Make X86::FsFLD0SS / FsFLD0SD real pseudo-instructions.

Like V_SET0, these instructions are expanded by ExpandPostRA to xorps /
vxorps so they can participate in execution domain swizzling.

This also makes the AVX variants redundant.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@145440 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/test/CodeGen/X86/bc-extract.ll b/test/CodeGen/X86/bc-extract.ll
index ac972a8..ceabcb7 100644
--- a/test/CodeGen/X86/bc-extract.ll
+++ b/test/CodeGen/X86/bc-extract.ll
@@ -11,7 +11,7 @@
 
 define float @extractFloat2() nounwind {
 entry:
-  ; CHECK: pxor	%xmm0, %xmm0
+  ; CHECK: xorps	%xmm0, %xmm0
   %tmp4 = bitcast <1 x double> <double 0x000000003F800000> to <2 x float>
   %tmp5 = extractelement <2 x float> %tmp4, i32 1
   ret float %tmp5