1 #ifndef CAPSTONE_ENGINE_H 2 #define CAPSTONE_ENGINE_H 3 4 /* Capstone Disassembly Engine */ 5 /* By Nguyen Anh Quynh <aquynh@gmail.com>, 2013-2016 */ 6 7 #ifdef __cplusplus 8 extern "C" { 9 #endif 10 11 #include <stdarg.h> 12 13 #if defined(CAPSTONE_HAS_OSXKERNEL) 14 #include <libkern/libkern.h> 15 #else 16 #include <stdlib.h> 17 #include <stdio.h> 18 #endif 19 20 #include "platform.h" 21 22 #ifdef _MSC_VER 23 #pragma warning(disable:4201) 24 #pragma warning(disable:4100) 25 #define CAPSTONE_API __cdecl 26 #ifdef CAPSTONE_SHARED 27 #define CAPSTONE_EXPORT __declspec(dllexport) 28 #else // defined(CAPSTONE_STATIC) 29 #define CAPSTONE_EXPORT 30 #endif 31 #else 32 #define CAPSTONE_API 33 #if defined(__GNUC__) && !defined(CAPSTONE_STATIC) 34 #define CAPSTONE_EXPORT __attribute__((visibility("default"))) 35 #else // defined(CAPSTONE_STATIC) 36 #define CAPSTONE_EXPORT 37 #endif 38 #endif 39 40 #ifdef __GNUC__ 41 #define CAPSTONE_DEPRECATED __attribute__((deprecated)) 42 #elif defined(_MSC_VER) 43 #define CAPSTONE_DEPRECATED __declspec(deprecated) 44 #else 45 #pragma message("WARNING: You need to implement CAPSTONE_DEPRECATED for this compiler") 46 #define CAPSTONE_DEPRECATED 47 #endif 48 49 // Capstone API version 50 #define CS_API_MAJOR 5 51 #define CS_API_MINOR 0 52 53 // Version for bleeding edge code of the Github's "next" branch. 54 // Use this if you want the absolutely latest development code. 55 // This version number will be bumped up whenever we have a new major change. 56 #define CS_NEXT_VERSION 5 57 58 // Capstone package version 59 #define CS_VERSION_MAJOR CS_API_MAJOR 60 #define CS_VERSION_MINOR CS_API_MINOR 61 #define CS_VERSION_EXTRA 0 62 63 /// Macro to create combined version which can be compared to 64 /// result of cs_version() API. 65 #define CS_MAKE_VERSION(major, minor) ((major << 8) + minor) 66 67 /// Maximum size of an instruction mnemonic string. 68 #define CS_MNEMONIC_SIZE 32 69 70 // Handle using with all API 71 typedef size_t csh; 72 73 /// Architecture type 74 typedef enum cs_arch { 75 CS_ARCH_ARM = 0, ///< ARM architecture (including Thumb, Thumb-2) 76 CS_ARCH_ARM64, ///< ARM-64, also called AArch64 77 CS_ARCH_MIPS, ///< Mips architecture 78 CS_ARCH_X86, ///< X86 architecture (including x86 & x86-64) 79 CS_ARCH_PPC, ///< PowerPC architecture 80 CS_ARCH_SPARC, ///< Sparc architecture 81 CS_ARCH_SYSZ, ///< SystemZ architecture 82 CS_ARCH_XCORE, ///< XCore architecture 83 CS_ARCH_M68K, ///< 68K architecture 84 CS_ARCH_TMS320C64X, ///< TMS320C64x architecture 85 CS_ARCH_M680X, ///< 680X architecture 86 CS_ARCH_EVM, ///< Ethereum architecture 87 CS_ARCH_MOS65XX, ///< MOS65XX architecture (including MOS6502) 88 CS_ARCH_MAX, 89 CS_ARCH_ALL = 0xFFFF, // All architectures - for cs_support() 90 } cs_arch; 91 92 // Support value to verify diet mode of the engine. 93 // If cs_support(CS_SUPPORT_DIET) return True, the engine was compiled 94 // in diet mode. 95 #define CS_SUPPORT_DIET (CS_ARCH_ALL + 1) 96 97 // Support value to verify X86 reduce mode of the engine. 98 // If cs_support(CS_SUPPORT_X86_REDUCE) return True, the engine was compiled 99 // in X86 reduce mode. 100 #define CS_SUPPORT_X86_REDUCE (CS_ARCH_ALL + 2) 101 102 /// Mode type 103 typedef enum cs_mode { 104 CS_MODE_LITTLE_ENDIAN = 0, ///< little-endian mode (default mode) 105 CS_MODE_ARM = 0, ///< 32-bit ARM 106 CS_MODE_16 = 1 << 1, ///< 16-bit mode (X86) 107 CS_MODE_32 = 1 << 2, ///< 32-bit mode (X86) 108 CS_MODE_64 = 1 << 3, ///< 64-bit mode (X86, PPC) 109 CS_MODE_THUMB = 1 << 4, ///< ARM's Thumb mode, including Thumb-2 110 CS_MODE_MCLASS = 1 << 5, ///< ARM's Cortex-M series 111 CS_MODE_V8 = 1 << 6, ///< ARMv8 A32 encodings for ARM 112 CS_MODE_MICRO = 1 << 4, ///< MicroMips mode (MIPS) 113 CS_MODE_MIPS3 = 1 << 5, ///< Mips III ISA 114 CS_MODE_MIPS32R6 = 1 << 6, ///< Mips32r6 ISA 115 CS_MODE_MIPS2 = 1 << 7, ///< Mips II ISA 116 CS_MODE_V9 = 1 << 4, ///< SparcV9 mode (Sparc) 117 CS_MODE_QPX = 1 << 4, ///< Quad Processing eXtensions mode (PPC) 118 CS_MODE_M68K_000 = 1 << 1, ///< M68K 68000 mode 119 CS_MODE_M68K_010 = 1 << 2, ///< M68K 68010 mode 120 CS_MODE_M68K_020 = 1 << 3, ///< M68K 68020 mode 121 CS_MODE_M68K_030 = 1 << 4, ///< M68K 68030 mode 122 CS_MODE_M68K_040 = 1 << 5, ///< M68K 68040 mode 123 CS_MODE_M68K_060 = 1 << 6, ///< M68K 68060 mode 124 CS_MODE_BIG_ENDIAN = 1U << 31, ///< big-endian mode 125 CS_MODE_MIPS32 = CS_MODE_32, ///< Mips32 ISA (Mips) 126 CS_MODE_MIPS64 = CS_MODE_64, ///< Mips64 ISA (Mips) 127 CS_MODE_M680X_6301 = 1 << 1, ///< M680X Hitachi 6301,6303 mode 128 CS_MODE_M680X_6309 = 1 << 2, ///< M680X Hitachi 6309 mode 129 CS_MODE_M680X_6800 = 1 << 3, ///< M680X Motorola 6800,6802 mode 130 CS_MODE_M680X_6801 = 1 << 4, ///< M680X Motorola 6801,6803 mode 131 CS_MODE_M680X_6805 = 1 << 5, ///< M680X Motorola/Freescale 6805 mode 132 CS_MODE_M680X_6808 = 1 << 6, ///< M680X Motorola/Freescale/NXP 68HC08 mode 133 CS_MODE_M680X_6809 = 1 << 7, ///< M680X Motorola 6809 mode 134 CS_MODE_M680X_6811 = 1 << 8, ///< M680X Motorola/Freescale/NXP 68HC11 mode 135 CS_MODE_M680X_CPU12 = 1 << 9, ///< M680X Motorola/Freescale/NXP CPU12 136 ///< used on M68HC12/HCS12 137 CS_MODE_M680X_HCS08 = 1 << 10, ///< M680X Freescale/NXP HCS08 mode 138 } cs_mode; 139 140 typedef void* (CAPSTONE_API *cs_malloc_t)(size_t size); 141 typedef void* (CAPSTONE_API *cs_calloc_t)(size_t nmemb, size_t size); 142 typedef void* (CAPSTONE_API *cs_realloc_t)(void *ptr, size_t size); 143 typedef void (CAPSTONE_API *cs_free_t)(void *ptr); 144 typedef int (CAPSTONE_API *cs_vsnprintf_t)(char *str, size_t size, const char *format, va_list ap); 145 146 147 /// User-defined dynamic memory related functions: malloc/calloc/realloc/free/vsnprintf() 148 /// By default, Capstone uses system's malloc(), calloc(), realloc(), free() & vsnprintf(). 149 typedef struct cs_opt_mem { 150 cs_malloc_t malloc; 151 cs_calloc_t calloc; 152 cs_realloc_t realloc; 153 cs_free_t free; 154 cs_vsnprintf_t vsnprintf; 155 } cs_opt_mem; 156 157 /// Customize mnemonic for instructions with alternative name. 158 /// To reset existing customized instruction to its default mnemonic, 159 /// call cs_option(CS_OPT_MNEMONIC) again with the same @id and NULL value 160 /// for @mnemonic. 161 typedef struct cs_opt_mnem { 162 /// ID of instruction to be customized. 163 unsigned int id; 164 /// Customized instruction mnemonic. 165 const char *mnemonic; 166 } cs_opt_mnem; 167 168 /// Runtime option for the disassembled engine 169 typedef enum cs_opt_type { 170 CS_OPT_INVALID = 0, ///< No option specified 171 CS_OPT_SYNTAX, ///< Assembly output syntax 172 CS_OPT_DETAIL, ///< Break down instruction structure into details 173 CS_OPT_MODE, ///< Change engine's mode at run-time 174 CS_OPT_MEM, ///< User-defined dynamic memory related functions 175 CS_OPT_SKIPDATA, ///< Skip data when disassembling. Then engine is in SKIPDATA mode. 176 CS_OPT_SKIPDATA_SETUP, ///< Setup user-defined function for SKIPDATA option 177 CS_OPT_MNEMONIC, ///< Customize instruction mnemonic 178 CS_OPT_UNSIGNED, ///< print immediate operands in unsigned form 179 } cs_opt_type; 180 181 /// Runtime option value (associated with option type above) 182 typedef enum cs_opt_value { 183 CS_OPT_OFF = 0, ///< Turn OFF an option - default for CS_OPT_DETAIL, CS_OPT_SKIPDATA, CS_OPT_UNSIGNED. 184 CS_OPT_ON = 3, ///< Turn ON an option (CS_OPT_DETAIL, CS_OPT_SKIPDATA). 185 CS_OPT_SYNTAX_DEFAULT = 0, ///< Default asm syntax (CS_OPT_SYNTAX). 186 CS_OPT_SYNTAX_INTEL, ///< X86 Intel asm syntax - default on X86 (CS_OPT_SYNTAX). 187 CS_OPT_SYNTAX_ATT, ///< X86 ATT asm syntax (CS_OPT_SYNTAX). 188 CS_OPT_SYNTAX_NOREGNAME, ///< Prints register name with only number (CS_OPT_SYNTAX) 189 CS_OPT_SYNTAX_MASM, ///< X86 Intel Masm syntax (CS_OPT_SYNTAX). 190 } cs_opt_value; 191 192 /// Common instruction operand types - to be consistent across all architectures. 193 typedef enum cs_op_type { 194 CS_OP_INVALID = 0, ///< uninitialized/invalid operand. 195 CS_OP_REG, ///< Register operand. 196 CS_OP_IMM, ///< Immediate operand. 197 CS_OP_MEM, ///< Memory operand. 198 CS_OP_FP, ///< Floating-Point operand. 199 } cs_op_type; 200 201 /// Common instruction operand access types - to be consistent across all architectures. 202 /// It is possible to combine access types, for example: CS_AC_READ | CS_AC_WRITE 203 typedef enum cs_ac_type { 204 CS_AC_INVALID = 0, ///< Uninitialized/invalid access type. 205 CS_AC_READ = 1 << 0, ///< Operand read from memory or register. 206 CS_AC_WRITE = 1 << 1, ///< Operand write to memory or register. 207 } cs_ac_type; 208 209 /// Common instruction groups - to be consistent across all architectures. 210 typedef enum cs_group_type { 211 CS_GRP_INVALID = 0, ///< uninitialized/invalid group. 212 CS_GRP_JUMP, ///< all jump instructions (conditional+direct+indirect jumps) 213 CS_GRP_CALL, ///< all call instructions 214 CS_GRP_RET, ///< all return instructions 215 CS_GRP_INT, ///< all interrupt instructions (int+syscall) 216 CS_GRP_IRET, ///< all interrupt return instructions 217 CS_GRP_PRIVILEGE, ///< all privileged instructions 218 CS_GRP_BRANCH_RELATIVE, ///< all relative branching instructions 219 } cs_group_type; 220 221 /** 222 User-defined callback function for SKIPDATA option. 223 See tests/test_skipdata.c for sample code demonstrating this API. 224 225 @code: the input buffer containing code to be disassembled. 226 This is the same buffer passed to cs_disasm(). 227 @code_size: size (in bytes) of the above @code buffer. 228 @offset: the position of the currently-examining byte in the input 229 buffer @code mentioned above. 230 @user_data: user-data passed to cs_option() via @user_data field in 231 cs_opt_skipdata struct below. 232 233 @return: return number of bytes to skip, or 0 to immediately stop disassembling. 234 */ 235 typedef size_t (CAPSTONE_API *cs_skipdata_cb_t)(const uint8_t *code, size_t code_size, size_t offset, void *user_data); 236 237 /// User-customized setup for SKIPDATA option 238 typedef struct cs_opt_skipdata { 239 /// Capstone considers data to skip as special "instructions". 240 /// User can specify the string for this instruction's "mnemonic" here. 241 /// By default (if @mnemonic is NULL), Capstone use ".byte". 242 const char *mnemonic; 243 244 /// User-defined callback function to be called when Capstone hits data. 245 /// If the returned value from this callback is positive (>0), Capstone 246 /// will skip exactly that number of bytes & continue. Otherwise, if 247 /// the callback returns 0, Capstone stops disassembling and returns 248 /// immediately from cs_disasm() 249 /// NOTE: if this callback pointer is NULL, Capstone would skip a number 250 /// of bytes depending on architectures, as following: 251 /// Arm: 2 bytes (Thumb mode) or 4 bytes. 252 /// Arm64: 4 bytes. 253 /// Mips: 4 bytes. 254 /// M680x: 1 byte. 255 /// PowerPC: 4 bytes. 256 /// Sparc: 4 bytes. 257 /// SystemZ: 2 bytes. 258 /// X86: 1 bytes. 259 /// XCore: 2 bytes. 260 /// EVM: 1 bytes. 261 /// MOS65XX: 1 bytes. 262 cs_skipdata_cb_t callback; // default value is NULL 263 264 /// User-defined data to be passed to @callback function pointer. 265 void *user_data; 266 } cs_opt_skipdata; 267 268 269 #include "arm.h" 270 #include "arm64.h" 271 #include "m68k.h" 272 #include "mips.h" 273 #include "ppc.h" 274 #include "sparc.h" 275 #include "systemz.h" 276 #include "x86.h" 277 #include "xcore.h" 278 #include "tms320c64x.h" 279 #include "m680x.h" 280 #include "evm.h" 281 #include "mos65xx.h" 282 283 /// NOTE: All information in cs_detail is only available when CS_OPT_DETAIL = CS_OPT_ON 284 /// Initialized as memset(., 0, offsetof(cs_detail, ARCH)+sizeof(cs_ARCH)) 285 /// by ARCH_getInstruction in arch/ARCH/ARCHDisassembler.c 286 /// if cs_detail changes, in particular if a field is added after the union, 287 /// then update arch/ARCH/ARCHDisassembler.c accordingly 288 typedef struct cs_detail { 289 uint16_t regs_read[16]; ///< list of implicit registers read by this insn 290 uint8_t regs_read_count; ///< number of implicit registers read by this insn 291 292 uint16_t regs_write[20]; ///< list of implicit registers modified by this insn 293 uint8_t regs_write_count; ///< number of implicit registers modified by this insn 294 295 uint8_t groups[8]; ///< list of group this instruction belong to 296 uint8_t groups_count; ///< number of groups this insn belongs to 297 298 /// Architecture-specific instruction info 299 union { 300 cs_x86 x86; ///< X86 architecture, including 16-bit, 32-bit & 64-bit mode 301 cs_arm64 arm64; ///< ARM64 architecture (aka AArch64) 302 cs_arm arm; ///< ARM architecture (including Thumb/Thumb2) 303 cs_m68k m68k; ///< M68K architecture 304 cs_mips mips; ///< MIPS architecture 305 cs_ppc ppc; ///< PowerPC architecture 306 cs_sparc sparc; ///< Sparc architecture 307 cs_sysz sysz; ///< SystemZ architecture 308 cs_xcore xcore; ///< XCore architecture 309 cs_tms320c64x tms320c64x; ///< TMS320C64x architecture 310 cs_m680x m680x; ///< M680X architecture 311 cs_evm evm; ///< Ethereum architecture 312 cs_mos65xx mos65xx; ///< MOS65XX architecture (including MOS6502) 313 }; 314 } cs_detail; 315 316 /// Detail information of disassembled instruction 317 typedef struct cs_insn { 318 /// Instruction ID (basically a numeric ID for the instruction mnemonic) 319 /// Find the instruction id in the '[ARCH]_insn' enum in the header file 320 /// of corresponding architecture, such as 'arm_insn' in arm.h for ARM, 321 /// 'x86_insn' in x86.h for X86, etc... 322 /// This information is available even when CS_OPT_DETAIL = CS_OPT_OFF 323 /// NOTE: in Skipdata mode, "data" instruction has 0 for this id field. 324 unsigned int id; 325 326 /// Address (EIP) of this instruction 327 /// This information is available even when CS_OPT_DETAIL = CS_OPT_OFF 328 uint64_t address; 329 330 /// Size of this instruction 331 /// This information is available even when CS_OPT_DETAIL = CS_OPT_OFF 332 uint16_t size; 333 334 /// Machine bytes of this instruction, with number of bytes indicated by @size above 335 /// This information is available even when CS_OPT_DETAIL = CS_OPT_OFF 336 uint8_t bytes[24]; 337 338 /// Ascii text of instruction mnemonic 339 /// This information is available even when CS_OPT_DETAIL = CS_OPT_OFF 340 char mnemonic[CS_MNEMONIC_SIZE]; 341 342 /// Ascii text of instruction operands 343 /// This information is available even when CS_OPT_DETAIL = CS_OPT_OFF 344 char op_str[160]; 345 346 /// Pointer to cs_detail. 347 /// NOTE: detail pointer is only valid when both requirements below are met: 348 /// (1) CS_OP_DETAIL = CS_OPT_ON 349 /// (2) Engine is not in Skipdata mode (CS_OP_SKIPDATA option set to CS_OPT_ON) 350 /// 351 /// NOTE 2: when in Skipdata mode, or when detail mode is OFF, even if this pointer 352 /// is not NULL, its content is still irrelevant. 353 cs_detail *detail; 354 } cs_insn; 355 356 357 /// Calculate the offset of a disassembled instruction in its buffer, given its position 358 /// in its array of disassembled insn 359 /// NOTE: this macro works with position (>=1), not index 360 #define CS_INSN_OFFSET(insns, post) (insns[post - 1].address - insns[0].address) 361 362 363 /// All type of errors encountered by Capstone API. 364 /// These are values returned by cs_errno() 365 typedef enum cs_err { 366 CS_ERR_OK = 0, ///< No error: everything was fine 367 CS_ERR_MEM, ///< Out-Of-Memory error: cs_open(), cs_disasm(), cs_disasm_iter() 368 CS_ERR_ARCH, ///< Unsupported architecture: cs_open() 369 CS_ERR_HANDLE, ///< Invalid handle: cs_op_count(), cs_op_index() 370 CS_ERR_CSH, ///< Invalid csh argument: cs_close(), cs_errno(), cs_option() 371 CS_ERR_MODE, ///< Invalid/unsupported mode: cs_open() 372 CS_ERR_OPTION, ///< Invalid/unsupported option: cs_option() 373 CS_ERR_DETAIL, ///< Information is unavailable because detail option is OFF 374 CS_ERR_MEMSETUP, ///< Dynamic memory management uninitialized (see CS_OPT_MEM) 375 CS_ERR_VERSION, ///< Unsupported version (bindings) 376 CS_ERR_DIET, ///< Access irrelevant data in "diet" engine 377 CS_ERR_SKIPDATA, ///< Access irrelevant data for "data" instruction in SKIPDATA mode 378 CS_ERR_X86_ATT, ///< X86 AT&T syntax is unsupported (opt-out at compile time) 379 CS_ERR_X86_INTEL, ///< X86 Intel syntax is unsupported (opt-out at compile time) 380 CS_ERR_X86_MASM, ///< X86 Masm syntax is unsupported (opt-out at compile time) 381 } cs_err; 382 383 /** 384 Return combined API version & major and minor version numbers. 385 386 @major: major number of API version 387 @minor: minor number of API version 388 389 @return hexical number as (major << 8 | minor), which encodes both 390 major & minor versions. 391 NOTE: This returned value can be compared with version number made 392 with macro CS_MAKE_VERSION 393 394 For example, second API version would return 1 in @major, and 1 in @minor 395 The return value would be 0x0101 396 397 NOTE: if you only care about returned value, but not major and minor values, 398 set both @major & @minor arguments to NULL. 399 */ 400 CAPSTONE_EXPORT 401 unsigned int CAPSTONE_API cs_version(int *major, int *minor); 402 403 404 /** 405 This API can be used to either ask for archs supported by this library, 406 or check to see if the library was compile with 'diet' option (or called 407 in 'diet' mode). 408 409 To check if a particular arch is supported by this library, set @query to 410 arch mode (CS_ARCH_* value). 411 To verify if this library supports all the archs, use CS_ARCH_ALL. 412 413 To check if this library is in 'diet' mode, set @query to CS_SUPPORT_DIET. 414 415 @return True if this library supports the given arch, or in 'diet' mode. 416 */ 417 CAPSTONE_EXPORT 418 bool CAPSTONE_API cs_support(int query); 419 420 /** 421 Initialize CS handle: this must be done before any usage of CS. 422 423 @arch: architecture type (CS_ARCH_*) 424 @mode: hardware mode. This is combined of CS_MODE_* 425 @handle: pointer to handle, which will be updated at return time 426 427 @return CS_ERR_OK on success, or other value on failure (refer to cs_err enum 428 for detailed error). 429 */ 430 CAPSTONE_EXPORT 431 cs_err CAPSTONE_API cs_open(cs_arch arch, cs_mode mode, csh *handle); 432 433 /** 434 Close CS handle: MUST do to release the handle when it is not used anymore. 435 NOTE: this must be only called when there is no longer usage of Capstone, 436 not even access to cs_insn array. The reason is the this API releases some 437 cached memory, thus access to any Capstone API after cs_close() might crash 438 your application. 439 440 In fact,this API invalidate @handle by ZERO out its value (i.e *handle = 0). 441 442 @handle: pointer to a handle returned by cs_open() 443 444 @return CS_ERR_OK on success, or other value on failure (refer to cs_err enum 445 for detailed error). 446 */ 447 CAPSTONE_EXPORT 448 cs_err CAPSTONE_API cs_close(csh *handle); 449 450 /** 451 Set option for disassembling engine at runtime 452 453 @handle: handle returned by cs_open() 454 @type: type of option to be set 455 @value: option value corresponding with @type 456 457 @return: CS_ERR_OK on success, or other value on failure. 458 Refer to cs_err enum for detailed error. 459 460 NOTE: in the case of CS_OPT_MEM, handle's value can be anything, 461 so that cs_option(handle, CS_OPT_MEM, value) can (i.e must) be called 462 even before cs_open() 463 */ 464 CAPSTONE_EXPORT 465 cs_err CAPSTONE_API cs_option(csh handle, cs_opt_type type, size_t value); 466 467 /** 468 Report the last error number when some API function fail. 469 Like glibc's errno, cs_errno might not retain its old value once accessed. 470 471 @handle: handle returned by cs_open() 472 473 @return: error code of cs_err enum type (CS_ERR_*, see above) 474 */ 475 CAPSTONE_EXPORT 476 cs_err CAPSTONE_API cs_errno(csh handle); 477 478 479 /** 480 Return a string describing given error code. 481 482 @code: error code (see CS_ERR_* above) 483 484 @return: returns a pointer to a string that describes the error code 485 passed in the argument @code 486 */ 487 CAPSTONE_EXPORT 488 const char * CAPSTONE_API cs_strerror(cs_err code); 489 490 /** 491 Disassemble binary code, given the code buffer, size, address and number 492 of instructions to be decoded. 493 This API dynamically allocate memory to contain disassembled instruction. 494 Resulting instructions will be put into @*insn 495 496 NOTE 1: this API will automatically determine memory needed to contain 497 output disassembled instructions in @insn. 498 499 NOTE 2: caller must free the allocated memory itself to avoid memory leaking. 500 501 NOTE 3: for system with scarce memory to be dynamically allocated such as 502 OS kernel or firmware, the API cs_disasm_iter() might be a better choice than 503 cs_disasm(). The reason is that with cs_disasm(), based on limited available 504 memory, we have to calculate in advance how many instructions to be disassembled, 505 which complicates things. This is especially troublesome for the case @count=0, 506 when cs_disasm() runs uncontrollably (until either end of input buffer, or 507 when it encounters an invalid instruction). 508 509 @handle: handle returned by cs_open() 510 @code: buffer containing raw binary code to be disassembled. 511 @code_size: size of the above code buffer. 512 @address: address of the first instruction in given raw code buffer. 513 @insn: array of instructions filled in by this API. 514 NOTE: @insn will be allocated by this function, and should be freed 515 with cs_free() API. 516 @count: number of instructions to be disassembled, or 0 to get all of them 517 518 @return: the number of successfully disassembled instructions, 519 or 0 if this function failed to disassemble the given code 520 521 On failure, call cs_errno() for error code. 522 */ 523 CAPSTONE_EXPORT 524 size_t CAPSTONE_API cs_disasm(csh handle, 525 const uint8_t *code, size_t code_size, 526 uint64_t address, 527 size_t count, 528 cs_insn **insn); 529 530 /** 531 Deprecated function - to be retired in the next version! 532 Use cs_disasm() instead of cs_disasm_ex() 533 */ 534 CAPSTONE_EXPORT 535 CAPSTONE_DEPRECATED 536 size_t CAPSTONE_API cs_disasm_ex(csh handle, 537 const uint8_t *code, size_t code_size, 538 uint64_t address, 539 size_t count, 540 cs_insn **insn); 541 542 /** 543 Free memory allocated by cs_malloc() or cs_disasm() (argument @insn) 544 545 @insn: pointer returned by @insn argument in cs_disasm() or cs_malloc() 546 @count: number of cs_insn structures returned by cs_disasm(), or 1 547 to free memory allocated by cs_malloc(). 548 */ 549 CAPSTONE_EXPORT 550 void CAPSTONE_API cs_free(cs_insn *insn, size_t count); 551 552 553 /** 554 Allocate memory for 1 instruction to be used by cs_disasm_iter(). 555 556 @handle: handle returned by cs_open() 557 558 NOTE: when no longer in use, you can reclaim the memory allocated for 559 this instruction with cs_free(insn, 1) 560 */ 561 CAPSTONE_EXPORT 562 cs_insn * CAPSTONE_API cs_malloc(csh handle); 563 564 /** 565 Fast API to disassemble binary code, given the code buffer, size, address 566 and number of instructions to be decoded. 567 This API puts the resulting instruction into a given cache in @insn. 568 See tests/test_iter.c for sample code demonstrating this API. 569 570 NOTE 1: this API will update @code, @size & @address to point to the next 571 instruction in the input buffer. Therefore, it is convenient to use 572 cs_disasm_iter() inside a loop to quickly iterate all the instructions. 573 While decoding one instruction at a time can also be achieved with 574 cs_disasm(count=1), some benchmarks shown that cs_disasm_iter() can be 30% 575 faster on random input. 576 577 NOTE 2: the cache in @insn can be created with cs_malloc() API. 578 579 NOTE 3: for system with scarce memory to be dynamically allocated such as 580 OS kernel or firmware, this API is recommended over cs_disasm(), which 581 allocates memory based on the number of instructions to be disassembled. 582 The reason is that with cs_disasm(), based on limited available memory, 583 we have to calculate in advance how many instructions to be disassembled, 584 which complicates things. This is especially troublesome for the case 585 @count=0, when cs_disasm() runs uncontrollably (until either end of input 586 buffer, or when it encounters an invalid instruction). 587 588 @handle: handle returned by cs_open() 589 @code: buffer containing raw binary code to be disassembled 590 @size: size of above code 591 @address: address of the first insn in given raw code buffer 592 @insn: pointer to instruction to be filled in by this API. 593 594 @return: true if this API successfully decode 1 instruction, 595 or false otherwise. 596 597 On failure, call cs_errno() for error code. 598 */ 599 CAPSTONE_EXPORT 600 bool CAPSTONE_API cs_disasm_iter(csh handle, 601 const uint8_t **code, size_t *size, 602 uint64_t *address, cs_insn *insn); 603 604 /** 605 Return friendly name of register in a string. 606 Find the instruction id from header file of corresponding architecture (arm.h for ARM, 607 x86.h for X86, ...) 608 609 WARN: when in 'diet' mode, this API is irrelevant because engine does not 610 store register name. 611 612 @handle: handle returned by cs_open() 613 @reg_id: register id 614 615 @return: string name of the register, or NULL if @reg_id is invalid. 616 */ 617 CAPSTONE_EXPORT 618 const char * CAPSTONE_API cs_reg_name(csh handle, unsigned int reg_id); 619 620 /** 621 Return friendly name of an instruction in a string. 622 Find the instruction id from header file of corresponding architecture (arm.h for ARM, x86.h for X86, ...) 623 624 WARN: when in 'diet' mode, this API is irrelevant because the engine does not 625 store instruction name. 626 627 @handle: handle returned by cs_open() 628 @insn_id: instruction id 629 630 @return: string name of the instruction, or NULL if @insn_id is invalid. 631 */ 632 CAPSTONE_EXPORT 633 const char * CAPSTONE_API cs_insn_name(csh handle, unsigned int insn_id); 634 635 /** 636 Return friendly name of a group id (that an instruction can belong to) 637 Find the group id from header file of corresponding architecture (arm.h for ARM, x86.h for X86, ...) 638 639 WARN: when in 'diet' mode, this API is irrelevant because the engine does not 640 store group name. 641 642 @handle: handle returned by cs_open() 643 @group_id: group id 644 645 @return: string name of the group, or NULL if @group_id is invalid. 646 */ 647 CAPSTONE_EXPORT 648 const char * CAPSTONE_API cs_group_name(csh handle, unsigned int group_id); 649 650 /** 651 Check if a disassembled instruction belong to a particular group. 652 Find the group id from header file of corresponding architecture (arm.h for ARM, x86.h for X86, ...) 653 Internally, this simply verifies if @group_id matches any member of insn->groups array. 654 655 NOTE: this API is only valid when detail option is ON (which is OFF by default). 656 657 WARN: when in 'diet' mode, this API is irrelevant because the engine does not 658 update @groups array. 659 660 @handle: handle returned by cs_open() 661 @insn: disassembled instruction structure received from cs_disasm() or cs_disasm_iter() 662 @group_id: group that you want to check if this instruction belong to. 663 664 @return: true if this instruction indeed belongs to the given group, or false otherwise. 665 */ 666 CAPSTONE_EXPORT 667 bool CAPSTONE_API cs_insn_group(csh handle, const cs_insn *insn, unsigned int group_id); 668 669 /** 670 Check if a disassembled instruction IMPLICITLY used a particular register. 671 Find the register id from header file of corresponding architecture (arm.h for ARM, x86.h for X86, ...) 672 Internally, this simply verifies if @reg_id matches any member of insn->regs_read array. 673 674 NOTE: this API is only valid when detail option is ON (which is OFF by default) 675 676 WARN: when in 'diet' mode, this API is irrelevant because the engine does not 677 update @regs_read array. 678 679 @insn: disassembled instruction structure received from cs_disasm() or cs_disasm_iter() 680 @reg_id: register that you want to check if this instruction used it. 681 682 @return: true if this instruction indeed implicitly used the given register, or false otherwise. 683 */ 684 CAPSTONE_EXPORT 685 bool CAPSTONE_API cs_reg_read(csh handle, const cs_insn *insn, unsigned int reg_id); 686 687 /** 688 Check if a disassembled instruction IMPLICITLY modified a particular register. 689 Find the register id from header file of corresponding architecture (arm.h for ARM, x86.h for X86, ...) 690 Internally, this simply verifies if @reg_id matches any member of insn->regs_write array. 691 692 NOTE: this API is only valid when detail option is ON (which is OFF by default) 693 694 WARN: when in 'diet' mode, this API is irrelevant because the engine does not 695 update @regs_write array. 696 697 @insn: disassembled instruction structure received from cs_disasm() or cs_disasm_iter() 698 @reg_id: register that you want to check if this instruction modified it. 699 700 @return: true if this instruction indeed implicitly modified the given register, or false otherwise. 701 */ 702 CAPSTONE_EXPORT 703 bool CAPSTONE_API cs_reg_write(csh handle, const cs_insn *insn, unsigned int reg_id); 704 705 /** 706 Count the number of operands of a given type. 707 Find the operand type in header file of corresponding architecture (arm.h for ARM, x86.h for X86, ...) 708 709 NOTE: this API is only valid when detail option is ON (which is OFF by default) 710 711 @handle: handle returned by cs_open() 712 @insn: disassembled instruction structure received from cs_disasm() or cs_disasm_iter() 713 @op_type: Operand type to be found. 714 715 @return: number of operands of given type @op_type in instruction @insn, 716 or -1 on failure. 717 */ 718 CAPSTONE_EXPORT 719 int CAPSTONE_API cs_op_count(csh handle, const cs_insn *insn, unsigned int op_type); 720 721 /** 722 Retrieve the position of operand of given type in <arch>.operands[] array. 723 Later, the operand can be accessed using the returned position. 724 Find the operand type in header file of corresponding architecture (arm.h for ARM, x86.h for X86, ...) 725 726 NOTE: this API is only valid when detail option is ON (which is OFF by default) 727 728 @handle: handle returned by cs_open() 729 @insn: disassembled instruction structure received from cs_disasm() or cs_disasm_iter() 730 @op_type: Operand type to be found. 731 @position: position of the operand to be found. This must be in the range 732 [1, cs_op_count(handle, insn, op_type)] 733 734 @return: index of operand of given type @op_type in <arch>.operands[] array 735 in instruction @insn, or -1 on failure. 736 */ 737 CAPSTONE_EXPORT 738 int CAPSTONE_API cs_op_index(csh handle, const cs_insn *insn, unsigned int op_type, 739 unsigned int position); 740 741 /// Type of array to keep the list of registers 742 typedef uint16_t cs_regs[64]; 743 744 /** 745 Retrieve all the registers accessed by an instruction, either explicitly or 746 implicitly. 747 748 WARN: when in 'diet' mode, this API is irrelevant because engine does not 749 store registers. 750 751 @handle: handle returned by cs_open() 752 @insn: disassembled instruction structure returned from cs_disasm() or cs_disasm_iter() 753 @regs_read: on return, this array contains all registers read by instruction. 754 @regs_read_count: number of registers kept inside @regs_read array. 755 @regs_write: on return, this array contains all registers written by instruction. 756 @regs_write_count: number of registers kept inside @regs_write array. 757 758 @return CS_ERR_OK on success, or other value on failure (refer to cs_err enum 759 for detailed error). 760 */ 761 CAPSTONE_EXPORT 762 cs_err CAPSTONE_API cs_regs_access(csh handle, const cs_insn *insn, 763 cs_regs regs_read, uint8_t *regs_read_count, 764 cs_regs regs_write, uint8_t *regs_write_count); 765 766 #ifdef __cplusplus 767 } 768 #endif 769 770 #endif 771