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
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
|
/**************************************************************************
*
* Copyright 2009 VMware, Inc.
* 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, sub license, 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 (including the
* next paragraph) 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 NON-INFRINGEMENT.
* IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS 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.
*
**************************************************************************/
#ifndef LP_RAST_PRIV_H
#define LP_RAST_PRIV_H
#include "os/os_thread.h"
#include "util/u_format.h"
#include "gallivm/lp_bld_debug.h"
#include "lp_memory.h"
#include "lp_rast.h"
#include "lp_scene.h"
#include "lp_state.h"
#include "lp_texture.h"
#include "lp_tile_soa.h"
#include "lp_limits.h"
/* If we crash in a jitted function, we can examine jit_line and jit_state
* to get some info. This is not thread-safe, however.
*/
#ifdef DEBUG
extern int jit_line;
extern const struct lp_rast_state *jit_state;
#define BEGIN_JIT_CALL(state) \
do { \
jit_line = __LINE__; \
jit_state = state; \
} while (0)
#define END_JIT_CALL() \
do { \
jit_line = 0; \
jit_state = NULL; \
} while (0)
#else
#define BEGIN_JIT_CALL(X)
#define END_JIT_CALL
#endif
struct lp_rasterizer;
/**
* Per-thread rasterization state
*/
struct lp_rasterizer_task
{
unsigned x, y; /**< Pos of this tile in framebuffer, in pixels */
uint8_t *color_tiles[PIPE_MAX_COLOR_BUFS];
uint8_t *depth_tile;
/** "back" pointer */
struct lp_rasterizer *rast;
/** "my" index */
unsigned thread_index;
/* occlude counter for visiable pixels */
uint32_t vis_counter;
pipe_semaphore work_ready;
pipe_semaphore work_done;
};
/**
* This is the state required while rasterizing tiles.
* Note that this contains per-thread information too.
* The tile size is TILE_SIZE x TILE_SIZE pixels.
*/
struct lp_rasterizer
{
boolean exit_flag;
/* Framebuffer stuff
*/
struct {
uint8_t *map;
unsigned stride;
unsigned blocksize;
} zsbuf;
struct {
unsigned nr_cbufs;
unsigned clear_color;
unsigned clear_depth;
char clear_stencil;
} state;
/** The incoming queue of scenes ready to rasterize */
struct lp_scene_queue *full_scenes;
/**
* The outgoing queue of processed scenes to return to setup module
*
* XXX: while scenes are per-context but the rasterizer is
* (potentially) shared, these empty scenes should be returned to
* the context which created them rather than retained here.
*/
/* struct lp_scene_queue *empty_scenes; */
/** The scene currently being rasterized by the threads */
struct lp_scene *curr_scene;
/** A task object for each rasterization thread */
struct lp_rasterizer_task tasks[LP_MAX_THREADS];
unsigned num_threads;
pipe_thread threads[LP_MAX_THREADS];
/** For synchronizing the rasterization threads */
pipe_barrier barrier;
};
void
lp_rast_shade_quads_mask(struct lp_rasterizer_task *task,
const struct lp_rast_shader_inputs *inputs,
unsigned x, unsigned y,
unsigned mask);
/**
* Get the pointer to a 4x4 depth/stencil block.
* We'll map the z/stencil buffer on demand here.
* Note that this may be called even when there's no z/stencil buffer - return
* NULL in that case.
* \param x, y location of 4x4 block in window coords
*/
static INLINE void *
lp_rast_get_depth_block_pointer(struct lp_rasterizer_task *task,
unsigned x, unsigned y)
{
const struct lp_rasterizer *rast = task->rast;
void *depth;
assert((x % TILE_VECTOR_WIDTH) == 0);
assert((y % TILE_VECTOR_HEIGHT) == 0);
if (!rast->zsbuf.map) {
/* Either out of memory or no zsbuf. Can't tell without access
* to the state. Just use dummy tile memory, but don't print
* the oom warning as this most likely because there is no
* zsbuf.
*/
return lp_dummy_tile;
}
depth = (rast->zsbuf.map +
rast->zsbuf.stride * y +
rast->zsbuf.blocksize * x * TILE_VECTOR_HEIGHT);
assert(lp_check_alignment(depth, 16));
return depth;
}
/**
* Get pointer to the swizzled color tile
*/
static INLINE uint8_t *
lp_rast_get_color_tile_pointer(struct lp_rasterizer_task *task,
unsigned buf, enum lp_texture_usage usage)
{
struct lp_rasterizer *rast = task->rast;
assert(task->x % TILE_SIZE == 0);
assert(task->y % TILE_SIZE == 0);
assert(buf < rast->state.nr_cbufs);
if (!task->color_tiles[buf]) {
struct pipe_surface *cbuf = rast->curr_scene->fb.cbufs[buf];
struct llvmpipe_resource *lpt;
assert(cbuf);
lpt = llvmpipe_resource(cbuf->texture);
task->color_tiles[buf] = lp_swizzled_cbuf[task->thread_index][buf];
if (usage != LP_TEX_USAGE_WRITE_ALL) {
llvmpipe_swizzle_cbuf_tile(lpt,
cbuf->face + cbuf->zslice,
cbuf->level,
task->x, task->y,
task->color_tiles[buf]);
}
}
return task->color_tiles[buf];
}
/**
* Get the pointer to a 4x4 color block (within a 64x64 tile).
* We'll map the color buffer on demand here.
* Note that this may be called even when there's no color buffers - return
* NULL in that case.
* \param x, y location of 4x4 block in window coords
*/
static INLINE uint8_t *
lp_rast_get_color_block_pointer(struct lp_rasterizer_task *task,
unsigned buf, unsigned x, unsigned y)
{
unsigned px, py, pixel_offset;
uint8_t *color;
assert((x % TILE_VECTOR_WIDTH) == 0);
assert((y % TILE_VECTOR_HEIGHT) == 0);
color = lp_rast_get_color_tile_pointer(task, buf, LP_TEX_USAGE_READ_WRITE);
assert(color);
px = x % TILE_SIZE;
py = y % TILE_SIZE;
pixel_offset = tile_pixel_offset(px, py, 0);
color = color + pixel_offset;
assert(lp_check_alignment(color, 16));
return color;
}
/**
* Shade all pixels in a 4x4 block. The fragment code omits the
* triangle in/out tests.
* \param x, y location of 4x4 block in window coords
*/
static INLINE void
lp_rast_shade_quads_all( struct lp_rasterizer_task *task,
const struct lp_rast_shader_inputs *inputs,
unsigned x, unsigned y )
{
const struct lp_rasterizer *rast = task->rast;
const struct lp_rast_state *state = inputs->state;
struct lp_fragment_shader_variant *variant = state->variant;
uint8_t *color[PIPE_MAX_COLOR_BUFS];
void *depth;
unsigned i;
/* color buffer */
for (i = 0; i < rast->state.nr_cbufs; i++)
color[i] = lp_rast_get_color_block_pointer(task, i, x, y);
depth = lp_rast_get_depth_block_pointer(task, x, y);
/* run shader on 4x4 block */
BEGIN_JIT_CALL(state);
variant->jit_function[RAST_WHOLE]( &state->jit_context,
x, y,
inputs->facing,
inputs->a0,
inputs->dadx,
inputs->dady,
color,
depth,
0xffff,
&task->vis_counter );
END_JIT_CALL();
}
#endif
|