Re: [PATCH v5 2/2] backlight: ktz8866: Add support for Kinetic KTZ8866 backlight

[Daniel Thompson]

On Tue, Jan 17, 2023 at 11:44:08PM +0800, Jianhua Lu wrote:
> Add support for Kinetic KTZ8866 backlight, which is used in
> Xiaomi tablet, Mi Pad 5 series. This driver lightly based on
> downstream implementation [1].
> [1] https://github.com/MiCode/Xiaomi_Kernel_OpenSource/blob/elish-r-oss/drivers/video/backlight/ktz8866.c
>
> Signed-off-by: Jianhua Lu <lujianhua000@xxxxxxxxx>
> ---
> Changes in v2:
>   - Add missing staitc modifier to ktz8866_write function.
>
> Changes in v3:
>   - Add 2022 to Copyright line.
>   - Sort headers.
>   - Remove meaningless comment.
>   - Use definitions instead of hardcoding.
>   - Add missing maintainer info.
>
> Changes in v4:
>   - Change 2022 to 2023.
>   - Remove useless macro and enum.
>   - Describe settings by devicetree.
>   - Move header file to C file.
> Changes in v5:
>   - Change "2023" to "2022, 2023" in Copyright line.
>   - Set scale property for backlight.
>
>  MAINTAINERS                       |   6 +
>  drivers/video/backlight/Kconfig   |   8 ++
>  drivers/video/backlight/Makefile  |   1 +
>  drivers/video/backlight/ktz8866.c | 201 ++++++++++++++++++++++++++++++
>  4 files changed, 216 insertions(+)
>  create mode 100644 drivers/video/backlight/ktz8866.c
>
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 42fc47c6edfd..2084e74e1b58 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -11674,6 +11674,12 @@ M:	John Hawley <warthog9@xxxxxxxxxxxxxx>
>  S:	Maintained
>  F:	tools/testing/ktest
>
> +KTZ8866 BACKLIGHT DRIVER
> +M:	Jianhua Lu <lujianhua000@xxxxxxxxx>
> +S:	Maintained
> +F:	Documentation/devicetree/bindings/leds/backlight/kinetic,ktz8866.yaml
> +F:	drivers/video/backlight/ktz8866.c
> +
>  L3MDEV
>  M:	David Ahern <dsahern@xxxxxxxxxx>
>  L:	netdev@xxxxxxxxxxxxxxx
> diff --git a/drivers/video/backlight/Kconfig b/drivers/video/backlight/Kconfig
> index 936ba1e4d35e..2845fd7e33ad 100644
> --- a/drivers/video/backlight/Kconfig
> +++ b/drivers/video/backlight/Kconfig
> @@ -190,6 +190,14 @@ config BACKLIGHT_KTD253
>  	  which is a 1-wire GPIO-controlled backlight found in some mobile
>  	  phones.
>
> +config BACKLIGHT_KTZ8866
> +	tristate "Backlight Driver for Kinetic KTZ8866"
> +	depends on I2C
> +	select REGMAP_I2C
> +	help
> +		Say Y to enabled the backlight driver for the Kinetic KTZ8866
> +		found in Xiaomi Mi Pad 5 series.

s/enabled/enable/ (and sorry for spotting this one so late)


> +++ b/drivers/video/backlight/ktz8866.c
> @@ -0,0 +1,201 @@
> [...]
> +static void ktz8866_init(struct ktz8866 *ktz)
> +{
> +	unsigned int val;
> +
> +	if(of_property_read_u32(ktz->client->dev.of_node, "current-num-sinks", &val))
> +		ktz8866_write(ktz, BL_EN, BIT(val) - 1);
> +	else
> +		/* Enable all 6 current sinks if the number of current sinks isn't specifed. */
> +		ktz8866_write(ktz, BL_EN, BIT(6) - 1);
> +
> +	if(of_property_read_u32(ktz->client->dev.of_node, "current-ramping-time-us", &val)) {
> +		if(val <= 128) {
> +			ktz8866_write(ktz, BL_CFG2, BIT(7) | (ilog2(val) << 3) | PWM_HYST);
> +		} else {
> +			ktz8866_write(ktz, BL_CFG2, BIT(7) | ((5 + val / 64) << 3) | PWM_HYST);
> +		}

This code is interpreting current-ramping-time-us as milliseconds rather
than microseconds!

I know I used microseconds in the example I proposed in the feedback for
v4 DT bindings but "something like" means I am merely providing an
example (mostly I was intending to show that the units should be
included both in the property name *and* description).

It is up to you whether you fix the mismatch by changing the DT bindings
document to current-ramping-time-ms or change this code to accept
values in microseconds.


> +	}
> +
> +	if(of_property_read_u32(ktz->client->dev.of_node, "led-ramping-time-us", &val)) {
> +		unsigned int ramp_off_time = ilog2(val) + 1;
> +		unsigned int ramp_on_time = ramp_off_time << 4;
> +		ktz8866_write(ktz, BL_DIMMING, ramp_on_time | ramp_off_time);
> +	}

Similarly, this code has not adopted the units specified in the
bindings documentation.

In this case 0 would map to 512us so if you decided to use milliseconds
you will need to add comment in the description saying that 0 will map
to 512us because the hardware cannot ramp faster than this!


Daniel.