1 //===-- SpecialCaseList.h - special case list for sanitizers ----*- 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 // This is a utility class used to parse user-provided text files with
9 // "special case lists" for code sanitizers. Such files are used to
10 // define an "ABI list" for DataFlowSanitizer and blacklists for sanitizers
11 // like AddressSanitizer or UndefinedBehaviorSanitizer.
12 //
13 // Empty lines and lines starting with "#" are ignored. Sections are defined
14 // using a '[section_name]' header and can be used to specify sanitizers the
15 // entries below it apply to. Section names are regular expressions, and
16 // entries without a section header match all sections (e.g. an '[*]' header
17 // is assumed.)
18 // The remaining lines should have the form:
19 //   prefix:wildcard_expression[=category]
20 // If category is not specified, it is assumed to be empty string.
21 // Definitions of "prefix" and "category" are sanitizer-specific. For example,
22 // sanitizer blacklists support prefixes "src", "fun" and "global".
23 // Wildcard expressions define, respectively, source files, functions or
24 // globals which shouldn't be instrumented.
25 // Examples of categories:
26 //   "functional": used in DFSan to list functions with pure functional
27 //                 semantics.
28 //   "init": used in ASan blacklist to disable initialization-order bugs
29 //           detection for certain globals or source files.
30 // Full special case list file example:
31 // ---
32 // [address]
33 // # Blacklisted items:
34 // fun:*_ZN4base6subtle*
35 // global:*global_with_bad_access_or_initialization*
36 // global:*global_with_initialization_issues*=init
37 // type:*Namespace::ClassName*=init
38 // src:file_with_tricky_code.cc
39 // src:ignore-global-initializers-issues.cc=init
40 //
41 // [dataflow]
42 // # Functions with pure functional semantics:
43 // fun:cos=functional
44 // fun:sin=functional
45 // ---
46 // Note that the wild card is in fact an llvm::Regex, but * is automatically
47 // replaced with .*
48 //
49 //===----------------------------------------------------------------------===//
50 
51 #ifndef LLVM_SUPPORT_SPECIALCASELIST_H
52 #define LLVM_SUPPORT_SPECIALCASELIST_H
53 
54 #include "llvm/ADT/StringMap.h"
55 #include "llvm/ADT/StringSet.h"
56 #include "llvm/Support/Regex.h"
57 #include "llvm/Support/TrigramIndex.h"
58 #include "llvm/Support/VirtualFileSystem.h"
59 #include <string>
60 #include <vector>
61 
62 namespace llvm {
63 class MemoryBuffer;
64 class Regex;
65 class StringRef;
66 
67 class SpecialCaseList {
68 public:
69   /// Parses the special case list entries from files. On failure, returns
70   /// 0 and writes an error message to string.
71   static std::unique_ptr<SpecialCaseList>
72   create(const std::vector<std::string> &Paths, llvm::vfs::FileSystem &FS,
73          std::string &Error);
74   /// Parses the special case list from a memory buffer. On failure, returns
75   /// 0 and writes an error message to string.
76   static std::unique_ptr<SpecialCaseList> create(const MemoryBuffer *MB,
77                                                  std::string &Error);
78   /// Parses the special case list entries from files. On failure, reports a
79   /// fatal error.
80   static std::unique_ptr<SpecialCaseList>
81   createOrDie(const std::vector<std::string> &Paths, llvm::vfs::FileSystem &FS);
82 
83   ~SpecialCaseList();
84 
85   /// Returns true, if special case list contains a line
86   /// \code
87   ///   @Prefix:<E>=@Category
88   /// \endcode
89   /// where @Query satisfies wildcard expression <E> in a given @Section.
90   bool inSection(StringRef Section, StringRef Prefix, StringRef Query,
91                  StringRef Category = StringRef()) const;
92 
93   /// Returns the line number corresponding to the special case list entry if
94   /// the special case list contains a line
95   /// \code
96   ///   @Prefix:<E>=@Category
97   /// \endcode
98   /// where @Query satisfies wildcard expression <E> in a given @Section.
99   /// Returns zero if there is no blacklist entry corresponding to this
100   /// expression.
101   unsigned inSectionBlame(StringRef Section, StringRef Prefix, StringRef Query,
102                           StringRef Category = StringRef()) const;
103 
104 protected:
105   // Implementations of the create*() functions that can also be used by derived
106   // classes.
107   bool createInternal(const std::vector<std::string> &Paths,
108                       vfs::FileSystem &VFS, std::string &Error);
109   bool createInternal(const MemoryBuffer *MB, std::string &Error);
110 
111   SpecialCaseList() = default;
112   SpecialCaseList(SpecialCaseList const &) = delete;
113   SpecialCaseList &operator=(SpecialCaseList const &) = delete;
114 
115   /// Represents a set of regular expressions.  Regular expressions which are
116   /// "literal" (i.e. no regex metacharacters) are stored in Strings.  The
117   /// reason for doing so is efficiency; StringMap is much faster at matching
118   /// literal strings than Regex.
119   class Matcher {
120   public:
121     bool insert(std::string Regexp, unsigned LineNumber, std::string &REError);
122     // Returns the line number in the source file that this query matches to.
123     // Returns zero if no match is found.
124     unsigned match(StringRef Query) const;
125 
126   private:
127     StringMap<unsigned> Strings;
128     TrigramIndex Trigrams;
129     std::vector<std::pair<std::unique_ptr<Regex>, unsigned>> RegExes;
130   };
131 
132   using SectionEntries = StringMap<StringMap<Matcher>>;
133 
134   struct Section {
135     Section(std::unique_ptr<Matcher> M) : SectionMatcher(std::move(M)){};
136 
137     std::unique_ptr<Matcher> SectionMatcher;
138     SectionEntries Entries;
139   };
140 
141   std::vector<Section> Sections;
142 
143   /// Parses just-constructed SpecialCaseList entries from a memory buffer.
144   bool parse(const MemoryBuffer *MB, StringMap<size_t> &SectionsMap,
145              std::string &Error);
146 
147   // Helper method for derived classes to search by Prefix, Query, and Category
148   // once they have already resolved a section entry.
149   unsigned inSectionBlame(const SectionEntries &Entries, StringRef Prefix,
150                           StringRef Query, StringRef Category) const;
151 };
152 
153 }  // namespace llvm
154 
155 #endif  // LLVM_SUPPORT_SPECIALCASELIST_H
156 
157