X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fglsl%2Fbuiltin_functions.cpp;h=9973a7630870bdaea24d24c421df1e23c02fc93a;hb=f821a3ec4f8dda55722f326f26855c2b24ca186c;hp=72054e0fe93c88f5248a04c55885f5987b405e85;hpb=10ef949424809d51c627008bb2feab5a067f8e08;p=mesa.git diff --git a/src/glsl/builtin_functions.cpp b/src/glsl/builtin_functions.cpp index 72054e0fe93..9973a763087 100644 --- a/src/glsl/builtin_functions.cpp +++ b/src/glsl/builtin_functions.cpp @@ -60,7 +60,11 @@ #include "ir_builder.h" #include "glsl_parser_extras.h" #include "program/prog_instruction.h" -#include +#include + +#define M_PIf ((float) M_PI) +#define M_PI_2f ((float) M_PI_2) +#define M_PI_4f ((float) M_PI_4) using namespace ir_builder; @@ -69,7 +73,7 @@ using namespace ir_builder; * @{ */ static bool -always_available(const _mesa_glsl_parse_state *state) +always_available(const _mesa_glsl_parse_state *) { return true; } @@ -77,7 +81,7 @@ always_available(const _mesa_glsl_parse_state *state) static bool compatibility_vs_only(const _mesa_glsl_parse_state *state) { - return state->target == vertex_shader && + return state->stage == MESA_SHADER_VERTEX && state->language_version <= 130 && !state->es_shader; } @@ -85,13 +89,13 @@ compatibility_vs_only(const _mesa_glsl_parse_state *state) static bool fs_only(const _mesa_glsl_parse_state *state) { - return state->target == fragment_shader; + return state->stage == MESA_SHADER_FRAGMENT; } static bool gs_only(const _mesa_glsl_parse_state *state) { - return state->target == geometry_shader; + return state->stage == MESA_SHADER_GEOMETRY; } static bool @@ -103,7 +107,7 @@ v110(const _mesa_glsl_parse_state *state) static bool v110_fs_only(const _mesa_glsl_parse_state *state) { - return !state->es_shader && state->target == fragment_shader; + return !state->es_shader && state->stage == MESA_SHADER_FRAGMENT; } static bool @@ -122,7 +126,7 @@ static bool v130_fs_only(const _mesa_glsl_parse_state *state) { return state->is_version(130, 300) && - state->target == fragment_shader; + state->stage == MESA_SHADER_FRAGMENT; } static bool @@ -131,6 +135,19 @@ v140(const _mesa_glsl_parse_state *state) return state->is_version(140, 0); } +static bool +v400_fs_only(const _mesa_glsl_parse_state *state) +{ + return state->is_version(400, 0) && + state->stage == MESA_SHADER_FRAGMENT; +} + +static bool +es31(const _mesa_glsl_parse_state *state) +{ + return state->is_version(0, 310); +} + static bool texture_rectangle(const _mesa_glsl_parse_state *state) { @@ -155,7 +172,7 @@ lod_exists_in_stage(const _mesa_glsl_parse_state *state) * Since ARB_shader_texture_lod can only be enabled on desktop GLSL, we * don't need to explicitly check state->es_shader. */ - return state->target == vertex_shader || + return state->stage == MESA_SHADER_VERTEX || state->is_version(130, 300) || state->ARB_shader_texture_lod_enable; } @@ -190,20 +207,22 @@ shader_bit_encoding(const _mesa_glsl_parse_state *state) static bool shader_integer_mix(const _mesa_glsl_parse_state *state) { - return v130(state) && state->EXT_shader_integer_mix_enable; + return state->is_version(450, 310) || + (v130(state) && state->EXT_shader_integer_mix_enable); } static bool -shader_packing(const _mesa_glsl_parse_state *state) +shader_packing_or_es3(const _mesa_glsl_parse_state *state) { return state->ARB_shading_language_packing_enable || - state->is_version(400, 0); + state->is_version(420, 300); } static bool -shader_packing_or_es3(const _mesa_glsl_parse_state *state) +shader_packing_or_es3_or_gpu_shader5(const _mesa_glsl_parse_state *state) { return state->ARB_shading_language_packing_enable || + state->ARB_gpu_shader5_enable || state->is_version(400, 300); } @@ -213,6 +232,28 @@ gpu_shader5(const _mesa_glsl_parse_state *state) return state->is_version(400, 0) || state->ARB_gpu_shader5_enable; } +static bool +gpu_shader5_or_es31(const _mesa_glsl_parse_state *state) +{ + return state->is_version(400, 310) || state->ARB_gpu_shader5_enable; +} + +static bool +shader_packing_or_es31_or_gpu_shader5(const _mesa_glsl_parse_state *state) +{ + return state->ARB_shading_language_packing_enable || + state->ARB_gpu_shader5_enable || + state->is_version(400, 310); +} + +static bool +fs_gpu_shader5(const _mesa_glsl_parse_state *state) +{ + return state->stage == MESA_SHADER_FRAGMENT && + (state->is_version(400, 0) || state->ARB_gpu_shader5_enable); +} + + static bool texture_array_lod(const _mesa_glsl_parse_state *state) { @@ -223,7 +264,7 @@ texture_array_lod(const _mesa_glsl_parse_state *state) static bool fs_texture_array(const _mesa_glsl_parse_state *state) { - return state->target == fragment_shader && + return state->stage == MESA_SHADER_FRAGMENT && state->EXT_texture_array_enable; } @@ -236,14 +277,36 @@ texture_array(const _mesa_glsl_parse_state *state) static bool texture_multisample(const _mesa_glsl_parse_state *state) { - return state->is_version(150, 0) || + return state->is_version(150, 310) || state->ARB_texture_multisample_enable; } +static bool +texture_multisample_array(const _mesa_glsl_parse_state *state) +{ + return state->is_version(150, 320) || + state->ARB_texture_multisample_enable || + state->OES_texture_storage_multisample_2d_array_enable; +} + +static bool +texture_samples_identical(const _mesa_glsl_parse_state *state) +{ + return texture_multisample(state) && + state->EXT_shader_samples_identical_enable; +} + +static bool +texture_samples_identical_array(const _mesa_glsl_parse_state *state) +{ + return texture_multisample_array(state) && + state->EXT_shader_samples_identical_enable; +} + static bool fs_texture_cube_map_array(const _mesa_glsl_parse_state *state) { - return state->target == fragment_shader && + return state->stage == MESA_SHADER_FRAGMENT && (state->is_version(400, 0) || state->ARB_texture_cube_map_array_enable); } @@ -255,19 +318,63 @@ texture_cube_map_array(const _mesa_glsl_parse_state *state) state->ARB_texture_cube_map_array_enable; } +static bool +texture_query_levels(const _mesa_glsl_parse_state *state) +{ + return state->is_version(430, 0) || + state->ARB_texture_query_levels_enable; +} + static bool texture_query_lod(const _mesa_glsl_parse_state *state) { - return state->target == fragment_shader && + return state->stage == MESA_SHADER_FRAGMENT && state->ARB_texture_query_lod_enable; } +static bool +texture_gather(const _mesa_glsl_parse_state *state) +{ + return state->is_version(400, 0) || + state->ARB_texture_gather_enable || + state->ARB_gpu_shader5_enable; +} + +static bool +texture_gather_or_es31(const _mesa_glsl_parse_state *state) +{ + return state->is_version(400, 310) || + state->ARB_texture_gather_enable || + state->ARB_gpu_shader5_enable; +} + +/* Only ARB_texture_gather but not GLSL 4.0 or ARB_gpu_shader5. + * used for relaxation of const offset requirements. + */ +static bool +texture_gather_only_or_es31(const _mesa_glsl_parse_state *state) +{ + return !state->is_version(400, 0) && + !state->ARB_gpu_shader5_enable && + (state->ARB_texture_gather_enable || + state->is_version(0, 310)); +} + /* Desktop GL or OES_standard_derivatives + fragment shader only */ static bool fs_oes_derivatives(const _mesa_glsl_parse_state *state) { - return state->target == fragment_shader && - (!state->es_shader || state->OES_standard_derivatives_enable); + return state->stage == MESA_SHADER_FRAGMENT && + (state->is_version(110, 300) || + state->OES_standard_derivatives_enable); +} + +static bool +fs_derivative_control(const _mesa_glsl_parse_state *state) +{ + return state->stage == MESA_SHADER_FRAGMENT && + (state->is_version(450, 0) || + state->ARB_derivative_control_enable); } static bool @@ -291,7 +398,7 @@ tex3d(const _mesa_glsl_parse_state *state) static bool fs_tex3d(const _mesa_glsl_parse_state *state) { - return state->target == fragment_shader && + return state->stage == MESA_SHADER_FRAGMENT && (!state->es_shader || state->OES_texture_3D_enable); } @@ -300,6 +407,90 @@ tex3d_lod(const _mesa_glsl_parse_state *state) { return tex3d(state) && lod_exists_in_stage(state); } + +static bool +shader_atomic_counters(const _mesa_glsl_parse_state *state) +{ + return state->has_atomic_counters(); +} + +static bool +shader_clock(const _mesa_glsl_parse_state *state) +{ + return state->ARB_shader_clock_enable; +} + +static bool +shader_storage_buffer_object(const _mesa_glsl_parse_state *state) +{ + return state->has_shader_storage_buffer_objects(); +} + +static bool +shader_trinary_minmax(const _mesa_glsl_parse_state *state) +{ + return state->AMD_shader_trinary_minmax_enable; +} + +static bool +shader_image_load_store(const _mesa_glsl_parse_state *state) +{ + return (state->is_version(420, 310) || + state->ARB_shader_image_load_store_enable); +} + +static bool +shader_image_atomic(const _mesa_glsl_parse_state *state) +{ + return (state->is_version(420, 0) || + state->ARB_shader_image_load_store_enable); +} + +static bool +shader_image_size(const _mesa_glsl_parse_state *state) +{ + return state->is_version(430, 310) || + state->ARB_shader_image_size_enable; +} + +static bool +shader_samples(const _mesa_glsl_parse_state *state) +{ + return state->is_version(450, 0) || + state->ARB_shader_texture_image_samples_enable; +} + +static bool +gs_streams(const _mesa_glsl_parse_state *state) +{ + return gpu_shader5(state) && gs_only(state); +} + +static bool +fp64(const _mesa_glsl_parse_state *state) +{ + return state->has_double(); +} + +static bool +compute_shader(const _mesa_glsl_parse_state *state) +{ + return state->stage == MESA_SHADER_COMPUTE; +} + +static bool +buffer_atomics_supported(const _mesa_glsl_parse_state *state) +{ + return compute_shader(state) || shader_storage_buffer_object(state); +} + +static bool +barrier_supported(const _mesa_glsl_parse_state *state) +{ + return compute_shader(state) || + state->stage == MESA_SHADER_TESS_CTRL; +} + /** @} */ /******************************************************************************/ @@ -323,8 +514,6 @@ public: ir_function_signature *find(_mesa_glsl_parse_state *state, const char *name, exec_list *actual_parameters); -private: - void *mem_ctx; /** * A shader to hold all the built-in signatures; created by this module. * @@ -334,11 +523,15 @@ private: */ gl_shader *shader; +private: + void *mem_ctx; + /** Global variables used by built-in functions. */ ir_variable *gl_ModelViewProjectionMatrix; ir_variable *gl_Vertex; void create_shader(); + void create_intrinsics(); void create_builtins(); /** @@ -353,16 +546,60 @@ private: ir_constant *imm(float f, unsigned vector_elements=1); ir_constant *imm(int i, unsigned vector_elements=1); ir_constant *imm(unsigned u, unsigned vector_elements=1); + ir_constant *imm(double d, unsigned vector_elements=1); ir_constant *imm(const glsl_type *type, const ir_constant_data &); ir_dereference_variable *var_ref(ir_variable *var); ir_dereference_array *array_ref(ir_variable *var, int i); ir_swizzle *matrix_elt(ir_variable *var, int col, int row); ir_expression *asin_expr(ir_variable *x); + void do_atan(ir_factory &body, const glsl_type *type, ir_variable *res, operand y_over_x); + + /** + * Call function \param f with parameters specified as the linked + * list \param params of \c ir_variable objects. \param ret should + * point to the ir_variable that will hold the function return + * value, or be \c NULL if the function has void return type. + */ + ir_call *call(ir_function *f, ir_variable *ret, exec_list params); /** Create a new function and add the given signatures. */ void add_function(const char *name, ...); + typedef ir_function_signature *(builtin_builder::*image_prototype_ctr)(const glsl_type *image_type, + unsigned num_arguments, + unsigned flags); + + enum image_function_flags { + IMAGE_FUNCTION_EMIT_STUB = (1 << 0), + IMAGE_FUNCTION_RETURNS_VOID = (1 << 1), + IMAGE_FUNCTION_HAS_VECTOR_DATA_TYPE = (1 << 2), + IMAGE_FUNCTION_SUPPORTS_FLOAT_DATA_TYPE = (1 << 3), + IMAGE_FUNCTION_READ_ONLY = (1 << 4), + IMAGE_FUNCTION_WRITE_ONLY = (1 << 5), + IMAGE_FUNCTION_AVAIL_ATOMIC = (1 << 6), + IMAGE_FUNCTION_MS_ONLY = (1 << 7), + }; + + /** + * Create a new image built-in function for all known image types. + * \p flags is a bitfield of \c image_function_flags flags. + */ + void add_image_function(const char *name, + const char *intrinsic_name, + image_prototype_ctr prototype, + unsigned num_arguments, + unsigned flags); + + /** + * Create new functions for all known image built-ins and types. + * If \p glsl is \c true, use the GLSL built-in names and emit code + * to call into the actual compiler intrinsic. If \p glsl is + * false, emit a function prototype with no body for each image + * intrinsic name. + */ + void add_image_functions(bool glsl); + ir_function_signature *new_sig(const glsl_type *return_type, builtin_available_predicate avail, int num_params, ...); @@ -375,8 +612,8 @@ private: ir_expression_operation opcode, const glsl_type *return_type, const glsl_type *param_type); - ir_function_signature *binop(ir_expression_operation opcode, - builtin_available_predicate avail, + ir_function_signature *binop(builtin_available_predicate avail, + ir_expression_operation opcode, const glsl_type *return_type, const glsl_type *param0_type, const glsl_type *param1_type); @@ -407,29 +644,29 @@ private: B1(log) B1(exp2) B1(log2) - B1(sqrt) - B1(inversesqrt) - B1(abs) - B1(sign) - B1(floor) - B1(trunc) - B1(round) - B1(roundEven) - B1(ceil) - B1(fract) + BA1(sqrt) + BA1(inversesqrt) + BA1(abs) + BA1(sign) + BA1(floor) + BA1(trunc) + BA1(round) + BA1(roundEven) + BA1(ceil) + BA1(fract) B2(mod) - B1(modf) + BA1(modf) BA2(min) BA2(max) BA2(clamp) - B2(mix_lrp) + BA2(mix_lrp) ir_function_signature *_mix_sel(builtin_available_predicate avail, const glsl_type *val_type, const glsl_type *blend_type); - B2(step) - B2(smoothstep) - B1(isnan) - B1(isinf) + BA2(step) + BA2(smoothstep) + BA1(isnan) + BA1(isinf) B1(floatBitsToInt) B1(floatBitsToUint) B1(intBitsToFloat) @@ -444,24 +681,27 @@ private: ir_function_signature *_unpackSnorm4x8(builtin_available_predicate avail); ir_function_signature *_packHalf2x16(builtin_available_predicate avail); ir_function_signature *_unpackHalf2x16(builtin_available_predicate avail); - B1(length) - B1(distance); - B1(dot); - B1(cross); - B1(normalize); + ir_function_signature *_packDouble2x32(builtin_available_predicate avail); + ir_function_signature *_unpackDouble2x32(builtin_available_predicate avail); + + BA1(length) + BA1(distance); + BA1(dot); + BA1(cross); + BA1(normalize); B0(ftransform); - B1(faceforward); - B1(reflect); - B1(refract); - B1(matrixCompMult); - B1(outerProduct); - B0(determinant_mat2); - B0(determinant_mat3); - B0(determinant_mat4); - B0(inverse_mat2); - B0(inverse_mat3); - B0(inverse_mat4); - B1(transpose); + BA1(faceforward); + BA1(reflect); + BA1(refract); + BA1(matrixCompMult); + BA1(outerProduct); + BA1(determinant_mat2); + BA1(determinant_mat3); + BA1(determinant_mat4); + BA1(inverse_mat2); + BA1(inverse_mat3); + BA1(inverse_mat4); + BA1(transpose); BA1(lessThan); BA1(lessThanEqual); BA1(greaterThan); @@ -471,14 +711,15 @@ private: B1(any); B1(all); B1(not); - B2(textureSize); - ir_function_signature *_textureSize(builtin_available_predicate avail, - const glsl_type *return_type, - const glsl_type *sampler_type); + BA2(textureSize); + B1(textureSamples); /** Flags to _texture() */ #define TEX_PROJECT 1 #define TEX_OFFSET 2 +#define TEX_COMPONENT 4 +#define TEX_OFFSET_NONCONST 8 +#define TEX_OFFSET_ARRAY 16 ir_function_signature *_texture(ir_texture_opcode opcode, builtin_available_predicate avail, @@ -495,11 +736,24 @@ private: B0(EmitVertex) B0(EndPrimitive) - - B2(textureQueryLod); + ir_function_signature *_EmitStreamVertex(builtin_available_predicate avail, + const glsl_type *stream_type); + ir_function_signature *_EndStreamPrimitive(builtin_available_predicate avail, + const glsl_type *stream_type); + B0(barrier) + + BA2(textureQueryLod); + B1(textureQueryLevels); + BA2(textureSamplesIdentical); B1(dFdx); B1(dFdy); B1(fwidth); + B1(dFdxCoarse); + B1(dFdyCoarse); + B1(fwidthCoarse); + B1(dFdxFine); + B1(dFdyFine); + B1(fwidthFine); B1(noise1); B1(noise2); B1(noise3); @@ -511,9 +765,61 @@ private: B1(bitCount) B1(findLSB) B1(findMSB) - B1(fma) + BA1(fma) B2(ldexp) B2(frexp) + B2(dfrexp) + B1(uaddCarry) + B1(usubBorrow) + B1(mulExtended) + B1(interpolateAtCentroid) + B1(interpolateAtOffset) + B1(interpolateAtSample) + + ir_function_signature *_atomic_counter_intrinsic(builtin_available_predicate avail); + ir_function_signature *_atomic_counter_op(const char *intrinsic, + builtin_available_predicate avail); + + ir_function_signature *_atomic_intrinsic2(builtin_available_predicate avail, + const glsl_type *type); + ir_function_signature *_atomic_op2(const char *intrinsic, + builtin_available_predicate avail, + const glsl_type *type); + ir_function_signature *_atomic_intrinsic3(builtin_available_predicate avail, + const glsl_type *type); + ir_function_signature *_atomic_op3(const char *intrinsic, + builtin_available_predicate avail, + const glsl_type *type); + + B1(min3) + B1(max3) + B1(mid3) + + ir_function_signature *_image_prototype(const glsl_type *image_type, + unsigned num_arguments, + unsigned flags); + ir_function_signature *_image_size_prototype(const glsl_type *image_type, + unsigned num_arguments, + unsigned flags); + ir_function_signature *_image_samples_prototype(const glsl_type *image_type, + unsigned num_arguments, + unsigned flags); + ir_function_signature *_image(image_prototype_ctr prototype, + const glsl_type *image_type, + const char *intrinsic_name, + unsigned num_arguments, + unsigned flags); + + ir_function_signature *_memory_barrier_intrinsic( + builtin_available_predicate avail); + ir_function_signature *_memory_barrier(const char *intrinsic_name, + builtin_available_predicate avail); + + ir_function_signature *_shader_clock_intrinsic(builtin_available_predicate avail, + const glsl_type *type); + ir_function_signature *_shader_clock(builtin_available_predicate avail, + const glsl_type *type); + #undef B0 #undef B1 #undef B2 @@ -553,14 +859,14 @@ builtin_builder::find(_mesa_glsl_parse_state *state, * that the "no matching signature" error will list potential candidates * from the available built-ins. */ - state->builtins_to_link[0] = shader; - state->num_builtins_to_link = 1; + state->uses_builtin_functions = true; ir_function *f = shader->symbols->get_function(name); if (f == NULL) return NULL; - ir_function_signature *sig = f->matching_signature(state, actual_parameters); + ir_function_signature *sig = + f->matching_signature(state, actual_parameters, true); if (sig == NULL) return NULL; @@ -576,6 +882,7 @@ builtin_builder::initialize() mem_ctx = ralloc_context(NULL); create_shader(); + create_intrinsics(); create_builtins(); } @@ -612,6 +919,99 @@ builtin_builder::create_shader() /** @} */ +/** + * Create ir_function and ir_function_signature objects for each + * intrinsic. + */ +void +builtin_builder::create_intrinsics() +{ + add_function("__intrinsic_atomic_read", + _atomic_counter_intrinsic(shader_atomic_counters), + NULL); + add_function("__intrinsic_atomic_increment", + _atomic_counter_intrinsic(shader_atomic_counters), + NULL); + add_function("__intrinsic_atomic_predecrement", + _atomic_counter_intrinsic(shader_atomic_counters), + NULL); + + add_function("__intrinsic_atomic_add", + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::uint_type), + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("__intrinsic_atomic_min", + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::uint_type), + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("__intrinsic_atomic_max", + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::uint_type), + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("__intrinsic_atomic_and", + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::uint_type), + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("__intrinsic_atomic_or", + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::uint_type), + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("__intrinsic_atomic_xor", + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::uint_type), + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("__intrinsic_atomic_exchange", + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::uint_type), + _atomic_intrinsic2(buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("__intrinsic_atomic_comp_swap", + _atomic_intrinsic3(buffer_atomics_supported, + glsl_type::uint_type), + _atomic_intrinsic3(buffer_atomics_supported, + glsl_type::int_type), + NULL); + + add_image_functions(false); + + add_function("__intrinsic_memory_barrier", + _memory_barrier_intrinsic(shader_image_load_store), + NULL); + add_function("__intrinsic_group_memory_barrier", + _memory_barrier_intrinsic(compute_shader), + NULL); + add_function("__intrinsic_memory_barrier_atomic_counter", + _memory_barrier_intrinsic(compute_shader), + NULL); + add_function("__intrinsic_memory_barrier_buffer", + _memory_barrier_intrinsic(compute_shader), + NULL); + add_function("__intrinsic_memory_barrier_image", + _memory_barrier_intrinsic(compute_shader), + NULL); + add_function("__intrinsic_memory_barrier_shared", + _memory_barrier_intrinsic(compute_shader), + NULL); + + add_function("__intrinsic_shader_clock", + _shader_clock_intrinsic(shader_clock, + glsl_type::uvec2_type), + NULL); +} + /** * Create ir_function and ir_function_signature objects for each built-in. * @@ -628,6 +1028,42 @@ builtin_builder::create_builtins() _##NAME(glsl_type::vec4_type), \ NULL); +#define FD(NAME) \ + add_function(#NAME, \ + _##NAME(always_available, glsl_type::float_type), \ + _##NAME(always_available, glsl_type::vec2_type), \ + _##NAME(always_available, glsl_type::vec3_type), \ + _##NAME(always_available, glsl_type::vec4_type), \ + _##NAME(fp64, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec2_type), \ + _##NAME(fp64, glsl_type::dvec3_type), \ + _##NAME(fp64, glsl_type::dvec4_type), \ + NULL); + +#define FD130(NAME) \ + add_function(#NAME, \ + _##NAME(v130, glsl_type::float_type), \ + _##NAME(v130, glsl_type::vec2_type), \ + _##NAME(v130, glsl_type::vec3_type), \ + _##NAME(v130, glsl_type::vec4_type), \ + _##NAME(fp64, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec2_type), \ + _##NAME(fp64, glsl_type::dvec3_type), \ + _##NAME(fp64, glsl_type::dvec4_type), \ + NULL); + +#define FDGS5(NAME) \ + add_function(#NAME, \ + _##NAME(gpu_shader5, glsl_type::float_type), \ + _##NAME(gpu_shader5, glsl_type::vec2_type), \ + _##NAME(gpu_shader5, glsl_type::vec3_type), \ + _##NAME(gpu_shader5, glsl_type::vec4_type), \ + _##NAME(fp64, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec2_type), \ + _##NAME(fp64, glsl_type::dvec3_type), \ + _##NAME(fp64, glsl_type::dvec4_type), \ + NULL); + #define FI(NAME) \ add_function(#NAME, \ _##NAME(glsl_type::float_type), \ @@ -640,7 +1076,23 @@ builtin_builder::create_builtins() _##NAME(glsl_type::ivec4_type), \ NULL); -#define FIU(NAME) \ +#define FID(NAME) \ + add_function(#NAME, \ + _##NAME(always_available, glsl_type::float_type), \ + _##NAME(always_available, glsl_type::vec2_type), \ + _##NAME(always_available, glsl_type::vec3_type), \ + _##NAME(always_available, glsl_type::vec4_type), \ + _##NAME(always_available, glsl_type::int_type), \ + _##NAME(always_available, glsl_type::ivec2_type), \ + _##NAME(always_available, glsl_type::ivec3_type), \ + _##NAME(always_available, glsl_type::ivec4_type), \ + _##NAME(fp64, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec2_type), \ + _##NAME(fp64, glsl_type::dvec3_type), \ + _##NAME(fp64, glsl_type::dvec4_type), \ + NULL); + +#define FIUD(NAME) \ add_function(#NAME, \ _##NAME(always_available, glsl_type::float_type), \ _##NAME(always_available, glsl_type::vec2_type), \ @@ -656,6 +1108,10 @@ builtin_builder::create_builtins() _##NAME(v130, glsl_type::uvec2_type), \ _##NAME(v130, glsl_type::uvec3_type), \ _##NAME(v130, glsl_type::uvec4_type), \ + _##NAME(fp64, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec2_type), \ + _##NAME(fp64, glsl_type::dvec3_type), \ + _##NAME(fp64, glsl_type::dvec4_type), \ NULL); #define IU(NAME) \ @@ -671,7 +1127,7 @@ builtin_builder::create_builtins() _##NAME(glsl_type::uvec4_type), \ NULL); -#define FIUB(NAME) \ +#define FIUBD(NAME) \ add_function(#NAME, \ _##NAME(always_available, glsl_type::float_type), \ _##NAME(always_available, glsl_type::vec2_type), \ @@ -692,9 +1148,14 @@ builtin_builder::create_builtins() _##NAME(always_available, glsl_type::bvec2_type), \ _##NAME(always_available, glsl_type::bvec3_type), \ _##NAME(always_available, glsl_type::bvec4_type), \ + \ + _##NAME(fp64, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec2_type), \ + _##NAME(fp64, glsl_type::dvec3_type), \ + _##NAME(fp64, glsl_type::dvec4_type), \ NULL); -#define FIU2_MIXED(NAME) \ +#define FIUD2_MIXED(NAME) \ add_function(#NAME, \ _##NAME(always_available, glsl_type::float_type, glsl_type::float_type), \ _##NAME(always_available, glsl_type::vec2_type, glsl_type::float_type), \ @@ -722,6 +1183,14 @@ builtin_builder::create_builtins() _##NAME(v130, glsl_type::uvec2_type, glsl_type::uvec2_type), \ _##NAME(v130, glsl_type::uvec3_type, glsl_type::uvec3_type), \ _##NAME(v130, glsl_type::uvec4_type, glsl_type::uvec4_type), \ + \ + _##NAME(fp64, glsl_type::double_type, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec2_type, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec3_type, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec4_type, glsl_type::double_type), \ + _##NAME(fp64, glsl_type::dvec2_type, glsl_type::dvec2_type), \ + _##NAME(fp64, glsl_type::dvec3_type, glsl_type::dvec3_type), \ + _##NAME(fp64, glsl_type::dvec4_type, glsl_type::dvec4_type), \ NULL); F(radians) @@ -754,16 +1223,16 @@ builtin_builder::create_builtins() F(log) F(exp2) F(log2) - F(sqrt) - F(inversesqrt) - FI(abs) - FI(sign) - F(floor) - F(trunc) - F(round) - F(roundEven) - F(ceil) - F(fract) + FD(sqrt) + FD(inversesqrt) + FID(abs) + FID(sign) + FD(floor) + FD(trunc) + FD(round) + FD(roundEven) + FD(ceil) + FD(fract) add_function("mod", _mod(glsl_type::float_type, glsl_type::float_type), @@ -774,29 +1243,52 @@ builtin_builder::create_builtins() _mod(glsl_type::vec2_type, glsl_type::vec2_type), _mod(glsl_type::vec3_type, glsl_type::vec3_type), _mod(glsl_type::vec4_type, glsl_type::vec4_type), + + _mod(glsl_type::double_type, glsl_type::double_type), + _mod(glsl_type::dvec2_type, glsl_type::double_type), + _mod(glsl_type::dvec3_type, glsl_type::double_type), + _mod(glsl_type::dvec4_type, glsl_type::double_type), + + _mod(glsl_type::dvec2_type, glsl_type::dvec2_type), + _mod(glsl_type::dvec3_type, glsl_type::dvec3_type), + _mod(glsl_type::dvec4_type, glsl_type::dvec4_type), NULL); - F(modf) + FD(modf) - FIU2_MIXED(min) - FIU2_MIXED(max) - FIU2_MIXED(clamp) + FIUD2_MIXED(min) + FIUD2_MIXED(max) + FIUD2_MIXED(clamp) add_function("mix", - _mix_lrp(glsl_type::float_type, glsl_type::float_type), - _mix_lrp(glsl_type::vec2_type, glsl_type::float_type), - _mix_lrp(glsl_type::vec3_type, glsl_type::float_type), - _mix_lrp(glsl_type::vec4_type, glsl_type::float_type), + _mix_lrp(always_available, glsl_type::float_type, glsl_type::float_type), + _mix_lrp(always_available, glsl_type::vec2_type, glsl_type::float_type), + _mix_lrp(always_available, glsl_type::vec3_type, glsl_type::float_type), + _mix_lrp(always_available, glsl_type::vec4_type, glsl_type::float_type), - _mix_lrp(glsl_type::vec2_type, glsl_type::vec2_type), - _mix_lrp(glsl_type::vec3_type, glsl_type::vec3_type), - _mix_lrp(glsl_type::vec4_type, glsl_type::vec4_type), + _mix_lrp(always_available, glsl_type::vec2_type, glsl_type::vec2_type), + _mix_lrp(always_available, glsl_type::vec3_type, glsl_type::vec3_type), + _mix_lrp(always_available, glsl_type::vec4_type, glsl_type::vec4_type), + + _mix_lrp(fp64, glsl_type::double_type, glsl_type::double_type), + _mix_lrp(fp64, glsl_type::dvec2_type, glsl_type::double_type), + _mix_lrp(fp64, glsl_type::dvec3_type, glsl_type::double_type), + _mix_lrp(fp64, glsl_type::dvec4_type, glsl_type::double_type), + + _mix_lrp(fp64, glsl_type::dvec2_type, glsl_type::dvec2_type), + _mix_lrp(fp64, glsl_type::dvec3_type, glsl_type::dvec3_type), + _mix_lrp(fp64, glsl_type::dvec4_type, glsl_type::dvec4_type), _mix_sel(v130, glsl_type::float_type, glsl_type::bool_type), _mix_sel(v130, glsl_type::vec2_type, glsl_type::bvec2_type), _mix_sel(v130, glsl_type::vec3_type, glsl_type::bvec3_type), _mix_sel(v130, glsl_type::vec4_type, glsl_type::bvec4_type), + _mix_sel(fp64, glsl_type::double_type, glsl_type::bool_type), + _mix_sel(fp64, glsl_type::dvec2_type, glsl_type::bvec2_type), + _mix_sel(fp64, glsl_type::dvec3_type, glsl_type::bvec3_type), + _mix_sel(fp64, glsl_type::dvec4_type, glsl_type::bvec4_type), + _mix_sel(shader_integer_mix, glsl_type::int_type, glsl_type::bool_type), _mix_sel(shader_integer_mix, glsl_type::ivec2_type, glsl_type::bvec2_type), _mix_sel(shader_integer_mix, glsl_type::ivec3_type, glsl_type::bvec3_type), @@ -814,29 +1306,45 @@ builtin_builder::create_builtins() NULL); add_function("step", - _step(glsl_type::float_type, glsl_type::float_type), - _step(glsl_type::float_type, glsl_type::vec2_type), - _step(glsl_type::float_type, glsl_type::vec3_type), - _step(glsl_type::float_type, glsl_type::vec4_type), - - _step(glsl_type::vec2_type, glsl_type::vec2_type), - _step(glsl_type::vec3_type, glsl_type::vec3_type), - _step(glsl_type::vec4_type, glsl_type::vec4_type), + _step(always_available, glsl_type::float_type, glsl_type::float_type), + _step(always_available, glsl_type::float_type, glsl_type::vec2_type), + _step(always_available, glsl_type::float_type, glsl_type::vec3_type), + _step(always_available, glsl_type::float_type, glsl_type::vec4_type), + + _step(always_available, glsl_type::vec2_type, glsl_type::vec2_type), + _step(always_available, glsl_type::vec3_type, glsl_type::vec3_type), + _step(always_available, glsl_type::vec4_type, glsl_type::vec4_type), + _step(fp64, glsl_type::double_type, glsl_type::double_type), + _step(fp64, glsl_type::double_type, glsl_type::dvec2_type), + _step(fp64, glsl_type::double_type, glsl_type::dvec3_type), + _step(fp64, glsl_type::double_type, glsl_type::dvec4_type), + + _step(fp64, glsl_type::dvec2_type, glsl_type::dvec2_type), + _step(fp64, glsl_type::dvec3_type, glsl_type::dvec3_type), + _step(fp64, glsl_type::dvec4_type, glsl_type::dvec4_type), NULL); add_function("smoothstep", - _smoothstep(glsl_type::float_type, glsl_type::float_type), - _smoothstep(glsl_type::float_type, glsl_type::vec2_type), - _smoothstep(glsl_type::float_type, glsl_type::vec3_type), - _smoothstep(glsl_type::float_type, glsl_type::vec4_type), - - _smoothstep(glsl_type::vec2_type, glsl_type::vec2_type), - _smoothstep(glsl_type::vec3_type, glsl_type::vec3_type), - _smoothstep(glsl_type::vec4_type, glsl_type::vec4_type), + _smoothstep(always_available, glsl_type::float_type, glsl_type::float_type), + _smoothstep(always_available, glsl_type::float_type, glsl_type::vec2_type), + _smoothstep(always_available, glsl_type::float_type, glsl_type::vec3_type), + _smoothstep(always_available, glsl_type::float_type, glsl_type::vec4_type), + + _smoothstep(always_available, glsl_type::vec2_type, glsl_type::vec2_type), + _smoothstep(always_available, glsl_type::vec3_type, glsl_type::vec3_type), + _smoothstep(always_available, glsl_type::vec4_type, glsl_type::vec4_type), + _smoothstep(fp64, glsl_type::double_type, glsl_type::double_type), + _smoothstep(fp64, glsl_type::double_type, glsl_type::dvec2_type), + _smoothstep(fp64, glsl_type::double_type, glsl_type::dvec3_type), + _smoothstep(fp64, glsl_type::double_type, glsl_type::dvec4_type), + + _smoothstep(fp64, glsl_type::dvec2_type, glsl_type::dvec2_type), + _smoothstep(fp64, glsl_type::dvec3_type, glsl_type::dvec3_type), + _smoothstep(fp64, glsl_type::dvec4_type, glsl_type::dvec4_type), NULL); - - F(isnan) - F(isinf) + + FD130(isnan) + FD130(isinf) F(floatBitsToInt) F(floatBitsToUint) @@ -853,78 +1361,116 @@ builtin_builder::create_builtins() _uintBitsToFloat(glsl_type::uvec4_type), NULL); - add_function("packUnorm2x16", _packUnorm2x16(shader_packing_or_es3), NULL); - add_function("packSnorm2x16", _packSnorm2x16(shader_packing_or_es3), NULL); - add_function("packUnorm4x8", _packUnorm4x8(shader_packing), NULL); - add_function("packSnorm4x8", _packSnorm4x8(shader_packing), NULL); - add_function("unpackUnorm2x16", _unpackUnorm2x16(shader_packing_or_es3), NULL); - add_function("unpackSnorm2x16", _unpackSnorm2x16(shader_packing_or_es3), NULL); - add_function("unpackUnorm4x8", _unpackUnorm4x8(shader_packing), NULL); - add_function("unpackSnorm4x8", _unpackSnorm4x8(shader_packing), NULL); - add_function("packHalf2x16", _packHalf2x16(shader_packing_or_es3), NULL); - add_function("unpackHalf2x16", _unpackHalf2x16(shader_packing_or_es3), NULL); + add_function("packUnorm2x16", _packUnorm2x16(shader_packing_or_es3_or_gpu_shader5), NULL); + add_function("packSnorm2x16", _packSnorm2x16(shader_packing_or_es3), NULL); + add_function("packUnorm4x8", _packUnorm4x8(shader_packing_or_es31_or_gpu_shader5), NULL); + add_function("packSnorm4x8", _packSnorm4x8(shader_packing_or_es31_or_gpu_shader5), NULL); + add_function("unpackUnorm2x16", _unpackUnorm2x16(shader_packing_or_es3_or_gpu_shader5), NULL); + add_function("unpackSnorm2x16", _unpackSnorm2x16(shader_packing_or_es3), NULL); + add_function("unpackUnorm4x8", _unpackUnorm4x8(shader_packing_or_es31_or_gpu_shader5), NULL); + add_function("unpackSnorm4x8", _unpackSnorm4x8(shader_packing_or_es31_or_gpu_shader5), NULL); + add_function("packHalf2x16", _packHalf2x16(shader_packing_or_es3), NULL); + add_function("unpackHalf2x16", _unpackHalf2x16(shader_packing_or_es3), NULL); + add_function("packDouble2x32", _packDouble2x32(fp64), NULL); + add_function("unpackDouble2x32", _unpackDouble2x32(fp64), NULL); - F(length) - F(distance) - F(dot) - add_function("cross", _cross(glsl_type::vec3_type), NULL); + FD(length) + FD(distance) + FD(dot) - F(normalize) + add_function("cross", _cross(always_available, glsl_type::vec3_type), + _cross(fp64, glsl_type::dvec3_type), NULL); + + FD(normalize) add_function("ftransform", _ftransform(), NULL); - F(faceforward) - F(reflect) - F(refract) + FD(faceforward) + FD(reflect) + FD(refract) // ... add_function("matrixCompMult", - _matrixCompMult(glsl_type::mat2_type), - _matrixCompMult(glsl_type::mat3_type), - _matrixCompMult(glsl_type::mat4_type), - _matrixCompMult(glsl_type::mat2x3_type), - _matrixCompMult(glsl_type::mat2x4_type), - _matrixCompMult(glsl_type::mat3x2_type), - _matrixCompMult(glsl_type::mat3x4_type), - _matrixCompMult(glsl_type::mat4x2_type), - _matrixCompMult(glsl_type::mat4x3_type), + _matrixCompMult(always_available, glsl_type::mat2_type), + _matrixCompMult(always_available, glsl_type::mat3_type), + _matrixCompMult(always_available, glsl_type::mat4_type), + _matrixCompMult(always_available, glsl_type::mat2x3_type), + _matrixCompMult(always_available, glsl_type::mat2x4_type), + _matrixCompMult(always_available, glsl_type::mat3x2_type), + _matrixCompMult(always_available, glsl_type::mat3x4_type), + _matrixCompMult(always_available, glsl_type::mat4x2_type), + _matrixCompMult(always_available, glsl_type::mat4x3_type), + _matrixCompMult(fp64, glsl_type::dmat2_type), + _matrixCompMult(fp64, glsl_type::dmat3_type), + _matrixCompMult(fp64, glsl_type::dmat4_type), + _matrixCompMult(fp64, glsl_type::dmat2x3_type), + _matrixCompMult(fp64, glsl_type::dmat2x4_type), + _matrixCompMult(fp64, glsl_type::dmat3x2_type), + _matrixCompMult(fp64, glsl_type::dmat3x4_type), + _matrixCompMult(fp64, glsl_type::dmat4x2_type), + _matrixCompMult(fp64, glsl_type::dmat4x3_type), NULL); add_function("outerProduct", - _outerProduct(glsl_type::mat2_type), - _outerProduct(glsl_type::mat3_type), - _outerProduct(glsl_type::mat4_type), - _outerProduct(glsl_type::mat2x3_type), - _outerProduct(glsl_type::mat2x4_type), - _outerProduct(glsl_type::mat3x2_type), - _outerProduct(glsl_type::mat3x4_type), - _outerProduct(glsl_type::mat4x2_type), - _outerProduct(glsl_type::mat4x3_type), + _outerProduct(v120, glsl_type::mat2_type), + _outerProduct(v120, glsl_type::mat3_type), + _outerProduct(v120, glsl_type::mat4_type), + _outerProduct(v120, glsl_type::mat2x3_type), + _outerProduct(v120, glsl_type::mat2x4_type), + _outerProduct(v120, glsl_type::mat3x2_type), + _outerProduct(v120, glsl_type::mat3x4_type), + _outerProduct(v120, glsl_type::mat4x2_type), + _outerProduct(v120, glsl_type::mat4x3_type), + _outerProduct(fp64, glsl_type::dmat2_type), + _outerProduct(fp64, glsl_type::dmat3_type), + _outerProduct(fp64, glsl_type::dmat4_type), + _outerProduct(fp64, glsl_type::dmat2x3_type), + _outerProduct(fp64, glsl_type::dmat2x4_type), + _outerProduct(fp64, glsl_type::dmat3x2_type), + _outerProduct(fp64, glsl_type::dmat3x4_type), + _outerProduct(fp64, glsl_type::dmat4x2_type), + _outerProduct(fp64, glsl_type::dmat4x3_type), NULL); add_function("determinant", - _determinant_mat2(), - _determinant_mat3(), - _determinant_mat4(), + _determinant_mat2(v120, glsl_type::mat2_type), + _determinant_mat3(v120, glsl_type::mat3_type), + _determinant_mat4(v120, glsl_type::mat4_type), + _determinant_mat2(fp64, glsl_type::dmat2_type), + _determinant_mat3(fp64, glsl_type::dmat3_type), + _determinant_mat4(fp64, glsl_type::dmat4_type), + NULL); add_function("inverse", - _inverse_mat2(), - _inverse_mat3(), - _inverse_mat4(), + _inverse_mat2(v120, glsl_type::mat2_type), + _inverse_mat3(v120, glsl_type::mat3_type), + _inverse_mat4(v120, glsl_type::mat4_type), + _inverse_mat2(fp64, glsl_type::dmat2_type), + _inverse_mat3(fp64, glsl_type::dmat3_type), + _inverse_mat4(fp64, glsl_type::dmat4_type), NULL); add_function("transpose", - _transpose(glsl_type::mat2_type), - _transpose(glsl_type::mat3_type), - _transpose(glsl_type::mat4_type), - _transpose(glsl_type::mat2x3_type), - _transpose(glsl_type::mat2x4_type), - _transpose(glsl_type::mat3x2_type), - _transpose(glsl_type::mat3x4_type), - _transpose(glsl_type::mat4x2_type), - _transpose(glsl_type::mat4x3_type), - NULL); - FIU(lessThan) - FIU(lessThanEqual) - FIU(greaterThan) - FIU(greaterThanEqual) - FIUB(notEqual) - FIUB(equal) + _transpose(v120, glsl_type::mat2_type), + _transpose(v120, glsl_type::mat3_type), + _transpose(v120, glsl_type::mat4_type), + _transpose(v120, glsl_type::mat2x3_type), + _transpose(v120, glsl_type::mat2x4_type), + _transpose(v120, glsl_type::mat3x2_type), + _transpose(v120, glsl_type::mat3x4_type), + _transpose(v120, glsl_type::mat4x2_type), + _transpose(v120, glsl_type::mat4x3_type), + _transpose(fp64, glsl_type::dmat2_type), + _transpose(fp64, glsl_type::dmat3_type), + _transpose(fp64, glsl_type::dmat4_type), + _transpose(fp64, glsl_type::dmat2x3_type), + _transpose(fp64, glsl_type::dmat2x4_type), + _transpose(fp64, glsl_type::dmat3x2_type), + _transpose(fp64, glsl_type::dmat3x4_type), + _transpose(fp64, glsl_type::dmat4x2_type), + _transpose(fp64, glsl_type::dmat4x3_type), + NULL); + FIUD(lessThan) + FIUD(lessThanEqual) + FIUD(greaterThan) + FIUD(greaterThanEqual) + FIUBD(notEqual) + FIUBD(equal) add_function("any", _any(glsl_type::bvec2_type), @@ -992,9 +1538,19 @@ builtin_builder::create_builtins() _textureSize(texture_multisample, glsl_type::ivec2_type, glsl_type::isampler2DMS_type), _textureSize(texture_multisample, glsl_type::ivec2_type, glsl_type::usampler2DMS_type), - _textureSize(texture_multisample, glsl_type::ivec3_type, glsl_type::sampler2DMSArray_type), - _textureSize(texture_multisample, glsl_type::ivec3_type, glsl_type::isampler2DMSArray_type), - _textureSize(texture_multisample, glsl_type::ivec3_type, glsl_type::usampler2DMSArray_type), + _textureSize(texture_multisample_array, glsl_type::ivec3_type, glsl_type::sampler2DMSArray_type), + _textureSize(texture_multisample_array, glsl_type::ivec3_type, glsl_type::isampler2DMSArray_type), + _textureSize(texture_multisample_array, glsl_type::ivec3_type, glsl_type::usampler2DMSArray_type), + NULL); + + add_function("textureSamples", + _textureSamples(glsl_type::sampler2DMS_type), + _textureSamples(glsl_type::isampler2DMS_type), + _textureSamples(glsl_type::usampler2DMS_type), + + _textureSamples(glsl_type::sampler2DMSArray_type), + _textureSamples(glsl_type::isampler2DMSArray_type), + _textureSamples(glsl_type::usampler2DMSArray_type), NULL); add_function("texture", @@ -1257,9 +1813,9 @@ builtin_builder::create_builtins() _texelFetch(texture_multisample, glsl_type::ivec4_type, glsl_type::isampler2DMS_type, glsl_type::ivec2_type), _texelFetch(texture_multisample, glsl_type::uvec4_type, glsl_type::usampler2DMS_type, glsl_type::ivec2_type), - _texelFetch(texture_multisample, glsl_type::vec4_type, glsl_type::sampler2DMSArray_type, glsl_type::ivec3_type), - _texelFetch(texture_multisample, glsl_type::ivec4_type, glsl_type::isampler2DMSArray_type, glsl_type::ivec3_type), - _texelFetch(texture_multisample, glsl_type::uvec4_type, glsl_type::usampler2DMSArray_type, glsl_type::ivec3_type), + _texelFetch(texture_multisample_array, glsl_type::vec4_type, glsl_type::sampler2DMSArray_type, glsl_type::ivec3_type), + _texelFetch(texture_multisample_array, glsl_type::ivec4_type, glsl_type::isampler2DMSArray_type, glsl_type::ivec3_type), + _texelFetch(texture_multisample_array, glsl_type::uvec4_type, glsl_type::usampler2DMSArray_type, glsl_type::ivec3_type), NULL); add_function("texelFetchOffset", @@ -1558,42 +2114,131 @@ builtin_builder::create_builtins() add_function("EmitVertex", _EmitVertex(), NULL); add_function("EndPrimitive", _EndPrimitive(), NULL); + add_function("EmitStreamVertex", + _EmitStreamVertex(gs_streams, glsl_type::uint_type), + _EmitStreamVertex(gs_streams, glsl_type::int_type), + NULL); + add_function("EndStreamPrimitive", + _EndStreamPrimitive(gs_streams, glsl_type::uint_type), + _EndStreamPrimitive(gs_streams, glsl_type::int_type), + NULL); + add_function("barrier", _barrier(), NULL); add_function("textureQueryLOD", - _textureQueryLod(glsl_type::sampler1D_type, glsl_type::float_type), - _textureQueryLod(glsl_type::isampler1D_type, glsl_type::float_type), - _textureQueryLod(glsl_type::usampler1D_type, glsl_type::float_type), - - _textureQueryLod(glsl_type::sampler2D_type, glsl_type::vec2_type), - _textureQueryLod(glsl_type::isampler2D_type, glsl_type::vec2_type), - _textureQueryLod(glsl_type::usampler2D_type, glsl_type::vec2_type), - - _textureQueryLod(glsl_type::sampler3D_type, glsl_type::vec3_type), - _textureQueryLod(glsl_type::isampler3D_type, glsl_type::vec3_type), - _textureQueryLod(glsl_type::usampler3D_type, glsl_type::vec3_type), + _textureQueryLod(texture_query_lod, glsl_type::sampler1D_type, glsl_type::float_type), + _textureQueryLod(texture_query_lod, glsl_type::isampler1D_type, glsl_type::float_type), + _textureQueryLod(texture_query_lod, glsl_type::usampler1D_type, glsl_type::float_type), + + _textureQueryLod(texture_query_lod, glsl_type::sampler2D_type, glsl_type::vec2_type), + _textureQueryLod(texture_query_lod, glsl_type::isampler2D_type, glsl_type::vec2_type), + _textureQueryLod(texture_query_lod, glsl_type::usampler2D_type, glsl_type::vec2_type), + + _textureQueryLod(texture_query_lod, glsl_type::sampler3D_type, glsl_type::vec3_type), + _textureQueryLod(texture_query_lod, glsl_type::isampler3D_type, glsl_type::vec3_type), + _textureQueryLod(texture_query_lod, glsl_type::usampler3D_type, glsl_type::vec3_type), + + _textureQueryLod(texture_query_lod, glsl_type::samplerCube_type, glsl_type::vec3_type), + _textureQueryLod(texture_query_lod, glsl_type::isamplerCube_type, glsl_type::vec3_type), + _textureQueryLod(texture_query_lod, glsl_type::usamplerCube_type, glsl_type::vec3_type), + + _textureQueryLod(texture_query_lod, glsl_type::sampler1DArray_type, glsl_type::float_type), + _textureQueryLod(texture_query_lod, glsl_type::isampler1DArray_type, glsl_type::float_type), + _textureQueryLod(texture_query_lod, glsl_type::usampler1DArray_type, glsl_type::float_type), + + _textureQueryLod(texture_query_lod, glsl_type::sampler2DArray_type, glsl_type::vec2_type), + _textureQueryLod(texture_query_lod, glsl_type::isampler2DArray_type, glsl_type::vec2_type), + _textureQueryLod(texture_query_lod, glsl_type::usampler2DArray_type, glsl_type::vec2_type), + + _textureQueryLod(texture_query_lod, glsl_type::samplerCubeArray_type, glsl_type::vec3_type), + _textureQueryLod(texture_query_lod, glsl_type::isamplerCubeArray_type, glsl_type::vec3_type), + _textureQueryLod(texture_query_lod, glsl_type::usamplerCubeArray_type, glsl_type::vec3_type), + + _textureQueryLod(texture_query_lod, glsl_type::sampler1DShadow_type, glsl_type::float_type), + _textureQueryLod(texture_query_lod, glsl_type::sampler2DShadow_type, glsl_type::vec2_type), + _textureQueryLod(texture_query_lod, glsl_type::samplerCubeShadow_type, glsl_type::vec3_type), + _textureQueryLod(texture_query_lod, glsl_type::sampler1DArrayShadow_type, glsl_type::float_type), + _textureQueryLod(texture_query_lod, glsl_type::sampler2DArrayShadow_type, glsl_type::vec2_type), + _textureQueryLod(texture_query_lod, glsl_type::samplerCubeArrayShadow_type, glsl_type::vec3_type), + NULL); - _textureQueryLod(glsl_type::samplerCube_type, glsl_type::vec3_type), - _textureQueryLod(glsl_type::isamplerCube_type, glsl_type::vec3_type), - _textureQueryLod(glsl_type::usamplerCube_type, glsl_type::vec3_type), + add_function("textureQueryLod", + _textureQueryLod(v400_fs_only, glsl_type::sampler1D_type, glsl_type::float_type), + _textureQueryLod(v400_fs_only, glsl_type::isampler1D_type, glsl_type::float_type), + _textureQueryLod(v400_fs_only, glsl_type::usampler1D_type, glsl_type::float_type), + + _textureQueryLod(v400_fs_only, glsl_type::sampler2D_type, glsl_type::vec2_type), + _textureQueryLod(v400_fs_only, glsl_type::isampler2D_type, glsl_type::vec2_type), + _textureQueryLod(v400_fs_only, glsl_type::usampler2D_type, glsl_type::vec2_type), + + _textureQueryLod(v400_fs_only, glsl_type::sampler3D_type, glsl_type::vec3_type), + _textureQueryLod(v400_fs_only, glsl_type::isampler3D_type, glsl_type::vec3_type), + _textureQueryLod(v400_fs_only, glsl_type::usampler3D_type, glsl_type::vec3_type), + + _textureQueryLod(v400_fs_only, glsl_type::samplerCube_type, glsl_type::vec3_type), + _textureQueryLod(v400_fs_only, glsl_type::isamplerCube_type, glsl_type::vec3_type), + _textureQueryLod(v400_fs_only, glsl_type::usamplerCube_type, glsl_type::vec3_type), + + _textureQueryLod(v400_fs_only, glsl_type::sampler1DArray_type, glsl_type::float_type), + _textureQueryLod(v400_fs_only, glsl_type::isampler1DArray_type, glsl_type::float_type), + _textureQueryLod(v400_fs_only, glsl_type::usampler1DArray_type, glsl_type::float_type), + + _textureQueryLod(v400_fs_only, glsl_type::sampler2DArray_type, glsl_type::vec2_type), + _textureQueryLod(v400_fs_only, glsl_type::isampler2DArray_type, glsl_type::vec2_type), + _textureQueryLod(v400_fs_only, glsl_type::usampler2DArray_type, glsl_type::vec2_type), + + _textureQueryLod(v400_fs_only, glsl_type::samplerCubeArray_type, glsl_type::vec3_type), + _textureQueryLod(v400_fs_only, glsl_type::isamplerCubeArray_type, glsl_type::vec3_type), + _textureQueryLod(v400_fs_only, glsl_type::usamplerCubeArray_type, glsl_type::vec3_type), + + _textureQueryLod(v400_fs_only, glsl_type::sampler1DShadow_type, glsl_type::float_type), + _textureQueryLod(v400_fs_only, glsl_type::sampler2DShadow_type, glsl_type::vec2_type), + _textureQueryLod(v400_fs_only, glsl_type::samplerCubeShadow_type, glsl_type::vec3_type), + _textureQueryLod(v400_fs_only, glsl_type::sampler1DArrayShadow_type, glsl_type::float_type), + _textureQueryLod(v400_fs_only, glsl_type::sampler2DArrayShadow_type, glsl_type::vec2_type), + _textureQueryLod(v400_fs_only, glsl_type::samplerCubeArrayShadow_type, glsl_type::vec3_type), + NULL); - _textureQueryLod(glsl_type::sampler1DArray_type, glsl_type::float_type), - _textureQueryLod(glsl_type::isampler1DArray_type, glsl_type::float_type), - _textureQueryLod(glsl_type::usampler1DArray_type, glsl_type::float_type), + add_function("textureQueryLevels", + _textureQueryLevels(glsl_type::sampler1D_type), + _textureQueryLevels(glsl_type::sampler2D_type), + _textureQueryLevels(glsl_type::sampler3D_type), + _textureQueryLevels(glsl_type::samplerCube_type), + _textureQueryLevels(glsl_type::sampler1DArray_type), + _textureQueryLevels(glsl_type::sampler2DArray_type), + _textureQueryLevels(glsl_type::samplerCubeArray_type), + _textureQueryLevels(glsl_type::sampler1DShadow_type), + _textureQueryLevels(glsl_type::sampler2DShadow_type), + _textureQueryLevels(glsl_type::samplerCubeShadow_type), + _textureQueryLevels(glsl_type::sampler1DArrayShadow_type), + _textureQueryLevels(glsl_type::sampler2DArrayShadow_type), + _textureQueryLevels(glsl_type::samplerCubeArrayShadow_type), + + _textureQueryLevels(glsl_type::isampler1D_type), + _textureQueryLevels(glsl_type::isampler2D_type), + _textureQueryLevels(glsl_type::isampler3D_type), + _textureQueryLevels(glsl_type::isamplerCube_type), + _textureQueryLevels(glsl_type::isampler1DArray_type), + _textureQueryLevels(glsl_type::isampler2DArray_type), + _textureQueryLevels(glsl_type::isamplerCubeArray_type), + + _textureQueryLevels(glsl_type::usampler1D_type), + _textureQueryLevels(glsl_type::usampler2D_type), + _textureQueryLevels(glsl_type::usampler3D_type), + _textureQueryLevels(glsl_type::usamplerCube_type), + _textureQueryLevels(glsl_type::usampler1DArray_type), + _textureQueryLevels(glsl_type::usampler2DArray_type), + _textureQueryLevels(glsl_type::usamplerCubeArray_type), - _textureQueryLod(glsl_type::sampler2DArray_type, glsl_type::vec2_type), - _textureQueryLod(glsl_type::isampler2DArray_type, glsl_type::vec2_type), - _textureQueryLod(glsl_type::usampler2DArray_type, glsl_type::vec2_type), + NULL); - _textureQueryLod(glsl_type::samplerCubeArray_type, glsl_type::vec3_type), - _textureQueryLod(glsl_type::isamplerCubeArray_type, glsl_type::vec3_type), - _textureQueryLod(glsl_type::usamplerCubeArray_type, glsl_type::vec3_type), + add_function("textureSamplesIdenticalEXT", + _textureSamplesIdentical(texture_samples_identical, glsl_type::sampler2DMS_type, glsl_type::ivec2_type), + _textureSamplesIdentical(texture_samples_identical, glsl_type::isampler2DMS_type, glsl_type::ivec2_type), + _textureSamplesIdentical(texture_samples_identical, glsl_type::usampler2DMS_type, glsl_type::ivec2_type), - _textureQueryLod(glsl_type::sampler1DShadow_type, glsl_type::float_type), - _textureQueryLod(glsl_type::sampler2DShadow_type, glsl_type::vec2_type), - _textureQueryLod(glsl_type::samplerCubeShadow_type, glsl_type::vec3_type), - _textureQueryLod(glsl_type::sampler1DArrayShadow_type, glsl_type::float_type), - _textureQueryLod(glsl_type::sampler2DArrayShadow_type, glsl_type::vec2_type), - _textureQueryLod(glsl_type::samplerCubeArrayShadow_type, glsl_type::vec3_type), + _textureSamplesIdentical(texture_samples_identical_array, glsl_type::sampler2DMSArray_type, glsl_type::ivec3_type), + _textureSamplesIdentical(texture_samples_identical_array, glsl_type::isampler2DMSArray_type, glsl_type::ivec3_type), + _textureSamplesIdentical(texture_samples_identical_array, glsl_type::usampler2DMSArray_type, glsl_type::ivec3_type), NULL); add_function("texture1D", @@ -1655,8 +2300,8 @@ builtin_builder::create_builtins() NULL); add_function("texture2DProjLod", - _texture(ir_txl, v110_lod, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec3_type, TEX_PROJECT), - _texture(ir_txl, v110_lod, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec4_type, TEX_PROJECT), + _texture(ir_txl, lod_exists_in_stage, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec3_type, TEX_PROJECT), + _texture(ir_txl, lod_exists_in_stage, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec4_type, TEX_PROJECT), NULL); add_function("texture3D", @@ -1683,7 +2328,7 @@ builtin_builder::create_builtins() NULL); add_function("textureCubeLod", - _texture(ir_txl, v110_lod, glsl_type::vec4_type, glsl_type::samplerCube_type, glsl_type::vec3_type), + _texture(ir_txl, lod_exists_in_stage, glsl_type::vec4_type, glsl_type::samplerCube_type, glsl_type::vec3_type), NULL); add_function("texture2DRect", @@ -1816,9 +2461,142 @@ builtin_builder::create_builtins() _texture(ir_txd, shader_texture_lod_and_rect, glsl_type::vec4_type, glsl_type::sampler2DRectShadow_type, glsl_type::vec4_type, TEX_PROJECT), NULL); + add_function("textureGather", + _texture(ir_tg4, texture_gather_or_es31, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec2_type), + _texture(ir_tg4, texture_gather_or_es31, glsl_type::ivec4_type, glsl_type::isampler2D_type, glsl_type::vec2_type), + _texture(ir_tg4, texture_gather_or_es31, glsl_type::uvec4_type, glsl_type::usampler2D_type, glsl_type::vec2_type), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec2_type), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec2_type), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DRect_type, glsl_type::vec2_type), + + _texture(ir_tg4, texture_gather_or_es31, glsl_type::vec4_type, glsl_type::sampler2DArray_type, glsl_type::vec3_type), + _texture(ir_tg4, texture_gather_or_es31, glsl_type::ivec4_type, glsl_type::isampler2DArray_type, glsl_type::vec3_type), + _texture(ir_tg4, texture_gather_or_es31, glsl_type::uvec4_type, glsl_type::usampler2DArray_type, glsl_type::vec3_type), + + _texture(ir_tg4, texture_gather_or_es31, glsl_type::vec4_type, glsl_type::samplerCube_type, glsl_type::vec3_type), + _texture(ir_tg4, texture_gather_or_es31, glsl_type::ivec4_type, glsl_type::isamplerCube_type, glsl_type::vec3_type), + _texture(ir_tg4, texture_gather_or_es31, glsl_type::uvec4_type, glsl_type::usamplerCube_type, glsl_type::vec3_type), + + _texture(ir_tg4, texture_gather, glsl_type::vec4_type, glsl_type::samplerCubeArray_type, glsl_type::vec4_type), + _texture(ir_tg4, texture_gather, glsl_type::ivec4_type, glsl_type::isamplerCubeArray_type, glsl_type::vec4_type), + _texture(ir_tg4, texture_gather, glsl_type::uvec4_type, glsl_type::usamplerCubeArray_type, glsl_type::vec4_type), + + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec2_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::ivec4_type, glsl_type::isampler2D_type, glsl_type::vec2_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::uvec4_type, glsl_type::usampler2D_type, glsl_type::vec2_type, TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec2_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec2_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DRect_type, glsl_type::vec2_type, TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::vec4_type, glsl_type::sampler2DArray_type, glsl_type::vec3_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::ivec4_type, glsl_type::isampler2DArray_type, glsl_type::vec3_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::uvec4_type, glsl_type::usampler2DArray_type, glsl_type::vec3_type, TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::vec4_type, glsl_type::samplerCube_type, glsl_type::vec3_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::ivec4_type, glsl_type::isamplerCube_type, glsl_type::vec3_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::uvec4_type, glsl_type::usamplerCube_type, glsl_type::vec3_type, TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::samplerCubeArray_type, glsl_type::vec4_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isamplerCubeArray_type, glsl_type::vec4_type, TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usamplerCubeArray_type, glsl_type::vec4_type, TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::vec4_type, glsl_type::sampler2DShadow_type, glsl_type::vec2_type), + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::vec4_type, glsl_type::sampler2DArrayShadow_type, glsl_type::vec3_type), + _texture(ir_tg4, gpu_shader5_or_es31, glsl_type::vec4_type, glsl_type::samplerCubeShadow_type, glsl_type::vec3_type), + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::samplerCubeArrayShadow_type, glsl_type::vec4_type), + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRectShadow_type, glsl_type::vec2_type), + NULL); + + add_function("textureGatherOffset", + _texture(ir_tg4, texture_gather_only_or_es31, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec2_type, TEX_OFFSET), + _texture(ir_tg4, texture_gather_only_or_es31, glsl_type::ivec4_type, glsl_type::isampler2D_type, glsl_type::vec2_type, TEX_OFFSET), + _texture(ir_tg4, texture_gather_only_or_es31, glsl_type::uvec4_type, glsl_type::usampler2D_type, glsl_type::vec2_type, TEX_OFFSET), + + _texture(ir_tg4, texture_gather_only_or_es31, glsl_type::vec4_type, glsl_type::sampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET), + _texture(ir_tg4, texture_gather_only_or_es31, glsl_type::ivec4_type, glsl_type::isampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET), + _texture(ir_tg4, texture_gather_only_or_es31, glsl_type::uvec4_type, glsl_type::usampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET), + + _texture(ir_tg4, es31, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec2_type, TEX_OFFSET | TEX_COMPONENT), + _texture(ir_tg4, es31, glsl_type::ivec4_type, glsl_type::isampler2D_type, glsl_type::vec2_type, TEX_OFFSET | TEX_COMPONENT), + _texture(ir_tg4, es31, glsl_type::uvec4_type, glsl_type::usampler2D_type, glsl_type::vec2_type, TEX_OFFSET | TEX_COMPONENT), + + _texture(ir_tg4, es31, glsl_type::vec4_type, glsl_type::sampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET | TEX_COMPONENT), + _texture(ir_tg4, es31, glsl_type::ivec4_type, glsl_type::isampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET | TEX_COMPONENT), + _texture(ir_tg4, es31, glsl_type::uvec4_type, glsl_type::usampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET | TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2D_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2D_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_NONCONST), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_NONCONST), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_NONCONST), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2D_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2D_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST | TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DShadow_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST), + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DArrayShadow_type, glsl_type::vec3_type, TEX_OFFSET_NONCONST), + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRectShadow_type, glsl_type::vec2_type, TEX_OFFSET_NONCONST), + + _texture(ir_tg4, es31, glsl_type::vec4_type, glsl_type::sampler2DShadow_type, glsl_type::vec2_type, TEX_OFFSET), + _texture(ir_tg4, es31, glsl_type::vec4_type, glsl_type::sampler2DArrayShadow_type, glsl_type::vec3_type, TEX_OFFSET), + NULL); + + add_function("textureGatherOffsets", + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2D_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2D_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2D_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2D_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2D_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_ARRAY), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_ARRAY), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_ARRAY), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DArray_type, glsl_type::vec3_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + _texture(ir_tg4, gpu_shader5, glsl_type::uvec4_type, glsl_type::usampler2DRect_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY | TEX_COMPONENT), + + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DShadow_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY), + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DArrayShadow_type, glsl_type::vec3_type, TEX_OFFSET_ARRAY), + _texture(ir_tg4, gpu_shader5, glsl_type::vec4_type, glsl_type::sampler2DRectShadow_type, glsl_type::vec2_type, TEX_OFFSET_ARRAY), + NULL); + F(dFdx) F(dFdy) F(fwidth) + F(dFdxCoarse) + F(dFdyCoarse) + F(fwidthCoarse) + F(dFdxFine) + F(dFdyFine) + F(fwidthFine) F(noise1) F(noise2) F(noise3) @@ -1830,13 +2608,17 @@ builtin_builder::create_builtins() IU(bitCount) IU(findLSB) IU(findMSB) - F(fma) + FDGS5(fma) add_function("ldexp", _ldexp(glsl_type::float_type, glsl_type::int_type), _ldexp(glsl_type::vec2_type, glsl_type::ivec2_type), _ldexp(glsl_type::vec3_type, glsl_type::ivec3_type), _ldexp(glsl_type::vec4_type, glsl_type::ivec4_type), + _ldexp(glsl_type::double_type, glsl_type::int_type), + _ldexp(glsl_type::dvec2_type, glsl_type::ivec2_type), + _ldexp(glsl_type::dvec3_type, glsl_type::ivec3_type), + _ldexp(glsl_type::dvec4_type, glsl_type::ivec4_type), NULL); add_function("frexp", @@ -1844,18 +2626,226 @@ builtin_builder::create_builtins() _frexp(glsl_type::vec2_type, glsl_type::ivec2_type), _frexp(glsl_type::vec3_type, glsl_type::ivec3_type), _frexp(glsl_type::vec4_type, glsl_type::ivec4_type), + _dfrexp(glsl_type::double_type, glsl_type::int_type), + _dfrexp(glsl_type::dvec2_type, glsl_type::ivec2_type), + _dfrexp(glsl_type::dvec3_type, glsl_type::ivec3_type), + _dfrexp(glsl_type::dvec4_type, glsl_type::ivec4_type), + NULL); + add_function("uaddCarry", + _uaddCarry(glsl_type::uint_type), + _uaddCarry(glsl_type::uvec2_type), + _uaddCarry(glsl_type::uvec3_type), + _uaddCarry(glsl_type::uvec4_type), + NULL); + add_function("usubBorrow", + _usubBorrow(glsl_type::uint_type), + _usubBorrow(glsl_type::uvec2_type), + _usubBorrow(glsl_type::uvec3_type), + _usubBorrow(glsl_type::uvec4_type), + NULL); + add_function("imulExtended", + _mulExtended(glsl_type::int_type), + _mulExtended(glsl_type::ivec2_type), + _mulExtended(glsl_type::ivec3_type), + _mulExtended(glsl_type::ivec4_type), + NULL); + add_function("umulExtended", + _mulExtended(glsl_type::uint_type), + _mulExtended(glsl_type::uvec2_type), + _mulExtended(glsl_type::uvec3_type), + _mulExtended(glsl_type::uvec4_type), + NULL); + add_function("interpolateAtCentroid", + _interpolateAtCentroid(glsl_type::float_type), + _interpolateAtCentroid(glsl_type::vec2_type), + _interpolateAtCentroid(glsl_type::vec3_type), + _interpolateAtCentroid(glsl_type::vec4_type), + NULL); + add_function("interpolateAtOffset", + _interpolateAtOffset(glsl_type::float_type), + _interpolateAtOffset(glsl_type::vec2_type), + _interpolateAtOffset(glsl_type::vec3_type), + _interpolateAtOffset(glsl_type::vec4_type), + NULL); + add_function("interpolateAtSample", + _interpolateAtSample(glsl_type::float_type), + _interpolateAtSample(glsl_type::vec2_type), + _interpolateAtSample(glsl_type::vec3_type), + _interpolateAtSample(glsl_type::vec4_type), NULL); -#undef F -#undef FI -#undef FIU -#undef FIUB -#undef FIU2_MIXED -} - -void -builtin_builder::add_function(const char *name, ...) -{ - va_list ap; + + add_function("atomicCounter", + _atomic_counter_op("__intrinsic_atomic_read", + shader_atomic_counters), + NULL); + add_function("atomicCounterIncrement", + _atomic_counter_op("__intrinsic_atomic_increment", + shader_atomic_counters), + NULL); + add_function("atomicCounterDecrement", + _atomic_counter_op("__intrinsic_atomic_predecrement", + shader_atomic_counters), + NULL); + + add_function("atomicAdd", + _atomic_op2("__intrinsic_atomic_add", + buffer_atomics_supported, + glsl_type::uint_type), + _atomic_op2("__intrinsic_atomic_add", + buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("atomicMin", + _atomic_op2("__intrinsic_atomic_min", + buffer_atomics_supported, + glsl_type::uint_type), + _atomic_op2("__intrinsic_atomic_min", + buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("atomicMax", + _atomic_op2("__intrinsic_atomic_max", + buffer_atomics_supported, + glsl_type::uint_type), + _atomic_op2("__intrinsic_atomic_max", + buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("atomicAnd", + _atomic_op2("__intrinsic_atomic_and", + buffer_atomics_supported, + glsl_type::uint_type), + _atomic_op2("__intrinsic_atomic_and", + buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("atomicOr", + _atomic_op2("__intrinsic_atomic_or", + buffer_atomics_supported, + glsl_type::uint_type), + _atomic_op2("__intrinsic_atomic_or", + buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("atomicXor", + _atomic_op2("__intrinsic_atomic_xor", + buffer_atomics_supported, + glsl_type::uint_type), + _atomic_op2("__intrinsic_atomic_xor", + buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("atomicExchange", + _atomic_op2("__intrinsic_atomic_exchange", + buffer_atomics_supported, + glsl_type::uint_type), + _atomic_op2("__intrinsic_atomic_exchange", + buffer_atomics_supported, + glsl_type::int_type), + NULL); + add_function("atomicCompSwap", + _atomic_op3("__intrinsic_atomic_comp_swap", + buffer_atomics_supported, + glsl_type::uint_type), + _atomic_op3("__intrinsic_atomic_comp_swap", + buffer_atomics_supported, + glsl_type::int_type), + NULL); + + add_function("min3", + _min3(glsl_type::float_type), + _min3(glsl_type::vec2_type), + _min3(glsl_type::vec3_type), + _min3(glsl_type::vec4_type), + + _min3(glsl_type::int_type), + _min3(glsl_type::ivec2_type), + _min3(glsl_type::ivec3_type), + _min3(glsl_type::ivec4_type), + + _min3(glsl_type::uint_type), + _min3(glsl_type::uvec2_type), + _min3(glsl_type::uvec3_type), + _min3(glsl_type::uvec4_type), + NULL); + + add_function("max3", + _max3(glsl_type::float_type), + _max3(glsl_type::vec2_type), + _max3(glsl_type::vec3_type), + _max3(glsl_type::vec4_type), + + _max3(glsl_type::int_type), + _max3(glsl_type::ivec2_type), + _max3(glsl_type::ivec3_type), + _max3(glsl_type::ivec4_type), + + _max3(glsl_type::uint_type), + _max3(glsl_type::uvec2_type), + _max3(glsl_type::uvec3_type), + _max3(glsl_type::uvec4_type), + NULL); + + add_function("mid3", + _mid3(glsl_type::float_type), + _mid3(glsl_type::vec2_type), + _mid3(glsl_type::vec3_type), + _mid3(glsl_type::vec4_type), + + _mid3(glsl_type::int_type), + _mid3(glsl_type::ivec2_type), + _mid3(glsl_type::ivec3_type), + _mid3(glsl_type::ivec4_type), + + _mid3(glsl_type::uint_type), + _mid3(glsl_type::uvec2_type), + _mid3(glsl_type::uvec3_type), + _mid3(glsl_type::uvec4_type), + NULL); + + add_image_functions(true); + + add_function("memoryBarrier", + _memory_barrier("__intrinsic_memory_barrier", + shader_image_load_store), + NULL); + add_function("groupMemoryBarrier", + _memory_barrier("__intrinsic_group_memory_barrier", + compute_shader), + NULL); + add_function("memoryBarrierAtomicCounter", + _memory_barrier("__intrinsic_memory_barrier_atomic_counter", + compute_shader), + NULL); + add_function("memoryBarrierBuffer", + _memory_barrier("__intrinsic_memory_barrier_buffer", + compute_shader), + NULL); + add_function("memoryBarrierImage", + _memory_barrier("__intrinsic_memory_barrier_image", + compute_shader), + NULL); + add_function("memoryBarrierShared", + _memory_barrier("__intrinsic_memory_barrier_shared", + compute_shader), + NULL); + + add_function("clock2x32ARB", + _shader_clock(shader_clock, + glsl_type::uvec2_type), + NULL); + +#undef F +#undef FI +#undef FIUD +#undef FIUBD +#undef FIU2_MIXED +} + +void +builtin_builder::add_function(const char *name, ...) +{ + va_list ap; ir_function *f = new(mem_ctx) ir_function(name); @@ -1865,8 +2855,6 @@ builtin_builder::add_function(const char *name, ...) if (sig == NULL) break; - sig->is_defined = true; - if (false) { exec_list stuff; stuff.push_tail(sig); @@ -1880,6 +2868,131 @@ builtin_builder::add_function(const char *name, ...) shader->symbols->add_function(f); } +void +builtin_builder::add_image_function(const char *name, + const char *intrinsic_name, + image_prototype_ctr prototype, + unsigned num_arguments, + unsigned flags) +{ + static const glsl_type *const types[] = { + glsl_type::image1D_type, + glsl_type::image2D_type, + glsl_type::image3D_type, + glsl_type::image2DRect_type, + glsl_type::imageCube_type, + glsl_type::imageBuffer_type, + glsl_type::image1DArray_type, + glsl_type::image2DArray_type, + glsl_type::imageCubeArray_type, + glsl_type::image2DMS_type, + glsl_type::image2DMSArray_type, + glsl_type::iimage1D_type, + glsl_type::iimage2D_type, + glsl_type::iimage3D_type, + glsl_type::iimage2DRect_type, + glsl_type::iimageCube_type, + glsl_type::iimageBuffer_type, + glsl_type::iimage1DArray_type, + glsl_type::iimage2DArray_type, + glsl_type::iimageCubeArray_type, + glsl_type::iimage2DMS_type, + glsl_type::iimage2DMSArray_type, + glsl_type::uimage1D_type, + glsl_type::uimage2D_type, + glsl_type::uimage3D_type, + glsl_type::uimage2DRect_type, + glsl_type::uimageCube_type, + glsl_type::uimageBuffer_type, + glsl_type::uimage1DArray_type, + glsl_type::uimage2DArray_type, + glsl_type::uimageCubeArray_type, + glsl_type::uimage2DMS_type, + glsl_type::uimage2DMSArray_type + }; + + ir_function *f = new(mem_ctx) ir_function(name); + + for (unsigned i = 0; i < ARRAY_SIZE(types); ++i) { + if ((types[i]->sampler_type != GLSL_TYPE_FLOAT || + (flags & IMAGE_FUNCTION_SUPPORTS_FLOAT_DATA_TYPE)) && + (types[i]->sampler_dimensionality == GLSL_SAMPLER_DIM_MS || + !(flags & IMAGE_FUNCTION_MS_ONLY))) + f->add_signature(_image(prototype, types[i], intrinsic_name, + num_arguments, flags)); + } + + shader->symbols->add_function(f); +} + +void +builtin_builder::add_image_functions(bool glsl) +{ + const unsigned flags = (glsl ? IMAGE_FUNCTION_EMIT_STUB : 0); + + add_image_function(glsl ? "imageLoad" : "__intrinsic_image_load", + "__intrinsic_image_load", + &builtin_builder::_image_prototype, 0, + (flags | IMAGE_FUNCTION_HAS_VECTOR_DATA_TYPE | + IMAGE_FUNCTION_SUPPORTS_FLOAT_DATA_TYPE | + IMAGE_FUNCTION_READ_ONLY)); + + add_image_function(glsl ? "imageStore" : "__intrinsic_image_store", + "__intrinsic_image_store", + &builtin_builder::_image_prototype, 1, + (flags | IMAGE_FUNCTION_RETURNS_VOID | + IMAGE_FUNCTION_HAS_VECTOR_DATA_TYPE | + IMAGE_FUNCTION_SUPPORTS_FLOAT_DATA_TYPE | + IMAGE_FUNCTION_WRITE_ONLY)); + + const unsigned atom_flags = flags | IMAGE_FUNCTION_AVAIL_ATOMIC; + + add_image_function(glsl ? "imageAtomicAdd" : "__intrinsic_image_atomic_add", + "__intrinsic_image_atomic_add", + &builtin_builder::_image_prototype, 1, atom_flags); + + add_image_function(glsl ? "imageAtomicMin" : "__intrinsic_image_atomic_min", + "__intrinsic_image_atomic_min", + &builtin_builder::_image_prototype, 1, atom_flags); + + add_image_function(glsl ? "imageAtomicMax" : "__intrinsic_image_atomic_max", + "__intrinsic_image_atomic_max", + &builtin_builder::_image_prototype, 1, atom_flags); + + add_image_function(glsl ? "imageAtomicAnd" : "__intrinsic_image_atomic_and", + "__intrinsic_image_atomic_and", + &builtin_builder::_image_prototype, 1, atom_flags); + + add_image_function(glsl ? "imageAtomicOr" : "__intrinsic_image_atomic_or", + "__intrinsic_image_atomic_or", + &builtin_builder::_image_prototype, 1, atom_flags); + + add_image_function(glsl ? "imageAtomicXor" : "__intrinsic_image_atomic_xor", + "__intrinsic_image_atomic_xor", + &builtin_builder::_image_prototype, 1, atom_flags); + + add_image_function((glsl ? "imageAtomicExchange" : + "__intrinsic_image_atomic_exchange"), + "__intrinsic_image_atomic_exchange", + &builtin_builder::_image_prototype, 1, atom_flags); + + add_image_function((glsl ? "imageAtomicCompSwap" : + "__intrinsic_image_atomic_comp_swap"), + "__intrinsic_image_atomic_comp_swap", + &builtin_builder::_image_prototype, 2, atom_flags); + + add_image_function(glsl ? "imageSize" : "__intrinsic_image_size", + "__intrinsic_image_size", + &builtin_builder::_image_size_prototype, 1, + flags | IMAGE_FUNCTION_SUPPORTS_FLOAT_DATA_TYPE); + + add_image_function(glsl ? "imageSamples" : "__intrinsic_image_samples", + "__intrinsic_image_samples", + &builtin_builder::_image_samples_prototype, 1, + flags | IMAGE_FUNCTION_SUPPORTS_FLOAT_DATA_TYPE | + IMAGE_FUNCTION_MS_ONLY); +} + ir_variable * builtin_builder::in_var(const glsl_type *type, const char *name) { @@ -1910,12 +3023,20 @@ builtin_builder::imm(unsigned u, unsigned vector_elements) return new(mem_ctx) ir_constant(u, vector_elements); } +ir_constant * +builtin_builder::imm(double d, unsigned vector_elements) +{ + return new(mem_ctx) ir_constant(d, vector_elements); +} + ir_constant * builtin_builder::imm(const glsl_type *type, const ir_constant_data &data) { return new(mem_ctx) ir_constant(type, &data); } +#define IMM_FP(type, val) (type->base_type == GLSL_TYPE_DOUBLE) ? imm(val) : imm((float)val) + ir_dereference_variable * builtin_builder::var_ref(ir_variable *var) { @@ -1964,7 +3085,13 @@ builtin_builder::new_sig(const glsl_type *return_type, #define MAKE_SIG(return_type, avail, ...) \ ir_function_signature *sig = \ new_sig(return_type, avail, __VA_ARGS__); \ - ir_factory body(&sig->body, mem_ctx); + ir_factory body(&sig->body, mem_ctx); \ + sig->is_defined = true; + +#define MAKE_INTRINSIC(return_type, avail, ...) \ + ir_function_signature *sig = \ + new_sig(return_type, avail, __VA_ARGS__); \ + sig->is_intrinsic = true; ir_function_signature * builtin_builder::unop(builtin_available_predicate avail, @@ -1985,9 +3112,16 @@ builtin_builder::_##NAME(const glsl_type *type) \ return unop(&AVAIL, OPCODE, type, type); \ } +#define UNOPA(NAME, OPCODE) \ +ir_function_signature * \ +builtin_builder::_##NAME(builtin_available_predicate avail, const glsl_type *type) \ +{ \ + return unop(avail, OPCODE, type, type); \ +} + ir_function_signature * -builtin_builder::binop(ir_expression_operation opcode, - builtin_available_predicate avail, +builtin_builder::binop(builtin_available_predicate avail, + ir_expression_operation opcode, const glsl_type *return_type, const glsl_type *param0_type, const glsl_type *param1_type) @@ -2046,16 +3180,35 @@ ir_expression * builtin_builder::asin_expr(ir_variable *x) { return mul(sign(x), - sub(imm(1.5707964f), + sub(imm(M_PI_2f), mul(sqrt(sub(imm(1.0f), abs(x))), - add(imm(1.5707964f), + add(imm(M_PI_2f), mul(abs(x), - add(imm(-0.21460183f), + add(imm(M_PI_4f - 1.0f), mul(abs(x), add(imm(0.086566724f), mul(abs(x), imm(-0.03102955f)))))))))); } +ir_call * +builtin_builder::call(ir_function *f, ir_variable *ret, exec_list params) +{ + exec_list actual_params; + + foreach_in_list(ir_variable, var, ¶ms) { + actual_params.push_tail(var_ref(var)); + } + + ir_function_signature *sig = + f->exact_matching_signature(NULL, &actual_params); + if (!sig) + return NULL; + + ir_dereference_variable *deref = + (sig->return_type->is_void() ? NULL : var_ref(ret)); + + return new(mem_ctx) ir_call(sig, deref, &actual_params); +} ir_function_signature * builtin_builder::_asin(const glsl_type *type) @@ -2074,7 +3227,7 @@ builtin_builder::_acos(const glsl_type *type) ir_variable *x = in_var(type, "x"); MAKE_SIG(type, always_available, 1, x); - body.emit(ret(sub(imm(1.5707964f), asin_expr(x)))); + body.emit(ret(sub(imm(M_PI_2f), asin_expr(x)))); return sig; } @@ -2101,23 +3254,19 @@ builtin_builder::_atan2(const glsl_type *type) ir_factory outer_then(&outer_if->then_instructions, mem_ctx); /* Then...call atan(y/x) */ - ir_variable *y_over_x = outer_then.make_temp(glsl_type::float_type, "y_over_x"); - outer_then.emit(assign(y_over_x, div(y, x))); - outer_then.emit(assign(r, mul(y_over_x, rsq(add(mul(y_over_x, y_over_x), - imm(1.0f)))))); - outer_then.emit(assign(r, asin_expr(r))); + do_atan(body, glsl_type::float_type, r, div(y, x)); /* ...and fix it up: */ ir_if *inner_if = new(mem_ctx) ir_if(less(x, imm(0.0f))); inner_if->then_instructions.push_tail( if_tree(gequal(y, imm(0.0f)), - assign(r, add(r, imm(3.141593f))), - assign(r, sub(r, imm(3.141593f))))); + assign(r, add(r, imm(M_PIf))), + assign(r, sub(r, imm(M_PIf))))); outer_then.emit(inner_if); /* Else... */ outer_if->else_instructions.push_tail( - assign(r, mul(sign(y), imm(1.5707965f)))); + assign(r, mul(sign(y), imm(M_PI_2f)))); body.emit(outer_if); @@ -2128,17 +3277,65 @@ builtin_builder::_atan2(const glsl_type *type) return sig; } +void +builtin_builder::do_atan(ir_factory &body, const glsl_type *type, ir_variable *res, operand y_over_x) +{ + /* + * range-reduction, first step: + * + * / y_over_x if |y_over_x| <= 1.0; + * x = < + * \ 1.0 / y_over_x otherwise + */ + ir_variable *x = body.make_temp(type, "atan_x"); + body.emit(assign(x, div(min2(abs(y_over_x), + imm(1.0f)), + max2(abs(y_over_x), + imm(1.0f))))); + + /* + * approximate atan by evaluating polynomial: + * + * x * 0.9999793128310355 - x^3 * 0.3326756418091246 + + * x^5 * 0.1938924977115610 - x^7 * 0.1173503194786851 + + * x^9 * 0.0536813784310406 - x^11 * 0.0121323213173444 + */ + ir_variable *tmp = body.make_temp(type, "atan_tmp"); + body.emit(assign(tmp, mul(x, x))); + body.emit(assign(tmp, mul(add(mul(sub(mul(add(mul(sub(mul(add(mul(imm(-0.0121323213173444f), + tmp), + imm(0.0536813784310406f)), + tmp), + imm(0.1173503194786851f)), + tmp), + imm(0.1938924977115610f)), + tmp), + imm(0.3326756418091246f)), + tmp), + imm(0.9999793128310355f)), + x))); + + /* range-reduction fixup */ + body.emit(assign(tmp, add(tmp, + mul(b2f(greater(abs(y_over_x), + imm(1.0f, type->components()))), + add(mul(tmp, + imm(-2.0f)), + imm(M_PI_2f)))))); + + /* sign fixup */ + body.emit(assign(res, mul(tmp, sign(y_over_x)))); +} + ir_function_signature * builtin_builder::_atan(const glsl_type *type) { ir_variable *y_over_x = in_var(type, "y_over_x"); MAKE_SIG(type, always_available, 1, y_over_x); - ir_variable *t = body.make_temp(type, "t"); - body.emit(assign(t, mul(y_over_x, rsq(add(mul(y_over_x, y_over_x), - imm(1.0f)))))); - - body.emit(ret(asin_expr(t))); + ir_variable *tmp = body.make_temp(type, "tmp"); + do_atan(body, type, tmp, y_over_x); + body.emit(ret(tmp)); return sig; } @@ -2220,39 +3417,39 @@ builtin_builder::_atanh(const glsl_type *type) ir_function_signature * builtin_builder::_pow(const glsl_type *type) { - return binop(ir_binop_pow, always_available, type, type, type); + return binop(always_available, ir_binop_pow, type, type, type); } UNOP(exp, ir_unop_exp, always_available) UNOP(log, ir_unop_log, always_available) UNOP(exp2, ir_unop_exp2, always_available) UNOP(log2, ir_unop_log2, always_available) -UNOP(sqrt, ir_unop_sqrt, always_available) -UNOP(inversesqrt, ir_unop_rsq, always_available) +UNOPA(sqrt, ir_unop_sqrt) +UNOPA(inversesqrt, ir_unop_rsq) /** @} */ -UNOP(abs, ir_unop_abs, always_available) -UNOP(sign, ir_unop_sign, always_available) -UNOP(floor, ir_unop_floor, always_available) -UNOP(trunc, ir_unop_trunc, v130) -UNOP(round, ir_unop_round_even, always_available) -UNOP(roundEven, ir_unop_round_even, always_available) -UNOP(ceil, ir_unop_ceil, always_available) -UNOP(fract, ir_unop_fract, always_available) +UNOPA(abs, ir_unop_abs) +UNOPA(sign, ir_unop_sign) +UNOPA(floor, ir_unop_floor) +UNOPA(trunc, ir_unop_trunc) +UNOPA(round, ir_unop_round_even) +UNOPA(roundEven, ir_unop_round_even) +UNOPA(ceil, ir_unop_ceil) +UNOPA(fract, ir_unop_fract) ir_function_signature * builtin_builder::_mod(const glsl_type *x_type, const glsl_type *y_type) { - return binop(ir_binop_mod, always_available, x_type, x_type, y_type); + return binop(always_available, ir_binop_mod, x_type, x_type, y_type); } ir_function_signature * -builtin_builder::_modf(const glsl_type *type) +builtin_builder::_modf(builtin_available_predicate avail, const glsl_type *type) { ir_variable *x = in_var(type, "x"); ir_variable *i = out_var(type, "i"); - MAKE_SIG(type, v130, 2, x, i); + MAKE_SIG(type, avail, 2, x, i); ir_variable *t = body.make_temp(type, "t"); body.emit(assign(t, expr(ir_unop_trunc, x))); @@ -2266,14 +3463,14 @@ ir_function_signature * builtin_builder::_min(builtin_available_predicate avail, const glsl_type *x_type, const glsl_type *y_type) { - return binop(ir_binop_min, avail, x_type, x_type, y_type); + return binop(avail, ir_binop_min, x_type, x_type, y_type); } ir_function_signature * builtin_builder::_max(builtin_available_predicate avail, const glsl_type *x_type, const glsl_type *y_type) { - return binop(ir_binop_max, avail, x_type, x_type, y_type); + return binop(avail, ir_binop_max, x_type, x_type, y_type); } ir_function_signature * @@ -2291,12 +3488,12 @@ builtin_builder::_clamp(builtin_available_predicate avail, } ir_function_signature * -builtin_builder::_mix_lrp(const glsl_type *val_type, const glsl_type *blend_type) +builtin_builder::_mix_lrp(builtin_available_predicate avail, const glsl_type *val_type, const glsl_type *blend_type) { ir_variable *x = in_var(val_type, "x"); ir_variable *y = in_var(val_type, "y"); ir_variable *a = in_var(blend_type, "a"); - MAKE_SIG(val_type, always_available, 3, x, y, a); + MAKE_SIG(val_type, avail, 3, x, y, a); body.emit(ret(lrp(x, y, a))); @@ -2326,26 +3523,37 @@ builtin_builder::_mix_sel(builtin_available_predicate avail, } ir_function_signature * -builtin_builder::_step(const glsl_type *edge_type, const glsl_type *x_type) +builtin_builder::_step(builtin_available_predicate avail, const glsl_type *edge_type, const glsl_type *x_type) { ir_variable *edge = in_var(edge_type, "edge"); ir_variable *x = in_var(x_type, "x"); - MAKE_SIG(x_type, always_available, 2, edge, x); + MAKE_SIG(x_type, avail, 2, edge, x); ir_variable *t = body.make_temp(x_type, "t"); if (x_type->vector_elements == 1) { /* Both are floats */ - body.emit(assign(t, b2f(gequal(x, edge)))); + if (edge_type->base_type == GLSL_TYPE_DOUBLE) + body.emit(assign(t, f2d(b2f(gequal(x, edge))))); + else + body.emit(assign(t, b2f(gequal(x, edge)))); } else if (edge_type->vector_elements == 1) { /* x is a vector but edge is a float */ for (int i = 0; i < x_type->vector_elements; i++) { - body.emit(assign(t, b2f(gequal(swizzle(x, i, 1), edge)), 1 << i)); + if (edge_type->base_type == GLSL_TYPE_DOUBLE) + body.emit(assign(t, f2d(b2f(gequal(swizzle(x, i, 1), edge))), 1 << i)); + else + body.emit(assign(t, b2f(gequal(swizzle(x, i, 1), edge)), 1 << i)); } } else { /* Both are vectors */ for (int i = 0; i < x_type->vector_elements; i++) { - body.emit(assign(t, b2f(gequal(swizzle(x, i, 1), swizzle(edge, i, 1))), - 1 << i)); + if (edge_type->base_type == GLSL_TYPE_DOUBLE) + body.emit(assign(t, f2d(b2f(gequal(swizzle(x, i, 1), swizzle(edge, i, 1)))), + 1 << i)); + else + body.emit(assign(t, b2f(gequal(swizzle(x, i, 1), swizzle(edge, i, 1))), + 1 << i)); + } } body.emit(ret(t)); @@ -2354,12 +3562,12 @@ builtin_builder::_step(const glsl_type *edge_type, const glsl_type *x_type) } ir_function_signature * -builtin_builder::_smoothstep(const glsl_type *edge_type, const glsl_type *x_type) +builtin_builder::_smoothstep(builtin_available_predicate avail, const glsl_type *edge_type, const glsl_type *x_type) { ir_variable *edge0 = in_var(edge_type, "edge0"); ir_variable *edge1 = in_var(edge_type, "edge1"); ir_variable *x = in_var(x_type, "x"); - MAKE_SIG(x_type, always_available, 3, edge0, edge1, x); + MAKE_SIG(x_type, avail, 3, edge0, edge1, x); /* From the GLSL 1.10 specification: * @@ -2370,18 +3578,18 @@ builtin_builder::_smoothstep(const glsl_type *edge_type, const glsl_type *x_type ir_variable *t = body.make_temp(x_type, "t"); body.emit(assign(t, clamp(div(sub(x, edge0), sub(edge1, edge0)), - imm(0.0f), imm(1.0f)))); + IMM_FP(x_type, 0.0), IMM_FP(x_type, 1.0)))); - body.emit(ret(mul(t, mul(t, sub(imm(3.0f), mul(imm(2.0f), t)))))); + body.emit(ret(mul(t, mul(t, sub(IMM_FP(x_type, 3.0), mul(IMM_FP(x_type, 2.0), t)))))); return sig; } ir_function_signature * -builtin_builder::_isnan(const glsl_type *type) +builtin_builder::_isnan(builtin_available_predicate avail, const glsl_type *type) { ir_variable *x = in_var(type, "x"); - MAKE_SIG(glsl_type::bvec(type->vector_elements), v130, 1, x); + MAKE_SIG(glsl_type::bvec(type->vector_elements), avail, 1, x); body.emit(ret(nequal(x, x))); @@ -2389,14 +3597,23 @@ builtin_builder::_isnan(const glsl_type *type) } ir_function_signature * -builtin_builder::_isinf(const glsl_type *type) +builtin_builder::_isinf(builtin_available_predicate avail, const glsl_type *type) { ir_variable *x = in_var(type, "x"); - MAKE_SIG(glsl_type::bvec(type->vector_elements), v130, 1, x); + MAKE_SIG(glsl_type::bvec(type->vector_elements), avail, 1, x); ir_constant_data infinities; for (int i = 0; i < type->vector_elements; i++) { - infinities.f[i] = std::numeric_limits::infinity(); + switch (type->base_type) { + case GLSL_TYPE_FLOAT: + infinities.f[i] = INFINITY; + break; + case GLSL_TYPE_DOUBLE: + infinities.d[i] = INFINITY; + break; + default: + unreachable("unknown type"); + } } body.emit(ret(equal(abs(x), imm(type, infinities)))); @@ -2532,22 +3749,40 @@ builtin_builder::_unpackHalf2x16(builtin_available_predicate avail) } ir_function_signature * -builtin_builder::_length(const glsl_type *type) +builtin_builder::_packDouble2x32(builtin_available_predicate avail) +{ + ir_variable *v = in_var(glsl_type::uvec2_type, "v"); + MAKE_SIG(glsl_type::double_type, avail, 1, v); + body.emit(ret(expr(ir_unop_pack_double_2x32, v))); + return sig; +} + +ir_function_signature * +builtin_builder::_unpackDouble2x32(builtin_available_predicate avail) +{ + ir_variable *p = in_var(glsl_type::double_type, "p"); + MAKE_SIG(glsl_type::uvec2_type, avail, 1, p); + body.emit(ret(expr(ir_unop_unpack_double_2x32, p))); + return sig; +} + +ir_function_signature * +builtin_builder::_length(builtin_available_predicate avail, const glsl_type *type) { ir_variable *x = in_var(type, "x"); - MAKE_SIG(glsl_type::float_type, always_available, 1, x); + MAKE_SIG(type->get_base_type(), avail, 1, x); - body.emit(ret(sqrt(dotlike(x, x)))); + body.emit(ret(sqrt(dot(x, x)))); return sig; } ir_function_signature * -builtin_builder::_distance(const glsl_type *type) +builtin_builder::_distance(builtin_available_predicate avail, const glsl_type *type) { ir_variable *p0 = in_var(type, "p0"); ir_variable *p1 = in_var(type, "p1"); - MAKE_SIG(glsl_type::float_type, always_available, 2, p0, p1); + MAKE_SIG(type->get_base_type(), avail, 2, p0, p1); if (type->vector_elements == 1) { body.emit(ret(abs(sub(p0, p1)))); @@ -2561,21 +3796,21 @@ builtin_builder::_distance(const glsl_type *type) } ir_function_signature * -builtin_builder::_dot(const glsl_type *type) +builtin_builder::_dot(builtin_available_predicate avail, const glsl_type *type) { if (type->vector_elements == 1) - return binop(ir_binop_mul, always_available, type, type, type); + return binop(avail, ir_binop_mul, type, type, type); - return binop(ir_binop_dot, always_available, - glsl_type::float_type, type, type); + return binop(avail, ir_binop_dot, + type->get_base_type(), type, type); } ir_function_signature * -builtin_builder::_cross(const glsl_type *type) +builtin_builder::_cross(builtin_available_predicate avail, const glsl_type *type) { ir_variable *a = in_var(type, "a"); ir_variable *b = in_var(type, "b"); - MAKE_SIG(type, always_available, 2, a, b); + MAKE_SIG(type, avail, 2, a, b); int yzx = MAKE_SWIZZLE4(SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_X, 0); int zxy = MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_X, SWIZZLE_Y, 0); @@ -2587,10 +3822,10 @@ builtin_builder::_cross(const glsl_type *type) } ir_function_signature * -builtin_builder::_normalize(const glsl_type *type) +builtin_builder::_normalize(builtin_available_predicate avail, const glsl_type *type) { ir_variable *x = in_var(type, "x"); - MAKE_SIG(type, always_available, 1, x); + MAKE_SIG(type, avail, 1, x); if (type->vector_elements == 1) { body.emit(ret(sign(x))); @@ -2620,42 +3855,42 @@ builtin_builder::_ftransform() } ir_function_signature * -builtin_builder::_faceforward(const glsl_type *type) +builtin_builder::_faceforward(builtin_available_predicate avail, const glsl_type *type) { ir_variable *N = in_var(type, "N"); ir_variable *I = in_var(type, "I"); ir_variable *Nref = in_var(type, "Nref"); - MAKE_SIG(type, always_available, 3, N, I, Nref); + MAKE_SIG(type, avail, 3, N, I, Nref); - body.emit(if_tree(less(dotlike(Nref, I), imm(0.0f)), + body.emit(if_tree(less(dot(Nref, I), IMM_FP(type, 0.0)), ret(N), ret(neg(N)))); return sig; } ir_function_signature * -builtin_builder::_reflect(const glsl_type *type) +builtin_builder::_reflect(builtin_available_predicate avail, const glsl_type *type) { ir_variable *I = in_var(type, "I"); ir_variable *N = in_var(type, "N"); - MAKE_SIG(type, always_available, 2, I, N); + MAKE_SIG(type, avail, 2, I, N); /* I - 2 * dot(N, I) * N */ - body.emit(ret(sub(I, mul(imm(2.0f), mul(dotlike(N, I), N))))); + body.emit(ret(sub(I, mul(IMM_FP(type, 2.0), mul(dot(N, I), N))))); return sig; } ir_function_signature * -builtin_builder::_refract(const glsl_type *type) +builtin_builder::_refract(builtin_available_predicate avail, const glsl_type *type) { ir_variable *I = in_var(type, "I"); ir_variable *N = in_var(type, "N"); - ir_variable *eta = in_var(glsl_type::float_type, "eta"); - MAKE_SIG(type, always_available, 3, I, N, eta); + ir_variable *eta = in_var(type->get_base_type(), "eta"); + MAKE_SIG(type, avail, 3, I, N, eta); - ir_variable *n_dot_i = body.make_temp(glsl_type::float_type, "n_dot_i"); - body.emit(assign(n_dot_i, dotlike(N, I))); + ir_variable *n_dot_i = body.make_temp(type->get_base_type(), "n_dot_i"); + body.emit(assign(n_dot_i, dot(N, I))); /* From the GLSL 1.10 specification: * k = 1.0 - eta * eta * (1.0 - dot(N, I) * dot(N, I)) @@ -2664,11 +3899,11 @@ builtin_builder::_refract(const glsl_type *type) * else * return eta * I - (eta * dot(N, I) + sqrt(k)) * N */ - ir_variable *k = body.make_temp(glsl_type::float_type, "k"); - body.emit(assign(k, sub(imm(1.0f), - mul(eta, mul(eta, sub(imm(1.0f), + ir_variable *k = body.make_temp(type->get_base_type(), "k"); + body.emit(assign(k, sub(IMM_FP(type, 1.0), + mul(eta, mul(eta, sub(IMM_FP(type, 1.0), mul(n_dot_i, n_dot_i))))))); - body.emit(if_tree(less(k, imm(0.0f)), + body.emit(if_tree(less(k, IMM_FP(type, 0.0)), ret(ir_constant::zero(mem_ctx, type)), ret(sub(mul(eta, I), mul(add(mul(eta, n_dot_i), sqrt(k)), N))))); @@ -2677,11 +3912,11 @@ builtin_builder::_refract(const glsl_type *type) } ir_function_signature * -builtin_builder::_matrixCompMult(const glsl_type *type) +builtin_builder::_matrixCompMult(builtin_available_predicate avail, const glsl_type *type) { ir_variable *x = in_var(type, "x"); ir_variable *y = in_var(type, "y"); - MAKE_SIG(type, always_available, 2, x, y); + MAKE_SIG(type, avail, 2, x, y); ir_variable *z = body.make_temp(type, "z"); for (int i = 0; i < type->matrix_columns; i++) { @@ -2693,11 +3928,19 @@ builtin_builder::_matrixCompMult(const glsl_type *type) } ir_function_signature * -builtin_builder::_outerProduct(const glsl_type *type) +builtin_builder::_outerProduct(builtin_available_predicate avail, const glsl_type *type) { - ir_variable *c = in_var(glsl_type::vec(type->vector_elements), "c"); - ir_variable *r = in_var(glsl_type::vec(type->matrix_columns), "r"); - MAKE_SIG(type, v120, 2, c, r); + ir_variable *c; + ir_variable *r; + + if (type->base_type == GLSL_TYPE_DOUBLE) { + r = in_var(glsl_type::dvec(type->matrix_columns), "r"); + c = in_var(glsl_type::dvec(type->vector_elements), "c"); + } else { + r = in_var(glsl_type::vec(type->matrix_columns), "r"); + c = in_var(glsl_type::vec(type->vector_elements), "c"); + } + MAKE_SIG(type, avail, 2, c, r); ir_variable *m = body.make_temp(type, "m"); for (int i = 0; i < type->matrix_columns; i++) { @@ -2709,15 +3952,15 @@ builtin_builder::_outerProduct(const glsl_type *type) } ir_function_signature * -builtin_builder::_transpose(const glsl_type *orig_type) +builtin_builder::_transpose(builtin_available_predicate avail, const glsl_type *orig_type) { const glsl_type *transpose_type = - glsl_type::get_instance(GLSL_TYPE_FLOAT, + glsl_type::get_instance(orig_type->base_type, orig_type->matrix_columns, orig_type->vector_elements); ir_variable *m = in_var(orig_type, "m"); - MAKE_SIG(transpose_type, v120, 1, m); + MAKE_SIG(transpose_type, avail, 1, m); ir_variable *t = body.make_temp(transpose_type, "t"); for (int i = 0; i < orig_type->matrix_columns; i++) { @@ -2733,10 +3976,10 @@ builtin_builder::_transpose(const glsl_type *orig_type) } ir_function_signature * -builtin_builder::_determinant_mat2() +builtin_builder::_determinant_mat2(builtin_available_predicate avail, const glsl_type *type) { - ir_variable *m = in_var(glsl_type::mat2_type, "m"); - MAKE_SIG(glsl_type::float_type, v120, 1, m); + ir_variable *m = in_var(type, "m"); + MAKE_SIG(type->get_base_type(), avail, 1, m); body.emit(ret(sub(mul(matrix_elt(m, 0, 0), matrix_elt(m, 1, 1)), mul(matrix_elt(m, 1, 0), matrix_elt(m, 0, 1))))); @@ -2745,10 +3988,10 @@ builtin_builder::_determinant_mat2() } ir_function_signature * -builtin_builder::_determinant_mat3() +builtin_builder::_determinant_mat3(builtin_available_predicate avail, const glsl_type *type) { - ir_variable *m = in_var(glsl_type::mat3_type, "m"); - MAKE_SIG(glsl_type::float_type, v120, 1, m); + ir_variable *m = in_var(type, "m"); + MAKE_SIG(type->get_base_type(), avail, 1, m); ir_expression *f1 = sub(mul(matrix_elt(m, 1, 1), matrix_elt(m, 2, 2)), @@ -2770,30 +4013,31 @@ builtin_builder::_determinant_mat3() } ir_function_signature * -builtin_builder::_determinant_mat4() -{ - ir_variable *m = in_var(glsl_type::mat4_type, "m"); - MAKE_SIG(glsl_type::float_type, v120, 1, m); - - ir_variable *SubFactor00 = body.make_temp(glsl_type::float_type, "SubFactor00"); - ir_variable *SubFactor01 = body.make_temp(glsl_type::float_type, "SubFactor01"); - ir_variable *SubFactor02 = body.make_temp(glsl_type::float_type, "SubFactor02"); - ir_variable *SubFactor03 = body.make_temp(glsl_type::float_type, "SubFactor03"); - ir_variable *SubFactor04 = body.make_temp(glsl_type::float_type, "SubFactor04"); - ir_variable *SubFactor05 = body.make_temp(glsl_type::float_type, "SubFactor05"); - ir_variable *SubFactor06 = body.make_temp(glsl_type::float_type, "SubFactor06"); - ir_variable *SubFactor07 = body.make_temp(glsl_type::float_type, "SubFactor07"); - ir_variable *SubFactor08 = body.make_temp(glsl_type::float_type, "SubFactor08"); - ir_variable *SubFactor09 = body.make_temp(glsl_type::float_type, "SubFactor09"); - ir_variable *SubFactor10 = body.make_temp(glsl_type::float_type, "SubFactor10"); - ir_variable *SubFactor11 = body.make_temp(glsl_type::float_type, "SubFactor11"); - ir_variable *SubFactor12 = body.make_temp(glsl_type::float_type, "SubFactor12"); - ir_variable *SubFactor13 = body.make_temp(glsl_type::float_type, "SubFactor13"); - ir_variable *SubFactor14 = body.make_temp(glsl_type::float_type, "SubFactor14"); - ir_variable *SubFactor15 = body.make_temp(glsl_type::float_type, "SubFactor15"); - ir_variable *SubFactor16 = body.make_temp(glsl_type::float_type, "SubFactor16"); - ir_variable *SubFactor17 = body.make_temp(glsl_type::float_type, "SubFactor17"); - ir_variable *SubFactor18 = body.make_temp(glsl_type::float_type, "SubFactor18"); +builtin_builder::_determinant_mat4(builtin_available_predicate avail, const glsl_type *type) +{ + ir_variable *m = in_var(type, "m"); + const glsl_type *btype = type->get_base_type(); + MAKE_SIG(btype, avail, 1, m); + + ir_variable *SubFactor00 = body.make_temp(btype, "SubFactor00"); + ir_variable *SubFactor01 = body.make_temp(btype, "SubFactor01"); + ir_variable *SubFactor02 = body.make_temp(btype, "SubFactor02"); + ir_variable *SubFactor03 = body.make_temp(btype, "SubFactor03"); + ir_variable *SubFactor04 = body.make_temp(btype, "SubFactor04"); + ir_variable *SubFactor05 = body.make_temp(btype, "SubFactor05"); + ir_variable *SubFactor06 = body.make_temp(btype, "SubFactor06"); + ir_variable *SubFactor07 = body.make_temp(btype, "SubFactor07"); + ir_variable *SubFactor08 = body.make_temp(btype, "SubFactor08"); + ir_variable *SubFactor09 = body.make_temp(btype, "SubFactor09"); + ir_variable *SubFactor10 = body.make_temp(btype, "SubFactor10"); + ir_variable *SubFactor11 = body.make_temp(btype, "SubFactor11"); + ir_variable *SubFactor12 = body.make_temp(btype, "SubFactor12"); + ir_variable *SubFactor13 = body.make_temp(btype, "SubFactor13"); + ir_variable *SubFactor14 = body.make_temp(btype, "SubFactor14"); + ir_variable *SubFactor15 = body.make_temp(btype, "SubFactor15"); + ir_variable *SubFactor16 = body.make_temp(btype, "SubFactor16"); + ir_variable *SubFactor17 = body.make_temp(btype, "SubFactor17"); + ir_variable *SubFactor18 = body.make_temp(btype, "SubFactor18"); body.emit(assign(SubFactor00, sub(mul(matrix_elt(m, 2, 2), matrix_elt(m, 3, 3)), mul(matrix_elt(m, 3, 2), matrix_elt(m, 2, 3))))); body.emit(assign(SubFactor01, sub(mul(matrix_elt(m, 2, 1), matrix_elt(m, 3, 3)), mul(matrix_elt(m, 3, 1), matrix_elt(m, 2, 3))))); @@ -2815,7 +4059,7 @@ builtin_builder::_determinant_mat4() body.emit(assign(SubFactor17, sub(mul(matrix_elt(m, 1, 0), matrix_elt(m, 2, 2)), mul(matrix_elt(m, 2, 0), matrix_elt(m, 1, 2))))); body.emit(assign(SubFactor18, sub(mul(matrix_elt(m, 1, 0), matrix_elt(m, 2, 1)), mul(matrix_elt(m, 2, 0), matrix_elt(m, 1, 1))))); - ir_variable *adj_0 = body.make_temp(glsl_type::vec4_type, "adj_0"); + ir_variable *adj_0 = body.make_temp(btype == glsl_type::float_type ? glsl_type::vec4_type : glsl_type::dvec4_type, "adj_0"); body.emit(assign(adj_0, add(sub(mul(matrix_elt(m, 1, 1), SubFactor00), @@ -2844,12 +4088,12 @@ builtin_builder::_determinant_mat4() } ir_function_signature * -builtin_builder::_inverse_mat2() +builtin_builder::_inverse_mat2(builtin_available_predicate avail, const glsl_type *type) { - ir_variable *m = in_var(glsl_type::mat2_type, "m"); - MAKE_SIG(glsl_type::mat2_type, v120, 1, m); + ir_variable *m = in_var(type, "m"); + MAKE_SIG(type, avail, 1, m); - ir_variable *adj = body.make_temp(glsl_type::mat2_type, "adj"); + ir_variable *adj = body.make_temp(type, "adj"); body.emit(assign(array_ref(adj, 0), matrix_elt(m, 1, 1), 1 << 0)); body.emit(assign(array_ref(adj, 0), neg(matrix_elt(m, 0, 1)), 1 << 1)); body.emit(assign(array_ref(adj, 1), neg(matrix_elt(m, 1, 0)), 1 << 0)); @@ -2864,14 +4108,15 @@ builtin_builder::_inverse_mat2() } ir_function_signature * -builtin_builder::_inverse_mat3() +builtin_builder::_inverse_mat3(builtin_available_predicate avail, const glsl_type *type) { - ir_variable *m = in_var(glsl_type::mat3_type, "m"); - MAKE_SIG(glsl_type::mat3_type, v120, 1, m); + ir_variable *m = in_var(type, "m"); + const glsl_type *btype = type->get_base_type(); + MAKE_SIG(type, avail, 1, m); - ir_variable *f11_22_21_12 = body.make_temp(glsl_type::float_type, "f11_22_21_12"); - ir_variable *f10_22_20_12 = body.make_temp(glsl_type::float_type, "f10_22_20_12"); - ir_variable *f10_21_20_11 = body.make_temp(glsl_type::float_type, "f10_21_20_11"); + ir_variable *f11_22_21_12 = body.make_temp(btype, "f11_22_21_12"); + ir_variable *f10_22_20_12 = body.make_temp(btype, "f10_22_20_12"); + ir_variable *f10_21_20_11 = body.make_temp(btype, "f10_21_20_11"); body.emit(assign(f11_22_21_12, sub(mul(matrix_elt(m, 1, 1), matrix_elt(m, 2, 2)), @@ -2883,7 +4128,7 @@ builtin_builder::_inverse_mat3() sub(mul(matrix_elt(m, 1, 0), matrix_elt(m, 2, 1)), mul(matrix_elt(m, 2, 0), matrix_elt(m, 1, 1))))); - ir_variable *adj = body.make_temp(glsl_type::mat3_type, "adj"); + ir_variable *adj = body.make_temp(type, "adj"); body.emit(assign(array_ref(adj, 0), f11_22_21_12, WRITEMASK_X)); body.emit(assign(array_ref(adj, 1), neg(f10_22_20_12), WRITEMASK_X)); body.emit(assign(array_ref(adj, 2), f10_21_20_11, WRITEMASK_X)); @@ -2925,30 +4170,31 @@ builtin_builder::_inverse_mat3() } ir_function_signature * -builtin_builder::_inverse_mat4() -{ - ir_variable *m = in_var(glsl_type::mat4_type, "m"); - MAKE_SIG(glsl_type::mat4_type, v120, 1, m); - - ir_variable *SubFactor00 = body.make_temp(glsl_type::float_type, "SubFactor00"); - ir_variable *SubFactor01 = body.make_temp(glsl_type::float_type, "SubFactor01"); - ir_variable *SubFactor02 = body.make_temp(glsl_type::float_type, "SubFactor02"); - ir_variable *SubFactor03 = body.make_temp(glsl_type::float_type, "SubFactor03"); - ir_variable *SubFactor04 = body.make_temp(glsl_type::float_type, "SubFactor04"); - ir_variable *SubFactor05 = body.make_temp(glsl_type::float_type, "SubFactor05"); - ir_variable *SubFactor06 = body.make_temp(glsl_type::float_type, "SubFactor06"); - ir_variable *SubFactor07 = body.make_temp(glsl_type::float_type, "SubFactor07"); - ir_variable *SubFactor08 = body.make_temp(glsl_type::float_type, "SubFactor08"); - ir_variable *SubFactor09 = body.make_temp(glsl_type::float_type, "SubFactor09"); - ir_variable *SubFactor10 = body.make_temp(glsl_type::float_type, "SubFactor10"); - ir_variable *SubFactor11 = body.make_temp(glsl_type::float_type, "SubFactor11"); - ir_variable *SubFactor12 = body.make_temp(glsl_type::float_type, "SubFactor12"); - ir_variable *SubFactor13 = body.make_temp(glsl_type::float_type, "SubFactor13"); - ir_variable *SubFactor14 = body.make_temp(glsl_type::float_type, "SubFactor14"); - ir_variable *SubFactor15 = body.make_temp(glsl_type::float_type, "SubFactor15"); - ir_variable *SubFactor16 = body.make_temp(glsl_type::float_type, "SubFactor16"); - ir_variable *SubFactor17 = body.make_temp(glsl_type::float_type, "SubFactor17"); - ir_variable *SubFactor18 = body.make_temp(glsl_type::float_type, "SubFactor18"); +builtin_builder::_inverse_mat4(builtin_available_predicate avail, const glsl_type *type) +{ + ir_variable *m = in_var(type, "m"); + const glsl_type *btype = type->get_base_type(); + MAKE_SIG(type, avail, 1, m); + + ir_variable *SubFactor00 = body.make_temp(btype, "SubFactor00"); + ir_variable *SubFactor01 = body.make_temp(btype, "SubFactor01"); + ir_variable *SubFactor02 = body.make_temp(btype, "SubFactor02"); + ir_variable *SubFactor03 = body.make_temp(btype, "SubFactor03"); + ir_variable *SubFactor04 = body.make_temp(btype, "SubFactor04"); + ir_variable *SubFactor05 = body.make_temp(btype, "SubFactor05"); + ir_variable *SubFactor06 = body.make_temp(btype, "SubFactor06"); + ir_variable *SubFactor07 = body.make_temp(btype, "SubFactor07"); + ir_variable *SubFactor08 = body.make_temp(btype, "SubFactor08"); + ir_variable *SubFactor09 = body.make_temp(btype, "SubFactor09"); + ir_variable *SubFactor10 = body.make_temp(btype, "SubFactor10"); + ir_variable *SubFactor11 = body.make_temp(btype, "SubFactor11"); + ir_variable *SubFactor12 = body.make_temp(btype, "SubFactor12"); + ir_variable *SubFactor13 = body.make_temp(btype, "SubFactor13"); + ir_variable *SubFactor14 = body.make_temp(btype, "SubFactor14"); + ir_variable *SubFactor15 = body.make_temp(btype, "SubFactor15"); + ir_variable *SubFactor16 = body.make_temp(btype, "SubFactor16"); + ir_variable *SubFactor17 = body.make_temp(btype, "SubFactor17"); + ir_variable *SubFactor18 = body.make_temp(btype, "SubFactor18"); body.emit(assign(SubFactor00, sub(mul(matrix_elt(m, 2, 2), matrix_elt(m, 3, 3)), mul(matrix_elt(m, 3, 2), matrix_elt(m, 2, 3))))); body.emit(assign(SubFactor01, sub(mul(matrix_elt(m, 2, 1), matrix_elt(m, 3, 3)), mul(matrix_elt(m, 3, 1), matrix_elt(m, 2, 3))))); @@ -2970,7 +4216,7 @@ builtin_builder::_inverse_mat4() body.emit(assign(SubFactor17, sub(mul(matrix_elt(m, 1, 0), matrix_elt(m, 2, 2)), mul(matrix_elt(m, 2, 0), matrix_elt(m, 1, 2))))); body.emit(assign(SubFactor18, sub(mul(matrix_elt(m, 1, 0), matrix_elt(m, 2, 1)), mul(matrix_elt(m, 2, 0), matrix_elt(m, 1, 1))))); - ir_variable *adj = body.make_temp(glsl_type::mat4_type, "adj"); + ir_variable *adj = body.make_temp(btype == glsl_type::float_type ? glsl_type::mat4_type : glsl_type::dmat4_type, "adj"); body.emit(assign(array_ref(adj, 0), add(sub(mul(matrix_elt(m, 1, 1), SubFactor00), mul(matrix_elt(m, 1, 2), SubFactor01)), @@ -3071,7 +4317,7 @@ ir_function_signature * builtin_builder::_lessThan(builtin_available_predicate avail, const glsl_type *type) { - return binop(ir_binop_less, avail, + return binop(avail, ir_binop_less, glsl_type::bvec(type->vector_elements), type, type); } @@ -3079,7 +4325,7 @@ ir_function_signature * builtin_builder::_lessThanEqual(builtin_available_predicate avail, const glsl_type *type) { - return binop(ir_binop_lequal, avail, + return binop(avail, ir_binop_lequal, glsl_type::bvec(type->vector_elements), type, type); } @@ -3087,7 +4333,7 @@ ir_function_signature * builtin_builder::_greaterThan(builtin_available_predicate avail, const glsl_type *type) { - return binop(ir_binop_greater, avail, + return binop(avail, ir_binop_greater, glsl_type::bvec(type->vector_elements), type, type); } @@ -3095,7 +4341,7 @@ ir_function_signature * builtin_builder::_greaterThanEqual(builtin_available_predicate avail, const glsl_type *type) { - return binop(ir_binop_gequal, avail, + return binop(avail, ir_binop_gequal, glsl_type::bvec(type->vector_elements), type, type); } @@ -3103,7 +4349,7 @@ ir_function_signature * builtin_builder::_equal(builtin_available_predicate avail, const glsl_type *type) { - return binop(ir_binop_equal, avail, + return binop(avail, ir_binop_equal, glsl_type::bvec(type->vector_elements), type, type); } @@ -3111,7 +4357,7 @@ ir_function_signature * builtin_builder::_notEqual(builtin_available_predicate avail, const glsl_type *type) { - return binop(ir_binop_nequal, avail, + return binop(avail, ir_binop_nequal, glsl_type::bvec(type->vector_elements), type, type); } @@ -3187,6 +4433,19 @@ builtin_builder::_textureSize(builtin_available_predicate avail, return sig; } +ir_function_signature * +builtin_builder::_textureSamples(const glsl_type *sampler_type) +{ + ir_variable *s = in_var(sampler_type, "sampler"); + MAKE_SIG(glsl_type::int_type, shader_samples, 1, s); + + ir_texture *tex = new(mem_ctx) ir_texture(ir_texture_samples); + tex->set_sampler(new(mem_ctx) ir_dereference_variable(s), glsl_type::int_type); + body.emit(ret(tex)); + + return sig; +} + ir_function_signature * builtin_builder::_texture(ir_texture_opcode opcode, builtin_available_predicate avail, @@ -3203,7 +4462,7 @@ builtin_builder::_texture(ir_texture_opcode opcode, ir_texture *tex = new(mem_ctx) ir_texture(opcode); tex->set_sampler(var_ref(s), return_type); - const int coord_size = sampler_type->sampler_coordinate_components(); + const int coord_size = sampler_type->coordinate_components(); if (coord_size == coord_type->vector_elements) { tex->coordinate = var_ref(P); @@ -3218,11 +4477,21 @@ builtin_builder::_texture(ir_texture_opcode opcode, if (flags & TEX_PROJECT) tex->projector = swizzle(P, coord_type->vector_elements - 1, 1); - /* The shadow comparitor is normally in the Z component, but a few types - * have sufficiently large coordinates that it's in W. - */ - if (sampler_type->sampler_shadow) - tex->shadow_comparitor = swizzle(P, MAX2(coord_size, SWIZZLE_Z), 1); + if (sampler_type->sampler_shadow) { + if (opcode == ir_tg4) { + /* gather has refz as a separate parameter, immediately after the + * coordinate + */ + ir_variable *refz = in_var(glsl_type::float_type, "refz"); + sig->parameters.push_tail(refz); + tex->shadow_comparitor = var_ref(refz); + } else { + /* The shadow comparitor is normally in the Z component, but a few types + * have sufficiently large coordinates that it's in W. + */ + tex->shadow_comparitor = swizzle(P, MAX2(coord_size, SWIZZLE_Z), 1); + } + } if (opcode == ir_txl) { ir_variable *lod = in_var(glsl_type::float_type, "lod"); @@ -3238,14 +4507,35 @@ builtin_builder::_texture(ir_texture_opcode opcode, tex->lod_info.grad.dPdy = var_ref(dPdy); } - if (flags & TEX_OFFSET) { + if (flags & (TEX_OFFSET | TEX_OFFSET_NONCONST)) { int offset_size = coord_size - (sampler_type->sampler_array ? 1 : 0); ir_variable *offset = - new(mem_ctx) ir_variable(glsl_type::ivec(offset_size), "offset", ir_var_const_in); + new(mem_ctx) ir_variable(glsl_type::ivec(offset_size), "offset", + (flags & TEX_OFFSET) ? ir_var_const_in : ir_var_function_in); sig->parameters.push_tail(offset); tex->offset = var_ref(offset); } + if (flags & TEX_OFFSET_ARRAY) { + ir_variable *offsets = + new(mem_ctx) ir_variable(glsl_type::get_array_instance(glsl_type::ivec2_type, 4), + "offsets", ir_var_const_in); + sig->parameters.push_tail(offsets); + tex->offset = var_ref(offsets); + } + + if (opcode == ir_tg4) { + if (flags & TEX_COMPONENT) { + ir_variable *component = + new(mem_ctx) ir_variable(glsl_type::int_type, "comp", ir_var_const_in); + sig->parameters.push_tail(component); + tex->lod_info.component = var_ref(component); + } + else { + tex->lod_info.component = imm(0); + } + } + /* The "bias" parameter comes /after/ the "offset" parameter, which is * inconsistent with both textureLodOffset and textureGradOffset. */ @@ -3325,7 +4615,28 @@ builtin_builder::_EmitVertex() { MAKE_SIG(glsl_type::void_type, gs_only, 0); - body.emit(new(mem_ctx) ir_emit_vertex()); + ir_rvalue *stream = new(mem_ctx) ir_constant(0, 1); + body.emit(new(mem_ctx) ir_emit_vertex(stream)); + + return sig; +} + +ir_function_signature * +builtin_builder::_EmitStreamVertex(builtin_available_predicate avail, + const glsl_type *stream_type) +{ + /* Section 8.12 (Geometry Shader Functions) of the GLSL 4.0 spec says: + * + * "Emit the current values of output variables to the current output + * primitive on stream stream. The argument to stream must be a constant + * integral expression." + */ + ir_variable *stream = + new(mem_ctx) ir_variable(stream_type, "stream", ir_var_const_in); + + MAKE_SIG(glsl_type::void_type, avail, 1, stream); + + body.emit(new(mem_ctx) ir_emit_vertex(var_ref(stream))); return sig; } @@ -3335,19 +4646,50 @@ builtin_builder::_EndPrimitive() { MAKE_SIG(glsl_type::void_type, gs_only, 0); - body.emit(new(mem_ctx) ir_end_primitive()); + ir_rvalue *stream = new(mem_ctx) ir_constant(0, 1); + body.emit(new(mem_ctx) ir_end_primitive(stream)); + + return sig; +} + +ir_function_signature * +builtin_builder::_EndStreamPrimitive(builtin_available_predicate avail, + const glsl_type *stream_type) +{ + /* Section 8.12 (Geometry Shader Functions) of the GLSL 4.0 spec says: + * + * "Completes the current output primitive on stream stream and starts + * a new one. The argument to stream must be a constant integral + * expression." + */ + ir_variable *stream = + new(mem_ctx) ir_variable(stream_type, "stream", ir_var_const_in); + + MAKE_SIG(glsl_type::void_type, avail, 1, stream); + + body.emit(new(mem_ctx) ir_end_primitive(var_ref(stream))); return sig; } ir_function_signature * -builtin_builder::_textureQueryLod(const glsl_type *sampler_type, +builtin_builder::_barrier() +{ + MAKE_SIG(glsl_type::void_type, barrier_supported, 0); + + body.emit(new(mem_ctx) ir_barrier()); + return sig; +} + +ir_function_signature * +builtin_builder::_textureQueryLod(builtin_available_predicate avail, + const glsl_type *sampler_type, const glsl_type *coord_type) { ir_variable *s = in_var(sampler_type, "sampler"); ir_variable *coord = in_var(coord_type, "coord"); /* The sampler and coordinate always exist; add optional parameters later. */ - MAKE_SIG(glsl_type::vec2_type, texture_query_lod, 2, s, coord); + MAKE_SIG(glsl_type::vec2_type, avail, 2, s, coord); ir_texture *tex = new(mem_ctx) ir_texture(ir_lod); tex->coordinate = var_ref(coord); @@ -3358,8 +4700,46 @@ builtin_builder::_textureQueryLod(const glsl_type *sampler_type, return sig; } +ir_function_signature * +builtin_builder::_textureQueryLevels(const glsl_type *sampler_type) +{ + ir_variable *s = in_var(sampler_type, "sampler"); + const glsl_type *return_type = glsl_type::int_type; + MAKE_SIG(return_type, texture_query_levels, 1, s); + + ir_texture *tex = new(mem_ctx) ir_texture(ir_query_levels); + tex->set_sampler(var_ref(s), return_type); + + body.emit(ret(tex)); + + return sig; +} + +ir_function_signature * +builtin_builder::_textureSamplesIdentical(builtin_available_predicate avail, + const glsl_type *sampler_type, + const glsl_type *coord_type) +{ + ir_variable *s = in_var(sampler_type, "sampler"); + ir_variable *P = in_var(coord_type, "P"); + const glsl_type *return_type = glsl_type::bool_type; + MAKE_SIG(return_type, avail, 2, s, P); + + ir_texture *tex = new(mem_ctx) ir_texture(ir_samples_identical); + tex->coordinate = var_ref(P); + tex->set_sampler(var_ref(s), return_type); + + body.emit(ret(tex)); + + return sig; +} + UNOP(dFdx, ir_unop_dFdx, fs_oes_derivatives) +UNOP(dFdxCoarse, ir_unop_dFdx_coarse, fs_derivative_control) +UNOP(dFdxFine, ir_unop_dFdx_fine, fs_derivative_control) UNOP(dFdy, ir_unop_dFdy, fs_oes_derivatives) +UNOP(dFdyCoarse, ir_unop_dFdy_coarse, fs_derivative_control) +UNOP(dFdyFine, ir_unop_dFdy_fine, fs_derivative_control) ir_function_signature * builtin_builder::_fwidth(const glsl_type *type) @@ -3372,6 +4752,30 @@ builtin_builder::_fwidth(const glsl_type *type) return sig; } +ir_function_signature * +builtin_builder::_fwidthCoarse(const glsl_type *type) +{ + ir_variable *p = in_var(type, "p"); + MAKE_SIG(type, fs_derivative_control, 1, p); + + body.emit(ret(add(abs(expr(ir_unop_dFdx_coarse, p)), + abs(expr(ir_unop_dFdy_coarse, p))))); + + return sig; +} + +ir_function_signature * +builtin_builder::_fwidthFine(const glsl_type *type) +{ + ir_variable *p = in_var(type, "p"); + MAKE_SIG(type, fs_derivative_control, 1, p); + + body.emit(ret(add(abs(expr(ir_unop_dFdx_fine, p)), + abs(expr(ir_unop_dFdy_fine, p))))); + + return sig; +} + ir_function_signature * builtin_builder::_noise1(const glsl_type *type) { @@ -3481,7 +4885,7 @@ builtin_builder::_bitfieldExtract(const glsl_type *type) ir_variable *value = in_var(type, "value"); ir_variable *offset = in_var(glsl_type::int_type, "offset"); ir_variable *bits = in_var(glsl_type::int_type, "bits"); - MAKE_SIG(type, gpu_shader5, 3, value, offset, bits); + MAKE_SIG(type, gpu_shader5_or_es31, 3, value, offset, bits); body.emit(ret(expr(ir_triop_bitfield_extract, value, offset, bits))); @@ -3495,45 +4899,45 @@ builtin_builder::_bitfieldInsert(const glsl_type *type) ir_variable *insert = in_var(type, "insert"); ir_variable *offset = in_var(glsl_type::int_type, "offset"); ir_variable *bits = in_var(glsl_type::int_type, "bits"); - MAKE_SIG(type, gpu_shader5, 4, base, insert, offset, bits); + MAKE_SIG(type, gpu_shader5_or_es31, 4, base, insert, offset, bits); body.emit(ret(bitfield_insert(base, insert, offset, bits))); return sig; } -UNOP(bitfieldReverse, ir_unop_bitfield_reverse, gpu_shader5) +UNOP(bitfieldReverse, ir_unop_bitfield_reverse, gpu_shader5_or_es31) ir_function_signature * builtin_builder::_bitCount(const glsl_type *type) { - return unop(gpu_shader5, ir_unop_bit_count, + return unop(gpu_shader5_or_es31, ir_unop_bit_count, glsl_type::ivec(type->vector_elements), type); } ir_function_signature * builtin_builder::_findLSB(const glsl_type *type) { - return unop(gpu_shader5, ir_unop_find_lsb, + return unop(gpu_shader5_or_es31, ir_unop_find_lsb, glsl_type::ivec(type->vector_elements), type); } ir_function_signature * builtin_builder::_findMSB(const glsl_type *type) { - return unop(gpu_shader5, ir_unop_find_msb, + return unop(gpu_shader5_or_es31, ir_unop_find_msb, glsl_type::ivec(type->vector_elements), type); } ir_function_signature * -builtin_builder::_fma(const glsl_type *type) +builtin_builder::_fma(builtin_available_predicate avail, const glsl_type *type) { ir_variable *a = in_var(type, "a"); ir_variable *b = in_var(type, "b"); ir_variable *c = in_var(type, "c"); - MAKE_SIG(type, gpu_shader5, 3, a, b, c); + MAKE_SIG(type, avail, 3, a, b, c); - body.emit(ret(fma(a, b, c))); + body.emit(ret(ir_builder::fma(a, b, c))); return sig; } @@ -3541,7 +4945,21 @@ builtin_builder::_fma(const glsl_type *type) ir_function_signature * builtin_builder::_ldexp(const glsl_type *x_type, const glsl_type *exp_type) { - return binop(ir_binop_ldexp, gpu_shader5, x_type, x_type, exp_type); + return binop(x_type->base_type == GLSL_TYPE_DOUBLE ? fp64 : gpu_shader5_or_es31, + ir_binop_ldexp, x_type, x_type, exp_type); +} + +ir_function_signature * +builtin_builder::_dfrexp(const glsl_type *x_type, const glsl_type *exp_type) +{ + ir_variable *x = in_var(x_type, "x"); + ir_variable *exponent = out_var(exp_type, "exp"); + MAKE_SIG(x_type, fp64, 2, x, exponent); + + body.emit(assign(exponent, expr(ir_unop_frexp_exp, x))); + + body.emit(ret(expr(ir_unop_frexp_sig, x))); + return sig; } ir_function_signature * @@ -3549,7 +4967,7 @@ builtin_builder::_frexp(const glsl_type *x_type, const glsl_type *exp_type) { ir_variable *x = in_var(x_type, "x"); ir_variable *exponent = out_var(exp_type, "exp"); - MAKE_SIG(x_type, gpu_shader5, 2, x, exponent); + MAKE_SIG(x_type, gpu_shader5_or_es31, 2, x, exponent); const unsigned vec_elem = x_type->vector_elements; const glsl_type *bvec = glsl_type::get_instance(GLSL_TYPE_BOOL, vec_elem, 1); @@ -3591,12 +5009,401 @@ builtin_builder::_frexp(const glsl_type *x_type, const glsl_type *exp_type) return sig; } + +ir_function_signature * +builtin_builder::_uaddCarry(const glsl_type *type) +{ + ir_variable *x = in_var(type, "x"); + ir_variable *y = in_var(type, "y"); + ir_variable *carry = out_var(type, "carry"); + MAKE_SIG(type, gpu_shader5_or_es31, 3, x, y, carry); + + body.emit(assign(carry, ir_builder::carry(x, y))); + body.emit(ret(add(x, y))); + + return sig; +} + +ir_function_signature * +builtin_builder::_usubBorrow(const glsl_type *type) +{ + ir_variable *x = in_var(type, "x"); + ir_variable *y = in_var(type, "y"); + ir_variable *borrow = out_var(type, "borrow"); + MAKE_SIG(type, gpu_shader5_or_es31, 3, x, y, borrow); + + body.emit(assign(borrow, ir_builder::borrow(x, y))); + body.emit(ret(sub(x, y))); + + return sig; +} + +/** + * For both imulExtended() and umulExtended() built-ins. + */ +ir_function_signature * +builtin_builder::_mulExtended(const glsl_type *type) +{ + ir_variable *x = in_var(type, "x"); + ir_variable *y = in_var(type, "y"); + ir_variable *msb = out_var(type, "msb"); + ir_variable *lsb = out_var(type, "lsb"); + MAKE_SIG(glsl_type::void_type, gpu_shader5_or_es31, 4, x, y, msb, lsb); + + body.emit(assign(msb, imul_high(x, y))); + body.emit(assign(lsb, mul(x, y))); + + return sig; +} + +ir_function_signature * +builtin_builder::_interpolateAtCentroid(const glsl_type *type) +{ + ir_variable *interpolant = in_var(type, "interpolant"); + interpolant->data.must_be_shader_input = 1; + MAKE_SIG(type, fs_gpu_shader5, 1, interpolant); + + body.emit(ret(interpolate_at_centroid(interpolant))); + + return sig; +} + +ir_function_signature * +builtin_builder::_interpolateAtOffset(const glsl_type *type) +{ + ir_variable *interpolant = in_var(type, "interpolant"); + interpolant->data.must_be_shader_input = 1; + ir_variable *offset = in_var(glsl_type::vec2_type, "offset"); + MAKE_SIG(type, fs_gpu_shader5, 2, interpolant, offset); + + body.emit(ret(interpolate_at_offset(interpolant, offset))); + + return sig; +} + +ir_function_signature * +builtin_builder::_interpolateAtSample(const glsl_type *type) +{ + ir_variable *interpolant = in_var(type, "interpolant"); + interpolant->data.must_be_shader_input = 1; + ir_variable *sample_num = in_var(glsl_type::int_type, "sample_num"); + MAKE_SIG(type, fs_gpu_shader5, 2, interpolant, sample_num); + + body.emit(ret(interpolate_at_sample(interpolant, sample_num))); + + return sig; +} + +ir_function_signature * +builtin_builder::_atomic_counter_intrinsic(builtin_available_predicate avail) +{ + ir_variable *counter = in_var(glsl_type::atomic_uint_type, "counter"); + MAKE_INTRINSIC(glsl_type::uint_type, avail, 1, counter); + return sig; +} + +ir_function_signature * +builtin_builder::_atomic_intrinsic2(builtin_available_predicate avail, + const glsl_type *type) +{ + ir_variable *atomic = in_var(type, "atomic"); + ir_variable *data = in_var(type, "data"); + MAKE_INTRINSIC(type, avail, 2, atomic, data); + return sig; +} + +ir_function_signature * +builtin_builder::_atomic_intrinsic3(builtin_available_predicate avail, + const glsl_type *type) +{ + ir_variable *atomic = in_var(type, "atomic"); + ir_variable *data1 = in_var(type, "data1"); + ir_variable *data2 = in_var(type, "data2"); + MAKE_INTRINSIC(type, avail, 3, atomic, data1, data2); + return sig; +} + +ir_function_signature * +builtin_builder::_atomic_counter_op(const char *intrinsic, + builtin_available_predicate avail) +{ + ir_variable *counter = in_var(glsl_type::atomic_uint_type, "atomic_counter"); + MAKE_SIG(glsl_type::uint_type, avail, 1, counter); + + ir_variable *retval = body.make_temp(glsl_type::uint_type, "atomic_retval"); + body.emit(call(shader->symbols->get_function(intrinsic), retval, + sig->parameters)); + body.emit(ret(retval)); + return sig; +} + +ir_function_signature * +builtin_builder::_atomic_op2(const char *intrinsic, + builtin_available_predicate avail, + const glsl_type *type) +{ + ir_variable *atomic = in_var(type, "atomic_var"); + ir_variable *data = in_var(type, "atomic_data"); + MAKE_SIG(type, avail, 2, atomic, data); + + ir_variable *retval = body.make_temp(type, "atomic_retval"); + body.emit(call(shader->symbols->get_function(intrinsic), retval, + sig->parameters)); + body.emit(ret(retval)); + return sig; +} + +ir_function_signature * +builtin_builder::_atomic_op3(const char *intrinsic, + builtin_available_predicate avail, + const glsl_type *type) +{ + ir_variable *atomic = in_var(type, "atomic_var"); + ir_variable *data1 = in_var(type, "atomic_data1"); + ir_variable *data2 = in_var(type, "atomic_data2"); + MAKE_SIG(type, avail, 3, atomic, data1, data2); + + ir_variable *retval = body.make_temp(type, "atomic_retval"); + body.emit(call(shader->symbols->get_function(intrinsic), retval, + sig->parameters)); + body.emit(ret(retval)); + return sig; +} + +ir_function_signature * +builtin_builder::_min3(const glsl_type *type) +{ + ir_variable *x = in_var(type, "x"); + ir_variable *y = in_var(type, "y"); + ir_variable *z = in_var(type, "z"); + MAKE_SIG(type, shader_trinary_minmax, 3, x, y, z); + + ir_expression *min3 = min2(x, min2(y,z)); + body.emit(ret(min3)); + + return sig; +} + +ir_function_signature * +builtin_builder::_max3(const glsl_type *type) +{ + ir_variable *x = in_var(type, "x"); + ir_variable *y = in_var(type, "y"); + ir_variable *z = in_var(type, "z"); + MAKE_SIG(type, shader_trinary_minmax, 3, x, y, z); + + ir_expression *max3 = max2(x, max2(y,z)); + body.emit(ret(max3)); + + return sig; +} + +ir_function_signature * +builtin_builder::_mid3(const glsl_type *type) +{ + ir_variable *x = in_var(type, "x"); + ir_variable *y = in_var(type, "y"); + ir_variable *z = in_var(type, "z"); + MAKE_SIG(type, shader_trinary_minmax, 3, x, y, z); + + ir_expression *mid3 = max2(min2(x, y), max2(min2(x, z), min2(y, z))); + body.emit(ret(mid3)); + + return sig; +} + +ir_function_signature * +builtin_builder::_image_prototype(const glsl_type *image_type, + unsigned num_arguments, + unsigned flags) +{ + const glsl_type *data_type = glsl_type::get_instance( + image_type->sampler_type, + (flags & IMAGE_FUNCTION_HAS_VECTOR_DATA_TYPE ? 4 : 1), + 1); + const glsl_type *ret_type = (flags & IMAGE_FUNCTION_RETURNS_VOID ? + glsl_type::void_type : data_type); + + /* Addressing arguments that are always present. */ + ir_variable *image = in_var(image_type, "image"); + ir_variable *coord = in_var( + glsl_type::ivec(image_type->coordinate_components()), "coord"); + + const builtin_available_predicate avail = + (flags & IMAGE_FUNCTION_AVAIL_ATOMIC ? shader_image_atomic : + shader_image_load_store); + ir_function_signature *sig = new_sig(ret_type, avail, 2, image, coord); + + /* Sample index for multisample images. */ + if (image_type->sampler_dimensionality == GLSL_SAMPLER_DIM_MS) + sig->parameters.push_tail(in_var(glsl_type::int_type, "sample")); + + /* Data arguments. */ + for (unsigned i = 0; i < num_arguments; ++i) { + char *arg_name = ralloc_asprintf(NULL, "arg%d", i); + sig->parameters.push_tail(in_var(data_type, arg_name)); + ralloc_free(arg_name); + } + + /* Set the maximal set of qualifiers allowed for this image + * built-in. Function calls with arguments having fewer + * qualifiers than present in the prototype are allowed by the + * spec, but not with more, i.e. this will make the compiler + * accept everything that needs to be accepted, and reject cases + * like loads from write-only or stores to read-only images. + */ + image->data.image_read_only = (flags & IMAGE_FUNCTION_READ_ONLY) != 0; + image->data.image_write_only = (flags & IMAGE_FUNCTION_WRITE_ONLY) != 0; + image->data.image_coherent = true; + image->data.image_volatile = true; + image->data.image_restrict = true; + + return sig; +} + +ir_function_signature * +builtin_builder::_image_size_prototype(const glsl_type *image_type, + unsigned /* num_arguments */, + unsigned /* flags */) +{ + const glsl_type *ret_type; + unsigned num_components = image_type->coordinate_components(); + + /* From the ARB_shader_image_size extension: + * "Cube images return the dimensions of one face." + */ + if (image_type->sampler_dimensionality == GLSL_SAMPLER_DIM_CUBE && + !image_type->sampler_array) { + num_components = 2; + } + + /* FIXME: Add the highp precision qualifier for GLES 3.10 when it is + * supported by mesa. + */ + ret_type = glsl_type::get_instance(GLSL_TYPE_INT, num_components, 1); + + ir_variable *image = in_var(image_type, "image"); + ir_function_signature *sig = new_sig(ret_type, shader_image_size, 1, image); + + /* Set the maximal set of qualifiers allowed for this image + * built-in. Function calls with arguments having fewer + * qualifiers than present in the prototype are allowed by the + * spec, but not with more, i.e. this will make the compiler + * accept everything that needs to be accepted, and reject cases + * like loads from write-only or stores to read-only images. + */ + image->data.image_read_only = true; + image->data.image_write_only = true; + image->data.image_coherent = true; + image->data.image_volatile = true; + image->data.image_restrict = true; + + return sig; +} + +ir_function_signature * +builtin_builder::_image_samples_prototype(const glsl_type *image_type, + unsigned /* num_arguments */, + unsigned /* flags */) +{ + ir_variable *image = in_var(image_type, "image"); + ir_function_signature *sig = + new_sig(glsl_type::int_type, shader_samples, 1, image); + + /* Set the maximal set of qualifiers allowed for this image + * built-in. Function calls with arguments having fewer + * qualifiers than present in the prototype are allowed by the + * spec, but not with more, i.e. this will make the compiler + * accept everything that needs to be accepted, and reject cases + * like loads from write-only or stores to read-only images. + */ + image->data.image_read_only = true; + image->data.image_write_only = true; + image->data.image_coherent = true; + image->data.image_volatile = true; + image->data.image_restrict = true; + + return sig; +} + +ir_function_signature * +builtin_builder::_image(image_prototype_ctr prototype, + const glsl_type *image_type, + const char *intrinsic_name, + unsigned num_arguments, + unsigned flags) +{ + ir_function_signature *sig = (this->*prototype)(image_type, + num_arguments, flags); + + if (flags & IMAGE_FUNCTION_EMIT_STUB) { + ir_factory body(&sig->body, mem_ctx); + ir_function *f = shader->symbols->get_function(intrinsic_name); + + if (flags & IMAGE_FUNCTION_RETURNS_VOID) { + body.emit(call(f, NULL, sig->parameters)); + } else { + ir_variable *ret_val = + body.make_temp(sig->return_type, "_ret_val"); + body.emit(call(f, ret_val, sig->parameters)); + body.emit(ret(ret_val)); + } + + sig->is_defined = true; + + } else { + sig->is_intrinsic = true; + } + + return sig; +} + +ir_function_signature * +builtin_builder::_memory_barrier_intrinsic(builtin_available_predicate avail) +{ + MAKE_INTRINSIC(glsl_type::void_type, avail, 0); + return sig; +} + +ir_function_signature * +builtin_builder::_memory_barrier(const char *intrinsic_name, + builtin_available_predicate avail) +{ + MAKE_SIG(glsl_type::void_type, avail, 0); + body.emit(call(shader->symbols->get_function(intrinsic_name), + NULL, sig->parameters)); + return sig; +} + +ir_function_signature * +builtin_builder::_shader_clock_intrinsic(builtin_available_predicate avail, + const glsl_type *type) +{ + MAKE_INTRINSIC(type, avail, 0); + return sig; +} + +ir_function_signature * +builtin_builder::_shader_clock(builtin_available_predicate avail, + const glsl_type *type) +{ + MAKE_SIG(type, avail, 0); + + ir_variable *retval = body.make_temp(type, "clock_retval"); + + body.emit(call(shader->symbols->get_function("__intrinsic_shader_clock"), + retval, sig->parameters)); + body.emit(ret(retval)); + return sig; +} + /** @} */ /******************************************************************************/ /* The singleton instance of builtin_builder. */ static builtin_builder builtins; +static mtx_t builtins_lock = _MTX_INITIALIZER_NP; /** * External API (exposing the built-in module to the rest of the compiler): @@ -3605,19 +5412,72 @@ static builtin_builder builtins; void _mesa_glsl_initialize_builtin_functions() { + mtx_lock(&builtins_lock); builtins.initialize(); + mtx_unlock(&builtins_lock); } void _mesa_glsl_release_builtin_functions() { + mtx_lock(&builtins_lock); builtins.release(); + mtx_unlock(&builtins_lock); } ir_function_signature * _mesa_glsl_find_builtin_function(_mesa_glsl_parse_state *state, const char *name, exec_list *actual_parameters) { - return builtins.find(state, name, actual_parameters); + ir_function_signature * s; + mtx_lock(&builtins_lock); + s = builtins.find(state, name, actual_parameters); + mtx_unlock(&builtins_lock); + return s; +} + +ir_function * +_mesa_glsl_find_builtin_function_by_name(const char *name) +{ + ir_function *f; + mtx_lock(&builtins_lock); + f = builtins.shader->symbols->get_function(name); + mtx_unlock(&builtins_lock); + return f; } + +gl_shader * +_mesa_glsl_get_builtin_function_shader() +{ + return builtins.shader; +} + + +/** + * Get the function signature for main from a shader + */ +ir_function_signature * +_mesa_get_main_function_signature(gl_shader *sh) +{ + ir_function *const f = sh->symbols->get_function("main"); + if (f != NULL) { + exec_list void_parameters; + + /* Look for the 'void main()' signature and ensure that it's defined. + * This keeps the linker from accidentally pick a shader that just + * contains a prototype for main. + * + * We don't have to check for multiple definitions of main (in multiple + * shaders) because that would have already been caught above. + */ + ir_function_signature *sig = + f->matching_signature(NULL, &void_parameters, false); + if ((sig != NULL) && sig->is_defined) { + return sig; + } + } + + return NULL; +} + /** @} */