i965/gen6: Fix segfault in transform feedback to DYNAMIC_DRAW buffers.

When storing data in a buffer of type DYNAMIC_DRAW, we don't create a
drm_intel_bo for it; instead we store the data in system memory and
defer allocation of the GPU buffer until it is needed.  Therefore, in
brw_update_sol_surface(), we can't just consult the "buffer" field of
the intel_buffer_object structure; we need to call
intel_bufferobj_buffer() to ensure that the deferred allocation
occurs.

This parallels a similar fix for gen7 (see commit ba6f4c9).

Fixes piglit test EXT_transform_feedback/buffer-usage on gen6.

This is a candidate for the 8.0 release branch.

Reviewed-by: Eric Anholt <eric@anholt.net>

diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
index 7fd83ea984c399961ac2fbacc362a1efb1a52b4e..c77d83a9509bc1a8db727ab6801a8fae45fc2aab 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -730,7 +730,10 @@ brw_update_sol_surface(struct brw_context *brw,
                        uint32_t *out_offset, unsigned num_vector_components,
                        unsigned stride_dwords, unsigned offset_dwords)
 {
-   drm_intel_bo *bo = intel_buffer_object(buffer_obj)->buffer;
+   struct intel_context *intel = &brw->intel;
+   struct intel_buffer_object *intel_bo = intel_buffer_object(buffer_obj);
+   drm_intel_bo *bo =
+      intel_bufferobj_buffer(intel, intel_bo, INTEL_WRITE_PART);
    uint32_t *surf = brw_state_batch(brw, AUB_TRACE_SURFACE_STATE, 6 * 4, 32,
                                     out_offset);
    uint32_t pitch_minus_1 = 4*stride_dwords - 1;