diff options
Diffstat (limited to 'lib/Target/X86/X86InstrInfo.td')
| -rw-r--r-- | lib/Target/X86/X86InstrInfo.td | 802 |
1 files changed, 748 insertions, 54 deletions
diff --git a/lib/Target/X86/X86InstrInfo.td b/lib/Target/X86/X86InstrInfo.td index 03df10db61c5..30b57d85d012 100644 --- a/lib/Target/X86/X86InstrInfo.td +++ b/lib/Target/X86/X86InstrInfo.td @@ -56,6 +56,10 @@ def SDT_X86CallSeqEnd : SDCallSeqEnd<[SDTCisVT<0, i32>, def SDT_X86Call : SDTypeProfile<0, -1, [SDTCisVT<0, iPTR>]>; +def SDT_X86VASTART_SAVE_XMM_REGS : SDTypeProfile<0, -1, [SDTCisVT<0, i8>, + SDTCisVT<1, iPTR>, + SDTCisVT<2, iPTR>]>; + def SDTX86RepStr : SDTypeProfile<0, 1, [SDTCisVT<0, OtherVT>]>; def SDTX86RdTsc : SDTypeProfile<0, 0, []>; @@ -114,6 +118,11 @@ def X86AtomSwap64 : SDNode<"X86ISD::ATOMSWAP64_DAG", SDTX86atomicBinary, def X86retflag : SDNode<"X86ISD::RET_FLAG", SDTX86Ret, [SDNPHasChain, SDNPOptInFlag]>; +def X86vastart_save_xmm_regs : + SDNode<"X86ISD::VASTART_SAVE_XMM_REGS", + SDT_X86VASTART_SAVE_XMM_REGS, + [SDNPHasChain]>; + def X86callseq_start : SDNode<"ISD::CALLSEQ_START", SDT_X86CallSeqStart, [SDNPHasChain, SDNPOutFlag]>; @@ -124,9 +133,6 @@ def X86callseq_end : def X86call : SDNode<"X86ISD::CALL", SDT_X86Call, [SDNPHasChain, SDNPOutFlag, SDNPOptInFlag]>; -def X86tailcall: SDNode<"X86ISD::TAILCALL", SDT_X86Call, - [SDNPHasChain, SDNPOutFlag, SDNPOptInFlag]>; - def X86rep_stos: SDNode<"X86ISD::REP_STOS", SDTX86RepStr, [SDNPHasChain, SDNPInFlag, SDNPOutFlag, SDNPMayStore]>; def X86rep_movs: SDNode<"X86ISD::REP_MOVS", SDTX86RepStr, @@ -156,6 +162,9 @@ def X86smul_flag : SDNode<"X86ISD::SMUL", SDTBinaryArithWithFlags>; def X86umul_flag : SDNode<"X86ISD::UMUL", SDTUnaryArithWithFlags>; def X86inc_flag : SDNode<"X86ISD::INC", SDTUnaryArithWithFlags>; def X86dec_flag : SDNode<"X86ISD::DEC", SDTUnaryArithWithFlags>; +def X86or_flag : SDNode<"X86ISD::OR", SDTBinaryArithWithFlags>; +def X86xor_flag : SDNode<"X86ISD::XOR", SDTBinaryArithWithFlags>; +def X86and_flag : SDNode<"X86ISD::AND", SDTBinaryArithWithFlags>; def X86mul_imm : SDNode<"X86ISD::MUL_IMM", SDTIntBinOp>; @@ -167,57 +176,80 @@ def i32imm_pcrel : Operand<i32> { let PrintMethod = "print_pcrel_imm"; } +// A version of ptr_rc which excludes SP, ESP, and RSP. This is used for +// the index operand of an address, to conform to x86 encoding restrictions. +def ptr_rc_nosp : PointerLikeRegClass<1>; // *mem - Operand definitions for the funky X86 addressing mode operands. // +def X86MemAsmOperand : AsmOperandClass { + let Name = "Mem"; + let SuperClass = ?; +} class X86MemOperand<string printMethod> : Operand<iPTR> { let PrintMethod = printMethod; - let MIOperandInfo = (ops ptr_rc, i8imm, ptr_rc, i32imm, i8imm); + let MIOperandInfo = (ops ptr_rc, i8imm, ptr_rc_nosp, i32imm, i8imm); + let ParserMatchClass = X86MemAsmOperand; } +def opaque32mem : X86MemOperand<"printopaquemem">; +def opaque48mem : X86MemOperand<"printopaquemem">; +def opaque80mem : X86MemOperand<"printopaquemem">; + def i8mem : X86MemOperand<"printi8mem">; def i16mem : X86MemOperand<"printi16mem">; def i32mem : X86MemOperand<"printi32mem">; def i64mem : X86MemOperand<"printi64mem">; def i128mem : X86MemOperand<"printi128mem">; -def i256mem : X86MemOperand<"printi256mem">; +//def i256mem : X86MemOperand<"printi256mem">; def f32mem : X86MemOperand<"printf32mem">; def f64mem : X86MemOperand<"printf64mem">; def f80mem : X86MemOperand<"printf80mem">; def f128mem : X86MemOperand<"printf128mem">; -def f256mem : X86MemOperand<"printf256mem">; +//def f256mem : X86MemOperand<"printf256mem">; // A version of i8mem for use on x86-64 that uses GR64_NOREX instead of // plain GR64, so that it doesn't potentially require a REX prefix. def i8mem_NOREX : Operand<i64> { let PrintMethod = "printi8mem"; - let MIOperandInfo = (ops GR64_NOREX, i8imm, GR64_NOREX, i32imm, i8imm); + let MIOperandInfo = (ops GR64_NOREX, i8imm, GR64_NOREX_NOSP, i32imm, i8imm); + let ParserMatchClass = X86MemAsmOperand; } def lea32mem : Operand<i32> { let PrintMethod = "printlea32mem"; - let MIOperandInfo = (ops GR32, i8imm, GR32, i32imm); + let MIOperandInfo = (ops GR32, i8imm, GR32_NOSP, i32imm); + let ParserMatchClass = X86MemAsmOperand; } def SSECC : Operand<i8> { let PrintMethod = "printSSECC"; } -def piclabel: Operand<i32> { - let PrintMethod = "printPICLabel"; +def ImmSExt8AsmOperand : AsmOperandClass { + let Name = "ImmSExt8"; + let SuperClass = ImmAsmOperand; } // A couple of more descriptive operand definitions. // 16-bits but only 8 bits are significant. -def i16i8imm : Operand<i16>; +def i16i8imm : Operand<i16> { + let ParserMatchClass = ImmSExt8AsmOperand; +} // 32-bits but only 8 bits are significant. -def i32i8imm : Operand<i32>; +def i32i8imm : Operand<i32> { + let ParserMatchClass = ImmSExt8AsmOperand; +} // Branch targets have OtherVT type and print as pc-relative values. def brtarget : Operand<OtherVT> { let PrintMethod = "print_pcrel_imm"; } +def brtarget8 : Operand<OtherVT> { + let PrintMethod = "print_pcrel_imm"; +} + //===----------------------------------------------------------------------===// // X86 Complex Pattern Definitions. // @@ -225,7 +257,8 @@ def brtarget : Operand<OtherVT> { // Define X86 specific addressing mode. def addr : ComplexPattern<iPTR, 5, "SelectAddr", [], []>; def lea32addr : ComplexPattern<i32, 4, "SelectLEAAddr", - [add, sub, mul, shl, or, frameindex], []>; + [add, sub, mul, X86mul_imm, shl, or, frameindex], + []>; def tls32addr : ComplexPattern<i32, 4, "SelectTLSADDRAddr", [tglobaltlsaddr], []>; @@ -246,8 +279,14 @@ def FPStackf32 : Predicate<"!Subtarget->hasSSE1()">; def FPStackf64 : Predicate<"!Subtarget->hasSSE2()">; def In32BitMode : Predicate<"!Subtarget->is64Bit()">; def In64BitMode : Predicate<"Subtarget->is64Bit()">; +def IsWin64 : Predicate<"Subtarget->isTargetWin64()">; +def NotWin64 : Predicate<"!Subtarget->isTargetWin64()">; def SmallCode : Predicate<"TM.getCodeModel() == CodeModel::Small">; -def NotSmallCode : Predicate<"TM.getCodeModel() != CodeModel::Small">; +def KernelCode : Predicate<"TM.getCodeModel() == CodeModel::Kernel">; +def FarData : Predicate<"TM.getCodeModel() != CodeModel::Small &&" + "TM.getCodeModel() != CodeModel::Kernel">; +def NearData : Predicate<"TM.getCodeModel() == CodeModel::Small ||" + "TM.getCodeModel() == CodeModel::Kernel">; def IsStatic : Predicate<"TM.getRelocationModel() == Reloc::Static">; def OptForSpeed : Predicate<"!OptForSize">; def FastBTMem : Predicate<"!Subtarget->isBTMemSlow()">; @@ -484,15 +523,35 @@ def ADJCALLSTACKUP32 : I<0, Pseudo, (outs), (ins i32imm:$amt1, i32imm:$amt2), Requires<[In32BitMode]>; } +// x86-64 va_start lowering magic. +let usesCustomDAGSchedInserter = 1 in +def VASTART_SAVE_XMM_REGS : I<0, Pseudo, + (outs), + (ins GR8:$al, + i64imm:$regsavefi, i64imm:$offset, + variable_ops), + "#VASTART_SAVE_XMM_REGS $al, $regsavefi, $offset", + [(X86vastart_save_xmm_regs GR8:$al, + imm:$regsavefi, + imm:$offset)]>; + // Nop -let neverHasSideEffects = 1 in +let neverHasSideEffects = 1 in { def NOOP : I<0x90, RawFrm, (outs), (ins), "nop", []>; + def NOOPL : I<0x1f, MRM0m, (outs), (ins i32mem:$zero), + "nopl\t$zero", []>, TB; +} -// PIC base +// Trap +def INT3 : I<0xcc, RawFrm, (outs), (ins), "int 3", []>; +def INT : I<0xcd, RawFrm, (outs), (ins i8imm:$trap), "int\t$trap", []>; + +// PIC base construction. This expands to code that looks like this: +// call $next_inst +// popl %destreg" let neverHasSideEffects = 1, isNotDuplicable = 1, Uses = [ESP] in - def MOVPC32r : Ii32<0xE8, Pseudo, (outs GR32:$reg), (ins piclabel:$label), - "call\t$label\n\t" - "pop{l}\t$reg", []>; + def MOVPC32r : Ii32<0xE8, Pseudo, (outs GR32:$reg), (ins i32imm:$label), + "", []>; //===----------------------------------------------------------------------===// // Control Flow Instructions... @@ -506,7 +565,11 @@ let isTerminator = 1, isReturn = 1, isBarrier = 1, [(X86retflag 0)]>; def RETI : Ii16<0xC2, RawFrm, (outs), (ins i16imm:$amt, variable_ops), "ret\t$amt", - [(X86retflag imm:$amt)]>; + [(X86retflag timm:$amt)]>; + def LRET : I <0xCB, RawFrm, (outs), (ins), + "lret", []>; + def LRETI : Ii16<0xCA, RawFrm, (outs), (ins i16imm:$amt), + "lret\t$amt", []>; } // All branches are RawFrm, Void, Branch, and Terminators @@ -514,8 +577,10 @@ let isBranch = 1, isTerminator = 1 in class IBr<bits<8> opcode, dag ins, string asm, list<dag> pattern> : I<opcode, RawFrm, (outs), ins, asm, pattern>; -let isBranch = 1, isBarrier = 1 in +let isBranch = 1, isBarrier = 1 in { def JMP : IBr<0xE9, (ins brtarget:$dst), "jmp\t$dst", [(br bb:$dst)]>; + def JMP8 : IBr<0xEB, (ins brtarget8:$dst), "jmp\t$dst", []>; +} // Indirect branches let isBranch = 1, isTerminator = 1, isBarrier = 1, isIndirectBranch = 1 in { @@ -523,10 +588,42 @@ let isBranch = 1, isTerminator = 1, isBarrier = 1, isIndirectBranch = 1 in { [(brind GR32:$dst)]>; def JMP32m : I<0xFF, MRM4m, (outs), (ins i32mem:$dst), "jmp{l}\t{*}$dst", [(brind (loadi32 addr:$dst))]>; + + def FARJMP16i : Iseg16<0xEA, RawFrm, (outs), + (ins i16imm:$seg, i16imm:$off), + "ljmp{w}\t$seg, $off", []>, OpSize; + def FARJMP32i : Iseg32<0xEA, RawFrm, (outs), + (ins i16imm:$seg, i32imm:$off), + "ljmp{l}\t$seg, $off", []>; + + def FARJMP16m : I<0xFF, MRM5m, (outs), (ins opaque32mem:$dst), + "ljmp{w}\t{*}$dst", []>, OpSize; + def FARJMP32m : I<0xFF, MRM5m, (outs), (ins opaque48mem:$dst), + "ljmp{l}\t{*}$dst", []>; } // Conditional branches let Uses = [EFLAGS] in { +// Short conditional jumps +def JO8 : IBr<0x70, (ins brtarget8:$dst), "jo\t$dst", []>; +def JNO8 : IBr<0x71, (ins brtarget8:$dst), "jno\t$dst", []>; +def JB8 : IBr<0x72, (ins brtarget8:$dst), "jb\t$dst", []>; +def JAE8 : IBr<0x73, (ins brtarget8:$dst), "jae\t$dst", []>; +def JE8 : IBr<0x74, (ins brtarget8:$dst), "je\t$dst", []>; +def JNE8 : IBr<0x75, (ins brtarget8:$dst), "jne\t$dst", []>; +def JBE8 : IBr<0x76, (ins brtarget8:$dst), "jbe\t$dst", []>; +def JA8 : IBr<0x77, (ins brtarget8:$dst), "ja\t$dst", []>; +def JS8 : IBr<0x78, (ins brtarget8:$dst), "js\t$dst", []>; +def JNS8 : IBr<0x79, (ins brtarget8:$dst), "jns\t$dst", []>; +def JP8 : IBr<0x7A, (ins brtarget8:$dst), "jp\t$dst", []>; +def JNP8 : IBr<0x7B, (ins brtarget8:$dst), "jnp\t$dst", []>; +def JL8 : IBr<0x7C, (ins brtarget8:$dst), "jl\t$dst", []>; +def JGE8 : IBr<0x7D, (ins brtarget8:$dst), "jge\t$dst", []>; +def JLE8 : IBr<0x7E, (ins brtarget8:$dst), "jle\t$dst", []>; +def JG8 : IBr<0x7F, (ins brtarget8:$dst), "jg\t$dst", []>; + +def JCXZ8 : IBr<0xE3, (ins brtarget8:$dst), "jcxz\t$dst", []>; + def JE : IBr<0x84, (ins brtarget:$dst), "je\t$dst", [(X86brcond bb:$dst, X86_COND_E, EFLAGS)]>, TB; def JNE : IBr<0x85, (ins brtarget:$dst), "jne\t$dst", @@ -563,6 +660,12 @@ def JNO : IBr<0x81, (ins brtarget:$dst), "jno\t$dst", [(X86brcond bb:$dst, X86_COND_NO, EFLAGS)]>, TB; } // Uses = [EFLAGS] +// Loop instructions + +def LOOP : I<0xE2, RawFrm, (ins brtarget8:$dst), (outs), "loop\t$dst", []>; +def LOOPE : I<0xE1, RawFrm, (ins brtarget8:$dst), (outs), "loope\t$dst", []>; +def LOOPNE : I<0xE0, RawFrm, (ins brtarget8:$dst), (outs), "loopne\t$dst", []>; + //===----------------------------------------------------------------------===// // Call Instructions... // @@ -583,13 +686,26 @@ let isCall = 1 in "call\t{*}$dst", [(X86call GR32:$dst)]>; def CALL32m : I<0xFF, MRM2m, (outs), (ins i32mem:$dst, variable_ops), "call\t{*}$dst", [(X86call (loadi32 addr:$dst))]>; + + def FARCALL16i : Iseg16<0x9A, RawFrm, (outs), + (ins i16imm:$seg, i16imm:$off), + "lcall{w}\t$seg, $off", []>, OpSize; + def FARCALL32i : Iseg32<0x9A, RawFrm, (outs), + (ins i16imm:$seg, i32imm:$off), + "lcall{l}\t$seg, $off", []>; + + def FARCALL16m : I<0xFF, MRM3m, (outs), (ins opaque32mem:$dst), + "lcall{w}\t{*}$dst", []>, OpSize; + def FARCALL32m : I<0xFF, MRM3m, (outs), (ins opaque48mem:$dst), + "lcall{l}\t{*}$dst", []>; } -// Tail call stuff. +// Constructing a stack frame. + +def ENTER : I<0xC8, RawFrm, (outs), (ins i16imm:$len, i8imm:$lvl), + "enter\t$len, $lvl", []>; -def TAILCALL : I<0, Pseudo, (outs), (ins), - "#TAILCALL", - []>; +// Tail call stuff. let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1 in def TCRETURNdi : I<0, Pseudo, (outs), (ins i32imm:$dst, i32imm:$offset, variable_ops), @@ -620,11 +736,29 @@ def LEAVE : I<0xC9, RawFrm, (outs), (ins), "leave", []>; let Defs = [ESP], Uses = [ESP], neverHasSideEffects=1 in { -let mayLoad = 1 in -def POP32r : I<0x58, AddRegFrm, (outs GR32:$reg), (ins), "pop{l}\t$reg", []>; +let mayLoad = 1 in { +def POP16r : I<0x58, AddRegFrm, (outs GR16:$reg), (ins), "pop{w}\t$reg", []>, + OpSize; +def POP32r : I<0x58, AddRegFrm, (outs GR32:$reg), (ins), "pop{l}\t$reg", []>; +def POP16rmr: I<0x8F, MRM0r, (outs GR16:$reg), (ins), "pop{w}\t$reg", []>, + OpSize; +def POP16rmm: I<0x8F, MRM0m, (outs i16mem:$dst), (ins), "pop{w}\t$dst", []>, + OpSize; +def POP32rmr: I<0x8F, MRM0r, (outs GR32:$reg), (ins), "pop{l}\t$reg", []>; +def POP32rmm: I<0x8F, MRM0m, (outs i32mem:$dst), (ins), "pop{l}\t$dst", []>; +} -let mayStore = 1 in +let mayStore = 1 in { +def PUSH16r : I<0x50, AddRegFrm, (outs), (ins GR16:$reg), "push{w}\t$reg",[]>, + OpSize; def PUSH32r : I<0x50, AddRegFrm, (outs), (ins GR32:$reg), "push{l}\t$reg",[]>; +def PUSH16rmr: I<0xFF, MRM6r, (outs), (ins GR16:$reg), "push{w}\t$reg",[]>, + OpSize; +def PUSH16rmm: I<0xFF, MRM6m, (outs), (ins i16mem:$src), "push{w}\t$src",[]>, + OpSize; +def PUSH32rmr: I<0xFF, MRM6r, (outs), (ins GR32:$reg), "push{l}\t$reg",[]>; +def PUSH32rmm: I<0xFF, MRM6m, (outs), (ins i32mem:$src), "push{l}\t$src",[]>; +} } let Defs = [ESP], Uses = [ESP], neverHasSideEffects = 1, mayStore = 1 in { @@ -710,6 +844,14 @@ let Defs = [ECX,EDI], Uses = [EAX,ECX,EDI] in def REP_STOSD : I<0xAB, RawFrm, (outs), (ins), "{rep;stosl|rep stosd}", [(X86rep_stos i32)]>, REP; +def SCAS8 : I<0xAE, RawFrm, (outs), (ins), "scas{b}", []>; +def SCAS16 : I<0xAF, RawFrm, (outs), (ins), "scas{w}", []>, OpSize; +def SCAS32 : I<0xAF, RawFrm, (outs), (ins), "scas{l}", []>; + +def CMPS8 : I<0xA6, RawFrm, (outs), (ins), "cmps{b}", []>; +def CMPS16 : I<0xA7, RawFrm, (outs), (ins), "cmps{w}", []>, OpSize; +def CMPS32 : I<0xA7, RawFrm, (outs), (ins), "cmps{l}", []>; + let Defs = [RAX, RDX] in def RDTSC : I<0x31, RawFrm, (outs), (ins), "rdtsc", [(X86rdtsc)]>, TB; @@ -718,6 +860,18 @@ let isBarrier = 1, hasCtrlDep = 1 in { def TRAP : I<0x0B, RawFrm, (outs), (ins), "ud2", [(trap)]>, TB; } +def SYSCALL : I<0x05, RawFrm, + (outs), (ins), "syscall", []>, TB; +def SYSRET : I<0x07, RawFrm, + (outs), (ins), "sysret", []>, TB; +def SYSENTER : I<0x34, RawFrm, + (outs), (ins), "sysenter", []>, TB; +def SYSEXIT : I<0x35, RawFrm, + (outs), (ins), "sysexit", []>, TB; + +def WAIT : I<0x9B, RawFrm, (outs), (ins), "wait", []>; + + //===----------------------------------------------------------------------===// // Input/Output Instructions... // @@ -793,6 +947,30 @@ def MOV32mi : Ii32<0xC7, MRM0m, (outs), (ins i32mem:$dst, i32imm:$src), "mov{l}\t{$src, $dst|$dst, $src}", [(store (i32 imm:$src), addr:$dst)]>; +def MOV8o8a : Ii8 <0xA0, RawFrm, (outs), (ins i8imm:$src), + "mov{b}\t{$src, %al|%al, $src}", []>; +def MOV16o16a : Ii16 <0xA1, RawFrm, (outs), (ins i16imm:$src), + "mov{w}\t{$src, %ax|%ax, $src}", []>, OpSize; +def MOV32o32a : Ii32 <0xA1, RawFrm, (outs), (ins i32imm:$src), + "mov{l}\t{$src, %eax|%eax, $src}", []>; + +def MOV8ao8 : Ii8 <0xA2, RawFrm, (outs i8imm:$dst), (ins), + "mov{b}\t{%al, $dst|$dst, %al}", []>; +def MOV16ao16 : Ii16 <0xA3, RawFrm, (outs i16imm:$dst), (ins), + "mov{w}\t{%ax, $dst|$dst, %ax}", []>, OpSize; +def MOV32ao32 : Ii32 <0xA3, RawFrm, (outs i32imm:$dst), (ins), + "mov{l}\t{%eax, $dst|$dst, %eax}", []>; + +// Moves to and from segment registers +def MOV16rs : I<0x8C, MRMDestReg, (outs GR16:$dst), (ins SEGMENT_REG:$src), + "mov{w}\t{$src, $dst|$dst, $src}", []>; +def MOV16ms : I<0x8C, MRMDestMem, (outs i16mem:$dst), (ins SEGMENT_REG:$src), + "mov{w}\t{$src, $dst|$dst, $src}", []>; +def MOV16sr : I<0x8E, MRMSrcReg, (outs SEGMENT_REG:$dst), (ins GR16:$src), + "mov{w}\t{$src, $dst|$dst, $src}", []>; +def MOV16sm : I<0x8E, MRMSrcMem, (outs SEGMENT_REG:$dst), (ins i16mem:$src), + "mov{w}\t{$src, $dst|$dst, $src}", []>; + let canFoldAsLoad = 1, isReMaterializable = 1, mayHaveSideEffects = 1 in { def MOV8rm : I<0x8A, MRMSrcMem, (outs GR8 :$dst), (ins i8mem :$src), "mov{b}\t{$src, $dst|$dst, $src}", @@ -950,6 +1128,20 @@ let isTwoAddress = 1 in { // Conditional moves let Uses = [EFLAGS] in { + +// X86 doesn't have 8-bit conditional moves. Use a customDAGSchedInserter to +// emit control flow. An alternative to this is to mark i8 SELECT as Promote, +// however that requires promoting the operands, and can induce additional +// i8 register pressure. Note that CMOV_GR8 is conservatively considered to +// clobber EFLAGS, because if one of the operands is zero, the expansion +// could involve an xor. +let usesCustomDAGSchedInserter = 1, isTwoAddress = 0, Defs = [EFLAGS] in +def CMOV_GR8 : I<0, Pseudo, + (outs GR8:$dst), (ins GR8:$src1, GR8:$src2, i8imm:$cond), + "#CMOV_GR8 PSEUDO!", + [(set GR8:$dst, (X86cmov GR8:$src1, GR8:$src2, + imm:$cond, EFLAGS))]>; + let isCommutable = 1 in { def CMOVB16rr : I<0x42, MRMSrcReg, // if <u, GR16 = GR16 (outs GR16:$dst), (ins GR16:$src1, GR16:$src2), @@ -1549,6 +1741,14 @@ let isTwoAddress = 0 in { "and{l}\t{$src, $dst|$dst, $src}", [(store (and (load addr:$dst), i32immSExt8:$src), addr:$dst), (implicit EFLAGS)]>; + + def AND8i8 : Ii8<0x24, RawFrm, (outs), (ins i8imm:$src), + "and{b}\t{$src, %al|%al, $src}", []>; + def AND16i16 : Ii16<0x25, RawFrm, (outs), (ins i16imm:$src), + "and{w}\t{$src, %ax|%ax, $src}", []>, OpSize; + def AND32i32 : Ii32<0x25, RawFrm, (outs), (ins i32imm:$src), + "and{l}\t{$src, %eax|%eax, $src}", []>; + } @@ -1635,6 +1835,13 @@ let isTwoAddress = 0 in { "or{l}\t{$src, $dst|$dst, $src}", [(store (or (load addr:$dst), i32immSExt8:$src), addr:$dst), (implicit EFLAGS)]>; + + def OR8i8 : Ii8 <0x0C, RawFrm, (outs), (ins i8imm:$src), + "or{b}\t{$src, %al|%al, $src}", []>; + def OR16i16 : Ii16 <0x0D, RawFrm, (outs), (ins i16imm:$src), + "or{w}\t{$src, %ax|%ax, $src}", []>, OpSize; + def OR32i32 : Ii32 <0x0D, RawFrm, (outs), (ins i32imm:$src), + "or{l}\t{$src, %eax|%eax, $src}", []>; } // isTwoAddress = 0 @@ -1744,6 +1951,13 @@ let isTwoAddress = 0 in { "xor{l}\t{$src, $dst|$dst, $src}", [(store (xor (load addr:$dst), i32immSExt8:$src), addr:$dst), (implicit EFLAGS)]>; + + def XOR8i8 : Ii8 <0x34, RawFrm, (outs), (ins i8imm:$src), + "xor{b}\t{$src, %al|%al, $src}", []>; + def XOR16i16 : Ii16 <0x35, RawFrm, (outs), (ins i16imm:$src), + "xor{w}\t{$src, %ax|%ax, $src}", []>, OpSize; + def XOR32i32 : Ii32 <0x35, RawFrm, (outs), (ins i32imm:$src), + "xor{l}\t{$src, %eax|%eax, $src}", []>; } // isTwoAddress = 0 } // Defs = [EFLAGS] @@ -1771,8 +1985,17 @@ def SHL16ri : Ii8<0xC1, MRM4r, (outs GR16:$dst), (ins GR16:$src1, i8imm:$src2), def SHL32ri : Ii8<0xC1, MRM4r, (outs GR32:$dst), (ins GR32:$src1, i8imm:$src2), "shl{l}\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (shl GR32:$src1, (i8 imm:$src2)))]>; -// NOTE: We don't use shifts of a register by one, because 'add reg,reg' is -// cheaper. + +// NOTE: We don't include patterns for shifts of a register by one, because +// 'add reg,reg' is cheaper. + +def SHL8r1 : I<0xD0, MRM4r, (outs GR8:$dst), (ins GR8:$src1), + "shl{b}\t$dst", []>; +def SHL16r1 : I<0xD1, MRM4r, (outs GR16:$dst), (ins GR16:$src1), + "shl{w}\t$dst", []>, OpSize; +def SHL32r1 : I<0xD1, MRM4r, (outs GR32:$dst), (ins GR32:$src1), + "shl{l}\t$dst", []>; + } // isConvertibleToThreeAddress = 1 let isTwoAddress = 0 in { @@ -1951,6 +2174,97 @@ let isTwoAddress = 0 in { } // Rotate instructions + +def RCL8r1 : I<0xD0, MRM2r, (outs GR8:$dst), (ins GR8:$src), + "rcl{b}\t{1, $dst|$dst, 1}", []>; +def RCL8m1 : I<0xD0, MRM2m, (outs i8mem:$dst), (ins i8mem:$src), + "rcl{b}\t{1, $dst|$dst, 1}", []>; +let Uses = [CL] in { +def RCL8rCL : I<0xD2, MRM2r, (outs GR8:$dst), (ins GR8:$src), + "rcl{b}\t{%cl, $dst|$dst, CL}", []>; +def RCL8mCL : I<0xD2, MRM2m, (outs i8mem:$dst), (ins i8mem:$src), + "rcl{b}\t{%cl, $dst|$dst, CL}", []>; +} +def RCL8ri : Ii8<0xC0, MRM2r, (outs GR8:$dst), (ins GR8:$src, i8imm:$cnt), + "rcl{b}\t{$cnt, $dst|$dst, $cnt}", []>; +def RCL8mi : Ii8<0xC0, MRM2m, (outs i8mem:$dst), (ins i8mem:$src, i8imm:$cnt), + "rcl{b}\t{$cnt, $dst|$dst, $cnt}", []>; + +def RCL16r1 : I<0xD1, MRM2r, (outs GR16:$dst), (ins GR16:$src), + "rcl{w}\t{1, $dst|$dst, 1}", []>, OpSize; +def RCL16m1 : I<0xD1, MRM2m, (outs i16mem:$dst), (ins i16mem:$src), + "rcl{w}\t{1, $dst|$dst, 1}", []>, OpSize; +let Uses = [CL] in { +def RCL16rCL : I<0xD3, MRM2r, (outs GR16:$dst), (ins GR16:$src), + "rcl{w}\t{%cl, $dst|$dst, CL}", []>, OpSize; +def RCL16mCL : I<0xD3, MRM2m, (outs i16mem:$dst), (ins i16mem:$src), + "rcl{w}\t{%cl, $dst|$dst, CL}", []>, OpSize; +} +def RCL16ri : Ii8<0xC1, MRM2r, (outs GR16:$dst), (ins GR16:$src, i8imm:$cnt), + "rcl{w}\t{$cnt, $dst|$dst, $cnt}", []>, OpSize; +def RCL16mi : Ii8<0xC1, MRM2m, (outs i16mem:$dst), (ins i16mem:$src, i8imm:$cnt), + "rcl{w}\t{$cnt, $dst|$dst, $cnt}", []>, OpSize; + +def RCL32r1 : I<0xD1, MRM2r, (outs GR32:$dst), (ins GR32:$src), + "rcl{l}\t{1, $dst|$dst, 1}", []>; +def RCL32m1 : I<0xD1, MRM2m, (outs i32mem:$dst), (ins i32mem:$src), + "rcl{l}\t{1, $dst|$dst, 1}", []>; +let Uses = [CL] in { +def RCL32rCL : I<0xD3, MRM2r, (outs GR32:$dst), (ins GR32:$src), + "rcl{l}\t{%cl, $dst|$dst, CL}", []>; +def RCL32mCL : I<0xD3, MRM2m, (outs i32mem:$dst), (ins i32mem:$src), + "rcl{l}\t{%cl, $dst|$dst, CL}", []>; +} +def RCL32ri : Ii8<0xC1, MRM2r, (outs GR32:$dst), (ins GR32:$src, i8imm:$cnt), + "rcl{l}\t{$cnt, $dst|$dst, $cnt}", []>; +def RCL32mi : Ii8<0xC1, MRM2m, (outs i32mem:$dst), (ins i32mem:$src, i8imm:$cnt), + "rcl{l}\t{$cnt, $dst|$dst, $cnt}", []>; + +def RCR8r1 : I<0xD0, MRM3r, (outs GR8:$dst), (ins GR8:$src), + "rcr{b}\t{1, $dst|$dst, 1}", []>; +def RCR8m1 : I<0xD0, MRM3m, (outs i8mem:$dst), (ins i8mem:$src), + "rcr{b}\t{1, $dst|$dst, 1}", []>; +let Uses = [CL] in { +def RCR8rCL : I<0xD2, MRM3r, (outs GR8:$dst), (ins GR8:$src), + "rcr{b}\t{%cl, $dst|$dst, CL}", []>; +def RCR8mCL : I<0xD2, MRM3m, (outs i8mem:$dst), (ins i8mem:$src), + "rcr{b}\t{%cl, $dst|$dst, CL}", []>; +} +def RCR8ri : Ii8<0xC0, MRM3r, (outs GR8:$dst), (ins GR8:$src, i8imm:$cnt), + "rcr{b}\t{$cnt, $dst|$dst, $cnt}", []>; +def RCR8mi : Ii8<0xC0, MRM3m, (outs i8mem:$dst), (ins i8mem:$src, i8imm:$cnt), + "rcr{b}\t{$cnt, $dst|$dst, $cnt}", []>; + +def RCR16r1 : I<0xD1, MRM3r, (outs GR16:$dst), (ins GR16:$src), + "rcr{w}\t{1, $dst|$dst, 1}", []>, OpSize; +def RCR16m1 : I<0xD1, MRM3m, (outs i16mem:$dst), (ins i16mem:$src), + "rcr{w}\t{1, $dst|$dst, 1}", []>, OpSize; +let Uses = [CL] in { +def RCR16rCL : I<0xD3, MRM3r, (outs GR16:$dst), (ins GR16:$src), + "rcr{w}\t{%cl, $dst|$dst, CL}", []>, OpSize; +def RCR16mCL : I<0xD3, MRM3m, (outs i16mem:$dst), (ins i16mem:$src), + "rcr{w}\t{%cl, $dst|$dst, CL}", []>, OpSize; +} +def RCR16ri : Ii8<0xC1, MRM3r, (outs GR16:$dst), (ins GR16:$src, i8imm:$cnt), + "rcr{w}\t{$cnt, $dst|$dst, $cnt}", []>, OpSize; +def RCR16mi : Ii8<0xC1, MRM3m, (outs i16mem:$dst), (ins i16mem:$src, i8imm:$cnt), + "rcr{w}\t{$cnt, $dst|$dst, $cnt}", []>, OpSize; + +def RCR32r1 : I<0xD1, MRM3r, (outs GR32:$dst), (ins GR32:$src), + "rcr{l}\t{1, $dst|$dst, 1}", []>; +def RCR32m1 : I<0xD1, MRM3m, (outs i32mem:$dst), (ins i32mem:$src), + "rcr{l}\t{1, $dst|$dst, 1}", []>; +let Uses = [CL] in { +def RCR32rCL : I<0xD3, MRM3r, (outs GR32:$dst), (ins GR32:$src), + "rcr{l}\t{%cl, $dst|$dst, CL}", []>; +def RCR32mCL : I<0xD3, MRM3m, (outs i32mem:$dst), (ins i32mem:$src), + "rcr{l}\t{%cl, $dst|$dst, CL}", []>; +} +def RCR32ri : Ii8<0xC1, MRM3r, (outs GR32:$dst), (ins GR32:$src, i8imm:$cnt), + "rcr{l}\t{$cnt, $dst|$dst, $cnt}", []>; +def RCR32mi : Ii8<0xC1, MRM3m, (outs i32mem:$dst), (ins i32mem:$src, i8imm:$cnt), + "rcr{l}\t{$cnt, $dst|$dst, $cnt}", []>; + // FIXME: provide shorter instructions when imm8 == 1 let Uses = [CL] in { def ROL8rCL : I<0xD2, MRM0r, (outs GR8 :$dst), (ins GR8 :$src), @@ -2228,6 +2542,15 @@ def ADD32rm : I<0x03, MRMSrcMem, (outs GR32:$dst), "add{l}\t{$src2, $dst|$dst, $src2}", [(set GR32:$dst, (add GR32:$src1, (load addr:$src2))), (implicit EFLAGS)]>; + +// Register-Register Addition - Equivalent to the normal rr forms (ADD8rr, +// ADD16rr, and ADD32rr), but differently encoded. +def ADD8mrmrr: I<0x02, MRMSrcReg, (outs GR8:$dst), (ins GR8:$src1, GR8:$src2), + "add{b}\t{$src2, $dst|$dst, $src2}", []>; +def ADD16mrmrr: I<0x03, MRMSrcReg,(outs GR16:$dst),(ins GR16:$src1, GR16:$src2), + "add{w}\t{$src2, $dst|$dst, $src2}", []>, OpSize; +def ADD32mrmrr: I<0x03, MRMSrcReg,(outs GR16:$dst),(ins GR16:$src1, GR16:$src2), + "add{l}\t{$src2, $dst|$dst, $src2}", []>; // Register-Integer Addition def ADD8ri : Ii8<0x80, MRM0r, (outs GR8:$dst), (ins GR8:$src1, i8imm:$src2), @@ -2295,6 +2618,14 @@ let isTwoAddress = 0 in { [(store (add (load addr:$dst), i32immSExt8:$src2), addr:$dst), (implicit EFLAGS)]>; + + // addition to rAX + def ADD8i8 : Ii8<0x04, RawFrm, (outs), (ins i8imm:$src), + "add{b}\t{$src, %al|%al, $src}", []>; + def ADD16i16 : Ii16<0x05, RawFrm, (outs), (ins i16imm:$src), + "add{w}\t{$src, %ax|%ax, $src}", []>, OpSize; + def ADD32i32 : Ii32<0x05, RawFrm, (outs), (ins i32imm:$src), + "add{l}\t{$src, %eax|%eax, $src}", []>; } let Uses = [EFLAGS] in { @@ -2373,6 +2704,13 @@ let isTwoAddress = 0 in { def ADC32mi8 : Ii8<0x83, MRM2m, (outs), (ins i32mem:$dst, i32i8imm :$src2), "adc{l}\t{$src2, $dst|$dst, $src2}", [(store (adde (load addr:$dst), i32immSExt8:$src2), addr:$dst)]>; + + def ADC8i8 : Ii8<0x14, RawFrm, (outs), (ins i8imm:$src), + "adc{b}\t{$src, %al|%al, $src}", []>; + def ADC16i16 : Ii16<0x15, RawFrm, (outs), (ins i16imm:$src), + "adc{w}\t{$src, %ax|%ax, $src}", []>, OpSize; + def ADC32i32 : Ii32<0x15, RawFrm, (outs), (ins i32imm:$src), + "adc{l}\t{$src, %eax|%eax, $src}", []>; } } // Uses = [EFLAGS] @@ -2472,6 +2810,13 @@ let isTwoAddress = 0 in { [(store (sub (load addr:$dst), i32immSExt8:$src2), addr:$dst), (implicit EFLAGS)]>; + + def SUB8i8 : Ii8<0x2C, RawFrm, (outs), (ins i8imm:$src), + "sub{b}\t{$src, %al|%al, $src}", []>; + def SUB16i16 : Ii16<0x2D, RawFrm, (outs), (ins i16imm:$src), + "sub{w}\t{$src, %ax|%ax, $src}", []>, OpSize; + def SUB32i32 : Ii32<0x2D, RawFrm, (outs), (ins i32imm:$src), + "sub{l}\t{$src, %eax|%eax, $src}", []>; } let Uses = [EFLAGS] in { @@ -2516,6 +2861,13 @@ let isTwoAddress = 0 in { def SBB32mi8 : Ii8<0x83, MRM3m, (outs), (ins i32mem:$dst, i32i8imm :$src2), "sbb{l}\t{$src2, $dst|$dst, $src2}", [(store (sube (load addr:$dst), i32immSExt8:$src2), addr:$dst)]>; + + def SBB8i8 : Ii8<0x1C, RawFrm, (outs), (ins i8imm:$src), + "sbb{b}\t{$src, %al|%al, $src}", []>; + def SBB16i16 : Ii16<0x1D, RawFrm, (outs), (ins i16imm:$src), + "sbb{w}\t{$src, %ax|%ax, $src}", []>, OpSize; + def SBB32i32 : Ii32<0x1D, RawFrm, (outs), (ins i32imm:$src), + "sbb{l}\t{$src, %eax|%eax, $src}", []>; } def SBB8rm : I<0x1A, MRMSrcMem, (outs GR8:$dst), (ins GR8:$src1, i8mem:$src2), "sbb{b}\t{$src2, $dst|$dst, $src2}", @@ -2647,6 +2999,13 @@ def TEST32rr : I<0x85, MRMDestReg, (outs), (ins GR32:$src1, GR32:$src2), (implicit EFLAGS)]>; } +def TEST8i8 : Ii8<0xA8, RawFrm, (outs), (ins i8imm:$src), + "test{b}\t{$src, %al|%al, $src}", []>; +def TEST16i16 : Ii16<0xA9, RawFrm, (outs), (ins i16imm:$src), + "test{w}\t{$src, %ax|%ax, $src}", []>, OpSize; +def TEST32i32 : Ii32<0xA9, RawFrm, (outs), (ins i32imm:$src), + "test{l}\t{$src, %eax|%eax, $src}", []>; + def TEST8rm : I<0x84, MRMSrcMem, (outs), (ins GR8 :$src1, i8mem :$src2), "test{b}\t{$src2, $src1|$src1, $src2}", [(X86cmp (and GR8:$src1, (loadi8 addr:$src2)), 0), @@ -2878,6 +3237,13 @@ def SETNOm : I<0x91, MRM0m, // Integer comparisons let Defs = [EFLAGS] in { +def CMP8i8 : Ii8<0x3C, RawFrm, (outs), (ins i8imm:$src), + "cmp{b}\t{$src, %al|%al, $src}", []>; +def CMP16i16 : Ii16<0x3D, RawFrm, (outs), (ins i16imm:$src), + "cmp{w}\t{$src, %ax|%ax, $src}", []>, OpSize; +def CMP32i32 : Ii32<0x3D, RawFrm, (outs), (ins i32imm:$src), + "cmp{l}\t{$src, %eax|%eax, $src}", []>; + def CMP8rr : I<0x38, MRMDestReg, (outs), (ins GR8 :$src1, GR8 :$src2), "cmp{b}\t{$src2, $src1|$src1, $src2}", @@ -2920,6 +3286,12 @@ def CMP32rm : I<0x3B, MRMSrcMem, "cmp{l}\t{$src2, $src1|$src1, $src2}", [(X86cmp GR32:$src1, (loadi32 addr:$src2)), (implicit EFLAGS)]>; +def CMP8mrmrr : I<0x3A, MRMSrcReg, (outs), (ins GR8:$src1, GR8:$src2), + "cmp{b}\t{$src2, $src1|$src1, $src2}", []>; +def CMP16mrmrr : I<0x3B, MRMSrcReg, (outs), (ins GR16:$src1, GR16:$src2), + "cmp{w}\t{$src2, $src1|$src1, $src2}", []>, OpSize; +def CMP32mrmrr : I<0x3B, MRMSrcReg, (outs), (ins GR32:$src1, GR32:$src2), + "cmp{l}\t{$src2, $src1|$src1, $src2}", []>; def CMP8ri : Ii8<0x80, MRM7r, (outs), (ins GR8:$src1, i8imm:$src2), "cmp{b}\t{$src2, $src1|$src1, $src2}", @@ -3095,7 +3467,8 @@ let neverHasSideEffects = 1 in { // Alias instructions that map movr0 to xor. // FIXME: remove when we can teach regalloc that xor reg, reg is ok. -let Defs = [EFLAGS], isReMaterializable = 1, isAsCheapAsAMove = 1 in { +let Defs = [EFLAGS], isReMaterializable = 1, isAsCheapAsAMove = 1, + isCodeGenOnly = 1 in { def MOV8r0 : I<0x30, MRMInitReg, (outs GR8 :$dst), (ins), "xor{b}\t$dst, $dst", [(set GR8:$dst, 0)]>; @@ -3127,12 +3500,12 @@ def TLS_addr32 : I<0, Pseudo, (outs), (ins lea32mem:$sym), [(X86tlsaddr tls32addr:$sym)]>, Requires<[In32BitMode]>; -let AddedComplexity = 5 in +let AddedComplexity = 5, isCodeGenOnly = 1 in def GS_MOV32rm : I<0x8B, MRMSrcMem, (outs GR32:$dst), (ins i32mem:$src), "movl\t%gs:$src, $dst", [(set GR32:$dst, (gsload addr:$src))]>, SegGS; -let AddedComplexity = 5 in +let AddedComplexity = 5, isCodeGenOnly = 1 in def FS_MOV32rm : I<0x8B, MRMSrcMem, (outs GR32:$dst), (ins i32mem:$src), "movl\t%fs:$src, $dst", [(set GR32:$dst, (fsload addr:$src))]>, SegFS; @@ -3143,7 +3516,7 @@ def FS_MOV32rm : I<0x8B, MRMSrcMem, (outs GR32:$dst), (ins i32mem:$src), def DWARF_LOC : I<0, Pseudo, (outs), (ins i32imm:$line, i32imm:$col, i32imm:$file), - ".loc\t${file:debug} ${line:debug} ${col:debug}", + ".loc\t$file $line $col", [(dwarf_loc (i32 imm:$line), (i32 imm:$col), (i32 imm:$file))]>; @@ -3151,7 +3524,7 @@ def DWARF_LOC : I<0, Pseudo, (outs), // EH Pseudo Instructions // let isTerminator = 1, isReturn = 1, isBarrier = 1, - hasCtrlDep = 1 in { + hasCtrlDep = 1, isCodeGenOnly = 1 in { def EH_RETURN : I<0xC3, RawFrm, (outs), (ins GR32:$addr), "ret\t#eh_return, addr: $addr", [(X86ehret GR32:$addr)]>; @@ -3223,6 +3596,78 @@ def LXADD8 : I<0xC0, MRMSrcMem, (outs GR8:$dst), (ins i8mem:$ptr, GR8:$val), TB, LOCK; } +// Optimized codegen when the non-memory output is not used. +// FIXME: Use normal add / sub instructions and add lock prefix dynamically. +def LOCK_ADD8mr : I<0x00, MRMDestMem, (outs), (ins i8mem:$dst, GR8:$src2), + "lock\n\t" + "add{b}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_ADD16mr : I<0x01, MRMDestMem, (outs), (ins i16mem:$dst, GR16:$src2), + "lock\n\t" + "add{w}\t{$src2, $dst|$dst, $src2}", []>, OpSize, LOCK; +def LOCK_ADD32mr : I<0x01, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src2), + "lock\n\t" + "add{l}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_ADD8mi : Ii8<0x80, MRM0m, (outs), (ins i8mem :$dst, i8imm :$src2), + "lock\n\t" + "add{b}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_ADD16mi : Ii16<0x81, MRM0m, (outs), (ins i16mem:$dst, i16imm:$src2), + "lock\n\t" + "add{w}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_ADD32mi : Ii32<0x81, MRM0m, (outs), (ins i32mem:$dst, i32imm:$src2), + "lock\n\t" + "add{l}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_ADD16mi8 : Ii8<0x83, MRM0m, (outs), (ins i16mem:$dst, i16i8imm :$src2), + "lock\n\t" + "add{w}\t{$src2, $dst|$dst, $src2}", []>, OpSize, LOCK; +def LOCK_ADD32mi8 : Ii8<0x83, MRM0m, (outs), (ins i32mem:$dst, i32i8imm :$src2), + "lock\n\t" + "add{l}\t{$src2, $dst|$dst, $src2}", []>, LOCK; + +def LOCK_INC8m : I<0xFE, MRM0m, (outs), (ins i8mem :$dst), + "lock\n\t" + "inc{b}\t$dst", []>, LOCK; +def LOCK_INC16m : I<0xFF, MRM0m, (outs), (ins i16mem:$dst), + "lock\n\t" + "inc{w}\t$dst", []>, OpSize, LOCK; +def LOCK_INC32m : I<0xFF, MRM0m, (outs), (ins i32mem:$dst), + "lock\n\t" + "inc{l}\t$dst", []>, LOCK; + +def LOCK_SUB8mr : I<0x28, MRMDestMem, (outs), (ins i8mem :$dst, GR8 :$src2), + "lock\n\t" + "sub{b}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_SUB16mr : I<0x29, MRMDestMem, (outs), (ins i16mem:$dst, GR16:$src2), + "lock\n\t" + "sub{w}\t{$src2, $dst|$dst, $src2}", []>, OpSize, LOCK; +def LOCK_SUB32mr : I<0x29, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src2), + "lock\n\t" + "sub{l}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_SUB8mi : Ii8<0x80, MRM5m, (outs), (ins i8mem :$dst, i8imm:$src2), + "lock\n\t" + "sub{b}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_SUB16mi : Ii16<0x81, MRM5m, (outs), (ins i16mem:$dst, i16imm:$src2), + "lock\n\t" + "sub{w}\t{$src2, $dst|$dst, $src2}", []>, OpSize, LOCK; +def LOCK_SUB32mi : Ii32<0x81, MRM5m, (outs), (ins i32mem:$dst, i32imm:$src2), + "lock\n\t" + "sub{l}\t{$src2, $dst|$dst, $src2}", []>, LOCK; +def LOCK_SUB16mi8 : Ii8<0x83, MRM5m, (outs), (ins i16mem:$dst, i16i8imm :$src2), + "lock\n\t" + "sub{w}\t{$src2, $dst|$dst, $src2}", []>, OpSize, LOCK; +def LOCK_SUB32mi8 : Ii8<0x83, MRM5m, (outs), (ins i32mem:$dst, i32i8imm :$src2), + "lock\n\t" + "sub{l}\t{$src2, $dst|$dst, $src2}", []>, LOCK; + +def LOCK_DEC8m : I<0xFE, MRM1m, (outs), (ins i8mem :$dst), + "lock\n\t" + "dec{b}\t$dst", []>, LOCK; +def LOCK_DEC16m : I<0xFF, MRM1m, (outs), (ins i16mem:$dst), + "lock\n\t" + "dec{w}\t$dst", []>, OpSize, LOCK; +def LOCK_DEC32m : I<0xFF, MRM1m, (outs), (ins i32mem:$dst), + "lock\n\t" + "dec{l}\t$dst", []>, LOCK; + // Atomic exchange, and, or, xor let Constraints = "$val = $dst", Defs = [EFLAGS], usesCustomDAGSchedInserter = 1 in { @@ -3318,6 +3763,25 @@ def ATOMSWAP6432 : I<0, Pseudo, (outs GR32:$dst1, GR32:$dst2), "#ATOMSWAP6432 PSEUDO!", []>; } +// Segmentation support instructions. + +def LAR16rm : I<0x02, MRMSrcMem, (outs GR16:$dst), (ins i16mem:$src), + "lar{w}\t{$src, $dst|$dst, $src}", []>, TB, OpSize; +def LAR16rr : I<0x02, MRMSrcReg, (outs GR16:$dst), (ins GR16:$src), + "lar{w}\t{$src, $dst|$dst, $src}", []>, TB, OpSize; + +// i16mem operand in LAR32rm and GR32 operand in LAR32rr is not a typo. +def LAR32rm : I<0x02, MRMSrcMem, (outs GR32:$dst), (ins i16mem:$src), + "lar{l}\t{$src, $dst|$dst, $src}", []>, TB; +def LAR32rr : I<0x02, MRMSrcReg, (outs GR32:$dst), (ins GR32:$src), + "lar{l}\t{$src, $dst|$dst, $src}", []>, TB; + +// String manipulation instructions + +def LODSB : I<0xAC, RawFrm, (outs), (ins), "lodsb", []>; +def LODSW : I<0xAD, RawFrm, (outs), (ins), "lodsw", []>, OpSize; +def LODSD : I<0xAD, RawFrm, (outs), (ins), "lodsd", []>; + //===----------------------------------------------------------------------===// // Non-Instruction Patterns //===----------------------------------------------------------------------===// @@ -3345,14 +3809,6 @@ def : Pat<(store (i32 (X86Wrapper texternalsym:$src)), addr:$dst), // Calls // tailcall stuff -def : Pat<(X86tailcall GR32:$dst), - (TAILCALL)>; - -def : Pat<(X86tailcall (i32 tglobaladdr:$dst)), - (TAILCALL)>; -def : Pat<(X86tailcall (i32 texternalsym:$dst)), - (TAILCALL)>; - def : Pat<(X86tcret GR32:$dst, imm:$off), (TCRETURNri GR32:$dst, imm:$off)>; @@ -3362,6 +3818,7 @@ def : Pat<(X86tcret (i32 tglobaladdr:$dst), imm:$off), def : Pat<(X86tcret (i32 texternalsym:$dst), imm:$off), (TCRETURNdi texternalsym:$dst, imm:$off)>; +// Normal calls, with various flavors of addresses. def : Pat<(X86call (i32 tglobaladdr:$dst)), (CALLpcrel32 tglobaladdr:$dst)>; def : Pat<(X86call (i32 texternalsym:$dst)), @@ -3472,21 +3929,17 @@ def : Pat<(zextloadi32i1 addr:$src), (MOVZX32rm8 addr:$src)>; // extload bool -> extload byte def : Pat<(extloadi8i1 addr:$src), (MOV8rm addr:$src)>; -def : Pat<(extloadi16i1 addr:$src), (MOVZX16rm8 addr:$src)>, - Requires<[In32BitMode]>; +def : Pat<(extloadi16i1 addr:$src), (MOVZX16rm8 addr:$src)>; def : Pat<(extloadi32i1 addr:$src), (MOVZX32rm8 addr:$src)>; -def : Pat<(extloadi16i8 addr:$src), (MOVZX16rm8 addr:$src)>, - Requires<[In32BitMode]>; +def : Pat<(extloadi16i8 addr:$src), (MOVZX16rm8 addr:$src)>; def : Pat<(extloadi32i8 addr:$src), (MOVZX32rm8 addr:$src)>; def : Pat<(extloadi32i16 addr:$src), (MOVZX32rm16 addr:$src)>; -// anyext -def : Pat<(i16 (anyext GR8 :$src)), (MOVZX16rr8 GR8 :$src)>, - Requires<[In32BitMode]>; -def : Pat<(i32 (anyext GR8 :$src)), (MOVZX32rr8 GR8 :$src)>, - Requires<[In32BitMode]>; -def : Pat<(i32 (anyext GR16:$src)), - (INSERT_SUBREG (i32 (IMPLICIT_DEF)), GR16:$src, x86_subreg_16bit)>; +// anyext. Define these to do an explicit zero-extend to +// avoid partial-register updates. +def : Pat<(i16 (anyext GR8 :$src)), (MOVZX16rr8 GR8 :$src)>; +def : Pat<(i32 (anyext GR8 :$src)), (MOVZX32rr8 GR8 :$src)>; +def : Pat<(i32 (anyext GR16:$src)), (MOVZX32rr16 GR16:$src)>; // (and (i32 load), 255) -> (zextload i8) def : Pat<(i32 (and (nvloadi32 addr:$src), (i32 255))), @@ -3567,6 +4020,10 @@ def : Pat<(i32 (zext (srl_su GR16:$src, (i8 8)))), (MOVZX32rr8 (EXTRACT_SUBREG (COPY_TO_REGCLASS GR16:$src, GR16_ABCD), x86_subreg_8bit_hi))>, Requires<[In32BitMode]>; +def : Pat<(i32 (anyext (srl_su GR16:$src, (i8 8)))), + (MOVZX32rr8 (EXTRACT_SUBREG (COPY_TO_REGCLASS GR16:$src, GR16_ABCD), + x86_subreg_8bit_hi))>, + Requires<[In32BitMode]>; def : Pat<(and (srl_su GR32:$src, (i8 8)), (i32 255)), (MOVZX32rr8 (EXTRACT_SUBREG (COPY_TO_REGCLASS GR32:$src, GR32_ABCD), x86_subreg_8bit_hi))>, @@ -3961,6 +4418,243 @@ def : Pat<(parallel (store (i32 (X86dec_flag (loadi32 addr:$dst))), addr:$dst), (implicit EFLAGS)), (DEC32m addr:$dst)>, Requires<[In32BitMode]>; +// Register-Register Or with EFLAGS result +def : Pat<(parallel (X86or_flag GR8:$src1, GR8:$src2), + (implicit EFLAGS)), + (OR8rr GR8:$src1, GR8:$src2)>; +def : Pat<(parallel (X86or_flag GR16:$src1, GR16:$src2), + (implicit EFLAGS)), + (OR16rr GR16:$src1, GR16:$src2)>; +def : Pat<(parallel (X86or_flag GR32:$src1, GR32:$src2), + (implicit EFLAGS)), + (OR32rr GR32:$src1, GR32:$src2)>; + +// Register-Memory Or with EFLAGS result +def : Pat<(parallel (X86or_flag GR8:$src1, (loadi8 addr:$src2)), + (implicit EFLAGS)), + (OR8rm GR8:$src1, addr:$src2)>; +def : Pat<(parallel (X86or_flag GR16:$src1, (loadi16 addr:$src2)), + (implicit EFLAGS)), + (OR16rm GR16:$src1, addr:$src2)>; +def : Pat<(parallel (X86or_flag GR32:$src1, (loadi32 addr:$src2)), + (implicit EFLAGS)), + (OR32rm GR32:$src1, addr:$src2)>; + +// Register-Integer Or with EFLAGS result +def : Pat<(parallel (X86or_flag GR8:$src1, imm:$src2), + (implicit EFLAGS)), + (OR8ri GR8:$src1, imm:$src2)>; +def : Pat<(parallel (X86or_flag GR16:$src1, imm:$src2), + (implicit EFLAGS)), + (OR16ri GR16:$src1, imm:$src2)>; +def : Pat<(parallel (X86or_flag GR32:$src1, imm:$src2), + (implicit EFLAGS)), + (OR32ri GR32:$src1, imm:$src2)>; +def : Pat<(parallel (X86or_flag GR16:$src1, i16immSExt8:$src2), + (implicit EFLAGS)), + (OR16ri8 GR16:$src1, i16immSExt8:$src2)>; +def : Pat<(parallel (X86or_flag GR32:$src1, i32immSExt8:$src2), + (implicit EFLAGS)), + (OR32ri8 GR32:$src1, i32immSExt8:$src2)>; + +// Memory-Register Or with EFLAGS result +def : Pat<(parallel (store (X86or_flag (loadi8 addr:$dst), GR8:$src2), + addr:$dst), + (implicit EFLAGS)), + (OR8mr addr:$dst, GR8:$src2)>; +def : Pat<(parallel (store (X86or_flag (loadi16 addr:$dst), GR16:$src2), + addr:$dst), + (implicit EFLAGS)), + (OR16mr addr:$dst, GR16:$src2)>; +def : Pat<(parallel (store (X86or_flag (loadi32 addr:$dst), GR32:$src2), + addr:$dst), + (implicit EFLAGS)), + (OR32mr addr:$dst, GR32:$src2)>; + +// Memory-Integer Or with EFLAGS result +def : Pat<(parallel (store (X86or_flag (loadi8 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (OR8mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86or_flag (loadi16 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (OR16mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86or_flag (loadi32 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (OR32mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86or_flag (loadi16 addr:$dst), i16immSExt8:$src2), + addr:$dst), + (implicit EFLAGS)), + (OR16mi8 addr:$dst, i16immSExt8:$src2)>; +def : Pat<(parallel (store (X86or_flag (loadi32 addr:$dst), i32immSExt8:$src2), + addr:$dst), + (implicit EFLAGS)), + (OR32mi8 addr:$dst, i32immSExt8:$src2)>; + +// Register-Register XOr with EFLAGS result +def : Pat<(parallel (X86xor_flag GR8:$src1, GR8:$src2), + (implicit EFLAGS)), + (XOR8rr GR8:$src1, GR8:$src2)>; +def : Pat<(parallel (X86xor_flag GR16:$src1, GR16:$src2), + (implicit EFLAGS)), + (XOR16rr GR16:$src1, GR16:$src2)>; +def : Pat<(parallel (X86xor_flag GR32:$src1, GR32:$src2), + (implicit EFLAGS)), + (XOR32rr GR32:$src1, GR32:$src2)>; + +// Register-Memory XOr with EFLAGS result +def : Pat<(parallel (X86xor_flag GR8:$src1, (loadi8 addr:$src2)), + (implicit EFLAGS)), + (XOR8rm GR8:$src1, addr:$src2)>; +def : Pat<(parallel (X86xor_flag GR16:$src1, (loadi16 addr:$src2)), + (implicit EFLAGS)), + (XOR16rm GR16:$src1, addr:$src2)>; +def : Pat<(parallel (X86xor_flag GR32:$src1, (loadi32 addr:$src2)), + (implicit EFLAGS)), + (XOR32rm GR32:$src1, addr:$src2)>; + +// Register-Integer XOr with EFLAGS result +def : Pat<(parallel (X86xor_flag GR8:$src1, imm:$src2), + (implicit EFLAGS)), + (XOR8ri GR8:$src1, imm:$src2)>; +def : Pat<(parallel (X86xor_flag GR16:$src1, imm:$src2), + (implicit EFLAGS)), + (XOR16ri GR16:$src1, imm:$src2)>; +def : Pat<(parallel (X86xor_flag GR32:$src1, imm:$src2), + (implicit EFLAGS)), + (XOR32ri GR32:$src1, imm:$src2)>; +def : Pat<(parallel (X86xor_flag GR16:$src1, i16immSExt8:$src2), + (implicit EFLAGS)), + (XOR16ri8 GR16:$src1, i16immSExt8:$src2)>; +def : Pat<(parallel (X86xor_flag GR32:$src1, i32immSExt8:$src2), + (implicit EFLAGS)), + (XOR32ri8 GR32:$src1, i32immSExt8:$src2)>; + +// Memory-Register XOr with EFLAGS result +def : Pat<(parallel (store (X86xor_flag (loadi8 addr:$dst), GR8:$src2), + addr:$dst), + (implicit EFLAGS)), + (XOR8mr addr:$dst, GR8:$src2)>; +def : Pat<(parallel (store (X86xor_flag (loadi16 addr:$dst), GR16:$src2), + addr:$dst), + (implicit EFLAGS)), + (XOR16mr addr:$dst, GR16:$src2)>; +def : Pat<(parallel (store (X86xor_flag (loadi32 addr:$dst), GR32:$src2), + addr:$dst), + (implicit EFLAGS)), + (XOR32mr addr:$dst, GR32:$src2)>; + +// Memory-Integer XOr with EFLAGS result +def : Pat<(parallel (store (X86xor_flag (loadi8 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (XOR8mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86xor_flag (loadi16 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (XOR16mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86xor_flag (loadi32 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (XOR32mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86xor_flag (loadi16 addr:$dst), i16immSExt8:$src2), + addr:$dst), + (implicit EFLAGS)), + (XOR16mi8 addr:$dst, i16immSExt8:$src2)>; +def : Pat<(parallel (store (X86xor_flag (loadi32 addr:$dst), i32immSExt8:$src2), + addr:$dst), + (implicit EFLAGS)), + (XOR32mi8 addr:$dst, i32immSExt8:$src2)>; + +// Register-Register And with EFLAGS result +def : Pat<(parallel (X86and_flag GR8:$src1, GR8:$src2), + (implicit EFLAGS)), + (AND8rr GR8:$src1, GR8:$src2)>; +def : Pat<(parallel (X86and_flag GR16:$src1, GR16:$src2), + (implicit EFLAGS)), + (AND16rr GR16:$src1, GR16:$src2)>; +def : Pat<(parallel (X86and_flag GR32:$src1, GR32:$src2), + (implicit EFLAGS)), + (AND32rr GR32:$src1, GR32:$src2)>; + +// Register-Memory And with EFLAGS result +def : Pat<(parallel (X86and_flag GR8:$src1, (loadi8 addr:$src2)), + (implicit EFLAGS)), + (AND8rm GR8:$src1, addr:$src2)>; +def : Pat<(parallel (X86and_flag GR16:$src1, (loadi16 addr:$src2)), + (implicit EFLAGS)), + (AND16rm GR16:$src1, addr:$src2)>; +def : Pat<(parallel (X86and_flag GR32:$src1, (loadi32 addr:$src2)), + (implicit EFLAGS)), + (AND32rm GR32:$src1, addr:$src2)>; + +// Register-Integer And with EFLAGS result +def : Pat<(parallel (X86and_flag GR8:$src1, imm:$src2), + (implicit EFLAGS)), + (AND8ri GR8:$src1, imm:$src2)>; +def : Pat<(parallel (X86and_flag GR16:$src1, imm:$src2), + (implicit EFLAGS)), + (AND16ri GR16:$src1, imm:$src2)>; +def : Pat<(parallel (X86and_flag GR32:$src1, imm:$src2), + (implicit EFLAGS)), + (AND32ri GR32:$src1, imm:$src2)>; +def : Pat<(parallel (X86and_flag GR16:$src1, i16immSExt8:$src2), + (implicit EFLAGS)), + (AND16ri8 GR16:$src1, i16immSExt8:$src2)>; +def : Pat<(parallel (X86and_flag GR32:$src1, i32immSExt8:$src2), + (implicit EFLAGS)), + (AND32ri8 GR32:$src1, i32immSExt8:$src2)>; + +// Memory-Register And with EFLAGS result +def : Pat<(parallel (store (X86and_flag (loadi8 addr:$dst), GR8:$src2), + addr:$dst), + (implicit EFLAGS)), + (AND8mr addr:$dst, GR8:$src2)>; +def : Pat<(parallel (store (X86and_flag (loadi16 addr:$dst), GR16:$src2), + addr:$dst), + (implicit EFLAGS)), + (AND16mr addr:$dst, GR16:$src2)>; +def : Pat<(parallel (store (X86and_flag (loadi32 addr:$dst), GR32:$src2), + addr:$dst), + (implicit EFLAGS)), + (AND32mr addr:$dst, GR32:$src2)>; + +// Memory-Integer And with EFLAGS result +def : Pat<(parallel (store (X86and_flag (loadi8 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (AND8mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86and_flag (loadi16 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (AND16mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86and_flag (loadi32 addr:$dst), imm:$src2), + addr:$dst), + (implicit EFLAGS)), + (AND32mi addr:$dst, imm:$src2)>; +def : Pat<(parallel (store (X86and_flag (loadi16 addr:$dst), i16immSExt8:$src2), + addr:$dst), + (implicit EFLAGS)), + (AND16mi8 addr:$dst, i16immSExt8:$src2)>; +def : Pat<(parallel (store (X86and_flag (loadi32 addr:$dst), i32immSExt8:$src2), + addr:$dst), + (implicit EFLAGS)), + (AND32mi8 addr:$dst, i32immSExt8:$src2)>; + +// -disable-16bit support. +def : Pat<(truncstorei16 (i32 imm:$src), addr:$dst), + (MOV16mi addr:$dst, imm:$src)>; +def : Pat<(truncstorei16 GR32:$src, addr:$dst), + (MOV16mr addr:$dst, (EXTRACT_SUBREG GR32:$src, x86_subreg_16bit))>; +def : Pat<(i32 (sextloadi16 addr:$dst)), + (MOVSX32rm16 addr:$dst)>; +def : Pat<(i32 (zextloadi16 addr:$dst)), + (MOVZX32rm16 addr:$dst)>; +def : Pat<(i32 (extloadi16 addr:$dst)), + (MOVZX32rm16 addr:$dst)>; + //===----------------------------------------------------------------------===// // Floating Point Stack Support //===----------------------------------------------------------------------===// |