Introduce an optimisation for special case lists with large numbers of literal entries.

Our internal regex implementation does not cope with large numbers
of anchors very efficiently.  Given a ~3600-entry special case list,
regex compilation can take on the order of seconds.  This patch solves
the problem for the special case of patterns matching literal global
names (i.e. patterns with no regex metacharacters).  Rather than
forming regexes from literal global name patterns, add them to
a StringSet which is checked before matching against the regex.
This reduces regex compilation time by an order of roughly thousands
when reading the aforementioned special case list, according to a
completely unscientific study.

No test cases.  I figure that any new tests for this code should
check that regex metacharacters are properly recognised.  However,
I could not find any documentation which documents the fact that the
syntax of global names in special case lists is based on regexes.
The extent to which regex syntax is supported in special case lists
should probably be decided on/documented before writing tests.

Differential Revision: http://llvm-reviews.chandlerc.com/D1150

llvm-svn: 187732
This commit is contained in:
Peter Collingbourne 2013-08-05 17:48:04 +00:00
parent 8df077ca5e
commit 42b450c977
2 changed files with 40 additions and 14 deletions
include/llvm/Transforms/Utils
lib/Transforms/Utils

View File

@ -89,7 +89,8 @@ class SpecialCaseList {
bool findCategory(const Module &M, StringRef &Category) const;
private:
StringMap<StringMap<Regex*> > Entries;
struct Entry;
StringMap<StringMap<Entry> > Entries;
void init(const MemoryBuffer *MB);
bool findCategory(const StringRef Section, const StringRef Query,

View File

@ -19,6 +19,7 @@
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/STLExtras.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"
@ -32,6 +33,22 @@
namespace llvm {
/// 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(0) {}
bool match(StringRef Query) const {
return Strings.count(Query) || (RegEx && RegEx->match(Query));
}
};
SpecialCaseList::SpecialCaseList(const StringRef Path) {
// Validate and open blacklist file.
if (Path.empty()) return;
@ -82,6 +99,12 @@ void SpecialCaseList::init(const MemoryBuffer *MB) {
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(".*")) {
@ -109,16 +132,20 @@ void SpecialCaseList::init(const MemoryBuffer *MB) {
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());
}
}
}
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;
}
}
}
@ -169,14 +196,13 @@ bool SpecialCaseList::isIn(const Module &M, const StringRef Category) const {
bool SpecialCaseList::findCategory(const StringRef Section,
const StringRef Query,
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;
for (StringMap<Regex *>::const_iterator II = I->second.begin(),
IE = I->second.end();
for (StringMap<Entry>::const_iterator II = I->second.begin(),
IE = I->second.end();
II != IE; ++II) {
Regex *FunctionRegex = II->getValue();
if (FunctionRegex->match(Query)) {
if (II->getValue().match(Query)) {
Category = II->first();
return true;
}
@ -188,13 +214,12 @@ bool SpecialCaseList::findCategory(const StringRef Section,
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