tstDisasm-2.cpp@ 27023

Last change on this file since 27023 was 26517, checked in by vboxsync, 15 years ago
*: RTGetOpt cleanup related to --help and --version (now standard option). Use RTGetOptPrintError.
Property svn:eol-style set to `native` Property svn:keywords set to `Id`
File size: 30.8 KB

Line
1	/* $Id: tstDisasm-2.cpp 26517 2010-02-14 21:39:00Z vboxsync $ */
2	/** @file
3	* Testcase - Generic Disassembler Tool.
4	*/
5
6	/*
7	* Copyright (C) 2008 Sun Microsystems, Inc.
8	*
9	* This file is part of VirtualBox Open Source Edition (OSE), as
10	* available from http://www.virtualbox.org. This file is free software;
11	* you can redistribute it and/or modify it under the terms of the GNU
12	* General Public License (GPL) as published by the Free Software
13	* Foundation, in version 2 as it comes in the "COPYING" file of the
14	* VirtualBox OSE distribution. VirtualBox OSE is distributed in the
15	* hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
16	*
17	* Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa
18	* Clara, CA 95054 USA or visit http://www.sun.com if you need
19	* additional information or have any questions.
20	*/
21
22	/*******************************************************************************
23	* Header Files *
24	*******************************************************************************/
25	#include <VBox/dis.h>
26	#include <VBox/err.h>
27	#include <iprt/alloc.h>
28	#include <iprt/assert.h>
29	#include <iprt/initterm.h>
30	#include <iprt/getopt.h>
31	#include <iprt/file.h>
32	#include <iprt/path.h>
33	#include <iprt/stream.h>
34	#include <iprt/string.h>
35	#include <iprt/ctype.h>
36
37
38	/*******************************************************************************
39	* Structures and Typedefs *
40	*******************************************************************************/
41	typedef enum { kAsmStyle_Default, kAsmStyle_yasm, kAsmStyle_masm, kAsmStyle_gas, kAsmStyle_invalid } ASMSTYLE;
42	typedef enum { kUndefOp_Fail, kUndefOp_All, kUndefOp_DefineByte, kUndefOp_End } UNDEFOPHANDLING;
43
44	typedef struct MYDISSTATE
45	{
46	DISCPUSTATE Cpu;
47	uint64_t uAddress; /*< The current instruction address. /
48	uint8_t pbInstr; /< The current instruction (pointer). /
49	uint32_t cbInstr; /*< The size of the current instruction. /
50	bool fUndefOp; /*< Whether the current instruction is really an undefined opcode./
51	UNDEFOPHANDLING enmUndefOp; /*< How to treat undefined opcodes. /
52	int rc; /*< Set if we hit EOF. /
53	size_t cbLeft; /*< The number of bytes left. (read) /
54	uint8_t pbNext; /< The next byte. (read) /
55	uint64_t uNextAddr; /*< The address of the next byte. (read) /
56	char szLine[256]; /*< The disassembler text output. /
57	} MYDISSTATE;
58	typedef MYDISSTATE *PMYDISSTATE;
59
60
61
62	/**
63	* Default style.
64	*
65	* @param pState The disassembler state.
66	*/
67	static void MyDisasDefaultFormatter(PMYDISSTATE pState)
68	{
69	RTPrintf("%s", pState->szLine);
70	}
71
72
73	/**
74	* Yasm style.
75	*
76	* @param pState The disassembler state.
77	*/
78	static void MyDisasYasmFormatter(PMYDISSTATE pState)
79	{
80	char szTmp[256];
81	#if 0
82	/* a very quick hack. */
83	strcpy(szTmp, RTStrStripL(strchr(pState->szLine, ':') + 1));
84
85	char *psz = strrchr(szTmp, '[');
86	*psz = '\0';
87	RTStrStripR(szTmp);
88
89	psz = strstr(szTmp, " ptr ");
90	if (psz)
91	memset(psz, ' ', 5);
92
93	char *pszEnd = strchr(szTmp, '\0');
94	while (pszEnd - &szTmp[0] < 71)
95	*pszEnd++ = ' ';
96	*pszEnd = '\0';
97
98	#else
99	size_t cch = DISFormatYasmEx(&pState->Cpu, szTmp, sizeof(szTmp),
100	DIS_FMT_FLAGS_STRICT \| DIS_FMT_FLAGS_ADDR_RIGHT \| DIS_FMT_FLAGS_ADDR_COMMENT
101	\| DIS_FMT_FLAGS_BYTES_RIGHT \| DIS_FMT_FLAGS_BYTES_COMMENT \| DIS_FMT_FLAGS_BYTES_SPACED,
102	NULL, NULL);
103	Assert(cch < sizeof(szTmp));
104	while (cch < 71)
105	szTmp[cch++] = ' ';
106	szTmp[cch] = '\0';
107	#endif
108
109	RTPrintf(" %s ; %08llu %s", szTmp, pState->uAddress, pState->szLine);
110	}
111
112
113	/**
114	* Checks if the encoding of the current instruction is something
115	* we can never get the assembler to produce.
116	*
117	* @returns true if it's odd, false if it isn't.
118	* @param pCpu The disassembler output.
119	*/
120	static bool MyDisasYasmFormatterIsOddEncoding(PMYDISSTATE pState)
121	{
122	/*
123	* Mod rm + SIB: Check for duplicate EBP encodings that yasm won't use for very good reasons.
124	*/
125	if ( pState->Cpu.addrmode != CPUMODE_16BIT ///@todo correct?
126	&& pState->Cpu.ModRM.Bits.Rm == 4
127	&& pState->Cpu.ModRM.Bits.Mod != 3)
128	{
129	/* No scaled index SIB (index=4), except for ESP. */
130	if ( pState->Cpu.SIB.Bits.Index == 4
131	&& pState->Cpu.SIB.Bits.Base != 4)
132	return true;
133
134	/* EBP + displacement */
135	if ( pState->Cpu.ModRM.Bits.Mod != 0
136	&& pState->Cpu.SIB.Bits.Base == 5
137	&& pState->Cpu.SIB.Bits.Scale == 0)
138	return true;
139	}
140
141	/*
142	* Seems to be an instruction alias here, but I cannot find any docs on it... hrmpf!
143	*/
144	if ( pState->Cpu.pCurInstr->opcode == OP_SHL
145	&& pState->Cpu.ModRM.Bits.Reg == 6)
146	return true;
147
148	/*
149	* Check for multiple prefixes of the same kind.
150	*/
151	uint32_t fPrefixes = 0;
152	for (uint8_t const *pu8 = pState->pbInstr;; pu8++)
153	{
154	uint32_t f;
155	switch (*pu8)
156	{
157	case 0xf0:
158	f = PREFIX_LOCK;
159	break;
160
161	case 0xf2:
162	case 0xf3:
163	f = PREFIX_REP; /* yes, both */
164	break;
165
166	case 0x2e:
167	case 0x3e:
168	case 0x26:
169	case 0x36:
170	case 0x64:
171	case 0x65:
172	f = PREFIX_SEG;
173	break;
174
175	case 0x66:
176	f = PREFIX_OPSIZE;
177	break;
178
179	case 0x67:
180	f = PREFIX_ADDRSIZE;
181	break;
182
183	case 0x40: case 0x41: case 0x42: case 0x43: case 0x44: case 0x45: case 0x46: case 0x47:
184	case 0x48: case 0x49: case 0x4a: case 0x4b: case 0x4c: case 0x4d: case 0x4e: case 0x4f:
185	f = pState->Cpu.mode == CPUMODE_64BIT ? PREFIX_REX : 0;
186	break;
187
188	default:
189	f = 0;
190	break;
191	}
192	if (!f)
193	break; /* done */
194	if (fPrefixes & f)
195	return true;
196	fPrefixes \|= f;
197	}
198
199	/* segment overrides are fun */
200	if (fPrefixes & PREFIX_SEG)
201	{
202	/* no effective address which it may apply to. */
203	Assert((pState->Cpu.prefix & PREFIX_SEG) \|\| pState->Cpu.mode == CPUMODE_64BIT);
204	if ( !DIS_IS_EFFECTIVE_ADDR(pState->Cpu.param1.flags)
205	&& !DIS_IS_EFFECTIVE_ADDR(pState->Cpu.param2.flags)
206	&& !DIS_IS_EFFECTIVE_ADDR(pState->Cpu.param3.flags))
207	return true;
208	}
209
210	/* fixed register + addr override doesn't go down all that well. */
211	if (fPrefixes & PREFIX_ADDRSIZE)
212	{
213	Assert(pState->Cpu.prefix & PREFIX_ADDRSIZE);
214	if ( pState->Cpu.pCurInstr->param3 == OP_PARM_NONE
215	&& pState->Cpu.pCurInstr->param2 == OP_PARM_NONE
216	&& ( pState->Cpu.pCurInstr->param1 >= OP_PARM_REG_GEN32_START
217	&& pState->Cpu.pCurInstr->param1 <= OP_PARM_REG_GEN32_END))
218	return true;
219	}
220
221	/* Almost all prefixes are bad. */
222	if (fPrefixes)
223	{
224	switch (pState->Cpu.pCurInstr->opcode)
225	{
226	/* nop w/ prefix(es). */
227	case OP_NOP:
228	return true;
229
230	case OP_JMP:
231	if ( pState->Cpu.pCurInstr->param1 != OP_PARM_Jb
232	&& pState->Cpu.pCurInstr->param1 != OP_PARM_Jv)
233	break;
234	/* fall thru */
235	case OP_JO:
236	case OP_JNO:
237	case OP_JC:
238	case OP_JNC:
239	case OP_JE:
240	case OP_JNE:
241	case OP_JBE:
242	case OP_JNBE:
243	case OP_JS:
244	case OP_JNS:
245	case OP_JP:
246	case OP_JNP:
247	case OP_JL:
248	case OP_JNL:
249	case OP_JLE:
250	case OP_JNLE:
251	/** @todo branch hinting 0x2e/0x3e... */
252	return true;
253	}
254
255	}
256
257	/* All but the segment prefix is bad news. */
258	if (fPrefixes & ~PREFIX_SEG)
259	{
260	switch (pState->Cpu.pCurInstr->opcode)
261	{
262	case OP_POP:
263	case OP_PUSH:
264	if ( pState->Cpu.pCurInstr->param1 >= OP_PARM_REG_SEG_START
265	&& pState->Cpu.pCurInstr->param1 <= OP_PARM_REG_SEG_END)
266	return true;
267	if ( (fPrefixes & ~PREFIX_OPSIZE)
268	&& pState->Cpu.pCurInstr->param1 >= OP_PARM_REG_GEN32_START
269	&& pState->Cpu.pCurInstr->param1 <= OP_PARM_REG_GEN32_END)
270	return true;
271	break;
272
273	case OP_POPA:
274	case OP_POPF:
275	case OP_PUSHA:
276	case OP_PUSHF:
277	if (fPrefixes & ~PREFIX_OPSIZE)
278	return true;
279	break;
280	}
281	}
282
283	/* Implicit 8-bit register instructions doesn't mix with operand size. */
284	if ( (fPrefixes & PREFIX_OPSIZE)
285	&& ( ( pState->Cpu.pCurInstr->param1 == OP_PARM_Gb /* r8 */
286	&& pState->Cpu.pCurInstr->param2 == OP_PARM_Eb /* r8/mem8 */)
287	\|\| ( pState->Cpu.pCurInstr->param2 == OP_PARM_Gb /* r8 */
288	&& pState->Cpu.pCurInstr->param1 == OP_PARM_Eb /* r8/mem8 */))
289	)
290	{
291	switch (pState->Cpu.pCurInstr->opcode)
292	{
293	case OP_ADD:
294	case OP_OR:
295	case OP_ADC:
296	case OP_SBB:
297	case OP_AND:
298	case OP_SUB:
299	case OP_XOR:
300	case OP_CMP:
301	return true;
302	default:
303	break;
304	}
305	}
306
307
308	/*
309	* Check for the version of xyz reg,reg instruction that the assembler doesn't use.
310	*
311	* For example:
312	* expected: 1aee sbb ch, dh ; SBB r8, r/m8
313	* yasm: 18F5 sbb ch, dh ; SBB r/m8, r8
314	*/
315	if (pState->Cpu.ModRM.Bits.Mod == 3 /* reg,reg */)
316	{
317	switch (pState->Cpu.pCurInstr->opcode)
318	{
319	case OP_ADD:
320	case OP_OR:
321	case OP_ADC:
322	case OP_SBB:
323	case OP_AND:
324	case OP_SUB:
325	case OP_XOR:
326	case OP_CMP:
327	if ( ( pState->Cpu.pCurInstr->param1 == OP_PARM_Gb /* r8 */
328	&& pState->Cpu.pCurInstr->param2 == OP_PARM_Eb /* r8/mem8 */)
329	\|\| ( pState->Cpu.pCurInstr->param1 == OP_PARM_Gv /* rX */
330	&& pState->Cpu.pCurInstr->param2 == OP_PARM_Ev /* rX/memX */))
331	return true;
332
333	/* 82 (see table A-6). */
334	if (pState->Cpu.opcode == 0x82)
335	return true;
336	break;
337
338	/* ff /0, fe /0, ff /1, fe /0 */
339	case OP_DEC:
340	case OP_INC:
341	return true;
342
343	case OP_POP:
344	case OP_PUSH:
345	Assert(pState->Cpu.opcode == 0x8f);
346	return true;
347
348	default:
349	break;
350	}
351	}
352
353	/* shl eax,1 will be assembled to the form without the immediate byte. */
354	if ( pState->Cpu.pCurInstr->param2 == OP_PARM_Ib
355	&& (uint8_t)pState->Cpu.param2.parval == 1)
356	{
357	switch (pState->Cpu.pCurInstr->opcode)
358	{
359	case OP_SHL:
360	case OP_SHR:
361	case OP_SAR:
362	case OP_RCL:
363	case OP_RCR:
364	case OP_ROL:
365	case OP_ROR:
366	return true;
367	}
368	}
369
370	/* And some more - see table A-6. */
371	if (pState->Cpu.opcode == 0x82)
372	{
373	switch (pState->Cpu.pCurInstr->opcode)
374	{
375	case OP_ADD:
376	case OP_OR:
377	case OP_ADC:
378	case OP_SBB:
379	case OP_AND:
380	case OP_SUB:
381	case OP_XOR:
382	case OP_CMP:
383	return true;
384	break;
385	}
386	}
387
388
389	/* check for REX.X = 1 without SIB. */
390
391	/* Yasm encodes setnbe al with /2 instead of /0 like the AMD manual
392	says (intel doesn't appear to care). */
393	switch (pState->Cpu.pCurInstr->opcode)
394	{
395	case OP_SETO:
396	case OP_SETNO:
397	case OP_SETC:
398	case OP_SETNC:
399	case OP_SETE:
400	case OP_SETNE:
401	case OP_SETBE:
402	case OP_SETNBE:
403	case OP_SETS:
404	case OP_SETNS:
405	case OP_SETP:
406	case OP_SETNP:
407	case OP_SETL:
408	case OP_SETNL:
409	case OP_SETLE:
410	case OP_SETNLE:
411	AssertMsg(pState->Cpu.opcode >= 0x90 && pState->Cpu.opcode <= 0x9f, ("%#x\n", pState->Cpu.opcode));
412	if (pState->Cpu.ModRM.Bits.Reg != 2)
413	return true;
414	break;
415	}
416
417	/*
418	* The MOVZX reg32,mem16 instruction without an operand size prefix
419	* doesn't quite make sense...
420	*/
421	if ( pState->Cpu.pCurInstr->opcode == OP_MOVZX
422	&& pState->Cpu.opcode == 0xB7
423	&& (pState->Cpu.mode == CPUMODE_16BIT) != !!(fPrefixes & PREFIX_OPSIZE))
424	return true;
425
426	return false;
427	}
428
429
430	/**
431	* Masm style.
432	*
433	* @param pState The disassembler state.
434	*/
435	static void MyDisasMasmFormatter(PMYDISSTATE pState)
436	{
437	RTPrintf("masm not implemented: %s", pState->szLine);
438	}
439
440
441	/**
442	* This is a temporary workaround for catching a few illegal opcodes
443	* that the disassembler is currently letting thru, just enough to make
444	* the assemblers happy.
445	*
446	* We're too close to a release to dare mess with these things now as
447	* they may consequences for performance and let alone introduce bugs.
448	*
449	* @returns true if it's valid. false if it isn't.
450	*
451	* @param pCpu The disassembler output.
452	*/
453	static bool MyDisasIsValidInstruction(DISCPUSTATE const *pCpu)
454	{
455	switch (pCpu->pCurInstr->opcode)
456	{
457	/* These doesn't take memory operands. */
458	case OP_MOV_CR:
459	case OP_MOV_DR:
460	case OP_MOV_TR:
461	if (pCpu->ModRM.Bits.Mod != 3)
462	return false;
463	break;
464
465	/* The 0x8f /0 variant of this instruction doesn't get its /r value verified. */
466	case OP_POP:
467	if ( pCpu->opcode == 0x8f
468	&& pCpu->ModRM.Bits.Reg != 0)
469	return false;
470	break;
471
472	/* The 0xc6 /0 and 0xc7 /0 variants of this instruction don't get their /r values verified. */
473	case OP_MOV:
474	if ( ( pCpu->opcode == 0xc6
475	\|\| pCpu->opcode == 0xc7)
476	&& pCpu->ModRM.Bits.Reg != 0)
477	return false;
478	break;
479
480	default:
481	break;
482	}
483
484	return true;
485	}
486
487
488	/**
489	* Callback for reading bytes.
490	*
491	* @todo This should check that the disassembler doesn't do unnecessary reads,
492	* however the current doesn't do this and is just complicated...
493	*/
494	static DECLCALLBACK(int) MyDisasInstrRead(RTUINTPTR uSrcAddr, uint8_t pbDst, uint32_t cbRead, void pvDisCpu)
495	{
496	PMYDISSTATE pState = (PMYDISSTATE)pvDisCpu;
497	if (RT_LIKELY( pState->uNextAddr == uSrcAddr
498	&& pState->cbLeft >= cbRead))
499	{
500	/*
501	* Straight forward reading.
502	*/
503	if (cbRead == 1)
504	{
505	pState->cbLeft--;
506	pbDst = pState->pbNext++;
507	pState->uNextAddr++;
508	}
509	else
510	{
511	memcpy(pbDst, pState->pbNext, cbRead);
512	pState->pbNext += cbRead;
513	pState->cbLeft -= cbRead;
514	pState->uNextAddr += cbRead;
515	}
516	}
517	else
518	{
519	/*
520	* Jumping up the stream.
521	* This occures when the byte sequence is added to the output string.
522	*/
523	uint64_t offReq64 = uSrcAddr - pState->uAddress;
524	if (offReq64 < 32)
525	{
526	uint32_t offReq = offReq64;
527	uintptr_t off = pState->pbNext - pState->pbInstr;
528	if (off + pState->cbLeft <= offReq)
529	{
530	pState->pbNext += pState->cbLeft;
531	pState->uNextAddr += pState->cbLeft;
532	pState->cbLeft = 0;
533
534	memset(pbDst, 0xcc, cbRead);
535	pState->rc = VERR_EOF;
536	return VERR_EOF;
537	}
538
539	/* reset the stream. */
540	pState->cbLeft += off;
541	pState->pbNext = pState->pbInstr;
542	pState->uNextAddr = pState->uAddress;
543
544	/* skip ahead. */
545	pState->cbLeft -= offReq;
546	pState->pbNext += offReq;
547	pState->uNextAddr += offReq;
548
549	/* do the reading. */
550	if (pState->cbLeft >= cbRead)
551	{
552	memcpy(pbDst, pState->pbNext, cbRead);
553	pState->cbLeft -= cbRead;
554	pState->pbNext += cbRead;
555	pState->uNextAddr += cbRead;
556	}
557	else
558	{
559	if (pState->cbLeft > 0)
560	{
561	memcpy(pbDst, pState->pbNext, pState->cbLeft);
562	pbDst += pState->cbLeft;
563	cbRead -= (uint32_t)pState->cbLeft;
564	pState->pbNext += pState->cbLeft;
565	pState->uNextAddr += pState->cbLeft;
566	pState->cbLeft = 0;
567	}
568	memset(pbDst, 0xcc, cbRead);
569	pState->rc = VERR_EOF;
570	return VERR_EOF;
571	}
572	}
573	else
574	{
575	RTStrmPrintf(g_pStdErr, "Reading before current instruction!\n");
576	memset(pbDst, 0x90, cbRead);
577	pState->rc = VERR_INTERNAL_ERROR;
578	return VERR_INTERNAL_ERROR;
579	}
580	}
581
582	return VINF_SUCCESS;
583	}
584
585
586	/**
587	* Disassembles a block of memory.
588	*
589	* @returns VBox status code.
590	* @param argv0 Program name (for errors and warnings).
591	* @param enmCpuMode The cpu mode to disassemble in.
592	* @param uAddress The address we're starting to disassemble at.
593	* @param pbFile Where to start disassemble.
594	* @param cbFile How much to disassemble.
595	* @param enmStyle The assembly output style.
596	* @param fListing Whether to print in a listing like mode.
597	* @param enmUndefOp How to deal with undefined opcodes.
598	*/
599	static int MyDisasmBlock(const char argv0, DISCPUMODE enmCpuMode, uint64_t uAddress, uint8_t pbFile, size_t cbFile,
600	ASMSTYLE enmStyle, bool fListing, UNDEFOPHANDLING enmUndefOp)
601	{
602	/*
603	* Initialize the CPU context.
604	*/
605	MYDISSTATE State;
606	State.Cpu.mode = enmCpuMode;
607	State.Cpu.pfnReadBytes = MyDisasInstrRead;
608	State.uAddress = uAddress;
609	State.pbInstr = pbFile;
610	State.cbInstr = 0;
611	State.enmUndefOp = enmUndefOp;
612	State.rc = VINF_SUCCESS;
613	State.cbLeft = cbFile;
614	State.pbNext = pbFile;
615	State.uNextAddr = uAddress;
616
617	void (*pfnFormatter)(PMYDISSTATE pState);
618	switch (enmStyle)
619	{
620	case kAsmStyle_Default:
621	pfnFormatter = MyDisasDefaultFormatter;
622	break;
623
624	case kAsmStyle_yasm:
625	RTPrintf(" BITS %d\n", enmCpuMode == CPUMODE_16BIT ? 16 : enmCpuMode == CPUMODE_32BIT ? 32 : 64);
626	pfnFormatter = MyDisasYasmFormatter;
627	break;
628
629	case kAsmStyle_masm:
630	pfnFormatter = MyDisasMasmFormatter;
631	break;
632
633	default:
634	AssertFailedReturn(VERR_INTERNAL_ERROR);
635	}
636
637	/*
638	* The loop.
639	*/
640	int rcRet = VINF_SUCCESS;
641	while (State.cbLeft > 0)
642	{
643	/*
644	* Disassemble it.
645	*/
646	State.cbInstr = 0;
647	State.cbLeft += State.pbNext - State.pbInstr;
648	State.uNextAddr = State.uAddress;
649	State.pbNext = State.pbInstr;
650
651	int rc = DISInstr(&State.Cpu, State.uAddress, 0, &State.cbInstr, State.szLine);
652	if ( RT_SUCCESS(rc)
653	\|\| ( ( rc == VERR_DIS_INVALID_OPCODE
654	\|\| rc == VERR_DIS_GEN_FAILURE)
655	&& State.enmUndefOp == kUndefOp_DefineByte))
656	{
657	State.fUndefOp = rc == VERR_DIS_INVALID_OPCODE
658	\|\| rc == VERR_DIS_GEN_FAILURE
659	\|\| State.Cpu.pCurInstr->opcode == OP_INVALID
660	\|\| State.Cpu.pCurInstr->opcode == OP_ILLUD2
661	\|\| ( State.enmUndefOp == kUndefOp_DefineByte
662	&& !MyDisasIsValidInstruction(&State.Cpu));
663	if (State.fUndefOp && State.enmUndefOp == kUndefOp_DefineByte)
664	{
665	RTPrintf(" db");
666	if (!State.cbInstr)
667	State.cbInstr = 1;
668	for (unsigned off = 0; off < State.cbInstr; off++)
669	{
670	uint8_t b;
671	State.Cpu.pfnReadBytes(State.uAddress + off, &b, 1, &State.Cpu);
672	RTPrintf(off ? ", %03xh" : " %03xh", b);
673	}
674	RTPrintf(" ; %s\n", State.szLine);
675	}
676	else if (!State.fUndefOp && State.enmUndefOp == kUndefOp_All)
677	{
678	RTPrintf("%s: error at %#RX64: unexpected valid instruction (op=%d)\n", argv0, State.uAddress, State.Cpu.pCurInstr->opcode);
679	pfnFormatter(&State);
680	rcRet = VERR_GENERAL_FAILURE;
681	}
682	else if (State.fUndefOp && State.enmUndefOp == kUndefOp_Fail)
683	{
684	RTPrintf("%s: error at %#RX64: undefined opcode (op=%d)\n", argv0, State.uAddress, State.Cpu.pCurInstr->opcode);
685	pfnFormatter(&State);
686	rcRet = VERR_GENERAL_FAILURE;
687	}
688	else
689	{
690	/* Use db for odd encodings that we can't make the assembler use. */
691	if ( State.enmUndefOp == kUndefOp_DefineByte
692	&& MyDisasYasmFormatterIsOddEncoding(&State))
693	{
694	RTPrintf(" db");
695	for (unsigned off = 0; off < State.cbInstr; off++)
696	{
697	uint8_t b;
698	State.Cpu.pfnReadBytes(State.uAddress + off, &b, 1, &State.Cpu);
699	RTPrintf(off ? ", %03xh" : " %03xh", b);
700	}
701	RTPrintf(" ; ");
702	}
703
704	pfnFormatter(&State);
705	}
706	}
707	else
708	{
709	State.cbInstr = State.pbNext - State.pbInstr;
710	if (!State.cbLeft)
711	RTPrintf("%s: error at %#RX64: read beyond the end (%Rrc)\n", argv0, State.uAddress, rc);
712	else if (State.cbInstr)
713	RTPrintf("%s: error at %#RX64: %Rrc cbInstr=%d\n", argv0, State.uAddress, rc, State.cbInstr);
714	else
715	{
716	RTPrintf("%s: error at %#RX64: %Rrc cbInstr=%d!\n", argv0, State.uAddress, rc, State.cbInstr);
717	if (rcRet == VINF_SUCCESS)
718	rcRet = rc;
719	break;
720	}
721	}
722
723
724	/* next */
725	State.uAddress += State.cbInstr;
726	State.pbInstr += State.cbInstr;
727	}
728
729	return rcRet;
730	}
731
732	/**
733	* Converts a hex char to a number.
734	*
735	* @returns 0..15 on success, -1 on failure.
736	* @param ch The character.
737	*/
738	static int HexDigitToNum(char ch)
739	{
740	switch (ch)
741	{
742	case '0': return 0;
743	case '1': return 1;
744	case '2': return 2;
745	case '3': return 3;
746	case '4': return 4;
747	case '5': return 5;
748	case '6': return 6;
749	case '7': return 7;
750	case '8': return 8;
751	case '9': return 9;
752	case 'A':
753	case 'a': return 0xa;
754	case 'B':
755	case 'b': return 0xb;
756	case 'C':
757	case 'c': return 0xc;
758	case 'D':
759	case 'd': return 0xd;
760	case 'E':
761	case 'e': return 0xe;
762	case 'F':
763	case 'f': return 0xf;
764	default:
765	RTPrintf("error: Invalid hex digig '%c'\n", ch);
766	return -1;
767	}
768	}
769
770	/**
771	* Prints usage info.
772	*
773	* @returns 1.
774	* @param argv0 The program name.
775	*/
776	static int Usage(const char *argv0)
777	{
778	RTStrmPrintf(g_pStdErr,
779	"usage: %s [options] <file1> [file2..fileN]\n"
780	" or: %s [options] <-x\|--hex-bytes> <hex byte> [more hex..]\n"
781	" or: %s <--help\|-h>\n"
782	"\n"
783	"Options:\n"
784	" --address\|-a <address>\n"
785	" The base address. Default: 0\n"
786	" --max-bytes\|-b <bytes>\n"
787	" The maximum number of bytes to disassemble. Default: 1GB\n"
788	" --cpumode\|-c <16\|32\|64>\n"
789	" The cpu mode. Default: 32\n"
790	" --listing\|-l, --no-listing\|-L\n"
791	" Enables or disables listing mode. Default: --no-listing\n"
792	" --offset\|-o <offset>\n"
793	" The file offset at which to start disassembling. Default: 0\n"
794	" --style\|-s <default\|yasm\|masm>\n"
795	" The assembly output style. Default: default\n"
796	" --undef-op\|-u <fail\|all\|db>\n"
797	" How to treat undefined opcodes. Default: fail\n"
798	, argv0, argv0);
799	return 1;
800	}
801
802
803	int main(int argc, char **argv)
804	{
805	RTR3Init();
806	const char * const argv0 = RTPathFilename(argv[0]);
807
808	/* options */
809	uint64_t uAddress = 0;
810	ASMSTYLE enmStyle = kAsmStyle_Default;
811	UNDEFOPHANDLING enmUndefOp = kUndefOp_Fail;
812	bool fListing = true;
813	DISCPUMODE enmCpuMode = CPUMODE_32BIT;
814	RTFOFF off = 0;
815	RTFOFF cbMax = _1G;
816	bool fHexBytes = false;
817
818	/*
819	* Parse arguments.
820	*/
821	static const RTGETOPTDEF g_aOptions[] =
822	{
823	{ "--address", 'a', RTGETOPT_REQ_UINT64 },
824	{ "--cpumode", 'c', RTGETOPT_REQ_UINT32 },
825	{ "--bytes", 'b', RTGETOPT_REQ_INT64 },
826	{ "--listing", 'l', RTGETOPT_REQ_NOTHING },
827	{ "--no-listing", 'L', RTGETOPT_REQ_NOTHING },
828	{ "--offset", 'o', RTGETOPT_REQ_INT64 },
829	{ "--style", 's', RTGETOPT_REQ_STRING },
830	{ "--undef-op", 'u', RTGETOPT_REQ_STRING },
831	{ "--hex-bytes", 'x', RTGETOPT_REQ_NOTHING },
832	};
833
834	int ch;
835	RTGETOPTUNION ValueUnion;
836	RTGETOPTSTATE GetState;
837	RTGetOptInit(&GetState, argc, argv, g_aOptions, RT_ELEMENTS(g_aOptions), 1, RTGETOPTINIT_FLAGS_OPTS_FIRST);
838	while ( (ch = RTGetOpt(&GetState, &ValueUnion))
839	&& ch != VINF_GETOPT_NOT_OPTION)
840	{
841	switch (ch)
842	{
843	case 'a':
844	uAddress = ValueUnion.u64;
845	break;
846
847	case 'b':
848	cbMax = ValueUnion.i;
849	break;
850
851	case 'c':
852	if (ValueUnion.u32 == 16)
853	enmCpuMode = CPUMODE_16BIT;
854	else if (ValueUnion.u32 == 32)
855	enmCpuMode = CPUMODE_32BIT;
856	else if (ValueUnion.u32 == 64)
857	enmCpuMode = CPUMODE_64BIT;
858	else
859	{
860	RTStrmPrintf(g_pStdErr, "%s: Invalid CPU mode value %RU32\n", argv0, ValueUnion.u32);
861	return 1;
862	}
863	break;
864
865	case 'h':
866	return Usage(argv0);
867
868	case 'l':
869	fListing = true;
870	break;
871
872	case 'L':
873	fListing = false;
874	break;
875
876	case 'o':
877	off = ValueUnion.i;
878	break;
879
880	case 's':
881	if (!strcmp(ValueUnion.psz, "default"))
882	enmStyle = kAsmStyle_Default;
883	else if (!strcmp(ValueUnion.psz, "yasm"))
884	enmStyle = kAsmStyle_yasm;
885	else if (!strcmp(ValueUnion.psz, "masm"))
886	{
887	enmStyle = kAsmStyle_masm;
888	RTStrmPrintf(g_pStdErr, "%s: masm style isn't implemented yet\n", argv0);
889	return 1;
890	}
891	else
892	{
893	RTStrmPrintf(g_pStdErr, "%s: unknown assembly style: %s\n", argv0, ValueUnion.psz);
894	return 1;
895	}
896	break;
897
898	case 'u':
899	if (!strcmp(ValueUnion.psz, "fail"))
900	enmUndefOp = kUndefOp_Fail;
901	else if (!strcmp(ValueUnion.psz, "all"))
902	enmUndefOp = kUndefOp_All;
903	else if (!strcmp(ValueUnion.psz, "db"))
904	enmUndefOp = kUndefOp_DefineByte;
905	else
906	{
907	RTStrmPrintf(g_pStdErr, "%s: unknown undefined opcode handling method: %s\n", argv0, ValueUnion.psz);
908	return 1;
909	}
910	break;
911
912	case 'x':
913	fHexBytes = true;
914	break;
915
916	case 'V':
917	RTPrintf("$Revision: $\n");
918	return 0;
919
920	default:
921	return RTGetOptPrintError(ch, &ValueUnion);
922	}
923	}
924	int iArg = GetState.iNext - 1; /** @todo Not pretty, add RTGetOptInit flag for this. */
925	if (iArg >= argc)
926	return Usage(argv0);
927
928	int rc = VINF_SUCCESS;
929	if (fHexBytes)
930	{
931	/*
932	* Convert the remaining arguments from a hex byte string into
933	* a buffer that we disassemble.
934	*/
935	size_t cb = 0;
936	uint8_t *pb = NULL;
937	for ( ; iArg < argc; iArg++)
938	{
939	const char *psz = argv[iArg];
940	while (*psz)
941	{
942	/** @todo this stuff belongs in IPRT, same stuff as mac address reading. Could be reused for IPv6 with a different item size.*/
943	/* skip white space */
944	while (RT_C_IS_SPACE(*psz))
945	psz++;
946	if (!*psz)
947	break;
948
949	/* one digit followed by a space or EOS, or two digits. */
950	int iNum = HexDigitToNum(*psz++);
951	if (iNum == -1)
952	return 1;
953	if (!RT_C_IS_SPACE(psz) && psz)
954	{
955	int iDigit = HexDigitToNum(*psz++);
956	if (iDigit == -1)
957	return 1;
958	iNum = iNum * 16 + iDigit;
959	}
960
961	/* add the byte */
962	if (!(cb % 4 /64/))
963	{
964	pb = (uint8_t *)RTMemRealloc(pb, cb + 64);
965	if (!pb)
966	{
967	RTPrintf("%s: error: RTMemRealloc failed\n", argv[0]);
968	return 1;
969	}
970	}
971	pb[cb++] = (uint8_t)iNum;
972	}
973	}
974
975	/*
976	* Disassemble it.
977	*/
978	rc = MyDisasmBlock(argv0, enmCpuMode, uAddress, pb, cb, enmStyle, fListing, enmUndefOp);
979	}
980	else
981	{
982	/*
983	* Process the files.
984	*/
985	for ( ; iArg < argc; iArg++)
986	{
987	/*
988	* Read the file into memory.
989	*/
990	void *pvFile;
991	size_t cbFile;
992	rc = RTFileReadAllEx(argv[iArg], off, cbMax, RTFILE_RDALL_O_DENY_NONE, &pvFile, &cbFile);
993	if (RT_FAILURE(rc))
994	{
995	RTStrmPrintf(g_pStdErr, "%s: %s: %Rrc\n", argv0, argv[iArg], rc);
996	break;
997	}
998
999	/*
1000	* Disassemble it.
1001	*/
1002	rc = MyDisasmBlock(argv0, enmCpuMode, uAddress, (uint8_t *)pvFile, cbFile, enmStyle, fListing, enmUndefOp);
1003	if (RT_FAILURE(rc))
1004	break;
1005	}
1006	}
1007
1008	return RT_SUCCESS(rc) ? 0 : 1;
1009	}
1010

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

source: vbox/trunk/src/VBox/Disassembler/testcase/tstDisasm-2.cpp@ 27023

Download in other formats: