llvm-project/lld/COFF/SymbolTable.h

113 lines
3.5 KiB
C
Raw Normal View History

//===- SymbolTable.h ------------------------------------------------------===//
//
// The LLVM Linker
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
#ifndef LLD_COFF_SYMBOL_TABLE_H
#define LLD_COFF_SYMBOL_TABLE_H
#include "InputFiles.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/DenseMapInfo.h"
#include "llvm/Support/Allocator.h"
#include "llvm/Support/raw_ostream.h"
namespace llvm {
struct LTOCodeGenerator;
}
namespace lld {
namespace coff {
class Chunk;
class Defined;
class Lazy;
class SymbolBody;
struct Symbol;
// SymbolTable is a bucket of all known symbols, including defined,
// undefined, or lazy symbols (the last one is symbols in archive
// files whose archive members are not yet loaded).
//
// We put all symbols of all files to a SymbolTable, and the
// SymbolTable selects the "best" symbols if there are name
// conflicts. For example, obviously, a defined symbol is better than
// an undefined symbol. Or, if there's a conflict between a lazy and a
// undefined, it'll read an archive member to read a real definition
// to replace the lazy symbol. The logic is implemented in resolve().
class SymbolTable {
public:
SymbolTable();
void addFile(std::unique_ptr<InputFile> File);
std::error_code run();
COFF: Change the order of adding symbols to the symbol table. Previously, the order of adding symbols to the symbol table was simple. We have a list of all input files. We read each file from beginning of the list and add all symbols in it to the symbol table. This patch changes that order. Now all archive files are added to the symbol table first, and then all the other object files are added. This shouldn't change the behavior in single-threading, and make room to parallelize in multi-threading. In the first step, only lazy symbols are added to the symbol table because archives contain only Lazy symbols. Member object files found to be necessary are queued. In the second step, defined and undefined symbols are added from object files. Adding an undefined symbol to the symbol table may cause more member files to be added to the queue. We simply continue reading all object files until the queue is empty. Finally, new archive or object files may be added to the queues by object files' directive sections (which contain new command line options). The above process is repeated until we get no new files. Symbols defined both in object files and in archives can make results undeterministic. If an archive is read before an object, a new member file gets linked, while in the other way, no new file would be added. That is the most popular cause of an undeterministic result or linking failure as I observed. Separating phases of adding lazy symbols and undefined symbols makes that deterministic. Adding symbols in each phase should be parallelizable. llvm-svn: 241107
2015-07-01 03:35:21 +08:00
std::error_code readArchives();
std::error_code readObjects();
bool queueEmpty();
// Print an error message on undefined symbols.
bool reportRemainingUndefines();
// Returns a list of chunks of selected symbols.
std::vector<Chunk *> getChunks();
// Returns a symbol for a given name. It's not guaranteed that the
// returned symbol actually has the same name (because of various
// mechanisms to allow aliases, a name can be resolved to a
// different symbol). Returns a nullptr if not found.
Defined *find(StringRef Name);
Symbol *findSymbol(StringRef Name);
// Occasionally we have to resolve an undefined symbol to its
// mangled symbol. This function tries to find a mangled name
// for U from the symbol table, and if found, set the symbol as
// a weak alias for U.
void mangleMaybe(Undefined *U);
// Print a layout map to OS.
void printMap(llvm::raw_ostream &OS);
// Build a COFF object representing the combined contents of BitcodeFiles
// and add it to the symbol table. Called after all files are added and
// before the writer writes results to a file.
std::error_code addCombinedLTOObject();
// The writer needs to handle DLL import libraries specially in
// order to create the import descriptor table.
std::vector<ImportFile *> ImportFiles;
// The writer needs to infer the machine type from the object files.
std::vector<ObjectFile *> ObjectFiles;
// Creates an Undefined symbol for a given name.
Undefined *addUndefined(StringRef Name);
// A list of chunks which to be added to .rdata.
std::vector<Chunk *> LocalImportChunks;
private:
COFF: Change the order of adding symbols to the symbol table. Previously, the order of adding symbols to the symbol table was simple. We have a list of all input files. We read each file from beginning of the list and add all symbols in it to the symbol table. This patch changes that order. Now all archive files are added to the symbol table first, and then all the other object files are added. This shouldn't change the behavior in single-threading, and make room to parallelize in multi-threading. In the first step, only lazy symbols are added to the symbol table because archives contain only Lazy symbols. Member object files found to be necessary are queued. In the second step, defined and undefined symbols are added from object files. Adding an undefined symbol to the symbol table may cause more member files to be added to the queue. We simply continue reading all object files until the queue is empty. Finally, new archive or object files may be added to the queues by object files' directive sections (which contain new command line options). The above process is repeated until we get no new files. Symbols defined both in object files and in archives can make results undeterministic. If an archive is read before an object, a new member file gets linked, while in the other way, no new file would be added. That is the most popular cause of an undeterministic result or linking failure as I observed. Separating phases of adding lazy symbols and undefined symbols makes that deterministic. Adding symbols in each phase should be parallelizable. llvm-svn: 241107
2015-07-01 03:35:21 +08:00
std::error_code addSymbol(SymbolBody *New);
void addLazy(Lazy *New, std::vector<Symbol *> *Accum);
std::error_code addMemberFile(Lazy *Body);
ErrorOr<ObjectFile *> createLTOObject(llvm::LTOCodeGenerator *CG);
llvm::DenseMap<StringRef, Symbol *> Symtab;
COFF: Change the order of adding symbols to the symbol table. Previously, the order of adding symbols to the symbol table was simple. We have a list of all input files. We read each file from beginning of the list and add all symbols in it to the symbol table. This patch changes that order. Now all archive files are added to the symbol table first, and then all the other object files are added. This shouldn't change the behavior in single-threading, and make room to parallelize in multi-threading. In the first step, only lazy symbols are added to the symbol table because archives contain only Lazy symbols. Member object files found to be necessary are queued. In the second step, defined and undefined symbols are added from object files. Adding an undefined symbol to the symbol table may cause more member files to be added to the queue. We simply continue reading all object files until the queue is empty. Finally, new archive or object files may be added to the queues by object files' directive sections (which contain new command line options). The above process is repeated until we get no new files. Symbols defined both in object files and in archives can make results undeterministic. If an archive is read before an object, a new member file gets linked, while in the other way, no new file would be added. That is the most popular cause of an undeterministic result or linking failure as I observed. Separating phases of adding lazy symbols and undefined symbols makes that deterministic. Adding symbols in each phase should be parallelizable. llvm-svn: 241107
2015-07-01 03:35:21 +08:00
std::vector<std::unique_ptr<InputFile>> Files;
COFF: Change the order of adding symbols to the symbol table. Previously, the order of adding symbols to the symbol table was simple. We have a list of all input files. We read each file from beginning of the list and add all symbols in it to the symbol table. This patch changes that order. Now all archive files are added to the symbol table first, and then all the other object files are added. This shouldn't change the behavior in single-threading, and make room to parallelize in multi-threading. In the first step, only lazy symbols are added to the symbol table because archives contain only Lazy symbols. Member object files found to be necessary are queued. In the second step, defined and undefined symbols are added from object files. Adding an undefined symbol to the symbol table may cause more member files to be added to the queue. We simply continue reading all object files until the queue is empty. Finally, new archive or object files may be added to the queues by object files' directive sections (which contain new command line options). The above process is repeated until we get no new files. Symbols defined both in object files and in archives can make results undeterministic. If an archive is read before an object, a new member file gets linked, while in the other way, no new file would be added. That is the most popular cause of an undeterministic result or linking failure as I observed. Separating phases of adding lazy symbols and undefined symbols makes that deterministic. Adding symbols in each phase should be parallelizable. llvm-svn: 241107
2015-07-01 03:35:21 +08:00
std::vector<ArchiveFile *> ArchiveQueue;
std::vector<InputFile *> ObjectQueue;
std::vector<BitcodeFile *> BitcodeFiles;
std::unique_ptr<MemoryBuffer> LTOMB;
llvm::BumpPtrAllocator Alloc;
};
} // namespace coff
} // namespace lld
#endif