DisasmFormatYasm.cpp@ 62956

Last change on this file since 62956 was 62594, checked in by vboxsync, 8 years ago
warnings
Property svn:eol-style set to `native` Property svn:keywords set to `Id Revision`
File size: 70.1 KB

Line
1	/* $Id: DisasmFormatYasm.cpp 62594 2016-07-27 14:32:14Z vboxsync $ */
2	/** @file
3	* VBox Disassembler - Yasm(/Nasm) Style Formatter.
4	*/
5
6	/*
7	* Copyright (C) 2008-2016 Oracle Corporation
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
18
19	/*********************************************************************************************************************************
20	* Header Files *
21	*********************************************************************************************************************************/
22	#include <VBox/dis.h>
23	#include "DisasmInternal.h"
24	#include <iprt/string.h>
25	#include <iprt/assert.h>
26	#include <iprt/ctype.h>
27
28
29	/*********************************************************************************************************************************
30	* Global Variables *
31	*********************************************************************************************************************************/
32	static const char g_szSpaces[] =
33	" ";
34	static const char g_aszYasmRegGen8[20][5] =
35	{
36	"al\0\0", "cl\0\0", "dl\0\0", "bl\0\0", "ah\0\0", "ch\0\0", "dh\0\0", "bh\0\0", "r8b\0", "r9b\0", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b", "spl\0", "bpl\0", "sil\0", "dil\0"
37	};
38	static const char g_aszYasmRegGen16[16][5] =
39	{
40	"ax\0\0", "cx\0\0", "dx\0\0", "bx\0\0", "sp\0\0", "bp\0\0", "si\0\0", "di\0\0", "r8w\0", "r9w\0", "r10w", "r11w", "r12w", "r13w", "r14w", "r15w"
41	};
42	static const char g_aszYasmRegGen1616[8][6] =
43	{
44	"bx+si", "bx+di", "bp+si", "bp+di", "si\0\0\0", "di\0\0\0", "bp\0\0\0", "bx\0\0\0"
45	};
46	static const char g_aszYasmRegGen32[16][5] =
47	{
48	"eax\0", "ecx\0", "edx\0", "ebx\0", "esp\0", "ebp\0", "esi\0", "edi\0", "r8d\0", "r9d\0", "r10d", "r11d", "r12d", "r13d", "r14d", "r15d"
49	};
50	static const char g_aszYasmRegGen64[16][4] =
51	{
52	"rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi", "r8\0", "r9\0", "r10", "r11", "r12", "r13", "r14", "r15"
53	};
54	static const char g_aszYasmRegSeg[6][3] =
55	{
56	"es", "cs", "ss", "ds", "fs", "gs"
57	};
58	static const char g_aszYasmRegFP[8][4] =
59	{
60	"st0", "st1", "st2", "st3", "st4", "st5", "st6", "st7"
61	};
62	static const char g_aszYasmRegMMX[8][4] =
63	{
64	"mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7"
65	};
66	static const char g_aszYasmRegXMM[16][6] =
67	{
68	"xmm0\0", "xmm1\0", "xmm2\0", "xmm3\0", "xmm4\0", "xmm5\0", "xmm6\0", "xmm7\0", "xmm8\0", "xmm9\0", "xmm10", "xmm11", "xmm12", "xmm13", "xmm14", "xmm15"
69	};
70	static const char g_aszYasmRegYMM[16][6] =
71	{
72	"ymm0\0", "ymm1\0", "ymm2\0", "ymm3\0", "ymm4\0", "ymm5\0", "ymm6\0", "ymm7\0", "ymm8\0", "ymm9\0", "ymm10", "ymm11", "ymm12", "ymm13", "ymm14", "ymm15"
73	};
74	static const char g_aszYasmRegCRx[16][5] =
75	{
76	"cr0\0", "cr1\0", "cr2\0", "cr3\0", "cr4\0", "cr5\0", "cr6\0", "cr7\0", "cr8\0", "cr9\0", "cr10", "cr11", "cr12", "cr13", "cr14", "cr15"
77	};
78	static const char g_aszYasmRegDRx[16][5] =
79	{
80	"dr0\0", "dr1\0", "dr2\0", "dr3\0", "dr4\0", "dr5\0", "dr6\0", "dr7\0", "dr8\0", "dr9\0", "dr10", "dr11", "dr12", "dr13", "dr14", "dr15"
81	};
82	static const char g_aszYasmRegTRx[16][5] =
83	{
84	"tr0\0", "tr1\0", "tr2\0", "tr3\0", "tr4\0", "tr5\0", "tr6\0", "tr7\0", "tr8\0", "tr9\0", "tr10", "tr11", "tr12", "tr13", "tr14", "tr15"
85	};
86
87
88
89	/**
90	* Gets the base register name for the given parameter.
91	*
92	* @returns Pointer to the register name.
93	* @param pDis The disassembler state.
94	* @param pParam The parameter.
95	* @param pcchReg Where to store the length of the name.
96	*/
97	static const char disasmFormatYasmBaseReg(PCDISSTATE pDis, PCDISOPPARAM pParam, size_t pcchReg)
98	{
99	RT_NOREF_PV(pDis);
100
101	switch (pParam->fUse & ( DISUSE_REG_GEN8 \| DISUSE_REG_GEN16 \| DISUSE_REG_GEN32 \| DISUSE_REG_GEN64
102	\| DISUSE_REG_FP \| DISUSE_REG_MMX \| DISUSE_REG_XMM \| DISUSE_REG_YMM
103	\| DISUSE_REG_CR \| DISUSE_REG_DBG \| DISUSE_REG_SEG \| DISUSE_REG_TEST))
104
105	{
106	case DISUSE_REG_GEN8:
107	{
108	Assert(pParam->Base.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen8));
109	const char *psz = g_aszYasmRegGen8[pParam->Base.idxGenReg];
110	*pcchReg = 2 + !!psz[2] + !!psz[3];
111	return psz;
112	}
113
114	case DISUSE_REG_GEN16:
115	{
116	Assert(pParam->Base.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen16));
117	const char *psz = g_aszYasmRegGen16[pParam->Base.idxGenReg];
118	*pcchReg = 2 + !!psz[2] + !!psz[3];
119	return psz;
120	}
121
122	// VSIB
123	case DISUSE_REG_XMM \| DISUSE_REG_GEN32:
124	case DISUSE_REG_YMM \| DISUSE_REG_GEN32:
125	case DISUSE_REG_GEN32:
126	{
127	Assert(pParam->Base.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen32));
128	const char *psz = g_aszYasmRegGen32[pParam->Base.idxGenReg];
129	*pcchReg = 2 + !!psz[2] + !!psz[3];
130	return psz;
131	}
132
133	// VSIB
134	case DISUSE_REG_XMM \| DISUSE_REG_GEN64:
135	case DISUSE_REG_YMM \| DISUSE_REG_GEN64:
136	case DISUSE_REG_GEN64:
137	{
138	Assert(pParam->Base.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen64));
139	const char *psz = g_aszYasmRegGen64[pParam->Base.idxGenReg];
140	*pcchReg = 2 + !!psz[2] + !!psz[3];
141	return psz;
142	}
143
144	case DISUSE_REG_FP:
145	{
146	Assert(pParam->Base.idxFpuReg < RT_ELEMENTS(g_aszYasmRegFP));
147	const char *psz = g_aszYasmRegFP[pParam->Base.idxFpuReg];
148	*pcchReg = 3;
149	return psz;
150	}
151
152	case DISUSE_REG_MMX:
153	{
154	Assert(pParam->Base.idxMmxReg < RT_ELEMENTS(g_aszYasmRegMMX));
155	const char *psz = g_aszYasmRegMMX[pParam->Base.idxMmxReg];
156	*pcchReg = 3;
157	return psz;
158	}
159
160	case DISUSE_REG_XMM:
161	{
162	Assert(pParam->Base.idxXmmReg < RT_ELEMENTS(g_aszYasmRegXMM));
163	const char *psz = g_aszYasmRegXMM[pParam->Base.idxXmmReg];
164	*pcchReg = 4 + !!psz[4];
165	return psz;
166	}
167
168	case DISUSE_REG_YMM:
169	{
170	Assert(pParam->Base.idxYmmReg < RT_ELEMENTS(g_aszYasmRegYMM));
171	const char *psz = g_aszYasmRegYMM[pParam->Base.idxYmmReg];
172	*pcchReg = 4 + !!psz[4];
173	return psz;
174	}
175
176	case DISUSE_REG_CR:
177	{
178	Assert(pParam->Base.idxCtrlReg < RT_ELEMENTS(g_aszYasmRegCRx));
179	const char *psz = g_aszYasmRegCRx[pParam->Base.idxCtrlReg];
180	*pcchReg = 3;
181	return psz;
182	}
183
184	case DISUSE_REG_DBG:
185	{
186	Assert(pParam->Base.idxDbgReg < RT_ELEMENTS(g_aszYasmRegDRx));
187	const char *psz = g_aszYasmRegDRx[pParam->Base.idxDbgReg];
188	*pcchReg = 3;
189	return psz;
190	}
191
192	case DISUSE_REG_SEG:
193	{
194	Assert(pParam->Base.idxSegReg < RT_ELEMENTS(g_aszYasmRegCRx));
195	const char *psz = g_aszYasmRegSeg[pParam->Base.idxSegReg];
196	*pcchReg = 2;
197	return psz;
198	}
199
200	case DISUSE_REG_TEST:
201	{
202	Assert(pParam->Base.idxTestReg < RT_ELEMENTS(g_aszYasmRegTRx));
203	const char *psz = g_aszYasmRegTRx[pParam->Base.idxTestReg];
204	*pcchReg = 3;
205	return psz;
206	}
207
208	default:
209	AssertMsgFailed(("%#x\n", pParam->fUse));
210	*pcchReg = 3;
211	return "r??";
212	}
213	}
214
215
216	/**
217	* Gets the index register name for the given parameter.
218	*
219	* @returns The index register name.
220	* @param pDis The disassembler state.
221	* @param pParam The parameter.
222	* @param pcchReg Where to store the length of the name.
223	*/
224	static const char disasmFormatYasmIndexReg(PCDISSTATE pDis, PCDISOPPARAM pParam, size_t pcchReg)
225	{
226	if (pParam->fUse & DISUSE_REG_XMM)
227	{
228	Assert(pParam->Index.idxXmmReg < RT_ELEMENTS(g_aszYasmRegXMM));
229	const char *psz = g_aszYasmRegXMM[pParam->Index.idxXmmReg];
230	*pcchReg = 4 + !!psz[4];
231	return psz;
232	}
233	else if (pParam->fUse & DISUSE_REG_YMM)
234	{
235	Assert(pParam->Index.idxYmmReg < RT_ELEMENTS(g_aszYasmRegYMM));
236	const char *psz = g_aszYasmRegYMM[pParam->Index.idxYmmReg];
237	*pcchReg = 4 + !!psz[4];
238	return psz;
239
240	}
241	else
242	switch (pDis->uAddrMode)
243	{
244	case DISCPUMODE_16BIT:
245	{
246	Assert(pParam->Index.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen16));
247	const char *psz = g_aszYasmRegGen16[pParam->Index.idxGenReg];
248	*pcchReg = 2 + !!psz[2] + !!psz[3];
249	return psz;
250	}
251
252	case DISCPUMODE_32BIT:
253	{
254	Assert(pParam->Index.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen32));
255	const char *psz = g_aszYasmRegGen32[pParam->Index.idxGenReg];
256	*pcchReg = 2 + !!psz[2] + !!psz[3];
257	return psz;
258	}
259
260	case DISCPUMODE_64BIT:
261	{
262	Assert(pParam->Index.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen64));
263	const char *psz = g_aszYasmRegGen64[pParam->Index.idxGenReg];
264	*pcchReg = 2 + !!psz[2] + !!psz[3];
265	return psz;
266	}
267
268	default:
269	AssertMsgFailed(("%#x %#x\n", pParam->fUse, pDis->uAddrMode));
270	*pcchReg = 3;
271	return "r??";
272	}
273	}
274
275
276	/**
277	* Formats the current instruction in Yasm (/ Nasm) style.
278	*
279	*
280	* @returns The number of output characters. If this is >= cchBuf, then the content
281	* of pszBuf will be truncated.
282	* @param pDis Pointer to the disassembler state.
283	* @param pszBuf The output buffer.
284	* @param cchBuf The size of the output buffer.
285	* @param fFlags Format flags, see DIS_FORMAT_FLAGS_*.
286	* @param pfnGetSymbol Get symbol name for a jmp or call target address. Optional.
287	* @param pvUser User argument for pfnGetSymbol.
288	*/
289	DISDECL(size_t) DISFormatYasmEx(PCDISSTATE pDis, char *pszBuf, size_t cchBuf, uint32_t fFlags,
290	PFNDISGETSYMBOL pfnGetSymbol, void *pvUser)
291	{
292	/** @todo monitor and mwait aren't formatted correctly in 64-bit mode. */
293	/*
294	* Input validation and massaging.
295	*/
296	AssertPtr(pDis);
297	AssertPtrNull(pszBuf);
298	Assert(pszBuf \|\| !cchBuf);
299	AssertPtrNull(pfnGetSymbol);
300	AssertMsg(DIS_FMT_FLAGS_IS_VALID(fFlags), ("%#x\n", fFlags));
301	if (fFlags & DIS_FMT_FLAGS_ADDR_COMMENT)
302	fFlags = (fFlags & ~DIS_FMT_FLAGS_ADDR_LEFT) \| DIS_FMT_FLAGS_ADDR_RIGHT;
303	if (fFlags & DIS_FMT_FLAGS_BYTES_COMMENT)
304	fFlags = (fFlags & ~DIS_FMT_FLAGS_BYTES_LEFT) \| DIS_FMT_FLAGS_BYTES_RIGHT;
305
306	PCDISOPCODE const pOp = pDis->pCurInstr;
307
308	/*
309	* Output macros
310	*/
311	char *pszDst = pszBuf;
312	size_t cchDst = cchBuf;
313	size_t cchOutput = 0;
314	#define PUT_C(ch) \
315	do { \
316	cchOutput++; \
317	if (cchDst > 1) \
318	{ \
319	cchDst--; \
320	*pszDst++ = (ch); \
321	} \
322	} while (0)
323	#define PUT_STR(pszSrc, cchSrc) \
324	do { \
325	cchOutput += (cchSrc); \
326	if (cchDst > (cchSrc)) \
327	{ \
328	memcpy(pszDst, (pszSrc), (cchSrc)); \
329	pszDst += (cchSrc); \
330	cchDst -= (cchSrc); \
331	} \
332	else if (cchDst > 1) \
333	{ \
334	memcpy(pszDst, (pszSrc), cchDst - 1); \
335	pszDst += cchDst - 1; \
336	cchDst = 1; \
337	} \
338	} while (0)
339	#define PUT_SZ(sz) \
340	PUT_STR((sz), sizeof(sz) - 1)
341	#define PUT_SZ_STRICT(szStrict, szRelaxed) \
342	do { if (fFlags & DIS_FMT_FLAGS_STRICT) PUT_SZ(szStrict); else PUT_SZ(szRelaxed); } while (0)
343	#define PUT_PSZ(psz) \
344	do { const size_t cchTmp = strlen(psz); PUT_STR((psz), cchTmp); } while (0)
345	#define PUT_NUM(cch, fmt, num) \
346	do { \
347	cchOutput += (cch); \
348	if (cchDst > 1) \
349	{ \
350	const size_t cchTmp = RTStrPrintf(pszDst, cchDst, fmt, (num)); \
351	pszDst += cchTmp; \
352	cchDst -= cchTmp; \
353	Assert(cchTmp == (cch) \|\| cchDst == 1); \
354	} \
355	} while (0)
356	/** @todo add two flags for choosing between %X / %x and h / 0x. */
357	#define PUT_NUM_8(num) PUT_NUM(4, "0%02xh", (uint8_t)(num))
358	#define PUT_NUM_16(num) PUT_NUM(6, "0%04xh", (uint16_t)(num))
359	#define PUT_NUM_32(num) PUT_NUM(10, "0%08xh", (uint32_t)(num))
360	#define PUT_NUM_64(num) PUT_NUM(18, "0%016RX64h", (uint64_t)(num))
361
362	#define PUT_NUM_SIGN(cch, fmt, num, stype, utype) \
363	do { \
364	if ((stype)(num) >= 0) \
365	{ \
366	PUT_C('+'); \
367	PUT_NUM(cch, fmt, (utype)(num)); \
368	} \
369	else \
370	{ \
371	PUT_C('-'); \
372	PUT_NUM(cch, fmt, (utype)-(stype)(num)); \
373	} \
374	} while (0)
375	#define PUT_NUM_S8(num) PUT_NUM_SIGN(4, "0%02xh", num, int8_t, uint8_t)
376	#define PUT_NUM_S16(num) PUT_NUM_SIGN(6, "0%04xh", num, int16_t, uint16_t)
377	#define PUT_NUM_S32(num) PUT_NUM_SIGN(10, "0%08xh", num, int32_t, uint32_t)
378	#define PUT_NUM_S64(num) PUT_NUM_SIGN(18, "0%016RX64h", num, int64_t, uint64_t)
379
380	#define PUT_SYMBOL_TWO(a_rcSym, a_szStart, a_chEnd) \
381	do { \
382	if (RT_SUCCESS(a_rcSym)) \
383	{ \
384	PUT_SZ(a_szStart); \
385	PUT_PSZ(szSymbol); \
386	if (off != 0) \
387	{ \
388	if ((int8_t)off == off) \
389	PUT_NUM_S8(off); \
390	else if ((int16_t)off == off) \
391	PUT_NUM_S16(off); \
392	else if ((int32_t)off == off) \
393	PUT_NUM_S32(off); \
394	else \
395	PUT_NUM_S64(off); \
396	} \
397	PUT_C(a_chEnd); \
398	} \
399	} while (0)
400
401	#define PUT_SYMBOL(a_uSeg, a_uAddr, a_szStart, a_chEnd) \
402	do { \
403	if (pfnGetSymbol) \
404	{ \
405	int rcSym = pfnGetSymbol(pDis, a_uSeg, a_uAddr, szSymbol, sizeof(szSymbol), &off, pvUser); \
406	PUT_SYMBOL_TWO(rcSym, a_szStart, a_chEnd); \
407	} \
408	} while (0)
409
410
411	/*
412	* The address?
413	*/
414	if (fFlags & DIS_FMT_FLAGS_ADDR_LEFT)
415	{
416	#if HC_ARCH_BITS == 64 \|\| GC_ARCH_BITS == 64
417	if (pDis->uInstrAddr >= _4G)
418	PUT_NUM(9, "%08x`", (uint32_t)(pDis->uInstrAddr >> 32));
419	#endif
420	PUT_NUM(8, "%08x", (uint32_t)pDis->uInstrAddr);
421	PUT_C(' ');
422	}
423
424	/*
425	* The opcode bytes?
426	*/
427	if (fFlags & DIS_FMT_FLAGS_BYTES_LEFT)
428	{
429	size_t cchTmp = disFormatBytes(pDis, pszDst, cchDst, fFlags);
430	cchOutput += cchTmp;
431	if (cchDst > 1)
432	{
433	if (cchTmp <= cchDst)
434	{
435	cchDst -= cchTmp;
436	pszDst += cchTmp;
437	}
438	else
439	{
440	pszDst += cchDst - 1;
441	cchDst = 1;
442	}
443	}
444
445	/* Some padding to align the instruction. */
446	size_t cchPadding = (7 * (2 + !!(fFlags & DIS_FMT_FLAGS_BYTES_SPACED)))
447	+ !!(fFlags & DIS_FMT_FLAGS_BYTES_BRACKETS) * 2
448	+ 2;
449	cchPadding = cchTmp + 1 >= cchPadding ? 1 : cchPadding - cchTmp;
450	PUT_STR(g_szSpaces, cchPadding);
451	}
452
453
454	/*
455	* Filter out invalid opcodes first as they need special
456	* treatment. UD2 is an exception and should be handled normally.
457	*/
458	size_t const offInstruction = cchOutput;
459	if ( pOp->uOpcode == OP_INVALID
460	\|\| ( pOp->uOpcode == OP_ILLUD2
461	&& (pDis->fPrefix & DISPREFIX_LOCK)))
462	PUT_SZ("Illegal opcode");
463	else
464	{
465	/*
466	* Prefixes
467	*/
468	if (pDis->fPrefix & DISPREFIX_LOCK)
469	PUT_SZ("lock ");
470	if (pDis->fPrefix & DISPREFIX_REP)
471	PUT_SZ("rep ");
472	else if(pDis->fPrefix & DISPREFIX_REPNE)
473	PUT_SZ("repne ");
474
475	/*
476	* Adjust the format string to the correct mnemonic
477	* or to avoid things the assembler cannot handle correctly.
478	*/
479	char szTmpFmt[48];
480	const char *pszFmt = pOp->pszOpcode;
481	bool fIgnoresOpSize = false;
482	bool fMayNeedAddrSize = false;
483	switch (pOp->uOpcode)
484	{
485	case OP_JECXZ:
486	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "jcxz %Jb" : pDis->uOpMode == DISCPUMODE_32BIT ? "jecxz %Jb" : "jrcxz %Jb";
487	break;
488	case OP_PUSHF:
489	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "pushfw" : pDis->uOpMode == DISCPUMODE_32BIT ? "pushfd" : "pushfq";
490	break;
491	case OP_POPF:
492	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "popfw" : pDis->uOpMode == DISCPUMODE_32BIT ? "popfd" : "popfq";
493	break;
494	case OP_PUSHA:
495	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "pushaw" : "pushad";
496	break;
497	case OP_POPA:
498	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "popaw" : "popad";
499	break;
500	case OP_INSB:
501	pszFmt = "insb";
502	fIgnoresOpSize = fMayNeedAddrSize = true;
503	break;
504	case OP_INSWD:
505	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "insw" : pDis->uOpMode == DISCPUMODE_32BIT ? "insd" : "insq";
506	fMayNeedAddrSize = true;
507	break;
508	case OP_OUTSB:
509	pszFmt = "outsb";
510	fIgnoresOpSize = fMayNeedAddrSize = true;
511	break;
512	case OP_OUTSWD:
513	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "outsw" : pDis->uOpMode == DISCPUMODE_32BIT ? "outsd" : "outsq";
514	fMayNeedAddrSize = true;
515	break;
516	case OP_MOVSB:
517	pszFmt = "movsb";
518	fIgnoresOpSize = fMayNeedAddrSize = true;
519	break;
520	case OP_MOVSWD:
521	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "movsw" : pDis->uOpMode == DISCPUMODE_32BIT ? "movsd" : "movsq";
522	fMayNeedAddrSize = true;
523	break;
524	case OP_CMPSB:
525	pszFmt = "cmpsb";
526	fIgnoresOpSize = fMayNeedAddrSize = true;
527	break;
528	case OP_CMPWD:
529	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "cmpsw" : pDis->uOpMode == DISCPUMODE_32BIT ? "cmpsd" : "cmpsq";
530	fMayNeedAddrSize = true;
531	break;
532	case OP_SCASB:
533	pszFmt = "scasb";
534	fIgnoresOpSize = fMayNeedAddrSize = true;
535	break;
536	case OP_SCASWD:
537	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "scasw" : pDis->uOpMode == DISCPUMODE_32BIT ? "scasd" : "scasq";
538	fMayNeedAddrSize = true;
539	break;
540	case OP_LODSB:
541	pszFmt = "lodsb";
542	fIgnoresOpSize = fMayNeedAddrSize = true;
543	break;
544	case OP_LODSWD:
545	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "lodsw" : pDis->uOpMode == DISCPUMODE_32BIT ? "lodsd" : "lodsq";
546	fMayNeedAddrSize = true;
547	break;
548	case OP_STOSB:
549	pszFmt = "stosb";
550	fIgnoresOpSize = fMayNeedAddrSize = true;
551	break;
552	case OP_STOSWD:
553	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "stosw" : pDis->uOpMode == DISCPUMODE_32BIT ? "stosd" : "stosq";
554	fMayNeedAddrSize = true;
555	break;
556	case OP_CBW:
557	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "cbw" : pDis->uOpMode == DISCPUMODE_32BIT ? "cwde" : "cdqe";
558	break;
559	case OP_CWD:
560	pszFmt = pDis->uOpMode == DISCPUMODE_16BIT ? "cwd" : pDis->uOpMode == DISCPUMODE_32BIT ? "cdq" : "cqo";
561	break;
562	case OP_SHL:
563	Assert(pszFmt[3] == '/');
564	pszFmt += 4;
565	break;
566	case OP_XLAT:
567	pszFmt = "xlatb";
568	break;
569	case OP_INT3:
570	pszFmt = "int3";
571	break;
572
573	/*
574	* Don't know how to tell yasm to generate complicated nop stuff, so 'db' it.
575	*/
576	case OP_NOP:
577	if (pDis->bOpCode == 0x90)
578	/* fine, fine */;
579	else if (pszFmt[sizeof("nop %Ev") - 1] == '/' && pszFmt[sizeof("nop %Ev")] == 'p')
580	pszFmt = "prefetch %Eb";
581	else if (pDis->bOpCode == 0x1f)
582	{
583	Assert(pDis->cbInstr >= 3);
584	PUT_SZ("db 00fh, 01fh,");
585	PUT_NUM_8(MAKE_MODRM(pDis->ModRM.Bits.Mod, pDis->ModRM.Bits.Reg, pDis->ModRM.Bits.Rm));
586	for (unsigned i = 3; i < pDis->cbInstr; i++)
587	{
588	PUT_C(',');
589	PUT_NUM_8(0x90); ///@todo fixme.
590	}
591	pszFmt = "";
592	}
593	break;
594
595	default:
596	/* ST(X) -> stX (floating point) */
597	if (*pszFmt == 'f' && strchr(pszFmt, '('))
598	{
599	char *pszFmtDst = szTmpFmt;
600	char ch;
601	do
602	{
603	ch = *pszFmt++;
604	if (ch == 'S' && pszFmt[0] == 'T' && pszFmt[1] == '(')
605	{
606	*pszFmtDst++ = 's';
607	*pszFmtDst++ = 't';
608	pszFmt += 2;
609	ch = *pszFmt;
610	Assert(pszFmt[1] == ')');
611	pszFmt += 2;
612	*pszFmtDst++ = ch;
613	}
614	else
615	*pszFmtDst++ = ch;
616	} while (ch != '\0');
617	pszFmt = szTmpFmt;
618	}
619	if (strchr ("#@&", *pszFmt))
620	{
621	const char *pszDelim = strchr(pszFmt, '/');
622	const char *pszSpace = (pszDelim ? strchr(pszDelim, ' ') : NULL);
623	if (pszDelim != NULL)
624	{
625	char *pszFmtDst = szTmpFmt;
626	if (pszSpace == NULL) pszSpace = strchr(pszDelim, 0);
627	if ( (*pszFmt == '#' && pDis->bVexWFlag)
628	\|\| (*pszFmt == '@' && !VEXREG_IS256B(pDis->bVexDestReg))
629	\|\| (*pszFmt == '&' && ( DISUSE_IS_EFFECTIVE_ADDR(pDis->Param1.fUse)
630	\|\| DISUSE_IS_EFFECTIVE_ADDR(pDis->Param2.fUse)
631	\|\| DISUSE_IS_EFFECTIVE_ADDR(pDis->Param3.fUse)
632	\|\| DISUSE_IS_EFFECTIVE_ADDR(pDis->Param4.fUse))))
633	{
634	strncpy(pszFmtDst, pszFmt + 1, pszDelim - pszFmt - 1);
635	pszFmtDst += pszDelim - pszFmt - 1;
636	}
637	else
638	{
639	strncpy(pszFmtDst, pszDelim + 1, pszSpace - pszDelim - 1);
640	pszFmtDst += pszSpace - pszDelim - 1;
641	}
642	strcpy (pszFmtDst, pszSpace);
643	pszFmt = szTmpFmt;
644	}
645	}
646	break;
647
648	/*
649	* Horrible hacks.
650	*/
651	case OP_FLD:
652	if (pDis->bOpCode == 0xdb) /* m80fp workaround. */
653	(int )&pDis->Param1.fParam &= ~0x1f; /* make it pure OP_PARM_M */
654	break;
655	case OP_LAR: /* hack w -> v, probably not correct. */
656	(int )&pDis->Param2.fParam &= ~0x1f;
657	(int )&pDis->Param2.fParam \|= OP_PARM_v;
658	break;
659	}
660
661	/*
662	* Add operand size and address prefixes for outsb, movsb, etc.
663	*/
664	if (pDis->fPrefix & (DISPREFIX_OPSIZE \| DISPREFIX_ADDRSIZE))
665	{
666	if (fIgnoresOpSize && (pDis->fPrefix & DISPREFIX_OPSIZE) )
667	{
668	if (pDis->uCpuMode == DISCPUMODE_16BIT)
669	PUT_SZ("o32 ");
670	else
671	PUT_SZ("o16 ");
672	}
673	if (fMayNeedAddrSize && (pDis->fPrefix & DISPREFIX_ADDRSIZE) )
674	{
675	if (pDis->uCpuMode == DISCPUMODE_16BIT)
676	PUT_SZ("a32 ");
677	else
678	PUT_SZ("a16 ");
679	}
680	}
681
682	/*
683	* Formatting context and associated macros.
684	*/
685	PCDISOPPARAM pParam = &pDis->Param1;
686	int iParam = 1;
687
688	#define PUT_FAR() \
689	do { \
690	if ( OP_PARM_VSUBTYPE(pParam->fParam) == OP_PARM_p \
691	&& pOp->uOpcode != OP_LDS /* table bugs? */ \
692	&& pOp->uOpcode != OP_LES \
693	&& pOp->uOpcode != OP_LFS \
694	&& pOp->uOpcode != OP_LGS \
695	&& pOp->uOpcode != OP_LSS ) \
696	PUT_SZ("far "); \
697	} while (0)
698	/** @todo mov ah,ch ends up with a byte 'override'... - check if this wasn't fixed. */
699	/** @todo drop the work/dword/qword override when the src/dst is a register (except for movsx/movzx). */
700	#define PUT_SIZE_OVERRIDE() \
701	do { \
702	switch (OP_PARM_VSUBTYPE(pParam->fParam)) \
703	{ \
704	case OP_PARM_v: \
705	case OP_PARM_y: \
706	switch (pDis->uOpMode) \
707	{ \
708	case DISCPUMODE_16BIT: if (OP_PARM_VSUBTYPE(pParam->fParam) != OP_PARM_y) PUT_SZ("word "); break; \
709	case DISCPUMODE_32BIT: \
710	if (pDis->pCurInstr->uOpcode != OP_GATHER \|\| pDis->bVexWFlag) { PUT_SZ("dword "); break; } \
711	case DISCPUMODE_64BIT: PUT_SZ("qword "); break; \
712	default: break; \
713	} \
714	break; \
715	case OP_PARM_b: PUT_SZ("byte "); break; \
716	case OP_PARM_w: \
717	if (OP_PARM_VTYPE(pParam->fParam) == OP_PARM_W \|\| \
718	OP_PARM_VTYPE(pParam->fParam) == OP_PARM_M) \
719	{ \
720	if (VEXREG_IS256B(pDis->bVexDestReg)) PUT_SZ("dword "); \
721	else PUT_SZ("word "); \
722	} \
723	break; \
724	case OP_PARM_d: \
725	if (OP_PARM_VTYPE(pParam->fParam) == OP_PARM_W \|\| \
726	OP_PARM_VTYPE(pParam->fParam) == OP_PARM_M) \
727	{ \
728	if (VEXREG_IS256B(pDis->bVexDestReg)) PUT_SZ("qword "); \
729	else PUT_SZ("dword "); \
730	} \
731	break; \
732	case OP_PARM_q: \
733	if (OP_PARM_VTYPE(pParam->fParam) == OP_PARM_W \|\| \
734	OP_PARM_VTYPE(pParam->fParam) == OP_PARM_M) \
735	{ \
736	if (VEXREG_IS256B(pDis->bVexDestReg)) PUT_SZ("oword "); \
737	else PUT_SZ("qword "); \
738	} \
739	break; \
740	case OP_PARM_ps: \
741	case OP_PARM_pd: \
742	case OP_PARM_x: if (VEXREG_IS256B(pDis->bVexDestReg)) { PUT_SZ("yword "); break; } \
743	case OP_PARM_ss: \
744	case OP_PARM_sd: \
745	case OP_PARM_dq: PUT_SZ("oword "); break; \
746	case OP_PARM_qq: PUT_SZ("yword "); break; \
747	case OP_PARM_p: break; /* see PUT_FAR */ \
748	case OP_PARM_s: if (pParam->fUse & DISUSE_REG_FP) PUT_SZ("tword "); break; /* ?? */ \
749	case OP_PARM_z: break; \
750	case OP_PARM_NONE: \
751	if ( OP_PARM_VTYPE(pParam->fParam) == OP_PARM_M \
752	&& ((pParam->fUse & DISUSE_REG_FP) \|\| pOp->uOpcode == OP_FLD)) \
753	PUT_SZ("tword "); \
754	break; \
755	default: break; /no pointer type specified/necessary/ \
756	} \
757	} while (0)
758	static const char s_szSegPrefix[6][4] = { "es:", "cs:", "ss:", "ds:", "fs:", "gs:" };
759	#define PUT_SEGMENT_OVERRIDE() \
760	do { \
761	if (pDis->fPrefix & DISPREFIX_SEG) \
762	PUT_STR(s_szSegPrefix[pDis->idxSegPrefix], 3); \
763	} while (0)
764
765
766	/*
767	* Segment prefixing for instructions that doesn't do memory access.
768	*/
769	if ( (pDis->fPrefix & DISPREFIX_SEG)
770	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param1.fUse)
771	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param2.fUse)
772	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param3.fUse))
773	{
774	PUT_STR(s_szSegPrefix[pDis->idxSegPrefix], 2);
775	PUT_C(' ');
776	}
777
778
779	/*
780	* The formatting loop.
781	*/
782	RTINTPTR off;
783	char szSymbol[128];
784	char ch;
785	while ((ch = *pszFmt++) != '\0')
786	{
787	if (ch == '%')
788	{
789	ch = *pszFmt++;
790	switch (ch)
791	{
792	/*
793	* ModRM - Register only.
794	*/
795	case 'C': /* Control register (ParseModRM / UseModRM). */
796	case 'D': /* Debug register (ParseModRM / UseModRM). */
797	case 'G': /* ModRM selects general register (ParseModRM / UseModRM). */
798	case 'S': /* ModRM byte selects a segment register (ParseModRM / UseModRM). */
799	case 'T': /* ModRM byte selects a test register (ParseModRM / UseModRM). */
800	case 'V': /* ModRM byte selects an XMM/SSE register (ParseModRM / UseModRM). */
801	case 'P': /* ModRM byte selects MMX register (ParseModRM / UseModRM). */
802	case 'H': /* The VEX.vvvv field of the VEX prefix selects a XMM/YMM register. */
803	case 'L': /* The upper 4 bits of the 8-bit immediate selects a XMM/YMM register. */
804	{
805	pszFmt += RT_C_IS_ALPHA(pszFmt[0]) ? RT_C_IS_ALPHA(pszFmt[1]) ? 2 : 1 : 0;
806	Assert(!(pParam->fUse & (DISUSE_INDEX \| DISUSE_SCALE) /* No SIB here... */));
807	Assert(!(pParam->fUse & (DISUSE_DISPLACEMENT8 \| DISUSE_DISPLACEMENT16 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT64 \| DISUSE_RIPDISPLACEMENT32)));
808
809	size_t cchReg;
810	const char *pszReg = disasmFormatYasmBaseReg(pDis, pParam, &cchReg);
811	PUT_STR(pszReg, cchReg);
812	break;
813	}
814
815	/*
816	* ModRM - Register or memory.
817	*/
818	case 'E': /* ModRM specifies parameter (ParseModRM / UseModRM / UseSIB). */
819	case 'Q': /* ModRM byte selects MMX register or memory address (ParseModRM / UseModRM). */
820	case 'R': /* ModRM byte may only refer to a general register (ParseModRM / UseModRM). */
821	case 'W': /* ModRM byte selects an XMM/SSE register or a memory address (ParseModRM / UseModRM). */
822	case 'M': /* ModRM may only refer to memory (ParseModRM / UseModRM). */
823	{
824	pszFmt += RT_C_IS_ALPHA(pszFmt[0]) ? RT_C_IS_ALPHA(pszFmt[1]) ? 2 : 1 : 0;
825
826	PUT_FAR();
827	uint32_t const fUse = pParam->fUse;
828	if (DISUSE_IS_EFFECTIVE_ADDR(fUse))
829	{
830	/* Work around mov seg,[mem16] and mov [mem16],seg as these always make a 16-bit mem
831	while the register variants deals with 16, 32 & 64 in the normal fashion. */
832	if ( pParam->fParam != OP_PARM_Ev
833	\|\| pOp->uOpcode != OP_MOV
834	\|\| ( pOp->fParam1 != OP_PARM_Sw
835	&& pOp->fParam2 != OP_PARM_Sw))
836	PUT_SIZE_OVERRIDE();
837	PUT_C('[');
838	}
839	if ( (fFlags & DIS_FMT_FLAGS_STRICT)
840	&& (fUse & (DISUSE_DISPLACEMENT8 \| DISUSE_DISPLACEMENT16 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT64 \| DISUSE_RIPDISPLACEMENT32)))
841	{
842	if ( (fUse & DISUSE_DISPLACEMENT8)
843	&& !pParam->uDisp.i8)
844	PUT_SZ("byte ");
845	else if ( (fUse & DISUSE_DISPLACEMENT16)
846	&& (int8_t)pParam->uDisp.i16 == (int16_t)pParam->uDisp.i16)
847	PUT_SZ("word ");
848	else if ( (fUse & DISUSE_DISPLACEMENT32)
849	&& (int16_t)pParam->uDisp.i32 == (int32_t)pParam->uDisp.i32) //??
850	PUT_SZ("dword ");
851	else if ( (fUse & DISUSE_DISPLACEMENT64)
852	&& (pDis->SIB.Bits.Base != 5 \|\| pDis->ModRM.Bits.Mod != 0)
853	&& (int32_t)pParam->uDisp.i64 == (int64_t)pParam->uDisp.i64) //??
854	PUT_SZ("qword ");
855	}
856	if (DISUSE_IS_EFFECTIVE_ADDR(fUse))
857	PUT_SEGMENT_OVERRIDE();
858
859	bool fBase = (fUse & DISUSE_BASE) /* When exactly is DISUSE_BASE supposed to be set? disasmModRMReg doesn't set it. */
860	\|\| ( (fUse & ( DISUSE_REG_GEN8
861	\| DISUSE_REG_GEN16
862	\| DISUSE_REG_GEN32
863	\| DISUSE_REG_GEN64
864	\| DISUSE_REG_FP
865	\| DISUSE_REG_MMX
866	\| DISUSE_REG_XMM
867	\| DISUSE_REG_YMM
868	\| DISUSE_REG_CR
869	\| DISUSE_REG_DBG
870	\| DISUSE_REG_SEG
871	\| DISUSE_REG_TEST ))
872	&& !DISUSE_IS_EFFECTIVE_ADDR(fUse));
873	if (fBase)
874	{
875	size_t cchReg;
876	const char *pszReg = disasmFormatYasmBaseReg(pDis, pParam, &cchReg);
877	PUT_STR(pszReg, cchReg);
878	}
879
880	if (fUse & DISUSE_INDEX)
881	{
882	if (fBase)
883	PUT_C('+');
884
885	size_t cchReg;
886	const char *pszReg = disasmFormatYasmIndexReg(pDis, pParam, &cchReg);
887	PUT_STR(pszReg, cchReg);
888
889	if (fUse & DISUSE_SCALE)
890	{
891	PUT_C('*');
892	PUT_C('0' + pParam->uScale);
893	}
894	}
895	else
896	Assert(!(fUse & DISUSE_SCALE));
897
898	int64_t off2 = 0;
899	if (fUse & (DISUSE_DISPLACEMENT8 \| DISUSE_DISPLACEMENT16 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT64 \| DISUSE_RIPDISPLACEMENT32))
900	{
901	if (fUse & DISUSE_DISPLACEMENT8)
902	off2 = pParam->uDisp.i8;
903	else if (fUse & DISUSE_DISPLACEMENT16)
904	off2 = pParam->uDisp.i16;
905	else if (fUse & (DISUSE_DISPLACEMENT32 \| DISUSE_RIPDISPLACEMENT32))
906	off2 = pParam->uDisp.i32;
907	else if (fUse & DISUSE_DISPLACEMENT64)
908	off2 = pParam->uDisp.i64;
909	else
910	{
911	AssertFailed();
912	off2 = 0;
913	}
914
915	int64_t off3 = off2;
916	if (fBase \|\| (fUse & (DISUSE_INDEX \| DISUSE_RIPDISPLACEMENT32)))
917	{
918	PUT_C(off3 >= 0 ? '+' : '-');
919	if (off3 < 0)
920	off3 = -off3;
921	}
922	if (fUse & DISUSE_DISPLACEMENT8)
923	PUT_NUM_8( off3);
924	else if (fUse & DISUSE_DISPLACEMENT16)
925	PUT_NUM_16(off3);
926	else if (fUse & DISUSE_DISPLACEMENT32)
927	PUT_NUM_32(off3);
928	else if (fUse & DISUSE_DISPLACEMENT64)
929	PUT_NUM_64(off3);
930	else
931	{
932	PUT_NUM_32(off3);
933	PUT_SZ(" wrt rip (");
934	off2 += pDis->uInstrAddr + pDis->cbInstr;
935	PUT_NUM_64(off2);
936	if (pfnGetSymbol)
937	PUT_SYMBOL((pDis->fPrefix & DISPREFIX_SEG)
938	? DIS_FMT_SEL_FROM_REG(pDis->idxSegPrefix)
939	: DIS_FMT_SEL_FROM_REG(DISSELREG_DS),
940	pDis->uAddrMode == DISCPUMODE_64BIT
941	? (uint64_t)off2
942	: pDis->uAddrMode == DISCPUMODE_32BIT
943	? (uint32_t)off2
944	: (uint16_t)off2,
945	" = ",
946	')');
947	else
948	PUT_C(')');
949	}
950	}
951
952	if (DISUSE_IS_EFFECTIVE_ADDR(fUse))
953	{
954	if (pfnGetSymbol && !fBase && !(fUse & (DISUSE_INDEX \| DISUSE_RIPDISPLACEMENT32)) && off2 != 0)
955	PUT_SYMBOL((pDis->fPrefix & DISPREFIX_SEG)
956	? DIS_FMT_SEL_FROM_REG(pDis->idxSegPrefix)
957	: DIS_FMT_SEL_FROM_REG(DISSELREG_DS),
958	pDis->uAddrMode == DISCPUMODE_64BIT
959	? (uint64_t)off2
960	: pDis->uAddrMode == DISCPUMODE_32BIT
961	? (uint32_t)off2
962	: (uint16_t)off2,
963	" (=",
964	')');
965	PUT_C(']');
966	}
967	break;
968	}
969
970	case 'F': /* Eflags register (0 - popf/pushf only, avoided in adjustments above). */
971	AssertFailed();
972	break;
973
974	case 'I': /* Immediate data (ParseImmByte, ParseImmByteSX, ParseImmV, ParseImmUshort, ParseImmZ). */
975	Assert(pszFmt == 'b' \|\| pszFmt == 'v' \|\| pszFmt == 'w' \|\| pszFmt == 'z'); pszFmt++;
976	switch (pParam->fUse & ( DISUSE_IMMEDIATE8 \| DISUSE_IMMEDIATE16 \| DISUSE_IMMEDIATE32 \| DISUSE_IMMEDIATE64
977	\| DISUSE_IMMEDIATE16_SX8 \| DISUSE_IMMEDIATE32_SX8 \| DISUSE_IMMEDIATE64_SX8))
978	{
979	case DISUSE_IMMEDIATE8:
980	if ( (fFlags & DIS_FMT_FLAGS_STRICT)
981	&& ( (pOp->fParam1 >= OP_PARM_REG_GEN8_START && pOp->fParam1 <= OP_PARM_REG_GEN8_END)
982	\|\| (pOp->fParam2 >= OP_PARM_REG_GEN8_START && pOp->fParam2 <= OP_PARM_REG_GEN8_END))
983	)
984	PUT_SZ("strict byte ");
985	PUT_NUM_8(pParam->uValue);
986	break;
987
988	case DISUSE_IMMEDIATE16:
989	if ( pDis->uCpuMode != pDis->uOpMode
990	\|\| ( (fFlags & DIS_FMT_FLAGS_STRICT)
991	&& ( (int8_t)pParam->uValue == (int16_t)pParam->uValue
992	\|\| (pOp->fParam1 >= OP_PARM_REG_GEN16_START && pOp->fParam1 <= OP_PARM_REG_GEN16_END)
993	\|\| (pOp->fParam2 >= OP_PARM_REG_GEN16_START && pOp->fParam2 <= OP_PARM_REG_GEN16_END))
994	)
995	)
996	{
997	if (OP_PARM_VSUBTYPE(pParam->fParam) == OP_PARM_b)
998	PUT_SZ_STRICT("strict byte ", "byte ");
999	else if ( OP_PARM_VSUBTYPE(pParam->fParam) == OP_PARM_v
1000	\|\| OP_PARM_VSUBTYPE(pParam->fParam) == OP_PARM_z)
1001	PUT_SZ_STRICT("strict word ", "word ");
1002	}
1003	PUT_NUM_16(pParam->uValue);
1004	break;
1005
1006	case DISUSE_IMMEDIATE16_SX8:
1007	if ( !(pDis->fPrefix & DISPREFIX_OPSIZE)
1008	\|\| pDis->pCurInstr->uOpcode != OP_PUSH)
1009	PUT_SZ_STRICT("strict byte ", "byte ");
1010	else
1011	PUT_SZ("word ");
1012	PUT_NUM_16(pParam->uValue);
1013	break;
1014
1015	case DISUSE_IMMEDIATE32:
1016	if ( pDis->uOpMode != (pDis->uCpuMode == DISCPUMODE_16BIT ? DISCPUMODE_16BIT : DISCPUMODE_32BIT) /* not perfect */
1017	\|\| ( (fFlags & DIS_FMT_FLAGS_STRICT)
1018	&& ( (int8_t)pParam->uValue == (int32_t)pParam->uValue
1019	\|\| (pOp->fParam1 >= OP_PARM_REG_GEN32_START && pOp->fParam1 <= OP_PARM_REG_GEN32_END)
1020	\|\| (pOp->fParam2 >= OP_PARM_REG_GEN32_START && pOp->fParam2 <= OP_PARM_REG_GEN32_END))
1021	)
1022	)
1023	{
1024	if (OP_PARM_VSUBTYPE(pParam->fParam) == OP_PARM_b)
1025	PUT_SZ_STRICT("strict byte ", "byte ");
1026	else if ( OP_PARM_VSUBTYPE(pParam->fParam) == OP_PARM_v
1027	\|\| OP_PARM_VSUBTYPE(pParam->fParam) == OP_PARM_z)
1028	PUT_SZ_STRICT("strict dword ", "dword ");
1029	}
1030	PUT_NUM_32(pParam->uValue);
1031	if (pDis->uCpuMode == DISCPUMODE_32BIT)
1032	PUT_SYMBOL(DIS_FMT_SEL_FROM_REG(DISSELREG_CS), pParam->uValue, " (=", ')');
1033	break;
1034
1035	case DISUSE_IMMEDIATE32_SX8:
1036	if ( !(pDis->fPrefix & DISPREFIX_OPSIZE)
1037	\|\| pDis->pCurInstr->uOpcode != OP_PUSH)
1038	PUT_SZ_STRICT("strict byte ", "byte ");
1039	else
1040	PUT_SZ("dword ");
1041	PUT_NUM_32(pParam->uValue);
1042	break;
1043
1044	case DISUSE_IMMEDIATE64_SX8:
1045	if ( !(pDis->fPrefix & DISPREFIX_OPSIZE)
1046	\|\| pDis->pCurInstr->uOpcode != OP_PUSH)
1047	PUT_SZ_STRICT("strict byte ", "byte ");
1048	else
1049	PUT_SZ("qword ");
1050	PUT_NUM_64(pParam->uValue);
1051	break;
1052
1053	case DISUSE_IMMEDIATE64:
1054	PUT_NUM_64(pParam->uValue);
1055	break;
1056
1057	default:
1058	AssertFailed();
1059	break;
1060	}
1061	break;
1062
1063	case 'J': /* Relative jump offset (ParseImmBRel + ParseImmVRel). */
1064	{
1065	int32_t offDisplacement;
1066	Assert(iParam == 1);
1067	bool fPrefix = (fFlags & DIS_FMT_FLAGS_STRICT)
1068	&& pOp->uOpcode != OP_CALL
1069	&& pOp->uOpcode != OP_LOOP
1070	&& pOp->uOpcode != OP_LOOPE
1071	&& pOp->uOpcode != OP_LOOPNE
1072	&& pOp->uOpcode != OP_JECXZ;
1073	if (pOp->uOpcode == OP_CALL)
1074	fFlags &= ~DIS_FMT_FLAGS_RELATIVE_BRANCH;
1075
1076	if (pParam->fUse & DISUSE_IMMEDIATE8_REL)
1077	{
1078	if (fPrefix)
1079	PUT_SZ("short ");
1080	offDisplacement = (int8_t)pParam->uValue;
1081	Assert(*pszFmt == 'b'); pszFmt++;
1082
1083	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1084	PUT_NUM_S8(offDisplacement);
1085	}
1086	else if (pParam->fUse & DISUSE_IMMEDIATE16_REL)
1087	{
1088	if (fPrefix)
1089	PUT_SZ("near ");
1090	offDisplacement = (int16_t)pParam->uValue;
1091	Assert(*pszFmt == 'v'); pszFmt++;
1092
1093	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1094	PUT_NUM_S16(offDisplacement);
1095	}
1096	else
1097	{
1098	if (fPrefix)
1099	PUT_SZ("near ");
1100	offDisplacement = (int32_t)pParam->uValue;
1101	Assert(pParam->fUse & (DISUSE_IMMEDIATE32_REL \| DISUSE_IMMEDIATE64_REL));
1102	Assert(*pszFmt == 'v'); pszFmt++;
1103
1104	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1105	PUT_NUM_S32(offDisplacement);
1106	}
1107	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1108	PUT_SZ(" (");
1109
1110	RTUINTPTR uTrgAddr = pDis->uInstrAddr + pDis->cbInstr + offDisplacement;
1111	if (pDis->uCpuMode == DISCPUMODE_16BIT)
1112	PUT_NUM_16(uTrgAddr);
1113	else if (pDis->uCpuMode == DISCPUMODE_32BIT)
1114	PUT_NUM_32(uTrgAddr);
1115	else
1116	PUT_NUM_64(uTrgAddr);
1117
1118	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1119	{
1120	PUT_SYMBOL(DIS_FMT_SEL_FROM_REG(DISSELREG_CS), uTrgAddr, " = ", ' ');
1121	PUT_C(')');
1122	}
1123	else
1124	PUT_SYMBOL(DIS_FMT_SEL_FROM_REG(DISSELREG_CS), uTrgAddr, " (", ')');
1125	break;
1126	}
1127
1128	case 'A': /* Direct (jump/call) address (ParseImmAddr). */
1129	{
1130	Assert(*pszFmt == 'p'); pszFmt++;
1131	PUT_FAR();
1132	PUT_SIZE_OVERRIDE();
1133	PUT_SEGMENT_OVERRIDE();
1134	off = 0;
1135	int rc = VERR_SYMBOL_NOT_FOUND;
1136	switch (pParam->fUse & (DISUSE_IMMEDIATE_ADDR_16_16 \| DISUSE_IMMEDIATE_ADDR_16_32 \| DISUSE_DISPLACEMENT64 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT16))
1137	{
1138	case DISUSE_IMMEDIATE_ADDR_16_16:
1139	PUT_NUM_16(pParam->uValue >> 16);
1140	PUT_C(':');
1141	PUT_NUM_16(pParam->uValue);
1142	if (pfnGetSymbol)
1143	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_VALUE(pParam->uValue >> 16), (uint16_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1144	break;
1145	case DISUSE_IMMEDIATE_ADDR_16_32:
1146	PUT_NUM_16(pParam->uValue >> 32);
1147	PUT_C(':');
1148	PUT_NUM_32(pParam->uValue);
1149	if (pfnGetSymbol)
1150	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_VALUE(pParam->uValue >> 16), (uint32_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1151	break;
1152	case DISUSE_DISPLACEMENT16:
1153	PUT_NUM_16(pParam->uValue);
1154	if (pfnGetSymbol)
1155	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), (uint16_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1156	break;
1157	case DISUSE_DISPLACEMENT32:
1158	PUT_NUM_32(pParam->uValue);
1159	if (pfnGetSymbol)
1160	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), (uint32_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1161	break;
1162	case DISUSE_DISPLACEMENT64:
1163	PUT_NUM_64(pParam->uValue);
1164	if (pfnGetSymbol)
1165	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), (uint64_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1166	break;
1167	default:
1168	AssertFailed();
1169	break;
1170	}
1171
1172	PUT_SYMBOL_TWO(rc, " [", ']');
1173	break;
1174	}
1175
1176	case 'O': /* No ModRM byte (ParseImmAddr). */
1177	{
1178	Assert(pszFmt == 'b' \|\| pszFmt == 'v'); pszFmt++;
1179	PUT_FAR();
1180	PUT_SIZE_OVERRIDE();
1181	PUT_C('[');
1182	PUT_SEGMENT_OVERRIDE();
1183	off = 0;
1184	int rc = VERR_SYMBOL_NOT_FOUND;
1185	switch (pParam->fUse & (DISUSE_IMMEDIATE_ADDR_16_16 \| DISUSE_IMMEDIATE_ADDR_16_32 \| DISUSE_DISPLACEMENT64 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT16))
1186	{
1187	case DISUSE_IMMEDIATE_ADDR_16_16:
1188	PUT_NUM_16(pParam->uValue >> 16);
1189	PUT_C(':');
1190	PUT_NUM_16(pParam->uValue);
1191	if (pfnGetSymbol)
1192	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_VALUE(pParam->uValue >> 16), (uint16_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1193	break;
1194	case DISUSE_IMMEDIATE_ADDR_16_32:
1195	PUT_NUM_16(pParam->uValue >> 32);
1196	PUT_C(':');
1197	PUT_NUM_32(pParam->uValue);
1198	if (pfnGetSymbol)
1199	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_VALUE(pParam->uValue >> 16), (uint32_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1200	break;
1201	case DISUSE_DISPLACEMENT16:
1202	PUT_NUM_16(pParam->uDisp.i16);
1203	if (pfnGetSymbol)
1204	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), pParam->uDisp.u16, szSymbol, sizeof(szSymbol), &off, pvUser);
1205	break;
1206	case DISUSE_DISPLACEMENT32:
1207	PUT_NUM_32(pParam->uDisp.i32);
1208	if (pfnGetSymbol)
1209	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), pParam->uDisp.u32, szSymbol, sizeof(szSymbol), &off, pvUser);
1210	break;
1211	case DISUSE_DISPLACEMENT64:
1212	PUT_NUM_64(pParam->uDisp.i64);
1213	if (pfnGetSymbol)
1214	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), pParam->uDisp.u64, szSymbol, sizeof(szSymbol), &off, pvUser);
1215	break;
1216	default:
1217	AssertFailed();
1218	break;
1219	}
1220	PUT_C(']');
1221
1222	PUT_SYMBOL_TWO(rc, " (", ')');
1223	break;
1224	}
1225
1226	case 'X': /* DS:SI (ParseXb, ParseXv). */
1227	case 'Y': /* ES:DI (ParseYb, ParseYv). */
1228	{
1229	Assert(pszFmt == 'b' \|\| pszFmt == 'v'); pszFmt++;
1230	PUT_FAR();
1231	PUT_SIZE_OVERRIDE();
1232	PUT_C('[');
1233	if (pParam->fUse & DISUSE_POINTER_DS_BASED)
1234	PUT_SZ("ds:");
1235	else
1236	PUT_SZ("es:");
1237
1238	size_t cchReg;
1239	const char *pszReg = disasmFormatYasmBaseReg(pDis, pParam, &cchReg);
1240	PUT_STR(pszReg, cchReg);
1241	PUT_C(']');
1242	break;
1243	}
1244
1245	case 'e': /* Register based on operand size (e.g. %eAX, %eAH) (ParseFixedReg). */
1246	{
1247	Assert(RT_C_IS_ALPHA(pszFmt[0]) && RT_C_IS_ALPHA(pszFmt[1]) && !RT_C_IS_ALPHA(pszFmt[2]));
1248	pszFmt += 2;
1249	size_t cchReg;
1250	const char *pszReg = disasmFormatYasmBaseReg(pDis, pParam, &cchReg);
1251	PUT_STR(pszReg, cchReg);
1252	break;
1253	}
1254
1255	default:
1256	AssertMsgFailed(("%c%s!\n", ch, pszFmt));
1257	break;
1258	}
1259	AssertMsg(pszFmt == ',' \|\| pszFmt == '\0', ("%c%s\n", ch, pszFmt));
1260	}
1261	else
1262	{
1263	PUT_C(ch);
1264	if (ch == ',')
1265	{
1266	Assert(*pszFmt != ' ');
1267	PUT_C(' ');
1268	switch (++iParam)
1269	{
1270	case 2: pParam = &pDis->Param2; break;
1271	case 3: pParam = &pDis->Param3; break;
1272	case 4: pParam = &pDis->Param4; break;
1273	default: pParam = NULL; break;
1274	}
1275	}
1276	}
1277	} /* while more to format */
1278	}
1279
1280	/*
1281	* Any additional output to the right of the instruction?
1282	*/
1283	if (fFlags & (DIS_FMT_FLAGS_BYTES_RIGHT \| DIS_FMT_FLAGS_ADDR_RIGHT))
1284	{
1285	/* some up front padding. */
1286	size_t cchPadding = cchOutput - offInstruction;
1287	cchPadding = cchPadding + 1 >= 42 ? 1 : 42 - cchPadding;
1288	PUT_STR(g_szSpaces, cchPadding);
1289
1290	/* comment? */
1291	if (fFlags & (DIS_FMT_FLAGS_BYTES_RIGHT \| DIS_FMT_FLAGS_ADDR_RIGHT))
1292	PUT_SZ(";");
1293
1294	/*
1295	* The address?
1296	*/
1297	if (fFlags & DIS_FMT_FLAGS_ADDR_RIGHT)
1298	{
1299	PUT_C(' ');
1300	#if HC_ARCH_BITS == 64 \|\| GC_ARCH_BITS == 64
1301	if (pDis->uInstrAddr >= _4G)
1302	PUT_NUM(9, "%08x`", (uint32_t)(pDis->uInstrAddr >> 32));
1303	#endif
1304	PUT_NUM(8, "%08x", (uint32_t)pDis->uInstrAddr);
1305	}
1306
1307	/*
1308	* Opcode bytes?
1309	*/
1310	if (fFlags & DIS_FMT_FLAGS_BYTES_RIGHT)
1311	{
1312	PUT_C(' ');
1313	size_t cchTmp = disFormatBytes(pDis, pszDst, cchDst, fFlags);
1314	cchOutput += cchTmp;
1315	if (cchTmp >= cchDst)
1316	cchTmp = cchDst - (cchDst != 0);
1317	cchDst -= cchTmp;
1318	pszDst += cchTmp;
1319	}
1320	}
1321
1322	/*
1323	* Terminate it - on overflow we'll have reserved one byte for this.
1324	*/
1325	if (cchDst > 0)
1326	*pszDst = '\0';
1327	else
1328	Assert(!cchBuf);
1329
1330	/* clean up macros */
1331	#undef PUT_PSZ
1332	#undef PUT_SZ
1333	#undef PUT_STR
1334	#undef PUT_C
1335	return cchOutput;
1336	}
1337
1338
1339	/**
1340	* Formats the current instruction in Yasm (/ Nasm) style.
1341	*
1342	* This is a simplified version of DISFormatYasmEx() provided for your convenience.
1343	*
1344	*
1345	* @returns The number of output characters. If this is >= cchBuf, then the content
1346	* of pszBuf will be truncated.
1347	* @param pDis Pointer to the disassembler state.
1348	* @param pszBuf The output buffer.
1349	* @param cchBuf The size of the output buffer.
1350	*/
1351	DISDECL(size_t) DISFormatYasm(PCDISSTATE pDis, char *pszBuf, size_t cchBuf)
1352	{
1353	return DISFormatYasmEx(pDis, pszBuf, cchBuf, 0 /* fFlags /, NULL / pfnGetSymbol /, NULL / pvUser */);
1354	}
1355
1356
1357	/**
1358	* Checks if the encoding of the given disassembled instruction is something we
1359	* can never get YASM to produce.
1360	*
1361	* @returns true if it's odd, false if it isn't.
1362	* @param pDis The disassembler output. The byte fetcher callback will
1363	* be used if present as we might need to fetch opcode
1364	* bytes.
1365	*/
1366	DISDECL(bool) DISFormatYasmIsOddEncoding(PDISSTATE pDis)
1367	{
1368	/*
1369	* Mod rm + SIB: Check for duplicate EBP encodings that yasm won't use for very good reasons.
1370	*/
1371	if ( pDis->uAddrMode != DISCPUMODE_16BIT ///@todo correct?
1372	&& pDis->ModRM.Bits.Rm == 4
1373	&& pDis->ModRM.Bits.Mod != 3)
1374	{
1375	/* No scaled index SIB (index=4), except for ESP. */
1376	if ( pDis->SIB.Bits.Index == 4
1377	&& pDis->SIB.Bits.Base != 4)
1378	return true;
1379
1380	/* EBP + displacement */
1381	if ( pDis->ModRM.Bits.Mod != 0
1382	&& pDis->SIB.Bits.Base == 5
1383	&& pDis->SIB.Bits.Scale == 0)
1384	return true;
1385	}
1386
1387	/*
1388	* Seems to be an instruction alias here, but I cannot find any docs on it... hrmpf!
1389	*/
1390	if ( pDis->pCurInstr->uOpcode == OP_SHL
1391	&& pDis->ModRM.Bits.Reg == 6)
1392	return true;
1393
1394	/*
1395	* Check for multiple prefixes of the same kind.
1396	*/
1397	uint8_t off1stSeg = UINT8_MAX;
1398	uint8_t offOpSize = UINT8_MAX;
1399	uint8_t offAddrSize = UINT8_MAX;
1400	uint32_t fPrefixes = 0;
1401	for (uint32_t offOpcode = 0; offOpcode < RT_ELEMENTS(pDis->abInstr); offOpcode++)
1402	{
1403	uint32_t f;
1404	switch (pDis->abInstr[offOpcode])
1405	{
1406	case 0xf0:
1407	f = DISPREFIX_LOCK;
1408	break;
1409
1410	case 0xf2:
1411	case 0xf3:
1412	f = DISPREFIX_REP; /* yes, both */
1413	break;
1414
1415	case 0x2e:
1416	case 0x3e:
1417	case 0x26:
1418	case 0x36:
1419	case 0x64:
1420	case 0x65:
1421	if (off1stSeg == UINT8_MAX)
1422	off1stSeg = offOpcode;
1423	f = DISPREFIX_SEG;
1424	break;
1425
1426	case 0x66:
1427	if (offOpSize == UINT8_MAX)
1428	offOpSize = offOpcode;
1429	f = DISPREFIX_OPSIZE;
1430	break;
1431
1432	case 0x67:
1433	if (offAddrSize == UINT8_MAX)
1434	offAddrSize = offOpcode;
1435	f = DISPREFIX_ADDRSIZE;
1436	break;
1437
1438	case 0x40: case 0x41: case 0x42: case 0x43: case 0x44: case 0x45: case 0x46: case 0x47:
1439	case 0x48: case 0x49: case 0x4a: case 0x4b: case 0x4c: case 0x4d: case 0x4e: case 0x4f:
1440	f = pDis->uCpuMode == DISCPUMODE_64BIT ? DISPREFIX_REX : 0;
1441	break;
1442
1443	default:
1444	f = 0;
1445	break;
1446	}
1447	if (!f)
1448	break; /* done */
1449	if (fPrefixes & f)
1450	return true;
1451	fPrefixes \|= f;
1452	}
1453
1454	/* segment overrides are fun */
1455	if (fPrefixes & DISPREFIX_SEG)
1456	{
1457	/* no effective address which it may apply to. */
1458	Assert((pDis->fPrefix & DISPREFIX_SEG) \|\| pDis->uCpuMode == DISCPUMODE_64BIT);
1459	if ( !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param1.fUse)
1460	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param2.fUse)
1461	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param3.fUse))
1462	return true;
1463
1464	/* Yasm puts the segment prefixes before the operand prefix with no
1465	way of overriding it. */
1466	if (offOpSize < off1stSeg)
1467	return true;
1468	}
1469
1470	/* fixed register + addr override doesn't go down all that well. */
1471	if (fPrefixes & DISPREFIX_ADDRSIZE)
1472	{
1473	Assert(pDis->fPrefix & DISPREFIX_ADDRSIZE);
1474	if ( pDis->pCurInstr->fParam3 == OP_PARM_NONE
1475	&& pDis->pCurInstr->fParam2 == OP_PARM_NONE
1476	&& ( pDis->pCurInstr->fParam1 >= OP_PARM_REG_GEN32_START
1477	&& pDis->pCurInstr->fParam1 <= OP_PARM_REG_GEN32_END))
1478	return true;
1479	}
1480
1481	/* Almost all prefixes are bad for jumps. */
1482	if (fPrefixes)
1483	{
1484	switch (pDis->pCurInstr->uOpcode)
1485	{
1486	/* nop w/ prefix(es). */
1487	case OP_NOP:
1488	return true;
1489
1490	case OP_JMP:
1491	if ( pDis->pCurInstr->fParam1 != OP_PARM_Jb
1492	&& pDis->pCurInstr->fParam1 != OP_PARM_Jv)
1493	break;
1494	/* fall thru */
1495	case OP_JO:
1496	case OP_JNO:
1497	case OP_JC:
1498	case OP_JNC:
1499	case OP_JE:
1500	case OP_JNE:
1501	case OP_JBE:
1502	case OP_JNBE:
1503	case OP_JS:
1504	case OP_JNS:
1505	case OP_JP:
1506	case OP_JNP:
1507	case OP_JL:
1508	case OP_JNL:
1509	case OP_JLE:
1510	case OP_JNLE:
1511	/** @todo branch hinting 0x2e/0x3e... */
1512	return true;
1513	}
1514
1515	}
1516
1517	/* All but the segment prefix is bad news for push/pop. */
1518	if (fPrefixes & ~DISPREFIX_SEG)
1519	{
1520	switch (pDis->pCurInstr->uOpcode)
1521	{
1522	case OP_POP:
1523	case OP_PUSH:
1524	if ( pDis->pCurInstr->fParam1 >= OP_PARM_REG_SEG_START
1525	&& pDis->pCurInstr->fParam1 <= OP_PARM_REG_SEG_END)
1526	return true;
1527	if ( (fPrefixes & ~DISPREFIX_OPSIZE)
1528	&& pDis->pCurInstr->fParam1 >= OP_PARM_REG_GEN32_START
1529	&& pDis->pCurInstr->fParam1 <= OP_PARM_REG_GEN32_END)
1530	return true;
1531	break;
1532
1533	case OP_POPA:
1534	case OP_POPF:
1535	case OP_PUSHA:
1536	case OP_PUSHF:
1537	if (fPrefixes & ~DISPREFIX_OPSIZE)
1538	return true;
1539	break;
1540	}
1541	}
1542
1543	/* Implicit 8-bit register instructions doesn't mix with operand size. */
1544	if ( (fPrefixes & DISPREFIX_OPSIZE)
1545	&& ( ( pDis->pCurInstr->fParam1 == OP_PARM_Gb /* r8 */
1546	&& pDis->pCurInstr->fParam2 == OP_PARM_Eb /* r8/mem8 */)
1547	\|\| ( pDis->pCurInstr->fParam2 == OP_PARM_Gb /* r8 */
1548	&& pDis->pCurInstr->fParam1 == OP_PARM_Eb /* r8/mem8 */))
1549	)
1550	{
1551	switch (pDis->pCurInstr->uOpcode)
1552	{
1553	case OP_ADD:
1554	case OP_OR:
1555	case OP_ADC:
1556	case OP_SBB:
1557	case OP_AND:
1558	case OP_SUB:
1559	case OP_XOR:
1560	case OP_CMP:
1561	return true;
1562	default:
1563	break;
1564	}
1565	}
1566
1567	/* Instructions taking no address or operand which thus may be annoyingly
1568	difficult to format for yasm. */
1569	if (fPrefixes)
1570	{
1571	switch (pDis->pCurInstr->uOpcode)
1572	{
1573	case OP_STI:
1574	case OP_STC:
1575	case OP_CLI:
1576	case OP_CLD:
1577	case OP_CLC:
1578	case OP_INT:
1579	case OP_INT3:
1580	case OP_INTO:
1581	case OP_HLT:
1582	/** @todo Many more to can be added here. */
1583	return true;
1584	default:
1585	break;
1586	}
1587	}
1588
1589	/* FPU and other instructions that ignores operand size override. */
1590	if (fPrefixes & DISPREFIX_OPSIZE)
1591	{
1592	switch (pDis->pCurInstr->uOpcode)
1593	{
1594	/* FPU: */
1595	case OP_FIADD:
1596	case OP_FIMUL:
1597	case OP_FISUB:
1598	case OP_FISUBR:
1599	case OP_FIDIV:
1600	case OP_FIDIVR:
1601	/** @todo there are many more. */
1602	return true;
1603
1604	case OP_MOV:
1605	/** @todo could be that we're not disassembling these correctly. */
1606	if (pDis->pCurInstr->fParam1 == OP_PARM_Sw)
1607	return true;
1608	/** @todo what about the other way? */
1609	break;
1610
1611	default:
1612	break;
1613	}
1614	}
1615
1616
1617	/*
1618	* Check for the version of xyz reg,reg instruction that the assembler doesn't use.
1619	*
1620	* For example:
1621	* expected: 1aee sbb ch, dh ; SBB r8, r/m8
1622	* yasm: 18F5 sbb ch, dh ; SBB r/m8, r8
1623	*/
1624	if (pDis->ModRM.Bits.Mod == 3 /* reg,reg */)
1625	{
1626	switch (pDis->pCurInstr->uOpcode)
1627	{
1628	case OP_ADD:
1629	case OP_OR:
1630	case OP_ADC:
1631	case OP_SBB:
1632	case OP_AND:
1633	case OP_SUB:
1634	case OP_XOR:
1635	case OP_CMP:
1636	if ( ( pDis->pCurInstr->fParam1 == OP_PARM_Gb /* r8 */
1637	&& pDis->pCurInstr->fParam2 == OP_PARM_Eb /* r8/mem8 */)
1638	\|\| ( pDis->pCurInstr->fParam1 == OP_PARM_Gv /* rX */
1639	&& pDis->pCurInstr->fParam2 == OP_PARM_Ev /* rX/memX */))
1640	return true;
1641
1642	/* 82 (see table A-6). */
1643	if (pDis->bOpCode == 0x82)
1644	return true;
1645	break;
1646
1647	/* ff /0, fe /0, ff /1, fe /0 */
1648	case OP_DEC:
1649	case OP_INC:
1650	return true;
1651
1652	case OP_POP:
1653	case OP_PUSH:
1654	Assert(pDis->bOpCode == 0x8f);
1655	return true;
1656
1657	case OP_MOV:
1658	if ( pDis->bOpCode == 0x8a
1659	\|\| pDis->bOpCode == 0x8b)
1660	return true;
1661	break;
1662
1663	default:
1664	break;
1665	}
1666	}
1667
1668	/* shl eax,1 will be assembled to the form without the immediate byte. */
1669	if ( pDis->pCurInstr->fParam2 == OP_PARM_Ib
1670	&& (uint8_t)pDis->Param2.uValue == 1)
1671	{
1672	switch (pDis->pCurInstr->uOpcode)
1673	{
1674	case OP_SHL:
1675	case OP_SHR:
1676	case OP_SAR:
1677	case OP_RCL:
1678	case OP_RCR:
1679	case OP_ROL:
1680	case OP_ROR:
1681	return true;
1682	}
1683	}
1684
1685	/* And some more - see table A-6. */
1686	if (pDis->bOpCode == 0x82)
1687	{
1688	switch (pDis->pCurInstr->uOpcode)
1689	{
1690	case OP_ADD:
1691	case OP_OR:
1692	case OP_ADC:
1693	case OP_SBB:
1694	case OP_AND:
1695	case OP_SUB:
1696	case OP_XOR:
1697	case OP_CMP:
1698	return true;
1699	break;
1700	}
1701	}
1702
1703
1704	/* check for REX.X = 1 without SIB. */
1705
1706	/* Yasm encodes setnbe al with /2 instead of /0 like the AMD manual
1707	says (intel doesn't appear to care). */
1708	switch (pDis->pCurInstr->uOpcode)
1709	{
1710	case OP_SETO:
1711	case OP_SETNO:
1712	case OP_SETC:
1713	case OP_SETNC:
1714	case OP_SETE:
1715	case OP_SETNE:
1716	case OP_SETBE:
1717	case OP_SETNBE:
1718	case OP_SETS:
1719	case OP_SETNS:
1720	case OP_SETP:
1721	case OP_SETNP:
1722	case OP_SETL:
1723	case OP_SETNL:
1724	case OP_SETLE:
1725	case OP_SETNLE:
1726	AssertMsg(pDis->bOpCode >= 0x90 && pDis->bOpCode <= 0x9f, ("%#x\n", pDis->bOpCode));
1727	if (pDis->ModRM.Bits.Reg != 2)
1728	return true;
1729	break;
1730	}
1731
1732	/*
1733	* The MOVZX reg32,mem16 instruction without an operand size prefix
1734	* doesn't quite make sense...
1735	*/
1736	if ( pDis->pCurInstr->uOpcode == OP_MOVZX
1737	&& pDis->bOpCode == 0xB7
1738	&& (pDis->uCpuMode == DISCPUMODE_16BIT) != !!(fPrefixes & DISPREFIX_OPSIZE))
1739	return true;
1740
1741	return false;
1742	}
1743

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

source: vbox/trunk/src/VBox/Disassembler/DisasmFormatYasm.cpp@ 62956

Download in other formats: