Skip to contentSkip to conversation

support DP MST audio

RFC PATCH Superseded [email protected] 3 participants 14 messages Sep 26, 2016 – Sep 28, 2016 Sound

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

cover support DP MST audio

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

 sound/pci/hda/hda_codec.c      | 129 ++++++++++++++++++-----
 sound/pci/hda/hda_codec.h      |  18 ++--
 sound/pci/hda/hda_generic.c    |  18 ++--
 sound/pci/hda/hda_proc.c       |   2 +-
 sound/pci/hda/patch_analog.c   |  18 ++--
 sound/pci/hda/patch_hdmi.c     | 228 +++++++++++++++++++++++++++++++----------
 sound/pci/hda/patch_realtek.c  |  22 ++--
 sound/pci/hda/patch_sigmatel.c |   4 +-
 sound/pci/hda/patch_via.c      |  10 +-
 9 files changed, 328 insertions(+), 121 deletions(-)
Signature · 1 line
1.9.1
#1 lore raw

RFC PATCH 1/3 ALSA: hda - codec add DP MST support for connection list

From: Libin Yang <[email protected]>

This patches adds the support of connection list for DP MST. With this, hdmi driver in DP MST mode can easily reuse the connection list mechanism.

---
 sound/pci/hda/hda_codec.c      | 52 +++++++++++++++++++++++++++---------------
 sound/pci/hda/hda_codec.h      | 15 ++++++------
 sound/pci/hda/hda_generic.c    | 18 +++++++--------
 sound/pci/hda/hda_proc.c       |  2 +-
 sound/pci/hda/patch_analog.c   | 18 ++++++++-------
 sound/pci/hda/patch_hdmi.c     |  8 ++++---
 sound/pci/hda/patch_realtek.c  | 22 +++++++++---------
 sound/pci/hda/patch_sigmatel.c |  4 ++--
 sound/pci/hda/patch_via.c      | 10 ++++----
 9 files changed, 85 insertions(+), 64 deletions(-)
sound/pci/hda/hda_codec.c +33 -19
index 9913be8..32e21e8 100644
@@ -110,23 +110,24 @@ struct hda_conn_list {
 	struct list_head list;
 	int len;
 	hda_nid_t nid;
+	int dev_id;
 	hda_nid_t conns[0];
 };
 
 /* look up the cached results */
 static struct hda_conn_list *
-lookup_conn_list(struct hda_codec *codec, hda_nid_t nid)
+lookup_conn_list(struct hda_codec *codec, hda_nid_t nid, int dev_id)
 {
 	struct hda_conn_list *p;
 	list_for_each_entry(p, &codec->conn_list, list) {
-		if (p->nid == nid)
+		if ((p->nid == nid) && (p->dev_id == dev_id))
 			return p;
 	}
 	return NULL;
 }
 
-static int add_conn_list(struct hda_codec *codec, hda_nid_t nid, int len,
-			 const hda_nid_t *list)
+static int add_conn_list(struct hda_codec *codec, hda_nid_t nid,
+			 int dev_id, int len, const hda_nid_t *list)
 {
 	struct hda_conn_list *p;
 
@@ -135,6 +136,7 @@ static int add_conn_list(struct hda_codec *codec, hda_nid_t nid, int len,
 		return -ENOMEM;
 	p->len = len;
 	p->nid = nid;
+	p->dev_id = dev_id;
 	memcpy(p->conns, list, len * sizeof(hda_nid_t));
 	list_add(&p->list, &codec->conn_list);
 	return 0;
@@ -150,8 +152,13 @@ static void remove_conn_list(struct hda_codec *codec)
 	}
 }
 
-/* read the connection and add to the cache */
-static int read_and_add_raw_conns(struct hda_codec *codec, hda_nid_t nid)
+/*
+ * read the connection and add to the cache
+ * the caller should select the device entry by sending the
+ * corresponding verb if necessary before calling this function
+ */
+static int read_and_add_raw_conns(struct hda_codec *codec, hda_nid_t nid,
+				  int dev_id)
 {
 	hda_nid_t list[32];
 	hda_nid_t *result = list;
@@ -166,7 +173,8 @@ static int read_and_add_raw_conns(struct hda_codec *codec, hda_nid_t nid)
 		len = snd_hda_get_raw_connections(codec, nid, result, len);
 	}
 	if (len >= 0)
-		len = snd_hda_override_conn_list(codec, nid, len, result);
+		len = snd_hda_override_conn_list(codec, nid, dev_id,
+						 len, result);
 	if (result != list)
 		kfree(result);
 	return len;
@@ -176,6 +184,7 @@ static int read_and_add_raw_conns(struct hda_codec *codec, hda_nid_t nid)
  * snd_hda_get_conn_list - get connection list
  * @codec: the HDA codec
  * @nid: NID to parse
+ * @dev_id: device entry id
  * @listp: the pointer to store NID list
  *
  * Parses the connection list of the given widget and stores the pointer
@@ -188,7 +197,7 @@ static int read_and_add_raw_conns(struct hda_codec *codec, hda_nid_t nid)
  * concurrently, protect with a mutex appropriately.
  */
 int snd_hda_get_conn_list(struct hda_codec *codec, hda_nid_t nid,
-			  const hda_nid_t **listp)
+			  int dev_id, const hda_nid_t **listp)
 {
 	bool added = false;
 
@@ -197,7 +206,7 @@ int snd_hda_get_conn_list(struct hda_codec *codec, hda_nid_t nid,
 		const struct hda_conn_list *p;
 
 		/* if the connection-list is already cached, read it */
-		p = lookup_conn_list(codec, nid);
+		p = lookup_conn_list(codec, nid, dev_id);
 		if (p) {
 			if (listp)
 				*listp = p->conns;
@@ -206,7 +215,7 @@ int snd_hda_get_conn_list(struct hda_codec *codec, hda_nid_t nid,
 		if (snd_BUG_ON(added))
 			return -EINVAL;
 
-		err = read_and_add_raw_conns(codec, nid);
+		err = read_and_add_raw_conns(codec, nid, dev_id);
 		if (err < 0)
 			return err;
 		added = true;
@@ -218,6 +227,7 @@ EXPORT_SYMBOL_GPL(snd_hda_get_conn_list);
  * snd_hda_get_connections - copy connection list
  * @codec: the HDA codec
  * @nid: NID to parse
+ * @dev_id: device entry id
  * @conn_list: connection list array; when NULL, checks only the size
  * @max_conns: max. number of connections to store
  *
@@ -227,10 +237,11 @@ EXPORT_SYMBOL_GPL(snd_hda_get_conn_list);
  * Returns the number of connections, or a negative error code.
  */
 int snd_hda_get_connections(struct hda_codec *codec, hda_nid_t nid,
-			    hda_nid_t *conn_list, int max_conns)
+			    int dev_id, hda_nid_t *conn_list,
+			    int max_conns)
 {
 	const hda_nid_t *list;
-	int len = snd_hda_get_conn_list(codec, nid, &list);
+	int len = snd_hda_get_conn_list(codec, nid, dev_id, &list);
 
 	if (len > 0 && conn_list) {
 		if (len > max_conns) {
@@ -249,6 +260,7 @@ EXPORT_SYMBOL_GPL(snd_hda_get_connections);
  * snd_hda_override_conn_list - add/modify the connection-list to cache
  * @codec: the HDA codec
  * @nid: NID to parse
+ * @dev_id: device entry id
  * @len: number of connection list entries
  * @list: the list of connection entries
  *
@@ -257,18 +269,18 @@ EXPORT_SYMBOL_GPL(snd_hda_get_connections);
  *
  * Returns zero or a negative error code.
  */
-int snd_hda_override_conn_list(struct hda_codec *codec, hda_nid_t nid, int len,
-			       const hda_nid_t *list)
+int snd_hda_override_conn_list(struct hda_codec *codec, hda_nid_t nid,
+			       int dev_id, int len, const hda_nid_t *list)
 {
 	struct hda_conn_list *p;
 
-	p = lookup_conn_list(codec, nid);
+	p = lookup_conn_list(codec, nid, dev_id);
 	if (p) {
 		list_del(&p->list);
 		kfree(p);
 	}
 
-	return add_conn_list(codec, nid, len, list);
+	return add_conn_list(codec, nid, dev_id, len, list);
 }
 EXPORT_SYMBOL_GPL(snd_hda_override_conn_list);
 
@@ -277,6 +289,7 @@ EXPORT_SYMBOL_GPL(snd_hda_override_conn_list);
  * @codec: the HDA codec
  * @mux: NID containing the list
  * @nid: NID to select
+ * @dev_id: device entry id
  * @recursive: 1 when searching NID recursively, otherwise 0
  *
  * Parses the connection list of the widget @mux and checks whether the
@@ -284,12 +297,12 @@ EXPORT_SYMBOL_GPL(snd_hda_override_conn_list);
  * Otherwise it returns -1.
  */
 int snd_hda_get_conn_index(struct hda_codec *codec, hda_nid_t mux,
-			   hda_nid_t nid, int recursive)
+			   hda_nid_t nid, int dev_id, int recursive)
 {
 	const hda_nid_t *conn;
 	int i, nums;
 
-	nums = snd_hda_get_conn_list(codec, mux, &conn);
+	nums = snd_hda_get_conn_list(codec, mux, dev_id, &conn);
 	for (i = 0; i < nums; i++)
 		if (conn[i] == nid)
 			return i;
@@ -304,7 +317,8 @@ int snd_hda_get_conn_index(struct hda_codec *codec, hda_nid_t mux,
 		unsigned int type = get_wcaps_type(get_wcaps(codec, conn[i]));
 		if (type == AC_WID_PIN || type == AC_WID_AUD_OUT)
 			continue;
-		if (snd_hda_get_conn_index(codec, conn[i], nid, recursive) >= 0)
+		if (snd_hda_get_conn_index(codec, conn[i], nid, dev_id,
+					   recursive) >= 0)
 			return i;
 	}
 	return -1;
sound/pci/hda/hda_codec.h +8 -7
index 373fcad..4852c1a 100644
@@ -329,11 +329,12 @@ snd_hda_codec_write(struct hda_codec *codec, hda_nid_t nid, int flags,
 #define snd_hda_get_sub_nodes(codec, nid, start_nid) \
 	snd_hdac_get_sub_nodes(&(codec)->core, nid, start_nid)
 int snd_hda_get_connections(struct hda_codec *codec, hda_nid_t nid,
-			    hda_nid_t *conn_list, int max_conns);
+			    int dev_id, hda_nid_t *conn_list,
+			    int max_conns);
 static inline int
-snd_hda_get_num_conns(struct hda_codec *codec, hda_nid_t nid)
+snd_hda_get_num_conns(struct hda_codec *codec, hda_nid_t nid, int dev_id)
 {
-	return snd_hda_get_connections(codec, nid, NULL, 0);
+	return snd_hda_get_connections(codec, nid, dev_id, NULL, 0);
 }
 
 #define snd_hda_get_raw_connections(codec, nid, list, max_conns) \
@@ -341,12 +342,12 @@ snd_hda_get_num_conns(struct hda_codec *codec, hda_nid_t nid)
 #define snd_hda_get_num_raw_conns(codec, nid) \
 	snd_hdac_get_connections(&(codec)->core, nid, NULL, 0);
 
-int snd_hda_get_conn_list(struct hda_codec *codec, hda_nid_t nid,
+int snd_hda_get_conn_list(struct hda_codec *codec, hda_nid_t nid, int dev_id,
 			  const hda_nid_t **listp);
-int snd_hda_override_conn_list(struct hda_codec *codec, hda_nid_t nid, int nums,
-			  const hda_nid_t *list);
+int snd_hda_override_conn_list(struct hda_codec *codec, hda_nid_t nid,
+			       int dev_id, int nums, const hda_nid_t *list);
 int snd_hda_get_conn_index(struct hda_codec *codec, hda_nid_t mux,
-			   hda_nid_t nid, int recursive);
+			   hda_nid_t nid, int dev_id, int recursive);
 int snd_hda_get_devices(struct hda_codec *codec, hda_nid_t nid,
 			u8 *dev_list, int max_devices);
 
sound/pci/hda/hda_generic.c +9 -9
index e7c8f4f..9a6e67b 100644
@@ -338,7 +338,7 @@ static bool is_reachable_path(struct hda_codec *codec,
 {
 	if (!from_nid || !to_nid)
 		return false;
-	return snd_hda_get_conn_index(codec, to_nid, from_nid, true) >= 0;
+	return snd_hda_get_conn_index(codec, to_nid, from_nid, 0, true) >= 0;
 }
 
 /* nid, dir and idx */
@@ -397,7 +397,7 @@ static bool __parse_nid_path(struct hda_codec *codec,
 	else if (to_nid == (hda_nid_t)(-anchor_nid))
 		return false; /* hit the exclusive nid */
 
-	nums = snd_hda_get_conn_list(codec, to_nid, &conn);
+	nums = snd_hda_get_conn_list(codec, to_nid, 0, &conn);
 	for (i = 0; i < nums; i++) {
 		if (conn[i] != from_nid) {
 			/* special case: when from_nid is 0,
@@ -696,9 +696,9 @@ static bool is_stereo_amps(struct hda_codec *codec, hda_nid_t nid, int dir)
 		return true;
 	if (dir != HDA_INPUT || get_wcaps_type(wcaps) != AC_WID_AUD_MIX)
 		return false;
-	if (snd_hda_get_num_conns(codec, nid) != 1)
+	if (snd_hda_get_num_conns(codec, nid, 0) != 1)
 		return false;
-	if (snd_hda_get_connections(codec, nid, &conn, 1) < 0)
+	if (snd_hda_get_connections(codec, nid, 0, &conn, 1) < 0)
 		return false;
 	return !!(get_wcaps(codec, conn) & AC_WCAP_STEREO);
 }
@@ -791,7 +791,7 @@ static void activate_amp_in(struct hda_codec *codec, struct nid_path *path,
 	int type;
 	hda_nid_t nid = path->path[i];
 
-	nums = snd_hda_get_conn_list(codec, nid, &conn);
+	nums = snd_hda_get_conn_list(codec, nid, 0, &conn);
 	type = get_wcaps_type(get_wcaps(codec, nid));
 	if (type == AC_WID_PIN ||
 	    (type == AC_WID_AUD_IN && codec->single_adc_amp)) {
@@ -1060,7 +1060,7 @@ static int add_sw_ctl(struct hda_codec *codec, const char *pfx, int cidx,
 	val = amp_val_replace_channels(val, chs);
 	if (get_amp_direction_(val) == HDA_INPUT) {
 		hda_nid_t nid = get_amp_nid_(val);
-		int nums = snd_hda_get_num_conns(codec, nid);
+		int nums = snd_hda_get_num_conns(codec, nid, 0);
 		if (nums > 1) {
 			type = HDA_CTL_BIND_MUTE;
 			val |= nums << 19;
@@ -3002,7 +3002,7 @@ static bool look_for_mix_leaf_ctls(struct hda_codec *codec, hda_nid_t mix_nid,
 	const hda_nid_t *list;
 	hda_nid_t nid;
 
-	idx = snd_hda_get_conn_index(codec, mix_nid, pin, true);
+	idx = snd_hda_get_conn_index(codec, mix_nid, pin, 0, true);
 	if (idx < 0)
 		return false;
 
@@ -3015,7 +3015,7 @@ static bool look_for_mix_leaf_ctls(struct hda_codec *codec, hda_nid_t mix_nid,
 		return true;
 
 	/* check leaf node */
-	num_conns = snd_hda_get_conn_list(codec, mix_nid, &list);
+	num_conns = snd_hda_get_conn_list(codec, mix_nid, 0, &list);
 	if (num_conns < idx)
 		return false;
 	nid = list[idx];
@@ -4760,7 +4760,7 @@ static void mute_all_mixer_nid(struct hda_codec *codec, hda_nid_t mix)
 	const hda_nid_t *conn;
 	bool has_amp;
 
-	nums = snd_hda_get_conn_list(codec, mix, &conn);
+	nums = snd_hda_get_conn_list(codec, mix, 0, &conn);
 	has_amp = nid_has_mute(codec, mix, HDA_INPUT);
 	for (i = 0; i < nums; i++) {
 		if (has_amp)
sound/pci/hda/hda_proc.c +1 -1
index 033aa84..c747183 100644
@@ -636,7 +636,7 @@ static void print_conn_list(struct snd_info_buffer *buffer,
 	}
 
 	/* Get Cache connections info */
-	cache_len = snd_hda_get_conn_list(codec, nid, &list);
+	cache_len = snd_hda_get_conn_list(codec, nid, 0, &list);
 	if (cache_len >= 0 && (cache_len != conn_len ||
 			      memcmp(list, conn, conn_len) != 0)) {
 		snd_iprintf(buffer, "  In-driver Connection: %d\n", cache_len);
sound/pci/hda/patch_analog.c +10 -8
index e0fb8c6..6f007ff 100644
@@ -466,7 +466,7 @@ static int ad1983_auto_smux_enum_info(struct snd_kcontrol *kcontrol,
 	static const char * const texts2[] = { "PCM", "ADC" };
 	static const char * const texts3[] = { "PCM", "ADC1", "ADC2" };
 	hda_nid_t dig_out = spec->gen.multiout.dig_out_nid;
-	int num_conns = snd_hda_get_num_conns(codec, dig_out);
+	int num_conns = snd_hda_get_num_conns(codec, dig_out, 0);
 
 	if (num_conns == 2)
 		return snd_hda_enum_helper_info(kcontrol, uinfo, 2, texts2);
@@ -493,7 +493,7 @@ static int ad1983_auto_smux_enum_put(struct snd_kcontrol *kcontrol,
 	struct ad198x_spec *spec = codec->spec;
 	unsigned int val = ucontrol->value.enumerated.item[0];
 	hda_nid_t dig_out = spec->gen.multiout.dig_out_nid;
-	int num_conns = snd_hda_get_num_conns(codec, dig_out);
+	int num_conns = snd_hda_get_num_conns(codec, dig_out, 0);
 
 	if (val >= num_conns)
 		return -EINVAL;
@@ -521,7 +521,7 @@ static int ad1983_add_spdif_mux_ctl(struct hda_codec *codec)
 
 	if (!dig_out)
 		return 0;
-	num_conns = snd_hda_get_num_conns(codec, dig_out);
+	num_conns = snd_hda_get_num_conns(codec, dig_out, 0);
 	if (num_conns != 2 && num_conns != 3)
 		return 0;
 	if (!snd_hda_gen_add_kctl(&spec->gen, NULL, &ad1983_auto_smux_mixer))
@@ -546,8 +546,10 @@ static int patch_ad1983(struct hda_codec *codec)
 	set_beep_amp(spec, 0x10, 0, HDA_OUTPUT);
 
 	/* limit the loopback routes not to confuse the parser */
-	snd_hda_override_conn_list(codec, 0x0c, ARRAY_SIZE(conn_0c), conn_0c);
-	snd_hda_override_conn_list(codec, 0x0d, ARRAY_SIZE(conn_0d), conn_0d);
+	snd_hda_override_conn_list(codec, 0x0c, 0, ARRAY_SIZE(conn_0c),
+				   conn_0c);
+	snd_hda_override_conn_list(codec, 0x0d, 0, ARRAY_SIZE(conn_0d),
+				   conn_0d);
 
 	err = ad198x_parse_auto_config(codec, false);
 	if (err < 0)
@@ -745,7 +747,7 @@ static int ad1988_auto_smux_enum_info(struct snd_kcontrol *kcontrol,
 	static const char * const texts[] = {
 		"PCM", "ADC1", "ADC2", "ADC3",
 	};
-	int num_conns = snd_hda_get_num_conns(codec, 0x0b) + 1;
+	int num_conns = snd_hda_get_num_conns(codec, 0x0b, 0) + 1;
 	if (num_conns > 4)
 		num_conns = 4;
 	return snd_hda_enum_helper_info(kcontrol, uinfo, num_conns, texts);
@@ -768,7 +770,7 @@ static int ad1988_auto_smux_enum_put(struct snd_kcontrol *kcontrol,
 	struct ad198x_spec *spec = codec->spec;
 	unsigned int val = ucontrol->value.enumerated.item[0];
 	struct nid_path *path;
-	int num_conns = snd_hda_get_num_conns(codec, 0x0b) + 1;
+	int num_conns = snd_hda_get_num_conns(codec, 0x0b, 0) + 1;
 
 	if (val >= num_conns)
 		return -EINVAL;
@@ -856,7 +858,7 @@ static int ad1988_add_spdif_mux_ctl(struct hda_codec *codec)
 	    get_wcaps_type(get_wcaps(codec, 0x1d)) != AC_WID_AUD_MIX)
 		return 0;
 
-	num_conns = snd_hda_get_num_conns(codec, 0x0b) + 1;
+	num_conns = snd_hda_get_num_conns(codec, 0x0b, 0) + 1;
 	if (num_conns != 3 && num_conns != 4)
 		return 0;
 
sound/pci/hda/patch_hdmi.c +5 -3
index cf9bc042..8a20cf4 100644
@@ -1198,7 +1198,7 @@ static int hdmi_read_pin_conn(struct hda_codec *codec, int pin_idx)
 		return -EINVAL;
 	}
 
-	per_pin->num_mux_nids = snd_hda_get_connections(codec, pin_nid,
+	per_pin->num_mux_nids = snd_hda_get_connections(codec, pin_nid, 0,
 							per_pin->mux_nids,
 							HDA_MAX_CONNECTIONS);
 
@@ -2224,14 +2224,16 @@ static void intel_haswell_fixup_connect_list(struct hda_codec *codec,
 	hda_nid_t conns[4];
 	int nconns;
 
-	nconns = snd_hda_get_connections(codec, nid, conns, ARRAY_SIZE(conns));
+	nconns = snd_hda_get_connections(codec, nid, 0, conns,
+					 ARRAY_SIZE(conns));
 	if (nconns == spec->num_cvts &&
 	    !memcmp(conns, spec->cvt_nids, spec->num_cvts * sizeof(hda_nid_t)))
 		return;
 
 	/* override pins connection list */
 	codec_dbg(codec, "hdmi: haswell: override pin connection 0x%x\n", nid);
-	snd_hda_override_conn_list(codec, nid, spec->num_cvts, spec->cvt_nids);
+	snd_hda_override_conn_list(codec, nid, 0, spec->num_cvts,
+				   spec->cvt_nids);
 }
 
 #define INTEL_VENDOR_NID 0x08
sound/pci/hda/patch_realtek.c +11 -11
index 1fe8750..51240d6 100644
@@ -1861,17 +1861,17 @@ static void alc889_fixup_dac_route(struct hda_codec *codec,
 		/* fake the connections during parsing the tree */
 		hda_nid_t conn1[2] = { 0x0c, 0x0d };
 		hda_nid_t conn2[2] = { 0x0e, 0x0f };
-		snd_hda_override_conn_list(codec, 0x14, 2, conn1);
-		snd_hda_override_conn_list(codec, 0x15, 2, conn1);
-		snd_hda_override_conn_list(codec, 0x18, 2, conn2);
-		snd_hda_override_conn_list(codec, 0x1a, 2, conn2);
+		snd_hda_override_conn_list(codec, 0x14, 0, 2, conn1);
+		snd_hda_override_conn_list(codec, 0x15, 0, 2, conn1);
+		snd_hda_override_conn_list(codec, 0x18, 0, 2, conn2);
+		snd_hda_override_conn_list(codec, 0x1a, 0, 2, conn2);
 	} else if (action == HDA_FIXUP_ACT_PROBE) {
 		/* restore the connections */
 		hda_nid_t conn[5] = { 0x0c, 0x0d, 0x0e, 0x0f, 0x26 };
-		snd_hda_override_conn_list(codec, 0x14, 5, conn);
-		snd_hda_override_conn_list(codec, 0x15, 5, conn);
-		snd_hda_override_conn_list(codec, 0x18, 5, conn);
-		snd_hda_override_conn_list(codec, 0x1a, 5, conn);
+		snd_hda_override_conn_list(codec, 0x14, 0, 5, conn);
+		snd_hda_override_conn_list(codec, 0x15, 0, 5, conn);
+		snd_hda_override_conn_list(codec, 0x18, 0, 5, conn);
+		snd_hda_override_conn_list(codec, 0x1a, 0, 5, conn);
 	}
 }
 
@@ -4684,8 +4684,8 @@ static void alc290_fixup_mono_speakers(struct hda_codec *codec,
 		   make sure 0x14 (front speaker) and 0x15 (headphones) use the
 		   stereo DAC, while leaving 0x17 (bass speaker) for node 0x03. */
 		hda_nid_t conn1[2] = { 0x0c };
-		snd_hda_override_conn_list(codec, 0x14, 1, conn1);
-		snd_hda_override_conn_list(codec, 0x15, 1, conn1);
+		snd_hda_override_conn_list(codec, 0x14, 0, 1, conn1);
+		snd_hda_override_conn_list(codec, 0x15, 0, 1, conn1);
 	}
 }
 
@@ -4701,7 +4701,7 @@ static void alc298_fixup_speaker_volume(struct hda_codec *codec,
 		   speaker's volume now. */
 
 		hda_nid_t conn1[1] = { 0x0c };
-		snd_hda_override_conn_list(codec, 0x17, 1, conn1);
+		snd_hda_override_conn_list(codec, 0x17, 0, 1, conn1);
 	}
 }
 
sound/pci/hda/patch_sigmatel.c +2 -2
index 37b70f8..5bff793 100644
@@ -984,7 +984,7 @@ static int stac_create_spdif_mux_ctls(struct hda_codec *codec)
 	if (cfg->dig_outs < 1)
 		return 0;
 
-	num_cons = snd_hda_get_num_conns(codec, cfg->dig_out_pins[0]);
+	num_cons = snd_hda_get_num_conns(codec, cfg->dig_out_pins[0], 0);
 	if (num_cons <= 1)
 		return 0;
 
@@ -4543,7 +4543,7 @@ static int patch_stac92hd73xx(struct hda_codec *codec)
 	spec->gen.mixer_nid = 0x1d;
 	spec->have_spdif_mux = 1;
 
-	num_dacs = snd_hda_get_num_conns(codec, 0x0a) - 1;
+	num_dacs = snd_hda_get_num_conns(codec, 0x0a, 0) - 1;
 	if (num_dacs < 3 || num_dacs > 5) {
 		codec_warn(codec,
 			   "Could not determine number of channels defaulting to DAC count\n");
sound/pci/hda/patch_via.c +6 -4
index fc30d1e..14d9010 100644
@@ -867,7 +867,7 @@ static int add_secret_dac_path(struct hda_codec *codec)
 
 	if (!spec->gen.mixer_nid)
 		return 0;
-	nums = snd_hda_get_connections(codec, spec->gen.mixer_nid, conn,
+	nums = snd_hda_get_connections(codec, spec->gen.mixer_nid, 0, conn,
 				       ARRAY_SIZE(conn) - 1);
 	for (i = 0; i < nums; i++) {
 		if (get_wcaps_type(get_wcaps(codec, conn[i])) == AC_WID_AUD_OUT)
@@ -882,7 +882,7 @@ static int add_secret_dac_path(struct hda_codec *codec)
 			conn[nums++] = nid;
 			return snd_hda_override_conn_list(codec,
 							  spec->gen.mixer_nid,
-							  nums, conn);
+							  0, nums, conn);
 		}
 	}
 	return 0;
@@ -1082,8 +1082,10 @@ static void fix_vt1802_connections(struct hda_codec *codec)
 	static hda_nid_t conn_24[] = { 0x14, 0x1c };
 	static hda_nid_t conn_33[] = { 0x1c };
 
-	snd_hda_override_conn_list(codec, 0x24, ARRAY_SIZE(conn_24), conn_24);
-	snd_hda_override_conn_list(codec, 0x33, ARRAY_SIZE(conn_33), conn_33);
+	snd_hda_override_conn_list(codec, 0x24, 0, ARRAY_SIZE(conn_24),
+				   conn_24);
+	snd_hda_override_conn_list(codec, 0x33, 0, ARRAY_SIZE(conn_33),
+				   conn_33);
 }
 
 /* patch for vt2002P */
Signature · 1 line
1.9.1
#2 lore raw

RFC PATCH 2/3 ALSA: hda - add DP mst verb support

In reply to [email protected]

From: Libin Yang <[email protected]>

Add snd_hda_get_dev_select() and snd_hda_set_dev_select() functions for DP MST audio support.

---
 sound/pci/hda/hda_codec.c | 73 ++++++++++++++++++++++++++++++++++++++++++++---
 sound/pci/hda/hda_codec.h |  3 ++
 2 files changed, 72 insertions(+), 4 deletions(-)
sound/pci/hda/hda_codec.c +69 -4
index 32e21e8..57c25b7 100644
@@ -325,9 +325,16 @@ int snd_hda_get_conn_index(struct hda_codec *codec, hda_nid_t mux,
 }
 EXPORT_SYMBOL_GPL(snd_hda_get_conn_index);
 
-
-/* return DEVLIST_LEN parameter of the given widget */
-static unsigned int get_num_devices(struct hda_codec *codec, hda_nid_t nid)
+/**
+ * snd_hda_get_num_devices - get DEVLIST_LEN parameter of the given widget
+ *  @codec: the HDA codec
+ *  @nid: NID of the pin to parse
+ *
+ * Get the device entry number on the given widget.
+ * This is a feature of DP MST audio. Each pin can
+ * have several device entries in it.
+ */
+unsigned int snd_hda_get_num_devices(struct hda_codec *codec, hda_nid_t nid)
 {
 	unsigned int wcaps = get_wcaps(codec, nid);
 	unsigned int parm;
@@ -341,6 +348,7 @@ static unsigned int get_num_devices(struct hda_codec *codec, hda_nid_t nid)
 		parm = 0;
 	return parm & AC_DEV_LIST_LEN_MASK;
 }
+EXPORT_SYMBOL_GPL(snd_hda_get_num_devices);
 
 /**
  * snd_hda_get_devices - copy device list without cache
@@ -358,7 +366,7 @@ int snd_hda_get_devices(struct hda_codec *codec, hda_nid_t nid,
 	unsigned int parm;
 	int i, dev_len, devices;
 
-	parm = get_num_devices(codec, nid);
+	parm = snd_hda_get_num_devices(codec, nid);
 	if (!parm)	/* not multi-stream capable */
 		return 0;
 
@@ -382,6 +390,63 @@ int snd_hda_get_devices(struct hda_codec *codec, hda_nid_t nid,
 	return devices;
 }
 
+/**
+ * snd_hda_get_dev_select - get device entry select on the pin
+ * @codec: the HDA codec
+ * @nid: NID of the pin to get device entry select
+ *
+ * Get the devcie entry select on the pin. Return the device entry
+ * id selected on the pin. Return 0 means the first device entry
+ * is selected or MST is not supported.
+ */
+int snd_hda_get_dev_select(struct hda_codec *codec, hda_nid_t nid)
+{
+	/* not support dp_mst will always return 0, using first dev_entry */
+	if (!codec->dp_mst)
+		return 0;
+
+	return snd_hda_codec_read(codec, nid, 0, AC_VERB_GET_DEVICE_SEL, 0);
+}
+EXPORT_SYMBOL_GPL(snd_hda_get_dev_select);
+
+/**
+ * snd_hda_set_dev_select - set device entry select on the pin
+ * @codec: the HDA codec
+ * @nid: NID of the pin to set device entry select
+ * @dev_id: device entry id to be set
+ *
+ * Set the device entry select on the pin nid.
+ */
+int snd_hda_set_dev_select(struct hda_codec *codec, hda_nid_t nid, int dev_id)
+{
+	int ret, num_devices;
+
+	/* not support dp_mst will always return 0, using first dev_entry */
+	if (!codec->dp_mst)
+		return 0;
+
+	/* AC_PAR_DEVLIST_LEN is 0 based. */
+	num_devices = snd_hda_get_num_devices(codec, nid) + 1;
+	/* If Device List Length is 0 (num_device = 1),
+	 * the pin is not multi stream capable.
+	 * Do nothing in this case.
+	 */
+	if (num_devices == 1)
+		return 0;
+
+	/* Behavior of setting index being equal to or greater than
+	 * Device List Length is not predictable
+	 */
+	if (num_devices <= dev_id)
+		return -EINVAL;
+
+	ret = snd_hda_codec_write(codec, nid, 0,
+			AC_VERB_SET_DEVICE_SEL, dev_id);
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(snd_hda_set_dev_select);
+
 /*
  * read widget caps for each widget and store in cache
  */
sound/pci/hda/hda_codec.h +3 -0
index 4852c1a..57d1659 100644
@@ -348,8 +348,11 @@ int snd_hda_override_conn_list(struct hda_codec *codec, hda_nid_t nid,
 			       int dev_id, int nums, const hda_nid_t *list);
 int snd_hda_get_conn_index(struct hda_codec *codec, hda_nid_t mux,
 			   hda_nid_t nid, int dev_id, int recursive);
+unsigned int snd_hda_get_num_devices(struct hda_codec *codec, hda_nid_t nid);
 int snd_hda_get_devices(struct hda_codec *codec, hda_nid_t nid,
 			u8 *dev_list, int max_devices);
+int snd_hda_get_dev_select(struct hda_codec *codec, hda_nid_t nid);
+int snd_hda_set_dev_select(struct hda_codec *codec, hda_nid_t nid, int dev_id);
 
 struct hda_verb {
 	hda_nid_t nid;
Signature · 1 line
1.9.1
#3 lore raw

RFC PATCH 3/3 ALSA: hda - add DP mst audio support

In reply to [email protected]

From: Libin Yang <[email protected]>

This patch adds the DP mst audio support.

---
 sound/pci/hda/hda_codec.c  |   4 +
 sound/pci/hda/patch_hdmi.c | 226 ++++++++++++++++++++++++++++++++++-----------
 2 files changed, 174 insertions(+), 56 deletions(-)
sound/pci/hda/hda_codec.c +4 -0
index 57c25b7..d624d77 100644
@@ -482,6 +482,10 @@ static int read_pin_defaults(struct hda_codec *codec)
 		pin->nid = nid;
 		pin->cfg = snd_hda_codec_read(codec, nid, 0,
 					      AC_VERB_GET_CONFIG_DEFAULT, 0);
+		/*
+		 * all device entries are the same widget control so far
+		 * fixme: if any codec is different, need fix here
+		 */
 		pin->ctrl = snd_hda_codec_read(codec, nid, 0,
 					       AC_VERB_GET_PIN_WIDGET_CONTROL,
 					       0);
sound/pci/hda/patch_hdmi.c +170 -56

470 lines not on this page · raw message

Signature · 1 line
1.9.1
#4 lore raw
In reply to [email protected]
Quoted 5 lines · On Mon, 26 Sep 2016 10:35:35 +0200, [email protected] wrote:

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

Quoted 7 lines

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

I read the patchset again, and I'm still not convinced enough by this change.

The connection list is coded in the current way under the assumption that connections are more or less static. The connections are cached in snd_array, which is only for growing up, and not suitable for the entries that might be frequently removed. The removal is done only at overriding, and it happens only once at boot.

OTOH, with DP-MST case, the connection list is basically dynamic. It may increase or decrease depending on the connected monitors... Is my understanding correct? Or is the DP-MST connection list is static, too? Then I do wonder how it covers the whole connections with arbitrary device indices.

So, the connection cache management is one thing. Another thing is that the patchset doesn't consider about the pin ELD notification. Basically we switched to ELD notification instead of the pin unsol event for Intel chips. How does it fit?

thanks,

Takashi

#5 lore raw

Hi Takashi,

Quoted 18 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Monday, September 26, 2016 11:11 PM
To: [email protected]
Cc: [email protected]; Lin, Mengdong <[email protected]>; Yang, Libin <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 8 lines · On Mon, 26 Sep 2016 10:35:35 +0200, [email protected] wrote:

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Quoted 6 lines
Quoted 4 lines

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

I read the patchset again, and I'm still not convinced enough by this change.

Yes and I'm sorry that this patchset is delayed for a long time. Our gfx driver has been redesigned and some APIs have been changed. The change is done in Pandiyan, Dhinakaran's patches:
Patchset: Prep. for DP audio MST support
drm/i915: Standardize port type for DVO encoders drm/i915: Store port enum in intel_encoder
drm/i915: Switch to using port stored in intel_encoder drm/i915: Move audio_connector to intel_encoder drm/i915: start adding dp mst audio
Patch: drm/i915/dp: DP audio API changes for MST

Quoted 11 lines

The connection list is coded in the current way under the assumption that connections are more or less static. The connections are cached in snd_array, which is only for growing up, and not suitable for the entries that might be frequently removed. The removal is done only at overriding, and it happens only once at boot.

OTOH, with DP-MST case, the connection list is basically dynamic. It may increase or decrease depending on the connected monitors... Is my understanding correct? Or is the DP-MST connection list is static, too? Then I do wonder how it covers the whole connections with arbitrary device indices.

For the connection list, the driver will setup all the connection list at the beginning for each device entry, even it is non-MST. We statically allocate the connection list and will never remove them.

Quoted 5 lines

So, the connection cache management is one thing. Another thing is that the patchset doesn't consider about the pin ELD notification. Basically we switched to ELD notification instead of the pin unsol event for Intel chips. How does it fit?

For the ELD notification, in the patch 0003, if there is monitor hotplug, gfx driver will notify audio driver with acomp. Gfx driver will tell which port (pin), pipe (device entry) occurs the hotplug event. And then audio driver will call snd_hdac_acomp_get_eld() to get the eld information.

Regards,
Libin

Quoted 5 lines

thanks,

Takashi

#6 lore raw
Quoted 46 lines · Libin

Hi Takashi,

Quoted 18 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Monday, September 26, 2016 11:11 PM
To: [email protected]
Cc: [email protected]; Lin, Mengdong <[email protected]>; Yang, Libin <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 8 lines · On Mon, 26 Sep 2016 10:35:35 +0200, [email protected] wrote:

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Quoted 6 lines
Quoted 4 lines

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

I read the patchset again, and I'm still not convinced enough by this change.

Yes and I'm sorry that this patchset is delayed for a long time. Our gfx driver has been redesigned and some APIs have been changed. The change is done in Pandiyan, Dhinakaran's patches:
Patchset: Prep. for DP audio MST support
drm/i915: Standardize port type for DVO encoders drm/i915: Store port enum in intel_encoder
drm/i915: Switch to using port stored in intel_encoder drm/i915: Move audio_connector to intel_encoder drm/i915: start adding dp mst audio
Patch: drm/i915/dp: DP audio API changes for MST

Quoted 7 lines

The connection list is coded in the current way under the assumption that connections are more or less static. The connections are cached in snd_array, which is only for growing up, and not suitable for the entries that might be frequently removed. The removal is done only at overriding, and it happens only once at boot.

Quoted 8 lines
Quoted 4 lines

OTOH, with DP-MST case, the connection list is basically dynamic. It may increase or decrease depending on the connected monitors... Is my understanding correct? Or is the DP-MST connection list is static, too? Then I do wonder how it covers the whole connections with arbitrary device indices.

For the connection list, the driver will setup all the connection list at the beginning for each device entry, even it is non-MST. We statically allocate the connection list and will never remove them.

Hrm, so how will it actually be? Are the device indices fixed through the whole operations, no matter which monitor is connected? How many will they be?

It'd be helpful if you give an example of the actual typical setup. Then we can judge whether the proposed implementation is suitable.

Quoted 9 lines
Quoted 4 lines

So, the connection cache management is one thing. Another thing is that the patchset doesn't consider about the pin ELD notification. Basically we switched to ELD notification instead of the pin unsol event for Intel chips. How does it fit?

For the ELD notification, in the patch 0003, if there is monitor hotplug, gfx driver will notify audio driver with acomp. Gfx driver will tell which port (pin), pipe (device entry) occurs the hotplug event. And then audio driver will call snd_hdac_acomp_get_eld() to get the eld information.

OK, then drop the patch 1 and try to implement without messing around the connection list. The patch 1 is what I really don't like from the beginning. It makes things complicated.

Is there anything else than haswell fixup that touches the connection list with the dev id? If it's the only place, there can be an alternative way to hack it.

thanks,

Takashi

#7 lore raw

Hi Takashi,

Quoted 67 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 1:51 AM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 27 lines · Libin

Hi Takashi,

Quoted 24 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Monday, September 26, 2016 11:11 PM
To: [email protected]
Cc: [email protected]; Lin, Mengdong <[email protected]>; Yang, Libin <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 5 lines · On Mon, 26 Sep 2016 10:35:35 +0200, [email protected] wrote:

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

Quoted 7 lines

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

I read the patchset again, and I'm still not convinced enough by this

change.

Quoted 23 lines

Yes and I'm sorry that this patchset is delayed for a long time. Our gfx driver has been redesigned and some APIs have been changed. The change is done in Pandiyan, Dhinakaran's patches:
Patchset: Prep. for DP audio MST support
drm/i915: Standardize port type for DVO encoders drm/i915: Store port enum in intel_encoder
drm/i915: Switch to using port stored in intel_encoder drm/i915: Move audio_connector to intel_encoder drm/i915: start adding dp mst audio
Patch: drm/i915/dp: DP audio API changes for MST

Quoted 11 lines

The connection list is coded in the current way under the assumption that connections are more or less static. The connections are cached in snd_array, which is only for growing up, and not suitable for the entries that might be frequently removed. The removal is done only at overriding, and it happens only once at boot.

OTOH, with DP-MST case, the connection list is basically dynamic. It may increase or decrease depending on the connected monitors... Is my understanding correct? Or is the DP-MST connection list is static, too? Then I do wonder how it covers the whole connections with

arbitrary device indices.

Quoted 4 lines

For the connection list, the driver will setup all the connection list at the beginning for each device entry, even it is non-MST. We statically allocate the connection list and will never remove them.

Quoted 6 lines

Hrm, so how will it actually be? Are the device indices fixed through the whole operations, no matter which monitor is connected? How many will they be?

It'd be helpful if you give an example of the actual typical setup. Then we can judge whether the proposed implementation is suitable.

Even it is in non-MST mode, the connection list is setup with the per_pin initialization. And as the device entries under the same pin have the same connection list, I use the device entry 0's connection list for other device entries. This avoids connection list is invalid for other device entries in non-MST. After initialization, the connection list will never change even the non-MST/MST mode is changed.

Quoted 11 lines
Quoted 11 lines
Quoted 6 lines

So, the connection cache management is one thing. Another thing is that the patchset doesn't consider about the pin ELD notification. Basically we switched to ELD notification instead of the pin unsol event for Intel chips. How does it fit?

For the ELD notification, in the patch 0003, if there is monitor hotplug, gfx driver will notify audio driver with acomp. Gfx driver will tell which port (pin), pipe (device entry) occurs the hotplug event. And then audio driver will call snd_hdac_acomp_get_eld() to get the

Quoted 5 lines

eld information.

OK, then drop the patch 1 and try to implement without messing around the connection list. The patch 1 is what I really don't like from the beginning. It makes things complicated.

As the connection list is the same for the device entries under same pin, what do you think if we use the original APIs, ignoring the device id. This will not change the connection list code or be a very small change.

If you agree on that, I will have a try on it.

On the same time, I will check with our silicon if my understanding "the connection list is the same for the device entries under same pin" is right.

Quoted 3 lines

Is there anything else than haswell fixup that touches the connection list with the dev id? If it's the only place, there can be an alternative way to hack it.

If we always use the device entry 0 as all the device entries under the same pin, I think we can ignore the dev id now.

Regards,
Libin

Quoted 5 lines

thanks,

Takashi

#8 lore raw
Quoted 77 lines · Libin

Hi Takashi,

Quoted 67 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 1:51 AM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 27 lines · Libin

Hi Takashi,

Quoted 24 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Monday, September 26, 2016 11:11 PM
To: [email protected]
Cc: [email protected]; Lin, Mengdong <[email protected]>; Yang, Libin <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 5 lines · On Mon, 26 Sep 2016 10:35:35 +0200, [email protected] wrote:

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

Quoted 7 lines

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

I read the patchset again, and I'm still not convinced enough by this

change.

Quoted 23 lines

Yes and I'm sorry that this patchset is delayed for a long time. Our gfx driver has been redesigned and some APIs have been changed. The change is done in Pandiyan, Dhinakaran's patches:
Patchset: Prep. for DP audio MST support
drm/i915: Standardize port type for DVO encoders drm/i915: Store port enum in intel_encoder
drm/i915: Switch to using port stored in intel_encoder drm/i915: Move audio_connector to intel_encoder drm/i915: start adding dp mst audio
Patch: drm/i915/dp: DP audio API changes for MST

Quoted 11 lines

The connection list is coded in the current way under the assumption that connections are more or less static. The connections are cached in snd_array, which is only for growing up, and not suitable for the entries that might be frequently removed. The removal is done only at overriding, and it happens only once at boot.

OTOH, with DP-MST case, the connection list is basically dynamic. It may increase or decrease depending on the connected monitors... Is my understanding correct? Or is the DP-MST connection list is static, too? Then I do wonder how it covers the whole connections with

arbitrary device indices.

Quoted 4 lines

For the connection list, the driver will setup all the connection list at the beginning for each device entry, even it is non-MST. We statically allocate the connection list and will never remove them.

Quoted 6 lines

Hrm, so how will it actually be? Are the device indices fixed through the whole operations, no matter which monitor is connected? How many will they be?

It'd be helpful if you give an example of the actual typical setup. Then we can judge whether the proposed implementation is suitable.

Quoted 6 lines

Even it is in non-MST mode, the connection list is setup with the per_pin initialization. And as the device entries under the same pin have the same connection list, I use the device entry 0's connection list for other device entries. This avoids connection list is invalid for other device entries in non-MST. After initialization, the connection list will never change even the non-MST/MST mode is changed.

The devid=0 case is fine. This is identical with the normal widget connections. But you wrote that these connections are static even for devid!=0. Is it correct?

If so, now my question again: for devid!=0, which connections are given statically? In other words, how many devid would be passed?

Quoted 25 lines
Quoted 9 lines
Quoted 9 lines
Quoted 4 lines

So, the connection cache management is one thing. Another thing is that the patchset doesn't consider about the pin ELD notification. Basically we switched to ELD notification instead of the pin unsol event for Intel chips. How does it fit?

For the ELD notification, in the patch 0003, if there is monitor hotplug, gfx driver will notify audio driver with acomp. Gfx driver will tell which port (pin), pipe (device entry) occurs the hotplug event. And then audio driver will call snd_hdac_acomp_get_eld() to get the

Quoted 5 lines

eld information.

OK, then drop the patch 1 and try to implement without messing around the connection list. The patch 1 is what I really don't like from the beginning. It makes things complicated.

As the connection list is the same for the device entries under same pin, what do you think if we use the original APIs, ignoring the device id. This will not change the connection list code or be a very small change.

If you agree on that, I will have a try on it.

On the same time, I will check with our silicon if my understanding "the connection list is the same for the device entries under same pin" is right.

Quoted 5 lines
Quoted 2 lines

Is there anything else than haswell fixup that touches the connection list with the dev id? If it's the only place, there can be an alternative way to hack it.

If we always use the device entry 0 as all the device entries under the same pin, I think we can ignore the dev id now.

Well, it is only about the pin/cvt connection inside HDMI/DP codec, and it should be easily covered in a different way even with devid!=0, I suppose. That said, the patchset looks way too complex than necessary.

Though, it might be a good idea to start with devid=0 assumption if it makes the changes simpler.

Takashi

#9 lore raw

Hi Takashi,

Quoted 97 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 3:59 PM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 60 lines · Libin

Hi Takashi,

Quoted 57 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 1:51 AM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 28 lines · Libin

Hi Takashi,

Quoted 25 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Monday, September 26, 2016 11:11 PM
To: [email protected]
Cc: [email protected]; Lin, Mengdong <[email protected]>; Yang, Libin <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 5 lines · On Mon, 26 Sep 2016 10:35:35 +0200, [email protected] wrote:

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

Quoted 7 lines

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

I read the patchset again, and I'm still not convinced enough by this

change.

Quoted 18 lines

Yes and I'm sorry that this patchset is delayed for a long time. Our gfx driver has been redesigned and some APIs have been changed. The change is done in Pandiyan, Dhinakaran's patches:
Patchset: Prep. for DP audio MST support
drm/i915: Standardize port type for DVO encoders drm/i915: Store port enum in intel_encoder
drm/i915: Switch to using port stored in intel_encoder drm/i915: Move audio_connector to intel_encoder drm/i915: start adding dp mst audio
Patch: drm/i915/dp: DP audio API changes for MST

Quoted 6 lines

The connection list is coded in the current way under the assumption that connections are more or less static. The connections are cached in snd_array, which is only for growing up, and not suitable for the entries that might be frequently removed. The removal is done only at overriding, and it happens only

once at boot.

Quoted 19 lines
Quoted 12 lines
Quoted 6 lines
Quoted 6 lines

OTOH, with DP-MST case, the connection list is basically dynamic. It may increase or decrease depending on the connected monitors... Is my understanding correct? Or is the DP-MST connection list is static, too? Then I do wonder how it covers the whole connections with

arbitrary device indices.

Quoted 4 lines

For the connection list, the driver will setup all the connection list at the beginning for each device entry, even it is non-MST. We statically allocate the connection list and will never remove them.

Quoted 6 lines

Hrm, so how will it actually be? Are the device indices fixed through the whole operations, no matter which monitor is connected? How many will they be?

It'd be helpful if you give an example of the actual typical setup. Then we can judge whether the proposed implementation is suitable.

Quoted 6 lines

Even it is in non-MST mode, the connection list is setup with the per_pin initialization. And as the device entries under the same pin have the same connection list, I use the device entry 0's connection list for other device entries. This avoids connection list is invalid for other device entries in non-MST. After initialization, the connection list will never change even the non-MST/MST mode is changed.

Quoted 6 lines

The devid=0 case is fine. This is identical with the normal widget connections. But you wrote that these connections are static even for devid!=0. Is it correct?

If so, now my question again: for devid!=0, which connections are given statically? In other words, how many devid would be passed?

For the connection list, it is static. It will be initialized at bootup.

For the pin and cvt actual connection, it is dynamically assigned. In hdmi_pcm_open(), it will call hdmi_choose_cvt() to pick up a cvt (based on the connection list and the cvt is used or not). Actually, I think the connection list is only useful when picking up the cvt (or other nodes). Please correct me if I'm wrong.

When we try to connect device entry 2, pin 1 to cvt A, for example, we will call snd_hda_set_dev_select(pin_nid, dev_id) to select the device entry on the pin. And then we will use the verb AC_VERB_SET_CONNECT_SEL to setup the connection as before.

Quoted 35 lines
Quoted 13 lines
Quoted 13 lines
Quoted 11 lines
Quoted 5 lines

So, the connection cache management is one thing. Another thing is that the patchset doesn't consider about the pin ELD notification. Basically we switched to ELD notification instead of the pin unsol event for Intel chips. How does it fit?

For the ELD notification, in the patch 0003, if there is monitor hotplug, gfx driver will notify audio driver with acomp. Gfx driver will tell which port (pin), pipe (device entry) occurs the hotplug event. And then audio driver will call snd_hdac_acomp_get_eld() to get the

eld information.

Quoted 7 lines
Quoted 3 lines

OK, then drop the patch 1 and try to implement without messing around the connection list. The patch 1 is what I really don't like from the beginning. It makes things complicated.

As the connection list is the same for the device entries under same pin, what do you think if we use the original APIs, ignoring the device id. This will not change the connection list code or be a very small

change.

Quoted 9 lines

If you agree on that, I will have a try on it.

On the same time, I will check with our silicon if my understanding "the connection list is the same for the device entries under same pin" is right.

Quoted 2 lines

Is there anything else than haswell fixup that touches the connection list with the dev id? If it's the only place, there can be an

alternative way to hack it.

Quoted 3 lines

If we always use the device entry 0 as all the device entries under the same pin, I think we can ignore the dev id now.

Quoted 6 lines

Well, it is only about the pin/cvt connection inside HDMI/DP codec, and it should be easily covered in a different way even with devid!=0, I suppose. That said, the patchset looks way too complex than necessary.

Though, it might be a good idea to start with devid=0 assumption if it makes the changes simpler.

Yes, the patches are too complex for this simple usage. I will try to see if it works without touching the connection list core.

Regards,
Libin

Quoted 3 lines

Takashi

#10 lore raw
Quoted 100 lines · Libin

Hi Takashi,

Quoted 97 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 3:59 PM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 60 lines · Libin

Hi Takashi,

Quoted 57 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 1:51 AM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 28 lines · Libin

Hi Takashi,

Quoted 25 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Monday, September 26, 2016 11:11 PM
To: [email protected]
Cc: [email protected]; Lin, Mengdong <[email protected]>; Yang, Libin <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 5 lines · On Mon, 26 Sep 2016 10:35:35 +0200, [email protected] wrote:

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

Quoted 7 lines

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

I read the patchset again, and I'm still not convinced enough by this

change.

Quoted 18 lines

Yes and I'm sorry that this patchset is delayed for a long time. Our gfx driver has been redesigned and some APIs have been changed. The change is done in Pandiyan, Dhinakaran's patches:
Patchset: Prep. for DP audio MST support
drm/i915: Standardize port type for DVO encoders drm/i915: Store port enum in intel_encoder
drm/i915: Switch to using port stored in intel_encoder drm/i915: Move audio_connector to intel_encoder drm/i915: start adding dp mst audio
Patch: drm/i915/dp: DP audio API changes for MST

Quoted 6 lines

The connection list is coded in the current way under the assumption that connections are more or less static. The connections are cached in snd_array, which is only for growing up, and not suitable for the entries that might be frequently removed. The removal is done only at overriding, and it happens only

once at boot.

Quoted 25 lines
Quoted 12 lines
Quoted 6 lines
Quoted 6 lines

OTOH, with DP-MST case, the connection list is basically dynamic. It may increase or decrease depending on the connected monitors... Is my understanding correct? Or is the DP-MST connection list is static, too? Then I do wonder how it covers the whole connections with

arbitrary device indices.

Quoted 4 lines

For the connection list, the driver will setup all the connection list at the beginning for each device entry, even it is non-MST. We statically allocate the connection list and will never remove them.

Quoted 6 lines

Hrm, so how will it actually be? Are the device indices fixed through the whole operations, no matter which monitor is connected? How many will they be?

It'd be helpful if you give an example of the actual typical setup. Then we can judge whether the proposed implementation is suitable.

Even it is in non-MST mode, the connection list is setup with the per_pin initialization. And as the device entries under the same pin have the same connection list, I use the device entry 0's connection list for other device entries. This avoids connection list is invalid for other device entries in non-MST. After initialization, the connection list will never change even the non-MST/MST mode is changed.

Quoted 8 lines
Quoted 6 lines

The devid=0 case is fine. This is identical with the normal widget connections. But you wrote that these connections are static even for devid!=0. Is it correct?

If so, now my question again: for devid!=0, which connections are given statically? In other words, how many devid would be passed?

For the connection list, it is static. It will be initialized at bootup.

So this is only about devid=0, right? For devid!=0, there wouldn't be anything else than pin/cvt connections, correct?

Quoted 1 line

For the pin and cvt actual connection, it is dynamically assigned.

... but you still want to cache this to an (ever-growing) array for the standard connection lists. That's what I was concerned.

Quoted 4 lines

In hdmi_pcm_open(), it will call hdmi_choose_cvt() to pick up a cvt (based on the connection list and the cvt is used or not). Actually, I think the connection list is only useful when picking up the cvt (or other nodes). Please correct me if I'm wrong.

Yes, since HDMI/DP codec has only pretty simple connections.

Quoted 4 lines

When we try to connect device entry 2, pin 1 to cvt A, for example, we will call snd_hda_set_dev_select(pin_nid, dev_id) to select the device entry on the pin. And then we will use the verb AC_VERB_SET_CONNECT_SEL to setup the connection as before.

OK, that sounds straightforward. So we just need to set up these verbs at opening the device (and likely at suspend/resume, too).

Quoted 38 lines
Quoted 38 lines
Quoted 12 lines
Quoted 12 lines
Quoted 10 lines
Quoted 4 lines

So, the connection cache management is one thing. Another thing is that the patchset doesn't consider about the pin ELD notification. Basically we switched to ELD notification instead of the pin unsol event for Intel chips. How does it fit?

For the ELD notification, in the patch 0003, if there is monitor hotplug, gfx driver will notify audio driver with acomp. Gfx driver will tell which port (pin), pipe (device entry) occurs the hotplug event. And then audio driver will call snd_hdac_acomp_get_eld() to get the

eld information.

Quoted 7 lines
Quoted 3 lines

OK, then drop the patch 1 and try to implement without messing around the connection list. The patch 1 is what I really don't like from the beginning. It makes things complicated.

As the connection list is the same for the device entries under same pin, what do you think if we use the original APIs, ignoring the device id. This will not change the connection list code or be a very small

change.

Quoted 9 lines

If you agree on that, I will have a try on it.

On the same time, I will check with our silicon if my understanding "the connection list is the same for the device entries under same pin" is right.

Quoted 2 lines

Is there anything else than haswell fixup that touches the connection list with the dev id? If it's the only place, there can be an

alternative way to hack it.

Quoted 3 lines

If we always use the device entry 0 as all the device entries under the same pin, I think we can ignore the dev id now.

Well, it is only about the pin/cvt connection inside HDMI/DP codec, and it should be easily covered in a different way even with devid!=0, I suppose. That said, the patchset looks way too complex than necessary.

Quoted 5 lines
Quoted 2 lines

Though, it might be a good idea to start with devid=0 assumption if it makes the changes simpler.

Yes, the patches are too complex for this simple usage. I will try to see if it works without touching the connection list core.

Yes, please. Let's keep KISS principle.

thanks,

Takashi

#11 lore raw

Hi Takashi,

Quoted 121 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 4:46 PM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 59 lines · Libin

Hi Takashi,

Quoted 56 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 3:59 PM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 46 lines · Libin

Hi Takashi,

Quoted 43 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Tuesday, September 27, 2016 1:51 AM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 30 lines · Libin

Hi Takashi,

Quoted 27 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Monday, September 26, 2016 11:11 PM
To: [email protected]
Cc: [email protected]; Lin, Mengdong <[email protected]>; Yang, Libin <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 5 lines · On Mon, 26 Sep 2016 10:35:35 +0200, [email protected] wrote:

From: Libin Yang <[email protected]>

This patchset starts to support DP MST audio.

Quoted 7 lines

This patchset is based on drm-intel-nightly on
git://anongit.freedesktop.org/drm-intel

Libin Yang (3):
ALSA: hda - codec add DP MST support for connection list
ALSA: hda - add DP mst verb support
ALSA: hda - add DP mst audio support

I read the patchset again, and I'm still not convinced enough by this

change.

Quoted 3 lines

Yes and I'm sorry that this patchset is delayed for a long time. Our gfx driver has been redesigned and some APIs have been

changed.

Quoted 16 lines
Quoted 16 lines
Quoted 16 lines
Quoted 16 lines
Quoted 16 lines

The change is done in Pandiyan, Dhinakaran's patches:
Patchset: Prep. for DP audio MST support
drm/i915: Standardize port type for DVO encoders drm/i915: Store port enum in intel_encoder drm/i915: Switch to using port stored in intel_encoder drm/i915: Move audio_connector to intel_encoder drm/i915: start adding dp mst audio
Patch: drm/i915/dp: DP audio API changes for MST

Quoted 7 lines

The connection list is coded in the current way under the assumption that connections are more or less static. The connections are cached in snd_array, which is only for growing up, and not suitable for the entries that might be frequently removed. The removal is done only at overriding, and it happens only

Quoted 4 lines
Quoted 4 lines

once at boot.

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

OTOH, with DP-MST case, the connection list is basically dynamic. It may increase or decrease depending on the connected

monitors...

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines

Is my understanding correct? Or is the DP-MST connection list is static, too? Then I do wonder how it covers the whole connections with

arbitrary device indices.

Quoted 3 lines

For the connection list, the driver will setup all the connection list at the beginning for each device entry, even it is non-

MST.

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines
Quoted 1 line

We statically allocate the connection list and will never remove them.

Hrm, so how will it actually be? Are the device indices fixed through the whole operations, no matter which monitor is connected? How many will they be?

It'd be helpful if you give an example of the actual typical setup. Then we can judge whether the proposed implementation is suitable.

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

Even it is in non-MST mode, the connection list is setup with the per_pin initialization. And as the device entries under the same pin have the same connection list, I use the device entry 0's connection list for other device entries. This avoids connection list is invalid for other device entries in non-MST. After initialization, the connection list will never change even the non-

MST/MST mode is changed.

Quoted 2 lines
Quoted 2 lines

The devid=0 case is fine. This is identical with the normal widget

connections.

Quoted 3 lines
Quoted 3 lines

But you wrote that these connections are static even for devid!=0. Is it correct?

Quoted 7 lines
Quoted 4 lines
Quoted 2 lines

If so, now my question again: for devid!=0, which connections are given statically? In other words, how many devid would be passed?

For the connection list, it is static. It will be initialized at bootup.

So this is only about devid=0, right? For devid!=0, there wouldn't be anything else than pin/cvt connections, correct?

Devid != 0 is the same as devid = 0. And there is only pin/cvt connections.

My basic idea is:
1. Every device entry use a virtual pin
2. Each device entry will be initialized at bootup even it is not in MST mode. This means after bootup, mode change will not add/remove the per_pin.
3. All device entries under the same pin will use the same connection list. And the connection list is initialized at bootup. It will be saved in per_pin->mux_nids. This will never be changed after bootup. (However the pin-cvt connection is dynamically assigned when necessary.)

Quoted 5 lines
Quoted 2 lines

For the pin and cvt actual connection, it is dynamically assigned.

... but you still want to cache this to an (ever-growing) array for the standard connection lists. That's what I was concerned.

Do you mean codec->conn_list? If we don't use the dev_id and all the device entries under the same pin use the same one, it will be exactly the same as before.

Or do you mean the array of spec->pins? spec->pins is setup at bootup and will not be changed later.

Regards,
Libin

Quoted 73 lines
Quoted 5 lines

In hdmi_pcm_open(), it will call hdmi_choose_cvt() to pick up a cvt (based on the connection list and the cvt is used or not). Actually, I think the connection list is only useful when picking up the cvt (or other nodes). Please correct me if I'm wrong.

Yes, since HDMI/DP codec has only pretty simple connections.

Quoted 4 lines

When we try to connect device entry 2, pin 1 to cvt A, for example, we will call snd_hda_set_dev_select(pin_nid, dev_id) to select the device entry on the pin. And then we will use the verb AC_VERB_SET_CONNECT_SEL to setup the connection as before.

OK, that sounds straightforward. So we just need to set up these verbs at opening the device (and likely at suspend/resume, too).

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

So, the connection cache management is one thing. Another thing is that the patchset doesn't consider about the pin ELD

notification.

Quoted 19 lines
Quoted 19 lines
Quoted 19 lines
Quoted 9 lines
Quoted 9 lines
Quoted 2 lines

Basically we switched to ELD notification instead of the pin unsol event for Intel chips. How does it fit?

For the ELD notification, in the patch 0003, if there is monitor hotplug, gfx driver will notify audio driver with acomp. Gfx driver will tell which port (pin), pipe (device entry) occurs the hotplug event. And then audio driver will call
snd_hdac_acomp_get_eld() to get the

Quoted 5 lines

eld information.

OK, then drop the patch 1 and try to implement without messing around the connection list. The patch 1 is what I really don't like from the beginning. It makes things complicated.

As the connection list is the same for the device entries under same pin, what do you think if we use the original APIs, ignoring the device id. This will not change the connection list code or be a very small

Quoted 6 lines
Quoted 6 lines

change.

Quoted 5 lines

If you agree on that, I will have a try on it.

On the same time, I will check with our silicon if my understanding "the connection list is the same for the device entries

under same pin"

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines

is right.

Quoted 3 lines

Is there anything else than haswell fixup that touches the connection list with the dev id? If it's the only place, there can be an

Quoted 7 lines
Quoted 7 lines

alternative way to hack it.

Quoted 3 lines

If we always use the device entry 0 as all the device entries under the same pin, I think we can ignore the dev id now.

Well, it is only about the pin/cvt connection inside HDMI/DP codec, and it should be easily covered in a different way even with devid!=0, I

suppose.

Quoted 7 lines
Quoted 4 lines

That said, the patchset looks way too complex than necessary.

Though, it might be a good idea to start with devid=0 assumption if it makes the changes simpler.

Yes, the patches are too complex for this simple usage. I will try to see if it works without touching the connection list core.

Yes, please. Let's keep KISS principle.

thanks,

Takashi

#12 lore raw
Quoted 8 lines · Libin

Hi Takashi,

Quoted 5 lines
Quoted 4 lines
Quoted 2 lines

If so, now my question again: for devid!=0, which connections are given statically? In other words, how many devid would be passed?

For the connection list, it is static. It will be initialized at bootup.

Quoted 7 lines
Quoted 2 lines

So this is only about devid=0, right? For devid!=0, there wouldn't be anything else than pin/cvt connections, correct?

Devid != 0 is the same as devid = 0. And there is only pin/cvt connections.

My basic idea is:
1. Every device entry use a virtual pin

OK.

Quoted 1 line

2. Each device entry will be initialized at bootup even it is not in MST mode.

Here starts unclearness. How "*each* device entry" will be enumerated at boot up time? The device id can be arbitrary numbers, per spec.

And "even it is not in MST" means the devices that aren't capable MST (like SandyBridge HDMI/DP)? Or it's meant "even when no monitor is connected via MST"?

Quoted 1 line

This means after bootup, mode change will not add/remove the per_pin.

What here "mode" means exactly?

Quoted 6 lines

3. All device entries under the same pin will use the same connection list. And the connection list is initialized at bootup. It will be saved in per_pin->mux_nids. This will never be changed after bootup. (However the pin-cvt connection is dynamically assigned when necessary.)

Quoted 1 line
Quoted 8 lines
Quoted 4 lines
Quoted 1 line

For the pin and cvt actual connection, it is dynamically assigned.

... but you still want to cache this to an (ever-growing) array for the standard connection lists. That's what I was concerned.

Do you mean codec->conn_list? If we don't use the dev_id and all the device entries under the same pin use the same one, it will be exactly the same as before.

But you touch the whole codes doing the connection list caching even by adding the dev_id field. Why this is needed if you don't use the dev_id...?

Again, try not to mess up the core code as a start. The virtual pin concept is OK as long as it's self-contained in the HDMI codec driver. And adding some basic calls to handle the device id is fine. But the way to change as in patch 1 can't be included without the enough justification.

thanks,

Takashi

#13 lore raw

Hi Takashi,

Quoted 27 lines

-----Original Message-----
From: Takashi Iwai [mailto:[email protected]]
Sent: Wednesday, September 28, 2016 3:57 PM
To: Yang, Libin <[email protected]>
Cc: [email protected]; [email protected]; Lin, Mengdong <[email protected]>
Subject: Re: [alsa-devel] [RFC PATCH 0/3] support DP MST audio

Quoted 5 lines · Libin

Hi Takashi,

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

If so, now my question again: for devid!=0, which connections are given statically? In other words, how many devid would be

passed?

Quoted 10 lines
Quoted 5 lines
Quoted 2 lines

For the connection list, it is static. It will be initialized at bootup.

So this is only about devid=0, right? For devid!=0, there wouldn't be anything else than pin/cvt connections, correct?

Devid != 0 is the same as devid = 0. And there is only pin/cvt connections.

My basic idea is:
1. Every device entry use a virtual pin

Quoted 6 lines

OK.

Quoted 1 line

2. Each device entry will be initialized at bootup even it is not in MST mode.

Here starts unclearness. How "*each* device entry" will be enumerated at boot up time? The device id can be arbitrary numbers, per spec.

This is a good question. We really met such issue. We currently is to judge whether it is support MST or not (if it is haswell_plus, then it supports MST). We didn't use snd_hda_get_num_devices() to get the device number.

And if it is MST, then everything works well, each per_pin represents a device entry.

If it is NON-MST, the device id means nothing, and snd_hda_set_dev_select() does nothing. So we will use the first device entry (dev_id = 0) to represent the pin.

Quoted 4 lines

And "even it is not in MST" means the devices that aren't capable MST (like SandyBridge HDMI/DP)? Or it's meant "even when no monitor is connected via MST"?

I mean " even when no monitor is connected via MST". Actually, when initialization, driver will judge the platform supports MST or not (for intel, only haswell_plus support MST).

Quoted 4 lines
Quoted 2 lines

This means after bootup, mode change will not add/remove the per_pin.

What here "mode" means exactly?

Mode means MST mode or NON_MST mode.

Quoted 17 lines
Quoted 8 lines

3. All device entries under the same pin will use the same connection list. And the connection list is initialized at bootup. It will be saved in per_pin->mux_nids. This will never be changed after bootup. (However the pin-cvt connection is dynamically assigned when
necessary.)

Quoted 1 line
Quoted 8 lines
Quoted 4 lines
Quoted 1 line

For the pin and cvt actual connection, it is dynamically assigned.

... but you still want to cache this to an (ever-growing) array for the standard connection lists. That's what I was concerned.

Do you mean codec->conn_list? If we don't use the dev_id and all the device entries under the same pin use the same one, it will be exactly the same as before.

Quoted 8 lines

But you touch the whole codes doing the connection list caching even by adding the dev_id field. Why this is needed if you don't use the dev_id...?

Again, try not to mess up the core code as a start. The virtual pin concept is OK as long as it's self-contained in the HDMI codec driver. And adding some basic calls to handle the device id is fine.
But the way to change as in patch 1 can't be included without the enough justification.

Yes, I understand and I will refine the patch and not touch the connection list core code in next patchset. In the new patch, dev_id will not be used in connection list. I will totally remove the patch 1.

Regards,
Libin

Quoted 5 lines

thanks,

Takashi