linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v2 0/4] ASoC: hdmi-codec: fixes and improvements
@ 2019-05-06  9:58 Jerome Brunet
  2019-05-06  9:58 ` [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces Jerome Brunet
                   ` (3 more replies)
  0 siblings, 4 replies; 16+ messages in thread
From: Jerome Brunet @ 2019-05-06  9:58 UTC (permalink / raw)
  To: Mark Brown, Liam Girdwood
  Cc: Jerome Brunet, alsa-devel, linux-kernel, patchwork-bot+notify

This patchset is a collection of fixes and improvement on the hdmi
codec driver. It should be completely transparent for the current users
of the codec.

The most important change is removal the current_substream pointer which
allows the codec to be used on codec-to-codec links. I plan to use this
for the HDMI sound support of the Amlogic g12a SoC family.

Changes since v1:
* Drop already applied patches
* Rebase patchset on sound/for-next

Jerome Brunet (4):
  ASoC: hdmi-codec: remove function name debug traces
  ASoC: hdmi-codec: remove reference to the current substream
  ASoC: hdmi-codec: remove reference to the dai drivers in the private
    data
  ASoC: hdmi-codec: remove ops dependency on the dai id

 sound/soc/codecs/hdmi-codec.c | 188 ++++++++++++++++------------------
 1 file changed, 91 insertions(+), 97 deletions(-)

-- 
2.20.1


^ permalink raw reply	[flat|nested] 16+ messages in thread

* [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces
  2019-05-06  9:58 [PATCH v2 0/4] ASoC: hdmi-codec: fixes and improvements Jerome Brunet
@ 2019-05-06  9:58 ` Jerome Brunet
  2019-05-08  6:50   ` Mark Brown
  2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove function name debug traces" to the asoc tree Mark Brown
  2019-05-06  9:58 ` [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream Jerome Brunet
                   ` (2 subsequent siblings)
  3 siblings, 2 replies; 16+ messages in thread
From: Jerome Brunet @ 2019-05-06  9:58 UTC (permalink / raw)
  To: Mark Brown, Liam Girdwood
  Cc: Jerome Brunet, alsa-devel, linux-kernel, patchwork-bot+notify

Remove the debug traces only showing the function name on entry.
The same can be obtained using ftrace.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
---
 sound/soc/codecs/hdmi-codec.c | 12 ------------
 1 file changed, 12 deletions(-)

diff --git a/sound/soc/codecs/hdmi-codec.c b/sound/soc/codecs/hdmi-codec.c
index 39caf19abb0b..eb31d7eddcbf 100644
--- a/sound/soc/codecs/hdmi-codec.c
+++ b/sound/soc/codecs/hdmi-codec.c
@@ -416,8 +416,6 @@ static int hdmi_codec_startup(struct snd_pcm_substream *substream,
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 	int ret = 0;
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	ret = hdmi_codec_new_stream(substream, dai);
 	if (ret)
 		return ret;
@@ -457,8 +455,6 @@ static void hdmi_codec_shutdown(struct snd_pcm_substream *substream,
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	WARN_ON(hcp->current_stream != substream);
 
 	hcp->chmap_idx = HDMI_CODEC_CHMAP_IDX_UNKNOWN;
@@ -527,8 +523,6 @@ static int hdmi_codec_set_fmt(struct snd_soc_dai *dai,
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 	struct hdmi_codec_daifmt cf = { 0 };
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	if (dai->id == DAI_ID_SPDIF)
 		return 0;
 
@@ -597,8 +591,6 @@ static int hdmi_codec_digital_mute(struct snd_soc_dai *dai, int mute)
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	if (hcp->hcd.ops->digital_mute)
 		return hcp->hcd.ops->digital_mute(dai->dev->parent,
 						  hcp->hcd.data, mute);
@@ -656,8 +648,6 @@ static int hdmi_codec_pcm_new(struct snd_soc_pcm_runtime *rtd,
 	};
 	int ret;
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	ret =  snd_pcm_add_chmap_ctls(rtd->pcm, SNDRV_PCM_STREAM_PLAYBACK,
 				      NULL, drv->playback.channels_max, 0,
 				      &hcp->chmap_info);
@@ -754,8 +744,6 @@ static int hdmi_codec_probe(struct platform_device *pdev)
 	int dai_count, i = 0;
 	int ret;
 
-	dev_dbg(dev, "%s()\n", __func__);
-
 	if (!hcd) {
 		dev_err(dev, "%s: No platform data\n", __func__);
 		return -EINVAL;
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream
  2019-05-06  9:58 [PATCH v2 0/4] ASoC: hdmi-codec: fixes and improvements Jerome Brunet
  2019-05-06  9:58 ` [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces Jerome Brunet
@ 2019-05-06  9:58 ` Jerome Brunet
  2019-05-08  7:00   ` Mark Brown
  2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove reference to the current substream" to the asoc tree Mark Brown
  2019-05-06  9:58 ` [PATCH v2 3/4] ASoC: hdmi-codec: remove reference to the dai drivers in the private data Jerome Brunet
  2019-05-06  9:58 ` [PATCH v2 4/4] ASoC: hdmi-codec: remove ops dependency on the dai id Jerome Brunet
  3 siblings, 2 replies; 16+ messages in thread
From: Jerome Brunet @ 2019-05-06  9:58 UTC (permalink / raw)
  To: Mark Brown, Liam Girdwood
  Cc: Jerome Brunet, alsa-devel, linux-kernel, patchwork-bot+notify

If the hdmi-codec is on a codec-to-codec link, the substream pointer
it receives is completely made up by snd_soc_dai_link_event().
The pointer will be different between startup() and shutdown().

The hdmi-codec complains when this happens even if it is not really a
problem. The current_substream pointer is not used for anything useful
apart from getting the exclusive ownership of the device.

Remove current_substream pointer and replace the exclusive locking
mechanism with a simple variable and some atomic operations.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
---
 sound/soc/codecs/hdmi-codec.c | 58 ++++++++++-------------------------
 1 file changed, 16 insertions(+), 42 deletions(-)

diff --git a/sound/soc/codecs/hdmi-codec.c b/sound/soc/codecs/hdmi-codec.c
index eb31d7eddcbf..4d32f93f6be6 100644
--- a/sound/soc/codecs/hdmi-codec.c
+++ b/sound/soc/codecs/hdmi-codec.c
@@ -280,11 +280,10 @@ struct hdmi_codec_priv {
 	struct hdmi_codec_pdata hcd;
 	struct snd_soc_dai_driver *daidrv;
 	struct hdmi_codec_daifmt daifmt[2];
-	struct mutex current_stream_lock;
-	struct snd_pcm_substream *current_stream;
 	uint8_t eld[MAX_ELD_BYTES];
 	struct snd_pcm_chmap *chmap_info;
 	unsigned int chmap_idx;
+	unsigned long busy;
 };
 
 static const struct snd_soc_dapm_widget hdmi_widgets[] = {
@@ -392,42 +391,22 @@ static int hdmi_codec_chmap_ctl_get(struct snd_kcontrol *kcontrol,
 	return 0;
 }
 
-static int hdmi_codec_new_stream(struct snd_pcm_substream *substream,
-				 struct snd_soc_dai *dai)
-{
-	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
-	int ret = 0;
-
-	mutex_lock(&hcp->current_stream_lock);
-	if (!hcp->current_stream) {
-		hcp->current_stream = substream;
-	} else if (hcp->current_stream != substream) {
-		dev_err(dai->dev, "Only one simultaneous stream supported!\n");
-		ret = -EINVAL;
-	}
-	mutex_unlock(&hcp->current_stream_lock);
-
-	return ret;
-}
-
 static int hdmi_codec_startup(struct snd_pcm_substream *substream,
 			      struct snd_soc_dai *dai)
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 	int ret = 0;
 
-	ret = hdmi_codec_new_stream(substream, dai);
-	if (ret)
-		return ret;
+	ret = test_and_set_bit(0, &hcp->busy);
+	if (ret) {
+		dev_err(dai->dev, "Only one simultaneous stream supported!\n");
+		return -EINVAL;
+	}
 
 	if (hcp->hcd.ops->audio_startup) {
 		ret = hcp->hcd.ops->audio_startup(dai->dev->parent, hcp->hcd.data);
-		if (ret) {
-			mutex_lock(&hcp->current_stream_lock);
-			hcp->current_stream = NULL;
-			mutex_unlock(&hcp->current_stream_lock);
-			return ret;
-		}
+		if (ret)
+			goto err;
 	}
 
 	if (hcp->hcd.ops->get_eld) {
@@ -437,17 +416,18 @@ static int hdmi_codec_startup(struct snd_pcm_substream *substream,
 		if (!ret) {
 			ret = snd_pcm_hw_constraint_eld(substream->runtime,
 							hcp->eld);
-			if (ret) {
-				mutex_lock(&hcp->current_stream_lock);
-				hcp->current_stream = NULL;
-				mutex_unlock(&hcp->current_stream_lock);
-				return ret;
-			}
+			if (ret)
+				goto err;
 		}
 		/* Select chmap supported */
 		hdmi_codec_eld_chmap(hcp);
 	}
 	return 0;
+
+err:
+	/* Release the exclusive lock on error */
+	clear_bit(0, &hcp->busy);
+	return ret;
 }
 
 static void hdmi_codec_shutdown(struct snd_pcm_substream *substream,
@@ -455,14 +435,10 @@ static void hdmi_codec_shutdown(struct snd_pcm_substream *substream,
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 
-	WARN_ON(hcp->current_stream != substream);
-
 	hcp->chmap_idx = HDMI_CODEC_CHMAP_IDX_UNKNOWN;
 	hcp->hcd.ops->audio_shutdown(dai->dev->parent, hcp->hcd.data);
 
-	mutex_lock(&hcp->current_stream_lock);
-	hcp->current_stream = NULL;
-	mutex_unlock(&hcp->current_stream_lock);
+	clear_bit(0, &hcp->busy);
 }
 
 static int hdmi_codec_hw_params(struct snd_pcm_substream *substream,
@@ -761,8 +737,6 @@ static int hdmi_codec_probe(struct platform_device *pdev)
 		return -ENOMEM;
 
 	hcp->hcd = *hcd;
-	mutex_init(&hcp->current_stream_lock);
-
 	hcp->daidrv = devm_kcalloc(dev, dai_count, sizeof(*hcp->daidrv),
 				   GFP_KERNEL);
 	if (!hcp->daidrv)
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* [PATCH v2 3/4] ASoC: hdmi-codec: remove reference to the dai drivers in the private data
  2019-05-06  9:58 [PATCH v2 0/4] ASoC: hdmi-codec: fixes and improvements Jerome Brunet
  2019-05-06  9:58 ` [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces Jerome Brunet
  2019-05-06  9:58 ` [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream Jerome Brunet
@ 2019-05-06  9:58 ` Jerome Brunet
  2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove reference to the dai drivers in the private data" to the asoc tree Mark Brown
  2019-05-06  9:58 ` [PATCH v2 4/4] ASoC: hdmi-codec: remove ops dependency on the dai id Jerome Brunet
  3 siblings, 1 reply; 16+ messages in thread
From: Jerome Brunet @ 2019-05-06  9:58 UTC (permalink / raw)
  To: Mark Brown, Liam Girdwood
  Cc: Jerome Brunet, alsa-devel, linux-kernel, patchwork-bot+notify

Keeping the a pointer to the dai drivers is not necessary. It is not used
by the hdmi_codec after the probe.

Even if it was used, the 'struct snd_soc_dai_driver' can accessed through
the 'struct snd_soc_dai' so keeping the pointer in the private data
structure is not useful.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
---
 sound/soc/codecs/hdmi-codec.c | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)

diff --git a/sound/soc/codecs/hdmi-codec.c b/sound/soc/codecs/hdmi-codec.c
index 4d32f93f6be6..9408e6bc4d3e 100644
--- a/sound/soc/codecs/hdmi-codec.c
+++ b/sound/soc/codecs/hdmi-codec.c
@@ -278,7 +278,6 @@ static const struct hdmi_codec_cea_spk_alloc hdmi_codec_channel_alloc[] = {
 
 struct hdmi_codec_priv {
 	struct hdmi_codec_pdata hcd;
-	struct snd_soc_dai_driver *daidrv;
 	struct hdmi_codec_daifmt daifmt[2];
 	uint8_t eld[MAX_ELD_BYTES];
 	struct snd_pcm_chmap *chmap_info;
@@ -715,6 +714,7 @@ static const struct snd_soc_component_driver hdmi_driver = {
 static int hdmi_codec_probe(struct platform_device *pdev)
 {
 	struct hdmi_codec_pdata *hcd = pdev->dev.platform_data;
+	struct snd_soc_dai_driver *daidrv;
 	struct device *dev = &pdev->dev;
 	struct hdmi_codec_priv *hcp;
 	int dai_count, i = 0;
@@ -737,27 +737,25 @@ static int hdmi_codec_probe(struct platform_device *pdev)
 		return -ENOMEM;
 
 	hcp->hcd = *hcd;
-	hcp->daidrv = devm_kcalloc(dev, dai_count, sizeof(*hcp->daidrv),
-				   GFP_KERNEL);
-	if (!hcp->daidrv)
+	daidrv = devm_kcalloc(dev, dai_count, sizeof(*daidrv), GFP_KERNEL);
+	if (!daidrv)
 		return -ENOMEM;
 
 	if (hcd->i2s) {
-		hcp->daidrv[i] = hdmi_i2s_dai;
-		hcp->daidrv[i].playback.channels_max =
-			hcd->max_i2s_channels;
+		daidrv[i] = hdmi_i2s_dai;
+		daidrv[i].playback.channels_max = hcd->max_i2s_channels;
 		i++;
 	}
 
 	if (hcd->spdif) {
-		hcp->daidrv[i] = hdmi_spdif_dai;
+		daidrv[i] = hdmi_spdif_dai;
 		hcp->daifmt[DAI_ID_SPDIF].fmt = HDMI_SPDIF;
 	}
 
 	dev_set_drvdata(dev, hcp);
 
-	ret = devm_snd_soc_register_component(dev, &hdmi_driver, hcp->daidrv,
-				     dai_count);
+	ret = devm_snd_soc_register_component(dev, &hdmi_driver, daidrv,
+					      dai_count);
 	if (ret) {
 		dev_err(dev, "%s: snd_soc_register_component() failed (%d)\n",
 			__func__, ret);
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* [PATCH v2 4/4] ASoC: hdmi-codec: remove ops dependency on the dai id
  2019-05-06  9:58 [PATCH v2 0/4] ASoC: hdmi-codec: fixes and improvements Jerome Brunet
                   ` (2 preceding siblings ...)
  2019-05-06  9:58 ` [PATCH v2 3/4] ASoC: hdmi-codec: remove reference to the dai drivers in the private data Jerome Brunet
@ 2019-05-06  9:58 ` Jerome Brunet
  2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove ops dependency on the dai id" to the asoc tree Mark Brown
  3 siblings, 1 reply; 16+ messages in thread
From: Jerome Brunet @ 2019-05-06  9:58 UTC (permalink / raw)
  To: Mark Brown, Liam Girdwood
  Cc: Jerome Brunet, alsa-devel, linux-kernel, patchwork-bot+notify

The dependency on the dai_id can be removed by setting different ops
for the i2s and spdif dai and storing the dai format information in
each dai structure. It simplies the code a bit.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
---
 sound/soc/codecs/hdmi-codec.c | 100 +++++++++++++++++++++++-----------
 1 file changed, 67 insertions(+), 33 deletions(-)

diff --git a/sound/soc/codecs/hdmi-codec.c b/sound/soc/codecs/hdmi-codec.c
index 9408e6bc4d3e..90a892766625 100644
--- a/sound/soc/codecs/hdmi-codec.c
+++ b/sound/soc/codecs/hdmi-codec.c
@@ -278,7 +278,6 @@ static const struct hdmi_codec_cea_spk_alloc hdmi_codec_channel_alloc[] = {
 
 struct hdmi_codec_priv {
 	struct hdmi_codec_pdata hcd;
-	struct hdmi_codec_daifmt daifmt[2];
 	uint8_t eld[MAX_ELD_BYTES];
 	struct snd_pcm_chmap *chmap_info;
 	unsigned int chmap_idx;
@@ -445,6 +444,7 @@ static int hdmi_codec_hw_params(struct snd_pcm_substream *substream,
 				struct snd_soc_dai *dai)
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
+	struct hdmi_codec_daifmt *cf = dai->playback_dma_data;
 	struct hdmi_codec_params hp = {
 		.iec = {
 			.status = { 0 },
@@ -489,28 +489,27 @@ static int hdmi_codec_hw_params(struct snd_pcm_substream *substream,
 	hp.channels = params_channels(params);
 
 	return hcp->hcd.ops->hw_params(dai->dev->parent, hcp->hcd.data,
-				       &hcp->daifmt[dai->id], &hp);
+				       cf, &hp);
 }
 
-static int hdmi_codec_set_fmt(struct snd_soc_dai *dai,
-			      unsigned int fmt)
+static int hdmi_codec_i2s_set_fmt(struct snd_soc_dai *dai,
+				  unsigned int fmt)
 {
-	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
-	struct hdmi_codec_daifmt cf = { 0 };
+	struct hdmi_codec_daifmt *cf = dai->playback_dma_data;
 
-	if (dai->id == DAI_ID_SPDIF)
-		return 0;
+	/* Reset daifmt */
+	memset(cf, 0, sizeof(*cf));
 
 	switch (fmt & SND_SOC_DAIFMT_MASTER_MASK) {
 	case SND_SOC_DAIFMT_CBM_CFM:
-		cf.bit_clk_master = 1;
-		cf.frame_clk_master = 1;
+		cf->bit_clk_master = 1;
+		cf->frame_clk_master = 1;
 		break;
 	case SND_SOC_DAIFMT_CBS_CFM:
-		cf.frame_clk_master = 1;
+		cf->frame_clk_master = 1;
 		break;
 	case SND_SOC_DAIFMT_CBM_CFS:
-		cf.bit_clk_master = 1;
+		cf->bit_clk_master = 1;
 		break;
 	case SND_SOC_DAIFMT_CBS_CFS:
 		break;
@@ -522,43 +521,41 @@ static int hdmi_codec_set_fmt(struct snd_soc_dai *dai,
 	case SND_SOC_DAIFMT_NB_NF:
 		break;
 	case SND_SOC_DAIFMT_NB_IF:
-		cf.frame_clk_inv = 1;
+		cf->frame_clk_inv = 1;
 		break;
 	case SND_SOC_DAIFMT_IB_NF:
-		cf.bit_clk_inv = 1;
+		cf->bit_clk_inv = 1;
 		break;
 	case SND_SOC_DAIFMT_IB_IF:
-		cf.frame_clk_inv = 1;
-		cf.bit_clk_inv = 1;
+		cf->frame_clk_inv = 1;
+		cf->bit_clk_inv = 1;
 		break;
 	}
 
 	switch (fmt & SND_SOC_DAIFMT_FORMAT_MASK) {
 	case SND_SOC_DAIFMT_I2S:
-		cf.fmt = HDMI_I2S;
+		cf->fmt = HDMI_I2S;
 		break;
 	case SND_SOC_DAIFMT_DSP_A:
-		cf.fmt = HDMI_DSP_A;
+		cf->fmt = HDMI_DSP_A;
 		break;
 	case SND_SOC_DAIFMT_DSP_B:
-		cf.fmt = HDMI_DSP_B;
+		cf->fmt = HDMI_DSP_B;
 		break;
 	case SND_SOC_DAIFMT_RIGHT_J:
-		cf.fmt = HDMI_RIGHT_J;
+		cf->fmt = HDMI_RIGHT_J;
 		break;
 	case SND_SOC_DAIFMT_LEFT_J:
-		cf.fmt = HDMI_LEFT_J;
+		cf->fmt = HDMI_LEFT_J;
 		break;
 	case SND_SOC_DAIFMT_AC97:
-		cf.fmt = HDMI_AC97;
+		cf->fmt = HDMI_AC97;
 		break;
 	default:
 		dev_err(dai->dev, "Invalid DAI interface format\n");
 		return -EINVAL;
 	}
 
-	hcp->daifmt[dai->id] = cf;
-
 	return 0;
 }
 
@@ -573,14 +570,20 @@ static int hdmi_codec_digital_mute(struct snd_soc_dai *dai, int mute)
 	return 0;
 }
 
-static const struct snd_soc_dai_ops hdmi_dai_ops = {
+static const struct snd_soc_dai_ops hdmi_codec_i2s_dai_ops = {
 	.startup	= hdmi_codec_startup,
 	.shutdown	= hdmi_codec_shutdown,
 	.hw_params	= hdmi_codec_hw_params,
-	.set_fmt	= hdmi_codec_set_fmt,
+	.set_fmt	= hdmi_codec_i2s_set_fmt,
 	.digital_mute	= hdmi_codec_digital_mute,
 };
 
+static const struct snd_soc_dai_ops hdmi_codec_spdif_dai_ops = {
+	.startup	= hdmi_codec_startup,
+	.shutdown	= hdmi_codec_shutdown,
+	.hw_params	= hdmi_codec_hw_params,
+	.digital_mute	= hdmi_codec_digital_mute,
+};
 
 #define HDMI_RATES	(SNDRV_PCM_RATE_32000 | SNDRV_PCM_RATE_44100 |\
 			 SNDRV_PCM_RATE_48000 | SNDRV_PCM_RATE_88200 |\
@@ -648,20 +651,52 @@ static int hdmi_codec_pcm_new(struct snd_soc_pcm_runtime *rtd,
 static int hdmi_dai_probe(struct snd_soc_dai *dai)
 {
 	struct snd_soc_dapm_context *dapm;
+	struct hdmi_codec_daifmt *daifmt;
 	struct snd_soc_dapm_route route = {
 		.sink = "TX",
 		.source = dai->driver->playback.stream_name,
 	};
+	int ret;
 
 	dapm = snd_soc_component_get_dapm(dai->component);
+	ret = snd_soc_dapm_add_routes(dapm, &route, 1);
+	if (ret)
+		return ret;
+
+	daifmt = kzalloc(sizeof(*daifmt), GFP_KERNEL);
+	if (!daifmt)
+		return -ENOMEM;
 
-	return snd_soc_dapm_add_routes(dapm, &route, 1);
+	dai->playback_dma_data = daifmt;
+	return 0;
+}
+
+static int hdmi_dai_spdif_probe(struct snd_soc_dai *dai)
+{
+	struct hdmi_codec_daifmt *cf = dai->playback_dma_data;
+	int ret;
+
+	ret = hdmi_dai_probe(dai);
+	if (ret)
+		return ret;
+
+	cf = dai->playback_dma_data;
+	cf->fmt = HDMI_SPDIF;
+
+	return 0;
+}
+
+static int hdmi_codec_dai_remove(struct snd_soc_dai *dai)
+{
+	kfree(dai->playback_dma_data);
+	return 0;
 }
 
 static const struct snd_soc_dai_driver hdmi_i2s_dai = {
 	.name = "i2s-hifi",
 	.id = DAI_ID_I2S,
 	.probe = hdmi_dai_probe,
+	.remove = hdmi_codec_dai_remove,
 	.playback = {
 		.stream_name = "I2S Playback",
 		.channels_min = 2,
@@ -670,14 +705,15 @@ static const struct snd_soc_dai_driver hdmi_i2s_dai = {
 		.formats = I2S_FORMATS,
 		.sig_bits = 24,
 	},
-	.ops = &hdmi_dai_ops,
+	.ops = &hdmi_codec_i2s_dai_ops,
 	.pcm_new = hdmi_codec_pcm_new,
 };
 
 static const struct snd_soc_dai_driver hdmi_spdif_dai = {
 	.name = "spdif-hifi",
 	.id = DAI_ID_SPDIF,
-	.probe = hdmi_dai_probe,
+	.probe = hdmi_dai_spdif_probe,
+	.remove = hdmi_codec_dai_remove,
 	.playback = {
 		.stream_name = "SPDIF Playback",
 		.channels_min = 2,
@@ -685,7 +721,7 @@ static const struct snd_soc_dai_driver hdmi_spdif_dai = {
 		.rates = HDMI_RATES,
 		.formats = SPDIF_FORMATS,
 	},
-	.ops = &hdmi_dai_ops,
+	.ops = &hdmi_codec_spdif_dai_ops,
 	.pcm_new = hdmi_codec_pcm_new,
 };
 
@@ -747,10 +783,8 @@ static int hdmi_codec_probe(struct platform_device *pdev)
 		i++;
 	}
 
-	if (hcd->spdif) {
+	if (hcd->spdif)
 		daidrv[i] = hdmi_spdif_dai;
-		hcp->daifmt[DAI_ID_SPDIF].fmt = HDMI_SPDIF;
-	}
 
 	dev_set_drvdata(dev, hcp);
 
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* Re: [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces
  2019-05-06  9:58 ` [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces Jerome Brunet
@ 2019-05-08  6:50   ` Mark Brown
  2019-05-08  8:08     ` Jerome Brunet
  2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove function name debug traces" to the asoc tree Mark Brown
  1 sibling, 1 reply; 16+ messages in thread
From: Mark Brown @ 2019-05-08  6:50 UTC (permalink / raw)
  To: Jerome Brunet
  Cc: Liam Girdwood, alsa-devel, linux-kernel, patchwork-bot+notify

[-- Attachment #1: Type: text/plain, Size: 339 bytes --]

On Mon, May 06, 2019 at 11:58:12AM +0200, Jerome Brunet wrote:
> Remove the debug traces only showing the function name on entry.
> The same can be obtained using ftrace.

This is not a bug fix and so shouldn't be the first patch in the series
in order to avoid dependencies from it on anything later in the series
that is actually a fix.

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

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream
  2019-05-06  9:58 ` [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream Jerome Brunet
@ 2019-05-08  7:00   ` Mark Brown
  2019-05-08  8:08     ` Jerome Brunet
  2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove reference to the current substream" to the asoc tree Mark Brown
  1 sibling, 1 reply; 16+ messages in thread
From: Mark Brown @ 2019-05-08  7:00 UTC (permalink / raw)
  To: Jerome Brunet
  Cc: Liam Girdwood, alsa-devel, linux-kernel, patchwork-bot+notify

[-- Attachment #1: Type: text/plain, Size: 437 bytes --]

On Mon, May 06, 2019 at 11:58:13AM +0200, Jerome Brunet wrote:

> Remove current_substream pointer and replace the exclusive locking
> mechanism with a simple variable and some atomic operations.

The advantage of mutexes is that they are very simple and clear to
reason about.  It is therefore unclear that this conversion to use
atomic variables is an improvement, they're generally more complex 
to reason about and fragile.

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

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream
  2019-05-08  7:00   ` Mark Brown
@ 2019-05-08  8:08     ` Jerome Brunet
  2019-05-08  8:57       ` Mark Brown
  0 siblings, 1 reply; 16+ messages in thread
From: Jerome Brunet @ 2019-05-08  8:08 UTC (permalink / raw)
  To: Mark Brown; +Cc: Liam Girdwood, alsa-devel, linux-kernel, patchwork-bot+notify

On Wed, 2019-05-08 at 16:00 +0900, Mark Brown wrote:
> On Mon, May 06, 2019 at 11:58:13AM +0200, Jerome Brunet wrote:
> 
> > Remove current_substream pointer and replace the exclusive locking
> > mechanism with a simple variable and some atomic operations.
> 
> The advantage of mutexes is that they are very simple and clear to
> reason about.  It is therefore unclear that this conversion to use
> atomic variables is an improvement, they're generally more complex 
> to reason about and fragile.

The point of this patch is not to remove the mutex in favor of atomic
operations

The point was to remove the current_substream pointer in favor of a
simple 'busy' flag. After that, I ended up with a mutex protecting
a flag and it seemed a bit overkill to me. Atomic op seemed like a
good fit for this but I don't really care about that particular
point.

I can put back mutex to protect the flag if you prefer.

Another solution would be to use the mutex as the 'busy' flag.
Get the ownership of the hdmi using try_lock() in startup() and
release it in shutdown()

Would you have a preference Mark ? 



^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces
  2019-05-08  6:50   ` Mark Brown
@ 2019-05-08  8:08     ` Jerome Brunet
  0 siblings, 0 replies; 16+ messages in thread
From: Jerome Brunet @ 2019-05-08  8:08 UTC (permalink / raw)
  To: Mark Brown; +Cc: Liam Girdwood, alsa-devel, linux-kernel, patchwork-bot+notify

On Wed, 2019-05-08 at 15:50 +0900, Mark Brown wrote:
> On Mon, May 06, 2019 at 11:58:12AM +0200, Jerome Brunet wrote:
> > Remove the debug traces only showing the function name on entry.
> > The same can be obtained using ftrace.
> 
> This is not a bug fix and so shouldn't be the first patch in the series
> in order to avoid dependencies from it on anything later in the series
> that is actually a fix.

Sure, I'll re-order



^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream
  2019-05-08  8:08     ` Jerome Brunet
@ 2019-05-08  8:57       ` Mark Brown
  2019-05-09  8:11         ` Jerome Brunet
  0 siblings, 1 reply; 16+ messages in thread
From: Mark Brown @ 2019-05-08  8:57 UTC (permalink / raw)
  To: Jerome Brunet
  Cc: Liam Girdwood, alsa-devel, linux-kernel, patchwork-bot+notify

[-- Attachment #1: Type: text/plain, Size: 949 bytes --]

On Wed, May 08, 2019 at 10:08:44AM +0200, Jerome Brunet wrote:
> On Wed, 2019-05-08 at 16:00 +0900, Mark Brown wrote:

> > The advantage of mutexes is that they are very simple and clear to
> > reason about.  It is therefore unclear that this conversion to use
> > atomic variables is an improvement, they're generally more complex 
> > to reason about and fragile.

> The point of this patch is not to remove the mutex in favor of atomic
> operations

Sure, but you mixed in a conversion to atomic operations as well.

> I can put back mutex to protect the flag if you prefer.

> Another solution would be to use the mutex as the 'busy' flag.
> Get the ownership of the hdmi using try_lock() in startup() and
> release it in shutdown()

> Would you have a preference Mark ? 

Probably using a mutex for the flag is clearer.  I'd rather keep things
as simple as absolutely possible to avoid people making mistakes in
future.

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

^ permalink raw reply	[flat|nested] 16+ messages in thread

* Applied "ASoC: hdmi-codec: remove ops dependency on the dai id" to the asoc tree
  2019-05-06  9:58 ` [PATCH v2 4/4] ASoC: hdmi-codec: remove ops dependency on the dai id Jerome Brunet
@ 2019-05-08  9:01   ` Mark Brown
  0 siblings, 0 replies; 16+ messages in thread
From: Mark Brown @ 2019-05-08  9:01 UTC (permalink / raw)
  To: Jerome Brunet
  Cc: alsa-devel, Liam Girdwood, linux-kernel, Mark Brown,
	patchwork-bot+notify

The patch

   ASoC: hdmi-codec: remove ops dependency on the dai id

has been applied to the asoc tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From 0cf4610b9f297e570da4d98514b310f076ecc8ab Mon Sep 17 00:00:00 2001
From: Jerome Brunet <jbrunet@baylibre.com>
Date: Mon, 6 May 2019 11:58:15 +0200
Subject: [PATCH] ASoC: hdmi-codec: remove ops dependency on the dai id

The dependency on the dai_id can be removed by setting different ops
for the i2s and spdif dai and storing the dai format information in
each dai structure. It simplies the code a bit.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 sound/soc/codecs/hdmi-codec.c | 100 +++++++++++++++++++++++-----------
 1 file changed, 67 insertions(+), 33 deletions(-)

diff --git a/sound/soc/codecs/hdmi-codec.c b/sound/soc/codecs/hdmi-codec.c
index 9408e6bc4d3e..90a892766625 100644
--- a/sound/soc/codecs/hdmi-codec.c
+++ b/sound/soc/codecs/hdmi-codec.c
@@ -278,7 +278,6 @@ static const struct hdmi_codec_cea_spk_alloc hdmi_codec_channel_alloc[] = {
 
 struct hdmi_codec_priv {
 	struct hdmi_codec_pdata hcd;
-	struct hdmi_codec_daifmt daifmt[2];
 	uint8_t eld[MAX_ELD_BYTES];
 	struct snd_pcm_chmap *chmap_info;
 	unsigned int chmap_idx;
@@ -445,6 +444,7 @@ static int hdmi_codec_hw_params(struct snd_pcm_substream *substream,
 				struct snd_soc_dai *dai)
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
+	struct hdmi_codec_daifmt *cf = dai->playback_dma_data;
 	struct hdmi_codec_params hp = {
 		.iec = {
 			.status = { 0 },
@@ -489,28 +489,27 @@ static int hdmi_codec_hw_params(struct snd_pcm_substream *substream,
 	hp.channels = params_channels(params);
 
 	return hcp->hcd.ops->hw_params(dai->dev->parent, hcp->hcd.data,
-				       &hcp->daifmt[dai->id], &hp);
+				       cf, &hp);
 }
 
-static int hdmi_codec_set_fmt(struct snd_soc_dai *dai,
-			      unsigned int fmt)
+static int hdmi_codec_i2s_set_fmt(struct snd_soc_dai *dai,
+				  unsigned int fmt)
 {
-	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
-	struct hdmi_codec_daifmt cf = { 0 };
+	struct hdmi_codec_daifmt *cf = dai->playback_dma_data;
 
-	if (dai->id == DAI_ID_SPDIF)
-		return 0;
+	/* Reset daifmt */
+	memset(cf, 0, sizeof(*cf));
 
 	switch (fmt & SND_SOC_DAIFMT_MASTER_MASK) {
 	case SND_SOC_DAIFMT_CBM_CFM:
-		cf.bit_clk_master = 1;
-		cf.frame_clk_master = 1;
+		cf->bit_clk_master = 1;
+		cf->frame_clk_master = 1;
 		break;
 	case SND_SOC_DAIFMT_CBS_CFM:
-		cf.frame_clk_master = 1;
+		cf->frame_clk_master = 1;
 		break;
 	case SND_SOC_DAIFMT_CBM_CFS:
-		cf.bit_clk_master = 1;
+		cf->bit_clk_master = 1;
 		break;
 	case SND_SOC_DAIFMT_CBS_CFS:
 		break;
@@ -522,43 +521,41 @@ static int hdmi_codec_set_fmt(struct snd_soc_dai *dai,
 	case SND_SOC_DAIFMT_NB_NF:
 		break;
 	case SND_SOC_DAIFMT_NB_IF:
-		cf.frame_clk_inv = 1;
+		cf->frame_clk_inv = 1;
 		break;
 	case SND_SOC_DAIFMT_IB_NF:
-		cf.bit_clk_inv = 1;
+		cf->bit_clk_inv = 1;
 		break;
 	case SND_SOC_DAIFMT_IB_IF:
-		cf.frame_clk_inv = 1;
-		cf.bit_clk_inv = 1;
+		cf->frame_clk_inv = 1;
+		cf->bit_clk_inv = 1;
 		break;
 	}
 
 	switch (fmt & SND_SOC_DAIFMT_FORMAT_MASK) {
 	case SND_SOC_DAIFMT_I2S:
-		cf.fmt = HDMI_I2S;
+		cf->fmt = HDMI_I2S;
 		break;
 	case SND_SOC_DAIFMT_DSP_A:
-		cf.fmt = HDMI_DSP_A;
+		cf->fmt = HDMI_DSP_A;
 		break;
 	case SND_SOC_DAIFMT_DSP_B:
-		cf.fmt = HDMI_DSP_B;
+		cf->fmt = HDMI_DSP_B;
 		break;
 	case SND_SOC_DAIFMT_RIGHT_J:
-		cf.fmt = HDMI_RIGHT_J;
+		cf->fmt = HDMI_RIGHT_J;
 		break;
 	case SND_SOC_DAIFMT_LEFT_J:
-		cf.fmt = HDMI_LEFT_J;
+		cf->fmt = HDMI_LEFT_J;
 		break;
 	case SND_SOC_DAIFMT_AC97:
-		cf.fmt = HDMI_AC97;
+		cf->fmt = HDMI_AC97;
 		break;
 	default:
 		dev_err(dai->dev, "Invalid DAI interface format\n");
 		return -EINVAL;
 	}
 
-	hcp->daifmt[dai->id] = cf;
-
 	return 0;
 }
 
@@ -573,14 +570,20 @@ static int hdmi_codec_digital_mute(struct snd_soc_dai *dai, int mute)
 	return 0;
 }
 
-static const struct snd_soc_dai_ops hdmi_dai_ops = {
+static const struct snd_soc_dai_ops hdmi_codec_i2s_dai_ops = {
 	.startup	= hdmi_codec_startup,
 	.shutdown	= hdmi_codec_shutdown,
 	.hw_params	= hdmi_codec_hw_params,
-	.set_fmt	= hdmi_codec_set_fmt,
+	.set_fmt	= hdmi_codec_i2s_set_fmt,
 	.digital_mute	= hdmi_codec_digital_mute,
 };
 
+static const struct snd_soc_dai_ops hdmi_codec_spdif_dai_ops = {
+	.startup	= hdmi_codec_startup,
+	.shutdown	= hdmi_codec_shutdown,
+	.hw_params	= hdmi_codec_hw_params,
+	.digital_mute	= hdmi_codec_digital_mute,
+};
 
 #define HDMI_RATES	(SNDRV_PCM_RATE_32000 | SNDRV_PCM_RATE_44100 |\
 			 SNDRV_PCM_RATE_48000 | SNDRV_PCM_RATE_88200 |\
@@ -648,20 +651,52 @@ static int hdmi_codec_pcm_new(struct snd_soc_pcm_runtime *rtd,
 static int hdmi_dai_probe(struct snd_soc_dai *dai)
 {
 	struct snd_soc_dapm_context *dapm;
+	struct hdmi_codec_daifmt *daifmt;
 	struct snd_soc_dapm_route route = {
 		.sink = "TX",
 		.source = dai->driver->playback.stream_name,
 	};
+	int ret;
 
 	dapm = snd_soc_component_get_dapm(dai->component);
+	ret = snd_soc_dapm_add_routes(dapm, &route, 1);
+	if (ret)
+		return ret;
+
+	daifmt = kzalloc(sizeof(*daifmt), GFP_KERNEL);
+	if (!daifmt)
+		return -ENOMEM;
 
-	return snd_soc_dapm_add_routes(dapm, &route, 1);
+	dai->playback_dma_data = daifmt;
+	return 0;
+}
+
+static int hdmi_dai_spdif_probe(struct snd_soc_dai *dai)
+{
+	struct hdmi_codec_daifmt *cf = dai->playback_dma_data;
+	int ret;
+
+	ret = hdmi_dai_probe(dai);
+	if (ret)
+		return ret;
+
+	cf = dai->playback_dma_data;
+	cf->fmt = HDMI_SPDIF;
+
+	return 0;
+}
+
+static int hdmi_codec_dai_remove(struct snd_soc_dai *dai)
+{
+	kfree(dai->playback_dma_data);
+	return 0;
 }
 
 static const struct snd_soc_dai_driver hdmi_i2s_dai = {
 	.name = "i2s-hifi",
 	.id = DAI_ID_I2S,
 	.probe = hdmi_dai_probe,
+	.remove = hdmi_codec_dai_remove,
 	.playback = {
 		.stream_name = "I2S Playback",
 		.channels_min = 2,
@@ -670,14 +705,15 @@ static const struct snd_soc_dai_driver hdmi_i2s_dai = {
 		.formats = I2S_FORMATS,
 		.sig_bits = 24,
 	},
-	.ops = &hdmi_dai_ops,
+	.ops = &hdmi_codec_i2s_dai_ops,
 	.pcm_new = hdmi_codec_pcm_new,
 };
 
 static const struct snd_soc_dai_driver hdmi_spdif_dai = {
 	.name = "spdif-hifi",
 	.id = DAI_ID_SPDIF,
-	.probe = hdmi_dai_probe,
+	.probe = hdmi_dai_spdif_probe,
+	.remove = hdmi_codec_dai_remove,
 	.playback = {
 		.stream_name = "SPDIF Playback",
 		.channels_min = 2,
@@ -685,7 +721,7 @@ static const struct snd_soc_dai_driver hdmi_spdif_dai = {
 		.rates = HDMI_RATES,
 		.formats = SPDIF_FORMATS,
 	},
-	.ops = &hdmi_dai_ops,
+	.ops = &hdmi_codec_spdif_dai_ops,
 	.pcm_new = hdmi_codec_pcm_new,
 };
 
@@ -747,10 +783,8 @@ static int hdmi_codec_probe(struct platform_device *pdev)
 		i++;
 	}
 
-	if (hcd->spdif) {
+	if (hcd->spdif)
 		daidrv[i] = hdmi_spdif_dai;
-		hcp->daifmt[DAI_ID_SPDIF].fmt = HDMI_SPDIF;
-	}
 
 	dev_set_drvdata(dev, hcp);
 
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* Applied "ASoC: hdmi-codec: remove reference to the dai drivers in the private data" to the asoc tree
  2019-05-06  9:58 ` [PATCH v2 3/4] ASoC: hdmi-codec: remove reference to the dai drivers in the private data Jerome Brunet
@ 2019-05-08  9:01   ` Mark Brown
  0 siblings, 0 replies; 16+ messages in thread
From: Mark Brown @ 2019-05-08  9:01 UTC (permalink / raw)
  To: Jerome Brunet
  Cc: alsa-devel, Liam Girdwood, linux-kernel, Mark Brown,
	patchwork-bot+notify

The patch

   ASoC: hdmi-codec: remove reference to the dai drivers in the private data

has been applied to the asoc tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From 1de005d47d90343666c5cc50a50929e05e52baac Mon Sep 17 00:00:00 2001
From: Jerome Brunet <jbrunet@baylibre.com>
Date: Mon, 6 May 2019 11:58:14 +0200
Subject: [PATCH] ASoC: hdmi-codec: remove reference to the dai drivers in the
 private data

Keeping the a pointer to the dai drivers is not necessary. It is not used
by the hdmi_codec after the probe.

Even if it was used, the 'struct snd_soc_dai_driver' can accessed through
the 'struct snd_soc_dai' so keeping the pointer in the private data
structure is not useful.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 sound/soc/codecs/hdmi-codec.c | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)

diff --git a/sound/soc/codecs/hdmi-codec.c b/sound/soc/codecs/hdmi-codec.c
index 4d32f93f6be6..9408e6bc4d3e 100644
--- a/sound/soc/codecs/hdmi-codec.c
+++ b/sound/soc/codecs/hdmi-codec.c
@@ -278,7 +278,6 @@ static const struct hdmi_codec_cea_spk_alloc hdmi_codec_channel_alloc[] = {
 
 struct hdmi_codec_priv {
 	struct hdmi_codec_pdata hcd;
-	struct snd_soc_dai_driver *daidrv;
 	struct hdmi_codec_daifmt daifmt[2];
 	uint8_t eld[MAX_ELD_BYTES];
 	struct snd_pcm_chmap *chmap_info;
@@ -715,6 +714,7 @@ static const struct snd_soc_component_driver hdmi_driver = {
 static int hdmi_codec_probe(struct platform_device *pdev)
 {
 	struct hdmi_codec_pdata *hcd = pdev->dev.platform_data;
+	struct snd_soc_dai_driver *daidrv;
 	struct device *dev = &pdev->dev;
 	struct hdmi_codec_priv *hcp;
 	int dai_count, i = 0;
@@ -737,27 +737,25 @@ static int hdmi_codec_probe(struct platform_device *pdev)
 		return -ENOMEM;
 
 	hcp->hcd = *hcd;
-	hcp->daidrv = devm_kcalloc(dev, dai_count, sizeof(*hcp->daidrv),
-				   GFP_KERNEL);
-	if (!hcp->daidrv)
+	daidrv = devm_kcalloc(dev, dai_count, sizeof(*daidrv), GFP_KERNEL);
+	if (!daidrv)
 		return -ENOMEM;
 
 	if (hcd->i2s) {
-		hcp->daidrv[i] = hdmi_i2s_dai;
-		hcp->daidrv[i].playback.channels_max =
-			hcd->max_i2s_channels;
+		daidrv[i] = hdmi_i2s_dai;
+		daidrv[i].playback.channels_max = hcd->max_i2s_channels;
 		i++;
 	}
 
 	if (hcd->spdif) {
-		hcp->daidrv[i] = hdmi_spdif_dai;
+		daidrv[i] = hdmi_spdif_dai;
 		hcp->daifmt[DAI_ID_SPDIF].fmt = HDMI_SPDIF;
 	}
 
 	dev_set_drvdata(dev, hcp);
 
-	ret = devm_snd_soc_register_component(dev, &hdmi_driver, hcp->daidrv,
-				     dai_count);
+	ret = devm_snd_soc_register_component(dev, &hdmi_driver, daidrv,
+					      dai_count);
 	if (ret) {
 		dev_err(dev, "%s: snd_soc_register_component() failed (%d)\n",
 			__func__, ret);
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* Applied "ASoC: hdmi-codec: remove reference to the current substream" to the asoc tree
  2019-05-06  9:58 ` [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream Jerome Brunet
  2019-05-08  7:00   ` Mark Brown
@ 2019-05-08  9:01   ` Mark Brown
  1 sibling, 0 replies; 16+ messages in thread
From: Mark Brown @ 2019-05-08  9:01 UTC (permalink / raw)
  To: Jerome Brunet
  Cc: alsa-devel, Liam Girdwood, linux-kernel, Mark Brown,
	patchwork-bot+notify

The patch

   ASoC: hdmi-codec: remove reference to the current substream

has been applied to the asoc tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From 3fcf94ef4d418668fa66e33ce9aabb05689b55f6 Mon Sep 17 00:00:00 2001
From: Jerome Brunet <jbrunet@baylibre.com>
Date: Mon, 6 May 2019 11:58:13 +0200
Subject: [PATCH] ASoC: hdmi-codec: remove reference to the current substream

If the hdmi-codec is on a codec-to-codec link, the substream pointer
it receives is completely made up by snd_soc_dai_link_event().
The pointer will be different between startup() and shutdown().

The hdmi-codec complains when this happens even if it is not really a
problem. The current_substream pointer is not used for anything useful
apart from getting the exclusive ownership of the device.

Remove current_substream pointer and replace the exclusive locking
mechanism with a simple variable and some atomic operations.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 sound/soc/codecs/hdmi-codec.c | 58 ++++++++++-------------------------
 1 file changed, 16 insertions(+), 42 deletions(-)

diff --git a/sound/soc/codecs/hdmi-codec.c b/sound/soc/codecs/hdmi-codec.c
index eb31d7eddcbf..4d32f93f6be6 100644
--- a/sound/soc/codecs/hdmi-codec.c
+++ b/sound/soc/codecs/hdmi-codec.c
@@ -280,11 +280,10 @@ struct hdmi_codec_priv {
 	struct hdmi_codec_pdata hcd;
 	struct snd_soc_dai_driver *daidrv;
 	struct hdmi_codec_daifmt daifmt[2];
-	struct mutex current_stream_lock;
-	struct snd_pcm_substream *current_stream;
 	uint8_t eld[MAX_ELD_BYTES];
 	struct snd_pcm_chmap *chmap_info;
 	unsigned int chmap_idx;
+	unsigned long busy;
 };
 
 static const struct snd_soc_dapm_widget hdmi_widgets[] = {
@@ -392,42 +391,22 @@ static int hdmi_codec_chmap_ctl_get(struct snd_kcontrol *kcontrol,
 	return 0;
 }
 
-static int hdmi_codec_new_stream(struct snd_pcm_substream *substream,
-				 struct snd_soc_dai *dai)
-{
-	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
-	int ret = 0;
-
-	mutex_lock(&hcp->current_stream_lock);
-	if (!hcp->current_stream) {
-		hcp->current_stream = substream;
-	} else if (hcp->current_stream != substream) {
-		dev_err(dai->dev, "Only one simultaneous stream supported!\n");
-		ret = -EINVAL;
-	}
-	mutex_unlock(&hcp->current_stream_lock);
-
-	return ret;
-}
-
 static int hdmi_codec_startup(struct snd_pcm_substream *substream,
 			      struct snd_soc_dai *dai)
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 	int ret = 0;
 
-	ret = hdmi_codec_new_stream(substream, dai);
-	if (ret)
-		return ret;
+	ret = test_and_set_bit(0, &hcp->busy);
+	if (ret) {
+		dev_err(dai->dev, "Only one simultaneous stream supported!\n");
+		return -EINVAL;
+	}
 
 	if (hcp->hcd.ops->audio_startup) {
 		ret = hcp->hcd.ops->audio_startup(dai->dev->parent, hcp->hcd.data);
-		if (ret) {
-			mutex_lock(&hcp->current_stream_lock);
-			hcp->current_stream = NULL;
-			mutex_unlock(&hcp->current_stream_lock);
-			return ret;
-		}
+		if (ret)
+			goto err;
 	}
 
 	if (hcp->hcd.ops->get_eld) {
@@ -437,17 +416,18 @@ static int hdmi_codec_startup(struct snd_pcm_substream *substream,
 		if (!ret) {
 			ret = snd_pcm_hw_constraint_eld(substream->runtime,
 							hcp->eld);
-			if (ret) {
-				mutex_lock(&hcp->current_stream_lock);
-				hcp->current_stream = NULL;
-				mutex_unlock(&hcp->current_stream_lock);
-				return ret;
-			}
+			if (ret)
+				goto err;
 		}
 		/* Select chmap supported */
 		hdmi_codec_eld_chmap(hcp);
 	}
 	return 0;
+
+err:
+	/* Release the exclusive lock on error */
+	clear_bit(0, &hcp->busy);
+	return ret;
 }
 
 static void hdmi_codec_shutdown(struct snd_pcm_substream *substream,
@@ -455,14 +435,10 @@ static void hdmi_codec_shutdown(struct snd_pcm_substream *substream,
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 
-	WARN_ON(hcp->current_stream != substream);
-
 	hcp->chmap_idx = HDMI_CODEC_CHMAP_IDX_UNKNOWN;
 	hcp->hcd.ops->audio_shutdown(dai->dev->parent, hcp->hcd.data);
 
-	mutex_lock(&hcp->current_stream_lock);
-	hcp->current_stream = NULL;
-	mutex_unlock(&hcp->current_stream_lock);
+	clear_bit(0, &hcp->busy);
 }
 
 static int hdmi_codec_hw_params(struct snd_pcm_substream *substream,
@@ -761,8 +737,6 @@ static int hdmi_codec_probe(struct platform_device *pdev)
 		return -ENOMEM;
 
 	hcp->hcd = *hcd;
-	mutex_init(&hcp->current_stream_lock);
-
 	hcp->daidrv = devm_kcalloc(dev, dai_count, sizeof(*hcp->daidrv),
 				   GFP_KERNEL);
 	if (!hcp->daidrv)
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* Applied "ASoC: hdmi-codec: remove function name debug traces" to the asoc tree
  2019-05-06  9:58 ` [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces Jerome Brunet
  2019-05-08  6:50   ` Mark Brown
@ 2019-05-08  9:01   ` Mark Brown
  1 sibling, 0 replies; 16+ messages in thread
From: Mark Brown @ 2019-05-08  9:01 UTC (permalink / raw)
  To: Jerome Brunet
  Cc: alsa-devel, Liam Girdwood, linux-kernel, Mark Brown,
	patchwork-bot+notify

The patch

   ASoC: hdmi-codec: remove function name debug traces

has been applied to the asoc tree at

   https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git for-5.3

All being well this means that it will be integrated into the linux-next
tree (usually sometime in the next 24 hours) and sent to Linus during
the next merge window (or sooner if it is a bug fix), however if
problems are discovered then the patch may be dropped or reverted.  

You may get further e-mails resulting from automated or manual testing
and review of the tree, please engage with people reporting problems and
send followup patches addressing any issues that are reported if needed.

If any updates are required or you are submitting further changes they
should be sent as incremental updates against current git, existing
patches will not be replaced.

Please add any relevant lists and maintainers to the CCs when replying
to this mail.

Thanks,
Mark

From 900e5daf7034cf65ce4072b86f297c42f9042433 Mon Sep 17 00:00:00 2001
From: Jerome Brunet <jbrunet@baylibre.com>
Date: Mon, 6 May 2019 11:58:12 +0200
Subject: [PATCH] ASoC: hdmi-codec: remove function name debug traces

Remove the debug traces only showing the function name on entry.
The same can be obtained using ftrace.

Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
---
 sound/soc/codecs/hdmi-codec.c | 12 ------------
 1 file changed, 12 deletions(-)

diff --git a/sound/soc/codecs/hdmi-codec.c b/sound/soc/codecs/hdmi-codec.c
index 39caf19abb0b..eb31d7eddcbf 100644
--- a/sound/soc/codecs/hdmi-codec.c
+++ b/sound/soc/codecs/hdmi-codec.c
@@ -416,8 +416,6 @@ static int hdmi_codec_startup(struct snd_pcm_substream *substream,
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 	int ret = 0;
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	ret = hdmi_codec_new_stream(substream, dai);
 	if (ret)
 		return ret;
@@ -457,8 +455,6 @@ static void hdmi_codec_shutdown(struct snd_pcm_substream *substream,
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	WARN_ON(hcp->current_stream != substream);
 
 	hcp->chmap_idx = HDMI_CODEC_CHMAP_IDX_UNKNOWN;
@@ -527,8 +523,6 @@ static int hdmi_codec_set_fmt(struct snd_soc_dai *dai,
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 	struct hdmi_codec_daifmt cf = { 0 };
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	if (dai->id == DAI_ID_SPDIF)
 		return 0;
 
@@ -597,8 +591,6 @@ static int hdmi_codec_digital_mute(struct snd_soc_dai *dai, int mute)
 {
 	struct hdmi_codec_priv *hcp = snd_soc_dai_get_drvdata(dai);
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	if (hcp->hcd.ops->digital_mute)
 		return hcp->hcd.ops->digital_mute(dai->dev->parent,
 						  hcp->hcd.data, mute);
@@ -656,8 +648,6 @@ static int hdmi_codec_pcm_new(struct snd_soc_pcm_runtime *rtd,
 	};
 	int ret;
 
-	dev_dbg(dai->dev, "%s()\n", __func__);
-
 	ret =  snd_pcm_add_chmap_ctls(rtd->pcm, SNDRV_PCM_STREAM_PLAYBACK,
 				      NULL, drv->playback.channels_max, 0,
 				      &hcp->chmap_info);
@@ -754,8 +744,6 @@ static int hdmi_codec_probe(struct platform_device *pdev)
 	int dai_count, i = 0;
 	int ret;
 
-	dev_dbg(dev, "%s()\n", __func__);
-
 	if (!hcd) {
 		dev_err(dev, "%s: No platform data\n", __func__);
 		return -EINVAL;
-- 
2.20.1


^ permalink raw reply related	[flat|nested] 16+ messages in thread

* Re: [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream
  2019-05-08  8:57       ` Mark Brown
@ 2019-05-09  8:11         ` Jerome Brunet
  2019-05-12  8:27           ` Mark Brown
  0 siblings, 1 reply; 16+ messages in thread
From: Jerome Brunet @ 2019-05-09  8:11 UTC (permalink / raw)
  To: Mark Brown; +Cc: Liam Girdwood, alsa-devel, linux-kernel, patchwork-bot+notify

On Wed, 2019-05-08 at 17:57 +0900, Mark Brown wrote:
> > Another solution would be to use the mutex as the 'busy' flag.
> > Get the ownership of the hdmi using try_lock() in startup() and
> > release it in shutdown()
> 
> > Would you have a preference Mark ? 
> 
> Probably using a mutex for the flag is clearer.  I'd rather keep things
> as simple as absolutely possible to avoid people making mistakes in
> future.

Hi Mark,

I received a notification that you applied this patch.
Just to confirm, you expect a follow up patch to re-introduce the mutex, right ?

Thanks
Jerome


^ permalink raw reply	[flat|nested] 16+ messages in thread

* Re: [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream
  2019-05-09  8:11         ` Jerome Brunet
@ 2019-05-12  8:27           ` Mark Brown
  0 siblings, 0 replies; 16+ messages in thread
From: Mark Brown @ 2019-05-12  8:27 UTC (permalink / raw)
  To: Jerome Brunet
  Cc: Liam Girdwood, alsa-devel, linux-kernel, patchwork-bot+notify

[-- Attachment #1: Type: text/plain, Size: 428 bytes --]

On Thu, May 09, 2019 at 10:11:48AM +0200, Jerome Brunet wrote:
> On Wed, 2019-05-08 at 17:57 +0900, Mark Brown wrote:

> > Probably using a mutex for the flag is clearer.  I'd rather keep things
> > as simple as absolutely possible to avoid people making mistakes in
> > future.

> I received a notification that you applied this patch.
> Just to confirm, you expect a follow up patch to re-introduce the mutex, right ?

Right.

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

^ permalink raw reply	[flat|nested] 16+ messages in thread

end of thread, other threads:[~2019-05-12 17:06 UTC | newest]

Thread overview: 16+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-05-06  9:58 [PATCH v2 0/4] ASoC: hdmi-codec: fixes and improvements Jerome Brunet
2019-05-06  9:58 ` [PATCH v2 1/4] ASoC: hdmi-codec: remove function name debug traces Jerome Brunet
2019-05-08  6:50   ` Mark Brown
2019-05-08  8:08     ` Jerome Brunet
2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove function name debug traces" to the asoc tree Mark Brown
2019-05-06  9:58 ` [PATCH v2 2/4] ASoC: hdmi-codec: remove reference to the current substream Jerome Brunet
2019-05-08  7:00   ` Mark Brown
2019-05-08  8:08     ` Jerome Brunet
2019-05-08  8:57       ` Mark Brown
2019-05-09  8:11         ` Jerome Brunet
2019-05-12  8:27           ` Mark Brown
2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove reference to the current substream" to the asoc tree Mark Brown
2019-05-06  9:58 ` [PATCH v2 3/4] ASoC: hdmi-codec: remove reference to the dai drivers in the private data Jerome Brunet
2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove reference to the dai drivers in the private data" to the asoc tree Mark Brown
2019-05-06  9:58 ` [PATCH v2 4/4] ASoC: hdmi-codec: remove ops dependency on the dai id Jerome Brunet
2019-05-08  9:01   ` Applied "ASoC: hdmi-codec: remove ops dependency on the dai id" to the asoc tree Mark Brown

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