Refactor

CMakeLists.txt

@@ -362,7 +362,6 @@ target_sources(mold PRIVATE
   lib/compress.cc
   lib/crc32.cc
   lib/demangle.cc
-  lib/elf.cc
   lib/filepath.cc
   lib/glob.cc
   lib/hyperloglog.cc
@@ -384,6 +383,7 @@ target_sources(mold PRIVATE
   src/arch-sparc64.cc
   src/arch-x86-64.cc
   src/config.cc
+  src/elf.cc
   third-party/rust-demangle/rust-demangle.c
   )
 

lib/archive-file.h

@@ -26,7 +26,6 @@
 #pragma once
 
 #include "common.h"
-#include "filetype.h"
 
 namespace mold {
 
@@ -164,14 +163,11 @@ std::vector<MappedFile *> read_fat_archive_members(Context &ctx, MappedFile *mf)
 
 template <typename Context, typename MappedFile>
 std::vector<MappedFile *> read_archive_members(Context &ctx, MappedFile *mf) {
-  switch (get_file_type(ctx, mf)) {
-  case FileType::AR:
+  std::string_view str = mf->get_contents();
+  if (str.starts_with("!<arch>\n"))
     return read_fat_archive_members(ctx, mf);
-  case FileType::THIN_AR:
-    return read_thin_archive_members(ctx, mf);
-  default:
-    unreachable();
-  }
+  assert(str.starts_with("!<thin>\n"));
+  return read_thin_archive_members(ctx, mf);
 }
 
 } // namespace mold

lib/integers.h

@@ -8,12 +8,12 @@
 //    big-endian SPARC machine to create a little-endian RV64 binary.
 //
 // 2. Even though data members in all ELF data strucutres are naturally
-//    aligned, they are not guaranteed to be aligned on memory. Because
-//    archive file (.a file) aligns each member only to a 2 byte boundary,
-//    anything larger than 2 bytes may be unaligned in an mmap'ed memory.
-//    Unaligned access is an undefined behavior in C/C++, so we shouldn't
-//    cast an arbitrary pointer to a uint32_t, for example, to read a
-//    32 bit value.
+//    aligned, they are not guaranteed to be aligned on memory because of
+//    archive files. Archive files (.a files) align each member only to a
+//    2 byte boundary, so anything larger than 2 bytes may be misaligned
+//    in an mmap'ed memory. Misaligned access is an undefined behavior in
+//    C/C++, so we shouldn't cast an arbitrary pointer to a uint32_t, for
+//    example, to read a 32 bit value.
 //
 // The data types defined in this file don't depend on host byte order and
 // don't do unaligned access.

lib/tar.cc

@@ -1,3 +1,5 @@
+// This file contains functions to create a tar file.
+
 #include "common.h"
 
 #ifdef _WIN32

src/elf.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#include "integers.h"
+#include "../lib/integers.h"
 
 #include <concepts>
 #include <ostream>

src/filetype.h

@@ -1,6 +1,6 @@
 #pragma once
 
-#include "common.h"
+#include "../lib/common.h"
 #include "elf.h"
 
 namespace mold {
@@ -140,23 +140,23 @@ FileType get_file_type(Context &ctx, MappedFile *mf) {
   return FileType::UNKNOWN;
 }
 
-inline std::string filetype_to_string(FileType type) {
-  switch (type) {
-  case FileType::UNKNOWN: return "UNKNOWN";
-  case FileType::EMPTY: return "EMPTY";
-  case FileType::ELF_OBJ: return "ELF_OBJ";
-  case FileType::ELF_DSO: return "ELF_DSO";
-  case FileType::AR: return "AR";
-  case FileType::THIN_AR: return "THIN_AR";
-  case FileType::TEXT: return "TEXT";
-  case FileType::GCC_LTO_OBJ: return "GCC_LTO_OBJ";
-  case FileType::LLVM_BITCODE: return "LLVM_BITCODE";
-  }
-  return "UNKNOWN";
-}
-
 inline std::ostream &operator<<(std::ostream &out, FileType type) {
-  out << filetype_to_string(type);
+  auto to_string = [&] {
+    switch (type) {
+    case FileType::UNKNOWN: return "UNKNOWN";
+    case FileType::EMPTY: return "EMPTY";
+    case FileType::ELF_OBJ: return "ELF_OBJ";
+    case FileType::ELF_DSO: return "ELF_DSO";
+    case FileType::AR: return "AR";
+    case FileType::THIN_AR: return "THIN_AR";
+    case FileType::TEXT: return "TEXT";
+    case FileType::GCC_LTO_OBJ: return "GCC_LTO_OBJ";
+    case FileType::LLVM_BITCODE: return "LLVM_BITCODE";
+    default: return "UNKNOWN";
+    }
+  };
+
+  out << to_string();
   return out;
 }
 

src/main.cc

@@ -1,4 +1,5 @@
 #include "mold.h"
+#include "filetype.h"
 #include "../lib/archive-file.h"
 #include "../lib/output-file.h"
 

src/mold.h

@@ -1,7 +1,7 @@
 #pragma once
 
 #include "../lib/common.h"
-#include "../lib/elf.h"
+#include "elf.h"
 
 #include <atomic>
 #include <bitset>