This patch makes File::parse() multi-thread safe. If one thread is running File::parse(), other threads will block if they try to call the same method. File::parse() is idempotent, so you can safely call multiple times. With this change, we don't have to wait for all worker threads to finish in Driver::link(). Previously, Driver::link() calls TaskGroup::sync() to wait for all threads running File::parse(). This was not ideal because we couldn't start the resolver until we parse all files. This patch increase parallelism by making Driver::link() to not wait for worker threads. The resolver calls parse() to make sure that the file being read has been parsed, and then uses the file. In this approach, the resolver can run with the parser threads in parallel. http://reviews.llvm.org/D6994 llvm-svn: 226281
501 lines
17 KiB
C++
501 lines
17 KiB
C++
//===- Core/Resolver.cpp - Resolves Atom References -----------------------===//
|
|
//
|
|
// The LLVM Linker
|
|
//
|
|
// This file is distributed under the University of Illinois Open Source
|
|
// License. See LICENSE.TXT for details.
|
|
//
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
#include "lld/Core/Atom.h"
|
|
#include "lld/Core/ArchiveLibraryFile.h"
|
|
#include "lld/Core/File.h"
|
|
#include "lld/Core/Instrumentation.h"
|
|
#include "lld/Core/LLVM.h"
|
|
#include "lld/Core/LinkingContext.h"
|
|
#include "lld/Core/Resolver.h"
|
|
#include "lld/Core/SharedLibraryFile.h"
|
|
#include "lld/Core/SymbolTable.h"
|
|
#include "lld/Core/UndefinedAtom.h"
|
|
#include "llvm/Support/Debug.h"
|
|
#include "llvm/Support/ErrorHandling.h"
|
|
#include "llvm/Support/Format.h"
|
|
#include "llvm/Support/raw_ostream.h"
|
|
#include <algorithm>
|
|
#include <cassert>
|
|
#include <vector>
|
|
|
|
namespace lld {
|
|
|
|
bool Resolver::handleFile(const File &file) {
|
|
bool undefAdded = false;
|
|
for (const DefinedAtom *atom : file.defined())
|
|
doDefinedAtom(*atom);
|
|
for (const UndefinedAtom *atom : file.undefined())
|
|
if (doUndefinedAtom(*atom))
|
|
undefAdded = true;
|
|
for (const SharedLibraryAtom *atom : file.sharedLibrary())
|
|
doSharedLibraryAtom(*atom);
|
|
for (const AbsoluteAtom *atom : file.absolute())
|
|
doAbsoluteAtom(*atom);
|
|
return undefAdded;
|
|
}
|
|
|
|
void Resolver::forEachUndefines(bool searchForOverrides,
|
|
UndefCallback callback) {
|
|
// Handle normal archives
|
|
unsigned undefineGenCount = 0;
|
|
do {
|
|
undefineGenCount = _symbolTable.size();
|
|
for (const UndefinedAtom *undefAtom : _symbolTable.undefines()) {
|
|
StringRef undefName = undefAtom->name();
|
|
// load for previous undefine may also have loaded this undefine
|
|
if (!_symbolTable.isDefined(undefName))
|
|
callback(undefName, false);
|
|
}
|
|
|
|
// search libraries for overrides of common symbols
|
|
if (searchForOverrides) {
|
|
for (StringRef tentDefName : _symbolTable.tentativeDefinitions()) {
|
|
// Load for previous tentative may also have loaded
|
|
// something that overrode this tentative, so always check.
|
|
const Atom *curAtom = _symbolTable.findByName(tentDefName);
|
|
assert(curAtom != nullptr);
|
|
if (const DefinedAtom *curDefAtom = dyn_cast<DefinedAtom>(curAtom)) {
|
|
if (curDefAtom->merge() == DefinedAtom::mergeAsTentative)
|
|
callback(tentDefName, true);
|
|
}
|
|
}
|
|
}
|
|
} while (undefineGenCount != _symbolTable.size());
|
|
}
|
|
|
|
bool Resolver::handleArchiveFile(const File &file) {
|
|
const ArchiveLibraryFile *archiveFile = cast<ArchiveLibraryFile>(&file);
|
|
bool searchForOverrides =
|
|
_context.searchArchivesToOverrideTentativeDefinitions();
|
|
bool undefAdded = false;
|
|
forEachUndefines(searchForOverrides,
|
|
[&](StringRef undefName, bool dataSymbolOnly) {
|
|
if (const File *member = archiveFile->find(undefName, dataSymbolOnly)) {
|
|
member->setOrdinal(_context.getNextOrdinalAndIncrement());
|
|
undefAdded = handleFile(*member) || undefAdded;
|
|
}
|
|
});
|
|
return undefAdded;
|
|
}
|
|
|
|
void Resolver::handleSharedLibrary(const File &file) {
|
|
// Add all the atoms from the shared library
|
|
const SharedLibraryFile *sharedLibrary = cast<SharedLibraryFile>(&file);
|
|
handleFile(*sharedLibrary);
|
|
bool searchForOverrides =
|
|
_context.searchSharedLibrariesToOverrideTentativeDefinitions();
|
|
forEachUndefines(searchForOverrides,
|
|
[&](StringRef undefName, bool dataSymbolOnly) {
|
|
if (const SharedLibraryAtom *atom =
|
|
sharedLibrary->exports(undefName, dataSymbolOnly))
|
|
doSharedLibraryAtom(*atom);
|
|
});
|
|
}
|
|
|
|
bool Resolver::doUndefinedAtom(const UndefinedAtom &atom) {
|
|
DEBUG_WITH_TYPE("resolver", llvm::dbgs()
|
|
<< " UndefinedAtom: "
|
|
<< llvm::format("0x%09lX", &atom)
|
|
<< ", name=" << atom.name() << "\n");
|
|
|
|
// add to list of known atoms
|
|
_atoms.push_back(&atom);
|
|
|
|
// tell symbol table
|
|
bool newUndefAdded = _symbolTable.add(atom);
|
|
|
|
// If the undefined symbol has an alternative name, try to resolve the
|
|
// symbol with the name to give it a second chance. This feature is used
|
|
// for COFF "weak external" symbol.
|
|
if (!_symbolTable.isDefined(atom.name())) {
|
|
if (const UndefinedAtom *fallbackAtom = atom.fallback()) {
|
|
doUndefinedAtom(*fallbackAtom);
|
|
_symbolTable.addReplacement(&atom, fallbackAtom);
|
|
}
|
|
}
|
|
return newUndefAdded;
|
|
}
|
|
|
|
/// \brief Add the section group and the group-child reference members.
|
|
void Resolver::maybeAddSectionGroupOrGnuLinkOnce(const DefinedAtom &atom) {
|
|
// First time adding a group?
|
|
bool isFirstTime = _symbolTable.addGroup(atom);
|
|
|
|
if (!isFirstTime) {
|
|
// If duplicate symbols are allowed, select the first group.
|
|
if (_context.getAllowDuplicates())
|
|
return;
|
|
auto *prevGroup = dyn_cast<DefinedAtom>(_symbolTable.findGroup(atom.name()));
|
|
assert(prevGroup &&
|
|
"Internal Error: The group atom could only be a defined atom");
|
|
// The atoms should be of the same content type, reject invalid group
|
|
// resolution behaviors.
|
|
if (atom.contentType() == prevGroup->contentType())
|
|
return;
|
|
llvm::errs() << "SymbolTable: error while merging " << atom.name()
|
|
<< "\n";
|
|
llvm::report_fatal_error("duplicate symbol error");
|
|
return;
|
|
}
|
|
|
|
for (const Reference *r : atom) {
|
|
if (r->kindNamespace() == lld::Reference::KindNamespace::all &&
|
|
r->kindValue() == lld::Reference::kindGroupChild) {
|
|
const DefinedAtom *target = dyn_cast<DefinedAtom>(r->target());
|
|
assert(target && "Internal Error: kindGroupChild references need to "
|
|
"be associated with Defined Atoms only");
|
|
_atoms.push_back(target);
|
|
_symbolTable.add(*target);
|
|
}
|
|
}
|
|
}
|
|
|
|
// Called on each atom when a file is added. Returns true if a given
|
|
// atom is added to the symbol table.
|
|
void Resolver::doDefinedAtom(const DefinedAtom &atom) {
|
|
DEBUG_WITH_TYPE("resolver", llvm::dbgs()
|
|
<< " DefinedAtom: "
|
|
<< llvm::format("0x%09lX", &atom)
|
|
<< ", file=#"
|
|
<< atom.file().ordinal()
|
|
<< ", atom=#"
|
|
<< atom.ordinal()
|
|
<< ", name="
|
|
<< atom.name()
|
|
<< "\n");
|
|
|
|
// Verify on zero-size atoms are pinned to start or end of section.
|
|
if (atom.sectionPosition() == DefinedAtom::sectionPositionStart ||
|
|
atom.sectionPosition() == DefinedAtom::sectionPositionEnd) {
|
|
assert(atom.size() == 0);
|
|
}
|
|
|
|
// add to list of known atoms
|
|
_atoms.push_back(&atom);
|
|
|
|
if (atom.isGroupParent()) {
|
|
maybeAddSectionGroupOrGnuLinkOnce(atom);
|
|
} else {
|
|
_symbolTable.add(atom);
|
|
}
|
|
|
|
// An atom that should never be dead-stripped is a dead-strip root.
|
|
if (_context.deadStrip() && atom.deadStrip() == DefinedAtom::deadStripNever) {
|
|
_deadStripRoots.insert(&atom);
|
|
}
|
|
}
|
|
|
|
void Resolver::doSharedLibraryAtom(const SharedLibraryAtom &atom) {
|
|
DEBUG_WITH_TYPE("resolver", llvm::dbgs()
|
|
<< " SharedLibraryAtom: "
|
|
<< llvm::format("0x%09lX", &atom)
|
|
<< ", name="
|
|
<< atom.name()
|
|
<< "\n");
|
|
|
|
// add to list of known atoms
|
|
_atoms.push_back(&atom);
|
|
|
|
// tell symbol table
|
|
_symbolTable.add(atom);
|
|
}
|
|
|
|
void Resolver::doAbsoluteAtom(const AbsoluteAtom &atom) {
|
|
DEBUG_WITH_TYPE("resolver", llvm::dbgs()
|
|
<< " AbsoluteAtom: "
|
|
<< llvm::format("0x%09lX", &atom)
|
|
<< ", name="
|
|
<< atom.name()
|
|
<< "\n");
|
|
|
|
// add to list of known atoms
|
|
_atoms.push_back(&atom);
|
|
|
|
// tell symbol table
|
|
if (atom.scope() != Atom::scopeTranslationUnit)
|
|
_symbolTable.add(atom);
|
|
}
|
|
|
|
// utility to add a vector of atoms
|
|
void Resolver::addAtoms(const std::vector<const DefinedAtom *> &newAtoms) {
|
|
for (const DefinedAtom *newAtom : newAtoms)
|
|
doDefinedAtom(*newAtom);
|
|
}
|
|
|
|
// Returns true if at least one of N previous files has created an
|
|
// undefined symbol.
|
|
bool Resolver::undefinesAdded(int begin, int end) {
|
|
std::vector<std::unique_ptr<Node>> &inputs = _context.getNodes();
|
|
for (int i = begin; i < end; ++i)
|
|
if (FileNode *node = dyn_cast<FileNode>(inputs[i].get()))
|
|
if (_newUndefinesAdded[node->getFile()])
|
|
return true;
|
|
return false;
|
|
}
|
|
|
|
File *Resolver::getFile(int &index, int &groupLevel) {
|
|
std::vector<std::unique_ptr<Node>> &inputs = _context.getNodes();
|
|
if ((size_t)index >= inputs.size())
|
|
return nullptr;
|
|
if (GroupEnd *group = dyn_cast<GroupEnd>(inputs[index].get())) {
|
|
// We are at the end of the current group. If one or more new
|
|
// undefined atom has been added in the last groupSize files, we
|
|
// reiterate over the files.
|
|
int size = group->getSize();
|
|
if (undefinesAdded(index - size, index)) {
|
|
index -= size;
|
|
++groupLevel;
|
|
return getFile(index, groupLevel);
|
|
}
|
|
++index;
|
|
--groupLevel;
|
|
return getFile(index, groupLevel);
|
|
}
|
|
return cast<FileNode>(inputs[index++].get())->getFile();
|
|
}
|
|
|
|
// Keep adding atoms until _context.getNextFile() returns an error. This
|
|
// function is where undefined atoms are resolved.
|
|
bool Resolver::resolveUndefines() {
|
|
ScopedTask task(getDefaultDomain(), "resolveUndefines");
|
|
int index = 0;
|
|
int groupLevel = 0;
|
|
for (;;) {
|
|
bool undefAdded = false;
|
|
File *file = getFile(index, groupLevel);
|
|
if (!file)
|
|
return true;
|
|
if (std::error_code ec = file->parse()) {
|
|
llvm::errs() << "Cannot open " + file->path()
|
|
<< ": " << ec.message() << "\n";
|
|
return false;
|
|
}
|
|
switch (file->kind()) {
|
|
case File::kindObject:
|
|
if (groupLevel > 0)
|
|
break;
|
|
assert(!file->hasOrdinal());
|
|
file->setOrdinal(_context.getNextOrdinalAndIncrement());
|
|
undefAdded = handleFile(*file);
|
|
break;
|
|
case File::kindArchiveLibrary:
|
|
if (!file->hasOrdinal())
|
|
file->setOrdinal(_context.getNextOrdinalAndIncrement());
|
|
undefAdded = handleArchiveFile(*file);
|
|
break;
|
|
case File::kindSharedLibrary:
|
|
if (!file->hasOrdinal())
|
|
file->setOrdinal(_context.getNextOrdinalAndIncrement());
|
|
handleSharedLibrary(*file);
|
|
break;
|
|
}
|
|
_newUndefinesAdded[file] = undefAdded;
|
|
}
|
|
}
|
|
|
|
// switch all references to undefined or coalesced away atoms
|
|
// to the new defined atom
|
|
void Resolver::updateReferences() {
|
|
ScopedTask task(getDefaultDomain(), "updateReferences");
|
|
for (const Atom *atom : _atoms) {
|
|
if (const DefinedAtom *defAtom = dyn_cast<DefinedAtom>(atom)) {
|
|
for (const Reference *ref : *defAtom) {
|
|
// A reference of type kindAssociate should't be updated.
|
|
// Instead, an atom having such reference will be removed
|
|
// if the target atom is coalesced away, so that they will
|
|
// go away as a group.
|
|
if (ref->kindNamespace() == lld::Reference::KindNamespace::all &&
|
|
ref->kindValue() == lld::Reference::kindAssociate) {
|
|
if (_symbolTable.isCoalescedAway(atom))
|
|
_deadAtoms.insert(ref->target());
|
|
continue;
|
|
}
|
|
const Atom *newTarget = _symbolTable.replacement(ref->target());
|
|
const_cast<Reference *>(ref)->setTarget(newTarget);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// For dead code stripping, recursively mark atoms "live"
|
|
void Resolver::markLive(const Atom *atom) {
|
|
// Mark the atom is live. If it's already marked live, then stop recursion.
|
|
auto exists = _liveAtoms.insert(atom);
|
|
if (!exists.second)
|
|
return;
|
|
|
|
// Mark all atoms it references as live
|
|
if (const DefinedAtom *defAtom = dyn_cast<DefinedAtom>(atom)) {
|
|
for (const Reference *ref : *defAtom)
|
|
markLive(ref->target());
|
|
for (const Atom *target : _reverseRef.lookup(defAtom))
|
|
markLive(target);
|
|
}
|
|
}
|
|
|
|
static bool isBackref(const Reference *ref) {
|
|
if (ref->kindNamespace() != lld::Reference::KindNamespace::all)
|
|
return false;
|
|
return (ref->kindValue() == lld::Reference::kindLayoutBefore ||
|
|
ref->kindValue() == lld::Reference::kindGroupChild);
|
|
}
|
|
|
|
// remove all atoms not actually used
|
|
void Resolver::deadStripOptimize() {
|
|
ScopedTask task(getDefaultDomain(), "deadStripOptimize");
|
|
// only do this optimization with -dead_strip
|
|
if (!_context.deadStrip())
|
|
return;
|
|
|
|
// Some type of references prevent referring atoms to be dead-striped.
|
|
// Make a reverse map of such references before traversing the graph.
|
|
for (const Atom *atom : _atoms)
|
|
if (const DefinedAtom *defAtom = dyn_cast<DefinedAtom>(atom))
|
|
for (const Reference *ref : *defAtom)
|
|
if (isBackref(ref))
|
|
_reverseRef[ref->target()].insert(atom);
|
|
|
|
// By default, shared libraries are built with all globals as dead strip roots
|
|
if (_context.globalsAreDeadStripRoots())
|
|
for (const Atom *atom : _atoms)
|
|
if (const DefinedAtom *defAtom = dyn_cast<DefinedAtom>(atom))
|
|
if (defAtom->scope() == DefinedAtom::scopeGlobal)
|
|
_deadStripRoots.insert(defAtom);
|
|
|
|
// Or, use list of names that are dead strip roots.
|
|
for (const StringRef &name : _context.deadStripRoots()) {
|
|
const Atom *symAtom = _symbolTable.findByName(name);
|
|
assert(symAtom);
|
|
_deadStripRoots.insert(symAtom);
|
|
}
|
|
|
|
// mark all roots as live, and recursively all atoms they reference
|
|
for (const Atom *dsrAtom : _deadStripRoots)
|
|
markLive(dsrAtom);
|
|
|
|
// now remove all non-live atoms from _atoms
|
|
_atoms.erase(std::remove_if(_atoms.begin(), _atoms.end(), [&](const Atom *a) {
|
|
return _liveAtoms.count(a) == 0;
|
|
}),
|
|
_atoms.end());
|
|
}
|
|
|
|
// error out if some undefines remain
|
|
bool Resolver::checkUndefines() {
|
|
// build vector of remaining undefined symbols
|
|
std::vector<const UndefinedAtom *> undefinedAtoms = _symbolTable.undefines();
|
|
if (_context.deadStrip()) {
|
|
// When dead code stripping, we don't care if dead atoms are undefined.
|
|
undefinedAtoms.erase(
|
|
std::remove_if(undefinedAtoms.begin(), undefinedAtoms.end(),
|
|
[&](const Atom *a) { return _liveAtoms.count(a) == 0; }),
|
|
undefinedAtoms.end());
|
|
}
|
|
|
|
// error message about missing symbols
|
|
if (!undefinedAtoms.empty()) {
|
|
// FIXME: need diagnostics interface for writing error messages
|
|
bool foundUndefines = false;
|
|
for (const UndefinedAtom *undefAtom : undefinedAtoms) {
|
|
const File &f = undefAtom->file();
|
|
|
|
// Skip over a weak symbol.
|
|
if (undefAtom->canBeNull() != UndefinedAtom::canBeNullNever)
|
|
continue;
|
|
|
|
// If this is a library and undefined symbols are allowed on the
|
|
// target platform, skip over it.
|
|
if (isa<SharedLibraryFile>(f) && _context.allowShlibUndefines())
|
|
continue;
|
|
|
|
// If the undefine is coalesced away, skip over it.
|
|
if (_symbolTable.isCoalescedAway(undefAtom))
|
|
continue;
|
|
|
|
// Seems like this symbol is undefined. Warn that.
|
|
foundUndefines = true;
|
|
if (_context.printRemainingUndefines()) {
|
|
llvm::errs() << "Undefined symbol: " << undefAtom->file().path()
|
|
<< ": " << _context.demangle(undefAtom->name())
|
|
<< "\n";
|
|
}
|
|
}
|
|
if (foundUndefines) {
|
|
if (_context.printRemainingUndefines())
|
|
llvm::errs() << "symbol(s) not found\n";
|
|
return true;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
// remove from _atoms all coaleseced away atoms
|
|
void Resolver::removeCoalescedAwayAtoms() {
|
|
ScopedTask task(getDefaultDomain(), "removeCoalescedAwayAtoms");
|
|
_atoms.erase(std::remove_if(_atoms.begin(), _atoms.end(), [&](const Atom *a) {
|
|
return _symbolTable.isCoalescedAway(a) || _deadAtoms.count(a);
|
|
}),
|
|
_atoms.end());
|
|
}
|
|
|
|
bool Resolver::resolve() {
|
|
if (!resolveUndefines())
|
|
return false;
|
|
updateReferences();
|
|
deadStripOptimize();
|
|
if (checkUndefines())
|
|
if (!_context.allowRemainingUndefines())
|
|
return false;
|
|
removeCoalescedAwayAtoms();
|
|
_result->addAtoms(_atoms);
|
|
return true;
|
|
}
|
|
|
|
void Resolver::MergedFile::addAtom(const Atom &atom) {
|
|
if (auto *def = dyn_cast<DefinedAtom>(&atom)) {
|
|
_definedAtoms._atoms.push_back(def);
|
|
} else if (auto *undef = dyn_cast<UndefinedAtom>(&atom)) {
|
|
_undefinedAtoms._atoms.push_back(undef);
|
|
} else if (auto *shared = dyn_cast<SharedLibraryAtom>(&atom)) {
|
|
_sharedLibraryAtoms._atoms.push_back(shared);
|
|
} else if (auto *abs = dyn_cast<AbsoluteAtom>(&atom)) {
|
|
_absoluteAtoms._atoms.push_back(abs);
|
|
} else {
|
|
llvm_unreachable("atom has unknown definition kind");
|
|
}
|
|
}
|
|
|
|
MutableFile::DefinedAtomRange Resolver::MergedFile::definedAtoms() {
|
|
return range<std::vector<const DefinedAtom *>::iterator>(
|
|
_definedAtoms._atoms.begin(), _definedAtoms._atoms.end());
|
|
}
|
|
|
|
void Resolver::MergedFile::removeDefinedAtomsIf(
|
|
std::function<bool(const DefinedAtom *)> pred) {
|
|
auto &atoms = _definedAtoms._atoms;
|
|
auto newEnd = std::remove_if(atoms.begin(), atoms.end(), pred);
|
|
atoms.erase(newEnd, atoms.end());
|
|
}
|
|
|
|
void Resolver::MergedFile::addAtoms(std::vector<const Atom *> &all) {
|
|
ScopedTask task(getDefaultDomain(), "addAtoms");
|
|
DEBUG_WITH_TYPE("resolver", llvm::dbgs() << "Resolver final atom list:\n");
|
|
for (const Atom *atom : all) {
|
|
DEBUG_WITH_TYPE("resolver", llvm::dbgs()
|
|
<< llvm::format(" 0x%09lX", atom)
|
|
<< ", name="
|
|
<< atom->name()
|
|
<< "\n");
|
|
addAtom(*atom);
|
|
}
|
|
}
|
|
|
|
} // namespace lld
|