1 /*
2 * Copyright (C) 2021 Alyssa Rosenzweig <alyssa@rosenzweig.io>
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
13 * Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 * SOFTWARE.
22 */
23
24 #include <stdio.h>
25 #include <assert.h>
26 #include <stdlib.h>
27 #include <stdbool.h>
28 #include <stdint.h>
29 #include "tiling.h"
30
31 /* Z-order with 64x64 tiles:
32 *
33 * [y5][x5][y4][x4][y3][x3][y2][x2][y1][x1][y0][x0]
34 *
35 * Efficient tiling algorithm described in
36 * https://fgiesen.wordpress.com/2011/01/17/texture-tiling-and-swizzling/ but
37 * for posterity, we split into X and Y parts, and are faced with the problem
38 * of incrementing:
39 *
40 * 0 [x5] 0 [x4] 0 [x3] 0 [x2] 0 [x1] 0 [x0]
41 *
42 * To do so, we fill in the "holes" with 1's by adding the bitwise inverse of
43 * the mask of bits we care about
44 *
45 * 0 [x5] 0 [x4] 0 [x3] 0 [x2] 0 [x1] 0 [x0]
46 * + 1 0 1 0 1 0 1 0 1 0 1 0
47 * ------------------------------------------
48 * 1 [x5] 1 [x4] 1 [x3] 1 [x2] 1 [x1] 1 [x0]
49 *
50 * Then when we add one, the holes are passed over by forcing carry bits high.
51 * Finally, we need to zero out the holes, by ANDing with the mask of bits we
52 * care about. In total, we get the expression (X + ~mask + 1) & mask, and
53 * applying the two's complement identity, we are left with (X - mask) & mask
54 */
55
56 #define TILE_WIDTH 64
57 #define TILE_HEIGHT 64
58 #define TILE_SHIFT 6
59 #define TILE_MASK ((1 << TILE_SHIFT) - 1)
60
61 /* mask of bits used for X coordinate in a tile */
62 #define SPACE_MASK 0x555 // 0b010101010101
63
64 #define MAX2(x, y) (((x) > (y)) ? (x) : (y))
65 #define MIN2(x, y) (((x) < (y)) ? (x) : (y))
66
67 static uint32_t
agx_space_bits(unsigned x)68 agx_space_bits(unsigned x)
69 {
70 assert(x < TILE_WIDTH);
71 return ((x & 1) << 0) | ((x & 2) << 1) | ((x & 4) << 2) |
72 ((x & 8) << 3) | ((x & 16) << 4) | ((x & 32) << 5);
73 }
74
75 #define TILED_UNALIGNED_TYPE(pixel_t, is_store) { \
76 unsigned tiles_per_row = (width + TILE_WIDTH - 1) >> TILE_SHIFT;\
77 unsigned y_offs = agx_space_bits(sy & TILE_MASK) << 1;\
78 unsigned x_offs_start = agx_space_bits(sx & TILE_MASK);\
79 \
80 for (unsigned y = sy; y < smaxy; ++y) {\
81 unsigned tile_y = (y >> TILE_SHIFT);\
82 unsigned tile_row = tile_y * tiles_per_row;\
83 unsigned x_offs = x_offs_start;\
84 \
85 pixel_t *linear_row = linear;\
86 \
87 for (unsigned x = sx; x < smaxx; ++x) {\
88 unsigned tile_x = (x >> TILE_SHIFT);\
89 unsigned tile_idx = (tile_row + tile_x);\
90 unsigned tile_base = tile_idx * (TILE_WIDTH * TILE_HEIGHT);\
91 \
92 pixel_t *ptiled = &tiled[tile_base + y_offs + x_offs];\
93 pixel_t *plinear = (linear_row++);\
94 pixel_t *outp = (pixel_t *) (is_store ? ptiled : plinear); \
95 pixel_t *inp = (pixel_t *) (is_store ? plinear : ptiled); \
96 *outp = *inp;\
97 x_offs = (x_offs - SPACE_MASK) & SPACE_MASK;\
98 }\
99 \
100 y_offs = (((y_offs >> 1) - SPACE_MASK) & SPACE_MASK) << 1;\
101 linear += linear_pitch;\
102 }\
103 }
104
105 void
agx_detile(void * _tiled,void * _linear,unsigned width,unsigned bpp,unsigned linear_pitch,unsigned sx,unsigned sy,unsigned smaxx,unsigned smaxy)106 agx_detile(void *_tiled, void *_linear,
107 unsigned width, unsigned bpp, unsigned linear_pitch,
108 unsigned sx, unsigned sy, unsigned smaxx, unsigned smaxy)
109 {
110 /* TODO: parametrize with macro magic */
111 assert(bpp == 32);
112
113 uint32_t *linear = _linear;
114 uint32_t *tiled = _tiled;
115 TILED_UNALIGNED_TYPE(uint32_t, false);
116 }
117
118 void
agx_tile(void * _tiled,void * _linear,unsigned width,unsigned bpp,unsigned linear_pitch,unsigned sx,unsigned sy,unsigned smaxx,unsigned smaxy)119 agx_tile(void *_tiled, void *_linear,
120 unsigned width, unsigned bpp, unsigned linear_pitch,
121 unsigned sx, unsigned sy, unsigned smaxx, unsigned smaxy)
122 {
123 /* TODO: parametrize with macro magic */
124 assert(bpp == 32);
125
126 uint32_t *linear = _linear;
127 uint32_t *tiled = _tiled;
128 TILED_UNALIGNED_TYPE(uint32_t, true);
129 }
130