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/SmallVector.h"
22#include "llvm/Support/DataTypes.h"
23#include "llvm/Support/ErrorHandling.h"
24#include <cassert>
25
26namespace llvm {
27namespace bitc {
28  enum StandardWidths {
29    BlockIDWidth   = 8,  // We use VBR-8 for block IDs.
30    CodeLenWidth   = 4,  // Codelen are VBR-4.
31    BlockSizeWidth = 32  // BlockSize up to 2^32 32-bit words = 16GB per block.
32  };
33
34  // The standard abbrev namespace always has a way to exit a block, enter a
35  // nested block, define abbrevs, and define an unabbreviated record.
36  enum FixedAbbrevIDs {
37    END_BLOCK = 0,  // Must be zero to guarantee termination for broken bitcode.
38    ENTER_SUBBLOCK = 1,
39
40    /// DEFINE_ABBREV - Defines an abbrev for the current block.  It consists
41    /// of a vbr5 for # operand infos.  Each operand info is emitted with a
42    /// single bit to indicate if it is a literal encoding.  If so, the value is
43    /// emitted with a vbr8.  If not, the encoding is emitted as 3 bits followed
44    /// by the info value as a vbr5 if needed.
45    DEFINE_ABBREV = 2,
46
47    // UNABBREV_RECORDs are emitted with a vbr6 for the record code, followed by
48    // a vbr6 for the # operands, followed by vbr6's for each operand.
49    UNABBREV_RECORD = 3,
50
51    // This is not a code, this is a marker for the first abbrev assignment.
52    FIRST_APPLICATION_ABBREV = 4
53  };
54
55  /// StandardBlockIDs - All bitcode files can optionally include a BLOCKINFO
56  /// block, which contains metadata about other blocks in the file.
57  enum StandardBlockIDs {
58    /// BLOCKINFO_BLOCK is used to define metadata about blocks, for example,
59    /// standard abbrevs that should be available to all blocks of a specified
60    /// ID.
61    BLOCKINFO_BLOCK_ID = 0,
62
63    // Block IDs 1-7 are reserved for future expansion.
64    FIRST_APPLICATION_BLOCKID = 8
65  };
66
67  /// BlockInfoCodes - The blockinfo block contains metadata about user-defined
68  /// blocks.
69  enum BlockInfoCodes {
70    // DEFINE_ABBREV has magic semantics here, applying to the current SETBID'd
71    // block, instead of the BlockInfo block.
72
73    BLOCKINFO_CODE_SETBID        = 1, // SETBID: [blockid#]
74    BLOCKINFO_CODE_BLOCKNAME     = 2, // BLOCKNAME: [name]
75    BLOCKINFO_CODE_SETRECORDNAME = 3  // BLOCKINFO_CODE_SETRECORDNAME:
76                                      //                             [id, name]
77  };
78
79} // End bitc namespace
80
81/// BitCodeAbbrevOp - This describes one or more operands in an abbreviation.
82/// This is actually a union of two different things:
83///   1. It could be a literal integer value ("the operand is always 17").
84///   2. It could be an encoding specification ("this operand encoded like so").
85///
86class BitCodeAbbrevOp {
87  uint64_t Val;           // A literal value or data for an encoding.
88  bool IsLiteral : 1;     // Indicate whether this is a literal value or not.
89  unsigned Enc   : 3;     // The encoding to use.
90public:
91  enum Encoding {
92    Fixed = 1,  // A fixed width field, Val specifies number of bits.
93    VBR   = 2,  // A VBR field where Val specifies the width of each chunk.
94    Array = 3,  // A sequence of fields, next field species elt encoding.
95    Char6 = 4,  // A 6-bit fixed field which maps to [a-zA-Z0-9._].
96    Blob  = 5   // 32-bit aligned array of 8-bit characters.
97  };
98
99  explicit BitCodeAbbrevOp(uint64_t V) :  Val(V), IsLiteral(true) {}
100  explicit BitCodeAbbrevOp(Encoding E, uint64_t Data = 0)
101    : Val(Data), IsLiteral(false), Enc(E) {}
102
103  bool isLiteral() const  { return IsLiteral; }
104  bool isEncoding() const { return !IsLiteral; }
105
106  // Accessors for literals.
107  uint64_t getLiteralValue() const { assert(isLiteral()); return Val; }
108
109  // Accessors for encoding info.
110  Encoding getEncoding() const { assert(isEncoding()); return (Encoding)Enc; }
111  uint64_t getEncodingData() const {
112    assert(isEncoding() && hasEncodingData());
113    return Val;
114  }
115
116  bool hasEncodingData() const { return hasEncodingData(getEncoding()); }
117  static bool hasEncodingData(Encoding E) {
118    switch (E) {
119    case Fixed:
120    case VBR:
121      return true;
122    case Array:
123    case Char6:
124    case Blob:
125      return false;
126    }
127    llvm_unreachable("Invalid encoding");
128  }
129
130  /// isChar6 - Return true if this character is legal in the Char6 encoding.
131  static bool isChar6(char C) {
132    if (C >= 'a' && C <= 'z') return true;
133    if (C >= 'A' && C <= 'Z') return true;
134    if (C >= '0' && C <= '9') return true;
135    if (C == '.' || C == '_') return true;
136    return false;
137  }
138  static unsigned EncodeChar6(char C) {
139    if (C >= 'a' && C <= 'z') return C-'a';
140    if (C >= 'A' && C <= 'Z') return C-'A'+26;
141    if (C >= '0' && C <= '9') return C-'0'+26+26;
142    if (C == '.')             return 62;
143    if (C == '_')             return 63;
144    llvm_unreachable("Not a value Char6 character!");
145  }
146
147  static char DecodeChar6(unsigned V) {
148    assert((V & ~63) == 0 && "Not a Char6 encoded character!");
149    if (V < 26)       return V+'a';
150    if (V < 26+26)    return V-26+'A';
151    if (V < 26+26+10) return V-26-26+'0';
152    if (V == 62)      return '.';
153    if (V == 63)      return '_';
154    llvm_unreachable("Not a value Char6 character!");
155  }
156
157};
158
159template <> struct isPodLike<BitCodeAbbrevOp> { static const bool value=true; };
160
161/// BitCodeAbbrev - This class represents an abbreviation record.  An
162/// abbreviation allows a complex record that has redundancy to be stored in a
163/// specialized format instead of the fully-general, fully-vbr, format.
164class BitCodeAbbrev {
165  SmallVector<BitCodeAbbrevOp, 32> OperandList;
166  unsigned char RefCount; // Number of things using this.
167  ~BitCodeAbbrev() {}
168public:
169  BitCodeAbbrev() : RefCount(1) {}
170
171  void addRef() { ++RefCount; }
172  void dropRef() { if (--RefCount == 0) delete this; }
173
174  unsigned getNumOperandInfos() const {
175    return static_cast<unsigned>(OperandList.size());
176  }
177  const BitCodeAbbrevOp &getOperandInfo(unsigned N) const {
178    return OperandList[N];
179  }
180
181  void Add(const BitCodeAbbrevOp &OpInfo) {
182    OperandList.push_back(OpInfo);
183  }
184};
185} // End llvm namespace
186
187#endif
188