1 | //===- MVEGatherScatterLowering.cpp - Gather/Scatter lowering -------------===// |
2 | // |
3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | // See https://llvm.org/LICENSE.txt for license information. |
5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | // |
7 | //===----------------------------------------------------------------------===// |
8 | // |
9 | /// This pass custom lowers llvm.gather and llvm.scatter instructions to |
10 | /// arm.mve.gather and arm.mve.scatter intrinsics, optimising the code to |
11 | /// produce a better final result as we go. |
12 | // |
13 | //===----------------------------------------------------------------------===// |
14 | |
15 | #include "ARM.h" |
16 | #include "ARMBaseInstrInfo.h" |
17 | #include "ARMSubtarget.h" |
18 | #include "llvm/Analysis/LoopInfo.h" |
19 | #include "llvm/Analysis/TargetTransformInfo.h" |
20 | #include "llvm/Analysis/ValueTracking.h" |
21 | #include "llvm/CodeGen/TargetLowering.h" |
22 | #include "llvm/CodeGen/TargetPassConfig.h" |
23 | #include "llvm/CodeGen/TargetSubtargetInfo.h" |
24 | #include "llvm/InitializePasses.h" |
25 | #include "llvm/IR/BasicBlock.h" |
26 | #include "llvm/IR/Constant.h" |
27 | #include "llvm/IR/Constants.h" |
28 | #include "llvm/IR/DerivedTypes.h" |
29 | #include "llvm/IR/Function.h" |
30 | #include "llvm/IR/InstrTypes.h" |
31 | #include "llvm/IR/Instruction.h" |
32 | #include "llvm/IR/Instructions.h" |
33 | #include "llvm/IR/IntrinsicInst.h" |
34 | #include "llvm/IR/Intrinsics.h" |
35 | #include "llvm/IR/IntrinsicsARM.h" |
36 | #include "llvm/IR/IRBuilder.h" |
37 | #include "llvm/IR/PatternMatch.h" |
38 | #include "llvm/IR/Type.h" |
39 | #include "llvm/IR/Value.h" |
40 | #include "llvm/Pass.h" |
41 | #include "llvm/Support/Casting.h" |
42 | #include "llvm/Transforms/Utils/Local.h" |
43 | #include <algorithm> |
44 | #include <cassert> |
45 | |
46 | using namespace llvm; |
47 | |
48 | #define DEBUG_TYPE "arm-mve-gather-scatter-lowering" |
49 | |
50 | cl::opt<bool> EnableMaskedGatherScatters( |
51 | "enable-arm-maskedgatscat" , cl::Hidden, cl::init(Val: true), |
52 | cl::desc("Enable the generation of masked gathers and scatters" )); |
53 | |
54 | namespace { |
55 | |
56 | class MVEGatherScatterLowering : public FunctionPass { |
57 | public: |
58 | static char ID; // Pass identification, replacement for typeid |
59 | |
60 | explicit MVEGatherScatterLowering() : FunctionPass(ID) { |
61 | initializeMVEGatherScatterLoweringPass(*PassRegistry::getPassRegistry()); |
62 | } |
63 | |
64 | bool runOnFunction(Function &F) override; |
65 | |
66 | StringRef getPassName() const override { |
67 | return "MVE gather/scatter lowering" ; |
68 | } |
69 | |
70 | void getAnalysisUsage(AnalysisUsage &AU) const override { |
71 | AU.setPreservesCFG(); |
72 | AU.addRequired<TargetPassConfig>(); |
73 | AU.addRequired<LoopInfoWrapperPass>(); |
74 | FunctionPass::getAnalysisUsage(AU); |
75 | } |
76 | |
77 | private: |
78 | LoopInfo *LI = nullptr; |
79 | const DataLayout *DL; |
80 | |
81 | // Check this is a valid gather with correct alignment |
82 | bool isLegalTypeAndAlignment(unsigned NumElements, unsigned ElemSize, |
83 | Align Alignment); |
84 | // Check whether Ptr is hidden behind a bitcast and look through it |
85 | void lookThroughBitcast(Value *&Ptr); |
86 | // Decompose a ptr into Base and Offsets, potentially using a GEP to return a |
87 | // scalar base and vector offsets, or else fallback to using a base of 0 and |
88 | // offset of Ptr where possible. |
89 | Value *decomposePtr(Value *Ptr, Value *&Offsets, int &Scale, |
90 | FixedVectorType *Ty, Type *MemoryTy, |
91 | IRBuilder<> &Builder); |
92 | // Check for a getelementptr and deduce base and offsets from it, on success |
93 | // returning the base directly and the offsets indirectly using the Offsets |
94 | // argument |
95 | Value *decomposeGEP(Value *&Offsets, FixedVectorType *Ty, |
96 | GetElementPtrInst *GEP, IRBuilder<> &Builder); |
97 | // Compute the scale of this gather/scatter instruction |
98 | int computeScale(unsigned GEPElemSize, unsigned MemoryElemSize); |
99 | // If the value is a constant, or derived from constants via additions |
100 | // and multilications, return its numeric value |
101 | std::optional<int64_t> getIfConst(const Value *V); |
102 | // If Inst is an add instruction, check whether one summand is a |
103 | // constant. If so, scale this constant and return it together with |
104 | // the other summand. |
105 | std::pair<Value *, int64_t> getVarAndConst(Value *Inst, int TypeScale); |
106 | |
107 | Instruction *lowerGather(IntrinsicInst *I); |
108 | // Create a gather from a base + vector of offsets |
109 | Instruction *tryCreateMaskedGatherOffset(IntrinsicInst *I, Value *Ptr, |
110 | Instruction *&Root, |
111 | IRBuilder<> &Builder); |
112 | // Create a gather from a vector of pointers |
113 | Instruction *tryCreateMaskedGatherBase(IntrinsicInst *I, Value *Ptr, |
114 | IRBuilder<> &Builder, |
115 | int64_t Increment = 0); |
116 | // Create an incrementing gather from a vector of pointers |
117 | Instruction *tryCreateMaskedGatherBaseWB(IntrinsicInst *I, Value *Ptr, |
118 | IRBuilder<> &Builder, |
119 | int64_t Increment = 0); |
120 | |
121 | Instruction *lowerScatter(IntrinsicInst *I); |
122 | // Create a scatter to a base + vector of offsets |
123 | Instruction *tryCreateMaskedScatterOffset(IntrinsicInst *I, Value *Offsets, |
124 | IRBuilder<> &Builder); |
125 | // Create a scatter to a vector of pointers |
126 | Instruction *tryCreateMaskedScatterBase(IntrinsicInst *I, Value *Ptr, |
127 | IRBuilder<> &Builder, |
128 | int64_t Increment = 0); |
129 | // Create an incrementing scatter from a vector of pointers |
130 | Instruction *tryCreateMaskedScatterBaseWB(IntrinsicInst *I, Value *Ptr, |
131 | IRBuilder<> &Builder, |
132 | int64_t Increment = 0); |
133 | |
134 | // QI gathers and scatters can increment their offsets on their own if |
135 | // the increment is a constant value (digit) |
136 | Instruction *tryCreateIncrementingGatScat(IntrinsicInst *I, Value *Ptr, |
137 | IRBuilder<> &Builder); |
138 | // QI gathers/scatters can increment their offsets on their own if the |
139 | // increment is a constant value (digit) - this creates a writeback QI |
140 | // gather/scatter |
141 | Instruction *tryCreateIncrementingWBGatScat(IntrinsicInst *I, Value *BasePtr, |
142 | Value *Ptr, unsigned TypeScale, |
143 | IRBuilder<> &Builder); |
144 | |
145 | // Optimise the base and offsets of the given address |
146 | bool optimiseAddress(Value *Address, BasicBlock *BB, LoopInfo *LI); |
147 | // Try to fold consecutive geps together into one |
148 | Value *foldGEP(GetElementPtrInst *GEP, Value *&Offsets, unsigned &Scale, |
149 | IRBuilder<> &Builder); |
150 | // Check whether these offsets could be moved out of the loop they're in |
151 | bool optimiseOffsets(Value *Offsets, BasicBlock *BB, LoopInfo *LI); |
152 | // Pushes the given add out of the loop |
153 | void pushOutAdd(PHINode *&Phi, Value *OffsSecondOperand, unsigned StartIndex); |
154 | // Pushes the given mul or shl out of the loop |
155 | void pushOutMulShl(unsigned Opc, PHINode *&Phi, Value *IncrementPerRound, |
156 | Value *OffsSecondOperand, unsigned LoopIncrement, |
157 | IRBuilder<> &Builder); |
158 | }; |
159 | |
160 | } // end anonymous namespace |
161 | |
162 | char MVEGatherScatterLowering::ID = 0; |
163 | |
164 | INITIALIZE_PASS(MVEGatherScatterLowering, DEBUG_TYPE, |
165 | "MVE gather/scattering lowering pass" , false, false) |
166 | |
167 | Pass *llvm::createMVEGatherScatterLoweringPass() { |
168 | return new MVEGatherScatterLowering(); |
169 | } |
170 | |
171 | bool MVEGatherScatterLowering::isLegalTypeAndAlignment(unsigned NumElements, |
172 | unsigned ElemSize, |
173 | Align Alignment) { |
174 | if (((NumElements == 4 && |
175 | (ElemSize == 32 || ElemSize == 16 || ElemSize == 8)) || |
176 | (NumElements == 8 && (ElemSize == 16 || ElemSize == 8)) || |
177 | (NumElements == 16 && ElemSize == 8)) && |
178 | Alignment >= ElemSize / 8) |
179 | return true; |
180 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: instruction does not have " |
181 | << "valid alignment or vector type \n" ); |
182 | return false; |
183 | } |
184 | |
185 | static bool checkOffsetSize(Value *Offsets, unsigned TargetElemCount) { |
186 | // Offsets that are not of type <N x i32> are sign extended by the |
187 | // getelementptr instruction, and MVE gathers/scatters treat the offset as |
188 | // unsigned. Thus, if the element size is smaller than 32, we can only allow |
189 | // positive offsets - i.e., the offsets are not allowed to be variables we |
190 | // can't look into. |
191 | // Additionally, <N x i32> offsets have to either originate from a zext of a |
192 | // vector with element types smaller or equal the type of the gather we're |
193 | // looking at, or consist of constants that we can check are small enough |
194 | // to fit into the gather type. |
195 | // Thus we check that 0 < value < 2^TargetElemSize. |
196 | unsigned TargetElemSize = 128 / TargetElemCount; |
197 | unsigned OffsetElemSize = cast<FixedVectorType>(Val: Offsets->getType()) |
198 | ->getElementType() |
199 | ->getScalarSizeInBits(); |
200 | if (OffsetElemSize != TargetElemSize || OffsetElemSize != 32) { |
201 | Constant *ConstOff = dyn_cast<Constant>(Val: Offsets); |
202 | if (!ConstOff) |
203 | return false; |
204 | int64_t TargetElemMaxSize = (1ULL << TargetElemSize); |
205 | auto CheckValueSize = [TargetElemMaxSize](Value *OffsetElem) { |
206 | ConstantInt *OConst = dyn_cast<ConstantInt>(Val: OffsetElem); |
207 | if (!OConst) |
208 | return false; |
209 | int SExtValue = OConst->getSExtValue(); |
210 | if (SExtValue >= TargetElemMaxSize || SExtValue < 0) |
211 | return false; |
212 | return true; |
213 | }; |
214 | if (isa<FixedVectorType>(Val: ConstOff->getType())) { |
215 | for (unsigned i = 0; i < TargetElemCount; i++) { |
216 | if (!CheckValueSize(ConstOff->getAggregateElement(Elt: i))) |
217 | return false; |
218 | } |
219 | } else { |
220 | if (!CheckValueSize(ConstOff)) |
221 | return false; |
222 | } |
223 | } |
224 | return true; |
225 | } |
226 | |
227 | Value *MVEGatherScatterLowering::decomposePtr(Value *Ptr, Value *&Offsets, |
228 | int &Scale, FixedVectorType *Ty, |
229 | Type *MemoryTy, |
230 | IRBuilder<> &Builder) { |
231 | if (auto *GEP = dyn_cast<GetElementPtrInst>(Val: Ptr)) { |
232 | if (Value *V = decomposeGEP(Offsets, Ty, GEP, Builder)) { |
233 | Scale = |
234 | computeScale(GEPElemSize: GEP->getSourceElementType()->getPrimitiveSizeInBits(), |
235 | MemoryElemSize: MemoryTy->getScalarSizeInBits()); |
236 | return Scale == -1 ? nullptr : V; |
237 | } |
238 | } |
239 | |
240 | // If we couldn't use the GEP (or it doesn't exist), attempt to use a |
241 | // BasePtr of 0 with Ptr as the Offsets, so long as there are only 4 |
242 | // elements. |
243 | FixedVectorType *PtrTy = cast<FixedVectorType>(Val: Ptr->getType()); |
244 | if (PtrTy->getNumElements() != 4 || MemoryTy->getScalarSizeInBits() == 32) |
245 | return nullptr; |
246 | Value *Zero = ConstantInt::get(Ty: Builder.getInt32Ty(), V: 0); |
247 | Value *BasePtr = Builder.CreateIntToPtr(V: Zero, DestTy: Builder.getPtrTy()); |
248 | Offsets = Builder.CreatePtrToInt( |
249 | V: Ptr, DestTy: FixedVectorType::get(ElementType: Builder.getInt32Ty(), NumElts: 4)); |
250 | Scale = 0; |
251 | return BasePtr; |
252 | } |
253 | |
254 | Value *MVEGatherScatterLowering::decomposeGEP(Value *&Offsets, |
255 | FixedVectorType *Ty, |
256 | GetElementPtrInst *GEP, |
257 | IRBuilder<> &Builder) { |
258 | if (!GEP) { |
259 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: no getelementpointer " |
260 | << "found\n" ); |
261 | return nullptr; |
262 | } |
263 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: getelementpointer found." |
264 | << " Looking at intrinsic for base + vector of offsets\n" ); |
265 | Value *GEPPtr = GEP->getPointerOperand(); |
266 | Offsets = GEP->getOperand(i_nocapture: 1); |
267 | if (GEPPtr->getType()->isVectorTy() || |
268 | !isa<FixedVectorType>(Val: Offsets->getType())) |
269 | return nullptr; |
270 | |
271 | if (GEP->getNumOperands() != 2) { |
272 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: getelementptr with too many" |
273 | << " operands. Expanding.\n" ); |
274 | return nullptr; |
275 | } |
276 | Offsets = GEP->getOperand(i_nocapture: 1); |
277 | unsigned OffsetsElemCount = |
278 | cast<FixedVectorType>(Val: Offsets->getType())->getNumElements(); |
279 | // Paranoid check whether the number of parallel lanes is the same |
280 | assert(Ty->getNumElements() == OffsetsElemCount); |
281 | |
282 | ZExtInst *ZextOffs = dyn_cast<ZExtInst>(Val: Offsets); |
283 | if (ZextOffs) |
284 | Offsets = ZextOffs->getOperand(i_nocapture: 0); |
285 | FixedVectorType *OffsetType = cast<FixedVectorType>(Val: Offsets->getType()); |
286 | |
287 | // If the offsets are already being zext-ed to <N x i32>, that relieves us of |
288 | // having to make sure that they won't overflow. |
289 | if (!ZextOffs || cast<FixedVectorType>(Val: ZextOffs->getDestTy()) |
290 | ->getElementType() |
291 | ->getScalarSizeInBits() != 32) |
292 | if (!checkOffsetSize(Offsets, TargetElemCount: OffsetsElemCount)) |
293 | return nullptr; |
294 | |
295 | // The offset sizes have been checked; if any truncating or zext-ing is |
296 | // required to fix them, do that now |
297 | if (Ty != Offsets->getType()) { |
298 | if ((Ty->getElementType()->getScalarSizeInBits() < |
299 | OffsetType->getElementType()->getScalarSizeInBits())) { |
300 | Offsets = Builder.CreateTrunc(V: Offsets, DestTy: Ty); |
301 | } else { |
302 | Offsets = Builder.CreateZExt(V: Offsets, DestTy: VectorType::getInteger(VTy: Ty)); |
303 | } |
304 | } |
305 | // If none of the checks failed, return the gep's base pointer |
306 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: found correct offsets\n" ); |
307 | return GEPPtr; |
308 | } |
309 | |
310 | void MVEGatherScatterLowering::lookThroughBitcast(Value *&Ptr) { |
311 | // Look through bitcast instruction if #elements is the same |
312 | if (auto *BitCast = dyn_cast<BitCastInst>(Val: Ptr)) { |
313 | auto *BCTy = cast<FixedVectorType>(Val: BitCast->getType()); |
314 | auto *BCSrcTy = cast<FixedVectorType>(Val: BitCast->getOperand(i_nocapture: 0)->getType()); |
315 | if (BCTy->getNumElements() == BCSrcTy->getNumElements()) { |
316 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: looking through " |
317 | << "bitcast\n" ); |
318 | Ptr = BitCast->getOperand(i_nocapture: 0); |
319 | } |
320 | } |
321 | } |
322 | |
323 | int MVEGatherScatterLowering::computeScale(unsigned GEPElemSize, |
324 | unsigned MemoryElemSize) { |
325 | // This can be a 32bit load/store scaled by 4, a 16bit load/store scaled by 2, |
326 | // or a 8bit, 16bit or 32bit load/store scaled by 1 |
327 | if (GEPElemSize == 32 && MemoryElemSize == 32) |
328 | return 2; |
329 | else if (GEPElemSize == 16 && MemoryElemSize == 16) |
330 | return 1; |
331 | else if (GEPElemSize == 8) |
332 | return 0; |
333 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: incorrect scale. Can't " |
334 | << "create intrinsic\n" ); |
335 | return -1; |
336 | } |
337 | |
338 | std::optional<int64_t> MVEGatherScatterLowering::getIfConst(const Value *V) { |
339 | const Constant *C = dyn_cast<Constant>(Val: V); |
340 | if (C && C->getSplatValue()) |
341 | return std::optional<int64_t>{C->getUniqueInteger().getSExtValue()}; |
342 | if (!isa<Instruction>(Val: V)) |
343 | return std::optional<int64_t>{}; |
344 | |
345 | const Instruction *I = cast<Instruction>(Val: V); |
346 | if (I->getOpcode() == Instruction::Add || I->getOpcode() == Instruction::Or || |
347 | I->getOpcode() == Instruction::Mul || |
348 | I->getOpcode() == Instruction::Shl) { |
349 | std::optional<int64_t> Op0 = getIfConst(V: I->getOperand(i: 0)); |
350 | std::optional<int64_t> Op1 = getIfConst(V: I->getOperand(i: 1)); |
351 | if (!Op0 || !Op1) |
352 | return std::optional<int64_t>{}; |
353 | if (I->getOpcode() == Instruction::Add) |
354 | return std::optional<int64_t>{*Op0 + *Op1}; |
355 | if (I->getOpcode() == Instruction::Mul) |
356 | return std::optional<int64_t>{*Op0 * *Op1}; |
357 | if (I->getOpcode() == Instruction::Shl) |
358 | return std::optional<int64_t>{*Op0 << *Op1}; |
359 | if (I->getOpcode() == Instruction::Or) |
360 | return std::optional<int64_t>{*Op0 | *Op1}; |
361 | } |
362 | return std::optional<int64_t>{}; |
363 | } |
364 | |
365 | // Return true if I is an Or instruction that is equivalent to an add, due to |
366 | // the operands having no common bits set. |
367 | static bool isAddLikeOr(Instruction *I, const DataLayout &DL) { |
368 | return I->getOpcode() == Instruction::Or && |
369 | haveNoCommonBitsSet(LHSCache: I->getOperand(i: 0), RHSCache: I->getOperand(i: 1), SQ: DL); |
370 | } |
371 | |
372 | std::pair<Value *, int64_t> |
373 | MVEGatherScatterLowering::getVarAndConst(Value *Inst, int TypeScale) { |
374 | std::pair<Value *, int64_t> ReturnFalse = |
375 | std::pair<Value *, int64_t>(nullptr, 0); |
376 | // At this point, the instruction we're looking at must be an add or an |
377 | // add-like-or. |
378 | Instruction *Add = dyn_cast<Instruction>(Val: Inst); |
379 | if (Add == nullptr || |
380 | (Add->getOpcode() != Instruction::Add && !isAddLikeOr(I: Add, DL: *DL))) |
381 | return ReturnFalse; |
382 | |
383 | Value *Summand; |
384 | std::optional<int64_t> Const; |
385 | // Find out which operand the value that is increased is |
386 | if ((Const = getIfConst(V: Add->getOperand(i: 0)))) |
387 | Summand = Add->getOperand(i: 1); |
388 | else if ((Const = getIfConst(V: Add->getOperand(i: 1)))) |
389 | Summand = Add->getOperand(i: 0); |
390 | else |
391 | return ReturnFalse; |
392 | |
393 | // Check that the constant is small enough for an incrementing gather |
394 | int64_t Immediate = *Const << TypeScale; |
395 | if (Immediate > 512 || Immediate < -512 || Immediate % 4 != 0) |
396 | return ReturnFalse; |
397 | |
398 | return std::pair<Value *, int64_t>(Summand, Immediate); |
399 | } |
400 | |
401 | Instruction *MVEGatherScatterLowering::lowerGather(IntrinsicInst *I) { |
402 | using namespace PatternMatch; |
403 | LLVM_DEBUG(dbgs() << "masked gathers: checking transform preconditions\n" |
404 | << *I << "\n" ); |
405 | |
406 | // @llvm.masked.gather.*(Ptrs, alignment, Mask, Src0) |
407 | // Attempt to turn the masked gather in I into a MVE intrinsic |
408 | // Potentially optimising the addressing modes as we do so. |
409 | auto *Ty = cast<FixedVectorType>(Val: I->getType()); |
410 | Value *Ptr = I->getArgOperand(i: 0); |
411 | Align Alignment = cast<ConstantInt>(Val: I->getArgOperand(i: 1))->getAlignValue(); |
412 | Value *Mask = I->getArgOperand(i: 2); |
413 | Value *PassThru = I->getArgOperand(i: 3); |
414 | |
415 | if (!isLegalTypeAndAlignment(NumElements: Ty->getNumElements(), ElemSize: Ty->getScalarSizeInBits(), |
416 | Alignment)) |
417 | return nullptr; |
418 | lookThroughBitcast(Ptr); |
419 | assert(Ptr->getType()->isVectorTy() && "Unexpected pointer type" ); |
420 | |
421 | IRBuilder<> Builder(I->getContext()); |
422 | Builder.SetInsertPoint(I); |
423 | Builder.SetCurrentDebugLocation(I->getDebugLoc()); |
424 | |
425 | Instruction *Root = I; |
426 | |
427 | Instruction *Load = tryCreateIncrementingGatScat(I, Ptr, Builder); |
428 | if (!Load) |
429 | Load = tryCreateMaskedGatherOffset(I, Ptr, Root, Builder); |
430 | if (!Load) |
431 | Load = tryCreateMaskedGatherBase(I, Ptr, Builder); |
432 | if (!Load) |
433 | return nullptr; |
434 | |
435 | if (!isa<UndefValue>(Val: PassThru) && !match(V: PassThru, P: m_Zero())) { |
436 | LLVM_DEBUG(dbgs() << "masked gathers: found non-trivial passthru - " |
437 | << "creating select\n" ); |
438 | Load = SelectInst::Create(C: Mask, S1: Load, S2: PassThru); |
439 | Builder.Insert(I: Load); |
440 | } |
441 | |
442 | Root->replaceAllUsesWith(V: Load); |
443 | Root->eraseFromParent(); |
444 | if (Root != I) |
445 | // If this was an extending gather, we need to get rid of the sext/zext |
446 | // sext/zext as well as of the gather itself |
447 | I->eraseFromParent(); |
448 | |
449 | LLVM_DEBUG(dbgs() << "masked gathers: successfully built masked gather\n" |
450 | << *Load << "\n" ); |
451 | return Load; |
452 | } |
453 | |
454 | Instruction *MVEGatherScatterLowering::tryCreateMaskedGatherBase( |
455 | IntrinsicInst *I, Value *Ptr, IRBuilder<> &Builder, int64_t Increment) { |
456 | using namespace PatternMatch; |
457 | auto *Ty = cast<FixedVectorType>(Val: I->getType()); |
458 | LLVM_DEBUG(dbgs() << "masked gathers: loading from vector of pointers\n" ); |
459 | if (Ty->getNumElements() != 4 || Ty->getScalarSizeInBits() != 32) |
460 | // Can't build an intrinsic for this |
461 | return nullptr; |
462 | Value *Mask = I->getArgOperand(i: 2); |
463 | if (match(V: Mask, P: m_One())) |
464 | return Builder.CreateIntrinsic(ID: Intrinsic::arm_mve_vldr_gather_base, |
465 | Types: {Ty, Ptr->getType()}, |
466 | Args: {Ptr, Builder.getInt32(C: Increment)}); |
467 | else |
468 | return Builder.CreateIntrinsic( |
469 | ID: Intrinsic::arm_mve_vldr_gather_base_predicated, |
470 | Types: {Ty, Ptr->getType(), Mask->getType()}, |
471 | Args: {Ptr, Builder.getInt32(C: Increment), Mask}); |
472 | } |
473 | |
474 | Instruction *MVEGatherScatterLowering::tryCreateMaskedGatherBaseWB( |
475 | IntrinsicInst *I, Value *Ptr, IRBuilder<> &Builder, int64_t Increment) { |
476 | using namespace PatternMatch; |
477 | auto *Ty = cast<FixedVectorType>(Val: I->getType()); |
478 | LLVM_DEBUG(dbgs() << "masked gathers: loading from vector of pointers with " |
479 | << "writeback\n" ); |
480 | if (Ty->getNumElements() != 4 || Ty->getScalarSizeInBits() != 32) |
481 | // Can't build an intrinsic for this |
482 | return nullptr; |
483 | Value *Mask = I->getArgOperand(i: 2); |
484 | if (match(V: Mask, P: m_One())) |
485 | return Builder.CreateIntrinsic(ID: Intrinsic::arm_mve_vldr_gather_base_wb, |
486 | Types: {Ty, Ptr->getType()}, |
487 | Args: {Ptr, Builder.getInt32(C: Increment)}); |
488 | else |
489 | return Builder.CreateIntrinsic( |
490 | ID: Intrinsic::arm_mve_vldr_gather_base_wb_predicated, |
491 | Types: {Ty, Ptr->getType(), Mask->getType()}, |
492 | Args: {Ptr, Builder.getInt32(C: Increment), Mask}); |
493 | } |
494 | |
495 | Instruction *MVEGatherScatterLowering::tryCreateMaskedGatherOffset( |
496 | IntrinsicInst *I, Value *Ptr, Instruction *&Root, IRBuilder<> &Builder) { |
497 | using namespace PatternMatch; |
498 | |
499 | Type *MemoryTy = I->getType(); |
500 | Type *ResultTy = MemoryTy; |
501 | |
502 | unsigned Unsigned = 1; |
503 | // The size of the gather was already checked in isLegalTypeAndAlignment; |
504 | // if it was not a full vector width an appropriate extend should follow. |
505 | auto *Extend = Root; |
506 | bool TruncResult = false; |
507 | if (MemoryTy->getPrimitiveSizeInBits() < 128) { |
508 | if (I->hasOneUse()) { |
509 | // If the gather has a single extend of the correct type, use an extending |
510 | // gather and replace the ext. In which case the correct root to replace |
511 | // is not the CallInst itself, but the instruction which extends it. |
512 | Instruction* User = cast<Instruction>(Val: *I->users().begin()); |
513 | if (isa<SExtInst>(Val: User) && |
514 | User->getType()->getPrimitiveSizeInBits() == 128) { |
515 | LLVM_DEBUG(dbgs() << "masked gathers: Incorporating extend: " |
516 | << *User << "\n" ); |
517 | Extend = User; |
518 | ResultTy = User->getType(); |
519 | Unsigned = 0; |
520 | } else if (isa<ZExtInst>(Val: User) && |
521 | User->getType()->getPrimitiveSizeInBits() == 128) { |
522 | LLVM_DEBUG(dbgs() << "masked gathers: Incorporating extend: " |
523 | << *ResultTy << "\n" ); |
524 | Extend = User; |
525 | ResultTy = User->getType(); |
526 | } |
527 | } |
528 | |
529 | // If an extend hasn't been found and the type is an integer, create an |
530 | // extending gather and truncate back to the original type. |
531 | if (ResultTy->getPrimitiveSizeInBits() < 128 && |
532 | ResultTy->isIntOrIntVectorTy()) { |
533 | ResultTy = ResultTy->getWithNewBitWidth( |
534 | NewBitWidth: 128 / cast<FixedVectorType>(Val: ResultTy)->getNumElements()); |
535 | TruncResult = true; |
536 | LLVM_DEBUG(dbgs() << "masked gathers: Small input type, truncing to: " |
537 | << *ResultTy << "\n" ); |
538 | } |
539 | |
540 | // The final size of the gather must be a full vector width |
541 | if (ResultTy->getPrimitiveSizeInBits() != 128) { |
542 | LLVM_DEBUG(dbgs() << "masked gathers: Extend needed but not provided " |
543 | "from the correct type. Expanding\n" ); |
544 | return nullptr; |
545 | } |
546 | } |
547 | |
548 | Value *Offsets; |
549 | int Scale; |
550 | Value *BasePtr = decomposePtr( |
551 | Ptr, Offsets, Scale, Ty: cast<FixedVectorType>(Val: ResultTy), MemoryTy, Builder); |
552 | if (!BasePtr) |
553 | return nullptr; |
554 | |
555 | Root = Extend; |
556 | Value *Mask = I->getArgOperand(i: 2); |
557 | Instruction *Load = nullptr; |
558 | if (!match(V: Mask, P: m_One())) |
559 | Load = Builder.CreateIntrinsic( |
560 | ID: Intrinsic::arm_mve_vldr_gather_offset_predicated, |
561 | Types: {ResultTy, BasePtr->getType(), Offsets->getType(), Mask->getType()}, |
562 | Args: {BasePtr, Offsets, Builder.getInt32(C: MemoryTy->getScalarSizeInBits()), |
563 | Builder.getInt32(C: Scale), Builder.getInt32(C: Unsigned), Mask}); |
564 | else |
565 | Load = Builder.CreateIntrinsic( |
566 | ID: Intrinsic::arm_mve_vldr_gather_offset, |
567 | Types: {ResultTy, BasePtr->getType(), Offsets->getType()}, |
568 | Args: {BasePtr, Offsets, Builder.getInt32(C: MemoryTy->getScalarSizeInBits()), |
569 | Builder.getInt32(C: Scale), Builder.getInt32(C: Unsigned)}); |
570 | |
571 | if (TruncResult) { |
572 | Load = TruncInst::Create(Instruction::Trunc, S: Load, Ty: MemoryTy); |
573 | Builder.Insert(I: Load); |
574 | } |
575 | return Load; |
576 | } |
577 | |
578 | Instruction *MVEGatherScatterLowering::lowerScatter(IntrinsicInst *I) { |
579 | using namespace PatternMatch; |
580 | LLVM_DEBUG(dbgs() << "masked scatters: checking transform preconditions\n" |
581 | << *I << "\n" ); |
582 | |
583 | // @llvm.masked.scatter.*(data, ptrs, alignment, mask) |
584 | // Attempt to turn the masked scatter in I into a MVE intrinsic |
585 | // Potentially optimising the addressing modes as we do so. |
586 | Value *Input = I->getArgOperand(i: 0); |
587 | Value *Ptr = I->getArgOperand(i: 1); |
588 | Align Alignment = cast<ConstantInt>(Val: I->getArgOperand(i: 2))->getAlignValue(); |
589 | auto *Ty = cast<FixedVectorType>(Val: Input->getType()); |
590 | |
591 | if (!isLegalTypeAndAlignment(NumElements: Ty->getNumElements(), ElemSize: Ty->getScalarSizeInBits(), |
592 | Alignment)) |
593 | return nullptr; |
594 | |
595 | lookThroughBitcast(Ptr); |
596 | assert(Ptr->getType()->isVectorTy() && "Unexpected pointer type" ); |
597 | |
598 | IRBuilder<> Builder(I->getContext()); |
599 | Builder.SetInsertPoint(I); |
600 | Builder.SetCurrentDebugLocation(I->getDebugLoc()); |
601 | |
602 | Instruction *Store = tryCreateIncrementingGatScat(I, Ptr, Builder); |
603 | if (!Store) |
604 | Store = tryCreateMaskedScatterOffset(I, Offsets: Ptr, Builder); |
605 | if (!Store) |
606 | Store = tryCreateMaskedScatterBase(I, Ptr, Builder); |
607 | if (!Store) |
608 | return nullptr; |
609 | |
610 | LLVM_DEBUG(dbgs() << "masked scatters: successfully built masked scatter\n" |
611 | << *Store << "\n" ); |
612 | I->eraseFromParent(); |
613 | return Store; |
614 | } |
615 | |
616 | Instruction *MVEGatherScatterLowering::tryCreateMaskedScatterBase( |
617 | IntrinsicInst *I, Value *Ptr, IRBuilder<> &Builder, int64_t Increment) { |
618 | using namespace PatternMatch; |
619 | Value *Input = I->getArgOperand(i: 0); |
620 | auto *Ty = cast<FixedVectorType>(Val: Input->getType()); |
621 | // Only QR variants allow truncating |
622 | if (!(Ty->getNumElements() == 4 && Ty->getScalarSizeInBits() == 32)) { |
623 | // Can't build an intrinsic for this |
624 | return nullptr; |
625 | } |
626 | Value *Mask = I->getArgOperand(i: 3); |
627 | // int_arm_mve_vstr_scatter_base(_predicated) addr, offset, data(, mask) |
628 | LLVM_DEBUG(dbgs() << "masked scatters: storing to a vector of pointers\n" ); |
629 | if (match(V: Mask, P: m_One())) |
630 | return Builder.CreateIntrinsic(ID: Intrinsic::arm_mve_vstr_scatter_base, |
631 | Types: {Ptr->getType(), Input->getType()}, |
632 | Args: {Ptr, Builder.getInt32(C: Increment), Input}); |
633 | else |
634 | return Builder.CreateIntrinsic( |
635 | ID: Intrinsic::arm_mve_vstr_scatter_base_predicated, |
636 | Types: {Ptr->getType(), Input->getType(), Mask->getType()}, |
637 | Args: {Ptr, Builder.getInt32(C: Increment), Input, Mask}); |
638 | } |
639 | |
640 | Instruction *MVEGatherScatterLowering::tryCreateMaskedScatterBaseWB( |
641 | IntrinsicInst *I, Value *Ptr, IRBuilder<> &Builder, int64_t Increment) { |
642 | using namespace PatternMatch; |
643 | Value *Input = I->getArgOperand(i: 0); |
644 | auto *Ty = cast<FixedVectorType>(Val: Input->getType()); |
645 | LLVM_DEBUG(dbgs() << "masked scatters: storing to a vector of pointers " |
646 | << "with writeback\n" ); |
647 | if (Ty->getNumElements() != 4 || Ty->getScalarSizeInBits() != 32) |
648 | // Can't build an intrinsic for this |
649 | return nullptr; |
650 | Value *Mask = I->getArgOperand(i: 3); |
651 | if (match(V: Mask, P: m_One())) |
652 | return Builder.CreateIntrinsic(ID: Intrinsic::arm_mve_vstr_scatter_base_wb, |
653 | Types: {Ptr->getType(), Input->getType()}, |
654 | Args: {Ptr, Builder.getInt32(C: Increment), Input}); |
655 | else |
656 | return Builder.CreateIntrinsic( |
657 | ID: Intrinsic::arm_mve_vstr_scatter_base_wb_predicated, |
658 | Types: {Ptr->getType(), Input->getType(), Mask->getType()}, |
659 | Args: {Ptr, Builder.getInt32(C: Increment), Input, Mask}); |
660 | } |
661 | |
662 | Instruction *MVEGatherScatterLowering::tryCreateMaskedScatterOffset( |
663 | IntrinsicInst *I, Value *Ptr, IRBuilder<> &Builder) { |
664 | using namespace PatternMatch; |
665 | Value *Input = I->getArgOperand(i: 0); |
666 | Value *Mask = I->getArgOperand(i: 3); |
667 | Type *InputTy = Input->getType(); |
668 | Type *MemoryTy = InputTy; |
669 | |
670 | LLVM_DEBUG(dbgs() << "masked scatters: getelementpointer found. Storing" |
671 | << " to base + vector of offsets\n" ); |
672 | // If the input has been truncated, try to integrate that trunc into the |
673 | // scatter instruction (we don't care about alignment here) |
674 | if (TruncInst *Trunc = dyn_cast<TruncInst>(Val: Input)) { |
675 | Value *PreTrunc = Trunc->getOperand(i_nocapture: 0); |
676 | Type *PreTruncTy = PreTrunc->getType(); |
677 | if (PreTruncTy->getPrimitiveSizeInBits() == 128) { |
678 | Input = PreTrunc; |
679 | InputTy = PreTruncTy; |
680 | } |
681 | } |
682 | bool ExtendInput = false; |
683 | if (InputTy->getPrimitiveSizeInBits() < 128 && |
684 | InputTy->isIntOrIntVectorTy()) { |
685 | // If we can't find a trunc to incorporate into the instruction, create an |
686 | // implicit one with a zext, so that we can still create a scatter. We know |
687 | // that the input type is 4x/8x/16x and of type i8/i16/i32, so any type |
688 | // smaller than 128 bits will divide evenly into a 128bit vector. |
689 | InputTy = InputTy->getWithNewBitWidth( |
690 | NewBitWidth: 128 / cast<FixedVectorType>(Val: InputTy)->getNumElements()); |
691 | ExtendInput = true; |
692 | LLVM_DEBUG(dbgs() << "masked scatters: Small input type, will extend:\n" |
693 | << *Input << "\n" ); |
694 | } |
695 | if (InputTy->getPrimitiveSizeInBits() != 128) { |
696 | LLVM_DEBUG(dbgs() << "masked scatters: cannot create scatters for " |
697 | "non-standard input types. Expanding.\n" ); |
698 | return nullptr; |
699 | } |
700 | |
701 | Value *Offsets; |
702 | int Scale; |
703 | Value *BasePtr = decomposePtr( |
704 | Ptr, Offsets, Scale, Ty: cast<FixedVectorType>(Val: InputTy), MemoryTy, Builder); |
705 | if (!BasePtr) |
706 | return nullptr; |
707 | |
708 | if (ExtendInput) |
709 | Input = Builder.CreateZExt(V: Input, DestTy: InputTy); |
710 | if (!match(V: Mask, P: m_One())) |
711 | return Builder.CreateIntrinsic( |
712 | ID: Intrinsic::arm_mve_vstr_scatter_offset_predicated, |
713 | Types: {BasePtr->getType(), Offsets->getType(), Input->getType(), |
714 | Mask->getType()}, |
715 | Args: {BasePtr, Offsets, Input, |
716 | Builder.getInt32(C: MemoryTy->getScalarSizeInBits()), |
717 | Builder.getInt32(C: Scale), Mask}); |
718 | else |
719 | return Builder.CreateIntrinsic( |
720 | ID: Intrinsic::arm_mve_vstr_scatter_offset, |
721 | Types: {BasePtr->getType(), Offsets->getType(), Input->getType()}, |
722 | Args: {BasePtr, Offsets, Input, |
723 | Builder.getInt32(C: MemoryTy->getScalarSizeInBits()), |
724 | Builder.getInt32(C: Scale)}); |
725 | } |
726 | |
727 | Instruction *MVEGatherScatterLowering::tryCreateIncrementingGatScat( |
728 | IntrinsicInst *I, Value *Ptr, IRBuilder<> &Builder) { |
729 | FixedVectorType *Ty; |
730 | if (I->getIntrinsicID() == Intrinsic::masked_gather) |
731 | Ty = cast<FixedVectorType>(Val: I->getType()); |
732 | else |
733 | Ty = cast<FixedVectorType>(Val: I->getArgOperand(i: 0)->getType()); |
734 | |
735 | // Incrementing gathers only exist for v4i32 |
736 | if (Ty->getNumElements() != 4 || Ty->getScalarSizeInBits() != 32) |
737 | return nullptr; |
738 | // Incrementing gathers are not beneficial outside of a loop |
739 | Loop *L = LI->getLoopFor(BB: I->getParent()); |
740 | if (L == nullptr) |
741 | return nullptr; |
742 | |
743 | // Decompose the GEP into Base and Offsets |
744 | GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(Val: Ptr); |
745 | Value *Offsets; |
746 | Value *BasePtr = decomposeGEP(Offsets, Ty, GEP, Builder); |
747 | if (!BasePtr) |
748 | return nullptr; |
749 | |
750 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: trying to build incrementing " |
751 | "wb gather/scatter\n" ); |
752 | |
753 | // The gep was in charge of making sure the offsets are scaled correctly |
754 | // - calculate that factor so it can be applied by hand |
755 | int TypeScale = |
756 | computeScale(GEPElemSize: DL->getTypeSizeInBits(Ty: GEP->getOperand(i_nocapture: 0)->getType()), |
757 | MemoryElemSize: DL->getTypeSizeInBits(Ty: GEP->getType()) / |
758 | cast<FixedVectorType>(Val: GEP->getType())->getNumElements()); |
759 | if (TypeScale == -1) |
760 | return nullptr; |
761 | |
762 | if (GEP->hasOneUse()) { |
763 | // Only in this case do we want to build a wb gather, because the wb will |
764 | // change the phi which does affect other users of the gep (which will still |
765 | // be using the phi in the old way) |
766 | if (auto *Load = tryCreateIncrementingWBGatScat(I, BasePtr, Ptr: Offsets, |
767 | TypeScale, Builder)) |
768 | return Load; |
769 | } |
770 | |
771 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: trying to build incrementing " |
772 | "non-wb gather/scatter\n" ); |
773 | |
774 | std::pair<Value *, int64_t> Add = getVarAndConst(Inst: Offsets, TypeScale); |
775 | if (Add.first == nullptr) |
776 | return nullptr; |
777 | Value *OffsetsIncoming = Add.first; |
778 | int64_t Immediate = Add.second; |
779 | |
780 | // Make sure the offsets are scaled correctly |
781 | Instruction *ScaledOffsets = BinaryOperator::Create( |
782 | Op: Instruction::Shl, S1: OffsetsIncoming, |
783 | S2: Builder.CreateVectorSplat(NumElts: Ty->getNumElements(), |
784 | V: Builder.getInt32(C: TypeScale)), |
785 | Name: "ScaledIndex" , InsertBefore: I->getIterator()); |
786 | // Add the base to the offsets |
787 | OffsetsIncoming = BinaryOperator::Create( |
788 | Op: Instruction::Add, S1: ScaledOffsets, |
789 | S2: Builder.CreateVectorSplat( |
790 | NumElts: Ty->getNumElements(), |
791 | V: Builder.CreatePtrToInt( |
792 | V: BasePtr, |
793 | DestTy: cast<VectorType>(Val: ScaledOffsets->getType())->getElementType())), |
794 | Name: "StartIndex" , InsertBefore: I->getIterator()); |
795 | |
796 | if (I->getIntrinsicID() == Intrinsic::masked_gather) |
797 | return tryCreateMaskedGatherBase(I, Ptr: OffsetsIncoming, Builder, Increment: Immediate); |
798 | else |
799 | return tryCreateMaskedScatterBase(I, Ptr: OffsetsIncoming, Builder, Increment: Immediate); |
800 | } |
801 | |
802 | Instruction *MVEGatherScatterLowering::tryCreateIncrementingWBGatScat( |
803 | IntrinsicInst *I, Value *BasePtr, Value *Offsets, unsigned TypeScale, |
804 | IRBuilder<> &Builder) { |
805 | // Check whether this gather's offset is incremented by a constant - if so, |
806 | // and the load is of the right type, we can merge this into a QI gather |
807 | Loop *L = LI->getLoopFor(BB: I->getParent()); |
808 | // Offsets that are worth merging into this instruction will be incremented |
809 | // by a constant, thus we're looking for an add of a phi and a constant |
810 | PHINode *Phi = dyn_cast<PHINode>(Val: Offsets); |
811 | if (Phi == nullptr || Phi->getNumIncomingValues() != 2 || |
812 | Phi->getParent() != L->getHeader() || Phi->getNumUses() != 2) |
813 | // No phi means no IV to write back to; if there is a phi, we expect it |
814 | // to have exactly two incoming values; the only phis we are interested in |
815 | // will be loop IV's and have exactly two uses, one in their increment and |
816 | // one in the gather's gep |
817 | return nullptr; |
818 | |
819 | unsigned IncrementIndex = |
820 | Phi->getIncomingBlock(i: 0) == L->getLoopLatch() ? 0 : 1; |
821 | // Look through the phi to the phi increment |
822 | Offsets = Phi->getIncomingValue(i: IncrementIndex); |
823 | |
824 | std::pair<Value *, int64_t> Add = getVarAndConst(Inst: Offsets, TypeScale); |
825 | if (Add.first == nullptr) |
826 | return nullptr; |
827 | Value *OffsetsIncoming = Add.first; |
828 | int64_t Immediate = Add.second; |
829 | if (OffsetsIncoming != Phi) |
830 | // Then the increment we are looking at is not an increment of the |
831 | // induction variable, and we don't want to do a writeback |
832 | return nullptr; |
833 | |
834 | Builder.SetInsertPoint(&Phi->getIncomingBlock(i: 1 - IncrementIndex)->back()); |
835 | unsigned NumElems = |
836 | cast<FixedVectorType>(Val: OffsetsIncoming->getType())->getNumElements(); |
837 | |
838 | // Make sure the offsets are scaled correctly |
839 | Instruction *ScaledOffsets = BinaryOperator::Create( |
840 | Op: Instruction::Shl, S1: Phi->getIncomingValue(i: 1 - IncrementIndex), |
841 | S2: Builder.CreateVectorSplat(NumElts: NumElems, V: Builder.getInt32(C: TypeScale)), |
842 | Name: "ScaledIndex" , |
843 | InsertBefore: Phi->getIncomingBlock(i: 1 - IncrementIndex)->back().getIterator()); |
844 | // Add the base to the offsets |
845 | OffsetsIncoming = BinaryOperator::Create( |
846 | Op: Instruction::Add, S1: ScaledOffsets, |
847 | S2: Builder.CreateVectorSplat( |
848 | NumElts: NumElems, |
849 | V: Builder.CreatePtrToInt( |
850 | V: BasePtr, |
851 | DestTy: cast<VectorType>(Val: ScaledOffsets->getType())->getElementType())), |
852 | Name: "StartIndex" , |
853 | InsertBefore: Phi->getIncomingBlock(i: 1 - IncrementIndex)->back().getIterator()); |
854 | // The gather is pre-incrementing |
855 | OffsetsIncoming = BinaryOperator::Create( |
856 | Op: Instruction::Sub, S1: OffsetsIncoming, |
857 | S2: Builder.CreateVectorSplat(NumElts: NumElems, V: Builder.getInt32(C: Immediate)), |
858 | Name: "PreIncrementStartIndex" , |
859 | InsertBefore: Phi->getIncomingBlock(i: 1 - IncrementIndex)->back().getIterator()); |
860 | Phi->setIncomingValue(i: 1 - IncrementIndex, V: OffsetsIncoming); |
861 | |
862 | Builder.SetInsertPoint(I); |
863 | |
864 | Instruction *EndResult; |
865 | Instruction *NewInduction; |
866 | if (I->getIntrinsicID() == Intrinsic::masked_gather) { |
867 | // Build the incrementing gather |
868 | Value *Load = tryCreateMaskedGatherBaseWB(I, Ptr: Phi, Builder, Increment: Immediate); |
869 | // One value to be handed to whoever uses the gather, one is the loop |
870 | // increment |
871 | EndResult = ExtractValueInst::Create(Agg: Load, Idxs: 0, NameStr: "Gather" ); |
872 | NewInduction = ExtractValueInst::Create(Agg: Load, Idxs: 1, NameStr: "GatherIncrement" ); |
873 | Builder.Insert(I: EndResult); |
874 | Builder.Insert(I: NewInduction); |
875 | } else { |
876 | // Build the incrementing scatter |
877 | EndResult = NewInduction = |
878 | tryCreateMaskedScatterBaseWB(I, Ptr: Phi, Builder, Increment: Immediate); |
879 | } |
880 | Instruction *AddInst = cast<Instruction>(Val: Offsets); |
881 | AddInst->replaceAllUsesWith(V: NewInduction); |
882 | AddInst->eraseFromParent(); |
883 | Phi->setIncomingValue(i: IncrementIndex, V: NewInduction); |
884 | |
885 | return EndResult; |
886 | } |
887 | |
888 | void MVEGatherScatterLowering::pushOutAdd(PHINode *&Phi, |
889 | Value *OffsSecondOperand, |
890 | unsigned StartIndex) { |
891 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: optimising add instruction\n" ); |
892 | BasicBlock::iterator InsertionPoint = |
893 | Phi->getIncomingBlock(i: StartIndex)->back().getIterator(); |
894 | // Initialize the phi with a vector that contains a sum of the constants |
895 | Instruction *NewIndex = BinaryOperator::Create( |
896 | Op: Instruction::Add, S1: Phi->getIncomingValue(i: StartIndex), S2: OffsSecondOperand, |
897 | Name: "PushedOutAdd" , InsertBefore: InsertionPoint); |
898 | unsigned IncrementIndex = StartIndex == 0 ? 1 : 0; |
899 | |
900 | // Order such that start index comes first (this reduces mov's) |
901 | Phi->addIncoming(V: NewIndex, BB: Phi->getIncomingBlock(i: StartIndex)); |
902 | Phi->addIncoming(V: Phi->getIncomingValue(i: IncrementIndex), |
903 | BB: Phi->getIncomingBlock(i: IncrementIndex)); |
904 | Phi->removeIncomingValue(Idx: 1); |
905 | Phi->removeIncomingValue(Idx: (unsigned)0); |
906 | } |
907 | |
908 | void MVEGatherScatterLowering::pushOutMulShl(unsigned Opcode, PHINode *&Phi, |
909 | Value *IncrementPerRound, |
910 | Value *OffsSecondOperand, |
911 | unsigned LoopIncrement, |
912 | IRBuilder<> &Builder) { |
913 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: optimising mul instruction\n" ); |
914 | |
915 | // Create a new scalar add outside of the loop and transform it to a splat |
916 | // by which loop variable can be incremented |
917 | BasicBlock::iterator InsertionPoint = |
918 | Phi->getIncomingBlock(i: LoopIncrement == 1 ? 0 : 1)->back().getIterator(); |
919 | |
920 | // Create a new index |
921 | Value *StartIndex = |
922 | BinaryOperator::Create(Op: (Instruction::BinaryOps)Opcode, |
923 | S1: Phi->getIncomingValue(i: LoopIncrement == 1 ? 0 : 1), |
924 | S2: OffsSecondOperand, Name: "PushedOutMul" , InsertBefore: InsertionPoint); |
925 | |
926 | Instruction *Product = |
927 | BinaryOperator::Create(Op: (Instruction::BinaryOps)Opcode, S1: IncrementPerRound, |
928 | S2: OffsSecondOperand, Name: "Product" , InsertBefore: InsertionPoint); |
929 | |
930 | BasicBlock::iterator NewIncrInsertPt = |
931 | Phi->getIncomingBlock(i: LoopIncrement)->back().getIterator(); |
932 | NewIncrInsertPt = std::prev(x: NewIncrInsertPt); |
933 | |
934 | // Increment NewIndex by Product instead of the multiplication |
935 | Instruction *NewIncrement = BinaryOperator::Create( |
936 | Op: Instruction::Add, S1: Phi, S2: Product, Name: "IncrementPushedOutMul" , InsertBefore: NewIncrInsertPt); |
937 | |
938 | Phi->addIncoming(V: StartIndex, |
939 | BB: Phi->getIncomingBlock(i: LoopIncrement == 1 ? 0 : 1)); |
940 | Phi->addIncoming(V: NewIncrement, BB: Phi->getIncomingBlock(i: LoopIncrement)); |
941 | Phi->removeIncomingValue(Idx: (unsigned)0); |
942 | Phi->removeIncomingValue(Idx: (unsigned)0); |
943 | } |
944 | |
945 | // Check whether all usages of this instruction are as offsets of |
946 | // gathers/scatters or simple arithmetics only used by gathers/scatters |
947 | static bool hasAllGatScatUsers(Instruction *I, const DataLayout &DL) { |
948 | if (I->hasNUses(N: 0)) { |
949 | return false; |
950 | } |
951 | bool Gatscat = true; |
952 | for (User *U : I->users()) { |
953 | if (!isa<Instruction>(Val: U)) |
954 | return false; |
955 | if (isa<GetElementPtrInst>(Val: U) || |
956 | isGatherScatter(IntInst: dyn_cast<IntrinsicInst>(Val: U))) { |
957 | return Gatscat; |
958 | } else { |
959 | unsigned OpCode = cast<Instruction>(Val: U)->getOpcode(); |
960 | if ((OpCode == Instruction::Add || OpCode == Instruction::Mul || |
961 | OpCode == Instruction::Shl || |
962 | isAddLikeOr(I: cast<Instruction>(Val: U), DL)) && |
963 | hasAllGatScatUsers(I: cast<Instruction>(Val: U), DL)) { |
964 | continue; |
965 | } |
966 | return false; |
967 | } |
968 | } |
969 | return Gatscat; |
970 | } |
971 | |
972 | bool MVEGatherScatterLowering::optimiseOffsets(Value *Offsets, BasicBlock *BB, |
973 | LoopInfo *LI) { |
974 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: trying to optimize: " |
975 | << *Offsets << "\n" ); |
976 | // Optimise the addresses of gathers/scatters by moving invariant |
977 | // calculations out of the loop |
978 | if (!isa<Instruction>(Val: Offsets)) |
979 | return false; |
980 | Instruction *Offs = cast<Instruction>(Val: Offsets); |
981 | if (Offs->getOpcode() != Instruction::Add && !isAddLikeOr(I: Offs, DL: *DL) && |
982 | Offs->getOpcode() != Instruction::Mul && |
983 | Offs->getOpcode() != Instruction::Shl) |
984 | return false; |
985 | Loop *L = LI->getLoopFor(BB); |
986 | if (L == nullptr) |
987 | return false; |
988 | if (!Offs->hasOneUse()) { |
989 | if (!hasAllGatScatUsers(I: Offs, DL: *DL)) |
990 | return false; |
991 | } |
992 | |
993 | // Find out which, if any, operand of the instruction |
994 | // is a phi node |
995 | PHINode *Phi; |
996 | int OffsSecondOp; |
997 | if (isa<PHINode>(Val: Offs->getOperand(i: 0))) { |
998 | Phi = cast<PHINode>(Val: Offs->getOperand(i: 0)); |
999 | OffsSecondOp = 1; |
1000 | } else if (isa<PHINode>(Val: Offs->getOperand(i: 1))) { |
1001 | Phi = cast<PHINode>(Val: Offs->getOperand(i: 1)); |
1002 | OffsSecondOp = 0; |
1003 | } else { |
1004 | bool Changed = false; |
1005 | if (isa<Instruction>(Val: Offs->getOperand(i: 0)) && |
1006 | L->contains(Inst: cast<Instruction>(Val: Offs->getOperand(i: 0)))) |
1007 | Changed |= optimiseOffsets(Offsets: Offs->getOperand(i: 0), BB, LI); |
1008 | if (isa<Instruction>(Val: Offs->getOperand(i: 1)) && |
1009 | L->contains(Inst: cast<Instruction>(Val: Offs->getOperand(i: 1)))) |
1010 | Changed |= optimiseOffsets(Offsets: Offs->getOperand(i: 1), BB, LI); |
1011 | if (!Changed) |
1012 | return false; |
1013 | if (isa<PHINode>(Val: Offs->getOperand(i: 0))) { |
1014 | Phi = cast<PHINode>(Val: Offs->getOperand(i: 0)); |
1015 | OffsSecondOp = 1; |
1016 | } else if (isa<PHINode>(Val: Offs->getOperand(i: 1))) { |
1017 | Phi = cast<PHINode>(Val: Offs->getOperand(i: 1)); |
1018 | OffsSecondOp = 0; |
1019 | } else { |
1020 | return false; |
1021 | } |
1022 | } |
1023 | // A phi node we want to perform this function on should be from the |
1024 | // loop header. |
1025 | if (Phi->getParent() != L->getHeader()) |
1026 | return false; |
1027 | |
1028 | // We're looking for a simple add recurrence. |
1029 | BinaryOperator *IncInstruction; |
1030 | Value *Start, *IncrementPerRound; |
1031 | if (!matchSimpleRecurrence(P: Phi, BO&: IncInstruction, Start, Step&: IncrementPerRound) || |
1032 | IncInstruction->getOpcode() != Instruction::Add) |
1033 | return false; |
1034 | |
1035 | int IncrementingBlock = Phi->getIncomingValue(i: 0) == IncInstruction ? 0 : 1; |
1036 | |
1037 | // Get the value that is added to/multiplied with the phi |
1038 | Value *OffsSecondOperand = Offs->getOperand(i: OffsSecondOp); |
1039 | |
1040 | if (IncrementPerRound->getType() != OffsSecondOperand->getType() || |
1041 | !L->isLoopInvariant(V: OffsSecondOperand)) |
1042 | // Something has gone wrong, abort |
1043 | return false; |
1044 | |
1045 | // Only proceed if the increment per round is a constant or an instruction |
1046 | // which does not originate from within the loop |
1047 | if (!isa<Constant>(Val: IncrementPerRound) && |
1048 | !(isa<Instruction>(Val: IncrementPerRound) && |
1049 | !L->contains(Inst: cast<Instruction>(Val: IncrementPerRound)))) |
1050 | return false; |
1051 | |
1052 | // If the phi is not used by anything else, we can just adapt it when |
1053 | // replacing the instruction; if it is, we'll have to duplicate it |
1054 | PHINode *NewPhi; |
1055 | if (Phi->getNumUses() == 2) { |
1056 | // No other users -> reuse existing phi (One user is the instruction |
1057 | // we're looking at, the other is the phi increment) |
1058 | if (IncInstruction->getNumUses() != 1) { |
1059 | // If the incrementing instruction does have more users than |
1060 | // our phi, we need to copy it |
1061 | IncInstruction = BinaryOperator::Create( |
1062 | Op: Instruction::BinaryOps(IncInstruction->getOpcode()), S1: Phi, |
1063 | S2: IncrementPerRound, Name: "LoopIncrement" , InsertBefore: IncInstruction->getIterator()); |
1064 | Phi->setIncomingValue(i: IncrementingBlock, V: IncInstruction); |
1065 | } |
1066 | NewPhi = Phi; |
1067 | } else { |
1068 | // There are other users -> create a new phi |
1069 | NewPhi = PHINode::Create(Ty: Phi->getType(), NumReservedValues: 2, NameStr: "NewPhi" , InsertBefore: Phi->getIterator()); |
1070 | // Copy the incoming values of the old phi |
1071 | NewPhi->addIncoming(V: Phi->getIncomingValue(i: IncrementingBlock == 1 ? 0 : 1), |
1072 | BB: Phi->getIncomingBlock(i: IncrementingBlock == 1 ? 0 : 1)); |
1073 | IncInstruction = BinaryOperator::Create( |
1074 | Op: Instruction::BinaryOps(IncInstruction->getOpcode()), S1: NewPhi, |
1075 | S2: IncrementPerRound, Name: "LoopIncrement" , InsertBefore: IncInstruction->getIterator()); |
1076 | NewPhi->addIncoming(V: IncInstruction, |
1077 | BB: Phi->getIncomingBlock(i: IncrementingBlock)); |
1078 | IncrementingBlock = 1; |
1079 | } |
1080 | |
1081 | IRBuilder<> Builder(BB->getContext()); |
1082 | Builder.SetInsertPoint(Phi); |
1083 | Builder.SetCurrentDebugLocation(Offs->getDebugLoc()); |
1084 | |
1085 | switch (Offs->getOpcode()) { |
1086 | case Instruction::Add: |
1087 | case Instruction::Or: |
1088 | pushOutAdd(Phi&: NewPhi, OffsSecondOperand, StartIndex: IncrementingBlock == 1 ? 0 : 1); |
1089 | break; |
1090 | case Instruction::Mul: |
1091 | case Instruction::Shl: |
1092 | pushOutMulShl(Opcode: Offs->getOpcode(), Phi&: NewPhi, IncrementPerRound, |
1093 | OffsSecondOperand, LoopIncrement: IncrementingBlock, Builder); |
1094 | break; |
1095 | default: |
1096 | return false; |
1097 | } |
1098 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: simplified loop variable " |
1099 | << "add/mul\n" ); |
1100 | |
1101 | // The instruction has now been "absorbed" into the phi value |
1102 | Offs->replaceAllUsesWith(V: NewPhi); |
1103 | if (Offs->hasNUses(N: 0)) |
1104 | Offs->eraseFromParent(); |
1105 | // Clean up the old increment in case it's unused because we built a new |
1106 | // one |
1107 | if (IncInstruction->hasNUses(N: 0)) |
1108 | IncInstruction->eraseFromParent(); |
1109 | |
1110 | return true; |
1111 | } |
1112 | |
1113 | static Value *CheckAndCreateOffsetAdd(Value *X, unsigned ScaleX, Value *Y, |
1114 | unsigned ScaleY, IRBuilder<> &Builder) { |
1115 | // Splat the non-vector value to a vector of the given type - if the value is |
1116 | // a constant (and its value isn't too big), we can even use this opportunity |
1117 | // to scale it to the size of the vector elements |
1118 | auto FixSummands = [&Builder](FixedVectorType *&VT, Value *&NonVectorVal) { |
1119 | ConstantInt *Const; |
1120 | if ((Const = dyn_cast<ConstantInt>(Val: NonVectorVal)) && |
1121 | VT->getElementType() != NonVectorVal->getType()) { |
1122 | unsigned TargetElemSize = VT->getElementType()->getPrimitiveSizeInBits(); |
1123 | uint64_t N = Const->getZExtValue(); |
1124 | if (N < (unsigned)(1 << (TargetElemSize - 1))) { |
1125 | NonVectorVal = Builder.CreateVectorSplat( |
1126 | NumElts: VT->getNumElements(), V: Builder.getIntN(N: TargetElemSize, C: N)); |
1127 | return; |
1128 | } |
1129 | } |
1130 | NonVectorVal = |
1131 | Builder.CreateVectorSplat(NumElts: VT->getNumElements(), V: NonVectorVal); |
1132 | }; |
1133 | |
1134 | FixedVectorType *XElType = dyn_cast<FixedVectorType>(Val: X->getType()); |
1135 | FixedVectorType *YElType = dyn_cast<FixedVectorType>(Val: Y->getType()); |
1136 | // If one of X, Y is not a vector, we have to splat it in order |
1137 | // to add the two of them. |
1138 | if (XElType && !YElType) { |
1139 | FixSummands(XElType, Y); |
1140 | YElType = cast<FixedVectorType>(Val: Y->getType()); |
1141 | } else if (YElType && !XElType) { |
1142 | FixSummands(YElType, X); |
1143 | XElType = cast<FixedVectorType>(Val: X->getType()); |
1144 | } |
1145 | assert(XElType && YElType && "Unknown vector types" ); |
1146 | // Check that the summands are of compatible types |
1147 | if (XElType != YElType) { |
1148 | LLVM_DEBUG(dbgs() << "masked gathers/scatters: incompatible gep offsets\n" ); |
1149 | return nullptr; |
1150 | } |
1151 | |
1152 | if (XElType->getElementType()->getScalarSizeInBits() != 32) { |
1153 | // Check that by adding the vectors we do not accidentally |
1154 | // create an overflow |
1155 | Constant *ConstX = dyn_cast<Constant>(Val: X); |
1156 | Constant *ConstY = dyn_cast<Constant>(Val: Y); |
1157 | if (!ConstX || !ConstY) |
1158 | return nullptr; |
1159 | unsigned TargetElemSize = 128 / XElType->getNumElements(); |
1160 | for (unsigned i = 0; i < XElType->getNumElements(); i++) { |
1161 | ConstantInt *ConstXEl = |
1162 | dyn_cast<ConstantInt>(Val: ConstX->getAggregateElement(Elt: i)); |
1163 | ConstantInt *ConstYEl = |
1164 | dyn_cast<ConstantInt>(Val: ConstY->getAggregateElement(Elt: i)); |
1165 | if (!ConstXEl || !ConstYEl || |
1166 | ConstXEl->getZExtValue() * ScaleX + |
1167 | ConstYEl->getZExtValue() * ScaleY >= |
1168 | (unsigned)(1 << (TargetElemSize - 1))) |
1169 | return nullptr; |
1170 | } |
1171 | } |
1172 | |
1173 | Value *XScale = Builder.CreateVectorSplat( |
1174 | NumElts: XElType->getNumElements(), |
1175 | V: Builder.getIntN(N: XElType->getScalarSizeInBits(), C: ScaleX)); |
1176 | Value *YScale = Builder.CreateVectorSplat( |
1177 | NumElts: YElType->getNumElements(), |
1178 | V: Builder.getIntN(N: YElType->getScalarSizeInBits(), C: ScaleY)); |
1179 | Value *Add = Builder.CreateAdd(LHS: Builder.CreateMul(LHS: X, RHS: XScale), |
1180 | RHS: Builder.CreateMul(LHS: Y, RHS: YScale)); |
1181 | |
1182 | if (checkOffsetSize(Offsets: Add, TargetElemCount: XElType->getNumElements())) |
1183 | return Add; |
1184 | else |
1185 | return nullptr; |
1186 | } |
1187 | |
1188 | Value *MVEGatherScatterLowering::foldGEP(GetElementPtrInst *GEP, |
1189 | Value *&Offsets, unsigned &Scale, |
1190 | IRBuilder<> &Builder) { |
1191 | Value *GEPPtr = GEP->getPointerOperand(); |
1192 | Offsets = GEP->getOperand(i_nocapture: 1); |
1193 | Scale = DL->getTypeAllocSize(Ty: GEP->getSourceElementType()); |
1194 | // We only merge geps with constant offsets, because only for those |
1195 | // we can make sure that we do not cause an overflow |
1196 | if (GEP->getNumIndices() != 1 || !isa<Constant>(Val: Offsets)) |
1197 | return nullptr; |
1198 | if (GetElementPtrInst *BaseGEP = dyn_cast<GetElementPtrInst>(Val: GEPPtr)) { |
1199 | // Merge the two geps into one |
1200 | Value *BaseBasePtr = foldGEP(GEP: BaseGEP, Offsets, Scale, Builder); |
1201 | if (!BaseBasePtr) |
1202 | return nullptr; |
1203 | Offsets = CheckAndCreateOffsetAdd( |
1204 | X: Offsets, ScaleX: Scale, Y: GEP->getOperand(i_nocapture: 1), |
1205 | ScaleY: DL->getTypeAllocSize(Ty: GEP->getSourceElementType()), Builder); |
1206 | if (Offsets == nullptr) |
1207 | return nullptr; |
1208 | Scale = 1; // Scale is always an i8 at this point. |
1209 | return BaseBasePtr; |
1210 | } |
1211 | return GEPPtr; |
1212 | } |
1213 | |
1214 | bool MVEGatherScatterLowering::optimiseAddress(Value *Address, BasicBlock *BB, |
1215 | LoopInfo *LI) { |
1216 | GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(Val: Address); |
1217 | if (!GEP) |
1218 | return false; |
1219 | bool Changed = false; |
1220 | if (GEP->hasOneUse() && isa<GetElementPtrInst>(Val: GEP->getPointerOperand())) { |
1221 | IRBuilder<> Builder(GEP->getContext()); |
1222 | Builder.SetInsertPoint(GEP); |
1223 | Builder.SetCurrentDebugLocation(GEP->getDebugLoc()); |
1224 | Value *Offsets; |
1225 | unsigned Scale; |
1226 | Value *Base = foldGEP(GEP, Offsets, Scale, Builder); |
1227 | // We only want to merge the geps if there is a real chance that they can be |
1228 | // used by an MVE gather; thus the offset has to have the correct size |
1229 | // (always i32 if it is not of vector type) and the base has to be a |
1230 | // pointer. |
1231 | if (Offsets && Base && Base != GEP) { |
1232 | assert(Scale == 1 && "Expected to fold GEP to a scale of 1" ); |
1233 | Type *BaseTy = Builder.getPtrTy(); |
1234 | if (auto *VecTy = dyn_cast<FixedVectorType>(Val: Base->getType())) |
1235 | BaseTy = FixedVectorType::get(ElementType: BaseTy, FVTy: VecTy); |
1236 | GetElementPtrInst *NewAddress = GetElementPtrInst::Create( |
1237 | PointeeType: Builder.getInt8Ty(), Ptr: Builder.CreateBitCast(V: Base, DestTy: BaseTy), IdxList: Offsets, |
1238 | NameStr: "gep.merged" , InsertBefore: GEP->getIterator()); |
1239 | LLVM_DEBUG(dbgs() << "Folded GEP: " << *GEP |
1240 | << "\n new : " << *NewAddress << "\n" ); |
1241 | GEP->replaceAllUsesWith( |
1242 | V: Builder.CreateBitCast(V: NewAddress, DestTy: GEP->getType())); |
1243 | GEP = NewAddress; |
1244 | Changed = true; |
1245 | } |
1246 | } |
1247 | Changed |= optimiseOffsets(Offsets: GEP->getOperand(i_nocapture: 1), BB: GEP->getParent(), LI); |
1248 | return Changed; |
1249 | } |
1250 | |
1251 | bool MVEGatherScatterLowering::runOnFunction(Function &F) { |
1252 | if (!EnableMaskedGatherScatters) |
1253 | return false; |
1254 | auto &TPC = getAnalysis<TargetPassConfig>(); |
1255 | auto &TM = TPC.getTM<TargetMachine>(); |
1256 | auto *ST = &TM.getSubtarget<ARMSubtarget>(F); |
1257 | if (!ST->hasMVEIntegerOps()) |
1258 | return false; |
1259 | LI = &getAnalysis<LoopInfoWrapperPass>().getLoopInfo(); |
1260 | DL = &F.getDataLayout(); |
1261 | SmallVector<IntrinsicInst *, 4> Gathers; |
1262 | SmallVector<IntrinsicInst *, 4> Scatters; |
1263 | |
1264 | bool Changed = false; |
1265 | |
1266 | for (BasicBlock &BB : F) { |
1267 | Changed |= SimplifyInstructionsInBlock(BB: &BB); |
1268 | |
1269 | for (Instruction &I : BB) { |
1270 | IntrinsicInst *II = dyn_cast<IntrinsicInst>(Val: &I); |
1271 | if (II && II->getIntrinsicID() == Intrinsic::masked_gather && |
1272 | isa<FixedVectorType>(Val: II->getType())) { |
1273 | Gathers.push_back(Elt: II); |
1274 | Changed |= optimiseAddress(Address: II->getArgOperand(i: 0), BB: II->getParent(), LI); |
1275 | } else if (II && II->getIntrinsicID() == Intrinsic::masked_scatter && |
1276 | isa<FixedVectorType>(Val: II->getArgOperand(i: 0)->getType())) { |
1277 | Scatters.push_back(Elt: II); |
1278 | Changed |= optimiseAddress(Address: II->getArgOperand(i: 1), BB: II->getParent(), LI); |
1279 | } |
1280 | } |
1281 | } |
1282 | for (IntrinsicInst *I : Gathers) { |
1283 | Instruction *L = lowerGather(I); |
1284 | if (L == nullptr) |
1285 | continue; |
1286 | |
1287 | // Get rid of any now dead instructions |
1288 | SimplifyInstructionsInBlock(BB: L->getParent()); |
1289 | Changed = true; |
1290 | } |
1291 | |
1292 | for (IntrinsicInst *I : Scatters) { |
1293 | Instruction *S = lowerScatter(I); |
1294 | if (S == nullptr) |
1295 | continue; |
1296 | |
1297 | // Get rid of any now dead instructions |
1298 | SimplifyInstructionsInBlock(BB: S->getParent()); |
1299 | Changed = true; |
1300 | } |
1301 | return Changed; |
1302 | } |
1303 | |