aboutsummaryrefslogtreecommitdiff
path: root/lib/Target/ARM/ARMISelDAGToDAG.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'lib/Target/ARM/ARMISelDAGToDAG.cpp')
-rw-r--r--lib/Target/ARM/ARMISelDAGToDAG.cpp371
1 files changed, 139 insertions, 232 deletions
diff --git a/lib/Target/ARM/ARMISelDAGToDAG.cpp b/lib/Target/ARM/ARMISelDAGToDAG.cpp
index c84d3ff81324..51a30c158dd1 100644
--- a/lib/Target/ARM/ARMISelDAGToDAG.cpp
+++ b/lib/Target/ARM/ARMISelDAGToDAG.cpp
@@ -36,6 +36,11 @@
using namespace llvm;
+static cl::opt<bool>
+DisableShifterOp("disable-shifter-op", cl::Hidden,
+ cl::desc("Disable isel of shifter-op"),
+ cl::init(false));
+
//===--------------------------------------------------------------------===//
/// ARMDAGToDAGISel - ARM specific code to select ARM machine
/// instructions for SelectionDAG operations.
@@ -113,6 +118,16 @@ public:
bool SelectT2AddrModeSoReg(SDNode *Op, SDValue N, SDValue &Base,
SDValue &OffReg, SDValue &ShImm);
+ inline bool Pred_so_imm(SDNode *inN) const {
+ ConstantSDNode *N = cast<ConstantSDNode>(inN);
+ return ARM_AM::getSOImmVal(N->getZExtValue()) != -1;
+ }
+
+ inline bool Pred_t2_so_imm(SDNode *inN) const {
+ ConstantSDNode *N = cast<ConstantSDNode>(inN);
+ return ARM_AM::getT2SOImmVal(N->getZExtValue()) != -1;
+ }
+
// Include the pieces autogenerated from the target description.
#include "ARMGenDAGISel.inc"
@@ -220,6 +235,9 @@ bool ARMDAGToDAGISel::SelectShifterOperandReg(SDNode *Op,
SDValue &BaseReg,
SDValue &ShReg,
SDValue &Opc) {
+ if (DisableShifterOp)
+ return false;
+
ARM_AM::ShiftOpc ShOpcVal = ARM_AM::getShiftOpcForNode(N);
// Don't match base register only case. That is matched to a separate
@@ -463,7 +481,7 @@ bool ARMDAGToDAGISel::SelectAddrMode3Offset(SDNode *Op, SDValue N,
bool ARMDAGToDAGISel::SelectAddrMode4(SDNode *Op, SDValue N,
SDValue &Addr, SDValue &Mode) {
Addr = N;
- Mode = CurDAG->getTargetConstant(0, MVT::i32);
+ Mode = CurDAG->getTargetConstant(ARM_AM::getAM4ModeImm(ARM_AM::ia), MVT::i32);
return true;
}
@@ -666,6 +684,9 @@ bool ARMDAGToDAGISel::SelectThumbAddrModeSP(SDNode *Op, SDValue N,
bool ARMDAGToDAGISel::SelectT2ShifterOperandReg(SDNode *Op, SDValue N,
SDValue &BaseReg,
SDValue &Opc) {
+ if (DisableShifterOp)
+ return false;
+
ARM_AM::ShiftOpc ShOpcVal = ARM_AM::getShiftOpcForNode(N);
// Don't match base register only case. That is matched to a separate
@@ -1090,110 +1111,79 @@ SDNode *ARMDAGToDAGISel::SelectVLD(SDNode *N, unsigned NumVecs,
break;
}
+ EVT ResTy;
+ if (NumVecs == 1)
+ ResTy = VT;
+ else {
+ unsigned ResTyElts = (NumVecs == 3) ? 4 : NumVecs;
+ if (!is64BitVector)
+ ResTyElts *= 2;
+ ResTy = EVT::getVectorVT(*CurDAG->getContext(), MVT::i64, ResTyElts);
+ }
+
SDValue Pred = getAL(CurDAG);
SDValue Reg0 = CurDAG->getRegister(0, MVT::i32);
+ SDValue SuperReg;
if (is64BitVector) {
unsigned Opc = DOpcodes[OpcodeIndex];
const SDValue Ops[] = { MemAddr, Align, Pred, Reg0, Chain };
- std::vector<EVT> ResTys(NumVecs, VT);
- ResTys.push_back(MVT::Other);
- SDNode *VLd = CurDAG->getMachineNode(Opc, dl, ResTys, Ops, 5);
- if (NumVecs < 2)
+ SDNode *VLd = CurDAG->getMachineNode(Opc, dl, ResTy, MVT::Other, Ops, 5);
+ if (NumVecs == 1)
return VLd;
- SDValue RegSeq;
- SDValue V0 = SDValue(VLd, 0);
- SDValue V1 = SDValue(VLd, 1);
-
- // Form a REG_SEQUENCE to force register allocation.
- if (NumVecs == 2)
- RegSeq = SDValue(PairDRegs(MVT::v2i64, V0, V1), 0);
- else {
- SDValue V2 = SDValue(VLd, 2);
- // If it's a vld3, form a quad D-register but discard the last part.
- SDValue V3 = (NumVecs == 3)
- ? SDValue(CurDAG->getMachineNode(TargetOpcode::IMPLICIT_DEF,dl,VT), 0)
- : SDValue(VLd, 3);
- RegSeq = SDValue(QuadDRegs(MVT::v4i64, V0, V1, V2, V3), 0);
- }
-
+ SuperReg = SDValue(VLd, 0);
assert(ARM::dsub_7 == ARM::dsub_0+7 && "Unexpected subreg numbering");
for (unsigned Vec = 0; Vec < NumVecs; ++Vec) {
SDValue D = CurDAG->getTargetExtractSubreg(ARM::dsub_0+Vec,
- dl, VT, RegSeq);
+ dl, VT, SuperReg);
ReplaceUses(SDValue(N, Vec), D);
}
- ReplaceUses(SDValue(N, NumVecs), SDValue(VLd, NumVecs));
+ ReplaceUses(SDValue(N, NumVecs), SDValue(VLd, 1));
return NULL;
}
- EVT RegVT = GetNEONSubregVT(VT);
if (NumVecs <= 2) {
// Quad registers are directly supported for VLD1 and VLD2,
// loading pairs of D regs.
unsigned Opc = QOpcodes0[OpcodeIndex];
const SDValue Ops[] = { MemAddr, Align, Pred, Reg0, Chain };
- std::vector<EVT> ResTys(2 * NumVecs, RegVT);
- ResTys.push_back(MVT::Other);
- SDNode *VLd = CurDAG->getMachineNode(Opc, dl, ResTys, Ops, 5);
- Chain = SDValue(VLd, 2 * NumVecs);
+ SDNode *VLd = CurDAG->getMachineNode(Opc, dl, ResTy, MVT::Other, Ops, 5);
+ if (NumVecs == 1)
+ return VLd;
+
+ SuperReg = SDValue(VLd, 0);
+ Chain = SDValue(VLd, 1);
- // Combine the even and odd subregs to produce the result.
- if (NumVecs == 1) {
- SDNode *Q = PairDRegs(VT, SDValue(VLd, 0), SDValue(VLd, 1));
- ReplaceUses(SDValue(N, 0), SDValue(Q, 0));
- } else {
- SDValue QQ = SDValue(QuadDRegs(MVT::v4i64,
- SDValue(VLd, 0), SDValue(VLd, 1),
- SDValue(VLd, 2), SDValue(VLd, 3)), 0);
- SDValue Q0 = CurDAG->getTargetExtractSubreg(ARM::qsub_0, dl, VT, QQ);
- SDValue Q1 = CurDAG->getTargetExtractSubreg(ARM::qsub_1, dl, VT, QQ);
- ReplaceUses(SDValue(N, 0), Q0);
- ReplaceUses(SDValue(N, 1), Q1);
- }
} else {
// Otherwise, quad registers are loaded with two separate instructions,
// where one loads the even registers and the other loads the odd registers.
-
- std::vector<EVT> ResTys(NumVecs, RegVT);
- ResTys.push_back(MemAddr.getValueType());
- ResTys.push_back(MVT::Other);
+ EVT AddrTy = MemAddr.getValueType();
// Load the even subregs.
unsigned Opc = QOpcodes0[OpcodeIndex];
- const SDValue OpsA[] = { MemAddr, Align, Reg0, Pred, Reg0, Chain };
- SDNode *VLdA = CurDAG->getMachineNode(Opc, dl, ResTys, OpsA, 6);
- Chain = SDValue(VLdA, NumVecs+1);
+ SDValue ImplDef =
+ SDValue(CurDAG->getMachineNode(TargetOpcode::IMPLICIT_DEF, dl, ResTy), 0);
+ const SDValue OpsA[] = { MemAddr, Align, Reg0, ImplDef, Pred, Reg0, Chain };
+ SDNode *VLdA =
+ CurDAG->getMachineNode(Opc, dl, ResTy, AddrTy, MVT::Other, OpsA, 7);
+ Chain = SDValue(VLdA, 2);
// Load the odd subregs.
Opc = QOpcodes1[OpcodeIndex];
- const SDValue OpsB[] = { SDValue(VLdA, NumVecs),
- Align, Reg0, Pred, Reg0, Chain };
- SDNode *VLdB = CurDAG->getMachineNode(Opc, dl, ResTys, OpsB, 6);
- Chain = SDValue(VLdB, NumVecs+1);
-
- SDValue V0 = SDValue(VLdA, 0);
- SDValue V1 = SDValue(VLdB, 0);
- SDValue V2 = SDValue(VLdA, 1);
- SDValue V3 = SDValue(VLdB, 1);
- SDValue V4 = SDValue(VLdA, 2);
- SDValue V5 = SDValue(VLdB, 2);
- SDValue V6 = (NumVecs == 3)
- ? SDValue(CurDAG->getMachineNode(TargetOpcode::IMPLICIT_DEF,dl,RegVT), 0)
- : SDValue(VLdA, 3);
- SDValue V7 = (NumVecs == 3)
- ? SDValue(CurDAG->getMachineNode(TargetOpcode::IMPLICIT_DEF,dl,RegVT), 0)
- : SDValue(VLdB, 3);
- SDValue RegSeq = SDValue(OctoDRegs(MVT::v8i64, V0, V1, V2, V3,
- V4, V5, V6, V7), 0);
-
- // Extract out the 3 / 4 Q registers.
- assert(ARM::qsub_3 == ARM::qsub_0+3 && "Unexpected subreg numbering");
- for (unsigned Vec = 0; Vec < NumVecs; ++Vec) {
- SDValue Q = CurDAG->getTargetExtractSubreg(ARM::qsub_0+Vec,
- dl, VT, RegSeq);
- ReplaceUses(SDValue(N, Vec), Q);
- }
+ const SDValue OpsB[] = { SDValue(VLdA, 1), Align, Reg0, SDValue(VLdA, 0),
+ Pred, Reg0, Chain };
+ SDNode *VLdB =
+ CurDAG->getMachineNode(Opc, dl, ResTy, AddrTy, MVT::Other, OpsB, 7);
+ SuperReg = SDValue(VLdB, 0);
+ Chain = SDValue(VLdB, 2);
+ }
+
+ // Extract out the Q registers.
+ assert(ARM::qsub_3 == ARM::qsub_0+3 && "Unexpected subreg numbering");
+ for (unsigned Vec = 0; Vec < NumVecs; ++Vec) {
+ SDValue Q = CurDAG->getTargetExtractSubreg(ARM::qsub_0+Vec,
+ dl, VT, SuperReg);
+ ReplaceUses(SDValue(N, Vec), Q);
}
ReplaceUses(SDValue(N, NumVecs), Chain);
return NULL;
@@ -1235,12 +1225,14 @@ SDNode *ARMDAGToDAGISel::SelectVST(SDNode *N, unsigned NumVecs,
SDValue Pred = getAL(CurDAG);
SDValue Reg0 = CurDAG->getRegister(0, MVT::i32);
- SmallVector<SDValue, 10> Ops;
+ SmallVector<SDValue, 7> Ops;
Ops.push_back(MemAddr);
Ops.push_back(Align);
if (is64BitVector) {
- if (NumVecs >= 2) {
+ if (NumVecs == 1) {
+ Ops.push_back(N->getOperand(3));
+ } else {
SDValue RegSeq;
SDValue V0 = N->getOperand(0+3);
SDValue V1 = N->getOperand(1+3);
@@ -1257,111 +1249,61 @@ SDNode *ARMDAGToDAGISel::SelectVST(SDNode *N, unsigned NumVecs,
: N->getOperand(3+3);
RegSeq = SDValue(QuadDRegs(MVT::v4i64, V0, V1, V2, V3), 0);
}
-
- // Now extract the D registers back out.
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_0, dl, VT,
- RegSeq));
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_1, dl, VT,
- RegSeq));
- if (NumVecs > 2)
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_2, dl, VT,
- RegSeq));
- if (NumVecs > 3)
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_3, dl, VT,
- RegSeq));
- } else {
- for (unsigned Vec = 0; Vec < NumVecs; ++Vec)
- Ops.push_back(N->getOperand(Vec+3));
+ Ops.push_back(RegSeq);
}
Ops.push_back(Pred);
Ops.push_back(Reg0); // predicate register
Ops.push_back(Chain);
unsigned Opc = DOpcodes[OpcodeIndex];
- return CurDAG->getMachineNode(Opc, dl, MVT::Other, Ops.data(), NumVecs+5);
+ return CurDAG->getMachineNode(Opc, dl, MVT::Other, Ops.data(), 6);
}
- EVT RegVT = GetNEONSubregVT(VT);
if (NumVecs <= 2) {
- // Quad registers are directly supported for VST1 and VST2,
- // storing pairs of D regs.
+ // Quad registers are directly supported for VST1 and VST2.
unsigned Opc = QOpcodes0[OpcodeIndex];
- if (NumVecs == 2) {
- // First extract the pair of Q registers.
+ if (NumVecs == 1) {
+ Ops.push_back(N->getOperand(3));
+ } else {
+ // Form a QQ register.
SDValue Q0 = N->getOperand(3);
SDValue Q1 = N->getOperand(4);
-
- // Form a QQ register.
- SDValue QQ = SDValue(PairQRegs(MVT::v4i64, Q0, Q1), 0);
-
- // Now extract the D registers back out.
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_0, dl, RegVT,
- QQ));
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_1, dl, RegVT,
- QQ));
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_2, dl, RegVT,
- QQ));
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_3, dl, RegVT,
- QQ));
- Ops.push_back(Pred);
- Ops.push_back(Reg0); // predicate register
- Ops.push_back(Chain);
- return CurDAG->getMachineNode(Opc, dl, MVT::Other, Ops.data(), 5 + 4);
- } else {
- for (unsigned Vec = 0; Vec < NumVecs; ++Vec) {
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_0, dl, RegVT,
- N->getOperand(Vec+3)));
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_1, dl, RegVT,
- N->getOperand(Vec+3)));
- }
- Ops.push_back(Pred);
- Ops.push_back(Reg0); // predicate register
- Ops.push_back(Chain);
- return CurDAG->getMachineNode(Opc, dl, MVT::Other, Ops.data(),
- 5 + 2 * NumVecs);
+ Ops.push_back(SDValue(PairQRegs(MVT::v4i64, Q0, Q1), 0));
}
+ Ops.push_back(Pred);
+ Ops.push_back(Reg0); // predicate register
+ Ops.push_back(Chain);
+ return CurDAG->getMachineNode(Opc, dl, MVT::Other, Ops.data(), 6);
}
// Otherwise, quad registers are stored with two separate instructions,
// where one stores the even registers and the other stores the odd registers.
// Form the QQQQ REG_SEQUENCE.
- SDValue V[8];
- for (unsigned Vec = 0, i = 0; Vec < NumVecs; ++Vec, i+=2) {
- V[i] = CurDAG->getTargetExtractSubreg(ARM::dsub_0, dl, RegVT,
- N->getOperand(Vec+3));
- V[i+1] = CurDAG->getTargetExtractSubreg(ARM::dsub_1, dl, RegVT,
- N->getOperand(Vec+3));
- }
- if (NumVecs == 3)
- V[6] = V[7] = SDValue(CurDAG->getMachineNode(TargetOpcode::IMPLICIT_DEF,
- dl, RegVT), 0);
-
- SDValue RegSeq = SDValue(OctoDRegs(MVT::v8i64, V[0], V[1], V[2], V[3],
- V[4], V[5], V[6], V[7]), 0);
+ SDValue V0 = N->getOperand(0+3);
+ SDValue V1 = N->getOperand(1+3);
+ SDValue V2 = N->getOperand(2+3);
+ SDValue V3 = (NumVecs == 3)
+ ? SDValue(CurDAG->getMachineNode(TargetOpcode::IMPLICIT_DEF, dl, VT), 0)
+ : N->getOperand(3+3);
+ SDValue RegSeq = SDValue(QuadQRegs(MVT::v8i64, V0, V1, V2, V3), 0);
// Store the even D registers.
- assert(ARM::dsub_7 == ARM::dsub_0+7 && "Unexpected subreg numbering");
Ops.push_back(Reg0); // post-access address offset
- for (unsigned Vec = 0; Vec < NumVecs; ++Vec)
- Ops.push_back(CurDAG->getTargetExtractSubreg(ARM::dsub_0+Vec*2, dl,
- RegVT, RegSeq));
+ Ops.push_back(RegSeq);
Ops.push_back(Pred);
Ops.push_back(Reg0); // predicate register
Ops.push_back(Chain);
unsigned Opc = QOpcodes0[OpcodeIndex];
SDNode *VStA = CurDAG->getMachineNode(Opc, dl, MemAddr.getValueType(),
- MVT::Other, Ops.data(), NumVecs+6);
+ MVT::Other, Ops.data(), 7);
Chain = SDValue(VStA, 1);
// Store the odd D registers.
Ops[0] = SDValue(VStA, 0); // MemAddr
- for (unsigned Vec = 0; Vec < NumVecs; ++Vec)
- Ops[Vec+3] = CurDAG->getTargetExtractSubreg(ARM::dsub_1+Vec*2, dl,
- RegVT, RegSeq);
- Ops[NumVecs+5] = Chain;
+ Ops[6] = Chain;
Opc = QOpcodes1[OpcodeIndex];
SDNode *VStB = CurDAG->getMachineNode(Opc, dl, MemAddr.getValueType(),
- MVT::Other, Ops.data(), NumVecs+6);
+ MVT::Other, Ops.data(), 7);
Chain = SDValue(VStB, 1);
ReplaceUses(SDValue(N, 0), Chain);
return NULL;
@@ -1675,7 +1617,7 @@ SelectT2CMOVSoImmOp(SDNode *N, SDValue FalseVal, SDValue TrueVal,
if (!T)
return 0;
- if (Predicate_t2_so_imm(TrueVal.getNode())) {
+ if (Pred_t2_so_imm(TrueVal.getNode())) {
SDValue True = CurDAG->getTargetConstant(T->getZExtValue(), MVT::i32);
SDValue CC = CurDAG->getTargetConstant(CCVal, MVT::i32);
SDValue Ops[] = { FalseVal, True, CC, CCR, InFlag };
@@ -1692,7 +1634,7 @@ SelectARMCMOVSoImmOp(SDNode *N, SDValue FalseVal, SDValue TrueVal,
if (!T)
return 0;
- if (Predicate_so_imm(TrueVal.getNode())) {
+ if (Pred_so_imm(TrueVal.getNode())) {
SDValue True = CurDAG->getTargetConstant(T->getZExtValue(), MVT::i32);
SDValue CC = CurDAG->getTargetConstant(CCVal, MVT::i32);
SDValue Ops[] = { FalseVal, True, CC, CCR, InFlag };
@@ -1740,7 +1682,7 @@ SDNode *ARMDAGToDAGISel::SelectCMOVOp(SDNode *N) {
}
// Pattern: (ARMcmov:i32 GPR:i32:$false,
- // (imm:i32)<<P:Predicate_so_imm>>:$true,
+ // (imm:i32)<<P:Pred_so_imm>>:$true,
// (imm:i32):$cc)
// Emits: (MOVCCi:i32 GPR:i32:$false,
// (so_imm:i32 (imm:i32):$true), (imm:i32):$cc)
@@ -2013,43 +1955,6 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
ResNode = SelectARMIndexedLoad(N);
if (ResNode)
return ResNode;
-
- // VLDMQ must be custom-selected for "v2f64 load" to set the AM5Opc value.
- if (Subtarget->hasVFP2() &&
- N->getValueType(0).getSimpleVT().SimpleTy == MVT::v2f64) {
- SDValue Chain = N->getOperand(0);
- SDValue AM5Opc =
- CurDAG->getTargetConstant(ARM_AM::getAM5Opc(ARM_AM::ia, 4), MVT::i32);
- SDValue Pred = getAL(CurDAG);
- SDValue PredReg = CurDAG->getRegister(0, MVT::i32);
- SDValue Ops[] = { N->getOperand(1), AM5Opc, Pred, PredReg, Chain };
- MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
- MemOp[0] = cast<MemSDNode>(N)->getMemOperand();
- SDNode *Ret = CurDAG->getMachineNode(ARM::VLDMQ, dl,
- MVT::v2f64, MVT::Other, Ops, 5);
- cast<MachineSDNode>(Ret)->setMemRefs(MemOp, MemOp + 1);
- return Ret;
- }
- // Other cases are autogenerated.
- break;
- }
- case ISD::STORE: {
- // VSTMQ must be custom-selected for "v2f64 store" to set the AM5Opc value.
- if (Subtarget->hasVFP2() &&
- N->getOperand(1).getValueType().getSimpleVT().SimpleTy == MVT::v2f64) {
- SDValue Chain = N->getOperand(0);
- SDValue AM5Opc =
- CurDAG->getTargetConstant(ARM_AM::getAM5Opc(ARM_AM::ia, 4), MVT::i32);
- SDValue Pred = getAL(CurDAG);
- SDValue PredReg = CurDAG->getRegister(0, MVT::i32);
- SDValue Ops[] = { N->getOperand(1), N->getOperand(2),
- AM5Opc, Pred, PredReg, Chain };
- MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
- MemOp[0] = cast<MemSDNode>(N)->getMemOperand();
- SDNode *Ret = CurDAG->getMachineNode(ARM::VSTMQ, dl, MVT::Other, Ops, 6);
- cast<MachineSDNode>(Ret)->setMemRefs(MemOp, MemOp + 1);
- return Ret;
- }
// Other cases are autogenerated.
break;
}
@@ -2206,39 +2111,40 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
case Intrinsic::arm_neon_vld1: {
unsigned DOpcodes[] = { ARM::VLD1d8, ARM::VLD1d16,
ARM::VLD1d32, ARM::VLD1d64 };
- unsigned QOpcodes[] = { ARM::VLD1q8, ARM::VLD1q16,
- ARM::VLD1q32, ARM::VLD1q64 };
+ unsigned QOpcodes[] = { ARM::VLD1q8Pseudo, ARM::VLD1q16Pseudo,
+ ARM::VLD1q32Pseudo, ARM::VLD1q64Pseudo };
return SelectVLD(N, 1, DOpcodes, QOpcodes, 0);
}
case Intrinsic::arm_neon_vld2: {
- unsigned DOpcodes[] = { ARM::VLD2d8, ARM::VLD2d16,
- ARM::VLD2d32, ARM::VLD1q64 };
- unsigned QOpcodes[] = { ARM::VLD2q8, ARM::VLD2q16, ARM::VLD2q32 };
+ unsigned DOpcodes[] = { ARM::VLD2d8Pseudo, ARM::VLD2d16Pseudo,
+ ARM::VLD2d32Pseudo, ARM::VLD1q64Pseudo };
+ unsigned QOpcodes[] = { ARM::VLD2q8Pseudo, ARM::VLD2q16Pseudo,
+ ARM::VLD2q32Pseudo };
return SelectVLD(N, 2, DOpcodes, QOpcodes, 0);
}
case Intrinsic::arm_neon_vld3: {
- unsigned DOpcodes[] = { ARM::VLD3d8, ARM::VLD3d16,
- ARM::VLD3d32, ARM::VLD1d64T };
- unsigned QOpcodes0[] = { ARM::VLD3q8_UPD,
- ARM::VLD3q16_UPD,
- ARM::VLD3q32_UPD };
- unsigned QOpcodes1[] = { ARM::VLD3q8odd_UPD,
- ARM::VLD3q16odd_UPD,
- ARM::VLD3q32odd_UPD };
+ unsigned DOpcodes[] = { ARM::VLD3d8Pseudo, ARM::VLD3d16Pseudo,
+ ARM::VLD3d32Pseudo, ARM::VLD1d64TPseudo };
+ unsigned QOpcodes0[] = { ARM::VLD3q8Pseudo_UPD,
+ ARM::VLD3q16Pseudo_UPD,
+ ARM::VLD3q32Pseudo_UPD };
+ unsigned QOpcodes1[] = { ARM::VLD3q8oddPseudo_UPD,
+ ARM::VLD3q16oddPseudo_UPD,
+ ARM::VLD3q32oddPseudo_UPD };
return SelectVLD(N, 3, DOpcodes, QOpcodes0, QOpcodes1);
}
case Intrinsic::arm_neon_vld4: {
- unsigned DOpcodes[] = { ARM::VLD4d8, ARM::VLD4d16,
- ARM::VLD4d32, ARM::VLD1d64Q };
- unsigned QOpcodes0[] = { ARM::VLD4q8_UPD,
- ARM::VLD4q16_UPD,
- ARM::VLD4q32_UPD };
- unsigned QOpcodes1[] = { ARM::VLD4q8odd_UPD,
- ARM::VLD4q16odd_UPD,
- ARM::VLD4q32odd_UPD };
+ unsigned DOpcodes[] = { ARM::VLD4d8Pseudo, ARM::VLD4d16Pseudo,
+ ARM::VLD4d32Pseudo, ARM::VLD1d64QPseudo };
+ unsigned QOpcodes0[] = { ARM::VLD4q8Pseudo_UPD,
+ ARM::VLD4q16Pseudo_UPD,
+ ARM::VLD4q32Pseudo_UPD };
+ unsigned QOpcodes1[] = { ARM::VLD4q8oddPseudo_UPD,
+ ARM::VLD4q16oddPseudo_UPD,
+ ARM::VLD4q32oddPseudo_UPD };
return SelectVLD(N, 4, DOpcodes, QOpcodes0, QOpcodes1);
}
@@ -2266,39 +2172,40 @@ SDNode *ARMDAGToDAGISel::Select(SDNode *N) {
case Intrinsic::arm_neon_vst1: {
unsigned DOpcodes[] = { ARM::VST1d8, ARM::VST1d16,
ARM::VST1d32, ARM::VST1d64 };
- unsigned QOpcodes[] = { ARM::VST1q8, ARM::VST1q16,
- ARM::VST1q32, ARM::VST1q64 };
+ unsigned QOpcodes[] = { ARM::VST1q8Pseudo, ARM::VST1q16Pseudo,
+ ARM::VST1q32Pseudo, ARM::VST1q64Pseudo };
return SelectVST(N, 1, DOpcodes, QOpcodes, 0);
}
case Intrinsic::arm_neon_vst2: {
- unsigned DOpcodes[] = { ARM::VST2d8, ARM::VST2d16,
- ARM::VST2d32, ARM::VST1q64 };
- unsigned QOpcodes[] = { ARM::VST2q8, ARM::VST2q16, ARM::VST2q32 };
+ unsigned DOpcodes[] = { ARM::VST2d8Pseudo, ARM::VST2d16Pseudo,
+ ARM::VST2d32Pseudo, ARM::VST1q64Pseudo };
+ unsigned QOpcodes[] = { ARM::VST2q8Pseudo, ARM::VST2q16Pseudo,
+ ARM::VST2q32Pseudo };
return SelectVST(N, 2, DOpcodes, QOpcodes, 0);
}
case Intrinsic::arm_neon_vst3: {
- unsigned DOpcodes[] = { ARM::VST3d8, ARM::VST3d16,
- ARM::VST3d32, ARM::VST1d64T };
- unsigned QOpcodes0[] = { ARM::VST3q8_UPD,
- ARM::VST3q16_UPD,
- ARM::VST3q32_UPD };
- unsigned QOpcodes1[] = { ARM::VST3q8odd_UPD,
- ARM::VST3q16odd_UPD,
- ARM::VST3q32odd_UPD };
+ unsigned DOpcodes[] = { ARM::VST3d8Pseudo, ARM::VST3d16Pseudo,
+ ARM::VST3d32Pseudo, ARM::VST1d64TPseudo };
+ unsigned QOpcodes0[] = { ARM::VST3q8Pseudo_UPD,
+ ARM::VST3q16Pseudo_UPD,
+ ARM::VST3q32Pseudo_UPD };
+ unsigned QOpcodes1[] = { ARM::VST3q8oddPseudo_UPD,
+ ARM::VST3q16oddPseudo_UPD,
+ ARM::VST3q32oddPseudo_UPD };
return SelectVST(N, 3, DOpcodes, QOpcodes0, QOpcodes1);
}
case Intrinsic::arm_neon_vst4: {
- unsigned DOpcodes[] = { ARM::VST4d8, ARM::VST4d16,
- ARM::VST4d32, ARM::VST1d64Q };
- unsigned QOpcodes0[] = { ARM::VST4q8_UPD,
- ARM::VST4q16_UPD,
- ARM::VST4q32_UPD };
- unsigned QOpcodes1[] = { ARM::VST4q8odd_UPD,
- ARM::VST4q16odd_UPD,
- ARM::VST4q32odd_UPD };
+ unsigned DOpcodes[] = { ARM::VST4d8Pseudo, ARM::VST4d16Pseudo,
+ ARM::VST4d32Pseudo, ARM::VST1d64QPseudo };
+ unsigned QOpcodes0[] = { ARM::VST4q8Pseudo_UPD,
+ ARM::VST4q16Pseudo_UPD,
+ ARM::VST4q32Pseudo_UPD };
+ unsigned QOpcodes1[] = { ARM::VST4q8oddPseudo_UPD,
+ ARM::VST4q16oddPseudo_UPD,
+ ARM::VST4q32oddPseudo_UPD };
return SelectVST(N, 4, DOpcodes, QOpcodes0, QOpcodes1);
}
generated by cgit v1.2.3 (git 2.25.1) at 2025-02-25 12:05:42 +0000