On Wed, 2005-05-04 at 07:20 +0100, David Woodhouse wrote:
On Wed, 2005-05-04 at 16:04 +1000, Benjamin Herrenschmidt wrote:
> I'm not sure what you are talking about, and I find bug #152648 more
> confusing than anything else anyway.
Yeah, I couldn't make much sense of it either.
The problem with dynamic clocks is already fixed in xorg-x11 6.8.2, but
the fix seems to cause problems with DRI for some i386 users. Those DRI
problems are bug #152648.
> That patch looks totally bogus, where does it come from ?
It came from our RPM, and is the supposed 'fix' for bug #152648.
I believe it's just an attempt to revert to what was in 6.8.1.
Ok, this is all not very clear.
What is happening with X.org 6.8.2 plus all my patches ? I have a hard
time understanding anything about what's in that bug report, it's messy
and confusing, Kristian talks for example about "identifyign the
regression" but doesn't say what he found nor doesn't attach a patch
etc...
One of my patches does indeed fix a bug which could happen with radeon
mobility M7 (7500) and dynamic clocks enable (among others), and fixes a
couple of bugs with dynamic clocks disable. There are still "issues"
with the dynamic clocks code imho, but it should be stable. So please
make sure you have all the patches in and get that user to try with
DynamicClocks both enabled or disabled (though if it's not a mobility
chip, the stuff won't do anything).
If that still doesn't work, I'll try to find out something more precise
in the dynamic clocks code on these models that may not be appropriate.
Ben.