DisasmFormatYasm.cpp@ 101540

Last change on this file since 101540 was 101539, checked in by vboxsync, 14 months ago
DIS,VMM,DBGC,IPRT,++: Some disassembler tweaks and TB disassembly work. bugref:10371 bugref:9898
Property svn:eol-style set to `native` Property svn:keywords set to `Id Revision`
File size: 71.5 KB

Line
1	/* $Id: DisasmFormatYasm.cpp 101539 2023-10-22 02:43:09Z vboxsync $ */
2	/** @file
3	* VBox Disassembler - Yasm(/Nasm) Style Formatter.
4	*/
5
6	/*
7	* Copyright (C) 2008-2023 Oracle and/or its affiliates.
8	*
9	* This file is part of VirtualBox base platform packages, as
10	* available from https://www.virtualbox.org.
11	*
12	* This program is free software; you can redistribute it and/or
13	* modify it under the terms of the GNU General Public License
14	* as published by the Free Software Foundation, in version 3 of the
15	* License.
16	*
17	* This program is distributed in the hope that it will be useful, but
18	* WITHOUT ANY WARRANTY; without even the implied warranty of
19	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20	* General Public License for more details.
21	*
22	* You should have received a copy of the GNU General Public License
23	* along with this program; if not, see <https://www.gnu.org/licenses>.
24	*
25	* SPDX-License-Identifier: GPL-3.0-only
26	*/
27
28
29	/*********************************************************************************************************************************
30	* Header Files *
31	*********************************************************************************************************************************/
32	#include <VBox/dis.h>
33	#include "DisasmInternal.h"
34	#include <iprt/assert.h>
35	#include <iprt/ctype.h>
36	#include <iprt/err.h>
37	#include <iprt/string.h>
38
39
40	/*********************************************************************************************************************************
41	* Global Variables *
42	*********************************************************************************************************************************/
43	static const char g_szSpaces[] =
44	" ";
45	static const char g_aszYasmRegGen8[20][5] =
46	{
47	"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"
48	};
49	static const char g_aszYasmRegGen16[16][5] =
50	{
51	"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"
52	};
53	#if 0 /* unused */
54	static const char g_aszYasmRegGen1616[8][6] =
55	{
56	"bx+si", "bx+di", "bp+si", "bp+di", "si\0\0\0", "di\0\0\0", "bp\0\0\0", "bx\0\0\0"
57	};
58	#endif
59	static const char g_aszYasmRegGen32[16][5] =
60	{
61	"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"
62	};
63	static const char g_aszYasmRegGen64[16][4] =
64	{
65	"rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi", "r8\0", "r9\0", "r10", "r11", "r12", "r13", "r14", "r15"
66	};
67	static const char g_aszYasmRegSeg[6][3] =
68	{
69	"es", "cs", "ss", "ds", "fs", "gs"
70	};
71	static const char g_aszYasmRegFP[8][4] =
72	{
73	"st0", "st1", "st2", "st3", "st4", "st5", "st6", "st7"
74	};
75	static const char g_aszYasmRegMMX[8][4] =
76	{
77	"mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7"
78	};
79	static const char g_aszYasmRegXMM[16][6] =
80	{
81	"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"
82	};
83	static const char g_aszYasmRegYMM[16][6] =
84	{
85	"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"
86	};
87	static const char g_aszYasmRegCRx[16][5] =
88	{
89	"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"
90	};
91	static const char g_aszYasmRegDRx[16][5] =
92	{
93	"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"
94	};
95	static const char g_aszYasmRegTRx[16][5] =
96	{
97	"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"
98	};
99
100
101
102	/**
103	* Gets the base register name for the given parameter.
104	*
105	* @returns Pointer to the register name.
106	* @param pDis The disassembler state.
107	* @param pParam The parameter.
108	* @param pcchReg Where to store the length of the name.
109	*/
110	static const char disasmFormatYasmBaseReg(PCDISSTATE pDis, PCDISOPPARAM pParam, size_t pcchReg)
111	{
112	RT_NOREF_PV(pDis);
113
114	switch (pParam->fUse & ( DISUSE_REG_GEN8 \| DISUSE_REG_GEN16 \| DISUSE_REG_GEN32 \| DISUSE_REG_GEN64
115	\| DISUSE_REG_FP \| DISUSE_REG_MMX \| DISUSE_REG_XMM \| DISUSE_REG_YMM
116	\| DISUSE_REG_CR \| DISUSE_REG_DBG \| DISUSE_REG_SEG \| DISUSE_REG_TEST))
117
118	{
119	case DISUSE_REG_GEN8:
120	{
121	Assert(pParam->x86.Base.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen8));
122	const char *psz = g_aszYasmRegGen8[pParam->x86.Base.idxGenReg];
123	*pcchReg = 2 + !!psz[2] + !!psz[3];
124	return psz;
125	}
126
127	case DISUSE_REG_GEN16:
128	{
129	Assert(pParam->x86.Base.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen16));
130	const char *psz = g_aszYasmRegGen16[pParam->x86.Base.idxGenReg];
131	*pcchReg = 2 + !!psz[2] + !!psz[3];
132	return psz;
133	}
134
135	// VSIB
136	case DISUSE_REG_XMM \| DISUSE_REG_GEN32:
137	case DISUSE_REG_YMM \| DISUSE_REG_GEN32:
138	case DISUSE_REG_GEN32:
139	{
140	Assert(pParam->x86.Base.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen32));
141	const char *psz = g_aszYasmRegGen32[pParam->x86.Base.idxGenReg];
142	*pcchReg = 2 + !!psz[2] + !!psz[3];
143	return psz;
144	}
145
146	// VSIB
147	case DISUSE_REG_XMM \| DISUSE_REG_GEN64:
148	case DISUSE_REG_YMM \| DISUSE_REG_GEN64:
149	case DISUSE_REG_GEN64:
150	{
151	Assert(pParam->x86.Base.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen64));
152	const char *psz = g_aszYasmRegGen64[pParam->x86.Base.idxGenReg];
153	*pcchReg = 2 + !!psz[2] + !!psz[3];
154	return psz;
155	}
156
157	case DISUSE_REG_FP:
158	{
159	Assert(pParam->x86.Base.idxFpuReg < RT_ELEMENTS(g_aszYasmRegFP));
160	const char *psz = g_aszYasmRegFP[pParam->x86.Base.idxFpuReg];
161	*pcchReg = 3;
162	return psz;
163	}
164
165	case DISUSE_REG_MMX:
166	{
167	Assert(pParam->x86.Base.idxMmxReg < RT_ELEMENTS(g_aszYasmRegMMX));
168	const char *psz = g_aszYasmRegMMX[pParam->x86.Base.idxMmxReg];
169	*pcchReg = 3;
170	return psz;
171	}
172
173	case DISUSE_REG_XMM:
174	{
175	Assert(pParam->x86.Base.idxXmmReg < RT_ELEMENTS(g_aszYasmRegXMM));
176	const char *psz = g_aszYasmRegXMM[pParam->x86.Base.idxXmmReg];
177	*pcchReg = 4 + !!psz[4];
178	return psz;
179	}
180
181	case DISUSE_REG_YMM:
182	{
183	Assert(pParam->x86.Base.idxYmmReg < RT_ELEMENTS(g_aszYasmRegYMM));
184	const char *psz = g_aszYasmRegYMM[pParam->x86.Base.idxYmmReg];
185	*pcchReg = 4 + !!psz[4];
186	return psz;
187	}
188
189	case DISUSE_REG_CR:
190	{
191	Assert(pParam->x86.Base.idxCtrlReg < RT_ELEMENTS(g_aszYasmRegCRx));
192	const char *psz = g_aszYasmRegCRx[pParam->x86.Base.idxCtrlReg];
193	*pcchReg = 3;
194	return psz;
195	}
196
197	case DISUSE_REG_DBG:
198	{
199	Assert(pParam->x86.Base.idxDbgReg < RT_ELEMENTS(g_aszYasmRegDRx));
200	const char *psz = g_aszYasmRegDRx[pParam->x86.Base.idxDbgReg];
201	*pcchReg = 3;
202	return psz;
203	}
204
205	case DISUSE_REG_SEG:
206	{
207	Assert(pParam->x86.Base.idxSegReg < RT_ELEMENTS(g_aszYasmRegCRx));
208	const char *psz = g_aszYasmRegSeg[pParam->x86.Base.idxSegReg];
209	*pcchReg = 2;
210	return psz;
211	}
212
213	case DISUSE_REG_TEST:
214	{
215	Assert(pParam->x86.Base.idxTestReg < RT_ELEMENTS(g_aszYasmRegTRx));
216	const char *psz = g_aszYasmRegTRx[pParam->x86.Base.idxTestReg];
217	*pcchReg = 3;
218	return psz;
219	}
220
221	default:
222	AssertMsgFailed(("%#x\n", pParam->fUse));
223	*pcchReg = 3;
224	return "r??";
225	}
226	}
227
228
229	/**
230	* Gets the index register name for the given parameter.
231	*
232	* @returns The index register name.
233	* @param pDis The disassembler state.
234	* @param pParam The parameter.
235	* @param pcchReg Where to store the length of the name.
236	*/
237	static const char disasmFormatYasmIndexReg(PCDISSTATE pDis, PCDISOPPARAM pParam, size_t pcchReg)
238	{
239	if (pParam->fUse & DISUSE_REG_XMM)
240	{
241	Assert(pParam->x86.Index.idxXmmReg < RT_ELEMENTS(g_aszYasmRegXMM));
242	const char *psz = g_aszYasmRegXMM[pParam->x86.Index.idxXmmReg];
243	*pcchReg = 4 + !!psz[4];
244	return psz;
245	}
246	else if (pParam->fUse & DISUSE_REG_YMM)
247	{
248	Assert(pParam->x86.Index.idxYmmReg < RT_ELEMENTS(g_aszYasmRegYMM));
249	const char *psz = g_aszYasmRegYMM[pParam->x86.Index.idxYmmReg];
250	*pcchReg = 4 + !!psz[4];
251	return psz;
252
253	}
254	else
255	switch (pDis->x86.uAddrMode)
256	{
257	case DISCPUMODE_16BIT:
258	{
259	Assert(pParam->x86.Index.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen16));
260	const char *psz = g_aszYasmRegGen16[pParam->x86.Index.idxGenReg];
261	*pcchReg = 2 + !!psz[2] + !!psz[3];
262	return psz;
263	}
264
265	case DISCPUMODE_32BIT:
266	{
267	Assert(pParam->x86.Index.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen32));
268	const char *psz = g_aszYasmRegGen32[pParam->x86.Index.idxGenReg];
269	*pcchReg = 2 + !!psz[2] + !!psz[3];
270	return psz;
271	}
272
273	case DISCPUMODE_64BIT:
274	{
275	Assert(pParam->x86.Index.idxGenReg < RT_ELEMENTS(g_aszYasmRegGen64));
276	const char *psz = g_aszYasmRegGen64[pParam->x86.Index.idxGenReg];
277	*pcchReg = 2 + !!psz[2] + !!psz[3];
278	return psz;
279	}
280
281	default:
282	AssertMsgFailed(("%#x %#x\n", pParam->fUse, pDis->x86.uAddrMode));
283	*pcchReg = 3;
284	return "r??";
285	}
286	}
287
288
289	/**
290	* Formats the current instruction in Yasm (/ Nasm) style.
291	*
292	*
293	* @returns The number of output characters. If this is >= cchBuf, then the content
294	* of pszBuf will be truncated.
295	* @param pDis Pointer to the disassembler state.
296	* @param pszBuf The output buffer.
297	* @param cchBuf The size of the output buffer.
298	* @param fFlags Format flags, see DIS_FORMAT_FLAGS_*.
299	* @param pfnGetSymbol Get symbol name for a jmp or call target address. Optional.
300	* @param pvUser User argument for pfnGetSymbol.
301	*/
302	DISDECL(size_t) DISFormatYasmEx(PCDISSTATE pDis, char *pszBuf, size_t cchBuf, uint32_t fFlags,
303	PFNDISGETSYMBOL pfnGetSymbol, void *pvUser)
304	{
305	/** @todo monitor and mwait aren't formatted correctly in 64-bit mode. */
306	/*
307	* Input validation and massaging.
308	*/
309	AssertPtr(pDis);
310	AssertPtrNull(pszBuf);
311	Assert(pszBuf \|\| !cchBuf);
312	AssertPtrNull(pfnGetSymbol);
313	AssertMsg(DIS_FMT_FLAGS_IS_VALID(fFlags), ("%#x\n", fFlags));
314	if (fFlags & DIS_FMT_FLAGS_ADDR_COMMENT)
315	fFlags = (fFlags & ~DIS_FMT_FLAGS_ADDR_LEFT) \| DIS_FMT_FLAGS_ADDR_RIGHT;
316	if (fFlags & DIS_FMT_FLAGS_BYTES_COMMENT)
317	fFlags = (fFlags & ~DIS_FMT_FLAGS_BYTES_LEFT) \| DIS_FMT_FLAGS_BYTES_RIGHT;
318
319	PCDISOPCODE const pOp = pDis->pCurInstr;
320
321	/*
322	* Output macros
323	*/
324	char *pszDst = pszBuf;
325	size_t cchDst = cchBuf;
326	size_t cchOutput = 0;
327	#define PUT_C(ch) \
328	do { \
329	cchOutput++; \
330	if (cchDst > 1) \
331	{ \
332	cchDst--; \
333	*pszDst++ = (ch); \
334	} \
335	} while (0)
336	#define PUT_STR(pszSrc, cchSrc) \
337	do { \
338	cchOutput += (cchSrc); \
339	if (cchDst > (cchSrc)) \
340	{ \
341	memcpy(pszDst, (pszSrc), (cchSrc)); \
342	pszDst += (cchSrc); \
343	cchDst -= (cchSrc); \
344	} \
345	else if (cchDst > 1) \
346	{ \
347	memcpy(pszDst, (pszSrc), cchDst - 1); \
348	pszDst += cchDst - 1; \
349	cchDst = 1; \
350	} \
351	} while (0)
352	#define PUT_SZ(sz) \
353	PUT_STR((sz), sizeof(sz) - 1)
354	#define PUT_SZ_STRICT(szStrict, szRelaxed) \
355	do { if (fFlags & DIS_FMT_FLAGS_STRICT) PUT_SZ(szStrict); else PUT_SZ(szRelaxed); } while (0)
356	#define PUT_PSZ(psz) \
357	do { const size_t cchTmp = strlen(psz); PUT_STR((psz), cchTmp); } while (0)
358	#define PUT_NUM(cch, fmt, num) \
359	do { \
360	cchOutput += (cch); \
361	if (cchDst > 1) \
362	{ \
363	const size_t cchTmp = RTStrPrintf(pszDst, cchDst, fmt, (num)); \
364	pszDst += cchTmp; \
365	cchDst -= cchTmp; \
366	Assert(cchTmp == (cch) \|\| cchDst == 1); \
367	} \
368	} while (0)
369	/** @todo add two flags for choosing between %X / %x and h / 0x. */
370	#define PUT_NUM_8(num) PUT_NUM(4, "0%02xh", (uint8_t)(num))
371	#define PUT_NUM_16(num) PUT_NUM(6, "0%04xh", (uint16_t)(num))
372	#define PUT_NUM_32(num) PUT_NUM(10, "0%08xh", (uint32_t)(num))
373	#define PUT_NUM_64(num) PUT_NUM(18, "0%016RX64h", (uint64_t)(num))
374
375	#define PUT_NUM_SIGN(cch, fmt, num, stype, utype) \
376	do { \
377	if ((stype)(num) >= 0) \
378	{ \
379	PUT_C('+'); \
380	PUT_NUM(cch, fmt, (utype)(num)); \
381	} \
382	else \
383	{ \
384	PUT_C('-'); \
385	PUT_NUM(cch, fmt, (utype)-(stype)(num)); \
386	} \
387	} while (0)
388	#define PUT_NUM_S8(num) PUT_NUM_SIGN(4, "0%02xh", num, int8_t, uint8_t)
389	#define PUT_NUM_S16(num) PUT_NUM_SIGN(6, "0%04xh", num, int16_t, uint16_t)
390	#define PUT_NUM_S32(num) PUT_NUM_SIGN(10, "0%08xh", num, int32_t, uint32_t)
391	#define PUT_NUM_S64(num) PUT_NUM_SIGN(18, "0%016RX64h", num, int64_t, uint64_t)
392
393	#define PUT_SYMBOL_TWO(a_rcSym, a_szStart, a_chEnd) \
394	do { \
395	if (RT_SUCCESS(a_rcSym)) \
396	{ \
397	PUT_SZ(a_szStart); \
398	PUT_PSZ(szSymbol); \
399	if (off != 0) \
400	{ \
401	if ((int8_t)off == off) \
402	PUT_NUM_S8(off); \
403	else if ((int16_t)off == off) \
404	PUT_NUM_S16(off); \
405	else if ((int32_t)off == off) \
406	PUT_NUM_S32(off); \
407	else \
408	PUT_NUM_S64(off); \
409	} \
410	PUT_C(a_chEnd); \
411	} \
412	} while (0)
413
414	#define PUT_SYMBOL(a_uSeg, a_uAddr, a_szStart, a_chEnd) \
415	do { \
416	if (pfnGetSymbol) \
417	{ \
418	int rcSym = pfnGetSymbol(pDis, a_uSeg, a_uAddr, szSymbol, sizeof(szSymbol), &off, pvUser); \
419	PUT_SYMBOL_TWO(rcSym, a_szStart, a_chEnd); \
420	} \
421	} while (0)
422
423
424	/*
425	* The address?
426	*/
427	if (fFlags & DIS_FMT_FLAGS_ADDR_LEFT)
428	{
429	#if HC_ARCH_BITS == 64 \|\| GC_ARCH_BITS == 64
430	if (pDis->uInstrAddr >= _4G)
431	PUT_NUM(9, "%08x`", (uint32_t)(pDis->uInstrAddr >> 32));
432	#endif
433	PUT_NUM(8, "%08x", (uint32_t)pDis->uInstrAddr);
434	PUT_C(' ');
435	}
436
437	/*
438	* The opcode bytes?
439	*/
440	if (fFlags & DIS_FMT_FLAGS_BYTES_LEFT)
441	{
442	size_t cchTmp = disFormatBytes(pDis, pszDst, cchDst, fFlags);
443	cchOutput += cchTmp;
444	if (cchDst > 1)
445	{
446	if (cchTmp <= cchDst)
447	{
448	cchDst -= cchTmp;
449	pszDst += cchTmp;
450	}
451	else
452	{
453	pszDst += cchDst - 1;
454	cchDst = 1;
455	}
456	}
457
458	/* Some padding to align the instruction. */
459	size_t cchPadding = (7 * (2 + !!(fFlags & DIS_FMT_FLAGS_BYTES_SPACED)))
460	+ !!(fFlags & DIS_FMT_FLAGS_BYTES_BRACKETS) * 2
461	+ 2;
462	cchPadding = cchTmp + 1 >= cchPadding ? 1 : cchPadding - cchTmp;
463	PUT_STR(g_szSpaces, cchPadding);
464	}
465
466
467	/*
468	* Filter out invalid opcodes first as they need special
469	* treatment. UD2 is an exception and should be handled normally.
470	*/
471	size_t const offInstruction = cchOutput;
472	if ( pOp->uOpcode == OP_INVALID
473	\|\| ( pOp->uOpcode == OP_ILLUD2
474	&& (pDis->x86.fPrefix & DISPREFIX_LOCK)))
475	PUT_SZ("Illegal opcode");
476	else
477	{
478	/*
479	* Prefixes
480	*/
481	if (pDis->x86.fPrefix & DISPREFIX_LOCK)
482	PUT_SZ("lock ");
483	if (pDis->x86.fPrefix & DISPREFIX_REP)
484	PUT_SZ("rep ");
485	else if(pDis->x86.fPrefix & DISPREFIX_REPNE)
486	PUT_SZ("repne ");
487
488	/*
489	* Adjust the format string to the correct mnemonic
490	* or to avoid things the assembler cannot handle correctly.
491	*/
492	char szTmpFmt[48];
493	const char *pszFmt = pOp->pszOpcode;
494	bool fIgnoresOpSize = false;
495	bool fMayNeedAddrSize = false;
496	switch (pOp->uOpcode)
497	{
498	case OP_JECXZ:
499	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "jcxz %Jb" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "jecxz %Jb" : "jrcxz %Jb";
500	break;
501	case OP_PUSHF:
502	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "pushfw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "pushfd" : "pushfq";
503	break;
504	case OP_POPF:
505	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "popfw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "popfd" : "popfq";
506	break;
507	case OP_PUSHA:
508	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "pushaw" : "pushad";
509	break;
510	case OP_POPA:
511	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "popaw" : "popad";
512	break;
513	case OP_INSB:
514	pszFmt = "insb";
515	fIgnoresOpSize = fMayNeedAddrSize = true;
516	break;
517	case OP_INSWD:
518	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "insw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "insd" : "insq";
519	fMayNeedAddrSize = true;
520	break;
521	case OP_OUTSB:
522	pszFmt = "outsb";
523	fIgnoresOpSize = fMayNeedAddrSize = true;
524	break;
525	case OP_OUTSWD:
526	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "outsw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "outsd" : "outsq";
527	fMayNeedAddrSize = true;
528	break;
529	case OP_MOVSB:
530	pszFmt = "movsb";
531	fIgnoresOpSize = fMayNeedAddrSize = true;
532	break;
533	case OP_MOVSWD:
534	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "movsw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "movsd" : "movsq";
535	fMayNeedAddrSize = true;
536	break;
537	case OP_CMPSB:
538	pszFmt = "cmpsb";
539	fIgnoresOpSize = fMayNeedAddrSize = true;
540	break;
541	case OP_CMPWD:
542	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "cmpsw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "cmpsd" : "cmpsq";
543	fMayNeedAddrSize = true;
544	break;
545	case OP_SCASB:
546	pszFmt = "scasb";
547	fIgnoresOpSize = fMayNeedAddrSize = true;
548	break;
549	case OP_SCASWD:
550	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "scasw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "scasd" : "scasq";
551	fMayNeedAddrSize = true;
552	break;
553	case OP_LODSB:
554	pszFmt = "lodsb";
555	fIgnoresOpSize = fMayNeedAddrSize = true;
556	break;
557	case OP_LODSWD:
558	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "lodsw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "lodsd" : "lodsq";
559	fMayNeedAddrSize = true;
560	break;
561	case OP_STOSB:
562	pszFmt = "stosb";
563	fIgnoresOpSize = fMayNeedAddrSize = true;
564	break;
565	case OP_STOSWD:
566	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "stosw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "stosd" : "stosq";
567	fMayNeedAddrSize = true;
568	break;
569	case OP_CBW:
570	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "cbw" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "cwde" : "cdqe";
571	break;
572	case OP_CWD:
573	pszFmt = pDis->x86.uOpMode == DISCPUMODE_16BIT ? "cwd" : pDis->x86.uOpMode == DISCPUMODE_32BIT ? "cdq" : "cqo";
574	break;
575	case OP_SHL:
576	Assert(pszFmt[3] == '/');
577	pszFmt += 4;
578	break;
579	case OP_XLAT:
580	pszFmt = "xlatb";
581	break;
582	case OP_INT3:
583	pszFmt = "int3";
584	break;
585
586	/*
587	* Don't know how to tell yasm to generate complicated nop stuff, so 'db' it.
588	*/
589	case OP_NOP:
590	if (pDis->x86.bOpCode == 0x90)
591	/* fine, fine */;
592	else if (pszFmt[sizeof("nop %Ev") - 1] == '/' && pszFmt[sizeof("nop %Ev")] == 'p')
593	pszFmt = "prefetch %Eb";
594	else if (pDis->x86.bOpCode == 0x1f)
595	{
596	Assert(pDis->cbInstr >= 3);
597	PUT_SZ("db 00fh, 01fh,");
598	PUT_NUM_8(MAKE_MODRM(pDis->x86.ModRM.Bits.Mod, pDis->x86.ModRM.Bits.Reg, pDis->x86.ModRM.Bits.Rm));
599	for (unsigned i = 3; i < pDis->cbInstr; i++)
600	{
601	PUT_C(',');
602	PUT_NUM_8(0x90); /// @todo fixme.
603	}
604	pszFmt = "";
605	}
606	break;
607
608	default:
609	/* ST(X) -> stX (floating point) */
610	if (*pszFmt == 'f' && strchr(pszFmt, '('))
611	{
612	char *pszFmtDst = szTmpFmt;
613	char ch;
614	do
615	{
616	ch = *pszFmt++;
617	if (ch == 'S' && pszFmt[0] == 'T' && pszFmt[1] == '(')
618	{
619	*pszFmtDst++ = 's';
620	*pszFmtDst++ = 't';
621	pszFmt += 2;
622	ch = *pszFmt;
623	Assert(pszFmt[1] == ')');
624	pszFmt += 2;
625	*pszFmtDst++ = ch;
626	}
627	else
628	*pszFmtDst++ = ch;
629	} while (ch != '\0');
630	pszFmt = szTmpFmt;
631	}
632	if (strchr("#@&", *pszFmt))
633	{
634	const char *pszDelim = strchr(pszFmt, '/');
635	const char *pszSpace = (pszDelim ? strchr(pszDelim, ' ') : NULL);
636	if (pszDelim != NULL)
637	{
638	char *pszFmtDst = szTmpFmt;
639	if (pszSpace == NULL) pszSpace = strchr(pszDelim, 0);
640	if ( (pszFmt == '#' && !pDis->x86.bVexWFlag) /* @todo check this*/
641	\|\| (*pszFmt == '@' && !VEXREG_IS256B(pDis->x86.bVexDestReg))
642	\|\| (*pszFmt == '&' && ( DISUSE_IS_EFFECTIVE_ADDR(pDis->Param1.fUse)
643	\|\| DISUSE_IS_EFFECTIVE_ADDR(pDis->Param2.fUse)
644	\|\| DISUSE_IS_EFFECTIVE_ADDR(pDis->Param3.fUse)
645	\|\| DISUSE_IS_EFFECTIVE_ADDR(pDis->Param4.fUse))))
646	{
647	strncpy(pszFmtDst, pszFmt + 1, pszDelim - pszFmt - 1);
648	pszFmtDst += pszDelim - pszFmt - 1;
649	}
650	else
651	{
652	strncpy(pszFmtDst, pszDelim + 1, pszSpace - pszDelim - 1);
653	pszFmtDst += pszSpace - pszDelim - 1;
654	}
655	strcpy (pszFmtDst, pszSpace);
656	pszFmt = szTmpFmt;
657	}
658	}
659	break;
660
661	/*
662	* Horrible hacks.
663	*/
664	case OP_FLD:
665	if (pDis->x86.bOpCode == 0xdb) /* m80fp workaround. */
666	(int )&pDis->Param1.x86.fParam &= ~0x1f; /* make it pure OP_PARM_M */
667	break;
668	case OP_LAR: /* hack w -> v, probably not correct. */
669	(int )&pDis->Param2.x86.fParam &= ~0x1f;
670	(int )&pDis->Param2.x86.fParam \|= OP_PARM_v;
671	break;
672	}
673
674	/*
675	* Add operand size and address prefixes for outsb, movsb, etc.
676	*/
677	if (pDis->x86.fPrefix & (DISPREFIX_OPSIZE \| DISPREFIX_ADDRSIZE))
678	{
679	if (fIgnoresOpSize && (pDis->x86.fPrefix & DISPREFIX_OPSIZE) )
680	{
681	if (pDis->uCpuMode == DISCPUMODE_16BIT)
682	PUT_SZ("o32 ");
683	else
684	PUT_SZ("o16 ");
685	}
686	if (fMayNeedAddrSize && (pDis->x86.fPrefix & DISPREFIX_ADDRSIZE) )
687	{
688	if (pDis->uCpuMode == DISCPUMODE_16BIT)
689	PUT_SZ("a32 ");
690	else
691	PUT_SZ("a16 ");
692	}
693	}
694
695	/*
696	* Formatting context and associated macros.
697	*/
698	PCDISOPPARAM pParam = &pDis->Param1;
699	int iParam = 1;
700
701	#define PUT_FAR() \
702	do { \
703	if ( OP_PARM_VSUBTYPE(pParam->x86.fParam) == OP_PARM_p \
704	&& pOp->uOpcode != OP_LDS /* table bugs? */ \
705	&& pOp->uOpcode != OP_LES \
706	&& pOp->uOpcode != OP_LFS \
707	&& pOp->uOpcode != OP_LGS \
708	&& pOp->uOpcode != OP_LSS ) \
709	PUT_SZ("far "); \
710	} while (0)
711	/** @todo mov ah,ch ends up with a byte 'override'... - check if this wasn't fixed. */
712	/** @todo drop the work/dword/qword override when the src/dst is a register (except for movsx/movzx). */
713	#define PUT_SIZE_OVERRIDE() \
714	do { \
715	switch (OP_PARM_VSUBTYPE(pParam->x86.fParam)) \
716	{ \
717	case OP_PARM_v: \
718	case OP_PARM_y: \
719	switch (pDis->x86.uOpMode) \
720	{ \
721	case DISCPUMODE_16BIT: if (OP_PARM_VSUBTYPE(pParam->x86.fParam) != OP_PARM_y) PUT_SZ("word "); break; \
722	case DISCPUMODE_32BIT: \
723	if (pDis->pCurInstr->uOpcode != OP_GATHER \|\| pDis->x86.bVexWFlag) { PUT_SZ("dword "); break; } \
724	RT_FALL_THRU(); \
725	case DISCPUMODE_64BIT: PUT_SZ("qword "); break; \
726	default: break; \
727	} \
728	break; \
729	case OP_PARM_b: PUT_SZ("byte "); break; \
730	case OP_PARM_w: \
731	if ( OP_PARM_VTYPE(pParam->x86.fParam) == OP_PARM_W \
732	\|\| OP_PARM_VTYPE(pParam->x86.fParam) == OP_PARM_M) \
733	{ \
734	if (VEXREG_IS256B(pDis->x86.bVexDestReg)) PUT_SZ("dword "); \
735	else PUT_SZ("word "); \
736	} \
737	break; \
738	case OP_PARM_d: \
739	if ( OP_PARM_VTYPE(pParam->x86.fParam) == OP_PARM_W \
740	\|\| OP_PARM_VTYPE(pParam->x86.fParam) == OP_PARM_M) \
741	{ \
742	if (VEXREG_IS256B(pDis->x86.bVexDestReg)) PUT_SZ("qword "); \
743	else PUT_SZ("dword "); \
744	} \
745	break; \
746	case OP_PARM_q: \
747	if ( OP_PARM_VTYPE(pParam->x86.fParam) == OP_PARM_W \
748	\|\| OP_PARM_VTYPE(pParam->x86.fParam) == OP_PARM_M) \
749	{ \
750	if (VEXREG_IS256B(pDis->x86.bVexDestReg)) PUT_SZ("oword "); \
751	else PUT_SZ("qword "); \
752	} \
753	break; \
754	case OP_PARM_ps: \
755	case OP_PARM_pd: \
756	case OP_PARM_x: if (VEXREG_IS256B(pDis->x86.bVexDestReg)) { PUT_SZ("yword "); break; } RT_FALL_THRU(); \
757	case OP_PARM_ss: \
758	case OP_PARM_sd: \
759	case OP_PARM_dq: PUT_SZ("oword "); break; \
760	case OP_PARM_qq: PUT_SZ("yword "); break; \
761	case OP_PARM_p: break; /* see PUT_FAR */ \
762	case OP_PARM_s: if (pParam->fUse & DISUSE_REG_FP) PUT_SZ("tword "); break; /* ?? */ \
763	case OP_PARM_z: break; \
764	case OP_PARM_NONE: \
765	if ( OP_PARM_VTYPE(pParam->x86.fParam) == OP_PARM_M \
766	&& ((pParam->fUse & DISUSE_REG_FP) \|\| pOp->uOpcode == OP_FLD)) \
767	PUT_SZ("tword "); \
768	break; \
769	default: break; /no pointer type specified/necessary/ \
770	} \
771	} while (0)
772	static const char s_szSegPrefix[6][4] = { "es:", "cs:", "ss:", "ds:", "fs:", "gs:" };
773	#define PUT_SEGMENT_OVERRIDE() \
774	do { \
775	if (pDis->x86.fPrefix & DISPREFIX_SEG) \
776	PUT_STR(s_szSegPrefix[pDis->x86.idxSegPrefix], 3); \
777	} while (0)
778
779
780	/*
781	* Segment prefixing for instructions that doesn't do memory access.
782	*/
783	if ( (pDis->x86.fPrefix & DISPREFIX_SEG)
784	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param1.fUse)
785	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param2.fUse)
786	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param3.fUse))
787	{
788	PUT_STR(s_szSegPrefix[pDis->x86.idxSegPrefix], 2);
789	PUT_C(' ');
790	}
791
792
793	/*
794	* The formatting loop.
795	*/
796	RTINTPTR off;
797	char szSymbol[128];
798	char ch;
799	while ((ch = *pszFmt++) != '\0')
800	{
801	if (ch == '%')
802	{
803	ch = *pszFmt++;
804	switch (ch)
805	{
806	/*
807	* ModRM - Register only / VEX.vvvv.
808	*/
809	case 'C': /* Control register (ParseModRM / UseModRM). */
810	case 'D': /* Debug register (ParseModRM / UseModRM). */
811	case 'G': /* ModRM selects general register (ParseModRM / UseModRM). */
812	case 'S': /* ModRM byte selects a segment register (ParseModRM / UseModRM). */
813	case 'T': /* ModRM byte selects a test register (ParseModRM / UseModRM). */
814	case 'V': /* ModRM byte selects an XMM/SSE register (ParseModRM / UseModRM). */
815	case 'P': /* ModRM byte selects MMX register (ParseModRM / UseModRM). */
816	case 'H': /* The VEX.vvvv field of the VEX prefix selects a XMM/YMM register. */
817	case 'B': /* The VEX.vvvv field of the VEX prefix selects a general register (ParseVexDest). */
818	case 'L': /* The upper 4 bits of the 8-bit immediate selects a XMM/YMM register. */
819	{
820	pszFmt += RT_C_IS_ALPHA(pszFmt[0]) ? RT_C_IS_ALPHA(pszFmt[1]) ? 2 : 1 : 0;
821	Assert(!(pParam->fUse & (DISUSE_INDEX \| DISUSE_SCALE) /* No SIB here... */));
822	Assert(!(pParam->fUse & (DISUSE_DISPLACEMENT8 \| DISUSE_DISPLACEMENT16 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT64 \| DISUSE_RIPDISPLACEMENT32)));
823
824	size_t cchReg;
825	const char *pszReg = disasmFormatYasmBaseReg(pDis, pParam, &cchReg);
826	PUT_STR(pszReg, cchReg);
827	break;
828	}
829
830	/*
831	* ModRM - Register or memory.
832	*/
833	case 'E': /* ModRM specifies parameter (ParseModRM / UseModRM / UseSIB). */
834	case 'Q': /* ModRM byte selects MMX register or memory address (ParseModRM / UseModRM). */
835	case 'R': /* ModRM byte may only refer to a general register (ParseModRM / UseModRM). */
836	case 'W': /* ModRM byte selects an XMM/SSE register or a memory address (ParseModRM / UseModRM). */
837	case 'U': /* ModRM byte may only refer to a XMM/SSE register (ParseModRM / UseModRM). */
838	case 'M': /* ModRM byte may only refer to memory (ParseModRM / UseModRM). */
839	{
840	pszFmt += RT_C_IS_ALPHA(pszFmt[0]) ? RT_C_IS_ALPHA(pszFmt[1]) ? 2 : 1 : 0;
841
842	PUT_FAR();
843	uint32_t const fUse = pParam->fUse;
844	if (DISUSE_IS_EFFECTIVE_ADDR(fUse))
845	{
846	/* Work around mov seg,[mem16] and mov [mem16],seg as these always make a 16-bit mem
847	while the register variants deals with 16, 32 & 64 in the normal fashion. */
848	if ( pParam->x86.fParam != OP_PARM_Ev
849	\|\| pOp->uOpcode != OP_MOV
850	\|\| ( pOp->fParam1 != OP_PARM_Sw
851	&& pOp->fParam2 != OP_PARM_Sw))
852	PUT_SIZE_OVERRIDE();
853	PUT_C('[');
854	}
855	if ( (fFlags & DIS_FMT_FLAGS_STRICT)
856	&& (fUse & (DISUSE_DISPLACEMENT8 \| DISUSE_DISPLACEMENT16 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT64 \| DISUSE_RIPDISPLACEMENT32)))
857	{
858	if ( (fUse & DISUSE_DISPLACEMENT8)
859	&& !pParam->x86.uDisp.i8)
860	PUT_SZ("byte ");
861	else if ( (fUse & DISUSE_DISPLACEMENT16)
862	&& (int8_t)pParam->x86.uDisp.i16 == (int16_t)pParam->x86.uDisp.i16)
863	PUT_SZ("word ");
864	else if ( (fUse & DISUSE_DISPLACEMENT32)
865	&& (int16_t)pParam->x86.uDisp.i32 == (int32_t)pParam->x86.uDisp.i32) //??
866	PUT_SZ("dword ");
867	else if ( (fUse & DISUSE_DISPLACEMENT64)
868	&& (pDis->x86.SIB.Bits.Base != 5 \|\| pDis->x86.ModRM.Bits.Mod != 0)
869	&& (int32_t)pParam->x86.uDisp.i64 == (int64_t)pParam->x86.uDisp.i64) //??
870	PUT_SZ("qword ");
871	}
872	if (DISUSE_IS_EFFECTIVE_ADDR(fUse))
873	PUT_SEGMENT_OVERRIDE();
874
875	bool fBase = (fUse & DISUSE_BASE) /* When exactly is DISUSE_BASE supposed to be set? disasmModRMReg doesn't set it. */
876	\|\| ( (fUse & ( DISUSE_REG_GEN8
877	\| DISUSE_REG_GEN16
878	\| DISUSE_REG_GEN32
879	\| DISUSE_REG_GEN64
880	\| DISUSE_REG_FP
881	\| DISUSE_REG_MMX
882	\| DISUSE_REG_XMM
883	\| DISUSE_REG_YMM
884	\| DISUSE_REG_CR
885	\| DISUSE_REG_DBG
886	\| DISUSE_REG_SEG
887	\| DISUSE_REG_TEST ))
888	&& !DISUSE_IS_EFFECTIVE_ADDR(fUse));
889	if (fBase)
890	{
891	size_t cchReg;
892	const char *pszReg = disasmFormatYasmBaseReg(pDis, pParam, &cchReg);
893	PUT_STR(pszReg, cchReg);
894	}
895
896	if (fUse & DISUSE_INDEX)
897	{
898	if (fBase)
899	PUT_C('+');
900
901	size_t cchReg;
902	const char *pszReg = disasmFormatYasmIndexReg(pDis, pParam, &cchReg);
903	PUT_STR(pszReg, cchReg);
904
905	if (fUse & DISUSE_SCALE)
906	{
907	PUT_C('*');
908	PUT_C('0' + pParam->x86.uScale);
909	}
910	}
911	else
912	Assert(!(fUse & DISUSE_SCALE));
913
914	int64_t off2 = 0;
915	if (fUse & (DISUSE_DISPLACEMENT8 \| DISUSE_DISPLACEMENT16 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT64 \| DISUSE_RIPDISPLACEMENT32))
916	{
917	if (fUse & DISUSE_DISPLACEMENT8)
918	off2 = pParam->x86.uDisp.i8;
919	else if (fUse & DISUSE_DISPLACEMENT16)
920	off2 = pParam->x86.uDisp.i16;
921	else if (fUse & (DISUSE_DISPLACEMENT32 \| DISUSE_RIPDISPLACEMENT32))
922	off2 = pParam->x86.uDisp.i32;
923	else if (fUse & DISUSE_DISPLACEMENT64)
924	off2 = pParam->x86.uDisp.i64;
925	else
926	{
927	AssertFailed();
928	off2 = 0;
929	}
930
931	int64_t off3 = off2;
932	if (fBase \|\| (fUse & (DISUSE_INDEX \| DISUSE_RIPDISPLACEMENT32)))
933	{
934	PUT_C(off3 >= 0 ? '+' : '-');
935	if (off3 < 0)
936	off3 = -off3;
937	}
938	if (fUse & DISUSE_DISPLACEMENT8)
939	PUT_NUM_8( off3);
940	else if (fUse & DISUSE_DISPLACEMENT16)
941	PUT_NUM_16(off3);
942	else if (fUse & DISUSE_DISPLACEMENT32)
943	PUT_NUM_32(off3);
944	else if (fUse & DISUSE_DISPLACEMENT64)
945	PUT_NUM_64(off3);
946	else
947	{
948	PUT_NUM_32(off3);
949	PUT_SZ(" wrt rip (");
950	off2 += pDis->uInstrAddr + pDis->cbInstr;
951	PUT_NUM_64(off2);
952	if (pfnGetSymbol)
953	PUT_SYMBOL((pDis->x86.fPrefix & DISPREFIX_SEG)
954	? DIS_FMT_SEL_FROM_REG(pDis->x86.idxSegPrefix)
955	: DIS_FMT_SEL_FROM_REG(DISSELREG_DS),
956	pDis->x86.uAddrMode == DISCPUMODE_64BIT
957	? (uint64_t)off2
958	: pDis->x86.uAddrMode == DISCPUMODE_32BIT
959	? (uint32_t)off2
960	: (uint16_t)off2,
961	" = ",
962	')');
963	else
964	PUT_C(')');
965	}
966	}
967
968	if (DISUSE_IS_EFFECTIVE_ADDR(fUse))
969	{
970	if (pfnGetSymbol && !fBase && !(fUse & (DISUSE_INDEX \| DISUSE_RIPDISPLACEMENT32)) && off2 != 0)
971	PUT_SYMBOL((pDis->x86.fPrefix & DISPREFIX_SEG)
972	? DIS_FMT_SEL_FROM_REG(pDis->x86.idxSegPrefix)
973	: DIS_FMT_SEL_FROM_REG(DISSELREG_DS),
974	pDis->x86.uAddrMode == DISCPUMODE_64BIT
975	? (uint64_t)off2
976	: pDis->x86.uAddrMode == DISCPUMODE_32BIT
977	? (uint32_t)off2
978	: (uint16_t)off2,
979	" (=",
980	')');
981	PUT_C(']');
982	}
983	break;
984	}
985
986	case 'F': /* Eflags register (0 - popf/pushf only, avoided in adjustments above). */
987	AssertFailed();
988	break;
989
990	case 'I': /* Immediate data (ParseImmByte, ParseImmByteSX, ParseImmV, ParseImmUshort, ParseImmZ). */
991	Assert(pszFmt == 'b' \|\| pszFmt == 'v' \|\| pszFmt == 'w' \|\| pszFmt == 'z'); pszFmt++;
992	switch (pParam->fUse & ( DISUSE_IMMEDIATE8 \| DISUSE_IMMEDIATE16 \| DISUSE_IMMEDIATE32 \| DISUSE_IMMEDIATE64
993	\| DISUSE_IMMEDIATE16_SX8 \| DISUSE_IMMEDIATE32_SX8 \| DISUSE_IMMEDIATE64_SX8))
994	{
995	case DISUSE_IMMEDIATE8:
996	if ( (fFlags & DIS_FMT_FLAGS_STRICT)
997	&& ( (pOp->fParam1 >= OP_PARM_REG_GEN8_START && pOp->fParam1 <= OP_PARM_REG_GEN8_END)
998	\|\| (pOp->fParam2 >= OP_PARM_REG_GEN8_START && pOp->fParam2 <= OP_PARM_REG_GEN8_END))
999	)
1000	PUT_SZ("strict byte ");
1001	PUT_NUM_8(pParam->uValue);
1002	break;
1003
1004	case DISUSE_IMMEDIATE16:
1005	if ( pDis->uCpuMode != pDis->x86.uOpMode
1006	\|\| ( (fFlags & DIS_FMT_FLAGS_STRICT)
1007	&& ( (int8_t)pParam->uValue == (int16_t)pParam->uValue
1008	\|\| (pOp->fParam1 >= OP_PARM_REG_GEN16_START && pOp->fParam1 <= OP_PARM_REG_GEN16_END)
1009	\|\| (pOp->fParam2 >= OP_PARM_REG_GEN16_START && pOp->fParam2 <= OP_PARM_REG_GEN16_END))
1010	)
1011	)
1012	{
1013	if (OP_PARM_VSUBTYPE(pParam->x86.fParam) == OP_PARM_b)
1014	PUT_SZ_STRICT("strict byte ", "byte ");
1015	else if ( OP_PARM_VSUBTYPE(pParam->x86.fParam) == OP_PARM_v
1016	\|\| OP_PARM_VSUBTYPE(pParam->x86.fParam) == OP_PARM_z)
1017	PUT_SZ_STRICT("strict word ", "word ");
1018	}
1019	PUT_NUM_16(pParam->uValue);
1020	break;
1021
1022	case DISUSE_IMMEDIATE16_SX8:
1023	if ( !(pDis->x86.fPrefix & DISPREFIX_OPSIZE)
1024	\|\| pDis->pCurInstr->uOpcode != OP_PUSH)
1025	PUT_SZ_STRICT("strict byte ", "byte ");
1026	else
1027	PUT_SZ("word ");
1028	PUT_NUM_16(pParam->uValue);
1029	break;
1030
1031	case DISUSE_IMMEDIATE32:
1032	if ( pDis->x86.uOpMode != (pDis->uCpuMode == DISCPUMODE_16BIT ? DISCPUMODE_16BIT : DISCPUMODE_32BIT) /* not perfect */
1033	\|\| ( (fFlags & DIS_FMT_FLAGS_STRICT)
1034	&& ( (int8_t)pParam->uValue == (int32_t)pParam->uValue
1035	\|\| (pOp->fParam1 >= OP_PARM_REG_GEN32_START && pOp->fParam1 <= OP_PARM_REG_GEN32_END)
1036	\|\| (pOp->fParam2 >= OP_PARM_REG_GEN32_START && pOp->fParam2 <= OP_PARM_REG_GEN32_END))
1037	)
1038	)
1039	{
1040	if (OP_PARM_VSUBTYPE(pParam->x86.fParam) == OP_PARM_b)
1041	PUT_SZ_STRICT("strict byte ", "byte ");
1042	else if ( OP_PARM_VSUBTYPE(pParam->x86.fParam) == OP_PARM_v
1043	\|\| OP_PARM_VSUBTYPE(pParam->x86.fParam) == OP_PARM_z)
1044	PUT_SZ_STRICT("strict dword ", "dword ");
1045	}
1046	PUT_NUM_32(pParam->uValue);
1047	if (pDis->uCpuMode == DISCPUMODE_32BIT)
1048	PUT_SYMBOL(DIS_FMT_SEL_FROM_REG(DISSELREG_CS), pParam->uValue, " (=", ')');
1049	break;
1050
1051	case DISUSE_IMMEDIATE32_SX8:
1052	if ( !(pDis->x86.fPrefix & DISPREFIX_OPSIZE)
1053	\|\| pDis->pCurInstr->uOpcode != OP_PUSH)
1054	PUT_SZ_STRICT("strict byte ", "byte ");
1055	else
1056	PUT_SZ("dword ");
1057	PUT_NUM_32(pParam->uValue);
1058	break;
1059
1060	case DISUSE_IMMEDIATE64_SX8:
1061	if ( !(pDis->x86.fPrefix & DISPREFIX_OPSIZE)
1062	\|\| pDis->pCurInstr->uOpcode != OP_PUSH)
1063	PUT_SZ_STRICT("strict byte ", "byte ");
1064	else
1065	PUT_SZ("qword ");
1066	PUT_NUM_64(pParam->uValue);
1067	break;
1068
1069	case DISUSE_IMMEDIATE64:
1070	PUT_NUM_64(pParam->uValue);
1071	break;
1072
1073	default:
1074	AssertFailed();
1075	break;
1076	}
1077	break;
1078
1079	case 'J': /* Relative jump offset (ParseImmBRel + ParseImmVRel). */
1080	{
1081	int32_t offDisplacement;
1082	Assert(iParam == 1);
1083	bool fPrefix = (fFlags & DIS_FMT_FLAGS_STRICT)
1084	&& pOp->uOpcode != OP_CALL
1085	&& pOp->uOpcode != OP_LOOP
1086	&& pOp->uOpcode != OP_LOOPE
1087	&& pOp->uOpcode != OP_LOOPNE
1088	&& pOp->uOpcode != OP_JECXZ;
1089	if (pOp->uOpcode == OP_CALL)
1090	fFlags &= ~DIS_FMT_FLAGS_RELATIVE_BRANCH;
1091
1092	if (pParam->fUse & DISUSE_IMMEDIATE8_REL)
1093	{
1094	if (fPrefix)
1095	PUT_SZ("short ");
1096	offDisplacement = (int8_t)pParam->uValue;
1097	Assert(*pszFmt == 'b'); pszFmt++;
1098
1099	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1100	PUT_NUM_S8(offDisplacement);
1101	}
1102	else if (pParam->fUse & DISUSE_IMMEDIATE16_REL)
1103	{
1104	if (fPrefix)
1105	PUT_SZ("near ");
1106	offDisplacement = (int16_t)pParam->uValue;
1107	Assert(*pszFmt == 'v'); pszFmt++;
1108
1109	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1110	PUT_NUM_S16(offDisplacement);
1111	}
1112	else
1113	{
1114	if (fPrefix)
1115	PUT_SZ("near ");
1116	offDisplacement = (int32_t)pParam->uValue;
1117	Assert(pParam->fUse & (DISUSE_IMMEDIATE32_REL \| DISUSE_IMMEDIATE64_REL));
1118	Assert(*pszFmt == 'v'); pszFmt++;
1119
1120	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1121	PUT_NUM_S32(offDisplacement);
1122	}
1123	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1124	PUT_SZ(" (");
1125
1126	RTUINTPTR uTrgAddr = pDis->uInstrAddr + pDis->cbInstr + offDisplacement;
1127	if (pDis->uCpuMode == DISCPUMODE_16BIT)
1128	PUT_NUM_16(uTrgAddr);
1129	else if (pDis->uCpuMode == DISCPUMODE_32BIT)
1130	PUT_NUM_32(uTrgAddr);
1131	else
1132	PUT_NUM_64(uTrgAddr);
1133
1134	if (fFlags & DIS_FMT_FLAGS_RELATIVE_BRANCH)
1135	{
1136	PUT_SYMBOL(DIS_FMT_SEL_FROM_REG(DISSELREG_CS), uTrgAddr, " = ", ' ');
1137	PUT_C(')');
1138	}
1139	else
1140	PUT_SYMBOL(DIS_FMT_SEL_FROM_REG(DISSELREG_CS), uTrgAddr, " (", ')');
1141	break;
1142	}
1143
1144	case 'A': /* Direct (jump/call) address (ParseImmAddr). */
1145	{
1146	Assert(*pszFmt == 'p'); pszFmt++;
1147	PUT_FAR();
1148	PUT_SIZE_OVERRIDE();
1149	PUT_SEGMENT_OVERRIDE();
1150	off = 0;
1151	int rc = VERR_SYMBOL_NOT_FOUND;
1152	switch (pParam->fUse & (DISUSE_IMMEDIATE_ADDR_16_16 \| DISUSE_IMMEDIATE_ADDR_16_32 \| DISUSE_DISPLACEMENT64 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT16))
1153	{
1154	case DISUSE_IMMEDIATE_ADDR_16_16:
1155	PUT_NUM_16(pParam->uValue >> 16);
1156	PUT_C(':');
1157	PUT_NUM_16(pParam->uValue);
1158	if (pfnGetSymbol)
1159	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_VALUE(pParam->uValue >> 16), (uint16_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1160	break;
1161	case DISUSE_IMMEDIATE_ADDR_16_32:
1162	PUT_NUM_16(pParam->uValue >> 32);
1163	PUT_C(':');
1164	PUT_NUM_32(pParam->uValue);
1165	if (pfnGetSymbol)
1166	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_VALUE(pParam->uValue >> 16), (uint32_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1167	break;
1168	case DISUSE_DISPLACEMENT16:
1169	PUT_NUM_16(pParam->uValue);
1170	if (pfnGetSymbol)
1171	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), (uint16_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1172	break;
1173	case DISUSE_DISPLACEMENT32:
1174	PUT_NUM_32(pParam->uValue);
1175	if (pfnGetSymbol)
1176	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), (uint32_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1177	break;
1178	case DISUSE_DISPLACEMENT64:
1179	PUT_NUM_64(pParam->uValue);
1180	if (pfnGetSymbol)
1181	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), (uint64_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1182	break;
1183	default:
1184	AssertFailed();
1185	break;
1186	}
1187
1188	PUT_SYMBOL_TWO(rc, " [", ']');
1189	break;
1190	}
1191
1192	case 'O': /* No ModRM byte (ParseImmAddr). */
1193	{
1194	Assert(pszFmt == 'b' \|\| pszFmt == 'v'); pszFmt++;
1195	PUT_FAR();
1196	PUT_SIZE_OVERRIDE();
1197	PUT_C('[');
1198	PUT_SEGMENT_OVERRIDE();
1199	off = 0;
1200	int rc = VERR_SYMBOL_NOT_FOUND;
1201	switch (pParam->fUse & (DISUSE_IMMEDIATE_ADDR_16_16 \| DISUSE_IMMEDIATE_ADDR_16_32 \| DISUSE_DISPLACEMENT64 \| DISUSE_DISPLACEMENT32 \| DISUSE_DISPLACEMENT16))
1202	{
1203	case DISUSE_IMMEDIATE_ADDR_16_16:
1204	PUT_NUM_16(pParam->uValue >> 16);
1205	PUT_C(':');
1206	PUT_NUM_16(pParam->uValue);
1207	if (pfnGetSymbol)
1208	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_VALUE(pParam->uValue >> 16), (uint16_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1209	break;
1210	case DISUSE_IMMEDIATE_ADDR_16_32:
1211	PUT_NUM_16(pParam->uValue >> 32);
1212	PUT_C(':');
1213	PUT_NUM_32(pParam->uValue);
1214	if (pfnGetSymbol)
1215	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_VALUE(pParam->uValue >> 16), (uint32_t)pParam->uValue, szSymbol, sizeof(szSymbol), &off, pvUser);
1216	break;
1217	case DISUSE_DISPLACEMENT16:
1218	PUT_NUM_16(pParam->x86.uDisp.i16);
1219	if (pfnGetSymbol)
1220	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), pParam->x86.uDisp.u16, szSymbol, sizeof(szSymbol), &off, pvUser);
1221	break;
1222	case DISUSE_DISPLACEMENT32:
1223	PUT_NUM_32(pParam->x86.uDisp.i32);
1224	if (pfnGetSymbol)
1225	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), pParam->x86.uDisp.u32, szSymbol, sizeof(szSymbol), &off, pvUser);
1226	break;
1227	case DISUSE_DISPLACEMENT64:
1228	PUT_NUM_64(pParam->x86.uDisp.i64);
1229	if (pfnGetSymbol)
1230	rc = pfnGetSymbol(pDis, DIS_FMT_SEL_FROM_REG(DISSELREG_CS), pParam->x86.uDisp.u64, szSymbol, sizeof(szSymbol), &off, pvUser);
1231	break;
1232	default:
1233	AssertFailed();
1234	break;
1235	}
1236	PUT_C(']');
1237
1238	PUT_SYMBOL_TWO(rc, " (", ')');
1239	break;
1240	}
1241
1242	case 'X': /* DS:SI (ParseXb, ParseXv). */
1243	case 'Y': /* ES:DI (ParseYb, ParseYv). */
1244	{
1245	Assert(pszFmt == 'b' \|\| pszFmt == 'v'); pszFmt++;
1246	PUT_FAR();
1247	PUT_SIZE_OVERRIDE();
1248	PUT_C('[');
1249	if (pParam->fUse & DISUSE_POINTER_DS_BASED)
1250	PUT_SZ("ds:");
1251	else
1252	PUT_SZ("es:");
1253
1254	size_t cchReg;
1255	const char *pszReg = disasmFormatYasmBaseReg(pDis, pParam, &cchReg);
1256	PUT_STR(pszReg, cchReg);
1257	PUT_C(']');
1258	break;
1259	}
1260
1261	case 'e': /* Register based on operand size (e.g. %eAX, %eAH) (ParseFixedReg). */
1262	{
1263	Assert(RT_C_IS_ALPHA(pszFmt[0]) && RT_C_IS_ALPHA(pszFmt[1]) && !RT_C_IS_ALPHA(pszFmt[2]));
1264	pszFmt += 2;
1265	size_t cchReg;
1266	const char *pszReg = disasmFormatYasmBaseReg(pDis, pParam, &cchReg);
1267	PUT_STR(pszReg, cchReg);
1268	break;
1269	}
1270
1271	default:
1272	AssertMsgFailed(("%c%s!\n", ch, pszFmt));
1273	break;
1274	}
1275	AssertMsg(pszFmt == ',' \|\| pszFmt == '\0', ("%c%s\n", ch, pszFmt));
1276	}
1277	else
1278	{
1279	PUT_C(ch);
1280	if (ch == ',')
1281	{
1282	Assert(*pszFmt != ' ');
1283	PUT_C(' ');
1284	switch (++iParam)
1285	{
1286	case 2: pParam = &pDis->Param2; break;
1287	case 3: pParam = &pDis->Param3; break;
1288	case 4: pParam = &pDis->Param4; break;
1289	default: pParam = NULL; break;
1290	}
1291	}
1292	}
1293	} /* while more to format */
1294	}
1295
1296	/*
1297	* Any additional output to the right of the instruction?
1298	*/
1299	if (fFlags & (DIS_FMT_FLAGS_BYTES_RIGHT \| DIS_FMT_FLAGS_ADDR_RIGHT))
1300	{
1301	/* some up front padding. */
1302	size_t cchPadding = cchOutput - offInstruction;
1303	cchPadding = cchPadding + 1 >= 42 ? 1 : 42 - cchPadding;
1304	PUT_STR(g_szSpaces, cchPadding);
1305
1306	/* comment? */
1307	if (fFlags & (DIS_FMT_FLAGS_BYTES_RIGHT \| DIS_FMT_FLAGS_ADDR_RIGHT))
1308	PUT_SZ(";");
1309
1310	/*
1311	* The address?
1312	*/
1313	if (fFlags & DIS_FMT_FLAGS_ADDR_RIGHT)
1314	{
1315	PUT_C(' ');
1316	#if HC_ARCH_BITS == 64 \|\| GC_ARCH_BITS == 64
1317	if (pDis->uInstrAddr >= _4G)
1318	PUT_NUM(9, "%08x`", (uint32_t)(pDis->uInstrAddr >> 32));
1319	#endif
1320	PUT_NUM(8, "%08x", (uint32_t)pDis->uInstrAddr);
1321	}
1322
1323	/*
1324	* Opcode bytes?
1325	*/
1326	if (fFlags & DIS_FMT_FLAGS_BYTES_RIGHT)
1327	{
1328	PUT_C(' ');
1329	size_t cchTmp = disFormatBytes(pDis, pszDst, cchDst, fFlags);
1330	cchOutput += cchTmp;
1331	if (cchTmp >= cchDst)
1332	cchTmp = cchDst - (cchDst != 0);
1333	cchDst -= cchTmp;
1334	pszDst += cchTmp;
1335	}
1336	}
1337
1338	/*
1339	* Terminate it - on overflow we'll have reserved one byte for this.
1340	*/
1341	if (cchDst > 0)
1342	*pszDst = '\0';
1343	else
1344	Assert(!cchBuf);
1345
1346	/* clean up macros */
1347	#undef PUT_PSZ
1348	#undef PUT_SZ
1349	#undef PUT_STR
1350	#undef PUT_C
1351	return cchOutput;
1352	}
1353
1354
1355	/**
1356	* Formats the current instruction in Yasm (/ Nasm) style.
1357	*
1358	* This is a simplified version of DISFormatYasmEx() provided for your convenience.
1359	*
1360	*
1361	* @returns The number of output characters. If this is >= cchBuf, then the content
1362	* of pszBuf will be truncated.
1363	* @param pDis Pointer to the disassembler state.
1364	* @param pszBuf The output buffer.
1365	* @param cchBuf The size of the output buffer.
1366	*/
1367	DISDECL(size_t) DISFormatYasm(PCDISSTATE pDis, char *pszBuf, size_t cchBuf)
1368	{
1369	return DISFormatYasmEx(pDis, pszBuf, cchBuf, 0 /* fFlags /, NULL / pfnGetSymbol /, NULL / pvUser */);
1370	}
1371
1372
1373	/**
1374	* Checks if the encoding of the given disassembled instruction is something we
1375	* can never get YASM to produce.
1376	*
1377	* @returns true if it's odd, false if it isn't.
1378	* @param pDis The disassembler output. The byte fetcher callback will
1379	* be used if present as we might need to fetch opcode
1380	* bytes.
1381	*/
1382	DISDECL(bool) DISFormatYasmIsOddEncoding(PDISSTATE pDis)
1383	{
1384	/*
1385	* Mod rm + SIB: Check for duplicate EBP encodings that yasm won't use for very good reasons.
1386	*/
1387	if ( pDis->x86.uAddrMode != DISCPUMODE_16BIT /// @todo correct?
1388	&& pDis->x86.ModRM.Bits.Rm == 4
1389	&& pDis->x86.ModRM.Bits.Mod != 3)
1390	{
1391	/* No scaled index SIB (index=4), except for ESP. */
1392	if ( pDis->x86.SIB.Bits.Index == 4
1393	&& pDis->x86.SIB.Bits.Base != 4)
1394	return true;
1395
1396	/* EBP + displacement */
1397	if ( pDis->x86.ModRM.Bits.Mod != 0
1398	&& pDis->x86.SIB.Bits.Base == 5
1399	&& pDis->x86.SIB.Bits.Scale == 0)
1400	return true;
1401	}
1402
1403	/*
1404	* Seems to be an instruction alias here, but I cannot find any docs on it... hrmpf!
1405	*/
1406	if ( pDis->pCurInstr->uOpcode == OP_SHL
1407	&& pDis->x86.ModRM.Bits.Reg == 6)
1408	return true;
1409
1410	/*
1411	* Check for multiple prefixes of the same kind.
1412	*/
1413	uint8_t off1stSeg = UINT8_MAX;
1414	uint8_t offOpSize = UINT8_MAX;
1415	uint8_t offAddrSize = UINT8_MAX;
1416	uint32_t fPrefixes = 0;
1417	for (uint32_t offOpcode = 0; offOpcode < RT_ELEMENTS(pDis->Instr.ab); offOpcode++)
1418	{
1419	uint32_t f;
1420	switch (pDis->Instr.ab[offOpcode])
1421	{
1422	case 0xf0:
1423	f = DISPREFIX_LOCK;
1424	break;
1425
1426	case 0xf2:
1427	case 0xf3:
1428	f = DISPREFIX_REP; /* yes, both */
1429	break;
1430
1431	case 0x2e:
1432	case 0x3e:
1433	case 0x26:
1434	case 0x36:
1435	case 0x64:
1436	case 0x65:
1437	if (off1stSeg == UINT8_MAX)
1438	off1stSeg = offOpcode;
1439	f = DISPREFIX_SEG;
1440	break;
1441
1442	case 0x66:
1443	if (offOpSize == UINT8_MAX)
1444	offOpSize = offOpcode;
1445	f = DISPREFIX_OPSIZE;
1446	break;
1447
1448	case 0x67:
1449	if (offAddrSize == UINT8_MAX)
1450	offAddrSize = offOpcode;
1451	f = DISPREFIX_ADDRSIZE;
1452	break;
1453
1454	case 0x40: case 0x41: case 0x42: case 0x43: case 0x44: case 0x45: case 0x46: case 0x47:
1455	case 0x48: case 0x49: case 0x4a: case 0x4b: case 0x4c: case 0x4d: case 0x4e: case 0x4f:
1456	f = pDis->uCpuMode == DISCPUMODE_64BIT ? DISPREFIX_REX : 0;
1457	break;
1458
1459	default:
1460	f = 0;
1461	break;
1462	}
1463	if (!f)
1464	break; /* done */
1465	if (fPrefixes & f)
1466	return true;
1467	fPrefixes \|= f;
1468	}
1469
1470	/* segment overrides are fun */
1471	if (fPrefixes & DISPREFIX_SEG)
1472	{
1473	/* no effective address which it may apply to. */
1474	Assert((pDis->x86.fPrefix & DISPREFIX_SEG) \|\| pDis->uCpuMode == DISCPUMODE_64BIT);
1475	if ( !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param1.fUse)
1476	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param2.fUse)
1477	&& !DISUSE_IS_EFFECTIVE_ADDR(pDis->Param3.fUse))
1478	return true;
1479
1480	/* Yasm puts the segment prefixes before the operand prefix with no
1481	way of overriding it. */
1482	if (offOpSize < off1stSeg)
1483	return true;
1484	}
1485
1486	/* fixed register + addr override doesn't go down all that well. */
1487	if (fPrefixes & DISPREFIX_ADDRSIZE)
1488	{
1489	Assert(pDis->x86.fPrefix & DISPREFIX_ADDRSIZE);
1490	if ( pDis->pCurInstr->fParam3 == OP_PARM_NONE
1491	&& pDis->pCurInstr->fParam2 == OP_PARM_NONE
1492	&& ( pDis->pCurInstr->fParam1 >= OP_PARM_REG_GEN32_START
1493	&& pDis->pCurInstr->fParam1 <= OP_PARM_REG_GEN32_END))
1494	return true;
1495	}
1496
1497	/* Almost all prefixes are bad for jumps. */
1498	if (fPrefixes)
1499	{
1500	switch (pDis->pCurInstr->uOpcode)
1501	{
1502	/* nop w/ prefix(es). */
1503	case OP_NOP:
1504	return true;
1505
1506	case OP_JMP:
1507	if ( pDis->pCurInstr->fParam1 != OP_PARM_Jb
1508	&& pDis->pCurInstr->fParam1 != OP_PARM_Jv)
1509	break;
1510	RT_FALL_THRU();
1511	case OP_JO:
1512	case OP_JNO:
1513	case OP_JC:
1514	case OP_JNC:
1515	case OP_JE:
1516	case OP_JNE:
1517	case OP_JBE:
1518	case OP_JNBE:
1519	case OP_JS:
1520	case OP_JNS:
1521	case OP_JP:
1522	case OP_JNP:
1523	case OP_JL:
1524	case OP_JNL:
1525	case OP_JLE:
1526	case OP_JNLE:
1527	/** @todo branch hinting 0x2e/0x3e... */
1528	return true;
1529	}
1530
1531	}
1532
1533	/* All but the segment prefix is bad news for push/pop. */
1534	if (fPrefixes & ~DISPREFIX_SEG)
1535	{
1536	switch (pDis->pCurInstr->uOpcode)
1537	{
1538	case OP_POP:
1539	case OP_PUSH:
1540	if ( pDis->pCurInstr->fParam1 >= OP_PARM_REG_SEG_START
1541	&& pDis->pCurInstr->fParam1 <= OP_PARM_REG_SEG_END)
1542	return true;
1543	if ( (fPrefixes & ~DISPREFIX_OPSIZE)
1544	&& pDis->pCurInstr->fParam1 >= OP_PARM_REG_GEN32_START
1545	&& pDis->pCurInstr->fParam1 <= OP_PARM_REG_GEN32_END)
1546	return true;
1547	break;
1548
1549	case OP_POPA:
1550	case OP_POPF:
1551	case OP_PUSHA:
1552	case OP_PUSHF:
1553	if (fPrefixes & ~DISPREFIX_OPSIZE)
1554	return true;
1555	break;
1556	}
1557	}
1558
1559	/* Implicit 8-bit register instructions doesn't mix with operand size. */
1560	if ( (fPrefixes & DISPREFIX_OPSIZE)
1561	&& ( ( pDis->pCurInstr->fParam1 == OP_PARM_Gb /* r8 */
1562	&& pDis->pCurInstr->fParam2 == OP_PARM_Eb /* r8/mem8 */)
1563	\|\| ( pDis->pCurInstr->fParam2 == OP_PARM_Gb /* r8 */
1564	&& pDis->pCurInstr->fParam1 == OP_PARM_Eb /* r8/mem8 */))
1565	)
1566	{
1567	switch (pDis->pCurInstr->uOpcode)
1568	{
1569	case OP_ADD:
1570	case OP_OR:
1571	case OP_ADC:
1572	case OP_SBB:
1573	case OP_AND:
1574	case OP_SUB:
1575	case OP_XOR:
1576	case OP_CMP:
1577	return true;
1578	default:
1579	break;
1580	}
1581	}
1582
1583	/* Instructions taking no address or operand which thus may be annoyingly
1584	difficult to format for yasm. */
1585	if (fPrefixes)
1586	{
1587	switch (pDis->pCurInstr->uOpcode)
1588	{
1589	case OP_STI:
1590	case OP_STC:
1591	case OP_CLI:
1592	case OP_CLD:
1593	case OP_CLC:
1594	case OP_INT:
1595	case OP_INT3:
1596	case OP_INTO:
1597	case OP_HLT:
1598	/** @todo Many more to can be added here. */
1599	return true;
1600	default:
1601	break;
1602	}
1603	}
1604
1605	/* FPU and other instructions that ignores operand size override. */
1606	if (fPrefixes & DISPREFIX_OPSIZE)
1607	{
1608	switch (pDis->pCurInstr->uOpcode)
1609	{
1610	/* FPU: */
1611	case OP_FIADD:
1612	case OP_FIMUL:
1613	case OP_FISUB:
1614	case OP_FISUBR:
1615	case OP_FIDIV:
1616	case OP_FIDIVR:
1617	/** @todo there are many more. */
1618	return true;
1619
1620	case OP_MOV:
1621	/** @todo could be that we're not disassembling these correctly. */
1622	if (pDis->pCurInstr->fParam1 == OP_PARM_Sw)
1623	return true;
1624	/** @todo what about the other way? */
1625	break;
1626
1627	default:
1628	break;
1629	}
1630	}
1631
1632
1633	/*
1634	* Check for the version of xyz reg,reg instruction that the assembler doesn't use.
1635	*
1636	* For example:
1637	* expected: 1aee sbb ch, dh ; SBB r8, r/m8
1638	* yasm: 18F5 sbb ch, dh ; SBB r/m8, r8
1639	*/
1640	if (pDis->x86.ModRM.Bits.Mod == 3 /* reg,reg */)
1641	{
1642	switch (pDis->pCurInstr->uOpcode)
1643	{
1644	case OP_ADD:
1645	case OP_OR:
1646	case OP_ADC:
1647	case OP_SBB:
1648	case OP_AND:
1649	case OP_SUB:
1650	case OP_XOR:
1651	case OP_CMP:
1652	if ( ( pDis->pCurInstr->fParam1 == OP_PARM_Gb /* r8 */
1653	&& pDis->pCurInstr->fParam2 == OP_PARM_Eb /* r8/mem8 */)
1654	\|\| ( pDis->pCurInstr->fParam1 == OP_PARM_Gv /* rX */
1655	&& pDis->pCurInstr->fParam2 == OP_PARM_Ev /* rX/memX */))
1656	return true;
1657
1658	/* 82 (see table A-6). */
1659	if (pDis->x86.bOpCode == 0x82)
1660	return true;
1661	break;
1662
1663	/* ff /0, fe /0, ff /1, fe /0 */
1664	case OP_DEC:
1665	case OP_INC:
1666	return true;
1667
1668	case OP_POP:
1669	case OP_PUSH:
1670	Assert(pDis->x86.bOpCode == 0x8f);
1671	return true;
1672
1673	case OP_MOV:
1674	if ( pDis->x86.bOpCode == 0x8a
1675	\|\| pDis->x86.bOpCode == 0x8b)
1676	return true;
1677	break;
1678
1679	default:
1680	break;
1681	}
1682	}
1683
1684	/* shl eax,1 will be assembled to the form without the immediate byte. */
1685	if ( pDis->pCurInstr->fParam2 == OP_PARM_Ib
1686	&& (uint8_t)pDis->Param2.uValue == 1)
1687	{
1688	switch (pDis->pCurInstr->uOpcode)
1689	{
1690	case OP_SHL:
1691	case OP_SHR:
1692	case OP_SAR:
1693	case OP_RCL:
1694	case OP_RCR:
1695	case OP_ROL:
1696	case OP_ROR:
1697	return true;
1698	}
1699	}
1700
1701	/* And some more - see table A-6. */
1702	if (pDis->x86.bOpCode == 0x82)
1703	{
1704	switch (pDis->pCurInstr->uOpcode)
1705	{
1706	case OP_ADD:
1707	case OP_OR:
1708	case OP_ADC:
1709	case OP_SBB:
1710	case OP_AND:
1711	case OP_SUB:
1712	case OP_XOR:
1713	case OP_CMP:
1714	return true;
1715	break;
1716	}
1717	}
1718
1719
1720	/* check for REX.X = 1 without SIB. */
1721
1722	/* Yasm encodes setnbe al with /2 instead of /0 like the AMD manual
1723	says (intel doesn't appear to care). */
1724	switch (pDis->pCurInstr->uOpcode)
1725	{
1726	case OP_SETO:
1727	case OP_SETNO:
1728	case OP_SETC:
1729	case OP_SETNC:
1730	case OP_SETE:
1731	case OP_SETNE:
1732	case OP_SETBE:
1733	case OP_SETNBE:
1734	case OP_SETS:
1735	case OP_SETNS:
1736	case OP_SETP:
1737	case OP_SETNP:
1738	case OP_SETL:
1739	case OP_SETNL:
1740	case OP_SETLE:
1741	case OP_SETNLE:
1742	AssertMsg(pDis->x86.bOpCode >= 0x90 && pDis->x86.bOpCode <= 0x9f, ("%#x\n", pDis->x86.bOpCode));
1743	if (pDis->x86.ModRM.Bits.Reg != 2)
1744	return true;
1745	break;
1746	}
1747
1748	/*
1749	* The MOVZX reg32,mem16 instruction without an operand size prefix
1750	* doesn't quite make sense...
1751	*/
1752	if ( pDis->pCurInstr->uOpcode == OP_MOVZX
1753	&& pDis->x86.bOpCode == 0xB7
1754	&& (pDis->uCpuMode == DISCPUMODE_16BIT) != !!(fPrefixes & DISPREFIX_OPSIZE))
1755	return true;
1756
1757	/*
1758	* YASM doesn't do ICEBP/INT1/INT01, unlike NASM.
1759	*/
1760	if (pDis->x86.bOpCode == 0xF1)
1761	return true;
1762
1763	return false;
1764	}
1765

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

source: vbox/trunk/src/VBox/Disassembler/DisasmFormatYasm.cpp@ 101540

Download in other formats: