Disasm.cpp@ 36219

Last change on this file since 36219 was 33540, checked in by vboxsync, 14 years ago
*: spelling fixes, thanks Timeless!
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 15.7 KB

Line
1	/** @file
2	*
3	* VBox disassembler:
4	* Main
5	*/
6
7	/*
8	* Copyright (C) 2006-2007 Oracle Corporation
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 structure 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 unsigned integers.
50	* @todo Need to extend this interface to include a code address so we
51	* can disassemble GC code. Perhaps a new function is better...
52	* @remark cbMax isn't respected as a boundary. 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 (RT_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 structure 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 structure 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	const OPCODE *paOneByteMap;
116	#ifdef __L4ENV__
117	jmp_buf jumpbuffer;
118	#endif
119
120	//reset instruction settings
121	pCpu->prefix = PREFIX_NONE;
122	pCpu->enmPrefixSeg = DIS_SELREG_DS;
123	pCpu->ModRM.u = 0;
124	pCpu->SIB.u = 0;
125	pCpu->lastprefix = 0;
126	pCpu->param1.parval = 0;
127	pCpu->param2.parval = 0;
128	pCpu->param3.parval = 0;
129	pCpu->param1.szParam[0] = 0;
130	pCpu->param2.szParam[0] = 0;
131	pCpu->param3.szParam[0] = 0;
132	pCpu->param1.size = 0;
133	pCpu->param2.size = 0;
134	pCpu->param3.size = 0;
135	pCpu->param1.flags = 0;
136	pCpu->param2.flags = 0;
137	pCpu->param3.flags = 0;
138	pCpu->uFilter = uFilter;
139	pCpu->pfnDisasmFnTable = pfnFullDisasm;
140
141	if (pszOutput)
142	*pszOutput = '\0';
143
144	if (pCpu->mode == CPUMODE_64BIT)
145	{
146	paOneByteMap = g_aOneByteMapX64;
147	pCpu->addrmode = CPUMODE_64BIT;
148	pCpu->opmode = CPUMODE_32BIT;
149	}
150	else
151	{
152	paOneByteMap = g_aOneByteMapX86;
153	pCpu->addrmode = pCpu->mode;
154	pCpu->opmode = pCpu->mode;
155	}
156
157	prefixbytes = 0;
158	#ifndef __L4ENV__ /* Unfortunately, we have no exception handling in l4env */
159	try
160	#else
161	pCpu->pJumpBuffer = &jumpbuffer;
162	if (setjmp(jumpbuffer) == 0)
163	#endif
164	{
165	while(1)
166	{
167	uint8_t codebyte = DISReadByte(pCpu, pu8Instruction+i);
168	uint8_t opcode = paOneByteMap[codebyte].opcode;
169
170	/* Hardcoded assumption about OP_* values!! */
171	if (opcode <= OP_LAST_PREFIX)
172	{
173	/* The REX prefix must precede the opcode byte(s). Any other placement is ignored. */
174	if (opcode != OP_REX)
175	{
176	pCpu->lastprefix = opcode;
177	pCpu->prefix &= ~PREFIX_REX;
178	}
179
180	switch(opcode)
181	{
182	case OP_INVALID:
183	#if 0 //defined (DEBUG_Sander)
184	AssertMsgFailed(("Invalid opcode!!\n"));
185	#endif
186	return VERR_DIS_INVALID_OPCODE;
187
188	// segment override prefix byte
189	case OP_SEG:
190	pCpu->enmPrefixSeg = (DIS_SELREG)(paOneByteMap[codebyte].param1 - OP_PARM_REG_SEG_START);
191	/* Segment prefixes for CS, DS, ES and SS are ignored in long mode. */
192	if ( pCpu->mode != CPUMODE_64BIT
193	\|\| pCpu->enmPrefixSeg >= DIS_SELREG_FS)
194	{
195	pCpu->prefix \|= PREFIX_SEG;
196	}
197	i += sizeof(uint8_t);
198	prefixbytes++;
199	continue; //fetch the next byte
200
201	// lock prefix byte
202	case OP_LOCK:
203	pCpu->prefix \|= PREFIX_LOCK;
204	i += sizeof(uint8_t);
205	prefixbytes++;
206	continue; //fetch the next byte
207
208	// address size override prefix byte
209	case OP_ADDRSIZE:
210	pCpu->prefix \|= PREFIX_ADDRSIZE;
211	if (pCpu->mode == CPUMODE_16BIT)
212	pCpu->addrmode = CPUMODE_32BIT;
213	else
214	if (pCpu->mode == CPUMODE_32BIT)
215	pCpu->addrmode = CPUMODE_16BIT;
216	else
217	pCpu->addrmode = CPUMODE_32BIT; /* 64 bits */
218
219	i += sizeof(uint8_t);
220	prefixbytes++;
221	continue; //fetch the next byte
222
223	// operand size override prefix byte
224	case OP_OPSIZE:
225	pCpu->prefix \|= PREFIX_OPSIZE;
226	if (pCpu->mode == CPUMODE_16BIT)
227	pCpu->opmode = CPUMODE_32BIT;
228	else
229	pCpu->opmode = CPUMODE_16BIT; /* for 32 and 64 bits mode (there is no 32 bits operand size override prefix) */
230
231	i += sizeof(uint8_t);
232	prefixbytes++;
233	continue; //fetch the next byte
234
235	// rep and repne are not really prefixes, but we'll treat them as such
236	case OP_REPE:
237	pCpu->prefix \|= PREFIX_REP;
238	i += sizeof(uint8_t);
239	prefixbytes += sizeof(uint8_t);
240	continue; //fetch the next byte
241
242	case OP_REPNE:
243	pCpu->prefix \|= PREFIX_REPNE;
244	i += sizeof(uint8_t);
245	prefixbytes += sizeof(uint8_t);
246	continue; //fetch the next byte
247
248	case OP_REX:
249	Assert(pCpu->mode == CPUMODE_64BIT);
250	/* REX prefix byte */
251	pCpu->prefix \|= PREFIX_REX;
252	pCpu->prefix_rex = PREFIX_REX_OP_2_FLAGS(paOneByteMap[codebyte].param1);
253	i += sizeof(uint8_t);
254	prefixbytes += sizeof(uint8_t);
255
256	if (pCpu->prefix_rex & PREFIX_REX_FLAGS_W)
257	pCpu->opmode = CPUMODE_64BIT; /* overrides size prefix byte */
258	continue; //fetch the next byte
259	}
260	}
261
262	idx = i;
263	i += sizeof(uint8_t); //first opcode byte
264
265	pCpu->opcode = codebyte;
266	/* Prefix byte(s) is/are part of the instruction. */
267	pCpu->opaddr = pu8Instruction + idx + u32EipOffset - prefixbytes;
268
269	inc = ParseInstruction(pu8Instruction + i, &paOneByteMap[pCpu->opcode], pCpu);
270
271	pCpu->opsize = prefixbytes + inc + sizeof(uint8_t);
272
273	if(pszOutput) {
274	disasmSprintf(pszOutput, pu8Instruction+i-1-prefixbytes, pCpu, &pCpu->param1, &pCpu->param2, &pCpu->param3);
275	}
276
277	i += inc;
278	prefixbytes = 0;
279	break;
280	}
281	}
282	#ifndef __L4ENV__
283	catch(...)
284	#else
285	else /* setjmp has returned a non-zero value: an exception occurred */
286	#endif
287	{
288	if (pcbSize)
289	*pcbSize = 0;
290	return VERR_DIS_GEN_FAILURE;
291	}
292
293	if (pcbSize)
294	*pcbSize = i;
295
296	if (pCpu->prefix & PREFIX_LOCK)
297	disValidateLockSequence(pCpu);
298
299	return VINF_SUCCESS;
300	}
301	//*****************************************************************************
302	//*****************************************************************************
303	char DbgBytesToString(PDISCPUSTATE pCpu, RTUINTPTR pBytes, int size, char pszOutput)
304	{
305	char *psz = strchr(pszOutput, '\0');
306	size_t len;
307	int i;
308
309	for(len = psz - pszOutput; len < 40; len++)
310	*psz++ = ' ';
311	*psz++ = ' ';
312	*psz++ = '[';
313
314	for(i = 0; (int)i < size; i++)
315	psz += RTStrPrintf(psz, 64, "%02X ", DISReadByte(pCpu, pBytes+i));
316
317	psz[-1] = ']'; // replaces space
318
319	return pszOutput;
320	}
321	//*****************************************************************************
322	//*****************************************************************************
323	void disasmSprintf(char pszOutput, RTUINTPTR pu8Instruction, PDISCPUSTATE pCpu, OP_PARAMETER pParam1, OP_PARAMETER pParam2, OP_PARAMETER pParam3)
324	{
325	const char *lpszFormat = pCpu->pszOpcode;
326	int param = 1;
327
328	RTStrPrintf(pszOutput, 64, "%08X: ", (unsigned)pCpu->opaddr);
329	if(pCpu->prefix & PREFIX_LOCK)
330	{
331	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "lock ");
332	}
333	if(pCpu->prefix & PREFIX_REP)
334	{
335	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "rep(e) ");
336	}
337	else
338	if(pCpu->prefix & PREFIX_REPNE)
339	{
340	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "repne ");
341	}
342
343	if(!strcmp("Invalid Opcode", lpszFormat))
344	{
345	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "Invalid Opcode [%02X][%02X]", DISReadByte(pCpu, pu8Instruction), DISReadByte(pCpu, pu8Instruction+1) );
346	}
347	else
348	while(*lpszFormat)
349	{
350	switch(*lpszFormat)
351	{
352	case '%':
353	switch(*(lpszFormat+1))
354	{
355	case 'J': //Relative jump offset
356	{
357	int32_t disp;
358
359	AssertMsg(param == 1, ("Invalid branch parameter nr"));
360	if(pParam1->flags & USE_IMMEDIATE8_REL)
361	{
362	disp = (int32_t)(char)pParam1->parval;
363	}
364	else
365	if(pParam1->flags & USE_IMMEDIATE16_REL)
366	{
367	disp = (int32_t)(uint16_t)pParam1->parval;
368	}
369	else
370	if(pParam1->flags & USE_IMMEDIATE32_REL)
371	{
372	disp = (int32_t)pParam1->parval;
373	}
374	else
375	if(pParam1->flags & USE_IMMEDIATE64_REL)
376	{
377	/* @todo: is this correct? */
378	disp = (int32_t)pParam1->parval;
379	}
380	else
381	{
382	AssertMsgFailed(("Oops!\n"));
383	return;
384	}
385	uint32_t addr = (uint32_t)(pCpu->opaddr + pCpu->opsize) + disp;
386	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "[%08X]", addr);
387	}
388
389	//no break;
390
391	case 'A': //direct address
392	case 'C': //control register
393	case 'D': //debug register
394	case 'E': //ModRM specifies parameter
395	case 'F': //Eflags register
396	case 'G': //ModRM selects general register
397	case 'I': //Immediate data
398	case 'M': //ModRM may only refer to memory
399	case 'O': //No ModRM byte
400	case 'P': //ModRM byte selects MMX register
401	case 'Q': //ModRM byte selects MMX register or memory address
402	case 'R': //ModRM byte may only refer to a general register
403	case 'S': //ModRM byte selects a segment register
404	case 'T': //ModRM byte selects a test register
405	case 'V': //ModRM byte selects an XMM/SSE register
406	case 'W': //ModRM byte selects an XMM/SSE register or a memory address
407	case 'X': //DS:SI
408	case 'Y': //ES:DI
409	switch(param)
410	{
411	case 1:
412	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam1->szParam);
413	break;
414	case 2:
415	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam2->szParam);
416	break;
417	case 3:
418	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam3->szParam);
419	break;
420	}
421	break;
422
423	case 'e': //register based on operand size (e.g. %eAX)
424	if(pCpu->opmode == CPUMODE_32BIT)
425	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "E");
426	if(pCpu->opmode == CPUMODE_64BIT)
427	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "R");
428
429	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "%c%c", lpszFormat[2], lpszFormat[3]);
430	break;
431
432	default:
433	AssertMsgFailed(("Oops!\n"));
434	break;
435	}
436
437	//Go to the next parameter in the format string
438	while(lpszFormat && lpszFormat != ',') lpszFormat++;
439	if(*lpszFormat == ',') lpszFormat--;
440
441	break;
442
443	case ',':
444	param++;
445	//no break
446
447	default:
448	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "%c", *lpszFormat);
449	break;
450	}
451
452	if(*lpszFormat) lpszFormat++;
453	}
454	DbgBytesToString(pCpu, pu8Instruction, pCpu->opsize, pszOutput);
455	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "\n");
456	}
457	//*****************************************************************************
458	//*****************************************************************************

Note: See TracBrowser for help on using the repository browser.

source: vbox/trunk/src/VBox/Disassembler/Disasm.cpp@ 36219

Download in other formats: