[clang] [clang-tools-extra] [clang-tidy][dataflow] Add `bugprone-null-check-after-dereference` check (PR #84166)
via cfe-commits
cfe-commits at lists.llvm.org
Wed Mar 20 08:55:26 PDT 2024
================
@@ -0,0 +1,625 @@
+//===-- NullPointerAnalysisModel.cpp ----------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines a generic null-pointer analysis model, used for finding
+// pointer null-checks after the pointer has already been dereferenced.
+//
+// Only a limited set of operations are currently recognized. Notably, pointer
+// arithmetic, null-pointer assignments and _nullable/_nonnull attributes are
+// missing as of yet.
+//
+//===----------------------------------------------------------------------===//
+
+#include "clang/Analysis/FlowSensitive/Models/NullPointerAnalysisModel.h"
+#include "clang/AST/ASTContext.h"
+#include "clang/AST/Decl.h"
+#include "clang/AST/Expr.h"
+#include "clang/AST/Stmt.h"
+#include "clang/ASTMatchers/ASTMatchFinder.h"
+#include "clang/ASTMatchers/ASTMatchers.h"
+#include "clang/Analysis/CFG.h"
+#include "clang/Analysis/FlowSensitive/CFGMatchSwitch.h"
+#include "clang/Analysis/FlowSensitive/DataflowAnalysis.h"
+#include "clang/Analysis/FlowSensitive/DataflowEnvironment.h"
+#include "clang/Analysis/FlowSensitive/DataflowLattice.h"
+#include "clang/Analysis/FlowSensitive/MapLattice.h"
+#include "clang/Analysis/FlowSensitive/NoopLattice.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/ADT/Twine.h"
+
+namespace clang::dataflow {
+
+namespace {
+using namespace ast_matchers;
+
+constexpr char kCond[] = "condition";
+constexpr char kVar[] = "var";
+constexpr char kValue[] = "value";
+constexpr char kIsNonnull[] = "is-nonnull";
+constexpr char kIsNull[] = "is-null";
+
+enum class SatisfiabilityResult {
+ // Returned when the value was not initialized yet.
+ Nullptr,
+ // Special value that signals that the boolean value can be anything.
+ // It signals that the underlying formulas are too complex to be calculated
+ // efficiently.
+ Top,
+ // Equivalent to the literal True in the current environment.
+ True,
+ // Equivalent to the literal False in the current environment.
+ False,
+ // Both True and False values could be produced with an appropriate set of
+ // conditions.
+ Unknown
+};
+
+using SR = SatisfiabilityResult;
+
+// FIXME: These AST matchers should also be exported via the
+// NullPointerAnalysisModel class, for tests
+auto ptrToVar(llvm::StringRef VarName = kVar) {
+ return traverse(TK_IgnoreUnlessSpelledInSource,
----------------
Discookie wrote:
Apparently `TK_IgnoreUnlessSpelledInSource` is still required here. While getting `Value`s out of the framework is propagated correctly, *setting* values is not.
When setting a new `Value`, only the outermost `Expr` will have that value set, which is not guaranteed to be a glvalue even if the underlying value is, and could be overwritten by the propagation done by `Transfer.cpp`. To set a value directly, we still need to bind to the root `Expr` in this case.
This traversal has weird behavior with `nullptr` and some other casts - on `nullptr` we need to match the `NullToPointer` cast, and not the nullptr directly, and explicit casts also have the chance to not bind correctly to the root pointer.
I'm not sure if there's a better way to bind to the "root" `Expr` that I haven't found yet, but the current solution has too many dragons. (Maybe this is also a case of X/Y problem, where if I assign atoms as `IsNull/IsNonnull` instead of formulas, I wouldn't need to re-assign `Value`s...)
https://github.com/llvm/llvm-project/pull/84166
More information about the cfe-commits
mailing list