ASoC: cs42l42: Assume 24-bit samples are in 32-bit slots

If the machine driver doesn't call snd_soc_dai_set_sysclk() the
SCLK is assumed to be sample_rate * sample_bits * 2 (that is, the
rate necessary for a standard I2S frame).

But 24-bit samples can be sent in either a 24-bit slot or a 32-bit
slot. If the PLL is configured for a 24-bit slot, but a 32-bit slot is
used, cs42l42 will be overclocked.

Ultimately it is the machine driver's responsibilty to call
snd_soc_dai_set_sysclk() if SLK will be different from the standard
I2S rate. However, it is convenient to assume 32-bit slots to allow
this common case without needing special machine driver support. The
machine driver then only has to set SCLK if the slots are 24-bit, but
if it fails to do this cs42l42 won't be overclocked.
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20210805161111.10410-7-rf@opensource.cirrus.comSigned-off-by: Mark Brown <broonie@kernel.org>

sound/soc/codecs/cs42l42.c

@@ -844,6 +844,13 @@ static int cs42l42_pcm_hw_params(struct snd_pcm_substream *substream,
 	if (channels == 1)
 		cs42l42->bclk *= 2;
 
+	/*
+	 * Assume 24-bit samples are in 32-bit slots, to prevent SCLK being
+	 * more than assumed (which would result in overclocking).
+	 */
+	if (params_width(params) == 24)
+		cs42l42->bclk = (cs42l42->bclk / 3) * 4;
+
 	switch(substream->stream) {
 	case SNDRV_PCM_STREAM_CAPTURE:
 		if (channels == 2) {