i965/fs: Rework representation of fragment output locations in NIR.
[mesa.git] / src / mesa / drivers / dri / i965 / brw_nir.h
1 /*
2 * Copyright © 2015 Intel Corporation
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
13 * Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
21 * IN THE SOFTWARE.
22 */
23
24 #pragma once
25
26 #include "brw_context.h"
27 #include "brw_reg.h"
28 #include "brw_shader.h"
29 #include "compiler/nir/nir.h"
30
31 #ifdef __cplusplus
32 extern "C" {
33 #endif
34
35 static inline int
36 type_size_scalar_bytes(const struct glsl_type *type)
37 {
38 return type_size_scalar(type) * 4;
39 }
40
41 static inline int
42 type_size_vec4_bytes(const struct glsl_type *type)
43 {
44 return type_size_vec4(type) * 16;
45 }
46
47 /* Flags set in the instr->pass_flags field by i965 analysis passes */
48 enum {
49 BRW_NIR_NON_BOOLEAN = 0x0,
50
51 /* Indicates that the given instruction's destination is a boolean
52 * value but that it needs to be resolved before it can be used.
53 * On Gen <= 5, CMP instructions return a 32-bit value where the bottom
54 * bit represents the actual true/false value of the compare and the top
55 * 31 bits are undefined. In order to use this value, we have to do a
56 * "resolve" operation by replacing the value of the CMP with -(x & 1)
57 * to sign-extend the bottom bit to 0/~0.
58 */
59 BRW_NIR_BOOLEAN_NEEDS_RESOLVE = 0x1,
60
61 /* Indicates that the given instruction's destination is a boolean
62 * value that has intentionally been left unresolved. Not all boolean
63 * values need to be resolved immediately. For instance, if we have
64 *
65 * CMP r1 r2 r3
66 * CMP r4 r5 r6
67 * AND r7 r1 r4
68 *
69 * We don't have to resolve the result of the two CMP instructions
70 * immediately because the AND still does an AND of the bottom bits.
71 * Instead, we can save ourselves instructions by delaying the resolve
72 * until after the AND. The result of the two CMP instructions is left
73 * as BRW_NIR_BOOLEAN_UNRESOLVED.
74 */
75 BRW_NIR_BOOLEAN_UNRESOLVED = 0x2,
76
77 /* Indicates a that the given instruction's destination is a boolean
78 * value that does not need a resolve. For instance, if you AND two
79 * values that are BRW_NIR_BOOLEAN_NEEDS_RESOLVE then we know that both
80 * values will be 0/~0 before we get them and the result of the AND is
81 * also guaranteed to be 0/~0 and does not need a resolve.
82 */
83 BRW_NIR_BOOLEAN_NO_RESOLVE = 0x3,
84
85 /* A mask to mask the boolean status values off of instr->pass_flags */
86 BRW_NIR_BOOLEAN_MASK = 0x3,
87 };
88
89 void brw_nir_analyze_boolean_resolves(nir_shader *nir);
90
91 nir_shader *brw_preprocess_nir(const struct brw_compiler *compiler,
92 nir_shader *nir);
93
94 bool brw_nir_lower_intrinsics(nir_shader *nir,
95 struct brw_stage_prog_data *prog_data);
96 void brw_nir_lower_vs_inputs(nir_shader *nir,
97 const struct brw_device_info *devinfo,
98 bool is_scalar,
99 bool use_legacy_snorm_formula,
100 const uint8_t *vs_attrib_wa_flags);
101 void brw_nir_lower_vue_inputs(nir_shader *nir, bool is_scalar,
102 const struct brw_vue_map *vue_map);
103 void brw_nir_lower_tes_inputs(nir_shader *nir, const struct brw_vue_map *vue);
104 void brw_nir_lower_fs_inputs(nir_shader *nir);
105 void brw_nir_lower_vue_outputs(nir_shader *nir, bool is_scalar);
106 void brw_nir_lower_tcs_outputs(nir_shader *nir, const struct brw_vue_map *vue);
107 void brw_nir_lower_fs_outputs(nir_shader *nir);
108 void brw_nir_lower_cs_shared(nir_shader *nir);
109
110 nir_shader *brw_postprocess_nir(nir_shader *nir,
111 const struct brw_device_info *devinfo,
112 bool is_scalar);
113
114 bool brw_nir_apply_attribute_workarounds(nir_shader *nir,
115 bool use_legacy_snorm_formula,
116 const uint8_t *attrib_wa_flags);
117
118 bool brw_nir_apply_trig_workarounds(nir_shader *nir);
119
120 void brw_nir_apply_tcs_quads_workaround(nir_shader *nir);
121
122 nir_shader *brw_nir_apply_sampler_key(nir_shader *nir,
123 const struct brw_device_info *devinfo,
124 const struct brw_sampler_prog_key_data *key,
125 bool is_scalar);
126
127 enum brw_reg_type brw_type_for_nir_type(nir_alu_type type);
128
129 enum glsl_base_type brw_glsl_base_type_for_nir_type(nir_alu_type type);
130
131 void brw_nir_setup_glsl_uniforms(nir_shader *shader,
132 struct gl_shader_program *shader_prog,
133 const struct gl_program *prog,
134 struct brw_stage_prog_data *stage_prog_data,
135 bool is_scalar);
136
137 void brw_nir_setup_arb_uniforms(nir_shader *shader, struct gl_program *prog,
138 struct brw_stage_prog_data *stage_prog_data);
139
140 bool brw_nir_opt_peephole_ffma(nir_shader *shader);
141
142 #define BRW_NIR_FRAG_OUTPUT_INDEX_SHIFT 0
143 #define BRW_NIR_FRAG_OUTPUT_INDEX_MASK INTEL_MASK(0, 0)
144 #define BRW_NIR_FRAG_OUTPUT_LOCATION_SHIFT 1
145 #define BRW_NIR_FRAG_OUTPUT_LOCATION_MASK INTEL_MASK(31, 1)
146
147 #ifdef __cplusplus
148 }
149 #endif