Re: [PATCH v2 3/6] ASoC: sdw_utils: add soc_sdw_es9356

[Pierre-Louis Bossart <pierre-louis.bossart@linux.dev>]

On 2/25/26 23:35, Zhang Yi wrote:
> Add a utility program for handling ES9356 in the universal machine driver
> 
> Signed-off-by: Zhang Yi <zhangyi@everest-semi.com>
> ---
>  sound/soc/sdw_utils/Makefile         |   1 +
>  sound/soc/sdw_utils/soc_sdw_es9356.c | 275 +++++++++++++++++++++++++++
>  2 files changed, 276 insertions(+)
>  create mode 100644 sound/soc/sdw_utils/soc_sdw_es9356.c
> 
> diff --git a/sound/soc/sdw_utils/Makefile b/sound/soc/sdw_utils/Makefile
> index e8bd5ffb1..f47b0cada 100644
> --- a/sound/soc/sdw_utils/Makefile
> +++ b/sound/soc/sdw_utils/Makefile
> @@ -7,6 +7,7 @@ snd-soc-sdw-utils-y := soc_sdw_utils.o soc_sdw_dmic.o soc_sdw_rt_dmic.o \
>  		       soc_sdw_cs42l42.o soc_sdw_cs42l43.o 		\
>  		       soc_sdw_cs42l45.o				\
>  		       soc_sdw_cs_amp.o					\
> +		       soc_sdw_es9356.o					\
>  		       soc_sdw_maxim.o \
>  		       soc_sdw_ti_amp.o
>  obj-$(CONFIG_SND_SOC_SDW_UTILS) += snd-soc-sdw-utils.o
> diff --git a/sound/soc/sdw_utils/soc_sdw_es9356.c b/sound/soc/sdw_utils/soc_sdw_es9356.c
> new file mode 100644
> index 000000000..d6649f0e9
> --- /dev/null
> +++ b/sound/soc/sdw_utils/soc_sdw_es9356.c
> @@ -0,0 +1,275 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +// Copyright (c) 2020 Intel Corporation

this doesn't seem right, and I mentioned it in the previous review...

> +int asoc_sdw_es9356_amp_init(struct snd_soc_card *card,
> +		       struct snd_soc_dai_link *dai_links,
> +		       struct asoc_sdw_codec_info *info,
> +		       bool playback)
> +{
> +	struct asoc_sdw_mc_private *ctx = snd_soc_card_get_drvdata(card);
> +	struct device *sdw_dev1, *sdw_dev2;
> +	int ret;
> +
> +	if (!playback)
> +		return 0;
> +
> +	info->amp_num++;
> +
> +	if (info->amp_num == 2) {
> +		sdw_dev1 = bus_find_device_by_name(&sdw_bus_type, NULL, dai_links->codecs[0].name);
> +
> +		if (!sdw_dev1)
> +			return -EPROBE_DEFER;
> +
> +		ret = es9356_add_codec_device_props(sdw_dev1, ctx->mc_quirk);
> +		if (ret < 0) {
> +			put_device(sdw_dev1);
> +			return ret;
> +		}
> +		ctx->amp_dev1 = sdw_dev1;
> +
> +		sdw_dev2 = bus_find_device_by_name(&sdw_bus_type, NULL, dai_links->codecs[1].name);
> +
> +		if (!sdw_dev2)
> +			return -EPROBE_DEFER;

that doesn't seem right, since you've already added stuff for sdw_dev1.

> +
> +		ret = es9356_add_codec_device_props(sdw_dev2, ctx->mc_quirk);
> +		if (ret < 0) {
> +			put_device(sdw_dev2);

this error handling seems weird, shouldn't you also do a put_device(sdw_dev1) here?

maybe the right thing is to check that both sdw_dev1 and sdw_dev2 can be found.
Then you can add properties for both.

I didn't find any other example of code where we deal with two sets of properties.


> +			return ret;
> +		}
> +		ctx->amp_dev2 = sdw_dev2;
> +	}
> +	return 0;
> +}
> +EXPORT_SYMBOL_NS(asoc_sdw_es9356_amp_init, "SND_SOC_SDW_UTILS");