Disasm.cpp@ 8361

Last change on this file since 8361 was 8361, checked in by vboxsync, 17 years ago
Even more disassembler updates
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 15.6 KB

Line
1	/** @file
2	*
3	* VBox disassembler:
4	* Main
5	*/
6
7	/*
8	* Copyright (C) 2006-2007 Sun Microsystems, Inc.
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	* Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa
19	* Clara, CA 95054 USA or visit http://www.sun.com if you need
20	* additional information or have any questions.
21	*/
22
23
24	/*******************************************************************************
25	* Header Files *
26	*******************************************************************************/
27	#ifdef USING_VISUAL_STUDIO
28	# include <stdafx.h>
29	#endif
30	#include <VBox/dis.h>
31	#include <VBox/disopcode.h>
32	#include <VBox/err.h>
33	#include <iprt/assert.h>
34	#include <iprt/string.h>
35	#include "DisasmInternal.h"
36	#include "DisasmTables.h"
37
38
39	/**
40	* Disassembles a code block.
41	*
42	* @returns VBox error code
43	* @param pCpu Pointer to cpu structure which have DISCPUSTATE::mode
44	* set correctly.
45	* @param pvCodeBlock Pointer to the strunction to disassemble.
46	* @param cbMax Maximum number of bytes to disassemble.
47	* @param pcbSize Where to store the size of the instruction.
48	* NULL is allowed.
49	*
50	*
51	* @todo Define output callback.
52	* @todo Using signed integers as sizes is a bit odd. There are still
53	* some GCC warnings about mixing signed and unsigend integers.
54	* @todo Need to extend this interface to include a code address so we
55	* can dissassemble GC code. Perhaps a new function is better...
56	* @remark cbMax isn't respected as a boundry. DISInstr() will read beyond cbMax.
57	* This means *pcbSize >= cbMax sometimes.
58	*/
59	DISDECL(int) DISBlock(PDISCPUSTATE pCpu, RTUINTPTR pvCodeBlock, unsigned cbMax, unsigned *pSize)
60	{
61	unsigned i = 0;
62	char szOutput[256];
63
64	while (i < cbMax)
65	{
66	unsigned cbInstr;
67	int rc = DISInstr(pCpu, pvCodeBlock + i, 0, &cbInstr, szOutput);
68	if (VBOX_FAILURE(rc))
69	return rc;
70
71	i += cbInstr;
72	}
73
74	if (pSize)
75	*pSize = i;
76	return true;
77	}
78
79	/**
80	* Disassembles one instruction
81	*
82	* @returns VBox error code
83	* @param pCpu Pointer to cpu structure which have DISCPUSTATE::mode
84	* set correctly.
85	* @param pu8Instruction Pointer to the strunction to disassemble.
86	* @param u32EipOffset Offset to add to instruction address to get the real virtual address
87	* @param pcbSize Where to store the size of the instruction.
88	* NULL is allowed.
89	* @param pszOutput Storage for disassembled instruction
90	*
91	* @todo Define output callback.
92	*/
93	DISDECL(int) DISInstr(PDISCPUSTATE pCpu, RTUINTPTR pu8Instruction, unsigned u32EipOffset, unsigned *pcbSize,
94	char *pszOutput)
95	{
96	return DISInstrEx(pCpu, pu8Instruction, u32EipOffset, pcbSize, pszOutput, OPTYPE_ALL);
97	}
98
99	/**
100	* Disassembles one instruction; only fully disassembly an instruction if it matches the filter criteria
101	*
102	* @returns VBox error code
103	* @param pCpu Pointer to cpu structure which have DISCPUSTATE::mode
104	* set correctly.
105	* @param pu8Instruction Pointer to the strunction to disassemble.
106	* @param u32EipOffset Offset to add to instruction address to get the real virtual address
107	* @param pcbSize Where to store the size of the instruction.
108	* NULL is allowed.
109	* @param pszOutput Storage for disassembled instruction
110	* @param uFilter Instruction type filter
111	*
112	* @todo Define output callback.
113	*/
114	DISDECL(int) DISInstrEx(PDISCPUSTATE pCpu, RTUINTPTR pu8Instruction, unsigned u32EipOffset, unsigned *pcbSize,
115	char *pszOutput, unsigned uFilter)
116	{
117	unsigned i = 0, prefixbytes;
118	unsigned idx, inc;
119	#ifdef __L4ENV__
120	jmp_buf jumpbuffer;
121	#endif
122
123	//reset instruction settings
124	pCpu->prefix = PREFIX_NONE;
125	pCpu->prefix_seg = 0;
126	pCpu->ModRM.u = 0;
127	pCpu->SIB.u = 0;
128	pCpu->lastprefix = 0;
129	pCpu->param1.parval = 0;
130	pCpu->param2.parval = 0;
131	pCpu->param3.parval = 0;
132	pCpu->param1.szParam[0] = 0;
133	pCpu->param2.szParam[0] = 0;
134	pCpu->param3.szParam[0] = 0;
135	pCpu->param1.size = 0;
136	pCpu->param2.size = 0;
137	pCpu->param3.size = 0;
138	pCpu->param1.flags = 0;
139	pCpu->param2.flags = 0;
140	pCpu->param3.flags = 0;
141	pCpu->uFilter = uFilter;
142	pCpu->pfnDisasmFnTable = pfnFullDisasm;
143
144	if (pszOutput)
145	*pszOutput = '\0';
146
147	if (pCpu->mode == CPUMODE_64BIT)
148	{
149	pCpu->addrmode = CPUMODE_64BIT;
150	pCpu->opmode = CPUMODE_32BIT;
151	}
152	else
153	{
154	pCpu->addrmode = pCpu->mode;
155	pCpu->opmode = pCpu->mode;
156	}
157
158	prefixbytes = 0;
159	#ifndef __L4ENV__ /* Unfortunately, we have no exception handling in l4env */
160	try
161	#else
162	pCpu->pJumpBuffer = &jumpbuffer;
163	if (setjmp(jumpbuffer) == 0)
164	#endif
165	{
166	while(1)
167	{
168	uint8_t codebyte = DISReadByte(pCpu, pu8Instruction+i);
169	uint8_t opcode = g_aOneByteMapX86[codebyte].opcode;
170
171	/* Hardcoded assumption about OP_* values!! */
172	if (opcode <= OP_LOCK)
173	{
174	pCpu->lastprefix = opcode;
175
176	/* The REX prefix must precede the opcode byte(s). Any other placement is ignored. */
177	if (opcode != OP_REX)
178	pCpu->prefix &= ~PREFIX_REX;
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->prefix_seg = g_aOneByteMapX86[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->prefix_seg >= OP_PARM_REG_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(opcode);
253
254	if (pCpu->prefix_rex & PREFIX_REX_FLAGS_W)
255	pCpu->opmode = CPUMODE_64BIT; /* overrides size prefix byte */
256	break;
257	}
258	}
259
260	idx = i;
261	i += sizeof(uint8_t); //first opcode byte
262
263	pCpu->opcode = codebyte;
264	/* Prefix byte(s) is/are part of the instruction. */
265	pCpu->opaddr = pu8Instruction + idx + u32EipOffset - prefixbytes;
266
267	if (pCpu->mode == CPUMODE_64BIT)
268	inc = ParseInstruction(pu8Instruction + i, &g_aOneByteMapX64[pCpu->opcode], pCpu);
269	else
270	inc = ParseInstruction(pu8Instruction + i, &g_aOneByteMapX86[pCpu->opcode], pCpu);
271
272	pCpu->opsize = prefixbytes + inc + sizeof(uint8_t);
273
274	if(pszOutput) {
275	disasmSprintf(pszOutput, pu8Instruction+i-1-prefixbytes, pCpu, &pCpu->param1, &pCpu->param2, &pCpu->param3);
276	}
277
278	i += inc;
279	prefixbytes = 0;
280	break;
281	}
282	}
283	#ifndef __L4ENV__
284	catch(...)
285	#else
286	else /* setjmp has returned a non-zero value: an exception occured */
287	#endif
288	{
289	if (pcbSize)
290	*pcbSize = 0;
291	return VERR_DIS_GEN_FAILURE;
292	}
293
294	if (pcbSize)
295	*pcbSize = i;
296
297	return VINF_SUCCESS;
298	}
299	//*****************************************************************************
300	//*****************************************************************************
301	char DbgBytesToString(PDISCPUSTATE pCpu, RTUINTPTR pBytes, int size, char pszOutput)
302	{
303	char szByte[4];
304	int len = strlen(pszOutput);
305	int i;
306
307	for(i = len; i < 40; i++)
308	{
309	strcat(pszOutput, " ");
310	}
311	strcat(pszOutput, " [");
312	for(i = 0; i < size; i++)
313	{
314	RTStrPrintf(szByte, sizeof(szByte), "%02X ", DISReadByte(pCpu, pBytes+i));
315	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, szByte);
316	}
317	len = strlen(pszOutput);
318	pszOutput[len - 1] = 0; //cut off last space
319
320	strcat(pszOutput, "]");
321	return pszOutput;
322	}
323	//*****************************************************************************
324	//*****************************************************************************
325	void disasmSprintf(char pszOutput, RTUINTPTR pu8Instruction, PDISCPUSTATE pCpu, OP_PARAMETER pParam1, OP_PARAMETER pParam2, OP_PARAMETER pParam3)
326	{
327	const char *lpszFormat = pCpu->pszOpcode;
328	int param = 1;
329
330	RTStrPrintf(pszOutput, 64, "%08X: ", (unsigned)pCpu->opaddr);
331	if(pCpu->prefix & PREFIX_LOCK)
332	{
333	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "lock ");
334	}
335	if(pCpu->prefix & PREFIX_REP)
336	{
337	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "rep(e) ");
338	}
339	else
340	if(pCpu->prefix & PREFIX_REPNE)
341	{
342	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "repne ");
343	}
344
345	if(!strcmp("Invalid Opcode", lpszFormat))
346	{
347	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "Invalid Opcode [%02X][%02X]", DISReadByte(pCpu, pu8Instruction), DISReadByte(pCpu, pu8Instruction+1) );
348	}
349	else
350	while(*lpszFormat)
351	{
352	switch(*lpszFormat)
353	{
354	case '%':
355	switch(*(lpszFormat+1))
356	{
357	case 'J': //Relative jump offset
358	{
359	int32_t disp;
360
361	AssertMsg(param == 1, ("Invalid branch parameter nr"));
362	if(pParam1->flags & USE_IMMEDIATE8_REL)
363	{
364	disp = (int32_t)(char)pParam1->parval;
365	}
366	else
367	if(pParam1->flags & USE_IMMEDIATE16_REL)
368	{
369	disp = (int32_t)(uint16_t)pParam1->parval;
370	}
371	else
372	if(pParam1->flags & USE_IMMEDIATE32_REL)
373	{
374	disp = (int32_t)pParam1->parval;
375	}
376	else
377	{
378	AssertMsgFailed(("Oops!\n"));
379	return;
380	}
381	uint32_t addr = (uint32_t)(pCpu->opaddr + pCpu->opsize) + disp;
382	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "[%08X]", addr);
383	}
384
385	//no break;
386
387	case 'A': //direct address
388	case 'C': //control register
389	case 'D': //debug register
390	case 'E': //ModRM specifies parameter
391	case 'F': //Eflags register
392	case 'G': //ModRM selects general register
393	case 'I': //Immediate data
394	case 'M': //ModRM may only refer to memory
395	case 'O': //No ModRM byte
396	case 'P': //ModRM byte selects MMX register
397	case 'Q': //ModRM byte selects MMX register or memory address
398	case 'R': //ModRM byte may only refer to a general register
399	case 'S': //ModRM byte selects a segment register
400	case 'T': //ModRM byte selects a test register
401	case 'V': //ModRM byte selects an XMM/SSE register
402	case 'W': //ModRM byte selects an XMM/SSE register or a memory address
403	case 'X': //DS:SI
404	case 'Y': //ES:DI
405	switch(param)
406	{
407	case 1:
408	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam1->szParam);
409	break;
410	case 2:
411	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam2->szParam);
412	break;
413	case 3:
414	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam3->szParam);
415	break;
416	}
417	break;
418
419	case 'e': //register based on operand size (e.g. %eAX)
420	if(pCpu->opmode == CPUMODE_32BIT)
421	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "E");
422	if(pCpu->opmode == CPUMODE_64BIT)
423	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "R");
424
425	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "%c%c", lpszFormat[2], lpszFormat[3]);
426	break;
427
428	default:
429	AssertMsgFailed(("Oops!\n"));
430	break;
431	}
432
433	//Go to the next parameter in the format string
434	while(lpszFormat && lpszFormat != ',') lpszFormat++;
435	if(*lpszFormat == ',') lpszFormat--;
436
437	break;
438
439	case ',':
440	param++;
441	//no break
442
443	default:
444	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "%c", *lpszFormat);
445	break;
446	}
447
448	if(*lpszFormat) lpszFormat++;
449	}
450	DbgBytesToString(pCpu, pu8Instruction, pCpu->opsize, pszOutput);
451	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "\n");
452	}
453	//*****************************************************************************
454	//*****************************************************************************

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

source: vbox/trunk/src/VBox/Disassembler/Disasm.cpp@ 8361

Download in other formats: