[PATCH] D24797: Only insert debug nodes after a phi if there is a valid insertion point

[Hans Wennborg via llvm-commits]

hans added a comment.

I'm not familiar with this code. I suppose it boils down to whether it's OK that ConvertDebugDeclareToDebugValue() is now best-effort instead of always inserting the dbg intrinsic. If that's the case, then lgtm.


https://reviews.llvm.org/D24797