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 "llvm/ExecutionEngine/RuntimeDyld.h"
  16 #include "ObjectImageCommon.h"
  17 #include "RuntimeDyldELF.h"
  18 #include "RuntimeDyldImpl.h"
  19 #include "RuntimeDyldMachO.h"
  20 #include "llvm/Support/MathExtras.h"
  21 #include "llvm/Support/Path.h"
  22
  23 using namespace llvm;
  24 using namespace llvm::object;
  25
  26 // Empty out-of-line virtual destructor as the key function.
  27 RuntimeDyldImpl::~RuntimeDyldImpl() {}
  28
  29 namespace llvm {
  30
  31 StringRef RuntimeDyldImpl::getEHFrameSection() {
  32   return StringRef();
  33 }
  34
  35 // Resolve the relocations for all symbols we currently know about.
  36 void RuntimeDyldImpl::resolveRelocations() {
  37   // First, resolve relocations associated with external symbols.
  38   resolveExternalSymbols();
  39
  40   // Just iterate over the sections we have and resolve all the relocations
  41   // in them. Gross overkill, but it gets the job done.
  42   for (int i = 0, e = Sections.size(); i != e; ++i) {
  43     uint64_t Addr = Sections[i].LoadAddress;
  44     DEBUG(dbgs() << "Resolving relocations Section #" << i
  45             << "\t" << format("%p", (uint8_t *)Addr)
  46             << "\n");
  47     resolveRelocationList(Relocations[i], Addr);
  48   }
  49 }
  50
  51 void RuntimeDyldImpl::mapSectionAddress(const void *LocalAddress,
  52                                         uint64_t TargetAddress) {
  53   for (unsigned i = 0, e = Sections.size(); i != e; ++i) {
  54     if (Sections[i].Address == LocalAddress) {
  55       reassignSectionAddress(i, TargetAddress);
  56       return;
  57     }
  58   }
  59   llvm_unreachable("Attempting to remap address of unknown section!");
  60 }
  61
  62 // Subclasses can implement this method to create specialized image instances.
  63 // The caller owns the pointer that is returned.
  64 ObjectImage *RuntimeDyldImpl::createObjectImage(ObjectBuffer *InputBuffer) {
  65   return new ObjectImageCommon(InputBuffer);
  66 }
  67
  68 ObjectImage *RuntimeDyldImpl::loadObject(ObjectBuffer *InputBuffer) {
  69   OwningPtr<ObjectImage> obj(createObjectImage(InputBuffer));
  70   if (!obj)
  71     report_fatal_error("Unable to create object image from memory buffer!");
  72
  73   Arch = (Triple::ArchType)obj->getArch();
  74
  75   // Symbols found in this object
  76   StringMap<SymbolLoc> LocalSymbols;
  77   // Used sections from the object file
  78   ObjSectionToIDMap LocalSections;
  79
  80   // Common symbols requiring allocation, with their sizes and alignments
  81   CommonSymbolMap CommonSymbols;
  82   // Maximum required total memory to allocate all common symbols
  83   uint64_t CommonSize = 0;
  84
  85   error_code err;
  86   // Parse symbols
  87   DEBUG(dbgs() << "Parse symbols:\n");
  88   for (symbol_iterator i = obj->begin_symbols(), e = obj->end_symbols();
  89        i != e; i.increment(err)) {
  90     Check(err);
  91     object::SymbolRef::Type SymType;
  92     StringRef Name;
  93     Check(i->getType(SymType));
  94     Check(i->getName(Name));
  95
  96     uint32_t flags;
  97     Check(i->getFlags(flags));
  98
  99     bool isCommon = flags & SymbolRef::SF_Common;
 100     if (isCommon) {
 101       // Add the common symbols to a list.  We'll allocate them all below.
 102       uint32_t Align;
 103       Check(i->getAlignment(Align));
 104       uint64_t Size = 0;
 105       Check(i->getSize(Size));
 106       CommonSize += Size + Align;
 107       CommonSymbols[*i] = CommonSymbolInfo(Size, Align);
 108     } else {
 109       if (SymType == object::SymbolRef::ST_Function ||
 110           SymType == object::SymbolRef::ST_Data ||
 111           SymType == object::SymbolRef::ST_Unknown) {
 112         uint64_t FileOffset;
 113         StringRef SectionData;
 114         bool IsCode;
 115         section_iterator si = obj->end_sections();
 116         Check(i->getFileOffset(FileOffset));
 117         Check(i->getSection(si));
 118         if (si == obj->end_sections()) continue;
 119         Check(si->getContents(SectionData));
 120         Check(si->isText(IsCode));
 121         const uint8_t* SymPtr = (const uint8_t*)InputBuffer->getBufferStart() +
 122                                 (uintptr_t)FileOffset;
 123         uintptr_t SectOffset = (uintptr_t)(SymPtr -
 124                                            (const uint8_t*)SectionData.begin());
 125         unsigned SectionID = findOrEmitSection(*obj, *si, IsCode, LocalSections);
 126         LocalSymbols[Name.data()] = SymbolLoc(SectionID, SectOffset);
 127         DEBUG(dbgs() << "\tFileOffset: " << format("%p", (uintptr_t)FileOffset)
 128                      << " flags: " << flags
 129                      << " SID: " << SectionID
 130                      << " Offset: " << format("%p", SectOffset));
 131         GlobalSymbolTable[Name] = SymbolLoc(SectionID, SectOffset);
 132       }
 133     }
 134     DEBUG(dbgs() << "\tType: " << SymType << " Name: " << Name << "\n");
 135   }
 136
 137   // Allocate common symbols
 138   if (CommonSize != 0)
 139     emitCommonSymbols(*obj, CommonSymbols, CommonSize, LocalSymbols);
 140
 141   // Parse and process relocations
 142   DEBUG(dbgs() << "Parse relocations:\n");
 143   for (section_iterator si = obj->begin_sections(),
 144        se = obj->end_sections(); si != se; si.increment(err)) {
 145     Check(err);
 146     bool isFirstRelocation = true;
 147     unsigned SectionID = 0;
 148     StubMap Stubs;
 149
 150     for (relocation_iterator i = si->begin_relocations(),
 151          e = si->end_relocations(); i != e; i.increment(err)) {
 152       Check(err);
 153
 154       // If it's the first relocation in this section, find its SectionID
 155       if (isFirstRelocation) {
 156         SectionID = findOrEmitSection(*obj, *si, true, LocalSections);
 157         DEBUG(dbgs() << "\tSectionID: " << SectionID << "\n");
 158         isFirstRelocation = false;
 159       }
 160
 161       processRelocationRef(SectionID, *i, *obj, LocalSections, LocalSymbols,
 162                            Stubs);
 163     }
 164   }
 165
 166   return obj.take();
 167 }
 168
 169 void RuntimeDyldImpl::emitCommonSymbols(ObjectImage &Obj,
 170                                         const CommonSymbolMap &CommonSymbols,
 171                                         uint64_t TotalSize,
 172                                         SymbolTableMap &SymbolTable) {
 173   // Allocate memory for the section
 174   unsigned SectionID = Sections.size();
 175   uint8_t *Addr = MemMgr->allocateDataSection(TotalSize, sizeof(void*),
 176                                               SectionID, false);
 177   if (!Addr)
 178     report_fatal_error("Unable to allocate memory for common symbols!");
 179   uint64_t Offset = 0;
 180   Sections.push_back(SectionEntry(StringRef(), Addr, TotalSize, 0));
 181   memset(Addr, 0, TotalSize);
 182
 183   DEBUG(dbgs() << "emitCommonSection SectionID: " << SectionID
 184                << " new addr: " << format("%p", Addr)
 185                << " DataSize: " << TotalSize
 186                << "\n");
 187
 188   // Assign the address of each symbol
 189   for (CommonSymbolMap::const_iterator it = CommonSymbols.begin(),
 190        itEnd = CommonSymbols.end(); it != itEnd; it++) {
 191     uint64_t Size = it->second.first;
 192     uint64_t Align = it->second.second;
 193     StringRef Name;
 194     it->first.getName(Name);
 195     if (Align) {
 196       // This symbol has an alignment requirement.
 197       uint64_t AlignOffset = OffsetToAlignment((uint64_t)Addr, Align);
 198       Addr += AlignOffset;
 199       Offset += AlignOffset;
 200       DEBUG(dbgs() << "Allocating common symbol " << Name << " address " <<
 201                       format("%p\n", Addr));
 202     }
 203     Obj.updateSymbolAddress(it->first, (uint64_t)Addr);
 204     SymbolTable[Name.data()] = SymbolLoc(SectionID, Offset);
 205     Offset += Size;
 206     Addr += Size;
 207   }
 208 }
 209
 210 unsigned RuntimeDyldImpl::emitSection(ObjectImage &Obj,
 211                                       const SectionRef &Section,
 212                                       bool IsCode) {
 213
 214   unsigned StubBufSize = 0,
 215            StubSize = getMaxStubSize();
 216   error_code err;
 217   if (StubSize > 0) {
 218     for (relocation_iterator i = Section.begin_relocations(),
 219          e = Section.end_relocations(); i != e; i.increment(err), Check(err))
 220       StubBufSize += StubSize;
 221   }
 222   StringRef data;
 223   uint64_t Alignment64;
 224   Check(Section.getContents(data));
 225   Check(Section.getAlignment(Alignment64));
 226
 227   unsigned Alignment = (unsigned)Alignment64 & 0xffffffffL;
 228   bool IsRequired;
 229   bool IsVirtual;
 230   bool IsZeroInit;
 231   bool IsReadOnly;
 232   uint64_t DataSize;
 233   StringRef Name;
 234   Check(Section.isRequiredForExecution(IsRequired));
 235   Check(Section.isVirtual(IsVirtual));
 236   Check(Section.isZeroInit(IsZeroInit));
 237   Check(Section.isReadOnlyData(IsReadOnly));
 238   Check(Section.getSize(DataSize));
 239   Check(Section.getName(Name));
 240   if (StubSize > 0) {
 241     unsigned StubAlignment = getStubAlignment();
 242     unsigned EndAlignment = (DataSize | Alignment) & -(DataSize | Alignment);
 243     if (StubAlignment > EndAlignment)
 244       StubBufSize += StubAlignment - EndAlignment;
 245   }
 246
 247   unsigned Allocate;
 248   unsigned SectionID = Sections.size();
 249   uint8_t *Addr;
 250   const char *pData = 0;
 251
 252   // Some sections, such as debug info, don't need to be loaded for execution.
 253   // Leave those where they are.
 254   if (IsRequired) {
 255     Allocate = DataSize + StubBufSize;
 256     Addr = IsCode
 257       ? MemMgr->allocateCodeSection(Allocate, Alignment, SectionID)
 258       : MemMgr->allocateDataSection(Allocate, Alignment, SectionID, IsReadOnly);
 259     if (!Addr)
 260       report_fatal_error("Unable to allocate section memory!");
 261
 262     // Virtual sections have no data in the object image, so leave pData = 0
 263     if (!IsVirtual)
 264       pData = data.data();
 265
 266     // Zero-initialize or copy the data from the image
 267     if (IsZeroInit || IsVirtual)
 268       memset(Addr, 0, DataSize);
 269     else
 270       memcpy(Addr, pData, DataSize);
 271
 272     DEBUG(dbgs() << "emitSection SectionID: " << SectionID
 273                  << " Name: " << Name
 274                  << " obj addr: " << format("%p", pData)
 275                  << " new addr: " << format("%p", Addr)
 276                  << " DataSize: " << DataSize
 277                  << " StubBufSize: " << StubBufSize
 278                  << " Allocate: " << Allocate
 279                  << "\n");
 280     Obj.updateSectionAddress(Section, (uint64_t)Addr);
 281   }
 282   else {
 283     // Even if we didn't load the section, we need to record an entry for it
 284     // to handle later processing (and by 'handle' I mean don't do anything
 285     // with these sections).
 286     Allocate = 0;
 287     Addr = 0;
 288     DEBUG(dbgs() << "emitSection SectionID: " << SectionID
 289                  << " Name: " << Name
 290                  << " obj addr: " << format("%p", data.data())
 291                  << " new addr: 0"
 292                  << " DataSize: " << DataSize
 293                  << " StubBufSize: " << StubBufSize
 294                  << " Allocate: " << Allocate
 295                  << "\n");
 296   }
 297
 298   Sections.push_back(SectionEntry(Name, Addr, DataSize, (uintptr_t)pData));
 299   return SectionID;
 300 }
 301
 302 unsigned RuntimeDyldImpl::findOrEmitSection(ObjectImage &Obj,
 303                                             const SectionRef &Section,
 304                                             bool IsCode,
 305                                             ObjSectionToIDMap &LocalSections) {
 306
 307   unsigned SectionID = 0;
 308   ObjSectionToIDMap::iterator i = LocalSections.find(Section);
 309   if (i != LocalSections.end())
 310     SectionID = i->second;
 311   else {
 312     SectionID = emitSection(Obj, Section, IsCode);
 313     LocalSections[Section] = SectionID;
 314   }
 315   return SectionID;
 316 }
 317
 318 void RuntimeDyldImpl::addRelocationForSection(const RelocationEntry &RE,
 319                                               unsigned SectionID) {
 320   Relocations[SectionID].push_back(RE);
 321 }
 322
 323 void RuntimeDyldImpl::addRelocationForSymbol(const RelocationEntry &RE,
 324                                              StringRef SymbolName) {
 325   // Relocation by symbol.  If the symbol is found in the global symbol table,
 326   // create an appropriate section relocation.  Otherwise, add it to
 327   // ExternalSymbolRelocations.
 328   SymbolTableMap::const_iterator Loc =
 329       GlobalSymbolTable.find(SymbolName);
 330   if (Loc == GlobalSymbolTable.end()) {
 331     ExternalSymbolRelocations[SymbolName].push_back(RE);
 332   } else {
 333     // Copy the RE since we want to modify its addend.
 334     RelocationEntry RECopy = RE;
 335     RECopy.Addend += Loc->second.second;
 336     Relocations[Loc->second.first].push_back(RECopy);
 337   }
 338 }
 339
 340 uint8_t *RuntimeDyldImpl::createStubFunction(uint8_t *Addr) {
 341   if (Arch == Triple::aarch64) {
 342     // This stub has to be able to access the full address space,
 343     // since symbol lookup won't necessarily find a handy, in-range,
 344     // PLT stub for functions which could be anywhere.
 345     uint32_t *StubAddr = (uint32_t*)Addr;
 346
 347     // Stub can use ip0 (== x16) to calculate address
 348     *StubAddr = 0xd2e00010; // movz ip0, #:abs_g3:<addr>
 349     StubAddr++;
 350     *StubAddr = 0xf2c00010; // movk ip0, #:abs_g2_nc:<addr>
 351     StubAddr++;
 352     *StubAddr = 0xf2a00010; // movk ip0, #:abs_g1_nc:<addr>
 353     StubAddr++;
 354     *StubAddr = 0xf2800010; // movk ip0, #:abs_g0_nc:<addr>
 355     StubAddr++;
 356     *StubAddr = 0xd61f0200; // br ip0
 357
 358     return Addr;
 359   } else if (Arch == Triple::arm) {
 360     // TODO: There is only ARM far stub now. We should add the Thumb stub,
 361     // and stubs for branches Thumb - ARM and ARM - Thumb.
 362     uint32_t *StubAddr = (uint32_t*)Addr;
 363     *StubAddr = 0xe51ff004; // ldr pc,<label>
 364     return (uint8_t*)++StubAddr;
 365   } else if (Arch == Triple::mipsel || Arch == Triple::mips) {
 366     uint32_t *StubAddr = (uint32_t*)Addr;
 367     // 0:   3c190000        lui     t9,%hi(addr).
 368     // 4:   27390000        addiu   t9,t9,%lo(addr).
 369     // 8:   03200008        jr      t9.
 370     // c:   00000000        nop.
 371     const unsigned LuiT9Instr = 0x3c190000, AdduiT9Instr = 0x27390000;
 372     const unsigned JrT9Instr = 0x03200008, NopInstr = 0x0;
 373
 374     *StubAddr = LuiT9Instr;
 375     StubAddr++;
 376     *StubAddr = AdduiT9Instr;
 377     StubAddr++;
 378     *StubAddr = JrT9Instr;
 379     StubAddr++;
 380     *StubAddr = NopInstr;
 381     return Addr;
 382   } else if (Arch == Triple::ppc64) {
 383     // PowerPC64 stub: the address points to a function descriptor
 384     // instead of the function itself. Load the function address
 385     // on r11 and sets it to control register. Also loads the function
 386     // TOC in r2 and environment pointer to r11.
 387     writeInt32BE(Addr,    0x3D800000); // lis   r12, highest(addr)
 388     writeInt32BE(Addr+4,  0x618C0000); // ori   r12, higher(addr)
 389     writeInt32BE(Addr+8,  0x798C07C6); // sldi  r12, r12, 32
 390     writeInt32BE(Addr+12, 0x658C0000); // oris  r12, r12, h(addr)
 391     writeInt32BE(Addr+16, 0x618C0000); // ori   r12, r12, l(addr)
 392     writeInt32BE(Addr+20, 0xF8410028); // std   r2,  40(r1)
 393     writeInt32BE(Addr+24, 0xE96C0000); // ld    r11, 0(r12)
 394     writeInt32BE(Addr+28, 0xE84C0008); // ld    r2,  0(r12)
 395     writeInt32BE(Addr+32, 0x7D6903A6); // mtctr r11
 396     writeInt32BE(Addr+36, 0xE96C0010); // ld    r11, 16(r2)
 397     writeInt32BE(Addr+40, 0x4E800420); // bctr
 398
 399     return Addr;
 400   } else if (Arch == Triple::systemz) {
 401     writeInt16BE(Addr,    0xC418);     // lgrl %r1,.+8
 402     writeInt16BE(Addr+2,  0x0000);
 403     writeInt16BE(Addr+4,  0x0004);
 404     writeInt16BE(Addr+6,  0x07F1);     // brc 15,%r1
 405     // 8-byte address stored at Addr + 8
 406     return Addr;
 407   }
 408   return Addr;
 409 }
 410
 411 // Assign an address to a symbol name and resolve all the relocations
 412 // associated with it.
 413 void RuntimeDyldImpl::reassignSectionAddress(unsigned SectionID,
 414                                              uint64_t Addr) {
 415   // The address to use for relocation resolution is not
 416   // the address of the local section buffer. We must be doing
 417   // a remote execution environment of some sort. Relocations can't
 418   // be applied until all the sections have been moved.  The client must
 419   // trigger this with a call to MCJIT::finalize() or
 420   // RuntimeDyld::resolveRelocations().
 421   //
 422   // Addr is a uint64_t because we can't assume the pointer width
 423   // of the target is the same as that of the host. Just use a generic
 424   // "big enough" type.
 425   Sections[SectionID].LoadAddress = Addr;
 426 }
 427
 428 void RuntimeDyldImpl::resolveRelocationList(const RelocationList &Relocs,
 429                                             uint64_t Value) {
 430   for (unsigned i = 0, e = Relocs.size(); i != e; ++i) {
 431     const RelocationEntry &RE = Relocs[i];
 432     // Ignore relocations for sections that were not loaded
 433     if (Sections[RE.SectionID].Address == 0)
 434       continue;
 435     resolveRelocation(RE, Value);
 436   }
 437 }
 438
 439 void RuntimeDyldImpl::resolveExternalSymbols() {
 440   StringMap<RelocationList>::iterator i = ExternalSymbolRelocations.begin(),
 441                                       e = ExternalSymbolRelocations.end();
 442   for (; i != e; i++) {
 443     StringRef Name = i->first();
 444     RelocationList &Relocs = i->second;
 445     SymbolTableMap::const_iterator Loc = GlobalSymbolTable.find(Name);
 446     if (Loc == GlobalSymbolTable.end()) {
 447       if (Name.size() == 0) {
 448         // This is an absolute symbol, use an address of zero.
 449         DEBUG(dbgs() << "Resolving absolute relocations." << "\n");
 450         resolveRelocationList(Relocs, 0);
 451       } else {
 452         // This is an external symbol, try to get its address from
 453         // MemoryManager.
 454         uint8_t *Addr = (uint8_t*) MemMgr->getPointerToNamedFunction(Name.data(),
 455                                                                    true);
 456         DEBUG(dbgs() << "Resolving relocations Name: " << Name
 457                 << "\t" << format("%p", Addr)
 458                 << "\n");
 459         resolveRelocationList(Relocs, (uintptr_t)Addr);
 460       }
 461     } else {
 462       report_fatal_error("Expected external symbol");
 463     }
 464   }
 465 }
 466
 467
 468 //===----------------------------------------------------------------------===//
 469 // RuntimeDyld class implementation
 470 RuntimeDyld::RuntimeDyld(RTDyldMemoryManager *mm) {
 471   // FIXME: There's a potential issue lurking here if a single instance of
 472   // RuntimeDyld is used to load multiple objects.  The current implementation
 473   // associates a single memory manager with a RuntimeDyld instance.  Even
 474   // though the public class spawns a new 'impl' instance for each load,
 475   // they share a single memory manager.  This can become a problem when page
 476   // permissions are applied.
 477   Dyld = 0;
 478   MM = mm;
 479 }
 480
 481 RuntimeDyld::~RuntimeDyld() {
 482   delete Dyld;
 483 }
 484
 485 ObjectImage *RuntimeDyld::loadObject(ObjectBuffer *InputBuffer) {
 486   if (!Dyld) {
 487     sys::LLVMFileType type = sys::IdentifyFileType(
 488             InputBuffer->getBufferStart(),
 489             static_cast<unsigned>(InputBuffer->getBufferSize()));
 490     switch (type) {
 491       case sys::ELF_Relocatable_FileType:
 492       case sys::ELF_Executable_FileType:
 493       case sys::ELF_SharedObject_FileType:
 494       case sys::ELF_Core_FileType:
 495         Dyld = new RuntimeDyldELF(MM);
 496         break;
 497       case sys::Mach_O_Object_FileType:
 498       case sys::Mach_O_Executable_FileType:
 499       case sys::Mach_O_FixedVirtualMemorySharedLib_FileType:
 500       case sys::Mach_O_Core_FileType:
 501       case sys::Mach_O_PreloadExecutable_FileType:
 502       case sys::Mach_O_DynamicallyLinkedSharedLib_FileType:
 503       case sys::Mach_O_DynamicLinker_FileType:
 504       case sys::Mach_O_Bundle_FileType:
 505       case sys::Mach_O_DynamicallyLinkedSharedLibStub_FileType:
 506       case sys::Mach_O_DSYMCompanion_FileType:
 507         Dyld = new RuntimeDyldMachO(MM);
 508         break;
 509       case sys::Unknown_FileType:
 510       case sys::Bitcode_FileType:
 511       case sys::Archive_FileType:
 512       case sys::COFF_FileType:
 513         report_fatal_error("Incompatible object format!");
 514     }
 515   } else {
 516     if (!Dyld->isCompatibleFormat(InputBuffer))
 517       report_fatal_error("Incompatible object format!");
 518   }
 519
 520   return Dyld->loadObject(InputBuffer);
 521 }
 522
 523 void *RuntimeDyld::getSymbolAddress(StringRef Name) {
 524   return Dyld->getSymbolAddress(Name);
 525 }
 526
 527 uint64_t RuntimeDyld::getSymbolLoadAddress(StringRef Name) {
 528   return Dyld->getSymbolLoadAddress(Name);
 529 }
 530
 531 void RuntimeDyld::resolveRelocations() {
 532   Dyld->resolveRelocations();
 533 }
 534
 535 void RuntimeDyld::reassignSectionAddress(unsigned SectionID,
 536                                          uint64_t Addr) {
 537   Dyld->reassignSectionAddress(SectionID, Addr);
 538 }
 539
 540 void RuntimeDyld::mapSectionAddress(const void *LocalAddress,
 541                                     uint64_t TargetAddress) {
 542   Dyld->mapSectionAddress(LocalAddress, TargetAddress);
 543 }
 544
 545 StringRef RuntimeDyld::getErrorString() {
 546   return Dyld->getErrorString();
 547 }
 548
 549 StringRef RuntimeDyld::getEHFrameSection() {
 550   return Dyld->getEHFrameSection();
 551 }
 552
 553 } // end namespace llvm