lib/ExecutionEngine/RuntimeDyld/RuntimeDyldMachO.cpp

   1 //===-- RuntimeDyldMachO.cpp - Run-time dynamic linker for MC-JIT -*- C++ -*-=//
   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 // Implementation of the MC-JIT runtime dynamic linker.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #define DEBUG_TYPE "dyld"
  15 #include "RuntimeDyldMachO.h"
  16 #include "llvm/ADT/OwningPtr.h"
  17 #include "llvm/ADT/STLExtras.h"
  18 #include "llvm/ADT/StringRef.h"
  19 using namespace llvm;
  20 using namespace llvm::object;
  21
  22 namespace llvm {
  23
  24 static unsigned char *processFDE(unsigned char *P, intptr_t DeltaForText, intptr_t DeltaForEH) {
  25   uint32_t Length = *((uint32_t*)P);
  26   P += 4;
  27   unsigned char *Ret = P + Length;
  28   uint32_t Offset = *((uint32_t*)P);
  29   if (Offset == 0) // is a CIE
  30     return Ret;
  31
  32   P += 4;
  33   intptr_t FDELocation = *((intptr_t*)P);
  34   intptr_t NewLocation = FDELocation - DeltaForText;
  35   *((intptr_t*)P) = NewLocation;
  36   P += sizeof(intptr_t);
  37
  38   // Skip the FDE address range
  39   P += sizeof(intptr_t);
  40
  41   uint8_t Augmentationsize = *P;
  42   P += 1;
  43   if (Augmentationsize != 0) {
  44     intptr_t LSDA = *((intptr_t*)P);
  45     intptr_t NewLSDA = LSDA - DeltaForEH;
  46     *((intptr_t*)P) = NewLSDA;
  47   }
  48
  49   return Ret;
  50 }
  51
  52 static intptr_t computeDelta(SectionEntry *A, SectionEntry *B) {
  53   intptr_t ObjDistance = A->ObjAddress  - B->ObjAddress;
  54   intptr_t MemDistance = A->LoadAddress - B->LoadAddress;
  55   return ObjDistance - MemDistance;
  56 }
  57
  58 void RuntimeDyldMachO::registerEHFrames() {
  59
  60   if (!MemMgr)
  61     return;
  62   for (int i = 0, e = UnregisteredEHFrameSections.size(); i != e; ++i) {
  63     EHFrameRelatedSections &SectionInfo = UnregisteredEHFrameSections[i];
  64     if (SectionInfo.EHFrameSID == RTDYLD_INVALID_SECTION_ID ||
  65         SectionInfo.TextSID == RTDYLD_INVALID_SECTION_ID)
  66       continue;
  67     SectionEntry *Text = &Sections[SectionInfo.TextSID];
  68     SectionEntry *EHFrame = &Sections[SectionInfo.EHFrameSID];
  69     SectionEntry *ExceptTab = NULL;
  70     if (SectionInfo.ExceptTabSID != RTDYLD_INVALID_SECTION_ID)
  71       ExceptTab = &Sections[SectionInfo.ExceptTabSID];
  72
  73     intptr_t DeltaForText = computeDelta(Text, EHFrame);
  74     intptr_t DeltaForEH = 0;
  75     if (ExceptTab)
  76       DeltaForEH = computeDelta(ExceptTab, EHFrame);
  77
  78     unsigned char *P = EHFrame->Address;
  79     unsigned char *End = P + EHFrame->Size;
  80     do  {
  81       P = processFDE(P, DeltaForText, DeltaForEH);
  82     } while(P != End);
  83
  84     MemMgr->registerEHFrames(EHFrame->Address,
  85                              EHFrame->LoadAddress,
  86                              EHFrame->Size);
  87   }
  88   UnregisteredEHFrameSections.clear();
  89 }
  90
  91 void RuntimeDyldMachO::finalizeLoad(ObjSectionToIDMap &SectionMap) {
  92   unsigned EHFrameSID = RTDYLD_INVALID_SECTION_ID;
  93   unsigned TextSID = RTDYLD_INVALID_SECTION_ID;
  94   unsigned ExceptTabSID = RTDYLD_INVALID_SECTION_ID;
  95   ObjSectionToIDMap::iterator i, e;
  96   for (i = SectionMap.begin(), e = SectionMap.end(); i != e; ++i) {
  97     const SectionRef &Section = i->first;
  98     StringRef Name;
  99     Section.getName(Name);
 100     if (Name == "__eh_frame")
 101       EHFrameSID = i->second;
 102     else if (Name == "__text")
 103       TextSID = i->second;
 104     else if (Name == "__gcc_except_tab")
 105       ExceptTabSID = i->second;
 106   }
 107   UnregisteredEHFrameSections.push_back(EHFrameRelatedSections(EHFrameSID,
 108                                                                TextSID,
 109                                                                ExceptTabSID));
 110 }
 111
 112 // The target location for the relocation is described by RE.SectionID and
 113 // RE.Offset.  RE.SectionID can be used to find the SectionEntry.  Each
 114 // SectionEntry has three members describing its location.
 115 // SectionEntry::Address is the address at which the section has been loaded
 116 // into memory in the current (host) process.  SectionEntry::LoadAddress is the
 117 // address that the section will have in the target process.
 118 // SectionEntry::ObjAddress is the address of the bits for this section in the
 119 // original emitted object image (also in the current address space).
 120 //
 121 // Relocations will be applied as if the section were loaded at
 122 // SectionEntry::LoadAddress, but they will be applied at an address based
 123 // on SectionEntry::Address.  SectionEntry::ObjAddress will be used to refer to
 124 // Target memory contents if they are required for value calculations.
 125 //
 126 // The Value parameter here is the load address of the symbol for the
 127 // relocation to be applied.  For relocations which refer to symbols in the
 128 // current object Value will be the LoadAddress of the section in which
 129 // the symbol resides (RE.Addend provides additional information about the
 130 // symbol location).  For external symbols, Value will be the address of the
 131 // symbol in the target address space.
 132 void RuntimeDyldMachO::resolveRelocation(const RelocationEntry &RE,
 133                                          uint64_t Value) {
 134   const SectionEntry &Section = Sections[RE.SectionID];
 135   return resolveRelocation(Section, RE.Offset, Value, RE.RelType, RE.Addend,
 136                            RE.IsPCRel, RE.Size);
 137 }
 138
 139 void RuntimeDyldMachO::resolveRelocation(const SectionEntry &Section,
 140                                          uint64_t Offset,
 141                                          uint64_t Value,
 142                                          uint32_t Type,
 143                                          int64_t Addend,
 144                                          bool isPCRel,
 145                                          unsigned LogSize) {
 146   uint8_t *LocalAddress = Section.Address + Offset;
 147   uint64_t FinalAddress = Section.LoadAddress + Offset;
 148   unsigned MachoType = Type;
 149   unsigned Size = 1 << LogSize;
 150
 151   DEBUG(dbgs() << "resolveRelocation LocalAddress: "
 152         << format("%p", LocalAddress)
 153         << " FinalAddress: " << format("%p", FinalAddress)
 154         << " Value: " << format("%p", Value)
 155         << " Addend: " << Addend
 156         << " isPCRel: " << isPCRel
 157         << " MachoType: " << MachoType
 158         << " Size: " << Size
 159         << "\n");
 160
 161   // This just dispatches to the proper target specific routine.
 162   switch (Arch) {
 163   default: llvm_unreachable("Unsupported CPU type!");
 164   case Triple::x86_64:
 165     resolveX86_64Relocation(LocalAddress,
 166                             FinalAddress,
 167                             (uintptr_t)Value,
 168                             isPCRel,
 169                             MachoType,
 170                             Size,
 171                             Addend);
 172     break;
 173   case Triple::x86:
 174     resolveI386Relocation(LocalAddress,
 175                           FinalAddress,
 176                           (uintptr_t)Value,
 177                           isPCRel,
 178                           MachoType,
 179                           Size,
 180                           Addend);
 181     break;
 182   case Triple::arm:    // Fall through.
 183   case Triple::thumb:
 184     resolveARMRelocation(LocalAddress,
 185                          FinalAddress,
 186                          (uintptr_t)Value,
 187                          isPCRel,
 188                          MachoType,
 189                          Size,
 190                          Addend);
 191     break;
 192   }
 193 }
 194
 195 bool RuntimeDyldMachO::resolveI386Relocation(uint8_t *LocalAddress,
 196                                              uint64_t FinalAddress,
 197                                              uint64_t Value,
 198                                              bool isPCRel,
 199                                              unsigned Type,
 200                                              unsigned Size,
 201                                              int64_t Addend) {
 202   if (isPCRel)
 203     Value -= FinalAddress + 4; // see resolveX86_64Relocation
 204
 205   switch (Type) {
 206   default:
 207     llvm_unreachable("Invalid relocation type!");
 208   case MachO::GENERIC_RELOC_VANILLA: {
 209     uint8_t *p = LocalAddress;
 210     uint64_t ValueToWrite = Value + Addend;
 211     for (unsigned i = 0; i < Size; ++i) {
 212       *p++ = (uint8_t)(ValueToWrite & 0xff);
 213       ValueToWrite >>= 8;
 214     }
 215     return false;
 216   }
 217   case MachO::GENERIC_RELOC_SECTDIFF:
 218   case MachO::GENERIC_RELOC_LOCAL_SECTDIFF:
 219   case MachO::GENERIC_RELOC_PB_LA_PTR:
 220     return Error("Relocation type not implemented yet!");
 221   }
 222 }
 223
 224 bool RuntimeDyldMachO::resolveX86_64Relocation(uint8_t *LocalAddress,
 225                                                uint64_t FinalAddress,
 226                                                uint64_t Value,
 227                                                bool isPCRel,
 228                                                unsigned Type,
 229                                                unsigned Size,
 230                                                int64_t Addend) {
 231   // If the relocation is PC-relative, the value to be encoded is the
 232   // pointer difference.
 233   if (isPCRel)
 234     // FIXME: It seems this value needs to be adjusted by 4 for an effective PC
 235     // address. Is that expected? Only for branches, perhaps?
 236     Value -= FinalAddress + 4;
 237
 238   switch(Type) {
 239   default:
 240     llvm_unreachable("Invalid relocation type!");
 241   case MachO::X86_64_RELOC_SIGNED_1:
 242   case MachO::X86_64_RELOC_SIGNED_2:
 243   case MachO::X86_64_RELOC_SIGNED_4:
 244   case MachO::X86_64_RELOC_SIGNED:
 245   case MachO::X86_64_RELOC_UNSIGNED:
 246   case MachO::X86_64_RELOC_BRANCH: {
 247     Value += Addend;
 248     // Mask in the target value a byte at a time (we don't have an alignment
 249     // guarantee for the target address, so this is safest).
 250     uint8_t *p = (uint8_t*)LocalAddress;
 251     for (unsigned i = 0; i < Size; ++i) {
 252       *p++ = (uint8_t)Value;
 253       Value >>= 8;
 254     }
 255     return false;
 256   }
 257   case MachO::X86_64_RELOC_GOT_LOAD:
 258   case MachO::X86_64_RELOC_GOT:
 259   case MachO::X86_64_RELOC_SUBTRACTOR:
 260   case MachO::X86_64_RELOC_TLV:
 261     return Error("Relocation type not implemented yet!");
 262   }
 263 }
 264
 265 bool RuntimeDyldMachO::resolveARMRelocation(uint8_t *LocalAddress,
 266                                             uint64_t FinalAddress,
 267                                             uint64_t Value,
 268                                             bool isPCRel,
 269                                             unsigned Type,
 270                                             unsigned Size,
 271                                             int64_t Addend) {
 272   // If the relocation is PC-relative, the value to be encoded is the
 273   // pointer difference.
 274   if (isPCRel) {
 275     Value -= FinalAddress;
 276     // ARM PCRel relocations have an effective-PC offset of two instructions
 277     // (four bytes in Thumb mode, 8 bytes in ARM mode).
 278     // FIXME: For now, assume ARM mode.
 279     Value -= 8;
 280   }
 281
 282   switch(Type) {
 283   default:
 284     llvm_unreachable("Invalid relocation type!");
 285   case MachO::ARM_RELOC_VANILLA: {
 286     // Mask in the target value a byte at a time (we don't have an alignment
 287     // guarantee for the target address, so this is safest).
 288     uint8_t *p = (uint8_t*)LocalAddress;
 289     for (unsigned i = 0; i < Size; ++i) {
 290       *p++ = (uint8_t)Value;
 291       Value >>= 8;
 292     }
 293     break;
 294   }
 295   case MachO::ARM_RELOC_BR24: {
 296     // Mask the value into the target address. We know instructions are
 297     // 32-bit aligned, so we can do it all at once.
 298     uint32_t *p = (uint32_t*)LocalAddress;
 299     // The low two bits of the value are not encoded.
 300     Value >>= 2;
 301     // Mask the value to 24 bits.
 302     Value &= 0xffffff;
 303     // FIXME: If the destination is a Thumb function (and the instruction
 304     // is a non-predicated BL instruction), we need to change it to a BLX
 305     // instruction instead.
 306
 307     // Insert the value into the instruction.
 308     *p = (*p & ~0xffffff) | Value;
 309     break;
 310   }
 311   case MachO::ARM_THUMB_RELOC_BR22:
 312   case MachO::ARM_THUMB_32BIT_BRANCH:
 313   case MachO::ARM_RELOC_HALF:
 314   case MachO::ARM_RELOC_HALF_SECTDIFF:
 315   case MachO::ARM_RELOC_PAIR:
 316   case MachO::ARM_RELOC_SECTDIFF:
 317   case MachO::ARM_RELOC_LOCAL_SECTDIFF:
 318   case MachO::ARM_RELOC_PB_LA_PTR:
 319     return Error("Relocation type not implemented yet!");
 320   }
 321   return false;
 322 }
 323
 324 void RuntimeDyldMachO::processRelocationRef(unsigned SectionID,
 325                                             RelocationRef RelI,
 326                                             ObjectImage &Obj,
 327                                             ObjSectionToIDMap &ObjSectionToID,
 328                                             const SymbolTableMap &Symbols,
 329                                             StubMap &Stubs) {
 330   const ObjectFile *OF = Obj.getObjectFile();
 331   const MachOObjectFile *MachO = static_cast<const MachOObjectFile*>(OF);
 332   MachO::any_relocation_info RE= MachO->getRelocation(RelI.getRawDataRefImpl());
 333
 334   uint32_t RelType = MachO->getAnyRelocationType(RE);
 335
 336   // FIXME: Properly handle scattered relocations.
 337   //        For now, optimistically skip these: they can often be ignored, as
 338   //        the static linker will already have applied the relocation, and it
 339   //        only needs to be reapplied if symbols move relative to one another.
 340   //        Note: This will fail horribly where the relocations *do* need to be
 341   //        applied, but that was already the case.
 342   if (MachO->isRelocationScattered(RE))
 343     return;
 344
 345   RelocationValueRef Value;
 346   SectionEntry &Section = Sections[SectionID];
 347
 348   bool isExtern = MachO->getPlainRelocationExternal(RE);
 349   bool IsPCRel = MachO->getAnyRelocationPCRel(RE);
 350   unsigned Size = MachO->getAnyRelocationLength(RE);
 351   uint64_t Offset;
 352   RelI.getOffset(Offset);
 353   uint8_t *LocalAddress = Section.Address + Offset;
 354   unsigned NumBytes = 1 << Size;
 355   uint64_t Addend = 0;
 356   memcpy(&Addend, LocalAddress, NumBytes);
 357
 358   if (isExtern) {
 359     // Obtain the symbol name which is referenced in the relocation
 360     symbol_iterator Symbol = RelI.getSymbol();
 361     StringRef TargetName;
 362     Symbol->getName(TargetName);
 363     // First search for the symbol in the local symbol table
 364     SymbolTableMap::const_iterator lsi = Symbols.find(TargetName.data());
 365     if (lsi != Symbols.end()) {
 366       Value.SectionID = lsi->second.first;
 367       Value.Addend = lsi->second.second + Addend;
 368     } else {
 369       // Search for the symbol in the global symbol table
 370       SymbolTableMap::const_iterator gsi = GlobalSymbolTable.find(TargetName.data());
 371       if (gsi != GlobalSymbolTable.end()) {
 372         Value.SectionID = gsi->second.first;
 373         Value.Addend = gsi->second.second + Addend;
 374       } else {
 375         Value.SymbolName = TargetName.data();
 376         Value.Addend = Addend;
 377       }
 378     }
 379   } else {
 380     SectionRef Sec = MachO->getRelocationSection(RE);
 381     Value.SectionID = findOrEmitSection(Obj, Sec, true, ObjSectionToID);
 382     uint64_t Addr;
 383     Sec.getAddress(Addr);
 384     Value.Addend = Addend - Addr;
 385   }
 386
 387   if (Arch == Triple::x86_64 && (RelType == MachO::X86_64_RELOC_GOT ||
 388                                  RelType == MachO::X86_64_RELOC_GOT_LOAD)) {
 389     assert(IsPCRel);
 390     assert(Size == 2);
 391     StubMap::const_iterator i = Stubs.find(Value);
 392     uint8_t *Addr;
 393     if (i != Stubs.end()) {
 394       Addr = Section.Address + i->second;
 395     } else {
 396       Stubs[Value] = Section.StubOffset;
 397       uint8_t *GOTEntry = Section.Address + Section.StubOffset;
 398       RelocationEntry RE(SectionID, Section.StubOffset,
 399                          MachO::X86_64_RELOC_UNSIGNED, 0, false, 3);
 400       if (Value.SymbolName)
 401         addRelocationForSymbol(RE, Value.SymbolName);
 402       else
 403         addRelocationForSection(RE, Value.SectionID);
 404       Section.StubOffset += 8;
 405       Addr = GOTEntry;
 406     }
 407     resolveRelocation(Section, Offset, (uint64_t)Addr,
 408                       MachO::X86_64_RELOC_UNSIGNED, Value.Addend, true, 2);
 409   } else if (Arch == Triple::arm &&
 410              (RelType & 0xf) == MachO::ARM_RELOC_BR24) {
 411     // This is an ARM branch relocation, need to use a stub function.
 412
 413     //  Look up for existing stub.
 414     StubMap::const_iterator i = Stubs.find(Value);
 415     if (i != Stubs.end())
 416       resolveRelocation(Section, Offset,
 417                         (uint64_t)Section.Address + i->second,
 418                         RelType, 0, IsPCRel, Size);
 419     else {
 420       // Create a new stub function.
 421       Stubs[Value] = Section.StubOffset;
 422       uint8_t *StubTargetAddr = createStubFunction(Section.Address +
 423                                                    Section.StubOffset);
 424       RelocationEntry RE(SectionID, StubTargetAddr - Section.Address,
 425                          MachO::GENERIC_RELOC_VANILLA, Value.Addend);
 426       if (Value.SymbolName)
 427         addRelocationForSymbol(RE, Value.SymbolName);
 428       else
 429         addRelocationForSection(RE, Value.SectionID);
 430       resolveRelocation(Section, Offset,
 431                         (uint64_t)Section.Address + Section.StubOffset,
 432                         RelType, 0, IsPCRel, Size);
 433       Section.StubOffset += getMaxStubSize();
 434     }
 435   } else {
 436     RelocationEntry RE(SectionID, Offset, RelType, Value.Addend,
 437                        IsPCRel, Size);
 438     if (Value.SymbolName)
 439       addRelocationForSymbol(RE, Value.SymbolName);
 440     else
 441       addRelocationForSection(RE, Value.SectionID);
 442   }
 443 }
 444
 445
 446 bool RuntimeDyldMachO::isCompatibleFormat(
 447         const ObjectBuffer *InputBuffer) const {
 448   if (InputBuffer->getBufferSize() < 4)
 449     return false;
 450   StringRef Magic(InputBuffer->getBufferStart(), 4);
 451   if (Magic == "\xFE\xED\xFA\xCE") return true;
 452   if (Magic == "\xCE\xFA\xED\xFE") return true;
 453   if (Magic == "\xFE\xED\xFA\xCF") return true;
 454   if (Magic == "\xCF\xFA\xED\xFE") return true;
 455   return false;
 456 }
 457
 458 bool RuntimeDyldMachO::isCompatibleFile(
 459         const object::ObjectFile *Obj) const {
 460   return Obj->isMachO();
 461 }
 462
 463 } // end namespace llvm