3 * Copyright (C) 2004 David Airlie All Rights Reserved.
5 * Permission is hereby granted, free of charge, to any person obtaining a
6 * copy of this software and associated documentation files (the "Software"),
7 * to deal in the Software without restriction, including without limitation
8 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
9 * and/or sell copies of the Software, and to permit persons to whom the
10 * Software is furnished to do so, subject to the following conditions:
12 * The above copyright notice and this permission notice shall be included
13 * in all copies or substantial portions of the Software.
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
16 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * DAVID AIRLIE BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
19 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
20 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
27 #include "shader/program.h"
28 #include "shader/atifragshader.h"
29 #include "swrast/s_atifragshader.h"
33 * State for executing ATI fragment shader.
37 GLfloat Registers
[6][4]; /** six temporary registers */
38 GLfloat PrevPassRegisters
[6][4];
39 GLfloat Inputs
[2][4]; /** Primary, secondary input colors */
48 fetch_texel(GLcontext
* ctx
, const GLfloat texcoord
[4], GLfloat lambda
,
49 GLuint unit
, GLfloat color
[4])
52 SWcontext
*swrast
= SWRAST_CONTEXT(ctx
);
54 /* XXX use a float-valued TextureSample routine here!!! */
55 swrast
->TextureSample
[unit
](ctx
, ctx
->Texture
.Unit
[unit
]._Current
,
56 1, (const GLfloat(*)[4]) texcoord
,
58 color
[0] = CHAN_TO_FLOAT(rgba
[0]);
59 color
[1] = CHAN_TO_FLOAT(rgba
[1]);
60 color
[2] = CHAN_TO_FLOAT(rgba
[2]);
61 color
[3] = CHAN_TO_FLOAT(rgba
[3]);
65 apply_swizzle(GLfloat values
[4], GLuint swizzle
)
75 case GL_SWIZZLE_STR_ATI
:
80 case GL_SWIZZLE_STQ_ATI
:
85 case GL_SWIZZLE_STR_DR_ATI
:
90 case GL_SWIZZLE_STQ_DQ_ATI
:
91 /* make sure q is not 0 to avoid problems later with infinite values (texture lookup)? */
92 if (q
== 0.0F
) q
= 0.000000001;
102 apply_src_rep(GLint optype
, GLuint rep
, GLfloat
* val
)
109 start
= optype
? 3 : 0;
112 for (i
= start
; i
< end
; i
++) {
131 apply_src_mod(GLint optype
, GLuint mod
, GLfloat
* val
)
139 start
= optype
? 3 : 0;
142 for (i
= start
; i
< end
; i
++) {
143 if (mod
& GL_COMP_BIT_ATI
)
146 if (mod
& GL_BIAS_BIT_ATI
)
147 val
[i
] = val
[i
] - 0.5;
149 if (mod
& GL_2X_BIT_ATI
)
152 if (mod
& GL_NEGATE_BIT_ATI
)
158 apply_dst_mod(GLuint optype
, GLuint mod
, GLfloat
* val
)
161 GLint has_sat
= mod
& GL_SATURATE_BIT_ATI
;
164 mod
&= ~GL_SATURATE_BIT_ATI
;
166 start
= optype
? 3 : 0;
167 end
= optype
? 4 : 3;
169 for (i
= start
; i
< end
; i
++) {
180 case GL_HALF_BIT_ATI
:
181 val
[i
] = val
[i
] * 0.5;
183 case GL_QUARTER_BIT_ATI
:
184 val
[i
] = val
[i
] * 0.25;
186 case GL_EIGHTH_BIT_ATI
:
187 val
[i
] = val
[i
] * 0.125;
194 else if (val
[i
] > 1.0)
200 else if (val
[i
] > 8.0)
208 write_dst_addr(GLuint optype
, GLuint mod
, GLuint mask
, GLfloat
* src
,
212 apply_dst_mod(optype
, mod
, src
);
214 if (optype
== ATI_FRAGMENT_SHADER_COLOR_OP
) {
216 if (mask
& GL_RED_BIT_ATI
)
219 if (mask
& GL_GREEN_BIT_ATI
)
222 if (mask
& GL_BLUE_BIT_ATI
)
226 for (i
= 0; i
< 3; i
++)
235 finish_pass(struct atifs_machine
*machine
)
239 for (i
= 0; i
< 6; i
++) {
240 COPY_4V(machine
->PrevPassRegisters
[i
], machine
->Registers
[i
]);
244 struct ati_fs_opcode_st ati_fs_opcodes
[] = {
253 {GL_DOT2_ADD_ATI
, 3},
261 handle_pass_op(struct atifs_machine
*machine
, struct atifs_setupinst
*texinst
,
262 const SWspan
*span
, GLuint column
, GLuint idx
)
264 GLuint swizzle
= texinst
->swizzle
;
265 GLuint pass_tex
= texinst
->src
;
267 if (pass_tex
>= GL_TEXTURE0_ARB
&& pass_tex
<= GL_TEXTURE7_ARB
) {
268 pass_tex
-= GL_TEXTURE0_ARB
;
269 COPY_4V(machine
->Registers
[idx
],
270 span
->array
->attribs
[FRAG_ATTRIB_TEX0
+ pass_tex
][column
]);
272 else if (pass_tex
>= GL_REG_0_ATI
&& pass_tex
<= GL_REG_5_ATI
) {
273 pass_tex
-= GL_REG_0_ATI
;
274 COPY_4V(machine
->Registers
[idx
], machine
->PrevPassRegisters
[pass_tex
]);
276 apply_swizzle(machine
->Registers
[idx
], swizzle
);
281 handle_sample_op(GLcontext
* ctx
, struct atifs_machine
*machine
,
282 struct atifs_setupinst
*texinst
, const SWspan
*span
,
283 GLuint column
, GLuint idx
)
285 /* sample from unit idx using texinst->src as coords */
286 GLuint swizzle
= texinst
->swizzle
;
287 GLuint coord_source
= texinst
->src
;
288 GLfloat tex_coords
[4];
290 if (coord_source
>= GL_TEXTURE0_ARB
&& coord_source
<= GL_TEXTURE7_ARB
) {
291 coord_source
-= GL_TEXTURE0_ARB
;
293 span
->array
->attribs
[FRAG_ATTRIB_TEX0
+ coord_source
][column
]);
295 else if (coord_source
>= GL_REG_0_ATI
&& coord_source
<= GL_REG_5_ATI
) {
296 coord_source
-= GL_REG_0_ATI
;
297 COPY_4V(tex_coords
, machine
->PrevPassRegisters
[coord_source
]);
299 apply_swizzle(tex_coords
, swizzle
);
300 fetch_texel(ctx
, tex_coords
, 0.0F
, idx
, machine
->Registers
[idx
]);
303 #define SETUP_SRC_REG(optype, i, x) \
305 COPY_4V(src[optype][i], x); \
311 * Execute the given fragment shader.
312 * NOTE: we do everything in single-precision floating point
313 * \param ctx - rendering context
314 * \param shader - the shader to execute
315 * \param machine - virtual machine state
316 * \param span - the SWspan we're operating on
317 * \param column - which pixel [i] we're operating on in the span
320 execute_shader(GLcontext
*ctx
, const struct ati_fragment_shader
*shader
,
321 struct atifs_machine
*machine
, const SWspan
*span
,
325 struct atifs_instruction
*inst
;
326 struct atifs_setupinst
*texinst
;
331 GLfloat src
[2][3][4];
332 GLfloat zeros
[4] = { 0.0, 0.0, 0.0, 0.0 };
333 GLfloat ones
[4] = { 1.0, 1.0, 1.0, 1.0 };
334 GLfloat dst
[2][4], *dstp
;
336 for (pass
= 0; pass
< shader
->NumPasses
; pass
++) {
338 finish_pass(machine
);
339 for (j
= 0; j
< MAX_NUM_FRAGMENT_REGISTERS_ATI
; j
++) {
340 texinst
= &shader
->SetupInst
[pass
][j
];
341 if (texinst
->Opcode
== ATI_FRAGMENT_SHADER_PASS_OP
)
342 handle_pass_op(machine
, texinst
, span
, column
, j
);
343 else if (texinst
->Opcode
== ATI_FRAGMENT_SHADER_SAMPLE_OP
)
344 handle_sample_op(ctx
, machine
, texinst
, span
, column
, j
);
347 for (pc
= 0; pc
< shader
->numArithInstr
[pass
]; pc
++) {
348 inst
= &shader
->Instructions
[pass
][pc
];
350 /* setup the source registers for color and alpha ops */
351 for (optype
= 0; optype
< 2; optype
++) {
352 for (i
= 0; i
< inst
->ArgCount
[optype
]; i
++) {
353 GLint index
= inst
->SrcReg
[optype
][i
].Index
;
355 if (index
>= GL_REG_0_ATI
&& index
<= GL_REG_5_ATI
)
356 SETUP_SRC_REG(optype
, i
,
357 machine
->Registers
[index
- GL_REG_0_ATI
]);
358 else if (index
>= GL_CON_0_ATI
&& index
<= GL_CON_7_ATI
) {
359 if (shader
->LocalConstDef
& (1 << (index
- GL_CON_0_ATI
))) {
360 SETUP_SRC_REG(optype
, i
,
361 shader
->Constants
[index
- GL_CON_0_ATI
]);
363 SETUP_SRC_REG(optype
, i
,
364 ctx
->ATIFragmentShader
.GlobalConstants
[index
- GL_CON_0_ATI
]);
367 else if (index
== GL_ONE
)
368 SETUP_SRC_REG(optype
, i
, ones
);
369 else if (index
== GL_ZERO
)
370 SETUP_SRC_REG(optype
, i
, zeros
);
371 else if (index
== GL_PRIMARY_COLOR_EXT
)
372 SETUP_SRC_REG(optype
, i
,
373 machine
->Inputs
[ATI_FS_INPUT_PRIMARY
]);
374 else if (index
== GL_SECONDARY_INTERPOLATOR_ATI
)
375 SETUP_SRC_REG(optype
, i
,
376 machine
->Inputs
[ATI_FS_INPUT_SECONDARY
]);
378 apply_src_rep(optype
, inst
->SrcReg
[optype
][i
].argRep
,
380 apply_src_mod(optype
, inst
->SrcReg
[optype
][i
].argMod
,
385 /* Execute the operations - color then alpha */
386 for (optype
= 0; optype
< 2; optype
++) {
387 if (inst
->Opcode
[optype
]) {
388 switch (inst
->Opcode
[optype
]) {
391 for (i
= 0; i
< 3; i
++) {
393 src
[optype
][0][i
] + src
[optype
][1][i
];
396 dst
[optype
][3] = src
[optype
][0][3] + src
[optype
][1][3];
400 for (i
= 0; i
< 3; i
++) {
402 src
[optype
][0][i
] - src
[optype
][1][i
];
405 dst
[optype
][3] = src
[optype
][0][3] - src
[optype
][1][3];
409 for (i
= 0; i
< 3; i
++) {
411 src
[optype
][0][i
] * src
[optype
][1][i
];
414 dst
[optype
][3] = src
[optype
][0][3] * src
[optype
][1][3];
418 for (i
= 0; i
< 3; i
++) {
420 src
[optype
][0][i
] * src
[optype
][1][i
] +
425 src
[optype
][0][3] * src
[optype
][1][3] +
430 for (i
= 0; i
< 3; i
++) {
432 src
[optype
][0][i
] * src
[optype
][1][i
] + (1 -
440 src
[optype
][0][3] * src
[optype
][1][3] + (1 -
448 for (i
= 0; i
< 3; i
++) {
449 dst
[optype
][i
] = src
[optype
][0][i
];
452 dst
[optype
][3] = src
[optype
][0][3];
456 for (i
= 0; i
< 3; i
++) {
459 0.5) ? src
[optype
][0][i
] : src
[optype
][1][i
];
465 0.5) ? src
[optype
][0][3] : src
[optype
][1][3];
471 for (i
= 0; i
< 3; i
++) {
473 (src
[optype
][2][i
] >=
474 0) ? src
[optype
][0][i
] : src
[optype
][1][i
];
478 (src
[optype
][2][3] >=
479 0) ? src
[optype
][0][3] : src
[optype
][1][3];
482 case GL_DOT2_ADD_ATI
:
486 /* DOT 2 always uses the source from the color op */
487 /* could save recalculation of dot products for alpha inst */
488 result
= src
[0][0][0] * src
[0][1][0] +
489 src
[0][0][1] * src
[0][1][1] + src
[0][2][2];
491 for (i
= 0; i
< 3; i
++) {
492 dst
[optype
][i
] = result
;
496 dst
[optype
][3] = result
;
503 /* DOT 3 always uses the source from the color op */
504 result
= src
[0][0][0] * src
[0][1][0] +
505 src
[0][0][1] * src
[0][1][1] +
506 src
[0][0][2] * src
[0][1][2];
509 for (i
= 0; i
< 3; i
++) {
510 dst
[optype
][i
] = result
;
514 dst
[optype
][3] = result
;
521 /* DOT 4 always uses the source from the color op */
522 result
= src
[0][0][0] * src
[0][1][0] +
523 src
[0][0][1] * src
[0][1][1] +
524 src
[0][0][2] * src
[0][1][2] +
525 src
[0][0][3] * src
[0][1][3];
527 for (i
= 0; i
< 3; i
++) {
528 dst
[optype
][i
] = result
;
532 dst
[optype
][3] = result
;
540 /* write out the destination registers */
541 for (optype
= 0; optype
< 2; optype
++) {
542 if (inst
->Opcode
[optype
]) {
543 dstreg
= inst
->DstReg
[optype
].Index
;
544 dstp
= machine
->Registers
[dstreg
- GL_REG_0_ATI
];
546 if ((optype
== 0) || ((inst
->Opcode
[1] != GL_DOT2_ADD_ATI
) &&
547 (inst
->Opcode
[1] != GL_DOT3_ATI
) && (inst
->Opcode
[1] != GL_DOT4_ATI
)))
548 write_dst_addr(optype
, inst
->DstReg
[optype
].dstMod
,
549 inst
->DstReg
[optype
].dstMask
, dst
[optype
],
552 write_dst_addr(1, inst
->DstReg
[0].dstMod
, 0, dst
[1], dstp
);
561 * Init fragment shader virtual machine state.
564 init_machine(GLcontext
* ctx
, struct atifs_machine
*machine
,
565 const struct ati_fragment_shader
*shader
,
566 const SWspan
*span
, GLuint col
)
568 GLfloat (*inputs
)[4] = machine
->Inputs
;
571 for (i
= 0; i
< 6; i
++) {
572 for (j
= 0; j
< 4; j
++)
573 machine
->Registers
[i
][j
] = 0.0;
576 COPY_4V(inputs
[ATI_FS_INPUT_PRIMARY
], span
->array
->attribs
[FRAG_ATTRIB_COL0
][col
]);
577 COPY_4V(inputs
[ATI_FS_INPUT_SECONDARY
], span
->array
->attribs
[FRAG_ATTRIB_COL1
][col
]);
583 * Execute the current ATI shader program, operating on the given span.
586 _swrast_exec_fragment_shader(GLcontext
* ctx
, SWspan
*span
)
588 const struct ati_fragment_shader
*shader
= ctx
->ATIFragmentShader
.Current
;
589 struct atifs_machine machine
;
592 /* incoming colors should be floats */
593 ASSERT(span
->array
->ChanType
== GL_FLOAT
);
595 ctx
->_CurrentProgram
= GL_FRAGMENT_SHADER_ATI
;
597 for (i
= 0; i
< span
->end
; i
++) {
598 if (span
->array
->mask
[i
]) {
599 init_machine(ctx
, &machine
, shader
, span
, i
);
601 execute_shader(ctx
, shader
, &machine
, span
, i
);
603 /* store result color */
605 const GLfloat
*colOut
= machine
.Registers
[0];
606 /*fprintf(stderr,"outputs %f %f %f %f\n",
607 colOut[0], colOut[1], colOut[2], colOut[3]); */
608 COPY_4V(span
->array
->attribs
[FRAG_ATTRIB_COL0
][i
], colOut
);
613 ctx
->_CurrentProgram
= 0;