On Tue, 9 Aug 2005, Benjamin Herrenschmidt wrote:
I can provide almost all the necessary informations. I have specs for most chips, and those I don't have spec for have Darwin drivers that should tell us all we need to know. There are several issues though, as I wrote earlier. More than just "fixups", we need some real re-architecture of the driver. We need to properly separate the transport layer (can be davbus or i2s) from the codecs, deal with multiple codecs (G5s have that typically) and multiple transport busses (some machines have codecs on the 2 i2s interfaces), we need to properly reprogram the i2s clocks, we need to do clock switching when fed with a digital input signal, etc... Also, on pretty much all recent machines, the only info that you have about the sound hardware is an "ID" value in the device-tree. Apple code matches that value with some XML structure in the driver property list that indicates what set of codecs etc... to use. I don't have time to actually do the work, but I'm willing to help.Adding the above info as a comment to the ALSA BTS bug I opened would already be a tremendous start. :)Please do it for me, I can't be bothered opening yet another account to some random BTS ...
Given how you already have a clear strategy in mind for the rearchitecturing of the ALSA PowerMac subsystem, it would probably be a good idea for you to come and participate in the thread on the ALSA BTS for this bug, so you can guide the ALSA team and explain to them exactly what you have in mind.
NOTE: I sympatize with the account creation complaint. I in fact pointed out to the ALSA people that they are not likely to get PowerMac users to report every broken ALSA item they encounter, because of the account creation obligation. I also suggested that they adopt an e-mail filtering based aproach similar to the Debian BTS, instead, as a way of making it easier for people to report bugs.
-- Martin-Eric Racine http://q-funk.iki.fi