On Thu, Jul 29, 2021 at 06:23:10PM +0100, Biju Das wrote:

> +config SND_SOC_RZ
> +	tristate "RZ/G2L series SSIF-2 support"
> +	depends on ARCH_R9A07G044 || COMPILE_TEST
> +	select SND_SIMPLE_CARD
> +	help

Why is the DAI selecting a specific sound card, and if it must then why
would it use simple-card and not the more modern audio-graph-card?  The
select should almost certainly just be removed entirely, this is not
something DAI drivers do - cards use DAIs, not the other way around.

> +++ b/sound/soc/sh/rz-ssi.c
> @@ -0,0 +1,871 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Renesas RZ/G2L ASoC Serial Sound Interface (SSIF-2) Driver
> + *
> + * Copyright (C) 2021 Renesas Electronics Corp.

Please make the entire comment a C++ one so things look more
intentional.

> +static int rz_ssi_stream_init(struct rz_ssi_priv *ssi,
> +			      struct rz_ssi_stream *strm,
> +			      struct snd_pcm_substream *substream)
> +{
> +	struct snd_pcm_runtime *runtime = substream->runtime;
> +
> +	strm->substream = substream;

> +	if (runtime->sample_bits != 16) {
> +		dev_err(ssi->dev, "Unsupported sample width: %d\n",
> +			runtime->sample_bits);
> +		strm->substream = NULL;
> +		return -EINVAL;
> +	}
> +
> +	if (runtime->frame_bits != 32) {
> +		dev_err(ssi->dev, "Unsupported frame width: %d\n",
> +			runtime->sample_bits);
> +		strm->substream = NULL;
> +		return -EINVAL;
> +	}
> +
> +	if (runtime->channels != 2) {
> +		dev_err(ssi->dev, "Number of channels not matched\n");
> +		strm->substream = NULL;
> +		return -EINVAL;
> +	}

Why are we only validating these here which is called from...

> +	switch (cmd) {
> +	case SNDRV_PCM_TRIGGER_START:
> +		/* Soft Reset */
> +		rz_ssi_reg_mask_setl(ssi, SSIFCR, 0, SSIFCR_SSIRST);
> +		rz_ssi_reg_mask_setl(ssi, SSIFCR, SSIFCR_SSIRST, 0);
> +		udelay(5);
> +
> +		ret = rz_ssi_stream_init(ssi, strm, substream);
> +		if (ret)
> +			goto done;

...trigger.  This stuff should be being validated when we set
parameters in hw_params() and can usefully report the error.

> +static int rz_ssi_dai_set_fmt(struct snd_soc_dai *dai, unsigned int fmt)
> +{
> +	struct rz_ssi_priv *ssi = snd_soc_dai_get_drvdata(dai);
> +
> +	/* set master/slave audio interface */
> +	switch (fmt & SND_SOC_DAIFMT_MASTER_MASK) {
> +	case SND_SOC_DAIFMT_CBS_CFS:
> +		asm("nop"); /* codec is slave */
> +		break;

That asm("nop") looks interesting...  I can't think why that'd be
needed?  Please also use the modern versions of these defines, consumer
and provider rather than slave and master.