[Intel-gfx] gen6 (SNB) depthbuffer issue with OpenGL games

[Ian Romanick]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 07/19/2011 01:21 PM, Nicolas Kalkhof wrote:
> Hi all,
> 
> ok I've nailed the issue down to 3.0.0-rc7 and 3.0.0-rc7-git1. I suspect that the changes made in
> drivers/gpu/drm/i915/i915_dma.c are the cause of the problem.
> 
> http://www.kernel.org/diff/diffview.cgi?file=%2Fpub%2Flinux%2Fkernel%2Fv3.0%2Fsnapshots%2Fpatch-3.0-rc7-git1.bz2;z=14
> 
> Any Clues?

Okay, so that's the commit below, which changes some error clean-up
paths.  That is also odd.  What *exact* GPU do you have?  Specificially,
what's the output of

    glxinfo | grep "OpenGL renderer"

and

    lspci -vn | grep VGA

Does this appear in all games or just certain games?  If it's just in
certain games, which ones?

commit a7b85d2aa63ed09cd5a4a640772b3272f5ac7caa
Author: Keith Packard <keithp at keithp.com>
Date:   Sun Jul 10 13:12:17 2011 -0700

    drm/i915: Clean up i915_driver_load failure path

    i915_driver_load adds a write-combining MTRR region for the GTT
    aperture to improve memory speeds through the aperture. If
    i915_driver_load fails after this, it would not have cleaned up the
    MTRR. This shouldn't cause any problems, except for consuming an MTRR
    register. Still, it's best to clean up completely in the failure path,
    which is easily done by calling mtrr_del if the mtrr was successfully
    allocated.

    i915_driver_load calls i915_gem_load which register
    i915_gem_inactive_shrink. If i915_driver_load fails after calling
    i915_gem_load, the shrinker will be left registered. When called, it
    will access freed memory and crash. The fix is to unregister the
shrinker in the
    failure path using code duplicated from i915_driver_unload.

    i915_driver_load also has some incorrect gotos in the error cleanup
    paths:

     * After failing to initialize the GTT (which cannot happen, btw,
       intel_gtt_get returns a fixed (non-NULL) value), it tries to
       free the uninitialized WC IO mapping. Fixed this by changing the
       target from out_iomapfree to out_rmmap

    Signed-off-by: Keith Packard <keithp at keithp.com>
    Tested-by: Lin Ming <ming.m.lin at intel.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/

iEYEARECAAYFAk4l8j4ACgkQX1gOwKyEAw+XRACaAzuKEihGpktWtA0UbTwx78NR
kBEAoJ7FxkXWEtxC5enkFwnFc19+sPGT
=abq+
-----END PGP SIGNATURE-----