//===-- RISCVInstrInfoZb.td - RISC-V Bitmanip instructions -*- tablegen -*-===// // // 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 // //===----------------------------------------------------------------------===// // // This file describes the RISC-V instructions from the standard Bitmanip // extensions, versions: // Zba - 1.0 // Zbb - 1.0 // Zbc - 1.0 // Zbs - 1.0 // // The experimental extensions appeared in an earlier draft of the Bitmanip // extensions. They are not ratified and subject to change. // // This file also describes RISC-V instructions from the Zbk* extensions in // Cryptography Extensions Volume I: Scalar & Entropy Source Instructions, // versions: // Zbkb - 1.0 // Zbkc - 1.0 // Zbkx - 1.0 // //===----------------------------------------------------------------------===// //===----------------------------------------------------------------------===// // Operand and SDNode transformation definitions. //===----------------------------------------------------------------------===// def riscv_clzw : SDNode<"RISCVISD::CLZW", SDT_RISCVIntUnaryOpW>; def riscv_ctzw : SDNode<"RISCVISD::CTZW", SDT_RISCVIntUnaryOpW>; def riscv_rolw : SDNode<"RISCVISD::ROLW", SDT_RISCVIntBinOpW>; def riscv_rorw : SDNode<"RISCVISD::RORW", SDT_RISCVIntBinOpW>; def riscv_brev8 : SDNode<"RISCVISD::BREV8", SDTIntUnaryOp>; def riscv_orc_b : SDNode<"RISCVISD::ORC_B", SDTIntUnaryOp>; def riscv_zip : SDNode<"RISCVISD::ZIP", SDTIntUnaryOp>; def riscv_unzip : SDNode<"RISCVISD::UNZIP", SDTIntUnaryOp>; def riscv_absw : SDNode<"RISCVISD::ABSW", SDTIntUnaryOp>; def UImmLog2XLenHalfAsmOperand : AsmOperandClass { let Name = "UImmLog2XLenHalf"; let RenderMethod = "addImmOperands"; let DiagnosticType = "InvalidUImmLog2XLenHalf"; } def shfl_uimm : Operand, ImmLeafis64Bit()) return isUInt<5>(Imm); return isUInt<4>(Imm); }]> { let ParserMatchClass = UImmLog2XLenHalfAsmOperand; let DecoderMethod = "decodeUImmOperand<5>"; let OperandType = "OPERAND_UIMM_SHFL"; let OperandNamespace = "RISCVOp"; let MCOperandPredicate = [{ int64_t Imm; if (!MCOp.evaluateAsConstantImm(Imm)) return false; if (STI.getTargetTriple().isArch64Bit()) return isUInt<5>(Imm); return isUInt<4>(Imm); }]; } def BCLRXForm : SDNodeXFormgetTargetConstant(countTrailingOnes(N->getZExtValue()), SDLoc(N), N->getValueType(0)); }]>; def SingleBitSetMaskToIndex : SDNodeXFormgetTargetConstant(countTrailingZeros(N->getZExtValue()), SDLoc(N), N->getValueType(0)); }]>; // Checks if this mask has a single 0 bit and cannot be used with ANDI. def BCLRMask : ImmLeafis64Bit()) return !isInt<12>(Imm) && isPowerOf2_64(~Imm); return !isInt<12>(Imm) && isPowerOf2_32(~Imm); }], BCLRXForm>; // Checks if this mask has a single 1 bit and cannot be used with ORI/XORI. def SingleBitSetMask : ImmLeafis64Bit()) return !isInt<12>(Imm) && isPowerOf2_64(Imm); return !isInt<12>(Imm) && isPowerOf2_32(Imm); }], SingleBitSetMaskToIndex>; // Check if (or r, i) can be optimized to (BSETI (BSETI r, i0), i1), // in which i = (1 << i0) | (1 << i1). def BSETINVTwoBitsMask : PatLeaf<(imm), [{ if (!N->hasOneUse()) return false; // The immediate should not be a simm12. if (isInt<12>(N->getSExtValue())) return false; // The immediate must have exactly two bits set. return llvm::popcount(N->getZExtValue()) == 2; }]>; def BSETINVTwoBitsMaskHigh : SDNodeXFormgetZExtValue(); return CurDAG->getTargetConstant(63 - countLeadingZeros(I), SDLoc(N), N->getValueType(0)); }]>; // Check if (or r, imm) can be optimized to (BSETI (ORI r, i0), i1), // in which imm = i0 | (1 << i1). def BSETINVORIMask : PatLeaf<(imm), [{ if (!N->hasOneUse()) return false; // The immediate should not be a simm12. if (isInt<12>(N->getSExtValue())) return false; // There should be only one set bit from bit 11 to the top. return isPowerOf2_64(N->getZExtValue() & ~0x7ff); }]>; def BSETINVORIMaskLow : SDNodeXFormgetTargetConstant(N->getZExtValue() & 0x7ff, SDLoc(N), N->getValueType(0)); }]>; // Check if (and r, i) can be optimized to (BCLRI (BCLRI r, i0), i1), // in which i = ~((1<hasOneUse()) return false; // The immediate should not be a simm12. if (isInt<12>(N->getSExtValue())) return false; // The immediate must have exactly two bits clear. return (unsigned)llvm::popcount(N->getZExtValue()) == Subtarget->getXLen() - 2; }]>; def BCLRITwoBitsMaskLow : SDNodeXFormgetTargetConstant(countTrailingZeros(~N->getZExtValue()), SDLoc(N), N->getValueType(0)); }]>; def BCLRITwoBitsMaskHigh : SDNodeXFormgetSExtValue(); if (!Subtarget->is64Bit()) I |= 0xffffffffull << 32; return CurDAG->getTargetConstant(63 - countLeadingZeros(~I), SDLoc(N), N->getValueType(0)); }]>; // Check if (and r, i) can be optimized to (BCLRI (ANDI r, i0), i1), // in which i = i0 & ~(1<hasOneUse()) return false; // The immediate should not be a simm12. if (isInt<12>(N->getSExtValue())) return false; // There should be only one clear bit from bit 11 to the top. uint64_t I = N->getZExtValue() | 0x7ff; return Subtarget->is64Bit() ? isPowerOf2_64(~I) : isPowerOf2_32(~I); }]>; def BCLRIANDIMaskLow : SDNodeXFormgetTargetConstant((N->getZExtValue() & 0x7ff) | ~0x7ffull, SDLoc(N), N->getValueType(0)); }]>; def C3LeftShift : PatLeaf<(imm), [{ uint64_t C = N->getZExtValue(); return C > 3 && ((C % 3) == 0) && isPowerOf2_64(C / 3); }]>; def C5LeftShift : PatLeaf<(imm), [{ uint64_t C = N->getZExtValue(); return C > 5 && ((C % 5) == 0) && isPowerOf2_64(C / 5); }]>; def C9LeftShift : PatLeaf<(imm), [{ uint64_t C = N->getZExtValue(); return C > 9 && ((C % 9) == 0) && isPowerOf2_64(C / 9); }]>; // Constant of the form (3 << C) where C is less than 32. def C3LeftShiftUW : PatLeaf<(imm), [{ uint64_t C = N->getZExtValue(); if (C <= 3 || (C % 3) != 0) return false; C /= 3; return isPowerOf2_64(C) && C < (1ULL << 32); }]>; // Constant of the form (5 << C) where C is less than 32. def C5LeftShiftUW : PatLeaf<(imm), [{ uint64_t C = N->getZExtValue(); if (C <= 5 || (C % 5) != 0) return false; C /= 5; return isPowerOf2_64(C) && C < (1ULL << 32); }]>; // Constant of the form (9 << C) where C is less than 32. def C9LeftShiftUW : PatLeaf<(imm), [{ uint64_t C = N->getZExtValue(); if (C <= 9 || (C % 9) != 0) return false; C /= 9; return isPowerOf2_64(C) && C < (1ULL << 32); }]>; def CSImm12MulBy4 : PatLeaf<(imm), [{ if (!N->hasOneUse()) return false; int64_t C = N->getSExtValue(); // Skip if C is simm12, an lui, or can be optimized by the PatLeaf AddiPair. return !isInt<13>(C) && !isShiftedInt<20, 12>(C) && isShiftedInt<12, 2>(C); }]>; def CSImm12MulBy8 : PatLeaf<(imm), [{ if (!N->hasOneUse()) return false; int64_t C = N->getSExtValue(); // Skip if C is simm12, an lui or can be optimized by the PatLeaf AddiPair or // CSImm12MulBy4. return !isInt<14>(C) && !isShiftedInt<20, 12>(C) && isShiftedInt<12, 3>(C); }]>; def SimmShiftRightBy2XForm : SDNodeXFormgetTargetConstant(N->getSExtValue() >> 2, SDLoc(N), N->getValueType(0)); }]>; def SimmShiftRightBy3XForm : SDNodeXFormgetTargetConstant(N->getSExtValue() >> 3, SDLoc(N), N->getValueType(0)); }]>; // Pattern to exclude simm12 immediates from matching. def non_imm12 : PatLeaf<(XLenVT GPR:$a), [{ auto *C = dyn_cast(N); return !C || !isInt<12>(C->getSExtValue()); }]>; def Shifted32OnesMask : PatLeaf<(imm), [{ uint64_t Imm = N->getZExtValue(); if (!isShiftedMask_64(Imm)) return false; unsigned TrailingZeros = countTrailingZeros(Imm); return TrailingZeros > 0 && TrailingZeros < 32 && Imm == UINT64_C(0xFFFFFFFF) << TrailingZeros; }], TrailingZeros>; def sh1add_op : ComplexPattern", [], [], 6>; def sh2add_op : ComplexPattern", [], [], 6>; def sh3add_op : ComplexPattern", [], [], 6>; def sh1add_uw_op : ComplexPattern", [], [], 6>; def sh2add_uw_op : ComplexPattern", [], [], 6>; def sh3add_uw_op : ComplexPattern", [], [], 6>; //===----------------------------------------------------------------------===// // Instruction class templates //===----------------------------------------------------------------------===// // Some of these templates should be moved to RISCVInstrFormats.td once the B // extension has been ratified. let hasSideEffects = 0, mayLoad = 0, mayStore = 0 in class RVBUnary funct7, bits<5> funct5, bits<3> funct3, RISCVOpcode opcode, string opcodestr> : RVInstR { let rs2 = funct5; } let hasSideEffects = 0, mayLoad = 0, mayStore = 0 in class RVBShift_ri imm11_7, bits<3> funct3, RISCVOpcode opcode, string opcodestr> : RVInstIShift; let hasSideEffects = 0, mayLoad = 0, mayStore = 0 in class RVBShiftW_ri imm11_5, bits<3> funct3, RISCVOpcode opcode, string opcodestr> : RVInstIShiftW; // Using RVInstIShiftW since it allocates 5 bits instead of 6 to shamt. let hasSideEffects = 0, mayLoad = 0, mayStore = 0 in class RVBShfl_ri imm11_5, bits<3> funct3, RISCVOpcode opcode, string opcodestr> : RVInstIShiftW; let hasSideEffects = 0, mayLoad = 0, mayStore = 0 in class RVBTernaryR funct2, bits<3> funct3, RISCVOpcode opcode, string opcodestr, string argstr> : RVInstR4; //===----------------------------------------------------------------------===// // Instructions //===----------------------------------------------------------------------===// let Predicates = [HasStdExtZbbOrZbkb] in { def ANDN : ALU_rr<0b0100000, 0b111, "andn">, Sched<[WriteIALU, ReadIALU, ReadIALU]>; def ORN : ALU_rr<0b0100000, 0b110, "orn">, Sched<[WriteIALU, ReadIALU, ReadIALU]>; def XNOR : ALU_rr<0b0100000, 0b100, "xnor">, Sched<[WriteIALU, ReadIALU, ReadIALU]>; } // Predicates = [HasStdExtZbbOrZbkb] let Predicates = [HasStdExtZba] in { def SH1ADD : ALU_rr<0b0010000, 0b010, "sh1add">, Sched<[WriteSHXADD, ReadSHXADD, ReadSHXADD]>; def SH2ADD : ALU_rr<0b0010000, 0b100, "sh2add">, Sched<[WriteSHXADD, ReadSHXADD, ReadSHXADD]>; def SH3ADD : ALU_rr<0b0010000, 0b110, "sh3add">, Sched<[WriteSHXADD, ReadSHXADD, ReadSHXADD]>; } // Predicates = [HasStdExtZba] let Predicates = [HasStdExtZba, IsRV64] in { def SLLI_UW : RVBShift_ri<0b00001, 0b001, OPC_OP_IMM_32, "slli.uw">, Sched<[WriteShiftImm32, ReadShiftImm32]>; def ADD_UW : ALUW_rr<0b0000100, 0b000, "add.uw">, Sched<[WriteIALU32, ReadIALU32, ReadIALU32]>; def SH1ADD_UW : ALUW_rr<0b0010000, 0b010, "sh1add.uw">, Sched<[WriteSHXADD32, ReadSHXADD32, ReadSHXADD32]>; def SH2ADD_UW : ALUW_rr<0b0010000, 0b100, "sh2add.uw">, Sched<[WriteSHXADD32, ReadSHXADD32, ReadSHXADD32]>; def SH3ADD_UW : ALUW_rr<0b0010000, 0b110, "sh3add.uw">, Sched<[WriteSHXADD32, ReadSHXADD32, ReadSHXADD32]>; } // Predicates = [HasStdExtZba, IsRV64] let Predicates = [HasStdExtZbbOrZbkb] in { def ROL : ALU_rr<0b0110000, 0b001, "rol">, Sched<[WriteRotateReg, ReadRotateReg, ReadRotateReg]>; def ROR : ALU_rr<0b0110000, 0b101, "ror">, Sched<[WriteRotateReg, ReadRotateReg, ReadRotateReg]>; def RORI : RVBShift_ri<0b01100, 0b101, OPC_OP_IMM, "rori">, Sched<[WriteRotateImm, ReadRotateImm]>; } // Predicates = [HasStdExtZbbOrZbkb] let Predicates = [HasStdExtZbbOrZbkb, IsRV64], IsSignExtendingOpW = 1 in { def ROLW : ALUW_rr<0b0110000, 0b001, "rolw">, Sched<[WriteRotateReg32, ReadRotateReg32, ReadRotateReg32]>; def RORW : ALUW_rr<0b0110000, 0b101, "rorw">, Sched<[WriteRotateReg32, ReadRotateReg32, ReadRotateReg32]>; def RORIW : RVBShiftW_ri<0b0110000, 0b101, OPC_OP_IMM_32, "roriw">, Sched<[WriteRotateImm32, ReadRotateImm32]>; } // Predicates = [HasStdExtZbbOrZbkb, IsRV64] let Predicates = [HasStdExtZbs] in { def BCLR : ALU_rr<0b0100100, 0b001, "bclr">, Sched<[WriteSingleBit, ReadSingleBit, ReadSingleBit]>; def BSET : ALU_rr<0b0010100, 0b001, "bset">, Sched<[WriteSingleBit, ReadSingleBit, ReadSingleBit]>; def BINV : ALU_rr<0b0110100, 0b001, "binv">, Sched<[WriteSingleBit, ReadSingleBit, ReadSingleBit]>; let IsSignExtendingOpW = 1 in def BEXT : ALU_rr<0b0100100, 0b101, "bext">, Sched<[WriteSingleBit, ReadSingleBit, ReadSingleBit]>; def BCLRI : RVBShift_ri<0b01001, 0b001, OPC_OP_IMM, "bclri">, Sched<[WriteSingleBitImm, ReadSingleBitImm]>; def BSETI : RVBShift_ri<0b00101, 0b001, OPC_OP_IMM, "bseti">, Sched<[WriteSingleBitImm, ReadSingleBitImm]>; def BINVI : RVBShift_ri<0b01101, 0b001, OPC_OP_IMM, "binvi">, Sched<[WriteSingleBitImm, ReadSingleBitImm]>; let IsSignExtendingOpW = 1 in def BEXTI : RVBShift_ri<0b01001, 0b101, OPC_OP_IMM, "bexti">, Sched<[WriteSingleBitImm, ReadSingleBitImm]>; } // Predicates = [HasStdExtZbs] // These instructions were named xperm.n and xperm.b in the last version of // the draft bit manipulation specification they were included in. However, we // use the mnemonics given to them in the ratified Zbkx extension. let Predicates = [HasStdExtZbkx] in { def XPERM4 : ALU_rr<0b0010100, 0b010, "xperm4">, Sched<[WriteXPERM, ReadXPERM, ReadXPERM]>; def XPERM8 : ALU_rr<0b0010100, 0b100, "xperm8">, Sched<[WriteXPERM, ReadXPERM, ReadXPERM]>; } // Predicates = [HasStdExtZbkx] let Predicates = [HasStdExtZbb], IsSignExtendingOpW = 1 in { def CLZ : RVBUnary<0b0110000, 0b00000, 0b001, OPC_OP_IMM, "clz">, Sched<[WriteCLZ, ReadCLZ]>; def CTZ : RVBUnary<0b0110000, 0b00001, 0b001, OPC_OP_IMM, "ctz">, Sched<[WriteCTZ, ReadCTZ]>; def CPOP : RVBUnary<0b0110000, 0b00010, 0b001, OPC_OP_IMM, "cpop">, Sched<[WriteCPOP, ReadCPOP]>; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbb, IsRV64], IsSignExtendingOpW = 1 in { def CLZW : RVBUnary<0b0110000, 0b00000, 0b001, OPC_OP_IMM_32, "clzw">, Sched<[WriteCLZ32, ReadCLZ32]>; def CTZW : RVBUnary<0b0110000, 0b00001, 0b001, OPC_OP_IMM_32, "ctzw">, Sched<[WriteCTZ32, ReadCTZ32]>; def CPOPW : RVBUnary<0b0110000, 0b00010, 0b001, OPC_OP_IMM_32, "cpopw">, Sched<[WriteCPOP32, ReadCPOP32]>; } // Predicates = [HasStdExtZbb, IsRV64] let Predicates = [HasStdExtZbb], IsSignExtendingOpW = 1 in { def SEXT_B : RVBUnary<0b0110000, 0b00100, 0b001, OPC_OP_IMM, "sext.b">, Sched<[WriteIALU, ReadIALU]>; def SEXT_H : RVBUnary<0b0110000, 0b00101, 0b001, OPC_OP_IMM, "sext.h">, Sched<[WriteIALU, ReadIALU]>; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbc] in { def CLMULR : ALU_rr<0b0000101, 0b010, "clmulr", /*Commutable*/1>, Sched<[WriteCLMUL, ReadCLMUL, ReadCLMUL]>; } // Predicates = [HasStdExtZbc] let Predicates = [HasStdExtZbcOrZbkc] in { def CLMUL : ALU_rr<0b0000101, 0b001, "clmul", /*Commutable*/1>, Sched<[WriteCLMUL, ReadCLMUL, ReadCLMUL]>; def CLMULH : ALU_rr<0b0000101, 0b011, "clmulh", /*Commutable*/1>, Sched<[WriteCLMUL, ReadCLMUL, ReadCLMUL]>; } // Predicates = [HasStdExtZbcOrZbkc] let Predicates = [HasStdExtZbb] in { def MIN : ALU_rr<0b0000101, 0b100, "min", /*Commutable*/1>, Sched<[WriteIALU, ReadIALU, ReadIALU]>; def MINU : ALU_rr<0b0000101, 0b101, "minu", /*Commutable*/1>, Sched<[WriteIALU, ReadIALU, ReadIALU]>; def MAX : ALU_rr<0b0000101, 0b110, "max", /*Commutable*/1>, Sched<[WriteIALU, ReadIALU, ReadIALU]>; def MAXU : ALU_rr<0b0000101, 0b111, "maxu", /*Commutable*/1>, Sched<[WriteIALU, ReadIALU, ReadIALU]>; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbkb] in { def PACK : ALU_rr<0b0000100, 0b100, "pack">, Sched<[WritePACK, ReadPACK, ReadPACK]>; let IsSignExtendingOpW = 1 in def PACKH : ALU_rr<0b0000100, 0b111, "packh">, Sched<[WritePACK, ReadPACK, ReadPACK]>; } // Predicates = [HasStdExtZbkb] let Predicates = [HasStdExtZbkb, IsRV64], IsSignExtendingOpW = 1 in def PACKW : ALUW_rr<0b0000100, 0b100, "packw">, Sched<[WritePACK32, ReadPACK32, ReadPACK32]>; let Predicates = [HasStdExtZbb, IsRV32] in { def ZEXT_H_RV32 : RVBUnary<0b0000100, 0b00000, 0b100, OPC_OP, "zext.h">, Sched<[WriteIALU, ReadIALU]>; } // Predicates = [HasStdExtZbb, IsRV32] let Predicates = [HasStdExtZbb, IsRV64], IsSignExtendingOpW = 1 in { def ZEXT_H_RV64 : RVBUnary<0b0000100, 0b00000, 0b100, OPC_OP_32, "zext.h">, Sched<[WriteIALU, ReadIALU]>; } // Predicates = [HasStdExtZbb, IsRV64] let Predicates = [HasStdExtZbbOrZbkb, IsRV32] in { def REV8_RV32 : RVBUnary<0b0110100, 0b11000, 0b101, OPC_OP_IMM, "rev8">, Sched<[WriteREV8, ReadREV8]>; } // Predicates = [HasStdExtZbbOrZbkb, IsRV32] let Predicates = [HasStdExtZbbOrZbkb, IsRV64] in { def REV8_RV64 : RVBUnary<0b0110101, 0b11000, 0b101, OPC_OP_IMM, "rev8">, Sched<[WriteREV8, ReadREV8]>; } // Predicates = [HasStdExtZbbOrZbkb, IsRV64] let Predicates = [HasStdExtZbb] in { def ORC_B : RVBUnary<0b0010100, 0b00111, 0b101, OPC_OP_IMM, "orc.b">, Sched<[WriteORCB, ReadORCB]>; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbkb] in def BREV8 : RVBUnary<0b0110100, 0b00111, 0b101, OPC_OP_IMM, "brev8">, Sched<[WriteBREV8, ReadBREV8]>; let Predicates = [HasStdExtZbkb, IsRV32] in { def ZIP_RV32 : RVBUnary<0b0000100, 0b01111, 0b001, OPC_OP_IMM, "zip">, Sched<[WriteZIP, ReadZIP]>; def UNZIP_RV32 : RVBUnary<0b0000100, 0b01111, 0b101, OPC_OP_IMM, "unzip">, Sched<[WriteZIP, ReadZIP]>; } // Predicates = [HasStdExtZbkb, IsRV32] //===----------------------------------------------------------------------===// // Pseudo Instructions //===----------------------------------------------------------------------===// let Predicates = [HasStdExtZba, IsRV64] in { def : InstAlias<"zext.w $rd, $rs", (ADD_UW GPR:$rd, GPR:$rs, X0)>; } // Predicates = [HasStdExtZba, IsRV64] let Predicates = [HasStdExtZbb] in { def : InstAlias<"ror $rd, $rs1, $shamt", (RORI GPR:$rd, GPR:$rs1, uimmlog2xlen:$shamt), 0>; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbb, IsRV64] in { def : InstAlias<"rorw $rd, $rs1, $shamt", (RORIW GPR:$rd, GPR:$rs1, uimm5:$shamt), 0>; } // Predicates = [HasStdExtZbb, IsRV64] let Predicates = [HasStdExtZbs] in { def : InstAlias<"bset $rd, $rs1, $shamt", (BSETI GPR:$rd, GPR:$rs1, uimmlog2xlen:$shamt), 0>; def : InstAlias<"bclr $rd, $rs1, $shamt", (BCLRI GPR:$rd, GPR:$rs1, uimmlog2xlen:$shamt), 0>; def : InstAlias<"binv $rd, $rs1, $shamt", (BINVI GPR:$rd, GPR:$rs1, uimmlog2xlen:$shamt), 0>; def : InstAlias<"bext $rd, $rs1, $shamt", (BEXTI GPR:$rd, GPR:$rs1, uimmlog2xlen:$shamt), 0>; } // Predicates = [HasStdExtZbs] //===----------------------------------------------------------------------===// // Codegen patterns //===----------------------------------------------------------------------===// let Predicates = [HasStdExtZbbOrZbkb] in { def : Pat<(and GPR:$rs1, (not GPR:$rs2)), (ANDN GPR:$rs1, GPR:$rs2)>; def : Pat<(or GPR:$rs1, (not GPR:$rs2)), (ORN GPR:$rs1, GPR:$rs2)>; def : Pat<(xor GPR:$rs1, (not GPR:$rs2)), (XNOR GPR:$rs1, GPR:$rs2)>; } // Predicates = [HasStdExtZbbOrZbkb] let Predicates = [HasStdExtZbbOrZbkb] in { def : PatGprGpr, ROL>; def : PatGprGpr, ROR>; def : PatGprImm; // There's no encoding for roli in the the 'B' extension as it can be // implemented with rori by negating the immediate. def : Pat<(rotl GPR:$rs1, uimmlog2xlen:$shamt), (RORI GPR:$rs1, (ImmSubFromXLen uimmlog2xlen:$shamt))>; } // Predicates = [HasStdExtZbbOrZbkb] let Predicates = [HasStdExtZbbOrZbkb, IsRV64] in { def : PatGprGpr, ROLW>; def : PatGprGpr, RORW>; def : PatGprImm; def : Pat<(riscv_rolw GPR:$rs1, uimm5:$rs2), (RORIW GPR:$rs1, (ImmSubFrom32 uimm5:$rs2))>; } // Predicates = [HasStdExtZbbOrZbkb, IsRV64] let Predicates = [HasStdExtZbs] in { def : Pat<(and (not (shiftop 1, GPR:$rs2)), GPR:$rs1), (BCLR GPR:$rs1, GPR:$rs2)>; def : Pat<(and (rotl -2, GPR:$rs2), GPR:$rs1), (BCLR GPR:$rs1, GPR:$rs2)>; def : Pat<(or (shiftop 1, GPR:$rs2), GPR:$rs1), (BSET GPR:$rs1, GPR:$rs2)>; def : Pat<(xor (shiftop 1, GPR:$rs2), GPR:$rs1), (BINV GPR:$rs1, GPR:$rs2)>; def : Pat<(and (shiftop GPR:$rs1, GPR:$rs2), 1), (BEXT GPR:$rs1, GPR:$rs2)>; def : Pat<(shiftop 1, GPR:$rs2), (BSET X0, GPR:$rs2)>; def : Pat<(and GPR:$rs1, BCLRMask:$mask), (BCLRI GPR:$rs1, BCLRMask:$mask)>; def : Pat<(or GPR:$rs1, SingleBitSetMask:$mask), (BSETI GPR:$rs1, SingleBitSetMask:$mask)>; def : Pat<(xor GPR:$rs1, SingleBitSetMask:$mask), (BINVI GPR:$rs1, SingleBitSetMask:$mask)>; def : Pat<(and (srl GPR:$rs1, uimmlog2xlen:$shamt), (XLenVT 1)), (BEXTI GPR:$rs1, uimmlog2xlen:$shamt)>; def : Pat<(seteq (and GPR:$rs1, SingleBitSetMask:$mask), 0), (BEXTI (XORI GPR:$rs1, -1), SingleBitSetMask:$mask)>; def : Pat<(or GPR:$r, BSETINVTwoBitsMask:$i), (BSETI (BSETI GPR:$r, (TrailingZeros BSETINVTwoBitsMask:$i)), (BSETINVTwoBitsMaskHigh BSETINVTwoBitsMask:$i))>; def : Pat<(xor GPR:$r, BSETINVTwoBitsMask:$i), (BINVI (BINVI GPR:$r, (TrailingZeros BSETINVTwoBitsMask:$i)), (BSETINVTwoBitsMaskHigh BSETINVTwoBitsMask:$i))>; def : Pat<(or GPR:$r, BSETINVORIMask:$i), (BSETI (ORI GPR:$r, (BSETINVORIMaskLow BSETINVORIMask:$i)), (BSETINVTwoBitsMaskHigh BSETINVORIMask:$i))>; def : Pat<(xor GPR:$r, BSETINVORIMask:$i), (BINVI (XORI GPR:$r, (BSETINVORIMaskLow BSETINVORIMask:$i)), (BSETINVTwoBitsMaskHigh BSETINVORIMask:$i))>; def : Pat<(and GPR:$r, BCLRITwoBitsMask:$i), (BCLRI (BCLRI GPR:$r, (BCLRITwoBitsMaskLow BCLRITwoBitsMask:$i)), (BCLRITwoBitsMaskHigh BCLRITwoBitsMask:$i))>; def : Pat<(and GPR:$r, BCLRIANDIMask:$i), (BCLRI (ANDI GPR:$r, (BCLRIANDIMaskLow BCLRIANDIMask:$i)), (BCLRITwoBitsMaskHigh BCLRIANDIMask:$i))>; } // Predicates = [HasStdExtZbs] let Predicates = [HasStdExtZbb] in { def : Pat<(riscv_orc_b GPR:$rs1), (ORC_B GPR:$rs1)>; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbkb] in { def : Pat<(riscv_brev8 GPR:$rs1), (BREV8 GPR:$rs1)>; } // Predicates = [HasStdExtZbkb] let Predicates = [HasStdExtZbkb, IsRV32] in { // We treat zip and unzip as separate instructions, so match it directly. def : Pat<(i32 (riscv_zip GPR:$rs1)), (ZIP_RV32 GPR:$rs1)>; def : Pat<(i32 (riscv_unzip GPR:$rs1)), (UNZIP_RV32 GPR:$rs1)>; } // Predicates = [HasStdExtZbkb, IsRV32] let Predicates = [HasStdExtZbb] in { def : PatGpr; def : PatGpr; def : PatGpr; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbb, IsRV64] in { def : PatGpr; def : PatGpr; def : Pat<(i64 (ctpop (i64 (zexti32 (i64 GPR:$rs1))))), (CPOPW GPR:$rs1)>; def : Pat<(i64 (riscv_absw GPR:$rs1)), (MAX GPR:$rs1, (SUBW X0, GPR:$rs1))>; } // Predicates = [HasStdExtZbb, IsRV64] let Predicates = [HasStdExtZbb] in { def : Pat<(sext_inreg GPR:$rs1, i8), (SEXT_B GPR:$rs1)>; def : Pat<(sext_inreg GPR:$rs1, i16), (SEXT_H GPR:$rs1)>; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbb] in { def : PatGprGpr; def : PatGprGpr; def : PatGprGpr; def : PatGprGpr; } // Predicates = [HasStdExtZbb] let Predicates = [HasStdExtZbbOrZbkb, IsRV32] in { def : Pat<(i32 (bswap GPR:$rs1)), (REV8_RV32 GPR:$rs1)>; } // Predicates = [HasStdExtZbbOrZbkb, IsRV32] let Predicates = [HasStdExtZbbOrZbkb, IsRV64] in { def : Pat<(i64 (bswap GPR:$rs1)), (REV8_RV64 GPR:$rs1)>; } // Predicates = [HasStdExtZbbOrZbkb, IsRV64] let Predicates = [HasStdExtZbkb] in { def : Pat<(or (and (shl GPR:$rs2, (XLenVT 8)), 0xFFFF), (zexti8 GPR:$rs1)), (PACKH GPR:$rs1, GPR:$rs2)>; def : Pat<(or (shl (zexti8 GPR:$rs2), (XLenVT 8)), (zexti8 GPR:$rs1)), (PACKH GPR:$rs1, GPR:$rs2)>; def : Pat<(and (or (shl GPR:$rs2, (XLenVT 8)), (zexti8 GPR:$rs1)), 0xFFFF), (PACKH GPR:$rs1, GPR:$rs2)>; } // Predicates = [HasStdExtZbkb] let Predicates = [HasStdExtZbkb, IsRV32] in def : Pat<(i32 (or (zexti16 GPR:$rs1), (shl GPR:$rs2, (i32 16)))), (PACK GPR:$rs1, GPR:$rs2)>; let Predicates = [HasStdExtZbkb, IsRV64] in { def : Pat<(i64 (or (zexti32 GPR:$rs1), (shl GPR:$rs2, (i64 32)))), (PACK GPR:$rs1, GPR:$rs2)>; def : Pat<(binop_allwusers (shl GPR:$rs2, (i64 16)), (zexti16 GPR:$rs1)), (PACKW GPR:$rs1, GPR:$rs2)>; def : Pat<(i64 (or (sext_inreg (shl GPR:$rs2, (i64 16)), i32), (zexti16 GPR:$rs1))), (PACKW GPR:$rs1, GPR:$rs2)>; } // Predicates = [HasStdExtZbkb, IsRV64] let Predicates = [HasStdExtZbb, IsRV32] in def : Pat<(i32 (and GPR:$rs, 0xFFFF)), (ZEXT_H_RV32 GPR:$rs)>; let Predicates = [HasStdExtZbb, IsRV64] in def : Pat<(i64 (and GPR:$rs, 0xFFFF)), (ZEXT_H_RV64 GPR:$rs)>; let Predicates = [HasStdExtZba] in { def : Pat<(add (shl GPR:$rs1, (XLenVT 1)), non_imm12:$rs2), (SH1ADD GPR:$rs1, GPR:$rs2)>; def : Pat<(add (shl GPR:$rs1, (XLenVT 2)), non_imm12:$rs2), (SH2ADD GPR:$rs1, GPR:$rs2)>; def : Pat<(add (shl GPR:$rs1, (XLenVT 3)), non_imm12:$rs2), (SH3ADD GPR:$rs1, GPR:$rs2)>; // More complex cases use a ComplexPattern. def : Pat<(add sh1add_op:$rs1, non_imm12:$rs2), (SH1ADD sh1add_op:$rs1, GPR:$rs2)>; def : Pat<(add sh2add_op:$rs1, non_imm12:$rs2), (SH2ADD sh2add_op:$rs1, GPR:$rs2)>; def : Pat<(add sh3add_op:$rs1, non_imm12:$rs2), (SH3ADD sh3add_op:$rs1, GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 6)), GPR:$rs2), (SH1ADD (SH1ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 10)), GPR:$rs2), (SH1ADD (SH2ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 18)), GPR:$rs2), (SH1ADD (SH3ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 12)), GPR:$rs2), (SH2ADD (SH1ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 20)), GPR:$rs2), (SH2ADD (SH2ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 36)), GPR:$rs2), (SH2ADD (SH3ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 24)), GPR:$rs2), (SH3ADD (SH1ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 40)), GPR:$rs2), (SH3ADD (SH2ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add (mul_oneuse GPR:$rs1, (XLenVT 72)), GPR:$rs2), (SH3ADD (SH3ADD GPR:$rs1, GPR:$rs1), GPR:$rs2)>; def : Pat<(add GPR:$r, CSImm12MulBy4:$i), (SH2ADD (ADDI X0, (SimmShiftRightBy2XForm CSImm12MulBy4:$i)), GPR:$r)>; def : Pat<(add GPR:$r, CSImm12MulBy8:$i), (SH3ADD (ADDI X0, (SimmShiftRightBy3XForm CSImm12MulBy8:$i)), GPR:$r)>; def : Pat<(mul GPR:$r, C3LeftShift:$i), (SLLI (SH1ADD GPR:$r, GPR:$r), (TrailingZeros C3LeftShift:$i))>; def : Pat<(mul GPR:$r, C5LeftShift:$i), (SLLI (SH2ADD GPR:$r, GPR:$r), (TrailingZeros C5LeftShift:$i))>; def : Pat<(mul GPR:$r, C9LeftShift:$i), (SLLI (SH3ADD GPR:$r, GPR:$r), (TrailingZeros C9LeftShift:$i))>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 11)), (SH1ADD (SH2ADD GPR:$r, GPR:$r), GPR:$r)>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 19)), (SH1ADD (SH3ADD GPR:$r, GPR:$r), GPR:$r)>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 13)), (SH2ADD (SH1ADD GPR:$r, GPR:$r), GPR:$r)>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 21)), (SH2ADD (SH2ADD GPR:$r, GPR:$r), GPR:$r)>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 37)), (SH2ADD (SH3ADD GPR:$r, GPR:$r), GPR:$r)>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 25)), (SH3ADD (SH1ADD GPR:$r, GPR:$r), GPR:$r)>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 41)), (SH3ADD (SH2ADD GPR:$r, GPR:$r), GPR:$r)>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 73)), (SH3ADD (SH3ADD GPR:$r, GPR:$r), GPR:$r)>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 27)), (SH1ADD (SH3ADD GPR:$r, GPR:$r), (SH3ADD GPR:$r, GPR:$r))>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 45)), (SH2ADD (SH3ADD GPR:$r, GPR:$r), (SH3ADD GPR:$r, GPR:$r))>; def : Pat<(mul_const_oneuse GPR:$r, (XLenVT 81)), (SH3ADD (SH3ADD GPR:$r, GPR:$r), (SH3ADD GPR:$r, GPR:$r))>; } // Predicates = [HasStdExtZba] let Predicates = [HasStdExtZba, IsRV64] in { def : Pat<(i64 (shl (and GPR:$rs1, 0xFFFFFFFF), uimm5:$shamt)), (SLLI_UW GPR:$rs1, uimm5:$shamt)>; // Match a shifted 0xffffffff mask. Use SRLI to clear the LSBs and SLLI_UW to // mask and shift. def : Pat<(i64 (and GPR:$rs1, Shifted32OnesMask:$mask)), (SLLI_UW (SRLI GPR:$rs1, Shifted32OnesMask:$mask), Shifted32OnesMask:$mask)>; def : Pat<(i64 (add (and GPR:$rs1, 0xFFFFFFFF), non_imm12:$rs2)), (ADD_UW GPR:$rs1, GPR:$rs2)>; def : Pat<(i64 (and GPR:$rs, 0xFFFFFFFF)), (ADD_UW GPR:$rs, X0)>; def : Pat<(i64 (add (shl (and GPR:$rs1, 0xFFFFFFFF), (i64 1)), non_imm12:$rs2)), (SH1ADD_UW GPR:$rs1, GPR:$rs2)>; def : Pat<(i64 (add (shl (and GPR:$rs1, 0xFFFFFFFF), (i64 2)), non_imm12:$rs2)), (SH2ADD_UW GPR:$rs1, GPR:$rs2)>; def : Pat<(i64 (add (shl (and GPR:$rs1, 0xFFFFFFFF), (i64 3)), non_imm12:$rs2)), (SH3ADD_UW GPR:$rs1, GPR:$rs2)>; def : Pat<(i64 (add (and (shl GPR:$rs1, (i64 1)), 0x1FFFFFFFF), non_imm12:$rs2)), (SH1ADD_UW GPR:$rs1, GPR:$rs2)>; def : Pat<(i64 (add (and (shl GPR:$rs1, (i64 2)), 0x3FFFFFFFF), non_imm12:$rs2)), (SH2ADD_UW GPR:$rs1, GPR:$rs2)>; def : Pat<(i64 (add (and (shl GPR:$rs1, (i64 3)), 0x7FFFFFFFF), non_imm12:$rs2)), (SH3ADD_UW GPR:$rs1, GPR:$rs2)>; // More complex cases use a ComplexPattern. def : Pat<(add sh1add_uw_op:$rs1, non_imm12:$rs2), (SH1ADD_UW sh1add_uw_op:$rs1, GPR:$rs2)>; def : Pat<(add sh2add_uw_op:$rs1, non_imm12:$rs2), (SH2ADD_UW sh2add_uw_op:$rs1, GPR:$rs2)>; def : Pat<(add sh3add_uw_op:$rs1, non_imm12:$rs2), (SH3ADD_UW sh3add_uw_op:$rs1, GPR:$rs2)>; def : Pat<(i64 (add (and GPR:$rs1, 0xFFFFFFFE), non_imm12:$rs2)), (SH1ADD (SRLIW GPR:$rs1, 1), GPR:$rs2)>; def : Pat<(i64 (add (and GPR:$rs1, 0xFFFFFFFC), non_imm12:$rs2)), (SH2ADD (SRLIW GPR:$rs1, 2), GPR:$rs2)>; def : Pat<(i64 (add (and GPR:$rs1, 0xFFFFFFF8), non_imm12:$rs2)), (SH3ADD (SRLIW GPR:$rs1, 3), GPR:$rs2)>; // Use SRLI to clear the LSBs and SHXADD_UW to mask and shift. def : Pat<(i64 (add (and GPR:$rs1, 0x1FFFFFFFE), non_imm12:$rs2)), (SH1ADD_UW (SRLI GPR:$rs1, 1), GPR:$rs2)>; def : Pat<(i64 (add (and GPR:$rs1, 0x3FFFFFFFC), non_imm12:$rs2)), (SH2ADD_UW (SRLI GPR:$rs1, 2), GPR:$rs2)>; def : Pat<(i64 (add (and GPR:$rs1, 0x7FFFFFFF8), non_imm12:$rs2)), (SH3ADD_UW (SRLI GPR:$rs1, 3), GPR:$rs2)>; def : Pat<(mul (binop_oneuse GPR:$r, 0xFFFFFFFF), C3LeftShiftUW:$i), (SH1ADD (SLLI_UW GPR:$r, (TrailingZeros C3LeftShiftUW:$i)), (SLLI_UW GPR:$r, (TrailingZeros C3LeftShiftUW:$i)))>; def : Pat<(mul (binop_oneuse GPR:$r, 0xFFFFFFFF), C5LeftShiftUW:$i), (SH2ADD (SLLI_UW GPR:$r, (TrailingZeros C5LeftShiftUW:$i)), (SLLI_UW GPR:$r, (TrailingZeros C5LeftShiftUW:$i)))>; def : Pat<(mul (binop_oneuse GPR:$r, 0xFFFFFFFF), C9LeftShiftUW:$i), (SH3ADD (SLLI_UW GPR:$r, (TrailingZeros C9LeftShiftUW:$i)), (SLLI_UW GPR:$r, (TrailingZeros C9LeftShiftUW:$i)))>; } // Predicates = [HasStdExtZba, IsRV64] let Predicates = [HasStdExtZbcOrZbkc] in { def : PatGprGpr; def : PatGprGpr; } // Predicates = [HasStdExtZbcOrZbkc] let Predicates = [HasStdExtZbc] in def : PatGprGpr; let Predicates = [HasStdExtZbkx] in { def : PatGprGpr; def : PatGprGpr; } // Predicates = [HasStdExtZbkx]