[llvm-branch-commits] [llvm] InstCombine: Fold out nanless canonicalize pattern (PR #172998)

[Joshua Cranmer via llvm-branch-commits]

================
@@ -4343,6 +4343,71 @@ Instruction *InstCombinerImpl::visitSelectInst(SelectInst &SI) {
           matchFMulByZeroIfResultEqZero(*this, Cmp0, Cmp1, MatchCmp1, MatchCmp0,
                                         SI, SIFPOp->hasNoSignedZeros()))
         return replaceInstUsesWith(SI, Cmp0);
+
+      if (Pred == CmpInst::FCMP_ORD || Pred == CmpInst::FCMP_UNO) {
+        // Fold out only-canonicalize-non-nans pattern. This implements a
+        // wrapper around llvm.canonicalize which is not required to quiet
+        // signaling nans or preserve nan payload bits.
+        //
+        //   %hard.canonical = call @llvm.canonicalize(%x)
+        //   %soft.canonical = fdiv 1.0, %x
+        //   %ord = fcmp ord %x, 0.0
+        //   %x.canon = select i1 %ord, %hard.canonical, %soft.canonical
+        //
+        // With known IEEE handling:
+        //   => %x
+        //
+        // With other denormal behaviors or exotic types:
----------------
jcranmer-intel wrote:

The comment here implies that we're doing the transform for the exotic long double types, but the code itself is not doing so (and I agree with the code over the comment here, I'm *extremely* wary of getting rid of canonicalizes for these types).

https://github.com/llvm/llvm-project/pull/172998