Merge pull request #520 from tapthaker/macho-tbd-files

[Mach-O] Adds support for objc and weak symbols in TBD files

macho/input-files.cc

@@ -679,8 +679,30 @@ void DylibFile<E>::parse(Context<E> &ctx) {
   switch (get_file_type(this->mf)) {
   case FileType::TAPI: {
     TextDylib tbd = parse_tbd(ctx, this->mf);
+    auto add_symbol_with_prefix = [&](std::string_view prefix,
+                                      std::string_view sym_name) {
+      std::string symbol_name = std::string(prefix);
+      symbol_name.append(sym_name);
+      Symbol<E> *sym = get_symbol(ctx, save_string(ctx, symbol_name));
+      this->syms.push_back(sym);
+    };
+
     for (std::string_view sym : tbd.exports)
       this->syms.push_back(get_symbol(ctx, sym));
+    for (std::string_view weak_sym : tbd.weak_exports) {
+      Symbol<E> *sym = get_symbol(ctx, weak_sym);
+      sym->is_weak_def = true;
+      this->syms.push_back(sym);
+    }
+    for (std::string_view objc_class : tbd.objc_classes) {
+      add_symbol_with_prefix(OBJC2_CLASS_NAME_PREFIX, objc_class);
+      add_symbol_with_prefix(OBJC2_METACLASS_NAME_PREFIX, objc_class);
+    }
+    for (std::string_view eh_type : tbd.objc_eh_types)
+      add_symbol_with_prefix(OBJC2_EHTYPE_PREFIX, eh_type);
+    for (std::string_view ivar : tbd.objc_ivars)
+      add_symbol_with_prefix(OBJC2_IVAR_PREFIX, ivar);
+
     install_name = tbd.install_name;
     break;
   }

macho/mold.h

@@ -702,6 +702,12 @@ parse_yaml(std::string_view str);
 // tapi.cc
 //
 
+static const std::string_view OBJC2_CLASS_NAME_PREFIX = "_OBJC_CLASS_$_";
+static const std::string_view OBJC2_METACLASS_NAME_PREFIX =
+    "_OBJC_METACLASS_$_";
+static const std::string_view OBJC2_EHTYPE_PREFIX = "_OBJC_EHTYPE_$_";
+static const std::string_view OBJC2_IVAR_PREFIX = "_OBJC_IVAR_$_";
+
 struct TextDylib {
   std::string_view uuid;
   std::string_view install_name;
@@ -709,6 +715,10 @@ struct TextDylib {
   std::string_view parent_umbrella;
   std::vector<std::string_view> reexported_libs;
   std::vector<std::string_view> exports;
+  std::vector<std::string_view> weak_exports;
+  std::vector<std::string_view> objc_classes;
+  std::vector<std::string_view> objc_eh_types;
+  std::vector<std::string_view> objc_ivars;
 };
 
 template <typename E>

macho/tapi.cc

@@ -72,10 +72,23 @@ static std::optional<TextDylib> to_tbd(YamlNode &node, std::string_view arch) {
 
   for (std::string_view key : {"exports", "reexports"})
     for (YamlNode &mem : get_vector(node, key))
-      if (contains(get_vector(mem, "targets"), arch))
+      if (contains(get_vector(mem, "targets"), arch)) {
         for (YamlNode &mem : get_vector(mem, "symbols"))
           if (auto *sym = std::get_if<std::string_view>(&mem.data))
             tbd.exports.push_back(*sym);
+        for (YamlNode &mem : get_vector(mem, "weak-symbols"))
+          if (auto *sym = std::get_if<std::string_view>(&mem.data))
+            tbd.weak_exports.push_back(*sym);
+        for (YamlNode &mem : get_vector(mem, "objc-classes"))
+          if (auto *clazz = std::get_if<std::string_view>(&mem.data))
+            tbd.objc_classes.push_back(*clazz);
+        for (YamlNode &mem : get_vector(mem, "objc-eh-types"))
+          if (auto *eh_type = std::get_if<std::string_view>(&mem.data))
+            tbd.objc_eh_types.push_back(*eh_type);
+        for (YamlNode &mem : get_vector(mem, "objc-ivars"))
+          if (auto *ivar = std::get_if<std::string_view>(&mem.data))
+            tbd.objc_ivars.push_back(*ivar);
+      }
 
   return tbd;
 }

test/macho/tbd.sh

@@ -0,0 +1,85 @@
+#!/bin/bash
+export LC_ALL=C
+set -e
+CC="${TEST_CC:-cc}"
+CXX="${TEST_CXX:-c++}"
+GCC="${TEST_GCC:-gcc}"
+GXX="${TEST_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 $t
+
+some_fw=$t/SF/SomeFramework.framework/
+mkdir -p $some_fw
+
+cat > $some_fw/SomeFramework.tbd <<EOF
+--- !tapi-tbd
+tbd-version:     4
+targets:         [ x86_64-macos, arm64-macos ]
+uuids:
+  - target:          x86_64-macos
+    value:           00000000-0000-0000-0000-000000000000
+  - target:          arm64-macos
+    value:           00000000-0000-0000-0000-000000000000
+install-name:    '/usr/frameworks/SomeFramework.framework/SomeFramework'
+current-version: 0000
+compatibility-version: 150
+reexported-libraries:
+  - targets:         [ x86_64-macos, arm64-macos ]
+    libraries:       [ ]
+exports:
+  - targets:         [ x86_64-macos, arm64-macos ]
+    symbols:         [ _some_framework_print ]
+    objc-classes:    [ SomeObjectiveC ]
+    weak-symbols:    [ _weak_some_framework_print ]
+...
+EOF
+
+
+cat <<EOF | clang -o $t/TestTBDFiles.o -c -xobjective-c -
+#include <stdio.h>
+#import <CoreFoundation/CoreFoundation.h>
+
+// Interface Declaration for SomeFramework.framework
+
+@interface SomeObjectiveC: NSObject
+
+@end
+
+void some_framework_print(char*);
+void weak_some_framework_print(char*) __attribute__((weak));;
+
+// End Interface Declaration for SomeFramework.framework
+
+@interface TestTBDFiles: SomeObjectiveC
+@end
+
+@implementation TestTBDFiles
+
+-(void) helloWorld {
+  some_framework_print("Hello World");
+  weak_some_framework_print("Hello World");
+}
+
+@end
+
+int main(int argc, const char * argv[]) {
+    @autoreleasepool {
+        TestTBDFiles *tbd = [[TestTBDFiles alloc]init];
+        [tbd helloWorld];
+    }
+    return 0;
+}
+EOF
+
+
+clang --ld-path=./ld64 -F$t/SF/ -Wl,-framework,SomeFramework \
+  -Wl,-framework,CoreFoundation -lobjc -o $t/exe $t/TestTBDFiles.o
+otool -L $t/exe > $t/install_paths.log
+grep -q '/usr/frameworks/SomeFramework.framework/SomeFramework' $t/install_paths.log
+
+echo OK