[SCEVExpander] Only create multiplication if needed.

9345ab3a4550 updated generateOverflowCheck to skip creating checks that
always evaluate to false. This in turn means that we only need to
compute |Step| * Trip count  if the result of the multiplication is
actually used.

Sink the multiplication into ComputeEndCheck, so it is only created
when there's an actual check.

1 files changed, 16 insertions, 15 deletions

diff --git a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
index d604623..70d284f 100644
--- a/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
+++ b/llvm/lib/Transforms/Utils/ScalarEvolutionExpander.cpp
@@ -2494,21 +2494,6 @@ Value *SCEVExpander::generateOverflowCheck(const SCEVAddRecExpr *AR,
   Value *TruncTripCount = Builder.CreateZExtOrTrunc(TripCountVal, Ty);
 
   // Compute |Step| * Backedge
-  Value *MulV, *OfMul;
-  if (Step->isOne()) {
-    // Special-case Step of one. Potentially-costly `umul_with_overflow` isn't
-    // needed, there is never an overflow, so to avoid artificially inflating
-    // the cost of the check, directly emit the optimized IR.
-    MulV = TruncTripCount;
-    OfMul = ConstantInt::getFalse(MulV->getContext());
-  } else {
-    auto *MulF = Intrinsic::getDeclaration(Loc->getModule(),
-                                           Intrinsic::umul_with_overflow, Ty);
-    CallInst *Mul = Builder.CreateCall(MulF, {AbsStep, TruncTripCount}, "mul");
-    MulV = Builder.CreateExtractValue(Mul, 0, "mul.result");
-    OfMul = Builder.CreateExtractValue(Mul, 1, "mul.overflow");
-  }
-
   // Compute:
   //   1. Start + |Step| * Backedge < Start
   //   2. Start - |Step| * Backedge > Start
@@ -2521,6 +2506,22 @@ Value *SCEVExpander::generateOverflowCheck(const SCEVAddRecExpr *AR,
     if (!Signed && Start->isZero() && SE.isKnownPositive(Step))
       return ConstantInt::getFalse(Loc->getContext());
 
+    Value *MulV, *OfMul;
+    if (Step->isOne()) {
+      // Special-case Step of one. Potentially-costly `umul_with_overflow` isn't
+      // needed, there is never an overflow, so to avoid artificially inflating
+      // the cost of the check, directly emit the optimized IR.
+      MulV = TruncTripCount;
+      OfMul = ConstantInt::getFalse(MulV->getContext());
+    } else {
+      auto *MulF = Intrinsic::getDeclaration(Loc->getModule(),
+                                             Intrinsic::umul_with_overflow, Ty);
+      CallInst *Mul =
+          Builder.CreateCall(MulF, {AbsStep, TruncTripCount}, "mul");
+      MulV = Builder.CreateExtractValue(Mul, 0, "mul.result");
+      OfMul = Builder.CreateExtractValue(Mul, 1, "mul.overflow");
+    }
+
     Value *Add = nullptr, *Sub = nullptr;
     bool NeedPosCheck = !SE.isKnownNegative(Step);
     bool NeedNegCheck = !SE.isKnownPositive(Step);