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
|
/*
* Copyright © 2015 Intel Corporation
*
* 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 (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 NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS 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.
*/
#include "anv_nir.h"
#include "nir/nir_builder.h"
static void
apply_dynamic_offsets_block(nir_block *block, nir_builder *b,
const struct anv_pipeline_layout *layout,
bool add_bounds_checks,
uint32_t indices_start)
{
struct anv_descriptor_set_layout *set_layout;
nir_foreach_instr_safe(instr, block) {
if (instr->type != nir_instr_type_intrinsic)
continue;
nir_intrinsic_instr *intrin = nir_instr_as_intrinsic(instr);
unsigned block_idx_src;
switch (intrin->intrinsic) {
case nir_intrinsic_load_ubo:
case nir_intrinsic_load_ssbo:
block_idx_src = 0;
break;
case nir_intrinsic_store_ssbo:
block_idx_src = 1;
break;
default:
continue; /* the loop */
}
nir_instr *res_instr = intrin->src[block_idx_src].ssa->parent_instr;
assert(res_instr->type == nir_instr_type_intrinsic);
nir_intrinsic_instr *res_intrin = nir_instr_as_intrinsic(res_instr);
assert(res_intrin->intrinsic == nir_intrinsic_vulkan_resource_index);
unsigned set = res_intrin->const_index[0];
unsigned binding = res_intrin->const_index[1];
set_layout = layout->set[set].layout;
if (set_layout->binding[binding].dynamic_offset_index < 0)
continue;
b->cursor = nir_before_instr(&intrin->instr);
/* First, we need to generate the uniform load for the buffer offset */
uint32_t index = layout->set[set].dynamic_offset_start +
set_layout->binding[binding].dynamic_offset_index;
uint32_t array_size = set_layout->binding[binding].array_size;
nir_intrinsic_instr *offset_load =
nir_intrinsic_instr_create(b->shader, nir_intrinsic_load_uniform);
offset_load->num_components = 2;
nir_intrinsic_set_base(offset_load, indices_start + index * 8);
nir_intrinsic_set_range(offset_load, array_size * 8);
offset_load->src[0] = nir_src_for_ssa(nir_imul(b, res_intrin->src[0].ssa,
nir_imm_int(b, 8)));
nir_ssa_dest_init(&offset_load->instr, &offset_load->dest, 2, 32, NULL);
nir_builder_instr_insert(b, &offset_load->instr);
nir_src *offset_src = nir_get_io_offset_src(intrin);
nir_ssa_def *old_offset = nir_ssa_for_src(b, *offset_src, 1);
nir_ssa_def *new_offset = nir_iadd(b, old_offset, &offset_load->dest.ssa);
nir_instr_rewrite_src(&intrin->instr, offset_src,
nir_src_for_ssa(new_offset));
if (!add_bounds_checks)
continue;
/* In order to avoid out-of-bounds access, we predicate */
nir_ssa_def *pred = nir_uge(b, nir_channel(b, &offset_load->dest.ssa, 1),
old_offset);
nir_if *if_stmt = nir_if_create(b->shader);
if_stmt->condition = nir_src_for_ssa(pred);
nir_cf_node_insert(b->cursor, &if_stmt->cf_node);
nir_instr_remove(&intrin->instr);
nir_instr_insert_after_cf_list(&if_stmt->then_list, &intrin->instr);
if (intrin->intrinsic != nir_intrinsic_store_ssbo) {
/* It's a load, we need a phi node */
nir_phi_instr *phi = nir_phi_instr_create(b->shader);
nir_ssa_dest_init(&phi->instr, &phi->dest,
intrin->num_components,
intrin->dest.ssa.bit_size, NULL);
nir_phi_src *src1 = ralloc(phi, nir_phi_src);
struct exec_node *tnode = exec_list_get_tail(&if_stmt->then_list);
src1->pred = exec_node_data(nir_block, tnode, cf_node.node);
src1->src = nir_src_for_ssa(&intrin->dest.ssa);
exec_list_push_tail(&phi->srcs, &src1->node);
b->cursor = nir_after_cf_list(&if_stmt->else_list);
nir_const_value zero_val = { .u32 = { 0, 0, 0, 0 } };
nir_ssa_def *zero = nir_build_imm(b, intrin->num_components,
intrin->dest.ssa.bit_size, zero_val);
nir_phi_src *src2 = ralloc(phi, nir_phi_src);
struct exec_node *enode = exec_list_get_tail(&if_stmt->else_list);
src2->pred = exec_node_data(nir_block, enode, cf_node.node);
src2->src = nir_src_for_ssa(zero);
exec_list_push_tail(&phi->srcs, &src2->node);
assert(intrin->dest.is_ssa);
nir_ssa_def_rewrite_uses(&intrin->dest.ssa,
nir_src_for_ssa(&phi->dest.ssa));
nir_instr_insert_after_cf(&if_stmt->cf_node, &phi->instr);
}
}
}
void
anv_nir_apply_dynamic_offsets(struct anv_pipeline *pipeline,
nir_shader *shader,
struct brw_stage_prog_data *prog_data)
{
const struct anv_pipeline_layout *layout = pipeline->layout;
if (!layout || !layout->stage[shader->stage].has_dynamic_offsets)
return;
const bool add_bounds_checks = pipeline->device->robust_buffer_access;
nir_foreach_function(function, shader) {
if (!function->impl)
continue;
nir_builder builder;
nir_builder_init(&builder, function->impl);
nir_foreach_block(block, function->impl) {
apply_dynamic_offsets_block(block, &builder, pipeline->layout,
add_bounds_checks, shader->num_uniforms);
}
nir_metadata_preserve(function->impl, nir_metadata_block_index |
nir_metadata_dominance);
}
struct anv_push_constants *null_data = NULL;
for (unsigned i = 0; i < MAX_DYNAMIC_BUFFERS; i++) {
prog_data->param[i * 2 + shader->num_uniforms / 4] =
(const union gl_constant_value *)&null_data->dynamic[i].offset;
prog_data->param[i * 2 + 1 + shader->num_uniforms / 4] =
(const union gl_constant_value *)&null_data->dynamic[i].range;
}
shader->num_uniforms += MAX_DYNAMIC_BUFFERS * 8;
}
|
