[Mach-O] Fix weak undef symbols

macho/input-files.cc

@@ -20,7 +20,7 @@ void InputFile<E>::clear_symbols() {
       sym->file = nullptr;
       sym->scope = SCOPE_LOCAL;
       sym->is_imported = false;
-      sym->is_weak_def = false;
+      sym->is_weak = false;
       sym->subsec = nullptr;
       sym->value = 0;
       sym->is_common = false;
@@ -138,7 +138,7 @@ void ObjectFile<E>::parse_symbols(Context<E> &ctx) {
       sym.subsec = nullptr;
       sym.scope = SCOPE_LOCAL;
       sym.is_common = false;
-      sym.is_weak_def = false;
+      sym.is_weak = false;
 
       switch (msym.type) {
       case N_UNDF:
@@ -533,7 +533,7 @@ template <typename E>
 static u64 get_rank(Symbol<E> &sym) {
   if (!sym.file)
     return 7 << 24;
-  return get_rank(sym.file, sym.is_common, sym.is_weak_def);
+  return get_rank(sym.file, sym.is_common, sym.is_weak);
 }
 
 template <typename E>
@@ -559,14 +559,14 @@ void ObjectFile<E>::resolve_symbols(Context<E> &ctx) {
 
     Symbol<E> &sym = *this->syms[i];
     std::scoped_lock lock(sym.mu);
-    bool is_weak_def = (msym.desc & N_WEAK_DEF);
+    bool is_weak = (msym.desc & N_WEAK_DEF);
 
     sym.scope = merge_scope(sym, msym);
 
-    if (get_rank(this, msym.is_common(), is_weak_def) < get_rank(sym)) {
+    if (get_rank(this, msym.is_common(), is_weak) < get_rank(sym)) {
       sym.file = this;
       sym.is_imported = false;
-      sym.is_weak_def = is_weak_def;
+      sym.is_weak = is_weak;
 
       switch (msym.type) {
       case N_UNDF:
@@ -647,7 +647,7 @@ void ObjectFile<E>::convert_common_symbols(Context<E> &ctx) {
       subsections.emplace_back(subsec);
 
       sym.is_imported = false;
-      sym.is_weak_def = false;
+      sym.is_weak = false;
       sym.subsec = subsec;
       sym.value = 0;
       sym.is_common = false;
@@ -832,7 +832,7 @@ void DylibFile<E>::resolve_symbols(Context<E> &ctx) {
       sym->file = this;
       sym->scope = SCOPE_LOCAL;
       sym->is_imported = true;
-      sym->is_weak_def = this->is_weak;
+      sym->is_weak = this->is_weak;
       sym->subsec = nullptr;
       sym->value = 0;
       sym->is_common = false;

macho/main.cc

@@ -255,9 +255,13 @@ template <typename E>
 static void claim_unresolved_symbols(Context<E> &ctx) {
   Timer t(ctx, "claim_unresolved_symbols");
 
-  for (std::string_view name : ctx.arg.U)
-    if (Symbol<E> *sym = get_symbol(ctx, name); !sym->file)
+  for (std::string_view name : ctx.arg.U) {
+    Symbol<E> *sym = get_symbol(ctx, name);
+    if (!sym->file) {
       sym->is_imported = true;
+      sym->is_weak = true;
+    }
+  }
 
   tbb::parallel_for_each(ctx.objs, [&](ObjectFile<E> *file) {
     for (i64 i = 0; i < file->mach_syms.size(); i++) {

macho/mold.h

@@ -281,7 +281,7 @@ struct Symbol {
 
   u8 scope : 2 = SCOPE_LOCAL;
   bool is_common : 1 = false;
-  bool is_weak_def : 1 = false;
+  bool is_weak : 1 = false;
   bool is_imported : 1 = false;
   bool referenced_dynamically : 1 = false;
 

macho/output-chunks.cc

@@ -680,7 +680,7 @@ static u32 get_dylib_idx(InputFile<E> *file) {
 template <typename E>
 void BindEncoder::add(Symbol<E> &sym, i64 seg_idx, i64 offset) {
   i64 dylib_idx = get_dylib_idx(sym.file);
-  i64 flags = (sym.file->is_weak ? BIND_SYMBOL_FLAGS_WEAK_IMPORT : 0);
+  i64 flags = (sym.is_weak ? BIND_SYMBOL_FLAGS_WEAK_IMPORT : 0);
 
   if (last_dylib != dylib_idx) {
     if (dylib_idx < 16) {
@@ -767,7 +767,7 @@ void LazyBindSection<E>::add(Context<E> &ctx, Symbol<E> &sym) {
     encode_uleb(contents, dylib_idx);
   }
 
-  i64 flags = (sym.file->is_weak ? BIND_SYMBOL_FLAGS_WEAK_IMPORT : 0);
+  i64 flags = (sym.is_weak ? BIND_SYMBOL_FLAGS_WEAK_IMPORT : 0);
   assert(flags < 16);
 
   emit(BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM | flags);

test/macho/weak-def.sh

@@ -0,0 +1,40 @@
+#!/bin/bash
+export LC_ALL=C
+set -e
+CC="${CC:-cc}"
+CXX="${CXX:-c++}"
+GCC="${GCC:-gcc}"
+GXX="${GXX:-g++}"
+OBJDUMP="${OBJDUMP:-objdump}"
+MACHINE="${MACHINE:-$(uname -m)}"
+testname=$(basename "$0" .sh)
+echo -n "Testing $testname ... "
+cd "$(dirname "$0")"/../..
+t=out/test/macho/$testname
+mkdir -p $t
+
+cat <<EOF | $CC -c -o $t/a.o -xc -
+#include <stdio.h>
+
+int foo() __attribute__((weak));
+
+int foo() {
+  return 3;
+}
+
+int main() {
+  printf("%d\n", foo());
+}
+EOF
+
+cat <<EOF | $CC -c -o $t/b.o -xc -
+int foo() { return 42; }
+EOF
+
+clang --ld-path=./ld64 -o $t/exe1 $t/a.o
+$t/exe1 | grep -q '^3$'
+
+clang --ld-path=./ld64 -o $t/exe1 $t/a.o $t/b.o
+$t/exe1 | grep -q '^42$'
+
+echo OK

test/macho/weak-undef.sh

@@ -25,10 +25,10 @@ cat <<EOF | $CC -c -o $t/b.o -xc -
 int foo() { return 42; }
 EOF
 
-clang -o $t/exe1 $t/a.o -Wl,-U,_foo
+clang --ld-path=./ld64 -o $t/exe1 $t/a.o -Wl,-U,_foo
 $t/exe1 | grep -q '^5$'
 
-clang -o $t/exe2 $t/a.o $t/b.o -Wl,-U,_foo
+clang --ld-path=./ld64 -o $t/exe2 $t/a.o $t/b.o -Wl,-U,_foo
 $t/exe2 | grep -q '^42$'
 
 echo OK