fix issue to create cl image from libva with non-zero offset

Beignet accepts buffer object name to share data between libva,
it is supposed to support to create cl image from the bo name
with a non-zero offset, but it does not work at some platforms.

The driver calls intel_bo_gem_create_from_name to retrieve the
dri_bo, and the offset of dri_bo is changed by the non-zero offset.
At some platforms, the change of the offset has side effect when
the kernel is executed again and so intel_bo_gem_create_from_name
is called for the second time.

So, do not change the offset of dri_bo, but maintain the non-zero
offset in cl_image, and use the non-zero offset until we fill the
surface state.

Signed-off-by: Guo Yejun <yejun.guo@intel.com>
Tested-by: Zhigang Gong <zhigang.gong@linux.intel.com>

diff --git a/src/cl_driver.h b/src/cl_driver.h
index 8697ff2..58bf489 100644 (file)
--- a/src/cl_driver.h
+++ b/src/cl_driver.h
@@ -305,7 +305,7 @@ extern cl_buffer_release_from_texture_cb *cl_buffer_release_from_texture;
 typedef cl_buffer (cl_buffer_get_buffer_from_libva_cb)(cl_context ctx, unsigned int bo_name, size_t *sz);
 extern cl_buffer_get_buffer_from_libva_cb *cl_buffer_get_buffer_from_libva;
 
-typedef cl_buffer (cl_buffer_get_image_from_libva_cb)(cl_context ctx, unsigned int bo_name, struct _cl_mem_image *image, unsigned int offset);
+typedef cl_buffer (cl_buffer_get_image_from_libva_cb)(cl_context ctx, unsigned int bo_name, struct _cl_mem_image *image);
 extern cl_buffer_get_image_from_libva_cb *cl_buffer_get_image_from_libva;
 
 /* Unref a buffer and destroy it if no more ref */

diff --git a/src/cl_mem.c b/src/cl_mem.c
index d319966..cbe7c5b 100644 (file)
--- a/src/cl_mem.c
+++ b/src/cl_mem.c
@@ -1910,7 +1910,7 @@ LOCAL cl_mem cl_mem_new_libva_image(cl_context ctx,
 
   image = cl_mem_image(mem);
 
-  mem->bo = cl_buffer_get_image_from_libva(ctx, bo_name, image, offset);
+  mem->bo = cl_buffer_get_image_from_libva(ctx, bo_name, image);
 
   image->w = width;
   image->h = height;

diff --git a/src/intel/intel_driver.c b/src/intel/intel_driver.c
index fc037cc..f627f98 100644 (file)
--- a/src/intel/intel_driver.c
+++ b/src/intel/intel_driver.c
@@ -675,15 +675,13 @@ cl_buffer intel_share_buffer_from_libva(cl_context ctx,
 
 cl_buffer intel_share_image_from_libva(cl_context ctx,
                                        unsigned int bo_name,
-                                       struct _cl_mem_image *image,
-                                       unsigned int offset)
+                                       struct _cl_mem_image *image)
 {
   drm_intel_bo *intel_bo;
   uint32_t intel_tiling, intel_swizzle_mode;
 
   intel_bo = intel_driver_share_buffer((intel_driver_t *)ctx->drv, "shared from libva", bo_name);
 
-  intel_bo->offset += offset;
   drm_intel_bo_get_tiling(intel_bo, &intel_tiling, &intel_swizzle_mode);
   image->tiling = get_cl_tiling(intel_tiling);
 

diff --git a/src/intel/intel_gpgpu.c b/src/intel/intel_gpgpu.c
index 5898906..f901330 100644 (file)
--- a/src/intel/intel_gpgpu.c
+++ b/src/intel/intel_gpgpu.c
@@ -1061,7 +1061,7 @@ intel_gpgpu_bind_image_gen7(intel_gpgpu_t *gpgpu,
     ss->ss0.surface_array_spacing = 1;
   }
   ss->ss0.surface_format = format;
-  ss->ss1.base_addr = obj_bo->offset;
+  ss->ss1.base_addr = obj_bo->offset + obj_bo_offset;
   ss->ss2.width = w - 1;
 
   ss->ss2.height = h - 1;
@@ -1078,7 +1078,7 @@ intel_gpgpu_bind_image_gen7(intel_gpgpu_t *gpgpu,
     ss->ss0.tile_walk = I965_TILEWALK_YMAJOR;
   }
   ss->ss0.render_cache_rw_mode = 1; /* XXX do we need to set it? */
-  intel_gpgpu_set_buf_reloc_gen7(gpgpu, index, obj_bo, obj_bo_offset);
+  intel_gpgpu_set_buf_reloc_gen7(gpgpu, index, obj_bo, obj_bo->offset + obj_bo_offset);
 
   assert(index < GEN_MAX_SURFACES);
 }
@@ -1106,7 +1106,7 @@ intel_gpgpu_bind_image_gen75(intel_gpgpu_t *gpgpu,
     ss->ss0.surface_array_spacing = 1;
   }
   ss->ss0.surface_format = format;
-  ss->ss1.base_addr = obj_bo->offset;
+  ss->ss1.base_addr = obj_bo->offset + obj_bo_offset;
   ss->ss2.width = w - 1;
   ss->ss2.height = h - 1;
   ss->ss3.depth = depth - 1;
@@ -1126,7 +1126,7 @@ intel_gpgpu_bind_image_gen75(intel_gpgpu_t *gpgpu,
     ss->ss0.tile_walk = I965_TILEWALK_YMAJOR;
   }
   ss->ss0.render_cache_rw_mode = 1; /* XXX do we need to set it? */
-  intel_gpgpu_set_buf_reloc_gen7(gpgpu, index, obj_bo, obj_bo_offset);
+  intel_gpgpu_set_buf_reloc_gen7(gpgpu, index, obj_bo, obj_bo->offset + obj_bo_offset);
 
   assert(index < GEN_MAX_SURFACES);
 }