Skip to contentSkip to conversation

ALSA: hda - Send CONNECT_SEL on capture start

PATCH In review Dylan Reid 2 participants 4 messages Nov 16, 2011 Sound

lore
Share & tools
lore

Dylan Reid

author
#0 lore raw

PATCH 1/1 ALSA: hda - Send CONNECT_SEL on capture start

cur_adc is set to zero in cs_capture_pcm_prepare. This causes the SET_CONNECT_SEL verb to be sent to node zero when doing automic after recording. Delay sending the command until we are about to start capture and cur_adc is set to the correct NID.

---
 sound/pci/hda/patch_cirrus.c |   16 ++++++++++------
 1 files changed, 10 insertions(+), 6 deletions(-)
sound/pci/hda/patch_cirrus.c +10 -6
index 2a2d864..30f2f6a 100644
@@ -250,6 +250,9 @@ static int cs_capture_pcm_prepare(struct hda_pcm_stream *hinfo,
 	spec->cur_adc = spec->adc_nid[spec->cur_input];
 	spec->cur_adc_stream_tag = stream_tag;
 	spec->cur_adc_format = format;
+	snd_hda_codec_write(codec, spec->cur_adc, 0,
+			    AC_VERB_SET_CONNECT_SEL,
+			    spec->adc_idx[spec->cur_input]);
 	snd_hda_codec_setup_stream(codec, spec->cur_adc, stream_tag, 0, format);
 	return 0;
 }
@@ -685,13 +688,13 @@ static int change_cur_input(struct hda_codec *codec, unsigned int idx,
 		/* stream is running, let's swap the current ADC */
 		__snd_hda_codec_cleanup_stream(codec, spec->cur_adc, 1);
 		spec->cur_adc = spec->adc_nid[idx];
+		snd_hda_codec_write(codec, spec->cur_adc, 0,
+				    AC_VERB_SET_CONNECT_SEL,
+				    spec->adc_idx[idx]);
 		snd_hda_codec_setup_stream(codec, spec->cur_adc,
 					   spec->cur_adc_stream_tag, 0,
 					   spec->cur_adc_format);
 	}
-	snd_hda_codec_write(codec, spec->cur_adc, 0,
-			    AC_VERB_SET_CONNECT_SEL,
-			    spec->adc_idx[idx]);
 	spec->cur_input = idx;
 	return 1;
 }
@@ -974,9 +977,10 @@ static void cs_automic(struct hda_codec *codec)
 			spec->cur_input = spec->last_input;
 		}
 
-		snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
-					AC_VERB_SET_CONNECT_SEL,
-					spec->adc_idx[spec->cur_input]);
+		if (spec->cur_adc)
+			snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
+				AC_VERB_SET_CONNECT_SEL,
+				spec->adc_idx[spec->cur_input]);
 	} else {
 		if (present)
 			change_cur_input(codec, spec->automic_idx, 0);
Signature · 1 line
1.7.5.185.g0b9dee
#1 lore raw

At Wed, 16 Nov 2011 00:25:49 -0800,

Quoted 2 lines · Dylan Reid

cur_adc is set to zero in cs_capture_pcm_prepare.

Not really. It's set to zero until cs_capture_pcm_preare() is called.

Quoted 28 lines

This causes the
SET_CONNECT_SEL verb to be sent to node zero when doing automic after recording. Delay sending the command until we are about to start capture and cur_adc is set to the correct NID.

---
 sound/pci/hda/patch_cirrus.c |   16 ++++++++++------
 1 files changed, 10 insertions(+), 6 deletions(-)
sound/pci/hda/patch_cirrus.c +3 -0
index 2a2d864..30f2f6a 100644
@@ -250,6 +250,9 @@ static int cs_capture_pcm_prepare(struct hda_pcm_stream *hinfo,
 	spec->cur_adc = spec->adc_nid[spec->cur_input];
 	spec->cur_adc_stream_tag = stream_tag;
 	spec->cur_adc_format = format;
+	snd_hda_codec_write(codec, spec->cur_adc, 0,
+			    AC_VERB_SET_CONNECT_SEL,
+			    spec->adc_idx[spec->cur_input]);
 	snd_hda_codec_setup_stream(codec, spec->cur_adc, stream_tag, 0, format);
 	return 0;
 }
@@ -685,13 +688,13 @@ static int change_cur_input(struct hda_codec *codec, unsigned int idx,
 		/* stream is running, let's swap the current ADC */
 		__snd_hda_codec_cleanup_stream(codec, spec->cur_adc, 1);
 		spec->cur_adc = spec->adc_nid[idx];

sound/pci/hda/patch_cirrus.c static int change_cur_input(struct hda_codec *codec, unsigned int idx, · line 694

Quoted 10 lines
sound/pci/hda/patch_cirrus.c +3 -3
+		snd_hda_codec_write(codec, spec->cur_adc, 0,
+				    AC_VERB_SET_CONNECT_SEL,
+				    spec->adc_idx[idx]);
 		snd_hda_codec_setup_stream(codec, spec->cur_adc,
 					   spec->cur_adc_stream_tag, 0,
 					   spec->cur_adc_format);
 	}
-	snd_hda_codec_write(codec, spec->cur_adc, 0,
-			    AC_VERB_SET_CONNECT_SEL,
-			    spec->adc_idx[idx]);

This breaks the case when the input-source is changed dynamically during the recording with CS420x.

Quoted 4 lines
sound/pci/hda/patch_cirrus.c +0 -0
 	spec->cur_input = idx;
 	return 1;
 }

sound/pci/hda/patch_cirrus.c static void cs_automic(struct hda_codec *codec) · line 983

Quoted 10 lines
sound/pci/hda/patch_cirrus.c +4 -3
@@ -974,9 +977,10 @@ static void cs_automic(struct hda_codec *codec)
 			spec->cur_input = spec->last_input;
 		}
 
-		snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
-					AC_VERB_SET_CONNECT_SEL,
-					spec->adc_idx[spec->cur_input]);
+		if (spec->cur_adc)
+			snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
+				AC_VERB_SET_CONNECT_SEL,
+				spec->adc_idx[spec->cur_input]);

Well, it'd be better to create a function instead of scattering the same verb call allover.

How about the (untested) patch below instead?

Takashi

---

sound/pci/hda/patch_cirrus.c +13 -10
index 2a2d864..2fbab8e 100644
@@ -237,6 +237,15 @@ static int cs_dig_playback_pcm_cleanup(struct hda_pcm_stream *hinfo,
 	return snd_hda_multi_out_dig_cleanup(codec, &spec->multiout);
 }
 
+static void cs_update_input_select(struct hda_codec *codec)
+{
+	struct cs_spec *spec = codec->spec;
+	if (spec->cur_adc)
+		snd_hda_codec_write(codec, spec->cur_adc, 0,
+				    AC_VERB_SET_CONNECT_SEL,
+				    spec->adc_idx[spec->cur_input]);
+}
+
 /*
  * Analog capture
  */
@@ -250,6 +259,7 @@ static int cs_capture_pcm_prepare(struct hda_pcm_stream *hinfo,
 	spec->cur_adc = spec->adc_nid[spec->cur_input];
 	spec->cur_adc_stream_tag = stream_tag;
 	spec->cur_adc_format = format;
+	cs_update_input_select(codec);
 	snd_hda_codec_setup_stream(codec, spec->cur_adc, stream_tag, 0, format);
 	return 0;
 }
@@ -689,10 +699,8 @@ static int change_cur_input(struct hda_codec *codec, unsigned int idx,
 					   spec->cur_adc_stream_tag, 0,
 					   spec->cur_adc_format);
 	}
-	snd_hda_codec_write(codec, spec->cur_adc, 0,
-			    AC_VERB_SET_CONNECT_SEL,
-			    spec->adc_idx[idx]);
 	spec->cur_input = idx;
+	cs_update_input_select(codec);
 	return 1;
 }
 
@@ -973,10 +981,7 @@ static void cs_automic(struct hda_codec *codec)
 		} else  {
 			spec->cur_input = spec->last_input;
 		}
-
-		snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
-					AC_VERB_SET_CONNECT_SEL,
-					spec->adc_idx[spec->cur_input]);
+		cs_update_input_select(codec);
 	} else {
 		if (present)
 			change_cur_input(codec, spec->automic_idx, 0);
@@ -1073,9 +1078,7 @@ static void init_input(struct hda_codec *codec)
 			cs_automic(codec);
 		else  {
 			spec->cur_adc = spec->adc_nid[spec->cur_input];
-			snd_hda_codec_write(codec, spec->cur_adc, 0,
-					AC_VERB_SET_CONNECT_SEL,
-					spec->adc_idx[spec->cur_input]);
+			cs_update_input_select(codec);
 		}
 	} else {
 		change_cur_input(codec, spec->cur_input, 1);
author
#2 lore raw
Quoted 68 lines · Takashi Iwai

At Wed, 16 Nov 2011 00:25:49 -0800,

Quoted 2 lines · Dylan Reid

cur_adc is set to zero in cs_capture_pcm_prepare.

Not really.  It's set to zero until cs_capture_pcm_preare() is called.

Quoted 28 lines

 This causes the
SET_CONNECT_SEL verb to be sent to node zero when doing automic after recording.  Delay sending the command until we are about to start capture and cur_adc is set to the correct NID.

---
 sound/pci/hda/patch_cirrus.c |   16 ++++++++++------

 1 files changed, 10 insertions(+), 6 deletions(-)
sound/pci/hda/patch_cirrus.c +3 -0
index 2a2d864..30f2f6a 100644
@@ -250,6 +250,9 @@ static int cs_capture_pcm_prepare(struct hda_pcm_stream *hinfo,
       spec->cur_adc = spec->adc_nid[spec->cur_input];
       spec->cur_adc_stream_tag = stream_tag;
       spec->cur_adc_format = format;
+     snd_hda_codec_write(codec, spec->cur_adc, 0,
+                         AC_VERB_SET_CONNECT_SEL,
+                         spec->adc_idx[spec->cur_input]);
       snd_hda_codec_setup_stream(codec, spec->cur_adc, stream_tag, 0, format);
       return 0;
  }
@@ -685,13 +688,13 @@ static int change_cur_input(struct hda_codec *codec, unsigned int idx,
               /* stream is running, let's swap the current ADC */
               __snd_hda_codec_cleanup_stream(codec, spec->cur_adc, 1);
               spec->cur_adc = spec->adc_nid[idx];

sound/pci/hda/patch_cirrus.c static int change_cur_input(struct hda_codec *codec, unsigned int idx, · line 694

Quoted 10 lines
sound/pci/hda/patch_cirrus.c +3 -3
+             snd_hda_codec_write(codec, spec->cur_adc, 0,
+                                 AC_VERB_SET_CONNECT_SEL,
+                                 spec->adc_idx[idx]);
               snd_hda_codec_setup_stream(codec, spec->cur_adc,
                                          spec->cur_adc_stream_tag, 0,
                                          spec->cur_adc_format);
       }
-     snd_hda_codec_write(codec, spec->cur_adc, 0,
-                         AC_VERB_SET_CONNECT_SEL,
-                         spec->adc_idx[idx]);

This breaks the case when the input-source is changed dynamically during the recording with CS420x.

Quoted 4 lines
sound/pci/hda/patch_cirrus.c +0 -0
       spec->cur_input = idx;
       return 1;
  }

sound/pci/hda/patch_cirrus.c static void cs_automic(struct hda_codec *codec) · line 983

Quoted 10 lines
sound/pci/hda/patch_cirrus.c +4 -3
@@ -974,9 +977,10 @@ static void cs_automic(struct hda_codec *codec)
                       spec->cur_input = spec->last_input;
               }
 
-             snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
-                                     AC_VERB_SET_CONNECT_SEL,
-                                     spec->adc_idx[spec->cur_input]);
+             if (spec->cur_adc)
+                     snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
+                             AC_VERB_SET_CONNECT_SEL,
+                             spec->adc_idx[spec->cur_input]);

Well, it'd be better to create a function instead of scattering the same verb call allover.

Quoted 1 line

How about the (untested) patch below instead?

That is better. I tested it this morning and it works like a charm. Can you apply that?

Thanks Takashi,

Dylan

Quoted 69 lines

Takashi

---

sound/pci/hda/patch_cirrus.c +13 -10
index 2a2d864..2fbab8e 100644
@@ -237,6 +237,15 @@ static int cs_dig_playback_pcm_cleanup(struct hda_pcm_stream *hinfo,
        return snd_hda_multi_out_dig_cleanup(codec, &spec->multiout);
  }
 
+static void cs_update_input_select(struct hda_codec *codec)
+{
+       struct cs_spec *spec = codec->spec;
+       if (spec->cur_adc)
+               snd_hda_codec_write(codec, spec->cur_adc, 0,
+                                   AC_VERB_SET_CONNECT_SEL,
+                                   spec->adc_idx[spec->cur_input]);
+}
+
  /*
  * Analog capture
  */
@@ -250,6 +259,7 @@ static int cs_capture_pcm_prepare(struct hda_pcm_stream *hinfo,
        spec->cur_adc = spec->adc_nid[spec->cur_input];
        spec->cur_adc_stream_tag = stream_tag;
        spec->cur_adc_format = format;
+       cs_update_input_select(codec);
        snd_hda_codec_setup_stream(codec, spec->cur_adc, stream_tag, 0, format);
        return 0;
  }
@@ -689,10 +699,8 @@ static int change_cur_input(struct hda_codec *codec, unsigned int idx,
                                           spec->cur_adc_stream_tag, 0,
                                           spec->cur_adc_format);
        }
-       snd_hda_codec_write(codec, spec->cur_adc, 0,
-                           AC_VERB_SET_CONNECT_SEL,
-                           spec->adc_idx[idx]);
        spec->cur_input = idx;
+       cs_update_input_select(codec);
        return 1;
  }
 
@@ -973,10 +981,7 @@ static void cs_automic(struct hda_codec *codec)
                } else  {
                        spec->cur_input = spec->last_input;
                }
-
-               snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
-                                       AC_VERB_SET_CONNECT_SEL,
-                                       spec->adc_idx[spec->cur_input]);
+               cs_update_input_select(codec);
        } else {
                if (present)
                        change_cur_input(codec, spec->automic_idx, 0);
@@ -1073,9 +1078,7 @@ static void init_input(struct hda_codec *codec)
                        cs_automic(codec);
                else  {
                        spec->cur_adc = spec->adc_nid[spec->cur_input];
-                       snd_hda_codec_write(codec, spec->cur_adc, 0,
-                                       AC_VERB_SET_CONNECT_SEL,
-                                       spec->adc_idx[spec->cur_input]);
+                       cs_update_input_select(codec);
                }
        } else {
                change_cur_input(codec, spec->cur_input, 1);
#3 lore raw

At Wed, 16 Nov 2011 09:02:32 -0800,

Quoted 70 lines · Dylan Reid
Quoted 68 lines · Takashi Iwai

At Wed, 16 Nov 2011 00:25:49 -0800,

Quoted 2 lines · Dylan Reid

cur_adc is set to zero in cs_capture_pcm_prepare.

Not really.  It's set to zero until cs_capture_pcm_preare() is called.

Quoted 28 lines

 This causes the
SET_CONNECT_SEL verb to be sent to node zero when doing automic after recording.  Delay sending the command until we are about to start capture and cur_adc is set to the correct NID.

---
 sound/pci/hda/patch_cirrus.c |   16 ++++++++++------

 1 files changed, 10 insertions(+), 6 deletions(-)
sound/pci/hda/patch_cirrus.c +3 -0
index 2a2d864..30f2f6a 100644
@@ -250,6 +250,9 @@ static int cs_capture_pcm_prepare(struct hda_pcm_stream *hinfo,
       spec->cur_adc = spec->adc_nid[spec->cur_input];
       spec->cur_adc_stream_tag = stream_tag;
       spec->cur_adc_format = format;
+     snd_hda_codec_write(codec, spec->cur_adc, 0,
+                         AC_VERB_SET_CONNECT_SEL,
+                         spec->adc_idx[spec->cur_input]);
       snd_hda_codec_setup_stream(codec, spec->cur_adc, stream_tag, 0, format);
       return 0;
  }
@@ -685,13 +688,13 @@ static int change_cur_input(struct hda_codec *codec, unsigned int idx,
               /* stream is running, let's swap the current ADC */
               __snd_hda_codec_cleanup_stream(codec, spec->cur_adc, 1);
               spec->cur_adc = spec->adc_nid[idx];

sound/pci/hda/patch_cirrus.c static int change_cur_input(struct hda_codec *codec, unsigned int idx, · line 694

Quoted 10 lines
sound/pci/hda/patch_cirrus.c +3 -3
+             snd_hda_codec_write(codec, spec->cur_adc, 0,
+                                 AC_VERB_SET_CONNECT_SEL,
+                                 spec->adc_idx[idx]);
               snd_hda_codec_setup_stream(codec, spec->cur_adc,
                                          spec->cur_adc_stream_tag, 0,
                                          spec->cur_adc_format);
       }
-     snd_hda_codec_write(codec, spec->cur_adc, 0,
-                         AC_VERB_SET_CONNECT_SEL,
-                         spec->adc_idx[idx]);

This breaks the case when the input-source is changed dynamically during the recording with CS420x.

Quoted 4 lines
sound/pci/hda/patch_cirrus.c +0 -0
       spec->cur_input = idx;
       return 1;
  }

sound/pci/hda/patch_cirrus.c static void cs_automic(struct hda_codec *codec) · line 983

Quoted 10 lines
sound/pci/hda/patch_cirrus.c +4 -3
@@ -974,9 +977,10 @@ static void cs_automic(struct hda_codec *codec)
                       spec->cur_input = spec->last_input;
               }
 
-             snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
-                                     AC_VERB_SET_CONNECT_SEL,
-                                     spec->adc_idx[spec->cur_input]);
+             if (spec->cur_adc)
+                     snd_hda_codec_write_cache(codec, spec->cur_adc, 0,
+                             AC_VERB_SET_CONNECT_SEL,
+                             spec->adc_idx[spec->cur_input]);

Well, it'd be better to create a function instead of scattering the same verb call allover.

Quoted 3 lines
Quoted 1 line

How about the (untested) patch below instead?

That is better. I tested it this morning and it works like a charm. Can you apply that?

Thanks for testing. Now applied to sound git tree.

Takashi
_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel