2017-10-20 05:37:38 +08:00
|
|
|
//===-- RISCVISelDAGToDAG.cpp - A dag to dag inst selector for RISCV ------===//
|
|
|
|
//
|
2019-01-19 16:50:56 +08:00
|
|
|
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
|
|
// See https://llvm.org/LICENSE.txt for license information.
|
|
|
|
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
2017-10-20 05:37:38 +08:00
|
|
|
//
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
//
|
|
|
|
// This file defines an instruction selector for the RISCV target.
|
|
|
|
//
|
|
|
|
//===----------------------------------------------------------------------===//
|
|
|
|
|
2020-03-31 14:28:24 +08:00
|
|
|
#include "RISCVISelDAGToDAG.h"
|
2017-10-20 05:37:38 +08:00
|
|
|
#include "MCTargetDesc/RISCVMCTargetDesc.h"
|
2021-01-15 03:44:02 +08:00
|
|
|
#include "MCTargetDesc/RISCVMatInt.h"
|
2017-12-11 19:53:54 +08:00
|
|
|
#include "llvm/CodeGen/MachineFrameInfo.h"
|
2020-12-19 04:08:27 +08:00
|
|
|
#include "llvm/IR/IntrinsicsRISCV.h"
|
2020-06-24 20:53:27 +08:00
|
|
|
#include "llvm/Support/Alignment.h"
|
2017-10-20 05:37:38 +08:00
|
|
|
#include "llvm/Support/Debug.h"
|
2021-01-28 12:36:21 +08:00
|
|
|
#include "llvm/Support/KnownBits.h"
|
2017-10-20 05:37:38 +08:00
|
|
|
#include "llvm/Support/MathExtras.h"
|
|
|
|
#include "llvm/Support/raw_ostream.h"
|
2020-03-31 14:28:24 +08:00
|
|
|
|
2017-10-20 05:37:38 +08:00
|
|
|
using namespace llvm;
|
|
|
|
|
|
|
|
#define DEBUG_TYPE "riscv-isel"
|
|
|
|
|
2018-04-12 13:34:25 +08:00
|
|
|
void RISCVDAGToDAGISel::PostprocessISelDAG() {
|
|
|
|
doPeepholeLoadStoreADDI();
|
|
|
|
}
|
2018-03-19 19:54:28 +08:00
|
|
|
|
2018-11-16 18:14:16 +08:00
|
|
|
static SDNode *selectImm(SelectionDAG *CurDAG, const SDLoc &DL, int64_t Imm,
|
|
|
|
MVT XLenVT) {
|
|
|
|
RISCVMatInt::InstSeq Seq;
|
|
|
|
RISCVMatInt::generateInstSeq(Imm, XLenVT == MVT::i64, Seq);
|
|
|
|
|
2019-09-20 21:48:02 +08:00
|
|
|
SDNode *Result = nullptr;
|
2018-11-16 18:14:16 +08:00
|
|
|
SDValue SrcReg = CurDAG->getRegister(RISCV::X0, XLenVT);
|
|
|
|
for (RISCVMatInt::Inst &Inst : Seq) {
|
|
|
|
SDValue SDImm = CurDAG->getTargetConstant(Inst.Imm, DL, XLenVT);
|
|
|
|
if (Inst.Opc == RISCV::LUI)
|
|
|
|
Result = CurDAG->getMachineNode(RISCV::LUI, DL, XLenVT, SDImm);
|
|
|
|
else
|
|
|
|
Result = CurDAG->getMachineNode(Inst.Opc, DL, XLenVT, SrcReg, SDImm);
|
|
|
|
|
|
|
|
// Only the first instruction has X0 as its source.
|
|
|
|
SrcReg = SDValue(Result, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
return Result;
|
|
|
|
}
|
|
|
|
|
2021-01-31 07:57:12 +08:00
|
|
|
static RISCVVLMUL getLMUL(MVT VT) {
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
switch (VT.getSizeInBits().getKnownMinValue() / 8) {
|
|
|
|
default:
|
|
|
|
llvm_unreachable("Invalid LMUL.");
|
|
|
|
case 1:
|
|
|
|
return RISCVVLMUL::LMUL_F8;
|
|
|
|
case 2:
|
|
|
|
return RISCVVLMUL::LMUL_F4;
|
|
|
|
case 4:
|
|
|
|
return RISCVVLMUL::LMUL_F2;
|
|
|
|
case 8:
|
|
|
|
return RISCVVLMUL::LMUL_1;
|
|
|
|
case 16:
|
|
|
|
return RISCVVLMUL::LMUL_2;
|
|
|
|
case 32:
|
|
|
|
return RISCVVLMUL::LMUL_4;
|
|
|
|
case 64:
|
|
|
|
return RISCVVLMUL::LMUL_8;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-02-12 06:19:30 +08:00
|
|
|
static unsigned getRegClassIDForLMUL(RISCVVLMUL LMul) {
|
|
|
|
switch (LMul) {
|
|
|
|
default:
|
|
|
|
llvm_unreachable("Invalid LMUL.");
|
|
|
|
case RISCVVLMUL::LMUL_F8:
|
|
|
|
case RISCVVLMUL::LMUL_F4:
|
|
|
|
case RISCVVLMUL::LMUL_F2:
|
|
|
|
case RISCVVLMUL::LMUL_1:
|
|
|
|
return RISCV::VRRegClassID;
|
|
|
|
case RISCVVLMUL::LMUL_2:
|
|
|
|
return RISCV::VRM2RegClassID;
|
|
|
|
case RISCVVLMUL::LMUL_4:
|
|
|
|
return RISCV::VRM4RegClassID;
|
|
|
|
case RISCVVLMUL::LMUL_8:
|
|
|
|
return RISCV::VRM8RegClassID;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-01-31 07:57:12 +08:00
|
|
|
static unsigned getSubregIndexByMVT(MVT VT, unsigned Index) {
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
RISCVVLMUL LMUL = getLMUL(VT);
|
|
|
|
if (LMUL == RISCVVLMUL::LMUL_F8 || LMUL == RISCVVLMUL::LMUL_F4 ||
|
|
|
|
LMUL == RISCVVLMUL::LMUL_F2 || LMUL == RISCVVLMUL::LMUL_1) {
|
|
|
|
static_assert(RISCV::sub_vrm1_7 == RISCV::sub_vrm1_0 + 7,
|
|
|
|
"Unexpected subreg numbering");
|
|
|
|
return RISCV::sub_vrm1_0 + Index;
|
|
|
|
} else if (LMUL == RISCVVLMUL::LMUL_2) {
|
|
|
|
static_assert(RISCV::sub_vrm2_3 == RISCV::sub_vrm2_0 + 3,
|
|
|
|
"Unexpected subreg numbering");
|
|
|
|
return RISCV::sub_vrm2_0 + Index;
|
|
|
|
} else if (LMUL == RISCVVLMUL::LMUL_4) {
|
|
|
|
static_assert(RISCV::sub_vrm4_1 == RISCV::sub_vrm4_0 + 1,
|
|
|
|
"Unexpected subreg numbering");
|
|
|
|
return RISCV::sub_vrm4_0 + Index;
|
|
|
|
}
|
|
|
|
llvm_unreachable("Invalid vector type.");
|
|
|
|
}
|
|
|
|
|
|
|
|
static SDValue createTupleImpl(SelectionDAG &CurDAG, ArrayRef<SDValue> Regs,
|
|
|
|
unsigned RegClassID, unsigned SubReg0) {
|
|
|
|
assert(Regs.size() >= 2 && Regs.size() <= 8);
|
|
|
|
|
|
|
|
SDLoc DL(Regs[0]);
|
|
|
|
SmallVector<SDValue, 8> Ops;
|
|
|
|
|
|
|
|
Ops.push_back(CurDAG.getTargetConstant(RegClassID, DL, MVT::i32));
|
|
|
|
|
|
|
|
for (unsigned I = 0; I < Regs.size(); ++I) {
|
|
|
|
Ops.push_back(Regs[I]);
|
|
|
|
Ops.push_back(CurDAG.getTargetConstant(SubReg0 + I, DL, MVT::i32));
|
|
|
|
}
|
|
|
|
SDNode *N =
|
|
|
|
CurDAG.getMachineNode(TargetOpcode::REG_SEQUENCE, DL, MVT::Untyped, Ops);
|
|
|
|
return SDValue(N, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static SDValue createM1Tuple(SelectionDAG &CurDAG, ArrayRef<SDValue> Regs,
|
|
|
|
unsigned NF) {
|
|
|
|
static const unsigned RegClassIDs[] = {
|
|
|
|
RISCV::VRN2M1RegClassID, RISCV::VRN3M1RegClassID, RISCV::VRN4M1RegClassID,
|
|
|
|
RISCV::VRN5M1RegClassID, RISCV::VRN6M1RegClassID, RISCV::VRN7M1RegClassID,
|
|
|
|
RISCV::VRN8M1RegClassID};
|
|
|
|
|
|
|
|
return createTupleImpl(CurDAG, Regs, RegClassIDs[NF - 2], RISCV::sub_vrm1_0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static SDValue createM2Tuple(SelectionDAG &CurDAG, ArrayRef<SDValue> Regs,
|
|
|
|
unsigned NF) {
|
|
|
|
static const unsigned RegClassIDs[] = {RISCV::VRN2M2RegClassID,
|
|
|
|
RISCV::VRN3M2RegClassID,
|
|
|
|
RISCV::VRN4M2RegClassID};
|
|
|
|
|
|
|
|
return createTupleImpl(CurDAG, Regs, RegClassIDs[NF - 2], RISCV::sub_vrm2_0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static SDValue createM4Tuple(SelectionDAG &CurDAG, ArrayRef<SDValue> Regs,
|
|
|
|
unsigned NF) {
|
|
|
|
return createTupleImpl(CurDAG, Regs, RISCV::VRN2M4RegClassID,
|
|
|
|
RISCV::sub_vrm4_0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static SDValue createTuple(SelectionDAG &CurDAG, ArrayRef<SDValue> Regs,
|
|
|
|
unsigned NF, RISCVVLMUL LMUL) {
|
|
|
|
switch (LMUL) {
|
|
|
|
default:
|
|
|
|
llvm_unreachable("Invalid LMUL.");
|
|
|
|
case RISCVVLMUL::LMUL_F8:
|
|
|
|
case RISCVVLMUL::LMUL_F4:
|
|
|
|
case RISCVVLMUL::LMUL_F2:
|
|
|
|
case RISCVVLMUL::LMUL_1:
|
|
|
|
return createM1Tuple(CurDAG, Regs, NF);
|
|
|
|
case RISCVVLMUL::LMUL_2:
|
|
|
|
return createM2Tuple(CurDAG, Regs, NF);
|
|
|
|
case RISCVVLMUL::LMUL_4:
|
|
|
|
return createM4Tuple(CurDAG, Regs, NF);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-02-17 14:58:14 +08:00
|
|
|
void RISCVDAGToDAGISel::selectVLSEG(SDNode *Node, unsigned IntNo, bool IsMasked,
|
2021-01-15 19:29:51 +08:00
|
|
|
bool IsStrided) {
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
SDLoc DL(Node);
|
|
|
|
unsigned NF = Node->getNumValues() - 1;
|
2021-01-31 07:57:12 +08:00
|
|
|
MVT VT = Node->getSimpleValueType(0);
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
unsigned ScalarSize = VT.getScalarSizeInBits();
|
|
|
|
MVT XLenVT = Subtarget->getXLenVT();
|
|
|
|
RISCVVLMUL LMUL = getLMUL(VT);
|
|
|
|
SDValue SEW = CurDAG->getTargetConstant(ScalarSize, DL, XLenVT);
|
2021-02-17 14:58:14 +08:00
|
|
|
unsigned CurOp = 2;
|
2021-01-15 19:29:51 +08:00
|
|
|
SmallVector<SDValue, 7> Operands;
|
2021-02-17 14:58:14 +08:00
|
|
|
if (IsMasked) {
|
|
|
|
SmallVector<SDValue, 8> Regs(Node->op_begin() + CurOp,
|
|
|
|
Node->op_begin() + CurOp + NF);
|
|
|
|
SDValue MaskedOff = createTuple(*CurDAG, Regs, NF, LMUL);
|
|
|
|
Operands.push_back(MaskedOff);
|
|
|
|
CurOp += NF;
|
2021-01-15 19:29:51 +08:00
|
|
|
}
|
2021-02-17 14:58:14 +08:00
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Base pointer.
|
|
|
|
if (IsStrided)
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Stride.
|
|
|
|
if (IsMasked)
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Mask.
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // VL.
|
2021-01-15 19:29:51 +08:00
|
|
|
Operands.push_back(SEW);
|
2021-02-17 14:58:14 +08:00
|
|
|
Operands.push_back(Node->getOperand(0)); // Chain.
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
const RISCVZvlssegTable::RISCVZvlsseg *P = RISCVZvlssegTable::getPseudo(
|
2021-01-18 10:02:40 +08:00
|
|
|
IntNo, ScalarSize, static_cast<unsigned>(LMUL),
|
|
|
|
static_cast<unsigned>(RISCVVLMUL::LMUL_1));
|
|
|
|
SDNode *Load =
|
|
|
|
CurDAG->getMachineNode(P->Pseudo, DL, MVT::Untyped, MVT::Other, Operands);
|
|
|
|
SDValue SuperReg = SDValue(Load, 0);
|
|
|
|
for (unsigned I = 0; I < NF; ++I)
|
|
|
|
ReplaceUses(SDValue(Node, I),
|
2021-01-31 07:57:12 +08:00
|
|
|
CurDAG->getTargetExtractSubreg(getSubregIndexByMVT(VT, I), DL,
|
2021-01-18 10:02:40 +08:00
|
|
|
VT, SuperReg));
|
|
|
|
|
|
|
|
ReplaceUses(SDValue(Node, NF), SDValue(Load, 1));
|
|
|
|
CurDAG->RemoveDeadNode(Node);
|
|
|
|
}
|
|
|
|
|
2021-02-17 14:58:14 +08:00
|
|
|
void RISCVDAGToDAGISel::selectVLSEGFF(SDNode *Node, bool IsMasked) {
|
2021-01-24 13:37:38 +08:00
|
|
|
SDLoc DL(Node);
|
|
|
|
unsigned IntNo = cast<ConstantSDNode>(Node->getOperand(1))->getZExtValue();
|
2021-01-28 03:01:07 +08:00
|
|
|
unsigned NF = Node->getNumValues() - 2; // Do not count VL and Chain.
|
2021-01-31 07:57:12 +08:00
|
|
|
MVT VT = Node->getSimpleValueType(0);
|
2021-01-24 13:37:38 +08:00
|
|
|
MVT XLenVT = Subtarget->getXLenVT();
|
|
|
|
unsigned ScalarSize = VT.getScalarSizeInBits();
|
|
|
|
RISCVVLMUL LMUL = getLMUL(VT);
|
|
|
|
SDValue SEW = CurDAG->getTargetConstant(ScalarSize, DL, XLenVT);
|
2021-02-17 14:58:14 +08:00
|
|
|
|
|
|
|
unsigned CurOp = 2;
|
2021-01-24 13:37:38 +08:00
|
|
|
SmallVector<SDValue, 7> Operands;
|
2021-02-17 14:58:14 +08:00
|
|
|
if (IsMasked) {
|
|
|
|
SmallVector<SDValue, 8> Regs(Node->op_begin() + CurOp,
|
|
|
|
Node->op_begin() + CurOp + NF);
|
|
|
|
SDValue MaskedOff = createTuple(*CurDAG, Regs, NF, LMUL);
|
|
|
|
Operands.push_back(MaskedOff);
|
|
|
|
CurOp += NF;
|
|
|
|
}
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Base pointer.
|
|
|
|
if (IsMasked)
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Mask.
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // VL.
|
2021-01-24 13:37:38 +08:00
|
|
|
Operands.push_back(SEW);
|
2021-02-17 14:58:14 +08:00
|
|
|
Operands.push_back(Node->getOperand(0)); // Chain.
|
2021-01-24 13:37:38 +08:00
|
|
|
const RISCVZvlssegTable::RISCVZvlsseg *P = RISCVZvlssegTable::getPseudo(
|
|
|
|
IntNo, ScalarSize, static_cast<unsigned>(LMUL),
|
|
|
|
static_cast<unsigned>(RISCVVLMUL::LMUL_1));
|
|
|
|
SDNode *Load = CurDAG->getMachineNode(P->Pseudo, DL, MVT::Untyped, MVT::Other,
|
|
|
|
MVT::Glue, Operands);
|
2021-01-28 03:01:07 +08:00
|
|
|
SDNode *ReadVL = CurDAG->getMachineNode(RISCV::PseudoReadVL, DL, XLenVT,
|
|
|
|
/*Glue*/ SDValue(Load, 2));
|
|
|
|
|
2021-01-24 13:37:38 +08:00
|
|
|
SDValue SuperReg = SDValue(Load, 0);
|
|
|
|
for (unsigned I = 0; I < NF; ++I)
|
|
|
|
ReplaceUses(SDValue(Node, I),
|
2021-01-31 07:57:12 +08:00
|
|
|
CurDAG->getTargetExtractSubreg(getSubregIndexByMVT(VT, I), DL,
|
2021-01-24 13:37:38 +08:00
|
|
|
VT, SuperReg));
|
|
|
|
|
2021-01-28 03:01:07 +08:00
|
|
|
ReplaceUses(SDValue(Node, NF), SDValue(ReadVL, 0)); // VL
|
|
|
|
ReplaceUses(SDValue(Node, NF + 1), SDValue(Load, 1)); // Chain
|
2021-01-24 13:37:38 +08:00
|
|
|
CurDAG->RemoveDeadNode(Node);
|
|
|
|
}
|
|
|
|
|
2021-02-17 14:58:14 +08:00
|
|
|
void RISCVDAGToDAGISel::selectVLXSEG(SDNode *Node, unsigned IntNo,
|
|
|
|
bool IsMasked) {
|
2021-01-18 10:02:40 +08:00
|
|
|
SDLoc DL(Node);
|
|
|
|
unsigned NF = Node->getNumValues() - 1;
|
2021-01-31 07:57:12 +08:00
|
|
|
MVT VT = Node->getSimpleValueType(0);
|
2021-01-18 10:02:40 +08:00
|
|
|
unsigned ScalarSize = VT.getScalarSizeInBits();
|
|
|
|
MVT XLenVT = Subtarget->getXLenVT();
|
|
|
|
RISCVVLMUL LMUL = getLMUL(VT);
|
|
|
|
SDValue SEW = CurDAG->getTargetConstant(ScalarSize, DL, XLenVT);
|
2021-02-17 14:58:14 +08:00
|
|
|
unsigned CurOp = 2;
|
|
|
|
SmallVector<SDValue, 7> Operands;
|
|
|
|
if (IsMasked) {
|
|
|
|
SmallVector<SDValue, 8> Regs(Node->op_begin() + CurOp,
|
|
|
|
Node->op_begin() + CurOp + NF);
|
|
|
|
SDValue MaskedOff = createTuple(*CurDAG, Regs, NF, LMUL);
|
|
|
|
Operands.push_back(MaskedOff);
|
|
|
|
CurOp += NF;
|
|
|
|
}
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Base pointer.
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Index.
|
|
|
|
MVT IndexVT = Operands.back()->getSimpleValueType(0);
|
|
|
|
if (IsMasked)
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Mask.
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // VL.
|
|
|
|
Operands.push_back(SEW);
|
|
|
|
Operands.push_back(Node->getOperand(0)); // Chain.
|
2021-01-18 10:02:40 +08:00
|
|
|
|
|
|
|
RISCVVLMUL IndexLMUL = getLMUL(IndexVT);
|
|
|
|
unsigned IndexScalarSize = IndexVT.getScalarSizeInBits();
|
|
|
|
const RISCVZvlssegTable::RISCVZvlsseg *P = RISCVZvlssegTable::getPseudo(
|
|
|
|
IntNo, IndexScalarSize, static_cast<unsigned>(LMUL),
|
|
|
|
static_cast<unsigned>(IndexLMUL));
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
SDNode *Load =
|
|
|
|
CurDAG->getMachineNode(P->Pseudo, DL, MVT::Untyped, MVT::Other, Operands);
|
|
|
|
SDValue SuperReg = SDValue(Load, 0);
|
|
|
|
for (unsigned I = 0; I < NF; ++I)
|
|
|
|
ReplaceUses(SDValue(Node, I),
|
2021-01-31 07:57:12 +08:00
|
|
|
CurDAG->getTargetExtractSubreg(getSubregIndexByMVT(VT, I), DL,
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
VT, SuperReg));
|
|
|
|
|
|
|
|
ReplaceUses(SDValue(Node, NF), SDValue(Load, 1));
|
|
|
|
CurDAG->RemoveDeadNode(Node);
|
|
|
|
}
|
|
|
|
|
2021-02-17 14:58:14 +08:00
|
|
|
void RISCVDAGToDAGISel::selectVSSEG(SDNode *Node, unsigned IntNo, bool IsMasked,
|
2021-01-16 21:40:41 +08:00
|
|
|
bool IsStrided) {
|
2021-01-14 17:07:18 +08:00
|
|
|
SDLoc DL(Node);
|
|
|
|
unsigned NF = Node->getNumOperands() - 4;
|
2021-01-16 21:40:41 +08:00
|
|
|
if (IsStrided)
|
|
|
|
NF--;
|
2021-02-17 14:58:14 +08:00
|
|
|
if (IsMasked)
|
|
|
|
NF--;
|
2021-01-31 07:57:12 +08:00
|
|
|
MVT VT = Node->getOperand(2)->getSimpleValueType(0);
|
2021-01-14 17:07:18 +08:00
|
|
|
unsigned ScalarSize = VT.getScalarSizeInBits();
|
|
|
|
MVT XLenVT = Subtarget->getXLenVT();
|
|
|
|
RISCVVLMUL LMUL = getLMUL(VT);
|
|
|
|
SDValue SEW = CurDAG->getTargetConstant(ScalarSize, DL, XLenVT);
|
|
|
|
SmallVector<SDValue, 8> Regs(Node->op_begin() + 2, Node->op_begin() + 2 + NF);
|
|
|
|
SDValue StoreVal = createTuple(*CurDAG, Regs, NF, LMUL);
|
2021-02-17 14:58:14 +08:00
|
|
|
SmallVector<SDValue, 7> Operands;
|
2021-01-16 21:40:41 +08:00
|
|
|
Operands.push_back(StoreVal);
|
2021-02-17 14:58:14 +08:00
|
|
|
unsigned CurOp = 2 + NF;
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Base pointer.
|
|
|
|
if (IsStrided)
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Stride.
|
|
|
|
if (IsMasked)
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Mask.
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // VL.
|
2021-01-16 21:40:41 +08:00
|
|
|
Operands.push_back(SEW);
|
|
|
|
Operands.push_back(Node->getOperand(0)); // Chain.
|
2021-01-14 17:07:18 +08:00
|
|
|
const RISCVZvlssegTable::RISCVZvlsseg *P = RISCVZvlssegTable::getPseudo(
|
2021-01-18 10:02:40 +08:00
|
|
|
IntNo, ScalarSize, static_cast<unsigned>(LMUL),
|
|
|
|
static_cast<unsigned>(RISCVVLMUL::LMUL_1));
|
2021-01-14 17:07:18 +08:00
|
|
|
SDNode *Store =
|
|
|
|
CurDAG->getMachineNode(P->Pseudo, DL, Node->getValueType(0), Operands);
|
|
|
|
ReplaceNode(Node, Store);
|
|
|
|
}
|
|
|
|
|
2021-02-17 14:58:14 +08:00
|
|
|
void RISCVDAGToDAGISel::selectVSXSEG(SDNode *Node, unsigned IntNo,
|
|
|
|
bool IsMasked) {
|
2021-01-14 17:07:18 +08:00
|
|
|
SDLoc DL(Node);
|
|
|
|
unsigned NF = Node->getNumOperands() - 5;
|
2021-02-17 14:58:14 +08:00
|
|
|
if (IsMasked)
|
|
|
|
--NF;
|
2021-01-31 07:57:12 +08:00
|
|
|
MVT VT = Node->getOperand(2)->getSimpleValueType(0);
|
2021-01-14 17:07:18 +08:00
|
|
|
unsigned ScalarSize = VT.getScalarSizeInBits();
|
|
|
|
MVT XLenVT = Subtarget->getXLenVT();
|
|
|
|
RISCVVLMUL LMUL = getLMUL(VT);
|
|
|
|
SDValue SEW = CurDAG->getTargetConstant(ScalarSize, DL, XLenVT);
|
2021-02-17 14:58:14 +08:00
|
|
|
SmallVector<SDValue, 7> Operands;
|
2021-01-14 17:07:18 +08:00
|
|
|
SmallVector<SDValue, 8> Regs(Node->op_begin() + 2, Node->op_begin() + 2 + NF);
|
|
|
|
SDValue StoreVal = createTuple(*CurDAG, Regs, NF, LMUL);
|
2021-01-16 21:40:41 +08:00
|
|
|
Operands.push_back(StoreVal);
|
2021-02-17 14:58:14 +08:00
|
|
|
unsigned CurOp = 2 + NF;
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Base pointer.
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Index.
|
|
|
|
MVT IndexVT = Operands.back()->getSimpleValueType(0);
|
|
|
|
if (IsMasked)
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // Mask.
|
|
|
|
Operands.push_back(Node->getOperand(CurOp++)); // VL.
|
2021-01-16 21:40:41 +08:00
|
|
|
Operands.push_back(SEW);
|
|
|
|
Operands.push_back(Node->getOperand(0)); // Chain.
|
2021-01-14 17:07:18 +08:00
|
|
|
|
2021-01-19 10:47:44 +08:00
|
|
|
RISCVVLMUL IndexLMUL = getLMUL(IndexVT);
|
|
|
|
unsigned IndexScalarSize = IndexVT.getScalarSizeInBits();
|
|
|
|
const RISCVZvlssegTable::RISCVZvlsseg *P = RISCVZvlssegTable::getPseudo(
|
|
|
|
IntNo, IndexScalarSize, static_cast<unsigned>(LMUL),
|
|
|
|
static_cast<unsigned>(IndexLMUL));
|
|
|
|
SDNode *Store =
|
|
|
|
CurDAG->getMachineNode(P->Pseudo, DL, Node->getValueType(0), Operands);
|
|
|
|
ReplaceNode(Node, Store);
|
|
|
|
}
|
|
|
|
|
2017-10-20 05:37:38 +08:00
|
|
|
void RISCVDAGToDAGISel::Select(SDNode *Node) {
|
2018-10-03 21:13:13 +08:00
|
|
|
// If we have a custom node, we have already selected.
|
2017-10-20 05:37:38 +08:00
|
|
|
if (Node->isMachineOpcode()) {
|
2018-05-14 20:53:11 +08:00
|
|
|
LLVM_DEBUG(dbgs() << "== "; Node->dump(CurDAG); dbgs() << "\n");
|
2017-10-20 05:37:38 +08:00
|
|
|
Node->setNodeId(-1);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2017-11-21 16:23:08 +08:00
|
|
|
// Instruction Selection not handled by the auto-generated tablegen selection
|
|
|
|
// should be handled here.
|
2018-10-03 21:13:13 +08:00
|
|
|
unsigned Opcode = Node->getOpcode();
|
|
|
|
MVT XLenVT = Subtarget->getXLenVT();
|
|
|
|
SDLoc DL(Node);
|
2021-01-31 07:57:12 +08:00
|
|
|
MVT VT = Node->getSimpleValueType(0);
|
2018-10-03 21:13:13 +08:00
|
|
|
|
|
|
|
switch (Opcode) {
|
2020-07-08 09:54:22 +08:00
|
|
|
case ISD::ADD: {
|
|
|
|
// Optimize (add r, imm) to (addi (addi r, imm0) imm1) if applicable. The
|
|
|
|
// immediate must be in specific ranges and have a single use.
|
|
|
|
if (auto *ConstOp = dyn_cast<ConstantSDNode>(Node->getOperand(1))) {
|
|
|
|
if (!(ConstOp->hasOneUse()))
|
|
|
|
break;
|
|
|
|
// The imm must be in range [-4096,-2049] or [2048,4094].
|
|
|
|
int64_t Imm = ConstOp->getSExtValue();
|
|
|
|
if (!(-4096 <= Imm && Imm <= -2049) && !(2048 <= Imm && Imm <= 4094))
|
|
|
|
break;
|
|
|
|
// Break the imm to imm0+imm1.
|
|
|
|
const SDValue ImmOp0 = CurDAG->getTargetConstant(Imm - Imm / 2, DL, VT);
|
|
|
|
const SDValue ImmOp1 = CurDAG->getTargetConstant(Imm / 2, DL, VT);
|
|
|
|
auto *NodeAddi0 = CurDAG->getMachineNode(RISCV::ADDI, DL, VT,
|
|
|
|
Node->getOperand(0), ImmOp0);
|
|
|
|
auto *NodeAddi1 = CurDAG->getMachineNode(RISCV::ADDI, DL, VT,
|
|
|
|
SDValue(NodeAddi0, 0), ImmOp1);
|
|
|
|
ReplaceNode(Node, NodeAddi1);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
2018-10-03 21:13:13 +08:00
|
|
|
case ISD::Constant: {
|
|
|
|
auto ConstNode = cast<ConstantSDNode>(Node);
|
|
|
|
if (VT == XLenVT && ConstNode->isNullValue()) {
|
2020-12-09 13:12:34 +08:00
|
|
|
SDValue New =
|
|
|
|
CurDAG->getCopyFromReg(CurDAG->getEntryNode(), DL, RISCV::X0, XLenVT);
|
2017-11-21 20:00:19 +08:00
|
|
|
ReplaceNode(Node, New.getNode());
|
|
|
|
return;
|
2017-11-21 16:23:08 +08:00
|
|
|
}
|
2018-11-16 18:14:16 +08:00
|
|
|
int64_t Imm = ConstNode->getSExtValue();
|
|
|
|
if (XLenVT == MVT::i64) {
|
2020-12-09 13:12:34 +08:00
|
|
|
ReplaceNode(Node, selectImm(CurDAG, DL, Imm, XLenVT));
|
2018-11-16 18:14:16 +08:00
|
|
|
return;
|
|
|
|
}
|
2018-10-03 21:13:13 +08:00
|
|
|
break;
|
2017-11-21 16:23:08 +08:00
|
|
|
}
|
2018-10-03 21:13:13 +08:00
|
|
|
case ISD::FrameIndex: {
|
2017-12-11 19:53:54 +08:00
|
|
|
SDValue Imm = CurDAG->getTargetConstant(0, DL, XLenVT);
|
2018-05-05 09:57:00 +08:00
|
|
|
int FI = cast<FrameIndexSDNode>(Node)->getIndex();
|
2017-12-11 19:53:54 +08:00
|
|
|
SDValue TFI = CurDAG->getTargetFrameIndex(FI, VT);
|
|
|
|
ReplaceNode(Node, CurDAG->getMachineNode(RISCV::ADDI, DL, VT, TFI, Imm));
|
|
|
|
return;
|
|
|
|
}
|
2021-02-02 01:21:43 +08:00
|
|
|
case ISD::SRL: {
|
|
|
|
// Optimize (srl (and X, 0xffff), C) -> (srli (slli X, 16), 16 + C).
|
2021-02-02 01:40:16 +08:00
|
|
|
// Taking into account that the 0xffff may have had lower bits unset by
|
|
|
|
// SimplifyDemandedBits. This avoids materializing the 0xffff immediate.
|
|
|
|
// This pattern occurs when type legalizing i16 right shifts.
|
2021-02-02 01:21:43 +08:00
|
|
|
// FIXME: This could be extended to other AND masks.
|
|
|
|
auto *N1C = dyn_cast<ConstantSDNode>(Node->getOperand(1));
|
|
|
|
if (N1C) {
|
|
|
|
uint64_t ShAmt = N1C->getZExtValue();
|
|
|
|
SDValue N0 = Node->getOperand(0);
|
|
|
|
if (ShAmt < 16 && N0.getOpcode() == ISD::AND && N0.hasOneUse() &&
|
|
|
|
isa<ConstantSDNode>(N0.getOperand(1))) {
|
|
|
|
uint64_t Mask = N0.getConstantOperandVal(1);
|
|
|
|
Mask |= maskTrailingOnes<uint64_t>(ShAmt);
|
|
|
|
if (Mask == 0xffff) {
|
|
|
|
SDLoc DL(Node);
|
|
|
|
unsigned SLLOpc = Subtarget->is64Bit() ? RISCV::SLLIW : RISCV::SLLI;
|
|
|
|
unsigned SRLOpc = Subtarget->is64Bit() ? RISCV::SRLIW : RISCV::SRLI;
|
|
|
|
SDNode *SLLI =
|
|
|
|
CurDAG->getMachineNode(SLLOpc, DL, VT, N0->getOperand(0),
|
|
|
|
CurDAG->getTargetConstant(16, DL, VT));
|
|
|
|
SDNode *SRLI = CurDAG->getMachineNode(
|
|
|
|
SRLOpc, DL, VT, SDValue(SLLI, 0),
|
|
|
|
CurDAG->getTargetConstant(16 + ShAmt, DL, VT));
|
|
|
|
ReplaceNode(Node, SRLI);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
break;
|
|
|
|
}
|
2020-12-19 04:08:27 +08:00
|
|
|
case ISD::INTRINSIC_W_CHAIN: {
|
|
|
|
unsigned IntNo = cast<ConstantSDNode>(Node->getOperand(1))->getZExtValue();
|
|
|
|
switch (IntNo) {
|
|
|
|
// By default we do not custom select any intrinsic.
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
|
2021-02-17 11:46:44 +08:00
|
|
|
case Intrinsic::riscv_vsetvli:
|
|
|
|
case Intrinsic::riscv_vsetvlimax: {
|
2020-12-19 04:08:27 +08:00
|
|
|
if (!Subtarget->hasStdExtV())
|
|
|
|
break;
|
|
|
|
|
2021-02-17 11:46:44 +08:00
|
|
|
bool VLMax = IntNo == Intrinsic::riscv_vsetvlimax;
|
|
|
|
unsigned Offset = VLMax ? 2 : 3;
|
|
|
|
|
|
|
|
assert(Node->getNumOperands() == Offset + 2 &&
|
|
|
|
"Unexpected number of operands");
|
2020-12-19 04:08:27 +08:00
|
|
|
|
|
|
|
RISCVVSEW VSEW =
|
2021-02-17 11:46:44 +08:00
|
|
|
static_cast<RISCVVSEW>(Node->getConstantOperandVal(Offset) & 0x7);
|
|
|
|
RISCVVLMUL VLMul = static_cast<RISCVVLMUL>(
|
|
|
|
Node->getConstantOperandVal(Offset + 1) & 0x7);
|
2020-12-19 04:08:27 +08:00
|
|
|
|
|
|
|
unsigned VTypeI = RISCVVType::encodeVTYPE(
|
|
|
|
VLMul, VSEW, /*TailAgnostic*/ true, /*MaskAgnostic*/ false);
|
|
|
|
SDValue VTypeIOp = CurDAG->getTargetConstant(VTypeI, DL, XLenVT);
|
|
|
|
|
2021-02-17 11:46:44 +08:00
|
|
|
SDValue VLOperand;
|
|
|
|
if (VLMax) {
|
|
|
|
VLOperand = CurDAG->getRegister(RISCV::X0, XLenVT);
|
|
|
|
} else {
|
|
|
|
VLOperand = Node->getOperand(2);
|
|
|
|
|
|
|
|
if (auto *C = dyn_cast<ConstantSDNode>(VLOperand)) {
|
|
|
|
uint64_t AVL = C->getZExtValue();
|
|
|
|
if (isUInt<5>(AVL)) {
|
|
|
|
SDValue VLImm = CurDAG->getTargetConstant(AVL, DL, XLenVT);
|
|
|
|
ReplaceNode(
|
|
|
|
Node, CurDAG->getMachineNode(RISCV::PseudoVSETIVLI, DL, XLenVT,
|
2021-02-01 16:08:46 +08:00
|
|
|
MVT::Other, VLImm, VTypeIOp,
|
|
|
|
/* Chain */ Node->getOperand(0)));
|
2021-02-17 11:46:44 +08:00
|
|
|
return;
|
|
|
|
}
|
2020-12-19 04:08:27 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
ReplaceNode(Node,
|
|
|
|
CurDAG->getMachineNode(RISCV::PseudoVSETVLI, DL, XLenVT,
|
|
|
|
MVT::Other, VLOperand, VTypeIOp,
|
|
|
|
/* Chain */ Node->getOperand(0)));
|
|
|
|
return;
|
|
|
|
}
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
case Intrinsic::riscv_vlseg2:
|
|
|
|
case Intrinsic::riscv_vlseg3:
|
|
|
|
case Intrinsic::riscv_vlseg4:
|
|
|
|
case Intrinsic::riscv_vlseg5:
|
|
|
|
case Intrinsic::riscv_vlseg6:
|
|
|
|
case Intrinsic::riscv_vlseg7:
|
|
|
|
case Intrinsic::riscv_vlseg8: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVLSEG(Node, IntNo, /*IsMasked*/ false, /*IsStrided*/ false);
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vlseg2_mask:
|
|
|
|
case Intrinsic::riscv_vlseg3_mask:
|
|
|
|
case Intrinsic::riscv_vlseg4_mask:
|
|
|
|
case Intrinsic::riscv_vlseg5_mask:
|
|
|
|
case Intrinsic::riscv_vlseg6_mask:
|
|
|
|
case Intrinsic::riscv_vlseg7_mask:
|
|
|
|
case Intrinsic::riscv_vlseg8_mask: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVLSEG(Node, IntNo, /*IsMasked*/ true, /*IsStrided*/ false);
|
2021-01-15 19:29:51 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vlsseg2:
|
|
|
|
case Intrinsic::riscv_vlsseg3:
|
|
|
|
case Intrinsic::riscv_vlsseg4:
|
|
|
|
case Intrinsic::riscv_vlsseg5:
|
|
|
|
case Intrinsic::riscv_vlsseg6:
|
|
|
|
case Intrinsic::riscv_vlsseg7:
|
|
|
|
case Intrinsic::riscv_vlsseg8: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVLSEG(Node, IntNo, /*IsMasked*/ false, /*IsStrided*/ true);
|
2021-01-15 19:29:51 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vlsseg2_mask:
|
|
|
|
case Intrinsic::riscv_vlsseg3_mask:
|
|
|
|
case Intrinsic::riscv_vlsseg4_mask:
|
|
|
|
case Intrinsic::riscv_vlsseg5_mask:
|
|
|
|
case Intrinsic::riscv_vlsseg6_mask:
|
|
|
|
case Intrinsic::riscv_vlsseg7_mask:
|
|
|
|
case Intrinsic::riscv_vlsseg8_mask: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVLSEG(Node, IntNo, /*IsMasked*/ true, /*IsStrided*/ true);
|
[RISCV] Implement vlseg intrinsics.
For Zvlsseg, we need continuous vector registers for the values. We need
to define new register classes for the different combinations of (number
of fields and LMUL). For example,
when the number of fields(NF) = 3, LMUL = 2, the values will be assigned
to (V0M2, V2M2, V4M2), (V2M2, V4M2, V6M2), (V4M2, V6M2, V8M2), ...
We define the vlseg intrinsics with multiple outputs. There is no way to
describe the codegen patterns with multiple outputs in the tablegen
files. We do the codegen in RISCVISelDAGToDAG and use EXTRACT_SUBREG to
extract the values of output.
The multiple scalable vector values will be put into a struct. This
patch is depended on the support for scalable vector struct.
Differential Revision: https://reviews.llvm.org/D94229
2020-12-31 17:14:15 +08:00
|
|
|
return;
|
|
|
|
}
|
2021-01-18 10:02:40 +08:00
|
|
|
case Intrinsic::riscv_vloxseg2:
|
|
|
|
case Intrinsic::riscv_vloxseg3:
|
|
|
|
case Intrinsic::riscv_vloxseg4:
|
|
|
|
case Intrinsic::riscv_vloxseg5:
|
|
|
|
case Intrinsic::riscv_vloxseg6:
|
|
|
|
case Intrinsic::riscv_vloxseg7:
|
|
|
|
case Intrinsic::riscv_vloxseg8:
|
|
|
|
case Intrinsic::riscv_vluxseg2:
|
|
|
|
case Intrinsic::riscv_vluxseg3:
|
|
|
|
case Intrinsic::riscv_vluxseg4:
|
|
|
|
case Intrinsic::riscv_vluxseg5:
|
|
|
|
case Intrinsic::riscv_vluxseg6:
|
|
|
|
case Intrinsic::riscv_vluxseg7:
|
|
|
|
case Intrinsic::riscv_vluxseg8: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVLXSEG(Node, IntNo, /*IsMasked*/ false);
|
2021-01-18 10:02:40 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vloxseg2_mask:
|
|
|
|
case Intrinsic::riscv_vloxseg3_mask:
|
|
|
|
case Intrinsic::riscv_vloxseg4_mask:
|
|
|
|
case Intrinsic::riscv_vloxseg5_mask:
|
|
|
|
case Intrinsic::riscv_vloxseg6_mask:
|
|
|
|
case Intrinsic::riscv_vloxseg7_mask:
|
|
|
|
case Intrinsic::riscv_vloxseg8_mask:
|
|
|
|
case Intrinsic::riscv_vluxseg2_mask:
|
|
|
|
case Intrinsic::riscv_vluxseg3_mask:
|
|
|
|
case Intrinsic::riscv_vluxseg4_mask:
|
|
|
|
case Intrinsic::riscv_vluxseg5_mask:
|
|
|
|
case Intrinsic::riscv_vluxseg6_mask:
|
|
|
|
case Intrinsic::riscv_vluxseg7_mask:
|
|
|
|
case Intrinsic::riscv_vluxseg8_mask: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVLXSEG(Node, IntNo, /*IsMasked*/ true);
|
2021-01-18 10:02:40 +08:00
|
|
|
return;
|
|
|
|
}
|
2021-01-28 03:01:07 +08:00
|
|
|
case Intrinsic::riscv_vlseg8ff:
|
|
|
|
case Intrinsic::riscv_vlseg7ff:
|
|
|
|
case Intrinsic::riscv_vlseg6ff:
|
|
|
|
case Intrinsic::riscv_vlseg5ff:
|
|
|
|
case Intrinsic::riscv_vlseg4ff:
|
|
|
|
case Intrinsic::riscv_vlseg3ff:
|
|
|
|
case Intrinsic::riscv_vlseg2ff: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVLSEGFF(Node, /*IsMasked*/ false);
|
2021-01-28 03:01:07 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vlseg8ff_mask:
|
|
|
|
case Intrinsic::riscv_vlseg7ff_mask:
|
|
|
|
case Intrinsic::riscv_vlseg6ff_mask:
|
|
|
|
case Intrinsic::riscv_vlseg5ff_mask:
|
|
|
|
case Intrinsic::riscv_vlseg4ff_mask:
|
|
|
|
case Intrinsic::riscv_vlseg3ff_mask:
|
|
|
|
case Intrinsic::riscv_vlseg2ff_mask: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVLSEGFF(Node, /*IsMasked*/ true);
|
2021-01-28 03:01:07 +08:00
|
|
|
return;
|
|
|
|
}
|
2020-12-19 04:08:27 +08:00
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
2021-01-14 17:07:18 +08:00
|
|
|
case ISD::INTRINSIC_VOID: {
|
|
|
|
unsigned IntNo = cast<ConstantSDNode>(Node->getOperand(1))->getZExtValue();
|
|
|
|
switch (IntNo) {
|
|
|
|
case Intrinsic::riscv_vsseg2:
|
|
|
|
case Intrinsic::riscv_vsseg3:
|
|
|
|
case Intrinsic::riscv_vsseg4:
|
|
|
|
case Intrinsic::riscv_vsseg5:
|
|
|
|
case Intrinsic::riscv_vsseg6:
|
|
|
|
case Intrinsic::riscv_vsseg7:
|
|
|
|
case Intrinsic::riscv_vsseg8: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVSSEG(Node, IntNo, /*IsMasked*/ false, /*IsStrided*/ false);
|
2021-01-14 17:07:18 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vsseg2_mask:
|
|
|
|
case Intrinsic::riscv_vsseg3_mask:
|
|
|
|
case Intrinsic::riscv_vsseg4_mask:
|
|
|
|
case Intrinsic::riscv_vsseg5_mask:
|
|
|
|
case Intrinsic::riscv_vsseg6_mask:
|
|
|
|
case Intrinsic::riscv_vsseg7_mask:
|
|
|
|
case Intrinsic::riscv_vsseg8_mask: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVSSEG(Node, IntNo, /*IsMasked*/ true, /*IsStrided*/ false);
|
2021-01-16 21:40:41 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vssseg2:
|
|
|
|
case Intrinsic::riscv_vssseg3:
|
|
|
|
case Intrinsic::riscv_vssseg4:
|
|
|
|
case Intrinsic::riscv_vssseg5:
|
|
|
|
case Intrinsic::riscv_vssseg6:
|
|
|
|
case Intrinsic::riscv_vssseg7:
|
|
|
|
case Intrinsic::riscv_vssseg8: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVSSEG(Node, IntNo, /*IsMasked*/ false, /*IsStrided*/ true);
|
2021-01-16 21:40:41 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vssseg2_mask:
|
|
|
|
case Intrinsic::riscv_vssseg3_mask:
|
|
|
|
case Intrinsic::riscv_vssseg4_mask:
|
|
|
|
case Intrinsic::riscv_vssseg5_mask:
|
|
|
|
case Intrinsic::riscv_vssseg6_mask:
|
|
|
|
case Intrinsic::riscv_vssseg7_mask:
|
|
|
|
case Intrinsic::riscv_vssseg8_mask: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVSSEG(Node, IntNo, /*IsMasked*/ true, /*IsStrided*/ true);
|
2021-01-14 17:07:18 +08:00
|
|
|
return;
|
|
|
|
}
|
2021-01-19 10:47:44 +08:00
|
|
|
case Intrinsic::riscv_vsoxseg2:
|
|
|
|
case Intrinsic::riscv_vsoxseg3:
|
|
|
|
case Intrinsic::riscv_vsoxseg4:
|
|
|
|
case Intrinsic::riscv_vsoxseg5:
|
|
|
|
case Intrinsic::riscv_vsoxseg6:
|
|
|
|
case Intrinsic::riscv_vsoxseg7:
|
|
|
|
case Intrinsic::riscv_vsoxseg8:
|
|
|
|
case Intrinsic::riscv_vsuxseg2:
|
|
|
|
case Intrinsic::riscv_vsuxseg3:
|
|
|
|
case Intrinsic::riscv_vsuxseg4:
|
|
|
|
case Intrinsic::riscv_vsuxseg5:
|
|
|
|
case Intrinsic::riscv_vsuxseg6:
|
|
|
|
case Intrinsic::riscv_vsuxseg7:
|
|
|
|
case Intrinsic::riscv_vsuxseg8: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVSXSEG(Node, IntNo, /*IsMasked*/ false);
|
2021-01-19 10:47:44 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
case Intrinsic::riscv_vsoxseg2_mask:
|
|
|
|
case Intrinsic::riscv_vsoxseg3_mask:
|
|
|
|
case Intrinsic::riscv_vsoxseg4_mask:
|
|
|
|
case Intrinsic::riscv_vsoxseg5_mask:
|
|
|
|
case Intrinsic::riscv_vsoxseg6_mask:
|
|
|
|
case Intrinsic::riscv_vsoxseg7_mask:
|
|
|
|
case Intrinsic::riscv_vsoxseg8_mask:
|
|
|
|
case Intrinsic::riscv_vsuxseg2_mask:
|
|
|
|
case Intrinsic::riscv_vsuxseg3_mask:
|
|
|
|
case Intrinsic::riscv_vsuxseg4_mask:
|
|
|
|
case Intrinsic::riscv_vsuxseg5_mask:
|
|
|
|
case Intrinsic::riscv_vsuxseg6_mask:
|
|
|
|
case Intrinsic::riscv_vsuxseg7_mask:
|
|
|
|
case Intrinsic::riscv_vsuxseg8_mask: {
|
2021-02-17 14:58:14 +08:00
|
|
|
selectVSXSEG(Node, IntNo, /*IsMasked*/ true);
|
2021-01-19 10:47:44 +08:00
|
|
|
return;
|
|
|
|
}
|
2021-01-14 17:07:18 +08:00
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
2021-02-12 06:19:30 +08:00
|
|
|
case ISD::INSERT_SUBVECTOR: {
|
|
|
|
// Bail when not a "cast" like insert_subvector.
|
|
|
|
if (Node->getConstantOperandVal(2) != 0)
|
|
|
|
break;
|
|
|
|
if (!Node->getOperand(0).isUndef())
|
|
|
|
break;
|
|
|
|
|
|
|
|
// Bail when normal isel should do the job.
|
2021-02-16 04:42:33 +08:00
|
|
|
MVT InVT = Node->getOperand(1).getSimpleValueType();
|
2021-02-12 06:19:30 +08:00
|
|
|
if (VT.isFixedLengthVector() || InVT.isScalableVector())
|
|
|
|
break;
|
|
|
|
|
2021-02-16 04:42:33 +08:00
|
|
|
unsigned RegClassID;
|
|
|
|
if (VT.getVectorElementType() == MVT::i1)
|
|
|
|
RegClassID = RISCV::VRRegClassID;
|
|
|
|
else
|
|
|
|
RegClassID = getRegClassIDForLMUL(getLMUL(VT));
|
|
|
|
|
2021-02-12 06:19:30 +08:00
|
|
|
SDValue V = Node->getOperand(1);
|
|
|
|
SDLoc DL(V);
|
|
|
|
SDValue RC =
|
|
|
|
CurDAG->getTargetConstant(RegClassID, DL, Subtarget->getXLenVT());
|
|
|
|
SDNode *NewNode =
|
|
|
|
CurDAG->getMachineNode(TargetOpcode::COPY_TO_REGCLASS, DL, VT, V, RC);
|
|
|
|
ReplaceNode(Node, NewNode);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
case ISD::EXTRACT_SUBVECTOR: {
|
|
|
|
// Bail when not a "cast" like extract_subvector.
|
|
|
|
if (Node->getConstantOperandVal(1) != 0)
|
|
|
|
break;
|
|
|
|
|
|
|
|
// Bail when normal isel can do the job.
|
2021-02-16 04:42:33 +08:00
|
|
|
MVT InVT = Node->getOperand(0).getSimpleValueType();
|
2021-02-12 06:19:30 +08:00
|
|
|
if (VT.isScalableVector() || InVT.isFixedLengthVector())
|
|
|
|
break;
|
|
|
|
|
2021-02-16 04:42:33 +08:00
|
|
|
unsigned RegClassID;
|
|
|
|
if (InVT.getVectorElementType() == MVT::i1)
|
|
|
|
RegClassID = RISCV::VRRegClassID;
|
|
|
|
else
|
|
|
|
RegClassID = getRegClassIDForLMUL(getLMUL(InVT));
|
|
|
|
|
2021-02-12 06:19:30 +08:00
|
|
|
SDValue V = Node->getOperand(0);
|
|
|
|
SDLoc DL(V);
|
|
|
|
SDValue RC =
|
|
|
|
CurDAG->getTargetConstant(RegClassID, DL, Subtarget->getXLenVT());
|
|
|
|
SDNode *NewNode =
|
|
|
|
CurDAG->getMachineNode(TargetOpcode::COPY_TO_REGCLASS, DL, VT, V, RC);
|
|
|
|
ReplaceNode(Node, NewNode);
|
|
|
|
return;
|
|
|
|
}
|
2018-10-03 21:13:13 +08:00
|
|
|
}
|
2017-11-21 16:23:08 +08:00
|
|
|
|
2017-10-20 05:37:38 +08:00
|
|
|
// Select the default instruction.
|
|
|
|
SelectCode(Node);
|
|
|
|
}
|
|
|
|
|
2018-01-11 04:05:09 +08:00
|
|
|
bool RISCVDAGToDAGISel::SelectInlineAsmMemoryOperand(
|
|
|
|
const SDValue &Op, unsigned ConstraintID, std::vector<SDValue> &OutOps) {
|
|
|
|
switch (ConstraintID) {
|
|
|
|
case InlineAsm::Constraint_m:
|
|
|
|
// We just support simple memory operands that have a single address
|
|
|
|
// operand and need no special handling.
|
|
|
|
OutOps.push_back(Op);
|
|
|
|
return false;
|
2019-08-16 18:28:34 +08:00
|
|
|
case InlineAsm::Constraint_A:
|
|
|
|
OutOps.push_back(Op);
|
|
|
|
return false;
|
2018-01-11 04:05:09 +08:00
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2017-12-11 19:53:54 +08:00
|
|
|
bool RISCVDAGToDAGISel::SelectAddrFI(SDValue Addr, SDValue &Base) {
|
2021-02-03 02:05:33 +08:00
|
|
|
if (auto *FIN = dyn_cast<FrameIndexSDNode>(Addr)) {
|
2017-12-11 19:53:54 +08:00
|
|
|
Base = CurDAG->getTargetFrameIndex(FIN->getIndex(), Subtarget->getXLenVT());
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2021-02-13 16:42:25 +08:00
|
|
|
bool RISCVDAGToDAGISel::SelectBaseAddr(SDValue Addr, SDValue &Base) {
|
2021-02-03 02:05:33 +08:00
|
|
|
// If this is FrameIndex, select it directly. Otherwise just let it get
|
|
|
|
// selected to a register independently.
|
|
|
|
if (auto *FIN = dyn_cast<FrameIndexSDNode>(Addr))
|
|
|
|
Base = CurDAG->getTargetFrameIndex(FIN->getIndex(), Subtarget->getXLenVT());
|
|
|
|
else
|
|
|
|
Base = Addr;
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2021-02-13 06:01:28 +08:00
|
|
|
bool RISCVDAGToDAGISel::selectShiftMask(SDValue N, unsigned ShiftWidth,
|
|
|
|
SDValue &ShAmt) {
|
|
|
|
// Shift instructions on RISCV only read the lower 5 or 6 bits of the shift
|
|
|
|
// amount. If there is an AND on the shift amount, we can bypass it if it
|
|
|
|
// doesn't affect any of those bits.
|
|
|
|
if (N.getOpcode() == ISD::AND && isa<ConstantSDNode>(N.getOperand(1))) {
|
|
|
|
const APInt &AndMask = N->getConstantOperandAPInt(1);
|
|
|
|
|
|
|
|
// Since the max shift amount is a power of 2 we can subtract 1 to make a
|
|
|
|
// mask that covers the bits needed to represent all shift amounts.
|
|
|
|
assert(isPowerOf2_32(ShiftWidth) && "Unexpected max shift amount!");
|
|
|
|
APInt ShMask(AndMask.getBitWidth(), ShiftWidth - 1);
|
|
|
|
|
|
|
|
if (ShMask.isSubsetOf(AndMask)) {
|
|
|
|
ShAmt = N.getOperand(0);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
// SimplifyDemandedBits may have optimized the mask so try restoring any
|
|
|
|
// bits that are known zero.
|
|
|
|
KnownBits Known = CurDAG->computeKnownBits(N->getOperand(0));
|
|
|
|
if (ShMask.isSubsetOf(AndMask | Known.Zero)) {
|
|
|
|
ShAmt = N.getOperand(0);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
2021-01-28 12:36:21 +08:00
|
|
|
|
2021-02-13 06:01:28 +08:00
|
|
|
ShAmt = N;
|
|
|
|
return true;
|
2021-01-28 12:36:21 +08:00
|
|
|
}
|
|
|
|
|
2021-01-06 03:16:50 +08:00
|
|
|
// Match (srl (and val, mask), imm) where the result would be a
|
|
|
|
// zero-extended 32-bit integer. i.e. the mask is 0xffffffff or the result
|
|
|
|
// is equivalent to this (SimplifyDemandedBits may have removed lower bits
|
|
|
|
// from the mask that aren't necessary due to the right-shifting).
|
|
|
|
bool RISCVDAGToDAGISel::MatchSRLIW(SDNode *N) const {
|
|
|
|
assert(N->getOpcode() == ISD::SRL);
|
|
|
|
assert(N->getOperand(0).getOpcode() == ISD::AND);
|
|
|
|
assert(isa<ConstantSDNode>(N->getOperand(1)));
|
|
|
|
assert(isa<ConstantSDNode>(N->getOperand(0).getOperand(1)));
|
|
|
|
|
|
|
|
// The IsRV64 predicate is checked after PatFrag predicates so we can get
|
|
|
|
// here even on RV32.
|
|
|
|
if (!Subtarget->is64Bit())
|
|
|
|
return false;
|
|
|
|
|
|
|
|
SDValue And = N->getOperand(0);
|
|
|
|
uint64_t ShAmt = N->getConstantOperandVal(1);
|
|
|
|
uint64_t Mask = And.getConstantOperandVal(1);
|
|
|
|
return (Mask | maskTrailingOnes<uint64_t>(ShAmt)) == 0xffffffff;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Check that it is a SLLIUW (Shift Logical Left Immediate Unsigned i32
|
|
|
|
// on RV64).
|
|
|
|
// SLLIUW is the same as SLLI except for the fact that it clears the bits
|
|
|
|
// XLEN-1:32 of the input RS1 before shifting.
|
|
|
|
// A PatFrag has already checked that it has the right structure:
|
2020-07-15 18:50:03 +08:00
|
|
|
//
|
2021-01-06 03:16:50 +08:00
|
|
|
// (AND (SHL RS1, VC2), VC1)
|
2020-07-15 18:50:03 +08:00
|
|
|
//
|
2021-01-06 03:16:50 +08:00
|
|
|
// We check that VC2, the shamt is less than 32, otherwise the pattern is
|
|
|
|
// exactly the same as SLLI and we give priority to that.
|
|
|
|
// Eventually we check that VC1, the mask used to clear the upper 32 bits
|
|
|
|
// of RS1, is correct:
|
2020-07-15 18:50:03 +08:00
|
|
|
//
|
2021-01-06 03:16:50 +08:00
|
|
|
// VC1 == (0xFFFFFFFF << VC2)
|
2020-11-17 01:22:42 +08:00
|
|
|
//
|
2021-01-06 03:16:50 +08:00
|
|
|
bool RISCVDAGToDAGISel::MatchSLLIUW(SDNode *N) const {
|
|
|
|
assert(N->getOpcode() == ISD::AND);
|
|
|
|
assert(N->getOperand(0).getOpcode() == ISD::SHL);
|
|
|
|
assert(isa<ConstantSDNode>(N->getOperand(1)));
|
|
|
|
assert(isa<ConstantSDNode>(N->getOperand(0).getOperand(1)));
|
|
|
|
|
|
|
|
// The IsRV64 predicate is checked after PatFrag predicates so we can get
|
|
|
|
// here even on RV32.
|
|
|
|
if (!Subtarget->is64Bit())
|
2020-11-17 01:22:42 +08:00
|
|
|
return false;
|
|
|
|
|
2021-01-06 03:16:50 +08:00
|
|
|
SDValue Shl = N->getOperand(0);
|
|
|
|
uint64_t VC1 = N->getConstantOperandVal(1);
|
|
|
|
uint64_t VC2 = Shl.getConstantOperandVal(1);
|
2020-11-17 01:22:42 +08:00
|
|
|
|
2021-01-06 03:16:50 +08:00
|
|
|
// Immediate range should be enforced by uimm5 predicate.
|
|
|
|
assert(VC2 < 32 && "Unexpected immediate");
|
2021-01-24 16:13:12 +08:00
|
|
|
return (VC1 >> VC2) == UINT64_C(0xFFFFFFFF);
|
2020-07-15 18:50:03 +08:00
|
|
|
}
|
|
|
|
|
2021-02-02 15:53:54 +08:00
|
|
|
// X0 has special meaning for vsetvl/vsetvli.
|
|
|
|
// rd | rs1 | AVL value | Effect on vl
|
|
|
|
//--------------------------------------------------------------
|
|
|
|
// !X0 | X0 | VLMAX | Set vl to VLMAX
|
|
|
|
// X0 | X0 | Value in vl | Keep current vl, just change vtype.
|
|
|
|
bool RISCVDAGToDAGISel::selectVLOp(SDValue N, SDValue &VL) {
|
|
|
|
// If the VL value is a constant 0, manually select it to an ADDI with 0
|
|
|
|
// immediate to prevent the default selection path from matching it to X0.
|
|
|
|
auto *C = dyn_cast<ConstantSDNode>(N);
|
|
|
|
if (C && C->isNullValue())
|
|
|
|
VL = SDValue(selectImm(CurDAG, SDLoc(N), 0, Subtarget->getXLenVT()), 0);
|
|
|
|
else
|
|
|
|
VL = N;
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2020-12-15 21:05:32 +08:00
|
|
|
bool RISCVDAGToDAGISel::selectVSplat(SDValue N, SDValue &SplatVal) {
|
|
|
|
if (N.getOpcode() != ISD::SPLAT_VECTOR &&
|
2021-02-13 01:09:22 +08:00
|
|
|
N.getOpcode() != RISCVISD::SPLAT_VECTOR_I64 &&
|
|
|
|
N.getOpcode() != RISCVISD::VMV_V_X_VL)
|
2020-12-15 21:05:32 +08:00
|
|
|
return false;
|
|
|
|
SplatVal = N.getOperand(0);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool RISCVDAGToDAGISel::selectVSplatSimm5(SDValue N, SDValue &SplatVal) {
|
|
|
|
if ((N.getOpcode() != ISD::SPLAT_VECTOR &&
|
2021-02-13 01:09:22 +08:00
|
|
|
N.getOpcode() != RISCVISD::SPLAT_VECTOR_I64 &&
|
|
|
|
N.getOpcode() != RISCVISD::VMV_V_X_VL) ||
|
2020-12-15 21:05:32 +08:00
|
|
|
!isa<ConstantSDNode>(N.getOperand(0)))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
int64_t SplatImm = cast<ConstantSDNode>(N.getOperand(0))->getSExtValue();
|
|
|
|
|
2020-12-24 21:20:51 +08:00
|
|
|
// Both ISD::SPLAT_VECTOR and RISCVISD::SPLAT_VECTOR_I64 share semantics when
|
|
|
|
// the operand type is wider than the resulting vector element type: an
|
|
|
|
// implicit truncation first takes place. Therefore, perform a manual
|
|
|
|
// truncation/sign-extension in order to ignore any truncated bits and catch
|
|
|
|
// any zero-extended immediate.
|
|
|
|
// For example, we wish to match (i8 -1) -> (XLenVT 255) as a simm5 by first
|
|
|
|
// sign-extending to (XLenVT -1).
|
2021-01-31 07:57:12 +08:00
|
|
|
MVT XLenVT = Subtarget->getXLenVT();
|
2020-12-24 21:20:51 +08:00
|
|
|
assert(XLenVT == N.getOperand(0).getSimpleValueType() &&
|
|
|
|
"Unexpected splat operand type");
|
2021-01-31 07:57:12 +08:00
|
|
|
MVT EltVT = N.getSimpleValueType().getVectorElementType();
|
2020-12-24 21:20:51 +08:00
|
|
|
if (EltVT.bitsLT(XLenVT)) {
|
|
|
|
SplatImm = SignExtend64(SplatImm, EltVT.getSizeInBits());
|
|
|
|
}
|
|
|
|
|
2020-12-15 21:05:32 +08:00
|
|
|
if (!isInt<5>(SplatImm))
|
|
|
|
return false;
|
|
|
|
|
2020-12-24 21:20:51 +08:00
|
|
|
SplatVal = CurDAG->getTargetConstant(SplatImm, SDLoc(N), XLenVT);
|
2020-12-15 21:05:32 +08:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool RISCVDAGToDAGISel::selectVSplatUimm5(SDValue N, SDValue &SplatVal) {
|
|
|
|
if ((N.getOpcode() != ISD::SPLAT_VECTOR &&
|
2021-02-13 01:09:22 +08:00
|
|
|
N.getOpcode() != RISCVISD::SPLAT_VECTOR_I64 &&
|
|
|
|
N.getOpcode() != RISCVISD::VMV_V_X_VL) ||
|
2020-12-15 21:05:32 +08:00
|
|
|
!isa<ConstantSDNode>(N.getOperand(0)))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
int64_t SplatImm = cast<ConstantSDNode>(N.getOperand(0))->getSExtValue();
|
|
|
|
|
|
|
|
if (!isUInt<5>(SplatImm))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
SplatVal =
|
|
|
|
CurDAG->getTargetConstant(SplatImm, SDLoc(N), Subtarget->getXLenVT());
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2021-02-13 01:09:22 +08:00
|
|
|
bool RISCVDAGToDAGISel::selectRVVSimm5(SDValue N, unsigned Width,
|
|
|
|
SDValue &Imm) {
|
|
|
|
if (auto *C = dyn_cast<ConstantSDNode>(N)) {
|
|
|
|
int64_t ImmVal = SignExtend64(C->getSExtValue(), Width);
|
|
|
|
|
|
|
|
if (!isInt<5>(ImmVal))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
Imm = CurDAG->getTargetConstant(ImmVal, SDLoc(N), Subtarget->getXLenVT());
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
bool RISCVDAGToDAGISel::selectRVVUimm5(SDValue N, unsigned Width,
|
|
|
|
SDValue &Imm) {
|
|
|
|
if (auto *C = dyn_cast<ConstantSDNode>(N)) {
|
|
|
|
int64_t ImmVal = C->getSExtValue();
|
|
|
|
|
|
|
|
if (!isUInt<5>(ImmVal))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
Imm = CurDAG->getTargetConstant(ImmVal, SDLoc(N), Subtarget->getXLenVT());
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2018-03-19 19:54:28 +08:00
|
|
|
// Merge an ADDI into the offset of a load/store instruction where possible.
|
2020-06-24 20:53:27 +08:00
|
|
|
// (load (addi base, off1), off2) -> (load base, off1+off2)
|
|
|
|
// (store val, (addi base, off1), off2) -> (store val, base, off1+off2)
|
|
|
|
// This is possible when off1+off2 fits a 12-bit immediate.
|
2018-03-19 19:54:28 +08:00
|
|
|
void RISCVDAGToDAGISel::doPeepholeLoadStoreADDI() {
|
|
|
|
SelectionDAG::allnodes_iterator Position(CurDAG->getRoot().getNode());
|
|
|
|
++Position;
|
|
|
|
|
|
|
|
while (Position != CurDAG->allnodes_begin()) {
|
|
|
|
SDNode *N = &*--Position;
|
|
|
|
// Skip dead nodes and any non-machine opcodes.
|
|
|
|
if (N->use_empty() || !N->isMachineOpcode())
|
|
|
|
continue;
|
|
|
|
|
|
|
|
int OffsetOpIdx;
|
|
|
|
int BaseOpIdx;
|
|
|
|
|
|
|
|
// Only attempt this optimisation for I-type loads and S-type stores.
|
|
|
|
switch (N->getMachineOpcode()) {
|
|
|
|
default:
|
|
|
|
continue;
|
|
|
|
case RISCV::LB:
|
|
|
|
case RISCV::LH:
|
|
|
|
case RISCV::LW:
|
|
|
|
case RISCV::LBU:
|
|
|
|
case RISCV::LHU:
|
|
|
|
case RISCV::LWU:
|
|
|
|
case RISCV::LD:
|
2020-07-03 22:57:59 +08:00
|
|
|
case RISCV::FLH:
|
2018-03-19 19:54:28 +08:00
|
|
|
case RISCV::FLW:
|
|
|
|
case RISCV::FLD:
|
|
|
|
BaseOpIdx = 0;
|
|
|
|
OffsetOpIdx = 1;
|
|
|
|
break;
|
|
|
|
case RISCV::SB:
|
|
|
|
case RISCV::SH:
|
|
|
|
case RISCV::SW:
|
|
|
|
case RISCV::SD:
|
2020-07-03 22:57:59 +08:00
|
|
|
case RISCV::FSH:
|
2018-03-19 19:54:28 +08:00
|
|
|
case RISCV::FSW:
|
|
|
|
case RISCV::FSD:
|
|
|
|
BaseOpIdx = 1;
|
|
|
|
OffsetOpIdx = 2;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2020-06-24 20:53:27 +08:00
|
|
|
if (!isa<ConstantSDNode>(N->getOperand(OffsetOpIdx)))
|
2018-03-19 19:54:28 +08:00
|
|
|
continue;
|
|
|
|
|
|
|
|
SDValue Base = N->getOperand(BaseOpIdx);
|
|
|
|
|
2018-04-19 03:02:31 +08:00
|
|
|
// If the base is an ADDI, we can merge it in to the load/store.
|
|
|
|
if (!Base.isMachineOpcode() || Base.getMachineOpcode() != RISCV::ADDI)
|
2018-03-19 19:54:28 +08:00
|
|
|
continue;
|
|
|
|
|
2018-04-19 03:02:31 +08:00
|
|
|
SDValue ImmOperand = Base.getOperand(1);
|
2020-06-24 20:53:27 +08:00
|
|
|
uint64_t Offset2 = N->getConstantOperandVal(OffsetOpIdx);
|
2018-04-19 03:02:31 +08:00
|
|
|
|
|
|
|
if (auto Const = dyn_cast<ConstantSDNode>(ImmOperand)) {
|
2020-06-24 20:53:27 +08:00
|
|
|
int64_t Offset1 = Const->getSExtValue();
|
|
|
|
int64_t CombinedOffset = Offset1 + Offset2;
|
|
|
|
if (!isInt<12>(CombinedOffset))
|
|
|
|
continue;
|
|
|
|
ImmOperand = CurDAG->getTargetConstant(CombinedOffset, SDLoc(ImmOperand),
|
|
|
|
ImmOperand.getValueType());
|
2018-04-19 03:02:31 +08:00
|
|
|
} else if (auto GA = dyn_cast<GlobalAddressSDNode>(ImmOperand)) {
|
2020-06-24 20:53:27 +08:00
|
|
|
// If the off1 in (addi base, off1) is a global variable's address (its
|
|
|
|
// low part, really), then we can rely on the alignment of that variable
|
|
|
|
// to provide a margin of safety before off1 can overflow the 12 bits.
|
|
|
|
// Check if off2 falls within that margin; if so off1+off2 can't overflow.
|
|
|
|
const DataLayout &DL = CurDAG->getDataLayout();
|
|
|
|
Align Alignment = GA->getGlobal()->getPointerAlignment(DL);
|
|
|
|
if (Offset2 != 0 && Alignment <= Offset2)
|
|
|
|
continue;
|
|
|
|
int64_t Offset1 = GA->getOffset();
|
|
|
|
int64_t CombinedOffset = Offset1 + Offset2;
|
2018-04-19 03:02:31 +08:00
|
|
|
ImmOperand = CurDAG->getTargetGlobalAddress(
|
|
|
|
GA->getGlobal(), SDLoc(ImmOperand), ImmOperand.getValueType(),
|
2020-06-24 20:53:27 +08:00
|
|
|
CombinedOffset, GA->getTargetFlags());
|
2020-05-12 01:52:42 +08:00
|
|
|
} else if (auto CP = dyn_cast<ConstantPoolSDNode>(ImmOperand)) {
|
2020-06-24 20:53:27 +08:00
|
|
|
// Ditto.
|
|
|
|
Align Alignment = CP->getAlign();
|
|
|
|
if (Offset2 != 0 && Alignment <= Offset2)
|
|
|
|
continue;
|
|
|
|
int64_t Offset1 = CP->getOffset();
|
|
|
|
int64_t CombinedOffset = Offset1 + Offset2;
|
2020-05-12 01:52:42 +08:00
|
|
|
ImmOperand = CurDAG->getTargetConstantPool(
|
|
|
|
CP->getConstVal(), ImmOperand.getValueType(), CP->getAlign(),
|
2020-06-24 20:53:27 +08:00
|
|
|
CombinedOffset, CP->getTargetFlags());
|
2018-03-19 19:54:28 +08:00
|
|
|
} else {
|
2018-04-19 03:02:31 +08:00
|
|
|
continue;
|
2018-03-19 19:54:28 +08:00
|
|
|
}
|
|
|
|
|
2018-05-14 20:53:11 +08:00
|
|
|
LLVM_DEBUG(dbgs() << "Folding add-immediate into mem-op:\nBase: ");
|
|
|
|
LLVM_DEBUG(Base->dump(CurDAG));
|
|
|
|
LLVM_DEBUG(dbgs() << "\nN: ");
|
|
|
|
LLVM_DEBUG(N->dump(CurDAG));
|
|
|
|
LLVM_DEBUG(dbgs() << "\n");
|
2018-03-19 19:54:28 +08:00
|
|
|
|
|
|
|
// Modify the offset operand of the load/store.
|
|
|
|
if (BaseOpIdx == 0) // Load
|
2018-04-19 03:02:31 +08:00
|
|
|
CurDAG->UpdateNodeOperands(N, Base.getOperand(0), ImmOperand,
|
|
|
|
N->getOperand(2));
|
2018-03-19 19:54:28 +08:00
|
|
|
else // Store
|
2018-04-19 03:02:31 +08:00
|
|
|
CurDAG->UpdateNodeOperands(N, N->getOperand(0), Base.getOperand(0),
|
|
|
|
ImmOperand, N->getOperand(3));
|
|
|
|
|
|
|
|
// The add-immediate may now be dead, in which case remove it.
|
|
|
|
if (Base.getNode()->use_empty())
|
|
|
|
CurDAG->RemoveDeadNode(Base.getNode());
|
2018-03-19 19:54:28 +08:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-10-20 05:37:38 +08:00
|
|
|
// This pass converts a legalized DAG into a RISCV-specific DAG, ready
|
|
|
|
// for instruction scheduling.
|
|
|
|
FunctionPass *llvm::createRISCVISelDag(RISCVTargetMachine &TM) {
|
|
|
|
return new RISCVDAGToDAGISel(TM);
|
|
|
|
}
|