llvm-project/lld/MachO/Driver.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1722 lines
63 KiB
C++
Raw Normal View History

//===- Driver.cpp ---------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#include "Driver.h"
#include "Config.h"
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
#include "ICF.h"
#include "InputFiles.h"
#include "LTO.h"
[lld/mac] Implement -dead_strip Also adds support for live_support sections, no_dead_strip sections, .no_dead_strip symbols. Chromium Framework 345MB unstripped -> 250MB stripped (vs 290MB unstripped -> 236M stripped with ld64). Doing dead stripping is a bit faster than not, because so much less data needs to be processed: % ministat lld_* x lld_nostrip.txt + lld_strip.txt N Min Max Median Avg Stddev x 10 3.929414 4.07692 4.0269079 4.0089678 0.044214794 + 10 3.8129408 3.9025559 3.8670411 3.8642573 0.024779651 Difference at 95.0% confidence -0.144711 +/- 0.0336749 -3.60967% +/- 0.839989% (Student's t, pooled s = 0.0358398) This interacts with many parts of the linker. I tried to add test coverage for all added `isLive()` checks, so that some test will fail if any of them is removed. I checked that the test expectations for the most part match ld64's behavior (except for live-support-iterations.s, see the comment in the test). Interacts with: - debug info - export tries - import opcodes - flags like -exported_symbol(s_list) - -U / dynamic_lookup - mod_init_funcs, mod_term_funcs - weak symbol handling - unwind info - stubs - map files - -sectcreate - undefined, dylib, common, defined (both absolute and normal) symbols It's possible it interacts with more features I didn't think of, of course. I also did some manual testing: - check-llvm check-clang check-lld work with lld with this patch as host linker and -dead_strip enabled - Chromium still starts - Chromium's base_unittests still pass, including unwind tests Implemenation-wise, this is InputSection-based, so it'll work for object files with .subsections_via_symbols (which includes all object files generated by clang). I first based this on the COFF implementation, but later realized that things are more similar to ELF. I think it'd be good to refactor MarkLive.cpp to look more like the ELF part at some point, but I'd like to get a working state checked in first. Mechanical parts: - Rename canOmitFromOutput to wasCoalesced (no behavior change) since it really is for weak coalesced symbols - Add noDeadStrip to Defined, corresponding to N_NO_DEAD_STRIP (`.no_dead_strip` in asm) Fixes PR49276. Differential Revision: https://reviews.llvm.org/D103324
2021-05-08 05:10:05 +08:00
#include "MarkLive.h"
#include "ObjC.h"
#include "OutputSection.h"
#include "OutputSegment.h"
#include "SectionPriorities.h"
#include "SymbolTable.h"
#include "Symbols.h"
#include "SyntheticSections.h"
#include "Target.h"
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
#include "UnwindInfoSection.h"
#include "Writer.h"
#include "lld/Common/Args.h"
#include "lld/Common/Driver.h"
#include "lld/Common/ErrorHandler.h"
#include "lld/Common/LLVM.h"
#include "lld/Common/Memory.h"
#include "lld/Common/Reproduce.h"
#include "lld/Common/Version.h"
#include "llvm/ADT/DenseSet.h"
#include "llvm/ADT/StringExtras.h"
#include "llvm/ADT/StringRef.h"
#include "llvm/BinaryFormat/MachO.h"
#include "llvm/BinaryFormat/Magic.h"
#include "llvm/Config/llvm-config.h"
#include "llvm/LTO/LTO.h"
#include "llvm/Object/Archive.h"
#include "llvm/Option/ArgList.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/FileSystem.h"
#include "llvm/Support/Host.h"
#include "llvm/Support/MemoryBuffer.h"
#include "llvm/Support/Parallel.h"
#include "llvm/Support/Path.h"
#include "llvm/Support/TarWriter.h"
#include "llvm/Support/TargetSelect.h"
#include "llvm/Support/TimeProfiler.h"
#include "llvm/TextAPI/PackedVersion.h"
#include <algorithm>
using namespace llvm;
using namespace llvm::MachO;
using namespace llvm::object;
using namespace llvm::opt;
using namespace llvm::sys;
using namespace lld;
using namespace lld::macho;
std::unique_ptr<Configuration> macho::config;
std::unique_ptr<DependencyTracker> macho::depTracker;
static HeaderFileType getOutputType(const InputArgList &args) {
// TODO: -r, -dylinker, -preload...
Arg *outputArg = args.getLastArg(OPT_bundle, OPT_dylib, OPT_execute);
if (outputArg == nullptr)
return MH_EXECUTE;
switch (outputArg->getOption().getID()) {
case OPT_bundle:
return MH_BUNDLE;
case OPT_dylib:
return MH_DYLIB;
case OPT_execute:
return MH_EXECUTE;
default:
llvm_unreachable("internal error");
}
}
static DenseMap<CachedHashStringRef, StringRef> resolvedLibraries;
static Optional<StringRef> findLibrary(StringRef name) {
CachedHashStringRef key(name);
auto entry = resolvedLibraries.find(key);
if (entry != resolvedLibraries.end())
return entry->second;
auto doFind = [&] {
if (config->searchDylibsFirst) {
if (Optional<StringRef> path = findPathCombination(
"lib" + name, config->librarySearchPaths, {".tbd", ".dylib"}))
return path;
return findPathCombination("lib" + name, config->librarySearchPaths,
{".a"});
}
return findPathCombination("lib" + name, config->librarySearchPaths,
{".tbd", ".dylib", ".a"});
};
Optional<StringRef> path = doFind();
if (path)
resolvedLibraries[key] = *path;
return path;
}
static DenseMap<CachedHashStringRef, StringRef> resolvedFrameworks;
static Optional<StringRef> findFramework(StringRef name) {
CachedHashStringRef key(name);
auto entry = resolvedFrameworks.find(key);
if (entry != resolvedFrameworks.end())
return entry->second;
SmallString<260> symlink;
StringRef suffix;
std::tie(name, suffix) = name.split(",");
for (StringRef dir : config->frameworkSearchPaths) {
symlink = dir;
path::append(symlink, name + ".framework", name);
if (!suffix.empty()) {
// NOTE: we must resolve the symlink before trying the suffixes, because
// there are no symlinks for the suffixed paths.
SmallString<260> location;
if (!fs::real_path(symlink, location)) {
// only append suffix if realpath() succeeds
Twine suffixed = location + suffix;
if (fs::exists(suffixed))
return resolvedFrameworks[key] = saver().save(suffixed.str());
}
// Suffix lookup failed, fall through to the no-suffix case.
}
if (Optional<StringRef> path = resolveDylibPath(symlink.str()))
return resolvedFrameworks[key] = *path;
}
return {};
}
static bool warnIfNotDirectory(StringRef option, StringRef path) {
if (!fs::exists(path)) {
warn("directory not found for option -" + option + path);
return false;
} else if (!fs::is_directory(path)) {
warn("option -" + option + path + " references a non-directory path");
return false;
}
return true;
}
static std::vector<StringRef>
getSearchPaths(unsigned optionCode, InputArgList &args,
const std::vector<StringRef> &roots,
const SmallVector<StringRef, 2> &systemPaths) {
std::vector<StringRef> paths;
StringRef optionLetter{optionCode == OPT_F ? "F" : "L"};
for (StringRef path : args::getStrings(args, optionCode)) {
// NOTE: only absolute paths are re-rooted to syslibroot(s)
bool found = false;
if (path::is_absolute(path, path::Style::posix)) {
for (StringRef root : roots) {
SmallString<261> buffer(root);
path::append(buffer, path);
// Do not warn about paths that are computed via the syslib roots
if (fs::is_directory(buffer)) {
paths.push_back(saver().save(buffer.str()));
found = true;
}
}
}
if (!found && warnIfNotDirectory(optionLetter, path))
paths.push_back(path);
}
// `-Z` suppresses the standard "system" search paths.
if (args.hasArg(OPT_Z))
return paths;
for (const StringRef &path : systemPaths) {
for (const StringRef &root : roots) {
SmallString<261> buffer(root);
path::append(buffer, path);
if (fs::is_directory(buffer))
paths.push_back(saver().save(buffer.str()));
}
}
return paths;
}
static std::vector<StringRef> getSystemLibraryRoots(InputArgList &args) {
std::vector<StringRef> roots;
for (const Arg *arg : args.filtered(OPT_syslibroot))
roots.push_back(arg->getValue());
// NOTE: the final `-syslibroot` being `/` will ignore all roots
if (!roots.empty() && roots.back() == "/")
roots.clear();
// NOTE: roots can never be empty - add an empty root to simplify the library
// and framework search path computation.
if (roots.empty())
roots.emplace_back("");
return roots;
}
static std::vector<StringRef>
getLibrarySearchPaths(InputArgList &args, const std::vector<StringRef> &roots) {
return getSearchPaths(OPT_L, args, roots, {"/usr/lib", "/usr/local/lib"});
}
static std::vector<StringRef>
getFrameworkSearchPaths(InputArgList &args,
const std::vector<StringRef> &roots) {
return getSearchPaths(OPT_F, args, roots,
{"/Library/Frameworks", "/System/Library/Frameworks"});
}
static llvm::CachePruningPolicy getLTOCachePolicy(InputArgList &args) {
SmallString<128> ltoPolicy;
auto add = [&ltoPolicy](Twine val) {
if (!ltoPolicy.empty())
ltoPolicy += ":";
val.toVector(ltoPolicy);
};
for (const Arg *arg :
args.filtered(OPT_thinlto_cache_policy, OPT_prune_interval_lto,
OPT_prune_after_lto, OPT_max_relative_cache_size_lto)) {
switch (arg->getOption().getID()) {
case OPT_thinlto_cache_policy:
add(arg->getValue());
break;
case OPT_prune_interval_lto:
if (!strcmp("-1", arg->getValue()))
add("prune_interval=87600h"); // 10 years
else
add(Twine("prune_interval=") + arg->getValue() + "s");
break;
case OPT_prune_after_lto:
add(Twine("prune_after=") + arg->getValue() + "s");
break;
case OPT_max_relative_cache_size_lto:
add(Twine("cache_size=") + arg->getValue() + "%");
break;
}
}
return CHECK(parseCachePruningPolicy(ltoPolicy), "invalid LTO cache policy");
}
// What caused a given library to be loaded. Only relevant for archives.
// Note that this does not tell us *how* we should load the library, i.e.
// whether we should do it lazily or eagerly (AKA force loading). The "how" is
// decided within addFile().
enum class LoadType {
CommandLine, // Library was passed as a regular CLI argument
CommandLineForce, // Library was passed via `-force_load`
LCLinkerOption, // Library was passed via LC_LINKER_OPTIONS
};
struct ArchiveFileInfo {
ArchiveFile *file;
bool isCommandLineLoad;
};
static DenseMap<StringRef, ArchiveFileInfo> loadedArchives;
static InputFile *addFile(StringRef path, LoadType loadType,
bool isLazy = false, bool isExplicit = true,
bool isBundleLoader = false,
bool isForceHidden = false) {
Optional<MemoryBufferRef> buffer = readFile(path);
if (!buffer)
return nullptr;
MemoryBufferRef mbref = *buffer;
InputFile *newFile = nullptr;
file_magic magic = identify_magic(mbref.getBuffer());
switch (magic) {
case file_magic::archive: {
bool isCommandLineLoad = loadType != LoadType::LCLinkerOption;
// Avoid loading archives twice. If the archives are being force-loaded,
// loading them twice would create duplicate symbol errors. In the
// non-force-loading case, this is just a minor performance optimization.
// We don't take a reference to cachedFile here because the
// loadArchiveMember() call below may recursively call addFile() and
// invalidate this reference.
auto entry = loadedArchives.find(path);
ArchiveFile *file;
if (entry == loadedArchives.end()) {
// No cached archive, we need to create a new one
std::unique_ptr<object::Archive> archive = CHECK(
object::Archive::create(mbref), path + ": failed to parse archive");
if (!archive->isEmpty() && !archive->hasSymbolTable())
error(path + ": archive has no index; run ranlib to add one");
file = make<ArchiveFile>(std::move(archive), isForceHidden);
} else {
file = entry->second.file;
// Command-line loads take precedence. If file is previously loaded via
// command line, or is loaded via LC_LINKER_OPTION and being loaded via
// LC_LINKER_OPTION again, using the cached archive is enough.
if (entry->second.isCommandLineLoad || !isCommandLineLoad)
return file;
}
bool isLCLinkerForceLoad = loadType == LoadType::LCLinkerOption &&
config->forceLoadSwift &&
path::filename(path).startswith("libswift");
if ((isCommandLineLoad && config->allLoad) ||
loadType == LoadType::CommandLineForce || isLCLinkerForceLoad) {
if (Optional<MemoryBufferRef> buffer = readFile(path)) {
Error e = Error::success();
for (const object::Archive::Child &c : file->getArchive().children(e)) {
StringRef reason;
switch (loadType) {
case LoadType::LCLinkerOption:
reason = "LC_LINKER_OPTION";
break;
case LoadType::CommandLineForce:
reason = "-force_load";
break;
case LoadType::CommandLine:
reason = "-all_load";
break;
}
if (Error e = file->fetch(c, reason))
error(toString(file) + ": " + reason +
" failed to load archive member: " + toString(std::move(e)));
}
if (e)
error(toString(file) +
": Archive::children failed: " + toString(std::move(e)));
}
} else if (isCommandLineLoad && config->forceLoadObjC) {
for (const object::Archive::Symbol &sym : file->getArchive().symbols())
if (sym.getName().startswith(objc::klass))
file->fetch(sym);
// TODO: no need to look for ObjC sections for a given archive member if
// we already found that it contains an ObjC symbol.
if (Optional<MemoryBufferRef> buffer = readFile(path)) {
Error e = Error::success();
for (const object::Archive::Child &c : file->getArchive().children(e)) {
Expected<MemoryBufferRef> mb = c.getMemoryBufferRef();
if (!mb || !hasObjCSection(*mb))
continue;
if (Error e = file->fetch(c, "-ObjC"))
error(toString(file) + ": -ObjC failed to load archive member: " +
toString(std::move(e)));
}
if (e)
error(toString(file) +
": Archive::children failed: " + toString(std::move(e)));
}
}
file->addLazySymbols();
loadedArchives[path] = ArchiveFileInfo{file, isCommandLineLoad};
newFile = file;
break;
}
case file_magic::macho_object:
newFile = make<ObjFile>(mbref, getModTime(path), "", isLazy);
break;
case file_magic::macho_dynamically_linked_shared_lib:
case file_magic::macho_dynamically_linked_shared_lib_stub:
case file_magic::tapi_file:
if (DylibFile *dylibFile =
loadDylib(mbref, nullptr, /*isBundleLoader=*/false, isExplicit))
newFile = dylibFile;
break;
case file_magic::bitcode:
newFile = make<BitcodeFile>(mbref, "", 0, isLazy);
break;
case file_magic::macho_executable:
case file_magic::macho_bundle:
// We only allow executable and bundle type here if it is used
// as a bundle loader.
if (!isBundleLoader)
error(path + ": unhandled file type");
if (DylibFile *dylibFile = loadDylib(mbref, nullptr, isBundleLoader))
newFile = dylibFile;
break;
default:
error(path + ": unhandled file type");
}
if (newFile && !isa<DylibFile>(newFile)) {
if ((isa<ObjFile>(newFile) || isa<BitcodeFile>(newFile)) && newFile->lazy &&
config->forceLoadObjC) {
for (Symbol *sym : newFile->symbols)
if (sym && sym->getName().startswith(objc::klass)) {
extract(*newFile, "-ObjC");
break;
}
if (newFile->lazy && hasObjCSection(mbref))
extract(*newFile, "-ObjC");
}
// printArchiveMemberLoad() prints both .a and .o names, so no need to
// print the .a name here. Similarly skip lazy files.
if (config->printEachFile && magic != file_magic::archive && !isLazy)
message(toString(newFile));
inputFiles.insert(newFile);
}
return newFile;
}
static void addLibrary(StringRef name, bool isNeeded, bool isWeak,
bool isReexport, bool isHidden, bool isExplicit,
LoadType loadType) {
if (Optional<StringRef> path = findLibrary(name)) {
if (auto *dylibFile = dyn_cast_or_null<DylibFile>(
addFile(*path, loadType, /*isLazy=*/false, isExplicit,
/*isBundleLoader=*/false, isHidden))) {
if (isNeeded)
dylibFile->forceNeeded = true;
if (isWeak)
dylibFile->forceWeakImport = true;
if (isReexport) {
config->hasReexports = true;
dylibFile->reexport = true;
}
}
return;
}
error("library not found for -l" + name);
}
static DenseSet<StringRef> loadedObjectFrameworks;
static void addFramework(StringRef name, bool isNeeded, bool isWeak,
bool isReexport, bool isExplicit, LoadType loadType) {
if (Optional<StringRef> path = findFramework(name)) {
if (loadedObjectFrameworks.contains(*path))
return;
InputFile *file =
addFile(*path, loadType, /*isLazy=*/false, isExplicit, false);
if (auto *dylibFile = dyn_cast_or_null<DylibFile>(file)) {
if (isNeeded)
dylibFile->forceNeeded = true;
if (isWeak)
dylibFile->forceWeakImport = true;
if (isReexport) {
config->hasReexports = true;
dylibFile->reexport = true;
}
} else if (isa_and_nonnull<ObjFile>(file) ||
isa_and_nonnull<BitcodeFile>(file)) {
// Cache frameworks containing object or bitcode files to avoid duplicate
// symbols. Frameworks containing static archives are cached separately
// in addFile() to share caching with libraries, and frameworks
// containing dylibs should allow overwriting of attributes such as
// forceNeeded by subsequent loads
loadedObjectFrameworks.insert(*path);
}
return;
}
error("framework not found for -framework " + name);
}
// Parses LC_LINKER_OPTION contents, which can add additional command line
// flags. This directly parses the flags instead of using the standard argument
// parser to improve performance.
void macho::parseLCLinkerOption(InputFile *f, unsigned argc, StringRef data) {
SmallVector<StringRef, 4> argv;
size_t offset = 0;
for (unsigned i = 0; i < argc && offset < data.size(); ++i) {
argv.push_back(data.data() + offset);
offset += strlen(data.data() + offset) + 1;
}
if (argv.size() != argc || offset > data.size())
fatal(toString(f) + ": invalid LC_LINKER_OPTION");
unsigned i = 0;
StringRef arg = argv[i];
if (arg.consume_front("-l")) {
addLibrary(arg, /*isNeeded=*/false, /*isWeak=*/false,
/*isReexport=*/false, /*isHidden=*/false, /*isExplicit=*/false,
LoadType::LCLinkerOption);
} else if (arg == "-framework") {
StringRef name = argv[++i];
addFramework(name, /*isNeeded=*/false, /*isWeak=*/false,
/*isReexport=*/false, /*isExplicit=*/false,
LoadType::LCLinkerOption);
} else {
error(arg + " is not allowed in LC_LINKER_OPTION");
}
}
static void addFileList(StringRef path, bool isLazy) {
Optional<MemoryBufferRef> buffer = readFile(path);
if (!buffer)
return;
MemoryBufferRef mbref = *buffer;
for (StringRef path : args::getLines(mbref))
addFile(rerootPath(path), LoadType::CommandLine, isLazy);
}
// We expect sub-library names of the form "libfoo", which will match a dylib
// with a path of .*/libfoo.{dylib, tbd}.
// XXX ld64 seems to ignore the extension entirely when matching sub-libraries;
// I'm not sure what the use case for that is.
static bool markReexport(StringRef searchName, ArrayRef<StringRef> extensions) {
for (InputFile *file : inputFiles) {
if (auto *dylibFile = dyn_cast<DylibFile>(file)) {
StringRef filename = path::filename(dylibFile->getName());
if (filename.consume_front(searchName) &&
2022-07-21 00:09:19 +08:00
(filename.empty() || llvm::is_contained(extensions, filename))) {
dylibFile->reexport = true;
return true;
}
}
}
return false;
}
// This function is called on startup. We need this for LTO since
// LTO calls LLVM functions to compile bitcode files to native code.
// Technically this can be delayed until we read bitcode files, but
// we don't bother to do lazily because the initialization is fast.
static void initLLVM() {
InitializeAllTargets();
InitializeAllTargetMCs();
InitializeAllAsmPrinters();
InitializeAllAsmParsers();
}
static bool compileBitcodeFiles() {
TimeTraceScope timeScope("LTO");
auto *lto = make<BitcodeCompiler>();
for (InputFile *file : inputFiles)
if (auto *bitcodeFile = dyn_cast<BitcodeFile>(file))
if (!file->lazy)
lto->add(*bitcodeFile);
std::vector<ObjFile *> compiled = lto->compile();
for (ObjFile *file : compiled)
inputFiles.insert(file);
return !compiled.empty();
}
// Replaces common symbols with defined symbols residing in __common sections.
// This function must be called after all symbol names are resolved (i.e. after
// all InputFiles have been loaded.) As a result, later operations won't see
// any CommonSymbols.
static void replaceCommonSymbols() {
TimeTraceScope timeScope("Replace common symbols");
2021-07-18 01:42:26 +08:00
ConcatOutputSection *osec = nullptr;
for (Symbol *sym : symtab->getSymbols()) {
auto *common = dyn_cast<CommonSymbol>(sym);
if (common == nullptr)
continue;
// Casting to size_t will truncate large values on 32-bit architectures,
// but it's not really worth supporting the linking of 64-bit programs on
// 32-bit archs.
ArrayRef<uint8_t> data = {nullptr, static_cast<size_t>(common->size)};
[lld-macho][nfc] Eliminate InputSection::Shared Earlier in LLD's evolution, I tried to create the illusion that subsections were indistinguishable from "top-level" sections. Thus, even though the subsections shared many common field values, I hid those common values away in a private Shared struct (see D105305). More recently, however, @gkm added a public `Section` struct in D113241 that served as an explicit way to store values that are common to an entire set of subsections (aka InputSections). Now that we have another "common value" struct, `Shared` has been rendered redundant. All its fields can be moved into `Section` instead, and the pointer to `Shared` can be replaced with a pointer to `Section`. This `Section` pointer also has the advantage of letting us inspect other subsections easily, simplifying the implementation of {D118798}. P.S. I do think that having both `Section` and `InputSection` makes for a slightly confusing naming scheme. I considered renaming `InputSection` to `Subsection`, but that would break the symmetry with `OutputSection`. It would also make us deviate from LLD-ELF's naming scheme. This change is perf-neutral on my 3.2 GHz 16-Core Intel Xeon W machine: base diff difference (95% CI) sys_time 1.258 ± 0.031 1.248 ± 0.023 [ -1.6% .. +0.1%] user_time 3.659 ± 0.047 3.658 ± 0.041 [ -0.5% .. +0.4%] wall_time 4.640 ± 0.085 4.625 ± 0.063 [ -1.0% .. +0.3%] samples 49 61 There's also no stat sig change in RSS (as measured by `time -l`): base diff difference (95% CI) time 998038627.097 ± 13567305.958 1003327715.556 ± 15210451.236 [ -0.2% .. +1.2%] samples 31 36 Reviewed By: #lld-macho, oontvoo Differential Revision: https://reviews.llvm.org/D118797
2022-02-04 08:53:29 +08:00
// FIXME avoid creating one Section per symbol?
auto *section =
make<Section>(common->getFile(), segment_names::data,
section_names::common, S_ZEROFILL, /*addr=*/0);
auto *isec = make<ConcatInputSection>(*section, data, common->align);
2021-07-18 01:42:26 +08:00
if (!osec)
osec = ConcatOutputSection::getOrCreateForInput(isec);
isec->parent = osec;
inputSections.push_back(isec);
[lld/mac] Implement -dead_strip Also adds support for live_support sections, no_dead_strip sections, .no_dead_strip symbols. Chromium Framework 345MB unstripped -> 250MB stripped (vs 290MB unstripped -> 236M stripped with ld64). Doing dead stripping is a bit faster than not, because so much less data needs to be processed: % ministat lld_* x lld_nostrip.txt + lld_strip.txt N Min Max Median Avg Stddev x 10 3.929414 4.07692 4.0269079 4.0089678 0.044214794 + 10 3.8129408 3.9025559 3.8670411 3.8642573 0.024779651 Difference at 95.0% confidence -0.144711 +/- 0.0336749 -3.60967% +/- 0.839989% (Student's t, pooled s = 0.0358398) This interacts with many parts of the linker. I tried to add test coverage for all added `isLive()` checks, so that some test will fail if any of them is removed. I checked that the test expectations for the most part match ld64's behavior (except for live-support-iterations.s, see the comment in the test). Interacts with: - debug info - export tries - import opcodes - flags like -exported_symbol(s_list) - -U / dynamic_lookup - mod_init_funcs, mod_term_funcs - weak symbol handling - unwind info - stubs - map files - -sectcreate - undefined, dylib, common, defined (both absolute and normal) symbols It's possible it interacts with more features I didn't think of, of course. I also did some manual testing: - check-llvm check-clang check-lld work with lld with this patch as host linker and -dead_strip enabled - Chromium still starts - Chromium's base_unittests still pass, including unwind tests Implemenation-wise, this is InputSection-based, so it'll work for object files with .subsections_via_symbols (which includes all object files generated by clang). I first based this on the COFF implementation, but later realized that things are more similar to ELF. I think it'd be good to refactor MarkLive.cpp to look more like the ELF part at some point, but I'd like to get a working state checked in first. Mechanical parts: - Rename canOmitFromOutput to wasCoalesced (no behavior change) since it really is for weak coalesced symbols - Add noDeadStrip to Defined, corresponding to N_NO_DEAD_STRIP (`.no_dead_strip` in asm) Fixes PR49276. Differential Revision: https://reviews.llvm.org/D103324
2021-05-08 05:10:05 +08:00
// FIXME: CommonSymbol should store isReferencedDynamically, noDeadStrip
// and pass them on here.
replaceSymbol<Defined>(
sym, sym->getName(), common->getFile(), isec, /*value=*/0, /*size=*/0,
/*isWeakDef=*/false, /*isExternal=*/true, common->privateExtern,
/*includeInSymtab=*/true, /*isThumb=*/false,
/*isReferencedDynamically=*/false, /*noDeadStrip=*/false);
}
}
static void initializeSectionRenameMap() {
if (config->dataConst) {
SmallVector<StringRef> v{section_names::got,
section_names::authGot,
section_names::authPtr,
section_names::nonLazySymbolPtr,
section_names::const_,
section_names::cfString,
section_names::moduleInitFunc,
section_names::moduleTermFunc,
section_names::objcClassList,
section_names::objcNonLazyClassList,
section_names::objcCatList,
section_names::objcNonLazyCatList,
section_names::objcProtoList,
section_names::objCImageInfo};
for (StringRef s : v)
config->sectionRenameMap[{segment_names::data, s}] = {
segment_names::dataConst, s};
}
config->sectionRenameMap[{segment_names::text, section_names::staticInit}] = {
segment_names::text, section_names::text};
config->sectionRenameMap[{segment_names::import, section_names::pointers}] = {
config->dataConst ? segment_names::dataConst : segment_names::data,
section_names::nonLazySymbolPtr};
}
static inline char toLowerDash(char x) {
if (x >= 'A' && x <= 'Z')
return x - 'A' + 'a';
else if (x == ' ')
return '-';
return x;
}
static std::string lowerDash(StringRef s) {
return std::string(map_iterator(s.begin(), toLowerDash),
map_iterator(s.end(), toLowerDash));
}
struct PlatformVersion {
PlatformType platform = PLATFORM_UNKNOWN;
llvm::VersionTuple minimum;
llvm::VersionTuple sdk;
};
static PlatformVersion parsePlatformVersion(const Arg *arg) {
assert(arg->getOption().getID() == OPT_platform_version);
StringRef platformStr = arg->getValue(0);
StringRef minVersionStr = arg->getValue(1);
StringRef sdkVersionStr = arg->getValue(2);
PlatformVersion platformVersion;
// TODO(compnerd) see if we can generate this case list via XMACROS
platformVersion.platform =
StringSwitch<PlatformType>(lowerDash(platformStr))
.Cases("macos", "1", PLATFORM_MACOS)
.Cases("ios", "2", PLATFORM_IOS)
.Cases("tvos", "3", PLATFORM_TVOS)
.Cases("watchos", "4", PLATFORM_WATCHOS)
.Cases("bridgeos", "5", PLATFORM_BRIDGEOS)
.Cases("mac-catalyst", "6", PLATFORM_MACCATALYST)
.Cases("ios-simulator", "7", PLATFORM_IOSSIMULATOR)
.Cases("tvos-simulator", "8", PLATFORM_TVOSSIMULATOR)
.Cases("watchos-simulator", "9", PLATFORM_WATCHOSSIMULATOR)
.Cases("driverkit", "10", PLATFORM_DRIVERKIT)
.Default(PLATFORM_UNKNOWN);
if (platformVersion.platform == PLATFORM_UNKNOWN)
error(Twine("malformed platform: ") + platformStr);
// TODO: check validity of version strings, which varies by platform
// NOTE: ld64 accepts version strings with 5 components
// llvm::VersionTuple accepts no more than 4 components
// Has Apple ever published version strings with 5 components?
if (platformVersion.minimum.tryParse(minVersionStr))
error(Twine("malformed minimum version: ") + minVersionStr);
if (platformVersion.sdk.tryParse(sdkVersionStr))
error(Twine("malformed sdk version: ") + sdkVersionStr);
return platformVersion;
}
// Has the side-effect of setting Config::platformInfo.
static PlatformType parsePlatformVersions(const ArgList &args) {
std::map<PlatformType, PlatformVersion> platformVersions;
const PlatformVersion *lastVersionInfo = nullptr;
for (const Arg *arg : args.filtered(OPT_platform_version)) {
PlatformVersion version = parsePlatformVersion(arg);
// For each platform, the last flag wins:
// `-platform_version macos 2 3 -platform_version macos 4 5` has the same
// effect as just passing `-platform_version macos 4 5`.
// FIXME: ld64 warns on multiple flags for one platform. Should we?
platformVersions[version.platform] = version;
lastVersionInfo = &platformVersions[version.platform];
}
if (platformVersions.empty()) {
error("must specify -platform_version");
return PLATFORM_UNKNOWN;
}
if (platformVersions.size() > 2) {
error("must specify -platform_version at most twice");
return PLATFORM_UNKNOWN;
}
if (platformVersions.size() == 2) {
[lld/mac] Support writing zippered dylibs and bundles With -platform_version flags for two distinct platforms, this writes a LC_BUILD_VERSION header for each. The motivation is that this is needed for self-hosting with lld as linker after D124059. To create a zippered output at the clang driver level, pass -target arm64-apple-macos -darwin-target-variant arm64-apple-ios-macabi to create a zippered dylib. (In Xcode's clang, `-darwin-target-variant` is spelled just `-target-variant`.) (If you pass `-target arm64-apple-ios-macabi -target-variant arm64-apple-macos` instead, ld64 crashes!) This results in two -platform_version flags being passed to the linker. ld64 also verifies that the iOS SDK version is at least 13.1. We don't do that yet. But ld64 also does that for other platforms and we don't. So we need to do that at some point, but not in this patch. Only dylib and bundle outputs can be zippered. I verified that a Catalyst app linked against a dylib created with clang -shared foo.cc -o libfoo.dylib \ -target arm64-apple-macos \ -target-variant arm64-apple-ios-macabi \ -Wl,-install_name,@rpath/libfoo.dylib \ -fuse-ld=$PWD/out/gn/bin/ld64.lld runs successfully. (The app calls a function `f()` in libfoo.dylib that returns a const char* "foo", and NSLog(@"%s")s it.) ld64 is a bit more permissive when writing zippered outputs, see references to "unzippered twins". That's not implemented yet. (If anybody wants to implement that, D124275 is a good start.) Differential Revision: https://reviews.llvm.org/D124887
2022-04-22 23:55:50 +08:00
bool isZipperedCatalyst = platformVersions.count(PLATFORM_MACOS) &&
platformVersions.count(PLATFORM_MACCATALYST);
if (!isZipperedCatalyst) {
error("lld supports writing zippered outputs only for "
"macos and mac-catalyst");
} else if (config->outputType != MH_DYLIB &&
config->outputType != MH_BUNDLE) {
error("writing zippered outputs only valid for -dylib and -bundle");
} else {
config->platformInfo.minimum = platformVersions[PLATFORM_MACOS].minimum;
config->platformInfo.sdk = platformVersions[PLATFORM_MACOS].sdk;
config->secondaryPlatformInfo = PlatformInfo{};
config->secondaryPlatformInfo->minimum =
platformVersions[PLATFORM_MACCATALYST].minimum;
config->secondaryPlatformInfo->sdk =
platformVersions[PLATFORM_MACCATALYST].sdk;
}
return PLATFORM_MACOS;
}
[lld/mac] Support writing zippered dylibs and bundles With -platform_version flags for two distinct platforms, this writes a LC_BUILD_VERSION header for each. The motivation is that this is needed for self-hosting with lld as linker after D124059. To create a zippered output at the clang driver level, pass -target arm64-apple-macos -darwin-target-variant arm64-apple-ios-macabi to create a zippered dylib. (In Xcode's clang, `-darwin-target-variant` is spelled just `-target-variant`.) (If you pass `-target arm64-apple-ios-macabi -target-variant arm64-apple-macos` instead, ld64 crashes!) This results in two -platform_version flags being passed to the linker. ld64 also verifies that the iOS SDK version is at least 13.1. We don't do that yet. But ld64 also does that for other platforms and we don't. So we need to do that at some point, but not in this patch. Only dylib and bundle outputs can be zippered. I verified that a Catalyst app linked against a dylib created with clang -shared foo.cc -o libfoo.dylib \ -target arm64-apple-macos \ -target-variant arm64-apple-ios-macabi \ -Wl,-install_name,@rpath/libfoo.dylib \ -fuse-ld=$PWD/out/gn/bin/ld64.lld runs successfully. (The app calls a function `f()` in libfoo.dylib that returns a const char* "foo", and NSLog(@"%s")s it.) ld64 is a bit more permissive when writing zippered outputs, see references to "unzippered twins". That's not implemented yet. (If anybody wants to implement that, D124275 is a good start.) Differential Revision: https://reviews.llvm.org/D124887
2022-04-22 23:55:50 +08:00
config->platformInfo.minimum = lastVersionInfo->minimum;
config->platformInfo.sdk = lastVersionInfo->sdk;
return lastVersionInfo->platform;
}
// Has the side-effect of setting Config::target.
static TargetInfo *createTargetInfo(InputArgList &args) {
StringRef archName = args.getLastArgValue(OPT_arch);
if (archName.empty()) {
error("must specify -arch");
return nullptr;
}
PlatformType platform = parsePlatformVersions(args);
config->platformInfo.target =
MachO::Target(getArchitectureFromName(archName), platform);
[lld/mac] Support writing zippered dylibs and bundles With -platform_version flags for two distinct platforms, this writes a LC_BUILD_VERSION header for each. The motivation is that this is needed for self-hosting with lld as linker after D124059. To create a zippered output at the clang driver level, pass -target arm64-apple-macos -darwin-target-variant arm64-apple-ios-macabi to create a zippered dylib. (In Xcode's clang, `-darwin-target-variant` is spelled just `-target-variant`.) (If you pass `-target arm64-apple-ios-macabi -target-variant arm64-apple-macos` instead, ld64 crashes!) This results in two -platform_version flags being passed to the linker. ld64 also verifies that the iOS SDK version is at least 13.1. We don't do that yet. But ld64 also does that for other platforms and we don't. So we need to do that at some point, but not in this patch. Only dylib and bundle outputs can be zippered. I verified that a Catalyst app linked against a dylib created with clang -shared foo.cc -o libfoo.dylib \ -target arm64-apple-macos \ -target-variant arm64-apple-ios-macabi \ -Wl,-install_name,@rpath/libfoo.dylib \ -fuse-ld=$PWD/out/gn/bin/ld64.lld runs successfully. (The app calls a function `f()` in libfoo.dylib that returns a const char* "foo", and NSLog(@"%s")s it.) ld64 is a bit more permissive when writing zippered outputs, see references to "unzippered twins". That's not implemented yet. (If anybody wants to implement that, D124275 is a good start.) Differential Revision: https://reviews.llvm.org/D124887
2022-04-22 23:55:50 +08:00
if (config->secondaryPlatformInfo) {
config->secondaryPlatformInfo->target =
MachO::Target(getArchitectureFromName(archName), PLATFORM_MACCATALYST);
}
uint32_t cpuType;
uint32_t cpuSubtype;
std::tie(cpuType, cpuSubtype) = getCPUTypeFromArchitecture(config->arch());
switch (cpuType) {
[lld-macho][nfc] Remove `MachO::` prefix where possible Previously, SyntheticSections.cpp did not have a top-level `using namespace llvm::MachO` because it caused a naming conflict: `llvm::MachO::Symbol` would collide with `lld::macho::Symbol`. `MachO::Symbol` represents the symbols defined in InterfaceFiles (TBDs). By moving the inclusion of InterfaceFile.h into our .cpp files, we can avoid this name collision in other files where we are only dealing with LLD's own symbols. Along the way, I removed all unnecessary "MachO::" prefixes in our code. Cons of this approach: If TextAPI/MachO/Symbol.h gets included via some other header file in the future, we could run into this collision again. Alternative 1: Have either TextAPI/MachO or BinaryFormat/MachO.h use a different namespace. Most of the benefit of `using namespace llvm::MachO` comes from being able to use things in BinaryFormat/MachO.h conveniently; if TextAPI was under a different (and fully-qualified) namespace like `llvm::tapi` that would solve our problems. Cons: lots of files across llvm-project will need to be updated, and folks who own the TextAPI code need to agree to the name change. Alternative 2: Rename our Symbol to something like `LldSymbol`. I think this is ugly. Personally I think alternative #1 is ideal, but I'm not sure the effort to do it is worthwhile, this diff's halfway solution seems good enough to me. Thoughts? Reviewed By: #lld-macho, oontvoo, MaskRay Differential Revision: https://reviews.llvm.org/D98149
2021-03-12 02:28:08 +08:00
case CPU_TYPE_X86_64:
return createX86_64TargetInfo();
[lld-macho][nfc] Remove `MachO::` prefix where possible Previously, SyntheticSections.cpp did not have a top-level `using namespace llvm::MachO` because it caused a naming conflict: `llvm::MachO::Symbol` would collide with `lld::macho::Symbol`. `MachO::Symbol` represents the symbols defined in InterfaceFiles (TBDs). By moving the inclusion of InterfaceFile.h into our .cpp files, we can avoid this name collision in other files where we are only dealing with LLD's own symbols. Along the way, I removed all unnecessary "MachO::" prefixes in our code. Cons of this approach: If TextAPI/MachO/Symbol.h gets included via some other header file in the future, we could run into this collision again. Alternative 1: Have either TextAPI/MachO or BinaryFormat/MachO.h use a different namespace. Most of the benefit of `using namespace llvm::MachO` comes from being able to use things in BinaryFormat/MachO.h conveniently; if TextAPI was under a different (and fully-qualified) namespace like `llvm::tapi` that would solve our problems. Cons: lots of files across llvm-project will need to be updated, and folks who own the TextAPI code need to agree to the name change. Alternative 2: Rename our Symbol to something like `LldSymbol`. I think this is ugly. Personally I think alternative #1 is ideal, but I'm not sure the effort to do it is worthwhile, this diff's halfway solution seems good enough to me. Thoughts? Reviewed By: #lld-macho, oontvoo, MaskRay Differential Revision: https://reviews.llvm.org/D98149
2021-03-12 02:28:08 +08:00
case CPU_TYPE_ARM64:
return createARM64TargetInfo();
case CPU_TYPE_ARM64_32:
return createARM64_32TargetInfo();
case CPU_TYPE_ARM:
return createARMTargetInfo(cpuSubtype);
default:
error("missing or unsupported -arch " + archName);
return nullptr;
}
}
static UndefinedSymbolTreatment
getUndefinedSymbolTreatment(const ArgList &args) {
StringRef treatmentStr = args.getLastArgValue(OPT_undefined);
auto treatment =
StringSwitch<UndefinedSymbolTreatment>(treatmentStr)
.Cases("error", "", UndefinedSymbolTreatment::error)
.Case("warning", UndefinedSymbolTreatment::warning)
.Case("suppress", UndefinedSymbolTreatment::suppress)
.Case("dynamic_lookup", UndefinedSymbolTreatment::dynamic_lookup)
.Default(UndefinedSymbolTreatment::unknown);
if (treatment == UndefinedSymbolTreatment::unknown) {
warn(Twine("unknown -undefined TREATMENT '") + treatmentStr +
"', defaulting to 'error'");
treatment = UndefinedSymbolTreatment::error;
} else if (config->namespaceKind == NamespaceKind::twolevel &&
(treatment == UndefinedSymbolTreatment::warning ||
treatment == UndefinedSymbolTreatment::suppress)) {
if (treatment == UndefinedSymbolTreatment::warning)
error("'-undefined warning' only valid with '-flat_namespace'");
else
error("'-undefined suppress' only valid with '-flat_namespace'");
treatment = UndefinedSymbolTreatment::error;
}
return treatment;
}
static ICFLevel getICFLevel(const ArgList &args) {
StringRef icfLevelStr = args.getLastArgValue(OPT_icf_eq);
auto icfLevel = StringSwitch<ICFLevel>(icfLevelStr)
.Cases("none", "", ICFLevel::none)
.Case("safe", ICFLevel::safe)
.Case("all", ICFLevel::all)
.Default(ICFLevel::unknown);
if (icfLevel == ICFLevel::unknown) {
warn(Twine("unknown --icf=OPTION `") + icfLevelStr +
"', defaulting to `none'");
icfLevel = ICFLevel::none;
}
return icfLevel;
}
static void warnIfDeprecatedOption(const Option &opt) {
if (!opt.getGroup().isValid())
return;
if (opt.getGroup().getID() == OPT_grp_deprecated) {
warn("Option `" + opt.getPrefixedName() + "' is deprecated in ld64:");
warn(opt.getHelpText());
}
}
static void warnIfUnimplementedOption(const Option &opt) {
if (!opt.getGroup().isValid() || !opt.hasFlag(DriverFlag::HelpHidden))
return;
switch (opt.getGroup().getID()) {
case OPT_grp_deprecated:
// warn about deprecated options elsewhere
break;
case OPT_grp_undocumented:
warn("Option `" + opt.getPrefixedName() +
"' is undocumented. Should lld implement it?");
break;
case OPT_grp_obsolete:
warn("Option `" + opt.getPrefixedName() +
"' is obsolete. Please modernize your usage.");
break;
case OPT_grp_ignored:
warn("Option `" + opt.getPrefixedName() + "' is ignored.");
break;
case OPT_grp_ignored_silently:
break;
default:
warn("Option `" + opt.getPrefixedName() +
"' is not yet implemented. Stay tuned...");
break;
}
}
static const char *getReproduceOption(InputArgList &args) {
if (const Arg *arg = args.getLastArg(OPT_reproduce))
return arg->getValue();
return getenv("LLD_REPRODUCE");
}
static void parseClangOption(StringRef opt, const Twine &msg) {
std::string err;
raw_string_ostream os(err);
const char *argv[] = {"lld", opt.data()};
if (cl::ParseCommandLineOptions(2, argv, "", &os))
return;
os.flush();
error(msg + ": " + StringRef(err).trim());
}
static uint32_t parseDylibVersion(const ArgList &args, unsigned id) {
const Arg *arg = args.getLastArg(id);
if (!arg)
return 0;
if (config->outputType != MH_DYLIB) {
error(arg->getAsString(args) + ": only valid with -dylib");
return 0;
}
PackedVersion version;
if (!version.parse32(arg->getValue())) {
error(arg->getAsString(args) + ": malformed version");
return 0;
}
return version.rawValue();
}
static uint32_t parseProtection(StringRef protStr) {
uint32_t prot = 0;
for (char c : protStr) {
switch (c) {
case 'r':
prot |= VM_PROT_READ;
break;
case 'w':
prot |= VM_PROT_WRITE;
break;
case 'x':
prot |= VM_PROT_EXECUTE;
break;
case '-':
break;
default:
error("unknown -segprot letter '" + Twine(c) + "' in " + protStr);
return 0;
}
}
return prot;
}
static std::vector<SectionAlign> parseSectAlign(const opt::InputArgList &args) {
std::vector<SectionAlign> sectAligns;
for (const Arg *arg : args.filtered(OPT_sectalign)) {
StringRef segName = arg->getValue(0);
StringRef sectName = arg->getValue(1);
StringRef alignStr = arg->getValue(2);
if (alignStr.startswith("0x") || alignStr.startswith("0X"))
alignStr = alignStr.drop_front(2);
uint32_t align;
if (alignStr.getAsInteger(16, align)) {
error("-sectalign: failed to parse '" + StringRef(arg->getValue(2)) +
"' as number");
continue;
}
if (!isPowerOf2_32(align)) {
error("-sectalign: '" + StringRef(arg->getValue(2)) +
"' (in base 16) not a power of two");
continue;
}
sectAligns.push_back({segName, sectName, align});
}
return sectAligns;
}
PlatformType macho::removeSimulator(PlatformType platform) {
switch (platform) {
case PLATFORM_IOSSIMULATOR:
return PLATFORM_IOS;
case PLATFORM_TVOSSIMULATOR:
return PLATFORM_TVOS;
case PLATFORM_WATCHOSSIMULATOR:
return PLATFORM_WATCHOS;
default:
return platform;
}
}
static bool dataConstDefault(const InputArgList &args) {
static const std::vector<std::pair<PlatformType, VersionTuple>> minVersion = {
{PLATFORM_MACOS, VersionTuple(10, 15)},
{PLATFORM_IOS, VersionTuple(13, 0)},
{PLATFORM_TVOS, VersionTuple(13, 0)},
{PLATFORM_WATCHOS, VersionTuple(6, 0)},
{PLATFORM_BRIDGEOS, VersionTuple(4, 0)}};
PlatformType platform = removeSimulator(config->platformInfo.target.Platform);
auto it = llvm::find_if(minVersion,
[&](const auto &p) { return p.first == platform; });
if (it != minVersion.end())
if (config->platformInfo.minimum < it->second)
return false;
switch (config->outputType) {
case MH_EXECUTE:
return !args.hasArg(OPT_no_pie);
case MH_BUNDLE:
// FIXME: return false when -final_name ...
// has prefix "/System/Library/UserEventPlugins/"
// or matches "/usr/libexec/locationd" "/usr/libexec/terminusd"
return true;
case MH_DYLIB:
return true;
case MH_OBJECT:
return false;
default:
llvm_unreachable(
"unsupported output type for determining data-const default");
}
return false;
}
void SymbolPatterns::clear() {
literals.clear();
globs.clear();
}
void SymbolPatterns::insert(StringRef symbolName) {
if (symbolName.find_first_of("*?[]") == StringRef::npos)
literals.insert(CachedHashStringRef(symbolName));
else if (Expected<GlobPattern> pattern = GlobPattern::create(symbolName))
globs.emplace_back(*pattern);
else
error("invalid symbol-name pattern: " + symbolName);
}
bool SymbolPatterns::matchLiteral(StringRef symbolName) const {
return literals.contains(CachedHashStringRef(symbolName));
}
bool SymbolPatterns::matchGlob(StringRef symbolName) const {
for (const GlobPattern &glob : globs)
if (glob.match(symbolName))
return true;
return false;
}
bool SymbolPatterns::match(StringRef symbolName) const {
return matchLiteral(symbolName) || matchGlob(symbolName);
}
static void parseSymbolPatternsFile(const Arg *arg,
SymbolPatterns &symbolPatterns) {
StringRef path = arg->getValue();
Optional<MemoryBufferRef> buffer = readFile(path);
if (!buffer) {
error("Could not read symbol file: " + path);
return;
}
MemoryBufferRef mbref = *buffer;
for (StringRef line : args::getLines(mbref)) {
line = line.take_until([](char c) { return c == '#'; }).trim();
if (!line.empty())
symbolPatterns.insert(line);
}
}
static void handleSymbolPatterns(InputArgList &args,
SymbolPatterns &symbolPatterns,
unsigned singleOptionCode,
unsigned listFileOptionCode) {
for (const Arg *arg : args.filtered(singleOptionCode))
symbolPatterns.insert(arg->getValue());
for (const Arg *arg : args.filtered(listFileOptionCode))
parseSymbolPatternsFile(arg, symbolPatterns);
}
static void createFiles(const InputArgList &args) {
TimeTraceScope timeScope("Load input files");
// This loop should be reserved for options whose exact ordering matters.
// Other options should be handled via filtered() and/or getLastArg().
bool isLazy = false;
for (const Arg *arg : args) {
const Option &opt = arg->getOption();
warnIfDeprecatedOption(opt);
warnIfUnimplementedOption(opt);
switch (opt.getID()) {
case OPT_INPUT:
addFile(rerootPath(arg->getValue()), LoadType::CommandLine, isLazy);
break;
case OPT_needed_library:
if (auto *dylibFile = dyn_cast_or_null<DylibFile>(
addFile(rerootPath(arg->getValue()), LoadType::CommandLine)))
dylibFile->forceNeeded = true;
break;
case OPT_reexport_library:
if (auto *dylibFile = dyn_cast_or_null<DylibFile>(
addFile(rerootPath(arg->getValue()), LoadType::CommandLine))) {
config->hasReexports = true;
dylibFile->reexport = true;
}
break;
case OPT_weak_library:
if (auto *dylibFile = dyn_cast_or_null<DylibFile>(
addFile(rerootPath(arg->getValue()), LoadType::CommandLine)))
dylibFile->forceWeakImport = true;
break;
case OPT_filelist:
addFileList(arg->getValue(), isLazy);
break;
case OPT_force_load:
addFile(rerootPath(arg->getValue()), LoadType::CommandLineForce);
break;
case OPT_load_hidden:
addFile(rerootPath(arg->getValue()), LoadType::CommandLine,
/*isLazy=*/false, /*isExplicit=*/true, /*isBundleLoader=*/false,
/*isForceHidden=*/true);
break;
case OPT_l:
case OPT_needed_l:
case OPT_reexport_l:
case OPT_weak_l:
case OPT_hidden_l:
addLibrary(arg->getValue(), opt.getID() == OPT_needed_l,
opt.getID() == OPT_weak_l, opt.getID() == OPT_reexport_l,
opt.getID() == OPT_hidden_l,
/*isExplicit=*/true, LoadType::CommandLine);
break;
case OPT_framework:
case OPT_needed_framework:
case OPT_reexport_framework:
case OPT_weak_framework:
addFramework(arg->getValue(), opt.getID() == OPT_needed_framework,
opt.getID() == OPT_weak_framework,
opt.getID() == OPT_reexport_framework, /*isExplicit=*/true,
LoadType::CommandLine);
break;
case OPT_start_lib:
if (isLazy)
error("nested --start-lib");
isLazy = true;
break;
case OPT_end_lib:
if (!isLazy)
error("stray --end-lib");
isLazy = false;
break;
default:
break;
}
}
}
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
static void gatherInputSections() {
TimeTraceScope timeScope("Gathering input sections");
int inputOrder = 0;
for (const InputFile *file : inputFiles) {
[lld-macho][nfc] Eliminate InputSection::Shared Earlier in LLD's evolution, I tried to create the illusion that subsections were indistinguishable from "top-level" sections. Thus, even though the subsections shared many common field values, I hid those common values away in a private Shared struct (see D105305). More recently, however, @gkm added a public `Section` struct in D113241 that served as an explicit way to store values that are common to an entire set of subsections (aka InputSections). Now that we have another "common value" struct, `Shared` has been rendered redundant. All its fields can be moved into `Section` instead, and the pointer to `Shared` can be replaced with a pointer to `Section`. This `Section` pointer also has the advantage of letting us inspect other subsections easily, simplifying the implementation of {D118798}. P.S. I do think that having both `Section` and `InputSection` makes for a slightly confusing naming scheme. I considered renaming `InputSection` to `Subsection`, but that would break the symmetry with `OutputSection`. It would also make us deviate from LLD-ELF's naming scheme. This change is perf-neutral on my 3.2 GHz 16-Core Intel Xeon W machine: base diff difference (95% CI) sys_time 1.258 ± 0.031 1.248 ± 0.023 [ -1.6% .. +0.1%] user_time 3.659 ± 0.047 3.658 ± 0.041 [ -0.5% .. +0.4%] wall_time 4.640 ± 0.085 4.625 ± 0.063 [ -1.0% .. +0.3%] samples 49 61 There's also no stat sig change in RSS (as measured by `time -l`): base diff difference (95% CI) time 998038627.097 ± 13567305.958 1003327715.556 ± 15210451.236 [ -0.2% .. +1.2%] samples 31 36 Reviewed By: #lld-macho, oontvoo Differential Revision: https://reviews.llvm.org/D118797
2022-02-04 08:53:29 +08:00
for (const Section *section : file->sections) {
// Compact unwind entries require special handling elsewhere. (In
// contrast, EH frames are handled like regular ConcatInputSections.)
if (section->name == section_names::compactUnwind)
continue;
2021-07-18 01:42:26 +08:00
ConcatOutputSection *osec = nullptr;
for (const Subsection &subsection : section->subsections) {
if (auto *isec = dyn_cast<ConcatInputSection>(subsection.isec)) {
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
if (isec->isCoalescedWeak())
continue;
isec->outSecOff = inputOrder++;
2021-07-18 01:42:26 +08:00
if (!osec)
osec = ConcatOutputSection::getOrCreateForInput(isec);
isec->parent = osec;
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
inputSections.push_back(isec);
} else if (auto *isec =
dyn_cast<CStringInputSection>(subsection.isec)) {
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
if (in.cStringSection->inputOrder == UnspecifiedInputOrder)
in.cStringSection->inputOrder = inputOrder++;
in.cStringSection->addInput(isec);
} else if (auto *isec =
dyn_cast<WordLiteralInputSection>(subsection.isec)) {
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
if (in.wordLiteralSection->inputOrder == UnspecifiedInputOrder)
in.wordLiteralSection->inputOrder = inputOrder++;
in.wordLiteralSection->addInput(isec);
} else {
llvm_unreachable("unexpected input section kind");
}
}
}
if (!file->objCImageInfo.empty())
in.objCImageInfo->addFile(file);
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
}
assert(inputOrder <= UnspecifiedInputOrder);
}
static void foldIdenticalLiterals() {
TimeTraceScope timeScope("Fold identical literals");
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
// We always create a cStringSection, regardless of whether dedupLiterals is
// true. If it isn't, we simply create a non-deduplicating CStringSection.
// Either way, we must unconditionally finalize it here.
in.cStringSection->finalizeContents();
if (in.wordLiteralSection)
in.wordLiteralSection->finalizeContents();
}
static void referenceStubBinder() {
bool needsStubHelper = config->outputType == MH_DYLIB ||
config->outputType == MH_EXECUTE ||
config->outputType == MH_BUNDLE;
if (!needsStubHelper || !symtab->find("dyld_stub_binder"))
return;
// dyld_stub_binder is used by dyld to resolve lazy bindings. This code here
// adds a opportunistic reference to dyld_stub_binder if it happens to exist.
// dyld_stub_binder is in libSystem.dylib, which is usually linked in. This
// isn't needed for correctness, but the presence of that symbol suppresses
// "no symbols" diagnostics from `nm`.
// StubHelperSection::setup() adds a reference and errors out if
// dyld_stub_binder doesn't exist in case it is actually needed.
symtab->addUndefined("dyld_stub_binder", /*file=*/nullptr, /*isWeak=*/false);
}
static void createAliases() {
for (const auto &pair : config->aliasedSymbols) {
if (const auto &sym = symtab->find(pair.first)) {
if (const auto &defined = dyn_cast<Defined>(sym)) {
symtab->aliasDefined(defined, pair.second);
continue;
}
}
warn("undefined base symbol '" + pair.first + "' for alias '" +
pair.second + "'\n");
}
}
static void handleExplicitExports() {
if (config->hasExplicitExports) {
parallelForEach(symtab->getSymbols(), [](Symbol *sym) {
if (auto *defined = dyn_cast<Defined>(sym)) {
StringRef symbolName = defined->getName();
if (config->exportedSymbols.match(symbolName)) {
if (defined->privateExtern) {
if (defined->weakDefCanBeHidden) {
// weak_def_can_be_hidden symbols behave similarly to
// private_extern symbols in most cases, except for when
// it is explicitly exported.
// The former can be exported but the latter cannot.
defined->privateExtern = false;
} else {
warn("cannot export hidden symbol " + toString(*defined) +
"\n>>> defined in " + toString(defined->getFile()));
}
}
} else {
defined->privateExtern = true;
}
}
});
} else if (!config->unexportedSymbols.empty()) {
parallelForEach(symtab->getSymbols(), [](Symbol *sym) {
if (auto *defined = dyn_cast<Defined>(sym))
if (config->unexportedSymbols.match(defined->getName()))
defined->privateExtern = true;
});
}
}
bool macho::link(ArrayRef<const char *> argsArr, llvm::raw_ostream &stdoutOS,
llvm::raw_ostream &stderrOS, bool exitEarly,
bool disableOutput) {
// This driver-specific context will be freed later by lldMain().
auto *ctx = new CommonLinkerContext;
ctx->e.initialize(stdoutOS, stderrOS, exitEarly, disableOutput);
ctx->e.cleanupCallback = []() {
resolvedFrameworks.clear();
resolvedLibraries.clear();
cachedReads.clear();
concatOutputSections.clear();
inputFiles.clear();
inputSections.clear();
loadedArchives.clear();
loadedObjectFrameworks.clear();
syntheticSections.clear();
thunkMap.clear();
firstTLVDataSection = nullptr;
tar = nullptr;
memset(&in, 0, sizeof(in));
resetLoadedDylibs();
resetOutputSegments();
resetWriter();
InputFile::resetIdCount();
};
ctx->e.logName = args::getFilenameWithoutExe(argsArr[0]);
MachOOptTable parser;
InputArgList args = parser.parse(argsArr.slice(1));
ctx->e.errorLimitExceededMsg = "too many errors emitted, stopping now "
"(use --error-limit=0 to see all errors)";
ctx->e.errorLimit = args::getInteger(args, OPT_error_limit_eq, 20);
ctx->e.verbose = args.hasArg(OPT_verbose);
if (args.hasArg(OPT_help_hidden)) {
parser.printHelp(argsArr[0], /*showHidden=*/true);
return true;
}
if (args.hasArg(OPT_help)) {
parser.printHelp(argsArr[0], /*showHidden=*/false);
return true;
}
if (args.hasArg(OPT_version)) {
message(getLLDVersion());
return true;
}
config = std::make_unique<Configuration>();
symtab = std::make_unique<SymbolTable>();
[lld/mac] Support writing zippered dylibs and bundles With -platform_version flags for two distinct platforms, this writes a LC_BUILD_VERSION header for each. The motivation is that this is needed for self-hosting with lld as linker after D124059. To create a zippered output at the clang driver level, pass -target arm64-apple-macos -darwin-target-variant arm64-apple-ios-macabi to create a zippered dylib. (In Xcode's clang, `-darwin-target-variant` is spelled just `-target-variant`.) (If you pass `-target arm64-apple-ios-macabi -target-variant arm64-apple-macos` instead, ld64 crashes!) This results in two -platform_version flags being passed to the linker. ld64 also verifies that the iOS SDK version is at least 13.1. We don't do that yet. But ld64 also does that for other platforms and we don't. So we need to do that at some point, but not in this patch. Only dylib and bundle outputs can be zippered. I verified that a Catalyst app linked against a dylib created with clang -shared foo.cc -o libfoo.dylib \ -target arm64-apple-macos \ -target-variant arm64-apple-ios-macabi \ -Wl,-install_name,@rpath/libfoo.dylib \ -fuse-ld=$PWD/out/gn/bin/ld64.lld runs successfully. (The app calls a function `f()` in libfoo.dylib that returns a const char* "foo", and NSLog(@"%s")s it.) ld64 is a bit more permissive when writing zippered outputs, see references to "unzippered twins". That's not implemented yet. (If anybody wants to implement that, D124275 is a good start.) Differential Revision: https://reviews.llvm.org/D124887
2022-04-22 23:55:50 +08:00
config->outputType = getOutputType(args);
target = createTargetInfo(args);
depTracker = std::make_unique<DependencyTracker>(
args.getLastArgValue(OPT_dependency_info));
if (errorCount())
return false;
if (args.hasArg(OPT_pagezero_size)) {
uint64_t pagezeroSize = args::getHex(args, OPT_pagezero_size, 0);
// ld64 does something really weird. It attempts to realign the value to the
// page size, but assumes the the page size is 4K. This doesn't work with
// most of Apple's ARM64 devices, which use a page size of 16K. This means
// that it will first 4K align it by rounding down, then round up to 16K.
// This probably only happened because no one using this arg with anything
// other then 0, so no one checked if it did what is what it says it does.
// So we are not copying this weird behavior and doing the it in a logical
// way, by always rounding down to page size.
if (!isAligned(Align(target->getPageSize()), pagezeroSize)) {
pagezeroSize -= pagezeroSize % target->getPageSize();
warn("__PAGEZERO size is not page aligned, rounding down to 0x" +
Twine::utohexstr(pagezeroSize));
}
target->pageZeroSize = pagezeroSize;
}
config->osoPrefix = args.getLastArgValue(OPT_oso_prefix);
if (!config->osoPrefix.empty()) {
// Expand special characters, such as ".", "..", or "~", if present.
// Note: LD64 only expands "." and not other special characters.
// That seems silly to imitate so we will not try to follow it, but rather
// just use real_path() to do it.
// The max path length is 4096, in theory. However that seems quite long
// and seems unlikely that any one would want to strip everything from the
// path. Hence we've picked a reasonably large number here.
SmallString<1024> expanded;
if (!fs::real_path(config->osoPrefix, expanded,
/*expand_tilde=*/true)) {
// Note: LD64 expands "." to be `<current_dir>/`
// (ie., it has a slash suffix) whereas real_path() doesn't.
// So we have to append '/' to be consistent.
StringRef sep = sys::path::get_separator();
// real_path removes trailing slashes as part of the normalization, but
// these are meaningful for our text based stripping
if (config->osoPrefix.equals(".") || config->osoPrefix.endswith(sep))
expanded += sep;
config->osoPrefix = saver().save(expanded.str());
}
}
[lld/mac] Implement -dead_strip Also adds support for live_support sections, no_dead_strip sections, .no_dead_strip symbols. Chromium Framework 345MB unstripped -> 250MB stripped (vs 290MB unstripped -> 236M stripped with ld64). Doing dead stripping is a bit faster than not, because so much less data needs to be processed: % ministat lld_* x lld_nostrip.txt + lld_strip.txt N Min Max Median Avg Stddev x 10 3.929414 4.07692 4.0269079 4.0089678 0.044214794 + 10 3.8129408 3.9025559 3.8670411 3.8642573 0.024779651 Difference at 95.0% confidence -0.144711 +/- 0.0336749 -3.60967% +/- 0.839989% (Student's t, pooled s = 0.0358398) This interacts with many parts of the linker. I tried to add test coverage for all added `isLive()` checks, so that some test will fail if any of them is removed. I checked that the test expectations for the most part match ld64's behavior (except for live-support-iterations.s, see the comment in the test). Interacts with: - debug info - export tries - import opcodes - flags like -exported_symbol(s_list) - -U / dynamic_lookup - mod_init_funcs, mod_term_funcs - weak symbol handling - unwind info - stubs - map files - -sectcreate - undefined, dylib, common, defined (both absolute and normal) symbols It's possible it interacts with more features I didn't think of, of course. I also did some manual testing: - check-llvm check-clang check-lld work with lld with this patch as host linker and -dead_strip enabled - Chromium still starts - Chromium's base_unittests still pass, including unwind tests Implemenation-wise, this is InputSection-based, so it'll work for object files with .subsections_via_symbols (which includes all object files generated by clang). I first based this on the COFF implementation, but later realized that things are more similar to ELF. I think it'd be good to refactor MarkLive.cpp to look more like the ELF part at some point, but I'd like to get a working state checked in first. Mechanical parts: - Rename canOmitFromOutput to wasCoalesced (no behavior change) since it really is for weak coalesced symbols - Add noDeadStrip to Defined, corresponding to N_NO_DEAD_STRIP (`.no_dead_strip` in asm) Fixes PR49276. Differential Revision: https://reviews.llvm.org/D103324
2021-05-08 05:10:05 +08:00
// Must be set before any InputSections and Symbols are created.
config->deadStrip = args.hasArg(OPT_dead_strip);
config->systemLibraryRoots = getSystemLibraryRoots(args);
if (const char *path = getReproduceOption(args)) {
// Note that --reproduce is a debug option so you can ignore it
// if you are trying to understand the whole picture of the code.
Expected<std::unique_ptr<TarWriter>> errOrWriter =
TarWriter::create(path, path::stem(path));
if (errOrWriter) {
tar = std::move(*errOrWriter);
tar->append("response.txt", createResponseFile(args));
tar->append("version.txt", getLLDVersion() + "\n");
} else {
error("--reproduce: " + toString(errOrWriter.takeError()));
}
}
if (auto *arg = args.getLastArg(OPT_threads_eq)) {
StringRef v(arg->getValue());
unsigned threads = 0;
if (!llvm::to_integer(v, threads, 0) || threads == 0)
error(arg->getSpelling() + ": expected a positive integer, but got '" +
arg->getValue() + "'");
parallel::strategy = hardware_concurrency(threads);
config->thinLTOJobs = v;
}
if (auto *arg = args.getLastArg(OPT_thinlto_jobs_eq))
config->thinLTOJobs = arg->getValue();
if (!get_threadpool_strategy(config->thinLTOJobs))
error("--thinlto-jobs: invalid job count: " + config->thinLTOJobs);
for (const Arg *arg : args.filtered(OPT_u)) {
config->explicitUndefineds.push_back(symtab->addUndefined(
arg->getValue(), /*file=*/nullptr, /*isWeakRef=*/false));
}
for (const Arg *arg : args.filtered(OPT_U))
config->explicitDynamicLookups.insert(arg->getValue());
config->mapFile = args.getLastArgValue(OPT_map);
config->optimize = args::getInteger(args, OPT_O, 1);
config->outputFile = args.getLastArgValue(OPT_o, "a.out");
config->finalOutput =
args.getLastArgValue(OPT_final_output, config->outputFile);
config->astPaths = args.getAllArgValues(OPT_add_ast_path);
config->headerPad = args::getHex(args, OPT_headerpad, /*Default=*/32);
config->headerPadMaxInstallNames =
args.hasArg(OPT_headerpad_max_install_names);
config->printDylibSearch =
args.hasArg(OPT_print_dylib_search) || getenv("RC_TRACE_DYLIB_SEARCHING");
config->printEachFile = args.hasArg(OPT_t);
config->printWhyLoad = args.hasArg(OPT_why_load);
config->omitDebugInfo = args.hasArg(OPT_S);
config->errorForArchMismatch = args.hasArg(OPT_arch_errors_fatal);
if (const Arg *arg = args.getLastArg(OPT_bundle_loader)) {
if (config->outputType != MH_BUNDLE)
error("-bundle_loader can only be used with MachO bundle output");
addFile(arg->getValue(), LoadType::CommandLine, /*isLazy=*/false,
/*isExplicit=*/false, /*isBundleLoader=*/true);
}
if (const Arg *arg = args.getLastArg(OPT_umbrella)) {
if (config->outputType != MH_DYLIB)
warn("-umbrella used, but not creating dylib");
config->umbrella = arg->getValue();
}
config->ltoObjPath = args.getLastArgValue(OPT_object_path_lto);
config->ltoo = args::getInteger(args, OPT_lto_O, 2);
if (config->ltoo > 3)
error("--lto-O: invalid optimization level: " + Twine(config->ltoo));
config->thinLTOCacheDir = args.getLastArgValue(OPT_cache_path_lto);
config->thinLTOCachePolicy = getLTOCachePolicy(args);
config->runtimePaths = args::getStrings(args, OPT_rpath);
config->allLoad = args.hasFlag(OPT_all_load, OPT_noall_load, false);
config->archMultiple = args.hasArg(OPT_arch_multiple);
config->applicationExtension = args.hasFlag(
OPT_application_extension, OPT_no_application_extension, false);
config->exportDynamic = args.hasArg(OPT_export_dynamic);
config->forceLoadObjC = args.hasArg(OPT_ObjC);
config->forceLoadSwift = args.hasArg(OPT_force_load_swift_libs);
config->deadStripDylibs = args.hasArg(OPT_dead_strip_dylibs);
config->demangle = args.hasArg(OPT_demangle);
config->implicitDylibs = !args.hasArg(OPT_no_implicit_dylibs);
config->emitFunctionStarts =
args.hasFlag(OPT_function_starts, OPT_no_function_starts, true);
config->emitBitcodeBundle = args.hasArg(OPT_bitcode_bundle);
config->emitDataInCodeInfo =
args.hasFlag(OPT_data_in_code_info, OPT_no_data_in_code_info, true);
config->icfLevel = getICFLevel(args);
config->dedupLiterals =
args.hasFlag(OPT_deduplicate_literals, OPT_icf_eq, false) ||
config->icfLevel != ICFLevel::none;
config->warnDylibInstallName = args.hasFlag(
OPT_warn_dylib_install_name, OPT_no_warn_dylib_install_name, false);
config->ignoreOptimizationHints = args.hasArg(OPT_ignore_optimization_hints);
config->callGraphProfileSort = args.hasFlag(
OPT_call_graph_profile_sort, OPT_no_call_graph_profile_sort, true);
config->printSymbolOrder = args.getLastArgValue(OPT_print_symbol_order);
config->forceExactCpuSubtypeMatch =
getenv("LD_DYLIB_CPU_SUBTYPES_MUST_MATCH");
for (const Arg *arg : args.filtered(OPT_alias)) {
config->aliasedSymbols.push_back(
std::make_pair(arg->getValue(0), arg->getValue(1)));
}
// FIXME: Add a commandline flag for this too.
config->zeroModTime = getenv("ZERO_AR_DATE");
std::array<PlatformType, 3> encryptablePlatforms{
PLATFORM_IOS, PLATFORM_WATCHOS, PLATFORM_TVOS};
config->emitEncryptionInfo =
args.hasFlag(OPT_encryptable, OPT_no_encryption,
is_contained(encryptablePlatforms, config->platform()));
#ifndef LLVM_HAVE_LIBXAR
if (config->emitBitcodeBundle)
error("-bitcode_bundle unsupported because LLD wasn't built with libxar");
#endif
if (const Arg *arg = args.getLastArg(OPT_install_name)) {
if (config->warnDylibInstallName && config->outputType != MH_DYLIB)
warn(
arg->getAsString(args) +
": ignored, only has effect with -dylib [--warn-dylib-install-name]");
else
config->installName = arg->getValue();
} else if (config->outputType == MH_DYLIB) {
config->installName = config->finalOutput;
}
if (args.hasArg(OPT_mark_dead_strippable_dylib)) {
if (config->outputType != MH_DYLIB)
warn("-mark_dead_strippable_dylib: ignored, only has effect with -dylib");
else
config->markDeadStrippableDylib = true;
}
if (const Arg *arg = args.getLastArg(OPT_static, OPT_dynamic))
config->staticLink = (arg->getOption().getID() == OPT_static);
if (const Arg *arg =
args.getLastArg(OPT_flat_namespace, OPT_twolevel_namespace))
config->namespaceKind = arg->getOption().getID() == OPT_twolevel_namespace
? NamespaceKind::twolevel
: NamespaceKind::flat;
config->undefinedSymbolTreatment = getUndefinedSymbolTreatment(args);
if (config->outputType == MH_EXECUTE)
config->entry = symtab->addUndefined(args.getLastArgValue(OPT_e, "_main"),
/*file=*/nullptr,
/*isWeakRef=*/false);
config->librarySearchPaths =
getLibrarySearchPaths(args, config->systemLibraryRoots);
config->frameworkSearchPaths =
getFrameworkSearchPaths(args, config->systemLibraryRoots);
if (const Arg *arg =
args.getLastArg(OPT_search_paths_first, OPT_search_dylibs_first))
config->searchDylibsFirst =
arg->getOption().getID() == OPT_search_dylibs_first;
config->dylibCompatibilityVersion =
parseDylibVersion(args, OPT_compatibility_version);
config->dylibCurrentVersion = parseDylibVersion(args, OPT_current_version);
config->dataConst =
args.hasFlag(OPT_data_const, OPT_no_data_const, dataConstDefault(args));
// Populate config->sectionRenameMap with builtin default renames.
// Options -rename_section and -rename_segment are able to override.
initializeSectionRenameMap();
// Reject every special character except '.' and '$'
// TODO(gkm): verify that this is the proper set of invalid chars
StringRef invalidNameChars("!\"#%&'()*+,-/:;<=>?@[\\]^`{|}~");
auto validName = [invalidNameChars](StringRef s) {
if (s.find_first_of(invalidNameChars) != StringRef::npos)
error("invalid name for segment or section: " + s);
return s;
};
for (const Arg *arg : args.filtered(OPT_rename_section)) {
config->sectionRenameMap[{validName(arg->getValue(0)),
validName(arg->getValue(1))}] = {
validName(arg->getValue(2)), validName(arg->getValue(3))};
}
for (const Arg *arg : args.filtered(OPT_rename_segment)) {
config->segmentRenameMap[validName(arg->getValue(0))] =
validName(arg->getValue(1));
}
config->sectionAlignments = parseSectAlign(args);
for (const Arg *arg : args.filtered(OPT_segprot)) {
StringRef segName = arg->getValue(0);
uint32_t maxProt = parseProtection(arg->getValue(1));
uint32_t initProt = parseProtection(arg->getValue(2));
if (maxProt != initProt && config->arch() != AK_i386)
error("invalid argument '" + arg->getAsString(args) +
"': max and init must be the same for non-i386 archs");
if (segName == segment_names::linkEdit)
error("-segprot cannot be used to change __LINKEDIT's protections");
config->segmentProtections.push_back({segName, maxProt, initProt});
}
config->hasExplicitExports =
args.hasArg(OPT_no_exported_symbols) ||
args.hasArgNoClaim(OPT_exported_symbol, OPT_exported_symbols_list);
handleSymbolPatterns(args, config->exportedSymbols, OPT_exported_symbol,
OPT_exported_symbols_list);
handleSymbolPatterns(args, config->unexportedSymbols, OPT_unexported_symbol,
OPT_unexported_symbols_list);
if (config->hasExplicitExports && !config->unexportedSymbols.empty())
error("cannot use both -exported_symbol* and -unexported_symbol* options");
if (args.hasArg(OPT_no_exported_symbols) && !config->exportedSymbols.empty())
error("cannot use both -exported_symbol* and -no_exported_symbols options");
// Imitating LD64's:
// -non_global_symbols_no_strip_list and -non_global_symbols_strip_list can't
// both be present.
// But -x can be used with either of these two, in which case, the last arg
// takes effect.
// (TODO: This is kind of confusing - considering disallowing using them
// together for a more straightforward behaviour)
{
bool includeLocal = false;
bool excludeLocal = false;
for (const Arg *arg :
args.filtered(OPT_x, OPT_non_global_symbols_no_strip_list,
OPT_non_global_symbols_strip_list)) {
switch (arg->getOption().getID()) {
case OPT_x:
config->localSymbolsPresence = SymtabPresence::None;
break;
case OPT_non_global_symbols_no_strip_list:
if (excludeLocal) {
error("cannot use both -non_global_symbols_no_strip_list and "
"-non_global_symbols_strip_list");
} else {
includeLocal = true;
config->localSymbolsPresence = SymtabPresence::SelectivelyIncluded;
parseSymbolPatternsFile(arg, config->localSymbolPatterns);
}
break;
case OPT_non_global_symbols_strip_list:
if (includeLocal) {
error("cannot use both -non_global_symbols_no_strip_list and "
"-non_global_symbols_strip_list");
} else {
excludeLocal = true;
config->localSymbolsPresence = SymtabPresence::SelectivelyExcluded;
parseSymbolPatternsFile(arg, config->localSymbolPatterns);
}
break;
default:
llvm_unreachable("unexpected option");
}
}
}
// Explicitly-exported literal symbols must be defined, but might
// languish in an archive if unreferenced elsewhere or if they are in the
// non-global strip list. Light a fire under those lazy symbols!
for (const CachedHashStringRef &cachedName : config->exportedSymbols.literals)
symtab->addUndefined(cachedName.val(), /*file=*/nullptr,
/*isWeakRef=*/false);
for (const Arg *arg : args.filtered(OPT_why_live))
config->whyLive.insert(arg->getValue());
if (!config->whyLive.empty() && !config->deadStrip)
warn("-why_live has no effect without -dead_strip, ignoring");
config->saveTemps = args.hasArg(OPT_save_temps);
config->adhocCodesign = args.hasFlag(
OPT_adhoc_codesign, OPT_no_adhoc_codesign,
(config->arch() == AK_arm64 || config->arch() == AK_arm64e) &&
config->platform() == PLATFORM_MACOS);
if (args.hasArg(OPT_v)) {
message(getLLDVersion(), lld::errs());
message(StringRef("Library search paths:") +
(config->librarySearchPaths.empty()
? ""
: "\n\t" + join(config->librarySearchPaths, "\n\t")),
lld::errs());
message(StringRef("Framework search paths:") +
(config->frameworkSearchPaths.empty()
? ""
: "\n\t" + join(config->frameworkSearchPaths, "\n\t")),
lld::errs());
}
config->progName = argsArr[0];
config->timeTraceEnabled = args.hasArg(OPT_time_trace_eq);
config->timeTraceGranularity =
args::getInteger(args, OPT_time_trace_granularity_eq, 500);
// Initialize time trace profiler.
if (config->timeTraceEnabled)
timeTraceProfilerInitialize(config->timeTraceGranularity, config->progName);
{
TimeTraceScope timeScope("ExecuteLinker");
initLLVM(); // must be run before any call to addFile()
createFiles(args);
config->isPic = config->outputType == MH_DYLIB ||
config->outputType == MH_BUNDLE ||
(config->outputType == MH_EXECUTE &&
args.hasFlag(OPT_pie, OPT_no_pie, true));
// Now that all dylibs have been loaded, search for those that should be
// re-exported.
{
auto reexportHandler = [](const Arg *arg,
const std::vector<StringRef> &extensions) {
config->hasReexports = true;
StringRef searchName = arg->getValue();
if (!markReexport(searchName, extensions))
error(arg->getSpelling() + " " + searchName +
" does not match a supplied dylib");
};
std::vector<StringRef> extensions = {".tbd"};
for (const Arg *arg : args.filtered(OPT_sub_umbrella))
reexportHandler(arg, extensions);
extensions.push_back(".dylib");
for (const Arg *arg : args.filtered(OPT_sub_library))
reexportHandler(arg, extensions);
}
cl::ResetAllOptionOccurrences();
// Parse LTO options.
if (const Arg *arg = args.getLastArg(OPT_mcpu))
parseClangOption(saver().save("-mcpu=" + StringRef(arg->getValue())),
arg->getSpelling());
for (const Arg *arg : args.filtered(OPT_mllvm))
parseClangOption(arg->getValue(), arg->getSpelling());
createSyntheticSections();
createSyntheticSymbols();
createAliases();
// If we are in "explicit exports" mode, hide everything that isn't
// explicitly exported. Do this before running LTO so that LTO can better
// optimize.
handleExplicitExports();
// LTO may emit a non-hidden (extern) object file symbol even if the
// corresponding bitcode symbol is hidden. In particular, this happens for
// cross-module references to hidden symbols under ThinLTO. Thus, if we
// compiled any bitcode files, we must redo the symbol hiding.
if (compileBitcodeFiles())
handleExplicitExports();
replaceCommonSymbols();
StringRef orderFile = args.getLastArgValue(OPT_order_file);
if (!orderFile.empty())
priorityBuilder.parseOrderFile(orderFile);
referenceStubBinder();
// FIXME: should terminate the link early based on errors encountered so
// far?
for (const Arg *arg : args.filtered(OPT_sectcreate)) {
StringRef segName = arg->getValue(0);
StringRef sectName = arg->getValue(1);
StringRef fileName = arg->getValue(2);
Optional<MemoryBufferRef> buffer = readFile(fileName);
if (buffer)
inputFiles.insert(make<OpaqueFile>(*buffer, segName, sectName));
}
for (const Arg *arg : args.filtered(OPT_add_empty_section)) {
StringRef segName = arg->getValue(0);
StringRef sectName = arg->getValue(1);
inputFiles.insert(make<OpaqueFile>(MemoryBufferRef(), segName, sectName));
}
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
gatherInputSections();
if (config->callGraphProfileSort)
priorityBuilder.extractCallGraphProfile();
[lld/mac] Implement -dead_strip Also adds support for live_support sections, no_dead_strip sections, .no_dead_strip symbols. Chromium Framework 345MB unstripped -> 250MB stripped (vs 290MB unstripped -> 236M stripped with ld64). Doing dead stripping is a bit faster than not, because so much less data needs to be processed: % ministat lld_* x lld_nostrip.txt + lld_strip.txt N Min Max Median Avg Stddev x 10 3.929414 4.07692 4.0269079 4.0089678 0.044214794 + 10 3.8129408 3.9025559 3.8670411 3.8642573 0.024779651 Difference at 95.0% confidence -0.144711 +/- 0.0336749 -3.60967% +/- 0.839989% (Student's t, pooled s = 0.0358398) This interacts with many parts of the linker. I tried to add test coverage for all added `isLive()` checks, so that some test will fail if any of them is removed. I checked that the test expectations for the most part match ld64's behavior (except for live-support-iterations.s, see the comment in the test). Interacts with: - debug info - export tries - import opcodes - flags like -exported_symbol(s_list) - -U / dynamic_lookup - mod_init_funcs, mod_term_funcs - weak symbol handling - unwind info - stubs - map files - -sectcreate - undefined, dylib, common, defined (both absolute and normal) symbols It's possible it interacts with more features I didn't think of, of course. I also did some manual testing: - check-llvm check-clang check-lld work with lld with this patch as host linker and -dead_strip enabled - Chromium still starts - Chromium's base_unittests still pass, including unwind tests Implemenation-wise, this is InputSection-based, so it'll work for object files with .subsections_via_symbols (which includes all object files generated by clang). I first based this on the COFF implementation, but later realized that things are more similar to ELF. I think it'd be good to refactor MarkLive.cpp to look more like the ELF part at some point, but I'd like to get a working state checked in first. Mechanical parts: - Rename canOmitFromOutput to wasCoalesced (no behavior change) since it really is for weak coalesced symbols - Add noDeadStrip to Defined, corresponding to N_NO_DEAD_STRIP (`.no_dead_strip` in asm) Fixes PR49276. Differential Revision: https://reviews.llvm.org/D103324
2021-05-08 05:10:05 +08:00
if (config->deadStrip)
markLive();
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
// ICF assumes that all literals have been folded already, so we must run
// foldIdenticalLiterals before foldIdenticalSections.
foldIdenticalLiterals();
if (config->icfLevel != ICFLevel::none) {
if (config->icfLevel == ICFLevel::safe)
markAddrSigSymbols();
foldIdenticalSections(/*onlyCfStrings=*/false);
} else if (config->dedupLiterals) {
foldIdenticalSections(/*onlyCfStrings=*/true);
}
[lld-macho] Move ICF earlier to avoid emitting redundant binds This is a pretty big refactoring diff, so here are the motivations: Previously, ICF ran after scanRelocations(), where we emitting bind/rebase opcodes etc. So we had a bunch of redundant leftovers after ICF. Having ICF run before Writer seems like a better design, and is what LLD-ELF does, so this diff refactors it accordingly. However, ICF had two dependencies on things occurring in Writer: 1) it needs literals to be deduplicated beforehand and 2) it needs to know which functions have unwind info, which was being handled by `UnwindInfoSection::prepareRelocations()`. In order to do literal deduplication earlier, we need to add literal input sections to their corresponding output sections. So instead of putting all input sections into the big `inputSections` vector, and then filtering them by type later on, I've changed things so that literal sections get added directly to their output sections during the 'gather' phase. Likewise for compact unwind sections -- they get added directly to the UnwindInfoSection now. This latter change is not strictly necessary, but makes it easier for ICF to determine which functions have unwind info. Adding literal sections directly to their output sections means that we can no longer determine `inputOrder` from iterating over `inputSections`. Instead, we store that order explicitly on InputSection. Bloating the size of InputSection for this purpose would be unfortunate -- but LLD-ELF has already solved this problem: it reuses `outSecOff` to store this order value. One downside of this refactor is that we now make an additional pass over the unwind info relocations to figure out which functions have unwind info, since want to know that before `processRelocations()`. I've made sure to run that extra loop only if ICF is enabled, so there should be no overhead in non-optimizing runs of the linker. The upside of all this is that the `inputSections` vector now contains only ConcatInputSections that are destined for ConcatOutputSections, so we can clean up a bunch of code that just existed to filter out other elements from that vector. I will test for the lack of redundant binds/rebases in the upcoming cfstring deduplication diff. While binds/rebases can also happen in the regular `.text` section, they're more common in `.data` sections, so it seems more natural to test it that way. This change is perf-neutral when linking chromium_framework. Reviewed By: oontvoo Differential Revision: https://reviews.llvm.org/D105044
2021-07-02 08:33:42 +08:00
// Write to an output file.
if (target->wordSize == 8)
writeResult<LP64>();
else
writeResult<ILP32>();
depTracker->write(getLLDVersion(), inputFiles, config->outputFile);
}
if (config->timeTraceEnabled) {
checkError(timeTraceProfilerWrite(
args.getLastArgValue(OPT_time_trace_eq).str(), config->outputFile));
timeTraceProfilerCleanup();
}
return errorCount() == 0;
}