1 //===- SymbolManager.h - Management of Symbolic Values --------------------===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 //  This file defines SymbolManager, a class that manages symbolic values
11 //  created for use by ExprEngine and related classes.
12 //
13 //===----------------------------------------------------------------------===//
14 
15 #include "clang/StaticAnalyzer/Core/PathSensitive/SymbolManager.h"
16 #include "clang/AST/ASTContext.h"
17 #include "clang/AST/Expr.h"
18 #include "clang/Analysis/Analyses/LiveVariables.h"
19 #include "clang/Analysis/AnalysisDeclContext.h"
20 #include "clang/Basic/LLVM.h"
21 #include "clang/StaticAnalyzer/Core/PathSensitive/MemRegion.h"
22 #include "clang/StaticAnalyzer/Core/PathSensitive/SVals.h"
23 #include "clang/StaticAnalyzer/Core/PathSensitive/Store.h"
24 #include "clang/StaticAnalyzer/Core/PathSensitive/SymExpr.h"
25 #include "llvm/ADT/FoldingSet.h"
26 #include "llvm/ADT/STLExtras.h"
27 #include "llvm/Support/Casting.h"
28 #include "llvm/Support/Compiler.h"
29 #include "llvm/Support/ErrorHandling.h"
30 #include "llvm/Support/raw_ostream.h"
31 #include <cassert>
32 
33 using namespace clang;
34 using namespace ento;
35 
36 void SymExpr::anchor() {}
37 
38 LLVM_DUMP_METHOD void SymExpr::dump() const {
39   dumpToStream(llvm::errs());
40 }
41 
42 void SymIntExpr::dumpToStream(raw_ostream &os) const {
43   os << '(';
44   getLHS()->dumpToStream(os);
45   os << ") "
46      << BinaryOperator::getOpcodeStr(getOpcode()) << ' ';
47   if (getRHS().isUnsigned())
48     os << getRHS().getZExtValue();
49   else
50     os << getRHS().getSExtValue();
51   if (getRHS().isUnsigned())
52     os << 'U';
53 }
54 
55 void IntSymExpr::dumpToStream(raw_ostream &os) const {
56   if (getLHS().isUnsigned())
57     os << getLHS().getZExtValue();
58   else
59     os << getLHS().getSExtValue();
60   if (getLHS().isUnsigned())
61     os << 'U';
62   os << ' '
63      << BinaryOperator::getOpcodeStr(getOpcode())
64      << " (";
65   getRHS()->dumpToStream(os);
66   os << ')';
67 }
68 
69 void SymSymExpr::dumpToStream(raw_ostream &os) const {
70   os << '(';
71   getLHS()->dumpToStream(os);
72   os << ") "
73      << BinaryOperator::getOpcodeStr(getOpcode())
74      << " (";
75   getRHS()->dumpToStream(os);
76   os << ')';
77 }
78 
79 void SymbolCast::dumpToStream(raw_ostream &os) const {
80   os << '(' << ToTy.getAsString() << ") (";
81   Operand->dumpToStream(os);
82   os << ')';
83 }
84 
85 void SymbolConjured::dumpToStream(raw_ostream &os) const {
86   os << "conj_$" << getSymbolID() << '{' << T.getAsString() << '}';
87 }
88 
89 void SymbolDerived::dumpToStream(raw_ostream &os) const {
90   os << "derived_$" << getSymbolID() << '{'
91      << getParentSymbol() << ',' << getRegion() << '}';
92 }
93 
94 void SymbolExtent::dumpToStream(raw_ostream &os) const {
95   os << "extent_$" << getSymbolID() << '{' << getRegion() << '}';
96 }
97 
98 void SymbolMetadata::dumpToStream(raw_ostream &os) const {
99   os << "meta_$" << getSymbolID() << '{'
100      << getRegion() << ',' << T.getAsString() << '}';
101 }
102 
103 void SymbolData::anchor() {}
104 
105 void SymbolRegionValue::dumpToStream(raw_ostream &os) const {
106   os << "reg_$" << getSymbolID()
107      << '<' << getType().getAsString() << ' ' << R << '>';
108 }
109 
110 bool SymExpr::symbol_iterator::operator==(const symbol_iterator &X) const {
111   return itr == X.itr;
112 }
113 
114 bool SymExpr::symbol_iterator::operator!=(const symbol_iterator &X) const {
115   return itr != X.itr;
116 }
117 
118 SymExpr::symbol_iterator::symbol_iterator(const SymExpr *SE) {
119   itr.push_back(SE);
120 }
121 
122 SymExpr::symbol_iterator &SymExpr::symbol_iterator::operator++() {
123   assert(!itr.empty() && "attempting to iterate on an 'end' iterator");
124   expand();
125   return *this;
126 }
127 
128 SymbolRef SymExpr::symbol_iterator::operator*() {
129   assert(!itr.empty() && "attempting to dereference an 'end' iterator");
130   return itr.back();
131 }
132 
133 void SymExpr::symbol_iterator::expand() {
134   const SymExpr *SE = itr.pop_back_val();
135 
136   switch (SE->getKind()) {
137     case SymExpr::SymbolRegionValueKind:
138     case SymExpr::SymbolConjuredKind:
139     case SymExpr::SymbolDerivedKind:
140     case SymExpr::SymbolExtentKind:
141     case SymExpr::SymbolMetadataKind:
142       return;
143     case SymExpr::SymbolCastKind:
144       itr.push_back(cast<SymbolCast>(SE)->getOperand());
145       return;
146     case SymExpr::SymIntExprKind:
147       itr.push_back(cast<SymIntExpr>(SE)->getLHS());
148       return;
149     case SymExpr::IntSymExprKind:
150       itr.push_back(cast<IntSymExpr>(SE)->getRHS());
151       return;
152     case SymExpr::SymSymExprKind: {
153       const auto *x = cast<SymSymExpr>(SE);
154       itr.push_back(x->getLHS());
155       itr.push_back(x->getRHS());
156       return;
157     }
158   }
159   llvm_unreachable("unhandled expansion case");
160 }
161 
162 const SymbolRegionValue*
163 SymbolManager::getRegionValueSymbol(const TypedValueRegion* R) {
164   llvm::FoldingSetNodeID profile;
165   SymbolRegionValue::Profile(profile, R);
166   void *InsertPos;
167   SymExpr *SD = DataSet.FindNodeOrInsertPos(profile, InsertPos);
168   if (!SD) {
169     SD = (SymExpr*) BPAlloc.Allocate<SymbolRegionValue>();
170     new (SD) SymbolRegionValue(SymbolCounter, R);
171     DataSet.InsertNode(SD, InsertPos);
172     ++SymbolCounter;
173   }
174 
175   return cast<SymbolRegionValue>(SD);
176 }
177 
178 const SymbolConjured* SymbolManager::conjureSymbol(const Stmt *E,
179                                                    const LocationContext *LCtx,
180                                                    QualType T,
181                                                    unsigned Count,
182                                                    const void *SymbolTag) {
183   llvm::FoldingSetNodeID profile;
184   SymbolConjured::Profile(profile, E, T, Count, LCtx, SymbolTag);
185   void *InsertPos;
186   SymExpr *SD = DataSet.FindNodeOrInsertPos(profile, InsertPos);
187   if (!SD) {
188     SD = (SymExpr*) BPAlloc.Allocate<SymbolConjured>();
189     new (SD) SymbolConjured(SymbolCounter, E, LCtx, T, Count, SymbolTag);
190     DataSet.InsertNode(SD, InsertPos);
191     ++SymbolCounter;
192   }
193 
194   return cast<SymbolConjured>(SD);
195 }
196 
197 const SymbolDerived*
198 SymbolManager::getDerivedSymbol(SymbolRef parentSymbol,
199                                 const TypedValueRegion *R) {
200   llvm::FoldingSetNodeID profile;
201   SymbolDerived::Profile(profile, parentSymbol, R);
202   void *InsertPos;
203   SymExpr *SD = DataSet.FindNodeOrInsertPos(profile, InsertPos);
204   if (!SD) {
205     SD = (SymExpr*) BPAlloc.Allocate<SymbolDerived>();
206     new (SD) SymbolDerived(SymbolCounter, parentSymbol, R);
207     DataSet.InsertNode(SD, InsertPos);
208     ++SymbolCounter;
209   }
210 
211   return cast<SymbolDerived>(SD);
212 }
213 
214 const SymbolExtent*
215 SymbolManager::getExtentSymbol(const SubRegion *R) {
216   llvm::FoldingSetNodeID profile;
217   SymbolExtent::Profile(profile, R);
218   void *InsertPos;
219   SymExpr *SD = DataSet.FindNodeOrInsertPos(profile, InsertPos);
220   if (!SD) {
221     SD = (SymExpr*) BPAlloc.Allocate<SymbolExtent>();
222     new (SD) SymbolExtent(SymbolCounter, R);
223     DataSet.InsertNode(SD, InsertPos);
224     ++SymbolCounter;
225   }
226 
227   return cast<SymbolExtent>(SD);
228 }
229 
230 const SymbolMetadata *
231 SymbolManager::getMetadataSymbol(const MemRegion* R, const Stmt *S, QualType T,
232                                  const LocationContext *LCtx,
233                                  unsigned Count, const void *SymbolTag) {
234   llvm::FoldingSetNodeID profile;
235   SymbolMetadata::Profile(profile, R, S, T, LCtx, Count, SymbolTag);
236   void *InsertPos;
237   SymExpr *SD = DataSet.FindNodeOrInsertPos(profile, InsertPos);
238   if (!SD) {
239     SD = (SymExpr*) BPAlloc.Allocate<SymbolMetadata>();
240     new (SD) SymbolMetadata(SymbolCounter, R, S, T, LCtx, Count, SymbolTag);
241     DataSet.InsertNode(SD, InsertPos);
242     ++SymbolCounter;
243   }
244 
245   return cast<SymbolMetadata>(SD);
246 }
247 
248 const SymbolCast*
249 SymbolManager::getCastSymbol(const SymExpr *Op,
250                              QualType From, QualType To) {
251   llvm::FoldingSetNodeID ID;
252   SymbolCast::Profile(ID, Op, From, To);
253   void *InsertPos;
254   SymExpr *data = DataSet.FindNodeOrInsertPos(ID, InsertPos);
255   if (!data) {
256     data = (SymbolCast*) BPAlloc.Allocate<SymbolCast>();
257     new (data) SymbolCast(Op, From, To);
258     DataSet.InsertNode(data, InsertPos);
259   }
260 
261   return cast<SymbolCast>(data);
262 }
263 
264 const SymIntExpr *SymbolManager::getSymIntExpr(const SymExpr *lhs,
265                                                BinaryOperator::Opcode op,
266                                                const llvm::APSInt& v,
267                                                QualType t) {
268   llvm::FoldingSetNodeID ID;
269   SymIntExpr::Profile(ID, lhs, op, v, t);
270   void *InsertPos;
271   SymExpr *data = DataSet.FindNodeOrInsertPos(ID, InsertPos);
272 
273   if (!data) {
274     data = (SymIntExpr*) BPAlloc.Allocate<SymIntExpr>();
275     new (data) SymIntExpr(lhs, op, v, t);
276     DataSet.InsertNode(data, InsertPos);
277   }
278 
279   return cast<SymIntExpr>(data);
280 }
281 
282 const IntSymExpr *SymbolManager::getIntSymExpr(const llvm::APSInt& lhs,
283                                                BinaryOperator::Opcode op,
284                                                const SymExpr *rhs,
285                                                QualType t) {
286   llvm::FoldingSetNodeID ID;
287   IntSymExpr::Profile(ID, lhs, op, rhs, t);
288   void *InsertPos;
289   SymExpr *data = DataSet.FindNodeOrInsertPos(ID, InsertPos);
290 
291   if (!data) {
292     data = (IntSymExpr*) BPAlloc.Allocate<IntSymExpr>();
293     new (data) IntSymExpr(lhs, op, rhs, t);
294     DataSet.InsertNode(data, InsertPos);
295   }
296 
297   return cast<IntSymExpr>(data);
298 }
299 
300 const SymSymExpr *SymbolManager::getSymSymExpr(const SymExpr *lhs,
301                                                BinaryOperator::Opcode op,
302                                                const SymExpr *rhs,
303                                                QualType t) {
304   llvm::FoldingSetNodeID ID;
305   SymSymExpr::Profile(ID, lhs, op, rhs, t);
306   void *InsertPos;
307   SymExpr *data = DataSet.FindNodeOrInsertPos(ID, InsertPos);
308 
309   if (!data) {
310     data = (SymSymExpr*) BPAlloc.Allocate<SymSymExpr>();
311     new (data) SymSymExpr(lhs, op, rhs, t);
312     DataSet.InsertNode(data, InsertPos);
313   }
314 
315   return cast<SymSymExpr>(data);
316 }
317 
318 QualType SymbolConjured::getType() const {
319   return T;
320 }
321 
322 QualType SymbolDerived::getType() const {
323   return R->getValueType();
324 }
325 
326 QualType SymbolExtent::getType() const {
327   ASTContext &Ctx = R->getMemRegionManager()->getContext();
328   return Ctx.getSizeType();
329 }
330 
331 QualType SymbolMetadata::getType() const {
332   return T;
333 }
334 
335 QualType SymbolRegionValue::getType() const {
336   return R->getValueType();
337 }
338 
339 SymbolManager::~SymbolManager() {
340   llvm::DeleteContainerSeconds(SymbolDependencies);
341 }
342 
343 bool SymbolManager::canSymbolicate(QualType T) {
344   T = T.getCanonicalType();
345 
346   if (Loc::isLocType(T))
347     return true;
348 
349   if (T->isIntegralOrEnumerationType())
350     return true;
351 
352   if (T->isRecordType() && !T->isUnionType())
353     return true;
354 
355   return false;
356 }
357 
358 void SymbolManager::addSymbolDependency(const SymbolRef Primary,
359                                         const SymbolRef Dependent) {
360   SymbolDependTy::iterator I = SymbolDependencies.find(Primary);
361   SymbolRefSmallVectorTy *dependencies = nullptr;
362   if (I == SymbolDependencies.end()) {
363     dependencies = new SymbolRefSmallVectorTy();
364     SymbolDependencies[Primary] = dependencies;
365   } else {
366     dependencies = I->second;
367   }
368   dependencies->push_back(Dependent);
369 }
370 
371 const SymbolRefSmallVectorTy *SymbolManager::getDependentSymbols(
372                                                      const SymbolRef Primary) {
373   SymbolDependTy::const_iterator I = SymbolDependencies.find(Primary);
374   if (I == SymbolDependencies.end())
375     return nullptr;
376   return I->second;
377 }
378 
379 void SymbolReaper::markDependentsLive(SymbolRef sym) {
380   // Do not mark dependents more then once.
381   SymbolMapTy::iterator LI = TheLiving.find(sym);
382   assert(LI != TheLiving.end() && "The primary symbol is not live.");
383   if (LI->second == HaveMarkedDependents)
384     return;
385   LI->second = HaveMarkedDependents;
386 
387   if (const SymbolRefSmallVectorTy *Deps = SymMgr.getDependentSymbols(sym)) {
388     for (const auto I : *Deps) {
389       if (TheLiving.find(I) != TheLiving.end())
390         continue;
391       markLive(I);
392     }
393   }
394 }
395 
396 void SymbolReaper::markLive(SymbolRef sym) {
397   TheLiving[sym] = NotProcessed;
398   TheDead.erase(sym);
399   markDependentsLive(sym);
400 }
401 
402 void SymbolReaper::markLive(const MemRegion *region) {
403   RegionRoots.insert(region);
404   markElementIndicesLive(region);
405 }
406 
407 void SymbolReaper::markElementIndicesLive(const MemRegion *region) {
408   for (auto SR = dyn_cast<SubRegion>(region); SR;
409        SR = dyn_cast<SubRegion>(SR->getSuperRegion())) {
410     if (const auto ER = dyn_cast<ElementRegion>(SR)) {
411       SVal Idx = ER->getIndex();
412       for (auto SI = Idx.symbol_begin(), SE = Idx.symbol_end(); SI != SE; ++SI)
413         markLive(*SI);
414     }
415   }
416 }
417 
418 void SymbolReaper::markInUse(SymbolRef sym) {
419   if (isa<SymbolMetadata>(sym))
420     MetadataInUse.insert(sym);
421 }
422 
423 bool SymbolReaper::maybeDead(SymbolRef sym) {
424   if (isLive(sym))
425     return false;
426 
427   TheDead.insert(sym);
428   return true;
429 }
430 
431 bool SymbolReaper::isLiveRegion(const MemRegion *MR) {
432   if (RegionRoots.count(MR))
433     return true;
434 
435   MR = MR->getBaseRegion();
436 
437   if (const auto *SR = dyn_cast<SymbolicRegion>(MR))
438     return isLive(SR->getSymbol());
439 
440   if (const auto *VR = dyn_cast<VarRegion>(MR))
441     return isLive(VR, true);
442 
443   // FIXME: This is a gross over-approximation. What we really need is a way to
444   // tell if anything still refers to this region. Unlike SymbolicRegions,
445   // AllocaRegions don't have associated symbols, though, so we don't actually
446   // have a way to track their liveness.
447   if (isa<AllocaRegion>(MR))
448     return true;
449 
450   if (isa<CXXThisRegion>(MR))
451     return true;
452 
453   if (isa<MemSpaceRegion>(MR))
454     return true;
455 
456   if (isa<CodeTextRegion>(MR))
457     return true;
458 
459   return false;
460 }
461 
462 bool SymbolReaper::isLive(SymbolRef sym) {
463   if (TheLiving.count(sym)) {
464     markDependentsLive(sym);
465     return true;
466   }
467 
468   bool KnownLive;
469 
470   switch (sym->getKind()) {
471   case SymExpr::SymbolRegionValueKind:
472     KnownLive = isLiveRegion(cast<SymbolRegionValue>(sym)->getRegion());
473     break;
474   case SymExpr::SymbolConjuredKind:
475     KnownLive = false;
476     break;
477   case SymExpr::SymbolDerivedKind:
478     KnownLive = isLive(cast<SymbolDerived>(sym)->getParentSymbol());
479     break;
480   case SymExpr::SymbolExtentKind:
481     KnownLive = isLiveRegion(cast<SymbolExtent>(sym)->getRegion());
482     break;
483   case SymExpr::SymbolMetadataKind:
484     KnownLive = MetadataInUse.count(sym) &&
485                 isLiveRegion(cast<SymbolMetadata>(sym)->getRegion());
486     if (KnownLive)
487       MetadataInUse.erase(sym);
488     break;
489   case SymExpr::SymIntExprKind:
490     KnownLive = isLive(cast<SymIntExpr>(sym)->getLHS());
491     break;
492   case SymExpr::IntSymExprKind:
493     KnownLive = isLive(cast<IntSymExpr>(sym)->getRHS());
494     break;
495   case SymExpr::SymSymExprKind:
496     KnownLive = isLive(cast<SymSymExpr>(sym)->getLHS()) &&
497                 isLive(cast<SymSymExpr>(sym)->getRHS());
498     break;
499   case SymExpr::SymbolCastKind:
500     KnownLive = isLive(cast<SymbolCast>(sym)->getOperand());
501     break;
502   }
503 
504   if (KnownLive)
505     markLive(sym);
506 
507   return KnownLive;
508 }
509 
510 bool
511 SymbolReaper::isLive(const Stmt *ExprVal, const LocationContext *ELCtx) const {
512   if (LCtx == nullptr)
513     return false;
514 
515   if (LCtx != ELCtx) {
516     // If the reaper's location context is a parent of the expression's
517     // location context, then the expression value is now "out of scope".
518     if (LCtx->isParentOf(ELCtx))
519       return false;
520     return true;
521   }
522 
523   // If no statement is provided, everything is this and parent contexts is live.
524   if (!Loc)
525     return true;
526 
527   return LCtx->getAnalysis<RelaxedLiveVariables>()->isLive(Loc, ExprVal);
528 }
529 
530 bool SymbolReaper::isLive(const VarRegion *VR, bool includeStoreBindings) const{
531   const StackFrameContext *VarContext = VR->getStackFrame();
532 
533   if (!VarContext)
534     return true;
535 
536   if (!LCtx)
537     return false;
538   const StackFrameContext *CurrentContext = LCtx->getStackFrame();
539 
540   if (VarContext == CurrentContext) {
541     // If no statement is provided, everything is live.
542     if (!Loc)
543       return true;
544 
545     if (LCtx->getAnalysis<RelaxedLiveVariables>()->isLive(Loc, VR->getDecl()))
546       return true;
547 
548     if (!includeStoreBindings)
549       return false;
550 
551     unsigned &cachedQuery =
552       const_cast<SymbolReaper *>(this)->includedRegionCache[VR];
553 
554     if (cachedQuery) {
555       return cachedQuery == 1;
556     }
557 
558     // Query the store to see if the region occurs in any live bindings.
559     if (Store store = reapedStore.getStore()) {
560       bool hasRegion =
561         reapedStore.getStoreManager().includedInBindings(store, VR);
562       cachedQuery = hasRegion ? 1 : 2;
563       return hasRegion;
564     }
565 
566     return false;
567   }
568 
569   return VarContext->isParentOf(CurrentContext);
570 }
571