diff options
Diffstat (limited to 'lib/Target/ARM/ARMInstrNEON.td')
| -rw-r--r-- | lib/Target/ARM/ARMInstrNEON.td | 99 |
1 files changed, 30 insertions, 69 deletions
diff --git a/lib/Target/ARM/ARMInstrNEON.td b/lib/Target/ARM/ARMInstrNEON.td index 197ec16..a84315f 100644 --- a/lib/Target/ARM/ARMInstrNEON.td +++ b/lib/Target/ARM/ARMInstrNEON.td @@ -98,17 +98,8 @@ def NEONfmin : SDNode<"ARMISD::FMIN", SDTARMFMAX>; // NEON operand definitions //===----------------------------------------------------------------------===// -def h8imm : Operand<i8> { - let PrintMethod = "printHex8ImmOperand"; -} -def h16imm : Operand<i16> { - let PrintMethod = "printHex16ImmOperand"; -} -def h32imm : Operand<i32> { - let PrintMethod = "printHex32ImmOperand"; -} -def h64imm : Operand<i64> { - let PrintMethod = "printHex64ImmOperand"; +def nModImm : Operand<i32> { + let PrintMethod = "printNEONModImmOperand"; } //===----------------------------------------------------------------------===// @@ -812,11 +803,6 @@ def DSubReg_f64_reg : SDNodeXForm<imm, [{ assert(ARM::dsub_7 == ARM::dsub_0+7 && "Unexpected subreg numbering"); return CurDAG->getTargetConstant(ARM::dsub_0 + N->getZExtValue(), MVT::i32); }]>; -def DSubReg_f64_other_reg : SDNodeXForm<imm, [{ - assert(ARM::dsub_7 == ARM::dsub_0+7 && "Unexpected subreg numbering"); - return CurDAG->getTargetConstant(ARM::dsub_0 + (1 - N->getZExtValue()), - MVT::i32); -}]>; // Extract S sub-registers of Q/D registers. def SSubReg_f32_reg : SDNodeXForm<imm, [{ @@ -2282,7 +2268,7 @@ def VCEQfq : N3VQ<0,0,0b00,0b1110,0, IIC_VBINQ, "vceq", "f32", v4i32, v4f32, NEONvceq, 1>; // For disassembly only. defm VCEQz : N2V_QHS_cmp<0b11, 0b11, 0b01, 0b00010, 0, "vceq", "i", - "$dst, $src, #0">; + "$dst, $src, #0">; // VCGE : Vector Compare Greater Than or Equal defm VCGEs : N3V_QHS<0, 0, 0b0011, 1, IIC_VSUBi4D, IIC_VSUBi4D, IIC_VSUBi4Q, @@ -2834,73 +2820,70 @@ def VMOVQQQQ : PseudoInst<(outs QQQQPR:$dst), (ins QQQQPR:$src), // VMOV_get_imm8 xform function: convert build_vector to VMOV.i8 imm. def VMOV_get_imm8 : SDNodeXForm<build_vector, [{ - return ARM::getVMOVImm(N, 1, *CurDAG); + return ARM::getNEONModImm(N, 1, true, *CurDAG); }]>; def vmovImm8 : PatLeaf<(build_vector), [{ - return ARM::getVMOVImm(N, 1, *CurDAG).getNode() != 0; + return ARM::getNEONModImm(N, 1, true, *CurDAG).getNode() != 0; }], VMOV_get_imm8>; // VMOV_get_imm16 xform function: convert build_vector to VMOV.i16 imm. def VMOV_get_imm16 : SDNodeXForm<build_vector, [{ - return ARM::getVMOVImm(N, 2, *CurDAG); + return ARM::getNEONModImm(N, 2, true, *CurDAG); }]>; def vmovImm16 : PatLeaf<(build_vector), [{ - return ARM::getVMOVImm(N, 2, *CurDAG).getNode() != 0; + return ARM::getNEONModImm(N, 2, true, *CurDAG).getNode() != 0; }], VMOV_get_imm16>; // VMOV_get_imm32 xform function: convert build_vector to VMOV.i32 imm. def VMOV_get_imm32 : SDNodeXForm<build_vector, [{ - return ARM::getVMOVImm(N, 4, *CurDAG); + return ARM::getNEONModImm(N, 4, true, *CurDAG); }]>; def vmovImm32 : PatLeaf<(build_vector), [{ - return ARM::getVMOVImm(N, 4, *CurDAG).getNode() != 0; + return ARM::getNEONModImm(N, 4, true, *CurDAG).getNode() != 0; }], VMOV_get_imm32>; // VMOV_get_imm64 xform function: convert build_vector to VMOV.i64 imm. def VMOV_get_imm64 : SDNodeXForm<build_vector, [{ - return ARM::getVMOVImm(N, 8, *CurDAG); + return ARM::getNEONModImm(N, 8, true, *CurDAG); }]>; def vmovImm64 : PatLeaf<(build_vector), [{ - return ARM::getVMOVImm(N, 8, *CurDAG).getNode() != 0; + return ARM::getNEONModImm(N, 8, true, *CurDAG).getNode() != 0; }], VMOV_get_imm64>; -// Note: Some of the cmode bits in the following VMOV instructions need to -// be encoded based on the immed values. - let isReMaterializable = 1 in { def VMOVv8i8 : N1ModImm<1, 0b000, 0b1110, 0, 0, 0, 1, (outs DPR:$dst), - (ins h8imm:$SIMM), IIC_VMOVImm, + (ins nModImm:$SIMM), IIC_VMOVImm, "vmov", "i8", "$dst, $SIMM", "", [(set DPR:$dst, (v8i8 vmovImm8:$SIMM))]>; def VMOVv16i8 : N1ModImm<1, 0b000, 0b1110, 0, 1, 0, 1, (outs QPR:$dst), - (ins h8imm:$SIMM), IIC_VMOVImm, + (ins nModImm:$SIMM), IIC_VMOVImm, "vmov", "i8", "$dst, $SIMM", "", [(set QPR:$dst, (v16i8 vmovImm8:$SIMM))]>; -def VMOVv4i16 : N1ModImm<1, 0b000, {1,0,?,?}, 0, 0, {?}, 1, (outs DPR:$dst), - (ins h16imm:$SIMM), IIC_VMOVImm, +def VMOVv4i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 0, 0, 1, (outs DPR:$dst), + (ins nModImm:$SIMM), IIC_VMOVImm, "vmov", "i16", "$dst, $SIMM", "", [(set DPR:$dst, (v4i16 vmovImm16:$SIMM))]>; -def VMOVv8i16 : N1ModImm<1, 0b000, {1,0,?,?}, 0, 1, {?}, 1, (outs QPR:$dst), - (ins h16imm:$SIMM), IIC_VMOVImm, +def VMOVv8i16 : N1ModImm<1, 0b000, {1,0,?,0}, 0, 1, 0, 1, (outs QPR:$dst), + (ins nModImm:$SIMM), IIC_VMOVImm, "vmov", "i16", "$dst, $SIMM", "", [(set QPR:$dst, (v8i16 vmovImm16:$SIMM))]>; -def VMOVv2i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 0, {?}, 1, (outs DPR:$dst), - (ins h32imm:$SIMM), IIC_VMOVImm, +def VMOVv2i32 : N1ModImm<1, 0b000, {0,?,?,0}, 0, 0, 0, 1, (outs DPR:$dst), + (ins nModImm:$SIMM), IIC_VMOVImm, "vmov", "i32", "$dst, $SIMM", "", [(set DPR:$dst, (v2i32 vmovImm32:$SIMM))]>; -def VMOVv4i32 : N1ModImm<1, 0b000, {?,?,?,?}, 0, 1, {?}, 1, (outs QPR:$dst), - (ins h32imm:$SIMM), IIC_VMOVImm, +def VMOVv4i32 : N1ModImm<1, 0b000, {0,?,?,0}, 0, 1, 0, 1, (outs QPR:$dst), + (ins nModImm:$SIMM), IIC_VMOVImm, "vmov", "i32", "$dst, $SIMM", "", [(set QPR:$dst, (v4i32 vmovImm32:$SIMM))]>; def VMOVv1i64 : N1ModImm<1, 0b000, 0b1110, 0, 0, 1, 1, (outs DPR:$dst), - (ins h64imm:$SIMM), IIC_VMOVImm, + (ins nModImm:$SIMM), IIC_VMOVImm, "vmov", "i64", "$dst, $SIMM", "", [(set DPR:$dst, (v1i64 vmovImm64:$SIMM))]>; def VMOVv2i64 : N1ModImm<1, 0b000, 0b1110, 0, 1, 1, 1, (outs QPR:$dst), - (ins h64imm:$SIMM), IIC_VMOVImm, + (ins nModImm:$SIMM), IIC_VMOVImm, "vmov", "i64", "$dst, $SIMM", "", [(set QPR:$dst, (v2i64 vmovImm64:$SIMM))]>; } // isReMaterializable @@ -3122,17 +3105,6 @@ def VDUPfqf : N2V<0b11, 0b11, {?,1}, {0,0}, 0b11000, 1, 0, IIC_VMOVD, "vdup", "32", "$dst, ${src:lane}", "", [(set QPR:$dst, (v4f32 (NEONvdup (f32 SPR:$src))))]>; -def : Pat<(v2i64 (NEONvduplane (v2i64 QPR:$src), imm:$lane)), - (INSERT_SUBREG QPR:$src, - (i64 (EXTRACT_SUBREG QPR:$src, - (DSubReg_f64_reg imm:$lane))), - (DSubReg_f64_other_reg imm:$lane))>; -def : Pat<(v2f64 (NEONvduplane (v2f64 QPR:$src), imm:$lane)), - (INSERT_SUBREG QPR:$src, - (f64 (EXTRACT_SUBREG QPR:$src, - (DSubReg_f64_reg imm:$lane))), - (DSubReg_f64_other_reg imm:$lane))>; - // VMOVN : Vector Narrowing Move defm VMOVN : N2VNInt_HSD<0b11,0b11,0b10,0b00100,0,0, IIC_VMOVD, "vmovn", "i", int_arm_neon_vmovn>; @@ -3319,22 +3291,16 @@ let hasExtraSrcRegAllocReq = 1 in { def VTBL2 : N3V<1,1,0b11,0b1001,0,0, (outs DPR:$dst), (ins DPR:$tbl1, DPR:$tbl2, DPR:$src), NVTBLFrm, IIC_VTB2, - "vtbl", "8", "$dst, \\{$tbl1, $tbl2\\}, $src", "", - [(set DPR:$dst, (v8i8 (int_arm_neon_vtbl2 - DPR:$tbl1, DPR:$tbl2, DPR:$src)))]>; + "vtbl", "8", "$dst, \\{$tbl1, $tbl2\\}, $src", "", []>; def VTBL3 : N3V<1,1,0b11,0b1010,0,0, (outs DPR:$dst), (ins DPR:$tbl1, DPR:$tbl2, DPR:$tbl3, DPR:$src), NVTBLFrm, IIC_VTB3, - "vtbl", "8", "$dst, \\{$tbl1, $tbl2, $tbl3\\}, $src", "", - [(set DPR:$dst, (v8i8 (int_arm_neon_vtbl3 - DPR:$tbl1, DPR:$tbl2, DPR:$tbl3, DPR:$src)))]>; + "vtbl", "8", "$dst, \\{$tbl1, $tbl2, $tbl3\\}, $src", "", []>; def VTBL4 : N3V<1,1,0b11,0b1011,0,0, (outs DPR:$dst), (ins DPR:$tbl1, DPR:$tbl2, DPR:$tbl3, DPR:$tbl4, DPR:$src), NVTBLFrm, IIC_VTB4, - "vtbl", "8", "$dst, \\{$tbl1, $tbl2, $tbl3, $tbl4\\}, $src", "", - [(set DPR:$dst, (v8i8 (int_arm_neon_vtbl4 DPR:$tbl1, DPR:$tbl2, - DPR:$tbl3, DPR:$tbl4, DPR:$src)))]>; + "vtbl", "8", "$dst, \\{$tbl1, $tbl2, $tbl3, $tbl4\\}, $src", "", []>; } // hasExtraSrcRegAllocReq = 1 // VTBX : Vector Table Extension @@ -3348,23 +3314,18 @@ let hasExtraSrcRegAllocReq = 1 in { def VTBX2 : N3V<1,1,0b11,0b1001,1,0, (outs DPR:$dst), (ins DPR:$orig, DPR:$tbl1, DPR:$tbl2, DPR:$src), NVTBLFrm, IIC_VTBX2, - "vtbx", "8", "$dst, \\{$tbl1, $tbl2\\}, $src", "$orig = $dst", - [(set DPR:$dst, (v8i8 (int_arm_neon_vtbx2 - DPR:$orig, DPR:$tbl1, DPR:$tbl2, DPR:$src)))]>; + "vtbx", "8", "$dst, \\{$tbl1, $tbl2\\}, $src", "$orig = $dst", []>; def VTBX3 : N3V<1,1,0b11,0b1010,1,0, (outs DPR:$dst), (ins DPR:$orig, DPR:$tbl1, DPR:$tbl2, DPR:$tbl3, DPR:$src), NVTBLFrm, IIC_VTBX3, - "vtbx", "8", "$dst, \\{$tbl1, $tbl2, $tbl3\\}, $src", "$orig = $dst", - [(set DPR:$dst, (v8i8 (int_arm_neon_vtbx3 DPR:$orig, DPR:$tbl1, - DPR:$tbl2, DPR:$tbl3, DPR:$src)))]>; + "vtbx", "8", "$dst, \\{$tbl1, $tbl2, $tbl3\\}, $src", + "$orig = $dst", []>; def VTBX4 : N3V<1,1,0b11,0b1011,1,0, (outs DPR:$dst), (ins DPR:$orig, DPR:$tbl1, DPR:$tbl2, DPR:$tbl3, DPR:$tbl4, DPR:$src), NVTBLFrm, IIC_VTBX4, "vtbx", "8", "$dst, \\{$tbl1, $tbl2, $tbl3, $tbl4\\}, $src", - "$orig = $dst", - [(set DPR:$dst, (v8i8 (int_arm_neon_vtbx4 DPR:$orig, DPR:$tbl1, - DPR:$tbl2, DPR:$tbl3, DPR:$tbl4, DPR:$src)))]>; + "$orig = $dst", []>; } // hasExtraSrcRegAllocReq = 1 //===----------------------------------------------------------------------===// |
