// FPStack specific DAG Nodes.
//===----------------------------------------------------------------------===//
-def SDTX86FpGet : SDTypeProfile<1, 0, [SDTCisFP<0>]>;
-def SDTX86FpSet : SDTypeProfile<0, 1, [SDTCisFP<0>]>;
+def SDTX86FpGet2 : SDTypeProfile<2, 0, [SDTCisVT<0, f80>,
+ SDTCisVT<1, f80>]>;
def SDTX86Fld : SDTypeProfile<1, 2, [SDTCisFP<0>,
SDTCisPtrTy<1>,
SDTCisVT<2, OtherVT>]>;
def SDTX86CwdStore : SDTypeProfile<0, 1, [SDTCisPtrTy<0>]>;
-def X86fpget : SDNode<"X86ISD::FP_GET_RESULT", SDTX86FpGet,
- [SDNPHasChain, SDNPInFlag, SDNPOutFlag]>;
-def X86fpset : SDNode<"X86ISD::FP_SET_RESULT", SDTX86FpSet,
- [SDNPHasChain, SDNPOutFlag]>;
def X86fld : SDNode<"X86ISD::FLD", SDTX86Fld,
[SDNPHasChain, SDNPMayLoad]>;
def X86fst : SDNode<"X86ISD::FST", SDTX86Fst,
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!",
+ "##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!",
+ "##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!",
+ "##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!",
+ "##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!",
+ "##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!",
+ "##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!",
+ "##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!",
+ "##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!",
+ "##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, (outs), (ins), "#FP_REG_KILL", []>;
+ 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"
// encoding and asm printing info).
// Pseudo Instructions for FP stack return values.
-def FpGETRESULT32 : FpI_<(outs RFP32:$dst), (ins), SpecialFP,
- [(set RFP32:$dst, X86fpget)]>; // FPR = ST(0)
-
-def FpGETRESULT64 : FpI_<(outs RFP64:$dst), (ins), SpecialFP,
- [(set RFP64:$dst, X86fpget)]>; // FPR = ST(0)
-
-def FpGETRESULT80 : FpI_<(outs RFP80:$dst), (ins), SpecialFP,
- [(set RFP80:$dst, X86fpget)]>; // FPR = ST(0)
+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 FpSETRESULT32 : FpI_<(outs), (ins RFP32:$src), SpecialFP,
- [(X86fpset RFP32:$src)]>;// ST(0) = FPR
-
-def FpSETRESULT64 : FpI_<(outs), (ins RFP64:$src), SpecialFP,
- [(X86fpset RFP64:$src)]>;// ST(0) = FPR
+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:$src), SpecialFP, []>; // ST(0) = FPR
+}
-def FpSETRESULT80 : FpI_<(outs), (ins RFP80:$src), SpecialFP,
- [(X86fpset RFP80:$src)]>;// ST(0) = FPR
+let Defs = [ST1] in {
+def FpSET_ST1_32 : FpI_<(outs), (ins RFP32:$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.
FpI_<outs, ins, fp, pattern>, Requires<[FPStackf64]>;
// Register copies. Just copies, the shortening ones do not truncate.
-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, []>;
+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<SDNode OpNode> {
defm SIN : FPUnary<fsin, 0xFE, "fsin">;
defm COS : FPUnary<fcos, 0xFF, "fcos">;
-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:$src), OneArgFP,
- []>;
+let neverHasSideEffects = 1 in {
+def TST_Fp32 : FpIf32<(outs), (ins RFP32:$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.
"fcmovnu\t{$op, %st(0)|%ST(0), $op}">, DB;
// Floating point loads & stores.
-let isSimpleLoad = 1 in {
+let canFoldAsLoad = 1 in {
def LD_Fp32m : FpIf32<(outs RFP32:$dst), (ins f32mem:$src), ZeroArgFP,
[(set RFP32:$dst, (loadf32 addr:$src))]>;
let isReMaterializable = 1, mayHaveSideEffects = 1 in
[(truncstoref64 RFP80:$src, addr:$op)]>;
// FST does not support 80-bit memory target; FSTP must be used.
-let mayStore = 1 in {
+let mayStore = 1, neverHasSideEffects = 1 in {
def ST_FpP32m : FpIf32<(outs), (ins f32mem:$op, RFP32:$src), OneArgFP, []>;
def ST_FpP64m32 : FpIf64<(outs), (ins f32mem:$op, RFP64:$src), OneArgFP, []>;
def ST_FpP64m : FpIf64<(outs), (ins f64mem:$op, RFP64:$src), OneArgFP, []>;
}
def ST_FpP80m : FpI_<(outs), (ins f80mem:$op, RFP80:$src), OneArgFP,
[(store RFP80:$src, addr:$op)]>;
-let mayStore = 1 in {
+let mayStore = 1, neverHasSideEffects = 1 in {
def IST_Fp16m32 : FpIf32<(outs), (ins i16mem:$op, RFP32:$src), OneArgFP, []>;
def IST_Fp32m32 : FpIf32<(outs), (ins i32mem:$op, RFP32:$src), OneArgFP, []>;
def IST_Fp64m32 : FpIf32<(outs), (ins i64mem:$op, RFP32:$src), OneArgFP, []>;
// Floating point compares.
let Defs = [EFLAGS] in {
def UCOM_Fpr32 : FpIf32<(outs), (ins RFP32:$lhs, RFP32:$rhs), CompareFP,
- []>; // FPSW = cmp ST(0) with ST(i)
+ []>; // FPSW = cmp ST(0) with ST(i)
+def UCOM_Fpr64 : FpIf64<(outs), (ins RFP64:$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:$rhs),
(implicit EFLAGS)]>; // CC = ST(0) cmp ST(i)
-def UCOM_Fpr64 : FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP,
- []>; // FPSW = cmp ST(0) with ST(i)
def UCOM_FpIr64: FpIf64<(outs), (ins RFP64:$lhs, RFP64:$rhs), CompareFP,
[(X86cmp RFP64:$lhs, RFP64:$rhs),
(implicit EFLAGS)]>; // CC = ST(0) cmp ST(i)
-def UCOM_Fpr80 : FpI_<(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP,
- []>; // FPSW = cmp ST(0) with ST(i)
def UCOM_FpIr80: FpI_<(outs), (ins RFP80:$lhs, RFP80:$rhs), CompareFP,
[(X86cmp RFP80:$lhs, RFP80:$rhs),
(implicit EFLAGS)]>; // CC = ST(0) cmp ST(i)
// Used to conv. i64 to f64 since there isn't a SSE version.
def : Pat<(X86fildflag addr:$src, i64), (ILD_Fp64m64 addr:$src)>;
-def : Pat<(f64 (fextend RFP32:$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:$src)>, Requires<[FPStackf64]>;
+// FP extensions map onto simple pseudo-value conversions if they are to/from
+// the FP stack.
+def : Pat<(f64 (fextend RFP32:$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]>;
projects / oota-llvm.git / blobdiff