Re: [PATCH v1 3/3] pinctrl: qcom: lpass-lpi: Resume clocks for GPIO access

[Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>]

On 4/13/26 2:22 PM, Ajay Kumar Nandam wrote:
> Ensure the LPI pinctrl device clocks are runtime resumed
> before accessing GPIO registers and autosuspended after
> the access completes.
> 
> Guard GPIO register read and write helpers with synchronous
> runtime PM calls so the device is active during MMIO
> operations.
> 
> Signed-off-by: Ajay Kumar Nandam <ajay.nandam@oss.qualcomm.com>
> ---
>  drivers/pinctrl/qcom/pinctrl-lpass-lpi.c | 18 ++++++++++++++++++
>  1 file changed, 18 insertions(+)
> 
> diff --git a/drivers/pinctrl/qcom/pinctrl-lpass-lpi.c b/drivers/pinctrl/qcom/pinctrl-lpass-lpi.c
> index d108e7321..4275f2734 100644
> --- a/drivers/pinctrl/qcom/pinctrl-lpass-lpi.c
> +++ b/drivers/pinctrl/qcom/pinctrl-lpass-lpi.c
> @@ -49,8 +49,17 @@ static int lpi_gpio_read(struct lpi_pinctrl *state, unsigned int pin,
>  	else
>  		pin_offset = LPI_TLMM_REG_OFFSET * pin;
>  
> +	ret = pm_runtime_get_sync(state->dev);
> +	if (ret < 0) {
> +		pm_runtime_put_noidle(state->dev);
> +		return ret;
> +	}

Okay that's how they can fail..

Please move from pm_runtime_get_sync() to pm_runtime_resume_and_get() or
someone will come around next week to "improve" it

> +
>  	*val = ioread32(state->tlmm_base + pin_offset + addr);
>  
> +	pm_runtime_mark_last_busy(state->dev);
> +	pm_runtime_put_autosuspend(state->dev);

put_autosuspend() already does mark_last_busy() nowadays

Konrad