[PATCH] D62979: [InstSimplify] enhance/fix fcmp fold with never-nan operand

[Cameron McInally via Phabricator via llvm-commits]

cameron.mcinally added a comment.

This LGTM, with one clarification inline...



================
Comment at: llvm/test/Transforms/InstSimplify/floating-point-compare.ll:371
+; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %fabs = tail call double @llvm.fabs.f64(double %x)
----------------
Why can't we infer that the operand here is >=0 (from the fabs) and not a NaN (from the fcmp's nnan)? Is it not safe to combine that information? 

I suppose a poison value could be lost...


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D62979/new/

https://reviews.llvm.org/D62979