Major New Features¶

• A new Implicit Conversion Sanitizer (-fsanitize=implicit-conversion) group was added. Please refer to the Undefined Behavior Sanitizer (UBSan) section of the release notes for the details.
• Preliminary/experimental support for DWARF v5 debugging information. If you compile with -gdwarf-5 -O0 you should get fully conforming DWARF v5 information, including the new .debug_names accelerator table. Type units and split DWARF are known not to conform, and higher optimization levels will likely get a mix of v4 and v5 formats.

Improvements to Clang’s diagnostics¶

• -Wc++98-compat-extra-semi is a new flag, which was previously inseparable from -Wc++98-compat-pedantic. The latter still controls the new flag.
• -Wextra-semi now also controls -Wc++98-compat-extra-semi. Please do note that if you pass -Wno-c++98-compat-pedantic, it implies -Wno-c++98-compat-extra-semi, so if you want that diagnostic, you need to explicitly re-enable it (e.g. by appending -Wextra-semi).
• -Wself-assign and -Wself-assign-field were extended to diagnose self-assignment operations using overloaded operators (i.e. classes). If you are doing such an assignment intentionally, e.g. in a unit test for a data structure, the first warning can be disabled by passing -Wno-self-assign-overloaded, also the warning can be suppressed by adding *& to the right-hand side or casting it to the appropriate reference type.

Non-comprehensive list of changes in this release¶

• Clang binary and libraries have been renamed from 7.0 to 7. For example, the clang binary will be called clang-7 instead of clang-7.0.
• The optimization flag to merge constants (-fmerge-all-constants) is no longer applied by default.
• Clang implements a collection of recent fixes to the C++ standard’s definition of “standard-layout”. In particular, a class is only considered to be standard-layout if all base classes and the first data member (or bit-field) can be laid out at offset zero.
• Clang’s handling of the GCC packed class attribute in C++ has been fixed to apply only to non-static data members and not to base classes. This fixes an ABI difference between Clang and GCC, but creates an ABI difference between Clang 7 and earlier versions. The old behavior can be restored by setting -fclang-abi-compat to 6 or lower.
• Clang implements the proposed resolution of LWG issue 2358, along with the corresponding change to the Itanium C++ ABI, which make classes containing only unnamed non-zero-length bit-fields be considered non-empty. This is an ABI break compared to prior Clang releases, but makes Clang generate code that is ABI-compatible with other compilers. The old behavior can be restored by setting -fclang-abi-compat to 6 or lower.
• An existing tool named diagtool has been added to the release. As the name suggests, it helps with dealing with diagnostics in clang, such as finding out the warning hierarchy, and which of them are enabled by default or for a particular compiler invocation.
• By default, Clang emits an address-significance table into every ELF object file when using the integrated assembler. Address-significance tables allow linkers to implement safe ICF without the false positives that can result from other implementation techniques such as relocation scanning. The -faddrsig and -fno-addrsig flags can be used to control whether to emit the address-significance table.
• The integrated assembler is enabled by default on OpenBSD / FreeBSD for MIPS 64-bit targets.
• On MIPS FreeBSD, default CPUs have been changed to mips2 for 32-bit targets and mips3 for 64-bit targets.

New Compiler Flags¶

• -fstrict-float-cast-overflow and -fno-strict-float-cast-overflow.

  When converting a floating-point value to int and the value is not representable in the destination integer type, the code has undefined behavior according to the language standard. By default, Clang will not guarantee any particular result in that case. With the ‘no-strict’ option, Clang attempts to match the overflowing behavior of the target’s native float-to-int conversion instructions.

• -fforce-emit-vtables and -fno-force-emit-vtables.

  In order to improve devirtualization, forces emission of vtables even in modules where it isn’t necessary. It causes more inline virtual functions to be emitted.

• Added the -mcrc and -mno-crc flags to enable/disable using of MIPS Cyclic Redundancy Check instructions.

• Added the -mvirt and -mno-virt flags to enable/disable using of MIPS Virtualization instructions.

• Added the -mginv and -mno-ginv flags to enable/disable using of MIPS Global INValidate instructions.

Modified Compiler Flags¶

• Before Clang 7, we prepended the # character to the --autocomplete argument to enable cc1 flags. For example, when the -cc1 or -Xclang flag is in the clang invocation, the shell executed clang --autocomplete=#-<flag to be completed>. Clang 7 now requires the whole invocation including all flags to be passed to the --autocomplete like this: clang --autocomplete=-cc1,-xc++,-fsyn.

Attribute Changes in Clang¶

• Clang now supports function multiversioning with attribute ‘target’ on ELF based x86/x86-64 environments by using indirect functions. This implementation has a few minor limitations over the GCC implementation for the sake of AST sanity, however it is otherwise compatible with existing code using this feature for GCC. Consult the documentation for the target attribute for more information.

Windows Support¶

• clang-cl’s support for precompiled headers has been much improved:

  • When using a pch file, clang-cl now no longer redundantly emits inline methods that are already stored in the obj that was built together with the pch file (matching cl.exe). This speeds up builds using pch files by around 30%.
  • The /Ycfoo.h and /Yufoo.h flags can now be used without /FIfoo.h when foo.h is instead included by an explicit #include directive. This means Visual Studio’s default stdafx.h setup now uses precompiled headers with clang-cl.

• The alternative entry point names (wmain/WinMain/wWinMain/DllMain) now are properly mangled as plain C names in C++ contexts when targeting MinGW, without having to explicit specify extern "C". (This was already the case for MSVC targets.)

Objective-C Language Changes in Clang¶

Clang now supports the GNUstep Objective-C ABI v2 on ELF platforms. This is enabled with the -fobjc-runtime=gnustep-2.0 flag. The new ABI is incompatible with the older GNUstep ABIs, which were incremental changes on the old GCC ABI. The new ABI provides richer reflection metadata and allows the linker to remove duplicate selector and protocol definitions, giving smaller binaries. Windows support for the new ABI is underway, but was not completed in time for the LLVM 7.0.0 release.

OpenCL C/C++ Language Changes in Clang¶

Miscellaneous changes in OpenCL C:

• Added cles_khr_int64 extension.
• Added bug fixes and simplifications to Clang blocks in OpenCL mode.
• Added compiler flag -cl-uniform-work-group-size to allow extra compile time optimisation.
• Propagate denorms-are-zero attribute to IR if -cl-denorms-are-zero is passed to the compiler.
• Separated read_only and write_only pipe IR types.
• Fixed address space for the __func__ predefined macro.
• Improved diagnostics of kernel argument types.

Started OpenCL C++ support:

• Added -std/-cl-std=c++.
• Added support for keywords.

OpenMP Support in Clang¶

• Clang gained basic support for OpenMP 4.5 offloading for NVPTX target.

  To compile your program for NVPTX target use the following options: -fopenmp -fopenmp-targets=nvptx64-nvidia-cuda for 64 bit platforms or -fopenmp -fopenmp-targets=nvptx-nvidia-cuda for 32 bit platform.

• Passing options to the OpenMP device offloading toolchain can be done using the -Xopenmp-target=<triple> -opt=val flag. In this way the -opt=val option will be forwarded to the respective OpenMP device offloading toolchain described by the triple. For example passing the compute capability to the OpenMP NVPTX offloading toolchain can be done as follows: -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_60. For the case when only one target offload toolchain is specified under the -fopenmp-targets=<triples> option, then the triple can be skipped: -Xopenmp-target -march=sm_60.

• Other bugfixes.

CUDA Support in Clang¶

• Clang will now try to locate the CUDA installation next to ptxas in the PATH environment variable. This behavior can be turned off by passing the new flag --cuda-path-ignore-env.
• Clang now supports generating object files with relocatable device code. This feature needs to be enabled with -fcuda-rdc and may result in performance penalties compared to whole program compilation. Please note that NVIDIA’s nvcc must be used for linking.

Internal API Changes¶

These are major API changes that have happened since the 6.0.0 release of Clang. If upgrading an external codebase that uses Clang as a library, this section should help get you past the largest hurdles of upgrading.

• The methods getLocStart, getStartLoc and getLocEnd in the AST classes are deprecated. New APIs getBeginLoc and getEndLoc should be used instead. While the old methods remain in this release, they will not be present in the next release of Clang.

clang-format¶

• Clang-format will now support detecting and formatting code snippets in raw string literals. This is configured through the RawStringFormats style option.

Static Analyzer¶

• The new MmapWriteExec checker had been introduced to detect attempts to map pages both writable and executable.

Undefined Behavior Sanitizer (UBSan)¶

• A new Implicit Conversion Sanitizer (-fsanitize=implicit-conversion) group was added.

  Currently, only one type of issues is caught - implicit integer truncation (-fsanitize=implicit-integer-truncation), also known as integer demotion. While there is a -Wconversion diagnostic group that catches this kind of issues, it is both noisy, and does not catch all the cases.

  unsigned char store = 0;
  
  bool consume(unsigned int val);
  
  void test(unsigned long val) {
    if (consume(val)) // the value may have been silently truncated.
      store = store + 768; // before addition, 'store' was promoted to int.
    (void)consume((unsigned int)val); // OK, the truncation is explicit.
  }
  

  Just like other -fsanitize=integer checks, these issues are not undefined behaviour. But they are not always intentional, and are somewhat hard to track down. This group is not enabled by -fsanitize=undefined, but the -fsanitize=implicit-integer-truncation check is enabled by -fsanitize=integer.