On Fri, 24 Apr 2020 17:44:27 +0200, Pierre-Louis Bossart wrote: > > > > >> Hmm, I wonder is this now doing more harm than good. Based on browsing > >> through the related code in hda-codec.c and friends, it would seem > >> "sofbus" as the default is mostly harmless, but I could have missed > >> something. > > > > That's currently harmless since no codec driver defines "sofbus" > > model, hence the HDA parser continues to match with the default > > quirks. OTOH, the fixed "sofbus" model is fairly useless. So, feel > > free to take my ack, too: > > For my education, are you saying that the default should be that the > modelname is NULL, and the hda auto parser will use known quirks based > on PCI/SSID information, and when the user sets the model name to a > non-NULL string it will force a specific quirk to be used? Yes. If the given string matches with the pre-defined table, the quirk entry is used and applied. If no string is given or it doesn't match, it continues to the fallback quirk, that is, matching with PCI SSID, then codec SSID. Takashi
