Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 1 | /* Capstone Disassembler Engine */ |
Nguyen Anh Quynh | 3edc30d | 2014-08-15 13:53:03 +0800 | [diff] [blame] | 2 | /* By David Hogarty, 2014 */ |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 3 | |
| 4 | // the following must precede stdio (woo, thanks msft) |
mrexodia | d9f96e7 | 2016-09-15 15:58:05 +0200 | [diff] [blame] | 5 | #if defined(_MSC_VER) && _MSC_VER < 1900 |
| 6 | #define _CRT_SECURE_NO_WARNINGS |
| 7 | #define snprintf _snprintf |
Nguyen Anh Quynh | 568718b | 2016-03-08 11:11:20 +0800 | [diff] [blame] | 8 | #endif |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 9 | #include <stdio.h> |
| 10 | #include <stdlib.h> |
| 11 | #include <string.h> |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 12 | |
tandasat | 45e5eab | 2016-05-11 21:48:32 -0700 | [diff] [blame] | 13 | #include <capstone/platform.h> |
learn_more | 04f9c32 | 2015-04-21 19:27:42 +0200 | [diff] [blame] | 14 | #include <capstone/capstone.h> |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 15 | |
| 16 | static csh handle; |
| 17 | |
| 18 | struct platform { |
| 19 | cs_arch arch; |
| 20 | cs_mode mode; |
| 21 | unsigned char *code; |
| 22 | size_t size; |
| 23 | char *comment; |
| 24 | int syntax; |
| 25 | }; |
| 26 | |
Nguyen Anh Quynh | 4b68d95 | 2015-02-25 18:01:02 +0800 | [diff] [blame] | 27 | static char *hex_string(unsigned char *str, size_t len) |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 28 | { |
| 29 | // returns a malloced string that has the hex version of the string in it |
| 30 | // null if failed to malloc |
Nguyen Anh Quynh | 967e987 | 2015-02-25 18:11:26 +0800 | [diff] [blame] | 31 | char *hex_out; |
| 32 | size_t i; |
| 33 | |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 34 | hex_out = (char *) malloc(len*2 + 1); // two ascii characters per input character, plus trailing null |
| 35 | if (!hex_out) { goto Exit; } |
| 36 | |
| 37 | for (i = 0; i < len; ++i) { |
| 38 | snprintf(hex_out + (i*2), 2, "%02x", str[i]); |
| 39 | } |
| 40 | |
| 41 | hex_out[len*2] = 0; // trailing null |
| 42 | |
| 43 | Exit: |
| 44 | return hex_out; |
| 45 | } |
| 46 | |
| 47 | static void snprint_insn_detail(char * buf, size_t * cur, size_t * left, cs_insn *ins) |
| 48 | { |
| 49 | size_t used = 0; |
| 50 | |
| 51 | #define _this_printf(...) \ |
| 52 | { \ |
| 53 | size_t used = 0; \ |
| 54 | used = snprintf(buf + *cur, *left, __VA_ARGS__); \ |
| 55 | *left -= used; \ |
| 56 | *cur += used; \ |
| 57 | } |
| 58 | |
| 59 | cs_arm *arm; |
| 60 | int i; |
| 61 | |
| 62 | // detail can be NULL on "data" instruction if SKIPDATA option is turned ON |
| 63 | if (ins->detail == NULL) |
| 64 | return; |
| 65 | |
| 66 | arm = &(ins->detail->arm); |
| 67 | |
| 68 | if (arm->op_count) |
| 69 | _this_printf("\top_count: %u\n", arm->op_count); |
| 70 | |
| 71 | for (i = 0; i < arm->op_count; i++) { |
| 72 | cs_arm_op *op = &(arm->operands[i]); |
| 73 | switch((int)op->type) { |
| 74 | default: |
| 75 | break; |
| 76 | case ARM_OP_REG: |
| 77 | _this_printf("\t\toperands[%u].type: REG = %s\n", i, cs_reg_name(handle, op->reg)); |
| 78 | break; |
| 79 | case ARM_OP_IMM: |
| 80 | _this_printf("\t\toperands[%u].type: IMM = 0x%x\n", i, op->imm); |
| 81 | break; |
| 82 | case ARM_OP_FP: |
| 83 | _this_printf("\t\toperands[%u].type: FP = %f\n", i, op->fp); |
| 84 | break; |
| 85 | case ARM_OP_MEM: |
| 86 | _this_printf("\t\toperands[%u].type: MEM\n", i); |
| 87 | if (op->mem.base != X86_REG_INVALID) |
| 88 | _this_printf("\t\t\toperands[%u].mem.base: REG = %s\n", |
| 89 | i, cs_reg_name(handle, op->mem.base)); |
| 90 | if (op->mem.index != X86_REG_INVALID) |
| 91 | _this_printf("\t\t\toperands[%u].mem.index: REG = %s\n", |
| 92 | i, cs_reg_name(handle, op->mem.index)); |
| 93 | if (op->mem.scale != 1) |
| 94 | _this_printf("\t\t\toperands[%u].mem.scale: %u\n", i, op->mem.scale); |
| 95 | if (op->mem.disp != 0) |
| 96 | _this_printf("\t\t\toperands[%u].mem.disp: 0x%x\n", i, op->mem.disp); |
| 97 | |
| 98 | break; |
| 99 | case ARM_OP_PIMM: |
| 100 | _this_printf("\t\toperands[%u].type: P-IMM = %u\n", i, op->imm); |
| 101 | break; |
| 102 | case ARM_OP_CIMM: |
| 103 | _this_printf("\t\toperands[%u].type: C-IMM = %u\n", i, op->imm); |
| 104 | break; |
| 105 | } |
| 106 | |
| 107 | if (op->shift.type != ARM_SFT_INVALID && op->shift.value) { |
| 108 | if (op->shift.type < ARM_SFT_ASR_REG) { |
| 109 | // shift with constant value |
| 110 | _this_printf("\t\t\tShift: %u = %u\n", op->shift.type, op->shift.value); |
| 111 | } else { |
| 112 | // shift with register |
| 113 | _this_printf("\t\t\tShift: %u = %s\n", op->shift.type, |
| 114 | cs_reg_name(handle, op->shift.value)); |
| 115 | } |
| 116 | } |
| 117 | } |
| 118 | |
| 119 | if (arm->cc != ARM_CC_AL && arm->cc != ARM_CC_INVALID) { |
| 120 | _this_printf("\tCode condition: %u\n", arm->cc); |
| 121 | } |
| 122 | |
| 123 | if (arm->update_flags) { |
| 124 | _this_printf("\tUpdate-flags: True\n"); |
| 125 | } |
| 126 | |
| 127 | if (arm->writeback) { |
| 128 | _this_printf("\tWrite-back: True\n"); |
| 129 | } |
| 130 | |
| 131 | #undef _this_printf |
| 132 | } |
| 133 | |
| 134 | static void print_insn_detail(cs_insn *ins) |
| 135 | { |
| 136 | char a_buf[2048]; |
| 137 | size_t cur=0, left=2048; |
| 138 | snprint_insn_detail(a_buf, &cur, &left, ins); |
| 139 | printf("%s\n", a_buf); |
| 140 | } |
| 141 | |
| 142 | struct invalid_code { |
| 143 | unsigned char *code; |
| 144 | size_t size; |
| 145 | char *comment; |
| 146 | }; |
| 147 | |
| 148 | #define MAX_INVALID_CODES 16 |
| 149 | |
| 150 | struct invalid_instructions { |
| 151 | cs_arch arch; |
| 152 | cs_mode mode; |
| 153 | char *platform_comment; |
| 154 | int num_invalid_codes; |
| 155 | struct invalid_code invalid_codes[MAX_INVALID_CODES]; |
| 156 | }; |
| 157 | |
| 158 | static void test_invalids() |
| 159 | { |
| 160 | struct invalid_instructions invalids[] = {{ |
| 161 | CS_ARCH_ARM, |
| 162 | CS_MODE_THUMB, |
| 163 | "Thumb", |
| 164 | 1, |
| 165 | {{ |
| 166 | (unsigned char *)"\xbd\xe8\x1e\xff", |
| 167 | 4, |
| 168 | "invalid thumb2 pop because sp used and because both pc and lr are " |
| 169 | "present at the same time" |
| 170 | }}, |
| 171 | }}; |
| 172 | |
| 173 | struct invalid_instructions * invalid = NULL; |
| 174 | |
| 175 | uint64_t address = 0x1000; |
| 176 | cs_insn *insn; |
| 177 | int i; |
| 178 | int j; |
| 179 | size_t count; |
| 180 | |
| 181 | printf("\nShould be invalid\n" |
| 182 | "-----------------\n"); |
| 183 | |
| 184 | for (i = 0; i < sizeof(invalids)/sizeof(invalids[0]); i++) { |
| 185 | cs_err err; |
| 186 | |
| 187 | invalid = invalids + i; |
| 188 | err = cs_open(invalid->arch, invalid->mode, &handle); |
| 189 | |
| 190 | if (err) { |
| 191 | printf("Failed on cs_open() with error returned: %u\n", err); |
| 192 | continue; |
| 193 | } |
| 194 | |
| 195 | cs_option(handle, CS_OPT_DETAIL, CS_OPT_ON); |
flyingsymbols | d91f964 | 2014-10-22 03:21:43 -0400 | [diff] [blame] | 196 | cs_option(handle, CS_OPT_SYNTAX, CS_OPT_SYNTAX_NOREGNAME); |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 197 | |
| 198 | for (j = 0; j < invalid->num_invalid_codes; ++j) { |
Nguyen Anh Quynh | 4b68d95 | 2015-02-25 18:01:02 +0800 | [diff] [blame] | 199 | struct invalid_code *invalid_code = NULL; |
| 200 | char *hex_str = NULL; |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 201 | |
| 202 | invalid_code = invalid->invalid_codes + j; |
| 203 | |
| 204 | hex_str = hex_string(invalid_code->code, invalid_code->size); |
| 205 | |
| 206 | printf("%s %s: %s\n", invalid->platform_comment, hex_str, invalid_code->comment); |
| 207 | |
| 208 | free(hex_str); |
| 209 | |
Nguyen Anh Quynh | ad449b5 | 2014-10-24 00:36:48 +0800 | [diff] [blame] | 210 | count = cs_disasm(handle, |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 211 | invalid_code->code, invalid_code->size, address, 0, &insn |
| 212 | ); |
| 213 | |
| 214 | if (count) { |
| 215 | size_t k; |
| 216 | printf(" ERROR:\n"); |
| 217 | |
| 218 | for (k = 0; k < count; k++) { |
| 219 | printf(" 0x%"PRIx64":\t%s\t%s\n", |
| 220 | insn[k].address, insn[k].mnemonic, insn[k].op_str); |
| 221 | print_insn_detail(&insn[k]); |
| 222 | } |
| 223 | cs_free(insn, count); |
| 224 | |
| 225 | } else { |
| 226 | printf(" SUCCESS: invalid\n"); |
| 227 | } |
| 228 | } |
| 229 | |
| 230 | cs_close(&handle); |
| 231 | } |
| 232 | } |
| 233 | |
| 234 | struct valid_code { |
| 235 | unsigned char *code; |
| 236 | size_t size; |
| 237 | uint32_t start_addr; |
Nguyen Anh Quynh | 4b68d95 | 2015-02-25 18:01:02 +0800 | [diff] [blame] | 238 | char *expected_out; |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 239 | char *comment; |
| 240 | }; |
| 241 | |
| 242 | #define MAX_VALID_CODES 16 |
| 243 | struct valid_instructions { |
| 244 | cs_arch arch; |
| 245 | cs_mode mode; |
| 246 | char *platform_comment; |
| 247 | int num_valid_codes; |
| 248 | struct valid_code valid_codes[MAX_VALID_CODES]; |
| 249 | }; |
| 250 | |
| 251 | static void test_valids() |
| 252 | { |
| 253 | struct valid_instructions valids[] = {{ |
| 254 | CS_ARCH_ARM, |
| 255 | CS_MODE_THUMB, |
| 256 | "Thumb", |
flyingsymbols | d91f964 | 2014-10-22 03:21:43 -0400 | [diff] [blame] | 257 | 3, |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 258 | {{ (unsigned char *)"\x00\xf0\x26\xe8", 4, 0x352, |
flyingsymbols | d91f964 | 2014-10-22 03:21:43 -0400 | [diff] [blame] | 259 | "0x352:\tblx\t#0x3a0\n" |
| 260 | "\top_count: 1\n" |
| 261 | "\t\toperands[0].type: IMM = 0x3a0\n", |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 262 | |
flyingsymbols | d91f964 | 2014-10-22 03:21:43 -0400 | [diff] [blame] | 263 | "thumb2 blx with misaligned immediate" |
| 264 | }, { (unsigned char *)"\x05\xdd", 2, 0x1f0, |
| 265 | "0x1f0:\tble\t#0x1fe\n" |
| 266 | "\top_count: 1\n" |
| 267 | "\t\toperands[0].type: IMM = 0x1fe\n" |
| 268 | "\tCode condition: 14\n", |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 269 | |
flyingsymbols | d91f964 | 2014-10-22 03:21:43 -0400 | [diff] [blame] | 270 | "thumb b cc with thumb-aligned target" |
| 271 | }, { (unsigned char *)"\xbd\xe8\xf0\x8f", 4, 0, |
| 272 | "0x0:\tpop.w\t{r4, r5, r6, r7, r8, r9, r10, r11, pc}\n" |
| 273 | "\top_count: 9\n" |
| 274 | "\t\toperands[0].type: REG = r4\n" |
| 275 | "\t\toperands[1].type: REG = r5\n" |
| 276 | "\t\toperands[2].type: REG = r6\n" |
| 277 | "\t\toperands[3].type: REG = r7\n" |
| 278 | "\t\toperands[4].type: REG = r8\n" |
| 279 | "\t\toperands[5].type: REG = r9\n" |
| 280 | "\t\toperands[6].type: REG = r10\n" |
| 281 | "\t\toperands[7].type: REG = r11\n" |
| 282 | "\t\toperands[8].type: REG = pc\n", |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 283 | |
flyingsymbols | d91f964 | 2014-10-22 03:21:43 -0400 | [diff] [blame] | 284 | "thumb2 pop that should be valid" |
| 285 | }, |
| 286 | } |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 287 | }}; |
| 288 | |
| 289 | struct valid_instructions * valid = NULL; |
| 290 | |
| 291 | uint64_t address = 0x1000; |
| 292 | cs_insn *insn; |
| 293 | int i; |
| 294 | int j; |
| 295 | size_t count; |
| 296 | |
| 297 | |
| 298 | for (i = 0; i < sizeof(valids)/sizeof(valids[0]); i++) { |
| 299 | cs_err err; |
| 300 | |
| 301 | valid = valids + i; |
| 302 | err = cs_open(valid->arch, valid->mode, &handle); |
| 303 | |
| 304 | if (err) { |
| 305 | printf("Failed on cs_open() with error returned: %u\n", err); |
| 306 | continue; |
| 307 | } |
| 308 | |
| 309 | cs_option(handle, CS_OPT_DETAIL, CS_OPT_ON); |
flyingsymbols | d91f964 | 2014-10-22 03:21:43 -0400 | [diff] [blame] | 310 | cs_option(handle, CS_OPT_SYNTAX, CS_OPT_SYNTAX_NOREGNAME); |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 311 | |
| 312 | #define _this_printf(...) \ |
| 313 | { \ |
| 314 | size_t used = 0; \ |
| 315 | used = snprintf(tmp_buf + cur, left, __VA_ARGS__); \ |
| 316 | left -= used; \ |
| 317 | cur += used; \ |
| 318 | } |
| 319 | printf("\nShould be valid\n" |
| 320 | "---------------\n"); |
| 321 | |
| 322 | for (j = 0; j < valid->num_valid_codes; ++j) { |
| 323 | char tmp_buf[2048]; |
| 324 | size_t left = 2048; |
| 325 | size_t cur = 0; |
| 326 | size_t used = 0; |
| 327 | int success = 0; |
| 328 | char * hex_str = NULL; |
| 329 | |
| 330 | struct valid_code * valid_code = NULL; |
| 331 | valid_code = valid->valid_codes + j; |
| 332 | |
| 333 | hex_str = hex_string(valid_code->code, valid_code->size); |
| 334 | |
| 335 | printf("%s %s @ 0x%04x: %s\n %s", |
| 336 | valid->platform_comment, hex_str, valid_code->start_addr, |
| 337 | valid_code->comment, valid_code->expected_out); |
| 338 | |
Nguyen Anh Quynh | ad449b5 | 2014-10-24 00:36:48 +0800 | [diff] [blame] | 339 | count = cs_disasm(handle, |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 340 | valid_code->code, valid_code->size, |
| 341 | valid_code->start_addr, 0, &insn |
| 342 | ); |
| 343 | |
| 344 | if (count) { |
| 345 | size_t k; |
| 346 | size_t max_len = 0; |
| 347 | size_t tmp_len = 0; |
| 348 | |
| 349 | for (k = 0; k < count; k++) { |
| 350 | _this_printf( |
| 351 | "0x%"PRIx64":\t%s\t%s\n", |
| 352 | insn[k].address, insn[k].mnemonic, |
| 353 | insn[k].op_str |
| 354 | ); |
| 355 | |
| 356 | snprint_insn_detail(tmp_buf, &cur, &left, &insn[k]); |
| 357 | } |
| 358 | |
| 359 | max_len = strlen(tmp_buf); |
| 360 | tmp_len = strlen(valid_code->expected_out); |
| 361 | if (tmp_len > max_len) { |
| 362 | max_len = tmp_len; |
| 363 | } |
| 364 | |
| 365 | if (memcmp(tmp_buf, valid_code->expected_out, max_len)) { |
| 366 | printf( |
flyingsymbols | d91f964 | 2014-10-22 03:21:43 -0400 | [diff] [blame] | 367 | " ERROR: '''\n%s''' does not match" |
| 368 | " expected '''\n%s'''\n", |
| 369 | tmp_buf, valid_code->expected_out |
| 370 | ); |
Nguyen Anh Quynh | 26dfbc6 | 2014-07-31 18:24:51 +0800 | [diff] [blame] | 371 | } else { |
| 372 | printf(" SUCCESS: valid\n"); |
| 373 | } |
| 374 | |
| 375 | cs_free(insn, count); |
| 376 | |
| 377 | } else { |
| 378 | printf("ERROR: invalid\n"); |
| 379 | } |
| 380 | } |
| 381 | |
| 382 | cs_close(&handle); |
| 383 | } |
| 384 | |
| 385 | #undef _this_prinf |
| 386 | } |
| 387 | |
| 388 | int main() |
| 389 | { |
| 390 | test_invalids(); |
| 391 | test_valids(); |
| 392 | return 0; |
| 393 | } |
| 394 | |