[clang] [libcxx] [C++17] Support __GCC_[CON|DE]STRUCTIVE_SIZE (PR #89446)

Aaron Ballman via cfe-commits cfe-commits at lists.llvm.org
Wed Apr 24 10:28:07 PDT 2024


AaronBallman wrote:

> I'd rather keep the libc++ tests green while landing this PR. I tried something out, let's see if that works.

It looks like some further adjustments may be needed as some stage1 builders are failing:
```
# .---command stderr------------
# | /home/runner/_work/llvm-project/llvm-project/libcxx/test/std/language.support/support.dynamic/hardware_inference_size.compile.pass.cpp:16:32: error: no member named 'hardware_destructive_interference_size' in namespace 'std'
# |    16 | ASSERT_SAME_TYPE(decltype(std::hardware_destructive_interference_size), const std::size_t);
# |       |                           ~~~~~^
# | /home/runner/_work/llvm-project/llvm-project/libcxx/test/support/test_macros.h:271:48: note: expanded from macro 'ASSERT_SAME_TYPE'
# |   271 |     static_assert((test_macros_detail::is_same<__VA_ARGS__>::value), \
# |       |                                                ^~~~~~~~~~~
# | /home/runner/_work/llvm-project/llvm-project/libcxx/test/std/language.support/support.dynamic/hardware_inference_size.compile.pass.cpp:17:32: error: no member named 'hardware_constructive_interference_size' in namespace 'std'
# |    17 | ASSERT_SAME_TYPE(decltype(std::hardware_constructive_interference_size), const std::size_t);
# |       |                           ~~~~~^
# | /home/runner/_work/llvm-project/llvm-project/libcxx/test/support/test_macros.h:271:48: note: expanded from macro 'ASSERT_SAME_TYPE'
# |   271 |     static_assert((test_macros_detail::is_same<__VA_ARGS__>::value), \
# |       |                                                ^~~~~~~~~~~
# | 2 errors generated.
# `-----------------------------
```
I'm not quite certain what's causing the failure though

> Otherwise LGTM, thanks for tackling this long-standing item on our TODO list :-)

Happy to help!


https://github.com/llvm/llvm-project/pull/89446


More information about the cfe-commits mailing list