Home | History | Annotate | Download | only in Support 
      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 SpecialCaseList *SpecialCaseList::create(
     52     const StringRef Path, std::string &Error) {
     53   if (Path.empty())
     54     return 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 SpecialCaseList *SpecialCaseList::create(
     65     const MemoryBuffer *MB, std::string &Error) {
     66   std::unique_ptr<SpecialCaseList> SCL(new SpecialCaseList());
     67   if (!SCL->parse(MB, Error))
     68     return nullptr;
     69   return SCL.release();
     70 }
     71 
     72 SpecialCaseList *SpecialCaseList::createOrDie(const StringRef Path) {
     73   std::string Error;
     74   if (SpecialCaseList *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     // Backwards compatibility.
    107     if (Prefix == "global-init") {
    108       Prefix = "global";
    109       Category = "init";
    110     } else if (Prefix == "global-init-type") {
    111       Prefix = "type";
    112       Category = "init";
    113     } else if (Prefix == "global-init-src") {
    114       Prefix = "src";
    115       Category = "init";
    116     }
    117 
    118     // See if we can store Regexp in Strings.
    119     if (Regex::isLiteralERE(Regexp)) {
    120       Entries[Prefix][Category].Strings.insert(Regexp);
    121       continue;
    122     }
    123 
    124     // Replace * with .*
    125     for (size_t pos = 0; (pos = Regexp.find("*", pos)) != std::string::npos;
    126          pos += strlen(".*")) {
    127       Regexp.replace(pos, strlen("*"), ".*");
    128     }
    129 
    130     // Check that the regexp is valid.
    131     Regex CheckRE(Regexp);
    132     std::string REError;
    133     if (!CheckRE.isValid(REError)) {
    134       Error = (Twine("Malformed regex in line ") + Twine(LineNo) + ": '" +
    135                SplitLine.second + "': " + REError).str();
    136       return false;
    137     }
    138 
    139     // Add this regexp into the proper group by its prefix.
    140     if (!Regexps[Prefix][Category].empty())
    141       Regexps[Prefix][Category] += "|";
    142     Regexps[Prefix][Category] += "^" + Regexp + "$";
    143   }
    144 
    145   // Iterate through each of the prefixes, and create Regexs for them.
    146   for (StringMap<StringMap<std::string> >::const_iterator I = Regexps.begin(),
    147                                                           E = Regexps.end();
    148        I != E; ++I) {
    149     for (StringMap<std::string>::const_iterator II = I->second.begin(),
    150                                                 IE = I->second.end();
    151          II != IE; ++II) {
    152       Entries[I->getKey()][II->getKey()].RegEx.reset(new Regex(II->getValue()));
    153     }
    154   }
    155   return true;
    156 }
    157 
    158 SpecialCaseList::~SpecialCaseList() {}
    159 
    160 bool SpecialCaseList::inSection(const StringRef Section, const StringRef Query,
    161                                 const StringRef Category) const {
    162   StringMap<StringMap<Entry> >::const_iterator I = Entries.find(Section);
    163   if (I == Entries.end()) return false;
    164   StringMap<Entry>::const_iterator II = I->second.find(Category);
    165   if (II == I->second.end()) return false;
    166 
    167   return II->getValue().match(Query);
    168 }
    169 
    170 }  // namespace llvm
    171