lib/ExecutionEngine/RuntimeDyld/RuntimeDyld.cpp

   1 //===-- RuntimeDyld.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 "ObjectImageCommon.h"
  16 #include "RuntimeDyldImpl.h"
  17 #include "RuntimeDyldELF.h"
  18 #include "RuntimeDyldMachO.h"
  19 #include "llvm/Support/Path.h"
  20 #include "llvm/Support/MathExtras.h"
  21
  22 using namespace llvm;
  23 using namespace llvm::object;
  24
  25 // Empty out-of-line virtual destructor as the key function.
  26 RTDyldMemoryManager::~RTDyldMemoryManager() {}
  27 RuntimeDyldImpl::~RuntimeDyldImpl() {}
  28
  29 namespace llvm {
  30
  31 // Resolve the relocations for all symbols we currently know about.
  32 void RuntimeDyldImpl::resolveRelocations() {
  33   // First, resolve relocations associated with external symbols.
  34   resolveExternalSymbols();
  35
  36   // Just iterate over the sections we have and resolve all the relocations
  37   // in them. Gross overkill, but it gets the job done.
  38   for (int i = 0, e = Sections.size(); i != e; ++i) {
  39     uint64_t Addr = Sections[i].LoadAddress;
  40     DEBUG(dbgs() << "Resolving relocations Section #" << i
  41             << "\t" << format("%p", (uint8_t *)Addr)
  42             << "\n");
  43     resolveRelocationList(Relocations[i], Addr);
  44   }
  45 }
  46
  47 void RuntimeDyldImpl::mapSectionAddress(const void *LocalAddress,
  48                                         uint64_t TargetAddress) {
  49   for (unsigned i = 0, e = Sections.size(); i != e; ++i) {
  50     if (Sections[i].Address == LocalAddress) {
  51       reassignSectionAddress(i, TargetAddress);
  52       return;
  53     }
  54   }
  55   llvm_unreachable("Attempting to remap address of unknown section!");
  56 }
  57
  58 // Subclasses can implement this method to create specialized image instances.
  59 // The caller owns the pointer that is returned.
  60 ObjectImage *RuntimeDyldImpl::createObjectImage(ObjectBuffer *InputBuffer) {
  61   return new ObjectImageCommon(InputBuffer);
  62 }
  63
  64 ObjectImage *RuntimeDyldImpl::loadObject(ObjectBuffer *InputBuffer) {
  65   OwningPtr<ObjectImage> obj(createObjectImage(InputBuffer));
  66   if (!obj)
  67     report_fatal_error("Unable to create object image from memory buffer!");
  68
  69   Arch = (Triple::ArchType)obj->getArch();
  70
  71   // Symbols found in this object
  72   StringMap<SymbolLoc> LocalSymbols;
  73   // Used sections from the object file
  74   ObjSectionToIDMap LocalSections;
  75
  76   // Common symbols requiring allocation, with their sizes and alignments
  77   CommonSymbolMap CommonSymbols;
  78   // Maximum required total memory to allocate all common symbols
  79   uint64_t CommonSize = 0;
  80
  81   error_code err;
  82   // Parse symbols
  83   DEBUG(dbgs() << "Parse symbols:\n");
  84   for (symbol_iterator i = obj->begin_symbols(), e = obj->end_symbols();
  85        i != e; i.increment(err)) {
  86     Check(err);
  87     object::SymbolRef::Type SymType;
  88     StringRef Name;
  89     Check(i->getType(SymType));
  90     Check(i->getName(Name));
  91
  92     uint32_t flags;
  93     Check(i->getFlags(flags));
  94
  95     bool isCommon = flags & SymbolRef::SF_Common;
  96     if (isCommon) {
  97       // Add the common symbols to a list.  We'll allocate them all below.
  98       uint64_t Align = getCommonSymbolAlignment(*i);
  99       uint64_t Size = 0;
 100       Check(i->getSize(Size));
 101       CommonSize += Size + Align;
 102       CommonSymbols[*i] = CommonSymbolInfo(Size, Align);
 103     } else {
 104       if (SymType == object::SymbolRef::ST_Function ||
 105           SymType == object::SymbolRef::ST_Data ||
 106           SymType == object::SymbolRef::ST_Unknown) {
 107         uint64_t FileOffset;
 108         StringRef SectionData;
 109         section_iterator si = obj->end_sections();
 110         Check(i->getFileOffset(FileOffset));
 111         Check(i->getSection(si));
 112         if (si == obj->end_sections()) continue;
 113         Check(si->getContents(SectionData));
 114         const uint8_t* SymPtr = (const uint8_t*)InputBuffer->getBufferStart() +
 115                                 (uintptr_t)FileOffset;
 116         uintptr_t SectOffset = (uintptr_t)(SymPtr -
 117                                            (const uint8_t*)SectionData.begin());
 118         unsigned SectionID =
 119           findOrEmitSection(*obj,
 120                             *si,
 121                             SymType == object::SymbolRef::ST_Function,
 122                             LocalSections);
 123         LocalSymbols[Name.data()] = SymbolLoc(SectionID, SectOffset);
 124         DEBUG(dbgs() << "\tFileOffset: " << format("%p", (uintptr_t)FileOffset)
 125                      << " flags: " << flags
 126                      << " SID: " << SectionID
 127                      << " Offset: " << format("%p", SectOffset));
 128         bool isGlobal = flags & SymbolRef::SF_Global;
 129         if (isGlobal)
 130           GlobalSymbolTable[Name] = SymbolLoc(SectionID, SectOffset);
 131       }
 132     }
 133     DEBUG(dbgs() << "\tType: " << SymType << " Name: " << Name << "\n");
 134   }
 135
 136   // Allocate common symbols
 137   if (CommonSize != 0)
 138     emitCommonSymbols(*obj, CommonSymbols, CommonSize, LocalSymbols);
 139
 140   // Parse and process relocations
 141   DEBUG(dbgs() << "Parse relocations:\n");
 142   for (section_iterator si = obj->begin_sections(),
 143        se = obj->end_sections(); si != se; si.increment(err)) {
 144     Check(err);
 145     bool isFirstRelocation = true;
 146     unsigned SectionID = 0;
 147     StubMap Stubs;
 148
 149     for (relocation_iterator i = si->begin_relocations(),
 150          e = si->end_relocations(); i != e; i.increment(err)) {
 151       Check(err);
 152
 153       // If it's the first relocation in this section, find its SectionID
 154       if (isFirstRelocation) {
 155         SectionID = findOrEmitSection(*obj, *si, true, LocalSections);
 156         DEBUG(dbgs() << "\tSectionID: " << SectionID << "\n");
 157         isFirstRelocation = false;
 158       }
 159
 160       ObjRelocationInfo RI;
 161       RI.SectionID = SectionID;
 162       Check(i->getAdditionalInfo(RI.AdditionalInfo));
 163       Check(i->getOffset(RI.Offset));
 164       Check(i->getSymbol(RI.Symbol));
 165       Check(i->getType(RI.Type));
 166
 167       DEBUG(dbgs() << "\t\tAddend: " << RI.AdditionalInfo
 168                    << " Offset: " << format("%p", (uintptr_t)RI.Offset)
 169                    << " Type: " << (uint32_t)(RI.Type & 0xffffffffL)
 170                    << "\n");
 171       processRelocationRef(RI, *obj, LocalSections, LocalSymbols, Stubs);
 172     }
 173   }
 174
 175   return obj.take();
 176 }
 177
 178 void RuntimeDyldImpl::emitCommonSymbols(ObjectImage &Obj,
 179                                         const CommonSymbolMap &CommonSymbols,
 180                                         uint64_t TotalSize,
 181                                         SymbolTableMap &SymbolTable) {
 182   // Allocate memory for the section
 183   unsigned SectionID = Sections.size();
 184   uint8_t *Addr = MemMgr->allocateDataSection(TotalSize, sizeof(void*),
 185                                               SectionID, false);
 186   if (!Addr)
 187     report_fatal_error("Unable to allocate memory for common symbols!");
 188   uint64_t Offset = 0;
 189   Sections.push_back(SectionEntry(StringRef(), Addr, TotalSize, TotalSize, 0));
 190   memset(Addr, 0, TotalSize);
 191
 192   DEBUG(dbgs() << "emitCommonSection SectionID: " << SectionID
 193                << " new addr: " << format("%p", Addr)
 194                << " DataSize: " << TotalSize
 195                << "\n");
 196
 197   // Assign the address of each symbol
 198   for (CommonSymbolMap::const_iterator it = CommonSymbols.begin(),
 199        itEnd = CommonSymbols.end(); it != itEnd; it++) {
 200     uint64_t Size = it->second.first;
 201     uint64_t Align = it->second.second;
 202     StringRef Name;
 203     it->first.getName(Name);
 204     if (Align) {
 205       // This symbol has an alignment requirement.
 206       uint64_t AlignOffset = OffsetToAlignment((uint64_t)Addr, Align);
 207       Addr += AlignOffset;
 208       Offset += AlignOffset;
 209       DEBUG(dbgs() << "Allocating common symbol " << Name << " address " <<
 210                       format("%p\n", Addr));
 211     }
 212     Obj.updateSymbolAddress(it->first, (uint64_t)Addr);
 213     SymbolTable[Name.data()] = SymbolLoc(SectionID, Offset);
 214     Offset += Size;
 215     Addr += Size;
 216   }
 217 }
 218
 219 unsigned RuntimeDyldImpl::emitSection(ObjectImage &Obj,
 220                                       const SectionRef &Section,
 221                                       bool IsCode) {
 222
 223   unsigned StubBufSize = 0,
 224            StubSize = getMaxStubSize();
 225   error_code err;
 226   if (StubSize > 0) {
 227     for (relocation_iterator i = Section.begin_relocations(),
 228          e = Section.end_relocations(); i != e; i.increment(err), Check(err))
 229       StubBufSize += StubSize;
 230   }
 231   StringRef data;
 232   uint64_t Alignment64;
 233   Check(Section.getContents(data));
 234   Check(Section.getAlignment(Alignment64));
 235
 236   unsigned Alignment = (unsigned)Alignment64 & 0xffffffffL;
 237   bool IsRequired;
 238   bool IsVirtual;
 239   bool IsZeroInit;
 240   bool IsReadOnly;
 241   uint64_t DataSize;
 242   StringRef Name;
 243   Check(Section.isRequiredForExecution(IsRequired));
 244   Check(Section.isVirtual(IsVirtual));
 245   Check(Section.isZeroInit(IsZeroInit));
 246   Check(Section.isReadOnlyData(IsReadOnly));
 247   Check(Section.getSize(DataSize));
 248   Check(Section.getName(Name));
 249
 250   unsigned Allocate;
 251   unsigned SectionID = Sections.size();
 252   uint8_t *Addr;
 253   const char *pData = 0;
 254
 255   // Some sections, such as debug info, don't need to be loaded for execution.
 256   // Leave those where they are.
 257   if (IsRequired) {
 258     Allocate = DataSize + StubBufSize;
 259     Addr = IsCode
 260       ? MemMgr->allocateCodeSection(Allocate, Alignment, SectionID)
 261       : MemMgr->allocateDataSection(Allocate, Alignment, SectionID, IsReadOnly);
 262     if (!Addr)
 263       report_fatal_error("Unable to allocate section memory!");
 264
 265     // Virtual sections have no data in the object image, so leave pData = 0
 266     if (!IsVirtual)
 267       pData = data.data();
 268
 269     // Zero-initialize or copy the data from the image
 270     if (IsZeroInit || IsVirtual)
 271       memset(Addr, 0, DataSize);
 272     else
 273       memcpy(Addr, pData, DataSize);
 274
 275     DEBUG(dbgs() << "emitSection SectionID: " << SectionID
 276                  << " Name: " << Name
 277                  << " obj addr: " << format("%p", pData)
 278                  << " new addr: " << format("%p", Addr)
 279                  << " DataSize: " << DataSize
 280                  << " StubBufSize: " << StubBufSize
 281                  << " Allocate: " << Allocate
 282                  << "\n");
 283     Obj.updateSectionAddress(Section, (uint64_t)Addr);
 284   }
 285   else {
 286     // Even if we didn't load the section, we need to record an entry for it
 287     // to handle later processing (and by 'handle' I mean don't do anything
 288     // with these sections).
 289     Allocate = 0;
 290     Addr = 0;
 291     DEBUG(dbgs() << "emitSection SectionID: " << SectionID
 292                  << " Name: " << Name
 293                  << " obj addr: " << format("%p", data.data())
 294                  << " new addr: 0"
 295                  << " DataSize: " << DataSize
 296                  << " StubBufSize: " << StubBufSize
 297                  << " Allocate: " << Allocate
 298                  << "\n");
 299   }
 300
 301   Sections.push_back(SectionEntry(Name, Addr, Allocate, DataSize,
 302                                   (uintptr_t)pData));
 303   return SectionID;
 304 }
 305
 306 unsigned RuntimeDyldImpl::findOrEmitSection(ObjectImage &Obj,
 307                                             const SectionRef &Section,
 308                                             bool IsCode,
 309                                             ObjSectionToIDMap &LocalSections) {
 310
 311   unsigned SectionID = 0;
 312   ObjSectionToIDMap::iterator i = LocalSections.find(Section);
 313   if (i != LocalSections.end())
 314     SectionID = i->second;
 315   else {
 316     SectionID = emitSection(Obj, Section, IsCode);
 317     LocalSections[Section] = SectionID;
 318   }
 319   return SectionID;
 320 }
 321
 322 void RuntimeDyldImpl::addRelocationForSection(const RelocationEntry &RE,
 323                                               unsigned SectionID) {
 324   Relocations[SectionID].push_back(RE);
 325 }
 326
 327 void RuntimeDyldImpl::addRelocationForSymbol(const RelocationEntry &RE,
 328                                              StringRef SymbolName) {
 329   // Relocation by symbol.  If the symbol is found in the global symbol table,
 330   // create an appropriate section relocation.  Otherwise, add it to
 331   // ExternalSymbolRelocations.
 332   SymbolTableMap::const_iterator Loc =
 333       GlobalSymbolTable.find(SymbolName);
 334   if (Loc == GlobalSymbolTable.end()) {
 335     ExternalSymbolRelocations[SymbolName].push_back(RE);
 336   } else {
 337     // Copy the RE since we want to modify its addend.
 338     RelocationEntry RECopy = RE;
 339     RECopy.Addend += Loc->second.second;
 340     Relocations[Loc->second.first].push_back(RECopy);
 341   }
 342 }
 343
 344 uint8_t *RuntimeDyldImpl::createStubFunction(uint8_t *Addr) {
 345   if (Arch == Triple::arm) {
 346     // TODO: There is only ARM far stub now. We should add the Thumb stub,
 347     // and stubs for branches Thumb - ARM and ARM - Thumb.
 348     uint32_t *StubAddr = (uint32_t*)Addr;
 349     *StubAddr = 0xe51ff004; // ldr pc,<label>
 350     return (uint8_t*)++StubAddr;
 351   } else if (Arch == Triple::mipsel) {
 352     uint32_t *StubAddr = (uint32_t*)Addr;
 353     // 0:   3c190000        lui     t9,%hi(addr).
 354     // 4:   27390000        addiu   t9,t9,%lo(addr).
 355     // 8:   03200008        jr      t9.
 356     // c:   00000000        nop.
 357     const unsigned LuiT9Instr = 0x3c190000, AdduiT9Instr = 0x27390000;
 358     const unsigned JrT9Instr = 0x03200008, NopInstr = 0x0;
 359
 360     *StubAddr = LuiT9Instr;
 361     StubAddr++;
 362     *StubAddr = AdduiT9Instr;
 363     StubAddr++;
 364     *StubAddr = JrT9Instr;
 365     StubAddr++;
 366     *StubAddr = NopInstr;
 367     return Addr;
 368   } else if (Arch == Triple::ppc64) {
 369     // PowerPC64 stub: the address points to a function descriptor
 370     // instead of the function itself. Load the function address
 371     // on r11 and sets it to control register. Also loads the function
 372     // TOC in r2 and environment pointer to r11.
 373     writeInt32BE(Addr,    0x3D800000); // lis   r12, highest(addr)
 374     writeInt32BE(Addr+4,  0x618C0000); // ori   r12, higher(addr)
 375     writeInt32BE(Addr+8,  0x798C07C6); // sldi  r12, r12, 32
 376     writeInt32BE(Addr+12, 0x658C0000); // oris  r12, r12, h(addr)
 377     writeInt32BE(Addr+16, 0x618C0000); // ori   r12, r12, l(addr)
 378     writeInt32BE(Addr+20, 0xF8410028); // std   r2,  40(r1)
 379     writeInt32BE(Addr+24, 0xE96C0000); // ld    r11, 0(r12)
 380     writeInt32BE(Addr+28, 0xE84C0008); // ld    r2,  0(r12)
 381     writeInt32BE(Addr+32, 0x7D6903A6); // mtctr r11
 382     writeInt32BE(Addr+36, 0xE96C0010); // ld    r11, 16(r2)
 383     writeInt32BE(Addr+40, 0x4E800420); // bctr
 384
 385     return Addr;
 386   }
 387   return Addr;
 388 }
 389
 390 // Assign an address to a symbol name and resolve all the relocations
 391 // associated with it.
 392 void RuntimeDyldImpl::reassignSectionAddress(unsigned SectionID,
 393                                              uint64_t Addr) {
 394   // The address to use for relocation resolution is not
 395   // the address of the local section buffer. We must be doing
 396   // a remote execution environment of some sort. Relocations can't
 397   // be applied until all the sections have been moved.  The client must
 398   // trigger this with a call to MCJIT::finalize() or
 399   // RuntimeDyld::resolveRelocations().
 400   //
 401   // Addr is a uint64_t because we can't assume the pointer width
 402   // of the target is the same as that of the host. Just use a generic
 403   // "big enough" type.
 404   Sections[SectionID].LoadAddress = Addr;
 405 }
 406
 407 void RuntimeDyldImpl::resolveRelocationEntry(const RelocationEntry &RE,
 408                                              uint64_t Value) {
 409   // Ignore relocations for sections that were not loaded
 410   if (Sections[RE.SectionID].Address != 0) {
 411     DEBUG(dbgs() << "\tSectionID: " << RE.SectionID
 412           << " + " << RE.Offset << " ("
 413           << format("%p", Sections[RE.SectionID].Address + RE.Offset) << ")"
 414           << " RelType: " << RE.RelType
 415           << " Addend: " << RE.Addend
 416           << "\n");
 417
 418     resolveRelocation(Sections[RE.SectionID], RE.Offset,
 419                       Value, RE.RelType, RE.Addend);
 420   }
 421 }
 422
 423 void RuntimeDyldImpl::resolveRelocationList(const RelocationList &Relocs,
 424                                             uint64_t Value) {
 425   for (unsigned i = 0, e = Relocs.size(); i != e; ++i) {
 426     resolveRelocationEntry(Relocs[i], Value);
 427   }
 428 }
 429
 430 void RuntimeDyldImpl::resolveExternalSymbols() {
 431   StringMap<RelocationList>::iterator i = ExternalSymbolRelocations.begin(),
 432                                       e = ExternalSymbolRelocations.end();
 433   for (; i != e; i++) {
 434     StringRef Name = i->first();
 435     RelocationList &Relocs = i->second;
 436     SymbolTableMap::const_iterator Loc = GlobalSymbolTable.find(Name);
 437     if (Loc == GlobalSymbolTable.end()) {
 438       // This is an external symbol, try to get it address from
 439       // MemoryManager.
 440       uint8_t *Addr = (uint8_t*) MemMgr->getPointerToNamedFunction(Name.data(),
 441                                                                    true);
 442       DEBUG(dbgs() << "Resolving relocations Name: " << Name
 443               << "\t" << format("%p", Addr)
 444               << "\n");
 445       resolveRelocationList(Relocs, (uintptr_t)Addr);
 446     } else {
 447       report_fatal_error("Expected external symbol");
 448     }
 449   }
 450 }
 451
 452
 453 //===----------------------------------------------------------------------===//
 454 // RuntimeDyld class implementation
 455 RuntimeDyld::RuntimeDyld(RTDyldMemoryManager *mm) {
 456   // FIXME: There's a potential issue lurking here if a single instance of
 457   // RuntimeDyld is used to load multiple objects.  The current implementation
 458   // associates a single memory manager with a RuntimeDyld instance.  Even
 459   // though the public class spawns a new 'impl' instance for each load,
 460   // they share a single memory manager.  This can become a problem when page
 461   // permissions are applied.
 462   Dyld = 0;
 463   MM = mm;
 464 }
 465
 466 RuntimeDyld::~RuntimeDyld() {
 467   delete Dyld;
 468 }
 469
 470 ObjectImage *RuntimeDyld::loadObject(ObjectBuffer *InputBuffer) {
 471   if (!Dyld) {
 472     sys::LLVMFileType type = sys::IdentifyFileType(
 473             InputBuffer->getBufferStart(),
 474             static_cast<unsigned>(InputBuffer->getBufferSize()));
 475     switch (type) {
 476       case sys::ELF_Relocatable_FileType:
 477       case sys::ELF_Executable_FileType:
 478       case sys::ELF_SharedObject_FileType:
 479       case sys::ELF_Core_FileType:
 480         Dyld = new RuntimeDyldELF(MM);
 481         break;
 482       case sys::Mach_O_Object_FileType:
 483       case sys::Mach_O_Executable_FileType:
 484       case sys::Mach_O_FixedVirtualMemorySharedLib_FileType:
 485       case sys::Mach_O_Core_FileType:
 486       case sys::Mach_O_PreloadExecutable_FileType:
 487       case sys::Mach_O_DynamicallyLinkedSharedLib_FileType:
 488       case sys::Mach_O_DynamicLinker_FileType:
 489       case sys::Mach_O_Bundle_FileType:
 490       case sys::Mach_O_DynamicallyLinkedSharedLibStub_FileType:
 491       case sys::Mach_O_DSYMCompanion_FileType:
 492         Dyld = new RuntimeDyldMachO(MM);
 493         break;
 494       case sys::Unknown_FileType:
 495       case sys::Bitcode_FileType:
 496       case sys::Archive_FileType:
 497       case sys::COFF_FileType:
 498         report_fatal_error("Incompatible object format!");
 499     }
 500   } else {
 501     if (!Dyld->isCompatibleFormat(InputBuffer))
 502       report_fatal_error("Incompatible object format!");
 503   }
 504
 505   return Dyld->loadObject(InputBuffer);
 506 }
 507
 508 void *RuntimeDyld::getSymbolAddress(StringRef Name) {
 509   return Dyld->getSymbolAddress(Name);
 510 }
 511
 512 uint64_t RuntimeDyld::getSymbolLoadAddress(StringRef Name) {
 513   return Dyld->getSymbolLoadAddress(Name);
 514 }
 515
 516 void RuntimeDyld::resolveRelocations() {
 517   Dyld->resolveRelocations();
 518 }
 519
 520 void RuntimeDyld::reassignSectionAddress(unsigned SectionID,
 521                                          uint64_t Addr) {
 522   Dyld->reassignSectionAddress(SectionID, Addr);
 523 }
 524
 525 void RuntimeDyld::mapSectionAddress(const void *LocalAddress,
 526                                     uint64_t TargetAddress) {
 527   Dyld->mapSectionAddress(LocalAddress, TargetAddress);
 528 }
 529
 530 StringRef RuntimeDyld::getErrorString() {
 531   return Dyld->getErrorString();
 532 }
 533
 534 } // end namespace llvm