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
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
|
//===-- ARMAsmBackend.cpp - ARM Assembler Backend -------------------------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
#include "ARM.h"
#include "ARMAddressingModes.h"
#include "ARMFixupKinds.h"
#include "llvm/ADT/Twine.h"
#include "llvm/MC/MCAssembler.h"
#include "llvm/MC/MCDirectives.h"
#include "llvm/MC/MCExpr.h"
#include "llvm/MC/MCObjectFormat.h"
#include "llvm/MC/MCObjectWriter.h"
#include "llvm/MC/MCSectionELF.h"
#include "llvm/MC/MCSectionMachO.h"
#include "llvm/Object/MachOFormat.h"
#include "llvm/Support/ELF.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Target/TargetAsmBackend.h"
#include "llvm/Target/TargetRegistry.h"
using namespace llvm;
namespace {
class ARMAsmBackend : public TargetAsmBackend {
bool isThumbMode; // Currently emitting Thumb code.
public:
ARMAsmBackend(const Target &T) : TargetAsmBackend(), isThumbMode(false) {}
bool MayNeedRelaxation(const MCInst &Inst) const;
void RelaxInstruction(const MCInst &Inst, MCInst &Res) const;
bool WriteNopData(uint64_t Count, MCObjectWriter *OW) const;
void HandleAssemblerFlag(MCAssemblerFlag Flag) {
switch (Flag) {
default: break;
case MCAF_Code16:
setIsThumb(true);
break;
case MCAF_Code32:
setIsThumb(false);
break;
}
}
unsigned getPointerSize() const { return 4; }
bool isThumb() const { return isThumbMode; }
void setIsThumb(bool it) { isThumbMode = it; }
};
} // end anonymous namespace
bool ARMAsmBackend::MayNeedRelaxation(const MCInst &Inst) const {
// FIXME: Thumb targets, different move constant targets..
return false;
}
void ARMAsmBackend::RelaxInstruction(const MCInst &Inst, MCInst &Res) const {
assert(0 && "ARMAsmBackend::RelaxInstruction() unimplemented");
return;
}
bool ARMAsmBackend::WriteNopData(uint64_t Count, MCObjectWriter *OW) const {
if (isThumb()) {
assert (((Count & 1) == 0) && "Unaligned Nop data fragment!");
// FIXME: 0xbf00 is the ARMv7 value. For v6 and before, we'll need to
// use 0x46c0 (which is a 'mov r8, r8' insn).
Count /= 2;
for (uint64_t i = 0; i != Count; ++i)
OW->Write16(0xbf00);
return true;
}
// ARM mode
Count /= 4;
for (uint64_t i = 0; i != Count; ++i)
OW->Write32(0xe1a00000);
return true;
}
static unsigned adjustFixupValue(unsigned Kind, uint64_t Value) {
switch (Kind) {
default:
llvm_unreachable("Unknown fixup kind!");
case FK_Data_4:
return Value;
case ARM::fixup_arm_movt_hi16:
case ARM::fixup_arm_movw_lo16: {
unsigned Hi4 = (Value & 0xF000) >> 12;
unsigned Lo12 = Value & 0x0FFF;
// inst{19-16} = Hi4;
// inst{11-0} = Lo12;
Value = (Hi4 << 16) | (Lo12);
return Value;
}
case ARM::fixup_arm_ldst_pcrel_12:
// ARM PC-relative values are offset by 8.
Value -= 4;
// FALLTHROUGH
case ARM::fixup_t2_ldst_pcrel_12: {
// Offset by 4, adjusted by two due to the half-word ordering of thumb.
Value -= 4;
bool isAdd = true;
if ((int64_t)Value < 0) {
Value = -Value;
isAdd = false;
}
assert ((Value < 4096) && "Out of range pc-relative fixup value!");
Value |= isAdd << 23;
// Same addressing mode as fixup_arm_pcrel_10,
// but with 16-bit halfwords swapped.
if (Kind == ARM::fixup_t2_ldst_pcrel_12) {
uint64_t swapped = (Value & 0xFFFF0000) >> 16;
swapped |= (Value & 0x0000FFFF) << 16;
return swapped;
}
return Value;
}
case ARM::fixup_arm_adr_pcrel_12: {
// ARM PC-relative values are offset by 8.
Value -= 8;
unsigned opc = 4; // bits {24-21}. Default to add: 0b0100
if ((int64_t)Value < 0) {
Value = -Value;
opc = 2; // 0b0010
}
assert(ARM_AM::getSOImmVal(Value) != -1 &&
"Out of range pc-relative fixup value!");
// Encode the immediate and shift the opcode into place.
return ARM_AM::getSOImmVal(Value) | (opc << 21);
}
case ARM::fixup_t2_adr_pcrel_12: {
Value -= 4;
unsigned opc = 0;
if ((int64_t)Value < 0) {
Value = -Value;
opc = 5;
}
uint32_t out = (opc << 21);
out |= (Value & 0x800) << 14;
out |= (Value & 0x700) << 4;
out |= (Value & 0x0FF);
uint64_t swapped = (out & 0xFFFF0000) >> 16;
swapped |= (out & 0x0000FFFF) << 16;
return swapped;
}
case ARM::fixup_arm_branch:
// These values don't encode the low two bits since they're always zero.
// Offset by 8 just as above.
return 0xffffff & ((Value - 8) >> 2);
case ARM::fixup_t2_uncondbranch: {
Value = Value - 4;
Value >>= 1; // Low bit is not encoded.
uint32_t out = 0;
bool I = Value & 0x800000;
bool J1 = Value & 0x400000;
bool J2 = Value & 0x200000;
J1 ^= I;
J2 ^= I;
out |= I << 26; // S bit
out |= !J1 << 13; // J1 bit
out |= !J2 << 11; // J2 bit
out |= (Value & 0x1FF800) << 5; // imm6 field
out |= (Value & 0x0007FF); // imm11 field
uint64_t swapped = (out & 0xFFFF0000) >> 16;
swapped |= (out & 0x0000FFFF) << 16;
return swapped;
}
case ARM::fixup_t2_condbranch: {
Value = Value - 4;
Value >>= 1; // Low bit is not encoded.
uint64_t out = 0;
out |= (Value & 0x80000) << 7; // S bit
out |= (Value & 0x40000) >> 7; // J2 bit
out |= (Value & 0x20000) >> 4; // J1 bit
out |= (Value & 0x1F800) << 5; // imm6 field
out |= (Value & 0x007FF); // imm11 field
uint32_t swapped = (out & 0xFFFF0000) >> 16;
swapped |= (out & 0x0000FFFF) << 16;
return swapped;
}
case ARM::fixup_arm_thumb_bl: {
// The value doesn't encode the low bit (always zero) and is offset by
// four. The value is encoded into disjoint bit positions in the destination
// opcode. x = unchanged, I = immediate value bit, S = sign extension bit
//
// BL: xxxxxSIIIIIIIIII xxxxxIIIIIIIIIII
//
// Note that the halfwords are stored high first, low second; so we need
// to transpose the fixup value here to map properly.
unsigned isNeg = (int64_t(Value) < 0) ? 1 : 0;
uint32_t Binary = 0;
Value = 0x3fffff & ((Value - 4) >> 1);
Binary = (Value & 0x7ff) << 16; // Low imm11 value.
Binary |= (Value & 0x1ffc00) >> 11; // High imm10 value.
Binary |= isNeg << 10; // Sign bit.
return Binary;
}
case ARM::fixup_arm_thumb_blx: {
// The value doesn't encode the low two bits (always zero) and is offset by
// four (see fixup_arm_thumb_cp). The value is encoded into disjoint bit
// positions in the destination opcode. x = unchanged, I = immediate value
// bit, S = sign extension bit, 0 = zero.
//
// BLX: xxxxxSIIIIIIIIII xxxxxIIIIIIIIII0
//
// Note that the halfwords are stored high first, low second; so we need
// to transpose the fixup value here to map properly.
unsigned isNeg = (int64_t(Value) < 0) ? 1 : 0;
uint32_t Binary = 0;
Value = 0xfffff & ((Value - 2) >> 2);
Binary = (Value & 0x3ff) << 17; // Low imm10L value.
Binary |= (Value & 0xffc00) >> 10; // High imm10H value.
Binary |= isNeg << 10; // Sign bit.
return Binary;
}
case ARM::fixup_arm_thumb_cp:
// Offset by 4, and don't encode the low two bits. Two bytes of that
// 'off by 4' is implicitly handled by the half-word ordering of the
// Thumb encoding, so we only need to adjust by 2 here.
return ((Value - 2) >> 2) & 0xff;
case ARM::fixup_arm_thumb_cb: {
// Offset by 4 and don't encode the lower bit, which is always 0.
uint32_t Binary = (Value - 4) >> 1;
return ((Binary & 0x20) << 9) | ((Binary & 0x1f) << 3);
}
case ARM::fixup_arm_thumb_br:
// Offset by 4 and don't encode the lower bit, which is always 0.
return ((Value - 4) >> 1) & 0x7ff;
case ARM::fixup_arm_thumb_bcc:
// Offset by 4 and don't encode the lower bit, which is always 0.
return ((Value - 4) >> 1) & 0xff;
case ARM::fixup_arm_pcrel_10:
Value = Value - 4; // ARM fixups offset by an additional word and don't
// need to adjust for the half-word ordering.
// Fall through.
case ARM::fixup_t2_pcrel_10: {
// Offset by 4, adjusted by two due to the half-word ordering of thumb.
Value = Value - 4;
bool isAdd = true;
if ((int64_t)Value < 0) {
Value = -Value;
isAdd = false;
}
// These values don't encode the low two bits since they're always zero.
Value >>= 2;
assert ((Value < 256) && "Out of range pc-relative fixup value!");
Value |= isAdd << 23;
// Same addressing mode as fixup_arm_pcrel_10,
// but with 16-bit halfwords swapped.
if (Kind == ARM::fixup_t2_pcrel_10) {
uint32_t swapped = (Value & 0xFFFF0000) >> 16;
swapped |= (Value & 0x0000FFFF) << 16;
return swapped;
}
return Value;
}
}
}
namespace {
// FIXME: This should be in a separate file.
// ELF is an ELF of course...
class ELFARMAsmBackend : public ARMAsmBackend {
MCELFObjectFormat Format;
public:
Triple::OSType OSType;
ELFARMAsmBackend(const Target &T, Triple::OSType _OSType)
: ARMAsmBackend(T), OSType(_OSType) {
HasScatteredSymbols = true;
}
virtual const MCObjectFormat &getObjectFormat() const {
return Format;
}
void ApplyFixup(const MCFixup &Fixup, char *Data, unsigned DataSize,
uint64_t Value) const;
MCObjectWriter *createObjectWriter(raw_ostream &OS) const {
return createELFObjectWriter(OS, /*Is64Bit=*/false,
OSType, ELF::EM_ARM,
/*IsLittleEndian=*/true,
/*HasRelocationAddend=*/false);
}
};
// FIXME: Raise this to share code between Darwin and ELF.
void ELFARMAsmBackend::ApplyFixup(const MCFixup &Fixup, char *Data,
unsigned DataSize, uint64_t Value) const {
unsigned NumBytes = 4; // FIXME: 2 for Thumb
Value = adjustFixupValue(Fixup.getKind(), Value);
if (!Value) return; // Doesn't change encoding.
unsigned Offset = Fixup.getOffset();
assert(Offset % NumBytes == 0 && "Offset mod NumBytes is nonzero!");
// For each byte of the fragment that the fixup touches, mask in the bits from
// the fixup value. The Value has been "split up" into the appropriate
// bitfields above.
for (unsigned i = 0; i != NumBytes; ++i)
Data[Offset + i] |= uint8_t((Value >> (i * 8)) & 0xff);
}
// FIXME: This should be in a separate file.
class DarwinARMAsmBackend : public ARMAsmBackend {
MCMachOObjectFormat Format;
public:
DarwinARMAsmBackend(const Target &T) : ARMAsmBackend(T) {
HasScatteredSymbols = true;
}
virtual const MCObjectFormat &getObjectFormat() const {
return Format;
}
void ApplyFixup(const MCFixup &Fixup, char *Data, unsigned DataSize,
uint64_t Value) const;
MCObjectWriter *createObjectWriter(raw_ostream &OS) const {
// FIXME: Subtarget info should be derived. Force v7 for now.
return createMachObjectWriter(OS, /*Is64Bit=*/false,
object::mach::CTM_ARM,
object::mach::CSARM_V7,
/*IsLittleEndian=*/true);
}
virtual bool doesSectionRequireSymbols(const MCSection &Section) const {
return false;
}
};
/// getFixupKindNumBytes - The number of bytes the fixup may change.
static unsigned getFixupKindNumBytes(unsigned Kind) {
switch (Kind) {
default:
llvm_unreachable("Unknown fixup kind!");
case ARM::fixup_arm_thumb_bcc:
case ARM::fixup_arm_thumb_cp:
return 1;
case ARM::fixup_arm_thumb_br:
case ARM::fixup_arm_thumb_cb:
return 2;
case ARM::fixup_arm_ldst_pcrel_12:
case ARM::fixup_arm_pcrel_10:
case ARM::fixup_arm_adr_pcrel_12:
case ARM::fixup_arm_branch:
return 3;
case FK_Data_4:
case ARM::fixup_t2_ldst_pcrel_12:
case ARM::fixup_t2_condbranch:
case ARM::fixup_t2_uncondbranch:
case ARM::fixup_t2_pcrel_10:
case ARM::fixup_t2_adr_pcrel_12:
case ARM::fixup_arm_thumb_bl:
case ARM::fixup_arm_thumb_blx:
return 4;
}
}
void DarwinARMAsmBackend::ApplyFixup(const MCFixup &Fixup, char *Data,
unsigned DataSize, uint64_t Value) const {
unsigned NumBytes = getFixupKindNumBytes(Fixup.getKind());
Value = adjustFixupValue(Fixup.getKind(), Value);
if (!Value) return; // Doesn't change encoding.
unsigned Offset = Fixup.getOffset();
assert(Offset + NumBytes <= DataSize && "Invalid fixup offset!");
// For each byte of the fragment that the fixup touches, mask in the
// bits from the fixup value.
for (unsigned i = 0; i != NumBytes; ++i)
Data[Offset + i] |= uint8_t((Value >> (i * 8)) & 0xff);
}
} // end anonymous namespace
TargetAsmBackend *llvm::createARMAsmBackend(const Target &T,
const std::string &TT) {
switch (Triple(TT).getOS()) {
case Triple::Darwin:
return new DarwinARMAsmBackend(T);
case Triple::MinGW32:
case Triple::Cygwin:
case Triple::Win32:
assert(0 && "Windows not supported on ARM");
default:
return new ELFARMAsmBackend(T, Triple(TT).getOS());
}
}
|