Bug 4615

Summary:	SiS driver broken by renderbuffer changes
Product:	Mesa	Reporter:	Eric Anholt <eric>
Component:	Drivers/DRI/SiS	Assignee:	Eric Anholt <eric>
Status:	RESOLVED FIXED	QA Contact:
Severity:	critical
Priority:	high	CC:	alexdeucher
Version:	git
Hardware:	Other
OS:	FreeBSD
Whiteboard:
i915 platform:		i915 features:
Attachments:	sis-renderbuffer.diff

Description Eric Anholt 2005-09-27 23:28:38 UTC

The SiS DRI driver now either crashes on startup or renders terribly.  It looks
like it wasn't converted well for the renderbuffer changes.  It allocates
back/depth buffers per context, like VIA does (and unlike all the other drivers
we have).

Comment 1 Alex Deucher 2005-09-28 06:01:47 UTC

Savage seems to have the same problems.

Comment 2 Brian Paul 2005-09-28 07:13:20 UTC

Please provide a stack trace from the problem when running a simple Mesa demo.

Comment 3 Eric Anholt 2005-10-15 20:25:13 UTC

Created attachment 3560 [details] [review]
sis-renderbuffer.diff

Untested patch to possibly unbreak sis post-renderbuffer (the testbox's network
card wandered off and should return in the next couple of days).  Notably,
previously it would (at least) result in all span rendering going to the front
buffer, as all the renderbuffers pointed to NULL and the hardware offsets were
set to 0.

Comment 4 Brian Paul 2005-10-18 07:57:39 UTC

The savage driver should be OK now (see bug 4590).  So, is it just the sis
driver that needs fixing?  Can someone please test Eric's patch ASAP?  We need
to get 6.4 wrapped up soon.

Comment 5 Eric Anholt 2005-10-18 12:45:36 UTC

Looks like the immediate problem I was having was elsewhere (dispatch changes
from idr), though the renderbuffer changes are still necessary afaik.  I'm doing
a little more testing before committing, which I'll insta-merge to 6_4
(performance with mesa test apps is shot -- trying to figure out if something's
still wrong with updates for mesa/dri changes, or if it's a local issue).

Comment 6 Eric Anholt 2005-10-18 20:54:56 UTC

Fixed in CVS, including 6.4 branch.  Furthermore, trivially tested 6.4-branch on
r200, r100, r128, mach64 (fixed), 3dfx, mga.  Didn't test on savage (oops,
haven't ported it yet).

Comment 7 Adam Jackson 2009-08-24 12:23:27 UTC

Mass version move, cvs -> git

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.