prog_to_nir: OPCODE_EXP is not nir_op_fexp
[mesa.git] / src / mesa / program / prog_instruction.h
1 /*
2 * Mesa 3-D graphics library
3 *
4 * Copyright (C) 1999-2008 Brian Paul All Rights Reserved.
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the "Software"),
8 * to deal in the Software without restriction, including without limitation
9 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
10 * and/or sell copies of the Software, and to permit persons to whom the
11 * Software is furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included
14 * in all copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
17 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
20 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
21 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
22 * OTHER DEALINGS IN THE SOFTWARE.
23 */
24
25
26 /**
27 * \file prog_instruction.h
28 *
29 * Vertex/fragment program instruction datatypes and constants.
30 *
31 * \author Brian Paul
32 * \author Keith Whitwell
33 * \author Ian Romanick <idr@us.ibm.com>
34 */
35
36
37 #ifndef PROG_INSTRUCTION_H
38 #define PROG_INSTRUCTION_H
39
40
41 #include "main/glheader.h"
42
43
44 /**
45 * Swizzle indexes.
46 * Do not change!
47 */
48 /*@{*/
49 #define SWIZZLE_X 0
50 #define SWIZZLE_Y 1
51 #define SWIZZLE_Z 2
52 #define SWIZZLE_W 3
53 #define SWIZZLE_ZERO 4 /**< For SWZ instruction only */
54 #define SWIZZLE_ONE 5 /**< For SWZ instruction only */
55 #define SWIZZLE_NIL 7 /**< used during shader code gen (undefined value) */
56 /*@}*/
57
58 #define MAKE_SWIZZLE4(a,b,c,d) (((a)<<0) | ((b)<<3) | ((c)<<6) | ((d)<<9))
59 #define SWIZZLE_NOOP MAKE_SWIZZLE4(0,1,2,3)
60 #define GET_SWZ(swz, idx) (((swz) >> ((idx)*3)) & 0x7)
61 #define GET_BIT(msk, idx) (((msk) >> (idx)) & 0x1)
62 /** Determine if swz contains SWIZZLE_ZERO/ONE/NIL for any components. */
63 #define HAS_EXTENDED_SWIZZLE(swz) (swz & 0x924)
64
65 #define SWIZZLE_XYZW MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W)
66 #define SWIZZLE_XXXX MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)
67 #define SWIZZLE_YYYY MAKE_SWIZZLE4(SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y)
68 #define SWIZZLE_ZZZZ MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z)
69 #define SWIZZLE_WWWW MAKE_SWIZZLE4(SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W)
70
71
72 /**
73 * Writemask values, 1 bit per component.
74 */
75 /*@{*/
76 #define WRITEMASK_X 0x1
77 #define WRITEMASK_Y 0x2
78 #define WRITEMASK_XY 0x3
79 #define WRITEMASK_Z 0x4
80 #define WRITEMASK_XZ 0x5
81 #define WRITEMASK_YZ 0x6
82 #define WRITEMASK_XYZ 0x7
83 #define WRITEMASK_W 0x8
84 #define WRITEMASK_XW 0x9
85 #define WRITEMASK_YW 0xa
86 #define WRITEMASK_XYW 0xb
87 #define WRITEMASK_ZW 0xc
88 #define WRITEMASK_XZW 0xd
89 #define WRITEMASK_YZW 0xe
90 #define WRITEMASK_XYZW 0xf
91 /*@}*/
92
93
94 /**
95 * Condition codes
96 */
97 /*@{*/
98 #define COND_GT 1 /**< greater than zero */
99 #define COND_EQ 2 /**< equal to zero */
100 #define COND_LT 3 /**< less than zero */
101 #define COND_UN 4 /**< unordered (NaN) */
102 #define COND_GE 5 /**< greater than or equal to zero */
103 #define COND_LE 6 /**< less than or equal to zero */
104 #define COND_NE 7 /**< not equal to zero */
105 #define COND_TR 8 /**< always true */
106 #define COND_FL 9 /**< always false */
107 /*@}*/
108
109
110 /**
111 * Instruction precision for GL_NV_fragment_program
112 */
113 /*@{*/
114 #define FLOAT32 0x1
115 #define FLOAT16 0x2
116 #define FIXED12 0x4
117 /*@}*/
118
119
120 /**
121 * Saturation modes when storing values.
122 */
123 /*@{*/
124 #define SATURATE_OFF 0
125 #define SATURATE_ZERO_ONE 1
126 /*@}*/
127
128
129 /**
130 * Per-component negation masks
131 */
132 /*@{*/
133 #define NEGATE_X 0x1
134 #define NEGATE_Y 0x2
135 #define NEGATE_Z 0x4
136 #define NEGATE_W 0x8
137 #define NEGATE_XYZ 0x7
138 #define NEGATE_XYZW 0xf
139 #define NEGATE_NONE 0x0
140 /*@}*/
141
142
143 /**
144 * Program instruction opcodes for vertex, fragment and geometry programs.
145 */
146 typedef enum prog_opcode {
147 /* ARB_vp ARB_fp NV_vp NV_fp GLSL */
148 /*------------------------------------------*/
149 OPCODE_NOP = 0, /* X */
150 OPCODE_ABS, /* X X 1.1 X */
151 OPCODE_ADD, /* X X X X X */
152 OPCODE_ARL, /* X X X */
153 OPCODE_BGNLOOP, /* opt */
154 OPCODE_BGNSUB, /* opt */
155 OPCODE_BRK, /* 2 opt */
156 OPCODE_CAL, /* 2 2 opt */
157 OPCODE_CMP, /* X X */
158 OPCODE_CONT, /* opt */
159 OPCODE_COS, /* X 2 X X */
160 OPCODE_DDX, /* X X */
161 OPCODE_DDY, /* X X */
162 OPCODE_DP2, /* 2 X */
163 OPCODE_DP3, /* X X X X X */
164 OPCODE_DP4, /* X X X X X */
165 OPCODE_DPH, /* X X 1.1 */
166 OPCODE_DST, /* X X X X */
167 OPCODE_ELSE, /* opt */
168 OPCODE_END, /* X X X X opt */
169 OPCODE_ENDIF, /* opt */
170 OPCODE_ENDLOOP, /* opt */
171 OPCODE_ENDSUB, /* opt */
172 OPCODE_EX2, /* X X 2 X X */
173 OPCODE_EXP, /* X X */
174 OPCODE_FLR, /* X X 2 X X */
175 OPCODE_FRC, /* X X 2 X X */
176 OPCODE_IF, /* opt */
177 OPCODE_KIL, /* X X */
178 OPCODE_KIL_NV, /* X X */
179 OPCODE_LG2, /* X X 2 X X */
180 OPCODE_LIT, /* X X X X */
181 OPCODE_LOG, /* X X */
182 OPCODE_LRP, /* X X */
183 OPCODE_MAD, /* X X X X X */
184 OPCODE_MAX, /* X X X X X */
185 OPCODE_MIN, /* X X X X X */
186 OPCODE_MOV, /* X X X X X */
187 OPCODE_MUL, /* X X X X X */
188 OPCODE_NOISE1, /* X */
189 OPCODE_NOISE2, /* X */
190 OPCODE_NOISE3, /* X */
191 OPCODE_NOISE4, /* X */
192 OPCODE_POW, /* X X X X */
193 OPCODE_RCP, /* X X X X X */
194 OPCODE_RET, /* 2 2 opt */
195 OPCODE_RSQ, /* X X X X X */
196 OPCODE_SCS, /* X X */
197 OPCODE_SEQ, /* 2 X X */
198 OPCODE_SGE, /* X X X X X */
199 OPCODE_SGT, /* 2 X X */
200 OPCODE_SIN, /* X 2 X X */
201 OPCODE_SLE, /* 2 X X */
202 OPCODE_SLT, /* X X X X X */
203 OPCODE_SNE, /* 2 X X */
204 OPCODE_SSG, /* 2 X */
205 OPCODE_SUB, /* X X 1.1 X X */
206 OPCODE_SWZ, /* X X X */
207 OPCODE_TEX, /* X 3 X X */
208 OPCODE_TXB, /* X 3 X */
209 OPCODE_TXD, /* X X */
210 OPCODE_TXL, /* 3 2 X */
211 OPCODE_TXP, /* X X */
212 OPCODE_TXP_NV, /* 3 X */
213 OPCODE_TRUNC, /* X */
214 OPCODE_XPD, /* X X */
215 MAX_OPCODE
216 } gl_inst_opcode;
217
218
219 /**
220 * Number of bits for the src/dst register Index field.
221 * This limits the size of temp/uniform register files.
222 */
223 #define INST_INDEX_BITS 12
224
225
226 /**
227 * Instruction source register.
228 */
229 struct prog_src_register
230 {
231 GLuint File:4; /**< One of the PROGRAM_* register file values. */
232 GLint Index:(INST_INDEX_BITS+1); /**< Extra bit here for sign bit.
233 * May be negative for relative addressing.
234 */
235 GLuint Swizzle:12;
236 GLuint RelAddr:1;
237
238 /** Take the component-wise absolute value */
239 GLuint Abs:1;
240
241 /**
242 * Post-Abs negation.
243 * This will either be NEGATE_NONE or NEGATE_XYZW, except for the SWZ
244 * instruction which allows per-component negation.
245 */
246 GLuint Negate:4;
247
248 /**
249 * Is the register two-dimensional.
250 * Two dimensional registers are of the
251 * REGISTER[index][index2] format.
252 * They are used by the geometry shaders where
253 * the first index is the index within an array
254 * and the second index is the semantic of the
255 * array, e.g. gl_PositionIn[index] would become
256 * INPUT[index][gl_PositionIn]
257 */
258 GLuint HasIndex2:1;
259 GLuint RelAddr2:1;
260 GLint Index2:(INST_INDEX_BITS+1); /**< Extra bit here for sign bit.
261 * May be negative for relative
262 * addressing. */
263 };
264
265
266 /**
267 * Instruction destination register.
268 */
269 struct prog_dst_register
270 {
271 GLuint File:4; /**< One of the PROGRAM_* register file values */
272 GLuint Index:INST_INDEX_BITS; /**< Unsigned, never negative */
273 GLuint WriteMask:4;
274 GLuint RelAddr:1;
275
276 /**
277 * \name Conditional destination update control.
278 *
279 * \since
280 * NV_fragment_program_option, NV_vertex_program2, NV_vertex_program2_option.
281 */
282 /*@{*/
283 /**
284 * Takes one of the 9 possible condition values (EQ, FL, GT, GE, LE, LT,
285 * NE, TR, or UN). Dest reg is only written to if the matching
286 * (swizzled) condition code value passes. When a conditional update mask
287 * is not specified, this will be \c COND_TR.
288 */
289 GLuint CondMask:4;
290
291 /**
292 * Condition code swizzle value.
293 */
294 GLuint CondSwizzle:12;
295 };
296
297
298 /**
299 * Vertex/fragment program instruction.
300 */
301 struct prog_instruction
302 {
303 gl_inst_opcode Opcode;
304 struct prog_src_register SrcReg[3];
305 struct prog_dst_register DstReg;
306
307 /**
308 * Indicates that the instruction should update the condition code
309 * register.
310 *
311 * \since
312 * NV_fragment_program_option, NV_vertex_program2, NV_vertex_program2_option.
313 */
314 GLuint CondUpdate:1;
315
316 /**
317 * If prog_instruction::CondUpdate is \c GL_TRUE, this value selects the
318 * condition code register that is to be updated.
319 *
320 * In GL_NV_fragment_program or GL_NV_vertex_program2 mode, only condition
321 * code register 0 is available. In GL_NV_vertex_program3 mode, condition
322 * code registers 0 and 1 are available.
323 *
324 * \since
325 * NV_fragment_program_option, NV_vertex_program2, NV_vertex_program2_option.
326 */
327 GLuint CondDst:1;
328
329 /**
330 * Saturate each value of the vectored result to the range [0,1] or the
331 * range [-1,1]. \c SSAT mode (i.e., saturation to the range [-1,1]) is
332 * only available in NV_fragment_program2 mode.
333 * Value is one of the SATURATE_* tokens.
334 *
335 * \since
336 * NV_fragment_program_option, NV_vertex_program3.
337 */
338 GLuint SaturateMode:2;
339
340 /**
341 * Per-instruction selectable precision: FLOAT32, FLOAT16, FIXED12.
342 *
343 * \since
344 * NV_fragment_program_option.
345 */
346 GLuint Precision:3;
347
348 /**
349 * \name Extra fields for TEX, TXB, TXD, TXL, TXP instructions.
350 */
351 /*@{*/
352 /** Source texture unit. */
353 GLuint TexSrcUnit:5;
354
355 /** Source texture target, one of TEXTURE_{1D,2D,3D,CUBE,RECT}_INDEX */
356 GLuint TexSrcTarget:4;
357
358 /** True if tex instruction should do shadow comparison */
359 GLuint TexShadow:1;
360 /*@}*/
361
362 /**
363 * For BRA and CAL instructions, the location to jump to.
364 * For BGNLOOP, points to ENDLOOP (and vice-versa).
365 * For BRK, points to ENDLOOP
366 * For IF, points to ELSE or ENDIF.
367 * For ELSE, points to ENDIF.
368 */
369 GLint BranchTarget;
370
371 /** for driver use (try to remove someday) */
372 GLint Aux;
373
374 /** for debugging purposes */
375 const char *Comment;
376 };
377
378
379 #ifdef __cplusplus
380 extern "C" {
381 #endif
382
383 extern void
384 _mesa_init_instructions(struct prog_instruction *inst, GLuint count);
385
386 extern struct prog_instruction *
387 _mesa_alloc_instructions(GLuint numInst);
388
389 extern struct prog_instruction *
390 _mesa_copy_instructions(struct prog_instruction *dest,
391 const struct prog_instruction *src, GLuint n);
392
393 extern void
394 _mesa_free_instructions(struct prog_instruction *inst, GLuint count);
395
396 extern GLuint
397 _mesa_num_inst_src_regs(gl_inst_opcode opcode);
398
399 extern GLuint
400 _mesa_num_inst_dst_regs(gl_inst_opcode opcode);
401
402 extern GLboolean
403 _mesa_is_tex_instruction(gl_inst_opcode opcode);
404
405 extern GLboolean
406 _mesa_check_soa_dependencies(const struct prog_instruction *inst);
407
408 extern const char *
409 _mesa_opcode_string(gl_inst_opcode opcode);
410
411
412 #ifdef __cplusplus
413 } /* extern "C" */
414 #endif
415
416 #endif /* PROG_INSTRUCTION_H */