-//==- X86InstrFPStack.td - Describe the X86 Instruction Set -------*- C++ -*-=//
+//==- X86InstrFPStack.td - Describe the X86 Instruction Set --*- tablegen -*-=//
//
// The LLVM Compiler Infrastructure
//
-// This file was developed by the Evan Cheng and is distributed under
-// the University of Illinois Open Source License. See LICENSE.TXT for details.
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
//
//===----------------------------------------------------------------------===//
+//===----------------------------------------------------------------------===//
+// FPStack specific DAG Nodes.
+//===----------------------------------------------------------------------===//
+
+def SDTX86FpGet2 : SDTypeProfile<2, 0, [SDTCisVT<0, f80>,
+ SDTCisVT<1, f80>]>;
+def SDTX86Fld : SDTypeProfile<1, 2, [SDTCisFP<0>,
+ SDTCisPtrTy<1>,
+ SDTCisVT<2, OtherVT>]>;
+def SDTX86Fst : SDTypeProfile<0, 3, [SDTCisFP<0>,
+ SDTCisPtrTy<1>,
+ SDTCisVT<2, OtherVT>]>;
+def SDTX86Fild : SDTypeProfile<1, 2, [SDTCisFP<0>, SDTCisPtrTy<1>,
+ SDTCisVT<2, OtherVT>]>;
+def SDTX86FpToIMem : SDTypeProfile<0, 2, [SDTCisFP<0>, SDTCisPtrTy<1>]>;
+
+def SDTX86CwdStore : SDTypeProfile<0, 1, [SDTCisPtrTy<0>]>;
+
+def X86fld : SDNode<"X86ISD::FLD", SDTX86Fld,
+ [SDNPHasChain, SDNPMayLoad]>;
+def X86fst : SDNode<"X86ISD::FST", SDTX86Fst,
+ [SDNPHasChain, SDNPInFlag, SDNPMayStore]>;
+def X86fild : SDNode<"X86ISD::FILD", SDTX86Fild,
+ [SDNPHasChain, SDNPMayLoad]>;
+def X86fildflag : SDNode<"X86ISD::FILD_FLAG", SDTX86Fild,
+ [SDNPHasChain, SDNPOutFlag, SDNPMayLoad]>;
+def X86fp_to_i16mem : SDNode<"X86ISD::FP_TO_INT16_IN_MEM", SDTX86FpToIMem,
+ [SDNPHasChain, SDNPMayStore]>;
+def X86fp_to_i32mem : SDNode<"X86ISD::FP_TO_INT32_IN_MEM", SDTX86FpToIMem,
+ [SDNPHasChain, SDNPMayStore]>;
+def X86fp_to_i64mem : SDNode<"X86ISD::FP_TO_INT64_IN_MEM", SDTX86FpToIMem,
+ [SDNPHasChain, SDNPMayStore]>;
+def X86fp_cwd_get16 : SDNode<"X86ISD::FNSTCW16m", SDTX86CwdStore,
+ [SDNPHasChain, SDNPMayStore, SDNPSideEffect]>;
+
+//===----------------------------------------------------------------------===//
+// FPStack pattern fragments
+//===----------------------------------------------------------------------===//
+
+def fpimm0 : PatLeaf<(fpimm), [{
+ return N->isExactlyValue(+0.0);
+}]>;
+
+def fpimmneg0 : PatLeaf<(fpimm), [{
+ return N->isExactlyValue(-0.0);
+}]>;
+
+def fpimm1 : PatLeaf<(fpimm), [{
+ return N->isExactlyValue(+1.0);
+}]>;
+
+def fpimmneg1 : PatLeaf<(fpimm), [{
+ return N->isExactlyValue(-1.0);
+}]>;
+
// Some 'special' instructions
let usesCustomDAGSchedInserter = 1 in { // Expanded by the scheduler.
- def FP_TO_INT16_IN_MEM : I<0, Pseudo,
- (ops i16mem:$dst, RFP:$src),
- "#FP_TO_INT16_IN_MEM PSEUDO!",
- [(X86fp_to_i16mem RFP:$src, addr:$dst)]>;
- def FP_TO_INT32_IN_MEM : I<0, Pseudo,
- (ops i32mem:$dst, RFP:$src),
- "#FP_TO_INT32_IN_MEM PSEUDO!",
- [(X86fp_to_i32mem RFP:$src, addr:$dst)]>;
- def FP_TO_INT64_IN_MEM : I<0, Pseudo,
- (ops i64mem:$dst, RFP:$src),
- "#FP_TO_INT64_IN_MEM PSEUDO!",
- [(X86fp_to_i64mem RFP:$src, addr:$dst)]>;
+ def FP32_TO_INT16_IN_MEM : I<0, Pseudo,
+ (outs), (ins i16mem:$dst, RFP32:$src),
+ "##FP32_TO_INT16_IN_MEM PSEUDO!",
+ [(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!",
+ [(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!",
+ [(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!",
+ [(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!",
+ [(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!",
+ [(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!",
+ [(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!",
+ [(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!",
+ [(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, (ops), "#FP_REG_KILL", []>;
-
-// All FP Stack operations are represented with two instructions here. The
-// first instruction, generated by the instruction selector, uses "RFP"
-// registers: a traditional register file to reference floating point values.
-// These instructions are all psuedo instructions and use the "Fp" prefix.
+ 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,
+// 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.
+// In some cases there are additional variants with a mixture of different
+// register sizes.
// The second instruction is defined with FPI, which is the actual instruction
-// emitted by the assembler. The FP stackifier pass converts one to the other
-// after register allocation occurs.
+// 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
+// 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).
-// FPI - Floating Point Instruction template.
-class FPI<bits<8> o, Format F, dag ops, string asm> : I<o, F, ops, asm, []> {}
-
-// FpI_ - Floating Point Psuedo Instruction template. Not Predicated.
-class FpI_<dag ops, FPFormat fp, list<dag> pattern>
- : X86Inst<0, Pseudo, NoImm, ops, ""> {
- let FPForm = fp; let FPFormBits = FPForm.Value;
- let Pattern = pattern;
+// 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:$dst), (ins), SpecialFP, []>; // FPR = ST(0)
+
+// FpGET_ST1* should only be issued *after* an FpGET_ST0* has been issued when
+// there are two values live out on the stack from a call or inlineasm. This
+// magic is handled by the stackifier. It is not valid to emit FpGET_ST1* and
+// then FpGET_ST0*. In addition, it is invalid for any FP-using operations to
+// occur between them.
+def FpGET_ST1_32 : FpI_<(outs RFP32:$dst), (ins), SpecialFP, []>; // FPR = ST(1)
+def FpGET_ST1_64 : FpI_<(outs RFP64:$dst), (ins), SpecialFP, []>; // FPR = ST(1)
+def FpGET_ST1_80 : FpI_<(outs RFP80:$dst), (ins), SpecialFP, []>; // 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:$src), SpecialFP, []>; // ST(0) = FPR
}
-// Random Pseudo Instructions.
-def FpGETRESULT : FpI_<(ops RFP:$dst), SpecialFP,
- [(set RFP:$dst, X86fpget)]>; // FPR = ST(0)
-
-let noResults = 1 in
- def FpSETRESULT : FpI_<(ops RFP:$src), SpecialFP,
- [(X86fpset RFP:$src)]>, Imp<[], [ST0]>; // ST(0) = FPR
-
-// FpI - Floating Point Psuedo Instruction template. Predicated on FPStack.
-class FpI<dag ops, FPFormat fp, list<dag> pattern> :
- FpI_<ops, fp, pattern>, Requires<[FPStack]>;
+let Defs = [ST1] in {
+def FpSET_ST1_32 : FpI_<(outs), (ins RFP32:$src), SpecialFP, []>; // ST(1) = FPR
+def FpSET_ST1_64 : FpI_<(outs), (ins RFP64:$src), SpecialFP, []>; // ST(1) = FPR
+def FpSET_ST1_80 : FpI_<(outs), (ins RFP80:$src), SpecialFP, []>; // ST(1) = FPR
+}
+// FpIf32, FpIf64 - Floating Point Psuedo 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.
+class FpIf32<dag outs, dag ins, FPFormat fp, list<dag> pattern> :
+ FpI_<outs, ins, fp, pattern>, Requires<[FPStackf32]>;
+class FpIf64<dag outs, dag ins, FPFormat fp, list<dag> pattern> :
+ FpI_<outs, ins, fp, pattern>, 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, []>;
+}
-def FpMOV : FpI<(ops RFP:$dst, RFP:$src), SpecialFP, []>; // f1 = fmov f2
+// Factoring for arithmetic.
+multiclass FPBinary_rr<SDNode OpNode> {
+// Register op register -> register
+// These are separated out because they have no reversed form.
+def _Fp32 : FpIf32<(outs RFP32:$dst), (ins RFP32:$src1, RFP32:$src2), TwoArgFP,
+ [(set RFP32:$dst, (OpNode RFP32:$src1, RFP32:$src2))]>;
+def _Fp64 : FpIf64<(outs RFP64:$dst), (ins RFP64:$src1, RFP64:$src2), TwoArgFP,
+ [(set RFP64:$dst, (OpNode RFP64:$src1, RFP64:$src2))]>;
+def _Fp80 : FpI_<(outs RFP80:$dst), (ins RFP80:$src1, RFP80:$src2), TwoArgFP,
+ [(set RFP80:$dst, (OpNode RFP80:$src1, RFP80:$src2))]>;
+}
+// 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<SDNode OpNode, Format fp, string asmstring> {
+// 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; }
+// 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; }
+}
-// Arithmetic
-// Add, Sub, Mul, Div.
-def FpADD : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), TwoArgFP,
- [(set RFP:$dst, (fadd RFP:$src1, RFP:$src2))]>;
-def FpSUB : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), TwoArgFP,
- [(set RFP:$dst, (fsub RFP:$src1, RFP:$src2))]>;
-def FpMUL : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), TwoArgFP,
- [(set RFP:$dst, (fmul RFP:$src1, RFP:$src2))]>;
-def FpDIV : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), TwoArgFP,
- [(set RFP:$dst, (fdiv RFP:$src1, RFP:$src2))]>;
+defm ADD : FPBinary_rr<fadd>;
+defm SUB : FPBinary_rr<fsub>;
+defm MUL : FPBinary_rr<fmul>;
+defm DIV : FPBinary_rr<fdiv>;
+defm ADD : FPBinary<fadd, MRM0m, "add">;
+defm SUB : FPBinary<fsub, MRM4m, "sub">;
+defm SUBR: FPBinary<fsub ,MRM5m, "subr">;
+defm MUL : FPBinary<fmul, MRM1m, "mul">;
+defm DIV : FPBinary<fdiv, MRM6m, "div">;
+defm DIVR: FPBinary<fdiv, MRM7m, "divr">;
class FPST0rInst<bits<8> o, string asm>
- : FPI<o, AddRegFrm, (ops RST:$op), asm>, D8;
+ : FPI<o, AddRegFrm, (outs), (ins RST:$op), asm>, D8;
class FPrST0Inst<bits<8> o, string asm>
- : FPI<o, AddRegFrm, (ops RST:$op), asm>, DC;
+ : FPI<o, AddRegFrm, (outs), (ins RST:$op), asm>, DC;
class FPrST0PInst<bits<8> o, string asm>
- : FPI<o, AddRegFrm, (ops RST:$op), asm>, DE;
-
-// Binary Ops with a memory source.
-def FpADD32m : FpI<(ops RFP:$dst, RFP:$src1, f32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fadd RFP:$src1,
- (extloadf64f32 addr:$src2)))]>;
- // ST(0) = ST(0) + [mem32]
-def FpADD64m : FpI<(ops RFP:$dst, RFP:$src1, f64mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fadd RFP:$src1, (loadf64 addr:$src2)))]>;
- // ST(0) = ST(0) + [mem64]
-def FpMUL32m : FpI<(ops RFP:$dst, RFP:$src1, f32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fmul RFP:$src1,
- (extloadf64f32 addr:$src2)))]>;
- // ST(0) = ST(0) * [mem32]
-def FpMUL64m : FpI<(ops RFP:$dst, RFP:$src1, f64mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fmul RFP:$src1, (loadf64 addr:$src2)))]>;
- // ST(0) = ST(0) * [mem64]
-def FpSUB32m : FpI<(ops RFP:$dst, RFP:$src1, f32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fsub RFP:$src1,
- (extloadf64f32 addr:$src2)))]>;
- // ST(0) = ST(0) - [mem32]
-def FpSUB64m : FpI<(ops RFP:$dst, RFP:$src1, f64mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fsub RFP:$src1, (loadf64 addr:$src2)))]>;
- // ST(0) = ST(0) - [mem64]
-def FpSUBR32m : FpI<(ops RFP:$dst, RFP:$src1, f32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fsub (extloadf64f32 addr:$src2),
- RFP:$src1))]>;
- // ST(0) = [mem32] - ST(0)
-def FpSUBR64m : FpI<(ops RFP:$dst, RFP:$src1, f64mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fsub (loadf64 addr:$src2), RFP:$src1))]>;
- // ST(0) = [mem64] - ST(0)
-def FpDIV32m : FpI<(ops RFP:$dst, RFP:$src1, f32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fdiv RFP:$src1,
- (extloadf64f32 addr:$src2)))]>;
- // ST(0) = ST(0) / [mem32]
-def FpDIV64m : FpI<(ops RFP:$dst, RFP:$src1, f64mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fdiv RFP:$src1, (loadf64 addr:$src2)))]>;
- // ST(0) = ST(0) / [mem64]
-def FpDIVR32m : FpI<(ops RFP:$dst, RFP:$src1, f32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fdiv (extloadf64f32 addr:$src2),
- RFP:$src1))]>;
- // ST(0) = [mem32] / ST(0)
-def FpDIVR64m : FpI<(ops RFP:$dst, RFP:$src1, f64mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fdiv (loadf64 addr:$src2), RFP:$src1))]>;
- // ST(0) = [mem64] / ST(0)
-
-
-def FADD32m : FPI<0xD8, MRM0m, (ops f32mem:$src), "fadd{s} $src">;
-def FADD64m : FPI<0xDC, MRM0m, (ops f64mem:$src), "fadd{l} $src">;
-def FMUL32m : FPI<0xD8, MRM1m, (ops f32mem:$src), "fmul{s} $src">;
-def FMUL64m : FPI<0xDC, MRM1m, (ops f64mem:$src), "fmul{l} $src">;
-def FSUB32m : FPI<0xD8, MRM4m, (ops f32mem:$src), "fsub{s} $src">;
-def FSUB64m : FPI<0xDC, MRM4m, (ops f64mem:$src), "fsub{l} $src">;
-def FSUBR32m : FPI<0xD8, MRM5m, (ops f32mem:$src), "fsubr{s} $src">;
-def FSUBR64m : FPI<0xDC, MRM5m, (ops f64mem:$src), "fsubr{l} $src">;
-def FDIV32m : FPI<0xD8, MRM6m, (ops f32mem:$src), "fdiv{s} $src">;
-def FDIV64m : FPI<0xDC, MRM6m, (ops f64mem:$src), "fdiv{l} $src">;
-def FDIVR32m : FPI<0xD8, MRM7m, (ops f32mem:$src), "fdivr{s} $src">;
-def FDIVR64m : FPI<0xDC, MRM7m, (ops f64mem:$src), "fdivr{l} $src">;
-
-def FpIADD16m : FpI<(ops RFP:$dst, RFP:$src1, i16mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fadd RFP:$src1,
- (X86fild addr:$src2, i16)))]>;
- // ST(0) = ST(0) + [mem16int]
-def FpIADD32m : FpI<(ops RFP:$dst, RFP:$src1, i32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fadd RFP:$src1,
- (X86fild addr:$src2, i32)))]>;
- // ST(0) = ST(0) + [mem32int]
-def FpIMUL16m : FpI<(ops RFP:$dst, RFP:$src1, i16mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fmul RFP:$src1,
- (X86fild addr:$src2, i16)))]>;
- // ST(0) = ST(0) * [mem16int]
-def FpIMUL32m : FpI<(ops RFP:$dst, RFP:$src1, i32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fmul RFP:$src1,
- (X86fild addr:$src2, i32)))]>;
- // ST(0) = ST(0) * [mem32int]
-def FpISUB16m : FpI<(ops RFP:$dst, RFP:$src1, i16mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fsub RFP:$src1,
- (X86fild addr:$src2, i16)))]>;
- // ST(0) = ST(0) - [mem16int]
-def FpISUB32m : FpI<(ops RFP:$dst, RFP:$src1, i32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fsub RFP:$src1,
- (X86fild addr:$src2, i32)))]>;
- // ST(0) = ST(0) - [mem32int]
-def FpISUBR16m : FpI<(ops RFP:$dst, RFP:$src1, i16mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fsub (X86fild addr:$src2, i16),
- RFP:$src1))]>;
- // ST(0) = [mem16int] - ST(0)
-def FpISUBR32m : FpI<(ops RFP:$dst, RFP:$src1, i32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fsub (X86fild addr:$src2, i32),
- RFP:$src1))]>;
- // ST(0) = [mem32int] - ST(0)
-def FpIDIV16m : FpI<(ops RFP:$dst, RFP:$src1, i16mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fdiv RFP:$src1,
- (X86fild addr:$src2, i16)))]>;
- // ST(0) = ST(0) / [mem16int]
-def FpIDIV32m : FpI<(ops RFP:$dst, RFP:$src1, i32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fdiv RFP:$src1,
- (X86fild addr:$src2, i32)))]>;
- // ST(0) = ST(0) / [mem32int]
-def FpIDIVR16m : FpI<(ops RFP:$dst, RFP:$src1, i16mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fdiv (X86fild addr:$src2, i16),
- RFP:$src1))]>;
- // ST(0) = [mem16int] / ST(0)
-def FpIDIVR32m : FpI<(ops RFP:$dst, RFP:$src1, i32mem:$src2), OneArgFPRW,
- [(set RFP:$dst, (fdiv (X86fild addr:$src2, i32),
- RFP:$src1))]>;
- // ST(0) = [mem32int] / ST(0)
-
-def FIADD16m : FPI<0xDE, MRM0m, (ops i16mem:$src), "fiadd{s} $src">;
-def FIADD32m : FPI<0xDA, MRM0m, (ops i32mem:$src), "fiadd{l} $src">;
-def FIMUL16m : FPI<0xDE, MRM1m, (ops i16mem:$src), "fimul{s} $src">;
-def FIMUL32m : FPI<0xDA, MRM1m, (ops i32mem:$src), "fimul{l} $src">;
-def FISUB16m : FPI<0xDE, MRM4m, (ops i16mem:$src), "fisub{s} $src">;
-def FISUB32m : FPI<0xDA, MRM4m, (ops i32mem:$src), "fisub{l} $src">;
-def FISUBR16m : FPI<0xDE, MRM5m, (ops i16mem:$src), "fisubr{s} $src">;
-def FISUBR32m : FPI<0xDA, MRM5m, (ops i32mem:$src), "fisubr{l} $src">;
-def FIDIV16m : FPI<0xDE, MRM6m, (ops i16mem:$src), "fidiv{s} $src">;
-def FIDIV32m : FPI<0xDA, MRM6m, (ops i32mem:$src), "fidiv{l} $src">;
-def FIDIVR16m : FPI<0xDE, MRM7m, (ops i16mem:$src), "fidivr{s} $src">;
-def FIDIVR32m : FPI<0xDA, MRM7m, (ops i32mem:$src), "fidivr{l} $src">;
+ : FPI<o, AddRegFrm, (outs), (ins RST:$op), asm>, DE;
// 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 FADDST0r : FPST0rInst <0xC0, "fadd $op">;
-def FADDrST0 : FPrST0Inst <0xC0, "fadd {%st(0), $op|$op, %ST(0)}">;
-def FADDPrST0 : FPrST0PInst<0xC0, "faddp $op">;
-def FSUBRST0r : FPST0rInst <0xE8, "fsubr $op">;
-def FSUBrST0 : FPrST0Inst <0xE8, "fsub{r} {%st(0), $op|$op, %ST(0)}">;
-def FSUBPrST0 : FPrST0PInst<0xE8, "fsub{r}p $op">;
-def FSUBST0r : FPST0rInst <0xE0, "fsub $op">;
-def FSUBRrST0 : FPrST0Inst <0xE0, "fsub{|r} {%st(0), $op|$op, %ST(0)}">;
-def FSUBRPrST0 : FPrST0PInst<0xE0, "fsub{|r}p $op">;
-def FMULST0r : FPST0rInst <0xC8, "fmul $op">;
-def FMULrST0 : FPrST0Inst <0xC8, "fmul {%st(0), $op|$op, %ST(0)}">;
-def FMULPrST0 : FPrST0PInst<0xC8, "fmulp $op">;
-def FDIVRST0r : FPST0rInst <0xF8, "fdivr $op">;
-def FDIVrST0 : FPrST0Inst <0xF8, "fdiv{r} {%st(0), $op|$op, %ST(0)}">;
-def FDIVPrST0 : FPrST0PInst<0xF8, "fdiv{r}p $op">;
-def FDIVST0r : FPST0rInst <0xF0, "fdiv $op">;
-def FDIVRrST0 : FPrST0Inst <0xF0, "fdiv{|r} {%st(0), $op|$op, %ST(0)}">;
-def FDIVRPrST0 : FPrST0PInst<0xF0, "fdiv{|r}p $op">;
-
+def ADD_FST0r : FPST0rInst <0xC0, "fadd\t$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$op">;
// Unary operations.
-def FpCHS : FpI<(ops RFP:$dst, RFP:$src), OneArgFPRW,
- [(set RFP:$dst, (fneg RFP:$src))]>;
-def FpABS : FpI<(ops RFP:$dst, RFP:$src), OneArgFPRW,
- [(set RFP:$dst, (fabs RFP:$src))]>;
-def FpSQRT : FpI<(ops RFP:$dst, RFP:$src), OneArgFPRW,
- [(set RFP:$dst, (fsqrt RFP:$src))]>;
-def FpSIN : FpI<(ops RFP:$dst, RFP:$src), OneArgFPRW,
- [(set RFP:$dst, (fsin RFP:$src))]>;
-def FpCOS : FpI<(ops RFP:$dst, RFP:$src), OneArgFPRW,
- [(set RFP:$dst, (fcos RFP:$src))]>;
-def FpTST : FpI<(ops RFP:$src), OneArgFP,
- []>;
-
-def FCHS : FPI<0xE0, RawFrm, (ops), "fchs">, D9;
-def FABS : FPI<0xE1, RawFrm, (ops), "fabs">, D9;
-def FSQRT : FPI<0xFA, RawFrm, (ops), "fsqrt">, D9;
-def FSIN : FPI<0xFE, RawFrm, (ops), "fsin">, D9;
-def FCOS : FPI<0xFF, RawFrm, (ops), "fcos">, D9;
-def FTST : FPI<0xE4, RawFrm, (ops), "ftst">, D9;
+multiclass FPUnary<SDNode OpNode, bits<8> opcode, string asmstring> {
+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<opcode, RawFrm, (outs), (ins), asmstring>, D9;
+}
+defm CHS : FPUnary<fneg, 0xE0, "fchs">;
+defm ABS : FPUnary<fabs, 0xE1, "fabs">;
+defm SQRT: FPUnary<fsqrt,0xFA, "fsqrt">;
+defm SIN : FPUnary<fsin, 0xFE, "fsin">;
+defm COS : FPUnary<fcos, 0xFF, "fcos">;
+
+let neverHasSideEffects = 1 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:$src), OneArgFP, []>;
+}
+def TST_F : FPI<0xE4, RawFrm, (outs), (ins), "ftst">, D9;
// Floating point cmovs.
-let isTwoAddress = 1 in {
- def FpCMOVB : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), CondMovFP,
- [(set RFP:$dst, (X86cmov RFP:$src1, RFP:$src2,
- X86_COND_B))]>;
- def FpCMOVBE : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), CondMovFP,
- [(set RFP:$dst, (X86cmov RFP:$src1, RFP:$src2,
- X86_COND_BE))]>;
- def FpCMOVE : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), CondMovFP,
- [(set RFP:$dst, (X86cmov RFP:$src1, RFP:$src2,
- X86_COND_E))]>;
- def FpCMOVP : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), CondMovFP,
- [(set RFP:$dst, (X86cmov RFP:$src1, RFP:$src2,
- X86_COND_P))]>;
- def FpCMOVNB : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), CondMovFP,
- [(set RFP:$dst, (X86cmov RFP:$src1, RFP:$src2,
- X86_COND_AE))]>;
- def FpCMOVNBE: FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), CondMovFP,
- [(set RFP:$dst, (X86cmov RFP:$src1, RFP:$src2,
- X86_COND_A))]>;
- def FpCMOVNE : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), CondMovFP,
- [(set RFP:$dst, (X86cmov RFP:$src1, RFP:$src2,
- X86_COND_NE))]>;
- def FpCMOVNP : FpI<(ops RFP:$dst, RFP:$src1, RFP:$src2), CondMovFP,
- [(set RFP:$dst, (X86cmov RFP:$src1, RFP:$src2,
- X86_COND_NP))]>;
+multiclass FPCMov<PatLeaf cc> {
+ def _Fp32 : FpIf32<(outs RFP32:$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),
+ 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:$src2,
+ cc, EFLAGS))]>;
+}
+let Uses = [EFLAGS], isTwoAddress = 1 in {
+defm CMOVB : FPCMov<X86_COND_B>;
+defm CMOVBE : FPCMov<X86_COND_BE>;
+defm CMOVE : FPCMov<X86_COND_E>;
+defm CMOVP : FPCMov<X86_COND_P>;
+defm CMOVNB : FPCMov<X86_COND_AE>;
+defm CMOVNBE: FPCMov<X86_COND_A>;
+defm CMOVNE : FPCMov<X86_COND_NE>;
+defm CMOVNP : FPCMov<X86_COND_NP>;
}
-def FCMOVB : FPI<0xC0, AddRegFrm, (ops RST:$op),
- "fcmovb {$op, %st(0)|%ST(0), $op}">, DA;
-def FCMOVBE : FPI<0xD0, AddRegFrm, (ops RST:$op),
- "fcmovbe {$op, %st(0)|%ST(0), $op}">, DA;
-def FCMOVE : FPI<0xC8, AddRegFrm, (ops RST:$op),
- "fcmove {$op, %st(0)|%ST(0), $op}">, DA;
-def FCMOVP : FPI<0xD8, AddRegFrm, (ops RST:$op),
- "fcmovu {$op, %st(0)|%ST(0), $op}">, DA;
-def FCMOVNB : FPI<0xC0, AddRegFrm, (ops RST:$op),
- "fcmovnb {$op, %st(0)|%ST(0), $op}">, DB;
-def FCMOVNBE : FPI<0xD0, AddRegFrm, (ops RST:$op),
- "fcmovnbe {$op, %st(0)|%ST(0), $op}">, DB;
-def FCMOVNE : FPI<0xC8, AddRegFrm, (ops RST:$op),
- "fcmovne {$op, %st(0)|%ST(0), $op}">, DB;
-def FCMOVNP : FPI<0xD8, AddRegFrm, (ops RST:$op),
- "fcmovnu {$op, %st(0)|%ST(0), $op}">, DB;
+// These are not factored because there's no clean way to pass DA/DB.
+def CMOVB_F : FPI<0xC0, AddRegFrm, (outs RST:$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), $op}">, DB;
// Floating point loads & stores.
-def FpLD32m : FpI<(ops RFP:$dst, f32mem:$src), ZeroArgFP,
- [(set RFP:$dst, (extloadf64f32 addr:$src))]>;
-def FpLD64m : FpI<(ops RFP:$dst, f64mem:$src), ZeroArgFP,
- [(set RFP:$dst, (loadf64 addr:$src))]>;
-def FpILD16m : FpI<(ops RFP:$dst, i16mem:$src), ZeroArgFP,
- [(set RFP:$dst, (X86fild addr:$src, i16))]>;
-def FpILD32m : FpI<(ops RFP:$dst, i32mem:$src), ZeroArgFP,
- [(set RFP:$dst, (X86fild addr:$src, i32))]>;
-def FpILD64m : FpI<(ops RFP:$dst, i64mem:$src), ZeroArgFP,
- [(set RFP:$dst, (X86fild addr:$src, i64))]>;
-
-def FpST32m : FpI<(ops f32mem:$op, RFP:$src), OneArgFP,
- [(truncstore RFP:$src, addr:$op, f32)]>;
-def FpST64m : FpI<(ops f64mem:$op, RFP:$src), OneArgFP,
- [(store RFP:$src, addr:$op)]>;
-
-def FpSTP32m : FpI<(ops f32mem:$op, RFP:$src), OneArgFP, []>;
-def FpSTP64m : FpI<(ops f64mem:$op, RFP:$src), OneArgFP, []>;
-def FpIST16m : FpI<(ops i16mem:$op, RFP:$src), OneArgFP, []>;
-def FpIST32m : FpI<(ops i32mem:$op, RFP:$src), OneArgFP, []>;
-def FpIST64m : FpI<(ops i64mem:$op, RFP:$src), OneArgFP, []>;
-
-def FLD32m : FPI<0xD9, MRM0m, (ops f32mem:$src), "fld{s} $src">;
-def FLD64m : FPI<0xDD, MRM0m, (ops f64mem:$src), "fld{l} $src">;
-def FILD16m : FPI<0xDF, MRM0m, (ops i16mem:$src), "fild{s} $src">;
-def FILD32m : FPI<0xDB, MRM0m, (ops i32mem:$src), "fild{l} $src">;
-def FILD64m : FPI<0xDF, MRM5m, (ops i64mem:$src), "fild{ll} $src">;
-def FST32m : FPI<0xD9, MRM2m, (ops f32mem:$dst), "fst{s} $dst">;
-def FST64m : FPI<0xDD, MRM2m, (ops f64mem:$dst), "fst{l} $dst">;
-def FSTP32m : FPI<0xD9, MRM3m, (ops f32mem:$dst), "fstp{s} $dst">;
-def FSTP64m : FPI<0xDD, MRM3m, (ops f64mem:$dst), "fstp{l} $dst">;
-def FIST16m : FPI<0xDF, MRM2m, (ops i16mem:$dst), "fist{s} $dst">;
-def FIST32m : FPI<0xDB, MRM2m, (ops i32mem:$dst), "fist{l} $dst">;
-def FISTP16m : FPI<0xDF, MRM3m, (ops i16mem:$dst), "fistp{s} $dst">;
-def FISTP32m : FPI<0xDB, MRM3m, (ops i32mem:$dst), "fistp{l} $dst">;
-def FISTP64m : FPI<0xDF, MRM7m, (ops i64mem:$dst), "fistp{ll} $dst">;
+let canFoldAsLoad = 1 in {
+def LD_Fp32m : FpIf32<(outs RFP32:$dst), (ins f32mem:$src), ZeroArgFP,
+ [(set RFP32:$dst, (loadf32 addr:$src))]>;
+let isReMaterializable = 1, mayHaveSideEffects = 1 in
+ def LD_Fp64m : FpIf64<(outs RFP64:$dst), (ins f64mem:$src), ZeroArgFP,
+ [(set RFP64:$dst, (loadf64 addr:$src))]>;
+def LD_Fp80m : FpI_<(outs RFP80:$dst), (ins f80mem:$src), ZeroArgFP,
+ [(set RFP80:$dst, (loadf80 addr:$src))]>;
+}
+def LD_Fp32m64 : FpIf64<(outs RFP64:$dst), (ins f32mem:$src), ZeroArgFP,
+ [(set RFP64:$dst, (f64 (extloadf32 addr:$src)))]>;
+def LD_Fp64m80 : FpI_<(outs RFP80:$dst), (ins f64mem:$src), ZeroArgFP,
+ [(set RFP80:$dst, (f80 (extloadf64 addr:$src)))]>;
+def LD_Fp32m80 : FpI_<(outs RFP80:$dst), (ins f32mem:$src), ZeroArgFP,
+ [(set RFP80:$dst, (f80 (extloadf32 addr:$src)))]>;
+def ILD_Fp16m32: FpIf32<(outs RFP32:$dst), (ins i16mem:$src), ZeroArgFP,
+ [(set RFP32:$dst, (X86fild addr:$src, i16))]>;
+def ILD_Fp32m32: FpIf32<(outs RFP32:$dst), (ins i32mem:$src), ZeroArgFP,
+ [(set RFP32:$dst, (X86fild addr:$src, i32))]>;
+def ILD_Fp64m32: FpIf32<(outs RFP32:$dst), (ins i64mem:$src), ZeroArgFP,
+ [(set RFP32:$dst, (X86fild addr:$src, i64))]>;
+def ILD_Fp16m64: FpIf64<(outs RFP64:$dst), (ins i16mem:$src), ZeroArgFP,
+ [(set RFP64:$dst, (X86fild addr:$src, i16))]>;
+def ILD_Fp32m64: FpIf64<(outs RFP64:$dst), (ins i32mem:$src), ZeroArgFP,
+ [(set RFP64:$dst, (X86fild addr:$src, i32))]>;
+def ILD_Fp64m64: FpIf64<(outs RFP64:$dst), (ins i64mem:$src), ZeroArgFP,
+ [(set RFP64:$dst, (X86fild addr:$src, i64))]>;
+def ILD_Fp16m80: FpI_<(outs RFP80:$dst), (ins i16mem:$src), ZeroArgFP,
+ [(set RFP80:$dst, (X86fild addr:$src, i16))]>;
+def ILD_Fp32m80: FpI_<(outs RFP80:$dst), (ins i32mem:$src), ZeroArgFP,
+ [(set RFP80:$dst, (X86fild addr:$src, i32))]>;
+def ILD_Fp64m80: FpI_<(outs RFP80:$dst), (ins i64mem:$src), ZeroArgFP,
+ [(set RFP80:$dst, (X86fild addr:$src, i64))]>;
+
+def ST_Fp32m : FpIf32<(outs), (ins f32mem:$op, RFP32:$src), OneArgFP,
+ [(store RFP32:$src, addr:$op)]>;
+def ST_Fp64m32 : FpIf64<(outs), (ins f32mem:$op, RFP64:$src), OneArgFP,
+ [(truncstoref32 RFP64:$src, addr:$op)]>;
+def ST_Fp64m : FpIf64<(outs), (ins f64mem:$op, RFP64:$src), OneArgFP,
+ [(store RFP64:$src, addr:$op)]>;
+def ST_Fp80m32 : FpI_<(outs), (ins f32mem:$op, RFP80:$src), OneArgFP,
+ [(truncstoref32 RFP80:$src, addr:$op)]>;
+def ST_Fp80m64 : FpI_<(outs), (ins f64mem:$op, RFP80:$src), OneArgFP,
+ [(truncstoref64 RFP80:$src, addr:$op)]>;
+// FST does not support 80-bit memory target; FSTP must be used.
+
+let mayStore = 1, neverHasSideEffects = 1 in {
+def ST_FpP32m : FpIf32<(outs), (ins f32mem:$op, RFP32:$src), OneArgFP, []>;
+def ST_FpP64m32 : FpIf64<(outs), (ins f32mem:$op, RFP64:$src), OneArgFP, []>;
+def ST_FpP64m : FpIf64<(outs), (ins f64mem:$op, RFP64:$src), OneArgFP, []>;
+def ST_FpP80m32 : FpI_<(outs), (ins f32mem:$op, RFP80:$src), OneArgFP, []>;
+def ST_FpP80m64 : FpI_<(outs), (ins f64mem:$op, RFP80:$src), OneArgFP, []>;
+}
+def ST_FpP80m : FpI_<(outs), (ins f80mem:$op, RFP80:$src), OneArgFP,
+ [(store RFP80:$src, addr:$op)]>;
+let mayStore = 1, neverHasSideEffects = 1 in {
+def IST_Fp16m32 : FpIf32<(outs), (ins i16mem:$op, RFP32:$src), OneArgFP, []>;
+def IST_Fp32m32 : FpIf32<(outs), (ins i32mem:$op, RFP32:$src), OneArgFP, []>;
+def IST_Fp64m32 : FpIf32<(outs), (ins i64mem:$op, RFP32:$src), OneArgFP, []>;
+def IST_Fp16m64 : FpIf64<(outs), (ins i16mem:$op, RFP64:$src), OneArgFP, []>;
+def IST_Fp32m64 : FpIf64<(outs), (ins i32mem:$op, RFP64:$src), OneArgFP, []>;
+def IST_Fp64m64 : FpIf64<(outs), (ins i64mem:$op, RFP64:$src), OneArgFP, []>;
+def IST_Fp16m80 : FpI_<(outs), (ins i16mem:$op, RFP80:$src), OneArgFP, []>;
+def IST_Fp32m80 : FpI_<(outs), (ins i32mem:$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$src">;
+}
+let mayStore = 1 in {
+def ST_F32m : FPI<0xD9, MRM2m, (outs), (ins f32mem:$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$dst">;
+}
// FISTTP requires SSE3 even though it's a FPStack op.
-def FpISTT16m : FpI_<(ops i16mem:$op, RFP:$src), OneArgFP,
- [(X86fp_to_i16mem RFP:$src, addr:$op)]>,
- Requires<[HasSSE3]>;
-def FpISTT32m : FpI_<(ops i32mem:$op, RFP:$src), OneArgFP,
- [(X86fp_to_i32mem RFP:$src, addr:$op)]>,
- Requires<[HasSSE3]>;
-def FpISTT64m : FpI_<(ops i64mem:$op, RFP:$src), OneArgFP,
- [(X86fp_to_i64mem RFP:$src, addr:$op)]>,
- Requires<[HasSSE3]>;
-
-def FISTTP16m : FPI<0xDF, MRM1m, (ops i16mem:$dst), "fisttp{s} $dst">;
-def FISTTP32m : FPI<0xDB, MRM1m, (ops i32mem:$dst), "fisttp{l} $dst">;
-def FISTTP64m : FPI<0xDD, MRM1m, (ops i64mem:$dst), "fisttp{ll} $dst">;
+def ISTT_Fp16m32 : FpI_<(outs), (ins i16mem:$op, RFP32:$src), OneArgFP,
+ [(X86fp_to_i16mem RFP32:$src, addr:$op)]>,
+ Requires<[HasSSE3]>;
+def ISTT_Fp32m32 : FpI_<(outs), (ins i32mem:$op, RFP32:$src), OneArgFP,
+ [(X86fp_to_i32mem RFP32:$src, addr:$op)]>,
+ Requires<[HasSSE3]>;
+def ISTT_Fp64m32 : FpI_<(outs), (ins i64mem:$op, RFP32:$src), OneArgFP,
+ [(X86fp_to_i64mem RFP32:$src, addr:$op)]>,
+ Requires<[HasSSE3]>;
+def ISTT_Fp16m64 : FpI_<(outs), (ins i16mem:$op, RFP64:$src), OneArgFP,
+ [(X86fp_to_i16mem RFP64:$src, addr:$op)]>,
+ Requires<[HasSSE3]>;
+def ISTT_Fp32m64 : FpI_<(outs), (ins i32mem:$op, RFP64:$src), OneArgFP,
+ [(X86fp_to_i32mem RFP64:$src, addr:$op)]>,
+ Requires<[HasSSE3]>;
+def ISTT_Fp64m64 : FpI_<(outs), (ins i64mem:$op, RFP64:$src), OneArgFP,
+ [(X86fp_to_i64mem RFP64:$src, addr:$op)]>,
+ Requires<[HasSSE3]>;
+def ISTT_Fp16m80 : FpI_<(outs), (ins i16mem:$op, RFP80:$src), OneArgFP,
+ [(X86fp_to_i16mem RFP80:$src, addr:$op)]>,
+ Requires<[HasSSE3]>;
+def ISTT_Fp32m80 : FpI_<(outs), (ins i32mem:$op, RFP80:$src), OneArgFP,
+ [(X86fp_to_i32mem RFP80:$src, addr:$op)]>,
+ Requires<[HasSSE3]>;
+def ISTT_Fp64m80 : FpI_<(outs), (ins i64mem:$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$dst">;
+}
// FP Stack manipulation instructions.
-def FLDrr : FPI<0xC0, AddRegFrm, (ops RST:$op), "fld $op">, D9;
-def FSTrr : FPI<0xD0, AddRegFrm, (ops RST:$op), "fst $op">, DD;
-def FSTPrr : FPI<0xD8, AddRegFrm, (ops RST:$op), "fstp $op">, DD;
-def FXCH : FPI<0xC8, AddRegFrm, (ops RST:$op), "fxch $op">, D9;
+def LD_Frr : FPI<0xC0, AddRegFrm, (outs), (ins RST:$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$op">, D9;
// Floating point constant loads.
-def FpLD0 : FpI<(ops RFP:$dst), ZeroArgFP,
- [(set RFP:$dst, fp64imm0)]>;
-def FpLD1 : FpI<(ops RFP:$dst), ZeroArgFP,
- [(set RFP:$dst, fp64imm1)]>;
+let isReMaterializable = 1 in {
+def LD_Fp032 : FpIf32<(outs RFP32:$dst), (ins), ZeroArgFP,
+ [(set RFP32:$dst, fpimm0)]>;
+def LD_Fp132 : FpIf32<(outs RFP32:$dst), (ins), ZeroArgFP,
+ [(set RFP32:$dst, fpimm1)]>;
+def LD_Fp064 : FpIf64<(outs RFP64:$dst), (ins), ZeroArgFP,
+ [(set RFP64:$dst, fpimm0)]>;
+def LD_Fp164 : FpIf64<(outs RFP64:$dst), (ins), ZeroArgFP,
+ [(set RFP64:$dst, fpimm1)]>;
+def LD_Fp080 : FpI_<(outs RFP80:$dst), (ins), ZeroArgFP,
+ [(set RFP80:$dst, fpimm0)]>;
+def LD_Fp180 : FpI_<(outs RFP80:$dst), (ins), ZeroArgFP,
+ [(set RFP80:$dst, fpimm1)]>;
+}
-def FLD0 : FPI<0xEE, RawFrm, (ops), "fldz">, D9;
-def FLD1 : FPI<0xE8, RawFrm, (ops), "fld1">, D9;
+def LD_F0 : FPI<0xEE, RawFrm, (outs), (ins), "fldz">, D9;
+def LD_F1 : FPI<0xE8, RawFrm, (outs), (ins), "fld1">, D9;
// Floating point compares.
-def FpUCOMr : FpI<(ops RFP:$lhs, RFP:$rhs), CompareFP,
- []>; // FPSW = cmp ST(0) with ST(i)
-def FpUCOMIr : FpI<(ops RFP:$lhs, RFP:$rhs), CompareFP,
- [(X86cmp RFP:$lhs, RFP:$rhs)]>; // CC = cmp ST(0) with ST(i)
-
-def FUCOMr : FPI<0xE0, AddRegFrm, // FPSW = cmp ST(0) with ST(i)
- (ops RST:$reg),
- "fucom $reg">, DD, Imp<[ST0],[]>;
-def FUCOMPr : FPI<0xE8, AddRegFrm, // FPSW = cmp ST(0) with ST(i), pop
- (ops RST:$reg),
- "fucomp $reg">, DD, Imp<[ST0],[]>;
-def FUCOMPPr : FPI<0xE9, RawFrm, // cmp ST(0) with ST(1), pop, pop
- (ops),
- "fucompp">, DA, Imp<[ST0],[]>;
-
-def FUCOMIr : FPI<0xE8, AddRegFrm, // CC = cmp ST(0) with ST(i)
- (ops RST:$reg),
- "fucomi {$reg, %st(0)|%ST(0), $reg}">, DB, Imp<[ST0],[]>;
-def FUCOMIPr : FPI<0xE8, AddRegFrm, // CC = cmp ST(0) with ST(i), pop
- (ops RST:$reg),
- "fucomip {$reg, %st(0)|%ST(0), $reg}">, DF, Imp<[ST0],[]>;
+let Defs = [EFLAGS] in {
+def UCOM_Fpr32 : FpIf32<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP,
+ []>; // FPSW = cmp ST(0) with ST(i)
+def UCOM_Fpr64 : FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP,
+ []>; // FPSW = cmp ST(0) with ST(i)
+def UCOM_Fpr80 : FpI_ <(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP,
+ []>; // FPSW = cmp ST(0) with ST(i)
+
+def UCOM_FpIr32: FpIf32<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP,
+ [(X86cmp RFP32:$lhs, RFP32:$rhs),
+ (implicit EFLAGS)]>; // CC = ST(0) cmp ST(i)
+def UCOM_FpIr64: FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP,
+ [(X86cmp RFP64:$lhs, RFP64:$rhs),
+ (implicit EFLAGS)]>; // CC = ST(0) cmp ST(i)
+def UCOM_FpIr80: FpI_<(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP,
+ [(X86cmp RFP80:$lhs, RFP80:$rhs),
+ (implicit EFLAGS)]>; // CC = ST(0) cmp ST(i)
+}
+let Defs = [EFLAGS], Uses = [ST0] in {
+def UCOM_Fr : FPI<0xE0, AddRegFrm, // FPSW = cmp ST(0) with ST(i)
+ (outs), (ins RST:$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), $reg}">, DF;
+}
// Floating point flag ops.
+let Defs = [AX] in
def FNSTSW8r : I<0xE0, RawFrm, // AX = fp flags
- (ops), "fnstsw", []>, DF, Imp<[],[AX]>;
+ (outs), (ins), "fnstsw", []>, DF;
def FNSTCW16m : I<0xD9, MRM7m, // [mem16] = X87 control world
- (ops i16mem:$dst), "fnstcw $dst", []>;
+ (outs), (ins i16mem:$dst), "fnstcw\t$dst",
+ [(X86fp_cwd_get16 addr:$dst)]>;
+
+let mayLoad = 1 in
def FLDCW16m : I<0xD9, MRM5m, // X87 control world = [mem16]
- (ops i16mem:$dst), "fldcw $dst", []>;
+ (outs), (ins i16mem:$dst), "fldcw\t$dst", []>;
//===----------------------------------------------------------------------===//
// Non-Instruction Patterns
//===----------------------------------------------------------------------===//
-// Required for RET of f32 / f64 values.
-def : Pat<(X86fld addr:$src, f32), (FpLD32m addr:$src)>;
-def : Pat<(X86fld addr:$src, f64), (FpLD64m addr:$src)>;
+// Required for RET of f32 / f64 / f80 values.
+def : Pat<(X86fld addr:$src, f32), (LD_Fp32m addr:$src)>;
+def : Pat<(X86fld addr:$src, f64), (LD_Fp64m addr:$src)>;
+def : Pat<(X86fld addr:$src, f80), (LD_Fp80m addr:$src)>;
-// Required for CALL which return f32 / f64 values.
-def : Pat<(X86fst RFP:$src, addr:$op, f32), (FpST32m addr:$op, RFP:$src)>;
-def : Pat<(X86fst RFP:$src, addr:$op, f64), (FpST64m addr:$op, RFP:$src)>;
+// Required for CALL which return f32 / f64 / f80 values.
+def : Pat<(X86fst RFP32:$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:$op, f64), (ST_Fp64m addr:$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:$src)>;
// Floating point constant -0.0 and -1.0
-def : Pat<(f64 fp64immneg0), (FpCHS (FpLD0))>, Requires<[FPStack]>;
-def : Pat<(f64 fp64immneg1), (FpCHS (FpLD1))>, Requires<[FPStack]>;
+def : Pat<(f32 fpimmneg0), (CHS_Fp32 (LD_Fp032))>, Requires<[FPStackf32]>;
+def : Pat<(f32 fpimmneg1), (CHS_Fp32 (LD_Fp132))>, Requires<[FPStackf32]>;
+def : Pat<(f64 fpimmneg0), (CHS_Fp64 (LD_Fp064))>, Requires<[FPStackf64]>;
+def : Pat<(f64 fpimmneg1), (CHS_Fp64 (LD_Fp164))>, Requires<[FPStackf64]>;
+def : Pat<(f80 fpimmneg0), (CHS_Fp80 (LD_Fp080))>;
+def : Pat<(f80 fpimmneg1), (CHS_Fp80 (LD_Fp180))>;
// Used to conv. i64 to f64 since there isn't a SSE version.
-def : Pat<(X86fildflag addr:$src, i64), (FpILD64m addr:$src)>;
+def : Pat<(X86fildflag addr:$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:$src)>,
+ Requires<[FPStackf32]>;
+def : Pat<(f80 (fextend RFP32:$src)), (MOV_Fp3280 RFP32:$src)>,
+ Requires<[FPStackf32]>;
+def : Pat<(f80 (fextend RFP64:$src)), (MOV_Fp6480 RFP64:$src)>,
+ Requires<[FPStackf64]>;
+
+// FP truncations map onto simple pseudo-value conversions if they are to/from
+// the FP stack. We have validated that only value-preserving truncations make
+// it through isel.
+def : Pat<(f32 (fround RFP64:$src)), (MOV_Fp6432 RFP64:$src)>,
+ Requires<[FPStackf32]>;
+def : Pat<(f32 (fround RFP80:$src)), (MOV_Fp8032 RFP80:$src)>,
+ Requires<[FPStackf32]>;
+def : Pat<(f64 (fround RFP80:$src)), (MOV_Fp8064 RFP80:$src)>,
+ Requires<[FPStackf64]>;