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 | |
26 | namespace llvm { |
27 | /// Offsets of the 32-bit fields of bitcode wrapper header. |
28 | static const unsigned BWH_MagicField = 0 * 4; |
29 | static const unsigned BWH_VersionField = 1 * 4; |
30 | static const unsigned BWH_OffsetField = 2 * 4; |
31 | static const unsigned BWH_SizeField = 3 * 4; |
32 | static const unsigned BWH_CPUTypeField = 4 * 4; |
33 | static const unsigned = 5 * 4; |
34 | |
35 | namespace bitc { |
36 | enum StandardWidths { |
37 | BlockIDWidth = 8, // We use VBR-8 for block IDs. |
38 | CodeLenWidth = 4, // Codelen are VBR-4. |
39 | BlockSizeWidth = 32 // BlockSize up to 2^32 32-bit words = 16GB per block. |
40 | }; |
41 | |
42 | // The standard abbrev namespace always has a way to exit a block, enter a |
43 | // nested block, define abbrevs, and define an unabbreviated record. |
44 | enum FixedAbbrevIDs { |
45 | END_BLOCK = 0, // Must be zero to guarantee termination for broken bitcode. |
46 | ENTER_SUBBLOCK = 1, |
47 | |
48 | /// DEFINE_ABBREV - Defines an abbrev for the current block. It consists |
49 | /// of a vbr5 for # operand infos. Each operand info is emitted with a |
50 | /// single bit to indicate if it is a literal encoding. If so, the value is |
51 | /// emitted with a vbr8. If not, the encoding is emitted as 3 bits followed |
52 | /// by the info value as a vbr5 if needed. |
53 | DEFINE_ABBREV = 2, |
54 | |
55 | // UNABBREV_RECORDs are emitted with a vbr6 for the record code, followed by |
56 | // a vbr6 for the # operands, followed by vbr6's for each operand. |
57 | UNABBREV_RECORD = 3, |
58 | |
59 | // This is not a code, this is a marker for the first abbrev assignment. |
60 | FIRST_APPLICATION_ABBREV = 4 |
61 | }; |
62 | |
63 | /// StandardBlockIDs - All bitcode files can optionally include a BLOCKINFO |
64 | /// block, which contains metadata about other blocks in the file. |
65 | enum StandardBlockIDs { |
66 | /// BLOCKINFO_BLOCK is used to define metadata about blocks, for example, |
67 | /// standard abbrevs that should be available to all blocks of a specified |
68 | /// ID. |
69 | BLOCKINFO_BLOCK_ID = 0, |
70 | |
71 | // Block IDs 1-7 are reserved for future expansion. |
72 | FIRST_APPLICATION_BLOCKID = 8 |
73 | }; |
74 | |
75 | /// BlockInfoCodes - The blockinfo block contains metadata about user-defined |
76 | /// blocks. |
77 | enum BlockInfoCodes { |
78 | // DEFINE_ABBREV has magic semantics here, applying to the current SETBID'd |
79 | // block, instead of the BlockInfo block. |
80 | |
81 | BLOCKINFO_CODE_SETBID = 1, // SETBID: [blockid#] |
82 | BLOCKINFO_CODE_BLOCKNAME = 2, // BLOCKNAME: [name] |
83 | BLOCKINFO_CODE_SETRECORDNAME = 3 // BLOCKINFO_CODE_SETRECORDNAME: |
84 | // [id, name] |
85 | }; |
86 | |
87 | } // End bitc namespace |
88 | |
89 | /// BitCodeAbbrevOp - This describes one or more operands in an abbreviation. |
90 | /// This is actually a union of two different things: |
91 | /// 1. It could be a literal integer value ("the operand is always 17"). |
92 | /// 2. It could be an encoding specification ("this operand encoded like so"). |
93 | /// |
94 | class BitCodeAbbrevOp { |
95 | uint64_t Val; // A literal value or data for an encoding. |
96 | bool IsLiteral : 1; // Indicate whether this is a literal value or not. |
97 | unsigned Enc : 3; // The encoding to use. |
98 | public: |
99 | enum Encoding { |
100 | Fixed = 1, // A fixed width field, Val specifies number of bits. |
101 | VBR = 2, // A VBR field where Val specifies the width of each chunk. |
102 | Array = 3, // A sequence of fields, next field species elt encoding. |
103 | Char6 = 4, // A 6-bit fixed field which maps to [a-zA-Z0-9._]. |
104 | Blob = 5 // 32-bit aligned array of 8-bit characters. |
105 | }; |
106 | |
107 | explicit BitCodeAbbrevOp(uint64_t V) : Val(V), IsLiteral(true) {} |
108 | explicit BitCodeAbbrevOp(Encoding E, uint64_t Data = 0) |
109 | : Val(Data), IsLiteral(false), Enc(E) {} |
110 | |
111 | bool isLiteral() const { return IsLiteral; } |
112 | bool isEncoding() const { return !IsLiteral; } |
113 | |
114 | // Accessors for literals. |
115 | uint64_t getLiteralValue() const { assert(isLiteral()); return Val; } |
116 | |
117 | // Accessors for encoding info. |
118 | Encoding getEncoding() const { assert(isEncoding()); return (Encoding)Enc; } |
119 | uint64_t getEncodingData() const { |
120 | assert(isEncoding() && hasEncodingData()); |
121 | return Val; |
122 | } |
123 | |
124 | bool hasEncodingData() const { return hasEncodingData(getEncoding()); } |
125 | static bool hasEncodingData(Encoding E) { |
126 | switch (E) { |
127 | case Fixed: |
128 | case VBR: |
129 | return true; |
130 | case Array: |
131 | case Char6: |
132 | case Blob: |
133 | return false; |
134 | } |
135 | report_fatal_error("Invalid encoding" ); |
136 | } |
137 | |
138 | /// isChar6 - Return true if this character is legal in the Char6 encoding. |
139 | static bool isChar6(char C) { |
140 | if (C >= 'a' && C <= 'z') return true; |
141 | if (C >= 'A' && C <= 'Z') return true; |
142 | if (C >= '0' && C <= '9') return true; |
143 | if (C == '.' || C == '_') return true; |
144 | return false; |
145 | } |
146 | static unsigned EncodeChar6(char C) { |
147 | if (C >= 'a' && C <= 'z') return C-'a'; |
148 | if (C >= 'A' && C <= 'Z') return C-'A'+26; |
149 | if (C >= '0' && C <= '9') return C-'0'+26+26; |
150 | if (C == '.') return 62; |
151 | if (C == '_') return 63; |
152 | llvm_unreachable("Not a value Char6 character!" ); |
153 | } |
154 | |
155 | static char DecodeChar6(unsigned V) { |
156 | assert((V & ~63) == 0 && "Not a Char6 encoded character!" ); |
157 | return "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789._" |
158 | [V]; |
159 | } |
160 | |
161 | }; |
162 | |
163 | template <> struct isPodLike<BitCodeAbbrevOp> { static const bool value=true; }; |
164 | |
165 | /// BitCodeAbbrev - This class represents an abbreviation record. An |
166 | /// abbreviation allows a complex record that has redundancy to be stored in a |
167 | /// specialized format instead of the fully-general, fully-vbr, format. |
168 | class BitCodeAbbrev { |
169 | SmallVector<BitCodeAbbrevOp, 32> OperandList; |
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 | |