Re: [PATCH v2 1/3] media: qcom: camss: Add support for TPG common

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

On 7/17/25 5:20 AM, Wenmeng Liu wrote:
> Add support for TPG common, unlike CSID TPG, this TPG can
> be seen as a combination of CSIPHY and sensor.
> 
> Signed-off-by: Wenmeng Liu <quic_wenmliu@quicinc.com>
> ---

[...]

> +++ b/drivers/media/platform/qcom/camss/camss-tpg.c
> @@ -0,0 +1,737 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * camss-tpg.c

I think the filename is redundant, especially since it may change in
the future

[...]

> +const struct tpg_format_info *tpg_get_fmt_entry(const struct tpg_format_info *formats,
> +						unsigned int nformats,
> +						u32 code)
> +{
> +	unsigned int i;

https://staticthinking.wordpress.com/2022/06/01/unsigned-int-i-is-stupid/

> +
> +	for (i = 0; i < nformats; i++)
> +		if (code == formats[i].code)
> +			return &formats[i];
> +
> +	WARN(1, "Unknown format\n");
> +
> +	return &formats[0];

Err.. that doesn't seem right, neither WARN (which usually signifies
some sort of a critical condition or hw failure), nor returning a format
different to the one the user requested

We should probably return some kind of -EOPNOTSUPP

> +}
> +
> +/*
> + * tpg_set_clock_rates - Calculate and set clock rates on tpg module
> + * @tpg: tpg device
> + */
> +static int tpg_set_clock_rates(struct tpg_device *tpg)
> +{
> +	struct device *dev = tpg->camss->dev;
> +	int i, j;
> +	int ret;
> +
> +	for (i = 0; i < tpg->nclocks; i++) {
> +		struct camss_clock *clock = &tpg->clock[i];
> +		u64 min_rate = 0;
> +		long round_rate;
> +
> +		camss_add_clock_margin(&min_rate);
> +
> +		for (j = 0; j < clock->nfreqs; j++)
> +			if (min_rate < clock->freq[j])
> +				break;
> +
> +		if (j == clock->nfreqs) {
> +			dev_err(dev,
> +				"clock is too high for TPG\n");

I really insist you don't have to break this line

It would probably be useful to print the rates (the one that's too
high and the maximum)

> +			return -EINVAL;
> +		}
> +
> +		/* if clock is not available */
> +		/* set highest possible tpg clock rate */
> +		if (min_rate == 0)
> +			j = clock->nfreqs - 1;

Well, you never assign anything nonzero to min_rate..

[...]

> +static void tpg_try_format(struct tpg_device *tpg,
> +			   struct v4l2_subdev_state *sd_state,
> +			   unsigned int pad,
> +			   struct v4l2_mbus_framefmt *fmt,
> +			   enum v4l2_subdev_format_whence which)
> +{
> +	unsigned int i;
> +
> +	switch (pad) {
> +	case MSM_TPG_PAD_SINK:
> +		/* Test generator is enabled, set format on source */
> +		/* pad to allow test generator usage */

This is a very strange way to write multiline comments

[...]

> +	/* Memory */
> +	tpg->base = devm_platform_ioremap_resource_byname(pdev, res->reg[0]);
> +	if (IS_ERR(tpg->base))
> +		return PTR_ERR(tpg->base);
> +
> +	/* Interrupt */
> +	ret = platform_get_irq_byname(pdev, res->interrupt[0]);
> +	if (ret < 0)
> +		return ret;

The comments are unnecessary

Konrad