//
//===----------------------------------------------------------------------===//
+// *********************************** NOTE ***********************************
+// ** For POWER8 Little Endian, the VSX swap optimization relies on knowing **
+// ** which VMX and VSX instructions are lane-sensitive and which are not. **
+// ** A lane-sensitive instruction relies, implicitly or explicitly, on **
+// ** whether lanes are numbered from left to right. An instruction like **
+// ** VADDFP is not lane-sensitive, because each lane of the result vector **
+// ** relies only on the corresponding lane of the source vectors. However, **
+// ** an instruction like VMULESB is lane-sensitive, because "even" and **
+// ** "odd" lanes are different for big-endian and little-endian numbering. **
+// ** **
+// ** When adding new VMX and VSX instructions, please consider whether they **
+// ** are lane-sensitive. If so, they must be added to a switch statement **
+// ** in PPCVSXSwapRemoval::gatherVectorInstructions(). **
+// ****************************************************************************
+
//===----------------------------------------------------------------------===//
// Altivec transformation functions and pattern fragments.
//
(vector_shuffle node:$lhs, node:$rhs), [{
return PPC::isVPKUWUMShuffleMask(cast<ShuffleVectorSDNode>(N), 0, *CurDAG);
}]>;
+def vpkudum_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVPKUDUMShuffleMask(cast<ShuffleVectorSDNode>(N), 0, *CurDAG);
+}]>;
def vpkuhum_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
return PPC::isVPKUHUMShuffleMask(cast<ShuffleVectorSDNode>(N), 1, *CurDAG);
(vector_shuffle node:$lhs, node:$rhs), [{
return PPC::isVPKUWUMShuffleMask(cast<ShuffleVectorSDNode>(N), 1, *CurDAG);
}]>;
+def vpkudum_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVPKUDUMShuffleMask(cast<ShuffleVectorSDNode>(N), 1, *CurDAG);
+}]>;
// These fragments are provided for little-endian, where the inputs must be
// swapped for correct semantics.
(vector_shuffle node:$lhs, node:$rhs), [{
return PPC::isVPKUWUMShuffleMask(cast<ShuffleVectorSDNode>(N), 2, *CurDAG);
}]>;
+def vpkudum_swapped_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVPKUDUMShuffleMask(cast<ShuffleVectorSDNode>(N), 2, *CurDAG);
+}]>;
def vmrglb_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle (v16i8 node:$lhs), node:$rhs), [{
}]>;
+def vmrgew_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVMRGEOShuffleMask(cast<ShuffleVectorSDNode>(N), true, 0, *CurDAG);
+}]>;
+def vmrgow_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVMRGEOShuffleMask(cast<ShuffleVectorSDNode>(N), false, 0, *CurDAG);
+}]>;
+def vmrgew_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVMRGEOShuffleMask(cast<ShuffleVectorSDNode>(N), true, 1, *CurDAG);
+}]>;
+def vmrgow_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVMRGEOShuffleMask(cast<ShuffleVectorSDNode>(N), false, 1, *CurDAG);
+}]>;
+def vmrgew_swapped_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVMRGEOShuffleMask(cast<ShuffleVectorSDNode>(N), true, 2, *CurDAG);
+}]>;
+def vmrgow_swapped_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
+ (vector_shuffle node:$lhs, node:$rhs), [{
+ return PPC::isVMRGEOShuffleMask(cast<ShuffleVectorSDNode>(N), false, 2, *CurDAG);
+}]>;
+
+
+
def VSLDOI_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::isVSLDOIShuffleMask(N, 0, *CurDAG));
+ return getI32Imm(PPC::isVSLDOIShuffleMask(N, 0, *CurDAG), SDLoc(N));
}]>;
def vsldoi_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
/// VSLDOI_unary* - These are used to match vsldoi(X,X), which is turned into
/// vector_shuffle(X,undef,mask) by the dag combiner.
def VSLDOI_unary_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::isVSLDOIShuffleMask(N, 1, *CurDAG));
+ return getI32Imm(PPC::isVSLDOIShuffleMask(N, 1, *CurDAG), SDLoc(N));
}]>;
def vsldoi_unary_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
/// VSLDOI_swapped* - These fragments are provided for little-endian, where
/// the inputs must be swapped for correct semantics.
def VSLDOI_swapped_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::isVSLDOIShuffleMask(N, 2, *CurDAG));
+ return getI32Imm(PPC::isVSLDOIShuffleMask(N, 2, *CurDAG), SDLoc(N));
}]>;
def vsldoi_swapped_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
// VSPLT*_get_imm xform function: convert vector_shuffle mask to VSPLT* imm.
def VSPLTB_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::getVSPLTImmediate(N, 1, *CurDAG));
+ return getI32Imm(PPC::getVSPLTImmediate(N, 1, *CurDAG), SDLoc(N));
}]>;
def vspltb_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
return PPC::isSplatShuffleMask(cast<ShuffleVectorSDNode>(N), 1);
}], VSPLTB_get_imm>;
def VSPLTH_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::getVSPLTImmediate(N, 2, *CurDAG));
+ return getI32Imm(PPC::getVSPLTImmediate(N, 2, *CurDAG), SDLoc(N));
}]>;
def vsplth_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
return PPC::isSplatShuffleMask(cast<ShuffleVectorSDNode>(N), 2);
}], VSPLTH_get_imm>;
def VSPLTW_get_imm : SDNodeXForm<vector_shuffle, [{
- return getI32Imm(PPC::getVSPLTImmediate(N, 4, *CurDAG));
+ return getI32Imm(PPC::getVSPLTImmediate(N, 4, *CurDAG), SDLoc(N));
}]>;
def vspltw_shuffle : PatFrag<(ops node:$lhs, node:$rhs),
(vector_shuffle node:$lhs, node:$rhs), [{
def : Pat<(v16i8 (bitconvert (v4i32 VRRC:$src))), (v16i8 VRRC:$src)>;
def : Pat<(v16i8 (bitconvert (v4f32 VRRC:$src))), (v16i8 VRRC:$src)>;
def : Pat<(v16i8 (bitconvert (v2i64 VRRC:$src))), (v16i8 VRRC:$src)>;
+def : Pat<(v16i8 (bitconvert (v1i128 VRRC:$src))), (v16i8 VRRC:$src)>;
def : Pat<(v8i16 (bitconvert (v16i8 VRRC:$src))), (v8i16 VRRC:$src)>;
def : Pat<(v8i16 (bitconvert (v4i32 VRRC:$src))), (v8i16 VRRC:$src)>;
def : Pat<(v8i16 (bitconvert (v4f32 VRRC:$src))), (v8i16 VRRC:$src)>;
def : Pat<(v8i16 (bitconvert (v2i64 VRRC:$src))), (v8i16 VRRC:$src)>;
+def : Pat<(v8i16 (bitconvert (v1i128 VRRC:$src))), (v8i16 VRRC:$src)>;
def : Pat<(v4i32 (bitconvert (v16i8 VRRC:$src))), (v4i32 VRRC:$src)>;
def : Pat<(v4i32 (bitconvert (v8i16 VRRC:$src))), (v4i32 VRRC:$src)>;
def : Pat<(v4i32 (bitconvert (v4f32 VRRC:$src))), (v4i32 VRRC:$src)>;
def : Pat<(v4i32 (bitconvert (v2i64 VRRC:$src))), (v4i32 VRRC:$src)>;
+def : Pat<(v4i32 (bitconvert (v1i128 VRRC:$src))), (v4i32 VRRC:$src)>;
def : Pat<(v4f32 (bitconvert (v16i8 VRRC:$src))), (v4f32 VRRC:$src)>;
def : Pat<(v4f32 (bitconvert (v8i16 VRRC:$src))), (v4f32 VRRC:$src)>;
def : Pat<(v4f32 (bitconvert (v4i32 VRRC:$src))), (v4f32 VRRC:$src)>;
def : Pat<(v4f32 (bitconvert (v2i64 VRRC:$src))), (v4f32 VRRC:$src)>;
+def : Pat<(v4f32 (bitconvert (v1i128 VRRC:$src))), (v4f32 VRRC:$src)>;
def : Pat<(v2i64 (bitconvert (v16i8 VRRC:$src))), (v2i64 VRRC:$src)>;
def : Pat<(v2i64 (bitconvert (v8i16 VRRC:$src))), (v2i64 VRRC:$src)>;
def : Pat<(v2i64 (bitconvert (v4i32 VRRC:$src))), (v2i64 VRRC:$src)>;
def : Pat<(v2i64 (bitconvert (v4f32 VRRC:$src))), (v2i64 VRRC:$src)>;
+def : Pat<(v2i64 (bitconvert (v1i128 VRRC:$src))), (v2i64 VRRC:$src)>;
+
+def : Pat<(v1i128 (bitconvert (v16i8 VRRC:$src))), (v1i128 VRRC:$src)>;
+def : Pat<(v1i128 (bitconvert (v8i16 VRRC:$src))), (v1i128 VRRC:$src)>;
+def : Pat<(v1i128 (bitconvert (v4i32 VRRC:$src))), (v1i128 VRRC:$src)>;
+def : Pat<(v1i128 (bitconvert (v4f32 VRRC:$src))), (v1i128 VRRC:$src)>;
+def : Pat<(v1i128 (bitconvert (v2i64 VRRC:$src))), (v1i128 VRRC:$src)>;
// Shuffles.
def VMAXSD : VX1_Int_Ty<450, "vmaxsd", int_ppc_altivec_vmaxsd, v2i64>;
def VMAXUD : VX1_Int_Ty<194, "vmaxud", int_ppc_altivec_vmaxud, v2i64>;
def VMINSD : VX1_Int_Ty<962, "vminsd", int_ppc_altivec_vminsd, v2i64>;
-def VMIDUD : VX1_Int_Ty<706, "vminud", int_ppc_altivec_vminud, v2i64>;
+def VMINUD : VX1_Int_Ty<706, "vminud", int_ppc_altivec_vminud, v2i64>;
} // isCommutable
+// Vector merge
+def VMRGEW : VXForm_1<1932, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
+ "vmrgew $vD, $vA, $vB", IIC_VecFP,
+ [(set v16i8:$vD, (vmrgew_shuffle v16i8:$vA, v16i8:$vB))]>;
+def VMRGOW : VXForm_1<1676, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
+ "vmrgow $vD, $vA, $vB", IIC_VecFP,
+ [(set v16i8:$vD, (vmrgow_shuffle v16i8:$vA, v16i8:$vB))]>;
+
+// Match vmrgew(x,x) and vmrgow(x,x)
+def:Pat<(vmrgew_unary_shuffle v16i8:$vA, undef),
+ (VMRGEW $vA, $vA)>;
+def:Pat<(vmrgow_unary_shuffle v16i8:$vA, undef),
+ (VMRGOW $vA, $vA)>;
+
+// Match vmrgew(y,x) and vmrgow(y,x), i.e., swapped operands. These fragments
+// are matched for little-endian, where the inputs must be swapped for correct
+// semantics.w
+def:Pat<(vmrgew_swapped_shuffle v16i8:$vA, v16i8:$vB),
+ (VMRGEW $vB, $vA)>;
+def:Pat<(vmrgow_swapped_shuffle v16i8:$vA, v16i8:$vB),
+ (VMRGOW $vB, $vA)>;
+
+
// Vector shifts
def VRLD : VX1_Int_Ty<196, "vrld", int_ppc_altivec_vrld, v2i64>;
def VSLD : VXForm_1<1476, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
def VADDUDM : VXForm_1<192, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
"vaddudm $vD, $vA, $vB", IIC_VecGeneral,
[(set v2i64:$vD, (add v2i64:$vA, v2i64:$vB))]>;
+def VADDUQM : VXForm_1<256, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
+ "vadduqm $vD, $vA, $vB", IIC_VecGeneral,
+ [(set v1i128:$vD, (add v1i128:$vA, v1i128:$vB))]>;
} // isCommutable
+// Vector Quadword Add
+def VADDEUQM : VA1a_Int_Ty<60, "vaddeuqm", int_ppc_altivec_vaddeuqm, v1i128>;
+def VADDCUQ : VX1_Int_Ty<320, "vaddcuq", int_ppc_altivec_vaddcuq, v1i128>;
+def VADDECUQ : VA1a_Int_Ty<61, "vaddecuq", int_ppc_altivec_vaddecuq, v1i128>;
+
+// Vector Doubleword Subtract
def VSUBUDM : VXForm_1<1216, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
"vsubudm $vD, $vA, $vB", IIC_VecGeneral,
[(set v2i64:$vD, (sub v2i64:$vA, v2i64:$vB))]>;
+// Vector Quadword Subtract
+def VSUBUQM : VXForm_1<1280, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
+ "vsubuqm $vD, $vA, $vB", IIC_VecGeneral,
+ [(set v1i128:$vD, (sub v1i128:$vA, v1i128:$vB))]>;
+def VSUBEUQM : VA1a_Int_Ty<62, "vsubeuqm", int_ppc_altivec_vsubeuqm, v1i128>;
+def VSUBCUQ : VX1_Int_Ty<1344, "vsubcuq", int_ppc_altivec_vsubcuq, v1i128>;
+def VSUBECUQ : VA1a_Int_Ty<63, "vsubecuq", int_ppc_altivec_vsubecuq, v1i128>;
+
// Count Leading Zeros
def VCLZB : VXForm_2<1794, (outs vrrc:$vD), (ins vrrc:$vB),
"vclzb $vD, $vB", IIC_VecGeneral,
def VPERMXOR : VA1a_Int_Ty<45, "vpermxor",
int_ppc_altivec_crypto_vpermxor, v16i8>;
+// Vector doubleword integer pack and unpack.
+def VPKSDSS : VX1_Int_Ty2<1486, "vpksdss", int_ppc_altivec_vpksdss,
+ v4i32, v2i64>;
+def VPKSDUS : VX1_Int_Ty2<1358, "vpksdus", int_ppc_altivec_vpksdus,
+ v4i32, v2i64>;
+def VPKUDUM : VXForm_1<1102, (outs vrrc:$vD), (ins vrrc:$vA, vrrc:$vB),
+ "vpkudum $vD, $vA, $vB", IIC_VecFP,
+ [(set v16i8:$vD,
+ (vpkudum_shuffle v16i8:$vA, v16i8:$vB))]>;
+def VPKUDUS : VX1_Int_Ty2<1230, "vpkudus", int_ppc_altivec_vpkudus,
+ v4i32, v2i64>;
+def VUPKHSW : VX2_Int_Ty2<1614, "vupkhsw", int_ppc_altivec_vupkhsw,
+ v2i64, v4i32>;
+def VUPKLSW : VX2_Int_Ty2<1742, "vupklsw", int_ppc_altivec_vupklsw,
+ v2i64, v4i32>;
+
+// Shuffle patterns for unary and swapped (LE) vector pack modulo.
+def:Pat<(vpkudum_unary_shuffle v16i8:$vA, undef),
+ (VPKUDUM $vA, $vA)>;
+def:Pat<(vpkudum_swapped_shuffle v16i8:$vA, v16i8:$vB),
+ (VPKUDUM $vB, $vA)>;
+
+def VGBBD : VX2_Int_Ty2<1292, "vgbbd", int_ppc_altivec_vgbbd, v16i8, v16i8>;
+def VBPERMQ : VX1_Int_Ty2<1356, "vbpermq", int_ppc_altivec_vbpermq,
+ v2i64, v16i8>;
} // end HasP8Altivec
// Crypto instructions (from builtins)