[PATCH] D39361: [SCEV] A different fix for PR33494

[Sanjoy Das via Phabricator via llvm-commits]

sanjoy added inline comments.


================
Comment at: lib/Analysis/ScalarEvolution.cpp:6384
+          if (LoopForUser && L->contains(LoopForUser))
+            Worklist.push_back(I);
+        }
----------------
mkazantsev wrote:
> Do we want to check whether the instruction is already enqueued before we push it?
That's what the existing behavior is -- I'll prefer fixing it in a separate NFC change if you don't mind.


https://reviews.llvm.org/D39361