Commit 4a6ef8e37c4d ("pwm: Add support referencing PWMs from ACPI")
made pwm_get unconditionally return the acpi_pwm_get return value if
the device passed to pwm_get has an ACPI fwnode.

But even if the passed in device has an ACPI fwnode, it does not
necessarily have the necessary ACPI package defining its pwm bindings,
especially since the binding / API of this ACPI package has only been
introduced very recently.

Up until now X86/ACPI devices which use a separate pwm controller for
controlling their LCD screen's backlight brightness have been relying
on the static lookup-list to get their pwm.

pwm_get unconditionally returning the acpi_pwm_get return value breaks
this, breaking backlight control on these devices.

This commit fixes this by making pwm_get fall back to the static
lookup-list if acpi_pwm_get returns -ENOENT.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=96571
Reported-by: youling257@gmail.com
Fixes: 4a6ef8e37c4d ("pwm: Add support referencing PWMs from ACPI")
Cc: Nikolaus Voss <nikolaus.voss@loewensteinmedical.de>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Nikolaus Voss <nikolaus.voss@loewensteinmedical.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

Pull pwm updates from Thierry Reding:
 "This set of changes contains a new driver for SiFive SoCs as well as
  enhancements to the core (device links are used to track dependencies
  between PWM providers and consumers, support for PWM controllers via
  ACPI, sysfs will now suspend/resume PWMs that it has claimed) and
  various existing drivers"

* tag 'pwm/for-5.3-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm: (37 commits)
  pwm: fsl-ftm: Make sure to unlock mutex on failure
  pwm: fsl-ftm: Use write protection for prescaler & polarity
  pwm: fsl-ftm: More relaxed permissions for updating period
  pwm: atmel-hlcdc: Add compatible for SAM9X60 HLCDC's PWM
  pwm: bcm2835: Improve precision of PWM
  leds: pwm: Support ACPI via firmware-node framework
  pwm: Add support referencing PWMs from ACPI
  pwm: rcar: Remove suspend/resume support
  pwm: sysfs: Add suspend/resume support
  pwm: Add power management descriptions
  pwm: meson: Add documentation to the driver
  pwm: meson: Add support PWM_POLARITY_INVERSED when disabling
  pwm: meson: Don't cache struct pwm_state internally
  pwm: meson: Read the full hardware state in meson_pwm_get_state()
  pwm: meson: Simplify the calculation of the pre-divider and count
  pwm: meson: Move pwm_set_chip_data() to meson_pwm_request()
  pwm: meson: Add the per-channel register offsets and bits in a struct
  pwm: meson: Add the meson_pwm_channel data to struct meson_pwm
  pwm: meson: Pass struct pwm_device to meson_pwm_calc()
  pwm: meson: Don't duplicate the polarity internally
  ...

Upon failure to enable clocks while trying to enable the PWM, make sure
to unlock the mutex that was taken to avoid a deadlock during subsequent
operations.

Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Julia Lawall <julia.lawall@lip6.fr>
Cc: Patrick Havelange <patrick.havelange@essensium.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

Modifying the prescaler or polarity value must be done with the
write protection disabled. Currently this is working by chance as
the write protection is in a disabled state by default.
This patch makes sure that we enable/disable the write protection
when needed.

Signed-off-by: Patrick Havelange <patrick.havelange@essensium.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

The Flextimer has only one period for several channels. The PWM
subsystem doesn't allow to model something like that. The current
implementation simply disallows changing the period once it has
been set, having as a side effect that you need to enable and
disable the PWM if you want to change the period.

The driver should allow as much freedom as possible for configuring
the period and duty cycle. Therefore, this patch reworks the code
to allow the following:

- period and duty_cycle can be set at will when the PWM is disabled;
- when enabling a PWM, verify that the period is either not set yet,
  or the same as the other already enabled PWM(s), and fail if not;
- allow to change the period on the fly when the PWM is the only one
  enabled.

It also allows to have different periods configured for different PWMs.
Only one period can be used at a time, thus the first PWM to be enabled
will set that period, only other PWMs with that same period can be
enabled at the same time. To use another PWM with another period, the
enabled PWMs must be disabled first.

Example scenario :
echo 5000000 > pwm0/period	#OK
echo 1000000 > pwm0/duty_cycle	#OK
echo 1000000 > pwm1/period	#OK
echo 1000000 > pwm1/duty_cycle	#OK
echo 1 > pwm0/enable		#OK
echo 1 > pwm1/enable		#FAIL (pwm0/period != pwm1/period)
echo 0 > pwm0/enable		#OK
echo 1 > pwm1/enable		#OK
echo 1000000 > pwm0/period	#OK
echo 2000000 > pwm0/period	#OK
echo 1 > pwm0/enable		#FAIL (pwm0/period != pwm1/period)
echo 2000000 > pwm1/period	#OK (pwm1 still running, changed on the fly)
echo 1 > pwm0/enable		#OK (now pwm0/period == pwm1/period)
echo 3000000 > pwm1/period	#FAIL (other PWMs running)
echo 0 > pwm0/enable		#OK
echo 3000000 > pwm1/period	#OK (only this PWM running)

Adapting the code to satisfy these constraints turned up a number of
additional issues with the current implementation:
- the prescaler value 0 was not used (when it could have been);
- when setting the period was not possible, the internal state was
  inconsistent;
- the maximal value for configuring the period was never used;

Since all of these interact with each other, rather than trying to fix
each individual issue, this patch reworks how the period and duty cycle
are set entirely, with the following additional improvements:
- implement the new apply() method instead of the individual methods;
- return the exact used period/duty_cycle values;
- more coherent argument types for period, duty_cycle;

Signed-off-by: Patrick Havelange <patrick.havelange@essensium.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

Add compatible string for SAM9X60 HLCDC's PWM.

Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

If sending IR with carrier of 455kHz using the pwm-ir-tx driver, the
carrier ends up being 476kHz. The clock is set to bcm2835-pwm with a
rate of 10MHz.

A carrier of 455kHz has a period of 2198ns, but the arithmetic truncates
this to 2100ns rather than 2200ns. So, use DIV_ROUND_CLOSEST() to reduce
rounding errors, and we have a much more accurate carrier of 454.5kHz.

Reported-by: Andreas Christ <andreas@christ-faesch.ch>
Signed-off-by: Sean Young <sean@mess.org>
Acked-by: Stefan Wahren <wahrenst@gmx.net>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

In analogy to referencing a GPIO using the "gpios" property from ACPI,
support referencing a PWM using the "pwms" property.

ACPI entries must look like
 Package () {"pwms", Package ()
     { <PWM device reference>, <PWM index>, <PWM period> [, <PWM flags>]}}

In contrast to the DT implementation, only _one_ PWM entry in the "pwms"
property is supported. As a consequence "pwm-names"-property and
con_id lookup aren't supported.

Support for ACPI is added via the firmware-node framework which is an
abstraction layer on top of ACPI/DT. To keep this patch clean, DT and
ACPI paths are kept separate. The firmware-node framework could be used
to unify both paths in a future patch.

To support leds-pwm driver, an additional method devm_fwnode_pwm_get()
which supports both ACPI and DT configuration is exported.

Signed-off-by: Nikolaus Voss <nikolaus.voss@loewensteinmedical.de>
[thierry.reding@gmail.com: fix build failures for !ACPI]
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

According to the Documentation/pwm.txt, all PWM consumers should
implement power management instead of the PWM driver. So, this
patch removes suspend/resume support.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

According to the Documentation/pwm.txt, all PWM consumers should have
power management. Since this sysfs interface is one of consumers so that
this patch adds suspend/resume support.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
[thierry.reding@gmail.com: fix build warnings for !PM]
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>