================
@@ -6418,6 +6418,12 @@ def warn_signed_bitfield_enum_conversion : Warning<
InGroup<BitFieldEnumConversion>, DefaultIgnore;
def note_change_bitfield_sign : Note<
"consider making the bitfield type %select{unsigned|signed}0">;
+def warn_ms_bitfield_mismatched_storage_packing : Warning<
+ "bit-field %0 of type %1 has a different storage size (%2 vs %3 bytes) than
the "
+ "preceding bit-field and may not be packed under MSVC ABI">,
+ InGroup<MSBitfieldCompatibility>, DefaultIgnore;
----------------
mstorsjo wrote:
Yes; mingw also uses the MS bitfield packing logic. (Originally, for mingw
targets, users would need to opt in to this with `-mms-bitfields`; GCC enabled
this option by default for mingw targets in 4.7 and we followed suit a bit
belatedly in Clang 11 - see e3fd9dc9734c5775dc6824d0a839702e8d43e7f6.)
As for whether this should be enabled by deault; I think it's entirely up to
each individual project whether they want it as an early portability headsup
that they're writing code which is behaving differently on Windows or not.
People writing windows-exclusive code might not care, and they may want to
intentionally keep their structs as they are to match some existing ABI.
https://github.com/llvm/llvm-project/pull/117428
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
