Attention is currently required from: Nicholas Sudsgaard, Paul Menzel.
Nico Huber has posted comments on this change by Nicholas Sudsgaard. ( https://review.coreboot.org/c/coreboot/+/83504?usp=email )
Change subject: southbridge/intel: Use azalia_audio_init() instead of duplicating code ......................................................................
Patch Set 3: Code-Review+1
(1 comment)
Commit Message:
https://review.coreboot.org/c/coreboot/+/83504/comment/87da0cf1_6e6718c3?usp... : PS3, Line 12: The southbridge's codect_detect() was written a long time ago and : mentions that it was "not working yet" in the commit message. Therefore, : I believe that the differences was due to it being a rough : implementation and did not follow the specification exactly in some : places (e.g. only using 4 bits for codec_mask instead of 15 bits). This awfully sounds like you are making excuses for changing the code. That shouldn't be necessary. The code may have been added with these words, but it's running on some of the most prominent coreboot platforms, hence I'd expect we'd know by now if it wasn't working. Also, what commit message?
Reading it, I see nothing too rough. The generic code, OTOH, looks wrong and overengineered. The additional wait loop seems to go back to commit be61a173. Probably was just mimicking the other loop. Both implementations seem to be missing all the delays around reset (de-)assertion btw. That might actually be what the mentioned BKDG had in mind...
The effective codec mask is given in each chip's datasheet btw.