draw: account for separate shader objects in geometry shader code

The geometry shader code seems to have been originally written with the
assumptions that there are the same number of VS outputs as GS outputs and
that VS outputs are in the same order as their corresponding GS inputs. Since
TGSI uses separate shader objects, these are both wrong assumptions. This
was causing several valid vertex/geometry shader combinations to either render
incorrectly or trigger an assertion.

Conflicts:
	src/gallium/auxiliary/draw/draw_gs.c

Reviewed-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>

src/gallium/auxiliary/draw/draw_gs.c

@@ -148,6 +148,22 @@ void draw_delete_geometry_shader(struct draw_context *draw,
    FREE(dgs);
 }
 
+static INLINE int
+draw_gs_get_input_index(int semantic, int index,
+                        const struct tgsi_shader_info *input_info)
+{
+   int i;
+   const ubyte *input_semantic_names = input_info->output_semantic_name;
+   const ubyte *input_semantic_indices = input_info->output_semantic_index;
+   for (i = 0; i < PIPE_MAX_SHADER_OUTPUTS; i++) {
+      if (input_semantic_names[i] == semantic &&
+          input_semantic_indices[i] == index)
+         return i;
+   }
+   debug_assert(0);
+   return -1;
+}
+
 /*#define DEBUG_OUTPUTS 1*/
 static INLINE void
 draw_geometry_fetch_outputs(struct draw_geometry_shader *shader,
@@ -228,6 +244,10 @@ static void draw_fetch_gs_input(struct draw_geometry_shader *shader,
             machine->Inputs[idx].xyzw[3].f[prim_idx] =
                (float)shader->in_prim_idx;
          } else {
+            vs_slot = draw_gs_get_input_index(
+                        shader->info.input_semantic_name[slot],
+                        shader->info.input_semantic_index[slot],
+                        shader->input_info);
 #if DEBUG_INPUTS
             debug_printf("\tSlot = %d, vs_slot = %d, idx = %d:\n",
                          slot, vs_slot, idx);
@@ -381,12 +401,14 @@ int draw_geometry_shader_run(struct draw_geometry_shader *shader,
                              const unsigned constants_size[PIPE_MAX_CONSTANT_BUFFERS], 
                              const struct draw_vertex_info *input_verts,
                              const struct draw_prim_info *input_prim,
+                             const struct tgsi_shader_info *input_info,
                              struct draw_vertex_info *output_verts,
                              struct draw_prim_info *output_prims )
 {
    const float (*input)[4] = (const float (*)[4])input_verts->verts->data;
    unsigned input_stride = input_verts->vertex_size;
-   unsigned vertex_size = input_verts->vertex_size;
+   unsigned num_outputs = shader->info.num_outputs;
+   unsigned vertex_size = sizeof(struct vertex_header) + num_outputs * 4 * sizeof(float);
    struct tgsi_exec_machine *machine = shader->machine;
    unsigned num_input_verts = input_prim->linear ?
                               input_verts->count :
@@ -398,10 +420,10 @@ int draw_geometry_shader_run(struct draw_geometry_shader *shader,
                                                     shader->max_output_vertices)
                             * num_in_primitives;
 
-   output_verts->vertex_size = input_verts->vertex_size;
-   output_verts->stride = input_verts->vertex_size;
+   output_verts->vertex_size = vertex_size;
+   output_verts->stride = output_verts->vertex_size;
    output_verts->verts =
-      (struct vertex_header *)MALLOC(input_verts->vertex_size *
+      (struct vertex_header *)MALLOC(output_verts->vertex_size *
                                      num_in_primitives *
                                      shader->max_output_vertices);
 
@@ -425,6 +447,7 @@ int draw_geometry_shader_run(struct draw_geometry_shader *shader,
    shader->in_prim_idx = 0;
    shader->input_vertex_stride = input_stride;
    shader->input = input;
+   shader->input_info = input_info;
    FREE(shader->primitive_lengths);
    shader->primitive_lengths = MALLOC(max_out_prims * sizeof(unsigned));
 

src/gallium/auxiliary/draw/draw_gs.h

@@ -64,6 +64,7 @@ struct draw_geometry_shader {
    unsigned in_prim_idx;
    unsigned input_vertex_stride;
    const float (*input)[4];
+   const struct tgsi_shader_info *input_info;
 };
 
 /*
@@ -76,6 +77,7 @@ int draw_geometry_shader_run(struct draw_geometry_shader *shader,
                              const unsigned constants_size[PIPE_MAX_CONSTANT_BUFFERS], 
                              const struct draw_vertex_info *input_verts,
                              const struct draw_prim_info *input_prim,
+                             const struct tgsi_shader_info *input_info,
                              struct draw_vertex_info *output_verts,
                              struct draw_prim_info *output_prims );
 

src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c

@@ -270,6 +270,7 @@ static void fetch_pipeline_generic( struct draw_pt_middle_end *middle,
                                draw->pt.user.gs_constants_size,
                                vert_info,
                                prim_info,
+                               &vshader->info,
                                &gs_vert_info,
                                &gs_prim_info);
 

src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline_llvm.c

@@ -289,11 +289,13 @@ llvm_pipeline_generic( struct draw_pt_middle_end *middle,
 
 
    if ((opt & PT_SHADE) && gshader) {
+      struct draw_vertex_shader *vshader = draw->vs.vertex_shader;
       draw_geometry_shader_run(gshader,
                                draw->pt.user.gs_constants,
                                draw->pt.user.gs_constants_size,
                                vert_info,
                                prim_info,
+                               &vshader->info,
                                &gs_vert_info,
                                &gs_prim_info);