1 //===- SROA.cpp - Scalar Replacement Of Aggregates ------------------------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 /// This transformation implements the well known scalar replacement of
11 /// aggregates transformation. It tries to identify promotable elements of an
12 /// aggregate alloca, and promote them to registers. It will also try to
13 /// convert uses of an element (or set of elements) of an alloca into a vector
14 /// or bitfield-style integer scalar if appropriate.
16 /// It works to do this with minimal slicing of the alloca so that regions
17 /// which are merely transferred in and out of external memory remain unchanged
18 /// and are not decomposed to scalar code.
20 /// Because this also performs alloca promotion, it can be thought of as also
21 /// serving the purpose of SSA formation. The algorithm iterates on the
22 /// function until all opportunities for promotion have been realized.
24 //===----------------------------------------------------------------------===//
26 #define DEBUG_TYPE "sroa"
27 #include "llvm/Transforms/Scalar.h"
28 #include "llvm/ADT/STLExtras.h"
29 #include "llvm/ADT/SetVector.h"
30 #include "llvm/ADT/SmallVector.h"
31 #include "llvm/ADT/Statistic.h"
32 #include "llvm/Analysis/Dominators.h"
33 #include "llvm/Analysis/Loads.h"
34 #include "llvm/Analysis/PtrUseVisitor.h"
35 #include "llvm/Analysis/ValueTracking.h"
36 #include "llvm/Constants.h"
37 #include "llvm/DIBuilder.h"
38 #include "llvm/DataLayout.h"
39 #include "llvm/DebugInfo.h"
40 #include "llvm/DerivedTypes.h"
41 #include "llvm/Function.h"
42 #include "llvm/IRBuilder.h"
43 #include "llvm/InstVisitor.h"
44 #include "llvm/Instructions.h"
45 #include "llvm/IntrinsicInst.h"
46 #include "llvm/LLVMContext.h"
47 #include "llvm/Module.h"
48 #include "llvm/Operator.h"
49 #include "llvm/Pass.h"
50 #include "llvm/Support/CommandLine.h"
51 #include "llvm/Support/Debug.h"
52 #include "llvm/Support/ErrorHandling.h"
53 #include "llvm/Support/GetElementPtrTypeIterator.h"
54 #include "llvm/Support/MathExtras.h"
55 #include "llvm/Support/raw_ostream.h"
56 #include "llvm/Transforms/Utils/Local.h"
57 #include "llvm/Transforms/Utils/PromoteMemToReg.h"
58 #include "llvm/Transforms/Utils/SSAUpdater.h"
61 STATISTIC(NumAllocasAnalyzed, "Number of allocas analyzed for replacement");
62 STATISTIC(NumNewAllocas, "Number of new, smaller allocas introduced");
63 STATISTIC(NumPromoted, "Number of allocas promoted to SSA values");
64 STATISTIC(NumLoadsSpeculated, "Number of loads speculated to allow promotion");
65 STATISTIC(NumDeleted, "Number of instructions deleted");
66 STATISTIC(NumVectorized, "Number of vectorized aggregates");
68 /// Hidden option to force the pass to not use DomTree and mem2reg, instead
69 /// forming SSA values through the SSAUpdater infrastructure.
71 ForceSSAUpdater("force-ssa-updater", cl::init(false), cl::Hidden);
74 /// \brief Alloca partitioning representation.
76 /// This class represents a partitioning of an alloca into slices, and
77 /// information about the nature of uses of each slice of the alloca. The goal
78 /// is that this information is sufficient to decide if and how to split the
79 /// alloca apart and replace slices with scalars. It is also intended that this
80 /// structure can capture the relevant information needed both to decide about
81 /// and to enact these transformations.
82 class AllocaPartitioning {
84 /// \brief A common base class for representing a half-open byte range.
86 /// \brief The beginning offset of the range.
89 /// \brief The ending offset, not included in the range.
92 ByteRange() : BeginOffset(), EndOffset() {}
93 ByteRange(uint64_t BeginOffset, uint64_t EndOffset)
94 : BeginOffset(BeginOffset), EndOffset(EndOffset) {}
96 /// \brief Support for ordering ranges.
98 /// This provides an ordering over ranges such that start offsets are
99 /// always increasing, and within equal start offsets, the end offsets are
100 /// decreasing. Thus the spanning range comes first in a cluster with the
101 /// same start position.
102 bool operator<(const ByteRange &RHS) const {
103 if (BeginOffset < RHS.BeginOffset) return true;
104 if (BeginOffset > RHS.BeginOffset) return false;
105 if (EndOffset > RHS.EndOffset) return true;
109 /// \brief Support comparison with a single offset to allow binary searches.
110 friend bool operator<(const ByteRange &LHS, uint64_t RHSOffset) {
111 return LHS.BeginOffset < RHSOffset;
114 friend LLVM_ATTRIBUTE_UNUSED bool operator<(uint64_t LHSOffset,
115 const ByteRange &RHS) {
116 return LHSOffset < RHS.BeginOffset;
119 bool operator==(const ByteRange &RHS) const {
120 return BeginOffset == RHS.BeginOffset && EndOffset == RHS.EndOffset;
122 bool operator!=(const ByteRange &RHS) const { return !operator==(RHS); }
125 /// \brief A partition of an alloca.
127 /// This structure represents a contiguous partition of the alloca. These are
128 /// formed by examining the uses of the alloca. During formation, they may
129 /// overlap but once an AllocaPartitioning is built, the Partitions within it
130 /// are all disjoint.
131 struct Partition : public ByteRange {
132 /// \brief Whether this partition is splittable into smaller partitions.
134 /// We flag partitions as splittable when they are formed entirely due to
135 /// accesses by trivially splittable operations such as memset and memcpy.
138 /// \brief Test whether a partition has been marked as dead.
139 bool isDead() const {
140 if (BeginOffset == UINT64_MAX) {
141 assert(EndOffset == UINT64_MAX);
147 /// \brief Kill a partition.
148 /// This is accomplished by setting both its beginning and end offset to
149 /// the maximum possible value.
151 assert(!isDead() && "He's Dead, Jim!");
152 BeginOffset = EndOffset = UINT64_MAX;
155 Partition() : ByteRange(), IsSplittable() {}
156 Partition(uint64_t BeginOffset, uint64_t EndOffset, bool IsSplittable)
157 : ByteRange(BeginOffset, EndOffset), IsSplittable(IsSplittable) {}
160 /// \brief A particular use of a partition of the alloca.
162 /// This structure is used to associate uses of a partition with it. They
163 /// mark the range of bytes which are referenced by a particular instruction,
164 /// and includes a handle to the user itself and the pointer value in use.
165 /// The bounds of these uses are determined by intersecting the bounds of the
166 /// memory use itself with a particular partition. As a consequence there is
167 /// intentionally overlap between various uses of the same partition.
168 struct PartitionUse : public ByteRange {
169 /// \brief The use in question. Provides access to both user and used value.
171 /// Note that this may be null if the partition use is *dead*, that is, it
172 /// should be ignored.
175 PartitionUse() : ByteRange(), U() {}
176 PartitionUse(uint64_t BeginOffset, uint64_t EndOffset, Use *U)
177 : ByteRange(BeginOffset, EndOffset), U(U) {}
180 /// \brief Construct a partitioning of a particular alloca.
182 /// Construction does most of the work for partitioning the alloca. This
183 /// performs the necessary walks of users and builds a partitioning from it.
184 AllocaPartitioning(const DataLayout &TD, AllocaInst &AI);
186 /// \brief Test whether a pointer to the allocation escapes our analysis.
188 /// If this is true, the partitioning is never fully built and should be
190 bool isEscaped() const { return PointerEscapingInstr; }
192 /// \brief Support for iterating over the partitions.
194 typedef SmallVectorImpl<Partition>::iterator iterator;
195 iterator begin() { return Partitions.begin(); }
196 iterator end() { return Partitions.end(); }
198 typedef SmallVectorImpl<Partition>::const_iterator const_iterator;
199 const_iterator begin() const { return Partitions.begin(); }
200 const_iterator end() const { return Partitions.end(); }
203 /// \brief Support for iterating over and manipulating a particular
204 /// partition's uses.
206 /// The iteration support provided for uses is more limited, but also
207 /// includes some manipulation routines to support rewriting the uses of
208 /// partitions during SROA.
210 typedef SmallVectorImpl<PartitionUse>::iterator use_iterator;
211 use_iterator use_begin(unsigned Idx) { return Uses[Idx].begin(); }
212 use_iterator use_begin(const_iterator I) { return Uses[I - begin()].begin(); }
213 use_iterator use_end(unsigned Idx) { return Uses[Idx].end(); }
214 use_iterator use_end(const_iterator I) { return Uses[I - begin()].end(); }
216 typedef SmallVectorImpl<PartitionUse>::const_iterator const_use_iterator;
217 const_use_iterator use_begin(unsigned Idx) const { return Uses[Idx].begin(); }
218 const_use_iterator use_begin(const_iterator I) const {
219 return Uses[I - begin()].begin();
221 const_use_iterator use_end(unsigned Idx) const { return Uses[Idx].end(); }
222 const_use_iterator use_end(const_iterator I) const {
223 return Uses[I - begin()].end();
226 unsigned use_size(unsigned Idx) const { return Uses[Idx].size(); }
227 unsigned use_size(const_iterator I) const { return Uses[I - begin()].size(); }
228 const PartitionUse &getUse(unsigned PIdx, unsigned UIdx) const {
229 return Uses[PIdx][UIdx];
231 const PartitionUse &getUse(const_iterator I, unsigned UIdx) const {
232 return Uses[I - begin()][UIdx];
235 void use_push_back(unsigned Idx, const PartitionUse &PU) {
236 Uses[Idx].push_back(PU);
238 void use_push_back(const_iterator I, const PartitionUse &PU) {
239 Uses[I - begin()].push_back(PU);
243 /// \brief Allow iterating the dead users for this alloca.
245 /// These are instructions which will never actually use the alloca as they
246 /// are outside the allocated range. They are safe to replace with undef and
249 typedef SmallVectorImpl<Instruction *>::const_iterator dead_user_iterator;
250 dead_user_iterator dead_user_begin() const { return DeadUsers.begin(); }
251 dead_user_iterator dead_user_end() const { return DeadUsers.end(); }
254 /// \brief Allow iterating the dead expressions referring to this alloca.
256 /// These are operands which have cannot actually be used to refer to the
257 /// alloca as they are outside its range and the user doesn't correct for
258 /// that. These mostly consist of PHI node inputs and the like which we just
259 /// need to replace with undef.
261 typedef SmallVectorImpl<Use *>::const_iterator dead_op_iterator;
262 dead_op_iterator dead_op_begin() const { return DeadOperands.begin(); }
263 dead_op_iterator dead_op_end() const { return DeadOperands.end(); }
266 /// \brief MemTransferInst auxiliary data.
267 /// This struct provides some auxiliary data about memory transfer
268 /// intrinsics such as memcpy and memmove. These intrinsics can use two
269 /// different ranges within the same alloca, and provide other challenges to
270 /// correctly represent. We stash extra data to help us untangle this
271 /// after the partitioning is complete.
272 struct MemTransferOffsets {
273 /// The destination begin and end offsets when the destination is within
274 /// this alloca. If the end offset is zero the destination is not within
276 uint64_t DestBegin, DestEnd;
278 /// The source begin and end offsets when the source is within this alloca.
279 /// If the end offset is zero, the source is not within this alloca.
280 uint64_t SourceBegin, SourceEnd;
282 /// Flag for whether an alloca is splittable.
285 MemTransferOffsets getMemTransferOffsets(MemTransferInst &II) const {
286 return MemTransferInstData.lookup(&II);
289 /// \brief Map from a PHI or select operand back to a partition.
291 /// When manipulating PHI nodes or selects, they can use more than one
292 /// partition of an alloca. We store a special mapping to allow finding the
293 /// partition referenced by each of these operands, if any.
294 iterator findPartitionForPHIOrSelectOperand(Use *U) {
295 SmallDenseMap<Use *, std::pair<unsigned, unsigned> >::const_iterator MapIt
296 = PHIOrSelectOpMap.find(U);
297 if (MapIt == PHIOrSelectOpMap.end())
300 return begin() + MapIt->second.first;
303 /// \brief Map from a PHI or select operand back to the specific use of
306 /// Similar to mapping these operands back to the partitions, this maps
307 /// directly to the use structure of that partition.
308 use_iterator findPartitionUseForPHIOrSelectOperand(Use *U) {
309 SmallDenseMap<Use *, std::pair<unsigned, unsigned> >::const_iterator MapIt
310 = PHIOrSelectOpMap.find(U);
311 assert(MapIt != PHIOrSelectOpMap.end());
312 return Uses[MapIt->second.first].begin() + MapIt->second.second;
315 /// \brief Compute a common type among the uses of a particular partition.
317 /// This routines walks all of the uses of a particular partition and tries
318 /// to find a common type between them. Untyped operations such as memset and
319 /// memcpy are ignored.
320 Type *getCommonType(iterator I) const;
322 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
323 void print(raw_ostream &OS, const_iterator I, StringRef Indent = " ") const;
324 void printUsers(raw_ostream &OS, const_iterator I,
325 StringRef Indent = " ") const;
326 void print(raw_ostream &OS) const;
327 void LLVM_ATTRIBUTE_NOINLINE LLVM_ATTRIBUTE_USED dump(const_iterator I) const;
328 void LLVM_ATTRIBUTE_NOINLINE LLVM_ATTRIBUTE_USED dump() const;
332 template <typename DerivedT, typename RetT = void> class BuilderBase;
333 class PartitionBuilder;
334 friend class AllocaPartitioning::PartitionBuilder;
336 friend class AllocaPartitioning::UseBuilder;
338 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
339 /// \brief Handle to alloca instruction to simplify method interfaces.
343 /// \brief The instruction responsible for this alloca having no partitioning.
345 /// When an instruction (potentially) escapes the pointer to the alloca, we
346 /// store a pointer to that here and abort trying to partition the alloca.
347 /// This will be null if the alloca is partitioned successfully.
348 Instruction *PointerEscapingInstr;
350 /// \brief The partitions of the alloca.
352 /// We store a vector of the partitions over the alloca here. This vector is
353 /// sorted by increasing begin offset, and then by decreasing end offset. See
354 /// the Partition inner class for more details. Initially (during
355 /// construction) there are overlaps, but we form a disjoint sequence of
356 /// partitions while finishing construction and a fully constructed object is
357 /// expected to always have this as a disjoint space.
358 SmallVector<Partition, 8> Partitions;
360 /// \brief The uses of the partitions.
362 /// This is essentially a mapping from each partition to a list of uses of
363 /// that partition. The mapping is done with a Uses vector that has the exact
364 /// same number of entries as the partition vector. Each entry is itself
365 /// a vector of the uses.
366 SmallVector<SmallVector<PartitionUse, 2>, 8> Uses;
368 /// \brief Instructions which will become dead if we rewrite the alloca.
370 /// Note that these are not separated by partition. This is because we expect
371 /// a partitioned alloca to be completely rewritten or not rewritten at all.
372 /// If rewritten, all these instructions can simply be removed and replaced
373 /// with undef as they come from outside of the allocated space.
374 SmallVector<Instruction *, 8> DeadUsers;
376 /// \brief Operands which will become dead if we rewrite the alloca.
378 /// These are operands that in their particular use can be replaced with
379 /// undef when we rewrite the alloca. These show up in out-of-bounds inputs
380 /// to PHI nodes and the like. They aren't entirely dead (there might be
381 /// a GEP back into the bounds using it elsewhere) and nor is the PHI, but we
382 /// want to swap this particular input for undef to simplify the use lists of
384 SmallVector<Use *, 8> DeadOperands;
386 /// \brief The underlying storage for auxiliary memcpy and memset info.
387 SmallDenseMap<MemTransferInst *, MemTransferOffsets, 4> MemTransferInstData;
389 /// \brief A side datastructure used when building up the partitions and uses.
391 /// This mapping is only really used during the initial building of the
392 /// partitioning so that we can retain information about PHI and select nodes
394 SmallDenseMap<Instruction *, std::pair<uint64_t, bool> > PHIOrSelectSizes;
396 /// \brief Auxiliary information for particular PHI or select operands.
397 SmallDenseMap<Use *, std::pair<unsigned, unsigned>, 4> PHIOrSelectOpMap;
399 /// \brief A utility routine called from the constructor.
401 /// This does what it says on the tin. It is the key of the alloca partition
402 /// splitting and merging. After it is called we have the desired disjoint
403 /// collection of partitions.
404 void splitAndMergePartitions();
408 static Value *foldSelectInst(SelectInst &SI) {
409 // If the condition being selected on is a constant or the same value is
410 // being selected between, fold the select. Yes this does (rarely) happen
412 if (ConstantInt *CI = dyn_cast<ConstantInt>(SI.getCondition()))
413 return SI.getOperand(1+CI->isZero());
414 if (SI.getOperand(1) == SI.getOperand(2)) {
415 return SI.getOperand(1);
420 /// \brief Builder for the alloca partitioning.
422 /// This class builds an alloca partitioning by recursively visiting the uses
423 /// of an alloca and splitting the partitions for each load and store at each
425 class AllocaPartitioning::PartitionBuilder
426 : public PtrUseVisitor<PartitionBuilder> {
427 friend class PtrUseVisitor<PartitionBuilder>;
428 friend class InstVisitor<PartitionBuilder>;
429 typedef PtrUseVisitor<PartitionBuilder> Base;
431 const uint64_t AllocSize;
432 AllocaPartitioning &P;
434 SmallDenseMap<Instruction *, unsigned> MemTransferPartitionMap;
437 PartitionBuilder(const DataLayout &DL, AllocaInst &AI, AllocaPartitioning &P)
438 : PtrUseVisitor<PartitionBuilder>(DL),
439 AllocSize(DL.getTypeAllocSize(AI.getAllocatedType())),
443 void insertUse(Instruction &I, const APInt &Offset, uint64_t Size,
444 bool IsSplittable = false) {
445 // Completely skip uses which have a zero size or start either before or
446 // past the end of the allocation.
447 if (Size == 0 || Offset.isNegative() || Offset.uge(AllocSize)) {
448 DEBUG(dbgs() << "WARNING: Ignoring " << Size << " byte use @" << Offset
449 << " which has zero size or starts outside of the "
450 << AllocSize << " byte alloca:\n"
451 << " alloca: " << P.AI << "\n"
452 << " use: " << I << "\n");
456 uint64_t BeginOffset = Offset.getZExtValue();
457 uint64_t EndOffset = BeginOffset + Size;
459 // Clamp the end offset to the end of the allocation. Note that this is
460 // formulated to handle even the case where "BeginOffset + Size" overflows.
461 // NOTE! This may appear superficially to be something we could ignore
462 // entirely, but that is not so! There may be PHI-node uses where some
463 // instructions are dead but not others. We can't completely ignore the
464 // PHI node, and so have to record at least the information here.
465 assert(AllocSize >= BeginOffset); // Established above.
466 if (Size > AllocSize - BeginOffset) {
467 DEBUG(dbgs() << "WARNING: Clamping a " << Size << " byte use @" << Offset
468 << " to remain within the " << AllocSize << " byte alloca:\n"
469 << " alloca: " << P.AI << "\n"
470 << " use: " << I << "\n");
471 EndOffset = AllocSize;
474 Partition New(BeginOffset, EndOffset, IsSplittable);
475 P.Partitions.push_back(New);
478 void handleLoadOrStore(Type *Ty, Instruction &I, const APInt &Offset,
480 uint64_t Size = DL.getTypeStoreSize(Ty);
482 // If this memory access can be shown to *statically* extend outside the
483 // bounds of of the allocation, it's behavior is undefined, so simply
484 // ignore it. Note that this is more strict than the generic clamping
485 // behavior of insertUse. We also try to handle cases which might run the
487 // FIXME: We should instead consider the pointer to have escaped if this
488 // function is being instrumented for addressing bugs or race conditions.
489 if (Offset.isNegative() || Size > AllocSize ||
490 Offset.ugt(AllocSize - Size)) {
491 DEBUG(dbgs() << "WARNING: Ignoring " << Size << " byte "
492 << (isa<LoadInst>(I) ? "load" : "store") << " @" << Offset
493 << " which extends past the end of the " << AllocSize
495 << " alloca: " << P.AI << "\n"
496 << " use: " << I << "\n");
500 // We allow splitting of loads and stores where the type is an integer type
501 // and which cover the entire alloca. Such integer loads and stores
502 // often require decomposition into fine grained loads and stores.
503 bool IsSplittable = false;
504 if (IntegerType *ITy = dyn_cast<IntegerType>(Ty))
505 IsSplittable = !IsVolatile && ITy->getBitWidth() == AllocSize*8;
507 insertUse(I, Offset, Size, IsSplittable);
510 void visitLoadInst(LoadInst &LI) {
511 assert((!LI.isSimple() || LI.getType()->isSingleValueType()) &&
512 "All simple FCA loads should have been pre-split");
515 return PI.setAborted(&LI);
517 return handleLoadOrStore(LI.getType(), LI, Offset, LI.isVolatile());
520 void visitStoreInst(StoreInst &SI) {
521 Value *ValOp = SI.getValueOperand();
523 return PI.setEscapedAndAborted(&SI);
525 return PI.setAborted(&SI);
527 assert((!SI.isSimple() || ValOp->getType()->isSingleValueType()) &&
528 "All simple FCA stores should have been pre-split");
529 handleLoadOrStore(ValOp->getType(), SI, Offset, SI.isVolatile());
533 void visitMemSetInst(MemSetInst &II) {
534 assert(II.getRawDest() == *U && "Pointer use is not the destination?");
535 ConstantInt *Length = dyn_cast<ConstantInt>(II.getLength());
536 if ((Length && Length->getValue() == 0) ||
537 (IsOffsetKnown && !Offset.isNegative() && Offset.uge(AllocSize)))
538 // Zero-length mem transfer intrinsics can be ignored entirely.
542 return PI.setAborted(&II);
544 insertUse(II, Offset,
545 Length ? Length->getLimitedValue()
546 : AllocSize - Offset.getLimitedValue(),
550 void visitMemTransferInst(MemTransferInst &II) {
551 ConstantInt *Length = dyn_cast<ConstantInt>(II.getLength());
552 if ((Length && Length->getValue() == 0) ||
553 (IsOffsetKnown && !Offset.isNegative() && Offset.uge(AllocSize)))
554 // Zero-length mem transfer intrinsics can be ignored entirely.
558 return PI.setAborted(&II);
560 uint64_t RawOffset = Offset.getLimitedValue();
561 uint64_t Size = Length ? Length->getLimitedValue()
562 : AllocSize - RawOffset;
564 MemTransferOffsets &Offsets = P.MemTransferInstData[&II];
566 // Only intrinsics with a constant length can be split.
567 Offsets.IsSplittable = Length;
569 if (*U == II.getRawDest()) {
570 Offsets.DestBegin = RawOffset;
571 Offsets.DestEnd = RawOffset + Size;
573 if (*U == II.getRawSource()) {
574 Offsets.SourceBegin = RawOffset;
575 Offsets.SourceEnd = RawOffset + Size;
578 // If we have set up end offsets for both the source and the destination,
579 // we have found both sides of this transfer pointing at the same alloca.
580 bool SeenBothEnds = Offsets.SourceEnd && Offsets.DestEnd;
581 if (SeenBothEnds && II.getRawDest() != II.getRawSource()) {
582 unsigned PrevIdx = MemTransferPartitionMap[&II];
584 // Check if the begin offsets match and this is a non-volatile transfer.
585 // In that case, we can completely elide the transfer.
586 if (!II.isVolatile() && Offsets.SourceBegin == Offsets.DestBegin) {
587 P.Partitions[PrevIdx].kill();
591 // Otherwise we have an offset transfer within the same alloca. We can't
593 P.Partitions[PrevIdx].IsSplittable = Offsets.IsSplittable = false;
594 } else if (SeenBothEnds) {
595 // Handle the case where this exact use provides both ends of the
597 assert(II.getRawDest() == II.getRawSource());
599 // For non-volatile transfers this is a no-op.
600 if (!II.isVolatile())
603 // Otherwise just suppress splitting.
604 Offsets.IsSplittable = false;
608 // Insert the use now that we've fixed up the splittable nature.
609 insertUse(II, Offset, Size, Offsets.IsSplittable);
611 // Setup the mapping from intrinsic to partition of we've not seen both
612 // ends of this transfer.
614 unsigned NewIdx = P.Partitions.size() - 1;
616 = MemTransferPartitionMap.insert(std::make_pair(&II, NewIdx)).second;
618 "Already have intrinsic in map but haven't seen both ends");
623 // Disable SRoA for any intrinsics except for lifetime invariants.
624 // FIXME: What about debug instrinsics? This matches old behavior, but
625 // doesn't make sense.
626 void visitIntrinsicInst(IntrinsicInst &II) {
628 return PI.setAborted(&II);
630 if (II.getIntrinsicID() == Intrinsic::lifetime_start ||
631 II.getIntrinsicID() == Intrinsic::lifetime_end) {
632 ConstantInt *Length = cast<ConstantInt>(II.getArgOperand(0));
633 uint64_t Size = std::min(AllocSize - Offset.getLimitedValue(),
634 Length->getLimitedValue());
635 insertUse(II, Offset, Size, true);
639 Base::visitIntrinsicInst(II);
642 Instruction *hasUnsafePHIOrSelectUse(Instruction *Root, uint64_t &Size) {
643 // We consider any PHI or select that results in a direct load or store of
644 // the same offset to be a viable use for partitioning purposes. These uses
645 // are considered unsplittable and the size is the maximum loaded or stored
647 SmallPtrSet<Instruction *, 4> Visited;
648 SmallVector<std::pair<Instruction *, Instruction *>, 4> Uses;
649 Visited.insert(Root);
650 Uses.push_back(std::make_pair(cast<Instruction>(*U), Root));
651 // If there are no loads or stores, the access is dead. We mark that as
652 // a size zero access.
655 Instruction *I, *UsedI;
656 llvm::tie(UsedI, I) = Uses.pop_back_val();
658 if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
659 Size = std::max(Size, DL.getTypeStoreSize(LI->getType()));
662 if (StoreInst *SI = dyn_cast<StoreInst>(I)) {
663 Value *Op = SI->getOperand(0);
666 Size = std::max(Size, DL.getTypeStoreSize(Op->getType()));
670 if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(I)) {
671 if (!GEP->hasAllZeroIndices())
673 } else if (!isa<BitCastInst>(I) && !isa<PHINode>(I) &&
674 !isa<SelectInst>(I)) {
678 for (Value::use_iterator UI = I->use_begin(), UE = I->use_end(); UI != UE;
680 if (Visited.insert(cast<Instruction>(*UI)))
681 Uses.push_back(std::make_pair(I, cast<Instruction>(*UI)));
682 } while (!Uses.empty());
687 void visitPHINode(PHINode &PN) {
691 return PI.setAborted(&PN);
693 // See if we already have computed info on this node.
694 std::pair<uint64_t, bool> &PHIInfo = P.PHIOrSelectSizes[&PN];
696 PHIInfo.second = true;
697 insertUse(PN, Offset, PHIInfo.first);
701 // Check for an unsafe use of the PHI node.
702 if (Instruction *UnsafeI = hasUnsafePHIOrSelectUse(&PN, PHIInfo.first))
703 return PI.setAborted(UnsafeI);
705 insertUse(PN, Offset, PHIInfo.first);
708 void visitSelectInst(SelectInst &SI) {
711 if (Value *Result = foldSelectInst(SI)) {
713 // If the result of the constant fold will be the pointer, recurse
714 // through the select as if we had RAUW'ed it.
720 return PI.setAborted(&SI);
722 // See if we already have computed info on this node.
723 std::pair<uint64_t, bool> &SelectInfo = P.PHIOrSelectSizes[&SI];
724 if (SelectInfo.first) {
725 SelectInfo.second = true;
726 insertUse(SI, Offset, SelectInfo.first);
730 // Check for an unsafe use of the PHI node.
731 if (Instruction *UnsafeI = hasUnsafePHIOrSelectUse(&SI, SelectInfo.first))
732 return PI.setAborted(UnsafeI);
734 insertUse(SI, Offset, SelectInfo.first);
737 /// \brief Disable SROA entirely if there are unhandled users of the alloca.
738 void visitInstruction(Instruction &I) {
743 /// \brief Use adder for the alloca partitioning.
745 /// This class adds the uses of an alloca to all of the partitions which they
746 /// use. For splittable partitions, this can end up doing essentially a linear
747 /// walk of the partitions, but the number of steps remains bounded by the
748 /// total result instruction size:
749 /// - The number of partitions is a result of the number unsplittable
750 /// instructions using the alloca.
751 /// - The number of users of each partition is at worst the total number of
752 /// splittable instructions using the alloca.
753 /// Thus we will produce N * M instructions in the end, where N are the number
754 /// of unsplittable uses and M are the number of splittable. This visitor does
755 /// the exact same number of updates to the partitioning.
757 /// In the more common case, this visitor will leverage the fact that the
758 /// partition space is pre-sorted, and do a logarithmic search for the
759 /// partition needed, making the total visit a classical ((N + M) * log(N))
760 /// complexity operation.
761 class AllocaPartitioning::UseBuilder : public PtrUseVisitor<UseBuilder> {
762 friend class PtrUseVisitor<UseBuilder>;
763 friend class InstVisitor<UseBuilder>;
764 typedef PtrUseVisitor<UseBuilder> Base;
766 const uint64_t AllocSize;
767 AllocaPartitioning &P;
769 /// \brief Set to de-duplicate dead instructions found in the use walk.
770 SmallPtrSet<Instruction *, 4> VisitedDeadInsts;
773 UseBuilder(const DataLayout &TD, AllocaInst &AI, AllocaPartitioning &P)
774 : PtrUseVisitor<UseBuilder>(TD),
775 AllocSize(TD.getTypeAllocSize(AI.getAllocatedType())),
779 void markAsDead(Instruction &I) {
780 if (VisitedDeadInsts.insert(&I))
781 P.DeadUsers.push_back(&I);
784 void insertUse(Instruction &User, const APInt &Offset, uint64_t Size) {
785 // If the use has a zero size or extends outside of the allocation, record
786 // it as a dead use for elimination later.
787 if (Size == 0 || Offset.isNegative() || Offset.uge(AllocSize))
788 return markAsDead(User);
790 uint64_t BeginOffset = Offset.getZExtValue();
791 uint64_t EndOffset = BeginOffset + Size;
793 // Clamp the end offset to the end of the allocation. Note that this is
794 // formulated to handle even the case where "BeginOffset + Size" overflows.
795 assert(AllocSize >= BeginOffset); // Established above.
796 if (Size > AllocSize - BeginOffset)
797 EndOffset = AllocSize;
799 // NB: This only works if we have zero overlapping partitions.
800 iterator B = std::lower_bound(P.begin(), P.end(), BeginOffset);
801 if (B != P.begin() && llvm::prior(B)->EndOffset > BeginOffset)
803 for (iterator I = B, E = P.end(); I != E && I->BeginOffset < EndOffset;
805 PartitionUse NewPU(std::max(I->BeginOffset, BeginOffset),
806 std::min(I->EndOffset, EndOffset), U);
807 P.use_push_back(I, NewPU);
808 if (isa<PHINode>(U->getUser()) || isa<SelectInst>(U->getUser()))
809 P.PHIOrSelectOpMap[U]
810 = std::make_pair(I - P.begin(), P.Uses[I - P.begin()].size() - 1);
814 void handleLoadOrStore(Type *Ty, Instruction &I, const APInt &Offset) {
815 uint64_t Size = DL.getTypeStoreSize(Ty);
817 // If this memory access can be shown to *statically* extend outside the
818 // bounds of of the allocation, it's behavior is undefined, so simply
819 // ignore it. Note that this is more strict than the generic clamping
820 // behavior of insertUse.
821 if (Offset.isNegative() || Size > AllocSize ||
822 Offset.ugt(AllocSize - Size))
823 return markAsDead(I);
825 insertUse(I, Offset, Size);
828 void visitBitCastInst(BitCastInst &BC) {
830 return markAsDead(BC);
832 return Base::visitBitCastInst(BC);
835 void visitGetElementPtrInst(GetElementPtrInst &GEPI) {
836 if (GEPI.use_empty())
837 return markAsDead(GEPI);
839 return Base::visitGetElementPtrInst(GEPI);
842 void visitLoadInst(LoadInst &LI) {
843 assert(IsOffsetKnown);
844 handleLoadOrStore(LI.getType(), LI, Offset);
847 void visitStoreInst(StoreInst &SI) {
848 assert(IsOffsetKnown);
849 handleLoadOrStore(SI.getOperand(0)->getType(), SI, Offset);
852 void visitMemSetInst(MemSetInst &II) {
853 ConstantInt *Length = dyn_cast<ConstantInt>(II.getLength());
854 if ((Length && Length->getValue() == 0) ||
855 (IsOffsetKnown && !Offset.isNegative() && Offset.uge(AllocSize)))
856 return markAsDead(II);
858 assert(IsOffsetKnown);
859 insertUse(II, Offset, Length ? Length->getLimitedValue()
860 : AllocSize - Offset.getLimitedValue());
863 void visitMemTransferInst(MemTransferInst &II) {
864 ConstantInt *Length = dyn_cast<ConstantInt>(II.getLength());
865 if ((Length && Length->getValue() == 0) ||
866 (IsOffsetKnown && !Offset.isNegative() && Offset.uge(AllocSize)))
867 return markAsDead(II);
869 assert(IsOffsetKnown);
870 uint64_t Size = Length ? Length->getLimitedValue()
871 : AllocSize - Offset.getLimitedValue();
873 MemTransferOffsets &Offsets = P.MemTransferInstData[&II];
874 if (!II.isVolatile() && Offsets.DestEnd && Offsets.SourceEnd &&
875 Offsets.DestBegin == Offsets.SourceBegin)
876 return markAsDead(II); // Skip identity transfers without side-effects.
878 insertUse(II, Offset, Size);
881 void visitIntrinsicInst(IntrinsicInst &II) {
882 assert(IsOffsetKnown);
883 assert(II.getIntrinsicID() == Intrinsic::lifetime_start ||
884 II.getIntrinsicID() == Intrinsic::lifetime_end);
886 ConstantInt *Length = cast<ConstantInt>(II.getArgOperand(0));
887 insertUse(II, Offset, std::min(Length->getLimitedValue(),
888 AllocSize - Offset.getLimitedValue()));
891 void insertPHIOrSelect(Instruction &User, const APInt &Offset) {
892 uint64_t Size = P.PHIOrSelectSizes.lookup(&User).first;
894 // For PHI and select operands outside the alloca, we can't nuke the entire
895 // phi or select -- the other side might still be relevant, so we special
896 // case them here and use a separate structure to track the operands
897 // themselves which should be replaced with undef.
898 if ((Offset.isNegative() && Offset.uge(Size)) ||
899 (!Offset.isNegative() && Offset.uge(AllocSize))) {
900 P.DeadOperands.push_back(U);
904 insertUse(User, Offset, Size);
907 void visitPHINode(PHINode &PN) {
909 return markAsDead(PN);
911 assert(IsOffsetKnown);
912 insertPHIOrSelect(PN, Offset);
915 void visitSelectInst(SelectInst &SI) {
917 return markAsDead(SI);
919 if (Value *Result = foldSelectInst(SI)) {
921 // If the result of the constant fold will be the pointer, recurse
922 // through the select as if we had RAUW'ed it.
925 // Otherwise the operand to the select is dead, and we can replace it
927 P.DeadOperands.push_back(U);
932 assert(IsOffsetKnown);
933 insertPHIOrSelect(SI, Offset);
936 /// \brief Unreachable, we've already visited the alloca once.
937 void visitInstruction(Instruction &I) {
938 llvm_unreachable("Unhandled instruction in use builder.");
942 void AllocaPartitioning::splitAndMergePartitions() {
943 size_t NumDeadPartitions = 0;
945 // Track the range of splittable partitions that we pass when accumulating
946 // overlapping unsplittable partitions.
947 uint64_t SplitEndOffset = 0ull;
949 Partition New(0ull, 0ull, false);
951 for (unsigned i = 0, j = i, e = Partitions.size(); i != e; i = j) {
954 if (!Partitions[i].IsSplittable || New.BeginOffset == New.EndOffset) {
955 assert(New.BeginOffset == New.EndOffset);
958 assert(New.IsSplittable);
959 New.EndOffset = std::max(New.EndOffset, Partitions[i].EndOffset);
961 assert(New.BeginOffset != New.EndOffset);
963 // Scan the overlapping partitions.
964 while (j != e && New.EndOffset > Partitions[j].BeginOffset) {
965 // If the new partition we are forming is splittable, stop at the first
966 // unsplittable partition.
967 if (New.IsSplittable && !Partitions[j].IsSplittable)
970 // Grow the new partition to include any equally splittable range. 'j' is
971 // always equally splittable when New is splittable, but when New is not
972 // splittable, we may subsume some (or part of some) splitable partition
973 // without growing the new one.
974 if (New.IsSplittable == Partitions[j].IsSplittable) {
975 New.EndOffset = std::max(New.EndOffset, Partitions[j].EndOffset);
977 assert(!New.IsSplittable);
978 assert(Partitions[j].IsSplittable);
979 SplitEndOffset = std::max(SplitEndOffset, Partitions[j].EndOffset);
982 Partitions[j].kill();
987 // If the new partition is splittable, chop off the end as soon as the
988 // unsplittable subsequent partition starts and ensure we eventually cover
989 // the splittable area.
990 if (j != e && New.IsSplittable) {
991 SplitEndOffset = std::max(SplitEndOffset, New.EndOffset);
992 New.EndOffset = std::min(New.EndOffset, Partitions[j].BeginOffset);
995 // Add the new partition if it differs from the original one and is
996 // non-empty. We can end up with an empty partition here if it was
997 // splittable but there is an unsplittable one that starts at the same
999 if (New != Partitions[i]) {
1000 if (New.BeginOffset != New.EndOffset)
1001 Partitions.push_back(New);
1002 // Mark the old one for removal.
1003 Partitions[i].kill();
1004 ++NumDeadPartitions;
1007 New.BeginOffset = New.EndOffset;
1008 if (!New.IsSplittable) {
1009 New.EndOffset = std::max(New.EndOffset, SplitEndOffset);
1010 if (j != e && !Partitions[j].IsSplittable)
1011 New.EndOffset = std::min(New.EndOffset, Partitions[j].BeginOffset);
1012 New.IsSplittable = true;
1013 // If there is a trailing splittable partition which won't be fused into
1014 // the next splittable partition go ahead and add it onto the partitions
1016 if (New.BeginOffset < New.EndOffset &&
1017 (j == e || !Partitions[j].IsSplittable ||
1018 New.EndOffset < Partitions[j].BeginOffset)) {
1019 Partitions.push_back(New);
1020 New.BeginOffset = New.EndOffset = 0ull;
1025 // Re-sort the partitions now that they have been split and merged into
1026 // disjoint set of partitions. Also remove any of the dead partitions we've
1027 // replaced in the process.
1028 std::sort(Partitions.begin(), Partitions.end());
1029 if (NumDeadPartitions) {
1030 assert(Partitions.back().isDead());
1031 assert((ptrdiff_t)NumDeadPartitions ==
1032 std::count(Partitions.begin(), Partitions.end(), Partitions.back()));
1034 Partitions.erase(Partitions.end() - NumDeadPartitions, Partitions.end());
1037 AllocaPartitioning::AllocaPartitioning(const DataLayout &TD, AllocaInst &AI)
1039 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
1042 PointerEscapingInstr(0) {
1043 PartitionBuilder PB(TD, AI, *this);
1044 PartitionBuilder::PtrInfo PtrI = PB.visitPtr(AI);
1045 if (PtrI.isEscaped() || PtrI.isAborted()) {
1046 // FIXME: We should sink the escape vs. abort info into the caller nicely,
1047 // possibly by just storing the PtrInfo in the AllocaPartitioning.
1048 PointerEscapingInstr = PtrI.getEscapingInst() ? PtrI.getEscapingInst()
1049 : PtrI.getAbortingInst();
1050 assert(PointerEscapingInstr && "Did not track a bad instruction");
1054 // Sort the uses. This arranges for the offsets to be in ascending order,
1055 // and the sizes to be in descending order.
1056 std::sort(Partitions.begin(), Partitions.end());
1058 // Remove any partitions from the back which are marked as dead.
1059 while (!Partitions.empty() && Partitions.back().isDead())
1060 Partitions.pop_back();
1062 if (Partitions.size() > 1) {
1063 // Intersect splittability for all partitions with equal offsets and sizes.
1064 // Then remove all but the first so that we have a sequence of non-equal but
1065 // potentially overlapping partitions.
1066 for (iterator I = Partitions.begin(), J = I, E = Partitions.end(); I != E;
1069 while (J != E && *I == *J) {
1070 I->IsSplittable &= J->IsSplittable;
1074 Partitions.erase(std::unique(Partitions.begin(), Partitions.end()),
1077 // Split splittable and merge unsplittable partitions into a disjoint set
1078 // of partitions over the used space of the allocation.
1079 splitAndMergePartitions();
1082 // Now build up the user lists for each of these disjoint partitions by
1083 // re-walking the recursive users of the alloca.
1084 Uses.resize(Partitions.size());
1085 UseBuilder UB(TD, AI, *this);
1086 PtrI = UB.visitPtr(AI);
1087 assert(!PtrI.isEscaped() && "Previously analyzed pointer now escapes!");
1088 assert(!PtrI.isAborted() && "Early aborted the visit of the pointer.");
1091 Type *AllocaPartitioning::getCommonType(iterator I) const {
1093 for (const_use_iterator UI = use_begin(I), UE = use_end(I); UI != UE; ++UI) {
1095 continue; // Skip dead uses.
1096 if (isa<IntrinsicInst>(*UI->U->getUser()))
1098 if (UI->BeginOffset != I->BeginOffset || UI->EndOffset != I->EndOffset)
1102 if (LoadInst *LI = dyn_cast<LoadInst>(UI->U->getUser())) {
1103 UserTy = LI->getType();
1104 } else if (StoreInst *SI = dyn_cast<StoreInst>(UI->U->getUser())) {
1105 UserTy = SI->getValueOperand()->getType();
1107 return 0; // Bail if we have weird uses.
1110 if (IntegerType *ITy = dyn_cast<IntegerType>(UserTy)) {
1111 // If the type is larger than the partition, skip it. We only encounter
1112 // this for split integer operations where we want to use the type of the
1113 // entity causing the split.
1114 if (ITy->getBitWidth() > (I->EndOffset - I->BeginOffset)*8)
1117 // If we have found an integer type use covering the alloca, use that
1118 // regardless of the other types, as integers are often used for a "bucket
1123 if (Ty && Ty != UserTy)
1131 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
1133 void AllocaPartitioning::print(raw_ostream &OS, const_iterator I,
1134 StringRef Indent) const {
1135 OS << Indent << "partition #" << (I - begin())
1136 << " [" << I->BeginOffset << "," << I->EndOffset << ")"
1137 << (I->IsSplittable ? " (splittable)" : "")
1138 << (Uses[I - begin()].empty() ? " (zero uses)" : "")
1142 void AllocaPartitioning::printUsers(raw_ostream &OS, const_iterator I,
1143 StringRef Indent) const {
1144 for (const_use_iterator UI = use_begin(I), UE = use_end(I);
1147 continue; // Skip dead uses.
1148 OS << Indent << " [" << UI->BeginOffset << "," << UI->EndOffset << ") "
1149 << "used by: " << *UI->U->getUser() << "\n";
1150 if (MemTransferInst *II = dyn_cast<MemTransferInst>(UI->U->getUser())) {
1151 const MemTransferOffsets &MTO = MemTransferInstData.lookup(II);
1153 if (!MTO.IsSplittable)
1154 IsDest = UI->BeginOffset == MTO.DestBegin;
1156 IsDest = MTO.DestBegin != 0u;
1157 OS << Indent << " (original " << (IsDest ? "dest" : "source") << ": "
1158 << "[" << (IsDest ? MTO.DestBegin : MTO.SourceBegin)
1159 << "," << (IsDest ? MTO.DestEnd : MTO.SourceEnd) << ")\n";
1164 void AllocaPartitioning::print(raw_ostream &OS) const {
1165 if (PointerEscapingInstr) {
1166 OS << "No partitioning for alloca: " << AI << "\n"
1167 << " A pointer to this alloca escaped by:\n"
1168 << " " << *PointerEscapingInstr << "\n";
1172 OS << "Partitioning of alloca: " << AI << "\n";
1174 for (const_iterator I = begin(), E = end(); I != E; ++I, ++Num) {
1180 void AllocaPartitioning::dump(const_iterator I) const { print(dbgs(), I); }
1181 void AllocaPartitioning::dump() const { print(dbgs()); }
1183 #endif // !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
1187 /// \brief Implementation of LoadAndStorePromoter for promoting allocas.
1189 /// This subclass of LoadAndStorePromoter adds overrides to handle promoting
1190 /// the loads and stores of an alloca instruction, as well as updating its
1191 /// debug information. This is used when a domtree is unavailable and thus
1192 /// mem2reg in its full form can't be used to handle promotion of allocas to
1194 class AllocaPromoter : public LoadAndStorePromoter {
1198 SmallVector<DbgDeclareInst *, 4> DDIs;
1199 SmallVector<DbgValueInst *, 4> DVIs;
1202 AllocaPromoter(const SmallVectorImpl<Instruction*> &Insts, SSAUpdater &S,
1203 AllocaInst &AI, DIBuilder &DIB)
1204 : LoadAndStorePromoter(Insts, S), AI(AI), DIB(DIB) {}
1206 void run(const SmallVectorImpl<Instruction*> &Insts) {
1207 // Remember which alloca we're promoting (for isInstInList).
1208 if (MDNode *DebugNode = MDNode::getIfExists(AI.getContext(), &AI)) {
1209 for (Value::use_iterator UI = DebugNode->use_begin(),
1210 UE = DebugNode->use_end();
1212 if (DbgDeclareInst *DDI = dyn_cast<DbgDeclareInst>(*UI))
1213 DDIs.push_back(DDI);
1214 else if (DbgValueInst *DVI = dyn_cast<DbgValueInst>(*UI))
1215 DVIs.push_back(DVI);
1218 LoadAndStorePromoter::run(Insts);
1219 AI.eraseFromParent();
1220 while (!DDIs.empty())
1221 DDIs.pop_back_val()->eraseFromParent();
1222 while (!DVIs.empty())
1223 DVIs.pop_back_val()->eraseFromParent();
1226 virtual bool isInstInList(Instruction *I,
1227 const SmallVectorImpl<Instruction*> &Insts) const {
1228 if (LoadInst *LI = dyn_cast<LoadInst>(I))
1229 return LI->getOperand(0) == &AI;
1230 return cast<StoreInst>(I)->getPointerOperand() == &AI;
1233 virtual void updateDebugInfo(Instruction *Inst) const {
1234 for (SmallVector<DbgDeclareInst *, 4>::const_iterator I = DDIs.begin(),
1235 E = DDIs.end(); I != E; ++I) {
1236 DbgDeclareInst *DDI = *I;
1237 if (StoreInst *SI = dyn_cast<StoreInst>(Inst))
1238 ConvertDebugDeclareToDebugValue(DDI, SI, DIB);
1239 else if (LoadInst *LI = dyn_cast<LoadInst>(Inst))
1240 ConvertDebugDeclareToDebugValue(DDI, LI, DIB);
1242 for (SmallVector<DbgValueInst *, 4>::const_iterator I = DVIs.begin(),
1243 E = DVIs.end(); I != E; ++I) {
1244 DbgValueInst *DVI = *I;
1246 if (StoreInst *SI = dyn_cast<StoreInst>(Inst)) {
1247 // If an argument is zero extended then use argument directly. The ZExt
1248 // may be zapped by an optimization pass in future.
1249 if (ZExtInst *ZExt = dyn_cast<ZExtInst>(SI->getOperand(0)))
1250 Arg = dyn_cast<Argument>(ZExt->getOperand(0));
1251 if (SExtInst *SExt = dyn_cast<SExtInst>(SI->getOperand(0)))
1252 Arg = dyn_cast<Argument>(SExt->getOperand(0));
1254 Arg = SI->getOperand(0);
1255 } else if (LoadInst *LI = dyn_cast<LoadInst>(Inst)) {
1256 Arg = LI->getOperand(0);
1260 Instruction *DbgVal =
1261 DIB.insertDbgValueIntrinsic(Arg, 0, DIVariable(DVI->getVariable()),
1263 DbgVal->setDebugLoc(DVI->getDebugLoc());
1267 } // end anon namespace
1271 /// \brief An optimization pass providing Scalar Replacement of Aggregates.
1273 /// This pass takes allocations which can be completely analyzed (that is, they
1274 /// don't escape) and tries to turn them into scalar SSA values. There are
1275 /// a few steps to this process.
1277 /// 1) It takes allocations of aggregates and analyzes the ways in which they
1278 /// are used to try to split them into smaller allocations, ideally of
1279 /// a single scalar data type. It will split up memcpy and memset accesses
1280 /// as necessary and try to isolate invidual scalar accesses.
1281 /// 2) It will transform accesses into forms which are suitable for SSA value
1282 /// promotion. This can be replacing a memset with a scalar store of an
1283 /// integer value, or it can involve speculating operations on a PHI or
1284 /// select to be a PHI or select of the results.
1285 /// 3) Finally, this will try to detect a pattern of accesses which map cleanly
1286 /// onto insert and extract operations on a vector value, and convert them to
1287 /// this form. By doing so, it will enable promotion of vector aggregates to
1288 /// SSA vector values.
1289 class SROA : public FunctionPass {
1290 const bool RequiresDomTree;
1293 const DataLayout *TD;
1296 /// \brief Worklist of alloca instructions to simplify.
1298 /// Each alloca in the function is added to this. Each new alloca formed gets
1299 /// added to it as well to recursively simplify unless that alloca can be
1300 /// directly promoted. Finally, each time we rewrite a use of an alloca other
1301 /// the one being actively rewritten, we add it back onto the list if not
1302 /// already present to ensure it is re-visited.
1303 SetVector<AllocaInst *, SmallVector<AllocaInst *, 16> > Worklist;
1305 /// \brief A collection of instructions to delete.
1306 /// We try to batch deletions to simplify code and make things a bit more
1308 SetVector<Instruction *, SmallVector<Instruction *, 8> > DeadInsts;
1310 /// \brief Post-promotion worklist.
1312 /// Sometimes we discover an alloca which has a high probability of becoming
1313 /// viable for SROA after a round of promotion takes place. In those cases,
1314 /// the alloca is enqueued here for re-processing.
1316 /// Note that we have to be very careful to clear allocas out of this list in
1317 /// the event they are deleted.
1318 SetVector<AllocaInst *, SmallVector<AllocaInst *, 16> > PostPromotionWorklist;
1320 /// \brief A collection of alloca instructions we can directly promote.
1321 std::vector<AllocaInst *> PromotableAllocas;
1324 SROA(bool RequiresDomTree = true)
1325 : FunctionPass(ID), RequiresDomTree(RequiresDomTree),
1326 C(0), TD(0), DT(0) {
1327 initializeSROAPass(*PassRegistry::getPassRegistry());
1329 bool runOnFunction(Function &F);
1330 void getAnalysisUsage(AnalysisUsage &AU) const;
1332 const char *getPassName() const { return "SROA"; }
1336 friend class PHIOrSelectSpeculator;
1337 friend class AllocaPartitionRewriter;
1338 friend class AllocaPartitionVectorRewriter;
1340 bool rewriteAllocaPartition(AllocaInst &AI,
1341 AllocaPartitioning &P,
1342 AllocaPartitioning::iterator PI);
1343 bool splitAlloca(AllocaInst &AI, AllocaPartitioning &P);
1344 bool runOnAlloca(AllocaInst &AI);
1345 void deleteDeadInstructions(SmallPtrSet<AllocaInst *, 4> &DeletedAllocas);
1346 bool promoteAllocas(Function &F);
1352 FunctionPass *llvm::createSROAPass(bool RequiresDomTree) {
1353 return new SROA(RequiresDomTree);
1356 INITIALIZE_PASS_BEGIN(SROA, "sroa", "Scalar Replacement Of Aggregates",
1358 INITIALIZE_PASS_DEPENDENCY(DominatorTree)
1359 INITIALIZE_PASS_END(SROA, "sroa", "Scalar Replacement Of Aggregates",
1363 /// \brief Visitor to speculate PHIs and Selects where possible.
1364 class PHIOrSelectSpeculator : public InstVisitor<PHIOrSelectSpeculator> {
1365 // Befriend the base class so it can delegate to private visit methods.
1366 friend class llvm::InstVisitor<PHIOrSelectSpeculator>;
1368 const DataLayout &TD;
1369 AllocaPartitioning &P;
1373 PHIOrSelectSpeculator(const DataLayout &TD, AllocaPartitioning &P, SROA &Pass)
1374 : TD(TD), P(P), Pass(Pass) {}
1376 /// \brief Visit the users of an alloca partition and rewrite them.
1377 void visitUsers(AllocaPartitioning::const_iterator PI) {
1378 // Note that we need to use an index here as the underlying vector of uses
1379 // may be grown during speculation. However, we never need to re-visit the
1380 // new uses, and so we can use the initial size bound.
1381 for (unsigned Idx = 0, Size = P.use_size(PI); Idx != Size; ++Idx) {
1382 const AllocaPartitioning::PartitionUse &PU = P.getUse(PI, Idx);
1384 continue; // Skip dead use.
1386 visit(cast<Instruction>(PU.U->getUser()));
1391 // By default, skip this instruction.
1392 void visitInstruction(Instruction &I) {}
1394 /// PHI instructions that use an alloca and are subsequently loaded can be
1395 /// rewritten to load both input pointers in the pred blocks and then PHI the
1396 /// results, allowing the load of the alloca to be promoted.
1398 /// %P2 = phi [i32* %Alloca, i32* %Other]
1399 /// %V = load i32* %P2
1401 /// %V1 = load i32* %Alloca -> will be mem2reg'd
1403 /// %V2 = load i32* %Other
1405 /// %V = phi [i32 %V1, i32 %V2]
1407 /// We can do this to a select if its only uses are loads and if the operands
1408 /// to the select can be loaded unconditionally.
1410 /// FIXME: This should be hoisted into a generic utility, likely in
1411 /// Transforms/Util/Local.h
1412 bool isSafePHIToSpeculate(PHINode &PN, SmallVectorImpl<LoadInst *> &Loads) {
1413 // For now, we can only do this promotion if the load is in the same block
1414 // as the PHI, and if there are no stores between the phi and load.
1415 // TODO: Allow recursive phi users.
1416 // TODO: Allow stores.
1417 BasicBlock *BB = PN.getParent();
1418 unsigned MaxAlign = 0;
1419 for (Value::use_iterator UI = PN.use_begin(), UE = PN.use_end();
1421 LoadInst *LI = dyn_cast<LoadInst>(*UI);
1422 if (LI == 0 || !LI->isSimple()) return false;
1424 // For now we only allow loads in the same block as the PHI. This is
1425 // a common case that happens when instcombine merges two loads through
1427 if (LI->getParent() != BB) return false;
1429 // Ensure that there are no instructions between the PHI and the load that
1431 for (BasicBlock::iterator BBI = &PN; &*BBI != LI; ++BBI)
1432 if (BBI->mayWriteToMemory())
1435 MaxAlign = std::max(MaxAlign, LI->getAlignment());
1436 Loads.push_back(LI);
1439 // We can only transform this if it is safe to push the loads into the
1440 // predecessor blocks. The only thing to watch out for is that we can't put
1441 // a possibly trapping load in the predecessor if it is a critical edge.
1442 for (unsigned Idx = 0, Num = PN.getNumIncomingValues(); Idx != Num;
1444 TerminatorInst *TI = PN.getIncomingBlock(Idx)->getTerminator();
1445 Value *InVal = PN.getIncomingValue(Idx);
1447 // If the value is produced by the terminator of the predecessor (an
1448 // invoke) or it has side-effects, there is no valid place to put a load
1449 // in the predecessor.
1450 if (TI == InVal || TI->mayHaveSideEffects())
1453 // If the predecessor has a single successor, then the edge isn't
1455 if (TI->getNumSuccessors() == 1)
1458 // If this pointer is always safe to load, or if we can prove that there
1459 // is already a load in the block, then we can move the load to the pred
1461 if (InVal->isDereferenceablePointer() ||
1462 isSafeToLoadUnconditionally(InVal, TI, MaxAlign, &TD))
1471 void visitPHINode(PHINode &PN) {
1472 DEBUG(dbgs() << " original: " << PN << "\n");
1474 SmallVector<LoadInst *, 4> Loads;
1475 if (!isSafePHIToSpeculate(PN, Loads))
1478 assert(!Loads.empty());
1480 Type *LoadTy = cast<PointerType>(PN.getType())->getElementType();
1481 IRBuilder<> PHIBuilder(&PN);
1482 PHINode *NewPN = PHIBuilder.CreatePHI(LoadTy, PN.getNumIncomingValues(),
1483 PN.getName() + ".sroa.speculated");
1485 // Get the TBAA tag and alignment to use from one of the loads. It doesn't
1486 // matter which one we get and if any differ, it doesn't matter.
1487 LoadInst *SomeLoad = cast<LoadInst>(Loads.back());
1488 MDNode *TBAATag = SomeLoad->getMetadata(LLVMContext::MD_tbaa);
1489 unsigned Align = SomeLoad->getAlignment();
1491 // Rewrite all loads of the PN to use the new PHI.
1493 LoadInst *LI = Loads.pop_back_val();
1494 LI->replaceAllUsesWith(NewPN);
1495 Pass.DeadInsts.insert(LI);
1496 } while (!Loads.empty());
1498 // Inject loads into all of the pred blocks.
1499 for (unsigned Idx = 0, Num = PN.getNumIncomingValues(); Idx != Num; ++Idx) {
1500 BasicBlock *Pred = PN.getIncomingBlock(Idx);
1501 TerminatorInst *TI = Pred->getTerminator();
1502 Use *InUse = &PN.getOperandUse(PN.getOperandNumForIncomingValue(Idx));
1503 Value *InVal = PN.getIncomingValue(Idx);
1504 IRBuilder<> PredBuilder(TI);
1507 = PredBuilder.CreateLoad(InVal, (PN.getName() + ".sroa.speculate.load." +
1509 ++NumLoadsSpeculated;
1510 Load->setAlignment(Align);
1512 Load->setMetadata(LLVMContext::MD_tbaa, TBAATag);
1513 NewPN->addIncoming(Load, Pred);
1515 Instruction *Ptr = dyn_cast<Instruction>(InVal);
1517 // No uses to rewrite.
1520 // Try to lookup and rewrite any partition uses corresponding to this phi
1522 AllocaPartitioning::iterator PI
1523 = P.findPartitionForPHIOrSelectOperand(InUse);
1527 // Replace the Use in the PartitionUse for this operand with the Use
1529 AllocaPartitioning::use_iterator UI
1530 = P.findPartitionUseForPHIOrSelectOperand(InUse);
1531 assert(isa<PHINode>(*UI->U->getUser()));
1532 UI->U = &Load->getOperandUse(Load->getPointerOperandIndex());
1534 DEBUG(dbgs() << " speculated to: " << *NewPN << "\n");
1537 /// Select instructions that use an alloca and are subsequently loaded can be
1538 /// rewritten to load both input pointers and then select between the result,
1539 /// allowing the load of the alloca to be promoted.
1541 /// %P2 = select i1 %cond, i32* %Alloca, i32* %Other
1542 /// %V = load i32* %P2
1544 /// %V1 = load i32* %Alloca -> will be mem2reg'd
1545 /// %V2 = load i32* %Other
1546 /// %V = select i1 %cond, i32 %V1, i32 %V2
1548 /// We can do this to a select if its only uses are loads and if the operand
1549 /// to the select can be loaded unconditionally.
1550 bool isSafeSelectToSpeculate(SelectInst &SI,
1551 SmallVectorImpl<LoadInst *> &Loads) {
1552 Value *TValue = SI.getTrueValue();
1553 Value *FValue = SI.getFalseValue();
1554 bool TDerefable = TValue->isDereferenceablePointer();
1555 bool FDerefable = FValue->isDereferenceablePointer();
1557 for (Value::use_iterator UI = SI.use_begin(), UE = SI.use_end();
1559 LoadInst *LI = dyn_cast<LoadInst>(*UI);
1560 if (LI == 0 || !LI->isSimple()) return false;
1562 // Both operands to the select need to be dereferencable, either
1563 // absolutely (e.g. allocas) or at this point because we can see other
1565 if (!TDerefable && !isSafeToLoadUnconditionally(TValue, LI,
1566 LI->getAlignment(), &TD))
1568 if (!FDerefable && !isSafeToLoadUnconditionally(FValue, LI,
1569 LI->getAlignment(), &TD))
1571 Loads.push_back(LI);
1577 void visitSelectInst(SelectInst &SI) {
1578 DEBUG(dbgs() << " original: " << SI << "\n");
1579 IRBuilder<> IRB(&SI);
1581 // If the select isn't safe to speculate, just use simple logic to emit it.
1582 SmallVector<LoadInst *, 4> Loads;
1583 if (!isSafeSelectToSpeculate(SI, Loads))
1586 Use *Ops[2] = { &SI.getOperandUse(1), &SI.getOperandUse(2) };
1587 AllocaPartitioning::iterator PIs[2];
1588 AllocaPartitioning::PartitionUse PUs[2];
1589 for (unsigned i = 0, e = 2; i != e; ++i) {
1590 PIs[i] = P.findPartitionForPHIOrSelectOperand(Ops[i]);
1591 if (PIs[i] != P.end()) {
1592 // If the pointer is within the partitioning, remove the select from
1593 // its uses. We'll add in the new loads below.
1594 AllocaPartitioning::use_iterator UI
1595 = P.findPartitionUseForPHIOrSelectOperand(Ops[i]);
1597 // Clear out the use here so that the offsets into the use list remain
1598 // stable but this use is ignored when rewriting.
1603 Value *TV = SI.getTrueValue();
1604 Value *FV = SI.getFalseValue();
1605 // Replace the loads of the select with a select of two loads.
1606 while (!Loads.empty()) {
1607 LoadInst *LI = Loads.pop_back_val();
1609 IRB.SetInsertPoint(LI);
1611 IRB.CreateLoad(TV, LI->getName() + ".sroa.speculate.load.true");
1613 IRB.CreateLoad(FV, LI->getName() + ".sroa.speculate.load.false");
1614 NumLoadsSpeculated += 2;
1616 // Transfer alignment and TBAA info if present.
1617 TL->setAlignment(LI->getAlignment());
1618 FL->setAlignment(LI->getAlignment());
1619 if (MDNode *Tag = LI->getMetadata(LLVMContext::MD_tbaa)) {
1620 TL->setMetadata(LLVMContext::MD_tbaa, Tag);
1621 FL->setMetadata(LLVMContext::MD_tbaa, Tag);
1624 Value *V = IRB.CreateSelect(SI.getCondition(), TL, FL,
1625 LI->getName() + ".sroa.speculated");
1627 LoadInst *Loads[2] = { TL, FL };
1628 for (unsigned i = 0, e = 2; i != e; ++i) {
1629 if (PIs[i] != P.end()) {
1630 Use *LoadUse = &Loads[i]->getOperandUse(0);
1631 assert(PUs[i].U->get() == LoadUse->get());
1633 P.use_push_back(PIs[i], PUs[i]);
1637 DEBUG(dbgs() << " speculated to: " << *V << "\n");
1638 LI->replaceAllUsesWith(V);
1639 Pass.DeadInsts.insert(LI);
1645 /// \brief Build a GEP out of a base pointer and indices.
1647 /// This will return the BasePtr if that is valid, or build a new GEP
1648 /// instruction using the IRBuilder if GEP-ing is needed.
1649 static Value *buildGEP(IRBuilder<> &IRB, Value *BasePtr,
1650 SmallVectorImpl<Value *> &Indices,
1651 const Twine &Prefix) {
1652 if (Indices.empty())
1655 // A single zero index is a no-op, so check for this and avoid building a GEP
1657 if (Indices.size() == 1 && cast<ConstantInt>(Indices.back())->isZero())
1660 return IRB.CreateInBoundsGEP(BasePtr, Indices, Prefix + ".idx");
1663 /// \brief Get a natural GEP off of the BasePtr walking through Ty toward
1664 /// TargetTy without changing the offset of the pointer.
1666 /// This routine assumes we've already established a properly offset GEP with
1667 /// Indices, and arrived at the Ty type. The goal is to continue to GEP with
1668 /// zero-indices down through type layers until we find one the same as
1669 /// TargetTy. If we can't find one with the same type, we at least try to use
1670 /// one with the same size. If none of that works, we just produce the GEP as
1671 /// indicated by Indices to have the correct offset.
1672 static Value *getNaturalGEPWithType(IRBuilder<> &IRB, const DataLayout &TD,
1673 Value *BasePtr, Type *Ty, Type *TargetTy,
1674 SmallVectorImpl<Value *> &Indices,
1675 const Twine &Prefix) {
1677 return buildGEP(IRB, BasePtr, Indices, Prefix);
1679 // See if we can descend into a struct and locate a field with the correct
1681 unsigned NumLayers = 0;
1682 Type *ElementTy = Ty;
1684 if (ElementTy->isPointerTy())
1686 if (SequentialType *SeqTy = dyn_cast<SequentialType>(ElementTy)) {
1687 ElementTy = SeqTy->getElementType();
1688 // Note that we use the default address space as this index is over an
1689 // array or a vector, not a pointer.
1690 Indices.push_back(IRB.getInt(APInt(TD.getPointerSizeInBits(0), 0)));
1691 } else if (StructType *STy = dyn_cast<StructType>(ElementTy)) {
1692 if (STy->element_begin() == STy->element_end())
1693 break; // Nothing left to descend into.
1694 ElementTy = *STy->element_begin();
1695 Indices.push_back(IRB.getInt32(0));
1700 } while (ElementTy != TargetTy);
1701 if (ElementTy != TargetTy)
1702 Indices.erase(Indices.end() - NumLayers, Indices.end());
1704 return buildGEP(IRB, BasePtr, Indices, Prefix);
1707 /// \brief Recursively compute indices for a natural GEP.
1709 /// This is the recursive step for getNaturalGEPWithOffset that walks down the
1710 /// element types adding appropriate indices for the GEP.
1711 static Value *getNaturalGEPRecursively(IRBuilder<> &IRB, const DataLayout &TD,
1712 Value *Ptr, Type *Ty, APInt &Offset,
1714 SmallVectorImpl<Value *> &Indices,
1715 const Twine &Prefix) {
1717 return getNaturalGEPWithType(IRB, TD, Ptr, Ty, TargetTy, Indices, Prefix);
1719 // We can't recurse through pointer types.
1720 if (Ty->isPointerTy())
1723 // We try to analyze GEPs over vectors here, but note that these GEPs are
1724 // extremely poorly defined currently. The long-term goal is to remove GEPing
1725 // over a vector from the IR completely.
1726 if (VectorType *VecTy = dyn_cast<VectorType>(Ty)) {
1727 unsigned ElementSizeInBits = TD.getTypeSizeInBits(VecTy->getScalarType());
1728 if (ElementSizeInBits % 8)
1729 return 0; // GEPs over non-multiple of 8 size vector elements are invalid.
1730 APInt ElementSize(Offset.getBitWidth(), ElementSizeInBits / 8);
1731 APInt NumSkippedElements = Offset.sdiv(ElementSize);
1732 if (NumSkippedElements.ugt(VecTy->getNumElements()))
1734 Offset -= NumSkippedElements * ElementSize;
1735 Indices.push_back(IRB.getInt(NumSkippedElements));
1736 return getNaturalGEPRecursively(IRB, TD, Ptr, VecTy->getElementType(),
1737 Offset, TargetTy, Indices, Prefix);
1740 if (ArrayType *ArrTy = dyn_cast<ArrayType>(Ty)) {
1741 Type *ElementTy = ArrTy->getElementType();
1742 APInt ElementSize(Offset.getBitWidth(), TD.getTypeAllocSize(ElementTy));
1743 APInt NumSkippedElements = Offset.sdiv(ElementSize);
1744 if (NumSkippedElements.ugt(ArrTy->getNumElements()))
1747 Offset -= NumSkippedElements * ElementSize;
1748 Indices.push_back(IRB.getInt(NumSkippedElements));
1749 return getNaturalGEPRecursively(IRB, TD, Ptr, ElementTy, Offset, TargetTy,
1753 StructType *STy = dyn_cast<StructType>(Ty);
1757 const StructLayout *SL = TD.getStructLayout(STy);
1758 uint64_t StructOffset = Offset.getZExtValue();
1759 if (StructOffset >= SL->getSizeInBytes())
1761 unsigned Index = SL->getElementContainingOffset(StructOffset);
1762 Offset -= APInt(Offset.getBitWidth(), SL->getElementOffset(Index));
1763 Type *ElementTy = STy->getElementType(Index);
1764 if (Offset.uge(TD.getTypeAllocSize(ElementTy)))
1765 return 0; // The offset points into alignment padding.
1767 Indices.push_back(IRB.getInt32(Index));
1768 return getNaturalGEPRecursively(IRB, TD, Ptr, ElementTy, Offset, TargetTy,
1772 /// \brief Get a natural GEP from a base pointer to a particular offset and
1773 /// resulting in a particular type.
1775 /// The goal is to produce a "natural" looking GEP that works with the existing
1776 /// composite types to arrive at the appropriate offset and element type for
1777 /// a pointer. TargetTy is the element type the returned GEP should point-to if
1778 /// possible. We recurse by decreasing Offset, adding the appropriate index to
1779 /// Indices, and setting Ty to the result subtype.
1781 /// If no natural GEP can be constructed, this function returns null.
1782 static Value *getNaturalGEPWithOffset(IRBuilder<> &IRB, const DataLayout &TD,
1783 Value *Ptr, APInt Offset, Type *TargetTy,
1784 SmallVectorImpl<Value *> &Indices,
1785 const Twine &Prefix) {
1786 PointerType *Ty = cast<PointerType>(Ptr->getType());
1788 // Don't consider any GEPs through an i8* as natural unless the TargetTy is
1790 if (Ty == IRB.getInt8PtrTy() && TargetTy->isIntegerTy(8))
1793 Type *ElementTy = Ty->getElementType();
1794 if (!ElementTy->isSized())
1795 return 0; // We can't GEP through an unsized element.
1796 APInt ElementSize(Offset.getBitWidth(), TD.getTypeAllocSize(ElementTy));
1797 if (ElementSize == 0)
1798 return 0; // Zero-length arrays can't help us build a natural GEP.
1799 APInt NumSkippedElements = Offset.sdiv(ElementSize);
1801 Offset -= NumSkippedElements * ElementSize;
1802 Indices.push_back(IRB.getInt(NumSkippedElements));
1803 return getNaturalGEPRecursively(IRB, TD, Ptr, ElementTy, Offset, TargetTy,
1807 /// \brief Compute an adjusted pointer from Ptr by Offset bytes where the
1808 /// resulting pointer has PointerTy.
1810 /// This tries very hard to compute a "natural" GEP which arrives at the offset
1811 /// and produces the pointer type desired. Where it cannot, it will try to use
1812 /// the natural GEP to arrive at the offset and bitcast to the type. Where that
1813 /// fails, it will try to use an existing i8* and GEP to the byte offset and
1814 /// bitcast to the type.
1816 /// The strategy for finding the more natural GEPs is to peel off layers of the
1817 /// pointer, walking back through bit casts and GEPs, searching for a base
1818 /// pointer from which we can compute a natural GEP with the desired
1819 /// properities. The algorithm tries to fold as many constant indices into
1820 /// a single GEP as possible, thus making each GEP more independent of the
1821 /// surrounding code.
1822 static Value *getAdjustedPtr(IRBuilder<> &IRB, const DataLayout &TD,
1823 Value *Ptr, APInt Offset, Type *PointerTy,
1824 const Twine &Prefix) {
1825 // Even though we don't look through PHI nodes, we could be called on an
1826 // instruction in an unreachable block, which may be on a cycle.
1827 SmallPtrSet<Value *, 4> Visited;
1828 Visited.insert(Ptr);
1829 SmallVector<Value *, 4> Indices;
1831 // We may end up computing an offset pointer that has the wrong type. If we
1832 // never are able to compute one directly that has the correct type, we'll
1833 // fall back to it, so keep it around here.
1834 Value *OffsetPtr = 0;
1836 // Remember any i8 pointer we come across to re-use if we need to do a raw
1839 APInt Int8PtrOffset(Offset.getBitWidth(), 0);
1841 Type *TargetTy = PointerTy->getPointerElementType();
1844 // First fold any existing GEPs into the offset.
1845 while (GEPOperator *GEP = dyn_cast<GEPOperator>(Ptr)) {
1846 APInt GEPOffset(Offset.getBitWidth(), 0);
1847 if (!GEP->accumulateConstantOffset(TD, GEPOffset))
1849 Offset += GEPOffset;
1850 Ptr = GEP->getPointerOperand();
1851 if (!Visited.insert(Ptr))
1855 // See if we can perform a natural GEP here.
1857 if (Value *P = getNaturalGEPWithOffset(IRB, TD, Ptr, Offset, TargetTy,
1859 if (P->getType() == PointerTy) {
1860 // Zap any offset pointer that we ended up computing in previous rounds.
1861 if (OffsetPtr && OffsetPtr->use_empty())
1862 if (Instruction *I = dyn_cast<Instruction>(OffsetPtr))
1863 I->eraseFromParent();
1871 // Stash this pointer if we've found an i8*.
1872 if (Ptr->getType()->isIntegerTy(8)) {
1874 Int8PtrOffset = Offset;
1877 // Peel off a layer of the pointer and update the offset appropriately.
1878 if (Operator::getOpcode(Ptr) == Instruction::BitCast) {
1879 Ptr = cast<Operator>(Ptr)->getOperand(0);
1880 } else if (GlobalAlias *GA = dyn_cast<GlobalAlias>(Ptr)) {
1881 if (GA->mayBeOverridden())
1883 Ptr = GA->getAliasee();
1887 assert(Ptr->getType()->isPointerTy() && "Unexpected operand type!");
1888 } while (Visited.insert(Ptr));
1892 Int8Ptr = IRB.CreateBitCast(Ptr, IRB.getInt8PtrTy(),
1893 Prefix + ".raw_cast");
1894 Int8PtrOffset = Offset;
1897 OffsetPtr = Int8PtrOffset == 0 ? Int8Ptr :
1898 IRB.CreateInBoundsGEP(Int8Ptr, IRB.getInt(Int8PtrOffset),
1899 Prefix + ".raw_idx");
1903 // On the off chance we were targeting i8*, guard the bitcast here.
1904 if (Ptr->getType() != PointerTy)
1905 Ptr = IRB.CreateBitCast(Ptr, PointerTy, Prefix + ".cast");
1910 /// \brief Test whether we can convert a value from the old to the new type.
1912 /// This predicate should be used to guard calls to convertValue in order to
1913 /// ensure that we only try to convert viable values. The strategy is that we
1914 /// will peel off single element struct and array wrappings to get to an
1915 /// underlying value, and convert that value.
1916 static bool canConvertValue(const DataLayout &DL, Type *OldTy, Type *NewTy) {
1919 if (DL.getTypeSizeInBits(NewTy) != DL.getTypeSizeInBits(OldTy))
1921 if (!NewTy->isSingleValueType() || !OldTy->isSingleValueType())
1924 if (NewTy->isPointerTy() || OldTy->isPointerTy()) {
1925 if (NewTy->isPointerTy() && OldTy->isPointerTy())
1927 if (NewTy->isIntegerTy() || OldTy->isIntegerTy())
1935 /// \brief Generic routine to convert an SSA value to a value of a different
1938 /// This will try various different casting techniques, such as bitcasts,
1939 /// inttoptr, and ptrtoint casts. Use the \c canConvertValue predicate to test
1940 /// two types for viability with this routine.
1941 static Value *convertValue(const DataLayout &DL, IRBuilder<> &IRB, Value *V,
1943 assert(canConvertValue(DL, V->getType(), Ty) &&
1944 "Value not convertable to type");
1945 if (V->getType() == Ty)
1947 if (V->getType()->isIntegerTy() && Ty->isPointerTy())
1948 return IRB.CreateIntToPtr(V, Ty);
1949 if (V->getType()->isPointerTy() && Ty->isIntegerTy())
1950 return IRB.CreatePtrToInt(V, Ty);
1952 return IRB.CreateBitCast(V, Ty);
1955 /// \brief Test whether the given alloca partition can be promoted to a vector.
1957 /// This is a quick test to check whether we can rewrite a particular alloca
1958 /// partition (and its newly formed alloca) into a vector alloca with only
1959 /// whole-vector loads and stores such that it could be promoted to a vector
1960 /// SSA value. We only can ensure this for a limited set of operations, and we
1961 /// don't want to do the rewrites unless we are confident that the result will
1962 /// be promotable, so we have an early test here.
1963 static bool isVectorPromotionViable(const DataLayout &TD,
1965 AllocaPartitioning &P,
1966 uint64_t PartitionBeginOffset,
1967 uint64_t PartitionEndOffset,
1968 AllocaPartitioning::const_use_iterator I,
1969 AllocaPartitioning::const_use_iterator E) {
1970 VectorType *Ty = dyn_cast<VectorType>(AllocaTy);
1974 uint64_t VecSize = TD.getTypeSizeInBits(Ty);
1975 uint64_t ElementSize = TD.getTypeSizeInBits(Ty->getScalarType());
1977 // While the definition of LLVM vectors is bitpacked, we don't support sizes
1978 // that aren't byte sized.
1979 if (ElementSize % 8)
1981 assert((VecSize % 8) == 0 && "vector size not a multiple of element size?");
1985 for (; I != E; ++I) {
1987 continue; // Skip dead use.
1989 uint64_t BeginOffset = I->BeginOffset - PartitionBeginOffset;
1990 uint64_t BeginIndex = BeginOffset / ElementSize;
1991 if (BeginIndex * ElementSize != BeginOffset ||
1992 BeginIndex >= Ty->getNumElements())
1994 uint64_t EndOffset = I->EndOffset - PartitionBeginOffset;
1995 uint64_t EndIndex = EndOffset / ElementSize;
1996 if (EndIndex * ElementSize != EndOffset ||
1997 EndIndex > Ty->getNumElements())
2000 assert(EndIndex > BeginIndex && "Empty vector!");
2001 uint64_t NumElements = EndIndex - BeginIndex;
2003 = (NumElements == 1) ? Ty->getElementType()
2004 : VectorType::get(Ty->getElementType(), NumElements);
2006 if (MemIntrinsic *MI = dyn_cast<MemIntrinsic>(I->U->getUser())) {
2007 if (MI->isVolatile())
2009 if (MemTransferInst *MTI = dyn_cast<MemTransferInst>(I->U->getUser())) {
2010 const AllocaPartitioning::MemTransferOffsets &MTO
2011 = P.getMemTransferOffsets(*MTI);
2012 if (!MTO.IsSplittable)
2015 } else if (I->U->get()->getType()->getPointerElementType()->isStructTy()) {
2016 // Disable vector promotion when there are loads or stores of an FCA.
2018 } else if (LoadInst *LI = dyn_cast<LoadInst>(I->U->getUser())) {
2019 if (LI->isVolatile())
2021 if (!canConvertValue(TD, PartitionTy, LI->getType()))
2023 } else if (StoreInst *SI = dyn_cast<StoreInst>(I->U->getUser())) {
2024 if (SI->isVolatile())
2026 if (!canConvertValue(TD, SI->getValueOperand()->getType(), PartitionTy))
2035 /// \brief Test whether the given alloca partition's integer operations can be
2036 /// widened to promotable ones.
2038 /// This is a quick test to check whether we can rewrite the integer loads and
2039 /// stores to a particular alloca into wider loads and stores and be able to
2040 /// promote the resulting alloca.
2041 static bool isIntegerWideningViable(const DataLayout &TD,
2043 uint64_t AllocBeginOffset,
2044 AllocaPartitioning &P,
2045 AllocaPartitioning::const_use_iterator I,
2046 AllocaPartitioning::const_use_iterator E) {
2047 uint64_t SizeInBits = TD.getTypeSizeInBits(AllocaTy);
2048 // Don't create integer types larger than the maximum bitwidth.
2049 if (SizeInBits > IntegerType::MAX_INT_BITS)
2052 // Don't try to handle allocas with bit-padding.
2053 if (SizeInBits != TD.getTypeStoreSizeInBits(AllocaTy))
2056 // We need to ensure that an integer type with the appropriate bitwidth can
2057 // be converted to the alloca type, whatever that is. We don't want to force
2058 // the alloca itself to have an integer type if there is a more suitable one.
2059 Type *IntTy = Type::getIntNTy(AllocaTy->getContext(), SizeInBits);
2060 if (!canConvertValue(TD, AllocaTy, IntTy) ||
2061 !canConvertValue(TD, IntTy, AllocaTy))
2064 uint64_t Size = TD.getTypeStoreSize(AllocaTy);
2066 // Check the uses to ensure the uses are (likely) promoteable integer uses.
2067 // Also ensure that the alloca has a covering load or store. We don't want
2068 // to widen the integer operotains only to fail to promote due to some other
2069 // unsplittable entry (which we may make splittable later).
2070 bool WholeAllocaOp = false;
2071 for (; I != E; ++I) {
2073 continue; // Skip dead use.
2075 uint64_t RelBegin = I->BeginOffset - AllocBeginOffset;
2076 uint64_t RelEnd = I->EndOffset - AllocBeginOffset;
2078 // We can't reasonably handle cases where the load or store extends past
2079 // the end of the aloca's type and into its padding.
2083 if (LoadInst *LI = dyn_cast<LoadInst>(I->U->getUser())) {
2084 if (LI->isVolatile())
2086 if (RelBegin == 0 && RelEnd == Size)
2087 WholeAllocaOp = true;
2088 if (IntegerType *ITy = dyn_cast<IntegerType>(LI->getType())) {
2089 if (ITy->getBitWidth() < TD.getTypeStoreSizeInBits(ITy))
2093 // Non-integer loads need to be convertible from the alloca type so that
2094 // they are promotable.
2095 if (RelBegin != 0 || RelEnd != Size ||
2096 !canConvertValue(TD, AllocaTy, LI->getType()))
2098 } else if (StoreInst *SI = dyn_cast<StoreInst>(I->U->getUser())) {
2099 Type *ValueTy = SI->getValueOperand()->getType();
2100 if (SI->isVolatile())
2102 if (RelBegin == 0 && RelEnd == Size)
2103 WholeAllocaOp = true;
2104 if (IntegerType *ITy = dyn_cast<IntegerType>(ValueTy)) {
2105 if (ITy->getBitWidth() < TD.getTypeStoreSizeInBits(ITy))
2109 // Non-integer stores need to be convertible to the alloca type so that
2110 // they are promotable.
2111 if (RelBegin != 0 || RelEnd != Size ||
2112 !canConvertValue(TD, ValueTy, AllocaTy))
2114 } else if (MemIntrinsic *MI = dyn_cast<MemIntrinsic>(I->U->getUser())) {
2115 if (MI->isVolatile() || !isa<Constant>(MI->getLength()))
2117 if (MemTransferInst *MTI = dyn_cast<MemTransferInst>(I->U->getUser())) {
2118 const AllocaPartitioning::MemTransferOffsets &MTO
2119 = P.getMemTransferOffsets(*MTI);
2120 if (!MTO.IsSplittable)
2123 } else if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(I->U->getUser())) {
2124 if (II->getIntrinsicID() != Intrinsic::lifetime_start &&
2125 II->getIntrinsicID() != Intrinsic::lifetime_end)
2131 return WholeAllocaOp;
2134 static Value *extractInteger(const DataLayout &DL, IRBuilder<> &IRB, Value *V,
2135 IntegerType *Ty, uint64_t Offset,
2136 const Twine &Name) {
2137 DEBUG(dbgs() << " start: " << *V << "\n");
2138 IntegerType *IntTy = cast<IntegerType>(V->getType());
2139 assert(DL.getTypeStoreSize(Ty) + Offset <= DL.getTypeStoreSize(IntTy) &&
2140 "Element extends past full value");
2141 uint64_t ShAmt = 8*Offset;
2142 if (DL.isBigEndian())
2143 ShAmt = 8*(DL.getTypeStoreSize(IntTy) - DL.getTypeStoreSize(Ty) - Offset);
2145 V = IRB.CreateLShr(V, ShAmt, Name + ".shift");
2146 DEBUG(dbgs() << " shifted: " << *V << "\n");
2148 assert(Ty->getBitWidth() <= IntTy->getBitWidth() &&
2149 "Cannot extract to a larger integer!");
2151 V = IRB.CreateTrunc(V, Ty, Name + ".trunc");
2152 DEBUG(dbgs() << " trunced: " << *V << "\n");
2157 static Value *insertInteger(const DataLayout &DL, IRBuilder<> &IRB, Value *Old,
2158 Value *V, uint64_t Offset, const Twine &Name) {
2159 IntegerType *IntTy = cast<IntegerType>(Old->getType());
2160 IntegerType *Ty = cast<IntegerType>(V->getType());
2161 assert(Ty->getBitWidth() <= IntTy->getBitWidth() &&
2162 "Cannot insert a larger integer!");
2163 DEBUG(dbgs() << " start: " << *V << "\n");
2165 V = IRB.CreateZExt(V, IntTy, Name + ".ext");
2166 DEBUG(dbgs() << " extended: " << *V << "\n");
2168 assert(DL.getTypeStoreSize(Ty) + Offset <= DL.getTypeStoreSize(IntTy) &&
2169 "Element store outside of alloca store");
2170 uint64_t ShAmt = 8*Offset;
2171 if (DL.isBigEndian())
2172 ShAmt = 8*(DL.getTypeStoreSize(IntTy) - DL.getTypeStoreSize(Ty) - Offset);
2174 V = IRB.CreateShl(V, ShAmt, Name + ".shift");
2175 DEBUG(dbgs() << " shifted: " << *V << "\n");
2178 if (ShAmt || Ty->getBitWidth() < IntTy->getBitWidth()) {
2179 APInt Mask = ~Ty->getMask().zext(IntTy->getBitWidth()).shl(ShAmt);
2180 Old = IRB.CreateAnd(Old, Mask, Name + ".mask");
2181 DEBUG(dbgs() << " masked: " << *Old << "\n");
2182 V = IRB.CreateOr(Old, V, Name + ".insert");
2183 DEBUG(dbgs() << " inserted: " << *V << "\n");
2188 static Value *extractVector(IRBuilder<> &IRB, Value *V,
2189 unsigned BeginIndex, unsigned EndIndex,
2190 const Twine &Name) {
2191 VectorType *VecTy = cast<VectorType>(V->getType());
2192 unsigned NumElements = EndIndex - BeginIndex;
2193 assert(NumElements <= VecTy->getNumElements() && "Too many elements!");
2195 if (NumElements == VecTy->getNumElements())
2198 if (NumElements == 1) {
2199 V = IRB.CreateExtractElement(V, IRB.getInt32(BeginIndex),
2201 DEBUG(dbgs() << " extract: " << *V << "\n");
2205 SmallVector<Constant*, 8> Mask;
2206 Mask.reserve(NumElements);
2207 for (unsigned i = BeginIndex; i != EndIndex; ++i)
2208 Mask.push_back(IRB.getInt32(i));
2209 V = IRB.CreateShuffleVector(V, UndefValue::get(V->getType()),
2210 ConstantVector::get(Mask),
2212 DEBUG(dbgs() << " shuffle: " << *V << "\n");
2216 static Value *insertVector(IRBuilder<> &IRB, Value *Old, Value *V,
2217 unsigned BeginIndex, const Twine &Name) {
2218 VectorType *VecTy = cast<VectorType>(Old->getType());
2219 assert(VecTy && "Can only insert a vector into a vector");
2221 VectorType *Ty = dyn_cast<VectorType>(V->getType());
2223 // Single element to insert.
2224 V = IRB.CreateInsertElement(Old, V, IRB.getInt32(BeginIndex),
2226 DEBUG(dbgs() << " insert: " << *V << "\n");
2230 assert(Ty->getNumElements() <= VecTy->getNumElements() &&
2231 "Too many elements!");
2232 if (Ty->getNumElements() == VecTy->getNumElements()) {
2233 assert(V->getType() == VecTy && "Vector type mismatch");
2236 unsigned EndIndex = BeginIndex + Ty->getNumElements();
2238 // When inserting a smaller vector into the larger to store, we first
2239 // use a shuffle vector to widen it with undef elements, and then
2240 // a second shuffle vector to select between the loaded vector and the
2242 SmallVector<Constant*, 8> Mask;
2243 Mask.reserve(VecTy->getNumElements());
2244 for (unsigned i = 0; i != VecTy->getNumElements(); ++i)
2245 if (i >= BeginIndex && i < EndIndex)
2246 Mask.push_back(IRB.getInt32(i - BeginIndex));
2248 Mask.push_back(UndefValue::get(IRB.getInt32Ty()));
2249 V = IRB.CreateShuffleVector(V, UndefValue::get(V->getType()),
2250 ConstantVector::get(Mask),
2252 DEBUG(dbgs() << " shuffle1: " << *V << "\n");
2255 for (unsigned i = 0; i != VecTy->getNumElements(); ++i)
2256 if (i >= BeginIndex && i < EndIndex)
2257 Mask.push_back(IRB.getInt32(i));
2259 Mask.push_back(IRB.getInt32(i + VecTy->getNumElements()));
2260 V = IRB.CreateShuffleVector(V, Old, ConstantVector::get(Mask),
2262 DEBUG(dbgs() << " shuffle2: " << *V << "\n");
2267 /// \brief Visitor to rewrite instructions using a partition of an alloca to
2268 /// use a new alloca.
2270 /// Also implements the rewriting to vector-based accesses when the partition
2271 /// passes the isVectorPromotionViable predicate. Most of the rewriting logic
2273 class AllocaPartitionRewriter : public InstVisitor<AllocaPartitionRewriter,
2275 // Befriend the base class so it can delegate to private visit methods.
2276 friend class llvm::InstVisitor<AllocaPartitionRewriter, bool>;
2278 const DataLayout &TD;
2279 AllocaPartitioning &P;
2281 AllocaInst &OldAI, &NewAI;
2282 const uint64_t NewAllocaBeginOffset, NewAllocaEndOffset;
2285 // If we are rewriting an alloca partition which can be written as pure
2286 // vector operations, we stash extra information here. When VecTy is
2287 // non-null, we have some strict guarantees about the rewriten alloca:
2288 // - The new alloca is exactly the size of the vector type here.
2289 // - The accesses all either map to the entire vector or to a single
2291 // - The set of accessing instructions is only one of those handled above
2292 // in isVectorPromotionViable. Generally these are the same access kinds
2293 // which are promotable via mem2reg.
2296 uint64_t ElementSize;
2298 // This is a convenience and flag variable that will be null unless the new
2299 // alloca's integer operations should be widened to this integer type due to
2300 // passing isIntegerWideningViable above. If it is non-null, the desired
2301 // integer type will be stored here for easy access during rewriting.
2304 // The offset of the partition user currently being rewritten.
2305 uint64_t BeginOffset, EndOffset;
2307 Instruction *OldPtr;
2309 // The name prefix to use when rewriting instructions for this alloca.
2310 std::string NamePrefix;
2313 AllocaPartitionRewriter(const DataLayout &TD, AllocaPartitioning &P,
2314 AllocaPartitioning::iterator PI,
2315 SROA &Pass, AllocaInst &OldAI, AllocaInst &NewAI,
2316 uint64_t NewBeginOffset, uint64_t NewEndOffset)
2317 : TD(TD), P(P), Pass(Pass),
2318 OldAI(OldAI), NewAI(NewAI),
2319 NewAllocaBeginOffset(NewBeginOffset),
2320 NewAllocaEndOffset(NewEndOffset),
2321 NewAllocaTy(NewAI.getAllocatedType()),
2322 VecTy(), ElementTy(), ElementSize(), IntTy(),
2323 BeginOffset(), EndOffset() {
2326 /// \brief Visit the users of the alloca partition and rewrite them.
2327 bool visitUsers(AllocaPartitioning::const_use_iterator I,
2328 AllocaPartitioning::const_use_iterator E) {
2329 if (isVectorPromotionViable(TD, NewAI.getAllocatedType(), P,
2330 NewAllocaBeginOffset, NewAllocaEndOffset,
2333 VecTy = cast<VectorType>(NewAI.getAllocatedType());
2334 ElementTy = VecTy->getElementType();
2335 assert((TD.getTypeSizeInBits(VecTy->getScalarType()) % 8) == 0 &&
2336 "Only multiple-of-8 sized vector elements are viable");
2337 ElementSize = TD.getTypeSizeInBits(VecTy->getScalarType()) / 8;
2338 } else if (isIntegerWideningViable(TD, NewAI.getAllocatedType(),
2339 NewAllocaBeginOffset, P, I, E)) {
2340 IntTy = Type::getIntNTy(NewAI.getContext(),
2341 TD.getTypeSizeInBits(NewAI.getAllocatedType()));
2343 bool CanSROA = true;
2344 for (; I != E; ++I) {
2346 continue; // Skip dead uses.
2347 BeginOffset = I->BeginOffset;
2348 EndOffset = I->EndOffset;
2350 OldPtr = cast<Instruction>(I->U->get());
2351 NamePrefix = (Twine(NewAI.getName()) + "." + Twine(BeginOffset)).str();
2352 CanSROA &= visit(cast<Instruction>(I->U->getUser()));
2368 // Every instruction which can end up as a user must have a rewrite rule.
2369 bool visitInstruction(Instruction &I) {
2370 DEBUG(dbgs() << " !!!! Cannot rewrite: " << I << "\n");
2371 llvm_unreachable("No rewrite rule for this instruction!");
2374 Twine getName(const Twine &Suffix) {
2375 return NamePrefix + Suffix;
2378 Value *getAdjustedAllocaPtr(IRBuilder<> &IRB, Type *PointerTy) {
2379 assert(BeginOffset >= NewAllocaBeginOffset);
2380 APInt Offset(TD.getPointerSizeInBits(), BeginOffset - NewAllocaBeginOffset);
2381 return getAdjustedPtr(IRB, TD, &NewAI, Offset, PointerTy, getName(""));
2384 /// \brief Compute suitable alignment to access an offset into the new alloca.
2385 unsigned getOffsetAlign(uint64_t Offset) {
2386 unsigned NewAIAlign = NewAI.getAlignment();
2388 NewAIAlign = TD.getABITypeAlignment(NewAI.getAllocatedType());
2389 return MinAlign(NewAIAlign, Offset);
2392 /// \brief Compute suitable alignment to access this partition of the new
2394 unsigned getPartitionAlign() {
2395 return getOffsetAlign(BeginOffset - NewAllocaBeginOffset);
2398 /// \brief Compute suitable alignment to access a type at an offset of the
2401 /// \returns zero if the type's ABI alignment is a suitable alignment,
2402 /// otherwise returns the maximal suitable alignment.
2403 unsigned getOffsetTypeAlign(Type *Ty, uint64_t Offset) {
2404 unsigned Align = getOffsetAlign(Offset);
2405 return Align == TD.getABITypeAlignment(Ty) ? 0 : Align;
2408 /// \brief Compute suitable alignment to access a type at the beginning of
2409 /// this partition of the new alloca.
2411 /// See \c getOffsetTypeAlign for details; this routine delegates to it.
2412 unsigned getPartitionTypeAlign(Type *Ty) {
2413 return getOffsetTypeAlign(Ty, BeginOffset - NewAllocaBeginOffset);
2416 unsigned getIndex(uint64_t Offset) {
2417 assert(VecTy && "Can only call getIndex when rewriting a vector");
2418 uint64_t RelOffset = Offset - NewAllocaBeginOffset;
2419 assert(RelOffset / ElementSize < UINT32_MAX && "Index out of bounds");
2420 uint32_t Index = RelOffset / ElementSize;
2421 assert(Index * ElementSize == RelOffset);
2425 void deleteIfTriviallyDead(Value *V) {
2426 Instruction *I = cast<Instruction>(V);
2427 if (isInstructionTriviallyDead(I))
2428 Pass.DeadInsts.insert(I);
2431 Value *rewriteVectorizedLoadInst(IRBuilder<> &IRB) {
2432 unsigned BeginIndex = getIndex(BeginOffset);
2433 unsigned EndIndex = getIndex(EndOffset);
2434 assert(EndIndex > BeginIndex && "Empty vector!");
2436 Value *V = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2438 return extractVector(IRB, V, BeginIndex, EndIndex, getName(".vec"));
2441 Value *rewriteIntegerLoad(IRBuilder<> &IRB, LoadInst &LI) {
2442 assert(IntTy && "We cannot insert an integer to the alloca");
2443 assert(!LI.isVolatile());
2444 Value *V = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2446 V = convertValue(TD, IRB, V, IntTy);
2447 assert(BeginOffset >= NewAllocaBeginOffset && "Out of bounds offset");
2448 uint64_t Offset = BeginOffset - NewAllocaBeginOffset;
2449 if (Offset > 0 || EndOffset < NewAllocaEndOffset)
2450 V = extractInteger(TD, IRB, V, cast<IntegerType>(LI.getType()), Offset,
2451 getName(".extract"));
2455 bool visitLoadInst(LoadInst &LI) {
2456 DEBUG(dbgs() << " original: " << LI << "\n");
2457 Value *OldOp = LI.getOperand(0);
2458 assert(OldOp == OldPtr);
2459 IRBuilder<> IRB(&LI);
2461 uint64_t Size = EndOffset - BeginOffset;
2462 bool IsSplitIntLoad = Size < TD.getTypeStoreSize(LI.getType());
2464 // If this memory access can be shown to *statically* extend outside the
2465 // bounds of the original allocation it's behavior is undefined. Rather
2466 // than trying to transform it, just replace it with undef.
2467 // FIXME: We should do something more clever for functions being
2468 // instrumented by asan.
2469 // FIXME: Eventually, once ASan and friends can flush out bugs here, this
2470 // should be transformed to a load of null making it unreachable.
2471 uint64_t OldAllocSize = TD.getTypeAllocSize(OldAI.getAllocatedType());
2472 if (TD.getTypeStoreSize(LI.getType()) > OldAllocSize) {
2473 LI.replaceAllUsesWith(UndefValue::get(LI.getType()));
2474 Pass.DeadInsts.insert(&LI);
2475 deleteIfTriviallyDead(OldOp);
2476 DEBUG(dbgs() << " to: undef!!\n");
2480 Type *TargetTy = IsSplitIntLoad ? Type::getIntNTy(LI.getContext(), Size * 8)
2482 bool IsPtrAdjusted = false;
2485 V = rewriteVectorizedLoadInst(IRB);
2486 } else if (IntTy && LI.getType()->isIntegerTy()) {
2487 V = rewriteIntegerLoad(IRB, LI);
2488 } else if (BeginOffset == NewAllocaBeginOffset &&
2489 canConvertValue(TD, NewAllocaTy, LI.getType())) {
2490 V = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2491 LI.isVolatile(), getName(".load"));
2493 Type *LTy = TargetTy->getPointerTo();
2494 V = IRB.CreateAlignedLoad(getAdjustedAllocaPtr(IRB, LTy),
2495 getPartitionTypeAlign(TargetTy),
2496 LI.isVolatile(), getName(".load"));
2497 IsPtrAdjusted = true;
2499 V = convertValue(TD, IRB, V, TargetTy);
2501 if (IsSplitIntLoad) {
2502 assert(!LI.isVolatile());
2503 assert(LI.getType()->isIntegerTy() &&
2504 "Only integer type loads and stores are split");
2505 assert(LI.getType()->getIntegerBitWidth() ==
2506 TD.getTypeStoreSizeInBits(LI.getType()) &&
2507 "Non-byte-multiple bit width");
2508 assert(LI.getType()->getIntegerBitWidth() ==
2509 TD.getTypeAllocSizeInBits(OldAI.getAllocatedType()) &&
2510 "Only alloca-wide loads can be split and recomposed");
2511 // Move the insertion point just past the load so that we can refer to it.
2512 IRB.SetInsertPoint(llvm::next(BasicBlock::iterator(&LI)));
2513 // Create a placeholder value with the same type as LI to use as the
2514 // basis for the new value. This allows us to replace the uses of LI with
2515 // the computed value, and then replace the placeholder with LI, leaving
2516 // LI only used for this computation.
2518 = new LoadInst(UndefValue::get(LI.getType()->getPointerTo()));
2519 V = insertInteger(TD, IRB, Placeholder, V, BeginOffset,
2520 getName(".insert"));
2521 LI.replaceAllUsesWith(V);
2522 Placeholder->replaceAllUsesWith(&LI);
2525 LI.replaceAllUsesWith(V);
2528 Pass.DeadInsts.insert(&LI);
2529 deleteIfTriviallyDead(OldOp);
2530 DEBUG(dbgs() << " to: " << *V << "\n");
2531 return !LI.isVolatile() && !IsPtrAdjusted;
2534 bool rewriteVectorizedStoreInst(IRBuilder<> &IRB, Value *V,
2535 StoreInst &SI, Value *OldOp) {
2536 unsigned BeginIndex = getIndex(BeginOffset);
2537 unsigned EndIndex = getIndex(EndOffset);
2538 assert(EndIndex > BeginIndex && "Empty vector!");
2539 unsigned NumElements = EndIndex - BeginIndex;
2540 assert(NumElements <= VecTy->getNumElements() && "Too many elements!");
2542 = (NumElements == 1) ? ElementTy
2543 : VectorType::get(ElementTy, NumElements);
2544 if (V->getType() != PartitionTy)
2545 V = convertValue(TD, IRB, V, PartitionTy);
2547 // Mix in the existing elements.
2548 Value *Old = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2550 V = insertVector(IRB, Old, V, BeginIndex, getName(".vec"));
2552 StoreInst *Store = IRB.CreateAlignedStore(V, &NewAI, NewAI.getAlignment());
2553 Pass.DeadInsts.insert(&SI);
2556 DEBUG(dbgs() << " to: " << *Store << "\n");
2560 bool rewriteIntegerStore(IRBuilder<> &IRB, Value *V, StoreInst &SI) {
2561 assert(IntTy && "We cannot extract an integer from the alloca");
2562 assert(!SI.isVolatile());
2563 if (TD.getTypeSizeInBits(V->getType()) != IntTy->getBitWidth()) {
2564 Value *Old = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2565 getName(".oldload"));
2566 Old = convertValue(TD, IRB, Old, IntTy);
2567 assert(BeginOffset >= NewAllocaBeginOffset && "Out of bounds offset");
2568 uint64_t Offset = BeginOffset - NewAllocaBeginOffset;
2569 V = insertInteger(TD, IRB, Old, SI.getValueOperand(), Offset,
2570 getName(".insert"));
2572 V = convertValue(TD, IRB, V, NewAllocaTy);
2573 StoreInst *Store = IRB.CreateAlignedStore(V, &NewAI, NewAI.getAlignment());
2574 Pass.DeadInsts.insert(&SI);
2576 DEBUG(dbgs() << " to: " << *Store << "\n");
2580 bool visitStoreInst(StoreInst &SI) {
2581 DEBUG(dbgs() << " original: " << SI << "\n");
2582 Value *OldOp = SI.getOperand(1);
2583 assert(OldOp == OldPtr);
2584 IRBuilder<> IRB(&SI);
2586 Value *V = SI.getValueOperand();
2588 // Strip all inbounds GEPs and pointer casts to try to dig out any root
2589 // alloca that should be re-examined after promoting this alloca.
2590 if (V->getType()->isPointerTy())
2591 if (AllocaInst *AI = dyn_cast<AllocaInst>(V->stripInBoundsOffsets()))
2592 Pass.PostPromotionWorklist.insert(AI);
2594 uint64_t Size = EndOffset - BeginOffset;
2595 if (Size < TD.getTypeStoreSize(V->getType())) {
2596 assert(!SI.isVolatile());
2597 assert(V->getType()->isIntegerTy() &&
2598 "Only integer type loads and stores are split");
2599 assert(V->getType()->getIntegerBitWidth() ==
2600 TD.getTypeStoreSizeInBits(V->getType()) &&
2601 "Non-byte-multiple bit width");
2602 assert(V->getType()->getIntegerBitWidth() ==
2603 TD.getTypeAllocSizeInBits(OldAI.getAllocatedType()) &&
2604 "Only alloca-wide stores can be split and recomposed");
2605 IntegerType *NarrowTy = Type::getIntNTy(SI.getContext(), Size * 8);
2606 V = extractInteger(TD, IRB, V, NarrowTy, BeginOffset,
2607 getName(".extract"));
2611 return rewriteVectorizedStoreInst(IRB, V, SI, OldOp);
2612 if (IntTy && V->getType()->isIntegerTy())
2613 return rewriteIntegerStore(IRB, V, SI);
2616 if (BeginOffset == NewAllocaBeginOffset &&
2617 canConvertValue(TD, V->getType(), NewAllocaTy)) {
2618 V = convertValue(TD, IRB, V, NewAllocaTy);
2619 NewSI = IRB.CreateAlignedStore(V, &NewAI, NewAI.getAlignment(),
2622 Value *NewPtr = getAdjustedAllocaPtr(IRB, V->getType()->getPointerTo());
2623 NewSI = IRB.CreateAlignedStore(V, NewPtr,
2624 getPartitionTypeAlign(V->getType()),
2628 Pass.DeadInsts.insert(&SI);
2629 deleteIfTriviallyDead(OldOp);
2631 DEBUG(dbgs() << " to: " << *NewSI << "\n");
2632 return NewSI->getPointerOperand() == &NewAI && !SI.isVolatile();
2635 /// \brief Compute an integer value from splatting an i8 across the given
2636 /// number of bytes.
2638 /// Note that this routine assumes an i8 is a byte. If that isn't true, don't
2639 /// call this routine.
2640 /// FIXME: Heed the abvice above.
2642 /// \param V The i8 value to splat.
2643 /// \param Size The number of bytes in the output (assuming i8 is one byte)
2644 Value *getIntegerSplat(IRBuilder<> &IRB, Value *V, unsigned Size) {
2645 assert(Size > 0 && "Expected a positive number of bytes.");
2646 IntegerType *VTy = cast<IntegerType>(V->getType());
2647 assert(VTy->getBitWidth() == 8 && "Expected an i8 value for the byte");
2651 Type *SplatIntTy = Type::getIntNTy(VTy->getContext(), Size*8);
2652 V = IRB.CreateMul(IRB.CreateZExt(V, SplatIntTy, getName(".zext")),
2653 ConstantExpr::getUDiv(
2654 Constant::getAllOnesValue(SplatIntTy),
2655 ConstantExpr::getZExt(
2656 Constant::getAllOnesValue(V->getType()),
2658 getName(".isplat"));
2662 /// \brief Compute a vector splat for a given element value.
2663 Value *getVectorSplat(IRBuilder<> &IRB, Value *V, unsigned NumElements) {
2664 assert(NumElements > 0 && "Cannot splat to an empty vector.");
2666 // First insert it into a one-element vector so we can shuffle it. It is
2667 // really silly that LLVM's IR requires this in order to form a splat.
2668 Value *Undef = UndefValue::get(VectorType::get(V->getType(), 1));
2669 V = IRB.CreateInsertElement(Undef, V, IRB.getInt32(0),
2670 getName(".splatinsert"));
2672 // Shuffle the value across the desired number of elements.
2673 SmallVector<Constant*, 8> Mask(NumElements, IRB.getInt32(0));
2674 V = IRB.CreateShuffleVector(V, Undef, ConstantVector::get(Mask),
2676 DEBUG(dbgs() << " splat: " << *V << "\n");
2680 bool visitMemSetInst(MemSetInst &II) {
2681 DEBUG(dbgs() << " original: " << II << "\n");
2682 IRBuilder<> IRB(&II);
2683 assert(II.getRawDest() == OldPtr);
2685 // If the memset has a variable size, it cannot be split, just adjust the
2686 // pointer to the new alloca.
2687 if (!isa<Constant>(II.getLength())) {
2688 II.setDest(getAdjustedAllocaPtr(IRB, II.getRawDest()->getType()));
2689 Type *CstTy = II.getAlignmentCst()->getType();
2690 II.setAlignment(ConstantInt::get(CstTy, getPartitionAlign()));
2692 deleteIfTriviallyDead(OldPtr);
2696 // Record this instruction for deletion.
2697 Pass.DeadInsts.insert(&II);
2699 Type *AllocaTy = NewAI.getAllocatedType();
2700 Type *ScalarTy = AllocaTy->getScalarType();
2702 // If this doesn't map cleanly onto the alloca type, and that type isn't
2703 // a single value type, just emit a memset.
2704 if (!VecTy && !IntTy &&
2705 (BeginOffset != NewAllocaBeginOffset ||
2706 EndOffset != NewAllocaEndOffset ||
2707 !AllocaTy->isSingleValueType() ||
2708 !TD.isLegalInteger(TD.getTypeSizeInBits(ScalarTy)) ||
2709 TD.getTypeSizeInBits(ScalarTy)%8 != 0)) {
2710 Type *SizeTy = II.getLength()->getType();
2711 Constant *Size = ConstantInt::get(SizeTy, EndOffset - BeginOffset);
2713 = IRB.CreateMemSet(getAdjustedAllocaPtr(IRB,
2714 II.getRawDest()->getType()),
2715 II.getValue(), Size, getPartitionAlign(),
2718 DEBUG(dbgs() << " to: " << *New << "\n");
2722 // If we can represent this as a simple value, we have to build the actual
2723 // value to store, which requires expanding the byte present in memset to
2724 // a sensible representation for the alloca type. This is essentially
2725 // splatting the byte to a sufficiently wide integer, splatting it across
2726 // any desired vector width, and bitcasting to the final type.
2727 uint64_t Size = EndOffset - BeginOffset;
2728 Value *V = getIntegerSplat(IRB, II.getValue(), Size);
2731 // If this is a memset of a vectorized alloca, insert it.
2732 assert(ElementTy == ScalarTy);
2734 unsigned BeginIndex = getIndex(BeginOffset);
2735 unsigned EndIndex = getIndex(EndOffset);
2736 assert(EndIndex > BeginIndex && "Empty vector!");
2737 unsigned NumElements = EndIndex - BeginIndex;
2738 assert(NumElements <= VecTy->getNumElements() && "Too many elements!");
2740 Value *Splat = getIntegerSplat(IRB, II.getValue(),
2741 TD.getTypeSizeInBits(ElementTy)/8);
2742 Splat = convertValue(TD, IRB, Splat, ElementTy);
2743 if (NumElements > 1)
2744 Splat = getVectorSplat(IRB, Splat, NumElements);
2746 Value *Old = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2747 getName(".oldload"));
2748 V = insertVector(IRB, Old, Splat, BeginIndex, getName(".vec"));
2750 // If this is a memset on an alloca where we can widen stores, insert the
2752 assert(!II.isVolatile());
2754 V = getIntegerSplat(IRB, II.getValue(), Size);
2756 if (IntTy && (BeginOffset != NewAllocaBeginOffset ||
2757 EndOffset != NewAllocaBeginOffset)) {
2758 Value *Old = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2759 getName(".oldload"));
2760 Old = convertValue(TD, IRB, Old, IntTy);
2761 assert(BeginOffset >= NewAllocaBeginOffset && "Out of bounds offset");
2762 uint64_t Offset = BeginOffset - NewAllocaBeginOffset;
2763 V = insertInteger(TD, IRB, Old, V, Offset, getName(".insert"));
2765 assert(V->getType() == IntTy &&
2766 "Wrong type for an alloca wide integer!");
2768 V = convertValue(TD, IRB, V, AllocaTy);
2770 // Established these invariants above.
2771 assert(BeginOffset == NewAllocaBeginOffset);
2772 assert(EndOffset == NewAllocaEndOffset);
2774 V = getIntegerSplat(IRB, II.getValue(),
2775 TD.getTypeSizeInBits(ScalarTy)/8);
2776 if (VectorType *AllocaVecTy = dyn_cast<VectorType>(AllocaTy))
2777 V = getVectorSplat(IRB, V, AllocaVecTy->getNumElements());
2779 V = convertValue(TD, IRB, V, AllocaTy);
2782 Value *New = IRB.CreateAlignedStore(V, &NewAI, NewAI.getAlignment(),
2785 DEBUG(dbgs() << " to: " << *New << "\n");
2786 return !II.isVolatile();
2789 bool visitMemTransferInst(MemTransferInst &II) {
2790 // Rewriting of memory transfer instructions can be a bit tricky. We break
2791 // them into two categories: split intrinsics and unsplit intrinsics.
2793 DEBUG(dbgs() << " original: " << II << "\n");
2794 IRBuilder<> IRB(&II);
2796 assert(II.getRawSource() == OldPtr || II.getRawDest() == OldPtr);
2797 bool IsDest = II.getRawDest() == OldPtr;
2799 const AllocaPartitioning::MemTransferOffsets &MTO
2800 = P.getMemTransferOffsets(II);
2802 // Compute the relative offset within the transfer.
2803 unsigned IntPtrWidth = TD.getPointerSizeInBits();
2804 APInt RelOffset(IntPtrWidth, BeginOffset - (IsDest ? MTO.DestBegin
2805 : MTO.SourceBegin));
2807 unsigned Align = II.getAlignment();
2809 Align = MinAlign(RelOffset.zextOrTrunc(64).getZExtValue(),
2810 MinAlign(II.getAlignment(), getPartitionAlign()));
2812 // For unsplit intrinsics, we simply modify the source and destination
2813 // pointers in place. This isn't just an optimization, it is a matter of
2814 // correctness. With unsplit intrinsics we may be dealing with transfers
2815 // within a single alloca before SROA ran, or with transfers that have
2816 // a variable length. We may also be dealing with memmove instead of
2817 // memcpy, and so simply updating the pointers is the necessary for us to
2818 // update both source and dest of a single call.
2819 if (!MTO.IsSplittable) {
2820 Value *OldOp = IsDest ? II.getRawDest() : II.getRawSource();
2822 II.setDest(getAdjustedAllocaPtr(IRB, II.getRawDest()->getType()));
2824 II.setSource(getAdjustedAllocaPtr(IRB, II.getRawSource()->getType()));
2826 Type *CstTy = II.getAlignmentCst()->getType();
2827 II.setAlignment(ConstantInt::get(CstTy, Align));
2829 DEBUG(dbgs() << " to: " << II << "\n");
2830 deleteIfTriviallyDead(OldOp);
2833 // For split transfer intrinsics we have an incredibly useful assurance:
2834 // the source and destination do not reside within the same alloca, and at
2835 // least one of them does not escape. This means that we can replace
2836 // memmove with memcpy, and we don't need to worry about all manner of
2837 // downsides to splitting and transforming the operations.
2839 // If this doesn't map cleanly onto the alloca type, and that type isn't
2840 // a single value type, just emit a memcpy.
2842 = !VecTy && !IntTy && (BeginOffset != NewAllocaBeginOffset ||
2843 EndOffset != NewAllocaEndOffset ||
2844 !NewAI.getAllocatedType()->isSingleValueType());
2846 // If we're just going to emit a memcpy, the alloca hasn't changed, and the
2847 // size hasn't been shrunk based on analysis of the viable range, this is
2849 if (EmitMemCpy && &OldAI == &NewAI) {
2850 uint64_t OrigBegin = IsDest ? MTO.DestBegin : MTO.SourceBegin;
2851 uint64_t OrigEnd = IsDest ? MTO.DestEnd : MTO.SourceEnd;
2852 // Ensure the start lines up.
2853 assert(BeginOffset == OrigBegin);
2856 // Rewrite the size as needed.
2857 if (EndOffset != OrigEnd)
2858 II.setLength(ConstantInt::get(II.getLength()->getType(),
2859 EndOffset - BeginOffset));
2862 // Record this instruction for deletion.
2863 Pass.DeadInsts.insert(&II);
2865 // Strip all inbounds GEPs and pointer casts to try to dig out any root
2866 // alloca that should be re-examined after rewriting this instruction.
2867 Value *OtherPtr = IsDest ? II.getRawSource() : II.getRawDest();
2869 = dyn_cast<AllocaInst>(OtherPtr->stripInBoundsOffsets()))
2870 Pass.Worklist.insert(AI);
2873 Type *OtherPtrTy = IsDest ? II.getRawSource()->getType()
2874 : II.getRawDest()->getType();
2876 // Compute the other pointer, folding as much as possible to produce
2877 // a single, simple GEP in most cases.
2878 OtherPtr = getAdjustedPtr(IRB, TD, OtherPtr, RelOffset, OtherPtrTy,
2879 getName("." + OtherPtr->getName()));
2882 = getAdjustedAllocaPtr(IRB, IsDest ? II.getRawDest()->getType()
2883 : II.getRawSource()->getType());
2884 Type *SizeTy = II.getLength()->getType();
2885 Constant *Size = ConstantInt::get(SizeTy, EndOffset - BeginOffset);
2887 CallInst *New = IRB.CreateMemCpy(IsDest ? OurPtr : OtherPtr,
2888 IsDest ? OtherPtr : OurPtr,
2889 Size, Align, II.isVolatile());
2891 DEBUG(dbgs() << " to: " << *New << "\n");
2895 // Note that we clamp the alignment to 1 here as a 0 alignment for a memcpy
2896 // is equivalent to 1, but that isn't true if we end up rewriting this as
2901 bool IsWholeAlloca = BeginOffset == NewAllocaBeginOffset &&
2902 EndOffset == NewAllocaEndOffset;
2903 uint64_t Size = EndOffset - BeginOffset;
2904 unsigned BeginIndex = VecTy ? getIndex(BeginOffset) : 0;
2905 unsigned EndIndex = VecTy ? getIndex(EndOffset) : 0;
2906 unsigned NumElements = EndIndex - BeginIndex;
2907 IntegerType *SubIntTy
2908 = IntTy ? Type::getIntNTy(IntTy->getContext(), Size*8) : 0;
2910 Type *OtherPtrTy = NewAI.getType();
2911 if (VecTy && !IsWholeAlloca) {
2912 if (NumElements == 1)
2913 OtherPtrTy = VecTy->getElementType();
2915 OtherPtrTy = VectorType::get(VecTy->getElementType(), NumElements);
2917 OtherPtrTy = OtherPtrTy->getPointerTo();
2918 } else if (IntTy && !IsWholeAlloca) {
2919 OtherPtrTy = SubIntTy->getPointerTo();
2922 Value *SrcPtr = getAdjustedPtr(IRB, TD, OtherPtr, RelOffset, OtherPtrTy,
2923 getName("." + OtherPtr->getName()));
2924 Value *DstPtr = &NewAI;
2926 std::swap(SrcPtr, DstPtr);
2929 if (VecTy && !IsWholeAlloca && !IsDest) {
2930 Src = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2932 Src = extractVector(IRB, Src, BeginIndex, EndIndex, getName(".vec"));
2933 } else if (IntTy && !IsWholeAlloca && !IsDest) {
2934 Src = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2936 Src = convertValue(TD, IRB, Src, IntTy);
2937 assert(BeginOffset >= NewAllocaBeginOffset && "Out of bounds offset");
2938 uint64_t Offset = BeginOffset - NewAllocaBeginOffset;
2939 Src = extractInteger(TD, IRB, Src, SubIntTy, Offset, getName(".extract"));
2941 Src = IRB.CreateAlignedLoad(SrcPtr, Align, II.isVolatile(),
2942 getName(".copyload"));
2945 if (VecTy && !IsWholeAlloca && IsDest) {
2946 Value *Old = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2947 getName(".oldload"));
2948 Src = insertVector(IRB, Old, Src, BeginIndex, getName(".vec"));
2949 } else if (IntTy && !IsWholeAlloca && IsDest) {
2950 Value *Old = IRB.CreateAlignedLoad(&NewAI, NewAI.getAlignment(),
2951 getName(".oldload"));
2952 Old = convertValue(TD, IRB, Old, IntTy);
2953 assert(BeginOffset >= NewAllocaBeginOffset && "Out of bounds offset");
2954 uint64_t Offset = BeginOffset - NewAllocaBeginOffset;
2955 Src = insertInteger(TD, IRB, Old, Src, Offset, getName(".insert"));
2956 Src = convertValue(TD, IRB, Src, NewAllocaTy);
2959 StoreInst *Store = cast<StoreInst>(
2960 IRB.CreateAlignedStore(Src, DstPtr, Align, II.isVolatile()));
2962 DEBUG(dbgs() << " to: " << *Store << "\n");
2963 return !II.isVolatile();
2966 bool visitIntrinsicInst(IntrinsicInst &II) {
2967 assert(II.getIntrinsicID() == Intrinsic::lifetime_start ||
2968 II.getIntrinsicID() == Intrinsic::lifetime_end);
2969 DEBUG(dbgs() << " original: " << II << "\n");
2970 IRBuilder<> IRB(&II);
2971 assert(II.getArgOperand(1) == OldPtr);
2973 // Record this instruction for deletion.
2974 Pass.DeadInsts.insert(&II);
2977 = ConstantInt::get(cast<IntegerType>(II.getArgOperand(0)->getType()),
2978 EndOffset - BeginOffset);
2979 Value *Ptr = getAdjustedAllocaPtr(IRB, II.getArgOperand(1)->getType());
2981 if (II.getIntrinsicID() == Intrinsic::lifetime_start)
2982 New = IRB.CreateLifetimeStart(Ptr, Size);
2984 New = IRB.CreateLifetimeEnd(Ptr, Size);
2986 DEBUG(dbgs() << " to: " << *New << "\n");
2990 bool visitPHINode(PHINode &PN) {
2991 DEBUG(dbgs() << " original: " << PN << "\n");
2993 // We would like to compute a new pointer in only one place, but have it be
2994 // as local as possible to the PHI. To do that, we re-use the location of
2995 // the old pointer, which necessarily must be in the right position to
2996 // dominate the PHI.
2997 IRBuilder<> PtrBuilder(cast<Instruction>(OldPtr));
2999 Value *NewPtr = getAdjustedAllocaPtr(PtrBuilder, OldPtr->getType());
3000 // Replace the operands which were using the old pointer.
3001 std::replace(PN.op_begin(), PN.op_end(), cast<Value>(OldPtr), NewPtr);
3003 DEBUG(dbgs() << " to: " << PN << "\n");
3004 deleteIfTriviallyDead(OldPtr);
3008 bool visitSelectInst(SelectInst &SI) {
3009 DEBUG(dbgs() << " original: " << SI << "\n");
3010 IRBuilder<> IRB(&SI);
3012 // Find the operand we need to rewrite here.
3013 bool IsTrueVal = SI.getTrueValue() == OldPtr;
3015 assert(SI.getFalseValue() != OldPtr && "Pointer is both operands!");
3017 assert(SI.getFalseValue() == OldPtr && "Pointer isn't an operand!");
3019 Value *NewPtr = getAdjustedAllocaPtr(IRB, OldPtr->getType());
3020 SI.setOperand(IsTrueVal ? 1 : 2, NewPtr);
3021 DEBUG(dbgs() << " to: " << SI << "\n");
3022 deleteIfTriviallyDead(OldPtr);
3030 /// \brief Visitor to rewrite aggregate loads and stores as scalar.
3032 /// This pass aggressively rewrites all aggregate loads and stores on
3033 /// a particular pointer (or any pointer derived from it which we can identify)
3034 /// with scalar loads and stores.
3035 class AggLoadStoreRewriter : public InstVisitor<AggLoadStoreRewriter, bool> {
3036 // Befriend the base class so it can delegate to private visit methods.
3037 friend class llvm::InstVisitor<AggLoadStoreRewriter, bool>;
3039 const DataLayout &TD;
3041 /// Queue of pointer uses to analyze and potentially rewrite.
3042 SmallVector<Use *, 8> Queue;
3044 /// Set to prevent us from cycling with phi nodes and loops.
3045 SmallPtrSet<User *, 8> Visited;
3047 /// The current pointer use being rewritten. This is used to dig up the used
3048 /// value (as opposed to the user).
3052 AggLoadStoreRewriter(const DataLayout &TD) : TD(TD) {}
3054 /// Rewrite loads and stores through a pointer and all pointers derived from
3056 bool rewrite(Instruction &I) {
3057 DEBUG(dbgs() << " Rewriting FCA loads and stores...\n");
3059 bool Changed = false;
3060 while (!Queue.empty()) {
3061 U = Queue.pop_back_val();
3062 Changed |= visit(cast<Instruction>(U->getUser()));
3068 /// Enqueue all the users of the given instruction for further processing.
3069 /// This uses a set to de-duplicate users.
3070 void enqueueUsers(Instruction &I) {
3071 for (Value::use_iterator UI = I.use_begin(), UE = I.use_end(); UI != UE;
3073 if (Visited.insert(*UI))
3074 Queue.push_back(&UI.getUse());
3077 // Conservative default is to not rewrite anything.
3078 bool visitInstruction(Instruction &I) { return false; }
3080 /// \brief Generic recursive split emission class.
3081 template <typename Derived>
3084 /// The builder used to form new instructions.
3086 /// The indices which to be used with insert- or extractvalue to select the
3087 /// appropriate value within the aggregate.
3088 SmallVector<unsigned, 4> Indices;
3089 /// The indices to a GEP instruction which will move Ptr to the correct slot
3090 /// within the aggregate.
3091 SmallVector<Value *, 4> GEPIndices;
3092 /// The base pointer of the original op, used as a base for GEPing the
3093 /// split operations.
3096 /// Initialize the splitter with an insertion point, Ptr and start with a
3097 /// single zero GEP index.
3098 OpSplitter(Instruction *InsertionPoint, Value *Ptr)
3099 : IRB(InsertionPoint), GEPIndices(1, IRB.getInt32(0)), Ptr(Ptr) {}
3102 /// \brief Generic recursive split emission routine.
3104 /// This method recursively splits an aggregate op (load or store) into
3105 /// scalar or vector ops. It splits recursively until it hits a single value
3106 /// and emits that single value operation via the template argument.
3108 /// The logic of this routine relies on GEPs and insertvalue and
3109 /// extractvalue all operating with the same fundamental index list, merely
3110 /// formatted differently (GEPs need actual values).
3112 /// \param Ty The type being split recursively into smaller ops.
3113 /// \param Agg The aggregate value being built up or stored, depending on
3114 /// whether this is splitting a load or a store respectively.
3115 void emitSplitOps(Type *Ty, Value *&Agg, const Twine &Name) {
3116 if (Ty->isSingleValueType())
3117 return static_cast<Derived *>(this)->emitFunc(Ty, Agg, Name);
3119 if (ArrayType *ATy = dyn_cast<ArrayType>(Ty)) {
3120 unsigned OldSize = Indices.size();
3122 for (unsigned Idx = 0, Size = ATy->getNumElements(); Idx != Size;
3124 assert(Indices.size() == OldSize && "Did not return to the old size");
3125 Indices.push_back(Idx);
3126 GEPIndices.push_back(IRB.getInt32(Idx));
3127 emitSplitOps(ATy->getElementType(), Agg, Name + "." + Twine(Idx));
3128 GEPIndices.pop_back();
3134 if (StructType *STy = dyn_cast<StructType>(Ty)) {
3135 unsigned OldSize = Indices.size();
3137 for (unsigned Idx = 0, Size = STy->getNumElements(); Idx != Size;
3139 assert(Indices.size() == OldSize && "Did not return to the old size");
3140 Indices.push_back(Idx);
3141 GEPIndices.push_back(IRB.getInt32(Idx));
3142 emitSplitOps(STy->getElementType(Idx), Agg, Name + "." + Twine(Idx));
3143 GEPIndices.pop_back();
3149 llvm_unreachable("Only arrays and structs are aggregate loadable types");
3153 struct LoadOpSplitter : public OpSplitter<LoadOpSplitter> {
3154 LoadOpSplitter(Instruction *InsertionPoint, Value *Ptr)
3155 : OpSplitter<LoadOpSplitter>(InsertionPoint, Ptr) {}
3157 /// Emit a leaf load of a single value. This is called at the leaves of the
3158 /// recursive emission to actually load values.
3159 void emitFunc(Type *Ty, Value *&Agg, const Twine &Name) {
3160 assert(Ty->isSingleValueType());
3161 // Load the single value and insert it using the indices.
3162 Value *Load = IRB.CreateLoad(IRB.CreateInBoundsGEP(Ptr, GEPIndices,
3165 Agg = IRB.CreateInsertValue(Agg, Load, Indices, Name + ".insert");
3166 DEBUG(dbgs() << " to: " << *Load << "\n");
3170 bool visitLoadInst(LoadInst &LI) {
3171 assert(LI.getPointerOperand() == *U);
3172 if (!LI.isSimple() || LI.getType()->isSingleValueType())
3175 // We have an aggregate being loaded, split it apart.
3176 DEBUG(dbgs() << " original: " << LI << "\n");
3177 LoadOpSplitter Splitter(&LI, *U);
3178 Value *V = UndefValue::get(LI.getType());
3179 Splitter.emitSplitOps(LI.getType(), V, LI.getName() + ".fca");
3180 LI.replaceAllUsesWith(V);
3181 LI.eraseFromParent();
3185 struct StoreOpSplitter : public OpSplitter<StoreOpSplitter> {
3186 StoreOpSplitter(Instruction *InsertionPoint, Value *Ptr)
3187 : OpSplitter<StoreOpSplitter>(InsertionPoint, Ptr) {}
3189 /// Emit a leaf store of a single value. This is called at the leaves of the
3190 /// recursive emission to actually produce stores.
3191 void emitFunc(Type *Ty, Value *&Agg, const Twine &Name) {
3192 assert(Ty->isSingleValueType());
3193 // Extract the single value and store it using the indices.
3194 Value *Store = IRB.CreateStore(
3195 IRB.CreateExtractValue(Agg, Indices, Name + ".extract"),
3196 IRB.CreateInBoundsGEP(Ptr, GEPIndices, Name + ".gep"));
3198 DEBUG(dbgs() << " to: " << *Store << "\n");
3202 bool visitStoreInst(StoreInst &SI) {
3203 if (!SI.isSimple() || SI.getPointerOperand() != *U)
3205 Value *V = SI.getValueOperand();
3206 if (V->getType()->isSingleValueType())
3209 // We have an aggregate being stored, split it apart.
3210 DEBUG(dbgs() << " original: " << SI << "\n");
3211 StoreOpSplitter Splitter(&SI, *U);
3212 Splitter.emitSplitOps(V->getType(), V, V->getName() + ".fca");
3213 SI.eraseFromParent();
3217 bool visitBitCastInst(BitCastInst &BC) {
3222 bool visitGetElementPtrInst(GetElementPtrInst &GEPI) {
3227 bool visitPHINode(PHINode &PN) {
3232 bool visitSelectInst(SelectInst &SI) {
3239 /// \brief Strip aggregate type wrapping.
3241 /// This removes no-op aggregate types wrapping an underlying type. It will
3242 /// strip as many layers of types as it can without changing either the type
3243 /// size or the allocated size.
3244 static Type *stripAggregateTypeWrapping(const DataLayout &DL, Type *Ty) {
3245 if (Ty->isSingleValueType())
3248 uint64_t AllocSize = DL.getTypeAllocSize(Ty);
3249 uint64_t TypeSize = DL.getTypeSizeInBits(Ty);
3252 if (ArrayType *ArrTy = dyn_cast<ArrayType>(Ty)) {
3253 InnerTy = ArrTy->getElementType();
3254 } else if (StructType *STy = dyn_cast<StructType>(Ty)) {
3255 const StructLayout *SL = DL.getStructLayout(STy);
3256 unsigned Index = SL->getElementContainingOffset(0);
3257 InnerTy = STy->getElementType(Index);
3262 if (AllocSize > DL.getTypeAllocSize(InnerTy) ||
3263 TypeSize > DL.getTypeSizeInBits(InnerTy))
3266 return stripAggregateTypeWrapping(DL, InnerTy);
3269 /// \brief Try to find a partition of the aggregate type passed in for a given
3270 /// offset and size.
3272 /// This recurses through the aggregate type and tries to compute a subtype
3273 /// based on the offset and size. When the offset and size span a sub-section
3274 /// of an array, it will even compute a new array type for that sub-section,
3275 /// and the same for structs.
3277 /// Note that this routine is very strict and tries to find a partition of the
3278 /// type which produces the *exact* right offset and size. It is not forgiving
3279 /// when the size or offset cause either end of type-based partition to be off.
3280 /// Also, this is a best-effort routine. It is reasonable to give up and not
3281 /// return a type if necessary.
3282 static Type *getTypePartition(const DataLayout &TD, Type *Ty,
3283 uint64_t Offset, uint64_t Size) {
3284 if (Offset == 0 && TD.getTypeAllocSize(Ty) == Size)
3285 return stripAggregateTypeWrapping(TD, Ty);
3286 if (Offset > TD.getTypeAllocSize(Ty) ||
3287 (TD.getTypeAllocSize(Ty) - Offset) < Size)
3290 if (SequentialType *SeqTy = dyn_cast<SequentialType>(Ty)) {
3291 // We can't partition pointers...
3292 if (SeqTy->isPointerTy())
3295 Type *ElementTy = SeqTy->getElementType();
3296 uint64_t ElementSize = TD.getTypeAllocSize(ElementTy);
3297 uint64_t NumSkippedElements = Offset / ElementSize;
3298 if (ArrayType *ArrTy = dyn_cast<ArrayType>(SeqTy))
3299 if (NumSkippedElements >= ArrTy->getNumElements())
3301 if (VectorType *VecTy = dyn_cast<VectorType>(SeqTy))
3302 if (NumSkippedElements >= VecTy->getNumElements())
3304 Offset -= NumSkippedElements * ElementSize;
3306 // First check if we need to recurse.
3307 if (Offset > 0 || Size < ElementSize) {
3308 // Bail if the partition ends in a different array element.
3309 if ((Offset + Size) > ElementSize)
3311 // Recurse through the element type trying to peel off offset bytes.
3312 return getTypePartition(TD, ElementTy, Offset, Size);
3314 assert(Offset == 0);
3316 if (Size == ElementSize)
3317 return stripAggregateTypeWrapping(TD, ElementTy);
3318 assert(Size > ElementSize);
3319 uint64_t NumElements = Size / ElementSize;
3320 if (NumElements * ElementSize != Size)
3322 return ArrayType::get(ElementTy, NumElements);
3325 StructType *STy = dyn_cast<StructType>(Ty);
3329 const StructLayout *SL = TD.getStructLayout(STy);
3330 if (Offset >= SL->getSizeInBytes())
3332 uint64_t EndOffset = Offset + Size;
3333 if (EndOffset > SL->getSizeInBytes())
3336 unsigned Index = SL->getElementContainingOffset(Offset);
3337 Offset -= SL->getElementOffset(Index);
3339 Type *ElementTy = STy->getElementType(Index);
3340 uint64_t ElementSize = TD.getTypeAllocSize(ElementTy);
3341 if (Offset >= ElementSize)
3342 return 0; // The offset points into alignment padding.
3344 // See if any partition must be contained by the element.
3345 if (Offset > 0 || Size < ElementSize) {
3346 if ((Offset + Size) > ElementSize)
3348 return getTypePartition(TD, ElementTy, Offset, Size);
3350 assert(Offset == 0);
3352 if (Size == ElementSize)
3353 return stripAggregateTypeWrapping(TD, ElementTy);
3355 StructType::element_iterator EI = STy->element_begin() + Index,
3356 EE = STy->element_end();
3357 if (EndOffset < SL->getSizeInBytes()) {
3358 unsigned EndIndex = SL->getElementContainingOffset(EndOffset);
3359 if (Index == EndIndex)
3360 return 0; // Within a single element and its padding.
3362 // Don't try to form "natural" types if the elements don't line up with the
3364 // FIXME: We could potentially recurse down through the last element in the
3365 // sub-struct to find a natural end point.
3366 if (SL->getElementOffset(EndIndex) != EndOffset)
3369 assert(Index < EndIndex);
3370 EE = STy->element_begin() + EndIndex;
3373 // Try to build up a sub-structure.
3374 StructType *SubTy = StructType::get(STy->getContext(), makeArrayRef(EI, EE),
3376 const StructLayout *SubSL = TD.getStructLayout(SubTy);
3377 if (Size != SubSL->getSizeInBytes())
3378 return 0; // The sub-struct doesn't have quite the size needed.
3383 /// \brief Rewrite an alloca partition's users.
3385 /// This routine drives both of the rewriting goals of the SROA pass. It tries
3386 /// to rewrite uses of an alloca partition to be conducive for SSA value
3387 /// promotion. If the partition needs a new, more refined alloca, this will
3388 /// build that new alloca, preserving as much type information as possible, and
3389 /// rewrite the uses of the old alloca to point at the new one and have the
3390 /// appropriate new offsets. It also evaluates how successful the rewrite was
3391 /// at enabling promotion and if it was successful queues the alloca to be
3393 bool SROA::rewriteAllocaPartition(AllocaInst &AI,
3394 AllocaPartitioning &P,
3395 AllocaPartitioning::iterator PI) {
3396 uint64_t AllocaSize = PI->EndOffset - PI->BeginOffset;
3397 bool IsLive = false;
3398 for (AllocaPartitioning::use_iterator UI = P.use_begin(PI),
3400 UI != UE && !IsLive; ++UI)
3404 return false; // No live uses left of this partition.
3406 DEBUG(dbgs() << "Speculating PHIs and selects in partition "
3407 << "[" << PI->BeginOffset << "," << PI->EndOffset << ")\n");
3409 PHIOrSelectSpeculator Speculator(*TD, P, *this);
3410 DEBUG(dbgs() << " speculating ");
3411 DEBUG(P.print(dbgs(), PI, ""));
3412 Speculator.visitUsers(PI);
3414 // Try to compute a friendly type for this partition of the alloca. This
3415 // won't always succeed, in which case we fall back to a legal integer type
3416 // or an i8 array of an appropriate size.
3418 if (Type *PartitionTy = P.getCommonType(PI))
3419 if (TD->getTypeAllocSize(PartitionTy) >= AllocaSize)
3420 AllocaTy = PartitionTy;
3422 if (Type *PartitionTy = getTypePartition(*TD, AI.getAllocatedType(),
3423 PI->BeginOffset, AllocaSize))
3424 AllocaTy = PartitionTy;
3426 (AllocaTy->isArrayTy() &&
3427 AllocaTy->getArrayElementType()->isIntegerTy())) &&
3428 TD->isLegalInteger(AllocaSize * 8))
3429 AllocaTy = Type::getIntNTy(*C, AllocaSize * 8);
3431 AllocaTy = ArrayType::get(Type::getInt8Ty(*C), AllocaSize);
3432 assert(TD->getTypeAllocSize(AllocaTy) >= AllocaSize);
3434 // Check for the case where we're going to rewrite to a new alloca of the
3435 // exact same type as the original, and with the same access offsets. In that
3436 // case, re-use the existing alloca, but still run through the rewriter to
3437 // performe phi and select speculation.
3439 if (AllocaTy == AI.getAllocatedType()) {
3440 assert(PI->BeginOffset == 0 &&
3441 "Non-zero begin offset but same alloca type");
3442 assert(PI == P.begin() && "Begin offset is zero on later partition");
3445 unsigned Alignment = AI.getAlignment();
3447 // The minimum alignment which users can rely on when the explicit
3448 // alignment is omitted or zero is that required by the ABI for this
3450 Alignment = TD->getABITypeAlignment(AI.getAllocatedType());
3452 Alignment = MinAlign(Alignment, PI->BeginOffset);
3453 // If we will get at least this much alignment from the type alone, leave
3454 // the alloca's alignment unconstrained.
3455 if (Alignment <= TD->getABITypeAlignment(AllocaTy))
3457 NewAI = new AllocaInst(AllocaTy, 0, Alignment,
3458 AI.getName() + ".sroa." + Twine(PI - P.begin()),
3463 DEBUG(dbgs() << "Rewriting alloca partition "
3464 << "[" << PI->BeginOffset << "," << PI->EndOffset << ") to: "
3467 // Track the high watermark of the post-promotion worklist. We will reset it
3468 // to this point if the alloca is not in fact scheduled for promotion.
3469 unsigned PPWOldSize = PostPromotionWorklist.size();
3471 AllocaPartitionRewriter Rewriter(*TD, P, PI, *this, AI, *NewAI,
3472 PI->BeginOffset, PI->EndOffset);
3473 DEBUG(dbgs() << " rewriting ");
3474 DEBUG(P.print(dbgs(), PI, ""));
3475 bool Promotable = Rewriter.visitUsers(P.use_begin(PI), P.use_end(PI));
3477 DEBUG(dbgs() << " and queuing for promotion\n");
3478 PromotableAllocas.push_back(NewAI);
3479 } else if (NewAI != &AI) {
3480 // If we can't promote the alloca, iterate on it to check for new
3481 // refinements exposed by splitting the current alloca. Don't iterate on an
3482 // alloca which didn't actually change and didn't get promoted.
3483 Worklist.insert(NewAI);
3486 // Drop any post-promotion work items if promotion didn't happen.
3488 while (PostPromotionWorklist.size() > PPWOldSize)
3489 PostPromotionWorklist.pop_back();
3494 /// \brief Walks the partitioning of an alloca rewriting uses of each partition.
3495 bool SROA::splitAlloca(AllocaInst &AI, AllocaPartitioning &P) {
3496 bool Changed = false;
3497 for (AllocaPartitioning::iterator PI = P.begin(), PE = P.end(); PI != PE;
3499 Changed |= rewriteAllocaPartition(AI, P, PI);
3504 /// \brief Analyze an alloca for SROA.
3506 /// This analyzes the alloca to ensure we can reason about it, builds
3507 /// a partitioning of the alloca, and then hands it off to be split and
3508 /// rewritten as needed.
3509 bool SROA::runOnAlloca(AllocaInst &AI) {
3510 DEBUG(dbgs() << "SROA alloca: " << AI << "\n");
3511 ++NumAllocasAnalyzed;
3513 // Special case dead allocas, as they're trivial.
3514 if (AI.use_empty()) {
3515 AI.eraseFromParent();
3519 // Skip alloca forms that this analysis can't handle.
3520 if (AI.isArrayAllocation() || !AI.getAllocatedType()->isSized() ||
3521 TD->getTypeAllocSize(AI.getAllocatedType()) == 0)
3524 bool Changed = false;
3526 // First, split any FCA loads and stores touching this alloca to promote
3527 // better splitting and promotion opportunities.
3528 AggLoadStoreRewriter AggRewriter(*TD);
3529 Changed |= AggRewriter.rewrite(AI);
3531 // Build the partition set using a recursive instruction-visiting builder.
3532 AllocaPartitioning P(*TD, AI);
3533 DEBUG(P.print(dbgs()));
3537 // Delete all the dead users of this alloca before splitting and rewriting it.
3538 for (AllocaPartitioning::dead_user_iterator DI = P.dead_user_begin(),
3539 DE = P.dead_user_end();
3542 (*DI)->replaceAllUsesWith(UndefValue::get((*DI)->getType()));
3543 DeadInsts.insert(*DI);
3545 for (AllocaPartitioning::dead_op_iterator DO = P.dead_op_begin(),
3546 DE = P.dead_op_end();
3549 // Clobber the use with an undef value.
3550 **DO = UndefValue::get(OldV->getType());
3551 if (Instruction *OldI = dyn_cast<Instruction>(OldV))
3552 if (isInstructionTriviallyDead(OldI)) {
3554 DeadInsts.insert(OldI);
3558 // No partitions to split. Leave the dead alloca for a later pass to clean up.
3559 if (P.begin() == P.end())
3562 return splitAlloca(AI, P) || Changed;
3565 /// \brief Delete the dead instructions accumulated in this run.
3567 /// Recursively deletes the dead instructions we've accumulated. This is done
3568 /// at the very end to maximize locality of the recursive delete and to
3569 /// minimize the problems of invalidated instruction pointers as such pointers
3570 /// are used heavily in the intermediate stages of the algorithm.
3572 /// We also record the alloca instructions deleted here so that they aren't
3573 /// subsequently handed to mem2reg to promote.
3574 void SROA::deleteDeadInstructions(SmallPtrSet<AllocaInst*, 4> &DeletedAllocas) {
3575 while (!DeadInsts.empty()) {
3576 Instruction *I = DeadInsts.pop_back_val();
3577 DEBUG(dbgs() << "Deleting dead instruction: " << *I << "\n");
3579 I->replaceAllUsesWith(UndefValue::get(I->getType()));
3581 for (User::op_iterator OI = I->op_begin(), E = I->op_end(); OI != E; ++OI)
3582 if (Instruction *U = dyn_cast<Instruction>(*OI)) {
3583 // Zero out the operand and see if it becomes trivially dead.
3585 if (isInstructionTriviallyDead(U))
3586 DeadInsts.insert(U);
3589 if (AllocaInst *AI = dyn_cast<AllocaInst>(I))
3590 DeletedAllocas.insert(AI);
3593 I->eraseFromParent();
3597 /// \brief Promote the allocas, using the best available technique.
3599 /// This attempts to promote whatever allocas have been identified as viable in
3600 /// the PromotableAllocas list. If that list is empty, there is nothing to do.
3601 /// If there is a domtree available, we attempt to promote using the full power
3602 /// of mem2reg. Otherwise, we build and use the AllocaPromoter above which is
3603 /// based on the SSAUpdater utilities. This function returns whether any
3604 /// promotion occured.
3605 bool SROA::promoteAllocas(Function &F) {
3606 if (PromotableAllocas.empty())
3609 NumPromoted += PromotableAllocas.size();
3611 if (DT && !ForceSSAUpdater) {
3612 DEBUG(dbgs() << "Promoting allocas with mem2reg...\n");
3613 PromoteMemToReg(PromotableAllocas, *DT);
3614 PromotableAllocas.clear();
3618 DEBUG(dbgs() << "Promoting allocas with SSAUpdater...\n");
3620 DIBuilder DIB(*F.getParent());
3621 SmallVector<Instruction*, 64> Insts;
3623 for (unsigned Idx = 0, Size = PromotableAllocas.size(); Idx != Size; ++Idx) {
3624 AllocaInst *AI = PromotableAllocas[Idx];
3625 for (Value::use_iterator UI = AI->use_begin(), UE = AI->use_end();
3627 Instruction *I = cast<Instruction>(*UI++);
3628 // FIXME: Currently the SSAUpdater infrastructure doesn't reason about
3629 // lifetime intrinsics and so we strip them (and the bitcasts+GEPs
3630 // leading to them) here. Eventually it should use them to optimize the
3631 // scalar values produced.
3632 if (isa<BitCastInst>(I) || isa<GetElementPtrInst>(I)) {
3633 assert(onlyUsedByLifetimeMarkers(I) &&
3634 "Found a bitcast used outside of a lifetime marker.");
3635 while (!I->use_empty())
3636 cast<Instruction>(*I->use_begin())->eraseFromParent();
3637 I->eraseFromParent();
3640 if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(I)) {
3641 assert(II->getIntrinsicID() == Intrinsic::lifetime_start ||
3642 II->getIntrinsicID() == Intrinsic::lifetime_end);
3643 II->eraseFromParent();
3649 AllocaPromoter(Insts, SSA, *AI, DIB).run(Insts);
3653 PromotableAllocas.clear();
3658 /// \brief A predicate to test whether an alloca belongs to a set.
3659 class IsAllocaInSet {
3660 typedef SmallPtrSet<AllocaInst *, 4> SetType;
3664 typedef AllocaInst *argument_type;
3666 IsAllocaInSet(const SetType &Set) : Set(Set) {}
3667 bool operator()(AllocaInst *AI) const { return Set.count(AI); }
3671 bool SROA::runOnFunction(Function &F) {
3672 DEBUG(dbgs() << "SROA function: " << F.getName() << "\n");
3673 C = &F.getContext();
3674 TD = getAnalysisIfAvailable<DataLayout>();
3676 DEBUG(dbgs() << " Skipping SROA -- no target data!\n");
3679 DT = getAnalysisIfAvailable<DominatorTree>();
3681 BasicBlock &EntryBB = F.getEntryBlock();
3682 for (BasicBlock::iterator I = EntryBB.begin(), E = llvm::prior(EntryBB.end());
3684 if (AllocaInst *AI = dyn_cast<AllocaInst>(I))
3685 Worklist.insert(AI);
3687 bool Changed = false;
3688 // A set of deleted alloca instruction pointers which should be removed from
3689 // the list of promotable allocas.
3690 SmallPtrSet<AllocaInst *, 4> DeletedAllocas;
3693 while (!Worklist.empty()) {
3694 Changed |= runOnAlloca(*Worklist.pop_back_val());
3695 deleteDeadInstructions(DeletedAllocas);
3697 // Remove the deleted allocas from various lists so that we don't try to
3698 // continue processing them.
3699 if (!DeletedAllocas.empty()) {
3700 Worklist.remove_if(IsAllocaInSet(DeletedAllocas));
3701 PostPromotionWorklist.remove_if(IsAllocaInSet(DeletedAllocas));
3702 PromotableAllocas.erase(std::remove_if(PromotableAllocas.begin(),
3703 PromotableAllocas.end(),
3704 IsAllocaInSet(DeletedAllocas)),
3705 PromotableAllocas.end());
3706 DeletedAllocas.clear();
3710 Changed |= promoteAllocas(F);
3712 Worklist = PostPromotionWorklist;
3713 PostPromotionWorklist.clear();
3714 } while (!Worklist.empty());
3719 void SROA::getAnalysisUsage(AnalysisUsage &AU) const {
3720 if (RequiresDomTree)
3721 AU.addRequired<DominatorTree>();
3722 AU.setPreservesCFG();