drm/i915: Don't tag kernel batches as user batches
authorJohn Harrison <John.C.Harrison@Intel.com>
Fri, 29 May 2015 16:43:38 +0000 (17:43 +0100)
committerDaniel Vetter <daniel.vetter@ffwll.ch>
Tue, 23 Jun 2015 12:02:07 +0000 (14:02 +0200)
The render state initialisation code does an explicit i915_add_request() call to
commit the init commands. It was passing in the initialisation batch buffer to
add_request() as the batch object parameter. However, the batch object entry in
the request structure (which is all that parameter is used for) is meant for
keeping track of user generated batch buffers for blame tagging during GPU
hangs.

This patch clears the batch object parameter so that kernel generated batch
buffers are not tagged as being user generated.

For: VIZ-5115
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: Tomas Elf <tomas.elf@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
drivers/gpu/drm/i915/i915_gem_render_state.c
drivers/gpu/drm/i915/intel_lrc.c

index 4418616301e755efa54a59e03ea06c2d10a59f01..a32a4b9492b65c482d00ed8ea1577b2528e6d035 100644 (file)
@@ -173,7 +173,7 @@ int i915_gem_render_state_init(struct intel_engine_cs *ring)
 
        i915_vma_move_to_active(i915_gem_obj_to_ggtt(so.obj), ring);
 
-       __i915_add_request(ring, NULL, so.obj, true);
+       __i915_add_request(ring, NULL, NULL, true);
        /* __i915_add_request moves object to inactive if it fails */
 out:
        i915_gem_render_state_fini(&so);
index 47443b9aa17216f9ab824224c1ebb1677b4e5617..550e854a350917161dacdcd935d03c9e7104a021 100644 (file)
@@ -1599,7 +1599,7 @@ static int intel_lr_context_render_state_init(struct intel_engine_cs *ring,
 
        i915_vma_move_to_active(i915_gem_obj_to_ggtt(so.obj), ring);
 
-       __i915_add_request(ring, file, so.obj, true);
+       __i915_add_request(ring, file, NULL, true);
        /* intel_logical_ring_add_request moves object to inactive if it
         * fails */
 out: