[PATCH] D69281: [FPEnv][WIP] Constrained FCmp intrinsics

[Ulrich Weigand via Phabricator via llvm-commits]

uweigand marked an inline comment as done.
uweigand added a comment.

Thanks for the review!



================
Comment at: llvm/lib/CodeGen/SelectionDAG/LegalizeVectorOps.cpp:1420
+    OpChains.push_back(ScalarChain);
   }
 
----------------
cameron.mcinally wrote:
> This throws away strict-ness for the expanded scalar selects. Is that deliberate?
> 
> I would have expected this to produce scalar STRICT_FSETCC nodes instead.
I don't see how this throws away strict-ness.   Note that we **do** produce scalar STRICT_FSETCC nodes (that's in the ScalarOp = DAG.getNode(Op->getOpcode() ...) line).   The additional select just ensures the (integral) output has the same value that a vector SETCC would have, instead of the (difference) value that a scalar SETCC has.   Since this is completely an integer operation, strict-ness doesn't apply.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D69281/new/

https://reviews.llvm.org/D69281