[alsa-devel] [PATCH] ALSA: hda - avoid unneccesary indices on "Headphone Jack" controls
In case there is one "Headphone Jack" and one "Dock Headphone Jack", one of them will get an index, even though that is not needed. This patch fixes that issue.
BugLink: https://bugs.launchpad.net/bugs/1060729 Signed-off-by: David Henningsson david.henningsson@canonical.com --- sound/pci/hda/hda_auto_parser.c | 49 +++++++++++++++++++++++++++++---------- 1 file changed, 37 insertions(+), 12 deletions(-)
Alsa-info at: https://launchpadlibrarian.net/118211803/AlsaInfo
diff --git a/sound/pci/hda/hda_auto_parser.c b/sound/pci/hda/hda_auto_parser.c index a98e25e..4ec6dc8 100644 --- a/sound/pci/hda/hda_auto_parser.c +++ b/sound/pci/hda/hda_auto_parser.c @@ -498,6 +498,38 @@ static const char *check_output_sfx(hda_nid_t nid, const hda_nid_t *pins, return channel_sfx[i]; }
+static const char *check_output_pfx(struct hda_codec *codec, hda_nid_t nid) +{ + unsigned int def_conf = snd_hda_codec_get_pincfg(codec, nid); + int attr = snd_hda_get_input_pin_attr(def_conf); + + /* check the location */ + switch (attr) { + case INPUT_PIN_ATTR_DOCK: + return "Dock "; + case INPUT_PIN_ATTR_FRONT: + return "Front "; + } + return ""; +} + +static int get_hp_label_index(struct hda_codec *codec, hda_nid_t nid, + const hda_nid_t *pins, int num_pins) +{ + int i, j, idx = 0; + + const char *pfx = check_output_pfx(codec, nid); + + i = find_idx_in_nid_list(nid, pins, num_pins); + if (i < 0) + return -1; + for (j = 0; j < i; j++) + if (pfx == check_output_pfx(codec, pins[j])) + idx++; + + return idx; +} + static int fill_audio_out_name(struct hda_codec *codec, hda_nid_t nid, const struct auto_pin_cfg *cfg, const char *name, char *label, int maxlen, @@ -505,20 +537,13 @@ static int fill_audio_out_name(struct hda_codec *codec, hda_nid_t nid, { unsigned int def_conf = snd_hda_codec_get_pincfg(codec, nid); int attr = snd_hda_get_input_pin_attr(def_conf); - const char *pfx = "", *sfx = ""; + const char *pfx, *sfx = "";
/* handle as a speaker if it's a fixed line-out */ if (!strcmp(name, "Line Out") && attr == INPUT_PIN_ATTR_INT) name = "Speaker"; - /* check the location */ - switch (attr) { - case INPUT_PIN_ATTR_DOCK: - pfx = "Dock "; - break; - case INPUT_PIN_ATTR_FRONT: - pfx = "Front "; - break; - } + pfx = check_output_pfx(codec, nid); + if (cfg) { /* try to give a unique suffix if needed */ sfx = check_output_sfx(nid, cfg->line_out_pins, cfg->line_outs, @@ -528,8 +553,8 @@ static int fill_audio_out_name(struct hda_codec *codec, hda_nid_t nid, indexp); if (!sfx) { /* don't add channel suffix for Headphone controls */ - int idx = find_idx_in_nid_list(nid, cfg->hp_pins, - cfg->hp_outs); + int idx = get_hp_label_index(codec, nid, cfg->hp_pins, + cfg->hp_outs); if (idx >= 0) *indexp = idx; sfx = "";
At Wed, 3 Oct 2012 11:12:53 +0200, David Henningsson wrote:
In case there is one "Headphone Jack" and one "Dock Headphone Jack", one of them will get an index, even though that is not needed. This patch fixes that issue.
BugLink: https://bugs.launchpad.net/bugs/1060729 Signed-off-by: David Henningsson david.henningsson@canonical.com
Applied. This one doesn't need to go to stable, right?
Takashi
sound/pci/hda/hda_auto_parser.c | 49 +++++++++++++++++++++++++++++---------- 1 file changed, 37 insertions(+), 12 deletions(-)
Alsa-info at: https://launchpadlibrarian.net/118211803/AlsaInfo
diff --git a/sound/pci/hda/hda_auto_parser.c b/sound/pci/hda/hda_auto_parser.c index a98e25e..4ec6dc8 100644 --- a/sound/pci/hda/hda_auto_parser.c +++ b/sound/pci/hda/hda_auto_parser.c @@ -498,6 +498,38 @@ static const char *check_output_sfx(hda_nid_t nid, const hda_nid_t *pins, return channel_sfx[i]; }
+static const char *check_output_pfx(struct hda_codec *codec, hda_nid_t nid) +{
- unsigned int def_conf = snd_hda_codec_get_pincfg(codec, nid);
- int attr = snd_hda_get_input_pin_attr(def_conf);
- /* check the location */
- switch (attr) {
- case INPUT_PIN_ATTR_DOCK:
return "Dock ";
- case INPUT_PIN_ATTR_FRONT:
return "Front ";
- }
- return "";
+}
+static int get_hp_label_index(struct hda_codec *codec, hda_nid_t nid,
const hda_nid_t *pins, int num_pins)
+{
- int i, j, idx = 0;
- const char *pfx = check_output_pfx(codec, nid);
- i = find_idx_in_nid_list(nid, pins, num_pins);
- if (i < 0)
return -1;
- for (j = 0; j < i; j++)
if (pfx == check_output_pfx(codec, pins[j]))
idx++;
- return idx;
+}
static int fill_audio_out_name(struct hda_codec *codec, hda_nid_t nid, const struct auto_pin_cfg *cfg, const char *name, char *label, int maxlen, @@ -505,20 +537,13 @@ static int fill_audio_out_name(struct hda_codec *codec, hda_nid_t nid, { unsigned int def_conf = snd_hda_codec_get_pincfg(codec, nid); int attr = snd_hda_get_input_pin_attr(def_conf);
- const char *pfx = "", *sfx = "";
const char *pfx, *sfx = "";
/* handle as a speaker if it's a fixed line-out */ if (!strcmp(name, "Line Out") && attr == INPUT_PIN_ATTR_INT) name = "Speaker";
- /* check the location */
- switch (attr) {
- case INPUT_PIN_ATTR_DOCK:
pfx = "Dock ";
break;
- case INPUT_PIN_ATTR_FRONT:
pfx = "Front ";
break;
- }
- pfx = check_output_pfx(codec, nid);
- if (cfg) { /* try to give a unique suffix if needed */ sfx = check_output_sfx(nid, cfg->line_out_pins, cfg->line_outs,
@@ -528,8 +553,8 @@ static int fill_audio_out_name(struct hda_codec *codec, hda_nid_t nid, indexp); if (!sfx) { /* don't add channel suffix for Headphone controls */
int idx = find_idx_in_nid_list(nid, cfg->hp_pins,
cfg->hp_outs);
int idx = get_hp_label_index(codec, nid, cfg->hp_pins,
cfg->hp_outs); if (idx >= 0) *indexp = idx; sfx = "";
-- 1.7.9.5
On 10/08/2012 10:18 AM, Takashi Iwai wrote:
At Wed, 3 Oct 2012 11:12:53 +0200, David Henningsson wrote:
In case there is one "Headphone Jack" and one "Dock Headphone Jack", one of them will get an index, even though that is not needed. This patch fixes that issue.
BugLink: https://bugs.launchpad.net/bugs/1060729 Signed-off-by: David Henningsson david.henningsson@canonical.com
Applied. This one doesn't need to go to stable, right?
Good question. I guess it depends on how we look upon kcontrol indices in general.
PulseAudio never picks anything up if it has an index, therefore it becomes a real problem/bug for users. (Not sure if one could argue that this is a problem in PulseAudio rather than the kernel. Haven't thought that through.)
OTOH, if the reasoning is that you would break custom made scripts that depend on this index, i e it becomes a change of behaviour.
To be pragmatic about it, the chances that people are depending on this index being non-zero are probably very low, so I don't mind sending this one to stable.
participants (2)
-
David Henningsson
-
Takashi Iwai