radv/shader_info: repack and compact struct radv_shader_info

using pahole.

Reduces the size of radv_shader_info from 760 bytes to 640 bytes.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37931>

src/amd/vulkan/radv_shader_info.h

@@ -79,98 +79,88 @@ struct gfx10_ngg_info {
 };
 
 struct radv_shader_info {
+   uint32_t workgroup_size;
+   uint32_t nir_shared_size;
    uint64_t inline_push_constant_mask;
    uint32_t push_constant_size;
-   bool can_inline_all_push_constants;
-   bool loads_push_constants;
-   bool loads_dynamic_offsets;
    uint32_t desc_set_used_mask;
-   bool uses_view_index;
+   uint32_t user_data_0;
-   bool uses_invocation_id;
-   bool uses_prim_id;
-   uint8_t wave_size;
-   struct radv_userdata_locations user_sgprs_locs;
-   bool is_ngg;
-   bool is_ngg_passthrough;
-   bool has_ngg_culling;
-   bool has_ngg_early_prim_export;
-   bool has_prim_query;
-   bool has_xfb_query;
-   uint8_t ngg_lds_scratch_size;
    uint32_t num_tess_patches;
    uint32_t esgs_itemsize;       /* Only for VS or TES as ES */
    uint32_t ngg_lds_vertex_size; /* VS,TES: Cull+XFB, GS: GSVS size */
-   struct radv_vs_output_info outinfo;
-   unsigned workgroup_size;
-   bool force_vrs_per_vertex;
-   mesa_shader_stage stage;
-   mesa_shader_stage next_stage;
-   enum radv_shader_type type;
-   uint32_t user_data_0;
-   bool inputs_linked;
-   bool outputs_linked;
-   bool merged_shader_compiled_separately; /* GFX9+ */
-   bool force_indirect_descriptors;
    uint64_t gs_inputs_read;  /* Mask of GS inputs read (only used by linked ES) */
-   unsigned nir_shared_size;
+
+   struct radv_userdata_locations user_sgprs_locs;
+   struct radv_vs_output_info outinfo;
+
+   uint8_t wave_size;
+   uint8_t ngg_lds_scratch_size;
+   mesa_shader_stage stage : 8;
+   mesa_shader_stage next_stage : 8;
+   enum radv_shader_type type : 4;
+
+   bool can_inline_all_push_constants : 1;
+   bool loads_push_constants : 1;
+   bool loads_dynamic_offsets : 1;
+   bool uses_view_index : 1;
+   bool uses_invocation_id : 1;
+   bool uses_prim_id : 1;
+   bool is_ngg : 1;
+   bool is_ngg_passthrough : 1;
+   bool has_ngg_culling : 1;
+   bool has_ngg_early_prim_export : 1;
+   bool has_prim_query : 1;
+   bool has_xfb_query : 1;
+   bool force_vrs_per_vertex : 1;
+   bool inputs_linked : 1;
+   bool outputs_linked : 1;
+   bool merged_shader_compiled_separately : 1; /* GFX9+ */
+   bool force_indirect_descriptors : 1;
 
    struct {
-      bool needs_draw_id;
-      bool needs_instance_id;
-      bool as_es;
-      bool as_ls;
-      bool tcs_in_out_eq;
       uint64_t tcs_inputs_via_temp;
       uint64_t tcs_inputs_via_lds;
-      uint8_t num_linked_outputs;
-      bool needs_base_instance;
-      bool use_per_attribute_vb_descs;
       uint32_t vb_desc_usage_mask;
       uint32_t input_slot_usage_mask;
-      uint8_t num_attributes;
-      bool has_prolog;
-      bool dynamic_inputs;
-      bool dynamic_num_verts_per_prim;
       uint32_t num_outputs; /* For NGG streamout only */
+      uint8_t num_linked_outputs;
+      uint8_t num_attributes;
+      bool needs_draw_id : 1;
+      bool needs_instance_id : 1;
+      bool as_es : 1;
+      bool as_ls : 1;
+      bool tcs_in_out_eq : 1;
+      bool needs_base_instance : 1;
+      bool use_per_attribute_vb_descs : 1;
+      bool has_prolog : 1;
+      bool dynamic_inputs : 1;
+      bool dynamic_num_verts_per_prim : 1;
    } vs;
    struct {
       uint8_t num_components_per_stream[4];
-      unsigned vertices_in;
+      uint32_t vertices_in;
-      unsigned vertices_out;
+      uint32_t vertices_out;
-      unsigned input_prim;
+      uint32_t input_prim;
-      unsigned output_prim;
+      uint32_t output_prim;
-      unsigned invocations;
+      uint32_t invocations;
-      unsigned es_type; /* GFX9: VS or TES */
+      uint32_t es_type; /* GFX9: VS or TES */
       uint8_t num_linked_inputs;
       bool has_pipeline_stat_query;
    } gs;
    struct {
-      bool as_es;
+      uint32_t tcs_vertices_out;
-      enum tess_primitive_mode _primitive_mode;
+      uint32_t num_outputs;            /* For NGG streamout only */
-      enum gl_tess_spacing spacing;
-      bool ccw;
-      bool point_mode;
-      bool reads_tess_factors;
-      unsigned tcs_vertices_out;
       uint8_t num_linked_inputs;       /* Number of reserved per-vertex input slots in VRAM. */
       uint8_t num_linked_patch_inputs; /* Number of reserved per-patch input slots in VRAM. */
       uint8_t num_linked_outputs;
-      uint32_t num_outputs; /* For NGG streamout only */
+      enum tess_primitive_mode _primitive_mode : 2;
+      enum gl_tess_spacing spacing : 2;
+      bool as_es : 1;
+      bool ccw : 1;
+      bool point_mode : 1;
+      bool reads_tess_factors : 1;
    } tes;
    struct {
-      bool uses_sample_shading;
-      bool needs_sample_positions;
-      bool needs_poly_line_smooth;
-      bool writes_memory;
-      bool writes_z;
-      bool writes_stencil;
-      bool writes_sample_mask;
-      bool writes_mrt0_alpha;
-      bool exports_mrtz_via_epilog;
-      bool has_pcoord;
-      bool prim_id_input;
-      bool viewport_index_input;
-      uint8_t input_clips_culls_mask;
       uint32_t input_mask;
       uint32_t input_per_primitive_mask;
       uint32_t float32_shaded_mask;
@@ -179,68 +169,82 @@ struct radv_shader_info {
       uint32_t float16_shaded_mask;
       uint32_t float16_hi_shaded_mask;
       uint32_t num_inputs;
-      bool can_discard;
+      uint8_t input_clips_culls_mask;
-      bool early_fragment_test;
+      bool uses_sample_shading : 1;
-      bool post_depth_coverage;
+      bool needs_sample_positions : 1;
-      bool reads_sample_mask_in;
+      bool needs_poly_line_smooth : 1;
-      bool reads_front_face;
+      bool writes_memory : 1;
-      bool reads_sample_id;
+      bool writes_z : 1;
-      bool reads_frag_shading_rate;
+      bool writes_stencil : 1;
-      bool reads_barycentric_model;
+      bool writes_sample_mask : 1;
-      bool reads_persp_sample;
+      bool writes_mrt0_alpha : 1;
-      bool reads_persp_center;
+      bool mrt0_is_dual_src : 1;
-      bool reads_persp_centroid;
+      bool exports_mrtz_via_epilog : 1;
-      bool reads_linear_sample;
+      bool has_pcoord : 1;
-      bool reads_linear_center;
+      bool prim_id_input : 1;
-      bool reads_linear_centroid;
+      bool viewport_index_input : 1;
-      bool reads_fully_covered;
+      bool can_discard : 1;
-      bool reads_pixel_coord;
+      bool early_fragment_test : 1;
-      bool reads_layer;
+      bool post_depth_coverage : 1;
       uint8_t reads_frag_coord_mask;
       uint8_t reads_sample_pos_mask;
       uint8_t depth_layout;
-      bool allow_flat_shading;
+      bool reads_sample_mask_in : 1;
-      bool pops; /* Uses Primitive Ordered Pixel Shading (fragment shader interlock) */
+      bool reads_front_face : 1;
-      bool pops_is_per_sample;
+      bool reads_sample_id : 1;
-      bool mrt0_is_dual_src;
+      bool reads_frag_shading_rate : 1;
+      bool reads_barycentric_model : 1;
+      bool reads_persp_sample : 1;
+      bool reads_persp_center : 1;
+      bool reads_persp_centroid : 1;
+      bool reads_linear_sample : 1;
+      bool reads_linear_center : 1;
+      bool reads_linear_centroid : 1;
+      bool reads_fully_covered : 1;
+      bool reads_pixel_coord : 1;
+      bool reads_layer : 1;
+      bool pops : 1; /* Uses Primitive Ordered Pixel Shading (fragment shader interlock) */
+      bool pops_is_per_sample : 1;
       uint32_t spi_ps_input_ena;
       uint32_t spi_ps_input_addr;
       uint32_t colors_written; /* Mask of outputs written */
       uint32_t spi_shader_col_format;
       uint32_t cb_shader_mask;
       uint8_t color0_written;
-      bool load_provoking_vtx;
+      bool load_provoking_vtx : 1;
-      bool load_rasterization_prim;
+      bool load_rasterization_prim : 1;
-      bool force_sample_iter_shading_rate;
+      bool force_sample_iter_shading_rate : 1;
-      bool uses_fbfetch_output;
+      bool uses_fbfetch_output : 1;
-      bool has_epilog;
+      bool allow_flat_shading : 1;
+
+      bool has_epilog : 1;
    } ps;
    struct {
-      bool uses_grid_size;
+      uint32_t block_size[3];
       bool uses_block_id[3];
       bool uses_thread_id[3];
-      bool uses_local_invocation_idx;
+      bool uses_grid_size : 1;
-      unsigned block_size[3];
+      bool uses_local_invocation_idx : 1;
 
-      bool uses_full_subgroups;
+      bool uses_full_subgroups : 1;
-      bool linear_taskmesh_dispatch;
+      bool linear_taskmesh_dispatch : 1;
-      bool has_query; /* Task shader only */
+      bool has_query : 1; /* Task shader only */
 
-      bool regalloc_hang_bug;
+      bool regalloc_hang_bug : 1;
 
-      unsigned derivative_group : 2;
+      uint32_t derivative_group : 2;
    } cs;
    struct {
       ac_nir_tess_io_info io_info;
       uint64_t tes_inputs_read;
       uint64_t tes_patch_inputs_read;
-      unsigned tcs_vertices_out;
+      uint32_t tcs_vertices_out;
       uint32_t lds_size;         /* in bytes */
       uint8_t num_linked_inputs; /* Number of reserved per-vertex input slots in LDS. */
       bool tes_reads_tess_factors : 1;
    } tcs;
    struct {
-      enum mesa_prim output_prim;
+      enum mesa_prim output_prim : 8; /* byte-size aligned */
       bool needs_ms_scratch_ring;
       bool has_task; /* If mesh shader is used together with a task shader. */
       bool has_query;