Skip to contentSkip to conversation

ASoC: Fix constraint return value handling

PATCH In review [email protected] 5 participants 39 messages Oct 15 – Oct 16

Subsystems (15)
lore
Share & tools
lore
#0 lore raw

v2 cover ASoC: Fix constraint return value handling

From: bui duc phuc <[email protected]>

Hi all,

snd_pcm_hw_constraint_minmax() and snd_pcm_hw_constraint_integer() may return positive values on success. However, ASoC only marks the DAI, link, or component as started or opened when the corresponding callback returns 0.

This series fixes affected ASoC drivers to return 0 on success and properly handle rollback paths.

Found by manual code inspection.

Compile-tested only.

Link v1:
https://lore.kernel.org/all/[email protected]/

Changes in v2:
- Add Reviewed-by tags from Vijendar Mukunda and Chancel Liu.
- Remove redundant integer constraints already applied by snd_dmaengine_pcm_open().

Best regards,
Phuc

bui duc phuc (25):
ASoC: amd: acp: Fix .close() being skipped on rollback
ASoC: amd: ps: Fix acp63_pdm_dma_close() being skipped on rollback
ASoC: amd: ps: Fix acp63_sdw_dma_close() being skipped on rollback
ASoC: amd: raven: Fix acp3x_dma_close() being skipped on rollback
ASoC: amd: renoir: Fix acp_pdm_dma_close() being skipped on rollback
ASoC: amd: vangogh: Fix acp5x_dma_close() being skipped on rollback
ASoC: amd: yc: Fix acp6x_pdm_dma_close() being skipped on rollback
ASoC: apple: mca: Ensure the DAI is marked as started on success
ASoC: atmel: atmel-pcm-pdc: Fix atmel_pcm_close() being skipped on
rollback
ASoC: codecs: cs42l42: Ensure the DAI is marked as started on success
ASoC: codecs: nau8325: Ensure the DAI is marked as started on success
ASoC: codecs: nau8540: Ensure the DAI is marked as started on success
ASoC: codecs: nau8821: Ensure the DAI is marked as started on success
ASoC: codecs: nau8824: Ensure the DAI is marked as started on success
ASoC: codecs: nau8825: Ensure the DAI is marked as started on success
ASoC: codecs: wm8580: Ensure the DAI is marked as started on success
ASoC: codecs: wm8782: Ensure the DAI is marked as started on success
ASoC: imx-pcm-rpmsg: Fix imx_rpmsg_pcm_close() being skipped on
rollback
ASoC: intel: atom: Fix .shutdown() being skipped on rollback
ASoC: renesas: fsi: Ensure the component is marked as opened on
success
ASoC: renesas: msiof: Remove redundant period constraint
ASoC: renesas: rz-ssi: Ensure the component is marked as opened on
success
ASoC: renesas: rz-ssi: Remove redundant component open callback
ASoC: rockchip: rk3399_gru_sound: Mark the link as started on success
ASoC: rockchip: rockchip_max98090: Mark the link as started on success

 sound/soc/amd/acp/acp-platform.c             |  7 ++++++-
 sound/soc/amd/ps/ps-pdm-dma.c                |  8 +++++++-
 sound/soc/amd/ps/ps-sdw-dma.c                |  8 +++++++-
 sound/soc/amd/raven/acp3x-pcm-dma.c          |  8 +++++++-
 sound/soc/amd/renoir/acp3x-pdm-dma.c         |  8 +++++++-
 sound/soc/amd/vangogh/acp5x-pcm-dma.c        |  8 +++++++-
 sound/soc/amd/yc/acp6x-pdm-dma.c             |  8 +++++++-
 sound/soc/apple/mca.c                        | 16 +++++++++++++---
 sound/soc/atmel/atmel-pcm-pdc.c              |  7 +++++++
 sound/soc/codecs/cs42l42.c                   | 16 +++++++++++++---
 sound/soc/codecs/nau8325.c                   | 16 +++++++++++++---
 sound/soc/codecs/nau8540.c                   | 16 +++++++++++++---
 sound/soc/codecs/nau8821.c                   | 16 +++++++++++++---
 sound/soc/codecs/nau8824.c                   | 16 +++++++++++++---
 sound/soc/codecs/nau8825.c                   | 16 +++++++++++++---
 sound/soc/codecs/wm8580.c                    | 15 +++++++++++++--
 sound/soc/codecs/wm8782.c                    | 12 +++++++++++-
 sound/soc/fsl/imx-pcm-rpmsg.c                |  8 +++++++-
 sound/soc/intel/atom/sst-mfld-platform-pcm.c |  7 ++++++-
 sound/soc/renesas/fsi.c                      | 11 +++++++++--
 sound/soc/renesas/rcar/msiof.c               | 16 ++++++----------
 sound/soc/renesas/rz-ssi.c                   | 19 +++++++++++--------
 sound/soc/rockchip/rk3399_gru_sound.c        | 14 ++++++++++++--
 sound/soc/rockchip/rockchip_max98090.c       | 15 +++++++++++++--
 24 files changed, 234 insertions(+), 57 deletions(-)
Signature · 1 line
2.43.0
#1 lore raw

PATCH v2 1/25 ASoC: amd: acp: Fix .close() being skipped on rollback

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from acp_dma_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes 623621a9f9e1 ("ASoC: amd: Add common framework to support I2S on ACP SOC")
---
 sound/soc/amd/acp/acp-platform.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
sound/soc/amd/acp/acp-platform.c +6 -1
index 6b1e18b31c1c..f2c8853775ad 100644
@@ -248,7 +248,12 @@ static int acp_dma_open(struct snd_soc_component *component, struct snd_pcm_subs
 	list_add_tail(&stream->list, &chip->stream_list);
 	spin_unlock_irq(&chip->acp_lock);
 
-	return ret;
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 static int acp_dma_hw_params(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#4 lore raw

PATCH v2 2/25 ASoC: amd: ps: Fix acp63_pdm_dma_close() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from acp63_pdm_dma_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes 5137305662ef ("ASoC: amd: add acp6.2 pdm driver dma ops")
---
 sound/soc/amd/ps/ps-pdm-dma.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
sound/soc/amd/ps/ps-pdm-dma.c +7 -1
index 04c014349347..12c341e89bce 100644
@@ -211,7 +211,13 @@ static int acp63_pdm_dma_open(struct snd_soc_component *component,
 
 	pdm_data->acp63_base = adata->acp63_base;
 	runtime->private_data = pdm_data;
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 static int acp63_pdm_dma_hw_params(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#5 lore raw

PATCH v2 3/25 ASoC: amd: ps: Fix acp63_sdw_dma_close() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from acp63_sdw_dma_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes f722917350ee ("ASoC: amd: ps: add SoundWire dma driver dma ops")
---
 sound/soc/amd/ps/ps-sdw-dma.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
sound/soc/amd/ps/ps-sdw-dma.c +7 -1
index f27ebbd21379..c2640985f54c 100644
@@ -337,7 +337,13 @@ static int acp63_sdw_dma_open(struct snd_soc_component *component,
 	stream->stream_id = cpu_dai->id;
 	stream->instance = amd_manager->instance;
 	runtime->private_data = stream;
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 static int acp63_sdw_dma_hw_params(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#6 lore raw

PATCH v2 4/25 ASoC: amd: raven: Fix acp3x_dma_close() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from acp3x_dma_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes 703a6e22888b ("ASoC: amd: Enabling I2S instance in DMA and DAI")
---
 sound/soc/amd/raven/acp3x-pcm-dma.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
sound/soc/amd/raven/acp3x-pcm-dma.c +7 -1
index 37ea5c572eb9..765057499021 100644
@@ -237,7 +237,13 @@ static int acp3x_dma_open(struct snd_soc_component *component,
 
 	i2s_data->acp3x_base = adata->acp3x_base;
 	runtime->private_data = i2s_data;
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 
Signature · 1 line
2.43.0
#8 lore raw

PATCH v2 5/25 ASoC: amd: renoir: Fix acp_pdm_dma_close() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from acp_pdm_dma_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes 4a767b1d039a ("ASoC: amd: add acp3x pdm driver dma ops")
---
 sound/soc/amd/renoir/acp3x-pdm-dma.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
sound/soc/amd/renoir/acp3x-pdm-dma.c +7 -1
index 3f59b753243d..89dc60849c06 100644
@@ -233,7 +233,13 @@ static int acp_pdm_dma_open(struct snd_soc_component *component,
 
 	pdm_data->acp_base = adata->acp_base;
 	runtime->private_data = pdm_data;
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 static int acp_pdm_dma_hw_params(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#9 lore raw

PATCH v2 6/25 ASoC: amd: vangogh: Fix acp5x_dma_close() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from acp5x_dma_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes cab396d8b22c ("ASoC: amd: add ACP5x pcm dma driver ops")
---
 sound/soc/amd/vangogh/acp5x-pcm-dma.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
sound/soc/amd/vangogh/acp5x-pcm-dma.c +7 -1
index 831e30e9b042..72050fa2de80 100644
@@ -231,7 +231,13 @@ static int acp5x_dma_open(struct snd_soc_component *component,
 	}
 	i2s_data->acp5x_base = adata->acp5x_base;
 	runtime->private_data = i2s_data;
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 static int acp5x_dma_hw_params(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#10 lore raw

PATCH v2 7/25 ASoC: amd: yc: Fix acp6x_pdm_dma_close() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from acp6x_pdm_dma_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes ceb4fcc13ae5 ("ASoC: amd: add acp6x pdm driver dma ops")
---
 sound/soc/amd/yc/acp6x-pdm-dma.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
sound/soc/amd/yc/acp6x-pdm-dma.c +7 -1
index 40c4d833f4ed..cba2509faaa1 100644
@@ -209,7 +209,13 @@ static int acp6x_pdm_dma_open(struct snd_soc_component *component,
 
 	pdm_data->acp6x_base = adata->acp6x_base;
 	runtime->private_data = pdm_data;
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 static int acp6x_pdm_dma_hw_params(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#11 lore raw

PATCH v2 8/25 ASoC: apple: mca: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from mca_fe_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as mca_fe_ops has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/apple/mca.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)
sound/soc/apple/mca.c +13 -3
index be702b2942a5..26022a42a95e 100644
@@ -516,6 +516,7 @@ static int mca_fe_startup(struct snd_pcm_substream *substream,
 {
 	struct mca_cluster *cl = mca_dai_to_cluster(dai);
 	unsigned int mask, nchannels;
+	int ret;
 
 	if (cl->tdm_slots) {
 		if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK)
@@ -528,9 +529,18 @@ static int mca_fe_startup(struct snd_pcm_substream *substream,
 		nchannels = 2;
 	}
 
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-					    SNDRV_PCM_HW_PARAM_CHANNELS,
-					    1, nchannels);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_CHANNELS,
+					   1, nchannels);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 static int mca_fe_set_tdm_slot(struct snd_soc_dai *dai, unsigned int tx_mask,
Signature · 1 line
2.43.0
#12 lore raw

PATCH v2 9/25 ASoC: atmel: atmel-pcm-pdc: Fix atmel_pcm_close() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from atmel_pcm_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes 92dfa6198623 ("ASoC: atmel-pcm: split into two file")
---
 sound/soc/atmel/atmel-pcm-pdc.c | 7 +++++++
 1 file changed, 7 insertions(+)
sound/soc/atmel/atmel-pcm-pdc.c +7 -0
index 374f93fa6e0b..4cedf2ad934d 100644
@@ -295,6 +295,13 @@ static int atmel_pcm_open(struct snd_soc_component *component,
 	}
 	runtime->private_data = prtd;
 
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
+
  out:
 	return ret;
 }
Signature · 1 line
2.43.0
#14 lore raw

PATCH v2 10/25 ASoC: codecs: cs42l42: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from cs42l42_dai_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as cs42l42_ops has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/codecs/cs42l42.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)
sound/soc/codecs/cs42l42.c +13 -3
index b5d69d5516a6..18d6cb9bfa2c 100644
@@ -890,6 +890,7 @@ static int cs42l42_dai_startup(struct snd_pcm_substream *substream, struct snd_s
 {
 	struct snd_soc_component *component = dai->component;
 	struct cs42l42_private *cs42l42 = snd_soc_component_get_drvdata(component);
+	int ret;
 
 	/*
 	 * Sample rates < 44.1 kHz would produce an out-of-range SCLK with
@@ -900,9 +901,18 @@ static int cs42l42_dai_startup(struct snd_pcm_substream *substream, struct snd_s
 		return 0;
 
 	/* Machine driver has not set a SCLK, limit bottom end to 44.1 kHz */
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-					    SNDRV_PCM_HW_PARAM_RATE,
-					    44100, 96000);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_RATE,
+					   44100, 96000);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 static int cs42l42_pcm_hw_params(struct snd_pcm_substream *substream,
Signature · 1 line
2.43.0
#15 lore raw

PATCH v2 11/25 ASoC: codecs: nau8325: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from nau8325_dai_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as nau8325_dai_ops has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/codecs/nau8325.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)
sound/soc/codecs/nau8325.c +13 -3
index 2c95ed00a681..5ccf4c110567 100644
@@ -489,14 +489,24 @@ static int nau8325_dai_startup(struct snd_pcm_substream *substream,
 	struct snd_soc_component *component = dai->component;
 	struct nau8325 *nau8325 = snd_soc_component_get_drvdata(component);
 	const struct nau8325_osr_attr *osr;
+	int ret;
 
 	osr = nau8325_get_osr(nau8325);
 	if (!osr || !osr->osr)
 		return -EINVAL;
 
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-					    SNDRV_PCM_HW_PARAM_RATE,
-					    0, CLK_DA_AD_MAX / osr->osr);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_RATE,
+					   0, CLK_DA_AD_MAX / osr->osr);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 static int nau8325_hw_params(struct snd_pcm_substream *substream,
Signature · 1 line
2.43.0
#16 lore raw

PATCH v2 12/25 ASoC: codecs: nau8540: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from nau8540_dai_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as nau8540_dai_ops has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/codecs/nau8540.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)
sound/soc/codecs/nau8540.c +13 -3
index d4211a76b096..09e7f31a904d 100644
@@ -419,14 +419,24 @@ static int nau8540_dai_startup(struct snd_pcm_substream *substream,
 	struct snd_soc_component *component = dai->component;
 	struct nau8540 *nau8540 = snd_soc_component_get_drvdata(component);
 	const struct nau8540_osr_attr *osr;
+	int ret;
 
 	osr = nau8540_get_osr(nau8540);
 	if (!osr || !osr->osr)
 		return -EINVAL;
 
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-					    SNDRV_PCM_HW_PARAM_RATE,
-					    0, CLK_ADC_MAX / osr->osr);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_RATE,
+					   0, CLK_ADC_MAX / osr->osr);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 static int nau8540_hw_params(struct snd_pcm_substream *substream,
Signature · 1 line
2.43.0
#17 lore raw

PATCH v2 13/25 ASoC: codecs: nau8821: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from nau8821_dai_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as nau8821_dai_ops has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/codecs/nau8821.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)
sound/soc/codecs/nau8821.c +13 -3
index ecadbaee1037..7b88ee3159d1 100644
@@ -829,14 +829,24 @@ static int nau8821_dai_startup(struct snd_pcm_substream *substream,
 	struct snd_soc_component *component = dai->component;
 	struct nau8821 *nau8821 = snd_soc_component_get_drvdata(component);
 	const struct nau8821_osr_attr *osr;
+	int ret;
 
 	osr = nau8821_get_osr(nau8821, substream->stream);
 	if (!osr || !osr->osr)
 		return -EINVAL;
 
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-					    SNDRV_PCM_HW_PARAM_RATE,
-					    0, CLK_DA_AD_MAX / osr->osr);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_RATE,
+					   0, CLK_DA_AD_MAX / osr->osr);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 static int nau8821_hw_params(struct snd_pcm_substream *substream,
Signature · 1 line
2.43.0
#18 lore raw

PATCH v2 14/25 ASoC: codecs: nau8824: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from nau8824_dai_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as nau8824_dai_ops has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/codecs/nau8824.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)
sound/soc/codecs/nau8824.c +13 -3
index 853eb9472af9..046cce9192b7 100644
@@ -1061,14 +1061,24 @@ static int nau8824_dai_startup(struct snd_pcm_substream *substream,
 	struct snd_soc_component *component = dai->component;
 	struct nau8824 *nau8824 = snd_soc_component_get_drvdata(component);
 	const struct nau8824_osr_attr *osr;
+	int ret;
 
 	osr = nau8824_get_osr(nau8824, substream->stream);
 	if (!osr || !osr->osr)
 		return -EINVAL;
 
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-					    SNDRV_PCM_HW_PARAM_RATE,
-					    0, CLK_DA_AD_MAX / osr->osr);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_RATE,
+					   0, CLK_DA_AD_MAX / osr->osr);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 static int nau8824_hw_params(struct snd_pcm_substream *substream,
Signature · 1 line
2.43.0
#19 lore raw

PATCH v2 15/25 ASoC: codecs: nau8825: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from nau8825_dai_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as nau8825_dai_ops has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/codecs/nau8825.c | 16 +++++++++++++---
 1 file changed, 13 insertions(+), 3 deletions(-)
sound/soc/codecs/nau8825.c +13 -3
index 63d927f31409..af2a49c893d3 100644
@@ -1261,14 +1261,24 @@ static int nau8825_dai_startup(struct snd_pcm_substream *substream,
 	struct snd_soc_component *component = dai->component;
 	struct nau8825 *nau8825 = snd_soc_component_get_drvdata(component);
 	const struct nau8825_osr_attr *osr;
+	int ret;
 
 	osr = nau8825_get_osr(nau8825, substream->stream);
 	if (!osr || !osr->osr)
 		return -EINVAL;
 
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-					    SNDRV_PCM_HW_PARAM_RATE,
-					    0, CLK_DA_AD_MAX / osr->osr);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_RATE,
+					   0, CLK_DA_AD_MAX / osr->osr);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 static int nau8825_hw_params(struct snd_pcm_substream *substream,
Signature · 1 line
2.43.0
#20 lore raw

PATCH v2 16/25 ASoC: codecs: wm8580: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from wm8580_playback_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as wm8580_dai_ops_playback has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/codecs/wm8580.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
sound/soc/codecs/wm8580.c +13 -2
index 9bc5d1e795dc..cee80da8f753 100644
@@ -852,9 +852,20 @@ static int wm8580_playback_startup(struct snd_pcm_substream *substream,
 {
 	struct snd_soc_component *component = dai->component;
 	struct wm8580_priv *wm8580 = snd_soc_component_get_drvdata(component);
+	int ret;
 
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-		SNDRV_PCM_HW_PARAM_CHANNELS, 1, wm8580->drvdata->num_dacs * 2);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_CHANNELS, 1,
+					   wm8580->drvdata->num_dacs * 2);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 #define WM8580_FORMATS (SNDRV_PCM_FMTBIT_S16_LE | SNDRV_PCM_FMTBIT_S20_3LE |\
Signature · 1 line
2.43.0
#21 lore raw

PATCH v2 17/25 ASoC: codecs: wm8782: Ensure the DAI is marked as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from wm8782_dai_startup(). ASoC only marks the DAI as started when .startup() returns 0.

This currently has no functional impact, as wm8782_dai_ops has no .shutdown() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI is correctly marked as started.

---
 sound/soc/codecs/wm8782.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)
sound/soc/codecs/wm8782.c +11 -1
index 3a2acdfa9b85..cbf0c823c022 100644
@@ -39,9 +39,19 @@ static int wm8782_dai_startup(struct snd_pcm_substream *sub, struct snd_soc_dai
 	struct snd_pcm_runtime *runtime = sub->runtime;
 	struct wm8782_priv *priv =
 		snd_soc_component_get_drvdata(dai->component);
+	int ret;
 
-	return snd_pcm_hw_constraint_minmax(runtime, SNDRV_PCM_HW_PARAM_RATE,
+	ret = snd_pcm_hw_constraint_minmax(runtime, SNDRV_PCM_HW_PARAM_RATE,
 					   8000, priv->max_rate);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI is correctly marked as started.
+	 */
+	return 0;
 }
 
 static const struct snd_soc_dapm_widget wm8782_dapm_widgets[] = {
Signature · 1 line
2.43.0
#22 lore raw

PATCH v2 18/25 ASoC: imx-pcm-rpmsg: Fix imx_rpmsg_pcm_close() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from imx_rpmsg_pcm_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback.

Return 0 explicitly on success.

Fixes 3c00eceb2a53 ("ASoC: imx-pcm-rpmsg: Add platform driver for audio base on rpmsg")
---
 sound/soc/fsl/imx-pcm-rpmsg.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
sound/soc/fsl/imx-pcm-rpmsg.c +7 -1
index 7086fb7fc1eb..85b7c929a28a 100644
@@ -269,7 +269,13 @@ static int imx_rpmsg_pcm_open(struct snd_soc_component *component,
 	info->stream_timer[substream->stream].substream = substream;
 	timer_setup(&info->stream_timer[substream->stream].timer,
 		    imx_rpmsg_timer_callback, 0);
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 static int imx_rpmsg_pcm_close(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#24 lore raw

PATCH v2 19/25 ASoC: intel: atom: Fix .shutdown() being skipped on rollback

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from sst_media_open(). ASoC only marks the DAI as started when .startup() returns 0, so .shutdown() can be skipped on rollback.

Return 0 explicitly on success.

Fixes 4889a8a73f65 ("ASoC: intel: atom: Use __free(kfree) for stream pointer")
Fixes 6cc0f4e63994 ("ASoC: Intel: mfld_pcm: move stream handling to dai_ops")
---
 sound/soc/intel/atom/sst-mfld-platform-pcm.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
sound/soc/intel/atom/sst-mfld-platform-pcm.c +6 -1
index a4c8cbfba096..58681e007abf 100644
@@ -341,7 +341,12 @@ static int sst_media_open(struct snd_pcm_substream *substream,
 
 	stream = NULL;
 
-	return ret_val;
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .shutdown() is not skipped on rollback.
+	 */
+	return 0;
 }
 
 static void sst_media_close(struct snd_pcm_substream *substream,
Signature · 1 line
2.43.0
#27 lore raw

PATCH v2 20/25 ASoC: renesas: fsi: Ensure the component is marked as opened on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from fsi_pcm_open(). ASoC only marks the component as opened when .open() returns 0.

This currently has no functional impact, as fsi_soc_component has no .close() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the component is correctly marked as opened.

---
 sound/soc/renesas/fsi.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
sound/soc/renesas/fsi.c +9 -2
index 23b7e63596d3..74e2ea002fbb 100644
@@ -1811,14 +1811,21 @@ static int fsi_pcm_open(struct snd_soc_component *component,
 			struct snd_pcm_substream *substream)
 {
 	struct snd_pcm_runtime *runtime = substream->runtime;
-	int ret = 0;
+	int ret;
 
 	snd_soc_set_runtime_hwparams(substream, &fsi_pcm_hardware);
 
 	ret = snd_pcm_hw_constraint_integer(runtime,
 					    SNDRV_PCM_HW_PARAM_PERIODS);
+	if (ret < 0)
+		return ret;
 
-	return ret;
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure the component is correctly marked as opened.
+	 */
+	return 0;
 }
 
 static snd_pcm_uframes_t fsi_pointer(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#29 lore raw

PATCH v2 21/25 ASoC: renesas: msiof: Remove redundant period constraint

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() is already called by snd_dmaengine_pcm_open(). Calling it again in msiof_open() is unnecessary, as the constraint is already applied and the second call returns 0.

Remove the redundant constraint call.

---

Changes in v2:
- Remove redundant integer constraints already applied by snd_dmaengine_pcm_open().

 sound/soc/renesas/rcar/msiof.c | 16 ++++++----------
 1 file changed, 6 insertions(+), 10 deletions(-)
sound/soc/renesas/rcar/msiof.c +6 -10
index 128543fc4fc9..11a6f22ec0ad 100644
@@ -419,18 +419,14 @@ static int msiof_open(struct snd_soc_component *component,
 		return PTR_ERR(chan);
 
 	ret = snd_dmaengine_pcm_open(substream, chan);
-	if (ret < 0)
-		goto open_err_dma;
-
-	snd_soc_set_runtime_hwparams(substream, &msiof_pcm_hardware);
-
-	ret = snd_pcm_hw_constraint_integer(substream->runtime, SNDRV_PCM_HW_PARAM_PERIODS);
-
-open_err_dma:
-	if (ret < 0)
+	if (ret < 0) {
 		dma_release_channel(chan);
+		return ret;
+	}
 
-	return ret;
+	snd_soc_set_runtime_hwparams(substream, &msiof_pcm_hardware);
+
+	return 0;
 }
 
 static int msiof_close(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#31 lore raw

PATCH v2 22/25 ASoC: renesas: rz-ssi: Ensure the component is marked as opened on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from rz_ssi_pcm_open_pio(). ASoC only marks the component as opened when .open() returns 0.

This currently has no functional impact, as the component driver has no .close() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the component is correctly marked as opened.

---
 sound/soc/renesas/rz-ssi.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)
sound/soc/renesas/rz-ssi.c +12 -1
index b1587f499cb8..5f4a53886541 100644
@@ -937,10 +937,21 @@ static const struct snd_pcm_hardware rz_ssi_pcm_hardware = {
 static int rz_ssi_pcm_open_pio(struct snd_soc_component *component,
 			       struct snd_pcm_substream *substream)
 {
+	int ret;
+
 	snd_soc_set_runtime_hwparams(substream, &rz_ssi_pcm_hardware);
 
-	return snd_pcm_hw_constraint_integer(substream->runtime,
+	ret = snd_pcm_hw_constraint_integer(substream->runtime,
 					    SNDRV_PCM_HW_PARAM_PERIODS);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure the component is correctly marked as opened.
+	 */
+	return 0;
 }
 
 static int rz_ssi_pcm_open_dma(struct snd_soc_component *component,
Signature · 1 line
2.43.0
#32 lore raw

PATCH v2 23/25 ASoC: renesas: rz-ssi: Remove redundant component open callback

In reply to [email protected]

From: bui duc phuc <[email protected]>

In DMA mode, the generic dmaengine PCM component already applies the integer PERIODS constraint in snd_dmaengine_pcm_open().

Since rz_ssi_pcm_open_dma() only applies this constraint, remove the redundant component .open callback.

---
 sound/soc/renesas/rz-ssi.c | 8 --------
 1 file changed, 8 deletions(-)
sound/soc/renesas/rz-ssi.c +0 -8
index 5f4a53886541..9a0781bb072d 100644
@@ -954,13 +954,6 @@ static int rz_ssi_pcm_open_pio(struct snd_soc_component *component,
 	return 0;
 }
 
-static int rz_ssi_pcm_open_dma(struct snd_soc_component *component,
-			       struct snd_pcm_substream *substream)
-{
-	return snd_pcm_hw_constraint_integer(substream->runtime,
-					     SNDRV_PCM_HW_PARAM_PERIODS);
-}
-
 static snd_pcm_uframes_t rz_ssi_pcm_pointer(struct snd_soc_component *component,
 					    struct snd_pcm_substream *substream)
 {
@@ -1011,7 +1004,6 @@ static const struct snd_soc_component_driver rz_ssi_soc_component_pio = {
 
 static const struct snd_soc_component_driver rz_ssi_soc_component_dma = {
 	.name			= "rz-ssi",
-	.open			= rz_ssi_pcm_open_dma,
 	.legacy_dai_naming	= 1,
 };
 
Signature · 1 line
2.43.0
#33 lore raw

PATCH v2 24/25 ASoC: rockchip: rk3399_gru_sound: Mark the link as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from rockchip_sound_startup(). ASoC only marks the DAI link as started when .startup() returns 0.

This currently has no functional impact, as there is no .shutdown() callback in the relevant snd_soc_ops. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI link is correctly marked as started.

---
 sound/soc/rockchip/rk3399_gru_sound.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)
sound/soc/rockchip/rk3399_gru_sound.c +12 -2
index 04e7bed8275d..d4a0daabfb42 100644
@@ -258,10 +258,20 @@ static int rockchip_sound_dmic_hw_params(struct snd_pcm_substream *substream,
 static int rockchip_sound_startup(struct snd_pcm_substream *substream)
 {
 	struct snd_pcm_runtime *runtime = substream->runtime;
+	int ret;
 
 	runtime->hw.formats = SNDRV_PCM_FMTBIT_S16_LE;
-	return snd_pcm_hw_constraint_minmax(runtime, SNDRV_PCM_HW_PARAM_RATE,
-			8000, 96000);
+	ret = snd_pcm_hw_constraint_minmax(runtime, SNDRV_PCM_HW_PARAM_RATE,
+					   8000, 96000);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI link as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI link is correctly marked as started.
+	 */
+	return 0;
 }
 
 static const struct snd_soc_ops rockchip_sound_max98357a_ops = {
Signature · 1 line
2.43.0
maintainer Acked-by on v2 20/25
#28 lore raw

Re: [PATCH v2 20/25] ASoC: renesas: fsi: Ensure the component is marked as opened on success

In reply to [email protected]

Hi

Quoted 11 lines

snd_pcm_hw_constraint_integer() may return 1, which is then returned from fsi_pcm_open(). ASoC only marks the component as opened when .open() returns 0.

This currently has no functional impact, as fsi_soc_component has no .close() callback. Still, return 0 explicitly on success to match what ASoC expects and ensure the component is correctly marked as opened.

---

Thank you for your help !!

Best regards
---
Kuninori Morimoto

maintainer Acked-by on v2 21/25
#30 lore raw

Re: [PATCH v2 21/25] ASoC: renesas: msiof: Remove redundant period constraint

In reply to [email protected]

Hi

Quoted 9 lines

snd_pcm_hw_constraint_integer() is already called by snd_dmaengine_pcm_open(). Calling it again in msiof_open() is unnecessary, as the constraint is already applied and the second call returns 0.

Remove the redundant constraint call.

---

Thank you for your help !!

Best regards
---
Kuninori Morimoto

maintainer Acked-by on v2 19/25
#25 lore raw

Re: [PATCH v2 19/25] ASoC: intel: atom: Fix .shutdown() being skipped on rollback

In reply to [email protected]
Quoted 11 lines · On 10/15/2026 3:43 AM, [email protected] wrote:

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_integer() may return 1, which is then returned from sst_media_open(). ASoC only marks the DAI as started when .startup() returns 0, so .shutdown() can be skipped on rollback.

Return 0 explicitly on success.

Fixes 4889a8a73f65 ("ASoC: intel: atom: Use __free(kfree) for stream pointer")
Fixes 6cc0f4e63994 ("ASoC: Intel: mfld_pcm: move stream handling to dai_ops")
maintainer
#34 lore raw
In reply to [email protected]
Quoted 4 lines · On Thu, Oct 15, 2026 at 08:42:59AM +0700, [email protected] wrote:

Changes in v2:
- Add Reviewed-by tags from Vijendar Mukunda and Chancel Liu.
- Remove redundant integer constraints already applied by snd_dmaengine_pcm_open().

v1 was applied, please send incremental changes for the differences.

#35 lore raw

Hi Mark,

Quoted 6 lines
Quoted 4 lines

Changes in v2:
- Add Reviewed-by tags from Vijendar Mukunda and Chancel Liu.
- Remove redundant integer constraints already applied by snd_dmaengine_pcm_open().

v1 was applied, please send incremental changes for the differences.

Thank you for applying v1.
I'll prepare and send the incremental changes soon.

Best regards,
Phuc

#36 lore raw

PATCH v2 25/25 ASoC: rockchip: rockchip_max98090: Mark the link as started on success

In reply to [email protected]

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from rk_aif1_startup(). ASoC only marks the DAI link as started when .startup() returns 0.

This currently has no functional impact, as there is no .shutdown() callback in the relevant snd_soc_ops. Still, return 0 explicitly on success to match what ASoC expects and ensure the DAI link is correctly marked as started.

---
 sound/soc/rockchip/rockchip_max98090.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)
sound/soc/rockchip/rockchip_max98090.c +13 -2
index 426506a8e18e..de392b420ebd 100644
@@ -191,12 +191,23 @@ static int rk_aif1_hw_params(struct snd_pcm_substream *substream,
 
 static int rk_aif1_startup(struct snd_pcm_substream *substream)
 {
+	int ret;
 	/*
 	 * Set period size to 240 because pl330 has issue
 	 * dealing with larger period in stress testing.
 	 */
-	return snd_pcm_hw_constraint_minmax(substream->runtime,
-			SNDRV_PCM_HW_PARAM_PERIOD_SIZE, 240, 240);
+	ret = snd_pcm_hw_constraint_minmax(substream->runtime,
+					   SNDRV_PCM_HW_PARAM_PERIOD_SIZE,
+					   240, 240);
+	if (ret < 0)
+		return ret;
+
+	/*
+	 * ASoC only marks the DAI link as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_minmax() may return 1, explicitly
+	 * return 0 to ensure the DAI link is correctly marked as started.
+	 */
+	return 0;
 }
 
 static const struct snd_soc_ops rk_aif1_ops = {
Signature · 1 line
2.43.0
#26 lore raw

Re: [PATCH v2 19/25] ASoC: intel: atom: Fix .shutdown() being skipped on rollback

In reply to [email protected]
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] ALSA: intel: atom: resource leak on sst_media_open() error path --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] ALSA: intel: atom: resource leak on sst_media_open() error path
--

commit fac36234f535190efd5d6e1b1740986dc014a4d6
Author: bui duc phuc <[email protected]>

ASoC: intel: atom: Fix .shutdown() being skipped on rollback

snd_pcm_hw_constraint_integer() may return 1, which is then returned from sst_media_open(). ASoC only marks the DAI as started when .startup() returns 0, so .shutdown() can be skipped on rollback. This commit returns 0 explicitly on success.

Quoted 6 lines
sound/soc/intel/atom/sst-mfld-platform-pcm.c +0 -0
index a4c8cbfba0963..58681e007abf9 100644
@@ -341,7 +341,12 @@ static int sst_media_open(struct snd_pcm_substream *substream,
 

sound/soc/intel/atom/sst-mfld-platform-pcm.c static int sst_media_open(struct snd_pcm_substream *substream, · line 350

Quoted 10 lines
sound/soc/intel/atom/sst-mfld-platform-pcm.c +6 -1
 	stream = NULL;
 
-	return ret_val;
+	/*
+	 * ASoC only marks the DAI as started when .startup() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .shutdown() is not skipped on rollback.
+	 */
+	return 0;
 }

[Severity: High]
Since ASoC only marks the DAI as started when .startup() returns 0, do the error paths earlier in sst_media_open() leak resources?

For example, if power_up_sst() fails:

	ret_val = power_up_sst(stream);
	if (ret_val < 0)
		return ret_val;

The function returns the error directly. Since .startup() returns an error, .shutdown() won't be called. Does this leak the module reference acquired earlier via try_module_get(sst->dev->driver->owner)?

Similarly, if snd_pcm_hw_constraint_integer() fails:

	ret_val = snd_pcm_hw_constraint_integer(runtime,
						SNDRV_PCM_HW_PARAM_PERIODS);

	if (ret_val < 0)
		return ret_val;

The function returns the error, skipping .shutdown(). Does this leak both the module reference and the power state from power_up_sst(stream)?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=19
#23 lore raw

Re: [PATCH v2 18/25] ASoC: imx-pcm-rpmsg: Fix imx_rpmsg_pcm_close() being skipped on rollback

In reply to [email protected]
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Medium] ASoC: fsl: rpmsg: remote state leak on hw_constraint error --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] ASoC: fsl: rpmsg: remote state leak on hw_constraint error
--

commit e67c435f012ba584c1ab6a6bd485da78b988ac05
Author: bui duc phuc <[email protected]>

ASoC: imx-pcm-rpmsg: Fix imx_rpmsg_pcm_close() being skipped on rollback

This commit fixes an issue where the close callback was being skipped when snd_pcm_hw_constraint_integer() returned a positive value. It ensures the component is properly marked as opened by explicitly returning 0 on success.

Quoted 7 lines
sound/soc/fsl/imx-pcm-rpmsg.c +0 -0
index 7086fb7fc1eba..85b7c929a28a4 100644
@@ -269,7 +269,13 @@ static int imx_rpmsg_pcm_open(struct snd_soc_component *component,
 	info->stream_timer[substream->stream].substream = substream;
 	timer_setup(&info->stream_timer[substream->stream].timer,

sound/soc/fsl/imx-pcm-rpmsg.c static int imx_rpmsg_pcm_open(struct snd_soc_component *component, · line 279

Quoted 10 lines
sound/soc/fsl/imx-pcm-rpmsg.c +7 -1
 		    imx_rpmsg_timer_callback, 0);
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }

[Severity: Medium]
If snd_pcm_hw_constraint_integer() returns an error, can this still leave the remote DSP in a leaked state?

Earlier in imx_rpmsg_pcm_open(), an open message is successfully sent to the remote DSP:

sound/soc/fsl/imx-pcm-rpmsg.c:imx_rpmsg_pcm_open() {
	...
	info->send_message(msg, info);
	...
	ret = snd_pcm_hw_constraint_integer(substream->runtime,
					    SNDRV_PCM_HW_PARAM_PERIODS);
	if (ret < 0)
		return ret;
	...
}

Because returning a negative error code prevents ASoC from marking the component as opened, snd_soc_component_close() will skip calling the close callback on error rollback.

Does the driver need to manually send a close message before returning the error, or is there another way to ensure the remote DSP state is cleaned up on this failure path?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=18
#7 lore raw

Re: [PATCH v2 04/25] ASoC: amd: raven: Fix acp3x_dma_close() being skipped on rollback

In reply to [email protected]
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] sound: memory leak of runtime->private_data in acp3x_dma_close() --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] sound: memory leak of runtime->private_data in acp3x_dma_close() --

commit f9bf5629ed83739ae2130cf6fe20e488a93aca99
Author: bui duc phuc <[email protected]>

ASoC: amd: raven: Fix acp3x_dma_close() being skipped on rollback

The commit ensures that acp3x_dma_open() explicitly returns 0 on success. This prevents ASoC from skipping the .close() callback during rollback when snd_pcm_hw_constraint_integer() returns 1.

Quoted 6 lines
sound/soc/amd/raven/acp3x-pcm-dma.c +0 -0
index 37ea5c572eb94..7650574990210 100644
@@ -237,7 +237,13 @@ static int acp3x_dma_open(struct snd_soc_component *component,
 

sound/soc/amd/raven/acp3x-pcm-dma.c static int acp3x_dma_open(struct snd_soc_component *component, · line 246

Quoted 10 lines
sound/soc/amd/raven/acp3x-pcm-dma.c +7 -1
 	i2s_data->acp3x_base = adata->acp3x_base;
 	runtime->private_data = i2s_data;
-	return ret;
+
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;

[Severity: High]
Does this code leak the i2s_data structure stored in runtime->private_data?

While this patch ensures that acp3x_dma_close() is not skipped on rollback, acp3x_dma_close() does not appear to free the memory allocated in acp3x_dma_open():

sound/soc/amd/raven/acp3x-pcm-dma.c:acp3x_dma_close() {
    ...
	ins = substream->runtime->private_data;
	if (!ins)
		return -EINVAL;

	if (substream->stream == SNDRV_PCM_STREAM_PLAYBACK) {
    ...
	}

	return 0;
}

Since there is no kfree(ins) in acp3x_dma_close(), repeatedly opening and closing the stream could cause a memory leak.

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=4
#13 lore raw

Re: [PATCH v2 09/25] ASoC: atmel: atmel-pcm-pdc: Fix atmel_pcm_close() being skipped on rollback

In reply to [email protected]
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] sound: atmel: Use-after-free and NULL deref in atmel_pcm_hw_free race --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] sound: atmel: Use-after-free and NULL deref in atmel_pcm_hw_free race --

commit 04962f72fa0bbcac4e1c99626f3cc0f25a8a7762
Author: bui duc phuc <[email protected]>

ASoC: atmel: atmel-pcm-pdc: Fix atmel_pcm_close() being skipped on rollback

This commit fixes an issue where atmel_pcm_close() might be skipped on rollback. It forces atmel_pcm_open() to explicitly return 0 on success, as snd_pcm_hw_constraint_integer() can return 1, which causes ASoC to skip marking the component as opened.

Quoted 5 lines
sound/soc/atmel/atmel-pcm-pdc.c +0 -0
index 374f93fa6e0b0..4cedf2ad934d6 100644

sound/soc/atmel/atmel-pcm-pdc.c static int atmel_pcm_open(struct snd_soc_component *component, · line 303

Quoted 9 lines
sound/soc/atmel/atmel-pcm-pdc.c +6 -0
@@ -295,6 +295,13 @@ static int atmel_pcm_open(struct snd_soc_component *component,
 	}
 	runtime->private_data = prtd;
 
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;

[Severity: High]
Could this change expose a race condition during teardown? By ensuring atmel_pcm_close() executes, it seems we can now hit a use-after-free or NULL pointer dereference.

When an audio stream is closed, ALSA triggers snd_pcm_drop() followed by trigger(STOP), which disables the PDC but does not disable SSC interrupts globally. This is followed by atmel_pcm_hw_free(), which clears the interrupt handler:

sound/soc/atmel/atmel-pcm-pdc.c:atmel_pcm_hw_free() {
    ...
    prtd->params->dma_intr_handler = NULL;
    ...
}

And finally atmel_pcm_close() is called, which frees the prtd structure:

sound/soc/atmel/atmel-pcm-pdc.c:atmel_pcm_close() {
    ...
    kfree(prtd);
    ...
}

Meanwhile, an interrupt can still fire, and atmel_ssc_interrupt() calls the handler without synchronization:

sound/soc/atmel/atmel_ssc_dai.c:atmel_ssc_interrupt() {
    ...
    dma_params->dma_intr_handler(ssc_sr, dma_params->substream);
    ...
}

Since there is no synchronize_irq() in the PCM teardown path, atmel_ssc_interrupt() can race with atmel_pcm_hw_free() and atmel_pcm_close(). This could lead to a NULL pointer dereference if dma_intr_handler is evaluated after being cleared, or a use-after-free of prtd if kfree(prtd) executes while atmel_pcm_dma_irq() is still actively processing the stream on another CPU.

Quoted 4 lines
sound/soc/atmel/atmel-pcm-pdc.c +1 -0
+
  out:
 	return ret;
 }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=9
#2 lore raw

Re: [PATCH v2 01/25] ASoC: amd: acp: Fix .close() being skipped on rollback

In reply to [email protected]
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] sound/soc/amd: raven: memory leak in acp3x_dma_close() --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] sound/soc/amd: raven: memory leak in acp3x_dma_close()
--

commit 85bba20e9cd8adc1bcd476d2122c1723541d8edd
Author: bui duc phuc <[email protected]>

ASoC: amd: acp: Fix .close() being skipped on rollback

snd_pcm_hw_constraint_integer() may return 1, which is then returned from acp_dma_open(). ASoC only marks the component as opened when .open() returns 0, so .close() can be skipped on rollback. Return 0 explicitly on success.

Quoted 6 lines
sound/soc/amd/acp/acp-platform.c +0 -0
index 6b1e18b31c1c6..f2c8853775ad3 100644
@@ -248,7 +248,12 @@ static int acp_dma_open(struct snd_soc_component *component, struct snd_pcm_subs
 	list_add_tail(&stream->list, &chip->stream_list);

sound/soc/amd/acp/acp-platform.c static int acp_dma_open(struct snd_soc_component *component, struct snd_pcm_subs · line 257

Quoted 10 lines
sound/soc/amd/acp/acp-platform.c +6 -1
 	spin_unlock_irq(&chip->acp_lock);
 
-	return ret;
+	/*
+	 * ASoC only marks the component as opened when .open() returns 0.
+	 * Since snd_pcm_hw_constraint_integer() may return 1, explicitly
+	 * return 0 to ensure .close() is not skipped on rollback.
+	 */
+	return 0;
 }

[Severity: High]
Is there a memory leak in acp3x_dma_close() related to this subsystem?

Looking at sound/soc/amd/raven/acp3x-pcm-dma.c, the stream instance structure is allocated on every stream open:

sound/soc/amd/raven/acp3x-pcm-dma.c:acp3x_dma_open() {
    ...
	i2s_data = kzalloc_obj(*i2s_data);
	if (!i2s_data)
		return -EINVAL;
    ...
}

However, when the stream is closed, acp3x_dma_close() returns without calling kfree() to free the allocated stream instance structure:

sound/soc/amd/raven/acp3x-pcm-dma.c:acp3x_dma_close() {
    ...
	return 0;
}

Could this lead to a memory leak of the i2s_data structure on every stream closure, allowing a user to drain system memory by repeatedly opening and closing the device?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
#3 lore raw

Re: [PATCH v2 01/25] ASoC: amd: acp: Fix .close() being skipped on rollback

Quoted 17 lines

[Severity: High]
Is there a memory leak in acp3x_dma_close() related to this subsystem?

Looking at sound/soc/amd/raven/acp3x-pcm-dma.c, the stream instance structure is allocated on every stream open:

sound/soc/amd/raven/acp3x-pcm-dma.c:acp3x_dma_open() {
    ...
        i2s_data = kzalloc_obj(*i2s_data);
        if (!i2s_data)
                return -EINVAL;
    ...
}

However, when the stream is closed, acp3x_dma_close() returns without calling kfree() to free the allocated stream instance structure:

Quoted 8 lines
sound/soc/amd/raven/acp3x-pcm-dma.c:acp3x_dma_close() {
    ...
        return 0;
}

Could this lead to a memory leak of the i2s_data structure on every stream closure, allowing a user to drain system memory by repeatedly opening and closing the device?

This appears to be a pre-existing issue unrelated to the current patch.

However, after reviewing the code, I agree with Sashiko's finding. I could not find any place where the i2s_data structure is freed after being allocated in .open().

I'll prepare a separate patch to address this issue.

maintainer
#37 lore raw

Re: [PATCH v2 25/25] ASoC: rockchip: rockchip_max98090: Mark the link as started on success

In reply to [email protected]
Quoted 5 lines · On Fri, Oct 16, 2026 at 11:00:48AM +0700, [email protected] wrote:

From: bui duc phuc <[email protected]>

snd_pcm_hw_constraint_minmax() may return 1, which is then returned from rk_aif1_startup(). ASoC only marks the DAI link as started when .startup() returns 0.

Can you please resubmit this as a separate patch as a new thread without the series numbering? This is really going to mess with the tooling :(

Please don't send new patches in reply to old patches or serieses, this makes it harder for both people and tools to understand what is going on - it can bury things in mailboxes and make it difficult to keep track of what current patches are, both for the new patches and the old ones.

Series numbering only makes sense within a series for ordering, if you're sending something separately to a series or a new series then preserving the ordering of the series just causes confusion.

#38 lore raw

Re: [PATCH v2 25/25] ASoC: rockchip: rockchip_max98090: Mark the link as started on success

Hi Mark,

Quoted 11 lines

Can you please resubmit this as a separate patch as a new thread without the series numbering? This is really going to mess with the tooling :(

Please don't send new patches in reply to old patches or serieses, this makes it harder for both people and tools to understand what is going on - it can bury things in mailboxes and make it difficult to keep track of what current patches are, both for the new patches and the old ones.

Series numbering only makes sense within a series for ordering, if you're sending something separately to a series or a new series then preserving the ordering of the series just causes confusion.

Sorry for the inconvenience.

This was patch 25 of v2. I had only managed to send the first 24 patches previously.
Patch 25 failed to send due to an error, so I sent it afterward to complete the series history.

Link to v2:
https://lore.kernel.org/all/[email protected]/

Since you had already applied v1, this patch does not necessarily need to be resubmitted,
so please feel free to ignore it.

Link v1 :
https://lore.kernel.org/all/[email protected]/

I have also sent the incremental changes in a separate thread, as you requested:

https://lore.kernel.org/all/[email protected]/

Best regards,
Phuc