ALSA: hda: Conexant: Allow different output types to share DAC

From 95ca8a0cd32ad781c54a349c5ce84a9119668f9c Mon Sep 17 00:00:00 2001

Quoted 2 lines

From 95ca8a0cd32ad781c54a349c5ce84a9119668f9c Mon Sep 17 00:00:00 2001

From: David Henningsson <[email protected]>
Date: Thu, 25 Aug 2011 13:16:02 +0200
Subject: [PATCH] ALSA: hda: Conexant: Allow different output types to share DAC

Headphones has stopped working for the original reported (a regression compared to 2.6.38). This is because Speaker and Headphones share the same DAC, in which case no Headphones volume control was created. This patch fixes so that both Speaker and Headphones volume controls are created in such scenario.

Buglink http://bugs.launchpad.net/bugs/817943

Signed-off-by David Henningsson <[email protected]>

---
 sound/pci/hda/patch_conexant.c |   40 ++++++++++++++++++++++------------------
 1 files changed, 22 insertions(+), 18 deletions(-)

sound/pci/hda/patch_conexant.c +3 -2

index 197ad93..a29b88d 100644
@@ -3327,22 +3327,23 @@ static int fill_cx_auto_dacs(struct hda_codec *codec, hda_nid_t *dacs)
 }
 
 /* fill pin_dac_pair list from the pin and dac list */
-static int fill_dacs_for_pins(struct hda_codec *codec, hda_nid_t *pins,
+static void fill_dacs_for_pins(struct hda_codec *codec, hda_nid_t *pins,
 			      int num_pins, hda_nid_t *dacs, int *rest,
-			      struct pin_dac_pair *filled, int type)
+			      struct pin_dac_pair *filled, int *saved_nums, 
+			      int type)
 {
 	int i, nums;

sound/pci/hda/patch_conexant.c +16 -13

 		nums++;
 	}
-	return nums;
+	*saved_nums = nums;
 }
 
 /* parse analog output paths */

@@ -3355,15 +3356,16 @@ static void cx_auto_parse_output(struct hda_codec *codec)
 
 	rest = fill_cx_auto_dacs(codec, dacs);
 	/* parse all analog output pins */
-	nums = fill_dacs_for_pins(codec, cfg->line_out_pins, cfg->line_outs,
-				  dacs, &rest, spec->dac_info,
-				  AUTO_PIN_LINE_OUT);
-	nums += fill_dacs_for_pins(codec, cfg->hp_pins, cfg->hp_outs,
-				  dacs, &rest, spec->dac_info + nums,
-				  AUTO_PIN_HP_OUT);
-	nums += fill_dacs_for_pins(codec, cfg->speaker_pins, cfg->speaker_outs,
-				  dacs, &rest, spec->dac_info + nums,
-				  AUTO_PIN_SPEAKER_OUT);
+	nums = 0;
+	fill_dacs_for_pins(codec, cfg->line_out_pins, cfg->line_outs,
+			  dacs, &rest, spec->dac_info, &nums,
+			  AUTO_PIN_LINE_OUT);
+	fill_dacs_for_pins(codec, cfg->hp_pins, cfg->hp_outs,
+			  dacs, &rest, spec->dac_info, &nums,
+			  AUTO_PIN_HP_OUT);
+	fill_dacs_for_pins(codec, cfg->speaker_pins, cfg->speaker_outs,
+			  dacs, &rest, spec->dac_info, &nums,
+			  AUTO_PIN_SPEAKER_OUT);
 	spec->dac_info_filled = nums;
 	/* fill multiout struct */
 	for (i = 0; i < nums; i++) {

@@ -4130,9 +4132,11 @@ static int try_add_pb_volume(struct hda_codec *codec, hda_nid_t dac,
 			     hda_nid_t pin, const char *name, int idx)
 {
 	unsigned int caps;
-	caps = query_amp_caps(codec, dac, HDA_OUTPUT);
-	if (caps & AC_AMPCAP_NUM_STEPS)
-		return cx_auto_add_pb_volume(codec, dac, name, idx);
+	if (!(dac & DAC_SLAVE_FLAG)) {
+		caps = query_amp_caps(codec, dac, HDA_OUTPUT);
+		if (caps & AC_AMPCAP_NUM_STEPS)
+			return cx_auto_add_pb_volume(codec, dac, name, idx);
+	}
 	caps = query_amp_caps(codec, pin, HDA_OUTPUT);
 	if (caps & AC_AMPCAP_NUM_STEPS)
 		return cx_auto_add_pb_volume(codec, pin, name, idx);

At Thu, 25 Aug 2011 13:26:26 +0200,

Quoted 34 lines · David Henningsson

Quoted 2 lines

From 95ca8a0cd32ad781c54a349c5ce84a9119668f9c Mon Sep 17 00:00:00 2001

From: David Henningsson <[email protected]>
Date: Thu, 25 Aug 2011 13:16:02 +0200
Subject: [PATCH] ALSA: hda: Conexant: Allow different output types to share DAC

Headphones has stopped working for the original reported (a regression compared to 2.6.38). This is because Speaker and Headphones share the same DAC, in which case no Headphones volume control was created. This patch fixes so that both Speaker and Headphones volume controls are created in such scenario.

Buglink http://bugs.launchpad.net/bugs/817943

Signed-off-by David Henningsson <[email protected]>

---
 sound/pci/hda/patch_conexant.c |   40 ++++++++++++++++++++++------------------
 1 files changed, 22 insertions(+), 18 deletions(-)

sound/pci/hda/patch_conexant.c +3 -2

index 197ad93..a29b88d 100644
@@ -3327,22 +3327,23 @@ static int fill_cx_auto_dacs(struct hda_codec *codec, hda_nid_t *dacs)
 }
 
 /* fill pin_dac_pair list from the pin and dac list */
-static int fill_dacs_for_pins(struct hda_codec *codec, hda_nid_t *pins,
+static void fill_dacs_for_pins(struct hda_codec *codec, hda_nid_t *pins,
 			      int num_pins, hda_nid_t *dacs, int *rest,
-			      struct pin_dac_pair *filled, int type)
+			      struct pin_dac_pair *filled, int *saved_nums, 
+			      int type)
 {
 	int i, nums;

sound/pci/hda/patch_conexant.c static int fill_cx_auto_dacs(struct hda_codec *codec, hda_nid_t *dacs) · line 3343

Quoted 10 lines

sound/pci/hda/patch_conexant.c +2 -2

 
-	nums = 0;
+	nums = *saved_nums;
 	for (i = 0; i < num_pins; i++) {
 		filled[nums].pin = pins[i];
 		filled[nums].type = type;
 		filled[nums].dac = get_unassigned_dac(codec, pins[i], dacs, rest);
-		if (!filled[nums].dac && i > 0 && filled[0].dac)
+		if (!filled[nums].dac && filled[0].dac)
 			filled[nums].dac = filled[0].dac | DAC_SLAVE_FLAG;

It should try the first DAC of the same type at first, not always to filled[0].

Quoted 49 lines

sound/pci/hda/patch_conexant.c +16 -13

 		nums++;
 	}
-	return nums;
+	*saved_nums = nums;
 }
 
 /* parse analog output paths */

@@ -3355,15 +3356,16 @@ static void cx_auto_parse_output(struct hda_codec *codec)
 
 	rest = fill_cx_auto_dacs(codec, dacs);
 	/* parse all analog output pins */
-	nums = fill_dacs_for_pins(codec, cfg->line_out_pins, cfg->line_outs,
-				  dacs, &rest, spec->dac_info,
-				  AUTO_PIN_LINE_OUT);
-	nums += fill_dacs_for_pins(codec, cfg->hp_pins, cfg->hp_outs,
-				  dacs, &rest, spec->dac_info + nums,
-				  AUTO_PIN_HP_OUT);
-	nums += fill_dacs_for_pins(codec, cfg->speaker_pins, cfg->speaker_outs,
-				  dacs, &rest, spec->dac_info + nums,
-				  AUTO_PIN_SPEAKER_OUT);
+	nums = 0;
+	fill_dacs_for_pins(codec, cfg->line_out_pins, cfg->line_outs,
+			  dacs, &rest, spec->dac_info, &nums,
+			  AUTO_PIN_LINE_OUT);
+	fill_dacs_for_pins(codec, cfg->hp_pins, cfg->hp_outs,
+			  dacs, &rest, spec->dac_info, &nums,
+			  AUTO_PIN_HP_OUT);
+	fill_dacs_for_pins(codec, cfg->speaker_pins, cfg->speaker_outs,
+			  dacs, &rest, spec->dac_info, &nums,
+			  AUTO_PIN_SPEAKER_OUT);
 	spec->dac_info_filled = nums;
 	/* fill multiout struct */
 	for (i = 0; i < nums; i++) {

@@ -4130,9 +4132,11 @@ static int try_add_pb_volume(struct hda_codec *codec, hda_nid_t dac,
 			     hda_nid_t pin, const char *name, int idx)
 {
 	unsigned int caps;
-	caps = query_amp_caps(codec, dac, HDA_OUTPUT);
-	if (caps & AC_AMPCAP_NUM_STEPS)
-		return cx_auto_add_pb_volume(codec, dac, name, idx);
+	if (!(dac & DAC_SLAVE_FLAG)) {
+		caps = query_amp_caps(codec, dac, HDA_OUTPUT);
+		if (caps & AC_AMPCAP_NUM_STEPS)
+			return cx_auto_add_pb_volume(codec, dac, name, idx);
+	}
 	caps = query_amp_caps(codec, pin, HDA_OUTPUT);
 	if (caps & AC_AMPCAP_NUM_STEPS)
 		return cx_auto_add_pb_volume(codec, pin, name, idx);

sound/pci/hda/patch_conexant.c static int cx_auto_build_output_controls(struct hda_codec *codec) · line 4163

Quoted 6 lines

sound/pci/hda/patch_conexant.c +1 -1

@@ -4155,7 +4159,7 @@ static int cx_auto_build_output_controls(struct hda_codec *codec)
 		const char *label;
 		int idx, type;
 		hda_nid_t dac = spec->dac_info[i].dac;
-		if (!dac || (dac & DAC_SLAVE_FLAG))
+		if (!dac)
 			continue;

With the fix like the above, the whole NULL-dac check should be skipped, too.

How about the revised patch below?

At Thu, 25 Aug 2011 13:26:26 +0200,

Quoted 35 lines · David Henningsson

Quoted 2 lines

From 95ca8a0cd32ad781c54a349c5ce84a9119668f9c Mon Sep 17 00:00:00 2001

From: David Henningsson<[email protected]>
Date: Thu, 25 Aug 2011 13:16:02 +0200
Subject: [PATCH] ALSA: hda: Conexant: Allow different output types to share DAC

Headphones has stopped working for the original reported (a regression compared to 2.6.38). This is because Speaker and Headphones share the same DAC, in which case no Headphones volume control was created. This patch fixes so that both Speaker and Headphones volume controls are created in such scenario.

Buglink http://bugs.launchpad.net/bugs/817943

Signed-off-by David Henningsson<[email protected]>

---
  sound/pci/hda/patch_conexant.c |   40 ++++++++++++++++++++++------------------
  1 files changed, 22 insertions(+), 18 deletions(-)

sound/pci/hda/patch_conexant.c +3 -2

index 197ad93..a29b88d 100644
@@ -3327,22 +3327,23 @@ static int fill_cx_auto_dacs(struct hda_codec *codec, hda_nid_t *dacs)
  }
 
  /* fill pin_dac_pair list from the pin and dac list */
-static int fill_dacs_for_pins(struct hda_codec *codec, hda_nid_t *pins,
+static void fill_dacs_for_pins(struct hda_codec *codec, hda_nid_t *pins,
  			      int num_pins, hda_nid_t *dacs, int *rest,
-			      struct pin_dac_pair *filled, int type)
+			      struct pin_dac_pair *filled, int *saved_nums,
+			      int type)
  {
  	int i, nums;
 

sound/pci/hda/patch_conexant.c static int fill_cx_auto_dacs(struct hda_codec *codec, hda_nid_t *dacs) · line 3343

Quoted 9 lines

sound/pci/hda/patch_conexant.c +2 -2

-	nums = 0;
+	nums = *saved_nums;
  	for (i = 0; i<  num_pins; i++) {
  		filled[nums].pin = pins[i];
  		filled[nums].type = type;
  		filled[nums].dac = get_unassigned_dac(codec, pins[i], dacs, rest);
-		if (!filled[nums].dac&&  i>  0&&  filled[0].dac)
+		if (!filled[nums].dac&&  filled[0].dac)
  			filled[nums].dac = filled[0].dac | DAC_SLAVE_FLAG;

It should try the first DAC of the same type at first, not always to filled[0].

sound/pci/hda/patch_conexant.c static int cx_auto_build_output_controls(struct hda_codec *codec) · line 4163

Quoted 7 lines

sound/pci/hda/patch_conexant.c +1 -1

@@ -4155,7 +4159,7 @@ static int cx_auto_build_output_controls(struct hda_codec *codec)
  		const char *label;
  		int idx, type;
  		hda_nid_t dac = spec->dac_info[i].dac;
-		if (!dac || (dac&  DAC_SLAVE_FLAG))
+		if (!dac)
  			continue;

With the fix like the above, the whole NULL-dac check should be skipped, too.

How about the revised patch below?

From 1947cd5075a43edb1901c3fc9e97ba77d4c7cca9 Mon Sep 17 00:00:00 2001

Quoted 37 lines · Takashi Iwai

At Thu, 25 Aug 2011 13:26:26 +0200,

Quoted 35 lines · David Henningsson

Quoted 2 lines

From 95ca8a0cd32ad781c54a349c5ce84a9119668f9c Mon Sep 17 00:00:00 2001

From: David Henningsson<[email protected]>
Date: Thu, 25 Aug 2011 13:16:02 +0200
Subject: [PATCH] ALSA: hda: Conexant: Allow different output types to share DAC

Headphones has stopped working for the original reported (a regression compared to 2.6.38). This is because Speaker and Headphones share the same DAC, in which case no Headphones volume control was created. This patch fixes so that both Speaker and Headphones volume controls are created in such scenario.

Buglink http://bugs.launchpad.net/bugs/817943

Signed-off-by David Henningsson<[email protected]>

---
  sound/pci/hda/patch_conexant.c |   40 ++++++++++++++++++++++------------------
  1 files changed, 22 insertions(+), 18 deletions(-)

sound/pci/hda/patch_conexant.c +3 -2

index 197ad93..a29b88d 100644
@@ -3327,22 +3327,23 @@ static int fill_cx_auto_dacs(struct hda_codec *codec, hda_nid_t *dacs)
  }
 
  /* fill pin_dac_pair list from the pin and dac list */
-static int fill_dacs_for_pins(struct hda_codec *codec, hda_nid_t *pins,
+static void fill_dacs_for_pins(struct hda_codec *codec, hda_nid_t *pins,
  			      int num_pins, hda_nid_t *dacs, int *rest,
-			      struct pin_dac_pair *filled, int type)
+			      struct pin_dac_pair *filled, int *saved_nums,
+			      int type)
  {
  	int i, nums;
 

Quoted 12 lines

sound/pci/hda/patch_conexant.c static int fill_cx_auto_dacs(struct hda_codec *codec, hda_nid_t *dacs) · line 3348

Quoted 9 lines

sound/pci/hda/patch_conexant.c +2 -2

-	nums = 0;
+	nums = *saved_nums;
  	for (i = 0; i<  num_pins; i++) {
  		filled[nums].pin = pins[i];
  		filled[nums].type = type;
  		filled[nums].dac = get_unassigned_dac(codec, pins[i], dacs, rest);
-		if (!filled[nums].dac&&  i>  0&&  filled[0].dac)
+		if (!filled[nums].dac&&  filled[0].dac)
  			filled[nums].dac = filled[0].dac | DAC_SLAVE_FLAG;

It should try the first DAC of the same type at first, not always to filled[0].

Ok, fair point. I'm attaching a revised patch that also checks that the DAC is actually reachable, which should be even better.

Quoted 10 lines

sound/pci/hda/patch_conexant.c static int cx_auto_build_output_controls(struct hda_codec *codec) · line 4163

Quoted 7 lines

sound/pci/hda/patch_conexant.c +1 -1

@@ -4155,7 +4159,7 @@ static int cx_auto_build_output_controls(struct hda_codec *codec)
  		const char *label;
  		int idx, type;
  		hda_nid_t dac = spec->dac_info[i].dac;
-		if (!dac || (dac&  DAC_SLAVE_FLAG))
+		if (!dac)
  			continue;

With the fix like the above, the whole NULL-dac check should be skipped, too.

Since NULL-dac is essentially an error, this is more a matter of taste IMO.

Quoted 1 line

How about the revised patch below?

Took some of it into the attached version of the patch.