SpecialCaseList.cpp revision 263508 
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/Transforms/Utils/SpecialCaseList.h"
18#include "llvm/ADT/OwningPtr.h"
19#include "llvm/ADT/SmallVector.h"
20#include "llvm/ADT/STLExtras.h"
21#include "llvm/ADT/StringExtras.h"
22#include "llvm/ADT/StringSet.h"
23#include "llvm/IR/DerivedTypes.h"
24#include "llvm/IR/Function.h"
25#include "llvm/IR/GlobalVariable.h"
26#include "llvm/IR/Module.h"
27#include "llvm/Support/MemoryBuffer.h"
28#include "llvm/Support/Regex.h"
29#include "llvm/Support/raw_ostream.h"
30#include "llvm/Support/system_error.h"
31#include <string>
32#include <utility>
33
34namespace llvm {
35
36/// Represents a set of regular expressions.  Regular expressions which are
37/// "literal" (i.e. no regex metacharacters) are stored in Strings, while all
38/// others are represented as a single pipe-separated regex in RegEx.  The
39/// reason for doing so is efficiency; StringSet is much faster at matching
40/// literal strings than Regex.
41struct SpecialCaseList::Entry {
42  StringSet<> Strings;
43  Regex *RegEx;
44
45  Entry() : RegEx(0) {}
46
47  bool match(StringRef Query) const {
48    return Strings.count(Query) || (RegEx && RegEx->match(Query));
49  }
50};
51
52SpecialCaseList::SpecialCaseList() : Entries() {}
53
54SpecialCaseList *SpecialCaseList::create(
55    const StringRef Path, std::string &Error) {
56  if (Path.empty())
57    return new SpecialCaseList();
58  OwningPtr<MemoryBuffer> File;
59  if (error_code EC = MemoryBuffer::getFile(Path, File)) {
60    Error = (Twine("Can't open file '") + Path + "': " + EC.message()).str();
61    return 0;
62  }
63  return create(File.get(), Error);
64}
65
66SpecialCaseList *SpecialCaseList::create(
67    const MemoryBuffer *MB, std::string &Error) {
68  OwningPtr<SpecialCaseList> SCL(new SpecialCaseList());
69  if (!SCL->parse(MB, Error))
70    return 0;
71  return SCL.take();
72}
73
74SpecialCaseList *SpecialCaseList::createOrDie(const StringRef Path) {
75  std::string Error;
76  if (SpecialCaseList *SCL = create(Path, Error))
77    return SCL;
78  report_fatal_error(Error);
79}
80
81bool SpecialCaseList::parse(const MemoryBuffer *MB, std::string &Error) {
82  // Iterate through each line in the blacklist file.
83  SmallVector<StringRef, 16> Lines;
84  SplitString(MB->getBuffer(), Lines, "\n\r");
85  StringMap<StringMap<std::string> > Regexps;
86  assert(Entries.empty() &&
87         "parse() should be called on an empty SpecialCaseList");
88  int LineNo = 1;
89  for (SmallVectorImpl<StringRef>::iterator I = Lines.begin(), E = Lines.end();
90       I != E; ++I, ++LineNo) {
91    // Ignore empty lines and lines starting with "#"
92    if (I->empty() || I->startswith("#"))
93      continue;
94    // Get our prefix and unparsed regexp.
95    std::pair<StringRef, StringRef> SplitLine = I->split(":");
96    StringRef Prefix = SplitLine.first;
97    if (SplitLine.second.empty()) {
98      // Missing ':' in the line.
99      Error = (Twine("Malformed line ") + Twine(LineNo) + ": '" +
100               SplitLine.first + "'").str();
101      return false;
102    }
103
104    std::pair<StringRef, StringRef> SplitRegexp = SplitLine.second.split("=");
105    std::string Regexp = SplitRegexp.first;
106    StringRef Category = SplitRegexp.second;
107
108    // Backwards compatibility.
109    if (Prefix == "global-init") {
110      Prefix = "global";
111      Category = "init";
112    } else if (Prefix == "global-init-type") {
113      Prefix = "type";
114      Category = "init";
115    } else if (Prefix == "global-init-src") {
116      Prefix = "src";
117      Category = "init";
118    }
119
120    // See if we can store Regexp in Strings.
121    if (Regex::isLiteralERE(Regexp)) {
122      Entries[Prefix][Category].Strings.insert(Regexp);
123      continue;
124    }
125
126    // Replace * with .*
127    for (size_t pos = 0; (pos = Regexp.find("*", pos)) != std::string::npos;
128         pos += strlen(".*")) {
129      Regexp.replace(pos, strlen("*"), ".*");
130    }
131
132    // Check that the regexp is valid.
133    Regex CheckRE(Regexp);
134    std::string REError;
135    if (!CheckRE.isValid(REError)) {
136      Error = (Twine("Malformed regex in line ") + Twine(LineNo) + ": '" +
137               SplitLine.second + "': " + REError).str();
138      return false;
139    }
140
141    // Add this regexp into the proper group by its prefix.
142    if (!Regexps[Prefix][Category].empty())
143      Regexps[Prefix][Category] += "|";
144    Regexps[Prefix][Category] += "^" + Regexp + "$";
145  }
146
147  // Iterate through each of the prefixes, and create Regexs for them.
148  for (StringMap<StringMap<std::string> >::const_iterator I = Regexps.begin(),
149                                                          E = Regexps.end();
150       I != E; ++I) {
151    for (StringMap<std::string>::const_iterator II = I->second.begin(),
152                                                IE = I->second.end();
153         II != IE; ++II) {
154      Entries[I->getKey()][II->getKey()].RegEx = new Regex(II->getValue());
155    }
156  }
157  return true;
158}
159
160SpecialCaseList::~SpecialCaseList() {
161  for (StringMap<StringMap<Entry> >::iterator I = Entries.begin(),
162                                              E = Entries.end();
163       I != E; ++I) {
164    for (StringMap<Entry>::const_iterator II = I->second.begin(),
165                                          IE = I->second.end();
166         II != IE; ++II) {
167      delete II->second.RegEx;
168    }
169  }
170}
171
172bool SpecialCaseList::isIn(const Function& F, const StringRef Category) const {
173  return isIn(*F.getParent(), Category) ||
174         inSectionCategory("fun", F.getName(), Category);
175}
176
177static StringRef GetGlobalTypeString(const GlobalValue &G) {
178  // Types of GlobalVariables are always pointer types.
179  Type *GType = G.getType()->getElementType();
180  // For now we support blacklisting struct types only.
181  if (StructType *SGType = dyn_cast<StructType>(GType)) {
182    if (!SGType->isLiteral())
183      return SGType->getName();
184  }
185  return "<unknown type>";
186}
187
188bool SpecialCaseList::isIn(const GlobalVariable &G,
189                           const StringRef Category) const {
190  return isIn(*G.getParent(), Category) ||
191         inSectionCategory("global", G.getName(), Category) ||
192         inSectionCategory("type", GetGlobalTypeString(G), Category);
193}
194
195bool SpecialCaseList::isIn(const GlobalAlias &GA,
196                           const StringRef Category) const {
197  if (isIn(*GA.getParent(), Category))
198    return true;
199
200  if (isa<FunctionType>(GA.getType()->getElementType()))
201    return inSectionCategory("fun", GA.getName(), Category);
202
203  return inSectionCategory("global", GA.getName(), Category) ||
204         inSectionCategory("type", GetGlobalTypeString(GA), Category);
205}
206
207bool SpecialCaseList::isIn(const Module &M, const StringRef Category) const {
208  return inSectionCategory("src", M.getModuleIdentifier(), Category);
209}
210
211bool SpecialCaseList::inSectionCategory(const StringRef Section,
212                                        const StringRef Query,
213                                        const StringRef Category) const {
214  StringMap<StringMap<Entry> >::const_iterator I = Entries.find(Section);
215  if (I == Entries.end()) return false;
216  StringMap<Entry>::const_iterator II = I->second.find(Category);
217  if (II == I->second.end()) return false;
218
219  return II->getValue().match(Query);
220}
221
222}  // namespace llvm
223