phoebewang wrote: > except for `bool` (which stays zeroext, as it requires zero-extension to 8 > bits) and Darwin.
Does i2~i7 follow the same rule? > However, this _does_ change behavior for non-i8/i16 types. Specifically, it > changes `_BitInt` types to have unspecified high bits, instead of performing > a partial sign/zero extension to 8, 16 or 32 bits. Does this change i32/i64 type or not? Is it limited to `_BitInt(N)` where N < 32? > Because the only case with actual behavior change is `_BitInt`, I'm not > bothering with `-fclang-abi-compat` for this change. We explicitly document > `_BitInt` as not ABI stable at this point. Sounds like if we exclude `_BitInt` type in the patch, it can be an NFC change? > This change does not affect handling of arguments. What about when passing the result immediately as an argument? https://github.com/llvm/llvm-project/pull/232729 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
