llvm-capstone/lld/MachO/Writer.cpp

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

763 lines
24 KiB
C++
Raw Normal View History

//===- Writer.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 "Writer.h"
#include "Config.h"
#include "InputFiles.h"
#include "InputSection.h"
#include "MergedOutputSection.h"
#include "OutputSection.h"
#include "OutputSegment.h"
#include "SymbolTable.h"
#include "Symbols.h"
#include "SyntheticSections.h"
#include "Target.h"
#include "UnwindInfoSection.h"
#include "lld/Common/ErrorHandler.h"
#include "lld/Common/Memory.h"
#include "llvm/BinaryFormat/MachO.h"
#include "llvm/Config/llvm-config.h"
#include "llvm/Support/LEB128.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/Path.h"
#include "llvm/Support/xxhash.h"
#include <algorithm>
using namespace llvm;
using namespace llvm::MachO;
using namespace lld;
using namespace lld::macho;
namespace {
class LCUuid;
class Writer {
public:
Writer() : buffer(errorHandler().outputBuffer) {}
void scanRelocations();
void createOutputSections();
void createLoadCommands();
void assignAddresses(OutputSegment *);
void openFile();
void writeSections();
void writeUuid();
void run();
std::unique_ptr<FileOutputBuffer> &buffer;
uint64_t addr = 0;
uint64_t fileOff = 0;
MachHeaderSection *header = nullptr;
[lld-macho][reland] Add basic symbol table output This diff implements basic support for writing a symbol table. Attributes are loosely supported for extern symbols and not at all for other types. Initial version by Kellie Medlin <kelliem@fb.com> Originally committed in a3d95a50ee33 and reverted in fbae153ca583 due to UBSAN erroring over unaligned writes. That has been fixed in the current diff with the following changes: ``` diff --git a/lld/MachO/SyntheticSections.cpp b/lld/MachO/SyntheticSections.cpp --- a/lld/MachO/SyntheticSections.cpp +++ b/lld/MachO/SyntheticSections.cpp @@ -133,6 +133,9 @@ SymtabSection::SymtabSection(StringTableSection &stringTableSection) : stringTableSection(stringTableSection) { segname = segment_names::linkEdit; name = section_names::symbolTable; + // TODO: When we introduce the SyntheticSections superclass, we should make + // all synthetic sections aligned to WordSize by default. + align = WordSize; } size_t SymtabSection::getSize() const { diff --git a/lld/MachO/Writer.cpp b/lld/MachO/Writer.cpp --- a/lld/MachO/Writer.cpp +++ b/lld/MachO/Writer.cpp @@ -371,6 +371,7 @@ void Writer::assignAddresses(OutputSegment *seg) { ArrayRef<InputSection *> sections = p.second; for (InputSection *isec : sections) { addr = alignTo(addr, isec->align); + // We must align the file offsets too to avoid misaligned writes of + // structs. + fileOff = alignTo(fileOff, isec->align); isec->addr = addr; addr += isec->getSize(); fileOff += isec->getFileSize(); @@ -396,6 +397,7 @@ void Writer::writeSections() { uint64_t fileOff = seg->fileOff; for (auto &sect : seg->getSections()) { for (InputSection *isec : sect.second) { + fileOff = alignTo(fileOff, isec->align); isec->writeTo(buf + fileOff); fileOff += isec->getFileSize(); } ``` I don't think it's easy to write a test for alignment (that doesn't involve brittly hard-coding file offsets), so there isn't one... but UBSAN builds pass now. Differential Revision: https://reviews.llvm.org/D79050
2020-04-28 23:58:19 +00:00
StringTableSection *stringTableSection = nullptr;
SymtabSection *symtabSection = nullptr;
IndirectSymtabSection *indirectSymtabSection = nullptr;
UnwindInfoSection *unwindInfoSection = nullptr;
LCUuid *uuidCommand = nullptr;
};
// LC_DYLD_INFO_ONLY stores the offsets of symbol import/export information.
class LCDyldInfo : public LoadCommand {
public:
LCDyldInfo(RebaseSection *rebaseSection, BindingSection *bindingSection,
WeakBindingSection *weakBindingSection,
LazyBindingSection *lazyBindingSection,
ExportSection *exportSection)
: rebaseSection(rebaseSection), bindingSection(bindingSection),
weakBindingSection(weakBindingSection),
lazyBindingSection(lazyBindingSection), exportSection(exportSection) {}
uint32_t getSize() const override { return sizeof(dyld_info_command); }
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<dyld_info_command *>(buf);
c->cmd = LC_DYLD_INFO_ONLY;
c->cmdsize = getSize();
if (rebaseSection->isNeeded()) {
c->rebase_off = rebaseSection->fileOff;
c->rebase_size = rebaseSection->getFileSize();
}
if (bindingSection->isNeeded()) {
c->bind_off = bindingSection->fileOff;
c->bind_size = bindingSection->getFileSize();
}
if (weakBindingSection->isNeeded()) {
c->weak_bind_off = weakBindingSection->fileOff;
c->weak_bind_size = weakBindingSection->getFileSize();
}
if (lazyBindingSection->isNeeded()) {
c->lazy_bind_off = lazyBindingSection->fileOff;
c->lazy_bind_size = lazyBindingSection->getFileSize();
}
if (exportSection->isNeeded()) {
c->export_off = exportSection->fileOff;
c->export_size = exportSection->getFileSize();
}
}
RebaseSection *rebaseSection;
BindingSection *bindingSection;
WeakBindingSection *weakBindingSection;
LazyBindingSection *lazyBindingSection;
ExportSection *exportSection;
};
class LCDysymtab : public LoadCommand {
public:
LCDysymtab(SymtabSection *symtabSection,
IndirectSymtabSection *indirectSymtabSection)
: symtabSection(symtabSection),
indirectSymtabSection(indirectSymtabSection) {}
uint32_t getSize() const override { return sizeof(dysymtab_command); }
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<dysymtab_command *>(buf);
c->cmd = LC_DYSYMTAB;
c->cmdsize = getSize();
c->ilocalsym = 0;
c->iextdefsym = c->nlocalsym = symtabSection->getNumLocalSymbols();
c->nextdefsym = symtabSection->getNumExternalSymbols();
c->iundefsym = c->iextdefsym + c->nextdefsym;
c->nundefsym = symtabSection->getNumUndefinedSymbols();
c->indirectsymoff = indirectSymtabSection->fileOff;
c->nindirectsyms = indirectSymtabSection->getNumSymbols();
}
SymtabSection *symtabSection;
IndirectSymtabSection *indirectSymtabSection;
};
class LCSegment : public LoadCommand {
public:
LCSegment(StringRef name, OutputSegment *seg) : name(name), seg(seg) {}
uint32_t getSize() const override {
return sizeof(segment_command_64) +
seg->numNonHiddenSections() * sizeof(section_64);
}
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<segment_command_64 *>(buf);
buf += sizeof(segment_command_64);
c->cmd = LC_SEGMENT_64;
c->cmdsize = getSize();
memcpy(c->segname, name.data(), name.size());
c->fileoff = seg->fileOff;
c->maxprot = seg->maxProt;
c->initprot = seg->initProt;
if (seg->getSections().empty())
return;
c->vmaddr = seg->firstSection()->addr;
c->vmsize =
seg->lastSection()->addr + seg->lastSection()->getSize() - c->vmaddr;
c->nsects = seg->numNonHiddenSections();
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
for (OutputSection *osec : seg->getSections()) {
if (!isZeroFill(osec->flags)) {
assert(osec->fileOff >= seg->fileOff);
c->filesize = std::max(
c->filesize, osec->fileOff + osec->getFileSize() - seg->fileOff);
}
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
if (osec->isHidden())
continue;
auto *sectHdr = reinterpret_cast<section_64 *>(buf);
buf += sizeof(section_64);
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
memcpy(sectHdr->sectname, osec->name.data(), osec->name.size());
memcpy(sectHdr->segname, name.data(), name.size());
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
sectHdr->addr = osec->addr;
sectHdr->offset = osec->fileOff;
sectHdr->align = Log2_32(osec->align);
sectHdr->flags = osec->flags;
sectHdr->size = osec->getSize();
sectHdr->reserved1 = osec->reserved1;
sectHdr->reserved2 = osec->reserved2;
}
}
private:
StringRef name;
OutputSegment *seg;
};
class LCMain : public LoadCommand {
uint32_t getSize() const override { return sizeof(entry_point_command); }
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<entry_point_command *>(buf);
c->cmd = LC_MAIN;
c->cmdsize = getSize();
if (config->entry->isInStubs())
c->entryoff =
in.stubs->fileOff + config->entry->stubsIndex * target->stubSize;
else
c->entryoff = config->entry->getFileOffset();
c->stacksize = 0;
}
};
class LCSymtab : public LoadCommand {
public:
[lld-macho][reland] Add basic symbol table output This diff implements basic support for writing a symbol table. Attributes are loosely supported for extern symbols and not at all for other types. Initial version by Kellie Medlin <kelliem@fb.com> Originally committed in a3d95a50ee33 and reverted in fbae153ca583 due to UBSAN erroring over unaligned writes. That has been fixed in the current diff with the following changes: ``` diff --git a/lld/MachO/SyntheticSections.cpp b/lld/MachO/SyntheticSections.cpp --- a/lld/MachO/SyntheticSections.cpp +++ b/lld/MachO/SyntheticSections.cpp @@ -133,6 +133,9 @@ SymtabSection::SymtabSection(StringTableSection &stringTableSection) : stringTableSection(stringTableSection) { segname = segment_names::linkEdit; name = section_names::symbolTable; + // TODO: When we introduce the SyntheticSections superclass, we should make + // all synthetic sections aligned to WordSize by default. + align = WordSize; } size_t SymtabSection::getSize() const { diff --git a/lld/MachO/Writer.cpp b/lld/MachO/Writer.cpp --- a/lld/MachO/Writer.cpp +++ b/lld/MachO/Writer.cpp @@ -371,6 +371,7 @@ void Writer::assignAddresses(OutputSegment *seg) { ArrayRef<InputSection *> sections = p.second; for (InputSection *isec : sections) { addr = alignTo(addr, isec->align); + // We must align the file offsets too to avoid misaligned writes of + // structs. + fileOff = alignTo(fileOff, isec->align); isec->addr = addr; addr += isec->getSize(); fileOff += isec->getFileSize(); @@ -396,6 +397,7 @@ void Writer::writeSections() { uint64_t fileOff = seg->fileOff; for (auto &sect : seg->getSections()) { for (InputSection *isec : sect.second) { + fileOff = alignTo(fileOff, isec->align); isec->writeTo(buf + fileOff); fileOff += isec->getFileSize(); } ``` I don't think it's easy to write a test for alignment (that doesn't involve brittly hard-coding file offsets), so there isn't one... but UBSAN builds pass now. Differential Revision: https://reviews.llvm.org/D79050
2020-04-28 23:58:19 +00:00
LCSymtab(SymtabSection *symtabSection, StringTableSection *stringTableSection)
: symtabSection(symtabSection), stringTableSection(stringTableSection) {}
uint32_t getSize() const override { return sizeof(symtab_command); }
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<symtab_command *>(buf);
c->cmd = LC_SYMTAB;
c->cmdsize = getSize();
c->symoff = symtabSection->fileOff;
[lld-macho][reland] Add basic symbol table output This diff implements basic support for writing a symbol table. Attributes are loosely supported for extern symbols and not at all for other types. Initial version by Kellie Medlin <kelliem@fb.com> Originally committed in a3d95a50ee33 and reverted in fbae153ca583 due to UBSAN erroring over unaligned writes. That has been fixed in the current diff with the following changes: ``` diff --git a/lld/MachO/SyntheticSections.cpp b/lld/MachO/SyntheticSections.cpp --- a/lld/MachO/SyntheticSections.cpp +++ b/lld/MachO/SyntheticSections.cpp @@ -133,6 +133,9 @@ SymtabSection::SymtabSection(StringTableSection &stringTableSection) : stringTableSection(stringTableSection) { segname = segment_names::linkEdit; name = section_names::symbolTable; + // TODO: When we introduce the SyntheticSections superclass, we should make + // all synthetic sections aligned to WordSize by default. + align = WordSize; } size_t SymtabSection::getSize() const { diff --git a/lld/MachO/Writer.cpp b/lld/MachO/Writer.cpp --- a/lld/MachO/Writer.cpp +++ b/lld/MachO/Writer.cpp @@ -371,6 +371,7 @@ void Writer::assignAddresses(OutputSegment *seg) { ArrayRef<InputSection *> sections = p.second; for (InputSection *isec : sections) { addr = alignTo(addr, isec->align); + // We must align the file offsets too to avoid misaligned writes of + // structs. + fileOff = alignTo(fileOff, isec->align); isec->addr = addr; addr += isec->getSize(); fileOff += isec->getFileSize(); @@ -396,6 +397,7 @@ void Writer::writeSections() { uint64_t fileOff = seg->fileOff; for (auto &sect : seg->getSections()) { for (InputSection *isec : sect.second) { + fileOff = alignTo(fileOff, isec->align); isec->writeTo(buf + fileOff); fileOff += isec->getFileSize(); } ``` I don't think it's easy to write a test for alignment (that doesn't involve brittly hard-coding file offsets), so there isn't one... but UBSAN builds pass now. Differential Revision: https://reviews.llvm.org/D79050
2020-04-28 23:58:19 +00:00
c->nsyms = symtabSection->getNumSymbols();
c->stroff = stringTableSection->fileOff;
[lld-macho][reland] Add basic symbol table output This diff implements basic support for writing a symbol table. Attributes are loosely supported for extern symbols and not at all for other types. Initial version by Kellie Medlin <kelliem@fb.com> Originally committed in a3d95a50ee33 and reverted in fbae153ca583 due to UBSAN erroring over unaligned writes. That has been fixed in the current diff with the following changes: ``` diff --git a/lld/MachO/SyntheticSections.cpp b/lld/MachO/SyntheticSections.cpp --- a/lld/MachO/SyntheticSections.cpp +++ b/lld/MachO/SyntheticSections.cpp @@ -133,6 +133,9 @@ SymtabSection::SymtabSection(StringTableSection &stringTableSection) : stringTableSection(stringTableSection) { segname = segment_names::linkEdit; name = section_names::symbolTable; + // TODO: When we introduce the SyntheticSections superclass, we should make + // all synthetic sections aligned to WordSize by default. + align = WordSize; } size_t SymtabSection::getSize() const { diff --git a/lld/MachO/Writer.cpp b/lld/MachO/Writer.cpp --- a/lld/MachO/Writer.cpp +++ b/lld/MachO/Writer.cpp @@ -371,6 +371,7 @@ void Writer::assignAddresses(OutputSegment *seg) { ArrayRef<InputSection *> sections = p.second; for (InputSection *isec : sections) { addr = alignTo(addr, isec->align); + // We must align the file offsets too to avoid misaligned writes of + // structs. + fileOff = alignTo(fileOff, isec->align); isec->addr = addr; addr += isec->getSize(); fileOff += isec->getFileSize(); @@ -396,6 +397,7 @@ void Writer::writeSections() { uint64_t fileOff = seg->fileOff; for (auto &sect : seg->getSections()) { for (InputSection *isec : sect.second) { + fileOff = alignTo(fileOff, isec->align); isec->writeTo(buf + fileOff); fileOff += isec->getFileSize(); } ``` I don't think it's easy to write a test for alignment (that doesn't involve brittly hard-coding file offsets), so there isn't one... but UBSAN builds pass now. Differential Revision: https://reviews.llvm.org/D79050
2020-04-28 23:58:19 +00:00
c->strsize = stringTableSection->getFileSize();
}
[lld-macho][reland] Add basic symbol table output This diff implements basic support for writing a symbol table. Attributes are loosely supported for extern symbols and not at all for other types. Initial version by Kellie Medlin <kelliem@fb.com> Originally committed in a3d95a50ee33 and reverted in fbae153ca583 due to UBSAN erroring over unaligned writes. That has been fixed in the current diff with the following changes: ``` diff --git a/lld/MachO/SyntheticSections.cpp b/lld/MachO/SyntheticSections.cpp --- a/lld/MachO/SyntheticSections.cpp +++ b/lld/MachO/SyntheticSections.cpp @@ -133,6 +133,9 @@ SymtabSection::SymtabSection(StringTableSection &stringTableSection) : stringTableSection(stringTableSection) { segname = segment_names::linkEdit; name = section_names::symbolTable; + // TODO: When we introduce the SyntheticSections superclass, we should make + // all synthetic sections aligned to WordSize by default. + align = WordSize; } size_t SymtabSection::getSize() const { diff --git a/lld/MachO/Writer.cpp b/lld/MachO/Writer.cpp --- a/lld/MachO/Writer.cpp +++ b/lld/MachO/Writer.cpp @@ -371,6 +371,7 @@ void Writer::assignAddresses(OutputSegment *seg) { ArrayRef<InputSection *> sections = p.second; for (InputSection *isec : sections) { addr = alignTo(addr, isec->align); + // We must align the file offsets too to avoid misaligned writes of + // structs. + fileOff = alignTo(fileOff, isec->align); isec->addr = addr; addr += isec->getSize(); fileOff += isec->getFileSize(); @@ -396,6 +397,7 @@ void Writer::writeSections() { uint64_t fileOff = seg->fileOff; for (auto &sect : seg->getSections()) { for (InputSection *isec : sect.second) { + fileOff = alignTo(fileOff, isec->align); isec->writeTo(buf + fileOff); fileOff += isec->getFileSize(); } ``` I don't think it's easy to write a test for alignment (that doesn't involve brittly hard-coding file offsets), so there isn't one... but UBSAN builds pass now. Differential Revision: https://reviews.llvm.org/D79050
2020-04-28 23:58:19 +00:00
SymtabSection *symtabSection = nullptr;
StringTableSection *stringTableSection = nullptr;
};
// There are several dylib load commands that share the same structure:
// * LC_LOAD_DYLIB
// * LC_ID_DYLIB
// * LC_REEXPORT_DYLIB
class LCDylib : public LoadCommand {
public:
LCDylib(LoadCommandType type, StringRef path,
uint32_t compatibilityVersion = 0, uint32_t currentVersion = 0)
: type(type), path(path), compatibilityVersion(compatibilityVersion),
currentVersion(currentVersion) {
instanceCount++;
}
uint32_t getSize() const override {
return alignTo(sizeof(dylib_command) + path.size() + 1, 8);
}
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<dylib_command *>(buf);
buf += sizeof(dylib_command);
c->cmd = type;
c->cmdsize = getSize();
c->dylib.name = sizeof(dylib_command);
c->dylib.timestamp = 0;
c->dylib.compatibility_version = compatibilityVersion;
c->dylib.current_version = currentVersion;
memcpy(buf, path.data(), path.size());
buf[path.size()] = '\0';
}
static uint32_t getInstanceCount() { return instanceCount; }
private:
LoadCommandType type;
StringRef path;
uint32_t compatibilityVersion;
uint32_t currentVersion;
static uint32_t instanceCount;
};
uint32_t LCDylib::instanceCount = 0;
class LCLoadDylinker : public LoadCommand {
public:
uint32_t getSize() const override {
return alignTo(sizeof(dylinker_command) + path.size() + 1, 8);
}
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<dylinker_command *>(buf);
buf += sizeof(dylinker_command);
c->cmd = LC_LOAD_DYLINKER;
c->cmdsize = getSize();
c->name = sizeof(dylinker_command);
memcpy(buf, path.data(), path.size());
buf[path.size()] = '\0';
}
private:
// Recent versions of Darwin won't run any binary that has dyld at a
// different location.
const StringRef path = "/usr/lib/dyld";
};
class LCRPath : public LoadCommand {
public:
LCRPath(StringRef path) : path(path) {}
uint32_t getSize() const override {
return alignTo(sizeof(rpath_command) + path.size() + 1, WordSize);
}
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<rpath_command *>(buf);
buf += sizeof(rpath_command);
c->cmd = LC_RPATH;
c->cmdsize = getSize();
c->path = sizeof(rpath_command);
memcpy(buf, path.data(), path.size());
buf[path.size()] = '\0';
}
private:
StringRef path;
};
class LCBuildVersion : public LoadCommand {
public:
LCBuildVersion(const PlatformInfo &platform) : platform(platform) {}
const int ntools = 1;
uint32_t getSize() const override {
return sizeof(build_version_command) + ntools * sizeof(build_tool_version);
}
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<build_version_command *>(buf);
c->cmd = LC_BUILD_VERSION;
c->cmdsize = getSize();
c->platform = static_cast<uint32_t>(platform.kind);
c->minos = ((platform.minimum.getMajor() << 020) |
(platform.minimum.getMinor().getValueOr(0) << 010) |
platform.minimum.getSubminor().getValueOr(0));
c->sdk = ((platform.sdk.getMajor() << 020) |
(platform.sdk.getMinor().getValueOr(0) << 010) |
platform.sdk.getSubminor().getValueOr(0));
c->ntools = ntools;
auto *t = reinterpret_cast<build_tool_version *>(&c[1]);
t->tool = TOOL_LD;
t->version = (LLVM_VERSION_MAJOR << 020) | (LLVM_VERSION_MINOR << 010) |
LLVM_VERSION_PATCH;
}
const PlatformInfo &platform;
};
// Stores a unique identifier for the output file based on an MD5 hash of its
// contents. In order to hash the contents, we must first write them, but
// LC_UUID itself must be part of the written contents in order for all the
// offsets to be calculated correctly. We resolve this circular paradox by
// first writing an LC_UUID with an all-zero UUID, then updating the UUID with
// its real value later.
class LCUuid : public LoadCommand {
public:
uint32_t getSize() const override { return sizeof(uuid_command); }
void writeTo(uint8_t *buf) const override {
auto *c = reinterpret_cast<uuid_command *>(buf);
c->cmd = LC_UUID;
c->cmdsize = getSize();
uuidBuf = c->uuid;
}
void writeUuid(uint64_t digest) const {
// xxhash only gives us 8 bytes, so put some fixed data in the other half.
static_assert(sizeof(uuid_command::uuid) == 16, "unexpected uuid size");
memcpy(uuidBuf, "LLD\xa1UU1D", 8);
memcpy(uuidBuf + 8, &digest, 8);
// RFC 4122 conformance. We need to fix 4 bits in byte 6 and 2 bits in
// byte 8. Byte 6 is already fine due to the fixed data we put in. We don't
// want to lose bits of the digest in byte 8, so swap that with a byte of
// fixed data that happens to have the right bits set.
std::swap(uuidBuf[3], uuidBuf[8]);
// Claim that this is an MD5-based hash. It isn't, but this signals that
// this is not a time-based and not a random hash. MD5 seems like the least
// bad lie we can put here.
assert((uuidBuf[6] & 0xf0) == 0x30 && "See RFC 4122 Sections 4.2.2, 4.1.3");
assert((uuidBuf[8] & 0xc0) == 0x80 && "See RFC 4122 Section 4.2.2");
}
mutable uint8_t *uuidBuf;
};
} // namespace
void Writer::scanRelocations() {
for (InputSection *isec : inputSections) {
// We do not wish to add rebase opcodes for __LD,__compact_unwind, because
// it doesn't actually end up in the final binary. TODO: filtering it out
// before Writer runs might be cleaner...
if (isec->segname == segment_names::ld)
continue;
for (Reloc &r : isec->relocs) {
2020-09-23 03:42:12 +00:00
if (auto *s = r.referent.dyn_cast<lld::macho::Symbol *>()) {
if (isa<Undefined>(s))
error("undefined symbol " + toString(*s) + ", referenced from " +
toString(isec->file));
else
target->prepareSymbolRelocation(s, isec, r);
} else {
assert(r.referent.is<InputSection *>());
if (!r.pcrel)
in.rebase->addEntry(isec, r.offset);
}
}
}
}
void Writer::createLoadCommands() {
in.header->addLoadCommand(make<LCDyldInfo>(
in.rebase, in.binding, in.weakBinding, in.lazyBinding, in.exports));
in.header->addLoadCommand(make<LCSymtab>(symtabSection, stringTableSection));
in.header->addLoadCommand(
make<LCDysymtab>(symtabSection, indirectSymtabSection));
for (StringRef path : config->runtimePaths)
in.header->addLoadCommand(make<LCRPath>(path));
switch (config->outputType) {
case MH_EXECUTE:
in.header->addLoadCommand(make<LCMain>());
in.header->addLoadCommand(make<LCLoadDylinker>());
break;
case MH_DYLIB:
in.header->addLoadCommand(make<LCDylib>(LC_ID_DYLIB, config->installName,
config->dylibCompatibilityVersion,
config->dylibCurrentVersion));
break;
case MH_BUNDLE:
break;
default:
llvm_unreachable("unhandled output file type");
}
in.header->addLoadCommand(make<LCBuildVersion>(config->platform));
uuidCommand = make<LCUuid>();
in.header->addLoadCommand(uuidCommand);
uint8_t segIndex = 0;
for (OutputSegment *seg : outputSegments) {
in.header->addLoadCommand(make<LCSegment>(seg->name, seg));
seg->index = segIndex++;
}
uint64_t dylibOrdinal = 1;
for (InputFile *file : inputFiles) {
if (auto *dylibFile = dyn_cast<DylibFile>(file)) {
// TODO: dylibs that are only referenced by weak refs should also be
// loaded via LC_LOAD_WEAK_DYLIB.
LoadCommandType lcType =
dylibFile->forceWeakImport ? LC_LOAD_WEAK_DYLIB : LC_LOAD_DYLIB;
in.header->addLoadCommand(make<LCDylib>(lcType, dylibFile->dylibName));
dylibFile->ordinal = dylibOrdinal++;
if (dylibFile->reexport)
in.header->addLoadCommand(
make<LCDylib>(LC_REEXPORT_DYLIB, dylibFile->dylibName));
}
}
const uint32_t MACOS_MAXPATHLEN = 1024;
config->headerPad = std::max(
config->headerPad, (config->headerPadMaxInstallNames
? LCDylib::getInstanceCount() * MACOS_MAXPATHLEN
: 0));
}
static size_t getSymbolPriority(const SymbolPriorityEntry &entry,
const InputFile &file) {
return std::max(entry.objectFiles.lookup(sys::path::filename(file.getName())),
entry.anyObjectFile);
}
// Each section gets assigned the priority of the highest-priority symbol it
// contains.
static DenseMap<const InputSection *, size_t> buildInputSectionPriorities() {
DenseMap<const InputSection *, size_t> sectionPriorities;
if (config->priorities.empty())
return sectionPriorities;
auto addSym = [&](Defined &sym) {
auto it = config->priorities.find(sym.getName());
if (it == config->priorities.end())
return;
SymbolPriorityEntry &entry = it->second;
size_t &priority = sectionPriorities[sym.isec];
priority = std::max(priority, getSymbolPriority(entry, *sym.isec->file));
};
// TODO: Make sure this handles weak symbols correctly.
for (InputFile *file : inputFiles)
if (isa<ObjFile>(file) || isa<ArchiveFile>(file))
for (lld::macho::Symbol *sym : file->symbols)
if (auto *d = dyn_cast<Defined>(sym))
addSym(*d);
return sectionPriorities;
}
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
static int segmentOrder(OutputSegment *seg) {
return StringSwitch<int>(seg->name)
.Case(segment_names::pageZero, -2)
.Case(segment_names::text, -1)
// Make sure __LINKEDIT is the last segment (i.e. all its hidden
// sections must be ordered after other sections).
.Case(segment_names::linkEdit, std::numeric_limits<int>::max())
.Default(0);
}
static int sectionOrder(OutputSection *osec) {
StringRef segname = osec->parent->name;
// Sections are uniquely identified by their segment + section name.
if (segname == segment_names::text) {
return StringSwitch<int>(osec->name)
.Case(section_names::header, -1)
.Case(section_names::unwindInfo, std::numeric_limits<int>::max() - 1)
.Case(section_names::ehFrame, std::numeric_limits<int>::max())
.Default(0);
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
} else if (segname == segment_names::linkEdit) {
return StringSwitch<int>(osec->name)
.Case(section_names::rebase, -8)
.Case(section_names::binding, -7)
.Case(section_names::weakBinding, -6)
.Case(section_names::lazyBinding, -5)
.Case(section_names::export_, -4)
.Case(section_names::symbolTable, -3)
.Case(section_names::indirectSymbolTable, -2)
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
.Case(section_names::stringTable, -1)
.Default(0);
}
// ZeroFill sections must always be the at the end of their segments,
// otherwise subsequent sections may get overwritten with zeroes at runtime.
if (isZeroFill(osec->flags))
return std::numeric_limits<int>::max();
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
return 0;
}
template <typename T, typename F>
static std::function<bool(T, T)> compareByOrder(F ord) {
return [=](T a, T b) { return ord(a) < ord(b); };
}
// Sorting only can happen once all outputs have been collected. Here we sort
// segments, output sections within each segment, and input sections within each
// output segment.
static void sortSegmentsAndSections() {
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
llvm::stable_sort(outputSegments,
compareByOrder<OutputSegment *>(segmentOrder));
DenseMap<const InputSection *, size_t> isecPriorities =
buildInputSectionPriorities();
uint32_t sectionIndex = 0;
for (OutputSegment *seg : outputSegments) {
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
seg->sortOutputSections(compareByOrder<OutputSection *>(sectionOrder));
for (auto *osec : seg->getSections()) {
// Now that the output sections are sorted, assign the final
// output section indices.
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
if (!osec->isHidden())
osec->index = ++sectionIndex;
if (!isecPriorities.empty()) {
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
if (auto *merged = dyn_cast<MergedOutputSection>(osec)) {
llvm::stable_sort(merged->inputs,
[&](InputSection *a, InputSection *b) {
return isecPriorities[a] > isecPriorities[b];
});
}
}
}
}
}
void Writer::createOutputSections() {
// First, create hidden sections
stringTableSection = make<StringTableSection>();
unwindInfoSection = make<UnwindInfoSection>(); // TODO(gkm): only when no -r
symtabSection = make<SymtabSection>(*stringTableSection);
indirectSymtabSection = make<IndirectSymtabSection>();
switch (config->outputType) {
case MH_EXECUTE:
make<PageZeroSection>();
break;
case MH_DYLIB:
case MH_BUNDLE:
break;
default:
llvm_unreachable("unhandled output file type");
}
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
// Then merge input sections into output sections.
MapVector<std::pair<StringRef, StringRef>, MergedOutputSection *>
mergedOutputSections;
for (InputSection *isec : inputSections) {
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
MergedOutputSection *&osec =
mergedOutputSections[{isec->segname, isec->name}];
if (osec == nullptr)
osec = make<MergedOutputSection>(isec->name);
osec->mergeInput(isec);
}
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
for (const auto &it : mergedOutputSections) {
StringRef segname = it.first.first;
MergedOutputSection *osec = it.second;
if (unwindInfoSection && segname == segment_names::ld) {
assert(osec->name == section_names::compactUnwind);
unwindInfoSection->setCompactUnwindSection(osec);
[lld-macho] Emit STABS symbols for debugging, and drop debug sections Debug sections contain a large amount of data. In order not to bloat the size of the final binary, we remove them and instead emit STABS symbols for `dsymutil` and the debugger to locate their contents in the object files. With this diff, `dsymutil` is able to locate the debug info. However, we need a few more features before `lldb` is able to work well with our binaries -- e.g. having `LC_DYSYMTAB` accurately reflect the number of local symbols, emitting `LC_UUID`, and more. Those will be handled in follow-up diffs. Note also that the STABS we emit differ slightly from what ld64 does. First, we emit the path to the source file as one `N_SO` symbol instead of two. (`ld64` emits one `N_SO` for the dirname and one of the basename.) Second, we do not emit `N_BNSYM` and `N_ENSYM` STABS to mark the start and end of functions, because the `N_FUN` STABS already serve that purpose. @clayborg recommended these changes based on his knowledge of what the debugging tools look for. Additionally, this current implementation doesn't accurately reflect the size of function symbols. It uses the size of their containing sectioins as a proxy, but that is only accurate if `.subsections_with_symbols` is set, and if there isn't an `N_ALT_ENTRY` in that particular subsection. I think we have two options to solve this: 1. We can split up subsections by symbol even if `.subsections_with_symbols` is not set, but include constraints to ensure those subsections retain their order in the final output. This is `ld64`'s approach. 2. We could just add a `size` field to our `Symbol` class. This seems simpler, and I'm more inclined toward it, but I'm not sure if there are use cases that it doesn't handle well. As such I'm punting on the decision for now. Reviewed By: clayborg Differential Revision: https://reviews.llvm.org/D89257
2020-12-01 22:45:01 +00:00
} else {
getOrCreateOutputSegment(segname)->addOutputSection(osec);
[lld-macho] Emit STABS symbols for debugging, and drop debug sections Debug sections contain a large amount of data. In order not to bloat the size of the final binary, we remove them and instead emit STABS symbols for `dsymutil` and the debugger to locate their contents in the object files. With this diff, `dsymutil` is able to locate the debug info. However, we need a few more features before `lldb` is able to work well with our binaries -- e.g. having `LC_DYSYMTAB` accurately reflect the number of local symbols, emitting `LC_UUID`, and more. Those will be handled in follow-up diffs. Note also that the STABS we emit differ slightly from what ld64 does. First, we emit the path to the source file as one `N_SO` symbol instead of two. (`ld64` emits one `N_SO` for the dirname and one of the basename.) Second, we do not emit `N_BNSYM` and `N_ENSYM` STABS to mark the start and end of functions, because the `N_FUN` STABS already serve that purpose. @clayborg recommended these changes based on his knowledge of what the debugging tools look for. Additionally, this current implementation doesn't accurately reflect the size of function symbols. It uses the size of their containing sectioins as a proxy, but that is only accurate if `.subsections_with_symbols` is set, and if there isn't an `N_ALT_ENTRY` in that particular subsection. I think we have two options to solve this: 1. We can split up subsections by symbol even if `.subsections_with_symbols` is not set, but include constraints to ensure those subsections retain their order in the final output. This is `ld64`'s approach. 2. We could just add a `size` field to our `Symbol` class. This seems simpler, and I'm more inclined toward it, but I'm not sure if there are use cases that it doesn't handle well. As such I'm punting on the decision for now. Reviewed By: clayborg Differential Revision: https://reviews.llvm.org/D89257
2020-12-01 22:45:01 +00:00
}
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
}
for (SyntheticSection *ssec : syntheticSections) {
auto it = mergedOutputSections.find({ssec->segname, ssec->name});
if (it == mergedOutputSections.end()) {
if (ssec->isNeeded())
getOrCreateOutputSegment(ssec->segname)->addOutputSection(ssec);
} else {
error("section from " + toString(it->second->firstSection()->file) +
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
" conflicts with synthetic section " + ssec->segname + "," +
ssec->name);
}
}
}
void Writer::assignAddresses(OutputSegment *seg) {
addr = alignTo(addr, PageSize);
fileOff = alignTo(fileOff, PageSize);
seg->fileOff = fileOff;
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
for (auto *osec : seg->getSections()) {
if (!osec->isNeeded())
continue;
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
addr = alignTo(addr, osec->align);
fileOff = alignTo(fileOff, osec->align);
osec->addr = addr;
osec->fileOff = isZeroFill(osec->flags) ? 0 : fileOff;
osec->finalize();
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
addr += osec->getSize();
fileOff += osec->getFileSize();
}
}
void Writer::openFile() {
Expected<std::unique_ptr<FileOutputBuffer>> bufferOrErr =
FileOutputBuffer::create(config->outputFile, fileOff,
FileOutputBuffer::F_executable);
if (!bufferOrErr)
error("failed to open " + config->outputFile + ": " +
llvm::toString(bufferOrErr.takeError()));
else
buffer = std::move(*bufferOrErr);
}
void Writer::writeSections() {
uint8_t *buf = buffer->getBufferStart();
[lld-macho] Refactor segment/section creation, sorting, and merging Summary: There were a few issues with the previous setup: 1. The section sorting comparator used a declarative map of section names to determine the correct order, but it turns out we need to match on more than just names -- in particular, an upcoming diff will sort based on whether the S_ZERO_FILL flag is set. This diff changes the sorter to a more imperative but flexible form. 2. We were sorting OutputSections stored in a MapVector, which left the MapVector in an inconsistent state -- the wrong keys map to the wrong values! In practice, we weren't doing key lookups (only container iteration) after the sort, so this was fine, but it was still a dubious state of affairs. This diff copies the OutputSections to a vector before sorting them. 3. We were adding unneeded OutputSections to OutputSegments and then filtering them out later, which meant that we had to remember whether an OutputSegment was in a pre- or post-filtered state. This diff only adds the sections to the segments if they are needed. In addition to those major changes, two minor ones worth noting: 1. I renamed all OutputSection variable names to `osec`, to parallel `isec`. Previously we were using some inconsistent combination of `osec`, `os`, and `section`. 2. I added a check (and a test) for InputSections with names that clashed with those of our synthetic OutputSections. Reviewers: #lld-macho Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D81887
2020-06-15 07:03:24 +00:00
for (OutputSegment *seg : outputSegments)
for (OutputSection *osec : seg->getSections())
osec->writeTo(buf + osec->fileOff);
}
void Writer::writeUuid() {
uint64_t digest =
xxHash64({buffer->getBufferStart(), buffer->getBufferEnd()});
uuidCommand->writeUuid(digest);
}
void Writer::run() {
// dyld requires __LINKEDIT segment to always exist (even if empty).
OutputSegment *linkEditSegment =
getOrCreateOutputSegment(segment_names::linkEdit);
prepareBranchTarget(config->entry);
scanRelocations();
if (in.stubHelper->isNeeded())
in.stubHelper->setup();
for (const macho::Symbol *sym : symtab->getSymbols())
if (const auto *defined = dyn_cast<Defined>(sym))
if (defined->overridesWeakDef)
in.weakBinding->addNonWeakDefinition(defined);
// Sort and assign sections to their respective segments. No more sections nor
// segments may be created after these methods run.
createOutputSections();
sortSegmentsAndSections();
createLoadCommands();
// Ensure that segments (and the sections they contain) are allocated
// addresses in ascending order, which dyld requires.
//
// Note that at this point, __LINKEDIT sections are empty, but we need to
// determine addresses of other segments/sections before generating its
// contents.
for (OutputSegment *seg : outputSegments)
if (seg != linkEditSegment)
assignAddresses(seg);
// Fill __LINKEDIT contents.
in.rebase->finalizeContents();
in.binding->finalizeContents();
in.weakBinding->finalizeContents();
in.lazyBinding->finalizeContents();
in.exports->finalizeContents();
[lld-macho][reland] Add basic symbol table output This diff implements basic support for writing a symbol table. Attributes are loosely supported for extern symbols and not at all for other types. Initial version by Kellie Medlin <kelliem@fb.com> Originally committed in a3d95a50ee33 and reverted in fbae153ca583 due to UBSAN erroring over unaligned writes. That has been fixed in the current diff with the following changes: ``` diff --git a/lld/MachO/SyntheticSections.cpp b/lld/MachO/SyntheticSections.cpp --- a/lld/MachO/SyntheticSections.cpp +++ b/lld/MachO/SyntheticSections.cpp @@ -133,6 +133,9 @@ SymtabSection::SymtabSection(StringTableSection &stringTableSection) : stringTableSection(stringTableSection) { segname = segment_names::linkEdit; name = section_names::symbolTable; + // TODO: When we introduce the SyntheticSections superclass, we should make + // all synthetic sections aligned to WordSize by default. + align = WordSize; } size_t SymtabSection::getSize() const { diff --git a/lld/MachO/Writer.cpp b/lld/MachO/Writer.cpp --- a/lld/MachO/Writer.cpp +++ b/lld/MachO/Writer.cpp @@ -371,6 +371,7 @@ void Writer::assignAddresses(OutputSegment *seg) { ArrayRef<InputSection *> sections = p.second; for (InputSection *isec : sections) { addr = alignTo(addr, isec->align); + // We must align the file offsets too to avoid misaligned writes of + // structs. + fileOff = alignTo(fileOff, isec->align); isec->addr = addr; addr += isec->getSize(); fileOff += isec->getFileSize(); @@ -396,6 +397,7 @@ void Writer::writeSections() { uint64_t fileOff = seg->fileOff; for (auto &sect : seg->getSections()) { for (InputSection *isec : sect.second) { + fileOff = alignTo(fileOff, isec->align); isec->writeTo(buf + fileOff); fileOff += isec->getFileSize(); } ``` I don't think it's easy to write a test for alignment (that doesn't involve brittly hard-coding file offsets), so there isn't one... but UBSAN builds pass now. Differential Revision: https://reviews.llvm.org/D79050
2020-04-28 23:58:19 +00:00
symtabSection->finalizeContents();
indirectSymtabSection->finalizeContents();
// Now that __LINKEDIT is filled out, do a proper calculation of its
// addresses and offsets.
assignAddresses(linkEditSegment);
openFile();
if (errorCount())
return;
writeSections();
writeUuid();
if (auto e = buffer->commit())
error("failed to write to the output file: " + toString(std::move(e)));
}
void macho::writeResult() { Writer().run(); }
void macho::createSyntheticSections() {
in.header = make<MachHeaderSection>();
in.rebase = make<RebaseSection>();
in.binding = make<BindingSection>();
in.weakBinding = make<WeakBindingSection>();
in.lazyBinding = make<LazyBindingSection>();
in.exports = make<ExportSection>();
in.got = make<GotSection>();
in.tlvPointers = make<TlvPointerSection>();
in.lazyPointers = make<LazyPointerSection>();
in.stubs = make<StubsSection>();
in.stubHelper = make<StubHelperSection>();
in.imageLoaderCache = make<ImageLoaderCacheSection>();
}