1 /*
2 * Copyright (C) 2022 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
25 #include "pan_ir.h"
26 #include "compiler/nir/nir_builder.h"
27
28 static void
lower_xfb_output(nir_builder * b,nir_intrinsic_instr * intr,unsigned start_component,unsigned num_components,unsigned buffer,unsigned offset_words)29 lower_xfb_output(nir_builder *b, nir_intrinsic_instr *intr,
30 unsigned start_component, unsigned num_components,
31 unsigned buffer, unsigned offset_words)
32 {
33 assert(buffer < MAX_XFB_BUFFERS);
34 assert(nir_intrinsic_component(intr) == 0); // TODO
35
36 /* Transform feedback info in units of words, convert to bytes. */
37 uint16_t stride = b->shader->info.xfb_stride[buffer] * 4;
38 assert(stride != 0);
39
40 uint16_t offset = offset_words * 4;
41
42 nir_ssa_def *index = nir_iadd(b,
43 nir_imul(b, nir_load_instance_id(b),
44 nir_load_num_vertices(b)),
45 nir_load_vertex_id_zero_base(b));
46
47 BITSET_SET(b->shader->info.system_values_read, SYSTEM_VALUE_VERTEX_ID_ZERO_BASE);
48 BITSET_SET(b->shader->info.system_values_read, SYSTEM_VALUE_INSTANCE_ID);
49
50 nir_ssa_def *buf = nir_load_xfb_address(b, 64, .base = buffer);
51 nir_ssa_def *addr =
52 nir_iadd(b, buf, nir_u2u64(b,
53 nir_iadd_imm(b,
54 nir_imul_imm(b, index, stride),
55 offset)));
56
57 assert(intr->src[0].is_ssa && "must lower XFB before lowering SSA");
58 nir_ssa_def *src = intr->src[0].ssa;
59 nir_ssa_def *value = nir_channels(b, src, BITFIELD_MASK(num_components) << start_component);
60 nir_store_global(b, addr, 4, value, BITFIELD_MASK(num_components));
61 }
62
63 static bool
lower_xfb(nir_builder * b,nir_instr * instr,UNUSED void * data)64 lower_xfb(nir_builder *b, nir_instr *instr, UNUSED void *data)
65 {
66 if (instr->type != nir_instr_type_intrinsic)
67 return false;
68
69 nir_intrinsic_instr *intr = nir_instr_as_intrinsic(instr);
70 if (intr->intrinsic != nir_intrinsic_store_output)
71 return false;
72
73 bool progress = false;
74
75 b->cursor = nir_before_instr(&intr->instr);
76
77 for (unsigned i = 0; i < 2; ++i) {
78 nir_io_xfb xfb = i ? nir_intrinsic_io_xfb2(intr) : nir_intrinsic_io_xfb(intr);
79 for (unsigned j = 0; j < 2; ++j) {
80 if (!xfb.out[j].num_components) continue;
81
82 lower_xfb_output(b, intr, i*2 + j,
83 xfb.out[j].num_components,
84 xfb.out[j].buffer,
85 xfb.out[j].offset);
86 progress = true;
87 }
88 }
89
90 nir_instr_remove(instr);
91 return progress;
92 }
93
94 bool
pan_lower_xfb(nir_shader * nir)95 pan_lower_xfb(nir_shader *nir)
96 {
97 return nir_shader_instructions_pass(nir, lower_xfb,
98 nir_metadata_block_index |
99 nir_metadata_dominance, NULL);
100 }
101
102