From ef96a8ecd6308cc31e43515c2ce0bae6ffbdbbc6 Mon Sep 17 00:00:00 2001 From: Vasileios Kalintiris Date: Mon, 26 Jan 2015 12:33:22 +0000 Subject: [PATCH] [mips] Enable arithmetic and binary operations for the i128 data type. Summary: This patch adds support for some operations that were missing from 128-bit integer types (add/sub/mul/sdiv/udiv... etc.). With these changes we can support the __int128_t and __uint128_t data types from C/C++. Depends on D7125 Reviewers: dsanders Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D7143 llvm-svn: 227089 --- llvm/lib/Target/Mips/Mips64InstrInfo.td | 10 +++++ llvm/lib/Target/Mips/MipsISelLowering.cpp | 54 +++++++++++++------------- llvm/lib/Target/Mips/MipsSEISelDAGToDAG.cpp | 28 ++++++++++++-- llvm/lib/Target/Mips/MipsSEISelLowering.cpp | 4 ++ llvm/test/CodeGen/Mips/llvm-ir/add.ll | 30 +++++++++++++++ llvm/test/CodeGen/Mips/llvm-ir/and.ll | 20 ++++++++++ llvm/test/CodeGen/Mips/llvm-ir/ashr.ll | 60 +++++++++++++++++++++++++++++ llvm/test/CodeGen/Mips/llvm-ir/lshr.ll | 55 ++++++++++++++++++++++++++ llvm/test/CodeGen/Mips/llvm-ir/mul.ll | 59 ++++++++++++++++++++-------- llvm/test/CodeGen/Mips/llvm-ir/or.ll | 20 ++++++++++ llvm/test/CodeGen/Mips/llvm-ir/sdiv.ll | 13 +++++++ llvm/test/CodeGen/Mips/llvm-ir/shl.ll | 55 ++++++++++++++++++++++++++ llvm/test/CodeGen/Mips/llvm-ir/srem.ll | 13 +++++++ llvm/test/CodeGen/Mips/llvm-ir/sub.ll | 29 ++++++++++++++ llvm/test/CodeGen/Mips/llvm-ir/udiv.ll | 13 +++++++ llvm/test/CodeGen/Mips/llvm-ir/urem.ll | 13 +++++++ llvm/test/CodeGen/Mips/llvm-ir/xor.ll | 20 ++++++++++ 17 files changed, 450 insertions(+), 46 deletions(-) diff --git a/llvm/lib/Target/Mips/Mips64InstrInfo.td b/llvm/lib/Target/Mips/Mips64InstrInfo.td index d32ac1b..9e1b9d0 100644 --- a/llvm/lib/Target/Mips/Mips64InstrInfo.td +++ b/llvm/lib/Target/Mips/Mips64InstrInfo.td @@ -498,6 +498,16 @@ def : MipsPat<(i64 (sext_inreg GPR64:$src, i32)), // bswap MipsPattern def : MipsPat<(bswap GPR64:$rt), (DSHD (DSBH GPR64:$rt))>; +// Carry pattern +def : MipsPat<(subc GPR64:$lhs, GPR64:$rhs), + (DSUBu GPR64:$lhs, GPR64:$rhs)>; +let AdditionalPredicates = [NotDSP] in { + def : MipsPat<(addc GPR64:$lhs, GPR64:$rhs), + (DADDu GPR64:$lhs, GPR64:$rhs)>; + def : MipsPat<(addc GPR64:$lhs, immSExt16:$imm), + (DADDiu GPR64:$lhs, imm:$imm)>; +} + // Octeon bbit0/bbit1 MipsPattern let Predicates = [HasMips64, HasCnMips] in { def : MipsPat<(brcond (i32 (seteq (and i64:$lhs, PowerOf2LO:$mask), 0)), bb:$dst), diff --git a/llvm/lib/Target/Mips/MipsISelLowering.cpp b/llvm/lib/Target/Mips/MipsISelLowering.cpp index d25f563..c6045fe 100644 --- a/llvm/lib/Target/Mips/MipsISelLowering.cpp +++ b/llvm/lib/Target/Mips/MipsISelLowering.cpp @@ -261,6 +261,9 @@ MipsTargetLowering::MipsTargetLowering(const MipsTargetMachine &TM, setOperationAction(ISD::LOAD, MVT::i64, Custom); setOperationAction(ISD::STORE, MVT::i64, Custom); setOperationAction(ISD::FP_TO_SINT, MVT::i64, Custom); + setOperationAction(ISD::SHL_PARTS, MVT::i64, Custom); + setOperationAction(ISD::SRA_PARTS, MVT::i64, Custom); + setOperationAction(ISD::SRL_PARTS, MVT::i64, Custom); } if (!Subtarget.isGP64bit()) { @@ -2017,10 +2020,11 @@ SDValue MipsTargetLowering::lowerATOMIC_FENCE(SDValue Op, SDValue MipsTargetLowering::lowerShiftLeftParts(SDValue Op, SelectionDAG &DAG) const { SDLoc DL(Op); + MVT VT = Subtarget.isGP64bit() ? MVT::i64 : MVT::i32; + SDValue Lo = Op.getOperand(0), Hi = Op.getOperand(1); SDValue Shamt = Op.getOperand(2); - - // if shamt < 32: + // if shamt < (VT.bits): // lo = (shl lo, shamt) // hi = (or (shl hi, shamt) (srl (srl lo, 1), ~shamt)) // else: @@ -2028,18 +2032,17 @@ SDValue MipsTargetLowering::lowerShiftLeftParts(SDValue Op, // hi = (shl lo, shamt[4:0]) SDValue Not = DAG.getNode(ISD::XOR, DL, MVT::i32, Shamt, DAG.getConstant(-1, MVT::i32)); - SDValue ShiftRight1Lo = DAG.getNode(ISD::SRL, DL, MVT::i32, Lo, - DAG.getConstant(1, MVT::i32)); - SDValue ShiftRightLo = DAG.getNode(ISD::SRL, DL, MVT::i32, ShiftRight1Lo, - Not); - SDValue ShiftLeftHi = DAG.getNode(ISD::SHL, DL, MVT::i32, Hi, Shamt); - SDValue Or = DAG.getNode(ISD::OR, DL, MVT::i32, ShiftLeftHi, ShiftRightLo); - SDValue ShiftLeftLo = DAG.getNode(ISD::SHL, DL, MVT::i32, Lo, Shamt); + SDValue ShiftRight1Lo = DAG.getNode(ISD::SRL, DL, VT, Lo, + DAG.getConstant(1, VT)); + SDValue ShiftRightLo = DAG.getNode(ISD::SRL, DL, VT, ShiftRight1Lo, Not); + SDValue ShiftLeftHi = DAG.getNode(ISD::SHL, DL, VT, Hi, Shamt); + SDValue Or = DAG.getNode(ISD::OR, DL, VT, ShiftLeftHi, ShiftRightLo); + SDValue ShiftLeftLo = DAG.getNode(ISD::SHL, DL, VT, Lo, Shamt); SDValue Cond = DAG.getNode(ISD::AND, DL, MVT::i32, Shamt, DAG.getConstant(0x20, MVT::i32)); - Lo = DAG.getNode(ISD::SELECT, DL, MVT::i32, Cond, - DAG.getConstant(0, MVT::i32), ShiftLeftLo); - Hi = DAG.getNode(ISD::SELECT, DL, MVT::i32, Cond, ShiftLeftLo, Or); + Lo = DAG.getNode(ISD::SELECT, DL, VT, Cond, + DAG.getConstant(0, VT), ShiftLeftLo); + Hi = DAG.getNode(ISD::SELECT, DL, VT, Cond, ShiftLeftLo, Or); SDValue Ops[2] = {Lo, Hi}; return DAG.getMergeValues(Ops, DL); @@ -2050,8 +2053,9 @@ SDValue MipsTargetLowering::lowerShiftRightParts(SDValue Op, SelectionDAG &DAG, SDLoc DL(Op); SDValue Lo = Op.getOperand(0), Hi = Op.getOperand(1); SDValue Shamt = Op.getOperand(2); + MVT VT = Subtarget.isGP64bit() ? MVT::i64 : MVT::i32; - // if shamt < 32: + // if shamt < (VT.bits): // lo = (or (shl (shl hi, 1), ~shamt) (srl lo, shamt)) // if isSRA: // hi = (sra hi, shamt) @@ -2066,21 +2070,19 @@ SDValue MipsTargetLowering::lowerShiftRightParts(SDValue Op, SelectionDAG &DAG, // hi = 0 SDValue Not = DAG.getNode(ISD::XOR, DL, MVT::i32, Shamt, DAG.getConstant(-1, MVT::i32)); - SDValue ShiftLeft1Hi = DAG.getNode(ISD::SHL, DL, MVT::i32, Hi, - DAG.getConstant(1, MVT::i32)); - SDValue ShiftLeftHi = DAG.getNode(ISD::SHL, DL, MVT::i32, ShiftLeft1Hi, Not); - SDValue ShiftRightLo = DAG.getNode(ISD::SRL, DL, MVT::i32, Lo, Shamt); - SDValue Or = DAG.getNode(ISD::OR, DL, MVT::i32, ShiftLeftHi, ShiftRightLo); - SDValue ShiftRightHi = DAG.getNode(IsSRA ? ISD::SRA : ISD::SRL, DL, MVT::i32, - Hi, Shamt); + SDValue ShiftLeft1Hi = DAG.getNode(ISD::SHL, DL, VT, Hi, + DAG.getConstant(1, VT)); + SDValue ShiftLeftHi = DAG.getNode(ISD::SHL, DL, VT, ShiftLeft1Hi, Not); + SDValue ShiftRightLo = DAG.getNode(ISD::SRL, DL, VT, Lo, Shamt); + SDValue Or = DAG.getNode(ISD::OR, DL, VT, ShiftLeftHi, ShiftRightLo); + SDValue ShiftRightHi = DAG.getNode(IsSRA ? ISD::SRA : ISD::SRL, + DL, VT, Hi, Shamt); SDValue Cond = DAG.getNode(ISD::AND, DL, MVT::i32, Shamt, DAG.getConstant(0x20, MVT::i32)); - SDValue Shift31 = DAG.getNode(ISD::SRA, DL, MVT::i32, Hi, - DAG.getConstant(31, MVT::i32)); - Lo = DAG.getNode(ISD::SELECT, DL, MVT::i32, Cond, ShiftRightHi, Or); - Hi = DAG.getNode(ISD::SELECT, DL, MVT::i32, Cond, - IsSRA ? Shift31 : DAG.getConstant(0, MVT::i32), - ShiftRightHi); + SDValue Shift31 = DAG.getNode(ISD::SRA, DL, VT, Hi, DAG.getConstant(31, VT)); + Lo = DAG.getNode(ISD::SELECT, DL, VT, Cond, ShiftRightHi, Or); + Hi = DAG.getNode(ISD::SELECT, DL, VT, Cond, + IsSRA ? Shift31 : DAG.getConstant(0, VT), ShiftRightHi); SDValue Ops[2] = {Lo, Hi}; return DAG.getMergeValues(Ops, DL); diff --git a/llvm/lib/Target/Mips/MipsSEISelDAGToDAG.cpp b/llvm/lib/Target/Mips/MipsSEISelDAGToDAG.cpp index 7096382..cb7ac40 100644 --- a/llvm/lib/Target/Mips/MipsSEISelDAGToDAG.cpp +++ b/llvm/lib/Target/Mips/MipsSEISelDAGToDAG.cpp @@ -236,13 +236,31 @@ SDNode *MipsSEDAGToDAGISel::selectAddESubE(unsigned MOp, SDValue InFlag, (Opc == ISD::SUBC || Opc == ISD::SUBE)) && "(ADD|SUB)E flag operand must come from (ADD|SUB)C/E insn"); + unsigned SLTuOp = Mips::SLTu, ADDuOp = Mips::ADDu; + if (Subtarget->isGP64bit()) { + SLTuOp = Mips::SLTu64; + ADDuOp = Mips::DADDu; + } + SDValue Ops[] = { CmpLHS, InFlag.getOperand(1) }; SDValue LHS = Node->getOperand(0), RHS = Node->getOperand(1); EVT VT = LHS.getValueType(); - SDNode *Carry = CurDAG->getMachineNode(Mips::SLTu, DL, VT, Ops); - SDNode *AddCarry = CurDAG->getMachineNode(Mips::ADDu, DL, VT, + SDNode *Carry = CurDAG->getMachineNode(SLTuOp, DL, VT, Ops); + + if (Subtarget->isGP64bit()) { + // On 64-bit targets, sltu produces an i64 but our backend currently says + // that SLTu64 produces an i32. We need to fix this in the long run but for + // now, just make the DAG type-correct by asserting the upper bits are zero. + Carry = CurDAG->getMachineNode(Mips::SUBREG_TO_REG, DL, VT, + CurDAG->getTargetConstant(0, VT), + SDValue(Carry, 0), + CurDAG->getTargetConstant(Mips::sub_32, VT)); + } + + SDNode *AddCarry = CurDAG->getMachineNode(ADDuOp, DL, VT, SDValue(Carry, 0), RHS); + return CurDAG->SelectNodeTo(Node, MOp, VT, MVT::Glue, LHS, SDValue(AddCarry, 0)); } @@ -641,7 +659,8 @@ std::pair MipsSEDAGToDAGISel::selectNode(SDNode *Node) { case ISD::SUBE: { SDValue InFlag = Node->getOperand(2); - Result = selectAddESubE(Mips::SUBu, InFlag, InFlag.getOperand(0), DL, Node); + unsigned Opc = Subtarget->isGP64bit() ? Mips::DSUBu : Mips::SUBu; + Result = selectAddESubE(Opc, InFlag, InFlag.getOperand(0), DL, Node); return std::make_pair(true, Result); } @@ -649,7 +668,8 @@ std::pair MipsSEDAGToDAGISel::selectNode(SDNode *Node) { if (Subtarget->hasDSP()) // Select DSP instructions, ADDSC and ADDWC. break; SDValue InFlag = Node->getOperand(2); - Result = selectAddESubE(Mips::ADDu, InFlag, InFlag.getValue(0), DL, Node); + unsigned Opc = Subtarget->isGP64bit() ? Mips::DADDu : Mips::ADDu; + Result = selectAddESubE(Opc, InFlag, InFlag.getValue(0), DL, Node); return std::make_pair(true, Result); } diff --git a/llvm/lib/Target/Mips/MipsSEISelLowering.cpp b/llvm/lib/Target/Mips/MipsSEISelLowering.cpp index 29aac2e..3f865af 100644 --- a/llvm/lib/Target/Mips/MipsSEISelLowering.cpp +++ b/llvm/lib/Target/Mips/MipsSEISelLowering.cpp @@ -122,6 +122,8 @@ MipsSETargetLowering::MipsSETargetLowering(const MipsTargetMachine &TM, setOperationAction(ISD::MUL, MVT::i64, Custom); if (Subtarget.isGP64bit()) { + setOperationAction(ISD::SMUL_LOHI, MVT::i64, Custom); + setOperationAction(ISD::UMUL_LOHI, MVT::i64, Custom); setOperationAction(ISD::MULHS, MVT::i64, Custom); setOperationAction(ISD::MULHU, MVT::i64, Custom); setOperationAction(ISD::SDIVREM, MVT::i64, Custom); @@ -200,6 +202,8 @@ MipsSETargetLowering::MipsSETargetLowering(const MipsTargetMachine &TM, if (Subtarget.hasMips64r6()) { // MIPS64r6 replaces the accumulator-based multiplies with a three register // instruction + setOperationAction(ISD::SMUL_LOHI, MVT::i64, Expand); + setOperationAction(ISD::UMUL_LOHI, MVT::i64, Expand); setOperationAction(ISD::MUL, MVT::i64, Legal); setOperationAction(ISD::MULHS, MVT::i64, Legal); setOperationAction(ISD::MULHU, MVT::i64, Legal); diff --git a/llvm/test/CodeGen/Mips/llvm-ir/add.ll b/llvm/test/CodeGen/Mips/llvm-ir/add.ll index 8890992..83774ed 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/add.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/add.ll @@ -83,3 +83,33 @@ entry: %r = add i64 %a, %b ret i64 %r } + +define signext i128 @add_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: add_i128: + + ; GP32: lw $[[T0:[0-9]+]], 28($sp) + ; GP32: addu $[[T1:[0-9]+]], $7, $[[T0]] + ; GP32: sltu $[[T2:[0-9]+]], $[[T1]], $[[T0]] + ; GP32: lw $[[T3:[0-9]+]], 24($sp) + ; GP32: addu $[[T4:[0-9]+]], $[[T2]], $[[T3]] + ; GP32: addu $[[T5:[0-9]+]], $6, $[[T4]] + ; GP32: sltu $[[T6:[0-9]+]], $[[T5]], $[[T3]] + ; GP32: lw $[[T7:[0-9]+]], 20($sp) + ; GP32: addu $[[T8:[0-9]+]], $[[T6]], $[[T7]] + ; GP32: lw $[[T9:[0-9]+]], 16($sp) + ; GP32: addu $3, $5, $[[T8]] + ; GP32: sltu $[[T10:[0-9]+]], $3, $[[T7]] + ; GP32: addu $[[T11:[0-9]+]], $[[T10]], $[[T9]] + ; GP32: addu $2, $4, $[[T11]] + ; GP32: move $4, $[[T5]] + ; GP32: move $5, $[[T1]] + + ; GP64: daddu $3, $5, $7 + ; GP64: sltu $[[T0:[0-9]+]], $3, $7 + ; GP64: daddu $[[T1:[0-9]+]], $[[T0]], $6 + ; GP64: daddu $2, $4, $[[T1]] + + %r = add i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/and.ll b/llvm/test/CodeGen/Mips/llvm-ir/and.ll index 3cb4957..09d0ef9 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/and.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/and.ll @@ -72,3 +72,23 @@ entry: %r = and i64 %a, %b ret i64 %r } + +define signext i128 @and_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: and_i128: + + ; GP32: lw $[[T0:[0-9]+]], 24($sp) + ; GP32: lw $[[T1:[0-9]+]], 20($sp) + ; GP32: lw $[[T2:[0-9]+]], 16($sp) + ; GP32: and $2, $4, $[[T2]] + ; GP32: and $3, $5, $[[T1]] + ; GP32: and $4, $6, $[[T0]] + ; GP32: lw $[[T3:[0-9]+]], 28($sp) + ; GP32: and $5, $7, $[[T3]] + + ; GP64: and $2, $4, $6 + ; GP64: and $3, $5, $7 + + %r = and i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/ashr.ll b/llvm/test/CodeGen/Mips/llvm-ir/ashr.ll index b3f023a..8866b8b 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/ashr.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/ashr.ll @@ -126,3 +126,63 @@ entry: %r = ashr i64 %a, %b ret i64 %r } + +define signext i128 @ashr_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: ashr_i128: + + ; GP32: lw $25, %call16(__ashrti3)($gp) + + ; M3: sll $[[T0:[0-9]+]], $7, 0 + ; M3: dsrav $[[T1:[0-9]+]], $4, $[[T0]] + ; M3: andi $[[T2:[0-9]+]], $[[T0]], 32 + ; M3: bnez $[[T3:[0-9]+]], $[[BB0:BB[0-9_]+]] + ; M3: move $3, $[[T1]] + ; M3: dsrlv $[[T4:[0-9]+]], $5, $[[T0]] + ; M3: dsll $[[T5:[0-9]+]], $4, 1 + ; M3: not $[[T6:[0-9]+]], $[[T0]] + ; M3: dsllv $[[T7:[0-9]+]], $[[T5]], $[[T6]] + ; M3: or $3, $[[T7]], $[[T4]] + ; M3: $[[BB0]]: + ; M3: beqz $[[T3]], $[[BB1:BB[0-9_]+]] + ; M3: nop + ; M3: dsra $2, $4, 31 + ; M3: $[[BB1]]: + ; M3: jr $ra + ; M3: nop + + ; GP64-NOT-R6: sll $[[T0:[0-9]+]], $7, 0 + ; GP64-NOT-R6: dsrlv $[[T1:[0-9]+]], $5, $[[T0]] + ; GP64-NOT-R6: dsll $[[T2:[0-9]+]], $4, 1 + ; GP64-NOT-R6: not $[[T3:[0-9]+]], $[[T0]] + ; GP64-NOT-R6: dsllv $[[T4:[0-9]+]], $[[T2]], $[[T3]] + ; GP64-NOT-R6: or $3, $[[T4]], $[[T1]] + ; GP64-NOT-R6: dsrav $2, $4, $[[T0]] + ; GP64-NOT-R6: andi $[[T5:[0-9]+]], $[[T0]], 32 + + ; GP64-NOT-R6: movn $3, $2, $[[T5]] + ; GP64-NOT-R6: dsra $[[T6:[0-9]+]], $4, 31 + ; GP64-NOT-R6: jr $ra + ; GP64-NOT-R6: movn $2, $[[T6]], $[[T5]] + + ; 64R6: sll $[[T0:[0-9]+]], $7, 0 + ; 64R6: dsrav $[[T1:[0-9]+]], $4, $[[T0]] + ; 64R6: andi $[[T2:[0-9]+]], $[[T0]], 32 + ; 64R6: sll $[[T3:[0-9]+]], $[[T2]], 0 + ; 64R6: seleqz $[[T4:[0-9]+]], $[[T1]], $[[T3]] + ; 64R6: dsra $[[T5:[0-9]+]], $4, 31 + ; 64R6: selnez $[[T6:[0-9]+]], $[[T5]], $[[T3]] + ; 64R6: or $2, $[[T6]], $[[T4]] + ; 64R6: dsrlv $[[T7:[0-9]+]], $5, $[[T0]] + ; 64R6: dsll $[[T8:[0-9]+]], $4, 1 + ; 64R6: not $[[T9:[0-9]+]], $[[T0]] + ; 64R6: dsllv $[[T10:[0-9]+]], $[[T8]], $[[T9]] + ; 64R6: or $[[T11:[0-9]+]], $[[T10]], $[[T7]] + ; 64R6: seleqz $[[T12:[0-9]+]], $[[T11]], $[[T3]] + ; 64R6: selnez $[[T13:[0-9]+]], $[[T1]], $[[T3]] + ; 64R6: jr $ra + ; 64R6: or $3, $[[T13]], $[[T12]] + + %r = ashr i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/lshr.ll b/llvm/test/CodeGen/Mips/llvm-ir/lshr.ll index 8975c8c..1679678 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/lshr.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/lshr.ll @@ -119,3 +119,58 @@ entry: %r = lshr i64 %a, %b ret i64 %r } + +define signext i128 @lshr_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: lshr_i128: + + ; GP32: lw $25, %call16(__lshrti3)($gp) + + ; M3: sll $[[T0:[0-9]+]], $7, 0 + ; M3: dsrlv $[[T1:[0-9]+]], $4, $[[T0]] + ; M3: andi $[[T2:[0-9]+]], $[[T0]], 32 + ; M3: bnez $[[T3:[0-9]+]], $[[BB0:BB[0-9_]+]] + ; M3: move $3, $[[T1]] + ; M3: dsrlv $[[T4:[0-9]+]], $5, $[[T0]] + ; M3: dsll $[[T5:[0-9]+]], $4, 1 + ; M3: not $[[T6:[0-9]+]], $[[T0]] + ; M3: dsllv $[[T7:[0-9]+]], $[[T5]], $[[T6]] + ; M3: or $3, $[[T7]], $[[T4]] + ; M3: $[[BB0]]: + ; M3: bnez $[[T3]], $[[BB1:BB[0-9_]+]] + ; M3: daddiu $2, $zero, 0 + ; M3: move $2, $[[T1]] + ; M3: $[[BB1]]: + ; M3: jr $ra + ; M3: nop + + ; GP64-NOT-R6: sll $[[T0:[0-9]+]], $7, 0 + ; GP64-NOT-R6: dsrlv $[[T1:[0-9]+]], $5, $[[T0]] + ; GP64-NOT-R6: dsll $[[T2:[0-9]+]], $4, 1 + ; GP64-NOT-R6: not $[[T3:[0-9]+]], $[[T0]] + ; GP64-NOT-R6: dsllv $[[T4:[0-9]+]], $[[T2]], $[[T3]] + ; GP64-NOT-R6: or $3, $[[T4]], $[[T1]] + ; GP64-NOT-R6: dsrlv $2, $4, $[[T0]] + ; GP64-NOT-R6: andi $[[T5:[0-9]+]], $[[T0]], 32 + ; GP64-NOT-R6: movn $3, $2, $[[T5]] + ; GP64-NOT-R6: jr $ra + ; GP64-NOT-R6: movn $2, $zero, $1 + + ; 64R6: sll $[[T0:[0-9]+]], $7, 0 + ; 64R6: dsrlv $[[T1:[0-9]+]], $5, $[[T0]] + ; 64R6: dsll $[[T2:[0-9]+]], $4, 1 + ; 64R6: not $[[T3:[0-9]+]], $[[T0]] + ; 64R6: dsllv $[[T4:[0-9]+]], $[[T2]], $[[T3]] + ; 64R6: or $[[T5:[0-9]+]], $[[T4]], $[[T1]] + ; 64R6: andi $[[T6:[0-9]+]], $[[T0]], 32 + ; 64R6: sll $[[T7:[0-9]+]], $[[T6]], 0 + ; 64R6: seleqz $[[T8:[0-9]+]], $[[T5]], $[[T7]] + ; 64R6: dsrlv $[[T9:[0-9]+]], $4, $[[T0]] + ; 64R6: selnez $[[T10:[0-9]+]], $[[T9]], $[[T7]] + ; 64R6: or $3, $[[T10]], $[[T8]] + ; 64R6: jr $ra + ; 64R6: seleqz $2, $[[T0]], $[[T7]] + + %r = lshr i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/mul.ll b/llvm/test/CodeGen/Mips/llvm-ir/mul.ll index 1674124..5f7f338 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/mul.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/mul.ll @@ -1,19 +1,19 @@ -; RUN: llc < %s -march=mips -mcpu=mips2 | FileCheck %s \ -; RUN: -check-prefix=ALL -check-prefix=M2 -; RUN: llc < %s -march=mips -mcpu=mips32 | FileCheck %s \ -; RUN: -check-prefix=ALL -check-prefix=32R1-R2 -check-prefix=32R1 -; RUN: llc < %s -march=mips -mcpu=mips32r2 | FileCheck %s \ -; RUN: -check-prefix=ALL -check-prefix=32R1-R2 -check-prefix=32R2 -; RUN: llc < %s -march=mips -mcpu=mips32r6 | FileCheck %s \ -; RUN: -check-prefix=ALL -check-prefix=32R6 -; RUN: llc < %s -march=mips64 -mcpu=mips4 | FileCheck %s \ -; RUN: -check-prefix=ALL -check-prefix=M4 -; RUN: llc < %s -march=mips64 -mcpu=mips64 | FileCheck %s \ -; RUN: -check-prefix=ALL -check-prefix=64R1-R2 -; RUN: llc < %s -march=mips64 -mcpu=mips64r2 | FileCheck %s \ -; RUN: -check-prefix=ALL -check-prefix=64R1-R2 -; RUN: llc < %s -march=mips64 -mcpu=mips64r6 | FileCheck %s \ -; RUN: -check-prefix=ALL -check-prefix=64R6 +; RUN: llc < %s -march=mips -mcpu=mips2 | FileCheck %s -check-prefix=ALL \ +; RUN: -check-prefix=M2 -check-prefix=GP32 +; RUN: llc < %s -march=mips -mcpu=mips32 | FileCheck %s -check-prefix=ALL \ +; RUN: -check-prefix=32R1-R2 -check-prefix=GP32 +; RUN: llc < %s -march=mips -mcpu=mips32r2 | FileCheck %s -check-prefix=ALL \ +; RUN: -check-prefix=32R1-R2 -check-prefix=32R2 -check-prefix=GP32 +; RUN: llc < %s -march=mips -mcpu=mips32r6 | FileCheck %s -check-prefix=ALL \ +; RUN: -check-prefix=32R6 -check-prefix=GP32 +; RUN: llc < %s -march=mips64 -mcpu=mips4 | FileCheck %s -check-prefix=ALL \ +; RUN: -check-prefix=M4 -check-prefix=GP64-NOT-R6 +; RUN: llc < %s -march=mips64 -mcpu=mips64 | FileCheck %s -check-prefix=ALL \ +; RUN: -check-prefix=64R1-R2 -check-prefix=GP64-NOT-R6 +; RUN: llc < %s -march=mips64 -mcpu=mips64r2 | FileCheck %s -check-prefix=ALL \ +; RUN: -check-prefix=64R1-R2 -check-prefix=GP64 -check-prefix=GP64-NOT-R6 +; RUN: llc < %s -march=mips64 -mcpu=mips64r6 | FileCheck %s -check-prefix=ALL \ +; RUN: -check-prefix=64R6 define signext i1 @mul_i1(i1 signext %a, i1 signext %b) { entry: @@ -179,3 +179,30 @@ entry: %r = mul i64 %a, %b ret i64 %r } + +define signext i128 @mul_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: mul_i128: + + ; GP32: lw $25, %call16(__multi3)($gp) + + ; GP64-NOT-R6: dmult $4, $7 + ; GP64-NOT-R6: mflo $[[T0:[0-9]+]] + ; GP64-NOT-R6: dmult $5, $6 + ; GP64-NOT-R6: mflo $[[T1:[0-9]+]] + ; GP64-NOT-R6: dmultu $5, $7 + ; GP64-NOT-R6: mflo $3 + ; GP64-NOT-R6: mfhi $[[T2:[0-9]+]] + ; GP64-NOT-R6: daddu $[[T3:[0-9]+]], $[[T2]], $[[T1]] + ; GP64-NOT-R6: daddu $2, $[[T3:[0-9]+]], $[[T0]] + + ; 64R6: dmul $[[T0:[0-9]+]], $5, $6 + ; 64R6: dmuhu $[[T1:[0-9]+]], $5, $7 + ; 64R6: daddu $[[T2:[0-9]+]], $[[T1]], $[[T0]] + ; 64R6: dmul $[[T3:[0-9]+]], $4, $7 + ; 64R6: daddu $2, $[[T2]], $[[T3]] + ; 64R6: dmul $3, $5, $7 + + %r = mul i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/or.ll b/llvm/test/CodeGen/Mips/llvm-ir/or.ll index d641b6e..21d1d4f 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/or.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/or.ll @@ -73,3 +73,23 @@ entry: %r = or i64 %a, %b ret i64 %r } + +define signext i128 @or_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: or_i128: + + ; GP32: lw $[[T0:[0-9]+]], 24($sp) + ; GP32: lw $[[T1:[0-9]+]], 20($sp) + ; GP32: lw $[[T2:[0-9]+]], 16($sp) + ; GP32: or $2, $4, $[[T2]] + ; GP32: or $3, $5, $[[T1]] + ; GP32: or $4, $6, $[[T0]] + ; GP32: lw $[[T3:[0-9]+]], 28($sp) + ; GP32: or $5, $7, $[[T3]] + + ; GP64: or $2, $4, $6 + ; GP64: or $3, $5, $7 + + %r = or i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/sdiv.ll b/llvm/test/CodeGen/Mips/llvm-ir/sdiv.ll index 9b93fcf..54b7f70 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/sdiv.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/sdiv.ll @@ -121,3 +121,16 @@ entry: %r = sdiv i64 %a, %b ret i64 %r } + +define signext i128 @sdiv_i128(i128 signext %a, i128 signext %b) { +entry: + ; ALL-LABEL: sdiv_i128: + + ; GP32: lw $25, %call16(__divti3)($gp) + + ; GP64-NOT-R6: ld $25, %call16(__divti3)($gp) + ; 64R6: ld $25, %call16(__divti3)($gp) + + %r = sdiv i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/shl.ll b/llvm/test/CodeGen/Mips/llvm-ir/shl.ll index a3948fe..072b44a 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/shl.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/shl.ll @@ -131,3 +131,58 @@ entry: %r = shl i64 %a, %b ret i64 %r } + +define signext i128 @shl_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: shl_i128: + + ; GP32: lw $25, %call16(__ashlti3)($gp) + + ; M3: sll $[[T0:[0-9]+]], $7, 0 + ; M3: dsllv $[[T1:[0-9]+]], $5, $[[T0]] + ; M3: andi $[[T2:[0-9]+]], $[[T0]], 32 + ; M3: bnez $[[T3:[0-9]+]], $[[BB0:BB[0-9_]+]] + ; M3: move $2, $[[T1]] + ; M3: dsllv $[[T4:[0-9]+]], $4, $[[T0]] + ; M3: dsrl $[[T5:[0-9]+]], $5, 1 + ; M3: not $[[T6:[0-9]+]], $[[T0]] + ; M3: dsrlv $[[T7:[0-9]+]], $[[T5]], $[[T6]] + ; M3: or $2, $[[T4]], $[[T7]] + ; M3: $[[BB0]]: + ; M3: bnez $[[T3]], $[[BB1:BB[0-9_]+]] + ; M3: daddiu $3, $zero, 0 + ; M3: move $3, $[[T1]] + ; M3: $[[BB1]]: + ; M3: jr $ra + ; M3: nop + + ; GP64-NOT-R6: sll $[[T0:[0-9]+]], $7, 0 + ; GP64-NOT-R6: dsllv $[[T1:[0-9]+]], $4, $[[T0]] + ; GP64-NOT-R6: dsrl $[[T2:[0-9]+]], $5, 1 + ; GP64-NOT-R6: not $[[T3:[0-9]+]], $[[T0]] + ; GP64-NOT-R6: dsrlv $[[T4:[0-9]+]], $[[T2]], $[[T3]] + ; GP64-NOT-R6: or $2, $[[T1]], $[[T4]] + ; GP64-NOT-R6: dsllv $3, $5, $[[T0]] + ; GP64-NOT-R6: andi $[[T5:[0-9]+]], $[[T0]], 32 + ; GP64-NOT-R6: movn $2, $3, $[[T5]] + ; GP64-NOT-R6: jr $ra + ; GP64-NOT-R6: movn $3, $zero, $1 + + ; 64R6: sll $[[T0:[0-9]+]], $7, 0 + ; 64R6: dsllv $[[T1:[0-9]+]], $4, $[[T0]] + ; 64R6: dsrl $[[T2:[0-9]+]], $5, 1 + ; 64R6: not $[[T3:[0-9]+]], $[[T0]] + ; 64R6: dsrlv $[[T4:[0-9]+]], $[[T2]], $[[T3]] + ; 64R6: or $[[T5:[0-9]+]], $[[T1]], $[[T4]] + ; 64R6: andi $[[T6:[0-9]+]], $[[T0]], 32 + ; 64R6: sll $[[T7:[0-9]+]], $[[T6]], 0 + ; 64R6: seleqz $[[T8:[0-9]+]], $[[T5]], $[[T7]] + ; 64R6: dsllv $[[T9:[0-9]+]], $5, $[[T0]] + ; 64R6: selnez $[[T10:[0-9]+]], $[[T9]], $[[T7]] + ; 64R6: or $2, $[[T10]], $[[T8]] + ; 64R6: jr $ra + ; 64R6: seleqz $3, $[[T0]], $[[T7]] + + %r = shl i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/srem.ll b/llvm/test/CodeGen/Mips/llvm-ir/srem.ll index 8ae7b07..1e949d2 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/srem.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/srem.ll @@ -114,3 +114,16 @@ entry: %r = srem i64 %a, %b ret i64 %r } + +define signext i128 @srem_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: srem_i128: + + ; GP32: lw $25, %call16(__modti3)($gp) + + ; GP64-NOT-R6: ld $25, %call16(__modti3)($gp) + ; 64-R6: ld $25, %call16(__modti3)($gp) + + %r = srem i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/sub.ll b/llvm/test/CodeGen/Mips/llvm-ir/sub.ll index 7f2238d..6d592be 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/sub.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/sub.ll @@ -83,3 +83,32 @@ entry: %r = sub i64 %a, %b ret i64 %r } + +define signext i128 @sub_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: sub_i128: + + ; GP32: lw $[[T0:[0-9]+]], 20($sp) + ; GP32: sltu $[[T1:[0-9]+]], $5, $[[T0]] + ; GP32: lw $[[T2:[0-9]+]], 16($sp) + ; GP32: addu $[[T3:[0-9]+]], $[[T1]], $[[T2]] + ; GP32: lw $[[T4:[0-9]+]], 24($sp) + ; GP32: lw $[[T5:[0-9]+]], 28($sp) + ; GP32: subu $[[T6:[0-9]+]], $7, $[[T5]] + ; GP32: subu $2, $4, $[[T3]] + ; GP32: sltu $[[T8:[0-9]+]], $6, $[[T4]] + ; GP32: addu $[[T9:[0-9]+]], $[[T8]], $[[T0]] + ; GP32: subu $3, $5, $[[T9]] + ; GP32: sltu $[[T10:[0-9]+]], $7, $[[T5]] + ; GP32: addu $[[T11:[0-9]+]], $[[T10]], $[[T4]] + ; GP32: subu $4, $6, $[[T11]] + ; GP32: move $5, $[[T6]] + + ; GP64: dsubu $3, $5, $7 + ; GP64: sltu $[[T0:[0-9]+]], $5, $7 + ; GP64: daddu $[[T1:[0-9]+]], $[[T0]], $6 + ; GP64: dsubu $2, $4, $[[T1]] + + %r = sub i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/udiv.ll b/llvm/test/CodeGen/Mips/llvm-ir/udiv.ll index 1603424..1f7aa0d 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/udiv.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/udiv.ll @@ -93,3 +93,16 @@ entry: %r = udiv i64 %a, %b ret i64 %r } + +define signext i128 @udiv_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: udiv_i128: + + ; GP32: lw $25, %call16(__udivti3)($gp) + + ; GP64-NOT-R6: ld $25, %call16(__udivti3)($gp) + ; 64-R6: ld $25, %call16(__udivti3)($gp) + + %r = udiv i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/urem.ll b/llvm/test/CodeGen/Mips/llvm-ir/urem.ll index 86c4605..7323534 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/urem.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/urem.ll @@ -130,3 +130,16 @@ entry: %r = urem i64 %a, %b ret i64 %r } + +define signext i128 @urem_i128(i128 signext %a, i128 signext %b) { +entry: + ; ALL-LABEL: urem_i128: + + ; GP32: lw $25, %call16(__umodti3)($gp) + + ; GP64-NOT-R6: ld $25, %call16(__umodti3)($gp) + ; 64-R6: ld $25, %call16(__umodti3)($gp) + + %r = urem i128 %a, %b + ret i128 %r +} diff --git a/llvm/test/CodeGen/Mips/llvm-ir/xor.ll b/llvm/test/CodeGen/Mips/llvm-ir/xor.ll index 69d3614..94dead1 100644 --- a/llvm/test/CodeGen/Mips/llvm-ir/xor.ll +++ b/llvm/test/CodeGen/Mips/llvm-ir/xor.ll @@ -72,3 +72,23 @@ entry: %r = xor i64 %a, %b ret i64 %r } + +define signext i128 @xor_i128(i128 signext %a, i128 signext %b) { +entry: +; ALL-LABEL: xor_i128: + + ; GP32: lw $[[T0:[0-9]+]], 24($sp) + ; GP32: lw $[[T1:[0-9]+]], 20($sp) + ; GP32: lw $[[T2:[0-9]+]], 16($sp) + ; GP32: xor $2, $4, $[[T2]] + ; GP32: xor $3, $5, $[[T1]] + ; GP32: xor $4, $6, $[[T0]] + ; GP32: lw $[[T3:[0-9]+]], 28($sp) + ; GP32: xor $5, $7, $[[T3]] + + ; GP64: xor $2, $4, $6 + ; GP64: xor $3, $5, $7 + + %r = xor i128 %a, %b + ret i128 %r +} -- 2.7.4