Re: [PATCH 09/10] ASoC: qcom: add sdm660 internal sound card support

[Richard Acayan <mailingradian@gmail.com>]

On Mon, Feb 16, 2026 at 01:34:20PM +0000, Srinivas Kandagatla wrote:
> On 2/11/26 2:03 AM, Richard Acayan wrote:
(snip)
> > +	case INT0_MI2S_RX:
> > +		data->int0_mi2s_clk_count++;
> > +		if (data->int0_mi2s_clk_count == 1)
> > +			snd_soc_dai_set_sysclk(cpu,
> > +				Q6AFE_LPASS_CLK_ID_INT0_MI2S_IBIT,
> > +				MI2S_BCLK_RATE, SNDRV_PCM_STREAM_PLAYBACK);
> > +
> > +		/*
> > +		 * Downstream specifies that the AFE is a clock consumer, but
> > +		 * the sound is distorted (loud on the right channel and sped
> > +		 * up) unless we set it as a producer.
> This comment does not make any value, Who is driving the bit clk and
> frame clock is totally depended on the hw setup on the platform, Driving
> incorrect clock would lead to such issues. Does the codec drive on the
> platform that you are testing?

I dug into the downstream Q6AFE driver a bit more. It actually inverts
the logic and specifies that the codec is a clock consumer (instead of
taking Q6AFE's perspective), so it is equivalent to setting
SND_SOC_DAIFMT_CBP_CFP.

Downstream (https://android.googlesource.com/kernel/msm-extra/+/530cffa4cc977a348753831b163eb9d3302b954a/asoc/msm-dai-q6-v2.c#4347):

	case SND_SOC_DAIFMT_CBS_CFS:
		mi2s_dai_data->rx_dai.mi2s_dai_data.port_config.i2s.ws_src = 1;
		mi2s_dai_data->tx_dai.mi2s_dai_data.port_config.i2s.ws_src = 1;
		break;

Mainline (sound/soc/qcom/q6afe.c):
	
	#define AFE_PORT_CONFIG_I2S_WS_SRC_INTERNAL	0x1
	...
	case SND_SOC_DAIFMT_BP_FP:
		pcfg->i2s_cfg.ws_src = AFE_PORT_CONFIG_I2S_WS_SRC_INTERNAL;
		break;

I don't think the downstream code needs to be explained so I would
simply drop these code comments.

> > +		 */
> > +		snd_soc_dai_set_fmt(cpu, SND_SOC_DAIFMT_CBP_CFP);