| 1 | //===- ExplodedGraph.cpp - Local, Path-Sens. "Exploded Graph" -------------===// |
| 2 | // |
| 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | // |
| 9 | // This file defines the template classes ExplodedNode and ExplodedGraph, |
| 10 | // which represent a path-sensitive, intra-procedural "exploded graph." |
| 11 | // |
| 12 | //===----------------------------------------------------------------------===// |
| 13 | |
| 14 | #include "clang/StaticAnalyzer/Core/PathSensitive/ExplodedGraph.h" |
| 15 | #include "clang/AST/Expr.h" |
| 16 | #include "clang/AST/ExprObjC.h" |
| 17 | #include "clang/AST/ParentMap.h" |
| 18 | #include "clang/AST/Stmt.h" |
| 19 | #include "clang/Analysis/CFGStmtMap.h" |
| 20 | #include "clang/Analysis/ProgramPoint.h" |
| 21 | #include "clang/Analysis/Support/BumpVector.h" |
| 22 | #include "clang/Basic/LLVM.h" |
| 23 | #include "clang/StaticAnalyzer/Core/PathSensitive/CallEvent.h" |
| 24 | #include "clang/StaticAnalyzer/Core/PathSensitive/ProgramState.h" |
| 25 | #include "clang/StaticAnalyzer/Core/PathSensitive/ProgramState_Fwd.h" |
| 26 | #include "llvm/ADT/DenseSet.h" |
| 27 | #include "llvm/ADT/FoldingSet.h" |
| 28 | #include "llvm/ADT/PointerUnion.h" |
| 29 | #include <cassert> |
| 30 | #include <memory> |
| 31 | #include <optional> |
| 32 | |
| 33 | using namespace clang; |
| 34 | using namespace ento; |
| 35 | |
| 36 | //===----------------------------------------------------------------------===// |
| 37 | // Cleanup. |
| 38 | //===----------------------------------------------------------------------===// |
| 39 | |
| 40 | ExplodedGraph::ExplodedGraph() = default; |
| 41 | |
| 42 | ExplodedGraph::~ExplodedGraph() = default; |
| 43 | |
| 44 | //===----------------------------------------------------------------------===// |
| 45 | // Node reclamation. |
| 46 | //===----------------------------------------------------------------------===// |
| 47 | |
| 48 | bool ExplodedGraph::isInterestingLValueExpr(const Expr *Ex) { |
| 49 | if (!Ex->isLValue()) |
| 50 | return false; |
| 51 | return isa<DeclRefExpr, MemberExpr, ObjCIvarRefExpr, ArraySubscriptExpr>(Val: Ex); |
| 52 | } |
| 53 | |
| 54 | bool ExplodedGraph::shouldCollect(const ExplodedNode *node) { |
| 55 | // First, we only consider nodes for reclamation of the following |
| 56 | // conditions apply: |
| 57 | // |
| 58 | // (1) 1 predecessor (that has one successor) |
| 59 | // (2) 1 successor (that has one predecessor) |
| 60 | // |
| 61 | // If a node has no successor it is on the "frontier", while a node |
| 62 | // with no predecessor is a root. |
| 63 | // |
| 64 | // After these prerequisites, we discard all "filler" nodes that |
| 65 | // are used only for intermediate processing, and are not essential |
| 66 | // for analyzer history: |
| 67 | // |
| 68 | // (a) PreStmtPurgeDeadSymbols |
| 69 | // |
| 70 | // We then discard all other nodes where *all* of the following conditions |
| 71 | // apply: |
| 72 | // |
| 73 | // (3) The ProgramPoint is for a PostStmt, but not a PostStore. |
| 74 | // (4) There is no 'tag' for the ProgramPoint. |
| 75 | // (5) The 'store' is the same as the predecessor. |
| 76 | // (6) The 'GDM' is the same as the predecessor. |
| 77 | // (7) The LocationContext is the same as the predecessor. |
| 78 | // (8) Expressions that are *not* lvalue expressions. |
| 79 | // (9) The PostStmt isn't for a non-consumed Stmt or Expr. |
| 80 | // (10) The successor is neither a CallExpr StmtPoint nor a CallEnter or |
| 81 | // PreImplicitCall (so that we would be able to find it when retrying a |
| 82 | // call with no inlining). |
| 83 | // FIXME: It may be safe to reclaim PreCall and PostCall nodes as well. |
| 84 | |
| 85 | // Conditions 1 and 2. |
| 86 | if (node->pred_size() != 1 || node->succ_size() != 1) |
| 87 | return false; |
| 88 | |
| 89 | const ExplodedNode *pred = *(node->pred_begin()); |
| 90 | if (pred->succ_size() != 1) |
| 91 | return false; |
| 92 | |
| 93 | const ExplodedNode *succ = *(node->succ_begin()); |
| 94 | if (succ->pred_size() != 1) |
| 95 | return false; |
| 96 | |
| 97 | // Now reclaim any nodes that are (by definition) not essential to |
| 98 | // analysis history and are not consulted by any client code. |
| 99 | ProgramPoint progPoint = node->getLocation(); |
| 100 | if (progPoint.getAs<PreStmtPurgeDeadSymbols>()) |
| 101 | return !progPoint.getTag(); |
| 102 | |
| 103 | // Condition 3. |
| 104 | if (!progPoint.getAs<PostStmt>() || progPoint.getAs<PostStore>()) |
| 105 | return false; |
| 106 | |
| 107 | // Condition 4. |
| 108 | if (progPoint.getTag()) |
| 109 | return false; |
| 110 | |
| 111 | // Conditions 5, 6, and 7. |
| 112 | ProgramStateRef state = node->getState(); |
| 113 | ProgramStateRef pred_state = pred->getState(); |
| 114 | if (state->store != pred_state->store || state->GDM != pred_state->GDM || |
| 115 | progPoint.getLocationContext() != pred->getLocationContext()) |
| 116 | return false; |
| 117 | |
| 118 | // All further checks require expressions. As per #3, we know that we have |
| 119 | // a PostStmt. |
| 120 | const Expr *Ex = dyn_cast<Expr>(Val: progPoint.castAs<PostStmt>().getStmt()); |
| 121 | if (!Ex) |
| 122 | return false; |
| 123 | |
| 124 | // Condition 8. |
| 125 | // Do not collect nodes for "interesting" lvalue expressions since they are |
| 126 | // used extensively for generating path diagnostics. |
| 127 | if (isInterestingLValueExpr(Ex)) |
| 128 | return false; |
| 129 | |
| 130 | // Condition 9. |
| 131 | // Do not collect nodes for non-consumed Stmt or Expr to ensure precise |
| 132 | // diagnostic generation; specifically, so that we could anchor arrows |
| 133 | // pointing to the beginning of statements (as written in code). |
| 134 | const ParentMap &PM = progPoint.getLocationContext()->getParentMap(); |
| 135 | if (!PM.isConsumedExpr(E: Ex)) |
| 136 | return false; |
| 137 | |
| 138 | // Condition 10. |
| 139 | const ProgramPoint SuccLoc = succ->getLocation(); |
| 140 | if (std::optional<StmtPoint> SP = SuccLoc.getAs<StmtPoint>()) |
| 141 | if (CallEvent::isCallStmt(S: SP->getStmt())) |
| 142 | return false; |
| 143 | |
| 144 | // Condition 10, continuation. |
| 145 | if (SuccLoc.getAs<CallEnter>() || SuccLoc.getAs<PreImplicitCall>()) |
| 146 | return false; |
| 147 | |
| 148 | return true; |
| 149 | } |
| 150 | |
| 151 | void ExplodedGraph::collectNode(ExplodedNode *node) { |
| 152 | // Removing a node means: |
| 153 | // (a) changing the predecessors successor to the successor of this node |
| 154 | // (b) changing the successors predecessor to the predecessor of this node |
| 155 | // (c) Putting 'node' onto freeNodes. |
| 156 | assert(node->pred_size() == 1 || node->succ_size() == 1); |
| 157 | ExplodedNode *pred = *(node->pred_begin()); |
| 158 | ExplodedNode *succ = *(node->succ_begin()); |
| 159 | pred->replaceSuccessor(node: succ); |
| 160 | succ->replacePredecessor(node: pred); |
| 161 | FreeNodes.push_back(x: node); |
| 162 | Nodes.RemoveNode(N: node); |
| 163 | --NumNodes; |
| 164 | node->~ExplodedNode(); |
| 165 | } |
| 166 | |
| 167 | void ExplodedGraph::reclaimRecentlyAllocatedNodes() { |
| 168 | if (ChangedNodes.empty()) |
| 169 | return; |
| 170 | |
| 171 | // Only periodically reclaim nodes so that we can build up a set of |
| 172 | // nodes that meet the reclamation criteria. Freshly created nodes |
| 173 | // by definition have no successor, and thus cannot be reclaimed (see below). |
| 174 | assert(ReclaimCounter > 0); |
| 175 | if (--ReclaimCounter != 0) |
| 176 | return; |
| 177 | ReclaimCounter = ReclaimNodeInterval; |
| 178 | |
| 179 | for (const auto node : ChangedNodes) |
| 180 | if (shouldCollect(node)) |
| 181 | collectNode(node); |
| 182 | ChangedNodes.clear(); |
| 183 | } |
| 184 | |
| 185 | //===----------------------------------------------------------------------===// |
| 186 | // ExplodedNode. |
| 187 | //===----------------------------------------------------------------------===// |
| 188 | |
| 189 | // An NodeGroup's storage type is actually very much like a TinyPtrVector: |
| 190 | // it can be either a pointer to a single ExplodedNode, or a pointer to a |
| 191 | // BumpVector allocated with the ExplodedGraph's allocator. This allows the |
| 192 | // common case of single-node NodeGroups to be implemented with no extra memory. |
| 193 | // |
| 194 | // Consequently, each of the NodeGroup methods have up to four cases to handle: |
| 195 | // 1. The flag is set and this group does not actually contain any nodes. |
| 196 | // 2. The group is empty, in which case the storage value is null. |
| 197 | // 3. The group contains a single node. |
| 198 | // 4. The group contains more than one node. |
| 199 | using ExplodedNodeVector = BumpVector<ExplodedNode *>; |
| 200 | using GroupStorage = llvm::PointerUnion<ExplodedNode *, ExplodedNodeVector *>; |
| 201 | |
| 202 | void ExplodedNode::addPredecessor(ExplodedNode *V, ExplodedGraph &G) { |
| 203 | assert(!V->isSink()); |
| 204 | Preds.addNode(N: V, G); |
| 205 | V->Succs.addNode(N: this, G); |
| 206 | } |
| 207 | |
| 208 | void ExplodedNode::NodeGroup::replaceNode(ExplodedNode *node) { |
| 209 | assert(!getFlag()); |
| 210 | |
| 211 | GroupStorage &Storage = reinterpret_cast<GroupStorage&>(P); |
| 212 | assert(isa<ExplodedNode *>(Storage)); |
| 213 | Storage = node; |
| 214 | assert(isa<ExplodedNode *>(Storage)); |
| 215 | } |
| 216 | |
| 217 | void ExplodedNode::NodeGroup::addNode(ExplodedNode *N, ExplodedGraph &G) { |
| 218 | assert(!getFlag()); |
| 219 | |
| 220 | GroupStorage &Storage = reinterpret_cast<GroupStorage&>(P); |
| 221 | if (Storage.isNull()) { |
| 222 | Storage = N; |
| 223 | assert(isa<ExplodedNode *>(Storage)); |
| 224 | return; |
| 225 | } |
| 226 | |
| 227 | ExplodedNodeVector *V = dyn_cast<ExplodedNodeVector *>(Val&: Storage); |
| 228 | |
| 229 | if (!V) { |
| 230 | // Switch from single-node to multi-node representation. |
| 231 | auto *Old = cast<ExplodedNode *>(Val&: Storage); |
| 232 | |
| 233 | BumpVectorContext &Ctx = G.getNodeAllocator(); |
| 234 | V = new (G.getAllocator()) ExplodedNodeVector(Ctx, 4); |
| 235 | V->push_back(Elt: Old, C&: Ctx); |
| 236 | |
| 237 | Storage = V; |
| 238 | assert(!getFlag()); |
| 239 | assert(isa<ExplodedNodeVector *>(Storage)); |
| 240 | } |
| 241 | |
| 242 | V->push_back(Elt: N, C&: G.getNodeAllocator()); |
| 243 | } |
| 244 | |
| 245 | unsigned ExplodedNode::NodeGroup::size() const { |
| 246 | if (getFlag()) |
| 247 | return 0; |
| 248 | |
| 249 | const GroupStorage &Storage = reinterpret_cast<const GroupStorage &>(P); |
| 250 | if (Storage.isNull()) |
| 251 | return 0; |
| 252 | if (ExplodedNodeVector *V = dyn_cast<ExplodedNodeVector *>(Val: Storage)) |
| 253 | return V->size(); |
| 254 | return 1; |
| 255 | } |
| 256 | |
| 257 | ExplodedNode * const *ExplodedNode::NodeGroup::begin() const { |
| 258 | if (getFlag()) |
| 259 | return nullptr; |
| 260 | |
| 261 | const GroupStorage &Storage = reinterpret_cast<const GroupStorage &>(P); |
| 262 | if (Storage.isNull()) |
| 263 | return nullptr; |
| 264 | if (ExplodedNodeVector *V = dyn_cast<ExplodedNodeVector *>(Val: Storage)) |
| 265 | return V->begin(); |
| 266 | return Storage.getAddrOfPtr1(); |
| 267 | } |
| 268 | |
| 269 | ExplodedNode * const *ExplodedNode::NodeGroup::end() const { |
| 270 | if (getFlag()) |
| 271 | return nullptr; |
| 272 | |
| 273 | const GroupStorage &Storage = reinterpret_cast<const GroupStorage &>(P); |
| 274 | if (Storage.isNull()) |
| 275 | return nullptr; |
| 276 | if (ExplodedNodeVector *V = dyn_cast<ExplodedNodeVector *>(Val: Storage)) |
| 277 | return V->end(); |
| 278 | return Storage.getAddrOfPtr1() + 1; |
| 279 | } |
| 280 | |
| 281 | bool ExplodedNode::isTrivial() const { |
| 282 | return pred_size() == 1 && succ_size() == 1 && |
| 283 | getFirstPred()->getState()->getID() == getState()->getID() && |
| 284 | getFirstPred()->succ_size() == 1; |
| 285 | } |
| 286 | |
| 287 | const CFGBlock *ExplodedNode::getCFGBlock() const { |
| 288 | ProgramPoint P = getLocation(); |
| 289 | if (auto BEP = P.getAs<BlockEntrance>()) |
| 290 | return BEP->getBlock(); |
| 291 | |
| 292 | // Find the node's current statement in the CFG. |
| 293 | // FIXME: getStmtForDiagnostics() does nasty things in order to provide |
| 294 | // a valid statement for body farms, do we need this behavior here? |
| 295 | if (const Stmt *S = getStmtForDiagnostics()) |
| 296 | return getLocationContext() |
| 297 | ->getAnalysisDeclContext() |
| 298 | ->getCFGStmtMap() |
| 299 | ->getBlock(S); |
| 300 | |
| 301 | return nullptr; |
| 302 | } |
| 303 | |
| 304 | static const LocationContext * |
| 305 | findTopAutosynthesizedParentContext(const LocationContext *LC) { |
| 306 | assert(LC->getAnalysisDeclContext()->isBodyAutosynthesized()); |
| 307 | const LocationContext *ParentLC = LC->getParent(); |
| 308 | assert(ParentLC && "We don't start analysis from autosynthesized code" ); |
| 309 | while (ParentLC->getAnalysisDeclContext()->isBodyAutosynthesized()) { |
| 310 | LC = ParentLC; |
| 311 | ParentLC = LC->getParent(); |
| 312 | assert(ParentLC && "We don't start analysis from autosynthesized code" ); |
| 313 | } |
| 314 | return LC; |
| 315 | } |
| 316 | |
| 317 | const Stmt *ExplodedNode::getStmtForDiagnostics() const { |
| 318 | // We cannot place diagnostics on autosynthesized code. |
| 319 | // Put them onto the call site through which we jumped into autosynthesized |
| 320 | // code for the first time. |
| 321 | const LocationContext *LC = getLocationContext(); |
| 322 | if (LC->getAnalysisDeclContext()->isBodyAutosynthesized()) { |
| 323 | // It must be a stack frame because we only autosynthesize functions. |
| 324 | return cast<StackFrameContext>(Val: findTopAutosynthesizedParentContext(LC)) |
| 325 | ->getCallSite(); |
| 326 | } |
| 327 | // Otherwise, see if the node's program point directly points to a statement. |
| 328 | // FIXME: Refactor into a ProgramPoint method? |
| 329 | ProgramPoint P = getLocation(); |
| 330 | if (auto SP = P.getAs<StmtPoint>()) |
| 331 | return SP->getStmt(); |
| 332 | if (auto BE = P.getAs<BlockEdge>()) |
| 333 | return BE->getSrc()->getTerminatorStmt(); |
| 334 | if (auto CE = P.getAs<CallEnter>()) |
| 335 | return CE->getCallExpr(); |
| 336 | if (auto CEE = P.getAs<CallExitEnd>()) |
| 337 | return CEE->getCalleeContext()->getCallSite(); |
| 338 | if (auto PIPP = P.getAs<PostInitializer>()) |
| 339 | return PIPP->getInitializer()->getInit(); |
| 340 | if (auto CEB = P.getAs<CallExitBegin>()) |
| 341 | return CEB->getReturnStmt(); |
| 342 | if (auto FEP = P.getAs<FunctionExitPoint>()) |
| 343 | return FEP->getStmt(); |
| 344 | |
| 345 | return nullptr; |
| 346 | } |
| 347 | |
| 348 | const Stmt *ExplodedNode::getNextStmtForDiagnostics() const { |
| 349 | for (const ExplodedNode *N = getFirstSucc(); N; N = N->getFirstSucc()) { |
| 350 | if (N->getLocation().isPurgeKind()) |
| 351 | continue; |
| 352 | if (const Stmt *S = N->getStmtForDiagnostics()) { |
| 353 | // Check if the statement is '?' or '&&'/'||'. These are "merges", |
| 354 | // not actual statement points. |
| 355 | switch (S->getStmtClass()) { |
| 356 | case Stmt::ChooseExprClass: |
| 357 | case Stmt::BinaryConditionalOperatorClass: |
| 358 | case Stmt::ConditionalOperatorClass: |
| 359 | continue; |
| 360 | case Stmt::BinaryOperatorClass: { |
| 361 | BinaryOperatorKind Op = cast<BinaryOperator>(Val: S)->getOpcode(); |
| 362 | if (Op == BO_LAnd || Op == BO_LOr) |
| 363 | continue; |
| 364 | break; |
| 365 | } |
| 366 | default: |
| 367 | break; |
| 368 | } |
| 369 | // We found the statement, so return it. |
| 370 | return S; |
| 371 | } |
| 372 | } |
| 373 | |
| 374 | return nullptr; |
| 375 | } |
| 376 | |
| 377 | const Stmt *ExplodedNode::getPreviousStmtForDiagnostics() const { |
| 378 | for (const ExplodedNode *N = getFirstPred(); N; N = N->getFirstPred()) |
| 379 | if (const Stmt *S = N->getStmtForDiagnostics(); S && !isa<CompoundStmt>(Val: S)) |
| 380 | return S; |
| 381 | |
| 382 | return nullptr; |
| 383 | } |
| 384 | |
| 385 | const Stmt *ExplodedNode::getCurrentOrPreviousStmtForDiagnostics() const { |
| 386 | if (const Stmt *S = getStmtForDiagnostics()) |
| 387 | return S; |
| 388 | |
| 389 | return getPreviousStmtForDiagnostics(); |
| 390 | } |
| 391 | |
| 392 | ExplodedNode *ExplodedGraph::getNode(const ProgramPoint &L, |
| 393 | ProgramStateRef State, |
| 394 | bool IsSink, |
| 395 | bool* IsNew) { |
| 396 | // Profile 'State' to determine if we already have an existing node. |
| 397 | llvm::FoldingSetNodeID profile; |
| 398 | void *InsertPos = nullptr; |
| 399 | |
| 400 | NodeTy::Profile(ID&: profile, Loc: L, state: State, IsSink); |
| 401 | NodeTy* V = Nodes.FindNodeOrInsertPos(ID: profile, InsertPos); |
| 402 | |
| 403 | if (!V) { |
| 404 | if (!FreeNodes.empty()) { |
| 405 | V = FreeNodes.back(); |
| 406 | FreeNodes.pop_back(); |
| 407 | } |
| 408 | else { |
| 409 | // Allocate a new node. |
| 410 | V = getAllocator().Allocate<NodeTy>(); |
| 411 | } |
| 412 | |
| 413 | ++NumNodes; |
| 414 | new (V) NodeTy(L, State, NumNodes, IsSink); |
| 415 | |
| 416 | if (ReclaimNodeInterval) |
| 417 | ChangedNodes.push_back(x: V); |
| 418 | |
| 419 | // Insert the node into the node set and return it. |
| 420 | Nodes.InsertNode(N: V, InsertPos); |
| 421 | |
| 422 | if (IsNew) *IsNew = true; |
| 423 | } |
| 424 | else |
| 425 | if (IsNew) *IsNew = false; |
| 426 | |
| 427 | return V; |
| 428 | } |
| 429 | |
| 430 | ExplodedNode *ExplodedGraph::createUncachedNode(const ProgramPoint &L, |
| 431 | ProgramStateRef State, |
| 432 | int64_t Id, |
| 433 | bool IsSink) { |
| 434 | NodeTy *V = getAllocator().Allocate<NodeTy>(); |
| 435 | new (V) NodeTy(L, State, Id, IsSink); |
| 436 | return V; |
| 437 | } |
| 438 | |
| 439 | std::unique_ptr<ExplodedGraph> |
| 440 | ExplodedGraph::trim(ArrayRef<const NodeTy *> Sinks, |
| 441 | InterExplodedGraphMap *ForwardMap, |
| 442 | InterExplodedGraphMap *InverseMap) const { |
| 443 | // FIXME: The two-pass algorithm of this function (which was introduced in |
| 444 | // 2008) is terribly overcomplicated and should be replaced by a single |
| 445 | // (backward) pass. |
| 446 | |
| 447 | if (Nodes.empty()) |
| 448 | return nullptr; |
| 449 | |
| 450 | using Pass1Ty = llvm::DenseSet<const ExplodedNode *>; |
| 451 | Pass1Ty Pass1; |
| 452 | |
| 453 | using Pass2Ty = InterExplodedGraphMap; |
| 454 | InterExplodedGraphMap Pass2Scratch; |
| 455 | Pass2Ty &Pass2 = ForwardMap ? *ForwardMap : Pass2Scratch; |
| 456 | |
| 457 | SmallVector<const ExplodedNode*, 10> WL1, WL2; |
| 458 | |
| 459 | // ===- Pass 1 (reverse DFS) -=== |
| 460 | for (const auto Sink : Sinks) |
| 461 | if (Sink) |
| 462 | WL1.push_back(Elt: Sink); |
| 463 | |
| 464 | // Process the first worklist until it is empty. |
| 465 | while (!WL1.empty()) { |
| 466 | const ExplodedNode *N = WL1.pop_back_val(); |
| 467 | |
| 468 | // Have we already visited this node? If so, continue to the next one. |
| 469 | if (!Pass1.insert(V: N).second) |
| 470 | continue; |
| 471 | |
| 472 | // If this is the root enqueue it to the second worklist. |
| 473 | if (N->Preds.empty()) { |
| 474 | assert(N == getRoot() && "Found non-root node with no predecessors!" ); |
| 475 | WL2.push_back(Elt: N); |
| 476 | continue; |
| 477 | } |
| 478 | |
| 479 | // Visit our predecessors and enqueue them. |
| 480 | WL1.append(in_start: N->Preds.begin(), in_end: N->Preds.end()); |
| 481 | } |
| 482 | |
| 483 | // We didn't hit the root? Return with a null pointer for the new graph. |
| 484 | if (WL2.empty()) |
| 485 | return nullptr; |
| 486 | |
| 487 | assert(WL2.size() == 1 && "There must be only one root!" ); |
| 488 | |
| 489 | // Create an empty graph. |
| 490 | std::unique_ptr<ExplodedGraph> G = std::make_unique<ExplodedGraph>(); |
| 491 | |
| 492 | // ===- Pass 2 (forward DFS to construct the new graph) -=== |
| 493 | while (!WL2.empty()) { |
| 494 | const ExplodedNode *N = WL2.pop_back_val(); |
| 495 | |
| 496 | auto [Place, Inserted] = Pass2.try_emplace(Key: N); |
| 497 | |
| 498 | // Skip this node if we have already processed it. |
| 499 | if (!Inserted) |
| 500 | continue; |
| 501 | |
| 502 | // Create the corresponding node in the new graph and record the mapping |
| 503 | // from the old node to the new node. |
| 504 | ExplodedNode *NewN = G->createUncachedNode(L: N->getLocation(), State: N->State, |
| 505 | Id: N->getID(), IsSink: N->isSink()); |
| 506 | Place->second = NewN; |
| 507 | |
| 508 | // Also record the reverse mapping from the new node to the old node. |
| 509 | if (InverseMap) (*InverseMap)[NewN] = N; |
| 510 | |
| 511 | // If this node is the root, designate it as such in the graph. |
| 512 | if (N->Preds.empty()) { |
| 513 | assert(N == getRoot()); |
| 514 | G->designateAsRoot(V: NewN); |
| 515 | } |
| 516 | |
| 517 | // In the case that some of the intended predecessors of NewN have already |
| 518 | // been created, we should hook them up as predecessors. |
| 519 | |
| 520 | // Walk through the predecessors of 'N' and hook up their corresponding |
| 521 | // nodes in the new graph (if any) to the freshly created node. |
| 522 | for (const ExplodedNode *Pred : N->Preds) { |
| 523 | Pass2Ty::iterator PI = Pass2.find(Val: Pred); |
| 524 | if (PI == Pass2.end()) |
| 525 | continue; |
| 526 | |
| 527 | NewN->addPredecessor(V: const_cast<ExplodedNode *>(PI->second), G&: *G); |
| 528 | } |
| 529 | |
| 530 | // In the case that some of the intended successors of NewN have already |
| 531 | // been created, we should hook them up as successors. Otherwise, enqueue |
| 532 | // the new nodes from the original graph that should have nodes created |
| 533 | // in the new graph. |
| 534 | for (const ExplodedNode *Succ : N->Succs) { |
| 535 | Pass2Ty::iterator PI = Pass2.find(Val: Succ); |
| 536 | if (PI != Pass2.end()) { |
| 537 | const_cast<ExplodedNode *>(PI->second)->addPredecessor(V: NewN, G&: *G); |
| 538 | continue; |
| 539 | } |
| 540 | |
| 541 | // Enqueue nodes to the worklist that were marked during pass 1. |
| 542 | if (Pass1.count(V: Succ)) |
| 543 | WL2.push_back(Elt: Succ); |
| 544 | } |
| 545 | } |
| 546 | |
| 547 | return G; |
| 548 | } |
| 549 | |