1//===- X86RecognizableInstr.h - Disassembler instruction spec ---*- C++ -*-===//
2//
3// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4// See https://llvm.org/LICENSE.txt for license information.
5// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6//
7//===----------------------------------------------------------------------===//
8//
9// This file is part of the X86 Disassembler Emitter.
10// It contains the interface of a single recognizable instruction.
11// Documentation for the disassembler emitter in general can be found in
12// X86DisassemblerEmitter.h.
13//
14//===----------------------------------------------------------------------===//
15
16#ifndef LLVM_UTILS_TABLEGEN_X86RECOGNIZABLEINSTR_H
17#define LLVM_UTILS_TABLEGEN_X86RECOGNIZABLEINSTR_H
18
19#include "Common/CodeGenInstruction.h"
20#include "llvm/Support/X86DisassemblerDecoderCommon.h"
21#include <cstdint>
22#include <string>
23#include <vector>
24
25struct InstructionSpecifier;
26
27namespace llvm {
28class Record;
29#define X86_INSTR_MRM_MAPPING \
30 MAP(C0, 64) \
31 MAP(C1, 65) \
32 MAP(C2, 66) \
33 MAP(C3, 67) \
34 MAP(C4, 68) \
35 MAP(C5, 69) \
36 MAP(C6, 70) \
37 MAP(C7, 71) \
38 MAP(C8, 72) \
39 MAP(C9, 73) \
40 MAP(CA, 74) \
41 MAP(CB, 75) \
42 MAP(CC, 76) \
43 MAP(CD, 77) \
44 MAP(CE, 78) \
45 MAP(CF, 79) \
46 MAP(D0, 80) \
47 MAP(D1, 81) \
48 MAP(D2, 82) \
49 MAP(D3, 83) \
50 MAP(D4, 84) \
51 MAP(D5, 85) \
52 MAP(D6, 86) \
53 MAP(D7, 87) \
54 MAP(D8, 88) \
55 MAP(D9, 89) \
56 MAP(DA, 90) \
57 MAP(DB, 91) \
58 MAP(DC, 92) \
59 MAP(DD, 93) \
60 MAP(DE, 94) \
61 MAP(DF, 95) \
62 MAP(E0, 96) \
63 MAP(E1, 97) \
64 MAP(E2, 98) \
65 MAP(E3, 99) \
66 MAP(E4, 100) \
67 MAP(E5, 101) \
68 MAP(E6, 102) \
69 MAP(E7, 103) \
70 MAP(E8, 104) \
71 MAP(E9, 105) \
72 MAP(EA, 106) \
73 MAP(EB, 107) \
74 MAP(EC, 108) \
75 MAP(ED, 109) \
76 MAP(EE, 110) \
77 MAP(EF, 111) \
78 MAP(F0, 112) \
79 MAP(F1, 113) \
80 MAP(F2, 114) \
81 MAP(F3, 115) \
82 MAP(F4, 116) \
83 MAP(F5, 117) \
84 MAP(F6, 118) \
85 MAP(F7, 119) \
86 MAP(F8, 120) \
87 MAP(F9, 121) \
88 MAP(FA, 122) \
89 MAP(FB, 123) \
90 MAP(FC, 124) \
91 MAP(FD, 125) \
92 MAP(FE, 126) \
93 MAP(FF, 127)
94
95// A clone of X86 since we can't depend on something that is generated.
96namespace X86Local {
97enum {
98 Pseudo = 0,
99 RawFrm = 1,
100 AddRegFrm = 2,
101 RawFrmMemOffs = 3,
102 RawFrmSrc = 4,
103 RawFrmDst = 5,
104 RawFrmDstSrc = 6,
105 RawFrmImm8 = 7,
106 RawFrmImm16 = 8,
107 AddCCFrm = 9,
108 PrefixByte = 10,
109 MRMDestRegCC = 18,
110 MRMDestMemCC = 19,
111 MRMDestMem4VOp3CC = 20,
112 MRMr0 = 21,
113 MRMSrcMemFSIB = 22,
114 MRMDestMemFSIB = 23,
115 MRMDestMem = 24,
116 MRMSrcMem = 25,
117 MRMSrcMem4VOp3 = 26,
118 MRMSrcMemOp4 = 27,
119 MRMSrcMemCC = 28,
120 MRMXmCC = 30,
121 MRMXm = 31,
122 MRM0m = 32,
123 MRM1m = 33,
124 MRM2m = 34,
125 MRM3m = 35,
126 MRM4m = 36,
127 MRM5m = 37,
128 MRM6m = 38,
129 MRM7m = 39,
130 MRMDestReg = 40,
131 MRMSrcReg = 41,
132 MRMSrcReg4VOp3 = 42,
133 MRMSrcRegOp4 = 43,
134 MRMSrcRegCC = 44,
135 MRMXrCC = 46,
136 MRMXr = 47,
137 MRM0r = 48,
138 MRM1r = 49,
139 MRM2r = 50,
140 MRM3r = 51,
141 MRM4r = 52,
142 MRM5r = 53,
143 MRM6r = 54,
144 MRM7r = 55,
145 MRM0X = 56,
146 MRM1X = 57,
147 MRM2X = 58,
148 MRM3X = 59,
149 MRM4X = 60,
150 MRM5X = 61,
151 MRM6X = 62,
152 MRM7X = 63,
153#define MAP(from, to) MRM_##from = to,
154 X86_INSTR_MRM_MAPPING
155#undef MAP
156};
157
158enum {
159 OB = 0,
160 TB = 1,
161 T8 = 2,
162 TA = 3,
163 XOP8 = 4,
164 XOP9 = 5,
165 XOPA = 6,
166 ThreeDNow = 7,
167 T_MAP4 = 8,
168 T_MAP5 = 9,
169 T_MAP6 = 10,
170 T_MAP7 = 11
171};
172
173enum { PD = 1, XS = 2, XD = 3, PS = 4 };
174enum { VEX = 1, XOP = 2, EVEX = 3 };
175enum { OpSize16 = 1, OpSize32 = 2 };
176enum { AdSize16 = 1, AdSize32 = 2, AdSize64 = 3 };
177enum { ExplicitREX2 = 1, ExplicitEVEX = 3 };
178} // namespace X86Local
179
180namespace X86Disassembler {
181class DisassemblerTables;
182/// Extract common fields of a single X86 instruction from a CodeGenInstruction
183struct RecognizableInstrBase {
184 /// The OpPrefix field from the record
185 uint8_t OpPrefix;
186 /// The OpMap field from the record
187 uint8_t OpMap;
188 /// The opcode field from the record; this is the opcode used in the Intel
189 /// encoding and therefore distinct from the UID
190 uint8_t Opcode;
191 /// The form field from the record
192 uint8_t Form;
193 // The encoding field from the record
194 uint8_t Encoding;
195 /// The OpSize field from the record
196 uint8_t OpSize;
197 /// The AdSize field from the record
198 uint8_t AdSize;
199 /// The hasREX_W field from the record
200 bool HasREX_W;
201 /// The hasVEX_4V field from the record
202 bool HasVEX_4V;
203 /// The IgnoresW field from the record
204 bool IgnoresW;
205 /// The hasVEX_L field from the record
206 bool HasVEX_L;
207 /// The ignoreVEX_L field from the record
208 bool IgnoresVEX_L;
209 /// The hasEVEX_L2Prefix field from the record
210 bool HasEVEX_L2;
211 /// The hasEVEX_K field from the record
212 bool HasEVEX_K;
213 /// The hasEVEX_KZ field from the record
214 bool HasEVEX_KZ;
215 /// The hasEVEX_B field from the record
216 bool HasEVEX_B;
217 /// The hasEVEX_U field from the record
218 bool HasEVEX_U;
219 /// The hasEVEX_NF field from the record
220 bool HasEVEX_NF;
221 /// The hasTwoConditionalOps field from the record
222 bool HasTwoConditionalOps;
223 /// Indicates that the instruction uses the L and L' fields for RC.
224 bool EncodeRC;
225 /// The isCodeGenOnly field from the record
226 bool IsCodeGenOnly;
227 /// The isAsmParserOnly field from the record
228 bool IsAsmParserOnly;
229 /// The ForceDisassemble field from the record
230 bool ForceDisassemble;
231 // The CD8_Scale field from the record
232 uint8_t CD8_Scale;
233 /// If explicitOpPrefix field from the record equals ExplicitREX2
234 bool ExplicitREX2Prefix;
235 /// \param insn The CodeGenInstruction to extract information from.
236 RecognizableInstrBase(const CodeGenInstruction &insn);
237 /// \returns true if this instruction should be emitted
238 bool shouldBeEmitted() const;
239};
240
241/// RecognizableInstr - Encapsulates all information required to decode a single
242/// instruction, as extracted from the LLVM instruction tables. Has methods
243/// to interpret the information available in the LLVM tables, and to emit the
244/// instruction into DisassemblerTables.
245class RecognizableInstr : public RecognizableInstrBase {
246private:
247 /// The record from the .td files corresponding to this instruction
248 const Record *Rec;
249 /// The instruction name as listed in the tables
250 std::string Name;
251 // Whether the instruction has the predicate "In32BitMode"
252 bool Is32Bit;
253 // Whether the instruction has the predicate "In64BitMode"
254 bool Is64Bit;
255 /// The operands of the instruction, as listed in the CodeGenInstruction.
256 /// They are not one-to-one with operands listed in the MCInst; for example,
257 /// memory operands expand to 5 operands in the MCInst
258 const std::vector<CGIOperandList::OperandInfo> *Operands;
259
260 /// The opcode of the instruction, as used in an MCInst
261 InstrUID UID;
262 /// The description of the instruction that is emitted into the instruction
263 /// info table
264 InstructionSpecifier *Spec;
265
266 /// insnContext - Returns the primary context in which the instruction is
267 /// valid.
268 ///
269 /// @return - The context in which the instruction is valid.
270 InstructionContext insnContext() const;
271
272 /// typeFromString - Translates an operand type from the string provided in
273 /// the LLVM tables to an OperandType for use in the operand specifier.
274 ///
275 /// @param s - The string, as extracted by calling Rec->getName()
276 /// on a CodeGenInstruction::OperandInfo.
277 /// @param hasREX_W - Indicates whether the instruction has a REX.W
278 /// prefix. If it does, 32-bit register operands stay
279 /// 32-bit regardless of the operand size.
280 /// @param OpSize Indicates the operand size of the instruction.
281 /// If register size does not match OpSize, then
282 /// register sizes keep their size.
283 /// @return - The operand's type.
284 static OperandType typeFromString(StringRef Str, bool hasREX_W,
285 uint8_t OpSize);
286
287 /// immediateEncodingFromString - Translates an immediate encoding from the
288 /// string provided in the LLVM tables to an OperandEncoding for use in
289 /// the operand specifier.
290 ///
291 /// @param s - See typeFromString().
292 /// @param OpSize - Indicates whether this is an OpSize16 instruction.
293 /// If it is not, then 16-bit immediate operands stay 16-bit.
294 /// @return - The operand's encoding.
295 static OperandEncoding immediateEncodingFromString(StringRef Str,
296 uint8_t OpSize);
297
298 /// rmRegisterEncodingFromString - Like immediateEncodingFromString, but
299 /// handles operands that are in the REG field of the ModR/M byte.
300 static OperandEncoding rmRegisterEncodingFromString(StringRef Str,
301 uint8_t OpSize);
302
303 /// rmRegisterEncodingFromString - Like immediateEncodingFromString, but
304 /// handles operands that are in the REG field of the ModR/M byte.
305 static OperandEncoding roRegisterEncodingFromString(StringRef Str,
306 uint8_t OpSize);
307 static OperandEncoding memoryEncodingFromString(StringRef Str,
308 uint8_t OpSize);
309 static OperandEncoding relocationEncodingFromString(StringRef Str,
310 uint8_t OpSize);
311 static OperandEncoding opcodeModifierEncodingFromString(StringRef Str,
312 uint8_t OpSize);
313 static OperandEncoding vvvvRegisterEncodingFromString(StringRef Str,
314 uint8_t OpSize);
315 static OperandEncoding writemaskRegisterEncodingFromString(StringRef Str,
316 uint8_t OpSize);
317
318 /// Adjust the encoding type for an operand based on the instruction.
319 void adjustOperandEncoding(OperandEncoding &encoding);
320
321 /// handleOperand - Converts a single operand from the LLVM table format to
322 /// the emitted table format, handling any duplicate operands it encounters
323 /// and then one non-duplicate.
324 ///
325 /// @param optional - Determines whether to assert that the
326 /// operand exists.
327 /// @param operandIndex - The index into the generated operand table.
328 /// Incremented by this function one or more
329 /// times to reflect possible duplicate
330 /// operands).
331 /// @param physicalOperandIndex - The index of the current operand into the
332 /// set of non-duplicate ('physical') operands.
333 /// Incremented by this function once.
334 /// @param numPhysicalOperands - The number of non-duplicate operands in the
335 /// instructions.
336 /// @param operandMapping - The operand mapping, which has an entry for
337 /// each operand that indicates whether it is a
338 /// duplicate, and of what.
339 using EncodingFn =
340 llvm::function_ref<OperandEncoding(StringRef s, uint8_t OpSize)>;
341 void handleOperand(bool optional, unsigned &operandIndex,
342 unsigned &physicalOperandIndex,
343 unsigned numPhysicalOperands,
344 const unsigned *operandMapping,
345 EncodingFn encodingFromString);
346
347 /// emitInstructionSpecifier - Loads the instruction specifier for the current
348 /// instruction into a DisassemblerTables.
349 ///
350 void emitInstructionSpecifier();
351
352 /// emitDecodePath - Populates the proper fields in the decode tables
353 /// corresponding to the decode paths for this instruction.
354 ///
355 /// \param tables The DisassemblerTables to populate with the decode
356 /// decode information for the current instruction.
357 void emitDecodePath(DisassemblerTables &tables) const;
358
359public:
360 /// Constructor - Initializes a RecognizableInstr with the appropriate fields
361 /// from a CodeGenInstruction.
362 ///
363 /// \param tables The DisassemblerTables that the specifier will be added to.
364 /// \param insn The CodeGenInstruction to extract information from.
365 /// \param uid The unique ID of the current instruction.
366 RecognizableInstr(DisassemblerTables &tables, const CodeGenInstruction &insn,
367 InstrUID uid);
368 /// processInstr - Accepts a CodeGenInstruction and loads decode information
369 /// for it into a DisassemblerTables if appropriate.
370 ///
371 /// \param tables The DiassemblerTables to be populated with decode
372 /// information.
373 /// \param insn The CodeGenInstruction to be used as a source for this
374 /// information.
375 /// \param uid The unique ID of the instruction.
376 static void processInstr(DisassemblerTables &tables,
377 const CodeGenInstruction &insn, InstrUID uid);
378};
379
380std::string getMnemonic(const CodeGenInstruction *I, unsigned Variant);
381bool isRegisterOperand(const Record *Rec);
382bool isMemoryOperand(const Record *Rec);
383bool isImmediateOperand(const Record *Rec);
384unsigned getRegOperandSize(const Record *RegRec);
385unsigned getMemOperandSize(const Record *MemRec);
386} // namespace X86Disassembler
387} // namespace llvm
388#endif
389