diff --git a/lib/Target/X86/X86InstrInfo.cpp b/lib/Target/X86/X86InstrInfo.cpp
index 53d5ee30224..9f5f66e37c8 100644
--- a/lib/Target/X86/X86InstrInfo.cpp
+++ b/lib/Target/X86/X86InstrInfo.cpp
@@ -539,6 +539,8 @@ X86InstrInfo::X86InstrInfo(X86TargetMachine &tm)
     { X86::VSQRTPSr_Int,    X86::VSQRTPSm_Int,        TB_ALIGN_16 },
     { X86::VUCOMISDrr,      X86::VUCOMISDrm,          0 },
     { X86::VUCOMISSrr,      X86::VUCOMISSrm,          0 },
+    { X86::VBROADCASTSSrr,  X86::VBROADCASTSSrm,      TB_NO_REVERSE },
+
     // AVX 256-bit foldable instructions
     { X86::VMOVAPDYrr,      X86::VMOVAPDYrm,          TB_ALIGN_32 },
     { X86::VMOVAPSYrr,      X86::VMOVAPSYrm,          TB_ALIGN_32 },
@@ -547,6 +549,7 @@ X86InstrInfo::X86InstrInfo(X86TargetMachine &tm)
     { X86::VMOVUPSYrr,      X86::VMOVUPSYrm,          0 },
     { X86::VPERMILPDYri,    X86::VPERMILPDYmi,        TB_ALIGN_32 },
     { X86::VPERMILPSYri,    X86::VPERMILPSYmi,        TB_ALIGN_32 },
+
     // AVX2 foldable instructions
     { X86::VPABSBrr256,     X86::VPABSBrm256,         TB_ALIGN_32 },
     { X86::VPABSDrr256,     X86::VPABSDrm256,         TB_ALIGN_32 },
@@ -562,6 +565,8 @@ X86InstrInfo::X86InstrInfo(X86TargetMachine &tm)
     { X86::VSQRTPDYr_Int,   X86::VSQRTPDYm_Int,       TB_ALIGN_32 },
     { X86::VSQRTPSYr,       X86::VSQRTPSYm,           TB_ALIGN_32 },
     { X86::VSQRTPSYr_Int,   X86::VSQRTPSYm_Int,       TB_ALIGN_32 },
+    { X86::VBROADCASTSSYrr, X86::VBROADCASTSSYrm,     TB_NO_REVERSE },
+    { X86::VBROADCASTSDYrr, X86::VBROADCASTSDYrm,     TB_NO_REVERSE },
   };
 
   for (unsigned i = 0, e = array_lengthof(OpTbl1); i != e; ++i) {
diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 5a224aff9a1..e4caace00ca 100644
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -7274,7 +7274,7 @@ let ExeDomain = SSEPackedSingle in {
                                       int_x86_avx_vbroadcast_ss_256>;
 }
 let ExeDomain = SSEPackedDouble in
-def VBROADCASTSDrm  : avx_broadcast<0x19, "vbroadcastsd", VR256, f64mem,
+def VBROADCASTSDYrm  : avx_broadcast<0x19, "vbroadcastsd", VR256, f64mem,
                                     int_x86_avx_vbroadcast_sd_256>;
 def VBROADCASTF128 : avx_broadcast<0x1A, "vbroadcastf128", VR256, f128mem,
                                    int_x86_avx_vbroadcastf128_pd_256>;
@@ -7753,11 +7753,11 @@ let Predicates = [HasAVX] in {
 def : Pat<(v8i32 (X86VBroadcast (loadi32 addr:$src))),
           (VBROADCASTSSYrm addr:$src)>;
 def : Pat<(v4i64 (X86VBroadcast (loadi64 addr:$src))),
-          (VBROADCASTSDrm addr:$src)>;
+          (VBROADCASTSDYrm addr:$src)>;
 def : Pat<(v8f32 (X86VBroadcast (loadf32 addr:$src))),
           (VBROADCASTSSYrm addr:$src)>;
 def : Pat<(v4f64 (X86VBroadcast (loadf64 addr:$src))),
-          (VBROADCASTSDrm addr:$src)>;
+          (VBROADCASTSDYrm addr:$src)>;
 def : Pat<(v4f32 (X86VBroadcast (loadf32 addr:$src))),
           (VBROADCASTSSrm addr:$src)>;
 def : Pat<(v4i32 (X86VBroadcast (loadi32 addr:$src))),
diff --git a/test/CodeGen/X86/2012-07-15-broadcastfold.ll b/test/CodeGen/X86/2012-07-15-broadcastfold.ll
new file mode 100644
index 00000000000..6a7bb63b88e
--- /dev/null
+++ b/test/CodeGen/X86/2012-07-15-broadcastfold.ll
@@ -0,0 +1,21 @@
+; RUN: llc < %s -march=x86 -mcpu=corei7 -mattr=+avx2 | FileCheck %s
+
+declare x86_fastcallcc i64 @barrier()
+
+;CHECK: bcast_fold
+;CHECK: vbroadcastss    -24(%ebp), %ymm0 # 16-byte Folded Reload
+;CHECK: ret
+define <8 x float> @bcast_fold( float* %A) {
+BB:
+  %A0 = load float* %A
+  %tt3 = call x86_fastcallcc i64 @barrier()
+  br i1 undef, label %work, label %exit
+
+work:
+  %A1 = insertelement <8 x float> undef, float %A0, i32 0
+  %A2 = shufflevector <8 x float> %A1, <8 x float> undef, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
+  ret <8 x float> %A2
+
+exit:
+  ret <8 x float> undef
+}