[llvm-dev] [cfe-dev] Zero length function pointer equality (original) (raw)
David Blaikie via llvm-dev llvm-dev at lists.llvm.org
Mon Mar 22 08:56:55 PDT 2021
- Previous message: [llvm-dev] [cfe-dev] Zero length function pointer equality
- Next message: [llvm-dev] [RFC] The `implements` attribute, or how to swap functions statically but late
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Mon, Mar 22, 2021 at 7:39 AM <paul.robinson at sony.com> wrote:
> I'm inclined to omit the trap after a call to a noreturn function > for brevity - even though it does leave the possibility of > violating the noreturn contract leading to that fallthrough UB.
This is specifically the case where we needed it, at least when the noreturn call is in the last block. Which is commonly the case, a noreturn call is likely to have its block laid out at the end of a function. --paulr
Yeah, not suggesting removing that option or changing those who've opted into that behavior - only about the default when platforms haven't otherwise cared/have already been living with zero length functions.
- Dave -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20210322/992c445e/attachment.html>
- Previous message: [llvm-dev] [cfe-dev] Zero length function pointer equality
- Next message: [llvm-dev] [RFC] The `implements` attribute, or how to swap functions statically but late
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]