1 /**************************************************************************
3 * Copyright 2007-2008 Tungsten Graphics, Inc., Cedar Park, Texas.
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sub license, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
19 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
21 * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
22 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
23 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
24 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
26 **************************************************************************/
31 #include "pipe/p_compiler.h"
33 #if defined __cplusplus
37 #define MAX_LABELS (4 * 1024) /**< basically, max instructions */
39 #define NUM_CHANNELS 4 /* R,G,B,A */
40 #define QUAD_SIZE 4 /* 4 pixel/quad */
43 * Registers may be treated as float, signed int or unsigned int.
45 union tgsi_exec_channel
49 unsigned u
[QUAD_SIZE
];
53 * A vector[RGBA] of channels[4 pixels]
55 struct tgsi_exec_vector
57 union tgsi_exec_channel xyzw
[NUM_CHANNELS
];
61 * For fragment programs, information for computing fragment input
62 * values from plane equation of the triangle/line.
64 struct tgsi_interp_coef
66 float a0
[NUM_CHANNELS
]; /* in an xyzw layout */
67 float dadx
[NUM_CHANNELS
];
68 float dady
[NUM_CHANNELS
];
72 * Information for sampling textures, which must be implemented
73 * by code outside the TGSI executor.
77 /** Get samples for four fragments in a quad */
78 void (*get_samples
)(struct tgsi_sampler
*sampler
,
79 const float s
[QUAD_SIZE
],
80 const float t
[QUAD_SIZE
],
81 const float p
[QUAD_SIZE
],
83 float rgba
[NUM_CHANNELS
][QUAD_SIZE
]);
87 * For branching/calling subroutines.
89 struct tgsi_exec_labels
91 unsigned labels
[MAX_LABELS
][2];
96 #define TGSI_EXEC_NUM_TEMPS 128
97 #define TGSI_EXEC_NUM_TEMP_EXTRAS 6
98 #define TGSI_EXEC_NUM_IMMEDIATES 256
101 * Locations of various utility registers (_I = Index, _C = Channel)
103 #define TGSI_EXEC_TEMP_00000000_I (TGSI_EXEC_NUM_TEMPS + 0)
104 #define TGSI_EXEC_TEMP_00000000_C 0
106 #define TGSI_EXEC_TEMP_7FFFFFFF_I (TGSI_EXEC_NUM_TEMPS + 0)
107 #define TGSI_EXEC_TEMP_7FFFFFFF_C 1
109 #define TGSI_EXEC_TEMP_80000000_I (TGSI_EXEC_NUM_TEMPS + 0)
110 #define TGSI_EXEC_TEMP_80000000_C 2
112 #define TGSI_EXEC_TEMP_FFFFFFFF_I (TGSI_EXEC_NUM_TEMPS + 0)
113 #define TGSI_EXEC_TEMP_FFFFFFFF_C 3
115 #define TGSI_EXEC_TEMP_ONE_I (TGSI_EXEC_NUM_TEMPS + 1)
116 #define TGSI_EXEC_TEMP_ONE_C 0
118 #define TGSI_EXEC_TEMP_TWO_I (TGSI_EXEC_NUM_TEMPS + 1)
119 #define TGSI_EXEC_TEMP_TWO_C 1
121 #define TGSI_EXEC_TEMP_128_I (TGSI_EXEC_NUM_TEMPS + 1)
122 #define TGSI_EXEC_TEMP_128_C 2
124 #define TGSI_EXEC_TEMP_MINUS_128_I (TGSI_EXEC_NUM_TEMPS + 1)
125 #define TGSI_EXEC_TEMP_MINUS_128_C 3
127 #define TGSI_EXEC_TEMP_KILMASK_I (TGSI_EXEC_NUM_TEMPS + 2)
128 #define TGSI_EXEC_TEMP_KILMASK_C 0
130 #define TGSI_EXEC_TEMP_OUTPUT_I (TGSI_EXEC_NUM_TEMPS + 2)
131 #define TGSI_EXEC_TEMP_OUTPUT_C 1
133 #define TGSI_EXEC_TEMP_PRIMITIVE_I (TGSI_EXEC_NUM_TEMPS + 2)
134 #define TGSI_EXEC_TEMP_PRIMITIVE_C 2
136 /* NVIDIA condition code (CC) vector
138 #define TGSI_EXEC_CC_GT 0x01
139 #define TGSI_EXEC_CC_EQ 0x02
140 #define TGSI_EXEC_CC_LT 0x04
141 #define TGSI_EXEC_CC_UN 0x08
143 #define TGSI_EXEC_CC_X_MASK 0x000000ff
144 #define TGSI_EXEC_CC_X_SHIFT 0
145 #define TGSI_EXEC_CC_Y_MASK 0x0000ff00
146 #define TGSI_EXEC_CC_Y_SHIFT 8
147 #define TGSI_EXEC_CC_Z_MASK 0x00ff0000
148 #define TGSI_EXEC_CC_Z_SHIFT 16
149 #define TGSI_EXEC_CC_W_MASK 0xff000000
150 #define TGSI_EXEC_CC_W_SHIFT 24
152 #define TGSI_EXEC_TEMP_CC_I (TGSI_EXEC_NUM_TEMPS + 2)
153 #define TGSI_EXEC_TEMP_CC_C 3
155 #define TGSI_EXEC_TEMP_THREE_I (TGSI_EXEC_NUM_TEMPS + 3)
156 #define TGSI_EXEC_TEMP_THREE_C 0
158 #define TGSI_EXEC_TEMP_HALF_I (TGSI_EXEC_NUM_TEMPS + 3)
159 #define TGSI_EXEC_TEMP_HALF_C 1
161 /* execution mask, each value is either 0 or ~0 */
162 #define TGSI_EXEC_MASK_I (TGSI_EXEC_NUM_TEMPS + 3)
163 #define TGSI_EXEC_MASK_C 2
165 #define TGSI_EXEC_TEMP_R0 (TGSI_EXEC_NUM_TEMPS + 4)
167 #define TGSI_EXEC_TEMP_ADDR (TGSI_EXEC_NUM_TEMPS + 5)
170 #define TGSI_EXEC_MAX_COND_NESTING 20
171 #define TGSI_EXEC_MAX_LOOP_NESTING 20
172 #define TGSI_EXEC_MAX_CALL_NESTING 20
174 /* The maximum number of input attributes per vertex. For 2D
175 * input register files, this is the stride between two 1D
178 #define TGSI_EXEC_MAX_INPUT_ATTRIBS 17
180 /* The maximum number of constant vectors per constant buffer.
182 #define TGSI_EXEC_MAX_CONST_BUFFER 4096
185 * Run-time virtual machine state for executing TGSI shader.
187 struct tgsi_exec_machine
189 /* Total = program temporaries + internal temporaries
190 * + 1 padding to align to 16 bytes
192 struct tgsi_exec_vector _Temps
[TGSI_EXEC_NUM_TEMPS
+
193 TGSI_EXEC_NUM_TEMP_EXTRAS
+ 1];
196 * This will point to _Temps after aligning to 16B boundary.
198 struct tgsi_exec_vector
*Temps
;
199 struct tgsi_exec_vector
*Addrs
;
201 struct tgsi_sampler
**Samplers
;
203 float Imms
[TGSI_EXEC_NUM_IMMEDIATES
][4];
205 const float (*Consts
)[4];
206 struct tgsi_exec_vector
*Inputs
;
207 struct tgsi_exec_vector
*Outputs
;
208 const struct tgsi_token
*Tokens
; /**< Declarations, instructions */
209 unsigned Processor
; /**< TGSI_PROCESSOR_x */
211 /* GEOMETRY processor only. */
212 unsigned *Primitives
;
214 /* FRAGMENT processor only. */
215 const struct tgsi_interp_coef
*InterpCoefs
;
216 struct tgsi_exec_vector QuadPos
;
218 /* Conditional execution masks */
219 uint CondMask
; /**< For IF/ELSE/ENDIF */
220 uint LoopMask
; /**< For BGNLOOP/ENDLOOP */
221 uint ContMask
; /**< For loop CONT statements */
222 uint FuncMask
; /**< For function calls */
223 uint ExecMask
; /**< = CondMask & LoopMask */
225 /** Condition mask stack (for nested conditionals) */
226 uint CondStack
[TGSI_EXEC_MAX_COND_NESTING
];
229 /** Loop mask stack (for nested loops) */
230 uint LoopStack
[TGSI_EXEC_MAX_LOOP_NESTING
];
233 /** Loop continue mask stack (see comments in tgsi_exec.c) */
234 uint ContStack
[TGSI_EXEC_MAX_LOOP_NESTING
];
237 /** Function execution mask stack (for executing subroutine code) */
238 uint FuncStack
[TGSI_EXEC_MAX_CALL_NESTING
];
241 /** Function call stack for saving/restoring the program counter */
242 uint CallStack
[TGSI_EXEC_MAX_CALL_NESTING
];
245 struct tgsi_full_instruction
*Instructions
;
246 uint NumInstructions
;
248 struct tgsi_full_declaration
*Declarations
;
249 uint NumDeclarations
;
251 struct tgsi_exec_labels Labels
;
255 tgsi_exec_machine_init(
256 struct tgsi_exec_machine
*mach
);
260 tgsi_exec_machine_bind_shader(
261 struct tgsi_exec_machine
*mach
,
262 const struct tgsi_token
*tokens
,
264 struct tgsi_sampler
**samplers
);
267 tgsi_exec_machine_run(
268 struct tgsi_exec_machine
*mach
);
272 tgsi_exec_machine_free_data(struct tgsi_exec_machine
*mach
);
276 tgsi_set_kill_mask(struct tgsi_exec_machine
*mach
, unsigned mask
)
278 mach
->Temps
[TGSI_EXEC_TEMP_KILMASK_I
].xyzw
[TGSI_EXEC_TEMP_KILMASK_C
].u
[0] =
283 /** Set execution mask values prior to executing the shader */
285 tgsi_set_exec_mask(struct tgsi_exec_machine
*mach
,
286 boolean ch0
, boolean ch1
, boolean ch2
, boolean ch3
)
288 int *mask
= mach
->Temps
[TGSI_EXEC_MASK_I
].xyzw
[TGSI_EXEC_MASK_C
].i
;
289 mask
[0] = ch0
? ~0 : 0;
290 mask
[1] = ch1
? ~0 : 0;
291 mask
[2] = ch2
? ~0 : 0;
292 mask
[3] = ch3
? ~0 : 0;
296 #if defined __cplusplus
300 #endif /* TGSI_EXEC_H */