[PATCH v3] pwm: renesas-tpu: Add suspend/resume function

[PATCH v3] pwm: renesas-tpu: Add suspend/resume function

[Cao Van Dong]

This patch adds suspend/resume function support for Renesas the 16-Bit Timer
Pulse Unit (TPU) driver. This has been tested on the Salvator-XS board 
with R-Car M3-N and H3 at renesas-drivers-2019-05-21-v5.2-rc1 tag.
I expect this to work on other SoCs.

Test procedure:
  - Enable TPU and pin control in DTS.
  - Make sure switches { SW29-[1-2] are switched off or 
    SW31-[1-4] are switched off(only for Salvator-xs) }.
  - Exercise userspace PWM control for pwm[2,3] 
    of /sys/class/pwm/pwmchip1/ .
  - Inspect PWM signals on the input side of { CN29-[58,60] 
    or SW31-[1,2] (only for Salvator-xs) }
    before and after suspend/resume using an oscilloscope. 

Signed-off-by: Cao Van Dong <cv-dong@jinso.co.jp>
Tested-by: Cao Van Dong <cv-dong@jinso.co.jp>
Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
---
Changes v2 -> v3:
  - Changes '3' -> TPU_CHANNEL_MAX in loop.
  - Remove pm_runtime_put() function in tpu_pwm_suspend() function.
---
Changes v1 -> v2:
  - Repair the handling code to cover case of using multiple timers.
---
 drivers/pwm/pwm-renesas-tpu.c | 53 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 53 insertions(+)

diff --git a/drivers/pwm/pwm-renesas-tpu.c b/drivers/pwm/pwm-renesas-tpu.c
index 4a855a2..cb97252 100644
--- a/drivers/pwm/pwm-renesas-tpu.c
+++ b/drivers/pwm/pwm-renesas-tpu.c
@@ -366,6 +366,58 @@ static void tpu_pwm_disable(struct pwm_chip *chip, struct pwm_device *_pwm)
 	tpu_pwm_timer_stop(pwm);
 }
 
+#ifdef CONFIG_PM_SLEEP
+static int tpu_pwm_restart_timer(struct pwm_device *pwm)
+{
+	if (!test_bit(PWMF_REQUESTED, &pwm->flags))
+		return 0;
+
+	/* Restart timer */
+	tpu_pwm_disable(pwm->chip,pwm);
+	tpu_pwm_enable(pwm->chip,pwm);
+
+	return 0;
+}
+
+static int tpu_pwm_suspend(struct device *dev)
+{
+	struct tpu_device *tpu = dev_get_drvdata(dev);
+	struct pwm_chip *chip = &tpu->chip;
+	struct pwm_device *pwm;
+	int i;
+
+	for (i = 0; i < TPU_CHANNEL_MAX; i++) {
+		if ((pwm_get_chip_data(&chip->pwms[i])) != NULL) {
+			pwm = &chip->pwms[i];
+			if (!test_bit(PWMF_REQUESTED, &pwm->flags))
+				return 0;
+		}
+	}
+
+	return 0;
+}
+
+static int tpu_pwm_resume(struct device *dev)
+{
+	struct tpu_device *tpu = dev_get_drvdata(dev);
+	struct pwm_chip *chip = &tpu->chip;
+	struct pwm_device *pwm;
+	int i;
+
+	pm_runtime_get_sync(dev);
+
+	for (i = 0; i < TPU_CHANNEL_MAX; i++) {
+		if ((pwm_get_chip_data(&chip->pwms[i])) != NULL) {
+			pwm = &chip->pwms[i];
+			tpu_pwm_restart_timer(pwm);
+		}
+	}
+
+	return 0;
+}
+#endif /* CONFIG_PM_SLEEP */
+static SIMPLE_DEV_PM_OPS(tpu_pwm_pm_ops, tpu_pwm_suspend, tpu_pwm_resume);
+
 static const struct pwm_ops tpu_pwm_ops = {
 	.request = tpu_pwm_request,
 	.free = tpu_pwm_free,
@@ -459,6 +511,7 @@ static struct platform_driver tpu_driver = {
 	.remove		= tpu_remove,
 	.driver		= {
 		.name	= "renesas-tpu-pwm",
+		.pm	= &tpu_pwm_pm_ops,
 		.of_match_table = of_match_ptr(tpu_of_table),
 	}
 };
-- 
2.7.4

Re: [PATCH v3] pwm: renesas-tpu: Add suspend/resume function

[Kuninori Morimoto]

Hi Cao

If you post v3 patches, you should post whole series

> This patch adds suspend/resume function support for Renesas the 16-Bit Timer
> Pulse Unit (TPU) driver. This has been tested on the Salvator-XS board 
> with R-Car M3-N and H3 at renesas-drivers-2019-05-21-v5.2-rc1 tag.
> I expect this to work on other SoCs.
> 
> Test procedure:
>   - Enable TPU and pin control in DTS.
>   - Make sure switches { SW29-[1-2] are switched off or 
>     SW31-[1-4] are switched off(only for Salvator-xs) }.
>   - Exercise userspace PWM control for pwm[2,3] 
>     of /sys/class/pwm/pwmchip1/ .
>   - Inspect PWM signals on the input side of { CN29-[58,60] 
>     or SW31-[1,2] (only for Salvator-xs) }
>     before and after suspend/resume using an oscilloscope. 
> 
> Signed-off-by: Cao Van Dong <cv-dong@jinso.co.jp>
> Tested-by: Cao Van Dong <cv-dong@jinso.co.jp>
> Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>

Drop my name from here, I'm not Acked yet.

> +#ifdef CONFIG_PM_SLEEP
> +static int tpu_pwm_restart_timer(struct pwm_device *pwm)
> +{
> +	if (!test_bit(PWMF_REQUESTED, &pwm->flags))
> +		return 0;
> +
> +	/* Restart timer */
> +	tpu_pwm_disable(pwm->chip,pwm);
> +	tpu_pwm_enable(pwm->chip,pwm);
> +
> +	return 0;
> +}
> +
> +static int tpu_pwm_suspend(struct device *dev)
> +{
> +	struct tpu_device *tpu = dev_get_drvdata(dev);
> +	struct pwm_chip *chip = &tpu->chip;
> +	struct pwm_device *pwm;
> +	int i;
> +
> +	for (i = 0; i < TPU_CHANNEL_MAX; i++) {
> +		if ((pwm_get_chip_data(&chip->pwms[i])) != NULL) {
> +			pwm = &chip->pwms[i];
> +			if (!test_bit(PWMF_REQUESTED, &pwm->flags))
> +				return 0;
> +		}
> +	}
> +
> +	return 0;
> +}
> +
> +static int tpu_pwm_resume(struct device *dev)
> +{
> +	struct tpu_device *tpu = dev_get_drvdata(dev);
> +	struct pwm_chip *chip = &tpu->chip;
> +	struct pwm_device *pwm;
> +	int i;
> +
> +	pm_runtime_get_sync(dev);
> +
> +	for (i = 0; i < TPU_CHANNEL_MAX; i++) {
> +		if ((pwm_get_chip_data(&chip->pwms[i])) != NULL) {
> +			pwm = &chip->pwms[i];
> +			tpu_pwm_restart_timer(pwm);
> +		}
> +	}
> +
> +	return 0;
> +}
> +#endif /* CONFIG_PM_SLEEP */

again

	pwm = &chip->pwms[i];
	if (pwm_get_chip_data(pwms)) {

Why you "restart" at resume ?
I think "disable" at suspend, and "enable" at resume is generic ?

You call pm_runtime_get_sync() at resume,
but removed pm_runtime_put() suspend ??

Thank you for your help !!
Best regards
---
Kuninori Morimoto

Re: [PATCH v3] pwm: renesas-tpu: Add suspend/resume function

[Cao Van Dong]

Dear Morimoto-san,

Thank for your feedback!

On 2019/05/24 10:21, Kuninori Morimoto wrote:
> Hi Cao
>
> If you post v3 patches, you should post whole series
>
>> This patch adds suspend/resume function support for Renesas the 16-Bit Timer
>> Pulse Unit (TPU) driver. This has been tested on the Salvator-XS board
>> with R-Car M3-N and H3 at renesas-drivers-2019-05-21-v5.2-rc1 tag.
>> I expect this to work on other SoCs.
>>
>> Test procedure:
>>    - Enable TPU and pin control in DTS.
>>    - Make sure switches { SW29-[1-2] are switched off or
>>      SW31-[1-4] are switched off(only for Salvator-xs) }.
>>    - Exercise userspace PWM control for pwm[2,3]
>>      of /sys/class/pwm/pwmchip1/ .
>>    - Inspect PWM signals on the input side of { CN29-[58,60]
>>      or SW31-[1,2] (only for Salvator-xs) }
>>      before and after suspend/resume using an oscilloscope.
>>
>> Signed-off-by: Cao Van Dong <cv-dong@jinso.co.jp>
>> Tested-by: Cao Van Dong <cv-dong@jinso.co.jp>
>> Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
> Drop my name from here, I'm not Acked yet.

I'm so sorry !!

>> +#ifdef CONFIG_PM_SLEEP
>> +static int tpu_pwm_restart_timer(struct pwm_device *pwm)
>> +{
>> +	if (!test_bit(PWMF_REQUESTED, &pwm->flags))
>> +		return 0;
>> +
>> +	/* Restart timer */
>> +	tpu_pwm_disable(pwm->chip,pwm);
>> +	tpu_pwm_enable(pwm->chip,pwm);
>> +
>> +	return 0;
>> +}
>> +
>> +static int tpu_pwm_suspend(struct device *dev)
>> +{
>> +	struct tpu_device *tpu = dev_get_drvdata(dev);
>> +	struct pwm_chip *chip = &tpu->chip;
>> +	struct pwm_device *pwm;
>> +	int i;
>> +
>> +	for (i = 0; i < TPU_CHANNEL_MAX; i++) {
>> +		if ((pwm_get_chip_data(&chip->pwms[i])) != NULL) {
>> +			pwm = &chip->pwms[i];
>> +			if (!test_bit(PWMF_REQUESTED, &pwm->flags))
>> +				return 0;
>> +		}
>> +	}
>> +
>> +	return 0;
>> +}
>> +
>> +static int tpu_pwm_resume(struct device *dev)
>> +{
>> +	struct tpu_device *tpu = dev_get_drvdata(dev);
>> +	struct pwm_chip *chip = &tpu->chip;
>> +	struct pwm_device *pwm;
>> +	int i;
>> +
>> +	pm_runtime_get_sync(dev);
>> +
>> +	for (i = 0; i < TPU_CHANNEL_MAX; i++) {
>> +		if ((pwm_get_chip_data(&chip->pwms[i])) != NULL) {
>> +			pwm = &chip->pwms[i];
>> +			tpu_pwm_restart_timer(pwm);
>> +		}
>> +	}
>> +
>> +	return 0;
>> +}
>> +#endif /* CONFIG_PM_SLEEP */
> again
>
> 	pwm = &chip->pwms[i];
> 	if (pwm_get_chip_data(pwms)) {
>
> Why you "restart" at resume ?
> I think "disable" at suspend, and "enable" at resume is generic ?

This is a good idea.

> You call pm_runtime_get_sync() at resume,
> but removed pm_runtime_put() suspend ??

Very sorry for the inconvenience. I'm mistake. I will remove 
pm_runtime_get_sync() at resume.

Thank you,
Dong

> Thank you for your help !!
> Best regards
> ---
> Kuninori Morimoto