mesa: Changes to 'mesa_7_0_branch'

[Michel Daenzer]

 src/mesa/drivers/dri/common/dri_util.c       |    2 
 src/mesa/drivers/dri/i915tex/intel_buffers.c |   10 +++-
 src/mesa/drivers/dri/i915tex/intel_context.c |   57 ++++++++++++++-------------
 src/mesa/x86/common_x86.c                    |    9 ----
 4 files changed, 41 insertions(+), 37 deletions(-)

commit diffs at http://gitweb.freedesktop.org/?p=mesa/mesa.git;a=summary

New commits:
commit fdefc2bbda9d76576a7473c7d76f9ac7c94ccc87
Author: Michel Dänzer <michel at tungstengraphics.com>
Date:   Tue Jul 10 11:02:18 2007 +0200

    i915tex: Better procedure for dropping batchbuffer on virtual resolution change.
    
    The previous procedure would often result in a GPU lockup.

commit 2079df85279cbf708ffd20dcfa642012f8416bde
Author: Michel Dänzer <michel at tungstengraphics.com>
Date:   Tue Jul 10 10:54:51 2007 +0200

    i915tex: Only wait for vblank when really necessary.
    
    This avoids superfluous waits for vblank timing out under some circumstances.

commit b53659452ccdf33df4bae77d142a86525d5db779
Author: Michel Dänzer <michel at tungstengraphics.com>
Date:   Tue Jul 10 10:49:28 2007 +0200

    Clear pointers to freed cliprects.
    
    Not doing this could lead to double frees under rare circumstances.

commit 46f1d6653ef11a4ce773d7838d037195e7ec1a93
Author: Michel Dänzer <michel at tungstengraphics.com>
Date:   Tue Jun 5 19:52:10 2007 +0200

    Remove dubious compile-time test for pre-2.4 Linux kernels.
    
    LINUX_VERSION_CODE shouldn't be used by userspace code, it can be defined empty
    these days.
    
    If anybody still cares about 2.2 kernels, they should reinstate this as a
    proper runtime test.