[PATCH] D49039: [BitcodeReader] Infer the correct runtime preemption for GlobalValue

[Vedant Kumar via Phabricator via llvm-commits]

vsk added a comment.

In https://reviews.llvm.org/D49039#1154893, @steven_wu wrote:

> > In that case, we can enable verification on checked-in bitcode files as a follow-up. How about testing this by round-tripping textual IR through llvm-as and llvm-dis? You just need a function, alias, and global var with local linkage and without dso_local set, right?
>
> Nope. dso_local is even implicit in assembly which is a completely different from bitcode reader/writer.
>
>   static void PrintDSOLocation(const GlobalValue &GV,
>                                formatted_raw_ostream &Out) {
>     // GVs with local linkage or non default visibility are implicitly dso_local,
>     // so we don't print it.
>     bool Implicit = GV.hasLocalLinkage() ||
>                     (!GV.hasExternalWeakLinkage() && !GV.hasDefaultVisibility());
>     if (GV.isDSOLocal() && !Implicit)
>       Out << "dso_local ";
>   }
>   
>
> I guess this patch is unifying the behavior, sort of.


Hm, if there's no simpler option, what about checking in bitcode produced by 6.0 which would have triggered the verifier?


Repository:
  rL LLVM

https://reviews.llvm.org/D49039