1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
|
/*
* Mesa 3-D graphics library
* Version: 6.5
*
* Copyright (C) 1999-2005 Brian Paul All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included
* in all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
* AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
/* Vertices are just an array of floats, with all the attributes
* packed. We currently assume a layout like:
*
* attr[0][0..3] - window position
* attr[1..n][0..3] - remaining attributes.
*
* Attributes are assumed to be 4 floats wide but are packed so that
* all the enabled attributes run contiguously.
*/
#include "pipe/p_util.h"
#include "pipe/p_defines.h"
#include "pipe/llvm/llvmtgsi.h"
#include "x86/rtasm/x86sse.h"
#include "sp_context.h"
#include "sp_state.h"
#include "sp_headers.h"
#include "sp_quad.h"
#include "sp_tex_sample.h"
struct quad_shade_stage
{
struct quad_stage stage;
struct tgsi_sampler samplers[PIPE_MAX_SAMPLERS];
struct tgsi_exec_machine machine;
struct tgsi_exec_vector *inputs, *outputs;
int colorOutSlot, depthOutSlot;
};
/** cast wrapper */
static INLINE struct quad_shade_stage *
quad_shade_stage(struct quad_stage *qs)
{
return (struct quad_shade_stage *) qs;
}
typedef void (XSTDCALL *codegen_function)(
const struct tgsi_exec_vector *input,
struct tgsi_exec_vector *output,
float (*constant)[4],
struct tgsi_exec_vector *temporary,
const struct tgsi_interp_coef *coef );
/* This should be done by the fragment shader execution unit (code
* generated from the decl instructions). Do it here for now.
*/
static void
shade_quad(
struct quad_stage *qs,
struct quad_header *quad )
{
struct quad_shade_stage *qss = quad_shade_stage( qs );
struct softpipe_context *softpipe = qs->softpipe;
const float fx = (float) quad->x0;
const float fy = (float) quad->y0;
struct tgsi_exec_machine *machine = &qss->machine;
/* Consts does not require 16 byte alignment. */
machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT];
machine->SamplerUnits = softpipe->sampler_units;
machine->InterpCoefs = quad->coef;
machine->Inputs[0].xyzw[0].f[0] = fx;
machine->Inputs[0].xyzw[0].f[1] = fx + 1.0f;
machine->Inputs[0].xyzw[0].f[2] = fx;
machine->Inputs[0].xyzw[0].f[3] = fx + 1.0f;
machine->Inputs[0].xyzw[1].f[0] = fy;
machine->Inputs[0].xyzw[1].f[1] = fy;
machine->Inputs[0].xyzw[1].f[2] = fy + 1.0f;
machine->Inputs[0].xyzw[1].f[3] = fy + 1.0f;
/* run shader */
/* XXX: Generated code effectively unusable until it handles quad->mask */
if( !quad->mask ) {
codegen_function func = (codegen_function) x86_get_func( &softpipe->fs->sse2_program );
func(
machine->Inputs,
machine->Outputs,
machine->Consts,
machine->Temps,
machine->InterpCoefs );
}
else {
#ifdef MESA_LLVM
/*ga_llvm_prog_exec(softpipe->fs->llvm_prog);*/
#endif
quad->mask &= tgsi_exec_machine_run( machine );
}
/* store result color */
if (qss->colorOutSlot >= 0) {
/* XXX need to handle multiple color outputs someday */
assert(qss->stage.softpipe->fs->shader.output_semantic_name[qss->colorOutSlot]
== TGSI_SEMANTIC_COLOR);
memcpy(
quad->outputs.color,
&machine->Outputs[qss->colorOutSlot].xyzw[0].f[0],
sizeof( quad->outputs.color ) );
}
/* store result Z */
if (qss->depthOutSlot >= 0) {
/* output[slot] is new Z */
uint i;
for (i = 0; i < 4; i++) {
quad->outputs.depth[i] = machine->Outputs[0].xyzw[2].f[i];
}
}
else {
/* copy input Z (which was interpolated by the executor) to output Z */
uint i;
for (i = 0; i < 4; i++) {
quad->outputs.depth[i] = machine->Inputs[0].xyzw[2].f[i];
}
}
/* shader may cull fragments */
if( quad->mask ) {
qs->next->run( qs->next, quad );
}
}
/**
* Per-primitive (or per-begin?) setup
*/
static void shade_begin(struct quad_stage *qs)
{
struct quad_shade_stage *qss = quad_shade_stage(qs);
struct softpipe_context *softpipe = qs->softpipe;
unsigned i;
/* set TGSI sampler state that varies */
for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
qss->samplers[i].state = softpipe->sampler[i];
qss->samplers[i].texture = softpipe->texture[i];
}
/* XXX only do this if the fragment shader changes... */
tgsi_exec_machine_init(&qss->machine,
softpipe->fs->shader.tokens,
PIPE_MAX_SAMPLERS,
qss->samplers );
/* find output slots for depth, color */
qss->colorOutSlot = -1;
qss->depthOutSlot = -1;
for (i = 0; i < qss->stage.softpipe->fs->shader.num_outputs; i++) {
switch (qss->stage.softpipe->fs->shader.output_semantic_name[i]) {
case TGSI_SEMANTIC_POSITION:
qss->depthOutSlot = i;
break;
case TGSI_SEMANTIC_COLOR:
qss->colorOutSlot = i;
break;
}
}
if (qs->next)
qs->next->begin(qs->next);
}
static void shade_destroy(struct quad_stage *qs)
{
free( qs );
}
struct quad_stage *sp_quad_shade_stage( struct softpipe_context *softpipe )
{
struct quad_shade_stage *qss = CALLOC_STRUCT(quad_shade_stage);
uint i;
/* allocate storage for program inputs/outputs, aligned to 16 bytes */
qss->inputs = malloc(PIPE_ATTRIB_MAX * sizeof(*qss->inputs) + 16);
qss->outputs = malloc(PIPE_ATTRIB_MAX * sizeof(*qss->outputs) + 16);
qss->machine.Inputs = align16(qss->inputs);
qss->machine.Outputs = align16(qss->outputs);
qss->stage.softpipe = softpipe;
qss->stage.begin = shade_begin;
qss->stage.run = shade_quad;
qss->stage.destroy = shade_destroy;
/* set TGSI sampler state that's constant */
for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
assert(softpipe->tex_cache[i]);
qss->samplers[i].get_samples = sp_get_samples;
qss->samplers[i].pipe = &softpipe->pipe;
qss->samplers[i].cache = softpipe->tex_cache[i];
}
return &qss->stage;
}
|