[PATCH] D52307: [LoopRotate] Fix lifetime handling.

[Florian Hahn via Phabricator via llvm-commits]

fhahn added a comment.

In https://reviews.llvm.org/D52307#1241165, @efriedma wrote:

> An alloca can have multiple associated lifetime start/ends.  Inlining routinely generates code like that, unrolling generates code like that, etc.  And stack coloring knows how to handle such code.
>
> Granted, the way the intrinsic is specified makes everything related to it pretty confusing.


Right! If everything is working as intended currently, I'd be happy to clarify the LangRef and track down where things go wrong with those lifetimes later.

I guess both interpretations of the lifetime behavior make sense, what's in the LangRef currently seems to make using the info provided by the intrinsics easier to use (no need to check for other lifetime markers to the same location), having multiple distinct ranges for the same location increases precision.


https://reviews.llvm.org/D52307