bool GlobalModuleIndexBuilder::loadModuleFile(const FileEntry *File) {
// Open the module file.
OwningPtr<llvm::MemoryBuffer> Buffer;
Buffer.reset(FileMgr.getBufferForFile(File));
if (!Buffer) {
return true;
}
// Initialize the input stream
llvm::BitstreamReader InStreamFile;
llvm::BitstreamCursor InStream;
InStreamFile.init((const unsigned char *)Buffer->getBufferStart(),
(const unsigned char *)Buffer->getBufferEnd());
InStream.init(InStreamFile);
// Sniff for the signature.
if (InStream.Read(8) != 'C' ||
InStream.Read(8) != 'P' ||
InStream.Read(8) != 'C' ||
InStream.Read(8) != 'H') {
return true;
}
// Record this module file and assign it a unique ID (if it doesn't have
// one already).
unsigned ID = getModuleFileInfo(File).ID;
// Search for the blocks and records we care about.
enum { Other, ControlBlock, ASTBlock } State = Other;
bool Done = false;
while (!Done) {
llvm::BitstreamEntry Entry = InStream.advance();
switch (Entry.Kind) {
case llvm::BitstreamEntry::Error:
Done = true;
continue;
case llvm::BitstreamEntry::Record:
// In the 'other' state, just skip the record. We don't care.
if (State == Other) {
InStream.skipRecord(Entry.ID);
continue;
}
// Handle potentially-interesting records below.
break;
case llvm::BitstreamEntry::SubBlock:
if (Entry.ID == CONTROL_BLOCK_ID) {
if (InStream.EnterSubBlock(CONTROL_BLOCK_ID))
return true;
// Found the control block.
State = ControlBlock;
continue;
}
if (Entry.ID == AST_BLOCK_ID) {
if (InStream.EnterSubBlock(AST_BLOCK_ID))
return true;
// Found the AST block.
State = ASTBlock;
continue;
}
if (InStream.SkipBlock())
return true;
continue;
case llvm::BitstreamEntry::EndBlock:
State = Other;
continue;
}
// Read the given record.
SmallVector<uint64_t, 64> Record;
StringRef Blob;
unsigned Code = InStream.readRecord(Entry.ID, Record, &Blob);
// Handle module dependencies.
if (State == ControlBlock && Code == IMPORTS) {
// Load each of the imported PCH files.
unsigned Idx = 0, N = Record.size();
while (Idx < N) {
// Read information about the AST file.
// Skip the imported kind
++Idx;
// Skip the import location
++Idx;
// Retrieve the imported file name.
unsigned Length = Record[Idx++];
SmallString<128> ImportedFile(Record.begin() + Idx,
Record.begin() + Idx + Length);
Idx += Length;
// Find the imported module file.
const FileEntry *DependsOnFile = FileMgr.getFile(ImportedFile);
if (!DependsOnFile)
return true;
// Record the dependency.
unsigned DependsOnID = getModuleFileInfo(DependsOnFile).ID;
getModuleFileInfo(File).Dependencies.push_back(DependsOnID);
}
continue;
}
// Handle the identifier table
if (State == ASTBlock && Code == IDENTIFIER_TABLE && Record[0] > 0) {
typedef OnDiskChainedHashTable<InterestingASTIdentifierLookupTrait>
InterestingIdentifierTable;
llvm::OwningPtr<InterestingIdentifierTable>
Table(InterestingIdentifierTable::Create(
(const unsigned char *)Blob.data() + Record[0],
(const unsigned char *)Blob.data()));
for (InterestingIdentifierTable::data_iterator D = Table->data_begin(),
DEnd = Table->data_end();
D != DEnd; ++D) {
std::pair<StringRef, bool> Ident = *D;
if (Ident.second)
InterestingIdentifiers[Ident.first].push_back(ID);
else
(void)InterestingIdentifiers[Ident.first];
}
}
// FIXME: Handle the selector table.
// We don't care about this record.
}
return false;
}
void BenchmarkIRParsing() {
outs() << "Benchmarking IR parsing...\n";
OwningPtr<MemoryBuffer> FileBuf;
error_code ec = MemoryBuffer::getFileOrSTDIN(InputFilename.c_str(), FileBuf);
if (ec) {
report_fatal_error("Could not open input file: " + ec.message());
}
size_t BufSize = FileBuf->getBufferSize();
const uint8_t *BufPtr =
reinterpret_cast<const uint8_t*>(FileBuf->getBufferStart());
const uint8_t *EndBufPtr =
reinterpret_cast<const uint8_t*>(FileBuf->getBufferEnd());
// Since MemoryBuffer may use mmap, make sure to first touch all bytes in the
// input buffer to make sure it's actually in memory.
volatile uint8_t *Slot = new uint8_t;
for (const uint8_t *S = BufPtr; S != EndBufPtr; ++S) {
*Slot = *S;
}
delete Slot;
outs() << "Read bitcode into buffer. Size=" << BufSize << "\n";
// Trivial copy into a new buffer with a cascading XOR that simulates
// "touching" every byte in the buffer in a simple way.
{
TimingOperationBlock T("Simple XOR copy", BufSize);
volatile uint8_t *OutBuf = new uint8_t[BufSize];
OutBuf[0] = 1;
size_t N = 1;
// Run over the input buffer from start to end-1; run over the output buffer
// from 1 to end.
for (const uint8_t *S = BufPtr; S != EndBufPtr - 1; ++S, ++N) {
OutBuf[N] = OutBuf[N - 1] ^ *S;
}
delete[] OutBuf;
}
// Bitcode parsing without any additional operations. This is the minimum
// required to actually extract information from PNaCl bitcode.
{
TimingOperationBlock T("Bitcode block parsing", BufSize);
NaClBitcodeHeader Header;
if (Header.Read(BufPtr, EndBufPtr)) {
report_fatal_error("Invalid PNaCl bitcode header");
}
if (!Header.IsSupported()) {
errs() << "Warning: " << Header.Unsupported() << "\n";
}
if (!Header.IsReadable()) {
report_fatal_error("Bitcode file is not readable");
}
NaClBitstreamReader StreamFile(BufPtr, EndBufPtr);
NaClBitstreamCursor Stream(StreamFile);
StreamFile.CollectBlockInfoNames();
DummyBitcodeParser Parser(Stream);
while (!Stream.AtEndOfStream()) {
if (Parser.Parse()) {
report_fatal_error("Parsing failed");
}
}
}
// Actual LLVM IR parsing and formation from the bitcode
{
TimingOperationBlock T("LLVM IR parsing", BufSize);
SMDiagnostic Err;
Module *M = NaClParseIRFile(InputFilename, PNaClFormat,
Err, getGlobalContext());
if (!M) {
report_fatal_error("Unable to NaClParseIRFile");
}
}
}
CXDiagnosticSet DiagLoader::load(const char *file) {
// Open the diagnostics file.
std::string ErrStr;
FileSystemOptions FO;
FileManager FileMgr(FO);
OwningPtr<llvm::MemoryBuffer> Buffer;
Buffer.reset(FileMgr.getBufferForFile(file));
if (!Buffer) {
reportBad(CXLoadDiag_CannotLoad, ErrStr);
return 0;
}
llvm::BitstreamReader StreamFile;
StreamFile.init((const unsigned char *)Buffer->getBufferStart(),
(const unsigned char *)Buffer->getBufferEnd());
llvm::BitstreamCursor Stream;
Stream.init(StreamFile);
// Sniff for the signature.
if (Stream.Read(8) != 'D' ||
Stream.Read(8) != 'I' ||
Stream.Read(8) != 'A' ||
Stream.Read(8) != 'G') {
reportBad(CXLoadDiag_InvalidFile,
"Bad header in diagnostics file");
return 0;
}
OwningPtr<CXLoadedDiagnosticSetImpl> Diags(new CXLoadedDiagnosticSetImpl());
while (true) {
unsigned BlockID = 0;
StreamResult Res = readToNextRecordOrBlock(Stream, "Top-level",
BlockID, true);
switch (Res) {
case Read_EndOfStream:
return (CXDiagnosticSet) Diags.take();
case Read_Failure:
return 0;
case Read_Record:
llvm_unreachable("Top-level does not have records");
case Read_BlockEnd:
continue;
case Read_BlockBegin:
break;
}
switch (BlockID) {
case serialized_diags::BLOCK_META:
if (readMetaBlock(Stream))
return 0;
break;
case serialized_diags::BLOCK_DIAG:
if (readDiagnosticBlock(Stream, *Diags.get(), *Diags.get()))
return 0;
break;
default:
if (!Stream.SkipBlock()) {
reportInvalidFile("Malformed block at top-level of diagnostics file");
return 0;
}
break;
}
}
}
error_code MemoryBuffer::getOpenFile(int FD, const char *Filename,
OwningPtr<MemoryBuffer> &result,
uint64_t FileSize, uint64_t MapSize,
int64_t Offset,
bool RequiresNullTerminator) {
static int PageSize = sys::Process::GetPageSize();
// Default is to map the full file.
if (MapSize == uint64_t(-1)) {
// If we don't know the file size, use fstat to find out. fstat on an open
// file descriptor is cheaper than stat on a random path.
if (FileSize == uint64_t(-1)) {
struct stat FileInfo;
// TODO: This should use fstat64 when available.
if (fstat(FD, &FileInfo) == -1) {
return error_code(errno, posix_category());
}
FileSize = FileInfo.st_size;
}
MapSize = FileSize;
}
if (shouldUseMmap(FD, FileSize, MapSize, Offset, RequiresNullTerminator,
PageSize)) {
off_t RealMapOffset = Offset & ~(PageSize - 1);
off_t Delta = Offset - RealMapOffset;
size_t RealMapSize = MapSize + Delta;
if (const char *Pages = sys::Path::MapInFilePages(FD,
RealMapSize,
RealMapOffset)) {
result.reset(GetNamedBuffer<MemoryBufferMMapFile>(
StringRef(Pages + Delta, MapSize), Filename, RequiresNullTerminator));
if (RequiresNullTerminator && result->getBufferEnd()[0] != '\0') {
// There could be a racing issue that resulted in the file being larger
// than the FileSize passed by the caller. We already have an assertion
// for this in MemoryBuffer::init() but have a runtime guarantee that
// the buffer will be null-terminated here, so do a copy that adds a
// null-terminator.
result.reset(MemoryBuffer::getMemBufferCopy(result->getBuffer(),
Filename));
}
return error_code::success();
}
}
MemoryBuffer *Buf = MemoryBuffer::getNewUninitMemBuffer(MapSize, Filename);
if (!Buf) {
// Failed to create a buffer. The only way it can fail is if
// new(std::nothrow) returns 0.
return make_error_code(errc::not_enough_memory);
}
OwningPtr<MemoryBuffer> SB(Buf);
char *BufPtr = const_cast<char*>(SB->getBufferStart());
size_t BytesLeft = MapSize;
#ifndef HAVE_PREAD
if (lseek(FD, Offset, SEEK_SET) == -1)
return error_code(errno, posix_category());
#endif
while (BytesLeft) {
#ifdef HAVE_PREAD
ssize_t NumRead = ::pread(FD, BufPtr, BytesLeft, MapSize-BytesLeft+Offset);
#else
ssize_t NumRead = ::read(FD, BufPtr, BytesLeft);
#endif
if (NumRead == -1) {
if (errno == EINTR)
continue;
// Error while reading.
return error_code(errno, posix_category());
}
if (NumRead == 0) {
assert(0 && "We got inaccurate FileSize value or fstat reported an "
"invalid file size.");
*BufPtr = '\0'; // null-terminate at the actual size.
break;
}
BytesLeft -= NumRead;
BufPtr += NumRead;
}
result.swap(SB);
return error_code::success();
}