src/llvm/lib/Support/SpecialCaseList.cpp

   1 //===-- SpecialCaseList.cpp - special case list for sanitizers ------------===//
   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 is a utility class for instrumentation passes (like AddressSanitizer
  11 // or ThreadSanitizer) to avoid instrumenting some functions or global
  12 // variables, or to instrument some functions or global variables in a specific
  13 // way, based on a user-supplied list.
  14 //
  15 //===----------------------------------------------------------------------===//
  16
  17 #include "llvm/Support/SpecialCaseList.h"
  18 #include "llvm/ADT/STLExtras.h"
  19 #include "llvm/ADT/SmallVector.h"
  20 #include "llvm/ADT/StringExtras.h"
  21 #include "llvm/ADT/StringSet.h"
  22 #include "llvm/Support/MemoryBuffer.h"
  23 #include "llvm/Support/Regex.h"
  24 #include "llvm/Support/raw_ostream.h"
  25 #include <string>
  26 #include <system_error>
  27 #include <utility>
  28
  29 namespace llvm {
  30
  31 /// Represents a set of regular expressions.  Regular expressions which are
  32 /// "literal" (i.e. no regex metacharacters) are stored in Strings, while all
  33 /// others are represented as a single pipe-separated regex in RegEx.  The
  34 /// reason for doing so is efficiency; StringSet is much faster at matching
  35 /// literal strings than Regex.
  36 struct SpecialCaseList::Entry {
  37   Entry() {}
  38   Entry(Entry &&Other)
  39       : Strings(std::move(Other.Strings)), RegEx(std::move(Other.RegEx)) {}
  40
  41   StringSet<> Strings;
  42   std::unique_ptr<Regex> RegEx;
  43
  44   bool match(StringRef Query) const {
  45     return Strings.count(Query) || (RegEx && RegEx->match(Query));
  46   }
  47 };
  48
  49 SpecialCaseList::SpecialCaseList() : Entries() {}
  50
  51 std::unique_ptr<SpecialCaseList> SpecialCaseList::create(StringRef Path,
  52                                                          std::string &Error) {
  53   if (Path.empty())
  54     return std::unique_ptr<SpecialCaseList>(new SpecialCaseList());
  55   ErrorOr<std::unique_ptr<MemoryBuffer>> FileOrErr =
  56       MemoryBuffer::getFile(Path);
  57   if (std::error_code EC = FileOrErr.getError()) {
  58     Error = (Twine("Can't open file '") + Path + "': " + EC.message()).str();
  59     return nullptr;
  60   }
  61   return create(FileOrErr.get().get(), Error);
  62 }
  63
  64 std::unique_ptr<SpecialCaseList> SpecialCaseList::create(const MemoryBuffer *MB,
  65                                                          std::string &Error) {
  66   std::unique_ptr<SpecialCaseList> SCL(new SpecialCaseList());
  67   if (!SCL->parse(MB, Error))
  68     return nullptr;
  69   return SCL;
  70 }
  71
  72 std::unique_ptr<SpecialCaseList> SpecialCaseList::createOrDie(StringRef Path) {
  73   std::string Error;
  74   if (auto SCL = create(Path, Error))
  75     return SCL;
  76   report_fatal_error(Error);
  77 }
  78
  79 bool SpecialCaseList::parse(const MemoryBuffer *MB, std::string &Error) {
  80   // Iterate through each line in the blacklist file.
  81   SmallVector<StringRef, 16> Lines;
  82   SplitString(MB->getBuffer(), Lines, "\n\r");
  83   StringMap<StringMap<std::string> > Regexps;
  84   assert(Entries.empty() &&
  85          "parse() should be called on an empty SpecialCaseList");
  86   int LineNo = 1;
  87   for (SmallVectorImpl<StringRef>::iterator I = Lines.begin(), E = Lines.end();
  88        I != E; ++I, ++LineNo) {
  89     // Ignore empty lines and lines starting with "#"
  90     if (I->empty() || I->startswith("#"))
  91       continue;
  92     // Get our prefix and unparsed regexp.
  93     std::pair<StringRef, StringRef> SplitLine = I->split(":");
  94     StringRef Prefix = SplitLine.first;
  95     if (SplitLine.second.empty()) {
  96       // Missing ':' in the line.
  97       Error = (Twine("Malformed line ") + Twine(LineNo) + ": '" +
  98                SplitLine.first + "'").str();
  99       return false;
 100     }
 101
 102     std::pair<StringRef, StringRef> SplitRegexp = SplitLine.second.split("=");
 103     std::string Regexp = SplitRegexp.first;
 104     StringRef Category = SplitRegexp.second;
 105
 106     // See if we can store Regexp in Strings.
 107     if (Regex::isLiteralERE(Regexp)) {
 108       Entries[Prefix][Category].Strings.insert(Regexp);
 109       continue;
 110     }
 111
 112     // Replace * with .*
 113     for (size_t pos = 0; (pos = Regexp.find("*", pos)) != std::string::npos;
 114          pos += strlen(".*")) {
 115       Regexp.replace(pos, strlen("*"), ".*");
 116     }
 117
 118     // Check that the regexp is valid.
 119     Regex CheckRE(Regexp);
 120     std::string REError;
 121     if (!CheckRE.isValid(REError)) {
 122       Error = (Twine("Malformed regex in line ") + Twine(LineNo) + ": '" +
 123                SplitLine.second + "': " + REError).str();
 124       return false;
 125     }
 126
 127     // Add this regexp into the proper group by its prefix.
 128     if (!Regexps[Prefix][Category].empty())
 129       Regexps[Prefix][Category] += "|";
 130     Regexps[Prefix][Category] += "^" + Regexp + "$";
 131   }
 132
 133   // Iterate through each of the prefixes, and create Regexs for them.
 134   for (StringMap<StringMap<std::string> >::const_iterator I = Regexps.begin(),
 135                                                           E = Regexps.end();
 136        I != E; ++I) {
 137     for (StringMap<std::string>::const_iterator II = I->second.begin(),
 138                                                 IE = I->second.end();
 139          II != IE; ++II) {
 140       Entries[I->getKey()][II->getKey()].RegEx.reset(new Regex(II->getValue()));
 141     }
 142   }
 143   return true;
 144 }
 145
 146 SpecialCaseList::~SpecialCaseList() {}
 147
 148 bool SpecialCaseList::inSection(StringRef Section, StringRef Query,
 149                                 StringRef Category) const {
 150   StringMap<StringMap<Entry> >::const_iterator I = Entries.find(Section);
 151   if (I == Entries.end()) return false;
 152   StringMap<Entry>::const_iterator II = I->second.find(Category);
 153   if (II == I->second.end()) return false;
 154
 155   return II->getValue().match(Query);
 156 }
 157
 158 }  // namespace llvm