1 /*
2  * Copyright © 2010 Intel Corporation
3  *
4  * Permission is hereby granted, free of charge, to any person obtaining a
5  * copy of this software and associated documentation files (the "Software"),
6  * to deal in the Software without restriction, including without limitation
7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8  * and/or sell copies of the Software, and to permit persons to whom the
9  * Software is furnished to do so, subject to the following conditions:
10  *
11  * The above copyright notice and this permission notice (including the next
12  * paragraph) shall be included in all copies or substantial portions of the
13  * Software.
14  *
15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
21  * DEALINGS IN THE SOFTWARE.
22  */
23 
24 /**
25  * \file ir_optimization.h
26  *
27  * Prototypes for optimization passes to be called by the compiler and drivers.
28  */
29 
30 #ifndef GLSL_IR_OPTIMIZATION_H
31 #define GLSL_IR_OPTIMIZATION_H
32 
33 struct gl_linked_shader;
34 struct gl_shader_program;
35 
36 /* Operations for lower_instructions() */
37 #define SUB_TO_ADD_NEG     0x01
38 #define FDIV_TO_MUL_RCP    0x02
39 #define EXP_TO_EXP2        0x04
40 #define POW_TO_EXP2        0x08
41 #define LOG_TO_LOG2        0x10
42 #define MOD_TO_FLOOR       0x20
43 #define INT_DIV_TO_MUL_RCP 0x40
44 #define LDEXP_TO_ARITH     0x80
45 #define CARRY_TO_ARITH     0x100
46 #define BORROW_TO_ARITH    0x200
47 #define SAT_TO_CLAMP       0x400
48 #define DOPS_TO_DFRAC      0x800
49 #define DFREXP_DLDEXP_TO_ARITH    0x1000
50 #define BIT_COUNT_TO_MATH         0x02000
51 #define EXTRACT_TO_SHIFTS         0x04000
52 #define INSERT_TO_SHIFTS          0x08000
53 #define REVERSE_TO_SHIFTS         0x10000
54 #define FIND_LSB_TO_FLOAT_CAST    0x20000
55 #define FIND_MSB_TO_FLOAT_CAST    0x40000
56 #define IMUL_HIGH_TO_MUL          0x80000
57 #define DDIV_TO_MUL_RCP           0x100000
58 #define DIV_TO_MUL_RCP            (FDIV_TO_MUL_RCP | DDIV_TO_MUL_RCP)
59 #define SQRT_TO_ABS_SQRT          0x200000
60 #define MUL64_TO_MUL_AND_MUL_HIGH 0x400000
61 
62 /* Opertaions for lower_64bit_integer_instructions() */
63 #define MUL64                     (1U << 0)
64 #define SIGN64                    (1U << 1)
65 #define DIV64                     (1U << 2)
66 #define MOD64                     (1U << 3)
67 
68 /**
69  * \see class lower_packing_builtins_visitor
70  */
71 enum lower_packing_builtins_op {
72    LOWER_PACK_UNPACK_NONE               = 0x0000,
73 
74    LOWER_PACK_SNORM_2x16                = 0x0001,
75    LOWER_UNPACK_SNORM_2x16              = 0x0002,
76 
77    LOWER_PACK_UNORM_2x16                = 0x0004,
78    LOWER_UNPACK_UNORM_2x16              = 0x0008,
79 
80    LOWER_PACK_HALF_2x16                 = 0x0010,
81    LOWER_UNPACK_HALF_2x16               = 0x0020,
82 
83    LOWER_PACK_SNORM_4x8                 = 0x0040,
84    LOWER_UNPACK_SNORM_4x8               = 0x0080,
85 
86    LOWER_PACK_UNORM_4x8                 = 0x0100,
87    LOWER_UNPACK_UNORM_4x8               = 0x0200,
88 
89    LOWER_PACK_USE_BFI                   = 0x0400,
90    LOWER_PACK_USE_BFE                   = 0x0800,
91 };
92 
93 bool do_common_optimization(exec_list *ir, bool linked,
94 			    bool uniform_locations_assigned,
95                             const struct gl_shader_compiler_options *options,
96                             bool native_integers);
97 
98 bool ir_constant_fold(ir_rvalue **rvalue);
99 
100 bool do_rebalance_tree(exec_list *instructions);
101 bool do_algebraic(exec_list *instructions, bool native_integers,
102                   const struct gl_shader_compiler_options *options);
103 bool opt_conditional_discard(exec_list *instructions);
104 bool do_constant_folding(exec_list *instructions);
105 bool do_constant_variable(exec_list *instructions);
106 bool do_constant_variable_unlinked(exec_list *instructions);
107 bool do_copy_propagation_elements(exec_list *instructions);
108 bool do_constant_propagation(exec_list *instructions);
109 void do_dead_builtin_varyings(struct gl_context *ctx,
110                               gl_linked_shader *producer,
111                               gl_linked_shader *consumer,
112                               unsigned num_tfeedback_decls,
113                               class tfeedback_decl *tfeedback_decls);
114 bool do_dead_code(exec_list *instructions, bool uniform_locations_assigned);
115 bool do_dead_code_local(exec_list *instructions);
116 bool do_dead_code_unlinked(exec_list *instructions);
117 bool do_dead_functions(exec_list *instructions);
118 bool opt_flip_matrices(exec_list *instructions);
119 bool do_function_inlining(exec_list *instructions);
120 bool do_lower_jumps(exec_list *instructions, bool pull_out_jumps = true, bool lower_sub_return = true, bool lower_main_return = false, bool lower_continue = false, bool lower_break = false);
121 bool do_lower_texture_projection(exec_list *instructions);
122 bool do_if_simplification(exec_list *instructions);
123 bool opt_flatten_nested_if_blocks(exec_list *instructions);
124 bool do_discard_simplification(exec_list *instructions);
125 bool lower_if_to_cond_assign(gl_shader_stage stage, exec_list *instructions,
126                              unsigned max_depth = 0, unsigned min_branch_cost = 0);
127 bool do_mat_op_to_vec(exec_list *instructions);
128 bool do_minmax_prune(exec_list *instructions);
129 bool do_structure_splitting(exec_list *instructions);
130 bool optimize_swizzles(exec_list *instructions);
131 bool do_vectorize(exec_list *instructions);
132 bool do_tree_grafting(exec_list *instructions);
133 bool do_vec_index_to_cond_assign(exec_list *instructions);
134 bool do_vec_index_to_swizzle(exec_list *instructions);
135 bool lower_discard(exec_list *instructions);
136 void lower_discard_flow(exec_list *instructions);
137 bool lower_instructions(exec_list *instructions, unsigned what_to_lower);
138 bool lower_variable_index_to_cond_assign(gl_shader_stage stage,
139     exec_list *instructions, bool lower_input, bool lower_output,
140     bool lower_temp, bool lower_uniform);
141 bool lower_quadop_vector(exec_list *instructions, bool dont_lower_swz);
142 bool lower_const_arrays_to_uniforms(exec_list *instructions, unsigned stage, unsigned max_uniform_components);
143 bool lower_clip_cull_distance(struct gl_shader_program *prog,
144                               gl_linked_shader *shader);
145 ir_variable * lower_xfb_varying(void *mem_ctx,
146                                 gl_linked_shader *shader,
147                                 const char *old_var_name);
148 void lower_output_reads(unsigned stage, exec_list *instructions);
149 bool lower_packing_builtins(exec_list *instructions, int op_mask);
150 void lower_shared_reference(struct gl_context *ctx,
151                             struct gl_shader_program *prog,
152                             struct gl_linked_shader *shader);
153 void lower_ubo_reference(struct gl_linked_shader *shader,
154                          bool clamp_block_indices, bool use_std430_as_default);
155 void lower_packed_varyings(void *mem_ctx,
156                            unsigned locations_used,
157                            const uint8_t *components,
158                            ir_variable_mode mode,
159                            unsigned gs_input_vertices,
160                            gl_linked_shader *shader,
161                            bool disable_varying_packing,
162                            bool disable_xfb_packing,
163                            bool xfb_enabled);
164 bool lower_vector_insert(exec_list *instructions, bool lower_nonconstant_index);
165 bool lower_vector_derefs(gl_linked_shader *shader);
166 void lower_named_interface_blocks(void *mem_ctx, gl_linked_shader *shader);
167 bool optimize_redundant_jumps(exec_list *instructions);
168 bool optimize_split_arrays(exec_list *instructions, bool linked);
169 bool lower_offset_arrays(exec_list *instructions);
170 void optimize_dead_builtin_variables(exec_list *instructions,
171                                      enum ir_variable_mode other);
172 bool lower_tess_level(gl_linked_shader *shader);
173 
174 bool lower_vertex_id(gl_linked_shader *shader);
175 bool lower_cs_derived(gl_linked_shader *shader);
176 bool lower_blend_equation_advanced(gl_linked_shader *shader, bool coherent);
177 
178 bool lower_builtins(exec_list *instructions);
179 bool lower_subroutine(exec_list *instructions, struct _mesa_glsl_parse_state *state);
180 void propagate_invariance(exec_list *instructions);
181 
182 namespace ir_builder { class ir_factory; };
183 
184 ir_variable *compare_index_block(ir_builder::ir_factory &body,
185                                  ir_variable *index,
186                                  unsigned base, unsigned components);
187 
188 bool lower_64bit_integer_instructions(exec_list *instructions,
189                                       unsigned what_to_lower);
190 
191 void lower_precision(const struct gl_shader_compiler_options *options,
192                      exec_list *instructions);
193 
194 #endif /* GLSL_IR_OPTIMIZATION_H */
195