[Mesa-stable] [PATCH] st/mesa: inform the driver of framebuffer changes before compute dispatches

Marek Olšák maraeo at gmail.com
Wed Feb 22 20:19:41 UTC 2017


Makes sense.

Reviewed-by: Marek Olšák <marek.olsak at amd.com>

Marek

On Wed, Feb 22, 2017 at 7:59 PM, Nicolai Hähnle <nhaehnle at gmail.com> wrote:
> From: Nicolai Hähnle <nicolai.haehnle at amd.com>
>
> Even though compute shaders cannot access the framebuffer, there is a
> synchronization issue when a compute dispatch accesses a texture that
> was previously bound and drawn to as a framebuffer.
>
> Section 9.3 (Feedback Loops Between Textures and the Framebuffer) of
> the OpenGL 4.5 spec rather implicitly clarifies that undefined behavior
> results if the texture is still attached to the currently bound
> framebuffer. However, the feedback loop is broken when the application
> changes the framebuffer binding before a compute dispatch, and the
> state tracker needs to let the driver known about this.
>
> Fixes GL45-CTS.compute_shader.pipeline-post-fs on SI family Radeons.
>
> Cc: mesa-stable at lists.freedesktop.org
> ---
>  src/mesa/state_tracker/st_atom.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
>
> diff --git a/src/mesa/state_tracker/st_atom.c b/src/mesa/state_tracker/st_atom.c
> index 65ac517..9e1137e 100644
> --- a/src/mesa/state_tracker/st_atom.c
> +++ b/src/mesa/state_tracker/st_atom.c
> @@ -181,21 +181,29 @@ void st_validate_state( struct st_context *st, enum st_pipeline pipeline )
>        struct gl_program *new_cp = ctx->ComputeProgram._Current;
>
>        if (new_cp != &old_cp->Base) {
>           if (old_cp)
>              st->dirty |= old_cp->affected_states;
>           assert(new_cp);
>           st->dirty |= st_compute_program(new_cp)->affected_states;
>        }
>
>        st->compute_shader_may_be_dirty = false;
> -      pipeline_mask = ST_PIPELINE_COMPUTE_STATE_MASK;
> +
> +      /*
> +       * We add the ST_NEW_FB_STATE bit here as well, because glBindFramebuffer
> +       * acts as a barrier that breaks feedback loops between the framebuffer
> +       * and textures bound to the framebuffer, even when those textures are
> +       * accessed by compute shaders; so we must inform the driver of new
> +       * framebuffer state.
> +       */
> +      pipeline_mask = ST_PIPELINE_COMPUTE_STATE_MASK | ST_NEW_FB_STATE;
>        break;
>     }
>
>     default:
>        unreachable("Invalid pipeline specified");
>     }
>
>     dirty = st->dirty & pipeline_mask;
>     if (!dirty)
>        return;
> --
> 2.9.3
>
> _______________________________________________
> mesa-stable mailing list
> mesa-stable at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-stable


More information about the mesa-stable mailing list