Bug 21565

Summary:	radeon-rewrite r200: Error messages during rendering
Product:	DRI	Reporter:	Stefan Dösinger <stefandoesinger>
Component:	DRM/Radeon	Assignee:	Default DRI bug account <dri-devel>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	medium
Version:	DRI git
Hardware:	Other
OS:	All
Whiteboard:
i915 platform:		i915 features:

Description Stefan Dösinger 2009-05-05 05:57:49 UTC

Since commit f616995e5ed56745c4470b2ca5aeeb8d89a8c9db, the r200 driver writes lots of error messages during rendering. This is reproducible with glxgears:

stefan@laptop ~/src/mesa/src/mesa/drivers/dri/r200 $ glxgears
CS section size missmatch start at (r200_state_init.c,ctx_emit_cs,510) 22 vs 26
CS section end at (r200_state_init.c,ctx_emit_cs,543)
CS section size missmatch start at (r200_state_init.c,ctx_emit_cs,510) 22 vs 26
CS section end at (r200_state_init.c,ctx_emit_cs,543)
CS section size missmatch start at (r200_state_init.c,ctx_emit_cs,510) 22 vs 26
CS section end at (r200_state_init.c,ctx_emit_cs,543)

etc, etc, etc

There are no visible rendering bugs in glxgears, but in other apps many vertex program and ATI_fragment_programs do not work correctly(The commit above made the extensions show up as expected). Do these messages indicate that something deeper is wrong, or are they just spam? The failing shaders include the DolphinVS.exe dx8 sdk sample when run with Wine, the Bumpmap samples and vertex blend sample. I think the directx8 sdk is not publically available any more, but I can try to write stand alone test apps to show these bugs if needed. (Wine uses GL_ATI_fragment_shader on r200 cards for fixed function fragment processing)

The error messages do not show up in mesa master, and the extensions work correctly.

Comment 1 Roland Scheidegger 2009-05-05 08:48:18 UTC

I've pushed a (totally untested) fix for the section size mismatch (though maybe should get rid of the conditional there since it's always going to be true).
I don't think though this will fix other issues.

Comment 2 Stefan Dösinger 2009-05-05 10:11:18 UTC

Works for glxgears(no more errors), but in other apps there are still similar complains:

This is etracer(in the main menu, before the glDrawArrays crash):

$ etracer
Extreme TuxRacer SVN Development --  http://www.extremetuxracer.com
(c) 2007 The ETRacer team
(c) 2004-2005 The PPRacer team
(c) 1999-2001 Jasmin F. Patry<jfpatry@sunspirestudios.com>
ETRacer comes with ABSOLUTELY NO WARRANTY. This is free software,
and you are welcome to redistribute it under certain conditions.
See http://www.gnu.org/copyleft/gpl.html for details.

%%% etracer warning: Attempt to bind to Texture unloaded texture: `b-herring_run_icon'

CS section size missmatch start at (r200_state_init.c,tex_emit_cs,603) 15 vs 13
CS section end at (r200_state_init.c,tex_emit_cs,625)
CS section size missmatch start at (r200_state_init.c,tex_emit_cs,603) 15 vs 13
CS section end at (r200_state_init.c,tex_emit_cs,625)
CS section size missmatch start at (r200_state_init.c,tex_emit_cs,603) 15 vs 13
CS section end at (r200_state_init.c,tex_emit_cs,625)

I'll see if I can find other error locations in other apps.

Comment 3 Roland Scheidegger 2009-05-05 12:07:20 UTC

I've pushed a fix for another section size mismatch, I think that should be the last one at least in that area...

Comment 4 Stefan Dösinger 2009-05-05 14:10:18 UTC

Yep, the warnings seem to be gone, so I consider this fixed. If I run across another warning I'll reopen.

Comment 5 Stefan Dösinger 2009-05-07 11:12:18 UTC

Here's one more. It was uncovered by the proposed patch to bug 21531:

CS section size missmatch start at (r200_cmdbuf.c,r200EmitVertexAOS,249) 5 vs 7
CS section end at (r200_cmdbuf.c,r200EmitVertexAOS,254)

I guess the section size mismatch is not caused by the same bug as the NULL ptr deref the patch fixes, so I reopen this bug.

Comment 6 Jerome Glisse 2009-05-20 05:16:37 UTC

Fixed with 8308bf9ee155b405ad42e6621daf33a108330418, closing the bug.

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.