llvm-project/clang/lib/Lex/PTHLexer.cpp

702 lines
23 KiB
C++
Raw Normal View History

//===--- PTHLexer.cpp - Lex from a token stream ---------------------------===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file implements the PTHLexer interface.
//
//===----------------------------------------------------------------------===//
#include "clang/Basic/TokenKinds.h"
#include "clang/Basic/FileManager.h"
#include "clang/Basic/IdentifierTable.h"
#include "clang/Basic/OnDiskHashTable.h"
#include "clang/Lex/PTHLexer.h"
#include "clang/Lex/Preprocessor.h"
#include "clang/Lex/PTHManager.h"
#include "clang/Lex/Token.h"
#include "clang/Lex/Preprocessor.h"
#include "llvm/ADT/StringMap.h"
#include "llvm/ADT/OwningPtr.h"
#include "llvm/Support/MemoryBuffer.h"
#include <sys/stat.h>
using namespace clang;
using namespace clang::io;
#define DISK_TOKEN_SIZE (1+1+2+4+4)
//===----------------------------------------------------------------------===//
// PTHLexer methods.
//===----------------------------------------------------------------------===//
PTHLexer::PTHLexer(Preprocessor &PP, FileID FID, const unsigned char *D,
const unsigned char *ppcond, PTHManager &PM)
: PreprocessorLexer(&PP, FID), TokBuf(D), CurPtr(D), LastHashTokPtr(0),
PPCond(ppcond), CurPPCondPtr(ppcond), PTHMgr(PM) {
FileStartLoc = PP.getSourceManager().getLocForStartOfFile(FID);
}
void PTHLexer::Lex(Token& Tok) {
LexNextToken:
//===--------------------------------------==//
// Read the raw token data.
//===--------------------------------------==//
// Shadow CurPtr into an automatic variable.
const unsigned char *CurPtrShadow = CurPtr;
// Read in the data for the token.
unsigned Word0 = ReadLE32(CurPtrShadow);
uint32_t IdentifierID = ReadLE32(CurPtrShadow);
uint32_t FileOffset = ReadLE32(CurPtrShadow);
tok::TokenKind TKind = (tok::TokenKind) (Word0 & 0xFF);
Token::TokenFlags TFlags = (Token::TokenFlags) ((Word0 >> 8) & 0xFF);
uint32_t Len = Word0 >> 16;
CurPtr = CurPtrShadow;
//===--------------------------------------==//
// Construct the token itself.
//===--------------------------------------==//
Tok.startToken();
Tok.setKind(TKind);
Tok.setFlag(TFlags);
assert(!LexingRawMode);
Tok.setLocation(FileStartLoc.getFileLocWithOffset(FileOffset));
Tok.setLength(Len);
// Handle identifiers.
if (Tok.isLiteral()) {
Tok.setLiteralData((const char*) (PTHMgr.SpellingBase + IdentifierID));
}
else if (IdentifierID) {
MIOpt.ReadToken();
IdentifierInfo *II = PTHMgr.GetIdentifierInfo(IdentifierID-1);
Tok.setIdentifierInfo(II);
// Change the kind of this identifier to the appropriate token kind, e.g.
// turning "for" into a keyword.
Tok.setKind(II->getTokenID());
if (II->isHandleIdentifierCase())
PP->HandleIdentifier(Tok);
return;
}
//===--------------------------------------==//
// Process the token.
//===--------------------------------------==//
#if 0
SourceManager& SM = PP->getSourceManager();
llvm::cerr << SM.getFileEntryForID(FileID)->getName()
<< ':' << SM.getLogicalLineNumber(Tok.getLocation())
<< ':' << SM.getLogicalColumnNumber(Tok.getLocation())
<< '\n';
#endif
if (TKind == tok::eof) {
// Save the end-of-file token.
EofToken = Tok;
Preprocessor *PPCache = PP;
assert(!ParsingPreprocessorDirective);
assert(!LexingRawMode);
// FIXME: Issue diagnostics similar to Lexer.
if (PP->HandleEndOfFile(Tok, false))
return;
assert(PPCache && "Raw buffer::LexEndOfFile should return a token");
return PPCache->Lex(Tok);
}
if (TKind == tok::hash && Tok.isAtStartOfLine()) {
LastHashTokPtr = CurPtr - DISK_TOKEN_SIZE;
assert(!LexingRawMode);
PP->HandleDirective(Tok);
if (PP->isCurrentLexer(this))
goto LexNextToken;
return PP->Lex(Tok);
}
if (TKind == tok::eom) {
assert(ParsingPreprocessorDirective);
ParsingPreprocessorDirective = false;
return;
}
MIOpt.ReadToken();
}
// FIXME: We can just grab the last token instead of storing a copy
// into EofToken.
void PTHLexer::getEOF(Token& Tok) {
assert(EofToken.is(tok::eof));
Tok = EofToken;
}
void PTHLexer::DiscardToEndOfLine() {
assert(ParsingPreprocessorDirective && ParsingFilename == false &&
"Must be in a preprocessing directive!");
// We assume that if the preprocessor wishes to discard to the end of
// the line that it also means to end the current preprocessor directive.
ParsingPreprocessorDirective = false;
// Skip tokens by only peeking at their token kind and the flags.
// We don't need to actually reconstruct full tokens from the token buffer.
// This saves some copies and it also reduces IdentifierInfo* lookup.
const unsigned char* p = CurPtr;
while (1) {
// Read the token kind. Are we at the end of the file?
tok::TokenKind x = (tok::TokenKind) (uint8_t) *p;
if (x == tok::eof) break;
// Read the token flags. Are we at the start of the next line?
Token::TokenFlags y = (Token::TokenFlags) (uint8_t) p[1];
if (y & Token::StartOfLine) break;
// Skip to the next token.
p += DISK_TOKEN_SIZE;
}
CurPtr = p;
}
/// SkipBlock - Used by Preprocessor to skip the current conditional block.
bool PTHLexer::SkipBlock() {
assert(CurPPCondPtr && "No cached PP conditional information.");
assert(LastHashTokPtr && "No known '#' token.");
const unsigned char* HashEntryI = 0;
uint32_t Offset;
uint32_t TableIdx;
do {
// Read the token offset from the side-table.
Offset = ReadLE32(CurPPCondPtr);
// Read the target table index from the side-table.
TableIdx = ReadLE32(CurPPCondPtr);
// Compute the actual memory address of the '#' token data for this entry.
HashEntryI = TokBuf + Offset;
// Optmization: "Sibling jumping". #if...#else...#endif blocks can
// contain nested blocks. In the side-table we can jump over these
// nested blocks instead of doing a linear search if the next "sibling"
// entry is not at a location greater than LastHashTokPtr.
if (HashEntryI < LastHashTokPtr && TableIdx) {
// In the side-table we are still at an entry for a '#' token that
// is earlier than the last one we saw. Check if the location we would
// stride gets us closer.
const unsigned char* NextPPCondPtr =
PPCond + TableIdx*(sizeof(uint32_t)*2);
assert(NextPPCondPtr >= CurPPCondPtr);
// Read where we should jump to.
uint32_t TmpOffset = ReadLE32(NextPPCondPtr);
const unsigned char* HashEntryJ = TokBuf + TmpOffset;
if (HashEntryJ <= LastHashTokPtr) {
// Jump directly to the next entry in the side table.
HashEntryI = HashEntryJ;
Offset = TmpOffset;
TableIdx = ReadLE32(NextPPCondPtr);
CurPPCondPtr = NextPPCondPtr;
}
}
}
while (HashEntryI < LastHashTokPtr);
assert(HashEntryI == LastHashTokPtr && "No PP-cond entry found for '#'");
assert(TableIdx && "No jumping from #endifs.");
// Update our side-table iterator.
const unsigned char* NextPPCondPtr = PPCond + TableIdx*(sizeof(uint32_t)*2);
assert(NextPPCondPtr >= CurPPCondPtr);
CurPPCondPtr = NextPPCondPtr;
// Read where we should jump to.
HashEntryI = TokBuf + ReadLE32(NextPPCondPtr);
uint32_t NextIdx = ReadLE32(NextPPCondPtr);
// By construction NextIdx will be zero if this is a #endif. This is useful
// to know to obviate lexing another token.
bool isEndif = NextIdx == 0;
// This case can occur when we see something like this:
//
// #if ...
// /* a comment or nothing */
// #elif
//
// If we are skipping the first #if block it will be the case that CurPtr
// already points 'elif'. Just return.
if (CurPtr > HashEntryI) {
assert(CurPtr == HashEntryI + DISK_TOKEN_SIZE);
// Did we reach a #endif? If so, go ahead and consume that token as well.
if (isEndif)
CurPtr += DISK_TOKEN_SIZE*2;
else
LastHashTokPtr = HashEntryI;
return isEndif;
}
// Otherwise, we need to advance. Update CurPtr to point to the '#' token.
CurPtr = HashEntryI;
// Update the location of the last observed '#'. This is useful if we
// are skipping multiple blocks.
LastHashTokPtr = CurPtr;
// Skip the '#' token.
assert(((tok::TokenKind)*CurPtr) == tok::hash);
CurPtr += DISK_TOKEN_SIZE;
// Did we reach a #endif? If so, go ahead and consume that token as well.
if (isEndif) { CurPtr += DISK_TOKEN_SIZE*2; }
return isEndif;
}
SourceLocation PTHLexer::getSourceLocation() {
// getSourceLocation is not on the hot path. It is used to get the location
// of the next token when transitioning back to this lexer when done
// handling a #included file. Just read the necessary data from the token
// data buffer to construct the SourceLocation object.
// NOTE: This is a virtual function; hence it is defined out-of-line.
const unsigned char *OffsetPtr = CurPtr + (DISK_TOKEN_SIZE - 4);
uint32_t Offset = ReadLE32(OffsetPtr);
return FileStartLoc.getFileLocWithOffset(Offset);
}
//===----------------------------------------------------------------------===//
// PTH file lookup: map from strings to file data.
//===----------------------------------------------------------------------===//
/// PTHFileLookup - This internal data structure is used by the PTHManager
/// to map from FileEntry objects managed by FileManager to offsets within
/// the PTH file.
namespace {
class VISIBILITY_HIDDEN PTHFileData {
const uint32_t TokenOff;
const uint32_t PPCondOff;
public:
PTHFileData(uint32_t tokenOff, uint32_t ppCondOff)
: TokenOff(tokenOff), PPCondOff(ppCondOff) {}
uint32_t getTokenOffset() const { return TokenOff; }
uint32_t getPPCondOffset() const { return PPCondOff; }
};
class VISIBILITY_HIDDEN PTHFileLookupCommonTrait {
public:
typedef std::pair<unsigned char, const char*> internal_key_type;
static unsigned ComputeHash(internal_key_type x) {
return BernsteinHash(x.second);
}
static std::pair<unsigned, unsigned>
ReadKeyDataLength(const unsigned char*& d) {
unsigned keyLen = (unsigned) ReadUnalignedLE16(d);
unsigned dataLen = (unsigned) *(d++);
return std::make_pair(keyLen, dataLen);
}
static internal_key_type ReadKey(const unsigned char* d, unsigned) {
unsigned char k = *(d++); // Read the entry kind.
return std::make_pair(k, (const char*) d);
}
};
class VISIBILITY_HIDDEN PTHFileLookupTrait : public PTHFileLookupCommonTrait {
public:
typedef const FileEntry* external_key_type;
typedef PTHFileData data_type;
static internal_key_type GetInternalKey(const FileEntry* FE) {
return std::make_pair((unsigned char) 0x1, FE->getName());
}
static bool EqualKey(internal_key_type a, internal_key_type b) {
return a.first == b.first && strcmp(a.second, b.second) == 0;
}
static PTHFileData ReadData(const internal_key_type& k,
const unsigned char* d, unsigned) {
assert(k.first == 0x1 && "Only file lookups can match!");
uint32_t x = ::ReadUnalignedLE32(d);
uint32_t y = ::ReadUnalignedLE32(d);
return PTHFileData(x, y);
}
};
class VISIBILITY_HIDDEN PTHStringLookupTrait {
public:
typedef uint32_t
data_type;
typedef const std::pair<const char*, unsigned>
external_key_type;
typedef external_key_type internal_key_type;
static bool EqualKey(const internal_key_type& a,
const internal_key_type& b) {
return (a.second == b.second) ? memcmp(a.first, b.first, a.second) == 0
: false;
}
static unsigned ComputeHash(const internal_key_type& a) {
return BernsteinHash(a.first, a.second);
}
// This hopefully will just get inlined and removed by the optimizer.
static const internal_key_type&
GetInternalKey(const external_key_type& x) { return x; }
static std::pair<unsigned, unsigned>
ReadKeyDataLength(const unsigned char*& d) {
return std::make_pair((unsigned) ReadUnalignedLE16(d), sizeof(uint32_t));
}
static std::pair<const char*, unsigned>
ReadKey(const unsigned char* d, unsigned n) {
assert(n >= 2 && d[n-1] == '\0');
return std::make_pair((const char*) d, n-1);
}
static uint32_t ReadData(const internal_key_type& k, const unsigned char* d,
unsigned) {
return ::ReadUnalignedLE32(d);
}
};
} // end anonymous namespace
typedef OnDiskChainedHashTable<PTHFileLookupTrait> PTHFileLookup;
typedef OnDiskChainedHashTable<PTHStringLookupTrait> PTHStringIdLookup;
//===----------------------------------------------------------------------===//
// PTHManager methods.
//===----------------------------------------------------------------------===//
PTHManager::PTHManager(const llvm::MemoryBuffer* buf, void* fileLookup,
const unsigned char* idDataTable,
IdentifierInfo** perIDCache,
void* stringIdLookup, unsigned numIds,
const unsigned char* spellingBase,
const char* originalSourceFile)
: Buf(buf), PerIDCache(perIDCache), FileLookup(fileLookup),
IdDataTable(idDataTable), StringIdLookup(stringIdLookup),
NumIds(numIds), PP(0), SpellingBase(spellingBase),
OriginalSourceFile(originalSourceFile) {}
PTHManager::~PTHManager() {
delete Buf;
delete (PTHFileLookup*) FileLookup;
delete (PTHStringIdLookup*) StringIdLookup;
free(PerIDCache);
}
static void InvalidPTH(Diagnostic *Diags, Diagnostic::Level level,
const char* Msg = 0) {
if (!Diags) return;
if (!Msg) Msg = "Invalid or corrupted PTH file";
unsigned DiagID = Diags->getCustomDiagID(level, Msg);
Diags->Report(FullSourceLoc(), DiagID);
}
PTHManager* PTHManager::Create(const std::string& file, Diagnostic* Diags,
Diagnostic::Level level) {
// Memory map the PTH file.
llvm::OwningPtr<llvm::MemoryBuffer>
File(llvm::MemoryBuffer::getFile(file.c_str()));
if (!File) {
if (Diags) {
unsigned DiagID = Diags->getCustomDiagID(level,
"PTH file %0 could not be read");
Diags->Report(FullSourceLoc(), DiagID) << file;
}
return 0;
}
// Get the buffer ranges and check if there are at least three 32-bit
// words at the end of the file.
const unsigned char* BufBeg = (unsigned char*)File->getBufferStart();
const unsigned char* BufEnd = (unsigned char*)File->getBufferEnd();
// Check the prologue of the file.
2009-01-27 06:16:12 +08:00
if ((BufEnd - BufBeg) < (signed) (sizeof("cfe-pth") + 3 + 4) ||
memcmp(BufBeg, "cfe-pth", sizeof("cfe-pth") - 1) != 0) {
InvalidPTH(Diags, level);
return 0;
}
// Read the PTH version.
const unsigned char *p = BufBeg + (sizeof("cfe-pth") - 1);
unsigned Version = ReadLE32(p);
if (Version != PTHManager::Version) {
InvalidPTH(Diags, level,
Version < PTHManager::Version
? "PTH file uses an older PTH format that is no longer supported"
: "PTH file uses a newer PTH format that cannot be read");
return 0;
}
// Compute the address of the index table at the end of the PTH file.
const unsigned char *PrologueOffset = p;
if (PrologueOffset >= BufEnd) {
InvalidPTH(Diags, level);
return 0;
}
// Construct the file lookup table. This will be used for mapping from
// FileEntry*'s to cached tokens.
const unsigned char* FileTableOffset = PrologueOffset + sizeof(uint32_t)*2;
const unsigned char* FileTable = BufBeg + ReadLE32(FileTableOffset);
if (!(FileTable > BufBeg && FileTable < BufEnd)) {
InvalidPTH(Diags, level);
return 0; // FIXME: Proper error diagnostic?
}
llvm::OwningPtr<PTHFileLookup> FL(PTHFileLookup::Create(FileTable, BufBeg));
// Warn if the PTH file is empty. We still want to create a PTHManager
// as the PTH could be used with -include-pth.
if (FL->isEmpty())
InvalidPTH(Diags, level, "PTH file contains no cached source data");
// Get the location of the table mapping from persistent ids to the
// data needed to reconstruct identifiers.
const unsigned char* IDTableOffset = PrologueOffset + sizeof(uint32_t)*0;
const unsigned char* IData = BufBeg + ReadLE32(IDTableOffset);
if (!(IData >= BufBeg && IData < BufEnd)) {
InvalidPTH(Diags, level);
return 0;
}
// Get the location of the hashtable mapping between strings and
// persistent IDs.
const unsigned char* StringIdTableOffset = PrologueOffset + sizeof(uint32_t)*1;
const unsigned char* StringIdTable = BufBeg + ReadLE32(StringIdTableOffset);
if (!(StringIdTable >= BufBeg && StringIdTable < BufEnd)) {
InvalidPTH(Diags, level);
return 0;
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
}
llvm::OwningPtr<PTHStringIdLookup> SL(PTHStringIdLookup::Create(StringIdTable,
BufBeg));
// Get the location of the spelling cache.
const unsigned char* spellingBaseOffset = PrologueOffset + sizeof(uint32_t)*3;
const unsigned char* spellingBase = BufBeg + ReadLE32(spellingBaseOffset);
if (!(spellingBase >= BufBeg && spellingBase < BufEnd)) {
InvalidPTH(Diags, level);
return 0;
}
// Get the number of IdentifierInfos and pre-allocate the identifier cache.
uint32_t NumIds = ReadLE32(IData);
// Pre-allocate the peristent ID -> IdentifierInfo* cache. We use calloc()
// so that we in the best case only zero out memory once when the OS returns
// us new pages.
IdentifierInfo** PerIDCache = 0;
if (NumIds) {
PerIDCache = (IdentifierInfo**)calloc(NumIds, sizeof(*PerIDCache));
if (!PerIDCache) {
InvalidPTH(Diags, level,
"Could not allocate memory for processing PTH file");
return 0;
}
}
// Compute the address of the original source file.
const unsigned char* originalSourceBase = PrologueOffset + sizeof(uint32_t)*4;
unsigned len = ReadUnalignedLE16(originalSourceBase);
if (!len) originalSourceBase = 0;
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
// Create the new PTHManager.
return new PTHManager(File.take(), FL.take(), IData, PerIDCache,
SL.take(), NumIds, spellingBase,
(const char*) originalSourceBase);
}
IdentifierInfo* PTHManager::LazilyCreateIdentifierInfo(unsigned PersistentID) {
// Look in the PTH file for the string data for the IdentifierInfo object.
const unsigned char* TableEntry = IdDataTable + sizeof(uint32_t)*PersistentID;
const unsigned char* IDData =
(const unsigned char*)Buf->getBufferStart() + ReadLE32(TableEntry);
assert(IDData < (const unsigned char*)Buf->getBufferEnd());
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
// Allocate the object.
std::pair<IdentifierInfo,const unsigned char*> *Mem =
Alloc.Allocate<std::pair<IdentifierInfo,const unsigned char*> >();
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
Mem->second = IDData;
assert(IDData[0] != '\0');
Fix: <rdar://problem/6510344> [pth] PTH slows down regular lexer considerably (when it has substantial work) Changes to IdentifierTable: - High-level summary: StringMap never owns IdentifierInfos. It just references them. - The string map now has StringMapEntry<IdentifierInfo*> instead of StringMapEntry<IdentifierInfo>. The IdentifierInfo object is allocated using the same bump pointer allocator as used by the StringMap. Changes to IdentifierInfo: - Added an extra pointer to point to the StringMapEntry<IdentifierInfo*> in the string map. This pointer will be null if the IdentifierInfo* is *only* used by the PTHLexer (that is it isn't in the StringMap). Algorithmic changes: - Non-PTH case: IdentifierInfo::get() will always consult the StringMap first to see if we have an IdentifierInfo object. If that StringMapEntry references a null pointer, we allocate a new one from the BumpPtrAllocator and update the reference in the StringMapEntry. - PTH case: We do the same lookup as with the non-PTH case, but if we don't get a hit in the StringMap we do a secondary lookup in the PTHManager for the IdentifierInfo. If we don't find an IdentifierInfo we create a new one as in the non-PTH case. If we do find and IdentifierInfo in the PTHManager, we update the StringMapEntry to refer to it so that the IdentifierInfo will be found on the next StringMap lookup. This way we only do a binary search in the PTH file at most once for a given IdentifierInfo. This greatly speeds things up for source files containing a non-trivial amount of code. Performance impact: While these changes do add some extra indirection in IdentifierTable to access an IdentifierInfo*, I saw speedups even in the non-PTH case as well. Non-PTH: For -fsyntax-only on Cocoa.h, we see a 6% speedup. PTH (with Cocoa.h in token cache): 11% speedup. I also did an experiment where we did -fsyntax-only on a source file including a large header and Cocoa.h, but the token cache did not contain the larger header. For this file, we were seeing a performance *regression* when using PTH of 3% over non-PTH. Now we are seeing a performance improvement of 9%! Tests: The serialization tests are now failing. I looked at this extensively, and I my belief is that this change is unmasking a bug rather than introducing a new one. I have disabled the serialization tests for now. llvm-svn: 62636
2009-01-21 07:28:34 +08:00
IdentifierInfo *II = new ((void*) Mem) IdentifierInfo();
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
// Store the new IdentifierInfo in the cache.
PerIDCache[PersistentID] = II;
assert(II->getName() && II->getName()[0] != '\0');
return II;
}
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
IdentifierInfo* PTHManager::get(const char *NameStart, const char *NameEnd) {
PTHStringIdLookup& SL = *((PTHStringIdLookup*)StringIdLookup);
// Double check our assumption that the last character isn't '\0'.
assert(NameEnd==NameStart || NameStart[NameEnd-NameStart-1] != '\0');
PTHStringIdLookup::iterator I = SL.find(std::make_pair(NameStart,
NameEnd - NameStart));
if (I == SL.end()) // No identifier found?
return 0;
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
// Match found. Return the identifier!
assert(*I > 0);
return GetIdentifierInfo(*I-1);
}
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
PTHLexer *PTHManager::CreateLexer(FileID FID) {
const FileEntry *FE = PP->getSourceManager().getFileEntryForID(FID);
if (!FE)
return 0;
// Lookup the FileEntry object in our file lookup data structure. It will
// return a variant that indicates whether or not there is an offset within
// the PTH file that contains cached tokens.
PTHFileLookup& PFL = *((PTHFileLookup*)FileLookup);
PTHFileLookup::iterator I = PFL.find(FE);
if (I == PFL.end()) // No tokens available?
return 0;
const PTHFileData& FileData = *I;
const unsigned char *BufStart = (const unsigned char *)Buf->getBufferStart();
// Compute the offset of the token data within the buffer.
const unsigned char* data = BufStart + FileData.getTokenOffset();
// Get the location of pp-conditional table.
const unsigned char* ppcond = BufStart + FileData.getPPCondOffset();
uint32_t Len = ReadLE32(ppcond);
if (Len == 0) ppcond = 0;
IdentifierInfo: - IdentifierInfo can now (optionally) have its string data not be co-located with itself. This is for use with PTH. This aspect is a little gross, as getName() and getLength() now make assumptions about a possible alternate representation of IdentifierInfo. Perhaps we should make IdentifierInfo have virtual methods? IdentifierTable: - Added class "IdentifierInfoLookup" that can be used by IdentifierTable to perform "string -> IdentifierInfo" lookups using an auxilliary data structure. This is used by PTH. - Perform tests show that IdentifierTable::get() does not slow down because of the extra check for the IdentiferInfoLookup object (the regular StringMap lookup does enough work to mitigate the impact of an extra null pointer check). - The upshot is that now that some IdentifierInfo objects might be owned by the IdentiferInfoLookup object. This should be reviewed. PTH: - Modified PTHManager::GetIdentifierInfo to *not* insert entries in IdentifierTable's string map, and instead create IdentifierInfo objects on the fly when mapping from persistent IDs to IdentifierInfos. This saves a ton of work with string copies, hashing, and StringMap lookup and resizing. This change was motivated because when processing source files in the PTH cache we don't need to do any string -> IdentifierInfo lookups. - PTHManager now subclasses IdentifierInfoLookup, allowing clients of IdentifierTable to transparently use IdentifierInfo objects managed by the PTH file. PTHManager resolves "string -> IdentifierInfo" queries by doing a binary search over a sorted table of identifier strings in the PTH file (the exact algorithm we use can be changed as needed). These changes lead to the following performance changes when using PTH on Cocoa.h: - fsyntax-only: 10% performance improvement - Eonly: 30% performance improvement llvm-svn: 62273
2009-01-16 02:47:46 +08:00
assert(PP && "No preprocessor set yet!");
return new PTHLexer(*PP, FID, data, ppcond, *this);
}
//===----------------------------------------------------------------------===//
// 'stat' caching.
//===----------------------------------------------------------------------===//
namespace {
class VISIBILITY_HIDDEN PTHStatData {
public:
const bool hasStat;
const ino_t ino;
const dev_t dev;
const mode_t mode;
const time_t mtime;
const off_t size;
PTHStatData(ino_t i, dev_t d, mode_t mo, time_t m, off_t s)
: hasStat(true), ino(i), dev(d), mode(mo), mtime(m), size(s) {}
PTHStatData()
: hasStat(false), ino(0), dev(0), mode(0), mtime(0), size(0) {}
};
class VISIBILITY_HIDDEN PTHStatLookupTrait : public PTHFileLookupCommonTrait {
public:
typedef const char* external_key_type; // const char*
typedef PTHStatData data_type;
static internal_key_type GetInternalKey(const char *path) {
// The key 'kind' doesn't matter here because it is ignored in EqualKey.
return std::make_pair((unsigned char) 0x0, path);
}
static bool EqualKey(internal_key_type a, internal_key_type b) {
// When doing 'stat' lookups we don't care about the kind of 'a' and 'b',
// just the paths.
return strcmp(a.second, b.second) == 0;
}
static data_type ReadData(const internal_key_type& k, const unsigned char* d,
unsigned) {
if (k.first /* File or Directory */) {
if (k.first == 0x1 /* File */) d += 4 * 2; // Skip the first 2 words.
ino_t ino = (ino_t) ReadUnalignedLE32(d);
dev_t dev = (dev_t) ReadUnalignedLE32(d);
mode_t mode = (mode_t) ReadUnalignedLE16(d);
time_t mtime = (time_t) ReadUnalignedLE64(d);
return data_type(ino, dev, mode, mtime, (off_t) ReadUnalignedLE64(d));
}
// Negative stat. Don't read anything.
return data_type();
}
};
class VISIBILITY_HIDDEN PTHStatCache : public StatSysCallCache {
typedef OnDiskChainedHashTable<PTHStatLookupTrait> CacheTy;
CacheTy Cache;
public:
PTHStatCache(PTHFileLookup &FL) :
Cache(FL.getNumBuckets(), FL.getNumEntries(), FL.getBuckets(),
FL.getBase()) {}
~PTHStatCache() {}
int stat(const char *path, struct stat *buf) {
// Do the lookup for the file's data in the PTH file.
CacheTy::iterator I = Cache.find(path);
// If we don't get a hit in the PTH file just forward to 'stat'.
if (I == Cache.end()) return ::stat(path, buf);
const PTHStatData& Data = *I;
if (!Data.hasStat)
return 1;
buf->st_ino = Data.ino;
buf->st_dev = Data.dev;
buf->st_mtime = Data.mtime;
buf->st_mode = Data.mode;
buf->st_size = Data.size;
return 0;
}
};
} // end anonymous namespace
StatSysCallCache *PTHManager::createStatCache() {
return new PTHStatCache(*((PTHFileLookup*) FileLookup));
}