[PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[libin.yang]

From: Libin Yang <libin.yang@linux.intel.com>

The patches are trying to support dynamically binding PCM to pin
in HDMI/DP audio.

The pin will not bind to any PCM if there is not monitor connected.
When there is a monitor connected to the port (pin), driver will
try to find a proper PCM to assign to the port and setup the pin
if necessary.

Libin Yang (5):
  ALSA: hda - hdmi begin to support dynamic PCM assignment
  ALSA: hda - hdmi playback without monitor in dynamic pcm bind mode
  ALSA: hda - hdmi operate spdif based on pcm
  ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
  ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
    assignment mode

 sound/pci/hda/patch_hdmi.c | 374 ++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 351 insertions(+), 23 deletions(-)

-- 
1.9.1

[PATCH v4 1/5] ALSA: hda - hdmi begin to support dynamic PCM assignment

[libin.yang]

From: Libin Yang <libin.yang@linux.intel.com>

Begin to support dynamic PCM assignment to pin in
hdmi audio driver.

This means PCM will not be statically bound with pin.
When there is a monitor connected, the corresponding pin
will try to find a proper PCM to bind. When the monitor
is disconnected, the corresponding pin will unbind
the PCM. This helps to reduce the PCM number when there
are many pins (device entries in DP MST mode) and only
a few of them work at the same time.

This patch adds the pcm member in struct hdmi_spec_per_pin.
When PCM is dynamically bound to the pin, the member pcm
will pointer to the corresponding pcm_rec[] in hdmi_spec,
which means the hda_pcm is bound to the pin.

Signed-off-by: Libin Yang <libin.yang@linux.intel.com>
---
 sound/pci/hda/patch_hdmi.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c
index cd9b0ff..f762c7d 100644
--- a/sound/pci/hda/patch_hdmi.c
+++ b/sound/pci/hda/patch_hdmi.c
@@ -84,6 +84,7 @@ struct hdmi_spec_per_pin {
 	struct delayed_work work;
 	struct snd_kcontrol *eld_ctl;
 	struct snd_jack *acomp_jack; /* jack via audio component */
+	struct hda_pcm *pcm; /* pointer to spec->pcm_rec[n] dynamically*/
 	int repoll_count;
 	bool setup; /* the stream has been set up by prepare callback */
 	int channels; /* current number of channels */
@@ -142,7 +143,7 @@ struct hdmi_spec {
 	struct hdmi_ops ops;
 
 	bool dyn_pin_out;
-
+	bool dyn_pcm_assign;
 	/*
 	 * Non-generic VIA/NVIDIA specific
 	 */
@@ -387,13 +388,16 @@ static int hinfo_to_pin_index(struct hda_codec *codec,
 			      struct hda_pcm_stream *hinfo)
 {
 	struct hdmi_spec *spec = codec->spec;
+	struct hdmi_spec_per_pin *per_pin;
 	int pin_idx;
 
-	for (pin_idx = 0; pin_idx < spec->num_pins; pin_idx++)
-		if (get_pcm_rec(spec, pin_idx)->stream == hinfo)
+	for (pin_idx = 0; pin_idx < spec->num_pins; pin_idx++) {
+		per_pin = get_pin(spec, pin_idx);
+		if (per_pin->pcm && per_pin->pcm->stream == hinfo)
 			return pin_idx;
+	}
 
-	codec_warn(codec, "HDMI: hinfo %p not registered\n", hinfo);
+	codec_dbg(codec, "HDMI: hinfo %p not registered\n", hinfo);
 	return -EINVAL;
 }
 
@@ -2114,6 +2118,7 @@ static int hdmi_chmap_ctl_put(struct snd_kcontrol *kcontrol,
 static int generic_hdmi_build_pcms(struct hda_codec *codec)
 {
 	struct hdmi_spec *spec = codec->spec;
+	struct hdmi_spec_per_pin *per_pin;
 	int pin_idx;
 
 	for (pin_idx = 0; pin_idx < spec->num_pins; pin_idx++) {
@@ -2123,6 +2128,10 @@ static int generic_hdmi_build_pcms(struct hda_codec *codec)
 		info = snd_hda_codec_pcm_new(codec, "HDMI %d", pin_idx);
 		if (!info)
 			return -ENOMEM;
+		if (!spec->dyn_pcm_assign) {
+			per_pin = get_pin(spec, pin_idx);
+			per_pin->pcm = info;
+		}
 		spec->pcm_rec[pin_idx] = info;
 		info->pcm_type = HDA_PCM_TYPE_HDMI;
 		info->own_chmap = true;
-- 
1.9.1

[PATCH v4 2/5] ALSA: hda - hdmi playback without monitor in dynamic pcm bind mode

[libin.yang]

From: Libin Yang <libin.yang@linux.intel.com>

Pulseaudio requires open pcm successfully when probing.

This patch handles playback without monitor in dynamic pcm assignment
mode. It tries to open/prepare/close pcm successfully even there is
no pin bound to the PCM. On the meantime, it will try to find a proper
converter for the PCM.

As pcm is This patch introduces a pcm_lock in struct hdmi_spec.
This lock is used to protect:
1. the variables in struct hdmi_spec;
2. other variables shared for dynamic pcm assignment mode
3. device entry selection. As each device entry is represented by
   a separate struct struct hdmi_spec_per_pin, the lock in per_pin
   is not enough. Please see details below.

MST audio device entry operation:
1. select device entry on the pin
2. operate on the pin nid

Signed-off-by: Libin Yang <libin.yang@linux.intel.com>
---
 sound/pci/hda/patch_hdmi.c | 171 +++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 157 insertions(+), 14 deletions(-)

diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c
index f762c7d..04450f9 100644
--- a/sound/pci/hda/patch_hdmi.c
+++ b/sound/pci/hda/patch_hdmi.c
@@ -137,6 +137,7 @@ struct hdmi_spec {
 	int num_pins;
 	struct snd_array pins; /* struct hdmi_spec_per_pin */
 	struct hda_pcm *pcm_rec[16];
+	struct mutex pcm_lock;
 	unsigned int channels_max; /* max over all cvts */
 
 	struct hdmi_eld temp_eld;
@@ -1339,6 +1340,11 @@ static int hdmi_setup_stream(struct hda_codec *codec, hda_nid_t cvt_nid,
 	return 0;
 }
 
+/* Try to find an available converter
+ * If pin_idx is less then zero, just try to find an available converter.
+ * Otherwise, try to find an available converter and get the cvt mux index
+ * of the pin.
+ */
 static int hdmi_choose_cvt(struct hda_codec *codec,
 			int pin_idx, int *cvt_id, int *mux_id)
 {
@@ -1347,7 +1353,11 @@ static int hdmi_choose_cvt(struct hda_codec *codec,
 	struct hdmi_spec_per_cvt *per_cvt = NULL;
 	int cvt_idx, mux_idx = 0;
 
-	per_pin = get_pin(spec, pin_idx);
+	/* pin_idx < 0 means no pin will be bound to the converter */
+	if (pin_idx < 0)
+		per_pin = NULL;
+	else
+		per_pin = get_pin(spec, pin_idx);
 
 	/* Dynamically assign converter to stream */
 	for (cvt_idx = 0; cvt_idx < spec->num_cvts; cvt_idx++) {
@@ -1356,6 +1366,8 @@ static int hdmi_choose_cvt(struct hda_codec *codec,
 		/* Must not already be assigned */
 		if (per_cvt->assigned)
 			continue;
+		if (per_pin == NULL)
+			break;
 		/* Must be in pin's mux's list of converters */
 		for (mux_idx = 0; mux_idx < per_pin->num_mux_nids; mux_idx++)
 			if (per_pin->mux_nids[mux_idx] == per_cvt->cvt_nid)
@@ -1368,9 +1380,10 @@ static int hdmi_choose_cvt(struct hda_codec *codec,
 
 	/* No free converters */
 	if (cvt_idx == spec->num_cvts)
-		return -ENODEV;
+		return -EBUSY;
 
-	per_pin->mux_idx = mux_idx;
+	if (per_pin != NULL)
+		per_pin->mux_idx = mux_idx;
 
 	if (cvt_id)
 		*cvt_id = cvt_idx;
@@ -1396,6 +1409,20 @@ static void intel_verify_pin_cvt_connect(struct hda_codec *codec,
 					    mux_idx);
 }
 
+/* get the mux index for the converter of the pins
+ * converter's mux index is the same for all pins on Intel platform
+ */
+static int intel_cvt_id_to_mux_idx(struct hdmi_spec *spec,
+			hda_nid_t cvt_nid)
+{
+	int i;
+
+	for (i = 0; i < spec->num_cvts; i++)
+		if (spec->cvt_nids[i] == cvt_nid)
+			return i;
+	return -EINVAL;
+}
+
 /* Intel HDMI workaround to fix audio routing issue:
  * For some Intel display codecs, pins share the same connection list.
  * So a conveter can be selected by multiple pins and playback on any of these
@@ -1447,6 +1474,69 @@ static void intel_not_share_assigned_cvt(struct hda_codec *codec,
 	}
 }
 
+/* A wrapper of intel_not_share_asigned_cvt() */
+static void intel_not_share_assigned_cvt_nid(struct hda_codec *codec,
+			hda_nid_t pin_nid, hda_nid_t cvt_nid)
+{
+	int mux_idx;
+	struct hdmi_spec *spec = codec->spec;
+
+	if (!is_haswell_plus(codec) && !is_valleyview_plus(codec))
+		return;
+
+	/* On Intel platform, the mapping of converter nid to
+	 * mux index of the pins are always the same.
+	 * The pin nid may be 0, this means all pins will not
+	 * share the converter.
+	 */
+	mux_idx = intel_cvt_id_to_mux_idx(spec, cvt_nid);
+	if (mux_idx >= 0)
+		intel_not_share_assigned_cvt(codec, pin_nid, mux_idx);
+}
+
+/* called in hdmi_pcm_open when no pin is assigned to the PCM
+ * in dyn_pcm_assign mode.
+ */
+static int hdmi_pcm_open_no_pin(struct hda_pcm_stream *hinfo,
+			 struct hda_codec *codec,
+			 struct snd_pcm_substream *substream)
+{
+	struct hdmi_spec *spec = codec->spec;
+	struct snd_pcm_runtime *runtime = substream->runtime;
+	int cvt_idx;
+	struct hdmi_spec_per_cvt *per_cvt = NULL;
+	int err;
+
+	err = hdmi_choose_cvt(codec, -1, &cvt_idx, NULL);
+	if (err)
+		return err;
+
+	per_cvt = get_cvt(spec, cvt_idx);
+	per_cvt->assigned = 1;
+	hinfo->nid = per_cvt->cvt_nid;
+
+	intel_not_share_assigned_cvt_nid(codec, 0, per_cvt->cvt_nid);
+
+	/* todo: setup spdif ctls assign */
+
+	/* Initially set the converter's capabilities */
+	hinfo->channels_min = per_cvt->channels_min;
+	hinfo->channels_max = per_cvt->channels_max;
+	hinfo->rates = per_cvt->rates;
+	hinfo->formats = per_cvt->formats;
+	hinfo->maxbps = per_cvt->maxbps;
+
+	/* Store the updated parameters */
+	runtime->hw.channels_min = hinfo->channels_min;
+	runtime->hw.channels_max = hinfo->channels_max;
+	runtime->hw.formats = hinfo->formats;
+	runtime->hw.rates = hinfo->rates;
+
+	snd_pcm_hw_constraint_step(substream->runtime, 0,
+				   SNDRV_PCM_HW_PARAM_CHANNELS, 2);
+	return 0;
+}
+
 /*
  * HDA PCM callbacks
  */
@@ -1463,19 +1553,36 @@ static int hdmi_pcm_open(struct hda_pcm_stream *hinfo,
 	int err;
 
 	/* Validate hinfo */
+	mutex_lock(&spec->pcm_lock);
 	pin_idx = hinfo_to_pin_index(codec, hinfo);
-	if (snd_BUG_ON(pin_idx < 0))
-		return -EINVAL;
-	per_pin = get_pin(spec, pin_idx);
-	eld = &per_pin->sink_eld;
+	if (!spec->dyn_pcm_assign) {
+		if (snd_BUG_ON(pin_idx < 0)) {
+			mutex_unlock(&spec->pcm_lock);
+			return -EINVAL;
+		}
+	} else {
+		/* no pin is assigned to the PCM
+		 * PA need pcm open successfully when probe
+		 */
+		if (pin_idx < 0) {
+			err = hdmi_pcm_open_no_pin(hinfo, codec, substream);
+			mutex_unlock(&spec->pcm_lock);
+			return err;
+		}
+	}
 
 	err = hdmi_choose_cvt(codec, pin_idx, &cvt_idx, &mux_idx);
-	if (err < 0)
+	if (err < 0) {
+		mutex_unlock(&spec->pcm_lock);
 		return err;
+	}
 
 	per_cvt = get_cvt(spec, cvt_idx);
 	/* Claim converter */
 	per_cvt->assigned = 1;
+
+
+	per_pin = get_pin(spec, pin_idx);
 	per_pin->cvt_nid = per_cvt->cvt_nid;
 	hinfo->nid = per_cvt->cvt_nid;
 
@@ -1496,6 +1603,7 @@ static int hdmi_pcm_open(struct hda_pcm_stream *hinfo,
 	hinfo->formats = per_cvt->formats;
 	hinfo->maxbps = per_cvt->maxbps;
 
+	eld = &per_pin->sink_eld;
 	/* Restrict capabilities by ELD if this isn't disabled */
 	if (!static_hdmi_pcm && eld->eld_valid) {
 		snd_hdmi_eld_update_pcm_info(&eld->info, hinfo);
@@ -1504,10 +1612,12 @@ static int hdmi_pcm_open(struct hda_pcm_stream *hinfo,
 			per_cvt->assigned = 0;
 			hinfo->nid = 0;
 			snd_hda_spdif_ctls_unassign(codec, pin_idx);
+			mutex_unlock(&spec->pcm_lock);
 			return -ENODEV;
 		}
 	}
 
+	mutex_unlock(&spec->pcm_lock);
 	/* Store the updated parameters */
 	runtime->hw.channels_min = hinfo->channels_min;
 	runtime->hw.channels_max = hinfo->channels_max;
@@ -1852,13 +1962,34 @@ static int generic_hdmi_playback_pcm_prepare(struct hda_pcm_stream *hinfo,
 {
 	hda_nid_t cvt_nid = hinfo->nid;
 	struct hdmi_spec *spec = codec->spec;
-	int pin_idx = hinfo_to_pin_index(codec, hinfo);
-	struct hdmi_spec_per_pin *per_pin = get_pin(spec, pin_idx);
-	hda_nid_t pin_nid = per_pin->pin_nid;
+	int pin_idx;
+	struct hdmi_spec_per_pin *per_pin;
+	hda_nid_t pin_nid;
 	struct snd_pcm_runtime *runtime = substream->runtime;
 	bool non_pcm;
 	int pinctl;
+	int err;
+
+	mutex_lock(&spec->pcm_lock);
+	pin_idx = hinfo_to_pin_index(codec, hinfo);
+	if (spec->dyn_pcm_assign && pin_idx < 0) {
+		/* when dyn_pcm_assign and pcm is not bound to a pin
+		 * skip pin setup and return 0 to make audio playback
+		 * be ongoing
+		 */
+		intel_not_share_assigned_cvt_nid(codec, 0, cvt_nid);
+		snd_hda_codec_setup_stream(codec, cvt_nid,
+					stream_tag, 0, format);
+		mutex_unlock(&spec->pcm_lock);
+		return 0;
+	}
 
+	if (snd_BUG_ON(pin_idx < 0)) {
+		mutex_unlock(&spec->pcm_lock);
+		return -EINVAL;
+	}
+	per_pin = get_pin(spec, pin_idx);
+	pin_nid = per_pin->pin_nid;
 	if (is_haswell_plus(codec) || is_valleyview_plus(codec)) {
 		/* Verify pin:cvt selections to avoid silent audio after S3.
 		 * After S3, the audio driver restores pin:cvt selections
@@ -1883,7 +2014,6 @@ static int generic_hdmi_playback_pcm_prepare(struct hda_pcm_stream *hinfo,
 
 	hdmi_setup_audio_infoframe(codec, per_pin, non_pcm);
 	mutex_unlock(&per_pin->lock);
-
 	if (spec->dyn_pin_out) {
 		pinctl = snd_hda_codec_read(codec, pin_nid, 0,
 					    AC_VERB_GET_PIN_WIDGET_CONTROL, 0);
@@ -1892,7 +2022,10 @@ static int generic_hdmi_playback_pcm_prepare(struct hda_pcm_stream *hinfo,
 				    pinctl | PIN_OUT);
 	}
 
-	return spec->ops.setup_stream(codec, cvt_nid, pin_nid, stream_tag, format);
+	err = spec->ops.setup_stream(codec, cvt_nid, pin_nid,
+				 stream_tag, format);
+	mutex_unlock(&spec->pcm_lock);
+	return err;
 }
 
 static int generic_hdmi_playback_pcm_cleanup(struct hda_pcm_stream *hinfo,
@@ -1923,9 +2056,17 @@ static int hdmi_pcm_close(struct hda_pcm_stream *hinfo,
 		per_cvt->assigned = 0;
 		hinfo->nid = 0;
 
+		mutex_lock(&spec->pcm_lock);
 		pin_idx = hinfo_to_pin_index(codec, hinfo);
-		if (snd_BUG_ON(pin_idx < 0))
+		if (spec->dyn_pcm_assign && pin_idx < 0) {
+			mutex_unlock(&spec->pcm_lock);
+			return 0;
+		}
+
+		if (snd_BUG_ON(pin_idx < 0)) {
+			mutex_unlock(&spec->pcm_lock);
 			return -EINVAL;
+		}
 		per_pin = get_pin(spec, pin_idx);
 
 		if (spec->dyn_pin_out) {
@@ -1945,6 +2086,7 @@ static int hdmi_pcm_close(struct hda_pcm_stream *hinfo,
 		per_pin->setup = false;
 		per_pin->channels = 0;
 		mutex_unlock(&per_pin->lock);
+		mutex_unlock(&spec->pcm_lock);
 	}
 
 	return 0;
@@ -2459,6 +2601,7 @@ static int patch_generic_hdmi(struct hda_codec *codec)
 		return -ENOMEM;
 
 	spec->ops = generic_standard_hdmi_ops;
+	mutex_init(&spec->pcm_lock);
 	codec->spec = spec;
 	hdmi_array_init(spec, 4);
 
-- 
1.9.1

[PATCH v4 3/5] ALSA: hda - hdmi operate spdif based on pcm

[libin.yang]

From: Libin Yang <libin.yang@linux.intel.com>

Currently, the driver operates the spdif based on pin.
This is ok for the current driver as pcm is statically
bound to the pin.

However, if the driver uses dynamically pcm assignment,
this will cause confusion for user space.

The patch changes spdif operation from pin based to pcm based.

Signed-off-by: Libin Yang <libin.yang@linux.intel.com>
---
 sound/pci/hda/patch_hdmi.c | 34 +++++++++++++++++++++++++++++-----
 1 file changed, 29 insertions(+), 5 deletions(-)

diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c
index 04450f9..19bf782 100644
--- a/sound/pci/hda/patch_hdmi.c
+++ b/sound/pci/hda/patch_hdmi.c
@@ -138,6 +138,7 @@ struct hdmi_spec {
 	struct snd_array pins; /* struct hdmi_spec_per_pin */
 	struct hda_pcm *pcm_rec[16];
 	struct mutex pcm_lock;
+	int pcm_used;	/* counter of pcm_rec[] */
 	unsigned int channels_max; /* max over all cvts */
 
 	struct hdmi_eld temp_eld;
@@ -385,6 +386,20 @@ static int pin_nid_to_pin_index(struct hda_codec *codec, hda_nid_t pin_nid)
 	return -EINVAL;
 }
 
+static int hinfo_to_pcm_index(struct hda_codec *codec,
+			struct hda_pcm_stream *hinfo)
+{
+	struct hdmi_spec *spec = codec->spec;
+	int pcm_idx;
+
+	for (pcm_idx = 0; pcm_idx < spec->pcm_used; pcm_idx++)
+		if (get_pcm_rec(spec, pcm_idx)->stream == hinfo)
+			return pcm_idx;
+
+	codec_warn(codec, "HDMI: hinfo %p not registered\n", hinfo);
+	return -EINVAL;
+}
+
 static int hinfo_to_pin_index(struct hda_codec *codec,
 			      struct hda_pcm_stream *hinfo)
 {
@@ -1546,13 +1561,17 @@ static int hdmi_pcm_open(struct hda_pcm_stream *hinfo,
 {
 	struct hdmi_spec *spec = codec->spec;
 	struct snd_pcm_runtime *runtime = substream->runtime;
-	int pin_idx, cvt_idx, mux_idx = 0;
+	int pin_idx, cvt_idx, pcm_idx, mux_idx = 0;
 	struct hdmi_spec_per_pin *per_pin;
 	struct hdmi_eld *eld;
 	struct hdmi_spec_per_cvt *per_cvt = NULL;
 	int err;
 
 	/* Validate hinfo */
+	pcm_idx = hinfo_to_pcm_index(codec, hinfo);
+	if (pcm_idx < 0)
+		return -EINVAL;
+
 	mutex_lock(&spec->pcm_lock);
 	pin_idx = hinfo_to_pin_index(codec, hinfo);
 	if (!spec->dyn_pcm_assign) {
@@ -1594,7 +1613,7 @@ static int hdmi_pcm_open(struct hda_pcm_stream *hinfo,
 	if (is_haswell_plus(codec) || is_valleyview_plus(codec))
 		intel_not_share_assigned_cvt(codec, per_pin->pin_nid, mux_idx);
 
-	snd_hda_spdif_ctls_assign(codec, pin_idx, per_cvt->cvt_nid);
+	snd_hda_spdif_ctls_assign(codec, pcm_idx, per_cvt->cvt_nid);
 
 	/* Initially set the converter's capabilities */
 	hinfo->channels_min = per_cvt->channels_min;
@@ -1611,7 +1630,7 @@ static int hdmi_pcm_open(struct hda_pcm_stream *hinfo,
 		    !hinfo->rates || !hinfo->formats) {
 			per_cvt->assigned = 0;
 			hinfo->nid = 0;
-			snd_hda_spdif_ctls_unassign(codec, pin_idx);
+			snd_hda_spdif_ctls_unassign(codec, pcm_idx);
 			mutex_unlock(&spec->pcm_lock);
 			return -ENODEV;
 		}
@@ -2041,12 +2060,15 @@ static int hdmi_pcm_close(struct hda_pcm_stream *hinfo,
 			  struct snd_pcm_substream *substream)
 {
 	struct hdmi_spec *spec = codec->spec;
-	int cvt_idx, pin_idx;
+	int cvt_idx, pin_idx, pcm_idx;
 	struct hdmi_spec_per_cvt *per_cvt;
 	struct hdmi_spec_per_pin *per_pin;
 	int pinctl;
 
 	if (hinfo->nid) {
+		pcm_idx = hinfo_to_pcm_index(codec, hinfo);
+		if (snd_BUG_ON(pcm_idx < 0))
+			return -EINVAL;
 		cvt_idx = cvt_nid_to_cvt_index(codec, hinfo->nid);
 		if (snd_BUG_ON(cvt_idx < 0))
 			return -EINVAL;
@@ -2077,7 +2099,7 @@ static int hdmi_pcm_close(struct hda_pcm_stream *hinfo,
 					    pinctl & ~PIN_OUT);
 		}
 
-		snd_hda_spdif_ctls_unassign(codec, pin_idx);
+		snd_hda_spdif_ctls_unassign(codec, pcm_idx);
 
 		mutex_lock(&per_pin->lock);
 		per_pin->chmap_set = false;
@@ -2275,6 +2297,7 @@ static int generic_hdmi_build_pcms(struct hda_codec *codec)
 			per_pin->pcm = info;
 		}
 		spec->pcm_rec[pin_idx] = info;
+		spec->pcm_used++;
 		info->pcm_type = HDA_PCM_TYPE_HDMI;
 		info->own_chmap = true;
 
@@ -2351,6 +2374,7 @@ static int generic_hdmi_build_controls(struct hda_codec *codec)
 						  HDA_PCM_TYPE_HDMI);
 		if (err < 0)
 			return err;
+		/* pin number is the same with pcm number so far */
 		snd_hda_spdif_ctls_unassign(codec, pin_idx);
 
 		/* add control for ELD Bytes */
-- 
1.9.1

[PATCH v4 4/5] ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug

[libin.yang]

From: Libin Yang <libin.yang@linux.intel.com>

Dynamically bind/unbind the PCM to pin when HDMI/DP monitor hotplug.

When monitor is connected, find a proper PCM for the monitor.
When monitor is disconnected, unbind the PCM from the pin.

The binding policy (use Intel platform as example) is:
1. Try to use the legacy pin-pcm mapping for the device entry 0
   of the pin.
2. If step 1 fails, try to bind pin to the backup PCMs. For example,
   on Intel platform, if DP MST is enabled, 5 PCMs will be created.
   PCM 3, PCM 7, PCM 8 are supposed to be used by device entry 0 of
   pin 5, pin 6 and pin 7. PCM 9 and PCM 10 are the backup PCMs.
3. If step 2 fails, try to find any PCM to bind to the pin.

Signed-off-by: Libin Yang <libin.yang@linux.intel.com>
---
 sound/pci/hda/patch_hdmi.c | 73 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 73 insertions(+)

diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c
index 19bf782..10edab8 100644
--- a/sound/pci/hda/patch_hdmi.c
+++ b/sound/pci/hda/patch_hdmi.c
@@ -73,6 +73,8 @@ struct hdmi_spec_per_cvt {
 
 struct hdmi_spec_per_pin {
 	hda_nid_t pin_nid;
+	/* pin idx, different device entries on the same pin use the same idx */
+	int pin_nid_idx;
 	int num_mux_nids;
 	hda_nid_t mux_nids[HDA_MAX_CONNECTIONS];
 	int mux_idx;
@@ -85,6 +87,7 @@ struct hdmi_spec_per_pin {
 	struct snd_kcontrol *eld_ctl;
 	struct snd_jack *acomp_jack; /* jack via audio component */
 	struct hda_pcm *pcm; /* pointer to spec->pcm_rec[n] dynamically*/
+	int pcm_idx; /* which pcm is attached. -1 means no pcm is attached */
 	int repoll_count;
 	bool setup; /* the stream has been set up by prepare callback */
 	int channels; /* current number of channels */
@@ -138,6 +141,8 @@ struct hdmi_spec {
 	struct snd_array pins; /* struct hdmi_spec_per_pin */
 	struct hda_pcm *pcm_rec[16];
 	struct mutex pcm_lock;
+	/* pcm_bitmap means which pcms have been assigned to pins*/
+	unsigned long pcm_bitmap;
 	int pcm_used;	/* counter of pcm_rec[] */
 	unsigned int channels_max; /* max over all cvts */
 
@@ -1786,6 +1791,60 @@ static bool hdmi_present_sense_via_verbs(struct hdmi_spec_per_pin *per_pin,
 	return ret;
 }
 
+static int hdmi_find_pcm_slot(struct hdmi_spec *spec,
+				struct hdmi_spec_per_pin *per_pin)
+{
+	int i;
+
+	/* try the prefer PCM */
+	if (!test_bit(per_pin->pin_nid_idx, &spec->pcm_bitmap))
+		return per_pin->pin_nid_idx;
+
+	/* have a second try; check the "reserved area" over num_pins */
+	for (i = spec->num_pins; i < spec->pcm_used; i++) {
+		if (!test_bit(i, &spec->pcm_bitmap))
+			return i;
+	}
+
+	/* the last try; check the empty slots in pins */
+	for (i = 0; i < spec->num_pins; i++) {
+		if (!test_bit(i, &spec->pcm_bitmap))
+			return i;
+	}
+	return -EBUSY;
+}
+
+static void hdmi_attach_hda_pcm(struct hdmi_spec *spec,
+				struct hdmi_spec_per_pin *per_pin)
+{
+	int idx;
+
+	/* pcm already be attached to the pin */
+	if (per_pin->pcm)
+		return;
+	idx = hdmi_find_pcm_slot(spec, per_pin);
+	if (idx == -ENODEV)
+		return;
+	per_pin->pcm_idx = idx;
+	per_pin->pcm = spec->pcm_rec[idx];
+	set_bit(idx, &spec->pcm_bitmap);
+}
+
+static void hdmi_detach_hda_pcm(struct hdmi_spec *spec,
+				struct hdmi_spec_per_pin *per_pin)
+{
+	int idx;
+
+	/* pcm already be detached from the pin */
+	if (!per_pin->pcm)
+		return;
+	idx = per_pin->pcm_idx;
+	per_pin->pcm_idx = -1;
+	per_pin->pcm = NULL;
+	if (idx >= 0 && idx < spec->pcm_used)
+		clear_bit(idx, &spec->pcm_bitmap);
+}
+
 /* update ELD and jack state via audio component */
 static void sync_eld_via_acomp(struct hda_codec *codec,
 			       struct hdmi_spec_per_pin *per_pin)
@@ -1794,6 +1853,7 @@ static void sync_eld_via_acomp(struct hda_codec *codec,
 	struct hdmi_eld *eld = &spec->temp_eld;
 	int size;
 
+	mutex_lock(&spec->pcm_lock);
 	mutex_lock(&per_pin->lock);
 	size = snd_hdac_acomp_get_eld(&codec->bus->core, per_pin->pin_nid,
 				      &eld->monitor_present, eld->eld_buffer,
@@ -1815,11 +1875,19 @@ static void sync_eld_via_acomp(struct hda_codec *codec,
 		eld->eld_size = 0;
 	}
 
+	if (spec->dyn_pcm_assign) {
+		if (eld->eld_valid)
+			hdmi_attach_hda_pcm(spec, per_pin);
+		else
+			hdmi_detach_hda_pcm(spec, per_pin);
+	}
+
 	update_eld(codec, per_pin, eld);
 	snd_jack_report(per_pin->acomp_jack,
 			eld->monitor_present ? SND_JACK_AVOUT : 0);
  unlock:
 	mutex_unlock(&per_pin->lock);
+	mutex_unlock(&spec->pcm_lock);
 }
 
 static bool hdmi_present_sense(struct hdmi_spec_per_pin *per_pin, int repoll)
@@ -1875,6 +1943,11 @@ static int hdmi_add_pin(struct hda_codec *codec, hda_nid_t pin_nid)
 
 	per_pin->pin_nid = pin_nid;
 	per_pin->non_pcm = false;
+	if (spec->dyn_pcm_assign)
+		per_pin->pcm_idx = -1;
+	else
+		per_pin->pcm_idx = pin_idx;
+	per_pin->pin_nid_idx = pin_idx;
 
 	err = hdmi_read_pin_conn(codec, pin_idx);
 	if (err < 0)
-- 
1.9.1

[PATCH v4 5/5] ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic assignment mode

[libin.yang]

From: Libin Yang <libin.yang@linux.intel.com>

Setup pin configuration when monitor is hotplugged
in pcm dynamic assignment if the PCM is in open state.

When monitor is disconnect, The pin will be reset.

Signed-off-by: Libin Yang <libin.yang@linux.intel.com>
---
 sound/pci/hda/patch_hdmi.c | 87 +++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 83 insertions(+), 4 deletions(-)

diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c
index 10edab8..b285988 100644
--- a/sound/pci/hda/patch_hdmi.c
+++ b/sound/pci/hda/patch_hdmi.c
@@ -144,6 +144,11 @@ struct hdmi_spec {
 	/* pcm_bitmap means which pcms have been assigned to pins*/
 	unsigned long pcm_bitmap;
 	int pcm_used;	/* counter of pcm_rec[] */
+	/* bitmap shows whether the pcm is opened in user space
+	 * bit 0 means the first playback PCM (PCM3);
+	 * bit 1 means the second playback PCM, and so on.
+	 */
+	unsigned long pcm_in_use;
 	unsigned int channels_max; /* max over all cvts */
 
 	struct hdmi_eld temp_eld;
@@ -1523,10 +1528,14 @@ static int hdmi_pcm_open_no_pin(struct hda_pcm_stream *hinfo,
 {
 	struct hdmi_spec *spec = codec->spec;
 	struct snd_pcm_runtime *runtime = substream->runtime;
-	int cvt_idx;
+	int cvt_idx, pcm_idx;
 	struct hdmi_spec_per_cvt *per_cvt = NULL;
 	int err;
 
+	pcm_idx = hinfo_to_pcm_index(codec, hinfo);
+	if (pcm_idx < 0)
+		return -EINVAL;
+
 	err = hdmi_choose_cvt(codec, -1, &cvt_idx, NULL);
 	if (err)
 		return err;
@@ -1537,6 +1546,7 @@ static int hdmi_pcm_open_no_pin(struct hda_pcm_stream *hinfo,
 
 	intel_not_share_assigned_cvt_nid(codec, 0, per_cvt->cvt_nid);
 
+	set_bit(pcm_idx, &spec->pcm_in_use);
 	/* todo: setup spdif ctls assign */
 
 	/* Initially set the converter's capabilities */
@@ -1605,7 +1615,7 @@ static int hdmi_pcm_open(struct hda_pcm_stream *hinfo,
 	/* Claim converter */
 	per_cvt->assigned = 1;
 
-
+	set_bit(pcm_idx, &spec->pcm_in_use);
 	per_pin = get_pin(spec, pin_idx);
 	per_pin->cvt_nid = per_cvt->cvt_nid;
 	hinfo->nid = per_cvt->cvt_nid;
@@ -1726,6 +1736,71 @@ static void update_eld(struct hda_codec *codec,
 			       &per_pin->eld_ctl->id);
 }
 
+static int hdmi_get_pin_cvt_mux(struct hdmi_spec *spec,
+		struct hdmi_spec_per_pin *per_pin, hda_nid_t cvt_nid)
+{
+	int mux_idx;
+
+	for (mux_idx = 0; mux_idx < per_pin->num_mux_nids; mux_idx++)
+		if (per_pin->mux_nids[mux_idx] == cvt_nid)
+			break;
+	return mux_idx;
+}
+
+static bool check_non_pcm_per_cvt(struct hda_codec *codec, hda_nid_t cvt_nid);
+
+static void hdmi_pcm_setup_pin(struct hdmi_spec *spec,
+			   struct hdmi_spec_per_pin *per_pin)
+{
+	struct hda_codec *codec = per_pin->codec;
+	struct hda_pcm *pcm;
+	struct hda_pcm_stream *hinfo;
+	struct snd_pcm_substream *substream;
+	int mux_idx;
+	bool non_pcm;
+
+	if (per_pin->pcm_idx >= 0 && per_pin->pcm_idx < spec->pcm_used)
+		pcm = spec->pcm_rec[per_pin->pcm_idx];
+	else
+		return;
+	if (!test_bit(per_pin->pcm_idx, &spec->pcm_in_use))
+		return;
+
+	/* hdmi audio only uses playback and one substream */
+	hinfo = pcm->stream;
+	substream = pcm->pcm->streams[0].substream;
+
+	per_pin->cvt_nid = hinfo->nid;
+
+	mux_idx = hdmi_get_pin_cvt_mux(spec, per_pin, hinfo->nid);
+	if (mux_idx < per_pin->num_mux_nids)
+		snd_hda_codec_write_cache(codec, per_pin->pin_nid, 0,
+				AC_VERB_SET_CONNECT_SEL,
+				mux_idx);
+	snd_hda_spdif_ctls_assign(codec, per_pin->pcm_idx, hinfo->nid);
+
+	non_pcm = check_non_pcm_per_cvt(codec, hinfo->nid);
+	if (substream->runtime)
+		per_pin->channels = substream->runtime->channels;
+	per_pin->setup = true;
+	per_pin->mux_idx = mux_idx;
+
+	hdmi_setup_audio_infoframe(codec, per_pin, non_pcm);
+}
+
+static void hdmi_pcm_reset_pin(struct hdmi_spec *spec,
+			   struct hdmi_spec_per_pin *per_pin)
+{
+	if (per_pin->pcm_idx >= 0 && per_pin->pcm_idx < spec->pcm_used)
+		snd_hda_spdif_ctls_unassign(per_pin->codec, per_pin->pcm_idx);
+
+	per_pin->chmap_set = false;
+	memset(per_pin->chmap, 0, sizeof(per_pin->chmap));
+
+	per_pin->setup = false;
+	per_pin->channels = 0;
+}
+
 /* update ELD and jack state via HD-audio verbs */
 static bool hdmi_present_sense_via_verbs(struct hdmi_spec_per_pin *per_pin,
 					 int repoll)
@@ -1876,10 +1951,13 @@ static void sync_eld_via_acomp(struct hda_codec *codec,
 	}
 
 	if (spec->dyn_pcm_assign) {
-		if (eld->eld_valid)
+		if (eld->eld_valid) {
 			hdmi_attach_hda_pcm(spec, per_pin);
-		else
+			hdmi_pcm_setup_pin(spec, per_pin);
+		} else {
+			hdmi_pcm_reset_pin(spec, per_pin);
 			hdmi_detach_hda_pcm(spec, per_pin);
+		}
 	}
 
 	update_eld(codec, per_pin, eld);
@@ -2152,6 +2230,7 @@ static int hdmi_pcm_close(struct hda_pcm_stream *hinfo,
 		hinfo->nid = 0;
 
 		mutex_lock(&spec->pcm_lock);
+		clear_bit(pcm_idx, &spec->pcm_in_use);
 		pin_idx = hinfo_to_pin_index(codec, hinfo);
 		if (spec->dyn_pcm_assign && pin_idx < 0) {
 			mutex_unlock(&spec->pcm_lock);
-- 
1.9.1

Re: [PATCH v4 4/5] ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug

[Takashi Iwai]

On Wed, 16 Dec 2015 06:42:44 +0100,
libin.yang@linux.intel.com wrote:
> 
> From: Libin Yang <libin.yang@linux.intel.com>
> 
> Dynamically bind/unbind the PCM to pin when HDMI/DP monitor hotplug.
> 
> When monitor is connected, find a proper PCM for the monitor.
> When monitor is disconnected, unbind the PCM from the pin.
> 
> The binding policy (use Intel platform as example) is:
> 1. Try to use the legacy pin-pcm mapping for the device entry 0
>    of the pin.
> 2. If step 1 fails, try to bind pin to the backup PCMs. For example,
>    on Intel platform, if DP MST is enabled, 5 PCMs will be created.
>    PCM 3, PCM 7, PCM 8 are supposed to be used by device entry 0 of
>    pin 5, pin 6 and pin 7. PCM 9 and PCM 10 are the backup PCMs.
> 3. If step 2 fails, try to find any PCM to bind to the pin.
> 
> Signed-off-by: Libin Yang <libin.yang@linux.intel.com>

This patch adds the code in sync_eld_via_acomp(), i.e. it's handled
only with component.  As the dynamic HDA attach/detach itself is
agnostic to the component, I'd suggest to put this rather in
update_eld().

Also pcm_lock can be put in hdmi_present_sense() to wrap both
sync_eld_via_acomp() and hdmi_present_sense_via_verbs().


Takashi

> ---
>  sound/pci/hda/patch_hdmi.c | 73 ++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 73 insertions(+)
> 
> diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c
> index 19bf782..10edab8 100644
> --- a/sound/pci/hda/patch_hdmi.c
> +++ b/sound/pci/hda/patch_hdmi.c
> @@ -73,6 +73,8 @@ struct hdmi_spec_per_cvt {
>  
>  struct hdmi_spec_per_pin {
>  	hda_nid_t pin_nid;
> +	/* pin idx, different device entries on the same pin use the same idx */
> +	int pin_nid_idx;
>  	int num_mux_nids;
>  	hda_nid_t mux_nids[HDA_MAX_CONNECTIONS];
>  	int mux_idx;
> @@ -85,6 +87,7 @@ struct hdmi_spec_per_pin {
>  	struct snd_kcontrol *eld_ctl;
>  	struct snd_jack *acomp_jack; /* jack via audio component */
>  	struct hda_pcm *pcm; /* pointer to spec->pcm_rec[n] dynamically*/
> +	int pcm_idx; /* which pcm is attached. -1 means no pcm is attached */
>  	int repoll_count;
>  	bool setup; /* the stream has been set up by prepare callback */
>  	int channels; /* current number of channels */
> @@ -138,6 +141,8 @@ struct hdmi_spec {
>  	struct snd_array pins; /* struct hdmi_spec_per_pin */
>  	struct hda_pcm *pcm_rec[16];
>  	struct mutex pcm_lock;
> +	/* pcm_bitmap means which pcms have been assigned to pins*/
> +	unsigned long pcm_bitmap;
>  	int pcm_used;	/* counter of pcm_rec[] */
>  	unsigned int channels_max; /* max over all cvts */
>  
> @@ -1786,6 +1791,60 @@ static bool hdmi_present_sense_via_verbs(struct hdmi_spec_per_pin *per_pin,
>  	return ret;
>  }
>  
> +static int hdmi_find_pcm_slot(struct hdmi_spec *spec,
> +				struct hdmi_spec_per_pin *per_pin)
> +{
> +	int i;
> +
> +	/* try the prefer PCM */
> +	if (!test_bit(per_pin->pin_nid_idx, &spec->pcm_bitmap))
> +		return per_pin->pin_nid_idx;
> +
> +	/* have a second try; check the "reserved area" over num_pins */
> +	for (i = spec->num_pins; i < spec->pcm_used; i++) {
> +		if (!test_bit(i, &spec->pcm_bitmap))
> +			return i;
> +	}
> +
> +	/* the last try; check the empty slots in pins */
> +	for (i = 0; i < spec->num_pins; i++) {
> +		if (!test_bit(i, &spec->pcm_bitmap))
> +			return i;
> +	}
> +	return -EBUSY;
> +}
> +
> +static void hdmi_attach_hda_pcm(struct hdmi_spec *spec,
> +				struct hdmi_spec_per_pin *per_pin)
> +{
> +	int idx;
> +
> +	/* pcm already be attached to the pin */
> +	if (per_pin->pcm)
> +		return;
> +	idx = hdmi_find_pcm_slot(spec, per_pin);
> +	if (idx == -ENODEV)
> +		return;
> +	per_pin->pcm_idx = idx;
> +	per_pin->pcm = spec->pcm_rec[idx];
> +	set_bit(idx, &spec->pcm_bitmap);
> +}
> +
> +static void hdmi_detach_hda_pcm(struct hdmi_spec *spec,
> +				struct hdmi_spec_per_pin *per_pin)
> +{
> +	int idx;
> +
> +	/* pcm already be detached from the pin */
> +	if (!per_pin->pcm)
> +		return;
> +	idx = per_pin->pcm_idx;
> +	per_pin->pcm_idx = -1;
> +	per_pin->pcm = NULL;
> +	if (idx >= 0 && idx < spec->pcm_used)
> +		clear_bit(idx, &spec->pcm_bitmap);
> +}
> +
>  /* update ELD and jack state via audio component */
>  static void sync_eld_via_acomp(struct hda_codec *codec,
>  			       struct hdmi_spec_per_pin *per_pin)
> @@ -1794,6 +1853,7 @@ static void sync_eld_via_acomp(struct hda_codec *codec,
>  	struct hdmi_eld *eld = &spec->temp_eld;
>  	int size;
>  
> +	mutex_lock(&spec->pcm_lock);
>  	mutex_lock(&per_pin->lock);
>  	size = snd_hdac_acomp_get_eld(&codec->bus->core, per_pin->pin_nid,
>  				      &eld->monitor_present, eld->eld_buffer,
> @@ -1815,11 +1875,19 @@ static void sync_eld_via_acomp(struct hda_codec *codec,
>  		eld->eld_size = 0;
>  	}
>  
> +	if (spec->dyn_pcm_assign) {
> +		if (eld->eld_valid)
> +			hdmi_attach_hda_pcm(spec, per_pin);
> +		else
> +			hdmi_detach_hda_pcm(spec, per_pin);
> +	}
> +
>  	update_eld(codec, per_pin, eld);
>  	snd_jack_report(per_pin->acomp_jack,
>  			eld->monitor_present ? SND_JACK_AVOUT : 0);
>   unlock:
>  	mutex_unlock(&per_pin->lock);
> +	mutex_unlock(&spec->pcm_lock);
>  }
>  
>  static bool hdmi_present_sense(struct hdmi_spec_per_pin *per_pin, int repoll)
> @@ -1875,6 +1943,11 @@ static int hdmi_add_pin(struct hda_codec *codec, hda_nid_t pin_nid)
>  
>  	per_pin->pin_nid = pin_nid;
>  	per_pin->non_pcm = false;
> +	if (spec->dyn_pcm_assign)
> +		per_pin->pcm_idx = -1;
> +	else
> +		per_pin->pcm_idx = pin_idx;
> +	per_pin->pin_nid_idx = pin_idx;
>  
>  	err = hdmi_read_pin_conn(codec, pin_idx);
>  	if (err < 0)
> -- 
> 1.9.1
> 

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Takashi Iwai]

On Wed, 16 Dec 2015 06:42:40 +0100,
libin.yang@linux.intel.com wrote:
> 
> From: Libin Yang <libin.yang@linux.intel.com>
> 
> The patches are trying to support dynamically binding PCM to pin
> in HDMI/DP audio.
> 
> The pin will not bind to any PCM if there is not monitor connected.
> When there is a monitor connected to the port (pin), driver will
> try to find a proper PCM to assign to the port and setup the pin
> if necessary.
> 
> Libin Yang (5):
>   ALSA: hda - hdmi begin to support dynamic PCM assignment
>   ALSA: hda - hdmi playback without monitor in dynamic pcm bind mode
>   ALSA: hda - hdmi operate spdif based on pcm
>   ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
>   ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
>     assignment mode

I updated topic/hda-mst branch with these patches.  But the patch 4
needs more fixes, as suggested in another mail.

Overall, I'd like to know whether MST audio really works in the
current state, or not.  Do you have *any* git branch where MST audio
is supposed to work at all?


thanks,

Takashi

Re: [PATCH v4 4/5] ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug

[Libin Yang]

On 12/16/2015 03:43 PM, Takashi Iwai wrote:
> On Wed, 16 Dec 2015 06:42:44 +0100,
> libin.yang@linux.intel.com wrote:
>>
>> From: Libin Yang <libin.yang@linux.intel.com>
>>
>> Dynamically bind/unbind the PCM to pin when HDMI/DP monitor hotplug.
>>
>> When monitor is connected, find a proper PCM for the monitor.
>> When monitor is disconnected, unbind the PCM from the pin.
>>
>> The binding policy (use Intel platform as example) is:
>> 1. Try to use the legacy pin-pcm mapping for the device entry 0
>>     of the pin.
>> 2. If step 1 fails, try to bind pin to the backup PCMs. For example,
>>     on Intel platform, if DP MST is enabled, 5 PCMs will be created.
>>     PCM 3, PCM 7, PCM 8 are supposed to be used by device entry 0 of
>>     pin 5, pin 6 and pin 7. PCM 9 and PCM 10 are the backup PCMs.
>> 3. If step 2 fails, try to find any PCM to bind to the pin.
>>
>> Signed-off-by: Libin Yang <libin.yang@linux.intel.com>
>
> This patch adds the code in sync_eld_via_acomp(), i.e. it's handled
> only with component.  As the dynamic HDA attach/detach itself is
> agnostic to the component, I'd suggest to put this rather in
> update_eld().
>
> Also pcm_lock can be put in hdmi_present_sense() to wrap both
> sync_eld_via_acomp() and hdmi_present_sense_via_verbs().

OK, I will put it in update_eld().

Best Regards,
Libin

>
>
> Takashi
>
>> ---
>>   sound/pci/hda/patch_hdmi.c | 73 ++++++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 73 insertions(+)
>>
>> diff --git a/sound/pci/hda/patch_hdmi.c b/sound/pci/hda/patch_hdmi.c
>> index 19bf782..10edab8 100644
>> --- a/sound/pci/hda/patch_hdmi.c
>> +++ b/sound/pci/hda/patch_hdmi.c
>> @@ -73,6 +73,8 @@ struct hdmi_spec_per_cvt {
>>
>>   struct hdmi_spec_per_pin {
>>   	hda_nid_t pin_nid;
>> +	/* pin idx, different device entries on the same pin use the same idx */
>> +	int pin_nid_idx;
>>   	int num_mux_nids;
>>   	hda_nid_t mux_nids[HDA_MAX_CONNECTIONS];
>>   	int mux_idx;
>> @@ -85,6 +87,7 @@ struct hdmi_spec_per_pin {
>>   	struct snd_kcontrol *eld_ctl;
>>   	struct snd_jack *acomp_jack; /* jack via audio component */
>>   	struct hda_pcm *pcm; /* pointer to spec->pcm_rec[n] dynamically*/
>> +	int pcm_idx; /* which pcm is attached. -1 means no pcm is attached */
>>   	int repoll_count;
>>   	bool setup; /* the stream has been set up by prepare callback */
>>   	int channels; /* current number of channels */
>> @@ -138,6 +141,8 @@ struct hdmi_spec {
>>   	struct snd_array pins; /* struct hdmi_spec_per_pin */
>>   	struct hda_pcm *pcm_rec[16];
>>   	struct mutex pcm_lock;
>> +	/* pcm_bitmap means which pcms have been assigned to pins*/
>> +	unsigned long pcm_bitmap;
>>   	int pcm_used;	/* counter of pcm_rec[] */
>>   	unsigned int channels_max; /* max over all cvts */
>>
>> @@ -1786,6 +1791,60 @@ static bool hdmi_present_sense_via_verbs(struct hdmi_spec_per_pin *per_pin,
>>   	return ret;
>>   }
>>
>> +static int hdmi_find_pcm_slot(struct hdmi_spec *spec,
>> +				struct hdmi_spec_per_pin *per_pin)
>> +{
>> +	int i;
>> +
>> +	/* try the prefer PCM */
>> +	if (!test_bit(per_pin->pin_nid_idx, &spec->pcm_bitmap))
>> +		return per_pin->pin_nid_idx;
>> +
>> +	/* have a second try; check the "reserved area" over num_pins */
>> +	for (i = spec->num_pins; i < spec->pcm_used; i++) {
>> +		if (!test_bit(i, &spec->pcm_bitmap))
>> +			return i;
>> +	}
>> +
>> +	/* the last try; check the empty slots in pins */
>> +	for (i = 0; i < spec->num_pins; i++) {
>> +		if (!test_bit(i, &spec->pcm_bitmap))
>> +			return i;
>> +	}
>> +	return -EBUSY;
>> +}
>> +
>> +static void hdmi_attach_hda_pcm(struct hdmi_spec *spec,
>> +				struct hdmi_spec_per_pin *per_pin)
>> +{
>> +	int idx;
>> +
>> +	/* pcm already be attached to the pin */
>> +	if (per_pin->pcm)
>> +		return;
>> +	idx = hdmi_find_pcm_slot(spec, per_pin);
>> +	if (idx == -ENODEV)
>> +		return;
>> +	per_pin->pcm_idx = idx;
>> +	per_pin->pcm = spec->pcm_rec[idx];
>> +	set_bit(idx, &spec->pcm_bitmap);
>> +}
>> +
>> +static void hdmi_detach_hda_pcm(struct hdmi_spec *spec,
>> +				struct hdmi_spec_per_pin *per_pin)
>> +{
>> +	int idx;
>> +
>> +	/* pcm already be detached from the pin */
>> +	if (!per_pin->pcm)
>> +		return;
>> +	idx = per_pin->pcm_idx;
>> +	per_pin->pcm_idx = -1;
>> +	per_pin->pcm = NULL;
>> +	if (idx >= 0 && idx < spec->pcm_used)
>> +		clear_bit(idx, &spec->pcm_bitmap);
>> +}
>> +
>>   /* update ELD and jack state via audio component */
>>   static void sync_eld_via_acomp(struct hda_codec *codec,
>>   			       struct hdmi_spec_per_pin *per_pin)
>> @@ -1794,6 +1853,7 @@ static void sync_eld_via_acomp(struct hda_codec *codec,
>>   	struct hdmi_eld *eld = &spec->temp_eld;
>>   	int size;
>>
>> +	mutex_lock(&spec->pcm_lock);
>>   	mutex_lock(&per_pin->lock);
>>   	size = snd_hdac_acomp_get_eld(&codec->bus->core, per_pin->pin_nid,
>>   				      &eld->monitor_present, eld->eld_buffer,
>> @@ -1815,11 +1875,19 @@ static void sync_eld_via_acomp(struct hda_codec *codec,
>>   		eld->eld_size = 0;
>>   	}
>>
>> +	if (spec->dyn_pcm_assign) {
>> +		if (eld->eld_valid)
>> +			hdmi_attach_hda_pcm(spec, per_pin);
>> +		else
>> +			hdmi_detach_hda_pcm(spec, per_pin);
>> +	}
>> +
>>   	update_eld(codec, per_pin, eld);
>>   	snd_jack_report(per_pin->acomp_jack,
>>   			eld->monitor_present ? SND_JACK_AVOUT : 0);
>>    unlock:
>>   	mutex_unlock(&per_pin->lock);
>> +	mutex_unlock(&spec->pcm_lock);
>>   }
>>
>>   static bool hdmi_present_sense(struct hdmi_spec_per_pin *per_pin, int repoll)
>> @@ -1875,6 +1943,11 @@ static int hdmi_add_pin(struct hda_codec *codec, hda_nid_t pin_nid)
>>
>>   	per_pin->pin_nid = pin_nid;
>>   	per_pin->non_pcm = false;
>> +	if (spec->dyn_pcm_assign)
>> +		per_pin->pcm_idx = -1;
>> +	else
>> +		per_pin->pcm_idx = pin_idx;
>> +	per_pin->pin_nid_idx = pin_idx;
>>
>>   	err = hdmi_read_pin_conn(codec, pin_idx);
>>   	if (err < 0)
>> --
>> 1.9.1
>>

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Libin Yang]

On 12/16/2015 03:47 PM, Takashi Iwai wrote:
> On Wed, 16 Dec 2015 06:42:40 +0100,
> libin.yang@linux.intel.com wrote:
>>
>> From: Libin Yang <libin.yang@linux.intel.com>
>>
>> The patches are trying to support dynamically binding PCM to pin
>> in HDMI/DP audio.
>>
>> The pin will not bind to any PCM if there is not monitor connected.
>> When there is a monitor connected to the port (pin), driver will
>> try to find a proper PCM to assign to the port and setup the pin
>> if necessary.
>>
>> Libin Yang (5):
>>    ALSA: hda - hdmi begin to support dynamic PCM assignment
>>    ALSA: hda - hdmi playback without monitor in dynamic pcm bind mode
>>    ALSA: hda - hdmi operate spdif based on pcm
>>    ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
>>    ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
>>      assignment mode
>
> I updated topic/hda-mst branch with these patches.  But the patch 4
> needs more fixes, as suggested in another mail.

Thanks. Do I need to resend all the patches? Or just patch 4 and patch 5?

>
> Overall, I'd like to know whether MST audio really works in the
> current state, or not.  Do you have *any* git branch where MST audio
> is supposed to work at all?

Yes, I have all the patches in my local branch. We tested the patches 
before. But since a lot of code in mainline is changed, I need rebase 
the patches. Some patches are conflicted with the jack changes you 
made. Some also fix some unsol event issues, which seems you have made 
a lot of changes on it currently.

Regards,
Libin

>
>
> thanks,
>
> Takashi
> _______________________________________________
> Alsa-devel mailing list
> Alsa-devel@alsa-project.org
> http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
>

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Takashi Iwai]

On Wed, 16 Dec 2015 08:54:50 +0100,
Libin Yang wrote:
> 
> 
> 
> On 12/16/2015 03:47 PM, Takashi Iwai wrote:
> > On Wed, 16 Dec 2015 06:42:40 +0100,
> > libin.yang@linux.intel.com wrote:
> >>
> >> From: Libin Yang <libin.yang@linux.intel.com>
> >>
> >> The patches are trying to support dynamically binding PCM to pin
> >> in HDMI/DP audio.
> >>
> >> The pin will not bind to any PCM if there is not monitor connected.
> >> When there is a monitor connected to the port (pin), driver will
> >> try to find a proper PCM to assign to the port and setup the pin
> >> if necessary.
> >>
> >> Libin Yang (5):
> >>    ALSA: hda - hdmi begin to support dynamic PCM assignment
> >>    ALSA: hda - hdmi playback without monitor in dynamic pcm bind mode
> >>    ALSA: hda - hdmi operate spdif based on pcm
> >>    ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
> >>    ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
> >>      assignment mode
> >
> > I updated topic/hda-mst branch with these patches.  But the patch 4
> > needs more fixes, as suggested in another mail.
> 
> Thanks. Do I need to resend all the patches? Or just patch 4 and patch 5?

Only 4 (and 5 if needed) should suffice.

> > Overall, I'd like to know whether MST audio really works in the
> > current state, or not.  Do you have *any* git branch where MST audio
> > is supposed to work at all?
> 
> Yes, I have all the patches in my local branch. We tested the patches 
> before. But since a lot of code in mainline is changed, I need rebase 
> the patches. Some patches are conflicted with the jack changes you 
> made. Some also fix some unsol event issues, which seems you have made 
> a lot of changes on it currently.

Well, it doesn't matter how old it is, but I need the branch to test
whether MST really works at all.  Could you publish your git tree and
your old branch that worked?  You can use github or whatever.


thanks,

Takashi

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Libin Yang]

On 12/16/2015 03:59 PM, Takashi Iwai wrote:
> On Wed, 16 Dec 2015 08:54:50 +0100,
> Libin Yang wrote:
>>
>>
>>
>> On 12/16/2015 03:47 PM, Takashi Iwai wrote:
>>> On Wed, 16 Dec 2015 06:42:40 +0100,
>>> libin.yang@linux.intel.com wrote:
>>>>
>>>> From: Libin Yang <libin.yang@linux.intel.com>
>>>>
>>>> The patches are trying to support dynamically binding PCM to pin
>>>> in HDMI/DP audio.
>>>>
>>>> The pin will not bind to any PCM if there is not monitor connected.
>>>> When there is a monitor connected to the port (pin), driver will
>>>> try to find a proper PCM to assign to the port and setup the pin
>>>> if necessary.
>>>>
>>>> Libin Yang (5):
>>>>     ALSA: hda - hdmi begin to support dynamic PCM assignment
>>>>     ALSA: hda - hdmi playback without monitor in dynamic pcm bind mode
>>>>     ALSA: hda - hdmi operate spdif based on pcm
>>>>     ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
>>>>     ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
>>>>       assignment mode
>>>
>>> I updated topic/hda-mst branch with these patches.  But the patch 4
>>> needs more fixes, as suggested in another mail.
>>
>> Thanks. Do I need to resend all the patches? Or just patch 4 and patch 5?
>
> Only 4 (and 5 if needed) should suffice.
>
>>> Overall, I'd like to know whether MST audio really works in the
>>> current state, or not.  Do you have *any* git branch where MST audio
>>> is supposed to work at all?
>>
>> Yes, I have all the patches in my local branch. We tested the patches
>> before. But since a lot of code in mainline is changed, I need rebase
>> the patches. Some patches are conflicted with the jack changes you
>> made. Some also fix some unsol event issues, which seems you have made
>> a lot of changes on it currently.
>
> Well, it doesn't matter how old it is, but I need the branch to test
> whether MST really works at all.  Could you publish your git tree and
> your old branch that worked?  You can use github or whatever.


OK. I will submit the patches to github and let you know the branch later.

Best Regards,
Libin

>
>
> thanks,
>
> Takashi
>

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Yang, Libin]

Hi Takashi,

> -----Original Message-----
> From: Takashi Iwai [mailto:tiwai@suse.de]
> Sent: Wednesday, December 16, 2015 3:47 PM
> To: libin.yang@linux.intel.com
> Cc: alsa-devel@alsa-project.org; Lin, Mengdong; Yang, Libin
> Subject: Re: [alsa-devel] [PATCH v4 0/5] ALSA: hda - hdmi support
> dynamic pcm assignment
> 
> On Wed, 16 Dec 2015 06:42:40 +0100,
> libin.yang@linux.intel.com wrote:
> >
> > From: Libin Yang <libin.yang@linux.intel.com>
> >
> > The patches are trying to support dynamically binding PCM to pin
> > in HDMI/DP audio.
> >
> > The pin will not bind to any PCM if there is not monitor connected.
> > When there is a monitor connected to the port (pin), driver will
> > try to find a proper PCM to assign to the port and setup the pin
> > if necessary.
> >
> > Libin Yang (5):
> >   ALSA: hda - hdmi begin to support dynamic PCM assignment
> >   ALSA: hda - hdmi playback without monitor in dynamic pcm bind
> mode
> >   ALSA: hda - hdmi operate spdif based on pcm
> >   ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
> >   ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
> >     assignment mode
> 
> I updated topic/hda-mst branch with these patches.  But the patch 4
> needs more fixes, as suggested in another mail.
> 
> Overall, I'd like to know whether MST audio really works in the
> current state, or not.  Do you have *any* git branch where MST audio
> is supposed to work at all?

I have uploaded the patches to: https://github.com/libinyang/linux.git
Branch is mst_virtual_pin_github. 

There is still an issue when doing the mst audio playback:
The later converter selection for pin will overwrite the previous setting.
This is the driver bug, I'm debugging it. 

Regards,
Libin

> 
> 
> thanks,
> 
> Takashi

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Takashi Iwai]

On Fri, 18 Dec 2015 03:53:32 +0100,
Yang, Libin wrote:
> 
> Hi Takashi,
> 
> > -----Original Message-----
> > From: Takashi Iwai [mailto:tiwai@suse.de]
> > Sent: Wednesday, December 16, 2015 3:47 PM
> > To: libin.yang@linux.intel.com
> > Cc: alsa-devel@alsa-project.org; Lin, Mengdong; Yang, Libin
> > Subject: Re: [alsa-devel] [PATCH v4 0/5] ALSA: hda - hdmi support
> > dynamic pcm assignment
> > 
> > On Wed, 16 Dec 2015 06:42:40 +0100,
> > libin.yang@linux.intel.com wrote:
> > >
> > > From: Libin Yang <libin.yang@linux.intel.com>
> > >
> > > The patches are trying to support dynamically binding PCM to pin
> > > in HDMI/DP audio.
> > >
> > > The pin will not bind to any PCM if there is not monitor connected.
> > > When there is a monitor connected to the port (pin), driver will
> > > try to find a proper PCM to assign to the port and setup the pin
> > > if necessary.
> > >
> > > Libin Yang (5):
> > >   ALSA: hda - hdmi begin to support dynamic PCM assignment
> > >   ALSA: hda - hdmi playback without monitor in dynamic pcm bind
> > mode
> > >   ALSA: hda - hdmi operate spdif based on pcm
> > >   ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
> > >   ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
> > >     assignment mode
> > 
> > I updated topic/hda-mst branch with these patches.  But the patch 4
> > needs more fixes, as suggested in another mail.
> > 
> > Overall, I'd like to know whether MST audio really works in the
> > current state, or not.  Do you have *any* git branch where MST audio
> > is supposed to work at all?
> 
> I have uploaded the patches to: https://github.com/libinyang/linux.git
> Branch is mst_virtual_pin_github. 
> 
> There is still an issue when doing the mst audio playback:
> The later converter selection for pin will overwrite the previous setting.
> This is the driver bug, I'm debugging it. 

Thanks.  I tried this branch, but the DP MST audio on my HP laptop
dock doesn't seem working.  It gets no notification (maybe expected),
but the accesses to any HDMI devices result in the error.  I tested
via aplay -Dhdmi:0,0 to 0,2.

With the latest patchset (e.g. my test/hdmi-jack branch) I get the
notification and ELD, but the audio still doesn't come out.  So,
something seems missing.


Takashi

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Takashi Iwai]

On Fri, 18 Dec 2015 11:48:49 +0100,
Takashi Iwai wrote:
> 
> On Fri, 18 Dec 2015 03:53:32 +0100,
> Yang, Libin wrote:
> > 
> > Hi Takashi,
> > 
> > > -----Original Message-----
> > > From: Takashi Iwai [mailto:tiwai@suse.de]
> > > Sent: Wednesday, December 16, 2015 3:47 PM
> > > To: libin.yang@linux.intel.com
> > > Cc: alsa-devel@alsa-project.org; Lin, Mengdong; Yang, Libin
> > > Subject: Re: [alsa-devel] [PATCH v4 0/5] ALSA: hda - hdmi support
> > > dynamic pcm assignment
> > > 
> > > On Wed, 16 Dec 2015 06:42:40 +0100,
> > > libin.yang@linux.intel.com wrote:
> > > >
> > > > From: Libin Yang <libin.yang@linux.intel.com>
> > > >
> > > > The patches are trying to support dynamically binding PCM to pin
> > > > in HDMI/DP audio.
> > > >
> > > > The pin will not bind to any PCM if there is not monitor connected.
> > > > When there is a monitor connected to the port (pin), driver will
> > > > try to find a proper PCM to assign to the port and setup the pin
> > > > if necessary.
> > > >
> > > > Libin Yang (5):
> > > >   ALSA: hda - hdmi begin to support dynamic PCM assignment
> > > >   ALSA: hda - hdmi playback without monitor in dynamic pcm bind
> > > mode
> > > >   ALSA: hda - hdmi operate spdif based on pcm
> > > >   ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
> > > >   ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
> > > >     assignment mode
> > > 
> > > I updated topic/hda-mst branch with these patches.  But the patch 4
> > > needs more fixes, as suggested in another mail.
> > > 
> > > Overall, I'd like to know whether MST audio really works in the
> > > current state, or not.  Do you have *any* git branch where MST audio
> > > is supposed to work at all?
> > 
> > I have uploaded the patches to: https://github.com/libinyang/linux.git
> > Branch is mst_virtual_pin_github. 
> > 
> > There is still an issue when doing the mst audio playback:
> > The later converter selection for pin will overwrite the previous setting.
> > This is the driver bug, I'm debugging it. 
> 
> Thanks.  I tried this branch, but the DP MST audio on my HP laptop
> dock doesn't seem working.  It gets no notification (maybe expected),
> but the accesses to any HDMI devices result in the error.  I tested
> via aplay -Dhdmi:0,0 to 0,2.
> 
> With the latest patchset (e.g. my test/hdmi-jack branch) I get the
> notification and ELD, but the audio still doesn't come out.  So,
> something seems missing.

Actually I'm thinking how the MST things are supposed to work here.
The machine here has a docking station connected to port D, and it has
to DP connectors.  Meanwhile it has a built-in DP port on port C.
The latter works fine while dock ports don't.

Suppose if I plug two monitors on the docking station.  How would the
audio in HD-audio side be set up?


thanks,

Takashi

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Libin Yang]

Hi Takashi,

On 12/18/2015 08:05 PM, Takashi Iwai wrote:
> On Fri, 18 Dec 2015 11:48:49 +0100,
> Takashi Iwai wrote:
>>
>> On Fri, 18 Dec 2015 03:53:32 +0100,
>> Yang, Libin wrote:
>>>
>>> Hi Takashi,
>>>
>>>> -----Original Message-----
>>>> From: Takashi Iwai [mailto:tiwai@suse.de]
>>>> Sent: Wednesday, December 16, 2015 3:47 PM
>>>> To: libin.yang@linux.intel.com
>>>> Cc: alsa-devel@alsa-project.org; Lin, Mengdong; Yang, Libin
>>>> Subject: Re: [alsa-devel] [PATCH v4 0/5] ALSA: hda - hdmi support
>>>> dynamic pcm assignment
>>>>
>>>> On Wed, 16 Dec 2015 06:42:40 +0100,
>>>> libin.yang@linux.intel.com wrote:
>>>>>
>>>>> From: Libin Yang <libin.yang@linux.intel.com>
>>>>>
>>>>> The patches are trying to support dynamically binding PCM to pin
>>>>> in HDMI/DP audio.
>>>>>
>>>>> The pin will not bind to any PCM if there is not monitor connected.
>>>>> When there is a monitor connected to the port (pin), driver will
>>>>> try to find a proper PCM to assign to the port and setup the pin
>>>>> if necessary.
>>>>>
>>>>> Libin Yang (5):
>>>>>    ALSA: hda - hdmi begin to support dynamic PCM assignment
>>>>>    ALSA: hda - hdmi playback without monitor in dynamic pcm bind
>>>> mode
>>>>>    ALSA: hda - hdmi operate spdif based on pcm
>>>>>    ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
>>>>>    ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
>>>>>      assignment mode
>>>>
>>>> I updated topic/hda-mst branch with these patches.  But the patch 4
>>>> needs more fixes, as suggested in another mail.
>>>>
>>>> Overall, I'd like to know whether MST audio really works in the
>>>> current state, or not.  Do you have *any* git branch where MST audio
>>>> is supposed to work at all?
>>>
>>> I have uploaded the patches to: https://github.com/libinyang/linux.git
>>> Branch is mst_virtual_pin_github.
>>>
>>> There is still an issue when doing the mst audio playback:
>>> The later converter selection for pin will overwrite the previous setting.
>>> This is the driver bug, I'm debugging it.
>>
>> Thanks.  I tried this branch, but the DP MST audio on my HP laptop
>> dock doesn't seem working.  It gets no notification (maybe expected),
>> but the accesses to any HDMI devices result in the error.  I tested
>> via aplay -Dhdmi:0,0 to 0,2.
>>
>> With the latest patchset (e.g. my test/hdmi-jack branch) I get the
>> notification and ELD, but the audio still doesn't come out.  So,
>> something seems missing.
>
> Actually I'm thinking how the MST things are supposed to work here.
> The machine here has a docking station connected to port D, and it has
> to DP connectors.  Meanwhile it has a built-in DP port on port C.
> The latter works fine while dock ports don't.

I have updated the code last Friday, which fixed a converter setting 
issue in MST mode. Without the latest code, there is no sound for DP 
MST audio.

The port D should be the same with port C for the HW connection.

>
> Suppose if I plug two monitors on the docking station.  How would the
> audio in HD-audio side be set up?

Not sure how the docking station works. If the docking works as it is 
a MST hub, there should be sound independently from both ports. But 
I'm not sure whether the docking is a MST hub or just duplicates 
context to the other port.

I think we can try to get the device entry number on the pin by 
calling snd_hda_get_num_devices().

The steps should be:
1. connect 2 monitors to the docking station and boot up the machine.
2. If xxx_get_num_devices() returns 0, this means it is not MST mode.
    If xxx_get_num_devices() returns 1, this means it works as MST hub.


Best Regards,
Libin
>
>
> thanks,
>
> Takashi
> _______________________________________________
> Alsa-devel mailing list
> Alsa-devel@alsa-project.org
> http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
>

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Takashi Iwai]

On Mon, 21 Dec 2015 01:51:02 +0100,
Libin Yang wrote:
> 
> Hi Takashi,
> 
> On 12/18/2015 08:05 PM, Takashi Iwai wrote:
> > On Fri, 18 Dec 2015 11:48:49 +0100,
> > Takashi Iwai wrote:
> >>
> >> On Fri, 18 Dec 2015 03:53:32 +0100,
> >> Yang, Libin wrote:
> >>>
> >>> Hi Takashi,
> >>>
> >>>> -----Original Message-----
> >>>> From: Takashi Iwai [mailto:tiwai@suse.de]
> >>>> Sent: Wednesday, December 16, 2015 3:47 PM
> >>>> To: libin.yang@linux.intel.com
> >>>> Cc: alsa-devel@alsa-project.org; Lin, Mengdong; Yang, Libin
> >>>> Subject: Re: [alsa-devel] [PATCH v4 0/5] ALSA: hda - hdmi support
> >>>> dynamic pcm assignment
> >>>>
> >>>> On Wed, 16 Dec 2015 06:42:40 +0100,
> >>>> libin.yang@linux.intel.com wrote:
> >>>>>
> >>>>> From: Libin Yang <libin.yang@linux.intel.com>
> >>>>>
> >>>>> The patches are trying to support dynamically binding PCM to pin
> >>>>> in HDMI/DP audio.
> >>>>>
> >>>>> The pin will not bind to any PCM if there is not monitor connected.
> >>>>> When there is a monitor connected to the port (pin), driver will
> >>>>> try to find a proper PCM to assign to the port and setup the pin
> >>>>> if necessary.
> >>>>>
> >>>>> Libin Yang (5):
> >>>>>    ALSA: hda - hdmi begin to support dynamic PCM assignment
> >>>>>    ALSA: hda - hdmi playback without monitor in dynamic pcm bind
> >>>> mode
> >>>>>    ALSA: hda - hdmi operate spdif based on pcm
> >>>>>    ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
> >>>>>    ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
> >>>>>      assignment mode
> >>>>
> >>>> I updated topic/hda-mst branch with these patches.  But the patch 4
> >>>> needs more fixes, as suggested in another mail.
> >>>>
> >>>> Overall, I'd like to know whether MST audio really works in the
> >>>> current state, or not.  Do you have *any* git branch where MST audio
> >>>> is supposed to work at all?
> >>>
> >>> I have uploaded the patches to: https://github.com/libinyang/linux.git
> >>> Branch is mst_virtual_pin_github.
> >>>
> >>> There is still an issue when doing the mst audio playback:
> >>> The later converter selection for pin will overwrite the previous setting.
> >>> This is the driver bug, I'm debugging it.
> >>
> >> Thanks.  I tried this branch, but the DP MST audio on my HP laptop
> >> dock doesn't seem working.  It gets no notification (maybe expected),
> >> but the accesses to any HDMI devices result in the error.  I tested
> >> via aplay -Dhdmi:0,0 to 0,2.
> >>
> >> With the latest patchset (e.g. my test/hdmi-jack branch) I get the
> >> notification and ELD, but the audio still doesn't come out.  So,
> >> something seems missing.
> >
> > Actually I'm thinking how the MST things are supposed to work here.
> > The machine here has a docking station connected to port D, and it has
> > to DP connectors.  Meanwhile it has a built-in DP port on port C.
> > The latter works fine while dock ports don't.
> 
> I have updated the code last Friday, which fixed a converter setting 
> issue in MST mode. Without the latest code, there is no sound for DP 
> MST audio.
> 
> The port D should be the same with port C for the HW connection.

I tested the latest branch, commit
d6203f02941c72e2737271d98cf382f34295307e
    bug fix
    
    1. converter overwritten
    2. not share the converter

This is the right branch and commit for testing, right?

> > Suppose if I plug two monitors on the docking station.  How would the
> > audio in HD-audio side be set up?
> 
> Not sure how the docking station works. If the docking works as it is 
> a MST hub, there should be sound independently from both ports. But 
> I'm not sure whether the docking is a MST hub or just duplicates 
> context to the other port.

It's supposed to be a MST hub.

> I think we can try to get the device entry number on the pin by 
> calling snd_hda_get_num_devices().
> 
> The steps should be:
> 1. connect 2 monitors to the docking station and boot up the machine.
> 2. If xxx_get_num_devices() returns 0, this means it is not MST mode.
>     If xxx_get_num_devices() returns 1, this means it works as MST hub.

Currently I have only one monitor to test, so I can't check this,
unfortunately.


Takashi

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Libin Yang]

On 12/21/2015 04:35 PM, Takashi Iwai wrote:
> On Mon, 21 Dec 2015 01:51:02 +0100,
> Libin Yang wrote:
>>
>> Hi Takashi,
>>
>> On 12/18/2015 08:05 PM, Takashi Iwai wrote:
>>> On Fri, 18 Dec 2015 11:48:49 +0100,
>>> Takashi Iwai wrote:
>>>>
>>>> On Fri, 18 Dec 2015 03:53:32 +0100,
>>>> Yang, Libin wrote:
>>>>>
>>>>> Hi Takashi,
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Takashi Iwai [mailto:tiwai@suse.de]
>>>>>> Sent: Wednesday, December 16, 2015 3:47 PM
>>>>>> To: libin.yang@linux.intel.com
>>>>>> Cc: alsa-devel@alsa-project.org; Lin, Mengdong; Yang, Libin
>>>>>> Subject: Re: [alsa-devel] [PATCH v4 0/5] ALSA: hda - hdmi support
>>>>>> dynamic pcm assignment
>>>>>>
>>>>>> On Wed, 16 Dec 2015 06:42:40 +0100,
>>>>>> libin.yang@linux.intel.com wrote:
>>>>>>>
>>>>>>> From: Libin Yang <libin.yang@linux.intel.com>
>>>>>>>
>>>>>>> The patches are trying to support dynamically binding PCM to pin
>>>>>>> in HDMI/DP audio.
>>>>>>>
>>>>>>> The pin will not bind to any PCM if there is not monitor connected.
>>>>>>> When there is a monitor connected to the port (pin), driver will
>>>>>>> try to find a proper PCM to assign to the port and setup the pin
>>>>>>> if necessary.
>>>>>>>
>>>>>>> Libin Yang (5):
>>>>>>>     ALSA: hda - hdmi begin to support dynamic PCM assignment
>>>>>>>     ALSA: hda - hdmi playback without monitor in dynamic pcm bind
>>>>>> mode
>>>>>>>     ALSA: hda - hdmi operate spdif based on pcm
>>>>>>>     ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
>>>>>>>     ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
>>>>>>>       assignment mode
>>>>>>
>>>>>> I updated topic/hda-mst branch with these patches.  But the patch 4
>>>>>> needs more fixes, as suggested in another mail.
>>>>>>
>>>>>> Overall, I'd like to know whether MST audio really works in the
>>>>>> current state, or not.  Do you have *any* git branch where MST audio
>>>>>> is supposed to work at all?
>>>>>
>>>>> I have uploaded the patches to: https://github.com/libinyang/linux.git
>>>>> Branch is mst_virtual_pin_github.
>>>>>
>>>>> There is still an issue when doing the mst audio playback:
>>>>> The later converter selection for pin will overwrite the previous setting.
>>>>> This is the driver bug, I'm debugging it.
>>>>
>>>> Thanks.  I tried this branch, but the DP MST audio on my HP laptop
>>>> dock doesn't seem working.  It gets no notification (maybe expected),
>>>> but the accesses to any HDMI devices result in the error.  I tested
>>>> via aplay -Dhdmi:0,0 to 0,2.
>>>>
>>>> With the latest patchset (e.g. my test/hdmi-jack branch) I get the
>>>> notification and ELD, but the audio still doesn't come out.  So,
>>>> something seems missing.
>>>
>>> Actually I'm thinking how the MST things are supposed to work here.
>>> The machine here has a docking station connected to port D, and it has
>>> to DP connectors.  Meanwhile it has a built-in DP port on port C.
>>> The latter works fine while dock ports don't.
>>
>> I have updated the code last Friday, which fixed a converter setting
>> issue in MST mode. Without the latest code, there is no sound for DP
>> MST audio.
>>
>> The port D should be the same with port C for the HW connection.
>
> I tested the latest branch, commit
> d6203f02941c72e2737271d98cf382f34295307e
>      bug fix
>
>      1. converter overwritten
>      2. not share the converter
>
> This is the right branch and commit for testing, right?

Yes. this is the latest code. It works on my side.

>
>>> Suppose if I plug two monitors on the docking station.  How would the
>>> audio in HD-audio side be set up?
>>
>> Not sure how the docking station works. If the docking works as it is
>> a MST hub, there should be sound independently from both ports. But
>> I'm not sure whether the docking is a MST hub or just duplicates
>> context to the other port.
>
> It's supposed to be a MST hub.
>
>> I think we can try to get the device entry number on the pin by
>> calling snd_hda_get_num_devices().
>>
>> The steps should be:
>> 1. connect 2 monitors to the docking station and boot up the machine.
>> 2. If xxx_get_num_devices() returns 0, this means it is not MST mode.
>>      If xxx_get_num_devices() returns 1, this means it works as MST hub.
>
> Currently I have only one monitor to test, so I can't check this,
> unfortunately.

If it is a MST hub:
1. when there are 2 monitors are connected to the port, 2 PCM are 
assigned to each monitor. snd_hda_get_num_devices() will return 1. 
When playing, each PCM will find a converter to use.
2. When there is one monitor is connected to the port, 
snd_hda_get_num_devices() will return 0 and one PCM is assigned.

Anyway, if only one monitor is connected, it should always work, no 
matter with or without the DP MST support.

Could you please tell me you HP laptop type? I will debug on it if I 
can get one.

>
>
> Takashi
> _______________________________________________
> Alsa-devel mailing list
> Alsa-devel@alsa-project.org
> http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
>

Re: [PATCH v4 0/5] ALSA: hda - hdmi support dynamic pcm assignment

[Takashi Iwai]

On Mon, 21 Dec 2015 09:54:23 +0100,
Libin Yang wrote:
> 
> 
> On 12/21/2015 04:35 PM, Takashi Iwai wrote:
> > On Mon, 21 Dec 2015 01:51:02 +0100,
> > Libin Yang wrote:
> >>
> >> Hi Takashi,
> >>
> >> On 12/18/2015 08:05 PM, Takashi Iwai wrote:
> >>> On Fri, 18 Dec 2015 11:48:49 +0100,
> >>> Takashi Iwai wrote:
> >>>>
> >>>> On Fri, 18 Dec 2015 03:53:32 +0100,
> >>>> Yang, Libin wrote:
> >>>>>
> >>>>> Hi Takashi,
> >>>>>
> >>>>>> -----Original Message-----
> >>>>>> From: Takashi Iwai [mailto:tiwai@suse.de]
> >>>>>> Sent: Wednesday, December 16, 2015 3:47 PM
> >>>>>> To: libin.yang@linux.intel.com
> >>>>>> Cc: alsa-devel@alsa-project.org; Lin, Mengdong; Yang, Libin
> >>>>>> Subject: Re: [alsa-devel] [PATCH v4 0/5] ALSA: hda - hdmi support
> >>>>>> dynamic pcm assignment
> >>>>>>
> >>>>>> On Wed, 16 Dec 2015 06:42:40 +0100,
> >>>>>> libin.yang@linux.intel.com wrote:
> >>>>>>>
> >>>>>>> From: Libin Yang <libin.yang@linux.intel.com>
> >>>>>>>
> >>>>>>> The patches are trying to support dynamically binding PCM to pin
> >>>>>>> in HDMI/DP audio.
> >>>>>>>
> >>>>>>> The pin will not bind to any PCM if there is not monitor connected.
> >>>>>>> When there is a monitor connected to the port (pin), driver will
> >>>>>>> try to find a proper PCM to assign to the port and setup the pin
> >>>>>>> if necessary.
> >>>>>>>
> >>>>>>> Libin Yang (5):
> >>>>>>>     ALSA: hda - hdmi begin to support dynamic PCM assignment
> >>>>>>>     ALSA: hda - hdmi playback without monitor in dynamic pcm bind
> >>>>>> mode
> >>>>>>>     ALSA: hda - hdmi operate spdif based on pcm
> >>>>>>>     ALSA: hda - hdmi dynamically bind PCM to pin when monitor hotplug
> >>>>>>>     ALSA: hda - hdmi setup pin when monitor hotplug in pcm dynamic
> >>>>>>>       assignment mode
> >>>>>>
> >>>>>> I updated topic/hda-mst branch with these patches.  But the patch 4
> >>>>>> needs more fixes, as suggested in another mail.
> >>>>>>
> >>>>>> Overall, I'd like to know whether MST audio really works in the
> >>>>>> current state, or not.  Do you have *any* git branch where MST audio
> >>>>>> is supposed to work at all?
> >>>>>
> >>>>> I have uploaded the patches to: https://github.com/libinyang/linux.git
> >>>>> Branch is mst_virtual_pin_github.
> >>>>>
> >>>>> There is still an issue when doing the mst audio playback:
> >>>>> The later converter selection for pin will overwrite the previous setting.
> >>>>> This is the driver bug, I'm debugging it.
> >>>>
> >>>> Thanks.  I tried this branch, but the DP MST audio on my HP laptop
> >>>> dock doesn't seem working.  It gets no notification (maybe expected),
> >>>> but the accesses to any HDMI devices result in the error.  I tested
> >>>> via aplay -Dhdmi:0,0 to 0,2.
> >>>>
> >>>> With the latest patchset (e.g. my test/hdmi-jack branch) I get the
> >>>> notification and ELD, but the audio still doesn't come out.  So,
> >>>> something seems missing.
> >>>
> >>> Actually I'm thinking how the MST things are supposed to work here.
> >>> The machine here has a docking station connected to port D, and it has
> >>> to DP connectors.  Meanwhile it has a built-in DP port on port C.
> >>> The latter works fine while dock ports don't.
> >>
> >> I have updated the code last Friday, which fixed a converter setting
> >> issue in MST mode. Without the latest code, there is no sound for DP
> >> MST audio.
> >>
> >> The port D should be the same with port C for the HW connection.
> >
> > I tested the latest branch, commit
> > d6203f02941c72e2737271d98cf382f34295307e
> >      bug fix
> >
> >      1. converter overwritten
> >      2. not share the converter
> >
> > This is the right branch and commit for testing, right?
> 
> Yes. this is the latest code. It works on my side.
> 
> >
> >>> Suppose if I plug two monitors on the docking station.  How would the
> >>> audio in HD-audio side be set up?
> >>
> >> Not sure how the docking station works. If the docking works as it is
> >> a MST hub, there should be sound independently from both ports. But
> >> I'm not sure whether the docking is a MST hub or just duplicates
> >> context to the other port.
> >
> > It's supposed to be a MST hub.
> >
> >> I think we can try to get the device entry number on the pin by
> >> calling snd_hda_get_num_devices().
> >>
> >> The steps should be:
> >> 1. connect 2 monitors to the docking station and boot up the machine.
> >> 2. If xxx_get_num_devices() returns 0, this means it is not MST mode.
> >>      If xxx_get_num_devices() returns 1, this means it works as MST hub.
> >
> > Currently I have only one monitor to test, so I can't check this,
> > unfortunately.
> 
> If it is a MST hub:
> 1. when there are 2 monitors are connected to the port, 2 PCM are 
> assigned to each monitor. snd_hda_get_num_devices() will return 1. 
> When playing, each PCM will find a converter to use.
> 2. When there is one monitor is connected to the port, 
> snd_hda_get_num_devices() will return 0 and one PCM is assigned.
> 
> Anyway, if only one monitor is connected, it should always work, no 
> matter with or without the DP MST support.
> 
> Could you please tell me you HP laptop type? I will debug on it if I 
> can get one.

I'm testing HP Folio 940 with Haswell.


Takashi