diff --git a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index af614937de3..647ff6b6b20 100644
--- a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -5770,6 +5770,10 @@ SDValue DAGCombiner::visitFMA(SDNode *N) {
   if (N1CFP && N1CFP->isExactlyValue(1.0))
     return DAG.getNode(ISD::FADD, N->getDebugLoc(), VT, N0, N2);
 
+  // Canonicalize (fma c, x, y) -> (fma x, c, y)
+  if (!N0CFP && N1CFP)
+    return DAG.getNode(ISD::FMA, N->getDebugLoc(), VT, N1, N0, N2);
+
   return SDValue();
 }
 
diff --git a/test/CodeGen/ARM/fusedMAC.ll b/test/CodeGen/ARM/fusedMAC.ll
index fd7fdd5405a..8ebca02df99 100644
--- a/test/CodeGen/ARM/fusedMAC.ll
+++ b/test/CodeGen/ARM/fusedMAC.ll
@@ -189,6 +189,15 @@ define float @test_fma_const_fold(float %a, float %b) nounwind {
   ret float %ret
 }
 
+define float @test_fma_canonicalize(float %a, float %b) nounwind {
+; CHECK: test_fma_canonicalize
+; CHECK: vmov.f32 s0
+; CHECK: vfma.f32 s2, s0, s1
+  %ret = call float @llvm.fma.f32(float 2.0, float %a, float %b)
+  ret float %ret
+}
+
+
 declare float @llvm.fma.f32(float, float, float) nounwind readnone
 declare double @llvm.fma.f64(double, double, double) nounwind readnone
 declare <2 x float> @llvm.fma.v2f32(<2 x float>, <2 x float>, <2 x float>) nounwind readnone