Disasm.cpp@ 8377

Last change on this file since 8377 was 8377, checked in by vboxsync, 17 years ago
Disassembler fixes + testcase for 64 bits
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 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	const OPCODE *paOneByteMap;
120	#ifdef __L4ENV__
121	jmp_buf jumpbuffer;
122	#endif
123
124	//reset instruction settings
125	pCpu->prefix = PREFIX_NONE;
126	pCpu->prefix_seg = 0;
127	pCpu->ModRM.u = 0;
128	pCpu->SIB.u = 0;
129	pCpu->lastprefix = 0;
130	pCpu->param1.parval = 0;
131	pCpu->param2.parval = 0;
132	pCpu->param3.parval = 0;
133	pCpu->param1.szParam[0] = 0;
134	pCpu->param2.szParam[0] = 0;
135	pCpu->param3.szParam[0] = 0;
136	pCpu->param1.size = 0;
137	pCpu->param2.size = 0;
138	pCpu->param3.size = 0;
139	pCpu->param1.flags = 0;
140	pCpu->param2.flags = 0;
141	pCpu->param3.flags = 0;
142	pCpu->uFilter = uFilter;
143	pCpu->pfnDisasmFnTable = pfnFullDisasm;
144
145	if (pszOutput)
146	*pszOutput = '\0';
147
148	if (pCpu->mode == CPUMODE_64BIT)
149	{
150	paOneByteMap = g_aOneByteMapX64;
151	pCpu->addrmode = CPUMODE_64BIT;
152	pCpu->opmode = CPUMODE_32BIT;
153	}
154	else
155	{
156	paOneByteMap = g_aOneByteMapX86;
157	pCpu->addrmode = pCpu->mode;
158	pCpu->opmode = pCpu->mode;
159	}
160
161	prefixbytes = 0;
162	#ifndef __L4ENV__ /* Unfortunately, we have no exception handling in l4env */
163	try
164	#else
165	pCpu->pJumpBuffer = &jumpbuffer;
166	if (setjmp(jumpbuffer) == 0)
167	#endif
168	{
169	while(1)
170	{
171	uint8_t codebyte = DISReadByte(pCpu, pu8Instruction+i);
172	uint8_t opcode = paOneByteMap[codebyte].opcode;
173
174	/* Hardcoded assumption about OP_* values!! */
175	if (opcode <= OP_LAST_PREFIX)
176	{
177	pCpu->lastprefix = opcode;
178
179	/* The REX prefix must precede the opcode byte(s). Any other placement is ignored. */
180	if (opcode != OP_REX)
181	pCpu->prefix &= ~PREFIX_REX;
182
183	switch(opcode)
184	{
185	case OP_INVALID:
186	#if 0 //defined (DEBUG_Sander)
187	AssertMsgFailed(("Invalid opcode!!\n"));
188	#endif
189	return VERR_DIS_INVALID_OPCODE;
190
191	// segment override prefix byte
192	case OP_SEG:
193	pCpu->prefix_seg = paOneByteMap[codebyte].param1 - OP_PARM_REG_SEG_START;
194	/* Segment prefixes for CS, DS, ES and SS are ignored in long mode. */
195	if ( pCpu->mode != CPUMODE_64BIT
196	\|\| pCpu->prefix_seg >= OP_PARM_REG_FS)
197	{
198	pCpu->prefix \|= PREFIX_SEG;
199	}
200	i += sizeof(uint8_t);
201	prefixbytes++;
202	continue; //fetch the next byte
203
204	// lock prefix byte
205	case OP_LOCK:
206	pCpu->prefix \|= PREFIX_LOCK;
207	i += sizeof(uint8_t);
208	prefixbytes++;
209	continue; //fetch the next byte
210
211	// address size override prefix byte
212	case OP_ADDRSIZE:
213	pCpu->prefix \|= PREFIX_ADDRSIZE;
214	if (pCpu->mode == CPUMODE_16BIT)
215	pCpu->addrmode = CPUMODE_32BIT;
216	else
217	if (pCpu->mode == CPUMODE_32BIT)
218	pCpu->addrmode = CPUMODE_16BIT;
219	else
220	pCpu->addrmode = CPUMODE_32BIT; /* 64 bits */
221
222	i += sizeof(uint8_t);
223	prefixbytes++;
224	continue; //fetch the next byte
225
226	// operand size override prefix byte
227	case OP_OPSIZE:
228	pCpu->prefix \|= PREFIX_OPSIZE;
229	if (pCpu->mode == CPUMODE_16BIT)
230	pCpu->opmode = CPUMODE_32BIT;
231	else
232	pCpu->opmode = CPUMODE_16BIT; /* for 32 and 64 bits mode (there is no 32 bits operand size override prefix) */
233
234	i += sizeof(uint8_t);
235	prefixbytes++;
236	continue; //fetch the next byte
237
238	// rep and repne are not really prefixes, but we'll treat them as such
239	case OP_REPE:
240	pCpu->prefix \|= PREFIX_REP;
241	i += sizeof(uint8_t);
242	prefixbytes += sizeof(uint8_t);
243	continue; //fetch the next byte
244
245	case OP_REPNE:
246	pCpu->prefix \|= PREFIX_REPNE;
247	i += sizeof(uint8_t);
248	prefixbytes += sizeof(uint8_t);
249	continue; //fetch the next byte
250
251	case OP_REX:
252	Assert(pCpu->mode == CPUMODE_64BIT);
253	/* REX prefix byte */
254	pCpu->prefix \|= PREFIX_REX;
255	pCpu->prefix_rex = PREFIX_REX_OP_2_FLAGS(paOneByteMap[codebyte].param1);
256	i += sizeof(uint8_t);
257	prefixbytes += sizeof(uint8_t);
258
259	if (pCpu->prefix_rex & PREFIX_REX_FLAGS_W)
260	pCpu->opmode = CPUMODE_64BIT; /* overrides size prefix byte */
261	continue; //fetch the next byte
262	}
263	}
264
265	idx = i;
266	i += sizeof(uint8_t); //first opcode byte
267
268	pCpu->opcode = codebyte;
269	/* Prefix byte(s) is/are part of the instruction. */
270	pCpu->opaddr = pu8Instruction + idx + u32EipOffset - prefixbytes;
271
272	inc = ParseInstruction(pu8Instruction + i, &paOneByteMap[pCpu->opcode], pCpu);
273
274	pCpu->opsize = prefixbytes + inc + sizeof(uint8_t);
275
276	if(pszOutput) {
277	disasmSprintf(pszOutput, pu8Instruction+i-1-prefixbytes, pCpu, &pCpu->param1, &pCpu->param2, &pCpu->param3);
278	}
279
280	i += inc;
281	prefixbytes = 0;
282	break;
283	}
284	}
285	#ifndef __L4ENV__
286	catch(...)
287	#else
288	else /* setjmp has returned a non-zero value: an exception occured */
289	#endif
290	{
291	if (pcbSize)
292	*pcbSize = 0;
293	return VERR_DIS_GEN_FAILURE;
294	}
295
296	if (pcbSize)
297	*pcbSize = i;
298
299	return VINF_SUCCESS;
300	}
301	//*****************************************************************************
302	//*****************************************************************************
303	char DbgBytesToString(PDISCPUSTATE pCpu, RTUINTPTR pBytes, int size, char pszOutput)
304	{
305	char szByte[4];
306	int len = strlen(pszOutput);
307	int i;
308
309	for(i = len; i < 40; i++)
310	{
311	strcat(pszOutput, " ");
312	}
313	strcat(pszOutput, " [");
314	for(i = 0; i < size; i++)
315	{
316	RTStrPrintf(szByte, sizeof(szByte), "%02X ", DISReadByte(pCpu, pBytes+i));
317	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, szByte);
318	}
319	len = strlen(pszOutput);
320	pszOutput[len - 1] = 0; //cut off last space
321
322	strcat(pszOutput, "]");
323	return pszOutput;
324	}
325	//*****************************************************************************
326	//*****************************************************************************
327	void disasmSprintf(char pszOutput, RTUINTPTR pu8Instruction, PDISCPUSTATE pCpu, OP_PARAMETER pParam1, OP_PARAMETER pParam2, OP_PARAMETER pParam3)
328	{
329	const char *lpszFormat = pCpu->pszOpcode;
330	int param = 1;
331
332	RTStrPrintf(pszOutput, 64, "%08X: ", (unsigned)pCpu->opaddr);
333	if(pCpu->prefix & PREFIX_LOCK)
334	{
335	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "lock ");
336	}
337	if(pCpu->prefix & PREFIX_REP)
338	{
339	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "rep(e) ");
340	}
341	else
342	if(pCpu->prefix & PREFIX_REPNE)
343	{
344	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "repne ");
345	}
346
347	if(!strcmp("Invalid Opcode", lpszFormat))
348	{
349	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "Invalid Opcode [%02X][%02X]", DISReadByte(pCpu, pu8Instruction), DISReadByte(pCpu, pu8Instruction+1) );
350	}
351	else
352	while(*lpszFormat)
353	{
354	switch(*lpszFormat)
355	{
356	case '%':
357	switch(*(lpszFormat+1))
358	{
359	case 'J': //Relative jump offset
360	{
361	int32_t disp;
362
363	AssertMsg(param == 1, ("Invalid branch parameter nr"));
364	if(pParam1->flags & USE_IMMEDIATE8_REL)
365	{
366	disp = (int32_t)(char)pParam1->parval;
367	}
368	else
369	if(pParam1->flags & USE_IMMEDIATE16_REL)
370	{
371	disp = (int32_t)(uint16_t)pParam1->parval;
372	}
373	else
374	if(pParam1->flags & USE_IMMEDIATE32_REL)
375	{
376	disp = (int32_t)pParam1->parval;
377	}
378	else
379	{
380	AssertMsgFailed(("Oops!\n"));
381	return;
382	}
383	uint32_t addr = (uint32_t)(pCpu->opaddr + pCpu->opsize) + disp;
384	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "[%08X]", addr);
385	}
386
387	//no break;
388
389	case 'A': //direct address
390	case 'C': //control register
391	case 'D': //debug register
392	case 'E': //ModRM specifies parameter
393	case 'F': //Eflags register
394	case 'G': //ModRM selects general register
395	case 'I': //Immediate data
396	case 'M': //ModRM may only refer to memory
397	case 'O': //No ModRM byte
398	case 'P': //ModRM byte selects MMX register
399	case 'Q': //ModRM byte selects MMX register or memory address
400	case 'R': //ModRM byte may only refer to a general register
401	case 'S': //ModRM byte selects a segment register
402	case 'T': //ModRM byte selects a test register
403	case 'V': //ModRM byte selects an XMM/SSE register
404	case 'W': //ModRM byte selects an XMM/SSE register or a memory address
405	case 'X': //DS:SI
406	case 'Y': //ES:DI
407	switch(param)
408	{
409	case 1:
410	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam1->szParam);
411	break;
412	case 2:
413	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam2->szParam);
414	break;
415	case 3:
416	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, pParam3->szParam);
417	break;
418	}
419	break;
420
421	case 'e': //register based on operand size (e.g. %eAX)
422	if(pCpu->opmode == CPUMODE_32BIT)
423	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "E");
424	if(pCpu->opmode == CPUMODE_64BIT)
425	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "R");
426
427	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "%c%c", lpszFormat[2], lpszFormat[3]);
428	break;
429
430	default:
431	AssertMsgFailed(("Oops!\n"));
432	break;
433	}
434
435	//Go to the next parameter in the format string
436	while(lpszFormat && lpszFormat != ',') lpszFormat++;
437	if(*lpszFormat == ',') lpszFormat--;
438
439	break;
440
441	case ',':
442	param++;
443	//no break
444
445	default:
446	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "%c", *lpszFormat);
447	break;
448	}
449
450	if(*lpszFormat) lpszFormat++;
451	}
452	DbgBytesToString(pCpu, pu8Instruction, pCpu->opsize, pszOutput);
453	RTStrPrintf(&pszOutput[strlen(pszOutput)], 64, "\n");
454	}
455	//*****************************************************************************
456	//*****************************************************************************

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

source: vbox/trunk/src/VBox/Disassembler/Disasm.cpp@ 8377

Download in other formats: