[Mesa-dev] [PATCH 18/30] i965: Move texturing to the new resolve functions

[Chad Versace]

On Fri 26 May 2017, Jason Ekstrand wrote:
> ---
>  src/mesa/drivers/dri/i965/brw_context.c       | 55 +++++----------------------
>  src/mesa/drivers/dri/i965/intel_mipmap_tree.c | 55 +++++++++++++++++++++++++++
>  src/mesa/drivers/dri/i965/intel_mipmap_tree.h |  6 +++
>  3 files changed, 71 insertions(+), 45 deletions(-)

>  static void
>  intel_update_state(struct gl_context * ctx, GLuint new_state)
>  {
> @@ -259,16 +224,16 @@ intel_update_state(struct gl_context * ctx, GLuint new_state)
>        /* We need inte_texture_object::_Format to be valid */
>        intel_finalize_mipmap_tree(brw, i);
>  
> -      if (intel_miptree_sample_with_hiz(brw, tex_obj->mt))
> -         intel_miptree_all_slices_resolve_hiz(brw, tex_obj->mt);
> -      else
> -         intel_miptree_all_slices_resolve_depth(brw, tex_obj->mt);
> -      /* Sampling engine understands lossless compression and resolving
> -       * those surfaces should be skipped for performance reasons.
> -       */
> -      const int flags = intel_texture_view_requires_resolve(brw, tex_obj) ?
> -                           0 : INTEL_MIPTREE_IGNORE_CCS_E;
> -      intel_miptree_all_slices_resolve_color(brw, tex_obj->mt, flags);
> +      bool aux_supported;
> +      intel_miptree_prepare_texture(brw, tex_obj->mt, tex_obj->_Format,
> +                                    &aux_supported);
> +
> +      if (!aux_supported && brw->gen >= 9 &&
> +          intel_disable_rb_aux_buffer(brw, tex_obj->mt->bo)) {
> +         perf_debug("Sampling renderbuffer with non-compressible format - "
> +                    "turning off compression");
> +      }
> +

I don't understand why intel_disable_rb_aux_buffer() is called here.
Specifically, I don't understand why the code disables aux for
a *framebuffer* attachment in response to the state of a *texture* view
in the *same* draw call. There is hidden subtlety here.

[time passes... tick tick tick]

In general, the GL spec claims undefined behavior when simultaneuosly
sampling and rendering to the same buffer. Self-blits are legal, though.
So, it seems to me that intel_disable_rb_aux_buffer() is applicable here
only during a glBlit* when src and dest are the same. Is that correct?
Do there exist other applicable scenarios that I'm failing to see?

Anyway, this is just a refactoring patch, so
Reviewed-by: Chad Versace <chadversary at chromium.org>