GPIO-only driver operation never clears the SLEEP bit, which can cause
the GPIOs to become unusable.
Example:
1. user requests first PWM -> driver clears SLEEP bit
2. user frees last PWM -> driver sets SLEEP bit
3. user requests GPIO
4. user switches GPIO on -> output does not turn on
because SLEEP bit is set
Prevent this behaviour by letting the runtime PM framework control the
SLEEP bit. This will put the chip to SLEEP if no PWMs/GPIOs are exported
or in use.
Fixes: bccec89f0a ("Allow any of the 16 PWMs to be used as a GPIO")
Reported-by: Sven Van Asbroeck <TheSven73@googlemail.com>
Signed-off-by: Sven Van Asbroeck <TheSven73@googlemail.com>
Suggested-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
When first implementing support for changing the output frequency, an
optimization was added to continue the PWM after changing the prescaler
without having to reprogram the ON and OFF registers for the duty cycle,
in case the duty cycle stayed the same. This was flawed, because we
compared the absolute value of the duty cycle in nanoseconds instead of
the ratio to the period.
Fix the problem by removing the shortcut.
Fixes: 01ec847200 ("pwm-pca9685: Support changing the output frequency")
Cc: <stable@vger.kernel.org> # v4.3+
Signed-off-by: Clemens Gruber <clemens.gruber@pqgruber.com>
Reviewed-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
The PCA9685 controller has full on/off bit for each PWM channel. Setting
this bit bypasses the PWM control and the line works just as it would be a
GPIO. Furthermore in Intel Galileo it is actually used as GPIO output for
discreet muxes on the board.
This patch adds GPIO output only support for the driver so that we can
control the muxes on Galileo using standard GPIO interfaces available in
the kernel. GPIO and PWM functionality is exclusive so only one can be
active at a time on a single PWM channel.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
All PWM devices have been marked as "might sleep" since v4.5, there is
no longer a need to differentiate on a per-chip basis.
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
There is a chip connected to i2c bus on Intel Galileo Gen2 board. Enable it via
ACPI ID INT3492.
Cc: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Previously, period_ns and duty_ns were only used to determine the
ratio of ON and OFF time, the default frequency of 200 Hz was never
changed.
The PCA9685 however is capable of changing the PWM output frequency,
which is expected when changing the period.
This patch configures the prescaler accordingly, using the formula
and notes provided in the PCA9685 datasheet.
Bounds checking for the minimum and maximum frequencies, last updated
in revision v.4 of said datasheet, is also added.
The prescaler is only touched if the period changed, because we have to
put the chip into sleep mode to unlock the prescale register.
If it is changed, the PWM output frequency changes for all outputs,
because there is one prescaler per chip. This is documented in the
PCA9685 datasheet and in the comments.
If the duty cycle is not changed at the same time as the period, then
we restart the PWM output using the duty cycle to period ratio from
before the period change.
When using LEDs for example, previously set brightness levels stay the
same when the frequency changes.
Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: Steffen Trumtrar <s.trumtrar@pengutronix.de>
Signed-off-by: Clemens Gruber <clemens.gruber@pqgruber.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Problems:
- When duty_ns == period_ns, the full OFF bit was not cleared and the
PWM output of the PCA9685 stayed off.
- When duty_ns == period_ns and the catch-all channel was used, the
ALL_LED_OFF_L register was not cleared.
- The full ON bit was not cleared when setting the OFF time, therefore
the exact OFF time was ignored when setting a duty_ns < period_ns
Solution: Clear both OFF registers when setting full ON and clear the
full ON bit when changing the OFF registers.
Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: Steffen Trumtrar <s.trumtrar@pengutronix.de>
Signed-off-by: Clemens Gruber <clemens.gruber@pqgruber.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
i2c_driver does not need to set an owner because i2c_register_driver()
will set it.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
The regmap_config struct may be const because it is not modified by the
driver and regmap_init() accepts pointer to const.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Current code actually does not set MODE1_SLEEP bit because the new value for
bitmask (0x1) is wrong. To set MODE1_SLEEP bit, we should pass MODE1_SLEEP
as the new value for bitmask.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Reviewed-by: Steffen Trumtrar <s.trumtrar@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Some drivers don't set the .owner fields of the struct device_driver or
struct pwm_ops, which causes the module usage count to become wrong.
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Add pwm driver for the NXP pca9685 16 channel pwm-led controller.
The driver is really barebones at this stage. E.g. the OE' pin and
therefore the corresponding registers are not supported.
The driver was tested on a HW where this pin is tied to GND.
Signed-off-by: Steffen Trumtrar <s.trumtrar@pengutronix.de>
[thierry.reding@gmail.com: style and whitespace cleanups]
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>