andykaylor wrote: > Does alloca operations emitted by CIRGen always have the alignment attribute > set? If so, maybe we could just make the alignment attribute mandatory. > Otherwise I hope we could have a CIRGen test for that.
Good question. I see that the alignment attribute is marked as optional, but we don't have a builder without an alignment argument. @Men-cotton How did you encounter the crash you are fixing here? Fuzz testing? The LLVM Language Reference says this about alignment on the `alloca` instruction: `The alignment is only optional when parsing textual IR; for in-memory IR, it is always present. An omitted align argument means that the operation has the ABI alignment for the target.` I'm not sure that defaulting to zero matches those semantics. We should probably just make it a required attribute in CIR. https://github.com/llvm/llvm-project/pull/172663 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
