linux-arm-msm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [v5] ASoC: qcom: lpass-cpu: Fix pop noise during audio capture begin
@ 2021-06-03 15:08 Judy Hsiao
  2021-06-03 22:40 ` Doug Anderson
  0 siblings, 1 reply; 2+ messages in thread
From: Judy Hsiao @ 2021-06-03 15:08 UTC (permalink / raw)
  To: broonie
  Cc: Taniya Das, Rohit kumar, Banajit Goswami, Patrick Lai,
	Andy Gross, Bjorn Andersson, Liam Girdwood, Rob Herring,
	Jaroslav Kysela, Takashi Iwai, Srini Kandagatla, Stephan Gerhold,
	dianders, dgreid, cychiang, judyhsiao, tzungbi, swboyd,
	linux-arm-kernel, linux-arm-msm, devicetree, alsa-devel,
	Srinivasa Rao Mandadapu, Judy Hsiao

From: Srinivasa Rao Mandadapu <srivasam@codeaurora.org>

This patch fixes PoP noise of around 15ms observed during audio
capture begin.
Enables BCLK and LRCLK in snd_soc_dai_ops prepare call for
introducing some delay before capture start.

Co-developed-by: Judy Hsiao <judyhsiao@chromium.org>
Signed-off-by: Judy Hsiao <judyhsiao@chromium.org>
Signed-off-by: Srinivasa Rao Mandadapu <srivasam@codeaurora.org>
(am from https://patchwork.kernel.org/patch/12276369/)
(also found at https://lore.kernel.org/r/20210524142114.18676-1-srivasam@codeaurora.org)

---
Changes Since V4:
        -- Replaced the __clk_is_enabled(BCLK) check by the self maintained 
           mi2s_was_prepared bool state.
        -- Removed unrelated changes.
        -- Refined comments.
Changes Since V3:
        -- Checked BCLK is off before enabling it in lpass_cpu_daiops_prepare as
           lpass_cpu_daiops_prepare can be called multiple times
        -- Checked BCLK is on before disabling it in lpass_cpu_daiops_shutdown to
           fix the WARN. It is because BCLK may not be enabled if
           lpass_cpu_daiops_prepare is not called before lpass_cpu_daiops_shutdown
        -- Added more comments
Changes Since V2:
        -- Updated comments as per linux style
        -- Removed unrelated changes
Changes Since V1:
        -- Enabled BCLK and LRCLK in dai ops prepare API instead of startup API
        -- Added comments

 sound/soc/qcom/lpass-cpu.c | 80 ++++++++++++++++++++++++++++++++++++++
 sound/soc/qcom/lpass.h     |  4 ++
 2 files changed, 84 insertions(+)

diff --git a/sound/soc/qcom/lpass-cpu.c b/sound/soc/qcom/lpass-cpu.c
index af8cb64924a0..3129e7cc25a4 100644
--- a/sound/soc/qcom/lpass-cpu.c
+++ b/sound/soc/qcom/lpass-cpu.c
@@ -93,8 +93,30 @@ static void lpass_cpu_daiops_shutdown(struct snd_pcm_substream *substream,
 		struct snd_soc_dai *dai)
 {
 	struct lpass_data *drvdata = snd_soc_dai_get_drvdata(dai);
+	struct lpaif_i2sctl *i2sctl = drvdata->i2sctl;
+	unsigned int id = dai->driver->id;
 
 	clk_disable_unprepare(drvdata->mi2s_osr_clk[dai->driver->id]);
+	/*
+	 * Ensure LRCLK is disabled even in device node validation.
+	 * Will not impact if disabled in lpass_cpu_daiops_trigger()
+	 * suspend.
+	 */
+	if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK)
+		regmap_fields_write(i2sctl->spken, id, LPAIF_I2SCTL_SPKEN_DISABLE);
+	else
+		regmap_fields_write(i2sctl->micen, id, LPAIF_I2SCTL_MICEN_DISABLE);
+
+	/*
+	 * BCLK may not be enabled if lpass_cpu_daiops_prepare is called before
+	 * lpass_cpu_daiops_shutdown. It's paired with the clk_enable in
+	 * lpass_cpu_daiops_prepare.
+	 */
+	if (drvdata->mi2s_was_prepared[dai->driver->id]) {
+		drvdata->mi2s_was_prepared[dai->driver->id] = false;
+		clk_disable(drvdata->mi2s_bit_clk[dai->driver->id]);
+	}
+
 	clk_unprepare(drvdata->mi2s_bit_clk[dai->driver->id]);
 }
 
@@ -275,6 +297,18 @@ static int lpass_cpu_daiops_trigger(struct snd_pcm_substream *substream,
 	case SNDRV_PCM_TRIGGER_START:
 	case SNDRV_PCM_TRIGGER_RESUME:
 	case SNDRV_PCM_TRIGGER_PAUSE_RELEASE:
+		/*
+		 * Ensure lpass BCLK/LRCLK is enabled during
+		 * device resume as lpass_cpu_daiops_prepare() is not called
+		 * after the device resumes. We don't check mi2s_was_prepared before
+		 * enable/disable BCLK in trigger events because:
+		 *  1. These trigger events are paired, so the BCLK
+		 *     enable_count is balanced.
+		 *  2. the BCLK can be shared (ex: headset and headset mic),
+		 *     we need to increase the enable_count so that we don't
+		 *     turn off the shared BCLK while other devices are using
+		 *     it.
+		 */
 		if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK) {
 			ret = regmap_fields_write(i2sctl->spken, id,
 						 LPAIF_I2SCTL_SPKEN_ENABLE);
@@ -296,6 +330,10 @@ static int lpass_cpu_daiops_trigger(struct snd_pcm_substream *substream,
 	case SNDRV_PCM_TRIGGER_STOP:
 	case SNDRV_PCM_TRIGGER_SUSPEND:
 	case SNDRV_PCM_TRIGGER_PAUSE_PUSH:
+		/*
+		 * To ensure lpass BCLK/LRCLK is disabled during
+		 * device suspend.
+		 */
 		if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK) {
 			ret = regmap_fields_write(i2sctl->spken, id,
 						 LPAIF_I2SCTL_SPKEN_DISABLE);
@@ -315,12 +353,54 @@ static int lpass_cpu_daiops_trigger(struct snd_pcm_substream *substream,
 	return ret;
 }
 
+static int lpass_cpu_daiops_prepare(struct snd_pcm_substream *substream,
+		struct snd_soc_dai *dai)
+{
+	struct lpass_data *drvdata = snd_soc_dai_get_drvdata(dai);
+	struct lpaif_i2sctl *i2sctl = drvdata->i2sctl;
+	unsigned int id = dai->driver->id;
+	int ret;
+	/*
+	 * Ensure lpass BCLK/LRCLK is enabled bit before playback/capture
+	 * data flow starts. This allows other codec to have some delay before
+	 * the data flow.
+	 * (ex: to drop start up pop noise before capture starts).
+	 */
+	if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK)
+		ret = regmap_fields_write(i2sctl->spken, id, LPAIF_I2SCTL_SPKEN_ENABLE);
+	else
+		ret = regmap_fields_write(i2sctl->micen, id, LPAIF_I2SCTL_MICEN_ENABLE);
+
+	if (ret) {
+		dev_err(dai->dev, "error writing to i2sctl reg: %d\n", ret);
+		return ret;
+	}
+
+	/*
+	 * Check mi2s_was_prepared before enabling BCLK as lpass_cpu_daiops_prepare can
+	 * be called multiple times. It's paired with the clk_disable in
+	 * lpass_cpu_daiops_shutdown.
+	 */
+	if (!drvdata->mi2s_was_prepared[dai->driver->id]) {
+		drvdata->mi2s_was_prepared[dai->driver->id] = true;
+		ret = clk_enable(drvdata->mi2s_bit_clk[id]);
+		if (ret) {
+			dev_err(dai->dev, "error in enabling mi2s bit clk: %d\n", ret);
+			clk_disable(drvdata->mi2s_osr_clk[id]);
+			drvdata->mi2s_was_prepared[dai->driver->id] = false;
+			return ret;
+		}
+	}
+	return 0;
+}
+
 const struct snd_soc_dai_ops asoc_qcom_lpass_cpu_dai_ops = {
 	.set_sysclk	= lpass_cpu_daiops_set_sysclk,
 	.startup	= lpass_cpu_daiops_startup,
 	.shutdown	= lpass_cpu_daiops_shutdown,
 	.hw_params	= lpass_cpu_daiops_hw_params,
 	.trigger	= lpass_cpu_daiops_trigger,
+	.prepare	= lpass_cpu_daiops_prepare,
 };
 EXPORT_SYMBOL_GPL(asoc_qcom_lpass_cpu_dai_ops);
 
diff --git a/sound/soc/qcom/lpass.h b/sound/soc/qcom/lpass.h
index 83b2e08ade06..7f72214404ba 100644
--- a/sound/soc/qcom/lpass.h
+++ b/sound/soc/qcom/lpass.h
@@ -67,6 +67,10 @@ struct lpass_data {
 	/* MI2S SD lines to use for playback/capture */
 	unsigned int mi2s_playback_sd_mode[LPASS_MAX_MI2S_PORTS];
 	unsigned int mi2s_capture_sd_mode[LPASS_MAX_MI2S_PORTS];
+
+	/* The state of MI2S prepare dai_ops was called */
+	bool mi2s_was_prepared[LPASS_MAX_MI2S_PORTS];
+
 	int hdmi_port_enable;
 
 	/* low-power audio interface (LPAIF) registers */
-- 
2.31.0


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [v5] ASoC: qcom: lpass-cpu: Fix pop noise during audio capture begin
  2021-06-03 15:08 [v5] ASoC: qcom: lpass-cpu: Fix pop noise during audio capture begin Judy Hsiao
@ 2021-06-03 22:40 ` Doug Anderson
  0 siblings, 0 replies; 2+ messages in thread
From: Doug Anderson @ 2021-06-03 22:40 UTC (permalink / raw)
  To: Judy Hsiao
  Cc: Mark Brown, Taniya Das, Rohit kumar, Banajit Goswami,
	Patrick Lai, Andy Gross, Bjorn Andersson, Liam Girdwood,
	Rob Herring, Jaroslav Kysela, Takashi Iwai, Srini Kandagatla,
	Stephan Gerhold, Dylan Reid, Jimmy Cheng-Yi Chiang, Judy Hsiao,
	Tzung-Bi Shih, Stephen Boyd, Linux ARM, linux-arm-msm,
	open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS,
	ALSA Development Mailing List, Srinivasa Rao Mandadapu

Judy,

On Thu, Jun 3, 2021 at 8:08 AM Judy Hsiao <judyhsiao@chromium.org> wrote:
>
> @@ -315,12 +353,54 @@ static int lpass_cpu_daiops_trigger(struct snd_pcm_substream *substream,
>         return ret;
>  }
>
> +static int lpass_cpu_daiops_prepare(struct snd_pcm_substream *substream,
> +               struct snd_soc_dai *dai)
> +{
> +       struct lpass_data *drvdata = snd_soc_dai_get_drvdata(dai);
> +       struct lpaif_i2sctl *i2sctl = drvdata->i2sctl;
> +       unsigned int id = dai->driver->id;
> +       int ret;
> +       /*
> +        * Ensure lpass BCLK/LRCLK is enabled bit before playback/capture
> +        * data flow starts. This allows other codec to have some delay before
> +        * the data flow.
> +        * (ex: to drop start up pop noise before capture starts).
> +        */

nit: there's usually a blank line between the variable declarations
and the first line of code, even if the first line of code is a
comment.


> +       if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK)
> +               ret = regmap_fields_write(i2sctl->spken, id, LPAIF_I2SCTL_SPKEN_ENABLE);
> +       else
> +               ret = regmap_fields_write(i2sctl->micen, id, LPAIF_I2SCTL_MICEN_ENABLE);
> +
> +       if (ret) {
> +               dev_err(dai->dev, "error writing to i2sctl reg: %d\n", ret);
> +               return ret;
> +       }
> +
> +       /*
> +        * Check mi2s_was_prepared before enabling BCLK as lpass_cpu_daiops_prepare can
> +        * be called multiple times. It's paired with the clk_disable in
> +        * lpass_cpu_daiops_shutdown.
> +        */
> +       if (!drvdata->mi2s_was_prepared[dai->driver->id]) {
> +               drvdata->mi2s_was_prepared[dai->driver->id] = true;
> +               ret = clk_enable(drvdata->mi2s_bit_clk[id]);
> +               if (ret) {
> +                       dev_err(dai->dev, "error in enabling mi2s bit clk: %d\n", ret);
> +                       clk_disable(drvdata->mi2s_osr_clk[id]);

Can you explain why this clk_disable() is here? Your function didn't
turn this clock on, so why should it be turning it off in the error
case?


> +                       drvdata->mi2s_was_prepared[dai->driver->id] = false;
> +                       return ret;
> +               }

Why not put the `drvdata->mi2s_was_prepared[dai->driver->id] = true;`
_after_ you check for errors. Then you don't need to undo it in the
error case. I presume that your prepare() function isn't reentrant and
can't be called at the same time as your shutdown (right?).

Other than that, I don't have any objections to this patch anymore. I
probably won't add a formal "Reviewed-by", though, since I _really_
don't know anything about the issue at hand or the code. I just
stumbled upon this because I was getting the clock splat at bootup. If
someone feels like this needs me to spin up enough to understand /
really review this patch then please yell.

-Doug

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2021-06-03 22:42 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-06-03 15:08 [v5] ASoC: qcom: lpass-cpu: Fix pop noise during audio capture begin Judy Hsiao
2021-06-03 22:40 ` Doug Anderson

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).