[PATCH] D70096: [strictfp] Replace dangling strictfp attrs with nobuiltin

[Kevin P. Neal via Phabricator via llvm-commits]

kpn added a comment.

In D70096#1741280 <https://reviews.llvm.org/D70096#1741280>, @andrew.w.kaylor wrote:

> In D70096#1741213 <https://reviews.llvm.org/D70096#1741213>, @kpn wrote:
>
> > Do we want to restrict this so it doesn't get upgraded if found in a future bitcode version?
>
>
> How would we do that?


BitcodeReaderBase knows the version number but it doesn't store it anywhere. It does use it to set UseStrtab, but that's it.

Thinking about it more, this incorrect bitcode case can only happen going forward if someone had a different bitcode writer implementation, or at least an llvm one that was hacked to allow this strictfp error. That seems like a stretch. So maybe it doesn't matter.

>> Do we want to not upgrade if a constrained intrinsic is used anywhere in the function? We do want to be able to distinguish between pre-constrained intrinsic uses and modern constrained uses that are wrong. I am assuming that the upgrade happens before the IR verifier runs.
> 
> Honestly, I'd be very surprised if anyone is using this in a product currently.

If nobody is using it then errors are pretty moot. When I asked for input from front-end guys of any language about strictfp changes for the IRBuilder I only got "don't care" responses because nobody who responded wanted to use strictfp. So it seems safe to just assume you are correct here.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D70096/new/

https://reviews.llvm.org/D70096