aobolensk wrote: > Looks reasonable to me, but I am not sure I fully comprehend the implications > of moving the undef/posion initializer lowering to earlier.
The lowering of undef/poison is correct, but it was performed only in case of `if (auto recordTy = mlir::dyn_cast<cir::RecordType>(ty))` So before my patch undef/poison were acceptable only as a part of the struct (field), but now they are lowered this way in any case https://github.com/llvm/llvm-project/pull/230901 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
