[alsa-devel] snd-hda-intel: Is there a way to specify a quirk for the default bdl_pos_adj value for the given chipset (PCI ID)?

[Aleksander Adamowski]

Hi!

I'm working on a patch against GIT for snd-hda-intel that would let my
card's capture work properly without losing ability to capture audio (that's
an old story available in this thread:
http://thread.gmane.org/gmane.linux.alsa.devel/66775).

I know how to make position_fix=1 permanent for the given PCI ID (the
snd_pci_quirk position_fix_list struct array in hda_intel.c), but there
seems to be no way of making a similar quirk for bdl_pos_adj - am I looking
in the wrong place?

My work on a patch goes in parallel with determining the minimal set of
modifications that will solve the problem.

Currently, I've determined that the combination of 3 module options
described in this forum post solves my problem:
http://www.linuxhardware.org/forums/viewtopic.php?p=66&sid=f6e8953338fd94c7c44a9d4955849e3f#66

I'm working on isolating the minimum set of these that solves the problem
(which requires a couple of hours to be reproduce, so that takes time).

I've observed that neither position_fix=1, nor bdl_pos_adj=0 solve the
problem by themselves in isolation.

I'm unsure whether model=3stack-6ch is necessary at all,
because patch_realtek.c already has the model mapping for my combination of
PCI IDs and it maps to a different model:
SND_PCI_QUIRK(0x1458, 0xa002, "Gigabyte P35 DS3R", ALC882_6ST_DIG),

Once the minimization is accomplished, I want to prepare a patch against GIT
(git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6.git is the
repo to work against, as far as I understand).

But since it seems that bdl_pos_adj=0 is required for proper operation, I
need a method to map the default for that option in the sources, if there is
such a method.

I'd be grateful for any pointers.

-- 
Best Regards,
 Aleksander Adamowski
 http://olo.org.pl