blob: 689aabc333df71f4ac6152826612686ec7055b92 [file] [log] [blame]
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001//===- lib/MC/MachObjectWriter.cpp - Mach-O File Writer -------------------===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9
10#include "llvm/MC/MachObjectWriter.h"
11#include "llvm/ADT/StringMap.h"
12#include "llvm/ADT/Twine.h"
13#include "llvm/MC/MCAssembler.h"
Daniel Dunbar207e06e2010-03-24 03:43:40 +000014#include "llvm/MC/MCAsmLayout.h"
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +000015#include "llvm/MC/MCExpr.h"
16#include "llvm/MC/MCObjectWriter.h"
17#include "llvm/MC/MCSectionMachO.h"
18#include "llvm/MC/MCSymbol.h"
Kevin Enderbya6eeb6e2010-05-07 21:44:23 +000019#include "llvm/MC/MCMachOSymbolFlags.h"
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +000020#include "llvm/MC/MCValue.h"
21#include "llvm/Support/ErrorHandling.h"
22#include "llvm/Support/MachO.h"
23#include "llvm/Target/TargetAsmBackend.h"
24
25// FIXME: Gross.
26#include "../Target/X86/X86FixupKinds.h"
27
28#include <vector>
29using namespace llvm;
30
Rafael Espindolaa8c02c32010-09-30 03:11:42 +000031// FIXME: this has been copied from (or to) X86AsmBackend.cpp
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +000032static unsigned getFixupKindLog2Size(unsigned Kind) {
33 switch (Kind) {
34 default: llvm_unreachable("invalid fixup kind!");
35 case X86::reloc_pcrel_1byte:
36 case FK_Data_1: return 0;
Chris Lattnerda3051a2010-07-07 22:35:13 +000037 case X86::reloc_pcrel_2byte:
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +000038 case FK_Data_2: return 1;
39 case X86::reloc_pcrel_4byte:
40 case X86::reloc_riprel_4byte:
Daniel Dunbar602b40f2010-03-19 18:07:55 +000041 case X86::reloc_riprel_4byte_movq_load:
Rafael Espindolaa8c02c32010-09-30 03:11:42 +000042 case X86::reloc_signed_4byte:
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +000043 case FK_Data_4: return 2;
44 case FK_Data_8: return 3;
45 }
46}
47
48static bool isFixupKindPCRel(unsigned Kind) {
49 switch (Kind) {
50 default:
51 return false;
52 case X86::reloc_pcrel_1byte:
Chris Lattnerda3051a2010-07-07 22:35:13 +000053 case X86::reloc_pcrel_2byte:
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +000054 case X86::reloc_pcrel_4byte:
55 case X86::reloc_riprel_4byte:
Daniel Dunbar602b40f2010-03-19 18:07:55 +000056 case X86::reloc_riprel_4byte_movq_load:
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +000057 return true;
58 }
59}
60
Daniel Dunbar602b40f2010-03-19 18:07:55 +000061static bool isFixupKindRIPRel(unsigned Kind) {
62 return Kind == X86::reloc_riprel_4byte ||
63 Kind == X86::reloc_riprel_4byte_movq_load;
64}
65
Daniel Dunbare9460ec2010-05-10 23:15:13 +000066static bool doesSymbolRequireExternRelocation(MCSymbolData *SD) {
67 // Undefined symbols are always extern.
68 if (SD->Symbol->isUndefined())
69 return true;
70
71 // References to weak definitions require external relocation entries; the
72 // definition may not always be the one in the same object file.
73 if (SD->getFlags() & SF_WeakDefinition)
74 return true;
75
76 // Otherwise, we can use an internal relocation.
77 return false;
78}
79
Rafael Espindola70703872010-09-30 02:22:20 +000080static bool isScatteredFixupFullyResolved(const MCAssembler &Asm,
81 const MCValue Target,
82 const MCSymbolData *BaseSymbol) {
83 // The effective fixup address is
84 // addr(atom(A)) + offset(A)
85 // - addr(atom(B)) - offset(B)
86 // - addr(BaseSymbol) + <fixup offset from base symbol>
87 // and the offsets are not relocatable, so the fixup is fully resolved when
88 // addr(atom(A)) - addr(atom(B)) - addr(BaseSymbol) == 0.
89 //
90 // Note that "false" is almost always conservatively correct (it means we emit
91 // a relocation which is unnecessary), except when it would force us to emit a
92 // relocation which the target cannot encode.
93
94 const MCSymbolData *A_Base = 0, *B_Base = 0;
95 if (const MCSymbolRefExpr *A = Target.getSymA()) {
96 // Modified symbol references cannot be resolved.
97 if (A->getKind() != MCSymbolRefExpr::VK_None)
98 return false;
99
100 A_Base = Asm.getAtom(&Asm.getSymbolData(A->getSymbol()));
101 if (!A_Base)
102 return false;
103 }
104
105 if (const MCSymbolRefExpr *B = Target.getSymB()) {
106 // Modified symbol references cannot be resolved.
107 if (B->getKind() != MCSymbolRefExpr::VK_None)
108 return false;
109
110 B_Base = Asm.getAtom(&Asm.getSymbolData(B->getSymbol()));
111 if (!B_Base)
112 return false;
113 }
114
115 // If there is no base, A and B have to be the same atom for this fixup to be
116 // fully resolved.
117 if (!BaseSymbol)
118 return A_Base == B_Base;
119
120 // Otherwise, B must be missing and A must be the base.
121 return !B_Base && BaseSymbol == A_Base;
122}
123
124static bool isScatteredFixupFullyResolvedSimple(const MCAssembler &Asm,
125 const MCValue Target,
126 const MCSection *BaseSection) {
127 // The effective fixup address is
128 // addr(atom(A)) + offset(A)
129 // - addr(atom(B)) - offset(B)
130 // - addr(<base symbol>) + <fixup offset from base symbol>
131 // and the offsets are not relocatable, so the fixup is fully resolved when
132 // addr(atom(A)) - addr(atom(B)) - addr(<base symbol>)) == 0.
133 //
134 // The simple (Darwin, except on x86_64) way of dealing with this was to
135 // assume that any reference to a temporary symbol *must* be a temporary
136 // symbol in the same atom, unless the sections differ. Therefore, any PCrel
137 // relocation to a temporary symbol (in the same section) is fully
138 // resolved. This also works in conjunction with absolutized .set, which
139 // requires the compiler to use .set to absolutize the differences between
140 // symbols which the compiler knows to be assembly time constants, so we don't
141 // need to worry about considering symbol differences fully resolved.
142
143 // Non-relative fixups are only resolved if constant.
144 if (!BaseSection)
145 return Target.isAbsolute();
146
147 // Otherwise, relative fixups are only resolved if not a difference and the
148 // target is a temporary in the same section.
149 if (Target.isAbsolute() || Target.getSymB())
150 return false;
151
152 const MCSymbol *A = &Target.getSymA()->getSymbol();
153 if (!A->isTemporary() || !A->isInSection() ||
154 &A->getSection() != BaseSection)
155 return false;
156
157 return true;
158}
159
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000160namespace {
161
162class MachObjectWriterImpl {
163 // See <mach-o/loader.h>.
164 enum {
165 Header_Magic32 = 0xFEEDFACE,
166 Header_Magic64 = 0xFEEDFACF
167 };
168
169 enum {
170 Header32Size = 28,
171 Header64Size = 32,
172 SegmentLoadCommand32Size = 56,
173 SegmentLoadCommand64Size = 72,
174 Section32Size = 68,
175 Section64Size = 80,
176 SymtabLoadCommandSize = 24,
177 DysymtabLoadCommandSize = 80,
178 Nlist32Size = 12,
179 Nlist64Size = 16,
180 RelocationInfoSize = 8
181 };
182
183 enum HeaderFileType {
184 HFT_Object = 0x1
185 };
186
187 enum HeaderFlags {
188 HF_SubsectionsViaSymbols = 0x2000
189 };
190
191 enum LoadCommandType {
192 LCT_Segment = 0x1,
193 LCT_Symtab = 0x2,
194 LCT_Dysymtab = 0xb,
195 LCT_Segment64 = 0x19
196 };
197
198 // See <mach-o/nlist.h>.
199 enum SymbolTypeType {
200 STT_Undefined = 0x00,
201 STT_Absolute = 0x02,
202 STT_Section = 0x0e
203 };
204
205 enum SymbolTypeFlags {
206 // If any of these bits are set, then the entry is a stab entry number (see
207 // <mach-o/stab.h>. Otherwise the other masks apply.
208 STF_StabsEntryMask = 0xe0,
209
210 STF_TypeMask = 0x0e,
211 STF_External = 0x01,
212 STF_PrivateExtern = 0x10
213 };
214
215 /// IndirectSymbolFlags - Flags for encoding special values in the indirect
216 /// symbol entry.
217 enum IndirectSymbolFlags {
218 ISF_Local = 0x80000000,
219 ISF_Absolute = 0x40000000
220 };
221
222 /// RelocationFlags - Special flags for addresses.
223 enum RelocationFlags {
224 RF_Scattered = 0x80000000
225 };
226
227 enum RelocationInfoType {
228 RIT_Vanilla = 0,
229 RIT_Pair = 1,
230 RIT_Difference = 2,
231 RIT_PreboundLazyPointer = 3,
Eric Christopher96ac5152010-05-26 00:02:12 +0000232 RIT_LocalDifference = 4,
233 RIT_TLV = 5
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000234 };
235
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000236 /// X86_64 uses its own relocation types.
237 enum RelocationInfoTypeX86_64 {
238 RIT_X86_64_Unsigned = 0,
239 RIT_X86_64_Signed = 1,
240 RIT_X86_64_Branch = 2,
241 RIT_X86_64_GOTLoad = 3,
242 RIT_X86_64_GOT = 4,
243 RIT_X86_64_Subtractor = 5,
244 RIT_X86_64_Signed1 = 6,
245 RIT_X86_64_Signed2 = 7,
Eric Christopher96ac5152010-05-26 00:02:12 +0000246 RIT_X86_64_Signed4 = 8,
247 RIT_X86_64_TLV = 9
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000248 };
249
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000250 /// MachSymbolData - Helper struct for containing some precomputed information
251 /// on symbols.
252 struct MachSymbolData {
253 MCSymbolData *SymbolData;
254 uint64_t StringIndex;
255 uint8_t SectionIndex;
256
257 // Support lexicographic sorting.
258 bool operator<(const MachSymbolData &RHS) const {
Benjamin Kramerc37791e2010-05-20 14:14:22 +0000259 return SymbolData->getSymbol().getName() <
260 RHS.SymbolData->getSymbol().getName();
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000261 }
262 };
263
264 /// @name Relocation Data
265 /// @{
266
267 struct MachRelocationEntry {
268 uint32_t Word0;
269 uint32_t Word1;
270 };
271
272 llvm::DenseMap<const MCSectionData*,
273 std::vector<MachRelocationEntry> > Relocations;
Daniel Dunbar2ae4bfd2010-05-18 17:28:24 +0000274 llvm::DenseMap<const MCSectionData*, unsigned> IndirectSymBase;
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000275
276 /// @}
277 /// @name Symbol Table Data
278 /// @{
279
280 SmallString<256> StringTable;
281 std::vector<MachSymbolData> LocalSymbolData;
282 std::vector<MachSymbolData> ExternalSymbolData;
283 std::vector<MachSymbolData> UndefinedSymbolData;
284
285 /// @}
286
287 MachObjectWriter *Writer;
288
289 raw_ostream &OS;
290
291 unsigned Is64Bit : 1;
292
293public:
294 MachObjectWriterImpl(MachObjectWriter *_Writer, bool _Is64Bit)
295 : Writer(_Writer), OS(Writer->getStream()), Is64Bit(_Is64Bit) {
296 }
297
298 void Write8(uint8_t Value) { Writer->Write8(Value); }
299 void Write16(uint16_t Value) { Writer->Write16(Value); }
300 void Write32(uint32_t Value) { Writer->Write32(Value); }
301 void Write64(uint64_t Value) { Writer->Write64(Value); }
302 void WriteZeros(unsigned N) { Writer->WriteZeros(N); }
303 void WriteBytes(StringRef Str, unsigned ZeroFillSize = 0) {
304 Writer->WriteBytes(Str, ZeroFillSize);
305 }
306
307 void WriteHeader(unsigned NumLoadCommands, unsigned LoadCommandsSize,
308 bool SubsectionsViaSymbols) {
309 uint32_t Flags = 0;
310
311 if (SubsectionsViaSymbols)
312 Flags |= HF_SubsectionsViaSymbols;
313
314 // struct mach_header (28 bytes) or
315 // struct mach_header_64 (32 bytes)
316
317 uint64_t Start = OS.tell();
318 (void) Start;
319
320 Write32(Is64Bit ? Header_Magic64 : Header_Magic32);
321
322 // FIXME: Support cputype.
323 Write32(Is64Bit ? MachO::CPUTypeX86_64 : MachO::CPUTypeI386);
324 // FIXME: Support cpusubtype.
325 Write32(MachO::CPUSubType_I386_ALL);
326 Write32(HFT_Object);
327 Write32(NumLoadCommands); // Object files have a single load command, the
328 // segment.
329 Write32(LoadCommandsSize);
330 Write32(Flags);
331 if (Is64Bit)
332 Write32(0); // reserved
333
334 assert(OS.tell() - Start == Is64Bit ? Header64Size : Header32Size);
335 }
336
337 /// WriteSegmentLoadCommand - Write a segment load command.
338 ///
339 /// \arg NumSections - The number of sections in this segment.
340 /// \arg SectionDataSize - The total size of the sections.
341 void WriteSegmentLoadCommand(unsigned NumSections,
342 uint64_t VMSize,
343 uint64_t SectionDataStartOffset,
344 uint64_t SectionDataSize) {
345 // struct segment_command (56 bytes) or
346 // struct segment_command_64 (72 bytes)
347
348 uint64_t Start = OS.tell();
349 (void) Start;
350
351 unsigned SegmentLoadCommandSize = Is64Bit ? SegmentLoadCommand64Size :
352 SegmentLoadCommand32Size;
353 Write32(Is64Bit ? LCT_Segment64 : LCT_Segment);
354 Write32(SegmentLoadCommandSize +
355 NumSections * (Is64Bit ? Section64Size : Section32Size));
356
357 WriteBytes("", 16);
358 if (Is64Bit) {
359 Write64(0); // vmaddr
360 Write64(VMSize); // vmsize
361 Write64(SectionDataStartOffset); // file offset
362 Write64(SectionDataSize); // file size
363 } else {
364 Write32(0); // vmaddr
365 Write32(VMSize); // vmsize
366 Write32(SectionDataStartOffset); // file offset
367 Write32(SectionDataSize); // file size
368 }
369 Write32(0x7); // maxprot
370 Write32(0x7); // initprot
371 Write32(NumSections);
372 Write32(0); // flags
373
374 assert(OS.tell() - Start == SegmentLoadCommandSize);
375 }
376
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000377 void WriteSection(const MCAssembler &Asm, const MCAsmLayout &Layout,
378 const MCSectionData &SD, uint64_t FileOffset,
379 uint64_t RelocationsStart, unsigned NumRelocations) {
Daniel Dunbar5d428512010-03-25 02:00:07 +0000380 uint64_t SectionSize = Layout.getSectionSize(&SD);
Daniel Dunbar5d428512010-03-25 02:00:07 +0000381
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000382 // The offset is unused for virtual sections.
383 if (Asm.getBackend().isVirtualSection(SD.getSection())) {
Daniel Dunbarb026d642010-03-25 07:10:05 +0000384 assert(Layout.getSectionFileSize(&SD) == 0 && "Invalid file size!");
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000385 FileOffset = 0;
386 }
387
388 // struct section (68 bytes) or
389 // struct section_64 (80 bytes)
390
391 uint64_t Start = OS.tell();
392 (void) Start;
393
Daniel Dunbar56279f42010-05-18 17:28:20 +0000394 const MCSectionMachO &Section = cast<MCSectionMachO>(SD.getSection());
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000395 WriteBytes(Section.getSectionName(), 16);
396 WriteBytes(Section.getSegmentName(), 16);
397 if (Is64Bit) {
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000398 Write64(Layout.getSectionAddress(&SD)); // address
Daniel Dunbar5d428512010-03-25 02:00:07 +0000399 Write64(SectionSize); // size
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000400 } else {
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000401 Write32(Layout.getSectionAddress(&SD)); // address
Daniel Dunbar5d428512010-03-25 02:00:07 +0000402 Write32(SectionSize); // size
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000403 }
404 Write32(FileOffset);
405
406 unsigned Flags = Section.getTypeAndAttributes();
407 if (SD.hasInstructions())
408 Flags |= MCSectionMachO::S_ATTR_SOME_INSTRUCTIONS;
409
410 assert(isPowerOf2_32(SD.getAlignment()) && "Invalid alignment!");
411 Write32(Log2_32(SD.getAlignment()));
412 Write32(NumRelocations ? RelocationsStart : 0);
413 Write32(NumRelocations);
414 Write32(Flags);
Daniel Dunbar2ae4bfd2010-05-18 17:28:24 +0000415 Write32(IndirectSymBase.lookup(&SD)); // reserved1
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000416 Write32(Section.getStubSize()); // reserved2
417 if (Is64Bit)
418 Write32(0); // reserved3
419
420 assert(OS.tell() - Start == Is64Bit ? Section64Size : Section32Size);
421 }
422
423 void WriteSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols,
424 uint32_t StringTableOffset,
425 uint32_t StringTableSize) {
426 // struct symtab_command (24 bytes)
427
428 uint64_t Start = OS.tell();
429 (void) Start;
430
431 Write32(LCT_Symtab);
432 Write32(SymtabLoadCommandSize);
433 Write32(SymbolOffset);
434 Write32(NumSymbols);
435 Write32(StringTableOffset);
436 Write32(StringTableSize);
437
438 assert(OS.tell() - Start == SymtabLoadCommandSize);
439 }
440
441 void WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol,
442 uint32_t NumLocalSymbols,
443 uint32_t FirstExternalSymbol,
444 uint32_t NumExternalSymbols,
445 uint32_t FirstUndefinedSymbol,
446 uint32_t NumUndefinedSymbols,
447 uint32_t IndirectSymbolOffset,
448 uint32_t NumIndirectSymbols) {
449 // struct dysymtab_command (80 bytes)
450
451 uint64_t Start = OS.tell();
452 (void) Start;
453
454 Write32(LCT_Dysymtab);
455 Write32(DysymtabLoadCommandSize);
456 Write32(FirstLocalSymbol);
457 Write32(NumLocalSymbols);
458 Write32(FirstExternalSymbol);
459 Write32(NumExternalSymbols);
460 Write32(FirstUndefinedSymbol);
461 Write32(NumUndefinedSymbols);
462 Write32(0); // tocoff
463 Write32(0); // ntoc
464 Write32(0); // modtaboff
465 Write32(0); // nmodtab
466 Write32(0); // extrefsymoff
467 Write32(0); // nextrefsyms
468 Write32(IndirectSymbolOffset);
469 Write32(NumIndirectSymbols);
470 Write32(0); // extreloff
471 Write32(0); // nextrel
472 Write32(0); // locreloff
473 Write32(0); // nlocrel
474
475 assert(OS.tell() - Start == DysymtabLoadCommandSize);
476 }
477
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000478 void WriteNlist(MachSymbolData &MSD, const MCAsmLayout &Layout) {
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000479 MCSymbolData &Data = *MSD.SymbolData;
480 const MCSymbol &Symbol = Data.getSymbol();
481 uint8_t Type = 0;
482 uint16_t Flags = Data.getFlags();
483 uint32_t Address = 0;
484
485 // Set the N_TYPE bits. See <mach-o/nlist.h>.
486 //
487 // FIXME: Are the prebound or indirect fields possible here?
488 if (Symbol.isUndefined())
489 Type = STT_Undefined;
490 else if (Symbol.isAbsolute())
491 Type = STT_Absolute;
492 else
493 Type = STT_Section;
494
495 // FIXME: Set STAB bits.
496
497 if (Data.isPrivateExtern())
498 Type |= STF_PrivateExtern;
499
500 // Set external bit.
501 if (Data.isExternal() || Symbol.isUndefined())
502 Type |= STF_External;
503
504 // Compute the symbol address.
505 if (Symbol.isDefined()) {
506 if (Symbol.isAbsolute()) {
Daniel Dunbar2d7fd612010-05-05 19:01:05 +0000507 Address = cast<MCConstantExpr>(Symbol.getVariableValue())->getValue();
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000508 } else {
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000509 Address = Layout.getSymbolAddress(&Data);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000510 }
511 } else if (Data.isCommon()) {
512 // Common symbols are encoded with the size in the address
513 // field, and their alignment in the flags.
514 Address = Data.getCommonSize();
515
516 // Common alignment is packed into the 'desc' bits.
517 if (unsigned Align = Data.getCommonAlignment()) {
518 unsigned Log2Size = Log2_32(Align);
519 assert((1U << Log2Size) == Align && "Invalid 'common' alignment!");
520 if (Log2Size > 15)
Chris Lattner75361b62010-04-07 22:58:41 +0000521 report_fatal_error("invalid 'common' alignment '" +
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000522 Twine(Align) + "'");
523 // FIXME: Keep this mask with the SymbolFlags enumeration.
524 Flags = (Flags & 0xF0FF) | (Log2Size << 8);
525 }
526 }
527
528 // struct nlist (12 bytes)
529
530 Write32(MSD.StringIndex);
531 Write8(Type);
532 Write8(MSD.SectionIndex);
533
534 // The Mach-O streamer uses the lowest 16-bits of the flags for the 'desc'
535 // value.
536 Write16(Flags);
537 if (Is64Bit)
538 Write64(Address);
539 else
540 Write32(Address);
541 }
542
Daniel Dunbar35b06572010-03-22 23:16:43 +0000543 // FIXME: We really need to improve the relocation validation. Basically, we
544 // want to implement a separate computation which evaluates the relocation
545 // entry as the linker would, and verifies that the resultant fixup value is
546 // exactly what the encoder wanted. This will catch several classes of
547 // problems:
548 //
549 // - Relocation entry bugs, the two algorithms are unlikely to have the same
550 // exact bug.
551 //
552 // - Relaxation issues, where we forget to relax something.
553 //
554 // - Input errors, where something cannot be correctly encoded. 'as' allows
555 // these through in many cases.
556
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000557 void RecordX86_64Relocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
Daniel Dunbarb7514182010-03-22 20:35:50 +0000558 const MCFragment *Fragment,
Daniel Dunbarc90e30a2010-05-26 15:18:56 +0000559 const MCFixup &Fixup, MCValue Target,
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000560 uint64_t &FixedValue) {
Daniel Dunbar482ad802010-05-26 15:18:31 +0000561 unsigned IsPCRel = isFixupKindPCRel(Fixup.getKind());
562 unsigned IsRIPRel = isFixupKindRIPRel(Fixup.getKind());
563 unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind());
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000564
565 // See <reloc.h>.
Daniel Dunbar482ad802010-05-26 15:18:31 +0000566 uint32_t FixupOffset =
567 Layout.getFragmentOffset(Fragment) + Fixup.getOffset();
568 uint32_t FixupAddress =
569 Layout.getFragmentAddress(Fragment) + Fixup.getOffset();
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000570 int64_t Value = 0;
571 unsigned Index = 0;
572 unsigned IsExtern = 0;
573 unsigned Type = 0;
574
575 Value = Target.getConstant();
576
577 if (IsPCRel) {
578 // Compensate for the relocation offset, Darwin x86_64 relocations only
579 // have the addend and appear to have attempted to define it to be the
580 // actual expression addend without the PCrel bias. However, instructions
581 // with data following the relocation are not accomodated for (see comment
582 // below regarding SIGNED{1,2,4}), so it isn't exactly that either.
Benjamin Kramer454c4ce2010-04-08 15:25:57 +0000583 Value += 1LL << Log2Size;
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000584 }
585
586 if (Target.isAbsolute()) { // constant
587 // SymbolNum of 0 indicates the absolute section.
588 Type = RIT_X86_64_Unsigned;
589 Index = 0;
590
591 // FIXME: I believe this is broken, I don't think the linker can
592 // understand it. I think it would require a local relocation, but I'm not
593 // sure if that would work either. The official way to get an absolute
594 // PCrel relocation is to use an absolute symbol (which we don't support
595 // yet).
596 if (IsPCRel) {
597 IsExtern = 1;
598 Type = RIT_X86_64_Branch;
599 }
600 } else if (Target.getSymB()) { // A - B + constant
601 const MCSymbol *A = &Target.getSymA()->getSymbol();
602 MCSymbolData &A_SD = Asm.getSymbolData(*A);
Rafael Espindolab8141102010-09-27 18:13:03 +0000603 const MCSymbolData *A_Base = Asm.getAtom(&A_SD);
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000604
605 const MCSymbol *B = &Target.getSymB()->getSymbol();
606 MCSymbolData &B_SD = Asm.getSymbolData(*B);
Rafael Espindolab8141102010-09-27 18:13:03 +0000607 const MCSymbolData *B_Base = Asm.getAtom(&B_SD);
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000608
609 // Neither symbol can be modified.
610 if (Target.getSymA()->getKind() != MCSymbolRefExpr::VK_None ||
611 Target.getSymB()->getKind() != MCSymbolRefExpr::VK_None)
Chris Lattner75361b62010-04-07 22:58:41 +0000612 report_fatal_error("unsupported relocation of modified symbol");
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000613
614 // We don't support PCrel relocations of differences. Darwin 'as' doesn't
615 // implement most of these correctly.
616 if (IsPCRel)
Chris Lattner75361b62010-04-07 22:58:41 +0000617 report_fatal_error("unsupported pc-relative relocation of difference");
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000618
619 // We don't currently support any situation where one or both of the
620 // symbols would require a local relocation. This is almost certainly
621 // unused and may not be possible to encode correctly.
622 if (!A_Base || !B_Base)
Chris Lattner75361b62010-04-07 22:58:41 +0000623 report_fatal_error("unsupported local relocations in difference");
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000624
625 // Darwin 'as' doesn't emit correct relocations for this (it ends up with
626 // a single SIGNED relocation); reject it for now.
627 if (A_Base == B_Base)
Chris Lattner75361b62010-04-07 22:58:41 +0000628 report_fatal_error("unsupported relocation with identical base");
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000629
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000630 Value += Layout.getSymbolAddress(&A_SD) - Layout.getSymbolAddress(A_Base);
631 Value -= Layout.getSymbolAddress(&B_SD) - Layout.getSymbolAddress(B_Base);
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000632
633 Index = A_Base->getIndex();
634 IsExtern = 1;
635 Type = RIT_X86_64_Unsigned;
636
637 MachRelocationEntry MRE;
Daniel Dunbar640e9482010-05-11 23:53:07 +0000638 MRE.Word0 = FixupOffset;
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000639 MRE.Word1 = ((Index << 0) |
640 (IsPCRel << 24) |
641 (Log2Size << 25) |
642 (IsExtern << 27) |
643 (Type << 28));
Daniel Dunbarb7514182010-03-22 20:35:50 +0000644 Relocations[Fragment->getParent()].push_back(MRE);
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000645
646 Index = B_Base->getIndex();
647 IsExtern = 1;
648 Type = RIT_X86_64_Subtractor;
649 } else {
650 const MCSymbol *Symbol = &Target.getSymA()->getSymbol();
651 MCSymbolData &SD = Asm.getSymbolData(*Symbol);
Rafael Espindolab8141102010-09-27 18:13:03 +0000652 const MCSymbolData *Base = Asm.getAtom(&SD);
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000653
Daniel Dunbarae7fb0b2010-05-05 17:22:39 +0000654 // Relocations inside debug sections always use local relocations when
655 // possible. This seems to be done because the debugger doesn't fully
656 // understand x86_64 relocation entries, and expects to find values that
657 // have already been fixed up.
Daniel Dunbar2d7fd612010-05-05 19:01:05 +0000658 if (Symbol->isInSection()) {
Daniel Dunbarae7fb0b2010-05-05 17:22:39 +0000659 const MCSectionMachO &Section = static_cast<const MCSectionMachO&>(
660 Fragment->getParent()->getSection());
661 if (Section.hasAttribute(MCSectionMachO::S_ATTR_DEBUG))
662 Base = 0;
663 }
664
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000665 // x86_64 almost always uses external relocations, except when there is no
666 // symbol to use as a base address (a local symbol with no preceeding
667 // non-local symbol).
668 if (Base) {
669 Index = Base->getIndex();
670 IsExtern = 1;
671
672 // Add the local offset, if needed.
673 if (Base != &SD)
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000674 Value += Layout.getSymbolAddress(&SD) - Layout.getSymbolAddress(Base);
Daniel Dunbaref4591e2010-05-11 23:53:05 +0000675 } else if (Symbol->isInSection()) {
Daniel Dunbar8fb04032010-03-25 08:08:54 +0000676 // The index is the section ordinal (1-based).
677 Index = SD.getFragment()->getParent()->getOrdinal() + 1;
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000678 IsExtern = 0;
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000679 Value += Layout.getSymbolAddress(&SD);
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000680
681 if (IsPCRel)
Daniel Dunbardb4c7e62010-05-11 23:53:11 +0000682 Value -= FixupAddress + (1 << Log2Size);
Daniel Dunbaref4591e2010-05-11 23:53:05 +0000683 } else {
684 report_fatal_error("unsupported relocation of undefined symbol '" +
685 Symbol->getName() + "'");
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000686 }
687
688 MCSymbolRefExpr::VariantKind Modifier = Target.getSymA()->getKind();
689 if (IsPCRel) {
690 if (IsRIPRel) {
691 if (Modifier == MCSymbolRefExpr::VK_GOTPCREL) {
692 // x86_64 distinguishes movq foo@GOTPCREL so that the linker can
693 // rewrite the movq to an leaq at link time if the symbol ends up in
694 // the same linkage unit.
Daniel Dunbar482ad802010-05-26 15:18:31 +0000695 if (unsigned(Fixup.getKind()) == X86::reloc_riprel_4byte_movq_load)
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000696 Type = RIT_X86_64_GOTLoad;
697 else
698 Type = RIT_X86_64_GOT;
Eric Christopheraeed4d82010-05-27 00:52:31 +0000699 } else if (Modifier == MCSymbolRefExpr::VK_TLVP) {
Eric Christopher96ac5152010-05-26 00:02:12 +0000700 Type = RIT_X86_64_TLV;
Eric Christopheraeed4d82010-05-27 00:52:31 +0000701 } else if (Modifier != MCSymbolRefExpr::VK_None) {
702 report_fatal_error("unsupported symbol modifier in relocation");
Daniel Dunbarf0f6cdb2010-05-14 18:53:40 +0000703 } else {
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000704 Type = RIT_X86_64_Signed;
Daniel Dunbarf0f6cdb2010-05-14 18:53:40 +0000705
706 // The Darwin x86_64 relocation format has a problem where it cannot
707 // encode an address (L<foo> + <constant>) which is outside the atom
708 // containing L<foo>. Generally, this shouldn't occur but it does
709 // happen when we have a RIPrel instruction with data following the
710 // relocation entry (e.g., movb $012, L0(%rip)). Even with the PCrel
711 // adjustment Darwin x86_64 uses, the offset is still negative and
712 // the linker has no way to recognize this.
713 //
714 // To work around this, Darwin uses several special relocation types
715 // to indicate the offsets. However, the specification or
716 // implementation of these seems to also be incomplete; they should
717 // adjust the addend as well based on the actual encoded instruction
718 // (the additional bias), but instead appear to just look at the
719 // final offset.
720 switch (-(Target.getConstant() + (1LL << Log2Size))) {
721 case 1: Type = RIT_X86_64_Signed1; break;
722 case 2: Type = RIT_X86_64_Signed2; break;
723 case 4: Type = RIT_X86_64_Signed4; break;
724 }
725 }
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000726 } else {
727 if (Modifier != MCSymbolRefExpr::VK_None)
Chris Lattner75361b62010-04-07 22:58:41 +0000728 report_fatal_error("unsupported symbol modifier in branch "
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000729 "relocation");
730
731 Type = RIT_X86_64_Branch;
732 }
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000733 } else {
Daniel Dunbar1de558b2010-03-29 23:56:40 +0000734 if (Modifier == MCSymbolRefExpr::VK_GOT) {
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000735 Type = RIT_X86_64_GOT;
Daniel Dunbar1de558b2010-03-29 23:56:40 +0000736 } else if (Modifier == MCSymbolRefExpr::VK_GOTPCREL) {
737 // GOTPCREL is allowed as a modifier on non-PCrel instructions, in
738 // which case all we do is set the PCrel bit in the relocation entry;
739 // this is used with exception handling, for example. The source is
740 // required to include any necessary offset directly.
741 Type = RIT_X86_64_GOT;
742 IsPCRel = 1;
Eric Christopher96ac5152010-05-26 00:02:12 +0000743 } else if (Modifier == MCSymbolRefExpr::VK_TLVP) {
744 report_fatal_error("TLVP symbol modifier should have been rip-rel");
Daniel Dunbar1de558b2010-03-29 23:56:40 +0000745 } else if (Modifier != MCSymbolRefExpr::VK_None)
Chris Lattner75361b62010-04-07 22:58:41 +0000746 report_fatal_error("unsupported symbol modifier in relocation");
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000747 else
748 Type = RIT_X86_64_Unsigned;
749 }
750 }
751
752 // x86_64 always writes custom values into the fixups.
753 FixedValue = Value;
754
755 // struct relocation_info (8 bytes)
756 MachRelocationEntry MRE;
Daniel Dunbar640e9482010-05-11 23:53:07 +0000757 MRE.Word0 = FixupOffset;
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000758 MRE.Word1 = ((Index << 0) |
759 (IsPCRel << 24) |
760 (Log2Size << 25) |
761 (IsExtern << 27) |
762 (Type << 28));
Daniel Dunbarb7514182010-03-22 20:35:50 +0000763 Relocations[Fragment->getParent()].push_back(MRE);
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000764 }
765
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000766 void RecordScatteredRelocation(const MCAssembler &Asm,
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000767 const MCAsmLayout &Layout,
Daniel Dunbarb7514182010-03-22 20:35:50 +0000768 const MCFragment *Fragment,
Daniel Dunbarc90e30a2010-05-26 15:18:56 +0000769 const MCFixup &Fixup, MCValue Target,
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000770 uint64_t &FixedValue) {
Daniel Dunbar482ad802010-05-26 15:18:31 +0000771 uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
772 unsigned IsPCRel = isFixupKindPCRel(Fixup.getKind());
773 unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind());
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000774 unsigned Type = RIT_Vanilla;
775
776 // See <reloc.h>.
777 const MCSymbol *A = &Target.getSymA()->getSymbol();
778 MCSymbolData *A_SD = &Asm.getSymbolData(*A);
779
780 if (!A_SD->getFragment())
Chris Lattner75361b62010-04-07 22:58:41 +0000781 report_fatal_error("symbol '" + A->getName() +
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000782 "' can not be undefined in a subtraction expression");
783
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000784 uint32_t Value = Layout.getSymbolAddress(A_SD);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000785 uint32_t Value2 = 0;
786
787 if (const MCSymbolRefExpr *B = Target.getSymB()) {
788 MCSymbolData *B_SD = &Asm.getSymbolData(B->getSymbol());
789
790 if (!B_SD->getFragment())
Chris Lattner75361b62010-04-07 22:58:41 +0000791 report_fatal_error("symbol '" + B->getSymbol().getName() +
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000792 "' can not be undefined in a subtraction expression");
793
794 // Select the appropriate difference relocation type.
795 //
796 // Note that there is no longer any semantic difference between these two
797 // relocation types from the linkers point of view, this is done solely
798 // for pedantic compatibility with 'as'.
799 Type = A_SD->isExternal() ? RIT_Difference : RIT_LocalDifference;
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000800 Value2 = Layout.getSymbolAddress(B_SD);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000801 }
802
803 // Relocations are written out in reverse order, so the PAIR comes first.
804 if (Type == RIT_Difference || Type == RIT_LocalDifference) {
805 MachRelocationEntry MRE;
806 MRE.Word0 = ((0 << 0) |
807 (RIT_Pair << 24) |
808 (Log2Size << 28) |
809 (IsPCRel << 30) |
810 RF_Scattered);
811 MRE.Word1 = Value2;
Daniel Dunbarb7514182010-03-22 20:35:50 +0000812 Relocations[Fragment->getParent()].push_back(MRE);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000813 }
814
815 MachRelocationEntry MRE;
Daniel Dunbar640e9482010-05-11 23:53:07 +0000816 MRE.Word0 = ((FixupOffset << 0) |
817 (Type << 24) |
818 (Log2Size << 28) |
819 (IsPCRel << 30) |
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000820 RF_Scattered);
821 MRE.Word1 = Value;
Daniel Dunbarb7514182010-03-22 20:35:50 +0000822 Relocations[Fragment->getParent()].push_back(MRE);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000823 }
824
Eric Christopherc9ada472010-06-15 22:59:05 +0000825 void RecordTLVPRelocation(const MCAssembler &Asm,
Eric Christophere48dbf82010-06-16 00:26:36 +0000826 const MCAsmLayout &Layout,
827 const MCFragment *Fragment,
828 const MCFixup &Fixup, MCValue Target,
829 uint64_t &FixedValue) {
Eric Christopherc9ada472010-06-15 22:59:05 +0000830 assert(Target.getSymA()->getKind() == MCSymbolRefExpr::VK_TLVP &&
831 !Is64Bit &&
832 "Should only be called with a 32-bit TLVP relocation!");
833
Eric Christopherc9ada472010-06-15 22:59:05 +0000834 unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind());
835 uint32_t Value = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
836 unsigned IsPCRel = 0;
837
838 // Get the symbol data.
839 MCSymbolData *SD_A = &Asm.getSymbolData(Target.getSymA()->getSymbol());
840 unsigned Index = SD_A->getIndex();
841
Eric Christopherbc067372010-06-16 21:32:38 +0000842 // We're only going to have a second symbol in pic mode and it'll be a
843 // subtraction from the picbase. For 32-bit pic the addend is the difference
Eric Christopher04b8d3c2010-06-17 00:49:46 +0000844 // between the picbase and the next address. For 32-bit static the addend
845 // is zero.
Eric Christopherbc067372010-06-16 21:32:38 +0000846 if (Target.getSymB()) {
Eric Christopher1008d352010-06-22 23:51:47 +0000847 // If this is a subtraction then we're pcrel.
848 uint32_t FixupAddress =
849 Layout.getFragmentAddress(Fragment) + Fixup.getOffset();
850 MCSymbolData *SD_B = &Asm.getSymbolData(Target.getSymB()->getSymbol());
Eric Christopherc9ada472010-06-15 22:59:05 +0000851 IsPCRel = 1;
Eric Christopher1008d352010-06-22 23:51:47 +0000852 FixedValue = (FixupAddress - Layout.getSymbolAddress(SD_B) +
853 Target.getConstant());
Chris Lattnerabf8f9c2010-08-16 16:35:20 +0000854 FixedValue += 1ULL << Log2Size;
Eric Christopherbc067372010-06-16 21:32:38 +0000855 } else {
856 FixedValue = 0;
857 }
Eric Christopherc9ada472010-06-15 22:59:05 +0000858
859 // struct relocation_info (8 bytes)
860 MachRelocationEntry MRE;
861 MRE.Word0 = Value;
862 MRE.Word1 = ((Index << 0) |
863 (IsPCRel << 24) |
864 (Log2Size << 25) |
865 (1 << 27) | // Extern
866 (RIT_TLV << 28)); // Type
867 Relocations[Fragment->getParent()].push_back(MRE);
868 }
869
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000870 void RecordRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
Daniel Dunbarc90e30a2010-05-26 15:18:56 +0000871 const MCFragment *Fragment, const MCFixup &Fixup,
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000872 MCValue Target, uint64_t &FixedValue) {
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000873 if (Is64Bit) {
Daniel Dunbar207e06e2010-03-24 03:43:40 +0000874 RecordX86_64Relocation(Asm, Layout, Fragment, Fixup, Target, FixedValue);
Daniel Dunbar602b40f2010-03-19 18:07:55 +0000875 return;
876 }
877
Daniel Dunbar482ad802010-05-26 15:18:31 +0000878 unsigned IsPCRel = isFixupKindPCRel(Fixup.getKind());
879 unsigned Log2Size = getFixupKindLog2Size(Fixup.getKind());
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000880
Eric Christopherc9ada472010-06-15 22:59:05 +0000881 // If this is a 32-bit TLVP reloc it's handled a bit differently.
Daniel Dunbar23bea412010-09-17 15:21:50 +0000882 if (Target.getSymA() &&
883 Target.getSymA()->getKind() == MCSymbolRefExpr::VK_TLVP) {
Eric Christopherc9ada472010-06-15 22:59:05 +0000884 RecordTLVPRelocation(Asm, Layout, Fragment, Fixup, Target, FixedValue);
885 return;
886 }
887
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000888 // If this is a difference or a defined symbol plus an offset, then we need
889 // a scattered relocation entry.
Daniel Dunbara8251fa2010-05-10 23:15:20 +0000890 // Differences always require scattered relocations.
891 if (Target.getSymB())
892 return RecordScatteredRelocation(Asm, Layout, Fragment, Fixup,
893 Target, FixedValue);
894
895 // Get the symbol data, if any.
896 MCSymbolData *SD = 0;
897 if (Target.getSymA())
898 SD = &Asm.getSymbolData(Target.getSymA()->getSymbol());
899
900 // If this is an internal relocation with an offset, it also needs a
901 // scattered relocation entry.
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000902 uint32_t Offset = Target.getConstant();
903 if (IsPCRel)
904 Offset += 1 << Log2Size;
Daniel Dunbara8251fa2010-05-10 23:15:20 +0000905 if (Offset && SD && !doesSymbolRequireExternRelocation(SD))
906 return RecordScatteredRelocation(Asm, Layout, Fragment, Fixup,
907 Target, FixedValue);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000908
909 // See <reloc.h>.
Daniel Dunbar482ad802010-05-26 15:18:31 +0000910 uint32_t FixupOffset = Layout.getFragmentOffset(Fragment)+Fixup.getOffset();
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000911 unsigned Index = 0;
912 unsigned IsExtern = 0;
913 unsigned Type = 0;
914
915 if (Target.isAbsolute()) { // constant
916 // SymbolNum of 0 indicates the absolute section.
917 //
918 // FIXME: Currently, these are never generated (see code below). I cannot
919 // find a case where they are actually emitted.
920 Type = RIT_Vanilla;
Michael J. Spencerb0f3b3e2010-08-10 16:00:49 +0000921 } else {
Daniel Dunbare9460ec2010-05-10 23:15:13 +0000922 // Check whether we need an external or internal relocation.
923 if (doesSymbolRequireExternRelocation(SD)) {
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000924 IsExtern = 1;
925 Index = SD->getIndex();
Daniel Dunbare9460ec2010-05-10 23:15:13 +0000926 // For external relocations, make sure to offset the fixup value to
927 // compensate for the addend of the symbol address, if it was
928 // undefined. This occurs with weak definitions, for example.
929 if (!SD->Symbol->isUndefined())
Kevin Enderbya6eeb6e2010-05-07 21:44:23 +0000930 FixedValue -= Layout.getSymbolAddress(SD);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000931 } else {
Daniel Dunbar8fb04032010-03-25 08:08:54 +0000932 // The index is the section ordinal (1-based).
933 Index = SD->getFragment()->getParent()->getOrdinal() + 1;
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000934 }
935
936 Type = RIT_Vanilla;
937 }
938
939 // struct relocation_info (8 bytes)
940 MachRelocationEntry MRE;
Daniel Dunbar640e9482010-05-11 23:53:07 +0000941 MRE.Word0 = FixupOffset;
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000942 MRE.Word1 = ((Index << 0) |
943 (IsPCRel << 24) |
944 (Log2Size << 25) |
945 (IsExtern << 27) |
946 (Type << 28));
Daniel Dunbarb7514182010-03-22 20:35:50 +0000947 Relocations[Fragment->getParent()].push_back(MRE);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000948 }
949
950 void BindIndirectSymbols(MCAssembler &Asm) {
951 // This is the point where 'as' creates actual symbols for indirect symbols
952 // (in the following two passes). It would be easier for us to do this
953 // sooner when we see the attribute, but that makes getting the order in the
954 // symbol table much more complicated than it is worth.
955 //
956 // FIXME: Revisit this when the dust settles.
957
958 // Bind non lazy symbol pointers first.
Daniel Dunbar2ae4bfd2010-05-18 17:28:24 +0000959 unsigned IndirectIndex = 0;
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000960 for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
Daniel Dunbar2ae4bfd2010-05-18 17:28:24 +0000961 ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000962 const MCSectionMachO &Section =
Daniel Dunbar56279f42010-05-18 17:28:20 +0000963 cast<MCSectionMachO>(it->SectionData->getSection());
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000964
965 if (Section.getType() != MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS)
966 continue;
967
Daniel Dunbar2ae4bfd2010-05-18 17:28:24 +0000968 // Initialize the section indirect symbol base, if necessary.
969 if (!IndirectSymBase.count(it->SectionData))
970 IndirectSymBase[it->SectionData] = IndirectIndex;
971
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000972 Asm.getOrCreateSymbolData(*it->Symbol);
973 }
974
975 // Then lazy symbol pointers and symbol stubs.
Daniel Dunbar2ae4bfd2010-05-18 17:28:24 +0000976 IndirectIndex = 0;
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000977 for (MCAssembler::indirect_symbol_iterator it = Asm.indirect_symbol_begin(),
Daniel Dunbar2ae4bfd2010-05-18 17:28:24 +0000978 ie = Asm.indirect_symbol_end(); it != ie; ++it, ++IndirectIndex) {
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000979 const MCSectionMachO &Section =
Daniel Dunbar56279f42010-05-18 17:28:20 +0000980 cast<MCSectionMachO>(it->SectionData->getSection());
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000981
982 if (Section.getType() != MCSectionMachO::S_LAZY_SYMBOL_POINTERS &&
983 Section.getType() != MCSectionMachO::S_SYMBOL_STUBS)
984 continue;
985
Daniel Dunbar2ae4bfd2010-05-18 17:28:24 +0000986 // Initialize the section indirect symbol base, if necessary.
987 if (!IndirectSymBase.count(it->SectionData))
988 IndirectSymBase[it->SectionData] = IndirectIndex;
989
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +0000990 // Set the symbol type to undefined lazy, but only on construction.
991 //
992 // FIXME: Do not hardcode.
993 bool Created;
994 MCSymbolData &Entry = Asm.getOrCreateSymbolData(*it->Symbol, &Created);
995 if (Created)
996 Entry.setFlags(Entry.getFlags() | 0x0001);
997 }
998 }
999
1000 /// ComputeSymbolTable - Compute the symbol table data
1001 ///
1002 /// \param StringTable [out] - The string table data.
1003 /// \param StringIndexMap [out] - Map from symbol names to offsets in the
1004 /// string table.
1005 void ComputeSymbolTable(MCAssembler &Asm, SmallString<256> &StringTable,
1006 std::vector<MachSymbolData> &LocalSymbolData,
1007 std::vector<MachSymbolData> &ExternalSymbolData,
1008 std::vector<MachSymbolData> &UndefinedSymbolData) {
1009 // Build section lookup table.
1010 DenseMap<const MCSection*, uint8_t> SectionIndexMap;
1011 unsigned Index = 1;
1012 for (MCAssembler::iterator it = Asm.begin(),
1013 ie = Asm.end(); it != ie; ++it, ++Index)
1014 SectionIndexMap[&it->getSection()] = Index;
1015 assert(Index <= 256 && "Too many sections!");
1016
1017 // Index 0 is always the empty string.
1018 StringMap<uint64_t> StringIndexMap;
1019 StringTable += '\x00';
1020
1021 // Build the symbol arrays and the string table, but only for non-local
1022 // symbols.
1023 //
1024 // The particular order that we collect the symbols and create the string
1025 // table, then sort the symbols is chosen to match 'as'. Even though it
1026 // doesn't matter for correctness, this is important for letting us diff .o
1027 // files.
1028 for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
1029 ie = Asm.symbol_end(); it != ie; ++it) {
1030 const MCSymbol &Symbol = it->getSymbol();
1031
1032 // Ignore non-linker visible symbols.
Daniel Dunbar843aa1f2010-06-16 20:04:29 +00001033 if (!Asm.isSymbolLinkerVisible(it->getSymbol()))
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001034 continue;
1035
1036 if (!it->isExternal() && !Symbol.isUndefined())
1037 continue;
1038
1039 uint64_t &Entry = StringIndexMap[Symbol.getName()];
1040 if (!Entry) {
1041 Entry = StringTable.size();
1042 StringTable += Symbol.getName();
1043 StringTable += '\x00';
1044 }
1045
1046 MachSymbolData MSD;
1047 MSD.SymbolData = it;
1048 MSD.StringIndex = Entry;
1049
1050 if (Symbol.isUndefined()) {
1051 MSD.SectionIndex = 0;
1052 UndefinedSymbolData.push_back(MSD);
1053 } else if (Symbol.isAbsolute()) {
1054 MSD.SectionIndex = 0;
1055 ExternalSymbolData.push_back(MSD);
1056 } else {
1057 MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
1058 assert(MSD.SectionIndex && "Invalid section index!");
1059 ExternalSymbolData.push_back(MSD);
1060 }
1061 }
1062
1063 // Now add the data for local symbols.
1064 for (MCAssembler::symbol_iterator it = Asm.symbol_begin(),
1065 ie = Asm.symbol_end(); it != ie; ++it) {
1066 const MCSymbol &Symbol = it->getSymbol();
1067
1068 // Ignore non-linker visible symbols.
Daniel Dunbar843aa1f2010-06-16 20:04:29 +00001069 if (!Asm.isSymbolLinkerVisible(it->getSymbol()))
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001070 continue;
1071
1072 if (it->isExternal() || Symbol.isUndefined())
1073 continue;
1074
1075 uint64_t &Entry = StringIndexMap[Symbol.getName()];
1076 if (!Entry) {
1077 Entry = StringTable.size();
1078 StringTable += Symbol.getName();
1079 StringTable += '\x00';
1080 }
1081
1082 MachSymbolData MSD;
1083 MSD.SymbolData = it;
1084 MSD.StringIndex = Entry;
1085
1086 if (Symbol.isAbsolute()) {
1087 MSD.SectionIndex = 0;
1088 LocalSymbolData.push_back(MSD);
1089 } else {
1090 MSD.SectionIndex = SectionIndexMap.lookup(&Symbol.getSection());
1091 assert(MSD.SectionIndex && "Invalid section index!");
1092 LocalSymbolData.push_back(MSD);
1093 }
1094 }
1095
1096 // External and undefined symbols are required to be in lexicographic order.
1097 std::sort(ExternalSymbolData.begin(), ExternalSymbolData.end());
1098 std::sort(UndefinedSymbolData.begin(), UndefinedSymbolData.end());
1099
1100 // Set the symbol indices.
1101 Index = 0;
1102 for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
1103 LocalSymbolData[i].SymbolData->setIndex(Index++);
1104 for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
1105 ExternalSymbolData[i].SymbolData->setIndex(Index++);
1106 for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
1107 UndefinedSymbolData[i].SymbolData->setIndex(Index++);
1108
1109 // The string table is padded to a multiple of 4.
1110 while (StringTable.size() % 4)
1111 StringTable += '\x00';
1112 }
1113
Daniel Dunbar873decb2010-03-20 01:58:40 +00001114 void ExecutePostLayoutBinding(MCAssembler &Asm) {
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001115 // Create symbol data for any indirect symbols.
1116 BindIndirectSymbols(Asm);
1117
1118 // Compute symbol table information and bind symbol indices.
1119 ComputeSymbolTable(Asm, StringTable, LocalSymbolData, ExternalSymbolData,
1120 UndefinedSymbolData);
1121 }
1122
Rafael Espindola70703872010-09-30 02:22:20 +00001123
1124 bool IsFixupFullyResolved(const MCAssembler &Asm,
1125 const MCValue Target,
1126 bool IsPCRel,
1127 const MCFragment *DF) const {
1128 // If we are using scattered symbols, determine whether this value is
1129 // actually resolved; scattering may cause atoms to move.
1130 if (Asm.getBackend().hasScatteredSymbols()) {
1131 if (Asm.getBackend().hasReliableSymbolDifference()) {
1132 // If this is a PCrel relocation, find the base atom (identified by its
1133 // symbol) that the fixup value is relative to.
1134 const MCSymbolData *BaseSymbol = 0;
1135 if (IsPCRel) {
1136 BaseSymbol = DF->getAtom();
1137 if (!BaseSymbol)
1138 return false;
1139 }
1140
1141 return isScatteredFixupFullyResolved(Asm, Target, BaseSymbol);
1142 } else {
1143 const MCSection *BaseSection = 0;
1144 if (IsPCRel)
1145 BaseSection = &DF->getParent()->getSection();
1146
1147 return isScatteredFixupFullyResolvedSimple(Asm, Target, BaseSection);
1148 }
1149 }
1150 return true;
1151 }
1152
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001153 void WriteObject(const MCAssembler &Asm, const MCAsmLayout &Layout) {
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001154 unsigned NumSections = Asm.size();
1155
1156 // The section data starts after the header, the segment load command (and
1157 // section headers) and the symbol table.
1158 unsigned NumLoadCommands = 1;
1159 uint64_t LoadCommandsSize = Is64Bit ?
1160 SegmentLoadCommand64Size + NumSections * Section64Size :
1161 SegmentLoadCommand32Size + NumSections * Section32Size;
1162
1163 // Add the symbol table load command sizes, if used.
1164 unsigned NumSymbols = LocalSymbolData.size() + ExternalSymbolData.size() +
1165 UndefinedSymbolData.size();
1166 if (NumSymbols) {
1167 NumLoadCommands += 2;
1168 LoadCommandsSize += SymtabLoadCommandSize + DysymtabLoadCommandSize;
1169 }
1170
1171 // Compute the total size of the section data, as well as its file size and
1172 // vm size.
1173 uint64_t SectionDataStart = (Is64Bit ? Header64Size : Header32Size)
1174 + LoadCommandsSize;
1175 uint64_t SectionDataSize = 0;
1176 uint64_t SectionDataFileSize = 0;
1177 uint64_t VMSize = 0;
1178 for (MCAssembler::const_iterator it = Asm.begin(),
1179 ie = Asm.end(); it != ie; ++it) {
1180 const MCSectionData &SD = *it;
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001181 uint64_t Address = Layout.getSectionAddress(&SD);
Daniel Dunbar5d428512010-03-25 02:00:07 +00001182 uint64_t Size = Layout.getSectionSize(&SD);
1183 uint64_t FileSize = Layout.getSectionFileSize(&SD);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001184
Daniel Dunbar5d428512010-03-25 02:00:07 +00001185 VMSize = std::max(VMSize, Address + Size);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001186
1187 if (Asm.getBackend().isVirtualSection(SD.getSection()))
1188 continue;
1189
Daniel Dunbar5d428512010-03-25 02:00:07 +00001190 SectionDataSize = std::max(SectionDataSize, Address + Size);
1191 SectionDataFileSize = std::max(SectionDataFileSize, Address + FileSize);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001192 }
1193
1194 // The section data is padded to 4 bytes.
1195 //
1196 // FIXME: Is this machine dependent?
1197 unsigned SectionDataPadding = OffsetToAlignment(SectionDataFileSize, 4);
1198 SectionDataFileSize += SectionDataPadding;
1199
1200 // Write the prolog, starting with the header and load command...
1201 WriteHeader(NumLoadCommands, LoadCommandsSize,
1202 Asm.getSubsectionsViaSymbols());
1203 WriteSegmentLoadCommand(NumSections, VMSize,
1204 SectionDataStart, SectionDataSize);
1205
1206 // ... and then the section headers.
1207 uint64_t RelocTableEnd = SectionDataStart + SectionDataFileSize;
1208 for (MCAssembler::const_iterator it = Asm.begin(),
1209 ie = Asm.end(); it != ie; ++it) {
1210 std::vector<MachRelocationEntry> &Relocs = Relocations[it];
1211 unsigned NumRelocs = Relocs.size();
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001212 uint64_t SectionStart = SectionDataStart + Layout.getSectionAddress(it);
1213 WriteSection(Asm, Layout, *it, SectionStart, RelocTableEnd, NumRelocs);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001214 RelocTableEnd += NumRelocs * RelocationInfoSize;
1215 }
1216
1217 // Write the symbol table load command, if used.
1218 if (NumSymbols) {
1219 unsigned FirstLocalSymbol = 0;
1220 unsigned NumLocalSymbols = LocalSymbolData.size();
1221 unsigned FirstExternalSymbol = FirstLocalSymbol + NumLocalSymbols;
1222 unsigned NumExternalSymbols = ExternalSymbolData.size();
1223 unsigned FirstUndefinedSymbol = FirstExternalSymbol + NumExternalSymbols;
1224 unsigned NumUndefinedSymbols = UndefinedSymbolData.size();
1225 unsigned NumIndirectSymbols = Asm.indirect_symbol_size();
1226 unsigned NumSymTabSymbols =
1227 NumLocalSymbols + NumExternalSymbols + NumUndefinedSymbols;
1228 uint64_t IndirectSymbolSize = NumIndirectSymbols * 4;
1229 uint64_t IndirectSymbolOffset = 0;
1230
1231 // If used, the indirect symbols are written after the section data.
1232 if (NumIndirectSymbols)
1233 IndirectSymbolOffset = RelocTableEnd;
1234
1235 // The symbol table is written after the indirect symbol data.
1236 uint64_t SymbolTableOffset = RelocTableEnd + IndirectSymbolSize;
1237
1238 // The string table is written after symbol table.
1239 uint64_t StringTableOffset =
1240 SymbolTableOffset + NumSymTabSymbols * (Is64Bit ? Nlist64Size :
1241 Nlist32Size);
1242 WriteSymtabLoadCommand(SymbolTableOffset, NumSymTabSymbols,
1243 StringTableOffset, StringTable.size());
1244
1245 WriteDysymtabLoadCommand(FirstLocalSymbol, NumLocalSymbols,
1246 FirstExternalSymbol, NumExternalSymbols,
1247 FirstUndefinedSymbol, NumUndefinedSymbols,
1248 IndirectSymbolOffset, NumIndirectSymbols);
1249 }
1250
1251 // Write the actual section data.
1252 for (MCAssembler::const_iterator it = Asm.begin(),
1253 ie = Asm.end(); it != ie; ++it)
Daniel Dunbar432cd5f2010-03-25 02:00:02 +00001254 Asm.WriteSectionData(it, Layout, Writer);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001255
1256 // Write the extra padding.
1257 WriteZeros(SectionDataPadding);
1258
1259 // Write the relocation entries.
1260 for (MCAssembler::const_iterator it = Asm.begin(),
1261 ie = Asm.end(); it != ie; ++it) {
1262 // Write the section relocation entries, in reverse order to match 'as'
1263 // (approximately, the exact algorithm is more complicated than this).
1264 std::vector<MachRelocationEntry> &Relocs = Relocations[it];
1265 for (unsigned i = 0, e = Relocs.size(); i != e; ++i) {
1266 Write32(Relocs[e - i - 1].Word0);
1267 Write32(Relocs[e - i - 1].Word1);
1268 }
1269 }
1270
1271 // Write the symbol table data, if used.
1272 if (NumSymbols) {
1273 // Write the indirect symbol entries.
1274 for (MCAssembler::const_indirect_symbol_iterator
1275 it = Asm.indirect_symbol_begin(),
1276 ie = Asm.indirect_symbol_end(); it != ie; ++it) {
1277 // Indirect symbols in the non lazy symbol pointer section have some
1278 // special handling.
1279 const MCSectionMachO &Section =
1280 static_cast<const MCSectionMachO&>(it->SectionData->getSection());
1281 if (Section.getType() == MCSectionMachO::S_NON_LAZY_SYMBOL_POINTERS) {
1282 // If this symbol is defined and internal, mark it as such.
1283 if (it->Symbol->isDefined() &&
1284 !Asm.getSymbolData(*it->Symbol).isExternal()) {
1285 uint32_t Flags = ISF_Local;
1286 if (it->Symbol->isAbsolute())
1287 Flags |= ISF_Absolute;
1288 Write32(Flags);
1289 continue;
1290 }
1291 }
1292
1293 Write32(Asm.getSymbolData(*it->Symbol).getIndex());
1294 }
1295
1296 // FIXME: Check that offsets match computed ones.
1297
1298 // Write the symbol table entries.
1299 for (unsigned i = 0, e = LocalSymbolData.size(); i != e; ++i)
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001300 WriteNlist(LocalSymbolData[i], Layout);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001301 for (unsigned i = 0, e = ExternalSymbolData.size(); i != e; ++i)
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001302 WriteNlist(ExternalSymbolData[i], Layout);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001303 for (unsigned i = 0, e = UndefinedSymbolData.size(); i != e; ++i)
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001304 WriteNlist(UndefinedSymbolData[i], Layout);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001305
1306 // Write the string table.
1307 OS << StringTable.str();
1308 }
1309 }
1310};
1311
1312}
1313
1314MachObjectWriter::MachObjectWriter(raw_ostream &OS,
1315 bool Is64Bit,
1316 bool IsLittleEndian)
1317 : MCObjectWriter(OS, IsLittleEndian)
1318{
1319 Impl = new MachObjectWriterImpl(this, Is64Bit);
1320}
1321
1322MachObjectWriter::~MachObjectWriter() {
1323 delete (MachObjectWriterImpl*) Impl;
1324}
1325
1326void MachObjectWriter::ExecutePostLayoutBinding(MCAssembler &Asm) {
1327 ((MachObjectWriterImpl*) Impl)->ExecutePostLayoutBinding(Asm);
1328}
1329
1330void MachObjectWriter::RecordRelocation(const MCAssembler &Asm,
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001331 const MCAsmLayout &Layout,
Daniel Dunbarb7514182010-03-22 20:35:50 +00001332 const MCFragment *Fragment,
Daniel Dunbarc90e30a2010-05-26 15:18:56 +00001333 const MCFixup &Fixup, MCValue Target,
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001334 uint64_t &FixedValue) {
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001335 ((MachObjectWriterImpl*) Impl)->RecordRelocation(Asm, Layout, Fragment, Fixup,
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001336 Target, FixedValue);
1337}
1338
Rafael Espindola70703872010-09-30 02:22:20 +00001339bool MachObjectWriter::IsFixupFullyResolved(const MCAssembler &Asm,
1340 const MCValue Target,
1341 bool IsPCRel,
1342 const MCFragment *DF) const {
1343 return ((MachObjectWriterImpl*) Impl)->IsFixupFullyResolved(Asm, Target,
1344 IsPCRel, DF);
1345}
1346
Daniel Dunbar207e06e2010-03-24 03:43:40 +00001347void MachObjectWriter::WriteObject(const MCAssembler &Asm,
1348 const MCAsmLayout &Layout) {
1349 ((MachObjectWriterImpl*) Impl)->WriteObject(Asm, Layout);
Daniel Dunbar2df4ceb2010-03-19 10:43:15 +00001350}