[Mesa-dev] [PATCH] mesa: Only validate SSO shader IO in OpenGL ES or debug context

[Ilia Mirkin]

This will cause st/mesa to break, no? Right now validate_io iterates
over the shader ir, which st/mesa frees after linking.

  -ilia

On Thu, May 26, 2016 at 5:59 PM, Ian Romanick <idr at freedesktop.org> wrote:
> From: Ian Romanick <ian.d.romanick at intel.com>
>
> Signed-off-by: Ian Romanick <ian.d.romanick at intel.com>
> Suggested-by: Timothy Arceri <timothy.arceri at collabora.com>
> Cc: Timothy Arceri <timothy.arceri at collabora.com>
> ---
>  src/mesa/main/pipelineobj.c    | 18 ++++++++++++++++--
>  src/mesa/main/shader_query.cpp | 12 ++++--------
>  2 files changed, 20 insertions(+), 10 deletions(-)
>
> diff --git a/src/mesa/main/pipelineobj.c b/src/mesa/main/pipelineobj.c
> index b150198..5a46cfe 100644
> --- a/src/mesa/main/pipelineobj.c
> +++ b/src/mesa/main/pipelineobj.c
> @@ -906,7 +906,8 @@ _mesa_validate_program_pipeline(struct gl_context* ctx,
>     /* Validate inputs against outputs, this cannot be done during linking
>      * since programs have been linked separately from each other.
>      *
> -    * From OpenGL 4.5 Core spec:
> +    * Section 11.1.3.11 (Validation) of the OpenGL 4.5 Core Profile spec says:
> +    *
>      *     "Separable program objects may have validation failures that cannot be
>      *     detected without the complete program pipeline. Mismatched interfaces,
>      *     improper usage of program objects together, and the same
> @@ -914,8 +915,21 @@ _mesa_validate_program_pipeline(struct gl_context* ctx,
>      *     program objects."
>      *
>      * OpenGL ES 3.1 specification has the same text.
> +    *
> +    * Section 11.1.3.11 (Validation) of the OpenGL ES spec also says:
> +    *
> +    *    An INVALID_OPERATION error is generated by any command that transfers
> +    *    vertices to the GL or launches compute work if the current set of
> +    *    active program objects cannot be executed, for reasons including:
> +    *
> +    *    * The current program pipeline object contains a shader interface
> +    *      that doesn't have an exact match (see section 7.4.1)
> +    *
> +    * Based on this, only perform the most-strict checking on ES or when the
> +    * application has created a debug context.
>      */
> -   if (!_mesa_validate_pipeline_io(pipe))
> +   if ((_mesa_is_gles(ctx) || (ctx->Const.ContextFlags & GL_CONTEXT_FLAG_DEBUG_BIT)) &&
> +       !_mesa_validate_pipeline_io(pipe))
>        return GL_FALSE;
>
>     pipe->Validated = GL_TRUE;
> diff --git a/src/mesa/main/shader_query.cpp b/src/mesa/main/shader_query.cpp
> index 9e18a1c..a37c179 100644
> --- a/src/mesa/main/shader_query.cpp
> +++ b/src/mesa/main/shader_query.cpp
> @@ -1372,7 +1372,7 @@ _mesa_get_program_resourceiv(struct gl_shader_program *shProg,
>
>  static bool
>  validate_io(const struct gl_shader *producer,
> -            const struct gl_shader *consumer, bool isES)
> +            const struct gl_shader *consumer)
>  {
>     assert(producer && consumer);
>     unsigned inputs = 0, outputs = 0;
> @@ -1416,10 +1416,6 @@ validate_io(const struct gl_shader *producer,
>      * packing makes this challenging.
>      */
>
> -   /* Currently no matching done for desktop. */
> -   if (!isES)
> -      return true;
> -
>     /* For each output in a, find input in b and do any required checks. */
>     foreach_in_list(ir_instruction, out, producer->ir) {
>        ir_variable *out_var = out->as_variable();
> @@ -1502,9 +1498,9 @@ _mesa_validate_pipeline_io(struct gl_pipeline_object *pipeline)
>              break;
>
>           if (!validate_io(shProg[prev]->_LinkedShaders[prev],
> -                          shProg[idx]->_LinkedShaders[idx],
> -                          shProg[prev]->IsES || shProg[idx]->IsES))
> -            return false;
> +                          shProg[idx]->_LinkedShaders[idx]))
> +               return false;
> +
>           prev = idx;
>        }
>     }
> --
> 2.5.5
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev