1 | //===-- llvm/CodeGen/GlobalISel/LegalizerHelper.cpp -----------------------===// |
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 | /// \file This file implements the LegalizerHelper class to legalize |
10 | /// individual instructions and the LegalizeMachineIR wrapper pass for the |
11 | /// primary legalization. |
12 | // |
13 | //===----------------------------------------------------------------------===// |
14 | |
15 | #include "llvm/CodeGen/GlobalISel/LegalizerHelper.h" |
16 | #include "llvm/CodeGen/GlobalISel/CallLowering.h" |
17 | #include "llvm/CodeGen/GlobalISel/GISelChangeObserver.h" |
18 | #include "llvm/CodeGen/GlobalISel/GISelKnownBits.h" |
19 | #include "llvm/CodeGen/GlobalISel/GenericMachineInstrs.h" |
20 | #include "llvm/CodeGen/GlobalISel/LegalizerInfo.h" |
21 | #include "llvm/CodeGen/GlobalISel/LostDebugLocObserver.h" |
22 | #include "llvm/CodeGen/GlobalISel/MIPatternMatch.h" |
23 | #include "llvm/CodeGen/GlobalISel/MachineIRBuilder.h" |
24 | #include "llvm/CodeGen/GlobalISel/Utils.h" |
25 | #include "llvm/CodeGen/MachineConstantPool.h" |
26 | #include "llvm/CodeGen/MachineFrameInfo.h" |
27 | #include "llvm/CodeGen/MachineRegisterInfo.h" |
28 | #include "llvm/CodeGen/RuntimeLibcallUtil.h" |
29 | #include "llvm/CodeGen/TargetFrameLowering.h" |
30 | #include "llvm/CodeGen/TargetInstrInfo.h" |
31 | #include "llvm/CodeGen/TargetLowering.h" |
32 | #include "llvm/CodeGen/TargetOpcodes.h" |
33 | #include "llvm/CodeGen/TargetSubtargetInfo.h" |
34 | #include "llvm/IR/Instructions.h" |
35 | #include "llvm/Support/Debug.h" |
36 | #include "llvm/Support/MathExtras.h" |
37 | #include "llvm/Support/raw_ostream.h" |
38 | #include "llvm/Target/TargetMachine.h" |
39 | #include <numeric> |
40 | #include <optional> |
41 | |
42 | #define DEBUG_TYPE "legalizer" |
43 | |
44 | using namespace llvm; |
45 | using namespace LegalizeActions; |
46 | using namespace MIPatternMatch; |
47 | |
48 | /// Try to break down \p OrigTy into \p NarrowTy sized pieces. |
49 | /// |
50 | /// Returns the number of \p NarrowTy elements needed to reconstruct \p OrigTy, |
51 | /// with any leftover piece as type \p LeftoverTy |
52 | /// |
53 | /// Returns -1 in the first element of the pair if the breakdown is not |
54 | /// satisfiable. |
55 | static std::pair<int, int> |
56 | getNarrowTypeBreakDown(LLT OrigTy, LLT NarrowTy, LLT &LeftoverTy) { |
57 | assert(!LeftoverTy.isValid() && "this is an out argument" ); |
58 | |
59 | unsigned Size = OrigTy.getSizeInBits(); |
60 | unsigned NarrowSize = NarrowTy.getSizeInBits(); |
61 | unsigned NumParts = Size / NarrowSize; |
62 | unsigned LeftoverSize = Size - NumParts * NarrowSize; |
63 | assert(Size > NarrowSize); |
64 | |
65 | if (LeftoverSize == 0) |
66 | return {NumParts, 0}; |
67 | |
68 | if (NarrowTy.isVector()) { |
69 | unsigned EltSize = OrigTy.getScalarSizeInBits(); |
70 | if (LeftoverSize % EltSize != 0) |
71 | return {-1, -1}; |
72 | LeftoverTy = |
73 | LLT::scalarOrVector(EC: ElementCount::getFixed(MinVal: LeftoverSize / EltSize), |
74 | ScalarTy: OrigTy.getElementType()); |
75 | } else { |
76 | LeftoverTy = LLT::scalar(SizeInBits: LeftoverSize); |
77 | } |
78 | |
79 | int NumLeftover = LeftoverSize / LeftoverTy.getSizeInBits(); |
80 | return std::make_pair(x&: NumParts, y&: NumLeftover); |
81 | } |
82 | |
83 | static Type *getFloatTypeForLLT(LLVMContext &Ctx, LLT Ty) { |
84 | |
85 | if (!Ty.isScalar()) |
86 | return nullptr; |
87 | |
88 | switch (Ty.getSizeInBits()) { |
89 | case 16: |
90 | return Type::getHalfTy(C&: Ctx); |
91 | case 32: |
92 | return Type::getFloatTy(C&: Ctx); |
93 | case 64: |
94 | return Type::getDoubleTy(C&: Ctx); |
95 | case 80: |
96 | return Type::getX86_FP80Ty(C&: Ctx); |
97 | case 128: |
98 | return Type::getFP128Ty(C&: Ctx); |
99 | default: |
100 | return nullptr; |
101 | } |
102 | } |
103 | |
104 | LegalizerHelper::LegalizerHelper(MachineFunction &MF, |
105 | GISelChangeObserver &Observer, |
106 | MachineIRBuilder &Builder) |
107 | : MIRBuilder(Builder), Observer(Observer), MRI(MF.getRegInfo()), |
108 | LI(*MF.getSubtarget().getLegalizerInfo()), |
109 | TLI(*MF.getSubtarget().getTargetLowering()), KB(nullptr) {} |
110 | |
111 | LegalizerHelper::LegalizerHelper(MachineFunction &MF, const LegalizerInfo &LI, |
112 | GISelChangeObserver &Observer, |
113 | MachineIRBuilder &B, GISelKnownBits *KB) |
114 | : MIRBuilder(B), Observer(Observer), MRI(MF.getRegInfo()), LI(LI), |
115 | TLI(*MF.getSubtarget().getTargetLowering()), KB(KB) {} |
116 | |
117 | LegalizerHelper::LegalizeResult |
118 | LegalizerHelper::legalizeInstrStep(MachineInstr &MI, |
119 | LostDebugLocObserver &LocObserver) { |
120 | LLVM_DEBUG(dbgs() << "Legalizing: " << MI); |
121 | |
122 | MIRBuilder.setInstrAndDebugLoc(MI); |
123 | |
124 | if (isa<GIntrinsic>(Val: MI)) |
125 | return LI.legalizeIntrinsic(Helper&: *this, MI) ? Legalized : UnableToLegalize; |
126 | auto Step = LI.getAction(MI, MRI); |
127 | switch (Step.Action) { |
128 | case Legal: |
129 | LLVM_DEBUG(dbgs() << ".. Already legal\n" ); |
130 | return AlreadyLegal; |
131 | case Libcall: |
132 | LLVM_DEBUG(dbgs() << ".. Convert to libcall\n" ); |
133 | return libcall(MI, LocObserver); |
134 | case NarrowScalar: |
135 | LLVM_DEBUG(dbgs() << ".. Narrow scalar\n" ); |
136 | return narrowScalar(MI, TypeIdx: Step.TypeIdx, NarrowTy: Step.NewType); |
137 | case WidenScalar: |
138 | LLVM_DEBUG(dbgs() << ".. Widen scalar\n" ); |
139 | return widenScalar(MI, TypeIdx: Step.TypeIdx, WideTy: Step.NewType); |
140 | case Bitcast: |
141 | LLVM_DEBUG(dbgs() << ".. Bitcast type\n" ); |
142 | return bitcast(MI, TypeIdx: Step.TypeIdx, Ty: Step.NewType); |
143 | case Lower: |
144 | LLVM_DEBUG(dbgs() << ".. Lower\n" ); |
145 | return lower(MI, TypeIdx: Step.TypeIdx, Ty: Step.NewType); |
146 | case FewerElements: |
147 | LLVM_DEBUG(dbgs() << ".. Reduce number of elements\n" ); |
148 | return fewerElementsVector(MI, TypeIdx: Step.TypeIdx, NarrowTy: Step.NewType); |
149 | case MoreElements: |
150 | LLVM_DEBUG(dbgs() << ".. Increase number of elements\n" ); |
151 | return moreElementsVector(MI, TypeIdx: Step.TypeIdx, MoreTy: Step.NewType); |
152 | case Custom: |
153 | LLVM_DEBUG(dbgs() << ".. Custom legalization\n" ); |
154 | return LI.legalizeCustom(Helper&: *this, MI, LocObserver) ? Legalized |
155 | : UnableToLegalize; |
156 | default: |
157 | LLVM_DEBUG(dbgs() << ".. Unable to legalize\n" ); |
158 | return UnableToLegalize; |
159 | } |
160 | } |
161 | |
162 | void LegalizerHelper::insertParts(Register DstReg, |
163 | LLT ResultTy, LLT PartTy, |
164 | ArrayRef<Register> PartRegs, |
165 | LLT LeftoverTy, |
166 | ArrayRef<Register> LeftoverRegs) { |
167 | if (!LeftoverTy.isValid()) { |
168 | assert(LeftoverRegs.empty()); |
169 | |
170 | if (!ResultTy.isVector()) { |
171 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: PartRegs); |
172 | return; |
173 | } |
174 | |
175 | if (PartTy.isVector()) |
176 | MIRBuilder.buildConcatVectors(Res: DstReg, Ops: PartRegs); |
177 | else |
178 | MIRBuilder.buildBuildVector(Res: DstReg, Ops: PartRegs); |
179 | return; |
180 | } |
181 | |
182 | // Merge sub-vectors with different number of elements and insert into DstReg. |
183 | if (ResultTy.isVector()) { |
184 | assert(LeftoverRegs.size() == 1 && "Expected one leftover register" ); |
185 | SmallVector<Register, 8> AllRegs; |
186 | for (auto Reg : concat<const Register>(Ranges&: PartRegs, Ranges&: LeftoverRegs)) |
187 | AllRegs.push_back(Elt: Reg); |
188 | return mergeMixedSubvectors(DstReg, PartRegs: AllRegs); |
189 | } |
190 | |
191 | SmallVector<Register> GCDRegs; |
192 | LLT GCDTy = getGCDType(OrigTy: getGCDType(OrigTy: ResultTy, TargetTy: LeftoverTy), TargetTy: PartTy); |
193 | for (auto PartReg : concat<const Register>(Ranges&: PartRegs, Ranges&: LeftoverRegs)) |
194 | extractGCDType(Parts&: GCDRegs, GCDTy, SrcReg: PartReg); |
195 | LLT ResultLCMTy = buildLCMMergePieces(DstTy: ResultTy, NarrowTy: LeftoverTy, GCDTy, VRegs&: GCDRegs); |
196 | buildWidenedRemergeToDst(DstReg, LCMTy: ResultLCMTy, RemergeRegs: GCDRegs); |
197 | } |
198 | |
199 | void LegalizerHelper::appendVectorElts(SmallVectorImpl<Register> &Elts, |
200 | Register Reg) { |
201 | LLT Ty = MRI.getType(Reg); |
202 | SmallVector<Register, 8> RegElts; |
203 | extractParts(Reg, Ty: Ty.getScalarType(), NumParts: Ty.getNumElements(), VRegs&: RegElts, |
204 | MIRBuilder, MRI); |
205 | Elts.append(RHS: RegElts); |
206 | } |
207 | |
208 | /// Merge \p PartRegs with different types into \p DstReg. |
209 | void LegalizerHelper::mergeMixedSubvectors(Register DstReg, |
210 | ArrayRef<Register> PartRegs) { |
211 | SmallVector<Register, 8> AllElts; |
212 | for (unsigned i = 0; i < PartRegs.size() - 1; ++i) |
213 | appendVectorElts(Elts&: AllElts, Reg: PartRegs[i]); |
214 | |
215 | Register Leftover = PartRegs[PartRegs.size() - 1]; |
216 | if (!MRI.getType(Reg: Leftover).isVector()) |
217 | AllElts.push_back(Elt: Leftover); |
218 | else |
219 | appendVectorElts(Elts&: AllElts, Reg: Leftover); |
220 | |
221 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: AllElts); |
222 | } |
223 | |
224 | /// Append the result registers of G_UNMERGE_VALUES \p MI to \p Regs. |
225 | static void getUnmergeResults(SmallVectorImpl<Register> &Regs, |
226 | const MachineInstr &MI) { |
227 | assert(MI.getOpcode() == TargetOpcode::G_UNMERGE_VALUES); |
228 | |
229 | const int StartIdx = Regs.size(); |
230 | const int NumResults = MI.getNumOperands() - 1; |
231 | Regs.resize(N: Regs.size() + NumResults); |
232 | for (int I = 0; I != NumResults; ++I) |
233 | Regs[StartIdx + I] = MI.getOperand(i: I).getReg(); |
234 | } |
235 | |
236 | void LegalizerHelper::(SmallVectorImpl<Register> &Parts, |
237 | LLT GCDTy, Register SrcReg) { |
238 | LLT SrcTy = MRI.getType(Reg: SrcReg); |
239 | if (SrcTy == GCDTy) { |
240 | // If the source already evenly divides the result type, we don't need to do |
241 | // anything. |
242 | Parts.push_back(Elt: SrcReg); |
243 | } else { |
244 | // Need to split into common type sized pieces. |
245 | auto Unmerge = MIRBuilder.buildUnmerge(Res: GCDTy, Op: SrcReg); |
246 | getUnmergeResults(Regs&: Parts, MI: *Unmerge); |
247 | } |
248 | } |
249 | |
250 | LLT LegalizerHelper::(SmallVectorImpl<Register> &Parts, LLT DstTy, |
251 | LLT NarrowTy, Register SrcReg) { |
252 | LLT SrcTy = MRI.getType(Reg: SrcReg); |
253 | LLT GCDTy = getGCDType(OrigTy: getGCDType(OrigTy: SrcTy, TargetTy: NarrowTy), TargetTy: DstTy); |
254 | extractGCDType(Parts, GCDTy, SrcReg); |
255 | return GCDTy; |
256 | } |
257 | |
258 | LLT LegalizerHelper::buildLCMMergePieces(LLT DstTy, LLT NarrowTy, LLT GCDTy, |
259 | SmallVectorImpl<Register> &VRegs, |
260 | unsigned PadStrategy) { |
261 | LLT LCMTy = getLCMType(OrigTy: DstTy, TargetTy: NarrowTy); |
262 | |
263 | int NumParts = LCMTy.getSizeInBits() / NarrowTy.getSizeInBits(); |
264 | int NumSubParts = NarrowTy.getSizeInBits() / GCDTy.getSizeInBits(); |
265 | int NumOrigSrc = VRegs.size(); |
266 | |
267 | Register PadReg; |
268 | |
269 | // Get a value we can use to pad the source value if the sources won't evenly |
270 | // cover the result type. |
271 | if (NumOrigSrc < NumParts * NumSubParts) { |
272 | if (PadStrategy == TargetOpcode::G_ZEXT) |
273 | PadReg = MIRBuilder.buildConstant(Res: GCDTy, Val: 0).getReg(Idx: 0); |
274 | else if (PadStrategy == TargetOpcode::G_ANYEXT) |
275 | PadReg = MIRBuilder.buildUndef(Res: GCDTy).getReg(Idx: 0); |
276 | else { |
277 | assert(PadStrategy == TargetOpcode::G_SEXT); |
278 | |
279 | // Shift the sign bit of the low register through the high register. |
280 | auto ShiftAmt = |
281 | MIRBuilder.buildConstant(Res: LLT::scalar(SizeInBits: 64), Val: GCDTy.getSizeInBits() - 1); |
282 | PadReg = MIRBuilder.buildAShr(Dst: GCDTy, Src0: VRegs.back(), Src1: ShiftAmt).getReg(Idx: 0); |
283 | } |
284 | } |
285 | |
286 | // Registers for the final merge to be produced. |
287 | SmallVector<Register, 4> Remerge(NumParts); |
288 | |
289 | // Registers needed for intermediate merges, which will be merged into a |
290 | // source for Remerge. |
291 | SmallVector<Register, 4> SubMerge(NumSubParts); |
292 | |
293 | // Once we've fully read off the end of the original source bits, we can reuse |
294 | // the same high bits for remaining padding elements. |
295 | Register AllPadReg; |
296 | |
297 | // Build merges to the LCM type to cover the original result type. |
298 | for (int I = 0; I != NumParts; ++I) { |
299 | bool AllMergePartsArePadding = true; |
300 | |
301 | // Build the requested merges to the requested type. |
302 | for (int J = 0; J != NumSubParts; ++J) { |
303 | int Idx = I * NumSubParts + J; |
304 | if (Idx >= NumOrigSrc) { |
305 | SubMerge[J] = PadReg; |
306 | continue; |
307 | } |
308 | |
309 | SubMerge[J] = VRegs[Idx]; |
310 | |
311 | // There are meaningful bits here we can't reuse later. |
312 | AllMergePartsArePadding = false; |
313 | } |
314 | |
315 | // If we've filled up a complete piece with padding bits, we can directly |
316 | // emit the natural sized constant if applicable, rather than a merge of |
317 | // smaller constants. |
318 | if (AllMergePartsArePadding && !AllPadReg) { |
319 | if (PadStrategy == TargetOpcode::G_ANYEXT) |
320 | AllPadReg = MIRBuilder.buildUndef(Res: NarrowTy).getReg(Idx: 0); |
321 | else if (PadStrategy == TargetOpcode::G_ZEXT) |
322 | AllPadReg = MIRBuilder.buildConstant(Res: NarrowTy, Val: 0).getReg(Idx: 0); |
323 | |
324 | // If this is a sign extension, we can't materialize a trivial constant |
325 | // with the right type and have to produce a merge. |
326 | } |
327 | |
328 | if (AllPadReg) { |
329 | // Avoid creating additional instructions if we're just adding additional |
330 | // copies of padding bits. |
331 | Remerge[I] = AllPadReg; |
332 | continue; |
333 | } |
334 | |
335 | if (NumSubParts == 1) |
336 | Remerge[I] = SubMerge[0]; |
337 | else |
338 | Remerge[I] = MIRBuilder.buildMergeLikeInstr(Res: NarrowTy, Ops: SubMerge).getReg(Idx: 0); |
339 | |
340 | // In the sign extend padding case, re-use the first all-signbit merge. |
341 | if (AllMergePartsArePadding && !AllPadReg) |
342 | AllPadReg = Remerge[I]; |
343 | } |
344 | |
345 | VRegs = std::move(Remerge); |
346 | return LCMTy; |
347 | } |
348 | |
349 | void LegalizerHelper::buildWidenedRemergeToDst(Register DstReg, LLT LCMTy, |
350 | ArrayRef<Register> RemergeRegs) { |
351 | LLT DstTy = MRI.getType(Reg: DstReg); |
352 | |
353 | // Create the merge to the widened source, and extract the relevant bits into |
354 | // the result. |
355 | |
356 | if (DstTy == LCMTy) { |
357 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: RemergeRegs); |
358 | return; |
359 | } |
360 | |
361 | auto Remerge = MIRBuilder.buildMergeLikeInstr(Res: LCMTy, Ops: RemergeRegs); |
362 | if (DstTy.isScalar() && LCMTy.isScalar()) { |
363 | MIRBuilder.buildTrunc(Res: DstReg, Op: Remerge); |
364 | return; |
365 | } |
366 | |
367 | if (LCMTy.isVector()) { |
368 | unsigned NumDefs = LCMTy.getSizeInBits() / DstTy.getSizeInBits(); |
369 | SmallVector<Register, 8> UnmergeDefs(NumDefs); |
370 | UnmergeDefs[0] = DstReg; |
371 | for (unsigned I = 1; I != NumDefs; ++I) |
372 | UnmergeDefs[I] = MRI.createGenericVirtualRegister(Ty: DstTy); |
373 | |
374 | MIRBuilder.buildUnmerge(Res: UnmergeDefs, |
375 | Op: MIRBuilder.buildMergeLikeInstr(Res: LCMTy, Ops: RemergeRegs)); |
376 | return; |
377 | } |
378 | |
379 | llvm_unreachable("unhandled case" ); |
380 | } |
381 | |
382 | static RTLIB::Libcall getRTLibDesc(unsigned Opcode, unsigned Size) { |
383 | #define RTLIBCASE_INT(LibcallPrefix) \ |
384 | do { \ |
385 | switch (Size) { \ |
386 | case 32: \ |
387 | return RTLIB::LibcallPrefix##32; \ |
388 | case 64: \ |
389 | return RTLIB::LibcallPrefix##64; \ |
390 | case 128: \ |
391 | return RTLIB::LibcallPrefix##128; \ |
392 | default: \ |
393 | llvm_unreachable("unexpected size"); \ |
394 | } \ |
395 | } while (0) |
396 | |
397 | #define RTLIBCASE(LibcallPrefix) \ |
398 | do { \ |
399 | switch (Size) { \ |
400 | case 32: \ |
401 | return RTLIB::LibcallPrefix##32; \ |
402 | case 64: \ |
403 | return RTLIB::LibcallPrefix##64; \ |
404 | case 80: \ |
405 | return RTLIB::LibcallPrefix##80; \ |
406 | case 128: \ |
407 | return RTLIB::LibcallPrefix##128; \ |
408 | default: \ |
409 | llvm_unreachable("unexpected size"); \ |
410 | } \ |
411 | } while (0) |
412 | |
413 | switch (Opcode) { |
414 | case TargetOpcode::G_MUL: |
415 | RTLIBCASE_INT(MUL_I); |
416 | case TargetOpcode::G_SDIV: |
417 | RTLIBCASE_INT(SDIV_I); |
418 | case TargetOpcode::G_UDIV: |
419 | RTLIBCASE_INT(UDIV_I); |
420 | case TargetOpcode::G_SREM: |
421 | RTLIBCASE_INT(SREM_I); |
422 | case TargetOpcode::G_UREM: |
423 | RTLIBCASE_INT(UREM_I); |
424 | case TargetOpcode::G_CTLZ_ZERO_UNDEF: |
425 | RTLIBCASE_INT(CTLZ_I); |
426 | case TargetOpcode::G_FADD: |
427 | RTLIBCASE(ADD_F); |
428 | case TargetOpcode::G_FSUB: |
429 | RTLIBCASE(SUB_F); |
430 | case TargetOpcode::G_FMUL: |
431 | RTLIBCASE(MUL_F); |
432 | case TargetOpcode::G_FDIV: |
433 | RTLIBCASE(DIV_F); |
434 | case TargetOpcode::G_FEXP: |
435 | RTLIBCASE(EXP_F); |
436 | case TargetOpcode::G_FEXP2: |
437 | RTLIBCASE(EXP2_F); |
438 | case TargetOpcode::G_FEXP10: |
439 | RTLIBCASE(EXP10_F); |
440 | case TargetOpcode::G_FREM: |
441 | RTLIBCASE(REM_F); |
442 | case TargetOpcode::G_FPOW: |
443 | RTLIBCASE(POW_F); |
444 | case TargetOpcode::G_FPOWI: |
445 | RTLIBCASE(POWI_F); |
446 | case TargetOpcode::G_FMA: |
447 | RTLIBCASE(FMA_F); |
448 | case TargetOpcode::G_FSIN: |
449 | RTLIBCASE(SIN_F); |
450 | case TargetOpcode::G_FCOS: |
451 | RTLIBCASE(COS_F); |
452 | case TargetOpcode::G_FTAN: |
453 | RTLIBCASE(TAN_F); |
454 | case TargetOpcode::G_FASIN: |
455 | RTLIBCASE(ASIN_F); |
456 | case TargetOpcode::G_FACOS: |
457 | RTLIBCASE(ACOS_F); |
458 | case TargetOpcode::G_FATAN: |
459 | RTLIBCASE(ATAN_F); |
460 | case TargetOpcode::G_FSINH: |
461 | RTLIBCASE(SINH_F); |
462 | case TargetOpcode::G_FCOSH: |
463 | RTLIBCASE(COSH_F); |
464 | case TargetOpcode::G_FTANH: |
465 | RTLIBCASE(TANH_F); |
466 | case TargetOpcode::G_FLOG10: |
467 | RTLIBCASE(LOG10_F); |
468 | case TargetOpcode::G_FLOG: |
469 | RTLIBCASE(LOG_F); |
470 | case TargetOpcode::G_FLOG2: |
471 | RTLIBCASE(LOG2_F); |
472 | case TargetOpcode::G_FLDEXP: |
473 | RTLIBCASE(LDEXP_F); |
474 | case TargetOpcode::G_FCEIL: |
475 | RTLIBCASE(CEIL_F); |
476 | case TargetOpcode::G_FFLOOR: |
477 | RTLIBCASE(FLOOR_F); |
478 | case TargetOpcode::G_FMINNUM: |
479 | RTLIBCASE(FMIN_F); |
480 | case TargetOpcode::G_FMAXNUM: |
481 | RTLIBCASE(FMAX_F); |
482 | case TargetOpcode::G_FSQRT: |
483 | RTLIBCASE(SQRT_F); |
484 | case TargetOpcode::G_FRINT: |
485 | RTLIBCASE(RINT_F); |
486 | case TargetOpcode::G_FNEARBYINT: |
487 | RTLIBCASE(NEARBYINT_F); |
488 | case TargetOpcode::G_INTRINSIC_ROUNDEVEN: |
489 | RTLIBCASE(ROUNDEVEN_F); |
490 | case TargetOpcode::G_INTRINSIC_LRINT: |
491 | RTLIBCASE(LRINT_F); |
492 | case TargetOpcode::G_INTRINSIC_LLRINT: |
493 | RTLIBCASE(LLRINT_F); |
494 | } |
495 | llvm_unreachable("Unknown libcall function" ); |
496 | } |
497 | |
498 | /// True if an instruction is in tail position in its caller. Intended for |
499 | /// legalizing libcalls as tail calls when possible. |
500 | static bool isLibCallInTailPosition(const CallLowering::ArgInfo &Result, |
501 | MachineInstr &MI, |
502 | const TargetInstrInfo &TII, |
503 | MachineRegisterInfo &MRI) { |
504 | MachineBasicBlock &MBB = *MI.getParent(); |
505 | const Function &F = MBB.getParent()->getFunction(); |
506 | |
507 | // Conservatively require the attributes of the call to match those of |
508 | // the return. Ignore NoAlias and NonNull because they don't affect the |
509 | // call sequence. |
510 | AttributeList CallerAttrs = F.getAttributes(); |
511 | if (AttrBuilder(F.getContext(), CallerAttrs.getRetAttrs()) |
512 | .removeAttribute(Val: Attribute::NoAlias) |
513 | .removeAttribute(Val: Attribute::NonNull) |
514 | .hasAttributes()) |
515 | return false; |
516 | |
517 | // It's not safe to eliminate the sign / zero extension of the return value. |
518 | if (CallerAttrs.hasRetAttr(Kind: Attribute::ZExt) || |
519 | CallerAttrs.hasRetAttr(Kind: Attribute::SExt)) |
520 | return false; |
521 | |
522 | // Only tail call if the following instruction is a standard return or if we |
523 | // have a `thisreturn` callee, and a sequence like: |
524 | // |
525 | // G_MEMCPY %0, %1, %2 |
526 | // $x0 = COPY %0 |
527 | // RET_ReallyLR implicit $x0 |
528 | auto Next = next_nodbg(It: MI.getIterator(), End: MBB.instr_end()); |
529 | if (Next != MBB.instr_end() && Next->isCopy()) { |
530 | if (MI.getOpcode() == TargetOpcode::G_BZERO) |
531 | return false; |
532 | |
533 | // For MEMCPY/MOMMOVE/MEMSET these will be the first use (the dst), as the |
534 | // mempy/etc routines return the same parameter. For other it will be the |
535 | // returned value. |
536 | Register VReg = MI.getOperand(i: 0).getReg(); |
537 | if (!VReg.isVirtual() || VReg != Next->getOperand(i: 1).getReg()) |
538 | return false; |
539 | |
540 | Register PReg = Next->getOperand(i: 0).getReg(); |
541 | if (!PReg.isPhysical()) |
542 | return false; |
543 | |
544 | auto Ret = next_nodbg(It: Next, End: MBB.instr_end()); |
545 | if (Ret == MBB.instr_end() || !Ret->isReturn()) |
546 | return false; |
547 | |
548 | if (Ret->getNumImplicitOperands() != 1) |
549 | return false; |
550 | |
551 | if (!Ret->getOperand(i: 0).isReg() || PReg != Ret->getOperand(i: 0).getReg()) |
552 | return false; |
553 | |
554 | // Skip over the COPY that we just validated. |
555 | Next = Ret; |
556 | } |
557 | |
558 | if (Next == MBB.instr_end() || TII.isTailCall(Inst: *Next) || !Next->isReturn()) |
559 | return false; |
560 | |
561 | return true; |
562 | } |
563 | |
564 | LegalizerHelper::LegalizeResult |
565 | llvm::createLibcall(MachineIRBuilder &MIRBuilder, const char *Name, |
566 | const CallLowering::ArgInfo &Result, |
567 | ArrayRef<CallLowering::ArgInfo> Args, |
568 | const CallingConv::ID CC, LostDebugLocObserver &LocObserver, |
569 | MachineInstr *MI) { |
570 | auto &CLI = *MIRBuilder.getMF().getSubtarget().getCallLowering(); |
571 | |
572 | CallLowering::CallLoweringInfo Info; |
573 | Info.CallConv = CC; |
574 | Info.Callee = MachineOperand::CreateES(SymName: Name); |
575 | Info.OrigRet = Result; |
576 | if (MI) |
577 | Info.IsTailCall = |
578 | (Result.Ty->isVoidTy() || |
579 | Result.Ty == MIRBuilder.getMF().getFunction().getReturnType()) && |
580 | isLibCallInTailPosition(Result, MI&: *MI, TII: MIRBuilder.getTII(), |
581 | MRI&: *MIRBuilder.getMRI()); |
582 | |
583 | std::copy(Args.begin(), Args.end(), std::back_inserter(x&: Info.OrigArgs)); |
584 | if (!CLI.lowerCall(MIRBuilder, Info)) |
585 | return LegalizerHelper::UnableToLegalize; |
586 | |
587 | if (MI && Info.LoweredTailCall) { |
588 | assert(Info.IsTailCall && "Lowered tail call when it wasn't a tail call?" ); |
589 | |
590 | // Check debug locations before removing the return. |
591 | LocObserver.checkpoint(CheckDebugLocs: true); |
592 | |
593 | // We must have a return following the call (or debug insts) to get past |
594 | // isLibCallInTailPosition. |
595 | do { |
596 | MachineInstr *Next = MI->getNextNode(); |
597 | assert(Next && |
598 | (Next->isCopy() || Next->isReturn() || Next->isDebugInstr()) && |
599 | "Expected instr following MI to be return or debug inst?" ); |
600 | // We lowered a tail call, so the call is now the return from the block. |
601 | // Delete the old return. |
602 | Next->eraseFromParent(); |
603 | } while (MI->getNextNode()); |
604 | |
605 | // We expect to lose the debug location from the return. |
606 | LocObserver.checkpoint(CheckDebugLocs: false); |
607 | } |
608 | return LegalizerHelper::Legalized; |
609 | } |
610 | |
611 | LegalizerHelper::LegalizeResult |
612 | llvm::createLibcall(MachineIRBuilder &MIRBuilder, RTLIB::Libcall Libcall, |
613 | const CallLowering::ArgInfo &Result, |
614 | ArrayRef<CallLowering::ArgInfo> Args, |
615 | LostDebugLocObserver &LocObserver, MachineInstr *MI) { |
616 | auto &TLI = *MIRBuilder.getMF().getSubtarget().getTargetLowering(); |
617 | const char *Name = TLI.getLibcallName(Call: Libcall); |
618 | if (!Name) |
619 | return LegalizerHelper::UnableToLegalize; |
620 | const CallingConv::ID CC = TLI.getLibcallCallingConv(Call: Libcall); |
621 | return createLibcall(MIRBuilder, Name, Result, Args, CC, LocObserver, MI); |
622 | } |
623 | |
624 | // Useful for libcalls where all operands have the same type. |
625 | static LegalizerHelper::LegalizeResult |
626 | simpleLibcall(MachineInstr &MI, MachineIRBuilder &MIRBuilder, unsigned Size, |
627 | Type *OpType, LostDebugLocObserver &LocObserver) { |
628 | auto Libcall = getRTLibDesc(Opcode: MI.getOpcode(), Size); |
629 | |
630 | // FIXME: What does the original arg index mean here? |
631 | SmallVector<CallLowering::ArgInfo, 3> Args; |
632 | for (const MachineOperand &MO : llvm::drop_begin(RangeOrContainer: MI.operands())) |
633 | Args.push_back(Elt: {MO.getReg(), OpType, 0}); |
634 | return createLibcall(MIRBuilder, Libcall, |
635 | Result: {MI.getOperand(i: 0).getReg(), OpType, 0}, Args, |
636 | LocObserver, MI: &MI); |
637 | } |
638 | |
639 | LegalizerHelper::LegalizeResult |
640 | llvm::createMemLibcall(MachineIRBuilder &MIRBuilder, MachineRegisterInfo &MRI, |
641 | MachineInstr &MI, LostDebugLocObserver &LocObserver) { |
642 | auto &Ctx = MIRBuilder.getMF().getFunction().getContext(); |
643 | |
644 | SmallVector<CallLowering::ArgInfo, 3> Args; |
645 | // Add all the args, except for the last which is an imm denoting 'tail'. |
646 | for (unsigned i = 0; i < MI.getNumOperands() - 1; ++i) { |
647 | Register Reg = MI.getOperand(i).getReg(); |
648 | |
649 | // Need derive an IR type for call lowering. |
650 | LLT OpLLT = MRI.getType(Reg); |
651 | Type *OpTy = nullptr; |
652 | if (OpLLT.isPointer()) |
653 | OpTy = PointerType::get(C&: Ctx, AddressSpace: OpLLT.getAddressSpace()); |
654 | else |
655 | OpTy = IntegerType::get(C&: Ctx, NumBits: OpLLT.getSizeInBits()); |
656 | Args.push_back(Elt: {Reg, OpTy, 0}); |
657 | } |
658 | |
659 | auto &CLI = *MIRBuilder.getMF().getSubtarget().getCallLowering(); |
660 | auto &TLI = *MIRBuilder.getMF().getSubtarget().getTargetLowering(); |
661 | RTLIB::Libcall RTLibcall; |
662 | unsigned Opc = MI.getOpcode(); |
663 | switch (Opc) { |
664 | case TargetOpcode::G_BZERO: |
665 | RTLibcall = RTLIB::BZERO; |
666 | break; |
667 | case TargetOpcode::G_MEMCPY: |
668 | RTLibcall = RTLIB::MEMCPY; |
669 | Args[0].Flags[0].setReturned(); |
670 | break; |
671 | case TargetOpcode::G_MEMMOVE: |
672 | RTLibcall = RTLIB::MEMMOVE; |
673 | Args[0].Flags[0].setReturned(); |
674 | break; |
675 | case TargetOpcode::G_MEMSET: |
676 | RTLibcall = RTLIB::MEMSET; |
677 | Args[0].Flags[0].setReturned(); |
678 | break; |
679 | default: |
680 | llvm_unreachable("unsupported opcode" ); |
681 | } |
682 | const char *Name = TLI.getLibcallName(Call: RTLibcall); |
683 | |
684 | // Unsupported libcall on the target. |
685 | if (!Name) { |
686 | LLVM_DEBUG(dbgs() << ".. .. Could not find libcall name for " |
687 | << MIRBuilder.getTII().getName(Opc) << "\n" ); |
688 | return LegalizerHelper::UnableToLegalize; |
689 | } |
690 | |
691 | CallLowering::CallLoweringInfo Info; |
692 | Info.CallConv = TLI.getLibcallCallingConv(Call: RTLibcall); |
693 | Info.Callee = MachineOperand::CreateES(SymName: Name); |
694 | Info.OrigRet = CallLowering::ArgInfo({0}, Type::getVoidTy(C&: Ctx), 0); |
695 | Info.IsTailCall = |
696 | MI.getOperand(i: MI.getNumOperands() - 1).getImm() && |
697 | isLibCallInTailPosition(Result: Info.OrigRet, MI, TII: MIRBuilder.getTII(), MRI); |
698 | |
699 | std::copy(Args.begin(), Args.end(), std::back_inserter(x&: Info.OrigArgs)); |
700 | if (!CLI.lowerCall(MIRBuilder, Info)) |
701 | return LegalizerHelper::UnableToLegalize; |
702 | |
703 | if (Info.LoweredTailCall) { |
704 | assert(Info.IsTailCall && "Lowered tail call when it wasn't a tail call?" ); |
705 | |
706 | // Check debug locations before removing the return. |
707 | LocObserver.checkpoint(CheckDebugLocs: true); |
708 | |
709 | // We must have a return following the call (or debug insts) to get past |
710 | // isLibCallInTailPosition. |
711 | do { |
712 | MachineInstr *Next = MI.getNextNode(); |
713 | assert(Next && |
714 | (Next->isCopy() || Next->isReturn() || Next->isDebugInstr()) && |
715 | "Expected instr following MI to be return or debug inst?" ); |
716 | // We lowered a tail call, so the call is now the return from the block. |
717 | // Delete the old return. |
718 | Next->eraseFromParent(); |
719 | } while (MI.getNextNode()); |
720 | |
721 | // We expect to lose the debug location from the return. |
722 | LocObserver.checkpoint(CheckDebugLocs: false); |
723 | } |
724 | |
725 | return LegalizerHelper::Legalized; |
726 | } |
727 | |
728 | static RTLIB::Libcall getOutlineAtomicLibcall(MachineInstr &MI) { |
729 | unsigned Opc = MI.getOpcode(); |
730 | auto &AtomicMI = cast<GMemOperation>(Val&: MI); |
731 | auto &MMO = AtomicMI.getMMO(); |
732 | auto Ordering = MMO.getMergedOrdering(); |
733 | LLT MemType = MMO.getMemoryType(); |
734 | uint64_t MemSize = MemType.getSizeInBytes(); |
735 | if (MemType.isVector()) |
736 | return RTLIB::UNKNOWN_LIBCALL; |
737 | |
738 | #define LCALLS(A, B) \ |
739 | { A##B##_RELAX, A##B##_ACQ, A##B##_REL, A##B##_ACQ_REL } |
740 | #define LCALL5(A) \ |
741 | LCALLS(A, 1), LCALLS(A, 2), LCALLS(A, 4), LCALLS(A, 8), LCALLS(A, 16) |
742 | switch (Opc) { |
743 | case TargetOpcode::G_ATOMIC_CMPXCHG: |
744 | case TargetOpcode::G_ATOMIC_CMPXCHG_WITH_SUCCESS: { |
745 | const RTLIB::Libcall LC[5][4] = {LCALL5(RTLIB::OUTLINE_ATOMIC_CAS)}; |
746 | return getOutlineAtomicHelper(LC, Order: Ordering, MemSize); |
747 | } |
748 | case TargetOpcode::G_ATOMICRMW_XCHG: { |
749 | const RTLIB::Libcall LC[5][4] = {LCALL5(RTLIB::OUTLINE_ATOMIC_SWP)}; |
750 | return getOutlineAtomicHelper(LC, Order: Ordering, MemSize); |
751 | } |
752 | case TargetOpcode::G_ATOMICRMW_ADD: |
753 | case TargetOpcode::G_ATOMICRMW_SUB: { |
754 | const RTLIB::Libcall LC[5][4] = {LCALL5(RTLIB::OUTLINE_ATOMIC_LDADD)}; |
755 | return getOutlineAtomicHelper(LC, Order: Ordering, MemSize); |
756 | } |
757 | case TargetOpcode::G_ATOMICRMW_AND: { |
758 | const RTLIB::Libcall LC[5][4] = {LCALL5(RTLIB::OUTLINE_ATOMIC_LDCLR)}; |
759 | return getOutlineAtomicHelper(LC, Order: Ordering, MemSize); |
760 | } |
761 | case TargetOpcode::G_ATOMICRMW_OR: { |
762 | const RTLIB::Libcall LC[5][4] = {LCALL5(RTLIB::OUTLINE_ATOMIC_LDSET)}; |
763 | return getOutlineAtomicHelper(LC, Order: Ordering, MemSize); |
764 | } |
765 | case TargetOpcode::G_ATOMICRMW_XOR: { |
766 | const RTLIB::Libcall LC[5][4] = {LCALL5(RTLIB::OUTLINE_ATOMIC_LDEOR)}; |
767 | return getOutlineAtomicHelper(LC, Order: Ordering, MemSize); |
768 | } |
769 | default: |
770 | return RTLIB::UNKNOWN_LIBCALL; |
771 | } |
772 | #undef LCALLS |
773 | #undef LCALL5 |
774 | } |
775 | |
776 | static LegalizerHelper::LegalizeResult |
777 | createAtomicLibcall(MachineIRBuilder &MIRBuilder, MachineInstr &MI) { |
778 | auto &Ctx = MIRBuilder.getMF().getFunction().getContext(); |
779 | |
780 | Type *RetTy; |
781 | SmallVector<Register> RetRegs; |
782 | SmallVector<CallLowering::ArgInfo, 3> Args; |
783 | unsigned Opc = MI.getOpcode(); |
784 | switch (Opc) { |
785 | case TargetOpcode::G_ATOMIC_CMPXCHG: |
786 | case TargetOpcode::G_ATOMIC_CMPXCHG_WITH_SUCCESS: { |
787 | Register Success; |
788 | LLT SuccessLLT; |
789 | auto [Ret, RetLLT, Mem, MemLLT, Cmp, CmpLLT, New, NewLLT] = |
790 | MI.getFirst4RegLLTs(); |
791 | RetRegs.push_back(Elt: Ret); |
792 | RetTy = IntegerType::get(C&: Ctx, NumBits: RetLLT.getSizeInBits()); |
793 | if (Opc == TargetOpcode::G_ATOMIC_CMPXCHG_WITH_SUCCESS) { |
794 | std::tie(args&: Ret, args&: RetLLT, args&: Success, args&: SuccessLLT, args&: Mem, args&: MemLLT, args&: Cmp, args&: CmpLLT, args&: New, |
795 | args&: NewLLT) = MI.getFirst5RegLLTs(); |
796 | RetRegs.push_back(Elt: Success); |
797 | RetTy = StructType::get( |
798 | Context&: Ctx, Elements: {RetTy, IntegerType::get(C&: Ctx, NumBits: SuccessLLT.getSizeInBits())}); |
799 | } |
800 | Args.push_back(Elt: {Cmp, IntegerType::get(C&: Ctx, NumBits: CmpLLT.getSizeInBits()), 0}); |
801 | Args.push_back(Elt: {New, IntegerType::get(C&: Ctx, NumBits: NewLLT.getSizeInBits()), 0}); |
802 | Args.push_back(Elt: {Mem, PointerType::get(C&: Ctx, AddressSpace: MemLLT.getAddressSpace()), 0}); |
803 | break; |
804 | } |
805 | case TargetOpcode::G_ATOMICRMW_XCHG: |
806 | case TargetOpcode::G_ATOMICRMW_ADD: |
807 | case TargetOpcode::G_ATOMICRMW_SUB: |
808 | case TargetOpcode::G_ATOMICRMW_AND: |
809 | case TargetOpcode::G_ATOMICRMW_OR: |
810 | case TargetOpcode::G_ATOMICRMW_XOR: { |
811 | auto [Ret, RetLLT, Mem, MemLLT, Val, ValLLT] = MI.getFirst3RegLLTs(); |
812 | RetRegs.push_back(Elt: Ret); |
813 | RetTy = IntegerType::get(C&: Ctx, NumBits: RetLLT.getSizeInBits()); |
814 | if (Opc == TargetOpcode::G_ATOMICRMW_AND) |
815 | Val = |
816 | MIRBuilder.buildXor(Dst: ValLLT, Src0: MIRBuilder.buildConstant(Res: ValLLT, Val: -1), Src1: Val) |
817 | .getReg(Idx: 0); |
818 | else if (Opc == TargetOpcode::G_ATOMICRMW_SUB) |
819 | Val = |
820 | MIRBuilder.buildSub(Dst: ValLLT, Src0: MIRBuilder.buildConstant(Res: ValLLT, Val: 0), Src1: Val) |
821 | .getReg(Idx: 0); |
822 | Args.push_back(Elt: {Val, IntegerType::get(C&: Ctx, NumBits: ValLLT.getSizeInBits()), 0}); |
823 | Args.push_back(Elt: {Mem, PointerType::get(C&: Ctx, AddressSpace: MemLLT.getAddressSpace()), 0}); |
824 | break; |
825 | } |
826 | default: |
827 | llvm_unreachable("unsupported opcode" ); |
828 | } |
829 | |
830 | auto &CLI = *MIRBuilder.getMF().getSubtarget().getCallLowering(); |
831 | auto &TLI = *MIRBuilder.getMF().getSubtarget().getTargetLowering(); |
832 | RTLIB::Libcall RTLibcall = getOutlineAtomicLibcall(MI); |
833 | const char *Name = TLI.getLibcallName(Call: RTLibcall); |
834 | |
835 | // Unsupported libcall on the target. |
836 | if (!Name) { |
837 | LLVM_DEBUG(dbgs() << ".. .. Could not find libcall name for " |
838 | << MIRBuilder.getTII().getName(Opc) << "\n" ); |
839 | return LegalizerHelper::UnableToLegalize; |
840 | } |
841 | |
842 | CallLowering::CallLoweringInfo Info; |
843 | Info.CallConv = TLI.getLibcallCallingConv(Call: RTLibcall); |
844 | Info.Callee = MachineOperand::CreateES(SymName: Name); |
845 | Info.OrigRet = CallLowering::ArgInfo(RetRegs, RetTy, 0); |
846 | |
847 | std::copy(Args.begin(), Args.end(), std::back_inserter(x&: Info.OrigArgs)); |
848 | if (!CLI.lowerCall(MIRBuilder, Info)) |
849 | return LegalizerHelper::UnableToLegalize; |
850 | |
851 | return LegalizerHelper::Legalized; |
852 | } |
853 | |
854 | static RTLIB::Libcall getConvRTLibDesc(unsigned Opcode, Type *ToType, |
855 | Type *FromType) { |
856 | auto ToMVT = MVT::getVT(Ty: ToType); |
857 | auto FromMVT = MVT::getVT(Ty: FromType); |
858 | |
859 | switch (Opcode) { |
860 | case TargetOpcode::G_FPEXT: |
861 | return RTLIB::getFPEXT(OpVT: FromMVT, RetVT: ToMVT); |
862 | case TargetOpcode::G_FPTRUNC: |
863 | return RTLIB::getFPROUND(OpVT: FromMVT, RetVT: ToMVT); |
864 | case TargetOpcode::G_FPTOSI: |
865 | return RTLIB::getFPTOSINT(OpVT: FromMVT, RetVT: ToMVT); |
866 | case TargetOpcode::G_FPTOUI: |
867 | return RTLIB::getFPTOUINT(OpVT: FromMVT, RetVT: ToMVT); |
868 | case TargetOpcode::G_SITOFP: |
869 | return RTLIB::getSINTTOFP(OpVT: FromMVT, RetVT: ToMVT); |
870 | case TargetOpcode::G_UITOFP: |
871 | return RTLIB::getUINTTOFP(OpVT: FromMVT, RetVT: ToMVT); |
872 | } |
873 | llvm_unreachable("Unsupported libcall function" ); |
874 | } |
875 | |
876 | static LegalizerHelper::LegalizeResult |
877 | conversionLibcall(MachineInstr &MI, MachineIRBuilder &MIRBuilder, Type *ToType, |
878 | Type *FromType, LostDebugLocObserver &LocObserver) { |
879 | RTLIB::Libcall Libcall = getConvRTLibDesc(Opcode: MI.getOpcode(), ToType, FromType); |
880 | return createLibcall( |
881 | MIRBuilder, Libcall, Result: {MI.getOperand(i: 0).getReg(), ToType, 0}, |
882 | Args: {{MI.getOperand(i: 1).getReg(), FromType, 0}}, LocObserver, MI: &MI); |
883 | } |
884 | |
885 | static RTLIB::Libcall |
886 | getStateLibraryFunctionFor(MachineInstr &MI, const TargetLowering &TLI) { |
887 | RTLIB::Libcall RTLibcall; |
888 | switch (MI.getOpcode()) { |
889 | case TargetOpcode::G_GET_FPENV: |
890 | RTLibcall = RTLIB::FEGETENV; |
891 | break; |
892 | case TargetOpcode::G_SET_FPENV: |
893 | case TargetOpcode::G_RESET_FPENV: |
894 | RTLibcall = RTLIB::FESETENV; |
895 | break; |
896 | case TargetOpcode::G_GET_FPMODE: |
897 | RTLibcall = RTLIB::FEGETMODE; |
898 | break; |
899 | case TargetOpcode::G_SET_FPMODE: |
900 | case TargetOpcode::G_RESET_FPMODE: |
901 | RTLibcall = RTLIB::FESETMODE; |
902 | break; |
903 | default: |
904 | llvm_unreachable("Unexpected opcode" ); |
905 | } |
906 | return RTLibcall; |
907 | } |
908 | |
909 | // Some library functions that read FP state (fegetmode, fegetenv) write the |
910 | // state into a region in memory. IR intrinsics that do the same operations |
911 | // (get_fpmode, get_fpenv) return the state as integer value. To implement these |
912 | // intrinsics via the library functions, we need to use temporary variable, |
913 | // for example: |
914 | // |
915 | // %0:_(s32) = G_GET_FPMODE |
916 | // |
917 | // is transformed to: |
918 | // |
919 | // %1:_(p0) = G_FRAME_INDEX %stack.0 |
920 | // BL &fegetmode |
921 | // %0:_(s32) = G_LOAD % 1 |
922 | // |
923 | LegalizerHelper::LegalizeResult |
924 | LegalizerHelper::createGetStateLibcall(MachineIRBuilder &MIRBuilder, |
925 | MachineInstr &MI, |
926 | LostDebugLocObserver &LocObserver) { |
927 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
928 | auto &MF = MIRBuilder.getMF(); |
929 | auto &MRI = *MIRBuilder.getMRI(); |
930 | auto &Ctx = MF.getFunction().getContext(); |
931 | |
932 | // Create temporary, where library function will put the read state. |
933 | Register Dst = MI.getOperand(i: 0).getReg(); |
934 | LLT StateTy = MRI.getType(Reg: Dst); |
935 | TypeSize StateSize = StateTy.getSizeInBytes(); |
936 | Align TempAlign = getStackTemporaryAlignment(Type: StateTy); |
937 | MachinePointerInfo TempPtrInfo; |
938 | auto Temp = createStackTemporary(Bytes: StateSize, Alignment: TempAlign, PtrInfo&: TempPtrInfo); |
939 | |
940 | // Create a call to library function, with the temporary as an argument. |
941 | unsigned TempAddrSpace = DL.getAllocaAddrSpace(); |
942 | Type *StatePtrTy = PointerType::get(C&: Ctx, AddressSpace: TempAddrSpace); |
943 | RTLIB::Libcall RTLibcall = getStateLibraryFunctionFor(MI, TLI); |
944 | auto Res = |
945 | createLibcall(MIRBuilder, Libcall: RTLibcall, |
946 | Result: CallLowering::ArgInfo({0}, Type::getVoidTy(C&: Ctx), 0), |
947 | Args: CallLowering::ArgInfo({Temp.getReg(Idx: 0), StatePtrTy, 0}), |
948 | LocObserver, MI: nullptr); |
949 | if (Res != LegalizerHelper::Legalized) |
950 | return Res; |
951 | |
952 | // Create a load from the temporary. |
953 | MachineMemOperand *MMO = MF.getMachineMemOperand( |
954 | PtrInfo: TempPtrInfo, f: MachineMemOperand::MOLoad, MemTy: StateTy, base_alignment: TempAlign); |
955 | MIRBuilder.buildLoadInstr(Opcode: TargetOpcode::G_LOAD, Res: Dst, Addr: Temp, MMO&: *MMO); |
956 | |
957 | return LegalizerHelper::Legalized; |
958 | } |
959 | |
960 | // Similar to `createGetStateLibcall` the function calls a library function |
961 | // using transient space in stack. In this case the library function reads |
962 | // content of memory region. |
963 | LegalizerHelper::LegalizeResult |
964 | LegalizerHelper::createSetStateLibcall(MachineIRBuilder &MIRBuilder, |
965 | MachineInstr &MI, |
966 | LostDebugLocObserver &LocObserver) { |
967 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
968 | auto &MF = MIRBuilder.getMF(); |
969 | auto &MRI = *MIRBuilder.getMRI(); |
970 | auto &Ctx = MF.getFunction().getContext(); |
971 | |
972 | // Create temporary, where library function will get the new state. |
973 | Register Src = MI.getOperand(i: 0).getReg(); |
974 | LLT StateTy = MRI.getType(Reg: Src); |
975 | TypeSize StateSize = StateTy.getSizeInBytes(); |
976 | Align TempAlign = getStackTemporaryAlignment(Type: StateTy); |
977 | MachinePointerInfo TempPtrInfo; |
978 | auto Temp = createStackTemporary(Bytes: StateSize, Alignment: TempAlign, PtrInfo&: TempPtrInfo); |
979 | |
980 | // Put the new state into the temporary. |
981 | MachineMemOperand *MMO = MF.getMachineMemOperand( |
982 | PtrInfo: TempPtrInfo, f: MachineMemOperand::MOStore, MemTy: StateTy, base_alignment: TempAlign); |
983 | MIRBuilder.buildStore(Val: Src, Addr: Temp, MMO&: *MMO); |
984 | |
985 | // Create a call to library function, with the temporary as an argument. |
986 | unsigned TempAddrSpace = DL.getAllocaAddrSpace(); |
987 | Type *StatePtrTy = PointerType::get(C&: Ctx, AddressSpace: TempAddrSpace); |
988 | RTLIB::Libcall RTLibcall = getStateLibraryFunctionFor(MI, TLI); |
989 | return createLibcall(MIRBuilder, Libcall: RTLibcall, |
990 | Result: CallLowering::ArgInfo({0}, Type::getVoidTy(C&: Ctx), 0), |
991 | Args: CallLowering::ArgInfo({Temp.getReg(Idx: 0), StatePtrTy, 0}), |
992 | LocObserver, MI: nullptr); |
993 | } |
994 | |
995 | // The function is used to legalize operations that set default environment |
996 | // state. In C library a call like `fesetmode(FE_DFL_MODE)` is used for that. |
997 | // On most targets supported in glibc FE_DFL_MODE is defined as |
998 | // `((const femode_t *) -1)`. Such assumption is used here. If for some target |
999 | // it is not true, the target must provide custom lowering. |
1000 | LegalizerHelper::LegalizeResult |
1001 | LegalizerHelper::createResetStateLibcall(MachineIRBuilder &MIRBuilder, |
1002 | MachineInstr &MI, |
1003 | LostDebugLocObserver &LocObserver) { |
1004 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
1005 | auto &MF = MIRBuilder.getMF(); |
1006 | auto &Ctx = MF.getFunction().getContext(); |
1007 | |
1008 | // Create an argument for the library function. |
1009 | unsigned AddrSpace = DL.getDefaultGlobalsAddressSpace(); |
1010 | Type *StatePtrTy = PointerType::get(C&: Ctx, AddressSpace: AddrSpace); |
1011 | unsigned PtrSize = DL.getPointerSizeInBits(AS: AddrSpace); |
1012 | LLT MemTy = LLT::pointer(AddressSpace: AddrSpace, SizeInBits: PtrSize); |
1013 | auto DefValue = MIRBuilder.buildConstant(Res: LLT::scalar(SizeInBits: PtrSize), Val: -1LL); |
1014 | DstOp Dest(MRI.createGenericVirtualRegister(Ty: MemTy)); |
1015 | MIRBuilder.buildIntToPtr(Dst: Dest, Src: DefValue); |
1016 | |
1017 | RTLIB::Libcall RTLibcall = getStateLibraryFunctionFor(MI, TLI); |
1018 | return createLibcall(MIRBuilder, Libcall: RTLibcall, |
1019 | Result: CallLowering::ArgInfo({0}, Type::getVoidTy(C&: Ctx), 0), |
1020 | Args: CallLowering::ArgInfo({Dest.getReg(), StatePtrTy, 0}), |
1021 | LocObserver, MI: &MI); |
1022 | } |
1023 | |
1024 | LegalizerHelper::LegalizeResult |
1025 | LegalizerHelper::libcall(MachineInstr &MI, LostDebugLocObserver &LocObserver) { |
1026 | auto &Ctx = MIRBuilder.getMF().getFunction().getContext(); |
1027 | |
1028 | switch (MI.getOpcode()) { |
1029 | default: |
1030 | return UnableToLegalize; |
1031 | case TargetOpcode::G_MUL: |
1032 | case TargetOpcode::G_SDIV: |
1033 | case TargetOpcode::G_UDIV: |
1034 | case TargetOpcode::G_SREM: |
1035 | case TargetOpcode::G_UREM: |
1036 | case TargetOpcode::G_CTLZ_ZERO_UNDEF: { |
1037 | LLT LLTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
1038 | unsigned Size = LLTy.getSizeInBits(); |
1039 | Type *HLTy = IntegerType::get(C&: Ctx, NumBits: Size); |
1040 | auto Status = simpleLibcall(MI, MIRBuilder, Size, OpType: HLTy, LocObserver); |
1041 | if (Status != Legalized) |
1042 | return Status; |
1043 | break; |
1044 | } |
1045 | case TargetOpcode::G_FADD: |
1046 | case TargetOpcode::G_FSUB: |
1047 | case TargetOpcode::G_FMUL: |
1048 | case TargetOpcode::G_FDIV: |
1049 | case TargetOpcode::G_FMA: |
1050 | case TargetOpcode::G_FPOW: |
1051 | case TargetOpcode::G_FREM: |
1052 | case TargetOpcode::G_FCOS: |
1053 | case TargetOpcode::G_FSIN: |
1054 | case TargetOpcode::G_FTAN: |
1055 | case TargetOpcode::G_FACOS: |
1056 | case TargetOpcode::G_FASIN: |
1057 | case TargetOpcode::G_FATAN: |
1058 | case TargetOpcode::G_FCOSH: |
1059 | case TargetOpcode::G_FSINH: |
1060 | case TargetOpcode::G_FTANH: |
1061 | case TargetOpcode::G_FLOG10: |
1062 | case TargetOpcode::G_FLOG: |
1063 | case TargetOpcode::G_FLOG2: |
1064 | case TargetOpcode::G_FLDEXP: |
1065 | case TargetOpcode::G_FEXP: |
1066 | case TargetOpcode::G_FEXP2: |
1067 | case TargetOpcode::G_FEXP10: |
1068 | case TargetOpcode::G_FCEIL: |
1069 | case TargetOpcode::G_FFLOOR: |
1070 | case TargetOpcode::G_FMINNUM: |
1071 | case TargetOpcode::G_FMAXNUM: |
1072 | case TargetOpcode::G_FSQRT: |
1073 | case TargetOpcode::G_FRINT: |
1074 | case TargetOpcode::G_FNEARBYINT: |
1075 | case TargetOpcode::G_INTRINSIC_ROUNDEVEN: { |
1076 | LLT LLTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
1077 | unsigned Size = LLTy.getSizeInBits(); |
1078 | Type *HLTy = getFloatTypeForLLT(Ctx, Ty: LLTy); |
1079 | if (!HLTy || (Size != 32 && Size != 64 && Size != 80 && Size != 128)) { |
1080 | LLVM_DEBUG(dbgs() << "No libcall available for type " << LLTy << ".\n" ); |
1081 | return UnableToLegalize; |
1082 | } |
1083 | auto Status = simpleLibcall(MI, MIRBuilder, Size, OpType: HLTy, LocObserver); |
1084 | if (Status != Legalized) |
1085 | return Status; |
1086 | break; |
1087 | } |
1088 | case TargetOpcode::G_INTRINSIC_LRINT: |
1089 | case TargetOpcode::G_INTRINSIC_LLRINT: { |
1090 | LLT LLTy = MRI.getType(Reg: MI.getOperand(i: 1).getReg()); |
1091 | unsigned Size = LLTy.getSizeInBits(); |
1092 | Type *HLTy = getFloatTypeForLLT(Ctx, Ty: LLTy); |
1093 | Type *ITy = IntegerType::get( |
1094 | C&: Ctx, NumBits: MRI.getType(Reg: MI.getOperand(i: 0).getReg()).getSizeInBits()); |
1095 | if (!HLTy || (Size != 32 && Size != 64 && Size != 80 && Size != 128)) { |
1096 | LLVM_DEBUG(dbgs() << "No libcall available for type " << LLTy << ".\n" ); |
1097 | return UnableToLegalize; |
1098 | } |
1099 | auto Libcall = getRTLibDesc(Opcode: MI.getOpcode(), Size); |
1100 | LegalizeResult Status = |
1101 | createLibcall(MIRBuilder, Libcall, Result: {MI.getOperand(i: 0).getReg(), ITy, 0}, |
1102 | Args: {{MI.getOperand(i: 1).getReg(), HLTy, 0}}, LocObserver, MI: &MI); |
1103 | if (Status != Legalized) |
1104 | return Status; |
1105 | MI.eraseFromParent(); |
1106 | return Legalized; |
1107 | } |
1108 | case TargetOpcode::G_FPOWI: { |
1109 | LLT LLTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
1110 | unsigned Size = LLTy.getSizeInBits(); |
1111 | Type *HLTy = getFloatTypeForLLT(Ctx, Ty: LLTy); |
1112 | Type *ITy = IntegerType::get( |
1113 | C&: Ctx, NumBits: MRI.getType(Reg: MI.getOperand(i: 2).getReg()).getSizeInBits()); |
1114 | if (!HLTy || (Size != 32 && Size != 64 && Size != 80 && Size != 128)) { |
1115 | LLVM_DEBUG(dbgs() << "No libcall available for type " << LLTy << ".\n" ); |
1116 | return UnableToLegalize; |
1117 | } |
1118 | auto Libcall = getRTLibDesc(Opcode: MI.getOpcode(), Size); |
1119 | std::initializer_list<CallLowering::ArgInfo> Args = { |
1120 | {MI.getOperand(i: 1).getReg(), HLTy, 0}, |
1121 | {MI.getOperand(i: 2).getReg(), ITy, 1}}; |
1122 | LegalizeResult Status = |
1123 | createLibcall(MIRBuilder, Libcall, Result: {MI.getOperand(i: 0).getReg(), HLTy, 0}, |
1124 | Args, LocObserver, MI: &MI); |
1125 | if (Status != Legalized) |
1126 | return Status; |
1127 | break; |
1128 | } |
1129 | case TargetOpcode::G_FPEXT: |
1130 | case TargetOpcode::G_FPTRUNC: { |
1131 | Type *FromTy = getFloatTypeForLLT(Ctx, Ty: MRI.getType(Reg: MI.getOperand(i: 1).getReg())); |
1132 | Type *ToTy = getFloatTypeForLLT(Ctx, Ty: MRI.getType(Reg: MI.getOperand(i: 0).getReg())); |
1133 | if (!FromTy || !ToTy) |
1134 | return UnableToLegalize; |
1135 | LegalizeResult Status = |
1136 | conversionLibcall(MI, MIRBuilder, ToType: ToTy, FromType: FromTy, LocObserver); |
1137 | if (Status != Legalized) |
1138 | return Status; |
1139 | break; |
1140 | } |
1141 | case TargetOpcode::G_FPTOSI: |
1142 | case TargetOpcode::G_FPTOUI: { |
1143 | // FIXME: Support other types |
1144 | Type *FromTy = |
1145 | getFloatTypeForLLT(Ctx, Ty: MRI.getType(Reg: MI.getOperand(i: 1).getReg())); |
1146 | unsigned ToSize = MRI.getType(Reg: MI.getOperand(i: 0).getReg()).getSizeInBits(); |
1147 | if ((ToSize != 32 && ToSize != 64 && ToSize != 128) || !FromTy) |
1148 | return UnableToLegalize; |
1149 | LegalizeResult Status = conversionLibcall( |
1150 | MI, MIRBuilder, ToType: Type::getIntNTy(C&: Ctx, N: ToSize), FromType: FromTy, LocObserver); |
1151 | if (Status != Legalized) |
1152 | return Status; |
1153 | break; |
1154 | } |
1155 | case TargetOpcode::G_SITOFP: |
1156 | case TargetOpcode::G_UITOFP: { |
1157 | unsigned FromSize = MRI.getType(Reg: MI.getOperand(i: 1).getReg()).getSizeInBits(); |
1158 | Type *ToTy = |
1159 | getFloatTypeForLLT(Ctx, Ty: MRI.getType(Reg: MI.getOperand(i: 0).getReg())); |
1160 | if ((FromSize != 32 && FromSize != 64 && FromSize != 128) || !ToTy) |
1161 | return UnableToLegalize; |
1162 | LegalizeResult Status = conversionLibcall( |
1163 | MI, MIRBuilder, ToType: ToTy, FromType: Type::getIntNTy(C&: Ctx, N: FromSize), LocObserver); |
1164 | if (Status != Legalized) |
1165 | return Status; |
1166 | break; |
1167 | } |
1168 | case TargetOpcode::G_ATOMICRMW_XCHG: |
1169 | case TargetOpcode::G_ATOMICRMW_ADD: |
1170 | case TargetOpcode::G_ATOMICRMW_SUB: |
1171 | case TargetOpcode::G_ATOMICRMW_AND: |
1172 | case TargetOpcode::G_ATOMICRMW_OR: |
1173 | case TargetOpcode::G_ATOMICRMW_XOR: |
1174 | case TargetOpcode::G_ATOMIC_CMPXCHG: |
1175 | case TargetOpcode::G_ATOMIC_CMPXCHG_WITH_SUCCESS: { |
1176 | auto Status = createAtomicLibcall(MIRBuilder, MI); |
1177 | if (Status != Legalized) |
1178 | return Status; |
1179 | break; |
1180 | } |
1181 | case TargetOpcode::G_BZERO: |
1182 | case TargetOpcode::G_MEMCPY: |
1183 | case TargetOpcode::G_MEMMOVE: |
1184 | case TargetOpcode::G_MEMSET: { |
1185 | LegalizeResult Result = |
1186 | createMemLibcall(MIRBuilder, MRI&: *MIRBuilder.getMRI(), MI, LocObserver); |
1187 | if (Result != Legalized) |
1188 | return Result; |
1189 | MI.eraseFromParent(); |
1190 | return Result; |
1191 | } |
1192 | case TargetOpcode::G_GET_FPENV: |
1193 | case TargetOpcode::G_GET_FPMODE: { |
1194 | LegalizeResult Result = createGetStateLibcall(MIRBuilder, MI, LocObserver); |
1195 | if (Result != Legalized) |
1196 | return Result; |
1197 | break; |
1198 | } |
1199 | case TargetOpcode::G_SET_FPENV: |
1200 | case TargetOpcode::G_SET_FPMODE: { |
1201 | LegalizeResult Result = createSetStateLibcall(MIRBuilder, MI, LocObserver); |
1202 | if (Result != Legalized) |
1203 | return Result; |
1204 | break; |
1205 | } |
1206 | case TargetOpcode::G_RESET_FPENV: |
1207 | case TargetOpcode::G_RESET_FPMODE: { |
1208 | LegalizeResult Result = |
1209 | createResetStateLibcall(MIRBuilder, MI, LocObserver); |
1210 | if (Result != Legalized) |
1211 | return Result; |
1212 | break; |
1213 | } |
1214 | } |
1215 | |
1216 | MI.eraseFromParent(); |
1217 | return Legalized; |
1218 | } |
1219 | |
1220 | LegalizerHelper::LegalizeResult LegalizerHelper::narrowScalar(MachineInstr &MI, |
1221 | unsigned TypeIdx, |
1222 | LLT NarrowTy) { |
1223 | uint64_t SizeOp0 = MRI.getType(Reg: MI.getOperand(i: 0).getReg()).getSizeInBits(); |
1224 | uint64_t NarrowSize = NarrowTy.getSizeInBits(); |
1225 | |
1226 | switch (MI.getOpcode()) { |
1227 | default: |
1228 | return UnableToLegalize; |
1229 | case TargetOpcode::G_IMPLICIT_DEF: { |
1230 | Register DstReg = MI.getOperand(i: 0).getReg(); |
1231 | LLT DstTy = MRI.getType(Reg: DstReg); |
1232 | |
1233 | // If SizeOp0 is not an exact multiple of NarrowSize, emit |
1234 | // G_ANYEXT(G_IMPLICIT_DEF). Cast result to vector if needed. |
1235 | // FIXME: Although this would also be legal for the general case, it causes |
1236 | // a lot of regressions in the emitted code (superfluous COPYs, artifact |
1237 | // combines not being hit). This seems to be a problem related to the |
1238 | // artifact combiner. |
1239 | if (SizeOp0 % NarrowSize != 0) { |
1240 | LLT ImplicitTy = NarrowTy; |
1241 | if (DstTy.isVector()) |
1242 | ImplicitTy = LLT::vector(EC: DstTy.getElementCount(), ScalarTy: ImplicitTy); |
1243 | |
1244 | Register ImplicitReg = MIRBuilder.buildUndef(Res: ImplicitTy).getReg(Idx: 0); |
1245 | MIRBuilder.buildAnyExt(Res: DstReg, Op: ImplicitReg); |
1246 | |
1247 | MI.eraseFromParent(); |
1248 | return Legalized; |
1249 | } |
1250 | |
1251 | int NumParts = SizeOp0 / NarrowSize; |
1252 | |
1253 | SmallVector<Register, 2> DstRegs; |
1254 | for (int i = 0; i < NumParts; ++i) |
1255 | DstRegs.push_back(Elt: MIRBuilder.buildUndef(Res: NarrowTy).getReg(Idx: 0)); |
1256 | |
1257 | if (DstTy.isVector()) |
1258 | MIRBuilder.buildBuildVector(Res: DstReg, Ops: DstRegs); |
1259 | else |
1260 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: DstRegs); |
1261 | MI.eraseFromParent(); |
1262 | return Legalized; |
1263 | } |
1264 | case TargetOpcode::G_CONSTANT: { |
1265 | LLT Ty = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
1266 | const APInt &Val = MI.getOperand(i: 1).getCImm()->getValue(); |
1267 | unsigned TotalSize = Ty.getSizeInBits(); |
1268 | unsigned NarrowSize = NarrowTy.getSizeInBits(); |
1269 | int NumParts = TotalSize / NarrowSize; |
1270 | |
1271 | SmallVector<Register, 4> PartRegs; |
1272 | for (int I = 0; I != NumParts; ++I) { |
1273 | unsigned Offset = I * NarrowSize; |
1274 | auto K = MIRBuilder.buildConstant(Res: NarrowTy, |
1275 | Val: Val.lshr(shiftAmt: Offset).trunc(width: NarrowSize)); |
1276 | PartRegs.push_back(Elt: K.getReg(Idx: 0)); |
1277 | } |
1278 | |
1279 | LLT LeftoverTy; |
1280 | unsigned LeftoverBits = TotalSize - NumParts * NarrowSize; |
1281 | SmallVector<Register, 1> LeftoverRegs; |
1282 | if (LeftoverBits != 0) { |
1283 | LeftoverTy = LLT::scalar(SizeInBits: LeftoverBits); |
1284 | auto K = MIRBuilder.buildConstant( |
1285 | Res: LeftoverTy, |
1286 | Val: Val.lshr(shiftAmt: NumParts * NarrowSize).trunc(width: LeftoverBits)); |
1287 | LeftoverRegs.push_back(Elt: K.getReg(Idx: 0)); |
1288 | } |
1289 | |
1290 | insertParts(DstReg: MI.getOperand(i: 0).getReg(), |
1291 | ResultTy: Ty, PartTy: NarrowTy, PartRegs, LeftoverTy, LeftoverRegs); |
1292 | |
1293 | MI.eraseFromParent(); |
1294 | return Legalized; |
1295 | } |
1296 | case TargetOpcode::G_SEXT: |
1297 | case TargetOpcode::G_ZEXT: |
1298 | case TargetOpcode::G_ANYEXT: |
1299 | return narrowScalarExt(MI, TypeIdx, Ty: NarrowTy); |
1300 | case TargetOpcode::G_TRUNC: { |
1301 | if (TypeIdx != 1) |
1302 | return UnableToLegalize; |
1303 | |
1304 | uint64_t SizeOp1 = MRI.getType(Reg: MI.getOperand(i: 1).getReg()).getSizeInBits(); |
1305 | if (NarrowTy.getSizeInBits() * 2 != SizeOp1) { |
1306 | LLVM_DEBUG(dbgs() << "Can't narrow trunc to type " << NarrowTy << "\n" ); |
1307 | return UnableToLegalize; |
1308 | } |
1309 | |
1310 | auto Unmerge = MIRBuilder.buildUnmerge(Res: NarrowTy, Op: MI.getOperand(i: 1)); |
1311 | MIRBuilder.buildCopy(Res: MI.getOperand(i: 0), Op: Unmerge.getReg(Idx: 0)); |
1312 | MI.eraseFromParent(); |
1313 | return Legalized; |
1314 | } |
1315 | case TargetOpcode::G_CONSTANT_FOLD_BARRIER: |
1316 | case TargetOpcode::G_FREEZE: { |
1317 | if (TypeIdx != 0) |
1318 | return UnableToLegalize; |
1319 | |
1320 | LLT Ty = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
1321 | // Should widen scalar first |
1322 | if (Ty.getSizeInBits() % NarrowTy.getSizeInBits() != 0) |
1323 | return UnableToLegalize; |
1324 | |
1325 | auto Unmerge = MIRBuilder.buildUnmerge(Res: NarrowTy, Op: MI.getOperand(i: 1).getReg()); |
1326 | SmallVector<Register, 8> Parts; |
1327 | for (unsigned i = 0; i < Unmerge->getNumDefs(); ++i) { |
1328 | Parts.push_back( |
1329 | Elt: MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {NarrowTy}, SrcOps: {Unmerge.getReg(Idx: i)}) |
1330 | .getReg(Idx: 0)); |
1331 | } |
1332 | |
1333 | MIRBuilder.buildMergeLikeInstr(Res: MI.getOperand(i: 0).getReg(), Ops: Parts); |
1334 | MI.eraseFromParent(); |
1335 | return Legalized; |
1336 | } |
1337 | case TargetOpcode::G_ADD: |
1338 | case TargetOpcode::G_SUB: |
1339 | case TargetOpcode::G_SADDO: |
1340 | case TargetOpcode::G_SSUBO: |
1341 | case TargetOpcode::G_SADDE: |
1342 | case TargetOpcode::G_SSUBE: |
1343 | case TargetOpcode::G_UADDO: |
1344 | case TargetOpcode::G_USUBO: |
1345 | case TargetOpcode::G_UADDE: |
1346 | case TargetOpcode::G_USUBE: |
1347 | return narrowScalarAddSub(MI, TypeIdx, NarrowTy); |
1348 | case TargetOpcode::G_MUL: |
1349 | case TargetOpcode::G_UMULH: |
1350 | return narrowScalarMul(MI, Ty: NarrowTy); |
1351 | case TargetOpcode::G_EXTRACT: |
1352 | return narrowScalarExtract(MI, TypeIdx, Ty: NarrowTy); |
1353 | case TargetOpcode::G_INSERT: |
1354 | return narrowScalarInsert(MI, TypeIdx, Ty: NarrowTy); |
1355 | case TargetOpcode::G_LOAD: { |
1356 | auto &LoadMI = cast<GLoad>(Val&: MI); |
1357 | Register DstReg = LoadMI.getDstReg(); |
1358 | LLT DstTy = MRI.getType(Reg: DstReg); |
1359 | if (DstTy.isVector()) |
1360 | return UnableToLegalize; |
1361 | |
1362 | if (8 * LoadMI.getMemSize().getValue() != DstTy.getSizeInBits()) { |
1363 | Register TmpReg = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
1364 | MIRBuilder.buildLoad(Res: TmpReg, Addr: LoadMI.getPointerReg(), MMO&: LoadMI.getMMO()); |
1365 | MIRBuilder.buildAnyExt(Res: DstReg, Op: TmpReg); |
1366 | LoadMI.eraseFromParent(); |
1367 | return Legalized; |
1368 | } |
1369 | |
1370 | return reduceLoadStoreWidth(MI&: LoadMI, TypeIdx, NarrowTy); |
1371 | } |
1372 | case TargetOpcode::G_ZEXTLOAD: |
1373 | case TargetOpcode::G_SEXTLOAD: { |
1374 | auto &LoadMI = cast<GExtLoad>(Val&: MI); |
1375 | Register DstReg = LoadMI.getDstReg(); |
1376 | Register PtrReg = LoadMI.getPointerReg(); |
1377 | |
1378 | Register TmpReg = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
1379 | auto &MMO = LoadMI.getMMO(); |
1380 | unsigned MemSize = MMO.getSizeInBits().getValue(); |
1381 | |
1382 | if (MemSize == NarrowSize) { |
1383 | MIRBuilder.buildLoad(Res: TmpReg, Addr: PtrReg, MMO); |
1384 | } else if (MemSize < NarrowSize) { |
1385 | MIRBuilder.buildLoadInstr(Opcode: LoadMI.getOpcode(), Res: TmpReg, Addr: PtrReg, MMO); |
1386 | } else if (MemSize > NarrowSize) { |
1387 | // FIXME: Need to split the load. |
1388 | return UnableToLegalize; |
1389 | } |
1390 | |
1391 | if (isa<GZExtLoad>(Val: LoadMI)) |
1392 | MIRBuilder.buildZExt(Res: DstReg, Op: TmpReg); |
1393 | else |
1394 | MIRBuilder.buildSExt(Res: DstReg, Op: TmpReg); |
1395 | |
1396 | LoadMI.eraseFromParent(); |
1397 | return Legalized; |
1398 | } |
1399 | case TargetOpcode::G_STORE: { |
1400 | auto &StoreMI = cast<GStore>(Val&: MI); |
1401 | |
1402 | Register SrcReg = StoreMI.getValueReg(); |
1403 | LLT SrcTy = MRI.getType(Reg: SrcReg); |
1404 | if (SrcTy.isVector()) |
1405 | return UnableToLegalize; |
1406 | |
1407 | int NumParts = SizeOp0 / NarrowSize; |
1408 | unsigned HandledSize = NumParts * NarrowTy.getSizeInBits(); |
1409 | unsigned LeftoverBits = SrcTy.getSizeInBits() - HandledSize; |
1410 | if (SrcTy.isVector() && LeftoverBits != 0) |
1411 | return UnableToLegalize; |
1412 | |
1413 | if (8 * StoreMI.getMemSize().getValue() != SrcTy.getSizeInBits()) { |
1414 | Register TmpReg = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
1415 | MIRBuilder.buildTrunc(Res: TmpReg, Op: SrcReg); |
1416 | MIRBuilder.buildStore(Val: TmpReg, Addr: StoreMI.getPointerReg(), MMO&: StoreMI.getMMO()); |
1417 | StoreMI.eraseFromParent(); |
1418 | return Legalized; |
1419 | } |
1420 | |
1421 | return reduceLoadStoreWidth(MI&: StoreMI, TypeIdx: 0, NarrowTy); |
1422 | } |
1423 | case TargetOpcode::G_SELECT: |
1424 | return narrowScalarSelect(MI, TypeIdx, Ty: NarrowTy); |
1425 | case TargetOpcode::G_AND: |
1426 | case TargetOpcode::G_OR: |
1427 | case TargetOpcode::G_XOR: { |
1428 | // Legalize bitwise operation: |
1429 | // A = BinOp<Ty> B, C |
1430 | // into: |
1431 | // B1, ..., BN = G_UNMERGE_VALUES B |
1432 | // C1, ..., CN = G_UNMERGE_VALUES C |
1433 | // A1 = BinOp<Ty/N> B1, C2 |
1434 | // ... |
1435 | // AN = BinOp<Ty/N> BN, CN |
1436 | // A = G_MERGE_VALUES A1, ..., AN |
1437 | return narrowScalarBasic(MI, TypeIdx, Ty: NarrowTy); |
1438 | } |
1439 | case TargetOpcode::G_SHL: |
1440 | case TargetOpcode::G_LSHR: |
1441 | case TargetOpcode::G_ASHR: |
1442 | return narrowScalarShift(MI, TypeIdx, Ty: NarrowTy); |
1443 | case TargetOpcode::G_CTLZ: |
1444 | case TargetOpcode::G_CTLZ_ZERO_UNDEF: |
1445 | case TargetOpcode::G_CTTZ: |
1446 | case TargetOpcode::G_CTTZ_ZERO_UNDEF: |
1447 | case TargetOpcode::G_CTPOP: |
1448 | if (TypeIdx == 1) |
1449 | switch (MI.getOpcode()) { |
1450 | case TargetOpcode::G_CTLZ: |
1451 | case TargetOpcode::G_CTLZ_ZERO_UNDEF: |
1452 | return narrowScalarCTLZ(MI, TypeIdx, Ty: NarrowTy); |
1453 | case TargetOpcode::G_CTTZ: |
1454 | case TargetOpcode::G_CTTZ_ZERO_UNDEF: |
1455 | return narrowScalarCTTZ(MI, TypeIdx, Ty: NarrowTy); |
1456 | case TargetOpcode::G_CTPOP: |
1457 | return narrowScalarCTPOP(MI, TypeIdx, Ty: NarrowTy); |
1458 | default: |
1459 | return UnableToLegalize; |
1460 | } |
1461 | |
1462 | Observer.changingInstr(MI); |
1463 | narrowScalarDst(MI, NarrowTy, OpIdx: 0, ExtOpcode: TargetOpcode::G_ZEXT); |
1464 | Observer.changedInstr(MI); |
1465 | return Legalized; |
1466 | case TargetOpcode::G_INTTOPTR: |
1467 | if (TypeIdx != 1) |
1468 | return UnableToLegalize; |
1469 | |
1470 | Observer.changingInstr(MI); |
1471 | narrowScalarSrc(MI, NarrowTy, OpIdx: 1); |
1472 | Observer.changedInstr(MI); |
1473 | return Legalized; |
1474 | case TargetOpcode::G_PTRTOINT: |
1475 | if (TypeIdx != 0) |
1476 | return UnableToLegalize; |
1477 | |
1478 | Observer.changingInstr(MI); |
1479 | narrowScalarDst(MI, NarrowTy, OpIdx: 0, ExtOpcode: TargetOpcode::G_ZEXT); |
1480 | Observer.changedInstr(MI); |
1481 | return Legalized; |
1482 | case TargetOpcode::G_PHI: { |
1483 | // FIXME: add support for when SizeOp0 isn't an exact multiple of |
1484 | // NarrowSize. |
1485 | if (SizeOp0 % NarrowSize != 0) |
1486 | return UnableToLegalize; |
1487 | |
1488 | unsigned NumParts = SizeOp0 / NarrowSize; |
1489 | SmallVector<Register, 2> DstRegs(NumParts); |
1490 | SmallVector<SmallVector<Register, 2>, 2> SrcRegs(MI.getNumOperands() / 2); |
1491 | Observer.changingInstr(MI); |
1492 | for (unsigned i = 1; i < MI.getNumOperands(); i += 2) { |
1493 | MachineBasicBlock &OpMBB = *MI.getOperand(i: i + 1).getMBB(); |
1494 | MIRBuilder.setInsertPt(MBB&: OpMBB, II: OpMBB.getFirstTerminatorForward()); |
1495 | extractParts(Reg: MI.getOperand(i).getReg(), Ty: NarrowTy, NumParts, |
1496 | VRegs&: SrcRegs[i / 2], MIRBuilder, MRI); |
1497 | } |
1498 | MachineBasicBlock &MBB = *MI.getParent(); |
1499 | MIRBuilder.setInsertPt(MBB, II: MI); |
1500 | for (unsigned i = 0; i < NumParts; ++i) { |
1501 | DstRegs[i] = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
1502 | MachineInstrBuilder MIB = |
1503 | MIRBuilder.buildInstr(Opcode: TargetOpcode::G_PHI).addDef(RegNo: DstRegs[i]); |
1504 | for (unsigned j = 1; j < MI.getNumOperands(); j += 2) |
1505 | MIB.addUse(RegNo: SrcRegs[j / 2][i]).add(MO: MI.getOperand(i: j + 1)); |
1506 | } |
1507 | MIRBuilder.setInsertPt(MBB, II: MBB.getFirstNonPHI()); |
1508 | MIRBuilder.buildMergeLikeInstr(Res: MI.getOperand(i: 0), Ops: DstRegs); |
1509 | Observer.changedInstr(MI); |
1510 | MI.eraseFromParent(); |
1511 | return Legalized; |
1512 | } |
1513 | case TargetOpcode::G_EXTRACT_VECTOR_ELT: |
1514 | case TargetOpcode::G_INSERT_VECTOR_ELT: { |
1515 | if (TypeIdx != 2) |
1516 | return UnableToLegalize; |
1517 | |
1518 | int OpIdx = MI.getOpcode() == TargetOpcode::G_EXTRACT_VECTOR_ELT ? 2 : 3; |
1519 | Observer.changingInstr(MI); |
1520 | narrowScalarSrc(MI, NarrowTy, OpIdx); |
1521 | Observer.changedInstr(MI); |
1522 | return Legalized; |
1523 | } |
1524 | case TargetOpcode::G_ICMP: { |
1525 | Register LHS = MI.getOperand(i: 2).getReg(); |
1526 | LLT SrcTy = MRI.getType(Reg: LHS); |
1527 | uint64_t SrcSize = SrcTy.getSizeInBits(); |
1528 | CmpInst::Predicate Pred = |
1529 | static_cast<CmpInst::Predicate>(MI.getOperand(i: 1).getPredicate()); |
1530 | |
1531 | // TODO: Handle the non-equality case for weird sizes. |
1532 | if (NarrowSize * 2 != SrcSize && !ICmpInst::isEquality(P: Pred)) |
1533 | return UnableToLegalize; |
1534 | |
1535 | LLT LeftoverTy; // Example: s88 -> s64 (NarrowTy) + s24 (leftover) |
1536 | SmallVector<Register, 4> LHSPartRegs, LHSLeftoverRegs; |
1537 | if (!extractParts(Reg: LHS, RegTy: SrcTy, MainTy: NarrowTy, LeftoverTy, VRegs&: LHSPartRegs, |
1538 | LeftoverVRegs&: LHSLeftoverRegs, MIRBuilder, MRI)) |
1539 | return UnableToLegalize; |
1540 | |
1541 | LLT Unused; // Matches LeftoverTy; G_ICMP LHS and RHS are the same type. |
1542 | SmallVector<Register, 4> RHSPartRegs, RHSLeftoverRegs; |
1543 | if (!extractParts(Reg: MI.getOperand(i: 3).getReg(), RegTy: SrcTy, MainTy: NarrowTy, LeftoverTy&: Unused, |
1544 | VRegs&: RHSPartRegs, LeftoverVRegs&: RHSLeftoverRegs, MIRBuilder, MRI)) |
1545 | return UnableToLegalize; |
1546 | |
1547 | // We now have the LHS and RHS of the compare split into narrow-type |
1548 | // registers, plus potentially some leftover type. |
1549 | Register Dst = MI.getOperand(i: 0).getReg(); |
1550 | LLT ResTy = MRI.getType(Reg: Dst); |
1551 | if (ICmpInst::isEquality(P: Pred)) { |
1552 | // For each part on the LHS and RHS, keep track of the result of XOR-ing |
1553 | // them together. For each equal part, the result should be all 0s. For |
1554 | // each non-equal part, we'll get at least one 1. |
1555 | auto Zero = MIRBuilder.buildConstant(Res: NarrowTy, Val: 0); |
1556 | SmallVector<Register, 4> Xors; |
1557 | for (auto LHSAndRHS : zip(t&: LHSPartRegs, u&: RHSPartRegs)) { |
1558 | auto LHS = std::get<0>(t&: LHSAndRHS); |
1559 | auto RHS = std::get<1>(t&: LHSAndRHS); |
1560 | auto Xor = MIRBuilder.buildXor(Dst: NarrowTy, Src0: LHS, Src1: RHS).getReg(Idx: 0); |
1561 | Xors.push_back(Elt: Xor); |
1562 | } |
1563 | |
1564 | // Build a G_XOR for each leftover register. Each G_XOR must be widened |
1565 | // to the desired narrow type so that we can OR them together later. |
1566 | SmallVector<Register, 4> WidenedXors; |
1567 | for (auto LHSAndRHS : zip(t&: LHSLeftoverRegs, u&: RHSLeftoverRegs)) { |
1568 | auto LHS = std::get<0>(t&: LHSAndRHS); |
1569 | auto RHS = std::get<1>(t&: LHSAndRHS); |
1570 | auto Xor = MIRBuilder.buildXor(Dst: LeftoverTy, Src0: LHS, Src1: RHS).getReg(Idx: 0); |
1571 | LLT GCDTy = extractGCDType(Parts&: WidenedXors, DstTy: NarrowTy, NarrowTy: LeftoverTy, SrcReg: Xor); |
1572 | buildLCMMergePieces(DstTy: LeftoverTy, NarrowTy, GCDTy, VRegs&: WidenedXors, |
1573 | /* PadStrategy = */ TargetOpcode::G_ZEXT); |
1574 | Xors.insert(I: Xors.end(), From: WidenedXors.begin(), To: WidenedXors.end()); |
1575 | } |
1576 | |
1577 | // Now, for each part we broke up, we know if they are equal/not equal |
1578 | // based off the G_XOR. We can OR these all together and compare against |
1579 | // 0 to get the result. |
1580 | assert(Xors.size() >= 2 && "Should have gotten at least two Xors?" ); |
1581 | auto Or = MIRBuilder.buildOr(Dst: NarrowTy, Src0: Xors[0], Src1: Xors[1]); |
1582 | for (unsigned I = 2, E = Xors.size(); I < E; ++I) |
1583 | Or = MIRBuilder.buildOr(Dst: NarrowTy, Src0: Or, Src1: Xors[I]); |
1584 | MIRBuilder.buildICmp(Pred, Res: Dst, Op0: Or, Op1: Zero); |
1585 | } else { |
1586 | // TODO: Handle non-power-of-two types. |
1587 | assert(LHSPartRegs.size() == 2 && "Expected exactly 2 LHS part regs?" ); |
1588 | assert(RHSPartRegs.size() == 2 && "Expected exactly 2 RHS part regs?" ); |
1589 | Register LHSL = LHSPartRegs[0]; |
1590 | Register LHSH = LHSPartRegs[1]; |
1591 | Register RHSL = RHSPartRegs[0]; |
1592 | Register RHSH = RHSPartRegs[1]; |
1593 | MachineInstrBuilder CmpH = MIRBuilder.buildICmp(Pred, Res: ResTy, Op0: LHSH, Op1: RHSH); |
1594 | MachineInstrBuilder CmpHEQ = |
1595 | MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_EQ, Res: ResTy, Op0: LHSH, Op1: RHSH); |
1596 | MachineInstrBuilder CmpLU = MIRBuilder.buildICmp( |
1597 | Pred: ICmpInst::getUnsignedPredicate(pred: Pred), Res: ResTy, Op0: LHSL, Op1: RHSL); |
1598 | MIRBuilder.buildSelect(Res: Dst, Tst: CmpHEQ, Op0: CmpLU, Op1: CmpH); |
1599 | } |
1600 | MI.eraseFromParent(); |
1601 | return Legalized; |
1602 | } |
1603 | case TargetOpcode::G_FCMP: |
1604 | if (TypeIdx != 0) |
1605 | return UnableToLegalize; |
1606 | |
1607 | Observer.changingInstr(MI); |
1608 | narrowScalarDst(MI, NarrowTy, OpIdx: 0, ExtOpcode: TargetOpcode::G_ZEXT); |
1609 | Observer.changedInstr(MI); |
1610 | return Legalized; |
1611 | |
1612 | case TargetOpcode::G_SEXT_INREG: { |
1613 | if (TypeIdx != 0) |
1614 | return UnableToLegalize; |
1615 | |
1616 | int64_t SizeInBits = MI.getOperand(i: 2).getImm(); |
1617 | |
1618 | // So long as the new type has more bits than the bits we're extending we |
1619 | // don't need to break it apart. |
1620 | if (NarrowTy.getScalarSizeInBits() > SizeInBits) { |
1621 | Observer.changingInstr(MI); |
1622 | // We don't lose any non-extension bits by truncating the src and |
1623 | // sign-extending the dst. |
1624 | MachineOperand &MO1 = MI.getOperand(i: 1); |
1625 | auto TruncMIB = MIRBuilder.buildTrunc(Res: NarrowTy, Op: MO1); |
1626 | MO1.setReg(TruncMIB.getReg(Idx: 0)); |
1627 | |
1628 | MachineOperand &MO2 = MI.getOperand(i: 0); |
1629 | Register DstExt = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
1630 | MIRBuilder.setInsertPt(MBB&: MIRBuilder.getMBB(), II: ++MIRBuilder.getInsertPt()); |
1631 | MIRBuilder.buildSExt(Res: MO2, Op: DstExt); |
1632 | MO2.setReg(DstExt); |
1633 | Observer.changedInstr(MI); |
1634 | return Legalized; |
1635 | } |
1636 | |
1637 | // Break it apart. Components below the extension point are unmodified. The |
1638 | // component containing the extension point becomes a narrower SEXT_INREG. |
1639 | // Components above it are ashr'd from the component containing the |
1640 | // extension point. |
1641 | if (SizeOp0 % NarrowSize != 0) |
1642 | return UnableToLegalize; |
1643 | int NumParts = SizeOp0 / NarrowSize; |
1644 | |
1645 | // List the registers where the destination will be scattered. |
1646 | SmallVector<Register, 2> DstRegs; |
1647 | // List the registers where the source will be split. |
1648 | SmallVector<Register, 2> SrcRegs; |
1649 | |
1650 | // Create all the temporary registers. |
1651 | for (int i = 0; i < NumParts; ++i) { |
1652 | Register SrcReg = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
1653 | |
1654 | SrcRegs.push_back(Elt: SrcReg); |
1655 | } |
1656 | |
1657 | // Explode the big arguments into smaller chunks. |
1658 | MIRBuilder.buildUnmerge(Res: SrcRegs, Op: MI.getOperand(i: 1)); |
1659 | |
1660 | Register AshrCstReg = |
1661 | MIRBuilder.buildConstant(Res: NarrowTy, Val: NarrowTy.getScalarSizeInBits() - 1) |
1662 | .getReg(Idx: 0); |
1663 | Register FullExtensionReg; |
1664 | Register PartialExtensionReg; |
1665 | |
1666 | // Do the operation on each small part. |
1667 | for (int i = 0; i < NumParts; ++i) { |
1668 | if ((i + 1) * NarrowTy.getScalarSizeInBits() <= SizeInBits) { |
1669 | DstRegs.push_back(Elt: SrcRegs[i]); |
1670 | PartialExtensionReg = DstRegs.back(); |
1671 | } else if (i * NarrowTy.getScalarSizeInBits() >= SizeInBits) { |
1672 | assert(PartialExtensionReg && |
1673 | "Expected to visit partial extension before full" ); |
1674 | if (FullExtensionReg) { |
1675 | DstRegs.push_back(Elt: FullExtensionReg); |
1676 | continue; |
1677 | } |
1678 | DstRegs.push_back( |
1679 | Elt: MIRBuilder.buildAShr(Dst: NarrowTy, Src0: PartialExtensionReg, Src1: AshrCstReg) |
1680 | .getReg(Idx: 0)); |
1681 | FullExtensionReg = DstRegs.back(); |
1682 | } else { |
1683 | DstRegs.push_back( |
1684 | Elt: MIRBuilder |
1685 | .buildInstr( |
1686 | Opc: TargetOpcode::G_SEXT_INREG, DstOps: {NarrowTy}, |
1687 | SrcOps: {SrcRegs[i], SizeInBits % NarrowTy.getScalarSizeInBits()}) |
1688 | .getReg(Idx: 0)); |
1689 | PartialExtensionReg = DstRegs.back(); |
1690 | } |
1691 | } |
1692 | |
1693 | // Gather the destination registers into the final destination. |
1694 | Register DstReg = MI.getOperand(i: 0).getReg(); |
1695 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: DstRegs); |
1696 | MI.eraseFromParent(); |
1697 | return Legalized; |
1698 | } |
1699 | case TargetOpcode::G_BSWAP: |
1700 | case TargetOpcode::G_BITREVERSE: { |
1701 | if (SizeOp0 % NarrowSize != 0) |
1702 | return UnableToLegalize; |
1703 | |
1704 | Observer.changingInstr(MI); |
1705 | SmallVector<Register, 2> SrcRegs, DstRegs; |
1706 | unsigned NumParts = SizeOp0 / NarrowSize; |
1707 | extractParts(Reg: MI.getOperand(i: 1).getReg(), Ty: NarrowTy, NumParts, VRegs&: SrcRegs, |
1708 | MIRBuilder, MRI); |
1709 | |
1710 | for (unsigned i = 0; i < NumParts; ++i) { |
1711 | auto DstPart = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {NarrowTy}, |
1712 | SrcOps: {SrcRegs[NumParts - 1 - i]}); |
1713 | DstRegs.push_back(Elt: DstPart.getReg(Idx: 0)); |
1714 | } |
1715 | |
1716 | MIRBuilder.buildMergeLikeInstr(Res: MI.getOperand(i: 0), Ops: DstRegs); |
1717 | |
1718 | Observer.changedInstr(MI); |
1719 | MI.eraseFromParent(); |
1720 | return Legalized; |
1721 | } |
1722 | case TargetOpcode::G_PTR_ADD: |
1723 | case TargetOpcode::G_PTRMASK: { |
1724 | if (TypeIdx != 1) |
1725 | return UnableToLegalize; |
1726 | Observer.changingInstr(MI); |
1727 | narrowScalarSrc(MI, NarrowTy, OpIdx: 2); |
1728 | Observer.changedInstr(MI); |
1729 | return Legalized; |
1730 | } |
1731 | case TargetOpcode::G_FPTOUI: |
1732 | case TargetOpcode::G_FPTOSI: |
1733 | return narrowScalarFPTOI(MI, TypeIdx, Ty: NarrowTy); |
1734 | case TargetOpcode::G_FPEXT: |
1735 | if (TypeIdx != 0) |
1736 | return UnableToLegalize; |
1737 | Observer.changingInstr(MI); |
1738 | narrowScalarDst(MI, NarrowTy, OpIdx: 0, ExtOpcode: TargetOpcode::G_FPEXT); |
1739 | Observer.changedInstr(MI); |
1740 | return Legalized; |
1741 | case TargetOpcode::G_FLDEXP: |
1742 | case TargetOpcode::G_STRICT_FLDEXP: |
1743 | return narrowScalarFLDEXP(MI, TypeIdx, Ty: NarrowTy); |
1744 | case TargetOpcode::G_VSCALE: { |
1745 | Register Dst = MI.getOperand(i: 0).getReg(); |
1746 | LLT Ty = MRI.getType(Reg: Dst); |
1747 | |
1748 | // Assume VSCALE(1) fits into a legal integer |
1749 | const APInt One(NarrowTy.getSizeInBits(), 1); |
1750 | auto VScaleBase = MIRBuilder.buildVScale(Res: NarrowTy, MinElts: One); |
1751 | auto ZExt = MIRBuilder.buildZExt(Res: Ty, Op: VScaleBase); |
1752 | auto C = MIRBuilder.buildConstant(Res: Ty, Val: *MI.getOperand(i: 1).getCImm()); |
1753 | MIRBuilder.buildMul(Dst, Src0: ZExt, Src1: C); |
1754 | |
1755 | MI.eraseFromParent(); |
1756 | return Legalized; |
1757 | } |
1758 | } |
1759 | } |
1760 | |
1761 | Register LegalizerHelper::coerceToScalar(Register Val) { |
1762 | LLT Ty = MRI.getType(Reg: Val); |
1763 | if (Ty.isScalar()) |
1764 | return Val; |
1765 | |
1766 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
1767 | LLT NewTy = LLT::scalar(SizeInBits: Ty.getSizeInBits()); |
1768 | if (Ty.isPointer()) { |
1769 | if (DL.isNonIntegralAddressSpace(AddrSpace: Ty.getAddressSpace())) |
1770 | return Register(); |
1771 | return MIRBuilder.buildPtrToInt(Dst: NewTy, Src: Val).getReg(Idx: 0); |
1772 | } |
1773 | |
1774 | Register NewVal = Val; |
1775 | |
1776 | assert(Ty.isVector()); |
1777 | if (Ty.isPointerVector()) |
1778 | NewVal = MIRBuilder.buildPtrToInt(Dst: NewTy, Src: NewVal).getReg(Idx: 0); |
1779 | return MIRBuilder.buildBitcast(Dst: NewTy, Src: NewVal).getReg(Idx: 0); |
1780 | } |
1781 | |
1782 | void LegalizerHelper::widenScalarSrc(MachineInstr &MI, LLT WideTy, |
1783 | unsigned OpIdx, unsigned ExtOpcode) { |
1784 | MachineOperand &MO = MI.getOperand(i: OpIdx); |
1785 | auto ExtB = MIRBuilder.buildInstr(Opc: ExtOpcode, DstOps: {WideTy}, SrcOps: {MO}); |
1786 | MO.setReg(ExtB.getReg(Idx: 0)); |
1787 | } |
1788 | |
1789 | void LegalizerHelper::narrowScalarSrc(MachineInstr &MI, LLT NarrowTy, |
1790 | unsigned OpIdx) { |
1791 | MachineOperand &MO = MI.getOperand(i: OpIdx); |
1792 | auto ExtB = MIRBuilder.buildTrunc(Res: NarrowTy, Op: MO); |
1793 | MO.setReg(ExtB.getReg(Idx: 0)); |
1794 | } |
1795 | |
1796 | void LegalizerHelper::widenScalarDst(MachineInstr &MI, LLT WideTy, |
1797 | unsigned OpIdx, unsigned TruncOpcode) { |
1798 | MachineOperand &MO = MI.getOperand(i: OpIdx); |
1799 | Register DstExt = MRI.createGenericVirtualRegister(Ty: WideTy); |
1800 | MIRBuilder.setInsertPt(MBB&: MIRBuilder.getMBB(), II: ++MIRBuilder.getInsertPt()); |
1801 | MIRBuilder.buildInstr(Opc: TruncOpcode, DstOps: {MO}, SrcOps: {DstExt}); |
1802 | MO.setReg(DstExt); |
1803 | } |
1804 | |
1805 | void LegalizerHelper::narrowScalarDst(MachineInstr &MI, LLT NarrowTy, |
1806 | unsigned OpIdx, unsigned ExtOpcode) { |
1807 | MachineOperand &MO = MI.getOperand(i: OpIdx); |
1808 | Register DstTrunc = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
1809 | MIRBuilder.setInsertPt(MBB&: MIRBuilder.getMBB(), II: ++MIRBuilder.getInsertPt()); |
1810 | MIRBuilder.buildInstr(Opc: ExtOpcode, DstOps: {MO}, SrcOps: {DstTrunc}); |
1811 | MO.setReg(DstTrunc); |
1812 | } |
1813 | |
1814 | void LegalizerHelper::moreElementsVectorDst(MachineInstr &MI, LLT WideTy, |
1815 | unsigned OpIdx) { |
1816 | MachineOperand &MO = MI.getOperand(i: OpIdx); |
1817 | MIRBuilder.setInsertPt(MBB&: MIRBuilder.getMBB(), II: ++MIRBuilder.getInsertPt()); |
1818 | Register Dst = MO.getReg(); |
1819 | Register DstExt = MRI.createGenericVirtualRegister(Ty: WideTy); |
1820 | MO.setReg(DstExt); |
1821 | MIRBuilder.buildDeleteTrailingVectorElements(Res: Dst, Op0: DstExt); |
1822 | } |
1823 | |
1824 | void LegalizerHelper::moreElementsVectorSrc(MachineInstr &MI, LLT MoreTy, |
1825 | unsigned OpIdx) { |
1826 | MachineOperand &MO = MI.getOperand(i: OpIdx); |
1827 | SmallVector<Register, 8> Regs; |
1828 | MO.setReg(MIRBuilder.buildPadVectorWithUndefElements(Res: MoreTy, Op0: MO).getReg(Idx: 0)); |
1829 | } |
1830 | |
1831 | void LegalizerHelper::bitcastSrc(MachineInstr &MI, LLT CastTy, unsigned OpIdx) { |
1832 | MachineOperand &Op = MI.getOperand(i: OpIdx); |
1833 | Op.setReg(MIRBuilder.buildBitcast(Dst: CastTy, Src: Op).getReg(Idx: 0)); |
1834 | } |
1835 | |
1836 | void LegalizerHelper::bitcastDst(MachineInstr &MI, LLT CastTy, unsigned OpIdx) { |
1837 | MachineOperand &MO = MI.getOperand(i: OpIdx); |
1838 | Register CastDst = MRI.createGenericVirtualRegister(Ty: CastTy); |
1839 | MIRBuilder.setInsertPt(MBB&: MIRBuilder.getMBB(), II: ++MIRBuilder.getInsertPt()); |
1840 | MIRBuilder.buildBitcast(Dst: MO, Src: CastDst); |
1841 | MO.setReg(CastDst); |
1842 | } |
1843 | |
1844 | LegalizerHelper::LegalizeResult |
1845 | LegalizerHelper::widenScalarMergeValues(MachineInstr &MI, unsigned TypeIdx, |
1846 | LLT WideTy) { |
1847 | if (TypeIdx != 1) |
1848 | return UnableToLegalize; |
1849 | |
1850 | auto [DstReg, DstTy, Src1Reg, Src1Ty] = MI.getFirst2RegLLTs(); |
1851 | if (DstTy.isVector()) |
1852 | return UnableToLegalize; |
1853 | |
1854 | LLT SrcTy = MRI.getType(Reg: Src1Reg); |
1855 | const int DstSize = DstTy.getSizeInBits(); |
1856 | const int SrcSize = SrcTy.getSizeInBits(); |
1857 | const int WideSize = WideTy.getSizeInBits(); |
1858 | const int NumMerge = (DstSize + WideSize - 1) / WideSize; |
1859 | |
1860 | unsigned NumOps = MI.getNumOperands(); |
1861 | unsigned NumSrc = MI.getNumOperands() - 1; |
1862 | unsigned PartSize = DstTy.getSizeInBits() / NumSrc; |
1863 | |
1864 | if (WideSize >= DstSize) { |
1865 | // Directly pack the bits in the target type. |
1866 | Register ResultReg = MIRBuilder.buildZExt(Res: WideTy, Op: Src1Reg).getReg(Idx: 0); |
1867 | |
1868 | for (unsigned I = 2; I != NumOps; ++I) { |
1869 | const unsigned Offset = (I - 1) * PartSize; |
1870 | |
1871 | Register SrcReg = MI.getOperand(i: I).getReg(); |
1872 | assert(MRI.getType(SrcReg) == LLT::scalar(PartSize)); |
1873 | |
1874 | auto ZextInput = MIRBuilder.buildZExt(Res: WideTy, Op: SrcReg); |
1875 | |
1876 | Register NextResult = I + 1 == NumOps && WideTy == DstTy ? DstReg : |
1877 | MRI.createGenericVirtualRegister(Ty: WideTy); |
1878 | |
1879 | auto ShiftAmt = MIRBuilder.buildConstant(Res: WideTy, Val: Offset); |
1880 | auto Shl = MIRBuilder.buildShl(Dst: WideTy, Src0: ZextInput, Src1: ShiftAmt); |
1881 | MIRBuilder.buildOr(Dst: NextResult, Src0: ResultReg, Src1: Shl); |
1882 | ResultReg = NextResult; |
1883 | } |
1884 | |
1885 | if (WideSize > DstSize) |
1886 | MIRBuilder.buildTrunc(Res: DstReg, Op: ResultReg); |
1887 | else if (DstTy.isPointer()) |
1888 | MIRBuilder.buildIntToPtr(Dst: DstReg, Src: ResultReg); |
1889 | |
1890 | MI.eraseFromParent(); |
1891 | return Legalized; |
1892 | } |
1893 | |
1894 | // Unmerge the original values to the GCD type, and recombine to the next |
1895 | // multiple greater than the original type. |
1896 | // |
1897 | // %3:_(s12) = G_MERGE_VALUES %0:_(s4), %1:_(s4), %2:_(s4) -> s6 |
1898 | // %4:_(s2), %5:_(s2) = G_UNMERGE_VALUES %0 |
1899 | // %6:_(s2), %7:_(s2) = G_UNMERGE_VALUES %1 |
1900 | // %8:_(s2), %9:_(s2) = G_UNMERGE_VALUES %2 |
1901 | // %10:_(s6) = G_MERGE_VALUES %4, %5, %6 |
1902 | // %11:_(s6) = G_MERGE_VALUES %7, %8, %9 |
1903 | // %12:_(s12) = G_MERGE_VALUES %10, %11 |
1904 | // |
1905 | // Padding with undef if necessary: |
1906 | // |
1907 | // %2:_(s8) = G_MERGE_VALUES %0:_(s4), %1:_(s4) -> s6 |
1908 | // %3:_(s2), %4:_(s2) = G_UNMERGE_VALUES %0 |
1909 | // %5:_(s2), %6:_(s2) = G_UNMERGE_VALUES %1 |
1910 | // %7:_(s2) = G_IMPLICIT_DEF |
1911 | // %8:_(s6) = G_MERGE_VALUES %3, %4, %5 |
1912 | // %9:_(s6) = G_MERGE_VALUES %6, %7, %7 |
1913 | // %10:_(s12) = G_MERGE_VALUES %8, %9 |
1914 | |
1915 | const int GCD = std::gcd(m: SrcSize, n: WideSize); |
1916 | LLT GCDTy = LLT::scalar(SizeInBits: GCD); |
1917 | |
1918 | SmallVector<Register, 8> Parts; |
1919 | SmallVector<Register, 8> NewMergeRegs; |
1920 | SmallVector<Register, 8> Unmerges; |
1921 | LLT WideDstTy = LLT::scalar(SizeInBits: NumMerge * WideSize); |
1922 | |
1923 | // Decompose the original operands if they don't evenly divide. |
1924 | for (const MachineOperand &MO : llvm::drop_begin(RangeOrContainer: MI.operands())) { |
1925 | Register SrcReg = MO.getReg(); |
1926 | if (GCD == SrcSize) { |
1927 | Unmerges.push_back(Elt: SrcReg); |
1928 | } else { |
1929 | auto Unmerge = MIRBuilder.buildUnmerge(Res: GCDTy, Op: SrcReg); |
1930 | for (int J = 0, JE = Unmerge->getNumOperands() - 1; J != JE; ++J) |
1931 | Unmerges.push_back(Elt: Unmerge.getReg(Idx: J)); |
1932 | } |
1933 | } |
1934 | |
1935 | // Pad with undef to the next size that is a multiple of the requested size. |
1936 | if (static_cast<int>(Unmerges.size()) != NumMerge * WideSize) { |
1937 | Register UndefReg = MIRBuilder.buildUndef(Res: GCDTy).getReg(Idx: 0); |
1938 | for (int I = Unmerges.size(); I != NumMerge * WideSize; ++I) |
1939 | Unmerges.push_back(Elt: UndefReg); |
1940 | } |
1941 | |
1942 | const int PartsPerGCD = WideSize / GCD; |
1943 | |
1944 | // Build merges of each piece. |
1945 | ArrayRef<Register> Slicer(Unmerges); |
1946 | for (int I = 0; I != NumMerge; ++I, Slicer = Slicer.drop_front(N: PartsPerGCD)) { |
1947 | auto Merge = |
1948 | MIRBuilder.buildMergeLikeInstr(Res: WideTy, Ops: Slicer.take_front(N: PartsPerGCD)); |
1949 | NewMergeRegs.push_back(Elt: Merge.getReg(Idx: 0)); |
1950 | } |
1951 | |
1952 | // A truncate may be necessary if the requested type doesn't evenly divide the |
1953 | // original result type. |
1954 | if (DstTy.getSizeInBits() == WideDstTy.getSizeInBits()) { |
1955 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: NewMergeRegs); |
1956 | } else { |
1957 | auto FinalMerge = MIRBuilder.buildMergeLikeInstr(Res: WideDstTy, Ops: NewMergeRegs); |
1958 | MIRBuilder.buildTrunc(Res: DstReg, Op: FinalMerge.getReg(Idx: 0)); |
1959 | } |
1960 | |
1961 | MI.eraseFromParent(); |
1962 | return Legalized; |
1963 | } |
1964 | |
1965 | LegalizerHelper::LegalizeResult |
1966 | LegalizerHelper::widenScalarUnmergeValues(MachineInstr &MI, unsigned TypeIdx, |
1967 | LLT WideTy) { |
1968 | if (TypeIdx != 0) |
1969 | return UnableToLegalize; |
1970 | |
1971 | int NumDst = MI.getNumOperands() - 1; |
1972 | Register SrcReg = MI.getOperand(i: NumDst).getReg(); |
1973 | LLT SrcTy = MRI.getType(Reg: SrcReg); |
1974 | if (SrcTy.isVector()) |
1975 | return UnableToLegalize; |
1976 | |
1977 | Register Dst0Reg = MI.getOperand(i: 0).getReg(); |
1978 | LLT DstTy = MRI.getType(Reg: Dst0Reg); |
1979 | if (!DstTy.isScalar()) |
1980 | return UnableToLegalize; |
1981 | |
1982 | if (WideTy.getSizeInBits() >= SrcTy.getSizeInBits()) { |
1983 | if (SrcTy.isPointer()) { |
1984 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
1985 | if (DL.isNonIntegralAddressSpace(AddrSpace: SrcTy.getAddressSpace())) { |
1986 | LLVM_DEBUG( |
1987 | dbgs() << "Not casting non-integral address space integer\n" ); |
1988 | return UnableToLegalize; |
1989 | } |
1990 | |
1991 | SrcTy = LLT::scalar(SizeInBits: SrcTy.getSizeInBits()); |
1992 | SrcReg = MIRBuilder.buildPtrToInt(Dst: SrcTy, Src: SrcReg).getReg(Idx: 0); |
1993 | } |
1994 | |
1995 | // Widen SrcTy to WideTy. This does not affect the result, but since the |
1996 | // user requested this size, it is probably better handled than SrcTy and |
1997 | // should reduce the total number of legalization artifacts. |
1998 | if (WideTy.getSizeInBits() > SrcTy.getSizeInBits()) { |
1999 | SrcTy = WideTy; |
2000 | SrcReg = MIRBuilder.buildAnyExt(Res: WideTy, Op: SrcReg).getReg(Idx: 0); |
2001 | } |
2002 | |
2003 | // Theres no unmerge type to target. Directly extract the bits from the |
2004 | // source type |
2005 | unsigned DstSize = DstTy.getSizeInBits(); |
2006 | |
2007 | MIRBuilder.buildTrunc(Res: Dst0Reg, Op: SrcReg); |
2008 | for (int I = 1; I != NumDst; ++I) { |
2009 | auto ShiftAmt = MIRBuilder.buildConstant(Res: SrcTy, Val: DstSize * I); |
2010 | auto Shr = MIRBuilder.buildLShr(Dst: SrcTy, Src0: SrcReg, Src1: ShiftAmt); |
2011 | MIRBuilder.buildTrunc(Res: MI.getOperand(i: I), Op: Shr); |
2012 | } |
2013 | |
2014 | MI.eraseFromParent(); |
2015 | return Legalized; |
2016 | } |
2017 | |
2018 | // Extend the source to a wider type. |
2019 | LLT LCMTy = getLCMType(OrigTy: SrcTy, TargetTy: WideTy); |
2020 | |
2021 | Register WideSrc = SrcReg; |
2022 | if (LCMTy.getSizeInBits() != SrcTy.getSizeInBits()) { |
2023 | // TODO: If this is an integral address space, cast to integer and anyext. |
2024 | if (SrcTy.isPointer()) { |
2025 | LLVM_DEBUG(dbgs() << "Widening pointer source types not implemented\n" ); |
2026 | return UnableToLegalize; |
2027 | } |
2028 | |
2029 | WideSrc = MIRBuilder.buildAnyExt(Res: LCMTy, Op: WideSrc).getReg(Idx: 0); |
2030 | } |
2031 | |
2032 | auto Unmerge = MIRBuilder.buildUnmerge(Res: WideTy, Op: WideSrc); |
2033 | |
2034 | // Create a sequence of unmerges and merges to the original results. Since we |
2035 | // may have widened the source, we will need to pad the results with dead defs |
2036 | // to cover the source register. |
2037 | // e.g. widen s48 to s64: |
2038 | // %1:_(s48), %2:_(s48) = G_UNMERGE_VALUES %0:_(s96) |
2039 | // |
2040 | // => |
2041 | // %4:_(s192) = G_ANYEXT %0:_(s96) |
2042 | // %5:_(s64), %6, %7 = G_UNMERGE_VALUES %4 ; Requested unmerge |
2043 | // ; unpack to GCD type, with extra dead defs |
2044 | // %8:_(s16), %9, %10, %11 = G_UNMERGE_VALUES %5:_(s64) |
2045 | // %12:_(s16), %13, dead %14, dead %15 = G_UNMERGE_VALUES %6:_(s64) |
2046 | // dead %16:_(s16), dead %17, dead %18, dead %18 = G_UNMERGE_VALUES %7:_(s64) |
2047 | // %1:_(s48) = G_MERGE_VALUES %8:_(s16), %9, %10 ; Remerge to destination |
2048 | // %2:_(s48) = G_MERGE_VALUES %11:_(s16), %12, %13 ; Remerge to destination |
2049 | const LLT GCDTy = getGCDType(OrigTy: WideTy, TargetTy: DstTy); |
2050 | const int NumUnmerge = Unmerge->getNumOperands() - 1; |
2051 | const int PartsPerRemerge = DstTy.getSizeInBits() / GCDTy.getSizeInBits(); |
2052 | |
2053 | // Directly unmerge to the destination without going through a GCD type |
2054 | // if possible |
2055 | if (PartsPerRemerge == 1) { |
2056 | const int PartsPerUnmerge = WideTy.getSizeInBits() / DstTy.getSizeInBits(); |
2057 | |
2058 | for (int I = 0; I != NumUnmerge; ++I) { |
2059 | auto MIB = MIRBuilder.buildInstr(Opcode: TargetOpcode::G_UNMERGE_VALUES); |
2060 | |
2061 | for (int J = 0; J != PartsPerUnmerge; ++J) { |
2062 | int Idx = I * PartsPerUnmerge + J; |
2063 | if (Idx < NumDst) |
2064 | MIB.addDef(RegNo: MI.getOperand(i: Idx).getReg()); |
2065 | else { |
2066 | // Create dead def for excess components. |
2067 | MIB.addDef(RegNo: MRI.createGenericVirtualRegister(Ty: DstTy)); |
2068 | } |
2069 | } |
2070 | |
2071 | MIB.addUse(RegNo: Unmerge.getReg(Idx: I)); |
2072 | } |
2073 | } else { |
2074 | SmallVector<Register, 16> Parts; |
2075 | for (int J = 0; J != NumUnmerge; ++J) |
2076 | extractGCDType(Parts, GCDTy, SrcReg: Unmerge.getReg(Idx: J)); |
2077 | |
2078 | SmallVector<Register, 8> RemergeParts; |
2079 | for (int I = 0; I != NumDst; ++I) { |
2080 | for (int J = 0; J < PartsPerRemerge; ++J) { |
2081 | const int Idx = I * PartsPerRemerge + J; |
2082 | RemergeParts.emplace_back(Args&: Parts[Idx]); |
2083 | } |
2084 | |
2085 | MIRBuilder.buildMergeLikeInstr(Res: MI.getOperand(i: I).getReg(), Ops: RemergeParts); |
2086 | RemergeParts.clear(); |
2087 | } |
2088 | } |
2089 | |
2090 | MI.eraseFromParent(); |
2091 | return Legalized; |
2092 | } |
2093 | |
2094 | LegalizerHelper::LegalizeResult |
2095 | LegalizerHelper::(MachineInstr &MI, unsigned TypeIdx, |
2096 | LLT WideTy) { |
2097 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
2098 | unsigned Offset = MI.getOperand(i: 2).getImm(); |
2099 | |
2100 | if (TypeIdx == 0) { |
2101 | if (SrcTy.isVector() || DstTy.isVector()) |
2102 | return UnableToLegalize; |
2103 | |
2104 | SrcOp Src(SrcReg); |
2105 | if (SrcTy.isPointer()) { |
2106 | // Extracts from pointers can be handled only if they are really just |
2107 | // simple integers. |
2108 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
2109 | if (DL.isNonIntegralAddressSpace(AddrSpace: SrcTy.getAddressSpace())) |
2110 | return UnableToLegalize; |
2111 | |
2112 | LLT SrcAsIntTy = LLT::scalar(SizeInBits: SrcTy.getSizeInBits()); |
2113 | Src = MIRBuilder.buildPtrToInt(Dst: SrcAsIntTy, Src); |
2114 | SrcTy = SrcAsIntTy; |
2115 | } |
2116 | |
2117 | if (DstTy.isPointer()) |
2118 | return UnableToLegalize; |
2119 | |
2120 | if (Offset == 0) { |
2121 | // Avoid a shift in the degenerate case. |
2122 | MIRBuilder.buildTrunc(Res: DstReg, |
2123 | Op: MIRBuilder.buildAnyExtOrTrunc(Res: WideTy, Op: Src)); |
2124 | MI.eraseFromParent(); |
2125 | return Legalized; |
2126 | } |
2127 | |
2128 | // Do a shift in the source type. |
2129 | LLT ShiftTy = SrcTy; |
2130 | if (WideTy.getSizeInBits() > SrcTy.getSizeInBits()) { |
2131 | Src = MIRBuilder.buildAnyExt(Res: WideTy, Op: Src); |
2132 | ShiftTy = WideTy; |
2133 | } |
2134 | |
2135 | auto LShr = MIRBuilder.buildLShr( |
2136 | Dst: ShiftTy, Src0: Src, Src1: MIRBuilder.buildConstant(Res: ShiftTy, Val: Offset)); |
2137 | MIRBuilder.buildTrunc(Res: DstReg, Op: LShr); |
2138 | MI.eraseFromParent(); |
2139 | return Legalized; |
2140 | } |
2141 | |
2142 | if (SrcTy.isScalar()) { |
2143 | Observer.changingInstr(MI); |
2144 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2145 | Observer.changedInstr(MI); |
2146 | return Legalized; |
2147 | } |
2148 | |
2149 | if (!SrcTy.isVector()) |
2150 | return UnableToLegalize; |
2151 | |
2152 | if (DstTy != SrcTy.getElementType()) |
2153 | return UnableToLegalize; |
2154 | |
2155 | if (Offset % SrcTy.getScalarSizeInBits() != 0) |
2156 | return UnableToLegalize; |
2157 | |
2158 | Observer.changingInstr(MI); |
2159 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2160 | |
2161 | MI.getOperand(i: 2).setImm((WideTy.getSizeInBits() / SrcTy.getSizeInBits()) * |
2162 | Offset); |
2163 | widenScalarDst(MI, WideTy: WideTy.getScalarType(), OpIdx: 0); |
2164 | Observer.changedInstr(MI); |
2165 | return Legalized; |
2166 | } |
2167 | |
2168 | LegalizerHelper::LegalizeResult |
2169 | LegalizerHelper::widenScalarInsert(MachineInstr &MI, unsigned TypeIdx, |
2170 | LLT WideTy) { |
2171 | if (TypeIdx != 0 || WideTy.isVector()) |
2172 | return UnableToLegalize; |
2173 | Observer.changingInstr(MI); |
2174 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2175 | widenScalarDst(MI, WideTy); |
2176 | Observer.changedInstr(MI); |
2177 | return Legalized; |
2178 | } |
2179 | |
2180 | LegalizerHelper::LegalizeResult |
2181 | LegalizerHelper::widenScalarAddSubOverflow(MachineInstr &MI, unsigned TypeIdx, |
2182 | LLT WideTy) { |
2183 | unsigned Opcode; |
2184 | unsigned ExtOpcode; |
2185 | std::optional<Register> CarryIn; |
2186 | switch (MI.getOpcode()) { |
2187 | default: |
2188 | llvm_unreachable("Unexpected opcode!" ); |
2189 | case TargetOpcode::G_SADDO: |
2190 | Opcode = TargetOpcode::G_ADD; |
2191 | ExtOpcode = TargetOpcode::G_SEXT; |
2192 | break; |
2193 | case TargetOpcode::G_SSUBO: |
2194 | Opcode = TargetOpcode::G_SUB; |
2195 | ExtOpcode = TargetOpcode::G_SEXT; |
2196 | break; |
2197 | case TargetOpcode::G_UADDO: |
2198 | Opcode = TargetOpcode::G_ADD; |
2199 | ExtOpcode = TargetOpcode::G_ZEXT; |
2200 | break; |
2201 | case TargetOpcode::G_USUBO: |
2202 | Opcode = TargetOpcode::G_SUB; |
2203 | ExtOpcode = TargetOpcode::G_ZEXT; |
2204 | break; |
2205 | case TargetOpcode::G_SADDE: |
2206 | Opcode = TargetOpcode::G_UADDE; |
2207 | ExtOpcode = TargetOpcode::G_SEXT; |
2208 | CarryIn = MI.getOperand(i: 4).getReg(); |
2209 | break; |
2210 | case TargetOpcode::G_SSUBE: |
2211 | Opcode = TargetOpcode::G_USUBE; |
2212 | ExtOpcode = TargetOpcode::G_SEXT; |
2213 | CarryIn = MI.getOperand(i: 4).getReg(); |
2214 | break; |
2215 | case TargetOpcode::G_UADDE: |
2216 | Opcode = TargetOpcode::G_UADDE; |
2217 | ExtOpcode = TargetOpcode::G_ZEXT; |
2218 | CarryIn = MI.getOperand(i: 4).getReg(); |
2219 | break; |
2220 | case TargetOpcode::G_USUBE: |
2221 | Opcode = TargetOpcode::G_USUBE; |
2222 | ExtOpcode = TargetOpcode::G_ZEXT; |
2223 | CarryIn = MI.getOperand(i: 4).getReg(); |
2224 | break; |
2225 | } |
2226 | |
2227 | if (TypeIdx == 1) { |
2228 | unsigned BoolExtOp = MIRBuilder.getBoolExtOp(IsVec: WideTy.isVector(), IsFP: false); |
2229 | |
2230 | Observer.changingInstr(MI); |
2231 | if (CarryIn) |
2232 | widenScalarSrc(MI, WideTy, OpIdx: 4, ExtOpcode: BoolExtOp); |
2233 | widenScalarDst(MI, WideTy, OpIdx: 1); |
2234 | |
2235 | Observer.changedInstr(MI); |
2236 | return Legalized; |
2237 | } |
2238 | |
2239 | auto LHSExt = MIRBuilder.buildInstr(Opc: ExtOpcode, DstOps: {WideTy}, SrcOps: {MI.getOperand(i: 2)}); |
2240 | auto RHSExt = MIRBuilder.buildInstr(Opc: ExtOpcode, DstOps: {WideTy}, SrcOps: {MI.getOperand(i: 3)}); |
2241 | // Do the arithmetic in the larger type. |
2242 | Register NewOp; |
2243 | if (CarryIn) { |
2244 | LLT CarryOutTy = MRI.getType(Reg: MI.getOperand(i: 1).getReg()); |
2245 | NewOp = MIRBuilder |
2246 | .buildInstr(Opc: Opcode, DstOps: {WideTy, CarryOutTy}, |
2247 | SrcOps: {LHSExt, RHSExt, *CarryIn}) |
2248 | .getReg(Idx: 0); |
2249 | } else { |
2250 | NewOp = MIRBuilder.buildInstr(Opc: Opcode, DstOps: {WideTy}, SrcOps: {LHSExt, RHSExt}).getReg(Idx: 0); |
2251 | } |
2252 | LLT OrigTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
2253 | auto TruncOp = MIRBuilder.buildTrunc(Res: OrigTy, Op: NewOp); |
2254 | auto ExtOp = MIRBuilder.buildInstr(Opc: ExtOpcode, DstOps: {WideTy}, SrcOps: {TruncOp}); |
2255 | // There is no overflow if the ExtOp is the same as NewOp. |
2256 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: MI.getOperand(i: 1), Op0: NewOp, Op1: ExtOp); |
2257 | // Now trunc the NewOp to the original result. |
2258 | MIRBuilder.buildTrunc(Res: MI.getOperand(i: 0), Op: NewOp); |
2259 | MI.eraseFromParent(); |
2260 | return Legalized; |
2261 | } |
2262 | |
2263 | LegalizerHelper::LegalizeResult |
2264 | LegalizerHelper::widenScalarAddSubShlSat(MachineInstr &MI, unsigned TypeIdx, |
2265 | LLT WideTy) { |
2266 | bool IsSigned = MI.getOpcode() == TargetOpcode::G_SADDSAT || |
2267 | MI.getOpcode() == TargetOpcode::G_SSUBSAT || |
2268 | MI.getOpcode() == TargetOpcode::G_SSHLSAT; |
2269 | bool IsShift = MI.getOpcode() == TargetOpcode::G_SSHLSAT || |
2270 | MI.getOpcode() == TargetOpcode::G_USHLSAT; |
2271 | // We can convert this to: |
2272 | // 1. Any extend iN to iM |
2273 | // 2. SHL by M-N |
2274 | // 3. [US][ADD|SUB|SHL]SAT |
2275 | // 4. L/ASHR by M-N |
2276 | // |
2277 | // It may be more efficient to lower this to a min and a max operation in |
2278 | // the higher precision arithmetic if the promoted operation isn't legal, |
2279 | // but this decision is up to the target's lowering request. |
2280 | Register DstReg = MI.getOperand(i: 0).getReg(); |
2281 | |
2282 | unsigned NewBits = WideTy.getScalarSizeInBits(); |
2283 | unsigned SHLAmount = NewBits - MRI.getType(Reg: DstReg).getScalarSizeInBits(); |
2284 | |
2285 | // Shifts must zero-extend the RHS to preserve the unsigned quantity, and |
2286 | // must not left shift the RHS to preserve the shift amount. |
2287 | auto LHS = MIRBuilder.buildAnyExt(Res: WideTy, Op: MI.getOperand(i: 1)); |
2288 | auto RHS = IsShift ? MIRBuilder.buildZExt(Res: WideTy, Op: MI.getOperand(i: 2)) |
2289 | : MIRBuilder.buildAnyExt(Res: WideTy, Op: MI.getOperand(i: 2)); |
2290 | auto ShiftK = MIRBuilder.buildConstant(Res: WideTy, Val: SHLAmount); |
2291 | auto ShiftL = MIRBuilder.buildShl(Dst: WideTy, Src0: LHS, Src1: ShiftK); |
2292 | auto ShiftR = IsShift ? RHS : MIRBuilder.buildShl(Dst: WideTy, Src0: RHS, Src1: ShiftK); |
2293 | |
2294 | auto WideInst = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {WideTy}, |
2295 | SrcOps: {ShiftL, ShiftR}, Flags: MI.getFlags()); |
2296 | |
2297 | // Use a shift that will preserve the number of sign bits when the trunc is |
2298 | // folded away. |
2299 | auto Result = IsSigned ? MIRBuilder.buildAShr(Dst: WideTy, Src0: WideInst, Src1: ShiftK) |
2300 | : MIRBuilder.buildLShr(Dst: WideTy, Src0: WideInst, Src1: ShiftK); |
2301 | |
2302 | MIRBuilder.buildTrunc(Res: DstReg, Op: Result); |
2303 | MI.eraseFromParent(); |
2304 | return Legalized; |
2305 | } |
2306 | |
2307 | LegalizerHelper::LegalizeResult |
2308 | LegalizerHelper::widenScalarMulo(MachineInstr &MI, unsigned TypeIdx, |
2309 | LLT WideTy) { |
2310 | if (TypeIdx == 1) { |
2311 | Observer.changingInstr(MI); |
2312 | widenScalarDst(MI, WideTy, OpIdx: 1); |
2313 | Observer.changedInstr(MI); |
2314 | return Legalized; |
2315 | } |
2316 | |
2317 | bool IsSigned = MI.getOpcode() == TargetOpcode::G_SMULO; |
2318 | auto [Result, OriginalOverflow, LHS, RHS] = MI.getFirst4Regs(); |
2319 | LLT SrcTy = MRI.getType(Reg: LHS); |
2320 | LLT OverflowTy = MRI.getType(Reg: OriginalOverflow); |
2321 | unsigned SrcBitWidth = SrcTy.getScalarSizeInBits(); |
2322 | |
2323 | // To determine if the result overflowed in the larger type, we extend the |
2324 | // input to the larger type, do the multiply (checking if it overflows), |
2325 | // then also check the high bits of the result to see if overflow happened |
2326 | // there. |
2327 | unsigned ExtOp = IsSigned ? TargetOpcode::G_SEXT : TargetOpcode::G_ZEXT; |
2328 | auto LeftOperand = MIRBuilder.buildInstr(Opc: ExtOp, DstOps: {WideTy}, SrcOps: {LHS}); |
2329 | auto RightOperand = MIRBuilder.buildInstr(Opc: ExtOp, DstOps: {WideTy}, SrcOps: {RHS}); |
2330 | |
2331 | // Multiplication cannot overflow if the WideTy is >= 2 * original width, |
2332 | // so we don't need to check the overflow result of larger type Mulo. |
2333 | bool WideMulCanOverflow = WideTy.getScalarSizeInBits() < 2 * SrcBitWidth; |
2334 | |
2335 | unsigned MulOpc = |
2336 | WideMulCanOverflow ? MI.getOpcode() : (unsigned)TargetOpcode::G_MUL; |
2337 | |
2338 | MachineInstrBuilder Mulo; |
2339 | if (WideMulCanOverflow) |
2340 | Mulo = MIRBuilder.buildInstr(Opc: MulOpc, DstOps: {WideTy, OverflowTy}, |
2341 | SrcOps: {LeftOperand, RightOperand}); |
2342 | else |
2343 | Mulo = MIRBuilder.buildInstr(Opc: MulOpc, DstOps: {WideTy}, SrcOps: {LeftOperand, RightOperand}); |
2344 | |
2345 | auto Mul = Mulo->getOperand(i: 0); |
2346 | MIRBuilder.buildTrunc(Res: Result, Op: Mul); |
2347 | |
2348 | MachineInstrBuilder ExtResult; |
2349 | // Overflow occurred if it occurred in the larger type, or if the high part |
2350 | // of the result does not zero/sign-extend the low part. Check this second |
2351 | // possibility first. |
2352 | if (IsSigned) { |
2353 | // For signed, overflow occurred when the high part does not sign-extend |
2354 | // the low part. |
2355 | ExtResult = MIRBuilder.buildSExtInReg(Res: WideTy, Op: Mul, ImmOp: SrcBitWidth); |
2356 | } else { |
2357 | // Unsigned overflow occurred when the high part does not zero-extend the |
2358 | // low part. |
2359 | ExtResult = MIRBuilder.buildZExtInReg(Res: WideTy, Op: Mul, ImmOp: SrcBitWidth); |
2360 | } |
2361 | |
2362 | if (WideMulCanOverflow) { |
2363 | auto Overflow = |
2364 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: OverflowTy, Op0: Mul, Op1: ExtResult); |
2365 | // Finally check if the multiplication in the larger type itself overflowed. |
2366 | MIRBuilder.buildOr(Dst: OriginalOverflow, Src0: Mulo->getOperand(i: 1), Src1: Overflow); |
2367 | } else { |
2368 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: OriginalOverflow, Op0: Mul, Op1: ExtResult); |
2369 | } |
2370 | MI.eraseFromParent(); |
2371 | return Legalized; |
2372 | } |
2373 | |
2374 | LegalizerHelper::LegalizeResult |
2375 | LegalizerHelper::widenScalar(MachineInstr &MI, unsigned TypeIdx, LLT WideTy) { |
2376 | switch (MI.getOpcode()) { |
2377 | default: |
2378 | return UnableToLegalize; |
2379 | case TargetOpcode::G_ATOMICRMW_XCHG: |
2380 | case TargetOpcode::G_ATOMICRMW_ADD: |
2381 | case TargetOpcode::G_ATOMICRMW_SUB: |
2382 | case TargetOpcode::G_ATOMICRMW_AND: |
2383 | case TargetOpcode::G_ATOMICRMW_OR: |
2384 | case TargetOpcode::G_ATOMICRMW_XOR: |
2385 | case TargetOpcode::G_ATOMICRMW_MIN: |
2386 | case TargetOpcode::G_ATOMICRMW_MAX: |
2387 | case TargetOpcode::G_ATOMICRMW_UMIN: |
2388 | case TargetOpcode::G_ATOMICRMW_UMAX: |
2389 | assert(TypeIdx == 0 && "atomicrmw with second scalar type" ); |
2390 | Observer.changingInstr(MI); |
2391 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ANYEXT); |
2392 | widenScalarDst(MI, WideTy, OpIdx: 0); |
2393 | Observer.changedInstr(MI); |
2394 | return Legalized; |
2395 | case TargetOpcode::G_ATOMIC_CMPXCHG: |
2396 | assert(TypeIdx == 0 && "G_ATOMIC_CMPXCHG with second scalar type" ); |
2397 | Observer.changingInstr(MI); |
2398 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ANYEXT); |
2399 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode: TargetOpcode::G_ANYEXT); |
2400 | widenScalarDst(MI, WideTy, OpIdx: 0); |
2401 | Observer.changedInstr(MI); |
2402 | return Legalized; |
2403 | case TargetOpcode::G_ATOMIC_CMPXCHG_WITH_SUCCESS: |
2404 | if (TypeIdx == 0) { |
2405 | Observer.changingInstr(MI); |
2406 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode: TargetOpcode::G_ANYEXT); |
2407 | widenScalarSrc(MI, WideTy, OpIdx: 4, ExtOpcode: TargetOpcode::G_ANYEXT); |
2408 | widenScalarDst(MI, WideTy, OpIdx: 0); |
2409 | Observer.changedInstr(MI); |
2410 | return Legalized; |
2411 | } |
2412 | assert(TypeIdx == 1 && |
2413 | "G_ATOMIC_CMPXCHG_WITH_SUCCESS with third scalar type" ); |
2414 | Observer.changingInstr(MI); |
2415 | widenScalarDst(MI, WideTy, OpIdx: 1); |
2416 | Observer.changedInstr(MI); |
2417 | return Legalized; |
2418 | case TargetOpcode::G_EXTRACT: |
2419 | return widenScalarExtract(MI, TypeIdx, WideTy); |
2420 | case TargetOpcode::G_INSERT: |
2421 | return widenScalarInsert(MI, TypeIdx, WideTy); |
2422 | case TargetOpcode::G_MERGE_VALUES: |
2423 | return widenScalarMergeValues(MI, TypeIdx, WideTy); |
2424 | case TargetOpcode::G_UNMERGE_VALUES: |
2425 | return widenScalarUnmergeValues(MI, TypeIdx, WideTy); |
2426 | case TargetOpcode::G_SADDO: |
2427 | case TargetOpcode::G_SSUBO: |
2428 | case TargetOpcode::G_UADDO: |
2429 | case TargetOpcode::G_USUBO: |
2430 | case TargetOpcode::G_SADDE: |
2431 | case TargetOpcode::G_SSUBE: |
2432 | case TargetOpcode::G_UADDE: |
2433 | case TargetOpcode::G_USUBE: |
2434 | return widenScalarAddSubOverflow(MI, TypeIdx, WideTy); |
2435 | case TargetOpcode::G_UMULO: |
2436 | case TargetOpcode::G_SMULO: |
2437 | return widenScalarMulo(MI, TypeIdx, WideTy); |
2438 | case TargetOpcode::G_SADDSAT: |
2439 | case TargetOpcode::G_SSUBSAT: |
2440 | case TargetOpcode::G_SSHLSAT: |
2441 | case TargetOpcode::G_UADDSAT: |
2442 | case TargetOpcode::G_USUBSAT: |
2443 | case TargetOpcode::G_USHLSAT: |
2444 | return widenScalarAddSubShlSat(MI, TypeIdx, WideTy); |
2445 | case TargetOpcode::G_CTTZ: |
2446 | case TargetOpcode::G_CTTZ_ZERO_UNDEF: |
2447 | case TargetOpcode::G_CTLZ: |
2448 | case TargetOpcode::G_CTLZ_ZERO_UNDEF: |
2449 | case TargetOpcode::G_CTPOP: { |
2450 | if (TypeIdx == 0) { |
2451 | Observer.changingInstr(MI); |
2452 | widenScalarDst(MI, WideTy, OpIdx: 0); |
2453 | Observer.changedInstr(MI); |
2454 | return Legalized; |
2455 | } |
2456 | |
2457 | Register SrcReg = MI.getOperand(i: 1).getReg(); |
2458 | |
2459 | // First extend the input. |
2460 | unsigned ExtOpc = MI.getOpcode() == TargetOpcode::G_CTTZ || |
2461 | MI.getOpcode() == TargetOpcode::G_CTTZ_ZERO_UNDEF |
2462 | ? TargetOpcode::G_ANYEXT |
2463 | : TargetOpcode::G_ZEXT; |
2464 | auto MIBSrc = MIRBuilder.buildInstr(Opc: ExtOpc, DstOps: {WideTy}, SrcOps: {SrcReg}); |
2465 | LLT CurTy = MRI.getType(Reg: SrcReg); |
2466 | unsigned NewOpc = MI.getOpcode(); |
2467 | if (NewOpc == TargetOpcode::G_CTTZ) { |
2468 | // The count is the same in the larger type except if the original |
2469 | // value was zero. This can be handled by setting the bit just off |
2470 | // the top of the original type. |
2471 | auto TopBit = |
2472 | APInt::getOneBitSet(numBits: WideTy.getSizeInBits(), BitNo: CurTy.getSizeInBits()); |
2473 | MIBSrc = MIRBuilder.buildOr( |
2474 | Dst: WideTy, Src0: MIBSrc, Src1: MIRBuilder.buildConstant(Res: WideTy, Val: TopBit)); |
2475 | // Now we know the operand is non-zero, use the more relaxed opcode. |
2476 | NewOpc = TargetOpcode::G_CTTZ_ZERO_UNDEF; |
2477 | } |
2478 | |
2479 | unsigned SizeDiff = WideTy.getSizeInBits() - CurTy.getSizeInBits(); |
2480 | |
2481 | if (MI.getOpcode() == TargetOpcode::G_CTLZ_ZERO_UNDEF) { |
2482 | // An optimization where the result is the CTLZ after the left shift by |
2483 | // (Difference in widety and current ty), that is, |
2484 | // MIBSrc = MIBSrc << (sizeinbits(WideTy) - sizeinbits(CurTy)) |
2485 | // Result = ctlz MIBSrc |
2486 | MIBSrc = MIRBuilder.buildShl(Dst: WideTy, Src0: MIBSrc, |
2487 | Src1: MIRBuilder.buildConstant(Res: WideTy, Val: SizeDiff)); |
2488 | } |
2489 | |
2490 | // Perform the operation at the larger size. |
2491 | auto MIBNewOp = MIRBuilder.buildInstr(Opc: NewOpc, DstOps: {WideTy}, SrcOps: {MIBSrc}); |
2492 | // This is already the correct result for CTPOP and CTTZs |
2493 | if (MI.getOpcode() == TargetOpcode::G_CTLZ) { |
2494 | // The correct result is NewOp - (Difference in widety and current ty). |
2495 | MIBNewOp = MIRBuilder.buildSub( |
2496 | Dst: WideTy, Src0: MIBNewOp, Src1: MIRBuilder.buildConstant(Res: WideTy, Val: SizeDiff)); |
2497 | } |
2498 | |
2499 | MIRBuilder.buildZExtOrTrunc(Res: MI.getOperand(i: 0), Op: MIBNewOp); |
2500 | MI.eraseFromParent(); |
2501 | return Legalized; |
2502 | } |
2503 | case TargetOpcode::G_BSWAP: { |
2504 | Observer.changingInstr(MI); |
2505 | Register DstReg = MI.getOperand(i: 0).getReg(); |
2506 | |
2507 | Register ShrReg = MRI.createGenericVirtualRegister(Ty: WideTy); |
2508 | Register DstExt = MRI.createGenericVirtualRegister(Ty: WideTy); |
2509 | Register ShiftAmtReg = MRI.createGenericVirtualRegister(Ty: WideTy); |
2510 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2511 | |
2512 | MI.getOperand(i: 0).setReg(DstExt); |
2513 | |
2514 | MIRBuilder.setInsertPt(MBB&: MIRBuilder.getMBB(), II: ++MIRBuilder.getInsertPt()); |
2515 | |
2516 | LLT Ty = MRI.getType(Reg: DstReg); |
2517 | unsigned DiffBits = WideTy.getScalarSizeInBits() - Ty.getScalarSizeInBits(); |
2518 | MIRBuilder.buildConstant(Res: ShiftAmtReg, Val: DiffBits); |
2519 | MIRBuilder.buildLShr(Dst: ShrReg, Src0: DstExt, Src1: ShiftAmtReg); |
2520 | |
2521 | MIRBuilder.buildTrunc(Res: DstReg, Op: ShrReg); |
2522 | Observer.changedInstr(MI); |
2523 | return Legalized; |
2524 | } |
2525 | case TargetOpcode::G_BITREVERSE: { |
2526 | Observer.changingInstr(MI); |
2527 | |
2528 | Register DstReg = MI.getOperand(i: 0).getReg(); |
2529 | LLT Ty = MRI.getType(Reg: DstReg); |
2530 | unsigned DiffBits = WideTy.getScalarSizeInBits() - Ty.getScalarSizeInBits(); |
2531 | |
2532 | Register DstExt = MRI.createGenericVirtualRegister(Ty: WideTy); |
2533 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2534 | MI.getOperand(i: 0).setReg(DstExt); |
2535 | MIRBuilder.setInsertPt(MBB&: MIRBuilder.getMBB(), II: ++MIRBuilder.getInsertPt()); |
2536 | |
2537 | auto ShiftAmt = MIRBuilder.buildConstant(Res: WideTy, Val: DiffBits); |
2538 | auto Shift = MIRBuilder.buildLShr(Dst: WideTy, Src0: DstExt, Src1: ShiftAmt); |
2539 | MIRBuilder.buildTrunc(Res: DstReg, Op: Shift); |
2540 | Observer.changedInstr(MI); |
2541 | return Legalized; |
2542 | } |
2543 | case TargetOpcode::G_FREEZE: |
2544 | case TargetOpcode::G_CONSTANT_FOLD_BARRIER: |
2545 | Observer.changingInstr(MI); |
2546 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2547 | widenScalarDst(MI, WideTy); |
2548 | Observer.changedInstr(MI); |
2549 | return Legalized; |
2550 | |
2551 | case TargetOpcode::G_ABS: |
2552 | Observer.changingInstr(MI); |
2553 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_SEXT); |
2554 | widenScalarDst(MI, WideTy); |
2555 | Observer.changedInstr(MI); |
2556 | return Legalized; |
2557 | |
2558 | case TargetOpcode::G_ADD: |
2559 | case TargetOpcode::G_AND: |
2560 | case TargetOpcode::G_MUL: |
2561 | case TargetOpcode::G_OR: |
2562 | case TargetOpcode::G_XOR: |
2563 | case TargetOpcode::G_SUB: |
2564 | case TargetOpcode::G_SHUFFLE_VECTOR: |
2565 | // Perform operation at larger width (any extension is fines here, high bits |
2566 | // don't affect the result) and then truncate the result back to the |
2567 | // original type. |
2568 | Observer.changingInstr(MI); |
2569 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2570 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ANYEXT); |
2571 | widenScalarDst(MI, WideTy); |
2572 | Observer.changedInstr(MI); |
2573 | return Legalized; |
2574 | |
2575 | case TargetOpcode::G_SBFX: |
2576 | case TargetOpcode::G_UBFX: |
2577 | Observer.changingInstr(MI); |
2578 | |
2579 | if (TypeIdx == 0) { |
2580 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2581 | widenScalarDst(MI, WideTy); |
2582 | } else { |
2583 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ZEXT); |
2584 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode: TargetOpcode::G_ZEXT); |
2585 | } |
2586 | |
2587 | Observer.changedInstr(MI); |
2588 | return Legalized; |
2589 | |
2590 | case TargetOpcode::G_SHL: |
2591 | Observer.changingInstr(MI); |
2592 | |
2593 | if (TypeIdx == 0) { |
2594 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2595 | widenScalarDst(MI, WideTy); |
2596 | } else { |
2597 | assert(TypeIdx == 1); |
2598 | // The "number of bits to shift" operand must preserve its value as an |
2599 | // unsigned integer: |
2600 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ZEXT); |
2601 | } |
2602 | |
2603 | Observer.changedInstr(MI); |
2604 | return Legalized; |
2605 | |
2606 | case TargetOpcode::G_ROTR: |
2607 | case TargetOpcode::G_ROTL: |
2608 | if (TypeIdx != 1) |
2609 | return UnableToLegalize; |
2610 | |
2611 | Observer.changingInstr(MI); |
2612 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ZEXT); |
2613 | Observer.changedInstr(MI); |
2614 | return Legalized; |
2615 | |
2616 | case TargetOpcode::G_SDIV: |
2617 | case TargetOpcode::G_SREM: |
2618 | case TargetOpcode::G_SMIN: |
2619 | case TargetOpcode::G_SMAX: |
2620 | Observer.changingInstr(MI); |
2621 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_SEXT); |
2622 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_SEXT); |
2623 | widenScalarDst(MI, WideTy); |
2624 | Observer.changedInstr(MI); |
2625 | return Legalized; |
2626 | |
2627 | case TargetOpcode::G_SDIVREM: |
2628 | Observer.changingInstr(MI); |
2629 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_SEXT); |
2630 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode: TargetOpcode::G_SEXT); |
2631 | widenScalarDst(MI, WideTy); |
2632 | widenScalarDst(MI, WideTy, OpIdx: 1); |
2633 | Observer.changedInstr(MI); |
2634 | return Legalized; |
2635 | |
2636 | case TargetOpcode::G_ASHR: |
2637 | case TargetOpcode::G_LSHR: |
2638 | Observer.changingInstr(MI); |
2639 | |
2640 | if (TypeIdx == 0) { |
2641 | unsigned CvtOp = MI.getOpcode() == TargetOpcode::G_ASHR ? |
2642 | TargetOpcode::G_SEXT : TargetOpcode::G_ZEXT; |
2643 | |
2644 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: CvtOp); |
2645 | widenScalarDst(MI, WideTy); |
2646 | } else { |
2647 | assert(TypeIdx == 1); |
2648 | // The "number of bits to shift" operand must preserve its value as an |
2649 | // unsigned integer: |
2650 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ZEXT); |
2651 | } |
2652 | |
2653 | Observer.changedInstr(MI); |
2654 | return Legalized; |
2655 | case TargetOpcode::G_UDIV: |
2656 | case TargetOpcode::G_UREM: |
2657 | case TargetOpcode::G_UMIN: |
2658 | case TargetOpcode::G_UMAX: |
2659 | Observer.changingInstr(MI); |
2660 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ZEXT); |
2661 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ZEXT); |
2662 | widenScalarDst(MI, WideTy); |
2663 | Observer.changedInstr(MI); |
2664 | return Legalized; |
2665 | |
2666 | case TargetOpcode::G_UDIVREM: |
2667 | Observer.changingInstr(MI); |
2668 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ZEXT); |
2669 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode: TargetOpcode::G_ZEXT); |
2670 | widenScalarDst(MI, WideTy); |
2671 | widenScalarDst(MI, WideTy, OpIdx: 1); |
2672 | Observer.changedInstr(MI); |
2673 | return Legalized; |
2674 | |
2675 | case TargetOpcode::G_SELECT: |
2676 | Observer.changingInstr(MI); |
2677 | if (TypeIdx == 0) { |
2678 | // Perform operation at larger width (any extension is fine here, high |
2679 | // bits don't affect the result) and then truncate the result back to the |
2680 | // original type. |
2681 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ANYEXT); |
2682 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode: TargetOpcode::G_ANYEXT); |
2683 | widenScalarDst(MI, WideTy); |
2684 | } else { |
2685 | bool IsVec = MRI.getType(Reg: MI.getOperand(i: 1).getReg()).isVector(); |
2686 | // Explicit extension is required here since high bits affect the result. |
2687 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: MIRBuilder.getBoolExtOp(IsVec, IsFP: false)); |
2688 | } |
2689 | Observer.changedInstr(MI); |
2690 | return Legalized; |
2691 | |
2692 | case TargetOpcode::G_FPTOSI: |
2693 | case TargetOpcode::G_FPTOUI: |
2694 | case TargetOpcode::G_INTRINSIC_LRINT: |
2695 | case TargetOpcode::G_INTRINSIC_LLRINT: |
2696 | case TargetOpcode::G_IS_FPCLASS: |
2697 | Observer.changingInstr(MI); |
2698 | |
2699 | if (TypeIdx == 0) |
2700 | widenScalarDst(MI, WideTy); |
2701 | else |
2702 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_FPEXT); |
2703 | |
2704 | Observer.changedInstr(MI); |
2705 | return Legalized; |
2706 | case TargetOpcode::G_SITOFP: |
2707 | Observer.changingInstr(MI); |
2708 | |
2709 | if (TypeIdx == 0) |
2710 | widenScalarDst(MI, WideTy, OpIdx: 0, TruncOpcode: TargetOpcode::G_FPTRUNC); |
2711 | else |
2712 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_SEXT); |
2713 | |
2714 | Observer.changedInstr(MI); |
2715 | return Legalized; |
2716 | case TargetOpcode::G_UITOFP: |
2717 | Observer.changingInstr(MI); |
2718 | |
2719 | if (TypeIdx == 0) |
2720 | widenScalarDst(MI, WideTy, OpIdx: 0, TruncOpcode: TargetOpcode::G_FPTRUNC); |
2721 | else |
2722 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ZEXT); |
2723 | |
2724 | Observer.changedInstr(MI); |
2725 | return Legalized; |
2726 | case TargetOpcode::G_LOAD: |
2727 | case TargetOpcode::G_SEXTLOAD: |
2728 | case TargetOpcode::G_ZEXTLOAD: |
2729 | Observer.changingInstr(MI); |
2730 | widenScalarDst(MI, WideTy); |
2731 | Observer.changedInstr(MI); |
2732 | return Legalized; |
2733 | |
2734 | case TargetOpcode::G_STORE: { |
2735 | if (TypeIdx != 0) |
2736 | return UnableToLegalize; |
2737 | |
2738 | LLT Ty = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
2739 | if (!Ty.isScalar()) |
2740 | return UnableToLegalize; |
2741 | |
2742 | Observer.changingInstr(MI); |
2743 | |
2744 | unsigned ExtType = Ty.getScalarSizeInBits() == 1 ? |
2745 | TargetOpcode::G_ZEXT : TargetOpcode::G_ANYEXT; |
2746 | widenScalarSrc(MI, WideTy, OpIdx: 0, ExtOpcode: ExtType); |
2747 | |
2748 | Observer.changedInstr(MI); |
2749 | return Legalized; |
2750 | } |
2751 | case TargetOpcode::G_CONSTANT: { |
2752 | MachineOperand &SrcMO = MI.getOperand(i: 1); |
2753 | LLVMContext &Ctx = MIRBuilder.getMF().getFunction().getContext(); |
2754 | unsigned ExtOpc = LI.getExtOpcodeForWideningConstant( |
2755 | SmallTy: MRI.getType(Reg: MI.getOperand(i: 0).getReg())); |
2756 | assert((ExtOpc == TargetOpcode::G_ZEXT || ExtOpc == TargetOpcode::G_SEXT || |
2757 | ExtOpc == TargetOpcode::G_ANYEXT) && |
2758 | "Illegal Extend" ); |
2759 | const APInt &SrcVal = SrcMO.getCImm()->getValue(); |
2760 | const APInt &Val = (ExtOpc == TargetOpcode::G_SEXT) |
2761 | ? SrcVal.sext(width: WideTy.getSizeInBits()) |
2762 | : SrcVal.zext(width: WideTy.getSizeInBits()); |
2763 | Observer.changingInstr(MI); |
2764 | SrcMO.setCImm(ConstantInt::get(Context&: Ctx, V: Val)); |
2765 | |
2766 | widenScalarDst(MI, WideTy); |
2767 | Observer.changedInstr(MI); |
2768 | return Legalized; |
2769 | } |
2770 | case TargetOpcode::G_FCONSTANT: { |
2771 | // To avoid changing the bits of the constant due to extension to a larger |
2772 | // type and then using G_FPTRUNC, we simply convert to a G_CONSTANT. |
2773 | MachineOperand &SrcMO = MI.getOperand(i: 1); |
2774 | APInt Val = SrcMO.getFPImm()->getValueAPF().bitcastToAPInt(); |
2775 | MIRBuilder.setInstrAndDebugLoc(MI); |
2776 | auto IntCst = MIRBuilder.buildConstant(Res: MI.getOperand(i: 0).getReg(), Val); |
2777 | widenScalarDst(MI&: *IntCst, WideTy, OpIdx: 0, TruncOpcode: TargetOpcode::G_TRUNC); |
2778 | MI.eraseFromParent(); |
2779 | return Legalized; |
2780 | } |
2781 | case TargetOpcode::G_IMPLICIT_DEF: { |
2782 | Observer.changingInstr(MI); |
2783 | widenScalarDst(MI, WideTy); |
2784 | Observer.changedInstr(MI); |
2785 | return Legalized; |
2786 | } |
2787 | case TargetOpcode::G_BRCOND: |
2788 | Observer.changingInstr(MI); |
2789 | widenScalarSrc(MI, WideTy, OpIdx: 0, ExtOpcode: MIRBuilder.getBoolExtOp(IsVec: false, IsFP: false)); |
2790 | Observer.changedInstr(MI); |
2791 | return Legalized; |
2792 | |
2793 | case TargetOpcode::G_FCMP: |
2794 | Observer.changingInstr(MI); |
2795 | if (TypeIdx == 0) |
2796 | widenScalarDst(MI, WideTy); |
2797 | else { |
2798 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_FPEXT); |
2799 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode: TargetOpcode::G_FPEXT); |
2800 | } |
2801 | Observer.changedInstr(MI); |
2802 | return Legalized; |
2803 | |
2804 | case TargetOpcode::G_ICMP: |
2805 | Observer.changingInstr(MI); |
2806 | if (TypeIdx == 0) |
2807 | widenScalarDst(MI, WideTy); |
2808 | else { |
2809 | unsigned ExtOpcode = CmpInst::isSigned(predicate: static_cast<CmpInst::Predicate>( |
2810 | MI.getOperand(i: 1).getPredicate())) |
2811 | ? TargetOpcode::G_SEXT |
2812 | : TargetOpcode::G_ZEXT; |
2813 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode); |
2814 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode); |
2815 | } |
2816 | Observer.changedInstr(MI); |
2817 | return Legalized; |
2818 | |
2819 | case TargetOpcode::G_PTR_ADD: |
2820 | assert(TypeIdx == 1 && "unable to legalize pointer of G_PTR_ADD" ); |
2821 | Observer.changingInstr(MI); |
2822 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_SEXT); |
2823 | Observer.changedInstr(MI); |
2824 | return Legalized; |
2825 | |
2826 | case TargetOpcode::G_PHI: { |
2827 | assert(TypeIdx == 0 && "Expecting only Idx 0" ); |
2828 | |
2829 | Observer.changingInstr(MI); |
2830 | for (unsigned I = 1; I < MI.getNumOperands(); I += 2) { |
2831 | MachineBasicBlock &OpMBB = *MI.getOperand(i: I + 1).getMBB(); |
2832 | MIRBuilder.setInsertPt(MBB&: OpMBB, II: OpMBB.getFirstTerminatorForward()); |
2833 | widenScalarSrc(MI, WideTy, OpIdx: I, ExtOpcode: TargetOpcode::G_ANYEXT); |
2834 | } |
2835 | |
2836 | MachineBasicBlock &MBB = *MI.getParent(); |
2837 | MIRBuilder.setInsertPt(MBB, II: --MBB.getFirstNonPHI()); |
2838 | widenScalarDst(MI, WideTy); |
2839 | Observer.changedInstr(MI); |
2840 | return Legalized; |
2841 | } |
2842 | case TargetOpcode::G_EXTRACT_VECTOR_ELT: { |
2843 | if (TypeIdx == 0) { |
2844 | Register VecReg = MI.getOperand(i: 1).getReg(); |
2845 | LLT VecTy = MRI.getType(Reg: VecReg); |
2846 | Observer.changingInstr(MI); |
2847 | |
2848 | widenScalarSrc( |
2849 | MI, WideTy: LLT::vector(EC: VecTy.getElementCount(), ScalarSizeInBits: WideTy.getSizeInBits()), OpIdx: 1, |
2850 | ExtOpcode: TargetOpcode::G_ANYEXT); |
2851 | |
2852 | widenScalarDst(MI, WideTy, OpIdx: 0); |
2853 | Observer.changedInstr(MI); |
2854 | return Legalized; |
2855 | } |
2856 | |
2857 | if (TypeIdx != 2) |
2858 | return UnableToLegalize; |
2859 | Observer.changingInstr(MI); |
2860 | // TODO: Probably should be zext |
2861 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_SEXT); |
2862 | Observer.changedInstr(MI); |
2863 | return Legalized; |
2864 | } |
2865 | case TargetOpcode::G_INSERT_VECTOR_ELT: { |
2866 | if (TypeIdx == 0) { |
2867 | Observer.changingInstr(MI); |
2868 | const LLT WideEltTy = WideTy.getElementType(); |
2869 | |
2870 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2871 | widenScalarSrc(MI, WideTy: WideEltTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ANYEXT); |
2872 | widenScalarDst(MI, WideTy, OpIdx: 0); |
2873 | Observer.changedInstr(MI); |
2874 | return Legalized; |
2875 | } |
2876 | |
2877 | if (TypeIdx == 1) { |
2878 | Observer.changingInstr(MI); |
2879 | |
2880 | Register VecReg = MI.getOperand(i: 1).getReg(); |
2881 | LLT VecTy = MRI.getType(Reg: VecReg); |
2882 | LLT WideVecTy = LLT::vector(EC: VecTy.getElementCount(), ScalarTy: WideTy); |
2883 | |
2884 | widenScalarSrc(MI, WideTy: WideVecTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
2885 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ANYEXT); |
2886 | widenScalarDst(MI, WideTy: WideVecTy, OpIdx: 0); |
2887 | Observer.changedInstr(MI); |
2888 | return Legalized; |
2889 | } |
2890 | |
2891 | if (TypeIdx == 2) { |
2892 | Observer.changingInstr(MI); |
2893 | // TODO: Probably should be zext |
2894 | widenScalarSrc(MI, WideTy, OpIdx: 3, ExtOpcode: TargetOpcode::G_SEXT); |
2895 | Observer.changedInstr(MI); |
2896 | return Legalized; |
2897 | } |
2898 | |
2899 | return UnableToLegalize; |
2900 | } |
2901 | case TargetOpcode::G_FADD: |
2902 | case TargetOpcode::G_FMUL: |
2903 | case TargetOpcode::G_FSUB: |
2904 | case TargetOpcode::G_FMA: |
2905 | case TargetOpcode::G_FMAD: |
2906 | case TargetOpcode::G_FNEG: |
2907 | case TargetOpcode::G_FABS: |
2908 | case TargetOpcode::G_FCANONICALIZE: |
2909 | case TargetOpcode::G_FMINNUM: |
2910 | case TargetOpcode::G_FMAXNUM: |
2911 | case TargetOpcode::G_FMINNUM_IEEE: |
2912 | case TargetOpcode::G_FMAXNUM_IEEE: |
2913 | case TargetOpcode::G_FMINIMUM: |
2914 | case TargetOpcode::G_FMAXIMUM: |
2915 | case TargetOpcode::G_FDIV: |
2916 | case TargetOpcode::G_FREM: |
2917 | case TargetOpcode::G_FCEIL: |
2918 | case TargetOpcode::G_FFLOOR: |
2919 | case TargetOpcode::G_FCOS: |
2920 | case TargetOpcode::G_FSIN: |
2921 | case TargetOpcode::G_FTAN: |
2922 | case TargetOpcode::G_FACOS: |
2923 | case TargetOpcode::G_FASIN: |
2924 | case TargetOpcode::G_FATAN: |
2925 | case TargetOpcode::G_FCOSH: |
2926 | case TargetOpcode::G_FSINH: |
2927 | case TargetOpcode::G_FTANH: |
2928 | case TargetOpcode::G_FLOG10: |
2929 | case TargetOpcode::G_FLOG: |
2930 | case TargetOpcode::G_FLOG2: |
2931 | case TargetOpcode::G_FRINT: |
2932 | case TargetOpcode::G_FNEARBYINT: |
2933 | case TargetOpcode::G_FSQRT: |
2934 | case TargetOpcode::G_FEXP: |
2935 | case TargetOpcode::G_FEXP2: |
2936 | case TargetOpcode::G_FEXP10: |
2937 | case TargetOpcode::G_FPOW: |
2938 | case TargetOpcode::G_INTRINSIC_TRUNC: |
2939 | case TargetOpcode::G_INTRINSIC_ROUND: |
2940 | case TargetOpcode::G_INTRINSIC_ROUNDEVEN: |
2941 | assert(TypeIdx == 0); |
2942 | Observer.changingInstr(MI); |
2943 | |
2944 | for (unsigned I = 1, E = MI.getNumOperands(); I != E; ++I) |
2945 | widenScalarSrc(MI, WideTy, OpIdx: I, ExtOpcode: TargetOpcode::G_FPEXT); |
2946 | |
2947 | widenScalarDst(MI, WideTy, OpIdx: 0, TruncOpcode: TargetOpcode::G_FPTRUNC); |
2948 | Observer.changedInstr(MI); |
2949 | return Legalized; |
2950 | case TargetOpcode::G_FPOWI: |
2951 | case TargetOpcode::G_FLDEXP: |
2952 | case TargetOpcode::G_STRICT_FLDEXP: { |
2953 | if (TypeIdx == 0) { |
2954 | if (MI.getOpcode() == TargetOpcode::G_STRICT_FLDEXP) |
2955 | return UnableToLegalize; |
2956 | |
2957 | Observer.changingInstr(MI); |
2958 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_FPEXT); |
2959 | widenScalarDst(MI, WideTy, OpIdx: 0, TruncOpcode: TargetOpcode::G_FPTRUNC); |
2960 | Observer.changedInstr(MI); |
2961 | return Legalized; |
2962 | } |
2963 | |
2964 | if (TypeIdx == 1) { |
2965 | // For some reason SelectionDAG tries to promote to a libcall without |
2966 | // actually changing the integer type for promotion. |
2967 | Observer.changingInstr(MI); |
2968 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_SEXT); |
2969 | Observer.changedInstr(MI); |
2970 | return Legalized; |
2971 | } |
2972 | |
2973 | return UnableToLegalize; |
2974 | } |
2975 | case TargetOpcode::G_FFREXP: { |
2976 | Observer.changingInstr(MI); |
2977 | |
2978 | if (TypeIdx == 0) { |
2979 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_FPEXT); |
2980 | widenScalarDst(MI, WideTy, OpIdx: 0, TruncOpcode: TargetOpcode::G_FPTRUNC); |
2981 | } else { |
2982 | widenScalarDst(MI, WideTy, OpIdx: 1); |
2983 | } |
2984 | |
2985 | Observer.changedInstr(MI); |
2986 | return Legalized; |
2987 | } |
2988 | case TargetOpcode::G_INTTOPTR: |
2989 | if (TypeIdx != 1) |
2990 | return UnableToLegalize; |
2991 | |
2992 | Observer.changingInstr(MI); |
2993 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ZEXT); |
2994 | Observer.changedInstr(MI); |
2995 | return Legalized; |
2996 | case TargetOpcode::G_PTRTOINT: |
2997 | if (TypeIdx != 0) |
2998 | return UnableToLegalize; |
2999 | |
3000 | Observer.changingInstr(MI); |
3001 | widenScalarDst(MI, WideTy, OpIdx: 0); |
3002 | Observer.changedInstr(MI); |
3003 | return Legalized; |
3004 | case TargetOpcode::G_BUILD_VECTOR: { |
3005 | Observer.changingInstr(MI); |
3006 | |
3007 | const LLT WideEltTy = TypeIdx == 1 ? WideTy : WideTy.getElementType(); |
3008 | for (int I = 1, E = MI.getNumOperands(); I != E; ++I) |
3009 | widenScalarSrc(MI, WideTy: WideEltTy, OpIdx: I, ExtOpcode: TargetOpcode::G_ANYEXT); |
3010 | |
3011 | // Avoid changing the result vector type if the source element type was |
3012 | // requested. |
3013 | if (TypeIdx == 1) { |
3014 | MI.setDesc(MIRBuilder.getTII().get(Opcode: TargetOpcode::G_BUILD_VECTOR_TRUNC)); |
3015 | } else { |
3016 | widenScalarDst(MI, WideTy, OpIdx: 0); |
3017 | } |
3018 | |
3019 | Observer.changedInstr(MI); |
3020 | return Legalized; |
3021 | } |
3022 | case TargetOpcode::G_SEXT_INREG: |
3023 | if (TypeIdx != 0) |
3024 | return UnableToLegalize; |
3025 | |
3026 | Observer.changingInstr(MI); |
3027 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
3028 | widenScalarDst(MI, WideTy, OpIdx: 0, TruncOpcode: TargetOpcode::G_TRUNC); |
3029 | Observer.changedInstr(MI); |
3030 | return Legalized; |
3031 | case TargetOpcode::G_PTRMASK: { |
3032 | if (TypeIdx != 1) |
3033 | return UnableToLegalize; |
3034 | Observer.changingInstr(MI); |
3035 | widenScalarSrc(MI, WideTy, OpIdx: 2, ExtOpcode: TargetOpcode::G_ZEXT); |
3036 | Observer.changedInstr(MI); |
3037 | return Legalized; |
3038 | } |
3039 | case TargetOpcode::G_VECREDUCE_FADD: |
3040 | case TargetOpcode::G_VECREDUCE_FMUL: |
3041 | case TargetOpcode::G_VECREDUCE_FMIN: |
3042 | case TargetOpcode::G_VECREDUCE_FMAX: |
3043 | case TargetOpcode::G_VECREDUCE_FMINIMUM: |
3044 | case TargetOpcode::G_VECREDUCE_FMAXIMUM: { |
3045 | if (TypeIdx != 0) |
3046 | return UnableToLegalize; |
3047 | Observer.changingInstr(MI); |
3048 | Register VecReg = MI.getOperand(i: 1).getReg(); |
3049 | LLT VecTy = MRI.getType(Reg: VecReg); |
3050 | LLT WideVecTy = VecTy.isVector() |
3051 | ? LLT::vector(EC: VecTy.getElementCount(), ScalarTy: WideTy) |
3052 | : WideTy; |
3053 | widenScalarSrc(MI, WideTy: WideVecTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_FPEXT); |
3054 | widenScalarDst(MI, WideTy, OpIdx: 0, TruncOpcode: TargetOpcode::G_FPTRUNC); |
3055 | Observer.changedInstr(MI); |
3056 | return Legalized; |
3057 | } |
3058 | case TargetOpcode::G_VSCALE: { |
3059 | MachineOperand &SrcMO = MI.getOperand(i: 1); |
3060 | LLVMContext &Ctx = MIRBuilder.getMF().getFunction().getContext(); |
3061 | const APInt &SrcVal = SrcMO.getCImm()->getValue(); |
3062 | // The CImm is always a signed value |
3063 | const APInt Val = SrcVal.sext(width: WideTy.getSizeInBits()); |
3064 | Observer.changingInstr(MI); |
3065 | SrcMO.setCImm(ConstantInt::get(Context&: Ctx, V: Val)); |
3066 | widenScalarDst(MI, WideTy); |
3067 | Observer.changedInstr(MI); |
3068 | return Legalized; |
3069 | } |
3070 | case TargetOpcode::G_SPLAT_VECTOR: { |
3071 | if (TypeIdx != 1) |
3072 | return UnableToLegalize; |
3073 | |
3074 | Observer.changingInstr(MI); |
3075 | widenScalarSrc(MI, WideTy, OpIdx: 1, ExtOpcode: TargetOpcode::G_ANYEXT); |
3076 | Observer.changedInstr(MI); |
3077 | return Legalized; |
3078 | } |
3079 | } |
3080 | } |
3081 | |
3082 | static void getUnmergePieces(SmallVectorImpl<Register> &Pieces, |
3083 | MachineIRBuilder &B, Register Src, LLT Ty) { |
3084 | auto Unmerge = B.buildUnmerge(Res: Ty, Op: Src); |
3085 | for (int I = 0, E = Unmerge->getNumOperands() - 1; I != E; ++I) |
3086 | Pieces.push_back(Elt: Unmerge.getReg(Idx: I)); |
3087 | } |
3088 | |
3089 | static void emitLoadFromConstantPool(Register DstReg, const Constant *ConstVal, |
3090 | MachineIRBuilder &MIRBuilder) { |
3091 | MachineRegisterInfo &MRI = *MIRBuilder.getMRI(); |
3092 | MachineFunction &MF = MIRBuilder.getMF(); |
3093 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
3094 | unsigned AddrSpace = DL.getDefaultGlobalsAddressSpace(); |
3095 | LLT AddrPtrTy = LLT::pointer(AddressSpace: AddrSpace, SizeInBits: DL.getPointerSizeInBits(AS: AddrSpace)); |
3096 | LLT DstLLT = MRI.getType(Reg: DstReg); |
3097 | |
3098 | Align Alignment(DL.getABITypeAlign(Ty: ConstVal->getType())); |
3099 | |
3100 | auto Addr = MIRBuilder.buildConstantPool( |
3101 | Res: AddrPtrTy, |
3102 | Idx: MF.getConstantPool()->getConstantPoolIndex(C: ConstVal, Alignment)); |
3103 | |
3104 | MachineMemOperand *MMO = |
3105 | MF.getMachineMemOperand(PtrInfo: MachinePointerInfo::getConstantPool(MF), |
3106 | f: MachineMemOperand::MOLoad, MemTy: DstLLT, base_alignment: Alignment); |
3107 | |
3108 | MIRBuilder.buildLoadInstr(Opcode: TargetOpcode::G_LOAD, Res: DstReg, Addr, MMO&: *MMO); |
3109 | } |
3110 | |
3111 | LegalizerHelper::LegalizeResult |
3112 | LegalizerHelper::lowerConstant(MachineInstr &MI) { |
3113 | const MachineOperand &ConstOperand = MI.getOperand(i: 1); |
3114 | const Constant *ConstantVal = ConstOperand.getCImm(); |
3115 | |
3116 | emitLoadFromConstantPool(DstReg: MI.getOperand(i: 0).getReg(), ConstVal: ConstantVal, MIRBuilder); |
3117 | MI.eraseFromParent(); |
3118 | |
3119 | return Legalized; |
3120 | } |
3121 | |
3122 | LegalizerHelper::LegalizeResult |
3123 | LegalizerHelper::lowerFConstant(MachineInstr &MI) { |
3124 | const MachineOperand &ConstOperand = MI.getOperand(i: 1); |
3125 | const Constant *ConstantVal = ConstOperand.getFPImm(); |
3126 | |
3127 | emitLoadFromConstantPool(DstReg: MI.getOperand(i: 0).getReg(), ConstVal: ConstantVal, MIRBuilder); |
3128 | MI.eraseFromParent(); |
3129 | |
3130 | return Legalized; |
3131 | } |
3132 | |
3133 | LegalizerHelper::LegalizeResult |
3134 | LegalizerHelper::lowerBitcast(MachineInstr &MI) { |
3135 | auto [Dst, DstTy, Src, SrcTy] = MI.getFirst2RegLLTs(); |
3136 | if (SrcTy.isVector()) { |
3137 | LLT SrcEltTy = SrcTy.getElementType(); |
3138 | SmallVector<Register, 8> SrcRegs; |
3139 | |
3140 | if (DstTy.isVector()) { |
3141 | int NumDstElt = DstTy.getNumElements(); |
3142 | int NumSrcElt = SrcTy.getNumElements(); |
3143 | |
3144 | LLT DstEltTy = DstTy.getElementType(); |
3145 | LLT DstCastTy = DstEltTy; // Intermediate bitcast result type |
3146 | LLT SrcPartTy = SrcEltTy; // Original unmerge result type. |
3147 | |
3148 | // If there's an element size mismatch, insert intermediate casts to match |
3149 | // the result element type. |
3150 | if (NumSrcElt < NumDstElt) { // Source element type is larger. |
3151 | // %1:_(<4 x s8>) = G_BITCAST %0:_(<2 x s16>) |
3152 | // |
3153 | // => |
3154 | // |
3155 | // %2:_(s16), %3:_(s16) = G_UNMERGE_VALUES %0 |
3156 | // %3:_(<2 x s8>) = G_BITCAST %2 |
3157 | // %4:_(<2 x s8>) = G_BITCAST %3 |
3158 | // %1:_(<4 x s16>) = G_CONCAT_VECTORS %3, %4 |
3159 | DstCastTy = LLT::fixed_vector(NumElements: NumDstElt / NumSrcElt, ScalarTy: DstEltTy); |
3160 | SrcPartTy = SrcEltTy; |
3161 | } else if (NumSrcElt > NumDstElt) { // Source element type is smaller. |
3162 | // |
3163 | // %1:_(<2 x s16>) = G_BITCAST %0:_(<4 x s8>) |
3164 | // |
3165 | // => |
3166 | // |
3167 | // %2:_(<2 x s8>), %3:_(<2 x s8>) = G_UNMERGE_VALUES %0 |
3168 | // %3:_(s16) = G_BITCAST %2 |
3169 | // %4:_(s16) = G_BITCAST %3 |
3170 | // %1:_(<2 x s16>) = G_BUILD_VECTOR %3, %4 |
3171 | SrcPartTy = LLT::fixed_vector(NumElements: NumSrcElt / NumDstElt, ScalarTy: SrcEltTy); |
3172 | DstCastTy = DstEltTy; |
3173 | } |
3174 | |
3175 | getUnmergePieces(Pieces&: SrcRegs, B&: MIRBuilder, Src, Ty: SrcPartTy); |
3176 | for (Register &SrcReg : SrcRegs) |
3177 | SrcReg = MIRBuilder.buildBitcast(Dst: DstCastTy, Src: SrcReg).getReg(Idx: 0); |
3178 | } else |
3179 | getUnmergePieces(Pieces&: SrcRegs, B&: MIRBuilder, Src, Ty: SrcEltTy); |
3180 | |
3181 | MIRBuilder.buildMergeLikeInstr(Res: Dst, Ops: SrcRegs); |
3182 | MI.eraseFromParent(); |
3183 | return Legalized; |
3184 | } |
3185 | |
3186 | if (DstTy.isVector()) { |
3187 | SmallVector<Register, 8> SrcRegs; |
3188 | getUnmergePieces(Pieces&: SrcRegs, B&: MIRBuilder, Src, Ty: DstTy.getElementType()); |
3189 | MIRBuilder.buildMergeLikeInstr(Res: Dst, Ops: SrcRegs); |
3190 | MI.eraseFromParent(); |
3191 | return Legalized; |
3192 | } |
3193 | |
3194 | return UnableToLegalize; |
3195 | } |
3196 | |
3197 | /// Figure out the bit offset into a register when coercing a vector index for |
3198 | /// the wide element type. This is only for the case when promoting vector to |
3199 | /// one with larger elements. |
3200 | // |
3201 | /// |
3202 | /// %offset_idx = G_AND %idx, ~(-1 << Log2(DstEltSize / SrcEltSize)) |
3203 | /// %offset_bits = G_SHL %offset_idx, Log2(SrcEltSize) |
3204 | static Register getBitcastWiderVectorElementOffset(MachineIRBuilder &B, |
3205 | Register Idx, |
3206 | unsigned NewEltSize, |
3207 | unsigned OldEltSize) { |
3208 | const unsigned Log2EltRatio = Log2_32(Value: NewEltSize / OldEltSize); |
3209 | LLT IdxTy = B.getMRI()->getType(Reg: Idx); |
3210 | |
3211 | // Now figure out the amount we need to shift to get the target bits. |
3212 | auto OffsetMask = B.buildConstant( |
3213 | Res: IdxTy, Val: ~(APInt::getAllOnes(numBits: IdxTy.getSizeInBits()) << Log2EltRatio)); |
3214 | auto OffsetIdx = B.buildAnd(Dst: IdxTy, Src0: Idx, Src1: OffsetMask); |
3215 | return B.buildShl(Dst: IdxTy, Src0: OffsetIdx, |
3216 | Src1: B.buildConstant(Res: IdxTy, Val: Log2_32(Value: OldEltSize))).getReg(Idx: 0); |
3217 | } |
3218 | |
3219 | /// Perform a G_EXTRACT_VECTOR_ELT in a different sized vector element. If this |
3220 | /// is casting to a vector with a smaller element size, perform multiple element |
3221 | /// extracts and merge the results. If this is coercing to a vector with larger |
3222 | /// elements, index the bitcasted vector and extract the target element with bit |
3223 | /// operations. This is intended to force the indexing in the native register |
3224 | /// size for architectures that can dynamically index the register file. |
3225 | LegalizerHelper::LegalizeResult |
3226 | LegalizerHelper::(MachineInstr &MI, unsigned TypeIdx, |
3227 | LLT CastTy) { |
3228 | if (TypeIdx != 1) |
3229 | return UnableToLegalize; |
3230 | |
3231 | auto [Dst, DstTy, SrcVec, SrcVecTy, Idx, IdxTy] = MI.getFirst3RegLLTs(); |
3232 | |
3233 | LLT SrcEltTy = SrcVecTy.getElementType(); |
3234 | unsigned NewNumElts = CastTy.isVector() ? CastTy.getNumElements() : 1; |
3235 | unsigned OldNumElts = SrcVecTy.getNumElements(); |
3236 | |
3237 | LLT NewEltTy = CastTy.isVector() ? CastTy.getElementType() : CastTy; |
3238 | Register CastVec = MIRBuilder.buildBitcast(Dst: CastTy, Src: SrcVec).getReg(Idx: 0); |
3239 | |
3240 | const unsigned NewEltSize = NewEltTy.getSizeInBits(); |
3241 | const unsigned OldEltSize = SrcEltTy.getSizeInBits(); |
3242 | if (NewNumElts > OldNumElts) { |
3243 | // Decreasing the vector element size |
3244 | // |
3245 | // e.g. i64 = extract_vector_elt x:v2i64, y:i32 |
3246 | // => |
3247 | // v4i32:castx = bitcast x:v2i64 |
3248 | // |
3249 | // i64 = bitcast |
3250 | // (v2i32 build_vector (i32 (extract_vector_elt castx, (2 * y))), |
3251 | // (i32 (extract_vector_elt castx, (2 * y + 1))) |
3252 | // |
3253 | if (NewNumElts % OldNumElts != 0) |
3254 | return UnableToLegalize; |
3255 | |
3256 | // Type of the intermediate result vector. |
3257 | const unsigned NewEltsPerOldElt = NewNumElts / OldNumElts; |
3258 | LLT MidTy = |
3259 | LLT::scalarOrVector(EC: ElementCount::getFixed(MinVal: NewEltsPerOldElt), ScalarTy: NewEltTy); |
3260 | |
3261 | auto NewEltsPerOldEltK = MIRBuilder.buildConstant(Res: IdxTy, Val: NewEltsPerOldElt); |
3262 | |
3263 | SmallVector<Register, 8> NewOps(NewEltsPerOldElt); |
3264 | auto NewBaseIdx = MIRBuilder.buildMul(Dst: IdxTy, Src0: Idx, Src1: NewEltsPerOldEltK); |
3265 | |
3266 | for (unsigned I = 0; I < NewEltsPerOldElt; ++I) { |
3267 | auto IdxOffset = MIRBuilder.buildConstant(Res: IdxTy, Val: I); |
3268 | auto TmpIdx = MIRBuilder.buildAdd(Dst: IdxTy, Src0: NewBaseIdx, Src1: IdxOffset); |
3269 | auto Elt = MIRBuilder.buildExtractVectorElement(Res: NewEltTy, Val: CastVec, Idx: TmpIdx); |
3270 | NewOps[I] = Elt.getReg(Idx: 0); |
3271 | } |
3272 | |
3273 | auto NewVec = MIRBuilder.buildBuildVector(Res: MidTy, Ops: NewOps); |
3274 | MIRBuilder.buildBitcast(Dst, Src: NewVec); |
3275 | MI.eraseFromParent(); |
3276 | return Legalized; |
3277 | } |
3278 | |
3279 | if (NewNumElts < OldNumElts) { |
3280 | if (NewEltSize % OldEltSize != 0) |
3281 | return UnableToLegalize; |
3282 | |
3283 | // This only depends on powers of 2 because we use bit tricks to figure out |
3284 | // the bit offset we need to shift to get the target element. A general |
3285 | // expansion could emit division/multiply. |
3286 | if (!isPowerOf2_32(Value: NewEltSize / OldEltSize)) |
3287 | return UnableToLegalize; |
3288 | |
3289 | // Increasing the vector element size. |
3290 | // %elt:_(small_elt) = G_EXTRACT_VECTOR_ELT %vec:_(<N x small_elt>), %idx |
3291 | // |
3292 | // => |
3293 | // |
3294 | // %cast = G_BITCAST %vec |
3295 | // %scaled_idx = G_LSHR %idx, Log2(DstEltSize / SrcEltSize) |
3296 | // %wide_elt = G_EXTRACT_VECTOR_ELT %cast, %scaled_idx |
3297 | // %offset_idx = G_AND %idx, ~(-1 << Log2(DstEltSize / SrcEltSize)) |
3298 | // %offset_bits = G_SHL %offset_idx, Log2(SrcEltSize) |
3299 | // %elt_bits = G_LSHR %wide_elt, %offset_bits |
3300 | // %elt = G_TRUNC %elt_bits |
3301 | |
3302 | const unsigned Log2EltRatio = Log2_32(Value: NewEltSize / OldEltSize); |
3303 | auto Log2Ratio = MIRBuilder.buildConstant(Res: IdxTy, Val: Log2EltRatio); |
3304 | |
3305 | // Divide to get the index in the wider element type. |
3306 | auto ScaledIdx = MIRBuilder.buildLShr(Dst: IdxTy, Src0: Idx, Src1: Log2Ratio); |
3307 | |
3308 | Register WideElt = CastVec; |
3309 | if (CastTy.isVector()) { |
3310 | WideElt = MIRBuilder.buildExtractVectorElement(Res: NewEltTy, Val: CastVec, |
3311 | Idx: ScaledIdx).getReg(Idx: 0); |
3312 | } |
3313 | |
3314 | // Compute the bit offset into the register of the target element. |
3315 | Register OffsetBits = getBitcastWiderVectorElementOffset( |
3316 | B&: MIRBuilder, Idx, NewEltSize, OldEltSize); |
3317 | |
3318 | // Shift the wide element to get the target element. |
3319 | auto = MIRBuilder.buildLShr(Dst: NewEltTy, Src0: WideElt, Src1: OffsetBits); |
3320 | MIRBuilder.buildTrunc(Res: Dst, Op: ExtractedBits); |
3321 | MI.eraseFromParent(); |
3322 | return Legalized; |
3323 | } |
3324 | |
3325 | return UnableToLegalize; |
3326 | } |
3327 | |
3328 | /// Emit code to insert \p InsertReg into \p TargetRet at \p OffsetBits in \p |
3329 | /// TargetReg, while preserving other bits in \p TargetReg. |
3330 | /// |
3331 | /// (InsertReg << Offset) | (TargetReg & ~(-1 >> InsertReg.size()) << Offset) |
3332 | static Register buildBitFieldInsert(MachineIRBuilder &B, |
3333 | Register TargetReg, Register InsertReg, |
3334 | Register OffsetBits) { |
3335 | LLT TargetTy = B.getMRI()->getType(Reg: TargetReg); |
3336 | LLT InsertTy = B.getMRI()->getType(Reg: InsertReg); |
3337 | auto ZextVal = B.buildZExt(Res: TargetTy, Op: InsertReg); |
3338 | auto ShiftedInsertVal = B.buildShl(Dst: TargetTy, Src0: ZextVal, Src1: OffsetBits); |
3339 | |
3340 | // Produce a bitmask of the value to insert |
3341 | auto EltMask = B.buildConstant( |
3342 | Res: TargetTy, Val: APInt::getLowBitsSet(numBits: TargetTy.getSizeInBits(), |
3343 | loBitsSet: InsertTy.getSizeInBits())); |
3344 | // Shift it into position |
3345 | auto ShiftedMask = B.buildShl(Dst: TargetTy, Src0: EltMask, Src1: OffsetBits); |
3346 | auto InvShiftedMask = B.buildNot(Dst: TargetTy, Src0: ShiftedMask); |
3347 | |
3348 | // Clear out the bits in the wide element |
3349 | auto MaskedOldElt = B.buildAnd(Dst: TargetTy, Src0: TargetReg, Src1: InvShiftedMask); |
3350 | |
3351 | // The value to insert has all zeros already, so stick it into the masked |
3352 | // wide element. |
3353 | return B.buildOr(Dst: TargetTy, Src0: MaskedOldElt, Src1: ShiftedInsertVal).getReg(Idx: 0); |
3354 | } |
3355 | |
3356 | /// Perform a G_INSERT_VECTOR_ELT in a different sized vector element. If this |
3357 | /// is increasing the element size, perform the indexing in the target element |
3358 | /// type, and use bit operations to insert at the element position. This is |
3359 | /// intended for architectures that can dynamically index the register file and |
3360 | /// want to force indexing in the native register size. |
3361 | LegalizerHelper::LegalizeResult |
3362 | LegalizerHelper::bitcastInsertVectorElt(MachineInstr &MI, unsigned TypeIdx, |
3363 | LLT CastTy) { |
3364 | if (TypeIdx != 0) |
3365 | return UnableToLegalize; |
3366 | |
3367 | auto [Dst, DstTy, SrcVec, SrcVecTy, Val, ValTy, Idx, IdxTy] = |
3368 | MI.getFirst4RegLLTs(); |
3369 | LLT VecTy = DstTy; |
3370 | |
3371 | LLT VecEltTy = VecTy.getElementType(); |
3372 | LLT NewEltTy = CastTy.isVector() ? CastTy.getElementType() : CastTy; |
3373 | const unsigned NewEltSize = NewEltTy.getSizeInBits(); |
3374 | const unsigned OldEltSize = VecEltTy.getSizeInBits(); |
3375 | |
3376 | unsigned NewNumElts = CastTy.isVector() ? CastTy.getNumElements() : 1; |
3377 | unsigned OldNumElts = VecTy.getNumElements(); |
3378 | |
3379 | Register CastVec = MIRBuilder.buildBitcast(Dst: CastTy, Src: SrcVec).getReg(Idx: 0); |
3380 | if (NewNumElts < OldNumElts) { |
3381 | if (NewEltSize % OldEltSize != 0) |
3382 | return UnableToLegalize; |
3383 | |
3384 | // This only depends on powers of 2 because we use bit tricks to figure out |
3385 | // the bit offset we need to shift to get the target element. A general |
3386 | // expansion could emit division/multiply. |
3387 | if (!isPowerOf2_32(Value: NewEltSize / OldEltSize)) |
3388 | return UnableToLegalize; |
3389 | |
3390 | const unsigned Log2EltRatio = Log2_32(Value: NewEltSize / OldEltSize); |
3391 | auto Log2Ratio = MIRBuilder.buildConstant(Res: IdxTy, Val: Log2EltRatio); |
3392 | |
3393 | // Divide to get the index in the wider element type. |
3394 | auto ScaledIdx = MIRBuilder.buildLShr(Dst: IdxTy, Src0: Idx, Src1: Log2Ratio); |
3395 | |
3396 | Register = CastVec; |
3397 | if (CastTy.isVector()) { |
3398 | ExtractedElt = MIRBuilder.buildExtractVectorElement(Res: NewEltTy, Val: CastVec, |
3399 | Idx: ScaledIdx).getReg(Idx: 0); |
3400 | } |
3401 | |
3402 | // Compute the bit offset into the register of the target element. |
3403 | Register OffsetBits = getBitcastWiderVectorElementOffset( |
3404 | B&: MIRBuilder, Idx, NewEltSize, OldEltSize); |
3405 | |
3406 | Register InsertedElt = buildBitFieldInsert(B&: MIRBuilder, TargetReg: ExtractedElt, |
3407 | InsertReg: Val, OffsetBits); |
3408 | if (CastTy.isVector()) { |
3409 | InsertedElt = MIRBuilder.buildInsertVectorElement( |
3410 | Res: CastTy, Val: CastVec, Elt: InsertedElt, Idx: ScaledIdx).getReg(Idx: 0); |
3411 | } |
3412 | |
3413 | MIRBuilder.buildBitcast(Dst, Src: InsertedElt); |
3414 | MI.eraseFromParent(); |
3415 | return Legalized; |
3416 | } |
3417 | |
3418 | return UnableToLegalize; |
3419 | } |
3420 | |
3421 | // This attempts to handle G_CONCAT_VECTORS with illegal operands, particularly |
3422 | // those that have smaller than legal operands. |
3423 | // |
3424 | // <16 x s8> = G_CONCAT_VECTORS <4 x s8>, <4 x s8>, <4 x s8>, <4 x s8> |
3425 | // |
3426 | // ===> |
3427 | // |
3428 | // s32 = G_BITCAST <4 x s8> |
3429 | // s32 = G_BITCAST <4 x s8> |
3430 | // s32 = G_BITCAST <4 x s8> |
3431 | // s32 = G_BITCAST <4 x s8> |
3432 | // <4 x s32> = G_BUILD_VECTOR s32, s32, s32, s32 |
3433 | // <16 x s8> = G_BITCAST <4 x s32> |
3434 | LegalizerHelper::LegalizeResult |
3435 | LegalizerHelper::bitcastConcatVector(MachineInstr &MI, unsigned TypeIdx, |
3436 | LLT CastTy) { |
3437 | // Convert it to CONCAT instruction |
3438 | auto ConcatMI = dyn_cast<GConcatVectors>(Val: &MI); |
3439 | if (!ConcatMI) { |
3440 | return UnableToLegalize; |
3441 | } |
3442 | |
3443 | // Check if bitcast is Legal |
3444 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
3445 | LLT SrcScalTy = LLT::scalar(SizeInBits: SrcTy.getSizeInBits()); |
3446 | |
3447 | // Check if the build vector is Legal |
3448 | if (!LI.isLegal(Query: {TargetOpcode::G_BUILD_VECTOR, {CastTy, SrcScalTy}})) { |
3449 | return UnableToLegalize; |
3450 | } |
3451 | |
3452 | // Bitcast the sources |
3453 | SmallVector<Register> BitcastRegs; |
3454 | for (unsigned i = 0; i < ConcatMI->getNumSources(); i++) { |
3455 | BitcastRegs.push_back( |
3456 | Elt: MIRBuilder.buildBitcast(Dst: SrcScalTy, Src: ConcatMI->getSourceReg(I: i)) |
3457 | .getReg(Idx: 0)); |
3458 | } |
3459 | |
3460 | // Build the scalar values into a vector |
3461 | Register BuildReg = |
3462 | MIRBuilder.buildBuildVector(Res: CastTy, Ops: BitcastRegs).getReg(Idx: 0); |
3463 | MIRBuilder.buildBitcast(Dst: DstReg, Src: BuildReg); |
3464 | |
3465 | MI.eraseFromParent(); |
3466 | return Legalized; |
3467 | } |
3468 | |
3469 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerLoad(GAnyLoad &LoadMI) { |
3470 | // Lower to a memory-width G_LOAD and a G_SEXT/G_ZEXT/G_ANYEXT |
3471 | Register DstReg = LoadMI.getDstReg(); |
3472 | Register PtrReg = LoadMI.getPointerReg(); |
3473 | LLT DstTy = MRI.getType(Reg: DstReg); |
3474 | MachineMemOperand &MMO = LoadMI.getMMO(); |
3475 | LLT MemTy = MMO.getMemoryType(); |
3476 | MachineFunction &MF = MIRBuilder.getMF(); |
3477 | |
3478 | unsigned MemSizeInBits = MemTy.getSizeInBits(); |
3479 | unsigned MemStoreSizeInBits = 8 * MemTy.getSizeInBytes(); |
3480 | |
3481 | if (MemSizeInBits != MemStoreSizeInBits) { |
3482 | if (MemTy.isVector()) |
3483 | return UnableToLegalize; |
3484 | |
3485 | // Promote to a byte-sized load if not loading an integral number of |
3486 | // bytes. For example, promote EXTLOAD:i20 -> EXTLOAD:i24. |
3487 | LLT WideMemTy = LLT::scalar(SizeInBits: MemStoreSizeInBits); |
3488 | MachineMemOperand *NewMMO = |
3489 | MF.getMachineMemOperand(MMO: &MMO, PtrInfo: MMO.getPointerInfo(), Ty: WideMemTy); |
3490 | |
3491 | Register LoadReg = DstReg; |
3492 | LLT LoadTy = DstTy; |
3493 | |
3494 | // If this wasn't already an extending load, we need to widen the result |
3495 | // register to avoid creating a load with a narrower result than the source. |
3496 | if (MemStoreSizeInBits > DstTy.getSizeInBits()) { |
3497 | LoadTy = WideMemTy; |
3498 | LoadReg = MRI.createGenericVirtualRegister(Ty: WideMemTy); |
3499 | } |
3500 | |
3501 | if (isa<GSExtLoad>(Val: LoadMI)) { |
3502 | auto NewLoad = MIRBuilder.buildLoad(Res: LoadTy, Addr: PtrReg, MMO&: *NewMMO); |
3503 | MIRBuilder.buildSExtInReg(Res: LoadReg, Op: NewLoad, ImmOp: MemSizeInBits); |
3504 | } else if (isa<GZExtLoad>(Val: LoadMI) || WideMemTy == LoadTy) { |
3505 | auto NewLoad = MIRBuilder.buildLoad(Res: LoadTy, Addr: PtrReg, MMO&: *NewMMO); |
3506 | // The extra bits are guaranteed to be zero, since we stored them that |
3507 | // way. A zext load from Wide thus automatically gives zext from MemVT. |
3508 | MIRBuilder.buildAssertZExt(Res: LoadReg, Op: NewLoad, Size: MemSizeInBits); |
3509 | } else { |
3510 | MIRBuilder.buildLoad(Res: LoadReg, Addr: PtrReg, MMO&: *NewMMO); |
3511 | } |
3512 | |
3513 | if (DstTy != LoadTy) |
3514 | MIRBuilder.buildTrunc(Res: DstReg, Op: LoadReg); |
3515 | |
3516 | LoadMI.eraseFromParent(); |
3517 | return Legalized; |
3518 | } |
3519 | |
3520 | // Big endian lowering not implemented. |
3521 | if (MIRBuilder.getDataLayout().isBigEndian()) |
3522 | return UnableToLegalize; |
3523 | |
3524 | // This load needs splitting into power of 2 sized loads. |
3525 | // |
3526 | // Our strategy here is to generate anyextending loads for the smaller |
3527 | // types up to next power-2 result type, and then combine the two larger |
3528 | // result values together, before truncating back down to the non-pow-2 |
3529 | // type. |
3530 | // E.g. v1 = i24 load => |
3531 | // v2 = i32 zextload (2 byte) |
3532 | // v3 = i32 load (1 byte) |
3533 | // v4 = i32 shl v3, 16 |
3534 | // v5 = i32 or v4, v2 |
3535 | // v1 = i24 trunc v5 |
3536 | // By doing this we generate the correct truncate which should get |
3537 | // combined away as an artifact with a matching extend. |
3538 | |
3539 | uint64_t LargeSplitSize, SmallSplitSize; |
3540 | |
3541 | if (!isPowerOf2_32(Value: MemSizeInBits)) { |
3542 | // This load needs splitting into power of 2 sized loads. |
3543 | LargeSplitSize = llvm::bit_floor(Value: MemSizeInBits); |
3544 | SmallSplitSize = MemSizeInBits - LargeSplitSize; |
3545 | } else { |
3546 | // This is already a power of 2, but we still need to split this in half. |
3547 | // |
3548 | // Assume we're being asked to decompose an unaligned load. |
3549 | // TODO: If this requires multiple splits, handle them all at once. |
3550 | auto &Ctx = MF.getFunction().getContext(); |
3551 | if (TLI.allowsMemoryAccess(Context&: Ctx, DL: MIRBuilder.getDataLayout(), Ty: MemTy, MMO)) |
3552 | return UnableToLegalize; |
3553 | |
3554 | SmallSplitSize = LargeSplitSize = MemSizeInBits / 2; |
3555 | } |
3556 | |
3557 | if (MemTy.isVector()) { |
3558 | // TODO: Handle vector extloads |
3559 | if (MemTy != DstTy) |
3560 | return UnableToLegalize; |
3561 | |
3562 | // TODO: We can do better than scalarizing the vector and at least split it |
3563 | // in half. |
3564 | return reduceLoadStoreWidth(MI&: LoadMI, TypeIdx: 0, NarrowTy: DstTy.getElementType()); |
3565 | } |
3566 | |
3567 | MachineMemOperand *LargeMMO = |
3568 | MF.getMachineMemOperand(MMO: &MMO, Offset: 0, Size: LargeSplitSize / 8); |
3569 | MachineMemOperand *SmallMMO = |
3570 | MF.getMachineMemOperand(MMO: &MMO, Offset: LargeSplitSize / 8, Size: SmallSplitSize / 8); |
3571 | |
3572 | LLT PtrTy = MRI.getType(Reg: PtrReg); |
3573 | unsigned AnyExtSize = PowerOf2Ceil(A: DstTy.getSizeInBits()); |
3574 | LLT AnyExtTy = LLT::scalar(SizeInBits: AnyExtSize); |
3575 | auto LargeLoad = MIRBuilder.buildLoadInstr(Opcode: TargetOpcode::G_ZEXTLOAD, Res: AnyExtTy, |
3576 | Addr: PtrReg, MMO&: *LargeMMO); |
3577 | |
3578 | auto OffsetCst = MIRBuilder.buildConstant(Res: LLT::scalar(SizeInBits: PtrTy.getSizeInBits()), |
3579 | Val: LargeSplitSize / 8); |
3580 | Register PtrAddReg = MRI.createGenericVirtualRegister(Ty: PtrTy); |
3581 | auto SmallPtr = MIRBuilder.buildPtrAdd(Res: PtrAddReg, Op0: PtrReg, Op1: OffsetCst); |
3582 | auto SmallLoad = MIRBuilder.buildLoadInstr(Opcode: LoadMI.getOpcode(), Res: AnyExtTy, |
3583 | Addr: SmallPtr, MMO&: *SmallMMO); |
3584 | |
3585 | auto ShiftAmt = MIRBuilder.buildConstant(Res: AnyExtTy, Val: LargeSplitSize); |
3586 | auto Shift = MIRBuilder.buildShl(Dst: AnyExtTy, Src0: SmallLoad, Src1: ShiftAmt); |
3587 | |
3588 | if (AnyExtTy == DstTy) |
3589 | MIRBuilder.buildOr(Dst: DstReg, Src0: Shift, Src1: LargeLoad); |
3590 | else if (AnyExtTy.getSizeInBits() != DstTy.getSizeInBits()) { |
3591 | auto Or = MIRBuilder.buildOr(Dst: AnyExtTy, Src0: Shift, Src1: LargeLoad); |
3592 | MIRBuilder.buildTrunc(Res: DstReg, Op: {Or}); |
3593 | } else { |
3594 | assert(DstTy.isPointer() && "expected pointer" ); |
3595 | auto Or = MIRBuilder.buildOr(Dst: AnyExtTy, Src0: Shift, Src1: LargeLoad); |
3596 | |
3597 | // FIXME: We currently consider this to be illegal for non-integral address |
3598 | // spaces, but we need still need a way to reinterpret the bits. |
3599 | MIRBuilder.buildIntToPtr(Dst: DstReg, Src: Or); |
3600 | } |
3601 | |
3602 | LoadMI.eraseFromParent(); |
3603 | return Legalized; |
3604 | } |
3605 | |
3606 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerStore(GStore &StoreMI) { |
3607 | // Lower a non-power of 2 store into multiple pow-2 stores. |
3608 | // E.g. split an i24 store into an i16 store + i8 store. |
3609 | // We do this by first extending the stored value to the next largest power |
3610 | // of 2 type, and then using truncating stores to store the components. |
3611 | // By doing this, likewise with G_LOAD, generate an extend that can be |
3612 | // artifact-combined away instead of leaving behind extracts. |
3613 | Register SrcReg = StoreMI.getValueReg(); |
3614 | Register PtrReg = StoreMI.getPointerReg(); |
3615 | LLT SrcTy = MRI.getType(Reg: SrcReg); |
3616 | MachineFunction &MF = MIRBuilder.getMF(); |
3617 | MachineMemOperand &MMO = **StoreMI.memoperands_begin(); |
3618 | LLT MemTy = MMO.getMemoryType(); |
3619 | |
3620 | unsigned StoreWidth = MemTy.getSizeInBits(); |
3621 | unsigned StoreSizeInBits = 8 * MemTy.getSizeInBytes(); |
3622 | |
3623 | if (StoreWidth != StoreSizeInBits) { |
3624 | if (SrcTy.isVector()) |
3625 | return UnableToLegalize; |
3626 | |
3627 | // Promote to a byte-sized store with upper bits zero if not |
3628 | // storing an integral number of bytes. For example, promote |
3629 | // TRUNCSTORE:i1 X -> TRUNCSTORE:i8 (and X, 1) |
3630 | LLT WideTy = LLT::scalar(SizeInBits: StoreSizeInBits); |
3631 | |
3632 | if (StoreSizeInBits > SrcTy.getSizeInBits()) { |
3633 | // Avoid creating a store with a narrower source than result. |
3634 | SrcReg = MIRBuilder.buildAnyExt(Res: WideTy, Op: SrcReg).getReg(Idx: 0); |
3635 | SrcTy = WideTy; |
3636 | } |
3637 | |
3638 | auto ZextInReg = MIRBuilder.buildZExtInReg(Res: SrcTy, Op: SrcReg, ImmOp: StoreWidth); |
3639 | |
3640 | MachineMemOperand *NewMMO = |
3641 | MF.getMachineMemOperand(MMO: &MMO, PtrInfo: MMO.getPointerInfo(), Ty: WideTy); |
3642 | MIRBuilder.buildStore(Val: ZextInReg, Addr: PtrReg, MMO&: *NewMMO); |
3643 | StoreMI.eraseFromParent(); |
3644 | return Legalized; |
3645 | } |
3646 | |
3647 | if (MemTy.isVector()) { |
3648 | // TODO: Handle vector trunc stores |
3649 | if (MemTy != SrcTy) |
3650 | return UnableToLegalize; |
3651 | |
3652 | // TODO: We can do better than scalarizing the vector and at least split it |
3653 | // in half. |
3654 | return reduceLoadStoreWidth(MI&: StoreMI, TypeIdx: 0, NarrowTy: SrcTy.getElementType()); |
3655 | } |
3656 | |
3657 | unsigned MemSizeInBits = MemTy.getSizeInBits(); |
3658 | uint64_t LargeSplitSize, SmallSplitSize; |
3659 | |
3660 | if (!isPowerOf2_32(Value: MemSizeInBits)) { |
3661 | LargeSplitSize = llvm::bit_floor<uint64_t>(Value: MemTy.getSizeInBits()); |
3662 | SmallSplitSize = MemTy.getSizeInBits() - LargeSplitSize; |
3663 | } else { |
3664 | auto &Ctx = MF.getFunction().getContext(); |
3665 | if (TLI.allowsMemoryAccess(Context&: Ctx, DL: MIRBuilder.getDataLayout(), Ty: MemTy, MMO)) |
3666 | return UnableToLegalize; // Don't know what we're being asked to do. |
3667 | |
3668 | SmallSplitSize = LargeSplitSize = MemSizeInBits / 2; |
3669 | } |
3670 | |
3671 | // Extend to the next pow-2. If this store was itself the result of lowering, |
3672 | // e.g. an s56 store being broken into s32 + s24, we might have a stored type |
3673 | // that's wider than the stored size. |
3674 | unsigned AnyExtSize = PowerOf2Ceil(A: MemTy.getSizeInBits()); |
3675 | const LLT NewSrcTy = LLT::scalar(SizeInBits: AnyExtSize); |
3676 | |
3677 | if (SrcTy.isPointer()) { |
3678 | const LLT IntPtrTy = LLT::scalar(SizeInBits: SrcTy.getSizeInBits()); |
3679 | SrcReg = MIRBuilder.buildPtrToInt(Dst: IntPtrTy, Src: SrcReg).getReg(Idx: 0); |
3680 | } |
3681 | |
3682 | auto ExtVal = MIRBuilder.buildAnyExtOrTrunc(Res: NewSrcTy, Op: SrcReg); |
3683 | |
3684 | // Obtain the smaller value by shifting away the larger value. |
3685 | auto ShiftAmt = MIRBuilder.buildConstant(Res: NewSrcTy, Val: LargeSplitSize); |
3686 | auto SmallVal = MIRBuilder.buildLShr(Dst: NewSrcTy, Src0: ExtVal, Src1: ShiftAmt); |
3687 | |
3688 | // Generate the PtrAdd and truncating stores. |
3689 | LLT PtrTy = MRI.getType(Reg: PtrReg); |
3690 | auto OffsetCst = MIRBuilder.buildConstant( |
3691 | Res: LLT::scalar(SizeInBits: PtrTy.getSizeInBits()), Val: LargeSplitSize / 8); |
3692 | auto SmallPtr = |
3693 | MIRBuilder.buildPtrAdd(Res: PtrTy, Op0: PtrReg, Op1: OffsetCst); |
3694 | |
3695 | MachineMemOperand *LargeMMO = |
3696 | MF.getMachineMemOperand(MMO: &MMO, Offset: 0, Size: LargeSplitSize / 8); |
3697 | MachineMemOperand *SmallMMO = |
3698 | MF.getMachineMemOperand(MMO: &MMO, Offset: LargeSplitSize / 8, Size: SmallSplitSize / 8); |
3699 | MIRBuilder.buildStore(Val: ExtVal, Addr: PtrReg, MMO&: *LargeMMO); |
3700 | MIRBuilder.buildStore(Val: SmallVal, Addr: SmallPtr, MMO&: *SmallMMO); |
3701 | StoreMI.eraseFromParent(); |
3702 | return Legalized; |
3703 | } |
3704 | |
3705 | LegalizerHelper::LegalizeResult |
3706 | LegalizerHelper::bitcast(MachineInstr &MI, unsigned TypeIdx, LLT CastTy) { |
3707 | switch (MI.getOpcode()) { |
3708 | case TargetOpcode::G_LOAD: { |
3709 | if (TypeIdx != 0) |
3710 | return UnableToLegalize; |
3711 | MachineMemOperand &MMO = **MI.memoperands_begin(); |
3712 | |
3713 | // Not sure how to interpret a bitcast of an extending load. |
3714 | if (MMO.getMemoryType().getSizeInBits() != CastTy.getSizeInBits()) |
3715 | return UnableToLegalize; |
3716 | |
3717 | Observer.changingInstr(MI); |
3718 | bitcastDst(MI, CastTy, OpIdx: 0); |
3719 | MMO.setType(CastTy); |
3720 | // The range metadata is no longer valid when reinterpreted as a different |
3721 | // type. |
3722 | MMO.clearRanges(); |
3723 | Observer.changedInstr(MI); |
3724 | return Legalized; |
3725 | } |
3726 | case TargetOpcode::G_STORE: { |
3727 | if (TypeIdx != 0) |
3728 | return UnableToLegalize; |
3729 | |
3730 | MachineMemOperand &MMO = **MI.memoperands_begin(); |
3731 | |
3732 | // Not sure how to interpret a bitcast of a truncating store. |
3733 | if (MMO.getMemoryType().getSizeInBits() != CastTy.getSizeInBits()) |
3734 | return UnableToLegalize; |
3735 | |
3736 | Observer.changingInstr(MI); |
3737 | bitcastSrc(MI, CastTy, OpIdx: 0); |
3738 | MMO.setType(CastTy); |
3739 | Observer.changedInstr(MI); |
3740 | return Legalized; |
3741 | } |
3742 | case TargetOpcode::G_SELECT: { |
3743 | if (TypeIdx != 0) |
3744 | return UnableToLegalize; |
3745 | |
3746 | if (MRI.getType(Reg: MI.getOperand(i: 1).getReg()).isVector()) { |
3747 | LLVM_DEBUG( |
3748 | dbgs() << "bitcast action not implemented for vector select\n" ); |
3749 | return UnableToLegalize; |
3750 | } |
3751 | |
3752 | Observer.changingInstr(MI); |
3753 | bitcastSrc(MI, CastTy, OpIdx: 2); |
3754 | bitcastSrc(MI, CastTy, OpIdx: 3); |
3755 | bitcastDst(MI, CastTy, OpIdx: 0); |
3756 | Observer.changedInstr(MI); |
3757 | return Legalized; |
3758 | } |
3759 | case TargetOpcode::G_AND: |
3760 | case TargetOpcode::G_OR: |
3761 | case TargetOpcode::G_XOR: { |
3762 | Observer.changingInstr(MI); |
3763 | bitcastSrc(MI, CastTy, OpIdx: 1); |
3764 | bitcastSrc(MI, CastTy, OpIdx: 2); |
3765 | bitcastDst(MI, CastTy, OpIdx: 0); |
3766 | Observer.changedInstr(MI); |
3767 | return Legalized; |
3768 | } |
3769 | case TargetOpcode::G_EXTRACT_VECTOR_ELT: |
3770 | return bitcastExtractVectorElt(MI, TypeIdx, CastTy); |
3771 | case TargetOpcode::G_INSERT_VECTOR_ELT: |
3772 | return bitcastInsertVectorElt(MI, TypeIdx, CastTy); |
3773 | case TargetOpcode::G_CONCAT_VECTORS: |
3774 | return bitcastConcatVector(MI, TypeIdx, CastTy); |
3775 | default: |
3776 | return UnableToLegalize; |
3777 | } |
3778 | } |
3779 | |
3780 | // Legalize an instruction by changing the opcode in place. |
3781 | void LegalizerHelper::changeOpcode(MachineInstr &MI, unsigned NewOpcode) { |
3782 | Observer.changingInstr(MI); |
3783 | MI.setDesc(MIRBuilder.getTII().get(Opcode: NewOpcode)); |
3784 | Observer.changedInstr(MI); |
3785 | } |
3786 | |
3787 | LegalizerHelper::LegalizeResult |
3788 | LegalizerHelper::lower(MachineInstr &MI, unsigned TypeIdx, LLT LowerHintTy) { |
3789 | using namespace TargetOpcode; |
3790 | |
3791 | switch(MI.getOpcode()) { |
3792 | default: |
3793 | return UnableToLegalize; |
3794 | case TargetOpcode::G_FCONSTANT: |
3795 | return lowerFConstant(MI); |
3796 | case TargetOpcode::G_BITCAST: |
3797 | return lowerBitcast(MI); |
3798 | case TargetOpcode::G_SREM: |
3799 | case TargetOpcode::G_UREM: { |
3800 | LLT Ty = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
3801 | auto Quot = |
3802 | MIRBuilder.buildInstr(Opc: MI.getOpcode() == G_SREM ? G_SDIV : G_UDIV, DstOps: {Ty}, |
3803 | SrcOps: {MI.getOperand(i: 1), MI.getOperand(i: 2)}); |
3804 | |
3805 | auto Prod = MIRBuilder.buildMul(Dst: Ty, Src0: Quot, Src1: MI.getOperand(i: 2)); |
3806 | MIRBuilder.buildSub(Dst: MI.getOperand(i: 0), Src0: MI.getOperand(i: 1), Src1: Prod); |
3807 | MI.eraseFromParent(); |
3808 | return Legalized; |
3809 | } |
3810 | case TargetOpcode::G_SADDO: |
3811 | case TargetOpcode::G_SSUBO: |
3812 | return lowerSADDO_SSUBO(MI); |
3813 | case TargetOpcode::G_UMULH: |
3814 | case TargetOpcode::G_SMULH: |
3815 | return lowerSMULH_UMULH(MI); |
3816 | case TargetOpcode::G_SMULO: |
3817 | case TargetOpcode::G_UMULO: { |
3818 | // Generate G_UMULH/G_SMULH to check for overflow and a normal G_MUL for the |
3819 | // result. |
3820 | auto [Res, Overflow, LHS, RHS] = MI.getFirst4Regs(); |
3821 | LLT Ty = MRI.getType(Reg: Res); |
3822 | |
3823 | unsigned Opcode = MI.getOpcode() == TargetOpcode::G_SMULO |
3824 | ? TargetOpcode::G_SMULH |
3825 | : TargetOpcode::G_UMULH; |
3826 | |
3827 | Observer.changingInstr(MI); |
3828 | const auto &TII = MIRBuilder.getTII(); |
3829 | MI.setDesc(TII.get(Opcode: TargetOpcode::G_MUL)); |
3830 | MI.removeOperand(OpNo: 1); |
3831 | Observer.changedInstr(MI); |
3832 | |
3833 | auto HiPart = MIRBuilder.buildInstr(Opc: Opcode, DstOps: {Ty}, SrcOps: {LHS, RHS}); |
3834 | auto Zero = MIRBuilder.buildConstant(Res: Ty, Val: 0); |
3835 | |
3836 | // Move insert point forward so we can use the Res register if needed. |
3837 | MIRBuilder.setInsertPt(MBB&: MIRBuilder.getMBB(), II: ++MIRBuilder.getInsertPt()); |
3838 | |
3839 | // For *signed* multiply, overflow is detected by checking: |
3840 | // (hi != (lo >> bitwidth-1)) |
3841 | if (Opcode == TargetOpcode::G_SMULH) { |
3842 | auto ShiftAmt = MIRBuilder.buildConstant(Res: Ty, Val: Ty.getSizeInBits() - 1); |
3843 | auto Shifted = MIRBuilder.buildAShr(Dst: Ty, Src0: Res, Src1: ShiftAmt); |
3844 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: Overflow, Op0: HiPart, Op1: Shifted); |
3845 | } else { |
3846 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: Overflow, Op0: HiPart, Op1: Zero); |
3847 | } |
3848 | return Legalized; |
3849 | } |
3850 | case TargetOpcode::G_FNEG: { |
3851 | auto [Res, SubByReg] = MI.getFirst2Regs(); |
3852 | LLT Ty = MRI.getType(Reg: Res); |
3853 | |
3854 | // TODO: Handle vector types once we are able to |
3855 | // represent them. |
3856 | if (Ty.isVector()) |
3857 | return UnableToLegalize; |
3858 | auto SignMask = |
3859 | MIRBuilder.buildConstant(Res: Ty, Val: APInt::getSignMask(BitWidth: Ty.getSizeInBits())); |
3860 | MIRBuilder.buildXor(Dst: Res, Src0: SubByReg, Src1: SignMask); |
3861 | MI.eraseFromParent(); |
3862 | return Legalized; |
3863 | } |
3864 | case TargetOpcode::G_FSUB: |
3865 | case TargetOpcode::G_STRICT_FSUB: { |
3866 | auto [Res, LHS, RHS] = MI.getFirst3Regs(); |
3867 | LLT Ty = MRI.getType(Reg: Res); |
3868 | |
3869 | // Lower (G_FSUB LHS, RHS) to (G_FADD LHS, (G_FNEG RHS)). |
3870 | auto Neg = MIRBuilder.buildFNeg(Dst: Ty, Src0: RHS); |
3871 | |
3872 | if (MI.getOpcode() == TargetOpcode::G_STRICT_FSUB) |
3873 | MIRBuilder.buildStrictFAdd(Dst: Res, Src0: LHS, Src1: Neg, Flags: MI.getFlags()); |
3874 | else |
3875 | MIRBuilder.buildFAdd(Dst: Res, Src0: LHS, Src1: Neg, Flags: MI.getFlags()); |
3876 | |
3877 | MI.eraseFromParent(); |
3878 | return Legalized; |
3879 | } |
3880 | case TargetOpcode::G_FMAD: |
3881 | return lowerFMad(MI); |
3882 | case TargetOpcode::G_FFLOOR: |
3883 | return lowerFFloor(MI); |
3884 | case TargetOpcode::G_INTRINSIC_ROUND: |
3885 | return lowerIntrinsicRound(MI); |
3886 | case TargetOpcode::G_FRINT: { |
3887 | // Since round even is the assumed rounding mode for unconstrained FP |
3888 | // operations, rint and roundeven are the same operation. |
3889 | changeOpcode(MI, NewOpcode: TargetOpcode::G_INTRINSIC_ROUNDEVEN); |
3890 | return Legalized; |
3891 | } |
3892 | case TargetOpcode::G_ATOMIC_CMPXCHG_WITH_SUCCESS: { |
3893 | auto [OldValRes, SuccessRes, Addr, CmpVal, NewVal] = MI.getFirst5Regs(); |
3894 | Register NewOldValRes = MRI.cloneVirtualRegister(VReg: OldValRes); |
3895 | MIRBuilder.buildAtomicCmpXchg(OldValRes: NewOldValRes, Addr, CmpVal, NewVal, |
3896 | MMO&: **MI.memoperands_begin()); |
3897 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_EQ, Res: SuccessRes, Op0: NewOldValRes, Op1: CmpVal); |
3898 | MIRBuilder.buildCopy(Res: OldValRes, Op: NewOldValRes); |
3899 | MI.eraseFromParent(); |
3900 | return Legalized; |
3901 | } |
3902 | case TargetOpcode::G_LOAD: |
3903 | case TargetOpcode::G_SEXTLOAD: |
3904 | case TargetOpcode::G_ZEXTLOAD: |
3905 | return lowerLoad(LoadMI&: cast<GAnyLoad>(Val&: MI)); |
3906 | case TargetOpcode::G_STORE: |
3907 | return lowerStore(StoreMI&: cast<GStore>(Val&: MI)); |
3908 | case TargetOpcode::G_CTLZ_ZERO_UNDEF: |
3909 | case TargetOpcode::G_CTTZ_ZERO_UNDEF: |
3910 | case TargetOpcode::G_CTLZ: |
3911 | case TargetOpcode::G_CTTZ: |
3912 | case TargetOpcode::G_CTPOP: |
3913 | return lowerBitCount(MI); |
3914 | case G_UADDO: { |
3915 | auto [Res, CarryOut, LHS, RHS] = MI.getFirst4Regs(); |
3916 | |
3917 | Register NewRes = MRI.cloneVirtualRegister(VReg: Res); |
3918 | |
3919 | MIRBuilder.buildAdd(Dst: NewRes, Src0: LHS, Src1: RHS); |
3920 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_ULT, Res: CarryOut, Op0: NewRes, Op1: RHS); |
3921 | |
3922 | MIRBuilder.buildCopy(Res, Op: NewRes); |
3923 | |
3924 | MI.eraseFromParent(); |
3925 | return Legalized; |
3926 | } |
3927 | case G_UADDE: { |
3928 | auto [Res, CarryOut, LHS, RHS, CarryIn] = MI.getFirst5Regs(); |
3929 | const LLT CondTy = MRI.getType(Reg: CarryOut); |
3930 | const LLT Ty = MRI.getType(Reg: Res); |
3931 | |
3932 | Register NewRes = MRI.cloneVirtualRegister(VReg: Res); |
3933 | |
3934 | // Initial add of the two operands. |
3935 | auto TmpRes = MIRBuilder.buildAdd(Dst: Ty, Src0: LHS, Src1: RHS); |
3936 | |
3937 | // Initial check for carry. |
3938 | auto Carry = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_ULT, Res: CondTy, Op0: TmpRes, Op1: LHS); |
3939 | |
3940 | // Add the sum and the carry. |
3941 | auto ZExtCarryIn = MIRBuilder.buildZExt(Res: Ty, Op: CarryIn); |
3942 | MIRBuilder.buildAdd(Dst: NewRes, Src0: TmpRes, Src1: ZExtCarryIn); |
3943 | |
3944 | // Second check for carry. We can only carry if the initial sum is all 1s |
3945 | // and the carry is set, resulting in a new sum of 0. |
3946 | auto Zero = MIRBuilder.buildConstant(Res: Ty, Val: 0); |
3947 | auto ResEqZero = |
3948 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_EQ, Res: CondTy, Op0: NewRes, Op1: Zero); |
3949 | auto Carry2 = MIRBuilder.buildAnd(Dst: CondTy, Src0: ResEqZero, Src1: CarryIn); |
3950 | MIRBuilder.buildOr(Dst: CarryOut, Src0: Carry, Src1: Carry2); |
3951 | |
3952 | MIRBuilder.buildCopy(Res, Op: NewRes); |
3953 | |
3954 | MI.eraseFromParent(); |
3955 | return Legalized; |
3956 | } |
3957 | case G_USUBO: { |
3958 | auto [Res, BorrowOut, LHS, RHS] = MI.getFirst4Regs(); |
3959 | |
3960 | MIRBuilder.buildSub(Dst: Res, Src0: LHS, Src1: RHS); |
3961 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_ULT, Res: BorrowOut, Op0: LHS, Op1: RHS); |
3962 | |
3963 | MI.eraseFromParent(); |
3964 | return Legalized; |
3965 | } |
3966 | case G_USUBE: { |
3967 | auto [Res, BorrowOut, LHS, RHS, BorrowIn] = MI.getFirst5Regs(); |
3968 | const LLT CondTy = MRI.getType(Reg: BorrowOut); |
3969 | const LLT Ty = MRI.getType(Reg: Res); |
3970 | |
3971 | // Initial subtract of the two operands. |
3972 | auto TmpRes = MIRBuilder.buildSub(Dst: Ty, Src0: LHS, Src1: RHS); |
3973 | |
3974 | // Initial check for borrow. |
3975 | auto Borrow = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_UGT, Res: CondTy, Op0: TmpRes, Op1: LHS); |
3976 | |
3977 | // Subtract the borrow from the first subtract. |
3978 | auto ZExtBorrowIn = MIRBuilder.buildZExt(Res: Ty, Op: BorrowIn); |
3979 | MIRBuilder.buildSub(Dst: Res, Src0: TmpRes, Src1: ZExtBorrowIn); |
3980 | |
3981 | // Second check for borrow. We can only borrow if the initial difference is |
3982 | // 0 and the borrow is set, resulting in a new difference of all 1s. |
3983 | auto Zero = MIRBuilder.buildConstant(Res: Ty, Val: 0); |
3984 | auto TmpResEqZero = |
3985 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_EQ, Res: CondTy, Op0: TmpRes, Op1: Zero); |
3986 | auto Borrow2 = MIRBuilder.buildAnd(Dst: CondTy, Src0: TmpResEqZero, Src1: BorrowIn); |
3987 | MIRBuilder.buildOr(Dst: BorrowOut, Src0: Borrow, Src1: Borrow2); |
3988 | |
3989 | MI.eraseFromParent(); |
3990 | return Legalized; |
3991 | } |
3992 | case G_UITOFP: |
3993 | return lowerUITOFP(MI); |
3994 | case G_SITOFP: |
3995 | return lowerSITOFP(MI); |
3996 | case G_FPTOUI: |
3997 | return lowerFPTOUI(MI); |
3998 | case G_FPTOSI: |
3999 | return lowerFPTOSI(MI); |
4000 | case G_FPTRUNC: |
4001 | return lowerFPTRUNC(MI); |
4002 | case G_FPOWI: |
4003 | return lowerFPOWI(MI); |
4004 | case G_SMIN: |
4005 | case G_SMAX: |
4006 | case G_UMIN: |
4007 | case G_UMAX: |
4008 | return lowerMinMax(MI); |
4009 | case G_SCMP: |
4010 | case G_UCMP: |
4011 | return lowerThreewayCompare(MI); |
4012 | case G_FCOPYSIGN: |
4013 | return lowerFCopySign(MI); |
4014 | case G_FMINNUM: |
4015 | case G_FMAXNUM: |
4016 | return lowerFMinNumMaxNum(MI); |
4017 | case G_MERGE_VALUES: |
4018 | return lowerMergeValues(MI); |
4019 | case G_UNMERGE_VALUES: |
4020 | return lowerUnmergeValues(MI); |
4021 | case TargetOpcode::G_SEXT_INREG: { |
4022 | assert(MI.getOperand(2).isImm() && "Expected immediate" ); |
4023 | int64_t SizeInBits = MI.getOperand(i: 2).getImm(); |
4024 | |
4025 | auto [DstReg, SrcReg] = MI.getFirst2Regs(); |
4026 | LLT DstTy = MRI.getType(Reg: DstReg); |
4027 | Register TmpRes = MRI.createGenericVirtualRegister(Ty: DstTy); |
4028 | |
4029 | auto MIBSz = MIRBuilder.buildConstant(Res: DstTy, Val: DstTy.getScalarSizeInBits() - SizeInBits); |
4030 | MIRBuilder.buildShl(Dst: TmpRes, Src0: SrcReg, Src1: MIBSz->getOperand(i: 0)); |
4031 | MIRBuilder.buildAShr(Dst: DstReg, Src0: TmpRes, Src1: MIBSz->getOperand(i: 0)); |
4032 | MI.eraseFromParent(); |
4033 | return Legalized; |
4034 | } |
4035 | case G_EXTRACT_VECTOR_ELT: |
4036 | case G_INSERT_VECTOR_ELT: |
4037 | return lowerExtractInsertVectorElt(MI); |
4038 | case G_SHUFFLE_VECTOR: |
4039 | return lowerShuffleVector(MI); |
4040 | case G_VECTOR_COMPRESS: |
4041 | return lowerVECTOR_COMPRESS(MI); |
4042 | case G_DYN_STACKALLOC: |
4043 | return lowerDynStackAlloc(MI); |
4044 | case G_STACKSAVE: |
4045 | return lowerStackSave(MI); |
4046 | case G_STACKRESTORE: |
4047 | return lowerStackRestore(MI); |
4048 | case G_EXTRACT: |
4049 | return lowerExtract(MI); |
4050 | case G_INSERT: |
4051 | return lowerInsert(MI); |
4052 | case G_BSWAP: |
4053 | return lowerBswap(MI); |
4054 | case G_BITREVERSE: |
4055 | return lowerBitreverse(MI); |
4056 | case G_READ_REGISTER: |
4057 | case G_WRITE_REGISTER: |
4058 | return lowerReadWriteRegister(MI); |
4059 | case G_UADDSAT: |
4060 | case G_USUBSAT: { |
4061 | // Try to make a reasonable guess about which lowering strategy to use. The |
4062 | // target can override this with custom lowering and calling the |
4063 | // implementation functions. |
4064 | LLT Ty = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
4065 | if (LI.isLegalOrCustom(Query: {G_UMIN, Ty})) |
4066 | return lowerAddSubSatToMinMax(MI); |
4067 | return lowerAddSubSatToAddoSubo(MI); |
4068 | } |
4069 | case G_SADDSAT: |
4070 | case G_SSUBSAT: { |
4071 | LLT Ty = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
4072 | |
4073 | // FIXME: It would probably make more sense to see if G_SADDO is preferred, |
4074 | // since it's a shorter expansion. However, we would need to figure out the |
4075 | // preferred boolean type for the carry out for the query. |
4076 | if (LI.isLegalOrCustom(Query: {G_SMIN, Ty}) && LI.isLegalOrCustom(Query: {G_SMAX, Ty})) |
4077 | return lowerAddSubSatToMinMax(MI); |
4078 | return lowerAddSubSatToAddoSubo(MI); |
4079 | } |
4080 | case G_SSHLSAT: |
4081 | case G_USHLSAT: |
4082 | return lowerShlSat(MI); |
4083 | case G_ABS: |
4084 | return lowerAbsToAddXor(MI); |
4085 | case G_SELECT: |
4086 | return lowerSelect(MI); |
4087 | case G_IS_FPCLASS: |
4088 | return lowerISFPCLASS(MI); |
4089 | case G_SDIVREM: |
4090 | case G_UDIVREM: |
4091 | return lowerDIVREM(MI); |
4092 | case G_FSHL: |
4093 | case G_FSHR: |
4094 | return lowerFunnelShift(MI); |
4095 | case G_ROTL: |
4096 | case G_ROTR: |
4097 | return lowerRotate(MI); |
4098 | case G_MEMSET: |
4099 | case G_MEMCPY: |
4100 | case G_MEMMOVE: |
4101 | return lowerMemCpyFamily(MI); |
4102 | case G_MEMCPY_INLINE: |
4103 | return lowerMemcpyInline(MI); |
4104 | case G_ZEXT: |
4105 | case G_SEXT: |
4106 | case G_ANYEXT: |
4107 | return lowerEXT(MI); |
4108 | case G_TRUNC: |
4109 | return lowerTRUNC(MI); |
4110 | GISEL_VECREDUCE_CASES_NONSEQ |
4111 | return lowerVectorReduction(MI); |
4112 | case G_VAARG: |
4113 | return lowerVAArg(MI); |
4114 | } |
4115 | } |
4116 | |
4117 | Align LegalizerHelper::getStackTemporaryAlignment(LLT Ty, |
4118 | Align MinAlign) const { |
4119 | // FIXME: We're missing a way to go back from LLT to llvm::Type to query the |
4120 | // datalayout for the preferred alignment. Also there should be a target hook |
4121 | // for this to allow targets to reduce the alignment and ignore the |
4122 | // datalayout. e.g. AMDGPU should always use a 4-byte alignment, regardless of |
4123 | // the type. |
4124 | return std::max(a: Align(PowerOf2Ceil(A: Ty.getSizeInBytes())), b: MinAlign); |
4125 | } |
4126 | |
4127 | MachineInstrBuilder |
4128 | LegalizerHelper::createStackTemporary(TypeSize Bytes, Align Alignment, |
4129 | MachinePointerInfo &PtrInfo) { |
4130 | MachineFunction &MF = MIRBuilder.getMF(); |
4131 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
4132 | int FrameIdx = MF.getFrameInfo().CreateStackObject(Size: Bytes, Alignment, isSpillSlot: false); |
4133 | |
4134 | unsigned AddrSpace = DL.getAllocaAddrSpace(); |
4135 | LLT FramePtrTy = LLT::pointer(AddressSpace: AddrSpace, SizeInBits: DL.getPointerSizeInBits(AS: AddrSpace)); |
4136 | |
4137 | PtrInfo = MachinePointerInfo::getFixedStack(MF, FI: FrameIdx); |
4138 | return MIRBuilder.buildFrameIndex(Res: FramePtrTy, Idx: FrameIdx); |
4139 | } |
4140 | |
4141 | static Register clampVectorIndex(MachineIRBuilder &B, Register IdxReg, |
4142 | LLT VecTy) { |
4143 | LLT IdxTy = B.getMRI()->getType(Reg: IdxReg); |
4144 | unsigned NElts = VecTy.getNumElements(); |
4145 | |
4146 | int64_t IdxVal; |
4147 | if (mi_match(R: IdxReg, MRI: *B.getMRI(), P: m_ICst(Cst&: IdxVal))) { |
4148 | if (IdxVal < VecTy.getNumElements()) |
4149 | return IdxReg; |
4150 | // If a constant index would be out of bounds, clamp it as well. |
4151 | } |
4152 | |
4153 | if (isPowerOf2_32(Value: NElts)) { |
4154 | APInt Imm = APInt::getLowBitsSet(numBits: IdxTy.getSizeInBits(), loBitsSet: Log2_32(Value: NElts)); |
4155 | return B.buildAnd(Dst: IdxTy, Src0: IdxReg, Src1: B.buildConstant(Res: IdxTy, Val: Imm)).getReg(Idx: 0); |
4156 | } |
4157 | |
4158 | return B.buildUMin(Dst: IdxTy, Src0: IdxReg, Src1: B.buildConstant(Res: IdxTy, Val: NElts - 1)) |
4159 | .getReg(Idx: 0); |
4160 | } |
4161 | |
4162 | Register LegalizerHelper::getVectorElementPointer(Register VecPtr, LLT VecTy, |
4163 | Register Index) { |
4164 | LLT EltTy = VecTy.getElementType(); |
4165 | |
4166 | // Calculate the element offset and add it to the pointer. |
4167 | unsigned EltSize = EltTy.getSizeInBits() / 8; // FIXME: should be ABI size. |
4168 | assert(EltSize * 8 == EltTy.getSizeInBits() && |
4169 | "Converting bits to bytes lost precision" ); |
4170 | |
4171 | Index = clampVectorIndex(B&: MIRBuilder, IdxReg: Index, VecTy); |
4172 | |
4173 | // Convert index to the correct size for the address space. |
4174 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
4175 | unsigned AS = MRI.getType(Reg: VecPtr).getAddressSpace(); |
4176 | unsigned IndexSizeInBits = DL.getIndexSize(AS) * 8; |
4177 | LLT IdxTy = MRI.getType(Reg: Index).changeElementSize(NewEltSize: IndexSizeInBits); |
4178 | if (IdxTy != MRI.getType(Reg: Index)) |
4179 | Index = MIRBuilder.buildSExtOrTrunc(Res: IdxTy, Op: Index).getReg(Idx: 0); |
4180 | |
4181 | auto Mul = MIRBuilder.buildMul(Dst: IdxTy, Src0: Index, |
4182 | Src1: MIRBuilder.buildConstant(Res: IdxTy, Val: EltSize)); |
4183 | |
4184 | LLT PtrTy = MRI.getType(Reg: VecPtr); |
4185 | return MIRBuilder.buildPtrAdd(Res: PtrTy, Op0: VecPtr, Op1: Mul).getReg(Idx: 0); |
4186 | } |
4187 | |
4188 | #ifndef NDEBUG |
4189 | /// Check that all vector operands have same number of elements. Other operands |
4190 | /// should be listed in NonVecOp. |
4191 | static bool hasSameNumEltsOnAllVectorOperands( |
4192 | GenericMachineInstr &MI, MachineRegisterInfo &MRI, |
4193 | std::initializer_list<unsigned> NonVecOpIndices) { |
4194 | if (MI.getNumMemOperands() != 0) |
4195 | return false; |
4196 | |
4197 | LLT VecTy = MRI.getType(MI.getReg(0)); |
4198 | if (!VecTy.isVector()) |
4199 | return false; |
4200 | unsigned NumElts = VecTy.getNumElements(); |
4201 | |
4202 | for (unsigned OpIdx = 1; OpIdx < MI.getNumOperands(); ++OpIdx) { |
4203 | MachineOperand &Op = MI.getOperand(OpIdx); |
4204 | if (!Op.isReg()) { |
4205 | if (!is_contained(NonVecOpIndices, OpIdx)) |
4206 | return false; |
4207 | continue; |
4208 | } |
4209 | |
4210 | LLT Ty = MRI.getType(Op.getReg()); |
4211 | if (!Ty.isVector()) { |
4212 | if (!is_contained(NonVecOpIndices, OpIdx)) |
4213 | return false; |
4214 | continue; |
4215 | } |
4216 | |
4217 | if (Ty.getNumElements() != NumElts) |
4218 | return false; |
4219 | } |
4220 | |
4221 | return true; |
4222 | } |
4223 | #endif |
4224 | |
4225 | /// Fill \p DstOps with DstOps that have same number of elements combined as |
4226 | /// the Ty. These DstOps have either scalar type when \p NumElts = 1 or are |
4227 | /// vectors with \p NumElts elements. When Ty.getNumElements() is not multiple |
4228 | /// of \p NumElts last DstOp (leftover) has fewer then \p NumElts elements. |
4229 | static void makeDstOps(SmallVectorImpl<DstOp> &DstOps, LLT Ty, |
4230 | unsigned NumElts) { |
4231 | LLT LeftoverTy; |
4232 | assert(Ty.isVector() && "Expected vector type" ); |
4233 | LLT EltTy = Ty.getElementType(); |
4234 | LLT NarrowTy = (NumElts == 1) ? EltTy : LLT::fixed_vector(NumElements: NumElts, ScalarTy: EltTy); |
4235 | int NumParts, NumLeftover; |
4236 | std::tie(args&: NumParts, args&: NumLeftover) = |
4237 | getNarrowTypeBreakDown(OrigTy: Ty, NarrowTy, LeftoverTy); |
4238 | |
4239 | assert(NumParts > 0 && "Error in getNarrowTypeBreakDown" ); |
4240 | for (int i = 0; i < NumParts; ++i) { |
4241 | DstOps.push_back(Elt: NarrowTy); |
4242 | } |
4243 | |
4244 | if (LeftoverTy.isValid()) { |
4245 | assert(NumLeftover == 1 && "expected exactly one leftover" ); |
4246 | DstOps.push_back(Elt: LeftoverTy); |
4247 | } |
4248 | } |
4249 | |
4250 | /// Operand \p Op is used on \p N sub-instructions. Fill \p Ops with \p N SrcOps |
4251 | /// made from \p Op depending on operand type. |
4252 | static void broadcastSrcOp(SmallVectorImpl<SrcOp> &Ops, unsigned N, |
4253 | MachineOperand &Op) { |
4254 | for (unsigned i = 0; i < N; ++i) { |
4255 | if (Op.isReg()) |
4256 | Ops.push_back(Elt: Op.getReg()); |
4257 | else if (Op.isImm()) |
4258 | Ops.push_back(Elt: Op.getImm()); |
4259 | else if (Op.isPredicate()) |
4260 | Ops.push_back(Elt: static_cast<CmpInst::Predicate>(Op.getPredicate())); |
4261 | else |
4262 | llvm_unreachable("Unsupported type" ); |
4263 | } |
4264 | } |
4265 | |
4266 | // Handle splitting vector operations which need to have the same number of |
4267 | // elements in each type index, but each type index may have a different element |
4268 | // type. |
4269 | // |
4270 | // e.g. <4 x s64> = G_SHL <4 x s64>, <4 x s32> -> |
4271 | // <2 x s64> = G_SHL <2 x s64>, <2 x s32> |
4272 | // <2 x s64> = G_SHL <2 x s64>, <2 x s32> |
4273 | // |
4274 | // Also handles some irregular breakdown cases, e.g. |
4275 | // e.g. <3 x s64> = G_SHL <3 x s64>, <3 x s32> -> |
4276 | // <2 x s64> = G_SHL <2 x s64>, <2 x s32> |
4277 | // s64 = G_SHL s64, s32 |
4278 | LegalizerHelper::LegalizeResult |
4279 | LegalizerHelper::fewerElementsVectorMultiEltType( |
4280 | GenericMachineInstr &MI, unsigned NumElts, |
4281 | std::initializer_list<unsigned> NonVecOpIndices) { |
4282 | assert(hasSameNumEltsOnAllVectorOperands(MI, MRI, NonVecOpIndices) && |
4283 | "Non-compatible opcode or not specified non-vector operands" ); |
4284 | unsigned OrigNumElts = MRI.getType(Reg: MI.getReg(Idx: 0)).getNumElements(); |
4285 | |
4286 | unsigned NumInputs = MI.getNumOperands() - MI.getNumDefs(); |
4287 | unsigned NumDefs = MI.getNumDefs(); |
4288 | |
4289 | // Create DstOps (sub-vectors with NumElts elts + Leftover) for each output. |
4290 | // Build instructions with DstOps to use instruction found by CSE directly. |
4291 | // CSE copies found instruction into given vreg when building with vreg dest. |
4292 | SmallVector<SmallVector<DstOp, 8>, 2> OutputOpsPieces(NumDefs); |
4293 | // Output registers will be taken from created instructions. |
4294 | SmallVector<SmallVector<Register, 8>, 2> OutputRegs(NumDefs); |
4295 | for (unsigned i = 0; i < NumDefs; ++i) { |
4296 | makeDstOps(DstOps&: OutputOpsPieces[i], Ty: MRI.getType(Reg: MI.getReg(Idx: i)), NumElts); |
4297 | } |
4298 | |
4299 | // Split vector input operands into sub-vectors with NumElts elts + Leftover. |
4300 | // Operands listed in NonVecOpIndices will be used as is without splitting; |
4301 | // examples: compare predicate in icmp and fcmp (op 1), vector select with i1 |
4302 | // scalar condition (op 1), immediate in sext_inreg (op 2). |
4303 | SmallVector<SmallVector<SrcOp, 8>, 3> InputOpsPieces(NumInputs); |
4304 | for (unsigned UseIdx = NumDefs, UseNo = 0; UseIdx < MI.getNumOperands(); |
4305 | ++UseIdx, ++UseNo) { |
4306 | if (is_contained(Set: NonVecOpIndices, Element: UseIdx)) { |
4307 | broadcastSrcOp(Ops&: InputOpsPieces[UseNo], N: OutputOpsPieces[0].size(), |
4308 | Op&: MI.getOperand(i: UseIdx)); |
4309 | } else { |
4310 | SmallVector<Register, 8> SplitPieces; |
4311 | extractVectorParts(Reg: MI.getReg(Idx: UseIdx), NumElts, VRegs&: SplitPieces, MIRBuilder, |
4312 | MRI); |
4313 | for (auto Reg : SplitPieces) |
4314 | InputOpsPieces[UseNo].push_back(Elt: Reg); |
4315 | } |
4316 | } |
4317 | |
4318 | unsigned NumLeftovers = OrigNumElts % NumElts ? 1 : 0; |
4319 | |
4320 | // Take i-th piece of each input operand split and build sub-vector/scalar |
4321 | // instruction. Set i-th DstOp(s) from OutputOpsPieces as destination(s). |
4322 | for (unsigned i = 0; i < OrigNumElts / NumElts + NumLeftovers; ++i) { |
4323 | SmallVector<DstOp, 2> Defs; |
4324 | for (unsigned DstNo = 0; DstNo < NumDefs; ++DstNo) |
4325 | Defs.push_back(Elt: OutputOpsPieces[DstNo][i]); |
4326 | |
4327 | SmallVector<SrcOp, 3> Uses; |
4328 | for (unsigned InputNo = 0; InputNo < NumInputs; ++InputNo) |
4329 | Uses.push_back(Elt: InputOpsPieces[InputNo][i]); |
4330 | |
4331 | auto I = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: Defs, SrcOps: Uses, Flags: MI.getFlags()); |
4332 | for (unsigned DstNo = 0; DstNo < NumDefs; ++DstNo) |
4333 | OutputRegs[DstNo].push_back(Elt: I.getReg(Idx: DstNo)); |
4334 | } |
4335 | |
4336 | // Merge small outputs into MI's output for each def operand. |
4337 | if (NumLeftovers) { |
4338 | for (unsigned i = 0; i < NumDefs; ++i) |
4339 | mergeMixedSubvectors(DstReg: MI.getReg(Idx: i), PartRegs: OutputRegs[i]); |
4340 | } else { |
4341 | for (unsigned i = 0; i < NumDefs; ++i) |
4342 | MIRBuilder.buildMergeLikeInstr(Res: MI.getReg(Idx: i), Ops: OutputRegs[i]); |
4343 | } |
4344 | |
4345 | MI.eraseFromParent(); |
4346 | return Legalized; |
4347 | } |
4348 | |
4349 | LegalizerHelper::LegalizeResult |
4350 | LegalizerHelper::fewerElementsVectorPhi(GenericMachineInstr &MI, |
4351 | unsigned NumElts) { |
4352 | unsigned OrigNumElts = MRI.getType(Reg: MI.getReg(Idx: 0)).getNumElements(); |
4353 | |
4354 | unsigned NumInputs = MI.getNumOperands() - MI.getNumDefs(); |
4355 | unsigned NumDefs = MI.getNumDefs(); |
4356 | |
4357 | SmallVector<DstOp, 8> OutputOpsPieces; |
4358 | SmallVector<Register, 8> OutputRegs; |
4359 | makeDstOps(DstOps&: OutputOpsPieces, Ty: MRI.getType(Reg: MI.getReg(Idx: 0)), NumElts); |
4360 | |
4361 | // Instructions that perform register split will be inserted in basic block |
4362 | // where register is defined (basic block is in the next operand). |
4363 | SmallVector<SmallVector<Register, 8>, 3> InputOpsPieces(NumInputs / 2); |
4364 | for (unsigned UseIdx = NumDefs, UseNo = 0; UseIdx < MI.getNumOperands(); |
4365 | UseIdx += 2, ++UseNo) { |
4366 | MachineBasicBlock &OpMBB = *MI.getOperand(i: UseIdx + 1).getMBB(); |
4367 | MIRBuilder.setInsertPt(MBB&: OpMBB, II: OpMBB.getFirstTerminatorForward()); |
4368 | extractVectorParts(Reg: MI.getReg(Idx: UseIdx), NumElts, VRegs&: InputOpsPieces[UseNo], |
4369 | MIRBuilder, MRI); |
4370 | } |
4371 | |
4372 | // Build PHIs with fewer elements. |
4373 | unsigned NumLeftovers = OrigNumElts % NumElts ? 1 : 0; |
4374 | MIRBuilder.setInsertPt(MBB&: *MI.getParent(), II: MI); |
4375 | for (unsigned i = 0; i < OrigNumElts / NumElts + NumLeftovers; ++i) { |
4376 | auto Phi = MIRBuilder.buildInstr(Opcode: TargetOpcode::G_PHI); |
4377 | Phi.addDef( |
4378 | RegNo: MRI.createGenericVirtualRegister(Ty: OutputOpsPieces[i].getLLTTy(MRI))); |
4379 | OutputRegs.push_back(Elt: Phi.getReg(Idx: 0)); |
4380 | |
4381 | for (unsigned j = 0; j < NumInputs / 2; ++j) { |
4382 | Phi.addUse(RegNo: InputOpsPieces[j][i]); |
4383 | Phi.add(MO: MI.getOperand(i: 1 + j * 2 + 1)); |
4384 | } |
4385 | } |
4386 | |
4387 | // Set the insert point after the existing PHIs |
4388 | MachineBasicBlock &MBB = *MI.getParent(); |
4389 | MIRBuilder.setInsertPt(MBB, II: MBB.getFirstNonPHI()); |
4390 | |
4391 | // Merge small outputs into MI's def. |
4392 | if (NumLeftovers) { |
4393 | mergeMixedSubvectors(DstReg: MI.getReg(Idx: 0), PartRegs: OutputRegs); |
4394 | } else { |
4395 | MIRBuilder.buildMergeLikeInstr(Res: MI.getReg(Idx: 0), Ops: OutputRegs); |
4396 | } |
4397 | |
4398 | MI.eraseFromParent(); |
4399 | return Legalized; |
4400 | } |
4401 | |
4402 | LegalizerHelper::LegalizeResult |
4403 | LegalizerHelper::fewerElementsVectorUnmergeValues(MachineInstr &MI, |
4404 | unsigned TypeIdx, |
4405 | LLT NarrowTy) { |
4406 | const int NumDst = MI.getNumOperands() - 1; |
4407 | const Register SrcReg = MI.getOperand(i: NumDst).getReg(); |
4408 | LLT DstTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
4409 | LLT SrcTy = MRI.getType(Reg: SrcReg); |
4410 | |
4411 | if (TypeIdx != 1 || NarrowTy == DstTy) |
4412 | return UnableToLegalize; |
4413 | |
4414 | // Requires compatible types. Otherwise SrcReg should have been defined by |
4415 | // merge-like instruction that would get artifact combined. Most likely |
4416 | // instruction that defines SrcReg has to perform more/fewer elements |
4417 | // legalization compatible with NarrowTy. |
4418 | assert(SrcTy.isVector() && NarrowTy.isVector() && "Expected vector types" ); |
4419 | assert((SrcTy.getScalarType() == NarrowTy.getScalarType()) && "bad type" ); |
4420 | |
4421 | if ((SrcTy.getSizeInBits() % NarrowTy.getSizeInBits() != 0) || |
4422 | (NarrowTy.getSizeInBits() % DstTy.getSizeInBits() != 0)) |
4423 | return UnableToLegalize; |
4424 | |
4425 | // This is most likely DstTy (smaller then register size) packed in SrcTy |
4426 | // (larger then register size) and since unmerge was not combined it will be |
4427 | // lowered to bit sequence extracts from register. Unpack SrcTy to NarrowTy |
4428 | // (register size) pieces first. Then unpack each of NarrowTy pieces to DstTy. |
4429 | |
4430 | // %1:_(DstTy), %2, %3, %4 = G_UNMERGE_VALUES %0:_(SrcTy) |
4431 | // |
4432 | // %5:_(NarrowTy), %6 = G_UNMERGE_VALUES %0:_(SrcTy) - reg sequence |
4433 | // %1:_(DstTy), %2 = G_UNMERGE_VALUES %5:_(NarrowTy) - sequence of bits in reg |
4434 | // %3:_(DstTy), %4 = G_UNMERGE_VALUES %6:_(NarrowTy) |
4435 | auto Unmerge = MIRBuilder.buildUnmerge(Res: NarrowTy, Op: SrcReg); |
4436 | const int NumUnmerge = Unmerge->getNumOperands() - 1; |
4437 | const int PartsPerUnmerge = NumDst / NumUnmerge; |
4438 | |
4439 | for (int I = 0; I != NumUnmerge; ++I) { |
4440 | auto MIB = MIRBuilder.buildInstr(Opcode: TargetOpcode::G_UNMERGE_VALUES); |
4441 | |
4442 | for (int J = 0; J != PartsPerUnmerge; ++J) |
4443 | MIB.addDef(RegNo: MI.getOperand(i: I * PartsPerUnmerge + J).getReg()); |
4444 | MIB.addUse(RegNo: Unmerge.getReg(Idx: I)); |
4445 | } |
4446 | |
4447 | MI.eraseFromParent(); |
4448 | return Legalized; |
4449 | } |
4450 | |
4451 | LegalizerHelper::LegalizeResult |
4452 | LegalizerHelper::fewerElementsVectorMerge(MachineInstr &MI, unsigned TypeIdx, |
4453 | LLT NarrowTy) { |
4454 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
4455 | // Requires compatible types. Otherwise user of DstReg did not perform unmerge |
4456 | // that should have been artifact combined. Most likely instruction that uses |
4457 | // DstReg has to do more/fewer elements legalization compatible with NarrowTy. |
4458 | assert(DstTy.isVector() && NarrowTy.isVector() && "Expected vector types" ); |
4459 | assert((DstTy.getScalarType() == NarrowTy.getScalarType()) && "bad type" ); |
4460 | if (NarrowTy == SrcTy) |
4461 | return UnableToLegalize; |
4462 | |
4463 | // This attempts to lower part of LCMTy merge/unmerge sequence. Intended use |
4464 | // is for old mir tests. Since the changes to more/fewer elements it should no |
4465 | // longer be possible to generate MIR like this when starting from llvm-ir |
4466 | // because LCMTy approach was replaced with merge/unmerge to vector elements. |
4467 | if (TypeIdx == 1) { |
4468 | assert(SrcTy.isVector() && "Expected vector types" ); |
4469 | assert((SrcTy.getScalarType() == NarrowTy.getScalarType()) && "bad type" ); |
4470 | if ((DstTy.getSizeInBits() % NarrowTy.getSizeInBits() != 0) || |
4471 | (NarrowTy.getNumElements() >= SrcTy.getNumElements())) |
4472 | return UnableToLegalize; |
4473 | // %2:_(DstTy) = G_CONCAT_VECTORS %0:_(SrcTy), %1:_(SrcTy) |
4474 | // |
4475 | // %3:_(EltTy), %4, %5 = G_UNMERGE_VALUES %0:_(SrcTy) |
4476 | // %6:_(EltTy), %7, %8 = G_UNMERGE_VALUES %1:_(SrcTy) |
4477 | // %9:_(NarrowTy) = G_BUILD_VECTOR %3:_(EltTy), %4 |
4478 | // %10:_(NarrowTy) = G_BUILD_VECTOR %5:_(EltTy), %6 |
4479 | // %11:_(NarrowTy) = G_BUILD_VECTOR %7:_(EltTy), %8 |
4480 | // %2:_(DstTy) = G_CONCAT_VECTORS %9:_(NarrowTy), %10, %11 |
4481 | |
4482 | SmallVector<Register, 8> Elts; |
4483 | LLT EltTy = MRI.getType(Reg: MI.getOperand(i: 1).getReg()).getScalarType(); |
4484 | for (unsigned i = 1; i < MI.getNumOperands(); ++i) { |
4485 | auto Unmerge = MIRBuilder.buildUnmerge(Res: EltTy, Op: MI.getOperand(i).getReg()); |
4486 | for (unsigned j = 0; j < Unmerge->getNumDefs(); ++j) |
4487 | Elts.push_back(Elt: Unmerge.getReg(Idx: j)); |
4488 | } |
4489 | |
4490 | SmallVector<Register, 8> NarrowTyElts; |
4491 | unsigned NumNarrowTyElts = NarrowTy.getNumElements(); |
4492 | unsigned NumNarrowTyPieces = DstTy.getNumElements() / NumNarrowTyElts; |
4493 | for (unsigned i = 0, Offset = 0; i < NumNarrowTyPieces; |
4494 | ++i, Offset += NumNarrowTyElts) { |
4495 | ArrayRef<Register> Pieces(&Elts[Offset], NumNarrowTyElts); |
4496 | NarrowTyElts.push_back( |
4497 | Elt: MIRBuilder.buildMergeLikeInstr(Res: NarrowTy, Ops: Pieces).getReg(Idx: 0)); |
4498 | } |
4499 | |
4500 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: NarrowTyElts); |
4501 | MI.eraseFromParent(); |
4502 | return Legalized; |
4503 | } |
4504 | |
4505 | assert(TypeIdx == 0 && "Bad type index" ); |
4506 | if ((NarrowTy.getSizeInBits() % SrcTy.getSizeInBits() != 0) || |
4507 | (DstTy.getSizeInBits() % NarrowTy.getSizeInBits() != 0)) |
4508 | return UnableToLegalize; |
4509 | |
4510 | // This is most likely SrcTy (smaller then register size) packed in DstTy |
4511 | // (larger then register size) and since merge was not combined it will be |
4512 | // lowered to bit sequence packing into register. Merge SrcTy to NarrowTy |
4513 | // (register size) pieces first. Then merge each of NarrowTy pieces to DstTy. |
4514 | |
4515 | // %0:_(DstTy) = G_MERGE_VALUES %1:_(SrcTy), %2, %3, %4 |
4516 | // |
4517 | // %5:_(NarrowTy) = G_MERGE_VALUES %1:_(SrcTy), %2 - sequence of bits in reg |
4518 | // %6:_(NarrowTy) = G_MERGE_VALUES %3:_(SrcTy), %4 |
4519 | // %0:_(DstTy) = G_MERGE_VALUES %5:_(NarrowTy), %6 - reg sequence |
4520 | SmallVector<Register, 8> NarrowTyElts; |
4521 | unsigned NumParts = DstTy.getNumElements() / NarrowTy.getNumElements(); |
4522 | unsigned NumSrcElts = SrcTy.isVector() ? SrcTy.getNumElements() : 1; |
4523 | unsigned NumElts = NarrowTy.getNumElements() / NumSrcElts; |
4524 | for (unsigned i = 0; i < NumParts; ++i) { |
4525 | SmallVector<Register, 8> Sources; |
4526 | for (unsigned j = 0; j < NumElts; ++j) |
4527 | Sources.push_back(Elt: MI.getOperand(i: 1 + i * NumElts + j).getReg()); |
4528 | NarrowTyElts.push_back( |
4529 | Elt: MIRBuilder.buildMergeLikeInstr(Res: NarrowTy, Ops: Sources).getReg(Idx: 0)); |
4530 | } |
4531 | |
4532 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: NarrowTyElts); |
4533 | MI.eraseFromParent(); |
4534 | return Legalized; |
4535 | } |
4536 | |
4537 | LegalizerHelper::LegalizeResult |
4538 | LegalizerHelper::(MachineInstr &MI, |
4539 | unsigned TypeIdx, |
4540 | LLT NarrowVecTy) { |
4541 | auto [DstReg, SrcVec] = MI.getFirst2Regs(); |
4542 | Register InsertVal; |
4543 | bool IsInsert = MI.getOpcode() == TargetOpcode::G_INSERT_VECTOR_ELT; |
4544 | |
4545 | assert((IsInsert ? TypeIdx == 0 : TypeIdx == 1) && "not a vector type index" ); |
4546 | if (IsInsert) |
4547 | InsertVal = MI.getOperand(i: 2).getReg(); |
4548 | |
4549 | Register Idx = MI.getOperand(i: MI.getNumOperands() - 1).getReg(); |
4550 | |
4551 | // TODO: Handle total scalarization case. |
4552 | if (!NarrowVecTy.isVector()) |
4553 | return UnableToLegalize; |
4554 | |
4555 | LLT VecTy = MRI.getType(Reg: SrcVec); |
4556 | |
4557 | // If the index is a constant, we can really break this down as you would |
4558 | // expect, and index into the target size pieces. |
4559 | int64_t IdxVal; |
4560 | auto MaybeCst = getIConstantVRegValWithLookThrough(VReg: Idx, MRI); |
4561 | if (MaybeCst) { |
4562 | IdxVal = MaybeCst->Value.getSExtValue(); |
4563 | // Avoid out of bounds indexing the pieces. |
4564 | if (IdxVal >= VecTy.getNumElements()) { |
4565 | MIRBuilder.buildUndef(Res: DstReg); |
4566 | MI.eraseFromParent(); |
4567 | return Legalized; |
4568 | } |
4569 | |
4570 | SmallVector<Register, 8> VecParts; |
4571 | LLT GCDTy = extractGCDType(Parts&: VecParts, DstTy: VecTy, NarrowTy: NarrowVecTy, SrcReg: SrcVec); |
4572 | |
4573 | // Build a sequence of NarrowTy pieces in VecParts for this operand. |
4574 | LLT LCMTy = buildLCMMergePieces(DstTy: VecTy, NarrowTy: NarrowVecTy, GCDTy, VRegs&: VecParts, |
4575 | PadStrategy: TargetOpcode::G_ANYEXT); |
4576 | |
4577 | unsigned NewNumElts = NarrowVecTy.getNumElements(); |
4578 | |
4579 | LLT IdxTy = MRI.getType(Reg: Idx); |
4580 | int64_t PartIdx = IdxVal / NewNumElts; |
4581 | auto NewIdx = |
4582 | MIRBuilder.buildConstant(Res: IdxTy, Val: IdxVal - NewNumElts * PartIdx); |
4583 | |
4584 | if (IsInsert) { |
4585 | LLT PartTy = MRI.getType(Reg: VecParts[PartIdx]); |
4586 | |
4587 | // Use the adjusted index to insert into one of the subvectors. |
4588 | auto InsertPart = MIRBuilder.buildInsertVectorElement( |
4589 | Res: PartTy, Val: VecParts[PartIdx], Elt: InsertVal, Idx: NewIdx); |
4590 | VecParts[PartIdx] = InsertPart.getReg(Idx: 0); |
4591 | |
4592 | // Recombine the inserted subvector with the others to reform the result |
4593 | // vector. |
4594 | buildWidenedRemergeToDst(DstReg, LCMTy, RemergeRegs: VecParts); |
4595 | } else { |
4596 | MIRBuilder.buildExtractVectorElement(Res: DstReg, Val: VecParts[PartIdx], Idx: NewIdx); |
4597 | } |
4598 | |
4599 | MI.eraseFromParent(); |
4600 | return Legalized; |
4601 | } |
4602 | |
4603 | // With a variable index, we can't perform the operation in a smaller type, so |
4604 | // we're forced to expand this. |
4605 | // |
4606 | // TODO: We could emit a chain of compare/select to figure out which piece to |
4607 | // index. |
4608 | return lowerExtractInsertVectorElt(MI); |
4609 | } |
4610 | |
4611 | LegalizerHelper::LegalizeResult |
4612 | LegalizerHelper::reduceLoadStoreWidth(GLoadStore &LdStMI, unsigned TypeIdx, |
4613 | LLT NarrowTy) { |
4614 | // FIXME: Don't know how to handle secondary types yet. |
4615 | if (TypeIdx != 0) |
4616 | return UnableToLegalize; |
4617 | |
4618 | // This implementation doesn't work for atomics. Give up instead of doing |
4619 | // something invalid. |
4620 | if (LdStMI.isAtomic()) |
4621 | return UnableToLegalize; |
4622 | |
4623 | bool IsLoad = isa<GLoad>(Val: LdStMI); |
4624 | Register ValReg = LdStMI.getReg(Idx: 0); |
4625 | Register AddrReg = LdStMI.getPointerReg(); |
4626 | LLT ValTy = MRI.getType(Reg: ValReg); |
4627 | |
4628 | // FIXME: Do we need a distinct NarrowMemory legalize action? |
4629 | if (ValTy.getSizeInBits() != 8 * LdStMI.getMemSize().getValue()) { |
4630 | LLVM_DEBUG(dbgs() << "Can't narrow extload/truncstore\n" ); |
4631 | return UnableToLegalize; |
4632 | } |
4633 | |
4634 | int NumParts = -1; |
4635 | int NumLeftover = -1; |
4636 | LLT LeftoverTy; |
4637 | SmallVector<Register, 8> NarrowRegs, NarrowLeftoverRegs; |
4638 | if (IsLoad) { |
4639 | std::tie(args&: NumParts, args&: NumLeftover) = getNarrowTypeBreakDown(OrigTy: ValTy, NarrowTy, LeftoverTy); |
4640 | } else { |
4641 | if (extractParts(Reg: ValReg, RegTy: ValTy, MainTy: NarrowTy, LeftoverTy, VRegs&: NarrowRegs, |
4642 | LeftoverVRegs&: NarrowLeftoverRegs, MIRBuilder, MRI)) { |
4643 | NumParts = NarrowRegs.size(); |
4644 | NumLeftover = NarrowLeftoverRegs.size(); |
4645 | } |
4646 | } |
4647 | |
4648 | if (NumParts == -1) |
4649 | return UnableToLegalize; |
4650 | |
4651 | LLT PtrTy = MRI.getType(Reg: AddrReg); |
4652 | const LLT OffsetTy = LLT::scalar(SizeInBits: PtrTy.getSizeInBits()); |
4653 | |
4654 | unsigned TotalSize = ValTy.getSizeInBits(); |
4655 | |
4656 | // Split the load/store into PartTy sized pieces starting at Offset. If this |
4657 | // is a load, return the new registers in ValRegs. For a store, each elements |
4658 | // of ValRegs should be PartTy. Returns the next offset that needs to be |
4659 | // handled. |
4660 | bool isBigEndian = MIRBuilder.getDataLayout().isBigEndian(); |
4661 | auto MMO = LdStMI.getMMO(); |
4662 | auto splitTypePieces = [=](LLT PartTy, SmallVectorImpl<Register> &ValRegs, |
4663 | unsigned NumParts, unsigned Offset) -> unsigned { |
4664 | MachineFunction &MF = MIRBuilder.getMF(); |
4665 | unsigned PartSize = PartTy.getSizeInBits(); |
4666 | for (unsigned Idx = 0, E = NumParts; Idx != E && Offset < TotalSize; |
4667 | ++Idx) { |
4668 | unsigned ByteOffset = Offset / 8; |
4669 | Register NewAddrReg; |
4670 | |
4671 | MIRBuilder.materializePtrAdd(Res&: NewAddrReg, Op0: AddrReg, ValueTy: OffsetTy, Value: ByteOffset); |
4672 | |
4673 | MachineMemOperand *NewMMO = |
4674 | MF.getMachineMemOperand(MMO: &MMO, Offset: ByteOffset, Ty: PartTy); |
4675 | |
4676 | if (IsLoad) { |
4677 | Register Dst = MRI.createGenericVirtualRegister(Ty: PartTy); |
4678 | ValRegs.push_back(Elt: Dst); |
4679 | MIRBuilder.buildLoad(Res: Dst, Addr: NewAddrReg, MMO&: *NewMMO); |
4680 | } else { |
4681 | MIRBuilder.buildStore(Val: ValRegs[Idx], Addr: NewAddrReg, MMO&: *NewMMO); |
4682 | } |
4683 | Offset = isBigEndian ? Offset - PartSize : Offset + PartSize; |
4684 | } |
4685 | |
4686 | return Offset; |
4687 | }; |
4688 | |
4689 | unsigned Offset = isBigEndian ? TotalSize - NarrowTy.getSizeInBits() : 0; |
4690 | unsigned HandledOffset = |
4691 | splitTypePieces(NarrowTy, NarrowRegs, NumParts, Offset); |
4692 | |
4693 | // Handle the rest of the register if this isn't an even type breakdown. |
4694 | if (LeftoverTy.isValid()) |
4695 | splitTypePieces(LeftoverTy, NarrowLeftoverRegs, NumLeftover, HandledOffset); |
4696 | |
4697 | if (IsLoad) { |
4698 | insertParts(DstReg: ValReg, ResultTy: ValTy, PartTy: NarrowTy, PartRegs: NarrowRegs, |
4699 | LeftoverTy, LeftoverRegs: NarrowLeftoverRegs); |
4700 | } |
4701 | |
4702 | LdStMI.eraseFromParent(); |
4703 | return Legalized; |
4704 | } |
4705 | |
4706 | LegalizerHelper::LegalizeResult |
4707 | LegalizerHelper::fewerElementsVector(MachineInstr &MI, unsigned TypeIdx, |
4708 | LLT NarrowTy) { |
4709 | using namespace TargetOpcode; |
4710 | GenericMachineInstr &GMI = cast<GenericMachineInstr>(Val&: MI); |
4711 | unsigned NumElts = NarrowTy.isVector() ? NarrowTy.getNumElements() : 1; |
4712 | |
4713 | switch (MI.getOpcode()) { |
4714 | case G_IMPLICIT_DEF: |
4715 | case G_TRUNC: |
4716 | case G_AND: |
4717 | case G_OR: |
4718 | case G_XOR: |
4719 | case G_ADD: |
4720 | case G_SUB: |
4721 | case G_MUL: |
4722 | case G_PTR_ADD: |
4723 | case G_SMULH: |
4724 | case G_UMULH: |
4725 | case G_FADD: |
4726 | case G_FMUL: |
4727 | case G_FSUB: |
4728 | case G_FNEG: |
4729 | case G_FABS: |
4730 | case G_FCANONICALIZE: |
4731 | case G_FDIV: |
4732 | case G_FREM: |
4733 | case G_FMA: |
4734 | case G_FMAD: |
4735 | case G_FPOW: |
4736 | case G_FEXP: |
4737 | case G_FEXP2: |
4738 | case G_FEXP10: |
4739 | case G_FLOG: |
4740 | case G_FLOG2: |
4741 | case G_FLOG10: |
4742 | case G_FLDEXP: |
4743 | case G_FNEARBYINT: |
4744 | case G_FCEIL: |
4745 | case G_FFLOOR: |
4746 | case G_FRINT: |
4747 | case G_INTRINSIC_ROUND: |
4748 | case G_INTRINSIC_ROUNDEVEN: |
4749 | case G_INTRINSIC_TRUNC: |
4750 | case G_FCOS: |
4751 | case G_FSIN: |
4752 | case G_FTAN: |
4753 | case G_FACOS: |
4754 | case G_FASIN: |
4755 | case G_FATAN: |
4756 | case G_FCOSH: |
4757 | case G_FSINH: |
4758 | case G_FTANH: |
4759 | case G_FSQRT: |
4760 | case G_BSWAP: |
4761 | case G_BITREVERSE: |
4762 | case G_SDIV: |
4763 | case G_UDIV: |
4764 | case G_SREM: |
4765 | case G_UREM: |
4766 | case G_SDIVREM: |
4767 | case G_UDIVREM: |
4768 | case G_SMIN: |
4769 | case G_SMAX: |
4770 | case G_UMIN: |
4771 | case G_UMAX: |
4772 | case G_ABS: |
4773 | case G_FMINNUM: |
4774 | case G_FMAXNUM: |
4775 | case G_FMINNUM_IEEE: |
4776 | case G_FMAXNUM_IEEE: |
4777 | case G_FMINIMUM: |
4778 | case G_FMAXIMUM: |
4779 | case G_FSHL: |
4780 | case G_FSHR: |
4781 | case G_ROTL: |
4782 | case G_ROTR: |
4783 | case G_FREEZE: |
4784 | case G_SADDSAT: |
4785 | case G_SSUBSAT: |
4786 | case G_UADDSAT: |
4787 | case G_USUBSAT: |
4788 | case G_UMULO: |
4789 | case G_SMULO: |
4790 | case G_SHL: |
4791 | case G_LSHR: |
4792 | case G_ASHR: |
4793 | case G_SSHLSAT: |
4794 | case G_USHLSAT: |
4795 | case G_CTLZ: |
4796 | case G_CTLZ_ZERO_UNDEF: |
4797 | case G_CTTZ: |
4798 | case G_CTTZ_ZERO_UNDEF: |
4799 | case G_CTPOP: |
4800 | case G_FCOPYSIGN: |
4801 | case G_ZEXT: |
4802 | case G_SEXT: |
4803 | case G_ANYEXT: |
4804 | case G_FPEXT: |
4805 | case G_FPTRUNC: |
4806 | case G_SITOFP: |
4807 | case G_UITOFP: |
4808 | case G_FPTOSI: |
4809 | case G_FPTOUI: |
4810 | case G_INTTOPTR: |
4811 | case G_PTRTOINT: |
4812 | case G_ADDRSPACE_CAST: |
4813 | case G_UADDO: |
4814 | case G_USUBO: |
4815 | case G_UADDE: |
4816 | case G_USUBE: |
4817 | case G_SADDO: |
4818 | case G_SSUBO: |
4819 | case G_SADDE: |
4820 | case G_SSUBE: |
4821 | case G_STRICT_FADD: |
4822 | case G_STRICT_FSUB: |
4823 | case G_STRICT_FMUL: |
4824 | case G_STRICT_FMA: |
4825 | case G_STRICT_FLDEXP: |
4826 | case G_FFREXP: |
4827 | return fewerElementsVectorMultiEltType(MI&: GMI, NumElts); |
4828 | case G_ICMP: |
4829 | case G_FCMP: |
4830 | return fewerElementsVectorMultiEltType(MI&: GMI, NumElts, NonVecOpIndices: {1 /*cpm predicate*/}); |
4831 | case G_IS_FPCLASS: |
4832 | return fewerElementsVectorMultiEltType(MI&: GMI, NumElts, NonVecOpIndices: {2, 3 /*mask,fpsem*/}); |
4833 | case G_SELECT: |
4834 | if (MRI.getType(Reg: MI.getOperand(i: 1).getReg()).isVector()) |
4835 | return fewerElementsVectorMultiEltType(MI&: GMI, NumElts); |
4836 | return fewerElementsVectorMultiEltType(MI&: GMI, NumElts, NonVecOpIndices: {1 /*scalar cond*/}); |
4837 | case G_PHI: |
4838 | return fewerElementsVectorPhi(MI&: GMI, NumElts); |
4839 | case G_UNMERGE_VALUES: |
4840 | return fewerElementsVectorUnmergeValues(MI, TypeIdx, NarrowTy); |
4841 | case G_BUILD_VECTOR: |
4842 | assert(TypeIdx == 0 && "not a vector type index" ); |
4843 | return fewerElementsVectorMerge(MI, TypeIdx, NarrowTy); |
4844 | case G_CONCAT_VECTORS: |
4845 | if (TypeIdx != 1) // TODO: This probably does work as expected already. |
4846 | return UnableToLegalize; |
4847 | return fewerElementsVectorMerge(MI, TypeIdx, NarrowTy); |
4848 | case G_EXTRACT_VECTOR_ELT: |
4849 | case G_INSERT_VECTOR_ELT: |
4850 | return fewerElementsVectorExtractInsertVectorElt(MI, TypeIdx, NarrowVecTy: NarrowTy); |
4851 | case G_LOAD: |
4852 | case G_STORE: |
4853 | return reduceLoadStoreWidth(LdStMI&: cast<GLoadStore>(Val&: MI), TypeIdx, NarrowTy); |
4854 | case G_SEXT_INREG: |
4855 | return fewerElementsVectorMultiEltType(MI&: GMI, NumElts, NonVecOpIndices: {2 /*imm*/}); |
4856 | GISEL_VECREDUCE_CASES_NONSEQ |
4857 | return fewerElementsVectorReductions(MI, TypeIdx, NarrowTy); |
4858 | case TargetOpcode::G_VECREDUCE_SEQ_FADD: |
4859 | case TargetOpcode::G_VECREDUCE_SEQ_FMUL: |
4860 | return fewerElementsVectorSeqReductions(MI, TypeIdx, NarrowTy); |
4861 | case G_SHUFFLE_VECTOR: |
4862 | return fewerElementsVectorShuffle(MI, TypeIdx, NarrowTy); |
4863 | case G_FPOWI: |
4864 | return fewerElementsVectorMultiEltType(MI&: GMI, NumElts, NonVecOpIndices: {2 /*pow*/}); |
4865 | case G_BITCAST: |
4866 | return fewerElementsBitcast(MI, TypeIdx, NarrowTy); |
4867 | case G_INTRINSIC_FPTRUNC_ROUND: |
4868 | return fewerElementsVectorMultiEltType(MI&: GMI, NumElts, NonVecOpIndices: {2}); |
4869 | default: |
4870 | return UnableToLegalize; |
4871 | } |
4872 | } |
4873 | |
4874 | LegalizerHelper::LegalizeResult |
4875 | LegalizerHelper::fewerElementsBitcast(MachineInstr &MI, unsigned int TypeIdx, |
4876 | LLT NarrowTy) { |
4877 | assert(MI.getOpcode() == TargetOpcode::G_BITCAST && |
4878 | "Not a bitcast operation" ); |
4879 | |
4880 | if (TypeIdx != 0) |
4881 | return UnableToLegalize; |
4882 | |
4883 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
4884 | |
4885 | unsigned SrcScalSize = SrcTy.getScalarSizeInBits(); |
4886 | LLT SrcNarrowTy = |
4887 | LLT::fixed_vector(NumElements: NarrowTy.getSizeInBits() / SrcScalSize, ScalarSizeInBits: SrcScalSize); |
4888 | |
4889 | // Split the Src and Dst Reg into smaller registers |
4890 | SmallVector<Register> SrcVRegs, BitcastVRegs; |
4891 | if (extractGCDType(Parts&: SrcVRegs, DstTy, NarrowTy: SrcNarrowTy, SrcReg) != SrcNarrowTy) |
4892 | return UnableToLegalize; |
4893 | |
4894 | // Build new smaller bitcast instructions |
4895 | // Not supporting Leftover types for now but will have to |
4896 | for (unsigned i = 0; i < SrcVRegs.size(); i++) |
4897 | BitcastVRegs.push_back( |
4898 | Elt: MIRBuilder.buildBitcast(Dst: NarrowTy, Src: SrcVRegs[i]).getReg(Idx: 0)); |
4899 | |
4900 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: BitcastVRegs); |
4901 | MI.eraseFromParent(); |
4902 | return Legalized; |
4903 | } |
4904 | |
4905 | LegalizerHelper::LegalizeResult LegalizerHelper::fewerElementsVectorShuffle( |
4906 | MachineInstr &MI, unsigned int TypeIdx, LLT NarrowTy) { |
4907 | assert(MI.getOpcode() == TargetOpcode::G_SHUFFLE_VECTOR); |
4908 | if (TypeIdx != 0) |
4909 | return UnableToLegalize; |
4910 | |
4911 | auto [DstReg, DstTy, Src1Reg, Src1Ty, Src2Reg, Src2Ty] = |
4912 | MI.getFirst3RegLLTs(); |
4913 | ArrayRef<int> Mask = MI.getOperand(i: 3).getShuffleMask(); |
4914 | // The shuffle should be canonicalized by now. |
4915 | if (DstTy != Src1Ty) |
4916 | return UnableToLegalize; |
4917 | if (DstTy != Src2Ty) |
4918 | return UnableToLegalize; |
4919 | |
4920 | if (!isPowerOf2_32(Value: DstTy.getNumElements())) |
4921 | return UnableToLegalize; |
4922 | |
4923 | // We only support splitting a shuffle into 2, so adjust NarrowTy accordingly. |
4924 | // Further legalization attempts will be needed to do split further. |
4925 | NarrowTy = |
4926 | DstTy.changeElementCount(EC: DstTy.getElementCount().divideCoefficientBy(RHS: 2)); |
4927 | unsigned NewElts = NarrowTy.getNumElements(); |
4928 | |
4929 | SmallVector<Register> SplitSrc1Regs, SplitSrc2Regs; |
4930 | extractParts(Reg: Src1Reg, Ty: NarrowTy, NumParts: 2, VRegs&: SplitSrc1Regs, MIRBuilder, MRI); |
4931 | extractParts(Reg: Src2Reg, Ty: NarrowTy, NumParts: 2, VRegs&: SplitSrc2Regs, MIRBuilder, MRI); |
4932 | Register Inputs[4] = {SplitSrc1Regs[0], SplitSrc1Regs[1], SplitSrc2Regs[0], |
4933 | SplitSrc2Regs[1]}; |
4934 | |
4935 | Register Hi, Lo; |
4936 | |
4937 | // If Lo or Hi uses elements from at most two of the four input vectors, then |
4938 | // express it as a vector shuffle of those two inputs. Otherwise extract the |
4939 | // input elements by hand and construct the Lo/Hi output using a BUILD_VECTOR. |
4940 | SmallVector<int, 16> Ops; |
4941 | for (unsigned High = 0; High < 2; ++High) { |
4942 | Register &Output = High ? Hi : Lo; |
4943 | |
4944 | // Build a shuffle mask for the output, discovering on the fly which |
4945 | // input vectors to use as shuffle operands (recorded in InputUsed). |
4946 | // If building a suitable shuffle vector proves too hard, then bail |
4947 | // out with useBuildVector set. |
4948 | unsigned InputUsed[2] = {-1U, -1U}; // Not yet discovered. |
4949 | unsigned FirstMaskIdx = High * NewElts; |
4950 | bool UseBuildVector = false; |
4951 | for (unsigned MaskOffset = 0; MaskOffset < NewElts; ++MaskOffset) { |
4952 | // The mask element. This indexes into the input. |
4953 | int Idx = Mask[FirstMaskIdx + MaskOffset]; |
4954 | |
4955 | // The input vector this mask element indexes into. |
4956 | unsigned Input = (unsigned)Idx / NewElts; |
4957 | |
4958 | if (Input >= std::size(Inputs)) { |
4959 | // The mask element does not index into any input vector. |
4960 | Ops.push_back(Elt: -1); |
4961 | continue; |
4962 | } |
4963 | |
4964 | // Turn the index into an offset from the start of the input vector. |
4965 | Idx -= Input * NewElts; |
4966 | |
4967 | // Find or create a shuffle vector operand to hold this input. |
4968 | unsigned OpNo; |
4969 | for (OpNo = 0; OpNo < std::size(InputUsed); ++OpNo) { |
4970 | if (InputUsed[OpNo] == Input) { |
4971 | // This input vector is already an operand. |
4972 | break; |
4973 | } else if (InputUsed[OpNo] == -1U) { |
4974 | // Create a new operand for this input vector. |
4975 | InputUsed[OpNo] = Input; |
4976 | break; |
4977 | } |
4978 | } |
4979 | |
4980 | if (OpNo >= std::size(InputUsed)) { |
4981 | // More than two input vectors used! Give up on trying to create a |
4982 | // shuffle vector. Insert all elements into a BUILD_VECTOR instead. |
4983 | UseBuildVector = true; |
4984 | break; |
4985 | } |
4986 | |
4987 | // Add the mask index for the new shuffle vector. |
4988 | Ops.push_back(Elt: Idx + OpNo * NewElts); |
4989 | } |
4990 | |
4991 | if (UseBuildVector) { |
4992 | LLT EltTy = NarrowTy.getElementType(); |
4993 | SmallVector<Register, 16> SVOps; |
4994 | |
4995 | // Extract the input elements by hand. |
4996 | for (unsigned MaskOffset = 0; MaskOffset < NewElts; ++MaskOffset) { |
4997 | // The mask element. This indexes into the input. |
4998 | int Idx = Mask[FirstMaskIdx + MaskOffset]; |
4999 | |
5000 | // The input vector this mask element indexes into. |
5001 | unsigned Input = (unsigned)Idx / NewElts; |
5002 | |
5003 | if (Input >= std::size(Inputs)) { |
5004 | // The mask element is "undef" or indexes off the end of the input. |
5005 | SVOps.push_back(Elt: MIRBuilder.buildUndef(Res: EltTy).getReg(Idx: 0)); |
5006 | continue; |
5007 | } |
5008 | |
5009 | // Turn the index into an offset from the start of the input vector. |
5010 | Idx -= Input * NewElts; |
5011 | |
5012 | // Extract the vector element by hand. |
5013 | SVOps.push_back(Elt: MIRBuilder |
5014 | .buildExtractVectorElement( |
5015 | Res: EltTy, Val: Inputs[Input], |
5016 | Idx: MIRBuilder.buildConstant(Res: LLT::scalar(SizeInBits: 32), Val: Idx)) |
5017 | .getReg(Idx: 0)); |
5018 | } |
5019 | |
5020 | // Construct the Lo/Hi output using a G_BUILD_VECTOR. |
5021 | Output = MIRBuilder.buildBuildVector(Res: NarrowTy, Ops: SVOps).getReg(Idx: 0); |
5022 | } else if (InputUsed[0] == -1U) { |
5023 | // No input vectors were used! The result is undefined. |
5024 | Output = MIRBuilder.buildUndef(Res: NarrowTy).getReg(Idx: 0); |
5025 | } else { |
5026 | Register Op0 = Inputs[InputUsed[0]]; |
5027 | // If only one input was used, use an undefined vector for the other. |
5028 | Register Op1 = InputUsed[1] == -1U |
5029 | ? MIRBuilder.buildUndef(Res: NarrowTy).getReg(Idx: 0) |
5030 | : Inputs[InputUsed[1]]; |
5031 | // At least one input vector was used. Create a new shuffle vector. |
5032 | Output = MIRBuilder.buildShuffleVector(Res: NarrowTy, Src1: Op0, Src2: Op1, Mask: Ops).getReg(Idx: 0); |
5033 | } |
5034 | |
5035 | Ops.clear(); |
5036 | } |
5037 | |
5038 | MIRBuilder.buildConcatVectors(Res: DstReg, Ops: {Lo, Hi}); |
5039 | MI.eraseFromParent(); |
5040 | return Legalized; |
5041 | } |
5042 | |
5043 | LegalizerHelper::LegalizeResult LegalizerHelper::fewerElementsVectorReductions( |
5044 | MachineInstr &MI, unsigned int TypeIdx, LLT NarrowTy) { |
5045 | auto &RdxMI = cast<GVecReduce>(Val&: MI); |
5046 | |
5047 | if (TypeIdx != 1) |
5048 | return UnableToLegalize; |
5049 | |
5050 | // The semantics of the normal non-sequential reductions allow us to freely |
5051 | // re-associate the operation. |
5052 | auto [DstReg, DstTy, SrcReg, SrcTy] = RdxMI.getFirst2RegLLTs(); |
5053 | |
5054 | if (NarrowTy.isVector() && |
5055 | (SrcTy.getNumElements() % NarrowTy.getNumElements() != 0)) |
5056 | return UnableToLegalize; |
5057 | |
5058 | unsigned ScalarOpc = RdxMI.getScalarOpcForReduction(); |
5059 | SmallVector<Register> SplitSrcs; |
5060 | // If NarrowTy is a scalar then we're being asked to scalarize. |
5061 | const unsigned NumParts = |
5062 | NarrowTy.isVector() ? SrcTy.getNumElements() / NarrowTy.getNumElements() |
5063 | : SrcTy.getNumElements(); |
5064 | |
5065 | extractParts(Reg: SrcReg, Ty: NarrowTy, NumParts, VRegs&: SplitSrcs, MIRBuilder, MRI); |
5066 | if (NarrowTy.isScalar()) { |
5067 | if (DstTy != NarrowTy) |
5068 | return UnableToLegalize; // FIXME: handle implicit extensions. |
5069 | |
5070 | if (isPowerOf2_32(Value: NumParts)) { |
5071 | // Generate a tree of scalar operations to reduce the critical path. |
5072 | SmallVector<Register> PartialResults; |
5073 | unsigned NumPartsLeft = NumParts; |
5074 | while (NumPartsLeft > 1) { |
5075 | for (unsigned Idx = 0; Idx < NumPartsLeft - 1; Idx += 2) { |
5076 | PartialResults.emplace_back( |
5077 | Args: MIRBuilder |
5078 | .buildInstr(Opc: ScalarOpc, DstOps: {NarrowTy}, |
5079 | SrcOps: {SplitSrcs[Idx], SplitSrcs[Idx + 1]}) |
5080 | .getReg(Idx: 0)); |
5081 | } |
5082 | SplitSrcs = PartialResults; |
5083 | PartialResults.clear(); |
5084 | NumPartsLeft = SplitSrcs.size(); |
5085 | } |
5086 | assert(SplitSrcs.size() == 1); |
5087 | MIRBuilder.buildCopy(Res: DstReg, Op: SplitSrcs[0]); |
5088 | MI.eraseFromParent(); |
5089 | return Legalized; |
5090 | } |
5091 | // If we can't generate a tree, then just do sequential operations. |
5092 | Register Acc = SplitSrcs[0]; |
5093 | for (unsigned Idx = 1; Idx < NumParts; ++Idx) |
5094 | Acc = MIRBuilder.buildInstr(Opc: ScalarOpc, DstOps: {NarrowTy}, SrcOps: {Acc, SplitSrcs[Idx]}) |
5095 | .getReg(Idx: 0); |
5096 | MIRBuilder.buildCopy(Res: DstReg, Op: Acc); |
5097 | MI.eraseFromParent(); |
5098 | return Legalized; |
5099 | } |
5100 | SmallVector<Register> PartialReductions; |
5101 | for (unsigned Part = 0; Part < NumParts; ++Part) { |
5102 | PartialReductions.push_back( |
5103 | Elt: MIRBuilder.buildInstr(Opc: RdxMI.getOpcode(), DstOps: {DstTy}, SrcOps: {SplitSrcs[Part]}) |
5104 | .getReg(Idx: 0)); |
5105 | } |
5106 | |
5107 | // If the types involved are powers of 2, we can generate intermediate vector |
5108 | // ops, before generating a final reduction operation. |
5109 | if (isPowerOf2_32(Value: SrcTy.getNumElements()) && |
5110 | isPowerOf2_32(Value: NarrowTy.getNumElements())) { |
5111 | return tryNarrowPow2Reduction(MI, SrcReg, SrcTy, NarrowTy, ScalarOpc); |
5112 | } |
5113 | |
5114 | Register Acc = PartialReductions[0]; |
5115 | for (unsigned Part = 1; Part < NumParts; ++Part) { |
5116 | if (Part == NumParts - 1) { |
5117 | MIRBuilder.buildInstr(Opc: ScalarOpc, DstOps: {DstReg}, |
5118 | SrcOps: {Acc, PartialReductions[Part]}); |
5119 | } else { |
5120 | Acc = MIRBuilder |
5121 | .buildInstr(Opc: ScalarOpc, DstOps: {DstTy}, SrcOps: {Acc, PartialReductions[Part]}) |
5122 | .getReg(Idx: 0); |
5123 | } |
5124 | } |
5125 | MI.eraseFromParent(); |
5126 | return Legalized; |
5127 | } |
5128 | |
5129 | LegalizerHelper::LegalizeResult |
5130 | LegalizerHelper::fewerElementsVectorSeqReductions(MachineInstr &MI, |
5131 | unsigned int TypeIdx, |
5132 | LLT NarrowTy) { |
5133 | auto [DstReg, DstTy, ScalarReg, ScalarTy, SrcReg, SrcTy] = |
5134 | MI.getFirst3RegLLTs(); |
5135 | if (!NarrowTy.isScalar() || TypeIdx != 2 || DstTy != ScalarTy || |
5136 | DstTy != NarrowTy) |
5137 | return UnableToLegalize; |
5138 | |
5139 | assert((MI.getOpcode() == TargetOpcode::G_VECREDUCE_SEQ_FADD || |
5140 | MI.getOpcode() == TargetOpcode::G_VECREDUCE_SEQ_FMUL) && |
5141 | "Unexpected vecreduce opcode" ); |
5142 | unsigned ScalarOpc = MI.getOpcode() == TargetOpcode::G_VECREDUCE_SEQ_FADD |
5143 | ? TargetOpcode::G_FADD |
5144 | : TargetOpcode::G_FMUL; |
5145 | |
5146 | SmallVector<Register> SplitSrcs; |
5147 | unsigned NumParts = SrcTy.getNumElements(); |
5148 | extractParts(Reg: SrcReg, Ty: NarrowTy, NumParts, VRegs&: SplitSrcs, MIRBuilder, MRI); |
5149 | Register Acc = ScalarReg; |
5150 | for (unsigned i = 0; i < NumParts; i++) |
5151 | Acc = MIRBuilder.buildInstr(Opc: ScalarOpc, DstOps: {NarrowTy}, SrcOps: {Acc, SplitSrcs[i]}) |
5152 | .getReg(Idx: 0); |
5153 | |
5154 | MIRBuilder.buildCopy(Res: DstReg, Op: Acc); |
5155 | MI.eraseFromParent(); |
5156 | return Legalized; |
5157 | } |
5158 | |
5159 | LegalizerHelper::LegalizeResult |
5160 | LegalizerHelper::tryNarrowPow2Reduction(MachineInstr &MI, Register SrcReg, |
5161 | LLT SrcTy, LLT NarrowTy, |
5162 | unsigned ScalarOpc) { |
5163 | SmallVector<Register> SplitSrcs; |
5164 | // Split the sources into NarrowTy size pieces. |
5165 | extractParts(Reg: SrcReg, Ty: NarrowTy, |
5166 | NumParts: SrcTy.getNumElements() / NarrowTy.getNumElements(), VRegs&: SplitSrcs, |
5167 | MIRBuilder, MRI); |
5168 | // We're going to do a tree reduction using vector operations until we have |
5169 | // one NarrowTy size value left. |
5170 | while (SplitSrcs.size() > 1) { |
5171 | SmallVector<Register> PartialRdxs; |
5172 | for (unsigned Idx = 0; Idx < SplitSrcs.size()-1; Idx += 2) { |
5173 | Register LHS = SplitSrcs[Idx]; |
5174 | Register RHS = SplitSrcs[Idx + 1]; |
5175 | // Create the intermediate vector op. |
5176 | Register Res = |
5177 | MIRBuilder.buildInstr(Opc: ScalarOpc, DstOps: {NarrowTy}, SrcOps: {LHS, RHS}).getReg(Idx: 0); |
5178 | PartialRdxs.push_back(Elt: Res); |
5179 | } |
5180 | SplitSrcs = std::move(PartialRdxs); |
5181 | } |
5182 | // Finally generate the requested NarrowTy based reduction. |
5183 | Observer.changingInstr(MI); |
5184 | MI.getOperand(i: 1).setReg(SplitSrcs[0]); |
5185 | Observer.changedInstr(MI); |
5186 | return Legalized; |
5187 | } |
5188 | |
5189 | LegalizerHelper::LegalizeResult |
5190 | LegalizerHelper::narrowScalarShiftByConstant(MachineInstr &MI, const APInt &Amt, |
5191 | const LLT HalfTy, const LLT AmtTy) { |
5192 | |
5193 | Register InL = MRI.createGenericVirtualRegister(Ty: HalfTy); |
5194 | Register InH = MRI.createGenericVirtualRegister(Ty: HalfTy); |
5195 | MIRBuilder.buildUnmerge(Res: {InL, InH}, Op: MI.getOperand(i: 1)); |
5196 | |
5197 | if (Amt.isZero()) { |
5198 | MIRBuilder.buildMergeLikeInstr(Res: MI.getOperand(i: 0), Ops: {InL, InH}); |
5199 | MI.eraseFromParent(); |
5200 | return Legalized; |
5201 | } |
5202 | |
5203 | LLT NVT = HalfTy; |
5204 | unsigned NVTBits = HalfTy.getSizeInBits(); |
5205 | unsigned VTBits = 2 * NVTBits; |
5206 | |
5207 | SrcOp Lo(Register(0)), Hi(Register(0)); |
5208 | if (MI.getOpcode() == TargetOpcode::G_SHL) { |
5209 | if (Amt.ugt(RHS: VTBits)) { |
5210 | Lo = Hi = MIRBuilder.buildConstant(Res: NVT, Val: 0); |
5211 | } else if (Amt.ugt(RHS: NVTBits)) { |
5212 | Lo = MIRBuilder.buildConstant(Res: NVT, Val: 0); |
5213 | Hi = MIRBuilder.buildShl(Dst: NVT, Src0: InL, |
5214 | Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: Amt - NVTBits)); |
5215 | } else if (Amt == NVTBits) { |
5216 | Lo = MIRBuilder.buildConstant(Res: NVT, Val: 0); |
5217 | Hi = InL; |
5218 | } else { |
5219 | Lo = MIRBuilder.buildShl(Dst: NVT, Src0: InL, Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: Amt)); |
5220 | auto OrLHS = |
5221 | MIRBuilder.buildShl(Dst: NVT, Src0: InH, Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: Amt)); |
5222 | auto OrRHS = MIRBuilder.buildLShr( |
5223 | Dst: NVT, Src0: InL, Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: -Amt + NVTBits)); |
5224 | Hi = MIRBuilder.buildOr(Dst: NVT, Src0: OrLHS, Src1: OrRHS); |
5225 | } |
5226 | } else if (MI.getOpcode() == TargetOpcode::G_LSHR) { |
5227 | if (Amt.ugt(RHS: VTBits)) { |
5228 | Lo = Hi = MIRBuilder.buildConstant(Res: NVT, Val: 0); |
5229 | } else if (Amt.ugt(RHS: NVTBits)) { |
5230 | Lo = MIRBuilder.buildLShr(Dst: NVT, Src0: InH, |
5231 | Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: Amt - NVTBits)); |
5232 | Hi = MIRBuilder.buildConstant(Res: NVT, Val: 0); |
5233 | } else if (Amt == NVTBits) { |
5234 | Lo = InH; |
5235 | Hi = MIRBuilder.buildConstant(Res: NVT, Val: 0); |
5236 | } else { |
5237 | auto ShiftAmtConst = MIRBuilder.buildConstant(Res: AmtTy, Val: Amt); |
5238 | |
5239 | auto OrLHS = MIRBuilder.buildLShr(Dst: NVT, Src0: InL, Src1: ShiftAmtConst); |
5240 | auto OrRHS = MIRBuilder.buildShl( |
5241 | Dst: NVT, Src0: InH, Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: -Amt + NVTBits)); |
5242 | |
5243 | Lo = MIRBuilder.buildOr(Dst: NVT, Src0: OrLHS, Src1: OrRHS); |
5244 | Hi = MIRBuilder.buildLShr(Dst: NVT, Src0: InH, Src1: ShiftAmtConst); |
5245 | } |
5246 | } else { |
5247 | if (Amt.ugt(RHS: VTBits)) { |
5248 | Hi = Lo = MIRBuilder.buildAShr( |
5249 | Dst: NVT, Src0: InH, Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: NVTBits - 1)); |
5250 | } else if (Amt.ugt(RHS: NVTBits)) { |
5251 | Lo = MIRBuilder.buildAShr(Dst: NVT, Src0: InH, |
5252 | Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: Amt - NVTBits)); |
5253 | Hi = MIRBuilder.buildAShr(Dst: NVT, Src0: InH, |
5254 | Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: NVTBits - 1)); |
5255 | } else if (Amt == NVTBits) { |
5256 | Lo = InH; |
5257 | Hi = MIRBuilder.buildAShr(Dst: NVT, Src0: InH, |
5258 | Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: NVTBits - 1)); |
5259 | } else { |
5260 | auto ShiftAmtConst = MIRBuilder.buildConstant(Res: AmtTy, Val: Amt); |
5261 | |
5262 | auto OrLHS = MIRBuilder.buildLShr(Dst: NVT, Src0: InL, Src1: ShiftAmtConst); |
5263 | auto OrRHS = MIRBuilder.buildShl( |
5264 | Dst: NVT, Src0: InH, Src1: MIRBuilder.buildConstant(Res: AmtTy, Val: -Amt + NVTBits)); |
5265 | |
5266 | Lo = MIRBuilder.buildOr(Dst: NVT, Src0: OrLHS, Src1: OrRHS); |
5267 | Hi = MIRBuilder.buildAShr(Dst: NVT, Src0: InH, Src1: ShiftAmtConst); |
5268 | } |
5269 | } |
5270 | |
5271 | MIRBuilder.buildMergeLikeInstr(Res: MI.getOperand(i: 0), Ops: {Lo, Hi}); |
5272 | MI.eraseFromParent(); |
5273 | |
5274 | return Legalized; |
5275 | } |
5276 | |
5277 | // TODO: Optimize if constant shift amount. |
5278 | LegalizerHelper::LegalizeResult |
5279 | LegalizerHelper::narrowScalarShift(MachineInstr &MI, unsigned TypeIdx, |
5280 | LLT RequestedTy) { |
5281 | if (TypeIdx == 1) { |
5282 | Observer.changingInstr(MI); |
5283 | narrowScalarSrc(MI, NarrowTy: RequestedTy, OpIdx: 2); |
5284 | Observer.changedInstr(MI); |
5285 | return Legalized; |
5286 | } |
5287 | |
5288 | Register DstReg = MI.getOperand(i: 0).getReg(); |
5289 | LLT DstTy = MRI.getType(Reg: DstReg); |
5290 | if (DstTy.isVector()) |
5291 | return UnableToLegalize; |
5292 | |
5293 | Register Amt = MI.getOperand(i: 2).getReg(); |
5294 | LLT ShiftAmtTy = MRI.getType(Reg: Amt); |
5295 | const unsigned DstEltSize = DstTy.getScalarSizeInBits(); |
5296 | if (DstEltSize % 2 != 0) |
5297 | return UnableToLegalize; |
5298 | |
5299 | // Ignore the input type. We can only go to exactly half the size of the |
5300 | // input. If that isn't small enough, the resulting pieces will be further |
5301 | // legalized. |
5302 | const unsigned NewBitSize = DstEltSize / 2; |
5303 | const LLT HalfTy = LLT::scalar(SizeInBits: NewBitSize); |
5304 | const LLT CondTy = LLT::scalar(SizeInBits: 1); |
5305 | |
5306 | if (auto VRegAndVal = getIConstantVRegValWithLookThrough(VReg: Amt, MRI)) { |
5307 | return narrowScalarShiftByConstant(MI, Amt: VRegAndVal->Value, HalfTy, |
5308 | AmtTy: ShiftAmtTy); |
5309 | } |
5310 | |
5311 | // TODO: Expand with known bits. |
5312 | |
5313 | // Handle the fully general expansion by an unknown amount. |
5314 | auto NewBits = MIRBuilder.buildConstant(Res: ShiftAmtTy, Val: NewBitSize); |
5315 | |
5316 | Register InL = MRI.createGenericVirtualRegister(Ty: HalfTy); |
5317 | Register InH = MRI.createGenericVirtualRegister(Ty: HalfTy); |
5318 | MIRBuilder.buildUnmerge(Res: {InL, InH}, Op: MI.getOperand(i: 1)); |
5319 | |
5320 | auto AmtExcess = MIRBuilder.buildSub(Dst: ShiftAmtTy, Src0: Amt, Src1: NewBits); |
5321 | auto AmtLack = MIRBuilder.buildSub(Dst: ShiftAmtTy, Src0: NewBits, Src1: Amt); |
5322 | |
5323 | auto Zero = MIRBuilder.buildConstant(Res: ShiftAmtTy, Val: 0); |
5324 | auto IsShort = MIRBuilder.buildICmp(Pred: ICmpInst::ICMP_ULT, Res: CondTy, Op0: Amt, Op1: NewBits); |
5325 | auto IsZero = MIRBuilder.buildICmp(Pred: ICmpInst::ICMP_EQ, Res: CondTy, Op0: Amt, Op1: Zero); |
5326 | |
5327 | Register ResultRegs[2]; |
5328 | switch (MI.getOpcode()) { |
5329 | case TargetOpcode::G_SHL: { |
5330 | // Short: ShAmt < NewBitSize |
5331 | auto LoS = MIRBuilder.buildShl(Dst: HalfTy, Src0: InL, Src1: Amt); |
5332 | |
5333 | auto LoOr = MIRBuilder.buildLShr(Dst: HalfTy, Src0: InL, Src1: AmtLack); |
5334 | auto HiOr = MIRBuilder.buildShl(Dst: HalfTy, Src0: InH, Src1: Amt); |
5335 | auto HiS = MIRBuilder.buildOr(Dst: HalfTy, Src0: LoOr, Src1: HiOr); |
5336 | |
5337 | // Long: ShAmt >= NewBitSize |
5338 | auto LoL = MIRBuilder.buildConstant(Res: HalfTy, Val: 0); // Lo part is zero. |
5339 | auto HiL = MIRBuilder.buildShl(Dst: HalfTy, Src0: InL, Src1: AmtExcess); // Hi from Lo part. |
5340 | |
5341 | auto Lo = MIRBuilder.buildSelect(Res: HalfTy, Tst: IsShort, Op0: LoS, Op1: LoL); |
5342 | auto Hi = MIRBuilder.buildSelect( |
5343 | Res: HalfTy, Tst: IsZero, Op0: InH, Op1: MIRBuilder.buildSelect(Res: HalfTy, Tst: IsShort, Op0: HiS, Op1: HiL)); |
5344 | |
5345 | ResultRegs[0] = Lo.getReg(Idx: 0); |
5346 | ResultRegs[1] = Hi.getReg(Idx: 0); |
5347 | break; |
5348 | } |
5349 | case TargetOpcode::G_LSHR: |
5350 | case TargetOpcode::G_ASHR: { |
5351 | // Short: ShAmt < NewBitSize |
5352 | auto HiS = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {HalfTy}, SrcOps: {InH, Amt}); |
5353 | |
5354 | auto LoOr = MIRBuilder.buildLShr(Dst: HalfTy, Src0: InL, Src1: Amt); |
5355 | auto HiOr = MIRBuilder.buildShl(Dst: HalfTy, Src0: InH, Src1: AmtLack); |
5356 | auto LoS = MIRBuilder.buildOr(Dst: HalfTy, Src0: LoOr, Src1: HiOr); |
5357 | |
5358 | // Long: ShAmt >= NewBitSize |
5359 | MachineInstrBuilder HiL; |
5360 | if (MI.getOpcode() == TargetOpcode::G_LSHR) { |
5361 | HiL = MIRBuilder.buildConstant(Res: HalfTy, Val: 0); // Hi part is zero. |
5362 | } else { |
5363 | auto ShiftAmt = MIRBuilder.buildConstant(Res: ShiftAmtTy, Val: NewBitSize - 1); |
5364 | HiL = MIRBuilder.buildAShr(Dst: HalfTy, Src0: InH, Src1: ShiftAmt); // Sign of Hi part. |
5365 | } |
5366 | auto LoL = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {HalfTy}, |
5367 | SrcOps: {InH, AmtExcess}); // Lo from Hi part. |
5368 | |
5369 | auto Lo = MIRBuilder.buildSelect( |
5370 | Res: HalfTy, Tst: IsZero, Op0: InL, Op1: MIRBuilder.buildSelect(Res: HalfTy, Tst: IsShort, Op0: LoS, Op1: LoL)); |
5371 | |
5372 | auto Hi = MIRBuilder.buildSelect(Res: HalfTy, Tst: IsShort, Op0: HiS, Op1: HiL); |
5373 | |
5374 | ResultRegs[0] = Lo.getReg(Idx: 0); |
5375 | ResultRegs[1] = Hi.getReg(Idx: 0); |
5376 | break; |
5377 | } |
5378 | default: |
5379 | llvm_unreachable("not a shift" ); |
5380 | } |
5381 | |
5382 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: ResultRegs); |
5383 | MI.eraseFromParent(); |
5384 | return Legalized; |
5385 | } |
5386 | |
5387 | LegalizerHelper::LegalizeResult |
5388 | LegalizerHelper::moreElementsVectorPhi(MachineInstr &MI, unsigned TypeIdx, |
5389 | LLT MoreTy) { |
5390 | assert(TypeIdx == 0 && "Expecting only Idx 0" ); |
5391 | |
5392 | Observer.changingInstr(MI); |
5393 | for (unsigned I = 1, E = MI.getNumOperands(); I != E; I += 2) { |
5394 | MachineBasicBlock &OpMBB = *MI.getOperand(i: I + 1).getMBB(); |
5395 | MIRBuilder.setInsertPt(MBB&: OpMBB, II: OpMBB.getFirstTerminator()); |
5396 | moreElementsVectorSrc(MI, MoreTy, OpIdx: I); |
5397 | } |
5398 | |
5399 | MachineBasicBlock &MBB = *MI.getParent(); |
5400 | MIRBuilder.setInsertPt(MBB, II: --MBB.getFirstNonPHI()); |
5401 | moreElementsVectorDst(MI, WideTy: MoreTy, OpIdx: 0); |
5402 | Observer.changedInstr(MI); |
5403 | return Legalized; |
5404 | } |
5405 | |
5406 | MachineInstrBuilder LegalizerHelper::getNeutralElementForVecReduce( |
5407 | unsigned Opcode, MachineIRBuilder &MIRBuilder, LLT Ty) { |
5408 | assert(Ty.isScalar() && "Expected scalar type to make neutral element for" ); |
5409 | |
5410 | switch (Opcode) { |
5411 | default: |
5412 | llvm_unreachable( |
5413 | "getNeutralElementForVecReduce called with invalid opcode!" ); |
5414 | case TargetOpcode::G_VECREDUCE_ADD: |
5415 | case TargetOpcode::G_VECREDUCE_OR: |
5416 | case TargetOpcode::G_VECREDUCE_XOR: |
5417 | case TargetOpcode::G_VECREDUCE_UMAX: |
5418 | return MIRBuilder.buildConstant(Res: Ty, Val: 0); |
5419 | case TargetOpcode::G_VECREDUCE_MUL: |
5420 | return MIRBuilder.buildConstant(Res: Ty, Val: 1); |
5421 | case TargetOpcode::G_VECREDUCE_AND: |
5422 | case TargetOpcode::G_VECREDUCE_UMIN: |
5423 | return MIRBuilder.buildConstant( |
5424 | Res: Ty, Val: APInt::getAllOnes(numBits: Ty.getScalarSizeInBits())); |
5425 | case TargetOpcode::G_VECREDUCE_SMAX: |
5426 | return MIRBuilder.buildConstant( |
5427 | Res: Ty, Val: APInt::getSignedMinValue(numBits: Ty.getSizeInBits())); |
5428 | case TargetOpcode::G_VECREDUCE_SMIN: |
5429 | return MIRBuilder.buildConstant( |
5430 | Res: Ty, Val: APInt::getSignedMaxValue(numBits: Ty.getSizeInBits())); |
5431 | case TargetOpcode::G_VECREDUCE_FADD: |
5432 | return MIRBuilder.buildFConstant(Res: Ty, Val: -0.0); |
5433 | case TargetOpcode::G_VECREDUCE_FMUL: |
5434 | return MIRBuilder.buildFConstant(Res: Ty, Val: 1.0); |
5435 | case TargetOpcode::G_VECREDUCE_FMINIMUM: |
5436 | case TargetOpcode::G_VECREDUCE_FMAXIMUM: |
5437 | assert(false && "getNeutralElementForVecReduce unimplemented for " |
5438 | "G_VECREDUCE_FMINIMUM and G_VECREDUCE_FMAXIMUM!" ); |
5439 | } |
5440 | llvm_unreachable("switch expected to return!" ); |
5441 | } |
5442 | |
5443 | LegalizerHelper::LegalizeResult |
5444 | LegalizerHelper::moreElementsVector(MachineInstr &MI, unsigned TypeIdx, |
5445 | LLT MoreTy) { |
5446 | unsigned Opc = MI.getOpcode(); |
5447 | switch (Opc) { |
5448 | case TargetOpcode::G_IMPLICIT_DEF: |
5449 | case TargetOpcode::G_LOAD: { |
5450 | if (TypeIdx != 0) |
5451 | return UnableToLegalize; |
5452 | Observer.changingInstr(MI); |
5453 | moreElementsVectorDst(MI, WideTy: MoreTy, OpIdx: 0); |
5454 | Observer.changedInstr(MI); |
5455 | return Legalized; |
5456 | } |
5457 | case TargetOpcode::G_STORE: |
5458 | if (TypeIdx != 0) |
5459 | return UnableToLegalize; |
5460 | Observer.changingInstr(MI); |
5461 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 0); |
5462 | Observer.changedInstr(MI); |
5463 | return Legalized; |
5464 | case TargetOpcode::G_AND: |
5465 | case TargetOpcode::G_OR: |
5466 | case TargetOpcode::G_XOR: |
5467 | case TargetOpcode::G_ADD: |
5468 | case TargetOpcode::G_SUB: |
5469 | case TargetOpcode::G_MUL: |
5470 | case TargetOpcode::G_FADD: |
5471 | case TargetOpcode::G_FSUB: |
5472 | case TargetOpcode::G_FMUL: |
5473 | case TargetOpcode::G_FDIV: |
5474 | case TargetOpcode::G_FCOPYSIGN: |
5475 | case TargetOpcode::G_UADDSAT: |
5476 | case TargetOpcode::G_USUBSAT: |
5477 | case TargetOpcode::G_SADDSAT: |
5478 | case TargetOpcode::G_SSUBSAT: |
5479 | case TargetOpcode::G_SMIN: |
5480 | case TargetOpcode::G_SMAX: |
5481 | case TargetOpcode::G_UMIN: |
5482 | case TargetOpcode::G_UMAX: |
5483 | case TargetOpcode::G_FMINNUM: |
5484 | case TargetOpcode::G_FMAXNUM: |
5485 | case TargetOpcode::G_FMINNUM_IEEE: |
5486 | case TargetOpcode::G_FMAXNUM_IEEE: |
5487 | case TargetOpcode::G_FMINIMUM: |
5488 | case TargetOpcode::G_FMAXIMUM: |
5489 | case TargetOpcode::G_STRICT_FADD: |
5490 | case TargetOpcode::G_STRICT_FSUB: |
5491 | case TargetOpcode::G_STRICT_FMUL: |
5492 | case TargetOpcode::G_SHL: |
5493 | case TargetOpcode::G_ASHR: |
5494 | case TargetOpcode::G_LSHR: { |
5495 | Observer.changingInstr(MI); |
5496 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 1); |
5497 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 2); |
5498 | moreElementsVectorDst(MI, WideTy: MoreTy, OpIdx: 0); |
5499 | Observer.changedInstr(MI); |
5500 | return Legalized; |
5501 | } |
5502 | case TargetOpcode::G_FMA: |
5503 | case TargetOpcode::G_STRICT_FMA: |
5504 | case TargetOpcode::G_FSHR: |
5505 | case TargetOpcode::G_FSHL: { |
5506 | Observer.changingInstr(MI); |
5507 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 1); |
5508 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 2); |
5509 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 3); |
5510 | moreElementsVectorDst(MI, WideTy: MoreTy, OpIdx: 0); |
5511 | Observer.changedInstr(MI); |
5512 | return Legalized; |
5513 | } |
5514 | case TargetOpcode::G_EXTRACT_VECTOR_ELT: |
5515 | case TargetOpcode::G_EXTRACT: |
5516 | if (TypeIdx != 1) |
5517 | return UnableToLegalize; |
5518 | Observer.changingInstr(MI); |
5519 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 1); |
5520 | Observer.changedInstr(MI); |
5521 | return Legalized; |
5522 | case TargetOpcode::G_INSERT: |
5523 | case TargetOpcode::G_INSERT_VECTOR_ELT: |
5524 | case TargetOpcode::G_FREEZE: |
5525 | case TargetOpcode::G_FNEG: |
5526 | case TargetOpcode::G_FABS: |
5527 | case TargetOpcode::G_FSQRT: |
5528 | case TargetOpcode::G_FCEIL: |
5529 | case TargetOpcode::G_FFLOOR: |
5530 | case TargetOpcode::G_FNEARBYINT: |
5531 | case TargetOpcode::G_FRINT: |
5532 | case TargetOpcode::G_INTRINSIC_ROUND: |
5533 | case TargetOpcode::G_INTRINSIC_ROUNDEVEN: |
5534 | case TargetOpcode::G_INTRINSIC_TRUNC: |
5535 | case TargetOpcode::G_BSWAP: |
5536 | case TargetOpcode::G_FCANONICALIZE: |
5537 | case TargetOpcode::G_SEXT_INREG: |
5538 | case TargetOpcode::G_ABS: |
5539 | if (TypeIdx != 0) |
5540 | return UnableToLegalize; |
5541 | Observer.changingInstr(MI); |
5542 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 1); |
5543 | moreElementsVectorDst(MI, WideTy: MoreTy, OpIdx: 0); |
5544 | Observer.changedInstr(MI); |
5545 | return Legalized; |
5546 | case TargetOpcode::G_SELECT: { |
5547 | auto [DstReg, DstTy, CondReg, CondTy] = MI.getFirst2RegLLTs(); |
5548 | if (TypeIdx == 1) { |
5549 | if (!CondTy.isScalar() || |
5550 | DstTy.getElementCount() != MoreTy.getElementCount()) |
5551 | return UnableToLegalize; |
5552 | |
5553 | // This is turning a scalar select of vectors into a vector |
5554 | // select. Broadcast the select condition. |
5555 | auto ShufSplat = MIRBuilder.buildShuffleSplat(Res: MoreTy, Src: CondReg); |
5556 | Observer.changingInstr(MI); |
5557 | MI.getOperand(i: 1).setReg(ShufSplat.getReg(Idx: 0)); |
5558 | Observer.changedInstr(MI); |
5559 | return Legalized; |
5560 | } |
5561 | |
5562 | if (CondTy.isVector()) |
5563 | return UnableToLegalize; |
5564 | |
5565 | Observer.changingInstr(MI); |
5566 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 2); |
5567 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 3); |
5568 | moreElementsVectorDst(MI, WideTy: MoreTy, OpIdx: 0); |
5569 | Observer.changedInstr(MI); |
5570 | return Legalized; |
5571 | } |
5572 | case TargetOpcode::G_UNMERGE_VALUES: |
5573 | return UnableToLegalize; |
5574 | case TargetOpcode::G_PHI: |
5575 | return moreElementsVectorPhi(MI, TypeIdx, MoreTy); |
5576 | case TargetOpcode::G_SHUFFLE_VECTOR: |
5577 | return moreElementsVectorShuffle(MI, TypeIdx, MoreTy); |
5578 | case TargetOpcode::G_BUILD_VECTOR: { |
5579 | SmallVector<SrcOp, 8> Elts; |
5580 | for (auto Op : MI.uses()) { |
5581 | Elts.push_back(Elt: Op.getReg()); |
5582 | } |
5583 | |
5584 | for (unsigned i = Elts.size(); i < MoreTy.getNumElements(); ++i) { |
5585 | Elts.push_back(Elt: MIRBuilder.buildUndef(Res: MoreTy.getScalarType())); |
5586 | } |
5587 | |
5588 | MIRBuilder.buildDeleteTrailingVectorElements( |
5589 | Res: MI.getOperand(i: 0).getReg(), Op0: MIRBuilder.buildInstr(Opc, DstOps: {MoreTy}, SrcOps: Elts)); |
5590 | MI.eraseFromParent(); |
5591 | return Legalized; |
5592 | } |
5593 | case TargetOpcode::G_SEXT: |
5594 | case TargetOpcode::G_ZEXT: |
5595 | case TargetOpcode::G_ANYEXT: |
5596 | case TargetOpcode::G_TRUNC: |
5597 | case TargetOpcode::G_FPTRUNC: |
5598 | case TargetOpcode::G_FPEXT: |
5599 | case TargetOpcode::G_FPTOSI: |
5600 | case TargetOpcode::G_FPTOUI: |
5601 | case TargetOpcode::G_SITOFP: |
5602 | case TargetOpcode::G_UITOFP: { |
5603 | Observer.changingInstr(MI); |
5604 | LLT SrcExtTy; |
5605 | LLT DstExtTy; |
5606 | if (TypeIdx == 0) { |
5607 | DstExtTy = MoreTy; |
5608 | SrcExtTy = LLT::fixed_vector( |
5609 | NumElements: MoreTy.getNumElements(), |
5610 | ScalarTy: MRI.getType(Reg: MI.getOperand(i: 1).getReg()).getElementType()); |
5611 | } else { |
5612 | DstExtTy = LLT::fixed_vector( |
5613 | NumElements: MoreTy.getNumElements(), |
5614 | ScalarTy: MRI.getType(Reg: MI.getOperand(i: 0).getReg()).getElementType()); |
5615 | SrcExtTy = MoreTy; |
5616 | } |
5617 | moreElementsVectorSrc(MI, MoreTy: SrcExtTy, OpIdx: 1); |
5618 | moreElementsVectorDst(MI, WideTy: DstExtTy, OpIdx: 0); |
5619 | Observer.changedInstr(MI); |
5620 | return Legalized; |
5621 | } |
5622 | case TargetOpcode::G_ICMP: |
5623 | case TargetOpcode::G_FCMP: { |
5624 | if (TypeIdx != 1) |
5625 | return UnableToLegalize; |
5626 | |
5627 | Observer.changingInstr(MI); |
5628 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 2); |
5629 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 3); |
5630 | LLT CondTy = LLT::fixed_vector( |
5631 | NumElements: MoreTy.getNumElements(), |
5632 | ScalarTy: MRI.getType(Reg: MI.getOperand(i: 0).getReg()).getElementType()); |
5633 | moreElementsVectorDst(MI, WideTy: CondTy, OpIdx: 0); |
5634 | Observer.changedInstr(MI); |
5635 | return Legalized; |
5636 | } |
5637 | case TargetOpcode::G_BITCAST: { |
5638 | if (TypeIdx != 0) |
5639 | return UnableToLegalize; |
5640 | |
5641 | LLT SrcTy = MRI.getType(Reg: MI.getOperand(i: 1).getReg()); |
5642 | LLT DstTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
5643 | |
5644 | unsigned coefficient = SrcTy.getNumElements() * MoreTy.getNumElements(); |
5645 | if (coefficient % DstTy.getNumElements() != 0) |
5646 | return UnableToLegalize; |
5647 | |
5648 | coefficient = coefficient / DstTy.getNumElements(); |
5649 | |
5650 | LLT NewTy = SrcTy.changeElementCount( |
5651 | EC: ElementCount::get(MinVal: coefficient, Scalable: MoreTy.isScalable())); |
5652 | Observer.changingInstr(MI); |
5653 | moreElementsVectorSrc(MI, MoreTy: NewTy, OpIdx: 1); |
5654 | moreElementsVectorDst(MI, WideTy: MoreTy, OpIdx: 0); |
5655 | Observer.changedInstr(MI); |
5656 | return Legalized; |
5657 | } |
5658 | case TargetOpcode::G_VECREDUCE_FADD: |
5659 | case TargetOpcode::G_VECREDUCE_FMUL: |
5660 | case TargetOpcode::G_VECREDUCE_ADD: |
5661 | case TargetOpcode::G_VECREDUCE_MUL: |
5662 | case TargetOpcode::G_VECREDUCE_AND: |
5663 | case TargetOpcode::G_VECREDUCE_OR: |
5664 | case TargetOpcode::G_VECREDUCE_XOR: |
5665 | case TargetOpcode::G_VECREDUCE_SMAX: |
5666 | case TargetOpcode::G_VECREDUCE_SMIN: |
5667 | case TargetOpcode::G_VECREDUCE_UMAX: |
5668 | case TargetOpcode::G_VECREDUCE_UMIN: { |
5669 | LLT OrigTy = MRI.getType(Reg: MI.getOperand(i: 1).getReg()); |
5670 | MachineOperand &MO = MI.getOperand(i: 1); |
5671 | auto NewVec = MIRBuilder.buildPadVectorWithUndefElements(Res: MoreTy, Op0: MO); |
5672 | auto NeutralElement = getNeutralElementForVecReduce( |
5673 | Opcode: MI.getOpcode(), MIRBuilder, Ty: MoreTy.getElementType()); |
5674 | |
5675 | LLT IdxTy(TLI.getVectorIdxTy(DL: MIRBuilder.getDataLayout())); |
5676 | for (size_t i = OrigTy.getNumElements(), e = MoreTy.getNumElements(); |
5677 | i != e; i++) { |
5678 | auto Idx = MIRBuilder.buildConstant(Res: IdxTy, Val: i); |
5679 | NewVec = MIRBuilder.buildInsertVectorElement(Res: MoreTy, Val: NewVec, |
5680 | Elt: NeutralElement, Idx); |
5681 | } |
5682 | |
5683 | Observer.changingInstr(MI); |
5684 | MO.setReg(NewVec.getReg(Idx: 0)); |
5685 | Observer.changedInstr(MI); |
5686 | return Legalized; |
5687 | } |
5688 | |
5689 | default: |
5690 | return UnableToLegalize; |
5691 | } |
5692 | } |
5693 | |
5694 | LegalizerHelper::LegalizeResult |
5695 | LegalizerHelper::equalizeVectorShuffleLengths(MachineInstr &MI) { |
5696 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
5697 | ArrayRef<int> Mask = MI.getOperand(i: 3).getShuffleMask(); |
5698 | unsigned MaskNumElts = Mask.size(); |
5699 | unsigned SrcNumElts = SrcTy.getNumElements(); |
5700 | LLT DestEltTy = DstTy.getElementType(); |
5701 | |
5702 | if (MaskNumElts == SrcNumElts) |
5703 | return Legalized; |
5704 | |
5705 | if (MaskNumElts < SrcNumElts) { |
5706 | // Extend mask to match new destination vector size with |
5707 | // undef values. |
5708 | SmallVector<int, 16> NewMask(Mask); |
5709 | for (unsigned I = MaskNumElts; I < SrcNumElts; ++I) |
5710 | NewMask.push_back(Elt: -1); |
5711 | |
5712 | moreElementsVectorDst(MI, WideTy: SrcTy, OpIdx: 0); |
5713 | MIRBuilder.setInstrAndDebugLoc(MI); |
5714 | MIRBuilder.buildShuffleVector(Res: MI.getOperand(i: 0).getReg(), |
5715 | Src1: MI.getOperand(i: 1).getReg(), |
5716 | Src2: MI.getOperand(i: 2).getReg(), Mask: NewMask); |
5717 | MI.eraseFromParent(); |
5718 | |
5719 | return Legalized; |
5720 | } |
5721 | |
5722 | unsigned PaddedMaskNumElts = alignTo(Value: MaskNumElts, Align: SrcNumElts); |
5723 | unsigned NumConcat = PaddedMaskNumElts / SrcNumElts; |
5724 | LLT PaddedTy = LLT::fixed_vector(NumElements: PaddedMaskNumElts, ScalarTy: DestEltTy); |
5725 | |
5726 | // Create new source vectors by concatenating the initial |
5727 | // source vectors with undefined vectors of the same size. |
5728 | auto Undef = MIRBuilder.buildUndef(Res: SrcTy); |
5729 | SmallVector<Register, 8> MOps1(NumConcat, Undef.getReg(Idx: 0)); |
5730 | SmallVector<Register, 8> MOps2(NumConcat, Undef.getReg(Idx: 0)); |
5731 | MOps1[0] = MI.getOperand(i: 1).getReg(); |
5732 | MOps2[0] = MI.getOperand(i: 2).getReg(); |
5733 | |
5734 | auto Src1 = MIRBuilder.buildConcatVectors(Res: PaddedTy, Ops: MOps1); |
5735 | auto Src2 = MIRBuilder.buildConcatVectors(Res: PaddedTy, Ops: MOps2); |
5736 | |
5737 | // Readjust mask for new input vector length. |
5738 | SmallVector<int, 8> MappedOps(PaddedMaskNumElts, -1); |
5739 | for (unsigned I = 0; I != MaskNumElts; ++I) { |
5740 | int Idx = Mask[I]; |
5741 | if (Idx >= static_cast<int>(SrcNumElts)) |
5742 | Idx += PaddedMaskNumElts - SrcNumElts; |
5743 | MappedOps[I] = Idx; |
5744 | } |
5745 | |
5746 | // If we got more elements than required, extract subvector. |
5747 | if (MaskNumElts != PaddedMaskNumElts) { |
5748 | auto Shuffle = |
5749 | MIRBuilder.buildShuffleVector(Res: PaddedTy, Src1, Src2, Mask: MappedOps); |
5750 | |
5751 | SmallVector<Register, 16> Elts(MaskNumElts); |
5752 | for (unsigned I = 0; I < MaskNumElts; ++I) { |
5753 | Elts[I] = |
5754 | MIRBuilder.buildExtractVectorElementConstant(Res: DestEltTy, Val: Shuffle, Idx: I) |
5755 | .getReg(Idx: 0); |
5756 | } |
5757 | MIRBuilder.buildBuildVector(Res: DstReg, Ops: Elts); |
5758 | } else { |
5759 | MIRBuilder.buildShuffleVector(Res: DstReg, Src1, Src2, Mask: MappedOps); |
5760 | } |
5761 | |
5762 | MI.eraseFromParent(); |
5763 | return LegalizerHelper::LegalizeResult::Legalized; |
5764 | } |
5765 | |
5766 | LegalizerHelper::LegalizeResult |
5767 | LegalizerHelper::moreElementsVectorShuffle(MachineInstr &MI, |
5768 | unsigned int TypeIdx, LLT MoreTy) { |
5769 | auto [DstTy, Src1Ty, Src2Ty] = MI.getFirst3LLTs(); |
5770 | ArrayRef<int> Mask = MI.getOperand(i: 3).getShuffleMask(); |
5771 | unsigned NumElts = DstTy.getNumElements(); |
5772 | unsigned WidenNumElts = MoreTy.getNumElements(); |
5773 | |
5774 | if (DstTy.isVector() && Src1Ty.isVector() && |
5775 | DstTy.getNumElements() != Src1Ty.getNumElements()) { |
5776 | return equalizeVectorShuffleLengths(MI); |
5777 | } |
5778 | |
5779 | if (TypeIdx != 0) |
5780 | return UnableToLegalize; |
5781 | |
5782 | // Expect a canonicalized shuffle. |
5783 | if (DstTy != Src1Ty || DstTy != Src2Ty) |
5784 | return UnableToLegalize; |
5785 | |
5786 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 1); |
5787 | moreElementsVectorSrc(MI, MoreTy, OpIdx: 2); |
5788 | |
5789 | // Adjust mask based on new input vector length. |
5790 | SmallVector<int, 16> NewMask; |
5791 | for (unsigned I = 0; I != NumElts; ++I) { |
5792 | int Idx = Mask[I]; |
5793 | if (Idx < static_cast<int>(NumElts)) |
5794 | NewMask.push_back(Elt: Idx); |
5795 | else |
5796 | NewMask.push_back(Elt: Idx - NumElts + WidenNumElts); |
5797 | } |
5798 | for (unsigned I = NumElts; I != WidenNumElts; ++I) |
5799 | NewMask.push_back(Elt: -1); |
5800 | moreElementsVectorDst(MI, WideTy: MoreTy, OpIdx: 0); |
5801 | MIRBuilder.setInstrAndDebugLoc(MI); |
5802 | MIRBuilder.buildShuffleVector(Res: MI.getOperand(i: 0).getReg(), |
5803 | Src1: MI.getOperand(i: 1).getReg(), |
5804 | Src2: MI.getOperand(i: 2).getReg(), Mask: NewMask); |
5805 | MI.eraseFromParent(); |
5806 | return Legalized; |
5807 | } |
5808 | |
5809 | void LegalizerHelper::multiplyRegisters(SmallVectorImpl<Register> &DstRegs, |
5810 | ArrayRef<Register> Src1Regs, |
5811 | ArrayRef<Register> Src2Regs, |
5812 | LLT NarrowTy) { |
5813 | MachineIRBuilder &B = MIRBuilder; |
5814 | unsigned SrcParts = Src1Regs.size(); |
5815 | unsigned DstParts = DstRegs.size(); |
5816 | |
5817 | unsigned DstIdx = 0; // Low bits of the result. |
5818 | Register FactorSum = |
5819 | B.buildMul(Dst: NarrowTy, Src0: Src1Regs[DstIdx], Src1: Src2Regs[DstIdx]).getReg(Idx: 0); |
5820 | DstRegs[DstIdx] = FactorSum; |
5821 | |
5822 | unsigned CarrySumPrevDstIdx; |
5823 | SmallVector<Register, 4> Factors; |
5824 | |
5825 | for (DstIdx = 1; DstIdx < DstParts; DstIdx++) { |
5826 | // Collect low parts of muls for DstIdx. |
5827 | for (unsigned i = DstIdx + 1 < SrcParts ? 0 : DstIdx - SrcParts + 1; |
5828 | i <= std::min(a: DstIdx, b: SrcParts - 1); ++i) { |
5829 | MachineInstrBuilder Mul = |
5830 | B.buildMul(Dst: NarrowTy, Src0: Src1Regs[DstIdx - i], Src1: Src2Regs[i]); |
5831 | Factors.push_back(Elt: Mul.getReg(Idx: 0)); |
5832 | } |
5833 | // Collect high parts of muls from previous DstIdx. |
5834 | for (unsigned i = DstIdx < SrcParts ? 0 : DstIdx - SrcParts; |
5835 | i <= std::min(a: DstIdx - 1, b: SrcParts - 1); ++i) { |
5836 | MachineInstrBuilder Umulh = |
5837 | B.buildUMulH(Dst: NarrowTy, Src0: Src1Regs[DstIdx - 1 - i], Src1: Src2Regs[i]); |
5838 | Factors.push_back(Elt: Umulh.getReg(Idx: 0)); |
5839 | } |
5840 | // Add CarrySum from additions calculated for previous DstIdx. |
5841 | if (DstIdx != 1) { |
5842 | Factors.push_back(Elt: CarrySumPrevDstIdx); |
5843 | } |
5844 | |
5845 | Register CarrySum; |
5846 | // Add all factors and accumulate all carries into CarrySum. |
5847 | if (DstIdx != DstParts - 1) { |
5848 | MachineInstrBuilder Uaddo = |
5849 | B.buildUAddo(Res: NarrowTy, CarryOut: LLT::scalar(SizeInBits: 1), Op0: Factors[0], Op1: Factors[1]); |
5850 | FactorSum = Uaddo.getReg(Idx: 0); |
5851 | CarrySum = B.buildZExt(Res: NarrowTy, Op: Uaddo.getReg(Idx: 1)).getReg(Idx: 0); |
5852 | for (unsigned i = 2; i < Factors.size(); ++i) { |
5853 | MachineInstrBuilder Uaddo = |
5854 | B.buildUAddo(Res: NarrowTy, CarryOut: LLT::scalar(SizeInBits: 1), Op0: FactorSum, Op1: Factors[i]); |
5855 | FactorSum = Uaddo.getReg(Idx: 0); |
5856 | MachineInstrBuilder Carry = B.buildZExt(Res: NarrowTy, Op: Uaddo.getReg(Idx: 1)); |
5857 | CarrySum = B.buildAdd(Dst: NarrowTy, Src0: CarrySum, Src1: Carry).getReg(Idx: 0); |
5858 | } |
5859 | } else { |
5860 | // Since value for the next index is not calculated, neither is CarrySum. |
5861 | FactorSum = B.buildAdd(Dst: NarrowTy, Src0: Factors[0], Src1: Factors[1]).getReg(Idx: 0); |
5862 | for (unsigned i = 2; i < Factors.size(); ++i) |
5863 | FactorSum = B.buildAdd(Dst: NarrowTy, Src0: FactorSum, Src1: Factors[i]).getReg(Idx: 0); |
5864 | } |
5865 | |
5866 | CarrySumPrevDstIdx = CarrySum; |
5867 | DstRegs[DstIdx] = FactorSum; |
5868 | Factors.clear(); |
5869 | } |
5870 | } |
5871 | |
5872 | LegalizerHelper::LegalizeResult |
5873 | LegalizerHelper::narrowScalarAddSub(MachineInstr &MI, unsigned TypeIdx, |
5874 | LLT NarrowTy) { |
5875 | if (TypeIdx != 0) |
5876 | return UnableToLegalize; |
5877 | |
5878 | Register DstReg = MI.getOperand(i: 0).getReg(); |
5879 | LLT DstType = MRI.getType(Reg: DstReg); |
5880 | // FIXME: add support for vector types |
5881 | if (DstType.isVector()) |
5882 | return UnableToLegalize; |
5883 | |
5884 | unsigned Opcode = MI.getOpcode(); |
5885 | unsigned OpO, OpE, OpF; |
5886 | switch (Opcode) { |
5887 | case TargetOpcode::G_SADDO: |
5888 | case TargetOpcode::G_SADDE: |
5889 | case TargetOpcode::G_UADDO: |
5890 | case TargetOpcode::G_UADDE: |
5891 | case TargetOpcode::G_ADD: |
5892 | OpO = TargetOpcode::G_UADDO; |
5893 | OpE = TargetOpcode::G_UADDE; |
5894 | OpF = TargetOpcode::G_UADDE; |
5895 | if (Opcode == TargetOpcode::G_SADDO || Opcode == TargetOpcode::G_SADDE) |
5896 | OpF = TargetOpcode::G_SADDE; |
5897 | break; |
5898 | case TargetOpcode::G_SSUBO: |
5899 | case TargetOpcode::G_SSUBE: |
5900 | case TargetOpcode::G_USUBO: |
5901 | case TargetOpcode::G_USUBE: |
5902 | case TargetOpcode::G_SUB: |
5903 | OpO = TargetOpcode::G_USUBO; |
5904 | OpE = TargetOpcode::G_USUBE; |
5905 | OpF = TargetOpcode::G_USUBE; |
5906 | if (Opcode == TargetOpcode::G_SSUBO || Opcode == TargetOpcode::G_SSUBE) |
5907 | OpF = TargetOpcode::G_SSUBE; |
5908 | break; |
5909 | default: |
5910 | llvm_unreachable("Unexpected add/sub opcode!" ); |
5911 | } |
5912 | |
5913 | // 1 for a plain add/sub, 2 if this is an operation with a carry-out. |
5914 | unsigned NumDefs = MI.getNumExplicitDefs(); |
5915 | Register Src1 = MI.getOperand(i: NumDefs).getReg(); |
5916 | Register Src2 = MI.getOperand(i: NumDefs + 1).getReg(); |
5917 | Register CarryDst, CarryIn; |
5918 | if (NumDefs == 2) |
5919 | CarryDst = MI.getOperand(i: 1).getReg(); |
5920 | if (MI.getNumOperands() == NumDefs + 3) |
5921 | CarryIn = MI.getOperand(i: NumDefs + 2).getReg(); |
5922 | |
5923 | LLT RegTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
5924 | LLT LeftoverTy, DummyTy; |
5925 | SmallVector<Register, 2> Src1Regs, Src2Regs, Src1Left, Src2Left, DstRegs; |
5926 | extractParts(Reg: Src1, RegTy, MainTy: NarrowTy, LeftoverTy, VRegs&: Src1Regs, LeftoverVRegs&: Src1Left, |
5927 | MIRBuilder, MRI); |
5928 | extractParts(Reg: Src2, RegTy, MainTy: NarrowTy, LeftoverTy&: DummyTy, VRegs&: Src2Regs, LeftoverVRegs&: Src2Left, MIRBuilder, |
5929 | MRI); |
5930 | |
5931 | int NarrowParts = Src1Regs.size(); |
5932 | for (int I = 0, E = Src1Left.size(); I != E; ++I) { |
5933 | Src1Regs.push_back(Elt: Src1Left[I]); |
5934 | Src2Regs.push_back(Elt: Src2Left[I]); |
5935 | } |
5936 | DstRegs.reserve(N: Src1Regs.size()); |
5937 | |
5938 | for (int i = 0, e = Src1Regs.size(); i != e; ++i) { |
5939 | Register DstReg = |
5940 | MRI.createGenericVirtualRegister(Ty: MRI.getType(Reg: Src1Regs[i])); |
5941 | Register CarryOut = MRI.createGenericVirtualRegister(Ty: LLT::scalar(SizeInBits: 1)); |
5942 | // Forward the final carry-out to the destination register |
5943 | if (i == e - 1 && CarryDst) |
5944 | CarryOut = CarryDst; |
5945 | |
5946 | if (!CarryIn) { |
5947 | MIRBuilder.buildInstr(Opc: OpO, DstOps: {DstReg, CarryOut}, |
5948 | SrcOps: {Src1Regs[i], Src2Regs[i]}); |
5949 | } else if (i == e - 1) { |
5950 | MIRBuilder.buildInstr(Opc: OpF, DstOps: {DstReg, CarryOut}, |
5951 | SrcOps: {Src1Regs[i], Src2Regs[i], CarryIn}); |
5952 | } else { |
5953 | MIRBuilder.buildInstr(Opc: OpE, DstOps: {DstReg, CarryOut}, |
5954 | SrcOps: {Src1Regs[i], Src2Regs[i], CarryIn}); |
5955 | } |
5956 | |
5957 | DstRegs.push_back(Elt: DstReg); |
5958 | CarryIn = CarryOut; |
5959 | } |
5960 | insertParts(DstReg: MI.getOperand(i: 0).getReg(), ResultTy: RegTy, PartTy: NarrowTy, |
5961 | PartRegs: ArrayRef(DstRegs).take_front(N: NarrowParts), LeftoverTy, |
5962 | LeftoverRegs: ArrayRef(DstRegs).drop_front(N: NarrowParts)); |
5963 | |
5964 | MI.eraseFromParent(); |
5965 | return Legalized; |
5966 | } |
5967 | |
5968 | LegalizerHelper::LegalizeResult |
5969 | LegalizerHelper::narrowScalarMul(MachineInstr &MI, LLT NarrowTy) { |
5970 | auto [DstReg, Src1, Src2] = MI.getFirst3Regs(); |
5971 | |
5972 | LLT Ty = MRI.getType(Reg: DstReg); |
5973 | if (Ty.isVector()) |
5974 | return UnableToLegalize; |
5975 | |
5976 | unsigned Size = Ty.getSizeInBits(); |
5977 | unsigned NarrowSize = NarrowTy.getSizeInBits(); |
5978 | if (Size % NarrowSize != 0) |
5979 | return UnableToLegalize; |
5980 | |
5981 | unsigned NumParts = Size / NarrowSize; |
5982 | bool IsMulHigh = MI.getOpcode() == TargetOpcode::G_UMULH; |
5983 | unsigned DstTmpParts = NumParts * (IsMulHigh ? 2 : 1); |
5984 | |
5985 | SmallVector<Register, 2> Src1Parts, Src2Parts; |
5986 | SmallVector<Register, 2> DstTmpRegs(DstTmpParts); |
5987 | extractParts(Reg: Src1, Ty: NarrowTy, NumParts, VRegs&: Src1Parts, MIRBuilder, MRI); |
5988 | extractParts(Reg: Src2, Ty: NarrowTy, NumParts, VRegs&: Src2Parts, MIRBuilder, MRI); |
5989 | multiplyRegisters(DstRegs&: DstTmpRegs, Src1Regs: Src1Parts, Src2Regs: Src2Parts, NarrowTy); |
5990 | |
5991 | // Take only high half of registers if this is high mul. |
5992 | ArrayRef<Register> DstRegs(&DstTmpRegs[DstTmpParts - NumParts], NumParts); |
5993 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: DstRegs); |
5994 | MI.eraseFromParent(); |
5995 | return Legalized; |
5996 | } |
5997 | |
5998 | LegalizerHelper::LegalizeResult |
5999 | LegalizerHelper::narrowScalarFPTOI(MachineInstr &MI, unsigned TypeIdx, |
6000 | LLT NarrowTy) { |
6001 | if (TypeIdx != 0) |
6002 | return UnableToLegalize; |
6003 | |
6004 | bool IsSigned = MI.getOpcode() == TargetOpcode::G_FPTOSI; |
6005 | |
6006 | Register Src = MI.getOperand(i: 1).getReg(); |
6007 | LLT SrcTy = MRI.getType(Reg: Src); |
6008 | |
6009 | // If all finite floats fit into the narrowed integer type, we can just swap |
6010 | // out the result type. This is practically only useful for conversions from |
6011 | // half to at least 16-bits, so just handle the one case. |
6012 | if (SrcTy.getScalarType() != LLT::scalar(SizeInBits: 16) || |
6013 | NarrowTy.getScalarSizeInBits() < (IsSigned ? 17u : 16u)) |
6014 | return UnableToLegalize; |
6015 | |
6016 | Observer.changingInstr(MI); |
6017 | narrowScalarDst(MI, NarrowTy, OpIdx: 0, |
6018 | ExtOpcode: IsSigned ? TargetOpcode::G_SEXT : TargetOpcode::G_ZEXT); |
6019 | Observer.changedInstr(MI); |
6020 | return Legalized; |
6021 | } |
6022 | |
6023 | LegalizerHelper::LegalizeResult |
6024 | LegalizerHelper::(MachineInstr &MI, unsigned TypeIdx, |
6025 | LLT NarrowTy) { |
6026 | if (TypeIdx != 1) |
6027 | return UnableToLegalize; |
6028 | |
6029 | uint64_t NarrowSize = NarrowTy.getSizeInBits(); |
6030 | |
6031 | int64_t SizeOp1 = MRI.getType(Reg: MI.getOperand(i: 1).getReg()).getSizeInBits(); |
6032 | // FIXME: add support for when SizeOp1 isn't an exact multiple of |
6033 | // NarrowSize. |
6034 | if (SizeOp1 % NarrowSize != 0) |
6035 | return UnableToLegalize; |
6036 | int NumParts = SizeOp1 / NarrowSize; |
6037 | |
6038 | SmallVector<Register, 2> SrcRegs, DstRegs; |
6039 | SmallVector<uint64_t, 2> Indexes; |
6040 | extractParts(Reg: MI.getOperand(i: 1).getReg(), Ty: NarrowTy, NumParts, VRegs&: SrcRegs, |
6041 | MIRBuilder, MRI); |
6042 | |
6043 | Register OpReg = MI.getOperand(i: 0).getReg(); |
6044 | uint64_t OpStart = MI.getOperand(i: 2).getImm(); |
6045 | uint64_t OpSize = MRI.getType(Reg: OpReg).getSizeInBits(); |
6046 | for (int i = 0; i < NumParts; ++i) { |
6047 | unsigned SrcStart = i * NarrowSize; |
6048 | |
6049 | if (SrcStart + NarrowSize <= OpStart || SrcStart >= OpStart + OpSize) { |
6050 | // No part of the extract uses this subregister, ignore it. |
6051 | continue; |
6052 | } else if (SrcStart == OpStart && NarrowTy == MRI.getType(Reg: OpReg)) { |
6053 | // The entire subregister is extracted, forward the value. |
6054 | DstRegs.push_back(Elt: SrcRegs[i]); |
6055 | continue; |
6056 | } |
6057 | |
6058 | // OpSegStart is where this destination segment would start in OpReg if it |
6059 | // extended infinitely in both directions. |
6060 | int64_t ; |
6061 | uint64_t SegSize; |
6062 | if (OpStart < SrcStart) { |
6063 | ExtractOffset = 0; |
6064 | SegSize = std::min(a: NarrowSize, b: OpStart + OpSize - SrcStart); |
6065 | } else { |
6066 | ExtractOffset = OpStart - SrcStart; |
6067 | SegSize = std::min(a: SrcStart + NarrowSize - OpStart, b: OpSize); |
6068 | } |
6069 | |
6070 | Register SegReg = SrcRegs[i]; |
6071 | if (ExtractOffset != 0 || SegSize != NarrowSize) { |
6072 | // A genuine extract is needed. |
6073 | SegReg = MRI.createGenericVirtualRegister(Ty: LLT::scalar(SizeInBits: SegSize)); |
6074 | MIRBuilder.buildExtract(Res: SegReg, Src: SrcRegs[i], Index: ExtractOffset); |
6075 | } |
6076 | |
6077 | DstRegs.push_back(Elt: SegReg); |
6078 | } |
6079 | |
6080 | Register DstReg = MI.getOperand(i: 0).getReg(); |
6081 | if (MRI.getType(Reg: DstReg).isVector()) |
6082 | MIRBuilder.buildBuildVector(Res: DstReg, Ops: DstRegs); |
6083 | else if (DstRegs.size() > 1) |
6084 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: DstRegs); |
6085 | else |
6086 | MIRBuilder.buildCopy(Res: DstReg, Op: DstRegs[0]); |
6087 | MI.eraseFromParent(); |
6088 | return Legalized; |
6089 | } |
6090 | |
6091 | LegalizerHelper::LegalizeResult |
6092 | LegalizerHelper::narrowScalarInsert(MachineInstr &MI, unsigned TypeIdx, |
6093 | LLT NarrowTy) { |
6094 | // FIXME: Don't know how to handle secondary types yet. |
6095 | if (TypeIdx != 0) |
6096 | return UnableToLegalize; |
6097 | |
6098 | SmallVector<Register, 2> SrcRegs, LeftoverRegs, DstRegs; |
6099 | SmallVector<uint64_t, 2> Indexes; |
6100 | LLT RegTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
6101 | LLT LeftoverTy; |
6102 | extractParts(Reg: MI.getOperand(i: 1).getReg(), RegTy, MainTy: NarrowTy, LeftoverTy, VRegs&: SrcRegs, |
6103 | LeftoverVRegs&: LeftoverRegs, MIRBuilder, MRI); |
6104 | |
6105 | for (Register Reg : LeftoverRegs) |
6106 | SrcRegs.push_back(Elt: Reg); |
6107 | |
6108 | uint64_t NarrowSize = NarrowTy.getSizeInBits(); |
6109 | Register OpReg = MI.getOperand(i: 2).getReg(); |
6110 | uint64_t OpStart = MI.getOperand(i: 3).getImm(); |
6111 | uint64_t OpSize = MRI.getType(Reg: OpReg).getSizeInBits(); |
6112 | for (int I = 0, E = SrcRegs.size(); I != E; ++I) { |
6113 | unsigned DstStart = I * NarrowSize; |
6114 | |
6115 | if (DstStart == OpStart && NarrowTy == MRI.getType(Reg: OpReg)) { |
6116 | // The entire subregister is defined by this insert, forward the new |
6117 | // value. |
6118 | DstRegs.push_back(Elt: OpReg); |
6119 | continue; |
6120 | } |
6121 | |
6122 | Register SrcReg = SrcRegs[I]; |
6123 | if (MRI.getType(Reg: SrcRegs[I]) == LeftoverTy) { |
6124 | // The leftover reg is smaller than NarrowTy, so we need to extend it. |
6125 | SrcReg = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
6126 | MIRBuilder.buildAnyExt(Res: SrcReg, Op: SrcRegs[I]); |
6127 | } |
6128 | |
6129 | if (DstStart + NarrowSize <= OpStart || DstStart >= OpStart + OpSize) { |
6130 | // No part of the insert affects this subregister, forward the original. |
6131 | DstRegs.push_back(Elt: SrcReg); |
6132 | continue; |
6133 | } |
6134 | |
6135 | // OpSegStart is where this destination segment would start in OpReg if it |
6136 | // extended infinitely in both directions. |
6137 | int64_t , InsertOffset; |
6138 | uint64_t SegSize; |
6139 | if (OpStart < DstStart) { |
6140 | InsertOffset = 0; |
6141 | ExtractOffset = DstStart - OpStart; |
6142 | SegSize = std::min(a: NarrowSize, b: OpStart + OpSize - DstStart); |
6143 | } else { |
6144 | InsertOffset = OpStart - DstStart; |
6145 | ExtractOffset = 0; |
6146 | SegSize = |
6147 | std::min(a: NarrowSize - InsertOffset, b: OpStart + OpSize - DstStart); |
6148 | } |
6149 | |
6150 | Register SegReg = OpReg; |
6151 | if (ExtractOffset != 0 || SegSize != OpSize) { |
6152 | // A genuine extract is needed. |
6153 | SegReg = MRI.createGenericVirtualRegister(Ty: LLT::scalar(SizeInBits: SegSize)); |
6154 | MIRBuilder.buildExtract(Res: SegReg, Src: OpReg, Index: ExtractOffset); |
6155 | } |
6156 | |
6157 | Register DstReg = MRI.createGenericVirtualRegister(Ty: NarrowTy); |
6158 | MIRBuilder.buildInsert(Res: DstReg, Src: SrcReg, Op: SegReg, Index: InsertOffset); |
6159 | DstRegs.push_back(Elt: DstReg); |
6160 | } |
6161 | |
6162 | uint64_t WideSize = DstRegs.size() * NarrowSize; |
6163 | Register DstReg = MI.getOperand(i: 0).getReg(); |
6164 | if (WideSize > RegTy.getSizeInBits()) { |
6165 | Register MergeReg = MRI.createGenericVirtualRegister(Ty: LLT::scalar(SizeInBits: WideSize)); |
6166 | MIRBuilder.buildMergeLikeInstr(Res: MergeReg, Ops: DstRegs); |
6167 | MIRBuilder.buildTrunc(Res: DstReg, Op: MergeReg); |
6168 | } else |
6169 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: DstRegs); |
6170 | |
6171 | MI.eraseFromParent(); |
6172 | return Legalized; |
6173 | } |
6174 | |
6175 | LegalizerHelper::LegalizeResult |
6176 | LegalizerHelper::narrowScalarBasic(MachineInstr &MI, unsigned TypeIdx, |
6177 | LLT NarrowTy) { |
6178 | Register DstReg = MI.getOperand(i: 0).getReg(); |
6179 | LLT DstTy = MRI.getType(Reg: DstReg); |
6180 | |
6181 | assert(MI.getNumOperands() == 3 && TypeIdx == 0); |
6182 | |
6183 | SmallVector<Register, 4> DstRegs, DstLeftoverRegs; |
6184 | SmallVector<Register, 4> Src0Regs, Src0LeftoverRegs; |
6185 | SmallVector<Register, 4> Src1Regs, Src1LeftoverRegs; |
6186 | LLT LeftoverTy; |
6187 | if (!extractParts(Reg: MI.getOperand(i: 1).getReg(), RegTy: DstTy, MainTy: NarrowTy, LeftoverTy, |
6188 | VRegs&: Src0Regs, LeftoverVRegs&: Src0LeftoverRegs, MIRBuilder, MRI)) |
6189 | return UnableToLegalize; |
6190 | |
6191 | LLT Unused; |
6192 | if (!extractParts(Reg: MI.getOperand(i: 2).getReg(), RegTy: DstTy, MainTy: NarrowTy, LeftoverTy&: Unused, |
6193 | VRegs&: Src1Regs, LeftoverVRegs&: Src1LeftoverRegs, MIRBuilder, MRI)) |
6194 | llvm_unreachable("inconsistent extractParts result" ); |
6195 | |
6196 | for (unsigned I = 0, E = Src1Regs.size(); I != E; ++I) { |
6197 | auto Inst = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {NarrowTy}, |
6198 | SrcOps: {Src0Regs[I], Src1Regs[I]}); |
6199 | DstRegs.push_back(Elt: Inst.getReg(Idx: 0)); |
6200 | } |
6201 | |
6202 | for (unsigned I = 0, E = Src1LeftoverRegs.size(); I != E; ++I) { |
6203 | auto Inst = MIRBuilder.buildInstr( |
6204 | Opc: MI.getOpcode(), |
6205 | DstOps: {LeftoverTy}, SrcOps: {Src0LeftoverRegs[I], Src1LeftoverRegs[I]}); |
6206 | DstLeftoverRegs.push_back(Elt: Inst.getReg(Idx: 0)); |
6207 | } |
6208 | |
6209 | insertParts(DstReg, ResultTy: DstTy, PartTy: NarrowTy, PartRegs: DstRegs, |
6210 | LeftoverTy, LeftoverRegs: DstLeftoverRegs); |
6211 | |
6212 | MI.eraseFromParent(); |
6213 | return Legalized; |
6214 | } |
6215 | |
6216 | LegalizerHelper::LegalizeResult |
6217 | LegalizerHelper::narrowScalarExt(MachineInstr &MI, unsigned TypeIdx, |
6218 | LLT NarrowTy) { |
6219 | if (TypeIdx != 0) |
6220 | return UnableToLegalize; |
6221 | |
6222 | auto [DstReg, SrcReg] = MI.getFirst2Regs(); |
6223 | |
6224 | LLT DstTy = MRI.getType(Reg: DstReg); |
6225 | if (DstTy.isVector()) |
6226 | return UnableToLegalize; |
6227 | |
6228 | SmallVector<Register, 8> Parts; |
6229 | LLT GCDTy = extractGCDType(Parts, DstTy, NarrowTy, SrcReg); |
6230 | LLT LCMTy = buildLCMMergePieces(DstTy, NarrowTy, GCDTy, VRegs&: Parts, PadStrategy: MI.getOpcode()); |
6231 | buildWidenedRemergeToDst(DstReg, LCMTy, RemergeRegs: Parts); |
6232 | |
6233 | MI.eraseFromParent(); |
6234 | return Legalized; |
6235 | } |
6236 | |
6237 | LegalizerHelper::LegalizeResult |
6238 | LegalizerHelper::narrowScalarSelect(MachineInstr &MI, unsigned TypeIdx, |
6239 | LLT NarrowTy) { |
6240 | if (TypeIdx != 0) |
6241 | return UnableToLegalize; |
6242 | |
6243 | Register CondReg = MI.getOperand(i: 1).getReg(); |
6244 | LLT CondTy = MRI.getType(Reg: CondReg); |
6245 | if (CondTy.isVector()) // TODO: Handle vselect |
6246 | return UnableToLegalize; |
6247 | |
6248 | Register DstReg = MI.getOperand(i: 0).getReg(); |
6249 | LLT DstTy = MRI.getType(Reg: DstReg); |
6250 | |
6251 | SmallVector<Register, 4> DstRegs, DstLeftoverRegs; |
6252 | SmallVector<Register, 4> Src1Regs, Src1LeftoverRegs; |
6253 | SmallVector<Register, 4> Src2Regs, Src2LeftoverRegs; |
6254 | LLT LeftoverTy; |
6255 | if (!extractParts(Reg: MI.getOperand(i: 2).getReg(), RegTy: DstTy, MainTy: NarrowTy, LeftoverTy, |
6256 | VRegs&: Src1Regs, LeftoverVRegs&: Src1LeftoverRegs, MIRBuilder, MRI)) |
6257 | return UnableToLegalize; |
6258 | |
6259 | LLT Unused; |
6260 | if (!extractParts(Reg: MI.getOperand(i: 3).getReg(), RegTy: DstTy, MainTy: NarrowTy, LeftoverTy&: Unused, |
6261 | VRegs&: Src2Regs, LeftoverVRegs&: Src2LeftoverRegs, MIRBuilder, MRI)) |
6262 | llvm_unreachable("inconsistent extractParts result" ); |
6263 | |
6264 | for (unsigned I = 0, E = Src1Regs.size(); I != E; ++I) { |
6265 | auto Select = MIRBuilder.buildSelect(Res: NarrowTy, |
6266 | Tst: CondReg, Op0: Src1Regs[I], Op1: Src2Regs[I]); |
6267 | DstRegs.push_back(Elt: Select.getReg(Idx: 0)); |
6268 | } |
6269 | |
6270 | for (unsigned I = 0, E = Src1LeftoverRegs.size(); I != E; ++I) { |
6271 | auto Select = MIRBuilder.buildSelect( |
6272 | Res: LeftoverTy, Tst: CondReg, Op0: Src1LeftoverRegs[I], Op1: Src2LeftoverRegs[I]); |
6273 | DstLeftoverRegs.push_back(Elt: Select.getReg(Idx: 0)); |
6274 | } |
6275 | |
6276 | insertParts(DstReg, ResultTy: DstTy, PartTy: NarrowTy, PartRegs: DstRegs, |
6277 | LeftoverTy, LeftoverRegs: DstLeftoverRegs); |
6278 | |
6279 | MI.eraseFromParent(); |
6280 | return Legalized; |
6281 | } |
6282 | |
6283 | LegalizerHelper::LegalizeResult |
6284 | LegalizerHelper::narrowScalarCTLZ(MachineInstr &MI, unsigned TypeIdx, |
6285 | LLT NarrowTy) { |
6286 | if (TypeIdx != 1) |
6287 | return UnableToLegalize; |
6288 | |
6289 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
6290 | unsigned NarrowSize = NarrowTy.getSizeInBits(); |
6291 | |
6292 | if (SrcTy.isScalar() && SrcTy.getSizeInBits() == 2 * NarrowSize) { |
6293 | const bool IsUndef = MI.getOpcode() == TargetOpcode::G_CTLZ_ZERO_UNDEF; |
6294 | |
6295 | MachineIRBuilder &B = MIRBuilder; |
6296 | auto UnmergeSrc = B.buildUnmerge(Res: NarrowTy, Op: SrcReg); |
6297 | // ctlz(Hi:Lo) -> Hi == 0 ? (NarrowSize + ctlz(Lo)) : ctlz(Hi) |
6298 | auto C_0 = B.buildConstant(Res: NarrowTy, Val: 0); |
6299 | auto HiIsZero = B.buildICmp(Pred: CmpInst::ICMP_EQ, Res: LLT::scalar(SizeInBits: 1), |
6300 | Op0: UnmergeSrc.getReg(Idx: 1), Op1: C_0); |
6301 | auto LoCTLZ = IsUndef ? |
6302 | B.buildCTLZ_ZERO_UNDEF(Dst: DstTy, Src0: UnmergeSrc.getReg(Idx: 0)) : |
6303 | B.buildCTLZ(Dst: DstTy, Src0: UnmergeSrc.getReg(Idx: 0)); |
6304 | auto C_NarrowSize = B.buildConstant(Res: DstTy, Val: NarrowSize); |
6305 | auto HiIsZeroCTLZ = B.buildAdd(Dst: DstTy, Src0: LoCTLZ, Src1: C_NarrowSize); |
6306 | auto HiCTLZ = B.buildCTLZ_ZERO_UNDEF(Dst: DstTy, Src0: UnmergeSrc.getReg(Idx: 1)); |
6307 | B.buildSelect(Res: DstReg, Tst: HiIsZero, Op0: HiIsZeroCTLZ, Op1: HiCTLZ); |
6308 | |
6309 | MI.eraseFromParent(); |
6310 | return Legalized; |
6311 | } |
6312 | |
6313 | return UnableToLegalize; |
6314 | } |
6315 | |
6316 | LegalizerHelper::LegalizeResult |
6317 | LegalizerHelper::narrowScalarCTTZ(MachineInstr &MI, unsigned TypeIdx, |
6318 | LLT NarrowTy) { |
6319 | if (TypeIdx != 1) |
6320 | return UnableToLegalize; |
6321 | |
6322 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
6323 | unsigned NarrowSize = NarrowTy.getSizeInBits(); |
6324 | |
6325 | if (SrcTy.isScalar() && SrcTy.getSizeInBits() == 2 * NarrowSize) { |
6326 | const bool IsUndef = MI.getOpcode() == TargetOpcode::G_CTTZ_ZERO_UNDEF; |
6327 | |
6328 | MachineIRBuilder &B = MIRBuilder; |
6329 | auto UnmergeSrc = B.buildUnmerge(Res: NarrowTy, Op: SrcReg); |
6330 | // cttz(Hi:Lo) -> Lo == 0 ? (cttz(Hi) + NarrowSize) : cttz(Lo) |
6331 | auto C_0 = B.buildConstant(Res: NarrowTy, Val: 0); |
6332 | auto LoIsZero = B.buildICmp(Pred: CmpInst::ICMP_EQ, Res: LLT::scalar(SizeInBits: 1), |
6333 | Op0: UnmergeSrc.getReg(Idx: 0), Op1: C_0); |
6334 | auto HiCTTZ = IsUndef ? |
6335 | B.buildCTTZ_ZERO_UNDEF(Dst: DstTy, Src0: UnmergeSrc.getReg(Idx: 1)) : |
6336 | B.buildCTTZ(Dst: DstTy, Src0: UnmergeSrc.getReg(Idx: 1)); |
6337 | auto C_NarrowSize = B.buildConstant(Res: DstTy, Val: NarrowSize); |
6338 | auto LoIsZeroCTTZ = B.buildAdd(Dst: DstTy, Src0: HiCTTZ, Src1: C_NarrowSize); |
6339 | auto LoCTTZ = B.buildCTTZ_ZERO_UNDEF(Dst: DstTy, Src0: UnmergeSrc.getReg(Idx: 0)); |
6340 | B.buildSelect(Res: DstReg, Tst: LoIsZero, Op0: LoIsZeroCTTZ, Op1: LoCTTZ); |
6341 | |
6342 | MI.eraseFromParent(); |
6343 | return Legalized; |
6344 | } |
6345 | |
6346 | return UnableToLegalize; |
6347 | } |
6348 | |
6349 | LegalizerHelper::LegalizeResult |
6350 | LegalizerHelper::narrowScalarCTPOP(MachineInstr &MI, unsigned TypeIdx, |
6351 | LLT NarrowTy) { |
6352 | if (TypeIdx != 1) |
6353 | return UnableToLegalize; |
6354 | |
6355 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
6356 | unsigned NarrowSize = NarrowTy.getSizeInBits(); |
6357 | |
6358 | if (SrcTy.isScalar() && SrcTy.getSizeInBits() == 2 * NarrowSize) { |
6359 | auto UnmergeSrc = MIRBuilder.buildUnmerge(Res: NarrowTy, Op: MI.getOperand(i: 1)); |
6360 | |
6361 | auto LoCTPOP = MIRBuilder.buildCTPOP(Dst: DstTy, Src0: UnmergeSrc.getReg(Idx: 0)); |
6362 | auto HiCTPOP = MIRBuilder.buildCTPOP(Dst: DstTy, Src0: UnmergeSrc.getReg(Idx: 1)); |
6363 | MIRBuilder.buildAdd(Dst: DstReg, Src0: HiCTPOP, Src1: LoCTPOP); |
6364 | |
6365 | MI.eraseFromParent(); |
6366 | return Legalized; |
6367 | } |
6368 | |
6369 | return UnableToLegalize; |
6370 | } |
6371 | |
6372 | LegalizerHelper::LegalizeResult |
6373 | LegalizerHelper::narrowScalarFLDEXP(MachineInstr &MI, unsigned TypeIdx, |
6374 | LLT NarrowTy) { |
6375 | if (TypeIdx != 1) |
6376 | return UnableToLegalize; |
6377 | |
6378 | MachineIRBuilder &B = MIRBuilder; |
6379 | Register ExpReg = MI.getOperand(i: 2).getReg(); |
6380 | LLT ExpTy = MRI.getType(Reg: ExpReg); |
6381 | |
6382 | unsigned ClampSize = NarrowTy.getScalarSizeInBits(); |
6383 | |
6384 | // Clamp the exponent to the range of the target type. |
6385 | auto MinExp = B.buildConstant(Res: ExpTy, Val: minIntN(N: ClampSize)); |
6386 | auto ClampMin = B.buildSMax(Dst: ExpTy, Src0: ExpReg, Src1: MinExp); |
6387 | auto MaxExp = B.buildConstant(Res: ExpTy, Val: maxIntN(N: ClampSize)); |
6388 | auto Clamp = B.buildSMin(Dst: ExpTy, Src0: ClampMin, Src1: MaxExp); |
6389 | |
6390 | auto Trunc = B.buildTrunc(Res: NarrowTy, Op: Clamp); |
6391 | Observer.changingInstr(MI); |
6392 | MI.getOperand(i: 2).setReg(Trunc.getReg(Idx: 0)); |
6393 | Observer.changedInstr(MI); |
6394 | return Legalized; |
6395 | } |
6396 | |
6397 | LegalizerHelper::LegalizeResult |
6398 | LegalizerHelper::lowerBitCount(MachineInstr &MI) { |
6399 | unsigned Opc = MI.getOpcode(); |
6400 | const auto &TII = MIRBuilder.getTII(); |
6401 | auto isSupported = [this](const LegalityQuery &Q) { |
6402 | auto QAction = LI.getAction(Query: Q).Action; |
6403 | return QAction == Legal || QAction == Libcall || QAction == Custom; |
6404 | }; |
6405 | switch (Opc) { |
6406 | default: |
6407 | return UnableToLegalize; |
6408 | case TargetOpcode::G_CTLZ_ZERO_UNDEF: { |
6409 | // This trivially expands to CTLZ. |
6410 | Observer.changingInstr(MI); |
6411 | MI.setDesc(TII.get(Opcode: TargetOpcode::G_CTLZ)); |
6412 | Observer.changedInstr(MI); |
6413 | return Legalized; |
6414 | } |
6415 | case TargetOpcode::G_CTLZ: { |
6416 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
6417 | unsigned Len = SrcTy.getSizeInBits(); |
6418 | |
6419 | if (isSupported({TargetOpcode::G_CTLZ_ZERO_UNDEF, {DstTy, SrcTy}})) { |
6420 | // If CTLZ_ZERO_UNDEF is supported, emit that and a select for zero. |
6421 | auto CtlzZU = MIRBuilder.buildCTLZ_ZERO_UNDEF(Dst: DstTy, Src0: SrcReg); |
6422 | auto ZeroSrc = MIRBuilder.buildConstant(Res: SrcTy, Val: 0); |
6423 | auto ICmp = MIRBuilder.buildICmp( |
6424 | Pred: CmpInst::ICMP_EQ, Res: SrcTy.changeElementSize(NewEltSize: 1), Op0: SrcReg, Op1: ZeroSrc); |
6425 | auto LenConst = MIRBuilder.buildConstant(Res: DstTy, Val: Len); |
6426 | MIRBuilder.buildSelect(Res: DstReg, Tst: ICmp, Op0: LenConst, Op1: CtlzZU); |
6427 | MI.eraseFromParent(); |
6428 | return Legalized; |
6429 | } |
6430 | // for now, we do this: |
6431 | // NewLen = NextPowerOf2(Len); |
6432 | // x = x | (x >> 1); |
6433 | // x = x | (x >> 2); |
6434 | // ... |
6435 | // x = x | (x >>16); |
6436 | // x = x | (x >>32); // for 64-bit input |
6437 | // Upto NewLen/2 |
6438 | // return Len - popcount(x); |
6439 | // |
6440 | // Ref: "Hacker's Delight" by Henry Warren |
6441 | Register Op = SrcReg; |
6442 | unsigned NewLen = PowerOf2Ceil(A: Len); |
6443 | for (unsigned i = 0; (1U << i) <= (NewLen / 2); ++i) { |
6444 | auto MIBShiftAmt = MIRBuilder.buildConstant(Res: SrcTy, Val: 1ULL << i); |
6445 | auto MIBOp = MIRBuilder.buildOr( |
6446 | Dst: SrcTy, Src0: Op, Src1: MIRBuilder.buildLShr(Dst: SrcTy, Src0: Op, Src1: MIBShiftAmt)); |
6447 | Op = MIBOp.getReg(Idx: 0); |
6448 | } |
6449 | auto MIBPop = MIRBuilder.buildCTPOP(Dst: DstTy, Src0: Op); |
6450 | MIRBuilder.buildSub(Dst: MI.getOperand(i: 0), Src0: MIRBuilder.buildConstant(Res: DstTy, Val: Len), |
6451 | Src1: MIBPop); |
6452 | MI.eraseFromParent(); |
6453 | return Legalized; |
6454 | } |
6455 | case TargetOpcode::G_CTTZ_ZERO_UNDEF: { |
6456 | // This trivially expands to CTTZ. |
6457 | Observer.changingInstr(MI); |
6458 | MI.setDesc(TII.get(Opcode: TargetOpcode::G_CTTZ)); |
6459 | Observer.changedInstr(MI); |
6460 | return Legalized; |
6461 | } |
6462 | case TargetOpcode::G_CTTZ: { |
6463 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
6464 | |
6465 | unsigned Len = SrcTy.getSizeInBits(); |
6466 | if (isSupported({TargetOpcode::G_CTTZ_ZERO_UNDEF, {DstTy, SrcTy}})) { |
6467 | // If CTTZ_ZERO_UNDEF is legal or custom, emit that and a select with |
6468 | // zero. |
6469 | auto CttzZU = MIRBuilder.buildCTTZ_ZERO_UNDEF(Dst: DstTy, Src0: SrcReg); |
6470 | auto Zero = MIRBuilder.buildConstant(Res: SrcTy, Val: 0); |
6471 | auto ICmp = MIRBuilder.buildICmp( |
6472 | Pred: CmpInst::ICMP_EQ, Res: DstTy.changeElementSize(NewEltSize: 1), Op0: SrcReg, Op1: Zero); |
6473 | auto LenConst = MIRBuilder.buildConstant(Res: DstTy, Val: Len); |
6474 | MIRBuilder.buildSelect(Res: DstReg, Tst: ICmp, Op0: LenConst, Op1: CttzZU); |
6475 | MI.eraseFromParent(); |
6476 | return Legalized; |
6477 | } |
6478 | // for now, we use: { return popcount(~x & (x - 1)); } |
6479 | // unless the target has ctlz but not ctpop, in which case we use: |
6480 | // { return 32 - nlz(~x & (x-1)); } |
6481 | // Ref: "Hacker's Delight" by Henry Warren |
6482 | auto MIBCstNeg1 = MIRBuilder.buildConstant(Res: SrcTy, Val: -1); |
6483 | auto MIBNot = MIRBuilder.buildXor(Dst: SrcTy, Src0: SrcReg, Src1: MIBCstNeg1); |
6484 | auto MIBTmp = MIRBuilder.buildAnd( |
6485 | Dst: SrcTy, Src0: MIBNot, Src1: MIRBuilder.buildAdd(Dst: SrcTy, Src0: SrcReg, Src1: MIBCstNeg1)); |
6486 | if (!isSupported({TargetOpcode::G_CTPOP, {SrcTy, SrcTy}}) && |
6487 | isSupported({TargetOpcode::G_CTLZ, {SrcTy, SrcTy}})) { |
6488 | auto MIBCstLen = MIRBuilder.buildConstant(Res: SrcTy, Val: Len); |
6489 | MIRBuilder.buildSub(Dst: MI.getOperand(i: 0), Src0: MIBCstLen, |
6490 | Src1: MIRBuilder.buildCTLZ(Dst: SrcTy, Src0: MIBTmp)); |
6491 | MI.eraseFromParent(); |
6492 | return Legalized; |
6493 | } |
6494 | Observer.changingInstr(MI); |
6495 | MI.setDesc(TII.get(Opcode: TargetOpcode::G_CTPOP)); |
6496 | MI.getOperand(i: 1).setReg(MIBTmp.getReg(Idx: 0)); |
6497 | Observer.changedInstr(MI); |
6498 | return Legalized; |
6499 | } |
6500 | case TargetOpcode::G_CTPOP: { |
6501 | Register SrcReg = MI.getOperand(i: 1).getReg(); |
6502 | LLT Ty = MRI.getType(Reg: SrcReg); |
6503 | unsigned Size = Ty.getSizeInBits(); |
6504 | MachineIRBuilder &B = MIRBuilder; |
6505 | |
6506 | // Count set bits in blocks of 2 bits. Default approach would be |
6507 | // B2Count = { val & 0x55555555 } + { (val >> 1) & 0x55555555 } |
6508 | // We use following formula instead: |
6509 | // B2Count = val - { (val >> 1) & 0x55555555 } |
6510 | // since it gives same result in blocks of 2 with one instruction less. |
6511 | auto C_1 = B.buildConstant(Res: Ty, Val: 1); |
6512 | auto B2Set1LoTo1Hi = B.buildLShr(Dst: Ty, Src0: SrcReg, Src1: C_1); |
6513 | APInt B2Mask1HiTo0 = APInt::getSplat(NewLen: Size, V: APInt(8, 0x55)); |
6514 | auto C_B2Mask1HiTo0 = B.buildConstant(Res: Ty, Val: B2Mask1HiTo0); |
6515 | auto B2Count1Hi = B.buildAnd(Dst: Ty, Src0: B2Set1LoTo1Hi, Src1: C_B2Mask1HiTo0); |
6516 | auto B2Count = B.buildSub(Dst: Ty, Src0: SrcReg, Src1: B2Count1Hi); |
6517 | |
6518 | // In order to get count in blocks of 4 add values from adjacent block of 2. |
6519 | // B4Count = { B2Count & 0x33333333 } + { (B2Count >> 2) & 0x33333333 } |
6520 | auto C_2 = B.buildConstant(Res: Ty, Val: 2); |
6521 | auto B4Set2LoTo2Hi = B.buildLShr(Dst: Ty, Src0: B2Count, Src1: C_2); |
6522 | APInt B4Mask2HiTo0 = APInt::getSplat(NewLen: Size, V: APInt(8, 0x33)); |
6523 | auto C_B4Mask2HiTo0 = B.buildConstant(Res: Ty, Val: B4Mask2HiTo0); |
6524 | auto B4HiB2Count = B.buildAnd(Dst: Ty, Src0: B4Set2LoTo2Hi, Src1: C_B4Mask2HiTo0); |
6525 | auto B4LoB2Count = B.buildAnd(Dst: Ty, Src0: B2Count, Src1: C_B4Mask2HiTo0); |
6526 | auto B4Count = B.buildAdd(Dst: Ty, Src0: B4HiB2Count, Src1: B4LoB2Count); |
6527 | |
6528 | // For count in blocks of 8 bits we don't have to mask high 4 bits before |
6529 | // addition since count value sits in range {0,...,8} and 4 bits are enough |
6530 | // to hold such binary values. After addition high 4 bits still hold count |
6531 | // of set bits in high 4 bit block, set them to zero and get 8 bit result. |
6532 | // B8Count = { B4Count + (B4Count >> 4) } & 0x0F0F0F0F |
6533 | auto C_4 = B.buildConstant(Res: Ty, Val: 4); |
6534 | auto B8HiB4Count = B.buildLShr(Dst: Ty, Src0: B4Count, Src1: C_4); |
6535 | auto B8CountDirty4Hi = B.buildAdd(Dst: Ty, Src0: B8HiB4Count, Src1: B4Count); |
6536 | APInt B8Mask4HiTo0 = APInt::getSplat(NewLen: Size, V: APInt(8, 0x0F)); |
6537 | auto C_B8Mask4HiTo0 = B.buildConstant(Res: Ty, Val: B8Mask4HiTo0); |
6538 | auto B8Count = B.buildAnd(Dst: Ty, Src0: B8CountDirty4Hi, Src1: C_B8Mask4HiTo0); |
6539 | |
6540 | assert(Size<=128 && "Scalar size is too large for CTPOP lower algorithm" ); |
6541 | // 8 bits can hold CTPOP result of 128 bit int or smaller. Mul with this |
6542 | // bitmask will set 8 msb in ResTmp to sum of all B8Counts in 8 bit blocks. |
6543 | auto MulMask = B.buildConstant(Res: Ty, Val: APInt::getSplat(NewLen: Size, V: APInt(8, 0x01))); |
6544 | |
6545 | // Shift count result from 8 high bits to low bits. |
6546 | auto C_SizeM8 = B.buildConstant(Res: Ty, Val: Size - 8); |
6547 | |
6548 | auto IsMulSupported = [this](const LLT Ty) { |
6549 | auto Action = LI.getAction(Query: {TargetOpcode::G_MUL, {Ty}}).Action; |
6550 | return Action == Legal || Action == WidenScalar || Action == Custom; |
6551 | }; |
6552 | if (IsMulSupported(Ty)) { |
6553 | auto ResTmp = B.buildMul(Dst: Ty, Src0: B8Count, Src1: MulMask); |
6554 | B.buildLShr(Dst: MI.getOperand(i: 0).getReg(), Src0: ResTmp, Src1: C_SizeM8); |
6555 | } else { |
6556 | auto ResTmp = B8Count; |
6557 | for (unsigned Shift = 8; Shift < Size; Shift *= 2) { |
6558 | auto ShiftC = B.buildConstant(Res: Ty, Val: Shift); |
6559 | auto Shl = B.buildShl(Dst: Ty, Src0: ResTmp, Src1: ShiftC); |
6560 | ResTmp = B.buildAdd(Dst: Ty, Src0: ResTmp, Src1: Shl); |
6561 | } |
6562 | B.buildLShr(Dst: MI.getOperand(i: 0).getReg(), Src0: ResTmp, Src1: C_SizeM8); |
6563 | } |
6564 | MI.eraseFromParent(); |
6565 | return Legalized; |
6566 | } |
6567 | } |
6568 | } |
6569 | |
6570 | // Check that (every element of) Reg is undef or not an exact multiple of BW. |
6571 | static bool isNonZeroModBitWidthOrUndef(const MachineRegisterInfo &MRI, |
6572 | Register Reg, unsigned BW) { |
6573 | return matchUnaryPredicate( |
6574 | MRI, Reg, |
6575 | Match: [=](const Constant *C) { |
6576 | // Null constant here means an undef. |
6577 | const ConstantInt *CI = dyn_cast_or_null<ConstantInt>(Val: C); |
6578 | return !CI || CI->getValue().urem(RHS: BW) != 0; |
6579 | }, |
6580 | /*AllowUndefs*/ true); |
6581 | } |
6582 | |
6583 | LegalizerHelper::LegalizeResult |
6584 | LegalizerHelper::lowerFunnelShiftWithInverse(MachineInstr &MI) { |
6585 | auto [Dst, X, Y, Z] = MI.getFirst4Regs(); |
6586 | LLT Ty = MRI.getType(Reg: Dst); |
6587 | LLT ShTy = MRI.getType(Reg: Z); |
6588 | |
6589 | unsigned BW = Ty.getScalarSizeInBits(); |
6590 | |
6591 | if (!isPowerOf2_32(Value: BW)) |
6592 | return UnableToLegalize; |
6593 | |
6594 | const bool IsFSHL = MI.getOpcode() == TargetOpcode::G_FSHL; |
6595 | unsigned RevOpcode = IsFSHL ? TargetOpcode::G_FSHR : TargetOpcode::G_FSHL; |
6596 | |
6597 | if (isNonZeroModBitWidthOrUndef(MRI, Reg: Z, BW)) { |
6598 | // fshl X, Y, Z -> fshr X, Y, -Z |
6599 | // fshr X, Y, Z -> fshl X, Y, -Z |
6600 | auto Zero = MIRBuilder.buildConstant(Res: ShTy, Val: 0); |
6601 | Z = MIRBuilder.buildSub(Dst: Ty, Src0: Zero, Src1: Z).getReg(Idx: 0); |
6602 | } else { |
6603 | // fshl X, Y, Z -> fshr (srl X, 1), (fshr X, Y, 1), ~Z |
6604 | // fshr X, Y, Z -> fshl (fshl X, Y, 1), (shl Y, 1), ~Z |
6605 | auto One = MIRBuilder.buildConstant(Res: ShTy, Val: 1); |
6606 | if (IsFSHL) { |
6607 | Y = MIRBuilder.buildInstr(Opc: RevOpcode, DstOps: {Ty}, SrcOps: {X, Y, One}).getReg(Idx: 0); |
6608 | X = MIRBuilder.buildLShr(Dst: Ty, Src0: X, Src1: One).getReg(Idx: 0); |
6609 | } else { |
6610 | X = MIRBuilder.buildInstr(Opc: RevOpcode, DstOps: {Ty}, SrcOps: {X, Y, One}).getReg(Idx: 0); |
6611 | Y = MIRBuilder.buildShl(Dst: Ty, Src0: Y, Src1: One).getReg(Idx: 0); |
6612 | } |
6613 | |
6614 | Z = MIRBuilder.buildNot(Dst: ShTy, Src0: Z).getReg(Idx: 0); |
6615 | } |
6616 | |
6617 | MIRBuilder.buildInstr(Opc: RevOpcode, DstOps: {Dst}, SrcOps: {X, Y, Z}); |
6618 | MI.eraseFromParent(); |
6619 | return Legalized; |
6620 | } |
6621 | |
6622 | LegalizerHelper::LegalizeResult |
6623 | LegalizerHelper::lowerFunnelShiftAsShifts(MachineInstr &MI) { |
6624 | auto [Dst, X, Y, Z] = MI.getFirst4Regs(); |
6625 | LLT Ty = MRI.getType(Reg: Dst); |
6626 | LLT ShTy = MRI.getType(Reg: Z); |
6627 | |
6628 | const unsigned BW = Ty.getScalarSizeInBits(); |
6629 | const bool IsFSHL = MI.getOpcode() == TargetOpcode::G_FSHL; |
6630 | |
6631 | Register ShX, ShY; |
6632 | Register ShAmt, InvShAmt; |
6633 | |
6634 | // FIXME: Emit optimized urem by constant instead of letting it expand later. |
6635 | if (isNonZeroModBitWidthOrUndef(MRI, Reg: Z, BW)) { |
6636 | // fshl: X << C | Y >> (BW - C) |
6637 | // fshr: X << (BW - C) | Y >> C |
6638 | // where C = Z % BW is not zero |
6639 | auto BitWidthC = MIRBuilder.buildConstant(Res: ShTy, Val: BW); |
6640 | ShAmt = MIRBuilder.buildURem(Dst: ShTy, Src0: Z, Src1: BitWidthC).getReg(Idx: 0); |
6641 | InvShAmt = MIRBuilder.buildSub(Dst: ShTy, Src0: BitWidthC, Src1: ShAmt).getReg(Idx: 0); |
6642 | ShX = MIRBuilder.buildShl(Dst: Ty, Src0: X, Src1: IsFSHL ? ShAmt : InvShAmt).getReg(Idx: 0); |
6643 | ShY = MIRBuilder.buildLShr(Dst: Ty, Src0: Y, Src1: IsFSHL ? InvShAmt : ShAmt).getReg(Idx: 0); |
6644 | } else { |
6645 | // fshl: X << (Z % BW) | Y >> 1 >> (BW - 1 - (Z % BW)) |
6646 | // fshr: X << 1 << (BW - 1 - (Z % BW)) | Y >> (Z % BW) |
6647 | auto Mask = MIRBuilder.buildConstant(Res: ShTy, Val: BW - 1); |
6648 | if (isPowerOf2_32(Value: BW)) { |
6649 | // Z % BW -> Z & (BW - 1) |
6650 | ShAmt = MIRBuilder.buildAnd(Dst: ShTy, Src0: Z, Src1: Mask).getReg(Idx: 0); |
6651 | // (BW - 1) - (Z % BW) -> ~Z & (BW - 1) |
6652 | auto NotZ = MIRBuilder.buildNot(Dst: ShTy, Src0: Z); |
6653 | InvShAmt = MIRBuilder.buildAnd(Dst: ShTy, Src0: NotZ, Src1: Mask).getReg(Idx: 0); |
6654 | } else { |
6655 | auto BitWidthC = MIRBuilder.buildConstant(Res: ShTy, Val: BW); |
6656 | ShAmt = MIRBuilder.buildURem(Dst: ShTy, Src0: Z, Src1: BitWidthC).getReg(Idx: 0); |
6657 | InvShAmt = MIRBuilder.buildSub(Dst: ShTy, Src0: Mask, Src1: ShAmt).getReg(Idx: 0); |
6658 | } |
6659 | |
6660 | auto One = MIRBuilder.buildConstant(Res: ShTy, Val: 1); |
6661 | if (IsFSHL) { |
6662 | ShX = MIRBuilder.buildShl(Dst: Ty, Src0: X, Src1: ShAmt).getReg(Idx: 0); |
6663 | auto ShY1 = MIRBuilder.buildLShr(Dst: Ty, Src0: Y, Src1: One); |
6664 | ShY = MIRBuilder.buildLShr(Dst: Ty, Src0: ShY1, Src1: InvShAmt).getReg(Idx: 0); |
6665 | } else { |
6666 | auto ShX1 = MIRBuilder.buildShl(Dst: Ty, Src0: X, Src1: One); |
6667 | ShX = MIRBuilder.buildShl(Dst: Ty, Src0: ShX1, Src1: InvShAmt).getReg(Idx: 0); |
6668 | ShY = MIRBuilder.buildLShr(Dst: Ty, Src0: Y, Src1: ShAmt).getReg(Idx: 0); |
6669 | } |
6670 | } |
6671 | |
6672 | MIRBuilder.buildOr(Dst, Src0: ShX, Src1: ShY); |
6673 | MI.eraseFromParent(); |
6674 | return Legalized; |
6675 | } |
6676 | |
6677 | LegalizerHelper::LegalizeResult |
6678 | LegalizerHelper::lowerFunnelShift(MachineInstr &MI) { |
6679 | // These operations approximately do the following (while avoiding undefined |
6680 | // shifts by BW): |
6681 | // G_FSHL: (X << (Z % BW)) | (Y >> (BW - (Z % BW))) |
6682 | // G_FSHR: (X << (BW - (Z % BW))) | (Y >> (Z % BW)) |
6683 | Register Dst = MI.getOperand(i: 0).getReg(); |
6684 | LLT Ty = MRI.getType(Reg: Dst); |
6685 | LLT ShTy = MRI.getType(Reg: MI.getOperand(i: 3).getReg()); |
6686 | |
6687 | bool IsFSHL = MI.getOpcode() == TargetOpcode::G_FSHL; |
6688 | unsigned RevOpcode = IsFSHL ? TargetOpcode::G_FSHR : TargetOpcode::G_FSHL; |
6689 | |
6690 | // TODO: Use smarter heuristic that accounts for vector legalization. |
6691 | if (LI.getAction(Query: {RevOpcode, {Ty, ShTy}}).Action == Lower) |
6692 | return lowerFunnelShiftAsShifts(MI); |
6693 | |
6694 | // This only works for powers of 2, fallback to shifts if it fails. |
6695 | LegalizerHelper::LegalizeResult Result = lowerFunnelShiftWithInverse(MI); |
6696 | if (Result == UnableToLegalize) |
6697 | return lowerFunnelShiftAsShifts(MI); |
6698 | return Result; |
6699 | } |
6700 | |
6701 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerEXT(MachineInstr &MI) { |
6702 | auto [Dst, Src] = MI.getFirst2Regs(); |
6703 | LLT DstTy = MRI.getType(Reg: Dst); |
6704 | LLT SrcTy = MRI.getType(Reg: Src); |
6705 | |
6706 | uint32_t DstTySize = DstTy.getSizeInBits(); |
6707 | uint32_t DstTyScalarSize = DstTy.getScalarSizeInBits(); |
6708 | uint32_t SrcTyScalarSize = SrcTy.getScalarSizeInBits(); |
6709 | |
6710 | if (!isPowerOf2_32(Value: DstTySize) || !isPowerOf2_32(Value: DstTyScalarSize) || |
6711 | !isPowerOf2_32(Value: SrcTyScalarSize)) |
6712 | return UnableToLegalize; |
6713 | |
6714 | // The step between extend is too large, split it by creating an intermediate |
6715 | // extend instruction |
6716 | if (SrcTyScalarSize * 2 < DstTyScalarSize) { |
6717 | LLT MidTy = SrcTy.changeElementSize(NewEltSize: SrcTyScalarSize * 2); |
6718 | // If the destination type is illegal, split it into multiple statements |
6719 | // zext x -> zext(merge(zext(unmerge), zext(unmerge))) |
6720 | auto NewExt = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {MidTy}, SrcOps: {Src}); |
6721 | // Unmerge the vector |
6722 | LLT EltTy = MidTy.changeElementCount( |
6723 | EC: MidTy.getElementCount().divideCoefficientBy(RHS: 2)); |
6724 | auto UnmergeSrc = MIRBuilder.buildUnmerge(Res: EltTy, Op: NewExt); |
6725 | |
6726 | // ZExt the vectors |
6727 | LLT ZExtResTy = DstTy.changeElementCount( |
6728 | EC: DstTy.getElementCount().divideCoefficientBy(RHS: 2)); |
6729 | auto ZExtRes1 = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {ZExtResTy}, |
6730 | SrcOps: {UnmergeSrc.getReg(Idx: 0)}); |
6731 | auto ZExtRes2 = MIRBuilder.buildInstr(Opc: MI.getOpcode(), DstOps: {ZExtResTy}, |
6732 | SrcOps: {UnmergeSrc.getReg(Idx: 1)}); |
6733 | |
6734 | // Merge the ending vectors |
6735 | MIRBuilder.buildMergeLikeInstr(Res: Dst, Ops: {ZExtRes1, ZExtRes2}); |
6736 | |
6737 | MI.eraseFromParent(); |
6738 | return Legalized; |
6739 | } |
6740 | return UnableToLegalize; |
6741 | } |
6742 | |
6743 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerTRUNC(MachineInstr &MI) { |
6744 | // MachineIRBuilder &MIRBuilder = Helper.MIRBuilder; |
6745 | MachineRegisterInfo &MRI = *MIRBuilder.getMRI(); |
6746 | // Similar to how operand splitting is done in SelectiondDAG, we can handle |
6747 | // %res(v8s8) = G_TRUNC %in(v8s32) by generating: |
6748 | // %inlo(<4x s32>), %inhi(<4 x s32>) = G_UNMERGE %in(<8 x s32>) |
6749 | // %lo16(<4 x s16>) = G_TRUNC %inlo |
6750 | // %hi16(<4 x s16>) = G_TRUNC %inhi |
6751 | // %in16(<8 x s16>) = G_CONCAT_VECTORS %lo16, %hi16 |
6752 | // %res(<8 x s8>) = G_TRUNC %in16 |
6753 | |
6754 | assert(MI.getOpcode() == TargetOpcode::G_TRUNC); |
6755 | |
6756 | Register DstReg = MI.getOperand(i: 0).getReg(); |
6757 | Register SrcReg = MI.getOperand(i: 1).getReg(); |
6758 | LLT DstTy = MRI.getType(Reg: DstReg); |
6759 | LLT SrcTy = MRI.getType(Reg: SrcReg); |
6760 | |
6761 | if (DstTy.isVector() && isPowerOf2_32(Value: DstTy.getNumElements()) && |
6762 | isPowerOf2_32(Value: DstTy.getScalarSizeInBits()) && |
6763 | isPowerOf2_32(Value: SrcTy.getNumElements()) && |
6764 | isPowerOf2_32(Value: SrcTy.getScalarSizeInBits())) { |
6765 | // Split input type. |
6766 | LLT SplitSrcTy = SrcTy.changeElementCount( |
6767 | EC: SrcTy.getElementCount().divideCoefficientBy(RHS: 2)); |
6768 | |
6769 | // First, split the source into two smaller vectors. |
6770 | SmallVector<Register, 2> SplitSrcs; |
6771 | extractParts(Reg: SrcReg, Ty: SplitSrcTy, NumParts: 2, VRegs&: SplitSrcs, MIRBuilder, MRI); |
6772 | |
6773 | // Truncate the splits into intermediate narrower elements. |
6774 | LLT InterTy; |
6775 | if (DstTy.getScalarSizeInBits() * 2 < SrcTy.getScalarSizeInBits()) |
6776 | InterTy = SplitSrcTy.changeElementSize(NewEltSize: DstTy.getScalarSizeInBits() * 2); |
6777 | else |
6778 | InterTy = SplitSrcTy.changeElementSize(NewEltSize: DstTy.getScalarSizeInBits()); |
6779 | for (unsigned I = 0; I < SplitSrcs.size(); ++I) { |
6780 | SplitSrcs[I] = MIRBuilder.buildTrunc(Res: InterTy, Op: SplitSrcs[I]).getReg(Idx: 0); |
6781 | } |
6782 | |
6783 | // Combine the new truncates into one vector |
6784 | auto Merge = MIRBuilder.buildMergeLikeInstr( |
6785 | Res: DstTy.changeElementSize(NewEltSize: InterTy.getScalarSizeInBits()), Ops: SplitSrcs); |
6786 | |
6787 | // Truncate the new vector to the final result type |
6788 | if (DstTy.getScalarSizeInBits() * 2 < SrcTy.getScalarSizeInBits()) |
6789 | MIRBuilder.buildTrunc(Res: MI.getOperand(i: 0).getReg(), Op: Merge.getReg(Idx: 0)); |
6790 | else |
6791 | MIRBuilder.buildCopy(Res: MI.getOperand(i: 0).getReg(), Op: Merge.getReg(Idx: 0)); |
6792 | |
6793 | MI.eraseFromParent(); |
6794 | |
6795 | return Legalized; |
6796 | } |
6797 | return UnableToLegalize; |
6798 | } |
6799 | |
6800 | LegalizerHelper::LegalizeResult |
6801 | LegalizerHelper::lowerRotateWithReverseRotate(MachineInstr &MI) { |
6802 | auto [Dst, DstTy, Src, SrcTy, Amt, AmtTy] = MI.getFirst3RegLLTs(); |
6803 | auto Zero = MIRBuilder.buildConstant(Res: AmtTy, Val: 0); |
6804 | bool IsLeft = MI.getOpcode() == TargetOpcode::G_ROTL; |
6805 | unsigned RevRot = IsLeft ? TargetOpcode::G_ROTR : TargetOpcode::G_ROTL; |
6806 | auto Neg = MIRBuilder.buildSub(Dst: AmtTy, Src0: Zero, Src1: Amt); |
6807 | MIRBuilder.buildInstr(Opc: RevRot, DstOps: {Dst}, SrcOps: {Src, Neg}); |
6808 | MI.eraseFromParent(); |
6809 | return Legalized; |
6810 | } |
6811 | |
6812 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerRotate(MachineInstr &MI) { |
6813 | auto [Dst, DstTy, Src, SrcTy, Amt, AmtTy] = MI.getFirst3RegLLTs(); |
6814 | |
6815 | unsigned EltSizeInBits = DstTy.getScalarSizeInBits(); |
6816 | bool IsLeft = MI.getOpcode() == TargetOpcode::G_ROTL; |
6817 | |
6818 | MIRBuilder.setInstrAndDebugLoc(MI); |
6819 | |
6820 | // If a rotate in the other direction is supported, use it. |
6821 | unsigned RevRot = IsLeft ? TargetOpcode::G_ROTR : TargetOpcode::G_ROTL; |
6822 | if (LI.isLegalOrCustom(Query: {RevRot, {DstTy, SrcTy}}) && |
6823 | isPowerOf2_32(Value: EltSizeInBits)) |
6824 | return lowerRotateWithReverseRotate(MI); |
6825 | |
6826 | // If a funnel shift is supported, use it. |
6827 | unsigned FShOpc = IsLeft ? TargetOpcode::G_FSHL : TargetOpcode::G_FSHR; |
6828 | unsigned RevFsh = !IsLeft ? TargetOpcode::G_FSHL : TargetOpcode::G_FSHR; |
6829 | bool IsFShLegal = false; |
6830 | if ((IsFShLegal = LI.isLegalOrCustom(Query: {FShOpc, {DstTy, AmtTy}})) || |
6831 | LI.isLegalOrCustom(Query: {RevFsh, {DstTy, AmtTy}})) { |
6832 | auto buildFunnelShift = [&](unsigned Opc, Register R1, Register R2, |
6833 | Register R3) { |
6834 | MIRBuilder.buildInstr(Opc, DstOps: {R1}, SrcOps: {R2, R2, R3}); |
6835 | MI.eraseFromParent(); |
6836 | return Legalized; |
6837 | }; |
6838 | // If a funnel shift in the other direction is supported, use it. |
6839 | if (IsFShLegal) { |
6840 | return buildFunnelShift(FShOpc, Dst, Src, Amt); |
6841 | } else if (isPowerOf2_32(Value: EltSizeInBits)) { |
6842 | Amt = MIRBuilder.buildNeg(Dst: DstTy, Src0: Amt).getReg(Idx: 0); |
6843 | return buildFunnelShift(RevFsh, Dst, Src, Amt); |
6844 | } |
6845 | } |
6846 | |
6847 | auto Zero = MIRBuilder.buildConstant(Res: AmtTy, Val: 0); |
6848 | unsigned ShOpc = IsLeft ? TargetOpcode::G_SHL : TargetOpcode::G_LSHR; |
6849 | unsigned RevShiftOpc = IsLeft ? TargetOpcode::G_LSHR : TargetOpcode::G_SHL; |
6850 | auto BitWidthMinusOneC = MIRBuilder.buildConstant(Res: AmtTy, Val: EltSizeInBits - 1); |
6851 | Register ShVal; |
6852 | Register RevShiftVal; |
6853 | if (isPowerOf2_32(Value: EltSizeInBits)) { |
6854 | // (rotl x, c) -> x << (c & (w - 1)) | x >> (-c & (w - 1)) |
6855 | // (rotr x, c) -> x >> (c & (w - 1)) | x << (-c & (w - 1)) |
6856 | auto NegAmt = MIRBuilder.buildSub(Dst: AmtTy, Src0: Zero, Src1: Amt); |
6857 | auto ShAmt = MIRBuilder.buildAnd(Dst: AmtTy, Src0: Amt, Src1: BitWidthMinusOneC); |
6858 | ShVal = MIRBuilder.buildInstr(Opc: ShOpc, DstOps: {DstTy}, SrcOps: {Src, ShAmt}).getReg(Idx: 0); |
6859 | auto RevAmt = MIRBuilder.buildAnd(Dst: AmtTy, Src0: NegAmt, Src1: BitWidthMinusOneC); |
6860 | RevShiftVal = |
6861 | MIRBuilder.buildInstr(Opc: RevShiftOpc, DstOps: {DstTy}, SrcOps: {Src, RevAmt}).getReg(Idx: 0); |
6862 | } else { |
6863 | // (rotl x, c) -> x << (c % w) | x >> 1 >> (w - 1 - (c % w)) |
6864 | // (rotr x, c) -> x >> (c % w) | x << 1 << (w - 1 - (c % w)) |
6865 | auto BitWidthC = MIRBuilder.buildConstant(Res: AmtTy, Val: EltSizeInBits); |
6866 | auto ShAmt = MIRBuilder.buildURem(Dst: AmtTy, Src0: Amt, Src1: BitWidthC); |
6867 | ShVal = MIRBuilder.buildInstr(Opc: ShOpc, DstOps: {DstTy}, SrcOps: {Src, ShAmt}).getReg(Idx: 0); |
6868 | auto RevAmt = MIRBuilder.buildSub(Dst: AmtTy, Src0: BitWidthMinusOneC, Src1: ShAmt); |
6869 | auto One = MIRBuilder.buildConstant(Res: AmtTy, Val: 1); |
6870 | auto Inner = MIRBuilder.buildInstr(Opc: RevShiftOpc, DstOps: {DstTy}, SrcOps: {Src, One}); |
6871 | RevShiftVal = |
6872 | MIRBuilder.buildInstr(Opc: RevShiftOpc, DstOps: {DstTy}, SrcOps: {Inner, RevAmt}).getReg(Idx: 0); |
6873 | } |
6874 | MIRBuilder.buildOr(Dst, Src0: ShVal, Src1: RevShiftVal); |
6875 | MI.eraseFromParent(); |
6876 | return Legalized; |
6877 | } |
6878 | |
6879 | // Expand s32 = G_UITOFP s64 using bit operations to an IEEE float |
6880 | // representation. |
6881 | LegalizerHelper::LegalizeResult |
6882 | LegalizerHelper::lowerU64ToF32BitOps(MachineInstr &MI) { |
6883 | auto [Dst, Src] = MI.getFirst2Regs(); |
6884 | const LLT S64 = LLT::scalar(SizeInBits: 64); |
6885 | const LLT S32 = LLT::scalar(SizeInBits: 32); |
6886 | const LLT S1 = LLT::scalar(SizeInBits: 1); |
6887 | |
6888 | assert(MRI.getType(Src) == S64 && MRI.getType(Dst) == S32); |
6889 | |
6890 | // unsigned cul2f(ulong u) { |
6891 | // uint lz = clz(u); |
6892 | // uint e = (u != 0) ? 127U + 63U - lz : 0; |
6893 | // u = (u << lz) & 0x7fffffffffffffffUL; |
6894 | // ulong t = u & 0xffffffffffUL; |
6895 | // uint v = (e << 23) | (uint)(u >> 40); |
6896 | // uint r = t > 0x8000000000UL ? 1U : (t == 0x8000000000UL ? v & 1U : 0U); |
6897 | // return as_float(v + r); |
6898 | // } |
6899 | |
6900 | auto Zero32 = MIRBuilder.buildConstant(Res: S32, Val: 0); |
6901 | auto Zero64 = MIRBuilder.buildConstant(Res: S64, Val: 0); |
6902 | |
6903 | auto LZ = MIRBuilder.buildCTLZ_ZERO_UNDEF(Dst: S32, Src0: Src); |
6904 | |
6905 | auto K = MIRBuilder.buildConstant(Res: S32, Val: 127U + 63U); |
6906 | auto Sub = MIRBuilder.buildSub(Dst: S32, Src0: K, Src1: LZ); |
6907 | |
6908 | auto NotZero = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: S1, Op0: Src, Op1: Zero64); |
6909 | auto E = MIRBuilder.buildSelect(Res: S32, Tst: NotZero, Op0: Sub, Op1: Zero32); |
6910 | |
6911 | auto Mask0 = MIRBuilder.buildConstant(Res: S64, Val: (-1ULL) >> 1); |
6912 | auto ShlLZ = MIRBuilder.buildShl(Dst: S64, Src0: Src, Src1: LZ); |
6913 | |
6914 | auto U = MIRBuilder.buildAnd(Dst: S64, Src0: ShlLZ, Src1: Mask0); |
6915 | |
6916 | auto Mask1 = MIRBuilder.buildConstant(Res: S64, Val: 0xffffffffffULL); |
6917 | auto T = MIRBuilder.buildAnd(Dst: S64, Src0: U, Src1: Mask1); |
6918 | |
6919 | auto UShl = MIRBuilder.buildLShr(Dst: S64, Src0: U, Src1: MIRBuilder.buildConstant(Res: S64, Val: 40)); |
6920 | auto ShlE = MIRBuilder.buildShl(Dst: S32, Src0: E, Src1: MIRBuilder.buildConstant(Res: S32, Val: 23)); |
6921 | auto V = MIRBuilder.buildOr(Dst: S32, Src0: ShlE, Src1: MIRBuilder.buildTrunc(Res: S32, Op: UShl)); |
6922 | |
6923 | auto C = MIRBuilder.buildConstant(Res: S64, Val: 0x8000000000ULL); |
6924 | auto RCmp = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_UGT, Res: S1, Op0: T, Op1: C); |
6925 | auto TCmp = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_EQ, Res: S1, Op0: T, Op1: C); |
6926 | auto One = MIRBuilder.buildConstant(Res: S32, Val: 1); |
6927 | |
6928 | auto VTrunc1 = MIRBuilder.buildAnd(Dst: S32, Src0: V, Src1: One); |
6929 | auto Select0 = MIRBuilder.buildSelect(Res: S32, Tst: TCmp, Op0: VTrunc1, Op1: Zero32); |
6930 | auto R = MIRBuilder.buildSelect(Res: S32, Tst: RCmp, Op0: One, Op1: Select0); |
6931 | MIRBuilder.buildAdd(Dst, Src0: V, Src1: R); |
6932 | |
6933 | MI.eraseFromParent(); |
6934 | return Legalized; |
6935 | } |
6936 | |
6937 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerUITOFP(MachineInstr &MI) { |
6938 | auto [Dst, DstTy, Src, SrcTy] = MI.getFirst2RegLLTs(); |
6939 | |
6940 | if (SrcTy == LLT::scalar(SizeInBits: 1)) { |
6941 | auto True = MIRBuilder.buildFConstant(Res: DstTy, Val: 1.0); |
6942 | auto False = MIRBuilder.buildFConstant(Res: DstTy, Val: 0.0); |
6943 | MIRBuilder.buildSelect(Res: Dst, Tst: Src, Op0: True, Op1: False); |
6944 | MI.eraseFromParent(); |
6945 | return Legalized; |
6946 | } |
6947 | |
6948 | if (SrcTy != LLT::scalar(SizeInBits: 64)) |
6949 | return UnableToLegalize; |
6950 | |
6951 | if (DstTy == LLT::scalar(SizeInBits: 32)) { |
6952 | // TODO: SelectionDAG has several alternative expansions to port which may |
6953 | // be more reasonble depending on the available instructions. If a target |
6954 | // has sitofp, does not have CTLZ, or can efficiently use f64 as an |
6955 | // intermediate type, this is probably worse. |
6956 | return lowerU64ToF32BitOps(MI); |
6957 | } |
6958 | |
6959 | return UnableToLegalize; |
6960 | } |
6961 | |
6962 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerSITOFP(MachineInstr &MI) { |
6963 | auto [Dst, DstTy, Src, SrcTy] = MI.getFirst2RegLLTs(); |
6964 | |
6965 | const LLT S64 = LLT::scalar(SizeInBits: 64); |
6966 | const LLT S32 = LLT::scalar(SizeInBits: 32); |
6967 | const LLT S1 = LLT::scalar(SizeInBits: 1); |
6968 | |
6969 | if (SrcTy == S1) { |
6970 | auto True = MIRBuilder.buildFConstant(Res: DstTy, Val: -1.0); |
6971 | auto False = MIRBuilder.buildFConstant(Res: DstTy, Val: 0.0); |
6972 | MIRBuilder.buildSelect(Res: Dst, Tst: Src, Op0: True, Op1: False); |
6973 | MI.eraseFromParent(); |
6974 | return Legalized; |
6975 | } |
6976 | |
6977 | if (SrcTy != S64) |
6978 | return UnableToLegalize; |
6979 | |
6980 | if (DstTy == S32) { |
6981 | // signed cl2f(long l) { |
6982 | // long s = l >> 63; |
6983 | // float r = cul2f((l + s) ^ s); |
6984 | // return s ? -r : r; |
6985 | // } |
6986 | Register L = Src; |
6987 | auto SignBit = MIRBuilder.buildConstant(Res: S64, Val: 63); |
6988 | auto S = MIRBuilder.buildAShr(Dst: S64, Src0: L, Src1: SignBit); |
6989 | |
6990 | auto LPlusS = MIRBuilder.buildAdd(Dst: S64, Src0: L, Src1: S); |
6991 | auto Xor = MIRBuilder.buildXor(Dst: S64, Src0: LPlusS, Src1: S); |
6992 | auto R = MIRBuilder.buildUITOFP(Dst: S32, Src0: Xor); |
6993 | |
6994 | auto RNeg = MIRBuilder.buildFNeg(Dst: S32, Src0: R); |
6995 | auto SignNotZero = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: S1, Op0: S, |
6996 | Op1: MIRBuilder.buildConstant(Res: S64, Val: 0)); |
6997 | MIRBuilder.buildSelect(Res: Dst, Tst: SignNotZero, Op0: RNeg, Op1: R); |
6998 | MI.eraseFromParent(); |
6999 | return Legalized; |
7000 | } |
7001 | |
7002 | return UnableToLegalize; |
7003 | } |
7004 | |
7005 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerFPTOUI(MachineInstr &MI) { |
7006 | auto [Dst, DstTy, Src, SrcTy] = MI.getFirst2RegLLTs(); |
7007 | const LLT S64 = LLT::scalar(SizeInBits: 64); |
7008 | const LLT S32 = LLT::scalar(SizeInBits: 32); |
7009 | |
7010 | if (SrcTy != S64 && SrcTy != S32) |
7011 | return UnableToLegalize; |
7012 | if (DstTy != S32 && DstTy != S64) |
7013 | return UnableToLegalize; |
7014 | |
7015 | // FPTOSI gives same result as FPTOUI for positive signed integers. |
7016 | // FPTOUI needs to deal with fp values that convert to unsigned integers |
7017 | // greater or equal to 2^31 for float or 2^63 for double. For brevity 2^Exp. |
7018 | |
7019 | APInt TwoPExpInt = APInt::getSignMask(BitWidth: DstTy.getSizeInBits()); |
7020 | APFloat TwoPExpFP(SrcTy.getSizeInBits() == 32 ? APFloat::IEEEsingle() |
7021 | : APFloat::IEEEdouble(), |
7022 | APInt::getZero(numBits: SrcTy.getSizeInBits())); |
7023 | TwoPExpFP.convertFromAPInt(Input: TwoPExpInt, IsSigned: false, RM: APFloat::rmNearestTiesToEven); |
7024 | |
7025 | MachineInstrBuilder FPTOSI = MIRBuilder.buildFPTOSI(Dst: DstTy, Src0: Src); |
7026 | |
7027 | MachineInstrBuilder Threshold = MIRBuilder.buildFConstant(Res: SrcTy, Val: TwoPExpFP); |
7028 | // For fp Value greater or equal to Threshold(2^Exp), we use FPTOSI on |
7029 | // (Value - 2^Exp) and add 2^Exp by setting highest bit in result to 1. |
7030 | MachineInstrBuilder FSub = MIRBuilder.buildFSub(Dst: SrcTy, Src0: Src, Src1: Threshold); |
7031 | MachineInstrBuilder ResLowBits = MIRBuilder.buildFPTOSI(Dst: DstTy, Src0: FSub); |
7032 | MachineInstrBuilder ResHighBit = MIRBuilder.buildConstant(Res: DstTy, Val: TwoPExpInt); |
7033 | MachineInstrBuilder Res = MIRBuilder.buildXor(Dst: DstTy, Src0: ResLowBits, Src1: ResHighBit); |
7034 | |
7035 | const LLT S1 = LLT::scalar(SizeInBits: 1); |
7036 | |
7037 | MachineInstrBuilder FCMP = |
7038 | MIRBuilder.buildFCmp(Pred: CmpInst::FCMP_ULT, Res: S1, Op0: Src, Op1: Threshold); |
7039 | MIRBuilder.buildSelect(Res: Dst, Tst: FCMP, Op0: FPTOSI, Op1: Res); |
7040 | |
7041 | MI.eraseFromParent(); |
7042 | return Legalized; |
7043 | } |
7044 | |
7045 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerFPTOSI(MachineInstr &MI) { |
7046 | auto [Dst, DstTy, Src, SrcTy] = MI.getFirst2RegLLTs(); |
7047 | const LLT S64 = LLT::scalar(SizeInBits: 64); |
7048 | const LLT S32 = LLT::scalar(SizeInBits: 32); |
7049 | |
7050 | // FIXME: Only f32 to i64 conversions are supported. |
7051 | if (SrcTy.getScalarType() != S32 || DstTy.getScalarType() != S64) |
7052 | return UnableToLegalize; |
7053 | |
7054 | // Expand f32 -> i64 conversion |
7055 | // This algorithm comes from compiler-rt's implementation of fixsfdi: |
7056 | // https://github.com/llvm/llvm-project/blob/main/compiler-rt/lib/builtins/fixsfdi.c |
7057 | |
7058 | unsigned SrcEltBits = SrcTy.getScalarSizeInBits(); |
7059 | |
7060 | auto ExponentMask = MIRBuilder.buildConstant(Res: SrcTy, Val: 0x7F800000); |
7061 | auto ExponentLoBit = MIRBuilder.buildConstant(Res: SrcTy, Val: 23); |
7062 | |
7063 | auto AndExpMask = MIRBuilder.buildAnd(Dst: SrcTy, Src0: Src, Src1: ExponentMask); |
7064 | auto ExponentBits = MIRBuilder.buildLShr(Dst: SrcTy, Src0: AndExpMask, Src1: ExponentLoBit); |
7065 | |
7066 | auto SignMask = MIRBuilder.buildConstant(Res: SrcTy, |
7067 | Val: APInt::getSignMask(BitWidth: SrcEltBits)); |
7068 | auto AndSignMask = MIRBuilder.buildAnd(Dst: SrcTy, Src0: Src, Src1: SignMask); |
7069 | auto SignLowBit = MIRBuilder.buildConstant(Res: SrcTy, Val: SrcEltBits - 1); |
7070 | auto Sign = MIRBuilder.buildAShr(Dst: SrcTy, Src0: AndSignMask, Src1: SignLowBit); |
7071 | Sign = MIRBuilder.buildSExt(Res: DstTy, Op: Sign); |
7072 | |
7073 | auto MantissaMask = MIRBuilder.buildConstant(Res: SrcTy, Val: 0x007FFFFF); |
7074 | auto AndMantissaMask = MIRBuilder.buildAnd(Dst: SrcTy, Src0: Src, Src1: MantissaMask); |
7075 | auto K = MIRBuilder.buildConstant(Res: SrcTy, Val: 0x00800000); |
7076 | |
7077 | auto R = MIRBuilder.buildOr(Dst: SrcTy, Src0: AndMantissaMask, Src1: K); |
7078 | R = MIRBuilder.buildZExt(Res: DstTy, Op: R); |
7079 | |
7080 | auto Bias = MIRBuilder.buildConstant(Res: SrcTy, Val: 127); |
7081 | auto Exponent = MIRBuilder.buildSub(Dst: SrcTy, Src0: ExponentBits, Src1: Bias); |
7082 | auto SubExponent = MIRBuilder.buildSub(Dst: SrcTy, Src0: Exponent, Src1: ExponentLoBit); |
7083 | auto ExponentSub = MIRBuilder.buildSub(Dst: SrcTy, Src0: ExponentLoBit, Src1: Exponent); |
7084 | |
7085 | auto Shl = MIRBuilder.buildShl(Dst: DstTy, Src0: R, Src1: SubExponent); |
7086 | auto Srl = MIRBuilder.buildLShr(Dst: DstTy, Src0: R, Src1: ExponentSub); |
7087 | |
7088 | const LLT S1 = LLT::scalar(SizeInBits: 1); |
7089 | auto CmpGt = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_SGT, |
7090 | Res: S1, Op0: Exponent, Op1: ExponentLoBit); |
7091 | |
7092 | R = MIRBuilder.buildSelect(Res: DstTy, Tst: CmpGt, Op0: Shl, Op1: Srl); |
7093 | |
7094 | auto XorSign = MIRBuilder.buildXor(Dst: DstTy, Src0: R, Src1: Sign); |
7095 | auto Ret = MIRBuilder.buildSub(Dst: DstTy, Src0: XorSign, Src1: Sign); |
7096 | |
7097 | auto ZeroSrcTy = MIRBuilder.buildConstant(Res: SrcTy, Val: 0); |
7098 | |
7099 | auto ExponentLt0 = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_SLT, |
7100 | Res: S1, Op0: Exponent, Op1: ZeroSrcTy); |
7101 | |
7102 | auto ZeroDstTy = MIRBuilder.buildConstant(Res: DstTy, Val: 0); |
7103 | MIRBuilder.buildSelect(Res: Dst, Tst: ExponentLt0, Op0: ZeroDstTy, Op1: Ret); |
7104 | |
7105 | MI.eraseFromParent(); |
7106 | return Legalized; |
7107 | } |
7108 | |
7109 | // f64 -> f16 conversion using round-to-nearest-even rounding mode. |
7110 | LegalizerHelper::LegalizeResult |
7111 | LegalizerHelper::lowerFPTRUNC_F64_TO_F16(MachineInstr &MI) { |
7112 | const LLT S1 = LLT::scalar(SizeInBits: 1); |
7113 | const LLT S32 = LLT::scalar(SizeInBits: 32); |
7114 | |
7115 | auto [Dst, Src] = MI.getFirst2Regs(); |
7116 | assert(MRI.getType(Dst).getScalarType() == LLT::scalar(16) && |
7117 | MRI.getType(Src).getScalarType() == LLT::scalar(64)); |
7118 | |
7119 | if (MRI.getType(Reg: Src).isVector()) // TODO: Handle vectors directly. |
7120 | return UnableToLegalize; |
7121 | |
7122 | if (MIRBuilder.getMF().getTarget().Options.UnsafeFPMath) { |
7123 | unsigned Flags = MI.getFlags(); |
7124 | auto Src32 = MIRBuilder.buildFPTrunc(Res: S32, Op: Src, Flags); |
7125 | MIRBuilder.buildFPTrunc(Res: Dst, Op: Src32, Flags); |
7126 | MI.eraseFromParent(); |
7127 | return Legalized; |
7128 | } |
7129 | |
7130 | const unsigned ExpMask = 0x7ff; |
7131 | const unsigned ExpBiasf64 = 1023; |
7132 | const unsigned ExpBiasf16 = 15; |
7133 | |
7134 | auto Unmerge = MIRBuilder.buildUnmerge(Res: S32, Op: Src); |
7135 | Register U = Unmerge.getReg(Idx: 0); |
7136 | Register UH = Unmerge.getReg(Idx: 1); |
7137 | |
7138 | auto E = MIRBuilder.buildLShr(Dst: S32, Src0: UH, Src1: MIRBuilder.buildConstant(Res: S32, Val: 20)); |
7139 | E = MIRBuilder.buildAnd(Dst: S32, Src0: E, Src1: MIRBuilder.buildConstant(Res: S32, Val: ExpMask)); |
7140 | |
7141 | // Subtract the fp64 exponent bias (1023) to get the real exponent and |
7142 | // add the f16 bias (15) to get the biased exponent for the f16 format. |
7143 | E = MIRBuilder.buildAdd( |
7144 | Dst: S32, Src0: E, Src1: MIRBuilder.buildConstant(Res: S32, Val: -ExpBiasf64 + ExpBiasf16)); |
7145 | |
7146 | auto M = MIRBuilder.buildLShr(Dst: S32, Src0: UH, Src1: MIRBuilder.buildConstant(Res: S32, Val: 8)); |
7147 | M = MIRBuilder.buildAnd(Dst: S32, Src0: M, Src1: MIRBuilder.buildConstant(Res: S32, Val: 0xffe)); |
7148 | |
7149 | auto MaskedSig = MIRBuilder.buildAnd(Dst: S32, Src0: UH, |
7150 | Src1: MIRBuilder.buildConstant(Res: S32, Val: 0x1ff)); |
7151 | MaskedSig = MIRBuilder.buildOr(Dst: S32, Src0: MaskedSig, Src1: U); |
7152 | |
7153 | auto Zero = MIRBuilder.buildConstant(Res: S32, Val: 0); |
7154 | auto SigCmpNE0 = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: S1, Op0: MaskedSig, Op1: Zero); |
7155 | auto Lo40Set = MIRBuilder.buildZExt(Res: S32, Op: SigCmpNE0); |
7156 | M = MIRBuilder.buildOr(Dst: S32, Src0: M, Src1: Lo40Set); |
7157 | |
7158 | // (M != 0 ? 0x0200 : 0) | 0x7c00; |
7159 | auto Bits0x200 = MIRBuilder.buildConstant(Res: S32, Val: 0x0200); |
7160 | auto CmpM_NE0 = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: S1, Op0: M, Op1: Zero); |
7161 | auto SelectCC = MIRBuilder.buildSelect(Res: S32, Tst: CmpM_NE0, Op0: Bits0x200, Op1: Zero); |
7162 | |
7163 | auto Bits0x7c00 = MIRBuilder.buildConstant(Res: S32, Val: 0x7c00); |
7164 | auto I = MIRBuilder.buildOr(Dst: S32, Src0: SelectCC, Src1: Bits0x7c00); |
7165 | |
7166 | // N = M | (E << 12); |
7167 | auto EShl12 = MIRBuilder.buildShl(Dst: S32, Src0: E, Src1: MIRBuilder.buildConstant(Res: S32, Val: 12)); |
7168 | auto N = MIRBuilder.buildOr(Dst: S32, Src0: M, Src1: EShl12); |
7169 | |
7170 | // B = clamp(1-E, 0, 13); |
7171 | auto One = MIRBuilder.buildConstant(Res: S32, Val: 1); |
7172 | auto OneSubExp = MIRBuilder.buildSub(Dst: S32, Src0: One, Src1: E); |
7173 | auto B = MIRBuilder.buildSMax(Dst: S32, Src0: OneSubExp, Src1: Zero); |
7174 | B = MIRBuilder.buildSMin(Dst: S32, Src0: B, Src1: MIRBuilder.buildConstant(Res: S32, Val: 13)); |
7175 | |
7176 | auto SigSetHigh = MIRBuilder.buildOr(Dst: S32, Src0: M, |
7177 | Src1: MIRBuilder.buildConstant(Res: S32, Val: 0x1000)); |
7178 | |
7179 | auto D = MIRBuilder.buildLShr(Dst: S32, Src0: SigSetHigh, Src1: B); |
7180 | auto D0 = MIRBuilder.buildShl(Dst: S32, Src0: D, Src1: B); |
7181 | |
7182 | auto D0_NE_SigSetHigh = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: S1, |
7183 | Op0: D0, Op1: SigSetHigh); |
7184 | auto D1 = MIRBuilder.buildZExt(Res: S32, Op: D0_NE_SigSetHigh); |
7185 | D = MIRBuilder.buildOr(Dst: S32, Src0: D, Src1: D1); |
7186 | |
7187 | auto CmpELtOne = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_SLT, Res: S1, Op0: E, Op1: One); |
7188 | auto V = MIRBuilder.buildSelect(Res: S32, Tst: CmpELtOne, Op0: D, Op1: N); |
7189 | |
7190 | auto VLow3 = MIRBuilder.buildAnd(Dst: S32, Src0: V, Src1: MIRBuilder.buildConstant(Res: S32, Val: 7)); |
7191 | V = MIRBuilder.buildLShr(Dst: S32, Src0: V, Src1: MIRBuilder.buildConstant(Res: S32, Val: 2)); |
7192 | |
7193 | auto VLow3Eq3 = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_EQ, Res: S1, Op0: VLow3, |
7194 | Op1: MIRBuilder.buildConstant(Res: S32, Val: 3)); |
7195 | auto V0 = MIRBuilder.buildZExt(Res: S32, Op: VLow3Eq3); |
7196 | |
7197 | auto VLow3Gt5 = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_SGT, Res: S1, Op0: VLow3, |
7198 | Op1: MIRBuilder.buildConstant(Res: S32, Val: 5)); |
7199 | auto V1 = MIRBuilder.buildZExt(Res: S32, Op: VLow3Gt5); |
7200 | |
7201 | V1 = MIRBuilder.buildOr(Dst: S32, Src0: V0, Src1: V1); |
7202 | V = MIRBuilder.buildAdd(Dst: S32, Src0: V, Src1: V1); |
7203 | |
7204 | auto CmpEGt30 = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_SGT, Res: S1, |
7205 | Op0: E, Op1: MIRBuilder.buildConstant(Res: S32, Val: 30)); |
7206 | V = MIRBuilder.buildSelect(Res: S32, Tst: CmpEGt30, |
7207 | Op0: MIRBuilder.buildConstant(Res: S32, Val: 0x7c00), Op1: V); |
7208 | |
7209 | auto CmpEGt1039 = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_EQ, Res: S1, |
7210 | Op0: E, Op1: MIRBuilder.buildConstant(Res: S32, Val: 1039)); |
7211 | V = MIRBuilder.buildSelect(Res: S32, Tst: CmpEGt1039, Op0: I, Op1: V); |
7212 | |
7213 | // Extract the sign bit. |
7214 | auto Sign = MIRBuilder.buildLShr(Dst: S32, Src0: UH, Src1: MIRBuilder.buildConstant(Res: S32, Val: 16)); |
7215 | Sign = MIRBuilder.buildAnd(Dst: S32, Src0: Sign, Src1: MIRBuilder.buildConstant(Res: S32, Val: 0x8000)); |
7216 | |
7217 | // Insert the sign bit |
7218 | V = MIRBuilder.buildOr(Dst: S32, Src0: Sign, Src1: V); |
7219 | |
7220 | MIRBuilder.buildTrunc(Res: Dst, Op: V); |
7221 | MI.eraseFromParent(); |
7222 | return Legalized; |
7223 | } |
7224 | |
7225 | LegalizerHelper::LegalizeResult |
7226 | LegalizerHelper::lowerFPTRUNC(MachineInstr &MI) { |
7227 | auto [DstTy, SrcTy] = MI.getFirst2LLTs(); |
7228 | const LLT S64 = LLT::scalar(SizeInBits: 64); |
7229 | const LLT S16 = LLT::scalar(SizeInBits: 16); |
7230 | |
7231 | if (DstTy.getScalarType() == S16 && SrcTy.getScalarType() == S64) |
7232 | return lowerFPTRUNC_F64_TO_F16(MI); |
7233 | |
7234 | return UnableToLegalize; |
7235 | } |
7236 | |
7237 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerFPOWI(MachineInstr &MI) { |
7238 | auto [Dst, Src0, Src1] = MI.getFirst3Regs(); |
7239 | LLT Ty = MRI.getType(Reg: Dst); |
7240 | |
7241 | auto CvtSrc1 = MIRBuilder.buildSITOFP(Dst: Ty, Src0: Src1); |
7242 | MIRBuilder.buildFPow(Dst, Src0, Src1: CvtSrc1, Flags: MI.getFlags()); |
7243 | MI.eraseFromParent(); |
7244 | return Legalized; |
7245 | } |
7246 | |
7247 | static CmpInst::Predicate minMaxToCompare(unsigned Opc) { |
7248 | switch (Opc) { |
7249 | case TargetOpcode::G_SMIN: |
7250 | return CmpInst::ICMP_SLT; |
7251 | case TargetOpcode::G_SMAX: |
7252 | return CmpInst::ICMP_SGT; |
7253 | case TargetOpcode::G_UMIN: |
7254 | return CmpInst::ICMP_ULT; |
7255 | case TargetOpcode::G_UMAX: |
7256 | return CmpInst::ICMP_UGT; |
7257 | default: |
7258 | llvm_unreachable("not in integer min/max" ); |
7259 | } |
7260 | } |
7261 | |
7262 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerMinMax(MachineInstr &MI) { |
7263 | auto [Dst, Src0, Src1] = MI.getFirst3Regs(); |
7264 | |
7265 | const CmpInst::Predicate Pred = minMaxToCompare(Opc: MI.getOpcode()); |
7266 | LLT CmpType = MRI.getType(Reg: Dst).changeElementSize(NewEltSize: 1); |
7267 | |
7268 | auto Cmp = MIRBuilder.buildICmp(Pred, Res: CmpType, Op0: Src0, Op1: Src1); |
7269 | MIRBuilder.buildSelect(Res: Dst, Tst: Cmp, Op0: Src0, Op1: Src1); |
7270 | |
7271 | MI.eraseFromParent(); |
7272 | return Legalized; |
7273 | } |
7274 | |
7275 | LegalizerHelper::LegalizeResult |
7276 | LegalizerHelper::lowerThreewayCompare(MachineInstr &MI) { |
7277 | GSUCmp *Cmp = cast<GSUCmp>(Val: &MI); |
7278 | |
7279 | Register Dst = Cmp->getReg(Idx: 0); |
7280 | LLT DstTy = MRI.getType(Reg: Dst); |
7281 | LLT CmpTy = DstTy.changeElementSize(NewEltSize: 1); |
7282 | |
7283 | CmpInst::Predicate LTPredicate = Cmp->isSigned() |
7284 | ? CmpInst::Predicate::ICMP_SLT |
7285 | : CmpInst::Predicate::ICMP_ULT; |
7286 | CmpInst::Predicate GTPredicate = Cmp->isSigned() |
7287 | ? CmpInst::Predicate::ICMP_SGT |
7288 | : CmpInst::Predicate::ICMP_UGT; |
7289 | |
7290 | auto One = MIRBuilder.buildConstant(Res: DstTy, Val: 1); |
7291 | auto Zero = MIRBuilder.buildConstant(Res: DstTy, Val: 0); |
7292 | auto IsGT = MIRBuilder.buildICmp(Pred: GTPredicate, Res: CmpTy, Op0: Cmp->getLHSReg(), |
7293 | Op1: Cmp->getRHSReg()); |
7294 | auto SelectZeroOrOne = MIRBuilder.buildSelect(Res: DstTy, Tst: IsGT, Op0: One, Op1: Zero); |
7295 | |
7296 | auto MinusOne = MIRBuilder.buildConstant(Res: DstTy, Val: -1); |
7297 | auto IsLT = MIRBuilder.buildICmp(Pred: LTPredicate, Res: CmpTy, Op0: Cmp->getLHSReg(), |
7298 | Op1: Cmp->getRHSReg()); |
7299 | MIRBuilder.buildSelect(Res: Dst, Tst: IsLT, Op0: MinusOne, Op1: SelectZeroOrOne); |
7300 | |
7301 | MI.eraseFromParent(); |
7302 | return Legalized; |
7303 | } |
7304 | |
7305 | LegalizerHelper::LegalizeResult |
7306 | LegalizerHelper::lowerFCopySign(MachineInstr &MI) { |
7307 | auto [Dst, DstTy, Src0, Src0Ty, Src1, Src1Ty] = MI.getFirst3RegLLTs(); |
7308 | const int Src0Size = Src0Ty.getScalarSizeInBits(); |
7309 | const int Src1Size = Src1Ty.getScalarSizeInBits(); |
7310 | |
7311 | auto SignBitMask = MIRBuilder.buildConstant( |
7312 | Res: Src0Ty, Val: APInt::getSignMask(BitWidth: Src0Size)); |
7313 | |
7314 | auto NotSignBitMask = MIRBuilder.buildConstant( |
7315 | Res: Src0Ty, Val: APInt::getLowBitsSet(numBits: Src0Size, loBitsSet: Src0Size - 1)); |
7316 | |
7317 | Register And0 = MIRBuilder.buildAnd(Dst: Src0Ty, Src0, Src1: NotSignBitMask).getReg(Idx: 0); |
7318 | Register And1; |
7319 | if (Src0Ty == Src1Ty) { |
7320 | And1 = MIRBuilder.buildAnd(Dst: Src1Ty, Src0: Src1, Src1: SignBitMask).getReg(Idx: 0); |
7321 | } else if (Src0Size > Src1Size) { |
7322 | auto ShiftAmt = MIRBuilder.buildConstant(Res: Src0Ty, Val: Src0Size - Src1Size); |
7323 | auto Zext = MIRBuilder.buildZExt(Res: Src0Ty, Op: Src1); |
7324 | auto Shift = MIRBuilder.buildShl(Dst: Src0Ty, Src0: Zext, Src1: ShiftAmt); |
7325 | And1 = MIRBuilder.buildAnd(Dst: Src0Ty, Src0: Shift, Src1: SignBitMask).getReg(Idx: 0); |
7326 | } else { |
7327 | auto ShiftAmt = MIRBuilder.buildConstant(Res: Src1Ty, Val: Src1Size - Src0Size); |
7328 | auto Shift = MIRBuilder.buildLShr(Dst: Src1Ty, Src0: Src1, Src1: ShiftAmt); |
7329 | auto Trunc = MIRBuilder.buildTrunc(Res: Src0Ty, Op: Shift); |
7330 | And1 = MIRBuilder.buildAnd(Dst: Src0Ty, Src0: Trunc, Src1: SignBitMask).getReg(Idx: 0); |
7331 | } |
7332 | |
7333 | // Be careful about setting nsz/nnan/ninf on every instruction, since the |
7334 | // constants are a nan and -0.0, but the final result should preserve |
7335 | // everything. |
7336 | unsigned Flags = MI.getFlags(); |
7337 | |
7338 | // We masked the sign bit and the not-sign bit, so these are disjoint. |
7339 | Flags |= MachineInstr::Disjoint; |
7340 | |
7341 | MIRBuilder.buildOr(Dst, Src0: And0, Src1: And1, Flags); |
7342 | |
7343 | MI.eraseFromParent(); |
7344 | return Legalized; |
7345 | } |
7346 | |
7347 | LegalizerHelper::LegalizeResult |
7348 | LegalizerHelper::lowerFMinNumMaxNum(MachineInstr &MI) { |
7349 | unsigned NewOp = MI.getOpcode() == TargetOpcode::G_FMINNUM ? |
7350 | TargetOpcode::G_FMINNUM_IEEE : TargetOpcode::G_FMAXNUM_IEEE; |
7351 | |
7352 | auto [Dst, Src0, Src1] = MI.getFirst3Regs(); |
7353 | LLT Ty = MRI.getType(Reg: Dst); |
7354 | |
7355 | if (!MI.getFlag(Flag: MachineInstr::FmNoNans)) { |
7356 | // Insert canonicalizes if it's possible we need to quiet to get correct |
7357 | // sNaN behavior. |
7358 | |
7359 | // Note this must be done here, and not as an optimization combine in the |
7360 | // absence of a dedicate quiet-snan instruction as we're using an |
7361 | // omni-purpose G_FCANONICALIZE. |
7362 | if (!isKnownNeverSNaN(Val: Src0, MRI)) |
7363 | Src0 = MIRBuilder.buildFCanonicalize(Dst: Ty, Src0, Flags: MI.getFlags()).getReg(Idx: 0); |
7364 | |
7365 | if (!isKnownNeverSNaN(Val: Src1, MRI)) |
7366 | Src1 = MIRBuilder.buildFCanonicalize(Dst: Ty, Src0: Src1, Flags: MI.getFlags()).getReg(Idx: 0); |
7367 | } |
7368 | |
7369 | // If there are no nans, it's safe to simply replace this with the non-IEEE |
7370 | // version. |
7371 | MIRBuilder.buildInstr(Opc: NewOp, DstOps: {Dst}, SrcOps: {Src0, Src1}, Flags: MI.getFlags()); |
7372 | MI.eraseFromParent(); |
7373 | return Legalized; |
7374 | } |
7375 | |
7376 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerFMad(MachineInstr &MI) { |
7377 | // Expand G_FMAD a, b, c -> G_FADD (G_FMUL a, b), c |
7378 | Register DstReg = MI.getOperand(i: 0).getReg(); |
7379 | LLT Ty = MRI.getType(Reg: DstReg); |
7380 | unsigned Flags = MI.getFlags(); |
7381 | |
7382 | auto Mul = MIRBuilder.buildFMul(Dst: Ty, Src0: MI.getOperand(i: 1), Src1: MI.getOperand(i: 2), |
7383 | Flags); |
7384 | MIRBuilder.buildFAdd(Dst: DstReg, Src0: Mul, Src1: MI.getOperand(i: 3), Flags); |
7385 | MI.eraseFromParent(); |
7386 | return Legalized; |
7387 | } |
7388 | |
7389 | LegalizerHelper::LegalizeResult |
7390 | LegalizerHelper::lowerIntrinsicRound(MachineInstr &MI) { |
7391 | auto [DstReg, X] = MI.getFirst2Regs(); |
7392 | const unsigned Flags = MI.getFlags(); |
7393 | const LLT Ty = MRI.getType(Reg: DstReg); |
7394 | const LLT CondTy = Ty.changeElementSize(NewEltSize: 1); |
7395 | |
7396 | // round(x) => |
7397 | // t = trunc(x); |
7398 | // d = fabs(x - t); |
7399 | // o = copysign(d >= 0.5 ? 1.0 : 0.0, x); |
7400 | // return t + o; |
7401 | |
7402 | auto T = MIRBuilder.buildIntrinsicTrunc(Dst: Ty, Src0: X, Flags); |
7403 | |
7404 | auto Diff = MIRBuilder.buildFSub(Dst: Ty, Src0: X, Src1: T, Flags); |
7405 | auto AbsDiff = MIRBuilder.buildFAbs(Dst: Ty, Src0: Diff, Flags); |
7406 | |
7407 | auto Half = MIRBuilder.buildFConstant(Res: Ty, Val: 0.5); |
7408 | auto Cmp = |
7409 | MIRBuilder.buildFCmp(Pred: CmpInst::FCMP_OGE, Res: CondTy, Op0: AbsDiff, Op1: Half, Flags); |
7410 | |
7411 | // Could emit G_UITOFP instead |
7412 | auto One = MIRBuilder.buildFConstant(Res: Ty, Val: 1.0); |
7413 | auto Zero = MIRBuilder.buildFConstant(Res: Ty, Val: 0.0); |
7414 | auto BoolFP = MIRBuilder.buildSelect(Res: Ty, Tst: Cmp, Op0: One, Op1: Zero); |
7415 | auto SignedOffset = MIRBuilder.buildFCopysign(Dst: Ty, Src0: BoolFP, Src1: X); |
7416 | |
7417 | MIRBuilder.buildFAdd(Dst: DstReg, Src0: T, Src1: SignedOffset, Flags); |
7418 | |
7419 | MI.eraseFromParent(); |
7420 | return Legalized; |
7421 | } |
7422 | |
7423 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerFFloor(MachineInstr &MI) { |
7424 | auto [DstReg, SrcReg] = MI.getFirst2Regs(); |
7425 | unsigned Flags = MI.getFlags(); |
7426 | LLT Ty = MRI.getType(Reg: DstReg); |
7427 | const LLT CondTy = Ty.changeElementSize(NewEltSize: 1); |
7428 | |
7429 | // result = trunc(src); |
7430 | // if (src < 0.0 && src != result) |
7431 | // result += -1.0. |
7432 | |
7433 | auto Trunc = MIRBuilder.buildIntrinsicTrunc(Dst: Ty, Src0: SrcReg, Flags); |
7434 | auto Zero = MIRBuilder.buildFConstant(Res: Ty, Val: 0.0); |
7435 | |
7436 | auto Lt0 = MIRBuilder.buildFCmp(Pred: CmpInst::FCMP_OLT, Res: CondTy, |
7437 | Op0: SrcReg, Op1: Zero, Flags); |
7438 | auto NeTrunc = MIRBuilder.buildFCmp(Pred: CmpInst::FCMP_ONE, Res: CondTy, |
7439 | Op0: SrcReg, Op1: Trunc, Flags); |
7440 | auto And = MIRBuilder.buildAnd(Dst: CondTy, Src0: Lt0, Src1: NeTrunc); |
7441 | auto AddVal = MIRBuilder.buildSITOFP(Dst: Ty, Src0: And); |
7442 | |
7443 | MIRBuilder.buildFAdd(Dst: DstReg, Src0: Trunc, Src1: AddVal, Flags); |
7444 | MI.eraseFromParent(); |
7445 | return Legalized; |
7446 | } |
7447 | |
7448 | LegalizerHelper::LegalizeResult |
7449 | LegalizerHelper::lowerMergeValues(MachineInstr &MI) { |
7450 | const unsigned NumOps = MI.getNumOperands(); |
7451 | auto [DstReg, DstTy, Src0Reg, Src0Ty] = MI.getFirst2RegLLTs(); |
7452 | unsigned PartSize = Src0Ty.getSizeInBits(); |
7453 | |
7454 | LLT WideTy = LLT::scalar(SizeInBits: DstTy.getSizeInBits()); |
7455 | Register ResultReg = MIRBuilder.buildZExt(Res: WideTy, Op: Src0Reg).getReg(Idx: 0); |
7456 | |
7457 | for (unsigned I = 2; I != NumOps; ++I) { |
7458 | const unsigned Offset = (I - 1) * PartSize; |
7459 | |
7460 | Register SrcReg = MI.getOperand(i: I).getReg(); |
7461 | auto ZextInput = MIRBuilder.buildZExt(Res: WideTy, Op: SrcReg); |
7462 | |
7463 | Register NextResult = I + 1 == NumOps && WideTy == DstTy ? DstReg : |
7464 | MRI.createGenericVirtualRegister(Ty: WideTy); |
7465 | |
7466 | auto ShiftAmt = MIRBuilder.buildConstant(Res: WideTy, Val: Offset); |
7467 | auto Shl = MIRBuilder.buildShl(Dst: WideTy, Src0: ZextInput, Src1: ShiftAmt); |
7468 | MIRBuilder.buildOr(Dst: NextResult, Src0: ResultReg, Src1: Shl); |
7469 | ResultReg = NextResult; |
7470 | } |
7471 | |
7472 | if (DstTy.isPointer()) { |
7473 | if (MIRBuilder.getDataLayout().isNonIntegralAddressSpace( |
7474 | AddrSpace: DstTy.getAddressSpace())) { |
7475 | LLVM_DEBUG(dbgs() << "Not casting nonintegral address space\n" ); |
7476 | return UnableToLegalize; |
7477 | } |
7478 | |
7479 | MIRBuilder.buildIntToPtr(Dst: DstReg, Src: ResultReg); |
7480 | } |
7481 | |
7482 | MI.eraseFromParent(); |
7483 | return Legalized; |
7484 | } |
7485 | |
7486 | LegalizerHelper::LegalizeResult |
7487 | LegalizerHelper::lowerUnmergeValues(MachineInstr &MI) { |
7488 | const unsigned NumDst = MI.getNumOperands() - 1; |
7489 | Register SrcReg = MI.getOperand(i: NumDst).getReg(); |
7490 | Register Dst0Reg = MI.getOperand(i: 0).getReg(); |
7491 | LLT DstTy = MRI.getType(Reg: Dst0Reg); |
7492 | if (DstTy.isPointer()) |
7493 | return UnableToLegalize; // TODO |
7494 | |
7495 | SrcReg = coerceToScalar(Val: SrcReg); |
7496 | if (!SrcReg) |
7497 | return UnableToLegalize; |
7498 | |
7499 | // Expand scalarizing unmerge as bitcast to integer and shift. |
7500 | LLT IntTy = MRI.getType(Reg: SrcReg); |
7501 | |
7502 | MIRBuilder.buildTrunc(Res: Dst0Reg, Op: SrcReg); |
7503 | |
7504 | const unsigned DstSize = DstTy.getSizeInBits(); |
7505 | unsigned Offset = DstSize; |
7506 | for (unsigned I = 1; I != NumDst; ++I, Offset += DstSize) { |
7507 | auto ShiftAmt = MIRBuilder.buildConstant(Res: IntTy, Val: Offset); |
7508 | auto Shift = MIRBuilder.buildLShr(Dst: IntTy, Src0: SrcReg, Src1: ShiftAmt); |
7509 | MIRBuilder.buildTrunc(Res: MI.getOperand(i: I), Op: Shift); |
7510 | } |
7511 | |
7512 | MI.eraseFromParent(); |
7513 | return Legalized; |
7514 | } |
7515 | |
7516 | /// Lower a vector extract or insert by writing the vector to a stack temporary |
7517 | /// and reloading the element or vector. |
7518 | /// |
7519 | /// %dst = G_EXTRACT_VECTOR_ELT %vec, %idx |
7520 | /// => |
7521 | /// %stack_temp = G_FRAME_INDEX |
7522 | /// G_STORE %vec, %stack_temp |
7523 | /// %idx = clamp(%idx, %vec.getNumElements()) |
7524 | /// %element_ptr = G_PTR_ADD %stack_temp, %idx |
7525 | /// %dst = G_LOAD %element_ptr |
7526 | LegalizerHelper::LegalizeResult |
7527 | LegalizerHelper::(MachineInstr &MI) { |
7528 | Register DstReg = MI.getOperand(i: 0).getReg(); |
7529 | Register SrcVec = MI.getOperand(i: 1).getReg(); |
7530 | Register InsertVal; |
7531 | if (MI.getOpcode() == TargetOpcode::G_INSERT_VECTOR_ELT) |
7532 | InsertVal = MI.getOperand(i: 2).getReg(); |
7533 | |
7534 | Register Idx = MI.getOperand(i: MI.getNumOperands() - 1).getReg(); |
7535 | |
7536 | LLT VecTy = MRI.getType(Reg: SrcVec); |
7537 | LLT EltTy = VecTy.getElementType(); |
7538 | unsigned NumElts = VecTy.getNumElements(); |
7539 | |
7540 | int64_t IdxVal; |
7541 | if (mi_match(R: Idx, MRI, P: m_ICst(Cst&: IdxVal)) && IdxVal <= NumElts) { |
7542 | SmallVector<Register, 8> SrcRegs; |
7543 | extractParts(Reg: SrcVec, Ty: EltTy, NumParts: NumElts, VRegs&: SrcRegs, MIRBuilder, MRI); |
7544 | |
7545 | if (InsertVal) { |
7546 | SrcRegs[IdxVal] = MI.getOperand(i: 2).getReg(); |
7547 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: SrcRegs); |
7548 | } else { |
7549 | MIRBuilder.buildCopy(Res: DstReg, Op: SrcRegs[IdxVal]); |
7550 | } |
7551 | |
7552 | MI.eraseFromParent(); |
7553 | return Legalized; |
7554 | } |
7555 | |
7556 | if (!EltTy.isByteSized()) { // Not implemented. |
7557 | LLVM_DEBUG(dbgs() << "Can't handle non-byte element vectors yet\n" ); |
7558 | return UnableToLegalize; |
7559 | } |
7560 | |
7561 | unsigned EltBytes = EltTy.getSizeInBytes(); |
7562 | Align VecAlign = getStackTemporaryAlignment(Ty: VecTy); |
7563 | Align EltAlign; |
7564 | |
7565 | MachinePointerInfo PtrInfo; |
7566 | auto StackTemp = createStackTemporary( |
7567 | Bytes: TypeSize::getFixed(ExactSize: VecTy.getSizeInBytes()), Alignment: VecAlign, PtrInfo); |
7568 | MIRBuilder.buildStore(Val: SrcVec, Addr: StackTemp, PtrInfo, Alignment: VecAlign); |
7569 | |
7570 | // Get the pointer to the element, and be sure not to hit undefined behavior |
7571 | // if the index is out of bounds. |
7572 | Register EltPtr = getVectorElementPointer(VecPtr: StackTemp.getReg(Idx: 0), VecTy, Index: Idx); |
7573 | |
7574 | if (mi_match(R: Idx, MRI, P: m_ICst(Cst&: IdxVal))) { |
7575 | int64_t Offset = IdxVal * EltBytes; |
7576 | PtrInfo = PtrInfo.getWithOffset(O: Offset); |
7577 | EltAlign = commonAlignment(A: VecAlign, Offset); |
7578 | } else { |
7579 | // We lose information with a variable offset. |
7580 | EltAlign = getStackTemporaryAlignment(Ty: EltTy); |
7581 | PtrInfo = MachinePointerInfo(MRI.getType(Reg: EltPtr).getAddressSpace()); |
7582 | } |
7583 | |
7584 | if (InsertVal) { |
7585 | // Write the inserted element |
7586 | MIRBuilder.buildStore(Val: InsertVal, Addr: EltPtr, PtrInfo, Alignment: EltAlign); |
7587 | |
7588 | // Reload the whole vector. |
7589 | MIRBuilder.buildLoad(Res: DstReg, Addr: StackTemp, PtrInfo, Alignment: VecAlign); |
7590 | } else { |
7591 | MIRBuilder.buildLoad(Res: DstReg, Addr: EltPtr, PtrInfo, Alignment: EltAlign); |
7592 | } |
7593 | |
7594 | MI.eraseFromParent(); |
7595 | return Legalized; |
7596 | } |
7597 | |
7598 | LegalizerHelper::LegalizeResult |
7599 | LegalizerHelper::lowerShuffleVector(MachineInstr &MI) { |
7600 | auto [DstReg, DstTy, Src0Reg, Src0Ty, Src1Reg, Src1Ty] = |
7601 | MI.getFirst3RegLLTs(); |
7602 | LLT IdxTy = LLT::scalar(SizeInBits: 32); |
7603 | |
7604 | ArrayRef<int> Mask = MI.getOperand(i: 3).getShuffleMask(); |
7605 | Register Undef; |
7606 | SmallVector<Register, 32> BuildVec; |
7607 | LLT EltTy = DstTy.getScalarType(); |
7608 | |
7609 | for (int Idx : Mask) { |
7610 | if (Idx < 0) { |
7611 | if (!Undef.isValid()) |
7612 | Undef = MIRBuilder.buildUndef(Res: EltTy).getReg(Idx: 0); |
7613 | BuildVec.push_back(Elt: Undef); |
7614 | continue; |
7615 | } |
7616 | |
7617 | if (Src0Ty.isScalar()) { |
7618 | BuildVec.push_back(Elt: Idx == 0 ? Src0Reg : Src1Reg); |
7619 | } else { |
7620 | int NumElts = Src0Ty.getNumElements(); |
7621 | Register SrcVec = Idx < NumElts ? Src0Reg : Src1Reg; |
7622 | int = Idx < NumElts ? Idx : Idx - NumElts; |
7623 | auto IdxK = MIRBuilder.buildConstant(Res: IdxTy, Val: ExtractIdx); |
7624 | auto = MIRBuilder.buildExtractVectorElement(Res: EltTy, Val: SrcVec, Idx: IdxK); |
7625 | BuildVec.push_back(Elt: Extract.getReg(Idx: 0)); |
7626 | } |
7627 | } |
7628 | |
7629 | if (DstTy.isScalar()) |
7630 | MIRBuilder.buildCopy(Res: DstReg, Op: BuildVec[0]); |
7631 | else |
7632 | MIRBuilder.buildBuildVector(Res: DstReg, Ops: BuildVec); |
7633 | MI.eraseFromParent(); |
7634 | return Legalized; |
7635 | } |
7636 | |
7637 | LegalizerHelper::LegalizeResult |
7638 | LegalizerHelper::lowerVECTOR_COMPRESS(llvm::MachineInstr &MI) { |
7639 | auto [Dst, DstTy, Vec, VecTy, Mask, MaskTy, Passthru, PassthruTy] = |
7640 | MI.getFirst4RegLLTs(); |
7641 | |
7642 | if (VecTy.isScalableVector()) |
7643 | report_fatal_error(reason: "Cannot expand masked_compress for scalable vectors." ); |
7644 | |
7645 | Align VecAlign = getStackTemporaryAlignment(Ty: VecTy); |
7646 | MachinePointerInfo PtrInfo; |
7647 | Register StackPtr = |
7648 | createStackTemporary(Bytes: TypeSize::getFixed(ExactSize: VecTy.getSizeInBytes()), Alignment: VecAlign, |
7649 | PtrInfo) |
7650 | .getReg(Idx: 0); |
7651 | MachinePointerInfo ValPtrInfo = |
7652 | MachinePointerInfo::getUnknownStack(MF&: *MI.getMF()); |
7653 | |
7654 | LLT IdxTy = LLT::scalar(SizeInBits: 32); |
7655 | LLT ValTy = VecTy.getElementType(); |
7656 | Align ValAlign = getStackTemporaryAlignment(Ty: ValTy); |
7657 | |
7658 | auto OutPos = MIRBuilder.buildConstant(Res: IdxTy, Val: 0); |
7659 | |
7660 | bool HasPassthru = |
7661 | MRI.getVRegDef(Reg: Passthru)->getOpcode() != TargetOpcode::G_IMPLICIT_DEF; |
7662 | |
7663 | if (HasPassthru) |
7664 | MIRBuilder.buildStore(Val: Passthru, Addr: StackPtr, PtrInfo, Alignment: VecAlign); |
7665 | |
7666 | Register LastWriteVal; |
7667 | std::optional<APInt> PassthruSplatVal = |
7668 | isConstantOrConstantSplatVector(MI&: *MRI.getVRegDef(Reg: Passthru), MRI); |
7669 | |
7670 | if (PassthruSplatVal.has_value()) { |
7671 | LastWriteVal = |
7672 | MIRBuilder.buildConstant(Res: ValTy, Val: PassthruSplatVal.value()).getReg(Idx: 0); |
7673 | } else if (HasPassthru) { |
7674 | auto Popcount = MIRBuilder.buildZExt(Res: MaskTy.changeElementSize(NewEltSize: 32), Op: Mask); |
7675 | Popcount = MIRBuilder.buildInstr(Opc: TargetOpcode::G_VECREDUCE_ADD, |
7676 | DstOps: {LLT::scalar(SizeInBits: 32)}, SrcOps: {Popcount}); |
7677 | |
7678 | Register LastElmtPtr = |
7679 | getVectorElementPointer(VecPtr: StackPtr, VecTy, Index: Popcount.getReg(Idx: 0)); |
7680 | LastWriteVal = |
7681 | MIRBuilder.buildLoad(Res: ValTy, Addr: LastElmtPtr, PtrInfo: ValPtrInfo, Alignment: ValAlign) |
7682 | .getReg(Idx: 0); |
7683 | } |
7684 | |
7685 | unsigned NumElmts = VecTy.getNumElements(); |
7686 | for (unsigned I = 0; I < NumElmts; ++I) { |
7687 | auto Idx = MIRBuilder.buildConstant(Res: IdxTy, Val: I); |
7688 | auto Val = MIRBuilder.buildExtractVectorElement(Res: ValTy, Val: Vec, Idx); |
7689 | Register ElmtPtr = |
7690 | getVectorElementPointer(VecPtr: StackPtr, VecTy, Index: OutPos.getReg(Idx: 0)); |
7691 | MIRBuilder.buildStore(Val, Addr: ElmtPtr, PtrInfo: ValPtrInfo, Alignment: ValAlign); |
7692 | |
7693 | LLT MaskITy = MaskTy.getElementType(); |
7694 | auto MaskI = MIRBuilder.buildExtractVectorElement(Res: MaskITy, Val: Mask, Idx); |
7695 | if (MaskITy.getSizeInBits() > 1) |
7696 | MaskI = MIRBuilder.buildTrunc(Res: LLT::scalar(SizeInBits: 1), Op: MaskI); |
7697 | |
7698 | MaskI = MIRBuilder.buildZExt(Res: IdxTy, Op: MaskI); |
7699 | OutPos = MIRBuilder.buildAdd(Dst: IdxTy, Src0: OutPos, Src1: MaskI); |
7700 | |
7701 | if (HasPassthru && I == NumElmts - 1) { |
7702 | auto EndOfVector = |
7703 | MIRBuilder.buildConstant(Res: IdxTy, Val: VecTy.getNumElements() - 1); |
7704 | auto AllLanesSelected = MIRBuilder.buildICmp( |
7705 | Pred: CmpInst::ICMP_UGT, Res: LLT::scalar(SizeInBits: 1), Op0: OutPos, Op1: EndOfVector); |
7706 | OutPos = MIRBuilder.buildInstr(Opc: TargetOpcode::G_UMIN, DstOps: {IdxTy}, |
7707 | SrcOps: {OutPos, EndOfVector}); |
7708 | ElmtPtr = getVectorElementPointer(VecPtr: StackPtr, VecTy, Index: OutPos.getReg(Idx: 0)); |
7709 | |
7710 | LastWriteVal = |
7711 | MIRBuilder.buildSelect(Res: ValTy, Tst: AllLanesSelected, Op0: Val, Op1: LastWriteVal) |
7712 | .getReg(Idx: 0); |
7713 | MIRBuilder.buildStore(Val: LastWriteVal, Addr: ElmtPtr, PtrInfo: ValPtrInfo, Alignment: ValAlign); |
7714 | } |
7715 | } |
7716 | |
7717 | // TODO: Use StackPtr's FrameIndex alignment. |
7718 | MIRBuilder.buildLoad(Res: Dst, Addr: StackPtr, PtrInfo, Alignment: VecAlign); |
7719 | |
7720 | MI.eraseFromParent(); |
7721 | return Legalized; |
7722 | } |
7723 | |
7724 | Register LegalizerHelper::getDynStackAllocTargetPtr(Register SPReg, |
7725 | Register AllocSize, |
7726 | Align Alignment, |
7727 | LLT PtrTy) { |
7728 | LLT IntPtrTy = LLT::scalar(SizeInBits: PtrTy.getSizeInBits()); |
7729 | |
7730 | auto SPTmp = MIRBuilder.buildCopy(Res: PtrTy, Op: SPReg); |
7731 | SPTmp = MIRBuilder.buildCast(Dst: IntPtrTy, Src: SPTmp); |
7732 | |
7733 | // Subtract the final alloc from the SP. We use G_PTRTOINT here so we don't |
7734 | // have to generate an extra instruction to negate the alloc and then use |
7735 | // G_PTR_ADD to add the negative offset. |
7736 | auto Alloc = MIRBuilder.buildSub(Dst: IntPtrTy, Src0: SPTmp, Src1: AllocSize); |
7737 | if (Alignment > Align(1)) { |
7738 | APInt AlignMask(IntPtrTy.getSizeInBits(), Alignment.value(), true); |
7739 | AlignMask.negate(); |
7740 | auto AlignCst = MIRBuilder.buildConstant(Res: IntPtrTy, Val: AlignMask); |
7741 | Alloc = MIRBuilder.buildAnd(Dst: IntPtrTy, Src0: Alloc, Src1: AlignCst); |
7742 | } |
7743 | |
7744 | return MIRBuilder.buildCast(Dst: PtrTy, Src: Alloc).getReg(Idx: 0); |
7745 | } |
7746 | |
7747 | LegalizerHelper::LegalizeResult |
7748 | LegalizerHelper::lowerDynStackAlloc(MachineInstr &MI) { |
7749 | const auto &MF = *MI.getMF(); |
7750 | const auto &TFI = *MF.getSubtarget().getFrameLowering(); |
7751 | if (TFI.getStackGrowthDirection() == TargetFrameLowering::StackGrowsUp) |
7752 | return UnableToLegalize; |
7753 | |
7754 | Register Dst = MI.getOperand(i: 0).getReg(); |
7755 | Register AllocSize = MI.getOperand(i: 1).getReg(); |
7756 | Align Alignment = assumeAligned(Value: MI.getOperand(i: 2).getImm()); |
7757 | |
7758 | LLT PtrTy = MRI.getType(Reg: Dst); |
7759 | Register SPReg = TLI.getStackPointerRegisterToSaveRestore(); |
7760 | Register SPTmp = |
7761 | getDynStackAllocTargetPtr(SPReg, AllocSize, Alignment, PtrTy); |
7762 | |
7763 | MIRBuilder.buildCopy(Res: SPReg, Op: SPTmp); |
7764 | MIRBuilder.buildCopy(Res: Dst, Op: SPTmp); |
7765 | |
7766 | MI.eraseFromParent(); |
7767 | return Legalized; |
7768 | } |
7769 | |
7770 | LegalizerHelper::LegalizeResult |
7771 | LegalizerHelper::lowerStackSave(MachineInstr &MI) { |
7772 | Register StackPtr = TLI.getStackPointerRegisterToSaveRestore(); |
7773 | if (!StackPtr) |
7774 | return UnableToLegalize; |
7775 | |
7776 | MIRBuilder.buildCopy(Res: MI.getOperand(i: 0), Op: StackPtr); |
7777 | MI.eraseFromParent(); |
7778 | return Legalized; |
7779 | } |
7780 | |
7781 | LegalizerHelper::LegalizeResult |
7782 | LegalizerHelper::lowerStackRestore(MachineInstr &MI) { |
7783 | Register StackPtr = TLI.getStackPointerRegisterToSaveRestore(); |
7784 | if (!StackPtr) |
7785 | return UnableToLegalize; |
7786 | |
7787 | MIRBuilder.buildCopy(Res: StackPtr, Op: MI.getOperand(i: 0)); |
7788 | MI.eraseFromParent(); |
7789 | return Legalized; |
7790 | } |
7791 | |
7792 | LegalizerHelper::LegalizeResult |
7793 | LegalizerHelper::(MachineInstr &MI) { |
7794 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
7795 | unsigned Offset = MI.getOperand(i: 2).getImm(); |
7796 | |
7797 | // Extract sub-vector or one element |
7798 | if (SrcTy.isVector()) { |
7799 | unsigned SrcEltSize = SrcTy.getElementType().getSizeInBits(); |
7800 | unsigned DstSize = DstTy.getSizeInBits(); |
7801 | |
7802 | if ((Offset % SrcEltSize == 0) && (DstSize % SrcEltSize == 0) && |
7803 | (Offset + DstSize <= SrcTy.getSizeInBits())) { |
7804 | // Unmerge and allow access to each Src element for the artifact combiner. |
7805 | auto Unmerge = MIRBuilder.buildUnmerge(Res: SrcTy.getElementType(), Op: SrcReg); |
7806 | |
7807 | // Take element(s) we need to extract and copy it (merge them). |
7808 | SmallVector<Register, 8> SubVectorElts; |
7809 | for (unsigned Idx = Offset / SrcEltSize; |
7810 | Idx < (Offset + DstSize) / SrcEltSize; ++Idx) { |
7811 | SubVectorElts.push_back(Elt: Unmerge.getReg(Idx)); |
7812 | } |
7813 | if (SubVectorElts.size() == 1) |
7814 | MIRBuilder.buildCopy(Res: DstReg, Op: SubVectorElts[0]); |
7815 | else |
7816 | MIRBuilder.buildMergeLikeInstr(Res: DstReg, Ops: SubVectorElts); |
7817 | |
7818 | MI.eraseFromParent(); |
7819 | return Legalized; |
7820 | } |
7821 | } |
7822 | |
7823 | if (DstTy.isScalar() && |
7824 | (SrcTy.isScalar() || |
7825 | (SrcTy.isVector() && DstTy == SrcTy.getElementType()))) { |
7826 | LLT SrcIntTy = SrcTy; |
7827 | if (!SrcTy.isScalar()) { |
7828 | SrcIntTy = LLT::scalar(SizeInBits: SrcTy.getSizeInBits()); |
7829 | SrcReg = MIRBuilder.buildBitcast(Dst: SrcIntTy, Src: SrcReg).getReg(Idx: 0); |
7830 | } |
7831 | |
7832 | if (Offset == 0) |
7833 | MIRBuilder.buildTrunc(Res: DstReg, Op: SrcReg); |
7834 | else { |
7835 | auto ShiftAmt = MIRBuilder.buildConstant(Res: SrcIntTy, Val: Offset); |
7836 | auto Shr = MIRBuilder.buildLShr(Dst: SrcIntTy, Src0: SrcReg, Src1: ShiftAmt); |
7837 | MIRBuilder.buildTrunc(Res: DstReg, Op: Shr); |
7838 | } |
7839 | |
7840 | MI.eraseFromParent(); |
7841 | return Legalized; |
7842 | } |
7843 | |
7844 | return UnableToLegalize; |
7845 | } |
7846 | |
7847 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerInsert(MachineInstr &MI) { |
7848 | auto [Dst, Src, InsertSrc] = MI.getFirst3Regs(); |
7849 | uint64_t Offset = MI.getOperand(i: 3).getImm(); |
7850 | |
7851 | LLT DstTy = MRI.getType(Reg: Src); |
7852 | LLT InsertTy = MRI.getType(Reg: InsertSrc); |
7853 | |
7854 | // Insert sub-vector or one element |
7855 | if (DstTy.isVector() && !InsertTy.isPointer()) { |
7856 | LLT EltTy = DstTy.getElementType(); |
7857 | unsigned EltSize = EltTy.getSizeInBits(); |
7858 | unsigned InsertSize = InsertTy.getSizeInBits(); |
7859 | |
7860 | if ((Offset % EltSize == 0) && (InsertSize % EltSize == 0) && |
7861 | (Offset + InsertSize <= DstTy.getSizeInBits())) { |
7862 | auto UnmergeSrc = MIRBuilder.buildUnmerge(Res: EltTy, Op: Src); |
7863 | SmallVector<Register, 8> DstElts; |
7864 | unsigned Idx = 0; |
7865 | // Elements from Src before insert start Offset |
7866 | for (; Idx < Offset / EltSize; ++Idx) { |
7867 | DstElts.push_back(Elt: UnmergeSrc.getReg(Idx)); |
7868 | } |
7869 | |
7870 | // Replace elements in Src with elements from InsertSrc |
7871 | if (InsertTy.getSizeInBits() > EltSize) { |
7872 | auto UnmergeInsertSrc = MIRBuilder.buildUnmerge(Res: EltTy, Op: InsertSrc); |
7873 | for (unsigned i = 0; Idx < (Offset + InsertSize) / EltSize; |
7874 | ++Idx, ++i) { |
7875 | DstElts.push_back(Elt: UnmergeInsertSrc.getReg(Idx: i)); |
7876 | } |
7877 | } else { |
7878 | DstElts.push_back(Elt: InsertSrc); |
7879 | ++Idx; |
7880 | } |
7881 | |
7882 | // Remaining elements from Src after insert |
7883 | for (; Idx < DstTy.getNumElements(); ++Idx) { |
7884 | DstElts.push_back(Elt: UnmergeSrc.getReg(Idx)); |
7885 | } |
7886 | |
7887 | MIRBuilder.buildMergeLikeInstr(Res: Dst, Ops: DstElts); |
7888 | MI.eraseFromParent(); |
7889 | return Legalized; |
7890 | } |
7891 | } |
7892 | |
7893 | if (InsertTy.isVector() || |
7894 | (DstTy.isVector() && DstTy.getElementType() != InsertTy)) |
7895 | return UnableToLegalize; |
7896 | |
7897 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
7898 | if ((DstTy.isPointer() && |
7899 | DL.isNonIntegralAddressSpace(AddrSpace: DstTy.getAddressSpace())) || |
7900 | (InsertTy.isPointer() && |
7901 | DL.isNonIntegralAddressSpace(AddrSpace: InsertTy.getAddressSpace()))) { |
7902 | LLVM_DEBUG(dbgs() << "Not casting non-integral address space integer\n" ); |
7903 | return UnableToLegalize; |
7904 | } |
7905 | |
7906 | LLT IntDstTy = DstTy; |
7907 | |
7908 | if (!DstTy.isScalar()) { |
7909 | IntDstTy = LLT::scalar(SizeInBits: DstTy.getSizeInBits()); |
7910 | Src = MIRBuilder.buildCast(Dst: IntDstTy, Src).getReg(Idx: 0); |
7911 | } |
7912 | |
7913 | if (!InsertTy.isScalar()) { |
7914 | const LLT IntInsertTy = LLT::scalar(SizeInBits: InsertTy.getSizeInBits()); |
7915 | InsertSrc = MIRBuilder.buildPtrToInt(Dst: IntInsertTy, Src: InsertSrc).getReg(Idx: 0); |
7916 | } |
7917 | |
7918 | Register ExtInsSrc = MIRBuilder.buildZExt(Res: IntDstTy, Op: InsertSrc).getReg(Idx: 0); |
7919 | if (Offset != 0) { |
7920 | auto ShiftAmt = MIRBuilder.buildConstant(Res: IntDstTy, Val: Offset); |
7921 | ExtInsSrc = MIRBuilder.buildShl(Dst: IntDstTy, Src0: ExtInsSrc, Src1: ShiftAmt).getReg(Idx: 0); |
7922 | } |
7923 | |
7924 | APInt MaskVal = APInt::getBitsSetWithWrap( |
7925 | numBits: DstTy.getSizeInBits(), loBit: Offset + InsertTy.getSizeInBits(), hiBit: Offset); |
7926 | |
7927 | auto Mask = MIRBuilder.buildConstant(Res: IntDstTy, Val: MaskVal); |
7928 | auto MaskedSrc = MIRBuilder.buildAnd(Dst: IntDstTy, Src0: Src, Src1: Mask); |
7929 | auto Or = MIRBuilder.buildOr(Dst: IntDstTy, Src0: MaskedSrc, Src1: ExtInsSrc); |
7930 | |
7931 | MIRBuilder.buildCast(Dst, Src: Or); |
7932 | MI.eraseFromParent(); |
7933 | return Legalized; |
7934 | } |
7935 | |
7936 | LegalizerHelper::LegalizeResult |
7937 | LegalizerHelper::lowerSADDO_SSUBO(MachineInstr &MI) { |
7938 | auto [Dst0, Dst0Ty, Dst1, Dst1Ty, LHS, LHSTy, RHS, RHSTy] = |
7939 | MI.getFirst4RegLLTs(); |
7940 | const bool IsAdd = MI.getOpcode() == TargetOpcode::G_SADDO; |
7941 | |
7942 | LLT Ty = Dst0Ty; |
7943 | LLT BoolTy = Dst1Ty; |
7944 | |
7945 | Register NewDst0 = MRI.cloneVirtualRegister(VReg: Dst0); |
7946 | |
7947 | if (IsAdd) |
7948 | MIRBuilder.buildAdd(Dst: NewDst0, Src0: LHS, Src1: RHS); |
7949 | else |
7950 | MIRBuilder.buildSub(Dst: NewDst0, Src0: LHS, Src1: RHS); |
7951 | |
7952 | // TODO: If SADDSAT/SSUBSAT is legal, compare results to detect overflow. |
7953 | |
7954 | auto Zero = MIRBuilder.buildConstant(Res: Ty, Val: 0); |
7955 | |
7956 | // For an addition, the result should be less than one of the operands (LHS) |
7957 | // if and only if the other operand (RHS) is negative, otherwise there will |
7958 | // be overflow. |
7959 | // For a subtraction, the result should be less than one of the operands |
7960 | // (LHS) if and only if the other operand (RHS) is (non-zero) positive, |
7961 | // otherwise there will be overflow. |
7962 | auto ResultLowerThanLHS = |
7963 | MIRBuilder.buildICmp(Pred: CmpInst::ICMP_SLT, Res: BoolTy, Op0: NewDst0, Op1: LHS); |
7964 | auto ConditionRHS = MIRBuilder.buildICmp( |
7965 | Pred: IsAdd ? CmpInst::ICMP_SLT : CmpInst::ICMP_SGT, Res: BoolTy, Op0: RHS, Op1: Zero); |
7966 | |
7967 | MIRBuilder.buildXor(Dst: Dst1, Src0: ConditionRHS, Src1: ResultLowerThanLHS); |
7968 | |
7969 | MIRBuilder.buildCopy(Res: Dst0, Op: NewDst0); |
7970 | MI.eraseFromParent(); |
7971 | |
7972 | return Legalized; |
7973 | } |
7974 | |
7975 | LegalizerHelper::LegalizeResult |
7976 | LegalizerHelper::lowerAddSubSatToMinMax(MachineInstr &MI) { |
7977 | auto [Res, LHS, RHS] = MI.getFirst3Regs(); |
7978 | LLT Ty = MRI.getType(Reg: Res); |
7979 | bool IsSigned; |
7980 | bool IsAdd; |
7981 | unsigned BaseOp; |
7982 | switch (MI.getOpcode()) { |
7983 | default: |
7984 | llvm_unreachable("unexpected addsat/subsat opcode" ); |
7985 | case TargetOpcode::G_UADDSAT: |
7986 | IsSigned = false; |
7987 | IsAdd = true; |
7988 | BaseOp = TargetOpcode::G_ADD; |
7989 | break; |
7990 | case TargetOpcode::G_SADDSAT: |
7991 | IsSigned = true; |
7992 | IsAdd = true; |
7993 | BaseOp = TargetOpcode::G_ADD; |
7994 | break; |
7995 | case TargetOpcode::G_USUBSAT: |
7996 | IsSigned = false; |
7997 | IsAdd = false; |
7998 | BaseOp = TargetOpcode::G_SUB; |
7999 | break; |
8000 | case TargetOpcode::G_SSUBSAT: |
8001 | IsSigned = true; |
8002 | IsAdd = false; |
8003 | BaseOp = TargetOpcode::G_SUB; |
8004 | break; |
8005 | } |
8006 | |
8007 | if (IsSigned) { |
8008 | // sadd.sat(a, b) -> |
8009 | // hi = 0x7fffffff - smax(a, 0) |
8010 | // lo = 0x80000000 - smin(a, 0) |
8011 | // a + smin(smax(lo, b), hi) |
8012 | // ssub.sat(a, b) -> |
8013 | // lo = smax(a, -1) - 0x7fffffff |
8014 | // hi = smin(a, -1) - 0x80000000 |
8015 | // a - smin(smax(lo, b), hi) |
8016 | // TODO: AMDGPU can use a "median of 3" instruction here: |
8017 | // a +/- med3(lo, b, hi) |
8018 | uint64_t NumBits = Ty.getScalarSizeInBits(); |
8019 | auto MaxVal = |
8020 | MIRBuilder.buildConstant(Res: Ty, Val: APInt::getSignedMaxValue(numBits: NumBits)); |
8021 | auto MinVal = |
8022 | MIRBuilder.buildConstant(Res: Ty, Val: APInt::getSignedMinValue(numBits: NumBits)); |
8023 | MachineInstrBuilder Hi, Lo; |
8024 | if (IsAdd) { |
8025 | auto Zero = MIRBuilder.buildConstant(Res: Ty, Val: 0); |
8026 | Hi = MIRBuilder.buildSub(Dst: Ty, Src0: MaxVal, Src1: MIRBuilder.buildSMax(Dst: Ty, Src0: LHS, Src1: Zero)); |
8027 | Lo = MIRBuilder.buildSub(Dst: Ty, Src0: MinVal, Src1: MIRBuilder.buildSMin(Dst: Ty, Src0: LHS, Src1: Zero)); |
8028 | } else { |
8029 | auto NegOne = MIRBuilder.buildConstant(Res: Ty, Val: -1); |
8030 | Lo = MIRBuilder.buildSub(Dst: Ty, Src0: MIRBuilder.buildSMax(Dst: Ty, Src0: LHS, Src1: NegOne), |
8031 | Src1: MaxVal); |
8032 | Hi = MIRBuilder.buildSub(Dst: Ty, Src0: MIRBuilder.buildSMin(Dst: Ty, Src0: LHS, Src1: NegOne), |
8033 | Src1: MinVal); |
8034 | } |
8035 | auto RHSClamped = |
8036 | MIRBuilder.buildSMin(Dst: Ty, Src0: MIRBuilder.buildSMax(Dst: Ty, Src0: Lo, Src1: RHS), Src1: Hi); |
8037 | MIRBuilder.buildInstr(Opc: BaseOp, DstOps: {Res}, SrcOps: {LHS, RHSClamped}); |
8038 | } else { |
8039 | // uadd.sat(a, b) -> a + umin(~a, b) |
8040 | // usub.sat(a, b) -> a - umin(a, b) |
8041 | Register Not = IsAdd ? MIRBuilder.buildNot(Dst: Ty, Src0: LHS).getReg(Idx: 0) : LHS; |
8042 | auto Min = MIRBuilder.buildUMin(Dst: Ty, Src0: Not, Src1: RHS); |
8043 | MIRBuilder.buildInstr(Opc: BaseOp, DstOps: {Res}, SrcOps: {LHS, Min}); |
8044 | } |
8045 | |
8046 | MI.eraseFromParent(); |
8047 | return Legalized; |
8048 | } |
8049 | |
8050 | LegalizerHelper::LegalizeResult |
8051 | LegalizerHelper::lowerAddSubSatToAddoSubo(MachineInstr &MI) { |
8052 | auto [Res, LHS, RHS] = MI.getFirst3Regs(); |
8053 | LLT Ty = MRI.getType(Reg: Res); |
8054 | LLT BoolTy = Ty.changeElementSize(NewEltSize: 1); |
8055 | bool IsSigned; |
8056 | bool IsAdd; |
8057 | unsigned OverflowOp; |
8058 | switch (MI.getOpcode()) { |
8059 | default: |
8060 | llvm_unreachable("unexpected addsat/subsat opcode" ); |
8061 | case TargetOpcode::G_UADDSAT: |
8062 | IsSigned = false; |
8063 | IsAdd = true; |
8064 | OverflowOp = TargetOpcode::G_UADDO; |
8065 | break; |
8066 | case TargetOpcode::G_SADDSAT: |
8067 | IsSigned = true; |
8068 | IsAdd = true; |
8069 | OverflowOp = TargetOpcode::G_SADDO; |
8070 | break; |
8071 | case TargetOpcode::G_USUBSAT: |
8072 | IsSigned = false; |
8073 | IsAdd = false; |
8074 | OverflowOp = TargetOpcode::G_USUBO; |
8075 | break; |
8076 | case TargetOpcode::G_SSUBSAT: |
8077 | IsSigned = true; |
8078 | IsAdd = false; |
8079 | OverflowOp = TargetOpcode::G_SSUBO; |
8080 | break; |
8081 | } |
8082 | |
8083 | auto OverflowRes = |
8084 | MIRBuilder.buildInstr(Opc: OverflowOp, DstOps: {Ty, BoolTy}, SrcOps: {LHS, RHS}); |
8085 | Register Tmp = OverflowRes.getReg(Idx: 0); |
8086 | Register Ov = OverflowRes.getReg(Idx: 1); |
8087 | MachineInstrBuilder Clamp; |
8088 | if (IsSigned) { |
8089 | // sadd.sat(a, b) -> |
8090 | // {tmp, ov} = saddo(a, b) |
8091 | // ov ? (tmp >>s 31) + 0x80000000 : r |
8092 | // ssub.sat(a, b) -> |
8093 | // {tmp, ov} = ssubo(a, b) |
8094 | // ov ? (tmp >>s 31) + 0x80000000 : r |
8095 | uint64_t NumBits = Ty.getScalarSizeInBits(); |
8096 | auto ShiftAmount = MIRBuilder.buildConstant(Res: Ty, Val: NumBits - 1); |
8097 | auto Sign = MIRBuilder.buildAShr(Dst: Ty, Src0: Tmp, Src1: ShiftAmount); |
8098 | auto MinVal = |
8099 | MIRBuilder.buildConstant(Res: Ty, Val: APInt::getSignedMinValue(numBits: NumBits)); |
8100 | Clamp = MIRBuilder.buildAdd(Dst: Ty, Src0: Sign, Src1: MinVal); |
8101 | } else { |
8102 | // uadd.sat(a, b) -> |
8103 | // {tmp, ov} = uaddo(a, b) |
8104 | // ov ? 0xffffffff : tmp |
8105 | // usub.sat(a, b) -> |
8106 | // {tmp, ov} = usubo(a, b) |
8107 | // ov ? 0 : tmp |
8108 | Clamp = MIRBuilder.buildConstant(Res: Ty, Val: IsAdd ? -1 : 0); |
8109 | } |
8110 | MIRBuilder.buildSelect(Res, Tst: Ov, Op0: Clamp, Op1: Tmp); |
8111 | |
8112 | MI.eraseFromParent(); |
8113 | return Legalized; |
8114 | } |
8115 | |
8116 | LegalizerHelper::LegalizeResult |
8117 | LegalizerHelper::lowerShlSat(MachineInstr &MI) { |
8118 | assert((MI.getOpcode() == TargetOpcode::G_SSHLSAT || |
8119 | MI.getOpcode() == TargetOpcode::G_USHLSAT) && |
8120 | "Expected shlsat opcode!" ); |
8121 | bool IsSigned = MI.getOpcode() == TargetOpcode::G_SSHLSAT; |
8122 | auto [Res, LHS, RHS] = MI.getFirst3Regs(); |
8123 | LLT Ty = MRI.getType(Reg: Res); |
8124 | LLT BoolTy = Ty.changeElementSize(NewEltSize: 1); |
8125 | |
8126 | unsigned BW = Ty.getScalarSizeInBits(); |
8127 | auto Result = MIRBuilder.buildShl(Dst: Ty, Src0: LHS, Src1: RHS); |
8128 | auto Orig = IsSigned ? MIRBuilder.buildAShr(Dst: Ty, Src0: Result, Src1: RHS) |
8129 | : MIRBuilder.buildLShr(Dst: Ty, Src0: Result, Src1: RHS); |
8130 | |
8131 | MachineInstrBuilder SatVal; |
8132 | if (IsSigned) { |
8133 | auto SatMin = MIRBuilder.buildConstant(Res: Ty, Val: APInt::getSignedMinValue(numBits: BW)); |
8134 | auto SatMax = MIRBuilder.buildConstant(Res: Ty, Val: APInt::getSignedMaxValue(numBits: BW)); |
8135 | auto Cmp = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_SLT, Res: BoolTy, Op0: LHS, |
8136 | Op1: MIRBuilder.buildConstant(Res: Ty, Val: 0)); |
8137 | SatVal = MIRBuilder.buildSelect(Res: Ty, Tst: Cmp, Op0: SatMin, Op1: SatMax); |
8138 | } else { |
8139 | SatVal = MIRBuilder.buildConstant(Res: Ty, Val: APInt::getMaxValue(numBits: BW)); |
8140 | } |
8141 | auto Ov = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_NE, Res: BoolTy, Op0: LHS, Op1: Orig); |
8142 | MIRBuilder.buildSelect(Res, Tst: Ov, Op0: SatVal, Op1: Result); |
8143 | |
8144 | MI.eraseFromParent(); |
8145 | return Legalized; |
8146 | } |
8147 | |
8148 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerBswap(MachineInstr &MI) { |
8149 | auto [Dst, Src] = MI.getFirst2Regs(); |
8150 | const LLT Ty = MRI.getType(Reg: Src); |
8151 | unsigned SizeInBytes = (Ty.getScalarSizeInBits() + 7) / 8; |
8152 | unsigned BaseShiftAmt = (SizeInBytes - 1) * 8; |
8153 | |
8154 | // Swap most and least significant byte, set remaining bytes in Res to zero. |
8155 | auto ShiftAmt = MIRBuilder.buildConstant(Res: Ty, Val: BaseShiftAmt); |
8156 | auto LSByteShiftedLeft = MIRBuilder.buildShl(Dst: Ty, Src0: Src, Src1: ShiftAmt); |
8157 | auto MSByteShiftedRight = MIRBuilder.buildLShr(Dst: Ty, Src0: Src, Src1: ShiftAmt); |
8158 | auto Res = MIRBuilder.buildOr(Dst: Ty, Src0: MSByteShiftedRight, Src1: LSByteShiftedLeft); |
8159 | |
8160 | // Set i-th high/low byte in Res to i-th low/high byte from Src. |
8161 | for (unsigned i = 1; i < SizeInBytes / 2; ++i) { |
8162 | // AND with Mask leaves byte i unchanged and sets remaining bytes to 0. |
8163 | APInt APMask(SizeInBytes * 8, 0xFF << (i * 8)); |
8164 | auto Mask = MIRBuilder.buildConstant(Res: Ty, Val: APMask); |
8165 | auto ShiftAmt = MIRBuilder.buildConstant(Res: Ty, Val: BaseShiftAmt - 16 * i); |
8166 | // Low byte shifted left to place of high byte: (Src & Mask) << ShiftAmt. |
8167 | auto LoByte = MIRBuilder.buildAnd(Dst: Ty, Src0: Src, Src1: Mask); |
8168 | auto LoShiftedLeft = MIRBuilder.buildShl(Dst: Ty, Src0: LoByte, Src1: ShiftAmt); |
8169 | Res = MIRBuilder.buildOr(Dst: Ty, Src0: Res, Src1: LoShiftedLeft); |
8170 | // High byte shifted right to place of low byte: (Src >> ShiftAmt) & Mask. |
8171 | auto SrcShiftedRight = MIRBuilder.buildLShr(Dst: Ty, Src0: Src, Src1: ShiftAmt); |
8172 | auto HiShiftedRight = MIRBuilder.buildAnd(Dst: Ty, Src0: SrcShiftedRight, Src1: Mask); |
8173 | Res = MIRBuilder.buildOr(Dst: Ty, Src0: Res, Src1: HiShiftedRight); |
8174 | } |
8175 | Res.getInstr()->getOperand(i: 0).setReg(Dst); |
8176 | |
8177 | MI.eraseFromParent(); |
8178 | return Legalized; |
8179 | } |
8180 | |
8181 | //{ (Src & Mask) >> N } | { (Src << N) & Mask } |
8182 | static MachineInstrBuilder SwapN(unsigned N, DstOp Dst, MachineIRBuilder &B, |
8183 | MachineInstrBuilder Src, const APInt &Mask) { |
8184 | const LLT Ty = Dst.getLLTTy(MRI: *B.getMRI()); |
8185 | MachineInstrBuilder C_N = B.buildConstant(Res: Ty, Val: N); |
8186 | MachineInstrBuilder MaskLoNTo0 = B.buildConstant(Res: Ty, Val: Mask); |
8187 | auto LHS = B.buildLShr(Dst: Ty, Src0: B.buildAnd(Dst: Ty, Src0: Src, Src1: MaskLoNTo0), Src1: C_N); |
8188 | auto RHS = B.buildAnd(Dst: Ty, Src0: B.buildShl(Dst: Ty, Src0: Src, Src1: C_N), Src1: MaskLoNTo0); |
8189 | return B.buildOr(Dst, Src0: LHS, Src1: RHS); |
8190 | } |
8191 | |
8192 | LegalizerHelper::LegalizeResult |
8193 | LegalizerHelper::lowerBitreverse(MachineInstr &MI) { |
8194 | auto [Dst, Src] = MI.getFirst2Regs(); |
8195 | const LLT Ty = MRI.getType(Reg: Src); |
8196 | unsigned Size = Ty.getScalarSizeInBits(); |
8197 | |
8198 | if (Size >= 8) { |
8199 | MachineInstrBuilder BSWAP = |
8200 | MIRBuilder.buildInstr(Opc: TargetOpcode::G_BSWAP, DstOps: {Ty}, SrcOps: {Src}); |
8201 | |
8202 | // swap high and low 4 bits in 8 bit blocks 7654|3210 -> 3210|7654 |
8203 | // [(val & 0xF0F0F0F0) >> 4] | [(val & 0x0F0F0F0F) << 4] |
8204 | // -> [(val & 0xF0F0F0F0) >> 4] | [(val << 4) & 0xF0F0F0F0] |
8205 | MachineInstrBuilder Swap4 = |
8206 | SwapN(N: 4, Dst: Ty, B&: MIRBuilder, Src: BSWAP, Mask: APInt::getSplat(NewLen: Size, V: APInt(8, 0xF0))); |
8207 | |
8208 | // swap high and low 2 bits in 4 bit blocks 32|10 76|54 -> 10|32 54|76 |
8209 | // [(val & 0xCCCCCCCC) >> 2] & [(val & 0x33333333) << 2] |
8210 | // -> [(val & 0xCCCCCCCC) >> 2] & [(val << 2) & 0xCCCCCCCC] |
8211 | MachineInstrBuilder Swap2 = |
8212 | SwapN(N: 2, Dst: Ty, B&: MIRBuilder, Src: Swap4, Mask: APInt::getSplat(NewLen: Size, V: APInt(8, 0xCC))); |
8213 | |
8214 | // swap high and low 1 bit in 2 bit blocks 1|0 3|2 5|4 7|6 -> 0|1 2|3 4|5 |
8215 | // 6|7 |
8216 | // [(val & 0xAAAAAAAA) >> 1] & [(val & 0x55555555) << 1] |
8217 | // -> [(val & 0xAAAAAAAA) >> 1] & [(val << 1) & 0xAAAAAAAA] |
8218 | SwapN(N: 1, Dst, B&: MIRBuilder, Src: Swap2, Mask: APInt::getSplat(NewLen: Size, V: APInt(8, 0xAA))); |
8219 | } else { |
8220 | // Expand bitreverse for types smaller than 8 bits. |
8221 | MachineInstrBuilder Tmp; |
8222 | for (unsigned I = 0, J = Size - 1; I < Size; ++I, --J) { |
8223 | MachineInstrBuilder Tmp2; |
8224 | if (I < J) { |
8225 | auto ShAmt = MIRBuilder.buildConstant(Res: Ty, Val: J - I); |
8226 | Tmp2 = MIRBuilder.buildShl(Dst: Ty, Src0: Src, Src1: ShAmt); |
8227 | } else { |
8228 | auto ShAmt = MIRBuilder.buildConstant(Res: Ty, Val: I - J); |
8229 | Tmp2 = MIRBuilder.buildLShr(Dst: Ty, Src0: Src, Src1: ShAmt); |
8230 | } |
8231 | |
8232 | auto Mask = MIRBuilder.buildConstant(Res: Ty, Val: 1ULL << J); |
8233 | Tmp2 = MIRBuilder.buildAnd(Dst: Ty, Src0: Tmp2, Src1: Mask); |
8234 | if (I == 0) |
8235 | Tmp = Tmp2; |
8236 | else |
8237 | Tmp = MIRBuilder.buildOr(Dst: Ty, Src0: Tmp, Src1: Tmp2); |
8238 | } |
8239 | MIRBuilder.buildCopy(Res: Dst, Op: Tmp); |
8240 | } |
8241 | |
8242 | MI.eraseFromParent(); |
8243 | return Legalized; |
8244 | } |
8245 | |
8246 | LegalizerHelper::LegalizeResult |
8247 | LegalizerHelper::lowerReadWriteRegister(MachineInstr &MI) { |
8248 | MachineFunction &MF = MIRBuilder.getMF(); |
8249 | |
8250 | bool IsRead = MI.getOpcode() == TargetOpcode::G_READ_REGISTER; |
8251 | int NameOpIdx = IsRead ? 1 : 0; |
8252 | int ValRegIndex = IsRead ? 0 : 1; |
8253 | |
8254 | Register ValReg = MI.getOperand(i: ValRegIndex).getReg(); |
8255 | const LLT Ty = MRI.getType(Reg: ValReg); |
8256 | const MDString *RegStr = cast<MDString>( |
8257 | Val: cast<MDNode>(Val: MI.getOperand(i: NameOpIdx).getMetadata())->getOperand(I: 0)); |
8258 | |
8259 | Register PhysReg = TLI.getRegisterByName(RegName: RegStr->getString().data(), Ty, MF); |
8260 | if (!PhysReg.isValid()) |
8261 | return UnableToLegalize; |
8262 | |
8263 | if (IsRead) |
8264 | MIRBuilder.buildCopy(Res: ValReg, Op: PhysReg); |
8265 | else |
8266 | MIRBuilder.buildCopy(Res: PhysReg, Op: ValReg); |
8267 | |
8268 | MI.eraseFromParent(); |
8269 | return Legalized; |
8270 | } |
8271 | |
8272 | LegalizerHelper::LegalizeResult |
8273 | LegalizerHelper::lowerSMULH_UMULH(MachineInstr &MI) { |
8274 | bool IsSigned = MI.getOpcode() == TargetOpcode::G_SMULH; |
8275 | unsigned ExtOp = IsSigned ? TargetOpcode::G_SEXT : TargetOpcode::G_ZEXT; |
8276 | Register Result = MI.getOperand(i: 0).getReg(); |
8277 | LLT OrigTy = MRI.getType(Reg: Result); |
8278 | auto SizeInBits = OrigTy.getScalarSizeInBits(); |
8279 | LLT WideTy = OrigTy.changeElementSize(NewEltSize: SizeInBits * 2); |
8280 | |
8281 | auto LHS = MIRBuilder.buildInstr(Opc: ExtOp, DstOps: {WideTy}, SrcOps: {MI.getOperand(i: 1)}); |
8282 | auto RHS = MIRBuilder.buildInstr(Opc: ExtOp, DstOps: {WideTy}, SrcOps: {MI.getOperand(i: 2)}); |
8283 | auto Mul = MIRBuilder.buildMul(Dst: WideTy, Src0: LHS, Src1: RHS); |
8284 | unsigned ShiftOp = IsSigned ? TargetOpcode::G_ASHR : TargetOpcode::G_LSHR; |
8285 | |
8286 | auto ShiftAmt = MIRBuilder.buildConstant(Res: WideTy, Val: SizeInBits); |
8287 | auto Shifted = MIRBuilder.buildInstr(Opc: ShiftOp, DstOps: {WideTy}, SrcOps: {Mul, ShiftAmt}); |
8288 | MIRBuilder.buildTrunc(Res: Result, Op: Shifted); |
8289 | |
8290 | MI.eraseFromParent(); |
8291 | return Legalized; |
8292 | } |
8293 | |
8294 | LegalizerHelper::LegalizeResult |
8295 | LegalizerHelper::lowerISFPCLASS(MachineInstr &MI) { |
8296 | auto [DstReg, DstTy, SrcReg, SrcTy] = MI.getFirst2RegLLTs(); |
8297 | FPClassTest Mask = static_cast<FPClassTest>(MI.getOperand(i: 2).getImm()); |
8298 | |
8299 | if (Mask == fcNone) { |
8300 | MIRBuilder.buildConstant(Res: DstReg, Val: 0); |
8301 | MI.eraseFromParent(); |
8302 | return Legalized; |
8303 | } |
8304 | if (Mask == fcAllFlags) { |
8305 | MIRBuilder.buildConstant(Res: DstReg, Val: 1); |
8306 | MI.eraseFromParent(); |
8307 | return Legalized; |
8308 | } |
8309 | |
8310 | // TODO: Try inverting the test with getInvertedFPClassTest like the DAG |
8311 | // version |
8312 | |
8313 | unsigned BitSize = SrcTy.getScalarSizeInBits(); |
8314 | const fltSemantics &Semantics = getFltSemanticForLLT(Ty: SrcTy.getScalarType()); |
8315 | |
8316 | LLT IntTy = LLT::scalar(SizeInBits: BitSize); |
8317 | if (SrcTy.isVector()) |
8318 | IntTy = LLT::vector(EC: SrcTy.getElementCount(), ScalarTy: IntTy); |
8319 | auto AsInt = MIRBuilder.buildCopy(Res: IntTy, Op: SrcReg); |
8320 | |
8321 | // Various masks. |
8322 | APInt SignBit = APInt::getSignMask(BitWidth: BitSize); |
8323 | APInt ValueMask = APInt::getSignedMaxValue(numBits: BitSize); // All bits but sign. |
8324 | APInt Inf = APFloat::getInf(Sem: Semantics).bitcastToAPInt(); // Exp and int bit. |
8325 | APInt ExpMask = Inf; |
8326 | APInt AllOneMantissa = APFloat::getLargest(Sem: Semantics).bitcastToAPInt() & ~Inf; |
8327 | APInt QNaNBitMask = |
8328 | APInt::getOneBitSet(numBits: BitSize, BitNo: AllOneMantissa.getActiveBits() - 1); |
8329 | APInt InvertionMask = APInt::getAllOnes(numBits: DstTy.getScalarSizeInBits()); |
8330 | |
8331 | auto SignBitC = MIRBuilder.buildConstant(Res: IntTy, Val: SignBit); |
8332 | auto ValueMaskC = MIRBuilder.buildConstant(Res: IntTy, Val: ValueMask); |
8333 | auto InfC = MIRBuilder.buildConstant(Res: IntTy, Val: Inf); |
8334 | auto ExpMaskC = MIRBuilder.buildConstant(Res: IntTy, Val: ExpMask); |
8335 | auto ZeroC = MIRBuilder.buildConstant(Res: IntTy, Val: 0); |
8336 | |
8337 | auto Abs = MIRBuilder.buildAnd(Dst: IntTy, Src0: AsInt, Src1: ValueMaskC); |
8338 | auto Sign = |
8339 | MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_NE, Res: DstTy, Op0: AsInt, Op1: Abs); |
8340 | |
8341 | auto Res = MIRBuilder.buildConstant(Res: DstTy, Val: 0); |
8342 | // Clang doesn't support capture of structured bindings: |
8343 | LLT DstTyCopy = DstTy; |
8344 | const auto appendToRes = [&](MachineInstrBuilder ToAppend) { |
8345 | Res = MIRBuilder.buildOr(Dst: DstTyCopy, Src0: Res, Src1: ToAppend); |
8346 | }; |
8347 | |
8348 | // Tests that involve more than one class should be processed first. |
8349 | if ((Mask & fcFinite) == fcFinite) { |
8350 | // finite(V) ==> abs(V) u< exp_mask |
8351 | appendToRes(MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_ULT, Res: DstTy, Op0: Abs, |
8352 | Op1: ExpMaskC)); |
8353 | Mask &= ~fcFinite; |
8354 | } else if ((Mask & fcFinite) == fcPosFinite) { |
8355 | // finite(V) && V > 0 ==> V u< exp_mask |
8356 | appendToRes(MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_ULT, Res: DstTy, Op0: AsInt, |
8357 | Op1: ExpMaskC)); |
8358 | Mask &= ~fcPosFinite; |
8359 | } else if ((Mask & fcFinite) == fcNegFinite) { |
8360 | // finite(V) && V < 0 ==> abs(V) u< exp_mask && signbit == 1 |
8361 | auto Cmp = MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_ULT, Res: DstTy, Op0: Abs, |
8362 | Op1: ExpMaskC); |
8363 | auto And = MIRBuilder.buildAnd(Dst: DstTy, Src0: Cmp, Src1: Sign); |
8364 | appendToRes(And); |
8365 | Mask &= ~fcNegFinite; |
8366 | } |
8367 | |
8368 | if (FPClassTest PartialCheck = Mask & (fcZero | fcSubnormal)) { |
8369 | // fcZero | fcSubnormal => test all exponent bits are 0 |
8370 | // TODO: Handle sign bit specific cases |
8371 | // TODO: Handle inverted case |
8372 | if (PartialCheck == (fcZero | fcSubnormal)) { |
8373 | auto ExpBits = MIRBuilder.buildAnd(Dst: IntTy, Src0: AsInt, Src1: ExpMaskC); |
8374 | appendToRes(MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_EQ, Res: DstTy, |
8375 | Op0: ExpBits, Op1: ZeroC)); |
8376 | Mask &= ~PartialCheck; |
8377 | } |
8378 | } |
8379 | |
8380 | // Check for individual classes. |
8381 | if (FPClassTest PartialCheck = Mask & fcZero) { |
8382 | if (PartialCheck == fcPosZero) |
8383 | appendToRes(MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_EQ, Res: DstTy, |
8384 | Op0: AsInt, Op1: ZeroC)); |
8385 | else if (PartialCheck == fcZero) |
8386 | appendToRes( |
8387 | MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_EQ, Res: DstTy, Op0: Abs, Op1: ZeroC)); |
8388 | else // fcNegZero |
8389 | appendToRes(MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_EQ, Res: DstTy, |
8390 | Op0: AsInt, Op1: SignBitC)); |
8391 | } |
8392 | |
8393 | if (FPClassTest PartialCheck = Mask & fcSubnormal) { |
8394 | // issubnormal(V) ==> unsigned(abs(V) - 1) u< (all mantissa bits set) |
8395 | // issubnormal(V) && V>0 ==> unsigned(V - 1) u< (all mantissa bits set) |
8396 | auto V = (PartialCheck == fcPosSubnormal) ? AsInt : Abs; |
8397 | auto OneC = MIRBuilder.buildConstant(Res: IntTy, Val: 1); |
8398 | auto VMinusOne = MIRBuilder.buildSub(Dst: IntTy, Src0: V, Src1: OneC); |
8399 | auto SubnormalRes = |
8400 | MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_ULT, Res: DstTy, Op0: VMinusOne, |
8401 | Op1: MIRBuilder.buildConstant(Res: IntTy, Val: AllOneMantissa)); |
8402 | if (PartialCheck == fcNegSubnormal) |
8403 | SubnormalRes = MIRBuilder.buildAnd(Dst: DstTy, Src0: SubnormalRes, Src1: Sign); |
8404 | appendToRes(SubnormalRes); |
8405 | } |
8406 | |
8407 | if (FPClassTest PartialCheck = Mask & fcInf) { |
8408 | if (PartialCheck == fcPosInf) |
8409 | appendToRes(MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_EQ, Res: DstTy, |
8410 | Op0: AsInt, Op1: InfC)); |
8411 | else if (PartialCheck == fcInf) |
8412 | appendToRes( |
8413 | MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_EQ, Res: DstTy, Op0: Abs, Op1: InfC)); |
8414 | else { // fcNegInf |
8415 | APInt NegInf = APFloat::getInf(Sem: Semantics, Negative: true).bitcastToAPInt(); |
8416 | auto NegInfC = MIRBuilder.buildConstant(Res: IntTy, Val: NegInf); |
8417 | appendToRes(MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_EQ, Res: DstTy, |
8418 | Op0: AsInt, Op1: NegInfC)); |
8419 | } |
8420 | } |
8421 | |
8422 | if (FPClassTest PartialCheck = Mask & fcNan) { |
8423 | auto InfWithQnanBitC = MIRBuilder.buildConstant(Res: IntTy, Val: Inf | QNaNBitMask); |
8424 | if (PartialCheck == fcNan) { |
8425 | // isnan(V) ==> abs(V) u> int(inf) |
8426 | appendToRes( |
8427 | MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_UGT, Res: DstTy, Op0: Abs, Op1: InfC)); |
8428 | } else if (PartialCheck == fcQNan) { |
8429 | // isquiet(V) ==> abs(V) u>= (unsigned(Inf) | quiet_bit) |
8430 | appendToRes(MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_UGE, Res: DstTy, Op0: Abs, |
8431 | Op1: InfWithQnanBitC)); |
8432 | } else { // fcSNan |
8433 | // issignaling(V) ==> abs(V) u> unsigned(Inf) && |
8434 | // abs(V) u< (unsigned(Inf) | quiet_bit) |
8435 | auto IsNan = |
8436 | MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_UGT, Res: DstTy, Op0: Abs, Op1: InfC); |
8437 | auto IsNotQnan = MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_ULT, Res: DstTy, |
8438 | Op0: Abs, Op1: InfWithQnanBitC); |
8439 | appendToRes(MIRBuilder.buildAnd(Dst: DstTy, Src0: IsNan, Src1: IsNotQnan)); |
8440 | } |
8441 | } |
8442 | |
8443 | if (FPClassTest PartialCheck = Mask & fcNormal) { |
8444 | // isnormal(V) ==> (0 u< exp u< max_exp) ==> (unsigned(exp-1) u< |
8445 | // (max_exp-1)) |
8446 | APInt ExpLSB = ExpMask & ~(ExpMask.shl(shiftAmt: 1)); |
8447 | auto ExpMinusOne = MIRBuilder.buildSub( |
8448 | Dst: IntTy, Src0: Abs, Src1: MIRBuilder.buildConstant(Res: IntTy, Val: ExpLSB)); |
8449 | APInt MaxExpMinusOne = ExpMask - ExpLSB; |
8450 | auto NormalRes = |
8451 | MIRBuilder.buildICmp(Pred: CmpInst::Predicate::ICMP_ULT, Res: DstTy, Op0: ExpMinusOne, |
8452 | Op1: MIRBuilder.buildConstant(Res: IntTy, Val: MaxExpMinusOne)); |
8453 | if (PartialCheck == fcNegNormal) |
8454 | NormalRes = MIRBuilder.buildAnd(Dst: DstTy, Src0: NormalRes, Src1: Sign); |
8455 | else if (PartialCheck == fcPosNormal) { |
8456 | auto PosSign = MIRBuilder.buildXor( |
8457 | Dst: DstTy, Src0: Sign, Src1: MIRBuilder.buildConstant(Res: DstTy, Val: InvertionMask)); |
8458 | NormalRes = MIRBuilder.buildAnd(Dst: DstTy, Src0: NormalRes, Src1: PosSign); |
8459 | } |
8460 | appendToRes(NormalRes); |
8461 | } |
8462 | |
8463 | MIRBuilder.buildCopy(Res: DstReg, Op: Res); |
8464 | MI.eraseFromParent(); |
8465 | return Legalized; |
8466 | } |
8467 | |
8468 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerSelect(MachineInstr &MI) { |
8469 | // Implement G_SELECT in terms of XOR, AND, OR. |
8470 | auto [DstReg, DstTy, MaskReg, MaskTy, Op1Reg, Op1Ty, Op2Reg, Op2Ty] = |
8471 | MI.getFirst4RegLLTs(); |
8472 | |
8473 | bool IsEltPtr = DstTy.isPointerOrPointerVector(); |
8474 | if (IsEltPtr) { |
8475 | LLT ScalarPtrTy = LLT::scalar(SizeInBits: DstTy.getScalarSizeInBits()); |
8476 | LLT NewTy = DstTy.changeElementType(NewEltTy: ScalarPtrTy); |
8477 | Op1Reg = MIRBuilder.buildPtrToInt(Dst: NewTy, Src: Op1Reg).getReg(Idx: 0); |
8478 | Op2Reg = MIRBuilder.buildPtrToInt(Dst: NewTy, Src: Op2Reg).getReg(Idx: 0); |
8479 | DstTy = NewTy; |
8480 | } |
8481 | |
8482 | if (MaskTy.isScalar()) { |
8483 | // Turn the scalar condition into a vector condition mask if needed. |
8484 | |
8485 | Register MaskElt = MaskReg; |
8486 | |
8487 | // The condition was potentially zero extended before, but we want a sign |
8488 | // extended boolean. |
8489 | if (MaskTy != LLT::scalar(SizeInBits: 1)) |
8490 | MaskElt = MIRBuilder.buildSExtInReg(Res: MaskTy, Op: MaskElt, ImmOp: 1).getReg(Idx: 0); |
8491 | |
8492 | // Continue the sign extension (or truncate) to match the data type. |
8493 | MaskElt = |
8494 | MIRBuilder.buildSExtOrTrunc(Res: DstTy.getScalarType(), Op: MaskElt).getReg(Idx: 0); |
8495 | |
8496 | if (DstTy.isVector()) { |
8497 | // Generate a vector splat idiom. |
8498 | auto ShufSplat = MIRBuilder.buildShuffleSplat(Res: DstTy, Src: MaskElt); |
8499 | MaskReg = ShufSplat.getReg(Idx: 0); |
8500 | } else { |
8501 | MaskReg = MaskElt; |
8502 | } |
8503 | MaskTy = DstTy; |
8504 | } else if (!DstTy.isVector()) { |
8505 | // Cannot handle the case that mask is a vector and dst is a scalar. |
8506 | return UnableToLegalize; |
8507 | } |
8508 | |
8509 | if (MaskTy.getSizeInBits() != DstTy.getSizeInBits()) { |
8510 | return UnableToLegalize; |
8511 | } |
8512 | |
8513 | auto NotMask = MIRBuilder.buildNot(Dst: MaskTy, Src0: MaskReg); |
8514 | auto NewOp1 = MIRBuilder.buildAnd(Dst: MaskTy, Src0: Op1Reg, Src1: MaskReg); |
8515 | auto NewOp2 = MIRBuilder.buildAnd(Dst: MaskTy, Src0: Op2Reg, Src1: NotMask); |
8516 | if (IsEltPtr) { |
8517 | auto Or = MIRBuilder.buildOr(Dst: DstTy, Src0: NewOp1, Src1: NewOp2); |
8518 | MIRBuilder.buildIntToPtr(Dst: DstReg, Src: Or); |
8519 | } else { |
8520 | MIRBuilder.buildOr(Dst: DstReg, Src0: NewOp1, Src1: NewOp2); |
8521 | } |
8522 | MI.eraseFromParent(); |
8523 | return Legalized; |
8524 | } |
8525 | |
8526 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerDIVREM(MachineInstr &MI) { |
8527 | // Split DIVREM into individual instructions. |
8528 | unsigned Opcode = MI.getOpcode(); |
8529 | |
8530 | MIRBuilder.buildInstr( |
8531 | Opc: Opcode == TargetOpcode::G_SDIVREM ? TargetOpcode::G_SDIV |
8532 | : TargetOpcode::G_UDIV, |
8533 | DstOps: {MI.getOperand(i: 0).getReg()}, SrcOps: {MI.getOperand(i: 2), MI.getOperand(i: 3)}); |
8534 | MIRBuilder.buildInstr( |
8535 | Opc: Opcode == TargetOpcode::G_SDIVREM ? TargetOpcode::G_SREM |
8536 | : TargetOpcode::G_UREM, |
8537 | DstOps: {MI.getOperand(i: 1).getReg()}, SrcOps: {MI.getOperand(i: 2), MI.getOperand(i: 3)}); |
8538 | MI.eraseFromParent(); |
8539 | return Legalized; |
8540 | } |
8541 | |
8542 | LegalizerHelper::LegalizeResult |
8543 | LegalizerHelper::lowerAbsToAddXor(MachineInstr &MI) { |
8544 | // Expand %res = G_ABS %a into: |
8545 | // %v1 = G_ASHR %a, scalar_size-1 |
8546 | // %v2 = G_ADD %a, %v1 |
8547 | // %res = G_XOR %v2, %v1 |
8548 | LLT DstTy = MRI.getType(Reg: MI.getOperand(i: 0).getReg()); |
8549 | Register OpReg = MI.getOperand(i: 1).getReg(); |
8550 | auto ShiftAmt = |
8551 | MIRBuilder.buildConstant(Res: DstTy, Val: DstTy.getScalarSizeInBits() - 1); |
8552 | auto Shift = MIRBuilder.buildAShr(Dst: DstTy, Src0: OpReg, Src1: ShiftAmt); |
8553 | auto Add = MIRBuilder.buildAdd(Dst: DstTy, Src0: OpReg, Src1: Shift); |
8554 | MIRBuilder.buildXor(Dst: MI.getOperand(i: 0).getReg(), Src0: Add, Src1: Shift); |
8555 | MI.eraseFromParent(); |
8556 | return Legalized; |
8557 | } |
8558 | |
8559 | LegalizerHelper::LegalizeResult |
8560 | LegalizerHelper::lowerAbsToMaxNeg(MachineInstr &MI) { |
8561 | // Expand %res = G_ABS %a into: |
8562 | // %v1 = G_CONSTANT 0 |
8563 | // %v2 = G_SUB %v1, %a |
8564 | // %res = G_SMAX %a, %v2 |
8565 | Register SrcReg = MI.getOperand(i: 1).getReg(); |
8566 | LLT Ty = MRI.getType(Reg: SrcReg); |
8567 | auto Zero = MIRBuilder.buildConstant(Res: Ty, Val: 0); |
8568 | auto Sub = MIRBuilder.buildSub(Dst: Ty, Src0: Zero, Src1: SrcReg); |
8569 | MIRBuilder.buildSMax(Dst: MI.getOperand(i: 0), Src0: SrcReg, Src1: Sub); |
8570 | MI.eraseFromParent(); |
8571 | return Legalized; |
8572 | } |
8573 | |
8574 | LegalizerHelper::LegalizeResult |
8575 | LegalizerHelper::lowerAbsToCNeg(MachineInstr &MI) { |
8576 | Register SrcReg = MI.getOperand(i: 1).getReg(); |
8577 | Register DestReg = MI.getOperand(i: 0).getReg(); |
8578 | LLT Ty = MRI.getType(Reg: SrcReg), IType = LLT::scalar(SizeInBits: 1); |
8579 | auto Zero = MIRBuilder.buildConstant(Res: Ty, Val: 0).getReg(Idx: 0); |
8580 | auto Sub = MIRBuilder.buildSub(Dst: Ty, Src0: Zero, Src1: SrcReg).getReg(Idx: 0); |
8581 | auto ICmp = MIRBuilder.buildICmp(Pred: CmpInst::ICMP_SGT, Res: IType, Op0: SrcReg, Op1: Zero); |
8582 | MIRBuilder.buildSelect(Res: DestReg, Tst: ICmp, Op0: SrcReg, Op1: Sub); |
8583 | MI.eraseFromParent(); |
8584 | return Legalized; |
8585 | } |
8586 | |
8587 | LegalizerHelper::LegalizeResult |
8588 | LegalizerHelper::lowerVectorReduction(MachineInstr &MI) { |
8589 | Register SrcReg = MI.getOperand(i: 1).getReg(); |
8590 | LLT SrcTy = MRI.getType(Reg: SrcReg); |
8591 | LLT DstTy = MRI.getType(Reg: SrcReg); |
8592 | |
8593 | // The source could be a scalar if the IR type was <1 x sN>. |
8594 | if (SrcTy.isScalar()) { |
8595 | if (DstTy.getSizeInBits() > SrcTy.getSizeInBits()) |
8596 | return UnableToLegalize; // FIXME: handle extension. |
8597 | // This can be just a plain copy. |
8598 | Observer.changingInstr(MI); |
8599 | MI.setDesc(MIRBuilder.getTII().get(Opcode: TargetOpcode::COPY)); |
8600 | Observer.changedInstr(MI); |
8601 | return Legalized; |
8602 | } |
8603 | return UnableToLegalize; |
8604 | } |
8605 | |
8606 | LegalizerHelper::LegalizeResult LegalizerHelper::lowerVAArg(MachineInstr &MI) { |
8607 | MachineFunction &MF = *MI.getMF(); |
8608 | const DataLayout &DL = MIRBuilder.getDataLayout(); |
8609 | LLVMContext &Ctx = MF.getFunction().getContext(); |
8610 | Register ListPtr = MI.getOperand(i: 1).getReg(); |
8611 | LLT PtrTy = MRI.getType(Reg: ListPtr); |
8612 | |
8613 | // LstPtr is a pointer to the head of the list. Get the address |
8614 | // of the head of the list. |
8615 | Align PtrAlignment = DL.getABITypeAlign(Ty: getTypeForLLT(Ty: PtrTy, C&: Ctx)); |
8616 | MachineMemOperand *PtrLoadMMO = MF.getMachineMemOperand( |
8617 | PtrInfo: MachinePointerInfo(), f: MachineMemOperand::MOLoad, MemTy: PtrTy, base_alignment: PtrAlignment); |
8618 | auto VAList = MIRBuilder.buildLoad(Res: PtrTy, Addr: ListPtr, MMO&: *PtrLoadMMO).getReg(Idx: 0); |
8619 | |
8620 | const Align A(MI.getOperand(i: 2).getImm()); |
8621 | LLT PtrTyAsScalarTy = LLT::scalar(SizeInBits: PtrTy.getSizeInBits()); |
8622 | if (A > TLI.getMinStackArgumentAlignment()) { |
8623 | Register AlignAmt = |
8624 | MIRBuilder.buildConstant(Res: PtrTyAsScalarTy, Val: A.value() - 1).getReg(Idx: 0); |
8625 | auto AddDst = MIRBuilder.buildPtrAdd(Res: PtrTy, Op0: VAList, Op1: AlignAmt); |
8626 | auto AndDst = MIRBuilder.buildMaskLowPtrBits(Res: PtrTy, Op0: AddDst, NumBits: Log2(A)); |
8627 | VAList = AndDst.getReg(Idx: 0); |
8628 | } |
8629 | |
8630 | // Increment the pointer, VAList, to the next vaarg |
8631 | // The list should be bumped by the size of element in the current head of |
8632 | // list. |
8633 | Register Dst = MI.getOperand(i: 0).getReg(); |
8634 | LLT LLTTy = MRI.getType(Reg: Dst); |
8635 | Type *Ty = getTypeForLLT(Ty: LLTTy, C&: Ctx); |
8636 | auto IncAmt = |
8637 | MIRBuilder.buildConstant(Res: PtrTyAsScalarTy, Val: DL.getTypeAllocSize(Ty)); |
8638 | auto Succ = MIRBuilder.buildPtrAdd(Res: PtrTy, Op0: VAList, Op1: IncAmt); |
8639 | |
8640 | // Store the increment VAList to the legalized pointer |
8641 | MachineMemOperand *StoreMMO = MF.getMachineMemOperand( |
8642 | PtrInfo: MachinePointerInfo(), f: MachineMemOperand::MOStore, MemTy: PtrTy, base_alignment: PtrAlignment); |
8643 | MIRBuilder.buildStore(Val: Succ, Addr: ListPtr, MMO&: *StoreMMO); |
8644 | // Load the actual argument out of the pointer VAList |
8645 | Align EltAlignment = DL.getABITypeAlign(Ty); |
8646 | MachineMemOperand *EltLoadMMO = MF.getMachineMemOperand( |
8647 | PtrInfo: MachinePointerInfo(), f: MachineMemOperand::MOLoad, MemTy: LLTTy, base_alignment: EltAlignment); |
8648 | MIRBuilder.buildLoad(Res: Dst, Addr: VAList, MMO&: *EltLoadMMO); |
8649 | |
8650 | MI.eraseFromParent(); |
8651 | return Legalized; |
8652 | } |
8653 | |
8654 | static bool shouldLowerMemFuncForSize(const MachineFunction &MF) { |
8655 | // On Darwin, -Os means optimize for size without hurting performance, so |
8656 | // only really optimize for size when -Oz (MinSize) is used. |
8657 | if (MF.getTarget().getTargetTriple().isOSDarwin()) |
8658 | return MF.getFunction().hasMinSize(); |
8659 | return MF.getFunction().hasOptSize(); |
8660 | } |
8661 | |
8662 | // Returns a list of types to use for memory op lowering in MemOps. A partial |
8663 | // port of findOptimalMemOpLowering in TargetLowering. |
8664 | static bool findGISelOptimalMemOpLowering(std::vector<LLT> &MemOps, |
8665 | unsigned Limit, const MemOp &Op, |
8666 | unsigned DstAS, unsigned SrcAS, |
8667 | const AttributeList &FuncAttributes, |
8668 | const TargetLowering &TLI) { |
8669 | if (Op.isMemcpyWithFixedDstAlign() && Op.getSrcAlign() < Op.getDstAlign()) |
8670 | return false; |
8671 | |
8672 | LLT Ty = TLI.getOptimalMemOpLLT(Op, FuncAttributes); |
8673 | |
8674 | if (Ty == LLT()) { |
8675 | // Use the largest scalar type whose alignment constraints are satisfied. |
8676 | // We only need to check DstAlign here as SrcAlign is always greater or |
8677 | // equal to DstAlign (or zero). |
8678 | Ty = LLT::scalar(SizeInBits: 64); |
8679 | if (Op.isFixedDstAlign()) |
8680 | while (Op.getDstAlign() < Ty.getSizeInBytes() && |
8681 | !TLI.allowsMisalignedMemoryAccesses(Ty, AddrSpace: DstAS, Alignment: Op.getDstAlign())) |
8682 | Ty = LLT::scalar(SizeInBits: Ty.getSizeInBytes()); |
8683 | assert(Ty.getSizeInBits() > 0 && "Could not find valid type" ); |
8684 | // FIXME: check for the largest legal type we can load/store to. |
8685 | } |
8686 | |
8687 | unsigned NumMemOps = 0; |
8688 | uint64_t Size = Op.size(); |
8689 | while (Size) { |
8690 | unsigned TySize = Ty.getSizeInBytes(); |
8691 | while (TySize > Size) { |
8692 | // For now, only use non-vector load / store's for the left-over pieces. |
8693 | LLT NewTy = Ty; |
8694 | // FIXME: check for mem op safety and legality of the types. Not all of |
8695 | // SDAGisms map cleanly to GISel concepts. |
8696 | if (NewTy.isVector()) |
8697 | NewTy = NewTy.getSizeInBits() > 64 ? LLT::scalar(SizeInBits: 64) : LLT::scalar(SizeInBits: 32); |
8698 | NewTy = LLT::scalar(SizeInBits: llvm::bit_floor(Value: NewTy.getSizeInBits() - 1)); |
8699 | unsigned NewTySize = NewTy.getSizeInBytes(); |
8700 | assert(NewTySize > 0 && "Could not find appropriate type" ); |
8701 | |
8702 | // If the new LLT cannot cover all of the remaining bits, then consider |
8703 | // issuing a (or a pair of) unaligned and overlapping load / store. |
8704 | unsigned Fast; |
8705 | // Need to get a VT equivalent for allowMisalignedMemoryAccesses(). |
8706 | MVT VT = getMVTForLLT(Ty); |
8707 | if (NumMemOps && Op.allowOverlap() && NewTySize < Size && |
8708 | TLI.allowsMisalignedMemoryAccesses( |
8709 | VT, AddrSpace: DstAS, Alignment: Op.isFixedDstAlign() ? Op.getDstAlign() : Align(1), |
8710 | Flags: MachineMemOperand::MONone, &Fast) && |
8711 | Fast) |
8712 | TySize = Size; |
8713 | else { |
8714 | Ty = NewTy; |
8715 | TySize = NewTySize; |
8716 | } |
8717 | } |
8718 | |
8719 | if (++NumMemOps > Limit) |
8720 | return false; |
8721 | |
8722 | MemOps.push_back(x: Ty); |
8723 | Size -= TySize; |
8724 | } |
8725 | |
8726 | return true; |
8727 | } |
8728 | |
8729 | // Get a vectorized representation of the memset value operand, GISel edition. |
8730 | static Register getMemsetValue(Register Val, LLT Ty, MachineIRBuilder &MIB) { |
8731 | MachineRegisterInfo &MRI = *MIB.getMRI(); |
8732 | unsigned NumBits = Ty.getScalarSizeInBits(); |
8733 | auto ValVRegAndVal = getIConstantVRegValWithLookThrough(VReg: Val, MRI); |
8734 | if (!Ty.isVector() && ValVRegAndVal) { |
8735 | APInt Scalar = ValVRegAndVal->Value.trunc(width: 8); |
8736 | APInt SplatVal = APInt::getSplat(NewLen: NumBits, V: Scalar); |
8737 | return MIB.buildConstant(Res: Ty, Val: SplatVal).getReg(Idx: 0); |
8738 | } |
8739 | |
8740 | // Extend the byte value to the larger type, and then multiply by a magic |
8741 | // value 0x010101... in order to replicate it across every byte. |
8742 | // Unless it's zero, in which case just emit a larger G_CONSTANT 0. |
8743 | if (ValVRegAndVal && ValVRegAndVal->Value == 0) { |
8744 | return MIB.buildConstant(Res: Ty, Val: 0).getReg(Idx: 0); |
8745 | } |
8746 | |
8747 | LLT ExtType = Ty.getScalarType(); |
8748 | auto ZExt = MIB.buildZExtOrTrunc(Res: ExtType, Op: Val); |
8749 | if (NumBits > 8) { |
8750 | APInt Magic = APInt::getSplat(NewLen: NumBits, V: APInt(8, 0x01)); |
8751 | auto MagicMI = MIB.buildConstant(Res: ExtType, Val: Magic); |
8752 | Val = MIB.buildMul(Dst: ExtType, Src0: ZExt, Src1: MagicMI).getReg(Idx: 0); |
8753 | } |
8754 | |
8755 | // For vector types create a G_BUILD_VECTOR. |
8756 | if (Ty.isVector()) |
8757 | Val = MIB.buildSplatBuildVector(Res: Ty, Src: Val).getReg(Idx: 0); |
8758 | |
8759 | return Val; |
8760 | } |
8761 | |
8762 | LegalizerHelper::LegalizeResult |
8763 | LegalizerHelper::lowerMemset(MachineInstr &MI, Register Dst, Register Val, |
8764 | uint64_t KnownLen, Align Alignment, |
8765 | bool IsVolatile) { |
8766 | auto &MF = *MI.getParent()->getParent(); |
8767 | const auto &TLI = *MF.getSubtarget().getTargetLowering(); |
8768 | auto &DL = MF.getDataLayout(); |
8769 | LLVMContext &C = MF.getFunction().getContext(); |
8770 | |
8771 | assert(KnownLen != 0 && "Have a zero length memset length!" ); |
8772 | |
8773 | bool DstAlignCanChange = false; |
8774 | MachineFrameInfo &MFI = MF.getFrameInfo(); |
8775 | bool OptSize = shouldLowerMemFuncForSize(MF); |
8776 | |
8777 | MachineInstr *FIDef = getOpcodeDef(Opcode: TargetOpcode::G_FRAME_INDEX, Reg: Dst, MRI); |
8778 | if (FIDef && !MFI.isFixedObjectIndex(ObjectIdx: FIDef->getOperand(i: 1).getIndex())) |
8779 | DstAlignCanChange = true; |
8780 | |
8781 | unsigned Limit = TLI.getMaxStoresPerMemset(OptSize); |
8782 | std::vector<LLT> MemOps; |
8783 | |
8784 | const auto &DstMMO = **MI.memoperands_begin(); |
8785 | MachinePointerInfo DstPtrInfo = DstMMO.getPointerInfo(); |
8786 | |
8787 | auto ValVRegAndVal = getIConstantVRegValWithLookThrough(VReg: Val, MRI); |
8788 | bool IsZeroVal = ValVRegAndVal && ValVRegAndVal->Value == 0; |
8789 | |
8790 | if (!findGISelOptimalMemOpLowering(MemOps, Limit, |
8791 | Op: MemOp::Set(Size: KnownLen, DstAlignCanChange, |
8792 | DstAlign: Alignment, |
8793 | /*IsZeroMemset=*/IsZeroVal, |
8794 | /*IsVolatile=*/IsVolatile), |
8795 | DstAS: DstPtrInfo.getAddrSpace(), SrcAS: ~0u, |
8796 | FuncAttributes: MF.getFunction().getAttributes(), TLI)) |
8797 | return UnableToLegalize; |
8798 | |
8799 | if (DstAlignCanChange) { |
8800 | // Get an estimate of the type from the LLT. |
8801 | Type *IRTy = getTypeForLLT(Ty: MemOps[0], C); |
8802 | Align NewAlign = DL.getABITypeAlign(Ty: IRTy); |
8803 | if (NewAlign > Alignment) { |
8804 | Alignment = NewAlign; |
8805 | unsigned FI = FIDef->getOperand(i: 1).getIndex(); |
8806 | // Give the stack frame object a larger alignment if needed. |
8807 | if (MFI.getObjectAlign(ObjectIdx: FI) < Alignment) |
8808 | MFI.setObjectAlignment(ObjectIdx: FI, Alignment); |
8809 | } |
8810 | } |
8811 | |
8812 | MachineIRBuilder MIB(MI); |
8813 | // Find the largest store and generate the bit pattern for it. |
8814 | LLT LargestTy = MemOps[0]; |
8815 | for (unsigned i = 1; i < MemOps.size(); i++) |
8816 | if (MemOps[i].getSizeInBits() > LargestTy.getSizeInBits()) |
8817 | LargestTy = MemOps[i]; |
8818 | |
8819 | // The memset stored value is always defined as an s8, so in order to make it |
8820 | // work with larger store types we need to repeat the bit pattern across the |
8821 | // wider type. |
8822 | Register MemSetValue = getMemsetValue(Val, Ty: LargestTy, MIB); |
8823 | |
8824 | if (!MemSetValue) |
8825 | return UnableToLegalize; |
8826 | |
8827 | // Generate the stores. For each store type in the list, we generate the |
8828 | // matching store of that type to the destination address. |
8829 | LLT PtrTy = MRI.getType(Reg: Dst); |
8830 | unsigned DstOff = 0; |
8831 | unsigned Size = KnownLen; |
8832 | for (unsigned I = 0; I < MemOps.size(); I++) { |
8833 | LLT Ty = MemOps[I]; |
8834 | unsigned TySize = Ty.getSizeInBytes(); |
8835 | if (TySize > Size) { |
8836 | // Issuing an unaligned load / store pair that overlaps with the previous |
8837 | // pair. Adjust the offset accordingly. |
8838 | assert(I == MemOps.size() - 1 && I != 0); |
8839 | DstOff -= TySize - Size; |
8840 | } |
8841 | |
8842 | // If this store is smaller than the largest store see whether we can get |
8843 | // the smaller value for free with a truncate. |
8844 | Register Value = MemSetValue; |
8845 | if (Ty.getSizeInBits() < LargestTy.getSizeInBits()) { |
8846 | MVT VT = getMVTForLLT(Ty); |
8847 | MVT LargestVT = getMVTForLLT(Ty: LargestTy); |
8848 | if (!LargestTy.isVector() && !Ty.isVector() && |
8849 | TLI.isTruncateFree(FromVT: LargestVT, ToVT: VT)) |
8850 | Value = MIB.buildTrunc(Res: Ty, Op: MemSetValue).getReg(Idx: 0); |
8851 | else |
8852 | Value = getMemsetValue(Val, Ty, MIB); |
8853 | if (!Value) |
8854 | return UnableToLegalize; |
8855 | } |
8856 | |
8857 | auto *StoreMMO = MF.getMachineMemOperand(MMO: &DstMMO, Offset: DstOff, Ty); |
8858 | |
8859 | Register Ptr = Dst; |
8860 | if (DstOff != 0) { |
8861 | auto Offset = |
8862 | MIB.buildConstant(Res: LLT::scalar(SizeInBits: PtrTy.getSizeInBits()), Val: DstOff); |
8863 | Ptr = MIB.buildPtrAdd(Res: PtrTy, Op0: Dst, Op1: Offset).getReg(Idx: 0); |
8864 | } |
8865 | |
8866 | MIB.buildStore(Val: Value, Addr: Ptr, MMO&: *StoreMMO); |
8867 | DstOff += Ty.getSizeInBytes(); |
8868 | Size -= TySize; |
8869 | } |
8870 | |
8871 | MI.eraseFromParent(); |
8872 | return Legalized; |
8873 | } |
8874 | |
8875 | LegalizerHelper::LegalizeResult |
8876 | LegalizerHelper::lowerMemcpyInline(MachineInstr &MI) { |
8877 | assert(MI.getOpcode() == TargetOpcode::G_MEMCPY_INLINE); |
8878 | |
8879 | auto [Dst, Src, Len] = MI.getFirst3Regs(); |
8880 | |
8881 | const auto *MMOIt = MI.memoperands_begin(); |
8882 | const MachineMemOperand *MemOp = *MMOIt; |
8883 | bool IsVolatile = MemOp->isVolatile(); |
8884 | |
8885 | // See if this is a constant length copy |
8886 | auto LenVRegAndVal = getIConstantVRegValWithLookThrough(VReg: Len, MRI); |
8887 | // FIXME: support dynamically sized G_MEMCPY_INLINE |
8888 | assert(LenVRegAndVal && |
8889 | "inline memcpy with dynamic size is not yet supported" ); |
8890 | uint64_t KnownLen = LenVRegAndVal->Value.getZExtValue(); |
8891 | if (KnownLen == 0) { |
8892 | MI.eraseFromParent(); |
8893 | return Legalized; |
8894 | } |
8895 | |
8896 | const auto &DstMMO = **MI.memoperands_begin(); |
8897 | const auto &SrcMMO = **std::next(x: MI.memoperands_begin()); |
8898 | Align DstAlign = DstMMO.getBaseAlign(); |
8899 | Align SrcAlign = SrcMMO.getBaseAlign(); |
8900 | |
8901 | return lowerMemcpyInline(MI, Dst, Src, KnownLen, DstAlign, SrcAlign, |
8902 | IsVolatile); |
8903 | } |
8904 | |
8905 | LegalizerHelper::LegalizeResult |
8906 | LegalizerHelper::lowerMemcpyInline(MachineInstr &MI, Register Dst, Register Src, |
8907 | uint64_t KnownLen, Align DstAlign, |
8908 | Align SrcAlign, bool IsVolatile) { |
8909 | assert(MI.getOpcode() == TargetOpcode::G_MEMCPY_INLINE); |
8910 | return lowerMemcpy(MI, Dst, Src, KnownLen, |
8911 | Limit: std::numeric_limits<uint64_t>::max(), DstAlign, SrcAlign, |
8912 | IsVolatile); |
8913 | } |
8914 | |
8915 | LegalizerHelper::LegalizeResult |
8916 | LegalizerHelper::lowerMemcpy(MachineInstr &MI, Register Dst, Register Src, |
8917 | uint64_t KnownLen, uint64_t Limit, Align DstAlign, |
8918 | Align SrcAlign, bool IsVolatile) { |
8919 | auto &MF = *MI.getParent()->getParent(); |
8920 | const auto &TLI = *MF.getSubtarget().getTargetLowering(); |
8921 | auto &DL = MF.getDataLayout(); |
8922 | LLVMContext &C = MF.getFunction().getContext(); |
8923 | |
8924 | assert(KnownLen != 0 && "Have a zero length memcpy length!" ); |
8925 | |
8926 | bool DstAlignCanChange = false; |
8927 | MachineFrameInfo &MFI = MF.getFrameInfo(); |
8928 | Align Alignment = std::min(a: DstAlign, b: SrcAlign); |
8929 | |
8930 | MachineInstr *FIDef = getOpcodeDef(Opcode: TargetOpcode::G_FRAME_INDEX, Reg: Dst, MRI); |
8931 | if (FIDef && !MFI.isFixedObjectIndex(ObjectIdx: FIDef->getOperand(i: 1).getIndex())) |
8932 | DstAlignCanChange = true; |
8933 | |
8934 | // FIXME: infer better src pointer alignment like SelectionDAG does here. |
8935 | // FIXME: also use the equivalent of isMemSrcFromConstant and alwaysinlining |
8936 | // if the memcpy is in a tail call position. |
8937 | |
8938 | std::vector<LLT> MemOps; |
8939 | |
8940 | const auto &DstMMO = **MI.memoperands_begin(); |
8941 | const auto &SrcMMO = **std::next(x: MI.memoperands_begin()); |
8942 | MachinePointerInfo DstPtrInfo = DstMMO.getPointerInfo(); |
8943 | MachinePointerInfo SrcPtrInfo = SrcMMO.getPointerInfo(); |
8944 | |
8945 | if (!findGISelOptimalMemOpLowering( |
8946 | MemOps, Limit, |
8947 | Op: MemOp::Copy(Size: KnownLen, DstAlignCanChange, DstAlign: Alignment, SrcAlign, |
8948 | IsVolatile), |
8949 | DstAS: DstPtrInfo.getAddrSpace(), SrcAS: SrcPtrInfo.getAddrSpace(), |
8950 | FuncAttributes: MF.getFunction().getAttributes(), TLI)) |
8951 | return UnableToLegalize; |
8952 | |
8953 | if (DstAlignCanChange) { |
8954 | // Get an estimate of the type from the LLT. |
8955 | Type *IRTy = getTypeForLLT(Ty: MemOps[0], C); |
8956 | Align NewAlign = DL.getABITypeAlign(Ty: IRTy); |
8957 | |
8958 | // Don't promote to an alignment that would require dynamic stack |
8959 | // realignment. |
8960 | const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo(); |
8961 | if (!TRI->hasStackRealignment(MF)) |
8962 | while (NewAlign > Alignment && DL.exceedsNaturalStackAlignment(Alignment: NewAlign)) |
8963 | NewAlign = NewAlign.previous(); |
8964 | |
8965 | if (NewAlign > Alignment) { |
8966 | Alignment = NewAlign; |
8967 | unsigned FI = FIDef->getOperand(i: 1).getIndex(); |
8968 | // Give the stack frame object a larger alignment if needed. |
8969 | if (MFI.getObjectAlign(ObjectIdx: FI) < Alignment) |
8970 | MFI.setObjectAlignment(ObjectIdx: FI, Alignment); |
8971 | } |
8972 | } |
8973 | |
8974 | LLVM_DEBUG(dbgs() << "Inlining memcpy: " << MI << " into loads & stores\n" ); |
8975 | |
8976 | MachineIRBuilder MIB(MI); |
8977 | // Now we need to emit a pair of load and stores for each of the types we've |
8978 | // collected. I.e. for each type, generate a load from the source pointer of |
8979 | // that type width, and then generate a corresponding store to the dest buffer |
8980 | // of that value loaded. This can result in a sequence of loads and stores |
8981 | // mixed types, depending on what the target specifies as good types to use. |
8982 | unsigned CurrOffset = 0; |
8983 | unsigned Size = KnownLen; |
8984 | for (auto CopyTy : MemOps) { |
8985 | // Issuing an unaligned load / store pair that overlaps with the previous |
8986 | // pair. Adjust the offset accordingly. |
8987 | if (CopyTy.getSizeInBytes() > Size) |
8988 | CurrOffset -= CopyTy.getSizeInBytes() - Size; |
8989 | |
8990 | // Construct MMOs for the accesses. |
8991 | auto *LoadMMO = |
8992 | MF.getMachineMemOperand(MMO: &SrcMMO, Offset: CurrOffset, Size: CopyTy.getSizeInBytes()); |
8993 | auto *StoreMMO = |
8994 | MF.getMachineMemOperand(MMO: &DstMMO, Offset: CurrOffset, Size: CopyTy.getSizeInBytes()); |
8995 | |
8996 | // Create the load. |
8997 | Register LoadPtr = Src; |
8998 | Register Offset; |
8999 | if (CurrOffset != 0) { |
9000 | LLT SrcTy = MRI.getType(Reg: Src); |
9001 | Offset = MIB.buildConstant(Res: LLT::scalar(SizeInBits: SrcTy.getSizeInBits()), Val: CurrOffset) |
9002 | .getReg(Idx: 0); |
9003 | LoadPtr = MIB.buildPtrAdd(Res: SrcTy, Op0: Src, Op1: Offset).getReg(Idx: 0); |
9004 | } |
9005 | auto LdVal = MIB.buildLoad(Res: CopyTy, Addr: LoadPtr, MMO&: *LoadMMO); |
9006 | |
9007 | // Create the store. |
9008 | Register StorePtr = Dst; |
9009 | if (CurrOffset != 0) { |
9010 | LLT DstTy = MRI.getType(Reg: Dst); |
9011 | StorePtr = MIB.buildPtrAdd(Res: DstTy, Op0: Dst, Op1: Offset).getReg(Idx: 0); |
9012 | } |
9013 | MIB.buildStore(Val: LdVal, Addr: StorePtr, MMO&: *StoreMMO); |
9014 | CurrOffset += CopyTy.getSizeInBytes(); |
9015 | Size -= CopyTy.getSizeInBytes(); |
9016 | } |
9017 | |
9018 | MI.eraseFromParent(); |
9019 | return Legalized; |
9020 | } |
9021 | |
9022 | LegalizerHelper::LegalizeResult |
9023 | LegalizerHelper::lowerMemmove(MachineInstr &MI, Register Dst, Register Src, |
9024 | uint64_t KnownLen, Align DstAlign, Align SrcAlign, |
9025 | bool IsVolatile) { |
9026 | auto &MF = *MI.getParent()->getParent(); |
9027 | const auto &TLI = *MF.getSubtarget().getTargetLowering(); |
9028 | auto &DL = MF.getDataLayout(); |
9029 | LLVMContext &C = MF.getFunction().getContext(); |
9030 | |
9031 | assert(KnownLen != 0 && "Have a zero length memmove length!" ); |
9032 | |
9033 | bool DstAlignCanChange = false; |
9034 | MachineFrameInfo &MFI = MF.getFrameInfo(); |
9035 | bool OptSize = shouldLowerMemFuncForSize(MF); |
9036 | Align Alignment = std::min(a: DstAlign, b: SrcAlign); |
9037 | |
9038 | MachineInstr *FIDef = getOpcodeDef(Opcode: TargetOpcode::G_FRAME_INDEX, Reg: Dst, MRI); |
9039 | if (FIDef && !MFI.isFixedObjectIndex(ObjectIdx: FIDef->getOperand(i: 1).getIndex())) |
9040 | DstAlignCanChange = true; |
9041 | |
9042 | unsigned Limit = TLI.getMaxStoresPerMemmove(OptSize); |
9043 | std::vector<LLT> MemOps; |
9044 | |
9045 | const auto &DstMMO = **MI.memoperands_begin(); |
9046 | const auto &SrcMMO = **std::next(x: MI.memoperands_begin()); |
9047 | MachinePointerInfo DstPtrInfo = DstMMO.getPointerInfo(); |
9048 | MachinePointerInfo SrcPtrInfo = SrcMMO.getPointerInfo(); |
9049 | |
9050 | // FIXME: SelectionDAG always passes false for 'AllowOverlap', apparently due |
9051 | // to a bug in it's findOptimalMemOpLowering implementation. For now do the |
9052 | // same thing here. |
9053 | if (!findGISelOptimalMemOpLowering( |
9054 | MemOps, Limit, |
9055 | Op: MemOp::Copy(Size: KnownLen, DstAlignCanChange, DstAlign: Alignment, SrcAlign, |
9056 | /*IsVolatile*/ true), |
9057 | DstAS: DstPtrInfo.getAddrSpace(), SrcAS: SrcPtrInfo.getAddrSpace(), |
9058 | FuncAttributes: MF.getFunction().getAttributes(), TLI)) |
9059 | return UnableToLegalize; |
9060 | |
9061 | if (DstAlignCanChange) { |
9062 | // Get an estimate of the type from the LLT. |
9063 | Type *IRTy = getTypeForLLT(Ty: MemOps[0], C); |
9064 | Align NewAlign = DL.getABITypeAlign(Ty: IRTy); |
9065 | |
9066 | // Don't promote to an alignment that would require dynamic stack |
9067 | // realignment. |
9068 | const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo(); |
9069 | if (!TRI->hasStackRealignment(MF)) |
9070 | while (NewAlign > Alignment && DL.exceedsNaturalStackAlignment(Alignment: NewAlign)) |
9071 | NewAlign = NewAlign.previous(); |
9072 | |
9073 | if (NewAlign > Alignment) { |
9074 | Alignment = NewAlign; |
9075 | unsigned FI = FIDef->getOperand(i: 1).getIndex(); |
9076 | // Give the stack frame object a larger alignment if needed. |
9077 | if (MFI.getObjectAlign(ObjectIdx: FI) < Alignment) |
9078 | MFI.setObjectAlignment(ObjectIdx: FI, Alignment); |
9079 | } |
9080 | } |
9081 | |
9082 | LLVM_DEBUG(dbgs() << "Inlining memmove: " << MI << " into loads & stores\n" ); |
9083 | |
9084 | MachineIRBuilder MIB(MI); |
9085 | // Memmove requires that we perform the loads first before issuing the stores. |
9086 | // Apart from that, this loop is pretty much doing the same thing as the |
9087 | // memcpy codegen function. |
9088 | unsigned CurrOffset = 0; |
9089 | SmallVector<Register, 16> LoadVals; |
9090 | for (auto CopyTy : MemOps) { |
9091 | // Construct MMO for the load. |
9092 | auto *LoadMMO = |
9093 | MF.getMachineMemOperand(MMO: &SrcMMO, Offset: CurrOffset, Size: CopyTy.getSizeInBytes()); |
9094 | |
9095 | // Create the load. |
9096 | Register LoadPtr = Src; |
9097 | if (CurrOffset != 0) { |
9098 | LLT SrcTy = MRI.getType(Reg: Src); |
9099 | auto Offset = |
9100 | MIB.buildConstant(Res: LLT::scalar(SizeInBits: SrcTy.getSizeInBits()), Val: CurrOffset); |
9101 | LoadPtr = MIB.buildPtrAdd(Res: SrcTy, Op0: Src, Op1: Offset).getReg(Idx: 0); |
9102 | } |
9103 | LoadVals.push_back(Elt: MIB.buildLoad(Res: CopyTy, Addr: LoadPtr, MMO&: *LoadMMO).getReg(Idx: 0)); |
9104 | CurrOffset += CopyTy.getSizeInBytes(); |
9105 | } |
9106 | |
9107 | CurrOffset = 0; |
9108 | for (unsigned I = 0; I < MemOps.size(); ++I) { |
9109 | LLT CopyTy = MemOps[I]; |
9110 | // Now store the values loaded. |
9111 | auto *StoreMMO = |
9112 | MF.getMachineMemOperand(MMO: &DstMMO, Offset: CurrOffset, Size: CopyTy.getSizeInBytes()); |
9113 | |
9114 | Register StorePtr = Dst; |
9115 | if (CurrOffset != 0) { |
9116 | LLT DstTy = MRI.getType(Reg: Dst); |
9117 | auto Offset = |
9118 | MIB.buildConstant(Res: LLT::scalar(SizeInBits: DstTy.getSizeInBits()), Val: CurrOffset); |
9119 | StorePtr = MIB.buildPtrAdd(Res: DstTy, Op0: Dst, Op1: Offset).getReg(Idx: 0); |
9120 | } |
9121 | MIB.buildStore(Val: LoadVals[I], Addr: StorePtr, MMO&: *StoreMMO); |
9122 | CurrOffset += CopyTy.getSizeInBytes(); |
9123 | } |
9124 | MI.eraseFromParent(); |
9125 | return Legalized; |
9126 | } |
9127 | |
9128 | LegalizerHelper::LegalizeResult |
9129 | LegalizerHelper::lowerMemCpyFamily(MachineInstr &MI, unsigned MaxLen) { |
9130 | const unsigned Opc = MI.getOpcode(); |
9131 | // This combine is fairly complex so it's not written with a separate |
9132 | // matcher function. |
9133 | assert((Opc == TargetOpcode::G_MEMCPY || Opc == TargetOpcode::G_MEMMOVE || |
9134 | Opc == TargetOpcode::G_MEMSET) && |
9135 | "Expected memcpy like instruction" ); |
9136 | |
9137 | auto MMOIt = MI.memoperands_begin(); |
9138 | const MachineMemOperand *MemOp = *MMOIt; |
9139 | |
9140 | Align DstAlign = MemOp->getBaseAlign(); |
9141 | Align SrcAlign; |
9142 | auto [Dst, Src, Len] = MI.getFirst3Regs(); |
9143 | |
9144 | if (Opc != TargetOpcode::G_MEMSET) { |
9145 | assert(MMOIt != MI.memoperands_end() && "Expected a second MMO on MI" ); |
9146 | MemOp = *(++MMOIt); |
9147 | SrcAlign = MemOp->getBaseAlign(); |
9148 | } |
9149 | |
9150 | // See if this is a constant length copy |
9151 | auto LenVRegAndVal = getIConstantVRegValWithLookThrough(VReg: Len, MRI); |
9152 | if (!LenVRegAndVal) |
9153 | return UnableToLegalize; |
9154 | uint64_t KnownLen = LenVRegAndVal->Value.getZExtValue(); |
9155 | |
9156 | if (KnownLen == 0) { |
9157 | MI.eraseFromParent(); |
9158 | return Legalized; |
9159 | } |
9160 | |
9161 | bool IsVolatile = MemOp->isVolatile(); |
9162 | if (Opc == TargetOpcode::G_MEMCPY_INLINE) |
9163 | return lowerMemcpyInline(MI, Dst, Src, KnownLen, DstAlign, SrcAlign, |
9164 | IsVolatile); |
9165 | |
9166 | // Don't try to optimize volatile. |
9167 | if (IsVolatile) |
9168 | return UnableToLegalize; |
9169 | |
9170 | if (MaxLen && KnownLen > MaxLen) |
9171 | return UnableToLegalize; |
9172 | |
9173 | if (Opc == TargetOpcode::G_MEMCPY) { |
9174 | auto &MF = *MI.getParent()->getParent(); |
9175 | const auto &TLI = *MF.getSubtarget().getTargetLowering(); |
9176 | bool OptSize = shouldLowerMemFuncForSize(MF); |
9177 | uint64_t Limit = TLI.getMaxStoresPerMemcpy(OptSize); |
9178 | return lowerMemcpy(MI, Dst, Src, KnownLen, Limit, DstAlign, SrcAlign, |
9179 | IsVolatile); |
9180 | } |
9181 | if (Opc == TargetOpcode::G_MEMMOVE) |
9182 | return lowerMemmove(MI, Dst, Src, KnownLen, DstAlign, SrcAlign, IsVolatile); |
9183 | if (Opc == TargetOpcode::G_MEMSET) |
9184 | return lowerMemset(MI, Dst, Val: Src, KnownLen, Alignment: DstAlign, IsVolatile); |
9185 | return UnableToLegalize; |
9186 | } |
9187 | |