[Mach-O] Emit __data_in_code

macho/macho.h

@@ -498,7 +498,7 @@ struct SourceVersionCommand {
 
 struct DataInCodeEntry {
   u32 offset;
-  u32 length;
+  u16 length;
   u16 kind;
 };
 

macho/main.cc

@@ -83,9 +83,11 @@ static bool compare_chunks(const Chunk *a, const Chunk *b) {
     // __LINKEDIT
     "__rebase",
     "__binding",
+    "__weak_binding",
     "__lazy_binding",
     "__export",
     "__func_starts",
+    "__data_in_code",
     "__symbol_table",
     "__ind_sym_tab",
     "__string_table",

macho/mold.h

@@ -80,6 +80,7 @@ public:
   std::vector<std::unique_ptr<InputSection>> sections;
   std::span<MachSym> mach_syms;
   std::vector<UnwindRecord> unwind_records;
+  std::span<DataInCodeEntry> data_in_code_entries;
 
 private:
   void override_symbol(Context &ctx, Symbol &sym, MachSym &msym);
@@ -468,6 +469,20 @@ public:
   static constexpr i64 BLOCK_SIZE = 4096;
 };
 
+class DataInCodeSection : public Chunk {
+public:
+  DataInCodeSection(Context &ctx)
+    : Chunk(ctx, "__LINKEDIT", "__data_in_code") {
+    is_hidden = true;
+    hdr.p2align = __builtin_ctz(alignof(DataInCodeEntry));
+  }
+
+  void compute_size(Context &ctx) override;
+  void copy_buf(Context &ctx) override;
+
+  std::vector<DataInCodeEntry> contents;
+};
+
 class StubsSection : public Chunk {
 public:
   StubsSection(Context &ctx) : Chunk(ctx, "__TEXT", "__stubs") {
@@ -714,6 +729,7 @@ struct Context {
   GotSection got{*this};
   LazySymbolPtrSection lazy_symbol_ptr{*this};
   CodeSignatureSection code_sig{*this};
+  DataInCodeSection data_in_code{*this};
 
   OutputRebaseSection rebase{*this};
   OutputBindSection bind{*this};

macho/object-file.cc

@@ -5,7 +5,10 @@
 namespace mold::macho {
 
 std::ostream &operator<<(std::ostream &out, const InputFile &file) {
-  out << path_clean(file.mf->name);
+  if (file.archive_name.empty())
+    out << path_clean(file.mf->name);
+  else
+    out << path_clean(file.archive_name) << "(" << path_clean(file.mf->name) + ")";
   return out;
 }
 
@@ -58,8 +61,13 @@ void ObjectFile::parse(Context &ctx) {
     case LC_DYSYMTAB:
     case LC_BUILD_VERSION:
     case LC_VERSION_MIN_MACOSX:
-    case LC_DATA_IN_CODE:
       break;
+    case LC_DATA_IN_CODE: {
+      LinkEditDataCommand &cmd = *(LinkEditDataCommand *)p;
+      data_in_code_entries = {(DataInCodeEntry *)(mf->data + cmd.dataoff),
+                              cmd.datasize / sizeof(DataInCodeEntry)};
+      break;
+    }
     default:
       Error(ctx) << *this << ": unknown load command: 0x" << std::hex << lc.cmd;
     }

macho/output-chunks.cc

@@ -170,6 +170,17 @@ static std::vector<u8> create_function_starts_cmd(Context &ctx) {
   return buf;
 }
 
+static std::vector<u8> create_data_in_code_cmd(Context &ctx) {
+  std::vector<u8> buf(sizeof(LinkEditDataCommand));
+  LinkEditDataCommand &cmd = *(LinkEditDataCommand *)buf.data();
+
+  cmd.cmd = LC_DATA_IN_CODE;
+  cmd.cmdsize = buf.size();
+  cmd.dataoff = ctx.data_in_code.hdr.offset;
+  cmd.datasize = ctx.data_in_code.hdr.size;
+  return buf;
+}
+
 static std::vector<u8> create_code_signature_cmd(Context &ctx) {
   std::vector<u8> buf(sizeof(LinkEditDataCommand));
   LinkEditDataCommand &cmd = *(LinkEditDataCommand *)buf.data();
@@ -224,6 +235,8 @@ static std::pair<i64, std::vector<u8>> create_load_commands(Context &ctx) {
   for (DylibFile *dylib : ctx.dylibs)
     vec.push_back(create_load_dylib_cmd(ctx, dylib->install_name));
   vec.push_back(create_function_starts_cmd(ctx));
+  if (!ctx.data_in_code.contents.empty())
+    vec.push_back(create_data_in_code_cmd(ctx));
   if (ctx.arg.adhoc_codesign)
     vec.push_back(create_code_signature_cmd(ctx));
   return {vec.size(), flatten(vec)};
@@ -867,6 +880,40 @@ void CodeSignatureSection::write_signature(Context &ctx) {
   }
 }
 
+void DataInCodeSection::compute_size(Context &ctx) {
+  assert(contents.empty());
+
+  for (ObjectFile *file : ctx.objs) {
+    std::span<DataInCodeEntry> entries = file->data_in_code_entries;
+
+    for (i64 i = 0; !entries.empty() && i < file->sections.size(); i++) {
+      InputSection &sec = *file->sections[i];
+
+      for (i64 j = 0; !entries.empty() && j < sec.subsections.size(); j++) {
+        Subsection &subsec = *sec.subsections[j];
+        DataInCodeEntry &ent = entries[0];
+
+        if (subsec.input_addr + subsec.input_size < ent.offset)
+          continue;
+
+        if (ent.offset < subsec.input_addr + subsec.input_size) {
+          u32 offset = subsec.get_addr(ctx) + subsec.input_addr - ent.offset -
+                       ctx.text_seg->cmd.vmaddr;
+          contents.push_back({offset, ent.length, ent.kind});
+        }
+
+        entries = entries.subspan(1);
+      }
+    }
+  }
+
+  hdr.size = contents.size() * sizeof(contents[0]);
+}
+
+void DataInCodeSection::copy_buf(Context &ctx) {
+  write_vector(ctx.buf + hdr.offset, contents);
+}
+
 void StubsSection::add(Context &ctx, Symbol *sym) {
   assert(sym->stub_idx == -1);
   sym->stub_idx = syms.size();