X-Git-Url: http://plrg.eecs.uci.edu/git/?p=oota-llvm.git;a=blobdiff_plain;f=lib%2FTarget%2FX86%2FX86InstrFPStack.td;h=03ae21125b0e8ebe61508eb6044fc277a9edc0a3;hp=d37ecf80899df834130b1452d7e92e575744cc9c;hb=a933a559f6e509c1a55cf3824ddf7635538ebf1d;hpb=8e6da15e54125db38c0ae32f7a6b2273c792c588 diff --git a/lib/Target/X86/X86InstrFPStack.td b/lib/Target/X86/X86InstrFPStack.td index d37ecf80899..03ae21125b0 100644 --- a/lib/Target/X86/X86InstrFPStack.td +++ b/lib/Target/X86/X86InstrFPStack.td @@ -1,10 +1,10 @@ -//==- X86InstrFPStack.td - Describe the X86 Instruction Set --*- tablegen -*-=// -// +//===- X86InstrFPStack.td - FPU Instruction Set ------------*- tablegen -*-===// +// // The LLVM Compiler Infrastructure // // This file is distributed under the University of Illinois Open Source // License. See LICENSE.TXT for details. -// +// //===----------------------------------------------------------------------===// // // This file describes the X86 x87 FPU instruction set, defining the @@ -17,36 +17,41 @@ // FPStack specific DAG Nodes. //===----------------------------------------------------------------------===// -def SDTX86FpGet2 : SDTypeProfile<2, 0, [SDTCisVT<0, f80>, +def SDTX86FpGet2 : SDTypeProfile<2, 0, [SDTCisVT<0, f80>, SDTCisVT<1, f80>]>; def SDTX86Fld : SDTypeProfile<1, 2, [SDTCisFP<0>, - SDTCisPtrTy<1>, + SDTCisPtrTy<1>, SDTCisVT<2, OtherVT>]>; def SDTX86Fst : SDTypeProfile<0, 3, [SDTCisFP<0>, - SDTCisPtrTy<1>, + SDTCisPtrTy<1>, SDTCisVT<2, OtherVT>]>; def SDTX86Fild : SDTypeProfile<1, 2, [SDTCisFP<0>, SDTCisPtrTy<1>, SDTCisVT<2, OtherVT>]>; +def SDTX86Fnstsw : SDTypeProfile<1, 1, [SDTCisVT<0, i16>, SDTCisVT<1, i16>]>; def SDTX86FpToIMem : SDTypeProfile<0, 2, [SDTCisFP<0>, SDTCisPtrTy<1>]>; def SDTX86CwdStore : SDTypeProfile<0, 1, [SDTCisPtrTy<0>]>; def X86fld : SDNode<"X86ISD::FLD", SDTX86Fld, - [SDNPHasChain, SDNPMayLoad]>; + [SDNPHasChain, SDNPMayLoad, SDNPMemOperand]>; def X86fst : SDNode<"X86ISD::FST", SDTX86Fst, - [SDNPHasChain, SDNPInFlag, SDNPMayStore]>; + [SDNPHasChain, SDNPInGlue, SDNPMayStore, + SDNPMemOperand]>; def X86fild : SDNode<"X86ISD::FILD", SDTX86Fild, - [SDNPHasChain, SDNPMayLoad]>; + [SDNPHasChain, SDNPMayLoad, SDNPMemOperand]>; def X86fildflag : SDNode<"X86ISD::FILD_FLAG", SDTX86Fild, - [SDNPHasChain, SDNPOutFlag, SDNPMayLoad]>; + [SDNPHasChain, SDNPOutGlue, SDNPMayLoad, + SDNPMemOperand]>; +def X86fp_stsw : SDNode<"X86ISD::FNSTSW16r", SDTX86Fnstsw>; def X86fp_to_i16mem : SDNode<"X86ISD::FP_TO_INT16_IN_MEM", SDTX86FpToIMem, - [SDNPHasChain, SDNPMayStore]>; + [SDNPHasChain, SDNPMayStore, SDNPMemOperand]>; def X86fp_to_i32mem : SDNode<"X86ISD::FP_TO_INT32_IN_MEM", SDTX86FpToIMem, - [SDNPHasChain, SDNPMayStore]>; + [SDNPHasChain, SDNPMayStore, SDNPMemOperand]>; def X86fp_to_i64mem : SDNode<"X86ISD::FP_TO_INT64_IN_MEM", SDTX86FpToIMem, - [SDNPHasChain, SDNPMayStore]>; + [SDNPHasChain, SDNPMayStore, SDNPMemOperand]>; def X86fp_cwd_get16 : SDNode<"X86ISD::FNSTCW16m", SDTX86CwdStore, - [SDNPHasChain, SDNPMayStore, SDNPSideEffect]>; + [SDNPHasChain, SDNPMayStore, SDNPSideEffect, + SDNPMemOperand]>; //===----------------------------------------------------------------------===// // FPStack pattern fragments @@ -69,53 +74,31 @@ def fpimmneg1 : PatLeaf<(fpimm), [{ }]>; // Some 'special' instructions -let usesCustomDAGSchedInserter = 1 in { // Expanded by the scheduler. - def FP32_TO_INT16_IN_MEM : I<0, Pseudo, - (outs), (ins i16mem:$dst, RFP32:$src), - "#FP32_TO_INT16_IN_MEM PSEUDO!", +let usesCustomInserter = 1 in { // Expanded after instruction selection. + def FP32_TO_INT16_IN_MEM : PseudoI<(outs), (ins i16mem:$dst, RFP32:$src), [(X86fp_to_i16mem RFP32:$src, addr:$dst)]>; - def FP32_TO_INT32_IN_MEM : I<0, Pseudo, - (outs), (ins i32mem:$dst, RFP32:$src), - "#FP32_TO_INT32_IN_MEM PSEUDO!", + def FP32_TO_INT32_IN_MEM : PseudoI<(outs), (ins i32mem:$dst, RFP32:$src), [(X86fp_to_i32mem RFP32:$src, addr:$dst)]>; - def FP32_TO_INT64_IN_MEM : I<0, Pseudo, - (outs), (ins i64mem:$dst, RFP32:$src), - "#FP32_TO_INT64_IN_MEM PSEUDO!", + def FP32_TO_INT64_IN_MEM : PseudoI<(outs), (ins i64mem:$dst, RFP32:$src), [(X86fp_to_i64mem RFP32:$src, addr:$dst)]>; - def FP64_TO_INT16_IN_MEM : I<0, Pseudo, - (outs), (ins i16mem:$dst, RFP64:$src), - "#FP64_TO_INT16_IN_MEM PSEUDO!", + def FP64_TO_INT16_IN_MEM : PseudoI<(outs), (ins i16mem:$dst, RFP64:$src), [(X86fp_to_i16mem RFP64:$src, addr:$dst)]>; - def FP64_TO_INT32_IN_MEM : I<0, Pseudo, - (outs), (ins i32mem:$dst, RFP64:$src), - "#FP64_TO_INT32_IN_MEM PSEUDO!", + def FP64_TO_INT32_IN_MEM : PseudoI<(outs), (ins i32mem:$dst, RFP64:$src), [(X86fp_to_i32mem RFP64:$src, addr:$dst)]>; - def FP64_TO_INT64_IN_MEM : I<0, Pseudo, - (outs), (ins i64mem:$dst, RFP64:$src), - "#FP64_TO_INT64_IN_MEM PSEUDO!", + def FP64_TO_INT64_IN_MEM : PseudoI<(outs), (ins i64mem:$dst, RFP64:$src), [(X86fp_to_i64mem RFP64:$src, addr:$dst)]>; - def FP80_TO_INT16_IN_MEM : I<0, Pseudo, - (outs), (ins i16mem:$dst, RFP80:$src), - "#FP80_TO_INT16_IN_MEM PSEUDO!", + def FP80_TO_INT16_IN_MEM : PseudoI<(outs), (ins i16mem:$dst, RFP80:$src), [(X86fp_to_i16mem RFP80:$src, addr:$dst)]>; - def FP80_TO_INT32_IN_MEM : I<0, Pseudo, - (outs), (ins i32mem:$dst, RFP80:$src), - "#FP80_TO_INT32_IN_MEM PSEUDO!", + def FP80_TO_INT32_IN_MEM : PseudoI<(outs), (ins i32mem:$dst, RFP80:$src), [(X86fp_to_i32mem RFP80:$src, addr:$dst)]>; - def FP80_TO_INT64_IN_MEM : I<0, Pseudo, - (outs), (ins i64mem:$dst, RFP80:$src), - "#FP80_TO_INT64_IN_MEM PSEUDO!", + def FP80_TO_INT64_IN_MEM : PseudoI<(outs), (ins i64mem:$dst, RFP80:$src), [(X86fp_to_i64mem RFP80:$src, addr:$dst)]>; } -let isTerminator = 1 in - let Defs = [FP0, FP1, FP2, FP3, FP4, FP5, FP6] in - def FP_REG_KILL : I<0, Pseudo, (outs), (ins), "#FP_REG_KILL", []>; - // All FP Stack operations are represented with four instructions here. The // first three instructions, generated by the instruction selector, use "RFP32" // "RFP64" or "RFP80" registers: traditional register files to reference 32-bit, -// 64-bit or 80-bit floating point values. These sizes apply to the values, +// 64-bit or 80-bit floating point values. These sizes apply to the values, // not the registers, which are always 80 bits; RFP32, RFP64 and RFP80 can be // copied to each other without losing information. These instructions are all // pseudo instructions and use the "_Fp" suffix. @@ -124,29 +107,14 @@ let isTerminator = 1 in // The second instruction is defined with FPI, which is the actual instruction // emitted by the assembler. These use "RST" registers, although frequently // the actual register(s) used are implicit. These are always 80 bits. -// The FP stackifier pass converts one to the other after register allocation +// The FP stackifier pass converts one to the other after register allocation // occurs. // // Note that the FpI instruction should have instruction selection info (e.g. // a pattern) and the FPI instruction should have emission info (e.g. opcode // encoding and asm printing info). -// Pseudo Instructions for FP stack return values. -def FpGET_ST0_32 : FpI_<(outs RFP32:$dst), (ins), SpecialFP, []>; // FPR = ST(0) -def FpGET_ST0_64 : FpI_<(outs RFP64:$dst), (ins), SpecialFP, []>; // FPR = ST(0) -def FpGET_ST0_80 : FpI_<(outs RFP80:$dst), (ins), SpecialFP, []>; // FPR = ST(0) - -def FpGET_ST0_ST1 : FpI_<(outs RFP80:$dst1, RFP80:$dst2), (ins), SpecialFP, - []>; // FPR = ST(0), FPR = ST(1) - - -let Defs = [ST0] in { -def FpSET_ST0_32 : FpI_<(outs), (ins RFP32:$src), SpecialFP, []>; // ST(0) = FPR -def FpSET_ST0_64 : FpI_<(outs), (ins RFP64:$src), SpecialFP, []>; // ST(0) = FPR -def FpSET_ST0_80 : FpI_<(outs), (ins RFP80:$src), SpecialFP, []>; // ST(0) = FPR -} - -// FpIf32, FpIf64 - Floating Point Psuedo Instruction template. +// FpIf32, FpIf64 - Floating Point Pseudo Instruction template. // f32 instructions can use SSE1 and are predicated on FPStackf32 == !SSE1. // f64 instructions can use SSE2 and are predicated on FPStackf64 == !SSE2. // f80 instructions cannot use SSE and use neither of these. @@ -155,19 +123,6 @@ class FpIf32 pattern> : class FpIf64 pattern> : FpI_, Requires<[FPStackf64]>; -// Register copies. Just copies, the shortening ones do not truncate. -let neverHasSideEffects = 1 in { - def MOV_Fp3232 : FpIf32<(outs RFP32:$dst), (ins RFP32:$src), SpecialFP, []>; - def MOV_Fp3264 : FpIf32<(outs RFP64:$dst), (ins RFP32:$src), SpecialFP, []>; - def MOV_Fp6432 : FpIf32<(outs RFP32:$dst), (ins RFP64:$src), SpecialFP, []>; - def MOV_Fp6464 : FpIf64<(outs RFP64:$dst), (ins RFP64:$src), SpecialFP, []>; - def MOV_Fp8032 : FpIf32<(outs RFP32:$dst), (ins RFP80:$src), SpecialFP, []>; - def MOV_Fp3280 : FpIf32<(outs RFP80:$dst), (ins RFP32:$src), SpecialFP, []>; - def MOV_Fp8064 : FpIf64<(outs RFP64:$dst), (ins RFP80:$src), SpecialFP, []>; - def MOV_Fp6480 : FpIf64<(outs RFP80:$dst), (ins RFP64:$src), SpecialFP, []>; - def MOV_Fp8080 : FpI_ <(outs RFP80:$dst), (ins RFP80:$src), SpecialFP, []>; -} - // Factoring for arithmetic. multiclass FPBinary_rr { // Register op register -> register @@ -182,132 +137,237 @@ def _Fp80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src1, RFP80:$src2), TwoArgFP, // The FopST0 series are not included here because of the irregularities // in where the 'r' goes in assembly output. // These instructions cannot address 80-bit memory. -multiclass FPBinary { +multiclass FPBinary { // ST(0) = ST(0) + [mem] -def _Fp32m : FpIf32<(outs RFP32:$dst), (ins RFP32:$src1, f32mem:$src2), OneArgFPRW, - [(set RFP32:$dst, - (OpNode RFP32:$src1, (loadf32 addr:$src2)))]>; -def _Fp64m : FpIf64<(outs RFP64:$dst), (ins RFP64:$src1, f64mem:$src2), OneArgFPRW, - [(set RFP64:$dst, - (OpNode RFP64:$src1, (loadf64 addr:$src2)))]>; -def _Fp64m32: FpIf64<(outs RFP64:$dst), (ins RFP64:$src1, f32mem:$src2), OneArgFPRW, - [(set RFP64:$dst, - (OpNode RFP64:$src1, (f64 (extloadf32 addr:$src2))))]>; -def _Fp80m32: FpI_<(outs RFP80:$dst), (ins RFP80:$src1, f32mem:$src2), OneArgFPRW, - [(set RFP80:$dst, - (OpNode RFP80:$src1, (f80 (extloadf32 addr:$src2))))]>; -def _Fp80m64: FpI_<(outs RFP80:$dst), (ins RFP80:$src1, f64mem:$src2), OneArgFPRW, - [(set RFP80:$dst, - (OpNode RFP80:$src1, (f80 (extloadf64 addr:$src2))))]>; -def _F32m : FPI<0xD8, fp, (outs), (ins f32mem:$src), - !strconcat("f", !strconcat(asmstring, "{s}\t$src"))> { let mayLoad = 1; } -def _F64m : FPI<0xDC, fp, (outs), (ins f64mem:$src), - !strconcat("f", !strconcat(asmstring, "{l}\t$src"))> { let mayLoad = 1; } +def _Fp32m : FpIf32<(outs RFP32:$dst), + (ins RFP32:$src1, f32mem:$src2), OneArgFPRW, + [!if(Forward, + (set RFP32:$dst, + (OpNode RFP32:$src1, (loadf32 addr:$src2))), + (set RFP32:$dst, + (OpNode (loadf32 addr:$src2), RFP32:$src1)))]>; +def _Fp64m : FpIf64<(outs RFP64:$dst), + (ins RFP64:$src1, f64mem:$src2), OneArgFPRW, + [!if(Forward, + (set RFP64:$dst, + (OpNode RFP64:$src1, (loadf64 addr:$src2))), + (set RFP64:$dst, + (OpNode (loadf64 addr:$src2), RFP64:$src1)))]>; +def _Fp64m32: FpIf64<(outs RFP64:$dst), + (ins RFP64:$src1, f32mem:$src2), OneArgFPRW, + [!if(Forward, + (set RFP64:$dst, + (OpNode RFP64:$src1, (f64 (extloadf32 addr:$src2)))), + (set RFP64:$dst, + (OpNode (f64 (extloadf32 addr:$src2)), RFP64:$src1)))]>; +def _Fp80m32: FpI_<(outs RFP80:$dst), + (ins RFP80:$src1, f32mem:$src2), OneArgFPRW, + [!if(Forward, + (set RFP80:$dst, + (OpNode RFP80:$src1, (f80 (extloadf32 addr:$src2)))), + (set RFP80:$dst, + (OpNode (f80 (extloadf32 addr:$src2)), RFP80:$src1)))]>; +def _Fp80m64: FpI_<(outs RFP80:$dst), + (ins RFP80:$src1, f64mem:$src2), OneArgFPRW, + [!if(Forward, + (set RFP80:$dst, + (OpNode RFP80:$src1, (f80 (extloadf64 addr:$src2)))), + (set RFP80:$dst, + (OpNode (f80 (extloadf64 addr:$src2)), RFP80:$src1)))]>; +let mayLoad = 1 in +def _F32m : FPI<0xD8, fp, (outs), (ins f32mem:$src), + !strconcat("f", asmstring, "{s}\t$src")>; +let mayLoad = 1 in +def _F64m : FPI<0xDC, fp, (outs), (ins f64mem:$src), + !strconcat("f", asmstring, "{l}\t$src")>; // ST(0) = ST(0) + [memint] -def _FpI16m32 : FpIf32<(outs RFP32:$dst), (ins RFP32:$src1, i16mem:$src2), OneArgFPRW, - [(set RFP32:$dst, (OpNode RFP32:$src1, - (X86fild addr:$src2, i16)))]>; -def _FpI32m32 : FpIf32<(outs RFP32:$dst), (ins RFP32:$src1, i32mem:$src2), OneArgFPRW, - [(set RFP32:$dst, (OpNode RFP32:$src1, - (X86fild addr:$src2, i32)))]>; -def _FpI16m64 : FpIf64<(outs RFP64:$dst), (ins RFP64:$src1, i16mem:$src2), OneArgFPRW, - [(set RFP64:$dst, (OpNode RFP64:$src1, - (X86fild addr:$src2, i16)))]>; -def _FpI32m64 : FpIf64<(outs RFP64:$dst), (ins RFP64:$src1, i32mem:$src2), OneArgFPRW, - [(set RFP64:$dst, (OpNode RFP64:$src1, - (X86fild addr:$src2, i32)))]>; -def _FpI16m80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src1, i16mem:$src2), OneArgFPRW, - [(set RFP80:$dst, (OpNode RFP80:$src1, - (X86fild addr:$src2, i16)))]>; -def _FpI32m80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src1, i32mem:$src2), OneArgFPRW, - [(set RFP80:$dst, (OpNode RFP80:$src1, - (X86fild addr:$src2, i32)))]>; -def _FI16m : FPI<0xDE, fp, (outs), (ins i16mem:$src), - !strconcat("fi", !strconcat(asmstring, "{s}\t$src"))> { let mayLoad = 1; } -def _FI32m : FPI<0xDA, fp, (outs), (ins i32mem:$src), - !strconcat("fi", !strconcat(asmstring, "{l}\t$src"))> { let mayLoad = 1; } +def _FpI16m32 : FpIf32<(outs RFP32:$dst), (ins RFP32:$src1, i16mem:$src2), + OneArgFPRW, + [!if(Forward, + (set RFP32:$dst, + (OpNode RFP32:$src1, (X86fild addr:$src2, i16))), + (set RFP32:$dst, + (OpNode (X86fild addr:$src2, i16), RFP32:$src1)))]>; +def _FpI32m32 : FpIf32<(outs RFP32:$dst), (ins RFP32:$src1, i32mem:$src2), + OneArgFPRW, + [!if(Forward, + (set RFP32:$dst, + (OpNode RFP32:$src1, (X86fild addr:$src2, i32))), + (set RFP32:$dst, + (OpNode (X86fild addr:$src2, i32), RFP32:$src1)))]>; +def _FpI16m64 : FpIf64<(outs RFP64:$dst), (ins RFP64:$src1, i16mem:$src2), + OneArgFPRW, + [!if(Forward, + (set RFP64:$dst, + (OpNode RFP64:$src1, (X86fild addr:$src2, i16))), + (set RFP64:$dst, + (OpNode (X86fild addr:$src2, i16), RFP64:$src1)))]>; +def _FpI32m64 : FpIf64<(outs RFP64:$dst), (ins RFP64:$src1, i32mem:$src2), + OneArgFPRW, + [!if(Forward, + (set RFP64:$dst, + (OpNode RFP64:$src1, (X86fild addr:$src2, i32))), + (set RFP64:$dst, + (OpNode (X86fild addr:$src2, i32), RFP64:$src1)))]>; +def _FpI16m80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src1, i16mem:$src2), + OneArgFPRW, + [!if(Forward, + (set RFP80:$dst, + (OpNode RFP80:$src1, (X86fild addr:$src2, i16))), + (set RFP80:$dst, + (OpNode (X86fild addr:$src2, i16), RFP80:$src1)))]>; +def _FpI32m80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src1, i32mem:$src2), + OneArgFPRW, + [!if(Forward, + (set RFP80:$dst, + (OpNode RFP80:$src1, (X86fild addr:$src2, i32))), + (set RFP80:$dst, + (OpNode (X86fild addr:$src2, i32), RFP80:$src1)))]>; +let mayLoad = 1 in +def _FI16m : FPI<0xDE, fp, (outs), (ins i16mem:$src), + !strconcat("fi", asmstring, "{s}\t$src")>; +let mayLoad = 1 in +def _FI32m : FPI<0xDA, fp, (outs), (ins i32mem:$src), + !strconcat("fi", asmstring, "{l}\t$src")>; } +let Defs = [FPSW] in { +// FPBinary_rr just defines pseudo-instructions, no need to set a scheduling +// resources. defm ADD : FPBinary_rr; defm SUB : FPBinary_rr; defm MUL : FPBinary_rr; defm DIV : FPBinary_rr; +// Sets the scheduling resources for the actual NAME#_Fm defintions. +let SchedRW = [WriteFAddLd] in { defm ADD : FPBinary; defm SUB : FPBinary; -defm SUBR: FPBinary; +defm SUBR: FPBinary; +} +let SchedRW = [WriteFMulLd] in { defm MUL : FPBinary; +} +let SchedRW = [WriteFDivLd] in { defm DIV : FPBinary; -defm DIVR: FPBinary; +defm DIVR: FPBinary; +} +} -class FPST0rInst o, string asm> - : FPI, D8; -class FPrST0Inst o, string asm> - : FPI, DC; -class FPrST0PInst o, string asm> - : FPI, DE; +class FPST0rInst + : FPI<0xD8, fp, (outs), (ins RST:$op), asm>; +class FPrST0Inst + : FPI<0xDC, fp, (outs), (ins RST:$op), asm>; +class FPrST0PInst + : FPI<0xDE, fp, (outs), (ins RST:$op), asm>; // NOTE: GAS and apparently all other AT&T style assemblers have a broken notion // of some of the 'reverse' forms of the fsub and fdiv instructions. As such, // we have to put some 'r's in and take them out of weird places. -def ADD_FST0r : FPST0rInst <0xC0, "fadd\t$op">; -def ADD_FrST0 : FPrST0Inst <0xC0, "fadd\t{%st(0), $op|$op, %ST(0)}">; -def ADD_FPrST0 : FPrST0PInst<0xC0, "faddp\t$op">; -def SUBR_FST0r : FPST0rInst <0xE8, "fsubr\t$op">; -def SUB_FrST0 : FPrST0Inst <0xE8, "fsub{r}\t{%st(0), $op|$op, %ST(0)}">; -def SUB_FPrST0 : FPrST0PInst<0xE8, "fsub{r}p\t$op">; -def SUB_FST0r : FPST0rInst <0xE0, "fsub\t$op">; -def SUBR_FrST0 : FPrST0Inst <0xE0, "fsub{|r}\t{%st(0), $op|$op, %ST(0)}">; -def SUBR_FPrST0 : FPrST0PInst<0xE0, "fsub{|r}p\t$op">; -def MUL_FST0r : FPST0rInst <0xC8, "fmul\t$op">; -def MUL_FrST0 : FPrST0Inst <0xC8, "fmul\t{%st(0), $op|$op, %ST(0)}">; -def MUL_FPrST0 : FPrST0PInst<0xC8, "fmulp\t$op">; -def DIVR_FST0r : FPST0rInst <0xF8, "fdivr\t$op">; -def DIV_FrST0 : FPrST0Inst <0xF8, "fdiv{r}\t{%st(0), $op|$op, %ST(0)}">; -def DIV_FPrST0 : FPrST0PInst<0xF8, "fdiv{r}p\t$op">; -def DIV_FST0r : FPST0rInst <0xF0, "fdiv\t$op">; -def DIVR_FrST0 : FPrST0Inst <0xF0, "fdiv{|r}\t{%st(0), $op|$op, %ST(0)}">; -def DIVR_FPrST0 : FPrST0PInst<0xF0, "fdiv{|r}p\t$op">; +let SchedRW = [WriteFAdd] in { +def ADD_FST0r : FPST0rInst ; +def ADD_FrST0 : FPrST0Inst ; +def ADD_FPrST0 : FPrST0PInst; +def SUBR_FST0r : FPST0rInst ; +def SUB_FrST0 : FPrST0Inst ; +def SUB_FPrST0 : FPrST0PInst; +def SUB_FST0r : FPST0rInst ; +def SUBR_FrST0 : FPrST0Inst ; +def SUBR_FPrST0 : FPrST0PInst; +} // SchedRW +let SchedRW = [WriteFMul] in { +def MUL_FST0r : FPST0rInst ; +def MUL_FrST0 : FPrST0Inst ; +def MUL_FPrST0 : FPrST0PInst; +} // SchedRW +let SchedRW = [WriteFDiv] in { +def DIVR_FST0r : FPST0rInst ; +def DIV_FrST0 : FPrST0Inst ; +def DIV_FPrST0 : FPrST0PInst; +def DIV_FST0r : FPST0rInst ; +def DIVR_FrST0 : FPrST0Inst ; +def DIVR_FPrST0 : FPrST0PInst; +} // SchedRW + +def COM_FST0r : FPST0rInst ; +def COMP_FST0r : FPST0rInst ; // Unary operations. -multiclass FPUnary opcode, string asmstring> { +multiclass FPUnary { def _Fp32 : FpIf32<(outs RFP32:$dst), (ins RFP32:$src), OneArgFPRW, [(set RFP32:$dst, (OpNode RFP32:$src))]>; def _Fp64 : FpIf64<(outs RFP64:$dst), (ins RFP64:$src), OneArgFPRW, [(set RFP64:$dst, (OpNode RFP64:$src))]>; def _Fp80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src), OneArgFPRW, [(set RFP80:$dst, (OpNode RFP80:$src))]>; -def _F : FPI, D9; +def _F : FPI<0xD9, fp, (outs), (ins), asmstring>; } -defm CHS : FPUnary; -defm ABS : FPUnary; -defm SQRT: FPUnary; -defm SIN : FPUnary; -defm COS : FPUnary; +let Defs = [FPSW] in { +defm CHS : FPUnary; +defm ABS : FPUnary; +let SchedRW = [WriteFSqrt] in { +defm SQRT: FPUnary; +} +defm SIN : FPUnary; +defm COS : FPUnary; -let neverHasSideEffects = 1 in { +let hasSideEffects = 0 in { def TST_Fp32 : FpIf32<(outs), (ins RFP32:$src), OneArgFP, []>; def TST_Fp64 : FpIf64<(outs), (ins RFP64:$src), OneArgFP, []>; def TST_Fp80 : FpI_<(outs), (ins RFP80:$src), OneArgFP, []>; } -def TST_F : FPI<0xE4, RawFrm, (outs), (ins), "ftst">, D9; +def TST_F : FPI<0xD9, MRM_E4, (outs), (ins), "ftst">; +} // Defs = [FPSW] + +// Versions of FP instructions that take a single memory operand. Added for the +// disassembler; remove as they are included with patterns elsewhere. +def FCOM32m : FPI<0xD8, MRM2m, (outs), (ins f32mem:$src), "fcom{s}\t$src">; +def FCOMP32m : FPI<0xD8, MRM3m, (outs), (ins f32mem:$src), "fcomp{s}\t$src">; + +def FLDENVm : FPI<0xD9, MRM4m, (outs), (ins f32mem:$src), "fldenv\t$src">; +def FSTENVm : FPI<0xD9, MRM6m, (outs f32mem:$dst), (ins), "fnstenv\t$dst">; + +def FICOM32m : FPI<0xDA, MRM2m, (outs), (ins i32mem:$src), "ficom{l}\t$src">; +def FICOMP32m: FPI<0xDA, MRM3m, (outs), (ins i32mem:$src), "ficomp{l}\t$src">; + +def FCOM64m : FPI<0xDC, MRM2m, (outs), (ins f64mem:$src), "fcom{l}\t$src">; +def FCOMP64m : FPI<0xDC, MRM3m, (outs), (ins f64mem:$src), "fcomp{l}\t$src">; + +def FRSTORm : FPI<0xDD, MRM4m, (outs f32mem:$dst), (ins), "frstor\t$dst">; +def FSAVEm : FPI<0xDD, MRM6m, (outs f32mem:$dst), (ins), "fnsave\t$dst">; +def FNSTSWm : FPI<0xDD, MRM7m, (outs i16mem:$dst), (ins), "fnstsw\t$dst">; + +def FICOM16m : FPI<0xDE, MRM2m, (outs), (ins i16mem:$src), "ficom{s}\t$src">; +def FICOMP16m: FPI<0xDE, MRM3m, (outs), (ins i16mem:$src), "ficomp{s}\t$src">; + +def FBLDm : FPI<0xDF, MRM4m, (outs), (ins f80mem:$src), "fbld\t$src">; +def FBSTPm : FPI<0xDF, MRM6m, (outs f80mem:$dst), (ins), "fbstp\t$dst">; // Floating point cmovs. +class FpIf32CMov pattern> : + FpI_, Requires<[FPStackf32, HasCMov]>; +class FpIf64CMov pattern> : + FpI_, Requires<[FPStackf64, HasCMov]>; + multiclass FPCMov { - def _Fp32 : FpIf32<(outs RFP32:$dst), (ins RFP32:$src1, RFP32:$src2), + def _Fp32 : FpIf32CMov<(outs RFP32:$dst), (ins RFP32:$src1, RFP32:$src2), CondMovFP, [(set RFP32:$dst, (X86cmov RFP32:$src1, RFP32:$src2, cc, EFLAGS))]>; - def _Fp64 : FpIf64<(outs RFP64:$dst), (ins RFP64:$src1, RFP64:$src2), + def _Fp64 : FpIf64CMov<(outs RFP64:$dst), (ins RFP64:$src1, RFP64:$src2), CondMovFP, [(set RFP64:$dst, (X86cmov RFP64:$src1, RFP64:$src2, cc, EFLAGS))]>; def _Fp80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src1, RFP80:$src2), CondMovFP, [(set RFP80:$dst, (X86cmov RFP80:$src1, RFP80:$src2, - cc, EFLAGS))]>; + cc, EFLAGS))]>, + Requires<[HasCMov]>; } -let Uses = [EFLAGS], isTwoAddress = 1 in { + +let Defs = [FPSW] in { +let Uses = [EFLAGS], Constraints = "$src1 = $dst" in { defm CMOVB : FPCMov; defm CMOVBE : FPCMov; defm CMOVE : FPCMov; @@ -316,31 +376,33 @@ defm CMOVNB : FPCMov; defm CMOVNBE: FPCMov; defm CMOVNE : FPCMov; defm CMOVNP : FPCMov; -} +} // Uses = [EFLAGS], Constraints = "$src1 = $dst" +let Predicates = [HasCMov] in { // These are not factored because there's no clean way to pass DA/DB. -def CMOVB_F : FPI<0xC0, AddRegFrm, (outs RST:$op), (ins), - "fcmovb\t{$op, %st(0)|%ST(0), $op}">, DA; -def CMOVBE_F : FPI<0xD0, AddRegFrm, (outs RST:$op), (ins), - "fcmovbe\t{$op, %st(0)|%ST(0), $op}">, DA; -def CMOVE_F : FPI<0xC8, AddRegFrm, (outs RST:$op), (ins), - "fcmove\t{$op, %st(0)|%ST(0), $op}">, DA; -def CMOVP_F : FPI<0xD8, AddRegFrm, (outs RST:$op), (ins), - "fcmovu\t {$op, %st(0)|%ST(0), $op}">, DA; -def CMOVNB_F : FPI<0xC0, AddRegFrm, (outs RST:$op), (ins), - "fcmovnb\t{$op, %st(0)|%ST(0), $op}">, DB; -def CMOVNBE_F: FPI<0xD0, AddRegFrm, (outs RST:$op), (ins), - "fcmovnbe\t{$op, %st(0)|%ST(0), $op}">, DB; -def CMOVNE_F : FPI<0xC8, AddRegFrm, (outs RST:$op), (ins), - "fcmovne\t{$op, %st(0)|%ST(0), $op}">, DB; -def CMOVNP_F : FPI<0xD8, AddRegFrm, (outs RST:$op), (ins), - "fcmovnu\t{$op, %st(0)|%ST(0), $op}">, DB; +def CMOVB_F : FPI<0xDA, MRM0r, (outs), (ins RST:$op), + "fcmovb\t{$op, %st(0)|st(0), $op}">; +def CMOVBE_F : FPI<0xDA, MRM2r, (outs), (ins RST:$op), + "fcmovbe\t{$op, %st(0)|st(0), $op}">; +def CMOVE_F : FPI<0xDA, MRM1r, (outs), (ins RST:$op), + "fcmove\t{$op, %st(0)|st(0), $op}">; +def CMOVP_F : FPI<0xDA, MRM3r, (outs), (ins RST:$op), + "fcmovu\t{$op, %st(0)|st(0), $op}">; +def CMOVNB_F : FPI<0xDB, MRM0r, (outs), (ins RST:$op), + "fcmovnb\t{$op, %st(0)|st(0), $op}">; +def CMOVNBE_F: FPI<0xDB, MRM2r, (outs), (ins RST:$op), + "fcmovnbe\t{$op, %st(0)|st(0), $op}">; +def CMOVNE_F : FPI<0xDB, MRM1r, (outs), (ins RST:$op), + "fcmovne\t{$op, %st(0)|st(0), $op}">; +def CMOVNP_F : FPI<0xDB, MRM3r, (outs), (ins RST:$op), + "fcmovnu\t{$op, %st(0)|st(0), $op}">; +} // Predicates = [HasCMov] // Floating point loads & stores. -let isSimpleLoad = 1 in { +let canFoldAsLoad = 1 in { def LD_Fp32m : FpIf32<(outs RFP32:$dst), (ins f32mem:$src), ZeroArgFP, [(set RFP32:$dst, (loadf32 addr:$src))]>; -let isReMaterializable = 1, mayHaveSideEffects = 1 in +let isReMaterializable = 1 in def LD_Fp64m : FpIf64<(outs RFP64:$dst), (ins f64mem:$src), ZeroArgFP, [(set RFP64:$dst, (loadf64 addr:$src))]>; def LD_Fp80m : FpI_<(outs RFP80:$dst), (ins f80mem:$src), ZeroArgFP, @@ -383,7 +445,7 @@ def ST_Fp80m64 : FpI_<(outs), (ins f64mem:$op, RFP80:$src), OneArgFP, [(truncstoref64 RFP80:$src, addr:$op)]>; // FST does not support 80-bit memory target; FSTP must be used. -let mayStore = 1, neverHasSideEffects = 1 in { +let mayStore = 1, hasSideEffects = 0 in { def ST_FpP32m : FpIf32<(outs), (ins f32mem:$op, RFP32:$src), OneArgFP, []>; def ST_FpP64m32 : FpIf64<(outs), (ins f32mem:$op, RFP64:$src), OneArgFP, []>; def ST_FpP64m : FpIf64<(outs), (ins f64mem:$op, RFP64:$src), OneArgFP, []>; @@ -392,7 +454,7 @@ def ST_FpP80m64 : FpI_<(outs), (ins f64mem:$op, RFP80:$src), OneArgFP, []>; } def ST_FpP80m : FpI_<(outs), (ins f80mem:$op, RFP80:$src), OneArgFP, [(store RFP80:$src, addr:$op)]>; -let mayStore = 1, neverHasSideEffects = 1 in { +let mayStore = 1, hasSideEffects = 0 in { def IST_Fp16m32 : FpIf32<(outs), (ins i16mem:$op, RFP32:$src), OneArgFP, []>; def IST_Fp32m32 : FpIf32<(outs), (ins i32mem:$op, RFP32:$src), OneArgFP, []>; def IST_Fp64m32 : FpIf32<(outs), (ins i64mem:$op, RFP32:$src), OneArgFP, []>; @@ -404,67 +466,81 @@ def IST_Fp32m80 : FpI_<(outs), (ins i32mem:$op, RFP80:$src), OneArgFP, []>; def IST_Fp64m80 : FpI_<(outs), (ins i64mem:$op, RFP80:$src), OneArgFP, []>; } -let mayLoad = 1 in { -def LD_F32m : FPI<0xD9, MRM0m, (outs), (ins f32mem:$src), "fld{s}\t$src">; -def LD_F64m : FPI<0xDD, MRM0m, (outs), (ins f64mem:$src), "fld{l}\t$src">; -def LD_F80m : FPI<0xDB, MRM5m, (outs), (ins f80mem:$src), "fld{t}\t$src">; -def ILD_F16m : FPI<0xDF, MRM0m, (outs), (ins i16mem:$src), "fild{s}\t$src">; -def ILD_F32m : FPI<0xDB, MRM0m, (outs), (ins i32mem:$src), "fild{l}\t$src">; -def ILD_F64m : FPI<0xDF, MRM5m, (outs), (ins i64mem:$src), "fild{ll}\t$src">; +let mayLoad = 1, SchedRW = [WriteLoad] in { +def LD_F32m : FPI<0xD9, MRM0m, (outs), (ins f32mem:$src), "fld{s}\t$src", + IIC_FLD>; +def LD_F64m : FPI<0xDD, MRM0m, (outs), (ins f64mem:$src), "fld{l}\t$src", + IIC_FLD>; +def LD_F80m : FPI<0xDB, MRM5m, (outs), (ins f80mem:$src), "fld{t}\t$src", + IIC_FLD80>; +def ILD_F16m : FPI<0xDF, MRM0m, (outs), (ins i16mem:$src), "fild{s}\t$src", + IIC_FILD>; +def ILD_F32m : FPI<0xDB, MRM0m, (outs), (ins i32mem:$src), "fild{l}\t$src", + IIC_FILD>; +def ILD_F64m : FPI<0xDF, MRM5m, (outs), (ins i64mem:$src), "fild{ll}\t$src", + IIC_FILD>; } -let mayStore = 1 in { -def ST_F32m : FPI<0xD9, MRM2m, (outs), (ins f32mem:$dst), "fst{s}\t$dst">; -def ST_F64m : FPI<0xDD, MRM2m, (outs), (ins f64mem:$dst), "fst{l}\t$dst">; -def ST_FP32m : FPI<0xD9, MRM3m, (outs), (ins f32mem:$dst), "fstp{s}\t$dst">; -def ST_FP64m : FPI<0xDD, MRM3m, (outs), (ins f64mem:$dst), "fstp{l}\t$dst">; -def ST_FP80m : FPI<0xDB, MRM7m, (outs), (ins f80mem:$dst), "fstp{t}\t$dst">; -def IST_F16m : FPI<0xDF, MRM2m, (outs), (ins i16mem:$dst), "fist{s}\t$dst">; -def IST_F32m : FPI<0xDB, MRM2m, (outs), (ins i32mem:$dst), "fist{l}\t$dst">; -def IST_FP16m : FPI<0xDF, MRM3m, (outs), (ins i16mem:$dst), "fistp{s}\t$dst">; -def IST_FP32m : FPI<0xDB, MRM3m, (outs), (ins i32mem:$dst), "fistp{l}\t$dst">; -def IST_FP64m : FPI<0xDF, MRM7m, (outs), (ins i64mem:$dst), "fistp{ll}\t$dst">; +let mayStore = 1, SchedRW = [WriteStore] in { +def ST_F32m : FPI<0xD9, MRM2m, (outs), (ins f32mem:$dst), "fst{s}\t$dst", + IIC_FST>; +def ST_F64m : FPI<0xDD, MRM2m, (outs), (ins f64mem:$dst), "fst{l}\t$dst", + IIC_FST>; +def ST_FP32m : FPI<0xD9, MRM3m, (outs), (ins f32mem:$dst), "fstp{s}\t$dst", + IIC_FST>; +def ST_FP64m : FPI<0xDD, MRM3m, (outs), (ins f64mem:$dst), "fstp{l}\t$dst", + IIC_FST>; +def ST_FP80m : FPI<0xDB, MRM7m, (outs), (ins f80mem:$dst), "fstp{t}\t$dst", + IIC_FST80>; +def IST_F16m : FPI<0xDF, MRM2m, (outs), (ins i16mem:$dst), "fist{s}\t$dst", + IIC_FIST>; +def IST_F32m : FPI<0xDB, MRM2m, (outs), (ins i32mem:$dst), "fist{l}\t$dst", + IIC_FIST>; +def IST_FP16m : FPI<0xDF, MRM3m, (outs), (ins i16mem:$dst), "fistp{s}\t$dst", + IIC_FIST>; +def IST_FP32m : FPI<0xDB, MRM3m, (outs), (ins i32mem:$dst), "fistp{l}\t$dst", + IIC_FIST>; +def IST_FP64m : FPI<0xDF, MRM7m, (outs), (ins i64mem:$dst), "fistp{ll}\t$dst", + IIC_FIST>; } // FISTTP requires SSE3 even though it's a FPStack op. +let Predicates = [HasSSE3] in { def ISTT_Fp16m32 : FpI_<(outs), (ins i16mem:$op, RFP32:$src), OneArgFP, - [(X86fp_to_i16mem RFP32:$src, addr:$op)]>, - Requires<[HasSSE3]>; + [(X86fp_to_i16mem RFP32:$src, addr:$op)]>; def ISTT_Fp32m32 : FpI_<(outs), (ins i32mem:$op, RFP32:$src), OneArgFP, - [(X86fp_to_i32mem RFP32:$src, addr:$op)]>, - Requires<[HasSSE3]>; + [(X86fp_to_i32mem RFP32:$src, addr:$op)]>; def ISTT_Fp64m32 : FpI_<(outs), (ins i64mem:$op, RFP32:$src), OneArgFP, - [(X86fp_to_i64mem RFP32:$src, addr:$op)]>, - Requires<[HasSSE3]>; + [(X86fp_to_i64mem RFP32:$src, addr:$op)]>; def ISTT_Fp16m64 : FpI_<(outs), (ins i16mem:$op, RFP64:$src), OneArgFP, - [(X86fp_to_i16mem RFP64:$src, addr:$op)]>, - Requires<[HasSSE3]>; + [(X86fp_to_i16mem RFP64:$src, addr:$op)]>; def ISTT_Fp32m64 : FpI_<(outs), (ins i32mem:$op, RFP64:$src), OneArgFP, - [(X86fp_to_i32mem RFP64:$src, addr:$op)]>, - Requires<[HasSSE3]>; + [(X86fp_to_i32mem RFP64:$src, addr:$op)]>; def ISTT_Fp64m64 : FpI_<(outs), (ins i64mem:$op, RFP64:$src), OneArgFP, - [(X86fp_to_i64mem RFP64:$src, addr:$op)]>, - Requires<[HasSSE3]>; + [(X86fp_to_i64mem RFP64:$src, addr:$op)]>; def ISTT_Fp16m80 : FpI_<(outs), (ins i16mem:$op, RFP80:$src), OneArgFP, - [(X86fp_to_i16mem RFP80:$src, addr:$op)]>, - Requires<[HasSSE3]>; + [(X86fp_to_i16mem RFP80:$src, addr:$op)]>; def ISTT_Fp32m80 : FpI_<(outs), (ins i32mem:$op, RFP80:$src), OneArgFP, - [(X86fp_to_i32mem RFP80:$src, addr:$op)]>, - Requires<[HasSSE3]>; + [(X86fp_to_i32mem RFP80:$src, addr:$op)]>; def ISTT_Fp64m80 : FpI_<(outs), (ins i64mem:$op, RFP80:$src), OneArgFP, - [(X86fp_to_i64mem RFP80:$src, addr:$op)]>, - Requires<[HasSSE3]>; - -let mayStore = 1 in { -def ISTT_FP16m : FPI<0xDF, MRM1m, (outs), (ins i16mem:$dst), "fisttp{s}\t$dst">; -def ISTT_FP32m : FPI<0xDB, MRM1m, (outs), (ins i32mem:$dst), "fisttp{l}\t$dst">; -def ISTT_FP64m : FPI<0xDD, MRM1m, (outs), (ins i64mem:$dst), "fisttp{ll}\t$dst">; + [(X86fp_to_i64mem RFP80:$src, addr:$op)]>; +} // Predicates = [HasSSE3] + +let mayStore = 1, SchedRW = [WriteStore] in { +def ISTT_FP16m : FPI<0xDF, MRM1m, (outs), (ins i16mem:$dst), "fisttp{s}\t$dst", + IIC_FST>; +def ISTT_FP32m : FPI<0xDB, MRM1m, (outs), (ins i32mem:$dst), "fisttp{l}\t$dst", + IIC_FST>; +def ISTT_FP64m : FPI<0xDD, MRM1m, (outs), (ins i64mem:$dst), + "fisttp{ll}\t$dst", IIC_FST>; } // FP Stack manipulation instructions. -def LD_Frr : FPI<0xC0, AddRegFrm, (outs), (ins RST:$op), "fld\t$op">, D9; -def ST_Frr : FPI<0xD0, AddRegFrm, (outs), (ins RST:$op), "fst\t$op">, DD; -def ST_FPrr : FPI<0xD8, AddRegFrm, (outs), (ins RST:$op), "fstp\t$op">, DD; -def XCH_F : FPI<0xC8, AddRegFrm, (outs), (ins RST:$op), "fxch\t$op">, D9; +let SchedRW = [WriteMove] in { +def LD_Frr : FPI<0xD9, MRM0r, (outs), (ins RST:$op), "fld\t$op", IIC_FLD>; +def ST_Frr : FPI<0xDD, MRM2r, (outs), (ins RST:$op), "fst\t$op", IIC_FST>; +def ST_FPrr : FPI<0xDD, MRM3r, (outs), (ins RST:$op), "fstp\t$op", IIC_FST>; +def XCH_F : FPI<0xD9, MRM1r, (outs), (ins RST:$op), "fxch\t$op", IIC_FXCH>; +} // Floating point constant loads. let isReMaterializable = 1 in { @@ -482,61 +558,124 @@ def LD_Fp180 : FpI_<(outs RFP80:$dst), (ins), ZeroArgFP, [(set RFP80:$dst, fpimm1)]>; } -def LD_F0 : FPI<0xEE, RawFrm, (outs), (ins), "fldz">, D9; -def LD_F1 : FPI<0xE8, RawFrm, (outs), (ins), "fld1">, D9; - +let SchedRW = [WriteZero] in { +def LD_F0 : FPI<0xD9, MRM_EE, (outs), (ins), "fldz", IIC_FLDZ>; +def LD_F1 : FPI<0xD9, MRM_E8, (outs), (ins), "fld1", IIC_FIST>; +} // Floating point compares. -let Defs = [EFLAGS] in { +let SchedRW = [WriteFAdd] in { def UCOM_Fpr32 : FpIf32<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP, - []>; // FPSW = cmp ST(0) with ST(i) + [(set FPSW, (trunc (X86cmp RFP32:$lhs, RFP32:$rhs)))]>; def UCOM_Fpr64 : FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP, - []>; // FPSW = cmp ST(0) with ST(i) + [(set FPSW, (trunc (X86cmp RFP64:$lhs, RFP64:$rhs)))]>; def UCOM_Fpr80 : FpI_ <(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP, - []>; // FPSW = cmp ST(0) with ST(i) - + [(set FPSW, (trunc (X86cmp RFP80:$lhs, RFP80:$rhs)))]>; +} // SchedRW +} // Defs = [FPSW] + +let SchedRW = [WriteFAdd] in { +// CC = ST(0) cmp ST(i) +let Defs = [EFLAGS, FPSW] in { def UCOM_FpIr32: FpIf32<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP, - [(X86cmp RFP32:$lhs, RFP32:$rhs), - (implicit EFLAGS)]>; // CC = ST(0) cmp ST(i) + [(set EFLAGS, (X86cmp RFP32:$lhs, RFP32:$rhs))]>; def UCOM_FpIr64: FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP, - [(X86cmp RFP64:$lhs, RFP64:$rhs), - (implicit EFLAGS)]>; // CC = ST(0) cmp ST(i) + [(set EFLAGS, (X86cmp RFP64:$lhs, RFP64:$rhs))]>; def UCOM_FpIr80: FpI_<(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP, - [(X86cmp RFP80:$lhs, RFP80:$rhs), - (implicit EFLAGS)]>; // CC = ST(0) cmp ST(i) + [(set EFLAGS, (X86cmp RFP80:$lhs, RFP80:$rhs))]>; +} + +let Defs = [FPSW], Uses = [ST0] in { +def UCOM_Fr : FPI<0xDD, MRM4r, // FPSW = cmp ST(0) with ST(i) + (outs), (ins RST:$reg), "fucom\t$reg", IIC_FUCOM>; +def UCOM_FPr : FPI<0xDD, MRM5r, // FPSW = cmp ST(0) with ST(i), pop + (outs), (ins RST:$reg), "fucomp\t$reg", IIC_FUCOM>; +def UCOM_FPPr : FPI<0xDA, MRM_E9, // cmp ST(0) with ST(1), pop, pop + (outs), (ins), "fucompp", IIC_FUCOM>; +} + +let Defs = [EFLAGS, FPSW], Uses = [ST0] in { +def UCOM_FIr : FPI<0xDB, MRM5r, // CC = cmp ST(0) with ST(i) + (outs), (ins RST:$reg), "fucomi\t$reg", IIC_FUCOMI>; +def UCOM_FIPr : FPI<0xDF, MRM5r, // CC = cmp ST(0) with ST(i), pop + (outs), (ins RST:$reg), "fucompi\t$reg", IIC_FUCOMI>; } -let Defs = [EFLAGS], Uses = [ST0] in { -def UCOM_Fr : FPI<0xE0, AddRegFrm, // FPSW = cmp ST(0) with ST(i) - (outs), (ins RST:$reg), - "fucom\t$reg">, DD; -def UCOM_FPr : FPI<0xE8, AddRegFrm, // FPSW = cmp ST(0) with ST(i), pop - (outs), (ins RST:$reg), - "fucomp\t$reg">, DD; -def UCOM_FPPr : FPI<0xE9, RawFrm, // cmp ST(0) with ST(1), pop, pop - (outs), (ins), - "fucompp">, DA; - -def UCOM_FIr : FPI<0xE8, AddRegFrm, // CC = cmp ST(0) with ST(i) - (outs), (ins RST:$reg), - "fucomi\t{$reg, %st(0)|%ST(0), $reg}">, DB; -def UCOM_FIPr : FPI<0xE8, AddRegFrm, // CC = cmp ST(0) with ST(i), pop - (outs), (ins RST:$reg), - "fucomip\t{$reg, %st(0)|%ST(0), $reg}">, DF; +let Defs = [EFLAGS, FPSW] in { +def COM_FIr : FPI<0xDB, MRM6r, (outs), (ins RST:$reg), + "fcomi\t$reg", IIC_FCOMI>; +def COM_FIPr : FPI<0xDF, MRM6r, (outs), (ins RST:$reg), + "fcompi\t$reg", IIC_FCOMI>; } +} // SchedRW // Floating point flag ops. -let Defs = [AX] in -def FNSTSW8r : I<0xE0, RawFrm, // AX = fp flags - (outs), (ins), "fnstsw", []>, DF; +let SchedRW = [WriteALU] in { +let Defs = [AX], Uses = [FPSW] in +def FNSTSW16r : I<0xDF, MRM_E0, // AX = fp flags + (outs), (ins), "fnstsw\t{%ax|ax}", + [(set AX, (X86fp_stsw FPSW))], IIC_FNSTSW>; def FNSTCW16m : I<0xD9, MRM7m, // [mem16] = X87 control world (outs), (ins i16mem:$dst), "fnstcw\t$dst", - [(X86fp_cwd_get16 addr:$dst)]>; - + [(X86fp_cwd_get16 addr:$dst)], IIC_FNSTCW>; +} // SchedRW let mayLoad = 1 in def FLDCW16m : I<0xD9, MRM5m, // X87 control world = [mem16] - (outs), (ins i16mem:$dst), "fldcw\t$dst", []>; + (outs), (ins i16mem:$dst), "fldcw\t$dst", [], IIC_FLDCW>, + Sched<[WriteLoad]>; + +// FPU control instructions +let SchedRW = [WriteMicrocoded] in { +let Defs = [FPSW] in +def FNINIT : I<0xDB, MRM_E3, (outs), (ins), "fninit", [], IIC_FNINIT>; +def FFREE : FPI<0xDD, MRM0r, (outs), (ins RST:$reg), + "ffree\t$reg", IIC_FFREE>; +// Clear exceptions + +let Defs = [FPSW] in +def FNCLEX : I<0xDB, MRM_E2, (outs), (ins), "fnclex", [], IIC_FNCLEX>; +} // SchedRW + +// Operandless floating-point instructions for the disassembler. +let SchedRW = [WriteMicrocoded] in { +def WAIT : I<0x9B, RawFrm, (outs), (ins), "wait", [], IIC_WAIT>; + +def FNOP : I<0xD9, MRM_D0, (outs), (ins), "fnop", [], IIC_FNOP>; +def FXAM : I<0xD9, MRM_E5, (outs), (ins), "fxam", [], IIC_FXAM>; +def FLDL2T : I<0xD9, MRM_E9, (outs), (ins), "fldl2t", [], IIC_FLDL>; +def FLDL2E : I<0xD9, MRM_EA, (outs), (ins), "fldl2e", [], IIC_FLDL>; +def FLDPI : I<0xD9, MRM_EB, (outs), (ins), "fldpi", [], IIC_FLDL>; +def FLDLG2 : I<0xD9, MRM_EC, (outs), (ins), "fldlg2", [], IIC_FLDL>; +def FLDLN2 : I<0xD9, MRM_ED, (outs), (ins), "fldln2", [], IIC_FLDL>; +def F2XM1 : I<0xD9, MRM_F0, (outs), (ins), "f2xm1", [], IIC_F2XM1>; +def FYL2X : I<0xD9, MRM_F1, (outs), (ins), "fyl2x", [], IIC_FYL2X>; +def FPTAN : I<0xD9, MRM_F2, (outs), (ins), "fptan", [], IIC_FPTAN>; +def FPATAN : I<0xD9, MRM_F3, (outs), (ins), "fpatan", [], IIC_FPATAN>; +def FXTRACT : I<0xD9, MRM_F4, (outs), (ins), "fxtract", [], IIC_FXTRACT>; +def FPREM1 : I<0xD9, MRM_F5, (outs), (ins), "fprem1", [], IIC_FPREM1>; +def FDECSTP : I<0xD9, MRM_F6, (outs), (ins), "fdecstp", [], IIC_FPSTP>; +def FINCSTP : I<0xD9, MRM_F7, (outs), (ins), "fincstp", [], IIC_FPSTP>; +def FPREM : I<0xD9, MRM_F8, (outs), (ins), "fprem", [], IIC_FPREM>; +def FYL2XP1 : I<0xD9, MRM_F9, (outs), (ins), "fyl2xp1", [], IIC_FYL2XP1>; +def FSINCOS : I<0xD9, MRM_FB, (outs), (ins), "fsincos", [], IIC_FSINCOS>; +def FRNDINT : I<0xD9, MRM_FC, (outs), (ins), "frndint", [], IIC_FRNDINT>; +def FSCALE : I<0xD9, MRM_FD, (outs), (ins), "fscale", [], IIC_FSCALE>; +def FCOMPP : I<0xDE, MRM_D9, (outs), (ins), "fcompp", [], IIC_FCOMPP>; + +let Predicates = [HasFXSR] in { + def FXSAVE : I<0xAE, MRM0m, (outs), (ins opaque512mem:$dst), + "fxsave\t$dst", [(int_x86_fxsave addr:$dst)], IIC_FXSAVE>, TB; + def FXSAVE64 : RI<0xAE, MRM0m, (outs), (ins opaque512mem:$dst), + "fxsave64\t$dst", [(int_x86_fxsave64 addr:$dst)], + IIC_FXSAVE>, TB, Requires<[In64BitMode]>; + def FXRSTOR : I<0xAE, MRM1m, (outs), (ins opaque512mem:$src), + "fxrstor\t$src", [(int_x86_fxrstor addr:$src)], IIC_FXRSTOR>, TB; + def FXRSTOR64 : RI<0xAE, MRM1m, (outs), (ins opaque512mem:$src), + "fxrstor64\t$src", [(int_x86_fxrstor64 addr:$src)], + IIC_FXRSTOR>, TB, Requires<[In64BitMode]>; +} // Predicates = [FeatureFXSR] +} // SchedRW //===----------------------------------------------------------------------===// // Non-Instruction Patterns @@ -549,11 +688,15 @@ def : Pat<(X86fld addr:$src, f80), (LD_Fp80m addr:$src)>; // Required for CALL which return f32 / f64 / f80 values. def : Pat<(X86fst RFP32:$src, addr:$op, f32), (ST_Fp32m addr:$op, RFP32:$src)>; -def : Pat<(X86fst RFP64:$src, addr:$op, f32), (ST_Fp64m32 addr:$op, RFP64:$src)>; +def : Pat<(X86fst RFP64:$src, addr:$op, f32), (ST_Fp64m32 addr:$op, + RFP64:$src)>; def : Pat<(X86fst RFP64:$src, addr:$op, f64), (ST_Fp64m addr:$op, RFP64:$src)>; -def : Pat<(X86fst RFP80:$src, addr:$op, f32), (ST_Fp80m32 addr:$op, RFP80:$src)>; -def : Pat<(X86fst RFP80:$src, addr:$op, f64), (ST_Fp80m64 addr:$op, RFP80:$src)>; -def : Pat<(X86fst RFP80:$src, addr:$op, f80), (ST_FpP80m addr:$op, RFP80:$src)>; +def : Pat<(X86fst RFP80:$src, addr:$op, f32), (ST_Fp80m32 addr:$op, + RFP80:$src)>; +def : Pat<(X86fst RFP80:$src, addr:$op, f64), (ST_Fp80m64 addr:$op, + RFP80:$src)>; +def : Pat<(X86fst RFP80:$src, addr:$op, f80), (ST_FpP80m addr:$op, + RFP80:$src)>; // Floating point constant -0.0 and -1.0 def : Pat<(f32 fpimmneg0), (CHS_Fp32 (LD_Fp032))>, Requires<[FPStackf32]>; @@ -568,19 +711,19 @@ def : Pat<(X86fildflag addr:$src, i64), (ILD_Fp64m64 addr:$src)>; // FP extensions map onto simple pseudo-value conversions if they are to/from // the FP stack. -def : Pat<(f64 (fextend RFP32:$src)), (MOV_Fp3264 RFP32:$src)>, +def : Pat<(f64 (fextend RFP32:$src)), (COPY_TO_REGCLASS RFP32:$src, RFP64)>, Requires<[FPStackf32]>; -def : Pat<(f80 (fextend RFP32:$src)), (MOV_Fp3280 RFP32:$src)>, +def : Pat<(f80 (fextend RFP32:$src)), (COPY_TO_REGCLASS RFP32:$src, RFP80)>, Requires<[FPStackf32]>; -def : Pat<(f80 (fextend RFP64:$src)), (MOV_Fp6480 RFP64:$src)>, +def : Pat<(f80 (fextend RFP64:$src)), (COPY_TO_REGCLASS RFP64:$src, RFP80)>, Requires<[FPStackf64]>; // FP truncations map onto simple pseudo-value conversions if they are to/from // the FP stack. We have validated that only value-preserving truncations make // it through isel. -def : Pat<(f32 (fround RFP64:$src)), (MOV_Fp6432 RFP64:$src)>, +def : Pat<(f32 (fround RFP64:$src)), (COPY_TO_REGCLASS RFP64:$src, RFP32)>, Requires<[FPStackf32]>; -def : Pat<(f32 (fround RFP80:$src)), (MOV_Fp8032 RFP80:$src)>, +def : Pat<(f32 (fround RFP80:$src)), (COPY_TO_REGCLASS RFP80:$src, RFP32)>, Requires<[FPStackf32]>; -def : Pat<(f64 (fround RFP80:$src)), (MOV_Fp8064 RFP80:$src)>, +def : Pat<(f64 (fround RFP80:$src)), (COPY_TO_REGCLASS RFP80:$src, RFP64)>, Requires<[FPStackf64]>;