[alsa-devel] imac27 12,2 (2011) model for patch_cirrus.c

[Takashi Iwai]

At Mon, 09 Jan 2012 15:26:07 +0100,
Jérémy Lal wrote:
> 
> On 09/01/2012 14:58, Takashi Iwai wrote:
> > At Mon, 09 Jan 2012 14:53:10 +0100,
> > Jérémy Lal wrote:
> >>
> >> On 09/01/2012 11:34, Takashi Iwai wrote:
> >>> At Thu, 29 Dec 2011 13:00:01 +0100,
> >>> Jérémy Lal wrote:
> >>>>
> >>>> Hi,
> >>>> it'd be glad to have patch_cirrus.c recognize model=imac27_122,
> >>>> (i'm adding a model because it's harder to auto-detect it),
> >>>> please find attached patch, tested on linux 3.2-rc7.
> >>>> I don't know anything about pin configs, just that it works with
> >>>> the ones in the patch.
> >>>> Note that i tried with model=auto (no sound) and model=imac27 (front speakers
> >>>> not muted when headphones are plugged in, and surround speakers not properly
> >>>> detected).
> >>>
> >>> Thanks for the patch.  But I see a problem in the code, i.e.
> >>>
> >>>> +static const struct cs_pincfg imac27_122_pincfgs[] = {
> >>>> +	{ 0x00, 0x821c9700 },
> >>>> +	{ 0x01, 0x011d9700 },
> >>> (snip)
> >>>> +	{ 0x27, 0x401f5701 },
> >>>> +	{} /* terminator */
> >>>> +};
> >>>
> >>> This looks obviously wrong.  The pincfg table should contain only the
> >>> default pin-configuration values for pin widgets while you are setting
> >>> some values to all widgets.  Please minimize this entry.
> >>
> >> It's working without overriding pin-configurations :)
> >> I tested almost everything except digital input/output.
> >>
> >> A kernel with the following patch and
> >> options snd-hda-intel model=imac27_122
> >> is enough to get sound working.
> > 
> > OK, that's much better.
> > Could you give the patch with a proper changelog and your sign-off
> > so that I can merge it now?
> 
> ALSA: hda/cirrus - support for iMac12,1 and 12,2 models
> 
> Those models just need to have headphones on GPI02
> instead of GPI01, and use BIOS pincfgs.
> 
> Signed-off-by: Jérémy Lal <kapouer at melix.org>
> ---
> 
> --- linux-source-3.2-rc7.orig/sound/pci/hda/patch_cirrus.c	2011-12-24 06:51:06.000000000 +0100
> +++ linux-source-3.2-rc7/sound/pci/hda/patch_cirrus.c	2012-01-09 12:40:53.026430475 +0100
> @@ -78,6 +78,7 @@
>  	CS420X_MBP53,
>  	CS420X_MBP55,
>  	CS420X_IMAC27,
> +	CS420X_IMAC27_122,
>  	CS420X_APPLE,
>  	CS420X_AUTO,
>  	CS420X_MODELS
> @@ -1278,6 +1279,7 @@
>  	[CS420X_MBP53] = "mbp53",
>  	[CS420X_MBP55] = "mbp55",
>  	[CS420X_IMAC27] = "imac27",
> +	[CS420X_IMAC27_122] = "imac27_122",
>  	[CS420X_APPLE] = "apple",
>  	[CS420X_AUTO] = "auto",
>  };
> @@ -1392,6 +1394,12 @@
>  		spec->gpio_eapd_speaker = 8; /* GPIO3 = speakers */
>  		spec->gpio_mask = spec->gpio_dir =
>  			spec->gpio_eapd_hp | spec->gpio_eapd_speaker;
> +		break;
> +	case CS420X_IMAC27_122:
> +		spec->gpio_eapd_hp = 4; /* GPIO2 = headphones */
> +		spec->gpio_eapd_speaker = 8; /* GPIO3 = speakers */
> +		spec->gpio_mask = spec->gpio_dir =
> +			spec->gpio_eapd_hp | spec->gpio_eapd_speaker;
>  		break;
>  	}
>  
> 
>  
> > Also, I guess your device should be identified via PCI SSID.
> > Could you check whether the device has a unique PCI SSID (check lspci
> > output), and add the corresponding entry to cs420x_cfg_tbl[]?  If it's
> > about 10de:xxxx, it should be unique.  If it's 8086:xxxx, then we'd
> > need to check the codec SSID instead of PCI SSID.
> 
> The PCI SSID is 8086:1c20
> How can i get the codec SSID ?

Check /proc/asound/card0/codec#* file.  This id should be added to
  cs420x_codec_cfg_tbl[], but before the entry
  SND_PCI_QUIRK_VENDOR(0x106b, "Apple", CS420X_APPLE)
which matches all 106b:*.


Takashi