[alsa-devel] [PATCH] ASoC: snd-sof-pci - add model parameter and pass it to HDA codec driver

[Takashi Iwai]

On Tue, 03 Dec 2019 20:47:32 +0100,
Pierre-Louis Bossart wrote:
> 
> 
> >>> It may be useful to pass the specific model to the generic HDA codec
> >>> routines like the legacy HDA driver (snd-hda-intel) allows.
> >>> The model name "sofbus" is tricky anyway.
> >>
> >> Humm, I must admit I have never looked at this for the legacy driver,
> >> and I am a bit confused on what this would be used for?
> >> The legacy driver uses codec->modelname but I can't figure out this
> >> part in hda_codec.c
> >>
> >> if (codec->bus->modelname) {
> >> 	codec->modelname = kstrdup(codec->bus->modelname, GFP_KERNEL);
> >>
> >> In theory there can be multiple codecs per bus (with different SDIs)
> >> so using the bus->modelname for the codec->modelname looks odd.
> >
> > That's true.  However, basically the model name is specific to the
> > whole device, hence it's usually OK to pass to all codecs.  The
> > mismatched model name is just ignored (that's why the current code
> > with model="sofbus" works).  So you can think it a kind of quirk
> > lookup with a system name given explicitly by user.
> >
> >> Is there an example of this being used for my education?
> >
> > You can find the list in Documentation/sound/hd-audio/models.rst
> > and the usage in Documentation/sound/hd-audio/notes.rst.  The
> > documents are a bit outdated, though.
> 
> ah, ok, thanks for the pointers.
> 
> So if you have a new machine that's not explicitly handled by quirks
> you can initially force existing tricks to be used, and in a second
> step the quirk is extended to handle that machine, yes?

Yes, that's the usual step.


Takashi