using the regulator name via desc->name instead, using the
device name as fallback.

Reviewed-by: Bartosz Golaszewski 
Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260122-s2mpg1x-regulators-v7-20-3b1f9831fffd@linaro.org
Signed-off-by: Mark Brown

regulator: s2mps11: add S2MPG11 regulator

2026-02-04T13:35:40+00:00

The S2MPG11 PMIC is a Power Management IC for mobile applications with
buck converters, various LDOs, power meters, and additional GPIO
interfaces. It typically complements an S2MPG10 PMIC in a main/sub
configuration as the sub-PMIC.

It has 12 buck, 1 buck-boost, and 15 LDO rails. Several of these can
either be controlled via software (register writes) or via external
signals, in particular by:
    * input pins connected to a main processor's:
        * GPIO pins
        * other pins that are e.g. firmware- or power-domain-controlled
          without explicit driver intervention
    * a combination of input pins and register writes.

Control via input pins allows PMIC rails to be controlled by firmware,
e.g. during standby/suspend or as part of power domain handling where
otherwise that would not be possible. Additionally toggling a pin is
faster than register writes, and it also allows the PMIC to ensure that
any necessary timing requirements between rails are respected
automatically if multiple rails are to be enabled or disabled quasi
simultaneously.

This commit implements support for all these rails and control
combination.

Note1: For an externally controlled rail, the regulator_ops provide an
empty ::enable() and no ::disable() implementations, even though Linux
can not enable the rail and one might think ::enable could be NULL.
Without ops->enable(), the regulator core will assume enabling such a
rail failed, though, and in turn never add a reference to its parent
(supplier) rail. Once a different (Linux-controlled) sibling (consumer)
rail on that same parent rail gets disabled, the parent gets disabled
(cutting power to the externally controlled rail although it should
stay on), and the system will misbehave.

Note2:  While external control via input pins appears to exist on other
versions of this PMIC, there is more flexibility in this version, in
particular there is a selection of input pins to choose from for each
rail (which must therefore be configured accordingly if in use),
whereas other versions don't have this flexibility.

Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260122-s2mpg1x-regulators-v7-19-3b1f9831fffd@linaro.org
Signed-off-by: Mark Brown

regulator: s2mps11: refactor S2MPG10 regulator macros for S2MPG11 reuse

2026-02-04T13:35:39+00:00

Rails in the S2MPG11 share a very similar set of properties with
S2MPG10 with slight differences. Update the existing macros to allow
reuse by the upcoming S2MPG11 driver.

Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260122-s2mpg1x-regulators-v7-18-3b1f9831fffd@linaro.org
Signed-off-by: Mark Brown

regulator: s2mps11: refactor S2MPG10 ::set_voltage_time() for S2MPG11 reuse

2026-02-04T13:35:38+00:00

The upcoming S2MPG11 support needs a similar, but different version of
::set_voltage_time(). For S2MPG10, the downwards and upwards ramps for
a rail are at different offsets at the same bit positions, while for
S2MPG11 the ramps are at the same offset at different bit positions.

Refactor the existing version slightly to allow reuse.

Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260122-s2mpg1x-regulators-v7-17-3b1f9831fffd@linaro.org
Signed-off-by: Mark Brown

regulator: s2mps11: add S2MPG10 regulator

2026-02-04T13:35:37+00:00

The S2MPG10 PMIC is a Power Management IC for mobile applications with
buck converters, various LDOs, power meters, RTC, clock outputs, and
additional GPIO interfaces.

It has 10 buck and 31 LDO rails. Several of these can either be
controlled via software (register writes) or via external signals, in
particular by:
    * one out of several input pins connected to a main processor's:
        * GPIO pins
        * other pins that are e.g. firmware- or power-domain-controlled
          without explicit driver intervention
    * a combination of input pins and register writes.

Control via input pins allows PMIC rails to be controlled by firmware,
e.g. during standby/suspend, or as part of power domain handling where
otherwise that would not be possible. Additionally toggling a pin is
faster than register writes, and it also allows the PMIC to ensure that
any necessary timing requirements between rails are respected
automatically if multiple rails are to be enabled or disabled quasi
simultaneously.

This commit implements support for all these rails and control
combinations.

Additional data needs to be stored for each regulator, e.g. the input
pin for external control, or a rail-specific ramp-rate for when
enabling a buck-rail. Therefore, probe() is updated slightly to make
that possible.

Note1: For an externally controlled rail, the regulator_ops provide an
empty ::enable() and no ::disable() implementations, even though Linux
can not enable the rail and one might think ::enable could be NULL.
Without ops->enable(), the regulator core will assume enabling such a
rail failed, though, and in turn never add a reference to its parent
(supplier) rail. Once a different (Linux-controlled) sibling (consumer)
rail on that same parent rail gets disabled, the parent gets disabled
(cutting power to the externally controlled rail although it should
stay on), and the system will misbehave.

Note2: While external control via input pins appears to exist on other
versions of this PMIC, there is more flexibility in this version, in
particular there is a selection of input pins to choose from for each
rail (which must therefore be configured accordingly if in use),
whereas other versions don't have this flexibility.

Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260122-s2mpg1x-regulators-v7-16-3b1f9831fffd@linaro.org
Signed-off-by: Mark Brown

regulator: s2mps11: refactor handling of external rail control

2026-02-04T13:35:36+00:00

Refactor s2mps14_pmic_enable_ext_control() and s2mps11_of_parse_cb()
slightly as a preparation for adding S2MPG10 and S2MPG11 support, as
both of those PMICs also support control of rails via GPIOs.

This also includes the following to avoid further updates in follow-up
commits:
* On S2MPG10 and S2MPG11, external rail control can be via GPIO or via
  non-GPIO signals, hence passing a GPIO is allowed to be optional.
  This avoids inappropriate verbose driver messages.
* Prepare to allow use of standard DT property name 'enable-gpios' for
  newer platforms instead of vendor-specific 'samsung,ext-control'.

Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260122-s2mpg1x-regulators-v7-15-3b1f9831fffd@linaro.org
Signed-off-by: Mark Brown

regulator: s2mps11: update node parsing (allow -supply properties)

2026-02-04T13:35:35+00:00

For the upcoming S2MPG10 and S2MPG11 support, we need to be able to
parse -supply properties in the PMIC's DT node.

This currently doesn't work, because the code here currently points the
regulator core at each individual regulator sub-node, and therefore the
regulator core is unable to find the -supply properties.

Update the code to simply let the regulator core handle all the parsing
by adding the ::of_match and ::regulators_node members to all existing
regulator descriptions, by adding ::of_parse_cb() to those
regulators which support the vendor-specific samsung,ext-control-gpios
to parse it (S2MPS14), and by dropping the explicit call to
of_regulator_match().

Configuring the PMIC to respect the external control GPIOs via
s2mps14_pmic_enable_ext_control() is left outside ::of_parse_cb()
because the regulator core ignores errors other than -EPROBE_DEFER from
that callback, while the code currently fails probe on register write
errors and I believe it should stay that way.

The driver can now avoid the devm_gpiod_unhinge() dance due to
simpler error handling of GPIO descriptor acquisition.

This change also has the advantage of reducing runtime memory
consumption by quite a bit as the driver doesn't need to allocate a
'struct of_regulator_match' and a 'struct gpio_desc *' for each
regulator for all PMICs as the regulator core does that. This saves
40+8 bytes on arm64 for each individual regulator on all supported
PMICs (even on non-S2MPS14 due to currently unnecessarily allocating
the extra memory unconditionally). With the upcoming S2MPG10 and
S2MPG11 support, this amounts to 1640+328 and 1120+224 bytes
respectively.

Signed-off-by: André Draszik 
Link: https://patch.msgid.link/20260122-s2mpg1x-regulators-v7-14-3b1f9831fffd@linaro.org
Signed-off-by: Mark Brown