Adds missing TLI check for library simplification of

* pow(x, 0.5) -> fabs(sqrt(x))
* pow(2.0, x) -> exp2(x)

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@188656 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Transforms/Utils/SimplifyLibCalls.cpp b/lib/Transforms/Utils/SimplifyLibCalls.cpp
index 93720be683e5af32f90e44cd1bc40b03258a02a6..ff0d5d9e8dd52688fcf1ce4e74b9c04f3ec1f9be 100644 (file)
--- a/lib/Transforms/Utils/SimplifyLibCalls.cpp
+++ b/lib/Transforms/Utils/SimplifyLibCalls.cpp
@@ -1133,9 +1133,11 @@ struct PowOpt : public UnsafeFPLibCallOptimization {
 
     Value *Op1 = CI->getArgOperand(0), *Op2 = CI->getArgOperand(1);
     if (ConstantFP *Op1C = dyn_cast<ConstantFP>(Op1)) {
 
     Value *Op1 = CI->getArgOperand(0), *Op2 = CI->getArgOperand(1);
     if (ConstantFP *Op1C = dyn_cast<ConstantFP>(Op1)) {

-      if (Op1C->isExactlyValue(1.0))  // pow(1.0, x) -> 1.0
+      // pow(1.0, x) -> 1.0
+      if (Op1C->isExactlyValue(1.0))

         return Op1C;
         return Op1C;

-      if (Op1C->isExactlyValue(2.0))  // pow(2.0, x) -> exp2(x)
+      // pow(2.0, x) -> exp2(x)
+      if (Op1C->isExactlyValue(2.0) && TLI->has(LibFunc::exp2))

         return EmitUnaryFloatFnCall(Op2, "exp2", B, Callee->getAttributes());
     }
 
         return EmitUnaryFloatFnCall(Op2, "exp2", B, Callee->getAttributes());
     }
 


@@ -1145,7 +1147,8 @@ struct PowOpt : public UnsafeFPLibCallOptimization {
     if (Op2C->getValueAPF().isZero())  // pow(x, 0.0) -> 1.0
       return ConstantFP::get(CI->getType(), 1.0);
 
     if (Op2C->getValueAPF().isZero())  // pow(x, 0.0) -> 1.0
       return ConstantFP::get(CI->getType(), 1.0);
 

-    if (Op2C->isExactlyValue(0.5)) {
+    if (Op2C->isExactlyValue(0.5) &&
+        TLI->has(LibFunc::sqrt) && TLI->has(LibFunc::fabs)) {

       // 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.
       // 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.

diff --git a/test/Transforms/InstCombine/pow-1.ll b/test/Transforms/InstCombine/pow-1.ll
index 0fdafebcd4a7c6042d5ede5e5fa55b9c2915e61b..9f1d073fe760987a9d3ca4117e12cb8686b48f01 100644 (file)
--- a/test/Transforms/InstCombine/pow-1.ll
+++ b/test/Transforms/InstCombine/pow-1.ll
@@ -151,4 +151,17 @@ define double @test_simplify16(double %x) {
 ; CHECK-NEXT: ret double [[RECIPROCAL]]
 }
 
 ; CHECK-NEXT: ret double [[RECIPROCAL]]
 }
 

+declare double @llvm.pow.f64(double %Val, double %Power)
+define double @test_simplify17(double %x) {
+; CHECK-LABEL: @test_simplify17(
+  %retval = call double @llvm.pow.f64(double %x, double 0.5)
+; CHECK-NEXT: [[SQRT:%[a-z0-9]+]] = call double @sqrt(double %x) [[NUW_RO]]
+; CHECK-NEXT: [[FABS:%[a-z0-9]+]] = call double @fabs(double [[SQRT]]) [[NUW_RO]]
+; CHECK-NEXT: [[FCMP:%[a-z0-9]+]] = fcmp oeq double %x, 0xFFF0000000000000
+; CHECK-NEXT: [[SELECT:%[a-z0-9]+]] = select i1 [[FCMP]], double 0x7FF0000000000000, double [[FABS]]
+  ret double %retval
+; CHECK-NEXT: ret double [[SELECT]]
+}
+

 ; CHECK: attributes [[NUW_RO]] = { nounwind readonly }
 ; CHECK: attributes [[NUW_RO]] = { nounwind readonly }

+

diff --git a/test/Transforms/InstCombine/pow-3.ll b/test/Transforms/InstCombine/pow-3.ll
new file mode 100644 (file)
index 0000000..1c5cf91
--- /dev/null
+++ b/test/Transforms/InstCombine/pow-3.ll
@@ -0,0 +1,12 @@
+; Test that the pow won't get simplified to sqrt(fabs) when they are not available.
+;
+; RUN: opt < %s -disable-simplify-libcalls -instcombine -S | FileCheck %s
+
+declare double @llvm.pow.f64(double %Val, double %Power)
+
+define double @test_simplify_unavailable(double %x) {
+; CHECK-LABEL: @test_simplify_unavailable(
+  %retval = call double @llvm.pow.f64(double %x, double 0.5)
+; CHECK-NEXT: call double @llvm.pow.f64(double %x, double 5.000000e-01)
+  ret double %retval
+}