[PATCH v3] Fix IMX PWM period setting

[Uwe Kleine-König]

Hello Olaf,

On Mon, Nov 03, 2025 at 06:50:20AM +0000, Krebs, Olaf wrote:
> From: Olaf Krebs <okr at smgw.emh-meter.de>
> 
> If a second PWM is requested by a driver before the first is configured, trying to configure any of these results in .user_count > 1 and thus the configuration fails.
> Fix that by only erroring out by additionally checking if the period is actually configured.
> 
> Exapmle: Using of 3 PWM channels to control a RGB LED.
> 
> DTS-Config for an imx93-Board:
> 	...
> 	led-controller {
> 		compatible = "pwm-leds-multicolor";
> 		multi-led {
> 			label = "RGBled";
> 			color = <LED_COLOR_ID_RGB>;
> 			function = LED_FUNCTION_INDICATOR;
> 			max-brightness = <255>;
> 			led-red {
> 				pwms = <&tpm5 0 1000000 PWM_POLARITY_INVERTED>;
> 				color = <LED_COLOR_ID_RED>;
> 			};
> 			led-green {
> 				pwms = <&tpm6 2 1000000 PWM_POLARITY_INVERTED>;
> 				color = <LED_COLOR_ID_GREEN>;
> 			};
> 			led-blue {
> 				pwms = <&tpm5 1 1000000 PWM_POLARITY_INVERTED>;
> 				color = <LED_COLOR_ID_BLUE>;
> 			};
> 		};
> 	};
> 	...
> 
> Without this patch, an BUSY-error message is generated during initialization.
> 
> [    7.395326] leds_pwm_multicolor led-controller: error -EBUSY: failed to set led PWM value for (null)
> [    7.405167] leds_pwm_multicolor led-controller: probe with driver leds_pwm_multicolor failed with error -16
> 
> Signed-off-by: Olaf krebs <olaf.krebs at emh-metering.com>

checkpatch is still unhappy with your patch. That might be (at least
partially) the fault of your MUA or MTA.

> ---
>  drivers/pwm/pwm-imx-tpm.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/pwm/pwm-imx-tpm.c b/drivers/pwm/pwm-imx-tpm.c index 5b399de16d60..411daa7711f1 100644
> --- a/drivers/pwm/pwm-imx-tpm.c
> +++ b/drivers/pwm/pwm-imx-tpm.c
> @@ -190,7 +190,7 @@ static int pwm_imx_tpm_apply_hw(struct pwm_chip *chip,
>  		 * there are multiple channels in use with different
>  		 * period settings.
>  		 */
> -		if (tpm->user_count > 1)
> +		if ((tpm->user_count > 1) && (tpm->real_period != 0))
>  			return -EBUSY;

Repeating my review for v2:
> Please drop the added parenthesis.

Have you considered "I wonder if a saner check would involve
enable_count instead of user_count."?

Best regards
Uwe
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20251103/8472594d/attachment.sig>