IEMAllAImpl.asm@ 45276

Last change on this file since 45276 was 42720, checked in by vboxsync, 12 years ago
IEM: Fixed BSF and BSR bug where they would set the destination register when the source was zero.
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 68.6 KB

Line
1	; $Id: IEMAllAImpl.asm 42720 2012-08-09 17:25:25Z vboxsync $
2	;; @file
3	; IEM - Instruction Implementation in Assembly.
4	;
5
6	; Copyright (C) 2011-2012 Oracle Corporation
7	;
8	; This file is part of VirtualBox Open Source Edition (OSE), as
9	; available from http://www.virtualbox.org. This file is free software;
10	; you can redistribute it and/or modify it under the terms of the GNU
11	; General Public License (GPL) as published by the Free Software
12	; Foundation, in version 2 as it comes in the "COPYING" file of the
13	; VirtualBox OSE distribution. VirtualBox OSE is distributed in the
14	; hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
15	;
16
17
18	;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
19	; Header Files ;
20	;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
21	%include "VBox/asmdefs.mac"
22	%include "VBox/err.mac"
23	%include "iprt/x86.mac"
24
25
26	;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
27	; Defined Constants And Macros ;
28	;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
29
30	;;
31	; RET XX / RET wrapper for fastcall.
32	;
33	%macro RET_FASTCALL 1
34	%ifdef RT_ARCH_X86
35	%ifdef RT_OS_WINDOWS
36	ret %1
37	%else
38	ret
39	%endif
40	%else
41	ret
42	%endif
43	%endmacro
44
45	;;
46	; NAME for fastcall functions.
47	;
48	;; @todo 'global @fastcall@12' is still broken in yasm and requires dollar
49	; escaping (or whatever the dollar is good for here). Thus the ugly
50	; prefix argument.
51	;
52	%define NAME_FASTCALL(a_Name, a_cbArgs, a_Prefix) NAME(a_Name)
53	%ifdef RT_ARCH_X86
54	%ifdef RT_OS_WINDOWS
55	%undef NAME_FASTCALL
56	%define NAME_FASTCALL(a_Name, a_cbArgs, a_Prefix) a_Prefix %+ a_Name %+ @ %+ a_cbArgs
57	%endif
58	%endif
59
60	;;
61	; BEGINPROC for fastcall functions.
62	;
63	; @param 1 The function name (C).
64	; @param 2 The argument size on x86.
65	;
66	%macro BEGINPROC_FASTCALL 2
67	%ifdef ASM_FORMAT_PE
68	export %1=NAME_FASTCALL(%1,%2,$@)
69	%endif
70	%ifdef __NASM__
71	%ifdef ASM_FORMAT_OMF
72	export NAME(%1) NAME_FASTCALL(%1,%2,$@)
73	%endif
74	%endif
75	%ifndef ASM_FORMAT_BIN
76	global NAME_FASTCALL(%1,%2,$@)
77	%endif
78	NAME_FASTCALL(%1,%2,@):
79	%endmacro
80
81
82	;
83	; We employ some macro assembly here to hid the calling convention differences.
84	;
85	%ifdef RT_ARCH_AMD64
86	%macro PROLOGUE_1_ARGS 0
87	%endmacro
88	%macro EPILOGUE_1_ARGS 0
89	ret
90	%endmacro
91	%macro EPILOGUE_1_ARGS_EX 0
92	ret
93	%endmacro
94
95	%macro PROLOGUE_2_ARGS 0
96	%endmacro
97	%macro EPILOGUE_2_ARGS 0
98	ret
99	%endmacro
100	%macro EPILOGUE_2_ARGS_EX 1
101	ret
102	%endmacro
103
104	%macro PROLOGUE_3_ARGS 0
105	%endmacro
106	%macro EPILOGUE_3_ARGS 0
107	ret
108	%endmacro
109	%macro EPILOGUE_3_ARGS_EX 1
110	ret
111	%endmacro
112
113	%macro PROLOGUE_4_ARGS 0
114	%endmacro
115	%macro EPILOGUE_4_ARGS 0
116	ret
117	%endmacro
118	%macro EPILOGUE_4_ARGS_EX 1
119	ret
120	%endmacro
121
122	%ifdef ASM_CALL64_GCC
123	%define A0 rdi
124	%define A0_32 edi
125	%define A0_16 di
126	%define A0_8 dil
127
128	%define A1 rsi
129	%define A1_32 esi
130	%define A1_16 si
131	%define A1_8 sil
132
133	%define A2 rdx
134	%define A2_32 edx
135	%define A2_16 dx
136	%define A2_8 dl
137
138	%define A3 rcx
139	%define A3_32 ecx
140	%define A3_16 cx
141	%endif
142
143	%ifdef ASM_CALL64_MSC
144	%define A0 rcx
145	%define A0_32 ecx
146	%define A0_16 cx
147	%define A0_8 cl
148
149	%define A1 rdx
150	%define A1_32 edx
151	%define A1_16 dx
152	%define A1_8 dl
153
154	%define A2 r8
155	%define A2_32 r8d
156	%define A2_16 r8w
157	%define A2_8 r8b
158
159	%define A3 r9
160	%define A3_32 r9d
161	%define A3_16 r9w
162	%endif
163
164	%define T0 rax
165	%define T0_32 eax
166	%define T0_16 ax
167	%define T0_8 al
168
169	%define T1 r11
170	%define T1_32 r11d
171	%define T1_16 r11w
172	%define T1_8 r11b
173
174	%else
175	; x86
176	%macro PROLOGUE_1_ARGS 0
177	push edi
178	%endmacro
179	%macro EPILOGUE_1_ARGS 0
180	pop edi
181	ret 0
182	%endmacro
183	%macro EPILOGUE_1_ARGS_EX 1
184	pop edi
185	ret %1
186	%endmacro
187
188	%macro PROLOGUE_2_ARGS 0
189	push edi
190	%endmacro
191	%macro EPILOGUE_2_ARGS 0
192	pop edi
193	ret 0
194	%endmacro
195	%macro EPILOGUE_2_ARGS_EX 1
196	pop edi
197	ret %1
198	%endmacro
199
200	%macro PROLOGUE_3_ARGS 0
201	push ebx
202	mov ebx, [esp + 4 + 4]
203	push edi
204	%endmacro
205	%macro EPILOGUE_3_ARGS_EX 1
206	%if (%1) < 4
207	%error "With three args, at least 4 bytes must be remove from the stack upon return (32-bit)."
208	%endif
209	pop edi
210	pop ebx
211	ret %1
212	%endmacro
213	%macro EPILOGUE_3_ARGS 0
214	EPILOGUE_3_ARGS_EX 4
215	%endmacro
216
217	%macro PROLOGUE_4_ARGS 0
218	push ebx
219	push edi
220	push esi
221	mov ebx, [esp + 12 + 4 + 0]
222	mov esi, [esp + 12 + 4 + 4]
223	%endmacro
224	%macro EPILOGUE_4_ARGS_EX 1
225	%if (%1) < 8
226	%error "With four args, at least 8 bytes must be remove from the stack upon return (32-bit)."
227	%endif
228	pop esi
229	pop edi
230	pop ebx
231	ret %1
232	%endmacro
233	%macro EPILOGUE_4_ARGS 0
234	EPILOGUE_4_ARGS_EX 8
235	%endmacro
236
237	%define A0 ecx
238	%define A0_32 ecx
239	%define A0_16 cx
240	%define A0_8 cl
241
242	%define A1 edx
243	%define A1_32 edx
244	%define A1_16 dx
245	%define A1_8 dl
246
247	%define A2 ebx
248	%define A2_32 ebx
249	%define A2_16 bx
250	%define A2_8 bl
251
252	%define A3 esi
253	%define A3_32 esi
254	%define A3_16 si
255
256	%define T0 eax
257	%define T0_32 eax
258	%define T0_16 ax
259	%define T0_8 al
260
261	%define T1 edi
262	%define T1_32 edi
263	%define T1_16 di
264	%endif
265
266
267	;;
268	; Load the relevant flags from [%1] if there are undefined flags (%3).
269	;
270	; @remarks Clobbers T0, stack. Changes EFLAGS.
271	; @param A2 The register pointing to the flags.
272	; @param 1 The parameter (A0..A3) pointing to the eflags.
273	; @param 2 The set of modified flags.
274	; @param 3 The set of undefined flags.
275	;
276	%macro IEM_MAYBE_LOAD_FLAGS 3
277	;%if (%3) != 0
278	pushf ; store current flags
279	mov T0_32, [%1] ; load the guest flags
280	and dword [xSP], ~(%2 \| %3) ; mask out the modified and undefined flags
281	and T0_32, (%2 \| %3) ; select the modified and undefined flags.
282	or [xSP], T0 ; merge guest flags with host flags.
283	popf ; load the mixed flags.
284	;%endif
285	%endmacro
286
287	;;
288	; Update the flag.
289	;
290	; @remarks Clobbers T0, T1, stack.
291	; @param 1 The register pointing to the EFLAGS.
292	; @param 2 The mask of modified flags to save.
293	; @param 3 The mask of undefined flags to (maybe) save.
294	;
295	%macro IEM_SAVE_FLAGS 3
296	%if (%2 \| %3) != 0
297	pushf
298	pop T1
299	mov T0_32, [%1] ; flags
300	and T0_32, ~(%2 \| %3) ; clear the modified & undefined flags.
301	and T1_32, (%2 \| %3) ; select the modified and undefined flags.
302	or T0_32, T1_32 ; combine the flags.
303	mov [%1], T0_32 ; save the flags.
304	%endif
305	%endmacro
306
307
308	;;
309	; Macro for implementing a binary operator.
310	;
311	; This will generate code for the 8, 16, 32 and 64 bit accesses with locked
312	; variants, except on 32-bit system where the 64-bit accesses requires hand
313	; coding.
314	;
315	; All the functions takes a pointer to the destination memory operand in A0,
316	; the source register operand in A1 and a pointer to eflags in A2.
317	;
318	; @param 1 The instruction mnemonic.
319	; @param 2 Non-zero if there should be a locked version.
320	; @param 3 The modified flags.
321	; @param 4 The undefined flags.
322	;
323	%macro IEMIMPL_BIN_OP 4
324	BEGINCODE
325	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u8, 12
326	PROLOGUE_3_ARGS
327	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
328	%1 byte [A0], A1_8
329	IEM_SAVE_FLAGS A2, %3, %4
330	EPILOGUE_3_ARGS
331	ENDPROC iemAImpl_ %+ %1 %+ _u8
332
333	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 12
334	PROLOGUE_3_ARGS
335	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
336	%1 word [A0], A1_16
337	IEM_SAVE_FLAGS A2, %3, %4
338	EPILOGUE_3_ARGS
339	ENDPROC iemAImpl_ %+ %1 %+ _u16
340
341	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 12
342	PROLOGUE_3_ARGS
343	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
344	%1 dword [A0], A1_32
345	IEM_SAVE_FLAGS A2, %3, %4
346	EPILOGUE_3_ARGS
347	ENDPROC iemAImpl_ %+ %1 %+ _u32
348
349	%ifdef RT_ARCH_AMD64
350	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 16
351	PROLOGUE_3_ARGS
352	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
353	%1 qword [A0], A1
354	IEM_SAVE_FLAGS A2, %3, %4
355	EPILOGUE_3_ARGS_EX 8
356	ENDPROC iemAImpl_ %+ %1 %+ _u64
357	%else ; stub it for now - later, replace with hand coded stuff.
358	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 16
359	int3
360	ret
361	ENDPROC iemAImpl_ %+ %1 %+ _u64
362	%endif ; !RT_ARCH_AMD64
363
364	%if %2 != 0 ; locked versions requested?
365
366	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u8_locked, 12
367	PROLOGUE_3_ARGS
368	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
369	lock %1 byte [A0], A1_8
370	IEM_SAVE_FLAGS A2, %3, %4
371	EPILOGUE_3_ARGS
372	ENDPROC iemAImpl_ %+ %1 %+ _u8_locked
373
374	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16_locked, 12
375	PROLOGUE_3_ARGS
376	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
377	lock %1 word [A0], A1_16
378	IEM_SAVE_FLAGS A2, %3, %4
379	EPILOGUE_3_ARGS
380	ENDPROC iemAImpl_ %+ %1 %+ _u16_locked
381
382	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32_locked, 12
383	PROLOGUE_3_ARGS
384	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
385	lock %1 dword [A0], A1_32
386	IEM_SAVE_FLAGS A2, %3, %4
387	EPILOGUE_3_ARGS
388	ENDPROC iemAImpl_ %+ %1 %+ _u32_locked
389
390	%ifdef RT_ARCH_AMD64
391	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64_locked, 16
392	PROLOGUE_3_ARGS
393	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
394	lock %1 qword [A0], A1
395	IEM_SAVE_FLAGS A2, %3, %4
396	EPILOGUE_3_ARGS_EX 8
397	ENDPROC iemAImpl_ %+ %1 %+ _u64_locked
398	%else ; stub it for now - later, replace with hand coded stuff.
399	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64_locked, 16
400	int3
401	ret 8
402	ENDPROC iemAImpl_ %+ %1 %+ _u64_locked
403	%endif ; !RT_ARCH_AMD64
404	%endif ; locked
405	%endmacro
406
407	; instr,lock,modified-flags.
408	IEMIMPL_BIN_OP add, 1, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
409	IEMIMPL_BIN_OP adc, 1, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
410	IEMIMPL_BIN_OP sub, 1, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
411	IEMIMPL_BIN_OP sbb, 1, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
412	IEMIMPL_BIN_OP or, 1, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), X86_EFL_AF,
413	IEMIMPL_BIN_OP xor, 1, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), X86_EFL_AF,
414	IEMIMPL_BIN_OP and, 1, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), X86_EFL_AF,
415	IEMIMPL_BIN_OP cmp, 0, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
416	IEMIMPL_BIN_OP test, 0, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), X86_EFL_AF,
417
418
419	;;
420	; Macro for implementing a bit operator.
421	;
422	; This will generate code for the 16, 32 and 64 bit accesses with locked
423	; variants, except on 32-bit system where the 64-bit accesses requires hand
424	; coding.
425	;
426	; All the functions takes a pointer to the destination memory operand in A0,
427	; the source register operand in A1 and a pointer to eflags in A2.
428	;
429	; @param 1 The instruction mnemonic.
430	; @param 2 Non-zero if there should be a locked version.
431	; @param 3 The modified flags.
432	; @param 4 The undefined flags.
433	;
434	%macro IEMIMPL_BIT_OP 4
435	BEGINCODE
436	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 12
437	PROLOGUE_3_ARGS
438	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
439	%1 word [A0], A1_16
440	IEM_SAVE_FLAGS A2, %3, %4
441	EPILOGUE_3_ARGS
442	ENDPROC iemAImpl_ %+ %1 %+ _u16
443
444	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 12
445	PROLOGUE_3_ARGS
446	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
447	%1 dword [A0], A1_32
448	IEM_SAVE_FLAGS A2, %3, %4
449	EPILOGUE_3_ARGS
450	ENDPROC iemAImpl_ %+ %1 %+ _u32
451
452	%ifdef RT_ARCH_AMD64
453	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 16
454	PROLOGUE_3_ARGS
455	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
456	%1 qword [A0], A1
457	IEM_SAVE_FLAGS A2, %3, %4
458	EPILOGUE_3_ARGS_EX 8
459	ENDPROC iemAImpl_ %+ %1 %+ _u64
460	%else ; stub it for now - later, replace with hand coded stuff.
461	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 16
462	int3
463	ret 8
464	ENDPROC iemAImpl_ %+ %1 %+ _u64
465	%endif ; !RT_ARCH_AMD64
466
467	%if %2 != 0 ; locked versions requested?
468
469	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16_locked, 12
470	PROLOGUE_3_ARGS
471	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
472	lock %1 word [A0], A1_16
473	IEM_SAVE_FLAGS A2, %3, %4
474	EPILOGUE_3_ARGS
475	ENDPROC iemAImpl_ %+ %1 %+ _u16_locked
476
477	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32_locked, 12
478	PROLOGUE_3_ARGS
479	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
480	lock %1 dword [A0], A1_32
481	IEM_SAVE_FLAGS A2, %3, %4
482	EPILOGUE_3_ARGS
483	ENDPROC iemAImpl_ %+ %1 %+ _u32_locked
484
485	%ifdef RT_ARCH_AMD64
486	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64_locked, 16
487	PROLOGUE_3_ARGS
488	IEM_MAYBE_LOAD_FLAGS A2, %3, %4
489	lock %1 qword [A0], A1
490	IEM_SAVE_FLAGS A2, %3, %4
491	EPILOGUE_3_ARGS_EX 8
492	ENDPROC iemAImpl_ %+ %1 %+ _u64_locked
493	%else ; stub it for now - later, replace with hand coded stuff.
494	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64_locked, 16
495	int3
496	ret 8
497	ENDPROC iemAImpl_ %+ %1 %+ _u64_locked
498	%endif ; !RT_ARCH_AMD64
499	%endif ; locked
500	%endmacro
501	IEMIMPL_BIT_OP bt, 0, (X86_EFL_CF), (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
502	IEMIMPL_BIT_OP btc, 1, (X86_EFL_CF), (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
503	IEMIMPL_BIT_OP bts, 1, (X86_EFL_CF), (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
504	IEMIMPL_BIT_OP btr, 1, (X86_EFL_CF), (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
505
506	;;
507	; Macro for implementing a bit search operator.
508	;
509	; This will generate code for the 16, 32 and 64 bit accesses, except on 32-bit
510	; system where the 64-bit accesses requires hand coding.
511	;
512	; All the functions takes a pointer to the destination memory operand in A0,
513	; the source register operand in A1 and a pointer to eflags in A2.
514	;
515	; @param 1 The instruction mnemonic.
516	; @param 2 The modified flags.
517	; @param 3 The undefined flags.
518	;
519	%macro IEMIMPL_BIT_OP 3
520	BEGINCODE
521	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 12
522	PROLOGUE_3_ARGS
523	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
524	%1 T0_16, A1_16
525	jz .unchanged_dst
526	mov [A0], T0_16
527	.unchanged_dst:
528	IEM_SAVE_FLAGS A2, %2, %3
529	EPILOGUE_3_ARGS
530	ENDPROC iemAImpl_ %+ %1 %+ _u16
531
532	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 12
533	PROLOGUE_3_ARGS
534	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
535	%1 T0_32, A1_32
536	jz .unchanged_dst
537	mov [A0], T0_32
538	.unchanged_dst:
539	IEM_SAVE_FLAGS A2, %2, %3
540	EPILOGUE_3_ARGS
541	ENDPROC iemAImpl_ %+ %1 %+ _u32
542
543	%ifdef RT_ARCH_AMD64
544	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 16
545	PROLOGUE_3_ARGS
546	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
547	%1 T0, A1
548	jz .unchanged_dst
549	mov [A0], T0
550	.unchanged_dst:
551	IEM_SAVE_FLAGS A2, %2, %3
552	EPILOGUE_3_ARGS_EX 8
553	ENDPROC iemAImpl_ %+ %1 %+ _u64
554	%else ; stub it for now - later, replace with hand coded stuff.
555	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 16
556	int3
557	ret 8
558	ENDPROC iemAImpl_ %+ %1 %+ _u64
559	%endif ; !RT_ARCH_AMD64
560	%endmacro
561	IEMIMPL_BIT_OP bsf, (X86_EFL_ZF), (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF)
562	IEMIMPL_BIT_OP bsr, (X86_EFL_ZF), (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF)
563
564
565	;
566	; IMUL is also a similar but yet different case (no lock, no mem dst).
567	; The rDX:rAX variant of imul is handled together with mul further down.
568	;
569	BEGINCODE
570	BEGINPROC_FASTCALL iemAImpl_imul_two_u16, 12
571	PROLOGUE_3_ARGS
572	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_CF), (X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
573	imul A1_16, word [A0]
574	mov [A0], A1_16
575	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_CF), (X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
576	EPILOGUE_3_ARGS
577	ENDPROC iemAImpl_imul_two_u16
578
579	BEGINPROC_FASTCALL iemAImpl_imul_two_u32, 12
580	PROLOGUE_3_ARGS
581	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_CF), (X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
582	imul A1_32, dword [A0]
583	mov [A0], A1_32
584	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_CF), (X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
585	EPILOGUE_3_ARGS
586	ENDPROC iemAImpl_imul_two_u32
587
588	BEGINPROC_FASTCALL iemAImpl_imul_two_u64, 16
589	PROLOGUE_3_ARGS
590	%ifdef RT_ARCH_AMD64
591	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_CF), (X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
592	imul A1, qword [A0]
593	mov [A0], A1
594	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_CF), (X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
595	%else
596	int3 ;; @todo implement me
597	%endif
598	EPILOGUE_3_ARGS_EX 8
599	ENDPROC iemAImpl_imul_two_u64
600
601
602	;
603	; XCHG for memory operands. This implies locking. No flag changes.
604	;
605	; Each function takes two arguments, first the pointer to the memory,
606	; then the pointer to the register. They all return void.
607	;
608	BEGINCODE
609	BEGINPROC_FASTCALL iemAImpl_xchg_u8, 8
610	PROLOGUE_2_ARGS
611	mov T0_8, [A1]
612	xchg [A0], T0_8
613	mov [A1], T0_8
614	EPILOGUE_2_ARGS
615	ENDPROC iemAImpl_xchg_u8
616
617	BEGINPROC_FASTCALL iemAImpl_xchg_u16, 8
618	PROLOGUE_2_ARGS
619	mov T0_16, [A1]
620	xchg [A0], T0_16
621	mov [A1], T0_16
622	EPILOGUE_2_ARGS
623	ENDPROC iemAImpl_xchg_u16
624
625	BEGINPROC_FASTCALL iemAImpl_xchg_u32, 8
626	PROLOGUE_2_ARGS
627	mov T0_32, [A1]
628	xchg [A0], T0_32
629	mov [A1], T0_32
630	EPILOGUE_2_ARGS
631	ENDPROC iemAImpl_xchg_u32
632
633	BEGINPROC_FASTCALL iemAImpl_xchg_u64, 8
634	%ifdef RT_ARCH_AMD64
635	PROLOGUE_2_ARGS
636	mov T0, [A1]
637	xchg [A0], T0
638	mov [A1], T0
639	EPILOGUE_2_ARGS
640	%else
641	int3
642	ret 0
643	%endif
644	ENDPROC iemAImpl_xchg_u64
645
646
647	;
648	; XADD for memory operands.
649	;
650	; Each function takes three arguments, first the pointer to the
651	; memory/register, then the pointer to the register, and finally a pointer to
652	; eflags. They all return void.
653	;
654	BEGINCODE
655	BEGINPROC_FASTCALL iemAImpl_xadd_u8, 12
656	PROLOGUE_3_ARGS
657	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
658	mov T0_8, [A1]
659	xadd [A0], T0_8
660	mov [A1], T0_8
661	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
662	EPILOGUE_3_ARGS
663	ENDPROC iemAImpl_xadd_u8
664
665	BEGINPROC_FASTCALL iemAImpl_xadd_u16, 12
666	PROLOGUE_3_ARGS
667	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
668	mov T0_16, [A1]
669	xadd [A0], T0_16
670	mov [A1], T0_16
671	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
672	EPILOGUE_3_ARGS
673	ENDPROC iemAImpl_xadd_u16
674
675	BEGINPROC_FASTCALL iemAImpl_xadd_u32, 12
676	PROLOGUE_3_ARGS
677	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
678	mov T0_32, [A1]
679	xadd [A0], T0_32
680	mov [A1], T0_32
681	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
682	EPILOGUE_3_ARGS
683	ENDPROC iemAImpl_xadd_u32
684
685	BEGINPROC_FASTCALL iemAImpl_xadd_u64, 12
686	%ifdef RT_ARCH_AMD64
687	PROLOGUE_3_ARGS
688	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
689	mov T0, [A1]
690	xadd [A0], T0
691	mov [A1], T0
692	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
693	EPILOGUE_3_ARGS
694	%else
695	int3
696	ret 4
697	%endif
698	ENDPROC iemAImpl_xadd_u64
699
700	BEGINPROC_FASTCALL iemAImpl_xadd_u8_locked, 12
701	PROLOGUE_3_ARGS
702	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
703	mov T0_8, [A1]
704	lock xadd [A0], T0_8
705	mov [A1], T0_8
706	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
707	EPILOGUE_3_ARGS
708	ENDPROC iemAImpl_xadd_u8_locked
709
710	BEGINPROC_FASTCALL iemAImpl_xadd_u16_locked, 12
711	PROLOGUE_3_ARGS
712	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
713	mov T0_16, [A1]
714	lock xadd [A0], T0_16
715	mov [A1], T0_16
716	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
717	EPILOGUE_3_ARGS
718	ENDPROC iemAImpl_xadd_u16_locked
719
720	BEGINPROC_FASTCALL iemAImpl_xadd_u32_locked, 12
721	PROLOGUE_3_ARGS
722	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
723	mov T0_32, [A1]
724	lock xadd [A0], T0_32
725	mov [A1], T0_32
726	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
727	EPILOGUE_3_ARGS
728	ENDPROC iemAImpl_xadd_u32_locked
729
730	BEGINPROC_FASTCALL iemAImpl_xadd_u64_locked, 12
731	%ifdef RT_ARCH_AMD64
732	PROLOGUE_3_ARGS
733	IEM_MAYBE_LOAD_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
734	mov T0, [A1]
735	lock xadd [A0], T0
736	mov [A1], T0
737	IEM_SAVE_FLAGS A2, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
738	EPILOGUE_3_ARGS
739	%else
740	int3
741	ret 4
742	%endif
743	ENDPROC iemAImpl_xadd_u64_locked
744
745
746	;
747	; CMPXCHG8B.
748	;
749	; These are tricky register wise, so the code is duplicated for each calling
750	; convention.
751	;
752	; WARNING! This code make ASSUMPTIONS about which registers T1 and T0 are mapped to!
753	;
754	; C-proto:
755	; IEM_DECL_IMPL_DEF(void, iemAImpl_cmpxchg8b,(uint64_t *pu64Dst, PRTUINT64U pu64EaxEdx, PRTUINT64U pu64EbxEcx,
756	; uint32_t *pEFlags));
757	;
758	BEGINCODE
759	BEGINPROC_FASTCALL iemAImpl_cmpxchg8b, 16
760	%ifdef RT_ARCH_AMD64
761	%ifdef ASM_CALL64_MSC
762	push rbx
763
764	mov r11, rdx ; pu64EaxEdx (is also T1)
765	mov r10, rcx ; pu64Dst
766
767	mov ebx, [r8]
768	mov ecx, [r8 + 4]
769	IEM_MAYBE_LOAD_FLAGS r9, (X86_EFL_ZF), 0 ; clobbers T0 (eax)
770	mov eax, [r11]
771	mov edx, [r11 + 4]
772
773	lock cmpxchg8b [r10]
774
775	mov [r11], eax
776	mov [r11 + 4], edx
777	IEM_SAVE_FLAGS r9, (X86_EFL_ZF), 0 ; clobbers T0+T1 (eax, r11)
778
779	pop rbx
780	ret
781	%else
782	push rbx
783
784	mov r10, rcx ; pEFlags
785	mov r11, rdx ; pu64EbxEcx (is also T1)
786
787	mov ebx, [r11]
788	mov ecx, [r11 + 4]
789	IEM_MAYBE_LOAD_FLAGS r10, (X86_EFL_ZF), 0 ; clobbers T0 (eax)
790	mov eax, [rsi]
791	mov edx, [rsi + 4]
792
793	lock cmpxchg8b [rdi]
794
795	mov [rsi], eax
796	mov [rsi + 4], edx
797	IEM_SAVE_FLAGS r10, (X86_EFL_ZF), 0 ; clobbers T0+T1 (eax, r11)
798
799	pop rbx
800	ret
801
802	%endif
803	%else
804	push esi
805	push edi
806	push ebx
807	push ebp
808
809	mov edi, ecx ; pu64Dst
810	mov esi, edx ; pu64EaxEdx
811	mov ecx, [esp + 16 + 4 + 0] ; pu64EbxEcx
812	mov ebp, [esp + 16 + 4 + 4] ; pEFlags
813
814	mov ebx, [ecx]
815	mov ecx, [ecx + 4]
816	IEM_MAYBE_LOAD_FLAGS ebp, (X86_EFL_ZF), 0 ; clobbers T0 (eax)
817	mov eax, [esi]
818	mov edx, [esi + 4]
819
820	lock cmpxchg8b [edi]
821
822	mov [esi], eax
823	mov [esi + 4], edx
824	IEM_SAVE_FLAGS ebp, (X86_EFL_ZF), 0 ; clobbers T0+T1 (eax, edi)
825
826	pop ebp
827	pop ebx
828	pop edi
829	pop esi
830	ret 8
831	%endif
832	ENDPROC iemAImpl_cmpxchg8b
833
834	BEGINPROC_FASTCALL iemAImpl_cmpxchg8b_locked, 16
835	; Lazy bird always lock prefixes cmpxchg8b.
836	jmp NAME_FASTCALL(iemAImpl_cmpxchg8b,16,$@)
837	ENDPROC iemAImpl_cmpxchg8b_locked
838
839
840
841	;
842	; CMPXCHG.
843	;
844	; WARNING! This code make ASSUMPTIONS about which registers T1 and T0 are mapped to!
845	;
846	; C-proto:
847	; IEM_DECL_IMPL_DEF(void, iemAImpl_cmpxchg,(uintX_t puXDst, uintX_t puEax, uintX_t uReg, uint32_t pEFlags));
848	;
849	BEGINCODE
850	%macro IEMIMPL_CMPXCHG 2
851	BEGINPROC_FASTCALL iemAImpl_cmpxchg_u8 %+ %2, 16
852	PROLOGUE_4_ARGS
853	IEM_MAYBE_LOAD_FLAGS A3, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0 (eax)
854	mov al, [A1]
855	%1 cmpxchg [A0], A2_8
856	mov [A1], al
857	IEM_SAVE_FLAGS A3, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0+T1 (eax, r11/edi)
858	EPILOGUE_4_ARGS
859	ENDPROC iemAImpl_cmpxchg_u8 %+ %2
860
861	BEGINPROC_FASTCALL iemAImpl_cmpxchg_u16 %+ %2, 16
862	PROLOGUE_4_ARGS
863	IEM_MAYBE_LOAD_FLAGS A3, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0 (eax)
864	mov ax, [A1]
865	%1 cmpxchg [A0], A2_16
866	mov [A1], ax
867	IEM_SAVE_FLAGS A3, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0+T1 (eax, r11/edi)
868	EPILOGUE_4_ARGS
869	ENDPROC iemAImpl_cmpxchg_u16 %+ %2
870
871	BEGINPROC_FASTCALL iemAImpl_cmpxchg_u32 %+ %2, 16
872	PROLOGUE_4_ARGS
873	IEM_MAYBE_LOAD_FLAGS A3, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0 (eax)
874	mov eax, [A1]
875	%1 cmpxchg [A0], A2_32
876	mov [A1], eax
877	IEM_SAVE_FLAGS A3, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0+T1 (eax, r11/edi)
878	EPILOGUE_4_ARGS
879	ENDPROC iemAImpl_cmpxchg_u32 %+ %2
880
881	BEGINPROC_FASTCALL iemAImpl_cmpxchg_u64 %+ %2, 16
882	%ifdef RT_ARCH_AMD64
883	PROLOGUE_4_ARGS
884	IEM_MAYBE_LOAD_FLAGS A3, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0 (eax)
885	mov ax, [A1]
886	%1 cmpxchg [A0], A2
887	mov [A1], ax
888	IEM_SAVE_FLAGS A3, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0+T1 (eax, r11/edi)
889	EPILOGUE_4_ARGS
890	%else
891	;
892	; Must use cmpxchg8b here. See also iemAImpl_cmpxchg8b.
893	;
894	push esi
895	push edi
896	push ebx
897	push ebp
898
899	mov edi, ecx ; pu64Dst
900	mov esi, edx ; pu64Rax
901	mov ecx, [esp + 16 + 4 + 0] ; pu64Reg - Note! Pointer on 32-bit hosts!
902	mov ebp, [esp + 16 + 4 + 4] ; pEFlags
903
904	mov ebx, [ecx]
905	mov ecx, [ecx + 4]
906	IEM_MAYBE_LOAD_FLAGS ebp, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0 (eax)
907	mov eax, [esi]
908	mov edx, [esi + 4]
909
910	lock cmpxchg8b [edi]
911
912	; cmpxchg8b doesn't set CF, PF, AF, SF and OF, so we have to do that.
913	jz .cmpxchg8b_not_equal
914	cmp eax, eax ; just set the other flags.
915	.store:
916	mov [esi], eax
917	mov [esi + 4], edx
918	IEM_SAVE_FLAGS ebp, (X86_EFL_ZF \| X86_EFL_CF \| X86_EFL_PF \| X86_EFL_AF \| X86_EFL_SF \| X86_EFL_OF), 0 ; clobbers T0+T1 (eax, edi)
919
920	pop ebp
921	pop ebx
922	pop edi
923	pop esi
924	ret 8
925
926	.cmpxchg8b_not_equal:
927	cmp [esi + 4], edx ;; @todo FIXME - verify 64-bit compare implementation
928	jne .store
929	cmp [esi], eax
930	jmp .store
931
932	%endif
933	ENDPROC iemAImpl_cmpxchg_u64 %+ %2
934	%endmacro ; IEMIMPL_CMPXCHG
935
936	IEMIMPL_CMPXCHG , ,
937	IEMIMPL_CMPXCHG lock, _locked
938
939	;;
940	; Macro for implementing a unary operator.
941	;
942	; This will generate code for the 8, 16, 32 and 64 bit accesses with locked
943	; variants, except on 32-bit system where the 64-bit accesses requires hand
944	; coding.
945	;
946	; All the functions takes a pointer to the destination memory operand in A0,
947	; the source register operand in A1 and a pointer to eflags in A2.
948	;
949	; @param 1 The instruction mnemonic.
950	; @param 2 The modified flags.
951	; @param 3 The undefined flags.
952	;
953	%macro IEMIMPL_UNARY_OP 3
954	BEGINCODE
955	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u8, 8
956	PROLOGUE_2_ARGS
957	IEM_MAYBE_LOAD_FLAGS A1, %2, %3
958	%1 byte [A0]
959	IEM_SAVE_FLAGS A1, %2, %3
960	EPILOGUE_2_ARGS
961	ENDPROC iemAImpl_ %+ %1 %+ _u8
962
963	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u8_locked, 8
964	PROLOGUE_2_ARGS
965	IEM_MAYBE_LOAD_FLAGS A1, %2, %3
966	lock %1 byte [A0]
967	IEM_SAVE_FLAGS A1, %2, %3
968	EPILOGUE_2_ARGS
969	ENDPROC iemAImpl_ %+ %1 %+ _u8_locked
970
971	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 8
972	PROLOGUE_2_ARGS
973	IEM_MAYBE_LOAD_FLAGS A1, %2, %3
974	%1 word [A0]
975	IEM_SAVE_FLAGS A1, %2, %3
976	EPILOGUE_2_ARGS
977	ENDPROC iemAImpl_ %+ %1 %+ _u16
978
979	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16_locked, 8
980	PROLOGUE_2_ARGS
981	IEM_MAYBE_LOAD_FLAGS A1, %2, %3
982	lock %1 word [A0]
983	IEM_SAVE_FLAGS A1, %2, %3
984	EPILOGUE_2_ARGS
985	ENDPROC iemAImpl_ %+ %1 %+ _u16_locked
986
987	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 8
988	PROLOGUE_2_ARGS
989	IEM_MAYBE_LOAD_FLAGS A1, %2, %3
990	%1 dword [A0]
991	IEM_SAVE_FLAGS A1, %2, %3
992	EPILOGUE_2_ARGS
993	ENDPROC iemAImpl_ %+ %1 %+ _u32
994
995	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32_locked, 8
996	PROLOGUE_2_ARGS
997	IEM_MAYBE_LOAD_FLAGS A1, %2, %3
998	lock %1 dword [A0]
999	IEM_SAVE_FLAGS A1, %2, %3
1000	EPILOGUE_2_ARGS
1001	ENDPROC iemAImpl_ %+ %1 %+ _u32_locked
1002
1003	%ifdef RT_ARCH_AMD64
1004	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 8
1005	PROLOGUE_2_ARGS
1006	IEM_MAYBE_LOAD_FLAGS A1, %2, %3
1007	%1 qword [A0]
1008	IEM_SAVE_FLAGS A1, %2, %3
1009	EPILOGUE_2_ARGS
1010	ENDPROC iemAImpl_ %+ %1 %+ _u64
1011
1012	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64_locked, 8
1013	PROLOGUE_2_ARGS
1014	IEM_MAYBE_LOAD_FLAGS A1, %2, %3
1015	lock %1 qword [A0]
1016	IEM_SAVE_FLAGS A1, %2, %3
1017	EPILOGUE_2_ARGS
1018	ENDPROC iemAImpl_ %+ %1 %+ _u64_locked
1019	%else
1020	; stub them for now.
1021	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 8
1022	int3
1023	ret 0
1024	ENDPROC iemAImpl_ %+ %1 %+ _u64
1025	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64_locked, 8
1026	int3
1027	ret 0
1028	ENDPROC iemAImpl_ %+ %1 %+ _u64_locked
1029	%endif
1030
1031	%endmacro
1032
1033	IEMIMPL_UNARY_OP inc, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF), 0
1034	IEMIMPL_UNARY_OP dec, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF), 0
1035	IEMIMPL_UNARY_OP neg, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
1036	IEMIMPL_UNARY_OP not, 0, 0
1037
1038
1039
1040	;;
1041	; Macro for implementing a shift operation.
1042	;
1043	; This will generate code for the 8, 16, 32 and 64 bit accesses, except on
1044	; 32-bit system where the 64-bit accesses requires hand coding.
1045	;
1046	; All the functions takes a pointer to the destination memory operand in A0,
1047	; the shift count in A1 and a pointer to eflags in A2.
1048	;
1049	; @param 1 The instruction mnemonic.
1050	; @param 2 The modified flags.
1051	; @param 3 The undefined flags.
1052	;
1053	; Makes ASSUMPTIONS about A0, A1 and A2 assignments.
1054	;
1055	%macro IEMIMPL_SHIFT_OP 3
1056	BEGINCODE
1057	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u8, 12
1058	PROLOGUE_3_ARGS
1059	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
1060	%ifdef ASM_CALL64_GCC
1061	mov cl, A1_8
1062	%1 byte [A0], cl
1063	%else
1064	xchg A1, A0
1065	%1 byte [A1], cl
1066	%endif
1067	IEM_SAVE_FLAGS A2, %2, %3
1068	EPILOGUE_3_ARGS
1069	ENDPROC iemAImpl_ %+ %1 %+ _u8
1070
1071	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 12
1072	PROLOGUE_3_ARGS
1073	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
1074	%ifdef ASM_CALL64_GCC
1075	mov cl, A1_8
1076	%1 word [A0], cl
1077	%else
1078	xchg A1, A0
1079	%1 word [A1], cl
1080	%endif
1081	IEM_SAVE_FLAGS A2, %2, %3
1082	EPILOGUE_3_ARGS
1083	ENDPROC iemAImpl_ %+ %1 %+ _u16
1084
1085	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 12
1086	PROLOGUE_3_ARGS
1087	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
1088	%ifdef ASM_CALL64_GCC
1089	mov cl, A1_8
1090	%1 dword [A0], cl
1091	%else
1092	xchg A1, A0
1093	%1 dword [A1], cl
1094	%endif
1095	IEM_SAVE_FLAGS A2, %2, %3
1096	EPILOGUE_3_ARGS
1097	ENDPROC iemAImpl_ %+ %1 %+ _u32
1098
1099	%ifdef RT_ARCH_AMD64
1100	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 12
1101	PROLOGUE_3_ARGS
1102	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
1103	%ifdef ASM_CALL64_GCC
1104	mov cl, A1_8
1105	%1 qword [A0], cl
1106	%else
1107	xchg A1, A0
1108	%1 qword [A1], cl
1109	%endif
1110	IEM_SAVE_FLAGS A2, %2, %3
1111	EPILOGUE_3_ARGS
1112	ENDPROC iemAImpl_ %+ %1 %+ _u64
1113	%else ; stub it for now - later, replace with hand coded stuff.
1114	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 12
1115	int3
1116	ret 4
1117	ENDPROC iemAImpl_ %+ %1 %+ _u64
1118	%endif ; !RT_ARCH_AMD64
1119
1120	%endmacro
1121
1122	IEMIMPL_SHIFT_OP rol, (X86_EFL_OF \| X86_EFL_CF), 0
1123	IEMIMPL_SHIFT_OP ror, (X86_EFL_OF \| X86_EFL_CF), 0
1124	IEMIMPL_SHIFT_OP rcl, (X86_EFL_OF \| X86_EFL_CF), 0
1125	IEMIMPL_SHIFT_OP rcr, (X86_EFL_OF \| X86_EFL_CF), 0
1126	IEMIMPL_SHIFT_OP shl, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), (X86_EFL_AF)
1127	IEMIMPL_SHIFT_OP shr, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), (X86_EFL_AF)
1128	IEMIMPL_SHIFT_OP sar, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), (X86_EFL_AF)
1129
1130
1131	;;
1132	; Macro for implementing a double precision shift operation.
1133	;
1134	; This will generate code for the 16, 32 and 64 bit accesses, except on
1135	; 32-bit system where the 64-bit accesses requires hand coding.
1136	;
1137	; The functions takes the destination operand (r/m) in A0, the source (reg) in
1138	; A1, the shift count in A2 and a pointer to the eflags variable/register in A3.
1139	;
1140	; @param 1 The instruction mnemonic.
1141	; @param 2 The modified flags.
1142	; @param 3 The undefined flags.
1143	;
1144	; Makes ASSUMPTIONS about A0, A1, A2 and A3 assignments.
1145	;
1146	%macro IEMIMPL_SHIFT_DBL_OP 3
1147	BEGINCODE
1148	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 16
1149	PROLOGUE_4_ARGS
1150	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1151	%ifdef ASM_CALL64_GCC
1152	xchg A3, A2
1153	%1 [A0], A1_16, cl
1154	xchg A3, A2
1155	%else
1156	xchg A0, A2
1157	%1 [A2], A1_16, cl
1158	%endif
1159	IEM_SAVE_FLAGS A3, %2, %3
1160	EPILOGUE_4_ARGS
1161	ENDPROC iemAImpl_ %+ %1 %+ _u16
1162
1163	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 16
1164	PROLOGUE_4_ARGS
1165	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1166	%ifdef ASM_CALL64_GCC
1167	xchg A3, A2
1168	%1 [A0], A1_32, cl
1169	xchg A3, A2
1170	%else
1171	xchg A0, A2
1172	%1 [A2], A1_32, cl
1173	%endif
1174	IEM_SAVE_FLAGS A3, %2, %3
1175	EPILOGUE_4_ARGS
1176	ENDPROC iemAImpl_ %+ %1 %+ _u32
1177
1178	%ifdef RT_ARCH_AMD64
1179	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 20
1180	PROLOGUE_4_ARGS
1181	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1182	%ifdef ASM_CALL64_GCC
1183	xchg A3, A2
1184	%1 [A0], A1, cl
1185	xchg A3, A2
1186	%else
1187	xchg A0, A2
1188	%1 [A2], A1, cl
1189	%endif
1190	IEM_SAVE_FLAGS A3, %2, %3
1191	EPILOGUE_4_ARGS_EX 12
1192	ENDPROC iemAImpl_ %+ %1 %+ _u64
1193	%else ; stub it for now - later, replace with hand coded stuff.
1194	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 20
1195	int3
1196	ret 12
1197	ENDPROC iemAImpl_ %+ %1 %+ _u64
1198	%endif ; !RT_ARCH_AMD64
1199
1200	%endmacro
1201
1202	IEMIMPL_SHIFT_DBL_OP shld, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), (X86_EFL_AF)
1203	IEMIMPL_SHIFT_DBL_OP shrd, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_PF \| X86_EFL_CF), (X86_EFL_AF)
1204
1205
1206	;;
1207	; Macro for implementing a multiplication operations.
1208	;
1209	; This will generate code for the 8, 16, 32 and 64 bit accesses, except on
1210	; 32-bit system where the 64-bit accesses requires hand coding.
1211	;
1212	; The 8-bit function only operates on AX, so it takes no DX pointer. The other
1213	; functions takes a pointer to rAX in A0, rDX in A1, the operand in A2 and a
1214	; pointer to eflags in A3.
1215	;
1216	; The functions all return 0 so the caller can be used for div/idiv as well as
1217	; for the mul/imul implementation.
1218	;
1219	; @param 1 The instruction mnemonic.
1220	; @param 2 The modified flags.
1221	; @param 3 The undefined flags.
1222	;
1223	; Makes ASSUMPTIONS about A0, A1, A2, A3, T0 and T1 assignments.
1224	;
1225	%macro IEMIMPL_MUL_OP 3
1226	BEGINCODE
1227	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u8, 12
1228	PROLOGUE_3_ARGS
1229	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
1230	mov al, [A0]
1231	%1 A1_8
1232	mov [A0], ax
1233	IEM_SAVE_FLAGS A2, %2, %3
1234	xor eax, eax
1235	EPILOGUE_3_ARGS
1236	ENDPROC iemAImpl_ %+ %1 %+ _u8
1237
1238	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 16
1239	PROLOGUE_4_ARGS
1240	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1241	mov ax, [A0]
1242	%ifdef ASM_CALL64_GCC
1243	%1 A2_16
1244	mov [A0], ax
1245	mov [A1], dx
1246	%else
1247	mov T1, A1
1248	%1 A2_16
1249	mov [A0], ax
1250	mov [T1], dx
1251	%endif
1252	IEM_SAVE_FLAGS A3, %2, %3
1253	xor eax, eax
1254	EPILOGUE_4_ARGS
1255	ENDPROC iemAImpl_ %+ %1 %+ _u16
1256
1257	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 16
1258	PROLOGUE_4_ARGS
1259	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1260	mov eax, [A0]
1261	%ifdef ASM_CALL64_GCC
1262	%1 A2_32
1263	mov [A0], eax
1264	mov [A1], edx
1265	%else
1266	mov T1, A1
1267	%1 A2_32
1268	mov [A0], eax
1269	mov [T1], edx
1270	%endif
1271	IEM_SAVE_FLAGS A3, %2, %3
1272	xor eax, eax
1273	EPILOGUE_4_ARGS
1274	ENDPROC iemAImpl_ %+ %1 %+ _u32
1275
1276	%ifdef RT_ARCH_AMD64
1277	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 20
1278	PROLOGUE_4_ARGS
1279	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1280	mov rax, [A0]
1281	%ifdef ASM_CALL64_GCC
1282	%1 A2
1283	mov [A0], rax
1284	mov [A1], rdx
1285	%else
1286	mov T1, A1
1287	%1 A2
1288	mov [A0], rax
1289	mov [T1], rdx
1290	%endif
1291	IEM_SAVE_FLAGS A3, %2, %3
1292	xor eax, eax
1293	EPILOGUE_4_ARGS_EX 12
1294	ENDPROC iemAImpl_ %+ %1 %+ _u64
1295	%else ; stub it for now - later, replace with hand coded stuff.
1296	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 20
1297	int3
1298	ret 12
1299	ENDPROC iemAImpl_ %+ %1 %+ _u64
1300	%endif ; !RT_ARCH_AMD64
1301
1302	%endmacro
1303
1304	IEMIMPL_MUL_OP mul, (X86_EFL_OF \| X86_EFL_CF), (X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
1305	IEMIMPL_MUL_OP imul, (X86_EFL_OF \| X86_EFL_CF), (X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF)
1306
1307
1308	;;
1309	; Macro for implementing a division operations.
1310	;
1311	; This will generate code for the 8, 16, 32 and 64 bit accesses, except on
1312	; 32-bit system where the 64-bit accesses requires hand coding.
1313	;
1314	; The 8-bit function only operates on AX, so it takes no DX pointer. The other
1315	; functions takes a pointer to rAX in A0, rDX in A1, the operand in A2 and a
1316	; pointer to eflags in A3.
1317	;
1318	; The functions all return 0 on success and -1 if a divide error should be
1319	; raised by the caller.
1320	;
1321	; @param 1 The instruction mnemonic.
1322	; @param 2 The modified flags.
1323	; @param 3 The undefined flags.
1324	;
1325	; Makes ASSUMPTIONS about A0, A1, A2, A3, T0 and T1 assignments.
1326	;
1327	%macro IEMIMPL_DIV_OP 3
1328	BEGINCODE
1329	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u8, 12
1330	PROLOGUE_3_ARGS
1331
1332	test A1_8, A1_8
1333	jz .div_zero
1334	;; @todo test for overflow
1335
1336	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
1337	mov ax, [A0]
1338	%1 A1_8
1339	mov [A0], ax
1340	IEM_SAVE_FLAGS A2, %2, %3
1341	xor eax, eax
1342
1343	.return:
1344	EPILOGUE_3_ARGS
1345
1346	.div_zero:
1347	mov eax, -1
1348	jmp .return
1349	ENDPROC iemAImpl_ %+ %1 %+ _u8
1350
1351	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 16
1352	PROLOGUE_4_ARGS
1353
1354	test A1_16, A1_16
1355	jz .div_zero
1356	;; @todo test for overflow
1357
1358	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1359	%ifdef ASM_CALL64_GCC
1360	mov T1, A2
1361	mov ax, [A0]
1362	mov dx, [A1]
1363	%1 T1_16
1364	mov [A0], ax
1365	mov [A1], dx
1366	%else
1367	mov T1, A1
1368	mov ax, [A0]
1369	mov dx, [T1]
1370	%1 A2_16
1371	mov [A0], ax
1372	mov [T1], dx
1373	%endif
1374	IEM_SAVE_FLAGS A3, %2, %3
1375	xor eax, eax
1376
1377	.return:
1378	EPILOGUE_4_ARGS
1379
1380	.div_zero:
1381	mov eax, -1
1382	jmp .return
1383	ENDPROC iemAImpl_ %+ %1 %+ _u16
1384
1385	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 16
1386	PROLOGUE_4_ARGS
1387
1388	test A1_32, A1_32
1389	jz .div_zero
1390	;; @todo test for overflow
1391
1392	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1393	mov eax, [A0]
1394	%ifdef ASM_CALL64_GCC
1395	mov T1, A2
1396	mov eax, [A0]
1397	mov edx, [A1]
1398	%1 T1_32
1399	mov [A0], eax
1400	mov [A1], edx
1401	%else
1402	mov T1, A1
1403	mov eax, [A0]
1404	mov edx, [T1]
1405	%1 A2_32
1406	mov [A0], eax
1407	mov [T1], edx
1408	%endif
1409	IEM_SAVE_FLAGS A3, %2, %3
1410	xor eax, eax
1411
1412	.return:
1413	EPILOGUE_4_ARGS
1414
1415	.div_zero:
1416	mov eax, -1
1417	jmp .return
1418	ENDPROC iemAImpl_ %+ %1 %+ _u32
1419
1420	%ifdef RT_ARCH_AMD64
1421	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 20
1422	PROLOGUE_4_ARGS
1423
1424	test A1, A1
1425	jz .div_zero
1426	;; @todo test for overflow
1427
1428	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1429	mov rax, [A0]
1430	%ifdef ASM_CALL64_GCC
1431	mov T1, A2
1432	mov rax, [A0]
1433	mov rdx, [A1]
1434	%1 T1
1435	mov [A0], rax
1436	mov [A1], rdx
1437	%else
1438	mov T1, A1
1439	mov rax, [A0]
1440	mov rdx, [T1]
1441	%1 A2
1442	mov [A0], rax
1443	mov [T1], rdx
1444	%endif
1445	IEM_SAVE_FLAGS A3, %2, %3
1446	xor eax, eax
1447
1448	.return:
1449	EPILOGUE_4_ARGS_EX 12
1450
1451	.div_zero:
1452	mov eax, -1
1453	jmp .return
1454	ENDPROC iemAImpl_ %+ %1 %+ _u64
1455	%else ; stub it for now - later, replace with hand coded stuff.
1456	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 20
1457	int3
1458	ret
1459	ENDPROC iemAImpl_ %+ %1 %+ _u64
1460	%endif ; !RT_ARCH_AMD64
1461
1462	%endmacro
1463
1464	IEMIMPL_DIV_OP div, 0, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF)
1465	IEMIMPL_DIV_OP idiv, 0, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF)
1466
1467
1468	;
1469	; BSWAP. No flag changes.
1470	;
1471	; Each function takes one argument, pointer to the value to bswap
1472	; (input/output). They all return void.
1473	;
1474	BEGINPROC_FASTCALL iemAImpl_bswap_u16, 4
1475	PROLOGUE_1_ARGS
1476	mov T0_32, [A0] ; just in case any of the upper bits are used.
1477	db 66h
1478	bswap T0_32
1479	mov [A0], T0_32
1480	EPILOGUE_1_ARGS
1481	ENDPROC iemAImpl_bswap_u16
1482
1483	BEGINPROC_FASTCALL iemAImpl_bswap_u32, 4
1484	PROLOGUE_1_ARGS
1485	mov T0_32, [A0]
1486	bswap T0_32
1487	mov [A0], T0_32
1488	EPILOGUE_1_ARGS
1489	ENDPROC iemAImpl_bswap_u32
1490
1491	BEGINPROC_FASTCALL iemAImpl_bswap_u64, 4
1492	%ifdef RT_ARCH_AMD64
1493	PROLOGUE_1_ARGS
1494	mov T0, [A0]
1495	bswap T0
1496	mov [A0], T0
1497	EPILOGUE_1_ARGS
1498	%else
1499	PROLOGUE_1_ARGS
1500	mov T0, [A0]
1501	mov T1, [A0 + 4]
1502	bswap T0
1503	bswap T1
1504	mov [A0 + 4], T0
1505	mov [A0], T1
1506	EPILOGUE_1_ARGS
1507	%endif
1508	ENDPROC iemAImpl_bswap_u64
1509
1510
1511	;;
1512	; Initialize the FPU for the actual instruction being emulated, this means
1513	; loading parts of the guest's control word and status word.
1514	;
1515	; @uses 24 bytes of stack.
1516	; @param 1 Expression giving the address of the FXSTATE of the guest.
1517	;
1518	%macro FPU_LD_FXSTATE_FCW_AND_SAFE_FSW 1
1519	fnstenv [xSP]
1520
1521	; FCW - for exception, precision and rounding control.
1522	movzx T0, word [%1 + X86FXSTATE.FCW]
1523	and T0, X86_FCW_MASK_ALL \| X86_FCW_PC_MASK \| X86_FCW_RC_MASK
1524	mov [xSP + X86FSTENV32P.FCW], T0_16
1525
1526	; FSW - for undefined C0, C1, C2, and C3.
1527	movzx T1, word [%1 + X86FXSTATE.FSW]
1528	and T1, X86_FSW_C_MASK
1529	movzx T0, word [xSP + X86FSTENV32P.FSW]
1530	and T0, X86_FSW_TOP_MASK
1531	or T0, T1
1532	mov [xSP + X86FSTENV32P.FSW], T0_16
1533
1534	fldenv [xSP]
1535	%endmacro
1536
1537
1538	;;
1539	; Need to move this as well somewhere better?
1540	;
1541	struc IEMFPURESULT
1542	.r80Result resw 5
1543	.FSW resw 1
1544	endstruc
1545
1546
1547	;;
1548	; Need to move this as well somewhere better?
1549	;
1550	struc IEMFPURESULTTWO
1551	.r80Result1 resw 5
1552	.FSW resw 1
1553	.r80Result2 resw 5
1554	endstruc
1555
1556
1557	;
1558	;---------------------- 16-bit signed integer operations ----------------------
1559	;
1560
1561
1562	;;
1563	; Converts a 16-bit floating point value to a 80-bit one (fpu register).
1564	;
1565	; @param A0 FPU context (fxsave).
1566	; @param A1 Pointer to a IEMFPURESULT for the output.
1567	; @param A2 Pointer to the 16-bit floating point value to convert.
1568	;
1569	BEGINPROC_FASTCALL iemAImpl_fild_i16_to_r80, 12
1570	PROLOGUE_3_ARGS
1571	sub xSP, 20h
1572
1573	fninit
1574	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1575	fild word [A2]
1576
1577	fnstsw word [A1 + IEMFPURESULT.FSW]
1578	fnclex
1579	fstp tword [A1 + IEMFPURESULT.r80Result]
1580
1581	fninit
1582	add xSP, 20h
1583	EPILOGUE_3_ARGS
1584	ENDPROC iemAImpl_fild_i16_to_r80
1585
1586
1587	;;
1588	; Store a 80-bit floating point value (register) as a 16-bit signed integer (memory).
1589	;
1590	; @param A0 FPU context (fxsave).
1591	; @param A1 Where to return the output FSW.
1592	; @param A2 Where to store the 16-bit signed integer value.
1593	; @param A3 Pointer to the 80-bit value.
1594	;
1595	BEGINPROC_FASTCALL iemAImpl_fist_r80_to_i16, 16
1596	PROLOGUE_4_ARGS
1597	sub xSP, 20h
1598
1599	fninit
1600	fld tword [A3]
1601	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1602	fistp word [A2]
1603
1604	fnstsw word [A1]
1605
1606	fninit
1607	add xSP, 20h
1608	EPILOGUE_4_ARGS
1609	ENDPROC iemAImpl_fist_r80_to_i16
1610
1611
1612	;;
1613	; Store a 80-bit floating point value (register) as a 16-bit signed integer
1614	; (memory) with truncation.
1615	;
1616	; @param A0 FPU context (fxsave).
1617	; @param A1 Where to return the output FSW.
1618	; @param A2 Where to store the 16-bit signed integer value.
1619	; @param A3 Pointer to the 80-bit value.
1620	;
1621	BEGINPROC_FASTCALL iemAImpl_fistt_r80_to_i16, 16
1622	PROLOGUE_4_ARGS
1623	sub xSP, 20h
1624
1625	fninit
1626	fld tword [A3]
1627	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1628	fisttp dword [A2]
1629
1630	fnstsw word [A1]
1631
1632	fninit
1633	add xSP, 20h
1634	EPILOGUE_4_ARGS
1635	ENDPROC iemAImpl_fistt_r80_to_i16
1636
1637
1638	;;
1639	; FPU instruction working on one 80-bit and one 16-bit signed integer value.
1640	;
1641	; @param 1 The instruction
1642	;
1643	; @param A0 FPU context (fxsave).
1644	; @param A1 Pointer to a IEMFPURESULT for the output.
1645	; @param A2 Pointer to the 80-bit value.
1646	; @param A3 Pointer to the 16-bit value.
1647	;
1648	%macro IEMIMPL_FPU_R80_BY_I16 1
1649	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_i16, 16
1650	PROLOGUE_4_ARGS
1651	sub xSP, 20h
1652
1653	fninit
1654	fld tword [A2]
1655	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1656	%1 word [A3]
1657
1658	fnstsw word [A1 + IEMFPURESULT.FSW]
1659	fnclex
1660	fstp tword [A1 + IEMFPURESULT.r80Result]
1661
1662	fninit
1663	add xSP, 20h
1664	EPILOGUE_4_ARGS
1665	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_i16
1666	%endmacro
1667
1668	IEMIMPL_FPU_R80_BY_I16 fiadd
1669	IEMIMPL_FPU_R80_BY_I16 fimul
1670	IEMIMPL_FPU_R80_BY_I16 fisub
1671	IEMIMPL_FPU_R80_BY_I16 fisubr
1672	IEMIMPL_FPU_R80_BY_I16 fidiv
1673	IEMIMPL_FPU_R80_BY_I16 fidivr
1674
1675
1676	;;
1677	; FPU instruction working on one 80-bit and one 16-bit signed integer value,
1678	; only returning FSW.
1679	;
1680	; @param 1 The instruction
1681	;
1682	; @param A0 FPU context (fxsave).
1683	; @param A1 Where to store the output FSW.
1684	; @param A2 Pointer to the 80-bit value.
1685	; @param A3 Pointer to the 64-bit value.
1686	;
1687	%macro IEMIMPL_FPU_R80_BY_I16_FSW 1
1688	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_i16, 16
1689	PROLOGUE_4_ARGS
1690	sub xSP, 20h
1691
1692	fninit
1693	fld tword [A2]
1694	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1695	%1 word [A3]
1696
1697	fnstsw word [A1]
1698
1699	fninit
1700	add xSP, 20h
1701	EPILOGUE_4_ARGS
1702	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_i16
1703	%endmacro
1704
1705	IEMIMPL_FPU_R80_BY_I16_FSW ficom
1706
1707
1708
1709	;
1710	;---------------------- 32-bit signed integer operations ----------------------
1711	;
1712
1713
1714	;;
1715	; Converts a 32-bit floating point value to a 80-bit one (fpu register).
1716	;
1717	; @param A0 FPU context (fxsave).
1718	; @param A1 Pointer to a IEMFPURESULT for the output.
1719	; @param A2 Pointer to the 32-bit floating point value to convert.
1720	;
1721	BEGINPROC_FASTCALL iemAImpl_fild_i32_to_r80, 12
1722	PROLOGUE_3_ARGS
1723	sub xSP, 20h
1724
1725	fninit
1726	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1727	fild dword [A2]
1728
1729	fnstsw word [A1 + IEMFPURESULT.FSW]
1730	fnclex
1731	fstp tword [A1 + IEMFPURESULT.r80Result]
1732
1733	fninit
1734	add xSP, 20h
1735	EPILOGUE_3_ARGS
1736	ENDPROC iemAImpl_fild_i32_to_r80
1737
1738
1739	;;
1740	; Store a 80-bit floating point value (register) as a 32-bit signed integer (memory).
1741	;
1742	; @param A0 FPU context (fxsave).
1743	; @param A1 Where to return the output FSW.
1744	; @param A2 Where to store the 32-bit signed integer value.
1745	; @param A3 Pointer to the 80-bit value.
1746	;
1747	BEGINPROC_FASTCALL iemAImpl_fist_r80_to_i32, 16
1748	PROLOGUE_4_ARGS
1749	sub xSP, 20h
1750
1751	fninit
1752	fld tword [A3]
1753	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1754	fistp dword [A2]
1755
1756	fnstsw word [A1]
1757
1758	fninit
1759	add xSP, 20h
1760	EPILOGUE_4_ARGS
1761	ENDPROC iemAImpl_fist_r80_to_i32
1762
1763
1764	;;
1765	; Store a 80-bit floating point value (register) as a 32-bit signed integer
1766	; (memory) with truncation.
1767	;
1768	; @param A0 FPU context (fxsave).
1769	; @param A1 Where to return the output FSW.
1770	; @param A2 Where to store the 32-bit signed integer value.
1771	; @param A3 Pointer to the 80-bit value.
1772	;
1773	BEGINPROC_FASTCALL iemAImpl_fistt_r80_to_i32, 16
1774	PROLOGUE_4_ARGS
1775	sub xSP, 20h
1776
1777	fninit
1778	fld tword [A3]
1779	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1780	fisttp dword [A2]
1781
1782	fnstsw word [A1]
1783
1784	fninit
1785	add xSP, 20h
1786	EPILOGUE_4_ARGS
1787	ENDPROC iemAImpl_fistt_r80_to_i32
1788
1789
1790	;;
1791	; FPU instruction working on one 80-bit and one 32-bit signed integer value.
1792	;
1793	; @param 1 The instruction
1794	;
1795	; @param A0 FPU context (fxsave).
1796	; @param A1 Pointer to a IEMFPURESULT for the output.
1797	; @param A2 Pointer to the 80-bit value.
1798	; @param A3 Pointer to the 32-bit value.
1799	;
1800	%macro IEMIMPL_FPU_R80_BY_I32 1
1801	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_i32, 16
1802	PROLOGUE_4_ARGS
1803	sub xSP, 20h
1804
1805	fninit
1806	fld tword [A2]
1807	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1808	%1 dword [A3]
1809
1810	fnstsw word [A1 + IEMFPURESULT.FSW]
1811	fnclex
1812	fstp tword [A1 + IEMFPURESULT.r80Result]
1813
1814	fninit
1815	add xSP, 20h
1816	EPILOGUE_4_ARGS
1817	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_i32
1818	%endmacro
1819
1820	IEMIMPL_FPU_R80_BY_I32 fiadd
1821	IEMIMPL_FPU_R80_BY_I32 fimul
1822	IEMIMPL_FPU_R80_BY_I32 fisub
1823	IEMIMPL_FPU_R80_BY_I32 fisubr
1824	IEMIMPL_FPU_R80_BY_I32 fidiv
1825	IEMIMPL_FPU_R80_BY_I32 fidivr
1826
1827
1828	;;
1829	; FPU instruction working on one 80-bit and one 32-bit signed integer value,
1830	; only returning FSW.
1831	;
1832	; @param 1 The instruction
1833	;
1834	; @param A0 FPU context (fxsave).
1835	; @param A1 Where to store the output FSW.
1836	; @param A2 Pointer to the 80-bit value.
1837	; @param A3 Pointer to the 64-bit value.
1838	;
1839	%macro IEMIMPL_FPU_R80_BY_I32_FSW 1
1840	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_i32, 16
1841	PROLOGUE_4_ARGS
1842	sub xSP, 20h
1843
1844	fninit
1845	fld tword [A2]
1846	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1847	%1 dword [A3]
1848
1849	fnstsw word [A1]
1850
1851	fninit
1852	add xSP, 20h
1853	EPILOGUE_4_ARGS
1854	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_i32
1855	%endmacro
1856
1857	IEMIMPL_FPU_R80_BY_I32_FSW ficom
1858
1859
1860
1861	;
1862	;---------------------- 64-bit signed integer operations ----------------------
1863	;
1864
1865
1866	;;
1867	; Converts a 64-bit floating point value to a 80-bit one (fpu register).
1868	;
1869	; @param A0 FPU context (fxsave).
1870	; @param A1 Pointer to a IEMFPURESULT for the output.
1871	; @param A2 Pointer to the 64-bit floating point value to convert.
1872	;
1873	BEGINPROC_FASTCALL iemAImpl_fild_i64_to_r80, 12
1874	PROLOGUE_3_ARGS
1875	sub xSP, 20h
1876
1877	fninit
1878	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1879	fild qword [A2]
1880
1881	fnstsw word [A1 + IEMFPURESULT.FSW]
1882	fnclex
1883	fstp tword [A1 + IEMFPURESULT.r80Result]
1884
1885	fninit
1886	add xSP, 20h
1887	EPILOGUE_3_ARGS
1888	ENDPROC iemAImpl_fild_i64_to_r80
1889
1890
1891	;;
1892	; Store a 80-bit floating point value (register) as a 64-bit signed integer (memory).
1893	;
1894	; @param A0 FPU context (fxsave).
1895	; @param A1 Where to return the output FSW.
1896	; @param A2 Where to store the 64-bit signed integer value.
1897	; @param A3 Pointer to the 80-bit value.
1898	;
1899	BEGINPROC_FASTCALL iemAImpl_fist_r80_to_i64, 16
1900	PROLOGUE_4_ARGS
1901	sub xSP, 20h
1902
1903	fninit
1904	fld tword [A3]
1905	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1906	fistp qword [A2]
1907
1908	fnstsw word [A1]
1909
1910	fninit
1911	add xSP, 20h
1912	EPILOGUE_4_ARGS
1913	ENDPROC iemAImpl_fist_r80_to_i64
1914
1915
1916	;;
1917	; Store a 80-bit floating point value (register) as a 64-bit signed integer
1918	; (memory) with truncation.
1919	;
1920	; @param A0 FPU context (fxsave).
1921	; @param A1 Where to return the output FSW.
1922	; @param A2 Where to store the 64-bit signed integer value.
1923	; @param A3 Pointer to the 80-bit value.
1924	;
1925	BEGINPROC_FASTCALL iemAImpl_fistt_r80_to_i64, 16
1926	PROLOGUE_4_ARGS
1927	sub xSP, 20h
1928
1929	fninit
1930	fld tword [A3]
1931	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1932	fisttp qword [A2]
1933
1934	fnstsw word [A1]
1935
1936	fninit
1937	add xSP, 20h
1938	EPILOGUE_4_ARGS
1939	ENDPROC iemAImpl_fistt_r80_to_i64
1940
1941
1942
1943	;
1944	;---------------------- 32-bit floating point operations ----------------------
1945	;
1946
1947	;;
1948	; Converts a 32-bit floating point value to a 80-bit one (fpu register).
1949	;
1950	; @param A0 FPU context (fxsave).
1951	; @param A1 Pointer to a IEMFPURESULT for the output.
1952	; @param A2 Pointer to the 32-bit floating point value to convert.
1953	;
1954	BEGINPROC_FASTCALL iemAImpl_fld_r32_to_r80, 12
1955	PROLOGUE_3_ARGS
1956	sub xSP, 20h
1957
1958	fninit
1959	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1960	fld dword [A2]
1961
1962	fnstsw word [A1 + IEMFPURESULT.FSW]
1963	fnclex
1964	fstp tword [A1 + IEMFPURESULT.r80Result]
1965
1966	fninit
1967	add xSP, 20h
1968	EPILOGUE_3_ARGS
1969	ENDPROC iemAImpl_fld_r32_to_r80
1970
1971
1972	;;
1973	; Store a 80-bit floating point value (register) as a 32-bit one (memory).
1974	;
1975	; @param A0 FPU context (fxsave).
1976	; @param A1 Where to return the output FSW.
1977	; @param A2 Where to store the 32-bit value.
1978	; @param A3 Pointer to the 80-bit value.
1979	;
1980	BEGINPROC_FASTCALL iemAImpl_fst_r80_to_r32, 16
1981	PROLOGUE_4_ARGS
1982	sub xSP, 20h
1983
1984	fninit
1985	fld tword [A3]
1986	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1987	fst dword [A2]
1988
1989	fnstsw word [A1]
1990
1991	fninit
1992	add xSP, 20h
1993	EPILOGUE_4_ARGS
1994	ENDPROC iemAImpl_fst_r80_to_r32
1995
1996
1997	;;
1998	; FPU instruction working on one 80-bit and one 32-bit floating point value.
1999	;
2000	; @param 1 The instruction
2001	;
2002	; @param A0 FPU context (fxsave).
2003	; @param A1 Pointer to a IEMFPURESULT for the output.
2004	; @param A2 Pointer to the 80-bit value.
2005	; @param A3 Pointer to the 32-bit value.
2006	;
2007	%macro IEMIMPL_FPU_R80_BY_R32 1
2008	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r32, 16
2009	PROLOGUE_4_ARGS
2010	sub xSP, 20h
2011
2012	fninit
2013	fld tword [A2]
2014	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2015	%1 dword [A3]
2016
2017	fnstsw word [A1 + IEMFPURESULT.FSW]
2018	fnclex
2019	fstp tword [A1 + IEMFPURESULT.r80Result]
2020
2021	fninit
2022	add xSP, 20h
2023	EPILOGUE_4_ARGS
2024	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r32
2025	%endmacro
2026
2027	IEMIMPL_FPU_R80_BY_R32 fadd
2028	IEMIMPL_FPU_R80_BY_R32 fmul
2029	IEMIMPL_FPU_R80_BY_R32 fsub
2030	IEMIMPL_FPU_R80_BY_R32 fsubr
2031	IEMIMPL_FPU_R80_BY_R32 fdiv
2032	IEMIMPL_FPU_R80_BY_R32 fdivr
2033
2034
2035	;;
2036	; FPU instruction working on one 80-bit and one 32-bit floating point value,
2037	; only returning FSW.
2038	;
2039	; @param 1 The instruction
2040	;
2041	; @param A0 FPU context (fxsave).
2042	; @param A1 Where to store the output FSW.
2043	; @param A2 Pointer to the 80-bit value.
2044	; @param A3 Pointer to the 64-bit value.
2045	;
2046	%macro IEMIMPL_FPU_R80_BY_R32_FSW 1
2047	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r32, 16
2048	PROLOGUE_4_ARGS
2049	sub xSP, 20h
2050
2051	fninit
2052	fld tword [A2]
2053	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2054	%1 dword [A3]
2055
2056	fnstsw word [A1]
2057
2058	fninit
2059	add xSP, 20h
2060	EPILOGUE_4_ARGS
2061	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r32
2062	%endmacro
2063
2064	IEMIMPL_FPU_R80_BY_R32_FSW fcom
2065
2066
2067
2068	;
2069	;---------------------- 64-bit floating point operations ----------------------
2070	;
2071
2072	;;
2073	; Converts a 64-bit floating point value to a 80-bit one (fpu register).
2074	;
2075	; @param A0 FPU context (fxsave).
2076	; @param A1 Pointer to a IEMFPURESULT for the output.
2077	; @param A2 Pointer to the 64-bit floating point value to convert.
2078	;
2079	BEGINPROC_FASTCALL iemAImpl_fld_r64_to_r80, 12
2080	PROLOGUE_3_ARGS
2081	sub xSP, 20h
2082
2083	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2084	fld qword [A2]
2085
2086	fnstsw word [A1 + IEMFPURESULT.FSW]
2087	fnclex
2088	fstp tword [A1 + IEMFPURESULT.r80Result]
2089
2090	fninit
2091	add xSP, 20h
2092	EPILOGUE_3_ARGS
2093	ENDPROC iemAImpl_fld_r64_to_r80
2094
2095
2096	;;
2097	; Store a 80-bit floating point value (register) as a 64-bit one (memory).
2098	;
2099	; @param A0 FPU context (fxsave).
2100	; @param A1 Where to return the output FSW.
2101	; @param A2 Where to store the 64-bit value.
2102	; @param A3 Pointer to the 80-bit value.
2103	;
2104	BEGINPROC_FASTCALL iemAImpl_fst_r80_to_r64, 16
2105	PROLOGUE_4_ARGS
2106	sub xSP, 20h
2107
2108	fninit
2109	fld tword [A3]
2110	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2111	fst qword [A2]
2112
2113	fnstsw word [A1]
2114
2115	fninit
2116	add xSP, 20h
2117	EPILOGUE_4_ARGS
2118	ENDPROC iemAImpl_fst_r80_to_r64
2119
2120
2121	;;
2122	; FPU instruction working on one 80-bit and one 64-bit floating point value.
2123	;
2124	; @param 1 The instruction
2125	;
2126	; @param A0 FPU context (fxsave).
2127	; @param A1 Pointer to a IEMFPURESULT for the output.
2128	; @param A2 Pointer to the 80-bit value.
2129	; @param A3 Pointer to the 64-bit value.
2130	;
2131	%macro IEMIMPL_FPU_R80_BY_R64 1
2132	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r64, 16
2133	PROLOGUE_4_ARGS
2134	sub xSP, 20h
2135
2136	fninit
2137	fld tword [A2]
2138	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2139	%1 qword [A3]
2140
2141	fnstsw word [A1 + IEMFPURESULT.FSW]
2142	fnclex
2143	fstp tword [A1 + IEMFPURESULT.r80Result]
2144
2145	fninit
2146	add xSP, 20h
2147	EPILOGUE_4_ARGS
2148	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r64
2149	%endmacro
2150
2151	IEMIMPL_FPU_R80_BY_R64 fadd
2152	IEMIMPL_FPU_R80_BY_R64 fmul
2153	IEMIMPL_FPU_R80_BY_R64 fsub
2154	IEMIMPL_FPU_R80_BY_R64 fsubr
2155	IEMIMPL_FPU_R80_BY_R64 fdiv
2156	IEMIMPL_FPU_R80_BY_R64 fdivr
2157
2158	;;
2159	; FPU instruction working on one 80-bit and one 64-bit floating point value,
2160	; only returning FSW.
2161	;
2162	; @param 1 The instruction
2163	;
2164	; @param A0 FPU context (fxsave).
2165	; @param A1 Where to store the output FSW.
2166	; @param A2 Pointer to the 80-bit value.
2167	; @param A3 Pointer to the 64-bit value.
2168	;
2169	%macro IEMIMPL_FPU_R80_BY_R64_FSW 1
2170	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r64, 16
2171	PROLOGUE_4_ARGS
2172	sub xSP, 20h
2173
2174	fninit
2175	fld tword [A2]
2176	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2177	%1 qword [A3]
2178
2179	fnstsw word [A1]
2180
2181	fninit
2182	add xSP, 20h
2183	EPILOGUE_4_ARGS
2184	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r64
2185	%endmacro
2186
2187	IEMIMPL_FPU_R80_BY_R64_FSW fcom
2188
2189
2190
2191	;
2192	;---------------------- 80-bit floating point operations ----------------------
2193	;
2194
2195	;;
2196	; Loads a 80-bit floating point register value from memory.
2197	;
2198	; @param A0 FPU context (fxsave).
2199	; @param A1 Pointer to a IEMFPURESULT for the output.
2200	; @param A2 Pointer to the 80-bit floating point value to load.
2201	;
2202	BEGINPROC_FASTCALL iemAImpl_fld_r80_from_r80, 12
2203	PROLOGUE_3_ARGS
2204	sub xSP, 20h
2205
2206	fninit
2207	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2208	fld tword [A2]
2209
2210	fnstsw word [A1 + IEMFPURESULT.FSW]
2211	fnclex
2212	fstp tword [A1 + IEMFPURESULT.r80Result]
2213
2214	fninit
2215	add xSP, 20h
2216	EPILOGUE_3_ARGS
2217	ENDPROC iemAImpl_fld_r80_from_r80
2218
2219
2220	;;
2221	; Store a 80-bit floating point register to memory
2222	;
2223	; @param A0 FPU context (fxsave).
2224	; @param A1 Where to return the output FSW.
2225	; @param A2 Where to store the 80-bit value.
2226	; @param A3 Pointer to the 80-bit register value.
2227	;
2228	BEGINPROC_FASTCALL iemAImpl_fst_r80_to_r80, 16
2229	PROLOGUE_4_ARGS
2230	sub xSP, 20h
2231
2232	fninit
2233	fld tword [A3]
2234	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2235	fstp tword [A2]
2236
2237	fnstsw word [A1]
2238
2239	fninit
2240	add xSP, 20h
2241	EPILOGUE_4_ARGS
2242	ENDPROC iemAImpl_fst_r80_to_r80
2243
2244
2245	;;
2246	; FPU instruction working on two 80-bit floating point values.
2247	;
2248	; @param 1 The instruction
2249	;
2250	; @param A0 FPU context (fxsave).
2251	; @param A1 Pointer to a IEMFPURESULT for the output.
2252	; @param A2 Pointer to the first 80-bit value (ST0)
2253	; @param A3 Pointer to the second 80-bit value (STn).
2254	;
2255	%macro IEMIMPL_FPU_R80_BY_R80 2
2256	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r80, 16
2257	PROLOGUE_4_ARGS
2258	sub xSP, 20h
2259
2260	fninit
2261	fld tword [A3]
2262	fld tword [A2]
2263	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2264	%1 %2
2265
2266	fnstsw word [A1 + IEMFPURESULT.FSW]
2267	fnclex
2268	fstp tword [A1 + IEMFPURESULT.r80Result]
2269
2270	fninit
2271	add xSP, 20h
2272	EPILOGUE_4_ARGS
2273	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r80
2274	%endmacro
2275
2276	IEMIMPL_FPU_R80_BY_R80 fadd, {st0, st1}
2277	IEMIMPL_FPU_R80_BY_R80 fmul, {st0, st1}
2278	IEMIMPL_FPU_R80_BY_R80 fsub, {st0, st1}
2279	IEMIMPL_FPU_R80_BY_R80 fsubr, {st0, st1}
2280	IEMIMPL_FPU_R80_BY_R80 fdiv, {st0, st1}
2281	IEMIMPL_FPU_R80_BY_R80 fdivr, {st0, st1}
2282	IEMIMPL_FPU_R80_BY_R80 fprem, {}
2283	IEMIMPL_FPU_R80_BY_R80 fprem1, {}
2284	IEMIMPL_FPU_R80_BY_R80 fscale, {}
2285
2286
2287	;;
2288	; FPU instruction working on two 80-bit floating point values, ST1 and ST0,
2289	; storing the result in ST1 and popping the stack.
2290	;
2291	; @param 1 The instruction
2292	;
2293	; @param A0 FPU context (fxsave).
2294	; @param A1 Pointer to a IEMFPURESULT for the output.
2295	; @param A2 Pointer to the first 80-bit value (ST1).
2296	; @param A3 Pointer to the second 80-bit value (ST0).
2297	;
2298	%macro IEMIMPL_FPU_R80_BY_R80_ST1_ST0_POP 1
2299	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r80, 16
2300	PROLOGUE_4_ARGS
2301	sub xSP, 20h
2302
2303	fninit
2304	fld tword [A2]
2305	fld tword [A3]
2306	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2307	%1
2308
2309	fnstsw word [A1 + IEMFPURESULT.FSW]
2310	fnclex
2311	fstp tword [A1 + IEMFPURESULT.r80Result]
2312
2313	fninit
2314	add xSP, 20h
2315	EPILOGUE_4_ARGS
2316	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r80
2317	%endmacro
2318
2319	IEMIMPL_FPU_R80_BY_R80_ST1_ST0_POP fpatan
2320	IEMIMPL_FPU_R80_BY_R80_ST1_ST0_POP fyl2xp1
2321
2322
2323	;;
2324	; FPU instruction working on two 80-bit floating point values, only
2325	; returning FSW.
2326	;
2327	; @param 1 The instruction
2328	;
2329	; @param A0 FPU context (fxsave).
2330	; @param A1 Pointer to a uint16_t for the resulting FSW.
2331	; @param A2 Pointer to the first 80-bit value.
2332	; @param A3 Pointer to the second 80-bit value.
2333	;
2334	%macro IEMIMPL_FPU_R80_BY_R80_FSW 1
2335	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r80, 16
2336	PROLOGUE_4_ARGS
2337	sub xSP, 20h
2338
2339	fninit
2340	fld tword [A3]
2341	fld tword [A2]
2342	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2343	%1 st0, st1
2344
2345	fnstsw word [A1]
2346
2347	fninit
2348	add xSP, 20h
2349	EPILOGUE_4_ARGS
2350	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r80
2351	%endmacro
2352
2353	IEMIMPL_FPU_R80_BY_R80_FSW fcom
2354	IEMIMPL_FPU_R80_BY_R80_FSW fucom
2355
2356
2357	;;
2358	; FPU instruction working on two 80-bit floating point values,
2359	; returning FSW and EFLAGS (eax).
2360	;
2361	; @param 1 The instruction
2362	;
2363	; @returns EFLAGS in EAX.
2364	; @param A0 FPU context (fxsave).
2365	; @param A1 Pointer to a uint16_t for the resulting FSW.
2366	; @param A2 Pointer to the first 80-bit value.
2367	; @param A3 Pointer to the second 80-bit value.
2368	;
2369	%macro IEMIMPL_FPU_R80_BY_R80_EFL 1
2370	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r80, 16
2371	PROLOGUE_4_ARGS
2372	sub xSP, 20h
2373
2374	fninit
2375	fld tword [A3]
2376	fld tword [A2]
2377	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2378	%1 st1
2379
2380	fnstsw word [A1]
2381	pushf
2382	pop xAX
2383
2384	fninit
2385	add xSP, 20h
2386	EPILOGUE_4_ARGS
2387	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r80
2388	%endmacro
2389
2390	IEMIMPL_FPU_R80_BY_R80_EFL fcomi
2391	IEMIMPL_FPU_R80_BY_R80_EFL fucomi
2392
2393
2394	;;
2395	; FPU instruction working on one 80-bit floating point value.
2396	;
2397	; @param 1 The instruction
2398	;
2399	; @param A0 FPU context (fxsave).
2400	; @param A1 Pointer to a IEMFPURESULT for the output.
2401	; @param A2 Pointer to the 80-bit value.
2402	;
2403	%macro IEMIMPL_FPU_R80 1
2404	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80, 12
2405	PROLOGUE_3_ARGS
2406	sub xSP, 20h
2407
2408	fninit
2409	fld tword [A2]
2410	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2411	%1
2412
2413	fnstsw word [A1 + IEMFPURESULT.FSW]
2414	fnclex
2415	fstp tword [A1 + IEMFPURESULT.r80Result]
2416
2417	fninit
2418	add xSP, 20h
2419	EPILOGUE_3_ARGS
2420	ENDPROC iemAImpl_ %+ %1 %+ _r80
2421	%endmacro
2422
2423	IEMIMPL_FPU_R80 fchs
2424	IEMIMPL_FPU_R80 fabs
2425	IEMIMPL_FPU_R80 f2xm1
2426	IEMIMPL_FPU_R80 fyl2x
2427	IEMIMPL_FPU_R80 fsqrt
2428	IEMIMPL_FPU_R80 frndint
2429	IEMIMPL_FPU_R80 fsin
2430	IEMIMPL_FPU_R80 fcos
2431
2432
2433	;;
2434	; FPU instruction working on one 80-bit floating point value, only
2435	; returning FSW.
2436	;
2437	; @param 1 The instruction
2438	;
2439	; @param A0 FPU context (fxsave).
2440	; @param A1 Pointer to a uint16_t for the resulting FSW.
2441	; @param A2 Pointer to the 80-bit value.
2442	;
2443	%macro IEMIMPL_FPU_R80_FSW 1
2444	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80, 12
2445	PROLOGUE_3_ARGS
2446	sub xSP, 20h
2447
2448	fninit
2449	fld tword [A2]
2450	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2451	%1
2452
2453	fnstsw word [A1]
2454
2455	fninit
2456	add xSP, 20h
2457	EPILOGUE_3_ARGS
2458	ENDPROC iemAImpl_ %+ %1 %+ _r80
2459	%endmacro
2460
2461	IEMIMPL_FPU_R80_FSW ftst
2462	IEMIMPL_FPU_R80_FSW fxam
2463
2464
2465
2466	;;
2467	; FPU instruction loading a 80-bit floating point constant.
2468	;
2469	; @param 1 The instruction
2470	;
2471	; @param A0 FPU context (fxsave).
2472	; @param A1 Pointer to a IEMFPURESULT for the output.
2473	;
2474	%macro IEMIMPL_FPU_R80_CONST 1
2475	BEGINPROC_FASTCALL iemAImpl_ %+ %1, 8
2476	PROLOGUE_2_ARGS
2477	sub xSP, 20h
2478
2479	fninit
2480	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2481	%1
2482
2483	fnstsw word [A1 + IEMFPURESULT.FSW]
2484	fnclex
2485	fstp tword [A1 + IEMFPURESULT.r80Result]
2486
2487	fninit
2488	add xSP, 20h
2489	EPILOGUE_2_ARGS
2490	ENDPROC iemAImpl_ %+ %1 %+
2491	%endmacro
2492
2493	IEMIMPL_FPU_R80_CONST fld1
2494	IEMIMPL_FPU_R80_CONST fldl2t
2495	IEMIMPL_FPU_R80_CONST fldl2e
2496	IEMIMPL_FPU_R80_CONST fldpi
2497	IEMIMPL_FPU_R80_CONST fldlg2
2498	IEMIMPL_FPU_R80_CONST fldln2
2499	IEMIMPL_FPU_R80_CONST fldz
2500
2501
2502	;;
2503	; FPU instruction working on one 80-bit floating point value, outputing two.
2504	;
2505	; @param 1 The instruction
2506	;
2507	; @param A0 FPU context (fxsave).
2508	; @param A1 Pointer to a IEMFPURESULTTWO for the output.
2509	; @param A2 Pointer to the 80-bit value.
2510	;
2511	%macro IEMIMPL_FPU_R80_R80 1
2512	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_r80, 12
2513	PROLOGUE_3_ARGS
2514	sub xSP, 20h
2515
2516	fninit
2517	fld tword [A2]
2518	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2519	%1
2520
2521	fnstsw word [A1 + IEMFPURESULTTWO.FSW]
2522	fnclex
2523	fstp tword [A1 + IEMFPURESULTTWO.r80Result2]
2524	fnclex
2525	fstp tword [A1 + IEMFPURESULTTWO.r80Result1]
2526
2527	fninit
2528	add xSP, 20h
2529	EPILOGUE_3_ARGS
2530	ENDPROC iemAImpl_ %+ %1 %+ _r80_r80
2531	%endmacro
2532
2533	IEMIMPL_FPU_R80_R80 fptan
2534	IEMIMPL_FPU_R80_R80 fxtract
2535	IEMIMPL_FPU_R80_R80 fsincos
2536

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

source: vbox/trunk/src/VBox/VMM/VMMAll/IEMAllAImpl.asm@ 45276

Download in other formats: