1/*
2 * Copyright (C) 2019 Collabora, Ltd.
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 "pan_device.h"
25#include "pan_bo.h"
26
27/* Sample positions are specified partially in hardware, partially in software
28 * on Mali. On Midgard, sample positions are completely fixed but need to be
29 * software accessible to implement gl_SamplePosition. On Bifrost, sample
30 * positions are part fixed, part programmable, and again need to be accessible
31 * for gl_SamplePosition. The mali_sample_position data structure is
32 * hardware-defined on Bifrost as a packed 8:8 fixed point format. The
33 * mali_sample_positions array is part-hardware, part-software-defined: it must
34 * give the sample position when indexed by the sample index, as well as
35 * reading back the origin when indexed by 32.
36 *
37 * The upshot is all of this is known at compile-time, so we can just hardcode
38 * the LUT, upload it to GPU memory on device initialization, and forget about
39 * it.
40 */
41
42struct mali_sample_position {
43        uint16_t x, y;
44} __attribute__((packed));
45
46struct mali_sample_positions {
47        struct mali_sample_position positions[32];
48        struct mali_sample_position origin;
49        struct mali_sample_position padding[64 - (32 + 1)];
50} __attribute__((packed));
51
52/* SAMPLE16 constructs a single sample in terms of 1/16's of the grid, centered
53 * at the origin. SAMPLE4/8 swap the units for legibility. */
54
55#define SAMPLE16(x, y) { \
56        (((x) + 8) * (256 / 16)), \
57        (((y) + 8) * (256 / 16)) \
58}
59
60#define SAMPLE8(x, y) SAMPLE16((x) * 2, (y) * 2)
61#define SAMPLE4(x, y) SAMPLE16((x) * 4, (y) * 4)
62
63const struct mali_sample_positions sample_position_lut[] = {
64        [MALI_SAMPLE_PATTERN_SINGLE_SAMPLED] = {
65                .positions = {
66                        SAMPLE4(0, 0)
67                },
68                .origin = SAMPLE4(0, 0)
69        },
70
71        [MALI_SAMPLE_PATTERN_ORDERED_4X_GRID] = {
72                .positions = {
73                        SAMPLE4(-1, -1),
74                        SAMPLE4( 1, -1),
75                        SAMPLE4(-1,  1),
76                        SAMPLE4( 1,  1),
77                },
78                .origin = SAMPLE4(0, 0)
79        },
80
81        [MALI_SAMPLE_PATTERN_ROTATED_4X_GRID] = {
82                .positions = {
83                        SAMPLE8(-1, -3),
84                        SAMPLE8( 3, -1),
85                        SAMPLE8(-3,  1),
86                        SAMPLE8( 1,  3),
87                },
88                .origin = SAMPLE8(0, 0)
89        },
90
91        [MALI_SAMPLE_PATTERN_D3D_8X_GRID] = {
92                .positions = {
93                        SAMPLE16( 1, -3),
94                        SAMPLE16(-1,  3),
95                        SAMPLE16( 5,  1),
96                        SAMPLE16(-3, -5),
97                        SAMPLE16(-5,  5),
98                        SAMPLE16(-7, -1),
99                        SAMPLE16( 3,  7),
100                        SAMPLE16( 7,  -7),
101                },
102                .origin = SAMPLE16(0, 0)
103        },
104
105        [MALI_SAMPLE_PATTERN_D3D_16X_GRID] = {
106                .positions = {
107                        SAMPLE16( 1,  1),
108                        SAMPLE16(-1, -3),
109                        SAMPLE16(-3,  2),
110                        SAMPLE16( 4, -1),
111                        SAMPLE16(-5, -2),
112                        SAMPLE16( 2,  5),
113                        SAMPLE16( 5,  3),
114                        SAMPLE16( 3, -5),
115                        SAMPLE16(-2,  6),
116                        SAMPLE16( 0,  7),
117                        SAMPLE16(-4, -6),
118                        SAMPLE16(-6,  4),
119                        SAMPLE16(-8,  0),
120                        SAMPLE16( 7, -4),
121                        SAMPLE16( 6,  7),
122                        SAMPLE16(-7, -8),
123
124                },
125                .origin = SAMPLE16(0, 0)
126        }
127};
128
129mali_ptr
130panfrost_sample_positions(const struct panfrost_device *dev,
131                enum mali_sample_pattern pattern)
132{
133        assert(pattern < ARRAY_SIZE(sample_position_lut));
134        unsigned offset = (pattern * sizeof(sample_position_lut[0]));
135        return dev->sample_positions->ptr.gpu + offset;
136}
137
138void
139panfrost_upload_sample_positions(struct panfrost_device *dev)
140{
141        STATIC_ASSERT(sizeof(sample_position_lut) < 4096);
142        dev->sample_positions = panfrost_bo_create(dev, 4096, 0, "Sample positions");
143
144        memcpy(dev->sample_positions->ptr.cpu, sample_position_lut,
145                        sizeof(sample_position_lut));
146}
147
148/* CPU side LUT query, to implement glGetMultisamplefv */
149
150void
151panfrost_query_sample_position(
152                enum mali_sample_pattern pattern,
153                unsigned sample_idx,
154                float *out)
155{
156        struct mali_sample_position pos =
157                sample_position_lut[pattern].positions[sample_idx];
158
159        out[0] = DECODE_FIXED_16(pos.x);
160        out[1] = DECODE_FIXED_16(pos.y);
161}
162