[SimplifyLibCalls] pow(x, 0.5) -> sqrt(x) in fast-math mode

[davide]

Summary

This (also) fixes a longstanding TODO.

[scanon]

LGTM

llvm/trunk/lib/Transforms/Utils/SimplifyLibCalls.cpp

if (ConstantFP *Op1C = dyn_cast<ConstantFP>(Op1)) {
     // pow(10.0, x) -> exp10(x)
     if (Op1C->isExactlyValue(10.0) &&
         hasUnaryFloatFn(TLI, Op1->getType(), LibFunc::exp10, LibFunc::exp10f,
                         LibFunc::exp10l))
       return EmitUnaryFloatFnCall(Op2, TLI->getName(LibFunc::exp10), B,
                                   Callee->getAttributes());
   }
 
+  bool unsafeFPMath = canUseUnsafeFPMath(CI->getParent()->getParent());
+
   // pow(exp(x), y) -> exp(x*y)
   // pow(exp2(x), y) -> exp2(x * y)
   // We enable these only under fast-math. Besides rounding
   // differences the transformation changes overflow and
   // underflow behavior quite dramatically.
   // Example: x = 1000, y = 0.001.
   // pow(exp(x), y) = pow(inf, 0.001) = inf, whereas exp(x*y) = exp(1).
-  if (canUseUnsafeFPMath(CI->getParent()->getParent())) {
+  if (unsafeFPMath) {
     if (auto *OpC = dyn_cast<CallInst>(Op1)) {
       IRBuilder<>::FastMathFlagGuard Guard(B);
       FastMathFlags FMF;
       FMF.setUnsafeAlgebra();
       B.SetFastMathFlags(FMF);
 
       LibFunc::Func Func;
       Function *Callee = OpC->getCalledFunction();
Value *LibCallSimplifier::optimizePow(CallInst *CI, IRBuilder<> &B) {
   if (Op2C->getValueAPF().isZero()) // pow(x, 0.0) -> 1.0
     return ConstantFP::get(CI->getType(), 1.0);
 
   if (Op2C->isExactlyValue(0.5) &&
       hasUnaryFloatFn(TLI, Op2->getType(), LibFunc::sqrt, LibFunc::sqrtf,
                       LibFunc::sqrtl) &&
       hasUnaryFloatFn(TLI, Op2->getType(), LibFunc::fabs, LibFunc::fabsf,
                       LibFunc::fabsl)) {
+
+    // In -ffast-math, pow(x, 0.5) -> sqrt(x).
+    if (unsafeFPMath)
+      return EmitUnaryFloatFnCall(Op1, TLI->getName(LibFunc::sqrt), B,
+                                  Callee->getAttributes());
+
     // Expand pow(x, 0.5) to (x == -infinity ? +infinity : fabs(sqrt(x))).
     // This is faster than calling pow, and still handles negative zero
     // and negative infinity correctly.
-    // TODO: In fast-math mode, this could be just sqrt(x).
     // TODO: In finite-only mode, this could be just fabs(sqrt(x)).
     Value *Inf = ConstantFP::getInfinity(CI->getType());
     Value *NegInf = ConstantFP::getInfinity(CI->getType(), true);
     Value *Sqrt = EmitUnaryFloatFnCall(Op1, "sqrt", B, Callee->getAttributes());
     Value *FAbs =
         EmitUnaryFloatFnCall(Sqrt, "fabs", B, Callee->getAttributes());
     Value *FCmp = B.CreateFCmpOEQ(Op1, NegInf);
     Value *Sel = B.CreateSelect(FCmp, Inf, FAbs);

llvm/trunk/test/Transforms/InstCombine/pow-sqrt.ll

+; RUN: opt < %s -instcombine -S | FileCheck %s
+
+define double @mypow(double %x) #0 {
+entry:
+  %pow = call double @llvm.pow.f64(double %x, double 5.000000e-01)
+  ret double %pow
+}
+
+; CHECK-LABEL: define double @mypow(
+; CHECK:   %sqrt = call double @sqrt(double %x) #1
+; CHECK:   ret double %sqrt
+; CHECK: }
+
+declare double @llvm.pow.f64(double, double)
+attributes #0 = { "unsafe-fp-math"="true" }