[PATCH] D68364: Implement C++20's P0784 (More constexpr containers)

[Yvan Roux via cfe-commits]

Hi Louis,

ARM and AArch64 libcxx bots are still broken after this commit (host
compiler is clang-8), logs are available here:
http://lab.llvm.org:8011/builders/libcxx-libcxxabi-libunwind-aarch64-linux/builds/3143

Cheers,
Yvan

On Mon, 28 Sep 2020 at 23:44, Louis Dionne via Phabricator via
cfe-commits <cfe-commits at lists.llvm.org> wrote:
>
> ldionne added a comment.
>
> In D68364#2294569 <https://reviews.llvm.org/D68364#2294569>, @jwakely wrote:
>
> > In D68364#2293971 <https://reviews.llvm.org/D68364#2293971>, @leonardchan wrote:
> >
> >> Is there a recommended way for working around this? We're using GCC 10.2.1. Thanks.
> >
> > I don't think your implementation is valid. I think P0784 only allows new-expressions and calls to `std::allocator::allocate` in constexpr functions, not calls to `operator new`.
>
> Hmm, right. It's actually impossible to implement `__libcpp_allocate` in a constexpr-friendly manner, since it's just allocating bytes.
>
> > Can you use something like `if (__builtin_is_constant_evaluated()) return new unsigned char[n];` and the equivalent in the corresponding deallocation function?
>
> That doesn't work, since we can't use `unsigned char`s as storage for arbitrary `T`s in `constexpr`.
>
> I fixed it by marking `__libcpp_allocate` as non-`constexpr` and calling `operator new` in `allocator<T>::allocate`, which seems to work on GCC.
>
>
> Repository:
>   rG LLVM Github Monorepo
>
> CHANGES SINCE LAST ACTION
>   https://reviews.llvm.org/D68364/new/
>
> https://reviews.llvm.org/D68364
>
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits