blob: 44ffe5fb5e68045f79ca8893fea362048be5fb5e [file] [log] [blame]
Steve Blocka7e24c12009-10-30 11:49:00 +00001// Copyright 2009 the V8 project authors. All rights reserved.
2// Redistribution and use in source and binary forms, with or without
3// modification, are permitted provided that the following conditions are
4// met:
5//
6// * Redistributions of source code must retain the above copyright
7// notice, this list of conditions and the following disclaimer.
8// * Redistributions in binary form must reproduce the above
9// copyright notice, this list of conditions and the following
10// disclaimer in the documentation and/or other materials provided
11// with the distribution.
12// * Neither the name of Google Inc. nor the names of its
13// contributors may be used to endorse or promote products derived
14// from this software without specific prior written permission.
15//
16// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
17// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
18// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
19// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
20// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
21// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
22// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
26// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27
28#include <assert.h>
29#include <stdio.h>
30#include <stdarg.h>
31
32#include "v8.h"
Leon Clarkef7060e22010-06-03 12:02:55 +010033
34#if defined(V8_TARGET_ARCH_X64)
35
Steve Blocka7e24c12009-10-30 11:49:00 +000036#include "disasm.h"
37
38namespace disasm {
39
40enum OperandType {
41 UNSET_OP_ORDER = 0,
42 // Operand size decides between 16, 32 and 64 bit operands.
43 REG_OPER_OP_ORDER = 1, // Register destination, operand source.
44 OPER_REG_OP_ORDER = 2, // Operand destination, register source.
45 // Fixed 8-bit operands.
46 BYTE_SIZE_OPERAND_FLAG = 4,
47 BYTE_REG_OPER_OP_ORDER = REG_OPER_OP_ORDER | BYTE_SIZE_OPERAND_FLAG,
48 BYTE_OPER_REG_OP_ORDER = OPER_REG_OP_ORDER | BYTE_SIZE_OPERAND_FLAG
49};
50
51//------------------------------------------------------------------
52// Tables
53//------------------------------------------------------------------
54struct ByteMnemonic {
55 int b; // -1 terminates, otherwise must be in range (0..255)
56 OperandType op_order_;
57 const char* mnem;
58};
59
60
61static ByteMnemonic two_operands_instr[] = {
62 { 0x00, BYTE_OPER_REG_OP_ORDER, "add" },
63 { 0x01, OPER_REG_OP_ORDER, "add" },
64 { 0x02, BYTE_REG_OPER_OP_ORDER, "add" },
65 { 0x03, REG_OPER_OP_ORDER, "add" },
66 { 0x08, BYTE_OPER_REG_OP_ORDER, "or" },
67 { 0x09, OPER_REG_OP_ORDER, "or" },
68 { 0x0A, BYTE_REG_OPER_OP_ORDER, "or" },
69 { 0x0B, REG_OPER_OP_ORDER, "or" },
70 { 0x10, BYTE_OPER_REG_OP_ORDER, "adc" },
71 { 0x11, OPER_REG_OP_ORDER, "adc" },
72 { 0x12, BYTE_REG_OPER_OP_ORDER, "adc" },
73 { 0x13, REG_OPER_OP_ORDER, "adc" },
74 { 0x18, BYTE_OPER_REG_OP_ORDER, "sbb" },
75 { 0x19, OPER_REG_OP_ORDER, "sbb" },
76 { 0x1A, BYTE_REG_OPER_OP_ORDER, "sbb" },
77 { 0x1B, REG_OPER_OP_ORDER, "sbb" },
78 { 0x20, BYTE_OPER_REG_OP_ORDER, "and" },
79 { 0x21, OPER_REG_OP_ORDER, "and" },
80 { 0x22, BYTE_REG_OPER_OP_ORDER, "and" },
81 { 0x23, REG_OPER_OP_ORDER, "and" },
82 { 0x28, BYTE_OPER_REG_OP_ORDER, "sub" },
83 { 0x29, OPER_REG_OP_ORDER, "sub" },
84 { 0x2A, BYTE_REG_OPER_OP_ORDER, "sub" },
85 { 0x2B, REG_OPER_OP_ORDER, "sub" },
86 { 0x30, BYTE_OPER_REG_OP_ORDER, "xor" },
87 { 0x31, OPER_REG_OP_ORDER, "xor" },
88 { 0x32, BYTE_REG_OPER_OP_ORDER, "xor" },
89 { 0x33, REG_OPER_OP_ORDER, "xor" },
90 { 0x38, BYTE_OPER_REG_OP_ORDER, "cmp" },
91 { 0x39, OPER_REG_OP_ORDER, "cmp" },
92 { 0x3A, BYTE_REG_OPER_OP_ORDER, "cmp" },
93 { 0x3B, REG_OPER_OP_ORDER, "cmp" },
94 { 0x63, REG_OPER_OP_ORDER, "movsxlq" },
95 { 0x84, BYTE_REG_OPER_OP_ORDER, "test" },
96 { 0x85, REG_OPER_OP_ORDER, "test" },
97 { 0x86, BYTE_REG_OPER_OP_ORDER, "xchg" },
98 { 0x87, REG_OPER_OP_ORDER, "xchg" },
99 { 0x88, BYTE_OPER_REG_OP_ORDER, "mov" },
100 { 0x89, OPER_REG_OP_ORDER, "mov" },
101 { 0x8A, BYTE_REG_OPER_OP_ORDER, "mov" },
102 { 0x8B, REG_OPER_OP_ORDER, "mov" },
103 { 0x8D, REG_OPER_OP_ORDER, "lea" },
104 { -1, UNSET_OP_ORDER, "" }
105};
106
107
108static ByteMnemonic zero_operands_instr[] = {
109 { 0xC3, UNSET_OP_ORDER, "ret" },
110 { 0xC9, UNSET_OP_ORDER, "leave" },
111 { 0xF4, UNSET_OP_ORDER, "hlt" },
112 { 0xCC, UNSET_OP_ORDER, "int3" },
113 { 0x60, UNSET_OP_ORDER, "pushad" },
114 { 0x61, UNSET_OP_ORDER, "popad" },
115 { 0x9C, UNSET_OP_ORDER, "pushfd" },
116 { 0x9D, UNSET_OP_ORDER, "popfd" },
117 { 0x9E, UNSET_OP_ORDER, "sahf" },
118 { 0x99, UNSET_OP_ORDER, "cdq" },
119 { 0x9B, UNSET_OP_ORDER, "fwait" },
Leon Clarked91b9f72010-01-27 17:25:45 +0000120 { 0xA4, UNSET_OP_ORDER, "movs" },
121 { 0xA5, UNSET_OP_ORDER, "movs" },
122 { 0xA6, UNSET_OP_ORDER, "cmps" },
123 { 0xA7, UNSET_OP_ORDER, "cmps" },
Steve Blocka7e24c12009-10-30 11:49:00 +0000124 { -1, UNSET_OP_ORDER, "" }
125};
126
127
128static ByteMnemonic call_jump_instr[] = {
129 { 0xE8, UNSET_OP_ORDER, "call" },
130 { 0xE9, UNSET_OP_ORDER, "jmp" },
131 { -1, UNSET_OP_ORDER, "" }
132};
133
134
135static ByteMnemonic short_immediate_instr[] = {
136 { 0x05, UNSET_OP_ORDER, "add" },
137 { 0x0D, UNSET_OP_ORDER, "or" },
138 { 0x15, UNSET_OP_ORDER, "adc" },
139 { 0x1D, UNSET_OP_ORDER, "sbb" },
140 { 0x25, UNSET_OP_ORDER, "and" },
141 { 0x2D, UNSET_OP_ORDER, "sub" },
142 { 0x35, UNSET_OP_ORDER, "xor" },
143 { 0x3D, UNSET_OP_ORDER, "cmp" },
144 { -1, UNSET_OP_ORDER, "" }
145};
146
147
148static const char* conditional_code_suffix[] = {
149 "o", "no", "c", "nc", "z", "nz", "na", "a",
150 "s", "ns", "pe", "po", "l", "ge", "le", "g"
151};
152
153
154enum InstructionType {
155 NO_INSTR,
156 ZERO_OPERANDS_INSTR,
157 TWO_OPERANDS_INSTR,
158 JUMP_CONDITIONAL_SHORT_INSTR,
159 REGISTER_INSTR,
160 PUSHPOP_INSTR, // Has implicit 64-bit operand size.
161 MOVE_REG_INSTR,
162 CALL_JUMP_INSTR,
163 SHORT_IMMEDIATE_INSTR
164};
165
166
Leon Clarked91b9f72010-01-27 17:25:45 +0000167enum Prefixes {
168 ESCAPE_PREFIX = 0x0F,
169 OPERAND_SIZE_OVERRIDE_PREFIX = 0x66,
170 ADDRESS_SIZE_OVERRIDE_PREFIX = 0x67,
171 REPNE_PREFIX = 0xF2,
172 REP_PREFIX = 0xF3,
173 REPEQ_PREFIX = REP_PREFIX
174};
175
176
Steve Blocka7e24c12009-10-30 11:49:00 +0000177struct InstructionDesc {
178 const char* mnem;
179 InstructionType type;
180 OperandType op_order_;
181 bool byte_size_operation; // Fixed 8-bit operation.
182};
183
184
185class InstructionTable {
186 public:
187 InstructionTable();
188 const InstructionDesc& Get(byte x) const {
189 return instructions_[x];
190 }
191
192 private:
193 InstructionDesc instructions_[256];
194 void Clear();
195 void Init();
196 void CopyTable(ByteMnemonic bm[], InstructionType type);
197 void SetTableRange(InstructionType type, byte start, byte end, bool byte_size,
198 const char* mnem);
199 void AddJumpConditionalShort();
200};
201
202
203InstructionTable::InstructionTable() {
204 Clear();
205 Init();
206}
207
208
209void InstructionTable::Clear() {
210 for (int i = 0; i < 256; i++) {
211 instructions_[i].mnem = "(bad)";
212 instructions_[i].type = NO_INSTR;
213 instructions_[i].op_order_ = UNSET_OP_ORDER;
214 instructions_[i].byte_size_operation = false;
215 }
216}
217
218
219void InstructionTable::Init() {
220 CopyTable(two_operands_instr, TWO_OPERANDS_INSTR);
221 CopyTable(zero_operands_instr, ZERO_OPERANDS_INSTR);
222 CopyTable(call_jump_instr, CALL_JUMP_INSTR);
223 CopyTable(short_immediate_instr, SHORT_IMMEDIATE_INSTR);
224 AddJumpConditionalShort();
225 SetTableRange(PUSHPOP_INSTR, 0x50, 0x57, false, "push");
226 SetTableRange(PUSHPOP_INSTR, 0x58, 0x5F, false, "pop");
227 SetTableRange(MOVE_REG_INSTR, 0xB8, 0xBF, false, "mov");
228}
229
230
231void InstructionTable::CopyTable(ByteMnemonic bm[], InstructionType type) {
232 for (int i = 0; bm[i].b >= 0; i++) {
233 InstructionDesc* id = &instructions_[bm[i].b];
234 id->mnem = bm[i].mnem;
235 OperandType op_order = bm[i].op_order_;
236 id->op_order_ =
237 static_cast<OperandType>(op_order & ~BYTE_SIZE_OPERAND_FLAG);
Steve Blockd0582a62009-12-15 09:54:21 +0000238 ASSERT_EQ(NO_INSTR, id->type); // Information not already entered
Steve Blocka7e24c12009-10-30 11:49:00 +0000239 id->type = type;
240 id->byte_size_operation = ((op_order & BYTE_SIZE_OPERAND_FLAG) != 0);
241 }
242}
243
244
245void InstructionTable::SetTableRange(InstructionType type,
246 byte start,
247 byte end,
248 bool byte_size,
249 const char* mnem) {
250 for (byte b = start; b <= end; b++) {
251 InstructionDesc* id = &instructions_[b];
Steve Blockd0582a62009-12-15 09:54:21 +0000252 ASSERT_EQ(NO_INSTR, id->type); // Information not already entered
Steve Blocka7e24c12009-10-30 11:49:00 +0000253 id->mnem = mnem;
254 id->type = type;
255 id->byte_size_operation = byte_size;
256 }
257}
258
259
260void InstructionTable::AddJumpConditionalShort() {
261 for (byte b = 0x70; b <= 0x7F; b++) {
262 InstructionDesc* id = &instructions_[b];
Steve Blockd0582a62009-12-15 09:54:21 +0000263 ASSERT_EQ(NO_INSTR, id->type); // Information not already entered
Steve Blocka7e24c12009-10-30 11:49:00 +0000264 id->mnem = NULL; // Computed depending on condition code.
265 id->type = JUMP_CONDITIONAL_SHORT_INSTR;
266 }
267}
268
269
270static InstructionTable instruction_table;
271
272static InstructionDesc cmov_instructions[16] = {
273 {"cmovo", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
274 {"cmovno", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
275 {"cmovc", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
276 {"cmovnc", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
277 {"cmovz", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
278 {"cmovnz", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
279 {"cmovna", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
280 {"cmova", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
281 {"cmovs", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
282 {"cmovns", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
283 {"cmovpe", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
284 {"cmovpo", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
285 {"cmovl", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
286 {"cmovge", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
287 {"cmovle", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false},
288 {"cmovg", TWO_OPERANDS_INSTR, REG_OPER_OP_ORDER, false}
289};
290
291//------------------------------------------------------------------------------
292// DisassemblerX64 implementation.
293
294enum UnimplementedOpcodeAction {
295 CONTINUE_ON_UNIMPLEMENTED_OPCODE,
296 ABORT_ON_UNIMPLEMENTED_OPCODE
297};
298
299// A new DisassemblerX64 object is created to disassemble each instruction.
300// The object can only disassemble a single instruction.
301class DisassemblerX64 {
302 public:
303 DisassemblerX64(const NameConverter& converter,
304 UnimplementedOpcodeAction unimplemented_action =
305 ABORT_ON_UNIMPLEMENTED_OPCODE)
306 : converter_(converter),
307 tmp_buffer_pos_(0),
308 abort_on_unimplemented_(
309 unimplemented_action == ABORT_ON_UNIMPLEMENTED_OPCODE),
310 rex_(0),
311 operand_size_(0),
312 group_1_prefix_(0),
313 byte_size_operand_(false) {
314 tmp_buffer_[0] = '\0';
315 }
316
317 virtual ~DisassemblerX64() {
318 }
319
320 // Writes one disassembled instruction into 'buffer' (0-terminated).
321 // Returns the length of the disassembled machine instruction in bytes.
322 int InstructionDecode(v8::internal::Vector<char> buffer, byte* instruction);
323
324 private:
325 enum OperandSize {
326 BYTE_SIZE = 0,
327 WORD_SIZE = 1,
328 DOUBLEWORD_SIZE = 2,
329 QUADWORD_SIZE = 3
330 };
331
332 const NameConverter& converter_;
333 v8::internal::EmbeddedVector<char, 128> tmp_buffer_;
334 unsigned int tmp_buffer_pos_;
335 bool abort_on_unimplemented_;
336 // Prefixes parsed
337 byte rex_;
338 byte operand_size_; // 0x66 or (if no group 3 prefix is present) 0x0.
339 byte group_1_prefix_; // 0xF2, 0xF3, or (if no group 1 prefix is present) 0.
340 // Byte size operand override.
341 bool byte_size_operand_;
342
343 void setRex(byte rex) {
344 ASSERT_EQ(0x40, rex & 0xF0);
345 rex_ = rex;
346 }
347
348 bool rex() { return rex_ != 0; }
349
350 bool rex_b() { return (rex_ & 0x01) != 0; }
351
352 // Actual number of base register given the low bits and the rex.b state.
353 int base_reg(int low_bits) { return low_bits | ((rex_ & 0x01) << 3); }
354
355 bool rex_x() { return (rex_ & 0x02) != 0; }
356
357 bool rex_r() { return (rex_ & 0x04) != 0; }
358
359 bool rex_w() { return (rex_ & 0x08) != 0; }
360
361 OperandSize operand_size() {
362 if (byte_size_operand_) return BYTE_SIZE;
363 if (rex_w()) return QUADWORD_SIZE;
364 if (operand_size_ != 0) return WORD_SIZE;
365 return DOUBLEWORD_SIZE;
366 }
367
368 char operand_size_code() {
369 return "bwlq"[operand_size()];
370 }
371
372 const char* NameOfCPURegister(int reg) const {
373 return converter_.NameOfCPURegister(reg);
374 }
375
376 const char* NameOfByteCPURegister(int reg) const {
377 return converter_.NameOfByteCPURegister(reg);
378 }
379
380 const char* NameOfXMMRegister(int reg) const {
381 return converter_.NameOfXMMRegister(reg);
382 }
383
384 const char* NameOfAddress(byte* addr) const {
385 return converter_.NameOfAddress(addr);
386 }
387
388 // Disassembler helper functions.
389 void get_modrm(byte data,
390 int* mod,
391 int* regop,
392 int* rm) {
393 *mod = (data >> 6) & 3;
394 *regop = ((data & 0x38) >> 3) | (rex_r() ? 8 : 0);
395 *rm = (data & 7) | (rex_b() ? 8 : 0);
396 }
397
398 void get_sib(byte data,
399 int* scale,
400 int* index,
401 int* base) {
402 *scale = (data >> 6) & 3;
403 *index = ((data >> 3) & 7) | (rex_x() ? 8 : 0);
404 *base = (data & 7) | (rex_b() ? 8 : 0);
405 }
406
407 typedef const char* (DisassemblerX64::*RegisterNameMapping)(int reg) const;
408
409 int PrintRightOperandHelper(byte* modrmp,
410 RegisterNameMapping register_name);
411 int PrintRightOperand(byte* modrmp);
412 int PrintRightByteOperand(byte* modrmp);
Steve Blockd0582a62009-12-15 09:54:21 +0000413 int PrintRightXMMOperand(byte* modrmp);
Steve Blocka7e24c12009-10-30 11:49:00 +0000414 int PrintOperands(const char* mnem,
415 OperandType op_order,
416 byte* data);
417 int PrintImmediate(byte* data, OperandSize size);
418 int PrintImmediateOp(byte* data);
419 const char* TwoByteMnemonic(byte opcode);
420 int TwoByteOpcodeInstruction(byte* data);
Steve Blockd0582a62009-12-15 09:54:21 +0000421 int F6F7Instruction(byte* data);
Steve Blocka7e24c12009-10-30 11:49:00 +0000422 int ShiftInstruction(byte* data);
423 int JumpShort(byte* data);
424 int JumpConditional(byte* data);
425 int JumpConditionalShort(byte* data);
426 int SetCC(byte* data);
427 int FPUInstruction(byte* data);
Steve Blockd0582a62009-12-15 09:54:21 +0000428 int MemoryFPUInstruction(int escape_opcode, int regop, byte* modrm_start);
429 int RegisterFPUInstruction(int escape_opcode, byte modrm_byte);
Steve Blocka7e24c12009-10-30 11:49:00 +0000430 void AppendToBuffer(const char* format, ...);
431
432 void UnimplementedInstruction() {
433 if (abort_on_unimplemented_) {
434 CHECK(false);
435 } else {
436 AppendToBuffer("'Unimplemented Instruction'");
437 }
438 }
439};
440
441
442void DisassemblerX64::AppendToBuffer(const char* format, ...) {
443 v8::internal::Vector<char> buf = tmp_buffer_ + tmp_buffer_pos_;
444 va_list args;
445 va_start(args, format);
446 int result = v8::internal::OS::VSNPrintF(buf, format, args);
447 va_end(args);
448 tmp_buffer_pos_ += result;
449}
450
451
452int DisassemblerX64::PrintRightOperandHelper(
453 byte* modrmp,
454 RegisterNameMapping register_name) {
455 int mod, regop, rm;
456 get_modrm(*modrmp, &mod, &regop, &rm);
457 switch (mod) {
458 case 0:
459 if ((rm & 7) == 5) {
460 int32_t disp = *reinterpret_cast<int32_t*>(modrmp + 1);
461 AppendToBuffer("[0x%x]", disp);
462 return 5;
463 } else if ((rm & 7) == 4) {
464 // Codes for SIB byte.
465 byte sib = *(modrmp + 1);
466 int scale, index, base;
467 get_sib(sib, &scale, &index, &base);
468 if (index == 4 && (base & 7) == 4 && scale == 0 /*times_1*/) {
469 // index == rsp means no index. Only use sib byte with no index for
470 // rsp and r12 base.
471 AppendToBuffer("[%s]", (this->*register_name)(base));
472 return 2;
473 } else if (base == 5) {
474 // base == rbp means no base register (when mod == 0).
475 int32_t disp = *reinterpret_cast<int32_t*>(modrmp + 2);
476 AppendToBuffer("[%s*%d+0x%x]",
477 (this->*register_name)(index),
478 1 << scale, disp);
479 return 6;
480 } else if (index != 4 && base != 5) {
481 // [base+index*scale]
482 AppendToBuffer("[%s+%s*%d]",
483 (this->*register_name)(base),
484 (this->*register_name)(index),
485 1 << scale);
486 return 2;
487 } else {
488 UnimplementedInstruction();
489 return 1;
490 }
491 } else {
492 AppendToBuffer("[%s]", (this->*register_name)(rm));
493 return 1;
494 }
495 break;
496 case 1: // fall through
497 case 2:
498 if ((rm & 7) == 4) {
499 byte sib = *(modrmp + 1);
500 int scale, index, base;
501 get_sib(sib, &scale, &index, &base);
502 int disp = (mod == 2) ? *reinterpret_cast<int32_t*>(modrmp + 2)
503 : *reinterpret_cast<char*>(modrmp + 2);
504 if (index == 4 && (base & 7) == 4 && scale == 0 /*times_1*/) {
505 if (-disp > 0) {
506 AppendToBuffer("[%s-0x%x]", (this->*register_name)(base), -disp);
507 } else {
508 AppendToBuffer("[%s+0x%x]", (this->*register_name)(base), disp);
509 }
510 } else {
511 if (-disp > 0) {
512 AppendToBuffer("[%s+%s*%d-0x%x]",
513 (this->*register_name)(base),
514 (this->*register_name)(index),
515 1 << scale,
516 -disp);
517 } else {
518 AppendToBuffer("[%s+%s*%d+0x%x]",
519 (this->*register_name)(base),
520 (this->*register_name)(index),
521 1 << scale,
522 disp);
523 }
524 }
525 return mod == 2 ? 6 : 3;
526 } else {
527 // No sib.
528 int disp = (mod == 2) ? *reinterpret_cast<int32_t*>(modrmp + 1)
529 : *reinterpret_cast<char*>(modrmp + 1);
530 if (-disp > 0) {
531 AppendToBuffer("[%s-0x%x]", (this->*register_name)(rm), -disp);
532 } else {
533 AppendToBuffer("[%s+0x%x]", (this->*register_name)(rm), disp);
534 }
535 return (mod == 2) ? 5 : 2;
536 }
537 break;
538 case 3:
539 AppendToBuffer("%s", (this->*register_name)(rm));
540 return 1;
541 default:
542 UnimplementedInstruction();
543 return 1;
544 }
545 UNREACHABLE();
546}
547
548
549int DisassemblerX64::PrintImmediate(byte* data, OperandSize size) {
550 int64_t value;
551 int count;
552 switch (size) {
553 case BYTE_SIZE:
554 value = *data;
555 count = 1;
556 break;
557 case WORD_SIZE:
558 value = *reinterpret_cast<int16_t*>(data);
559 count = 2;
560 break;
561 case DOUBLEWORD_SIZE:
562 value = *reinterpret_cast<uint32_t*>(data);
563 count = 4;
564 break;
565 case QUADWORD_SIZE:
566 value = *reinterpret_cast<int32_t*>(data);
567 count = 4;
568 break;
569 default:
570 UNREACHABLE();
571 value = 0; // Initialize variables on all paths to satisfy the compiler.
572 count = 0;
573 }
574 AppendToBuffer("%" V8_PTR_PREFIX "x", value);
575 return count;
576}
577
578
579int DisassemblerX64::PrintRightOperand(byte* modrmp) {
580 return PrintRightOperandHelper(modrmp,
581 &DisassemblerX64::NameOfCPURegister);
582}
583
584
585int DisassemblerX64::PrintRightByteOperand(byte* modrmp) {
586 return PrintRightOperandHelper(modrmp,
587 &DisassemblerX64::NameOfByteCPURegister);
588}
589
590
Steve Blockd0582a62009-12-15 09:54:21 +0000591int DisassemblerX64::PrintRightXMMOperand(byte* modrmp) {
592 return PrintRightOperandHelper(modrmp,
593 &DisassemblerX64::NameOfXMMRegister);
594}
595
596
Steve Blocka7e24c12009-10-30 11:49:00 +0000597// Returns number of bytes used including the current *data.
598// Writes instruction's mnemonic, left and right operands to 'tmp_buffer_'.
599int DisassemblerX64::PrintOperands(const char* mnem,
600 OperandType op_order,
601 byte* data) {
602 byte modrm = *data;
603 int mod, regop, rm;
604 get_modrm(modrm, &mod, &regop, &rm);
605 int advance = 0;
606 const char* register_name =
607 byte_size_operand_ ? NameOfByteCPURegister(regop)
608 : NameOfCPURegister(regop);
609 switch (op_order) {
610 case REG_OPER_OP_ORDER: {
611 AppendToBuffer("%s%c %s,",
612 mnem,
613 operand_size_code(),
614 register_name);
615 advance = byte_size_operand_ ? PrintRightByteOperand(data)
616 : PrintRightOperand(data);
617 break;
618 }
619 case OPER_REG_OP_ORDER: {
620 AppendToBuffer("%s%c ", mnem, operand_size_code());
621 advance = byte_size_operand_ ? PrintRightByteOperand(data)
622 : PrintRightOperand(data);
623 AppendToBuffer(",%s", register_name);
624 break;
625 }
626 default:
627 UNREACHABLE();
628 break;
629 }
630 return advance;
631}
632
633
634// Returns number of bytes used by machine instruction, including *data byte.
635// Writes immediate instructions to 'tmp_buffer_'.
636int DisassemblerX64::PrintImmediateOp(byte* data) {
637 bool byte_size_immediate = (*data & 0x02) != 0;
638 byte modrm = *(data + 1);
639 int mod, regop, rm;
640 get_modrm(modrm, &mod, &regop, &rm);
641 const char* mnem = "Imm???";
642 switch (regop) {
643 case 0:
644 mnem = "add";
645 break;
646 case 1:
647 mnem = "or";
648 break;
649 case 2:
650 mnem = "adc";
651 break;
652 case 4:
653 mnem = "and";
654 break;
655 case 5:
656 mnem = "sub";
657 break;
658 case 6:
659 mnem = "xor";
660 break;
661 case 7:
662 mnem = "cmp";
663 break;
664 default:
665 UnimplementedInstruction();
666 }
667 AppendToBuffer("%s%c ", mnem, operand_size_code());
668 int count = PrintRightOperand(data + 1);
669 AppendToBuffer(",0x");
670 OperandSize immediate_size = byte_size_immediate ? BYTE_SIZE : operand_size();
671 count += PrintImmediate(data + 1 + count, immediate_size);
672 return 1 + count;
673}
674
675
676// Returns number of bytes used, including *data.
Steve Blockd0582a62009-12-15 09:54:21 +0000677int DisassemblerX64::F6F7Instruction(byte* data) {
678 ASSERT(*data == 0xF7 || *data == 0xF6);
Steve Blocka7e24c12009-10-30 11:49:00 +0000679 byte modrm = *(data + 1);
680 int mod, regop, rm;
681 get_modrm(modrm, &mod, &regop, &rm);
682 if (mod == 3 && regop != 0) {
683 const char* mnem = NULL;
684 switch (regop) {
685 case 2:
686 mnem = "not";
687 break;
688 case 3:
689 mnem = "neg";
690 break;
691 case 4:
692 mnem = "mul";
693 break;
694 case 7:
695 mnem = "idiv";
696 break;
697 default:
698 UnimplementedInstruction();
699 }
700 AppendToBuffer("%s%c %s",
701 mnem,
702 operand_size_code(),
703 NameOfCPURegister(rm));
704 return 2;
Steve Blocka7e24c12009-10-30 11:49:00 +0000705 } else if (regop == 0) {
706 AppendToBuffer("test%c ", operand_size_code());
Steve Blockd0582a62009-12-15 09:54:21 +0000707 int count = PrintRightOperand(data + 1); // Use name of 64-bit register.
708 AppendToBuffer(",0x");
709 count += PrintImmediate(data + 1 + count, operand_size());
710 return 1 + count;
Steve Blocka7e24c12009-10-30 11:49:00 +0000711 } else {
712 UnimplementedInstruction();
713 return 2;
714 }
715}
716
717
718int DisassemblerX64::ShiftInstruction(byte* data) {
719 byte op = *data & (~1);
720 if (op != 0xD0 && op != 0xD2 && op != 0xC0) {
721 UnimplementedInstruction();
722 return 1;
723 }
724 byte modrm = *(data + 1);
725 int mod, regop, rm;
726 get_modrm(modrm, &mod, &regop, &rm);
727 regop &= 0x7; // The REX.R bit does not affect the operation.
728 int imm8 = -1;
729 int num_bytes = 2;
730 if (mod != 3) {
731 UnimplementedInstruction();
732 return num_bytes;
733 }
734 const char* mnem = NULL;
735 switch (regop) {
736 case 0:
737 mnem = "rol";
738 break;
739 case 1:
740 mnem = "ror";
741 break;
742 case 2:
743 mnem = "rcl";
744 break;
745 case 3:
746 mnem = "rcr";
747 break;
748 case 4:
749 mnem = "shl";
750 break;
751 case 5:
752 mnem = "shr";
753 break;
754 case 7:
755 mnem = "sar";
756 break;
757 default:
758 UnimplementedInstruction();
759 return num_bytes;
760 }
Steve Blockd0582a62009-12-15 09:54:21 +0000761 ASSERT_NE(NULL, mnem);
Steve Blocka7e24c12009-10-30 11:49:00 +0000762 if (op == 0xD0) {
763 imm8 = 1;
764 } else if (op == 0xC0) {
765 imm8 = *(data + 2);
766 num_bytes = 3;
767 }
768 AppendToBuffer("%s%c %s,",
769 mnem,
770 operand_size_code(),
771 byte_size_operand_ ? NameOfByteCPURegister(rm)
772 : NameOfCPURegister(rm));
773 if (op == 0xD2) {
774 AppendToBuffer("cl");
775 } else {
776 AppendToBuffer("%d", imm8);
777 }
778 return num_bytes;
779}
780
781
782// Returns number of bytes used, including *data.
783int DisassemblerX64::JumpShort(byte* data) {
Steve Blockd0582a62009-12-15 09:54:21 +0000784 ASSERT_EQ(0xEB, *data);
Steve Blocka7e24c12009-10-30 11:49:00 +0000785 byte b = *(data + 1);
786 byte* dest = data + static_cast<int8_t>(b) + 2;
787 AppendToBuffer("jmp %s", NameOfAddress(dest));
788 return 2;
789}
790
791
792// Returns number of bytes used, including *data.
793int DisassemblerX64::JumpConditional(byte* data) {
Steve Blockd0582a62009-12-15 09:54:21 +0000794 ASSERT_EQ(0x0F, *data);
Steve Blocka7e24c12009-10-30 11:49:00 +0000795 byte cond = *(data + 1) & 0x0F;
796 byte* dest = data + *reinterpret_cast<int32_t*>(data + 2) + 6;
797 const char* mnem = conditional_code_suffix[cond];
798 AppendToBuffer("j%s %s", mnem, NameOfAddress(dest));
799 return 6; // includes 0x0F
800}
801
802
803// Returns number of bytes used, including *data.
804int DisassemblerX64::JumpConditionalShort(byte* data) {
805 byte cond = *data & 0x0F;
806 byte b = *(data + 1);
807 byte* dest = data + static_cast<int8_t>(b) + 2;
808 const char* mnem = conditional_code_suffix[cond];
809 AppendToBuffer("j%s %s", mnem, NameOfAddress(dest));
810 return 2;
811}
812
813
814// Returns number of bytes used, including *data.
815int DisassemblerX64::SetCC(byte* data) {
Steve Blockd0582a62009-12-15 09:54:21 +0000816 ASSERT_EQ(0x0F, *data);
Steve Blocka7e24c12009-10-30 11:49:00 +0000817 byte cond = *(data + 1) & 0x0F;
818 const char* mnem = conditional_code_suffix[cond];
819 AppendToBuffer("set%s%c ", mnem, operand_size_code());
820 PrintRightByteOperand(data + 2);
821 return 3; // includes 0x0F
822}
823
824
825// Returns number of bytes used, including *data.
826int DisassemblerX64::FPUInstruction(byte* data) {
Steve Blockd0582a62009-12-15 09:54:21 +0000827 byte escape_opcode = *data;
828 ASSERT_EQ(0xD8, escape_opcode & 0xF8);
829 byte modrm_byte = *(data+1);
830
831 if (modrm_byte >= 0xC0) {
832 return RegisterFPUInstruction(escape_opcode, modrm_byte);
833 } else {
834 return MemoryFPUInstruction(escape_opcode, modrm_byte, data+1);
Steve Blocka7e24c12009-10-30 11:49:00 +0000835 }
Steve Blockd0582a62009-12-15 09:54:21 +0000836}
837
838int DisassemblerX64::MemoryFPUInstruction(int escape_opcode,
839 int modrm_byte,
840 byte* modrm_start) {
841 const char* mnem = "?";
842 int regop = (modrm_byte >> 3) & 0x7; // reg/op field of modrm byte.
843 switch (escape_opcode) {
844 case 0xD9: switch (regop) {
845 case 0: mnem = "fld_s"; break;
846 case 3: mnem = "fstp_s"; break;
847 case 7: mnem = "fstcw"; break;
848 default: UnimplementedInstruction();
849 }
850 break;
851
852 case 0xDB: switch (regop) {
853 case 0: mnem = "fild_s"; break;
854 case 1: mnem = "fisttp_s"; break;
855 case 2: mnem = "fist_s"; break;
856 case 3: mnem = "fistp_s"; break;
857 default: UnimplementedInstruction();
858 }
859 break;
860
861 case 0xDD: switch (regop) {
862 case 0: mnem = "fld_d"; break;
863 case 3: mnem = "fstp_d"; break;
864 default: UnimplementedInstruction();
865 }
866 break;
867
868 case 0xDF: switch (regop) {
869 case 5: mnem = "fild_d"; break;
870 case 7: mnem = "fistp_d"; break;
871 default: UnimplementedInstruction();
872 }
873 break;
874
875 default: UnimplementedInstruction();
876 }
877 AppendToBuffer("%s ", mnem);
878 int count = PrintRightOperand(modrm_start);
879 return count + 1;
880}
881
882int DisassemblerX64::RegisterFPUInstruction(int escape_opcode,
883 byte modrm_byte) {
884 bool has_register = false; // Is the FPU register encoded in modrm_byte?
885 const char* mnem = "?";
886
887 switch (escape_opcode) {
888 case 0xD8:
889 UnimplementedInstruction();
890 break;
891
892 case 0xD9:
893 switch (modrm_byte & 0xF8) {
894 case 0xC8:
895 mnem = "fxch";
896 has_register = true;
897 break;
898 default:
899 switch (modrm_byte) {
900 case 0xE0: mnem = "fchs"; break;
901 case 0xE1: mnem = "fabs"; break;
902 case 0xE4: mnem = "ftst"; break;
903 case 0xE8: mnem = "fld1"; break;
904 case 0xEE: mnem = "fldz"; break;
905 case 0xF5: mnem = "fprem1"; break;
906 case 0xF7: mnem = "fincstp"; break;
907 case 0xF8: mnem = "fprem"; break;
908 case 0xFE: mnem = "fsin"; break;
909 case 0xFF: mnem = "fcos"; break;
910 default: UnimplementedInstruction();
911 }
912 }
913 break;
914
915 case 0xDA:
916 if (modrm_byte == 0xE9) {
917 mnem = "fucompp";
918 } else {
919 UnimplementedInstruction();
920 }
921 break;
922
923 case 0xDB:
924 if ((modrm_byte & 0xF8) == 0xE8) {
925 mnem = "fucomi";
926 has_register = true;
927 } else if (modrm_byte == 0xE2) {
928 mnem = "fclex";
929 } else {
930 UnimplementedInstruction();
931 }
932 break;
933
934 case 0xDC:
935 has_register = true;
936 switch (modrm_byte & 0xF8) {
937 case 0xC0: mnem = "fadd"; break;
938 case 0xE8: mnem = "fsub"; break;
939 case 0xC8: mnem = "fmul"; break;
940 case 0xF8: mnem = "fdiv"; break;
941 default: UnimplementedInstruction();
942 }
943 break;
944
945 case 0xDD:
946 has_register = true;
947 switch (modrm_byte & 0xF8) {
948 case 0xC0: mnem = "ffree"; break;
949 case 0xD8: mnem = "fstp"; break;
950 default: UnimplementedInstruction();
951 }
952 break;
953
954 case 0xDE:
955 if (modrm_byte == 0xD9) {
956 mnem = "fcompp";
957 } else {
958 has_register = true;
959 switch (modrm_byte & 0xF8) {
960 case 0xC0: mnem = "faddp"; break;
961 case 0xE8: mnem = "fsubp"; break;
962 case 0xC8: mnem = "fmulp"; break;
963 case 0xF8: mnem = "fdivp"; break;
964 default: UnimplementedInstruction();
965 }
966 }
967 break;
968
969 case 0xDF:
970 if (modrm_byte == 0xE0) {
971 mnem = "fnstsw_ax";
972 } else if ((modrm_byte & 0xF8) == 0xE8) {
973 mnem = "fucomip";
974 has_register = true;
975 }
976 break;
977
978 default: UnimplementedInstruction();
979 }
980
981 if (has_register) {
982 AppendToBuffer("%s st%d", mnem, modrm_byte & 0x7);
983 } else {
984 AppendToBuffer("%s", mnem);
985 }
Steve Blocka7e24c12009-10-30 11:49:00 +0000986 return 2;
987}
988
989
Steve Blockd0582a62009-12-15 09:54:21 +0000990
Steve Blocka7e24c12009-10-30 11:49:00 +0000991// Handle all two-byte opcodes, which start with 0x0F.
992// These instructions may be affected by an 0x66, 0xF2, or 0xF3 prefix.
993// We do not use any three-byte opcodes, which start with 0x0F38 or 0x0F3A.
994int DisassemblerX64::TwoByteOpcodeInstruction(byte* data) {
995 byte opcode = *(data + 1);
996 byte* current = data + 2;
997 // At return, "current" points to the start of the next instruction.
998 const char* mnemonic = TwoByteMnemonic(opcode);
Andrei Popescu402d9372010-02-26 13:31:12 +0000999 if (operand_size_ == 0x66) {
1000 // 0x66 0x0F prefix.
Steve Blocka7e24c12009-10-30 11:49:00 +00001001 int mod, regop, rm;
Steve Block6ded16b2010-05-10 14:33:55 +01001002 if (opcode == 0x3A) {
1003 byte third_byte = *current;
1004 current = data + 3;
1005 if (third_byte == 0x17) {
1006 get_modrm(*current, &mod, &regop, &rm);
1007 AppendToBuffer("extractps "); // reg/m32, xmm, imm8
1008 current += PrintRightOperand(current);
1009 AppendToBuffer(", %s, %d", NameOfCPURegister(regop), (*current) & 3);
1010 current += 1;
1011 } else {
1012 UnimplementedInstruction();
1013 }
Steve Blocka7e24c12009-10-30 11:49:00 +00001014 } else {
Steve Block6ded16b2010-05-10 14:33:55 +01001015 get_modrm(*current, &mod, &regop, &rm);
1016 if (opcode == 0x6E) {
1017 AppendToBuffer("mov%c %s,",
1018 rex_w() ? 'q' : 'd',
1019 NameOfXMMRegister(regop));
1020 current += PrintRightOperand(current);
1021 } else if (opcode == 0x7E) {
1022 AppendToBuffer("mov%c %s,",
1023 rex_w() ? 'q' : 'd',
1024 NameOfCPURegister(regop));
1025 current += PrintRightXMMOperand(current);
1026 } else {
1027 const char* mnemonic = "?";
1028 if (opcode == 0x57) {
1029 mnemonic = "xorpd";
1030 } else if (opcode == 0x2E) {
1031 mnemonic = "comisd";
1032 } else if (opcode == 0x2F) {
1033 mnemonic = "ucomisd";
1034 } else {
1035 UnimplementedInstruction();
1036 }
1037 AppendToBuffer("%s %s,", mnemonic, NameOfXMMRegister(regop));
1038 current += PrintRightXMMOperand(current);
1039 }
Steve Blocka7e24c12009-10-30 11:49:00 +00001040 }
1041 } else if (group_1_prefix_ == 0xF2) {
1042 // Beginning of instructions with prefix 0xF2.
1043
1044 if (opcode == 0x11 || opcode == 0x10) {
1045 // MOVSD: Move scalar double-precision fp to/from/between XMM registers.
1046 AppendToBuffer("movsd ");
1047 int mod, regop, rm;
1048 get_modrm(*current, &mod, &regop, &rm);
1049 if (opcode == 0x11) {
1050 current += PrintRightOperand(current);
1051 AppendToBuffer(",%s", NameOfXMMRegister(regop));
1052 } else {
1053 AppendToBuffer("%s,", NameOfXMMRegister(regop));
1054 current += PrintRightOperand(current);
1055 }
1056 } else if (opcode == 0x2A) {
1057 // CVTSI2SD: integer to XMM double conversion.
1058 int mod, regop, rm;
1059 get_modrm(*current, &mod, &regop, &rm);
1060 AppendToBuffer("%s %s,", mnemonic, NameOfXMMRegister(regop));
Steve Blockd0582a62009-12-15 09:54:21 +00001061 current += PrintRightOperand(current);
Steve Block6ded16b2010-05-10 14:33:55 +01001062 } else if ((opcode & 0xF8) == 0x58 || opcode == 0x51) {
Steve Blocka7e24c12009-10-30 11:49:00 +00001063 // XMM arithmetic. Mnemonic was retrieved at the start of this function.
1064 int mod, regop, rm;
1065 get_modrm(*current, &mod, &regop, &rm);
Steve Blockd0582a62009-12-15 09:54:21 +00001066 AppendToBuffer("%s %s,", mnemonic, NameOfXMMRegister(regop));
1067 current += PrintRightXMMOperand(current);
Steve Blocka7e24c12009-10-30 11:49:00 +00001068 } else {
1069 UnimplementedInstruction();
1070 }
Steve Block6ded16b2010-05-10 14:33:55 +01001071 } else if (group_1_prefix_ == 0xF3) {
1072 // Instructions with prefix 0xF3.
1073 if (opcode == 0x2C) {
1074 // CVTTSS2SI: Convert scalar single-precision FP to dword integer.
1075 // Assert that mod is not 3, so source is memory, not an XMM register.
1076 ASSERT_NE(0xC0, *current & 0xC0);
1077 current += PrintOperands("cvttss2si", REG_OPER_OP_ORDER, current);
1078 } else if (opcode == 0x5A) {
1079 int mod, regop, rm;
1080 get_modrm(*current, &mod, &regop, &rm);
1081 AppendToBuffer("cvtss2sd %s,", NameOfXMMRegister(regop));
1082 current += PrintRightXMMOperand(current);
1083 } else {
1084 UnimplementedInstruction();
1085 }
Andrei Popescu402d9372010-02-26 13:31:12 +00001086 } else if (opcode == 0x1F) {
1087 // NOP
1088 int mod, regop, rm;
1089 get_modrm(*current, &mod, &regop, &rm);
1090 current++;
1091 if (regop == 4) { // SIB byte present.
1092 current++;
1093 }
1094 if (mod == 1) { // Byte displacement.
1095 current += 1;
1096 } else if (mod == 2) { // 32-bit displacement.
1097 current += 4;
1098 } // else no immediate displacement.
1099 AppendToBuffer("nop");
1100 } else if (opcode == 0xA2 || opcode == 0x31) {
1101 // RDTSC or CPUID
1102 AppendToBuffer("%s", mnemonic);
1103
1104 } else if ((opcode & 0xF0) == 0x40) {
1105 // CMOVcc: conditional move.
1106 int condition = opcode & 0x0F;
1107 const InstructionDesc& idesc = cmov_instructions[condition];
1108 byte_size_operand_ = idesc.byte_size_operation;
1109 current += PrintOperands(idesc.mnem, idesc.op_order_, current);
1110
1111 } else if ((opcode & 0xF0) == 0x80) {
1112 // Jcc: Conditional jump (branch).
1113 current = data + JumpConditional(data);
1114
1115 } else if (opcode == 0xBE || opcode == 0xBF || opcode == 0xB6 ||
1116 opcode == 0xB7 || opcode == 0xAF) {
1117 // Size-extending moves, IMUL.
1118 current += PrintOperands(mnemonic, REG_OPER_OP_ORDER, current);
1119
1120 } else if ((opcode & 0xF0) == 0x90) {
1121 // SETcc: Set byte on condition. Needs pointer to beginning of instruction.
1122 current = data + SetCC(data);
1123
1124 } else if (opcode == 0xAB || opcode == 0xA5 || opcode == 0xAD) {
1125 // SHLD, SHRD (double-precision shift), BTS (bit set).
1126 AppendToBuffer("%s ", mnemonic);
1127 int mod, regop, rm;
1128 get_modrm(*current, &mod, &regop, &rm);
1129 current += PrintRightOperand(current);
1130 if (opcode == 0xAB) {
1131 AppendToBuffer(",%s", NameOfCPURegister(regop));
1132 } else {
1133 AppendToBuffer(",%s,cl", NameOfCPURegister(regop));
1134 }
Steve Blocka7e24c12009-10-30 11:49:00 +00001135 } else {
1136 UnimplementedInstruction();
1137 }
Steve Blockd0582a62009-12-15 09:54:21 +00001138 return static_cast<int>(current - data);
Steve Blocka7e24c12009-10-30 11:49:00 +00001139}
1140
1141
1142// Mnemonics for two-byte opcode instructions starting with 0x0F.
1143// The argument is the second byte of the two-byte opcode.
1144// Returns NULL if the instruction is not handled here.
1145const char* DisassemblerX64::TwoByteMnemonic(byte opcode) {
1146 switch (opcode) {
1147 case 0x1F:
1148 return "nop";
1149 case 0x2A: // F2 prefix.
1150 return "cvtsi2sd";
1151 case 0x31:
1152 return "rdtsc";
Steve Block6ded16b2010-05-10 14:33:55 +01001153 case 0x51: // F2 prefix.
1154 return "sqrtsd";
Steve Blocka7e24c12009-10-30 11:49:00 +00001155 case 0x58: // F2 prefix.
1156 return "addsd";
1157 case 0x59: // F2 prefix.
1158 return "mulsd";
1159 case 0x5C: // F2 prefix.
1160 return "subsd";
1161 case 0x5E: // F2 prefix.
1162 return "divsd";
1163 case 0xA2:
1164 return "cpuid";
1165 case 0xA5:
1166 return "shld";
1167 case 0xAB:
1168 return "bts";
1169 case 0xAD:
1170 return "shrd";
1171 case 0xAF:
1172 return "imul";
1173 case 0xB6:
1174 return "movzxb";
1175 case 0xB7:
1176 return "movzxw";
1177 case 0xBE:
1178 return "movsxb";
1179 case 0xBF:
1180 return "movsxw";
1181 default:
1182 return NULL;
1183 }
1184}
1185
1186
1187// Disassembles the instruction at instr, and writes it into out_buffer.
1188int DisassemblerX64::InstructionDecode(v8::internal::Vector<char> out_buffer,
1189 byte* instr) {
1190 tmp_buffer_pos_ = 0; // starting to write as position 0
1191 byte* data = instr;
1192 bool processed = true; // Will be set to false if the current instruction
1193 // is not in 'instructions' table.
1194 byte current;
1195
1196 // Scan for prefixes.
1197 while (true) {
1198 current = *data;
Leon Clarked91b9f72010-01-27 17:25:45 +00001199 if (current == OPERAND_SIZE_OVERRIDE_PREFIX) { // Group 3 prefix.
Steve Blocka7e24c12009-10-30 11:49:00 +00001200 operand_size_ = current;
1201 } else if ((current & 0xF0) == 0x40) { // REX prefix.
1202 setRex(current);
1203 if (rex_w()) AppendToBuffer("REX.W ");
Leon Clarked91b9f72010-01-27 17:25:45 +00001204 } else if ((current & 0xFE) == 0xF2) { // Group 1 prefix (0xF2 or 0xF3).
Steve Blocka7e24c12009-10-30 11:49:00 +00001205 group_1_prefix_ = current;
1206 } else { // Not a prefix - an opcode.
1207 break;
1208 }
1209 data++;
1210 }
1211
1212 const InstructionDesc& idesc = instruction_table.Get(current);
1213 byte_size_operand_ = idesc.byte_size_operation;
1214 switch (idesc.type) {
1215 case ZERO_OPERANDS_INSTR:
Leon Clarked91b9f72010-01-27 17:25:45 +00001216 if (current >= 0xA4 && current <= 0xA7) {
1217 // String move or compare operations.
1218 if (group_1_prefix_ == REP_PREFIX) {
1219 // REP.
1220 AppendToBuffer("rep ");
1221 }
1222 if (rex_w()) AppendToBuffer("REX.W ");
1223 AppendToBuffer("%s%c", idesc.mnem, operand_size_code());
1224 } else {
1225 AppendToBuffer("%s", idesc.mnem, operand_size_code());
1226 }
Steve Blocka7e24c12009-10-30 11:49:00 +00001227 data++;
1228 break;
1229
1230 case TWO_OPERANDS_INSTR:
1231 data++;
1232 data += PrintOperands(idesc.mnem, idesc.op_order_, data);
1233 break;
1234
1235 case JUMP_CONDITIONAL_SHORT_INSTR:
1236 data += JumpConditionalShort(data);
1237 break;
1238
1239 case REGISTER_INSTR:
1240 AppendToBuffer("%s%c %s",
1241 idesc.mnem,
1242 operand_size_code(),
1243 NameOfCPURegister(base_reg(current & 0x07)));
1244 data++;
1245 break;
1246 case PUSHPOP_INSTR:
1247 AppendToBuffer("%s %s",
1248 idesc.mnem,
1249 NameOfCPURegister(base_reg(current & 0x07)));
1250 data++;
1251 break;
1252 case MOVE_REG_INSTR: {
1253 byte* addr = NULL;
1254 switch (operand_size()) {
1255 case WORD_SIZE:
1256 addr = reinterpret_cast<byte*>(*reinterpret_cast<int16_t*>(data + 1));
1257 data += 3;
1258 break;
1259 case DOUBLEWORD_SIZE:
1260 addr = reinterpret_cast<byte*>(*reinterpret_cast<int32_t*>(data + 1));
1261 data += 5;
1262 break;
1263 case QUADWORD_SIZE:
1264 addr = reinterpret_cast<byte*>(*reinterpret_cast<int64_t*>(data + 1));
1265 data += 9;
1266 break;
1267 default:
1268 UNREACHABLE();
1269 }
1270 AppendToBuffer("mov%c %s,%s",
1271 operand_size_code(),
1272 NameOfCPURegister(base_reg(current & 0x07)),
1273 NameOfAddress(addr));
1274 break;
1275 }
1276
1277 case CALL_JUMP_INSTR: {
1278 byte* addr = data + *reinterpret_cast<int32_t*>(data + 1) + 5;
1279 AppendToBuffer("%s %s", idesc.mnem, NameOfAddress(addr));
1280 data += 5;
1281 break;
1282 }
1283
1284 case SHORT_IMMEDIATE_INSTR: {
1285 byte* addr =
1286 reinterpret_cast<byte*>(*reinterpret_cast<int32_t*>(data + 1));
1287 AppendToBuffer("%s rax, %s", idesc.mnem, NameOfAddress(addr));
1288 data += 5;
1289 break;
1290 }
1291
1292 case NO_INSTR:
1293 processed = false;
1294 break;
1295
1296 default:
1297 UNIMPLEMENTED(); // This type is not implemented.
1298 }
1299
1300 // The first byte didn't match any of the simple opcodes, so we
1301 // need to do special processing on it.
1302 if (!processed) {
1303 switch (*data) {
1304 case 0xC2:
1305 AppendToBuffer("ret 0x%x", *reinterpret_cast<uint16_t*>(data + 1));
1306 data += 3;
1307 break;
1308
1309 case 0x69: // fall through
1310 case 0x6B: {
1311 int mod, regop, rm;
1312 get_modrm(*(data + 1), &mod, &regop, &rm);
1313 int32_t imm = *data == 0x6B ? *(data + 2)
1314 : *reinterpret_cast<int32_t*>(data + 2);
Steve Block6ded16b2010-05-10 14:33:55 +01001315 AppendToBuffer("imul%c %s,%s,0x%x",
1316 operand_size_code(),
1317 NameOfCPURegister(regop),
Steve Blocka7e24c12009-10-30 11:49:00 +00001318 NameOfCPURegister(rm), imm);
1319 data += 2 + (*data == 0x6B ? 1 : 4);
1320 break;
1321 }
1322
Steve Blocka7e24c12009-10-30 11:49:00 +00001323 case 0x81: // fall through
1324 case 0x83: // 0x81 with sign extension bit set
1325 data += PrintImmediateOp(data);
1326 break;
1327
1328 case 0x0F:
1329 data += TwoByteOpcodeInstruction(data);
1330 break;
1331
1332 case 0x8F: {
1333 data++;
1334 int mod, regop, rm;
1335 get_modrm(*data, &mod, &regop, &rm);
1336 if (regop == 0) {
1337 AppendToBuffer("pop ");
1338 data += PrintRightOperand(data);
1339 }
1340 }
1341 break;
1342
1343 case 0xFF: {
1344 data++;
1345 int mod, regop, rm;
1346 get_modrm(*data, &mod, &regop, &rm);
1347 const char* mnem = NULL;
1348 switch (regop) {
1349 case 0:
1350 mnem = "inc";
1351 break;
1352 case 1:
1353 mnem = "dec";
1354 break;
1355 case 2:
1356 mnem = "call";
1357 break;
1358 case 4:
1359 mnem = "jmp";
1360 break;
1361 case 6:
1362 mnem = "push";
1363 break;
1364 default:
1365 mnem = "???";
1366 }
1367 AppendToBuffer(((regop <= 1) ? "%s%c " : "%s "),
1368 mnem,
1369 operand_size_code());
1370 data += PrintRightOperand(data);
1371 }
1372 break;
1373
1374 case 0xC7: // imm32, fall through
1375 case 0xC6: // imm8
1376 {
1377 bool is_byte = *data == 0xC6;
1378 data++;
1379
1380 AppendToBuffer("mov%c ", is_byte ? 'b' : operand_size_code());
1381 data += PrintRightOperand(data);
1382 int32_t imm = is_byte ? *data : *reinterpret_cast<int32_t*>(data);
1383 AppendToBuffer(",0x%x", imm);
1384 data += is_byte ? 1 : 4;
1385 }
1386 break;
1387
1388 case 0x80: {
1389 data++;
1390 AppendToBuffer("cmpb ");
1391 data += PrintRightOperand(data);
1392 int32_t imm = *data;
1393 AppendToBuffer(",0x%x", imm);
1394 data++;
1395 }
1396 break;
1397
1398 case 0x88: // 8bit, fall through
1399 case 0x89: // 32bit
1400 {
1401 bool is_byte = *data == 0x88;
1402 int mod, regop, rm;
1403 data++;
1404 get_modrm(*data, &mod, &regop, &rm);
1405 AppendToBuffer("mov%c ", is_byte ? 'b' : operand_size_code());
1406 data += PrintRightOperand(data);
1407 AppendToBuffer(",%s", NameOfCPURegister(regop));
1408 }
1409 break;
1410
1411 case 0x90:
1412 case 0x91:
1413 case 0x92:
1414 case 0x93:
1415 case 0x94:
1416 case 0x95:
1417 case 0x96:
1418 case 0x97: {
Steve Blockd0582a62009-12-15 09:54:21 +00001419 int reg = (*data & 0x7) | (rex_b() ? 8 : 0);
Steve Blocka7e24c12009-10-30 11:49:00 +00001420 if (reg == 0) {
1421 AppendToBuffer("nop"); // Common name for xchg rax,rax.
1422 } else {
1423 AppendToBuffer("xchg%c rax, %s",
1424 operand_size_code(),
1425 NameOfCPURegister(reg));
1426 }
Steve Blockd0582a62009-12-15 09:54:21 +00001427 data++;
Steve Blocka7e24c12009-10-30 11:49:00 +00001428 }
Steve Blockd0582a62009-12-15 09:54:21 +00001429 break;
Steve Blocka7e24c12009-10-30 11:49:00 +00001430
1431 case 0xFE: {
1432 data++;
1433 int mod, regop, rm;
1434 get_modrm(*data, &mod, &regop, &rm);
1435 if (mod == 3 && regop == 1) {
1436 AppendToBuffer("decb %s", NameOfCPURegister(rm));
1437 } else {
1438 UnimplementedInstruction();
1439 }
1440 data++;
1441 }
1442 break;
1443
1444 case 0x68:
1445 AppendToBuffer("push 0x%x", *reinterpret_cast<int32_t*>(data + 1));
1446 data += 5;
1447 break;
1448
1449 case 0x6A:
1450 AppendToBuffer("push 0x%x", *reinterpret_cast<int8_t*>(data + 1));
1451 data += 2;
1452 break;
1453
1454 case 0xA1: // Fall through.
1455 case 0xA3:
1456 switch (operand_size()) {
1457 case DOUBLEWORD_SIZE: {
1458 const char* memory_location = NameOfAddress(
1459 reinterpret_cast<byte*>(
1460 *reinterpret_cast<int32_t*>(data + 1)));
1461 if (*data == 0xA1) { // Opcode 0xA1
1462 AppendToBuffer("movzxlq rax,(%s)", memory_location);
1463 } else { // Opcode 0xA3
1464 AppendToBuffer("movzxlq (%s),rax", memory_location);
1465 }
1466 data += 5;
1467 break;
1468 }
1469 case QUADWORD_SIZE: {
1470 // New x64 instruction mov rax,(imm_64).
1471 const char* memory_location = NameOfAddress(
1472 *reinterpret_cast<byte**>(data + 1));
1473 if (*data == 0xA1) { // Opcode 0xA1
1474 AppendToBuffer("movq rax,(%s)", memory_location);
1475 } else { // Opcode 0xA3
1476 AppendToBuffer("movq (%s),rax", memory_location);
1477 }
1478 data += 9;
1479 break;
1480 }
1481 default:
1482 UnimplementedInstruction();
1483 data += 2;
1484 }
1485 break;
1486
1487 case 0xA8:
1488 AppendToBuffer("test al,0x%x", *reinterpret_cast<uint8_t*>(data + 1));
1489 data += 2;
1490 break;
1491
1492 case 0xA9: {
1493 int64_t value = 0;
1494 switch (operand_size()) {
1495 case WORD_SIZE:
1496 value = *reinterpret_cast<uint16_t*>(data + 1);
1497 data += 3;
1498 break;
1499 case DOUBLEWORD_SIZE:
1500 value = *reinterpret_cast<uint32_t*>(data + 1);
1501 data += 5;
1502 break;
1503 case QUADWORD_SIZE:
1504 value = *reinterpret_cast<int32_t*>(data + 1);
1505 data += 5;
1506 break;
1507 default:
1508 UNREACHABLE();
1509 }
1510 AppendToBuffer("test%c rax,0x%"V8_PTR_PREFIX"x",
1511 operand_size_code(),
1512 value);
1513 break;
1514 }
1515 case 0xD1: // fall through
1516 case 0xD3: // fall through
1517 case 0xC1:
1518 data += ShiftInstruction(data);
1519 break;
1520 case 0xD0: // fall through
1521 case 0xD2: // fall through
1522 case 0xC0:
1523 byte_size_operand_ = true;
1524 data += ShiftInstruction(data);
1525 break;
1526
1527 case 0xD9: // fall through
1528 case 0xDA: // fall through
1529 case 0xDB: // fall through
1530 case 0xDC: // fall through
1531 case 0xDD: // fall through
1532 case 0xDE: // fall through
1533 case 0xDF:
1534 data += FPUInstruction(data);
1535 break;
1536
1537 case 0xEB:
1538 data += JumpShort(data);
1539 break;
1540
Steve Blockd0582a62009-12-15 09:54:21 +00001541 case 0xF6:
1542 byte_size_operand_ = true; // fall through
Steve Blocka7e24c12009-10-30 11:49:00 +00001543 case 0xF7:
Steve Blockd0582a62009-12-15 09:54:21 +00001544 data += F6F7Instruction(data);
Steve Blocka7e24c12009-10-30 11:49:00 +00001545 break;
1546
1547 default:
1548 UnimplementedInstruction();
1549 data += 1;
1550 }
1551 } // !processed
1552
1553 if (tmp_buffer_pos_ < sizeof tmp_buffer_) {
1554 tmp_buffer_[tmp_buffer_pos_] = '\0';
1555 }
1556
Steve Blockd0582a62009-12-15 09:54:21 +00001557 int instr_len = static_cast<int>(data - instr);
Steve Blocka7e24c12009-10-30 11:49:00 +00001558 ASSERT(instr_len > 0); // Ensure progress.
1559
1560 int outp = 0;
1561 // Instruction bytes.
1562 for (byte* bp = instr; bp < data; bp++) {
1563 outp += v8::internal::OS::SNPrintF(out_buffer + outp, "%02x", *bp);
1564 }
1565 for (int i = 6 - instr_len; i >= 0; i--) {
1566 outp += v8::internal::OS::SNPrintF(out_buffer + outp, " ");
1567 }
1568
1569 outp += v8::internal::OS::SNPrintF(out_buffer + outp, " %s",
1570 tmp_buffer_.start());
1571 return instr_len;
1572}
1573
1574//------------------------------------------------------------------------------
1575
1576
1577static const char* cpu_regs[16] = {
1578 "rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi",
1579 "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15"
1580};
1581
1582
1583static const char* byte_cpu_regs[16] = {
1584 "al", "cl", "dl", "bl", "spl", "bpl", "sil", "dil",
1585 "r8l", "r9l", "r10l", "r11l", "r12l", "r13l", "r14l", "r15l"
1586};
1587
1588
1589static const char* xmm_regs[16] = {
1590 "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7",
1591 "xmm8", "xmm9", "xmm10", "xmm11", "xmm12", "xmm13", "xmm14", "xmm15"
1592};
1593
1594
1595const char* NameConverter::NameOfAddress(byte* addr) const {
1596 static v8::internal::EmbeddedVector<char, 32> tmp_buffer;
1597 v8::internal::OS::SNPrintF(tmp_buffer, "%p", addr);
1598 return tmp_buffer.start();
1599}
1600
1601
1602const char* NameConverter::NameOfConstant(byte* addr) const {
1603 return NameOfAddress(addr);
1604}
1605
1606
1607const char* NameConverter::NameOfCPURegister(int reg) const {
1608 if (0 <= reg && reg < 16)
1609 return cpu_regs[reg];
1610 return "noreg";
1611}
1612
1613
1614const char* NameConverter::NameOfByteCPURegister(int reg) const {
1615 if (0 <= reg && reg < 16)
1616 return byte_cpu_regs[reg];
1617 return "noreg";
1618}
1619
1620
1621const char* NameConverter::NameOfXMMRegister(int reg) const {
1622 if (0 <= reg && reg < 16)
1623 return xmm_regs[reg];
1624 return "noxmmreg";
1625}
1626
1627
1628const char* NameConverter::NameInCode(byte* addr) const {
1629 // X64 does not embed debug strings at the moment.
1630 UNREACHABLE();
1631 return "";
1632}
1633
1634//------------------------------------------------------------------------------
1635
1636Disassembler::Disassembler(const NameConverter& converter)
1637 : converter_(converter) { }
1638
1639Disassembler::~Disassembler() { }
1640
1641
1642int Disassembler::InstructionDecode(v8::internal::Vector<char> buffer,
1643 byte* instruction) {
1644 DisassemblerX64 d(converter_, CONTINUE_ON_UNIMPLEMENTED_OPCODE);
1645 return d.InstructionDecode(buffer, instruction);
1646}
1647
1648
1649// The X64 assembler does not use constant pools.
1650int Disassembler::ConstantPoolSizeAt(byte* instruction) {
1651 return -1;
1652}
1653
1654
1655void Disassembler::Disassemble(FILE* f, byte* begin, byte* end) {
1656 NameConverter converter;
1657 Disassembler d(converter);
1658 for (byte* pc = begin; pc < end;) {
1659 v8::internal::EmbeddedVector<char, 128> buffer;
1660 buffer[0] = '\0';
1661 byte* prev_pc = pc;
1662 pc += d.InstructionDecode(buffer, pc);
1663 fprintf(f, "%p", prev_pc);
1664 fprintf(f, " ");
1665
1666 for (byte* bp = prev_pc; bp < pc; bp++) {
1667 fprintf(f, "%02x", *bp);
1668 }
Steve Blockd0582a62009-12-15 09:54:21 +00001669 for (int i = 6 - static_cast<int>(pc - prev_pc); i >= 0; i--) {
Steve Blocka7e24c12009-10-30 11:49:00 +00001670 fprintf(f, " ");
1671 }
1672 fprintf(f, " %s\n", buffer.start());
1673 }
1674}
1675
1676} // namespace disasm
Leon Clarkef7060e22010-06-03 12:02:55 +01001677
1678#endif // V8_TARGET_ARCH_X64