I'm not entirely sure that the code you have is the latest. Could you compare your patch_conexant.c with this one: http://members.dsl-only.net/~tdavis/patch_conexant.c

This is my latest version (it's also in the alsa hg tree).

Tobin

On Tue, 2007-03-27 at 13:48 -0300, Rodrigo Luiz wrote:

Hi Tobin,

I'm using a x86-64 arch with a SMP support.

More details in the link of output from my alsa-info.sh using the original patch_conexant.c

http://pastebin.ca/412287

Rodrigo.

Tobin Davis wrote:

...

What system do you have? Can you post the output from this script: http://bulletproof.servebeer.com/alsa/scripts/alsa-info.sh

I'm working on that driver, and am interested in knowing what I need to add to fix it.

Tobin

On Tue, 2007-03-27 at 12:17 -0300, Rodrigo Luiz wrote:

...

Hi.

I have a HDA NVidia, with a Conexant chipset CX20549 Venice (CXT5045)

I tried to test the last alsa-kernel in -mm tree to support my mic. (2.6.21-rc4-mm1 and 2.6.21-rc5-mm2)

When I try to load the hda-intel, I receive this message:

Mar 27 11:29:08 dig kernel: hda_intel: azx_get_response timeout, switching to polling mode Mar 27 11:29:09 dig kernel: hda_intel: azx_get_response timeout, switching to single_cmd mode

After loaded, the speakers doesn't work. Headphone works ok. Mic too.

Look in code, I apply this comment and now the speakers works for me:

--- patch_conexant.c.orig 2007-03-27 11:37:43.000000000 -0300 +++ patch_conexant.c 2007-03-27 11:38:02.000000000 -0300 @@ -822,7 +822,7 @@ case CXT5045_FUJITSU: spec->input_mux = &cxt5045_capture_source; spec->num_init_verbs = 2;

•           spec->init_verbs[1] = cxt5045_mic_sense_init_verbs;
  

+// spec->init_verbs[1] = cxt5045_mic_sense_init_verbs; spec->mixers[0] = cxt5045_mixers; codec->patch_ops.init = cxt5045_init; break;

But, I don't understand why, because this code is not executed for my chipset.

Thanks,

Rodrigo Luiz _______________________________________________ Alsa-devel mailing list Alsa-devel@alsa-project.org mailto:Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

-- Tobin Davis <tdavis@dsl-only.net mailto:tdavis@dsl-only.net>

---

Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

The other thing to check is that you are not specifying a "model=" parameter in the modprobe.conf file. The module should be allowed to autodetect your system. Your previous comment about commenting out the "spec->init_verbs[1] = cxt5045_mic_sense_init_verbs;" line indicates that your system is using the "model=fujitsu" option for some reason.

The diff below was actually a test for a user, and is not intended to be added to the main driver, sorry.

Tobin

On Tue, 2007-03-27 at 14:33 -0300, Rodrigo Luiz wrote:

This file in your home page is more recent in that -mm tree.

However, the changes is just for cxt5047 chipsets.

But I tried to load the hda-intel with the last one patch_conexant.c and I have the same issue.

Rodrigo.

The diff:

--- sound/pci/hda/patch_conexant.c 2007-03-27 13:21:30.000000000 -0300 +++ patch_conexant.c 2007-03-27 14:10:08.000000000 -0300 @@ -978,22 +978,10 @@ HDA_CODEC_MUTE("PCM Switch", 0x10, 0x00, HDA_OUTPUT), HDA_CODEC_VOLUME("PCM-2 Volume", 0x1c, 0x00, HDA_OUTPUT), HDA_CODEC_MUTE("PCM-2 Switch", 0x1c, 0x00, HDA_OUTPUT),

•   {
  

•           .iface = SNDRV_CTL_ELEM_IFACE_MIXER,
  

•           .name = "Master Playback Volume",
  

•           .info = snd_hda_mixer_amp_volume_info,
  

•           .get = snd_hda_mixer_amp_volume_get,
  

•           .put = cxt5047_hp_master_vol_put,
  

•           .private_value = HDA_COMPOSE_AMP_VAL(0x13, 3, 0, 
  

HDA_OUTPUT),

•   },
  

•   {
  

•           .iface = SNDRV_CTL_ELEM_IFACE_MIXER,
  

•           .name = "Master Playback Switch",
  

•           .info = cxt_eapd_info,
  

•           .get = cxt_eapd_get,
  

•           .put = cxt5047_hp_master_sw_put,
  

•           .private_value = 0x13,
  

•   },
  

•   HDA_CODEC_VOLUME("Speaker Volume", 0x1d, 0x00, HDA_OUTPUT),
  

•   HDA_CODEC_MUTE("Speaker Switch", 0x1d, 0x00, HDA_OUTPUT),
  

•   HDA_CODEC_VOLUME("Headphone Volume", 0x13, 0x00, HDA_OUTPUT),
  

•   HDA_CODEC_MUTE("Headphone Switch", 0x13, 0x00, HDA_OUTPUT),
  
     {}
  

  };

Tobin Davis wrote:

...

I'm not entirely sure that the code you have is the latest. Could you compare your patch_conexant.c with this one: http://members.dsl-only.net/~tdavis/patch_conexant.c

This is my latest version (it's also in the alsa hg tree).

Tobin

At Tue, 27 Mar 2007 15:50:09 -0300, Rodrigo Luiz wrote:

Tobin Davis wrote:

...

The other thing to check is that you are not specifying a "model=" parameter in the modprobe.conf file. The module should be allowed to autodetect your system. Your previous comment about commenting out the "spec->init_verbs[1] = cxt5045_mic_sense_init_verbs;" line indicates that your system is using the "model=fujitsu" option for some reason.

My modprobe.conf is empty and I dont' give any paramter for this module.

But I also am not understanding. My test code:

     switch (board_config) {
     case CXT5045_LAPTOP:
             codec->patch_ops.unsol_event = cxt5045_hp_unsol_event;
             spec->input_mux = &cxt5045_capture_source;
             spec->num_init_verbs = 2;
             spec->init_verbs[1] = cxt5045_hp_sense_init_verbs;
             spec->mixers[0] = cxt5045_mixers;
             codec->patch_ops.init = cxt5045_init;
             printk(KERN_DEBUG"cxt5045_laptop\n"); //for test
             break;
     case CXT5045_FUJITSU:
             spec->input_mux = &cxt5045_capture_source;
             spec->num_init_verbs = 2;

// spec->init_verbs[1] = cxt5045_mic_sense_init_verbs; spec->mixers[0] = cxt5045_mixers; codec->patch_ops.init = cxt5045_init; printk(KERN_DEBUG"cxt5045_fujitsu\n"); //for test break;

And I see in log:

Mar 27 15:36:48 dig kernel: cxt5045_laptop

But if I uncomment the line "spec->init_verbs[1] = cxt5045_mic_sense_init_verbs;", my speaker don't work and I receive these messages in log too.

Mar 27 15:44:59 dig kernel: hda_intel: azx_get_response timeout, switching to polling mode... Mar 27 15:45:00 dig kernel: hda_intel: azx_get_response timeout, switching to single_cmd mode...

It's very strange. The case CXT5045_FUJITSU isn't executed, but it is influencing in the module. It is truth. :/

It must be a coincidence. The compiler optimization stripped out the verbs because no one is referring. Try the patch below.

Takashi

diff -r 92288445bcd3 pci/hda/patch_conexant.c --- a/pci/hda/patch_conexant.c Wed Mar 28 13:42:25 2007 +0200 +++ b/pci/hda/patch_conexant.c Wed Mar 28 15:06:52 2007 +0200 @@ -632,11 +632,13 @@ static struct hda_verb cxt5045_hp_sense_ static struct hda_verb cxt5045_hp_sense_init_verbs[] = { /* pin sensing on HP jack */ {0x11, AC_VERB_SET_UNSOLICITED_ENABLE, AC_USRSP_EN | CONEXANT_HP_EVENT}, + { } /* end */ };

static struct hda_verb cxt5045_mic_sense_init_verbs[] = { /* pin sensing on HP jack */ {0x12, AC_VERB_SET_UNSOLICITED_ENABLE, AC_USRSP_EN | CONEXANT_MIC_EVENT}, + { } /* end */ };

#ifdef CONFIG_SND_DEBUG