Changes :- a) Functionality in InputGraph to insert Input elements at any position b) Functionality in the Resolver to use nextFile c) Move the functionality of assigning file ordinals to InputGraph d) Changes all inputs to MemoryBuffers e) Remove LinkerInput, InputFiles, ReaderArchive llvm-svn: 192081
990 lines
36 KiB
C++
990 lines
36 KiB
C++
//===- lib/ReaderWriter/PECOFF/WriterPECOFF.cpp ---------------------------===//
|
|
//
|
|
// The LLVM Linker
|
|
//
|
|
// This file is distributed under the University of Illinois Open Source
|
|
// License. See LICENSE.TXT for details.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
///
|
|
/// \file
|
|
///
|
|
/// PE/COFF file consists of DOS Header, PE Header, COFF Header and Section
|
|
/// Tables followed by raw section data.
|
|
///
|
|
/// This writer is reponsible for writing Core Linker results to an Windows
|
|
/// executable file. Currently it can only output ".text" section; other
|
|
/// sections including the symbol table are silently ignored.
|
|
///
|
|
/// This writer currently supports 32 bit PE/COFF for x86 processor only.
|
|
///
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#define DEBUG_TYPE "WriterPECOFF"
|
|
|
|
#include <map>
|
|
#include <time.h>
|
|
#include <vector>
|
|
|
|
#include "Atoms.h"
|
|
|
|
#include "lld/Core/DefinedAtom.h"
|
|
#include "lld/Core/File.h"
|
|
#include "lld/ReaderWriter/AtomLayout.h"
|
|
#include "lld/ReaderWriter/PECOFFLinkingContext.h"
|
|
#include "lld/ReaderWriter/Writer.h"
|
|
|
|
#include "llvm/ADT/ArrayRef.h"
|
|
#include "llvm/Object/COFF.h"
|
|
#include "llvm/Support/COFF.h"
|
|
#include "llvm/Support/Debug.h"
|
|
#include "llvm/Support/Endian.h"
|
|
#include "llvm/Support/ErrorHandling.h"
|
|
#include "llvm/Support/ErrorOr.h"
|
|
#include "llvm/Support/FileOutputBuffer.h"
|
|
#include "llvm/Support/Format.h"
|
|
|
|
using llvm::support::ulittle16_t;
|
|
using llvm::support::ulittle32_t;
|
|
|
|
namespace lld {
|
|
namespace pecoff {
|
|
|
|
namespace {
|
|
class SectionChunk;
|
|
|
|
// Page size of x86 processor. Some data needs to be aligned at page boundary
|
|
// when loaded into memory.
|
|
const int PAGE_SIZE = 4096;
|
|
|
|
// Disk sector size. Some data needs to be aligned at disk sector boundary in
|
|
// file.
|
|
const int SECTOR_SIZE = 512;
|
|
|
|
// The address of the executable when loaded into memory.
|
|
const int32_t IMAGE_BASE = 0x400000;
|
|
|
|
/// A Chunk is an abstrace contiguous range in an output file.
|
|
class Chunk {
|
|
public:
|
|
enum Kind {
|
|
kindHeader,
|
|
kindSection,
|
|
kindDataDirectory
|
|
};
|
|
|
|
explicit Chunk(Kind kind) : _kind(kind), _size(0), _align(1) {}
|
|
virtual ~Chunk() {};
|
|
virtual void write(uint8_t *fileBuffer) = 0;
|
|
|
|
virtual uint64_t fileOffset() const { return _fileOffset; }
|
|
virtual uint64_t size() const { return _size; }
|
|
virtual uint64_t align() const { return _align; }
|
|
|
|
virtual void setFileOffset(uint64_t fileOffset) {
|
|
_fileOffset = fileOffset;
|
|
}
|
|
|
|
Kind getKind() const { return _kind; }
|
|
|
|
protected:
|
|
Kind _kind;
|
|
uint64_t _size;
|
|
uint64_t _fileOffset;
|
|
uint64_t _align;
|
|
};
|
|
|
|
/// A HeaderChunk is an abstract class to represent a file header for
|
|
/// PE/COFF. The data in the header chunk is metadata about program and will
|
|
/// be consumed by the windows loader. HeaderChunks are not mapped to memory
|
|
/// when executed.
|
|
class HeaderChunk : public Chunk {
|
|
public:
|
|
HeaderChunk() : Chunk(kindHeader) {}
|
|
|
|
static bool classof(const Chunk *c) { return c->getKind() == kindHeader; }
|
|
};
|
|
|
|
/// A DOSStubChunk represents the DOS compatible header at the beginning
|
|
/// of PE/COFF files.
|
|
class DOSStubChunk : public HeaderChunk {
|
|
public:
|
|
DOSStubChunk() : HeaderChunk() {
|
|
// Make the DOS stub occupy the first 128 bytes of an exe. Technically
|
|
// this can be as small as 64 bytes, but GNU binutil's objdump cannot
|
|
// parse such irregular header.
|
|
_size = 128;
|
|
|
|
// A DOS stub is usually a small valid DOS program that prints out a message
|
|
// "This program requires Microsoft Windows" to help user who accidentally
|
|
// run a Windows executable on DOS. That's not a technical requirement, so
|
|
// we don't bother to emit such code, at least for now. We simply fill the
|
|
// DOS stub with null bytes.
|
|
std::memset(&_dosHeader, 0, sizeof(_dosHeader));
|
|
|
|
_dosHeader.Magic = 'M' | ('Z' << 8);
|
|
_dosHeader.AddressOfNewExeHeader = _size;
|
|
}
|
|
|
|
virtual void write(uint8_t *fileBuffer) {
|
|
std::memcpy(fileBuffer, &_dosHeader, sizeof(_dosHeader));
|
|
}
|
|
|
|
private:
|
|
llvm::object::dos_header _dosHeader;
|
|
};
|
|
|
|
/// A PEHeaderChunk represents PE header including COFF header.
|
|
class PEHeaderChunk : public HeaderChunk {
|
|
public:
|
|
explicit PEHeaderChunk(const PECOFFLinkingContext &context) : HeaderChunk() {
|
|
// Set the size of the chunk and initialize the header with null bytes.
|
|
_size = sizeof(llvm::COFF::PEMagic) + sizeof(_coffHeader)
|
|
+ sizeof(_peHeader);
|
|
std::memset(&_coffHeader, 0, sizeof(_coffHeader));
|
|
std::memset(&_peHeader, 0, sizeof(_peHeader));
|
|
|
|
_coffHeader.Machine = context.getMachineType();
|
|
_coffHeader.TimeDateStamp = time(NULL);
|
|
|
|
// The size of PE header including optional data directory is always 224.
|
|
_coffHeader.SizeOfOptionalHeader = 224;
|
|
|
|
// Attributes of the executable.
|
|
uint16_t characteristics = llvm::COFF::IMAGE_FILE_32BIT_MACHINE |
|
|
llvm::COFF::IMAGE_FILE_EXECUTABLE_IMAGE;
|
|
if (context.getLargeAddressAware())
|
|
characteristics |= llvm::COFF::IMAGE_FILE_LARGE_ADDRESS_AWARE;
|
|
if (context.getSwapRunFromCD())
|
|
characteristics |= llvm::COFF::IMAGE_FILE_REMOVABLE_RUN_FROM_SWAP;
|
|
if (context.getSwapRunFromNet())
|
|
characteristics |= llvm::COFF::IMAGE_FILE_NET_RUN_FROM_SWAP;
|
|
if (!context.getBaseRelocationEnabled())
|
|
characteristics |= llvm::COFF::IMAGE_FILE_RELOCS_STRIPPED;
|
|
|
|
_coffHeader.Characteristics = characteristics;
|
|
|
|
// 0x10b indicates a normal PE32 executable. For PE32+ it should be 0x20b.
|
|
_peHeader.Magic = 0x10b;
|
|
|
|
// The address of entry point relative to ImageBase. Windows executable
|
|
// usually starts at address 0x401000.
|
|
_peHeader.AddressOfEntryPoint = 0x1000;
|
|
|
|
// The address of the executable when loaded into memory. The default for
|
|
// DLLs is 0x10000000. The default for executables is 0x400000.
|
|
_peHeader.ImageBase = context.getBaseAddress();
|
|
|
|
// Sections should be page-aligned when loaded into memory, which is 4KB on
|
|
// x86.
|
|
_peHeader.SectionAlignment = context.getSectionAlignment();
|
|
|
|
// Sections in an executable file on disk should be sector-aligned (512 byte).
|
|
_peHeader.FileAlignment = SECTOR_SIZE;
|
|
|
|
// The version number of the resultant executable/DLL. The number is purely
|
|
// informative, and neither the linker nor the loader won't use it. User can
|
|
// set the value using /version command line option. Default is 0.0.
|
|
PECOFFLinkingContext::Version imageVersion = context.getImageVersion();
|
|
_peHeader.MajorImageVersion = imageVersion.majorVersion;
|
|
_peHeader.MinorImageVersion = imageVersion.minorVersion;
|
|
|
|
// The required Windows version number. This is the internal version and
|
|
// shouldn't be confused with product name. Windows 7 is version 6.1 and
|
|
// Windows 8 is 6.2, for example.
|
|
PECOFFLinkingContext::Version minOSVersion = context.getMinOSVersion();
|
|
_peHeader.MajorOperatingSystemVersion = minOSVersion.majorVersion;
|
|
_peHeader.MinorOperatingSystemVersion = minOSVersion.minorVersion;
|
|
_peHeader.MajorSubsystemVersion = minOSVersion.majorVersion;
|
|
_peHeader.MinorSubsystemVersion = minOSVersion.minorVersion;
|
|
|
|
_peHeader.Subsystem = context.getSubsystem();
|
|
|
|
// Despite its name, DLL characteristics field has meaning both for
|
|
// executables and DLLs. We are not very sure if the following bits must
|
|
// be set, but regular binaries seem to have these bits, so we follow
|
|
// them.
|
|
uint16_t dllCharacteristics =
|
|
llvm::COFF::IMAGE_DLL_CHARACTERISTICS_NO_SEH;
|
|
if (context.isTerminalServerAware())
|
|
dllCharacteristics |=
|
|
llvm::COFF::IMAGE_DLL_CHARACTERISTICS_TERMINAL_SERVER_AWARE;
|
|
if (context.isNxCompat())
|
|
dllCharacteristics |= llvm::COFF::IMAGE_DLL_CHARACTERISTICS_NX_COMPAT;
|
|
if (context.getDynamicBaseEnabled())
|
|
dllCharacteristics |= llvm::COFF::IMAGE_DLL_CHARACTERISTICS_DYNAMIC_BASE;
|
|
if (!context.getAllowBind())
|
|
dllCharacteristics |= llvm::COFF::IMAGE_DLL_CHARACTERISTICS_NO_BIND;
|
|
if (!context.getAllowIsolation())
|
|
dllCharacteristics |= llvm::COFF::IMAGE_DLL_CHARACTERISTICS_NO_ISOLATION;
|
|
_peHeader.DLLCharacteristics = dllCharacteristics;
|
|
|
|
_peHeader.SizeOfStackReserve = context.getStackReserve();
|
|
_peHeader.SizeOfStackCommit = context.getStackCommit();
|
|
_peHeader.SizeOfHeapReserve = context.getHeapReserve();
|
|
_peHeader.SizeOfHeapCommit = context.getHeapCommit();
|
|
|
|
// The number of data directory entries. We always have 16 entries.
|
|
_peHeader.NumberOfRvaAndSize = 16;
|
|
}
|
|
|
|
virtual void write(uint8_t *fileBuffer) {
|
|
fileBuffer += fileOffset();
|
|
std::memcpy(fileBuffer, llvm::COFF::PEMagic, sizeof(llvm::COFF::PEMagic));
|
|
fileBuffer += sizeof(llvm::COFF::PEMagic);
|
|
std::memcpy(fileBuffer, &_coffHeader, sizeof(_coffHeader));
|
|
fileBuffer += sizeof(_coffHeader);
|
|
std::memcpy(fileBuffer, &_peHeader, sizeof(_peHeader));
|
|
}
|
|
|
|
virtual void setSizeOfHeaders(uint64_t size) {
|
|
// Must be multiple of FileAlignment.
|
|
_peHeader.SizeOfHeaders = llvm::RoundUpToAlignment(size, SECTOR_SIZE);
|
|
}
|
|
|
|
virtual void setSizeOfCode(uint64_t size) {
|
|
_peHeader.SizeOfCode = size;
|
|
}
|
|
|
|
virtual void setSizeOfInitializedData(uint64_t size) {
|
|
_peHeader.SizeOfInitializedData = size;
|
|
}
|
|
|
|
virtual void setSizeOfUninitializedData(uint64_t size) {
|
|
_peHeader.SizeOfUninitializedData = size;
|
|
}
|
|
|
|
virtual void setNumberOfSections(uint32_t num) {
|
|
_coffHeader.NumberOfSections = num;
|
|
}
|
|
|
|
virtual void setBaseOfCode(uint32_t rva) { _peHeader.BaseOfCode = rva; }
|
|
|
|
virtual void setBaseOfData(uint32_t rva) { _peHeader.BaseOfData = rva; }
|
|
|
|
virtual void setSizeOfImage(uint32_t size) { _peHeader.SizeOfImage = size; }
|
|
|
|
virtual void setAddressOfEntryPoint(uint32_t address) {
|
|
_peHeader.AddressOfEntryPoint = address;
|
|
}
|
|
|
|
private:
|
|
llvm::object::coff_file_header _coffHeader;
|
|
llvm::object::pe32_header _peHeader;
|
|
};
|
|
|
|
/// A SectionHeaderTableChunk represents Section Table Header of PE/COFF
|
|
/// format, which is a list of section headers.
|
|
class SectionHeaderTableChunk : public HeaderChunk {
|
|
public:
|
|
SectionHeaderTableChunk() : HeaderChunk() {}
|
|
void addSection(SectionChunk *chunk);
|
|
virtual uint64_t size() const;
|
|
virtual void write(uint8_t *fileBuffer);
|
|
|
|
private:
|
|
std::vector<SectionChunk *> _sections;
|
|
};
|
|
|
|
/// An AtomChunk represents a section containing atoms.
|
|
class AtomChunk : public Chunk {
|
|
public:
|
|
virtual void write(uint8_t *fileBuffer) {
|
|
for (const auto *layout : _atomLayouts) {
|
|
const DefinedAtom *atom = cast<DefinedAtom>(layout->_atom);
|
|
ArrayRef<uint8_t> rawContent = atom->rawContent();
|
|
std::memcpy(fileBuffer + layout->_fileOffset, rawContent.data(),
|
|
rawContent.size());
|
|
}
|
|
}
|
|
|
|
/// Add all atoms to the given map. This data will be used to do relocation.
|
|
void buildAtomToVirtualAddr(std::map<const Atom *, uint64_t> &atomRva) {
|
|
for (const auto *layout : _atomLayouts)
|
|
atomRva[layout->_atom] = layout->_virtualAddr;
|
|
}
|
|
|
|
void applyRelocations(uint8_t *fileBuffer,
|
|
std::map<const Atom *, uint64_t> &atomRva,
|
|
uint64_t imageBaseAddress) {
|
|
for (const auto *layout : _atomLayouts) {
|
|
const DefinedAtom *atom = cast<DefinedAtom>(layout->_atom);
|
|
for (const Reference *ref : *atom) {
|
|
auto relocSite = reinterpret_cast<ulittle32_t *>(
|
|
fileBuffer + layout->_fileOffset + ref->offsetInAtom());
|
|
uint64_t targetAddr = atomRva[ref->target()];
|
|
// Also account for whatever offset is already stored at the relocation site.
|
|
targetAddr += *relocSite;
|
|
|
|
// Skip if this reference is not for relocation.
|
|
if (ref->kind() < lld::Reference::kindTargetLow)
|
|
continue;
|
|
|
|
switch (ref->kind()) {
|
|
case llvm::COFF::IMAGE_REL_I386_ABSOLUTE:
|
|
// This relocation is no-op.
|
|
break;
|
|
case llvm::COFF::IMAGE_REL_I386_DIR32:
|
|
// Set target's 32-bit VA.
|
|
*relocSite = targetAddr + imageBaseAddress;
|
|
break;
|
|
case llvm::COFF::IMAGE_REL_I386_DIR32NB:
|
|
// Set target's 32-bit RVA.
|
|
*relocSite = targetAddr;
|
|
break;
|
|
case llvm::COFF::IMAGE_REL_I386_REL32: {
|
|
// Set 32-bit relative address of the target. This relocation is
|
|
// usually used for relative branch or call instruction.
|
|
uint32_t disp = atomRva[atom] + ref->offsetInAtom() + 4;
|
|
*relocSite = targetAddr - disp;
|
|
break;
|
|
}
|
|
default:
|
|
llvm_unreachable("Unsupported relocation kind");
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
/// Print atom VAs. Used only for debugging.
|
|
void printAtomAddresses(uint64_t baseAddr) {
|
|
for (const auto *layout : _atomLayouts) {
|
|
const DefinedAtom *atom = cast<DefinedAtom>(layout->_atom);
|
|
uint64_t addr = layout->_virtualAddr;
|
|
llvm::dbgs() << llvm::format("0x%08llx: ", addr + baseAddr)
|
|
<< (atom->name().empty() ? "(anonymous)" : atom->name())
|
|
<< "\n";
|
|
}
|
|
}
|
|
|
|
/// List all virtual addresses (and not relative virtual addresses) that need
|
|
/// to be fixed up if image base is relocated. The only relocation type that
|
|
/// needs to be fixed is DIR32 on i386. REL32 is not (and should not be)
|
|
/// fixed up because it's PC-relative.
|
|
void addBaseRelocations(std::vector<uint64_t> &relocSites) {
|
|
// TODO: llvm-objdump doesn't support parsing the base relocation table, so
|
|
// we can't really test this at the moment. As a temporary solution, we
|
|
// should output debug messages with atom names and addresses so that we
|
|
// can inspect relocations, and fix the tests (base-reloc.test, maybe
|
|
// others) to use those messages.
|
|
for (const auto *layout : _atomLayouts) {
|
|
const DefinedAtom *atom = cast<DefinedAtom>(layout->_atom);
|
|
for (const Reference *ref : *atom)
|
|
if (ref->kind() == llvm::COFF::IMAGE_REL_I386_DIR32)
|
|
relocSites.push_back(layout->_virtualAddr + ref->offsetInAtom());
|
|
}
|
|
}
|
|
|
|
// Set the file offset of the beginning of this section.
|
|
virtual void setFileOffset(uint64_t fileOffset) {
|
|
Chunk::setFileOffset(fileOffset);
|
|
for (AtomLayout *layout : _atomLayouts)
|
|
layout->_fileOffset += fileOffset;
|
|
}
|
|
|
|
uint64_t getSectionRva() {
|
|
assert(_atomLayouts.size() > 0);
|
|
return _atomLayouts[0]->_virtualAddr;
|
|
}
|
|
|
|
virtual void setVirtualAddress(uint32_t rva) {
|
|
for (AtomLayout *layout : _atomLayouts)
|
|
layout->_virtualAddr += rva;
|
|
}
|
|
|
|
uint64_t getAtomVirtualAddress(StringRef name) {
|
|
for (auto atomLayout : _atomLayouts)
|
|
if (atomLayout->_atom->name() == name)
|
|
return atomLayout->_virtualAddr;
|
|
return 0;
|
|
}
|
|
|
|
static bool classof(const Chunk *c) {
|
|
Kind kind = c->getKind();
|
|
return kind == kindSection || kind == kindDataDirectory;
|
|
}
|
|
|
|
protected:
|
|
AtomChunk(Kind kind) : Chunk(kind) {}
|
|
std::vector<AtomLayout *> _atomLayouts;
|
|
};
|
|
|
|
/// A DataDirectoryChunk represents data directory entries that follows the PE
|
|
/// header in the output file. An entry consists of an 8 byte field that
|
|
/// indicates a relative virtual address (the starting address of the entry data
|
|
/// in memory) and 8 byte entry data size.
|
|
class DataDirectoryChunk : public AtomChunk {
|
|
public:
|
|
DataDirectoryChunk(const File &linkedFile)
|
|
: AtomChunk(kindDataDirectory), _file(linkedFile) {
|
|
// Extract atoms from the linked file and append them to this section.
|
|
for (const DefinedAtom *atom : linkedFile.defined()) {
|
|
if (atom->contentType() == DefinedAtom::typeDataDirectoryEntry) {
|
|
uint64_t offset = atom->ordinal() * sizeof(llvm::object::data_directory);
|
|
_atomLayouts.push_back(new (_alloc) AtomLayout(atom, offset, offset));
|
|
}
|
|
}
|
|
}
|
|
|
|
virtual uint64_t size() const {
|
|
return sizeof(llvm::object::data_directory) * 16;
|
|
}
|
|
|
|
void setBaseRelocField(uint32_t addr, uint32_t size) {
|
|
auto *atom = new (_alloc) coff::COFFDataDirectoryAtom(
|
|
_file, llvm::COFF::DataDirectoryIndex::BASE_RELOCATION_TABLE, size,
|
|
addr);
|
|
uint64_t offset = atom->ordinal() * sizeof(llvm::object::data_directory);
|
|
_atomLayouts.push_back(new (_alloc) AtomLayout(atom, offset, offset));
|
|
}
|
|
|
|
virtual void write(uint8_t *fileBuffer) {
|
|
for (const AtomLayout *layout : _atomLayouts) {
|
|
if (!layout)
|
|
continue;
|
|
ArrayRef<uint8_t> content = static_cast<const DefinedAtom *>(layout->_atom)->rawContent();
|
|
std::memcpy(fileBuffer + layout->_fileOffset, content.data(), content.size());
|
|
}
|
|
}
|
|
|
|
private:
|
|
const File &_file;
|
|
mutable llvm::BumpPtrAllocator _alloc;
|
|
};
|
|
|
|
/// A SectionChunk represents a section containing atoms. It consists of a
|
|
/// section header that to be written to PECOFF header and atoms which to be
|
|
/// written to the raw data section.
|
|
class SectionChunk : public AtomChunk {
|
|
public:
|
|
/// Returns the size of the section on disk. The returned value is multiple
|
|
/// of disk sector, so the size may include the null padding at the end of
|
|
/// section.
|
|
virtual uint64_t size() const {
|
|
return llvm::RoundUpToAlignment(_size, _align);
|
|
}
|
|
|
|
virtual uint64_t rawSize() const {
|
|
return _size;
|
|
}
|
|
|
|
// Set the file offset of the beginning of this section.
|
|
virtual void setFileOffset(uint64_t fileOffset) {
|
|
AtomChunk::setFileOffset(fileOffset);
|
|
_sectionHeader.PointerToRawData = fileOffset;
|
|
}
|
|
|
|
virtual void setVirtualAddress(uint32_t rva) {
|
|
_sectionHeader.VirtualAddress = rva;
|
|
AtomChunk::setVirtualAddress(rva);
|
|
}
|
|
|
|
virtual uint32_t getVirtualAddress() { return _sectionHeader.VirtualAddress; }
|
|
|
|
virtual llvm::object::coff_section &getSectionHeader() {
|
|
// Fix up section size before returning it. VirtualSize should be the size
|
|
// of the actual content, and SizeOfRawData should be aligned to the section
|
|
// alignment.
|
|
_sectionHeader.VirtualSize = _size;
|
|
_sectionHeader.SizeOfRawData = size();
|
|
return _sectionHeader;
|
|
}
|
|
|
|
ulittle32_t getSectionCharacteristics() {
|
|
return _sectionHeader.Characteristics;
|
|
}
|
|
|
|
void appendAtom(const DefinedAtom *atom) {
|
|
// Atom may have to be at a proper alignment boundary. If so, move the
|
|
// pointer to make a room after the last atom before adding new one.
|
|
_size = llvm::RoundUpToAlignment(_size, 1 << atom->alignment().powerOf2);
|
|
|
|
// Create an AtomLayout and move the current pointer.
|
|
auto *layout = new (_alloc) AtomLayout(atom, _size, _size);
|
|
_atomLayouts.push_back(layout);
|
|
_size += atom->size();
|
|
}
|
|
|
|
static bool classof(const Chunk *c) { return c->getKind() == kindSection; }
|
|
|
|
protected:
|
|
SectionChunk(StringRef sectionName, uint32_t characteristics)
|
|
: AtomChunk(kindSection),
|
|
_sectionHeader(createSectionHeader(sectionName, characteristics)) {
|
|
// The section should be aligned to disk sector.
|
|
_align = SECTOR_SIZE;
|
|
}
|
|
|
|
void buildContents(const File &linkedFile,
|
|
bool (*isEligible)(const DefinedAtom *)) {
|
|
// Extract atoms from the linked file and append them to this section.
|
|
for (const DefinedAtom *atom : linkedFile.defined()) {
|
|
assert(atom->sectionChoice() == DefinedAtom::sectionBasedOnContent);
|
|
if (isEligible(atom))
|
|
appendAtom(atom);
|
|
}
|
|
|
|
// Now that we have a list of atoms that to be written in this section,
|
|
// and we know the size of the section. Let's write them to the section
|
|
// header. VirtualSize should be the size of the actual content, and
|
|
// SizeOfRawData should be aligned to the section alignment.
|
|
_sectionHeader.VirtualSize = _size;
|
|
_sectionHeader.SizeOfRawData = size();
|
|
}
|
|
|
|
private:
|
|
llvm::object::coff_section
|
|
createSectionHeader(StringRef sectionName, uint32_t characteristics) const {
|
|
llvm::object::coff_section header;
|
|
|
|
// Section name equal to or shorter than 8 byte fits in the section
|
|
// header. Longer names should be stored to string table, which is not
|
|
// implemented yet.
|
|
if (sizeof(header.Name) < sectionName.size())
|
|
llvm_unreachable("Cannot handle section name longer than 8 byte");
|
|
|
|
// Name field must be NUL-padded. If the name is exactly 8 byte long,
|
|
// there's no terminating NUL.
|
|
std::memset(header.Name, 0, sizeof(header.Name));
|
|
std::strncpy(header.Name, sectionName.data(), sizeof(header.Name));
|
|
|
|
header.VirtualSize = 0;
|
|
header.VirtualAddress = 0;
|
|
header.SizeOfRawData = 0;
|
|
header.PointerToRawData = 0;
|
|
header.PointerToRelocations = 0;
|
|
header.PointerToLinenumbers = 0;
|
|
header.NumberOfRelocations = 0;
|
|
header.NumberOfLinenumbers = 0;
|
|
header.Characteristics = characteristics;
|
|
return header;
|
|
}
|
|
|
|
llvm::object::coff_section _sectionHeader;
|
|
mutable llvm::BumpPtrAllocator _alloc;
|
|
};
|
|
|
|
void SectionHeaderTableChunk::addSection(SectionChunk *chunk) {
|
|
_sections.push_back(chunk);
|
|
}
|
|
|
|
uint64_t SectionHeaderTableChunk::size() const {
|
|
return _sections.size() * sizeof(llvm::object::coff_section);
|
|
}
|
|
|
|
void SectionHeaderTableChunk::write(uint8_t *fileBuffer) {
|
|
uint64_t offset = 0;
|
|
fileBuffer += fileOffset();
|
|
for (const auto &chunk : _sections) {
|
|
const llvm::object::coff_section &header = chunk->getSectionHeader();
|
|
std::memcpy(fileBuffer + offset, &header, sizeof(header));
|
|
offset += sizeof(header);
|
|
}
|
|
}
|
|
|
|
// \brief A TextSectionChunk represents a .text section.
|
|
class TextSectionChunk : public SectionChunk {
|
|
public:
|
|
TextSectionChunk(const File &linkedFile)
|
|
: SectionChunk(".text", characteristics) {
|
|
buildContents(linkedFile, [](const DefinedAtom *atom) {
|
|
return atom->contentType() == DefinedAtom::typeCode;
|
|
});
|
|
}
|
|
|
|
private:
|
|
// When loaded into memory, text section should be readable and executable.
|
|
static const uint32_t characteristics =
|
|
llvm::COFF::IMAGE_SCN_CNT_CODE | llvm::COFF::IMAGE_SCN_MEM_EXECUTE |
|
|
llvm::COFF::IMAGE_SCN_MEM_READ;
|
|
};
|
|
|
|
// \brief A RDataSectionChunk represents a .rdata section.
|
|
class RDataSectionChunk : public SectionChunk {
|
|
public:
|
|
RDataSectionChunk(const File &linkedFile)
|
|
: SectionChunk(".rdata", characteristics) {
|
|
buildContents(linkedFile, [](const DefinedAtom *atom) {
|
|
return (atom->contentType() == DefinedAtom::typeData &&
|
|
atom->permissions() == DefinedAtom::permR__);
|
|
});
|
|
}
|
|
|
|
private:
|
|
// When loaded into memory, rdata section should be readable.
|
|
static const uint32_t characteristics =
|
|
llvm::COFF::IMAGE_SCN_MEM_READ |
|
|
llvm::COFF::IMAGE_SCN_CNT_INITIALIZED_DATA;
|
|
};
|
|
|
|
// \brief A DataSectionChunk represents a .data section.
|
|
class DataSectionChunk : public SectionChunk {
|
|
public:
|
|
DataSectionChunk(const File &linkedFile)
|
|
: SectionChunk(".data", characteristics) {
|
|
buildContents(linkedFile, [](const DefinedAtom *atom) {
|
|
return (atom->contentType() == DefinedAtom::typeData &&
|
|
atom->permissions() == DefinedAtom::permRW_);
|
|
});
|
|
}
|
|
|
|
private:
|
|
// When loaded into memory, data section should be readable and writable.
|
|
static const uint32_t characteristics =
|
|
llvm::COFF::IMAGE_SCN_MEM_READ |
|
|
llvm::COFF::IMAGE_SCN_CNT_INITIALIZED_DATA |
|
|
llvm::COFF::IMAGE_SCN_MEM_WRITE;
|
|
};
|
|
|
|
// \brief A BSSSectionChunk represents a .bss section.
|
|
//
|
|
// Seems link.exe does not emit .bss section but instead merges it with .data
|
|
// section. In COFF, if the size of the section in the header is greater than
|
|
// the size of the actual data on disk, the section on memory is zero-padded.
|
|
// That's why .bss can be merge with .data just by appending it at the end of
|
|
// the section.
|
|
//
|
|
// The executable with .bss is also valid and easier to understand. So we chose
|
|
// to create .bss in LLD.
|
|
class BssSectionChunk : public SectionChunk {
|
|
public:
|
|
// BSS section does not have contents, so write should be no-op.
|
|
virtual void write(uint8_t *fileBuffer) {}
|
|
|
|
virtual llvm::object::coff_section &getSectionHeader() {
|
|
llvm::object::coff_section §ionHeader =
|
|
SectionChunk::getSectionHeader();
|
|
sectionHeader.VirtualSize = 0;
|
|
sectionHeader.PointerToRawData = 0;
|
|
return sectionHeader;
|
|
}
|
|
|
|
BssSectionChunk(const File &linkedFile)
|
|
: SectionChunk(".bss", characteristics) {
|
|
buildContents(linkedFile, [](const DefinedAtom *atom) {
|
|
return atom->contentType() == DefinedAtom::typeZeroFill;
|
|
});
|
|
}
|
|
|
|
private:
|
|
// When loaded into memory, bss section should be readable and writable.
|
|
static const uint32_t characteristics =
|
|
llvm::COFF::IMAGE_SCN_MEM_READ |
|
|
llvm::COFF::IMAGE_SCN_CNT_UNINITIALIZED_DATA |
|
|
llvm::COFF::IMAGE_SCN_MEM_WRITE;
|
|
};
|
|
|
|
/// A BaseRelocAtom represents a base relocation block in ".reloc" section.
|
|
class BaseRelocAtom : public coff::COFFLinkerInternalAtom {
|
|
public:
|
|
BaseRelocAtom(const File &file, std::vector<uint8_t> data)
|
|
: COFFLinkerInternalAtom(file, std::move(data)) {}
|
|
|
|
virtual ContentType contentType() const { return typeData; }
|
|
virtual Alignment alignment() const { return Alignment(2); }
|
|
};
|
|
|
|
/// A BaseRelocChunk represents ".reloc" section.
|
|
///
|
|
/// .reloc section contains a list of addresses. If the PE/COFF loader decides
|
|
/// to load the binary at a memory address different from its preferred base
|
|
/// address, which is specified by ImageBase field in the COFF header, the
|
|
/// loader needs to relocate the binary, so that all the addresses in the binary
|
|
/// point to new locations. The loader will do that by fixing up the addresses
|
|
/// specified by .reloc section.
|
|
///
|
|
/// The executable is almost always loaded at the preferred base address because
|
|
/// it's loaded into an empty address space. The DLL is however an subject of
|
|
/// load-time relocation because it may conflict with other DLLs or the
|
|
/// executable.
|
|
class BaseRelocChunk : public SectionChunk {
|
|
typedef std::vector<std::unique_ptr<Chunk>> ChunkVectorT;
|
|
typedef std::map<uint64_t, std::vector<uint16_t>> PageOffsetT;
|
|
|
|
public:
|
|
BaseRelocChunk(const File &linkedFile)
|
|
: SectionChunk(".reloc", characteristics), _file(linkedFile) {}
|
|
|
|
/// Creates .reloc section content from the other sections. The content of
|
|
/// .reloc is basically a list of relocation sites. The relocation sites are
|
|
/// divided into blocks. Each block represents the base relocation for a 4K
|
|
/// page.
|
|
///
|
|
/// By dividing 32 bit RVAs into blocks, COFF saves disk and memory space for
|
|
/// the base relocation. A block consists of a 32 bit page RVA and 16 bit
|
|
/// relocation entries which represent offsets in the page. That is a more
|
|
/// compact representation than a simple vector of 32 bit RVAs.
|
|
void setContents(ChunkVectorT &chunks) {
|
|
std::vector<uint64_t> relocSites = listRelocSites(chunks);
|
|
PageOffsetT blocks = groupByPage(relocSites);
|
|
for (auto &i : blocks) {
|
|
uint64_t pageAddr = i.first;
|
|
const std::vector<uint16_t> &offsetsInPage = i.second;
|
|
appendAtom(createBaseRelocBlock(_file, pageAddr, offsetsInPage));
|
|
}
|
|
}
|
|
|
|
private:
|
|
// When loaded into memory, reloc section should be readable and writable.
|
|
static const uint32_t characteristics =
|
|
llvm::COFF::IMAGE_SCN_MEM_READ |
|
|
llvm::COFF::IMAGE_SCN_CNT_INITIALIZED_DATA |
|
|
llvm::COFF::IMAGE_SCN_MEM_DISCARDABLE;
|
|
|
|
// Returns a list of RVAs that needs to be relocated if the binary is loaded
|
|
// at an address different from its preferred one.
|
|
std::vector<uint64_t> listRelocSites(ChunkVectorT &chunks) {
|
|
std::vector<uint64_t> ret;
|
|
for (auto &cp : chunks)
|
|
if (SectionChunk *chunk = dyn_cast<SectionChunk>(&*cp))
|
|
chunk->addBaseRelocations(ret);
|
|
return std::move(ret);
|
|
}
|
|
|
|
// Divide the given RVAs into blocks.
|
|
PageOffsetT groupByPage(std::vector<uint64_t> relocSites) {
|
|
PageOffsetT blocks;
|
|
uint64_t mask = static_cast<uint64_t>(PAGE_SIZE) - 1;
|
|
for (uint64_t addr : relocSites)
|
|
blocks[addr & ~mask].push_back(addr & mask);
|
|
return std::move(blocks);
|
|
}
|
|
|
|
// Create the content of a relocation block.
|
|
DefinedAtom *createBaseRelocBlock(const File &file, uint64_t pageAddr,
|
|
const std::vector<uint16_t> &offsets) {
|
|
// Relocation blocks should be padded with IMAGE_REL_I386_ABSOLUTE to be
|
|
// aligned to a DWORD size boundary.
|
|
uint32_t size = llvm::RoundUpToAlignment(sizeof(ulittle32_t) * 2
|
|
+ sizeof(ulittle16_t) * offsets.size(), sizeof(ulittle32_t));
|
|
std::vector<uint8_t> contents(size);
|
|
uint8_t *ptr = &contents[0];
|
|
|
|
// The first four bytes is the page RVA.
|
|
*reinterpret_cast<ulittle32_t *>(ptr) = pageAddr;
|
|
ptr += sizeof(ulittle32_t);
|
|
|
|
// The second four bytes is the size of the block, including the the page
|
|
// RVA and this size field.
|
|
*reinterpret_cast<ulittle32_t *>(ptr) = size;
|
|
ptr += sizeof(ulittle32_t);
|
|
|
|
// The rest of the block consists of offsets in the page.
|
|
for (uint16_t offset : offsets) {
|
|
assert(offset < PAGE_SIZE);
|
|
uint16_t val = (llvm::COFF::IMAGE_REL_BASED_HIGHLOW << 12) | offset;
|
|
*reinterpret_cast<ulittle16_t *>(ptr) = val;
|
|
ptr += sizeof(ulittle16_t);
|
|
}
|
|
return new (_alloc) BaseRelocAtom(file, std::move(contents));
|
|
}
|
|
|
|
mutable llvm::BumpPtrAllocator _alloc;
|
|
const File &_file;
|
|
};
|
|
|
|
} // end anonymous namespace
|
|
|
|
class ExecutableWriter : public Writer {
|
|
public:
|
|
explicit ExecutableWriter(const PECOFFLinkingContext &context)
|
|
: _PECOFFLinkingContext(context), _numSections(0),
|
|
_imageSizeInMemory(PAGE_SIZE), _imageSizeOnDisk(0) {}
|
|
|
|
// Create all chunks that consist of the output file.
|
|
void build(const File &linkedFile) {
|
|
// Create file chunks and add them to the list.
|
|
auto *dosStub = new DOSStubChunk();
|
|
auto *peHeader = new PEHeaderChunk(_PECOFFLinkingContext);
|
|
auto *dataDirectory = new DataDirectoryChunk(linkedFile);
|
|
auto *sectionTable = new SectionHeaderTableChunk();
|
|
auto *text = new TextSectionChunk(linkedFile);
|
|
auto *rdata = new RDataSectionChunk(linkedFile);
|
|
auto *data = new DataSectionChunk(linkedFile);
|
|
auto *bss = new BssSectionChunk(linkedFile);
|
|
BaseRelocChunk *baseReloc = nullptr;
|
|
if (_PECOFFLinkingContext.getBaseRelocationEnabled())
|
|
baseReloc = new BaseRelocChunk(linkedFile);
|
|
|
|
addChunk(dosStub);
|
|
addChunk(peHeader);
|
|
addChunk(dataDirectory);
|
|
addChunk(sectionTable);
|
|
|
|
// Do not add the empty section. Windows loader does not like a section of
|
|
// size zero and rejects such executable.
|
|
if (text->size())
|
|
addSectionChunk(text, sectionTable);
|
|
if (rdata->size())
|
|
addSectionChunk(rdata, sectionTable);
|
|
if (data->size())
|
|
addSectionChunk(data, sectionTable);
|
|
if (bss->size())
|
|
addSectionChunk(bss, sectionTable);
|
|
|
|
// Now that we know the addresses of all defined atoms that needs to be
|
|
// relocated. So we can create the ".reloc" section which contains all the
|
|
// relocation sites.
|
|
if (baseReloc) {
|
|
baseReloc->setContents(_chunks);
|
|
if (baseReloc->size()) {
|
|
addSectionChunk(baseReloc, sectionTable);
|
|
dataDirectory->setBaseRelocField(baseReloc->getSectionRva(),
|
|
baseReloc->rawSize());
|
|
}
|
|
}
|
|
|
|
setImageSizeOnDisk();
|
|
|
|
// Now that we know the size and file offset of sections. Set the file
|
|
// header accordingly.
|
|
peHeader->setSizeOfCode(calcSizeOfCode());
|
|
if (text->size()) {
|
|
peHeader->setBaseOfCode(text->getVirtualAddress());
|
|
}
|
|
if (rdata->size()) {
|
|
peHeader->setBaseOfData(rdata->getVirtualAddress());
|
|
} else if (data->size()) {
|
|
peHeader->setBaseOfData(data->getVirtualAddress());
|
|
}
|
|
peHeader->setSizeOfInitializedData(calcSizeOfInitializedData());
|
|
peHeader->setSizeOfUninitializedData(calcSizeOfUninitializedData());
|
|
peHeader->setNumberOfSections(_numSections);
|
|
peHeader->setSizeOfImage(_imageSizeInMemory);
|
|
|
|
// The combined size of the DOS, PE and section headers including garbage
|
|
// between the end of the header and the beginning of the first section.
|
|
peHeader->setSizeOfHeaders(dosStub->size() + peHeader->size() +
|
|
sectionTable->size() + dataDirectory->size());
|
|
|
|
setAddressOfEntryPoint(text, peHeader);
|
|
}
|
|
|
|
virtual error_code writeFile(const File &linkedFile, StringRef path) {
|
|
this->build(linkedFile);
|
|
|
|
uint64_t totalSize = _chunks.back()->fileOffset() + _chunks.back()->size();
|
|
OwningPtr<llvm::FileOutputBuffer> buffer;
|
|
error_code ec = llvm::FileOutputBuffer::create(
|
|
path, totalSize, buffer, llvm::FileOutputBuffer::F_executable);
|
|
if (ec)
|
|
return ec;
|
|
|
|
for (const auto &chunk : _chunks)
|
|
chunk->write(buffer->getBufferStart());
|
|
applyAllRelocations(buffer->getBufferStart());
|
|
DEBUG(printAllAtomAddresses());
|
|
return buffer->commit();
|
|
}
|
|
|
|
private:
|
|
/// Apply relocations to the output file buffer. This two pass. In the first
|
|
/// pass, we visit all atoms to create a map from atom to its virtual
|
|
/// address. In the second pass, we visit all relocation references to fix
|
|
/// up addresses in the buffer.
|
|
void applyAllRelocations(uint8_t *bufferStart) {
|
|
for (auto &cp : _chunks)
|
|
if (AtomChunk *chunk = dyn_cast<AtomChunk>(&*cp))
|
|
chunk->applyRelocations(bufferStart, atomRva,
|
|
_PECOFFLinkingContext.getBaseAddress());
|
|
}
|
|
|
|
/// Print atom VAs. Used only for debugging.
|
|
void printAllAtomAddresses() {
|
|
for (auto &cp : _chunks)
|
|
if (AtomChunk *chunk = dyn_cast<AtomChunk>(&*cp))
|
|
chunk->printAtomAddresses(_PECOFFLinkingContext.getBaseAddress());
|
|
}
|
|
|
|
void addChunk(Chunk *chunk) {
|
|
_chunks.push_back(std::unique_ptr<Chunk>(chunk));
|
|
}
|
|
|
|
void addSectionChunk(SectionChunk *chunk,
|
|
SectionHeaderTableChunk *table) {
|
|
_chunks.push_back(std::unique_ptr<Chunk>(chunk));
|
|
table->addSection(chunk);
|
|
_numSections++;
|
|
|
|
// Compute and set the starting address of sections when loaded in
|
|
// memory. They are different from positions on disk because sections need
|
|
// to be sector-aligned on disk but page-aligned in memory.
|
|
chunk->setVirtualAddress(_imageSizeInMemory);
|
|
chunk->buildAtomToVirtualAddr(atomRva);
|
|
_imageSizeInMemory = llvm::RoundUpToAlignment(
|
|
_imageSizeInMemory + chunk->size(), PAGE_SIZE);
|
|
}
|
|
|
|
void setImageSizeOnDisk() {
|
|
for (auto &chunk : _chunks) {
|
|
// Compute and set the offset of the chunk in the output file.
|
|
_imageSizeOnDisk = llvm::RoundUpToAlignment(_imageSizeOnDisk,
|
|
chunk->align());
|
|
chunk->setFileOffset(_imageSizeOnDisk);
|
|
_imageSizeOnDisk += chunk->size();
|
|
}
|
|
}
|
|
|
|
void setAddressOfEntryPoint(TextSectionChunk *text, PEHeaderChunk *peHeader) {
|
|
// Find the virtual address of the entry point symbol if any.
|
|
// PECOFF spec says that entry point for dll images is optional, in which
|
|
// case it must be set to 0.
|
|
if (_PECOFFLinkingContext.entrySymbolName().empty() &&
|
|
_PECOFFLinkingContext.getImageType()
|
|
== PECOFFLinkingContext::IMAGE_DLL) {
|
|
peHeader->setAddressOfEntryPoint(0);
|
|
} else {
|
|
uint64_t entryPointAddress = text->getAtomVirtualAddress(
|
|
_PECOFFLinkingContext.entrySymbolName());
|
|
if (entryPointAddress != 0)
|
|
peHeader->setAddressOfEntryPoint(entryPointAddress);
|
|
}
|
|
}
|
|
|
|
uint64_t calcSectionSize(llvm::COFF::SectionCharacteristics sectionType) {
|
|
uint64_t ret = 0;
|
|
for (auto &cp : _chunks)
|
|
if (SectionChunk *chunk = dyn_cast<SectionChunk>(&*cp))
|
|
if (chunk->getSectionCharacteristics() & sectionType)
|
|
ret += chunk->size();
|
|
return ret;
|
|
}
|
|
|
|
uint64_t calcSizeOfInitializedData() {
|
|
return calcSectionSize(llvm::COFF::IMAGE_SCN_CNT_INITIALIZED_DATA);
|
|
}
|
|
|
|
uint64_t calcSizeOfUninitializedData() {
|
|
return calcSectionSize(llvm::COFF::IMAGE_SCN_CNT_UNINITIALIZED_DATA);
|
|
}
|
|
|
|
uint64_t calcSizeOfCode() {
|
|
return calcSectionSize(llvm::COFF::IMAGE_SCN_CNT_CODE);
|
|
}
|
|
|
|
std::vector<std::unique_ptr<Chunk>> _chunks;
|
|
const PECOFFLinkingContext &_PECOFFLinkingContext;
|
|
uint32_t _numSections;
|
|
|
|
// The size of the image in memory. This is initialized with PAGE_SIZE, as the
|
|
// first page starting at ImageBase is usually left unmapped. IIUC there's no
|
|
// technical reason to do so, but we'll follow that convention so that we
|
|
// don't produce odd-looking binary.
|
|
uint32_t _imageSizeInMemory;
|
|
|
|
// The size of the image on disk. This is basically the sum of all chunks in
|
|
// the output file with paddings between them.
|
|
uint32_t _imageSizeOnDisk;
|
|
|
|
// The map from defined atoms to its RVAs. Will be used for relocation.
|
|
std::map<const Atom *, uint64_t> atomRva;
|
|
};
|
|
|
|
} // end namespace pecoff
|
|
|
|
std::unique_ptr<Writer> createWriterPECOFF(const PECOFFLinkingContext &info) {
|
|
return std::unique_ptr<Writer>(new pecoff::ExecutableWriter(info));
|
|
}
|
|
|
|
} // end namespace lld
|