| | 1 | #define _GNU_SOURCE |
| | 2 | #include <stdio.h> |
| | 3 | #include <stdlib.h> |
| | 4 | #include <string.h> |
| | 5 | |
| | 6 | #include "my_assert.h" |
| | 7 | #include "my_str.h" |
| | 8 | |
| | 9 | #define ARRAY_SIZE(x) (sizeof(x) / sizeof(x[0])) |
| | 10 | #define IS(w, y) !strcmp(w, y) |
| | 11 | |
| | 12 | #include "protoparse.h" |
| | 13 | |
| | 14 | const char *asmfn; |
| | 15 | static int asmln; |
| | 16 | |
| | 17 | #define anote(fmt, ...) \ |
| | 18 | printf("%s:%d: note: " fmt, asmfn, asmln, ##__VA_ARGS__) |
| | 19 | #define awarn(fmt, ...) \ |
| | 20 | printf("%s:%d: warning: " fmt, asmfn, asmln, ##__VA_ARGS__) |
| | 21 | #define aerr(fmt, ...) do { \ |
| | 22 | printf("%s:%d: error: " fmt, asmfn, asmln, ##__VA_ARGS__); \ |
| | 23 | fcloseall(); \ |
| | 24 | exit(1); \ |
| | 25 | } while (0) |
| | 26 | |
| | 27 | enum op_flags { |
| | 28 | OPF_RMD = (1 << 0), /* removed or optimized out */ |
| | 29 | OPF_DATA = (1 << 1), /* data processing - writes to dst opr */ |
| | 30 | OPF_FLAGS = (1 << 2), /* sets flags */ |
| | 31 | OPF_JMP = (1 << 3), /* branches, ret and call */ |
| | 32 | OPF_CC = (1 << 4), /* uses flags */ |
| | 33 | OPF_TAIL = (1 << 5), /* ret or tail call */ |
| | 34 | OPF_REP = (1 << 6), /* prefixed by rep */ |
| | 35 | }; |
| | 36 | |
| | 37 | enum op_op { |
| | 38 | OP_INVAL, |
| | 39 | OP_NOP, |
| | 40 | OP_PUSH, |
| | 41 | OP_POP, |
| | 42 | OP_MOV, |
| | 43 | OP_LEA, |
| | 44 | OP_MOVZX, |
| | 45 | OP_MOVSX, |
| | 46 | OP_NOT, |
| | 47 | OP_CDQ, |
| | 48 | OP_STOS, |
| | 49 | OP_RET, |
| | 50 | OP_ADD, |
| | 51 | OP_SUB, |
| | 52 | OP_AND, |
| | 53 | OP_OR, |
| | 54 | OP_XOR, |
| | 55 | OP_SHL, |
| | 56 | OP_SHR, |
| | 57 | OP_SAR, |
| | 58 | OP_ADC, |
| | 59 | OP_SBB, |
| | 60 | OP_INC, |
| | 61 | OP_DEC, |
| | 62 | OP_NEG, |
| | 63 | OP_MUL, |
| | 64 | OP_IMUL, |
| | 65 | OP_DIV, |
| | 66 | OP_IDIV, |
| | 67 | OP_TEST, |
| | 68 | OP_CMP, |
| | 69 | OP_CALL, |
| | 70 | OP_JMP, |
| | 71 | OP_JO, |
| | 72 | OP_JNO, |
| | 73 | OP_JC, |
| | 74 | OP_JNC, |
| | 75 | OP_JZ, |
| | 76 | OP_JNZ, |
| | 77 | OP_JBE, |
| | 78 | OP_JA, |
| | 79 | OP_JS, |
| | 80 | OP_JNS, |
| | 81 | OP_JP, |
| | 82 | OP_JNP, |
| | 83 | OP_JL, |
| | 84 | OP_JGE, |
| | 85 | OP_JLE, |
| | 86 | OP_JG, |
| | 87 | }; |
| | 88 | |
| | 89 | enum opr_type { |
| | 90 | OPT_UNSPEC, |
| | 91 | OPT_REG, |
| | 92 | OPT_REGMEM, |
| | 93 | OPT_LABEL, |
| | 94 | OPT_OFFSET, |
| | 95 | OPT_CONST, |
| | 96 | }; |
| | 97 | |
| | 98 | enum opr_lenmod { |
| | 99 | OPLM_UNSPEC, |
| | 100 | OPLM_BYTE, |
| | 101 | OPLM_WORD, |
| | 102 | OPLM_DWORD, |
| | 103 | }; |
| | 104 | |
| | 105 | #define MAX_OPERANDS 3 |
| | 106 | |
| | 107 | struct parsed_opr { |
| | 108 | enum opr_type type; |
| | 109 | enum opr_lenmod lmod; |
| | 110 | int reg; |
| | 111 | unsigned int val; |
| | 112 | char name[256]; |
| | 113 | }; |
| | 114 | |
| | 115 | struct parsed_op { |
| | 116 | enum op_op op; |
| | 117 | struct parsed_opr operand[MAX_OPERANDS]; |
| | 118 | unsigned int flags; |
| | 119 | int operand_cnt; |
| | 120 | int regmask_src; // all referensed regs |
| | 121 | int regmask_dst; |
| | 122 | int pfomask; // flagop: parsed_flag_op that can't be delayed |
| | 123 | int argmask; // push: args that are altered before call |
| | 124 | int cc_scratch; // scratch storage during analysis |
| | 125 | int bt_i; // branch target (for branches) |
| | 126 | struct parsed_op *lrl; // label reference list entry |
| | 127 | void *datap; |
| | 128 | }; |
| | 129 | |
| | 130 | // datap: |
| | 131 | // OP_CALL - ptr to parsed_proto |
| | 132 | // (OPF_CC) - point to corresponding (OPF_FLAGS) |
| | 133 | |
| | 134 | struct parsed_equ { |
| | 135 | char name[64]; |
| | 136 | enum opr_lenmod lmod; |
| | 137 | int offset; |
| | 138 | }; |
| | 139 | |
| | 140 | #define MAX_OPS 1024 |
| | 141 | |
| | 142 | static struct parsed_op ops[MAX_OPS]; |
| | 143 | static struct parsed_equ *g_eqs; |
| | 144 | static int g_eqcnt; |
| | 145 | static char g_labels[MAX_OPS][32]; |
| | 146 | static struct parsed_op *g_label_refs[MAX_OPS]; |
| | 147 | static struct parsed_proto g_func_pp; |
| | 148 | static char g_func[256]; |
| | 149 | static char g_comment[256]; |
| | 150 | static int g_bp_frame; |
| | 151 | static int g_bp_stack; |
| | 152 | #define ferr(op_, fmt, ...) do { \ |
| | 153 | printf("error:%s:#%ld: '%s': " fmt, g_func, (op_) - ops, \ |
| | 154 | dump_op(op_), ##__VA_ARGS__); \ |
| | 155 | fcloseall(); \ |
| | 156 | exit(1); \ |
| | 157 | } while (0) |
| | 158 | |
| | 159 | #define MAX_REGS 8 |
| | 160 | |
| | 161 | const char *regs_r32[] = { "eax", "ebx", "ecx", "edx", "esi", "edi", "ebp", "esp" }; |
| | 162 | const char *regs_r16[] = { "ax", "bx", "cx", "dx", "si", "di", "bp", "sp" }; |
| | 163 | const char *regs_r8l[] = { "al", "bl", "cl", "dl" }; |
| | 164 | const char *regs_r8h[] = { "ah", "bh", "ch", "dh" }; |
| | 165 | |
| | 166 | enum x86_regs { xUNSPEC = -1, xAX, xBX, xCX, xDX, xSI, xDI, xBP, xSP }; |
| | 167 | |
| | 168 | // possible basic comparison types (without inversion) |
| | 169 | enum parsed_flag_op { |
| | 170 | PFO_O, // 0 OF=1 |
| | 171 | PFO_C, // 2 CF=1 |
| | 172 | PFO_Z, // 4 ZF=1 |
| | 173 | PFO_BE, // 6 CF=1||ZF=1 |
| | 174 | PFO_S, // 8 SF=1 |
| | 175 | PFO_P, // a PF=1 |
| | 176 | PFO_L, // c SF!=OF |
| | 177 | PFO_LE, // e ZF=1||SF!=OF |
| | 178 | }; |
| | 179 | |
| | 180 | static const char *parsed_flag_op_names[] = { |
| | 181 | "o", "c", "z", "be", "s", "p", "l", "le" |
| | 182 | }; |
| | 183 | |
| | 184 | static int char_array_i(const char *array[], size_t len, const char *s) |
| | 185 | { |
| | 186 | int i; |
| | 187 | |
| | 188 | for (i = 0; i < len; i++) |
| | 189 | if (IS(s, array[i])) |
| | 190 | return i; |
| | 191 | |
| | 192 | return -1; |
| | 193 | } |
| | 194 | |
| | 195 | static void printf_number(char *buf, size_t buf_size, long number) |
| | 196 | { |
| | 197 | // output in C-friendly form |
| | 198 | snprintf(buf, buf_size, number < 10 ? "%lu" : "0x%02lx", number); |
| | 199 | } |
| | 200 | |
| | 201 | static int parse_reg(enum opr_lenmod *reg_lmod, const char *s) |
| | 202 | { |
| | 203 | int reg; |
| | 204 | |
| | 205 | reg = char_array_i(regs_r32, ARRAY_SIZE(regs_r32), s); |
| | 206 | if (reg >= 0) { |
| | 207 | *reg_lmod = OPLM_DWORD; |
| | 208 | return reg; |
| | 209 | } |
| | 210 | reg = char_array_i(regs_r16, ARRAY_SIZE(regs_r16), s); |
| | 211 | if (reg >= 0) { |
| | 212 | *reg_lmod = OPLM_WORD; |
| | 213 | return reg; |
| | 214 | } |
| | 215 | reg = char_array_i(regs_r8h, ARRAY_SIZE(regs_r8h), s); |
| | 216 | if (reg >= 0) { |
| | 217 | *reg_lmod = OPLM_BYTE; |
| | 218 | return reg; |
| | 219 | } |
| | 220 | reg = char_array_i(regs_r8l, ARRAY_SIZE(regs_r8l), s); |
| | 221 | if (reg >= 0) { |
| | 222 | *reg_lmod = OPLM_BYTE; |
| | 223 | return reg; |
| | 224 | } |
| | 225 | |
| | 226 | return -1; |
| | 227 | } |
| | 228 | |
| | 229 | static long parse_number(const char *number) |
| | 230 | { |
| | 231 | int len = strlen(number); |
| | 232 | const char *p = number; |
| | 233 | char *endp = NULL; |
| | 234 | int neg = 0; |
| | 235 | int bad; |
| | 236 | long ret; |
| | 237 | |
| | 238 | if (*p == '-') { |
| | 239 | neg = 1; |
| | 240 | p++; |
| | 241 | } |
| | 242 | if (len > 1 && *p == '0') |
| | 243 | p++; |
| | 244 | if (number[len - 1] == 'h') { |
| | 245 | ret = strtol(p, &endp, 16); |
| | 246 | bad = (*endp != 'h'); |
| | 247 | } |
| | 248 | else { |
| | 249 | ret = strtol(p, &endp, 10); |
| | 250 | bad = (*endp != 0); |
| | 251 | } |
| | 252 | if (bad) |
| | 253 | aerr("number parsing failed\n"); |
| | 254 | if (neg) |
| | 255 | ret = -ret; |
| | 256 | return ret; |
| | 257 | } |
| | 258 | |
| | 259 | static int parse_indmode(char *name, int *regmask, int need_c_cvt) |
| | 260 | { |
| | 261 | enum opr_lenmod lmod; |
| | 262 | char cvtbuf[256]; |
| | 263 | char *d = cvtbuf; |
| | 264 | char *s = name; |
| | 265 | char w[64]; |
| | 266 | long number; |
| | 267 | int reg; |
| | 268 | int c = 0; |
| | 269 | |
| | 270 | *d = 0; |
| | 271 | |
| | 272 | while (*s != 0) { |
| | 273 | d += strlen(d); |
| | 274 | while (my_isblank(*s)) |
| | 275 | s++; |
| | 276 | for (; my_issep(*s); d++, s++) |
| | 277 | *d = *s; |
| | 278 | while (my_isblank(*s)) |
| | 279 | s++; |
| | 280 | *d = 0; |
| | 281 | |
| | 282 | // skip 'ds:' prefix |
| | 283 | if (!strncmp(s, "ds:", 3)) |
| | 284 | s += 3; |
| | 285 | |
| | 286 | s = next_idt(w, sizeof(w), s); |
| | 287 | if (w[0] == 0) |
| | 288 | break; |
| | 289 | c++; |
| | 290 | |
| | 291 | reg = parse_reg(&lmod, w); |
| | 292 | if (reg >= 0) { |
| | 293 | *regmask |= 1 << reg; |
| | 294 | goto pass; |
| | 295 | } |
| | 296 | |
| | 297 | if ('0' <= w[0] && w[0] <= '9') { |
| | 298 | number = parse_number(w); |
| | 299 | printf_number(d, sizeof(cvtbuf) - (d - cvtbuf), number); |
| | 300 | continue; |
| | 301 | } |
| | 302 | |
| | 303 | // probably some label/identifier - pass |
| | 304 | |
| | 305 | pass: |
| | 306 | snprintf(d, sizeof(cvtbuf) - (d - cvtbuf), "%s", w); |
| | 307 | } |
| | 308 | |
| | 309 | if (need_c_cvt) |
| | 310 | strcpy(name, cvtbuf); |
| | 311 | |
| | 312 | return c; |
| | 313 | } |
| | 314 | |
| | 315 | static int guess_lmod_from_name(struct parsed_opr *opr) |
| | 316 | { |
| | 317 | if (!strncmp(opr->name, "dword_", 6)) { |
| | 318 | opr->lmod = OPLM_DWORD; |
| | 319 | return 1; |
| | 320 | } |
| | 321 | if (!strncmp(opr->name, "word_", 5)) { |
| | 322 | opr->lmod = OPLM_WORD; |
| | 323 | return 1; |
| | 324 | } |
| | 325 | if (!strncmp(opr->name, "byte_", 5)) { |
| | 326 | opr->lmod = OPLM_BYTE; |
| | 327 | return 1; |
| | 328 | } |
| | 329 | return 0; |
| | 330 | } |
| | 331 | |
| | 332 | static void setup_reg_opr(struct parsed_opr *opr, int reg, enum opr_lenmod lmod, |
| | 333 | int *regmask) |
| | 334 | { |
| | 335 | opr->type = OPT_REG; |
| | 336 | opr->reg = reg; |
| | 337 | opr->lmod = lmod; |
| | 338 | *regmask |= 1 << reg; |
| | 339 | } |
| | 340 | |
| | 341 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name); |
| | 342 | |
| | 343 | static int parse_operand(struct parsed_opr *opr, |
| | 344 | int *regmask, int *regmask_indirect, |
| | 345 | char words[16][256], int wordc, int w, unsigned int op_flags) |
| | 346 | { |
| | 347 | enum opr_lenmod tmplmod; |
| | 348 | int ret, len; |
| | 349 | long number; |
| | 350 | int i; |
| | 351 | |
| | 352 | if (w >= wordc) |
| | 353 | aerr("parse_operand w %d, wordc %d\n", w, wordc); |
| | 354 | |
| | 355 | opr->reg = xUNSPEC; |
| | 356 | |
| | 357 | for (i = w; i < wordc; i++) { |
| | 358 | len = strlen(words[i]); |
| | 359 | if (words[i][len - 1] == ',') { |
| | 360 | words[i][len - 1] = 0; |
| | 361 | wordc = i + 1; |
| | 362 | break; |
| | 363 | } |
| | 364 | } |
| | 365 | |
| | 366 | if (op_flags & OPF_JMP) { |
| | 367 | const char *label; |
| | 368 | |
| | 369 | if (wordc - w == 3 && IS(words[w + 1], "ptr")) |
| | 370 | label = words[w + 2]; |
| | 371 | else if (wordc - w == 2 && IS(words[w], "short")) |
| | 372 | label = words[w + 1]; |
| | 373 | else if (wordc - w == 1) |
| | 374 | label = words[w]; |
| | 375 | else |
| | 376 | aerr("jump parse error"); |
| | 377 | |
| | 378 | opr->type = OPT_LABEL; |
| | 379 | strcpy(opr->name, label); |
| | 380 | return wordc; |
| | 381 | } |
| | 382 | |
| | 383 | if (wordc - w >= 3) { |
| | 384 | if (IS(words[w + 1], "ptr")) { |
| | 385 | if (IS(words[w], "dword")) |
| | 386 | opr->lmod = OPLM_DWORD; |
| | 387 | else if (IS(words[w], "word")) |
| | 388 | opr->lmod = OPLM_WORD; |
| | 389 | else if (IS(words[w], "byte")) |
| | 390 | opr->lmod = OPLM_BYTE; |
| | 391 | else |
| | 392 | aerr("type parsing failed\n"); |
| | 393 | w += 2; |
| | 394 | } |
| | 395 | } |
| | 396 | |
| | 397 | if (wordc - w == 2 && IS(words[w], "offset")) { |
| | 398 | opr->type = OPT_OFFSET; |
| | 399 | strcpy(opr->name, words[w + 1]); |
| | 400 | return wordc; |
| | 401 | } |
| | 402 | |
| | 403 | if (wordc - w != 1) |
| | 404 | aerr("parse_operand 1 word expected\n"); |
| | 405 | |
| | 406 | strcpy(opr->name, words[w]); |
| | 407 | |
| | 408 | if (words[w][0] == '[') { |
| | 409 | opr->type = OPT_REGMEM; |
| | 410 | ret = sscanf(words[w], "[%[^]]]", opr->name); |
| | 411 | if (ret != 1) |
| | 412 | aerr("[] parse failure\n"); |
| | 413 | |
| | 414 | parse_indmode(opr->name, regmask_indirect, 1); |
| | 415 | if (opr->lmod == OPLM_UNSPEC && !strncmp(opr->name, "ebp+", 4)) { |
| | 416 | // might be an equ |
| | 417 | struct parsed_equ *eq = equ_find(NULL, opr->name + 4); |
| | 418 | if (eq) |
| | 419 | opr->lmod = eq->lmod; |
| | 420 | } |
| | 421 | return wordc; |
| | 422 | } |
| | 423 | else if (strchr(words[w], '[')) { |
| | 424 | // label[reg] form |
| | 425 | opr->type = OPT_REGMEM; |
| | 426 | if (opr->lmod == OPLM_UNSPEC) |
| | 427 | guess_lmod_from_name(opr); |
| | 428 | parse_indmode(strchr(words[w], '['), regmask_indirect, 0); |
| | 429 | return wordc; |
| | 430 | } |
| | 431 | else if (('0' <= words[w][0] && words[w][0] <= '9') |
| | 432 | || words[w][0] == '-') |
| | 433 | { |
| | 434 | number = parse_number(words[w]); |
| | 435 | opr->type = OPT_CONST; |
| | 436 | opr->val = number; |
| | 437 | printf_number(opr->name, sizeof(opr->name), number); |
| | 438 | return wordc; |
| | 439 | } |
| | 440 | |
| | 441 | ret = parse_reg(&tmplmod, opr->name); |
| | 442 | if (ret >= 0) { |
| | 443 | setup_reg_opr(opr, ret, tmplmod, regmask); |
| | 444 | return wordc; |
| | 445 | } |
| | 446 | |
| | 447 | // most likely var in data segment |
| | 448 | opr->type = OPT_LABEL; |
| | 449 | if (opr->lmod == OPLM_UNSPEC) |
| | 450 | guess_lmod_from_name(opr); |
| | 451 | if (opr->lmod != OPLM_UNSPEC) |
| | 452 | return wordc; |
| | 453 | |
| | 454 | // TODO: scan data seg to determine type? |
| | 455 | return wordc; |
| | 456 | } |
| | 457 | |
| | 458 | static const struct { |
| | 459 | const char *name; |
| | 460 | unsigned int flags; |
| | 461 | } pref_table[] = { |
| | 462 | { "rep", OPF_REP }, |
| | 463 | }; |
| | 464 | |
| | 465 | static const struct { |
| | 466 | const char *name; |
| | 467 | enum op_op op; |
| | 468 | unsigned int minopr; |
| | 469 | unsigned int maxopr; |
| | 470 | unsigned int flags; |
| | 471 | } op_table[] = { |
| | 472 | { "nop", OP_NOP, 0, 0, 0 }, |
| | 473 | { "push", OP_PUSH, 1, 1, 0 }, |
| | 474 | { "pop", OP_POP, 1, 1, OPF_DATA }, |
| | 475 | { "mov" , OP_MOV, 2, 2, OPF_DATA }, |
| | 476 | { "lea", OP_LEA, 2, 2, OPF_DATA }, |
| | 477 | { "movzx",OP_MOVZX, 2, 2, OPF_DATA }, |
| | 478 | { "movsx",OP_MOVSX, 2, 2, OPF_DATA }, |
| | 479 | { "not", OP_NOT, 1, 1, OPF_DATA }, |
| | 480 | { "cdq", OP_CDQ, 0, 0, OPF_DATA }, |
| | 481 | { "stosb",OP_STOS, 0, 0, OPF_DATA }, |
| | 482 | { "stosw",OP_STOS, 0, 0, OPF_DATA }, |
| | 483 | { "stosd",OP_STOS, 0, 0, OPF_DATA }, |
| | 484 | { "add", OP_ADD, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 485 | { "sub", OP_SUB, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 486 | { "and", OP_AND, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 487 | { "or", OP_OR, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 488 | { "xor", OP_XOR, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 489 | { "shl", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 490 | { "shr", OP_SHR, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 491 | { "sal", OP_SHL, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 492 | { "sar", OP_SAR, 2, 2, OPF_DATA|OPF_FLAGS }, |
| | 493 | { "adc", OP_ADC, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
| | 494 | { "sbb", OP_SBB, 2, 2, OPF_DATA|OPF_FLAGS|OPF_CC }, |
| | 495 | { "inc", OP_INC, 1, 1, OPF_DATA|OPF_FLAGS }, |
| | 496 | { "dec", OP_DEC, 1, 1, OPF_DATA|OPF_FLAGS }, |
| | 497 | { "neg", OP_NEG, 1, 1, OPF_DATA|OPF_FLAGS }, |
| | 498 | { "mul", OP_MUL, 1, 1, OPF_DATA|OPF_FLAGS }, |
| | 499 | { "imul", OP_IMUL, 1, 3, OPF_DATA|OPF_FLAGS }, |
| | 500 | { "div", OP_DIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
| | 501 | { "idiv", OP_IDIV, 1, 1, OPF_DATA|OPF_FLAGS }, |
| | 502 | { "test", OP_TEST, 2, 2, OPF_FLAGS }, |
| | 503 | { "cmp", OP_CMP, 2, 2, OPF_FLAGS }, |
| | 504 | { "retn", OP_RET, 0, 1, OPF_JMP|OPF_TAIL }, |
| | 505 | { "call", OP_CALL, 1, 1, OPF_JMP|OPF_FLAGS }, |
| | 506 | { "jmp", OP_JMP, 1, 1, OPF_JMP }, |
| | 507 | { "jo", OP_JO, 1, 1, OPF_JMP|OPF_CC }, // 70 OF=1 |
| | 508 | { "jno", OP_JNO, 1, 1, OPF_JMP|OPF_CC }, // 71 OF=0 |
| | 509 | { "jc", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 CF=1 |
| | 510 | { "jb", OP_JC, 1, 1, OPF_JMP|OPF_CC }, // 72 |
| | 511 | { "jnc", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 CF=0 |
| | 512 | { "jnb", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
| | 513 | { "jae", OP_JNC, 1, 1, OPF_JMP|OPF_CC }, // 73 |
| | 514 | { "jz", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 ZF=1 |
| | 515 | { "je", OP_JZ, 1, 1, OPF_JMP|OPF_CC }, // 74 |
| | 516 | { "jnz", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 ZF=0 |
| | 517 | { "jne", OP_JNZ, 1, 1, OPF_JMP|OPF_CC }, // 75 |
| | 518 | { "jbe", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 CF=1 || ZF=1 |
| | 519 | { "jna", OP_JBE, 1, 1, OPF_JMP|OPF_CC }, // 76 |
| | 520 | { "ja", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 CF=0 && ZF=0 |
| | 521 | { "jnbe", OP_JA, 1, 1, OPF_JMP|OPF_CC }, // 77 |
| | 522 | { "js", OP_JS, 1, 1, OPF_JMP|OPF_CC }, // 78 SF=1 |
| | 523 | { "jns", OP_JNS, 1, 1, OPF_JMP|OPF_CC }, // 79 SF=0 |
| | 524 | { "jp", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a PF=1 |
| | 525 | { "jpe", OP_JP, 1, 1, OPF_JMP|OPF_CC }, // 7a |
| | 526 | { "jnp", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b PF=0 |
| | 527 | { "jpo", OP_JNP, 1, 1, OPF_JMP|OPF_CC }, // 7b |
| | 528 | { "jl", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c SF!=OF |
| | 529 | { "jnge", OP_JL, 1, 1, OPF_JMP|OPF_CC }, // 7c |
| | 530 | { "jge", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d SF=OF |
| | 531 | { "jnl", OP_JGE, 1, 1, OPF_JMP|OPF_CC }, // 7d |
| | 532 | { "jle", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e ZF=1 || SF!=OF |
| | 533 | { "jng", OP_JLE, 1, 1, OPF_JMP|OPF_CC }, // 7e |
| | 534 | { "jg", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f ZF=0 && SF=OF |
| | 535 | { "jnle", OP_JG, 1, 1, OPF_JMP|OPF_CC }, // 7f |
| | 536 | { "seto", OP_JO, 1, 1, OPF_DATA|OPF_CC }, |
| | 537 | { "setno", OP_JNO, 1, 1, OPF_DATA|OPF_CC }, |
| | 538 | { "setc", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
| | 539 | { "setb", OP_JC, 1, 1, OPF_DATA|OPF_CC }, |
| | 540 | { "setnc", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
| | 541 | { "setae", OP_JNC, 1, 1, OPF_DATA|OPF_CC }, |
| | 542 | { "setz", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
| | 543 | { "sete", OP_JZ, 1, 1, OPF_DATA|OPF_CC }, |
| | 544 | { "setnz", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
| | 545 | { "setne", OP_JNZ, 1, 1, OPF_DATA|OPF_CC }, |
| | 546 | { "setbe", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
| | 547 | { "setna", OP_JBE, 1, 1, OPF_DATA|OPF_CC }, |
| | 548 | { "seta", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
| | 549 | { "setnbe", OP_JA, 1, 1, OPF_DATA|OPF_CC }, |
| | 550 | { "sets", OP_JS, 1, 1, OPF_DATA|OPF_CC }, |
| | 551 | { "setns", OP_JNS, 1, 1, OPF_DATA|OPF_CC }, |
| | 552 | { "setp", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
| | 553 | { "setpe", OP_JP, 1, 1, OPF_DATA|OPF_CC }, |
| | 554 | { "setnp", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
| | 555 | { "setpo", OP_JNP, 1, 1, OPF_DATA|OPF_CC }, |
| | 556 | { "setl", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
| | 557 | { "setnge", OP_JL, 1, 1, OPF_DATA|OPF_CC }, |
| | 558 | { "setge", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
| | 559 | { "setnl", OP_JGE, 1, 1, OPF_DATA|OPF_CC }, |
| | 560 | { "setle", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
| | 561 | { "setng", OP_JLE, 1, 1, OPF_DATA|OPF_CC }, |
| | 562 | { "setg", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
| | 563 | { "setnle", OP_JG, 1, 1, OPF_DATA|OPF_CC }, |
| | 564 | }; |
| | 565 | |
| | 566 | static void parse_op(struct parsed_op *op, char words[16][256], int wordc) |
| | 567 | { |
| | 568 | enum opr_lenmod lmod; |
| | 569 | int prefix_flags = 0; |
| | 570 | int regmask_ind; |
| | 571 | int regmask; |
| | 572 | int op_w = 0; |
| | 573 | int opr = 0; |
| | 574 | int w = 0; |
| | 575 | int i; |
| | 576 | |
| | 577 | for (i = 0; i < ARRAY_SIZE(pref_table); i++) { |
| | 578 | if (IS(words[w], pref_table[i].name)) { |
| | 579 | prefix_flags = pref_table[i].flags; |
| | 580 | break; |
| | 581 | } |
| | 582 | } |
| | 583 | |
| | 584 | if (prefix_flags) { |
| | 585 | if (wordc <= 1) |
| | 586 | aerr("lone prefix: '%s'\n", words[0]); |
| | 587 | w++; |
| | 588 | } |
| | 589 | |
| | 590 | op_w = w; |
| | 591 | for (i = 0; i < ARRAY_SIZE(op_table); i++) { |
| | 592 | if (IS(words[w], op_table[i].name)) |
| | 593 | break; |
| | 594 | } |
| | 595 | |
| | 596 | if (i == ARRAY_SIZE(op_table)) |
| | 597 | aerr("unhandled op: '%s'\n", words[0]); |
| | 598 | w++; |
| | 599 | |
| | 600 | op->op = op_table[i].op; |
| | 601 | op->flags = op_table[i].flags | prefix_flags; |
| | 602 | op->regmask_src = op->regmask_dst = 0; |
| | 603 | |
| | 604 | for (opr = 0; opr < op_table[i].minopr; opr++) { |
| | 605 | regmask = regmask_ind = 0; |
| | 606 | w = parse_operand(&op->operand[opr], ®mask, ®mask_ind, |
| | 607 | words, wordc, w, op->flags); |
| | 608 | |
| | 609 | if (opr == 0 && (op->flags & OPF_DATA)) |
| | 610 | op->regmask_dst = regmask; |
| | 611 | // for now, mark dst as src too |
| | 612 | op->regmask_src |= regmask | regmask_ind; |
| | 613 | } |
| | 614 | |
| | 615 | for (; w < wordc && opr < op_table[i].maxopr; opr++) { |
| | 616 | w = parse_operand(&op->operand[opr], |
| | 617 | &op->regmask_src, &op->regmask_src, |
| | 618 | words, wordc, w, op->flags); |
| | 619 | } |
| | 620 | |
| | 621 | if (w < wordc) |
| | 622 | aerr("parse_op %s incomplete: %d/%d\n", |
| | 623 | words[0], w, wordc); |
| | 624 | |
| | 625 | // special cases |
| | 626 | op->operand_cnt = opr; |
| | 627 | if (!strncmp(op_table[i].name, "set", 3)) |
| | 628 | op->operand[0].lmod = OPLM_BYTE; |
| | 629 | |
| | 630 | // ops with implicit argumets |
| | 631 | switch (op->op) { |
| | 632 | case OP_CDQ: |
| | 633 | op->operand_cnt = 2; |
| | 634 | setup_reg_opr(&op->operand[0], xDX, OPLM_DWORD, &op->regmask_dst); |
| | 635 | setup_reg_opr(&op->operand[1], xAX, OPLM_DWORD, &op->regmask_src); |
| | 636 | break; |
| | 637 | |
| | 638 | case OP_STOS: |
| | 639 | if (op->operand_cnt != 0) |
| | 640 | break; |
| | 641 | if (IS(words[op_w], "stosb")) |
| | 642 | lmod = OPLM_BYTE; |
| | 643 | else if (IS(words[op_w], "stosw")) |
| | 644 | lmod = OPLM_WORD; |
| | 645 | else if (IS(words[op_w], "stosd")) |
| | 646 | lmod = OPLM_DWORD; |
| | 647 | op->operand_cnt = 3; |
| | 648 | setup_reg_opr(&op->operand[0], xDI, lmod, &op->regmask_dst); |
| | 649 | setup_reg_opr(&op->operand[1], xCX, OPLM_DWORD, &op->regmask_dst); |
| | 650 | setup_reg_opr(&op->operand[2], xAX, OPLM_DWORD, &op->regmask_src); |
| | 651 | break; |
| | 652 | |
| | 653 | case OP_IMUL: |
| | 654 | if (op->operand_cnt != 1) |
| | 655 | break; |
| | 656 | // fallthrough |
| | 657 | case OP_MUL: |
| | 658 | // singleop mul |
| | 659 | op->regmask_dst = (1 << xDX) | (1 << xAX); |
| | 660 | op->regmask_src |= (1 << xAX); |
| | 661 | if (op->operand[0].lmod == OPLM_UNSPEC) |
| | 662 | op->operand[0].lmod = OPLM_DWORD; |
| | 663 | break; |
| | 664 | |
| | 665 | case OP_DIV: |
| | 666 | case OP_IDIV: |
| | 667 | // we could set up operands for edx:eax, but there is no real need to |
| | 668 | // (see is_opr_modified()) |
| | 669 | regmask = (1 << xDX) | (1 << xAX); |
| | 670 | op->regmask_dst = regmask; |
| | 671 | op->regmask_src |= regmask; |
| | 672 | if (op->operand[0].lmod == OPLM_UNSPEC) |
| | 673 | op->operand[0].lmod = OPLM_DWORD; |
| | 674 | break; |
| | 675 | |
| | 676 | case OP_SHL: |
| | 677 | case OP_SHR: |
| | 678 | case OP_SAR: |
| | 679 | if (op->operand[1].lmod == OPLM_UNSPEC) |
| | 680 | op->operand[1].lmod = OPLM_BYTE; |
| | 681 | break; |
| | 682 | |
| | 683 | default: |
| | 684 | break; |
| | 685 | } |
| | 686 | } |
| | 687 | |
| | 688 | static const char *op_name(enum op_op op) |
| | 689 | { |
| | 690 | int i; |
| | 691 | |
| | 692 | for (i = 0; i < ARRAY_SIZE(op_table); i++) |
| | 693 | if (op_table[i].op == op) |
| | 694 | return op_table[i].name; |
| | 695 | |
| | 696 | return "???"; |
| | 697 | } |
| | 698 | |
| | 699 | // debug |
| | 700 | static const char *dump_op(struct parsed_op *po) |
| | 701 | { |
| | 702 | static char out[128]; |
| | 703 | char *p = out; |
| | 704 | int i; |
| | 705 | |
| | 706 | snprintf(out, sizeof(out), "%s", op_name(po->op)); |
| | 707 | for (i = 0; i < po->operand_cnt; i++) { |
| | 708 | p += strlen(p); |
| | 709 | if (i > 0) |
| | 710 | *p++ = ','; |
| | 711 | snprintf(p, sizeof(out) - (p - out), |
| | 712 | po->operand[i].type == OPT_REGMEM ? " [%s]" : " %s", |
| | 713 | po->operand[i].name); |
| | 714 | } |
| | 715 | |
| | 716 | return out; |
| | 717 | } |
| | 718 | |
| | 719 | static const char *opr_name(struct parsed_op *po, int opr_num) |
| | 720 | { |
| | 721 | if (opr_num >= po->operand_cnt) |
| | 722 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
| | 723 | return po->operand[opr_num].name; |
| | 724 | } |
| | 725 | |
| | 726 | static unsigned int opr_const(struct parsed_op *po, int opr_num) |
| | 727 | { |
| | 728 | if (opr_num >= po->operand_cnt) |
| | 729 | ferr(po, "opr OOR: %d/%d\n", opr_num, po->operand_cnt); |
| | 730 | if (po->operand[opr_num].type != OPT_CONST) |
| | 731 | ferr(po, "opr %d: const expected\n", opr_num); |
| | 732 | return po->operand[opr_num].val; |
| | 733 | } |
| | 734 | |
| | 735 | static const char *opr_reg_p(struct parsed_op *po, struct parsed_opr *popr) |
| | 736 | { |
| | 737 | if ((unsigned int)popr->reg >= MAX_REGS) |
| | 738 | ferr(po, "invalid reg: %d\n", popr->reg); |
| | 739 | return regs_r32[popr->reg]; |
| | 740 | } |
| | 741 | |
| | 742 | static struct parsed_equ *equ_find(struct parsed_op *po, const char *name) |
| | 743 | { |
| | 744 | int i; |
| | 745 | |
| | 746 | for (i = 0; i < g_eqcnt; i++) |
| | 747 | if (IS(g_eqs[i].name, name)) |
| | 748 | break; |
| | 749 | if (i >= g_eqcnt) { |
| | 750 | if (po != NULL) |
| | 751 | ferr(po, "unresolved equ name: '%s'\n", name); |
| | 752 | return NULL; |
| | 753 | } |
| | 754 | |
| | 755 | return &g_eqs[i]; |
| | 756 | } |
| | 757 | |
| | 758 | static void bg_frame_access(struct parsed_op *po, enum opr_lenmod lmod, |
| | 759 | char *buf, size_t buf_size, const char *bp_arg, |
| | 760 | int is_src, int is_lea) |
| | 761 | { |
| | 762 | const char *prefix = ""; |
| | 763 | struct parsed_equ *eq; |
| | 764 | int i, arg_i, arg_s; |
| | 765 | int sf_ofs; |
| | 766 | |
| | 767 | snprintf(g_comment, sizeof(g_comment), "%s", bp_arg); |
| | 768 | |
| | 769 | eq = equ_find(po, bp_arg); |
| | 770 | |
| | 771 | if (eq->offset >= 0) { |
| | 772 | arg_i = eq->offset / 4 - 2; |
| | 773 | if (arg_i < 0 || arg_i >= g_func_pp.argc_stack) |
| | 774 | ferr(po, "offset %d doesn't map to any arg\n", eq->offset); |
| | 775 | |
| | 776 | for (i = arg_s = 0; i < g_func_pp.argc; i++) { |
| | 777 | if (g_func_pp.arg[i].reg != NULL) |
| | 778 | continue; |
| | 779 | if (arg_s == arg_i) |
| | 780 | break; |
| | 781 | arg_s++; |
| | 782 | } |
| | 783 | if (i == g_func_pp.argc) |
| | 784 | ferr(po, "arg %d not in prototype?\n", arg_i); |
| | 785 | if (is_lea) |
| | 786 | ferr(po, "lea to arg?\n"); |
| | 787 | |
| | 788 | snprintf(buf, buf_size, "%sa%d", is_src ? "(u32)" : "", i + 1); |
| | 789 | } |
| | 790 | else { |
| | 791 | if (g_bp_stack == 0) |
| | 792 | ferr(po, "bp_stack access after it was not detected\n"); |
| | 793 | |
| | 794 | sf_ofs = g_bp_stack + eq->offset; |
| | 795 | if (sf_ofs < 0) |
| | 796 | ferr(po, "bp_stack offset %d/%d\n", eq->offset, g_bp_stack); |
| | 797 | |
| | 798 | if (is_lea) |
| | 799 | prefix = "(u32)&"; |
| | 800 | |
| | 801 | switch (lmod) |
| | 802 | { |
| | 803 | case OPLM_BYTE: |
| | 804 | snprintf(buf, buf_size, "%ssf.b[%d]", prefix, sf_ofs); |
| | 805 | break; |
| | 806 | case OPLM_WORD: |
| | 807 | snprintf(buf, buf_size, "%ssf.w[%d]", prefix, sf_ofs / 2); |
| | 808 | break; |
| | 809 | case OPLM_DWORD: |
| | 810 | snprintf(buf, buf_size, "%ssf.d[%d]", prefix, sf_ofs / 4); |
| | 811 | break; |
| | 812 | default: |
| | 813 | ferr(po, "bp_stack bad lmod: %d\n", lmod); |
| | 814 | } |
| | 815 | } |
| | 816 | } |
| | 817 | |
| | 818 | static char *out_src_opr(char *buf, size_t buf_size, |
| | 819 | struct parsed_op *po, struct parsed_opr *popr, int is_lea) |
| | 820 | { |
| | 821 | const char *cast = ""; |
| | 822 | char tmp1[256], tmp2[256]; |
| | 823 | char expr[256]; |
| | 824 | int ret; |
| | 825 | |
| | 826 | switch (popr->type) { |
| | 827 | case OPT_REG: |
| | 828 | if (is_lea) |
| | 829 | ferr(po, "lea from reg?\n"); |
| | 830 | |
| | 831 | switch (popr->lmod) { |
| | 832 | case OPLM_DWORD: |
| | 833 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
| | 834 | break; |
| | 835 | case OPLM_WORD: |
| | 836 | snprintf(buf, buf_size, "(u16)%s", opr_reg_p(po, popr)); |
| | 837 | break; |
| | 838 | case OPLM_BYTE: |
| | 839 | if (popr->name[1] == 'h') // XXX.. |
| | 840 | snprintf(buf, buf_size, "(u8)(%s >> 8)", opr_reg_p(po, popr)); |
| | 841 | else |
| | 842 | snprintf(buf, buf_size, "(u8)%s", opr_reg_p(po, popr)); |
| | 843 | break; |
| | 844 | default: |
| | 845 | ferr(po, "invalid src lmod: %d\n", popr->lmod); |
| | 846 | } |
| | 847 | break; |
| | 848 | |
| | 849 | case OPT_REGMEM: |
| | 850 | if (g_bp_frame && !strncmp(popr->name, "ebp+", 4)) { |
| | 851 | bg_frame_access(po, popr->lmod, buf, buf_size, |
| | 852 | popr->name + 4, 1, is_lea); |
| | 853 | break; |
| | 854 | } |
| | 855 | |
| | 856 | strcpy(expr, popr->name); |
| | 857 | if (strchr(expr, '[')) { |
| | 858 | // special case: '[' can only be left for label[reg] form |
| | 859 | ret = sscanf(expr, "%[^[][%[^]]]", tmp1, tmp2); |
| | 860 | if (ret != 2) |
| | 861 | ferr(po, "parse failure for '%s'\n", expr); |
| | 862 | snprintf(expr, sizeof(expr), "(u32)&%s + %s", tmp1, tmp2); |
| | 863 | } |
| | 864 | |
| | 865 | // XXX: do we need more parsing? |
| | 866 | if (is_lea) { |
| | 867 | snprintf(buf, buf_size, "%s", expr); |
| | 868 | break; |
| | 869 | } |
| | 870 | |
| | 871 | switch (popr->lmod) { |
| | 872 | case OPLM_DWORD: |
| | 873 | cast = "*(u32 *)"; |
| | 874 | break; |
| | 875 | case OPLM_WORD: |
| | 876 | cast = "*(u16 *)"; |
| | 877 | break; |
| | 878 | case OPLM_BYTE: |
| | 879 | cast = "*(u8 *)"; |
| | 880 | break; |
| | 881 | default: |
| | 882 | ferr(po, "invalid lmod: %d\n", popr->lmod); |
| | 883 | } |
| | 884 | snprintf(buf, buf_size, "%s(%s)", cast, expr); |
| | 885 | break; |
| | 886 | |
| | 887 | case OPT_LABEL: |
| | 888 | if (is_lea) |
| | 889 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
| | 890 | else |
| | 891 | snprintf(buf, buf_size, "(u32)%s", popr->name); |
| | 892 | break; |
| | 893 | |
| | 894 | case OPT_OFFSET: |
| | 895 | if (is_lea) |
| | 896 | ferr(po, "lea an offset?\n"); |
| | 897 | snprintf(buf, buf_size, "(u32)&%s", popr->name); |
| | 898 | break; |
| | 899 | |
| | 900 | case OPT_CONST: |
| | 901 | if (is_lea) |
| | 902 | ferr(po, "lea from const?\n"); |
| | 903 | |
| | 904 | printf_number(buf, buf_size, popr->val); |
| | 905 | break; |
| | 906 | |
| | 907 | default: |
| | 908 | ferr(po, "invalid src type: %d\n", popr->type); |
| | 909 | } |
| | 910 | |
| | 911 | return buf; |
| | 912 | } |
| | 913 | |
| | 914 | static char *out_dst_opr(char *buf, size_t buf_size, |
| | 915 | struct parsed_op *po, struct parsed_opr *popr) |
| | 916 | { |
| | 917 | switch (popr->type) { |
| | 918 | case OPT_REG: |
| | 919 | switch (popr->lmod) { |
| | 920 | case OPLM_DWORD: |
| | 921 | snprintf(buf, buf_size, "%s", opr_reg_p(po, popr)); |
| | 922 | break; |
| | 923 | case OPLM_WORD: |
| | 924 | // ugh.. |
| | 925 | snprintf(buf, buf_size, "LOWORD(%s)", opr_reg_p(po, popr)); |
| | 926 | break; |
| | 927 | case OPLM_BYTE: |
| | 928 | // ugh.. |
| | 929 | if (popr->name[1] == 'h') // XXX.. |
| | 930 | snprintf(buf, buf_size, "BYTE1(%s)", opr_reg_p(po, popr)); |
| | 931 | else |
| | 932 | snprintf(buf, buf_size, "LOBYTE(%s)", opr_reg_p(po, popr)); |
| | 933 | break; |
| | 934 | default: |
| | 935 | ferr(po, "invalid dst lmod: %d\n", popr->lmod); |
| | 936 | } |
| | 937 | break; |
| | 938 | |
| | 939 | case OPT_REGMEM: |
| | 940 | if (g_bp_frame && !strncmp(popr->name, "ebp+", 4)) { |
| | 941 | bg_frame_access(po, popr->lmod, buf, buf_size, |
| | 942 | popr->name + 4, 0, 0); |
| | 943 | break; |
| | 944 | } |
| | 945 | |
| | 946 | return out_src_opr(buf, buf_size, po, popr, 0); |
| | 947 | |
| | 948 | default: |
| | 949 | ferr(po, "invalid dst type: %d\n", popr->type); |
| | 950 | } |
| | 951 | |
| | 952 | return buf; |
| | 953 | } |
| | 954 | |
| | 955 | static const char *lmod_cast_u(struct parsed_op *po, |
| | 956 | enum opr_lenmod lmod) |
| | 957 | { |
| | 958 | switch (lmod) { |
| | 959 | case OPLM_DWORD: |
| | 960 | return ""; |
| | 961 | case OPLM_WORD: |
| | 962 | return "(u16)"; |
| | 963 | case OPLM_BYTE: |
| | 964 | return "(u8)"; |
| | 965 | default: |
| | 966 | ferr(po, "invalid lmod: %d\n", lmod); |
| | 967 | return "(_invalid_)"; |
| | 968 | } |
| | 969 | } |
| | 970 | |
| | 971 | static const char *lmod_cast_s(struct parsed_op *po, |
| | 972 | enum opr_lenmod lmod) |
| | 973 | { |
| | 974 | switch (lmod) { |
| | 975 | case OPLM_DWORD: |
| | 976 | return "(s32)"; |
| | 977 | case OPLM_WORD: |
| | 978 | return "(s16)"; |
| | 979 | case OPLM_BYTE: |
| | 980 | return "(s8)"; |
| | 981 | default: |
| | 982 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
| | 983 | return "(_invalid_)"; |
| | 984 | } |
| | 985 | } |
| | 986 | |
| | 987 | static const char *lmod_cast(struct parsed_op *po, |
| | 988 | enum opr_lenmod lmod, int is_signed) |
| | 989 | { |
| | 990 | return is_signed ? |
| | 991 | lmod_cast_s(po, lmod) : |
| | 992 | lmod_cast_u(po, lmod); |
| | 993 | } |
| | 994 | |
| | 995 | static int lmod_bytes(struct parsed_op *po, enum opr_lenmod lmod) |
| | 996 | { |
| | 997 | switch (lmod) { |
| | 998 | case OPLM_DWORD: |
| | 999 | return 4; |
| | 1000 | case OPLM_WORD: |
| | 1001 | return 2; |
| | 1002 | case OPLM_BYTE: |
| | 1003 | return 1; |
| | 1004 | default: |
| | 1005 | ferr(po, "%s: invalid lmod: %d\n", __func__, lmod); |
| | 1006 | return 0; |
| | 1007 | } |
| | 1008 | } |
| | 1009 | |
| | 1010 | static enum parsed_flag_op split_cond(struct parsed_op *po, |
| | 1011 | enum op_op op, int *is_inv) |
| | 1012 | { |
| | 1013 | *is_inv = 0; |
| | 1014 | |
| | 1015 | switch (op) { |
| | 1016 | case OP_JO: |
| | 1017 | return PFO_O; |
| | 1018 | case OP_JC: |
| | 1019 | return PFO_C; |
| | 1020 | case OP_JZ: |
| | 1021 | return PFO_Z; |
| | 1022 | case OP_JBE: |
| | 1023 | return PFO_BE; |
| | 1024 | case OP_JS: |
| | 1025 | return PFO_S; |
| | 1026 | case OP_JP: |
| | 1027 | return PFO_P; |
| | 1028 | case OP_JL: |
| | 1029 | return PFO_L; |
| | 1030 | case OP_JLE: |
| | 1031 | return PFO_LE; |
| | 1032 | |
| | 1033 | case OP_JNO: |
| | 1034 | *is_inv = 1; |
| | 1035 | return PFO_O; |
| | 1036 | case OP_JNC: |
| | 1037 | *is_inv = 1; |
| | 1038 | return PFO_C; |
| | 1039 | case OP_JNZ: |
| | 1040 | *is_inv = 1; |
| | 1041 | return PFO_Z; |
| | 1042 | case OP_JA: |
| | 1043 | *is_inv = 1; |
| | 1044 | return PFO_BE; |
| | 1045 | case OP_JNS: |
| | 1046 | *is_inv = 1; |
| | 1047 | return PFO_S; |
| | 1048 | case OP_JNP: |
| | 1049 | *is_inv = 1; |
| | 1050 | return PFO_P; |
| | 1051 | case OP_JGE: |
| | 1052 | *is_inv = 1; |
| | 1053 | return PFO_L; |
| | 1054 | case OP_JG: |
| | 1055 | *is_inv = 1; |
| | 1056 | return PFO_LE; |
| | 1057 | |
| | 1058 | case OP_ADC: |
| | 1059 | case OP_SBB: |
| | 1060 | return PFO_C; |
| | 1061 | |
| | 1062 | default: |
| | 1063 | ferr(po, "split_cond: bad op %d\n", op); |
| | 1064 | return -1; |
| | 1065 | } |
| | 1066 | } |
| | 1067 | |
| | 1068 | static void out_test_for_cc(char *buf, size_t buf_size, |
| | 1069 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
| | 1070 | enum opr_lenmod lmod, const char *expr) |
| | 1071 | { |
| | 1072 | const char *cast, *scast; |
| | 1073 | |
| | 1074 | cast = lmod_cast_u(po, lmod); |
| | 1075 | scast = lmod_cast_s(po, lmod); |
| | 1076 | |
| | 1077 | switch (pfo) { |
| | 1078 | case PFO_Z: |
| | 1079 | case PFO_BE: // CF=1||ZF=1; CF=0 |
| | 1080 | snprintf(buf, buf_size, "(%s%s %s 0)", |
| | 1081 | cast, expr, is_inv ? "!=" : "=="); |
| | 1082 | break; |
| | 1083 | |
| | 1084 | case PFO_S: |
| | 1085 | case PFO_L: // SF!=OF; OF=0 |
| | 1086 | snprintf(buf, buf_size, "(%s%s %s 0)", |
| | 1087 | scast, expr, is_inv ? ">=" : "<"); |
| | 1088 | break; |
| | 1089 | |
| | 1090 | case PFO_LE: // ZF=1||SF!=OF; OF=0 |
| | 1091 | snprintf(buf, buf_size, "(%s%s %s 0)", |
| | 1092 | scast, expr, is_inv ? ">" : "<="); |
| | 1093 | break; |
| | 1094 | |
| | 1095 | default: |
| | 1096 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
| | 1097 | } |
| | 1098 | } |
| | 1099 | |
| | 1100 | static void out_cmp_for_cc(char *buf, size_t buf_size, |
| | 1101 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv, |
| | 1102 | enum opr_lenmod lmod, const char *expr1, const char *expr2) |
| | 1103 | { |
| | 1104 | const char *cast, *scast; |
| | 1105 | |
| | 1106 | cast = lmod_cast_u(po, lmod); |
| | 1107 | scast = lmod_cast_s(po, lmod); |
| | 1108 | |
| | 1109 | switch (pfo) { |
| | 1110 | case PFO_C: |
| | 1111 | // note: must be unsigned compare |
| | 1112 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
| | 1113 | cast, expr1, is_inv ? ">=" : "<", cast, expr2); |
| | 1114 | break; |
| | 1115 | |
| | 1116 | case PFO_Z: |
| | 1117 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
| | 1118 | cast, expr1, is_inv ? "!=" : "==", cast, expr2); |
| | 1119 | break; |
| | 1120 | |
| | 1121 | case PFO_BE: // !a |
| | 1122 | // note: must be unsigned compare |
| | 1123 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
| | 1124 | cast, expr1, is_inv ? ">" : "<=", cast, expr2); |
| | 1125 | break; |
| | 1126 | |
| | 1127 | // note: must be signed compare |
| | 1128 | case PFO_S: |
| | 1129 | snprintf(buf, buf_size, "(%s(%s - %s) %s 0)", |
| | 1130 | scast, expr1, expr2, is_inv ? ">=" : "<"); |
| | 1131 | break; |
| | 1132 | |
| | 1133 | case PFO_L: // !ge |
| | 1134 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
| | 1135 | scast, expr1, is_inv ? ">=" : "<", scast, expr2); |
| | 1136 | break; |
| | 1137 | |
| | 1138 | case PFO_LE: |
| | 1139 | snprintf(buf, buf_size, "(%s%s %s %s%s)", |
| | 1140 | scast, expr1, is_inv ? ">" : "<=", scast, expr2); |
| | 1141 | break; |
| | 1142 | |
| | 1143 | default: |
| | 1144 | ferr(po, "%s: unhandled parsed_flag_op: %d\n", __func__, pfo); |
| | 1145 | } |
| | 1146 | } |
| | 1147 | |
| | 1148 | static void out_cmp_test(char *buf, size_t buf_size, |
| | 1149 | struct parsed_op *po, enum parsed_flag_op pfo, int is_inv) |
| | 1150 | { |
| | 1151 | char buf1[256], buf2[256], buf3[256]; |
| | 1152 | |
| | 1153 | if (po->op == OP_TEST) { |
| | 1154 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
| | 1155 | out_src_opr(buf3, sizeof(buf3), po, &po->operand[0], 0); |
| | 1156 | } |
| | 1157 | else { |
| | 1158 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
| | 1159 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0); |
| | 1160 | snprintf(buf3, sizeof(buf3), "(%s & %s)", buf1, buf2); |
| | 1161 | } |
| | 1162 | out_test_for_cc(buf, buf_size, po, pfo, is_inv, |
| | 1163 | po->operand[0].lmod, buf3); |
| | 1164 | } |
| | 1165 | else if (po->op == OP_CMP) { |
| | 1166 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0); |
| | 1167 | out_src_opr(buf3, sizeof(buf3), po, &po->operand[1], 0); |
| | 1168 | out_cmp_for_cc(buf, buf_size, po, pfo, is_inv, |
| | 1169 | po->operand[0].lmod, buf2, buf3); |
| | 1170 | } |
| | 1171 | else |
| | 1172 | ferr(po, "%s: unhandled op: %d\n", __func__, po->op); |
| | 1173 | } |
| | 1174 | |
| | 1175 | static void propagate_lmod(struct parsed_op *po, struct parsed_opr *popr1, |
| | 1176 | struct parsed_opr *popr2) |
| | 1177 | { |
| | 1178 | if (popr1->lmod == OPLM_UNSPEC && popr2->lmod == OPLM_UNSPEC) |
| | 1179 | ferr(po, "missing lmod for both operands\n"); |
| | 1180 | |
| | 1181 | if (popr1->lmod == OPLM_UNSPEC) |
| | 1182 | popr1->lmod = popr2->lmod; |
| | 1183 | else if (popr2->lmod == OPLM_UNSPEC) |
| | 1184 | popr2->lmod = popr1->lmod; |
| | 1185 | else if (popr1->lmod != popr2->lmod) |
| | 1186 | ferr(po, "conflicting lmods: %d vs %d\n", popr1->lmod, popr2->lmod); |
| | 1187 | } |
| | 1188 | |
| | 1189 | static const char *op_to_c(struct parsed_op *po) |
| | 1190 | { |
| | 1191 | switch (po->op) |
| | 1192 | { |
| | 1193 | case OP_ADD: |
| | 1194 | case OP_ADC: |
| | 1195 | return "+"; |
| | 1196 | case OP_SUB: |
| | 1197 | case OP_SBB: |
| | 1198 | return "-"; |
| | 1199 | case OP_AND: |
| | 1200 | return "&"; |
| | 1201 | case OP_OR: |
| | 1202 | return "|"; |
| | 1203 | case OP_XOR: |
| | 1204 | return "^"; |
| | 1205 | case OP_SHL: |
| | 1206 | return "<<"; |
| | 1207 | case OP_SHR: |
| | 1208 | return ">>"; |
| | 1209 | case OP_MUL: |
| | 1210 | case OP_IMUL: |
| | 1211 | return "*"; |
| | 1212 | default: |
| | 1213 | ferr(po, "op_to_c was supplied with %d\n", po->op); |
| | 1214 | } |
| | 1215 | } |
| | 1216 | |
| | 1217 | static int scan_for_pop(int i, int opcnt, const char *reg, |
| | 1218 | int magic, int do_patch) |
| | 1219 | { |
| | 1220 | struct parsed_op *po; |
| | 1221 | int ret = 0; |
| | 1222 | |
| | 1223 | for (; i < opcnt; i++) { |
| | 1224 | po = &ops[i]; |
| | 1225 | if (po->cc_scratch == magic) |
| | 1226 | break; // already checked |
| | 1227 | po->cc_scratch = magic; |
| | 1228 | |
| | 1229 | if (po->flags & OPF_TAIL) |
| | 1230 | return -1; // deadend |
| | 1231 | |
| | 1232 | if (po->flags & OPF_RMD) |
| | 1233 | continue; |
| | 1234 | |
| | 1235 | if ((po->flags & OPF_JMP) && po->op != OP_CALL) { |
| | 1236 | if (po->bt_i < 0) { |
| | 1237 | ferr(po, "dead branch\n"); |
| | 1238 | return -1; |
| | 1239 | } |
| | 1240 | |
| | 1241 | if (po->flags & OPF_CC) { |
| | 1242 | ret |= scan_for_pop(po->bt_i, opcnt, reg, magic, do_patch); |
| | 1243 | if (ret < 0) |
| | 1244 | return ret; // dead end |
| | 1245 | } |
| | 1246 | else { |
| | 1247 | i = po->bt_i - 1; |
| | 1248 | } |
| | 1249 | continue; |
| | 1250 | } |
| | 1251 | |
| | 1252 | if (po->op == OP_POP && po->operand[0].type == OPT_REG |
| | 1253 | && IS(po->operand[0].name, reg)) |
| | 1254 | { |
| | 1255 | if (do_patch) |
| | 1256 | po->flags |= OPF_RMD; |
| | 1257 | return 1; |
| | 1258 | } |
| | 1259 | } |
| | 1260 | |
| | 1261 | return ret; |
| | 1262 | } |
| | 1263 | |
| | 1264 | // scan for pop starting from 'ret' op (all paths) |
| | 1265 | static int scan_for_pop_ret(int i, int opcnt, const char *reg, int do_patch) |
| | 1266 | { |
| | 1267 | int found = 0; |
| | 1268 | int j; |
| | 1269 | |
| | 1270 | for (; i < opcnt; i++) { |
| | 1271 | if (!(ops[i].flags & OPF_TAIL)) |
| | 1272 | continue; |
| | 1273 | |
| | 1274 | for (j = i - 1; j >= 0; j--) { |
| | 1275 | if (ops[j].flags & OPF_RMD) |
| | 1276 | continue; |
| | 1277 | if (ops[j].flags & OPF_JMP) |
| | 1278 | return -1; |
| | 1279 | |
| | 1280 | if (ops[j].op == OP_POP && ops[j].operand[0].type == OPT_REG |
| | 1281 | && IS(ops[j].operand[0].name, reg)) |
| | 1282 | { |
| | 1283 | found = 1; |
| | 1284 | if (do_patch) |
| | 1285 | ops[j].flags |= OPF_RMD; |
| | 1286 | break; |
| | 1287 | } |
| | 1288 | |
| | 1289 | if (g_labels[j][0] != 0) |
| | 1290 | return -1; |
| | 1291 | } |
| | 1292 | } |
| | 1293 | |
| | 1294 | return found ? 0 : -1; |
| | 1295 | } |
| | 1296 | |
| | 1297 | // is operand 'opr modified' by parsed_op 'po'? |
| | 1298 | static int is_opr_modified(const struct parsed_opr *opr, |
| | 1299 | const struct parsed_op *po) |
| | 1300 | { |
| | 1301 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
| | 1302 | return 0; |
| | 1303 | |
| | 1304 | if (opr->type == OPT_REG && po->operand[0].type == OPT_REG) { |
| | 1305 | if (po->regmask_dst & (1 << opr->reg)) |
| | 1306 | return 1; |
| | 1307 | else |
| | 1308 | return 0; |
| | 1309 | } |
| | 1310 | |
| | 1311 | return IS(po->operand[0].name, opr->name); |
| | 1312 | } |
| | 1313 | |
| | 1314 | // is any operand of parsed_op 'po_test' modified by parsed_op 'po'? |
| | 1315 | static int is_any_opr_modified(const struct parsed_op *po_test, |
| | 1316 | const struct parsed_op *po) |
| | 1317 | { |
| | 1318 | int i; |
| | 1319 | |
| | 1320 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_DATA)) |
| | 1321 | return 0; |
| | 1322 | |
| | 1323 | if (po_test->regmask_src & po->regmask_dst) |
| | 1324 | return 1; |
| | 1325 | |
| | 1326 | for (i = 0; i < po_test->operand_cnt; i++) |
| | 1327 | if (IS(po_test->operand[i].name, po->operand[0].name)) |
| | 1328 | return 1; |
| | 1329 | |
| | 1330 | return 0; |
| | 1331 | } |
| | 1332 | |
| | 1333 | // scan for any po_test operand modification in range given |
| | 1334 | static int scan_for_mod(struct parsed_op *po_test, int i, int opcnt) |
| | 1335 | { |
| | 1336 | for (; i < opcnt; i++) { |
| | 1337 | if (is_any_opr_modified(po_test, &ops[i])) |
| | 1338 | return i; |
| | 1339 | } |
| | 1340 | |
| | 1341 | return -1; |
| | 1342 | } |
| | 1343 | |
| | 1344 | // scan for po_test operand[0] modification in range given |
| | 1345 | static int scan_for_mod_opr0(struct parsed_op *po_test, |
| | 1346 | int i, int opcnt) |
| | 1347 | { |
| | 1348 | for (; i < opcnt; i++) { |
| | 1349 | if (is_opr_modified(&po_test->operand[0], &ops[i])) |
| | 1350 | return i; |
| | 1351 | } |
| | 1352 | |
| | 1353 | return -1; |
| | 1354 | } |
| | 1355 | |
| | 1356 | static int scan_for_flag_set(int i) |
| | 1357 | { |
| | 1358 | for (; i >= 0; i--) { |
| | 1359 | if (ops[i].flags & OPF_FLAGS) |
| | 1360 | return i; |
| | 1361 | |
| | 1362 | if ((ops[i].flags & OPF_JMP) && !(ops[i].flags & OPF_CC)) |
| | 1363 | return -1; |
| | 1364 | if (g_labels[i][0] != 0) |
| | 1365 | return -1; |
| | 1366 | } |
| | 1367 | |
| | 1368 | return -1; |
| | 1369 | } |
| | 1370 | |
| | 1371 | // scan back for cdq, if anything modifies edx, fail |
| | 1372 | static int scan_for_cdq_edx(int i) |
| | 1373 | { |
| | 1374 | for (; i >= 0; i--) { |
| | 1375 | if (ops[i].op == OP_CDQ) |
| | 1376 | return i; |
| | 1377 | |
| | 1378 | if (ops[i].regmask_dst & (1 << xDX)) |
| | 1379 | return -1; |
| | 1380 | if (g_labels[i][0] != 0) |
| | 1381 | return -1; |
| | 1382 | } |
| | 1383 | |
| | 1384 | return -1; |
| | 1385 | } |
| | 1386 | |
| | 1387 | static void gen_func(FILE *fout, FILE *fhdr, const char *funcn, int opcnt) |
| | 1388 | { |
| | 1389 | struct parsed_op *po, *delayed_flag_op = NULL, *tmp_op; |
| | 1390 | struct parsed_opr *last_arith_dst = NULL; |
| | 1391 | char buf1[256], buf2[256], buf3[256]; |
| | 1392 | struct parsed_proto *pp, *pp_tmp; |
| | 1393 | const char *tmpname; |
| | 1394 | enum parsed_flag_op pfo; |
| | 1395 | int save_arg_vars = 0; |
| | 1396 | int cmp_result_vars = 0; |
| | 1397 | int need_mul_var = 0; |
| | 1398 | int had_decl = 0; |
| | 1399 | int regmask_arg = 0; |
| | 1400 | int regmask = 0; |
| | 1401 | int pfomask = 0; |
| | 1402 | int no_output; |
| | 1403 | int dummy; |
| | 1404 | int arg; |
| | 1405 | int i, j; |
| | 1406 | int reg; |
| | 1407 | int ret; |
| | 1408 | |
| | 1409 | g_bp_frame = g_bp_stack = 0; |
| | 1410 | |
| | 1411 | ret = proto_parse(fhdr, funcn, &g_func_pp); |
| | 1412 | if (ret) |
| | 1413 | ferr(ops, "proto_parse failed for '%s'\n", funcn); |
| | 1414 | |
| | 1415 | fprintf(fout, "%s %s(", g_func_pp.ret_type, funcn); |
| | 1416 | for (i = 0; i < g_func_pp.argc; i++) { |
| | 1417 | if (i > 0) |
| | 1418 | fprintf(fout, ", "); |
| | 1419 | fprintf(fout, "%s a%d", g_func_pp.arg[i].type, i + 1); |
| | 1420 | } |
| | 1421 | fprintf(fout, ")\n{\n"); |
| | 1422 | |
| | 1423 | // pass1: |
| | 1424 | // - handle ebp frame, remove ops related to it |
| | 1425 | if (ops[0].op == OP_PUSH && IS(opr_name(&ops[0], 0), "ebp") |
| | 1426 | && ops[1].op == OP_MOV |
| | 1427 | && IS(opr_name(&ops[1], 0), "ebp") |
| | 1428 | && IS(opr_name(&ops[1], 1), "esp")) |
| | 1429 | { |
| | 1430 | int ecx_push = 0; |
| | 1431 | |
| | 1432 | g_bp_frame = 1; |
| | 1433 | ops[0].flags |= OPF_RMD; |
| | 1434 | ops[1].flags |= OPF_RMD; |
| | 1435 | |
| | 1436 | if (ops[2].op == OP_SUB && IS(opr_name(&ops[2], 0), "esp")) { |
| | 1437 | g_bp_stack = opr_const(&ops[2], 1); |
| | 1438 | ops[2].flags |= OPF_RMD; |
| | 1439 | } |
| | 1440 | else { |
| | 1441 | // another way msvc builds stack frame.. |
| | 1442 | i = 2; |
| | 1443 | while (ops[i].op == OP_PUSH && IS(opr_name(&ops[i], 0), "ecx")) { |
| | 1444 | g_bp_stack += 4; |
| | 1445 | ops[i].flags |= OPF_RMD; |
| | 1446 | ecx_push++; |
| | 1447 | i++; |
| | 1448 | } |
| | 1449 | } |
| | 1450 | |
| | 1451 | i = 2; |
| | 1452 | do { |
| | 1453 | for (; i < opcnt; i++) |
| | 1454 | if (ops[i].op == OP_RET) |
| | 1455 | break; |
| | 1456 | if (ops[i - 1].op != OP_POP || !IS(opr_name(&ops[i - 1], 0), "ebp")) |
| | 1457 | ferr(&ops[i - 1], "'pop ebp' expected\n"); |
| | 1458 | ops[i - 1].flags |= OPF_RMD; |
| | 1459 | |
| | 1460 | if (g_bp_stack != 0) { |
| | 1461 | if (ops[i - 2].op != OP_MOV |
| | 1462 | || !IS(opr_name(&ops[i - 2], 0), "esp") |
| | 1463 | || !IS(opr_name(&ops[i - 2], 1), "ebp")) |
| | 1464 | { |
| | 1465 | ferr(&ops[i - 2], "esp restore expected\n"); |
| | 1466 | } |
| | 1467 | ops[i - 2].flags |= OPF_RMD; |
| | 1468 | |
| | 1469 | if (ecx_push && ops[i - 3].op == OP_POP |
| | 1470 | && IS(opr_name(&ops[i - 3], 0), "ecx")) |
| | 1471 | { |
| | 1472 | ferr(&ops[i - 3], "unexpected ecx pop\n"); |
| | 1473 | } |
| | 1474 | } |
| | 1475 | |
| | 1476 | i++; |
| | 1477 | } while (i < opcnt); |
| | 1478 | } |
| | 1479 | |
| | 1480 | // pass2: |
| | 1481 | // - resolve all branches |
| | 1482 | for (i = 0; i < opcnt; i++) { |
| | 1483 | po = &ops[i]; |
| | 1484 | po->bt_i = -1; |
| | 1485 | |
| | 1486 | if ((po->flags & OPF_RMD) || !(po->flags & OPF_JMP) |
| | 1487 | || po->op == OP_CALL || po->op == OP_RET) |
| | 1488 | continue; |
| | 1489 | |
| | 1490 | for (j = 0; j < opcnt; j++) { |
| | 1491 | if (g_labels[j][0] && IS(po->operand[0].name, g_labels[j])) { |
| | 1492 | po->bt_i = j; |
| | 1493 | po->lrl = g_label_refs[j]; |
| | 1494 | g_label_refs[j] = po; |
| | 1495 | break; |
| | 1496 | } |
| | 1497 | } |
| | 1498 | |
| | 1499 | if (po->bt_i == -1) { |
| | 1500 | // assume tail call |
| | 1501 | po->op = OP_CALL; |
| | 1502 | po->flags |= OPF_TAIL; |
| | 1503 | } |
| | 1504 | } |
| | 1505 | |
| | 1506 | // pass3: |
| | 1507 | // - find POPs for PUSHes, rm both |
| | 1508 | // - scan for all used registers |
| | 1509 | // - find flag set ops for their users |
| | 1510 | // - process calls |
| | 1511 | for (i = 0; i < opcnt; i++) { |
| | 1512 | po = &ops[i]; |
| | 1513 | if (po->flags & OPF_RMD) |
| | 1514 | continue; |
| | 1515 | |
| | 1516 | if (po->op == OP_PUSH && po->operand[0].type == OPT_REG) { |
| | 1517 | if (po->operand[0].reg < 0) |
| | 1518 | ferr(po, "reg not set for push?\n"); |
| | 1519 | if (!(regmask & (1 << po->operand[0].reg))) { // reg save |
| | 1520 | ret = scan_for_pop(i + 1, opcnt, |
| | 1521 | po->operand[0].name, i + opcnt, 0); |
| | 1522 | if (ret == 1) { |
| | 1523 | po->flags |= OPF_RMD; |
| | 1524 | scan_for_pop(i + 1, opcnt, po->operand[0].name, |
| | 1525 | i + opcnt * 2, 1); |
| | 1526 | continue; |
| | 1527 | } |
| | 1528 | ret = scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 0); |
| | 1529 | if (ret == 0) { |
| | 1530 | po->flags |= OPF_RMD; |
| | 1531 | scan_for_pop_ret(i + 1, opcnt, po->operand[0].name, 1); |
| | 1532 | continue; |
| | 1533 | } |
| | 1534 | } |
| | 1535 | } |
| | 1536 | |
| | 1537 | regmask |= po->regmask_src | po->regmask_dst; |
| | 1538 | |
| | 1539 | if (po->flags & OPF_CC) |
| | 1540 | { |
| | 1541 | ret = scan_for_flag_set(i - 1); |
| | 1542 | if (ret < 0) |
| | 1543 | ferr(po, "unable to trace flag setter\n"); |
| | 1544 | |
| | 1545 | tmp_op = &ops[ret]; // flag setter |
| | 1546 | pfo = split_cond(po, po->op, &dummy); |
| | 1547 | pfomask = 0; |
| | 1548 | |
| | 1549 | // to get nicer code, we try to delay test and cmp; |
| | 1550 | // if we can't because of operand modification, or if we |
| | 1551 | // have math op, make it calculate flags explicitly |
| | 1552 | if (tmp_op->op == OP_TEST || tmp_op->op == OP_CMP) { |
| | 1553 | if (scan_for_mod(tmp_op, ret + 1, i) >= 0) |
| | 1554 | pfomask = 1 << pfo; |
| | 1555 | } |
| | 1556 | else { |
| | 1557 | if ((pfo != PFO_Z && pfo != PFO_S && pfo != PFO_P) |
| | 1558 | || scan_for_mod_opr0(tmp_op, ret + 1, i) >= 0) |
| | 1559 | pfomask = 1 << pfo; |
| | 1560 | } |
| | 1561 | if (pfomask) { |
| | 1562 | tmp_op->pfomask |= pfomask; |
| | 1563 | cmp_result_vars |= pfomask; |
| | 1564 | po->datap = tmp_op; |
| | 1565 | } |
| | 1566 | |
| | 1567 | if (po->op == OP_ADC || po->op == OP_SBB) |
| | 1568 | cmp_result_vars |= 1 << PFO_C; |
| | 1569 | } |
| | 1570 | else if (po->op == OP_CALL) |
| | 1571 | { |
| | 1572 | pp = malloc(sizeof(*pp)); |
| | 1573 | my_assert_not(pp, NULL); |
| | 1574 | tmpname = opr_name(&ops[i], 0); |
| | 1575 | ret = proto_parse(fhdr, tmpname, pp); |
| | 1576 | if (ret) |
| | 1577 | ferr(po, "proto_parse failed for '%s'\n", tmpname); |
| | 1578 | |
| | 1579 | for (arg = 0; arg < pp->argc; arg++) |
| | 1580 | if (pp->arg[arg].reg == NULL) |
| | 1581 | break; |
| | 1582 | |
| | 1583 | for (j = i - 1; j >= 0 && arg < pp->argc; j--) { |
| | 1584 | if (ops[j].op == OP_CALL) { |
| | 1585 | pp_tmp = ops[j].datap; |
| | 1586 | if (pp_tmp == NULL) |
| | 1587 | ferr(po, "arg collect hit unparsed call\n"); |
| | 1588 | if (pp_tmp->argc_stack > 0) |
| | 1589 | ferr(po, "arg collect hit '%s' with %d stack args\n", |
| | 1590 | opr_name(&ops[j], 0), pp_tmp->argc_stack); |
| | 1591 | } |
| | 1592 | else if ((ops[j].flags & OPF_TAIL) |
| | 1593 | || (ops[j].flags & (OPF_JMP|OPF_CC)) == OPF_JMP) |
| | 1594 | { |
| | 1595 | break; |
| | 1596 | } |
| | 1597 | |
| | 1598 | if (ops[j].op == OP_PUSH) { |
| | 1599 | pp->arg[arg].datap = &ops[j]; |
| | 1600 | ret = scan_for_mod(&ops[j], j + 1, i); |
| | 1601 | if (ret >= 0) { |
| | 1602 | // mark this push as one that needs operand saving |
| | 1603 | ops[j].argmask |= 1 << arg; |
| | 1604 | save_arg_vars |= 1 << arg; |
| | 1605 | } |
| | 1606 | else |
| | 1607 | ops[j].flags |= OPF_RMD; |
| | 1608 | |
| | 1609 | // next arg |
| | 1610 | for (arg++; arg < pp->argc; arg++) |
| | 1611 | if (pp->arg[arg].reg == NULL) |
| | 1612 | break; |
| | 1613 | } |
| | 1614 | |
| | 1615 | if (g_labels[j][0] != 0) { |
| | 1616 | if (j > 0 && ((ops[j - 1].flags & OPF_TAIL) |
| | 1617 | || (ops[j - 1].flags & (OPF_JMP|OPF_CC)) == OPF_JMP)) |
| | 1618 | { |
| | 1619 | // follow the branch in reverse |
| | 1620 | if (g_label_refs[j] == NULL) |
| | 1621 | ferr(po, "no refs for '%s'?\n", g_labels[j]); |
| | 1622 | if (g_label_refs[j]->lrl != NULL) |
| | 1623 | ferr(po, "unhandled multiple fefs to '%s'\n", g_labels[j]); |
| | 1624 | j = (g_label_refs[j] - ops) + 1; |
| | 1625 | continue; |
| | 1626 | } |
| | 1627 | break; |
| | 1628 | } |
| | 1629 | } |
| | 1630 | if (arg < pp->argc) |
| | 1631 | ferr(po, "arg collect failed for '%s'\n", tmpname); |
| | 1632 | po->datap = pp; |
| | 1633 | } |
| | 1634 | else if (po->op == OP_MUL |
| | 1635 | || (po->op == OP_IMUL && po->operand_cnt == 1)) |
| | 1636 | { |
| | 1637 | need_mul_var = 1; |
| | 1638 | } |
| | 1639 | } |
| | 1640 | |
| | 1641 | // declare stack frame |
| | 1642 | if (g_bp_stack) |
| | 1643 | fprintf(fout, " union { u32 d[%d]; u16 w[%d]; u8 b[%d]; } sf;\n", |
| | 1644 | (g_bp_stack + 3) / 4, (g_bp_stack + 1) / 2, g_bp_stack); |
| | 1645 | |
| | 1646 | // declare arg-registers |
| | 1647 | for (i = 0; i < g_func_pp.argc; i++) { |
| | 1648 | if (g_func_pp.arg[i].reg != NULL) { |
| | 1649 | reg = char_array_i(regs_r32, |
| | 1650 | ARRAY_SIZE(regs_r32), g_func_pp.arg[i].reg); |
| | 1651 | if (reg < 0) |
| | 1652 | ferr(ops, "arg '%s' is not a reg?\n", g_func_pp.arg[i].reg); |
| | 1653 | |
| | 1654 | regmask_arg |= 1 << reg; |
| | 1655 | fprintf(fout, " u32 %s = (u32)a%d;\n", |
| | 1656 | g_func_pp.arg[i].reg, i + 1); |
| | 1657 | had_decl = 1; |
| | 1658 | } |
| | 1659 | } |
| | 1660 | |
| | 1661 | // declare other regs - special case for eax |
| | 1662 | if (!((regmask | regmask_arg) & 1) && !IS(g_func_pp.ret_type, "void")) { |
| | 1663 | fprintf(fout, " u32 eax = 0;\n"); |
| | 1664 | had_decl = 1; |
| | 1665 | } |
| | 1666 | |
| | 1667 | regmask &= ~regmask_arg; |
| | 1668 | if (g_bp_frame) |
| | 1669 | regmask &= ~(1 << xBP); |
| | 1670 | if (regmask) { |
| | 1671 | for (reg = 0; reg < 8; reg++) { |
| | 1672 | if (regmask & (1 << reg)) { |
| | 1673 | fprintf(fout, " u32 %s;\n", regs_r32[reg]); |
| | 1674 | had_decl = 1; |
| | 1675 | } |
| | 1676 | } |
| | 1677 | } |
| | 1678 | |
| | 1679 | if (save_arg_vars) { |
| | 1680 | for (reg = 0; reg < 32; reg++) { |
| | 1681 | if (save_arg_vars & (1 << reg)) { |
| | 1682 | fprintf(fout, " u32 s_a%d;\n", reg + 1); |
| | 1683 | had_decl = 1; |
| | 1684 | } |
| | 1685 | } |
| | 1686 | } |
| | 1687 | |
| | 1688 | if (cmp_result_vars) { |
| | 1689 | for (i = 0; i < 8; i++) { |
| | 1690 | if (cmp_result_vars & (1 << i)) { |
| | 1691 | fprintf(fout, " u32 cond_%s;\n", parsed_flag_op_names[i]); |
| | 1692 | had_decl = 1; |
| | 1693 | } |
| | 1694 | } |
| | 1695 | } |
| | 1696 | |
| | 1697 | if (need_mul_var) { |
| | 1698 | fprintf(fout, " u64 mul_tmp;\n"); |
| | 1699 | had_decl = 1; |
| | 1700 | } |
| | 1701 | |
| | 1702 | if (had_decl) |
| | 1703 | fprintf(fout, "\n"); |
| | 1704 | |
| | 1705 | // output ops |
| | 1706 | for (i = 0; i < opcnt; i++) |
| | 1707 | { |
| | 1708 | if (g_labels[i][0] != 0) |
| | 1709 | fprintf(fout, "\n%s:\n", g_labels[i]); |
| | 1710 | |
| | 1711 | po = &ops[i]; |
| | 1712 | if (po->flags & OPF_RMD) |
| | 1713 | continue; |
| | 1714 | |
| | 1715 | no_output = 0; |
| | 1716 | |
| | 1717 | #define assert_operand_cnt(n_) \ |
| | 1718 | if (po->operand_cnt != n_) \ |
| | 1719 | ferr(po, "operand_cnt is %d/%d\n", po->operand_cnt, n_) |
| | 1720 | |
| | 1721 | // conditional/flag using op? |
| | 1722 | if (po->flags & OPF_CC) |
| | 1723 | { |
| | 1724 | int is_delayed = 0; |
| | 1725 | int is_inv = 0; |
| | 1726 | |
| | 1727 | pfo = split_cond(po, po->op, &is_inv); |
| | 1728 | |
| | 1729 | // we go through all this trouble to avoid using parsed_flag_op, |
| | 1730 | // which makes generated code much nicer |
| | 1731 | if (delayed_flag_op != NULL) |
| | 1732 | { |
| | 1733 | out_cmp_test(buf1, sizeof(buf1), delayed_flag_op, pfo, is_inv); |
| | 1734 | is_delayed = 1; |
| | 1735 | } |
| | 1736 | else if (last_arith_dst != NULL |
| | 1737 | && (pfo == PFO_Z || pfo == PFO_S || pfo == PFO_P)) |
| | 1738 | { |
| | 1739 | out_src_opr(buf3, sizeof(buf3), po, last_arith_dst, 0); |
| | 1740 | out_test_for_cc(buf1, sizeof(buf1), po, pfo, is_inv, |
| | 1741 | last_arith_dst->lmod, buf3); |
| | 1742 | is_delayed = 1; |
| | 1743 | } |
| | 1744 | else if (po->datap != NULL) { |
| | 1745 | // use preprocessed results |
| | 1746 | tmp_op = po->datap; |
| | 1747 | if (!tmp_op || !(tmp_op->pfomask & (1 << pfo))) |
| | 1748 | ferr(po, "not prepared for pfo %d\n", pfo); |
| | 1749 | |
| | 1750 | // note: is_inv was not yet applied |
| | 1751 | snprintf(buf1, sizeof(buf1), "(%scond_%s)", |
| | 1752 | is_inv ? "!" : "", parsed_flag_op_names[pfo]); |
| | 1753 | } |
| | 1754 | else { |
| | 1755 | ferr(po, "all methods of finding comparison failed\n"); |
| | 1756 | } |
| | 1757 | |
| | 1758 | if (po->flags & OPF_JMP) { |
| | 1759 | fprintf(fout, " if %s\n", buf1); |
| | 1760 | } |
| | 1761 | else if (po->op == OP_ADC || po->op == OP_SBB) { |
| | 1762 | if (is_delayed) |
| | 1763 | fprintf(fout, " cond_%s = %s;\n", |
| | 1764 | parsed_flag_op_names[pfo], buf1); |
| | 1765 | } |
| | 1766 | else if (po->flags & OPF_DATA) { // SETcc |
| | 1767 | out_dst_opr(buf2, sizeof(buf2), po, &po->operand[0]); |
| | 1768 | fprintf(fout, " %s = %s;", buf2, buf1); |
| | 1769 | } |
| | 1770 | else { |
| | 1771 | ferr(po, "unhandled conditional op\n"); |
| | 1772 | } |
| | 1773 | } |
| | 1774 | |
| | 1775 | pfomask = po->pfomask; |
| | 1776 | |
| | 1777 | switch (po->op) |
| | 1778 | { |
| | 1779 | case OP_MOV: |
| | 1780 | assert_operand_cnt(2); |
| | 1781 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
| | 1782 | fprintf(fout, " %s = %s;", |
| | 1783 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
| | 1784 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
| | 1785 | break; |
| | 1786 | |
| | 1787 | case OP_LEA: |
| | 1788 | assert_operand_cnt(2); |
| | 1789 | po->operand[1].lmod = OPLM_DWORD; // always |
| | 1790 | fprintf(fout, " %s = %s;", |
| | 1791 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
| | 1792 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 1)); |
| | 1793 | break; |
| | 1794 | |
| | 1795 | case OP_MOVZX: |
| | 1796 | assert_operand_cnt(2); |
| | 1797 | fprintf(fout, " %s = %s;", |
| | 1798 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
| | 1799 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
| | 1800 | break; |
| | 1801 | |
| | 1802 | case OP_MOVSX: |
| | 1803 | assert_operand_cnt(2); |
| | 1804 | switch (po->operand[1].lmod) { |
| | 1805 | case OPLM_BYTE: |
| | 1806 | strcpy(buf3, "(s8)"); |
| | 1807 | break; |
| | 1808 | case OPLM_WORD: |
| | 1809 | strcpy(buf3, "(s16)"); |
| | 1810 | break; |
| | 1811 | default: |
| | 1812 | ferr(po, "invalid src lmod: %d\n", po->operand[1].lmod); |
| | 1813 | } |
| | 1814 | fprintf(fout, " %s = %s%s;", |
| | 1815 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
| | 1816 | buf3, |
| | 1817 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
| | 1818 | break; |
| | 1819 | |
| | 1820 | case OP_NOT: |
| | 1821 | assert_operand_cnt(1); |
| | 1822 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
| | 1823 | fprintf(fout, " %s = ~%s;", buf1, buf1); |
| | 1824 | break; |
| | 1825 | |
| | 1826 | case OP_CDQ: |
| | 1827 | assert_operand_cnt(2); |
| | 1828 | fprintf(fout, " %s = (s32)%s >> 31;", |
| | 1829 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
| | 1830 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
| | 1831 | strcpy(g_comment, "cdq"); |
| | 1832 | break; |
| | 1833 | |
| | 1834 | case OP_STOS: |
| | 1835 | // assumes DF=0 |
| | 1836 | assert_operand_cnt(3); |
| | 1837 | if (po->flags & OPF_REP) { |
| | 1838 | fprintf(fout, " for (; ecx != 0; ecx--, edi += %d)\n", |
| | 1839 | lmod_bytes(po, po->operand[0].lmod)); |
| | 1840 | fprintf(fout, " *(u32 *)edi = eax;"); |
| | 1841 | strcpy(g_comment, "rep stos"); |
| | 1842 | } |
| | 1843 | else { |
| | 1844 | fprintf(fout, " *(u32 *)edi = eax; edi += %d;", |
| | 1845 | lmod_bytes(po, po->operand[0].lmod)); |
| | 1846 | strcpy(g_comment, "stos"); |
| | 1847 | } |
| | 1848 | break; |
| | 1849 | |
| | 1850 | // arithmetic w/flags |
| | 1851 | case OP_ADD: |
| | 1852 | case OP_SUB: |
| | 1853 | case OP_AND: |
| | 1854 | case OP_OR: |
| | 1855 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
| | 1856 | // fallthrough |
| | 1857 | case OP_SHL: |
| | 1858 | case OP_SHR: |
| | 1859 | dualop_arith: |
| | 1860 | assert_operand_cnt(2); |
| | 1861 | fprintf(fout, " %s %s= %s;", |
| | 1862 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
| | 1863 | op_to_c(po), |
| | 1864 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
| | 1865 | last_arith_dst = &po->operand[0]; |
| | 1866 | delayed_flag_op = NULL; |
| | 1867 | break; |
| | 1868 | |
| | 1869 | case OP_XOR: |
| | 1870 | assert_operand_cnt(2); |
| | 1871 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
| | 1872 | if (IS(opr_name(po, 0), opr_name(po, 1))) { |
| | 1873 | // special case for XOR |
| | 1874 | fprintf(fout, " %s = 0;", |
| | 1875 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0])); |
| | 1876 | last_arith_dst = &po->operand[0]; |
| | 1877 | delayed_flag_op = NULL; |
| | 1878 | break; |
| | 1879 | } |
| | 1880 | goto dualop_arith; |
| | 1881 | |
| | 1882 | case OP_SAR: |
| | 1883 | assert_operand_cnt(2); |
| | 1884 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
| | 1885 | fprintf(fout, " %s = %s%s >> %s;", buf1, |
| | 1886 | lmod_cast_s(po, po->operand[0].lmod), buf1, |
| | 1887 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
| | 1888 | last_arith_dst = &po->operand[0]; |
| | 1889 | delayed_flag_op = NULL; |
| | 1890 | break; |
| | 1891 | |
| | 1892 | case OP_ADC: |
| | 1893 | case OP_SBB: |
| | 1894 | assert_operand_cnt(2); |
| | 1895 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
| | 1896 | fprintf(fout, " %s %s= %s + cond_c;", |
| | 1897 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]), |
| | 1898 | op_to_c(po), |
| | 1899 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[1], 0)); |
| | 1900 | last_arith_dst = &po->operand[0]; |
| | 1901 | delayed_flag_op = NULL; |
| | 1902 | break; |
| | 1903 | |
| | 1904 | case OP_INC: |
| | 1905 | case OP_DEC: |
| | 1906 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
| | 1907 | strcpy(buf2, po->op == OP_INC ? "++" : "--"); |
| | 1908 | fprintf(fout, " %s%s;", buf1, buf2); |
| | 1909 | last_arith_dst = &po->operand[0]; |
| | 1910 | delayed_flag_op = NULL; |
| | 1911 | break; |
| | 1912 | |
| | 1913 | case OP_NEG: |
| | 1914 | out_dst_opr(buf1, sizeof(buf1), po, &po->operand[0]); |
| | 1915 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0); |
| | 1916 | fprintf(fout, " %s = -%s%s;", buf1, |
| | 1917 | lmod_cast_s(po, po->operand[0].lmod), buf2); |
| | 1918 | last_arith_dst = &po->operand[0]; |
| | 1919 | delayed_flag_op = NULL; |
| | 1920 | if (pfomask & (1 << PFO_C)) { |
| | 1921 | fprintf(fout, "\n cond_c = (%s != 0);", buf1); |
| | 1922 | pfomask &= ~(1 << PFO_C); |
| | 1923 | } |
| | 1924 | break; |
| | 1925 | |
| | 1926 | case OP_IMUL: |
| | 1927 | if (po->operand_cnt == 2) |
| | 1928 | goto dualop_arith; |
| | 1929 | if (po->operand_cnt == 3) |
| | 1930 | ferr(po, "TODO imul3\n"); |
| | 1931 | // fallthrough |
| | 1932 | case OP_MUL: |
| | 1933 | assert_operand_cnt(1); |
| | 1934 | strcpy(buf1, po->op == OP_IMUL ? "(s64)(s32)" : "(u64)"); |
| | 1935 | fprintf(fout, " mul_tmp = %seax * %s%s;\n", buf1, buf1, |
| | 1936 | out_src_opr(buf2, sizeof(buf2), po, &po->operand[0], 0)); |
| | 1937 | fprintf(fout, " edx = mul_tmp >> 32;\n"); |
| | 1938 | fprintf(fout, " eax = mul_tmp;"); |
| | 1939 | last_arith_dst = NULL; |
| | 1940 | delayed_flag_op = NULL; |
| | 1941 | break; |
| | 1942 | |
| | 1943 | case OP_DIV: |
| | 1944 | case OP_IDIV: |
| | 1945 | assert_operand_cnt(1); |
| | 1946 | if (po->operand[0].lmod != OPLM_DWORD) |
| | 1947 | ferr(po, "unhandled lmod %d\n", po->operand[0].lmod); |
| | 1948 | |
| | 1949 | // 32bit division is common, look for it |
| | 1950 | if (scan_for_cdq_edx(i - 1) >= 0) { |
| | 1951 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0); |
| | 1952 | strcpy(buf2, lmod_cast(po, po->operand[0].lmod, |
| | 1953 | po->op == OP_IDIV)); |
| | 1954 | fprintf(fout, " edx = %seax %% %s%s;\n", buf2, buf2, buf1); |
| | 1955 | fprintf(fout, " eax = %seax / %s%s;", buf2, buf2, buf1); |
| | 1956 | } |
| | 1957 | else |
| | 1958 | ferr(po, "TODO 64bit divident\n"); |
| | 1959 | last_arith_dst = NULL; |
| | 1960 | delayed_flag_op = NULL; |
| | 1961 | break; |
| | 1962 | |
| | 1963 | case OP_TEST: |
| | 1964 | case OP_CMP: |
| | 1965 | propagate_lmod(po, &po->operand[0], &po->operand[1]); |
| | 1966 | if (pfomask != 0) { |
| | 1967 | for (j = 0; j < 8; j++) { |
| | 1968 | if (pfomask & (1 << j)) { |
| | 1969 | out_cmp_test(buf1, sizeof(buf1), po, j, 0); |
| | 1970 | fprintf(fout, " cond_%s = %s;", |
| | 1971 | parsed_flag_op_names[j], buf1); |
| | 1972 | } |
| | 1973 | } |
| | 1974 | pfomask = 0; |
| | 1975 | } |
| | 1976 | else |
| | 1977 | no_output = 1; |
| | 1978 | delayed_flag_op = po; |
| | 1979 | break; |
| | 1980 | |
| | 1981 | // note: we reuse OP_Jcc for SETcc, only flags differ |
| | 1982 | case OP_JO ... OP_JG: |
| | 1983 | if (po->flags & OPF_JMP) |
| | 1984 | fprintf(fout, " goto %s;", po->operand[0].name); |
| | 1985 | // else SETcc - should already be handled |
| | 1986 | break; |
| | 1987 | |
| | 1988 | case OP_JMP: |
| | 1989 | assert_operand_cnt(1); |
| | 1990 | if (po->operand[0].type != OPT_LABEL) |
| | 1991 | ferr(po, "unhandled call type\n"); |
| | 1992 | |
| | 1993 | fprintf(fout, " goto %s;", po->operand[0].name); |
| | 1994 | break; |
| | 1995 | |
| | 1996 | case OP_CALL: |
| | 1997 | assert_operand_cnt(1); |
| | 1998 | if (po->operand[0].type != OPT_LABEL) |
| | 1999 | ferr(po, "unhandled call type\n"); |
| | 2000 | |
| | 2001 | pp = po->datap; |
| | 2002 | if (pp == NULL) |
| | 2003 | ferr(po, "NULL pp\n"); |
| | 2004 | |
| | 2005 | fprintf(fout, " "); |
| | 2006 | if (!IS(pp->ret_type, "void")) { |
| | 2007 | if (po->flags & OPF_TAIL) |
| | 2008 | fprintf(fout, "return "); |
| | 2009 | else |
| | 2010 | fprintf(fout, "eax = "); |
| | 2011 | if (strchr(pp->ret_type, '*')) |
| | 2012 | fprintf(fout, "(u32)"); |
| | 2013 | } |
| | 2014 | |
| | 2015 | fprintf(fout, "%s(", opr_name(po, 0)); |
| | 2016 | for (arg = 0; arg < pp->argc; arg++) { |
| | 2017 | if (arg > 0) |
| | 2018 | fprintf(fout, ", "); |
| | 2019 | |
| | 2020 | if (strchr(pp->arg[arg].type, '*')) |
| | 2021 | fprintf(fout, "(%s)", pp->arg[arg].type); |
| | 2022 | |
| | 2023 | if (pp->arg[arg].reg != NULL) { |
| | 2024 | fprintf(fout, "%s", pp->arg[arg].reg); |
| | 2025 | continue; |
| | 2026 | } |
| | 2027 | |
| | 2028 | // stack arg |
| | 2029 | tmp_op = pp->arg[arg].datap; |
| | 2030 | if (tmp_op == NULL) |
| | 2031 | ferr(po, "parsed_op missing for arg%d\n", arg); |
| | 2032 | if (tmp_op->argmask) { |
| | 2033 | fprintf(fout, "s_a%d", arg + 1); |
| | 2034 | } |
| | 2035 | else { |
| | 2036 | fprintf(fout, "%s", |
| | 2037 | out_src_opr(buf1, sizeof(buf1), |
| | 2038 | tmp_op, &tmp_op->operand[0], 0)); |
| | 2039 | } |
| | 2040 | } |
| | 2041 | fprintf(fout, ");"); |
| | 2042 | |
| | 2043 | if (po->flags & OPF_TAIL) { |
| | 2044 | strcpy(g_comment, "tailcall"); |
| | 2045 | if (IS(pp->ret_type, "void")) |
| | 2046 | fprintf(fout, "\n return;"); |
| | 2047 | } |
| | 2048 | delayed_flag_op = NULL; |
| | 2049 | last_arith_dst = NULL; |
| | 2050 | break; |
| | 2051 | |
| | 2052 | case OP_RET: |
| | 2053 | if (IS(g_func_pp.ret_type, "void")) |
| | 2054 | fprintf(fout, " return;"); |
| | 2055 | else |
| | 2056 | fprintf(fout, " return eax;"); |
| | 2057 | break; |
| | 2058 | |
| | 2059 | case OP_PUSH: |
| | 2060 | if (po->argmask) { |
| | 2061 | // special case - saved func arg |
| | 2062 | for (j = 0; j < 32; j++) { |
| | 2063 | if (po->argmask & (1 << j)) { |
| | 2064 | fprintf(fout, " s_a%d = %s;", j + 1, |
| | 2065 | out_src_opr(buf1, sizeof(buf1), po, &po->operand[0], 0)); |
| | 2066 | } |
| | 2067 | } |
| | 2068 | break; |
| | 2069 | } |
| | 2070 | ferr(po, "push encountered\n"); |
| | 2071 | break; |
| | 2072 | |
| | 2073 | case OP_POP: |
| | 2074 | ferr(po, "pop encountered\n"); |
| | 2075 | break; |
| | 2076 | |
| | 2077 | case OP_NOP: |
| | 2078 | break; |
| | 2079 | |
| | 2080 | default: |
| | 2081 | no_output = 1; |
| | 2082 | ferr(po, "unhandled op type %d, flags %x\n", |
| | 2083 | po->op, po->flags); |
| | 2084 | break; |
| | 2085 | } |
| | 2086 | |
| | 2087 | if (g_comment[0] != 0) { |
| | 2088 | fprintf(fout, " // %s", g_comment); |
| | 2089 | g_comment[0] = 0; |
| | 2090 | no_output = 0; |
| | 2091 | } |
| | 2092 | if (!no_output) |
| | 2093 | fprintf(fout, "\n"); |
| | 2094 | |
| | 2095 | if (pfomask != 0) |
| | 2096 | ferr(po, "missed flag calc, pfomask=%x\n", pfomask); |
| | 2097 | |
| | 2098 | // see is delayed flag stuff is still valid |
| | 2099 | if (delayed_flag_op != NULL && delayed_flag_op != po) { |
| | 2100 | if (is_any_opr_modified(delayed_flag_op, po)) |
| | 2101 | delayed_flag_op = NULL; |
| | 2102 | } |
| | 2103 | |
| | 2104 | if (last_arith_dst != NULL && last_arith_dst != &po->operand[0]) { |
| | 2105 | if (is_opr_modified(last_arith_dst, po)) |
| | 2106 | last_arith_dst = NULL; |
| | 2107 | } |
| | 2108 | } |
| | 2109 | |
| | 2110 | fprintf(fout, "}\n\n"); |
| | 2111 | |
| | 2112 | // cleanup |
| | 2113 | for (i = 0; i < opcnt; i++) { |
| | 2114 | if (ops[i].op == OP_CALL) { |
| | 2115 | pp = ops[i].datap; |
| | 2116 | if (pp) { |
| | 2117 | proto_release(pp); |
| | 2118 | free(pp); |
| | 2119 | } |
| | 2120 | } |
| | 2121 | } |
| | 2122 | proto_release(&g_func_pp); |
| | 2123 | } |
| | 2124 | |
| | 2125 | int main(int argc, char *argv[]) |
| | 2126 | { |
| | 2127 | FILE *fout, *fasm, *fhdr; |
| | 2128 | char line[256]; |
| | 2129 | char words[16][256]; |
| | 2130 | int in_func = 0; |
| | 2131 | int skip_warned = 0; |
| | 2132 | int eq_alloc; |
| | 2133 | int pi = 0; |
| | 2134 | int len; |
| | 2135 | char *p; |
| | 2136 | int wordc; |
| | 2137 | |
| | 2138 | if (argc != 4) { |
| | 2139 | printf("usage:\n%s <.c> <.asm> <hdrf>\n", |
| | 2140 | argv[0]); |
| | 2141 | return 1; |
| | 2142 | } |
| | 2143 | |
| | 2144 | hdrfn = argv[3]; |
| | 2145 | fhdr = fopen(hdrfn, "r"); |
| | 2146 | my_assert_not(fhdr, NULL); |
| | 2147 | |
| | 2148 | asmfn = argv[2]; |
| | 2149 | fasm = fopen(asmfn, "r"); |
| | 2150 | my_assert_not(fasm, NULL); |
| | 2151 | |
| | 2152 | fout = fopen(argv[1], "w"); |
| | 2153 | my_assert_not(fout, NULL); |
| | 2154 | |
| | 2155 | eq_alloc = 128; |
| | 2156 | g_eqs = malloc(eq_alloc * sizeof(g_eqs[0])); |
| | 2157 | my_assert_not(g_eqs, NULL); |
| | 2158 | |
| | 2159 | while (fgets(line, sizeof(line), fasm)) |
| | 2160 | { |
| | 2161 | asmln++; |
| | 2162 | |
| | 2163 | p = sskip(line); |
| | 2164 | if (*p == 0 || *p == ';') |
| | 2165 | continue; |
| | 2166 | |
| | 2167 | memset(words, 0, sizeof(words)); |
| | 2168 | for (wordc = 0; wordc < 16; wordc++) { |
| | 2169 | p = sskip(next_word(words[wordc], sizeof(words[0]), p)); |
| | 2170 | if (*p == 0 || *p == ';') { |
| | 2171 | wordc++; |
| | 2172 | break; |
| | 2173 | } |
| | 2174 | } |
| | 2175 | |
| | 2176 | if (wordc == 0) { |
| | 2177 | // shouldn't happen |
| | 2178 | awarn("wordc == 0?\n"); |
| | 2179 | continue; |
| | 2180 | } |
| | 2181 | |
| | 2182 | // don't care about this: |
| | 2183 | if (words[0][0] == '.' |
| | 2184 | || IS(words[0], "include") |
| | 2185 | || IS(words[0], "assume") || IS(words[1], "segment") |
| | 2186 | || IS(words[0], "align")) |
| | 2187 | { |
| | 2188 | continue; |
| | 2189 | } |
| | 2190 | |
| | 2191 | if (IS(words[1], "proc")) { |
| | 2192 | if (in_func) |
| | 2193 | aerr("proc '%s' while in_func '%s'?\n", |
| | 2194 | words[0], g_func); |
| | 2195 | strcpy(g_func, words[0]); |
| | 2196 | in_func = 1; |
| | 2197 | continue; |
| | 2198 | } |
| | 2199 | |
| | 2200 | if (IS(words[1], "endp")) { |
| | 2201 | if (!in_func) |
| | 2202 | aerr("endp '%s' while not in_func?\n", words[0]); |
| | 2203 | if (!IS(g_func, words[0])) |
| | 2204 | aerr("endp '%s' while in_func '%s'?\n", |
| | 2205 | words[0], g_func); |
| | 2206 | gen_func(fout, fhdr, g_func, pi); |
| | 2207 | in_func = 0; |
| | 2208 | skip_warned = 0; |
| | 2209 | g_func[0] = 0; |
| | 2210 | if (pi != 0) { |
| | 2211 | memset(&ops, 0, pi * sizeof(ops[0])); |
| | 2212 | memset(g_labels, 0, pi * sizeof(g_labels[0])); |
| | 2213 | memset(g_label_refs, 0, pi * sizeof(g_label_refs[0])); |
| | 2214 | pi = 0; |
| | 2215 | } |
| | 2216 | g_eqcnt = 0; |
| | 2217 | continue; |
| | 2218 | } |
| | 2219 | |
| | 2220 | if (IS(words[1], "=")) { |
| | 2221 | if (wordc != 5) |
| | 2222 | aerr("unhandled equ, wc=%d\n", wordc); |
| | 2223 | if (g_eqcnt >= eq_alloc) { |
| | 2224 | eq_alloc *= 2; |
| | 2225 | g_eqs = realloc(g_eqs, eq_alloc * sizeof(g_eqs[0])); |
| | 2226 | my_assert_not(g_eqs, NULL); |
| | 2227 | } |
| | 2228 | |
| | 2229 | len = strlen(words[0]); |
| | 2230 | if (len > sizeof(g_eqs[0].name) - 1) |
| | 2231 | aerr("equ name too long: %d\n", len); |
| | 2232 | strcpy(g_eqs[g_eqcnt].name, words[0]); |
| | 2233 | |
| | 2234 | if (!IS(words[3], "ptr")) |
| | 2235 | aerr("unhandled equ\n"); |
| | 2236 | if (IS(words[2], "dword")) |
| | 2237 | g_eqs[g_eqcnt].lmod = OPLM_DWORD; |
| | 2238 | else if (IS(words[2], "word")) |
| | 2239 | g_eqs[g_eqcnt].lmod = OPLM_WORD; |
| | 2240 | else if (IS(words[2], "byte")) |
| | 2241 | g_eqs[g_eqcnt].lmod = OPLM_BYTE; |
| | 2242 | else |
| | 2243 | aerr("bad lmod: '%s'\n", words[2]); |
| | 2244 | |
| | 2245 | g_eqs[g_eqcnt].offset = parse_number(words[4]); |
| | 2246 | g_eqcnt++; |
| | 2247 | continue; |
| | 2248 | } |
| | 2249 | |
| | 2250 | if (pi >= ARRAY_SIZE(ops)) |
| | 2251 | aerr("too many ops\n"); |
| | 2252 | |
| | 2253 | p = strchr(words[0], ':'); |
| | 2254 | if (p != NULL) { |
| | 2255 | len = p - words[0]; |
| | 2256 | if (len > sizeof(g_labels[0]) - 1) |
| | 2257 | aerr("label too long: %d\n", len); |
| | 2258 | if (g_labels[pi][0] != 0) |
| | 2259 | aerr("dupe label?\n"); |
| | 2260 | memcpy(g_labels[pi], words[0], len); |
| | 2261 | g_labels[pi][len] = 0; |
| | 2262 | continue; |
| | 2263 | } |
| | 2264 | |
| | 2265 | if (!in_func) { |
| | 2266 | if (!skip_warned && g_labels[pi][0] != 0) { |
| | 2267 | anote("skipping from '%s'\n", g_labels[pi]); |
| | 2268 | skip_warned = 1; |
| | 2269 | } |
| | 2270 | g_labels[pi][0] = 0; |
| | 2271 | continue; |
| | 2272 | } |
| | 2273 | |
| | 2274 | parse_op(&ops[pi], words, wordc); |
| | 2275 | pi++; |
| | 2276 | } |
| | 2277 | |
| | 2278 | fclose(fout); |
| | 2279 | fclose(fasm); |
| | 2280 | fclose(fhdr); |
| | 2281 | |
| | 2282 | return 0; |
| | 2283 | } |
| | 2284 | |
| | 2285 | // vim:ts=2:shiftwidth=2:expandtab |
notaz.gp2x.de / ia32rtools.git / blame_incremental