This improves the driver's behavior in several ways:
- The lock is held for shorter periods and so a channel that is currently
waited for doesn't block disabling another channel.
- It's easier to understand because the procedure is split into more
semantic units and documentation is improved
- A channel is only set to pending when such an event is actually
scheduled in hardware (by writing the CUPD register).
- Also wait in .get_state() to report the last configured state instead
of (maybe) the previous one. This fixes the read back duty cycle and so
prevents a warning being emitted when PWM_DEBUG is on.
Tested on an AriettaG25.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
With the previous commit there is no need for the lowlevel driver any
more to specify it it uses two or three cells. So simplify accordingly.
The only non-trival change affects the pwm-rockchip driver: It used to only
support three cells if the hardware supports polarity. Now the default
number depends on the device tree which has to match hardware anyhow
(and if it doesn't the error is just a bit delayed as a PWM handle with
an inverted setting is catched when pwm_apply_state() is called).
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
In the calculation of the register value determining the duty cycle the
requested period is used instead of the actually implemented period which
results in suboptimal settings.
The following example assumes an input clock of 133333333 Hz on one of
the SoCs with 16 bit period.
When the following state is to be applied:
.period = 414727681
.duty_cycle = 652806
the following register values used to be calculated:
PRES = 10
CPRD = 54000
CDTY = 53916
which yields an actual duty cycle of a bit more than 645120 ns.
The setting
PRES = 10
CPRD = 54000
CDTY = 53915
however yields a duty of 652800 ns which is between the current result
and the requested value and so is a better approximation.
The reason for this error is that for the calculation of CDTY the
requested period was used instead of the actually implemented one.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
The CDTY register contains the number of inactive cycles. .apply() does
this correctly, however .get_state() got this wrong.
Fixes: 651b510a74 ("pwm: atmel: Implement .get_state()")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Before pwmchip_remove() returns the PWM is expected to be functional. So
remove the pwmchip before disabling the clock.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Since commit 5e5da1e9fb ("pwm: ab8500: Explicitly allocate pwm chip
base dynamically") all drivers use dynamic ID allocation explicitly. New
drivers are supposed to do the same, so remove support for driver
specified base IDs and drop all assignments in the low-level drivers.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
This function reads back the configured parameters from the hardware. As
.apply() rounds down (mostly) I'm rounding up in .get_state() to achieve
that applying a state just read from hardware is a no-op.
Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Acked-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
This makes it a bit easier when instrumenting register access to only
have to add code in one place.
Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
This documents the my findings while reading through the driver and the
reference manual.
Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
The calculated values are the same with the modified algorithm. The only
difference is that the calculation is a bit more efficient.
Acked-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
The maximal prescale value is 10 for all supported variants. So drop the
member in the variant description and introduce a global constant
instead.
This reduces the size of the variant descriptions and the .apply()
callback can be compiled a bit more effectively.
Acked-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Most Microchip (formerly Atmel) chips have publicly available manuals.
A comprehensive list is already contained in the documentation folder.
Reference this list in the header of the driver to allow reviewers to
find the relevant manuals.
Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Since the driver is now exclusively DT, it only binds if it finds a
match in the of_device_id table. But in that case the associated data
can never be NULL, so drop the unnecessary check.
While at it, drop the extra local variable and store the pointer to
this per-SoC data in the driver data directly.
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Since commit 26202873bb ("avr32: remove support for AVR32
architecture") there is no more user of platform_device_id and we
should only use dt bindings
Signed-off-by: Kamel Bouhara <kamel.bouhara@bootlin.com>
Acked-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Acked-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
It is surprising for a PWM consumer when the variable holding the
requested state is modified by pwm_apply_state(). Consider for example a
driver doing:
#define PERIOD 5000000
#define DUTY_LITTLE 10
...
struct pwm_state state = {
.period = PERIOD,
.duty_cycle = DUTY_LITTLE,
.polarity = PWM_POLARITY_NORMAL,
.enabled = true,
};
pwm_apply_state(mypwm, &state);
...
state.duty_cycle = PERIOD / 2;
pwm_apply_state(mypwm, &state);
For sure the second call to pwm_apply_state() should still have
state.period = PERIOD and not something the hardware driver chose for a
reason that doesn't necessarily apply to the second call.
So declare the state argument as a pointer to a const type and adapt all
drivers' .apply callbacks.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Based on 1 normalized pattern(s):
licensed under gplv2
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-only
has been chosen to replace the boilerplate/reference in 99 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Alexios Zavras <alexios.zavras@intel.com>
Reviewed-by: Richard Fontana <rfontana@redhat.com>
Reviewed-by: Allison Randal <allison@lohutok.net>
Reviewed-by: Steve Winslow <swinslow@gmail.com>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190528170027.163048684@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The values that these symbols define are only assigned to the per-SoC
structure where the context is clear, so there's no need for the extra
symbolic name.
Acked-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Rename objects of type atmel_pwm_data to contain chip name instead of
version number.
Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
SAM9X60's PWM controller uses 32 bits counters thus it could generate
signals with higher period and duty cycles than the old ones. Prepare
the current driver to be able to work with old controllers (that uses
16 bits counters) and with the new SAM9X60's controller, by providing
counters information based on compatible string.
Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Add struct atmel_pwm_data to embed different per controller information.
It prepares adding support for another similar controller that needs
additional information. At this stage, embed a member of type struct
atmel_pwm_registers in it.
Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
sama5d2 can use the same atmel_pwm_data as sama5d3.
Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Acked-by: Rob Herring <robh@kernel.org>
Reviewed-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
The currently Atmel PWM controllers supported by this driver could
change period or duty factor without channel disable, for regular
channels (sama5d3 support this by using period or duty factor update
registers, sam9rl support this by writing channel update register and
select the corresponding update: period or duty factor). The chip
doesn't support run time changings of signal polarity. To take advantage
of atomic PWM framework and let controller works without glitches, in
this patch only the duty factor could be changed without disabling PWM
channel. For period and signal polarity the atomic PWM is simulated by
disabling + enabling the right PWM channel.
Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Reviewed-by: Boris Brezillon <boris.brezillon@free-electrons.com>
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>
When disabling a PWM channel, the PWM clock was being stopped
immediately after writing to PWM_DIS. As a result, the disabling
of the PWM channel did not complete properly, and the PWM output
might be left at the wrong level.
Fix this by waiting for the channel to be effectively disabled
(by checking the PWM_SR register) before disabling the clock.
Signed-off-by: Guillermo Rodriguez <guille.rodriguez@gmail.com>
Acked-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Use the pwm_get_xxx() helpers instead of directly accessing the fields
in struct pwm_device. This will allow us to smoothly move to the atomic
update approach.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Some PWM drivers are testing the PWMF_ENABLED flag. Create a helper
function to hide the logic behind enabled test. This will allow us to
smoothly move from the current approach to an atomic PWM update
approach.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
pwm-leds calls .config() and .disable() in a row. This exhibits that it
may happen that the channel gets disabled before CDTY has been updated
with CUPD. The issue gets quite worse with long periods. So, ensure that
at least one period has past before disabling the channel by polling
ISR.
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Tested-by: Gaël PORTAY <gael.portay@gmail.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
CUPD is not flushed before enabling the channel so it will update
CDTY/CPRD just after one period. So we always set CUPD, even when the
channel is not enabled.
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
The prescale value used for calculating the period was incremented
afterwards, thus the resulting prescale value is by one too high.
This resulted in a PWM frequency only half as high as requested.
This patch moves the 64 bit division out of the prescale loop to
correct the above issue and make the calculation more efficient.
Signed-off-by: Nikolaus Voss <n.voss@weinmann-emt.de>
Tested-by: Bo Shen <voice.shen@atmel.com>
Acked-by: Bo Shen <voice.shen@atmel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
atmel_pwm_config() calls clk_get_rate() which might sleep, so we need to
set pwm_chip can_sleep flag.
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
From the datasheet, the actual duty cycle is:
(period - (1 / clk) * CDTY) / period
This actually correct the polarity of the PWM and solves the issue that
pwm-leds exhibits: when setting a duty cycle of 0 and then disabling a
channel, the level was wrong (1 when the polarity was normal and 0 when
the polarity was inversed).
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
When atmel_pwm_config() calculates and then sets the prescaler, it is
overwriting the channel's CMR register so we are losing the CPOL
configuration.
As atmel_pwm_config() is always called before enabling a channel,
inverting the polarity doesn't work.
Fix that by reading CMR first and only overwriting the prescaler bits.
Signed-off-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
When the PWM controller is registered successfully, the clock can not
unprepare, so fix it.
Signed-off-by: Bo Shen <voice.shen@atmel.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Add a PWM framework driver for the PWM controller found on Atmel SoCs.
Signed-off-by: Bo Shen <voice.shen@atmel.com>
Acked-by: Alexandre Belloni <alexandre.belloni@free-electrons.com>
Acked-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
[thierry.reding: coding style and other minor cleanups]
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Uwe Kleine-König
Yangtao Li
Uwe Kleine-König
Thierry Reding
Kamel Bouhara
Thomas Gleixner
Claudiu Beznea
Guillermo Rodriguez
Boris Brezillon
Alexandre Belloni
Nikolaus Voss
Bo Shen