diff --git a/lib/Target/X86/X86CallingConv.td b/lib/Target/X86/X86CallingConv.td
index 549eb9b9a20..e720b0b3b22 100644
--- a/lib/Target/X86/X86CallingConv.td
+++ b/lib/Target/X86/X86CallingConv.td
@@ -352,6 +352,12 @@ def CC_X86_32_FastCC : CallingConv<[
   // The first 2 integer arguments are passed in ECX/EDX
   CCIfType<[i32], CCAssignToReg<[ECX, EDX]>>,
 
+  // The first 3 float or double arguments, if the call is not a vararg
+  // call and if SSE2 is available, are passed in SSE registers.
+  CCIfNotVarArg<CCIfType<[f32,f64],
+                CCIfSubtarget<"hasSSE2()",
+                CCAssignToReg<[XMM0,XMM1,XMM2]>>>>,
+
   // Doubles get 8-byte slots that are 8-byte aligned.
   CCIfType<[f64], CCAssignToStack<8, 8>>,
 
diff --git a/test/CodeGen/X86/constant-pool-remat-0.ll b/test/CodeGen/X86/constant-pool-remat-0.ll
index 144d442d50d..40caaa6b936 100644
--- a/test/CodeGen/X86/constant-pool-remat-0.ll
+++ b/test/CodeGen/X86/constant-pool-remat-0.ll
@@ -1,13 +1,13 @@
 ; RUN: llvm-as < %s | llc -march=x86-64 | grep LCPI | count 3
 ; RUN: llvm-as < %s | llc -march=x86-64 -stats  -info-output-file - | grep asm-printer | grep 6
 ; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 | grep LCPI | count 3
-; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 -stats  -info-output-file - | grep asm-printer | grep 8
+; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse2 -stats  -info-output-file - | grep asm-printer | grep 12
 
-declare fastcc float @qux(float %y)
+declare float @qux(float %y)
 
-define fastcc float @array(float %a) nounwind {
+define float @array(float %a) nounwind {
   %n = mul float %a, 9.0
-  %m = call fastcc float @qux(float %n)
+  %m = call float @qux(float %n)
   %o = mul float %m, 9.0
   ret float %o
 }
diff --git a/test/CodeGen/X86/fastcc-2.ll b/test/CodeGen/X86/fastcc-2.ll
new file mode 100644
index 00000000000..40c753ee304
--- /dev/null
+++ b/test/CodeGen/X86/fastcc-2.ll
@@ -0,0 +1,10 @@
+; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin -mattr=+sse2 | grep movsd
+; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin -mattr=+sse2 | grep mov | count 1
+
+define i32 @foo() nounwind {
+entry:
+	tail call fastcc void @bar( double 1.000000e+00 ) nounwind
+	ret i32 0
+}
+
+declare fastcc void @bar(double)
diff --git a/test/CodeGen/X86/fastcc.ll b/test/CodeGen/X86/fastcc.ll
index 13068bada27..07af8051176 100644
--- a/test/CodeGen/X86/fastcc.ll
+++ b/test/CodeGen/X86/fastcc.ll
@@ -1,5 +1,5 @@
 ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin | grep mov | grep ecx | grep 0
-; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin | grep mov | grep xmm0 | grep 16
+; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin | grep mov | grep xmm0 | grep 8
 
 @d = external global double		; <double*> [#uses=1]
 @c = external global double		; <double*> [#uses=1]