On Tuesday, January 19, 2016 9:40:25 AM PST Timothy Arceri wrote:
> On Mon, 2016-01-18 at 07:05 -0800, Kenneth Graunke wrote:
> > @@ -1047,6 +1061,32 @@ link_program(struct gl_context *ctx, GLuint
> > program)
> >
> > _mesa_glsl_link_shader(ctx, shProg);
> >
> > + /* Capture .shader_test files. */
> > + const char *capture_path = _mesa_get_shader_capture_path();
> > + if (shProg->Name != 0 && capture_path != NULL) {
> > + FILE *file;
> > + char filename[PATH_MAX];
> > +
> > + _mesa_snprintf(filename, sizeof(filename),
> > "%s/%u.shader_test",
> > + capture_path, shProg->Name);
> > +
> > + file = fopen(filename, "w");
> > + if (file) {
> > + fprintf(file, "[require]\nGLSL >= %u.%02u\n",
> > + shProg->Version / 100, shProg->Version % 100);
> > + if (shProg->SeparateShader)
> > + fprintf(file, "GL_ARB_separate_shader_objects\n");
>
> This just adds the GL_ARB_separate_shader_objects requirement it
> doesn't enable SSO in shader_runner, is that what you want?
>
> Note that if you do enable it then each stage will be a separate
> program which is not necessarily what the application is doing although
> it will be much closer than not enabling it at all. Yeah...I wrote this code before you added "SSO ENABLED" to shader_runner. I originally had patches to shader-db's run.c that made it use glCreateProgram when the [require] block specified GL_ARB_separate_shader_objects, but I never merged it because I'm pretty sure it was buggy. I think specifying "SSO ENABLED" here (or maybe both?) and making run.c handle that is the right thing to do. I'll look into that. --Ken
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ mesa-dev mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/mesa-dev
