| 1 |
/* |
| 2 |
* HT Editor |
| 3 |
* x86dis.cc |
| 4 |
* |
| 5 |
* Copyright (C) 1999-2002 Stefan Weyergraf |
| 6 |
* |
| 7 |
* This program is free software; you can redistribute it and/or modify |
| 8 |
* it under the terms of the GNU General Public License version 2 as |
| 9 |
* published by the Free Software Foundation. |
| 10 |
* |
| 11 |
* This program is distributed in the hope that it will be useful, |
| 12 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 |
* GNU General Public License for more details. |
| 15 |
* |
| 16 |
* You should have received a copy of the GNU General Public License |
| 17 |
* along with this program; if not, write to the Free Software |
| 18 |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
| 19 |
*/ |
| 20 |
|
| 21 |
#include <cstdio> |
| 22 |
#include <cstring> |
| 23 |
#include <cstdlib> |
| 24 |
|
| 25 |
#include "x86dis.h" |
| 26 |
|
| 27 |
#define mkmod(modrm) (((modrm)>>6)&3) |
| 28 |
#define mkreg(modrm) (((modrm)>>3)&7) |
| 29 |
#define mkrm(modrm) ((modrm)&7) |
| 30 |
|
| 31 |
#define mkscale(modrm) (((modrm)>>6)&3) |
| 32 |
#define mkindex(modrm) (((modrm)>>3)&7) |
| 33 |
#define mkbase(modrm) ((modrm)&7) |
| 34 |
|
| 35 |
int modrm16_1[8] = { X86_REG_BX, X86_REG_BX, X86_REG_BP, X86_REG_BP, |
| 36 |
X86_REG_SI, X86_REG_DI, X86_REG_BP, X86_REG_BX}; |
| 37 |
int modrm16_2[8] = { X86_REG_SI, X86_REG_DI, X86_REG_SI, X86_REG_DI, |
| 38 |
X86_REG_NO, X86_REG_NO, X86_REG_NO, X86_REG_NO}; |
| 39 |
|
| 40 |
int sibbase[8] = { X86_REG_AX, X86_REG_CX, X86_REG_DX, X86_REG_BX, |
| 41 |
X86_REG_SP, X86_REG_BP, X86_REG_SI, X86_REG_DI }; |
| 42 |
|
| 43 |
int sibindex[8] = { X86_REG_AX, X86_REG_CX, X86_REG_DX, X86_REG_BX, |
| 44 |
X86_REG_NO, X86_REG_BP, X86_REG_SI, X86_REG_DI }; |
| 45 |
|
| 46 |
int sibscale[4] = {1, 2, 4, 8}; |
| 47 |
|
| 48 |
/* |
| 49 |
* CLASS X86Disassembler |
| 50 |
*/ |
| 51 |
|
| 52 |
X86Disassembler::X86Disassembler() |
| 53 |
{ |
| 54 |
} |
| 55 |
|
| 56 |
X86Disassembler::X86Disassembler(int _opsize, int _addrsize) |
| 57 |
{ |
| 58 |
opsize=_opsize; |
| 59 |
addrsize=_addrsize; |
| 60 |
} |
| 61 |
|
| 62 |
X86Disassembler::~X86Disassembler() |
| 63 |
{ |
| 64 |
} |
| 65 |
|
| 66 |
dis_insn *X86Disassembler::decode(const byte *code, int Maxlen, CPU_ADDR Addr) |
| 67 |
{ |
| 68 |
ocodep=code; |
| 69 |
/* initialize */ |
| 70 |
codep=ocodep; |
| 71 |
maxlen=Maxlen; |
| 72 |
seg=Addr.addr32.seg; |
| 73 |
addr=Addr.addr32.offset; |
| 74 |
modrm=-1; |
| 75 |
sib=-1; |
| 76 |
memset(&insn, 0, sizeof(insn)); |
| 77 |
insn.invalid=false; |
| 78 |
insn.eopsize=opsize; |
| 79 |
insn.eaddrsize=opsize; |
| 80 |
|
| 81 |
prefixes(); |
| 82 |
|
| 83 |
insn.opcode=c; |
| 84 |
decode_insn(&x86_insns[insn.opcode]); |
| 85 |
|
| 86 |
if (insn.invalid) { |
| 87 |
insn.name="db"; |
| 88 |
insn.size=1; |
| 89 |
insn.op[0].type=X86_OPTYPE_IMM; |
| 90 |
insn.op[0].size=1; |
| 91 |
insn.op[0].imm=*code; |
| 92 |
insn.lockprefix=X86_PREFIX_NO; |
| 93 |
insn.repprefix=X86_PREFIX_NO; |
| 94 |
insn.segprefix=X86_PREFIX_NO; |
| 95 |
for (int i=1; i<3; i++) insn.op[i].type=X86_OPTYPE_EMPTY; |
| 96 |
} else { |
| 97 |
insn.size=codep-ocodep; |
| 98 |
} |
| 99 |
return &insn; |
| 100 |
} |
| 101 |
|
| 102 |
void X86Disassembler::decode_modrm(x86_insn_op *op, char size, int allow_reg, int allow_mem, int mmx) |
| 103 |
{ |
| 104 |
int modrm=getmodrm(); |
| 105 |
int mod=mkmod(modrm); |
| 106 |
int rm=mkrm(modrm); |
| 107 |
if (mod==3) { |
| 108 |
/* reg */ |
| 109 |
if (!allow_reg) { |
| 110 |
invalidate(); |
| 111 |
return; |
| 112 |
} |
| 113 |
if (mmx) { |
| 114 |
op->type=X86_OPTYPE_MMX; |
| 115 |
op->size=8; |
| 116 |
op->mmx=rm; |
| 117 |
} else { |
| 118 |
op->type=X86_OPTYPE_REG; |
| 119 |
op->size=esizeop(size); |
| 120 |
op->reg=rm; |
| 121 |
} |
| 122 |
} else { |
| 123 |
/* mem */ |
| 124 |
if (!allow_mem) { |
| 125 |
invalidate(); |
| 126 |
return; |
| 127 |
} |
| 128 |
op->mem.addrsize=insn.eaddrsize; |
| 129 |
if (insn.eaddrsize==X86_ADDRSIZE16) { |
| 130 |
op->type=X86_OPTYPE_MEM; |
| 131 |
op->size=esizeop(size); |
| 132 |
op->mem.floatptr=isfloat(size); |
| 133 |
if ((mod==0) && (rm==6)) { |
| 134 |
op->mem.hasdisp=1; |
| 135 |
op->mem.disp=getword(); |
| 136 |
op->mem.base=X86_REG_NO; |
| 137 |
op->mem.index=X86_REG_NO; |
| 138 |
op->mem.scale=0; |
| 139 |
} else { |
| 140 |
op->mem.base=modrm16_1[rm]; |
| 141 |
op->mem.index=modrm16_2[rm]; |
| 142 |
op->mem.scale=1; |
| 143 |
switch (mod) { |
| 144 |
case 0: |
| 145 |
op->mem.hasdisp=0; |
| 146 |
op->mem.disp=0; |
| 147 |
break; |
| 148 |
case 1: |
| 149 |
op->mem.hasdisp=1; |
| 150 |
op->mem.disp=(char)getbyte(); |
| 151 |
break; |
| 152 |
case 2: |
| 153 |
op->mem.hasdisp=1; |
| 154 |
op->mem.disp=(short)getword(); |
| 155 |
break; |
| 156 |
} |
| 157 |
} |
| 158 |
} else { |
| 159 |
op->type=X86_OPTYPE_MEM; |
| 160 |
op->size=esizeop(size); |
| 161 |
op->mem.floatptr=isfloat(size); |
| 162 |
if ((mod==0) && (rm==5)) { |
| 163 |
op->mem.hasdisp=1; |
| 164 |
op->mem.disp=getdword(); |
| 165 |
op->mem.base=X86_REG_NO; |
| 166 |
op->mem.index=X86_REG_NO; |
| 167 |
op->mem.scale=0; |
| 168 |
} else if (rm==4) { |
| 169 |
decode_sib(op, mod); |
| 170 |
} else { |
| 171 |
op->mem.base=rm; |
| 172 |
op->mem.index=X86_REG_NO; |
| 173 |
op->mem.scale=1; |
| 174 |
switch (mod) { |
| 175 |
case 0: |
| 176 |
op->mem.hasdisp=0; |
| 177 |
op->mem.disp=0; |
| 178 |
break; |
| 179 |
case 1: |
| 180 |
op->mem.hasdisp=1; |
| 181 |
op->mem.disp=(char)getbyte(); |
| 182 |
break; |
| 183 |
case 2: |
| 184 |
op->mem.hasdisp=1; |
| 185 |
op->mem.disp=getdword(); |
| 186 |
break; |
| 187 |
} |
| 188 |
} |
| 189 |
} |
| 190 |
} |
| 191 |
} |
| 192 |
|
| 193 |
void X86Disassembler::decode_insn(x86opc_insn *xinsn) |
| 194 |
{ |
| 195 |
if (!xinsn->name) { |
| 196 |
x86opc_insn_op_special special=*((x86opc_insn_op_special*)(&xinsn->op[0])); |
| 197 |
switch (special.type) { |
| 198 |
case SPECIAL_TYPE_INVALID: |
| 199 |
invalidate(); |
| 200 |
break; |
| 201 |
case SPECIAL_TYPE_PREFIX: |
| 202 |
switch (c) { |
| 203 |
case 0x0f: |
| 204 |
if (insn.opcodeclass == X86DIS_OPCODE_CLASS_STD) { |
| 205 |
insn.opcode = getbyte(); |
| 206 |
insn.opcodeclass = X86DIS_OPCODE_CLASS_EXT; |
| 207 |
decode_insn(&x86_insns_ext[insn.opcode]); |
| 208 |
break; |
| 209 |
} |
| 210 |
default: |
| 211 |
invalidate(); |
| 212 |
break; |
| 213 |
} |
| 214 |
break; |
| 215 |
case SPECIAL_TYPE_GROUP: { |
| 216 |
int m = mkreg(getmodrm()); |
| 217 |
insn.opcode |= m<<8; |
| 218 |
decode_insn(&x86_group_insns[(int)special.data][m]); |
| 219 |
break; |
| 220 |
} |
| 221 |
case SPECIAL_TYPE_FGROUP: { |
| 222 |
int m = getmodrm(); |
| 223 |
if (mkmod(m) == 3) { |
| 224 |
x86opc_finsn f = x86_float_group_insns[(int)special.data][mkreg(m)]; |
| 225 |
/* fprintf(stderr, "special.data=%d, m=%d, mkreg(m)=%d, mkrm(m)=%d\n", special.data, m, mkreg(m), mkrm(m));*/ |
| 226 |
if (f.group) { |
| 227 |
decode_insn(&f.group[mkrm(m)]); |
| 228 |
} else if (f.insn.name) { |
| 229 |
decode_insn(&f.insn); |
| 230 |
} else invalidate(); |
| 231 |
} else { |
| 232 |
decode_insn(&x86_modfloat_group_insns[(int)special.data][mkreg(m)]); |
| 233 |
} |
| 234 |
break; |
| 235 |
} |
| 236 |
} |
| 237 |
} else { |
| 238 |
insn.name=xinsn->name; |
| 239 |
for (int i=0; i<3; i++) { |
| 240 |
decode_op(&insn.op[i], &xinsn->op[i]); |
| 241 |
} |
| 242 |
} |
| 243 |
} |
| 244 |
|
| 245 |
void X86Disassembler::decode_op(x86_insn_op *op, x86opc_insn_op *xop) |
| 246 |
{ |
| 247 |
switch (xop->type) { |
| 248 |
case TYPE_0: |
| 249 |
return; |
| 250 |
case TYPE_A: { |
| 251 |
/* direct address without ModR/M */ |
| 252 |
op->type=X86_OPTYPE_FARPTR; |
| 253 |
op->size=esizeop(xop->size); |
| 254 |
switch (op->size) { |
| 255 |
case 4: |
| 256 |
op->farptr.offset=getword(); |
| 257 |
op->farptr.seg=getword(); |
| 258 |
break; |
| 259 |
case 6: |
| 260 |
op->farptr.offset=getdword(); |
| 261 |
op->farptr.seg=getword(); |
| 262 |
break; |
| 263 |
} |
| 264 |
break; |
| 265 |
} |
| 266 |
case TYPE_C: { |
| 267 |
/* reg of ModR/M picks control register */ |
| 268 |
op->type=X86_OPTYPE_CRX; |
| 269 |
op->size=esizeop(xop->size); |
| 270 |
op->crx=mkreg(getmodrm()); |
| 271 |
break; |
| 272 |
} |
| 273 |
case TYPE_D: { |
| 274 |
/* reg of ModR/M picks debug register */ |
| 275 |
op->type=X86_OPTYPE_DRX; |
| 276 |
op->size=esizeop(xop->size); |
| 277 |
op->drx=mkreg(getmodrm()); |
| 278 |
break; |
| 279 |
} |
| 280 |
case TYPE_E: { |
| 281 |
/* ModR/M (general reg or memory) */ |
| 282 |
decode_modrm(op, xop->size, (xop->size!=SIZE_P), 1, 0); |
| 283 |
break; |
| 284 |
} |
| 285 |
case TYPE_F: { |
| 286 |
/* r/m of ModR/M picks a fpu register */ |
| 287 |
op->type=X86_OPTYPE_STX; |
| 288 |
op->size=10; |
| 289 |
op->stx=mkrm(getmodrm()); |
| 290 |
break; |
| 291 |
} |
| 292 |
case TYPE_Fx: { |
| 293 |
/* extra picks a fpu register */ |
| 294 |
op->type=X86_OPTYPE_STX; |
| 295 |
op->size=10; |
| 296 |
op->stx=xop->extra; |
| 297 |
break; |
| 298 |
} |
| 299 |
case TYPE_G: { |
| 300 |
/* reg of ModR/M picks general register */ |
| 301 |
op->type=X86_OPTYPE_REG; |
| 302 |
op->size=esizeop(xop->size); |
| 303 |
op->reg=mkreg(getmodrm()); |
| 304 |
break; |
| 305 |
} |
| 306 |
case TYPE_Is: { |
| 307 |
/* signed immediate */ |
| 308 |
op->type=X86_OPTYPE_IMM; |
| 309 |
op->size=esizeop(xop->extendedsize); |
| 310 |
int s=esizeop(xop->size); |
| 311 |
switch (s) { |
| 312 |
case 1: |
| 313 |
op->imm=(char)getbyte(); |
| 314 |
break; |
| 315 |
case 2: |
| 316 |
op->imm=(short)getword(); |
| 317 |
break; |
| 318 |
case 4: |
| 319 |
op->imm=getdword(); |
| 320 |
break; |
| 321 |
} |
| 322 |
switch (op->size) { |
| 323 |
case 1: |
| 324 |
op->imm&=0xff; |
| 325 |
break; |
| 326 |
case 2: |
| 327 |
op->imm&=0xffff; |
| 328 |
break; |
| 329 |
} |
| 330 |
break; |
| 331 |
} |
| 332 |
case TYPE_I: { |
| 333 |
/* unsigned immediate */ |
| 334 |
op->type=X86_OPTYPE_IMM; |
| 335 |
op->size=esizeop(xop->extendedsize); |
| 336 |
int s=esizeop(xop->size); |
| 337 |
switch (s) { |
| 338 |
case 1: |
| 339 |
op->imm=getbyte(); |
| 340 |
break; |
| 341 |
case 2: |
| 342 |
op->imm=getword(); |
| 343 |
break; |
| 344 |
case 4: |
| 345 |
op->imm=getdword(); |
| 346 |
break; |
| 347 |
} |
| 348 |
break; |
| 349 |
} |
| 350 |
case TYPE_Ix: { |
| 351 |
/* fixed immediate */ |
| 352 |
op->type=X86_OPTYPE_IMM; |
| 353 |
op->size=esizeop(xop->extendedsize); |
| 354 |
op->imm=xop->extra; |
| 355 |
break; |
| 356 |
} |
| 357 |
case TYPE_J: { |
| 358 |
/* relative branch offset */ |
| 359 |
op->type=X86_OPTYPE_IMM; |
| 360 |
// op->size=esizeaddr(xop->extendedsize); |
| 361 |
op->size=(addrsize==X86_ADDRSIZE32) ? 4 : 2; |
| 362 |
int s=esizeop(xop->size); |
| 363 |
switch (s) { |
| 364 |
case 1: |
| 365 |
op->imm=(char)getbyte()+addr; |
| 366 |
break; |
| 367 |
case 2: |
| 368 |
op->imm=(short)getword()+addr; |
| 369 |
break; |
| 370 |
case 4: |
| 371 |
op->imm=getdword()+addr; |
| 372 |
break; |
| 373 |
} |
| 374 |
break; |
| 375 |
} |
| 376 |
case TYPE_M: { |
| 377 |
/* ModR/M (memory only) */ |
| 378 |
decode_modrm(op, xop->size, 0, 1, 0); |
| 379 |
break; |
| 380 |
} |
| 381 |
case TYPE_O: { |
| 382 |
/* direct memory without ModR/M */ |
| 383 |
op->type=X86_OPTYPE_MEM; |
| 384 |
op->size=esizeop(xop->size); |
| 385 |
op->mem.floatptr=isfloat(xop->size); |
| 386 |
op->mem.addrsize=insn.eaddrsize; |
| 387 |
switch (insn.eaddrsize) { |
| 388 |
case X86_ADDRSIZE16: |
| 389 |
op->mem.hasdisp=1; |
| 390 |
op->mem.disp=getword(); |
| 391 |
break; |
| 392 |
case X86_ADDRSIZE32: |
| 393 |
op->mem.hasdisp=1; |
| 394 |
op->mem.disp=getdword(); |
| 395 |
break; |
| 396 |
} |
| 397 |
op->mem.base=X86_REG_NO; |
| 398 |
op->mem.index=X86_REG_NO; |
| 399 |
op->mem.scale=1; |
| 400 |
break; |
| 401 |
} |
| 402 |
case TYPE_P: { |
| 403 |
/* reg of ModR/M picks MMX register */ |
| 404 |
op->type=X86_OPTYPE_MMX; |
| 405 |
op->size=8; |
| 406 |
op->mmx=mkreg(getmodrm()); |
| 407 |
break; |
| 408 |
} |
| 409 |
case TYPE_Q: { |
| 410 |
/* ModR/M (MMX reg or memory) */ |
| 411 |
decode_modrm(op, xop->size, 1, 1, 1); |
| 412 |
break; |
| 413 |
} |
| 414 |
case TYPE_R: { |
| 415 |
/* rm of ModR/M picks general register */ |
| 416 |
op->type=X86_OPTYPE_REG; |
| 417 |
op->size=esizeop(xop->size); |
| 418 |
op->reg=mkrm(getmodrm()); |
| 419 |
break; |
| 420 |
} |
| 421 |
case TYPE_Rx: { |
| 422 |
/* extra picks register */ |
| 423 |
op->type=X86_OPTYPE_REG; |
| 424 |
op->size=esizeop(xop->size); |
| 425 |
op->reg=xop->extra; |
| 426 |
break; |
| 427 |
} |
| 428 |
case TYPE_S: { |
| 429 |
/* reg of ModR/M picks segment register */ |
| 430 |
op->type=X86_OPTYPE_SEG; |
| 431 |
op->size=esizeop(xop->size); |
| 432 |
op->seg=mkreg(getmodrm()); |
| 433 |
if (op->seg>5) invalidate(); |
| 434 |
break; |
| 435 |
} |
| 436 |
case TYPE_Sx: { |
| 437 |
/* extra picks segment register */ |
| 438 |
op->type=X86_OPTYPE_SEG; |
| 439 |
op->size=esizeop(xop->size); |
| 440 |
op->seg=xop->extra; |
| 441 |
if (op->seg>5) invalidate(); |
| 442 |
break; |
| 443 |
} |
| 444 |
case TYPE_T: { |
| 445 |
/* reg of ModR/M picks test register */ |
| 446 |
op->type=X86_OPTYPE_TRX; |
| 447 |
op->size=esizeop(xop->size); |
| 448 |
op->trx=mkreg(getmodrm()); |
| 449 |
break; |
| 450 |
} |
| 451 |
} |
| 452 |
} |
| 453 |
|
| 454 |
void X86Disassembler::decode_sib(x86_insn_op *op, int mod) |
| 455 |
{ |
| 456 |
int sib=getsib(); |
| 457 |
int scale=mkscale(sib); |
| 458 |
int index=mkindex(sib); |
| 459 |
int base=mkbase(sib); |
| 460 |
int disp=mod; |
| 461 |
if (base==5) { |
| 462 |
if (mod==0) { |
| 463 |
op->mem.base=X86_REG_NO; |
| 464 |
disp=2; |
| 465 |
} else { |
| 466 |
op->mem.base=X86_REG_BP; |
| 467 |
} |
| 468 |
op->mem.index=sibindex[index]; |
| 469 |
op->mem.scale=sibscale[scale]; |
| 470 |
} else { |
| 471 |
op->mem.base=sibbase[base]; |
| 472 |
op->mem.index=sibindex[index]; |
| 473 |
op->mem.scale=sibscale[scale]; |
| 474 |
} |
| 475 |
switch (disp) { |
| 476 |
case 0: |
| 477 |
op->mem.hasdisp=0; |
| 478 |
op->mem.disp=0; |
| 479 |
break; |
| 480 |
case 1: |
| 481 |
op->mem.hasdisp=1; |
| 482 |
op->mem.disp=(char)getbyte(); |
| 483 |
break; |
| 484 |
case 2: |
| 485 |
op->mem.hasdisp=1; |
| 486 |
op->mem.disp=getdword(); |
| 487 |
break; |
| 488 |
} |
| 489 |
} |
| 490 |
|
| 491 |
dis_insn *X86Disassembler::duplicateInsn(dis_insn *disasm_insn) |
| 492 |
{ |
| 493 |
x86dis_insn *insn = (x86dis_insn *)malloc(sizeof (x86dis_insn)); |
| 494 |
*insn = *(x86dis_insn *)disasm_insn; |
| 495 |
return insn; |
| 496 |
} |
| 497 |
|
| 498 |
int X86Disassembler::esizeaddr(char c) |
| 499 |
{ |
| 500 |
switch (c) { |
| 501 |
case SIZE_B: |
| 502 |
return 1; |
| 503 |
case SIZE_W: |
| 504 |
return 2; |
| 505 |
case SIZE_D: |
| 506 |
return 4; |
| 507 |
case SIZE_Q: |
| 508 |
return 8; |
| 509 |
case SIZE_S: |
| 510 |
return 4; |
| 511 |
case SIZE_L: |
| 512 |
return 8; |
| 513 |
case SIZE_T: |
| 514 |
return 10; |
| 515 |
case SIZE_C: |
| 516 |
if (insn.eaddrsize==X86_ADDRSIZE16) return 1; else return 2; |
| 517 |
case SIZE_V: |
| 518 |
if (insn.eaddrsize==X86_ADDRSIZE16) return 2; else return 4; |
| 519 |
case SIZE_P: |
| 520 |
if (insn.eaddrsize==X86_ADDRSIZE16) return 4; else return 6; |
| 521 |
} |
| 522 |
return 0; |
| 523 |
} |
| 524 |
|
| 525 |
int X86Disassembler::esizeop(char c) |
| 526 |
{ |
| 527 |
switch (c) { |
| 528 |
case SIZE_B: |
| 529 |
return 1; |
| 530 |
case SIZE_W: |
| 531 |
return 2; |
| 532 |
case SIZE_D: |
| 533 |
return 4; |
| 534 |
case SIZE_Q: |
| 535 |
return 8; |
| 536 |
case SIZE_S: |
| 537 |
return 4; |
| 538 |
case SIZE_L: |
| 539 |
return 8; |
| 540 |
case SIZE_T: |
| 541 |
return 10; |
| 542 |
case SIZE_C: |
| 543 |
if (insn.eopsize==X86_OPSIZE16) return 1; else return 2; |
| 544 |
case SIZE_V: |
| 545 |
if (insn.eopsize==X86_OPSIZE16) return 2; else return 4; |
| 546 |
case SIZE_P: |
| 547 |
if (insn.eopsize==X86_OPSIZE16) return 4; else return 6; |
| 548 |
} |
| 549 |
return 0; |
| 550 |
} |
| 551 |
|
| 552 |
byte X86Disassembler::getbyte() |
| 553 |
{ |
| 554 |
if (codep-ocodep+1<=maxlen) { |
| 555 |
addr+=1; |
| 556 |
return *(codep++); |
| 557 |
} else { |
| 558 |
invalidate(); |
| 559 |
return 0; |
| 560 |
} |
| 561 |
} |
| 562 |
|
| 563 |
word X86Disassembler::getword() |
| 564 |
{ |
| 565 |
if (codep-ocodep+2<=maxlen) { |
| 566 |
word w; |
| 567 |
addr += 2; |
| 568 |
w = codep[0] | (codep[1]<<8); |
| 569 |
codep += 2; |
| 570 |
return w; |
| 571 |
} else { |
| 572 |
invalidate(); |
| 573 |
return 0; |
| 574 |
} |
| 575 |
} |
| 576 |
|
| 577 |
dword X86Disassembler::getdword() |
| 578 |
{ |
| 579 |
if (codep-ocodep+4<=maxlen) { |
| 580 |
dword w; |
| 581 |
addr += 4; |
| 582 |
w = codep[0] | (codep[1]<<8) | (codep[2]<<16) | (codep[3]<<24); |
| 583 |
codep += 4; |
| 584 |
return w; |
| 585 |
} else { |
| 586 |
invalidate(); |
| 587 |
return 0; |
| 588 |
} |
| 589 |
} |
| 590 |
|
| 591 |
void X86Disassembler::getOpcodeMetrics(int &min_length, int &max_length, int &min_look_ahead, int &avg_look_ahead, int &addr_align) |
| 592 |
{ |
| 593 |
min_length = 1; |
| 594 |
max_length = 15; |
| 595 |
min_look_ahead = 120; // 1/2/3/4/5/6/8/10/12/15 |
| 596 |
avg_look_ahead = 24; // 1/2/3/4/6/8/12/24 |
| 597 |
addr_align = 1; |
| 598 |
} |
| 599 |
|
| 600 |
int X86Disassembler::getmodrm() |
| 601 |
{ |
| 602 |
if (modrm==-1) modrm=getbyte(); |
| 603 |
return modrm; |
| 604 |
} |
| 605 |
|
| 606 |
char *X86Disassembler::getName() |
| 607 |
{ |
| 608 |
return "x86/Disassembler"; |
| 609 |
} |
| 610 |
|
| 611 |
int X86Disassembler::getsib() |
| 612 |
{ |
| 613 |
if (sib==-1) sib=getbyte(); |
| 614 |
return sib; |
| 615 |
} |
| 616 |
|
| 617 |
byte X86Disassembler::getSize(dis_insn *disasm_insn) |
| 618 |
{ |
| 619 |
return ((x86dis_insn*)disasm_insn)->size; |
| 620 |
} |
| 621 |
|
| 622 |
void X86Disassembler::invalidate() |
| 623 |
{ |
| 624 |
insn.invalid=1; |
| 625 |
} |
| 626 |
|
| 627 |
int X86Disassembler::isfloat(char c) |
| 628 |
{ |
| 629 |
switch (c) { |
| 630 |
case SIZE_S: |
| 631 |
case SIZE_L: |
| 632 |
case SIZE_T: |
| 633 |
return 1; |
| 634 |
} |
| 635 |
return 0; |
| 636 |
} |
| 637 |
|
| 638 |
void X86Disassembler::prefixes() |
| 639 |
{ |
| 640 |
bool notprefix=0; |
| 641 |
insn.lockprefix=X86_PREFIX_NO; |
| 642 |
insn.repprefix=X86_PREFIX_NO; |
| 643 |
insn.segprefix=X86_PREFIX_NO; |
| 644 |
while ((codep-ocodep<15) && (!notprefix)) { |
| 645 |
c=getbyte(); |
| 646 |
switch (c) { |
| 647 |
case 0x26: |
| 648 |
insn.segprefix=X86_PREFIX_ES; |
| 649 |
break; |
| 650 |
case 0x2e: |
| 651 |
insn.segprefix=X86_PREFIX_CS; |
| 652 |
break; |
| 653 |
case 0x36: |
| 654 |
insn.segprefix=X86_PREFIX_SS; |
| 655 |
break; |
| 656 |
case 0x3e: |
| 657 |
insn.segprefix=X86_PREFIX_DS; |
| 658 |
break; |
| 659 |
case 0x64: |
| 660 |
insn.segprefix=X86_PREFIX_FS; |
| 661 |
break; |
| 662 |
case 0x65: |
| 663 |
insn.segprefix=X86_PREFIX_GS; |
| 664 |
break; |
| 665 |
case 0x66: |
| 666 |
insn.eopsize=(opsize==X86_OPSIZE16) ? X86_OPSIZE32 : X86_OPSIZE16; |
| 667 |
break; |
| 668 |
case 0x67: |
| 669 |
insn.eaddrsize=(addrsize==X86_ADDRSIZE16) ? X86_ADDRSIZE32 : X86_ADDRSIZE16; |
| 670 |
break; |
| 671 |
case 0xf0: |
| 672 |
insn.lockprefix=X86_PREFIX_LOCK; |
| 673 |
break; |
| 674 |
case 0xf2: |
| 675 |
insn.repprefix=X86_PREFIX_REPNZ; |
| 676 |
break; |
| 677 |
case 0xf3: |
| 678 |
insn.repprefix=X86_PREFIX_REPZ; |
| 679 |
break; |
| 680 |
default: |
| 681 |
notprefix=1; |
| 682 |
} |
| 683 |
} |
| 684 |
} |
| 685 |
|
| 686 |
int X86Disassembler::special_param_ambiguity(x86dis_insn *disasm_insn) |
| 687 |
{ |
| 688 |
int regc=0, memc=0, segc=0; |
| 689 |
for (int i=0; i<3; i++) { |
| 690 |
switch (disasm_insn->op[i].type) { |
| 691 |
case X86_OPTYPE_SEG: |
| 692 |
segc++; |
| 693 |
case X86_OPTYPE_REG: |
| 694 |
case X86_OPTYPE_CRX: |
| 695 |
case X86_OPTYPE_DRX: |
| 696 |
case X86_OPTYPE_TRX: |
| 697 |
case X86_OPTYPE_STX: |
| 698 |
case X86_OPTYPE_MMX: |
| 699 |
regc++; |
| 700 |
break; |
| 701 |
case X86_OPTYPE_MEM: |
| 702 |
memc++; |
| 703 |
break; |
| 704 |
} |
| 705 |
} |
| 706 |
return (memc && !regc) |
| 707 |
|| (memc && segc) |
| 708 |
|| (strcmp(disasm_insn->name, "movzx") == 0) |
| 709 |
|| (strcmp(disasm_insn->name, "movsx") == 0); |
| 710 |
} |
| 711 |
|
| 712 |
void X86Disassembler::str_op(char *opstr, int *opstrlen, x86dis_insn *insn, x86_insn_op *op, bool explicit_params) |
| 713 |
{ |
| 714 |
const char *cs_default = get_cs(e_cs_default); |
| 715 |
const char *cs_number = get_cs(e_cs_number); |
| 716 |
const char *cs_symbol = get_cs(e_cs_symbol); |
| 717 |
|
| 718 |
*opstrlen=0; |
| 719 |
switch (op->type) { |
| 720 |
case X86_OPTYPE_IMM: { |
| 721 |
CPU_ADDR a; |
| 722 |
// FIXME: hack |
| 723 |
if ((insn->name[0]=='j') || strcmp(insn->name, "call")==0) { |
| 724 |
a.addr32.seg = seg; |
| 725 |
} else { |
| 726 |
a.addr32.seg = 0; |
| 727 |
} |
| 728 |
a.addr32.offset = op->imm; |
| 729 |
int slen; |
| 730 |
char *s=(addr_sym_func) ? addr_sym_func(a, &slen, addr_sym_func_context) : NULL; |
| 731 |
if (s) { |
| 732 |
memmove(opstr, s, slen); |
| 733 |
opstr[slen]=0; |
| 734 |
*opstrlen=slen; |
| 735 |
} else { |
| 736 |
char *g=opstr; |
| 737 |
strcpy(g, cs_number); g += strlen(cs_number); |
| 738 |
switch (op->size) { |
| 739 |
case 1: |
| 740 |
hexd(&g, 2, options, op->imm); |
| 741 |
break; |
| 742 |
case 2: |
| 743 |
hexd(&g, 4, options, op->imm); |
| 744 |
break; |
| 745 |
case 4: |
| 746 |
hexd(&g, 8, options, op->imm); |
| 747 |
break; |
| 748 |
} |
| 749 |
strcpy(g, cs_default); g += strlen(cs_default); |
| 750 |
} |
| 751 |
break; |
| 752 |
} |
| 753 |
case X86_OPTYPE_REG: { |
| 754 |
int j; |
| 755 |
switch (op->size) { |
| 756 |
case 1: j=0; break; |
| 757 |
case 2: j=1; break; |
| 758 |
case 4: j=2; break; |
| 759 |
default: |
| 760 |
j=-1; |
| 761 |
} |
| 762 |
if (j!=-1) sprintf(opstr, x86_regs[j][op->reg]); |
| 763 |
break; |
| 764 |
} |
| 765 |
case X86_OPTYPE_SEG: |
| 766 |
if (x86_segs[op->seg]) { |
| 767 |
sprintf(opstr, x86_segs[op->seg]); |
| 768 |
} |
| 769 |
break; |
| 770 |
case X86_OPTYPE_CRX: |
| 771 |
sprintf(opstr, "cr%d", op->crx); |
| 772 |
break; |
| 773 |
case X86_OPTYPE_DRX: |
| 774 |
sprintf(opstr, "dr%d", op->drx); |
| 775 |
break; |
| 776 |
case X86_OPTYPE_TRX: |
| 777 |
sprintf(opstr, "tr%d", op->trx); |
| 778 |
break; |
| 779 |
case X86_OPTYPE_STX: |
| 780 |
if (op->stx) { |
| 781 |
sprintf(opstr, "st(%d)", op->stx); |
| 782 |
} else { |
| 783 |
strcpy(opstr, "st"); |
| 784 |
} |
| 785 |
break; |
| 786 |
case X86_OPTYPE_MMX: |
| 787 |
sprintf(opstr, "mm%d", op->mmx); |
| 788 |
break; |
| 789 |
case X86_OPTYPE_MEM: { |
| 790 |
char *d=opstr; |
| 791 |
int reg; |
| 792 |
if (explicit_params) { |
| 793 |
if (op->mem.floatptr) { |
| 794 |
switch (op->size) { |
| 795 |
case 4: |
| 796 |
d+=sprintf(d, "single ptr "); |
| 797 |
break; |
| 798 |
case 8: |
| 799 |
d+=sprintf(d, "double ptr "); |
| 800 |
break; |
| 801 |
case 10: |
| 802 |
d+=sprintf(d, "extended ptr "); |
| 803 |
break; |
| 804 |
} |
| 805 |
} else { |
| 806 |
switch (op->size) { |
| 807 |
case 1: |
| 808 |
d+=sprintf(d, "byte ptr "); |
| 809 |
break; |
| 810 |
case 2: |
| 811 |
d+=sprintf(d, "word ptr "); |
| 812 |
break; |
| 813 |
case 4: |
| 814 |
d+=sprintf(d, "dword ptr "); |
| 815 |
break; |
| 816 |
case 6: |
| 817 |
d+=sprintf(d, "pword ptr "); |
| 818 |
break; |
| 819 |
case 8: |
| 820 |
d+=sprintf(d, "qword ptr "); |
| 821 |
break; |
| 822 |
} |
| 823 |
} |
| 824 |
} |
| 825 |
if (insn->eaddrsize==X86_ADDRSIZE16) reg=1; else reg=2; |
| 826 |
if (insn->segprefix!=X86_PREFIX_NO) d+=sprintf(d, "%s%s:%s", x86_segs[insn->segprefix], cs_symbol, cs_default); |
| 827 |
strcpy(d, cs_symbol); d += strlen(cs_symbol); |
| 828 |
*(d++)='['; |
| 829 |
strcpy(d, cs_default); d += strlen(cs_default); |
| 830 |
|
| 831 |
int optimize_addr=options & X86DIS_STYLE_OPTIMIZE_ADDR; |
| 832 |
int first=1; |
| 833 |
if (optimize_addr && (op->mem.base!=X86_REG_NO) && (op->mem.base==op->mem.index)) { |
| 834 |
d+=sprintf(d, "%s%s*%s%d%s", x86_regs[reg][op->mem.index], cs_symbol, cs_number, op->mem.scale+1, cs_default); |
| 835 |
first=0; |
| 836 |
} else { |
| 837 |
if (op->mem.base!=X86_REG_NO) { |
| 838 |
d+=sprintf(d, "%s", x86_regs[reg][op->mem.base]); |
| 839 |
first=0; |
| 840 |
} |
| 841 |
if (op->mem.index!=X86_REG_NO) { |
| 842 |
if (!first) *(d++)='+'; |
| 843 |
if ((op->mem.scale==1)) { |
| 844 |
d+=sprintf(d, "%s", x86_regs[reg][op->mem.index]); |
| 845 |
} else { |
| 846 |
d+=sprintf(d, "%s%s*%s%d%s", x86_regs[reg][op->mem.index], cs_symbol, cs_number, op->mem.scale, cs_default); |
| 847 |
} |
| 848 |
first=0; |
| 849 |
} |
| 850 |
} |
| 851 |
if ((!optimize_addr && op->mem.hasdisp) || (optimize_addr && op->mem.disp) || first) { |
| 852 |
CPU_ADDR a; |
| 853 |
a.addr32.seg = 0; // FIXME: not ok |
| 854 |
a.addr32.offset=op->mem.disp; |
| 855 |
int slen; |
| 856 |
char *s=(addr_sym_func) ? addr_sym_func(a, &slen, addr_sym_func_context) : 0; |
| 857 |
if (s) { |
| 858 |
if (!first) { |
| 859 |
strcpy(d, cs_symbol); d += strlen(cs_symbol); |
| 860 |
*(d++)='+'; |
| 861 |
strcpy(d, cs_default); d += strlen(cs_default); |
| 862 |
} |
| 863 |
memmove(d, s, slen); |
| 864 |
d+=slen; |
| 865 |
*opstrlen=d-opstr; |
| 866 |
} else { |
| 867 |
dword q; |
| 868 |
switch (op->mem.addrsize) { |
| 869 |
case X86_ADDRSIZE16: |
| 870 |
q=op->mem.disp; |
| 871 |
if (!first) { |
| 872 |
strcpy(d, cs_symbol); d += strlen(cs_symbol); |
| 873 |
if (op->mem.disp&0x80000000) { |
| 874 |
*(d++)='-'; |
| 875 |
q=-q; |
| 876 |
} else *(d++)='+'; |
| 877 |
} |
| 878 |
strcpy(d, cs_number); d += strlen(cs_number); |
| 879 |
hexd(&d, 4, options, q); |
| 880 |
strcpy(d, cs_default); d += strlen(cs_default); |
| 881 |
break; |
| 882 |
case X86_ADDRSIZE32: |
| 883 |
q=op->mem.disp; |
| 884 |
if (!first) { |
| 885 |
strcpy(d, cs_symbol); d += strlen(cs_symbol); |
| 886 |
if (op->mem.disp&0x80000000) { |
| 887 |
*(d++)='-'; |
| 888 |
q=-q; |
| 889 |
} else *(d++)='+'; |
| 890 |
} |
| 891 |
strcpy(d, cs_number); d += strlen(cs_number); |
| 892 |
hexd(&d, 8, options, q); |
| 893 |
strcpy(d, cs_default); d += strlen(cs_default); |
| 894 |
break; |
| 895 |
} |
| 896 |
} |
| 897 |
} |
| 898 |
strcpy(d, cs_symbol); d += strlen(cs_symbol); |
| 899 |
*(d++)=']'; |
| 900 |
strcpy(d, cs_default); d += strlen(cs_default); |
| 901 |
if (*opstrlen) |
| 902 |
*opstrlen += strlen(cs_symbol) + 1 + strlen(cs_default); |
| 903 |
*d=0; |
| 904 |
break; |
| 905 |
} |
| 906 |
case X86_OPTYPE_FARPTR: { |
| 907 |
CPU_ADDR a; |
| 908 |
/* a.addr32.seg = op->farptr.seg; |
| 909 |
a.addr32.offset = op->farptr.offset; |
| 910 |
*/ |
| 911 |
a.addr32.seg = op->farptr.seg; |
| 912 |
a.addr32.offset = op->farptr.offset; |
| 913 |
int slen; |
| 914 |
char *s=(addr_sym_func) ? addr_sym_func(a, &slen, addr_sym_func_context) : 0; |
| 915 |
if (s) { |
| 916 |
memmove(opstr, s, slen); |
| 917 |
opstr[slen]=0; |
| 918 |
*opstrlen=slen; |
| 919 |
} else { |
| 920 |
char *g=opstr; |
| 921 |
hexd(&g, 4, options, op->farptr.seg); |
| 922 |
strcpy(g, cs_symbol); g += strlen(cs_symbol); |
| 923 |
*(g++)=':'; |
| 924 |
strcpy(g, cs_default); g += strlen(cs_default); |
| 925 |
switch (op->size) { |
| 926 |
case 4: |
| 927 |
hexd(&g, 4, options, op->farptr.offset); |
| 928 |
break; |
| 929 |
case 6: |
| 930 |
hexd(&g, 8, options, op->farptr.offset); |
| 931 |
break; |
| 932 |
} |
| 933 |
} |
| 934 |
break; |
| 935 |
} |
| 936 |
default: |
| 937 |
opstr[0]=0; |
| 938 |
} |
| 939 |
} |
| 940 |
|
| 941 |
void X86Disassembler::str_format(char **str, char **format, char *p, char *n, char *op[3], int oplen[3], char stopchar, int print) |
| 942 |
{ |
| 943 |
const char *cs_default = get_cs(e_cs_default); |
| 944 |
const char *cs_symbol = get_cs(e_cs_symbol); |
| 945 |
|
| 946 |
char *f=*format; |
| 947 |
char *s=*str; |
| 948 |
while (*f) { |
| 949 |
if (*f==stopchar) break; |
| 950 |
switch (*f) { |
| 951 |
case '\t': |
| 952 |
if (print) do *(s++)=' '; while ((s-insnstr) % DIS_STYLE_TABSIZE); |
| 953 |
break; |
| 954 |
case DISASM_STRF_VAR: |
| 955 |
f++; |
| 956 |
if (print) { |
| 957 |
char *t=0; |
| 958 |
int tl=0; |
| 959 |
switch (*f) { |
| 960 |
case DISASM_STRF_PREFIX: |
| 961 |
t=p; |
| 962 |
break; |
| 963 |
case DISASM_STRF_NAME: |
| 964 |
t=n; |
| 965 |
break; |
| 966 |
case DISASM_STRF_FIRST: |
| 967 |
t=op[0]; |
| 968 |
tl=oplen[0]; |
| 969 |
break; |
| 970 |
case DISASM_STRF_SECOND: |
| 971 |
t=op[1]; |
| 972 |
tl=oplen[1]; |
| 973 |
break; |
| 974 |
case DISASM_STRF_THIRD: |
| 975 |
t=op[2]; |
| 976 |
tl=oplen[2]; |
| 977 |
break; |
| 978 |
} |
| 979 |
if (tl) { |
| 980 |
memmove(s, t, tl); |
| 981 |
s+=tl; |
| 982 |
*s=0; |
| 983 |
} else { |
| 984 |
strcpy(s, t); |
| 985 |
s += strlen(s); |
| 986 |
} |
| 987 |
} |
| 988 |
break; |
| 989 |
case DISASM_STRF_COND: { |
| 990 |
char *t=0; |
| 991 |
f++; |
| 992 |
switch (*f) { |
| 993 |
case DISASM_STRF_PREFIX: |
| 994 |
t=p; |
| 995 |
break; |
| 996 |
case DISASM_STRF_NAME: |
| 997 |
t=n; |
| 998 |
break; |
| 999 |
case DISASM_STRF_FIRST: |
| 1000 |
t=op[0]; |
| 1001 |
break; |
| 1002 |
case DISASM_STRF_SECOND: |
| 1003 |
t=op[1]; |
| 1004 |
break; |
| 1005 |
case DISASM_STRF_THIRD: |
| 1006 |
t=op[2]; |
| 1007 |
break; |
| 1008 |
} |
| 1009 |
f+=2; |
| 1010 |
if ((t) && (t[0])) { |
| 1011 |
str_format(&s, &f, p, n, op, oplen, *(f-1), 1); |
| 1012 |
} else { |
| 1013 |
str_format(&s, &f, p, n, op, oplen, *(f-1), 0); |
| 1014 |
} |
| 1015 |
break; |
| 1016 |
} |
| 1017 |
default: |
| 1018 |
if (print) { |
| 1019 |
bool x = (strchr(",.-=+-*/[]()", *f) != NULL) && *f; |
| 1020 |
if (x) { strcpy(s, cs_symbol); s += strlen(cs_symbol); } |
| 1021 |
*(s++) = *f; |
| 1022 |
if (x) { strcpy(s, cs_default); s += strlen(cs_default); } |
| 1023 |
} |
| 1024 |
} |
| 1025 |
f++; |
| 1026 |
} |
| 1027 |
*s=0; |
| 1028 |
*format=f; |
| 1029 |
*str=s; |
| 1030 |
} |
| 1031 |
|
| 1032 |
char *X86Disassembler::str(dis_insn *disasm_insn, int options) |
| 1033 |
{ |
| 1034 |
return strf(disasm_insn, options, DISASM_STRF_DEFAULT_FORMAT); |
| 1035 |
} |
| 1036 |
|
| 1037 |
char *X86Disassembler::strf(dis_insn *disasm_insn, int opt, char *format) |
| 1038 |
{ |
| 1039 |
x86dis_insn *insn = (x86dis_insn*)disasm_insn; |
| 1040 |
char prefix[64]; |
| 1041 |
char *p = prefix; |
| 1042 |
options = opt; |
| 1043 |
*p = 0; |
| 1044 |
if (insn->lockprefix == X86_PREFIX_LOCK) p+=sprintf(p, "lock "); |
| 1045 |
if (insn->repprefix == X86_PREFIX_REPZ) { |
| 1046 |
p += sprintf(p, "repz "); |
| 1047 |
} else if (insn->repprefix == X86_PREFIX_REPNZ) { |
| 1048 |
p += sprintf(p, "repnz "); |
| 1049 |
} |
| 1050 |
if ((p != prefix) && (*(p-1) == ' ')) { |
| 1051 |
p--; |
| 1052 |
*p = 0; |
| 1053 |
} |
| 1054 |
|
| 1055 |
bool explicit_params = ((options & X86DIS_STYLE_EXPLICIT_MEMSIZE) || |
| 1056 |
special_param_ambiguity(insn)); |
| 1057 |
|
| 1058 |
char ops[3][512]; /* FIXME: possible buffer overflow ! */ |
| 1059 |
char *op[3]; |
| 1060 |
int oplen[3]; |
| 1061 |
|
| 1062 |
if (options & DIS_STYLE_HIGHLIGHT) enable_highlighting(); |
| 1063 |
for (int i=0; i<3; i++) { |
| 1064 |
op[i]=(char*)&ops[i]; |
| 1065 |
str_op(op[i], &oplen[i], insn, &insn->op[i], explicit_params); |
| 1066 |
} |
| 1067 |
char *s=insnstr; |
| 1068 |
char n[32]; |
| 1069 |
switch (insn->eopsize) { |
| 1070 |
case X86_OPSIZE16: |
| 1071 |
sprintf(n, insn->name, 'w'); |
| 1072 |
break; |
| 1073 |
case X86_OPSIZE32: |
| 1074 |
sprintf(n, insn->name, 'd'); |
| 1075 |
break; |
| 1076 |
default: |
| 1077 |
strcpy(n, insn->name); |
| 1078 |
break; |
| 1079 |
} |
| 1080 |
str_format(&s, &format, prefix, n, op, oplen, 0, 1); |
| 1081 |
disable_highlighting(); |
| 1082 |
return insnstr; |
| 1083 |
} |
| 1084 |
|
| 1085 |
bool X86Disassembler::validInsn(dis_insn *disasm_insn) |
| 1086 |
{ |
| 1087 |
return !((x86dis_insn *)disasm_insn)->invalid; |
| 1088 |
} |
| 1089 |
|
Parent Directory
| 
Revision Log