Re: [PATCH v4 1/1] pwm: pca9685: fix gpio-only operation.

From: Andy Shevchenko
Date: Tue Apr 18 2017 - 05:17:07 EST



+Cc: Rafael (one question to you below)

On Thu, 2017-04-13 at 08:58 -0400, Sven Van Asbroeck wrote:
> gpio-only driver operation never clears the SLEEP bit, which can
> cause the gpios to become unusable.
>
> Example:
> 1. user requests first pwmÂÂ->ÂÂÂÂÂÂdriver clears SLEEP bit
> 2. user frees last pwmÂÂÂÂÂÂ->ÂÂÂÂÂÂdriver sets SLEEP bit
> 3. user requests gpio
> 4. user switches gpio onÂÂÂÂ->ÂÂÂÂÂÂoutput does not turn on
> ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂbecause 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/in use.
>

I know the patch is applied already, though please consider below to be
addressed as usual (w/o Fixes tag).

> +static void pca9685_set_sleep_mode(struct pca9685 *pca, int sleep)
> +{
> + regmap_update_bits(pca->regmap, PCA9685_MODE1,
> + ÂÂÂMODE1_SLEEP, sleep ? MODE1_SLEEP : 0);

> + if (!sleep) {
> + /* Wait 500us for the oscillator to be back up */
> + udelay(500);
> + }

I would go with

/* Wait for @sleep microseconds for the oscillator to be back up */
if (sleep)
udelay(sleep);

Otherwise int sleep is oddly here.

Or

bool sleep

/* Wait 500us ... */
if (sleep)
udelay(500);

> +}


> +#ifdef CONFIG_PM
> +static int pca9685_pwm_runtime_suspend(struct device *dev)

__maybe_unused and remove ugly #ifdef:ery.

> +{
> + struct i2c_client *client = to_i2c_client(dev);
> + struct pca9685 *pca = i2c_get_clientdata(client);
> +
> + pca9685_set_sleep_mode(pca, 1);
> + return 0;
> Â}
> Â
> +static int pca9685_pwm_runtime_resume(struct device *dev)

Ditto.

> +{
> + struct i2c_client *client = to_i2c_client(dev);
> + struct pca9685 *pca = i2c_get_clientdata(client);
> +
> + pca9685_set_sleep_mode(pca, 0);
> + return 0;
> +}
> +#endif

> +static const struct dev_pm_ops pca9685_pwm_pm = {
> + SET_RUNTIME_PM_OPS(pca9685_pwm_runtime_suspend,
> + ÂÂÂpca9685_pwm_runtime_resume, NULL)
> +};
> +

Perhaps we may introduce RUNTIME_DEV_PM_OPS() macro and re-use it here.
Rafael?

--
Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>
Intel Finland Oy