1 //===- BitCodes.h - Enum values for the bitcode format ----------*- C++ -*-===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 // 10 // This header Bitcode enum values. 11 // 12 // The enum values defined in this file should be considered permanent. If 13 // new features are added, they should have values added at the end of the 14 // respective lists. 15 // 16 //===----------------------------------------------------------------------===// 17 18 #ifndef LLVM_BITCODE_BITCODES_H 19 #define LLVM_BITCODE_BITCODES_H 20 21 #include "llvm/ADT/IntrusiveRefCntPtr.h" 22 #include "llvm/ADT/SmallVector.h" 23 #include "llvm/Support/DataTypes.h" 24 #include "llvm/Support/ErrorHandling.h" 25 #include <cassert> 26 27 namespace llvm { 28 namespace bitc { 29 enum StandardWidths { 30 BlockIDWidth = 8, // We use VBR-8 for block IDs. 31 CodeLenWidth = 4, // Codelen are VBR-4. 32 BlockSizeWidth = 32 // BlockSize up to 2^32 32-bit words = 16GB per block. 33 }; 34 35 // The standard abbrev namespace always has a way to exit a block, enter a 36 // nested block, define abbrevs, and define an unabbreviated record. 37 enum FixedAbbrevIDs { 38 END_BLOCK = 0, // Must be zero to guarantee termination for broken bitcode. 39 ENTER_SUBBLOCK = 1, 40 41 /// DEFINE_ABBREV - Defines an abbrev for the current block. It consists 42 /// of a vbr5 for # operand infos. Each operand info is emitted with a 43 /// single bit to indicate if it is a literal encoding. If so, the value is 44 /// emitted with a vbr8. If not, the encoding is emitted as 3 bits followed 45 /// by the info value as a vbr5 if needed. 46 DEFINE_ABBREV = 2, 47 48 // UNABBREV_RECORDs are emitted with a vbr6 for the record code, followed by 49 // a vbr6 for the # operands, followed by vbr6's for each operand. 50 UNABBREV_RECORD = 3, 51 52 // This is not a code, this is a marker for the first abbrev assignment. 53 FIRST_APPLICATION_ABBREV = 4 54 }; 55 56 /// StandardBlockIDs - All bitcode files can optionally include a BLOCKINFO 57 /// block, which contains metadata about other blocks in the file. 58 enum StandardBlockIDs { 59 /// BLOCKINFO_BLOCK is used to define metadata about blocks, for example, 60 /// standard abbrevs that should be available to all blocks of a specified 61 /// ID. 62 BLOCKINFO_BLOCK_ID = 0, 63 64 // Block IDs 1-7 are reserved for future expansion. 65 FIRST_APPLICATION_BLOCKID = 8 66 }; 67 68 /// BlockInfoCodes - The blockinfo block contains metadata about user-defined 69 /// blocks. 70 enum BlockInfoCodes { 71 // DEFINE_ABBREV has magic semantics here, applying to the current SETBID'd 72 // block, instead of the BlockInfo block. 73 74 BLOCKINFO_CODE_SETBID = 1, // SETBID: [blockid#] 75 BLOCKINFO_CODE_BLOCKNAME = 2, // BLOCKNAME: [name] 76 BLOCKINFO_CODE_SETRECORDNAME = 3 // BLOCKINFO_CODE_SETRECORDNAME: 77 // [id, name] 78 }; 79 80 } // End bitc namespace 81 82 /// BitCodeAbbrevOp - This describes one or more operands in an abbreviation. 83 /// This is actually a union of two different things: 84 /// 1. It could be a literal integer value ("the operand is always 17"). 85 /// 2. It could be an encoding specification ("this operand encoded like so"). 86 /// 87 class BitCodeAbbrevOp { 88 uint64_t Val; // A literal value or data for an encoding. 89 bool IsLiteral : 1; // Indicate whether this is a literal value or not. 90 unsigned Enc : 3; // The encoding to use. 91 public: 92 enum Encoding { 93 Fixed = 1, // A fixed width field, Val specifies number of bits. 94 VBR = 2, // A VBR field where Val specifies the width of each chunk. 95 Array = 3, // A sequence of fields, next field species elt encoding. 96 Char6 = 4, // A 6-bit fixed field which maps to [a-zA-Z0-9._]. 97 Blob = 5 // 32-bit aligned array of 8-bit characters. 98 }; 99 BitCodeAbbrevOp(uint64_t V)100 explicit BitCodeAbbrevOp(uint64_t V) : Val(V), IsLiteral(true) {} 101 explicit BitCodeAbbrevOp(Encoding E, uint64_t Data = 0) Val(Data)102 : Val(Data), IsLiteral(false), Enc(E) {} 103 isLiteral()104 bool isLiteral() const { return IsLiteral; } isEncoding()105 bool isEncoding() const { return !IsLiteral; } 106 107 // Accessors for literals. getLiteralValue()108 uint64_t getLiteralValue() const { assert(isLiteral()); return Val; } 109 110 // Accessors for encoding info. getEncoding()111 Encoding getEncoding() const { assert(isEncoding()); return (Encoding)Enc; } getEncodingData()112 uint64_t getEncodingData() const { 113 assert(isEncoding() && hasEncodingData()); 114 return Val; 115 } 116 hasEncodingData()117 bool hasEncodingData() const { return hasEncodingData(getEncoding()); } hasEncodingData(Encoding E)118 static bool hasEncodingData(Encoding E) { 119 switch (E) { 120 case Fixed: 121 case VBR: 122 return true; 123 case Array: 124 case Char6: 125 case Blob: 126 return false; 127 } 128 report_fatal_error("Invalid encoding"); 129 } 130 131 /// isChar6 - Return true if this character is legal in the Char6 encoding. isChar6(char C)132 static bool isChar6(char C) { 133 if (C >= 'a' && C <= 'z') return true; 134 if (C >= 'A' && C <= 'Z') return true; 135 if (C >= '0' && C <= '9') return true; 136 if (C == '.' || C == '_') return true; 137 return false; 138 } EncodeChar6(char C)139 static unsigned EncodeChar6(char C) { 140 if (C >= 'a' && C <= 'z') return C-'a'; 141 if (C >= 'A' && C <= 'Z') return C-'A'+26; 142 if (C >= '0' && C <= '9') return C-'0'+26+26; 143 if (C == '.') return 62; 144 if (C == '_') return 63; 145 llvm_unreachable("Not a value Char6 character!"); 146 } 147 DecodeChar6(unsigned V)148 static char DecodeChar6(unsigned V) { 149 assert((V & ~63) == 0 && "Not a Char6 encoded character!"); 150 if (V < 26) return V+'a'; 151 if (V < 26+26) return V-26+'A'; 152 if (V < 26+26+10) return V-26-26+'0'; 153 if (V == 62) return '.'; 154 if (V == 63) return '_'; 155 llvm_unreachable("Not a value Char6 character!"); 156 } 157 158 }; 159 160 template <> struct isPodLike<BitCodeAbbrevOp> { static const bool value=true; }; 161 162 /// BitCodeAbbrev - This class represents an abbreviation record. An 163 /// abbreviation allows a complex record that has redundancy to be stored in a 164 /// specialized format instead of the fully-general, fully-vbr, format. 165 class BitCodeAbbrev : public RefCountedBase<BitCodeAbbrev> { 166 SmallVector<BitCodeAbbrevOp, 32> OperandList; 167 // Only RefCountedBase is allowed to delete. 168 ~BitCodeAbbrev() = default; 169 friend class RefCountedBase<BitCodeAbbrev>; 170 171 public: 172 unsigned getNumOperandInfos() const { 173 return static_cast<unsigned>(OperandList.size()); 174 } 175 const BitCodeAbbrevOp &getOperandInfo(unsigned N) const { 176 return OperandList[N]; 177 } 178 179 void Add(const BitCodeAbbrevOp &OpInfo) { 180 OperandList.push_back(OpInfo); 181 } 182 }; 183 } // End llvm namespace 184 185 #endif 186