ALSA: HDA: Virtual master gets wrong dB values on some Realtek codecs

[David Henningsson]

Please do not triage this bug, it's being used for tracking purposes.

What's happening here is that the wrong node is being used as a template when the vmaster ("Master" alsa mixer control) is created. This patch makes the parser use the mixer nid that the auto-parser has detected instead.

[David Henningsson (diwic)]

With the auto-parser we can choose the dac nid for vmaster from
the DACs we already know, instead of hard-coding it. This is more
future-proof and was actually wrong on one machine.

Signed-off-by: David Henningsson <email address hidden>
Signed-off-by: Takashi Iwai <email address hidden>
(backported from commit fde48a1f808e2bb6aaad5709d2470d814a157c86
Conflicts:
 sound/pci/hda/patch_realtek.c)

BugLink: https://bugs.launchpad.net/bugs/974090
---

After a quick chat with apw, I hope this patch can be committed. Feel free to
grab me on IRC if you have more questions - I know I'm late, sorry for that.
I wrote the patch a while back, but it did not make it into 3.2 (it is
in 3.3).

It fixes a volume slider problem for a machine we want to certify.
It has been tested on the actual hardware, and I've done a regression test
on one Realtek machine I have here.
Should you wish to do more regression tests, there is a ready-made DKMS package
here:
http://people.canonical.com/~diwic/temp/alsa-hda-dkms-vmaster-realtek_0.1_all.deb

 sound/pci/hda/patch_realtek.c | 42 ++++++++++++++++++----------------------
 1 files changed, 19 insertions(+), 23 deletions(-)

diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c
index 13fa5f2..91e767a 100644
--- a/sound/pci/hda/patch_realtek.c
+++ b/sound/pci/hda/patch_realtek.c
@@ -2937,6 +2937,8 @@ static int alc_auto_fill_extra_dacs(struct hda_codec *codec, int num_outs,

 static int alc_auto_fill_multi_ios(struct hda_codec *codec,
        unsigned int location);
+static hda_nid_t alc_look_for_out_vol_nid(struct hda_codec *codec,
+ hda_nid_t pin, hda_nid_t dac);

 /* fill in the dac_nids table from the parsed pin configuration */
 static int alc_auto_fill_dac_nids(struct hda_codec *codec)
@@ -3037,6 +3039,11 @@ static int alc_auto_fill_dac_nids(struct hda_codec *codec)
   }
  }

+ if (cfg->line_out_pins[0])
+ spec->vmaster_nid =
+ alc_look_for_out_vol_nid(codec, cfg->line_out_pins[0],
+ spec->multiout.dac_nids[0]);
+
  return 0;
 }

@@ -4000,8 +4007,10 @@ static int patch_alc880(struct hda_codec *codec)
 #endif
  }

- if (board_config != ALC_MODEL_AUTO)
+ if (board_config != ALC_MODEL_AUTO) {
+ spec->vmaster_nid = 0x0c;
   setup_preset(codec, &alc880_presets[board_config]);
+ }

  if (!spec->no_analog && !spec->adc_nids) {
   alc_auto_fill_adc_caps(codec);
@@ -4019,8 +4028,6 @@ static int patch_alc880(struct hda_codec *codec)
   set_beep_amp(spec, 0x0b, 0x05, HDA_INPUT);
  }

- spec->vmaster_nid = 0x0c;
-
  codec->patch_ops = alc_patch_ops;
  if (board_config == ALC_MODEL_AUTO)
   spec->init_hook = alc_auto_init_std;
@@ -4129,8 +4136,10 @@ static int patch_alc260(struct hda_codec *codec)
 #endif
  }

- if (board_config != ALC_MODEL_AUTO)
+ if (board_config != ALC_MODEL_AUTO) {
   setup_preset(codec, &alc260_presets[board_config]);
+ spec->vmaster_nid = 0x08;
+ }

  if (!spec->no_analog && !spec->adc_nids) {
   alc_auto_fill_adc_caps(codec);
@@ -4150,8 +4159,6 @@ static int patch_alc260(struct hda_codec *codec)

  alc_apply_fixup(codec, ALC_FIXUP_ACT_PROBE);

- spec->vmaster_nid = 0x08;
-
  codec->patch_ops = alc_patch_ops;
  if (board_config == ALC_MODEL_AUTO)
   ...

[Leann Ogasawara (leannogasawara)]

On Thu, 2012-04-05 at 13:56 +0200, David Henningsson wrote:
> With the auto-parser we can choose the dac nid for vmaster from
> the DACs we already know, instead of hard-coding it. This is more
> future-proof and was actually wrong on one machine.
>
> Signed-off-by: David Henningsson <email address hidden>
> Signed-off-by: Takashi Iwai <email address hidden>
> (backported from commit fde48a1f808e2bb6aaad5709d2470d814a157c86
> Conflicts:
> sound/pci/hda/patch_realtek.c)
>
> BugLink: https://bugs.launchpad.net/bugs/974090

Patch is upstream and tested.

Acked-by: Leann Ogasawara <email address hidden>

> ---
>
> After a quick chat with apw, I hope this patch can be committed. Feel free to
> grab me on IRC if you have more questions - I know I'm late, sorry for that.
> I wrote the patch a while back, but it did not make it into 3.2 (it is
> in 3.3).
>
> It fixes a volume slider problem for a machine we want to certify.
> It has been tested on the actual hardware, and I've done a regression test
> on one Realtek machine I have here.
> Should you wish to do more regression tests, there is a ready-made DKMS package
> here:
> http://people.canonical.com/~diwic/temp/alsa-hda-dkms-vmaster-realtek_0.1_all.deb
>
> sound/pci/hda/patch_realtek.c | 42 ++++++++++++++++++----------------------
> 1 files changed, 19 insertions(+), 23 deletions(-)
>
> diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c
> index 13fa5f2..91e767a 100644
> --- a/sound/pci/hda/patch_realtek.c
> +++ b/sound/pci/hda/patch_realtek.c
> @@ -2937,6 +2937,8 @@ static int alc_auto_fill_extra_dacs(struct hda_codec *codec, int num_outs,
>
> static int alc_auto_fill_multi_ios(struct hda_codec *codec,
> unsigned int location);
> +static hda_nid_t alc_look_for_out_vol_nid(struct hda_codec *codec,
> + hda_nid_t pin, hda_nid_t dac);
>
> /* fill in the dac_nids table from the parsed pin configuration */
> static int alc_auto_fill_dac_nids(struct hda_codec *codec)
> @@ -3037,6 +3039,11 @@ static int alc_auto_fill_dac_nids(struct hda_codec *codec)
> }
> }
>
> + if (cfg->line_out_pins[0])
> + spec->vmaster_nid =
> + alc_look_for_out_vol_nid(codec, cfg->line_out_pins[0],
> + spec->multiout.dac_nids[0]);
> +
> return 0;
> }
>
> @@ -4000,8 +4007,10 @@ static int patch_alc880(struct hda_codec *codec)
> #endif
> }
>
> - if (board_config != ALC_MODEL_AUTO)
> + if (board_config != ALC_MODEL_AUTO) {
> + spec->vmaster_nid = 0x0c;
> setup_preset(codec, &alc880_presets[board_config]);
> + }
>
> if (!spec->no_analog && !spec->adc_nids) {
> alc_auto_fill_adc_caps(codec);
> @@ -4019,8 +4028,6 @@ static int patch_alc880(struct hda_codec *codec)
> set_beep_amp(spec, 0x0b, 0x05, HDA_INPUT);
> }
>
> - spec->vmaster_nid = 0x0c;
> -
> codec->patch_ops = alc_patch_ops;
> if (board_config == ALC_MODEL_AUTO)
> spec->init_hook = alc_auto_init_std;
> @@ -4129,8 +4136,10 @@ static int patch_alc260(struct hda_codec *codec)
> #endif
> }
>
> - if (board_config != ALC_MODEL_AUTO)
> + if (board_config != ALC_MODEL_AUTO) {
> setup_preset(codec, &alc260_presets[board_config]);
> + spec->vmaster_nid = 0x08;...

[Seth Forshee (sforshee)]

On Thu, Apr 05, 2012 at 01:56:28PM +0200, David Henningsson wrote:
> With the auto-parser we can choose the dac nid for vmaster from
> the DACs we already know, instead of hard-coding it. This is more
> future-proof and was actually wrong on one machine.
>
> Signed-off-by: David Henningsson <email address hidden>
> Signed-off-by: Takashi Iwai <email address hidden>
> (backported from commit fde48a1f808e2bb6aaad5709d2470d814a157c86
> Conflicts:
> sound/pci/hda/patch_realtek.c)
>
> BugLink: https://bugs.launchpad.net/bugs/974090
> ---
>
> After a quick chat with apw, I hope this patch can be committed. Feel free to
> grab me on IRC if you have more questions - I know I'm late, sorry for that.
> I wrote the patch a while back, but it did not make it into 3.2 (it is
> in 3.3).
>
> It fixes a volume slider problem for a machine we want to certify.
> It has been tested on the actual hardware, and I've done a regression test
> on one Realtek machine I have here.
> Should you wish to do more regression tests, there is a ready-made DKMS package
> here:
> http://people.canonical.com/~diwic/temp/alsa-hda-dkms-vmaster-realtek_0.1_all.deb
>
> sound/pci/hda/patch_realtek.c | 42 ++++++++++++++++++----------------------
> 1 files changed, 19 insertions(+), 23 deletions(-)
>
> diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c
> index 13fa5f2..91e767a 100644
> --- a/sound/pci/hda/patch_realtek.c
> +++ b/sound/pci/hda/patch_realtek.c
> @@ -2937,6 +2937,8 @@ static int alc_auto_fill_extra_dacs(struct hda_codec *codec, int num_outs,
>
> static int alc_auto_fill_multi_ios(struct hda_codec *codec,
> unsigned int location);
> +static hda_nid_t alc_look_for_out_vol_nid(struct hda_codec *codec,
> + hda_nid_t pin, hda_nid_t dac);
>
> /* fill in the dac_nids table from the parsed pin configuration */
> static int alc_auto_fill_dac_nids(struct hda_codec *codec)
> @@ -3037,6 +3039,11 @@ static int alc_auto_fill_dac_nids(struct hda_codec *codec)
> }
> }
>
> + if (cfg->line_out_pins[0])
> + spec->vmaster_nid =
> + alc_look_for_out_vol_nid(codec, cfg->line_out_pins[0],
> + spec->multiout.dac_nids[0]);
> +
> return 0;
> }
>
> @@ -4000,8 +4007,10 @@ static int patch_alc880(struct hda_codec *codec)
> #endif
> }
>
> - if (board_config != ALC_MODEL_AUTO)
> + if (board_config != ALC_MODEL_AUTO) {
> + spec->vmaster_nid = 0x0c;
> setup_preset(codec, &alc880_presets[board_config]);
> + }
>
> if (!spec->no_analog && !spec->adc_nids) {
> alc_auto_fill_adc_caps(codec);
> @@ -4019,8 +4028,6 @@ static int patch_alc880(struct hda_codec *codec)
> set_beep_amp(spec, 0x0b, 0x05, HDA_INPUT);
> }
>
> - spec->vmaster_nid = 0x0c;
> -
> codec->patch_ops = alc_patch_ops;
> if (board_config == ALC_MODEL_AUTO)
> spec->init_hook = alc_auto_init_std;
> @@ -4129,8 +4136,10 @@ static int patch_alc260(struct hda_codec *codec)
> #endif
> }
>
> - if (board_config != ALC_MODEL_AUTO)
> + if (board_config != ALC_MODEL_AUTO) {
> setup_preset(codec, &alc260_presets[board_config]);
> + spec->vmaster_nid = 0x08;
> + }
>
> if (!spec->no_analog && !spec->adc_nids) {
> alc_auto_fill_adc_cap...

[Leann Ogasawara (leannogasawara)]

Applied to Precise master-next.

Thanks,
Leann

On Thu, 2012-04-05 at 13:56 +0200, David Henningsson wrote:
> With the auto-parser we can choose the dac nid for vmaster from
> the DACs we already know, instead of hard-coding it. This is more
> future-proof and was actually wrong on one machine.
>
> Signed-off-by: David Henningsson <email address hidden>
> Signed-off-by: Takashi Iwai <email address hidden>
> (backported from commit fde48a1f808e2bb6aaad5709d2470d814a157c86
> Conflicts:
> sound/pci/hda/patch_realtek.c)
>
> BugLink: https://bugs.launchpad.net/bugs/974090
> ---
>
> After a quick chat with apw, I hope this patch can be committed. Feel free to
> grab me on IRC if you have more questions - I know I'm late, sorry for that.
> I wrote the patch a while back, but it did not make it into 3.2 (it is
> in 3.3).
>
> It fixes a volume slider problem for a machine we want to certify.
> It has been tested on the actual hardware, and I've done a regression test
> on one Realtek machine I have here.
> Should you wish to do more regression tests, there is a ready-made DKMS package
> here:
> http://people.canonical.com/~diwic/temp/alsa-hda-dkms-vmaster-realtek_0.1_all.deb
>
> sound/pci/hda/patch_realtek.c | 42 ++++++++++++++++++----------------------
> 1 files changed, 19 insertions(+), 23 deletions(-)
>
> diff --git a/sound/pci/hda/patch_realtek.c b/sound/pci/hda/patch_realtek.c
> index 13fa5f2..91e767a 100644
> --- a/sound/pci/hda/patch_realtek.c
> +++ b/sound/pci/hda/patch_realtek.c
> @@ -2937,6 +2937,8 @@ static int alc_auto_fill_extra_dacs(struct hda_codec *codec, int num_outs,
>
> static int alc_auto_fill_multi_ios(struct hda_codec *codec,
> unsigned int location);
> +static hda_nid_t alc_look_for_out_vol_nid(struct hda_codec *codec,
> + hda_nid_t pin, hda_nid_t dac);
>
> /* fill in the dac_nids table from the parsed pin configuration */
> static int alc_auto_fill_dac_nids(struct hda_codec *codec)
> @@ -3037,6 +3039,11 @@ static int alc_auto_fill_dac_nids(struct hda_codec *codec)
> }
> }
>
> + if (cfg->line_out_pins[0])
> + spec->vmaster_nid =
> + alc_look_for_out_vol_nid(codec, cfg->line_out_pins[0],
> + spec->multiout.dac_nids[0]);
> +
> return 0;
> }
>
> @@ -4000,8 +4007,10 @@ static int patch_alc880(struct hda_codec *codec)
> #endif
> }
>
> - if (board_config != ALC_MODEL_AUTO)
> + if (board_config != ALC_MODEL_AUTO) {
> + spec->vmaster_nid = 0x0c;
> setup_preset(codec, &alc880_presets[board_config]);
> + }
>
> if (!spec->no_analog && !spec->adc_nids) {
> alc_auto_fill_adc_caps(codec);
> @@ -4019,8 +4028,6 @@ static int patch_alc880(struct hda_codec *codec)
> set_beep_amp(spec, 0x0b, 0x05, HDA_INPUT);
> }
>
> - spec->vmaster_nid = 0x0c;
> -
> codec->patch_ops = alc_patch_ops;
> if (board_config == ALC_MODEL_AUTO)
> spec->init_hook = alc_auto_init_std;
> @@ -4129,8 +4136,10 @@ static int patch_alc260(struct hda_codec *codec)
> #endif
> }
>
> - if (board_config != ALC_MODEL_AUTO)
> + if (board_config != ALC_MODEL_AUTO) {
> setup_preset(codec, &alc260_presets[board_config]);
> + spec->vmaster_nid = 0x08;
> + }
>
> if (!spec->no_analog && !spe...

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 3.2.0-23.36

---------------
linux (3.2.0-23.36) precise-proposed; urgency=low

  [ Andy Whitcroft ]

  * tools/hv: add basic Makefile
    - LP: #977246
  * tools/hv: add basic manual pages
    - LP: #977246
  * Hyper-V KVP daemon -- add to the tools package
    - LP: #977246

  [ David Henningsson ]

  * SAUCE: (drop after 3.3) ALSA: HDA: Remove quirk for Gigabyte
    GA-H61M-S2PV
    - LP: #948360

  [ Leann Ogasawara ]

  * d-i: Add dm-mirror and dm-raid to md-modules
    - LP: #919281

  [ Tim Gardner ]

  * [Config] Fix invalid linux-headers link
    - LP: #974403
  * Remove headers asm symlink entirely
    - LP: #974403

  [ Upstream Kernel Changes ]

  * Revert "x86/ioapic: Add register level checks to detect bogus io-apic
    entries"
    - LP: #974982
  * ALSA: HDA: Realtek: Take vmaster dac from multiout dac list
    - LP: #974090
  * kexec: remove KMSG_DUMP_KEXEC
    - LP: #944772
  * kmsg_dump: don't run on non-error paths by default
    - LP: #944772
  * pstore: Introduce get_reason_str() to pstore
    - LP: #944772
 -- Leann Ogasawara <email address hidden> Tue, 10 Apr 2012 08:16:09 -0700

[Aleksejs (zer0byte)]

They kill with wars, alcohol and abortions. Save us!!!