src/compiler/shader_info.h

   1 /*
   2  * Copyright © 2016 Intel Corporation
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining a
   5  * copy of this software and associated documentation files (the "Software"),
   6  * to deal in the Software without restriction, including without limitation
   7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
   8  * and/or sell copies of the Software, and to permit persons to whom the
   9  * Software is furnished to do so, subject to the following conditions:
  10  *
  11  * The above copyright notice and this permission notice (including the next
  12  * paragraph) shall be included in all copies or substantial portions of the
  13  * Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  21  * IN THE SOFTWARE.
  22  *
  23  */
  24
  25 #ifndef SHADER_INFO_H
  26 #define SHADER_INFO_H
  27
  28 #include "shader_enums.h"
  29 #include <stdint.h>
  30
  31 #ifdef __cplusplus
  32 extern "C" {
  33 #endif
  34
  35 struct spirv_supported_capabilities {
  36    bool address;
  37    bool atomic_storage;
  38    bool demote_to_helper_invocation;
  39    bool derivative_group;
  40    bool descriptor_array_dynamic_indexing;
  41    bool descriptor_array_non_uniform_indexing;
  42    bool descriptor_indexing;
  43    bool device_group;
  44    bool draw_parameters;
  45    bool float64;
  46    bool fragment_shader_sample_interlock;
  47    bool fragment_shader_pixel_interlock;
  48    bool geometry_streams;
  49    bool image_ms_array;
  50    bool image_read_without_format;
  51    bool image_write_without_format;
  52    bool int8;
  53    bool int16;
  54    bool int64;
  55    bool int64_atomics;
  56    bool kernel;
  57    bool min_lod;
  58    bool multiview;
  59    bool physical_storage_buffer_address;
  60    bool post_depth_coverage;
  61    bool runtime_descriptor_array;
  62    bool float_controls;
  63    bool shader_clock;
  64    bool shader_viewport_index_layer;
  65    bool stencil_export;
  66    bool storage_8bit;
  67    bool storage_16bit;
  68    bool storage_image_ms;
  69    bool subgroup_arithmetic;
  70    bool subgroup_ballot;
  71    bool subgroup_basic;
  72    bool subgroup_quad;
  73    bool subgroup_shuffle;
  74    bool subgroup_vote;
  75    bool tessellation;
  76    bool transform_feedback;
  77    bool variable_pointers;
  78    bool float16;
  79    bool amd_gcn_shader;
  80    bool amd_shader_ballot;
  81    bool amd_trinary_minmax;
  82 };
  83
  84 typedef struct shader_info {
  85    const char *name;
  86
  87    /* Descriptive name provided by the client; may be NULL */
  88    const char *label;
  89
  90    /** The shader stage, such as MESA_SHADER_VERTEX. */
  91    gl_shader_stage stage;
  92
  93    /** The shader stage in a non SSO linked program that follows this stage,
  94      * such as MESA_SHADER_FRAGMENT.
  95      */
  96    gl_shader_stage next_stage;
  97
  98    /* Number of textures used by this shader */
  99    unsigned num_textures;
 100    /* Number of uniform buffers used by this shader */
 101    unsigned num_ubos;
 102    /* Number of atomic buffers used by this shader */
 103    unsigned num_abos;
 104    /* Number of shader storage buffers used by this shader */
 105    unsigned num_ssbos;
 106    /* Number of images used by this shader */
 107    unsigned num_images;
 108    /* Index of the last MSAA image. */
 109    int last_msaa_image;
 110
 111    /* Which inputs are actually read */
 112    uint64_t inputs_read;
 113    /* Which outputs are actually written */
 114    uint64_t outputs_written;
 115    /* Which outputs are actually read */
 116    uint64_t outputs_read;
 117    /* Which system values are actually read */
 118    uint64_t system_values_read;
 119
 120    /* Which patch inputs are actually read */
 121    uint32_t patch_inputs_read;
 122    /* Which patch outputs are actually written */
 123    uint32_t patch_outputs_written;
 124    /* Which patch outputs are read */
 125    uint32_t patch_outputs_read;
 126
 127    /* Whether or not this shader ever uses textureGather() */
 128    bool uses_texture_gather;
 129
 130    /** Bitfield of which textures are used */
 131    uint32_t textures_used;
 132
 133    /** Bitfield of which textures are used by texelFetch() */
 134    uint32_t textures_used_by_txf;
 135
 136    /**
 137     * True if this shader uses the fddx/fddy opcodes.
 138     *
 139     * Note that this does not include the "fine" and "coarse" variants.
 140     */
 141    bool uses_fddx_fddy;
 142
 143    /**
 144     * True if this shader uses 64-bit ALU operations
 145     */
 146    bool uses_64bit;
 147
 148    /* The size of the gl_ClipDistance[] array, if declared. */
 149    unsigned clip_distance_array_size;
 150
 151    /* The size of the gl_CullDistance[] array, if declared. */
 152    unsigned cull_distance_array_size;
 153
 154    /* Whether the first UBO is the default uniform buffer, i.e. uniforms. */
 155    bool first_ubo_is_default_ubo;
 156
 157    /* Whether or not separate shader objects were used */
 158    bool separate_shader;
 159
 160    /** Was this shader linked with any transform feedback varyings? */
 161    bool has_transform_feedback_varyings;
 162
 163    /* SPV_KHR_float_controls: execution mode for floating point ops */
 164    unsigned float_controls_execution_mode;
 165
 166    union {
 167       struct {
 168          /* Which inputs are doubles */
 169          uint64_t double_inputs;
 170
 171          /* For AMD-specific driver-internal shaders. It replaces vertex
 172           * buffer loads with code generating VS inputs from scalar registers.
 173           *
 174           * Valid values: SI_VS_BLIT_SGPRS_POS_*
 175           */
 176          unsigned blit_sgprs_amd;
 177
 178          /* True if the shader writes position in window space coordinates pre-transform */
 179          bool window_space_position;
 180       } vs;
 181
 182       struct {
 183          /** The number of vertices recieves per input primitive */
 184          unsigned vertices_in;
 185
 186          /** The output primitive type (GL enum value) */
 187          unsigned output_primitive;
 188
 189          /** The input primitive type (GL enum value) */
 190          unsigned input_primitive;
 191
 192          /** The maximum number of vertices the geometry shader might write. */
 193          unsigned vertices_out;
 194
 195          /** 1 .. MAX_GEOMETRY_SHADER_INVOCATIONS */
 196          unsigned invocations;
 197
 198          /** Whether or not this shader uses EndPrimitive */
 199          bool uses_end_primitive;
 200
 201          /** Whether or not this shader uses non-zero streams */
 202          bool uses_streams;
 203       } gs;
 204
 205       struct {
 206          bool uses_discard;
 207
 208          /**
 209           * True if this fragment shader requires helper invocations.  This
 210           * can be caused by the use of ALU derivative ops, texture
 211           * instructions which do implicit derivatives, and the use of quad
 212           * subgroup operations.
 213           */
 214          bool needs_helper_invocations;
 215
 216          /**
 217           * Whether any inputs are declared with the "sample" qualifier.
 218           */
 219          bool uses_sample_qualifier;
 220
 221          /**
 222           * Whether early fragment tests are enabled as defined by
 223           * ARB_shader_image_load_store.
 224           */
 225          bool early_fragment_tests;
 226
 227          /**
 228           * Defined by INTEL_conservative_rasterization.
 229           */
 230          bool inner_coverage;
 231
 232          bool post_depth_coverage;
 233
 234          /**
 235           * \name ARB_fragment_coord_conventions
 236           * @{
 237           */
 238          bool pixel_center_integer;
 239          bool origin_upper_left;
 240          /*@}*/
 241
 242          bool pixel_interlock_ordered;
 243          bool pixel_interlock_unordered;
 244          bool sample_interlock_ordered;
 245          bool sample_interlock_unordered;
 246
 247          /**
 248           * Flags whether NIR's base types on the FS color outputs should be
 249           * ignored.
 250           *
 251           * GLSL requires that fragment shader output base types match the
 252           * render target's base types for the behavior to be defined.  From
 253           * the GL 4.6 spec:
 254           *
 255           *     "If the values written by the fragment shader do not match the
 256           *      format(s) of the corresponding color buffer(s), the result is
 257           *      undefined."
 258           *
 259           * However, for NIR shaders translated from TGSI, we don't have the
 260           * output types any more, so the driver will need to do whatever
 261           * fixups are necessary to handle effectively untyped data being
 262           * output from the FS.
 263           */
 264          bool untyped_color_outputs;
 265
 266          /** gl_FragDepth layout for ARB_conservative_depth. */
 267          enum gl_frag_depth_layout depth_layout;
 268       } fs;
 269
 270       struct {
 271          unsigned local_size[3];
 272
 273          bool local_size_variable;
 274          char user_data_components_amd;
 275
 276          /**
 277           * Size of shared variables accessed by the compute shader.
 278           */
 279          unsigned shared_size;
 280
 281
 282          /**
 283           * pointer size is:
 284           *   AddressingModelLogical:    0    (default)
 285           *   AddressingModelPhysical32: 32
 286           *   AddressingModelPhysical64: 64
 287           */
 288          unsigned ptr_size;
 289
 290          /*
 291           * Arrangement of invocations used to calculate derivatives in a compute
 292           * shader.  From NV_compute_shader_derivatives.
 293           */
 294          enum gl_derivative_group derivative_group;
 295       } cs;
 296
 297       /* Applies to both TCS and TES. */
 298       struct {
 299          /** The number of vertices in the TCS output patch. */
 300          unsigned tcs_vertices_out;
 301
 302          uint32_t primitive_mode; /* GL_TRIANGLES, GL_QUADS or GL_ISOLINES */
 303          enum gl_tess_spacing spacing;
 304          /** Is the vertex order counterclockwise? */
 305          bool ccw;
 306          bool point_mode;
 307       } tess;
 308    };
 309 } shader_info;
 310
 311 #ifdef __cplusplus
 312 }
 313 #endif
 314
 315 #endif /* SHADER_INFO_H */