//===----------------------------------------------------------------------===//
#include "llvm/Transforms/Utils/SpecialCaseList.h"
-#include "llvm/ADT/OwningPtr.h"
-#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/StringExtras.h"
+#include "llvm/ADT/StringSet.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/GlobalVariable.h"
#include "llvm/Support/MemoryBuffer.h"
#include "llvm/Support/Regex.h"
#include "llvm/Support/raw_ostream.h"
-#include "llvm/Support/system_error.h"
#include <string>
+#include <system_error>
#include <utility>
namespace llvm {
+using std::error_code;
+
+/// Represents a set of regular expressions. Regular expressions which are
+/// "literal" (i.e. no regex metacharacters) are stored in Strings, while all
+/// others are represented as a single pipe-separated regex in RegEx. The
+/// reason for doing so is efficiency; StringSet is much faster at matching
+/// literal strings than Regex.
+struct SpecialCaseList::Entry {
+ StringSet<> Strings;
+ Regex *RegEx;
+
+ Entry() : RegEx(nullptr) {}
-SpecialCaseList::SpecialCaseList(const StringRef Path) {
- // Validate and open blacklist file.
- if (Path.empty()) return;
- OwningPtr<MemoryBuffer> File;
+ bool match(StringRef Query) const {
+ return Strings.count(Query) || (RegEx && RegEx->match(Query));
+ }
+};
+
+SpecialCaseList::SpecialCaseList() : Entries() {}
+
+SpecialCaseList *SpecialCaseList::create(
+ const StringRef Path, std::string &Error) {
+ if (Path.empty())
+ return new SpecialCaseList();
+ std::unique_ptr<MemoryBuffer> File;
if (error_code EC = MemoryBuffer::getFile(Path, File)) {
- report_fatal_error("Can't open blacklist file: " + Path + ": " +
- EC.message());
+ Error = (Twine("Can't open file '") + Path + "': " + EC.message()).str();
+ return nullptr;
}
+ return create(File.get(), Error);
+}
- init(File.get());
+SpecialCaseList *SpecialCaseList::create(
+ const MemoryBuffer *MB, std::string &Error) {
+ std::unique_ptr<SpecialCaseList> SCL(new SpecialCaseList());
+ if (!SCL->parse(MB, Error))
+ return nullptr;
+ return SCL.release();
}
-SpecialCaseList::SpecialCaseList(const MemoryBuffer *MB) {
- init(MB);
+SpecialCaseList *SpecialCaseList::createOrDie(const StringRef Path) {
+ std::string Error;
+ if (SpecialCaseList *SCL = create(Path, Error))
+ return SCL;
+ report_fatal_error(Error);
}
-void SpecialCaseList::init(const MemoryBuffer *MB) {
+bool SpecialCaseList::parse(const MemoryBuffer *MB, std::string &Error) {
// Iterate through each line in the blacklist file.
SmallVector<StringRef, 16> Lines;
SplitString(MB->getBuffer(), Lines, "\n\r");
StringMap<StringMap<std::string> > Regexps;
+ assert(Entries.empty() &&
+ "parse() should be called on an empty SpecialCaseList");
+ int LineNo = 1;
for (SmallVectorImpl<StringRef>::iterator I = Lines.begin(), E = Lines.end();
- I != E; ++I) {
+ I != E; ++I, ++LineNo) {
// Ignore empty lines and lines starting with "#"
if (I->empty() || I->startswith("#"))
continue;
StringRef Prefix = SplitLine.first;
if (SplitLine.second.empty()) {
// Missing ':' in the line.
- report_fatal_error("malformed blacklist line: " + SplitLine.first);
+ Error = (Twine("Malformed line ") + Twine(LineNo) + ": '" +
+ SplitLine.first + "'").str();
+ return false;
}
std::pair<StringRef, StringRef> SplitRegexp = SplitLine.second.split("=");
Category = "init";
}
+ // See if we can store Regexp in Strings.
+ if (Regex::isLiteralERE(Regexp)) {
+ Entries[Prefix][Category].Strings.insert(Regexp);
+ continue;
+ }
+
// Replace * with .*
for (size_t pos = 0; (pos = Regexp.find("*", pos)) != std::string::npos;
pos += strlen(".*")) {
// Check that the regexp is valid.
Regex CheckRE(Regexp);
- std::string Error;
- if (!CheckRE.isValid(Error)) {
- report_fatal_error("malformed blacklist regex: " + SplitLine.second +
- ": " + Error);
+ std::string REError;
+ if (!CheckRE.isValid(REError)) {
+ Error = (Twine("Malformed regex in line ") + Twine(LineNo) + ": '" +
+ SplitLine.second + "': " + REError).str();
+ return false;
}
// Add this regexp into the proper group by its prefix.
for (StringMap<std::string>::const_iterator II = I->second.begin(),
IE = I->second.end();
II != IE; ++II) {
- Entries[I->getKey()][II->getKey()] = new Regex(II->getValue());
+ Entries[I->getKey()][II->getKey()].RegEx = new Regex(II->getValue());
}
}
+ return true;
}
SpecialCaseList::~SpecialCaseList() {
- for (StringMap<StringMap<Regex*> >::iterator I = Entries.begin(),
- E = Entries.end();
+ for (StringMap<StringMap<Entry> >::iterator I = Entries.begin(),
+ E = Entries.end();
I != E; ++I) {
- DeleteContainerSeconds(I->second);
+ for (StringMap<Entry>::const_iterator II = I->second.begin(),
+ IE = I->second.end();
+ II != IE; ++II) {
+ delete II->second.RegEx;
+ }
}
}
-bool SpecialCaseList::findCategory(const Function &F,
- StringRef &Category) const {
- return findCategory(*F.getParent(), Category) ||
- findCategory("fun", F.getName(), Category);
-}
-
bool SpecialCaseList::isIn(const Function& F, const StringRef Category) const {
return isIn(*F.getParent(), Category) ||
inSectionCategory("fun", F.getName(), Category);
}
-static StringRef GetGVTypeString(const GlobalVariable &G) {
+static StringRef GetGlobalTypeString(const GlobalValue &G) {
// Types of GlobalVariables are always pointer types.
Type *GType = G.getType()->getElementType();
// For now we support blacklisting struct types only.
return "<unknown type>";
}
-bool SpecialCaseList::findCategory(const GlobalVariable &G,
- StringRef &Category) const {
- return findCategory(*G.getParent(), Category) ||
- findCategory("global", G.getName(), Category) ||
- findCategory("type", GetGVTypeString(G), Category);
-}
-
bool SpecialCaseList::isIn(const GlobalVariable &G,
const StringRef Category) const {
return isIn(*G.getParent(), Category) ||
inSectionCategory("global", G.getName(), Category) ||
- inSectionCategory("type", GetGVTypeString(G), Category);
+ inSectionCategory("type", GetGlobalTypeString(G), Category);
}
-bool SpecialCaseList::findCategory(const Module &M, StringRef &Category) const {
- return findCategory("src", M.getModuleIdentifier(), Category);
+bool SpecialCaseList::isIn(const GlobalAlias &GA,
+ const StringRef Category) const {
+ if (isIn(*GA.getParent(), Category))
+ return true;
+
+ if (isa<FunctionType>(GA.getType()->getElementType()))
+ return inSectionCategory("fun", GA.getName(), Category);
+
+ return inSectionCategory("global", GA.getName(), Category) ||
+ inSectionCategory("type", GetGlobalTypeString(GA), Category);
}
bool SpecialCaseList::isIn(const Module &M, const StringRef Category) const {
return inSectionCategory("src", M.getModuleIdentifier(), Category);
}
-bool SpecialCaseList::findCategory(const StringRef Section,
- const StringRef Query,
- StringRef &Category) const {
- StringMap<StringMap<Regex *> >::const_iterator I = Entries.find(Section);
- if (I == Entries.end()) return false;
-
- for (StringMap<Regex *>::const_iterator II = I->second.begin(),
- IE = I->second.end();
- II != IE; ++II) {
- Regex *FunctionRegex = II->getValue();
- if (FunctionRegex->match(Query)) {
- Category = II->first();
- return true;
- }
- }
-
- return false;
-}
-
bool SpecialCaseList::inSectionCategory(const StringRef Section,
const StringRef Query,
const StringRef Category) const {
- StringMap<StringMap<Regex *> >::const_iterator I = Entries.find(Section);
+ StringMap<StringMap<Entry> >::const_iterator I = Entries.find(Section);
if (I == Entries.end()) return false;
- StringMap<Regex *>::const_iterator II = I->second.find(Category);
+ StringMap<Entry>::const_iterator II = I->second.find(Category);
if (II == I->second.end()) return false;
- Regex *FunctionRegex = II->getValue();
- return FunctionRegex->match(Query);
+ return II->getValue().match(Query);
}
} // namespace llvm