summaryrefslogtreecommitdiffstats
path: root/src/glsl
diff options
context:
space:
mode:
Diffstat (limited to 'src/glsl')
-rw-r--r--src/glsl/SConscript4
-rw-r--r--src/glsl/ast_function.cpp35
-rw-r--r--src/glsl/ast_to_hir.cpp16
-rw-r--r--src/glsl/builtins/ir/abs20
-rwxr-xr-xsrc/glsl/builtins/tools/generate_builtins.py4
-rw-r--r--src/glsl/glsl_parser_extras.cpp327
-rw-r--r--src/glsl/glsl_parser_extras.h44
-rw-r--r--src/glsl/ir.cpp8
-rw-r--r--src/glsl/ir.h4
-rw-r--r--src/glsl/ir_constant_expression.cpp13
-rw-r--r--src/glsl/ir_function.cpp9
-rw-r--r--src/glsl/ir_validate.cpp14
-rw-r--r--src/glsl/linker.cpp41
-rw-r--r--src/glsl/lower_if_to_cond_assign.cpp4
-rw-r--r--src/glsl/lower_instructions.cpp11
-rw-r--r--src/glsl/lower_mat_op_to_vec.cpp298
-rw-r--r--src/glsl/main.cpp1
-rw-r--r--src/glsl/opt_discard_simplification.cpp20
18 files changed, 523 insertions, 350 deletions
diff --git a/src/glsl/SConscript b/src/glsl/SConscript
index c3255835fb4..1441cc74bd8 100644
--- a/src/glsl/SConscript
+++ b/src/glsl/SConscript
@@ -102,7 +102,7 @@ if env['msvc']:
env.Prepend(CPPPATH = ['#/src/getopt'])
env.PrependUnique(LIBS = [getopt])
-if env['crosscompile'] and env['platform'] != 'embedded':
+if env['crosscompile'] and not env['embedded']:
Import('builtin_glsl_function')
else:
# Copy these files to avoid generation object files into src/mesa/program
@@ -156,7 +156,7 @@ Export('glsl')
# Skip building these programs as they will cause SCons error "Two environments
# with different actions were specified for the same target"
-if env['crosscompile'] or env['platform'] == 'embedded':
+if env['crosscompile'] or env['embedded']:
Return()
env = env.Clone()
diff --git a/src/glsl/ast_function.cpp b/src/glsl/ast_function.cpp
index 3ba699ad971..60a2c617f70 100644
--- a/src/glsl/ast_function.cpp
+++ b/src/glsl/ast_function.cpp
@@ -271,17 +271,36 @@ convert_component(ir_rvalue *src, const glsl_type *desired_type)
assert(a <= GLSL_TYPE_BOOL);
assert(b <= GLSL_TYPE_BOOL);
- if ((a == b) || (src->type->is_integer() && desired_type->is_integer()))
+ if (a == b)
return src;
switch (a) {
case GLSL_TYPE_UINT:
+ switch (b) {
+ case GLSL_TYPE_INT:
+ result = new(ctx) ir_expression(ir_unop_i2u, src);
+ break;
+ case GLSL_TYPE_FLOAT:
+ result = new(ctx) ir_expression(ir_unop_i2u,
+ new(ctx) ir_expression(ir_unop_f2i, src));
+ break;
+ case GLSL_TYPE_BOOL:
+ result = new(ctx) ir_expression(ir_unop_i2u,
+ new(ctx) ir_expression(ir_unop_b2i, src));
+ break;
+ }
+ break;
case GLSL_TYPE_INT:
- if (b == GLSL_TYPE_FLOAT)
- result = new(ctx) ir_expression(ir_unop_f2i, desired_type, src, NULL);
- else {
- assert(b == GLSL_TYPE_BOOL);
- result = new(ctx) ir_expression(ir_unop_b2i, desired_type, src, NULL);
+ switch (b) {
+ case GLSL_TYPE_UINT:
+ result = new(ctx) ir_expression(ir_unop_u2i, src);
+ break;
+ case GLSL_TYPE_FLOAT:
+ result = new(ctx) ir_expression(ir_unop_f2i, src);
+ break;
+ case GLSL_TYPE_BOOL:
+ result = new(ctx) ir_expression(ir_unop_b2i, src);
+ break;
}
break;
case GLSL_TYPE_FLOAT:
@@ -300,6 +319,9 @@ convert_component(ir_rvalue *src, const glsl_type *desired_type)
case GLSL_TYPE_BOOL:
switch (b) {
case GLSL_TYPE_UINT:
+ result = new(ctx) ir_expression(ir_unop_i2b,
+ new(ctx) ir_expression(ir_unop_u2i, src));
+ break;
case GLSL_TYPE_INT:
result = new(ctx) ir_expression(ir_unop_i2b, desired_type, src, NULL);
break;
@@ -311,6 +333,7 @@ convert_component(ir_rvalue *src, const glsl_type *desired_type)
}
assert(result != NULL);
+ assert(result->type == desired_type);
/* Try constant folding; it may fold in the conversion we just added. */
ir_constant *const constant = result->constant_expression_value();
diff --git a/src/glsl/ast_to_hir.cpp b/src/glsl/ast_to_hir.cpp
index 3b87f0d56de..2e54e8c22d8 100644
--- a/src/glsl/ast_to_hir.cpp
+++ b/src/glsl/ast_to_hir.cpp
@@ -447,9 +447,17 @@ modulus_result_type(const struct glsl_type *type_a,
* integer vectors. The operand types must both be signed or both be
* unsigned."
*/
- if (!type_a->is_integer() || !type_b->is_integer()
- || (type_a->base_type != type_b->base_type)) {
- _mesa_glsl_error(loc, state, "type mismatch");
+ if (!type_a->is_integer()) {
+ _mesa_glsl_error(loc, state, "LHS of operator %% must be an integer.");
+ return glsl_type::error_type;
+ }
+ if (!type_b->is_integer()) {
+ _mesa_glsl_error(loc, state, "RHS of operator %% must be an integer.");
+ return glsl_type::error_type;
+ }
+ if (type_a->base_type != type_b->base_type) {
+ _mesa_glsl_error(loc, state,
+ "operands of %% must have the same base type");
return glsl_type::error_type;
}
@@ -1940,7 +1948,7 @@ apply_type_qualifier_to_variable(const struct ast_type_qualifier *qual,
break;
case fragment_shader:
- if (!global_scope || (var->mode != ir_var_in)) {
+ if (!global_scope || (var->mode != ir_var_out)) {
fail = true;
string = "output";
}
diff --git a/src/glsl/builtins/ir/abs b/src/glsl/builtins/ir/abs
index 904845307c4..d07d1d99e9f 100644
--- a/src/glsl/builtins/ir/abs
+++ b/src/glsl/builtins/ir/abs
@@ -18,4 +18,24 @@
(parameters
(declare (in) vec4 arg0))
((return (expression vec4 abs (var_ref arg0)))))
+
+ (signature int
+ (parameters
+ (declare (in) int arg0))
+ ((return (expression int abs (var_ref arg0)))))
+
+ (signature ivec2
+ (parameters
+ (declare (in) ivec2 arg0))
+ ((return (expression ivec2 abs (var_ref arg0)))))
+
+ (signature ivec3
+ (parameters
+ (declare (in) ivec3 arg0))
+ ((return (expression ivec3 abs (var_ref arg0)))))
+
+ (signature ivec4
+ (parameters
+ (declare (in) ivec4 arg0))
+ ((return (expression ivec4 abs (var_ref arg0)))))
))
diff --git a/src/glsl/builtins/tools/generate_builtins.py b/src/glsl/builtins/tools/generate_builtins.py
index edd3c70e00b..17d528c2180 100755
--- a/src/glsl/builtins/tools/generate_builtins.py
+++ b/src/glsl/builtins/tools/generate_builtins.py
@@ -82,10 +82,6 @@ def write_profile(filename, profile):
kill_globals = re.compile(r'^\(declare.*\n', re.MULTILINE)
proto_ir = kill_globals.sub('', proto_ir)
- # Kill pointer addresses. They're not necessary in prototypes and just
- # clutter the diff output.
- proto_ir = re.sub(r'@0x[0-9a-f]+', '', proto_ir)
-
print 'static const char prototypes_for_' + profile + '[] ='
print stringify(proto_ir), ';'
diff --git a/src/glsl/glsl_parser_extras.cpp b/src/glsl/glsl_parser_extras.cpp
index d9aa300bbe4..cc781378d76 100644
--- a/src/glsl/glsl_parser_extras.cpp
+++ b/src/glsl/glsl_parser_extras.cpp
@@ -165,133 +165,242 @@ _mesa_glsl_warning(const YYLTYPE *locp, _mesa_glsl_parse_state *state,
}
+/**
+ * Enum representing the possible behaviors that can be specified in
+ * an #extension directive.
+ */
+enum ext_behavior {
+ extension_disable,
+ extension_enable,
+ extension_require,
+ extension_warn
+};
+
+/**
+ * Element type for _mesa_glsl_supported_extensions
+ */
+struct _mesa_glsl_extension {
+ /**
+ * Name of the extension when referred to in a GLSL extension
+ * statement
+ */
+ const char *name;
+
+ /** True if this extension is available to vertex shaders */
+ bool avail_in_VS;
+
+ /** True if this extension is available to geometry shaders */
+ bool avail_in_GS;
+
+ /** True if this extension is available to fragment shaders */
+ bool avail_in_FS;
+
+ /** True if this extension is available to desktop GL shaders */
+ bool avail_in_GL;
+
+ /** True if this extension is available to GLES shaders */
+ bool avail_in_ES;
+
+ /**
+ * Flag in the gl_extensions struct indicating whether this
+ * extension is supported by the driver, or
+ * &gl_extensions::dummy_true if supported by all drivers.
+ *
+ * Note: the type (GLboolean gl_extensions::*) is a "pointer to
+ * member" type, the type-safe alternative to the "offsetof" macro.
+ * In a nutshell:
+ *
+ * - foo bar::* p declares p to be an "offset" to a field of type
+ * foo that exists within struct bar
+ * - &bar::baz computes the "offset" of field baz within struct bar
+ * - x.*p accesses the field of x that exists at "offset" p
+ * - x->*p is equivalent to (*x).*p
+ */
+ const GLboolean gl_extensions::* supported_flag;
+
+ /**
+ * Flag in the _mesa_glsl_parse_state struct that should be set
+ * when this extension is enabled.
+ *
+ * See note in _mesa_glsl_extension::supported_flag about "pointer
+ * to member" types.
+ */
+ bool _mesa_glsl_parse_state::* enable_flag;
+
+ /**
+ * Flag in the _mesa_glsl_parse_state struct that should be set
+ * when the shader requests "warn" behavior for this extension.
+ *
+ * See note in _mesa_glsl_extension::supported_flag about "pointer
+ * to member" types.
+ */
+ bool _mesa_glsl_parse_state::* warn_flag;
+
+
+ bool compatible_with_state(const _mesa_glsl_parse_state *state) const;
+ void set_flags(_mesa_glsl_parse_state *state, ext_behavior behavior) const;
+};
+
+#define EXT(NAME, VS, GS, FS, GL, ES, SUPPORTED_FLAG) \
+ { "GL_" #NAME, VS, GS, FS, GL, ES, &gl_extensions::SUPPORTED_FLAG, \
+ &_mesa_glsl_parse_state::NAME##_enable, \
+ &_mesa_glsl_parse_state::NAME##_warn }
+
+/**
+ * Table of extensions that can be enabled/disabled within a shader,
+ * and the conditions under which they are supported.
+ */
+static const _mesa_glsl_extension _mesa_glsl_supported_extensions[] = {
+ /* target availability API availability */
+ /* name VS GS FS GL ES supported flag */
+ EXT(ARB_draw_buffers, false, false, true, true, false, dummy_true),
+ EXT(ARB_draw_instanced, true, false, false, true, false, ARB_draw_instanced),
+ EXT(ARB_explicit_attrib_location, true, false, true, true, false, ARB_explicit_attrib_location),
+ EXT(ARB_fragment_coord_conventions, true, false, true, true, false, ARB_fragment_coord_conventions),
+ EXT(ARB_texture_rectangle, true, false, true, true, false, dummy_true),
+ EXT(EXT_texture_array, true, false, true, true, false, EXT_texture_array),
+ EXT(ARB_shader_texture_lod, true, false, true, true, false, ARB_shader_texture_lod),
+ EXT(ARB_shader_stencil_export, false, false, true, true, false, ARB_shader_stencil_export),
+ EXT(AMD_conservative_depth, true, false, true, true, false, AMD_conservative_depth),
+ EXT(AMD_shader_stencil_export, false, false, true, true, false, ARB_shader_stencil_export),
+ EXT(OES_texture_3D, true, false, true, false, true, EXT_texture3D),
+};
+
+#undef EXT
+
+
+/**
+ * Determine whether a given extension is compatible with the target,
+ * API, and extension information in the current parser state.
+ */
+bool _mesa_glsl_extension::compatible_with_state(const _mesa_glsl_parse_state *
+ state) const
+{
+ /* Check that this extension matches the type of shader we are
+ * compiling to.
+ */
+ switch (state->target) {
+ case vertex_shader:
+ if (!this->avail_in_VS) {
+ return false;
+ }
+ break;
+ case geometry_shader:
+ if (!this->avail_in_GS) {
+ return false;
+ }
+ break;
+ case fragment_shader:
+ if (!this->avail_in_FS) {
+ return false;
+ }
+ break;
+ default:
+ assert (!"Unrecognized shader target");
+ return false;
+ }
+
+ /* Check that this extension matches whether we are compiling
+ * for desktop GL or GLES.
+ */
+ if (state->es_shader) {
+ if (!this->avail_in_ES) return false;
+ } else {
+ if (!this->avail_in_GL) return false;
+ }
+
+ /* Check that this extension is supported by the OpenGL
+ * implementation.
+ *
+ * Note: the ->* operator indexes into state->extensions by the
+ * offset this->supported_flag. See
+ * _mesa_glsl_extension::supported_flag for more info.
+ */
+ return state->extensions->*(this->supported_flag);
+}
+
+/**
+ * Set the appropriate flags in the parser state to establish the
+ * given behavior for this extension.
+ */
+void _mesa_glsl_extension::set_flags(_mesa_glsl_parse_state *state,
+ ext_behavior behavior) const
+{
+ /* Note: the ->* operator indexes into state by the
+ * offsets this->enable_flag and this->warn_flag. See
+ * _mesa_glsl_extension::supported_flag for more info.
+ */
+ state->*(this->enable_flag) = (behavior != extension_disable);
+ state->*(this->warn_flag) = (behavior == extension_warn);
+}
+
+/**
+ * Find an extension by name in _mesa_glsl_supported_extensions. If
+ * the name is not found, return NULL.
+ */
+static const _mesa_glsl_extension *find_extension(const char *name)
+{
+ for (unsigned i = 0; i < Elements(_mesa_glsl_supported_extensions); ++i) {
+ if (strcmp(name, _mesa_glsl_supported_extensions[i].name) == 0) {
+ return &_mesa_glsl_supported_extensions[i];
+ }
+ }
+ return NULL;
+}
+
+
bool
_mesa_glsl_process_extension(const char *name, YYLTYPE *name_locp,
- const char *behavior, YYLTYPE *behavior_locp,
+ const char *behavior_string, YYLTYPE *behavior_locp,
_mesa_glsl_parse_state *state)
{
- enum {
- extension_disable,
- extension_enable,
- extension_require,
- extension_warn
- } ext_mode;
-
- if (strcmp(behavior, "warn") == 0) {
- ext_mode = extension_warn;
- } else if (strcmp(behavior, "require") == 0) {
- ext_mode = extension_require;
- } else if (strcmp(behavior, "enable") == 0) {
- ext_mode = extension_enable;
- } else if (strcmp(behavior, "disable") == 0) {
- ext_mode = extension_disable;
+ ext_behavior behavior;
+ if (strcmp(behavior_string, "warn") == 0) {
+ behavior = extension_warn;
+ } else if (strcmp(behavior_string, "require") == 0) {
+ behavior = extension_require;
+ } else if (strcmp(behavior_string, "enable") == 0) {
+ behavior = extension_enable;
+ } else if (strcmp(behavior_string, "disable") == 0) {
+ behavior = extension_disable;
} else {
_mesa_glsl_error(behavior_locp, state,
"Unknown extension behavior `%s'",
- behavior);
+ behavior_string);
return false;
}
- bool unsupported = false;
-
if (strcmp(name, "all") == 0) {
- if ((ext_mode == extension_enable) || (ext_mode == extension_require)) {
+ if ((behavior == extension_enable) || (behavior == extension_require)) {
_mesa_glsl_error(name_locp, state, "Cannot %s all extensions",
- (ext_mode == extension_enable)
+ (behavior == extension_enable)
? "enable" : "require");
return false;
- }
- } else if (strcmp(name, "GL_ARB_draw_buffers") == 0) {
- /* This extension is only supported in fragment shaders.
- */
- if (state->target != fragment_shader) {
- unsupported = true;
} else {
- state->ARB_draw_buffers_enable = (ext_mode != extension_disable);
- state->ARB_draw_buffers_warn = (ext_mode == extension_warn);
+ for (unsigned i = 0;
+ i < Elements(_mesa_glsl_supported_extensions); ++i) {
+ const _mesa_glsl_extension *extension
+ = &_mesa_glsl_supported_extensions[i];
+ if (extension->compatible_with_state(state)) {
+ _mesa_glsl_supported_extensions[i].set_flags(state, behavior);
+ }
+ }
}
- } else if (strcmp(name, "GL_ARB_draw_instanced") == 0) {
- state->ARB_draw_instanced_enable = (ext_mode != extension_disable);
- state->ARB_draw_instanced_warn = (ext_mode == extension_warn);
-
- /* This extension is only supported in vertex shaders.
- */
- unsupported = (state->target != vertex_shader)
- || !state->extensions->ARB_draw_instanced;
- } else if (strcmp(name, "GL_ARB_explicit_attrib_location") == 0) {
- state->ARB_explicit_attrib_location_enable =
- (ext_mode != extension_disable);
- state->ARB_explicit_attrib_location_warn =
- (ext_mode == extension_warn);
-
- unsupported = !state->extensions->ARB_explicit_attrib_location;
- } else if (strcmp(name, "GL_ARB_fragment_coord_conventions") == 0) {
- state->ARB_fragment_coord_conventions_enable =
- (ext_mode != extension_disable);
- state->ARB_fragment_coord_conventions_warn =
- (ext_mode == extension_warn);
-
- unsupported = !state->extensions->ARB_fragment_coord_conventions;
- } else if (strcmp(name, "GL_ARB_texture_rectangle") == 0) {
- state->ARB_texture_rectangle_enable = (ext_mode != extension_disable);
- state->ARB_texture_rectangle_warn = (ext_mode == extension_warn);
- } else if (strcmp(name, "GL_EXT_texture_array") == 0) {
- state->EXT_texture_array_enable = (ext_mode != extension_disable);
- state->EXT_texture_array_warn = (ext_mode == extension_warn);
-
- unsupported = !state->extensions->EXT_texture_array;
- } else if (strcmp(name, "GL_ARB_shader_texture_lod") == 0) {
- /* Force ARB_texture_rectangle to be on so sampler2DRects are defined */
- state->ARB_texture_rectangle_enable = true;
-
- state->ARB_shader_texture_lod_enable = (ext_mode != extension_disable);
- state->ARB_shader_texture_lod_warn = (ext_mode == extension_warn);
-
- unsupported = !state->extensions->ARB_shader_texture_lod;
- } else if (strcmp(name, "GL_ARB_shader_stencil_export") == 0) {
- state->ARB_shader_stencil_export_enable = (ext_mode != extension_disable);
- state->ARB_shader_stencil_export_warn = (ext_mode == extension_warn);
-
- /* This extension is only supported in fragment shaders.
- */
- unsupported = (state->target != fragment_shader)
- || !state->extensions->ARB_shader_stencil_export;
- } else if (strcmp(name, "GL_AMD_conservative_depth") == 0) {
- /* The AMD_conservative spec does not forbid requiring the extension in
- * the vertex shader.
- */
- state->AMD_conservative_depth_enable = (ext_mode != extension_disable);
- state->AMD_conservative_depth_warn = (ext_mode == extension_warn);
- unsupported = !state->extensions->AMD_conservative_depth;
- } else if (strcmp(name, "GL_AMD_shader_stencil_export") == 0) {
- state->AMD_shader_stencil_export_enable = (ext_mode != extension_disable);
- state->AMD_shader_stencil_export_warn = (ext_mode == extension_warn);
-
- /* This extension is only supported in fragment shaders.
- * Both the ARB and AMD variants share the same ARB flag
- * in gl_extensions.
- */
- unsupported = (state->target != fragment_shader)
- || !state->extensions->ARB_shader_stencil_export;
- } else if (strcmp(name, "GL_OES_texture_3D") == 0 && state->es_shader) {
- state->OES_texture_3D_enable = (ext_mode != extension_disable);
- state->OES_texture_3D_warn = (ext_mode == extension_warn);
-
- unsupported = !state->extensions->EXT_texture3D;
} else {
- unsupported = true;
- }
-
- if (unsupported) {
- static const char *const fmt = "extension `%s' unsupported in %s shader";
-
- if (ext_mode == extension_require) {
- _mesa_glsl_error(name_locp, state, fmt,
- name, _mesa_glsl_shader_target_name(state->target));
- return false;
+ const _mesa_glsl_extension *extension = find_extension(name);
+ if (extension && extension->compatible_with_state(state)) {
+ extension->set_flags(state, behavior);
} else {
- _mesa_glsl_warning(name_locp, state, fmt,
- name, _mesa_glsl_shader_target_name(state->target));
+ static const char *const fmt = "extension `%s' unsupported in %s shader";
+
+ if (behavior == extension_require) {
+ _mesa_glsl_error(name_locp, state, fmt,
+ name, _mesa_glsl_shader_target_name(state->target));
+ return false;
+ } else {
+ _mesa_glsl_warning(name_locp, state, fmt,
+ name, _mesa_glsl_shader_target_name(state->target));
+ }
}
}
diff --git a/src/glsl/glsl_parser_extras.h b/src/glsl/glsl_parser_extras.h
index 878d2ae3f4d..2f4d3cba77f 100644
--- a/src/glsl/glsl_parser_extras.h
+++ b/src/glsl/glsl_parser_extras.h
@@ -156,28 +156,28 @@ struct _mesa_glsl_parse_state {
* \name Enable bits for GLSL extensions
*/
/*@{*/
- unsigned ARB_draw_buffers_enable:1;
- unsigned ARB_draw_buffers_warn:1;
- unsigned ARB_draw_instanced_enable:1;
- unsigned ARB_draw_instanced_warn:1;
- unsigned ARB_explicit_attrib_location_enable:1;
- unsigned ARB_explicit_attrib_location_warn:1;
- unsigned ARB_fragment_coord_conventions_enable:1;
- unsigned ARB_fragment_coord_conventions_warn:1;
- unsigned ARB_texture_rectangle_enable:1;
- unsigned ARB_texture_rectangle_warn:1;
- unsigned EXT_texture_array_enable:1;
- unsigned EXT_texture_array_warn:1;
- unsigned ARB_shader_texture_lod_enable:1;
- unsigned ARB_shader_texture_lod_warn:1;
- unsigned ARB_shader_stencil_export_enable:1;
- unsigned ARB_shader_stencil_export_warn:1;
- unsigned AMD_conservative_depth_enable:1;
- unsigned AMD_conservative_depth_warn:1;
- unsigned AMD_shader_stencil_export_enable:1;
- unsigned AMD_shader_stencil_export_warn:1;
- unsigned OES_texture_3D_enable:1;
- unsigned OES_texture_3D_warn:1;
+ bool ARB_draw_buffers_enable;
+ bool ARB_draw_buffers_warn;
+ bool ARB_draw_instanced_enable;
+ bool ARB_draw_instanced_warn;
+ bool ARB_explicit_attrib_location_enable;
+ bool ARB_explicit_attrib_location_warn;
+ bool ARB_fragment_coord_conventions_enable;
+ bool ARB_fragment_coord_conventions_warn;
+ bool ARB_texture_rectangle_enable;
+ bool ARB_texture_rectangle_warn;
+ bool EXT_texture_array_enable;
+ bool EXT_texture_array_warn;
+ bool ARB_shader_texture_lod_enable;
+ bool ARB_shader_texture_lod_warn;
+ bool ARB_shader_stencil_export_enable;
+ bool ARB_shader_stencil_export_warn;
+ bool AMD_conservative_depth_enable;
+ bool AMD_conservative_depth_warn;
+ bool AMD_shader_stencil_export_enable;
+ bool AMD_shader_stencil_export_warn;
+ bool OES_texture_3D_enable;
+ bool OES_texture_3D_warn;
/*@}*/
/** Extensions supported by the OpenGL implementation. */
diff --git a/src/glsl/ir.cpp b/src/glsl/ir.cpp
index a3623b31e5d..95689dc10b0 100644
--- a/src/glsl/ir.cpp
+++ b/src/glsl/ir.cpp
@@ -272,6 +272,7 @@ ir_expression::ir_expression(int op, ir_rvalue *op0)
case ir_unop_f2i:
case ir_unop_b2i:
+ case ir_unop_u2i:
this->type = glsl_type::get_instance(GLSL_TYPE_INT,
op0->type->vector_elements, 1);
break;
@@ -289,6 +290,11 @@ ir_expression::ir_expression(int op, ir_rvalue *op0)
op0->type->vector_elements, 1);
break;
+ case ir_unop_i2u:
+ this->type = glsl_type::get_instance(GLSL_TYPE_UINT,
+ op0->type->vector_elements, 1);
+ break;
+
case ir_unop_noise:
this->type = glsl_type::float_type;
break;
@@ -419,6 +425,8 @@ static const char *const operator_strs[] = {
"i2b",
"b2i",
"u2f",
+ "i2u",
+ "u2i",
"any",
"trunc",
"ceil",
diff --git a/src/glsl/ir.h b/src/glsl/ir.h
index a41984310b3..42a3936976a 100644
--- a/src/glsl/ir.h
+++ b/src/glsl/ir.h
@@ -682,7 +682,7 @@ public:
class ir_assignment : public ir_instruction {
public:
- ir_assignment(ir_rvalue *lhs, ir_rvalue *rhs, ir_rvalue *condition);
+ ir_assignment(ir_rvalue *lhs, ir_rvalue *rhs, ir_rvalue *condition = NULL);
/**
* Construct an assignment with an explicit write mask
@@ -789,6 +789,8 @@ enum ir_expression_operation {
ir_unop_i2b, /**< int-to-boolean conversion */
ir_unop_b2i, /**< Boolean-to-int conversion */
ir_unop_u2f, /**< Unsigned-to-float conversion. */
+ ir_unop_i2u, /**< Integer-to-unsigned conversion. */
+ ir_unop_u2i, /**< Unsigned-to-integer conversion. */
ir_unop_any,
/**
diff --git a/src/glsl/ir_constant_expression.cpp b/src/glsl/ir_constant_expression.cpp
index 2a308489656..f0299a2c4a0 100644
--- a/src/glsl/ir_constant_expression.cpp
+++ b/src/glsl/ir_constant_expression.cpp
@@ -166,7 +166,18 @@ ir_expression::constant_expression_value()
data.b[c] = op[0]->value.u[c] ? true : false;
}
break;
-
+ case ir_unop_u2i:
+ assert(op[0]->type->base_type == GLSL_TYPE_UINT);
+ for (unsigned c = 0; c < op[0]->type->components(); c++) {
+ data.i[c] = op[0]->value.u[c];
+ }
+ break;
+ case ir_unop_i2u:
+ assert(op[0]->type->base_type == GLSL_TYPE_INT);
+ for (unsigned c = 0; c < op[0]->type->components(); c++) {
+ data.u[c] = op[0]->value.i[c];
+ }
+ break;
case ir_unop_any:
assert(op[0]->type->is_boolean());
data.b[0] = false;
diff --git a/src/glsl/ir_function.cpp b/src/glsl/ir_function.cpp
index caee9296af9..ef8d4fcfcd4 100644
--- a/src/glsl/ir_function.cpp
+++ b/src/glsl/ir_function.cpp
@@ -165,6 +165,7 @@ ir_function_signature *
ir_function::matching_signature(const exec_list *actual_parameters)
{
ir_function_signature *match = NULL;
+ int matched_score;
foreach_iter(exec_list_iterator, iter, signatures) {
ir_function_signature *const sig =
@@ -173,14 +174,14 @@ ir_function::matching_signature(const exec_list *actual_parameters)
const int score = parameter_lists_match(& sig->parameters,
actual_parameters);
+ /* If we found an exact match, simply return it */
if (score == 0)
return sig;
- if (score > 0) {
- if (match != NULL)
- return NULL;
-
+ /* If we found a match with fewer conversions, use that instead */
+ if (score > 0 && (match == NULL || score < matched_score)) {
match = sig;
+ matched_score = score;
}
}
diff --git a/src/glsl/ir_validate.cpp b/src/glsl/ir_validate.cpp
index 7b1c19d65aa..f3fceb2a57d 100644
--- a/src/glsl/ir_validate.cpp
+++ b/src/glsl/ir_validate.cpp
@@ -254,7 +254,7 @@ ir_validate::visit_leave(ir_expression *ir)
case ir_unop_f2i:
assert(ir->operands[0]->type->base_type == GLSL_TYPE_FLOAT);
- assert(ir->type->is_integer());
+ assert(ir->type->base_type == GLSL_TYPE_INT);
break;
case ir_unop_i2f:
assert(ir->operands[0]->type->base_type == GLSL_TYPE_INT);
@@ -269,17 +269,25 @@ ir_validate::visit_leave(ir_expression *ir)
assert(ir->type->base_type == GLSL_TYPE_FLOAT);
break;
case ir_unop_i2b:
- assert(ir->operands[0]->type->is_integer());
+ assert(ir->operands[0]->type->base_type == GLSL_TYPE_INT);
assert(ir->type->base_type == GLSL_TYPE_BOOL);
break;
case ir_unop_b2i:
assert(ir->operands[0]->type->base_type == GLSL_TYPE_BOOL);
- assert(ir->type->is_integer());
+ assert(ir->type->base_type == GLSL_TYPE_INT);
break;
case ir_unop_u2f:
assert(ir->operands[0]->type->base_type == GLSL_TYPE_UINT);
assert(ir->type->base_type == GLSL_TYPE_FLOAT);
break;
+ case ir_unop_i2u:
+ assert(ir->operands[0]->type->base_type == GLSL_TYPE_INT);
+ assert(ir->type->base_type == GLSL_TYPE_UINT);
+ break;
+ case ir_unop_u2i:
+ assert(ir->operands[0]->type->base_type == GLSL_TYPE_UINT);
+ assert(ir->type->base_type == GLSL_TYPE_INT);
+ break;
case ir_unop_any:
assert(ir->operands[0]->type->base_type == GLSL_TYPE_BOOL);
diff --git a/src/glsl/linker.cpp b/src/glsl/linker.cpp
index 255edc6a76f..b6479e7a3a4 100644
--- a/src/glsl/linker.cpp
+++ b/src/glsl/linker.cpp
@@ -1405,8 +1405,9 @@ demote_shader_inputs_and_outputs(gl_shader *sh, enum ir_variable_mode mode)
}
-void
-assign_varying_locations(struct gl_shader_program *prog,
+bool
+assign_varying_locations(struct gl_context *ctx,
+ struct gl_shader_program *prog,
gl_shader *producer, gl_shader *consumer)
{
/* FINISHME: Set dynamically when geometry shader support is added. */
@@ -1462,6 +1463,8 @@ assign_varying_locations(struct gl_shader_program *prog,
}
}
+ unsigned varying_vectors = 0;
+
foreach_list(node, consumer->ir) {
ir_variable *const var = ((ir_instruction *) node)->as_variable();
@@ -1492,8 +1495,32 @@ assign_varying_locations(struct gl_shader_program *prog,
* value is written by the previous stage.
*/
var->mode = ir_var_auto;
+ } else {
+ /* The packing rules are used for vertex shader inputs are also used
+ * for fragment shader inputs.
+ */
+ varying_vectors += count_attribute_slots(var->type);
}
}
+
+ if (ctx->API == API_OPENGLES2 || prog->Version == 100) {
+ if (varying_vectors > ctx->Const.MaxVarying) {
+ linker_error_printf(prog, "shader uses too many varying vectors "
+ "(%u > %u)\n",
+ varying_vectors, ctx->Const.MaxVarying);
+ return false;
+ }
+ } else {
+ const unsigned float_components = varying_vectors * 4;
+ if (float_components > ctx->Const.MaxVarying * 4) {
+ linker_error_printf(prog, "shader uses too many varying components "
+ "(%u > %u)\n",
+ float_components, ctx->Const.MaxVarying * 4);
+ return false;
+ }
+ }
+
+ return true;
}
@@ -1666,9 +1693,13 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
if (prog->_LinkedShaders[i] == NULL)
continue;
- assign_varying_locations(prog,
- prog->_LinkedShaders[prev],
- prog->_LinkedShaders[i]);
+ if (!assign_varying_locations(ctx, prog,
+ prog->_LinkedShaders[prev],
+ prog->_LinkedShaders[i])) {
+ prog->LinkStatus = false;
+ goto done;
+ }
+
prev = i;
}
diff --git a/src/glsl/lower_if_to_cond_assign.cpp b/src/glsl/lower_if_to_cond_assign.cpp
index e3a1065d996..b637eb4fe1d 100644
--- a/src/glsl/lower_if_to_cond_assign.cpp
+++ b/src/glsl/lower_if_to_cond_assign.cpp
@@ -149,11 +149,9 @@ ir_visitor_status
ir_if_to_cond_assign_visitor::visit_leave(ir_if *ir)
{
/* Only flatten when beyond the GPU's maximum supported nesting depth. */
- if (this->depth <= this->max_depth)
+ if (this->depth-- <= this->max_depth)
return visit_continue;
- this->depth--;
-
bool found_control_flow = false;
ir_variable *cond_var;
ir_assignment *assign;
diff --git a/src/glsl/lower_instructions.cpp b/src/glsl/lower_instructions.cpp
index a5f61f213d0..806f8639959 100644
--- a/src/glsl/lower_instructions.cpp
+++ b/src/glsl/lower_instructions.cpp
@@ -168,8 +168,13 @@ lower_instructions_visitor::div_to_mul_rcp(ir_expression *ir)
op0 = new(ir) ir_expression(ir_binop_mul, vec_type, op0, op1);
- ir->operation = ir_unop_f2i;
- ir->operands[0] = op0;
+ if (ir->operands[1]->type->base_type == GLSL_TYPE_INT) {
+ ir->operation = ir_unop_f2i;
+ ir->operands[0] = op0;
+ } else {
+ ir->operation = ir_unop_i2u;
+ ir->operands[0] = new(ir) ir_expression(ir_unop_f2i, op0);
+ }
ir->operands[1] = NULL;
}
@@ -271,7 +276,7 @@ lower_instructions_visitor::visit_leave(ir_expression *ir)
break;
case ir_binop_mod:
- if (lowering(MOD_TO_FRACT))
+ if (lowering(MOD_TO_FRACT) && ir->type->is_float())
mod_to_fract(ir);
break;
diff --git a/src/glsl/lower_mat_op_to_vec.cpp b/src/glsl/lower_mat_op_to_vec.cpp
index 8cbbfa713c9..a371afc14c2 100644
--- a/src/glsl/lower_mat_op_to_vec.cpp
+++ b/src/glsl/lower_mat_op_to_vec.cpp
@@ -45,19 +45,19 @@ public:
ir_visitor_status visit_leave(ir_assignment *);
- ir_dereference *get_column(ir_variable *var, int col);
- ir_rvalue *get_element(ir_variable *var, int col, int row);
-
- void do_mul_mat_mat(ir_variable *result_var,
- ir_variable *a_var, ir_variable *b_var);
- void do_mul_mat_vec(ir_variable *result_var,
- ir_variable *a_var, ir_variable *b_var);
- void do_mul_vec_mat(ir_variable *result_var,
- ir_variable *a_var, ir_variable *b_var);
- void do_mul_mat_scalar(ir_variable *result_var,
- ir_variable *a_var, ir_variable *b_var);
- void do_equal_mat_mat(ir_variable *result_var, ir_variable *a_var,
- ir_variable *b_var, bool test_equal);
+ ir_dereference *get_column(ir_dereference *val, int col);
+ ir_rvalue *get_element(ir_dereference *val, int col, int row);
+
+ void do_mul_mat_mat(ir_dereference *result,
+ ir_dereference *a, ir_dereference *b);
+ void do_mul_mat_vec(ir_dereference *result,
+ ir_dereference *a, ir_dereference *b);
+ void do_mul_vec_mat(ir_dereference *result,
+ ir_dereference *a, ir_dereference *b);
+ void do_mul_mat_scalar(ir_dereference *result,
+ ir_dereference *a, ir_dereference *b);
+ void do_equal_mat_mat(ir_dereference *result, ir_dereference *a,
+ ir_dereference *b, bool test_equal);
void *mem_ctx;
bool made_progress;
@@ -97,182 +97,137 @@ do_mat_op_to_vec(exec_list *instructions)
}
ir_rvalue *
-ir_mat_op_to_vec_visitor::get_element(ir_variable *var, int col, int row)
+ir_mat_op_to_vec_visitor::get_element(ir_dereference *val, int col, int row)
{
- ir_dereference *deref;
+ val = get_column(val, col);
- deref = new(mem_ctx) ir_dereference_variable(var);
-
- if (var->type->is_matrix()) {
- deref = new(mem_ctx) ir_dereference_array(var,
- new(mem_ctx) ir_constant(col));
- } else {
- assert(col == 0);
- }
-
- return new(mem_ctx) ir_swizzle(deref, row, 0, 0, 0, 1);
+ return new(mem_ctx) ir_swizzle(val, row, 0, 0, 0, 1);
}
ir_dereference *
-ir_mat_op_to_vec_visitor::get_column(ir_variable *var, int row)
+ir_mat_op_to_vec_visitor::get_column(ir_dereference *val, int row)
{
- ir_dereference *deref;
-
- if (!var->type->is_matrix()) {
- deref = new(mem_ctx) ir_dereference_variable(var);
- } else {
- deref = new(mem_ctx) ir_dereference_variable(var);
- deref = new(mem_ctx) ir_dereference_array(deref,
- new(mem_ctx) ir_constant(row));
+ val = val->clone(mem_ctx, NULL);
+
+ if (val->type->is_matrix()) {
+ val = new(mem_ctx) ir_dereference_array(val,
+ new(mem_ctx) ir_constant(row));
}
- return deref;
+ return val;
}
void
-ir_mat_op_to_vec_visitor::do_mul_mat_mat(ir_variable *result_var,
- ir_variable *a_var,
- ir_variable *b_var)
+ir_mat_op_to_vec_visitor::do_mul_mat_mat(ir_dereference *result,
+ ir_dereference *a,
+ ir_dereference *b)
{
int b_col, i;
ir_assignment *assign;
ir_expression *expr;
- for (b_col = 0; b_col < b_var->type->matrix_columns; b_col++) {
- ir_rvalue *a = get_column(a_var, 0);
- ir_rvalue *b = get_element(b_var, b_col, 0);
-
+ for (b_col = 0; b_col < b->type->matrix_columns; b_col++) {
/* first column */
expr = new(mem_ctx) ir_expression(ir_binop_mul,
- a->type,
- a,
- b);
+ get_column(a, 0),
+ get_element(b, b_col, 0));
/* following columns */
- for (i = 1; i < a_var->type->matrix_columns; i++) {
+ for (i = 1; i < a->type->matrix_columns; i++) {
ir_expression *mul_expr;
- a = get_column(a_var, i);
- b = get_element(b_var, b_col, i);
-
mul_expr = new(mem_ctx) ir_expression(ir_binop_mul,
- a->type,
- a,
- b);
+ get_column(a, i),
+ get_element(b, b_col, i));
expr = new(mem_ctx) ir_expression(ir_binop_add,
- a->type,
expr,
mul_expr);
}
- ir_rvalue *result = get_column(result_var, b_col);
- assign = new(mem_ctx) ir_assignment(result,
- expr,
- NULL);
+ assign = new(mem_ctx) ir_assignment(get_column(result, b_col), expr);
base_ir->insert_before(assign);
}
}
void
-ir_mat_op_to_vec_visitor::do_mul_mat_vec(ir_variable *result_var,
- ir_variable *a_var,
- ir_variable *b_var)
+ir_mat_op_to_vec_visitor::do_mul_mat_vec(ir_dereference *result,
+ ir_dereference *a,
+ ir_dereference *b)
{
int i;
- ir_rvalue *a = get_column(a_var, 0);
- ir_rvalue *b = get_element(b_var, 0, 0);
ir_assignment *assign;
ir_expression *expr;
/* first column */
expr = new(mem_ctx) ir_expression(ir_binop_mul,
- result_var->type,
- a,
- b);
+ get_column(a, 0),
+ get_element(b, 0, 0));
/* following columns */
- for (i = 1; i < a_var->type->matrix_columns; i++) {
+ for (i = 1; i < a->type->matrix_columns; i++) {
ir_expression *mul_expr;
- a = get_column(a_var, i);
- b = get_element(b_var, 0, i);
-
mul_expr = new(mem_ctx) ir_expression(ir_binop_mul,
- result_var->type,
- a,
- b);
- expr = new(mem_ctx) ir_expression(ir_binop_add,
- result_var->type,
- expr,
- mul_expr);
+ get_column(a, i),
+ get_element(b, 0, i));
+ expr = new(mem_ctx) ir_expression(ir_binop_add, expr, mul_expr);
}
- ir_rvalue *result = new(mem_ctx) ir_dereference_variable(result_var);
- assign = new(mem_ctx) ir_assignment(result,
- expr,
- NULL);
+ result = result->clone(mem_ctx, NULL);
+ assign = new(mem_ctx) ir_assignment(result, expr);
base_ir->insert_before(assign);
}
void
-ir_mat_op_to_vec_visitor::do_mul_vec_mat(ir_variable *result_var,
- ir_variable *a_var,
- ir_variable *b_var)
+ir_mat_op_to_vec_visitor::do_mul_vec_mat(ir_dereference *result,
+ ir_dereference *a,
+ ir_dereference *b)
{
int i;
- for (i = 0; i < b_var->type->matrix_columns; i++) {
- ir_rvalue *a = new(mem_ctx) ir_dereference_variable(a_var);
- ir_rvalue *b = get_column(b_var, i);
- ir_rvalue *result;
+ for (i = 0; i < b->type->matrix_columns; i++) {
+ ir_rvalue *column_result;
ir_expression *column_expr;
ir_assignment *column_assign;
- result = new(mem_ctx) ir_dereference_variable(result_var);
- result = new(mem_ctx) ir_swizzle(result, i, 0, 0, 0, 1);
+ column_result = result->clone(mem_ctx, NULL);
+ column_result = new(mem_ctx) ir_swizzle(column_result, i, 0, 0, 0, 1);
column_expr = new(mem_ctx) ir_expression(ir_binop_dot,
- result->type,
- a,
- b);
+ a->clone(mem_ctx, NULL),
+ get_column(b, i));
- column_assign = new(mem_ctx) ir_assignment(result,
- column_expr,
- NULL);
+ column_assign = new(mem_ctx) ir_assignment(column_result,
+ column_expr);
base_ir->insert_before(column_assign);
}
}
void
-ir_mat_op_to_vec_visitor::do_mul_mat_scalar(ir_variable *result_var,
- ir_variable *a_var,
- ir_variable *b_var)
+ir_mat_op_to_vec_visitor::do_mul_mat_scalar(ir_dereference *result,
+ ir_dereference *a,
+ ir_dereference *b)
{
int i;
- for (i = 0; i < a_var->type->matrix_columns; i++) {
- ir_rvalue *a = get_column(a_var, i);
- ir_rvalue *b = new(mem_ctx) ir_dereference_variable(b_var);
- ir_rvalue *result = get_column(result_var, i);
+ for (i = 0; i < a->type->matrix_columns; i++) {
ir_expression *column_expr;
ir_assignment *column_assign;
column_expr = new(mem_ctx) ir_expression(ir_binop_mul,
- result->type,
- a,
- b);
+ get_column(a, i),
+ b->clone(mem_ctx, NULL));
- column_assign = new(mem_ctx) ir_assignment(result,
- column_expr,
- NULL);
+ column_assign = new(mem_ctx) ir_assignment(get_column(result, i),
+ column_expr);
base_ir->insert_before(column_assign);
}
}
void
-ir_mat_op_to_vec_visitor::do_equal_mat_mat(ir_variable *result_var,
- ir_variable *a_var,
- ir_variable *b_var,
+ir_mat_op_to_vec_visitor::do_equal_mat_mat(ir_dereference *result,
+ ir_dereference *a,
+ ir_dereference *b,
bool test_equal)
{
/* This essentially implements the following GLSL:
@@ -293,7 +248,7 @@ ir_mat_op_to_vec_visitor::do_equal_mat_mat(ir_variable *result_var,
* a[3] != b[3]);
* }
*/
- const unsigned columns = a_var->type->matrix_columns;
+ const unsigned columns = a->type->matrix_columns;
const glsl_type *const bvec_type =
glsl_type::get_instance(GLSL_TYPE_BOOL, columns, 1);
@@ -303,12 +258,10 @@ ir_mat_op_to_vec_visitor::do_equal_mat_mat(ir_variable *result_var,
this->base_ir->insert_before(tmp_bvec);
for (unsigned i = 0; i < columns; i++) {
- ir_dereference *const op0 = get_column(a_var, i);
- ir_dereference *const op1 = get_column(b_var, i);
-
ir_expression *const cmp =
new(this->mem_ctx) ir_expression(ir_binop_any_nequal,
- glsl_type::bool_type, op0, op1);
+ get_column(a, i),
+ get_column(b, i));
ir_dereference *const lhs =
new(this->mem_ctx) ir_dereference_variable(tmp_bvec);
@@ -319,23 +272,14 @@ ir_mat_op_to_vec_visitor::do_equal_mat_mat(ir_variable *result_var,
this->base_ir->insert_before(assign);
}
- ir_rvalue *const val =
- new(this->mem_ctx) ir_dereference_variable(tmp_bvec);
-
- ir_expression *any =
- new(this->mem_ctx) ir_expression(ir_unop_any, glsl_type::bool_type,
- val, NULL);
+ ir_rvalue *const val = new(this->mem_ctx) ir_dereference_variable(tmp_bvec);
+ ir_expression *any = new(this->mem_ctx) ir_expression(ir_unop_any, val);
if (test_equal)
- any = new(this->mem_ctx) ir_expression(ir_unop_logic_not,
- glsl_type::bool_type,
- any, NULL);
-
- ir_rvalue *const result =
- new(this->mem_ctx) ir_dereference_variable(result_var);
+ any = new(this->mem_ctx) ir_expression(ir_unop_logic_not, any);
ir_assignment *const assign =
- new(mem_ctx) ir_assignment(result, any, NULL);
+ new(mem_ctx) ir_assignment(result->clone(mem_ctx, NULL), any);
base_ir->insert_before(assign);
}
@@ -358,7 +302,7 @@ ir_mat_op_to_vec_visitor::visit_leave(ir_assignment *orig_assign)
{
ir_expression *orig_expr = orig_assign->rhs->as_expression();
unsigned int i, matrix_columns = 1;
- ir_variable *op_var[2];
+ ir_dereference *op[2];
if (!orig_expr)
return visit_continue;
@@ -370,51 +314,53 @@ ir_mat_op_to_vec_visitor::visit_leave(ir_assignment *orig_assign)
mem_ctx = ralloc_parent(orig_assign);
- ir_dereference_variable *lhs_deref =
+ ir_dereference_variable *result =
orig_assign->lhs->as_dereference_variable();
- assert(lhs_deref);
-
- ir_variable *result_var = lhs_deref->var;
+ assert(result);
/* Store the expression operands in temps so we can use them
* multiple times.
*/
for (i = 0; i < orig_expr->get_num_operands(); i++) {
ir_assignment *assign;
+ ir_dereference *deref = orig_expr->operands[i]->as_dereference();
- op_var[i] = new(mem_ctx) ir_variable(orig_expr->operands[i]->type,
- "mat_op_to_vec",
- ir_var_temporary);
- base_ir->insert_before(op_var[i]);
+ /* Avoid making a temporary if we don't need to to avoid aliasing. */
+ if (deref &&
+ deref->variable_referenced() != result->variable_referenced()) {
+ op[i] = deref;
+ continue;
+ }
- lhs_deref = new(mem_ctx) ir_dereference_variable(op_var[i]);
- assign = new(mem_ctx) ir_assignment(lhs_deref,
- orig_expr->operands[i],
- NULL);
+ /* Otherwise, store the operand in a temporary generally if it's
+ * not a dereference.
+ */
+ ir_variable *var = new(mem_ctx) ir_variable(orig_expr->operands[i]->type,
+ "mat_op_to_vec",
+ ir_var_temporary);
+ base_ir->insert_before(var);
+
+ /* Note that we use this dereference for the assignment. That means
+ * that others that want to use op[i] have to clone the deref.
+ */
+ op[i] = new(mem_ctx) ir_dereference_variable(var);
+ assign = new(mem_ctx) ir_assignment(op[i], orig_expr->operands[i]);
base_ir->insert_before(assign);
}
/* OK, time to break down this matrix operation. */
switch (orig_expr->operation) {
case ir_unop_neg: {
- const unsigned mask = (1U << result_var->type->vector_elements) - 1;
-
/* Apply the operation to each column.*/
for (i = 0; i < matrix_columns; i++) {
- ir_rvalue *op0 = get_column(op_var[0], i);
- ir_dereference *result = get_column(result_var, i);
ir_expression *column_expr;
ir_assignment *column_assign;
column_expr = new(mem_ctx) ir_expression(orig_expr->operation,
- result->type,
- op0,
- NULL);
-
- column_assign = new(mem_ctx) ir_assignment(result,
- column_expr,
- NULL,
- mask);
+ get_column(op[0], i));
+
+ column_assign = new(mem_ctx) ir_assignment(get_column(result, i),
+ column_expr);
assert(column_assign->write_mask != 0);
base_ir->insert_before(column_assign);
}
@@ -424,57 +370,49 @@ ir_mat_op_to_vec_visitor::visit_leave(ir_assignment *orig_assign)
case ir_binop_sub:
case ir_binop_div:
case ir_binop_mod: {
- const unsigned mask = (1U << result_var->type->vector_elements) - 1;
-
/* For most operations, the matrix version is just going
* column-wise through and applying the operation to each column
* if available.
*/
for (i = 0; i < matrix_columns; i++) {
- ir_rvalue *op0 = get_column(op_var[0], i);
- ir_rvalue *op1 = get_column(op_var[1], i);
- ir_dereference *result = get_column(result_var, i);
ir_expression *column_expr;
ir_assignment *column_assign;
column_expr = new(mem_ctx) ir_expression(orig_expr->operation,
- result->type,
- op0,
- op1);
-
- column_assign = new(mem_ctx) ir_assignment(result,
- column_expr,
- NULL,
- mask);
+ get_column(op[0], i),
+ get_column(op[1], i));
+
+ column_assign = new(mem_ctx) ir_assignment(get_column(result, i),
+ column_expr);
assert(column_assign->write_mask != 0);
base_ir->insert_before(column_assign);
}
break;
}
case ir_binop_mul:
- if (op_var[0]->type->is_matrix()) {
- if (op_var[1]->type->is_matrix()) {
- do_mul_mat_mat(result_var, op_var[0], op_var[1]);
- } else if (op_var[1]->type->is_vector()) {
- do_mul_mat_vec(result_var, op_var[0], op_var[1]);
+ if (op[0]->type->is_matrix()) {
+ if (op[1]->type->is_matrix()) {
+ do_mul_mat_mat(result, op[0], op[1]);
+ } else if (op[1]->type->is_vector()) {
+ do_mul_mat_vec(result, op[0], op[1]);
} else {
- assert(op_var[1]->type->is_scalar());
- do_mul_mat_scalar(result_var, op_var[0], op_var[1]);
+ assert(op[1]->type->is_scalar());
+ do_mul_mat_scalar(result, op[0], op[1]);
}
} else {
- assert(op_var[1]->type->is_matrix());
- if (op_var[0]->type->is_vector()) {
- do_mul_vec_mat(result_var, op_var[0], op_var[1]);
+ assert(op[1]->type->is_matrix());
+ if (op[0]->type->is_vector()) {
+ do_mul_vec_mat(result, op[0], op[1]);
} else {
- assert(op_var[0]->type->is_scalar());
- do_mul_mat_scalar(result_var, op_var[1], op_var[0]);
+ assert(op[0]->type->is_scalar());
+ do_mul_mat_scalar(result, op[1], op[0]);
}
}
break;
case ir_binop_all_equal:
case ir_binop_any_nequal:
- do_equal_mat_mat(result_var, op_var[1], op_var[0],
+ do_equal_mat_mat(result, op[1], op[0],
(orig_expr->operation == ir_binop_all_equal));
break;
diff --git a/src/glsl/main.cpp b/src/glsl/main.cpp
index 096da93dcef..7952bb1a3e3 100644
--- a/src/glsl/main.cpp
+++ b/src/glsl/main.cpp
@@ -76,6 +76,7 @@ initialize_context(struct gl_context *ctx, gl_api api)
ctx->Extensions.ARB_fragment_coord_conventions = GL_TRUE;
ctx->Extensions.EXT_texture_array = GL_TRUE;
ctx->Extensions.NV_texture_rectangle = GL_TRUE;
+ ctx->Extensions.EXT_texture3D = GL_TRUE;
/* GLSL 1.30 isn't fully supported, but we need to advertise 1.30 so that
* the built-in functions for 1.30 can be built.
diff --git a/src/glsl/opt_discard_simplification.cpp b/src/glsl/opt_discard_simplification.cpp
index 7c2928d271c..a19947ddd6c 100644
--- a/src/glsl/opt_discard_simplification.cpp
+++ b/src/glsl/opt_discard_simplification.cpp
@@ -104,9 +104,23 @@ static ir_discard *
find_unconditional_discard(exec_list &instructions)
{
foreach_list(n, &instructions) {
- ir_discard *ir = ((ir_instruction *) n)->as_discard();
- if (ir != NULL && ir->condition == NULL)
- return ir;
+ ir_instruction *ir = (ir_instruction *)n;
+
+ if (ir->ir_type == ir_type_return ||
+ ir->ir_type == ir_type_loop_jump)
+ return NULL;
+
+ /* So far, this code doesn't know how to look inside of flow
+ * control to see if a discard later on at this level is
+ * unconditional.
+ */
+ if (ir->ir_type == ir_type_if ||
+ ir->ir_type == ir_type_loop)
+ return NULL;
+
+ ir_discard *discard = ir->as_discard();
+ if (discard != NULL && discard->condition == NULL)
+ return discard;
}
return NULL;
}
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-16 15:34:18 +0000