iOS LLVM 中的宏定义
在阅读 Objc 库源码时常常会遇到很多宏定义,比如宏 SUPPORT_INDEXED_ISA、SUPPORT_PACKED_ISA,代码如下所示:
- // Define SUPPORT_INDEXED_ISA=1 on platforms that store the class in the isa
- // field as an index into a class table.
- // Note, keep this in sync with any .s files which also define it.
- // Be sure to edit objc-abi.h as well.
- #if __ARM_ARCH_7K__ >= 2 || (__arm64__ && !__LP64__)
- # define SUPPORT_INDEXED_ISA 1
- #else
- # define SUPPORT_INDEXED_ISA 0
- #endif
- // Define SUPPORT_PACKED_ISA=1 on platforms that store the class in the isa
- // field as a maskable pointer with other data around it.
- #if (!__LP64__ || TARGET_OS_WIN32 || \
- (TARGET_OS_SIMULATOR && !TARGET_OS_MACCATALYST && !__arm64__))
- # define SUPPORT_PACKED_ISA 0
- #else
- # define SUPPORT_PACKED_ISA 1
- #endif
在上面的宏定义中,__ARM_ARCH_7K__、__arm64__、__LP64__ 这些宏在 Objc 库中找不到定义的源码。如果不清楚这些宏的意义,对阅读源码会带来一定的障碍。实际上,这些宏都定义在 LLVM 源码中(基本上找不到定义的宏,都可以在 LLVM 源码中找到)。
__ARM_ARCH_7K__
在 LLVM 源码 ARM.cpp 中,可以看到对 __ARM_ARCH_7K__ 的定义,源码如下:
- // file: ARM.cpp
- // Unfortunately, __ARM_ARCH_7K__ is now more of an ABI descriptor. The CPU
- // happens to be Cortex-A7 though, so it should still get __ARM_ARCH_7A__.
- if (getTriple().isWatchABI()) // 判断是否是 Watch 的ABI
- Builder.defineMacro("__ARM_ARCH_7K__", "2");
从源码看到,这个宏是在 Apple Watch 下生效,在 iPhone 设备上该宏不会生效。
__arm64__
在 LLVM 源码 AArch64.cpp 中,可以看到对 __arm64__ 的定义,源码如下:
- void DarwinAArch64TargetInfo::getOSDefines(const LangOptions &Opts,
- const llvm::Triple &Triple,
- MacroBuilder &Builder) const {
- Builder.defineMacro("__AARCH64_SIMD__");
- if (Triple.isArch32Bit())
- Builder.defineMacro("__ARM64_ARCH_8_32__");
- else
- Builder.defineMacro("__ARM64_ARCH_8__");
- Builder.defineMacro("__ARM_NEON__");
- Builder.defineMacro("__LITTLE_ENDIAN__");
- Builder.defineMacro("__REGISTER_PREFIX__", "");
- Builder.defineMacro("__arm64", "1");
- Builder.defineMacro("__arm64__", "1"); // __arm64__ 定义
- if (Triple.isArm64e())
- Builder.defineMacro("__arm64e__", "1");
- getDarwinDefines(Builder, Opts, Triple, PlatformName, PlatformMinVersion);
- }
从源码上可以看到,只要 ARM CPU 是 64bit,就会定义 __arm64__ 宏,虽然可能这个 CPU 使用的是 ILP32(见下文)。
__LP64__
在 LLVM 源码 InitPreprocessor.cpp 中,可以看到对 __LP64__ 的定义,源码如下:
- // file: InitPreprocessor.cpp
static void InitializePredefinedMacros(const TargetInfo &TI,- const LangOptions &LangOpts,
- const FrontendOptions &FEOpts,
- const PreprocessorOptions &PPOpts,
- MacroBuilder &Builder) {
- ...
- if (TI.getPointerWidth(0) == 64 && TI.getLongWidth() == 64
- && TI.getIntWidth() == 32) {
- Builder.defineMacro("_LP64");
- Builder.defineMacro("__LP64__"); // 定义 __LP64__
- }
- if (TI.getPointerWidth(0) == 32 && TI.getLongWidth() == 32
- && TI.getIntWidth() == 32) {
- Builder.defineMacro("_ILP32");
- Builder.defineMacro("__ILP32__"); // 定义 __ILP32__
- }
- ...
- // Get other target #defines.
- TI.getTargetDefines(LangOpts, Builder); // 该方法会重新定义 __LP64__
- }
从上面源码可以看到,如果指针 pointer 的长度是 64bit,long 类型的长度是 64bit,int 类型的长度是 32bit,那么就定义宏 __LP64__。
如果指针 pointer 的长度是 32bit,long 类型的长度是 32bit,int 类型的长度是 32bit,那么就定义宏 __ILP32__。
源码最后一行 TI.getTargetDefines(LangOpts, Builder) 在 ARM 架构下重新定义 __LP64__,相关源码位于 AArch64.cpp:
- // file: AArch64.cpp
void AArch64TargetInfo::getTargetDefines(const LangOptions &Opts,- MacroBuilder &Builder) const {
- // Target identification.
- Builder.defineMacro("__aarch64__");
- // For bare-metal.
- if (getTriple().getOS() == llvm::Triple::UnknownOS &&
- getTriple().isOSBinFormatELF())
- Builder.defineMacro("__ELF__");
- // Target properties.
- if (!getTriple().isOSWindows() && getTriple().isArch64Bit()) { // 在非 Windows 下,并且真正支持 64bit 指针的 CPU 架构下才定义 __LP64__
- Builder.defineMacro("_LP64");
- Builder.defineMacro("__LP64__");
- }
- ...
- }
从上面的源码可以看到,非 Windows 系统 & 真正支持 64bit 指针的 CPU 架构才会定义 __LP64__。那么哪些类型的 CPU 支持 64bit 的指针呢? 通过查看 Triple::isArch64Bit 方法可以得到答案:
- // file: Triple.cpp
- bool Triple::isArch64Bit() const {
- return getArchPointerBitWidth(getArch()) == 64;
- }
- static unsigned getArchPointerBitWidth(llvm::Triple::ArchType Arch) {
- switch (Arch) {
- case llvm::Triple::UnknownArch:
- return 0;
- case llvm::Triple::avr:
- case llvm::Triple::msp430:
- return 16;
- case llvm::Triple::aarch64_32:
- case llvm::Triple::amdil:
- case llvm::Triple::arc:
- case llvm::Triple::arm:
- case llvm::Triple::armeb:
- case llvm::Triple::csky:
- case llvm::Triple::dxil:
- case llvm::Triple::hexagon:
- case llvm::Triple::hsail:
- case llvm::Triple::kalimba:
- case llvm::Triple::lanai:
- case llvm::Triple::le32:
- case llvm::Triple::loongarch32:
- case llvm::Triple::m68k:
- case llvm::Triple::mips:
- case llvm::Triple::mipsel:
- case llvm::Triple::nvptx:
- case llvm::Triple::ppc:
- case llvm::Triple::ppcle:
- case llvm::Triple::r600:
- case llvm::Triple::renderscript32:
- case llvm::Triple::riscv32:
- case llvm::Triple::shave:
- case llvm::Triple::sparc:
- case llvm::Triple::sparcel:
- case llvm::Triple::spir:
- case llvm::Triple::spirv32:
- case llvm::Triple::tce:
- case llvm::Triple::tcele:
- case llvm::Triple::thumb:
- case llvm::Triple::thumbeb:
- case llvm::Triple::wasm32:
- case llvm::Triple::x86:
- case llvm::Triple::xcore:
- return 32;
- case llvm::Triple::aarch64:
- case llvm::Triple::aarch64_be:
- case llvm::Triple::amdgcn:
- case llvm::Triple::amdil64:
- case llvm::Triple::bpfeb:
- case llvm::Triple::bpfel:
- case llvm::Triple::hsail64:
- case llvm::Triple::le64:
- case llvm::Triple::loongarch64:
- case llvm::Triple::mips64:
- case llvm::Triple::mips64el:
- case llvm::Triple::nvptx64:
- case llvm::Triple::ppc64:
- case llvm::Triple::ppc64le:
- case llvm::Triple::renderscript64:
- case llvm::Triple::riscv64:
- case llvm::Triple::sparcv9:
- case llvm::Triple::spir64:
- case llvm::Triple::spirv64:
- case llvm::Triple::systemz:
- case llvm::Triple::ve:
- case llvm::Triple::wasm64:
- case llvm::Triple::x86_64:
- return 64;
- }
- llvm_unreachable("Invalid architecture value");
- }
上面源码需要注意的一个 CPU 架构是 aarch64_32,这种 ARM 架构的 CPU 虽然是 64bit 的,但是 int、long、pointer 都使用 32bit 表示(即 ILP32)。这种 CPU 通常用在嵌入式里面,Apple Watch Series 4/5 就是使用的这种 CPU:
由于 Apple 从 iPhone 5S 就开始支持 64bit 的 CPU,因此在 >= iPhone 5S 的设备上,SUPPORT_INDEXED_ISA 定义为0,SUPPORT_PACKED_ISA 定义为1。
__OBJC__
___OBJC__ 宏定义在 LLVM 源码的 InitPreprocessor.cpp 文件,源码如下:
- // file: InitPreprocessor.cpp
static void InitializeStandardPredefinedMacros(const TargetInfo &TI,- const LangOptions &LangOpts,
- const FrontendOptions &FEOpts,
- MacroBuilder &Builder) {
- ...
- if (LangOpts.ObjC)
- Builder.defineMacro("__OBJC__"); // 定义 __OBJC__ 宏
- ...
- }
从源码可以看到,如果编译的语言是Objective-C,那么这个宏就会被定义。
__OBJC2__
__OBJC2__ 宏定义在 LLVM 源码的 InitPreprocessor.cpp 文件,源码如下:
- // file: InitPreprocessor.cpp
- static void InitializePredefinedMacros(const TargetInfo &TI,
- const LangOptions &LangOpts,
- const FrontendOptions &FEOpts,
- const PreprocessorOptions &PPOpts,
- MacroBuilder &Builder) {
- ...
- if (LangOpts.ObjC) {
- if (LangOpts.ObjCRuntime.isNonFragile()) {
- Builder.defineMacro("__OBJC2__"); // 如果是 Objective-C 语言,并且满足 non fragile,就定义 __OBJC2__
- ...
- }
对于 __OBJC2__ 宏的定义中,除了判断是 Objectvie-C 语言,还需要判断 non-fragile 条件。该条件判断的源码如下:
- // file: ObjcRuntime.h
- bool isNonFragile() const {
- switch (getKind()) {
- case FragileMacOSX: return false;
- case GCC: return false;
- case MacOSX: return true; // Mac
- case GNUstep: return true;
- case ObjFW: return true;
- case iOS: return true; // iOS
- case WatchOS: return true; // Watch
- }
- llvm_unreachable("bad kind");
- }
从源码可以看到,对于 iOS 系统和 Watch OS 系统,__OBJC2__ 宏是一定会定义的。但是对于 MAC 系统就要区分 MacOSX 与 FragileMacOSX。这些类型的定义源码如下:
- // file: ObjcRuntime.h
- class ObjCRuntime {
- public:
- /// The basic Objective-C runtimes that we know about.
- enum Kind {
- /// 'macosx' is the Apple-provided NeXT-derived runtime on Mac OS
- /// X platforms that use the non-fragile ABI; the version is a
- /// release of that OS.
- MacOSX,
- /// 'macosx-fragile' is the Apple-provided NeXT-derived runtime on
- /// Mac OS X platforms that use the fragile ABI; the version is a
- /// release of that OS.
- FragileMacOSX,
- /// 'ios' is the Apple-provided NeXT-derived runtime on iOS or the iOS
- /// simulator; it is always non-fragile. The version is a release
- /// version of iOS.
- iOS,
- /// 'watchos' is a variant of iOS for Apple's watchOS. The version
- /// is a release version of watchOS.
- WatchOS,
- /// 'gcc' is the Objective-C runtime shipped with GCC, implementing a
- /// fragile Objective-C ABI
- GCC,
- /// 'gnustep' is the modern non-fragile GNUstep runtime.
- GNUstep,
- /// 'objfw' is the Objective-C runtime included in ObjFW
- ObjFW
- };
- ...
- }
__has_feature
__has_feature 宏可以帮助我们判断一个功能是否可以由 Clang 编译器支持,Clang 文档原文如下:
These function-like macros take a single identifier argument that is the name of a feature.
__has_feature
evaluates to 1 if the feature is both supported by Clang and standardized in the current language standard or 0 if not
那么它的实现是怎样的呢?
首先 Clang 会注册 __has_feature 宏,注册的结果被保存在 Preprocessor 对象的实例变量 Ident__has_feature 中。源码如下所示:
- /// file: PPMacroExpansion.cpp
- /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the
- /// identifier table.
- void Preprocessor::RegisterBuiltinMacros() {
- // 注入了许多常见的内置宏
- Ident__LINE__ = RegisterBuiltinMacro(*this, "__LINE__");
- Ident__FILE__ = RegisterBuiltinMacro(*this, "__FILE__");
- Ident__DATE__ = RegisterBuiltinMacro(*this, "__DATE__");
- Ident__TIME__ = RegisterBuiltinMacro(*this, "__TIME__");
- Ident__COUNTER__ = RegisterBuiltinMacro(*this, "__COUNTER__");
- Ident_Pragma = RegisterBuiltinMacro(*this, "_Pragma");
- ...
- // Clang Extensions.
- Ident__FILE_NAME__ = RegisterBuiltinMacro(*this, "__FILE_NAME__");
- Ident__has_feature = RegisterBuiltinMacro(*this, "__has_feature"); // __has_feature 被注入
- Ident__has_extension = RegisterBuiltinMacro(*this, "__has_extension"); // __has_extension 被注入
- Ident__has_builtin = RegisterBuiltinMacro(*this, "__has_builtin"); // __has_builtin 被注入
- ...
- }
当 Clang 预编译源文件时如果遇到了 __has_feature 标识符,就会进行扩展,扩展的代码如下所示:
- // file: PPMacroExpansion.cpp
- void Preprocessor::ExpandBuiltinMacro(Token &Tok) {
- ...
- } else if (II == Ident__has_feature) {
- EvaluateFeatureLikeBuiltinMacro(OS, Tok, II, *this, false,
- [this](Token &Tok, bool &HasLexedNextToken) -> int {
- IdentifierInfo *II = ExpectFeatureIdentifierInfo(Tok, *this,
- diag::err_feature_check_malformed);
- return II && HasFeature(*this, II->getName()); // 最终 Clang 编译器调用 HasFeature 函数进行判断
- });
- } else if (II == Ident__has_extension) {
- ...
- }
- ...
- }
从源码可以看到,Clang 的扩展结果通过调用 HasFeature 函数获取,HasFeature 函数接收要检测的功能名作为参数,源码如下:
- // file: PPMacroExpansion.cpp
- static bool HasFeature(const Preprocessor &PP, StringRef Feature) {
- const LangOptions &LangOpts = PP.getLangOpts();
- // Normalize the feature name, __foo__ becomes foo.
- if (Feature.startswith("__") && Feature.endswith("__") && Feature.size() >= 4)
- Feature = Feature.substr(2, Feature.size() - 4);
- #define FEATURE(Name, Predicate) .Case(#Name, Predicate) // 下面的 Feature.def 里面使用了大量的 FEATURE 宏
- return llvm::StringSwitch<bool>(Feature) // StringSwitch 是一个类,它支持对字符串进行 switch-case 操作
- #include "clang/Basic/Features.def" // 所有 Clang 支持的功能都定义在这个文件
- .Default(false); // 默认返回 false
- #undef FEATURE
- }
从源码可以看到,HasFeature 函数内部定义了一个 FEATURE 宏,这个宏在 Feature.def 文件中被大量使用,下面截取部分 Feature.def 文件内容:
- // file: Feature.def
- ...
- // Objective-C features
- FEATURE(objc_arr, LangOpts.ObjCAutoRefCount) // FIXME: REMOVE?
- FEATURE(objc_arc, LangOpts.ObjCAutoRefCount)
- FEATURE(objc_arc_fields, true) // ARC
- FEATURE(objc_arc_weak, LangOpts.ObjCWeak) // weak
- FEATURE(objc_default_synthesize_properties, LangOpts.ObjC)
- FEATURE(objc_fixed_enum, LangOpts.ObjC)
- FEATURE(objc_instancetype, LangOpts.ObjC) // instancetype
- FEATURE(objc_kindof, LangOpts.ObjC)
- ...
经过宏扩展之后,HasFeature 函数最后的 return 语句实际上变成为:
- return llvm::StringSwitch<bool>(Feature)
- ...
- .Case("objc_arr", LangOpts.ObjCAutoRefCount)
- .Case("objc_arc", LangOpts.ObjCAutoRefCount)
- .Case("objc_arc_fields", true)
- .Case("objc_arc_weak", LangOpts.ObjCWeak)
- .Case("objc_default_synthesize_properties", LangOpts.ObjC)
- .Case("objc_fixed_enum", LangOpts.ObjC)
- .Case("objc_instancetype", LangOpts.ObjC)
- .Case("objc_kindof", LangOpts.ObjC)
- ...
- .Default(false)
return 语句首先传入待检测的功能名,调用 StringSwitch 的构造函数生成一个 StringSwitch 对象,这个 StringSwitch 对象用来对 string 进行 switch-case 操作,它内部有 Case 和 Default 两个方法,定义如下:
- template<typename T, typename R = T>
- class StringSwitch {
- ...
- // Case-sensitive case matchers
- StringSwitch &Case(StringLiteral S, T Value) {
- if (!Result && Str == S) { // 如果 switch-case 没有匹配的结果,本次 Case 方法才进行比较,否则如果已经匹配出结果,直接返回对象本身
- Result = std::move(Value);
- }
- return *this; // 返回对象本身,形成链式调用
- }
- ..
- R Default(T Value) {
- if (Result)
- return std::move(*Result); // 匹配除了结果,直接返回结果
- return Value; // 未匹配出结果,返回默认值
- }
- ...
- }
从源码可以看到,Case 方法只有在未匹配出结果时,才进行匹配操作,如果结果已经匹配,Case 方法直接返回对象本身,这样就可以形成链式调用。链式调用最后,会调用到 Default 方法,如果已经匹配到结果,Default 方法直接返回对应的匹配结果,否则就返回默认值。