* name and parameters.
*/
+
+/**
+ * Unfortunately, some versions of MinGW produce bad code if this file
+ * is compiled with -O2 or -O3. The resulting driver will crash in random
+ * places if the app uses GLSL.
+ * The work-around is to disable optimizations for just this file. Luckily,
+ * this code is basically just executed once.
+ *
+ * MinGW 4.6.3 (in Ubuntu 13.10) does not have this bug.
+ * MinGW 5.3.1 (in Ubuntu 16.04) definitely has this bug.
+ * MinGW 6.2.0 (in Ubuntu 16.10) definitely has this bug.
+ * MinGW x.y.z - don't know. Assume versions after 4.6.x are buggy
+ */
+
+#if defined(__MINGW32__) && ((__GNUC__ * 100) + __GNUC_MINOR >= 407)
+#warning "disabling optimizations for this file to work around compiler bug"
+#pragma GCC optimize("O1")
+#endif
+
+
#include <stdarg.h>
#include <stdio.h>
-#include "main/core.h" /* for struct gl_shader */
+#include "main/mtypes.h"
#include "main/shaderobj.h"
#include "ir_builder.h"
#include "glsl_parser_extras.h"
#include "program/prog_instruction.h"
#include <math.h>
#include "builtin_functions.h"
+#include "util/hash_table.h"
#define M_PIf ((float) M_PI)
#define M_PI_2f ((float) M_PI_2)
compatibility_vs_only(const _mesa_glsl_parse_state *state)
{
return state->stage == MESA_SHADER_VERTEX &&
- state->language_version <= 130 &&
+ (state->compat_shader || state->ARB_compatibility_enable) &&
!state->es_shader;
}
return state->is_version(130, 0);
}
+static bool
+v460_desktop(const _mesa_glsl_parse_state *state)
+{
+ return state->is_version(460, 0);
+}
+
static bool
v130_fs_only(const _mesa_glsl_parse_state *state)
{
return state->OES_EGL_image_external_enable;
}
+static bool
+texture_external_es3(const _mesa_glsl_parse_state *state)
+{
+ return state->OES_EGL_image_external_essl3_enable &&
+ state->es_shader &&
+ state->is_version(0, 300);
+}
+
/** True if texturing functions with explicit LOD are allowed. */
static bool
lod_exists_in_stage(const _mesa_glsl_parse_state *state)
return state->ARB_shader_atomic_counter_ops_enable;
}
+static bool
+shader_atomic_counter_ops_or_v460_desktop(const _mesa_glsl_parse_state *state)
+{
+ return state->ARB_shader_atomic_counter_ops_enable || v460_desktop(state);
+}
+
+static bool
+shader_ballot(const _mesa_glsl_parse_state *state)
+{
+ return state->ARB_shader_ballot_enable;
+}
+
static bool
shader_clock(const _mesa_glsl_parse_state *state)
{
return state->stage == MESA_SHADER_COMPUTE;
}
+static bool
+compute_shader_supported(const _mesa_glsl_parse_state *state)
+{
+ return state->has_compute_shader();
+}
+
static bool
buffer_atomics_supported(const _mesa_glsl_parse_state *state)
{
return state->ARB_shader_group_vote_enable;
}
+static bool
+vote_or_v460_desktop(const _mesa_glsl_parse_state *state)
+{
+ return state->ARB_shader_group_vote_enable || v460_desktop(state);
+}
+
static bool
integer_functions_supported(const _mesa_glsl_parse_state *state)
{
ir_expression_operation opcode,
const glsl_type *return_type,
const glsl_type *param0_type,
- const glsl_type *param1_type);
+ const glsl_type *param1_type,
+ bool swap_operands = false);
#define B0(X) ir_function_signature *_##X();
#define B1(X) ir_function_signature *_##X(const glsl_type *);
B1(all);
B1(not);
BA2(textureSize);
- B1(textureSamples);
+ BA1(textureSamples);
/** Flags to _texture() */
#define TEX_PROJECT 1
const glsl_type *sampler_type,
const glsl_type *coord_type,
int flags = 0);
- B0(textureCubeArrayShadow);
+ BA1(textureCubeArrayShadow);
ir_function_signature *_texelFetch(builtin_available_predicate avail,
const glsl_type *return_type,
const glsl_type *sampler_type,
B0(barrier)
BA2(textureQueryLod);
- B1(textureQueryLevels);
+ BA1(textureQueryLevels);
BA2(textureSamplesIdentical);
B1(dFdx);
B1(dFdy);
ir_function_signature *_memory_barrier(const char *intrinsic_name,
builtin_available_predicate avail);
+ ir_function_signature *_ballot_intrinsic();
+ ir_function_signature *_ballot();
+ ir_function_signature *_read_first_invocation_intrinsic(const glsl_type *type);
+ ir_function_signature *_read_first_invocation(const glsl_type *type);
+ ir_function_signature *_read_invocation_intrinsic(const glsl_type *type);
+ ir_function_signature *_read_invocation(const glsl_type *type);
+
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);
- ir_function_signature *_vote(enum ir_expression_operation opcode);
+ ir_function_signature *_vote_intrinsic(builtin_available_predicate avail,
+ enum ir_intrinsic_id id);
+ ir_function_signature *_vote(const char *intrinsic_name,
+ builtin_available_predicate avail);
#undef B0
#undef B1
_atomic_intrinsic2(buffer_atomics_supported,
glsl_type::int_type,
ir_intrinsic_generic_atomic_add),
- _atomic_counter_intrinsic1(shader_atomic_counter_ops,
+ _atomic_counter_intrinsic1(shader_atomic_counter_ops_or_v460_desktop,
ir_intrinsic_atomic_counter_add),
NULL);
add_function("__intrinsic_atomic_min",
_atomic_intrinsic2(buffer_atomics_supported,
glsl_type::int_type,
ir_intrinsic_generic_atomic_min),
- _atomic_counter_intrinsic1(shader_atomic_counter_ops,
+ _atomic_counter_intrinsic1(shader_atomic_counter_ops_or_v460_desktop,
ir_intrinsic_atomic_counter_min),
NULL);
add_function("__intrinsic_atomic_max",
_atomic_intrinsic2(buffer_atomics_supported,
glsl_type::int_type,
ir_intrinsic_generic_atomic_max),
- _atomic_counter_intrinsic1(shader_atomic_counter_ops,
+ _atomic_counter_intrinsic1(shader_atomic_counter_ops_or_v460_desktop,
ir_intrinsic_atomic_counter_max),
NULL);
add_function("__intrinsic_atomic_and",
_atomic_intrinsic2(buffer_atomics_supported,
glsl_type::int_type,
ir_intrinsic_generic_atomic_and),
- _atomic_counter_intrinsic1(shader_atomic_counter_ops,
+ _atomic_counter_intrinsic1(shader_atomic_counter_ops_or_v460_desktop,
ir_intrinsic_atomic_counter_and),
NULL);
add_function("__intrinsic_atomic_or",
_atomic_intrinsic2(buffer_atomics_supported,
glsl_type::int_type,
ir_intrinsic_generic_atomic_or),
- _atomic_counter_intrinsic1(shader_atomic_counter_ops,
+ _atomic_counter_intrinsic1(shader_atomic_counter_ops_or_v460_desktop,
ir_intrinsic_atomic_counter_or),
NULL);
add_function("__intrinsic_atomic_xor",
_atomic_intrinsic2(buffer_atomics_supported,
glsl_type::int_type,
ir_intrinsic_generic_atomic_xor),
- _atomic_counter_intrinsic1(shader_atomic_counter_ops,
+ _atomic_counter_intrinsic1(shader_atomic_counter_ops_or_v460_desktop,
ir_intrinsic_atomic_counter_xor),
NULL);
add_function("__intrinsic_atomic_exchange",
_atomic_intrinsic2(buffer_atomics_supported,
glsl_type::int_type,
ir_intrinsic_generic_atomic_exchange),
- _atomic_counter_intrinsic1(shader_atomic_counter_ops,
+ _atomic_counter_intrinsic1(shader_atomic_counter_ops_or_v460_desktop,
ir_intrinsic_atomic_counter_exchange),
NULL);
add_function("__intrinsic_atomic_comp_swap",
_atomic_intrinsic3(buffer_atomics_supported,
glsl_type::int_type,
ir_intrinsic_generic_atomic_comp_swap),
- _atomic_counter_intrinsic2(shader_atomic_counter_ops,
+ _atomic_counter_intrinsic2(shader_atomic_counter_ops_or_v460_desktop,
ir_intrinsic_atomic_counter_comp_swap),
NULL);
ir_intrinsic_group_memory_barrier),
NULL);
add_function("__intrinsic_memory_barrier_atomic_counter",
- _memory_barrier_intrinsic(compute_shader,
+ _memory_barrier_intrinsic(compute_shader_supported,
ir_intrinsic_memory_barrier_atomic_counter),
NULL);
add_function("__intrinsic_memory_barrier_buffer",
- _memory_barrier_intrinsic(compute_shader,
+ _memory_barrier_intrinsic(compute_shader_supported,
ir_intrinsic_memory_barrier_buffer),
NULL);
add_function("__intrinsic_memory_barrier_image",
- _memory_barrier_intrinsic(compute_shader,
+ _memory_barrier_intrinsic(compute_shader_supported,
ir_intrinsic_memory_barrier_image),
NULL);
add_function("__intrinsic_memory_barrier_shared",
_shader_clock_intrinsic(shader_clock,
glsl_type::uvec2_type),
NULL);
+
+ add_function("__intrinsic_vote_all",
+ _vote_intrinsic(vote_or_v460_desktop, ir_intrinsic_vote_all),
+ NULL);
+ add_function("__intrinsic_vote_any",
+ _vote_intrinsic(vote_or_v460_desktop, ir_intrinsic_vote_any),
+ NULL);
+ add_function("__intrinsic_vote_eq",
+ _vote_intrinsic(vote_or_v460_desktop, ir_intrinsic_vote_eq),
+ NULL);
+
+ add_function("__intrinsic_ballot", _ballot_intrinsic(), NULL);
+
+ add_function("__intrinsic_read_invocation",
+ _read_invocation_intrinsic(glsl_type::float_type),
+ _read_invocation_intrinsic(glsl_type::vec2_type),
+ _read_invocation_intrinsic(glsl_type::vec3_type),
+ _read_invocation_intrinsic(glsl_type::vec4_type),
+
+ _read_invocation_intrinsic(glsl_type::int_type),
+ _read_invocation_intrinsic(glsl_type::ivec2_type),
+ _read_invocation_intrinsic(glsl_type::ivec3_type),
+ _read_invocation_intrinsic(glsl_type::ivec4_type),
+
+ _read_invocation_intrinsic(glsl_type::uint_type),
+ _read_invocation_intrinsic(glsl_type::uvec2_type),
+ _read_invocation_intrinsic(glsl_type::uvec3_type),
+ _read_invocation_intrinsic(glsl_type::uvec4_type),
+ NULL);
+
+ add_function("__intrinsic_read_first_invocation",
+ _read_first_invocation_intrinsic(glsl_type::float_type),
+ _read_first_invocation_intrinsic(glsl_type::vec2_type),
+ _read_first_invocation_intrinsic(glsl_type::vec3_type),
+ _read_first_invocation_intrinsic(glsl_type::vec4_type),
+
+ _read_first_invocation_intrinsic(glsl_type::int_type),
+ _read_first_invocation_intrinsic(glsl_type::ivec2_type),
+ _read_first_invocation_intrinsic(glsl_type::ivec3_type),
+ _read_first_invocation_intrinsic(glsl_type::ivec4_type),
+
+ _read_first_invocation_intrinsic(glsl_type::uint_type),
+ _read_first_invocation_intrinsic(glsl_type::uvec2_type),
+ _read_first_invocation_intrinsic(glsl_type::uvec3_type),
+ _read_first_invocation_intrinsic(glsl_type::uvec4_type),
+ NULL);
+
}
/**
NULL);
add_function("textureSamples",
- _textureSamples(glsl_type::sampler2DMS_type),
- _textureSamples(glsl_type::isampler2DMS_type),
- _textureSamples(glsl_type::usampler2DMS_type),
+ _textureSamples(shader_samples, glsl_type::sampler2DMS_type),
+ _textureSamples(shader_samples, glsl_type::isampler2DMS_type),
+ _textureSamples(shader_samples, glsl_type::usampler2DMS_type),
- _textureSamples(glsl_type::sampler2DMSArray_type),
- _textureSamples(glsl_type::isampler2DMSArray_type),
- _textureSamples(glsl_type::usampler2DMSArray_type),
+ _textureSamples(shader_samples, glsl_type::sampler2DMSArray_type),
+ _textureSamples(shader_samples, glsl_type::isampler2DMSArray_type),
+ _textureSamples(shader_samples, glsl_type::usampler2DMSArray_type),
NULL);
add_function("texture",
/* samplerCubeArrayShadow is special; it has an extra parameter
* for the shadow comparator since there is no vec5 type.
*/
- _textureCubeArrayShadow(),
+ _textureCubeArrayShadow(texture_cube_map_array, glsl_type::samplerCubeArrayShadow_type),
_texture(ir_tex, v130, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec2_type),
_texture(ir_tex, v130, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec2_type),
_texture(ir_tex, v130, glsl_type::float_type, glsl_type::sampler2DRectShadow_type, glsl_type::vec3_type),
+ _texture(ir_tex, texture_external_es3, glsl_type::vec4_type, glsl_type::samplerExternalOES_type, glsl_type::vec2_type),
+
_texture(ir_txb, v130_fs_only, glsl_type::vec4_type, glsl_type::sampler1D_type, glsl_type::float_type),
_texture(ir_txb, v130_fs_only, glsl_type::ivec4_type, glsl_type::isampler1D_type, glsl_type::float_type),
_texture(ir_txb, v130_fs_only, glsl_type::uvec4_type, glsl_type::usampler1D_type, glsl_type::float_type),
_texture(ir_tex, v130, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec3_type, TEX_PROJECT),
_texture(ir_tex, v130, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec3_type, TEX_PROJECT),
+ _texture(ir_tex, texture_external_es3, glsl_type::vec4_type, glsl_type::samplerExternalOES_type, glsl_type::vec3_type, TEX_PROJECT),
+ _texture(ir_tex, texture_external_es3, glsl_type::vec4_type, glsl_type::samplerExternalOES_type, glsl_type::vec4_type, TEX_PROJECT),
+
_texture(ir_tex, v130, glsl_type::uvec4_type, glsl_type::usampler2DRect_type, glsl_type::vec3_type, TEX_PROJECT),
_texture(ir_tex, v130, glsl_type::vec4_type, glsl_type::sampler2DRect_type, glsl_type::vec4_type, TEX_PROJECT),
_texture(ir_tex, v130, glsl_type::ivec4_type, glsl_type::isampler2DRect_type, glsl_type::vec4_type, TEX_PROJECT),
_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),
+
+ _texelFetch(texture_external_es3, glsl_type::vec4_type, glsl_type::samplerExternalOES_type, glsl_type::ivec2_type),
+
NULL);
+
add_function("texelFetchOffset",
_texelFetch(v130, glsl_type::vec4_type, glsl_type::sampler1D_type, glsl_type::int_type, glsl_type::int_type),
_texelFetch(v130, glsl_type::ivec4_type, glsl_type::isampler1D_type, glsl_type::int_type, glsl_type::int_type),
NULL);
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),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler1D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler2D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler3D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::samplerCube_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler1DArray_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler2DArray_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::samplerCubeArray_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler1DShadow_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler2DShadow_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::samplerCubeShadow_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler1DArrayShadow_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::sampler2DArrayShadow_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::samplerCubeArrayShadow_type),
+
+ _textureQueryLevels(texture_query_levels, glsl_type::isampler1D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::isampler2D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::isampler3D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::isamplerCube_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::isampler1DArray_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::isampler2DArray_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::isamplerCubeArray_type),
+
+ _textureQueryLevels(texture_query_levels, glsl_type::usampler1D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::usampler2D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::usampler3D_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::usamplerCube_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::usampler1DArray_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::usampler2DArray_type),
+ _textureQueryLevels(texture_query_levels, glsl_type::usamplerCubeArray_type),
NULL);
shader_atomic_counter_ops),
NULL);
+ add_function("atomicCounterAdd",
+ _atomic_counter_op1("__intrinsic_atomic_add",
+ v460_desktop),
+ NULL);
+ add_function("atomicCounterSubtract",
+ _atomic_counter_op1("__intrinsic_atomic_sub",
+ v460_desktop),
+ NULL);
+ add_function("atomicCounterMin",
+ _atomic_counter_op1("__intrinsic_atomic_min",
+ v460_desktop),
+ NULL);
+ add_function("atomicCounterMax",
+ _atomic_counter_op1("__intrinsic_atomic_max",
+ v460_desktop),
+ NULL);
+ add_function("atomicCounterAnd",
+ _atomic_counter_op1("__intrinsic_atomic_and",
+ v460_desktop),
+ NULL);
+ add_function("atomicCounterOr",
+ _atomic_counter_op1("__intrinsic_atomic_or",
+ v460_desktop),
+ NULL);
+ add_function("atomicCounterXor",
+ _atomic_counter_op1("__intrinsic_atomic_xor",
+ v460_desktop),
+ NULL);
+ add_function("atomicCounterExchange",
+ _atomic_counter_op1("__intrinsic_atomic_exchange",
+ v460_desktop),
+ NULL);
+ add_function("atomicCounterCompSwap",
+ _atomic_counter_op2("__intrinsic_atomic_comp_swap",
+ v460_desktop),
+ NULL);
+
add_function("atomicAdd",
_atomic_op2("__intrinsic_atomic_add",
buffer_atomics_supported,
NULL);
add_function("memoryBarrierAtomicCounter",
_memory_barrier("__intrinsic_memory_barrier_atomic_counter",
- compute_shader),
+ compute_shader_supported),
NULL);
add_function("memoryBarrierBuffer",
_memory_barrier("__intrinsic_memory_barrier_buffer",
- compute_shader),
+ compute_shader_supported),
NULL);
add_function("memoryBarrierImage",
_memory_barrier("__intrinsic_memory_barrier_image",
- compute_shader),
+ compute_shader_supported),
NULL);
add_function("memoryBarrierShared",
_memory_barrier("__intrinsic_memory_barrier_shared",
compute_shader),
NULL);
+ add_function("ballotARB", _ballot(), NULL);
+
+ add_function("readInvocationARB",
+ _read_invocation(glsl_type::float_type),
+ _read_invocation(glsl_type::vec2_type),
+ _read_invocation(glsl_type::vec3_type),
+ _read_invocation(glsl_type::vec4_type),
+
+ _read_invocation(glsl_type::int_type),
+ _read_invocation(glsl_type::ivec2_type),
+ _read_invocation(glsl_type::ivec3_type),
+ _read_invocation(glsl_type::ivec4_type),
+
+ _read_invocation(glsl_type::uint_type),
+ _read_invocation(glsl_type::uvec2_type),
+ _read_invocation(glsl_type::uvec3_type),
+ _read_invocation(glsl_type::uvec4_type),
+ NULL);
+
+ add_function("readFirstInvocationARB",
+ _read_first_invocation(glsl_type::float_type),
+ _read_first_invocation(glsl_type::vec2_type),
+ _read_first_invocation(glsl_type::vec3_type),
+ _read_first_invocation(glsl_type::vec4_type),
+
+ _read_first_invocation(glsl_type::int_type),
+ _read_first_invocation(glsl_type::ivec2_type),
+ _read_first_invocation(glsl_type::ivec3_type),
+ _read_first_invocation(glsl_type::ivec4_type),
+
+ _read_first_invocation(glsl_type::uint_type),
+ _read_first_invocation(glsl_type::uvec2_type),
+ _read_first_invocation(glsl_type::uvec3_type),
+ _read_first_invocation(glsl_type::uvec4_type),
+ NULL);
+
add_function("clock2x32ARB",
_shader_clock(shader_clock,
glsl_type::uvec2_type),
glsl_type::uint64_t_type),
NULL);
- add_function("anyInvocationARB", _vote(ir_unop_vote_any), NULL);
- add_function("allInvocationsARB", _vote(ir_unop_vote_all), NULL);
- add_function("allInvocationsEqualARB", _vote(ir_unop_vote_eq), NULL);
+ add_function("anyInvocationARB",
+ _vote("__intrinsic_vote_any", vote),
+ NULL);
+
+ add_function("allInvocationsARB",
+ _vote("__intrinsic_vote_all", vote),
+ NULL);
+
+ add_function("allInvocationsEqualARB",
+ _vote("__intrinsic_vote_eq", vote),
+ NULL);
+
+ add_function("anyInvocation",
+ _vote("__intrinsic_vote_any", v460_desktop),
+ NULL);
+
+ add_function("allInvocations",
+ _vote("__intrinsic_vote_all", v460_desktop),
+ NULL);
+
+ add_function("allInvocationsEqual",
+ _vote("__intrinsic_vote_eq", v460_desktop),
+ NULL);
+
+ add_function("__builtin_idiv64",
+ generate_ir::idiv64(mem_ctx, integer_functions_supported),
+ NULL);
+
+ add_function("__builtin_imod64",
+ generate_ir::imod64(mem_ctx, integer_functions_supported),
+ NULL);
+
+ add_function("__builtin_sign64",
+ generate_ir::sign64(mem_ctx, integer_functions_supported),
+ NULL);
+
+ add_function("__builtin_udiv64",
+ generate_ir::udiv64(mem_ctx, integer_functions_supported),
+ NULL);
+
+ add_function("__builtin_umod64",
+ generate_ir::umod64(mem_ctx, integer_functions_supported),
+ NULL);
add_function("__builtin_umul64",
generate_ir::umul64(mem_ctx, integer_functions_supported),
return new(mem_ctx) ir_constant(type, &data);
}
-#define IMM_FP(type, val) (type->base_type == GLSL_TYPE_DOUBLE) ? imm(val) : imm((float)val)
+#define IMM_FP(type, val) (type->is_double()) ? imm(val) : imm((float)val)
ir_dereference_variable *
builtin_builder::var_ref(ir_variable *var)
ir_expression_operation opcode,
const glsl_type *return_type,
const glsl_type *param0_type,
- const glsl_type *param1_type)
+ const glsl_type *param1_type,
+ bool swap_operands)
{
ir_variable *x = in_var(param0_type, "x");
ir_variable *y = in_var(param1_type, "y");
MAKE_SIG(return_type, avail, 2, x, y);
- body.emit(ret(expr(opcode, x, y)));
+
+ if (swap_operands)
+ body.emit(ret(expr(opcode, y, x)));
+ else
+ body.emit(ret(expr(opcode, x, y)));
+
return sig;
}
ir_function_signature *
builtin_builder::_atan2(const glsl_type *type)
{
- ir_variable *vec_y = in_var(type, "vec_y");
- ir_variable *vec_x = in_var(type, "vec_x");
- MAKE_SIG(type, always_available, 2, vec_y, vec_x);
-
- ir_variable *vec_result = body.make_temp(type, "vec_result");
- ir_variable *r = body.make_temp(glsl_type::float_type, "r");
- for (int i = 0; i < type->vector_elements; i++) {
- ir_variable *y = body.make_temp(glsl_type::float_type, "y");
- ir_variable *x = body.make_temp(glsl_type::float_type, "x");
- body.emit(assign(y, swizzle(vec_y, i, 1)));
- body.emit(assign(x, swizzle(vec_x, i, 1)));
-
- /* If |x| >= 1.0e-8 * |y|: */
- ir_if *outer_if =
- new(mem_ctx) ir_if(greater(abs(x), mul(imm(1.0e-8f), abs(y))));
-
- ir_factory outer_then(&outer_if->then_instructions, mem_ctx);
-
- /* Then...call atan(y/x) */
- do_atan(outer_then, 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(M_PIf))),
- assign(r, sub(r, imm(M_PIf)))));
- outer_then.emit(inner_if);
+ const unsigned n = type->vector_elements;
+ ir_variable *y = in_var(type, "y");
+ ir_variable *x = in_var(type, "x");
+ MAKE_SIG(type, always_available, 2, y, x);
- /* Else... */
- outer_if->else_instructions.push_tail(
- assign(r, mul(sign(y), imm(M_PI_2f))));
+ /* If we're on the left half-plane rotate the coordinates π/2 clock-wise
+ * for the y=0 discontinuity to end up aligned with the vertical
+ * discontinuity of atan(s/t) along t=0. This also makes sure that we
+ * don't attempt to divide by zero along the vertical line, which may give
+ * unspecified results on non-GLSL 4.1-capable hardware.
+ */
+ ir_variable *flip = body.make_temp(glsl_type::bvec(n), "flip");
+ body.emit(assign(flip, gequal(imm(0.0f, n), x)));
+ ir_variable *s = body.make_temp(type, "s");
+ body.emit(assign(s, csel(flip, abs(x), y)));
+ ir_variable *t = body.make_temp(type, "t");
+ body.emit(assign(t, csel(flip, y, abs(x))));
- body.emit(outer_if);
+ /* If the magnitude of the denominator exceeds some huge value, scale down
+ * the arguments in order to prevent the reciprocal operation from flushing
+ * its result to zero, which would cause precision problems, and for s
+ * infinite would cause us to return a NaN instead of the correct finite
+ * value.
+ *
+ * If fmin and fmax are respectively the smallest and largest positive
+ * normalized floating point values representable by the implementation,
+ * the constants below should be in agreement with:
+ *
+ * huge <= 1 / fmin
+ * scale <= 1 / fmin / fmax (for |t| >= huge)
+ *
+ * In addition scale should be a negative power of two in order to avoid
+ * loss of precision. The values chosen below should work for most usual
+ * floating point representations with at least the dynamic range of ATI's
+ * 24-bit representation.
+ */
+ ir_constant *huge = imm(1e18f, n);
+ ir_variable *scale = body.make_temp(type, "scale");
+ body.emit(assign(scale, csel(gequal(abs(t), huge),
+ imm(0.25f, n), imm(1.0f, n))));
+ ir_variable *rcp_scaled_t = body.make_temp(type, "rcp_scaled_t");
+ body.emit(assign(rcp_scaled_t, rcp(mul(t, scale))));
+ ir_expression *s_over_t = mul(mul(s, scale), rcp_scaled_t);
+
+ /* For |x| = |y| assume tan = 1 even if infinite (i.e. pretend momentarily
+ * that ∞/∞ = 1) in order to comply with the rather artificial rules
+ * inherited from IEEE 754-2008, namely:
+ *
+ * "atan2(±∞, −∞) is ±3π/4
+ * atan2(±∞, +∞) is ±π/4"
+ *
+ * Note that this is inconsistent with the rules for the neighborhood of
+ * zero that are based on iterated limits:
+ *
+ * "atan2(±0, −0) is ±π
+ * atan2(±0, +0) is ±0"
+ *
+ * but GLSL specifically allows implementations to deviate from IEEE rules
+ * at (0,0), so we take that license (i.e. pretend that 0/0 = 1 here as
+ * well).
+ */
+ ir_expression *tan = csel(equal(abs(x), abs(y)),
+ imm(1.0f, n), abs(s_over_t));
- body.emit(assign(vec_result, r, 1 << i));
- }
- body.emit(ret(vec_result));
+ /* Calculate the arctangent and fix up the result if we had flipped the
+ * coordinate system.
+ */
+ ir_variable *arc = body.make_temp(type, "arc");
+ do_atan(body, type, arc, tan);
+ body.emit(assign(arc, add(arc, mul(b2f(flip), imm(M_PI_2f)))));
+
+ /* Rather convoluted calculation of the sign of the result. When x < 0 we
+ * cannot use fsign because we need to be able to distinguish between
+ * negative and positive zero. Unfortunately we cannot use bitwise
+ * arithmetic tricks either because of back-ends without integer support.
+ * When x >= 0 rcp_scaled_t will always be non-negative so this won't be
+ * able to distinguish between negative and positive zero, but we don't
+ * care because atan2 is continuous along the whole positive y = 0
+ * half-line, so it won't affect the result significantly.
+ */
+ body.emit(ret(csel(less(min2(y, rcp_scaled_t), imm(0.0f, n)),
+ neg(arc), arc)));
return sig;
}
ir_variable *t = body.make_temp(x_type, "t");
if (x_type->vector_elements == 1) {
/* Both are floats */
- if (edge_type->base_type == GLSL_TYPE_DOUBLE)
+ if (edge_type->is_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++) {
- if (edge_type->base_type == GLSL_TYPE_DOUBLE)
+ if (edge_type->is_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++) {
- if (edge_type->base_type == GLSL_TYPE_DOUBLE)
+ if (edge_type->is_double())
body.emit(assign(t, f2d(b2f(gequal(swizzle(x, i, 1), swizzle(edge, i, 1)))),
1 << i));
else
ir_variable *c;
ir_variable *r;
- if (type->base_type == GLSL_TYPE_DOUBLE) {
+ if (type->is_double()) {
r = in_var(glsl_type::dvec(type->matrix_columns), "r");
c = in_var(glsl_type::dvec(type->vector_elements), "c");
} else {
builtin_builder::_lessThanEqual(builtin_available_predicate avail,
const glsl_type *type)
{
- return binop(avail, ir_binop_lequal,
- glsl_type::bvec(type->vector_elements), type, type);
+ return binop(avail, ir_binop_gequal,
+ glsl_type::bvec(type->vector_elements), type, type,
+ true);
}
ir_function_signature *
builtin_builder::_greaterThan(builtin_available_predicate avail,
const glsl_type *type)
{
- return binop(avail, ir_binop_greater,
- glsl_type::bvec(type->vector_elements), type, type);
+ return binop(avail, ir_binop_less,
+ glsl_type::bvec(type->vector_elements), type, type,
+ true);
}
ir_function_signature *
}
ir_function_signature *
-builtin_builder::_textureSamples(const glsl_type *sampler_type)
+builtin_builder::_textureSamples(builtin_available_predicate avail,
+ const glsl_type *sampler_type)
{
ir_variable *s = in_var(sampler_type, "sampler");
- MAKE_SIG(glsl_type::int_type, shader_samples, 1, s);
+ MAKE_SIG(glsl_type::int_type, avail, 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);
}
ir_function_signature *
-builtin_builder::_textureCubeArrayShadow()
+builtin_builder::_textureCubeArrayShadow(builtin_available_predicate avail,
+ const glsl_type *sampler_type)
{
- ir_variable *s = in_var(glsl_type::samplerCubeArrayShadow_type, "sampler");
+ ir_variable *s = in_var(sampler_type, "sampler");
ir_variable *P = in_var(glsl_type::vec4_type, "P");
ir_variable *compare = in_var(glsl_type::float_type, "compare");
- MAKE_SIG(glsl_type::float_type, texture_cube_map_array, 3, s, P, compare);
+ MAKE_SIG(glsl_type::float_type, avail, 3, s, P, compare);
ir_texture *tex = new(mem_ctx) ir_texture(ir_tex);
tex->set_sampler(var_ref(s), glsl_type::float_type);
}
ir_function_signature *
-builtin_builder::_textureQueryLevels(const glsl_type *sampler_type)
+builtin_builder::_textureQueryLevels(builtin_available_predicate avail,
+ 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);
+ MAKE_SIG(return_type, avail, 1, s);
ir_texture *tex = new(mem_ctx) ir_texture(ir_query_levels);
tex->set_sampler(var_ref(s), return_type);
ir_function_signature *
builtin_builder::_ldexp(const glsl_type *x_type, const glsl_type *exp_type)
{
- return binop(x_type->base_type == GLSL_TYPE_DOUBLE ? fp64 : gpu_shader5_or_es31_or_integer_functions,
+ return binop(x_type->is_double() ? fp64 : gpu_shader5_or_es31_or_integer_functions,
ir_binop_ldexp, x_type, x_type, exp_type);
}
* 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;
+ image->data.memory_read_only = (flags & IMAGE_FUNCTION_READ_ONLY) != 0;
+ image->data.memory_write_only = (flags & IMAGE_FUNCTION_WRITE_ONLY) != 0;
+ image->data.memory_coherent = true;
+ image->data.memory_volatile = true;
+ image->data.memory_restrict = true;
return sig;
}
* 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;
+ image->data.memory_read_only = true;
+ image->data.memory_write_only = true;
+ image->data.memory_coherent = true;
+ image->data.memory_volatile = true;
+ image->data.memory_restrict = true;
return sig;
}
* 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;
+ image->data.memory_read_only = true;
+ image->data.memory_write_only = true;
+ image->data.memory_coherent = true;
+ image->data.memory_volatile = true;
+ image->data.memory_restrict = true;
return sig;
}
return sig;
}
+ir_function_signature *
+builtin_builder::_ballot_intrinsic()
+{
+ ir_variable *value = in_var(glsl_type::bool_type, "value");
+ MAKE_INTRINSIC(glsl_type::uint64_t_type, ir_intrinsic_ballot, shader_ballot,
+ 1, value);
+ return sig;
+}
+
+ir_function_signature *
+builtin_builder::_ballot()
+{
+ ir_variable *value = in_var(glsl_type::bool_type, "value");
+
+ MAKE_SIG(glsl_type::uint64_t_type, shader_ballot, 1, value);
+ ir_variable *retval = body.make_temp(glsl_type::uint64_t_type, "retval");
+
+ body.emit(call(shader->symbols->get_function("__intrinsic_ballot"),
+ retval, sig->parameters));
+ body.emit(ret(retval));
+ return sig;
+}
+
+ir_function_signature *
+builtin_builder::_read_first_invocation_intrinsic(const glsl_type *type)
+{
+ ir_variable *value = in_var(type, "value");
+ MAKE_INTRINSIC(type, ir_intrinsic_read_first_invocation, shader_ballot,
+ 1, value);
+ return sig;
+}
+
+ir_function_signature *
+builtin_builder::_read_first_invocation(const glsl_type *type)
+{
+ ir_variable *value = in_var(type, "value");
+
+ MAKE_SIG(type, shader_ballot, 1, value);
+ ir_variable *retval = body.make_temp(type, "retval");
+
+ body.emit(call(shader->symbols->get_function("__intrinsic_read_first_invocation"),
+ retval, sig->parameters));
+ body.emit(ret(retval));
+ return sig;
+}
+
+ir_function_signature *
+builtin_builder::_read_invocation_intrinsic(const glsl_type *type)
+{
+ ir_variable *value = in_var(type, "value");
+ ir_variable *invocation = in_var(glsl_type::uint_type, "invocation");
+ MAKE_INTRINSIC(type, ir_intrinsic_read_invocation, shader_ballot,
+ 2, value, invocation);
+ return sig;
+}
+
+ir_function_signature *
+builtin_builder::_read_invocation(const glsl_type *type)
+{
+ ir_variable *value = in_var(type, "value");
+ ir_variable *invocation = in_var(glsl_type::uint_type, "invocation");
+
+ MAKE_SIG(type, shader_ballot, 2, value, invocation);
+ ir_variable *retval = body.make_temp(type, "retval");
+
+ body.emit(call(shader->symbols->get_function("__intrinsic_read_invocation"),
+ retval, sig->parameters));
+ body.emit(ret(retval));
+ return sig;
+}
+
ir_function_signature *
builtin_builder::_shader_clock_intrinsic(builtin_available_predicate avail,
const glsl_type *type)
{
MAKE_SIG(type, avail, 0);
- ir_variable *retval = body.make_temp(type, "clock_retval");
+ ir_variable *retval = body.make_temp(glsl_type::uvec2_type, "clock_retval");
body.emit(call(shader->symbols->get_function("__intrinsic_shader_clock"),
retval, sig->parameters));
}
ir_function_signature *
-builtin_builder::_vote(enum ir_expression_operation opcode)
+builtin_builder::_vote_intrinsic(builtin_available_predicate avail,
+ enum ir_intrinsic_id id)
+{
+ ir_variable *value = in_var(glsl_type::bool_type, "value");
+ MAKE_INTRINSIC(glsl_type::bool_type, id, avail, 1, value);
+ return sig;
+}
+
+ir_function_signature *
+builtin_builder::_vote(const char *intrinsic_name,
+ builtin_available_predicate avail)
{
ir_variable *value = in_var(glsl_type::bool_type, "value");
- MAKE_SIG(glsl_type::bool_type, vote, 1, value);
- body.emit(ret(expr(opcode, value)));
+ MAKE_SIG(glsl_type::bool_type, avail, 1, value);
+
+ ir_variable *retval = body.make_temp(glsl_type::bool_type, "retval");
+
+ body.emit(call(shader->symbols->get_function(intrinsic_name),
+ retval, sig->parameters));
+ body.emit(ret(retval));
return sig;
}
_mesa_glsl_find_builtin_function(_mesa_glsl_parse_state *state,
const char *name, exec_list *actual_parameters)
{
- ir_function_signature * s;
+ 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)
+bool
+_mesa_glsl_has_builtin_function(_mesa_glsl_parse_state *state, const char *name)
{
ir_function *f;
+ bool ret = false;
mtx_lock(&builtins_lock);
f = builtins.shader->symbols->get_function(name);
+ if (f != NULL) {
+ foreach_in_list(ir_function_signature, sig, &f->signatures) {
+ if (sig->is_builtin_available(state)) {
+ ret = true;
+ break;
+ }
+ }
+ }
mtx_unlock(&builtins_lock);
- return f;
+
+ return ret;
}
gl_shader *