1 |
gbeauche |
1.6 |
/* |
2 |
|
|
* compiler/codegen_x86.cpp - IA-32 code generator |
3 |
|
|
* |
4 |
|
|
* Original 68040 JIT compiler for UAE, copyright 2000-2002 Bernd Meyer |
5 |
|
|
* |
6 |
|
|
* Adaptation for Basilisk II and improvements, copyright 2000-2002 |
7 |
|
|
* Gwenole Beauchesne |
8 |
|
|
* |
9 |
|
|
* Basilisk II (C) 1997-2002 Christian Bauer |
10 |
|
|
* |
11 |
|
|
* This program is free software; you can redistribute it and/or modify |
12 |
|
|
* it under the terms of the GNU General Public License as published by |
13 |
|
|
* the Free Software Foundation; either version 2 of the License, or |
14 |
|
|
* (at your option) any later version. |
15 |
|
|
* |
16 |
|
|
* This program is distributed in the hope that it will be useful, |
17 |
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
18 |
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
19 |
|
|
* GNU General Public License for more details. |
20 |
|
|
* |
21 |
|
|
* You should have received a copy of the GNU General Public License |
22 |
|
|
* along with this program; if not, write to the Free Software |
23 |
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
24 |
|
|
*/ |
25 |
|
|
|
26 |
gbeauche |
1.1 |
/* This should eventually end up in machdep/, but for now, x86 is the |
27 |
|
|
only target, and it's easier this way... */ |
28 |
|
|
|
29 |
gbeauche |
1.5 |
#include "flags_x86.h" |
30 |
|
|
|
31 |
gbeauche |
1.1 |
/************************************************************************* |
32 |
|
|
* Some basic information about the the target CPU * |
33 |
|
|
*************************************************************************/ |
34 |
|
|
|
35 |
|
|
#define EAX_INDEX 0 |
36 |
|
|
#define ECX_INDEX 1 |
37 |
|
|
#define EDX_INDEX 2 |
38 |
|
|
#define EBX_INDEX 3 |
39 |
|
|
#define ESP_INDEX 4 |
40 |
|
|
#define EBP_INDEX 5 |
41 |
|
|
#define ESI_INDEX 6 |
42 |
|
|
#define EDI_INDEX 7 |
43 |
|
|
|
44 |
|
|
/* The register in which subroutines return an integer return value */ |
45 |
|
|
#define REG_RESULT 0 |
46 |
|
|
|
47 |
|
|
/* The registers subroutines take their first and second argument in */ |
48 |
|
|
#if defined( _MSC_VER ) && !defined( USE_NORMAL_CALLING_CONVENTION ) |
49 |
|
|
/* Handle the _fastcall parameters of ECX and EDX */ |
50 |
|
|
#define REG_PAR1 1 |
51 |
|
|
#define REG_PAR2 2 |
52 |
|
|
#else |
53 |
|
|
#define REG_PAR1 0 |
54 |
|
|
#define REG_PAR2 2 |
55 |
|
|
#endif |
56 |
|
|
|
57 |
|
|
/* 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 */ |
63 |
|
|
#if defined( _MSC_VER ) && !defined( USE_NORMAL_CALLING_CONVENTION ) |
64 |
|
|
#define REG_PC_TMP 0 |
65 |
|
|
#else |
66 |
|
|
#define REG_PC_TMP 1 /* Another register that is not the above */ |
67 |
|
|
#endif |
68 |
|
|
|
69 |
|
|
#define SHIFTCOUNT_NREG 1 /* Register that can be used for shiftcount. |
70 |
|
|
-1 if any reg will do */ |
71 |
|
|
#define MUL_NREG1 0 /* %eax will hold the low 32 bits after a 32x32 mul */ |
72 |
|
|
#define MUL_NREG2 2 /* %edx will hold the high 32 bits */ |
73 |
|
|
|
74 |
|
|
uae_s8 always_used[]={4,-1}; |
75 |
|
|
uae_s8 can_byte[]={0,1,2,3,-1}; |
76 |
|
|
uae_s8 can_word[]={0,1,2,3,5,6,7,-1}; |
77 |
|
|
|
78 |
|
|
/* cpuopti mutate instruction handlers to assume registers are saved |
79 |
|
|
by the caller */ |
80 |
|
|
uae_u8 call_saved[]={0,0,0,0,1,0,0,0}; |
81 |
|
|
|
82 |
|
|
/* This *should* be the same as call_saved. But: |
83 |
|
|
- We might not really know which registers are saved, and which aren't, |
84 |
|
|
so we need to preserve some, but don't want to rely on everyone else |
85 |
|
|
also saving those registers |
86 |
|
|
- Special registers (such like the stack pointer) should not be "preserved" |
87 |
|
|
by pushing, even though they are "saved" across function calls |
88 |
|
|
*/ |
89 |
|
|
uae_u8 need_to_preserve[]={1,1,1,1,0,1,1,1}; |
90 |
|
|
|
91 |
|
|
/* Whether classes of instructions do or don't clobber the native flags */ |
92 |
|
|
#define CLOBBER_MOV |
93 |
|
|
#define CLOBBER_LEA |
94 |
|
|
#define CLOBBER_CMOV |
95 |
|
|
#define CLOBBER_POP |
96 |
|
|
#define CLOBBER_PUSH |
97 |
|
|
#define CLOBBER_SUB clobber_flags() |
98 |
|
|
#define CLOBBER_SBB clobber_flags() |
99 |
|
|
#define CLOBBER_CMP clobber_flags() |
100 |
|
|
#define CLOBBER_ADD clobber_flags() |
101 |
|
|
#define CLOBBER_ADC clobber_flags() |
102 |
|
|
#define CLOBBER_AND clobber_flags() |
103 |
|
|
#define CLOBBER_OR clobber_flags() |
104 |
|
|
#define CLOBBER_XOR clobber_flags() |
105 |
|
|
|
106 |
|
|
#define CLOBBER_ROL clobber_flags() |
107 |
|
|
#define CLOBBER_ROR clobber_flags() |
108 |
|
|
#define CLOBBER_SHLL clobber_flags() |
109 |
|
|
#define CLOBBER_SHRL clobber_flags() |
110 |
|
|
#define CLOBBER_SHRA clobber_flags() |
111 |
|
|
#define CLOBBER_TEST clobber_flags() |
112 |
|
|
#define CLOBBER_CL16 |
113 |
|
|
#define CLOBBER_CL8 |
114 |
|
|
#define CLOBBER_SE16 |
115 |
|
|
#define CLOBBER_SE8 |
116 |
|
|
#define CLOBBER_ZE16 |
117 |
|
|
#define CLOBBER_ZE8 |
118 |
|
|
#define CLOBBER_SW16 clobber_flags() |
119 |
|
|
#define CLOBBER_SW32 |
120 |
|
|
#define CLOBBER_SETCC |
121 |
|
|
#define CLOBBER_MUL clobber_flags() |
122 |
|
|
#define CLOBBER_BT clobber_flags() |
123 |
|
|
#define CLOBBER_BSF clobber_flags() |
124 |
|
|
|
125 |
gbeauche |
1.2 |
const bool optimize_accum = true; |
126 |
gbeauche |
1.1 |
const bool optimize_imm8 = true; |
127 |
|
|
const bool optimize_shift_once = true; |
128 |
|
|
|
129 |
|
|
/************************************************************************* |
130 |
|
|
* Actual encoding of the instructions on the target CPU * |
131 |
|
|
*************************************************************************/ |
132 |
|
|
|
133 |
gbeauche |
1.2 |
static __inline__ int isaccum(int r) |
134 |
|
|
{ |
135 |
|
|
return (r == EAX_INDEX); |
136 |
|
|
} |
137 |
|
|
|
138 |
gbeauche |
1.1 |
static __inline__ int isbyte(uae_s32 x) |
139 |
|
|
{ |
140 |
|
|
return (x>=-128 && x<=127); |
141 |
|
|
} |
142 |
|
|
|
143 |
|
|
static __inline__ int isword(uae_s32 x) |
144 |
|
|
{ |
145 |
|
|
return (x>=-32768 && x<=32767); |
146 |
|
|
} |
147 |
|
|
|
148 |
|
|
LOWFUNC(NONE,WRITE,1,raw_push_l_r,(R4 r)) |
149 |
|
|
{ |
150 |
|
|
emit_byte(0x50+r); |
151 |
|
|
} |
152 |
|
|
LENDFUNC(NONE,WRITE,1,raw_push_l_r,(R4 r)) |
153 |
|
|
|
154 |
|
|
LOWFUNC(NONE,READ,1,raw_pop_l_r,(R4 r)) |
155 |
|
|
{ |
156 |
|
|
emit_byte(0x58+r); |
157 |
|
|
} |
158 |
|
|
LENDFUNC(NONE,READ,1,raw_pop_l_r,(R4 r)) |
159 |
|
|
|
160 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bt_l_ri,(R4 r, IMM i)) |
161 |
|
|
{ |
162 |
|
|
emit_byte(0x0f); |
163 |
|
|
emit_byte(0xba); |
164 |
|
|
emit_byte(0xe0+r); |
165 |
|
|
emit_byte(i); |
166 |
|
|
} |
167 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bt_l_ri,(R4 r, IMM i)) |
168 |
|
|
|
169 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bt_l_rr,(R4 r, R4 b)) |
170 |
|
|
{ |
171 |
|
|
emit_byte(0x0f); |
172 |
|
|
emit_byte(0xa3); |
173 |
|
|
emit_byte(0xc0+8*b+r); |
174 |
|
|
} |
175 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bt_l_rr,(R4 r, R4 b)) |
176 |
|
|
|
177 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btc_l_ri,(RW4 r, IMM i)) |
178 |
|
|
{ |
179 |
|
|
emit_byte(0x0f); |
180 |
|
|
emit_byte(0xba); |
181 |
|
|
emit_byte(0xf8+r); |
182 |
|
|
emit_byte(i); |
183 |
|
|
} |
184 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btc_l_ri,(RW4 r, IMM i)) |
185 |
|
|
|
186 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btc_l_rr,(RW4 r, R4 b)) |
187 |
|
|
{ |
188 |
|
|
emit_byte(0x0f); |
189 |
|
|
emit_byte(0xbb); |
190 |
|
|
emit_byte(0xc0+8*b+r); |
191 |
|
|
} |
192 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btc_l_rr,(RW4 r, R4 b)) |
193 |
|
|
|
194 |
|
|
|
195 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btr_l_ri,(RW4 r, IMM i)) |
196 |
|
|
{ |
197 |
|
|
emit_byte(0x0f); |
198 |
|
|
emit_byte(0xba); |
199 |
|
|
emit_byte(0xf0+r); |
200 |
|
|
emit_byte(i); |
201 |
|
|
} |
202 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btr_l_ri,(RW4 r, IMM i)) |
203 |
|
|
|
204 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btr_l_rr,(RW4 r, R4 b)) |
205 |
|
|
{ |
206 |
|
|
emit_byte(0x0f); |
207 |
|
|
emit_byte(0xb3); |
208 |
|
|
emit_byte(0xc0+8*b+r); |
209 |
|
|
} |
210 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btr_l_rr,(RW4 r, R4 b)) |
211 |
|
|
|
212 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bts_l_ri,(RW4 r, IMM i)) |
213 |
|
|
{ |
214 |
|
|
emit_byte(0x0f); |
215 |
|
|
emit_byte(0xba); |
216 |
|
|
emit_byte(0xe8+r); |
217 |
|
|
emit_byte(i); |
218 |
|
|
} |
219 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bts_l_ri,(RW4 r, IMM i)) |
220 |
|
|
|
221 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bts_l_rr,(RW4 r, R4 b)) |
222 |
|
|
{ |
223 |
|
|
emit_byte(0x0f); |
224 |
|
|
emit_byte(0xab); |
225 |
|
|
emit_byte(0xc0+8*b+r); |
226 |
|
|
} |
227 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bts_l_rr,(RW4 r, R4 b)) |
228 |
|
|
|
229 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_w_ri,(RW2 d, IMM i)) |
230 |
|
|
{ |
231 |
|
|
emit_byte(0x66); |
232 |
|
|
if (isbyte(i)) { |
233 |
|
|
emit_byte(0x83); |
234 |
|
|
emit_byte(0xe8+d); |
235 |
|
|
emit_byte(i); |
236 |
|
|
} |
237 |
|
|
else { |
238 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
239 |
|
|
emit_byte(0x2d); |
240 |
|
|
else { |
241 |
gbeauche |
1.1 |
emit_byte(0x81); |
242 |
|
|
emit_byte(0xe8+d); |
243 |
gbeauche |
1.2 |
} |
244 |
gbeauche |
1.1 |
emit_word(i); |
245 |
|
|
} |
246 |
|
|
} |
247 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_w_ri,(RW2 d, IMM i)) |
248 |
|
|
|
249 |
|
|
|
250 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_l_rm,(W4 d, MEMR s)) |
251 |
|
|
{ |
252 |
|
|
emit_byte(0x8b); |
253 |
|
|
emit_byte(0x05+8*d); |
254 |
|
|
emit_long(s); |
255 |
|
|
} |
256 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_l_rm,(W4 d, MEMR s)) |
257 |
|
|
|
258 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_l_mi,(MEMW d, IMM s)) |
259 |
|
|
{ |
260 |
|
|
emit_byte(0xc7); |
261 |
|
|
emit_byte(0x05); |
262 |
|
|
emit_long(d); |
263 |
|
|
emit_long(s); |
264 |
|
|
} |
265 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_l_mi,(MEMW d, IMM s)) |
266 |
|
|
|
267 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_w_mi,(MEMW d, IMM s)) |
268 |
|
|
{ |
269 |
|
|
emit_byte(0x66); |
270 |
|
|
emit_byte(0xc7); |
271 |
|
|
emit_byte(0x05); |
272 |
|
|
emit_long(d); |
273 |
|
|
emit_word(s); |
274 |
|
|
} |
275 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_w_mi,(MEMW d, IMM s)) |
276 |
|
|
|
277 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_b_mi,(MEMW d, IMM s)) |
278 |
|
|
{ |
279 |
|
|
emit_byte(0xc6); |
280 |
|
|
emit_byte(0x05); |
281 |
|
|
emit_long(d); |
282 |
|
|
emit_byte(s); |
283 |
|
|
} |
284 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_b_mi,(MEMW d, IMM s)) |
285 |
|
|
|
286 |
|
|
LOWFUNC(WRITE,RMW,2,raw_rol_b_mi,(MEMRW d, IMM i)) |
287 |
|
|
{ |
288 |
|
|
if (optimize_shift_once && (i == 1)) { |
289 |
|
|
emit_byte(0xd0); |
290 |
|
|
emit_byte(0x05); |
291 |
|
|
emit_long(d); |
292 |
|
|
} |
293 |
|
|
else { |
294 |
|
|
emit_byte(0xc0); |
295 |
|
|
emit_byte(0x05); |
296 |
|
|
emit_long(d); |
297 |
|
|
emit_byte(i); |
298 |
|
|
} |
299 |
|
|
} |
300 |
|
|
LENDFUNC(WRITE,RMW,2,raw_rol_b_mi,(MEMRW d, IMM i)) |
301 |
|
|
|
302 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_b_ri,(RW1 r, IMM i)) |
303 |
|
|
{ |
304 |
|
|
if (optimize_shift_once && (i == 1)) { |
305 |
|
|
emit_byte(0xd0); |
306 |
|
|
emit_byte(0xc0+r); |
307 |
|
|
} |
308 |
|
|
else { |
309 |
|
|
emit_byte(0xc0); |
310 |
|
|
emit_byte(0xc0+r); |
311 |
|
|
emit_byte(i); |
312 |
|
|
} |
313 |
|
|
} |
314 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_b_ri,(RW1 r, IMM i)) |
315 |
|
|
|
316 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_w_ri,(RW2 r, IMM i)) |
317 |
|
|
{ |
318 |
|
|
emit_byte(0x66); |
319 |
|
|
emit_byte(0xc1); |
320 |
|
|
emit_byte(0xc0+r); |
321 |
|
|
emit_byte(i); |
322 |
|
|
} |
323 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_w_ri,(RW2 r, IMM i)) |
324 |
|
|
|
325 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_l_ri,(RW4 r, IMM i)) |
326 |
|
|
{ |
327 |
|
|
if (optimize_shift_once && (i == 1)) { |
328 |
|
|
emit_byte(0xd1); |
329 |
|
|
emit_byte(0xc0+r); |
330 |
|
|
} |
331 |
|
|
else { |
332 |
|
|
emit_byte(0xc1); |
333 |
|
|
emit_byte(0xc0+r); |
334 |
|
|
emit_byte(i); |
335 |
|
|
} |
336 |
|
|
} |
337 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_l_ri,(RW4 r, IMM i)) |
338 |
|
|
|
339 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_l_rr,(RW4 d, R1 r)) |
340 |
|
|
{ |
341 |
|
|
emit_byte(0xd3); |
342 |
|
|
emit_byte(0xc0+d); |
343 |
|
|
} |
344 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_l_rr,(RW4 d, R1 r)) |
345 |
|
|
|
346 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_w_rr,(RW2 d, R1 r)) |
347 |
|
|
{ |
348 |
|
|
emit_byte(0x66); |
349 |
|
|
emit_byte(0xd3); |
350 |
|
|
emit_byte(0xc0+d); |
351 |
|
|
} |
352 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_w_rr,(RW2 d, R1 r)) |
353 |
|
|
|
354 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_b_rr,(RW1 d, R1 r)) |
355 |
|
|
{ |
356 |
|
|
emit_byte(0xd2); |
357 |
|
|
emit_byte(0xc0+d); |
358 |
|
|
} |
359 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_b_rr,(RW1 d, R1 r)) |
360 |
|
|
|
361 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_l_rr,(RW4 d, R1 r)) |
362 |
|
|
{ |
363 |
|
|
emit_byte(0xd3); |
364 |
|
|
emit_byte(0xe0+d); |
365 |
|
|
} |
366 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_l_rr,(RW4 d, R1 r)) |
367 |
|
|
|
368 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_w_rr,(RW2 d, R1 r)) |
369 |
|
|
{ |
370 |
|
|
emit_byte(0x66); |
371 |
|
|
emit_byte(0xd3); |
372 |
|
|
emit_byte(0xe0+d); |
373 |
|
|
} |
374 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_w_rr,(RW2 d, R1 r)) |
375 |
|
|
|
376 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_b_rr,(RW1 d, R1 r)) |
377 |
|
|
{ |
378 |
|
|
emit_byte(0xd2); |
379 |
|
|
emit_byte(0xe0+d); |
380 |
|
|
} |
381 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_b_rr,(RW1 d, R1 r)) |
382 |
|
|
|
383 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_b_ri,(RW1 r, IMM i)) |
384 |
|
|
{ |
385 |
|
|
if (optimize_shift_once && (i == 1)) { |
386 |
|
|
emit_byte(0xd0); |
387 |
|
|
emit_byte(0xc8+r); |
388 |
|
|
} |
389 |
|
|
else { |
390 |
|
|
emit_byte(0xc0); |
391 |
|
|
emit_byte(0xc8+r); |
392 |
|
|
emit_byte(i); |
393 |
|
|
} |
394 |
|
|
} |
395 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_b_ri,(RW1 r, IMM i)) |
396 |
|
|
|
397 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_w_ri,(RW2 r, IMM i)) |
398 |
|
|
{ |
399 |
|
|
emit_byte(0x66); |
400 |
|
|
emit_byte(0xc1); |
401 |
|
|
emit_byte(0xc8+r); |
402 |
|
|
emit_byte(i); |
403 |
|
|
} |
404 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_w_ri,(RW2 r, IMM i)) |
405 |
|
|
|
406 |
|
|
// gb-- used for making an fpcr value in compemu_fpp.cpp |
407 |
|
|
LOWFUNC(WRITE,READ,2,raw_or_l_rm,(RW4 d, MEMR s)) |
408 |
|
|
{ |
409 |
|
|
emit_byte(0x0b); |
410 |
|
|
emit_byte(0x05+8*d); |
411 |
|
|
emit_long(s); |
412 |
|
|
} |
413 |
|
|
LENDFUNC(WRITE,READ,2,raw_or_l_rm,(RW4 d, MEMR s)) |
414 |
|
|
|
415 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_l_ri,(RW4 r, IMM i)) |
416 |
|
|
{ |
417 |
|
|
if (optimize_shift_once && (i == 1)) { |
418 |
|
|
emit_byte(0xd1); |
419 |
|
|
emit_byte(0xc8+r); |
420 |
|
|
} |
421 |
|
|
else { |
422 |
|
|
emit_byte(0xc1); |
423 |
|
|
emit_byte(0xc8+r); |
424 |
|
|
emit_byte(i); |
425 |
|
|
} |
426 |
|
|
} |
427 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_l_ri,(RW4 r, IMM i)) |
428 |
|
|
|
429 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_l_rr,(RW4 d, R1 r)) |
430 |
|
|
{ |
431 |
|
|
emit_byte(0xd3); |
432 |
|
|
emit_byte(0xc8+d); |
433 |
|
|
} |
434 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_l_rr,(RW4 d, R1 r)) |
435 |
|
|
|
436 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_w_rr,(RW2 d, R1 r)) |
437 |
|
|
{ |
438 |
|
|
emit_byte(0x66); |
439 |
|
|
emit_byte(0xd3); |
440 |
|
|
emit_byte(0xc8+d); |
441 |
|
|
} |
442 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_w_rr,(RW2 d, R1 r)) |
443 |
|
|
|
444 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_b_rr,(RW1 d, R1 r)) |
445 |
|
|
{ |
446 |
|
|
emit_byte(0xd2); |
447 |
|
|
emit_byte(0xc8+d); |
448 |
|
|
} |
449 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_b_rr,(RW1 d, R1 r)) |
450 |
|
|
|
451 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_l_rr,(RW4 d, R1 r)) |
452 |
|
|
{ |
453 |
|
|
emit_byte(0xd3); |
454 |
|
|
emit_byte(0xe8+d); |
455 |
|
|
} |
456 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_l_rr,(RW4 d, R1 r)) |
457 |
|
|
|
458 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_w_rr,(RW2 d, R1 r)) |
459 |
|
|
{ |
460 |
|
|
emit_byte(0x66); |
461 |
|
|
emit_byte(0xd3); |
462 |
|
|
emit_byte(0xe8+d); |
463 |
|
|
} |
464 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_w_rr,(RW2 d, R1 r)) |
465 |
|
|
|
466 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_b_rr,(RW1 d, R1 r)) |
467 |
|
|
{ |
468 |
|
|
emit_byte(0xd2); |
469 |
|
|
emit_byte(0xe8+d); |
470 |
|
|
} |
471 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_b_rr,(RW1 d, R1 r)) |
472 |
|
|
|
473 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_l_rr,(RW4 d, R1 r)) |
474 |
|
|
{ |
475 |
|
|
emit_byte(0xd3); |
476 |
|
|
emit_byte(0xf8+d); |
477 |
|
|
} |
478 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_l_rr,(RW4 d, R1 r)) |
479 |
|
|
|
480 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_w_rr,(RW2 d, R1 r)) |
481 |
|
|
{ |
482 |
|
|
emit_byte(0x66); |
483 |
|
|
emit_byte(0xd3); |
484 |
|
|
emit_byte(0xf8+d); |
485 |
|
|
} |
486 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_w_rr,(RW2 d, R1 r)) |
487 |
|
|
|
488 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_b_rr,(RW1 d, R1 r)) |
489 |
|
|
{ |
490 |
|
|
emit_byte(0xd2); |
491 |
|
|
emit_byte(0xf8+d); |
492 |
|
|
} |
493 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_b_rr,(RW1 d, R1 r)) |
494 |
|
|
|
495 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_l_ri,(RW4 r, IMM i)) |
496 |
|
|
{ |
497 |
|
|
if (optimize_shift_once && (i == 1)) { |
498 |
|
|
emit_byte(0xd1); |
499 |
|
|
emit_byte(0xe0+r); |
500 |
|
|
} |
501 |
|
|
else { |
502 |
|
|
emit_byte(0xc1); |
503 |
|
|
emit_byte(0xe0+r); |
504 |
|
|
emit_byte(i); |
505 |
|
|
} |
506 |
|
|
} |
507 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_l_ri,(RW4 r, IMM i)) |
508 |
|
|
|
509 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_w_ri,(RW2 r, IMM i)) |
510 |
|
|
{ |
511 |
|
|
emit_byte(0x66); |
512 |
|
|
emit_byte(0xc1); |
513 |
|
|
emit_byte(0xe0+r); |
514 |
|
|
emit_byte(i); |
515 |
|
|
} |
516 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_w_ri,(RW2 r, IMM i)) |
517 |
|
|
|
518 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_b_ri,(RW1 r, IMM i)) |
519 |
|
|
{ |
520 |
|
|
if (optimize_shift_once && (i == 1)) { |
521 |
|
|
emit_byte(0xd0); |
522 |
|
|
emit_byte(0xe0+r); |
523 |
|
|
} |
524 |
|
|
else { |
525 |
|
|
emit_byte(0xc0); |
526 |
|
|
emit_byte(0xe0+r); |
527 |
|
|
emit_byte(i); |
528 |
|
|
} |
529 |
|
|
} |
530 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_b_ri,(RW1 r, IMM i)) |
531 |
|
|
|
532 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_l_ri,(RW4 r, IMM i)) |
533 |
|
|
{ |
534 |
|
|
if (optimize_shift_once && (i == 1)) { |
535 |
|
|
emit_byte(0xd1); |
536 |
|
|
emit_byte(0xe8+r); |
537 |
|
|
} |
538 |
|
|
else { |
539 |
|
|
emit_byte(0xc1); |
540 |
|
|
emit_byte(0xe8+r); |
541 |
|
|
emit_byte(i); |
542 |
|
|
} |
543 |
|
|
} |
544 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_l_ri,(RW4 r, IMM i)) |
545 |
|
|
|
546 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_w_ri,(RW2 r, IMM i)) |
547 |
|
|
{ |
548 |
|
|
emit_byte(0x66); |
549 |
|
|
emit_byte(0xc1); |
550 |
|
|
emit_byte(0xe8+r); |
551 |
|
|
emit_byte(i); |
552 |
|
|
} |
553 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_w_ri,(RW2 r, IMM i)) |
554 |
|
|
|
555 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_b_ri,(RW1 r, IMM i)) |
556 |
|
|
{ |
557 |
|
|
if (optimize_shift_once && (i == 1)) { |
558 |
|
|
emit_byte(0xd0); |
559 |
|
|
emit_byte(0xe8+r); |
560 |
|
|
} |
561 |
|
|
else { |
562 |
|
|
emit_byte(0xc0); |
563 |
|
|
emit_byte(0xe8+r); |
564 |
|
|
emit_byte(i); |
565 |
|
|
} |
566 |
|
|
} |
567 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_b_ri,(RW1 r, IMM i)) |
568 |
|
|
|
569 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_l_ri,(RW4 r, IMM i)) |
570 |
|
|
{ |
571 |
|
|
if (optimize_shift_once && (i == 1)) { |
572 |
|
|
emit_byte(0xd1); |
573 |
|
|
emit_byte(0xf8+r); |
574 |
|
|
} |
575 |
|
|
else { |
576 |
|
|
emit_byte(0xc1); |
577 |
|
|
emit_byte(0xf8+r); |
578 |
|
|
emit_byte(i); |
579 |
|
|
} |
580 |
|
|
} |
581 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_l_ri,(RW4 r, IMM i)) |
582 |
|
|
|
583 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_w_ri,(RW2 r, IMM i)) |
584 |
|
|
{ |
585 |
|
|
emit_byte(0x66); |
586 |
|
|
emit_byte(0xc1); |
587 |
|
|
emit_byte(0xf8+r); |
588 |
|
|
emit_byte(i); |
589 |
|
|
} |
590 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_w_ri,(RW2 r, IMM i)) |
591 |
|
|
|
592 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_b_ri,(RW1 r, IMM i)) |
593 |
|
|
{ |
594 |
|
|
if (optimize_shift_once && (i == 1)) { |
595 |
|
|
emit_byte(0xd0); |
596 |
|
|
emit_byte(0xf8+r); |
597 |
|
|
} |
598 |
|
|
else { |
599 |
|
|
emit_byte(0xc0); |
600 |
|
|
emit_byte(0xf8+r); |
601 |
|
|
emit_byte(i); |
602 |
|
|
} |
603 |
|
|
} |
604 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_b_ri,(RW1 r, IMM i)) |
605 |
|
|
|
606 |
|
|
LOWFUNC(WRITE,NONE,1,raw_sahf,(R2 dummy_ah)) |
607 |
|
|
{ |
608 |
|
|
emit_byte(0x9e); |
609 |
|
|
} |
610 |
|
|
LENDFUNC(WRITE,NONE,1,raw_sahf,(R2 dummy_ah)) |
611 |
|
|
|
612 |
|
|
LOWFUNC(NONE,NONE,1,raw_cpuid,(R4 dummy_eax)) |
613 |
|
|
{ |
614 |
|
|
emit_byte(0x0f); |
615 |
|
|
emit_byte(0xa2); |
616 |
|
|
} |
617 |
|
|
LENDFUNC(NONE,NONE,1,raw_cpuid,(R4 dummy_eax)) |
618 |
|
|
|
619 |
|
|
LOWFUNC(READ,NONE,1,raw_lahf,(W2 dummy_ah)) |
620 |
|
|
{ |
621 |
|
|
emit_byte(0x9f); |
622 |
|
|
} |
623 |
|
|
LENDFUNC(READ,NONE,1,raw_lahf,(W2 dummy_ah)) |
624 |
|
|
|
625 |
|
|
LOWFUNC(READ,NONE,2,raw_setcc,(W1 d, IMM cc)) |
626 |
|
|
{ |
627 |
|
|
emit_byte(0x0f); |
628 |
|
|
emit_byte(0x90+cc); |
629 |
|
|
emit_byte(0xc0+d); |
630 |
|
|
} |
631 |
|
|
LENDFUNC(READ,NONE,2,raw_setcc,(W1 d, IMM cc)) |
632 |
|
|
|
633 |
|
|
LOWFUNC(READ,WRITE,2,raw_setcc_m,(MEMW d, IMM cc)) |
634 |
|
|
{ |
635 |
|
|
emit_byte(0x0f); |
636 |
|
|
emit_byte(0x90+cc); |
637 |
|
|
emit_byte(0x05); |
638 |
|
|
emit_long(d); |
639 |
|
|
} |
640 |
|
|
LENDFUNC(READ,WRITE,2,raw_setcc_m,(MEMW d, IMM cc)) |
641 |
|
|
|
642 |
|
|
LOWFUNC(READ,NONE,3,raw_cmov_l_rr,(RW4 d, R4 s, IMM cc)) |
643 |
|
|
{ |
644 |
|
|
if (have_cmov) { |
645 |
|
|
emit_byte(0x0f); |
646 |
|
|
emit_byte(0x40+cc); |
647 |
|
|
emit_byte(0xc0+8*d+s); |
648 |
|
|
} |
649 |
|
|
else { /* replacement using branch and mov */ |
650 |
|
|
int uncc=(cc^1); |
651 |
|
|
emit_byte(0x70+uncc); |
652 |
|
|
emit_byte(2); /* skip next 2 bytes if not cc=true */ |
653 |
|
|
emit_byte(0x89); |
654 |
|
|
emit_byte(0xc0+8*s+d); |
655 |
|
|
} |
656 |
|
|
} |
657 |
|
|
LENDFUNC(READ,NONE,3,raw_cmov_l_rr,(RW4 d, R4 s, IMM cc)) |
658 |
|
|
|
659 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bsf_l_rr,(W4 d, R4 s)) |
660 |
|
|
{ |
661 |
|
|
emit_byte(0x0f); |
662 |
|
|
emit_byte(0xbc); |
663 |
|
|
emit_byte(0xc0+8*d+s); |
664 |
|
|
} |
665 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bsf_l_rr,(W4 d, R4 s)) |
666 |
|
|
|
667 |
|
|
LOWFUNC(NONE,NONE,2,raw_sign_extend_16_rr,(W4 d, R2 s)) |
668 |
|
|
{ |
669 |
|
|
emit_byte(0x0f); |
670 |
|
|
emit_byte(0xbf); |
671 |
|
|
emit_byte(0xc0+8*d+s); |
672 |
|
|
} |
673 |
|
|
LENDFUNC(NONE,NONE,2,raw_sign_extend_16_rr,(W4 d, R2 s)) |
674 |
|
|
|
675 |
|
|
LOWFUNC(NONE,NONE,2,raw_sign_extend_8_rr,(W4 d, R1 s)) |
676 |
|
|
{ |
677 |
|
|
emit_byte(0x0f); |
678 |
|
|
emit_byte(0xbe); |
679 |
|
|
emit_byte(0xc0+8*d+s); |
680 |
|
|
} |
681 |
|
|
LENDFUNC(NONE,NONE,2,raw_sign_extend_8_rr,(W4 d, R1 s)) |
682 |
|
|
|
683 |
|
|
LOWFUNC(NONE,NONE,2,raw_zero_extend_16_rr,(W4 d, R2 s)) |
684 |
|
|
{ |
685 |
|
|
emit_byte(0x0f); |
686 |
|
|
emit_byte(0xb7); |
687 |
|
|
emit_byte(0xc0+8*d+s); |
688 |
|
|
} |
689 |
|
|
LENDFUNC(NONE,NONE,2,raw_zero_extend_16_rr,(W4 d, R2 s)) |
690 |
|
|
|
691 |
|
|
LOWFUNC(NONE,NONE,2,raw_zero_extend_8_rr,(W4 d, R1 s)) |
692 |
|
|
{ |
693 |
|
|
emit_byte(0x0f); |
694 |
|
|
emit_byte(0xb6); |
695 |
|
|
emit_byte(0xc0+8*d+s); |
696 |
|
|
} |
697 |
|
|
LENDFUNC(NONE,NONE,2,raw_zero_extend_8_rr,(W4 d, R1 s)) |
698 |
|
|
|
699 |
|
|
LOWFUNC(NONE,NONE,2,raw_imul_32_32,(RW4 d, R4 s)) |
700 |
|
|
{ |
701 |
|
|
emit_byte(0x0f); |
702 |
|
|
emit_byte(0xaf); |
703 |
|
|
emit_byte(0xc0+8*d+s); |
704 |
|
|
} |
705 |
|
|
LENDFUNC(NONE,NONE,2,raw_imul_32_32,(RW4 d, R4 s)) |
706 |
|
|
|
707 |
|
|
LOWFUNC(NONE,NONE,2,raw_imul_64_32,(RW4 d, RW4 s)) |
708 |
|
|
{ |
709 |
|
|
if (d!=MUL_NREG1 || s!=MUL_NREG2) |
710 |
|
|
abort(); |
711 |
|
|
emit_byte(0xf7); |
712 |
|
|
emit_byte(0xea); |
713 |
|
|
} |
714 |
|
|
LENDFUNC(NONE,NONE,2,raw_imul_64_32,(RW4 d, RW4 s)) |
715 |
|
|
|
716 |
|
|
LOWFUNC(NONE,NONE,2,raw_mul_64_32,(RW4 d, RW4 s)) |
717 |
|
|
{ |
718 |
|
|
if (d!=MUL_NREG1 || s!=MUL_NREG2) { |
719 |
|
|
printf("Bad register in MUL: d=%d, s=%d\n",d,s); |
720 |
|
|
abort(); |
721 |
|
|
} |
722 |
|
|
emit_byte(0xf7); |
723 |
|
|
emit_byte(0xe2); |
724 |
|
|
} |
725 |
|
|
LENDFUNC(NONE,NONE,2,raw_mul_64_32,(RW4 d, RW4 s)) |
726 |
|
|
|
727 |
|
|
LOWFUNC(NONE,NONE,2,raw_mul_32_32,(RW4 d, R4 s)) |
728 |
|
|
{ |
729 |
|
|
abort(); /* %^$&%^$%#^ x86! */ |
730 |
|
|
emit_byte(0x0f); |
731 |
|
|
emit_byte(0xaf); |
732 |
|
|
emit_byte(0xc0+8*d+s); |
733 |
|
|
} |
734 |
|
|
LENDFUNC(NONE,NONE,2,raw_mul_32_32,(RW4 d, R4 s)) |
735 |
|
|
|
736 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_b_rr,(W1 d, R1 s)) |
737 |
|
|
{ |
738 |
|
|
emit_byte(0x88); |
739 |
|
|
emit_byte(0xc0+8*s+d); |
740 |
|
|
} |
741 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_b_rr,(W1 d, R1 s)) |
742 |
|
|
|
743 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_w_rr,(W2 d, R2 s)) |
744 |
|
|
{ |
745 |
|
|
emit_byte(0x66); |
746 |
|
|
emit_byte(0x89); |
747 |
|
|
emit_byte(0xc0+8*s+d); |
748 |
|
|
} |
749 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_w_rr,(W2 d, R2 s)) |
750 |
|
|
|
751 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_l_rrm_indexed,(W4 d,R4 baser, R4 index, IMM factor)) |
752 |
|
|
{ |
753 |
|
|
int isebp=(baser==5)?0x40:0; |
754 |
|
|
int fi; |
755 |
|
|
|
756 |
|
|
switch(factor) { |
757 |
|
|
case 1: fi=0; break; |
758 |
|
|
case 2: fi=1; break; |
759 |
|
|
case 4: fi=2; break; |
760 |
|
|
case 8: fi=3; break; |
761 |
|
|
default: abort(); |
762 |
|
|
} |
763 |
|
|
|
764 |
|
|
|
765 |
|
|
emit_byte(0x8b); |
766 |
|
|
emit_byte(0x04+8*d+isebp); |
767 |
|
|
emit_byte(baser+8*index+0x40*fi); |
768 |
|
|
if (isebp) |
769 |
|
|
emit_byte(0x00); |
770 |
|
|
} |
771 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_l_rrm_indexed,(W4 d,R4 baser, R4 index, IMM factor)) |
772 |
|
|
|
773 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_w_rrm_indexed,(W2 d, R4 baser, R4 index, IMM factor)) |
774 |
|
|
{ |
775 |
|
|
int fi; |
776 |
|
|
int isebp; |
777 |
|
|
|
778 |
|
|
switch(factor) { |
779 |
|
|
case 1: fi=0; break; |
780 |
|
|
case 2: fi=1; break; |
781 |
|
|
case 4: fi=2; break; |
782 |
|
|
case 8: fi=3; break; |
783 |
|
|
default: abort(); |
784 |
|
|
} |
785 |
|
|
isebp=(baser==5)?0x40:0; |
786 |
|
|
|
787 |
|
|
emit_byte(0x66); |
788 |
|
|
emit_byte(0x8b); |
789 |
|
|
emit_byte(0x04+8*d+isebp); |
790 |
|
|
emit_byte(baser+8*index+0x40*fi); |
791 |
|
|
if (isebp) |
792 |
|
|
emit_byte(0x00); |
793 |
|
|
} |
794 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_w_rrm_indexed,(W2 d, R4 baser, R4 index, IMM factor)) |
795 |
|
|
|
796 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_b_rrm_indexed,(W1 d, R4 baser, R4 index, IMM factor)) |
797 |
|
|
{ |
798 |
|
|
int fi; |
799 |
|
|
int isebp; |
800 |
|
|
|
801 |
|
|
switch(factor) { |
802 |
|
|
case 1: fi=0; break; |
803 |
|
|
case 2: fi=1; break; |
804 |
|
|
case 4: fi=2; break; |
805 |
|
|
case 8: fi=3; break; |
806 |
|
|
default: abort(); |
807 |
|
|
} |
808 |
|
|
isebp=(baser==5)?0x40:0; |
809 |
|
|
|
810 |
|
|
emit_byte(0x8a); |
811 |
|
|
emit_byte(0x04+8*d+isebp); |
812 |
|
|
emit_byte(baser+8*index+0x40*fi); |
813 |
|
|
if (isebp) |
814 |
|
|
emit_byte(0x00); |
815 |
|
|
} |
816 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_b_rrm_indexed,(W1 d, R4 baser, R4 index, IMM factor)) |
817 |
|
|
|
818 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_l_mrr_indexed,(R4 baser, R4 index, IMM factor, R4 s)) |
819 |
|
|
{ |
820 |
|
|
int fi; |
821 |
|
|
int isebp; |
822 |
|
|
|
823 |
|
|
switch(factor) { |
824 |
|
|
case 1: fi=0; break; |
825 |
|
|
case 2: fi=1; break; |
826 |
|
|
case 4: fi=2; break; |
827 |
|
|
case 8: fi=3; break; |
828 |
|
|
default: abort(); |
829 |
|
|
} |
830 |
|
|
|
831 |
|
|
|
832 |
|
|
isebp=(baser==5)?0x40:0; |
833 |
|
|
|
834 |
|
|
emit_byte(0x89); |
835 |
|
|
emit_byte(0x04+8*s+isebp); |
836 |
|
|
emit_byte(baser+8*index+0x40*fi); |
837 |
|
|
if (isebp) |
838 |
|
|
emit_byte(0x00); |
839 |
|
|
} |
840 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_l_mrr_indexed,(R4 baser, R4 index, IMM factor, R4 s)) |
841 |
|
|
|
842 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_w_mrr_indexed,(R4 baser, R4 index, IMM factor, R2 s)) |
843 |
|
|
{ |
844 |
|
|
int fi; |
845 |
|
|
int isebp; |
846 |
|
|
|
847 |
|
|
switch(factor) { |
848 |
|
|
case 1: fi=0; break; |
849 |
|
|
case 2: fi=1; break; |
850 |
|
|
case 4: fi=2; break; |
851 |
|
|
case 8: fi=3; break; |
852 |
|
|
default: abort(); |
853 |
|
|
} |
854 |
|
|
isebp=(baser==5)?0x40:0; |
855 |
|
|
|
856 |
|
|
emit_byte(0x66); |
857 |
|
|
emit_byte(0x89); |
858 |
|
|
emit_byte(0x04+8*s+isebp); |
859 |
|
|
emit_byte(baser+8*index+0x40*fi); |
860 |
|
|
if (isebp) |
861 |
|
|
emit_byte(0x00); |
862 |
|
|
} |
863 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_w_mrr_indexed,(R4 baser, R4 index, IMM factor, R2 s)) |
864 |
|
|
|
865 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_b_mrr_indexed,(R4 baser, R4 index, IMM factor, R1 s)) |
866 |
|
|
{ |
867 |
|
|
int fi; |
868 |
|
|
int isebp; |
869 |
|
|
|
870 |
|
|
switch(factor) { |
871 |
|
|
case 1: fi=0; break; |
872 |
|
|
case 2: fi=1; break; |
873 |
|
|
case 4: fi=2; break; |
874 |
|
|
case 8: fi=3; break; |
875 |
|
|
default: abort(); |
876 |
|
|
} |
877 |
|
|
isebp=(baser==5)?0x40:0; |
878 |
|
|
|
879 |
|
|
emit_byte(0x88); |
880 |
|
|
emit_byte(0x04+8*s+isebp); |
881 |
|
|
emit_byte(baser+8*index+0x40*fi); |
882 |
|
|
if (isebp) |
883 |
|
|
emit_byte(0x00); |
884 |
|
|
} |
885 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_b_mrr_indexed,(R4 baser, R4 index, IMM factor, R1 s)) |
886 |
|
|
|
887 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_l_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R4 s)) |
888 |
|
|
{ |
889 |
|
|
int fi; |
890 |
|
|
|
891 |
|
|
switch(factor) { |
892 |
|
|
case 1: fi=0; break; |
893 |
|
|
case 2: fi=1; break; |
894 |
|
|
case 4: fi=2; break; |
895 |
|
|
case 8: fi=3; break; |
896 |
|
|
default: abort(); |
897 |
|
|
} |
898 |
|
|
|
899 |
|
|
emit_byte(0x89); |
900 |
|
|
emit_byte(0x84+8*s); |
901 |
|
|
emit_byte(baser+8*index+0x40*fi); |
902 |
|
|
emit_long(base); |
903 |
|
|
} |
904 |
|
|
LENDFUNC(NONE,WRITE,5,raw_mov_l_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R4 s)) |
905 |
|
|
|
906 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_w_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R2 s)) |
907 |
|
|
{ |
908 |
|
|
int fi; |
909 |
|
|
|
910 |
|
|
switch(factor) { |
911 |
|
|
case 1: fi=0; break; |
912 |
|
|
case 2: fi=1; break; |
913 |
|
|
case 4: fi=2; break; |
914 |
|
|
case 8: fi=3; break; |
915 |
|
|
default: abort(); |
916 |
|
|
} |
917 |
|
|
|
918 |
|
|
emit_byte(0x66); |
919 |
|
|
emit_byte(0x89); |
920 |
|
|
emit_byte(0x84+8*s); |
921 |
|
|
emit_byte(baser+8*index+0x40*fi); |
922 |
|
|
emit_long(base); |
923 |
|
|
} |
924 |
|
|
LENDFUNC(NONE,WRITE,5,raw_mov_w_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R2 s)) |
925 |
|
|
|
926 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_b_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R1 s)) |
927 |
|
|
{ |
928 |
|
|
int fi; |
929 |
|
|
|
930 |
|
|
switch(factor) { |
931 |
|
|
case 1: fi=0; break; |
932 |
|
|
case 2: fi=1; break; |
933 |
|
|
case 4: fi=2; break; |
934 |
|
|
case 8: fi=3; break; |
935 |
|
|
default: abort(); |
936 |
|
|
} |
937 |
|
|
|
938 |
|
|
emit_byte(0x88); |
939 |
|
|
emit_byte(0x84+8*s); |
940 |
|
|
emit_byte(baser+8*index+0x40*fi); |
941 |
|
|
emit_long(base); |
942 |
|
|
} |
943 |
|
|
LENDFUNC(NONE,WRITE,5,raw_mov_b_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R1 s)) |
944 |
|
|
|
945 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_l_brrm_indexed,(W4 d, IMM base, R4 baser, R4 index, IMM factor)) |
946 |
|
|
{ |
947 |
|
|
int fi; |
948 |
|
|
|
949 |
|
|
switch(factor) { |
950 |
|
|
case 1: fi=0; break; |
951 |
|
|
case 2: fi=1; break; |
952 |
|
|
case 4: fi=2; break; |
953 |
|
|
case 8: fi=3; break; |
954 |
|
|
default: abort(); |
955 |
|
|
} |
956 |
|
|
|
957 |
|
|
emit_byte(0x8b); |
958 |
|
|
emit_byte(0x84+8*d); |
959 |
|
|
emit_byte(baser+8*index+0x40*fi); |
960 |
|
|
emit_long(base); |
961 |
|
|
} |
962 |
|
|
LENDFUNC(NONE,READ,5,raw_mov_l_brrm_indexed,(W4 d, IMM base, R4 baser, R4 index, IMM factor)) |
963 |
|
|
|
964 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_w_brrm_indexed,(W2 d, IMM base, R4 baser, R4 index, IMM factor)) |
965 |
|
|
{ |
966 |
|
|
int fi; |
967 |
|
|
|
968 |
|
|
switch(factor) { |
969 |
|
|
case 1: fi=0; break; |
970 |
|
|
case 2: fi=1; break; |
971 |
|
|
case 4: fi=2; break; |
972 |
|
|
case 8: fi=3; break; |
973 |
|
|
default: abort(); |
974 |
|
|
} |
975 |
|
|
|
976 |
|
|
emit_byte(0x66); |
977 |
|
|
emit_byte(0x8b); |
978 |
|
|
emit_byte(0x84+8*d); |
979 |
|
|
emit_byte(baser+8*index+0x40*fi); |
980 |
|
|
emit_long(base); |
981 |
|
|
} |
982 |
|
|
LENDFUNC(NONE,READ,5,raw_mov_w_brrm_indexed,(W2 d, IMM base, R4 baser, R4 index, IMM factor)) |
983 |
|
|
|
984 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_b_brrm_indexed,(W1 d, IMM base, R4 baser, R4 index, IMM factor)) |
985 |
|
|
{ |
986 |
|
|
int fi; |
987 |
|
|
|
988 |
|
|
switch(factor) { |
989 |
|
|
case 1: fi=0; break; |
990 |
|
|
case 2: fi=1; break; |
991 |
|
|
case 4: fi=2; break; |
992 |
|
|
case 8: fi=3; break; |
993 |
|
|
default: abort(); |
994 |
|
|
} |
995 |
|
|
|
996 |
|
|
emit_byte(0x8a); |
997 |
|
|
emit_byte(0x84+8*d); |
998 |
|
|
emit_byte(baser+8*index+0x40*fi); |
999 |
|
|
emit_long(base); |
1000 |
|
|
} |
1001 |
|
|
LENDFUNC(NONE,READ,5,raw_mov_b_brrm_indexed,(W1 d, IMM base, R4 baser, R4 index, IMM factor)) |
1002 |
|
|
|
1003 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor)) |
1004 |
|
|
{ |
1005 |
|
|
int fi; |
1006 |
|
|
switch(factor) { |
1007 |
|
|
case 1: fi=0; break; |
1008 |
|
|
case 2: fi=1; break; |
1009 |
|
|
case 4: fi=2; break; |
1010 |
|
|
case 8: fi=3; break; |
1011 |
|
|
default: |
1012 |
|
|
fprintf(stderr,"Bad factor %d in mov_l_rm_indexed!\n",factor); |
1013 |
|
|
abort(); |
1014 |
|
|
} |
1015 |
|
|
emit_byte(0x8b); |
1016 |
|
|
emit_byte(0x04+8*d); |
1017 |
|
|
emit_byte(0x05+8*index+64*fi); |
1018 |
|
|
emit_long(base); |
1019 |
|
|
} |
1020 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor)) |
1021 |
|
|
|
1022 |
|
|
LOWFUNC(NONE,READ,5,raw_cmov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor, IMM cond)) |
1023 |
|
|
{ |
1024 |
|
|
int fi; |
1025 |
|
|
switch(factor) { |
1026 |
|
|
case 1: fi=0; break; |
1027 |
|
|
case 2: fi=1; break; |
1028 |
|
|
case 4: fi=2; break; |
1029 |
|
|
case 8: fi=3; break; |
1030 |
|
|
default: |
1031 |
|
|
fprintf(stderr,"Bad factor %d in mov_l_rm_indexed!\n",factor); |
1032 |
|
|
abort(); |
1033 |
|
|
} |
1034 |
|
|
if (have_cmov) { |
1035 |
|
|
emit_byte(0x0f); |
1036 |
|
|
emit_byte(0x40+cond); |
1037 |
|
|
emit_byte(0x04+8*d); |
1038 |
|
|
emit_byte(0x05+8*index+64*fi); |
1039 |
|
|
emit_long(base); |
1040 |
|
|
} |
1041 |
|
|
else { /* replacement using branch and mov */ |
1042 |
|
|
int uncc=(cond^1); |
1043 |
|
|
emit_byte(0x70+uncc); |
1044 |
|
|
emit_byte(7); /* skip next 7 bytes if not cc=true */ |
1045 |
|
|
emit_byte(0x8b); |
1046 |
|
|
emit_byte(0x04+8*d); |
1047 |
|
|
emit_byte(0x05+8*index+64*fi); |
1048 |
|
|
emit_long(base); |
1049 |
|
|
} |
1050 |
|
|
} |
1051 |
|
|
LENDFUNC(NONE,READ,5,raw_cmov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor, IMM cond)) |
1052 |
|
|
|
1053 |
|
|
LOWFUNC(NONE,READ,3,raw_cmov_l_rm,(W4 d, IMM mem, IMM cond)) |
1054 |
|
|
{ |
1055 |
|
|
if (have_cmov) { |
1056 |
|
|
emit_byte(0x0f); |
1057 |
|
|
emit_byte(0x40+cond); |
1058 |
|
|
emit_byte(0x05+8*d); |
1059 |
|
|
emit_long(mem); |
1060 |
|
|
} |
1061 |
|
|
else { /* replacement using branch and mov */ |
1062 |
|
|
int uncc=(cond^1); |
1063 |
|
|
emit_byte(0x70+uncc); |
1064 |
|
|
emit_byte(6); /* skip next 6 bytes if not cc=true */ |
1065 |
|
|
emit_byte(0x8b); |
1066 |
|
|
emit_byte(0x05+8*d); |
1067 |
|
|
emit_long(mem); |
1068 |
|
|
} |
1069 |
|
|
} |
1070 |
|
|
LENDFUNC(NONE,READ,3,raw_cmov_l_rm,(W4 d, IMM mem, IMM cond)) |
1071 |
|
|
|
1072 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
1073 |
|
|
{ |
1074 |
|
|
emit_byte(0x8b); |
1075 |
|
|
emit_byte(0x40+8*d+s); |
1076 |
|
|
emit_byte(offset); |
1077 |
|
|
} |
1078 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
1079 |
|
|
|
1080 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
1081 |
|
|
{ |
1082 |
|
|
emit_byte(0x66); |
1083 |
|
|
emit_byte(0x8b); |
1084 |
|
|
emit_byte(0x40+8*d+s); |
1085 |
|
|
emit_byte(offset); |
1086 |
|
|
} |
1087 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
1088 |
|
|
|
1089 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
1090 |
|
|
{ |
1091 |
|
|
emit_byte(0x8a); |
1092 |
|
|
emit_byte(0x40+8*d+s); |
1093 |
|
|
emit_byte(offset); |
1094 |
|
|
} |
1095 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
1096 |
|
|
|
1097 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_l_brR,(W4 d, R4 s, IMM offset)) |
1098 |
|
|
{ |
1099 |
|
|
emit_byte(0x8b); |
1100 |
|
|
emit_byte(0x80+8*d+s); |
1101 |
|
|
emit_long(offset); |
1102 |
|
|
} |
1103 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_l_brR,(W4 d, R4 s, IMM offset)) |
1104 |
|
|
|
1105 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_w_brR,(W2 d, R4 s, IMM offset)) |
1106 |
|
|
{ |
1107 |
|
|
emit_byte(0x66); |
1108 |
|
|
emit_byte(0x8b); |
1109 |
|
|
emit_byte(0x80+8*d+s); |
1110 |
|
|
emit_long(offset); |
1111 |
|
|
} |
1112 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_w_brR,(W2 d, R4 s, IMM offset)) |
1113 |
|
|
|
1114 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_b_brR,(W1 d, R4 s, IMM offset)) |
1115 |
|
|
{ |
1116 |
|
|
emit_byte(0x8a); |
1117 |
|
|
emit_byte(0x80+8*d+s); |
1118 |
|
|
emit_long(offset); |
1119 |
|
|
} |
1120 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_b_brR,(W1 d, R4 s, IMM offset)) |
1121 |
|
|
|
1122 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
1123 |
|
|
{ |
1124 |
|
|
emit_byte(0xc7); |
1125 |
|
|
emit_byte(0x40+d); |
1126 |
|
|
emit_byte(offset); |
1127 |
|
|
emit_long(i); |
1128 |
|
|
} |
1129 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
1130 |
|
|
|
1131 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
1132 |
|
|
{ |
1133 |
|
|
emit_byte(0x66); |
1134 |
|
|
emit_byte(0xc7); |
1135 |
|
|
emit_byte(0x40+d); |
1136 |
|
|
emit_byte(offset); |
1137 |
|
|
emit_word(i); |
1138 |
|
|
} |
1139 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
1140 |
|
|
|
1141 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
1142 |
|
|
{ |
1143 |
|
|
emit_byte(0xc6); |
1144 |
|
|
emit_byte(0x40+d); |
1145 |
|
|
emit_byte(offset); |
1146 |
|
|
emit_byte(i); |
1147 |
|
|
} |
1148 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
1149 |
|
|
|
1150 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
1151 |
|
|
{ |
1152 |
|
|
emit_byte(0x89); |
1153 |
|
|
emit_byte(0x40+8*s+d); |
1154 |
|
|
emit_byte(offset); |
1155 |
|
|
} |
1156 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
1157 |
|
|
|
1158 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
1159 |
|
|
{ |
1160 |
|
|
emit_byte(0x66); |
1161 |
|
|
emit_byte(0x89); |
1162 |
|
|
emit_byte(0x40+8*s+d); |
1163 |
|
|
emit_byte(offset); |
1164 |
|
|
} |
1165 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
1166 |
|
|
|
1167 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
1168 |
|
|
{ |
1169 |
|
|
emit_byte(0x88); |
1170 |
|
|
emit_byte(0x40+8*s+d); |
1171 |
|
|
emit_byte(offset); |
1172 |
|
|
} |
1173 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
1174 |
|
|
|
1175 |
|
|
LOWFUNC(NONE,NONE,3,raw_lea_l_brr,(W4 d, R4 s, IMM offset)) |
1176 |
|
|
{ |
1177 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1178 |
|
|
emit_byte(0x8d); |
1179 |
|
|
emit_byte(0x40+8*d+s); |
1180 |
|
|
emit_byte(offset); |
1181 |
|
|
} |
1182 |
|
|
else { |
1183 |
|
|
emit_byte(0x8d); |
1184 |
|
|
emit_byte(0x80+8*d+s); |
1185 |
|
|
emit_long(offset); |
1186 |
|
|
} |
1187 |
|
|
} |
1188 |
|
|
LENDFUNC(NONE,NONE,3,raw_lea_l_brr,(W4 d, R4 s, IMM offset)) |
1189 |
|
|
|
1190 |
|
|
LOWFUNC(NONE,NONE,5,raw_lea_l_brr_indexed,(W4 d, R4 s, R4 index, IMM factor, IMM offset)) |
1191 |
|
|
{ |
1192 |
|
|
int fi; |
1193 |
|
|
|
1194 |
|
|
switch(factor) { |
1195 |
|
|
case 1: fi=0; break; |
1196 |
|
|
case 2: fi=1; break; |
1197 |
|
|
case 4: fi=2; break; |
1198 |
|
|
case 8: fi=3; break; |
1199 |
|
|
default: abort(); |
1200 |
|
|
} |
1201 |
|
|
|
1202 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1203 |
|
|
emit_byte(0x8d); |
1204 |
|
|
emit_byte(0x44+8*d); |
1205 |
|
|
emit_byte(0x40*fi+8*index+s); |
1206 |
|
|
emit_byte(offset); |
1207 |
|
|
} |
1208 |
|
|
else { |
1209 |
|
|
emit_byte(0x8d); |
1210 |
|
|
emit_byte(0x84+8*d); |
1211 |
|
|
emit_byte(0x40*fi+8*index+s); |
1212 |
|
|
emit_long(offset); |
1213 |
|
|
} |
1214 |
|
|
} |
1215 |
|
|
LENDFUNC(NONE,NONE,5,raw_lea_l_brr_indexed,(W4 d, R4 s, R4 index, IMM factor, IMM offset)) |
1216 |
|
|
|
1217 |
|
|
LOWFUNC(NONE,NONE,4,raw_lea_l_rr_indexed,(W4 d, R4 s, R4 index, IMM factor)) |
1218 |
|
|
{ |
1219 |
|
|
int isebp=(s==5)?0x40:0; |
1220 |
|
|
int fi; |
1221 |
|
|
|
1222 |
|
|
switch(factor) { |
1223 |
|
|
case 1: fi=0; break; |
1224 |
|
|
case 2: fi=1; break; |
1225 |
|
|
case 4: fi=2; break; |
1226 |
|
|
case 8: fi=3; break; |
1227 |
|
|
default: abort(); |
1228 |
|
|
} |
1229 |
|
|
|
1230 |
|
|
emit_byte(0x8d); |
1231 |
|
|
emit_byte(0x04+8*d+isebp); |
1232 |
|
|
emit_byte(0x40*fi+8*index+s); |
1233 |
|
|
if (isebp) |
1234 |
|
|
emit_byte(0); |
1235 |
|
|
} |
1236 |
|
|
LENDFUNC(NONE,NONE,4,raw_lea_l_rr_indexed,(W4 d, R4 s, R4 index, IMM factor)) |
1237 |
|
|
|
1238 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_bRr,(R4 d, R4 s, IMM offset)) |
1239 |
|
|
{ |
1240 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1241 |
|
|
emit_byte(0x89); |
1242 |
|
|
emit_byte(0x40+8*s+d); |
1243 |
|
|
emit_byte(offset); |
1244 |
|
|
} |
1245 |
|
|
else { |
1246 |
|
|
emit_byte(0x89); |
1247 |
|
|
emit_byte(0x80+8*s+d); |
1248 |
|
|
emit_long(offset); |
1249 |
|
|
} |
1250 |
|
|
} |
1251 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_bRr,(R4 d, R4 s, IMM offset)) |
1252 |
|
|
|
1253 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_bRr,(R4 d, R2 s, IMM offset)) |
1254 |
|
|
{ |
1255 |
|
|
emit_byte(0x66); |
1256 |
|
|
emit_byte(0x89); |
1257 |
|
|
emit_byte(0x80+8*s+d); |
1258 |
|
|
emit_long(offset); |
1259 |
|
|
} |
1260 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_bRr,(R4 d, R2 s, IMM offset)) |
1261 |
|
|
|
1262 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_bRr,(R4 d, R1 s, IMM offset)) |
1263 |
|
|
{ |
1264 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1265 |
|
|
emit_byte(0x88); |
1266 |
|
|
emit_byte(0x40+8*s+d); |
1267 |
|
|
emit_byte(offset); |
1268 |
|
|
} |
1269 |
|
|
else { |
1270 |
|
|
emit_byte(0x88); |
1271 |
|
|
emit_byte(0x80+8*s+d); |
1272 |
|
|
emit_long(offset); |
1273 |
|
|
} |
1274 |
|
|
} |
1275 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_bRr,(R4 d, R1 s, IMM offset)) |
1276 |
|
|
|
1277 |
|
|
LOWFUNC(NONE,NONE,1,raw_bswap_32,(RW4 r)) |
1278 |
|
|
{ |
1279 |
|
|
emit_byte(0x0f); |
1280 |
|
|
emit_byte(0xc8+r); |
1281 |
|
|
} |
1282 |
|
|
LENDFUNC(NONE,NONE,1,raw_bswap_32,(RW4 r)) |
1283 |
|
|
|
1284 |
|
|
LOWFUNC(WRITE,NONE,1,raw_bswap_16,(RW2 r)) |
1285 |
|
|
{ |
1286 |
|
|
emit_byte(0x66); |
1287 |
|
|
emit_byte(0xc1); |
1288 |
|
|
emit_byte(0xc0+r); |
1289 |
|
|
emit_byte(0x08); |
1290 |
|
|
} |
1291 |
|
|
LENDFUNC(WRITE,NONE,1,raw_bswap_16,(RW2 r)) |
1292 |
|
|
|
1293 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_l_rr,(W4 d, R4 s)) |
1294 |
|
|
{ |
1295 |
|
|
emit_byte(0x89); |
1296 |
|
|
emit_byte(0xc0+8*s+d); |
1297 |
|
|
} |
1298 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_l_rr,(W4 d, R4 s)) |
1299 |
|
|
|
1300 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_l_mr,(IMM d, R4 s)) |
1301 |
|
|
{ |
1302 |
|
|
emit_byte(0x89); |
1303 |
|
|
emit_byte(0x05+8*s); |
1304 |
|
|
emit_long(d); |
1305 |
|
|
} |
1306 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_l_mr,(IMM d, R4 s)) |
1307 |
|
|
|
1308 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_w_mr,(IMM d, R2 s)) |
1309 |
|
|
{ |
1310 |
|
|
emit_byte(0x66); |
1311 |
|
|
emit_byte(0x89); |
1312 |
|
|
emit_byte(0x05+8*s); |
1313 |
|
|
emit_long(d); |
1314 |
|
|
} |
1315 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_w_mr,(IMM d, R2 s)) |
1316 |
|
|
|
1317 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_w_rm,(W2 d, IMM s)) |
1318 |
|
|
{ |
1319 |
|
|
emit_byte(0x66); |
1320 |
|
|
emit_byte(0x8b); |
1321 |
|
|
emit_byte(0x05+8*d); |
1322 |
|
|
emit_long(s); |
1323 |
|
|
} |
1324 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_w_rm,(W2 d, IMM s)) |
1325 |
|
|
|
1326 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_b_mr,(IMM d, R1 s)) |
1327 |
|
|
{ |
1328 |
|
|
emit_byte(0x88); |
1329 |
|
|
emit_byte(0x05+8*s); |
1330 |
|
|
emit_long(d); |
1331 |
|
|
} |
1332 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_b_mr,(IMM d, R1 s)) |
1333 |
|
|
|
1334 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_b_rm,(W1 d, IMM s)) |
1335 |
|
|
{ |
1336 |
|
|
emit_byte(0x8a); |
1337 |
|
|
emit_byte(0x05+8*d); |
1338 |
|
|
emit_long(s); |
1339 |
|
|
} |
1340 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_b_rm,(W1 d, IMM s)) |
1341 |
|
|
|
1342 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_l_ri,(W4 d, IMM s)) |
1343 |
|
|
{ |
1344 |
|
|
emit_byte(0xb8+d); |
1345 |
|
|
emit_long(s); |
1346 |
|
|
} |
1347 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_l_ri,(W4 d, IMM s)) |
1348 |
|
|
|
1349 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_w_ri,(W2 d, IMM s)) |
1350 |
|
|
{ |
1351 |
|
|
emit_byte(0x66); |
1352 |
|
|
emit_byte(0xb8+d); |
1353 |
|
|
emit_word(s); |
1354 |
|
|
} |
1355 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_w_ri,(W2 d, IMM s)) |
1356 |
|
|
|
1357 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_b_ri,(W1 d, IMM s)) |
1358 |
|
|
{ |
1359 |
|
|
emit_byte(0xb0+d); |
1360 |
|
|
emit_byte(s); |
1361 |
|
|
} |
1362 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_b_ri,(W1 d, IMM s)) |
1363 |
|
|
|
1364 |
|
|
LOWFUNC(RMW,RMW,2,raw_adc_l_mi,(MEMRW d, IMM s)) |
1365 |
|
|
{ |
1366 |
|
|
emit_byte(0x81); |
1367 |
|
|
emit_byte(0x15); |
1368 |
|
|
emit_long(d); |
1369 |
|
|
emit_long(s); |
1370 |
|
|
} |
1371 |
|
|
LENDFUNC(RMW,RMW,2,raw_adc_l_mi,(MEMRW d, IMM s)) |
1372 |
|
|
|
1373 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_l_mi,(IMM d, IMM s)) |
1374 |
|
|
{ |
1375 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1376 |
|
|
emit_byte(0x83); |
1377 |
|
|
emit_byte(0x05); |
1378 |
|
|
emit_long(d); |
1379 |
|
|
emit_byte(s); |
1380 |
|
|
} |
1381 |
|
|
else { |
1382 |
|
|
emit_byte(0x81); |
1383 |
|
|
emit_byte(0x05); |
1384 |
|
|
emit_long(d); |
1385 |
|
|
emit_long(s); |
1386 |
|
|
} |
1387 |
|
|
} |
1388 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_l_mi,(IMM d, IMM s)) |
1389 |
|
|
|
1390 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_w_mi,(IMM d, IMM s)) |
1391 |
|
|
{ |
1392 |
|
|
emit_byte(0x66); |
1393 |
|
|
emit_byte(0x81); |
1394 |
|
|
emit_byte(0x05); |
1395 |
|
|
emit_long(d); |
1396 |
|
|
emit_word(s); |
1397 |
|
|
} |
1398 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_w_mi,(IMM d, IMM s)) |
1399 |
|
|
|
1400 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_b_mi,(IMM d, IMM s)) |
1401 |
|
|
{ |
1402 |
|
|
emit_byte(0x80); |
1403 |
|
|
emit_byte(0x05); |
1404 |
|
|
emit_long(d); |
1405 |
|
|
emit_byte(s); |
1406 |
|
|
} |
1407 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_b_mi,(IMM d, IMM s)) |
1408 |
|
|
|
1409 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_l_ri,(R4 d, IMM i)) |
1410 |
|
|
{ |
1411 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1412 |
|
|
emit_byte(0xa9); |
1413 |
|
|
else { |
1414 |
gbeauche |
1.1 |
emit_byte(0xf7); |
1415 |
|
|
emit_byte(0xc0+d); |
1416 |
gbeauche |
1.2 |
} |
1417 |
gbeauche |
1.1 |
emit_long(i); |
1418 |
|
|
} |
1419 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_l_ri,(R4 d, IMM i)) |
1420 |
|
|
|
1421 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_l_rr,(R4 d, R4 s)) |
1422 |
|
|
{ |
1423 |
|
|
emit_byte(0x85); |
1424 |
|
|
emit_byte(0xc0+8*s+d); |
1425 |
|
|
} |
1426 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_l_rr,(R4 d, R4 s)) |
1427 |
|
|
|
1428 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_w_rr,(R2 d, R2 s)) |
1429 |
|
|
{ |
1430 |
|
|
emit_byte(0x66); |
1431 |
|
|
emit_byte(0x85); |
1432 |
|
|
emit_byte(0xc0+8*s+d); |
1433 |
|
|
} |
1434 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_w_rr,(R2 d, R2 s)) |
1435 |
|
|
|
1436 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_b_rr,(R1 d, R1 s)) |
1437 |
|
|
{ |
1438 |
|
|
emit_byte(0x84); |
1439 |
|
|
emit_byte(0xc0+8*s+d); |
1440 |
|
|
} |
1441 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_b_rr,(R1 d, R1 s)) |
1442 |
|
|
|
1443 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_l_ri,(RW4 d, IMM i)) |
1444 |
|
|
{ |
1445 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1446 |
gbeauche |
1.2 |
emit_byte(0x83); |
1447 |
|
|
emit_byte(0xe0+d); |
1448 |
|
|
emit_byte(i); |
1449 |
gbeauche |
1.1 |
} |
1450 |
|
|
else { |
1451 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1452 |
|
|
emit_byte(0x25); |
1453 |
|
|
else { |
1454 |
|
|
emit_byte(0x81); |
1455 |
|
|
emit_byte(0xe0+d); |
1456 |
|
|
} |
1457 |
|
|
emit_long(i); |
1458 |
gbeauche |
1.1 |
} |
1459 |
|
|
} |
1460 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_l_ri,(RW4 d, IMM i)) |
1461 |
|
|
|
1462 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_w_ri,(RW2 d, IMM i)) |
1463 |
|
|
{ |
1464 |
gbeauche |
1.2 |
emit_byte(0x66); |
1465 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1466 |
|
|
emit_byte(0x83); |
1467 |
|
|
emit_byte(0xe0+d); |
1468 |
|
|
emit_byte(i); |
1469 |
|
|
} |
1470 |
|
|
else { |
1471 |
|
|
if (optimize_accum && isaccum(d)) |
1472 |
|
|
emit_byte(0x25); |
1473 |
|
|
else { |
1474 |
|
|
emit_byte(0x81); |
1475 |
|
|
emit_byte(0xe0+d); |
1476 |
|
|
} |
1477 |
|
|
emit_word(i); |
1478 |
|
|
} |
1479 |
gbeauche |
1.1 |
} |
1480 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_w_ri,(RW2 d, IMM i)) |
1481 |
|
|
|
1482 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_l,(RW4 d, R4 s)) |
1483 |
|
|
{ |
1484 |
|
|
emit_byte(0x21); |
1485 |
|
|
emit_byte(0xc0+8*s+d); |
1486 |
|
|
} |
1487 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_l,(RW4 d, R4 s)) |
1488 |
|
|
|
1489 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_w,(RW2 d, R2 s)) |
1490 |
|
|
{ |
1491 |
|
|
emit_byte(0x66); |
1492 |
|
|
emit_byte(0x21); |
1493 |
|
|
emit_byte(0xc0+8*s+d); |
1494 |
|
|
} |
1495 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_w,(RW2 d, R2 s)) |
1496 |
|
|
|
1497 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_b,(RW1 d, R1 s)) |
1498 |
|
|
{ |
1499 |
|
|
emit_byte(0x20); |
1500 |
|
|
emit_byte(0xc0+8*s+d); |
1501 |
|
|
} |
1502 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_b,(RW1 d, R1 s)) |
1503 |
|
|
|
1504 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_l_ri,(RW4 d, IMM i)) |
1505 |
|
|
{ |
1506 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1507 |
|
|
emit_byte(0x83); |
1508 |
|
|
emit_byte(0xc8+d); |
1509 |
|
|
emit_byte(i); |
1510 |
|
|
} |
1511 |
|
|
else { |
1512 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1513 |
|
|
emit_byte(0x0d); |
1514 |
|
|
else { |
1515 |
gbeauche |
1.1 |
emit_byte(0x81); |
1516 |
|
|
emit_byte(0xc8+d); |
1517 |
gbeauche |
1.2 |
} |
1518 |
gbeauche |
1.1 |
emit_long(i); |
1519 |
|
|
} |
1520 |
|
|
} |
1521 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_l_ri,(RW4 d, IMM i)) |
1522 |
|
|
|
1523 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_l,(RW4 d, R4 s)) |
1524 |
|
|
{ |
1525 |
|
|
emit_byte(0x09); |
1526 |
|
|
emit_byte(0xc0+8*s+d); |
1527 |
|
|
} |
1528 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_l,(RW4 d, R4 s)) |
1529 |
|
|
|
1530 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_w,(RW2 d, R2 s)) |
1531 |
|
|
{ |
1532 |
|
|
emit_byte(0x66); |
1533 |
|
|
emit_byte(0x09); |
1534 |
|
|
emit_byte(0xc0+8*s+d); |
1535 |
|
|
} |
1536 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_w,(RW2 d, R2 s)) |
1537 |
|
|
|
1538 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_b,(RW1 d, R1 s)) |
1539 |
|
|
{ |
1540 |
|
|
emit_byte(0x08); |
1541 |
|
|
emit_byte(0xc0+8*s+d); |
1542 |
|
|
} |
1543 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_b,(RW1 d, R1 s)) |
1544 |
|
|
|
1545 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_l,(RW4 d, R4 s)) |
1546 |
|
|
{ |
1547 |
|
|
emit_byte(0x11); |
1548 |
|
|
emit_byte(0xc0+8*s+d); |
1549 |
|
|
} |
1550 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_l,(RW4 d, R4 s)) |
1551 |
|
|
|
1552 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_w,(RW2 d, R2 s)) |
1553 |
|
|
{ |
1554 |
|
|
emit_byte(0x66); |
1555 |
|
|
emit_byte(0x11); |
1556 |
|
|
emit_byte(0xc0+8*s+d); |
1557 |
|
|
} |
1558 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_w,(RW2 d, R2 s)) |
1559 |
|
|
|
1560 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_b,(RW1 d, R1 s)) |
1561 |
|
|
{ |
1562 |
|
|
emit_byte(0x10); |
1563 |
|
|
emit_byte(0xc0+8*s+d); |
1564 |
|
|
} |
1565 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_b,(RW1 d, R1 s)) |
1566 |
|
|
|
1567 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_l,(RW4 d, R4 s)) |
1568 |
|
|
{ |
1569 |
|
|
emit_byte(0x01); |
1570 |
|
|
emit_byte(0xc0+8*s+d); |
1571 |
|
|
} |
1572 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_l,(RW4 d, R4 s)) |
1573 |
|
|
|
1574 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_w,(RW2 d, R2 s)) |
1575 |
|
|
{ |
1576 |
|
|
emit_byte(0x66); |
1577 |
|
|
emit_byte(0x01); |
1578 |
|
|
emit_byte(0xc0+8*s+d); |
1579 |
|
|
} |
1580 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_w,(RW2 d, R2 s)) |
1581 |
|
|
|
1582 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_b,(RW1 d, R1 s)) |
1583 |
|
|
{ |
1584 |
|
|
emit_byte(0x00); |
1585 |
|
|
emit_byte(0xc0+8*s+d); |
1586 |
|
|
} |
1587 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_b,(RW1 d, R1 s)) |
1588 |
|
|
|
1589 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_l_ri,(RW4 d, IMM i)) |
1590 |
|
|
{ |
1591 |
|
|
if (isbyte(i)) { |
1592 |
|
|
emit_byte(0x83); |
1593 |
|
|
emit_byte(0xe8+d); |
1594 |
|
|
emit_byte(i); |
1595 |
|
|
} |
1596 |
|
|
else { |
1597 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1598 |
|
|
emit_byte(0x2d); |
1599 |
|
|
else { |
1600 |
gbeauche |
1.1 |
emit_byte(0x81); |
1601 |
|
|
emit_byte(0xe8+d); |
1602 |
gbeauche |
1.2 |
} |
1603 |
gbeauche |
1.1 |
emit_long(i); |
1604 |
|
|
} |
1605 |
|
|
} |
1606 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_l_ri,(RW4 d, IMM i)) |
1607 |
|
|
|
1608 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_b_ri,(RW1 d, IMM i)) |
1609 |
|
|
{ |
1610 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1611 |
|
|
emit_byte(0x2c); |
1612 |
|
|
else { |
1613 |
gbeauche |
1.1 |
emit_byte(0x80); |
1614 |
|
|
emit_byte(0xe8+d); |
1615 |
gbeauche |
1.2 |
} |
1616 |
gbeauche |
1.1 |
emit_byte(i); |
1617 |
|
|
} |
1618 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_b_ri,(RW1 d, IMM i)) |
1619 |
|
|
|
1620 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_l_ri,(RW4 d, IMM i)) |
1621 |
|
|
{ |
1622 |
|
|
if (isbyte(i)) { |
1623 |
|
|
emit_byte(0x83); |
1624 |
|
|
emit_byte(0xc0+d); |
1625 |
|
|
emit_byte(i); |
1626 |
|
|
} |
1627 |
|
|
else { |
1628 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1629 |
|
|
emit_byte(0x05); |
1630 |
|
|
else { |
1631 |
gbeauche |
1.1 |
emit_byte(0x81); |
1632 |
|
|
emit_byte(0xc0+d); |
1633 |
gbeauche |
1.2 |
} |
1634 |
gbeauche |
1.1 |
emit_long(i); |
1635 |
|
|
} |
1636 |
|
|
} |
1637 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_l_ri,(RW4 d, IMM i)) |
1638 |
|
|
|
1639 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_w_ri,(RW2 d, IMM i)) |
1640 |
|
|
{ |
1641 |
gbeauche |
1.2 |
emit_byte(0x66); |
1642 |
gbeauche |
1.1 |
if (isbyte(i)) { |
1643 |
|
|
emit_byte(0x83); |
1644 |
|
|
emit_byte(0xc0+d); |
1645 |
|
|
emit_byte(i); |
1646 |
|
|
} |
1647 |
|
|
else { |
1648 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1649 |
|
|
emit_byte(0x05); |
1650 |
|
|
else { |
1651 |
gbeauche |
1.1 |
emit_byte(0x81); |
1652 |
|
|
emit_byte(0xc0+d); |
1653 |
gbeauche |
1.2 |
} |
1654 |
gbeauche |
1.1 |
emit_word(i); |
1655 |
|
|
} |
1656 |
|
|
} |
1657 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_w_ri,(RW2 d, IMM i)) |
1658 |
|
|
|
1659 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_b_ri,(RW1 d, IMM i)) |
1660 |
|
|
{ |
1661 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1662 |
|
|
emit_byte(0x04); |
1663 |
|
|
else { |
1664 |
|
|
emit_byte(0x80); |
1665 |
|
|
emit_byte(0xc0+d); |
1666 |
|
|
} |
1667 |
gbeauche |
1.1 |
emit_byte(i); |
1668 |
|
|
} |
1669 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_b_ri,(RW1 d, IMM i)) |
1670 |
|
|
|
1671 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_l,(RW4 d, R4 s)) |
1672 |
|
|
{ |
1673 |
|
|
emit_byte(0x19); |
1674 |
|
|
emit_byte(0xc0+8*s+d); |
1675 |
|
|
} |
1676 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_l,(RW4 d, R4 s)) |
1677 |
|
|
|
1678 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_w,(RW2 d, R2 s)) |
1679 |
|
|
{ |
1680 |
|
|
emit_byte(0x66); |
1681 |
|
|
emit_byte(0x19); |
1682 |
|
|
emit_byte(0xc0+8*s+d); |
1683 |
|
|
} |
1684 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_w,(RW2 d, R2 s)) |
1685 |
|
|
|
1686 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_b,(RW1 d, R1 s)) |
1687 |
|
|
{ |
1688 |
|
|
emit_byte(0x18); |
1689 |
|
|
emit_byte(0xc0+8*s+d); |
1690 |
|
|
} |
1691 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_b,(RW1 d, R1 s)) |
1692 |
|
|
|
1693 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_l,(RW4 d, R4 s)) |
1694 |
|
|
{ |
1695 |
|
|
emit_byte(0x29); |
1696 |
|
|
emit_byte(0xc0+8*s+d); |
1697 |
|
|
} |
1698 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_l,(RW4 d, R4 s)) |
1699 |
|
|
|
1700 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_w,(RW2 d, R2 s)) |
1701 |
|
|
{ |
1702 |
|
|
emit_byte(0x66); |
1703 |
|
|
emit_byte(0x29); |
1704 |
|
|
emit_byte(0xc0+8*s+d); |
1705 |
|
|
} |
1706 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_w,(RW2 d, R2 s)) |
1707 |
|
|
|
1708 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_b,(RW1 d, R1 s)) |
1709 |
|
|
{ |
1710 |
|
|
emit_byte(0x28); |
1711 |
|
|
emit_byte(0xc0+8*s+d); |
1712 |
|
|
} |
1713 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_b,(RW1 d, R1 s)) |
1714 |
|
|
|
1715 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_l,(R4 d, R4 s)) |
1716 |
|
|
{ |
1717 |
|
|
emit_byte(0x39); |
1718 |
|
|
emit_byte(0xc0+8*s+d); |
1719 |
|
|
} |
1720 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_l,(R4 d, R4 s)) |
1721 |
|
|
|
1722 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_l_ri,(R4 r, IMM i)) |
1723 |
|
|
{ |
1724 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1725 |
|
|
emit_byte(0x83); |
1726 |
|
|
emit_byte(0xf8+r); |
1727 |
|
|
emit_byte(i); |
1728 |
|
|
} |
1729 |
|
|
else { |
1730 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(r)) |
1731 |
|
|
emit_byte(0x3d); |
1732 |
|
|
else { |
1733 |
gbeauche |
1.1 |
emit_byte(0x81); |
1734 |
|
|
emit_byte(0xf8+r); |
1735 |
gbeauche |
1.2 |
} |
1736 |
gbeauche |
1.1 |
emit_long(i); |
1737 |
|
|
} |
1738 |
|
|
} |
1739 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_l_ri,(R4 r, IMM i)) |
1740 |
|
|
|
1741 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_w,(R2 d, R2 s)) |
1742 |
|
|
{ |
1743 |
|
|
emit_byte(0x66); |
1744 |
|
|
emit_byte(0x39); |
1745 |
|
|
emit_byte(0xc0+8*s+d); |
1746 |
|
|
} |
1747 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_w,(R2 d, R2 s)) |
1748 |
|
|
|
1749 |
gbeauche |
1.5 |
LOWFUNC(WRITE,READ,2,raw_cmp_b_mi,(MEMR d, IMM s)) |
1750 |
|
|
{ |
1751 |
|
|
emit_byte(0x80); |
1752 |
|
|
emit_byte(0x3d); |
1753 |
|
|
emit_long(d); |
1754 |
|
|
emit_byte(s); |
1755 |
|
|
} |
1756 |
|
|
LENDFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1757 |
|
|
|
1758 |
gbeauche |
1.1 |
LOWFUNC(WRITE,NONE,2,raw_cmp_b_ri,(R1 d, IMM i)) |
1759 |
|
|
{ |
1760 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1761 |
|
|
emit_byte(0x3c); |
1762 |
|
|
else { |
1763 |
gbeauche |
1.1 |
emit_byte(0x80); |
1764 |
|
|
emit_byte(0xf8+d); |
1765 |
gbeauche |
1.2 |
} |
1766 |
gbeauche |
1.1 |
emit_byte(i); |
1767 |
|
|
} |
1768 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_b_ri,(R1 d, IMM i)) |
1769 |
|
|
|
1770 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_b,(R1 d, R1 s)) |
1771 |
|
|
{ |
1772 |
|
|
emit_byte(0x38); |
1773 |
|
|
emit_byte(0xc0+8*s+d); |
1774 |
|
|
} |
1775 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_b,(R1 d, R1 s)) |
1776 |
|
|
|
1777 |
|
|
LOWFUNC(WRITE,READ,4,raw_cmp_l_rm_indexed,(R4 d, IMM offset, R4 index, IMM factor)) |
1778 |
|
|
{ |
1779 |
|
|
int fi; |
1780 |
|
|
|
1781 |
|
|
switch(factor) { |
1782 |
|
|
case 1: fi=0; break; |
1783 |
|
|
case 2: fi=1; break; |
1784 |
|
|
case 4: fi=2; break; |
1785 |
|
|
case 8: fi=3; break; |
1786 |
|
|
default: abort(); |
1787 |
|
|
} |
1788 |
|
|
emit_byte(0x39); |
1789 |
|
|
emit_byte(0x04+8*d); |
1790 |
|
|
emit_byte(5+8*index+0x40*fi); |
1791 |
|
|
emit_long(offset); |
1792 |
|
|
} |
1793 |
|
|
LENDFUNC(WRITE,READ,4,raw_cmp_l_rm_indexed,(R4 d, IMM offset, R4 index, IMM factor)) |
1794 |
|
|
|
1795 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_l,(RW4 d, R4 s)) |
1796 |
|
|
{ |
1797 |
|
|
emit_byte(0x31); |
1798 |
|
|
emit_byte(0xc0+8*s+d); |
1799 |
|
|
} |
1800 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_l,(RW4 d, R4 s)) |
1801 |
|
|
|
1802 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_w,(RW2 d, R2 s)) |
1803 |
|
|
{ |
1804 |
|
|
emit_byte(0x66); |
1805 |
|
|
emit_byte(0x31); |
1806 |
|
|
emit_byte(0xc0+8*s+d); |
1807 |
|
|
} |
1808 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_w,(RW2 d, R2 s)) |
1809 |
|
|
|
1810 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_b,(RW1 d, R1 s)) |
1811 |
|
|
{ |
1812 |
|
|
emit_byte(0x30); |
1813 |
|
|
emit_byte(0xc0+8*s+d); |
1814 |
|
|
} |
1815 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_b,(RW1 d, R1 s)) |
1816 |
|
|
|
1817 |
|
|
LOWFUNC(WRITE,RMW,2,raw_sub_l_mi,(MEMRW d, IMM s)) |
1818 |
|
|
{ |
1819 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1820 |
|
|
emit_byte(0x83); |
1821 |
|
|
emit_byte(0x2d); |
1822 |
|
|
emit_long(d); |
1823 |
|
|
emit_byte(s); |
1824 |
|
|
} |
1825 |
|
|
else { |
1826 |
|
|
emit_byte(0x81); |
1827 |
|
|
emit_byte(0x2d); |
1828 |
|
|
emit_long(d); |
1829 |
|
|
emit_long(s); |
1830 |
|
|
} |
1831 |
|
|
} |
1832 |
|
|
LENDFUNC(WRITE,RMW,2,raw_sub_l_mi,(MEMRW d, IMM s)) |
1833 |
|
|
|
1834 |
|
|
LOWFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1835 |
|
|
{ |
1836 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1837 |
|
|
emit_byte(0x83); |
1838 |
|
|
emit_byte(0x3d); |
1839 |
|
|
emit_long(d); |
1840 |
|
|
emit_byte(s); |
1841 |
|
|
} |
1842 |
|
|
else { |
1843 |
|
|
emit_byte(0x81); |
1844 |
|
|
emit_byte(0x3d); |
1845 |
|
|
emit_long(d); |
1846 |
|
|
emit_long(s); |
1847 |
|
|
} |
1848 |
|
|
} |
1849 |
|
|
LENDFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1850 |
|
|
|
1851 |
|
|
LOWFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
1852 |
|
|
{ |
1853 |
|
|
emit_byte(0x87); |
1854 |
|
|
emit_byte(0xc0+8*r1+r2); |
1855 |
|
|
} |
1856 |
|
|
LENDFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
1857 |
|
|
|
1858 |
|
|
/************************************************************************* |
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 |
|
|
/************************************************************************* |
1907 |
|
|
* FIXME: mem access modes probably wrong * |
1908 |
|
|
*************************************************************************/ |
1909 |
|
|
|
1910 |
|
|
LOWFUNC(READ,WRITE,0,raw_pushfl,(void)) |
1911 |
|
|
{ |
1912 |
|
|
emit_byte(0x9c); |
1913 |
|
|
} |
1914 |
|
|
LENDFUNC(READ,WRITE,0,raw_pushfl,(void)) |
1915 |
|
|
|
1916 |
|
|
LOWFUNC(WRITE,READ,0,raw_popfl,(void)) |
1917 |
|
|
{ |
1918 |
|
|
emit_byte(0x9d); |
1919 |
|
|
} |
1920 |
|
|
LENDFUNC(WRITE,READ,0,raw_popfl,(void)) |
1921 |
|
|
|
1922 |
|
|
/************************************************************************* |
1923 |
|
|
* Unoptimizable stuff --- jump * |
1924 |
|
|
*************************************************************************/ |
1925 |
|
|
|
1926 |
|
|
static __inline__ void raw_call_r(R4 r) |
1927 |
|
|
{ |
1928 |
|
|
emit_byte(0xff); |
1929 |
|
|
emit_byte(0xd0+r); |
1930 |
gbeauche |
1.5 |
} |
1931 |
|
|
|
1932 |
|
|
static __inline__ void raw_call_m_indexed(uae_u32 base, uae_u32 r, uae_u32 m) |
1933 |
|
|
{ |
1934 |
|
|
int mu; |
1935 |
|
|
switch(m) { |
1936 |
|
|
case 1: mu=0; break; |
1937 |
|
|
case 2: mu=1; break; |
1938 |
|
|
case 4: mu=2; break; |
1939 |
|
|
case 8: mu=3; break; |
1940 |
|
|
default: abort(); |
1941 |
|
|
} |
1942 |
|
|
emit_byte(0xff); |
1943 |
|
|
emit_byte(0x14); |
1944 |
|
|
emit_byte(0x05+8*r+0x40*mu); |
1945 |
|
|
emit_long(base); |
1946 |
gbeauche |
1.1 |
} |
1947 |
|
|
|
1948 |
|
|
static __inline__ void raw_jmp_r(R4 r) |
1949 |
|
|
{ |
1950 |
|
|
emit_byte(0xff); |
1951 |
|
|
emit_byte(0xe0+r); |
1952 |
|
|
} |
1953 |
|
|
|
1954 |
|
|
static __inline__ void raw_jmp_m_indexed(uae_u32 base, uae_u32 r, uae_u32 m) |
1955 |
|
|
{ |
1956 |
|
|
int mu; |
1957 |
|
|
switch(m) { |
1958 |
|
|
case 1: mu=0; break; |
1959 |
|
|
case 2: mu=1; break; |
1960 |
|
|
case 4: mu=2; break; |
1961 |
|
|
case 8: mu=3; break; |
1962 |
|
|
default: abort(); |
1963 |
|
|
} |
1964 |
|
|
emit_byte(0xff); |
1965 |
|
|
emit_byte(0x24); |
1966 |
|
|
emit_byte(0x05+8*r+0x40*mu); |
1967 |
|
|
emit_long(base); |
1968 |
|
|
} |
1969 |
|
|
|
1970 |
|
|
static __inline__ void raw_jmp_m(uae_u32 base) |
1971 |
|
|
{ |
1972 |
|
|
emit_byte(0xff); |
1973 |
|
|
emit_byte(0x25); |
1974 |
|
|
emit_long(base); |
1975 |
|
|
} |
1976 |
|
|
|
1977 |
|
|
|
1978 |
|
|
static __inline__ void raw_call(uae_u32 t) |
1979 |
|
|
{ |
1980 |
|
|
emit_byte(0xe8); |
1981 |
|
|
emit_long(t-(uae_u32)target-4); |
1982 |
|
|
} |
1983 |
|
|
|
1984 |
|
|
static __inline__ void raw_jmp(uae_u32 t) |
1985 |
|
|
{ |
1986 |
|
|
emit_byte(0xe9); |
1987 |
|
|
emit_long(t-(uae_u32)target-4); |
1988 |
|
|
} |
1989 |
|
|
|
1990 |
|
|
static __inline__ void raw_jl(uae_u32 t) |
1991 |
|
|
{ |
1992 |
|
|
emit_byte(0x0f); |
1993 |
|
|
emit_byte(0x8c); |
1994 |
|
|
emit_long(t-(uae_u32)target-4); |
1995 |
|
|
} |
1996 |
|
|
|
1997 |
|
|
static __inline__ void raw_jz(uae_u32 t) |
1998 |
|
|
{ |
1999 |
|
|
emit_byte(0x0f); |
2000 |
|
|
emit_byte(0x84); |
2001 |
|
|
emit_long(t-(uae_u32)target-4); |
2002 |
|
|
} |
2003 |
|
|
|
2004 |
|
|
static __inline__ void raw_jnz(uae_u32 t) |
2005 |
|
|
{ |
2006 |
|
|
emit_byte(0x0f); |
2007 |
|
|
emit_byte(0x85); |
2008 |
|
|
emit_long(t-(uae_u32)target-4); |
2009 |
|
|
} |
2010 |
|
|
|
2011 |
|
|
static __inline__ void raw_jnz_l_oponly(void) |
2012 |
|
|
{ |
2013 |
|
|
emit_byte(0x0f); |
2014 |
|
|
emit_byte(0x85); |
2015 |
|
|
} |
2016 |
|
|
|
2017 |
|
|
static __inline__ void raw_jcc_l_oponly(int cc) |
2018 |
|
|
{ |
2019 |
|
|
emit_byte(0x0f); |
2020 |
|
|
emit_byte(0x80+cc); |
2021 |
|
|
} |
2022 |
|
|
|
2023 |
|
|
static __inline__ void raw_jnz_b_oponly(void) |
2024 |
|
|
{ |
2025 |
|
|
emit_byte(0x75); |
2026 |
|
|
} |
2027 |
|
|
|
2028 |
|
|
static __inline__ void raw_jz_b_oponly(void) |
2029 |
|
|
{ |
2030 |
|
|
emit_byte(0x74); |
2031 |
|
|
} |
2032 |
|
|
|
2033 |
|
|
static __inline__ void raw_jcc_b_oponly(int cc) |
2034 |
|
|
{ |
2035 |
|
|
emit_byte(0x70+cc); |
2036 |
|
|
} |
2037 |
|
|
|
2038 |
|
|
static __inline__ void raw_jmp_l_oponly(void) |
2039 |
|
|
{ |
2040 |
|
|
emit_byte(0xe9); |
2041 |
|
|
} |
2042 |
|
|
|
2043 |
|
|
static __inline__ void raw_jmp_b_oponly(void) |
2044 |
|
|
{ |
2045 |
|
|
emit_byte(0xeb); |
2046 |
|
|
} |
2047 |
|
|
|
2048 |
|
|
static __inline__ void raw_ret(void) |
2049 |
|
|
{ |
2050 |
|
|
emit_byte(0xc3); |
2051 |
|
|
} |
2052 |
|
|
|
2053 |
|
|
static __inline__ void raw_nop(void) |
2054 |
|
|
{ |
2055 |
|
|
emit_byte(0x90); |
2056 |
|
|
} |
2057 |
|
|
|
2058 |
|
|
|
2059 |
|
|
/************************************************************************* |
2060 |
|
|
* Flag handling, to and fro UAE flag register * |
2061 |
|
|
*************************************************************************/ |
2062 |
|
|
|
2063 |
|
|
#ifdef SAHF_SETO_PROFITABLE |
2064 |
|
|
|
2065 |
|
|
#define FLAG_NREG1 0 /* Set to -1 if any register will do */ |
2066 |
|
|
|
2067 |
|
|
static __inline__ void raw_flags_to_reg(int r) |
2068 |
|
|
{ |
2069 |
|
|
raw_lahf(0); /* Most flags in AH */ |
2070 |
|
|
//raw_setcc(r,0); /* V flag in AL */ |
2071 |
|
|
raw_setcc_m((uae_u32)live.state[FLAGTMP].mem,0); |
2072 |
|
|
|
2073 |
|
|
#if 1 /* Let's avoid those nasty partial register stalls */ |
2074 |
|
|
//raw_mov_b_mr((uae_u32)live.state[FLAGTMP].mem,r); |
2075 |
|
|
raw_mov_b_mr(((uae_u32)live.state[FLAGTMP].mem)+1,r+4); |
2076 |
|
|
//live.state[FLAGTMP].status=CLEAN; |
2077 |
|
|
live.state[FLAGTMP].status=INMEM; |
2078 |
|
|
live.state[FLAGTMP].realreg=-1; |
2079 |
|
|
/* We just "evicted" FLAGTMP. */ |
2080 |
|
|
if (live.nat[r].nholds!=1) { |
2081 |
|
|
/* Huh? */ |
2082 |
|
|
abort(); |
2083 |
|
|
} |
2084 |
|
|
live.nat[r].nholds=0; |
2085 |
|
|
#endif |
2086 |
|
|
} |
2087 |
|
|
|
2088 |
|
|
#define FLAG_NREG2 0 /* Set to -1 if any register will do */ |
2089 |
|
|
static __inline__ void raw_reg_to_flags(int r) |
2090 |
|
|
{ |
2091 |
|
|
raw_cmp_b_ri(r,-127); /* set V */ |
2092 |
|
|
raw_sahf(0); |
2093 |
|
|
} |
2094 |
|
|
|
2095 |
|
|
#else |
2096 |
|
|
|
2097 |
|
|
#define FLAG_NREG1 -1 /* Set to -1 if any register will do */ |
2098 |
|
|
static __inline__ void raw_flags_to_reg(int r) |
2099 |
|
|
{ |
2100 |
|
|
raw_pushfl(); |
2101 |
|
|
raw_pop_l_r(r); |
2102 |
|
|
raw_mov_l_mr((uae_u32)live.state[FLAGTMP].mem,r); |
2103 |
|
|
// live.state[FLAGTMP].status=CLEAN; |
2104 |
|
|
live.state[FLAGTMP].status=INMEM; |
2105 |
|
|
live.state[FLAGTMP].realreg=-1; |
2106 |
|
|
/* We just "evicted" FLAGTMP. */ |
2107 |
|
|
if (live.nat[r].nholds!=1) { |
2108 |
|
|
/* Huh? */ |
2109 |
|
|
abort(); |
2110 |
|
|
} |
2111 |
|
|
live.nat[r].nholds=0; |
2112 |
|
|
} |
2113 |
|
|
|
2114 |
|
|
#define FLAG_NREG2 -1 /* Set to -1 if any register will do */ |
2115 |
|
|
static __inline__ void raw_reg_to_flags(int r) |
2116 |
|
|
{ |
2117 |
|
|
raw_push_l_r(r); |
2118 |
|
|
raw_popfl(); |
2119 |
|
|
} |
2120 |
|
|
|
2121 |
|
|
#endif |
2122 |
|
|
|
2123 |
|
|
/* Apparently, there are enough instructions between flag store and |
2124 |
|
|
flag reload to avoid the partial memory stall */ |
2125 |
|
|
static __inline__ void raw_load_flagreg(uae_u32 target, uae_u32 r) |
2126 |
|
|
{ |
2127 |
|
|
#if 1 |
2128 |
|
|
raw_mov_l_rm(target,(uae_u32)live.state[r].mem); |
2129 |
|
|
#else |
2130 |
|
|
raw_mov_b_rm(target,(uae_u32)live.state[r].mem); |
2131 |
|
|
raw_mov_b_rm(target+4,((uae_u32)live.state[r].mem)+1); |
2132 |
|
|
#endif |
2133 |
|
|
} |
2134 |
|
|
|
2135 |
|
|
/* FLAGX is byte sized, and we *do* write it at that size */ |
2136 |
|
|
static __inline__ void raw_load_flagx(uae_u32 target, uae_u32 r) |
2137 |
|
|
{ |
2138 |
|
|
if (live.nat[target].canbyte) |
2139 |
|
|
raw_mov_b_rm(target,(uae_u32)live.state[r].mem); |
2140 |
|
|
else if (live.nat[target].canword) |
2141 |
|
|
raw_mov_w_rm(target,(uae_u32)live.state[r].mem); |
2142 |
|
|
else |
2143 |
|
|
raw_mov_l_rm(target,(uae_u32)live.state[r].mem); |
2144 |
|
|
} |
2145 |
|
|
|
2146 |
|
|
|
2147 |
|
|
static __inline__ void raw_inc_sp(int off) |
2148 |
|
|
{ |
2149 |
gbeauche |
1.2 |
raw_add_l_ri(ESP_INDEX,off); |
2150 |
gbeauche |
1.1 |
} |
2151 |
|
|
|
2152 |
|
|
/************************************************************************* |
2153 |
|
|
* Handling mistaken direct memory access * |
2154 |
|
|
*************************************************************************/ |
2155 |
|
|
|
2156 |
|
|
// gb-- I don't need that part for JIT Basilisk II |
2157 |
|
|
#if defined(NATMEM_OFFSET) && 0 |
2158 |
|
|
#include <asm/sigcontext.h> |
2159 |
|
|
#include <signal.h> |
2160 |
|
|
|
2161 |
|
|
#define SIG_READ 1 |
2162 |
|
|
#define SIG_WRITE 2 |
2163 |
|
|
|
2164 |
|
|
static int in_handler=0; |
2165 |
|
|
static uae_u8 veccode[256]; |
2166 |
|
|
|
2167 |
|
|
static void vec(int x, struct sigcontext sc) |
2168 |
|
|
{ |
2169 |
|
|
uae_u8* i=(uae_u8*)sc.eip; |
2170 |
|
|
uae_u32 addr=sc.cr2; |
2171 |
|
|
int r=-1; |
2172 |
|
|
int size=4; |
2173 |
|
|
int dir=-1; |
2174 |
|
|
int len=0; |
2175 |
|
|
int j; |
2176 |
|
|
|
2177 |
|
|
write_log("fault address is %08x at %08x\n",sc.cr2,sc.eip); |
2178 |
|
|
if (!canbang) |
2179 |
|
|
write_log("Not happy! Canbang is 0 in SIGSEGV handler!\n"); |
2180 |
|
|
if (in_handler) |
2181 |
|
|
write_log("Argh --- Am already in a handler. Shouldn't happen!\n"); |
2182 |
|
|
|
2183 |
|
|
if (canbang && i>=compiled_code && i<=current_compile_p) { |
2184 |
|
|
if (*i==0x66) { |
2185 |
|
|
i++; |
2186 |
|
|
size=2; |
2187 |
|
|
len++; |
2188 |
|
|
} |
2189 |
|
|
|
2190 |
|
|
switch(i[0]) { |
2191 |
|
|
case 0x8a: |
2192 |
|
|
if ((i[1]&0xc0)==0x80) { |
2193 |
|
|
r=(i[1]>>3)&7; |
2194 |
|
|
dir=SIG_READ; |
2195 |
|
|
size=1; |
2196 |
|
|
len+=6; |
2197 |
|
|
break; |
2198 |
|
|
} |
2199 |
|
|
break; |
2200 |
|
|
case 0x88: |
2201 |
|
|
if ((i[1]&0xc0)==0x80) { |
2202 |
|
|
r=(i[1]>>3)&7; |
2203 |
|
|
dir=SIG_WRITE; |
2204 |
|
|
size=1; |
2205 |
|
|
len+=6; |
2206 |
|
|
break; |
2207 |
|
|
} |
2208 |
|
|
break; |
2209 |
|
|
case 0x8b: |
2210 |
|
|
if ((i[1]&0xc0)==0x80) { |
2211 |
|
|
r=(i[1]>>3)&7; |
2212 |
|
|
dir=SIG_READ; |
2213 |
|
|
len+=6; |
2214 |
|
|
break; |
2215 |
|
|
} |
2216 |
|
|
if ((i[1]&0xc0)==0x40) { |
2217 |
|
|
r=(i[1]>>3)&7; |
2218 |
|
|
dir=SIG_READ; |
2219 |
|
|
len+=3; |
2220 |
|
|
break; |
2221 |
|
|
} |
2222 |
|
|
break; |
2223 |
|
|
case 0x89: |
2224 |
|
|
if ((i[1]&0xc0)==0x80) { |
2225 |
|
|
r=(i[1]>>3)&7; |
2226 |
|
|
dir=SIG_WRITE; |
2227 |
|
|
len+=6; |
2228 |
|
|
break; |
2229 |
|
|
} |
2230 |
|
|
if ((i[1]&0xc0)==0x40) { |
2231 |
|
|
r=(i[1]>>3)&7; |
2232 |
|
|
dir=SIG_WRITE; |
2233 |
|
|
len+=3; |
2234 |
|
|
break; |
2235 |
|
|
} |
2236 |
|
|
break; |
2237 |
|
|
} |
2238 |
|
|
} |
2239 |
|
|
|
2240 |
|
|
if (r!=-1) { |
2241 |
|
|
void* pr=NULL; |
2242 |
|
|
write_log("register was %d, direction was %d, size was %d\n",r,dir,size); |
2243 |
|
|
|
2244 |
|
|
switch(r) { |
2245 |
|
|
case 0: pr=&(sc.eax); break; |
2246 |
|
|
case 1: pr=&(sc.ecx); break; |
2247 |
|
|
case 2: pr=&(sc.edx); break; |
2248 |
|
|
case 3: pr=&(sc.ebx); break; |
2249 |
|
|
case 4: pr=(size>1)?NULL:(((uae_u8*)&(sc.eax))+1); break; |
2250 |
|
|
case 5: pr=(size>1)? |
2251 |
|
|
(void*)(&(sc.ebp)): |
2252 |
|
|
(void*)(((uae_u8*)&(sc.ecx))+1); break; |
2253 |
|
|
case 6: pr=(size>1)? |
2254 |
|
|
(void*)(&(sc.esi)): |
2255 |
|
|
(void*)(((uae_u8*)&(sc.edx))+1); break; |
2256 |
|
|
case 7: pr=(size>1)? |
2257 |
|
|
(void*)(&(sc.edi)): |
2258 |
|
|
(void*)(((uae_u8*)&(sc.ebx))+1); break; |
2259 |
|
|
default: abort(); |
2260 |
|
|
} |
2261 |
|
|
if (pr) { |
2262 |
|
|
blockinfo* bi; |
2263 |
|
|
|
2264 |
|
|
if (currprefs.comp_oldsegv) { |
2265 |
|
|
addr-=NATMEM_OFFSET; |
2266 |
|
|
|
2267 |
|
|
if ((addr>=0x10000000 && addr<0x40000000) || |
2268 |
|
|
(addr>=0x50000000)) { |
2269 |
|
|
write_log("Suspicious address in %x SEGV handler.\n",addr); |
2270 |
|
|
} |
2271 |
|
|
if (dir==SIG_READ) { |
2272 |
|
|
switch(size) { |
2273 |
|
|
case 1: *((uae_u8*)pr)=get_byte(addr); break; |
2274 |
|
|
case 2: *((uae_u16*)pr)=get_word(addr); break; |
2275 |
|
|
case 4: *((uae_u32*)pr)=get_long(addr); break; |
2276 |
|
|
default: abort(); |
2277 |
|
|
} |
2278 |
|
|
} |
2279 |
|
|
else { /* write */ |
2280 |
|
|
switch(size) { |
2281 |
|
|
case 1: put_byte(addr,*((uae_u8*)pr)); break; |
2282 |
|
|
case 2: put_word(addr,*((uae_u16*)pr)); break; |
2283 |
|
|
case 4: put_long(addr,*((uae_u32*)pr)); break; |
2284 |
|
|
default: abort(); |
2285 |
|
|
} |
2286 |
|
|
} |
2287 |
|
|
write_log("Handled one access!\n"); |
2288 |
|
|
fflush(stdout); |
2289 |
|
|
segvcount++; |
2290 |
|
|
sc.eip+=len; |
2291 |
|
|
} |
2292 |
|
|
else { |
2293 |
|
|
void* tmp=target; |
2294 |
|
|
int i; |
2295 |
|
|
uae_u8 vecbuf[5]; |
2296 |
|
|
|
2297 |
|
|
addr-=NATMEM_OFFSET; |
2298 |
|
|
|
2299 |
|
|
if ((addr>=0x10000000 && addr<0x40000000) || |
2300 |
|
|
(addr>=0x50000000)) { |
2301 |
|
|
write_log("Suspicious address in %x SEGV handler.\n",addr); |
2302 |
|
|
} |
2303 |
|
|
|
2304 |
|
|
target=(uae_u8*)sc.eip; |
2305 |
|
|
for (i=0;i<5;i++) |
2306 |
|
|
vecbuf[i]=target[i]; |
2307 |
|
|
emit_byte(0xe9); |
2308 |
|
|
emit_long((uae_u32)veccode-(uae_u32)target-4); |
2309 |
|
|
write_log("Create jump to %p\n",veccode); |
2310 |
|
|
|
2311 |
|
|
write_log("Handled one access!\n"); |
2312 |
|
|
fflush(stdout); |
2313 |
|
|
segvcount++; |
2314 |
|
|
|
2315 |
|
|
target=veccode; |
2316 |
|
|
|
2317 |
|
|
if (dir==SIG_READ) { |
2318 |
|
|
switch(size) { |
2319 |
|
|
case 1: raw_mov_b_ri(r,get_byte(addr)); break; |
2320 |
|
|
case 2: raw_mov_w_ri(r,get_byte(addr)); break; |
2321 |
|
|
case 4: raw_mov_l_ri(r,get_byte(addr)); break; |
2322 |
|
|
default: abort(); |
2323 |
|
|
} |
2324 |
|
|
} |
2325 |
|
|
else { /* write */ |
2326 |
|
|
switch(size) { |
2327 |
|
|
case 1: put_byte(addr,*((uae_u8*)pr)); break; |
2328 |
|
|
case 2: put_word(addr,*((uae_u16*)pr)); break; |
2329 |
|
|
case 4: put_long(addr,*((uae_u32*)pr)); break; |
2330 |
|
|
default: abort(); |
2331 |
|
|
} |
2332 |
|
|
} |
2333 |
|
|
for (i=0;i<5;i++) |
2334 |
|
|
raw_mov_b_mi(sc.eip+i,vecbuf[i]); |
2335 |
|
|
raw_mov_l_mi((uae_u32)&in_handler,0); |
2336 |
|
|
emit_byte(0xe9); |
2337 |
|
|
emit_long(sc.eip+len-(uae_u32)target-4); |
2338 |
|
|
in_handler=1; |
2339 |
|
|
target=tmp; |
2340 |
|
|
} |
2341 |
|
|
bi=active; |
2342 |
|
|
while (bi) { |
2343 |
|
|
if (bi->handler && |
2344 |
|
|
(uae_u8*)bi->direct_handler<=i && |
2345 |
|
|
(uae_u8*)bi->nexthandler>i) { |
2346 |
|
|
write_log("deleted trigger (%p<%p<%p) %p\n", |
2347 |
|
|
bi->handler, |
2348 |
|
|
i, |
2349 |
|
|
bi->nexthandler, |
2350 |
|
|
bi->pc_p); |
2351 |
|
|
invalidate_block(bi); |
2352 |
|
|
raise_in_cl_list(bi); |
2353 |
|
|
set_special(0); |
2354 |
|
|
return; |
2355 |
|
|
} |
2356 |
|
|
bi=bi->next; |
2357 |
|
|
} |
2358 |
|
|
/* Not found in the active list. Might be a rom routine that |
2359 |
|
|
is in the dormant list */ |
2360 |
|
|
bi=dormant; |
2361 |
|
|
while (bi) { |
2362 |
|
|
if (bi->handler && |
2363 |
|
|
(uae_u8*)bi->direct_handler<=i && |
2364 |
|
|
(uae_u8*)bi->nexthandler>i) { |
2365 |
|
|
write_log("deleted trigger (%p<%p<%p) %p\n", |
2366 |
|
|
bi->handler, |
2367 |
|
|
i, |
2368 |
|
|
bi->nexthandler, |
2369 |
|
|
bi->pc_p); |
2370 |
|
|
invalidate_block(bi); |
2371 |
|
|
raise_in_cl_list(bi); |
2372 |
|
|
set_special(0); |
2373 |
|
|
return; |
2374 |
|
|
} |
2375 |
|
|
bi=bi->next; |
2376 |
|
|
} |
2377 |
|
|
write_log("Huh? Could not find trigger!\n"); |
2378 |
|
|
return; |
2379 |
|
|
} |
2380 |
|
|
} |
2381 |
|
|
write_log("Can't handle access!\n"); |
2382 |
|
|
for (j=0;j<10;j++) { |
2383 |
|
|
write_log("instruction byte %2d is %02x\n",j,i[j]); |
2384 |
|
|
} |
2385 |
|
|
write_log("Please send the above info (starting at \"fault address\") to\n" |
2386 |
|
|
"bmeyer@csse.monash.edu.au\n" |
2387 |
|
|
"This shouldn't happen ;-)\n"); |
2388 |
|
|
fflush(stdout); |
2389 |
|
|
signal(SIGSEGV,SIG_DFL); /* returning here will cause a "real" SEGV */ |
2390 |
|
|
} |
2391 |
|
|
#endif |
2392 |
|
|
|
2393 |
|
|
|
2394 |
|
|
/************************************************************************* |
2395 |
|
|
* Checking for CPU features * |
2396 |
|
|
*************************************************************************/ |
2397 |
|
|
|
2398 |
gbeauche |
1.3 |
struct cpuinfo_x86 { |
2399 |
|
|
uae_u8 x86; // CPU family |
2400 |
|
|
uae_u8 x86_vendor; // CPU vendor |
2401 |
|
|
uae_u8 x86_processor; // CPU canonical processor type |
2402 |
|
|
uae_u8 x86_brand_id; // CPU BrandID if supported, yield 0 otherwise |
2403 |
|
|
uae_u32 x86_hwcap; |
2404 |
|
|
uae_u8 x86_model; |
2405 |
|
|
uae_u8 x86_mask; |
2406 |
|
|
int cpuid_level; // Maximum supported CPUID level, -1=no CPUID |
2407 |
|
|
char x86_vendor_id[16]; |
2408 |
|
|
}; |
2409 |
|
|
struct cpuinfo_x86 cpuinfo; |
2410 |
|
|
|
2411 |
|
|
enum { |
2412 |
|
|
X86_VENDOR_INTEL = 0, |
2413 |
|
|
X86_VENDOR_CYRIX = 1, |
2414 |
|
|
X86_VENDOR_AMD = 2, |
2415 |
|
|
X86_VENDOR_UMC = 3, |
2416 |
|
|
X86_VENDOR_NEXGEN = 4, |
2417 |
|
|
X86_VENDOR_CENTAUR = 5, |
2418 |
|
|
X86_VENDOR_RISE = 6, |
2419 |
|
|
X86_VENDOR_TRANSMETA = 7, |
2420 |
|
|
X86_VENDOR_NSC = 8, |
2421 |
|
|
X86_VENDOR_UNKNOWN = 0xff |
2422 |
|
|
}; |
2423 |
|
|
|
2424 |
|
|
enum { |
2425 |
|
|
X86_PROCESSOR_I386, /* 80386 */ |
2426 |
|
|
X86_PROCESSOR_I486, /* 80486DX, 80486SX, 80486DX[24] */ |
2427 |
|
|
X86_PROCESSOR_PENTIUM, |
2428 |
|
|
X86_PROCESSOR_PENTIUMPRO, |
2429 |
|
|
X86_PROCESSOR_K6, |
2430 |
|
|
X86_PROCESSOR_ATHLON, |
2431 |
|
|
X86_PROCESSOR_PENTIUM4, |
2432 |
|
|
X86_PROCESSOR_max |
2433 |
|
|
}; |
2434 |
|
|
|
2435 |
|
|
static const char * x86_processor_string_table[X86_PROCESSOR_max] = { |
2436 |
|
|
"80386", |
2437 |
|
|
"80486", |
2438 |
|
|
"Pentium", |
2439 |
|
|
"PentiumPro", |
2440 |
|
|
"K6", |
2441 |
|
|
"Athlon", |
2442 |
|
|
"Pentium4" |
2443 |
|
|
}; |
2444 |
|
|
|
2445 |
|
|
static struct ptt { |
2446 |
|
|
const int align_loop; |
2447 |
|
|
const int align_loop_max_skip; |
2448 |
|
|
const int align_jump; |
2449 |
|
|
const int align_jump_max_skip; |
2450 |
|
|
const int align_func; |
2451 |
|
|
} |
2452 |
|
|
x86_alignments[X86_PROCESSOR_max] = { |
2453 |
|
|
{ 4, 3, 4, 3, 4 }, |
2454 |
|
|
{ 16, 15, 16, 15, 16 }, |
2455 |
|
|
{ 16, 7, 16, 7, 16 }, |
2456 |
|
|
{ 16, 15, 16, 7, 16 }, |
2457 |
|
|
{ 32, 7, 32, 7, 32 }, |
2458 |
gbeauche |
1.4 |
{ 16, 7, 16, 7, 16 }, |
2459 |
gbeauche |
1.3 |
{ 0, 0, 0, 0, 0 } |
2460 |
|
|
}; |
2461 |
gbeauche |
1.1 |
|
2462 |
gbeauche |
1.3 |
static void |
2463 |
|
|
x86_get_cpu_vendor(struct cpuinfo_x86 *c) |
2464 |
gbeauche |
1.1 |
{ |
2465 |
gbeauche |
1.3 |
char *v = c->x86_vendor_id; |
2466 |
|
|
|
2467 |
|
|
if (!strcmp(v, "GenuineIntel")) |
2468 |
|
|
c->x86_vendor = X86_VENDOR_INTEL; |
2469 |
|
|
else if (!strcmp(v, "AuthenticAMD")) |
2470 |
|
|
c->x86_vendor = X86_VENDOR_AMD; |
2471 |
|
|
else if (!strcmp(v, "CyrixInstead")) |
2472 |
|
|
c->x86_vendor = X86_VENDOR_CYRIX; |
2473 |
|
|
else if (!strcmp(v, "Geode by NSC")) |
2474 |
|
|
c->x86_vendor = X86_VENDOR_NSC; |
2475 |
|
|
else if (!strcmp(v, "UMC UMC UMC ")) |
2476 |
|
|
c->x86_vendor = X86_VENDOR_UMC; |
2477 |
|
|
else if (!strcmp(v, "CentaurHauls")) |
2478 |
|
|
c->x86_vendor = X86_VENDOR_CENTAUR; |
2479 |
|
|
else if (!strcmp(v, "NexGenDriven")) |
2480 |
|
|
c->x86_vendor = X86_VENDOR_NEXGEN; |
2481 |
|
|
else if (!strcmp(v, "RiseRiseRise")) |
2482 |
|
|
c->x86_vendor = X86_VENDOR_RISE; |
2483 |
|
|
else if (!strcmp(v, "GenuineTMx86") || |
2484 |
|
|
!strcmp(v, "TransmetaCPU")) |
2485 |
|
|
c->x86_vendor = X86_VENDOR_TRANSMETA; |
2486 |
|
|
else |
2487 |
|
|
c->x86_vendor = X86_VENDOR_UNKNOWN; |
2488 |
|
|
} |
2489 |
gbeauche |
1.1 |
|
2490 |
gbeauche |
1.3 |
static void |
2491 |
|
|
cpuid(uae_u32 op, uae_u32 *eax, uae_u32 *ebx, uae_u32 *ecx, uae_u32 *edx) |
2492 |
|
|
{ |
2493 |
|
|
static uae_u8 cpuid_space[256]; |
2494 |
|
|
uae_u8* tmp=get_target(); |
2495 |
gbeauche |
1.1 |
|
2496 |
gbeauche |
1.3 |
set_target(cpuid_space); |
2497 |
|
|
raw_push_l_r(0); /* eax */ |
2498 |
|
|
raw_push_l_r(1); /* ecx */ |
2499 |
|
|
raw_push_l_r(2); /* edx */ |
2500 |
|
|
raw_push_l_r(3); /* ebx */ |
2501 |
|
|
raw_mov_l_rm(0,(uae_u32)&op); |
2502 |
|
|
raw_cpuid(0); |
2503 |
|
|
if (eax != NULL) raw_mov_l_mr((uae_u32)eax,0); |
2504 |
|
|
if (ebx != NULL) raw_mov_l_mr((uae_u32)ebx,3); |
2505 |
|
|
if (ecx != NULL) raw_mov_l_mr((uae_u32)ecx,1); |
2506 |
|
|
if (edx != NULL) raw_mov_l_mr((uae_u32)edx,2); |
2507 |
|
|
raw_pop_l_r(3); |
2508 |
|
|
raw_pop_l_r(2); |
2509 |
|
|
raw_pop_l_r(1); |
2510 |
|
|
raw_pop_l_r(0); |
2511 |
|
|
raw_ret(); |
2512 |
|
|
set_target(tmp); |
2513 |
gbeauche |
1.1 |
|
2514 |
gbeauche |
1.3 |
((cpuop_func*)cpuid_space)(0); |
2515 |
gbeauche |
1.1 |
} |
2516 |
|
|
|
2517 |
gbeauche |
1.3 |
static void |
2518 |
|
|
raw_init_cpu(void) |
2519 |
gbeauche |
1.1 |
{ |
2520 |
gbeauche |
1.3 |
struct cpuinfo_x86 *c = &cpuinfo; |
2521 |
|
|
|
2522 |
|
|
/* Defaults */ |
2523 |
|
|
c->x86_vendor = X86_VENDOR_UNKNOWN; |
2524 |
|
|
c->cpuid_level = -1; /* CPUID not detected */ |
2525 |
|
|
c->x86_model = c->x86_mask = 0; /* So far unknown... */ |
2526 |
|
|
c->x86_vendor_id[0] = '\0'; /* Unset */ |
2527 |
|
|
c->x86_hwcap = 0; |
2528 |
|
|
|
2529 |
|
|
/* Get vendor name */ |
2530 |
|
|
c->x86_vendor_id[12] = '\0'; |
2531 |
|
|
cpuid(0x00000000, |
2532 |
|
|
(uae_u32 *)&c->cpuid_level, |
2533 |
|
|
(uae_u32 *)&c->x86_vendor_id[0], |
2534 |
|
|
(uae_u32 *)&c->x86_vendor_id[8], |
2535 |
|
|
(uae_u32 *)&c->x86_vendor_id[4]); |
2536 |
|
|
x86_get_cpu_vendor(c); |
2537 |
|
|
|
2538 |
|
|
/* Intel-defined flags: level 0x00000001 */ |
2539 |
|
|
c->x86_brand_id = 0; |
2540 |
|
|
if ( c->cpuid_level >= 0x00000001 ) { |
2541 |
|
|
uae_u32 tfms, brand_id; |
2542 |
|
|
cpuid(0x00000001, &tfms, &brand_id, NULL, &c->x86_hwcap); |
2543 |
|
|
c->x86 = (tfms >> 8) & 15; |
2544 |
|
|
c->x86_model = (tfms >> 4) & 15; |
2545 |
|
|
c->x86_brand_id = brand_id & 0xff; |
2546 |
|
|
if ( (c->x86_vendor == X86_VENDOR_AMD) && |
2547 |
|
|
(c->x86 == 0xf)) { |
2548 |
|
|
/* AMD Extended Family and Model Values */ |
2549 |
|
|
c->x86 += (tfms >> 20) & 0xff; |
2550 |
|
|
c->x86_model += (tfms >> 12) & 0xf0; |
2551 |
|
|
} |
2552 |
|
|
c->x86_mask = tfms & 15; |
2553 |
|
|
} else { |
2554 |
|
|
/* Have CPUID level 0 only - unheard of */ |
2555 |
|
|
c->x86 = 4; |
2556 |
|
|
} |
2557 |
|
|
|
2558 |
|
|
/* Canonicalize processor ID */ |
2559 |
|
|
c->x86_processor = X86_PROCESSOR_max; |
2560 |
|
|
switch (c->x86) { |
2561 |
|
|
case 3: |
2562 |
|
|
c->x86_processor = X86_PROCESSOR_I386; |
2563 |
|
|
break; |
2564 |
|
|
case 4: |
2565 |
|
|
c->x86_processor = X86_PROCESSOR_I486; |
2566 |
|
|
break; |
2567 |
|
|
case 5: |
2568 |
|
|
if (c->x86_vendor == X86_VENDOR_AMD) |
2569 |
|
|
c->x86_processor = X86_PROCESSOR_K6; |
2570 |
|
|
else |
2571 |
|
|
c->x86_processor = X86_PROCESSOR_PENTIUM; |
2572 |
|
|
break; |
2573 |
|
|
case 6: |
2574 |
|
|
if (c->x86_vendor == X86_VENDOR_AMD) |
2575 |
|
|
c->x86_processor = X86_PROCESSOR_ATHLON; |
2576 |
|
|
else |
2577 |
|
|
c->x86_processor = X86_PROCESSOR_PENTIUMPRO; |
2578 |
|
|
break; |
2579 |
|
|
case 15: |
2580 |
|
|
if (c->x86_vendor == X86_VENDOR_INTEL) { |
2581 |
|
|
/* Assume any BranID >= 8 and family == 15 yields a Pentium 4 */ |
2582 |
|
|
if (c->x86_brand_id >= 8) |
2583 |
|
|
c->x86_processor = X86_PROCESSOR_PENTIUM4; |
2584 |
|
|
} |
2585 |
|
|
break; |
2586 |
|
|
} |
2587 |
|
|
if (c->x86_processor == X86_PROCESSOR_max) { |
2588 |
|
|
fprintf(stderr, "Error: unknown processor type\n"); |
2589 |
|
|
fprintf(stderr, " Family : %d\n", c->x86); |
2590 |
|
|
fprintf(stderr, " Model : %d\n", c->x86_model); |
2591 |
|
|
fprintf(stderr, " Mask : %d\n", c->x86_mask); |
2592 |
|
|
if (c->x86_brand_id) |
2593 |
|
|
fprintf(stderr, " BrandID : %02x\n", c->x86_brand_id); |
2594 |
|
|
abort(); |
2595 |
|
|
} |
2596 |
|
|
|
2597 |
|
|
/* Have CMOV support? */ |
2598 |
|
|
have_cmov = (c->x86_hwcap & (1 << 15)) && true; |
2599 |
|
|
|
2600 |
|
|
/* Can the host CPU suffer from partial register stalls? */ |
2601 |
|
|
have_rat_stall = (c->x86_vendor == X86_VENDOR_INTEL); |
2602 |
|
|
#if 1 |
2603 |
|
|
/* It appears that partial register writes are a bad idea even on |
2604 |
gbeauche |
1.1 |
AMD K7 cores, even though they are not supposed to have the |
2605 |
|
|
dreaded rat stall. Why? Anyway, that's why we lie about it ;-) */ |
2606 |
gbeauche |
1.3 |
if (c->x86_processor == X86_PROCESSOR_ATHLON) |
2607 |
|
|
have_rat_stall = true; |
2608 |
gbeauche |
1.1 |
#endif |
2609 |
gbeauche |
1.3 |
|
2610 |
|
|
/* Alignments */ |
2611 |
|
|
if (tune_alignment) { |
2612 |
|
|
align_loops = x86_alignments[c->x86_processor].align_loop; |
2613 |
|
|
align_jumps = x86_alignments[c->x86_processor].align_jump; |
2614 |
|
|
} |
2615 |
|
|
|
2616 |
|
|
write_log("Max CPUID level=%d Processor is %s [%s]\n", |
2617 |
|
|
c->cpuid_level, c->x86_vendor_id, |
2618 |
|
|
x86_processor_string_table[c->x86_processor]); |
2619 |
gbeauche |
1.1 |
} |
2620 |
|
|
|
2621 |
|
|
|
2622 |
|
|
/************************************************************************* |
2623 |
|
|
* FPU stuff * |
2624 |
|
|
*************************************************************************/ |
2625 |
|
|
|
2626 |
|
|
|
2627 |
|
|
static __inline__ void raw_fp_init(void) |
2628 |
|
|
{ |
2629 |
|
|
int i; |
2630 |
|
|
|
2631 |
|
|
for (i=0;i<N_FREGS;i++) |
2632 |
|
|
live.spos[i]=-2; |
2633 |
|
|
live.tos=-1; /* Stack is empty */ |
2634 |
|
|
} |
2635 |
|
|
|
2636 |
|
|
static __inline__ void raw_fp_cleanup_drop(void) |
2637 |
|
|
{ |
2638 |
|
|
#if 0 |
2639 |
|
|
/* using FINIT instead of popping all the entries. |
2640 |
|
|
Seems to have side effects --- there is display corruption in |
2641 |
|
|
Quake when this is used */ |
2642 |
|
|
if (live.tos>1) { |
2643 |
|
|
emit_byte(0x9b); |
2644 |
|
|
emit_byte(0xdb); |
2645 |
|
|
emit_byte(0xe3); |
2646 |
|
|
live.tos=-1; |
2647 |
|
|
} |
2648 |
|
|
#endif |
2649 |
|
|
while (live.tos>=1) { |
2650 |
|
|
emit_byte(0xde); |
2651 |
|
|
emit_byte(0xd9); |
2652 |
|
|
live.tos-=2; |
2653 |
|
|
} |
2654 |
|
|
while (live.tos>=0) { |
2655 |
|
|
emit_byte(0xdd); |
2656 |
|
|
emit_byte(0xd8); |
2657 |
|
|
live.tos--; |
2658 |
|
|
} |
2659 |
|
|
raw_fp_init(); |
2660 |
|
|
} |
2661 |
|
|
|
2662 |
|
|
static __inline__ void make_tos(int r) |
2663 |
|
|
{ |
2664 |
|
|
int p,q; |
2665 |
|
|
|
2666 |
|
|
if (live.spos[r]<0) { /* Register not yet on stack */ |
2667 |
|
|
emit_byte(0xd9); |
2668 |
|
|
emit_byte(0xe8); /* Push '1' on the stack, just to grow it */ |
2669 |
|
|
live.tos++; |
2670 |
|
|
live.spos[r]=live.tos; |
2671 |
|
|
live.onstack[live.tos]=r; |
2672 |
|
|
return; |
2673 |
|
|
} |
2674 |
|
|
/* Register is on stack */ |
2675 |
|
|
if (live.tos==live.spos[r]) |
2676 |
|
|
return; |
2677 |
|
|
p=live.spos[r]; |
2678 |
|
|
q=live.onstack[live.tos]; |
2679 |
|
|
|
2680 |
|
|
emit_byte(0xd9); |
2681 |
|
|
emit_byte(0xc8+live.tos-live.spos[r]); /* exchange it with top of stack */ |
2682 |
|
|
live.onstack[live.tos]=r; |
2683 |
|
|
live.spos[r]=live.tos; |
2684 |
|
|
live.onstack[p]=q; |
2685 |
|
|
live.spos[q]=p; |
2686 |
|
|
} |
2687 |
|
|
|
2688 |
|
|
static __inline__ void make_tos2(int r, int r2) |
2689 |
|
|
{ |
2690 |
|
|
int q; |
2691 |
|
|
|
2692 |
|
|
make_tos(r2); /* Put the reg that's supposed to end up in position2 |
2693 |
|
|
on top */ |
2694 |
|
|
|
2695 |
|
|
if (live.spos[r]<0) { /* Register not yet on stack */ |
2696 |
|
|
make_tos(r); /* This will extend the stack */ |
2697 |
|
|
return; |
2698 |
|
|
} |
2699 |
|
|
/* Register is on stack */ |
2700 |
|
|
emit_byte(0xd9); |
2701 |
|
|
emit_byte(0xc9); /* Move r2 into position 2 */ |
2702 |
|
|
|
2703 |
|
|
q=live.onstack[live.tos-1]; |
2704 |
|
|
live.onstack[live.tos]=q; |
2705 |
|
|
live.spos[q]=live.tos; |
2706 |
|
|
live.onstack[live.tos-1]=r2; |
2707 |
|
|
live.spos[r2]=live.tos-1; |
2708 |
|
|
|
2709 |
|
|
make_tos(r); /* And r into 1 */ |
2710 |
|
|
} |
2711 |
|
|
|
2712 |
|
|
static __inline__ int stackpos(int r) |
2713 |
|
|
{ |
2714 |
|
|
if (live.spos[r]<0) |
2715 |
|
|
abort(); |
2716 |
|
|
if (live.tos<live.spos[r]) { |
2717 |
|
|
printf("Looking for spos for fnreg %d\n",r); |
2718 |
|
|
abort(); |
2719 |
|
|
} |
2720 |
|
|
return live.tos-live.spos[r]; |
2721 |
|
|
} |
2722 |
|
|
|
2723 |
|
|
static __inline__ void usereg(int r) |
2724 |
|
|
{ |
2725 |
|
|
if (live.spos[r]<0) |
2726 |
|
|
make_tos(r); |
2727 |
|
|
} |
2728 |
|
|
|
2729 |
|
|
/* This is called with one FP value in a reg *above* tos, which it will |
2730 |
|
|
pop off the stack if necessary */ |
2731 |
|
|
static __inline__ void tos_make(int r) |
2732 |
|
|
{ |
2733 |
|
|
if (live.spos[r]<0) { |
2734 |
|
|
live.tos++; |
2735 |
|
|
live.spos[r]=live.tos; |
2736 |
|
|
live.onstack[live.tos]=r; |
2737 |
|
|
return; |
2738 |
|
|
} |
2739 |
|
|
emit_byte(0xdd); |
2740 |
|
|
emit_byte(0xd8+(live.tos+1)-live.spos[r]); /* store top of stack in reg, |
2741 |
|
|
and pop it*/ |
2742 |
|
|
} |
2743 |
|
|
|
2744 |
|
|
|
2745 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2746 |
|
|
{ |
2747 |
|
|
make_tos(r); |
2748 |
|
|
emit_byte(0xdd); |
2749 |
|
|
emit_byte(0x15); |
2750 |
|
|
emit_long(m); |
2751 |
|
|
} |
2752 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2753 |
|
|
|
2754 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_mr_drop,(MEMW m, FR r)) |
2755 |
|
|
{ |
2756 |
|
|
make_tos(r); |
2757 |
|
|
emit_byte(0xdd); |
2758 |
|
|
emit_byte(0x1d); |
2759 |
|
|
emit_long(m); |
2760 |
|
|
live.onstack[live.tos]=-1; |
2761 |
|
|
live.tos--; |
2762 |
|
|
live.spos[r]=-2; |
2763 |
|
|
} |
2764 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2765 |
|
|
|
2766 |
|
|
LOWFUNC(NONE,READ,2,raw_fmov_rm,(FW r, MEMR m)) |
2767 |
|
|
{ |
2768 |
|
|
emit_byte(0xdd); |
2769 |
|
|
emit_byte(0x05); |
2770 |
|
|
emit_long(m); |
2771 |
|
|
tos_make(r); |
2772 |
|
|
} |
2773 |
|
|
LENDFUNC(NONE,READ,2,raw_fmov_rm,(FW r, MEMR m)) |
2774 |
|
|
|
2775 |
|
|
LOWFUNC(NONE,READ,2,raw_fmovi_rm,(FW r, MEMR m)) |
2776 |
|
|
{ |
2777 |
|
|
emit_byte(0xdb); |
2778 |
|
|
emit_byte(0x05); |
2779 |
|
|
emit_long(m); |
2780 |
|
|
tos_make(r); |
2781 |
|
|
} |
2782 |
|
|
LENDFUNC(NONE,READ,2,raw_fmovi_rm,(FW r, MEMR m)) |
2783 |
|
|
|
2784 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmovi_mr,(MEMW m, FR r)) |
2785 |
|
|
{ |
2786 |
|
|
make_tos(r); |
2787 |
|
|
emit_byte(0xdb); |
2788 |
|
|
emit_byte(0x15); |
2789 |
|
|
emit_long(m); |
2790 |
|
|
} |
2791 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmovi_mr,(MEMW m, FR r)) |
2792 |
|
|
|
2793 |
|
|
LOWFUNC(NONE,READ,2,raw_fmovs_rm,(FW r, MEMR m)) |
2794 |
|
|
{ |
2795 |
|
|
emit_byte(0xd9); |
2796 |
|
|
emit_byte(0x05); |
2797 |
|
|
emit_long(m); |
2798 |
|
|
tos_make(r); |
2799 |
|
|
} |
2800 |
|
|
LENDFUNC(NONE,READ,2,raw_fmovs_rm,(FW r, MEMR m)) |
2801 |
|
|
|
2802 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmovs_mr,(MEMW m, FR r)) |
2803 |
|
|
{ |
2804 |
|
|
make_tos(r); |
2805 |
|
|
emit_byte(0xd9); |
2806 |
|
|
emit_byte(0x15); |
2807 |
|
|
emit_long(m); |
2808 |
|
|
} |
2809 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmovs_mr,(MEMW m, FR r)) |
2810 |
|
|
|
2811 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2812 |
|
|
{ |
2813 |
|
|
int rs; |
2814 |
|
|
|
2815 |
|
|
/* Stupid x87 can't write a long double to mem without popping the |
2816 |
|
|
stack! */ |
2817 |
|
|
usereg(r); |
2818 |
|
|
rs=stackpos(r); |
2819 |
|
|
emit_byte(0xd9); /* Get a copy to the top of stack */ |
2820 |
|
|
emit_byte(0xc0+rs); |
2821 |
|
|
|
2822 |
|
|
emit_byte(0xdb); /* store and pop it */ |
2823 |
|
|
emit_byte(0x3d); |
2824 |
|
|
emit_long(m); |
2825 |
|
|
} |
2826 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2827 |
|
|
|
2828 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_ext_mr_drop,(MEMW m, FR r)) |
2829 |
|
|
{ |
2830 |
|
|
int rs; |
2831 |
|
|
|
2832 |
|
|
make_tos(r); |
2833 |
|
|
emit_byte(0xdb); /* store and pop it */ |
2834 |
|
|
emit_byte(0x3d); |
2835 |
|
|
emit_long(m); |
2836 |
|
|
live.onstack[live.tos]=-1; |
2837 |
|
|
live.tos--; |
2838 |
|
|
live.spos[r]=-2; |
2839 |
|
|
} |
2840 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2841 |
|
|
|
2842 |
|
|
LOWFUNC(NONE,READ,2,raw_fmov_ext_rm,(FW r, MEMR m)) |
2843 |
|
|
{ |
2844 |
|
|
emit_byte(0xdb); |
2845 |
|
|
emit_byte(0x2d); |
2846 |
|
|
emit_long(m); |
2847 |
|
|
tos_make(r); |
2848 |
|
|
} |
2849 |
|
|
LENDFUNC(NONE,READ,2,raw_fmov_ext_rm,(FW r, MEMR m)) |
2850 |
|
|
|
2851 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_pi,(FW r)) |
2852 |
|
|
{ |
2853 |
|
|
emit_byte(0xd9); |
2854 |
|
|
emit_byte(0xeb); |
2855 |
|
|
tos_make(r); |
2856 |
|
|
} |
2857 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_pi,(FW r)) |
2858 |
|
|
|
2859 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_log10_2,(FW r)) |
2860 |
|
|
{ |
2861 |
|
|
emit_byte(0xd9); |
2862 |
|
|
emit_byte(0xec); |
2863 |
|
|
tos_make(r); |
2864 |
|
|
} |
2865 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_log10_2,(FW r)) |
2866 |
|
|
|
2867 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_log2_e,(FW r)) |
2868 |
|
|
{ |
2869 |
|
|
emit_byte(0xd9); |
2870 |
|
|
emit_byte(0xea); |
2871 |
|
|
tos_make(r); |
2872 |
|
|
} |
2873 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_log2_e,(FW r)) |
2874 |
|
|
|
2875 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_loge_2,(FW r)) |
2876 |
|
|
{ |
2877 |
|
|
emit_byte(0xd9); |
2878 |
|
|
emit_byte(0xed); |
2879 |
|
|
tos_make(r); |
2880 |
|
|
} |
2881 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_loge_2,(FW r)) |
2882 |
|
|
|
2883 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_1,(FW r)) |
2884 |
|
|
{ |
2885 |
|
|
emit_byte(0xd9); |
2886 |
|
|
emit_byte(0xe8); |
2887 |
|
|
tos_make(r); |
2888 |
|
|
} |
2889 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_1,(FW r)) |
2890 |
|
|
|
2891 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_0,(FW r)) |
2892 |
|
|
{ |
2893 |
|
|
emit_byte(0xd9); |
2894 |
|
|
emit_byte(0xee); |
2895 |
|
|
tos_make(r); |
2896 |
|
|
} |
2897 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_0,(FW r)) |
2898 |
|
|
|
2899 |
|
|
LOWFUNC(NONE,NONE,2,raw_fmov_rr,(FW d, FR s)) |
2900 |
|
|
{ |
2901 |
|
|
int ds; |
2902 |
|
|
|
2903 |
|
|
usereg(s); |
2904 |
|
|
ds=stackpos(s); |
2905 |
|
|
if (ds==0 && live.spos[d]>=0) { |
2906 |
|
|
/* source is on top of stack, and we already have the dest */ |
2907 |
|
|
int dd=stackpos(d); |
2908 |
|
|
emit_byte(0xdd); |
2909 |
|
|
emit_byte(0xd0+dd); |
2910 |
|
|
} |
2911 |
|
|
else { |
2912 |
|
|
emit_byte(0xd9); |
2913 |
|
|
emit_byte(0xc0+ds); /* duplicate source on tos */ |
2914 |
|
|
tos_make(d); /* store to destination, pop if necessary */ |
2915 |
|
|
} |
2916 |
|
|
} |
2917 |
|
|
LENDFUNC(NONE,NONE,2,raw_fmov_rr,(FW d, FR s)) |
2918 |
|
|
|
2919 |
|
|
LOWFUNC(NONE,READ,4,raw_fldcw_m_indexed,(R4 index, IMM base)) |
2920 |
|
|
{ |
2921 |
|
|
emit_byte(0xd9); |
2922 |
|
|
emit_byte(0xa8+index); |
2923 |
|
|
emit_long(base); |
2924 |
|
|
} |
2925 |
|
|
LENDFUNC(NONE,READ,4,raw_fldcw_m_indexed,(R4 index, IMM base)) |
2926 |
|
|
|
2927 |
|
|
|
2928 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsqrt_rr,(FW d, FR s)) |
2929 |
|
|
{ |
2930 |
|
|
int ds; |
2931 |
|
|
|
2932 |
|
|
if (d!=s) { |
2933 |
|
|
usereg(s); |
2934 |
|
|
ds=stackpos(s); |
2935 |
|
|
emit_byte(0xd9); |
2936 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2937 |
|
|
emit_byte(0xd9); |
2938 |
|
|
emit_byte(0xfa); /* take square root */ |
2939 |
|
|
tos_make(d); /* store to destination */ |
2940 |
|
|
} |
2941 |
|
|
else { |
2942 |
|
|
make_tos(d); |
2943 |
|
|
emit_byte(0xd9); |
2944 |
|
|
emit_byte(0xfa); /* take square root */ |
2945 |
|
|
} |
2946 |
|
|
} |
2947 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsqrt_rr,(FW d, FR s)) |
2948 |
|
|
|
2949 |
|
|
LOWFUNC(NONE,NONE,2,raw_fabs_rr,(FW d, FR s)) |
2950 |
|
|
{ |
2951 |
|
|
int ds; |
2952 |
|
|
|
2953 |
|
|
if (d!=s) { |
2954 |
|
|
usereg(s); |
2955 |
|
|
ds=stackpos(s); |
2956 |
|
|
emit_byte(0xd9); |
2957 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2958 |
|
|
emit_byte(0xd9); |
2959 |
|
|
emit_byte(0xe1); /* take fabs */ |
2960 |
|
|
tos_make(d); /* store to destination */ |
2961 |
|
|
} |
2962 |
|
|
else { |
2963 |
|
|
make_tos(d); |
2964 |
|
|
emit_byte(0xd9); |
2965 |
|
|
emit_byte(0xe1); /* take fabs */ |
2966 |
|
|
} |
2967 |
|
|
} |
2968 |
|
|
LENDFUNC(NONE,NONE,2,raw_fabs_rr,(FW d, FR s)) |
2969 |
|
|
|
2970 |
|
|
LOWFUNC(NONE,NONE,2,raw_frndint_rr,(FW d, FR s)) |
2971 |
|
|
{ |
2972 |
|
|
int ds; |
2973 |
|
|
|
2974 |
|
|
if (d!=s) { |
2975 |
|
|
usereg(s); |
2976 |
|
|
ds=stackpos(s); |
2977 |
|
|
emit_byte(0xd9); |
2978 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2979 |
|
|
emit_byte(0xd9); |
2980 |
|
|
emit_byte(0xfc); /* take frndint */ |
2981 |
|
|
tos_make(d); /* store to destination */ |
2982 |
|
|
} |
2983 |
|
|
else { |
2984 |
|
|
make_tos(d); |
2985 |
|
|
emit_byte(0xd9); |
2986 |
|
|
emit_byte(0xfc); /* take frndint */ |
2987 |
|
|
} |
2988 |
|
|
} |
2989 |
|
|
LENDFUNC(NONE,NONE,2,raw_frndint_rr,(FW d, FR s)) |
2990 |
|
|
|
2991 |
|
|
LOWFUNC(NONE,NONE,2,raw_fcos_rr,(FW d, FR s)) |
2992 |
|
|
{ |
2993 |
|
|
int ds; |
2994 |
|
|
|
2995 |
|
|
if (d!=s) { |
2996 |
|
|
usereg(s); |
2997 |
|
|
ds=stackpos(s); |
2998 |
|
|
emit_byte(0xd9); |
2999 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3000 |
|
|
emit_byte(0xd9); |
3001 |
|
|
emit_byte(0xff); /* take cos */ |
3002 |
|
|
tos_make(d); /* store to destination */ |
3003 |
|
|
} |
3004 |
|
|
else { |
3005 |
|
|
make_tos(d); |
3006 |
|
|
emit_byte(0xd9); |
3007 |
|
|
emit_byte(0xff); /* take cos */ |
3008 |
|
|
} |
3009 |
|
|
} |
3010 |
|
|
LENDFUNC(NONE,NONE,2,raw_fcos_rr,(FW d, FR s)) |
3011 |
|
|
|
3012 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsin_rr,(FW d, FR s)) |
3013 |
|
|
{ |
3014 |
|
|
int ds; |
3015 |
|
|
|
3016 |
|
|
if (d!=s) { |
3017 |
|
|
usereg(s); |
3018 |
|
|
ds=stackpos(s); |
3019 |
|
|
emit_byte(0xd9); |
3020 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3021 |
|
|
emit_byte(0xd9); |
3022 |
|
|
emit_byte(0xfe); /* take sin */ |
3023 |
|
|
tos_make(d); /* store to destination */ |
3024 |
|
|
} |
3025 |
|
|
else { |
3026 |
|
|
make_tos(d); |
3027 |
|
|
emit_byte(0xd9); |
3028 |
|
|
emit_byte(0xfe); /* take sin */ |
3029 |
|
|
} |
3030 |
|
|
} |
3031 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsin_rr,(FW d, FR s)) |
3032 |
|
|
|
3033 |
|
|
double one=1; |
3034 |
|
|
LOWFUNC(NONE,NONE,2,raw_ftwotox_rr,(FW d, FR s)) |
3035 |
|
|
{ |
3036 |
|
|
int ds; |
3037 |
|
|
|
3038 |
|
|
usereg(s); |
3039 |
|
|
ds=stackpos(s); |
3040 |
|
|
emit_byte(0xd9); |
3041 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3042 |
|
|
|
3043 |
|
|
emit_byte(0xd9); |
3044 |
|
|
emit_byte(0xc0); /* duplicate top of stack. Now up to 8 high */ |
3045 |
|
|
emit_byte(0xd9); |
3046 |
|
|
emit_byte(0xfc); /* rndint */ |
3047 |
|
|
emit_byte(0xd9); |
3048 |
|
|
emit_byte(0xc9); /* swap top two elements */ |
3049 |
|
|
emit_byte(0xd8); |
3050 |
|
|
emit_byte(0xe1); /* subtract rounded from original */ |
3051 |
|
|
emit_byte(0xd9); |
3052 |
|
|
emit_byte(0xf0); /* f2xm1 */ |
3053 |
|
|
emit_byte(0xdc); |
3054 |
|
|
emit_byte(0x05); |
3055 |
|
|
emit_long((uae_u32)&one); /* Add '1' without using extra stack space */ |
3056 |
|
|
emit_byte(0xd9); |
3057 |
|
|
emit_byte(0xfd); /* and scale it */ |
3058 |
|
|
emit_byte(0xdd); |
3059 |
|
|
emit_byte(0xd9); /* take he rounded value off */ |
3060 |
|
|
tos_make(d); /* store to destination */ |
3061 |
|
|
} |
3062 |
|
|
LENDFUNC(NONE,NONE,2,raw_ftwotox_rr,(FW d, FR s)) |
3063 |
|
|
|
3064 |
|
|
LOWFUNC(NONE,NONE,2,raw_fetox_rr,(FW d, FR s)) |
3065 |
|
|
{ |
3066 |
|
|
int ds; |
3067 |
|
|
|
3068 |
|
|
usereg(s); |
3069 |
|
|
ds=stackpos(s); |
3070 |
|
|
emit_byte(0xd9); |
3071 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3072 |
|
|
emit_byte(0xd9); |
3073 |
|
|
emit_byte(0xea); /* fldl2e */ |
3074 |
|
|
emit_byte(0xde); |
3075 |
|
|
emit_byte(0xc9); /* fmulp --- multiply source by log2(e) */ |
3076 |
|
|
|
3077 |
|
|
emit_byte(0xd9); |
3078 |
|
|
emit_byte(0xc0); /* duplicate top of stack. Now up to 8 high */ |
3079 |
|
|
emit_byte(0xd9); |
3080 |
|
|
emit_byte(0xfc); /* rndint */ |
3081 |
|
|
emit_byte(0xd9); |
3082 |
|
|
emit_byte(0xc9); /* swap top two elements */ |
3083 |
|
|
emit_byte(0xd8); |
3084 |
|
|
emit_byte(0xe1); /* subtract rounded from original */ |
3085 |
|
|
emit_byte(0xd9); |
3086 |
|
|
emit_byte(0xf0); /* f2xm1 */ |
3087 |
|
|
emit_byte(0xdc); |
3088 |
|
|
emit_byte(0x05); |
3089 |
|
|
emit_long((uae_u32)&one); /* Add '1' without using extra stack space */ |
3090 |
|
|
emit_byte(0xd9); |
3091 |
|
|
emit_byte(0xfd); /* and scale it */ |
3092 |
|
|
emit_byte(0xdd); |
3093 |
|
|
emit_byte(0xd9); /* take he rounded value off */ |
3094 |
|
|
tos_make(d); /* store to destination */ |
3095 |
|
|
} |
3096 |
|
|
LENDFUNC(NONE,NONE,2,raw_fetox_rr,(FW d, FR s)) |
3097 |
|
|
|
3098 |
|
|
LOWFUNC(NONE,NONE,2,raw_flog2_rr,(FW d, FR s)) |
3099 |
|
|
{ |
3100 |
|
|
int ds; |
3101 |
|
|
|
3102 |
|
|
usereg(s); |
3103 |
|
|
ds=stackpos(s); |
3104 |
|
|
emit_byte(0xd9); |
3105 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3106 |
|
|
emit_byte(0xd9); |
3107 |
|
|
emit_byte(0xe8); /* push '1' */ |
3108 |
|
|
emit_byte(0xd9); |
3109 |
|
|
emit_byte(0xc9); /* swap top two */ |
3110 |
|
|
emit_byte(0xd9); |
3111 |
|
|
emit_byte(0xf1); /* take 1*log2(x) */ |
3112 |
|
|
tos_make(d); /* store to destination */ |
3113 |
|
|
} |
3114 |
|
|
LENDFUNC(NONE,NONE,2,raw_flog2_rr,(FW d, FR s)) |
3115 |
|
|
|
3116 |
|
|
|
3117 |
|
|
LOWFUNC(NONE,NONE,2,raw_fneg_rr,(FW d, FR s)) |
3118 |
|
|
{ |
3119 |
|
|
int ds; |
3120 |
|
|
|
3121 |
|
|
if (d!=s) { |
3122 |
|
|
usereg(s); |
3123 |
|
|
ds=stackpos(s); |
3124 |
|
|
emit_byte(0xd9); |
3125 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3126 |
|
|
emit_byte(0xd9); |
3127 |
|
|
emit_byte(0xe0); /* take fchs */ |
3128 |
|
|
tos_make(d); /* store to destination */ |
3129 |
|
|
} |
3130 |
|
|
else { |
3131 |
|
|
make_tos(d); |
3132 |
|
|
emit_byte(0xd9); |
3133 |
|
|
emit_byte(0xe0); /* take fchs */ |
3134 |
|
|
} |
3135 |
|
|
} |
3136 |
|
|
LENDFUNC(NONE,NONE,2,raw_fneg_rr,(FW d, FR s)) |
3137 |
|
|
|
3138 |
|
|
LOWFUNC(NONE,NONE,2,raw_fadd_rr,(FRW d, FR s)) |
3139 |
|
|
{ |
3140 |
|
|
int ds; |
3141 |
|
|
|
3142 |
|
|
usereg(s); |
3143 |
|
|
usereg(d); |
3144 |
|
|
|
3145 |
|
|
if (live.spos[s]==live.tos) { |
3146 |
|
|
/* Source is on top of stack */ |
3147 |
|
|
ds=stackpos(d); |
3148 |
|
|
emit_byte(0xdc); |
3149 |
|
|
emit_byte(0xc0+ds); /* add source to dest*/ |
3150 |
|
|
} |
3151 |
|
|
else { |
3152 |
|
|
make_tos(d); |
3153 |
|
|
ds=stackpos(s); |
3154 |
|
|
|
3155 |
|
|
emit_byte(0xd8); |
3156 |
|
|
emit_byte(0xc0+ds); /* add source to dest*/ |
3157 |
|
|
} |
3158 |
|
|
} |
3159 |
|
|
LENDFUNC(NONE,NONE,2,raw_fadd_rr,(FRW d, FR s)) |
3160 |
|
|
|
3161 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsub_rr,(FRW d, FR s)) |
3162 |
|
|
{ |
3163 |
|
|
int ds; |
3164 |
|
|
|
3165 |
|
|
usereg(s); |
3166 |
|
|
usereg(d); |
3167 |
|
|
|
3168 |
|
|
if (live.spos[s]==live.tos) { |
3169 |
|
|
/* Source is on top of stack */ |
3170 |
|
|
ds=stackpos(d); |
3171 |
|
|
emit_byte(0xdc); |
3172 |
|
|
emit_byte(0xe8+ds); /* sub source from dest*/ |
3173 |
|
|
} |
3174 |
|
|
else { |
3175 |
|
|
make_tos(d); |
3176 |
|
|
ds=stackpos(s); |
3177 |
|
|
|
3178 |
|
|
emit_byte(0xd8); |
3179 |
|
|
emit_byte(0xe0+ds); /* sub src from dest */ |
3180 |
|
|
} |
3181 |
|
|
} |
3182 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsub_rr,(FRW d, FR s)) |
3183 |
|
|
|
3184 |
|
|
LOWFUNC(NONE,NONE,2,raw_fcmp_rr,(FR d, FR s)) |
3185 |
|
|
{ |
3186 |
|
|
int ds; |
3187 |
|
|
|
3188 |
|
|
usereg(s); |
3189 |
|
|
usereg(d); |
3190 |
|
|
|
3191 |
|
|
make_tos(d); |
3192 |
|
|
ds=stackpos(s); |
3193 |
|
|
|
3194 |
|
|
emit_byte(0xdd); |
3195 |
|
|
emit_byte(0xe0+ds); /* cmp dest with source*/ |
3196 |
|
|
} |
3197 |
|
|
LENDFUNC(NONE,NONE,2,raw_fcmp_rr,(FR d, FR s)) |
3198 |
|
|
|
3199 |
|
|
LOWFUNC(NONE,NONE,2,raw_fmul_rr,(FRW d, FR s)) |
3200 |
|
|
{ |
3201 |
|
|
int ds; |
3202 |
|
|
|
3203 |
|
|
usereg(s); |
3204 |
|
|
usereg(d); |
3205 |
|
|
|
3206 |
|
|
if (live.spos[s]==live.tos) { |
3207 |
|
|
/* Source is on top of stack */ |
3208 |
|
|
ds=stackpos(d); |
3209 |
|
|
emit_byte(0xdc); |
3210 |
|
|
emit_byte(0xc8+ds); /* mul dest by source*/ |
3211 |
|
|
} |
3212 |
|
|
else { |
3213 |
|
|
make_tos(d); |
3214 |
|
|
ds=stackpos(s); |
3215 |
|
|
|
3216 |
|
|
emit_byte(0xd8); |
3217 |
|
|
emit_byte(0xc8+ds); /* mul dest by source*/ |
3218 |
|
|
} |
3219 |
|
|
} |
3220 |
|
|
LENDFUNC(NONE,NONE,2,raw_fmul_rr,(FRW d, FR s)) |
3221 |
|
|
|
3222 |
|
|
LOWFUNC(NONE,NONE,2,raw_fdiv_rr,(FRW d, FR s)) |
3223 |
|
|
{ |
3224 |
|
|
int ds; |
3225 |
|
|
|
3226 |
|
|
usereg(s); |
3227 |
|
|
usereg(d); |
3228 |
|
|
|
3229 |
|
|
if (live.spos[s]==live.tos) { |
3230 |
|
|
/* Source is on top of stack */ |
3231 |
|
|
ds=stackpos(d); |
3232 |
|
|
emit_byte(0xdc); |
3233 |
|
|
emit_byte(0xf8+ds); /* div dest by source */ |
3234 |
|
|
} |
3235 |
|
|
else { |
3236 |
|
|
make_tos(d); |
3237 |
|
|
ds=stackpos(s); |
3238 |
|
|
|
3239 |
|
|
emit_byte(0xd8); |
3240 |
|
|
emit_byte(0xf0+ds); /* div dest by source*/ |
3241 |
|
|
} |
3242 |
|
|
} |
3243 |
|
|
LENDFUNC(NONE,NONE,2,raw_fdiv_rr,(FRW d, FR s)) |
3244 |
|
|
|
3245 |
|
|
LOWFUNC(NONE,NONE,2,raw_frem_rr,(FRW d, FR s)) |
3246 |
|
|
{ |
3247 |
|
|
int ds; |
3248 |
|
|
|
3249 |
|
|
usereg(s); |
3250 |
|
|
usereg(d); |
3251 |
|
|
|
3252 |
|
|
make_tos2(d,s); |
3253 |
|
|
ds=stackpos(s); |
3254 |
|
|
|
3255 |
|
|
if (ds!=1) { |
3256 |
|
|
printf("Failed horribly in raw_frem_rr! ds is %d\n",ds); |
3257 |
|
|
abort(); |
3258 |
|
|
} |
3259 |
|
|
emit_byte(0xd9); |
3260 |
|
|
emit_byte(0xf8); /* take rem from dest by source */ |
3261 |
|
|
} |
3262 |
|
|
LENDFUNC(NONE,NONE,2,raw_frem_rr,(FRW d, FR s)) |
3263 |
|
|
|
3264 |
|
|
LOWFUNC(NONE,NONE,2,raw_frem1_rr,(FRW d, FR s)) |
3265 |
|
|
{ |
3266 |
|
|
int ds; |
3267 |
|
|
|
3268 |
|
|
usereg(s); |
3269 |
|
|
usereg(d); |
3270 |
|
|
|
3271 |
|
|
make_tos2(d,s); |
3272 |
|
|
ds=stackpos(s); |
3273 |
|
|
|
3274 |
|
|
if (ds!=1) { |
3275 |
|
|
printf("Failed horribly in raw_frem1_rr! ds is %d\n",ds); |
3276 |
|
|
abort(); |
3277 |
|
|
} |
3278 |
|
|
emit_byte(0xd9); |
3279 |
|
|
emit_byte(0xf5); /* take rem1 from dest by source */ |
3280 |
|
|
} |
3281 |
|
|
LENDFUNC(NONE,NONE,2,raw_frem1_rr,(FRW d, FR s)) |
3282 |
|
|
|
3283 |
|
|
|
3284 |
|
|
LOWFUNC(NONE,NONE,1,raw_ftst_r,(FR r)) |
3285 |
|
|
{ |
3286 |
|
|
make_tos(r); |
3287 |
|
|
emit_byte(0xd9); /* ftst */ |
3288 |
|
|
emit_byte(0xe4); |
3289 |
|
|
} |
3290 |
|
|
LENDFUNC(NONE,NONE,1,raw_ftst_r,(FR r)) |
3291 |
|
|
|
3292 |
|
|
/* %eax register is clobbered if target processor doesn't support fucomi */ |
3293 |
|
|
#define FFLAG_NREG_CLOBBER_CONDITION !have_cmov |
3294 |
|
|
#define FFLAG_NREG EAX_INDEX |
3295 |
|
|
|
3296 |
|
|
static __inline__ void raw_fflags_into_flags(int r) |
3297 |
|
|
{ |
3298 |
|
|
int p; |
3299 |
|
|
|
3300 |
|
|
usereg(r); |
3301 |
|
|
p=stackpos(r); |
3302 |
|
|
|
3303 |
|
|
emit_byte(0xd9); |
3304 |
|
|
emit_byte(0xee); /* Push 0 */ |
3305 |
|
|
emit_byte(0xd9); |
3306 |
|
|
emit_byte(0xc9+p); /* swap top two around */ |
3307 |
|
|
if (have_cmov) { |
3308 |
|
|
// gb-- fucomi is for P6 cores only, not K6-2 then... |
3309 |
|
|
emit_byte(0xdb); |
3310 |
|
|
emit_byte(0xe9+p); /* fucomi them */ |
3311 |
|
|
} |
3312 |
|
|
else { |
3313 |
|
|
emit_byte(0xdd); |
3314 |
|
|
emit_byte(0xe1+p); /* fucom them */ |
3315 |
|
|
emit_byte(0x9b); |
3316 |
|
|
emit_byte(0xdf); |
3317 |
|
|
emit_byte(0xe0); /* fstsw ax */ |
3318 |
|
|
raw_sahf(0); /* sahf */ |
3319 |
|
|
} |
3320 |
|
|
emit_byte(0xdd); |
3321 |
|
|
emit_byte(0xd9+p); /* store value back, and get rid of 0 */ |
3322 |
|
|
} |