Sorry for not jumping in earlier, LinuxFest NW & job interviews have been taking my time recently.

First, don't make any changes to cxt5047 code as it doesn't pertain to your system and is currently working for everyone that has tested it. Your patch would break their functionality.

Second, the proposed patch has the exact opposite effect that other systems need, which I find rather interesting.

I'll do some experimenting with my test board here and see what it reports. I'd like to hold off on any patches for the immediate moment though, as it would seem to break more systems than it would fix. This may require a separation of functions for each system.

Tobin

On Wed, 2007-05-02 at 14:28 +0300, Andrei Popa wrote:

On Wed, 2007-05-02 at 13:01 +0200, Takashi Iwai wrote:

...

[keep cc to alsa-devel please]

I am not subscribed to alsa-devel and when I send there a mail I get a replay: " Your mail to 'Alsa-devel' with the subject

Re: [ALSA][hda-codec] fix-connexant-headphone-mute

Is being held until the list moderator can review it for approval.

The reason it is being held:

Post by non-member to a members-only list

"

...

At Wed, 02 May 2007 13:47:53 +0300, Andrei Popa wrote:

...

On Wed, 2007-05-02 at 12:16 +0200, Takashi Iwai wrote:

...

At Sun, 29 Apr 2007 05:19:50 +0300, Andrei Popa wrote:

...

On Sun, 2007-04-29 at 04:43 +0300, Andrei Popa wrote:

...

Hi,

The sound works ok with the built-in speakers on my Fujitsu laptop(amilo pro v3205). But if I plug in headphones in the audio out jack there is no sound in the headphones. With kernel 2.6.20 it worked ok. With 2.6.21 it doesn't. This BUG is reported in ALSA bugtracking system with ID 0003039.

Patch that fixes this bug:

diff --git a/sound/pci/hda/patch_conexant.c b/sound/pci/hda/patch_conexant.c index 46e93c6..5edd330 100644 --- a/sound/pci/hda/patch_conexant.c +++ b/sound/pci/hda/patch_conexant.c @@ -636,7 +636,7 @@ static void cxt5045_hp_automute(struct hda_codec *codec)

    bits = (spec->hp_present || !spec->cur_eapd) ? 0x80 : 0;
    snd_hda_codec_amp_update(codec, 0x10, 0, HDA_OUTPUT, 0, 0x80,

bits);

•   snd_hda_codec_amp_update(codec, 0x10, 1, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x10, 1, HDA_OUTPUT, 0, 0,
  

bits); }

/* unsolicited event for HP jack sensing */ @@ -988,10 +988,10 @@ static void cxt5047_hp_automute(struct hda_codec *codec)

    bits = (spec->hp_present || !spec->cur_eapd) ? 0x80 : 0;
    snd_hda_codec_amp_update(codec, 0x1d, 0, HDA_OUTPUT, 0, 0x80,

bits);

•   snd_hda_codec_amp_update(codec, 0x1d, 1, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x1d, 1, HDA_OUTPUT, 0, 0,
  

bits); /* Mute/Unmute PCM 2 for good measure - some systems need this */ snd_hda_codec_amp_update(codec, 0x1c, 0, HDA_OUTPUT, 0, 0x80, bits);

•   snd_hda_codec_amp_update(codec, 0x1c, 1, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x1c, 1, HDA_OUTPUT, 0, 0,
  

bits); }

/* toggle input of built-in and mic jack appropriately */

first patch was broken. it only worked for the right channel. this fixes this:

diff --git a/sound/pci/hda/patch_conexant.c b/sound/pci/hda/patch_conexant.c index 46e93c6..0822df7 100644 --- a/sound/pci/hda/patch_conexant.c +++ b/sound/pci/hda/patch_conexant.c @@ -635,8 +635,8 @@ static void cxt5045_hp_automute(struct hda_codec *codec) AC_VERB_GET_PIN_SENSE, 0) & 0x80000000;

    bits = (spec->hp_present || !spec->cur_eapd) ? 0x80 : 0;

•   snd_hda_codec_amp_update(codec, 0x10, 0, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x10, 1, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x10, 0, HDA_OUTPUT, 0, 0,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x10, 1, HDA_OUTPUT, 0, 0,
  

bits); }

/* unsolicited event for HP jack sensing */ @@ -987,11 +987,11 @@ static void cxt5047_hp_automute(struct hda_codec *codec) AC_VERB_GET_PIN_SENSE, 0) & 0x80000000;

    bits = (spec->hp_present || !spec->cur_eapd) ? 0x80 : 0;

•   snd_hda_codec_amp_update(codec, 0x1d, 0, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x1d, 1, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x1d, 0, HDA_OUTPUT, 0, 0,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x1d, 1, HDA_OUTPUT, 0, 0,
  

bits); /* Mute/Unmute PCM 2 for good measure - some systems need this */

•   snd_hda_codec_amp_update(codec, 0x1c, 0, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x1c, 1, HDA_OUTPUT, 0, 0x80,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x1c, 0, HDA_OUTPUT, 0, 0,
  

bits);

•   snd_hda_codec_amp_update(codec, 0x1c, 1, HDA_OUTPUT, 0, 0,
  

bits); }

/* toggle input of built-in and mic jack appropriately */

But this means to _unmute_ the speaker (0x1d) when you plug HP. So, something else should be wrong. Is hp_present detected properly per plugging or unplugging?

when I debugged the problem I think hp_present was "0" when the HP were plugged in and "-28953535"(a big negative number) when the HP were unplugged.

Weird, then this returns exactly the inversed bit. Could you double-check?

I was mistaken, it's how you said it must be:

static void cxt5045_hp_automute(struct hda_codec *codec) { struct conexant_spec *spec = codec->spec; unsigned int bits;

    spec->hp_present = snd_hda_codec_read(codec, 0x11, 0,
                                 AC_VERB_GET_PIN_SENSE, 0) &

0x80000000;

    printk(KERN_INFO "hp_present=%d\n",spec->hp_present);

    bits = (spec->hp_present || !spec->cur_eapd) ? 0x80 : 0;
    snd_hda_codec_amp_update(codec, 0x10, 0, HDA_OUTPUT, 0, 0,

bits); snd_hda_codec_amp_update(codec, 0x10, 1, HDA_OUTPUT, 0, 0, bits); }

hp_present=0 /* HP unplugged */ hp_present=-2147483648 /* HP plugged */

...

Anyway, having both this auto-mute and manual switches is confusing. If they exist, we'd need to add notifier to the control at unsol event handler.

Takashi