drm/i915/dpt: Treat the DPT BO as a framebuffer
authorVille Syrjälä <ville.syrjala@linux.intel.com>
Mon, 20 Mar 2023 09:05:17 +0000 (11:05 +0200)
committerVille Syrjälä <ville.syrjala@linux.intel.com>
Fri, 24 Mar 2023 15:14:34 +0000 (17:14 +0200)
commit779cb5ba64ec7df80675a956c9022929514f517a
tree5d70b40cda6b4c510c9587f5f34d3999d91077ce
parent883631771038d1b0c10c0929e31bbd5ffb5e682c
drm/i915/dpt: Treat the DPT BO as a framebuffer

Currently i915_gem_object_is_framebuffer() doesn't treat the
BO containing the framebuffer's DPT as a framebuffer itself.
This means eg. that the shrinker can evict the DPT BO while
leaving the actual FB BO bound, when the DPT is allocated
from regular shmem.

That causes an immediate oops during hibernate as we
try to rewrite the PTEs inside the already evicted
DPT obj.

TODO: presumably this might also be the reason for the
DPT related display faults under heavy memory pressure,
but I'm still not sure how that would happen as the object
should be pinned by intel_dpt_pin() while in active use by
the display engine...

Cc: stable@vger.kernel.org
Cc: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Fixes: 0dc987b699ce ("drm/i915/display: Add smem fallback allocation for dpt")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230320090522.9909-2-ville.syrjala@linux.intel.com
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
drivers/gpu/drm/i915/display/intel_dpt.c
drivers/gpu/drm/i915/gem/i915_gem_object.h
drivers/gpu/drm/i915/gem/i915_gem_object_types.h