1 /* Capstone Disassembler Engine */
2 /* By Nguyen Anh Quynh <aquynh@gmail.com>, 2013 */
3
4 #include <stdio.h>
5 #include <stdlib.h>
6
7 #include <capstone/platform.h>
8 #include <capstone/capstone.h>
9
10 static csh handle;
11
12 struct platform {
13 cs_arch arch;
14 cs_mode mode;
15 unsigned char *code;
16 size_t size;
17 const char *comment;
18 cs_opt_type opt_type;
19 cs_opt_value opt_value;
20 };
21
print_string_hex(const char * comment,unsigned char * str,size_t len)22 static void print_string_hex(const char *comment, unsigned char *str, size_t len)
23 {
24 unsigned char *c;
25
26 printf("%s", comment);
27 for (c = str; c < str + len; c++) {
28 printf("0x%02x ", *c & 0xff);
29 }
30
31 printf("\n");
32 }
33
get_eflag_name(uint64_t flag)34 static const char *get_eflag_name(uint64_t flag)
35 {
36 switch(flag) {
37 default:
38 return NULL;
39 case X86_EFLAGS_UNDEFINED_OF:
40 return "UNDEF_OF";
41 case X86_EFLAGS_UNDEFINED_SF:
42 return "UNDEF_SF";
43 case X86_EFLAGS_UNDEFINED_ZF:
44 return "UNDEF_ZF";
45 case X86_EFLAGS_MODIFY_AF:
46 return "MOD_AF";
47 case X86_EFLAGS_UNDEFINED_PF:
48 return "UNDEF_PF";
49 case X86_EFLAGS_MODIFY_CF:
50 return "MOD_CF";
51 case X86_EFLAGS_MODIFY_SF:
52 return "MOD_SF";
53 case X86_EFLAGS_MODIFY_ZF:
54 return "MOD_ZF";
55 case X86_EFLAGS_UNDEFINED_AF:
56 return "UNDEF_AF";
57 case X86_EFLAGS_MODIFY_PF:
58 return "MOD_PF";
59 case X86_EFLAGS_UNDEFINED_CF:
60 return "UNDEF_CF";
61 case X86_EFLAGS_MODIFY_OF:
62 return "MOD_OF";
63 case X86_EFLAGS_RESET_OF:
64 return "RESET_OF";
65 case X86_EFLAGS_RESET_CF:
66 return "RESET_CF";
67 case X86_EFLAGS_RESET_DF:
68 return "RESET_DF";
69 case X86_EFLAGS_RESET_IF:
70 return "RESET_IF";
71 case X86_EFLAGS_TEST_OF:
72 return "TEST_OF";
73 case X86_EFLAGS_TEST_SF:
74 return "TEST_SF";
75 case X86_EFLAGS_TEST_ZF:
76 return "TEST_ZF";
77 case X86_EFLAGS_TEST_PF:
78 return "TEST_PF";
79 case X86_EFLAGS_TEST_CF:
80 return "TEST_CF";
81 case X86_EFLAGS_RESET_SF:
82 return "RESET_SF";
83 case X86_EFLAGS_RESET_AF:
84 return "RESET_AF";
85 case X86_EFLAGS_RESET_TF:
86 return "RESET_TF";
87 case X86_EFLAGS_RESET_NT:
88 return "RESET_NT";
89 case X86_EFLAGS_PRIOR_OF:
90 return "PRIOR_OF";
91 case X86_EFLAGS_PRIOR_SF:
92 return "PRIOR_SF";
93 case X86_EFLAGS_PRIOR_ZF:
94 return "PRIOR_ZF";
95 case X86_EFLAGS_PRIOR_AF:
96 return "PRIOR_AF";
97 case X86_EFLAGS_PRIOR_PF:
98 return "PRIOR_PF";
99 case X86_EFLAGS_PRIOR_CF:
100 return "PRIOR_CF";
101 case X86_EFLAGS_PRIOR_TF:
102 return "PRIOR_TF";
103 case X86_EFLAGS_PRIOR_IF:
104 return "PRIOR_IF";
105 case X86_EFLAGS_PRIOR_DF:
106 return "PRIOR_DF";
107 case X86_EFLAGS_TEST_NT:
108 return "TEST_NT";
109 case X86_EFLAGS_TEST_DF:
110 return "TEST_DF";
111 case X86_EFLAGS_RESET_PF:
112 return "RESET_PF";
113 case X86_EFLAGS_PRIOR_NT:
114 return "PRIOR_NT";
115 case X86_EFLAGS_MODIFY_TF:
116 return "MOD_TF";
117 case X86_EFLAGS_MODIFY_IF:
118 return "MOD_IF";
119 case X86_EFLAGS_MODIFY_DF:
120 return "MOD_DF";
121 case X86_EFLAGS_MODIFY_NT:
122 return "MOD_NT";
123 case X86_EFLAGS_MODIFY_RF:
124 return "MOD_RF";
125 case X86_EFLAGS_SET_CF:
126 return "SET_CF";
127 case X86_EFLAGS_SET_DF:
128 return "SET_DF";
129 case X86_EFLAGS_SET_IF:
130 return "SET_IF";
131 }
132 }
133
print_insn_detail(csh ud,cs_mode mode,cs_insn * ins)134 static void print_insn_detail(csh ud, cs_mode mode, cs_insn *ins)
135 {
136 int count, i;
137 cs_x86 *x86;
138 cs_regs regs_read, regs_write;
139 uint8_t regs_read_count, regs_write_count;
140
141 // detail can be NULL on "data" instruction if SKIPDATA option is turned ON
142 if (ins->detail == NULL)
143 return;
144
145 x86 = &(ins->detail->x86);
146
147 print_string_hex("\tPrefix:", x86->prefix, 4);
148
149 print_string_hex("\tOpcode:", x86->opcode, 4);
150
151 printf("\trex: 0x%x\n", x86->rex);
152
153 printf("\taddr_size: %u\n", x86->addr_size);
154 printf("\tmodrm: 0x%x\n", x86->modrm);
155 if (x86->encoding.modrm_offset != 0) {
156 printf("\tmodrm_offset: 0x%x\n", x86->encoding.modrm_offset);
157 }
158
159 printf("\tdisp: 0x%" PRIx64 "\n", x86->disp);
160 if (x86->encoding.disp_offset != 0) {
161 printf("\tdisp_offset: 0x%x\n", x86->encoding.disp_offset);
162 }
163
164 if (x86->encoding.disp_size != 0) {
165 printf("\tdisp_size: 0x%x\n", x86->encoding.disp_size);
166 }
167
168 // SIB is not available in 16-bit mode
169 if ((mode & CS_MODE_16) == 0) {
170 printf("\tsib: 0x%x\n", x86->sib);
171 if (x86->sib_base != X86_REG_INVALID)
172 printf("\t\tsib_base: %s\n", cs_reg_name(handle, x86->sib_base));
173 if (x86->sib_index != X86_REG_INVALID)
174 printf("\t\tsib_index: %s\n", cs_reg_name(handle, x86->sib_index));
175 if (x86->sib_scale != 0)
176 printf("\t\tsib_scale: %d\n", x86->sib_scale);
177 }
178
179 // XOP code condition
180 if (x86->xop_cc != X86_XOP_CC_INVALID) {
181 printf("\txop_cc: %u\n", x86->xop_cc);
182 }
183
184 // SSE code condition
185 if (x86->sse_cc != X86_SSE_CC_INVALID) {
186 printf("\tsse_cc: %u\n", x86->sse_cc);
187 }
188
189 // AVX code condition
190 if (x86->avx_cc != X86_AVX_CC_INVALID) {
191 printf("\tavx_cc: %u\n", x86->avx_cc);
192 }
193
194 // AVX Suppress All Exception
195 if (x86->avx_sae) {
196 printf("\tavx_sae: %u\n", x86->avx_sae);
197 }
198
199 // AVX Rounding Mode
200 if (x86->avx_rm != X86_AVX_RM_INVALID) {
201 printf("\tavx_rm: %u\n", x86->avx_rm);
202 }
203
204 // Print out all immediate operands
205 count = cs_op_count(ud, ins, X86_OP_IMM);
206 if (count) {
207 printf("\timm_count: %u\n", count);
208 for (i = 1; i < count + 1; i++) {
209 int index = cs_op_index(ud, ins, X86_OP_IMM, i);
210 printf("\t\timms[%u]: 0x%" PRIx64 "\n", i, x86->operands[index].imm);
211 if (x86->encoding.imm_offset != 0) {
212 printf("\timm_offset: 0x%x\n", x86->encoding.imm_offset);
213 }
214
215 if (x86->encoding.imm_size != 0) {
216 printf("\timm_size: 0x%x\n", x86->encoding.imm_size);
217 }
218 }
219 }
220
221 if (x86->op_count)
222 printf("\top_count: %u\n", x86->op_count);
223
224 // Print out all operands
225 for (i = 0; i < x86->op_count; i++) {
226 cs_x86_op *op = &(x86->operands[i]);
227
228 switch((int)op->type) {
229 case X86_OP_REG:
230 printf("\t\toperands[%u].type: REG = %s\n", i, cs_reg_name(handle, op->reg));
231 break;
232 case X86_OP_IMM:
233 printf("\t\toperands[%u].type: IMM = 0x%" PRIx64 "\n", i, op->imm);
234 break;
235 case X86_OP_MEM:
236 printf("\t\toperands[%u].type: MEM\n", i);
237 if (op->mem.segment != X86_REG_INVALID)
238 printf("\t\t\toperands[%u].mem.segment: REG = %s\n", i, cs_reg_name(handle, op->mem.segment));
239 if (op->mem.base != X86_REG_INVALID)
240 printf("\t\t\toperands[%u].mem.base: REG = %s\n", i, cs_reg_name(handle, op->mem.base));
241 if (op->mem.index != X86_REG_INVALID)
242 printf("\t\t\toperands[%u].mem.index: REG = %s\n", i, cs_reg_name(handle, op->mem.index));
243 if (op->mem.scale != 1)
244 printf("\t\t\toperands[%u].mem.scale: %u\n", i, op->mem.scale);
245 if (op->mem.disp != 0)
246 printf("\t\t\toperands[%u].mem.disp: 0x%" PRIx64 "\n", i, op->mem.disp);
247 break;
248 default:
249 break;
250 }
251
252 // AVX broadcast type
253 if (op->avx_bcast != X86_AVX_BCAST_INVALID)
254 printf("\t\toperands[%u].avx_bcast: %u\n", i, op->avx_bcast);
255
256 // AVX zero opmask {z}
257 if (op->avx_zero_opmask != false)
258 printf("\t\toperands[%u].avx_zero_opmask: TRUE\n", i);
259
260 printf("\t\toperands[%u].size: %u\n", i, op->size);
261
262 switch(op->access) {
263 default:
264 break;
265 case CS_AC_READ:
266 printf("\t\toperands[%u].access: READ\n", i);
267 break;
268 case CS_AC_WRITE:
269 printf("\t\toperands[%u].access: WRITE\n", i);
270 break;
271 case CS_AC_READ | CS_AC_WRITE:
272 printf("\t\toperands[%u].access: READ | WRITE\n", i);
273 break;
274 }
275 }
276
277 // Print out all registers accessed by this instruction (either implicit or explicit)
278 if (!cs_regs_access(ud, ins,
279 regs_read, ®s_read_count,
280 regs_write, ®s_write_count)) {
281 if (regs_read_count) {
282 printf("\tRegisters read:");
283 for(i = 0; i < regs_read_count; i++) {
284 printf(" %s", cs_reg_name(handle, regs_read[i]));
285 }
286 printf("\n");
287 }
288
289 if (regs_write_count) {
290 printf("\tRegisters modified:");
291 for(i = 0; i < regs_write_count; i++) {
292 printf(" %s", cs_reg_name(handle, regs_write[i]));
293 }
294 printf("\n");
295 }
296 }
297
298 if (x86->eflags) {
299 printf("\tEFLAGS:");
300 for(i = 0; i <= 45; i++)
301 if (x86->eflags & ((uint64_t)1 << i)) {
302 printf(" %s", get_eflag_name((uint64_t)1 << i));
303 }
304 printf("\n");
305 }
306
307 printf("\n");
308 }
309
test()310 static void test()
311 {
312 //#define X86_CODE32 "\x01\xd8\x81\xc6\x34\x12\x00\x00\x05\x78\x56\x00\x00"
313 //#define X86_CODE32 "\x05\x78\x56\x00\x00"
314 //#define X86_CODE32 "\x01\xd8"
315 //#define X86_CODE32 "\x05\x23\x01\x00\x00"
316 //#define X86_CODE32 "\x8d\x87\x89\x67\x00\x00"
317 //#define X86_CODE32 "\xa1\x13\x48\x6d\x3a\x8b\x81\x23\x01\x00\x00\x8b\x84\x39\x23\x01\x00\x00"
318 //#define X86_CODE32 "\xb4\xc6" // mov ah, 0x6c
319 //#define X86_CODE32 "\x77\x04" // ja +6
320 #define X86_CODE64 "\x55\x48\x8b\x05\xb8\x13\x00\x00\xe9\xea\xbe\xad\xde\xff\x25\x23\x01\x00\x00\xe8\xdf\xbe\xad\xde\x74\xff"
321 //#define X86_CODE64 "\xe9\x79\xff\xff\xff" // jmp 0xf7e
322
323 #define X86_CODE16 "\x8d\x4c\x32\x08\x01\xd8\x81\xc6\x34\x12\x00\x00\x05\x23\x01\x00\x00\x36\x8b\x84\x91\x23\x01\x00\x00\x41\x8d\x84\x39\x89\x67\x00\x00\x8d\x87\x89\x67\x00\x00\xb4\xc6\x66\xe9\xb8\x00\x00\x00\x67\xff\xa0\x23\x01\x00\x00\x66\xe8\xcb\x00\x00\x00\x74\xfc"
324 #define X86_CODE32 "\x8d\x4c\x32\x08\x01\xd8\x81\xc6\x34\x12\x00\x00\x05\x23\x01\x00\x00\x36\x8b\x84\x91\x23\x01\x00\x00\x41\x8d\x84\x39\x89\x67\x00\x00\x8d\x87\x89\x67\x00\x00\xb4\xc6\xe9\xea\xbe\xad\xde\xff\xa0\x23\x01\x00\x00\xe8\xdf\xbe\xad\xde\x74\xff"
325 //#define X86_CODE32 "\x05\x23\x01\x00\x00\x0f\x01\xda"
326 //#define X86_CODE32 "\x0f\xa7\xc0" // xstorerng
327 //#define X86_CODE32 "\x64\xa1\x18\x00\x00\x00" // mov eax, dword ptr fs:[18]
328 //#define X86_CODE32 "\x64\xa3\x00\x00\x00\x00" // mov [fs:0x0], eax
329 //#define X86_CODE32 "\xd1\xe1" // shl ecx, 1
330 //#define X86_CODE32 "\xd1\xc8" // ror eax, 1
331 //#define X86_CODE32 "\x83\xC0\x80" // add eax, -x80
332 //#define X86_CODE32 "\xe8\x26\xfe\xff\xff" // call 0xe2b
333 //#define X86_CODE32 "\xcd\x80" // int 0x80
334 //#define X86_CODE32 "\x24\xb8" // and $0xb8,%al
335 //#define X86_CODE32 "\xf0\x01\xd8" // lock add eax,ebx
336 //#define X86_CODE32 "\xf3\xaa" // rep stosb
337 //#define X86_CODE32 "\x81\xc6\x23\x01\x00\x00"
338
339 struct platform platforms[] = {
340 {
341 CS_ARCH_X86,
342 CS_MODE_16,
343 (unsigned char *)X86_CODE16,
344 sizeof(X86_CODE16) - 1,
345 "X86 16bit (Intel syntax)"
346 },
347 {
348 CS_ARCH_X86,
349 CS_MODE_32,
350 (unsigned char *)X86_CODE32,
351 sizeof(X86_CODE32) - 1,
352 "X86 32 (AT&T syntax)",
353 CS_OPT_SYNTAX,
354 CS_OPT_SYNTAX_ATT,
355 },
356 {
357 CS_ARCH_X86,
358 CS_MODE_32,
359 (unsigned char *)X86_CODE32,
360 sizeof(X86_CODE32) - 1,
361 "X86 32 (Intel syntax)"
362 },
363 {
364 CS_ARCH_X86,
365 CS_MODE_64,
366 (unsigned char *)X86_CODE64,
367 sizeof(X86_CODE64) - 1,
368 "X86 64 (Intel syntax)"
369 },
370 };
371
372 uint64_t address = 0x1000;
373 cs_insn *insn;
374 int i;
375 size_t count;
376
377 for (i = 0; i < sizeof(platforms)/sizeof(platforms[0]); i++) {
378 cs_err err = cs_open(platforms[i].arch, platforms[i].mode, &handle);
379 if (err) {
380 printf("Failed on cs_open() with error returned: %u\n", err);
381 abort();
382 }
383
384 if (platforms[i].opt_type)
385 cs_option(handle, platforms[i].opt_type, platforms[i].opt_value);
386
387 cs_option(handle, CS_OPT_DETAIL, CS_OPT_ON);
388
389 count = cs_disasm(handle, platforms[i].code, platforms[i].size, address, 0, &insn);
390 if (count) {
391 size_t j;
392
393 printf("****************\n");
394 printf("Platform: %s\n", platforms[i].comment);
395 print_string_hex("Code:", platforms[i].code, platforms[i].size);
396 printf("Disasm:\n");
397
398 for (j = 0; j < count; j++) {
399 printf("0x%" PRIx64 ":\t%s\t%s\n\n", insn[j].address, insn[j].mnemonic, insn[j].op_str);
400 print_insn_detail(handle, platforms[i].mode, &insn[j]);
401 }
402 printf("0x%" PRIx64 ":\n", insn[j-1].address + insn[j-1].size);
403
404 // free memory allocated by cs_disasm()
405 cs_free(insn, count);
406 } else {
407 printf("****************\n");
408 printf("Platform: %s\n", platforms[i].comment);
409 print_string_hex("Code:", platforms[i].code, platforms[i].size);
410 printf("ERROR: Failed to disasm given code!\n");
411 abort();
412 }
413
414 printf("\n");
415
416 cs_close(&handle);
417 }
418 }
419
main()420 int main()
421 {
422 test();
423
424 return 0;
425 }
426