On Fri 10 Apr 2015, Pohjolainen, Topi wrote:
On Thu, Apr 09, 2015 at 08:57:04PM -0700, Chad Versace wrote:
diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c index a906460..492338b 100644 --- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c +++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c @@ -403,7 +403,8 @@ intel_miptree_create_layout(struct brw_context *brw, if (!for_bo && _mesa_get_format_base_format(format) == GL_DEPTH_STENCIL && (brw->must_use_separate_stencil || - (brw->has_separate_stencil && brw_is_hiz_depth_format(brw, format)))) { + (brw->has_separate_stencil && + intel_miptree_wants_hiz_buffer(brw, mt)))) {I had to check that format == mt->format, and indeed that is the case. I noticed that we probably want to make intel_miptree_create_layout() a little more consistent - it uses randomly both the stack variable and the one in miptree.
I agree. For large functions like intel_miptree_create_layout(), I like to group together all assignments to the struct at the bottom of the function, and use stack variables until then. Otherwise, it's too easy to accidentally use an unitialized struct field in the middle of the function. _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
