[alsa-devel] [ALC262_FUJITSU patch_realtek.c] Use jack sense for both headphone outputs
Good morning,
Could you please consider the following patch for 2.6.27. The driver starts up with the jacks in an inconsistent state (speakers on, laptop headphone output enabled, dock headphone output enabled) and does not consider their state until the first unsolicited event arrives. In some cases it takes *headphone insert* *headphone removal* *headphone insert* to get the driver to mute the speakers successfully.
Also, both HP outputs are enabled together, rather then individually enabled based on jack sense results.
This patch addresses both problems. Open to comments as usual & sign-off in diff.
Regards, Tony V. LINX sysadmin
At Mon, 19 May 2008 01:00:19 +0100, Tony Vroon wrote:
--- sound/pci/hda/patch_realtek.c.orig 2008-05-17 15:07:35.000000000 +0100 +++ sound/pci/hda/patch_realtek.c 2008-05-19 00:43:35.000000000 +0100 @@ -8757,7 +8757,7 @@ }, };
-/* mute/unmute internal speaker according to the hp jack and mute state */ +/* mute/unmute internal speaker according to the hp jacks and mute state */ static void alc262_fujitsu_automute(struct hda_codec *codec, int force) { struct alc_spec *spec = codec->spec; @@ -8765,13 +8765,28 @@
if (force || !spec->sense_updated) { unsigned int present_int_hp, present_dock_hp;
/* need to execute and sync at first */
snd_hda_codec_read(codec, 0x14, 0, AC_VERB_SET_PIN_SENSE, 0); present_int_hp = snd_hda_codec_read(codec, 0x14, 0, AC_VERB_GET_PIN_SENSE, 0);/* sense laptop headphone jack, enable amp if required */
if ((present_int_hp & 0x80000000) != 0)
snd_hda_codec_amp_stereo(codec, 0x14, HDA_OUTPUT, 0,
HDA_AMP_MUTE, 0);
else
snd_hda_codec_amp_stereo(codec, 0x14, HDA_OUTPUT, 0,
HDA_AMP_MUTE, HDA_AMP_MUTE);
snd_hda_codec_read(codec, 0x1B, 0, AC_VERB_SET_PIN_SENSE, 0); present_dock_hp = snd_hda_codec_read(codec, 0x1b, 0, AC_VERB_GET_PIN_SENSE, 0);/* sense dock headphone jack, enable amp if required */
if ((present_dock_hp & 0x80000000) != 0)
snd_hda_codec_amp_stereo(codec, 0x1b, HDA_OUTPUT, 0,
HDA_AMP_MUTE, 0);
else
snd_hda_codec_amp_stereo(codec, 0x1b, HDA_OUTPUT, 0,
HDA_AMP_MUTE, HDA_AMP_MUTE);
- spec->jack_present = (present_int_hp & 0x80000000) != 0; spec->jack_present |= (present_dock_hp & 0x80000000) != 0; spec->sense_updated = 1;
I get a compile warning, "unused variable mute". Could you fix it?
@@ -8781,10 +8796,9 @@ snd_hda_codec_amp_stereo(codec, 0x15, HDA_OUTPUT, 0, HDA_AMP_MUTE, HDA_AMP_MUTE); } else {
/* unmute internal speaker if necessary */
mute = snd_hda_codec_amp_read(codec, 0x14, 0, HDA_OUTPUT, 0);
Any reason to remove this? Otherwise the internal speaker won't react together with the Master volume.
Maybe a cleaner fix is to remove the bind_sw and bind_vol hacks and use vmaster instead.
thanks,
Takashi
On Mon, 2008-05-19 at 12:23 +0200, Takashi Iwai wrote:
Any reason to remove this? Otherwise the internal speaker won't react together with the Master volume.
It seemed to only take the internal headphone output into account, not the dock headphone output. But if it's just to link the two volumes together I'll add it back. It will fix the compile warning as well. Its purpose was unclear to me.
Will the existing mechanism cope if 0x14 (internal HP output) is muted and 0x1B (dock HP output) is not?
Maybe a cleaner fix is to remove the bind_sw and bind_vol hacks and use vmaster instead.
Is there example code I can look at please?
thanks, Takashi
Regards, Tony V.
At Mon, 19 May 2008 11:31:09 +0100, Tony Vroon wrote:
On Mon, 2008-05-19 at 12:23 +0200, Takashi Iwai wrote:
Any reason to remove this? Otherwise the internal speaker won't react together with the Master volume.
It seemed to only take the internal headphone output into account, not the dock headphone output. But if it's just to link the two volumes together I'll add it back. It will fix the compile warning as well. Its purpose was unclear to me.
Will the existing mechanism cope if 0x14 (internal HP output) is muted and 0x1B (dock HP output) is not?
The code there was to synchronize both HP and speaker outputs with the master switch. The master switch changes 0x14 mute status and calls automute. When no HP is plugged, the internal speaker is unmuted only when master switch is on -- that is, 0x14 is also unmuted.
The logic is a bit complex, and thus I suggested to replace with vmaster.
Maybe a cleaner fix is to remove the bind_sw and bind_vol hacks and use vmaster instead.
Is there example code I can look at please?
Add the individual mixer elements corresponding to HP, dock and PC-speaker switches instead of hackish master switch. Then a virtual master switch will be automatically created.
I'll create a test patch soon later...
thanks,
Takashi
At Mon, 19 May 2008 12:37:10 +0200, I wrote:
At Mon, 19 May 2008 11:31:09 +0100, Tony Vroon wrote:
On Mon, 2008-05-19 at 12:23 +0200, Takashi Iwai wrote:
Any reason to remove this? Otherwise the internal speaker won't react together with the Master volume.
It seemed to only take the internal headphone output into account, not the dock headphone output. But if it's just to link the two volumes together I'll add it back. It will fix the compile warning as well. Its purpose was unclear to me.
Will the existing mechanism cope if 0x14 (internal HP output) is muted and 0x1B (dock HP output) is not?
The code there was to synchronize both HP and speaker outputs with the master switch. The master switch changes 0x14 mute status and calls automute. When no HP is plugged, the internal speaker is unmuted only when master switch is on -- that is, 0x14 is also unmuted.
The logic is a bit complex, and thus I suggested to replace with vmaster.
Maybe a cleaner fix is to remove the bind_sw and bind_vol hacks and use vmaster instead.
Is there example code I can look at please?
Add the individual mixer elements corresponding to HP, dock and PC-speaker switches instead of hackish master switch. Then a virtual master switch will be automatically created.
I'll create a test patch soon later...
On the second thought, it's also not so trivial in this case because of the combination with the jack detection code.
How about the patch below?
Note that the mute bits of 0x14 and 0x1b follow only to the state of master switch. It's on/off regardless of jack detection. Only the internal speaker is affected via jack-detection.
Takashi
---
diff -r e2b60934d1c8 pci/hda/patch_realtek.c --- a/pci/hda/patch_realtek.c Mon May 19 12:32:25 2008 +0200 +++ b/pci/hda/patch_realtek.c Mon May 19 12:54:26 2008 +0200 @@ -8757,35 +8757,39 @@ }, };
-/* mute/unmute internal speaker according to the hp jack and mute state */ +/* mute/unmute internal speaker according to the hp jacks and mute state */ static void alc262_fujitsu_automute(struct hda_codec *codec, int force) { struct alc_spec *spec = codec->spec; unsigned int mute;
if (force || !spec->sense_updated) { - unsigned int present_int_hp, present_dock_hp; + unsigned int present; /* need to execute and sync at first */ snd_hda_codec_read(codec, 0x14, 0, AC_VERB_SET_PIN_SENSE, 0); - present_int_hp = snd_hda_codec_read(codec, 0x14, 0, - AC_VERB_GET_PIN_SENSE, 0); - snd_hda_codec_read(codec, 0x1B, 0, AC_VERB_SET_PIN_SENSE, 0); - present_dock_hp = snd_hda_codec_read(codec, 0x1b, 0, - AC_VERB_GET_PIN_SENSE, 0); - spec->jack_present = (present_int_hp & 0x80000000) != 0; - spec->jack_present |= (present_dock_hp & 0x80000000) != 0; + /* check laptop HP jack */ + present = snd_hda_codec_read(codec, 0x14, 0, + AC_VERB_GET_PIN_SENSE, 0); + /* need to execute and sync at first */ + snd_hda_codec_read(codec, 0x1b, 0, AC_VERB_SET_PIN_SENSE, 0); + /* check docking HP jack */ + present |= snd_hda_codec_read(codec, 0x1b, 0, + AC_VERB_GET_PIN_SENSE, 0); + if (present & AC_PINSENSE_PRESENCE) + spec->jack_present = 1; + else + spec->jack_present = 0; spec->sense_updated = 1; } - if (spec->jack_present) { - /* mute internal speaker */ - snd_hda_codec_amp_stereo(codec, 0x15, HDA_OUTPUT, 0, - HDA_AMP_MUTE, HDA_AMP_MUTE); - } else { - /* unmute internal speaker if necessary */ + /* unmute internal speaker only if both HPs are unplugged and + * master switch is on + */ + if (spec->jack_present) + mute = HDA_AMP_MUTE; + else mute = snd_hda_codec_amp_read(codec, 0x14, 0, HDA_OUTPUT, 0); - snd_hda_codec_amp_stereo(codec, 0x15, HDA_OUTPUT, 0, - HDA_AMP_MUTE, mute); - } + snd_hda_codec_amp_stereo(codec, 0x15, HDA_OUTPUT, 0, + HDA_AMP_MUTE, mute); }
/* unsolicited event for HP jack sensing */ @@ -8794,6 +8798,11 @@ { if ((res >> 26) != ALC_HP_EVENT) return; + alc262_fujitsu_automute(codec, 1); +} + +static void alc262_fujitsu_init_hook(struct hda_codec *codec) +{ alc262_fujitsu_automute(codec, 1); }
@@ -9570,6 +9579,7 @@ .channel_mode = alc262_modes, .input_mux = &alc262_fujitsu_capture_source, .unsol_event = alc262_fujitsu_unsol_event, + .init_hook = alc262_fujitsu_init_hook, }, [ALC262_HP_BPC] = { .mixers = { alc262_HP_BPC_mixer },
On Mon, 2008-05-19 at 12:56 +0200, Takashi Iwai wrote:
On the second thought, it's also not so trivial in this case because of the combination with the jack detection code.
How about the patch below?
Acked-by: Tony Vroon tony@linx.net
The jacks are no longer inconsistent at boot-up. Correct state is maintained, and no repeated plug/unplug is required before the speaker mutes/unmutes.
Takashi
Regards, Tony V.
participants (2)
-
Takashi Iwai
-
Tony Vroon