Re: PMUD

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

>
>That wasn't directed at you, as I hope you knew well enough. But we could
>need a few more eyes looking at this problem. Even though it seems to fall
>into the 'hopeless' class :-)

Well, we are close to it. Darwin source has sleep support for the
machine, so except for some binary only drivers, we have the
informations. For some reasons it doesn't work properly however.

I'm suspecting a problem with the r128 M3 chip. I've been doing some
hacking with it and discovered that the chip is actually not going to D2
(suspend) state when asked to do so. That could explain at least part of
the problem.

I've asked ATI and am waiting for an answer from them. In the meantime,
I've already begun comparing things between macos and linux (register
content mostly) but didn't find the solution yet.

>All this Pismo weirdness wasn't really known at the time the last pmud was
>released, that's how the Core99 stuff slipped in. Call it undue optimism
>if you like.