#ifndef BITSTREAM_READER_H
#define BITSTREAM_READER_H
+#include "llvm/ADT/OwningPtr.h"
#include "llvm/Bitcode/BitCodes.h"
#include <climits>
+#include <string>
#include <vector>
+#include "llvm/Support/StreamableMemoryObject.h"
namespace llvm {
struct BlockInfo {
unsigned BlockID;
std::vector<BitCodeAbbrev*> Abbrevs;
+ std::string Name;
+
+ std::vector<std::pair<unsigned, std::string> > RecordNames;
};
private:
- /// FirstChar/LastChar - This remembers the first and last bytes of the
- /// stream.
- const unsigned char *FirstChar, *LastChar;
+ OwningPtr<StreamableMemoryObject> BitcodeBytes;
std::vector<BlockInfo> BlockInfoRecords;
+ /// IgnoreBlockInfoNames - This is set to true if we don't care about the
+ /// block/record name information in the BlockInfo block. Only llvm-bcanalyzer
+ /// uses this.
+ bool IgnoreBlockInfoNames;
+
+ BitstreamReader(const BitstreamReader&); // DO NOT IMPLEMENT
+ void operator=(const BitstreamReader&); // DO NOT IMPLEMENT
public:
- BitstreamReader() : FirstChar(0), LastChar(0) {
+ BitstreamReader() : IgnoreBlockInfoNames(true) {
}
BitstreamReader(const unsigned char *Start, const unsigned char *End) {
+ IgnoreBlockInfoNames = true;
init(Start, End);
}
+ BitstreamReader(StreamableMemoryObject *bytes) {
+ BitcodeBytes.reset(bytes);
+ }
+
void init(const unsigned char *Start, const unsigned char *End) {
- FirstChar = Start;
- LastChar = End;
assert(((End-Start) & 3) == 0 &&"Bitcode stream not a multiple of 4 bytes");
+ BitcodeBytes.reset(getNonStreamedMemoryObject(Start, End));
}
+ StreamableMemoryObject &getBitcodeBytes() { return *BitcodeBytes; }
+
~BitstreamReader() {
// Free the BlockInfoRecords.
while (!BlockInfoRecords.empty()) {
BlockInfoRecords.pop_back();
}
}
-
- const unsigned char *getFirstChar() const { return FirstChar; }
- const unsigned char *getLastChar() const { return LastChar; }
+ /// CollectBlockInfoNames - This is called by clients that want block/record
+ /// name information.
+ void CollectBlockInfoNames() { IgnoreBlockInfoNames = false; }
+ bool isIgnoringBlockInfoNames() { return IgnoreBlockInfoNames; }
+
//===--------------------------------------------------------------------===//
// Block Manipulation
//===--------------------------------------------------------------------===//
/// getBlockInfo - If there is block info for the specified ID, return it,
/// otherwise return null.
- BlockInfo *getBlockInfo(unsigned BlockID) {
+ const BlockInfo *getBlockInfo(unsigned BlockID) const {
// Common case, the most recent entry matches BlockID.
if (!BlockInfoRecords.empty() && BlockInfoRecords.back().BlockID == BlockID)
return &BlockInfoRecords.back();
}
BlockInfo &getOrCreateBlockInfo(unsigned BlockID) {
- if (BlockInfo *BI = getBlockInfo(BlockID))
- return *BI;
+ if (const BlockInfo *BI = getBlockInfo(BlockID))
+ return *const_cast<BlockInfo*>(BI);
// Otherwise, add a new record.
BlockInfoRecords.push_back(BlockInfo());
class BitstreamCursor {
friend class Deserializer;
BitstreamReader *BitStream;
- const unsigned char *NextChar;
+ size_t NextChar;
/// CurWord - This is the current data we have pulled from the stream but have
/// not returned to the client.
}
explicit BitstreamCursor(BitstreamReader &R) : BitStream(&R) {
- NextChar = R.getFirstChar();
- assert(NextChar && "Bitstream not initialized yet");
+ NextChar = 0;
CurWord = 0;
BitsInCurWord = 0;
CurCodeSize = 2;
freeState();
BitStream = &R;
- NextChar = R.getFirstChar();
- assert(NextChar && "Bitstream not initialized yet");
+ NextChar = 0;
CurWord = 0;
BitsInCurWord = 0;
CurCodeSize = 2;
CurAbbrevs[i]->addRef();
// Copy block scope and bump ref counts.
+ BlockScope = RHS.BlockScope;
for (unsigned S = 0, e = static_cast<unsigned>(BlockScope.size());
S != e; ++S) {
std::vector<BitCodeAbbrev*> &Abbrevs = BlockScope[S].PrevAbbrevs;
/// GetAbbrevIDWidth - Return the number of bits used to encode an abbrev #.
unsigned GetAbbrevIDWidth() const { return CurCodeSize; }
- bool AtEndOfStream() const {
- return NextChar == BitStream->getLastChar() && BitsInCurWord == 0;
+ bool isEndPos(size_t pos) {
+ return BitStream->getBitcodeBytes().isObjectEnd(static_cast<uint64_t>(pos));
+ }
+
+ bool canSkipToPos(size_t pos) const {
+ // pos can be skipped to if it is a valid address or one byte past the end.
+ return pos == 0 || BitStream->getBitcodeBytes().isValidAddress(
+ static_cast<uint64_t>(pos - 1));
+ }
+
+ unsigned char getByte(size_t pos) {
+ uint8_t byte = -1;
+ BitStream->getBitcodeBytes().readByte(pos, &byte);
+ return byte;
+ }
+
+ uint32_t getWord(size_t pos) {
+ uint32_t word = -1;
+ BitStream->getBitcodeBytes().readBytes(pos,
+ sizeof(word),
+ reinterpret_cast<uint8_t *>(&word),
+ NULL);
+ return word;
+ }
+
+ bool AtEndOfStream() {
+ return isEndPos(NextChar) && BitsInCurWord == 0;
}
/// GetCurrentBitNo - Return the bit # of the bit we are reading.
uint64_t GetCurrentBitNo() const {
- return (NextChar-BitStream->getFirstChar())*CHAR_BIT - BitsInCurWord;
+ return NextChar*CHAR_BIT - BitsInCurWord;
+ }
+
+ BitstreamReader *getBitStreamReader() {
+ return BitStream;
+ }
+ const BitstreamReader *getBitStreamReader() const {
+ return BitStream;
}
void JumpToBit(uint64_t BitNo) {
uintptr_t ByteNo = uintptr_t(BitNo/8) & ~3;
uintptr_t WordBitNo = uintptr_t(BitNo) & 31;
- assert(ByteNo <= (uintptr_t)(BitStream->getLastChar()-
- BitStream->getFirstChar()) &&
- "Invalid location");
+ assert(canSkipToPos(ByteNo) && "Invalid location");
// Move the cursor to the right word.
- NextChar = BitStream->getFirstChar()+ByteNo;
+ NextChar = ByteNo;
BitsInCurWord = 0;
CurWord = 0;
uint32_t Read(unsigned NumBits) {
+ assert(NumBits <= 32 && "Cannot return more than 32 bits!");
// If the field is fully contained by CurWord, return it quickly.
if (BitsInCurWord >= NumBits) {
uint32_t R = CurWord & ((1U << NumBits)-1);
}
// If we run out of data, stop at the end of the stream.
- if (NextChar == BitStream->getLastChar()) {
+ if (isEndPos(NextChar)) {
CurWord = 0;
BitsInCurWord = 0;
return 0;
unsigned R = CurWord;
// Read the next word from the stream.
- CurWord = (NextChar[0] << 0) | (NextChar[1] << 8) |
- (NextChar[2] << 16) | (NextChar[3] << 24);
+ CurWord = getWord(NextChar);
NextChar += 4;
// Extract NumBits-BitsInCurWord from what we just read.
}
}
+ // ReadVBR64 - Read a VBR that may have a value up to 64-bits in size. The
+ // chunk size of the VBR must still be <= 32 bits though.
uint64_t ReadVBR64(unsigned NumBits) {
- uint64_t Piece = Read(NumBits);
+ uint32_t Piece = Read(NumBits);
if ((Piece & (1U << (NumBits-1))) == 0)
- return Piece;
+ return uint64_t(Piece);
uint64_t Result = 0;
unsigned NextBit = 0;
while (1) {
- Result |= (Piece & ((1U << (NumBits-1))-1)) << NextBit;
+ Result |= uint64_t(Piece & ((1U << (NumBits-1))-1)) << NextBit;
if ((Piece & (1U << (NumBits-1))) == 0)
return Result;
// Check that the block wasn't partially defined, and that the offset isn't
// bogus.
- if (AtEndOfStream() || NextChar+NumWords*4 > BitStream->getLastChar())
+ size_t SkipTo = NextChar + NumWords*4;
+ if (AtEndOfStream() || !canSkipToPos(SkipTo))
return true;
- NextChar += NumWords*4;
+ NextChar = SkipTo;
return false;
}
BlockScope.back().PrevAbbrevs.swap(CurAbbrevs);
// Add the abbrevs specific to this block to the CurAbbrevs list.
- if (BitstreamReader::BlockInfo *Info = BitStream->getBlockInfo(BlockID)) {
+ if (const BitstreamReader::BlockInfo *Info =
+ BitStream->getBlockInfo(BlockID)) {
for (unsigned i = 0, e = static_cast<unsigned>(Info->Abbrevs.size());
i != e; ++i) {
CurAbbrevs.push_back(Info->Abbrevs[i]);
if (NumWordsP) *NumWordsP = NumWords;
// Validate that this block is sane.
- if (CurCodeSize == 0 || AtEndOfStream() ||
- NextChar+NumWords*4 > BitStream->getLastChar())
+ if (CurCodeSize == 0 || AtEndOfStream())
return true;
return false;
void ReadAbbreviatedField(const BitCodeAbbrevOp &Op,
SmallVectorImpl<uint64_t> &Vals) {
assert(!Op.isLiteral() && "Use ReadAbbreviatedLiteral for literals!");
-
+
// Decode the value as we are commanded.
switch (Op.getEncoding()) {
- default: assert(0 && "Unknown encoding!");
+ default: llvm_unreachable("Unknown encoding!");
case BitCodeAbbrevOp::Fixed:
Vals.push_back(Read((unsigned)Op.getEncodingData()));
break;
SkipToWord(); // 32-bit alignment
// Figure out where the end of this blob will be including tail padding.
- const unsigned char *NewEnd = NextChar+((NumElts+3)&~3);
+ size_t NewEnd = NextChar+((NumElts+3)&~3);
// If this would read off the end of the bitcode file, just set the
// record to empty and return.
- if (NewEnd > BitStream->getLastChar()) {
+ if (!canSkipToPos(NewEnd)) {
Vals.append(NumElts, 0);
- NextChar = BitStream->getLastChar();
+ NextChar = BitStream->getBitcodeBytes().getExtent();
break;
}
// Otherwise, read the number of bytes. If we can return a reference to
// the data, do so to avoid copying it.
if (BlobStart) {
- *BlobStart = (const char*)NextChar;
+ *BlobStart = (const char*)BitStream->getBitcodeBytes().getPointer(
+ NextChar, NumElts);
*BlobLen = NumElts;
} else {
for (; NumElts; ++NextChar, --NumElts)
- Vals.push_back(*NextChar);
+ Vals.push_back(getByte(NextChar));
}
// Skip over tail padding.
NextChar = NewEnd;
if (Record.size() < 1) return true;
CurBlockInfo = &BitStream->getOrCreateBlockInfo((unsigned)Record[0]);
break;
+ case bitc::BLOCKINFO_CODE_BLOCKNAME: {
+ if (!CurBlockInfo) return true;
+ if (BitStream->isIgnoringBlockInfoNames()) break; // Ignore name.
+ std::string Name;
+ for (unsigned i = 0, e = Record.size(); i != e; ++i)
+ Name += (char)Record[i];
+ CurBlockInfo->Name = Name;
+ break;
+ }
+ case bitc::BLOCKINFO_CODE_SETRECORDNAME: {
+ if (!CurBlockInfo) return true;
+ if (BitStream->isIgnoringBlockInfoNames()) break; // Ignore name.
+ std::string Name;
+ for (unsigned i = 1, e = Record.size(); i != e; ++i)
+ Name += (char)Record[i];
+ CurBlockInfo->RecordNames.push_back(std::make_pair((unsigned)Record[0],
+ Name));
+ break;
+ }
}
}
}
projects / oota-llvm.git / blobdiff