Re: Status updates for feature freeze

On Sat, Jul 31, 2004 at 01:36:55AM -0700, Eric Anholt wrote:
> On Fri, 2004-07-30 at 21:31, Kevin E Martin wrote:
> > On Fri, Jul 30, 2004 at 07:48:47PM -0400, Kevin E Martin wrote:
> > > Below I've listed my current understanding of where we are today.  I
> > > would like to get confirmation and/or updates on the status of each to
> > > make sure that they are still on schedule.
> > > 
> > > 1. Composite: The XAA wrappers plus the Composite extension will be
> > >    integrated by Eric Anholt today, and Deron Johnson will review the
> > >    code for any bug fixes or missing pieces later.  Testing can begin
> > >    this weekend.
> > > 
> > >    After talking with Keith Packard earlier today, he said that he would
> > >    work toward exposing the extra visuals from the extension instead of
> > >    globally to avoid the problems with Mozilla/Flash, etc.  There is
> > >    also a hack that could be put into place in Xlib to work around this
> > >    problem, but that is not the desired solution.
> 
> We've got a composite wrapper in HEAD now.  Compiling of Composite on
> the server is disabled by default for the moment because the build
> breaks with Xprt, and I haven't figured out why yet.

Okay.  I can look into that tomorrow as well.  I'm just finishing off
the xorg.conf changes now, which will be ready for Composite when it is
enabled.

> I'm going to stall on fixing the composite wrapper for a little while to
> work on other issues.  I'd like to pull the new trapezoids in still.  I
> also need to get my XAA diff in,

Okay.  The new traps look to be a relatively straighforward merge.

>                                  but I'm getting a server crash on my
> Radeon (but not a SiS) even without the patch, so I'm not sure if I
> should go mucking about in there while things are broken (assuming a
> commit to ati was originally at fault).

I'm seeing a crash on server exit, which I've traced to a problem with
the ordering of CloseScreen unwrapping: RADEONDisplayPowerManagementSet
(called from DPMSClose) is being called after RADEONCloseScreen.  I
haven't tracked it down further than that though.  Not sure if this is
what you're seeing in your crash.

>                                          What are thoughts on the XAA
> hook?

I haven't had a chance to take a close look yet and probably won't have
time until tomorrow.