1 | //===-- HexagonISelDAGToDAG.cpp - A dag to dag inst selector for Hexagon --===// |
2 | // |
3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | // See https://llvm.org/LICENSE.txt for license information. |
5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | // |
7 | //===----------------------------------------------------------------------===// |
8 | // |
9 | // This file defines an instruction selector for the Hexagon target. |
10 | // |
11 | //===----------------------------------------------------------------------===// |
12 | |
13 | #include "HexagonISelDAGToDAG.h" |
14 | #include "Hexagon.h" |
15 | #include "HexagonISelLowering.h" |
16 | #include "HexagonMachineFunctionInfo.h" |
17 | #include "HexagonTargetMachine.h" |
18 | #include "llvm/CodeGen/FunctionLoweringInfo.h" |
19 | #include "llvm/CodeGen/MachineInstrBuilder.h" |
20 | #include "llvm/CodeGen/SelectionDAGISel.h" |
21 | #include "llvm/IR/Intrinsics.h" |
22 | #include "llvm/IR/IntrinsicsHexagon.h" |
23 | #include "llvm/Support/CommandLine.h" |
24 | #include "llvm/Support/Debug.h" |
25 | using namespace llvm; |
26 | |
27 | #define DEBUG_TYPE "hexagon-isel" |
28 | #define PASS_NAME "Hexagon DAG->DAG Pattern Instruction Selection" |
29 | |
30 | static |
31 | cl::opt<bool> |
32 | EnableAddressRebalancing("isel-rebalance-addr" , cl::Hidden, cl::init(Val: true), |
33 | cl::desc("Rebalance address calculation trees to improve " |
34 | "instruction selection" )); |
35 | |
36 | // Rebalance only if this allows e.g. combining a GA with an offset or |
37 | // factoring out a shift. |
38 | static |
39 | cl::opt<bool> |
40 | RebalanceOnlyForOptimizations("rebalance-only-opt" , cl::Hidden, cl::init(Val: false), |
41 | cl::desc("Rebalance address tree only if this allows optimizations" )); |
42 | |
43 | static |
44 | cl::opt<bool> |
45 | RebalanceOnlyImbalancedTrees("rebalance-only-imbal" , cl::Hidden, |
46 | cl::init(Val: false), cl::desc("Rebalance address tree only if it is imbalanced" )); |
47 | |
48 | static cl::opt<bool> CheckSingleUse("hexagon-isel-su" , cl::Hidden, |
49 | cl::init(Val: true), cl::desc("Enable checking of SDNode's single-use status" )); |
50 | |
51 | //===----------------------------------------------------------------------===// |
52 | // Instruction Selector Implementation |
53 | //===----------------------------------------------------------------------===// |
54 | |
55 | #define GET_DAGISEL_BODY HexagonDAGToDAGISel |
56 | #include "HexagonGenDAGISel.inc" |
57 | |
58 | namespace llvm { |
59 | /// createHexagonISelDag - This pass converts a legalized DAG into a |
60 | /// Hexagon-specific DAG, ready for instruction scheduling. |
61 | FunctionPass *createHexagonISelDag(HexagonTargetMachine &TM, |
62 | CodeGenOptLevel OptLevel) { |
63 | return new HexagonDAGToDAGISelLegacy(TM, OptLevel); |
64 | } |
65 | } |
66 | |
67 | HexagonDAGToDAGISelLegacy::HexagonDAGToDAGISelLegacy(HexagonTargetMachine &tm, |
68 | CodeGenOptLevel OptLevel) |
69 | : SelectionDAGISelLegacy( |
70 | ID, std::make_unique<HexagonDAGToDAGISel>(args&: tm, args&: OptLevel)) {} |
71 | |
72 | char HexagonDAGToDAGISelLegacy::ID = 0; |
73 | |
74 | INITIALIZE_PASS(HexagonDAGToDAGISelLegacy, DEBUG_TYPE, PASS_NAME, false, false) |
75 | |
76 | void HexagonDAGToDAGISel::SelectIndexedLoad(LoadSDNode *LD, const SDLoc &dl) { |
77 | SDValue Chain = LD->getChain(); |
78 | SDValue Base = LD->getBasePtr(); |
79 | SDValue Offset = LD->getOffset(); |
80 | int32_t Inc = cast<ConstantSDNode>(Val: Offset.getNode())->getSExtValue(); |
81 | EVT LoadedVT = LD->getMemoryVT(); |
82 | unsigned Opcode = 0; |
83 | |
84 | // Check for zero extended loads. Treat any-extend loads as zero extended |
85 | // loads. |
86 | ISD::LoadExtType ExtType = LD->getExtensionType(); |
87 | bool IsZeroExt = (ExtType == ISD::ZEXTLOAD || ExtType == ISD::EXTLOAD); |
88 | bool IsValidInc = HII->isValidAutoIncImm(VT: LoadedVT, Offset: Inc); |
89 | |
90 | assert(LoadedVT.isSimple()); |
91 | switch (LoadedVT.getSimpleVT().SimpleTy) { |
92 | case MVT::i8: |
93 | if (IsZeroExt) |
94 | Opcode = IsValidInc ? Hexagon::L2_loadrub_pi : Hexagon::L2_loadrub_io; |
95 | else |
96 | Opcode = IsValidInc ? Hexagon::L2_loadrb_pi : Hexagon::L2_loadrb_io; |
97 | break; |
98 | case MVT::i16: |
99 | if (IsZeroExt) |
100 | Opcode = IsValidInc ? Hexagon::L2_loadruh_pi : Hexagon::L2_loadruh_io; |
101 | else |
102 | Opcode = IsValidInc ? Hexagon::L2_loadrh_pi : Hexagon::L2_loadrh_io; |
103 | break; |
104 | case MVT::i32: |
105 | case MVT::f32: |
106 | case MVT::v2i16: |
107 | case MVT::v4i8: |
108 | Opcode = IsValidInc ? Hexagon::L2_loadri_pi : Hexagon::L2_loadri_io; |
109 | break; |
110 | case MVT::i64: |
111 | case MVT::f64: |
112 | case MVT::v2i32: |
113 | case MVT::v4i16: |
114 | case MVT::v8i8: |
115 | Opcode = IsValidInc ? Hexagon::L2_loadrd_pi : Hexagon::L2_loadrd_io; |
116 | break; |
117 | case MVT::v64i8: |
118 | case MVT::v32i16: |
119 | case MVT::v16i32: |
120 | case MVT::v8i64: |
121 | case MVT::v128i8: |
122 | case MVT::v64i16: |
123 | case MVT::v32i32: |
124 | case MVT::v16i64: |
125 | if (isAlignedMemNode(N: LD)) { |
126 | if (LD->isNonTemporal()) |
127 | Opcode = IsValidInc ? Hexagon::V6_vL32b_nt_pi : Hexagon::V6_vL32b_nt_ai; |
128 | else |
129 | Opcode = IsValidInc ? Hexagon::V6_vL32b_pi : Hexagon::V6_vL32b_ai; |
130 | } else { |
131 | Opcode = IsValidInc ? Hexagon::V6_vL32Ub_pi : Hexagon::V6_vL32Ub_ai; |
132 | } |
133 | break; |
134 | default: |
135 | llvm_unreachable("Unexpected memory type in indexed load" ); |
136 | } |
137 | |
138 | SDValue IncV = CurDAG->getTargetConstant(Val: Inc, DL: dl, VT: MVT::i32); |
139 | MachineMemOperand *MemOp = LD->getMemOperand(); |
140 | |
141 | auto getExt64 = [this,ExtType] (MachineSDNode *N, const SDLoc &dl) |
142 | -> MachineSDNode* { |
143 | if (ExtType == ISD::ZEXTLOAD || ExtType == ISD::EXTLOAD) { |
144 | SDValue Zero = CurDAG->getTargetConstant(Val: 0, DL: dl, VT: MVT::i32); |
145 | return CurDAG->getMachineNode(Opcode: Hexagon::A4_combineir, dl, VT: MVT::i64, |
146 | Op1: Zero, Op2: SDValue(N, 0)); |
147 | } |
148 | if (ExtType == ISD::SEXTLOAD) |
149 | return CurDAG->getMachineNode(Opcode: Hexagon::A2_sxtw, dl, VT: MVT::i64, |
150 | Op1: SDValue(N, 0)); |
151 | return N; |
152 | }; |
153 | |
154 | // Loaded value Next address Chain |
155 | SDValue From[3] = { SDValue(LD,0), SDValue(LD,1), SDValue(LD,2) }; |
156 | SDValue To[3]; |
157 | |
158 | EVT ValueVT = LD->getValueType(ResNo: 0); |
159 | if (ValueVT == MVT::i64 && ExtType != ISD::NON_EXTLOAD) { |
160 | // A load extending to i64 will actually produce i32, which will then |
161 | // need to be extended to i64. |
162 | assert(LoadedVT.getSizeInBits() <= 32); |
163 | ValueVT = MVT::i32; |
164 | } |
165 | |
166 | if (IsValidInc) { |
167 | MachineSDNode *L = CurDAG->getMachineNode(Opcode, dl, VT1: ValueVT, |
168 | VT2: MVT::i32, VT3: MVT::Other, Op1: Base, |
169 | Op2: IncV, Op3: Chain); |
170 | CurDAG->setNodeMemRefs(N: L, NewMemRefs: {MemOp}); |
171 | To[1] = SDValue(L, 1); // Next address. |
172 | To[2] = SDValue(L, 2); // Chain. |
173 | // Handle special case for extension to i64. |
174 | if (LD->getValueType(ResNo: 0) == MVT::i64) |
175 | L = getExt64(L, dl); |
176 | To[0] = SDValue(L, 0); // Loaded (extended) value. |
177 | } else { |
178 | SDValue Zero = CurDAG->getTargetConstant(Val: 0, DL: dl, VT: MVT::i32); |
179 | MachineSDNode *L = CurDAG->getMachineNode(Opcode, dl, VT1: ValueVT, VT2: MVT::Other, |
180 | Op1: Base, Op2: Zero, Op3: Chain); |
181 | CurDAG->setNodeMemRefs(N: L, NewMemRefs: {MemOp}); |
182 | To[2] = SDValue(L, 1); // Chain. |
183 | MachineSDNode *A = CurDAG->getMachineNode(Opcode: Hexagon::A2_addi, dl, VT: MVT::i32, |
184 | Op1: Base, Op2: IncV); |
185 | To[1] = SDValue(A, 0); // Next address. |
186 | // Handle special case for extension to i64. |
187 | if (LD->getValueType(ResNo: 0) == MVT::i64) |
188 | L = getExt64(L, dl); |
189 | To[0] = SDValue(L, 0); // Loaded (extended) value. |
190 | } |
191 | ReplaceUses(F: From, T: To, Num: 3); |
192 | CurDAG->RemoveDeadNode(N: LD); |
193 | } |
194 | |
195 | MachineSDNode *HexagonDAGToDAGISel::LoadInstrForLoadIntrinsic(SDNode *IntN) { |
196 | if (IntN->getOpcode() != ISD::INTRINSIC_W_CHAIN) |
197 | return nullptr; |
198 | |
199 | SDLoc dl(IntN); |
200 | unsigned IntNo = IntN->getConstantOperandVal(Num: 1); |
201 | |
202 | static std::map<unsigned,unsigned> LoadPciMap = { |
203 | { Intrinsic::hexagon_circ_ldb, Hexagon::L2_loadrb_pci }, |
204 | { Intrinsic::hexagon_circ_ldub, Hexagon::L2_loadrub_pci }, |
205 | { Intrinsic::hexagon_circ_ldh, Hexagon::L2_loadrh_pci }, |
206 | { Intrinsic::hexagon_circ_lduh, Hexagon::L2_loadruh_pci }, |
207 | { Intrinsic::hexagon_circ_ldw, Hexagon::L2_loadri_pci }, |
208 | { Intrinsic::hexagon_circ_ldd, Hexagon::L2_loadrd_pci }, |
209 | }; |
210 | auto FLC = LoadPciMap.find(x: IntNo); |
211 | if (FLC != LoadPciMap.end()) { |
212 | EVT ValTy = (IntNo == Intrinsic::hexagon_circ_ldd) ? MVT::i64 : MVT::i32; |
213 | EVT RTys[] = { ValTy, MVT::i32, MVT::Other }; |
214 | // Operands: { Base, Increment, Modifier, Chain } |
215 | auto Inc = cast<ConstantSDNode>(Val: IntN->getOperand(Num: 5)); |
216 | SDValue I = CurDAG->getTargetConstant(Val: Inc->getSExtValue(), DL: dl, VT: MVT::i32); |
217 | MachineSDNode *Res = CurDAG->getMachineNode(Opcode: FLC->second, dl, ResultTys: RTys, |
218 | Ops: { IntN->getOperand(Num: 2), I, IntN->getOperand(Num: 4), |
219 | IntN->getOperand(Num: 0) }); |
220 | return Res; |
221 | } |
222 | |
223 | return nullptr; |
224 | } |
225 | |
226 | SDNode *HexagonDAGToDAGISel::StoreInstrForLoadIntrinsic(MachineSDNode *LoadN, |
227 | SDNode *IntN) { |
228 | // The "LoadN" is just a machine load instruction. The intrinsic also |
229 | // involves storing it. Generate an appropriate store to the location |
230 | // given in the intrinsic's operand(3). |
231 | uint64_t F = HII->get(Opcode: LoadN->getMachineOpcode()).TSFlags; |
232 | unsigned SizeBits = (F >> HexagonII::MemAccessSizePos) & |
233 | HexagonII::MemAccesSizeMask; |
234 | unsigned Size = 1U << (SizeBits-1); |
235 | |
236 | SDLoc dl(IntN); |
237 | MachinePointerInfo PI; |
238 | SDValue TS; |
239 | SDValue Loc = IntN->getOperand(Num: 3); |
240 | |
241 | if (Size >= 4) |
242 | TS = CurDAG->getStore(Chain: SDValue(LoadN, 2), dl, Val: SDValue(LoadN, 0), Ptr: Loc, PtrInfo: PI, |
243 | Alignment: Align(Size)); |
244 | else |
245 | TS = CurDAG->getTruncStore(Chain: SDValue(LoadN, 2), dl, Val: SDValue(LoadN, 0), Ptr: Loc, |
246 | PtrInfo: PI, SVT: MVT::getIntegerVT(BitWidth: Size * 8), Alignment: Align(Size)); |
247 | |
248 | SDNode *StoreN; |
249 | { |
250 | HandleSDNode Handle(TS); |
251 | SelectStore(N: TS.getNode()); |
252 | StoreN = Handle.getValue().getNode(); |
253 | } |
254 | |
255 | // Load's results are { Loaded value, Updated pointer, Chain } |
256 | ReplaceUses(F: SDValue(IntN, 0), T: SDValue(LoadN, 1)); |
257 | ReplaceUses(F: SDValue(IntN, 1), T: SDValue(StoreN, 0)); |
258 | return StoreN; |
259 | } |
260 | |
261 | bool HexagonDAGToDAGISel::tryLoadOfLoadIntrinsic(LoadSDNode *N) { |
262 | // The intrinsics for load circ/brev perform two operations: |
263 | // 1. Load a value V from the specified location, using the addressing |
264 | // mode corresponding to the intrinsic. |
265 | // 2. Store V into a specified location. This location is typically a |
266 | // local, temporary object. |
267 | // In many cases, the program using these intrinsics will immediately |
268 | // load V again from the local object. In those cases, when certain |
269 | // conditions are met, the last load can be removed. |
270 | // This function identifies and optimizes this pattern. If the pattern |
271 | // cannot be optimized, it returns nullptr, which will cause the load |
272 | // to be selected separately from the intrinsic (which will be handled |
273 | // in SelectIntrinsicWChain). |
274 | |
275 | SDValue Ch = N->getOperand(Num: 0); |
276 | SDValue Loc = N->getOperand(Num: 1); |
277 | |
278 | // Assume that the load and the intrinsic are connected directly with a |
279 | // chain: |
280 | // t1: i32,ch = int.load ..., ..., ..., Loc, ... // <-- C |
281 | // t2: i32,ch = load t1:1, Loc, ... |
282 | SDNode *C = Ch.getNode(); |
283 | |
284 | if (C->getOpcode() != ISD::INTRINSIC_W_CHAIN) |
285 | return false; |
286 | |
287 | // The second load can only be eliminated if its extension type matches |
288 | // that of the load instruction corresponding to the intrinsic. The user |
289 | // can provide an address of an unsigned variable to store the result of |
290 | // a sign-extending intrinsic into (or the other way around). |
291 | ISD::LoadExtType IntExt; |
292 | switch (C->getConstantOperandVal(Num: 1)) { |
293 | case Intrinsic::hexagon_circ_ldub: |
294 | case Intrinsic::hexagon_circ_lduh: |
295 | IntExt = ISD::ZEXTLOAD; |
296 | break; |
297 | case Intrinsic::hexagon_circ_ldw: |
298 | case Intrinsic::hexagon_circ_ldd: |
299 | IntExt = ISD::NON_EXTLOAD; |
300 | break; |
301 | default: |
302 | IntExt = ISD::SEXTLOAD; |
303 | break; |
304 | } |
305 | if (N->getExtensionType() != IntExt) |
306 | return false; |
307 | |
308 | // Make sure the target location for the loaded value in the load intrinsic |
309 | // is the location from which LD (or N) is loading. |
310 | if (C->getNumOperands() < 4 || Loc.getNode() != C->getOperand(Num: 3).getNode()) |
311 | return false; |
312 | |
313 | if (MachineSDNode *L = LoadInstrForLoadIntrinsic(IntN: C)) { |
314 | SDNode *S = StoreInstrForLoadIntrinsic(LoadN: L, IntN: C); |
315 | SDValue F[] = { SDValue(N,0), SDValue(N,1), SDValue(C,0), SDValue(C,1) }; |
316 | SDValue T[] = { SDValue(L,0), SDValue(S,0), SDValue(L,1), SDValue(S,0) }; |
317 | ReplaceUses(F, T, Num: std::size(T)); |
318 | // This transformation will leave the intrinsic dead. If it remains in |
319 | // the DAG, the selection code will see it again, but without the load, |
320 | // and it will generate a store that is normally required for it. |
321 | CurDAG->RemoveDeadNode(N: C); |
322 | return true; |
323 | } |
324 | return false; |
325 | } |
326 | |
327 | // Convert the bit-reverse load intrinsic to appropriate target instruction. |
328 | bool HexagonDAGToDAGISel::SelectBrevLdIntrinsic(SDNode *IntN) { |
329 | if (IntN->getOpcode() != ISD::INTRINSIC_W_CHAIN) |
330 | return false; |
331 | |
332 | const SDLoc &dl(IntN); |
333 | unsigned IntNo = IntN->getConstantOperandVal(Num: 1); |
334 | |
335 | static const std::map<unsigned, unsigned> LoadBrevMap = { |
336 | { Intrinsic::hexagon_L2_loadrb_pbr, Hexagon::L2_loadrb_pbr }, |
337 | { Intrinsic::hexagon_L2_loadrub_pbr, Hexagon::L2_loadrub_pbr }, |
338 | { Intrinsic::hexagon_L2_loadrh_pbr, Hexagon::L2_loadrh_pbr }, |
339 | { Intrinsic::hexagon_L2_loadruh_pbr, Hexagon::L2_loadruh_pbr }, |
340 | { Intrinsic::hexagon_L2_loadri_pbr, Hexagon::L2_loadri_pbr }, |
341 | { Intrinsic::hexagon_L2_loadrd_pbr, Hexagon::L2_loadrd_pbr } |
342 | }; |
343 | auto FLI = LoadBrevMap.find(x: IntNo); |
344 | if (FLI != LoadBrevMap.end()) { |
345 | EVT ValTy = |
346 | (IntNo == Intrinsic::hexagon_L2_loadrd_pbr) ? MVT::i64 : MVT::i32; |
347 | EVT RTys[] = { ValTy, MVT::i32, MVT::Other }; |
348 | // Operands of Intrinsic: {chain, enum ID of intrinsic, baseptr, |
349 | // modifier}. |
350 | // Operands of target instruction: { Base, Modifier, Chain }. |
351 | MachineSDNode *Res = CurDAG->getMachineNode( |
352 | Opcode: FLI->second, dl, ResultTys: RTys, |
353 | Ops: {IntN->getOperand(Num: 2), IntN->getOperand(Num: 3), IntN->getOperand(Num: 0)}); |
354 | |
355 | MachineMemOperand *MemOp = cast<MemIntrinsicSDNode>(Val: IntN)->getMemOperand(); |
356 | CurDAG->setNodeMemRefs(N: Res, NewMemRefs: {MemOp}); |
357 | |
358 | ReplaceUses(F: SDValue(IntN, 0), T: SDValue(Res, 0)); |
359 | ReplaceUses(F: SDValue(IntN, 1), T: SDValue(Res, 1)); |
360 | ReplaceUses(F: SDValue(IntN, 2), T: SDValue(Res, 2)); |
361 | CurDAG->RemoveDeadNode(N: IntN); |
362 | return true; |
363 | } |
364 | return false; |
365 | } |
366 | |
367 | /// Generate a machine instruction node for the new circular buffer intrinsics. |
368 | /// The new versions use a CSx register instead of the K field. |
369 | bool HexagonDAGToDAGISel::SelectNewCircIntrinsic(SDNode *IntN) { |
370 | if (IntN->getOpcode() != ISD::INTRINSIC_W_CHAIN) |
371 | return false; |
372 | |
373 | SDLoc DL(IntN); |
374 | unsigned IntNo = IntN->getConstantOperandVal(Num: 1); |
375 | SmallVector<SDValue, 7> Ops; |
376 | |
377 | static std::map<unsigned,unsigned> LoadNPcMap = { |
378 | { Intrinsic::hexagon_L2_loadrub_pci, Hexagon::PS_loadrub_pci }, |
379 | { Intrinsic::hexagon_L2_loadrb_pci, Hexagon::PS_loadrb_pci }, |
380 | { Intrinsic::hexagon_L2_loadruh_pci, Hexagon::PS_loadruh_pci }, |
381 | { Intrinsic::hexagon_L2_loadrh_pci, Hexagon::PS_loadrh_pci }, |
382 | { Intrinsic::hexagon_L2_loadri_pci, Hexagon::PS_loadri_pci }, |
383 | { Intrinsic::hexagon_L2_loadrd_pci, Hexagon::PS_loadrd_pci }, |
384 | { Intrinsic::hexagon_L2_loadrub_pcr, Hexagon::PS_loadrub_pcr }, |
385 | { Intrinsic::hexagon_L2_loadrb_pcr, Hexagon::PS_loadrb_pcr }, |
386 | { Intrinsic::hexagon_L2_loadruh_pcr, Hexagon::PS_loadruh_pcr }, |
387 | { Intrinsic::hexagon_L2_loadrh_pcr, Hexagon::PS_loadrh_pcr }, |
388 | { Intrinsic::hexagon_L2_loadri_pcr, Hexagon::PS_loadri_pcr }, |
389 | { Intrinsic::hexagon_L2_loadrd_pcr, Hexagon::PS_loadrd_pcr } |
390 | }; |
391 | auto FLI = LoadNPcMap.find (x: IntNo); |
392 | if (FLI != LoadNPcMap.end()) { |
393 | EVT ValTy = MVT::i32; |
394 | if (IntNo == Intrinsic::hexagon_L2_loadrd_pci || |
395 | IntNo == Intrinsic::hexagon_L2_loadrd_pcr) |
396 | ValTy = MVT::i64; |
397 | EVT RTys[] = { ValTy, MVT::i32, MVT::Other }; |
398 | // Handle load.*_pci case which has 6 operands. |
399 | if (IntN->getNumOperands() == 6) { |
400 | auto Inc = cast<ConstantSDNode>(Val: IntN->getOperand(Num: 3)); |
401 | SDValue I = CurDAG->getTargetConstant(Val: Inc->getSExtValue(), DL, VT: MVT::i32); |
402 | // Operands: { Base, Increment, Modifier, Start, Chain }. |
403 | Ops = { IntN->getOperand(Num: 2), I, IntN->getOperand(Num: 4), IntN->getOperand(Num: 5), |
404 | IntN->getOperand(Num: 0) }; |
405 | } else |
406 | // Handle load.*_pcr case which has 5 operands. |
407 | // Operands: { Base, Modifier, Start, Chain }. |
408 | Ops = { IntN->getOperand(Num: 2), IntN->getOperand(Num: 3), IntN->getOperand(Num: 4), |
409 | IntN->getOperand(Num: 0) }; |
410 | MachineSDNode *Res = CurDAG->getMachineNode(Opcode: FLI->second, dl: DL, ResultTys: RTys, Ops); |
411 | ReplaceUses(F: SDValue(IntN, 0), T: SDValue(Res, 0)); |
412 | ReplaceUses(F: SDValue(IntN, 1), T: SDValue(Res, 1)); |
413 | ReplaceUses(F: SDValue(IntN, 2), T: SDValue(Res, 2)); |
414 | CurDAG->RemoveDeadNode(N: IntN); |
415 | return true; |
416 | } |
417 | |
418 | static std::map<unsigned,unsigned> StoreNPcMap = { |
419 | { Intrinsic::hexagon_S2_storerb_pci, Hexagon::PS_storerb_pci }, |
420 | { Intrinsic::hexagon_S2_storerh_pci, Hexagon::PS_storerh_pci }, |
421 | { Intrinsic::hexagon_S2_storerf_pci, Hexagon::PS_storerf_pci }, |
422 | { Intrinsic::hexagon_S2_storeri_pci, Hexagon::PS_storeri_pci }, |
423 | { Intrinsic::hexagon_S2_storerd_pci, Hexagon::PS_storerd_pci }, |
424 | { Intrinsic::hexagon_S2_storerb_pcr, Hexagon::PS_storerb_pcr }, |
425 | { Intrinsic::hexagon_S2_storerh_pcr, Hexagon::PS_storerh_pcr }, |
426 | { Intrinsic::hexagon_S2_storerf_pcr, Hexagon::PS_storerf_pcr }, |
427 | { Intrinsic::hexagon_S2_storeri_pcr, Hexagon::PS_storeri_pcr }, |
428 | { Intrinsic::hexagon_S2_storerd_pcr, Hexagon::PS_storerd_pcr } |
429 | }; |
430 | auto FSI = StoreNPcMap.find (x: IntNo); |
431 | if (FSI != StoreNPcMap.end()) { |
432 | EVT RTys[] = { MVT::i32, MVT::Other }; |
433 | // Handle store.*_pci case which has 7 operands. |
434 | if (IntN->getNumOperands() == 7) { |
435 | auto Inc = cast<ConstantSDNode>(Val: IntN->getOperand(Num: 3)); |
436 | SDValue I = CurDAG->getTargetConstant(Val: Inc->getSExtValue(), DL, VT: MVT::i32); |
437 | // Operands: { Base, Increment, Modifier, Value, Start, Chain }. |
438 | Ops = { IntN->getOperand(Num: 2), I, IntN->getOperand(Num: 4), IntN->getOperand(Num: 5), |
439 | IntN->getOperand(Num: 6), IntN->getOperand(Num: 0) }; |
440 | } else |
441 | // Handle store.*_pcr case which has 6 operands. |
442 | // Operands: { Base, Modifier, Value, Start, Chain }. |
443 | Ops = { IntN->getOperand(Num: 2), IntN->getOperand(Num: 3), IntN->getOperand(Num: 4), |
444 | IntN->getOperand(Num: 5), IntN->getOperand(Num: 0) }; |
445 | MachineSDNode *Res = CurDAG->getMachineNode(Opcode: FSI->second, dl: DL, ResultTys: RTys, Ops); |
446 | ReplaceUses(F: SDValue(IntN, 0), T: SDValue(Res, 0)); |
447 | ReplaceUses(F: SDValue(IntN, 1), T: SDValue(Res, 1)); |
448 | CurDAG->RemoveDeadNode(N: IntN); |
449 | return true; |
450 | } |
451 | |
452 | return false; |
453 | } |
454 | |
455 | void HexagonDAGToDAGISel::SelectLoad(SDNode *N) { |
456 | SDLoc dl(N); |
457 | LoadSDNode *LD = cast<LoadSDNode>(Val: N); |
458 | |
459 | // Handle indexed loads. |
460 | ISD::MemIndexedMode AM = LD->getAddressingMode(); |
461 | if (AM != ISD::UNINDEXED) { |
462 | SelectIndexedLoad(LD, dl); |
463 | return; |
464 | } |
465 | |
466 | // Handle patterns using circ/brev load intrinsics. |
467 | if (tryLoadOfLoadIntrinsic(N: LD)) |
468 | return; |
469 | |
470 | SelectCode(N: LD); |
471 | } |
472 | |
473 | void HexagonDAGToDAGISel::SelectIndexedStore(StoreSDNode *ST, const SDLoc &dl) { |
474 | SDValue Chain = ST->getChain(); |
475 | SDValue Base = ST->getBasePtr(); |
476 | SDValue Offset = ST->getOffset(); |
477 | SDValue Value = ST->getValue(); |
478 | // Get the constant value. |
479 | int32_t Inc = cast<ConstantSDNode>(Val: Offset.getNode())->getSExtValue(); |
480 | EVT StoredVT = ST->getMemoryVT(); |
481 | EVT ValueVT = Value.getValueType(); |
482 | |
483 | bool IsValidInc = HII->isValidAutoIncImm(VT: StoredVT, Offset: Inc); |
484 | unsigned Opcode = 0; |
485 | |
486 | assert(StoredVT.isSimple()); |
487 | switch (StoredVT.getSimpleVT().SimpleTy) { |
488 | case MVT::i8: |
489 | Opcode = IsValidInc ? Hexagon::S2_storerb_pi : Hexagon::S2_storerb_io; |
490 | break; |
491 | case MVT::i16: |
492 | Opcode = IsValidInc ? Hexagon::S2_storerh_pi : Hexagon::S2_storerh_io; |
493 | break; |
494 | case MVT::i32: |
495 | case MVT::f32: |
496 | case MVT::v2i16: |
497 | case MVT::v4i8: |
498 | Opcode = IsValidInc ? Hexagon::S2_storeri_pi : Hexagon::S2_storeri_io; |
499 | break; |
500 | case MVT::i64: |
501 | case MVT::f64: |
502 | case MVT::v2i32: |
503 | case MVT::v4i16: |
504 | case MVT::v8i8: |
505 | Opcode = IsValidInc ? Hexagon::S2_storerd_pi : Hexagon::S2_storerd_io; |
506 | break; |
507 | case MVT::v64i8: |
508 | case MVT::v32i16: |
509 | case MVT::v16i32: |
510 | case MVT::v8i64: |
511 | case MVT::v128i8: |
512 | case MVT::v64i16: |
513 | case MVT::v32i32: |
514 | case MVT::v16i64: |
515 | if (isAlignedMemNode(N: ST)) { |
516 | if (ST->isNonTemporal()) |
517 | Opcode = IsValidInc ? Hexagon::V6_vS32b_nt_pi : Hexagon::V6_vS32b_nt_ai; |
518 | else |
519 | Opcode = IsValidInc ? Hexagon::V6_vS32b_pi : Hexagon::V6_vS32b_ai; |
520 | } else { |
521 | Opcode = IsValidInc ? Hexagon::V6_vS32Ub_pi : Hexagon::V6_vS32Ub_ai; |
522 | } |
523 | break; |
524 | default: |
525 | llvm_unreachable("Unexpected memory type in indexed store" ); |
526 | } |
527 | |
528 | if (ST->isTruncatingStore() && ValueVT.getSizeInBits() == 64) { |
529 | assert(StoredVT.getSizeInBits() < 64 && "Not a truncating store" ); |
530 | Value = CurDAG->getTargetExtractSubreg(SRIdx: Hexagon::isub_lo, |
531 | DL: dl, VT: MVT::i32, Operand: Value); |
532 | } |
533 | |
534 | SDValue IncV = CurDAG->getTargetConstant(Val: Inc, DL: dl, VT: MVT::i32); |
535 | MachineMemOperand *MemOp = ST->getMemOperand(); |
536 | |
537 | // Next address Chain |
538 | SDValue From[2] = { SDValue(ST,0), SDValue(ST,1) }; |
539 | SDValue To[2]; |
540 | |
541 | if (IsValidInc) { |
542 | // Build post increment store. |
543 | SDValue Ops[] = { Base, IncV, Value, Chain }; |
544 | MachineSDNode *S = CurDAG->getMachineNode(Opcode, dl, VT1: MVT::i32, VT2: MVT::Other, |
545 | Ops); |
546 | CurDAG->setNodeMemRefs(N: S, NewMemRefs: {MemOp}); |
547 | To[0] = SDValue(S, 0); |
548 | To[1] = SDValue(S, 1); |
549 | } else { |
550 | SDValue Zero = CurDAG->getTargetConstant(Val: 0, DL: dl, VT: MVT::i32); |
551 | SDValue Ops[] = { Base, Zero, Value, Chain }; |
552 | MachineSDNode *S = CurDAG->getMachineNode(Opcode, dl, VT: MVT::Other, Ops); |
553 | CurDAG->setNodeMemRefs(N: S, NewMemRefs: {MemOp}); |
554 | To[1] = SDValue(S, 0); |
555 | MachineSDNode *A = CurDAG->getMachineNode(Opcode: Hexagon::A2_addi, dl, VT: MVT::i32, |
556 | Op1: Base, Op2: IncV); |
557 | To[0] = SDValue(A, 0); |
558 | } |
559 | |
560 | ReplaceUses(F: From, T: To, Num: 2); |
561 | CurDAG->RemoveDeadNode(N: ST); |
562 | } |
563 | |
564 | void HexagonDAGToDAGISel::SelectStore(SDNode *N) { |
565 | SDLoc dl(N); |
566 | StoreSDNode *ST = cast<StoreSDNode>(Val: N); |
567 | |
568 | // Handle indexed stores. |
569 | ISD::MemIndexedMode AM = ST->getAddressingMode(); |
570 | if (AM != ISD::UNINDEXED) { |
571 | SelectIndexedStore(ST, dl); |
572 | return; |
573 | } |
574 | |
575 | SelectCode(N: ST); |
576 | } |
577 | |
578 | void HexagonDAGToDAGISel::SelectSHL(SDNode *N) { |
579 | SDLoc dl(N); |
580 | SDValue Shl_0 = N->getOperand(Num: 0); |
581 | SDValue Shl_1 = N->getOperand(Num: 1); |
582 | |
583 | auto Default = [this,N] () -> void { SelectCode(N); }; |
584 | |
585 | if (N->getValueType(ResNo: 0) != MVT::i32 || Shl_1.getOpcode() != ISD::Constant) |
586 | return Default(); |
587 | |
588 | // RHS is const. |
589 | int32_t ShlConst = cast<ConstantSDNode>(Val&: Shl_1)->getSExtValue(); |
590 | |
591 | if (Shl_0.getOpcode() == ISD::MUL) { |
592 | SDValue Mul_0 = Shl_0.getOperand(i: 0); // Val |
593 | SDValue Mul_1 = Shl_0.getOperand(i: 1); // Const |
594 | // RHS of mul is const. |
595 | if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Val&: Mul_1)) { |
596 | int32_t ValConst = C->getSExtValue() << ShlConst; |
597 | if (isInt<9>(x: ValConst)) { |
598 | SDValue Val = CurDAG->getTargetConstant(Val: ValConst, DL: dl, VT: MVT::i32); |
599 | SDNode *Result = CurDAG->getMachineNode(Opcode: Hexagon::M2_mpysmi, dl, |
600 | VT: MVT::i32, Op1: Mul_0, Op2: Val); |
601 | ReplaceNode(F: N, T: Result); |
602 | return; |
603 | } |
604 | } |
605 | return Default(); |
606 | } |
607 | |
608 | if (Shl_0.getOpcode() == ISD::SUB) { |
609 | SDValue Sub_0 = Shl_0.getOperand(i: 0); // Const 0 |
610 | SDValue Sub_1 = Shl_0.getOperand(i: 1); // Val |
611 | if (ConstantSDNode *C1 = dyn_cast<ConstantSDNode>(Val&: Sub_0)) { |
612 | if (C1->getSExtValue() != 0 || Sub_1.getOpcode() != ISD::SHL) |
613 | return Default(); |
614 | SDValue Shl2_0 = Sub_1.getOperand(i: 0); // Val |
615 | SDValue Shl2_1 = Sub_1.getOperand(i: 1); // Const |
616 | if (ConstantSDNode *C2 = dyn_cast<ConstantSDNode>(Val&: Shl2_1)) { |
617 | int32_t ValConst = 1 << (ShlConst + C2->getSExtValue()); |
618 | if (isInt<9>(x: -ValConst)) { |
619 | SDValue Val = CurDAG->getTargetConstant(Val: -ValConst, DL: dl, VT: MVT::i32); |
620 | SDNode *Result = CurDAG->getMachineNode(Opcode: Hexagon::M2_mpysmi, dl, |
621 | VT: MVT::i32, Op1: Shl2_0, Op2: Val); |
622 | ReplaceNode(F: N, T: Result); |
623 | return; |
624 | } |
625 | } |
626 | } |
627 | } |
628 | |
629 | return Default(); |
630 | } |
631 | |
632 | // |
633 | // Handling intrinsics for circular load and bitreverse load. |
634 | // |
635 | void HexagonDAGToDAGISel::SelectIntrinsicWChain(SDNode *N) { |
636 | if (MachineSDNode *L = LoadInstrForLoadIntrinsic(IntN: N)) { |
637 | StoreInstrForLoadIntrinsic(LoadN: L, IntN: N); |
638 | CurDAG->RemoveDeadNode(N); |
639 | return; |
640 | } |
641 | |
642 | // Handle bit-reverse load intrinsics. |
643 | if (SelectBrevLdIntrinsic(IntN: N)) |
644 | return; |
645 | |
646 | if (SelectNewCircIntrinsic(IntN: N)) |
647 | return; |
648 | |
649 | unsigned IntNo = N->getConstantOperandVal(Num: 1); |
650 | if (IntNo == Intrinsic::hexagon_V6_vgathermw || |
651 | IntNo == Intrinsic::hexagon_V6_vgathermw_128B || |
652 | IntNo == Intrinsic::hexagon_V6_vgathermh || |
653 | IntNo == Intrinsic::hexagon_V6_vgathermh_128B || |
654 | IntNo == Intrinsic::hexagon_V6_vgathermhw || |
655 | IntNo == Intrinsic::hexagon_V6_vgathermhw_128B) { |
656 | SelectV65Gather(N); |
657 | return; |
658 | } |
659 | if (IntNo == Intrinsic::hexagon_V6_vgathermwq || |
660 | IntNo == Intrinsic::hexagon_V6_vgathermwq_128B || |
661 | IntNo == Intrinsic::hexagon_V6_vgathermhq || |
662 | IntNo == Intrinsic::hexagon_V6_vgathermhq_128B || |
663 | IntNo == Intrinsic::hexagon_V6_vgathermhwq || |
664 | IntNo == Intrinsic::hexagon_V6_vgathermhwq_128B) { |
665 | SelectV65GatherPred(N); |
666 | return; |
667 | } |
668 | |
669 | SelectCode(N); |
670 | } |
671 | |
672 | void HexagonDAGToDAGISel::SelectIntrinsicWOChain(SDNode *N) { |
673 | unsigned IID = N->getConstantOperandVal(Num: 0); |
674 | unsigned Bits; |
675 | switch (IID) { |
676 | case Intrinsic::hexagon_S2_vsplatrb: |
677 | Bits = 8; |
678 | break; |
679 | case Intrinsic::hexagon_S2_vsplatrh: |
680 | Bits = 16; |
681 | break; |
682 | case Intrinsic::hexagon_V6_vaddcarry: |
683 | case Intrinsic::hexagon_V6_vaddcarry_128B: |
684 | case Intrinsic::hexagon_V6_vsubcarry: |
685 | case Intrinsic::hexagon_V6_vsubcarry_128B: |
686 | SelectHVXDualOutput(N); |
687 | return; |
688 | default: |
689 | SelectCode(N); |
690 | return; |
691 | } |
692 | |
693 | SDValue V = N->getOperand(Num: 1); |
694 | SDValue U; |
695 | // Splat intrinsics. |
696 | if (keepsLowBits(Val: V, NumBits: Bits, Src&: U)) { |
697 | SDValue R = CurDAG->getNode(Opcode: N->getOpcode(), DL: SDLoc(N), VT: N->getValueType(ResNo: 0), |
698 | N1: N->getOperand(Num: 0), N2: U); |
699 | ReplaceNode(F: N, T: R.getNode()); |
700 | SelectCode(N: R.getNode()); |
701 | return; |
702 | } |
703 | SelectCode(N); |
704 | } |
705 | |
706 | void HexagonDAGToDAGISel::(SDNode *N) { |
707 | SDValue Inp = N->getOperand(Num: 0); |
708 | MVT ResTy = N->getValueType(ResNo: 0).getSimpleVT(); |
709 | unsigned Idx = N->getConstantOperandVal(Num: 1); |
710 | |
711 | [[maybe_unused]] MVT InpTy = Inp.getValueType().getSimpleVT(); |
712 | [[maybe_unused]] unsigned ResLen = ResTy.getVectorNumElements(); |
713 | assert(InpTy.getVectorElementType() == ResTy.getVectorElementType()); |
714 | assert(2 * ResLen == InpTy.getVectorNumElements()); |
715 | assert(ResTy.getSizeInBits() == 32); |
716 | assert(Idx == 0 || Idx == ResLen); |
717 | |
718 | unsigned SubReg = Idx == 0 ? Hexagon::isub_lo : Hexagon::isub_hi; |
719 | SDValue Ext = CurDAG->getTargetExtractSubreg(SRIdx: SubReg, DL: SDLoc(N), VT: ResTy, Operand: Inp); |
720 | |
721 | ReplaceNode(F: N, T: Ext.getNode()); |
722 | } |
723 | |
724 | // |
725 | // Map floating point constant values. |
726 | // |
727 | void HexagonDAGToDAGISel::SelectConstantFP(SDNode *N) { |
728 | SDLoc dl(N); |
729 | auto *CN = cast<ConstantFPSDNode>(Val: N); |
730 | APInt A = CN->getValueAPF().bitcastToAPInt(); |
731 | if (N->getValueType(ResNo: 0) == MVT::f32) { |
732 | SDValue V = CurDAG->getTargetConstant(Val: A.getZExtValue(), DL: dl, VT: MVT::i32); |
733 | ReplaceNode(F: N, T: CurDAG->getMachineNode(Opcode: Hexagon::A2_tfrsi, dl, VT: MVT::f32, Op1: V)); |
734 | return; |
735 | } |
736 | if (N->getValueType(ResNo: 0) == MVT::f64) { |
737 | SDValue V = CurDAG->getTargetConstant(Val: A.getZExtValue(), DL: dl, VT: MVT::i64); |
738 | ReplaceNode(F: N, T: CurDAG->getMachineNode(Opcode: Hexagon::CONST64, dl, VT: MVT::f64, Op1: V)); |
739 | return; |
740 | } |
741 | |
742 | SelectCode(N); |
743 | } |
744 | |
745 | // |
746 | // Map boolean values. |
747 | // |
748 | void HexagonDAGToDAGISel::SelectConstant(SDNode *N) { |
749 | if (N->getValueType(ResNo: 0) == MVT::i1) { |
750 | assert(!(N->getAsZExtVal() >> 1)); |
751 | unsigned Opc = (cast<ConstantSDNode>(Val: N)->getSExtValue() != 0) |
752 | ? Hexagon::PS_true |
753 | : Hexagon::PS_false; |
754 | ReplaceNode(F: N, T: CurDAG->getMachineNode(Opcode: Opc, dl: SDLoc(N), VT: MVT::i1)); |
755 | return; |
756 | } |
757 | |
758 | SelectCode(N); |
759 | } |
760 | |
761 | void HexagonDAGToDAGISel::SelectFrameIndex(SDNode *N) { |
762 | MachineFrameInfo &MFI = MF->getFrameInfo(); |
763 | const HexagonFrameLowering *HFI = HST->getFrameLowering(); |
764 | int FX = cast<FrameIndexSDNode>(Val: N)->getIndex(); |
765 | Align StkA = HFI->getStackAlign(); |
766 | Align MaxA = MFI.getMaxAlign(); |
767 | SDValue FI = CurDAG->getTargetFrameIndex(FI: FX, VT: MVT::i32); |
768 | SDLoc DL(N); |
769 | SDValue Zero = CurDAG->getTargetConstant(Val: 0, DL, VT: MVT::i32); |
770 | SDNode *R = nullptr; |
771 | |
772 | // Use PS_fi when: |
773 | // - the object is fixed, or |
774 | // - there are no objects with higher-than-default alignment, or |
775 | // - there are no dynamically allocated objects. |
776 | // Otherwise, use PS_fia. |
777 | if (FX < 0 || MaxA <= StkA || !MFI.hasVarSizedObjects()) { |
778 | R = CurDAG->getMachineNode(Opcode: Hexagon::PS_fi, dl: DL, VT: MVT::i32, Op1: FI, Op2: Zero); |
779 | } else { |
780 | auto &HMFI = *MF->getInfo<HexagonMachineFunctionInfo>(); |
781 | Register AR = HMFI.getStackAlignBaseReg(); |
782 | SDValue CH = CurDAG->getEntryNode(); |
783 | SDValue Ops[] = { CurDAG->getCopyFromReg(Chain: CH, dl: DL, Reg: AR, VT: MVT::i32), FI, Zero }; |
784 | R = CurDAG->getMachineNode(Opcode: Hexagon::PS_fia, dl: DL, VT: MVT::i32, Ops); |
785 | } |
786 | |
787 | ReplaceNode(F: N, T: R); |
788 | } |
789 | |
790 | void HexagonDAGToDAGISel::SelectAddSubCarry(SDNode *N) { |
791 | unsigned OpcCarry = N->getOpcode() == HexagonISD::ADDC ? Hexagon::A4_addp_c |
792 | : Hexagon::A4_subp_c; |
793 | SDNode *C = CurDAG->getMachineNode(Opcode: OpcCarry, dl: SDLoc(N), VTs: N->getVTList(), |
794 | Ops: { N->getOperand(Num: 0), N->getOperand(Num: 1), |
795 | N->getOperand(Num: 2) }); |
796 | ReplaceNode(F: N, T: C); |
797 | } |
798 | |
799 | void HexagonDAGToDAGISel::SelectVAlign(SDNode *N) { |
800 | MVT ResTy = N->getValueType(ResNo: 0).getSimpleVT(); |
801 | if (HST->isHVXVectorType(VecTy: ResTy, IncludeBool: true)) |
802 | return SelectHvxVAlign(N); |
803 | |
804 | const SDLoc &dl(N); |
805 | unsigned VecLen = ResTy.getSizeInBits(); |
806 | if (VecLen == 32) { |
807 | SDValue Ops[] = { |
808 | CurDAG->getTargetConstant(Val: Hexagon::DoubleRegsRegClassID, DL: dl, VT: MVT::i32), |
809 | N->getOperand(Num: 0), |
810 | CurDAG->getTargetConstant(Val: Hexagon::isub_hi, DL: dl, VT: MVT::i32), |
811 | N->getOperand(Num: 1), |
812 | CurDAG->getTargetConstant(Val: Hexagon::isub_lo, DL: dl, VT: MVT::i32) |
813 | }; |
814 | SDNode *R = CurDAG->getMachineNode(Opcode: TargetOpcode::REG_SEQUENCE, dl, |
815 | VT: MVT::i64, Ops); |
816 | |
817 | // Shift right by "(Addr & 0x3) * 8" bytes. |
818 | SDNode *C; |
819 | SDValue M0 = CurDAG->getTargetConstant(Val: 0x18, DL: dl, VT: MVT::i32); |
820 | SDValue M1 = CurDAG->getTargetConstant(Val: 0x03, DL: dl, VT: MVT::i32); |
821 | if (HST->useCompound()) { |
822 | C = CurDAG->getMachineNode(Opcode: Hexagon::S4_andi_asl_ri, dl, VT: MVT::i32, |
823 | Op1: M0, Op2: N->getOperand(Num: 2), Op3: M1); |
824 | } else { |
825 | SDNode *T = CurDAG->getMachineNode(Opcode: Hexagon::S2_asl_i_r, dl, VT: MVT::i32, |
826 | Op1: N->getOperand(Num: 2), Op2: M1); |
827 | C = CurDAG->getMachineNode(Opcode: Hexagon::A2_andir, dl, VT: MVT::i32, |
828 | Op1: SDValue(T, 0), Op2: M0); |
829 | } |
830 | SDNode *S = CurDAG->getMachineNode(Opcode: Hexagon::S2_lsr_r_p, dl, VT: MVT::i64, |
831 | Op1: SDValue(R, 0), Op2: SDValue(C, 0)); |
832 | SDValue E = CurDAG->getTargetExtractSubreg(SRIdx: Hexagon::isub_lo, DL: dl, VT: ResTy, |
833 | Operand: SDValue(S, 0)); |
834 | ReplaceNode(F: N, T: E.getNode()); |
835 | } else { |
836 | assert(VecLen == 64); |
837 | SDNode *Pu = CurDAG->getMachineNode(Opcode: Hexagon::C2_tfrrp, dl, VT: MVT::v8i1, |
838 | Op1: N->getOperand(Num: 2)); |
839 | SDNode *VA = CurDAG->getMachineNode(Opcode: Hexagon::S2_valignrb, dl, VT: ResTy, |
840 | Op1: N->getOperand(Num: 0), Op2: N->getOperand(Num: 1), |
841 | Op3: SDValue(Pu,0)); |
842 | ReplaceNode(F: N, T: VA); |
843 | } |
844 | } |
845 | |
846 | void HexagonDAGToDAGISel::SelectVAlignAddr(SDNode *N) { |
847 | const SDLoc &dl(N); |
848 | SDValue A = N->getOperand(Num: 1); |
849 | int Mask = -cast<ConstantSDNode>(Val: A.getNode())->getSExtValue(); |
850 | assert(isPowerOf2_32(-Mask)); |
851 | |
852 | SDValue M = CurDAG->getTargetConstant(Val: Mask, DL: dl, VT: MVT::i32); |
853 | SDNode *AA = CurDAG->getMachineNode(Opcode: Hexagon::A2_andir, dl, VT: MVT::i32, |
854 | Op1: N->getOperand(Num: 0), Op2: M); |
855 | ReplaceNode(F: N, T: AA); |
856 | } |
857 | |
858 | // Handle these nodes here to avoid having to write patterns for all |
859 | // combinations of input/output types. In all cases, the resulting |
860 | // instruction is the same. |
861 | void HexagonDAGToDAGISel::SelectTypecast(SDNode *N) { |
862 | SDValue Op = N->getOperand(Num: 0); |
863 | MVT OpTy = Op.getValueType().getSimpleVT(); |
864 | SDNode *T = CurDAG->MorphNodeTo(N, Opc: N->getOpcode(), |
865 | VTs: CurDAG->getVTList(VT: OpTy), Ops: {Op}); |
866 | ReplaceNode(F: T, T: Op.getNode()); |
867 | } |
868 | |
869 | void HexagonDAGToDAGISel::SelectP2D(SDNode *N) { |
870 | MVT ResTy = N->getValueType(ResNo: 0).getSimpleVT(); |
871 | SDNode *T = CurDAG->getMachineNode(Opcode: Hexagon::C2_mask, dl: SDLoc(N), VT: ResTy, |
872 | Op1: N->getOperand(Num: 0)); |
873 | ReplaceNode(F: N, T); |
874 | } |
875 | |
876 | void HexagonDAGToDAGISel::SelectD2P(SDNode *N) { |
877 | const SDLoc &dl(N); |
878 | MVT ResTy = N->getValueType(ResNo: 0).getSimpleVT(); |
879 | SDValue Zero = CurDAG->getTargetConstant(Val: 0, DL: dl, VT: MVT::i32); |
880 | SDNode *T = CurDAG->getMachineNode(Opcode: Hexagon::A4_vcmpbgtui, dl, VT: ResTy, |
881 | Op1: N->getOperand(Num: 0), Op2: Zero); |
882 | ReplaceNode(F: N, T); |
883 | } |
884 | |
885 | void HexagonDAGToDAGISel::SelectV2Q(SDNode *N) { |
886 | const SDLoc &dl(N); |
887 | MVT ResTy = N->getValueType(ResNo: 0).getSimpleVT(); |
888 | // The argument to V2Q should be a single vector. |
889 | MVT OpTy = N->getOperand(Num: 0).getValueType().getSimpleVT(); (void)OpTy; |
890 | assert(HST->getVectorLength() * 8 == OpTy.getSizeInBits()); |
891 | |
892 | SDValue C = CurDAG->getTargetConstant(Val: -1, DL: dl, VT: MVT::i32); |
893 | SDNode *R = CurDAG->getMachineNode(Opcode: Hexagon::A2_tfrsi, dl, VT: MVT::i32, Op1: C); |
894 | SDNode *T = CurDAG->getMachineNode(Opcode: Hexagon::V6_vandvrt, dl, VT: ResTy, |
895 | Op1: N->getOperand(Num: 0), Op2: SDValue(R,0)); |
896 | ReplaceNode(F: N, T); |
897 | } |
898 | |
899 | void HexagonDAGToDAGISel::SelectQ2V(SDNode *N) { |
900 | const SDLoc &dl(N); |
901 | MVT ResTy = N->getValueType(ResNo: 0).getSimpleVT(); |
902 | // The result of V2Q should be a single vector. |
903 | assert(HST->getVectorLength() * 8 == ResTy.getSizeInBits()); |
904 | |
905 | SDValue C = CurDAG->getTargetConstant(Val: -1, DL: dl, VT: MVT::i32); |
906 | SDNode *R = CurDAG->getMachineNode(Opcode: Hexagon::A2_tfrsi, dl, VT: MVT::i32, Op1: C); |
907 | SDNode *T = CurDAG->getMachineNode(Opcode: Hexagon::V6_vandqrt, dl, VT: ResTy, |
908 | Op1: N->getOperand(Num: 0), Op2: SDValue(R,0)); |
909 | ReplaceNode(F: N, T); |
910 | } |
911 | |
912 | void HexagonDAGToDAGISel::FDiv(SDNode *N) { |
913 | const SDLoc &dl(N); |
914 | ArrayRef<EVT> ResultType(N->value_begin(), N->value_end()); |
915 | SmallVector<SDValue, 2> Ops; |
916 | Ops = {N->getOperand(Num: 0), N->getOperand(Num: 1)}; |
917 | SDVTList VTs; |
918 | VTs = CurDAG->getVTList(VT1: MVT::f32, VT2: MVT::f32); |
919 | SDNode *ResScale = CurDAG->getMachineNode(Opcode: Hexagon::F2_sfrecipa, dl, VTs, Ops); |
920 | SDNode *D = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffixupd, dl, VT: MVT::f32, Ops); |
921 | |
922 | SDValue C = CurDAG->getTargetConstant(Val: 0x3f800000, DL: dl, VT: MVT::i32); |
923 | SDNode *constNode = |
924 | CurDAG->getMachineNode(Opcode: Hexagon::A2_tfrsi, dl, VT: MVT::f32, Op1: C); |
925 | |
926 | SDNode *n = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffixupn, dl, VT: MVT::f32, Ops); |
927 | SDNode *Err = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffms_lib, dl, VT: MVT::f32, |
928 | Op1: SDValue(constNode, 0), Op2: SDValue(D, 0), |
929 | Op3: SDValue(ResScale, 0)); |
930 | SDNode *NewRec = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffma_lib, dl, VT: MVT::f32, |
931 | Op1: SDValue(ResScale, 0), Op2: SDValue(Err, 0), |
932 | Op3: SDValue(ResScale, 0)); |
933 | SDNode *newErr = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffms_lib, dl, VT: MVT::f32, |
934 | Op1: SDValue(constNode, 0), Op2: SDValue(D, 0), |
935 | Op3: SDValue(NewRec, 0)); |
936 | SDNode *q = CurDAG->getMachineNode( |
937 | Opcode: Hexagon::A2_andir, dl, VT: MVT::f32, Op1: SDValue(n, 0), |
938 | Op2: CurDAG->getTargetConstant(Val: 0x80000000, DL: dl, VT: MVT::i32)); |
939 | SDNode *NewQ = |
940 | CurDAG->getMachineNode(Opcode: Hexagon::F2_sffma_lib, dl, VT: MVT::f32, Op1: SDValue(q, 0), |
941 | Op2: SDValue(n, 0), Op3: SDValue(NewRec, 0)); |
942 | SDNode *NNewRec = CurDAG->getMachineNode( |
943 | Opcode: Hexagon::F2_sffma_lib, dl, VT: MVT::f32, Op1: SDValue(NewRec, 0), |
944 | Op2: SDValue(newErr, 0), Op3: SDValue(NewRec, 0)); |
945 | SDNode *qErr = |
946 | CurDAG->getMachineNode(Opcode: Hexagon::F2_sffms_lib, dl, VT: MVT::f32, Op1: SDValue(n, 0), |
947 | Op2: SDValue(D, 0), Op3: SDValue(NewQ, 0)); |
948 | SDNode *NNewQ = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffma_lib, dl, VT: MVT::f32, |
949 | Op1: SDValue(NewQ, 0), Op2: SDValue(qErr, 0), |
950 | Op3: SDValue(NNewRec, 0)); |
951 | |
952 | SDNode *NqErr = |
953 | CurDAG->getMachineNode(Opcode: Hexagon::F2_sffms_lib, dl, VT: MVT::f32, Op1: SDValue(n, 0), |
954 | Op2: SDValue(NNewQ, 0), Op3: SDValue(D, 0)); |
955 | std::array<SDValue, 4> temp1 = {SDValue(NNewQ, 0), SDValue(NqErr, 0), |
956 | SDValue(NNewRec, 0), SDValue(ResScale, 1)}; |
957 | ArrayRef<SDValue> OpValue1(temp1); |
958 | SDNode *FinalNewQ = |
959 | CurDAG->getMachineNode(Opcode: Hexagon::F2_sffma_sc, dl, VT: MVT::f32, Ops: OpValue1); |
960 | ReplaceNode(F: N, T: FinalNewQ); |
961 | } |
962 | |
963 | void HexagonDAGToDAGISel::FastFDiv(SDNode *N) { |
964 | const SDLoc &dl(N); |
965 | ArrayRef<EVT> ResultType(N->value_begin(), N->value_end()); |
966 | SmallVector<SDValue, 2> Ops; |
967 | Ops = {N->getOperand(Num: 0), N->getOperand(Num: 1)}; |
968 | SDVTList VTs; |
969 | VTs = CurDAG->getVTList(VT1: MVT::f32, VT2: MVT::f32); |
970 | SDNode *ResScale = CurDAG->getMachineNode(Opcode: Hexagon::F2_sfrecipa, dl, VTs, Ops); |
971 | SDNode *D = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffixupd, dl, VT: MVT::f32, Ops); |
972 | |
973 | SDValue C = CurDAG->getTargetConstant(Val: 0x3f800000, DL: dl, VT: MVT::i32); |
974 | SDNode *constNode = |
975 | CurDAG->getMachineNode(Opcode: Hexagon::A2_tfrsi, dl, VT: MVT::f32, Op1: C); |
976 | |
977 | SDNode *n = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffixupn, dl, VT: MVT::f32, Ops); |
978 | SDNode *Err = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffms_lib, dl, VT: MVT::f32, |
979 | Op1: SDValue(constNode, 0), Op2: SDValue(D, 0), |
980 | Op3: SDValue(ResScale, 0)); |
981 | SDNode *NewRec = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffma_lib, dl, VT: MVT::f32, |
982 | Op1: SDValue(ResScale, 0), Op2: SDValue(Err, 0), |
983 | Op3: SDValue(ResScale, 0)); |
984 | SDNode *newErr = CurDAG->getMachineNode(Opcode: Hexagon::F2_sffms_lib, dl, VT: MVT::f32, |
985 | Op1: SDValue(constNode, 0), Op2: SDValue(D, 0), |
986 | Op3: SDValue(NewRec, 0)); |
987 | |
988 | SDNode *NNewRec = CurDAG->getMachineNode( |
989 | Opcode: Hexagon::F2_sffma_lib, dl, VT: MVT::f32, Op1: SDValue(NewRec, 0), |
990 | Op2: SDValue(newErr, 0), Op3: SDValue(NewRec, 0)); |
991 | SDNode *FinalNewQ = CurDAG->getMachineNode( |
992 | Opcode: Hexagon::F2_sfmpy, dl, VT: MVT::f32, Op1: SDValue(NNewRec, 0), Op2: SDValue(n, 0)); |
993 | ReplaceNode(F: N, T: FinalNewQ); |
994 | } |
995 | |
996 | void HexagonDAGToDAGISel::SelectFDiv(SDNode *N) { |
997 | if (N->getFlags().hasAllowReassociation()) |
998 | FastFDiv(N); |
999 | else |
1000 | FDiv(N); |
1001 | return; |
1002 | } |
1003 | |
1004 | void HexagonDAGToDAGISel::Select(SDNode *N) { |
1005 | if (N->isMachineOpcode()) |
1006 | return N->setNodeId(-1); // Already selected. |
1007 | |
1008 | auto isHvxOp = [this](SDNode *N) { |
1009 | for (unsigned i = 0, e = N->getNumValues(); i != e; ++i) { |
1010 | if (HST->isHVXVectorType(VecTy: N->getValueType(ResNo: i), IncludeBool: true)) |
1011 | return true; |
1012 | } |
1013 | for (SDValue I : N->ops()) { |
1014 | if (HST->isHVXVectorType(VecTy: I.getValueType(), IncludeBool: true)) |
1015 | return true; |
1016 | } |
1017 | return false; |
1018 | }; |
1019 | |
1020 | if (HST->useHVXOps() && isHvxOp(N)) { |
1021 | switch (N->getOpcode()) { |
1022 | case ISD::EXTRACT_SUBVECTOR: return SelectHvxExtractSubvector(N); |
1023 | case ISD::VECTOR_SHUFFLE: return SelectHvxShuffle(N); |
1024 | |
1025 | case HexagonISD::VROR: return SelectHvxRor(N); |
1026 | } |
1027 | } |
1028 | |
1029 | switch (N->getOpcode()) { |
1030 | case ISD::Constant: return SelectConstant(N); |
1031 | case ISD::ConstantFP: return SelectConstantFP(N); |
1032 | case ISD::FrameIndex: return SelectFrameIndex(N); |
1033 | case ISD::SHL: return SelectSHL(N); |
1034 | case ISD::LOAD: return SelectLoad(N); |
1035 | case ISD::STORE: return SelectStore(N); |
1036 | case ISD::INTRINSIC_W_CHAIN: return SelectIntrinsicWChain(N); |
1037 | case ISD::INTRINSIC_WO_CHAIN: return SelectIntrinsicWOChain(N); |
1038 | case ISD::EXTRACT_SUBVECTOR: return SelectExtractSubvector(N); |
1039 | |
1040 | case HexagonISD::ADDC: |
1041 | case HexagonISD::SUBC: return SelectAddSubCarry(N); |
1042 | case HexagonISD::VALIGN: return SelectVAlign(N); |
1043 | case HexagonISD::VALIGNADDR: return SelectVAlignAddr(N); |
1044 | case HexagonISD::TYPECAST: return SelectTypecast(N); |
1045 | case HexagonISD::P2D: return SelectP2D(N); |
1046 | case HexagonISD::D2P: return SelectD2P(N); |
1047 | case HexagonISD::Q2V: return SelectQ2V(N); |
1048 | case HexagonISD::V2Q: return SelectV2Q(N); |
1049 | case ISD::FDIV: |
1050 | return SelectFDiv(N); |
1051 | } |
1052 | |
1053 | SelectCode(N); |
1054 | } |
1055 | |
1056 | bool HexagonDAGToDAGISel::SelectInlineAsmMemoryOperand( |
1057 | const SDValue &Op, InlineAsm::ConstraintCode ConstraintID, |
1058 | std::vector<SDValue> &OutOps) { |
1059 | SDValue Inp = Op, Res; |
1060 | |
1061 | switch (ConstraintID) { |
1062 | default: |
1063 | return true; |
1064 | case InlineAsm::ConstraintCode::o: // Offsetable. |
1065 | case InlineAsm::ConstraintCode::v: // Not offsetable. |
1066 | case InlineAsm::ConstraintCode::m: // Memory. |
1067 | if (SelectAddrFI(N&: Inp, R&: Res)) |
1068 | OutOps.push_back(x: Res); |
1069 | else |
1070 | OutOps.push_back(x: Inp); |
1071 | break; |
1072 | } |
1073 | |
1074 | OutOps.push_back(x: CurDAG->getTargetConstant(Val: 0, DL: SDLoc(Op), VT: MVT::i32)); |
1075 | return false; |
1076 | } |
1077 | |
1078 | static bool isMemOPCandidate(SDNode *I, SDNode *U) { |
1079 | // I is an operand of U. Check if U is an arithmetic (binary) operation |
1080 | // usable in a memop, where the other operand is a loaded value, and the |
1081 | // result of U is stored in the same location. |
1082 | |
1083 | if (!U->hasOneUse()) |
1084 | return false; |
1085 | unsigned Opc = U->getOpcode(); |
1086 | switch (Opc) { |
1087 | case ISD::ADD: |
1088 | case ISD::SUB: |
1089 | case ISD::AND: |
1090 | case ISD::OR: |
1091 | break; |
1092 | default: |
1093 | return false; |
1094 | } |
1095 | |
1096 | SDValue S0 = U->getOperand(Num: 0); |
1097 | SDValue S1 = U->getOperand(Num: 1); |
1098 | SDValue SY = (S0.getNode() == I) ? S1 : S0; |
1099 | |
1100 | SDNode *UUse = *U->use_begin(); |
1101 | if (UUse->getNumValues() != 1) |
1102 | return false; |
1103 | |
1104 | // Check if one of the inputs to U is a load instruction and the output |
1105 | // is used by a store instruction. If so and they also have the same |
1106 | // base pointer, then don't preoprocess this node sequence as it |
1107 | // can be matched to a memop. |
1108 | SDNode *SYNode = SY.getNode(); |
1109 | if (UUse->getOpcode() == ISD::STORE && SYNode->getOpcode() == ISD::LOAD) { |
1110 | SDValue LDBasePtr = cast<MemSDNode>(Val: SYNode)->getBasePtr(); |
1111 | SDValue STBasePtr = cast<MemSDNode>(Val: UUse)->getBasePtr(); |
1112 | if (LDBasePtr == STBasePtr) |
1113 | return true; |
1114 | } |
1115 | return false; |
1116 | } |
1117 | |
1118 | |
1119 | // Transform: (or (select c x 0) z) -> (select c (or x z) z) |
1120 | // (or (select c 0 y) z) -> (select c z (or y z)) |
1121 | void HexagonDAGToDAGISel::ppSimplifyOrSelect0(std::vector<SDNode*> &&Nodes) { |
1122 | SelectionDAG &DAG = *CurDAG; |
1123 | |
1124 | for (auto *I : Nodes) { |
1125 | if (I->getOpcode() != ISD::OR) |
1126 | continue; |
1127 | |
1128 | auto IsSelect0 = [](const SDValue &Op) -> bool { |
1129 | if (Op.getOpcode() != ISD::SELECT) |
1130 | return false; |
1131 | return isNullConstant(V: Op.getOperand(i: 1)) || |
1132 | isNullConstant(V: Op.getOperand(i: 2)); |
1133 | }; |
1134 | |
1135 | SDValue N0 = I->getOperand(Num: 0), N1 = I->getOperand(Num: 1); |
1136 | EVT VT = I->getValueType(ResNo: 0); |
1137 | bool SelN0 = IsSelect0(N0); |
1138 | SDValue SOp = SelN0 ? N0 : N1; |
1139 | SDValue VOp = SelN0 ? N1 : N0; |
1140 | |
1141 | if (SOp.getOpcode() == ISD::SELECT && SOp.getNode()->hasOneUse()) { |
1142 | SDValue SC = SOp.getOperand(i: 0); |
1143 | SDValue SX = SOp.getOperand(i: 1); |
1144 | SDValue SY = SOp.getOperand(i: 2); |
1145 | SDLoc DLS = SOp; |
1146 | if (isNullConstant(V: SY)) { |
1147 | SDValue NewOr = DAG.getNode(Opcode: ISD::OR, DL: DLS, VT, N1: SX, N2: VOp); |
1148 | SDValue NewSel = DAG.getNode(Opcode: ISD::SELECT, DL: DLS, VT, N1: SC, N2: NewOr, N3: VOp); |
1149 | DAG.ReplaceAllUsesWith(From: I, To: NewSel.getNode()); |
1150 | } else if (isNullConstant(V: SX)) { |
1151 | SDValue NewOr = DAG.getNode(Opcode: ISD::OR, DL: DLS, VT, N1: SY, N2: VOp); |
1152 | SDValue NewSel = DAG.getNode(Opcode: ISD::SELECT, DL: DLS, VT, N1: SC, N2: VOp, N3: NewOr); |
1153 | DAG.ReplaceAllUsesWith(From: I, To: NewSel.getNode()); |
1154 | } |
1155 | } |
1156 | } |
1157 | } |
1158 | |
1159 | // Transform: (store ch val (add x (add (shl y c) e))) |
1160 | // to: (store ch val (add x (shl (add y d) c))), |
1161 | // where e = (shl d c) for some integer d. |
1162 | // The purpose of this is to enable generation of loads/stores with |
1163 | // shifted addressing mode, i.e. mem(x+y<<#c). For that, the shift |
1164 | // value c must be 0, 1 or 2. |
1165 | void HexagonDAGToDAGISel::ppAddrReorderAddShl(std::vector<SDNode*> &&Nodes) { |
1166 | SelectionDAG &DAG = *CurDAG; |
1167 | |
1168 | for (auto *I : Nodes) { |
1169 | if (I->getOpcode() != ISD::STORE) |
1170 | continue; |
1171 | |
1172 | // I matched: (store ch val Off) |
1173 | SDValue Off = I->getOperand(Num: 2); |
1174 | // Off needs to match: (add x (add (shl y c) (shl d c)))) |
1175 | if (Off.getOpcode() != ISD::ADD) |
1176 | continue; |
1177 | // Off matched: (add x T0) |
1178 | SDValue T0 = Off.getOperand(i: 1); |
1179 | // T0 needs to match: (add T1 T2): |
1180 | if (T0.getOpcode() != ISD::ADD) |
1181 | continue; |
1182 | // T0 matched: (add T1 T2) |
1183 | SDValue T1 = T0.getOperand(i: 0); |
1184 | SDValue T2 = T0.getOperand(i: 1); |
1185 | // T1 needs to match: (shl y c) |
1186 | if (T1.getOpcode() != ISD::SHL) |
1187 | continue; |
1188 | SDValue C = T1.getOperand(i: 1); |
1189 | ConstantSDNode *CN = dyn_cast<ConstantSDNode>(Val: C.getNode()); |
1190 | if (CN == nullptr) |
1191 | continue; |
1192 | unsigned CV = CN->getZExtValue(); |
1193 | if (CV > 2) |
1194 | continue; |
1195 | // T2 needs to match e, where e = (shl d c) for some d. |
1196 | ConstantSDNode *EN = dyn_cast<ConstantSDNode>(Val: T2.getNode()); |
1197 | if (EN == nullptr) |
1198 | continue; |
1199 | unsigned EV = EN->getZExtValue(); |
1200 | if (EV % (1 << CV) != 0) |
1201 | continue; |
1202 | unsigned DV = EV / (1 << CV); |
1203 | |
1204 | // Replace T0 with: (shl (add y d) c) |
1205 | SDLoc DL = SDLoc(I); |
1206 | EVT VT = T0.getValueType(); |
1207 | SDValue D = DAG.getConstant(Val: DV, DL, VT); |
1208 | // NewAdd = (add y d) |
1209 | SDValue NewAdd = DAG.getNode(Opcode: ISD::ADD, DL, VT, N1: T1.getOperand(i: 0), N2: D); |
1210 | // NewShl = (shl NewAdd c) |
1211 | SDValue NewShl = DAG.getNode(Opcode: ISD::SHL, DL, VT, N1: NewAdd, N2: C); |
1212 | ReplaceNode(F: T0.getNode(), T: NewShl.getNode()); |
1213 | } |
1214 | } |
1215 | |
1216 | // Transform: (load ch (add x (and (srl y c) Mask))) |
1217 | // to: (load ch (add x (shl (srl y d) d-c))) |
1218 | // where |
1219 | // Mask = 00..0 111..1 0.0 |
1220 | // | | +-- d-c 0s, and d-c is 0, 1 or 2. |
1221 | // | +-------- 1s |
1222 | // +-------------- at most c 0s |
1223 | // Motivating example: |
1224 | // DAG combiner optimizes (add x (shl (srl y 5) 2)) |
1225 | // to (add x (and (srl y 3) 1FFFFFFC)) |
1226 | // which results in a constant-extended and(##...,lsr). This transformation |
1227 | // undoes this simplification for cases where the shl can be folded into |
1228 | // an addressing mode. |
1229 | void HexagonDAGToDAGISel::ppAddrRewriteAndSrl(std::vector<SDNode*> &&Nodes) { |
1230 | SelectionDAG &DAG = *CurDAG; |
1231 | |
1232 | for (SDNode *N : Nodes) { |
1233 | unsigned Opc = N->getOpcode(); |
1234 | if (Opc != ISD::LOAD && Opc != ISD::STORE) |
1235 | continue; |
1236 | SDValue Addr = Opc == ISD::LOAD ? N->getOperand(Num: 1) : N->getOperand(Num: 2); |
1237 | // Addr must match: (add x T0) |
1238 | if (Addr.getOpcode() != ISD::ADD) |
1239 | continue; |
1240 | SDValue T0 = Addr.getOperand(i: 1); |
1241 | // T0 must match: (and T1 Mask) |
1242 | if (T0.getOpcode() != ISD::AND) |
1243 | continue; |
1244 | |
1245 | // We have an AND. |
1246 | // |
1247 | // Check the first operand. It must be: (srl y c). |
1248 | SDValue S = T0.getOperand(i: 0); |
1249 | if (S.getOpcode() != ISD::SRL) |
1250 | continue; |
1251 | ConstantSDNode *SN = dyn_cast<ConstantSDNode>(Val: S.getOperand(i: 1).getNode()); |
1252 | if (SN == nullptr) |
1253 | continue; |
1254 | if (SN->getAPIntValue().getBitWidth() != 32) |
1255 | continue; |
1256 | uint32_t CV = SN->getZExtValue(); |
1257 | |
1258 | // Check the second operand: the supposed mask. |
1259 | ConstantSDNode *MN = dyn_cast<ConstantSDNode>(Val: T0.getOperand(i: 1).getNode()); |
1260 | if (MN == nullptr) |
1261 | continue; |
1262 | if (MN->getAPIntValue().getBitWidth() != 32) |
1263 | continue; |
1264 | uint32_t Mask = MN->getZExtValue(); |
1265 | // Examine the mask. |
1266 | uint32_t TZ = llvm::countr_zero(Val: Mask); |
1267 | uint32_t M1 = llvm::countr_one(Value: Mask >> TZ); |
1268 | uint32_t LZ = llvm::countl_zero(Val: Mask); |
1269 | // Trailing zeros + middle ones + leading zeros must equal the width. |
1270 | if (TZ + M1 + LZ != 32) |
1271 | continue; |
1272 | // The number of trailing zeros will be encoded in the addressing mode. |
1273 | if (TZ > 2) |
1274 | continue; |
1275 | // The number of leading zeros must be at most c. |
1276 | if (LZ > CV) |
1277 | continue; |
1278 | |
1279 | // All looks good. |
1280 | SDValue Y = S.getOperand(i: 0); |
1281 | EVT VT = Addr.getValueType(); |
1282 | SDLoc dl(S); |
1283 | // TZ = D-C, so D = TZ+C. |
1284 | SDValue D = DAG.getConstant(Val: TZ+CV, DL: dl, VT); |
1285 | SDValue DC = DAG.getConstant(Val: TZ, DL: dl, VT); |
1286 | SDValue NewSrl = DAG.getNode(Opcode: ISD::SRL, DL: dl, VT, N1: Y, N2: D); |
1287 | SDValue NewShl = DAG.getNode(Opcode: ISD::SHL, DL: dl, VT, N1: NewSrl, N2: DC); |
1288 | ReplaceNode(F: T0.getNode(), T: NewShl.getNode()); |
1289 | } |
1290 | } |
1291 | |
1292 | // Transform: (op ... (zext i1 c) ...) -> (select c (op ... 0 ...) |
1293 | // (op ... 1 ...)) |
1294 | void HexagonDAGToDAGISel::ppHoistZextI1(std::vector<SDNode*> &&Nodes) { |
1295 | SelectionDAG &DAG = *CurDAG; |
1296 | |
1297 | for (SDNode *N : Nodes) { |
1298 | unsigned Opc = N->getOpcode(); |
1299 | if (Opc != ISD::ZERO_EXTEND) |
1300 | continue; |
1301 | SDValue OpI1 = N->getOperand(Num: 0); |
1302 | EVT OpVT = OpI1.getValueType(); |
1303 | if (!OpVT.isSimple() || OpVT.getSimpleVT() != MVT::i1) |
1304 | continue; |
1305 | for (auto I = N->use_begin(), E = N->use_end(); I != E; ++I) { |
1306 | SDNode *U = *I; |
1307 | if (U->getNumValues() != 1) |
1308 | continue; |
1309 | EVT UVT = U->getValueType(ResNo: 0); |
1310 | if (!UVT.isSimple() || !UVT.isInteger() || UVT.getSimpleVT() == MVT::i1) |
1311 | continue; |
1312 | // Do not generate select for all i1 vector type. |
1313 | if (UVT.isVector() && UVT.getVectorElementType() == MVT::i1) |
1314 | continue; |
1315 | if (isMemOPCandidate(I: N, U)) |
1316 | continue; |
1317 | |
1318 | // Potentially simplifiable operation. |
1319 | unsigned I1N = I.getOperandNo(); |
1320 | SmallVector<SDValue,2> Ops(U->getNumOperands()); |
1321 | for (unsigned i = 0, n = U->getNumOperands(); i != n; ++i) |
1322 | Ops[i] = U->getOperand(Num: i); |
1323 | EVT BVT = Ops[I1N].getValueType(); |
1324 | |
1325 | const SDLoc &dl(U); |
1326 | SDValue C0 = DAG.getConstant(Val: 0, DL: dl, VT: BVT); |
1327 | SDValue C1 = DAG.getConstant(Val: 1, DL: dl, VT: BVT); |
1328 | SDValue If0, If1; |
1329 | |
1330 | if (isa<MachineSDNode>(Val: U)) { |
1331 | unsigned UseOpc = U->getMachineOpcode(); |
1332 | Ops[I1N] = C0; |
1333 | If0 = SDValue(DAG.getMachineNode(Opcode: UseOpc, dl, VT: UVT, Ops), 0); |
1334 | Ops[I1N] = C1; |
1335 | If1 = SDValue(DAG.getMachineNode(Opcode: UseOpc, dl, VT: UVT, Ops), 0); |
1336 | } else { |
1337 | unsigned UseOpc = U->getOpcode(); |
1338 | Ops[I1N] = C0; |
1339 | If0 = DAG.getNode(Opcode: UseOpc, DL: dl, VT: UVT, Ops); |
1340 | Ops[I1N] = C1; |
1341 | If1 = DAG.getNode(Opcode: UseOpc, DL: dl, VT: UVT, Ops); |
1342 | } |
1343 | // We're generating a SELECT way after legalization, so keep the types |
1344 | // simple. |
1345 | unsigned UW = UVT.getSizeInBits(); |
1346 | EVT SVT = (UW == 32 || UW == 64) ? MVT::getIntegerVT(BitWidth: UW) : UVT; |
1347 | SDValue Sel = DAG.getNode(Opcode: ISD::SELECT, DL: dl, VT: SVT, N1: OpI1, |
1348 | N2: DAG.getBitcast(VT: SVT, V: If1), |
1349 | N3: DAG.getBitcast(VT: SVT, V: If0)); |
1350 | SDValue Ret = DAG.getBitcast(VT: UVT, V: Sel); |
1351 | DAG.ReplaceAllUsesWith(From: U, To: Ret.getNode()); |
1352 | } |
1353 | } |
1354 | } |
1355 | |
1356 | void HexagonDAGToDAGISel::PreprocessISelDAG() { |
1357 | // Repack all nodes before calling each preprocessing function, |
1358 | // because each of them can modify the set of nodes. |
1359 | auto getNodes = [this]() -> std::vector<SDNode *> { |
1360 | std::vector<SDNode *> T; |
1361 | T.reserve(n: CurDAG->allnodes_size()); |
1362 | for (SDNode &N : CurDAG->allnodes()) |
1363 | T.push_back(x: &N); |
1364 | return T; |
1365 | }; |
1366 | |
1367 | if (HST->useHVXOps()) |
1368 | PreprocessHvxISelDAG(); |
1369 | |
1370 | // Transform: (or (select c x 0) z) -> (select c (or x z) z) |
1371 | // (or (select c 0 y) z) -> (select c z (or y z)) |
1372 | ppSimplifyOrSelect0(Nodes: getNodes()); |
1373 | |
1374 | // Transform: (store ch val (add x (add (shl y c) e))) |
1375 | // to: (store ch val (add x (shl (add y d) c))), |
1376 | // where e = (shl d c) for some integer d. |
1377 | // The purpose of this is to enable generation of loads/stores with |
1378 | // shifted addressing mode, i.e. mem(x+y<<#c). For that, the shift |
1379 | // value c must be 0, 1 or 2. |
1380 | ppAddrReorderAddShl(Nodes: getNodes()); |
1381 | |
1382 | // Transform: (load ch (add x (and (srl y c) Mask))) |
1383 | // to: (load ch (add x (shl (srl y d) d-c))) |
1384 | // where |
1385 | // Mask = 00..0 111..1 0.0 |
1386 | // | | +-- d-c 0s, and d-c is 0, 1 or 2. |
1387 | // | +-------- 1s |
1388 | // +-------------- at most c 0s |
1389 | // Motivating example: |
1390 | // DAG combiner optimizes (add x (shl (srl y 5) 2)) |
1391 | // to (add x (and (srl y 3) 1FFFFFFC)) |
1392 | // which results in a constant-extended and(##...,lsr). This transformation |
1393 | // undoes this simplification for cases where the shl can be folded into |
1394 | // an addressing mode. |
1395 | ppAddrRewriteAndSrl(Nodes: getNodes()); |
1396 | |
1397 | // Transform: (op ... (zext i1 c) ...) -> (select c (op ... 0 ...) |
1398 | // (op ... 1 ...)) |
1399 | ppHoistZextI1(Nodes: getNodes()); |
1400 | |
1401 | DEBUG_WITH_TYPE("isel" , { |
1402 | dbgs() << "Preprocessed (Hexagon) selection DAG:" ; |
1403 | CurDAG->dump(); |
1404 | }); |
1405 | |
1406 | if (EnableAddressRebalancing) { |
1407 | rebalanceAddressTrees(); |
1408 | |
1409 | DEBUG_WITH_TYPE("isel" , { |
1410 | dbgs() << "Address tree balanced selection DAG:" ; |
1411 | CurDAG->dump(); |
1412 | }); |
1413 | } |
1414 | } |
1415 | |
1416 | void HexagonDAGToDAGISel::emitFunctionEntryCode() { |
1417 | auto &HST = MF->getSubtarget<HexagonSubtarget>(); |
1418 | auto &HFI = *HST.getFrameLowering(); |
1419 | if (!HFI.needsAligna(MF: *MF)) |
1420 | return; |
1421 | |
1422 | MachineFrameInfo &MFI = MF->getFrameInfo(); |
1423 | MachineBasicBlock *EntryBB = &MF->front(); |
1424 | Align EntryMaxA = MFI.getMaxAlign(); |
1425 | |
1426 | // Reserve the first non-volatile register. |
1427 | Register AP = 0; |
1428 | auto &HRI = *HST.getRegisterInfo(); |
1429 | BitVector Reserved = HRI.getReservedRegs(MF: *MF); |
1430 | for (const MCPhysReg *R = HRI.getCalleeSavedRegs(MF); *R; ++R) { |
1431 | if (Reserved[*R]) |
1432 | continue; |
1433 | AP = *R; |
1434 | break; |
1435 | } |
1436 | assert(AP.isValid() && "Couldn't reserve stack align register" ); |
1437 | BuildMI(BB: EntryBB, MIMD: DebugLoc(), MCID: HII->get(Opcode: Hexagon::PS_aligna), DestReg: AP) |
1438 | .addImm(Val: EntryMaxA.value()); |
1439 | MF->getInfo<HexagonMachineFunctionInfo>()->setStackAlignBaseReg(AP); |
1440 | } |
1441 | |
1442 | void HexagonDAGToDAGISel::updateAligna() { |
1443 | auto &HFI = *MF->getSubtarget<HexagonSubtarget>().getFrameLowering(); |
1444 | if (!HFI.needsAligna(MF: *MF)) |
1445 | return; |
1446 | auto *AlignaI = const_cast<MachineInstr*>(HFI.getAlignaInstr(MF: *MF)); |
1447 | assert(AlignaI != nullptr); |
1448 | unsigned MaxA = MF->getFrameInfo().getMaxAlign().value(); |
1449 | if (AlignaI->getOperand(i: 1).getImm() < MaxA) |
1450 | AlignaI->getOperand(i: 1).setImm(MaxA); |
1451 | } |
1452 | |
1453 | // Match a frame index that can be used in an addressing mode. |
1454 | bool HexagonDAGToDAGISel::SelectAddrFI(SDValue &N, SDValue &R) { |
1455 | if (N.getOpcode() != ISD::FrameIndex) |
1456 | return false; |
1457 | auto &HFI = *HST->getFrameLowering(); |
1458 | MachineFrameInfo &MFI = MF->getFrameInfo(); |
1459 | int FX = cast<FrameIndexSDNode>(Val&: N)->getIndex(); |
1460 | if (!MFI.isFixedObjectIndex(ObjectIdx: FX) && HFI.needsAligna(MF: *MF)) |
1461 | return false; |
1462 | R = CurDAG->getTargetFrameIndex(FI: FX, VT: MVT::i32); |
1463 | return true; |
1464 | } |
1465 | |
1466 | inline bool HexagonDAGToDAGISel::SelectAddrGA(SDValue &N, SDValue &R) { |
1467 | return SelectGlobalAddress(N, R, UseGP: false, Alignment: Align(1)); |
1468 | } |
1469 | |
1470 | inline bool HexagonDAGToDAGISel::SelectAddrGP(SDValue &N, SDValue &R) { |
1471 | return SelectGlobalAddress(N, R, UseGP: true, Alignment: Align(1)); |
1472 | } |
1473 | |
1474 | inline bool HexagonDAGToDAGISel::SelectAnyImm(SDValue &N, SDValue &R) { |
1475 | return SelectAnyImmediate(N, R, Alignment: Align(1)); |
1476 | } |
1477 | |
1478 | inline bool HexagonDAGToDAGISel::SelectAnyImm0(SDValue &N, SDValue &R) { |
1479 | return SelectAnyImmediate(N, R, Alignment: Align(1)); |
1480 | } |
1481 | inline bool HexagonDAGToDAGISel::SelectAnyImm1(SDValue &N, SDValue &R) { |
1482 | return SelectAnyImmediate(N, R, Alignment: Align(2)); |
1483 | } |
1484 | inline bool HexagonDAGToDAGISel::SelectAnyImm2(SDValue &N, SDValue &R) { |
1485 | return SelectAnyImmediate(N, R, Alignment: Align(4)); |
1486 | } |
1487 | inline bool HexagonDAGToDAGISel::SelectAnyImm3(SDValue &N, SDValue &R) { |
1488 | return SelectAnyImmediate(N, R, Alignment: Align(8)); |
1489 | } |
1490 | |
1491 | inline bool HexagonDAGToDAGISel::SelectAnyInt(SDValue &N, SDValue &R) { |
1492 | EVT T = N.getValueType(); |
1493 | if (!T.isInteger() || T.getSizeInBits() != 32 || !isa<ConstantSDNode>(Val: N)) |
1494 | return false; |
1495 | int32_t V = cast<const ConstantSDNode>(Val&: N)->getZExtValue(); |
1496 | R = CurDAG->getTargetConstant(Val: V, DL: SDLoc(N), VT: N.getValueType()); |
1497 | return true; |
1498 | } |
1499 | |
1500 | bool HexagonDAGToDAGISel::SelectAnyImmediate(SDValue &N, SDValue &R, |
1501 | Align Alignment) { |
1502 | switch (N.getOpcode()) { |
1503 | case ISD::Constant: { |
1504 | if (N.getValueType() != MVT::i32) |
1505 | return false; |
1506 | int32_t V = cast<const ConstantSDNode>(Val&: N)->getZExtValue(); |
1507 | if (!isAligned(Lhs: Alignment, SizeInBytes: V)) |
1508 | return false; |
1509 | R = CurDAG->getTargetConstant(Val: V, DL: SDLoc(N), VT: N.getValueType()); |
1510 | return true; |
1511 | } |
1512 | case HexagonISD::JT: |
1513 | case HexagonISD::CP: |
1514 | // These are assumed to always be aligned at least 8-byte boundary. |
1515 | if (Alignment > Align(8)) |
1516 | return false; |
1517 | R = N.getOperand(i: 0); |
1518 | return true; |
1519 | case ISD::ExternalSymbol: |
1520 | // Symbols may be aligned at any boundary. |
1521 | if (Alignment > Align(1)) |
1522 | return false; |
1523 | R = N; |
1524 | return true; |
1525 | case ISD::BlockAddress: |
1526 | // Block address is always aligned at least 4-byte boundary. |
1527 | if (Alignment > Align(4) || |
1528 | !isAligned(Lhs: Alignment, SizeInBytes: cast<BlockAddressSDNode>(Val&: N)->getOffset())) |
1529 | return false; |
1530 | R = N; |
1531 | return true; |
1532 | } |
1533 | |
1534 | if (SelectGlobalAddress(N, R, UseGP: false, Alignment) || |
1535 | SelectGlobalAddress(N, R, UseGP: true, Alignment)) |
1536 | return true; |
1537 | |
1538 | return false; |
1539 | } |
1540 | |
1541 | bool HexagonDAGToDAGISel::SelectGlobalAddress(SDValue &N, SDValue &R, |
1542 | bool UseGP, Align Alignment) { |
1543 | switch (N.getOpcode()) { |
1544 | case ISD::ADD: { |
1545 | SDValue N0 = N.getOperand(i: 0); |
1546 | SDValue N1 = N.getOperand(i: 1); |
1547 | unsigned GAOpc = N0.getOpcode(); |
1548 | if (UseGP && GAOpc != HexagonISD::CONST32_GP) |
1549 | return false; |
1550 | if (!UseGP && GAOpc != HexagonISD::CONST32) |
1551 | return false; |
1552 | if (ConstantSDNode *Const = dyn_cast<ConstantSDNode>(Val&: N1)) { |
1553 | if (!isAligned(Lhs: Alignment, SizeInBytes: Const->getZExtValue())) |
1554 | return false; |
1555 | SDValue Addr = N0.getOperand(i: 0); |
1556 | if (GlobalAddressSDNode *GA = dyn_cast<GlobalAddressSDNode>(Val&: Addr)) { |
1557 | if (GA->getOpcode() == ISD::TargetGlobalAddress) { |
1558 | uint64_t NewOff = GA->getOffset() + (uint64_t)Const->getSExtValue(); |
1559 | R = CurDAG->getTargetGlobalAddress(GV: GA->getGlobal(), DL: SDLoc(Const), |
1560 | VT: N.getValueType(), offset: NewOff); |
1561 | return true; |
1562 | } |
1563 | } |
1564 | } |
1565 | break; |
1566 | } |
1567 | case HexagonISD::CP: |
1568 | case HexagonISD::JT: |
1569 | case HexagonISD::CONST32: |
1570 | // The operand(0) of CONST32 is TargetGlobalAddress, which is what we |
1571 | // want in the instruction. |
1572 | if (!UseGP) |
1573 | R = N.getOperand(i: 0); |
1574 | return !UseGP; |
1575 | case HexagonISD::CONST32_GP: |
1576 | if (UseGP) |
1577 | R = N.getOperand(i: 0); |
1578 | return UseGP; |
1579 | default: |
1580 | return false; |
1581 | } |
1582 | |
1583 | return false; |
1584 | } |
1585 | |
1586 | bool HexagonDAGToDAGISel::DetectUseSxtw(SDValue &N, SDValue &R) { |
1587 | // This (complex pattern) function is meant to detect a sign-extension |
1588 | // i32->i64 on a per-operand basis. This would allow writing single |
1589 | // patterns that would cover a number of combinations of different ways |
1590 | // a sign-extensions could be written. For example: |
1591 | // (mul (DetectUseSxtw x) (DetectUseSxtw y)) -> (M2_dpmpyss_s0 x y) |
1592 | // could match either one of these: |
1593 | // (mul (sext x) (sext_inreg y)) |
1594 | // (mul (sext-load *p) (sext_inreg y)) |
1595 | // (mul (sext_inreg x) (sext y)) |
1596 | // etc. |
1597 | // |
1598 | // The returned value will have type i64 and its low word will |
1599 | // contain the value being extended. The high bits are not specified. |
1600 | // The returned type is i64 because the original type of N was i64, |
1601 | // but the users of this function should only use the low-word of the |
1602 | // result, e.g. |
1603 | // (mul sxtw:x, sxtw:y) -> (M2_dpmpyss_s0 (LoReg sxtw:x), (LoReg sxtw:y)) |
1604 | |
1605 | if (N.getValueType() != MVT::i64) |
1606 | return false; |
1607 | unsigned Opc = N.getOpcode(); |
1608 | switch (Opc) { |
1609 | case ISD::SIGN_EXTEND: |
1610 | case ISD::SIGN_EXTEND_INREG: { |
1611 | // sext_inreg has the source type as a separate operand. |
1612 | EVT T = Opc == ISD::SIGN_EXTEND |
1613 | ? N.getOperand(i: 0).getValueType() |
1614 | : cast<VTSDNode>(Val: N.getOperand(i: 1))->getVT(); |
1615 | unsigned SW = T.getSizeInBits(); |
1616 | if (SW == 32) |
1617 | R = N.getOperand(i: 0); |
1618 | else if (SW < 32) |
1619 | R = N; |
1620 | else |
1621 | return false; |
1622 | break; |
1623 | } |
1624 | case ISD::LOAD: { |
1625 | LoadSDNode *L = cast<LoadSDNode>(Val&: N); |
1626 | if (L->getExtensionType() != ISD::SEXTLOAD) |
1627 | return false; |
1628 | // All extending loads extend to i32, so even if the value in |
1629 | // memory is shorter than 32 bits, it will be i32 after the load. |
1630 | if (L->getMemoryVT().getSizeInBits() > 32) |
1631 | return false; |
1632 | R = N; |
1633 | break; |
1634 | } |
1635 | case ISD::SRA: { |
1636 | auto *S = dyn_cast<ConstantSDNode>(Val: N.getOperand(i: 1)); |
1637 | if (!S || S->getZExtValue() != 32) |
1638 | return false; |
1639 | R = N; |
1640 | break; |
1641 | } |
1642 | default: |
1643 | return false; |
1644 | } |
1645 | EVT RT = R.getValueType(); |
1646 | if (RT == MVT::i64) |
1647 | return true; |
1648 | assert(RT == MVT::i32); |
1649 | // This is only to produce a value of type i64. Do not rely on the |
1650 | // high bits produced by this. |
1651 | const SDLoc &dl(N); |
1652 | SDValue Ops[] = { |
1653 | CurDAG->getTargetConstant(Val: Hexagon::DoubleRegsRegClassID, DL: dl, VT: MVT::i32), |
1654 | R, CurDAG->getTargetConstant(Val: Hexagon::isub_hi, DL: dl, VT: MVT::i32), |
1655 | R, CurDAG->getTargetConstant(Val: Hexagon::isub_lo, DL: dl, VT: MVT::i32) |
1656 | }; |
1657 | SDNode *T = CurDAG->getMachineNode(Opcode: TargetOpcode::REG_SEQUENCE, dl, |
1658 | VT: MVT::i64, Ops); |
1659 | R = SDValue(T, 0); |
1660 | return true; |
1661 | } |
1662 | |
1663 | bool HexagonDAGToDAGISel::keepsLowBits(const SDValue &Val, unsigned NumBits, |
1664 | SDValue &Src) { |
1665 | unsigned Opc = Val.getOpcode(); |
1666 | switch (Opc) { |
1667 | case ISD::SIGN_EXTEND: |
1668 | case ISD::ZERO_EXTEND: |
1669 | case ISD::ANY_EXTEND: { |
1670 | const SDValue &Op0 = Val.getOperand(i: 0); |
1671 | EVT T = Op0.getValueType(); |
1672 | if (T.isInteger() && T.getSizeInBits() == NumBits) { |
1673 | Src = Op0; |
1674 | return true; |
1675 | } |
1676 | break; |
1677 | } |
1678 | case ISD::SIGN_EXTEND_INREG: |
1679 | case ISD::AssertSext: |
1680 | case ISD::AssertZext: |
1681 | if (Val.getOperand(i: 0).getValueType().isInteger()) { |
1682 | VTSDNode *T = cast<VTSDNode>(Val: Val.getOperand(i: 1)); |
1683 | if (T->getVT().getSizeInBits() == NumBits) { |
1684 | Src = Val.getOperand(i: 0); |
1685 | return true; |
1686 | } |
1687 | } |
1688 | break; |
1689 | case ISD::AND: { |
1690 | // Check if this is an AND with NumBits of lower bits set to 1. |
1691 | uint64_t Mask = (1ULL << NumBits) - 1; |
1692 | if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Val: Val.getOperand(i: 0))) { |
1693 | if (C->getZExtValue() == Mask) { |
1694 | Src = Val.getOperand(i: 1); |
1695 | return true; |
1696 | } |
1697 | } |
1698 | if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Val: Val.getOperand(i: 1))) { |
1699 | if (C->getZExtValue() == Mask) { |
1700 | Src = Val.getOperand(i: 0); |
1701 | return true; |
1702 | } |
1703 | } |
1704 | break; |
1705 | } |
1706 | case ISD::OR: |
1707 | case ISD::XOR: { |
1708 | // OR/XOR with the lower NumBits bits set to 0. |
1709 | uint64_t Mask = (1ULL << NumBits) - 1; |
1710 | if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Val: Val.getOperand(i: 0))) { |
1711 | if ((C->getZExtValue() & Mask) == 0) { |
1712 | Src = Val.getOperand(i: 1); |
1713 | return true; |
1714 | } |
1715 | } |
1716 | if (ConstantSDNode *C = dyn_cast<ConstantSDNode>(Val: Val.getOperand(i: 1))) { |
1717 | if ((C->getZExtValue() & Mask) == 0) { |
1718 | Src = Val.getOperand(i: 0); |
1719 | return true; |
1720 | } |
1721 | } |
1722 | break; |
1723 | } |
1724 | default: |
1725 | break; |
1726 | } |
1727 | return false; |
1728 | } |
1729 | |
1730 | bool HexagonDAGToDAGISel::isAlignedMemNode(const MemSDNode *N) const { |
1731 | return N->getAlign().value() >= N->getMemoryVT().getStoreSize(); |
1732 | } |
1733 | |
1734 | bool HexagonDAGToDAGISel::isSmallStackStore(const StoreSDNode *N) const { |
1735 | unsigned StackSize = MF->getFrameInfo().estimateStackSize(MF: *MF); |
1736 | switch (N->getMemoryVT().getStoreSize()) { |
1737 | case 1: |
1738 | return StackSize <= 56; // 1*2^6 - 8 |
1739 | case 2: |
1740 | return StackSize <= 120; // 2*2^6 - 8 |
1741 | case 4: |
1742 | return StackSize <= 248; // 4*2^6 - 8 |
1743 | default: |
1744 | return false; |
1745 | } |
1746 | } |
1747 | |
1748 | // Return true when the given node fits in a positive half word. |
1749 | bool HexagonDAGToDAGISel::isPositiveHalfWord(const SDNode *N) const { |
1750 | if (const ConstantSDNode *CN = dyn_cast<const ConstantSDNode>(Val: N)) { |
1751 | int64_t V = CN->getSExtValue(); |
1752 | return V > 0 && isInt<16>(x: V); |
1753 | } |
1754 | if (N->getOpcode() == ISD::SIGN_EXTEND_INREG) { |
1755 | const VTSDNode *VN = dyn_cast<const VTSDNode>(Val: N->getOperand(Num: 1)); |
1756 | return VN->getVT().getSizeInBits() <= 16; |
1757 | } |
1758 | return false; |
1759 | } |
1760 | |
1761 | bool HexagonDAGToDAGISel::hasOneUse(const SDNode *N) const { |
1762 | return !CheckSingleUse || N->hasOneUse(); |
1763 | } |
1764 | |
1765 | //////////////////////////////////////////////////////////////////////////////// |
1766 | // Rebalancing of address calculation trees |
1767 | |
1768 | static bool isOpcodeHandled(const SDNode *N) { |
1769 | switch (N->getOpcode()) { |
1770 | case ISD::ADD: |
1771 | case ISD::MUL: |
1772 | return true; |
1773 | case ISD::SHL: |
1774 | // We only handle constant shifts because these can be easily flattened |
1775 | // into multiplications by 2^Op1. |
1776 | return isa<ConstantSDNode>(Val: N->getOperand(Num: 1).getNode()); |
1777 | default: |
1778 | return false; |
1779 | } |
1780 | } |
1781 | |
1782 | /// Return the weight of an SDNode |
1783 | int HexagonDAGToDAGISel::getWeight(SDNode *N) { |
1784 | if (!isOpcodeHandled(N)) |
1785 | return 1; |
1786 | assert(RootWeights.count(N) && "Cannot get weight of unseen root!" ); |
1787 | assert(RootWeights[N] != -1 && "Cannot get weight of unvisited root!" ); |
1788 | assert(RootWeights[N] != -2 && "Cannot get weight of RAWU'd root!" ); |
1789 | return RootWeights[N]; |
1790 | } |
1791 | |
1792 | int HexagonDAGToDAGISel::getHeight(SDNode *N) { |
1793 | if (!isOpcodeHandled(N)) |
1794 | return 0; |
1795 | assert(RootWeights.count(N) && RootWeights[N] >= 0 && |
1796 | "Cannot query height of unvisited/RAUW'd node!" ); |
1797 | return RootHeights[N]; |
1798 | } |
1799 | |
1800 | namespace { |
1801 | struct WeightedLeaf { |
1802 | SDValue Value; |
1803 | int Weight; |
1804 | int InsertionOrder; |
1805 | |
1806 | WeightedLeaf() {} |
1807 | |
1808 | WeightedLeaf(SDValue Value, int Weight, int InsertionOrder) : |
1809 | Value(Value), Weight(Weight), InsertionOrder(InsertionOrder) { |
1810 | assert(Weight >= 0 && "Weight must be >= 0" ); |
1811 | } |
1812 | |
1813 | static bool Compare(const WeightedLeaf &A, const WeightedLeaf &B) { |
1814 | assert(A.Value.getNode() && B.Value.getNode()); |
1815 | return A.Weight == B.Weight ? |
1816 | (A.InsertionOrder > B.InsertionOrder) : |
1817 | (A.Weight > B.Weight); |
1818 | } |
1819 | }; |
1820 | |
1821 | /// A specialized priority queue for WeigthedLeaves. It automatically folds |
1822 | /// constants and allows removal of non-top elements while maintaining the |
1823 | /// priority order. |
1824 | class LeafPrioQueue { |
1825 | SmallVector<WeightedLeaf, 8> Q; |
1826 | bool HaveConst; |
1827 | WeightedLeaf ConstElt; |
1828 | unsigned Opcode; |
1829 | |
1830 | public: |
1831 | bool empty() { |
1832 | return (!HaveConst && Q.empty()); |
1833 | } |
1834 | |
1835 | size_t size() { |
1836 | return Q.size() + HaveConst; |
1837 | } |
1838 | |
1839 | bool hasConst() { |
1840 | return HaveConst; |
1841 | } |
1842 | |
1843 | const WeightedLeaf &top() { |
1844 | if (HaveConst) |
1845 | return ConstElt; |
1846 | return Q.front(); |
1847 | } |
1848 | |
1849 | WeightedLeaf pop() { |
1850 | if (HaveConst) { |
1851 | HaveConst = false; |
1852 | return ConstElt; |
1853 | } |
1854 | std::pop_heap(first: Q.begin(), last: Q.end(), comp: WeightedLeaf::Compare); |
1855 | return Q.pop_back_val(); |
1856 | } |
1857 | |
1858 | void push(WeightedLeaf L, bool SeparateConst=true) { |
1859 | if (!HaveConst && SeparateConst && isa<ConstantSDNode>(Val: L.Value)) { |
1860 | if (Opcode == ISD::MUL && |
1861 | cast<ConstantSDNode>(Val&: L.Value)->getSExtValue() == 1) |
1862 | return; |
1863 | if (Opcode == ISD::ADD && |
1864 | cast<ConstantSDNode>(Val&: L.Value)->getSExtValue() == 0) |
1865 | return; |
1866 | |
1867 | HaveConst = true; |
1868 | ConstElt = L; |
1869 | } else { |
1870 | Q.push_back(Elt: L); |
1871 | std::push_heap(first: Q.begin(), last: Q.end(), comp: WeightedLeaf::Compare); |
1872 | } |
1873 | } |
1874 | |
1875 | /// Push L to the bottom of the queue regardless of its weight. If L is |
1876 | /// constant, it will not be folded with other constants in the queue. |
1877 | void pushToBottom(WeightedLeaf L) { |
1878 | L.Weight = 1000; |
1879 | push(L, SeparateConst: false); |
1880 | } |
1881 | |
1882 | /// Search for a SHL(x, [<=MaxAmount]) subtree in the queue, return the one of |
1883 | /// lowest weight and remove it from the queue. |
1884 | WeightedLeaf findSHL(uint64_t MaxAmount); |
1885 | |
1886 | WeightedLeaf findMULbyConst(); |
1887 | |
1888 | LeafPrioQueue(unsigned Opcode) : |
1889 | HaveConst(false), Opcode(Opcode) { } |
1890 | }; |
1891 | } // end anonymous namespace |
1892 | |
1893 | WeightedLeaf LeafPrioQueue::findSHL(uint64_t MaxAmount) { |
1894 | int ResultPos; |
1895 | WeightedLeaf Result; |
1896 | |
1897 | for (int Pos = 0, End = Q.size(); Pos != End; ++Pos) { |
1898 | const WeightedLeaf &L = Q[Pos]; |
1899 | const SDValue &Val = L.Value; |
1900 | if (Val.getOpcode() != ISD::SHL || |
1901 | !isa<ConstantSDNode>(Val: Val.getOperand(i: 1)) || |
1902 | Val.getConstantOperandVal(i: 1) > MaxAmount) |
1903 | continue; |
1904 | if (!Result.Value.getNode() || Result.Weight > L.Weight || |
1905 | (Result.Weight == L.Weight && Result.InsertionOrder > L.InsertionOrder)) |
1906 | { |
1907 | Result = L; |
1908 | ResultPos = Pos; |
1909 | } |
1910 | } |
1911 | |
1912 | if (Result.Value.getNode()) { |
1913 | Q.erase(CI: &Q[ResultPos]); |
1914 | std::make_heap(first: Q.begin(), last: Q.end(), comp: WeightedLeaf::Compare); |
1915 | } |
1916 | |
1917 | return Result; |
1918 | } |
1919 | |
1920 | WeightedLeaf LeafPrioQueue::findMULbyConst() { |
1921 | int ResultPos; |
1922 | WeightedLeaf Result; |
1923 | |
1924 | for (int Pos = 0, End = Q.size(); Pos != End; ++Pos) { |
1925 | const WeightedLeaf &L = Q[Pos]; |
1926 | const SDValue &Val = L.Value; |
1927 | if (Val.getOpcode() != ISD::MUL || |
1928 | !isa<ConstantSDNode>(Val: Val.getOperand(i: 1)) || |
1929 | Val.getConstantOperandVal(i: 1) > 127) |
1930 | continue; |
1931 | if (!Result.Value.getNode() || Result.Weight > L.Weight || |
1932 | (Result.Weight == L.Weight && Result.InsertionOrder > L.InsertionOrder)) |
1933 | { |
1934 | Result = L; |
1935 | ResultPos = Pos; |
1936 | } |
1937 | } |
1938 | |
1939 | if (Result.Value.getNode()) { |
1940 | Q.erase(CI: &Q[ResultPos]); |
1941 | std::make_heap(first: Q.begin(), last: Q.end(), comp: WeightedLeaf::Compare); |
1942 | } |
1943 | |
1944 | return Result; |
1945 | } |
1946 | |
1947 | SDValue HexagonDAGToDAGISel::getMultiplierForSHL(SDNode *N) { |
1948 | uint64_t MulFactor = 1ull << N->getConstantOperandVal(Num: 1); |
1949 | return CurDAG->getConstant(Val: MulFactor, DL: SDLoc(N), |
1950 | VT: N->getOperand(Num: 1).getValueType()); |
1951 | } |
1952 | |
1953 | /// @returns the value x for which 2^x is a factor of Val |
1954 | static unsigned getPowerOf2Factor(SDValue Val) { |
1955 | if (Val.getOpcode() == ISD::MUL) { |
1956 | unsigned MaxFactor = 0; |
1957 | for (int i = 0; i < 2; ++i) { |
1958 | ConstantSDNode *C = dyn_cast<ConstantSDNode>(Val: Val.getOperand(i)); |
1959 | if (!C) |
1960 | continue; |
1961 | const APInt &CInt = C->getAPIntValue(); |
1962 | if (CInt.getBoolValue()) |
1963 | MaxFactor = CInt.countr_zero(); |
1964 | } |
1965 | return MaxFactor; |
1966 | } |
1967 | if (Val.getOpcode() == ISD::SHL) { |
1968 | if (!isa<ConstantSDNode>(Val: Val.getOperand(i: 1).getNode())) |
1969 | return 0; |
1970 | return (unsigned) Val.getConstantOperandVal(i: 1); |
1971 | } |
1972 | |
1973 | return 0; |
1974 | } |
1975 | |
1976 | /// @returns true if V>>Amount will eliminate V's operation on its child |
1977 | static bool willShiftRightEliminate(SDValue V, unsigned Amount) { |
1978 | if (V.getOpcode() == ISD::MUL) { |
1979 | SDValue Ops[] = { V.getOperand(i: 0), V.getOperand(i: 1) }; |
1980 | for (int i = 0; i < 2; ++i) |
1981 | if (isa<ConstantSDNode>(Val: Ops[i].getNode()) && |
1982 | V.getConstantOperandVal(i) % (1ULL << Amount) == 0) { |
1983 | uint64_t NewConst = V.getConstantOperandVal(i) >> Amount; |
1984 | return (NewConst == 1); |
1985 | } |
1986 | } else if (V.getOpcode() == ISD::SHL) { |
1987 | return (Amount == V.getConstantOperandVal(i: 1)); |
1988 | } |
1989 | |
1990 | return false; |
1991 | } |
1992 | |
1993 | SDValue HexagonDAGToDAGISel::factorOutPowerOf2(SDValue V, unsigned Power) { |
1994 | SDValue Ops[] = { V.getOperand(i: 0), V.getOperand(i: 1) }; |
1995 | if (V.getOpcode() == ISD::MUL) { |
1996 | for (int i=0; i < 2; ++i) { |
1997 | if (isa<ConstantSDNode>(Val: Ops[i].getNode()) && |
1998 | V.getConstantOperandVal(i) % ((uint64_t)1 << Power) == 0) { |
1999 | uint64_t NewConst = V.getConstantOperandVal(i) >> Power; |
2000 | if (NewConst == 1) |
2001 | return Ops[!i]; |
2002 | Ops[i] = CurDAG->getConstant(Val: NewConst, |
2003 | DL: SDLoc(V), VT: V.getValueType()); |
2004 | break; |
2005 | } |
2006 | } |
2007 | } else if (V.getOpcode() == ISD::SHL) { |
2008 | uint64_t ShiftAmount = V.getConstantOperandVal(i: 1); |
2009 | if (ShiftAmount == Power) |
2010 | return Ops[0]; |
2011 | Ops[1] = CurDAG->getConstant(Val: ShiftAmount - Power, |
2012 | DL: SDLoc(V), VT: V.getValueType()); |
2013 | } |
2014 | |
2015 | return CurDAG->getNode(Opcode: V.getOpcode(), DL: SDLoc(V), VT: V.getValueType(), Ops); |
2016 | } |
2017 | |
2018 | static bool isTargetConstant(const SDValue &V) { |
2019 | return V.getOpcode() == HexagonISD::CONST32 || |
2020 | V.getOpcode() == HexagonISD::CONST32_GP; |
2021 | } |
2022 | |
2023 | unsigned HexagonDAGToDAGISel::getUsesInFunction(const Value *V) { |
2024 | if (GAUsesInFunction.count(Val: V)) |
2025 | return GAUsesInFunction[V]; |
2026 | |
2027 | unsigned Result = 0; |
2028 | const Function &CurF = CurDAG->getMachineFunction().getFunction(); |
2029 | for (const User *U : V->users()) { |
2030 | if (isa<Instruction>(Val: U) && |
2031 | cast<Instruction>(Val: U)->getParent()->getParent() == &CurF) |
2032 | ++Result; |
2033 | } |
2034 | |
2035 | GAUsesInFunction[V] = Result; |
2036 | |
2037 | return Result; |
2038 | } |
2039 | |
2040 | /// Note - After calling this, N may be dead. It may have been replaced by a |
2041 | /// new node, so always use the returned value in place of N. |
2042 | /// |
2043 | /// @returns The SDValue taking the place of N (which could be N if it is |
2044 | /// unchanged) |
2045 | SDValue HexagonDAGToDAGISel::balanceSubTree(SDNode *N, bool TopLevel) { |
2046 | assert(RootWeights.count(N) && "Cannot balance non-root node." ); |
2047 | assert(RootWeights[N] != -2 && "This node was RAUW'd!" ); |
2048 | assert(!TopLevel || N->getOpcode() == ISD::ADD); |
2049 | |
2050 | // Return early if this node was already visited |
2051 | if (RootWeights[N] != -1) |
2052 | return SDValue(N, 0); |
2053 | |
2054 | assert(isOpcodeHandled(N)); |
2055 | |
2056 | SDValue Op0 = N->getOperand(Num: 0); |
2057 | SDValue Op1 = N->getOperand(Num: 1); |
2058 | |
2059 | // Return early if the operands will remain unchanged or are all roots |
2060 | if ((!isOpcodeHandled(N: Op0.getNode()) || RootWeights.count(Val: Op0.getNode())) && |
2061 | (!isOpcodeHandled(N: Op1.getNode()) || RootWeights.count(Val: Op1.getNode()))) { |
2062 | SDNode *Op0N = Op0.getNode(); |
2063 | int Weight; |
2064 | if (isOpcodeHandled(N: Op0N) && RootWeights[Op0N] == -1) { |
2065 | Weight = getWeight(N: balanceSubTree(N: Op0N).getNode()); |
2066 | // Weight = calculateWeight(Op0N); |
2067 | } else |
2068 | Weight = getWeight(N: Op0N); |
2069 | |
2070 | SDNode *Op1N = N->getOperand(Num: 1).getNode(); // Op1 may have been RAUWd |
2071 | if (isOpcodeHandled(N: Op1N) && RootWeights[Op1N] == -1) { |
2072 | Weight += getWeight(N: balanceSubTree(N: Op1N).getNode()); |
2073 | // Weight += calculateWeight(Op1N); |
2074 | } else |
2075 | Weight += getWeight(N: Op1N); |
2076 | |
2077 | RootWeights[N] = Weight; |
2078 | RootHeights[N] = std::max(a: getHeight(N: N->getOperand(Num: 0).getNode()), |
2079 | b: getHeight(N: N->getOperand(Num: 1).getNode())) + 1; |
2080 | |
2081 | LLVM_DEBUG(dbgs() << "--> No need to balance root (Weight=" << Weight |
2082 | << " Height=" << RootHeights[N] << "): " ); |
2083 | LLVM_DEBUG(N->dump(CurDAG)); |
2084 | |
2085 | return SDValue(N, 0); |
2086 | } |
2087 | |
2088 | LLVM_DEBUG(dbgs() << "** Balancing root node: " ); |
2089 | LLVM_DEBUG(N->dump(CurDAG)); |
2090 | |
2091 | unsigned NOpcode = N->getOpcode(); |
2092 | |
2093 | LeafPrioQueue Leaves(NOpcode); |
2094 | SmallVector<SDValue, 4> Worklist; |
2095 | Worklist.push_back(Elt: SDValue(N, 0)); |
2096 | |
2097 | // SHL nodes will be converted to MUL nodes |
2098 | if (NOpcode == ISD::SHL) |
2099 | NOpcode = ISD::MUL; |
2100 | |
2101 | bool CanFactorize = false; |
2102 | WeightedLeaf Mul1, Mul2; |
2103 | unsigned MaxPowerOf2 = 0; |
2104 | WeightedLeaf GA; |
2105 | |
2106 | // Do not try to factor out a shift if there is already a shift at the tip of |
2107 | // the tree. |
2108 | bool HaveTopLevelShift = false; |
2109 | if (TopLevel && |
2110 | ((isOpcodeHandled(N: Op0.getNode()) && Op0.getOpcode() == ISD::SHL && |
2111 | Op0.getConstantOperandVal(i: 1) < 4) || |
2112 | (isOpcodeHandled(N: Op1.getNode()) && Op1.getOpcode() == ISD::SHL && |
2113 | Op1.getConstantOperandVal(i: 1) < 4))) |
2114 | HaveTopLevelShift = true; |
2115 | |
2116 | // Flatten the subtree into an ordered list of leaves; at the same time |
2117 | // determine whether the tree is already balanced. |
2118 | int InsertionOrder = 0; |
2119 | SmallDenseMap<SDValue, int> NodeHeights; |
2120 | bool Imbalanced = false; |
2121 | int CurrentWeight = 0; |
2122 | while (!Worklist.empty()) { |
2123 | SDValue Child = Worklist.pop_back_val(); |
2124 | |
2125 | if (Child.getNode() != N && RootWeights.count(Val: Child.getNode())) { |
2126 | // CASE 1: Child is a root note |
2127 | |
2128 | int Weight = RootWeights[Child.getNode()]; |
2129 | if (Weight == -1) { |
2130 | Child = balanceSubTree(N: Child.getNode()); |
2131 | // calculateWeight(Child.getNode()); |
2132 | Weight = getWeight(N: Child.getNode()); |
2133 | } else if (Weight == -2) { |
2134 | // Whoops, this node was RAUWd by one of the balanceSubTree calls we |
2135 | // made. Our worklist isn't up to date anymore. |
2136 | // Restart the whole process. |
2137 | LLVM_DEBUG(dbgs() << "--> Subtree was RAUWd. Restarting...\n" ); |
2138 | return balanceSubTree(N, TopLevel); |
2139 | } |
2140 | |
2141 | NodeHeights[Child] = 1; |
2142 | CurrentWeight += Weight; |
2143 | |
2144 | unsigned PowerOf2; |
2145 | if (TopLevel && !CanFactorize && !HaveTopLevelShift && |
2146 | (Child.getOpcode() == ISD::MUL || Child.getOpcode() == ISD::SHL) && |
2147 | Child.hasOneUse() && (PowerOf2 = getPowerOf2Factor(Val: Child))) { |
2148 | // Try to identify two factorizable MUL/SHL children greedily. Leave |
2149 | // them out of the priority queue for now so we can deal with them |
2150 | // after. |
2151 | if (!Mul1.Value.getNode()) { |
2152 | Mul1 = WeightedLeaf(Child, Weight, InsertionOrder++); |
2153 | MaxPowerOf2 = PowerOf2; |
2154 | } else { |
2155 | Mul2 = WeightedLeaf(Child, Weight, InsertionOrder++); |
2156 | MaxPowerOf2 = std::min(a: MaxPowerOf2, b: PowerOf2); |
2157 | |
2158 | // Our addressing modes can only shift by a maximum of 3 |
2159 | if (MaxPowerOf2 > 3) |
2160 | MaxPowerOf2 = 3; |
2161 | |
2162 | CanFactorize = true; |
2163 | } |
2164 | } else |
2165 | Leaves.push(L: WeightedLeaf(Child, Weight, InsertionOrder++)); |
2166 | } else if (!isOpcodeHandled(N: Child.getNode())) { |
2167 | // CASE 2: Child is an unhandled kind of node (e.g. constant) |
2168 | int Weight = getWeight(N: Child.getNode()); |
2169 | |
2170 | NodeHeights[Child] = getHeight(N: Child.getNode()); |
2171 | CurrentWeight += Weight; |
2172 | |
2173 | if (isTargetConstant(V: Child) && !GA.Value.getNode()) |
2174 | GA = WeightedLeaf(Child, Weight, InsertionOrder++); |
2175 | else |
2176 | Leaves.push(L: WeightedLeaf(Child, Weight, InsertionOrder++)); |
2177 | } else { |
2178 | // CASE 3: Child is a subtree of same opcode |
2179 | // Visit children first, then flatten. |
2180 | unsigned ChildOpcode = Child.getOpcode(); |
2181 | assert(ChildOpcode == NOpcode || |
2182 | (NOpcode == ISD::MUL && ChildOpcode == ISD::SHL)); |
2183 | |
2184 | // Convert SHL to MUL |
2185 | SDValue Op1; |
2186 | if (ChildOpcode == ISD::SHL) |
2187 | Op1 = getMultiplierForSHL(N: Child.getNode()); |
2188 | else |
2189 | Op1 = Child->getOperand(Num: 1); |
2190 | |
2191 | if (!NodeHeights.count(Val: Op1) || !NodeHeights.count(Val: Child->getOperand(Num: 0))) { |
2192 | assert(!NodeHeights.count(Child) && "Parent visited before children?" ); |
2193 | // Visit children first, then re-visit this node |
2194 | Worklist.push_back(Elt: Child); |
2195 | Worklist.push_back(Elt: Op1); |
2196 | Worklist.push_back(Elt: Child->getOperand(Num: 0)); |
2197 | } else { |
2198 | // Back at this node after visiting the children |
2199 | if (std::abs(x: NodeHeights[Op1] - NodeHeights[Child->getOperand(Num: 0)]) > 1) |
2200 | Imbalanced = true; |
2201 | |
2202 | NodeHeights[Child] = std::max(a: NodeHeights[Op1], |
2203 | b: NodeHeights[Child->getOperand(Num: 0)]) + 1; |
2204 | } |
2205 | } |
2206 | } |
2207 | |
2208 | LLVM_DEBUG(dbgs() << "--> Current height=" << NodeHeights[SDValue(N, 0)] |
2209 | << " weight=" << CurrentWeight |
2210 | << " imbalanced=" << Imbalanced << "\n" ); |
2211 | |
2212 | // Transform MUL(x, C * 2^Y) + SHL(z, Y) -> SHL(ADD(MUL(x, C), z), Y) |
2213 | // This factors out a shift in order to match memw(a<<Y+b). |
2214 | if (CanFactorize && (willShiftRightEliminate(V: Mul1.Value, Amount: MaxPowerOf2) || |
2215 | willShiftRightEliminate(V: Mul2.Value, Amount: MaxPowerOf2))) { |
2216 | LLVM_DEBUG(dbgs() << "--> Found common factor for two MUL children!\n" ); |
2217 | int Weight = Mul1.Weight + Mul2.Weight; |
2218 | int Height = std::max(a: NodeHeights[Mul1.Value], b: NodeHeights[Mul2.Value]) + 1; |
2219 | SDValue Mul1Factored = factorOutPowerOf2(V: Mul1.Value, Power: MaxPowerOf2); |
2220 | SDValue Mul2Factored = factorOutPowerOf2(V: Mul2.Value, Power: MaxPowerOf2); |
2221 | SDValue Sum = CurDAG->getNode(Opcode: ISD::ADD, DL: SDLoc(N), VT: Mul1.Value.getValueType(), |
2222 | N1: Mul1Factored, N2: Mul2Factored); |
2223 | SDValue Const = CurDAG->getConstant(Val: MaxPowerOf2, DL: SDLoc(N), |
2224 | VT: Mul1.Value.getValueType()); |
2225 | SDValue New = CurDAG->getNode(Opcode: ISD::SHL, DL: SDLoc(N), VT: Mul1.Value.getValueType(), |
2226 | N1: Sum, N2: Const); |
2227 | NodeHeights[New] = Height; |
2228 | Leaves.push(L: WeightedLeaf(New, Weight, Mul1.InsertionOrder)); |
2229 | } else if (Mul1.Value.getNode()) { |
2230 | // We failed to factorize two MULs, so now the Muls are left outside the |
2231 | // queue... add them back. |
2232 | Leaves.push(L: Mul1); |
2233 | if (Mul2.Value.getNode()) |
2234 | Leaves.push(L: Mul2); |
2235 | CanFactorize = false; |
2236 | } |
2237 | |
2238 | // Combine GA + Constant -> GA+Offset, but only if GA is not used elsewhere |
2239 | // and the root node itself is not used more than twice. This reduces the |
2240 | // amount of additional constant extenders introduced by this optimization. |
2241 | bool CombinedGA = false; |
2242 | if (NOpcode == ISD::ADD && GA.Value.getNode() && Leaves.hasConst() && |
2243 | GA.Value.hasOneUse() && N->use_size() < 3) { |
2244 | GlobalAddressSDNode *GANode = |
2245 | cast<GlobalAddressSDNode>(Val: GA.Value.getOperand(i: 0)); |
2246 | ConstantSDNode *Offset = cast<ConstantSDNode>(Val: Leaves.top().Value); |
2247 | |
2248 | if (getUsesInFunction(V: GANode->getGlobal()) == 1 && Offset->hasOneUse() && |
2249 | getTargetLowering()->isOffsetFoldingLegal(GA: GANode)) { |
2250 | LLVM_DEBUG(dbgs() << "--> Combining GA and offset (" |
2251 | << Offset->getSExtValue() << "): " ); |
2252 | LLVM_DEBUG(GANode->dump(CurDAG)); |
2253 | |
2254 | SDValue NewTGA = |
2255 | CurDAG->getTargetGlobalAddress(GV: GANode->getGlobal(), DL: SDLoc(GA.Value), |
2256 | VT: GANode->getValueType(ResNo: 0), |
2257 | offset: GANode->getOffset() + (uint64_t)Offset->getSExtValue()); |
2258 | GA.Value = CurDAG->getNode(Opcode: GA.Value.getOpcode(), DL: SDLoc(GA.Value), |
2259 | VT: GA.Value.getValueType(), Operand: NewTGA); |
2260 | GA.Weight += Leaves.top().Weight; |
2261 | |
2262 | NodeHeights[GA.Value] = getHeight(N: GA.Value.getNode()); |
2263 | CombinedGA = true; |
2264 | |
2265 | Leaves.pop(); // Remove the offset constant from the queue |
2266 | } |
2267 | } |
2268 | |
2269 | if ((RebalanceOnlyForOptimizations && !CanFactorize && !CombinedGA) || |
2270 | (RebalanceOnlyImbalancedTrees && !Imbalanced)) { |
2271 | RootWeights[N] = CurrentWeight; |
2272 | RootHeights[N] = NodeHeights[SDValue(N, 0)]; |
2273 | |
2274 | return SDValue(N, 0); |
2275 | } |
2276 | |
2277 | // Combine GA + SHL(x, C<=31) so we will match Rx=add(#u8,asl(Rx,#U5)) |
2278 | if (NOpcode == ISD::ADD && GA.Value.getNode()) { |
2279 | WeightedLeaf SHL = Leaves.findSHL(MaxAmount: 31); |
2280 | if (SHL.Value.getNode()) { |
2281 | int Height = std::max(a: NodeHeights[GA.Value], b: NodeHeights[SHL.Value]) + 1; |
2282 | GA.Value = CurDAG->getNode(Opcode: ISD::ADD, DL: SDLoc(GA.Value), |
2283 | VT: GA.Value.getValueType(), |
2284 | N1: GA.Value, N2: SHL.Value); |
2285 | GA.Weight = SHL.Weight; // Specifically ignore the GA weight here |
2286 | NodeHeights[GA.Value] = Height; |
2287 | } |
2288 | } |
2289 | |
2290 | if (GA.Value.getNode()) |
2291 | Leaves.push(L: GA); |
2292 | |
2293 | // If this is the top level and we haven't factored out a shift, we should try |
2294 | // to move a constant to the bottom to match addressing modes like memw(rX+C) |
2295 | if (TopLevel && !CanFactorize && Leaves.hasConst()) { |
2296 | LLVM_DEBUG(dbgs() << "--> Pushing constant to tip of tree." ); |
2297 | Leaves.pushToBottom(L: Leaves.pop()); |
2298 | } |
2299 | |
2300 | const DataLayout &DL = CurDAG->getDataLayout(); |
2301 | const TargetLowering &TLI = *getTargetLowering(); |
2302 | |
2303 | // Rebuild the tree using Huffman's algorithm |
2304 | while (Leaves.size() > 1) { |
2305 | WeightedLeaf L0 = Leaves.pop(); |
2306 | |
2307 | // See whether we can grab a MUL to form an add(Rx,mpyi(Ry,#u6)), |
2308 | // otherwise just get the next leaf |
2309 | WeightedLeaf L1 = Leaves.findMULbyConst(); |
2310 | if (!L1.Value.getNode()) |
2311 | L1 = Leaves.pop(); |
2312 | |
2313 | assert(L0.Weight <= L1.Weight && "Priority queue is broken!" ); |
2314 | |
2315 | SDValue V0 = L0.Value; |
2316 | int V0Weight = L0.Weight; |
2317 | SDValue V1 = L1.Value; |
2318 | int V1Weight = L1.Weight; |
2319 | |
2320 | // Make sure that none of these nodes have been RAUW'd |
2321 | if ((RootWeights.count(Val: V0.getNode()) && RootWeights[V0.getNode()] == -2) || |
2322 | (RootWeights.count(Val: V1.getNode()) && RootWeights[V1.getNode()] == -2)) { |
2323 | LLVM_DEBUG(dbgs() << "--> Subtree was RAUWd. Restarting...\n" ); |
2324 | return balanceSubTree(N, TopLevel); |
2325 | } |
2326 | |
2327 | ConstantSDNode *V0C = dyn_cast<ConstantSDNode>(Val&: V0); |
2328 | ConstantSDNode *V1C = dyn_cast<ConstantSDNode>(Val&: V1); |
2329 | EVT VT = N->getValueType(ResNo: 0); |
2330 | SDValue NewNode; |
2331 | |
2332 | if (V0C && !V1C) { |
2333 | std::swap(a&: V0, b&: V1); |
2334 | std::swap(a&: V0C, b&: V1C); |
2335 | } |
2336 | |
2337 | // Calculate height of this node |
2338 | assert(NodeHeights.count(V0) && NodeHeights.count(V1) && |
2339 | "Children must have been visited before re-combining them!" ); |
2340 | int Height = std::max(a: NodeHeights[V0], b: NodeHeights[V1]) + 1; |
2341 | |
2342 | // Rebuild this node (and restore SHL from MUL if needed) |
2343 | if (V1C && NOpcode == ISD::MUL && V1C->getAPIntValue().isPowerOf2()) |
2344 | NewNode = CurDAG->getNode( |
2345 | Opcode: ISD::SHL, DL: SDLoc(V0), VT, N1: V0, |
2346 | N2: CurDAG->getConstant( |
2347 | Val: V1C->getAPIntValue().logBase2(), DL: SDLoc(N), |
2348 | VT: TLI.getScalarShiftAmountTy(DL, V0.getValueType()))); |
2349 | else |
2350 | NewNode = CurDAG->getNode(Opcode: NOpcode, DL: SDLoc(N), VT, N1: V0, N2: V1); |
2351 | |
2352 | NodeHeights[NewNode] = Height; |
2353 | |
2354 | int Weight = V0Weight + V1Weight; |
2355 | Leaves.push(L: WeightedLeaf(NewNode, Weight, L0.InsertionOrder)); |
2356 | |
2357 | LLVM_DEBUG(dbgs() << "--> Built new node (Weight=" << Weight |
2358 | << ",Height=" << Height << "):\n" ); |
2359 | LLVM_DEBUG(NewNode.dump()); |
2360 | } |
2361 | |
2362 | assert(Leaves.size() == 1); |
2363 | SDValue NewRoot = Leaves.top().Value; |
2364 | |
2365 | assert(NodeHeights.count(NewRoot)); |
2366 | int Height = NodeHeights[NewRoot]; |
2367 | |
2368 | // Restore SHL if we earlier converted it to a MUL |
2369 | if (NewRoot.getOpcode() == ISD::MUL) { |
2370 | ConstantSDNode *V1C = dyn_cast<ConstantSDNode>(Val: NewRoot.getOperand(i: 1)); |
2371 | if (V1C && V1C->getAPIntValue().isPowerOf2()) { |
2372 | EVT VT = NewRoot.getValueType(); |
2373 | SDValue V0 = NewRoot.getOperand(i: 0); |
2374 | NewRoot = CurDAG->getNode( |
2375 | Opcode: ISD::SHL, DL: SDLoc(NewRoot), VT, N1: V0, |
2376 | N2: CurDAG->getConstant( |
2377 | Val: V1C->getAPIntValue().logBase2(), DL: SDLoc(NewRoot), |
2378 | VT: TLI.getScalarShiftAmountTy(DL, V0.getValueType()))); |
2379 | } |
2380 | } |
2381 | |
2382 | if (N != NewRoot.getNode()) { |
2383 | LLVM_DEBUG(dbgs() << "--> Root is now: " ); |
2384 | LLVM_DEBUG(NewRoot.dump()); |
2385 | |
2386 | // Replace all uses of old root by new root |
2387 | CurDAG->ReplaceAllUsesWith(From: N, To: NewRoot.getNode()); |
2388 | // Mark that we have RAUW'd N |
2389 | RootWeights[N] = -2; |
2390 | } else { |
2391 | LLVM_DEBUG(dbgs() << "--> Root unchanged.\n" ); |
2392 | } |
2393 | |
2394 | RootWeights[NewRoot.getNode()] = Leaves.top().Weight; |
2395 | RootHeights[NewRoot.getNode()] = Height; |
2396 | |
2397 | return NewRoot; |
2398 | } |
2399 | |
2400 | void HexagonDAGToDAGISel::rebalanceAddressTrees() { |
2401 | for (SDNode &Node : llvm::make_early_inc_range(Range: CurDAG->allnodes())) { |
2402 | SDNode *N = &Node; |
2403 | if (N->getOpcode() != ISD::LOAD && N->getOpcode() != ISD::STORE) |
2404 | continue; |
2405 | |
2406 | SDValue BasePtr = cast<MemSDNode>(Val: N)->getBasePtr(); |
2407 | if (BasePtr.getOpcode() != ISD::ADD) |
2408 | continue; |
2409 | |
2410 | // We've already processed this node |
2411 | if (RootWeights.count(Val: BasePtr.getNode())) |
2412 | continue; |
2413 | |
2414 | LLVM_DEBUG(dbgs() << "** Rebalancing address calculation in node: " ); |
2415 | LLVM_DEBUG(N->dump(CurDAG)); |
2416 | |
2417 | // FindRoots |
2418 | SmallVector<SDNode *, 4> Worklist; |
2419 | |
2420 | Worklist.push_back(Elt: BasePtr.getOperand(i: 0).getNode()); |
2421 | Worklist.push_back(Elt: BasePtr.getOperand(i: 1).getNode()); |
2422 | |
2423 | while (!Worklist.empty()) { |
2424 | SDNode *N = Worklist.pop_back_val(); |
2425 | unsigned Opcode = N->getOpcode(); |
2426 | |
2427 | if (!isOpcodeHandled(N)) |
2428 | continue; |
2429 | |
2430 | Worklist.push_back(Elt: N->getOperand(Num: 0).getNode()); |
2431 | Worklist.push_back(Elt: N->getOperand(Num: 1).getNode()); |
2432 | |
2433 | // Not a root if it has only one use and same opcode as its parent |
2434 | if (N->hasOneUse() && Opcode == N->use_begin()->getOpcode()) |
2435 | continue; |
2436 | |
2437 | // This root node has already been processed |
2438 | if (RootWeights.count(Val: N)) |
2439 | continue; |
2440 | |
2441 | RootWeights[N] = -1; |
2442 | } |
2443 | |
2444 | // Balance node itself |
2445 | RootWeights[BasePtr.getNode()] = -1; |
2446 | SDValue NewBasePtr = balanceSubTree(N: BasePtr.getNode(), /*TopLevel=*/ true); |
2447 | |
2448 | if (N->getOpcode() == ISD::LOAD) |
2449 | N = CurDAG->UpdateNodeOperands(N, Op1: N->getOperand(Num: 0), |
2450 | Op2: NewBasePtr, Op3: N->getOperand(Num: 2)); |
2451 | else |
2452 | N = CurDAG->UpdateNodeOperands(N, Op1: N->getOperand(Num: 0), Op2: N->getOperand(Num: 1), |
2453 | Op3: NewBasePtr, Op4: N->getOperand(Num: 3)); |
2454 | |
2455 | LLVM_DEBUG(dbgs() << "--> Final node: " ); |
2456 | LLVM_DEBUG(N->dump(CurDAG)); |
2457 | } |
2458 | |
2459 | CurDAG->RemoveDeadNodes(); |
2460 | GAUsesInFunction.clear(); |
2461 | RootHeights.clear(); |
2462 | RootWeights.clear(); |
2463 | } |
2464 | |