1 //===-- Support/FoldingSet.cpp - Uniquing Hash Set --------------*- C++ -*-===// 2 // 3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 4 // See https://llvm.org/LICENSE.txt for license information. 5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6 // 7 //===----------------------------------------------------------------------===// 8 // 9 // This file implements a hash set that can be used to remove duplication of 10 // nodes in a graph. 11 // 12 //===----------------------------------------------------------------------===// 13 14 #include "llvm/ADT/FoldingSet.h" 15 #include "llvm/ADT/Hashing.h" 16 #include "llvm/ADT/StringRef.h" 17 #include "llvm/Support/Allocator.h" 18 #include "llvm/Support/ErrorHandling.h" 19 #include "llvm/Support/Host.h" 20 #include "llvm/Support/MathExtras.h" 21 #include <cassert> 22 #include <cstring> 23 using namespace llvm; 24 25 //===----------------------------------------------------------------------===// 26 // FoldingSetNodeIDRef Implementation 27 28 /// ComputeHash - Compute a strong hash value for this FoldingSetNodeIDRef, 29 /// used to lookup the node in the FoldingSetBase. 30 unsigned FoldingSetNodeIDRef::ComputeHash() const { 31 return static_cast<unsigned>(hash_combine_range(Data, Data+Size)); 32 } 33 34 bool FoldingSetNodeIDRef::operator==(FoldingSetNodeIDRef RHS) const { 35 if (Size != RHS.Size) return false; 36 return memcmp(Data, RHS.Data, Size*sizeof(*Data)) == 0; 37 } 38 39 /// Used to compare the "ordering" of two nodes as defined by the 40 /// profiled bits and their ordering defined by memcmp(). 41 bool FoldingSetNodeIDRef::operator<(FoldingSetNodeIDRef RHS) const { 42 if (Size != RHS.Size) 43 return Size < RHS.Size; 44 return memcmp(Data, RHS.Data, Size*sizeof(*Data)) < 0; 45 } 46 47 //===----------------------------------------------------------------------===// 48 // FoldingSetNodeID Implementation 49 50 /// Add* - Add various data types to Bit data. 51 /// 52 void FoldingSetNodeID::AddPointer(const void *Ptr) { 53 // Note: this adds pointers to the hash using sizes and endianness that 54 // depend on the host. It doesn't matter, however, because hashing on 55 // pointer values is inherently unstable. Nothing should depend on the 56 // ordering of nodes in the folding set. 57 static_assert(sizeof(uintptr_t) <= sizeof(unsigned long long), 58 "unexpected pointer size"); 59 AddInteger(reinterpret_cast<uintptr_t>(Ptr)); 60 } 61 void FoldingSetNodeID::AddInteger(signed I) { 62 Bits.push_back(I); 63 } 64 void FoldingSetNodeID::AddInteger(unsigned I) { 65 Bits.push_back(I); 66 } 67 void FoldingSetNodeID::AddInteger(long I) { 68 AddInteger((unsigned long)I); 69 } 70 void FoldingSetNodeID::AddInteger(unsigned long I) { 71 if (sizeof(long) == sizeof(int)) 72 AddInteger(unsigned(I)); 73 else if (sizeof(long) == sizeof(long long)) { 74 AddInteger((unsigned long long)I); 75 } else { 76 llvm_unreachable("unexpected sizeof(long)"); 77 } 78 } 79 void FoldingSetNodeID::AddInteger(long long I) { 80 AddInteger((unsigned long long)I); 81 } 82 void FoldingSetNodeID::AddInteger(unsigned long long I) { 83 AddInteger(unsigned(I)); 84 AddInteger(unsigned(I >> 32)); 85 } 86 87 void FoldingSetNodeID::AddString(StringRef String) { 88 unsigned Size = String.size(); 89 90 unsigned NumInserts = 1 + divideCeil(Size, 4); 91 Bits.reserve(Bits.size() + NumInserts); 92 93 Bits.push_back(Size); 94 if (!Size) return; 95 96 unsigned Units = Size / 4; 97 unsigned Pos = 0; 98 const unsigned *Base = (const unsigned*) String.data(); 99 100 // If the string is aligned do a bulk transfer. 101 if (!((intptr_t)Base & 3)) { 102 Bits.append(Base, Base + Units); 103 Pos = (Units + 1) * 4; 104 } else { 105 // Otherwise do it the hard way. 106 // To be compatible with above bulk transfer, we need to take endianness 107 // into account. 108 static_assert(sys::IsBigEndianHost || sys::IsLittleEndianHost, 109 "Unexpected host endianness"); 110 if (sys::IsBigEndianHost) { 111 for (Pos += 4; Pos <= Size; Pos += 4) { 112 unsigned V = ((unsigned char)String[Pos - 4] << 24) | 113 ((unsigned char)String[Pos - 3] << 16) | 114 ((unsigned char)String[Pos - 2] << 8) | 115 (unsigned char)String[Pos - 1]; 116 Bits.push_back(V); 117 } 118 } else { // Little-endian host 119 for (Pos += 4; Pos <= Size; Pos += 4) { 120 unsigned V = ((unsigned char)String[Pos - 1] << 24) | 121 ((unsigned char)String[Pos - 2] << 16) | 122 ((unsigned char)String[Pos - 3] << 8) | 123 (unsigned char)String[Pos - 4]; 124 Bits.push_back(V); 125 } 126 } 127 } 128 129 // With the leftover bits. 130 unsigned V = 0; 131 // Pos will have overshot size by 4 - #bytes left over. 132 // No need to take endianness into account here - this is always executed. 133 switch (Pos - Size) { 134 case 1: V = (V << 8) | (unsigned char)String[Size - 3]; LLVM_FALLTHROUGH; 135 case 2: V = (V << 8) | (unsigned char)String[Size - 2]; LLVM_FALLTHROUGH; 136 case 3: V = (V << 8) | (unsigned char)String[Size - 1]; break; 137 default: return; // Nothing left. 138 } 139 140 Bits.push_back(V); 141 } 142 143 // AddNodeID - Adds the Bit data of another ID to *this. 144 void FoldingSetNodeID::AddNodeID(const FoldingSetNodeID &ID) { 145 Bits.append(ID.Bits.begin(), ID.Bits.end()); 146 } 147 148 /// ComputeHash - Compute a strong hash value for this FoldingSetNodeID, used to 149 /// lookup the node in the FoldingSetBase. 150 unsigned FoldingSetNodeID::ComputeHash() const { 151 return FoldingSetNodeIDRef(Bits.data(), Bits.size()).ComputeHash(); 152 } 153 154 /// operator== - Used to compare two nodes to each other. 155 /// 156 bool FoldingSetNodeID::operator==(const FoldingSetNodeID &RHS) const { 157 return *this == FoldingSetNodeIDRef(RHS.Bits.data(), RHS.Bits.size()); 158 } 159 160 /// operator== - Used to compare two nodes to each other. 161 /// 162 bool FoldingSetNodeID::operator==(FoldingSetNodeIDRef RHS) const { 163 return FoldingSetNodeIDRef(Bits.data(), Bits.size()) == RHS; 164 } 165 166 /// Used to compare the "ordering" of two nodes as defined by the 167 /// profiled bits and their ordering defined by memcmp(). 168 bool FoldingSetNodeID::operator<(const FoldingSetNodeID &RHS) const { 169 return *this < FoldingSetNodeIDRef(RHS.Bits.data(), RHS.Bits.size()); 170 } 171 172 bool FoldingSetNodeID::operator<(FoldingSetNodeIDRef RHS) const { 173 return FoldingSetNodeIDRef(Bits.data(), Bits.size()) < RHS; 174 } 175 176 /// Intern - Copy this node's data to a memory region allocated from the 177 /// given allocator and return a FoldingSetNodeIDRef describing the 178 /// interned data. 179 FoldingSetNodeIDRef 180 FoldingSetNodeID::Intern(BumpPtrAllocator &Allocator) const { 181 unsigned *New = Allocator.Allocate<unsigned>(Bits.size()); 182 std::uninitialized_copy(Bits.begin(), Bits.end(), New); 183 return FoldingSetNodeIDRef(New, Bits.size()); 184 } 185 186 //===----------------------------------------------------------------------===// 187 /// Helper functions for FoldingSetBase. 188 189 /// GetNextPtr - In order to save space, each bucket is a 190 /// singly-linked-list. In order to make deletion more efficient, we make 191 /// the list circular, so we can delete a node without computing its hash. 192 /// The problem with this is that the start of the hash buckets are not 193 /// Nodes. If NextInBucketPtr is a bucket pointer, this method returns null: 194 /// use GetBucketPtr when this happens. 195 static FoldingSetBase::Node *GetNextPtr(void *NextInBucketPtr) { 196 // The low bit is set if this is the pointer back to the bucket. 197 if (reinterpret_cast<intptr_t>(NextInBucketPtr) & 1) 198 return nullptr; 199 200 return static_cast<FoldingSetBase::Node*>(NextInBucketPtr); 201 } 202 203 204 /// testing. 205 static void **GetBucketPtr(void *NextInBucketPtr) { 206 intptr_t Ptr = reinterpret_cast<intptr_t>(NextInBucketPtr); 207 assert((Ptr & 1) && "Not a bucket pointer"); 208 return reinterpret_cast<void**>(Ptr & ~intptr_t(1)); 209 } 210 211 /// GetBucketFor - Hash the specified node ID and return the hash bucket for 212 /// the specified ID. 213 static void **GetBucketFor(unsigned Hash, void **Buckets, unsigned NumBuckets) { 214 // NumBuckets is always a power of 2. 215 unsigned BucketNum = Hash & (NumBuckets-1); 216 return Buckets + BucketNum; 217 } 218 219 /// AllocateBuckets - Allocated initialized bucket memory. 220 static void **AllocateBuckets(unsigned NumBuckets) { 221 void **Buckets = static_cast<void**>(safe_calloc(NumBuckets + 1, 222 sizeof(void*))); 223 // Set the very last bucket to be a non-null "pointer". 224 Buckets[NumBuckets] = reinterpret_cast<void*>(-1); 225 return Buckets; 226 } 227 228 //===----------------------------------------------------------------------===// 229 // FoldingSetBase Implementation 230 231 FoldingSetBase::FoldingSetBase(unsigned Log2InitSize) { 232 assert(5 < Log2InitSize && Log2InitSize < 32 && 233 "Initial hash table size out of range"); 234 NumBuckets = 1 << Log2InitSize; 235 Buckets = AllocateBuckets(NumBuckets); 236 NumNodes = 0; 237 } 238 239 FoldingSetBase::FoldingSetBase(FoldingSetBase &&Arg) 240 : Buckets(Arg.Buckets), NumBuckets(Arg.NumBuckets), NumNodes(Arg.NumNodes) { 241 Arg.Buckets = nullptr; 242 Arg.NumBuckets = 0; 243 Arg.NumNodes = 0; 244 } 245 246 FoldingSetBase &FoldingSetBase::operator=(FoldingSetBase &&RHS) { 247 free(Buckets); // This may be null if the set is in a moved-from state. 248 Buckets = RHS.Buckets; 249 NumBuckets = RHS.NumBuckets; 250 NumNodes = RHS.NumNodes; 251 RHS.Buckets = nullptr; 252 RHS.NumBuckets = 0; 253 RHS.NumNodes = 0; 254 return *this; 255 } 256 257 FoldingSetBase::~FoldingSetBase() { 258 free(Buckets); 259 } 260 261 void FoldingSetBase::clear() { 262 // Set all but the last bucket to null pointers. 263 memset(Buckets, 0, NumBuckets*sizeof(void*)); 264 265 // Set the very last bucket to be a non-null "pointer". 266 Buckets[NumBuckets] = reinterpret_cast<void*>(-1); 267 268 // Reset the node count to zero. 269 NumNodes = 0; 270 } 271 272 void FoldingSetBase::GrowBucketCount(unsigned NewBucketCount, 273 const FoldingSetInfo &Info) { 274 assert((NewBucketCount > NumBuckets) && 275 "Can't shrink a folding set with GrowBucketCount"); 276 assert(isPowerOf2_32(NewBucketCount) && "Bad bucket count!"); 277 void **OldBuckets = Buckets; 278 unsigned OldNumBuckets = NumBuckets; 279 280 // Clear out new buckets. 281 Buckets = AllocateBuckets(NewBucketCount); 282 // Set NumBuckets only if allocation of new buckets was successful. 283 NumBuckets = NewBucketCount; 284 NumNodes = 0; 285 286 // Walk the old buckets, rehashing nodes into their new place. 287 FoldingSetNodeID TempID; 288 for (unsigned i = 0; i != OldNumBuckets; ++i) { 289 void *Probe = OldBuckets[i]; 290 if (!Probe) continue; 291 while (Node *NodeInBucket = GetNextPtr(Probe)) { 292 // Figure out the next link, remove NodeInBucket from the old link. 293 Probe = NodeInBucket->getNextInBucket(); 294 NodeInBucket->SetNextInBucket(nullptr); 295 296 // Insert the node into the new bucket, after recomputing the hash. 297 InsertNode(NodeInBucket, 298 GetBucketFor(Info.ComputeNodeHash(this, NodeInBucket, TempID), 299 Buckets, NumBuckets), 300 Info); 301 TempID.clear(); 302 } 303 } 304 305 free(OldBuckets); 306 } 307 308 /// GrowHashTable - Double the size of the hash table and rehash everything. 309 /// 310 void FoldingSetBase::GrowHashTable(const FoldingSetInfo &Info) { 311 GrowBucketCount(NumBuckets * 2, Info); 312 } 313 314 void FoldingSetBase::reserve(unsigned EltCount, const FoldingSetInfo &Info) { 315 // This will give us somewhere between EltCount / 2 and 316 // EltCount buckets. This puts us in the load factor 317 // range of 1.0 - 2.0. 318 if(EltCount < capacity()) 319 return; 320 GrowBucketCount(PowerOf2Floor(EltCount), Info); 321 } 322 323 /// FindNodeOrInsertPos - Look up the node specified by ID. If it exists, 324 /// return it. If not, return the insertion token that will make insertion 325 /// faster. 326 FoldingSetBase::Node *FoldingSetBase::FindNodeOrInsertPos( 327 const FoldingSetNodeID &ID, void *&InsertPos, const FoldingSetInfo &Info) { 328 unsigned IDHash = ID.ComputeHash(); 329 void **Bucket = GetBucketFor(IDHash, Buckets, NumBuckets); 330 void *Probe = *Bucket; 331 332 InsertPos = nullptr; 333 334 FoldingSetNodeID TempID; 335 while (Node *NodeInBucket = GetNextPtr(Probe)) { 336 if (Info.NodeEquals(this, NodeInBucket, ID, IDHash, TempID)) 337 return NodeInBucket; 338 TempID.clear(); 339 340 Probe = NodeInBucket->getNextInBucket(); 341 } 342 343 // Didn't find the node, return null with the bucket as the InsertPos. 344 InsertPos = Bucket; 345 return nullptr; 346 } 347 348 /// InsertNode - Insert the specified node into the folding set, knowing that it 349 /// is not already in the map. InsertPos must be obtained from 350 /// FindNodeOrInsertPos. 351 void FoldingSetBase::InsertNode(Node *N, void *InsertPos, 352 const FoldingSetInfo &Info) { 353 assert(!N->getNextInBucket()); 354 // Do we need to grow the hashtable? 355 if (NumNodes+1 > capacity()) { 356 GrowHashTable(Info); 357 FoldingSetNodeID TempID; 358 InsertPos = GetBucketFor(Info.ComputeNodeHash(this, N, TempID), Buckets, 359 NumBuckets); 360 } 361 362 ++NumNodes; 363 364 /// The insert position is actually a bucket pointer. 365 void **Bucket = static_cast<void**>(InsertPos); 366 367 void *Next = *Bucket; 368 369 // If this is the first insertion into this bucket, its next pointer will be 370 // null. Pretend as if it pointed to itself, setting the low bit to indicate 371 // that it is a pointer to the bucket. 372 if (!Next) 373 Next = reinterpret_cast<void*>(reinterpret_cast<intptr_t>(Bucket)|1); 374 375 // Set the node's next pointer, and make the bucket point to the node. 376 N->SetNextInBucket(Next); 377 *Bucket = N; 378 } 379 380 /// RemoveNode - Remove a node from the folding set, returning true if one was 381 /// removed or false if the node was not in the folding set. 382 bool FoldingSetBase::RemoveNode(Node *N) { 383 // Because each bucket is a circular list, we don't need to compute N's hash 384 // to remove it. 385 void *Ptr = N->getNextInBucket(); 386 if (!Ptr) return false; // Not in folding set. 387 388 --NumNodes; 389 N->SetNextInBucket(nullptr); 390 391 // Remember what N originally pointed to, either a bucket or another node. 392 void *NodeNextPtr = Ptr; 393 394 // Chase around the list until we find the node (or bucket) which points to N. 395 while (true) { 396 if (Node *NodeInBucket = GetNextPtr(Ptr)) { 397 // Advance pointer. 398 Ptr = NodeInBucket->getNextInBucket(); 399 400 // We found a node that points to N, change it to point to N's next node, 401 // removing N from the list. 402 if (Ptr == N) { 403 NodeInBucket->SetNextInBucket(NodeNextPtr); 404 return true; 405 } 406 } else { 407 void **Bucket = GetBucketPtr(Ptr); 408 Ptr = *Bucket; 409 410 // If we found that the bucket points to N, update the bucket to point to 411 // whatever is next. 412 if (Ptr == N) { 413 *Bucket = NodeNextPtr; 414 return true; 415 } 416 } 417 } 418 } 419 420 /// GetOrInsertNode - If there is an existing simple Node exactly 421 /// equal to the specified node, return it. Otherwise, insert 'N' and it 422 /// instead. 423 FoldingSetBase::Node * 424 FoldingSetBase::GetOrInsertNode(FoldingSetBase::Node *N, 425 const FoldingSetInfo &Info) { 426 FoldingSetNodeID ID; 427 Info.GetNodeProfile(this, N, ID); 428 void *IP; 429 if (Node *E = FindNodeOrInsertPos(ID, IP, Info)) 430 return E; 431 InsertNode(N, IP, Info); 432 return N; 433 } 434 435 //===----------------------------------------------------------------------===// 436 // FoldingSetIteratorImpl Implementation 437 438 FoldingSetIteratorImpl::FoldingSetIteratorImpl(void **Bucket) { 439 // Skip to the first non-null non-self-cycle bucket. 440 while (*Bucket != reinterpret_cast<void*>(-1) && 441 (!*Bucket || !GetNextPtr(*Bucket))) 442 ++Bucket; 443 444 NodePtr = static_cast<FoldingSetNode*>(*Bucket); 445 } 446 447 void FoldingSetIteratorImpl::advance() { 448 // If there is another link within this bucket, go to it. 449 void *Probe = NodePtr->getNextInBucket(); 450 451 if (FoldingSetNode *NextNodeInBucket = GetNextPtr(Probe)) 452 NodePtr = NextNodeInBucket; 453 else { 454 // Otherwise, this is the last link in this bucket. 455 void **Bucket = GetBucketPtr(Probe); 456 457 // Skip to the next non-null non-self-cycle bucket. 458 do { 459 ++Bucket; 460 } while (*Bucket != reinterpret_cast<void*>(-1) && 461 (!*Bucket || !GetNextPtr(*Bucket))); 462 463 NodePtr = static_cast<FoldingSetNode*>(*Bucket); 464 } 465 } 466 467 //===----------------------------------------------------------------------===// 468 // FoldingSetBucketIteratorImpl Implementation 469 470 FoldingSetBucketIteratorImpl::FoldingSetBucketIteratorImpl(void **Bucket) { 471 Ptr = (!*Bucket || !GetNextPtr(*Bucket)) ? (void*) Bucket : *Bucket; 472 } 473