Dear Blair, dear folks, On Wed, Jan 16, 2008 at 04:32:47PM -0500, Blair Sadewitz wrote: > Porting the whole thing shouldn't be too difficult, as there is a > portability layer. IIRC, there are calls to the dreaded vtophys() in > there somewhere, so that probably would have to be addressed. I major > showstopper for me, though, is that I have no idea how to teach it to > work with our new in-kernel linker. do you really want to load this foreign code as kernel modules? I think it would be better/easier to learn from the drivers and write the wrappers manually. > Another possibility that I thought about is taking the virtual mixer > code and running it in userspace under pud(4). In theory, we could > also dramatically improve our OSS compatability. Others who know more > than I do will be able to better comment on this--I have my hands full > trying to comprehend the insanity of DRM right now. :( Isn't pad a kernel based audio mixer that mixes audiostreams to the output channel? Why a userland mixer? Its bound to have serous timing and performance problems and audio is very sensitive for that. With regards, Reinoud
Attachment:
pgpF3t_KJ2mAY.pgp
Description: PGP signature