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
|
//===- AMDGPU.cpp ---------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#include "InputFiles.h"
#include "Symbols.h"
#include "Target.h"
#include "lld/Common/ErrorHandler.h"
#include "llvm/Object/ELF.h"
#include "llvm/Support/Endian.h"
using namespace llvm;
using namespace llvm::object;
using namespace llvm::support::endian;
using namespace llvm::ELF;
using namespace lld;
using namespace lld::elf;
namespace {
class AMDGPU final : public TargetInfo {
private:
uint32_t calcEFlagsV3() const;
uint32_t calcEFlagsV4() const;
public:
AMDGPU();
uint32_t calcEFlags() const override;
void relocate(uint8_t *loc, const Relocation &rel,
uint64_t val) const override;
RelExpr getRelExpr(RelType type, const Symbol &s,
const uint8_t *loc) const override;
RelType getDynRel(RelType type) const override;
};
} // namespace
AMDGPU::AMDGPU() {
relativeRel = R_AMDGPU_RELATIVE64;
gotRel = R_AMDGPU_ABS64;
noneRel = R_AMDGPU_NONE;
symbolicRel = R_AMDGPU_ABS64;
}
static uint32_t getEFlags(InputFile *file) {
return cast<ObjFile<ELF64LE>>(file)->getObj().getHeader().e_flags;
}
uint32_t AMDGPU::calcEFlagsV3() const {
uint32_t ret = getEFlags(objectFiles[0]);
// Verify that all input files have the same e_flags.
for (InputFile *f : makeArrayRef(objectFiles).slice(1)) {
if (ret == getEFlags(f))
continue;
error("incompatible e_flags: " + toString(f));
return 0;
}
return ret;
}
uint32_t AMDGPU::calcEFlagsV4() const {
uint32_t retMach = getEFlags(objectFiles[0]) & EF_AMDGPU_MACH;
uint32_t retXnack = getEFlags(objectFiles[0]) & EF_AMDGPU_FEATURE_XNACK_V4;
uint32_t retSramEcc =
getEFlags(objectFiles[0]) & EF_AMDGPU_FEATURE_SRAMECC_V4;
// Verify that all input files have compatible e_flags (same mach, all
// features in the same category are either ANY, ANY and ON, or ANY and OFF).
for (InputFile *f : makeArrayRef(objectFiles).slice(1)) {
if (retMach != (getEFlags(f) & EF_AMDGPU_MACH)) {
error("incompatible mach: " + toString(f));
return 0;
}
if (retXnack == EF_AMDGPU_FEATURE_XNACK_UNSUPPORTED_V4 ||
(retXnack != EF_AMDGPU_FEATURE_XNACK_ANY_V4 &&
(getEFlags(f) & EF_AMDGPU_FEATURE_XNACK_V4)
!= EF_AMDGPU_FEATURE_XNACK_ANY_V4)) {
if (retXnack != (getEFlags(f) & EF_AMDGPU_FEATURE_XNACK_V4)) {
error("incompatible xnack: " + toString(f));
return 0;
}
} else {
if (retXnack == EF_AMDGPU_FEATURE_XNACK_ANY_V4)
retXnack = getEFlags(f) & EF_AMDGPU_FEATURE_XNACK_V4;
}
if (retSramEcc == EF_AMDGPU_FEATURE_SRAMECC_UNSUPPORTED_V4 ||
(retSramEcc != EF_AMDGPU_FEATURE_SRAMECC_ANY_V4 &&
(getEFlags(f) & EF_AMDGPU_FEATURE_SRAMECC_V4) !=
EF_AMDGPU_FEATURE_SRAMECC_ANY_V4)) {
if (retSramEcc != (getEFlags(f) & EF_AMDGPU_FEATURE_SRAMECC_V4)) {
error("incompatible sramecc: " + toString(f));
return 0;
}
} else {
if (retSramEcc == EF_AMDGPU_FEATURE_SRAMECC_ANY_V4)
retSramEcc = getEFlags(f) & EF_AMDGPU_FEATURE_SRAMECC_V4;
}
}
return retMach | retXnack | retSramEcc;
}
uint32_t AMDGPU::calcEFlags() const {
assert(!objectFiles.empty());
uint8_t abiVersion = cast<ObjFile<ELF64LE>>(objectFiles[0])->getObj()
.getHeader().e_ident[EI_ABIVERSION];
switch (abiVersion) {
case ELFABIVERSION_AMDGPU_HSA_V2:
case ELFABIVERSION_AMDGPU_HSA_V3:
return calcEFlagsV3();
case ELFABIVERSION_AMDGPU_HSA_V4:
return calcEFlagsV4();
default:
error("unknown abi version: " + Twine(abiVersion));
return 0;
}
}
void AMDGPU::relocate(uint8_t *loc, const Relocation &rel, uint64_t val) const {
switch (rel.type) {
case R_AMDGPU_ABS32:
case R_AMDGPU_GOTPCREL:
case R_AMDGPU_GOTPCREL32_LO:
case R_AMDGPU_REL32:
case R_AMDGPU_REL32_LO:
write32le(loc, val);
break;
case R_AMDGPU_ABS64:
case R_AMDGPU_REL64:
write64le(loc, val);
break;
case R_AMDGPU_GOTPCREL32_HI:
case R_AMDGPU_REL32_HI:
write32le(loc, val >> 32);
break;
case R_AMDGPU_REL16: {
int64_t simm = (static_cast<int64_t>(val) - 4) / 4;
checkInt(loc, simm, 16, rel);
write16le(loc, simm);
break;
}
default:
llvm_unreachable("unknown relocation");
}
}
RelExpr AMDGPU::getRelExpr(RelType type, const Symbol &s,
const uint8_t *loc) const {
switch (type) {
case R_AMDGPU_ABS32:
case R_AMDGPU_ABS64:
return R_ABS;
case R_AMDGPU_REL32:
case R_AMDGPU_REL32_LO:
case R_AMDGPU_REL32_HI:
case R_AMDGPU_REL64:
case R_AMDGPU_REL16:
return R_PC;
case R_AMDGPU_GOTPCREL:
case R_AMDGPU_GOTPCREL32_LO:
case R_AMDGPU_GOTPCREL32_HI:
return R_GOT_PC;
default:
error(getErrorLocation(loc) + "unknown relocation (" + Twine(type) +
") against symbol " + toString(s));
return R_NONE;
}
}
RelType AMDGPU::getDynRel(RelType type) const {
if (type == R_AMDGPU_ABS64)
return type;
return R_AMDGPU_NONE;
}
TargetInfo *elf::getAMDGPUTargetInfo() {
static AMDGPU target;
return ⌖
}
|
