diff options
-rw-r--r-- | lib/Target/X86/X86ISelLowering.cpp | 22 | ||||
-rw-r--r-- | test/CodeGen/X86/2012-04-26-sdglue.ll | 3 |
2 files changed, 14 insertions, 11 deletions
diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp index 8404c7a3d5..32fb5ceebe 100644 --- a/lib/Target/X86/X86ISelLowering.cpp +++ b/lib/Target/X86/X86ISelLowering.cpp @@ -13050,16 +13050,18 @@ static SDValue PerformShuffleCombine256(SDNode *N, SelectionDAG &DAG, // If V1 is coming from a vector load then just fold to a VZEXT_LOAD. if (LoadSDNode *Ld = dyn_cast<LoadSDNode>(V1.getOperand(0))) { - SDVTList Tys = DAG.getVTList(MVT::v4i64, MVT::Other); - SDValue Ops[] = { Ld->getChain(), Ld->getBasePtr() }; - SDValue ResNode = - DAG.getMemIntrinsicNode(X86ISD::VZEXT_LOAD, dl, Tys, Ops, 2, - Ld->getMemoryVT(), - Ld->getPointerInfo(), - Ld->getAlignment(), - false/*isVolatile*/, true/*ReadMem*/, - false/*WriteMem*/); - return DAG.getNode(ISD::BITCAST, dl, VT, ResNode); + if (Ld->hasNUsesOfValue(1, 0)) { + SDVTList Tys = DAG.getVTList(MVT::v4i64, MVT::Other); + SDValue Ops[] = { Ld->getChain(), Ld->getBasePtr() }; + SDValue ResNode = + DAG.getMemIntrinsicNode(X86ISD::VZEXT_LOAD, dl, Tys, Ops, 2, + Ld->getMemoryVT(), + Ld->getPointerInfo(), + Ld->getAlignment(), + false/*isVolatile*/, true/*ReadMem*/, + false/*WriteMem*/); + return DAG.getNode(ISD::BITCAST, dl, VT, ResNode); + } } // Emit a zeroed vector and insert the desired subvector on its diff --git a/test/CodeGen/X86/2012-04-26-sdglue.ll b/test/CodeGen/X86/2012-04-26-sdglue.ll index 9543587747..9a66b670c7 100644 --- a/test/CodeGen/X86/2012-04-26-sdglue.ll +++ b/test/CodeGen/X86/2012-04-26-sdglue.ll @@ -5,7 +5,8 @@ ; It's hard to test for the ISEL condition because CodeGen optimizes ; away the bugpointed code. Just ensure the basics are still there. ;CHECK: func: -;CHECK: vmovups +;CHECK: vpxor +;CHECK: vinsertf128 ;CHECK: vpshufd ;CHECK: vpshufd ;CHECK: vmulps |