X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=include%2Fllvm%2FBitcode%2FBitstreamReader.h;h=779ef5fa2d839b7cb832ffe68c3c5d5a1eb7de33;hb=70c2fc0823bcc3970fdebf18ba69449d7822af86;hp=fe51ade4cb3fb182e6f77ae22fce296d33eb3c39;hpb=123f3848794b333063b478d38ff12e1bd3969125;p=oota-llvm.git diff --git a/include/llvm/Bitcode/BitstreamReader.h b/include/llvm/Bitcode/BitstreamReader.h index fe51ade4cb3..779ef5fa2d8 100644 --- a/include/llvm/Bitcode/BitstreamReader.h +++ b/include/llvm/Bitcode/BitstreamReader.h @@ -17,6 +17,7 @@ #include "llvm/Bitcode/BitCodes.h" #include +#include #include namespace llvm { @@ -24,122 +25,243 @@ namespace llvm { class Deserializer; class BitstreamReader { - const unsigned char *NextChar; - const unsigned char *LastChar; - friend class Deserializer; +public: + /// BlockInfo - This contains information emitted to BLOCKINFO_BLOCK blocks. + /// These describe abbreviations that all blocks of the specified ID inherit. + struct BlockInfo { + unsigned BlockID; + std::vector Abbrevs; + std::string Name; + + std::vector > RecordNames; + }; +private: + /// FirstChar/LastChar - This remembers the first and last bytes of the + /// stream. + const unsigned char *FirstChar, *LastChar; + + std::vector 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&); // NOT IMPLEMENTED + void operator=(const BitstreamReader&); // NOT IMPLEMENTED +public: + BitstreamReader() : FirstChar(0), LastChar(0), IgnoreBlockInfoNames(true) { + } + + BitstreamReader(const unsigned char *Start, const unsigned char *End) { + IgnoreBlockInfoNames = true; + init(Start, End); + } + + 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"); + } + + ~BitstreamReader() { + // Free the BlockInfoRecords. + while (!BlockInfoRecords.empty()) { + BlockInfo &Info = BlockInfoRecords.back(); + // Free blockinfo abbrev info. + for (unsigned i = 0, e = static_cast(Info.Abbrevs.size()); + i != e; ++i) + Info.Abbrevs[i]->dropRef(); + 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 + //===--------------------------------------------------------------------===// + + /// hasBlockInfoRecords - Return true if we've already read and processed the + /// block info block for this Bitstream. We only process it for the first + /// cursor that walks over it. + bool hasBlockInfoRecords() const { return !BlockInfoRecords.empty(); } + + /// getBlockInfo - If there is block info for the specified ID, return it, + /// otherwise return null. + const BlockInfo *getBlockInfo(unsigned BlockID) const { + // Common case, the most recent entry matches BlockID. + if (!BlockInfoRecords.empty() && BlockInfoRecords.back().BlockID == BlockID) + return &BlockInfoRecords.back(); + + for (unsigned i = 0, e = static_cast(BlockInfoRecords.size()); + i != e; ++i) + if (BlockInfoRecords[i].BlockID == BlockID) + return &BlockInfoRecords[i]; + return 0; + } + + BlockInfo &getOrCreateBlockInfo(unsigned BlockID) { + if (const BlockInfo *BI = getBlockInfo(BlockID)) + return *const_cast(BI); + + // Otherwise, add a new record. + BlockInfoRecords.push_back(BlockInfo()); + BlockInfoRecords.back().BlockID = BlockID; + return BlockInfoRecords.back(); + } + +}; + +class BitstreamCursor { + friend class Deserializer; + BitstreamReader *BitStream; + const unsigned char *NextChar; + /// CurWord - This is the current data we have pulled from the stream but have /// not returned to the client. uint32_t CurWord; - + /// BitsInCurWord - This is the number of bits in CurWord that are valid. This /// is always from [0...31] inclusive. unsigned BitsInCurWord; - + // CurCodeSize - This is the declared size of code values used for the current // block, in bits. unsigned CurCodeSize; - + /// CurAbbrevs - Abbrevs installed at in this block. std::vector CurAbbrevs; - + struct Block { unsigned PrevCodeSize; std::vector PrevAbbrevs; explicit Block(unsigned PCS) : PrevCodeSize(PCS) {} }; - + /// BlockScope - This tracks the codesize of parent blocks. SmallVector BlockScope; - - /// BlockInfo - This contains information emitted to BLOCKINFO_BLOCK blocks. - /// These describe abbreviations that all blocks of the specified ID inherit. - struct BlockInfo { - unsigned BlockID; - std::vector Abbrevs; - }; - std::vector BlockInfoRecords; - - /// FirstChar - This remembers the first byte of the stream. - const unsigned char *FirstChar; + public: - BitstreamReader() { - NextChar = FirstChar = LastChar = 0; + BitstreamCursor() : BitStream(0), NextChar(0) { + } + BitstreamCursor(const BitstreamCursor &RHS) : BitStream(0), NextChar(0) { + operator=(RHS); + } + + explicit BitstreamCursor(BitstreamReader &R) : BitStream(&R) { + NextChar = R.getFirstChar(); + assert(NextChar && "Bitstream not initialized yet"); CurWord = 0; BitsInCurWord = 0; - CurCodeSize = 0; - } - - BitstreamReader(const unsigned char *Start, const unsigned char *End) { - init(Start, End); + CurCodeSize = 2; } - - void init(const unsigned char *Start, const unsigned char *End) { - NextChar = FirstChar = Start; - LastChar = End; - assert(((End-Start) & 3) == 0 &&"Bitcode stream not a multiple of 4 bytes"); + + void init(BitstreamReader &R) { + freeState(); + + BitStream = &R; + NextChar = R.getFirstChar(); + assert(NextChar && "Bitstream not initialized yet"); CurWord = 0; BitsInCurWord = 0; CurCodeSize = 2; } - - ~BitstreamReader() { - // Abbrevs could still exist if the stream was broken. If so, don't leak - // them. + + ~BitstreamCursor() { + freeState(); + } + + void operator=(const BitstreamCursor &RHS) { + freeState(); + + BitStream = RHS.BitStream; + NextChar = RHS.NextChar; + CurWord = RHS.CurWord; + BitsInCurWord = RHS.BitsInCurWord; + CurCodeSize = RHS.CurCodeSize; + + // Copy abbreviations, and bump ref counts. + CurAbbrevs = RHS.CurAbbrevs; for (unsigned i = 0, e = static_cast(CurAbbrevs.size()); i != e; ++i) - CurAbbrevs[i]->dropRef(); - + CurAbbrevs[i]->addRef(); + + // Copy block scope and bump ref counts. for (unsigned S = 0, e = static_cast(BlockScope.size()); S != e; ++S) { std::vector &Abbrevs = BlockScope[S].PrevAbbrevs; for (unsigned i = 0, e = static_cast(Abbrevs.size()); i != e; ++i) - Abbrevs[i]->dropRef(); + Abbrevs[i]->addRef(); } - - // Free the BlockInfoRecords. - while (!BlockInfoRecords.empty()) { - BlockInfo &Info = BlockInfoRecords.back(); - // Free blockinfo abbrev info. - for (unsigned i = 0, e = static_cast(Info.Abbrevs.size()); + } + + void freeState() { + // Free all the Abbrevs. + for (unsigned i = 0, e = static_cast(CurAbbrevs.size()); + i != e; ++i) + CurAbbrevs[i]->dropRef(); + CurAbbrevs.clear(); + + // Free all the Abbrevs in the block scope. + for (unsigned S = 0, e = static_cast(BlockScope.size()); + S != e; ++S) { + std::vector &Abbrevs = BlockScope[S].PrevAbbrevs; + for (unsigned i = 0, e = static_cast(Abbrevs.size()); i != e; ++i) - Info.Abbrevs[i]->dropRef(); - BlockInfoRecords.pop_back(); + Abbrevs[i]->dropRef(); } + BlockScope.clear(); } - + + /// GetAbbrevIDWidth - Return the number of bits used to encode an abbrev #. + unsigned GetAbbrevIDWidth() const { return CurCodeSize; } + bool AtEndOfStream() const { - return NextChar == LastChar && BitsInCurWord == 0; + return NextChar == BitStream->getLastChar() && BitsInCurWord == 0; } - + /// GetCurrentBitNo - Return the bit # of the bit we are reading. uint64_t GetCurrentBitNo() const { - return (NextChar-FirstChar)*CHAR_BIT + ((32-BitsInCurWord) & 31); + return (NextChar-BitStream->getFirstChar())*CHAR_BIT - BitsInCurWord; } - + + BitstreamReader *getBitStreamReader() { + return BitStream; + } + const BitstreamReader *getBitStreamReader() const { + return BitStream; + } + + /// JumpToBit - Reset the stream to the specified bit number. void JumpToBit(uint64_t BitNo) { uintptr_t ByteNo = uintptr_t(BitNo/8) & ~3; uintptr_t WordBitNo = uintptr_t(BitNo) & 31; - assert(ByteNo < (uintptr_t)(LastChar-FirstChar) && "Invalid location"); - + assert(ByteNo <= (uintptr_t)(BitStream->getLastChar()- + BitStream->getFirstChar()) && + "Invalid location"); + // Move the cursor to the right word. - NextChar = FirstChar+ByteNo; + NextChar = BitStream->getFirstChar()+ByteNo; BitsInCurWord = 0; CurWord = 0; - + // Skip over any bits that are already consumed. - if (WordBitNo) { - NextChar -= 4; + if (WordBitNo) Read(static_cast(WordBitNo)); - } } - - /// GetAbbrevIDWidth - Return the number of bits used to encode an abbrev #. - unsigned GetAbbrevIDWidth() const { return CurCodeSize; } - + + 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); @@ -149,7 +271,7 @@ public: } // If we run out of data, stop at the end of the stream. - if (NextChar == LastChar) { + if (NextChar == BitStream->getLastChar()) { CurWord = 0; BitsInCurWord = 0; return 0; @@ -202,15 +324,17 @@ public: } } + // 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; @@ -225,31 +349,10 @@ public: CurWord = 0; } - unsigned ReadCode() { return Read(CurCodeSize); } - //===--------------------------------------------------------------------===// - // Block Manipulation - //===--------------------------------------------------------------------===// - -private: - /// getBlockInfo - If there is block info for the specified ID, return it, - /// otherwise return null. - BlockInfo *getBlockInfo(unsigned BlockID) { - // Common case, the most recent entry matches BlockID. - if (!BlockInfoRecords.empty() && BlockInfoRecords.back().BlockID == BlockID) - return &BlockInfoRecords.back(); - - for (unsigned i = 0, e = static_cast(BlockInfoRecords.size()); - i != e; ++i) - if (BlockInfoRecords[i].BlockID == BlockID) - return &BlockInfoRecords[i]; - return 0; - } -public: - // Block header: // [ENTER_SUBBLOCK, blockid, newcodelen, , blocklen] @@ -272,7 +375,7 @@ public: // Check that the block wasn't partially defined, and that the offset isn't // bogus. - if (AtEndOfStream() || NextChar+NumWords*4 > LastChar) + if (AtEndOfStream() || NextChar+NumWords*4 > BitStream->getLastChar()) return true; NextChar += NumWords*4; @@ -287,7 +390,8 @@ public: BlockScope.back().PrevAbbrevs.swap(CurAbbrevs); // Add the abbrevs specific to this block to the CurAbbrevs list. - if (BlockInfo *Info = getBlockInfo(BlockID)) { + if (const BitstreamReader::BlockInfo *Info = + BitStream->getBlockInfo(BlockID)) { for (unsigned i = 0, e = static_cast(Info->Abbrevs.size()); i != e; ++i) { CurAbbrevs.push_back(Info->Abbrevs[i]); @@ -302,7 +406,8 @@ public: if (NumWordsP) *NumWordsP = NumWords; // Validate that this block is sane. - if (CurCodeSize == 0 || AtEndOfStream() || NextChar+NumWords*4 > LastChar) + if (CurCodeSize == 0 || AtEndOfStream() || + NextChar+NumWords*4 > BitStream->getLastChar()) return true; return false; @@ -332,7 +437,7 @@ private: BlockScope.pop_back(); } - //===--------------------------------------------------------------------===// + //===--------------------------------------------------------------------===// // Record Processing //===--------------------------------------------------------------------===// @@ -371,7 +476,8 @@ public: return CurAbbrevs[AbbrevNo]; } - unsigned ReadRecord(unsigned AbbrevID, SmallVectorImpl &Vals) { + unsigned ReadRecord(unsigned AbbrevID, SmallVectorImpl &Vals, + const char **BlobStart = 0, unsigned *BlobLen = 0) { if (AbbrevID == bitc::UNABBREV_RECORD) { unsigned Code = ReadVBR(6); unsigned NumElts = ReadVBR(6); @@ -407,15 +513,21 @@ public: // If this would read off the end of the bitcode file, just set the // record to empty and return. - if (NewEnd > LastChar) { + if (NewEnd > BitStream->getLastChar()) { Vals.append(NumElts, 0); - NextChar = LastChar; + NextChar = BitStream->getLastChar(); break; } - // Otherwise, read the number of bytes. - for (; NumElts; ++NextChar, --NumElts) - Vals.push_back(*NextChar); + // 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; + *BlobLen = NumElts; + } else { + for (; NumElts; ++NextChar, --NumElts) + Vals.push_back(*NextChar); + } // Skip over tail padding. NextChar = NewEnd; } else { @@ -428,6 +540,12 @@ public: return Code; } + unsigned ReadRecord(unsigned AbbrevID, SmallVectorImpl &Vals, + const char *&BlobStart, unsigned &BlobLen) { + return ReadRecord(AbbrevID, Vals, &BlobStart, &BlobLen); + } + + //===--------------------------------------------------------------------===// // Abbrev Processing //===--------------------------------------------------------------------===// @@ -450,29 +568,18 @@ public: } CurAbbrevs.push_back(Abbv); } - - //===--------------------------------------------------------------------===// - // BlockInfo Block Reading - //===--------------------------------------------------------------------===// - -private: - BlockInfo &getOrCreateBlockInfo(unsigned BlockID) { - if (BlockInfo *BI = getBlockInfo(BlockID)) - return *BI; - - // Otherwise, add a new record. - BlockInfoRecords.push_back(BlockInfo()); - BlockInfoRecords.back().BlockID = BlockID; - return BlockInfoRecords.back(); - } - + public: bool ReadBlockInfoBlock() { + // If this is the second stream to get to the block info block, skip it. + if (BitStream->hasBlockInfoRecords()) + return SkipBlock(); + if (EnterSubBlock(bitc::BLOCKINFO_BLOCK_ID)) return true; SmallVector Record; - BlockInfo *CurBlockInfo = 0; + BitstreamReader::BlockInfo *CurBlockInfo = 0; // Read all the records for this module. while (1) { @@ -504,13 +611,32 @@ public: default: break; // Default behavior, ignore unknown content. case bitc::BLOCKINFO_CODE_SETBID: if (Record.size() < 1) return true; - CurBlockInfo = &getOrCreateBlockInfo((unsigned)Record[0]); + 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; + } + } } } }; - + } // End llvm namespace #endif