diff --git a/clang/lib/AST/Interp/Floating.h b/clang/lib/AST/Interp/Floating.h
--- a/clang/lib/AST/Interp/Floating.h
+++ b/clang/lib/AST/Interp/Floating.h
@@ -119,6 +119,13 @@
     return Status;
   }
 
+  static Floating abs(const Floating &F) {
+    APFloat V = F.F;
+    if (V.isNegative())
+      V.changeSign();
+    return Floating(V);
+  }
+
   // -------
 
   static APFloat::opStatus add(const Floating &A, const Floating &B,
diff --git a/clang/lib/AST/Interp/InterpBuiltin.cpp b/clang/lib/AST/Interp/InterpBuiltin.cpp
--- a/clang/lib/AST/Interp/InterpBuiltin.cpp
+++ b/clang/lib/AST/Interp/InterpBuiltin.cpp
@@ -266,6 +266,20 @@
   return true;
 }
 
+// The C standard says "fabs raises no floating-point exceptions,
+// even if x is a signaling NaN. The returned value is independent of
+// the current rounding direction mode."  Therefore constant folding can
+// proceed without regard to the floating point settings.
+// Reference, WG14 N2478 F.10.4.3
+static bool interp__builtin_fabs(InterpState &S, CodePtr OpPC,
+                                 const InterpFrame *Frame,
+                                 const Function *Func) {
+  const Floating &Val = getParam<Floating>(Frame, 0);
+
+  S.Stk.push<Floating>(Floating::abs(Val));
+  return true;
+}
+
 bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const Function *F) {
   InterpFrame *Frame = S.Current;
   APValue Dummy;
@@ -359,6 +373,14 @@
       return Ret<PT_Sint32>(S, OpPC, Dummy);
     break;
 
+  case Builtin::BI__builtin_fabs:
+  case Builtin::BI__builtin_fabsf:
+  case Builtin::BI__builtin_fabsl:
+  case Builtin::BI__builtin_fabsf128:
+    if (interp__builtin_fabs(S, OpPC, Frame, F))
+      return Ret<PT_Float>(S, OpPC, Dummy);
+    break;
+
   default:
     return false;
   }
diff --git a/clang/test/AST/Interp/builtin-functions.cpp b/clang/test/AST/Interp/builtin-functions.cpp
--- a/clang/test/AST/Interp/builtin-functions.cpp
+++ b/clang/test/AST/Interp/builtin-functions.cpp
@@ -140,3 +140,7 @@
   char classify_neg_zero[__builtin_fpclassify(-1, -1, -1, -1, +1, -0.0)];
   char classify_subnorm [__builtin_fpclassify(-1, -1, -1, +1, -1, 1.0e-38f)];
 }
+
+namespace fabs {
+  static_assert(__builtin_fabs(-14.0) == 14.0, "");
+}