1 | /** @file
|
---|
2 | *
|
---|
3 | * VBox disassembler:
|
---|
4 | * Main
|
---|
5 | */
|
---|
6 |
|
---|
7 | /*
|
---|
8 | * Copyright (C) 2006-2007 innotek GmbH
|
---|
9 | *
|
---|
10 | * This file is part of VirtualBox Open Source Edition (OSE), as
|
---|
11 | * available from http://www.virtualbox.org. This file is free software;
|
---|
12 | * you can redistribute it and/or modify it under the terms of the GNU
|
---|
13 | * General Public License (GPL) as published by the Free Software
|
---|
14 | * Foundation, in version 2 as it comes in the "COPYING" file of the
|
---|
15 | * VirtualBox OSE distribution. VirtualBox OSE is distributed in the
|
---|
16 | * hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
|
---|
17 | */
|
---|
18 |
|
---|
19 |
|
---|
20 | /*******************************************************************************
|
---|
21 | * Header Files *
|
---|
22 | *******************************************************************************/
|
---|
23 | #ifdef USING_VISUAL_STUDIO
|
---|
24 | # include <stdafx.h>
|
---|
25 | #endif
|
---|
26 | #include <VBox/dis.h>
|
---|
27 | #include <VBox/disopcode.h>
|
---|
28 | #include <VBox/err.h>
|
---|
29 | #include <iprt/assert.h>
|
---|
30 | #include <iprt/string.h>
|
---|
31 | #include "DisasmInternal.h"
|
---|
32 | #include "DisasmTables.h"
|
---|
33 |
|
---|
34 |
|
---|
35 | /**
|
---|
36 | * Disassembles a code block.
|
---|
37 | *
|
---|
38 | * @returns VBox error code
|
---|
39 | * @param pCpu Pointer to cpu structure which have DISCPUSTATE::mode
|
---|
40 | * set correctly.
|
---|
41 | * @param pvCodeBlock Pointer to the strunction to disassemble.
|
---|
42 | * @param cbMax Maximum number of bytes to disassemble.
|
---|
43 | * @param pcbSize Where to store the size of the instruction.
|
---|
44 | * NULL is allowed.
|
---|
45 | *
|
---|
46 | *
|
---|
47 | * @todo Define output callback.
|
---|
48 | * @todo Using signed integers as sizes is a bit odd. There are still
|
---|
49 | * some GCC warnings about mixing signed and unsigend integers.
|
---|
50 | * @todo Need to extend this interface to include a code address so we
|
---|
51 | * can dissassemble GC code. Perhaps a new function is better...
|
---|
52 | * @remark cbMax isn't respected as a boundry. DISInstr() will read beyond cbMax.
|
---|
53 | * This means *pcbSize >= cbMax sometimes.
|
---|
54 | */
|
---|
55 | DISDECL(int) DISBlock(PDISCPUSTATE pCpu, RTUINTPTR pvCodeBlock, unsigned cbMax, unsigned *pSize)
|
---|
56 | {
|
---|
57 | unsigned i = 0;
|
---|
58 | char szOutput[256];
|
---|
59 |
|
---|
60 | while (i < cbMax)
|
---|
61 | {
|
---|
62 | unsigned cbInstr;
|
---|
63 | int rc = DISInstr(pCpu, pvCodeBlock + i, 0, &cbInstr, szOutput);
|
---|
64 | if (VBOX_FAILURE(rc))
|
---|
65 | return rc;
|
---|
66 |
|
---|
67 | i += cbInstr;
|
---|
68 | }
|
---|
69 |
|
---|
70 | if (pSize)
|
---|
71 | *pSize = i;
|
---|
72 | return true;
|
---|
73 | }
|
---|
74 |
|
---|
75 | /**
|
---|
76 | * Disassembles one instruction
|
---|
77 | *
|
---|
78 | * @returns VBox error code
|
---|
79 | * @param pCpu Pointer to cpu structure which have DISCPUSTATE::mode
|
---|
80 | * set correctly.
|
---|
81 | * @param pu8Instruction Pointer to the strunction to disassemble.
|
---|
82 | * @param u32EipOffset Offset to add to instruction address to get the real virtual address
|
---|
83 | * @param pcbSize Where to store the size of the instruction.
|
---|
84 | * NULL is allowed.
|
---|
85 | * @param pszOutput Storage for disassembled instruction
|
---|
86 | *
|
---|
87 | * @todo Define output callback.
|
---|
88 | */
|
---|
89 | DISDECL(int) DISInstr(PDISCPUSTATE pCpu, RTUINTPTR pu8Instruction, unsigned u32EipOffset, unsigned *pcbSize,
|
---|
90 | char *pszOutput)
|
---|
91 | {
|
---|
92 | return DISInstrEx(pCpu, pu8Instruction, u32EipOffset, pcbSize, pszOutput, OPTYPE_ALL);
|
---|
93 | }
|
---|
94 |
|
---|
95 | /**
|
---|
96 | * Disassembles one instruction; only fully disassembly an instruction if it matches the filter criteria
|
---|
97 | *
|
---|
98 | * @returns VBox error code
|
---|
99 | * @param pCpu Pointer to cpu structure which have DISCPUSTATE::mode
|
---|
100 | * set correctly.
|
---|
101 | * @param pu8Instruction Pointer to the strunction to disassemble.
|
---|
102 | * @param u32EipOffset Offset to add to instruction address to get the real virtual address
|
---|
103 | * @param pcbSize Where to store the size of the instruction.
|
---|
104 | * NULL is allowed.
|
---|
105 | * @param pszOutput Storage for disassembled instruction
|
---|
106 | * @param uFilter Instruction type filter
|
---|
107 | *
|
---|
108 | * @todo Define output callback.
|
---|
109 | */
|
---|
110 | DISDECL(int) DISInstrEx(PDISCPUSTATE pCpu, RTUINTPTR pu8Instruction, unsigned u32EipOffset, unsigned *pcbSize,
|
---|
111 | char *pszOutput, unsigned uFilter)
|
---|
112 | {
|
---|
113 | unsigned i = 0, prefixbytes;
|
---|
114 | unsigned idx, inc;
|
---|
115 | #ifdef __L4ENV__
|
---|
116 | jmp_buf jumpbuffer;
|
---|
117 | #endif
|
---|
118 |
|
---|
119 | //reset instruction settings
|
---|
120 | pCpu->prefix = PREFIX_NONE;
|
---|
121 | pCpu->prefix_seg = 0;
|
---|
122 | pCpu->addrmode = pCpu->mode;
|
---|
123 | pCpu->opmode = pCpu->mode;
|
---|
124 | pCpu->ModRM = 0;
|
---|
125 | pCpu->SIB = 0;
|
---|
126 | pCpu->lastprefix = 0;
|
---|
127 | pCpu->param1.parval = 0;
|
---|
128 | pCpu->param2.parval = 0;
|
---|
129 | pCpu->param3.parval = 0;
|
---|
130 | pCpu->param1.szParam[0] = 0;
|
---|
131 | pCpu->param2.szParam[0] = 0;
|
---|
132 | pCpu->param3.szParam[0] = 0;
|
---|
133 | pCpu->param1.size = 0;
|
---|
134 | pCpu->param2.size = 0;
|
---|
135 | pCpu->param3.size = 0;
|
---|
136 | pCpu->param1.flags = 0;
|
---|
137 | pCpu->param2.flags = 0;
|
---|
138 | pCpu->param3.flags = 0;
|
---|
139 | pCpu->uFilter = uFilter;
|
---|
140 | pCpu->pfnDisasmFnTable = pfnFullDisasm;
|
---|
141 |
|
---|
142 | if (pszOutput)
|
---|
143 | *pszOutput = '\0';
|
---|
144 |
|
---|
145 | prefixbytes = 0;
|
---|
146 | #ifndef __L4ENV__ /* Unfortunately, we have no exception handling in l4env */
|
---|
147 | try
|
---|
148 | #else
|
---|
149 | pCpu->pJumpBuffer = &jumpbuffer;
|
---|
150 | if (setjmp(jumpbuffer) == 0)
|
---|
151 | #endif
|
---|
152 | {
|
---|
153 | while(1)
|
---|
154 | {
|
---|
155 | uint8_t codebyte = DISReadByte(pCpu, pu8Instruction+i);
|
---|
156 | uint8_t opcode = g_aOneByteMapX86[codebyte].opcode;
|
---|
157 |
|
---|
158 | /* Hardcoded assumption about OP_* values!! */
|
---|
159 | if (opcode <= OP_LOCK)
|
---|
160 | {
|
---|
161 | pCpu->lastprefix = opcode;
|
---|
162 | switch(opcode)
|
---|
163 | {
|
---|
164 | case OP_INVALID:
|
---|
165 | #if 0 //defined (DEBUG_Sander)
|
---|
166 | AssertMsgFailed(("Invalid opcode!!\n"));
|
---|
167 | #endif
|
---|
168 | return VERR_DIS_INVALID_OPCODE;
|
---|
169 |
|
---|
170 | // segment override prefix byte
|
---|
171 | case OP_SEG:
|
---|
172 | pCpu->prefix_seg = g_aOneByteMapX86[codebyte].param1 - OP_PARM_REG_SEG_START;
|
---|
173 | pCpu->prefix |= PREFIX_SEG;
|
---|
174 | i += sizeof(uint8_t);
|
---|
175 | prefixbytes++;
|
---|
176 | continue; //fetch the next byte
|
---|
177 |
|
---|
178 | // lock prefix byte
|
---|
179 | case OP_LOCK:
|
---|
180 | pCpu->prefix |= PREFIX_LOCK;
|
---|
181 | i += sizeof(uint8_t);
|
---|
182 | prefixbytes++;
|
---|
183 | continue; //fetch the next byte
|
---|
184 |
|
---|
185 | // address size override prefix byte
|
---|
186 | case OP_ADRSIZE:
|
---|
187 | pCpu->prefix |= PREFIX_ADDRSIZE;
|
---|
188 | if(pCpu->mode == CPUMODE_16BIT)
|
---|
189 | pCpu->addrmode = CPUMODE_32BIT;
|
---|
190 | else pCpu->addrmode = CPUMODE_16BIT;
|
---|
191 | i += sizeof(uint8_t);
|
---|
192 | prefixbytes++;
|
---|
193 | continue; //fetch the next byte
|
---|
194 |
|
---|
195 | // operand size override prefix byte
|
---|
196 | case OP_OPSIZE:
|
---|
197 | pCpu->prefix |= PREFIX_OPSIZE;
|
---|
198 | if(pCpu->mode == CPUMODE_16BIT)
|
---|
199 | pCpu->opmode = CPUMODE_32BIT;
|
---|
200 | else pCpu->opmode = CPUMODE_16BIT;
|
---|
201 | i += sizeof(uint8_t);
|
---|
202 | prefixbytes++;
|
---|
203 | continue; //fetch the next byte
|
---|
204 |
|
---|
205 | // rep and repne are not really prefixes, but we'll treat them as such
|
---|
206 | case OP_REPE:
|
---|
207 | pCpu->prefix |= PREFIX_REP;
|
---|
208 | i += sizeof(uint8_t);
|
---|
209 | prefixbytes += sizeof(uint8_t);
|
---|
210 | continue; //fetch the next byte
|
---|
211 |
|
---|
212 | case OP_REPNE:
|
---|
213 | pCpu->prefix |= PREFIX_REPNE;
|
---|
214 | i += sizeof(uint8_t);
|
---|
215 | prefixbytes += sizeof(uint8_t);
|
---|
216 | continue; //fetch the next byte
|
---|
217 | }
|
---|
218 | }
|
---|
219 |
|
---|
220 | idx = i;
|
---|
221 | i += sizeof(uint8_t); //first opcode byte
|
---|
222 |
|
---|
223 | pCpu->opcode = codebyte;
|
---|
224 | /* Prefix byte(s) is/are part of the instruction. */
|
---|
225 | pCpu->opaddr = pu8Instruction + idx + u32EipOffset - prefixbytes;
|
---|
226 |
|
---|
227 | inc = ParseInstruction(pu8Instruction + i, &g_aOneByteMapX86[pCpu->opcode], pCpu);
|
---|
228 |
|
---|
229 | pCpu->opsize = prefixbytes + inc + sizeof(uint8_t);
|
---|
230 |
|
---|
231 | if(pszOutput) {
|
---|
232 | disasmSprintf(pszOutput, pu8Instruction+i-1-prefixbytes, pCpu, &pCpu->param1, &pCpu->param2, &pCpu->param3);
|
---|
233 | }
|
---|
234 |
|
---|
235 | i += inc;
|
---|
236 | prefixbytes = 0;
|
---|
237 | break;
|
---|
238 | }
|
---|
239 | }
|
---|
240 | #ifndef __L4ENV__
|
---|
241 | catch(...)
|
---|
242 | #else
|
---|
243 | else /* setjmp has returned a non-zero value: an exception occured */
|
---|
244 | #endif
|
---|
245 | {
|
---|
246 | if (pcbSize)
|
---|
247 | *pcbSize = 0;
|
---|
248 | return VERR_DIS_GEN_FAILURE;
|
---|
249 | }
|
---|
250 |
|
---|
251 | if (pcbSize)
|
---|
252 | *pcbSize = i;
|
---|
253 |
|
---|
254 | return VINF_SUCCESS;
|
---|
255 | }
|
---|
256 | //*****************************************************************************
|
---|
257 | //*****************************************************************************
|
---|
258 | char *DbgBytesToString(PDISCPUSTATE pCpu, RTUINTPTR pBytes, int size, char *pszOutput)
|
---|
259 | {
|
---|
260 | char szByte[4];
|
---|
261 | int len = strlen(pszOutput);
|
---|
262 | int i;
|
---|
263 |
|
---|
264 | for(i = len; i < 40; i++)
|
---|
265 | {
|
---|
266 | strcat(pszOutput, " ");
|
---|
267 | }
|
---|
268 | strcat(pszOutput, " [");
|
---|
269 | for(i = 0; i < size; i++)
|
---|
270 | {
|
---|
271 | RTStrPrintf(szByte, sizeof(szByte), "%02X ", DISReadByte(pCpu, pBytes+i));
|
---|
272 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, szByte);
|
---|
273 | }
|
---|
274 | len = strlen(pszOutput);
|
---|
275 | pszOutput[len - 1] = 0; //cut off last space
|
---|
276 |
|
---|
277 | strcat(pszOutput, "]");
|
---|
278 | return pszOutput;
|
---|
279 | }
|
---|
280 | //*****************************************************************************
|
---|
281 | //*****************************************************************************
|
---|
282 | void disasmSprintf(char *pszOutput, RTUINTPTR pu8Instruction, PDISCPUSTATE pCpu, OP_PARAMETER *pParam1, OP_PARAMETER *pParam2, OP_PARAMETER *pParam3)
|
---|
283 | {
|
---|
284 | const char *lpszFormat = pCpu->pszOpcode;
|
---|
285 | int param = 1;
|
---|
286 |
|
---|
287 | RTStrPrintf(pszOutput, 64, "%08X: ", (unsigned)pCpu->opaddr);
|
---|
288 | if(pCpu->prefix & PREFIX_LOCK)
|
---|
289 | {
|
---|
290 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "lock ");
|
---|
291 | }
|
---|
292 | if(pCpu->prefix & PREFIX_REP)
|
---|
293 | {
|
---|
294 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "rep(e) ");
|
---|
295 | }
|
---|
296 | else
|
---|
297 | if(pCpu->prefix & PREFIX_REPNE)
|
---|
298 | {
|
---|
299 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "repne ");
|
---|
300 | }
|
---|
301 |
|
---|
302 | if(!strcmp("Invalid Opcode", lpszFormat))
|
---|
303 | {
|
---|
304 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "Invalid Opcode [%02X][%02X]", DISReadByte(pCpu, pu8Instruction), DISReadByte(pCpu, pu8Instruction+1) );
|
---|
305 | }
|
---|
306 | else
|
---|
307 | while(*lpszFormat)
|
---|
308 | {
|
---|
309 | switch(*lpszFormat)
|
---|
310 | {
|
---|
311 | case '%':
|
---|
312 | switch(*(lpszFormat+1))
|
---|
313 | {
|
---|
314 | case 'J': //Relative jump offset
|
---|
315 | {
|
---|
316 | int32_t disp;
|
---|
317 |
|
---|
318 | AssertMsg(param == 1, ("Invalid branch parameter nr"));
|
---|
319 | if(pParam1->flags & USE_IMMEDIATE8_REL)
|
---|
320 | {
|
---|
321 | disp = (int32_t)(char)pParam1->parval;
|
---|
322 | }
|
---|
323 | else
|
---|
324 | if(pParam1->flags & USE_IMMEDIATE16_REL)
|
---|
325 | {
|
---|
326 | disp = (int32_t)(uint16_t)pParam1->parval;
|
---|
327 | }
|
---|
328 | else
|
---|
329 | if(pParam1->flags & USE_IMMEDIATE32_REL)
|
---|
330 | {
|
---|
331 | disp = (int32_t)pParam1->parval;
|
---|
332 | }
|
---|
333 | else
|
---|
334 | {
|
---|
335 | AssertMsgFailed(("Oops!\n"));
|
---|
336 | return;
|
---|
337 | }
|
---|
338 | uint32_t addr = (uint32_t)(pCpu->opaddr + pCpu->opsize) + disp;
|
---|
339 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "[%08X]", addr);
|
---|
340 | }
|
---|
341 |
|
---|
342 | //no break;
|
---|
343 |
|
---|
344 | case 'A': //direct address
|
---|
345 | case 'C': //control register
|
---|
346 | case 'D': //debug register
|
---|
347 | case 'E': //ModRM specifies parameter
|
---|
348 | case 'F': //Eflags register
|
---|
349 | case 'G': //ModRM selects general register
|
---|
350 | case 'I': //Immediate data
|
---|
351 | case 'M': //ModRM may only refer to memory
|
---|
352 | case 'O': //No ModRM byte
|
---|
353 | case 'P': //ModRM byte selects MMX register
|
---|
354 | case 'Q': //ModRM byte selects MMX register or memory address
|
---|
355 | case 'R': //ModRM byte may only refer to a general register
|
---|
356 | case 'S': //ModRM byte selects a segment register
|
---|
357 | case 'T': //ModRM byte selects a test register
|
---|
358 | case 'V': //ModRM byte selects an XMM/SSE register
|
---|
359 | case 'W': //ModRM byte selects an XMM/SSE register or a memory address
|
---|
360 | case 'X': //DS:SI
|
---|
361 | case 'Y': //ES:DI
|
---|
362 | switch(param)
|
---|
363 | {
|
---|
364 | case 1:
|
---|
365 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam1->szParam);
|
---|
366 | break;
|
---|
367 | case 2:
|
---|
368 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam2->szParam);
|
---|
369 | break;
|
---|
370 | case 3:
|
---|
371 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam3->szParam);
|
---|
372 | break;
|
---|
373 | }
|
---|
374 | break;
|
---|
375 |
|
---|
376 | case 'e': //register based on operand size (e.g. %eAX)
|
---|
377 | if(pCpu->opmode == CPUMODE_32BIT)
|
---|
378 | {
|
---|
379 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "E");
|
---|
380 | }
|
---|
381 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "%c%c", lpszFormat[2], lpszFormat[3]);
|
---|
382 | break;
|
---|
383 |
|
---|
384 | default:
|
---|
385 | AssertMsgFailed(("Oops!\n"));
|
---|
386 | break;
|
---|
387 | }
|
---|
388 |
|
---|
389 | //Go to the next parameter in the format string
|
---|
390 | while(*lpszFormat && *lpszFormat != ',') lpszFormat++;
|
---|
391 | if(*lpszFormat == ',') lpszFormat--;
|
---|
392 |
|
---|
393 | break;
|
---|
394 |
|
---|
395 | case ',':
|
---|
396 | param++;
|
---|
397 | //no break
|
---|
398 |
|
---|
399 | default:
|
---|
400 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "%c", *lpszFormat);
|
---|
401 | break;
|
---|
402 | }
|
---|
403 |
|
---|
404 | if(*lpszFormat) lpszFormat++;
|
---|
405 | }
|
---|
406 | DbgBytesToString(pCpu, pu8Instruction, pCpu->opsize, pszOutput);
|
---|
407 | RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "\n");
|
---|
408 | }
|
---|
409 | //*****************************************************************************
|
---|
410 | //*****************************************************************************
|
---|