================ @@ -525,13 +519,12 @@ Bug Fixes in This Version - Clang now emits an error for friend declarations of lambda members. (#GH26540) - Fixed a crash caused by lambda capture handling in delayed default arguments. (#GH176534) - Fixed a crash when parsing invalid ``static_assert`` declarations with string-literal messages (#GH187690). +- Fixed an assert/crash in ``__builtin_va_arg`` when the argument type is an enum without a correspoding signed type (#GH191698). ---------------- DavidSpickett wrote:
I haven't reproduced the bug myself but I think users would think of this purely as a crash so you can say "a crash" rather than "assert/crash". You're not wrong to say that but most people won't care that the assert was the cause. Fix the spelling of: correspoding -> corresponding Can you explain a bit more what corresponding means here? Does it mean that the crash happened when you have an enum with a underlying type that doesn't have an equivalent signed type? So looking at the test case, I see `char16_t` and I assume there is no `char16_signed_t`. In which case I would expand the wording a bit: > when the argument type is an enum whose underlying type does not have a > corresponding signed type I haven't read the standards but I think the type of the enum is the enum itself, then there's the underlying type and the underlying type is the key here. https://github.com/llvm/llvm-project/pull/195945 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
