IEMAllAImpl.asm@ 46992

Last change on this file since 46992 was 46992, checked in by vboxsync, 11 years ago
8-bit idiv overflow checks.
Property svn:eol-style set to `native` Property svn:keywords set to `Author Date Id Revision`
File size: 70.3 KB

Line
1	; $Id: IEMAllAImpl.asm 46992 2013-07-05 08:00:06Z 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	; @param 4 1 if signed, 0 if unsigned.
1325	;
1326	; Makes ASSUMPTIONS about A0, A1, A2, A3, T0 and T1 assignments.
1327	;
1328	%macro IEMIMPL_DIV_OP 4
1329	BEGINCODE
1330	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u8, 12
1331	PROLOGUE_3_ARGS
1332
1333	; div by chainsaw check.
1334	test A1_8, A1_8
1335	jz .div_zero
1336
1337	; Overflow check - unsigned division is simple to verify, haven't
1338	; found a simple way to check signed division yet unfortunately.
1339	%if %4 == 0
1340	cmp [A0 + 1], A1_8
1341	jae .div_overflow
1342	%else
1343	mov T0_16, [A0] ; T0 = dividend
1344	mov T1_8, A1_8 ; T1 = divisor
1345	test T1_16, T1_16
1346	js .divisor_negative
1347	test T0_16, T0_16
1348	jns .both_positive
1349	neg T0_16
1350	.one_of_each: ; OK range is 2^(result-with - 1) + (divisor - 1).
1351	push T0 ; Start off like unsigned below.
1352	shr T0_16, 7
1353	cmp T0_8, T1_8
1354	pop T0
1355	jb .div_no_overflow
1356	ja .div_overflow
1357	and T0_8, 0x7f ; Special case for covering (divisor - 1).
1358	cmp T0_8, T1_8
1359	jae .div_overflow
1360	jmp .div_no_overflow
1361
1362	.divisor_negative:
1363	neg T1_8
1364	test T0_16, T0_16
1365	jns .one_of_each
1366	neg T0_16
1367	.both_positive: ; Same as unsigned shifted by sign indicator bit.
1368	shr T0_16, 7
1369	cmp T0_8, T1_8
1370	jae .div_overflow
1371	.div_no_overflow:
1372	%endif
1373
1374	IEM_MAYBE_LOAD_FLAGS A2, %2, %3
1375	mov ax, [A0]
1376	%1 A1_8
1377	mov [A0], ax
1378	IEM_SAVE_FLAGS A2, %2, %3
1379	xor eax, eax
1380
1381	.return:
1382	EPILOGUE_3_ARGS
1383
1384	.div_zero:
1385	.div_overflow:
1386	mov eax, -1
1387	jmp .return
1388	ENDPROC iemAImpl_ %+ %1 %+ _u8
1389
1390	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u16, 16
1391	PROLOGUE_4_ARGS
1392
1393	test A2_16, A2_16
1394	jz .div_zero
1395	%if %4 == 0
1396	cmp [A1], A2_16
1397	jae .div_overflow
1398	%else
1399	;; @todo idiv overflow checking.
1400	%endif
1401
1402	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1403	%ifdef ASM_CALL64_GCC
1404	mov T1, A2
1405	mov ax, [A0]
1406	mov dx, [A1]
1407	%1 T1_16
1408	mov [A0], ax
1409	mov [A1], dx
1410	%else
1411	mov T1, A1
1412	mov ax, [A0]
1413	mov dx, [T1]
1414	%1 A2_16
1415	mov [A0], ax
1416	mov [T1], dx
1417	%endif
1418	IEM_SAVE_FLAGS A3, %2, %3
1419	xor eax, eax
1420
1421	.return:
1422	EPILOGUE_4_ARGS
1423
1424	.div_zero:
1425	.div_overflow:
1426	mov eax, -1
1427	jmp .return
1428	ENDPROC iemAImpl_ %+ %1 %+ _u16
1429
1430	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u32, 16
1431	PROLOGUE_4_ARGS
1432
1433	test A2_32, A2_32
1434	jz .div_zero
1435	%if %4 == 0
1436	cmp [A1], A2_32
1437	jae .div_overflow
1438	%else
1439	;; @todo idiv overflow checking.
1440	%endif
1441
1442	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1443	mov eax, [A0]
1444	%ifdef ASM_CALL64_GCC
1445	mov T1, A2
1446	mov eax, [A0]
1447	mov edx, [A1]
1448	%1 T1_32
1449	mov [A0], eax
1450	mov [A1], edx
1451	%else
1452	mov T1, A1
1453	mov eax, [A0]
1454	mov edx, [T1]
1455	%1 A2_32
1456	mov [A0], eax
1457	mov [T1], edx
1458	%endif
1459	IEM_SAVE_FLAGS A3, %2, %3
1460	xor eax, eax
1461
1462	.return:
1463	EPILOGUE_4_ARGS
1464
1465	.div_zero:
1466	.div_overflow:
1467	mov eax, -1
1468	jmp .return
1469	ENDPROC iemAImpl_ %+ %1 %+ _u32
1470
1471	%ifdef RT_ARCH_AMD64
1472	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 20
1473	PROLOGUE_4_ARGS
1474
1475	test A2, A2
1476	jz .div_zero
1477	%if %4 == 0
1478	cmp [A1], A2
1479	jae .div_overflow
1480	%else
1481	;; @todo idiv overflow checking.
1482	%endif
1483
1484	IEM_MAYBE_LOAD_FLAGS A3, %2, %3
1485	mov rax, [A0]
1486	%ifdef ASM_CALL64_GCC
1487	mov T1, A2
1488	mov rax, [A0]
1489	mov rdx, [A1]
1490	%1 T1
1491	mov [A0], rax
1492	mov [A1], rdx
1493	%else
1494	mov T1, A1
1495	mov rax, [A0]
1496	mov rdx, [T1]
1497	%1 A2
1498	mov [A0], rax
1499	mov [T1], rdx
1500	%endif
1501	IEM_SAVE_FLAGS A3, %2, %3
1502	xor eax, eax
1503
1504	.return:
1505	EPILOGUE_4_ARGS_EX 12
1506
1507	.div_zero:
1508	.div_overflow:
1509	mov eax, -1
1510	jmp .return
1511	ENDPROC iemAImpl_ %+ %1 %+ _u64
1512	%else ; stub it for now - later, replace with hand coded stuff.
1513	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _u64, 20
1514	int3
1515	ret
1516	ENDPROC iemAImpl_ %+ %1 %+ _u64
1517	%endif ; !RT_ARCH_AMD64
1518
1519	%endmacro
1520
1521	IEMIMPL_DIV_OP div, 0, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 0
1522	IEMIMPL_DIV_OP idiv, 0, (X86_EFL_OF \| X86_EFL_SF \| X86_EFL_ZF \| X86_EFL_AF \| X86_EFL_PF \| X86_EFL_CF), 1
1523
1524
1525	;
1526	; BSWAP. No flag changes.
1527	;
1528	; Each function takes one argument, pointer to the value to bswap
1529	; (input/output). They all return void.
1530	;
1531	BEGINPROC_FASTCALL iemAImpl_bswap_u16, 4
1532	PROLOGUE_1_ARGS
1533	mov T0_32, [A0] ; just in case any of the upper bits are used.
1534	db 66h
1535	bswap T0_32
1536	mov [A0], T0_32
1537	EPILOGUE_1_ARGS
1538	ENDPROC iemAImpl_bswap_u16
1539
1540	BEGINPROC_FASTCALL iemAImpl_bswap_u32, 4
1541	PROLOGUE_1_ARGS
1542	mov T0_32, [A0]
1543	bswap T0_32
1544	mov [A0], T0_32
1545	EPILOGUE_1_ARGS
1546	ENDPROC iemAImpl_bswap_u32
1547
1548	BEGINPROC_FASTCALL iemAImpl_bswap_u64, 4
1549	%ifdef RT_ARCH_AMD64
1550	PROLOGUE_1_ARGS
1551	mov T0, [A0]
1552	bswap T0
1553	mov [A0], T0
1554	EPILOGUE_1_ARGS
1555	%else
1556	PROLOGUE_1_ARGS
1557	mov T0, [A0]
1558	mov T1, [A0 + 4]
1559	bswap T0
1560	bswap T1
1561	mov [A0 + 4], T0
1562	mov [A0], T1
1563	EPILOGUE_1_ARGS
1564	%endif
1565	ENDPROC iemAImpl_bswap_u64
1566
1567
1568	;;
1569	; Initialize the FPU for the actual instruction being emulated, this means
1570	; loading parts of the guest's control word and status word.
1571	;
1572	; @uses 24 bytes of stack.
1573	; @param 1 Expression giving the address of the FXSTATE of the guest.
1574	;
1575	%macro FPU_LD_FXSTATE_FCW_AND_SAFE_FSW 1
1576	fnstenv [xSP]
1577
1578	; FCW - for exception, precision and rounding control.
1579	movzx T0, word [%1 + X86FXSTATE.FCW]
1580	and T0, X86_FCW_MASK_ALL \| X86_FCW_PC_MASK \| X86_FCW_RC_MASK
1581	mov [xSP + X86FSTENV32P.FCW], T0_16
1582
1583	; FSW - for undefined C0, C1, C2, and C3.
1584	movzx T1, word [%1 + X86FXSTATE.FSW]
1585	and T1, X86_FSW_C_MASK
1586	movzx T0, word [xSP + X86FSTENV32P.FSW]
1587	and T0, X86_FSW_TOP_MASK
1588	or T0, T1
1589	mov [xSP + X86FSTENV32P.FSW], T0_16
1590
1591	fldenv [xSP]
1592	%endmacro
1593
1594
1595	;;
1596	; Need to move this as well somewhere better?
1597	;
1598	struc IEMFPURESULT
1599	.r80Result resw 5
1600	.FSW resw 1
1601	endstruc
1602
1603
1604	;;
1605	; Need to move this as well somewhere better?
1606	;
1607	struc IEMFPURESULTTWO
1608	.r80Result1 resw 5
1609	.FSW resw 1
1610	.r80Result2 resw 5
1611	endstruc
1612
1613
1614	;
1615	;---------------------- 16-bit signed integer operations ----------------------
1616	;
1617
1618
1619	;;
1620	; Converts a 16-bit floating point value to a 80-bit one (fpu register).
1621	;
1622	; @param A0 FPU context (fxsave).
1623	; @param A1 Pointer to a IEMFPURESULT for the output.
1624	; @param A2 Pointer to the 16-bit floating point value to convert.
1625	;
1626	BEGINPROC_FASTCALL iemAImpl_fild_i16_to_r80, 12
1627	PROLOGUE_3_ARGS
1628	sub xSP, 20h
1629
1630	fninit
1631	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1632	fild word [A2]
1633
1634	fnstsw word [A1 + IEMFPURESULT.FSW]
1635	fnclex
1636	fstp tword [A1 + IEMFPURESULT.r80Result]
1637
1638	fninit
1639	add xSP, 20h
1640	EPILOGUE_3_ARGS
1641	ENDPROC iemAImpl_fild_i16_to_r80
1642
1643
1644	;;
1645	; Store a 80-bit floating point value (register) as a 16-bit signed integer (memory).
1646	;
1647	; @param A0 FPU context (fxsave).
1648	; @param A1 Where to return the output FSW.
1649	; @param A2 Where to store the 16-bit signed integer value.
1650	; @param A3 Pointer to the 80-bit value.
1651	;
1652	BEGINPROC_FASTCALL iemAImpl_fist_r80_to_i16, 16
1653	PROLOGUE_4_ARGS
1654	sub xSP, 20h
1655
1656	fninit
1657	fld tword [A3]
1658	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1659	fistp word [A2]
1660
1661	fnstsw word [A1]
1662
1663	fninit
1664	add xSP, 20h
1665	EPILOGUE_4_ARGS
1666	ENDPROC iemAImpl_fist_r80_to_i16
1667
1668
1669	;;
1670	; Store a 80-bit floating point value (register) as a 16-bit signed integer
1671	; (memory) with truncation.
1672	;
1673	; @param A0 FPU context (fxsave).
1674	; @param A1 Where to return the output FSW.
1675	; @param A2 Where to store the 16-bit signed integer value.
1676	; @param A3 Pointer to the 80-bit value.
1677	;
1678	BEGINPROC_FASTCALL iemAImpl_fistt_r80_to_i16, 16
1679	PROLOGUE_4_ARGS
1680	sub xSP, 20h
1681
1682	fninit
1683	fld tword [A3]
1684	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1685	fisttp dword [A2]
1686
1687	fnstsw word [A1]
1688
1689	fninit
1690	add xSP, 20h
1691	EPILOGUE_4_ARGS
1692	ENDPROC iemAImpl_fistt_r80_to_i16
1693
1694
1695	;;
1696	; FPU instruction working on one 80-bit and one 16-bit signed integer value.
1697	;
1698	; @param 1 The instruction
1699	;
1700	; @param A0 FPU context (fxsave).
1701	; @param A1 Pointer to a IEMFPURESULT for the output.
1702	; @param A2 Pointer to the 80-bit value.
1703	; @param A3 Pointer to the 16-bit value.
1704	;
1705	%macro IEMIMPL_FPU_R80_BY_I16 1
1706	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_i16, 16
1707	PROLOGUE_4_ARGS
1708	sub xSP, 20h
1709
1710	fninit
1711	fld tword [A2]
1712	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1713	%1 word [A3]
1714
1715	fnstsw word [A1 + IEMFPURESULT.FSW]
1716	fnclex
1717	fstp tword [A1 + IEMFPURESULT.r80Result]
1718
1719	fninit
1720	add xSP, 20h
1721	EPILOGUE_4_ARGS
1722	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_i16
1723	%endmacro
1724
1725	IEMIMPL_FPU_R80_BY_I16 fiadd
1726	IEMIMPL_FPU_R80_BY_I16 fimul
1727	IEMIMPL_FPU_R80_BY_I16 fisub
1728	IEMIMPL_FPU_R80_BY_I16 fisubr
1729	IEMIMPL_FPU_R80_BY_I16 fidiv
1730	IEMIMPL_FPU_R80_BY_I16 fidivr
1731
1732
1733	;;
1734	; FPU instruction working on one 80-bit and one 16-bit signed integer value,
1735	; only returning FSW.
1736	;
1737	; @param 1 The instruction
1738	;
1739	; @param A0 FPU context (fxsave).
1740	; @param A1 Where to store the output FSW.
1741	; @param A2 Pointer to the 80-bit value.
1742	; @param A3 Pointer to the 64-bit value.
1743	;
1744	%macro IEMIMPL_FPU_R80_BY_I16_FSW 1
1745	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_i16, 16
1746	PROLOGUE_4_ARGS
1747	sub xSP, 20h
1748
1749	fninit
1750	fld tword [A2]
1751	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1752	%1 word [A3]
1753
1754	fnstsw word [A1]
1755
1756	fninit
1757	add xSP, 20h
1758	EPILOGUE_4_ARGS
1759	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_i16
1760	%endmacro
1761
1762	IEMIMPL_FPU_R80_BY_I16_FSW ficom
1763
1764
1765
1766	;
1767	;---------------------- 32-bit signed integer operations ----------------------
1768	;
1769
1770
1771	;;
1772	; Converts a 32-bit floating point value to a 80-bit one (fpu register).
1773	;
1774	; @param A0 FPU context (fxsave).
1775	; @param A1 Pointer to a IEMFPURESULT for the output.
1776	; @param A2 Pointer to the 32-bit floating point value to convert.
1777	;
1778	BEGINPROC_FASTCALL iemAImpl_fild_i32_to_r80, 12
1779	PROLOGUE_3_ARGS
1780	sub xSP, 20h
1781
1782	fninit
1783	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1784	fild dword [A2]
1785
1786	fnstsw word [A1 + IEMFPURESULT.FSW]
1787	fnclex
1788	fstp tword [A1 + IEMFPURESULT.r80Result]
1789
1790	fninit
1791	add xSP, 20h
1792	EPILOGUE_3_ARGS
1793	ENDPROC iemAImpl_fild_i32_to_r80
1794
1795
1796	;;
1797	; Store a 80-bit floating point value (register) as a 32-bit signed integer (memory).
1798	;
1799	; @param A0 FPU context (fxsave).
1800	; @param A1 Where to return the output FSW.
1801	; @param A2 Where to store the 32-bit signed integer value.
1802	; @param A3 Pointer to the 80-bit value.
1803	;
1804	BEGINPROC_FASTCALL iemAImpl_fist_r80_to_i32, 16
1805	PROLOGUE_4_ARGS
1806	sub xSP, 20h
1807
1808	fninit
1809	fld tword [A3]
1810	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1811	fistp dword [A2]
1812
1813	fnstsw word [A1]
1814
1815	fninit
1816	add xSP, 20h
1817	EPILOGUE_4_ARGS
1818	ENDPROC iemAImpl_fist_r80_to_i32
1819
1820
1821	;;
1822	; Store a 80-bit floating point value (register) as a 32-bit signed integer
1823	; (memory) with truncation.
1824	;
1825	; @param A0 FPU context (fxsave).
1826	; @param A1 Where to return the output FSW.
1827	; @param A2 Where to store the 32-bit signed integer value.
1828	; @param A3 Pointer to the 80-bit value.
1829	;
1830	BEGINPROC_FASTCALL iemAImpl_fistt_r80_to_i32, 16
1831	PROLOGUE_4_ARGS
1832	sub xSP, 20h
1833
1834	fninit
1835	fld tword [A3]
1836	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1837	fisttp dword [A2]
1838
1839	fnstsw word [A1]
1840
1841	fninit
1842	add xSP, 20h
1843	EPILOGUE_4_ARGS
1844	ENDPROC iemAImpl_fistt_r80_to_i32
1845
1846
1847	;;
1848	; FPU instruction working on one 80-bit and one 32-bit signed integer value.
1849	;
1850	; @param 1 The instruction
1851	;
1852	; @param A0 FPU context (fxsave).
1853	; @param A1 Pointer to a IEMFPURESULT for the output.
1854	; @param A2 Pointer to the 80-bit value.
1855	; @param A3 Pointer to the 32-bit value.
1856	;
1857	%macro IEMIMPL_FPU_R80_BY_I32 1
1858	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_i32, 16
1859	PROLOGUE_4_ARGS
1860	sub xSP, 20h
1861
1862	fninit
1863	fld tword [A2]
1864	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1865	%1 dword [A3]
1866
1867	fnstsw word [A1 + IEMFPURESULT.FSW]
1868	fnclex
1869	fstp tword [A1 + IEMFPURESULT.r80Result]
1870
1871	fninit
1872	add xSP, 20h
1873	EPILOGUE_4_ARGS
1874	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_i32
1875	%endmacro
1876
1877	IEMIMPL_FPU_R80_BY_I32 fiadd
1878	IEMIMPL_FPU_R80_BY_I32 fimul
1879	IEMIMPL_FPU_R80_BY_I32 fisub
1880	IEMIMPL_FPU_R80_BY_I32 fisubr
1881	IEMIMPL_FPU_R80_BY_I32 fidiv
1882	IEMIMPL_FPU_R80_BY_I32 fidivr
1883
1884
1885	;;
1886	; FPU instruction working on one 80-bit and one 32-bit signed integer value,
1887	; only returning FSW.
1888	;
1889	; @param 1 The instruction
1890	;
1891	; @param A0 FPU context (fxsave).
1892	; @param A1 Where to store the output FSW.
1893	; @param A2 Pointer to the 80-bit value.
1894	; @param A3 Pointer to the 64-bit value.
1895	;
1896	%macro IEMIMPL_FPU_R80_BY_I32_FSW 1
1897	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_i32, 16
1898	PROLOGUE_4_ARGS
1899	sub xSP, 20h
1900
1901	fninit
1902	fld tword [A2]
1903	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1904	%1 dword [A3]
1905
1906	fnstsw word [A1]
1907
1908	fninit
1909	add xSP, 20h
1910	EPILOGUE_4_ARGS
1911	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_i32
1912	%endmacro
1913
1914	IEMIMPL_FPU_R80_BY_I32_FSW ficom
1915
1916
1917
1918	;
1919	;---------------------- 64-bit signed integer operations ----------------------
1920	;
1921
1922
1923	;;
1924	; Converts a 64-bit floating point value to a 80-bit one (fpu register).
1925	;
1926	; @param A0 FPU context (fxsave).
1927	; @param A1 Pointer to a IEMFPURESULT for the output.
1928	; @param A2 Pointer to the 64-bit floating point value to convert.
1929	;
1930	BEGINPROC_FASTCALL iemAImpl_fild_i64_to_r80, 12
1931	PROLOGUE_3_ARGS
1932	sub xSP, 20h
1933
1934	fninit
1935	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1936	fild qword [A2]
1937
1938	fnstsw word [A1 + IEMFPURESULT.FSW]
1939	fnclex
1940	fstp tword [A1 + IEMFPURESULT.r80Result]
1941
1942	fninit
1943	add xSP, 20h
1944	EPILOGUE_3_ARGS
1945	ENDPROC iemAImpl_fild_i64_to_r80
1946
1947
1948	;;
1949	; Store a 80-bit floating point value (register) as a 64-bit signed integer (memory).
1950	;
1951	; @param A0 FPU context (fxsave).
1952	; @param A1 Where to return the output FSW.
1953	; @param A2 Where to store the 64-bit signed integer value.
1954	; @param A3 Pointer to the 80-bit value.
1955	;
1956	BEGINPROC_FASTCALL iemAImpl_fist_r80_to_i64, 16
1957	PROLOGUE_4_ARGS
1958	sub xSP, 20h
1959
1960	fninit
1961	fld tword [A3]
1962	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1963	fistp qword [A2]
1964
1965	fnstsw word [A1]
1966
1967	fninit
1968	add xSP, 20h
1969	EPILOGUE_4_ARGS
1970	ENDPROC iemAImpl_fist_r80_to_i64
1971
1972
1973	;;
1974	; Store a 80-bit floating point value (register) as a 64-bit signed integer
1975	; (memory) with truncation.
1976	;
1977	; @param A0 FPU context (fxsave).
1978	; @param A1 Where to return the output FSW.
1979	; @param A2 Where to store the 64-bit signed integer value.
1980	; @param A3 Pointer to the 80-bit value.
1981	;
1982	BEGINPROC_FASTCALL iemAImpl_fistt_r80_to_i64, 16
1983	PROLOGUE_4_ARGS
1984	sub xSP, 20h
1985
1986	fninit
1987	fld tword [A3]
1988	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
1989	fisttp qword [A2]
1990
1991	fnstsw word [A1]
1992
1993	fninit
1994	add xSP, 20h
1995	EPILOGUE_4_ARGS
1996	ENDPROC iemAImpl_fistt_r80_to_i64
1997
1998
1999
2000	;
2001	;---------------------- 32-bit floating point operations ----------------------
2002	;
2003
2004	;;
2005	; Converts a 32-bit floating point value to a 80-bit one (fpu register).
2006	;
2007	; @param A0 FPU context (fxsave).
2008	; @param A1 Pointer to a IEMFPURESULT for the output.
2009	; @param A2 Pointer to the 32-bit floating point value to convert.
2010	;
2011	BEGINPROC_FASTCALL iemAImpl_fld_r32_to_r80, 12
2012	PROLOGUE_3_ARGS
2013	sub xSP, 20h
2014
2015	fninit
2016	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2017	fld dword [A2]
2018
2019	fnstsw word [A1 + IEMFPURESULT.FSW]
2020	fnclex
2021	fstp tword [A1 + IEMFPURESULT.r80Result]
2022
2023	fninit
2024	add xSP, 20h
2025	EPILOGUE_3_ARGS
2026	ENDPROC iemAImpl_fld_r32_to_r80
2027
2028
2029	;;
2030	; Store a 80-bit floating point value (register) as a 32-bit one (memory).
2031	;
2032	; @param A0 FPU context (fxsave).
2033	; @param A1 Where to return the output FSW.
2034	; @param A2 Where to store the 32-bit value.
2035	; @param A3 Pointer to the 80-bit value.
2036	;
2037	BEGINPROC_FASTCALL iemAImpl_fst_r80_to_r32, 16
2038	PROLOGUE_4_ARGS
2039	sub xSP, 20h
2040
2041	fninit
2042	fld tword [A3]
2043	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2044	fst dword [A2]
2045
2046	fnstsw word [A1]
2047
2048	fninit
2049	add xSP, 20h
2050	EPILOGUE_4_ARGS
2051	ENDPROC iemAImpl_fst_r80_to_r32
2052
2053
2054	;;
2055	; FPU instruction working on one 80-bit and one 32-bit floating point value.
2056	;
2057	; @param 1 The instruction
2058	;
2059	; @param A0 FPU context (fxsave).
2060	; @param A1 Pointer to a IEMFPURESULT for the output.
2061	; @param A2 Pointer to the 80-bit value.
2062	; @param A3 Pointer to the 32-bit value.
2063	;
2064	%macro IEMIMPL_FPU_R80_BY_R32 1
2065	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r32, 16
2066	PROLOGUE_4_ARGS
2067	sub xSP, 20h
2068
2069	fninit
2070	fld tword [A2]
2071	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2072	%1 dword [A3]
2073
2074	fnstsw word [A1 + IEMFPURESULT.FSW]
2075	fnclex
2076	fstp tword [A1 + IEMFPURESULT.r80Result]
2077
2078	fninit
2079	add xSP, 20h
2080	EPILOGUE_4_ARGS
2081	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r32
2082	%endmacro
2083
2084	IEMIMPL_FPU_R80_BY_R32 fadd
2085	IEMIMPL_FPU_R80_BY_R32 fmul
2086	IEMIMPL_FPU_R80_BY_R32 fsub
2087	IEMIMPL_FPU_R80_BY_R32 fsubr
2088	IEMIMPL_FPU_R80_BY_R32 fdiv
2089	IEMIMPL_FPU_R80_BY_R32 fdivr
2090
2091
2092	;;
2093	; FPU instruction working on one 80-bit and one 32-bit floating point value,
2094	; only returning FSW.
2095	;
2096	; @param 1 The instruction
2097	;
2098	; @param A0 FPU context (fxsave).
2099	; @param A1 Where to store the output FSW.
2100	; @param A2 Pointer to the 80-bit value.
2101	; @param A3 Pointer to the 64-bit value.
2102	;
2103	%macro IEMIMPL_FPU_R80_BY_R32_FSW 1
2104	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r32, 16
2105	PROLOGUE_4_ARGS
2106	sub xSP, 20h
2107
2108	fninit
2109	fld tword [A2]
2110	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2111	%1 dword [A3]
2112
2113	fnstsw word [A1]
2114
2115	fninit
2116	add xSP, 20h
2117	EPILOGUE_4_ARGS
2118	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r32
2119	%endmacro
2120
2121	IEMIMPL_FPU_R80_BY_R32_FSW fcom
2122
2123
2124
2125	;
2126	;---------------------- 64-bit floating point operations ----------------------
2127	;
2128
2129	;;
2130	; Converts a 64-bit floating point value to a 80-bit one (fpu register).
2131	;
2132	; @param A0 FPU context (fxsave).
2133	; @param A1 Pointer to a IEMFPURESULT for the output.
2134	; @param A2 Pointer to the 64-bit floating point value to convert.
2135	;
2136	BEGINPROC_FASTCALL iemAImpl_fld_r64_to_r80, 12
2137	PROLOGUE_3_ARGS
2138	sub xSP, 20h
2139
2140	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2141	fld qword [A2]
2142
2143	fnstsw word [A1 + IEMFPURESULT.FSW]
2144	fnclex
2145	fstp tword [A1 + IEMFPURESULT.r80Result]
2146
2147	fninit
2148	add xSP, 20h
2149	EPILOGUE_3_ARGS
2150	ENDPROC iemAImpl_fld_r64_to_r80
2151
2152
2153	;;
2154	; Store a 80-bit floating point value (register) as a 64-bit one (memory).
2155	;
2156	; @param A0 FPU context (fxsave).
2157	; @param A1 Where to return the output FSW.
2158	; @param A2 Where to store the 64-bit value.
2159	; @param A3 Pointer to the 80-bit value.
2160	;
2161	BEGINPROC_FASTCALL iemAImpl_fst_r80_to_r64, 16
2162	PROLOGUE_4_ARGS
2163	sub xSP, 20h
2164
2165	fninit
2166	fld tword [A3]
2167	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2168	fst qword [A2]
2169
2170	fnstsw word [A1]
2171
2172	fninit
2173	add xSP, 20h
2174	EPILOGUE_4_ARGS
2175	ENDPROC iemAImpl_fst_r80_to_r64
2176
2177
2178	;;
2179	; FPU instruction working on one 80-bit and one 64-bit floating point value.
2180	;
2181	; @param 1 The instruction
2182	;
2183	; @param A0 FPU context (fxsave).
2184	; @param A1 Pointer to a IEMFPURESULT for the output.
2185	; @param A2 Pointer to the 80-bit value.
2186	; @param A3 Pointer to the 64-bit value.
2187	;
2188	%macro IEMIMPL_FPU_R80_BY_R64 1
2189	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r64, 16
2190	PROLOGUE_4_ARGS
2191	sub xSP, 20h
2192
2193	fninit
2194	fld tword [A2]
2195	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2196	%1 qword [A3]
2197
2198	fnstsw word [A1 + IEMFPURESULT.FSW]
2199	fnclex
2200	fstp tword [A1 + IEMFPURESULT.r80Result]
2201
2202	fninit
2203	add xSP, 20h
2204	EPILOGUE_4_ARGS
2205	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r64
2206	%endmacro
2207
2208	IEMIMPL_FPU_R80_BY_R64 fadd
2209	IEMIMPL_FPU_R80_BY_R64 fmul
2210	IEMIMPL_FPU_R80_BY_R64 fsub
2211	IEMIMPL_FPU_R80_BY_R64 fsubr
2212	IEMIMPL_FPU_R80_BY_R64 fdiv
2213	IEMIMPL_FPU_R80_BY_R64 fdivr
2214
2215	;;
2216	; FPU instruction working on one 80-bit and one 64-bit floating point value,
2217	; only returning FSW.
2218	;
2219	; @param 1 The instruction
2220	;
2221	; @param A0 FPU context (fxsave).
2222	; @param A1 Where to store the output FSW.
2223	; @param A2 Pointer to the 80-bit value.
2224	; @param A3 Pointer to the 64-bit value.
2225	;
2226	%macro IEMIMPL_FPU_R80_BY_R64_FSW 1
2227	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r64, 16
2228	PROLOGUE_4_ARGS
2229	sub xSP, 20h
2230
2231	fninit
2232	fld tword [A2]
2233	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2234	%1 qword [A3]
2235
2236	fnstsw word [A1]
2237
2238	fninit
2239	add xSP, 20h
2240	EPILOGUE_4_ARGS
2241	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r64
2242	%endmacro
2243
2244	IEMIMPL_FPU_R80_BY_R64_FSW fcom
2245
2246
2247
2248	;
2249	;---------------------- 80-bit floating point operations ----------------------
2250	;
2251
2252	;;
2253	; Loads a 80-bit floating point register value from memory.
2254	;
2255	; @param A0 FPU context (fxsave).
2256	; @param A1 Pointer to a IEMFPURESULT for the output.
2257	; @param A2 Pointer to the 80-bit floating point value to load.
2258	;
2259	BEGINPROC_FASTCALL iemAImpl_fld_r80_from_r80, 12
2260	PROLOGUE_3_ARGS
2261	sub xSP, 20h
2262
2263	fninit
2264	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2265	fld tword [A2]
2266
2267	fnstsw word [A1 + IEMFPURESULT.FSW]
2268	fnclex
2269	fstp tword [A1 + IEMFPURESULT.r80Result]
2270
2271	fninit
2272	add xSP, 20h
2273	EPILOGUE_3_ARGS
2274	ENDPROC iemAImpl_fld_r80_from_r80
2275
2276
2277	;;
2278	; Store a 80-bit floating point register to memory
2279	;
2280	; @param A0 FPU context (fxsave).
2281	; @param A1 Where to return the output FSW.
2282	; @param A2 Where to store the 80-bit value.
2283	; @param A3 Pointer to the 80-bit register value.
2284	;
2285	BEGINPROC_FASTCALL iemAImpl_fst_r80_to_r80, 16
2286	PROLOGUE_4_ARGS
2287	sub xSP, 20h
2288
2289	fninit
2290	fld tword [A3]
2291	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2292	fstp tword [A2]
2293
2294	fnstsw word [A1]
2295
2296	fninit
2297	add xSP, 20h
2298	EPILOGUE_4_ARGS
2299	ENDPROC iemAImpl_fst_r80_to_r80
2300
2301
2302	;;
2303	; FPU instruction working on two 80-bit floating point values.
2304	;
2305	; @param 1 The instruction
2306	;
2307	; @param A0 FPU context (fxsave).
2308	; @param A1 Pointer to a IEMFPURESULT for the output.
2309	; @param A2 Pointer to the first 80-bit value (ST0)
2310	; @param A3 Pointer to the second 80-bit value (STn).
2311	;
2312	%macro IEMIMPL_FPU_R80_BY_R80 2
2313	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r80, 16
2314	PROLOGUE_4_ARGS
2315	sub xSP, 20h
2316
2317	fninit
2318	fld tword [A3]
2319	fld tword [A2]
2320	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2321	%1 %2
2322
2323	fnstsw word [A1 + IEMFPURESULT.FSW]
2324	fnclex
2325	fstp tword [A1 + IEMFPURESULT.r80Result]
2326
2327	fninit
2328	add xSP, 20h
2329	EPILOGUE_4_ARGS
2330	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r80
2331	%endmacro
2332
2333	IEMIMPL_FPU_R80_BY_R80 fadd, {st0, st1}
2334	IEMIMPL_FPU_R80_BY_R80 fmul, {st0, st1}
2335	IEMIMPL_FPU_R80_BY_R80 fsub, {st0, st1}
2336	IEMIMPL_FPU_R80_BY_R80 fsubr, {st0, st1}
2337	IEMIMPL_FPU_R80_BY_R80 fdiv, {st0, st1}
2338	IEMIMPL_FPU_R80_BY_R80 fdivr, {st0, st1}
2339	IEMIMPL_FPU_R80_BY_R80 fprem, {}
2340	IEMIMPL_FPU_R80_BY_R80 fprem1, {}
2341	IEMIMPL_FPU_R80_BY_R80 fscale, {}
2342
2343
2344	;;
2345	; FPU instruction working on two 80-bit floating point values, ST1 and ST0,
2346	; storing the result in ST1 and popping the stack.
2347	;
2348	; @param 1 The instruction
2349	;
2350	; @param A0 FPU context (fxsave).
2351	; @param A1 Pointer to a IEMFPURESULT for the output.
2352	; @param A2 Pointer to the first 80-bit value (ST1).
2353	; @param A3 Pointer to the second 80-bit value (ST0).
2354	;
2355	%macro IEMIMPL_FPU_R80_BY_R80_ST1_ST0_POP 1
2356	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r80, 16
2357	PROLOGUE_4_ARGS
2358	sub xSP, 20h
2359
2360	fninit
2361	fld tword [A2]
2362	fld tword [A3]
2363	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2364	%1
2365
2366	fnstsw word [A1 + IEMFPURESULT.FSW]
2367	fnclex
2368	fstp tword [A1 + IEMFPURESULT.r80Result]
2369
2370	fninit
2371	add xSP, 20h
2372	EPILOGUE_4_ARGS
2373	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r80
2374	%endmacro
2375
2376	IEMIMPL_FPU_R80_BY_R80_ST1_ST0_POP fpatan
2377	IEMIMPL_FPU_R80_BY_R80_ST1_ST0_POP fyl2xp1
2378
2379
2380	;;
2381	; FPU instruction working on two 80-bit floating point values, only
2382	; returning FSW.
2383	;
2384	; @param 1 The instruction
2385	;
2386	; @param A0 FPU context (fxsave).
2387	; @param A1 Pointer to a uint16_t for the resulting FSW.
2388	; @param A2 Pointer to the first 80-bit value.
2389	; @param A3 Pointer to the second 80-bit value.
2390	;
2391	%macro IEMIMPL_FPU_R80_BY_R80_FSW 1
2392	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r80, 16
2393	PROLOGUE_4_ARGS
2394	sub xSP, 20h
2395
2396	fninit
2397	fld tword [A3]
2398	fld tword [A2]
2399	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2400	%1 st0, st1
2401
2402	fnstsw word [A1]
2403
2404	fninit
2405	add xSP, 20h
2406	EPILOGUE_4_ARGS
2407	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r80
2408	%endmacro
2409
2410	IEMIMPL_FPU_R80_BY_R80_FSW fcom
2411	IEMIMPL_FPU_R80_BY_R80_FSW fucom
2412
2413
2414	;;
2415	; FPU instruction working on two 80-bit floating point values,
2416	; returning FSW and EFLAGS (eax).
2417	;
2418	; @param 1 The instruction
2419	;
2420	; @returns EFLAGS in EAX.
2421	; @param A0 FPU context (fxsave).
2422	; @param A1 Pointer to a uint16_t for the resulting FSW.
2423	; @param A2 Pointer to the first 80-bit value.
2424	; @param A3 Pointer to the second 80-bit value.
2425	;
2426	%macro IEMIMPL_FPU_R80_BY_R80_EFL 1
2427	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_by_r80, 16
2428	PROLOGUE_4_ARGS
2429	sub xSP, 20h
2430
2431	fninit
2432	fld tword [A3]
2433	fld tword [A2]
2434	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2435	%1 st1
2436
2437	fnstsw word [A1]
2438	pushf
2439	pop xAX
2440
2441	fninit
2442	add xSP, 20h
2443	EPILOGUE_4_ARGS
2444	ENDPROC iemAImpl_ %+ %1 %+ _r80_by_r80
2445	%endmacro
2446
2447	IEMIMPL_FPU_R80_BY_R80_EFL fcomi
2448	IEMIMPL_FPU_R80_BY_R80_EFL fucomi
2449
2450
2451	;;
2452	; FPU instruction working on one 80-bit floating point value.
2453	;
2454	; @param 1 The instruction
2455	;
2456	; @param A0 FPU context (fxsave).
2457	; @param A1 Pointer to a IEMFPURESULT for the output.
2458	; @param A2 Pointer to the 80-bit value.
2459	;
2460	%macro IEMIMPL_FPU_R80 1
2461	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80, 12
2462	PROLOGUE_3_ARGS
2463	sub xSP, 20h
2464
2465	fninit
2466	fld tword [A2]
2467	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2468	%1
2469
2470	fnstsw word [A1 + IEMFPURESULT.FSW]
2471	fnclex
2472	fstp tword [A1 + IEMFPURESULT.r80Result]
2473
2474	fninit
2475	add xSP, 20h
2476	EPILOGUE_3_ARGS
2477	ENDPROC iemAImpl_ %+ %1 %+ _r80
2478	%endmacro
2479
2480	IEMIMPL_FPU_R80 fchs
2481	IEMIMPL_FPU_R80 fabs
2482	IEMIMPL_FPU_R80 f2xm1
2483	IEMIMPL_FPU_R80 fyl2x
2484	IEMIMPL_FPU_R80 fsqrt
2485	IEMIMPL_FPU_R80 frndint
2486	IEMIMPL_FPU_R80 fsin
2487	IEMIMPL_FPU_R80 fcos
2488
2489
2490	;;
2491	; FPU instruction working on one 80-bit floating point value, only
2492	; returning FSW.
2493	;
2494	; @param 1 The instruction
2495	;
2496	; @param A0 FPU context (fxsave).
2497	; @param A1 Pointer to a uint16_t for the resulting FSW.
2498	; @param A2 Pointer to the 80-bit value.
2499	;
2500	%macro IEMIMPL_FPU_R80_FSW 1
2501	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80, 12
2502	PROLOGUE_3_ARGS
2503	sub xSP, 20h
2504
2505	fninit
2506	fld tword [A2]
2507	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2508	%1
2509
2510	fnstsw word [A1]
2511
2512	fninit
2513	add xSP, 20h
2514	EPILOGUE_3_ARGS
2515	ENDPROC iemAImpl_ %+ %1 %+ _r80
2516	%endmacro
2517
2518	IEMIMPL_FPU_R80_FSW ftst
2519	IEMIMPL_FPU_R80_FSW fxam
2520
2521
2522
2523	;;
2524	; FPU instruction loading a 80-bit floating point constant.
2525	;
2526	; @param 1 The instruction
2527	;
2528	; @param A0 FPU context (fxsave).
2529	; @param A1 Pointer to a IEMFPURESULT for the output.
2530	;
2531	%macro IEMIMPL_FPU_R80_CONST 1
2532	BEGINPROC_FASTCALL iemAImpl_ %+ %1, 8
2533	PROLOGUE_2_ARGS
2534	sub xSP, 20h
2535
2536	fninit
2537	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2538	%1
2539
2540	fnstsw word [A1 + IEMFPURESULT.FSW]
2541	fnclex
2542	fstp tword [A1 + IEMFPURESULT.r80Result]
2543
2544	fninit
2545	add xSP, 20h
2546	EPILOGUE_2_ARGS
2547	ENDPROC iemAImpl_ %+ %1 %+
2548	%endmacro
2549
2550	IEMIMPL_FPU_R80_CONST fld1
2551	IEMIMPL_FPU_R80_CONST fldl2t
2552	IEMIMPL_FPU_R80_CONST fldl2e
2553	IEMIMPL_FPU_R80_CONST fldpi
2554	IEMIMPL_FPU_R80_CONST fldlg2
2555	IEMIMPL_FPU_R80_CONST fldln2
2556	IEMIMPL_FPU_R80_CONST fldz
2557
2558
2559	;;
2560	; FPU instruction working on one 80-bit floating point value, outputing two.
2561	;
2562	; @param 1 The instruction
2563	;
2564	; @param A0 FPU context (fxsave).
2565	; @param A1 Pointer to a IEMFPURESULTTWO for the output.
2566	; @param A2 Pointer to the 80-bit value.
2567	;
2568	%macro IEMIMPL_FPU_R80_R80 1
2569	BEGINPROC_FASTCALL iemAImpl_ %+ %1 %+ _r80_r80, 12
2570	PROLOGUE_3_ARGS
2571	sub xSP, 20h
2572
2573	fninit
2574	fld tword [A2]
2575	FPU_LD_FXSTATE_FCW_AND_SAFE_FSW A0
2576	%1
2577
2578	fnstsw word [A1 + IEMFPURESULTTWO.FSW]
2579	fnclex
2580	fstp tword [A1 + IEMFPURESULTTWO.r80Result2]
2581	fnclex
2582	fstp tword [A1 + IEMFPURESULTTWO.r80Result1]
2583
2584	fninit
2585	add xSP, 20h
2586	EPILOGUE_3_ARGS
2587	ENDPROC iemAImpl_ %+ %1 %+ _r80_r80
2588	%endmacro
2589
2590	IEMIMPL_FPU_R80_R80 fptan
2591	IEMIMPL_FPU_R80_R80 fxtract
2592	IEMIMPL_FPU_R80_R80 fsincos
2593

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

source: vbox/trunk/src/VBox/VMM/VMMAll/IEMAllAImpl.asm@ 46992

Download in other formats: