3 |
|
* |
4 |
|
* Original 68040 JIT compiler for UAE, copyright 2000-2002 Bernd Meyer |
5 |
|
* |
6 |
< |
* Adaptation for Basilisk II and improvements, copyright 2000-2002 |
6 |
> |
* Adaptation for Basilisk II and improvements, copyright 2000-2004 |
7 |
|
* Gwenole Beauchesne |
8 |
|
* |
9 |
< |
* Basilisk II (C) 1997-2002 Christian Bauer |
9 |
> |
* Basilisk II (C) 1997-2004 Christian Bauer |
10 |
> |
* |
11 |
> |
* Portions related to CPU detection come from linux/arch/i386/kernel/setup.c |
12 |
|
* |
13 |
|
* This program is free software; you can redistribute it and/or modify |
14 |
|
* it under the terms of the GNU General Public License as published by |
42 |
|
#define EBP_INDEX 5 |
43 |
|
#define ESI_INDEX 6 |
44 |
|
#define EDI_INDEX 7 |
45 |
+ |
#if defined(__x86_64__) |
46 |
+ |
#define R8_INDEX 8 |
47 |
+ |
#define R9_INDEX 9 |
48 |
+ |
#define R10_INDEX 10 |
49 |
+ |
#define R11_INDEX 11 |
50 |
+ |
#define R12_INDEX 12 |
51 |
+ |
#define R13_INDEX 13 |
52 |
+ |
#define R14_INDEX 14 |
53 |
+ |
#define R15_INDEX 15 |
54 |
+ |
#endif |
55 |
|
|
56 |
|
/* The register in which subroutines return an integer return value */ |
57 |
< |
#define REG_RESULT 0 |
57 |
> |
#define REG_RESULT EAX_INDEX |
58 |
|
|
59 |
|
/* The registers subroutines take their first and second argument in */ |
60 |
|
#if defined( _MSC_VER ) && !defined( USE_NORMAL_CALLING_CONVENTION ) |
61 |
|
/* Handle the _fastcall parameters of ECX and EDX */ |
62 |
< |
#define REG_PAR1 1 |
63 |
< |
#define REG_PAR2 2 |
62 |
> |
#define REG_PAR1 ECX_INDEX |
63 |
> |
#define REG_PAR2 EDX_INDEX |
64 |
> |
#elif defined(__x86_64__) |
65 |
> |
#define REG_PAR1 EDI_INDEX |
66 |
> |
#define REG_PAR2 ESI_INDEX |
67 |
|
#else |
68 |
< |
#define REG_PAR1 0 |
69 |
< |
#define REG_PAR2 2 |
68 |
> |
#define REG_PAR1 EAX_INDEX |
69 |
> |
#define REG_PAR2 EDX_INDEX |
70 |
|
#endif |
71 |
|
|
72 |
< |
/* Three registers that are not used for any of the above */ |
58 |
< |
#define REG_NOPAR1 6 |
59 |
< |
#define REG_NOPAR2 5 |
60 |
< |
#define REG_NOPAR3 3 |
61 |
< |
|
62 |
< |
#define REG_PC_PRE 0 /* The register we use for preloading regs.pc_p */ |
72 |
> |
#define REG_PC_PRE EAX_INDEX /* The register we use for preloading regs.pc_p */ |
73 |
|
#if defined( _MSC_VER ) && !defined( USE_NORMAL_CALLING_CONVENTION ) |
74 |
< |
#define REG_PC_TMP 0 |
74 |
> |
#define REG_PC_TMP EAX_INDEX |
75 |
|
#else |
76 |
< |
#define REG_PC_TMP 1 /* Another register that is not the above */ |
76 |
> |
#define REG_PC_TMP ECX_INDEX /* Another register that is not the above */ |
77 |
|
#endif |
78 |
|
|
79 |
< |
#define SHIFTCOUNT_NREG 1 /* Register that can be used for shiftcount. |
79 |
> |
#define SHIFTCOUNT_NREG ECX_INDEX /* Register that can be used for shiftcount. |
80 |
|
-1 if any reg will do */ |
81 |
< |
#define MUL_NREG1 0 /* %eax will hold the low 32 bits after a 32x32 mul */ |
82 |
< |
#define MUL_NREG2 2 /* %edx will hold the high 32 bits */ |
81 |
> |
#define MUL_NREG1 EAX_INDEX /* %eax will hold the low 32 bits after a 32x32 mul */ |
82 |
> |
#define MUL_NREG2 EDX_INDEX /* %edx will hold the high 32 bits */ |
83 |
|
|
84 |
|
uae_s8 always_used[]={4,-1}; |
85 |
+ |
#if defined(__x86_64__) |
86 |
+ |
uae_s8 can_byte[]={0,1,2,3,5,6,7,8,9,10,11,12,13,14,15,-1}; |
87 |
+ |
uae_s8 can_word[]={0,1,2,3,5,6,7,8,9,10,11,12,13,14,15,-1}; |
88 |
+ |
#else |
89 |
|
uae_s8 can_byte[]={0,1,2,3,-1}; |
90 |
|
uae_s8 can_word[]={0,1,2,3,5,6,7,-1}; |
91 |
+ |
#endif |
92 |
|
|
93 |
+ |
#if USE_OPTIMIZED_CALLS |
94 |
+ |
/* Make sure interpretive core does not use cpuopti */ |
95 |
+ |
uae_u8 call_saved[]={0,0,0,1,1,1,1,1}; |
96 |
+ |
#error FIXME: code not ready |
97 |
+ |
#else |
98 |
|
/* cpuopti mutate instruction handlers to assume registers are saved |
99 |
|
by the caller */ |
100 |
< |
uae_u8 call_saved[]={0,0,0,0,1,0,0,0}; |
100 |
> |
uae_u8 call_saved[]={0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0}; |
101 |
> |
#endif |
102 |
|
|
103 |
|
/* This *should* be the same as call_saved. But: |
104 |
|
- We might not really know which registers are saved, and which aren't, |
107 |
|
- Special registers (such like the stack pointer) should not be "preserved" |
108 |
|
by pushing, even though they are "saved" across function calls |
109 |
|
*/ |
110 |
< |
uae_u8 need_to_preserve[]={1,1,1,1,0,1,1,1}; |
110 |
> |
#if defined(__x86_64__) |
111 |
> |
/* callee-saved registers as defined by Linux/x86_64 ABI: rbx, rbp, rsp, r12 - r15 */ |
112 |
> |
static const uae_u8 need_to_preserve[]={0,0,0,1,0,1,0,0,0,0,0,0,1,1,1,1}; |
113 |
> |
#else |
114 |
> |
static const uae_u8 need_to_preserve[]={1,1,1,1,0,1,1,1}; |
115 |
> |
#endif |
116 |
|
|
117 |
|
/* Whether classes of instructions do or don't clobber the native flags */ |
118 |
|
#define CLOBBER_MOV |
137 |
|
#define CLOBBER_TEST clobber_flags() |
138 |
|
#define CLOBBER_CL16 |
139 |
|
#define CLOBBER_CL8 |
140 |
+ |
#define CLOBBER_SE32 |
141 |
|
#define CLOBBER_SE16 |
142 |
|
#define CLOBBER_SE8 |
143 |
+ |
#define CLOBBER_ZE32 |
144 |
|
#define CLOBBER_ZE16 |
145 |
|
#define CLOBBER_ZE8 |
146 |
|
#define CLOBBER_SW16 clobber_flags() |
150 |
|
#define CLOBBER_BT clobber_flags() |
151 |
|
#define CLOBBER_BSF clobber_flags() |
152 |
|
|
153 |
+ |
/* FIXME: disabled until that's proofread. */ |
154 |
+ |
#if defined(__x86_64__) |
155 |
+ |
#define USE_NEW_RTASM 1 |
156 |
+ |
#endif |
157 |
+ |
|
158 |
+ |
#if USE_NEW_RTASM |
159 |
+ |
|
160 |
+ |
#if defined(__x86_64__) |
161 |
+ |
#define X86_TARGET_64BIT 1 |
162 |
+ |
#endif |
163 |
+ |
#define X86_FLAT_REGISTERS 0 |
164 |
+ |
#define X86_OPTIMIZE_ALU 1 |
165 |
+ |
#define X86_OPTIMIZE_ROTSHI 1 |
166 |
+ |
#include "codegen_x86.h" |
167 |
+ |
|
168 |
+ |
#define x86_emit_byte(B) emit_byte(B) |
169 |
+ |
#define x86_emit_word(W) emit_word(W) |
170 |
+ |
#define x86_emit_long(L) emit_long(L) |
171 |
+ |
#define x86_emit_quad(Q) emit_quad(Q) |
172 |
+ |
#define x86_get_target() get_target() |
173 |
+ |
#define x86_emit_failure(MSG) jit_fail(MSG, __FILE__, __LINE__, __FUNCTION__) |
174 |
+ |
|
175 |
+ |
static void jit_fail(const char *msg, const char *file, int line, const char *function) |
176 |
+ |
{ |
177 |
+ |
fprintf(stderr, "JIT failure in function %s from file %s at line %d: %s\n", |
178 |
+ |
function, file, line, msg); |
179 |
+ |
abort(); |
180 |
+ |
} |
181 |
+ |
|
182 |
+ |
LOWFUNC(NONE,WRITE,1,raw_push_l_r,(R4 r)) |
183 |
+ |
{ |
184 |
+ |
#if defined(__x86_64__) |
185 |
+ |
PUSHQr(r); |
186 |
+ |
#else |
187 |
+ |
PUSHLr(r); |
188 |
+ |
#endif |
189 |
+ |
} |
190 |
+ |
LENDFUNC(NONE,WRITE,1,raw_push_l_r,(R4 r)) |
191 |
+ |
|
192 |
+ |
LOWFUNC(NONE,READ,1,raw_pop_l_r,(R4 r)) |
193 |
+ |
{ |
194 |
+ |
#if defined(__x86_64__) |
195 |
+ |
POPQr(r); |
196 |
+ |
#else |
197 |
+ |
POPLr(r); |
198 |
+ |
#endif |
199 |
+ |
} |
200 |
+ |
LENDFUNC(NONE,READ,1,raw_pop_l_r,(R4 r)) |
201 |
+ |
|
202 |
+ |
LOWFUNC(WRITE,NONE,2,raw_bt_l_ri,(R4 r, IMM i)) |
203 |
+ |
{ |
204 |
+ |
BTLir(i, r); |
205 |
+ |
} |
206 |
+ |
LENDFUNC(WRITE,NONE,2,raw_bt_l_ri,(R4 r, IMM i)) |
207 |
+ |
|
208 |
+ |
LOWFUNC(WRITE,NONE,2,raw_bt_l_rr,(R4 r, R4 b)) |
209 |
+ |
{ |
210 |
+ |
BTLrr(b, r); |
211 |
+ |
} |
212 |
+ |
LENDFUNC(WRITE,NONE,2,raw_bt_l_rr,(R4 r, R4 b)) |
213 |
+ |
|
214 |
+ |
LOWFUNC(WRITE,NONE,2,raw_btc_l_ri,(RW4 r, IMM i)) |
215 |
+ |
{ |
216 |
+ |
BTCLir(i, r); |
217 |
+ |
} |
218 |
+ |
LENDFUNC(WRITE,NONE,2,raw_btc_l_ri,(RW4 r, IMM i)) |
219 |
+ |
|
220 |
+ |
LOWFUNC(WRITE,NONE,2,raw_btc_l_rr,(RW4 r, R4 b)) |
221 |
+ |
{ |
222 |
+ |
BTCLrr(b, r); |
223 |
+ |
} |
224 |
+ |
LENDFUNC(WRITE,NONE,2,raw_btc_l_rr,(RW4 r, R4 b)) |
225 |
+ |
|
226 |
+ |
LOWFUNC(WRITE,NONE,2,raw_btr_l_ri,(RW4 r, IMM i)) |
227 |
+ |
{ |
228 |
+ |
BTRLir(i, r); |
229 |
+ |
} |
230 |
+ |
LENDFUNC(WRITE,NONE,2,raw_btr_l_ri,(RW4 r, IMM i)) |
231 |
+ |
|
232 |
+ |
LOWFUNC(WRITE,NONE,2,raw_btr_l_rr,(RW4 r, R4 b)) |
233 |
+ |
{ |
234 |
+ |
BTRLrr(b, r); |
235 |
+ |
} |
236 |
+ |
LENDFUNC(WRITE,NONE,2,raw_btr_l_rr,(RW4 r, R4 b)) |
237 |
+ |
|
238 |
+ |
LOWFUNC(WRITE,NONE,2,raw_bts_l_ri,(RW4 r, IMM i)) |
239 |
+ |
{ |
240 |
+ |
BTSLir(i, r); |
241 |
+ |
} |
242 |
+ |
LENDFUNC(WRITE,NONE,2,raw_bts_l_ri,(RW4 r, IMM i)) |
243 |
+ |
|
244 |
+ |
LOWFUNC(WRITE,NONE,2,raw_bts_l_rr,(RW4 r, R4 b)) |
245 |
+ |
{ |
246 |
+ |
BTSLrr(b, r); |
247 |
+ |
} |
248 |
+ |
LENDFUNC(WRITE,NONE,2,raw_bts_l_rr,(RW4 r, R4 b)) |
249 |
+ |
|
250 |
+ |
LOWFUNC(WRITE,NONE,2,raw_sub_w_ri,(RW2 d, IMM i)) |
251 |
+ |
{ |
252 |
+ |
SUBWir(i, d); |
253 |
+ |
} |
254 |
+ |
LENDFUNC(WRITE,NONE,2,raw_sub_w_ri,(RW2 d, IMM i)) |
255 |
+ |
|
256 |
+ |
LOWFUNC(NONE,READ,2,raw_mov_l_rm,(W4 d, MEMR s)) |
257 |
+ |
{ |
258 |
+ |
MOVLmr(s, X86_NOREG, X86_NOREG, 1, d); |
259 |
+ |
} |
260 |
+ |
LENDFUNC(NONE,READ,2,raw_mov_l_rm,(W4 d, MEMR s)) |
261 |
+ |
|
262 |
+ |
LOWFUNC(NONE,WRITE,2,raw_mov_l_mi,(MEMW d, IMM s)) |
263 |
+ |
{ |
264 |
+ |
MOVLim(s, d, X86_NOREG, X86_NOREG, 1); |
265 |
+ |
} |
266 |
+ |
LENDFUNC(NONE,WRITE,2,raw_mov_l_mi,(MEMW d, IMM s)) |
267 |
+ |
|
268 |
+ |
LOWFUNC(NONE,WRITE,2,raw_mov_w_mi,(MEMW d, IMM s)) |
269 |
+ |
{ |
270 |
+ |
MOVWim(s, d, X86_NOREG, X86_NOREG, 1); |
271 |
+ |
} |
272 |
+ |
LENDFUNC(NONE,WRITE,2,raw_mov_w_mi,(MEMW d, IMM s)) |
273 |
+ |
|
274 |
+ |
LOWFUNC(NONE,WRITE,2,raw_mov_b_mi,(MEMW d, IMM s)) |
275 |
+ |
{ |
276 |
+ |
MOVBim(s, d, X86_NOREG, X86_NOREG, 1); |
277 |
+ |
} |
278 |
+ |
LENDFUNC(NONE,WRITE,2,raw_mov_b_mi,(MEMW d, IMM s)) |
279 |
+ |
|
280 |
+ |
LOWFUNC(WRITE,RMW,2,raw_rol_b_mi,(MEMRW d, IMM i)) |
281 |
+ |
{ |
282 |
+ |
ROLBim(i, d, X86_NOREG, X86_NOREG, 1); |
283 |
+ |
} |
284 |
+ |
LENDFUNC(WRITE,RMW,2,raw_rol_b_mi,(MEMRW d, IMM i)) |
285 |
+ |
|
286 |
+ |
LOWFUNC(WRITE,NONE,2,raw_rol_b_ri,(RW1 r, IMM i)) |
287 |
+ |
{ |
288 |
+ |
ROLBir(i, r); |
289 |
+ |
} |
290 |
+ |
LENDFUNC(WRITE,NONE,2,raw_rol_b_ri,(RW1 r, IMM i)) |
291 |
+ |
|
292 |
+ |
LOWFUNC(WRITE,NONE,2,raw_rol_w_ri,(RW2 r, IMM i)) |
293 |
+ |
{ |
294 |
+ |
ROLWir(i, r); |
295 |
+ |
} |
296 |
+ |
LENDFUNC(WRITE,NONE,2,raw_rol_w_ri,(RW2 r, IMM i)) |
297 |
+ |
|
298 |
+ |
LOWFUNC(WRITE,NONE,2,raw_rol_l_ri,(RW4 r, IMM i)) |
299 |
+ |
{ |
300 |
+ |
ROLLir(i, r); |
301 |
+ |
} |
302 |
+ |
LENDFUNC(WRITE,NONE,2,raw_rol_l_ri,(RW4 r, IMM i)) |
303 |
+ |
|
304 |
+ |
LOWFUNC(WRITE,NONE,2,raw_rol_l_rr,(RW4 d, R1 r)) |
305 |
+ |
{ |
306 |
+ |
ROLLrr(r, d); |
307 |
+ |
} |
308 |
+ |
LENDFUNC(WRITE,NONE,2,raw_rol_l_rr,(RW4 d, R1 r)) |
309 |
+ |
|
310 |
+ |
LOWFUNC(WRITE,NONE,2,raw_rol_w_rr,(RW2 d, R1 r)) |
311 |
+ |
{ |
312 |
+ |
ROLWrr(r, d); |
313 |
+ |
} |
314 |
+ |
LENDFUNC(WRITE,NONE,2,raw_rol_w_rr,(RW2 d, R1 r)) |
315 |
+ |
|
316 |
+ |
LOWFUNC(WRITE,NONE,2,raw_rol_b_rr,(RW1 d, R1 r)) |
317 |
+ |
{ |
318 |
+ |
ROLBrr(r, d); |
319 |
+ |
} |
320 |
+ |
LENDFUNC(WRITE,NONE,2,raw_rol_b_rr,(RW1 d, R1 r)) |
321 |
+ |
|
322 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shll_l_rr,(RW4 d, R1 r)) |
323 |
+ |
{ |
324 |
+ |
SHLLrr(r, d); |
325 |
+ |
} |
326 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shll_l_rr,(RW4 d, R1 r)) |
327 |
+ |
|
328 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shll_w_rr,(RW2 d, R1 r)) |
329 |
+ |
{ |
330 |
+ |
SHLWrr(r, d); |
331 |
+ |
} |
332 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shll_w_rr,(RW2 d, R1 r)) |
333 |
+ |
|
334 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shll_b_rr,(RW1 d, R1 r)) |
335 |
+ |
{ |
336 |
+ |
SHLBrr(r, d); |
337 |
+ |
} |
338 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shll_b_rr,(RW1 d, R1 r)) |
339 |
+ |
|
340 |
+ |
LOWFUNC(WRITE,NONE,2,raw_ror_b_ri,(RW1 r, IMM i)) |
341 |
+ |
{ |
342 |
+ |
RORBir(i, r); |
343 |
+ |
} |
344 |
+ |
LENDFUNC(WRITE,NONE,2,raw_ror_b_ri,(RW1 r, IMM i)) |
345 |
+ |
|
346 |
+ |
LOWFUNC(WRITE,NONE,2,raw_ror_w_ri,(RW2 r, IMM i)) |
347 |
+ |
{ |
348 |
+ |
RORWir(i, r); |
349 |
+ |
} |
350 |
+ |
LENDFUNC(WRITE,NONE,2,raw_ror_w_ri,(RW2 r, IMM i)) |
351 |
+ |
|
352 |
+ |
LOWFUNC(WRITE,READ,2,raw_or_l_rm,(RW4 d, MEMR s)) |
353 |
+ |
{ |
354 |
+ |
ORLmr(s, X86_NOREG, X86_NOREG, 1, d); |
355 |
+ |
} |
356 |
+ |
LENDFUNC(WRITE,READ,2,raw_or_l_rm,(RW4 d, MEMR s)) |
357 |
+ |
|
358 |
+ |
LOWFUNC(WRITE,NONE,2,raw_ror_l_ri,(RW4 r, IMM i)) |
359 |
+ |
{ |
360 |
+ |
RORLir(i, r); |
361 |
+ |
} |
362 |
+ |
LENDFUNC(WRITE,NONE,2,raw_ror_l_ri,(RW4 r, IMM i)) |
363 |
+ |
|
364 |
+ |
LOWFUNC(WRITE,NONE,2,raw_ror_l_rr,(RW4 d, R1 r)) |
365 |
+ |
{ |
366 |
+ |
RORLrr(r, d); |
367 |
+ |
} |
368 |
+ |
LENDFUNC(WRITE,NONE,2,raw_ror_l_rr,(RW4 d, R1 r)) |
369 |
+ |
|
370 |
+ |
LOWFUNC(WRITE,NONE,2,raw_ror_w_rr,(RW2 d, R1 r)) |
371 |
+ |
{ |
372 |
+ |
RORWrr(r, d); |
373 |
+ |
} |
374 |
+ |
LENDFUNC(WRITE,NONE,2,raw_ror_w_rr,(RW2 d, R1 r)) |
375 |
+ |
|
376 |
+ |
LOWFUNC(WRITE,NONE,2,raw_ror_b_rr,(RW1 d, R1 r)) |
377 |
+ |
{ |
378 |
+ |
RORBrr(r, d); |
379 |
+ |
} |
380 |
+ |
LENDFUNC(WRITE,NONE,2,raw_ror_b_rr,(RW1 d, R1 r)) |
381 |
+ |
|
382 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shrl_l_rr,(RW4 d, R1 r)) |
383 |
+ |
{ |
384 |
+ |
SHRLrr(r, d); |
385 |
+ |
} |
386 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shrl_l_rr,(RW4 d, R1 r)) |
387 |
+ |
|
388 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shrl_w_rr,(RW2 d, R1 r)) |
389 |
+ |
{ |
390 |
+ |
SHRWrr(r, d); |
391 |
+ |
} |
392 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shrl_w_rr,(RW2 d, R1 r)) |
393 |
+ |
|
394 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shrl_b_rr,(RW1 d, R1 r)) |
395 |
+ |
{ |
396 |
+ |
SHRBrr(r, d); |
397 |
+ |
} |
398 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shrl_b_rr,(RW1 d, R1 r)) |
399 |
+ |
|
400 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shra_l_rr,(RW4 d, R1 r)) |
401 |
+ |
{ |
402 |
+ |
SARLrr(r, d); |
403 |
+ |
} |
404 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shra_l_rr,(RW4 d, R1 r)) |
405 |
+ |
|
406 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shra_w_rr,(RW2 d, R1 r)) |
407 |
+ |
{ |
408 |
+ |
SARWrr(r, d); |
409 |
+ |
} |
410 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shra_w_rr,(RW2 d, R1 r)) |
411 |
+ |
|
412 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shra_b_rr,(RW1 d, R1 r)) |
413 |
+ |
{ |
414 |
+ |
SARBrr(r, d); |
415 |
+ |
} |
416 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shra_b_rr,(RW1 d, R1 r)) |
417 |
+ |
|
418 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shll_l_ri,(RW4 r, IMM i)) |
419 |
+ |
{ |
420 |
+ |
SHLLir(i, r); |
421 |
+ |
} |
422 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shll_l_ri,(RW4 r, IMM i)) |
423 |
+ |
|
424 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shll_w_ri,(RW2 r, IMM i)) |
425 |
+ |
{ |
426 |
+ |
SHLWir(i, r); |
427 |
+ |
} |
428 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shll_w_ri,(RW2 r, IMM i)) |
429 |
+ |
|
430 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shll_b_ri,(RW1 r, IMM i)) |
431 |
+ |
{ |
432 |
+ |
SHLBir(i, r); |
433 |
+ |
} |
434 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shll_b_ri,(RW1 r, IMM i)) |
435 |
+ |
|
436 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shrl_l_ri,(RW4 r, IMM i)) |
437 |
+ |
{ |
438 |
+ |
SHRLir(i, r); |
439 |
+ |
} |
440 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shrl_l_ri,(RW4 r, IMM i)) |
441 |
+ |
|
442 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shrl_w_ri,(RW2 r, IMM i)) |
443 |
+ |
{ |
444 |
+ |
SHRWir(i, r); |
445 |
+ |
} |
446 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shrl_w_ri,(RW2 r, IMM i)) |
447 |
+ |
|
448 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shrl_b_ri,(RW1 r, IMM i)) |
449 |
+ |
{ |
450 |
+ |
SHRBir(i, r); |
451 |
+ |
} |
452 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shrl_b_ri,(RW1 r, IMM i)) |
453 |
+ |
|
454 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shra_l_ri,(RW4 r, IMM i)) |
455 |
+ |
{ |
456 |
+ |
SARLir(i, r); |
457 |
+ |
} |
458 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shra_l_ri,(RW4 r, IMM i)) |
459 |
+ |
|
460 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shra_w_ri,(RW2 r, IMM i)) |
461 |
+ |
{ |
462 |
+ |
SARWir(i, r); |
463 |
+ |
} |
464 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shra_w_ri,(RW2 r, IMM i)) |
465 |
+ |
|
466 |
+ |
LOWFUNC(WRITE,NONE,2,raw_shra_b_ri,(RW1 r, IMM i)) |
467 |
+ |
{ |
468 |
+ |
SARBir(i, r); |
469 |
+ |
} |
470 |
+ |
LENDFUNC(WRITE,NONE,2,raw_shra_b_ri,(RW1 r, IMM i)) |
471 |
+ |
|
472 |
+ |
LOWFUNC(WRITE,NONE,1,raw_sahf,(R2 dummy_ah)) |
473 |
+ |
{ |
474 |
+ |
SAHF(); |
475 |
+ |
} |
476 |
+ |
LENDFUNC(WRITE,NONE,1,raw_sahf,(R2 dummy_ah)) |
477 |
+ |
|
478 |
+ |
LOWFUNC(NONE,NONE,1,raw_cpuid,(R4 dummy_eax)) |
479 |
+ |
{ |
480 |
+ |
CPUID(); |
481 |
+ |
} |
482 |
+ |
LENDFUNC(NONE,NONE,1,raw_cpuid,(R4 dummy_eax)) |
483 |
+ |
|
484 |
+ |
LOWFUNC(READ,NONE,1,raw_lahf,(W2 dummy_ah)) |
485 |
+ |
{ |
486 |
+ |
LAHF(); |
487 |
+ |
} |
488 |
+ |
LENDFUNC(READ,NONE,1,raw_lahf,(W2 dummy_ah)) |
489 |
+ |
|
490 |
+ |
LOWFUNC(READ,NONE,2,raw_setcc,(W1 d, IMM cc)) |
491 |
+ |
{ |
492 |
+ |
SETCCir(cc, d); |
493 |
+ |
} |
494 |
+ |
LENDFUNC(READ,NONE,2,raw_setcc,(W1 d, IMM cc)) |
495 |
+ |
|
496 |
+ |
LOWFUNC(READ,WRITE,2,raw_setcc_m,(MEMW d, IMM cc)) |
497 |
+ |
{ |
498 |
+ |
SETCCim(cc, d, X86_NOREG, X86_NOREG, 1); |
499 |
+ |
} |
500 |
+ |
LENDFUNC(READ,WRITE,2,raw_setcc_m,(MEMW d, IMM cc)) |
501 |
+ |
|
502 |
+ |
LOWFUNC(READ,NONE,3,raw_cmov_l_rr,(RW4 d, R4 s, IMM cc)) |
503 |
+ |
{ |
504 |
+ |
if (have_cmov) |
505 |
+ |
CMOVLrr(cc, s, d); |
506 |
+ |
else { /* replacement using branch and mov */ |
507 |
+ |
#if defined(__x86_64__) |
508 |
+ |
write_log("x86-64 implementations are bound to have CMOV!\n"); |
509 |
+ |
abort(); |
510 |
+ |
#endif |
511 |
+ |
JCCSii(cc^1, 2); |
512 |
+ |
MOVLrr(s, d); |
513 |
+ |
} |
514 |
+ |
} |
515 |
+ |
LENDFUNC(READ,NONE,3,raw_cmov_l_rr,(RW4 d, R4 s, IMM cc)) |
516 |
+ |
|
517 |
+ |
LOWFUNC(WRITE,NONE,2,raw_bsf_l_rr,(W4 d, R4 s)) |
518 |
+ |
{ |
519 |
+ |
BSFLrr(s, d); |
520 |
+ |
} |
521 |
+ |
LENDFUNC(WRITE,NONE,2,raw_bsf_l_rr,(W4 d, R4 s)) |
522 |
+ |
|
523 |
+ |
LOWFUNC(NONE,NONE,2,raw_sign_extend_32_rr,(W4 d, R4 s)) |
524 |
+ |
{ |
525 |
+ |
MOVSLQrr(s, d); |
526 |
+ |
} |
527 |
+ |
LENDFUNC(NONE,NONE,2,raw_sign_extend_32_rr,(W4 d, R4 s)) |
528 |
+ |
|
529 |
+ |
LOWFUNC(NONE,NONE,2,raw_sign_extend_16_rr,(W4 d, R2 s)) |
530 |
+ |
{ |
531 |
+ |
MOVSWLrr(s, d); |
532 |
+ |
} |
533 |
+ |
LENDFUNC(NONE,NONE,2,raw_sign_extend_16_rr,(W4 d, R2 s)) |
534 |
+ |
|
535 |
+ |
LOWFUNC(NONE,NONE,2,raw_sign_extend_8_rr,(W4 d, R1 s)) |
536 |
+ |
{ |
537 |
+ |
MOVSBLrr(s, d); |
538 |
+ |
} |
539 |
+ |
LENDFUNC(NONE,NONE,2,raw_sign_extend_8_rr,(W4 d, R1 s)) |
540 |
+ |
|
541 |
+ |
LOWFUNC(NONE,NONE,2,raw_zero_extend_16_rr,(W4 d, R2 s)) |
542 |
+ |
{ |
543 |
+ |
MOVZWLrr(s, d); |
544 |
+ |
} |
545 |
+ |
LENDFUNC(NONE,NONE,2,raw_zero_extend_16_rr,(W4 d, R2 s)) |
546 |
+ |
|
547 |
+ |
LOWFUNC(NONE,NONE,2,raw_zero_extend_8_rr,(W4 d, R1 s)) |
548 |
+ |
{ |
549 |
+ |
MOVZBLrr(s, d); |
550 |
+ |
} |
551 |
+ |
LENDFUNC(NONE,NONE,2,raw_zero_extend_8_rr,(W4 d, R1 s)) |
552 |
+ |
|
553 |
+ |
LOWFUNC(NONE,NONE,2,raw_imul_32_32,(RW4 d, R4 s)) |
554 |
+ |
{ |
555 |
+ |
IMULLrr(s, d); |
556 |
+ |
} |
557 |
+ |
LENDFUNC(NONE,NONE,2,raw_imul_32_32,(RW4 d, R4 s)) |
558 |
+ |
|
559 |
+ |
LOWFUNC(NONE,NONE,2,raw_imul_64_32,(RW4 d, RW4 s)) |
560 |
+ |
{ |
561 |
+ |
if (d!=MUL_NREG1 || s!=MUL_NREG2) { |
562 |
+ |
write_log("Bad register in IMUL: d=%d, s=%d\n",d,s); |
563 |
+ |
abort(); |
564 |
+ |
} |
565 |
+ |
IMULLr(s); |
566 |
+ |
} |
567 |
+ |
LENDFUNC(NONE,NONE,2,raw_imul_64_32,(RW4 d, RW4 s)) |
568 |
+ |
|
569 |
+ |
LOWFUNC(NONE,NONE,2,raw_mul_64_32,(RW4 d, RW4 s)) |
570 |
+ |
{ |
571 |
+ |
if (d!=MUL_NREG1 || s!=MUL_NREG2) { |
572 |
+ |
write_log("Bad register in MUL: d=%d, s=%d\n",d,s); |
573 |
+ |
abort(); |
574 |
+ |
} |
575 |
+ |
MULLr(s); |
576 |
+ |
} |
577 |
+ |
LENDFUNC(NONE,NONE,2,raw_mul_64_32,(RW4 d, RW4 s)) |
578 |
+ |
|
579 |
+ |
LOWFUNC(NONE,NONE,2,raw_mul_32_32,(RW4 d, R4 s)) |
580 |
+ |
{ |
581 |
+ |
abort(); /* %^$&%^$%#^ x86! */ |
582 |
+ |
} |
583 |
+ |
LENDFUNC(NONE,NONE,2,raw_mul_32_32,(RW4 d, R4 s)) |
584 |
+ |
|
585 |
+ |
LOWFUNC(NONE,NONE,2,raw_mov_b_rr,(W1 d, R1 s)) |
586 |
+ |
{ |
587 |
+ |
MOVBrr(s, d); |
588 |
+ |
} |
589 |
+ |
LENDFUNC(NONE,NONE,2,raw_mov_b_rr,(W1 d, R1 s)) |
590 |
+ |
|
591 |
+ |
LOWFUNC(NONE,NONE,2,raw_mov_w_rr,(W2 d, R2 s)) |
592 |
+ |
{ |
593 |
+ |
MOVWrr(s, d); |
594 |
+ |
} |
595 |
+ |
LENDFUNC(NONE,NONE,2,raw_mov_w_rr,(W2 d, R2 s)) |
596 |
+ |
|
597 |
+ |
LOWFUNC(NONE,READ,4,raw_mov_l_rrm_indexed,(W4 d,R4 baser, R4 index, IMM factor)) |
598 |
+ |
{ |
599 |
+ |
MOVLmr(0, baser, index, factor, d); |
600 |
+ |
} |
601 |
+ |
LENDFUNC(NONE,READ,4,raw_mov_l_rrm_indexed,(W4 d,R4 baser, R4 index, IMM factor)) |
602 |
+ |
|
603 |
+ |
LOWFUNC(NONE,READ,4,raw_mov_w_rrm_indexed,(W2 d, R4 baser, R4 index, IMM factor)) |
604 |
+ |
{ |
605 |
+ |
MOVWmr(0, baser, index, factor, d); |
606 |
+ |
} |
607 |
+ |
LENDFUNC(NONE,READ,4,raw_mov_w_rrm_indexed,(W2 d, R4 baser, R4 index, IMM factor)) |
608 |
+ |
|
609 |
+ |
LOWFUNC(NONE,READ,4,raw_mov_b_rrm_indexed,(W1 d, R4 baser, R4 index, IMM factor)) |
610 |
+ |
{ |
611 |
+ |
MOVBmr(0, baser, index, factor, d); |
612 |
+ |
} |
613 |
+ |
LENDFUNC(NONE,READ,4,raw_mov_b_rrm_indexed,(W1 d, R4 baser, R4 index, IMM factor)) |
614 |
+ |
|
615 |
+ |
LOWFUNC(NONE,WRITE,4,raw_mov_l_mrr_indexed,(R4 baser, R4 index, IMM factor, R4 s)) |
616 |
+ |
{ |
617 |
+ |
MOVLrm(s, 0, baser, index, factor); |
618 |
+ |
} |
619 |
+ |
LENDFUNC(NONE,WRITE,4,raw_mov_l_mrr_indexed,(R4 baser, R4 index, IMM factor, R4 s)) |
620 |
+ |
|
621 |
+ |
LOWFUNC(NONE,WRITE,4,raw_mov_w_mrr_indexed,(R4 baser, R4 index, IMM factor, R2 s)) |
622 |
+ |
{ |
623 |
+ |
MOVWrm(s, 0, baser, index, factor); |
624 |
+ |
} |
625 |
+ |
LENDFUNC(NONE,WRITE,4,raw_mov_w_mrr_indexed,(R4 baser, R4 index, IMM factor, R2 s)) |
626 |
+ |
|
627 |
+ |
LOWFUNC(NONE,WRITE,4,raw_mov_b_mrr_indexed,(R4 baser, R4 index, IMM factor, R1 s)) |
628 |
+ |
{ |
629 |
+ |
MOVBrm(s, 0, baser, index, factor); |
630 |
+ |
} |
631 |
+ |
LENDFUNC(NONE,WRITE,4,raw_mov_b_mrr_indexed,(R4 baser, R4 index, IMM factor, R1 s)) |
632 |
+ |
|
633 |
+ |
LOWFUNC(NONE,WRITE,5,raw_mov_l_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R4 s)) |
634 |
+ |
{ |
635 |
+ |
MOVLrm(s, base, baser, index, factor); |
636 |
+ |
} |
637 |
+ |
LENDFUNC(NONE,WRITE,5,raw_mov_l_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R4 s)) |
638 |
+ |
|
639 |
+ |
LOWFUNC(NONE,WRITE,5,raw_mov_w_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R2 s)) |
640 |
+ |
{ |
641 |
+ |
MOVWrm(s, base, baser, index, factor); |
642 |
+ |
} |
643 |
+ |
LENDFUNC(NONE,WRITE,5,raw_mov_w_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R2 s)) |
644 |
+ |
|
645 |
+ |
LOWFUNC(NONE,WRITE,5,raw_mov_b_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R1 s)) |
646 |
+ |
{ |
647 |
+ |
MOVBrm(s, base, baser, index, factor); |
648 |
+ |
} |
649 |
+ |
LENDFUNC(NONE,WRITE,5,raw_mov_b_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R1 s)) |
650 |
+ |
|
651 |
+ |
LOWFUNC(NONE,READ,5,raw_mov_l_brrm_indexed,(W4 d, IMM base, R4 baser, R4 index, IMM factor)) |
652 |
+ |
{ |
653 |
+ |
MOVLmr(base, baser, index, factor, d); |
654 |
+ |
} |
655 |
+ |
LENDFUNC(NONE,READ,5,raw_mov_l_brrm_indexed,(W4 d, IMM base, R4 baser, R4 index, IMM factor)) |
656 |
+ |
|
657 |
+ |
LOWFUNC(NONE,READ,5,raw_mov_w_brrm_indexed,(W2 d, IMM base, R4 baser, R4 index, IMM factor)) |
658 |
+ |
{ |
659 |
+ |
MOVWmr(base, baser, index, factor, d); |
660 |
+ |
} |
661 |
+ |
LENDFUNC(NONE,READ,5,raw_mov_w_brrm_indexed,(W2 d, IMM base, R4 baser, R4 index, IMM factor)) |
662 |
+ |
|
663 |
+ |
LOWFUNC(NONE,READ,5,raw_mov_b_brrm_indexed,(W1 d, IMM base, R4 baser, R4 index, IMM factor)) |
664 |
+ |
{ |
665 |
+ |
MOVBmr(base, baser, index, factor, d); |
666 |
+ |
} |
667 |
+ |
LENDFUNC(NONE,READ,5,raw_mov_b_brrm_indexed,(W1 d, IMM base, R4 baser, R4 index, IMM factor)) |
668 |
+ |
|
669 |
+ |
LOWFUNC(NONE,READ,4,raw_mov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor)) |
670 |
+ |
{ |
671 |
+ |
MOVLmr(base, X86_NOREG, index, factor, d); |
672 |
+ |
} |
673 |
+ |
LENDFUNC(NONE,READ,4,raw_mov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor)) |
674 |
+ |
|
675 |
+ |
LOWFUNC(NONE,READ,5,raw_cmov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor, IMM cond)) |
676 |
+ |
{ |
677 |
+ |
if (have_cmov) |
678 |
+ |
CMOVLmr(cond, base, X86_NOREG, index, factor, d); |
679 |
+ |
else { /* replacement using branch and mov */ |
680 |
+ |
#if defined(__x86_64__) |
681 |
+ |
write_log("x86-64 implementations are bound to have CMOV!\n"); |
682 |
+ |
abort(); |
683 |
+ |
#endif |
684 |
+ |
JCCSii(cond^1, 7); |
685 |
+ |
MOVLmr(base, X86_NOREG, index, factor, d); |
686 |
+ |
} |
687 |
+ |
} |
688 |
+ |
LENDFUNC(NONE,READ,5,raw_cmov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor, IMM cond)) |
689 |
+ |
|
690 |
+ |
LOWFUNC(NONE,READ,3,raw_cmov_l_rm,(W4 d, IMM mem, IMM cond)) |
691 |
+ |
{ |
692 |
+ |
if (have_cmov) |
693 |
+ |
CMOVLmr(cond, mem, X86_NOREG, X86_NOREG, 1, d); |
694 |
+ |
else { /* replacement using branch and mov */ |
695 |
+ |
#if defined(__x86_64__) |
696 |
+ |
write_log("x86-64 implementations are bound to have CMOV!\n"); |
697 |
+ |
abort(); |
698 |
+ |
#endif |
699 |
+ |
JCCSii(cond^1, 6); |
700 |
+ |
MOVLmr(mem, X86_NOREG, X86_NOREG, 1, d); |
701 |
+ |
} |
702 |
+ |
} |
703 |
+ |
LENDFUNC(NONE,READ,3,raw_cmov_l_rm,(W4 d, IMM mem, IMM cond)) |
704 |
+ |
|
705 |
+ |
LOWFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
706 |
+ |
{ |
707 |
+ |
MOVLmr(offset, s, X86_NOREG, 1, d); |
708 |
+ |
} |
709 |
+ |
LENDFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
710 |
+ |
|
711 |
+ |
LOWFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
712 |
+ |
{ |
713 |
+ |
MOVWmr(offset, s, X86_NOREG, 1, d); |
714 |
+ |
} |
715 |
+ |
LENDFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
716 |
+ |
|
717 |
+ |
LOWFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
718 |
+ |
{ |
719 |
+ |
MOVBmr(offset, s, X86_NOREG, 1, d); |
720 |
+ |
} |
721 |
+ |
LENDFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
722 |
+ |
|
723 |
+ |
LOWFUNC(NONE,READ,3,raw_mov_l_brR,(W4 d, R4 s, IMM offset)) |
724 |
+ |
{ |
725 |
+ |
MOVLmr(offset, s, X86_NOREG, 1, d); |
726 |
+ |
} |
727 |
+ |
LENDFUNC(NONE,READ,3,raw_mov_l_brR,(W4 d, R4 s, IMM offset)) |
728 |
+ |
|
729 |
+ |
LOWFUNC(NONE,READ,3,raw_mov_w_brR,(W2 d, R4 s, IMM offset)) |
730 |
+ |
{ |
731 |
+ |
MOVWmr(offset, s, X86_NOREG, 1, d); |
732 |
+ |
} |
733 |
+ |
LENDFUNC(NONE,READ,3,raw_mov_w_brR,(W2 d, R4 s, IMM offset)) |
734 |
+ |
|
735 |
+ |
LOWFUNC(NONE,READ,3,raw_mov_b_brR,(W1 d, R4 s, IMM offset)) |
736 |
+ |
{ |
737 |
+ |
MOVBmr(offset, s, X86_NOREG, 1, d); |
738 |
+ |
} |
739 |
+ |
LENDFUNC(NONE,READ,3,raw_mov_b_brR,(W1 d, R4 s, IMM offset)) |
740 |
+ |
|
741 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
742 |
+ |
{ |
743 |
+ |
MOVLim(i, offset, d, X86_NOREG, 1); |
744 |
+ |
} |
745 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
746 |
+ |
|
747 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
748 |
+ |
{ |
749 |
+ |
MOVWim(i, offset, d, X86_NOREG, 1); |
750 |
+ |
} |
751 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
752 |
+ |
|
753 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
754 |
+ |
{ |
755 |
+ |
MOVBim(i, offset, d, X86_NOREG, 1); |
756 |
+ |
} |
757 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
758 |
+ |
|
759 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
760 |
+ |
{ |
761 |
+ |
MOVLrm(s, offset, d, X86_NOREG, 1); |
762 |
+ |
} |
763 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
764 |
+ |
|
765 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
766 |
+ |
{ |
767 |
+ |
MOVWrm(s, offset, d, X86_NOREG, 1); |
768 |
+ |
} |
769 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
770 |
+ |
|
771 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
772 |
+ |
{ |
773 |
+ |
MOVBrm(s, offset, d, X86_NOREG, 1); |
774 |
+ |
} |
775 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
776 |
+ |
|
777 |
+ |
LOWFUNC(NONE,NONE,3,raw_lea_l_brr,(W4 d, R4 s, IMM offset)) |
778 |
+ |
{ |
779 |
+ |
LEALmr(offset, s, X86_NOREG, 1, d); |
780 |
+ |
} |
781 |
+ |
LENDFUNC(NONE,NONE,3,raw_lea_l_brr,(W4 d, R4 s, IMM offset)) |
782 |
+ |
|
783 |
+ |
LOWFUNC(NONE,NONE,5,raw_lea_l_brr_indexed,(W4 d, R4 s, R4 index, IMM factor, IMM offset)) |
784 |
+ |
{ |
785 |
+ |
LEALmr(offset, s, index, factor, d); |
786 |
+ |
} |
787 |
+ |
LENDFUNC(NONE,NONE,5,raw_lea_l_brr_indexed,(W4 d, R4 s, R4 index, IMM factor, IMM offset)) |
788 |
+ |
|
789 |
+ |
LOWFUNC(NONE,NONE,4,raw_lea_l_rr_indexed,(W4 d, R4 s, R4 index, IMM factor)) |
790 |
+ |
{ |
791 |
+ |
LEALmr(0, s, index, factor, d); |
792 |
+ |
} |
793 |
+ |
LENDFUNC(NONE,NONE,4,raw_lea_l_rr_indexed,(W4 d, R4 s, R4 index, IMM factor)) |
794 |
+ |
|
795 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_l_bRr,(R4 d, R4 s, IMM offset)) |
796 |
+ |
{ |
797 |
+ |
MOVLrm(s, offset, d, X86_NOREG, 1); |
798 |
+ |
} |
799 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_l_bRr,(R4 d, R4 s, IMM offset)) |
800 |
+ |
|
801 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_w_bRr,(R4 d, R2 s, IMM offset)) |
802 |
+ |
{ |
803 |
+ |
MOVWrm(s, offset, d, X86_NOREG, 1); |
804 |
+ |
} |
805 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_w_bRr,(R4 d, R2 s, IMM offset)) |
806 |
+ |
|
807 |
+ |
LOWFUNC(NONE,WRITE,3,raw_mov_b_bRr,(R4 d, R1 s, IMM offset)) |
808 |
+ |
{ |
809 |
+ |
MOVBrm(s, offset, d, X86_NOREG, 1); |
810 |
+ |
} |
811 |
+ |
LENDFUNC(NONE,WRITE,3,raw_mov_b_bRr,(R4 d, R1 s, IMM offset)) |
812 |
+ |
|
813 |
+ |
LOWFUNC(NONE,NONE,1,raw_bswap_32,(RW4 r)) |
814 |
+ |
{ |
815 |
+ |
BSWAPLr(r); |
816 |
+ |
} |
817 |
+ |
LENDFUNC(NONE,NONE,1,raw_bswap_32,(RW4 r)) |
818 |
+ |
|
819 |
+ |
LOWFUNC(WRITE,NONE,1,raw_bswap_16,(RW2 r)) |
820 |
+ |
{ |
821 |
+ |
ROLWir(8, r); |
822 |
+ |
} |
823 |
+ |
LENDFUNC(WRITE,NONE,1,raw_bswap_16,(RW2 r)) |
824 |
+ |
|
825 |
+ |
LOWFUNC(NONE,NONE,2,raw_mov_l_rr,(W4 d, R4 s)) |
826 |
+ |
{ |
827 |
+ |
MOVLrr(s, d); |
828 |
+ |
} |
829 |
+ |
LENDFUNC(NONE,NONE,2,raw_mov_l_rr,(W4 d, R4 s)) |
830 |
+ |
|
831 |
+ |
LOWFUNC(NONE,WRITE,2,raw_mov_l_mr,(IMM d, R4 s)) |
832 |
+ |
{ |
833 |
+ |
MOVLrm(s, d, X86_NOREG, X86_NOREG, 1); |
834 |
+ |
} |
835 |
+ |
LENDFUNC(NONE,WRITE,2,raw_mov_l_mr,(IMM d, R4 s)) |
836 |
+ |
|
837 |
+ |
LOWFUNC(NONE,WRITE,2,raw_mov_w_mr,(IMM d, R2 s)) |
838 |
+ |
{ |
839 |
+ |
MOVWrm(s, d, X86_NOREG, X86_NOREG, 1); |
840 |
+ |
} |
841 |
+ |
LENDFUNC(NONE,WRITE,2,raw_mov_w_mr,(IMM d, R2 s)) |
842 |
+ |
|
843 |
+ |
LOWFUNC(NONE,READ,2,raw_mov_w_rm,(W2 d, IMM s)) |
844 |
+ |
{ |
845 |
+ |
MOVWmr(s, X86_NOREG, X86_NOREG, 1, d); |
846 |
+ |
} |
847 |
+ |
LENDFUNC(NONE,READ,2,raw_mov_w_rm,(W2 d, IMM s)) |
848 |
+ |
|
849 |
+ |
LOWFUNC(NONE,WRITE,2,raw_mov_b_mr,(IMM d, R1 s)) |
850 |
+ |
{ |
851 |
+ |
MOVBrm(s, d, X86_NOREG, X86_NOREG, 1); |
852 |
+ |
} |
853 |
+ |
LENDFUNC(NONE,WRITE,2,raw_mov_b_mr,(IMM d, R1 s)) |
854 |
+ |
|
855 |
+ |
LOWFUNC(NONE,READ,2,raw_mov_b_rm,(W1 d, IMM s)) |
856 |
+ |
{ |
857 |
+ |
MOVBmr(s, X86_NOREG, X86_NOREG, 1, d); |
858 |
+ |
} |
859 |
+ |
LENDFUNC(NONE,READ,2,raw_mov_b_rm,(W1 d, IMM s)) |
860 |
+ |
|
861 |
+ |
LOWFUNC(NONE,NONE,2,raw_mov_l_ri,(W4 d, IMM s)) |
862 |
+ |
{ |
863 |
+ |
MOVLir(s, d); |
864 |
+ |
} |
865 |
+ |
LENDFUNC(NONE,NONE,2,raw_mov_l_ri,(W4 d, IMM s)) |
866 |
+ |
|
867 |
+ |
LOWFUNC(NONE,NONE,2,raw_mov_w_ri,(W2 d, IMM s)) |
868 |
+ |
{ |
869 |
+ |
MOVWir(s, d); |
870 |
+ |
} |
871 |
+ |
LENDFUNC(NONE,NONE,2,raw_mov_w_ri,(W2 d, IMM s)) |
872 |
+ |
|
873 |
+ |
LOWFUNC(NONE,NONE,2,raw_mov_b_ri,(W1 d, IMM s)) |
874 |
+ |
{ |
875 |
+ |
MOVBir(s, d); |
876 |
+ |
} |
877 |
+ |
LENDFUNC(NONE,NONE,2,raw_mov_b_ri,(W1 d, IMM s)) |
878 |
+ |
|
879 |
+ |
LOWFUNC(RMW,RMW,2,raw_adc_l_mi,(MEMRW d, IMM s)) |
880 |
+ |
{ |
881 |
+ |
ADCLim(s, d, X86_NOREG, X86_NOREG, 1); |
882 |
+ |
} |
883 |
+ |
LENDFUNC(RMW,RMW,2,raw_adc_l_mi,(MEMRW d, IMM s)) |
884 |
+ |
|
885 |
+ |
LOWFUNC(WRITE,RMW,2,raw_add_l_mi,(IMM d, IMM s)) |
886 |
+ |
{ |
887 |
+ |
ADDLim(s, d, X86_NOREG, X86_NOREG, 1); |
888 |
+ |
} |
889 |
+ |
LENDFUNC(WRITE,RMW,2,raw_add_l_mi,(IMM d, IMM s)) |
890 |
+ |
|
891 |
+ |
LOWFUNC(WRITE,RMW,2,raw_add_w_mi,(IMM d, IMM s)) |
892 |
+ |
{ |
893 |
+ |
ADDWim(s, d, X86_NOREG, X86_NOREG, 1); |
894 |
+ |
} |
895 |
+ |
LENDFUNC(WRITE,RMW,2,raw_add_w_mi,(IMM d, IMM s)) |
896 |
+ |
|
897 |
+ |
LOWFUNC(WRITE,RMW,2,raw_add_b_mi,(IMM d, IMM s)) |
898 |
+ |
{ |
899 |
+ |
ADDBim(s, d, X86_NOREG, X86_NOREG, 1); |
900 |
+ |
} |
901 |
+ |
LENDFUNC(WRITE,RMW,2,raw_add_b_mi,(IMM d, IMM s)) |
902 |
+ |
|
903 |
+ |
LOWFUNC(WRITE,NONE,2,raw_test_l_ri,(R4 d, IMM i)) |
904 |
+ |
{ |
905 |
+ |
TESTLir(i, d); |
906 |
+ |
} |
907 |
+ |
LENDFUNC(WRITE,NONE,2,raw_test_l_ri,(R4 d, IMM i)) |
908 |
+ |
|
909 |
+ |
LOWFUNC(WRITE,NONE,2,raw_test_l_rr,(R4 d, R4 s)) |
910 |
+ |
{ |
911 |
+ |
TESTLrr(s, d); |
912 |
+ |
} |
913 |
+ |
LENDFUNC(WRITE,NONE,2,raw_test_l_rr,(R4 d, R4 s)) |
914 |
+ |
|
915 |
+ |
LOWFUNC(WRITE,NONE,2,raw_test_w_rr,(R2 d, R2 s)) |
916 |
+ |
{ |
917 |
+ |
TESTWrr(s, d); |
918 |
+ |
} |
919 |
+ |
LENDFUNC(WRITE,NONE,2,raw_test_w_rr,(R2 d, R2 s)) |
920 |
+ |
|
921 |
+ |
LOWFUNC(WRITE,NONE,2,raw_test_b_rr,(R1 d, R1 s)) |
922 |
+ |
{ |
923 |
+ |
TESTBrr(s, d); |
924 |
+ |
} |
925 |
+ |
LENDFUNC(WRITE,NONE,2,raw_test_b_rr,(R1 d, R1 s)) |
926 |
+ |
|
927 |
+ |
LOWFUNC(WRITE,NONE,2,raw_and_l_ri,(RW4 d, IMM i)) |
928 |
+ |
{ |
929 |
+ |
ANDLir(i, d); |
930 |
+ |
} |
931 |
+ |
LENDFUNC(WRITE,NONE,2,raw_and_l_ri,(RW4 d, IMM i)) |
932 |
+ |
|
933 |
+ |
LOWFUNC(WRITE,NONE,2,raw_and_w_ri,(RW2 d, IMM i)) |
934 |
+ |
{ |
935 |
+ |
ANDWir(i, d); |
936 |
+ |
} |
937 |
+ |
LENDFUNC(WRITE,NONE,2,raw_and_w_ri,(RW2 d, IMM i)) |
938 |
+ |
|
939 |
+ |
LOWFUNC(WRITE,NONE,2,raw_and_l,(RW4 d, R4 s)) |
940 |
+ |
{ |
941 |
+ |
ANDLrr(s, d); |
942 |
+ |
} |
943 |
+ |
LENDFUNC(WRITE,NONE,2,raw_and_l,(RW4 d, R4 s)) |
944 |
+ |
|
945 |
+ |
LOWFUNC(WRITE,NONE,2,raw_and_w,(RW2 d, R2 s)) |
946 |
+ |
{ |
947 |
+ |
ANDWrr(s, d); |
948 |
+ |
} |
949 |
+ |
LENDFUNC(WRITE,NONE,2,raw_and_w,(RW2 d, R2 s)) |
950 |
+ |
|
951 |
+ |
LOWFUNC(WRITE,NONE,2,raw_and_b,(RW1 d, R1 s)) |
952 |
+ |
{ |
953 |
+ |
ANDBrr(s, d); |
954 |
+ |
} |
955 |
+ |
LENDFUNC(WRITE,NONE,2,raw_and_b,(RW1 d, R1 s)) |
956 |
+ |
|
957 |
+ |
LOWFUNC(WRITE,NONE,2,raw_or_l_ri,(RW4 d, IMM i)) |
958 |
+ |
{ |
959 |
+ |
ORLir(i, d); |
960 |
+ |
} |
961 |
+ |
LENDFUNC(WRITE,NONE,2,raw_or_l_ri,(RW4 d, IMM i)) |
962 |
+ |
|
963 |
+ |
LOWFUNC(WRITE,NONE,2,raw_or_l,(RW4 d, R4 s)) |
964 |
+ |
{ |
965 |
+ |
ORLrr(s, d); |
966 |
+ |
} |
967 |
+ |
LENDFUNC(WRITE,NONE,2,raw_or_l,(RW4 d, R4 s)) |
968 |
+ |
|
969 |
+ |
LOWFUNC(WRITE,NONE,2,raw_or_w,(RW2 d, R2 s)) |
970 |
+ |
{ |
971 |
+ |
ORWrr(s, d); |
972 |
+ |
} |
973 |
+ |
LENDFUNC(WRITE,NONE,2,raw_or_w,(RW2 d, R2 s)) |
974 |
+ |
|
975 |
+ |
LOWFUNC(WRITE,NONE,2,raw_or_b,(RW1 d, R1 s)) |
976 |
+ |
{ |
977 |
+ |
ORBrr(s, d); |
978 |
+ |
} |
979 |
+ |
LENDFUNC(WRITE,NONE,2,raw_or_b,(RW1 d, R1 s)) |
980 |
+ |
|
981 |
+ |
LOWFUNC(RMW,NONE,2,raw_adc_l,(RW4 d, R4 s)) |
982 |
+ |
{ |
983 |
+ |
ADCLrr(s, d); |
984 |
+ |
} |
985 |
+ |
LENDFUNC(RMW,NONE,2,raw_adc_l,(RW4 d, R4 s)) |
986 |
+ |
|
987 |
+ |
LOWFUNC(RMW,NONE,2,raw_adc_w,(RW2 d, R2 s)) |
988 |
+ |
{ |
989 |
+ |
ADCWrr(s, d); |
990 |
+ |
} |
991 |
+ |
LENDFUNC(RMW,NONE,2,raw_adc_w,(RW2 d, R2 s)) |
992 |
+ |
|
993 |
+ |
LOWFUNC(RMW,NONE,2,raw_adc_b,(RW1 d, R1 s)) |
994 |
+ |
{ |
995 |
+ |
ADCBrr(s, d); |
996 |
+ |
} |
997 |
+ |
LENDFUNC(RMW,NONE,2,raw_adc_b,(RW1 d, R1 s)) |
998 |
+ |
|
999 |
+ |
LOWFUNC(WRITE,NONE,2,raw_add_l,(RW4 d, R4 s)) |
1000 |
+ |
{ |
1001 |
+ |
ADDLrr(s, d); |
1002 |
+ |
} |
1003 |
+ |
LENDFUNC(WRITE,NONE,2,raw_add_l,(RW4 d, R4 s)) |
1004 |
+ |
|
1005 |
+ |
LOWFUNC(WRITE,NONE,2,raw_add_w,(RW2 d, R2 s)) |
1006 |
+ |
{ |
1007 |
+ |
ADDWrr(s, d); |
1008 |
+ |
} |
1009 |
+ |
LENDFUNC(WRITE,NONE,2,raw_add_w,(RW2 d, R2 s)) |
1010 |
+ |
|
1011 |
+ |
LOWFUNC(WRITE,NONE,2,raw_add_b,(RW1 d, R1 s)) |
1012 |
+ |
{ |
1013 |
+ |
ADDBrr(s, d); |
1014 |
+ |
} |
1015 |
+ |
LENDFUNC(WRITE,NONE,2,raw_add_b,(RW1 d, R1 s)) |
1016 |
+ |
|
1017 |
+ |
LOWFUNC(WRITE,NONE,2,raw_sub_l_ri,(RW4 d, IMM i)) |
1018 |
+ |
{ |
1019 |
+ |
SUBLir(i, d); |
1020 |
+ |
} |
1021 |
+ |
LENDFUNC(WRITE,NONE,2,raw_sub_l_ri,(RW4 d, IMM i)) |
1022 |
+ |
|
1023 |
+ |
LOWFUNC(WRITE,NONE,2,raw_sub_b_ri,(RW1 d, IMM i)) |
1024 |
+ |
{ |
1025 |
+ |
SUBBir(i, d); |
1026 |
+ |
} |
1027 |
+ |
LENDFUNC(WRITE,NONE,2,raw_sub_b_ri,(RW1 d, IMM i)) |
1028 |
+ |
|
1029 |
+ |
LOWFUNC(WRITE,NONE,2,raw_add_l_ri,(RW4 d, IMM i)) |
1030 |
+ |
{ |
1031 |
+ |
ADDLir(i, d); |
1032 |
+ |
} |
1033 |
+ |
LENDFUNC(WRITE,NONE,2,raw_add_l_ri,(RW4 d, IMM i)) |
1034 |
+ |
|
1035 |
+ |
LOWFUNC(WRITE,NONE,2,raw_add_w_ri,(RW2 d, IMM i)) |
1036 |
+ |
{ |
1037 |
+ |
ADDWir(i, d); |
1038 |
+ |
} |
1039 |
+ |
LENDFUNC(WRITE,NONE,2,raw_add_w_ri,(RW2 d, IMM i)) |
1040 |
+ |
|
1041 |
+ |
LOWFUNC(WRITE,NONE,2,raw_add_b_ri,(RW1 d, IMM i)) |
1042 |
+ |
{ |
1043 |
+ |
ADDBir(i, d); |
1044 |
+ |
} |
1045 |
+ |
LENDFUNC(WRITE,NONE,2,raw_add_b_ri,(RW1 d, IMM i)) |
1046 |
+ |
|
1047 |
+ |
LOWFUNC(RMW,NONE,2,raw_sbb_l,(RW4 d, R4 s)) |
1048 |
+ |
{ |
1049 |
+ |
SBBLrr(s, d); |
1050 |
+ |
} |
1051 |
+ |
LENDFUNC(RMW,NONE,2,raw_sbb_l,(RW4 d, R4 s)) |
1052 |
+ |
|
1053 |
+ |
LOWFUNC(RMW,NONE,2,raw_sbb_w,(RW2 d, R2 s)) |
1054 |
+ |
{ |
1055 |
+ |
SBBWrr(s, d); |
1056 |
+ |
} |
1057 |
+ |
LENDFUNC(RMW,NONE,2,raw_sbb_w,(RW2 d, R2 s)) |
1058 |
+ |
|
1059 |
+ |
LOWFUNC(RMW,NONE,2,raw_sbb_b,(RW1 d, R1 s)) |
1060 |
+ |
{ |
1061 |
+ |
SBBBrr(s, d); |
1062 |
+ |
} |
1063 |
+ |
LENDFUNC(RMW,NONE,2,raw_sbb_b,(RW1 d, R1 s)) |
1064 |
+ |
|
1065 |
+ |
LOWFUNC(WRITE,NONE,2,raw_sub_l,(RW4 d, R4 s)) |
1066 |
+ |
{ |
1067 |
+ |
SUBLrr(s, d); |
1068 |
+ |
} |
1069 |
+ |
LENDFUNC(WRITE,NONE,2,raw_sub_l,(RW4 d, R4 s)) |
1070 |
+ |
|
1071 |
+ |
LOWFUNC(WRITE,NONE,2,raw_sub_w,(RW2 d, R2 s)) |
1072 |
+ |
{ |
1073 |
+ |
SUBWrr(s, d); |
1074 |
+ |
} |
1075 |
+ |
LENDFUNC(WRITE,NONE,2,raw_sub_w,(RW2 d, R2 s)) |
1076 |
+ |
|
1077 |
+ |
LOWFUNC(WRITE,NONE,2,raw_sub_b,(RW1 d, R1 s)) |
1078 |
+ |
{ |
1079 |
+ |
SUBBrr(s, d); |
1080 |
+ |
} |
1081 |
+ |
LENDFUNC(WRITE,NONE,2,raw_sub_b,(RW1 d, R1 s)) |
1082 |
+ |
|
1083 |
+ |
LOWFUNC(WRITE,NONE,2,raw_cmp_l,(R4 d, R4 s)) |
1084 |
+ |
{ |
1085 |
+ |
CMPLrr(s, d); |
1086 |
+ |
} |
1087 |
+ |
LENDFUNC(WRITE,NONE,2,raw_cmp_l,(R4 d, R4 s)) |
1088 |
+ |
|
1089 |
+ |
LOWFUNC(WRITE,NONE,2,raw_cmp_l_ri,(R4 r, IMM i)) |
1090 |
+ |
{ |
1091 |
+ |
CMPLir(i, r); |
1092 |
+ |
} |
1093 |
+ |
LENDFUNC(WRITE,NONE,2,raw_cmp_l_ri,(R4 r, IMM i)) |
1094 |
+ |
|
1095 |
+ |
LOWFUNC(WRITE,NONE,2,raw_cmp_w,(R2 d, R2 s)) |
1096 |
+ |
{ |
1097 |
+ |
CMPWrr(s, d); |
1098 |
+ |
} |
1099 |
+ |
LENDFUNC(WRITE,NONE,2,raw_cmp_w,(R2 d, R2 s)) |
1100 |
+ |
|
1101 |
+ |
LOWFUNC(WRITE,READ,2,raw_cmp_b_mi,(MEMR d, IMM s)) |
1102 |
+ |
{ |
1103 |
+ |
CMPBim(s, d, X86_NOREG, X86_NOREG, 1); |
1104 |
+ |
} |
1105 |
+ |
LENDFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1106 |
+ |
|
1107 |
+ |
LOWFUNC(WRITE,NONE,2,raw_cmp_b_ri,(R1 d, IMM i)) |
1108 |
+ |
{ |
1109 |
+ |
CMPBir(i, d); |
1110 |
+ |
} |
1111 |
+ |
LENDFUNC(WRITE,NONE,2,raw_cmp_b_ri,(R1 d, IMM i)) |
1112 |
+ |
|
1113 |
+ |
LOWFUNC(WRITE,NONE,2,raw_cmp_b,(R1 d, R1 s)) |
1114 |
+ |
{ |
1115 |
+ |
CMPBrr(s, d); |
1116 |
+ |
} |
1117 |
+ |
LENDFUNC(WRITE,NONE,2,raw_cmp_b,(R1 d, R1 s)) |
1118 |
+ |
|
1119 |
+ |
LOWFUNC(WRITE,READ,4,raw_cmp_l_rm_indexed,(R4 d, IMM offset, R4 index, IMM factor)) |
1120 |
+ |
{ |
1121 |
+ |
CMPLmr(offset, X86_NOREG, index, factor, d); |
1122 |
+ |
} |
1123 |
+ |
LENDFUNC(WRITE,READ,4,raw_cmp_l_rm_indexed,(R4 d, IMM offset, R4 index, IMM factor)) |
1124 |
+ |
|
1125 |
+ |
LOWFUNC(WRITE,NONE,2,raw_xor_l,(RW4 d, R4 s)) |
1126 |
+ |
{ |
1127 |
+ |
XORLrr(s, d); |
1128 |
+ |
} |
1129 |
+ |
LENDFUNC(WRITE,NONE,2,raw_xor_l,(RW4 d, R4 s)) |
1130 |
+ |
|
1131 |
+ |
LOWFUNC(WRITE,NONE,2,raw_xor_w,(RW2 d, R2 s)) |
1132 |
+ |
{ |
1133 |
+ |
XORWrr(s, d); |
1134 |
+ |
} |
1135 |
+ |
LENDFUNC(WRITE,NONE,2,raw_xor_w,(RW2 d, R2 s)) |
1136 |
+ |
|
1137 |
+ |
LOWFUNC(WRITE,NONE,2,raw_xor_b,(RW1 d, R1 s)) |
1138 |
+ |
{ |
1139 |
+ |
XORBrr(s, d); |
1140 |
+ |
} |
1141 |
+ |
LENDFUNC(WRITE,NONE,2,raw_xor_b,(RW1 d, R1 s)) |
1142 |
+ |
|
1143 |
+ |
LOWFUNC(WRITE,RMW,2,raw_sub_l_mi,(MEMRW d, IMM s)) |
1144 |
+ |
{ |
1145 |
+ |
SUBLim(s, d, X86_NOREG, X86_NOREG, 1); |
1146 |
+ |
} |
1147 |
+ |
LENDFUNC(WRITE,RMW,2,raw_sub_l_mi,(MEMRW d, IMM s)) |
1148 |
+ |
|
1149 |
+ |
LOWFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1150 |
+ |
{ |
1151 |
+ |
CMPLim(s, d, X86_NOREG, X86_NOREG, 1); |
1152 |
+ |
} |
1153 |
+ |
LENDFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1154 |
+ |
|
1155 |
+ |
LOWFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
1156 |
+ |
{ |
1157 |
+ |
XCHGLrr(r2, r1); |
1158 |
+ |
} |
1159 |
+ |
LENDFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
1160 |
+ |
|
1161 |
+ |
LOWFUNC(READ,WRITE,0,raw_pushfl,(void)) |
1162 |
+ |
{ |
1163 |
+ |
PUSHF(); |
1164 |
+ |
} |
1165 |
+ |
LENDFUNC(READ,WRITE,0,raw_pushfl,(void)) |
1166 |
+ |
|
1167 |
+ |
LOWFUNC(WRITE,READ,0,raw_popfl,(void)) |
1168 |
+ |
{ |
1169 |
+ |
POPF(); |
1170 |
+ |
} |
1171 |
+ |
LENDFUNC(WRITE,READ,0,raw_popfl,(void)) |
1172 |
+ |
|
1173 |
+ |
#else |
1174 |
+ |
|
1175 |
|
const bool optimize_accum = true; |
1176 |
|
const bool optimize_imm8 = true; |
1177 |
|
const bool optimize_shift_once = true; |
2121 |
|
|
2122 |
|
LOWFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
2123 |
|
{ |
2124 |
+ |
Dif(!isbyte(offset)) abort(); |
2125 |
|
emit_byte(0x8b); |
2126 |
|
emit_byte(0x40+8*d+s); |
2127 |
|
emit_byte(offset); |
2130 |
|
|
2131 |
|
LOWFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
2132 |
|
{ |
2133 |
+ |
Dif(!isbyte(offset)) abort(); |
2134 |
|
emit_byte(0x66); |
2135 |
|
emit_byte(0x8b); |
2136 |
|
emit_byte(0x40+8*d+s); |
2140 |
|
|
2141 |
|
LOWFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
2142 |
|
{ |
2143 |
+ |
Dif(!isbyte(offset)) abort(); |
2144 |
|
emit_byte(0x8a); |
2145 |
|
emit_byte(0x40+8*d+s); |
2146 |
|
emit_byte(offset); |
2174 |
|
|
2175 |
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
2176 |
|
{ |
2177 |
+ |
Dif(!isbyte(offset)) abort(); |
2178 |
|
emit_byte(0xc7); |
2179 |
|
emit_byte(0x40+d); |
2180 |
|
emit_byte(offset); |
2184 |
|
|
2185 |
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
2186 |
|
{ |
2187 |
+ |
Dif(!isbyte(offset)) abort(); |
2188 |
|
emit_byte(0x66); |
2189 |
|
emit_byte(0xc7); |
2190 |
|
emit_byte(0x40+d); |
2195 |
|
|
2196 |
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
2197 |
|
{ |
2198 |
+ |
Dif(!isbyte(offset)) abort(); |
2199 |
|
emit_byte(0xc6); |
2200 |
|
emit_byte(0x40+d); |
2201 |
|
emit_byte(offset); |
2205 |
|
|
2206 |
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
2207 |
|
{ |
2208 |
+ |
Dif(!isbyte(offset)) abort(); |
2209 |
|
emit_byte(0x89); |
2210 |
|
emit_byte(0x40+8*s+d); |
2211 |
|
emit_byte(offset); |
2214 |
|
|
2215 |
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
2216 |
|
{ |
2217 |
+ |
Dif(!isbyte(offset)) abort(); |
2218 |
|
emit_byte(0x66); |
2219 |
|
emit_byte(0x89); |
2220 |
|
emit_byte(0x40+8*s+d); |
2224 |
|
|
2225 |
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
2226 |
|
{ |
2227 |
+ |
Dif(!isbyte(offset)) abort(); |
2228 |
|
emit_byte(0x88); |
2229 |
|
emit_byte(0x40+8*s+d); |
2230 |
|
emit_byte(offset); |
2915 |
|
LENDFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
2916 |
|
|
2917 |
|
/************************************************************************* |
1859 |
– |
* FIXME: string-related instructions * |
1860 |
– |
*************************************************************************/ |
1861 |
– |
|
1862 |
– |
LOWFUNC(WRITE,NONE,0,raw_cld,(void)) |
1863 |
– |
{ |
1864 |
– |
emit_byte(0xfc); |
1865 |
– |
} |
1866 |
– |
LENDFUNC(WRITE,NONE,0,raw_cld,(void)) |
1867 |
– |
|
1868 |
– |
LOWFUNC(WRITE,NONE,0,raw_std,(void)) |
1869 |
– |
{ |
1870 |
– |
emit_byte(0xfd); |
1871 |
– |
} |
1872 |
– |
LENDFUNC(WRITE,NONE,0,raw_std,(void)) |
1873 |
– |
|
1874 |
– |
LOWFUNC(NONE,RMW,0,raw_movs_b,(void)) |
1875 |
– |
{ |
1876 |
– |
emit_byte(0xa4); |
1877 |
– |
} |
1878 |
– |
LENDFUNC(NONE,RMW,0,raw_movs_b,(void)) |
1879 |
– |
|
1880 |
– |
LOWFUNC(NONE,RMW,0,raw_movs_l,(void)) |
1881 |
– |
{ |
1882 |
– |
emit_byte(0xa5); |
1883 |
– |
} |
1884 |
– |
LENDFUNC(NONE,RMW,0,raw_movs_l,(void)) |
1885 |
– |
|
1886 |
– |
LOWFUNC(NONE,RMW,0,raw_rep,(void)) |
1887 |
– |
{ |
1888 |
– |
emit_byte(0xf3); |
1889 |
– |
} |
1890 |
– |
LENDFUNC(NONE,RMW,0,raw_rep,(void)) |
1891 |
– |
|
1892 |
– |
LOWFUNC(NONE,RMW,0,raw_rep_movsb,(void)) |
1893 |
– |
{ |
1894 |
– |
raw_rep(); |
1895 |
– |
raw_movs_b(); |
1896 |
– |
} |
1897 |
– |
LENDFUNC(NONE,RMW,0,raw_rep_movsb,(void)) |
1898 |
– |
|
1899 |
– |
LOWFUNC(NONE,RMW,0,raw_rep_movsl,(void)) |
1900 |
– |
{ |
1901 |
– |
raw_rep(); |
1902 |
– |
raw_movs_l(); |
1903 |
– |
} |
1904 |
– |
LENDFUNC(NONE,RMW,0,raw_rep_movsl,(void)) |
1905 |
– |
|
1906 |
– |
/************************************************************************* |
2918 |
|
* FIXME: mem access modes probably wrong * |
2919 |
|
*************************************************************************/ |
2920 |
|
|
2930 |
|
} |
2931 |
|
LENDFUNC(WRITE,READ,0,raw_popfl,(void)) |
2932 |
|
|
2933 |
+ |
#endif |
2934 |
+ |
|
2935 |
|
/************************************************************************* |
2936 |
|
* Unoptimizable stuff --- jump * |
2937 |
|
*************************************************************************/ |
2938 |
|
|
2939 |
|
static __inline__ void raw_call_r(R4 r) |
2940 |
|
{ |
2941 |
+ |
#if USE_NEW_RTASM |
2942 |
+ |
CALLsr(r); |
2943 |
+ |
#else |
2944 |
|
emit_byte(0xff); |
2945 |
|
emit_byte(0xd0+r); |
2946 |
+ |
#endif |
2947 |
|
} |
2948 |
|
|
2949 |
|
static __inline__ void raw_call_m_indexed(uae_u32 base, uae_u32 r, uae_u32 m) |
2950 |
|
{ |
2951 |
+ |
#if USE_NEW_RTASM |
2952 |
+ |
CALLsm(base, X86_NOREG, r, m); |
2953 |
+ |
#else |
2954 |
|
int mu; |
2955 |
|
switch(m) { |
2956 |
|
case 1: mu=0; break; |
2963 |
|
emit_byte(0x14); |
2964 |
|
emit_byte(0x05+8*r+0x40*mu); |
2965 |
|
emit_long(base); |
2966 |
+ |
#endif |
2967 |
|
} |
2968 |
|
|
2969 |
|
static __inline__ void raw_jmp_r(R4 r) |
2970 |
|
{ |
2971 |
+ |
#if USE_NEW_RTASM |
2972 |
+ |
JMPsr(r); |
2973 |
+ |
#else |
2974 |
|
emit_byte(0xff); |
2975 |
|
emit_byte(0xe0+r); |
2976 |
+ |
#endif |
2977 |
|
} |
2978 |
|
|
2979 |
|
static __inline__ void raw_jmp_m_indexed(uae_u32 base, uae_u32 r, uae_u32 m) |
2980 |
|
{ |
2981 |
+ |
#if USE_NEW_RTASM |
2982 |
+ |
JMPsm(base, X86_NOREG, r, m); |
2983 |
+ |
#else |
2984 |
|
int mu; |
2985 |
|
switch(m) { |
2986 |
|
case 1: mu=0; break; |
2993 |
|
emit_byte(0x24); |
2994 |
|
emit_byte(0x05+8*r+0x40*mu); |
2995 |
|
emit_long(base); |
2996 |
+ |
#endif |
2997 |
|
} |
2998 |
|
|
2999 |
|
static __inline__ void raw_jmp_m(uae_u32 base) |
3006 |
|
|
3007 |
|
static __inline__ void raw_call(uae_u32 t) |
3008 |
|
{ |
3009 |
+ |
#if USE_NEW_RTASM |
3010 |
+ |
CALLm(t); |
3011 |
+ |
#else |
3012 |
|
emit_byte(0xe8); |
3013 |
|
emit_long(t-(uae_u32)target-4); |
3014 |
+ |
#endif |
3015 |
|
} |
3016 |
|
|
3017 |
|
static __inline__ void raw_jmp(uae_u32 t) |
3018 |
|
{ |
3019 |
+ |
#if USE_NEW_RTASM |
3020 |
+ |
JMPm(t); |
3021 |
+ |
#else |
3022 |
|
emit_byte(0xe9); |
3023 |
|
emit_long(t-(uae_u32)target-4); |
3024 |
+ |
#endif |
3025 |
|
} |
3026 |
|
|
3027 |
|
static __inline__ void raw_jl(uae_u32 t) |
3028 |
|
{ |
3029 |
|
emit_byte(0x0f); |
3030 |
|
emit_byte(0x8c); |
3031 |
< |
emit_long(t-(uae_u32)target-4); |
3031 |
> |
emit_long(t-(uintptr)target-4); |
3032 |
|
} |
3033 |
|
|
3034 |
|
static __inline__ void raw_jz(uae_u32 t) |
3035 |
|
{ |
3036 |
|
emit_byte(0x0f); |
3037 |
|
emit_byte(0x84); |
3038 |
< |
emit_long(t-(uae_u32)target-4); |
3038 |
> |
emit_long(t-(uintptr)target-4); |
3039 |
|
} |
3040 |
|
|
3041 |
|
static __inline__ void raw_jnz(uae_u32 t) |
3042 |
|
{ |
3043 |
|
emit_byte(0x0f); |
3044 |
|
emit_byte(0x85); |
3045 |
< |
emit_long(t-(uae_u32)target-4); |
3045 |
> |
emit_long(t-(uintptr)target-4); |
3046 |
|
} |
3047 |
|
|
3048 |
|
static __inline__ void raw_jnz_l_oponly(void) |
3092 |
|
emit_byte(0x90); |
3093 |
|
} |
3094 |
|
|
3095 |
+ |
static __inline__ void raw_emit_nop_filler(int nbytes) |
3096 |
+ |
{ |
3097 |
+ |
/* Source: GNU Binutils 2.12.90.0.15 */ |
3098 |
+ |
/* Various efficient no-op patterns for aligning code labels. |
3099 |
+ |
Note: Don't try to assemble the instructions in the comments. |
3100 |
+ |
0L and 0w are not legal. */ |
3101 |
+ |
static const uae_u8 f32_1[] = |
3102 |
+ |
{0x90}; /* nop */ |
3103 |
+ |
static const uae_u8 f32_2[] = |
3104 |
+ |
{0x89,0xf6}; /* movl %esi,%esi */ |
3105 |
+ |
static const uae_u8 f32_3[] = |
3106 |
+ |
{0x8d,0x76,0x00}; /* leal 0(%esi),%esi */ |
3107 |
+ |
static const uae_u8 f32_4[] = |
3108 |
+ |
{0x8d,0x74,0x26,0x00}; /* leal 0(%esi,1),%esi */ |
3109 |
+ |
static const uae_u8 f32_5[] = |
3110 |
+ |
{0x90, /* nop */ |
3111 |
+ |
0x8d,0x74,0x26,0x00}; /* leal 0(%esi,1),%esi */ |
3112 |
+ |
static const uae_u8 f32_6[] = |
3113 |
+ |
{0x8d,0xb6,0x00,0x00,0x00,0x00}; /* leal 0L(%esi),%esi */ |
3114 |
+ |
static const uae_u8 f32_7[] = |
3115 |
+ |
{0x8d,0xb4,0x26,0x00,0x00,0x00,0x00}; /* leal 0L(%esi,1),%esi */ |
3116 |
+ |
static const uae_u8 f32_8[] = |
3117 |
+ |
{0x90, /* nop */ |
3118 |
+ |
0x8d,0xb4,0x26,0x00,0x00,0x00,0x00}; /* leal 0L(%esi,1),%esi */ |
3119 |
+ |
static const uae_u8 f32_9[] = |
3120 |
+ |
{0x89,0xf6, /* movl %esi,%esi */ |
3121 |
+ |
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
3122 |
+ |
static const uae_u8 f32_10[] = |
3123 |
+ |
{0x8d,0x76,0x00, /* leal 0(%esi),%esi */ |
3124 |
+ |
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
3125 |
+ |
static const uae_u8 f32_11[] = |
3126 |
+ |
{0x8d,0x74,0x26,0x00, /* leal 0(%esi,1),%esi */ |
3127 |
+ |
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
3128 |
+ |
static const uae_u8 f32_12[] = |
3129 |
+ |
{0x8d,0xb6,0x00,0x00,0x00,0x00, /* leal 0L(%esi),%esi */ |
3130 |
+ |
0x8d,0xbf,0x00,0x00,0x00,0x00}; /* leal 0L(%edi),%edi */ |
3131 |
+ |
static const uae_u8 f32_13[] = |
3132 |
+ |
{0x8d,0xb6,0x00,0x00,0x00,0x00, /* leal 0L(%esi),%esi */ |
3133 |
+ |
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
3134 |
+ |
static const uae_u8 f32_14[] = |
3135 |
+ |
{0x8d,0xb4,0x26,0x00,0x00,0x00,0x00, /* leal 0L(%esi,1),%esi */ |
3136 |
+ |
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
3137 |
+ |
static const uae_u8 f32_15[] = |
3138 |
+ |
{0xeb,0x0d,0x90,0x90,0x90,0x90,0x90, /* jmp .+15; lotsa nops */ |
3139 |
+ |
0x90,0x90,0x90,0x90,0x90,0x90,0x90,0x90}; |
3140 |
+ |
static const uae_u8 f32_16[] = |
3141 |
+ |
{0xeb,0x0d,0x90,0x90,0x90,0x90,0x90, /* jmp .+15; lotsa nops */ |
3142 |
+ |
0x90,0x90,0x90,0x90,0x90,0x90,0x90,0x90,0x90}; |
3143 |
+ |
static const uae_u8 *const f32_patt[] = { |
3144 |
+ |
f32_1, f32_2, f32_3, f32_4, f32_5, f32_6, f32_7, f32_8, |
3145 |
+ |
f32_9, f32_10, f32_11, f32_12, f32_13, f32_14, f32_15 |
3146 |
+ |
}; |
3147 |
+ |
static const uae_u8 prefixes[4] = { 0x66, 0x66, 0x66, 0x66 }; |
3148 |
+ |
|
3149 |
+ |
#if defined(__x86_64__) |
3150 |
+ |
/* The recommended way to pad 64bit code is to use NOPs preceded by |
3151 |
+ |
maximally four 0x66 prefixes. Balance the size of nops. */ |
3152 |
+ |
if (nbytes == 0) |
3153 |
+ |
return; |
3154 |
+ |
|
3155 |
+ |
int i; |
3156 |
+ |
int nnops = (nbytes + 3) / 4; |
3157 |
+ |
int len = nbytes / nnops; |
3158 |
+ |
int remains = nbytes - nnops * len; |
3159 |
+ |
|
3160 |
+ |
for (i = 0; i < remains; i++) { |
3161 |
+ |
emit_block(prefixes, len); |
3162 |
+ |
raw_nop(); |
3163 |
+ |
} |
3164 |
+ |
for (; i < nnops; i++) { |
3165 |
+ |
emit_block(prefixes, len - 1); |
3166 |
+ |
raw_nop(); |
3167 |
+ |
} |
3168 |
+ |
#else |
3169 |
+ |
int nloops = nbytes / 16; |
3170 |
+ |
while (nloops-- > 0) |
3171 |
+ |
emit_block(f32_16, sizeof(f32_16)); |
3172 |
+ |
|
3173 |
+ |
nbytes %= 16; |
3174 |
+ |
if (nbytes) |
3175 |
+ |
emit_block(f32_patt[nbytes - 1], nbytes); |
3176 |
+ |
#endif |
3177 |
+ |
} |
3178 |
+ |
|
3179 |
|
|
3180 |
|
/************************************************************************* |
3181 |
|
* Flag handling, to and fro UAE flag register * |
3189 |
|
{ |
3190 |
|
raw_lahf(0); /* Most flags in AH */ |
3191 |
|
//raw_setcc(r,0); /* V flag in AL */ |
3192 |
< |
raw_setcc_m((uae_u32)live.state[FLAGTMP].mem,0); |
3192 |
> |
raw_setcc_m((uintptr)live.state[FLAGTMP].mem,0); |
3193 |
|
|
3194 |
|
#if 1 /* Let's avoid those nasty partial register stalls */ |
3195 |
< |
//raw_mov_b_mr((uae_u32)live.state[FLAGTMP].mem,r); |
3196 |
< |
raw_mov_b_mr(((uae_u32)live.state[FLAGTMP].mem)+1,r+4); |
3195 |
> |
//raw_mov_b_mr((uintptr)live.state[FLAGTMP].mem,r); |
3196 |
> |
raw_mov_b_mr(((uintptr)live.state[FLAGTMP].mem)+1,r+4); |
3197 |
|
//live.state[FLAGTMP].status=CLEAN; |
3198 |
|
live.state[FLAGTMP].status=INMEM; |
3199 |
|
live.state[FLAGTMP].realreg=-1; |
3220 |
|
{ |
3221 |
|
raw_pushfl(); |
3222 |
|
raw_pop_l_r(r); |
3223 |
< |
raw_mov_l_mr((uae_u32)live.state[FLAGTMP].mem,r); |
3223 |
> |
raw_mov_l_mr((uintptr)live.state[FLAGTMP].mem,r); |
3224 |
|
// live.state[FLAGTMP].status=CLEAN; |
3225 |
|
live.state[FLAGTMP].status=INMEM; |
3226 |
|
live.state[FLAGTMP].realreg=-1; |
3246 |
|
static __inline__ void raw_load_flagreg(uae_u32 target, uae_u32 r) |
3247 |
|
{ |
3248 |
|
#if 1 |
3249 |
< |
raw_mov_l_rm(target,(uae_u32)live.state[r].mem); |
3249 |
> |
raw_mov_l_rm(target,(uintptr)live.state[r].mem); |
3250 |
|
#else |
3251 |
< |
raw_mov_b_rm(target,(uae_u32)live.state[r].mem); |
3252 |
< |
raw_mov_b_rm(target+4,((uae_u32)live.state[r].mem)+1); |
3251 |
> |
raw_mov_b_rm(target,(uintptr)live.state[r].mem); |
3252 |
> |
raw_mov_b_rm(target+4,((uintptr)live.state[r].mem)+1); |
3253 |
|
#endif |
3254 |
|
} |
3255 |
|
|
3257 |
|
static __inline__ void raw_load_flagx(uae_u32 target, uae_u32 r) |
3258 |
|
{ |
3259 |
|
if (live.nat[target].canbyte) |
3260 |
< |
raw_mov_b_rm(target,(uae_u32)live.state[r].mem); |
3260 |
> |
raw_mov_b_rm(target,(uintptr)live.state[r].mem); |
3261 |
|
else if (live.nat[target].canword) |
3262 |
< |
raw_mov_w_rm(target,(uae_u32)live.state[r].mem); |
3262 |
> |
raw_mov_w_rm(target,(uintptr)live.state[r].mem); |
3263 |
|
else |
3264 |
< |
raw_mov_l_rm(target,(uae_u32)live.state[r].mem); |
3264 |
> |
raw_mov_l_rm(target,(uintptr)live.state[r].mem); |
3265 |
|
} |
3266 |
|
|
3267 |
+ |
#define NATIVE_FLAG_Z 0x40 |
3268 |
+ |
static __inline__ void raw_flags_set_zero(int f, int r, int t) |
3269 |
+ |
{ |
3270 |
+ |
// FIXME: this is really suboptimal |
3271 |
+ |
raw_pushfl(); |
3272 |
+ |
raw_pop_l_r(f); |
3273 |
+ |
raw_and_l_ri(f,~NATIVE_FLAG_Z); |
3274 |
+ |
raw_test_l_rr(r,r); |
3275 |
+ |
raw_mov_l_ri(r,0); |
3276 |
+ |
raw_mov_l_ri(t,NATIVE_FLAG_Z); |
3277 |
+ |
raw_cmov_l_rr(r,t,NATIVE_CC_EQ); |
3278 |
+ |
raw_or_l(f,r); |
3279 |
+ |
raw_push_l_r(f); |
3280 |
+ |
raw_popfl(); |
3281 |
+ |
} |
3282 |
|
|
3283 |
|
static __inline__ void raw_inc_sp(int off) |
3284 |
|
{ |
3441 |
|
for (i=0;i<5;i++) |
3442 |
|
vecbuf[i]=target[i]; |
3443 |
|
emit_byte(0xe9); |
3444 |
< |
emit_long((uae_u32)veccode-(uae_u32)target-4); |
3444 |
> |
emit_long((uintptr)veccode-(uintptr)target-4); |
3445 |
|
write_log("Create jump to %p\n",veccode); |
3446 |
|
|
3447 |
|
write_log("Handled one access!\n"); |
3468 |
|
} |
3469 |
|
for (i=0;i<5;i++) |
3470 |
|
raw_mov_b_mi(sc.eip+i,vecbuf[i]); |
3471 |
< |
raw_mov_l_mi((uae_u32)&in_handler,0); |
3471 |
> |
raw_mov_l_mi((uintptr)&in_handler,0); |
3472 |
|
emit_byte(0xe9); |
3473 |
< |
emit_long(sc.eip+len-(uae_u32)target-4); |
3473 |
> |
emit_long(sc.eip+len-(uintptr)target-4); |
3474 |
|
in_handler=1; |
3475 |
|
target=tmp; |
3476 |
|
} |
3565 |
|
X86_PROCESSOR_K6, |
3566 |
|
X86_PROCESSOR_ATHLON, |
3567 |
|
X86_PROCESSOR_PENTIUM4, |
3568 |
+ |
X86_PROCESSOR_K8, |
3569 |
|
X86_PROCESSOR_max |
3570 |
|
}; |
3571 |
|
|
3576 |
|
"PentiumPro", |
3577 |
|
"K6", |
3578 |
|
"Athlon", |
3579 |
< |
"Pentium4" |
3579 |
> |
"Pentium4", |
3580 |
> |
"K8" |
3581 |
|
}; |
3582 |
|
|
3583 |
|
static struct ptt { |
3594 |
|
{ 16, 15, 16, 7, 16 }, |
3595 |
|
{ 32, 7, 32, 7, 32 }, |
3596 |
|
{ 16, 7, 16, 7, 16 }, |
3597 |
< |
{ 0, 0, 0, 0, 0 } |
3597 |
> |
{ 0, 0, 0, 0, 0 }, |
3598 |
> |
{ 16, 7, 16, 7, 16 } |
3599 |
|
}; |
3600 |
|
|
3601 |
|
static void |
3630 |
|
cpuid(uae_u32 op, uae_u32 *eax, uae_u32 *ebx, uae_u32 *ecx, uae_u32 *edx) |
3631 |
|
{ |
3632 |
|
static uae_u8 cpuid_space[256]; |
3633 |
+ |
static uae_u32 s_op, s_eax, s_ebx, s_ecx, s_edx; |
3634 |
|
uae_u8* tmp=get_target(); |
3635 |
|
|
3636 |
+ |
s_op = op; |
3637 |
|
set_target(cpuid_space); |
3638 |
|
raw_push_l_r(0); /* eax */ |
3639 |
|
raw_push_l_r(1); /* ecx */ |
3640 |
|
raw_push_l_r(2); /* edx */ |
3641 |
|
raw_push_l_r(3); /* ebx */ |
3642 |
< |
raw_mov_l_rm(0,(uae_u32)&op); |
3642 |
> |
raw_mov_l_rm(0,(uintptr)&s_op); |
3643 |
|
raw_cpuid(0); |
3644 |
< |
if (eax != NULL) raw_mov_l_mr((uae_u32)eax,0); |
3645 |
< |
if (ebx != NULL) raw_mov_l_mr((uae_u32)ebx,3); |
3646 |
< |
if (ecx != NULL) raw_mov_l_mr((uae_u32)ecx,1); |
3647 |
< |
if (edx != NULL) raw_mov_l_mr((uae_u32)edx,2); |
3644 |
> |
raw_mov_l_mr((uintptr)&s_eax,0); |
3645 |
> |
raw_mov_l_mr((uintptr)&s_ebx,3); |
3646 |
> |
raw_mov_l_mr((uintptr)&s_ecx,1); |
3647 |
> |
raw_mov_l_mr((uintptr)&s_edx,2); |
3648 |
|
raw_pop_l_r(3); |
3649 |
|
raw_pop_l_r(2); |
3650 |
|
raw_pop_l_r(1); |
3653 |
|
set_target(tmp); |
3654 |
|
|
3655 |
|
((cpuop_func*)cpuid_space)(0); |
3656 |
+ |
if (eax != NULL) *eax = s_eax; |
3657 |
+ |
if (ebx != NULL) *ebx = s_ebx; |
3658 |
+ |
if (ecx != NULL) *ecx = s_ecx; |
3659 |
+ |
if (edx != NULL) *edx = s_edx; |
3660 |
|
} |
3661 |
|
|
3662 |
|
static void |
3665 |
|
struct cpuinfo_x86 *c = &cpuinfo; |
3666 |
|
|
3667 |
|
/* Defaults */ |
3668 |
+ |
c->x86_processor = X86_PROCESSOR_max; |
3669 |
|
c->x86_vendor = X86_VENDOR_UNKNOWN; |
3670 |
|
c->cpuid_level = -1; /* CPUID not detected */ |
3671 |
|
c->x86_model = c->x86_mask = 0; /* So far unknown... */ |
3701 |
|
c->x86 = 4; |
3702 |
|
} |
3703 |
|
|
3704 |
+ |
/* AMD-defined flags: level 0x80000001 */ |
3705 |
+ |
uae_u32 xlvl; |
3706 |
+ |
cpuid(0x80000000, &xlvl, NULL, NULL, NULL); |
3707 |
+ |
if ( (xlvl & 0xffff0000) == 0x80000000 ) { |
3708 |
+ |
if ( xlvl >= 0x80000001 ) { |
3709 |
+ |
uae_u32 features; |
3710 |
+ |
cpuid(0x80000001, NULL, NULL, NULL, &features); |
3711 |
+ |
if (features & (1 << 29)) { |
3712 |
+ |
/* Assume x86-64 if long mode is supported */ |
3713 |
+ |
c->x86_processor = X86_PROCESSOR_K8; |
3714 |
+ |
} |
3715 |
+ |
} |
3716 |
+ |
} |
3717 |
+ |
|
3718 |
|
/* Canonicalize processor ID */ |
2559 |
– |
c->x86_processor = X86_PROCESSOR_max; |
3719 |
|
switch (c->x86) { |
3720 |
|
case 3: |
3721 |
|
c->x86_processor = X86_PROCESSOR_I386; |
3737 |
|
break; |
3738 |
|
case 15: |
3739 |
|
if (c->x86_vendor == X86_VENDOR_INTEL) { |
3740 |
< |
/* Assume any BranID >= 8 and family == 15 yields a Pentium 4 */ |
3740 |
> |
/* Assume any BrandID >= 8 and family == 15 yields a Pentium 4 */ |
3741 |
|
if (c->x86_brand_id >= 8) |
3742 |
|
c->x86_processor = X86_PROCESSOR_PENTIUM4; |
3743 |
|
} |
3744 |
+ |
if (c->x86_vendor == X86_VENDOR_AMD) { |
3745 |
+ |
/* Assume an Athlon processor if family == 15 and it was not |
3746 |
+ |
detected as an x86-64 so far */ |
3747 |
+ |
if (c->x86_processor == X86_PROCESSOR_max) |
3748 |
+ |
c->x86_processor = X86_PROCESSOR_ATHLON; |
3749 |
+ |
} |
3750 |
|
break; |
3751 |
|
} |
3752 |
|
if (c->x86_processor == X86_PROCESSOR_max) { |
3754 |
|
fprintf(stderr, " Family : %d\n", c->x86); |
3755 |
|
fprintf(stderr, " Model : %d\n", c->x86_model); |
3756 |
|
fprintf(stderr, " Mask : %d\n", c->x86_mask); |
3757 |
+ |
fprintf(stderr, " Vendor : %s [%d]\n", c->x86_vendor_id, c->x86_vendor); |
3758 |
|
if (c->x86_brand_id) |
3759 |
|
fprintf(stderr, " BrandID : %02x\n", c->x86_brand_id); |
3760 |
|
abort(); |
3761 |
|
} |
3762 |
|
|
3763 |
|
/* Have CMOV support? */ |
3764 |
< |
have_cmov = (c->x86_hwcap & (1 << 15)) && true; |
3764 |
> |
have_cmov = c->x86_hwcap & (1 << 15); |
3765 |
|
|
3766 |
|
/* Can the host CPU suffer from partial register stalls? */ |
3767 |
|
have_rat_stall = (c->x86_vendor == X86_VENDOR_INTEL); |
3784 |
|
x86_processor_string_table[c->x86_processor]); |
3785 |
|
} |
3786 |
|
|
3787 |
+ |
static bool target_check_bsf(void) |
3788 |
+ |
{ |
3789 |
+ |
bool mismatch = false; |
3790 |
+ |
for (int g_ZF = 0; g_ZF <= 1; g_ZF++) { |
3791 |
+ |
for (int g_CF = 0; g_CF <= 1; g_CF++) { |
3792 |
+ |
for (int g_OF = 0; g_OF <= 1; g_OF++) { |
3793 |
+ |
for (int g_SF = 0; g_SF <= 1; g_SF++) { |
3794 |
+ |
for (int value = -1; value <= 1; value++) { |
3795 |
+ |
int flags = (g_SF << 7) | (g_OF << 11) | (g_ZF << 6) | g_CF; |
3796 |
+ |
int tmp = value; |
3797 |
+ |
__asm__ __volatile__ ("push %0; popf; bsf %1,%1; pushf; pop %0" |
3798 |
+ |
: "+r" (flags), "+r" (tmp) : : "cc"); |
3799 |
+ |
int OF = (flags >> 11) & 1; |
3800 |
+ |
int SF = (flags >> 7) & 1; |
3801 |
+ |
int ZF = (flags >> 6) & 1; |
3802 |
+ |
int CF = flags & 1; |
3803 |
+ |
tmp = (value == 0); |
3804 |
+ |
if (ZF != tmp || SF != g_SF || OF != g_OF || CF != g_CF) |
3805 |
+ |
mismatch = true; |
3806 |
+ |
} |
3807 |
+ |
}}}} |
3808 |
+ |
if (mismatch) |
3809 |
+ |
write_log("Target CPU defines all flags on BSF instruction\n"); |
3810 |
+ |
return !mismatch; |
3811 |
+ |
} |
3812 |
+ |
|
3813 |
|
|
3814 |
|
/************************************************************************* |
3815 |
|
* FPU stuff * |
4244 |
|
emit_byte(0xf0); /* f2xm1 */ |
4245 |
|
emit_byte(0xdc); |
4246 |
|
emit_byte(0x05); |
4247 |
< |
emit_long((uae_u32)&one); /* Add '1' without using extra stack space */ |
4247 |
> |
emit_long((uintptr)&one); /* Add '1' without using extra stack space */ |
4248 |
|
emit_byte(0xd9); |
4249 |
|
emit_byte(0xfd); /* and scale it */ |
4250 |
|
emit_byte(0xdd); |
4278 |
|
emit_byte(0xf0); /* f2xm1 */ |
4279 |
|
emit_byte(0xdc); |
4280 |
|
emit_byte(0x05); |
4281 |
< |
emit_long((uae_u32)&one); /* Add '1' without using extra stack space */ |
4281 |
> |
emit_long((uintptr)&one); /* Add '1' without using extra stack space */ |
4282 |
|
emit_byte(0xd9); |
4283 |
|
emit_byte(0xfd); /* and scale it */ |
4284 |
|
emit_byte(0xdd); |