[PATCH v3 0/8] Speakers for Pixel 3 / 3 XL

[PATCH v3 0/8] Speakers for Pixel 3 / 3 XL

[David Heidelberg via B4 Relay]

This series address:

0. demistify the hex to BIT() macro and add left/right rx spk
1. most likely wrongly set DSP_B in sdm845.c
   my guess this is because TDM was never used in the mainline with
   sdm845
2. applying correct mask to left/right speaker for the TDM in sdm845.c,
   so both speakers can work simultaneusly
3. setting sysclk at startups, which some codecs need (such as cs35l36)
4. setting proper tdm slot in the cs35l36 driver
5. device-tree wiring for Pixel 3 / 3 XL

As per previous discussion not seding patch removing the "Input Mux"
which now interferes with TDM callback. If wanted, I can send it as a
follow-up.

Signed-off-by: David Heidelberg <david@ixit.cz>
---
Changes in v3:
- De-RFC, applied Reviewed-by tag.
- Link to v2: https://patch.msgid.link/20260903-pixel3-audio-v2-0-7c22ed915291@ixit.cz

Changes in v2:
- Device-tree name sorting, with exception keeping mm1-dai-link in it's
  place. (Konrad)
- Make set_tdm_slot cleaner and less leveled. (Konrad)
- Added TDM MIC defines in sdm845.c.
  Should we in capture use speaker mask?
- Renamed model from 'Google Pixel 3' to 'blueline', I would love to
  hear feedback on this one.
- Link to v1: https://lore.kernel.org/r/20260705-pixel3-audio-v1-0-3b66f33859f1@ixit.cz

---
David Heidelberg (8):
      ASoC: qcom: sdm845: Demystify TDM masks a bit
      ASoC: qcom: sdm845: use DSP_A format for TDM codec DAIs
      ASoC: qcom: sdm845: Use per-speaker RX masks for TDM slot assignment
      ASoC: qcom: sdm845: Set codec dai and component sysclk during startup
      ASoC: cs35l36: Implement set_tdm_slot to program RX and TX slots
      arm64: dts: qcom: sdm845-google: Add basic audio support
      slimbus: qcom-ngd-ctrl: Remove data channels on stream disable
      arm64: dts: qcom: sdm845-google: Add WCD9340 codec and microphone capture

 arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi | 195 ++++++++++++++++++++-
 drivers/slimbus/qcom-ngd-ctrl.c                    |  81 +++++++++
 sound/soc/codecs/cs35l36.c                         |  44 +++++
 sound/soc/qcom/sdm845.c                            |  49 ++++--
 4 files changed, 357 insertions(+), 12 deletions(-)
---
base-commit: 7079a12d7506b07fb53b54a664bfad5fa9b16d70
change-id: 20260613-pixel3-audio-bdbfb49e8037

Best regards,
--  
David Heidelberg <david@ixit.cz>

[PATCH v3 1/8] ASoC: qcom: sdm845: Demystify TDM masks a bit

[David Heidelberg via B4 Relay]

From: David Heidelberg <david@ixit.cz>

Describe the mask with the bits used for each RX/TX.

Signed-off-by: David Heidelberg <david@ixit.cz>
---
 sound/soc/qcom/sdm845.c | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/sound/soc/qcom/sdm845.c b/sound/soc/qcom/sdm845.c
index 6843ab8ba017e..0ce8265ab1c1f 100644
--- a/sound/soc/qcom/sdm845.c
+++ b/sound/soc/qcom/sdm845.c
@@ -18,19 +18,24 @@
 #include "sdw.h"
 #include "../codecs/rt5663.h"
 
 #define DRIVER_NAME	"sdm845"
 #define DEFAULT_SAMPLE_RATE_48K		48000
 #define DEFAULT_MCLK_RATE		24576000
 #define TDM_BCLK_RATE		6144000
 #define MI2S_BCLK_RATE		1536000
-#define LEFT_SPK_TDM_TX_MASK    0x30
-#define RIGHT_SPK_TDM_TX_MASK   0xC0
-#define SPK_TDM_RX_MASK         0x03
+#define LEFT_SPK_TDM_RX_MASK	BIT(0)
+#define RIGHT_SPK_TDM_RX_MASK	BIT(1)
+#define SPK_TDM_RX_MASK		(LEFT_SPK_TDM_RX_MASK | RIGHT_SPK_TDM_RX_MASK)
+#define MIC1_TDM_RX_MASK	BIT(2)
+#define MIC2_TDM_RX_MASK	BIT(3)
+#define MIC_TDM_RX_MASK		(MIC1_TDM_RX_MASK | MIC2_TDM_RX_MASK)
+#define LEFT_SPK_TDM_TX_MASK	(BIT(4) | BIT(5))
+#define RIGHT_SPK_TDM_TX_MASK	(BIT(6) | BIT(7))
 #define NUM_TDM_SLOTS           8
 #define SLIM_MAX_TX_PORTS 16
 #define SLIM_MAX_RX_PORTS 13
 #define WCD934X_DEFAULT_MCLK_RATE	9600000
 
 struct sdm845_snd_data {
 	struct snd_soc_jack jack;
 	bool jack_setup;
@@ -107,34 +112,36 @@ static int sdm845_tdm_snd_hw_params(struct snd_pcm_substream *substream,
 	default:
 		dev_err(rtd->dev, "%s: invalid param format 0x%x\n",
 				__func__, params_format(params));
 		return -EINVAL;
 	}
 
 	channels = params_channels(params);
 	if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK) {
-		ret = snd_soc_dai_set_tdm_slot(cpu_dai, 0, 0x3,
-				8, slot_width);
+		ret = snd_soc_dai_set_tdm_slot(cpu_dai, 0, SPK_TDM_RX_MASK,
+					       NUM_TDM_SLOTS, slot_width);
 		if (ret < 0) {
 			dev_err(rtd->dev, "%s: failed to set tdm slot, err:%d\n",
 					__func__, ret);
 			goto end;
 		}
 
 		ret = snd_soc_dai_set_channel_map(cpu_dai, 0, NULL,
 				channels, tdm_slot_offset);
 		if (ret < 0) {
 			dev_err(rtd->dev, "%s: failed to set channel map, err:%d\n",
 					__func__, ret);
 			goto end;
 		}
 	} else {
-		ret = snd_soc_dai_set_tdm_slot(cpu_dai, 0xf, 0,
-				8, slot_width);
+		ret = snd_soc_dai_set_tdm_slot(cpu_dai,
+					       SPK_TDM_RX_MASK |
+					       MIC_TDM_RX_MASK, 0,
+					       NUM_TDM_SLOTS, slot_width);
 		if (ret < 0) {
 			dev_err(rtd->dev, "%s: failed to set tdm slot, err:%d\n",
 					__func__, ret);
 			goto end;
 		}
 
 		ret = snd_soc_dai_set_channel_map(cpu_dai, channels,
 				tdm_slot_offset, 0, NULL);

-- 
2.55.0

Re: [PATCH v3 1/8] ASoC: qcom: sdm845: Demystify TDM masks a bit

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] sdm845: MIC_TDM_RX_MASK incorrectly implies microphone receives data
--

commit 0baf5b5f350c1d7b76ce347afed1d60796da5d93
Author: David Heidelberg <david@ixit.cz>

ASoC: qcom: sdm845: Demystify TDM masks a bit

Describe the mask with the bits used for each RX/TX.

> diff --git a/sound/soc/qcom/sdm845.c b/sound/soc/qcom/sdm845.c
> index 6843ab8ba017e..0ce8265ab1c1f 100644
> --- a/sound/soc/qcom/sdm845.c
> +++ b/sound/soc/qcom/sdm845.c
> @@ -23,9 +23,14 @@
>  #define DEFAULT_MCLK_RATE		24576000
>  #define TDM_BCLK_RATE		6144000
>  #define MI2S_BCLK_RATE		1536000
> -#define LEFT_SPK_TDM_TX_MASK    0x30
> -#define RIGHT_SPK_TDM_TX_MASK   0xC0
> -#define SPK_TDM_RX_MASK         0x03
> +#define LEFT_SPK_TDM_RX_MASK	BIT(0)
> +#define RIGHT_SPK_TDM_RX_MASK	BIT(1)
> +#define SPK_TDM_RX_MASK		(LEFT_SPK_TDM_RX_MASK | RIGHT_SPK_TDM_RX_MASK)
> +#define MIC1_TDM_RX_MASK	BIT(2)
> +#define MIC2_TDM_RX_MASK	BIT(3)
> +#define MIC_TDM_RX_MASK		(MIC1_TDM_RX_MASK | MIC2_TDM_RX_MASK)

[Severity: Low]
Does this naming convention accurately reflect the data direction?

The file appears to establish a peripheral-centric naming convention, where
SPK_TDM_RX_MASK is used for playback (speaker receives) and
LEFT_SPK_TDM_TX_MASK is used for IV sense (speaker transmits).

By this convention, since a microphone is an input device that transmits data
to the SoC, using an RX mask implies the microphone is receiving data. Would
it be more accurate to name these microphone slots MIC_TDM_TX_MASK to align
with the physical direction of the data flow?

> +#define LEFT_SPK_TDM_TX_MASK	(BIT(4) | BIT(5))
> +#define RIGHT_SPK_TDM_TX_MASK	(BIT(6) | BIT(7))
>  #define NUM_TDM_SLOTS           8

[ ... ]

> @@ -128,8 +133,10 @@ static int sdm845_tdm_snd_hw_params(struct snd_pcm_substream *substream,
>  			goto end;
>  		}
>  	} else {
> -		ret = snd_soc_dai_set_tdm_slot(cpu_dai, 0xf, 0,
> -				8, slot_width);
> +		ret = snd_soc_dai_set_tdm_slot(cpu_dai,
> +					       SPK_TDM_RX_MASK |
> +					       MIC_TDM_RX_MASK, 0,

[Severity: Low]
Is this using the correct mask name for the CPU capture path?

This forms the CPU capture mask using the inappropriately named 
MIC_TDM_RX_MASK, which contradicts the commit's stated intent to clarify 
RX/TX directions. Could this be updated to use a TX mask instead?

> +					       NUM_TDM_SLOTS, slot_width);
>  		if (ret < 0) {
>  			dev_err(rtd->dev, "%s: failed to set tdm slot, err:%d\n",
>  					__func__, ret);

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260924-pixel3-audio-v3-0-5d06460fc8a9@ixit.cz?part=1

[PATCH v3 2/8] ASoC: qcom: sdm845: use DSP_A format for TDM codec DAIs

[David Heidelberg via B4 Relay]

From: David Heidelberg <david@ixit.cz>

Before the DSP_B only worked because the only close-to-mainline consumer
cs35l36 codec was patched to map both DSP_A and DSP_B to the same
hardware register value (asp_fmt = 0), which is inherently DSP_A timing.
Use the right codec (DSP_A) which works as expected.

Signed-off-by: David Heidelberg <david@ixit.cz>
---
 sound/soc/qcom/sdm845.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sound/soc/qcom/sdm845.c b/sound/soc/qcom/sdm845.c
index 0ce8265ab1c1f..d1cc825fc5cdc 100644
--- a/sound/soc/qcom/sdm845.c
+++ b/sound/soc/qcom/sdm845.c
@@ -382,17 +382,17 @@ static int sdm845_snd_startup(struct snd_pcm_substream *substream)
 	case QUATERNARY_TDM_RX_0:
 	case QUATERNARY_TDM_TX_0:
 		if (++(data->quat_tdm_clk_count) == 1) {
 			snd_soc_dai_set_sysclk(cpu_dai,
 				Q6AFE_LPASS_CLK_ID_QUAD_TDM_IBIT,
 				TDM_BCLK_RATE, SNDRV_PCM_STREAM_PLAYBACK);
 		}
 
-		codec_dai_fmt |= SND_SOC_DAIFMT_IB_NF | SND_SOC_DAIFMT_DSP_B;
+		codec_dai_fmt |= SND_SOC_DAIFMT_IB_NF | SND_SOC_DAIFMT_DSP_A;
 
 		for_each_rtd_codec_dais(rtd, j, codec_dai) {
 
 			if (!strcmp(codec_dai->component->name_prefix,
 				    "Left")) {
 				ret = snd_soc_dai_set_fmt(
 						codec_dai, codec_dai_fmt);
 				if (ret < 0) {

-- 
2.55.0

[PATCH v3 3/8] ASoC: qcom: sdm845: Use per-speaker RX masks for TDM slot assignment

[David Heidelberg via B4 Relay]

From: David Heidelberg <david@ixit.cz>

Both Left and Right codec DAIs were passing the same SPK_TDM_RX_MASK,
both speakers ended up on slot 0, breaking the one speaker in
configuration such as on Pixel 3.

Split SPK_TDM_RX_MASK into per-speaker masks so that the Left codec
gets slot 0 (rx_mask=0x01) and the Right codec gets slot 1
(rx_mask=0x02).

This commit is here, so later CS35L36 receives correct slot for right
and left speakers.

Assisted-by: Claude:claude-4.6-opus
Signed-off-by: David Heidelberg <david@ixit.cz>
---
 sound/soc/qcom/sdm845.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/sound/soc/qcom/sdm845.c b/sound/soc/qcom/sdm845.c
index d1cc825fc5cdc..59cfa3b26cd4d 100644
--- a/sound/soc/qcom/sdm845.c
+++ b/sound/soc/qcom/sdm845.c
@@ -152,29 +152,29 @@ static int sdm845_tdm_snd_hw_params(struct snd_pcm_substream *substream,
 		}
 	}
 
 	for_each_rtd_codec_dais(rtd, j, codec_dai) {
 
 		if (!strcmp(codec_dai->component->name_prefix, "Left")) {
 			ret = snd_soc_dai_set_tdm_slot(
 					codec_dai, LEFT_SPK_TDM_TX_MASK,
-					SPK_TDM_RX_MASK, NUM_TDM_SLOTS,
+					LEFT_SPK_TDM_RX_MASK, NUM_TDM_SLOTS,
 					slot_width);
 			if (ret < 0) {
 				dev_err(rtd->dev,
 					"DEV0 TDM slot err:%d\n", ret);
 				return ret;
 			}
 		}
 
 		if (!strcmp(codec_dai->component->name_prefix, "Right")) {
 			ret = snd_soc_dai_set_tdm_slot(
 					codec_dai, RIGHT_SPK_TDM_TX_MASK,
-					SPK_TDM_RX_MASK, NUM_TDM_SLOTS,
+					RIGHT_SPK_TDM_RX_MASK, NUM_TDM_SLOTS,
 					slot_width);
 			if (ret < 0) {
 				dev_err(rtd->dev,
 					"DEV1 TDM slot err:%d\n", ret);
 				return ret;
 			}
 		}
 	}

-- 
2.55.0

[PATCH v3 4/8] ASoC: qcom: sdm845: Set codec dai and component sysclk during startup

[David Heidelberg via B4 Relay]

From: David Heidelberg <david@ixit.cz>

The cs35l36 codec needs the codec dai and component sysclk to be set
during TDM startup. Set these for all codec DAIs on the QUATERNARY_TDM
path, gracefully handling codecs that don't support sysclk by ignoring
-ENOTSUPP returns.

Based on work of Joel Selvaraj.

Signed-off-by: David Heidelberg <david@ixit.cz>
---
 sound/soc/qcom/sdm845.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/sound/soc/qcom/sdm845.c b/sound/soc/qcom/sdm845.c
index 59cfa3b26cd4d..f0fbec9c8df25 100644
--- a/sound/soc/qcom/sdm845.c
+++ b/sound/soc/qcom/sdm845.c
@@ -407,16 +407,38 @@ static int sdm845_snd_startup(struct snd_pcm_substream *substream)
 				ret = snd_soc_dai_set_fmt(
 						codec_dai, codec_dai_fmt);
 				if (ret < 0) {
 					dev_err(rtd->dev,
 						"Right TDM slot err:%d\n", ret);
 					return ret;
 				}
 			}
+
+			/* Set codec sysclk needed by codecs like cs35l36. */
+			ret = snd_soc_dai_set_sysclk(codec_dai, 0,
+						     TDM_BCLK_RATE,
+						     SND_SOC_CLOCK_IN);
+			if (ret < 0 && ret != -ENOTSUPP) {
+				dev_err(codec_dai->dev,
+					"Failed to set codec dai sysclk: %d\n",
+					ret);
+				return ret;
+			}
+
+			ret = snd_soc_component_set_sysclk(codec_dai->component,
+							   0, 0,
+							   TDM_BCLK_RATE,
+							   SND_SOC_CLOCK_IN);
+			if (ret < 0 && ret != -ENOTSUPP) {
+				dev_err(codec_dai->dev,
+					"Failed to set codec component sysclk: %d\n",
+					ret);
+				return ret;
+			}
 		}
 		break;
 	case SLIMBUS_0_RX...SLIMBUS_6_TX:
 		break;
 
 	default:
 		pr_err("%s: invalid dai id 0x%x\n", __func__, cpu_dai->id);
 		break;

-- 
2.55.0

[PATCH v3 5/8] ASoC: cs35l36: Implement set_tdm_slot to program RX and TX slots

[David Heidelberg via B4 Relay]

From: David Heidelberg <david@ixit.cz>

Program the ASP RX and TX slot registers from the TDM masks passed by
the machine driver. Each set bit in a mask names a slot; codec channels
are assigned to those slots in order, ASPRX1 taking the first RX slot
and ASPTX1..TX8 the first eight TX slots, with a warning if the mask
names more slots than the device has channels.

Passing slots == 0 or an empty mask restores the hardware defaults,
ASPRX1 in slot 0 and ASPTX1..TX8 in slots 0..7.

This lets a machine driver with several amplifiers on one bus, such as
sdm845 with two CS35L36, put each amplifier on its own RX slot and keep
their TX slots from colliding.

Assisted-by: Claude:claude-4.6-opus
Reviewed-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Signed-off-by: David Heidelberg <david@ixit.cz>
---
 sound/soc/codecs/cs35l36.c | 44 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 44 insertions(+)

diff --git a/sound/soc/codecs/cs35l36.c b/sound/soc/codecs/cs35l36.c
index 89645327945f1..4d35ca6291548 100644
--- a/sound/soc/codecs/cs35l36.c
+++ b/sound/soc/codecs/cs35l36.c
@@ -942,20 +942,64 @@ static const struct cs35l36_pll_config *cs35l36_get_clk_config(
 	for (i = 0; i < ARRAY_SIZE(cs35l36_pll_sysclk); i++) {
 		if (cs35l36_pll_sysclk[i].freq == freq)
 			return &cs35l36_pll_sysclk[i];
 	}
 
 	return NULL;
 }
 
+static void cs35l36_mask_to_slots(struct cs35l36_private *cs35l36,
+				  unsigned long mask, unsigned int base_reg,
+				  unsigned int nchan)
+{
+	unsigned int chan = 0, shift;
+	int slot;
+
+	/* Two 6-bit slot fields per register, at bits 0 and 16 */
+	for_each_set_bit(slot, &mask, BITS_PER_TYPE(mask)) {
+		if (chan == nchan) {
+			dev_warn(cs35l36->dev,
+				 "Too many slots in TDM mask: %lx\n", mask);
+			return;
+		}
+
+		shift = (chan % 2) * CS35L36_ASP_TX2_SLOT_SHIFT;
+		regmap_update_bits(cs35l36->regmap, base_reg + (chan / 2) * 4,
+				   CS35L36_ASP_RX1_SLOT_MASK << shift,
+				   slot << shift);
+		chan++;
+	}
+}
+
+static int cs35l36_set_tdm_slot(struct snd_soc_dai *dai,
+				unsigned int tx_mask, unsigned int rx_mask,
+				int slots, int slot_width)
+{
+	struct cs35l36_private *cs35l36 =
+			snd_soc_component_get_drvdata(dai->component);
+
+	/* Note: rx/tx is from point of view of the CPU end */
+	if (!slots || !rx_mask)
+		rx_mask = BIT(0);		/* ASPRX1 in slot 0 */
+
+	if (!slots || !tx_mask)
+		tx_mask = GENMASK(7, 0);	/* ASPTX1..8 in slots 0..7 */
+
+	cs35l36_mask_to_slots(cs35l36, rx_mask, CS35L36_ASP_RX1_SLOT, 1);
+	cs35l36_mask_to_slots(cs35l36, tx_mask, CS35L36_ASP_TX1_TX2_SLOT, 8);
+
+	return 0;
+}
+
 static const struct snd_soc_dai_ops cs35l36_ops = {
 	.set_fmt = cs35l36_set_dai_fmt,
 	.hw_params = cs35l36_pcm_hw_params,
 	.set_sysclk = cs35l36_dai_set_sysclk,
+	.set_tdm_slot = cs35l36_set_tdm_slot,
 };
 
 #define CS35L36_RATES (		    \
 	SNDRV_PCM_RATE_8000_48000 | \
 	SNDRV_PCM_RATE_12000 |	    \
 	SNDRV_PCM_RATE_24000 |	    \
 	SNDRV_PCM_RATE_88200 |	    \
 	SNDRV_PCM_RATE_96000 |	    \

-- 
2.55.0

Re: [PATCH v3 5/8] ASoC: cs35l36: Implement set_tdm_slot to program RX and TX slots

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 4 potential issue(s) to consider:
- [High] ASoC: cs35l36: I2S bus contention from unmapped TDM channels
- [High] ASoC: cs35l36: DAPM kcontrol corrupts TDM slot configuration
- [Medium] ASoC: cs35l36: missing error propagation for invalid TDM mask
- [Low] ASoC: cs35l36: inverted TDM mask perspective in comment
--

commit caaab40c1180fd182ef9f5dce19a698193666d09
Author: David Heidelberg <david@ixit.cz>

ASoC: cs35l36: Implement set_tdm_slot to program RX and TX slots

This commit introduces the set_tdm_slot callback for the CS35L36 codec to
allow machine drivers to dynamically program ASP RX and TX slot assignments.
The new configuration maps each bit set in the provided TDM masks to
consecutive hardware channels, enforcing maximum channel limits.

> diff --git a/sound/soc/codecs/cs35l36.c b/sound/soc/codecs/cs35l36.c
> index 89645327945f1..4d35ca6291548 100644
> --- a/sound/soc/codecs/cs35l36.c
> +++ b/sound/soc/codecs/cs35l36.c
> @@ -947,10 +947,54 @@ static const struct cs35l36_pll_config *cs35l36_get_clk_config(
>  	return NULL;
>  }
>  
> +static void cs35l36_mask_to_slots(struct cs35l36_private *cs35l36,
> +				  unsigned long mask, unsigned int base_reg,
> +				  unsigned int nchan)
> +{
> +	unsigned int chan = 0, shift;
> +	int slot;
> +
> +	/* Two 6-bit slot fields per register, at bits 0 and 16 */
> +	for_each_set_bit(slot, &mask, BITS_PER_TYPE(mask)) {
> +		if (chan == nchan) {
> +			dev_warn(cs35l36->dev,
> +				 "Too many slots in TDM mask: %lx\n", mask);
> +			return;

[Severity: Medium]
Will this error state be lost since this function returns void?

If a machine driver passes a mask with more bits set than supported channels,
this warning prints, but cs35l36_set_tdm_slot() continues and unconditionally
returns 0.

Should this configuration failure be propagated back to the sound framework
by returning a negative error code instead?

> +		}
> +
> +		shift = (chan % 2) * CS35L36_ASP_TX2_SLOT_SHIFT;
> +		regmap_update_bits(cs35l36->regmap, base_reg + (chan / 2) * 4,
> +				   CS35L36_ASP_RX1_SLOT_MASK << shift,
> +				   slot << shift);
> +		chan++;
> +	}

[Severity: High]
Does this leave remaining channels unmapped on their default slots if the
provided mask specifies fewer slots than the total channels?

In a multi-codec setup, the CPU DAI typically opens the stream with the total
number of channels on the bus. DAPM then powers on all asptx widgets up to
that channel count.

If we only map a subset of slots here, the remaining powered-on channels
will retain their hardware default slots and actively transmit on them.

Could this lead to hardware contention on the I2S bus and data corruption by
colliding with properly mapped channels or other codecs?

> +}
> +
> +static int cs35l36_set_tdm_slot(struct snd_soc_dai *dai,
> +				unsigned int tx_mask, unsigned int rx_mask,
> +				int slots, int slot_width)
> +{
> +	struct cs35l36_private *cs35l36 =
> +			snd_soc_component_get_drvdata(dai->component);
> +
> +	/* Note: rx/tx is from point of view of the CPU end */

[Severity: Low]
Is this comment accurate? The implementation below maps rx_mask to the
codec's rx register and tx_mask to the codec's tx register.

If it were from the CPU's point of view, the CPU's rx mask would map to the
codec's tx. The code correctly follows ASoC semantics (from the codec's
perspective), which makes the comment appear inverted.

> +	if (!slots || !rx_mask)
> +		rx_mask = BIT(0);		/* ASPRX1 in slot 0 */
> +
> +	if (!slots || !tx_mask)
> +		tx_mask = GENMASK(7, 0);	/* ASPTX1..8 in slots 0..7 */
> +
> +	cs35l36_mask_to_slots(cs35l36, rx_mask, CS35L36_ASP_RX1_SLOT, 1);

[Severity: High]
Could programming CS35L36_ASP_RX1_SLOT here conflict with the existing
DAPM kcontrol?

There appears to be a legacy DAPM kcontrol (cs35l36_chan_mux / chansel_enum)
that operates on bit 0 of this exact same register to toggle between
rx1 and rx2.

If a machine driver sets the rx slot to a value like 2 (binary 000010) here,
and userspace later restores the Input Mux kcontrol, could the framework
forcefully toggle bit 0 and inadvertently change the slot configuration?

> +	cs35l36_mask_to_slots(cs35l36, tx_mask, CS35L36_ASP_TX1_TX2_SLOT, 8);
> +
> +	return 0;
> +}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260924-pixel3-audio-v3-0-5d06460fc8a9@ixit.cz?part=5

[PATCH v3 6/8] arm64: dts: qcom: sdm845-google: Add basic audio support

[David Heidelberg via B4 Relay]

From: David Heidelberg <david@ixit.cz>

Introduce support for sound card and wire two CS35L36 audio codecs for
top and bottom speakers.

Inspired by commit from Joel Selvaraj.

Signed-off-by: David Heidelberg <david@ixit.cz>
---
 arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi | 132 ++++++++++++++++++++-
 1 file changed, 130 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi b/arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi
index 9e952f9862f24..cf87b582e3997 100644
--- a/arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi
+++ b/arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi
@@ -2,16 +2,18 @@
 
 /dts-v1/;
 
 #include <dt-bindings/arm/qcom,ids.h>
 #include <dt-bindings/dma/qcom-gpi.h>
 #include <dt-bindings/input/linux-event-codes.h>
 #include <dt-bindings/interrupt-controller/irq.h>
 #include <dt-bindings/regulator/qcom,rpmh-regulator.h>
+#include <dt-bindings/sound/qcom,q6afe.h>
+#include <dt-bindings/sound/qcom,q6asm.h>
 
 #include "sdm845.dtsi"
 #include "pm8998.dtsi"
 #include "pmi8998.dtsi"
 
 /delete-node/ &mpss_region;
 /delete-node/ &venus_mem;
 /delete-node/ &cdsp_mem;
@@ -404,19 +406,61 @@ nfc@28 {
 		firmware-gpios = <&tlmm 79 GPIO_ACTIVE_HIGH>;
 
 		pinctrl-0 = <&nfc_int_default &nfc_enable_default>;
 		pinctrl-names = "default";
 	};
 };
 
 &i2c12 {
-	/* Bottom spkr (right) CS35L36 @ 40 */
+	status = "okay";
+
+	cs35l36_bottom: audio-codec@40 {
+		compatible = "cirrus,cs35l36";
+		reg = <0x40>;
+		reset-gpios = <&tlmm 112 GPIO_ACTIVE_HIGH>;
+		#sound-dai-cells = <0>;
+		sound-name-prefix = "Right"; /* Bottom */
+		VA-supply = <&vreg_s4a_1p8>;
+		interrupts-extended = <&tlmm 115 IRQ_TYPE_LEVEL_LOW>;
+
+		pinctrl-0 = <&cs35l36_bottom_default_state>;
+		pinctrl-names = "default";
+
+		cirrus,boost-ind-nanohenry = <1000>;
+		cirrus,boost-ctl-select = <1>;
+		cirrus,boost-ctl-millivolt = <10000>;
+		cirrus,boost-peak-milliamp = <3700>;
+		cirrus,temp-warn-threshold = <1>;
+		cirrus,multi-amp-mode;
+		cirrus,irq-drive-select = <0>;
+		cirrus,irq-gpio-select = <0>;
+	};
 
-	/* Top spkr (left) CS35L36 @ 41 */
+	cs35l36_top: audio-codec@41 {
+		compatible = "cirrus,cs35l36";
+		reg = <0x41>;
+		reset-gpios = <&tlmm 75 GPIO_ACTIVE_HIGH>;
+		#sound-dai-cells = <0>;
+		sound-name-prefix = "Left"; /* Top */
+		VA-supply = <&vreg_s4a_1p8>;
+		interrupts-extended = <&tlmm 40 IRQ_TYPE_LEVEL_LOW>;
+
+		pinctrl-0 = <&cs35l36_top_default_state>;
+		pinctrl-names = "default";
+
+		cirrus,boost-ind-nanohenry = <1000>;
+		cirrus,boost-ctl-select = <0x01>;
+		cirrus,boost-ctl-millivolt = <10000>;
+		cirrus,boost-peak-milliamp = <3700>;
+		cirrus,temp-warn-threshold = <1>;
+		cirrus,multi-amp-mode;
+		cirrus,irq-drive-select = <0>;
+		cirrus,irq-gpio-select = <0>;
+	};
 };
 
 &ipa {
 	firmware-name = "qcom/sdm845/Google/blueline/ipa_fws.mbn";
 	memory-region = <&ipa_fw_mem>;
 
 	status = "okay";
 };
@@ -449,16 +493,35 @@ &pm8998_resin {
 };
 
 &pmi8998_charger {
 	monitored-battery = <&battery>;
 
 	status = "okay";
 };
 
+&q6afedai {
+	dai@72 {
+		reg = <QUATERNARY_TDM_RX_0>;
+
+		qcom,tdm-sync-mode = <0>;
+		qcom,tdm-sync-src = <1>;
+		qcom,tdm-data-out = <0>;
+		qcom,tdm-invert-sync = <0>;
+		qcom,tdm-data-delay = <1>;
+		qcom,tdm-data-align = <0>;
+	};
+};
+
+&q6asmdai {
+	dai@0 {
+		reg = <MSM_FRONTEND_DAI_MULTIMEDIA1>;
+	};
+};
+
 &qupv3_id_0 {
 	status = "okay";
 };
 
 &qupv3_id_1 {
 	status = "okay";
 };
 
@@ -467,16 +530,49 @@ &qup_uart9_rx {
 	bias-pull-up;
 };
 
 &qup_uart9_tx {
 	drive-strength = <2>;
 	bias-disable;
 };
 
+&sound {
+	compatible = "qcom,sdm845-sndcard";
+	model = "blueline";
+	pinctrl-0 = <&quat_mi2s_active>, <&quat_mi2s_sd0_active>,
+		    <&quat_mi2s_sd1_active>;
+	pinctrl-names = "default";
+
+	/* mm[0-9]-dai-link nodes must be positioned first regardless of sorting */
+	mm1-dai-link {
+		link-name = "MultiMedia1";
+
+		cpu {
+			sound-dai = <&q6asmdai MSM_FRONTEND_DAI_MULTIMEDIA1>;
+		};
+	};
+
+	cs35l36-dai-link {
+		link-name = "Speaker Playback";
+
+		codec {
+			sound-dai = <&cs35l36_bottom>, <&cs35l36_top>;
+		};
+
+		cpu {
+			sound-dai = <&q6afedai QUATERNARY_TDM_RX_0>;
+		};
+
+		platform {
+			sound-dai = <&q6routing>;
+		};
+	};
+};
+
 &tlmm {
 	gpio-reserved-ranges = < 0 4>, /* SPI (Intel MNH Pixel Visual Core) */
 			       <81 4>; /* SPI (most likely Fingerprint Cards FPC1075) */
 
 	nfc_int_default: nfc-int-default-state {
 		pins = "gpio63";
 		function = "gpio";
 		drive-strength = <2>;
@@ -505,16 +601,48 @@ touchscreen_pins: ts-pins-gpio-state {
 	};
 
 	touchscreen_i2c_pins: qup-i2c2-gpio-state {
 		pins = "gpio27", "gpio28";
 		function = "gpio";
 		drive-strength = <2>;
 		bias-disable;
 	};
+
+	cs35l36_bottom_default_state: cs35l36-bottom-default-state {
+		reset-pins {
+			pins = "gpio112";
+			function = "gpio";
+			drive-strength = <2>;
+			bias-disable;
+		};
+
+		irq-pins {
+			pins = "gpio115";
+			function = "gpio";
+			drive-strength = <2>;
+			bias-pull-up;
+		};
+	};
+
+	cs35l36_top_default_state: cs35l36-top-default-state {
+		irq-pins {
+			pins = "gpio40";
+			function = "gpio";
+			drive-strength = <2>;
+			bias-pull-up;
+		};
+
+		reset-pins {
+			pins = "gpio75";
+			function = "gpio";
+			drive-strength = <2>;
+			bias-disable;
+		};
+	};
 };
 
 &uart6 {
 	pinctrl-0 = <&qup_uart6_4pin>;
 
 	status = "okay";
 
 	bluetooth {

-- 
2.55.0

[PATCH v3 7/8] slimbus: qcom-ngd-ctrl: Remove data channels on stream disable

[David Heidelberg via B4 Relay]

From: David Heidelberg <david@ixit.cz>

qcom_slim_ngd_xfer_msg() silently returns success for every core
reconfiguration message, so the generic channel removal done by
slim_stream_disable() never reaches the ADSP SLIMbus master. The
master therefore keeps the channels of a closed stream. When the next
stream defines the same channel numbers again, the WCD9340 capture
completes without any error but records only zeros; removing the
channels explicitly when the stream is disabled makes it work again.

Implement the disable_stream controller op with the sequence the
downstream NGD driver sends for channel removal, CHAN_CTRL(REMOVE)
listing the stream's channels followed by RECONFIG_NOW. Like
downstream, treat -EREMOTEIO from the removal as success: it means the
controller is being restarted and the channels are gone anyway.

Based on work done by Casey Tunturi.

Assisted-by: Claude:claude-fable-5-1
Signed-off-by: David Heidelberg <david@ixit.cz>
---
 drivers/slimbus/qcom-ngd-ctrl.c | 81 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 81 insertions(+)

diff --git a/drivers/slimbus/qcom-ngd-ctrl.c b/drivers/slimbus/qcom-ngd-ctrl.c
index 934c44c5bc1a9..c91f55036dd55 100644
--- a/drivers/slimbus/qcom-ngd-ctrl.c
+++ b/drivers/slimbus/qcom-ngd-ctrl.c
@@ -1080,16 +1080,96 @@ static int qcom_slim_ngd_enable_stream(struct slim_stream_runtime *rt)
 		slim_free_txn_tid(ctrl, &txn);
 		dev_err(&sdev->dev, "TX timed out:MC:0x%x,mt:0x%x", txn.mc,
 				txn.mt);
 	}
 
 	return ret;
 }
 
+/* CHAN_CTRL operation, bits [7:6] of the first payload byte */
+#define SLIM_USR_CHAN_CTRL_REMOVE	2
+
+static int qcom_slim_ngd_disable_stream(struct slim_stream_runtime *rt)
+{
+	struct slim_device *sdev = rt->dev;
+	struct slim_controller *ctrl = sdev->ctrl;
+	struct slim_val_inf msg =  {0};
+	u8 wbuf[SLIM_MSGQ_BUF_LEN];
+	u8 rbuf[SLIM_MSGQ_BUF_LEN];
+	struct slim_msg_txn txn = {0,};
+	int i, ret;
+
+	txn.mt = SLIM_MSG_MT_DEST_REFERRED_USER;
+	txn.dt = SLIM_MSG_DEST_LOGICALADDR;
+	txn.la = SLIM_LA_MGR;
+	txn.ec = 0;
+	txn.msg = &msg;
+	txn.msg->num_bytes = 0;
+	txn.msg->wbuf = wbuf;
+	txn.msg->rbuf = rbuf;
+
+	/*
+	 * The NGD drops the core reconfiguration messages that
+	 * slim_stream_disable() uses to remove channels, so ask the master
+	 * to remove them the way the downstream NGD driver does:
+	 * CHAN_CTRL(REMOVE) listing every channel, then RECONFIG_NOW.
+	 * Without this the master keeps the channels active and a later
+	 * DEF_ACT_CHAN with the same channel numbers is a no-op, so the
+	 * data path is never re-armed and a reopened stream moves no data.
+	 */
+	/* 5-bit client number, operation in bits [7:6] */
+	wbuf[txn.msg->num_bytes++] = (SLIM_USR_CHAN_CTRL_REMOVE << 6) |
+				     (sdev->laddr & 0x1f);
+
+	ret = slim_alloc_txn_tid(ctrl, &txn);
+	if (ret) {
+		dev_err(&sdev->dev, "Fail to allocate TID\n");
+		return ret;
+	}
+	wbuf[txn.msg->num_bytes++] = txn.tid;
+
+	for (i = 0; i < rt->num_ports; i++)
+		wbuf[txn.msg->num_bytes++] = rt->ports[i].ch.id;
+
+	txn.mc = SLIM_USR_MC_CHAN_CTRL;
+	txn.rl = txn.msg->num_bytes + 4;
+	ret = qcom_slim_ngd_xfer_msg_sync(ctrl, &txn);
+	if (ret) {
+		slim_free_txn_tid(ctrl, &txn);
+		/* Controller restarting, the channels are gone anyway */
+		if (ret == -EREMOTEIO)
+			return 0;
+		dev_err(&sdev->dev, "TX timed out:MC:0x%x,mt:0x%x", txn.mc,
+				txn.mt);
+		return ret;
+	}
+
+	txn.mc = SLIM_USR_MC_RECONFIG_NOW;
+	txn.msg->num_bytes = 2;
+	wbuf[1] = sdev->laddr;
+	txn.rl = txn.msg->num_bytes + 4;
+
+	ret = slim_alloc_txn_tid(ctrl, &txn);
+	if (ret) {
+		dev_err(ctrl->dev, "Fail to allocate TID\n");
+		return ret;
+	}
+
+	wbuf[0] = txn.tid;
+	ret = qcom_slim_ngd_xfer_msg_sync(ctrl, &txn);
+	if (ret) {
+		slim_free_txn_tid(ctrl, &txn);
+		dev_err(&sdev->dev, "TX timed out:MC:0x%x,mt:0x%x", txn.mc,
+				txn.mt);
+	}
+
+	return ret;
+}
+
 static int qcom_slim_ngd_get_laddr(struct slim_controller *ctrl,
 				   struct slim_eaddr *ea, u8 *laddr)
 {
 	struct slim_val_inf msg =  {0};
 	u8 failed_ea[6] = {0, 0, 0, 0, 0, 0};
 	struct slim_msg_txn txn;
 	u8 wbuf[10] = {0};
 	u8 rbuf[10] = {0};
@@ -1619,16 +1699,17 @@ static int qcom_slim_ngd_ctrl_probe(struct platform_device *pdev)
 	ctrl->framer.rootfreq = SLIM_ROOT_FREQ >> 3;
 	ctrl->framer.superfreq =
 		ctrl->framer.rootfreq / SLIM_CL_PER_SUPERFRAME_DIV8;
 
 	ctrl->ctrl.a_framer = &ctrl->framer;
 	ctrl->ctrl.clkgear = SLIM_MAX_CLK_GEAR;
 	ctrl->ctrl.get_laddr = qcom_slim_ngd_get_laddr;
 	ctrl->ctrl.enable_stream = qcom_slim_ngd_enable_stream;
+	ctrl->ctrl.disable_stream = qcom_slim_ngd_disable_stream;
 	ctrl->ctrl.xfer_msg = qcom_slim_ngd_xfer_msg;
 	ctrl->ctrl.wakeup = NULL;
 	ctrl->state = QCOM_SLIM_NGD_CTRL_DOWN;
 
 	mutex_init(&ctrl->tx_lock);
 	mutex_init(&ctrl->ssr_lock);
 	spin_lock_init(&ctrl->tx_buf_lock);
 	init_completion(&ctrl->reconf);

-- 
2.55.0

[PATCH v3 8/8] arm64: dts: qcom: sdm845-google: Add WCD9340 codec and microphone capture

[David Heidelberg via B4 Relay]

From: David Heidelberg <david@ixit.cz>

The Pixel 3 sound card so far only drove the two CS35L36 speaker
amplifiers. Wire in the WCD9340 codec , add a SLIMbus capture link
for it and route the three digital microphones

Based on work done by Casey Tunturi.

Assisted-by: Claude:claude-fable-5-1
Signed-off-by: David Heidelberg <david@ixit.cz>
---
 arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi | 63 ++++++++++++++++++++++
 1 file changed, 63 insertions(+)

diff --git a/arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi b/arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi
index cf87b582e3997..5393aee0370ac 100644
--- a/arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi
+++ b/arch/arm64/boot/dts/qcom/sdm845-google-common.dtsi
@@ -4,18 +4,20 @@
 
 #include <dt-bindings/arm/qcom,ids.h>
 #include <dt-bindings/dma/qcom-gpi.h>
 #include <dt-bindings/input/linux-event-codes.h>
 #include <dt-bindings/interrupt-controller/irq.h>
 #include <dt-bindings/regulator/qcom,rpmh-regulator.h>
 #include <dt-bindings/sound/qcom,q6afe.h>
 #include <dt-bindings/sound/qcom,q6asm.h>
+#include <dt-bindings/sound/qcom,wcd934x.h>
 
 #include "sdm845.dtsi"
+#include "sdm845-wcd9340.dtsi"
 #include "pm8998.dtsi"
 #include "pmi8998.dtsi"
 
 /delete-node/ &mpss_region;
 /delete-node/ &venus_mem;
 /delete-node/ &cdsp_mem;
 /delete-node/ &mba_region;
 /delete-node/ &slpi_mem;
@@ -510,16 +512,20 @@ dai@72 {
 		qcom,tdm-data-align = <0>;
 	};
 };
 
 &q6asmdai {
 	dai@0 {
 		reg = <MSM_FRONTEND_DAI_MULTIMEDIA1>;
 	};
+
+	dai@1 {
+		reg = <MSM_FRONTEND_DAI_MULTIMEDIA2>;
+	};
 };
 
 &qupv3_id_0 {
 	status = "okay";
 };
 
 &qupv3_id_1 {
 	status = "okay";
@@ -537,40 +543,81 @@ &qup_uart9_tx {
 
 &sound {
 	compatible = "qcom,sdm845-sndcard";
 	model = "blueline";
 	pinctrl-0 = <&quat_mi2s_active>, <&quat_mi2s_sd0_active>,
 		    <&quat_mi2s_sd1_active>;
 	pinctrl-names = "default";
 
+	audio-routing = "DMIC0", "MCLK", /* Top mic */
+			"DMIC0", "MIC BIAS1",
+			"DMIC3", "MCLK", /* Top-edge mic */
+			"DMIC3", "MIC BIAS3",
+			"DMIC4", "MCLK", /* Bottom mic */
+			"DMIC4", "MIC BIAS4";
+
 	/* mm[0-9]-dai-link nodes must be positioned first regardless of sorting */
 	mm1-dai-link {
 		link-name = "MultiMedia1";
 
 		cpu {
 			sound-dai = <&q6asmdai MSM_FRONTEND_DAI_MULTIMEDIA1>;
 		};
 	};
 
+	mm2-dai-link {
+		link-name = "MultiMedia2";
+
+		cpu {
+			sound-dai = <&q6asmdai MSM_FRONTEND_DAI_MULTIMEDIA2>;
+		};
+	};
+
 	cs35l36-dai-link {
 		link-name = "Speaker Playback";
 
 		codec {
 			sound-dai = <&cs35l36_bottom>, <&cs35l36_top>;
 		};
 
 		cpu {
 			sound-dai = <&q6afedai QUATERNARY_TDM_RX_0>;
 		};
 
 		platform {
 			sound-dai = <&q6routing>;
 		};
 	};
+
+	slimcap-dai-link {
+		link-name = "SLIM Capture 1";
+
+		codec {
+			sound-dai = <&wcd9340 AIF1_CAP>;
+		};
+
+		cpu {
+			sound-dai = <&q6afedai SLIMBUS_0_TX>;
+		};
+
+		platform {
+			sound-dai = <&q6routing>;
+		};
+	};
+};
+
+&swm {
+	/*
+	 * Nothing hangs off the codec's SoundWire bus on this board: the
+	 * speakers are CS35L36 amplifiers on QUAT_TDM. With an empty bus the
+	 * controller reports port mismatches and frame-generator timeouts on
+	 * every boot.
+	 */
+	status = "disabled";
 };
 
 &tlmm {
 	gpio-reserved-ranges = < 0 4>, /* SPI (Intel MNH Pixel Visual Core) */
 			       <81 4>; /* SPI (most likely Fingerprint Cards FPC1075) */
 
 	nfc_int_default: nfc-int-default-state {
 		pins = "gpio63";
@@ -706,16 +753,32 @@ &usb_1_qmpphy {
 };
 
 &venus {
 	firmware-name = "qcom/sdm845/Google/blueline/venus.mbn";
 
 	status = "okay";
 };
 
+&wcd9340 {
+	/*
+	 * PM8998 LN_BB_CLK2 feeds the Pixel Visual Core on this board; the
+	 * codec MCLK comes from LN_BB_CLK3. With the wrong buffer the codec
+	 * never sees 19.2 MHz: eFuse sensing times out and every capture
+	 * path reads zeros while SLIMbus control traffic still works.
+	 */
+	clocks = <&rpmhcc RPMH_LN_BB_CLK3>;
+	reset-gpios = <&tlmm 64 GPIO_ACTIVE_HIGH>;
+	vdd-buck-supply = <&vreg_s4a_1p8>;
+	vdd-buck-sido-supply = <&vreg_s4a_1p8>;
+	vdd-tx-supply = <&vreg_s4a_1p8>;
+	vdd-rx-supply = <&vreg_s4a_1p8>;
+	vdd-io-supply = <&vreg_s4a_1p8>;
+};
+
 &wifi {
 	vdd-0.8-cx-mx-supply = <&vreg_l5a_0p8>;
 	vdd-1.8-xo-supply = <&vreg_pmu_xo>;
 	vdd-1.3-rfa-supply = <&vreg_pmu_rf>;
 	vdd-3.3-ch0-supply = <&vreg_pmu_ch0>;
 
 	qcom,snoc-host-cap-8bit-quirk;
 

-- 
2.55.0

Re: [PATCH v3 0/8] Speakers for Pixel 3 / 3 XL

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 242 bytes --]

On Thu, Sep 24, 2026 at 02:46:44PM +0200, David Heidelberg via B4 Relay wrote:
> This series address:
> 
> 0. demistify the hex to BIT() macro and add left/right rx spk

This doesn't apply against current code, please check and resend.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH v3 0/8] Speakers for Pixel 3 / 3 XL

[David Heidelberg]

On 24/09/2026 15:04, Mark Brown wrote:
> On Thu, Sep 24, 2026 at 02:46:44PM +0200, David Heidelberg via B4 Relay wrote:
>> This series address:
>>
>> 0. demistify the hex to BIT() macro and add left/right rx spk
> 
> This doesn't apply against current code, please check and resend.

rebased on top of the breaking commit and next-20260923 :)

Thanks
David