| 1 | //===- EntityPointerLevel.cpp -----------------------------------*- 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 | #include "clang/ScalableStaticAnalysis/Analyses/EntityPointerLevel/EntityPointerLevel.h" |
| 10 | #include "SSAFAnalysesCommon.h" |
| 11 | #include "clang/AST/ASTContext.h" |
| 12 | #include "clang/AST/Decl.h" |
| 13 | #include "clang/AST/Expr.h" |
| 14 | #include "clang/AST/ExprCXX.h" |
| 15 | #include "clang/AST/StmtVisitor.h" |
| 16 | #include "clang/ScalableStaticAnalysis/Core/TUSummary/TUSummaryExtractor.h" |
| 17 | #include "llvm/ADT/STLExtras.h" |
| 18 | #include <optional> |
| 19 | |
| 20 | using namespace clang; |
| 21 | using namespace ssaf; |
| 22 | |
| 23 | namespace clang::ssaf { |
| 24 | // Translate a pointer type expression 'E' to a (set of) EntityPointerLevel(s) |
| 25 | // associated with the declared type of the base address of `E`. If the base |
| 26 | // address of `E` is not associated with an entity, the translation result is an |
| 27 | // empty set. |
| 28 | // |
| 29 | // The translation is a process of traversing into the pointer 'E' until its |
| 30 | // base address can be represented by an entity, with the number of dereferences |
| 31 | // tracked by incrementing the pointer level. Naturally, taking address of, as |
| 32 | // the inverse operation of dereference, is tracked by decrementing the pointer |
| 33 | // level. |
| 34 | // |
| 35 | // For example, suppose there are pointers and arrays declared as |
| 36 | // int *ptr, **p1, **p2; |
| 37 | // int arr[10][10]; |
| 38 | // , the translation of expressions involving these base addresses will be: |
| 39 | // Translate(ptr + 5) -> {(ptr, 1)} |
| 40 | // Translate(arr[5]) -> {(arr, 2)} |
| 41 | // Translate(cond ? p1[5] : p2) -> {(p1, 2), (p2, 1)} |
| 42 | // Translate(&arr[5]) -> {(arr, 1)} |
| 43 | class EntityPointerLevelTranslator |
| 44 | : ConstStmtVisitor<EntityPointerLevelTranslator, |
| 45 | Expected<DeclPointerLevelVec>> { |
| 46 | friend class StmtVisitorBase; |
| 47 | |
| 48 | // Fallback method for all unsupported expression kind: |
| 49 | Expected<DeclPointerLevelVec> fallback(const Stmt *S) { |
| 50 | // Report an error/warning (at least in debug mode) for any unsupported kind |
| 51 | // of pointer/array typed expression, because we want to understand every |
| 52 | // pointer/array expression. But for non-pointer/array typed expressions, we |
| 53 | // could silently ignore unsupported kinds. This translator visits |
| 54 | // non-pointer/array typed expressions because of address-of expressions. |
| 55 | if (const Expr *E = dyn_cast<Expr>(Val: S); E && hasPtrOrArrType(E)) |
| 56 | return makeErrAtNode(Ctx, N: E, |
| 57 | Fmt: "attempt to translate %s to EntityPointerLevels" , |
| 58 | Args: E->getStmtClassName()); |
| 59 | return DeclPointerLevelVec{}; |
| 60 | } |
| 61 | |
| 62 | Expected<EntityPointerLevel> |
| 63 | createEntityPointerLevelFor(const NamedDecl *ND) { |
| 64 | std::optional<EntityId> Id = Extractor.addEntity(D: ND); |
| 65 | if (!Id) |
| 66 | return makeErrAtNode(Ctx, N: ND, Fmt: "failed to create EntityId for %s" , |
| 67 | Args: ND->getDeclKindName()); |
| 68 | return EntityPointerLevel{buildEntityPointerLevel(*Id, 1)}; |
| 69 | } |
| 70 | |
| 71 | Expected<EntityPointerLevel> |
| 72 | createEntityPointerLevelForReturn(const FunctionDecl *FD) { |
| 73 | std::optional<EntityId> Id = Extractor.addEntityForReturn(FD); |
| 74 | if (!Id) { |
| 75 | return makeErrAtNode(Ctx, N: FD, Fmt: "failed to create EntityId for function %s" , |
| 76 | Args: cast<NamedDecl>(Val: FD)->getNameAsString().c_str()); |
| 77 | } |
| 78 | return EntityPointerLevel{buildEntityPointerLevel(*Id, 1)}; |
| 79 | } |
| 80 | |
| 81 | // The common helper function for Translate(*base): |
| 82 | // Translate(*base) -> Translate(base) with .pointerLevel + 1 |
| 83 | Expected<DeclPointerLevelVec> translateDereferencePointer(const Expr *Ptr) { |
| 84 | assert(hasPtrOrArrType(Ptr)); |
| 85 | |
| 86 | Expected<DeclPointerLevelVec> SubResult = Visit(S: Ptr); |
| 87 | if (!SubResult) |
| 88 | return SubResult.takeError(); |
| 89 | |
| 90 | llvm::for_each(Range&: *SubResult, F: [](DeclPointerLevel &D) { ++D.PointerLevel; }); |
| 91 | return SubResult; |
| 92 | } |
| 93 | |
| 94 | TUSummaryExtractor &; |
| 95 | ASTContext &Ctx; |
| 96 | |
| 97 | public: |
| 98 | (TUSummaryExtractor &, ASTContext &Ctx) |
| 99 | : Extractor(Extractor), Ctx(Ctx) {} |
| 100 | |
| 101 | Expected<DeclPointerLevelVec> translate(const Expr *E) { return Visit(S: E); } |
| 102 | Expected<EntityPointerLevel> translate(const NamedDecl *D, bool IsRet) { |
| 103 | if (!IsRet) |
| 104 | return createEntityPointerLevelFor(ND: D); |
| 105 | |
| 106 | if (const auto *FD = dyn_cast<FunctionDecl>(Val: D)) |
| 107 | return createEntityPointerLevelForReturn(FD); |
| 108 | |
| 109 | return makeErrAtNode(Ctx, N: D, Fmt: "attempt to get entity for return of %s" , |
| 110 | Args: D->getDeclKindName()); |
| 111 | } |
| 112 | |
| 113 | // Converts a `DeclPointerLevel` to an `EntityPointerLevel` |
| 114 | Expected<EntityPointerLevel> toEntityPointerLevel(const DeclPointerLevel &D) { |
| 115 | Expected<EntityPointerLevel> Base = translate(D: D.Decl, IsRet: D.IsReturn); |
| 116 | if (!Base) |
| 117 | return Base.takeError(); |
| 118 | return buildEntityPointerLevel(Base->getEntity(), D.PointerLevel); |
| 119 | } |
| 120 | |
| 121 | private: |
| 122 | Expected<DeclPointerLevelVec> VisitStmt(const Stmt *E) { return fallback(S: E); } |
| 123 | |
| 124 | // Translate(base + x) -> Translate(base) |
| 125 | // Translate(x + base) -> Translate(base) |
| 126 | // Translate(base - x) -> Translate(base) |
| 127 | // Translate(base {+=, -=, =} x) -> Translate(base) |
| 128 | // Translate(x, base) -> Translate(base) |
| 129 | Expected<DeclPointerLevelVec> VisitBinaryOperator(const BinaryOperator *E) { |
| 130 | switch (E->getOpcode()) { |
| 131 | case clang::BO_Add: |
| 132 | if (hasPtrOrArrType(E: E->getLHS())) |
| 133 | return Visit(S: E->getLHS()); |
| 134 | return Visit(S: E->getRHS()); |
| 135 | case clang::BO_Sub: |
| 136 | case clang::BO_AddAssign: |
| 137 | case clang::BO_SubAssign: |
| 138 | case clang::BO_Assign: |
| 139 | return Visit(S: E->getLHS()); |
| 140 | case clang::BO_Comma: |
| 141 | return Visit(S: E->getRHS()); |
| 142 | default: |
| 143 | return fallback(S: E); |
| 144 | } |
| 145 | } |
| 146 | |
| 147 | // Translate({++, --}base) -> Translate(base) |
| 148 | // Translate(base{++, --}) -> Translate(base) |
| 149 | // Translate(*base) -> Translate(base) with .pointerLevel += 1 |
| 150 | // Translate(&base) -> {}, if Translate(base) is {} |
| 151 | // -> Translate(base) with .pointerLevel -= 1 |
| 152 | // Translate(+base) -> Translate(base) |
| 153 | Expected<DeclPointerLevelVec> VisitUnaryOperator(const UnaryOperator *E) { |
| 154 | switch (E->getOpcode()) { |
| 155 | case clang::UO_PostInc: |
| 156 | case clang::UO_PostDec: |
| 157 | case clang::UO_PreInc: |
| 158 | case clang::UO_PreDec: |
| 159 | return Visit(S: E->getSubExpr()); |
| 160 | case clang::UO_AddrOf: { |
| 161 | Expected<DeclPointerLevelVec> SubResult = Visit(S: E->getSubExpr()); |
| 162 | if (!SubResult) |
| 163 | return SubResult.takeError(); |
| 164 | |
| 165 | llvm::for_each(Range&: *SubResult, F: [](DeclPointerLevel &D) { |
| 166 | assert(D.PointerLevel > 0); |
| 167 | --D.PointerLevel; |
| 168 | }); |
| 169 | return SubResult; |
| 170 | } |
| 171 | case clang::UO_Deref: |
| 172 | return translateDereferencePointer(Ptr: E->getSubExpr()); |
| 173 | case clang::UO_Plus: |
| 174 | return Visit(S: E->getSubExpr()); |
| 175 | default: |
| 176 | return fallback(S: E); |
| 177 | } |
| 178 | } |
| 179 | |
| 180 | // Translate((T*)base) -> Translate(base) if base has pointer type |
| 181 | // -> {} otherwise |
| 182 | Expected<DeclPointerLevelVec> VisitCastExpr(const CastExpr *E) { |
| 183 | if (hasPtrOrArrType(E: E->getSubExpr())) |
| 184 | return Visit(S: E->getSubExpr()); |
| 185 | return DeclPointerLevelVec{}; |
| 186 | } |
| 187 | |
| 188 | // Translate(f(...)) -> {} if it is an indirect call |
| 189 | // -> {(f_return, 1)}, otherwise |
| 190 | Expected<DeclPointerLevelVec> VisitCallExpr(const CallExpr *E) { |
| 191 | if (auto *FD = E->getDirectCallee()) |
| 192 | return DeclPointerLevelVec{{.Decl: FD, /*PointerLevel=*/1, /*IsReturn=*/true}}; |
| 193 | return DeclPointerLevelVec{}; |
| 194 | } |
| 195 | |
| 196 | // Translate(base[x]) -> Translate(*base) |
| 197 | Expected<DeclPointerLevelVec> |
| 198 | VisitArraySubscriptExpr(const ArraySubscriptExpr *E) { |
| 199 | return translateDereferencePointer(Ptr: E->getBase()); |
| 200 | } |
| 201 | |
| 202 | // Translate(cond ? base1 : base2) := Translate(base1) U Translate(base2) |
| 203 | Expected<DeclPointerLevelVec> |
| 204 | VisitAbstractConditionalOperator(const AbstractConditionalOperator *E) { |
| 205 | Expected<DeclPointerLevelVec> ReT = Visit(S: E->getTrueExpr()); |
| 206 | Expected<DeclPointerLevelVec> ReF = Visit(S: E->getFalseExpr()); |
| 207 | |
| 208 | if (ReT && ReF) { |
| 209 | ReT->insert(I: ReT->end(), From: ReF->begin(), To: ReF->end()); |
| 210 | return ReT; |
| 211 | } |
| 212 | if (!ReF && !ReT) |
| 213 | return llvm::joinErrors(E1: ReT.takeError(), E2: ReF.takeError()); |
| 214 | if (!ReF) |
| 215 | return ReF.takeError(); |
| 216 | return ReT.takeError(); |
| 217 | } |
| 218 | |
| 219 | Expected<DeclPointerLevelVec> VisitParenExpr(const ParenExpr *E) { |
| 220 | return Visit(S: E->getSubExpr()); |
| 221 | } |
| 222 | |
| 223 | // Translate("string-literal") -> {} // no entity involved |
| 224 | Expected<DeclPointerLevelVec> VisitStringLiteral(const StringLiteral *E) { |
| 225 | return DeclPointerLevelVec{}; |
| 226 | } |
| 227 | |
| 228 | // Translate(predefined-expr) -> {} // treated the same as string literals |
| 229 | Expected<DeclPointerLevelVec> VisitPredefinedExpr(const PredefinedExpr *E) { |
| 230 | return DeclPointerLevelVec{}; |
| 231 | } |
| 232 | |
| 233 | // Translate(integer-literal) -> {} // no entity involved |
| 234 | Expected<DeclPointerLevelVec> VisitIntegerLiteral(const IntegerLiteral *E) { |
| 235 | return DeclPointerLevelVec{}; |
| 236 | } |
| 237 | |
| 238 | // Translate(DRE) -> {(Decl, 1)} |
| 239 | Expected<DeclPointerLevelVec> VisitDeclRefExpr(const DeclRefExpr *E) { |
| 240 | return DeclPointerLevelVec{ |
| 241 | {.Decl: E->getDecl(), /*PointerLevel=*/1, /*IsReturn=*/false}}; |
| 242 | } |
| 243 | |
| 244 | // Translate({., ->}f) -> {(MemberDecl, 1)} |
| 245 | Expected<DeclPointerLevelVec> VisitMemberExpr(const MemberExpr *E) { |
| 246 | return DeclPointerLevelVec{ |
| 247 | {.Decl: E->getMemberDecl(), /*PointerLevel=*/1, /*IsReturn=*/false}}; |
| 248 | } |
| 249 | |
| 250 | // Unwrap CXXDefaultArgExpr |
| 251 | Expected<DeclPointerLevelVec> |
| 252 | VisitCXXDefaultArgExpr(const CXXDefaultArgExpr *E) { |
| 253 | return Visit(S: E->getExpr()); |
| 254 | } |
| 255 | |
| 256 | // Unwrap OpaqueValueExpr |
| 257 | Expected<DeclPointerLevelVec> VisitOpaqueValueExpr(const OpaqueValueExpr *S) { |
| 258 | return Visit(S: S->getSourceExpr()); |
| 259 | } |
| 260 | |
| 261 | // Unwrap ExprWithCleanups |
| 262 | Expected<DeclPointerLevelVec> |
| 263 | VisitExprWithCleanups(const ExprWithCleanups *S) { |
| 264 | return Visit(S: S->getSubExpr()); |
| 265 | } |
| 266 | |
| 267 | // Unwrap MaterializeTemporaryExpr |
| 268 | Expected<DeclPointerLevelVec> |
| 269 | VisitMaterializeTemporaryExpr(const MaterializeTemporaryExpr *S) { |
| 270 | return Visit(S: S->getSubExpr()); |
| 271 | } |
| 272 | |
| 273 | // Unwrap CXXDefaultInitExpr |
| 274 | Expected<DeclPointerLevelVec> |
| 275 | VisitCXXDefaultInitExpr(const CXXDefaultInitExpr *E) { |
| 276 | return Visit(S: E->getExpr()); |
| 277 | } |
| 278 | |
| 279 | // Translate(`nullptr`) -> {} |
| 280 | Expected<DeclPointerLevelVec> |
| 281 | VisitCXXNullPtrLiteralExpr(const CXXNullPtrLiteralExpr *S) { |
| 282 | return DeclPointerLevelVec{}; |
| 283 | } |
| 284 | |
| 285 | // Translate(`this`) -> {} |
| 286 | Expected<DeclPointerLevelVec> VisitCXXThisExpr(const CXXThisExpr *S) { |
| 287 | return DeclPointerLevelVec{}; |
| 288 | } |
| 289 | |
| 290 | // Translate(`new`/`new [*]`) -> {} |
| 291 | Expected<DeclPointerLevelVec> VisitCXXNewExpr(const CXXNewExpr *S) { |
| 292 | return DeclPointerLevelVec{}; |
| 293 | } |
| 294 | |
| 295 | // ImplicitValueInitExpr, for raw pointer type, |
| 296 | // evaluates to a compile-time constant zero (or null). So no EPL in the |
| 297 | // result. |
| 298 | Expected<DeclPointerLevelVec> |
| 299 | VisitImplicitValueInitExpr(const ImplicitValueInitExpr *S) { |
| 300 | return DeclPointerLevelVec{}; |
| 301 | } |
| 302 | |
| 303 | // The InitListExpr must be an empty or singleton list that |
| 304 | // initializes a pointer scalar. Other cases are unexpected thus an error. |
| 305 | Expected<DeclPointerLevelVec> VisitInitListExpr(const InitListExpr *E) { |
| 306 | if (E->getNumInits() < 1) |
| 307 | return DeclPointerLevelVec{}; |
| 308 | if (E->getType()->isPointerType()) |
| 309 | return Visit(S: E->getInit(Init: 0)); |
| 310 | return llvm::createStringError( |
| 311 | Fmt: "Cannot translate an InitListExpr to EntityPointerLevels if it is not " |
| 312 | "an empty or singleton list that initializes a pointer scalar" ); |
| 313 | } |
| 314 | |
| 315 | // Clang may default initializes an array with a CXXConstructExpr. Fallback on |
| 316 | // other cases, if they exist. |
| 317 | // When a CXXConstructExpr has an array type, clang is initializing an array |
| 318 | // of class-type objects with default values. In this case, no entity is |
| 319 | // associated with the initializer. |
| 320 | Expected<DeclPointerLevelVec> |
| 321 | VisitCXXConstructExpr(const CXXConstructExpr *E) { |
| 322 | if (E->getType()->isArrayType()) { |
| 323 | return DeclPointerLevelVec{}; |
| 324 | } |
| 325 | return fallback(S: E); |
| 326 | } |
| 327 | |
| 328 | // No entity is associated with a CXXScalarValueInitExpr: |
| 329 | Expected<DeclPointerLevelVec> |
| 330 | VisitCXXScalarValueInitExpr(const CXXScalarValueInitExpr *E) { |
| 331 | return DeclPointerLevelVec{}; |
| 332 | } |
| 333 | }; |
| 334 | } // namespace clang::ssaf |
| 335 | |
| 336 | Expected<DeclPointerLevelVec> |
| 337 | clang::ssaf::(const Expr *E, ASTContext &Ctx, |
| 338 | TUSummaryExtractor &) { |
| 339 | EntityPointerLevelTranslator Translator(Extractor, Ctx); |
| 340 | |
| 341 | return Translator.translate(E); |
| 342 | } |
| 343 | |
| 344 | Expected<EntityPointerLevelSet> |
| 345 | clang::ssaf::(const Expr *E, ASTContext &Ctx, |
| 346 | TUSummaryExtractor &) { |
| 347 | EntityPointerLevelTranslator Translator(Extractor, Ctx); |
| 348 | auto DPLs = Translator.translate(E); |
| 349 | if (!DPLs) |
| 350 | return DPLs.takeError(); |
| 351 | return toEntityPointerLevels(DPLs: *DPLs, Ctx, Extractor); |
| 352 | } |
| 353 | |
| 354 | DeclPointerLevel clang::ssaf::createDeclPointerLevel(const NamedDecl *ND, |
| 355 | bool IsFunRet) { |
| 356 | return {.Decl: ND, .PointerLevel: 1, .IsReturn: IsFunRet}; |
| 357 | } |
| 358 | |
| 359 | /// Create an EntityPointerLevel from a ValueDecl of a pointer type. |
| 360 | Expected<EntityPointerLevel> clang::ssaf::( |
| 361 | const NamedDecl *ND, TUSummaryExtractor &, bool IsFunRet) { |
| 362 | EntityPointerLevelTranslator Translator(Extractor, ND->getASTContext()); |
| 363 | |
| 364 | return Translator.translate(D: ND, IsRet: IsFunRet); |
| 365 | } |
| 366 | |
| 367 | DeclPointerLevelVec |
| 368 | clang::ssaf::elaborateHigherDeclPointerLevels(const DeclPointerLevel &DPL) { |
| 369 | DeclPointerLevelVec Result{DPL}; |
| 370 | QualType T; |
| 371 | |
| 372 | if (DPL.IsReturn) { |
| 373 | if (const auto *FD = dyn_cast<FunctionDecl>(Val: DPL.Decl)) |
| 374 | T = FD->getReturnType().getNonReferenceType(); |
| 375 | } else if (const auto *VD = dyn_cast<ValueDecl>(Val: DPL.Decl)) { |
| 376 | T = VD->getType().getNonReferenceType(); |
| 377 | } |
| 378 | if (T.isNull()) |
| 379 | return Result; |
| 380 | |
| 381 | // Count the max pointer/array levels of `T`: |
| 382 | unsigned MaxLevel = 0; |
| 383 | |
| 384 | T = T.getCanonicalType(); |
| 385 | while (!T.isNull() && (T->isPointerType() || T->isArrayType())) { |
| 386 | if (const auto *PT = dyn_cast<PointerType>(Val&: T)) |
| 387 | T = PT->getPointeeType(); |
| 388 | else |
| 389 | T = cast<ArrayType>(Val&: T)->getElementType(); |
| 390 | ++MaxLevel; |
| 391 | } |
| 392 | assert(MaxLevel > 0); |
| 393 | Result.reserve(N: MaxLevel); |
| 394 | for (unsigned Level = DPL.PointerLevel + 1; Level <= MaxLevel; ++Level) |
| 395 | Result.push_back(Elt: {.Decl: DPL.Decl, .PointerLevel: Level, .IsReturn: DPL.IsReturn}); |
| 396 | return Result; |
| 397 | } |
| 398 | |
| 399 | Expected<EntityPointerLevelSet> |
| 400 | clang::ssaf::(const DeclPointerLevelVec &DPLs, |
| 401 | ASTContext &Ctx, |
| 402 | TUSummaryExtractor &) { |
| 403 | EntityPointerLevelTranslator Translator(Extractor, Ctx); |
| 404 | EntityPointerLevelSet Result; |
| 405 | |
| 406 | for (const auto &DPL : DPLs) { |
| 407 | Expected<EntityPointerLevel> EPL = Translator.toEntityPointerLevel(D: DPL); |
| 408 | if (!EPL) |
| 409 | return EPL.takeError(); |
| 410 | Result.insert(x: *EPL); |
| 411 | } |
| 412 | return Result; |
| 413 | } |
| 414 | |
| 415 | Expected<EntityPointerLevel> |
| 416 | clang::ssaf::(const DeclPointerLevel &DPL, ASTContext &Ctx, |
| 417 | TUSummaryExtractor &) { |
| 418 | EntityPointerLevelTranslator Translator(Extractor, Ctx); |
| 419 | return Translator.toEntityPointerLevel(D: DPL); |
| 420 | } |
| 421 | |
| 422 | EntityPointerLevel clang::ssaf::buildEntityPointerLevel(EntityId Id, |
| 423 | unsigned PtrLv) { |
| 424 | return EntityPointerLevel({Id, PtrLv}); |
| 425 | } |
| 426 | |