[prev in list] [next in list] [prev in thread] [next in thread] 

List:       freedesktop-xorg
Subject:    Re: OO.o render crash (Was: Moving rendercheck to git)
From:       Eric Anholt <eric () anholt ! net>
Date:       2006-04-28 3:31:40
Message-ID: 1146195100.65330.9.camel () leguin
[Download RAW message or body]


On Wed, 2006-04-19 at 13:43 +0200, Matthias Hopf wrote:
> On Apr 18, 06 13:59:53 -0700, Eric Anholt wrote:
> > I've migrated rendercheck to git, since I'm working on it again to come
> > up with some new basic-functionality tests.  If you've got outstanding
> > changes and don't want to deal with git, I'll handle pushing things over
> > for the next couple of days.  But since lars was the only one other than
> > me to have ever committed to it, I'm not too worried.
> 
> Eric, we're seeing massive render problems with openoffice.org (see
> https://bugzilla.novell.com/show_bug.cgi?id=152730 ) which I tracked
> down quite a bit. It seems that rendering to hardware surfaces
> (pDrawable->x/y != 0) is somewhat broken.
> I don't have an easy test case for that, so this could be an idea for
> additional tests.

Sorry for being slow in responding to this.  That bug is also quite
something to try and read :/

As far as I understand, XAA starts pixmaps out in the offscreen cache
(pDrawable->x/y != 0), and they get kicked out later.  If so,
rendercheck should be testing non-zero x/y in pixmaps pretty well.

Perhaps a better way would be trying to figure out from the debugger
what exactly that failing Render operation is, so it could be replicated
outside of OO.o.  Format of source, mask, and dest, and the
transform/repeat settings on them would be important.

-- 
Eric Anholt                     anholt@FreeBSD.org
eric@anholt.net                 eric.anholt@intel.com

["signature.asc" (application/pgp-signature)]

[prev in list] [next in list] [prev in thread] [next in thread]