Re: [PATCH v2 01/16] drivers: pwm: core: use a single of xlate function

[Claudiu Beznea <Claudiu.Beznea-UWL1GkI3JZL3oGB3hsPCZA@public.gmane.org>]

Hi Boris,

Thanks for your review. See below my answers.

On 12.01.2018 20:35, Brian Norris wrote:
> Hi,
> 
> On Fri, Jan 12, 2018 at 04:22:48PM +0200, Claudiu Beznea wrote:
>> Remove of_pwm_simple_xlate() and of_pwm_xlate_with_flags() functions
>> and add of_pwm_xlate() which is used in all cases no mather if the OF
>> bindings are with PWM flags or not. This should not affect the old
>> behavior since the xlate will be based on #pwm-cells property of the
>> PWM controller. Based on #pwm-cells property the xlate will consider
>> the flags or not. This will permit the addition of other inputs to OF
>> xlate by just adding proper code at the end of of_pwm_xlate() and a new
>> input to enum pwm_args_xlate_options. With this changes there will be
>> no need to fill of_xlate and of_pwm_n_cells of struct pwm_chip from
>> the drivers probe methods. References in drives to references to of_xlate
>> and of_pwm_n_cells were removed. Drivers which used private of_xlate
>> functions switched to the generic of_pwm_xlate() function which fits
>> for it but with little changes in device trees (these drivers translated
>> differently the "pwms" bindings; the "pwms" bindings now are generic to
>> all drivers and all drivers should provide them in the format described
>> in pwm documentation).
>>
>> Cc: Thierry Reding <thierry.reding-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
>> Cc: Mike Dunn <mikedunn-kFrNdAxtuftBDgjK7y7TUQ@public.gmane.org>
>> Cc: Brian Norris <briannorris-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
>> Cc: Alexander Shiyan <shc_work-JGs/UdohzUI@public.gmane.org>
>> Signed-off-by: Claudiu Beznea <claudiu.beznea-UWL1GkI3JZL3oGB3hsPCZA@public.gmane.org>
>> ---
>>
>> This patch (and the next 7) could be applied independetly by this series, if
>> any, but I choosed to have it here since it makes easy the PWM modes parsing.
>> If you feel it could be independently of this series I could send a new version.
>>
>> Also, Thierry, Mike, Brian, Shiyan, please take an extra look over pwm-pxa.c,
>> pwm-cros-ec.c and pwm-clps711x.c since these were moved to use the generic
>> pwms (minimum 2 pwm-cells).
>>
>>  drivers/pwm/core.c             | 56 +++++++++++-------------------------------
>>  drivers/pwm/pwm-atmel-hlcdc.c  |  2 --
>>  drivers/pwm/pwm-atmel-tcb.c    |  2 --
>>  drivers/pwm/pwm-atmel.c        |  6 -----
>>  drivers/pwm/pwm-bcm-iproc.c    |  2 --
>>  drivers/pwm/pwm-bcm-kona.c     |  2 --
>>  drivers/pwm/pwm-bcm2835.c      |  2 --
>>  drivers/pwm/pwm-berlin.c       |  2 --
>>  drivers/pwm/pwm-clps711x.c     | 11 ---------
>>  drivers/pwm/pwm-cros-ec.c      | 20 ---------------
> 
> For pwm-cros-ec.c:
> 
> Nacked-by: Brian Norris <briannorris-F7+t8E8rja9g9hUCZPvPmw@public.gmane.org>
> 
> This is a fiat change of the documented binding, which breaks the RK3399
> Kevin board. That's not how we do device tree.
> 
> You can extend the binding if you want, so you can represent the period
> in the device tree if you'd like (though the value won't mean anything;
> it can't be changed by the kernel), but don't break existing device
> trees.

That wasn't the idea, I wasn't intended to break something. The idea was
to have a generic device tree parsing function since all the drivers,
except pwm-pxa.c, pwm-cros-ec.c and pwm-clps711x.c, uses the same function
to parse DT bindings. And I think, these 3 drivers could use this way of
parsing, which is not something new, is what all the current PWM drivers
uses (except pwm-pxa.c, pwm-cros-ec.c and pwm-clps711x.c). It is true
I have no RK3399 board to run any tests.

pwm-cross-ec.c it is true that it's period cannot be changed. It is fixed, as
I saw in the driver, at EC_PWM_MAX_DUTY=0xffff. The driver itself won't apply
any PWM state if the period is different from EC_PWM_MAX_DUTY.
For this driver, the PWM bindings were changed (I did a grep by "google,cros-ec-pwm"
and located only:
arch/arm64/boot/dts/rockchip/rk3399-gru-kevin.dts
arch/arm64/boot/dts/rockchip/rk3399-gru.dtsi
files) and changed the bindings in this series, as follows, patch 7 from this series:
diff --git a/arch/arm64/boot/dts/rockchip/rk3399-gru-kevin.dts b/arch/arm64/boot/dts/rockchip/rk3399-gru-kevin.dts
index 0384e3121f18..0c790ec387eb 100644
--- a/arch/arm64/boot/dts/rockchip/rk3399-gru-kevin.dts
+++ b/arch/arm64/boot/dts/rockchip/rk3399-gru-kevin.dts
@@ -77,7 +77,7 @@
 
        backlight: backlight {
                compatible = "pwm-backlight";
-               pwms = <&cros_ec_pwm 1>;
+               pwms = <&cros_ec_pwm 1 65535>;
                brightness-levels = <0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
                                     17 18 19 20 21 22 23 24 25 26 27 28 29 30
                                     31 32 33 34 35 36 37 38 39 40 41 42 43 44
diff --git a/arch/arm64/boot/dts/rockchip/rk3399-gru.dtsi b/arch/arm64/boot/dts/rockchip/rk3399-gru.dtsi
index 5772c52fbfd3..aa377f9ae6ad 100644
--- a/arch/arm64/boot/dts/rockchip/rk3399-gru.dtsi
+++ b/arch/arm64/boot/dts/rockchip/rk3399-gru.dtsi
@@ -853,7 +853,7 @@ ap_i2c_audio: &i2c8 {
 
                cros_ec_pwm: ec-pwm {
                        compatible = "google,cros-ec-pwm";
-                       #pwm-cells = <1>;
+                       #pwm-cells = <2>;
                };
        };
};

The code that was removed requests a PWM, the one that was set in the bindings, and
then set pwm->args.period:
-static struct pwm_device *
-cros_ec_pwm_xlate(struct pwm_chip *pc, const struct of_phandle_args *args)
-{
-	struct pwm_device *pwm;
-
-	if (args->args[0] >= pc->npwm)
-		return ERR_PTR(-EINVAL);
-
-	pwm = pwm_request_from_chip(pc, args->args[0], NULL);
-	if (IS_ERR(pwm))
-		return pwm;
-
-	/* The EC won't let us change the period */
-	pwm->args.period = EC_PWM_MAX_DUTY;
-
-	return pwm;
-}

The old flow is as follows:
of_pwm_get() -> cros_ec_pwm_xlate() { request chip and set constant period }

The new flow uses of_pwm_xlate():
of_pwm_get() -> of_pwm_xlate() -> { parse PWM args: channel number, period, flags +
request PWM chip + set pwm->args; }

This path is only used at DT parsing.

In case of PWM channel requested by PWM backlight driver it looks good to me
with the changes in rk3399-gru-kevin.dts (please correct me if I'm wrong).

Since this driver accepts only EC_PWM_MAX_DUTY period maybe the documentation should
be updated regarding this value?

Please, let me know what you think!

Thanks,
Claudiu
> 
>>  drivers/pwm/pwm-fsl-ftm.c      |  2 --
>>  drivers/pwm/pwm-hibvt.c        |  2 --
>>  drivers/pwm/pwm-imx.c          |  8 ------
>>  drivers/pwm/pwm-lpc18xx-sct.c  |  2 --
>>  drivers/pwm/pwm-meson.c        |  2 --
>>  drivers/pwm/pwm-omap-dmtimer.c |  2 --
>>  drivers/pwm/pwm-pxa.c          | 19 --------------
>>  drivers/pwm/pwm-renesas-tpu.c  |  2 --
>>  drivers/pwm/pwm-rockchip.c     |  5 ----
>>  drivers/pwm/pwm-samsung.c      |  3 ---
>>  drivers/pwm/pwm-sun4i.c        |  2 --
>>  drivers/pwm/pwm-tiecap.c       |  2 --
>>  drivers/pwm/pwm-tiehrpwm.c     |  2 --
>>  drivers/pwm/pwm-vt8500.c       |  2 --
>>  drivers/pwm/pwm-zx.c           |  2 --
>>  include/linux/pwm.h            | 23 ++++++++++-------
>>  26 files changed, 29 insertions(+), 156 deletions(-)
>>
> ...
> 
>> diff --git a/drivers/pwm/pwm-cros-ec.c b/drivers/pwm/pwm-cros-ec.c
>> index 9c13694eaa24..692298693768 100644
>> --- a/drivers/pwm/pwm-cros-ec.c
>> +++ b/drivers/pwm/pwm-cros-ec.c
>> @@ -133,24 +133,6 @@ static void cros_ec_pwm_get_state(struct pwm_chip *chip, struct pwm_device *pwm,
>>  	state->duty_cycle = ret;
>>  }
>>  
>> -static struct pwm_device *
>> -cros_ec_pwm_xlate(struct pwm_chip *pc, const struct of_phandle_args *args)
>> -{
>> -	struct pwm_device *pwm;
>> -
>> -	if (args->args[0] >= pc->npwm)
>> -		return ERR_PTR(-EINVAL);
>> -
>> -	pwm = pwm_request_from_chip(pc, args->args[0], NULL);
>> -	if (IS_ERR(pwm))
>> -		return pwm;
>> -
>> -	/* The EC won't let us change the period */
>> -	pwm->args.period = EC_PWM_MAX_DUTY;
>> -
>> -	return pwm;
>> -}
>> -
>>  static const struct pwm_ops cros_ec_pwm_ops = {
>>  	.get_state	= cros_ec_pwm_get_state,
>>  	.apply		= cros_ec_pwm_apply,
>> @@ -207,8 +189,6 @@ static int cros_ec_pwm_probe(struct platform_device *pdev)
>>  	/* PWM chip */
>>  	chip->dev = dev;
>>  	chip->ops = &cros_ec_pwm_ops;
>> -	chip->of_xlate = cros_ec_pwm_xlate;
>> -	chip->of_pwm_n_cells = 1;
>>  	chip->base = -1;
>>  	ret = cros_ec_num_pwms(ec);
>>  	if (ret < 0) {
> 
> ...
> 
> Brian
> 
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html