Bug 13327

Summary: Xorg crashes on attempting to run fragment shader
Product: Mesa Reporter: Ben Gamari <bgamari>
Component: Drivers/DRI/i915Assignee: Ian Romanick <idr>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: medium CC: dri-devel
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:
Attachments: Xorg.log of a crashed session after trying water plugin
A GDB backtrace of the crash
GDB Backtrace and texImage dump
try to avoid dereferencing null image->Data pointer
GDB backtrace from patched mesa

Description Ben Gamari 2007-11-20 16:23:02 UTC
When one attempts to use compiz's "water" plugin, Xorg crashes. Log attached.
Comment 1 Ben Gamari 2007-11-20 16:23:35 UTC
Created attachment 12659 [details]
Xorg.log of a crashed session after trying water plugin
Comment 2 Brian Paul 2007-11-23 07:16:10 UTC
If you could recompile with -g to get more info from the stack trace, that'd help.  
Maybe you could also use gdb to inspect suspected variables, etc.  I don't have time to try to repoduce this myself.
Comment 3 Ben Gamari 2007-12-04 11:20:23 UTC
Created attachment 12935 [details]
A GDB backtrace of the crash

Here's a gdb backtrace
Comment 4 Brian Paul 2007-12-05 17:05:17 UTC
At the crash point, could you "print *texImage"?
Comment 5 Ben Gamari 2007-12-06 11:55:11 UTC
Created attachment 12979 [details]
GDB Backtrace and texImage dump

Here's a dump of texImage and another backtrace (just in case)
Comment 6 Brian Paul 2007-12-06 15:34:38 UTC
Created attachment 12985 [details]
try to avoid dereferencing null image->Data pointer

Here's a work-around.  Let's see what happens.

Apparently, we're try to texture without having a texture image defined.
Comment 7 Ben Gamari 2007-12-06 19:01:40 UTC
Created attachment 12986 [details]
GDB backtrace from patched mesa

Tried the workaround. No dice. Looks like it just fails later on down the line, although this time it's not nearly as straightforward of a fix. SIGSEGV in memset but none of the parameters are obviously wrong as far as I can tell (no NULL pointers at least).
Comment 8 Ben Gamari 2009-02-05 19:20:11 UTC
This seems to be gone on master. Marking as FIXED.

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.