//===--- TargetCXXABI.h - C++ ABI Target Configuration ----------*- C++ -*-===// // // The LLVM Compiler Infrastructure // // This file is distributed under the University of Illinois Open Source // License. See LICENSE.TXT for details. // //===----------------------------------------------------------------------===// /// /// \file /// Defines the TargetCXXABI class, which abstracts details of the /// C++ ABI that we're targeting. /// //===----------------------------------------------------------------------===// #ifndef LLVM_CLANG_BASIC_TARGETCXXABI_H #define LLVM_CLANG_BASIC_TARGETCXXABI_H #include "llvm/Support/ErrorHandling.h" namespace clang { /// The basic abstraction for the target C++ ABI. class TargetCXXABI { public: /// The basic C++ ABI kind. enum Kind { /// The generic Itanium ABI is the standard ABI of most open-source /// and Unix-like platforms. It is the primary ABI targeted by /// many compilers, including Clang and GCC. /// /// It is documented here: /// http://www.codesourcery.com/public/cxx-abi/ GenericItanium, /// The generic ARM ABI is a modified version of the Itanium ABI /// proposed by ARM for use on ARM-based platforms. /// /// These changes include: /// - the representation of member function pointers is adjusted /// to not conflict with the 'thumb' bit of ARM function pointers; /// - constructors and destructors return 'this'; /// - guard variables are smaller; /// - inline functions are never key functions; /// - array cookies have a slightly different layout; /// - additional convenience functions are specified; /// - and more! /// /// It is documented here: /// http://infocenter.arm.com /// /help/topic/com.arm.doc.ihi0041c/IHI0041C_cppabi.pdf GenericARM, /// The iOS ABI is a partial implementation of the ARM ABI. /// Several of the features of the ARM ABI were not fully implemented /// in the compilers that iOS was launched with. /// /// Essentially, the iOS ABI includes the ARM changes to: /// - member function pointers, /// - guard variables, /// - array cookies, and /// - constructor/destructor signatures. iOS, /// The iOS 64-bit ABI is follows ARM's published 64-bit ABI more /// closely, but we don't guarantee to follow it perfectly. /// /// It is documented here: /// http://infocenter.arm.com /// /help/topic/com.arm.doc.ihi0059a/IHI0059A_cppabi64.pdf iOS64, /// WatchOS is a modernisation of the iOS ABI, which roughly means it's /// the iOS64 ABI ported to 32-bits. The primary difference from iOS64 is /// that RTTI objects must still be unique at the moment. WatchOS, /// The generic AArch64 ABI is also a modified version of the Itanium ABI, /// but it has fewer divergences than the 32-bit ARM ABI. /// /// The relevant changes from the generic ABI in this case are: /// - representation of member function pointers adjusted as in ARM. /// - guard variables are smaller. GenericAArch64, /// The generic Mips ABI is a modified version of the Itanium ABI. /// /// At the moment, only change from the generic ABI in this case is: /// - representation of member function pointers adjusted as in ARM. GenericMIPS, /// The WebAssembly ABI is a modified version of the Itanium ABI. /// /// The changes from the Itanium ABI are: /// - representation of member function pointers is adjusted, as in ARM; /// - member functions are not specially aligned; /// - constructors and destructors return 'this', as in ARM; /// - guard variables are 32-bit on wasm32, as in ARM; /// - unused bits of guard variables are reserved, as in ARM; /// - inline functions are never key functions, as in ARM; /// - C++11 POD rules are used for tail padding, as in iOS64. /// /// TODO: At present the WebAssembly ABI is not considered stable, so none /// of these details is necessarily final yet. WebAssembly, /// The Microsoft ABI is the ABI used by Microsoft Visual Studio (and /// compatible compilers). /// /// FIXME: should this be split into Win32 and Win64 variants? /// /// Only scattered and incomplete official documentation exists. Microsoft }; private: // Right now, this class is passed around as a cheap value type. // If you add more members, especially non-POD members, please // audit the users to pass it by reference instead. Kind TheKind; public: /// A bogus initialization of the platform ABI. TargetCXXABI() : TheKind(GenericItanium) {} TargetCXXABI(Kind kind) : TheKind(kind) {} void set(Kind kind) { TheKind = kind; } Kind getKind() const { return TheKind; } /// Does this ABI generally fall into the Itanium family of ABIs? bool isItaniumFamily() const { switch (getKind()) { case GenericAArch64: case GenericItanium: case GenericARM: case iOS: case iOS64: case WatchOS: case GenericMIPS: case WebAssembly: return true; case Microsoft: return false; } llvm_unreachable("bad ABI kind"); } /// Is this ABI an MSVC-compatible ABI? bool isMicrosoft() const { switch (getKind()) { case GenericAArch64: case GenericItanium: case GenericARM: case iOS: case iOS64: case WatchOS: case GenericMIPS: case WebAssembly: return false; case Microsoft: return true; } llvm_unreachable("bad ABI kind"); } /// Are member functions differently aligned? /// /// Many Itanium-style C++ ABIs require member functions to be aligned, so /// that a pointer to such a function is guaranteed to have a zero in the /// least significant bit, so that pointers to member functions can use that /// bit to distinguish between virtual and non-virtual functions. However, /// some Itanium-style C++ ABIs differentiate between virtual and non-virtual /// functions via other means, and consequently don't require that member /// functions be aligned. bool areMemberFunctionsAligned() const { switch (getKind()) { case WebAssembly: // WebAssembly doesn't require any special alignment for member functions. return false; case GenericARM: case GenericAArch64: case GenericMIPS: // TODO: ARM-style pointers to member functions put the discriminator in // the this adjustment, so they don't require functions to have any // special alignment and could therefore also return false. case GenericItanium: case iOS: case iOS64: case WatchOS: case Microsoft: return true; } llvm_unreachable("bad ABI kind"); } /// Are arguments to a call destroyed left to right in the callee? /// This is a fundamental language change, since it implies that objects /// passed by value do *not* live to the end of the full expression. /// Temporaries passed to a function taking a const reference live to the end /// of the full expression as usual. Both the caller and the callee must /// have access to the destructor, while only the caller needs the /// destructor if this is false. bool areArgsDestroyedLeftToRightInCallee() const { return isMicrosoft(); } /// Does this ABI have different entrypoints for complete-object /// and base-subobject constructors? bool hasConstructorVariants() const { return isItaniumFamily(); } /// Does this ABI allow virtual bases to be primary base classes? bool hasPrimaryVBases() const { return isItaniumFamily(); } /// Does this ABI use key functions? If so, class data such as the /// vtable is emitted with strong linkage by the TU containing the key /// function. bool hasKeyFunctions() const { return isItaniumFamily(); } /// Can an out-of-line inline function serve as a key function? /// /// This flag is only useful in ABIs where type data (for example, /// vtables and type_info objects) are emitted only after processing /// the definition of a special "key" virtual function. (This is safe /// because the ODR requires that every virtual function be defined /// somewhere in a program.) This usually permits such data to be /// emitted in only a single object file, as opposed to redundantly /// in every object file that requires it. /// /// One simple and common definition of "key function" is the first /// virtual function in the class definition which is not defined there. /// This rule works very well when that function has a non-inline /// definition in some non-header file. Unfortunately, when that /// function is defined inline, this rule requires the type data /// to be emitted weakly, as if there were no key function. /// /// The ARM ABI observes that the ODR provides an additional guarantee: /// a virtual function is always ODR-used, so if it is defined inline, /// that definition must appear in every translation unit that defines /// the class. Therefore, there is no reason to allow such functions /// to serve as key functions. /// /// Because this changes the rules for emitting type data, /// it can cause type data to be emitted with both weak and strong /// linkage, which is not allowed on all platforms. Therefore, /// exploiting this observation requires an ABI break and cannot be /// done on a generic Itanium platform. bool canKeyFunctionBeInline() const { switch (getKind()) { case GenericARM: case iOS64: case WebAssembly: case WatchOS: return false; case GenericAArch64: case GenericItanium: case iOS: // old iOS compilers did not follow this rule case Microsoft: case GenericMIPS: return true; } llvm_unreachable("bad ABI kind"); } /// When is record layout allowed to allocate objects in the tail /// padding of a base class? /// /// This decision cannot be changed without breaking platform ABI /// compatibility, and yet it is tied to language guarantees which /// the committee has so far seen fit to strengthen no less than /// three separate times: /// - originally, there were no restrictions at all; /// - C++98 declared that objects could not be allocated in the /// tail padding of a POD type; /// - C++03 extended the definition of POD to include classes /// containing member pointers; and /// - C++11 greatly broadened the definition of POD to include /// all trivial standard-layout classes. /// Each of these changes technically took several existing /// platforms and made them permanently non-conformant. enum TailPaddingUseRules { /// The tail-padding of a base class is always theoretically /// available, even if it's POD. This is not strictly conforming /// in any language mode. AlwaysUseTailPadding, /// Only allocate objects in the tail padding of a base class if /// the base class is not POD according to the rules of C++ TR1. /// This is non-strictly conforming in C++11 mode. UseTailPaddingUnlessPOD03, /// Only allocate objects in the tail padding of a base class if /// the base class is not POD according to the rules of C++11. UseTailPaddingUnlessPOD11 }; TailPaddingUseRules getTailPaddingUseRules() const { switch (getKind()) { // To preserve binary compatibility, the generic Itanium ABI has // permanently locked the definition of POD to the rules of C++ TR1, // and that trickles down to derived ABIs. case GenericItanium: case GenericAArch64: case GenericARM: case iOS: case GenericMIPS: return UseTailPaddingUnlessPOD03; // iOS on ARM64 and WebAssembly use the C++11 POD rules. They do not honor // the Itanium exception about classes with over-large bitfields. case iOS64: case WebAssembly: case WatchOS: return UseTailPaddingUnlessPOD11; // MSVC always allocates fields in the tail-padding of a base class // subobject, even if they're POD. case Microsoft: return AlwaysUseTailPadding; } llvm_unreachable("bad ABI kind"); } friend bool operator==(const TargetCXXABI &left, const TargetCXXABI &right) { return left.getKind() == right.getKind(); } friend bool operator!=(const TargetCXXABI &left, const TargetCXXABI &right) { return !(left == right); } }; } // end namespace clang #endif