X86DisassemblerTables.h revision 208954 
1//===- X86DisassemblerTables.h - Disassembler tables ------------*- 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 file is part of the X86 Disassembler Emitter.
11// It contains the interface of the disassembler tables.
12// Documentation for the disassembler emitter in general can be found in
13//  X86DisasemblerEmitter.h.
14//
15//===----------------------------------------------------------------------===//
16
17#ifndef X86DISASSEMBLERTABLES_H
18#define X86DISASSEMBLERTABLES_H
19
20#include "X86DisassemblerShared.h"
21#include "X86ModRMFilters.h"
22
23#include "llvm/Support/raw_ostream.h"
24
25#include <vector>
26
27namespace llvm {
28
29namespace X86Disassembler {
30
31/// DisassemblerTables - Encapsulates all the decode tables being generated by
32///   the table emitter.  Contains functions to populate the tables as well as
33///   to emit them as hierarchical C structures suitable for consumption by the
34///   runtime.
35class DisassemblerTables {
36private:
37  /// The decoder tables.  There is one for each opcode type:
38  /// [0] one-byte opcodes
39  /// [1] two-byte opcodes of the form 0f __
40  /// [2] three-byte opcodes of the form 0f 38 __
41  /// [3] three-byte opcodes of the form 0f 3a __
42  ContextDecision* Tables[4];
43
44  /// The instruction information table
45  std::vector<InstructionSpecifier> InstructionSpecifiers;
46
47  /// True if there are primary decode conflicts in the instruction set
48  bool HasConflicts;
49
50  /// emitOneID - Emits a table entry for a single instruction entry, at the
51  ///   innermost level of the structure hierarchy.  The entry is printed out
52  ///   in the format "nnnn, /* MNEMONIC */" where nnnn is the ID in decimal,
53  ///   the comma is printed if addComma is true, and the menonic is the name
54  ///   of the instruction as listed in the LLVM tables.
55  ///
56  /// @param o        - The output stream to print the entry on.
57  /// @param i        - The indentation level for o.
58  /// @param id       - The unique ID of the instruction to print.
59  /// @param addComma - Whether or not to print a comma after the ID.  True if
60  ///                    additional items will follow.
61  void emitOneID(raw_ostream &o,
62                 uint32_t &i,
63                 InstrUID id,
64                 bool addComma) const;
65
66  /// emitModRMDecision - Emits a table of entries corresponding to a single
67  ///   ModR/M decision.  Compacts the ModR/M decision if possible.  ModR/M
68  ///   decisions are printed as:
69  ///
70  ///   { /* struct ModRMDecision */
71  ///     TYPE,
72  ///     modRMTablennnn
73  ///   }
74  ///
75  ///   where nnnn is a unique ID for the corresponding table of IDs.
76  ///   TYPE indicates whether the table has one entry that is the same
77  ///   regardless of ModR/M byte, two entries - one for bytes 0x00-0xbf and one
78  ///   for bytes 0xc0-0xff -, or 256 entries, one for each possible byte.
79  ///   nnnn is the number of a table for looking up these values.  The tables
80  ///   are writen separately so that tables consisting entirely of zeros will
81  ///   not be duplicated.  (These all have the name modRMEmptyTable.)  A table
82  ///   is printed as:
83  ///
84  ///   InstrUID modRMTablennnn[k] = {
85  ///     nnnn, /* MNEMONIC */
86  ///     ...
87  ///     nnnn /* MNEMONIC */
88  ///   };
89  ///
90  /// @param o1       - The output stream to print the ID table to.
91  /// @param o2       - The output stream to print the decision structure to.
92  /// @param i1       - The indentation level to use with stream o1.
93  /// @param i2       - The indentation level to use with stream o2.
94  /// @param decision - The ModR/M decision to emit.  This decision has 256
95  ///                   entries - emitModRMDecision decides how to compact it.
96  void emitModRMDecision(raw_ostream &o1,
97                         raw_ostream &o2,
98                         uint32_t &i1,
99                         uint32_t &i2,
100                         ModRMDecision &decision) const;
101
102  /// emitOpcodeDecision - Emits an OpcodeDecision and all its subsidiary ModR/M
103  ///   decisions.  An OpcodeDecision is printed as:
104  ///
105  ///   { /* struct OpcodeDecision */
106  ///     /* 0x00 */
107  ///     { /* struct ModRMDecision */
108  ///       ...
109  ///     }
110  ///     ...
111  ///   }
112  ///
113  ///   where the ModRMDecision structure is printed as described in the
114  ///   documentation for emitModRMDecision().  emitOpcodeDecision() passes on a
115  ///   stream and indent level for the UID tables generated by
116  ///   emitModRMDecision(), but does not use them itself.
117  ///
118  /// @param o1       - The output stream to print the ID tables generated by
119  ///                   emitModRMDecision() to.
120  /// @param o2       - The output stream for the decision structure itself.
121  /// @param i1       - The indent level to use with stream o1.
122  /// @param i2       - The indent level to use with stream o2.
123  /// @param decision - The OpcodeDecision to emit along with its subsidiary
124  ///                    structures.
125  void emitOpcodeDecision(raw_ostream &o1,
126                          raw_ostream &o2,
127                          uint32_t &i1,
128                          uint32_t &i2,
129                          OpcodeDecision &decision) const;
130
131  /// emitContextDecision - Emits a ContextDecision and all its subsidiary
132  ///   Opcode and ModRMDecisions.  A ContextDecision is printed as:
133  ///
134  ///   struct ContextDecision NAME = {
135  ///     { /* OpcodeDecisions */
136  ///       /* IC */
137  ///       { /* struct OpcodeDecision */
138  ///         ...
139  ///       },
140  ///       ...
141  ///     }
142  ///   }
143  ///
144  ///   NAME is the name of the ContextDecision (typically one of the four names
145  ///   ONEBYTE_SYM, TWOBYTE_SYM, THREEBYTE38_SYM, and THREEBYTE3A_SYM from
146  ///   X86DisassemblerDecoderCommon.h).
147  ///   IC is one of the contexts in InstructionContext.  There is an opcode
148  ///   decision for each possible context.
149  ///   The OpcodeDecision structures are printed as described in the
150  ///   documentation for emitOpcodeDecision.
151  ///
152  /// @param o1       - The output stream to print the ID tables generated by
153  ///                   emitModRMDecision() to.
154  /// @param o2       - The output stream to print the decision structure to.
155  /// @param i1       - The indent level to use with stream o1.
156  /// @param i2       - The indent level to use with stream o2.
157  /// @param decision - The ContextDecision to emit along with its subsidiary
158  ///                   structures.
159  /// @param name     - The name for the ContextDecision.
160  void emitContextDecision(raw_ostream &o1,
161                           raw_ostream &o2,
162                           uint32_t &i1,
163                           uint32_t &i2,
164                           ContextDecision &decision,
165                           const char* name) const;
166
167  /// emitInstructionInfo - Prints the instruction specifier table, which has
168  ///   one entry for each instruction, and contains name and operand
169  ///   information.  This table is printed as:
170  ///
171  ///   struct InstructionSpecifier CONTEXTS_SYM[k] = {
172  ///     {
173  ///       /* nnnn */
174  ///       "MNEMONIC",
175  ///       0xnn,
176  ///       {
177  ///         {
178  ///           ENCODING,
179  ///           TYPE
180  ///         },
181  ///         ...
182  ///       }
183  ///     },
184  ///   };
185  ///
186  ///   k is the total number of instructions.
187  ///   nnnn is the ID of the current instruction (0-based).  This table
188  ///   includes entries for non-instructions like PHINODE.
189  ///   0xnn is the lowest possible opcode for the current instruction, used for
190  ///   AddRegFrm instructions to compute the operand's value.
191  ///   ENCODING and TYPE describe the encoding and type for a single operand.
192  ///
193  /// @param o  - The output stream to which the instruction table should be
194  ///             written.
195  /// @param i  - The indent level for use with the stream.
196  void emitInstructionInfo(raw_ostream &o, uint32_t &i) const;
197
198  /// emitContextTable - Prints the table that is used to translate from an
199  ///   instruction attribute mask to an instruction context.  This table is
200  ///   printed as:
201  ///
202  ///   InstructionContext CONTEXTS_STR[256] = {
203  ///     IC, /* 0x00 */
204  ///     ...
205  ///   };
206  ///
207  ///   IC is the context corresponding to the mask 0x00, and there are 256
208  ///   possible masks.
209  ///
210  /// @param o  - The output stream to which the context table should be written.
211  /// @param i  - The indent level for use with the stream.
212  void emitContextTable(raw_ostream &o, uint32_t &i) const;
213
214  /// emitContextDecisions - Prints all four ContextDecision structures using
215  ///   emitContextDecision().
216  ///
217  /// @param o1 - The output stream to print the ID tables generated by
218  ///             emitModRMDecision() to.
219  /// @param o2 - The output stream to print the decision structures to.
220  /// @param i1 - The indent level to use with stream o1.
221  /// @param i2 - The indent level to use with stream o2.
222  void emitContextDecisions(raw_ostream &o1,
223                            raw_ostream &o2,
224                            uint32_t &i1,
225                            uint32_t &i2) const;
226
227  /// setTableFields - Uses a ModRMFilter to set the appropriate entries in a
228  ///   ModRMDecision to refer to a particular instruction ID.
229  ///
230  /// @param decision - The ModRMDecision to populate.
231  /// @param filter   - The filter to use in deciding which entries to populate.
232  /// @param uid      - The unique ID to set matching entries to.
233  /// @param opcode   - The opcode of the instruction, for error reporting.
234  void setTableFields(ModRMDecision &decision,
235                      const ModRMFilter &filter,
236                      InstrUID uid,
237                      uint8_t opcode);
238public:
239  /// Constructor - Allocates space for the class decisions and clears them.
240  DisassemblerTables();
241
242  ~DisassemblerTables();
243
244  /// emit - Emits the instruction table, context table, and class decisions.
245  ///
246  /// @param o  - The output stream to print the tables to.
247  void emit(raw_ostream &o) const;
248
249  /// setTableFields - Uses the opcode type, instruction context, opcode, and a
250  ///   ModRMFilter as criteria to set a particular set of entries in the
251  ///   decode tables to point to a specific uid.
252  ///
253  /// @param type         - The opcode type (ONEBYTE, TWOBYTE, etc.)
254  /// @param insnContext  - The context to use (IC, IC_64BIT, etc.)
255  /// @param opcode       - The last byte of the opcode (not counting any escape
256  ///                       or extended opcodes).
257  /// @param filter       - The ModRMFilter that decides which ModR/M byte values
258  ///                       correspond to the desired instruction.
259  /// @param uid          - The unique ID of the instruction.
260  void setTableFields(OpcodeType type,
261                      InstructionContext insnContext,
262                      uint8_t opcode,
263                      const ModRMFilter &filter,
264                      InstrUID uid);
265
266  /// specForUID - Returns the instruction specifier for a given unique
267  ///   instruction ID.  Used when resolving collisions.
268  ///
269  /// @param uid  - The unique ID of the instruction.
270  /// @return     - A reference to the instruction specifier.
271  InstructionSpecifier& specForUID(InstrUID uid) {
272    if (uid >= InstructionSpecifiers.size())
273      InstructionSpecifiers.resize(uid + 1);
274
275    return InstructionSpecifiers[uid];
276  }
277
278  // hasConflicts - Reports whether there were primary decode conflicts
279  //   from any instructions added to the tables.
280  // @return  - true if there were; false otherwise.
281
282  bool hasConflicts() {
283    return HasConflicts;
284  }
285};
286
287} // namespace X86Disassembler
288
289} // namespace llvm
290
291#endif
292