compiler: add shader_info.vs.blit_sgprs_amd
[mesa.git] / src / compiler / shader_info.h
1 /*
2 * Copyright © 2016 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
25 #ifndef SHADER_INFO_H
26 #define SHADER_INFO_H
27
28 #include "shader_enums.h"
29 #include <stdint.h>
30
31 #ifdef __cplusplus
32 extern "C" {
33 #endif
34
35 struct spirv_supported_capabilities {
36 bool address;
37 bool atomic_storage;
38 bool demote_to_helper_invocation;
39 bool derivative_group;
40 bool descriptor_array_dynamic_indexing;
41 bool descriptor_array_non_uniform_indexing;
42 bool descriptor_indexing;
43 bool device_group;
44 bool draw_parameters;
45 bool float64;
46 bool fragment_shader_sample_interlock;
47 bool fragment_shader_pixel_interlock;
48 bool geometry_streams;
49 bool image_ms_array;
50 bool image_read_without_format;
51 bool image_write_without_format;
52 bool int8;
53 bool int16;
54 bool int64;
55 bool int64_atomics;
56 bool kernel;
57 bool min_lod;
58 bool multiview;
59 bool physical_storage_buffer_address;
60 bool post_depth_coverage;
61 bool runtime_descriptor_array;
62 bool shader_viewport_index_layer;
63 bool stencil_export;
64 bool storage_8bit;
65 bool storage_16bit;
66 bool storage_image_ms;
67 bool subgroup_arithmetic;
68 bool subgroup_ballot;
69 bool subgroup_basic;
70 bool subgroup_quad;
71 bool subgroup_shuffle;
72 bool subgroup_vote;
73 bool tessellation;
74 bool transform_feedback;
75 bool variable_pointers;
76 bool float16;
77 bool amd_gcn_shader;
78 bool amd_shader_ballot;
79 bool amd_trinary_minmax;
80 };
81
82 typedef struct shader_info {
83 const char *name;
84
85 /* Descriptive name provided by the client; may be NULL */
86 const char *label;
87
88 /** The shader stage, such as MESA_SHADER_VERTEX. */
89 gl_shader_stage stage;
90
91 /** The shader stage in a non SSO linked program that follows this stage,
92 * such as MESA_SHADER_FRAGMENT.
93 */
94 gl_shader_stage next_stage;
95
96 /* Number of textures used by this shader */
97 unsigned num_textures;
98 /* Number of uniform buffers used by this shader */
99 unsigned num_ubos;
100 /* Number of atomic buffers used by this shader */
101 unsigned num_abos;
102 /* Number of shader storage buffers used by this shader */
103 unsigned num_ssbos;
104 /* Number of images used by this shader */
105 unsigned num_images;
106
107 /* Which inputs are actually read */
108 uint64_t inputs_read;
109 /* Which outputs are actually written */
110 uint64_t outputs_written;
111 /* Which outputs are actually read */
112 uint64_t outputs_read;
113 /* Which system values are actually read */
114 uint64_t system_values_read;
115
116 /* Which patch inputs are actually read */
117 uint32_t patch_inputs_read;
118 /* Which patch outputs are actually written */
119 uint32_t patch_outputs_written;
120 /* Which patch outputs are read */
121 uint32_t patch_outputs_read;
122
123 /* Whether or not this shader ever uses textureGather() */
124 bool uses_texture_gather;
125
126 /** Bitfield of which textures are used */
127 uint32_t textures_used;
128
129 /** Bitfield of which textures are used by texelFetch() */
130 uint32_t textures_used_by_txf;
131
132 /**
133 * True if this shader uses the fddx/fddy opcodes.
134 *
135 * Note that this does not include the "fine" and "coarse" variants.
136 */
137 bool uses_fddx_fddy;
138
139 /**
140 * True if this shader uses 64-bit ALU operations
141 */
142 bool uses_64bit;
143
144 /* The size of the gl_ClipDistance[] array, if declared. */
145 unsigned clip_distance_array_size;
146
147 /* The size of the gl_CullDistance[] array, if declared. */
148 unsigned cull_distance_array_size;
149
150 /* Whether or not separate shader objects were used */
151 bool separate_shader;
152
153 /** Was this shader linked with any transform feedback varyings? */
154 bool has_transform_feedback_varyings;
155
156 union {
157 struct {
158 /* Which inputs are doubles */
159 uint64_t double_inputs;
160
161 /* For AMD-specific driver-internal shaders. It replaces vertex
162 * buffer loads with code generating VS inputs from scalar registers.
163 *
164 * Valid values: SI_VS_BLIT_SGPRS_POS_*
165 */
166 unsigned blit_sgprs_amd;
167
168 /* True if the shader writes position in window space coordinates pre-transform */
169 bool window_space_position;
170 } vs;
171
172 struct {
173 /** The number of vertices recieves per input primitive */
174 unsigned vertices_in;
175
176 /** The output primitive type (GL enum value) */
177 unsigned output_primitive;
178
179 /** The input primitive type (GL enum value) */
180 unsigned input_primitive;
181
182 /** The maximum number of vertices the geometry shader might write. */
183 unsigned vertices_out;
184
185 /** 1 .. MAX_GEOMETRY_SHADER_INVOCATIONS */
186 unsigned invocations;
187
188 /** Whether or not this shader uses EndPrimitive */
189 bool uses_end_primitive;
190
191 /** Whether or not this shader uses non-zero streams */
192 bool uses_streams;
193 } gs;
194
195 struct {
196 bool uses_discard;
197
198 /**
199 * True if this fragment shader requires helper invocations. This
200 * can be caused by the use of ALU derivative ops, texture
201 * instructions which do implicit derivatives, and the use of quad
202 * subgroup operations.
203 */
204 bool needs_helper_invocations;
205
206 /**
207 * Whether any inputs are declared with the "sample" qualifier.
208 */
209 bool uses_sample_qualifier;
210
211 /**
212 * Whether early fragment tests are enabled as defined by
213 * ARB_shader_image_load_store.
214 */
215 bool early_fragment_tests;
216
217 /**
218 * Defined by INTEL_conservative_rasterization.
219 */
220 bool inner_coverage;
221
222 bool post_depth_coverage;
223
224 /**
225 * \name ARB_fragment_coord_conventions
226 * @{
227 */
228 bool pixel_center_integer;
229 bool origin_upper_left;
230 /*@}*/
231
232 bool pixel_interlock_ordered;
233 bool pixel_interlock_unordered;
234 bool sample_interlock_ordered;
235 bool sample_interlock_unordered;
236
237 /**
238 * Flags whether NIR's base types on the FS color outputs should be
239 * ignored.
240 *
241 * GLSL requires that fragment shader output base types match the
242 * render target's base types for the behavior to be defined. From
243 * the GL 4.6 spec:
244 *
245 * "If the values written by the fragment shader do not match the
246 * format(s) of the corresponding color buffer(s), the result is
247 * undefined."
248 *
249 * However, for NIR shaders translated from TGSI, we don't have the
250 * output types any more, so the driver will need to do whatever
251 * fixups are necessary to handle effectively untyped data being
252 * output from the FS.
253 */
254 bool untyped_color_outputs;
255
256 /** gl_FragDepth layout for ARB_conservative_depth. */
257 enum gl_frag_depth_layout depth_layout;
258 } fs;
259
260 struct {
261 unsigned local_size[3];
262
263 bool local_size_variable;
264
265 /**
266 * Size of shared variables accessed by the compute shader.
267 */
268 unsigned shared_size;
269
270
271 /**
272 * pointer size is:
273 * AddressingModelLogical: 0 (default)
274 * AddressingModelPhysical32: 32
275 * AddressingModelPhysical64: 64
276 */
277 unsigned ptr_size;
278
279 /*
280 * Arrangement of invocations used to calculate derivatives in a compute
281 * shader. From NV_compute_shader_derivatives.
282 */
283 enum gl_derivative_group derivative_group;
284 } cs;
285
286 /* Applies to both TCS and TES. */
287 struct {
288 /** The number of vertices in the TCS output patch. */
289 unsigned tcs_vertices_out;
290
291 uint32_t primitive_mode; /* GL_TRIANGLES, GL_QUADS or GL_ISOLINES */
292 enum gl_tess_spacing spacing;
293 /** Is the vertex order counterclockwise? */
294 bool ccw;
295 bool point_mode;
296 } tess;
297 };
298 } shader_info;
299
300 #ifdef __cplusplus
301 }
302 #endif
303
304 #endif /* SHADER_INFO_H */