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