[Mach-O] Implement -dead_strip

macho/dead-strip.cc

@@ -2,7 +2,58 @@
 
 namespace mold::macho {
 
+static std::vector<Subsection *> collect_root_set(Context &ctx) {
+  std::vector<Subsection *> rootset;
+
+  if (Symbol *sym = intern(ctx, ctx.arg.entry))
+    if (sym->subsec)
+      rootset.push_back(sym->subsec);
+
+  for (ObjectFile *file : ctx.objs)
+    for (std::unique_ptr<Subsection> &subsec : file->subsections)
+      if (subsec->isec.hdr.match("__TEXT", "__eh_frame") ||
+          subsec->isec.hdr.match("__TEXT", "__gcc_except_tab"))
+        rootset.push_back(subsec.get());
+
+  return rootset;
+}
+
+static void visit(Context &ctx, Subsection &subsec) {
+  if (subsec.is_visited.exchange(true))
+    return;
+
+  for (Relocation &rel : subsec.get_rels()) {
+    if (rel.sym) {
+      if (rel.sym->subsec)
+        visit(ctx, *rel.sym->subsec);
+    } else {
+      visit(ctx, *rel.subsec);
+    }
+  }
+}
+
+static void mark(Context &ctx, std::span<Subsection *> rootset) {
+  for (Subsection *subsec : rootset)
+    visit(ctx, *subsec);
+}
+
+static void sweep(Context &ctx) {
+  for (ObjectFile *file : ctx.objs) {
+    erase(file->subsections, [](const std::unique_ptr<Subsection> &subsec) {
+      return !subsec->is_visited;
+    });
+  }
+
+  for (ObjectFile *file : ctx.objs)
+    for (Symbol *&sym : file->syms)
+      if (sym->file == file && sym->subsec && !sym->subsec->is_visited)
+        sym = nullptr;
+}
+
 void dead_strip(Context &ctx) {
+  std::vector<Subsection *> rootset = collect_root_set(ctx);
+  mark(ctx, rootset);
+  sweep(ctx);
 }
 
 } // namespace mold::macho

macho/macho.h

@@ -379,7 +379,7 @@ struct MachSection {
     return {sectname, strnlen(sectname, sizeof(sectname))};
   }
 
-  bool match(std::string_view segname, std::string_view sectname) {
+  bool match(std::string_view segname, std::string_view sectname) const {
     return get_segname() == segname && get_sectname() == sectname;
   }
 

macho/main.cc

@@ -140,7 +140,7 @@ static void export_symbols(Context &ctx) {
 
   for (ObjectFile *file : ctx.objs) {
     for (Symbol *sym : file->syms) {
-      if (sym->file == file) {
+      if (sym && sym->file == file) {
         if (sym->flags & NEEDS_GOT)
           ctx.got.add(ctx, sym);
         if (sym->flags & NEEDS_THREAD_PTR)
@@ -151,6 +151,9 @@ static void export_symbols(Context &ctx) {
 
   for (DylibFile *file : ctx.dylibs) {
     for (Symbol *sym : file->syms) {
+      if (!sym)
+        continue;
+
       if (sym->file == file)
         if (sym->flags & NEEDS_STUB)
           ctx.stubs.add(ctx, sym);

macho/mold.h

@@ -177,6 +177,7 @@ public:
   u32 nunwind = 0;
   u32 raddr = -1;
   u16 p2align = 0;
+  std::atomic_bool is_visited = false;
 };
 
 //

macho/object-file.cc

@@ -443,11 +443,11 @@ void ObjectFile::convert_common_symbols(Context &ctx) {
 
 void ObjectFile::check_duplicate_symbols(Context &ctx) {
   for (i64 i = 0; i < syms.size(); i++) {
-    Symbol &sym = *syms[i];
+    Symbol *sym = syms[i];
     MachSym &msym = mach_syms[i];
-    if (!msym.is_undef() && !msym.is_common() && sym.file != this)
-      Error(ctx) << "duplicate symbol: " << *this << ": " << *sym.file
-                 << ": " << sym;
+    if (sym && !msym.is_undef() && !msym.is_common() && sym->file != this)
+      Error(ctx) << "duplicate symbol: " << *this << ": " << *sym->file
+                 << ": " << *sym;
   }
 }
 

macho/output-chunks.cc

@@ -723,7 +723,7 @@ void ExportEncoder::write_trie(u8 *start, TrieNode &node) {
 void OutputExportSection::compute_size(Context &ctx) {
   for (ObjectFile *file : ctx.objs)
     for (Symbol *sym : file->syms)
-      if (sym->is_extern && sym->file == file)
+      if (sym && sym->is_extern && sym->file == file)
         enc.add(sym->name, 0, sym->get_addr(ctx) - PAGE_ZERO_SIZE);
 
   hdr.size = align_to(enc.finish(), 8);
@@ -736,9 +736,10 @@ void OutputExportSection::copy_buf(Context &ctx) {
 void OutputFunctionStartsSection::compute_size(Context &ctx) {
   std::vector<u64> addrs;
 
-  for (ObjectFile *obj : ctx.objs)
-    for (Symbol *sym : obj->syms)
-      if (sym->file == obj && sym->subsec && &sym->subsec->isec.osec == ctx.text)
+  for (ObjectFile *file : ctx.objs)
+    for (Symbol *sym : file->syms)
+      if (sym && sym->file == file && sym->subsec &&
+          &sym->subsec->isec.osec == ctx.text)
         addrs.push_back(sym->get_addr(ctx));
 
   std::sort(addrs.begin(), addrs.end());
@@ -762,16 +763,16 @@ void OutputFunctionStartsSection::copy_buf(Context &ctx) {
 }
 
 void OutputSymtabSection::compute_size(Context &ctx) {
-  for (ObjectFile *obj : ctx.objs)
-    for (Symbol *sym : obj->syms)
-      if (sym->file == obj)
+  for (ObjectFile *file : ctx.objs)
+    for (Symbol *sym : file->syms)
+      if (sym && sym->file == file)
         globals.push_back({sym, ctx.strtab.add_string(sym->name)});
 
   i64 idx = globals.size();
 
   for (DylibFile *dylib : ctx.dylibs) {
     for (Symbol *sym : dylib->syms) {
-      if (sym->file == dylib) {
+      if (sym && sym->file == dylib) {
         if (sym->stub_idx != -1 || sym->got_idx != -1) {
           undefs.push_back({sym, ctx.strtab.add_string(sym->name)});
 

test/macho/dead-strip.sh

@@ -27,5 +27,8 @@ EOF
 
 clang -fuse-ld=$mold -o $t/exe $t/a.o -Wl,-dead_strip
 $t/exe | grep -q 'Hello world'
+otool -tVj $t/exe > $t/log
+grep -q 'hello:' $t/log
+! grep -q 'howdy:' $t/log || false
 
 echo OK