SpecialCaseList.cpp revision cd81d94322a39503e4a3e87b6ee03d4fcb3465fb
1474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org//===-- SpecialCaseList.cpp - special case list for sanitizers ------------===// 2474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// 3474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// The LLVM Compiler Infrastructure 4474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// 5474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// This file is distributed under the University of Illinois Open Source 6474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// License. See LICENSE.TXT for details. 7474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// 8474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org//===----------------------------------------------------------------------===// 9474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// 10474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// This is a utility class for instrumentation passes (like AddressSanitizer 11474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// or ThreadSanitizer) to avoid instrumenting some functions or global 128b26fe55f3e4daa2311dbd2d95e8ac2b4e080685johannkoenig@chromium.org// variables, or to instrument some functions or global variables in a specific 138b26fe55f3e4daa2311dbd2d95e8ac2b4e080685johannkoenig@chromium.org// way, based on a user-supplied list. 14474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org// 15474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org//===----------------------------------------------------------------------===// 16474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org 17474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include "llvm/Support/SpecialCaseList.h" 18474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include "llvm/ADT/STLExtras.h" 19474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include "llvm/ADT/SmallVector.h" 20474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include "llvm/ADT/StringExtras.h" 21474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include "llvm/ADT/StringSet.h" 22474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include "llvm/Support/MemoryBuffer.h" 23474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include "llvm/Support/Regex.h" 24474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include "llvm/Support/raw_ostream.h" 25ed759d81a39febed3a8a395386639d54307504aagrunell@chromium.org#include <string> 26474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org#include <system_error> 27167514562bbce1eb0566271d6cb41d90d2b5ffa0hclam@chromium.org#include <utility> 28167514562bbce1eb0566271d6cb41d90d2b5ffa0hclam@chromium.org 29474eb7536515fb785e925cc9375d22817c416851hclam@chromium.orgnamespace llvm { 30474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org 31474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org/// Represents a set of regular expressions. Regular expressions which are 32474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org/// "literal" (i.e. no regex metacharacters) are stored in Strings, while all 33474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org/// others are represented as a single pipe-separated regex in RegEx. The 34474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org/// reason for doing so is efficiency; StringSet is much faster at matching 35474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org/// literal strings than Regex. 36474eb7536515fb785e925cc9375d22817c416851hclam@chromium.orgstruct SpecialCaseList::Entry { 37474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org Entry() {} 38474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org Entry(Entry &&Other) 39474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org : Strings(std::move(Other.Strings)), RegEx(std::move(Other.RegEx)) {} 40474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org 41474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org StringSet<> Strings; 42474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org std::unique_ptr<Regex> RegEx; 43474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org 44474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org bool match(StringRef Query) const { 45474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org return Strings.count(Query) || (RegEx && RegEx->match(Query)); 46167514562bbce1eb0566271d6cb41d90d2b5ffa0hclam@chromium.org } 47474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org}; 48167514562bbce1eb0566271d6cb41d90d2b5ffa0hclam@chromium.org 49474eb7536515fb785e925cc9375d22817c416851hclam@chromium.orgSpecialCaseList::SpecialCaseList() : Entries() {} 50ed759d81a39febed3a8a395386639d54307504aagrunell@chromium.org 51ed759d81a39febed3a8a395386639d54307504aagrunell@chromium.orgSpecialCaseList *SpecialCaseList::create( 52ed759d81a39febed3a8a395386639d54307504aagrunell@chromium.org const StringRef Path, std::string &Error) { 53167514562bbce1eb0566271d6cb41d90d2b5ffa0hclam@chromium.org if (Path.empty()) 54474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org return new SpecialCaseList(); 55ed759d81a39febed3a8a395386639d54307504aagrunell@chromium.org ErrorOr<std::unique_ptr<MemoryBuffer>> FileOrErr = 56ed759d81a39febed3a8a395386639d54307504aagrunell@chromium.org MemoryBuffer::getFile(Path); 57474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org if (std::error_code EC = FileOrErr.getError()) { 58474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org Error = (Twine("Can't open file '") + Path + "': " + EC.message()).str(); 59474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org return nullptr; 60474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org } 61474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org return create(FileOrErr.get().get(), Error); 62474eb7536515fb785e925cc9375d22817c416851hclam@chromium.org} 638b26fe55f3e4daa2311dbd2d95e8ac2b4e080685johannkoenig@chromium.org 64SpecialCaseList *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 72SpecialCaseList *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 79bool 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 158SpecialCaseList::~SpecialCaseList() {} 159 160bool 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