Bisect shows:91473485fcf3e2cef465784ae5581787a2a8a3b3 is the first bad commit.
commit 91473485fcf3e2cef465784ae5581787a2a8a3b3
Author:     Ian Romanick <ian.d.romanick@intel.com>
AuthorDate: Tue Aug 7 12:46:22 2012 -0700
Commit:     Ian Romanick <ian.d.romanick@intel.com>
CommitDate: Wed Aug 29 15:09:37 2012 -0700

    intel: Clean up bits of cruft in intelCreateContext

    This and the previous three commits should probably be squashed together...

    Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>

Created attachment 66421 [details] [review]
Proposed fix

Can you try this patch?

(In reply to comment #2)
> Created attachment 66421 [details] [review] [review]
> Proposed fix
> 
> Can you try this patch?

Add this patch to the latest mesa master branch(commit a96119cc8c998).It still happens.

Created attachment 66700 [details] [review]
Proposed fix, v2

Can you try this patch instead?

(In reply to comment #4)
> Created attachment 66700 [details] [review] [review]
> Proposed fix, v2
> 
> Can you try this patch instead?


Add this patch to commit d220e2de7fe7a75e70d68f2a, It still happens.

(In reply to comment #5)
> (In reply to comment #4)
> > Created attachment 66700 [details] [review] [review] [review]
> > Proposed fix, v2
> > 
> > Can you try this patch instead?
> 
> 
> Add this patch to commit d220e2de7fe7a75e70d68f2a, It still happens.

Are you sure you applied the updated patch correctly?  I don't have access to an Ironlake machine right now, but I simulated it with the patch below.  This causes creation of any core-profile or forward-compatible context to fail.  Without attachment #66700 [details] [review] this causes an assertion failure at line 798 of intel_context.c (this would be the segfault you saw if I built without debug).  With attachment #66700 [details] [review] the test passes.


diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965
index 60b6454..00c9079 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -110,7 +110,7 @@ brwCreateContext(int api,
    case API_OPENGLES2:
       break;
    case API_OPENGL_CORE: {
-#ifdef TEXTURE_FLOAT_ENABLED
+#if 0 && defined TEXTURE_FLOAT_ENABLED
       const unsigned max_version =
          (screen->gen == 6 ||
           (screen->gen == 7 && screen->kernel_has_gen7_sol_reset))

Created attachment 67741 [details] [review]
The rest of the fix

Okay... both patches are required to fix the problem.  I've tested it on my ILK system.  Can you verify it on yours and give me a Tested-by?

Add both patches 66700 and 66741 into the latest commit ff947c6d6583. This issue goes away.

Fixed by the following series on master.  These changes have also been picked to the 9.0 branch.

commit e87c63f2889fcbeb5a8bbd91eda1333d7ed44bf2
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Fri Sep 28 15:38:26 2012 -0700

    i965: brwInitVtbl needs to know the chipset generation
    
    Fixes major regressions since de958de.
    
    Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>

commit de958de71b1450952e021af4e729c87406353db6
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Wed Sep 26 17:57:01 2012 -0700

    i915: Don't free the intel_context structure when intelCreateContext fails.
    
    intelDestroyContext will eventually be called, and it will clean things up.
    
    NOTE: This is a candidate for the 9.0 branch.
    
    Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
    Reviewed-by: Eric Anholt <eric@anholt.net>
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=53618

commit 87f26214d6bdeb439b30615ec53c293c5141cf11
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Fri Aug 31 14:55:21 2012 -0700

    i965: Don't free the intel_context structure when intelCreateContext fails.
    
    intelDestroyContext will eventually be called, and it will clean things
    up.  The call to brwInitVtbl is moved earlier so that
    intelDestroyContext can call the device-specific destructor.  This also
    makes the code look more like the i915 code.
    
    NOTE: This is a candidate for the 9.0 branch.
    
    Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
    Reviewed-by: Eric Anholt <eric@anholt.net>
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54301

commit 22897c74979aa02facdd5cd729db8dadf86924f5
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Fri Sep 28 08:29:12 2012 -0700

    intel: Don't call intelDestroyContext if there is no context to destroy
    
    Some error paths in the device-specific context creation functions can exit
    before the deintel_context structure is allocated.
    
    NOTE: This is a candidate for the 9.0 branch.
    
    Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
    Reviewed-by: Eric Anholt <eric@anholt.net>
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=53618
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54301

commit f93cb0bebb10e3e3e5df099be51021b211650356
Author: Ian Romanick <ian.d.romanick@intel.com>
Date:   Wed Sep 26 11:08:11 2012 -0700

    dri_util: Use calloc to allocate __DRIcontext
    
    The __DRIcontext contains some pointers, and some drivers check for them to 
    NULL in some failure paths.  Instead of sprinkling NULL assignments across t
    various drivers, just zero out the whole thing.
    
    NOTE: This is a candidate for the 9.0 branch.
    
    Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
    Reviewed-by: Eric Anholt <eric@anholt.net>
    Reviewed-and-tested-by: Kenneth Graunke <kenneth@whitecape.org>
    Tested-by: Lu Hua <huax.lu@intel.com>
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=53618
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54301

Verified. Fixed on master branch commit:259fc154f1fdcabbc0a6c02c524962b063f9dee6.