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 |
gbeauche |
1.7 |
* |
11 |
|
|
* Portions related to CPU detection come from linux/arch/i386/kernel/setup.c |
12 |
gbeauche |
1.6 |
* |
13 |
|
|
* This program is free software; you can redistribute it and/or modify |
14 |
|
|
* it under the terms of the GNU General Public License as published by |
15 |
|
|
* the Free Software Foundation; either version 2 of the License, or |
16 |
|
|
* (at your option) any later version. |
17 |
|
|
* |
18 |
|
|
* This program is distributed in the hope that it will be useful, |
19 |
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
20 |
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
21 |
|
|
* GNU General Public License for more details. |
22 |
|
|
* |
23 |
|
|
* You should have received a copy of the GNU General Public License |
24 |
|
|
* along with this program; if not, write to the Free Software |
25 |
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
26 |
|
|
*/ |
27 |
|
|
|
28 |
gbeauche |
1.1 |
/* This should eventually end up in machdep/, but for now, x86 is the |
29 |
|
|
only target, and it's easier this way... */ |
30 |
|
|
|
31 |
gbeauche |
1.5 |
#include "flags_x86.h" |
32 |
|
|
|
33 |
gbeauche |
1.1 |
/************************************************************************* |
34 |
|
|
* Some basic information about the the target CPU * |
35 |
|
|
*************************************************************************/ |
36 |
|
|
|
37 |
|
|
#define EAX_INDEX 0 |
38 |
|
|
#define ECX_INDEX 1 |
39 |
|
|
#define EDX_INDEX 2 |
40 |
|
|
#define EBX_INDEX 3 |
41 |
|
|
#define ESP_INDEX 4 |
42 |
|
|
#define EBP_INDEX 5 |
43 |
|
|
#define ESI_INDEX 6 |
44 |
|
|
#define EDI_INDEX 7 |
45 |
|
|
|
46 |
|
|
/* The register in which subroutines return an integer return value */ |
47 |
|
|
#define REG_RESULT 0 |
48 |
|
|
|
49 |
|
|
/* The registers subroutines take their first and second argument in */ |
50 |
|
|
#if defined( _MSC_VER ) && !defined( USE_NORMAL_CALLING_CONVENTION ) |
51 |
|
|
/* Handle the _fastcall parameters of ECX and EDX */ |
52 |
|
|
#define REG_PAR1 1 |
53 |
|
|
#define REG_PAR2 2 |
54 |
|
|
#else |
55 |
|
|
#define REG_PAR1 0 |
56 |
|
|
#define REG_PAR2 2 |
57 |
|
|
#endif |
58 |
|
|
|
59 |
|
|
/* Three registers that are not used for any of the above */ |
60 |
|
|
#define REG_NOPAR1 6 |
61 |
|
|
#define REG_NOPAR2 5 |
62 |
|
|
#define REG_NOPAR3 3 |
63 |
|
|
|
64 |
|
|
#define REG_PC_PRE 0 /* The register we use for preloading regs.pc_p */ |
65 |
|
|
#if defined( _MSC_VER ) && !defined( USE_NORMAL_CALLING_CONVENTION ) |
66 |
|
|
#define REG_PC_TMP 0 |
67 |
|
|
#else |
68 |
|
|
#define REG_PC_TMP 1 /* Another register that is not the above */ |
69 |
|
|
#endif |
70 |
|
|
|
71 |
|
|
#define SHIFTCOUNT_NREG 1 /* Register that can be used for shiftcount. |
72 |
|
|
-1 if any reg will do */ |
73 |
|
|
#define MUL_NREG1 0 /* %eax will hold the low 32 bits after a 32x32 mul */ |
74 |
|
|
#define MUL_NREG2 2 /* %edx will hold the high 32 bits */ |
75 |
|
|
|
76 |
|
|
uae_s8 always_used[]={4,-1}; |
77 |
|
|
uae_s8 can_byte[]={0,1,2,3,-1}; |
78 |
|
|
uae_s8 can_word[]={0,1,2,3,5,6,7,-1}; |
79 |
|
|
|
80 |
|
|
/* cpuopti mutate instruction handlers to assume registers are saved |
81 |
|
|
by the caller */ |
82 |
|
|
uae_u8 call_saved[]={0,0,0,0,1,0,0,0}; |
83 |
|
|
|
84 |
|
|
/* This *should* be the same as call_saved. But: |
85 |
|
|
- We might not really know which registers are saved, and which aren't, |
86 |
|
|
so we need to preserve some, but don't want to rely on everyone else |
87 |
|
|
also saving those registers |
88 |
|
|
- Special registers (such like the stack pointer) should not be "preserved" |
89 |
|
|
by pushing, even though they are "saved" across function calls |
90 |
|
|
*/ |
91 |
|
|
uae_u8 need_to_preserve[]={1,1,1,1,0,1,1,1}; |
92 |
|
|
|
93 |
|
|
/* Whether classes of instructions do or don't clobber the native flags */ |
94 |
|
|
#define CLOBBER_MOV |
95 |
|
|
#define CLOBBER_LEA |
96 |
|
|
#define CLOBBER_CMOV |
97 |
|
|
#define CLOBBER_POP |
98 |
|
|
#define CLOBBER_PUSH |
99 |
|
|
#define CLOBBER_SUB clobber_flags() |
100 |
|
|
#define CLOBBER_SBB clobber_flags() |
101 |
|
|
#define CLOBBER_CMP clobber_flags() |
102 |
|
|
#define CLOBBER_ADD clobber_flags() |
103 |
|
|
#define CLOBBER_ADC clobber_flags() |
104 |
|
|
#define CLOBBER_AND clobber_flags() |
105 |
|
|
#define CLOBBER_OR clobber_flags() |
106 |
|
|
#define CLOBBER_XOR clobber_flags() |
107 |
|
|
|
108 |
|
|
#define CLOBBER_ROL clobber_flags() |
109 |
|
|
#define CLOBBER_ROR clobber_flags() |
110 |
|
|
#define CLOBBER_SHLL clobber_flags() |
111 |
|
|
#define CLOBBER_SHRL clobber_flags() |
112 |
|
|
#define CLOBBER_SHRA clobber_flags() |
113 |
|
|
#define CLOBBER_TEST clobber_flags() |
114 |
|
|
#define CLOBBER_CL16 |
115 |
|
|
#define CLOBBER_CL8 |
116 |
|
|
#define CLOBBER_SE16 |
117 |
|
|
#define CLOBBER_SE8 |
118 |
|
|
#define CLOBBER_ZE16 |
119 |
|
|
#define CLOBBER_ZE8 |
120 |
|
|
#define CLOBBER_SW16 clobber_flags() |
121 |
|
|
#define CLOBBER_SW32 |
122 |
|
|
#define CLOBBER_SETCC |
123 |
|
|
#define CLOBBER_MUL clobber_flags() |
124 |
|
|
#define CLOBBER_BT clobber_flags() |
125 |
|
|
#define CLOBBER_BSF clobber_flags() |
126 |
|
|
|
127 |
gbeauche |
1.2 |
const bool optimize_accum = true; |
128 |
gbeauche |
1.1 |
const bool optimize_imm8 = true; |
129 |
|
|
const bool optimize_shift_once = true; |
130 |
|
|
|
131 |
|
|
/************************************************************************* |
132 |
|
|
* Actual encoding of the instructions on the target CPU * |
133 |
|
|
*************************************************************************/ |
134 |
|
|
|
135 |
gbeauche |
1.2 |
static __inline__ int isaccum(int r) |
136 |
|
|
{ |
137 |
|
|
return (r == EAX_INDEX); |
138 |
|
|
} |
139 |
|
|
|
140 |
gbeauche |
1.1 |
static __inline__ int isbyte(uae_s32 x) |
141 |
|
|
{ |
142 |
|
|
return (x>=-128 && x<=127); |
143 |
|
|
} |
144 |
|
|
|
145 |
|
|
static __inline__ int isword(uae_s32 x) |
146 |
|
|
{ |
147 |
|
|
return (x>=-32768 && x<=32767); |
148 |
|
|
} |
149 |
|
|
|
150 |
|
|
LOWFUNC(NONE,WRITE,1,raw_push_l_r,(R4 r)) |
151 |
|
|
{ |
152 |
|
|
emit_byte(0x50+r); |
153 |
|
|
} |
154 |
|
|
LENDFUNC(NONE,WRITE,1,raw_push_l_r,(R4 r)) |
155 |
|
|
|
156 |
|
|
LOWFUNC(NONE,READ,1,raw_pop_l_r,(R4 r)) |
157 |
|
|
{ |
158 |
|
|
emit_byte(0x58+r); |
159 |
|
|
} |
160 |
|
|
LENDFUNC(NONE,READ,1,raw_pop_l_r,(R4 r)) |
161 |
|
|
|
162 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bt_l_ri,(R4 r, IMM i)) |
163 |
|
|
{ |
164 |
|
|
emit_byte(0x0f); |
165 |
|
|
emit_byte(0xba); |
166 |
|
|
emit_byte(0xe0+r); |
167 |
|
|
emit_byte(i); |
168 |
|
|
} |
169 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bt_l_ri,(R4 r, IMM i)) |
170 |
|
|
|
171 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bt_l_rr,(R4 r, R4 b)) |
172 |
|
|
{ |
173 |
|
|
emit_byte(0x0f); |
174 |
|
|
emit_byte(0xa3); |
175 |
|
|
emit_byte(0xc0+8*b+r); |
176 |
|
|
} |
177 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bt_l_rr,(R4 r, R4 b)) |
178 |
|
|
|
179 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btc_l_ri,(RW4 r, IMM i)) |
180 |
|
|
{ |
181 |
|
|
emit_byte(0x0f); |
182 |
|
|
emit_byte(0xba); |
183 |
|
|
emit_byte(0xf8+r); |
184 |
|
|
emit_byte(i); |
185 |
|
|
} |
186 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btc_l_ri,(RW4 r, IMM i)) |
187 |
|
|
|
188 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btc_l_rr,(RW4 r, R4 b)) |
189 |
|
|
{ |
190 |
|
|
emit_byte(0x0f); |
191 |
|
|
emit_byte(0xbb); |
192 |
|
|
emit_byte(0xc0+8*b+r); |
193 |
|
|
} |
194 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btc_l_rr,(RW4 r, R4 b)) |
195 |
|
|
|
196 |
|
|
|
197 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btr_l_ri,(RW4 r, IMM i)) |
198 |
|
|
{ |
199 |
|
|
emit_byte(0x0f); |
200 |
|
|
emit_byte(0xba); |
201 |
|
|
emit_byte(0xf0+r); |
202 |
|
|
emit_byte(i); |
203 |
|
|
} |
204 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btr_l_ri,(RW4 r, IMM i)) |
205 |
|
|
|
206 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btr_l_rr,(RW4 r, R4 b)) |
207 |
|
|
{ |
208 |
|
|
emit_byte(0x0f); |
209 |
|
|
emit_byte(0xb3); |
210 |
|
|
emit_byte(0xc0+8*b+r); |
211 |
|
|
} |
212 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btr_l_rr,(RW4 r, R4 b)) |
213 |
|
|
|
214 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bts_l_ri,(RW4 r, IMM i)) |
215 |
|
|
{ |
216 |
|
|
emit_byte(0x0f); |
217 |
|
|
emit_byte(0xba); |
218 |
|
|
emit_byte(0xe8+r); |
219 |
|
|
emit_byte(i); |
220 |
|
|
} |
221 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bts_l_ri,(RW4 r, IMM i)) |
222 |
|
|
|
223 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bts_l_rr,(RW4 r, R4 b)) |
224 |
|
|
{ |
225 |
|
|
emit_byte(0x0f); |
226 |
|
|
emit_byte(0xab); |
227 |
|
|
emit_byte(0xc0+8*b+r); |
228 |
|
|
} |
229 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bts_l_rr,(RW4 r, R4 b)) |
230 |
|
|
|
231 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_w_ri,(RW2 d, IMM i)) |
232 |
|
|
{ |
233 |
|
|
emit_byte(0x66); |
234 |
|
|
if (isbyte(i)) { |
235 |
|
|
emit_byte(0x83); |
236 |
|
|
emit_byte(0xe8+d); |
237 |
|
|
emit_byte(i); |
238 |
|
|
} |
239 |
|
|
else { |
240 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
241 |
|
|
emit_byte(0x2d); |
242 |
|
|
else { |
243 |
gbeauche |
1.1 |
emit_byte(0x81); |
244 |
|
|
emit_byte(0xe8+d); |
245 |
gbeauche |
1.2 |
} |
246 |
gbeauche |
1.1 |
emit_word(i); |
247 |
|
|
} |
248 |
|
|
} |
249 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_w_ri,(RW2 d, IMM i)) |
250 |
|
|
|
251 |
|
|
|
252 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_l_rm,(W4 d, MEMR s)) |
253 |
|
|
{ |
254 |
|
|
emit_byte(0x8b); |
255 |
|
|
emit_byte(0x05+8*d); |
256 |
|
|
emit_long(s); |
257 |
|
|
} |
258 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_l_rm,(W4 d, MEMR s)) |
259 |
|
|
|
260 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_l_mi,(MEMW d, IMM s)) |
261 |
|
|
{ |
262 |
|
|
emit_byte(0xc7); |
263 |
|
|
emit_byte(0x05); |
264 |
|
|
emit_long(d); |
265 |
|
|
emit_long(s); |
266 |
|
|
} |
267 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_l_mi,(MEMW d, IMM s)) |
268 |
|
|
|
269 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_w_mi,(MEMW d, IMM s)) |
270 |
|
|
{ |
271 |
|
|
emit_byte(0x66); |
272 |
|
|
emit_byte(0xc7); |
273 |
|
|
emit_byte(0x05); |
274 |
|
|
emit_long(d); |
275 |
|
|
emit_word(s); |
276 |
|
|
} |
277 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_w_mi,(MEMW d, IMM s)) |
278 |
|
|
|
279 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_b_mi,(MEMW d, IMM s)) |
280 |
|
|
{ |
281 |
|
|
emit_byte(0xc6); |
282 |
|
|
emit_byte(0x05); |
283 |
|
|
emit_long(d); |
284 |
|
|
emit_byte(s); |
285 |
|
|
} |
286 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_b_mi,(MEMW d, IMM s)) |
287 |
|
|
|
288 |
|
|
LOWFUNC(WRITE,RMW,2,raw_rol_b_mi,(MEMRW d, IMM i)) |
289 |
|
|
{ |
290 |
|
|
if (optimize_shift_once && (i == 1)) { |
291 |
|
|
emit_byte(0xd0); |
292 |
|
|
emit_byte(0x05); |
293 |
|
|
emit_long(d); |
294 |
|
|
} |
295 |
|
|
else { |
296 |
|
|
emit_byte(0xc0); |
297 |
|
|
emit_byte(0x05); |
298 |
|
|
emit_long(d); |
299 |
|
|
emit_byte(i); |
300 |
|
|
} |
301 |
|
|
} |
302 |
|
|
LENDFUNC(WRITE,RMW,2,raw_rol_b_mi,(MEMRW d, IMM i)) |
303 |
|
|
|
304 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_b_ri,(RW1 r, IMM i)) |
305 |
|
|
{ |
306 |
|
|
if (optimize_shift_once && (i == 1)) { |
307 |
|
|
emit_byte(0xd0); |
308 |
|
|
emit_byte(0xc0+r); |
309 |
|
|
} |
310 |
|
|
else { |
311 |
|
|
emit_byte(0xc0); |
312 |
|
|
emit_byte(0xc0+r); |
313 |
|
|
emit_byte(i); |
314 |
|
|
} |
315 |
|
|
} |
316 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_b_ri,(RW1 r, IMM i)) |
317 |
|
|
|
318 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_w_ri,(RW2 r, IMM i)) |
319 |
|
|
{ |
320 |
|
|
emit_byte(0x66); |
321 |
|
|
emit_byte(0xc1); |
322 |
|
|
emit_byte(0xc0+r); |
323 |
|
|
emit_byte(i); |
324 |
|
|
} |
325 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_w_ri,(RW2 r, IMM i)) |
326 |
|
|
|
327 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_l_ri,(RW4 r, IMM i)) |
328 |
|
|
{ |
329 |
|
|
if (optimize_shift_once && (i == 1)) { |
330 |
|
|
emit_byte(0xd1); |
331 |
|
|
emit_byte(0xc0+r); |
332 |
|
|
} |
333 |
|
|
else { |
334 |
|
|
emit_byte(0xc1); |
335 |
|
|
emit_byte(0xc0+r); |
336 |
|
|
emit_byte(i); |
337 |
|
|
} |
338 |
|
|
} |
339 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_l_ri,(RW4 r, IMM i)) |
340 |
|
|
|
341 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_l_rr,(RW4 d, R1 r)) |
342 |
|
|
{ |
343 |
|
|
emit_byte(0xd3); |
344 |
|
|
emit_byte(0xc0+d); |
345 |
|
|
} |
346 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_l_rr,(RW4 d, R1 r)) |
347 |
|
|
|
348 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_w_rr,(RW2 d, R1 r)) |
349 |
|
|
{ |
350 |
|
|
emit_byte(0x66); |
351 |
|
|
emit_byte(0xd3); |
352 |
|
|
emit_byte(0xc0+d); |
353 |
|
|
} |
354 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_w_rr,(RW2 d, R1 r)) |
355 |
|
|
|
356 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_b_rr,(RW1 d, R1 r)) |
357 |
|
|
{ |
358 |
|
|
emit_byte(0xd2); |
359 |
|
|
emit_byte(0xc0+d); |
360 |
|
|
} |
361 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_b_rr,(RW1 d, R1 r)) |
362 |
|
|
|
363 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_l_rr,(RW4 d, R1 r)) |
364 |
|
|
{ |
365 |
|
|
emit_byte(0xd3); |
366 |
|
|
emit_byte(0xe0+d); |
367 |
|
|
} |
368 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_l_rr,(RW4 d, R1 r)) |
369 |
|
|
|
370 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_w_rr,(RW2 d, R1 r)) |
371 |
|
|
{ |
372 |
|
|
emit_byte(0x66); |
373 |
|
|
emit_byte(0xd3); |
374 |
|
|
emit_byte(0xe0+d); |
375 |
|
|
} |
376 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_w_rr,(RW2 d, R1 r)) |
377 |
|
|
|
378 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_b_rr,(RW1 d, R1 r)) |
379 |
|
|
{ |
380 |
|
|
emit_byte(0xd2); |
381 |
|
|
emit_byte(0xe0+d); |
382 |
|
|
} |
383 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_b_rr,(RW1 d, R1 r)) |
384 |
|
|
|
385 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_b_ri,(RW1 r, IMM i)) |
386 |
|
|
{ |
387 |
|
|
if (optimize_shift_once && (i == 1)) { |
388 |
|
|
emit_byte(0xd0); |
389 |
|
|
emit_byte(0xc8+r); |
390 |
|
|
} |
391 |
|
|
else { |
392 |
|
|
emit_byte(0xc0); |
393 |
|
|
emit_byte(0xc8+r); |
394 |
|
|
emit_byte(i); |
395 |
|
|
} |
396 |
|
|
} |
397 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_b_ri,(RW1 r, IMM i)) |
398 |
|
|
|
399 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_w_ri,(RW2 r, IMM i)) |
400 |
|
|
{ |
401 |
|
|
emit_byte(0x66); |
402 |
|
|
emit_byte(0xc1); |
403 |
|
|
emit_byte(0xc8+r); |
404 |
|
|
emit_byte(i); |
405 |
|
|
} |
406 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_w_ri,(RW2 r, IMM i)) |
407 |
|
|
|
408 |
|
|
// gb-- used for making an fpcr value in compemu_fpp.cpp |
409 |
|
|
LOWFUNC(WRITE,READ,2,raw_or_l_rm,(RW4 d, MEMR s)) |
410 |
|
|
{ |
411 |
|
|
emit_byte(0x0b); |
412 |
|
|
emit_byte(0x05+8*d); |
413 |
|
|
emit_long(s); |
414 |
|
|
} |
415 |
|
|
LENDFUNC(WRITE,READ,2,raw_or_l_rm,(RW4 d, MEMR s)) |
416 |
|
|
|
417 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_l_ri,(RW4 r, IMM i)) |
418 |
|
|
{ |
419 |
|
|
if (optimize_shift_once && (i == 1)) { |
420 |
|
|
emit_byte(0xd1); |
421 |
|
|
emit_byte(0xc8+r); |
422 |
|
|
} |
423 |
|
|
else { |
424 |
|
|
emit_byte(0xc1); |
425 |
|
|
emit_byte(0xc8+r); |
426 |
|
|
emit_byte(i); |
427 |
|
|
} |
428 |
|
|
} |
429 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_l_ri,(RW4 r, IMM i)) |
430 |
|
|
|
431 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_l_rr,(RW4 d, R1 r)) |
432 |
|
|
{ |
433 |
|
|
emit_byte(0xd3); |
434 |
|
|
emit_byte(0xc8+d); |
435 |
|
|
} |
436 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_l_rr,(RW4 d, R1 r)) |
437 |
|
|
|
438 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_w_rr,(RW2 d, R1 r)) |
439 |
|
|
{ |
440 |
|
|
emit_byte(0x66); |
441 |
|
|
emit_byte(0xd3); |
442 |
|
|
emit_byte(0xc8+d); |
443 |
|
|
} |
444 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_w_rr,(RW2 d, R1 r)) |
445 |
|
|
|
446 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_b_rr,(RW1 d, R1 r)) |
447 |
|
|
{ |
448 |
|
|
emit_byte(0xd2); |
449 |
|
|
emit_byte(0xc8+d); |
450 |
|
|
} |
451 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_b_rr,(RW1 d, R1 r)) |
452 |
|
|
|
453 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_l_rr,(RW4 d, R1 r)) |
454 |
|
|
{ |
455 |
|
|
emit_byte(0xd3); |
456 |
|
|
emit_byte(0xe8+d); |
457 |
|
|
} |
458 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_l_rr,(RW4 d, R1 r)) |
459 |
|
|
|
460 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_w_rr,(RW2 d, R1 r)) |
461 |
|
|
{ |
462 |
|
|
emit_byte(0x66); |
463 |
|
|
emit_byte(0xd3); |
464 |
|
|
emit_byte(0xe8+d); |
465 |
|
|
} |
466 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_w_rr,(RW2 d, R1 r)) |
467 |
|
|
|
468 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_b_rr,(RW1 d, R1 r)) |
469 |
|
|
{ |
470 |
|
|
emit_byte(0xd2); |
471 |
|
|
emit_byte(0xe8+d); |
472 |
|
|
} |
473 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_b_rr,(RW1 d, R1 r)) |
474 |
|
|
|
475 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_l_rr,(RW4 d, R1 r)) |
476 |
|
|
{ |
477 |
|
|
emit_byte(0xd3); |
478 |
|
|
emit_byte(0xf8+d); |
479 |
|
|
} |
480 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_l_rr,(RW4 d, R1 r)) |
481 |
|
|
|
482 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_w_rr,(RW2 d, R1 r)) |
483 |
|
|
{ |
484 |
|
|
emit_byte(0x66); |
485 |
|
|
emit_byte(0xd3); |
486 |
|
|
emit_byte(0xf8+d); |
487 |
|
|
} |
488 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_w_rr,(RW2 d, R1 r)) |
489 |
|
|
|
490 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_b_rr,(RW1 d, R1 r)) |
491 |
|
|
{ |
492 |
|
|
emit_byte(0xd2); |
493 |
|
|
emit_byte(0xf8+d); |
494 |
|
|
} |
495 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_b_rr,(RW1 d, R1 r)) |
496 |
|
|
|
497 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_l_ri,(RW4 r, IMM i)) |
498 |
|
|
{ |
499 |
|
|
if (optimize_shift_once && (i == 1)) { |
500 |
|
|
emit_byte(0xd1); |
501 |
|
|
emit_byte(0xe0+r); |
502 |
|
|
} |
503 |
|
|
else { |
504 |
|
|
emit_byte(0xc1); |
505 |
|
|
emit_byte(0xe0+r); |
506 |
|
|
emit_byte(i); |
507 |
|
|
} |
508 |
|
|
} |
509 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_l_ri,(RW4 r, IMM i)) |
510 |
|
|
|
511 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_w_ri,(RW2 r, IMM i)) |
512 |
|
|
{ |
513 |
|
|
emit_byte(0x66); |
514 |
|
|
emit_byte(0xc1); |
515 |
|
|
emit_byte(0xe0+r); |
516 |
|
|
emit_byte(i); |
517 |
|
|
} |
518 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_w_ri,(RW2 r, IMM i)) |
519 |
|
|
|
520 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_b_ri,(RW1 r, IMM i)) |
521 |
|
|
{ |
522 |
|
|
if (optimize_shift_once && (i == 1)) { |
523 |
|
|
emit_byte(0xd0); |
524 |
|
|
emit_byte(0xe0+r); |
525 |
|
|
} |
526 |
|
|
else { |
527 |
|
|
emit_byte(0xc0); |
528 |
|
|
emit_byte(0xe0+r); |
529 |
|
|
emit_byte(i); |
530 |
|
|
} |
531 |
|
|
} |
532 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_b_ri,(RW1 r, IMM i)) |
533 |
|
|
|
534 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_l_ri,(RW4 r, IMM i)) |
535 |
|
|
{ |
536 |
|
|
if (optimize_shift_once && (i == 1)) { |
537 |
|
|
emit_byte(0xd1); |
538 |
|
|
emit_byte(0xe8+r); |
539 |
|
|
} |
540 |
|
|
else { |
541 |
|
|
emit_byte(0xc1); |
542 |
|
|
emit_byte(0xe8+r); |
543 |
|
|
emit_byte(i); |
544 |
|
|
} |
545 |
|
|
} |
546 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_l_ri,(RW4 r, IMM i)) |
547 |
|
|
|
548 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_w_ri,(RW2 r, IMM i)) |
549 |
|
|
{ |
550 |
|
|
emit_byte(0x66); |
551 |
|
|
emit_byte(0xc1); |
552 |
|
|
emit_byte(0xe8+r); |
553 |
|
|
emit_byte(i); |
554 |
|
|
} |
555 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_w_ri,(RW2 r, IMM i)) |
556 |
|
|
|
557 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_b_ri,(RW1 r, IMM i)) |
558 |
|
|
{ |
559 |
|
|
if (optimize_shift_once && (i == 1)) { |
560 |
|
|
emit_byte(0xd0); |
561 |
|
|
emit_byte(0xe8+r); |
562 |
|
|
} |
563 |
|
|
else { |
564 |
|
|
emit_byte(0xc0); |
565 |
|
|
emit_byte(0xe8+r); |
566 |
|
|
emit_byte(i); |
567 |
|
|
} |
568 |
|
|
} |
569 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_b_ri,(RW1 r, IMM i)) |
570 |
|
|
|
571 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_l_ri,(RW4 r, IMM i)) |
572 |
|
|
{ |
573 |
|
|
if (optimize_shift_once && (i == 1)) { |
574 |
|
|
emit_byte(0xd1); |
575 |
|
|
emit_byte(0xf8+r); |
576 |
|
|
} |
577 |
|
|
else { |
578 |
|
|
emit_byte(0xc1); |
579 |
|
|
emit_byte(0xf8+r); |
580 |
|
|
emit_byte(i); |
581 |
|
|
} |
582 |
|
|
} |
583 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_l_ri,(RW4 r, IMM i)) |
584 |
|
|
|
585 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_w_ri,(RW2 r, IMM i)) |
586 |
|
|
{ |
587 |
|
|
emit_byte(0x66); |
588 |
|
|
emit_byte(0xc1); |
589 |
|
|
emit_byte(0xf8+r); |
590 |
|
|
emit_byte(i); |
591 |
|
|
} |
592 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_w_ri,(RW2 r, IMM i)) |
593 |
|
|
|
594 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_b_ri,(RW1 r, IMM i)) |
595 |
|
|
{ |
596 |
|
|
if (optimize_shift_once && (i == 1)) { |
597 |
|
|
emit_byte(0xd0); |
598 |
|
|
emit_byte(0xf8+r); |
599 |
|
|
} |
600 |
|
|
else { |
601 |
|
|
emit_byte(0xc0); |
602 |
|
|
emit_byte(0xf8+r); |
603 |
|
|
emit_byte(i); |
604 |
|
|
} |
605 |
|
|
} |
606 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_b_ri,(RW1 r, IMM i)) |
607 |
|
|
|
608 |
|
|
LOWFUNC(WRITE,NONE,1,raw_sahf,(R2 dummy_ah)) |
609 |
|
|
{ |
610 |
|
|
emit_byte(0x9e); |
611 |
|
|
} |
612 |
|
|
LENDFUNC(WRITE,NONE,1,raw_sahf,(R2 dummy_ah)) |
613 |
|
|
|
614 |
|
|
LOWFUNC(NONE,NONE,1,raw_cpuid,(R4 dummy_eax)) |
615 |
|
|
{ |
616 |
|
|
emit_byte(0x0f); |
617 |
|
|
emit_byte(0xa2); |
618 |
|
|
} |
619 |
|
|
LENDFUNC(NONE,NONE,1,raw_cpuid,(R4 dummy_eax)) |
620 |
|
|
|
621 |
|
|
LOWFUNC(READ,NONE,1,raw_lahf,(W2 dummy_ah)) |
622 |
|
|
{ |
623 |
|
|
emit_byte(0x9f); |
624 |
|
|
} |
625 |
|
|
LENDFUNC(READ,NONE,1,raw_lahf,(W2 dummy_ah)) |
626 |
|
|
|
627 |
|
|
LOWFUNC(READ,NONE,2,raw_setcc,(W1 d, IMM cc)) |
628 |
|
|
{ |
629 |
|
|
emit_byte(0x0f); |
630 |
|
|
emit_byte(0x90+cc); |
631 |
|
|
emit_byte(0xc0+d); |
632 |
|
|
} |
633 |
|
|
LENDFUNC(READ,NONE,2,raw_setcc,(W1 d, IMM cc)) |
634 |
|
|
|
635 |
|
|
LOWFUNC(READ,WRITE,2,raw_setcc_m,(MEMW d, IMM cc)) |
636 |
|
|
{ |
637 |
|
|
emit_byte(0x0f); |
638 |
|
|
emit_byte(0x90+cc); |
639 |
|
|
emit_byte(0x05); |
640 |
|
|
emit_long(d); |
641 |
|
|
} |
642 |
|
|
LENDFUNC(READ,WRITE,2,raw_setcc_m,(MEMW d, IMM cc)) |
643 |
|
|
|
644 |
|
|
LOWFUNC(READ,NONE,3,raw_cmov_l_rr,(RW4 d, R4 s, IMM cc)) |
645 |
|
|
{ |
646 |
|
|
if (have_cmov) { |
647 |
|
|
emit_byte(0x0f); |
648 |
|
|
emit_byte(0x40+cc); |
649 |
|
|
emit_byte(0xc0+8*d+s); |
650 |
|
|
} |
651 |
|
|
else { /* replacement using branch and mov */ |
652 |
|
|
int uncc=(cc^1); |
653 |
|
|
emit_byte(0x70+uncc); |
654 |
|
|
emit_byte(2); /* skip next 2 bytes if not cc=true */ |
655 |
|
|
emit_byte(0x89); |
656 |
|
|
emit_byte(0xc0+8*s+d); |
657 |
|
|
} |
658 |
|
|
} |
659 |
|
|
LENDFUNC(READ,NONE,3,raw_cmov_l_rr,(RW4 d, R4 s, IMM cc)) |
660 |
|
|
|
661 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bsf_l_rr,(W4 d, R4 s)) |
662 |
|
|
{ |
663 |
|
|
emit_byte(0x0f); |
664 |
|
|
emit_byte(0xbc); |
665 |
|
|
emit_byte(0xc0+8*d+s); |
666 |
|
|
} |
667 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bsf_l_rr,(W4 d, R4 s)) |
668 |
|
|
|
669 |
|
|
LOWFUNC(NONE,NONE,2,raw_sign_extend_16_rr,(W4 d, R2 s)) |
670 |
|
|
{ |
671 |
|
|
emit_byte(0x0f); |
672 |
|
|
emit_byte(0xbf); |
673 |
|
|
emit_byte(0xc0+8*d+s); |
674 |
|
|
} |
675 |
|
|
LENDFUNC(NONE,NONE,2,raw_sign_extend_16_rr,(W4 d, R2 s)) |
676 |
|
|
|
677 |
|
|
LOWFUNC(NONE,NONE,2,raw_sign_extend_8_rr,(W4 d, R1 s)) |
678 |
|
|
{ |
679 |
|
|
emit_byte(0x0f); |
680 |
|
|
emit_byte(0xbe); |
681 |
|
|
emit_byte(0xc0+8*d+s); |
682 |
|
|
} |
683 |
|
|
LENDFUNC(NONE,NONE,2,raw_sign_extend_8_rr,(W4 d, R1 s)) |
684 |
|
|
|
685 |
|
|
LOWFUNC(NONE,NONE,2,raw_zero_extend_16_rr,(W4 d, R2 s)) |
686 |
|
|
{ |
687 |
|
|
emit_byte(0x0f); |
688 |
|
|
emit_byte(0xb7); |
689 |
|
|
emit_byte(0xc0+8*d+s); |
690 |
|
|
} |
691 |
|
|
LENDFUNC(NONE,NONE,2,raw_zero_extend_16_rr,(W4 d, R2 s)) |
692 |
|
|
|
693 |
|
|
LOWFUNC(NONE,NONE,2,raw_zero_extend_8_rr,(W4 d, R1 s)) |
694 |
|
|
{ |
695 |
|
|
emit_byte(0x0f); |
696 |
|
|
emit_byte(0xb6); |
697 |
|
|
emit_byte(0xc0+8*d+s); |
698 |
|
|
} |
699 |
|
|
LENDFUNC(NONE,NONE,2,raw_zero_extend_8_rr,(W4 d, R1 s)) |
700 |
|
|
|
701 |
|
|
LOWFUNC(NONE,NONE,2,raw_imul_32_32,(RW4 d, R4 s)) |
702 |
|
|
{ |
703 |
|
|
emit_byte(0x0f); |
704 |
|
|
emit_byte(0xaf); |
705 |
|
|
emit_byte(0xc0+8*d+s); |
706 |
|
|
} |
707 |
|
|
LENDFUNC(NONE,NONE,2,raw_imul_32_32,(RW4 d, R4 s)) |
708 |
|
|
|
709 |
|
|
LOWFUNC(NONE,NONE,2,raw_imul_64_32,(RW4 d, RW4 s)) |
710 |
|
|
{ |
711 |
|
|
if (d!=MUL_NREG1 || s!=MUL_NREG2) |
712 |
|
|
abort(); |
713 |
|
|
emit_byte(0xf7); |
714 |
|
|
emit_byte(0xea); |
715 |
|
|
} |
716 |
|
|
LENDFUNC(NONE,NONE,2,raw_imul_64_32,(RW4 d, RW4 s)) |
717 |
|
|
|
718 |
|
|
LOWFUNC(NONE,NONE,2,raw_mul_64_32,(RW4 d, RW4 s)) |
719 |
|
|
{ |
720 |
|
|
if (d!=MUL_NREG1 || s!=MUL_NREG2) { |
721 |
|
|
printf("Bad register in MUL: d=%d, s=%d\n",d,s); |
722 |
|
|
abort(); |
723 |
|
|
} |
724 |
|
|
emit_byte(0xf7); |
725 |
|
|
emit_byte(0xe2); |
726 |
|
|
} |
727 |
|
|
LENDFUNC(NONE,NONE,2,raw_mul_64_32,(RW4 d, RW4 s)) |
728 |
|
|
|
729 |
|
|
LOWFUNC(NONE,NONE,2,raw_mul_32_32,(RW4 d, R4 s)) |
730 |
|
|
{ |
731 |
|
|
abort(); /* %^$&%^$%#^ x86! */ |
732 |
|
|
emit_byte(0x0f); |
733 |
|
|
emit_byte(0xaf); |
734 |
|
|
emit_byte(0xc0+8*d+s); |
735 |
|
|
} |
736 |
|
|
LENDFUNC(NONE,NONE,2,raw_mul_32_32,(RW4 d, R4 s)) |
737 |
|
|
|
738 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_b_rr,(W1 d, R1 s)) |
739 |
|
|
{ |
740 |
|
|
emit_byte(0x88); |
741 |
|
|
emit_byte(0xc0+8*s+d); |
742 |
|
|
} |
743 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_b_rr,(W1 d, R1 s)) |
744 |
|
|
|
745 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_w_rr,(W2 d, R2 s)) |
746 |
|
|
{ |
747 |
|
|
emit_byte(0x66); |
748 |
|
|
emit_byte(0x89); |
749 |
|
|
emit_byte(0xc0+8*s+d); |
750 |
|
|
} |
751 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_w_rr,(W2 d, R2 s)) |
752 |
|
|
|
753 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_l_rrm_indexed,(W4 d,R4 baser, R4 index, IMM factor)) |
754 |
|
|
{ |
755 |
|
|
int isebp=(baser==5)?0x40:0; |
756 |
|
|
int fi; |
757 |
|
|
|
758 |
|
|
switch(factor) { |
759 |
|
|
case 1: fi=0; break; |
760 |
|
|
case 2: fi=1; break; |
761 |
|
|
case 4: fi=2; break; |
762 |
|
|
case 8: fi=3; break; |
763 |
|
|
default: abort(); |
764 |
|
|
} |
765 |
|
|
|
766 |
|
|
|
767 |
|
|
emit_byte(0x8b); |
768 |
|
|
emit_byte(0x04+8*d+isebp); |
769 |
|
|
emit_byte(baser+8*index+0x40*fi); |
770 |
|
|
if (isebp) |
771 |
|
|
emit_byte(0x00); |
772 |
|
|
} |
773 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_l_rrm_indexed,(W4 d,R4 baser, R4 index, IMM factor)) |
774 |
|
|
|
775 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_w_rrm_indexed,(W2 d, R4 baser, R4 index, IMM factor)) |
776 |
|
|
{ |
777 |
|
|
int fi; |
778 |
|
|
int isebp; |
779 |
|
|
|
780 |
|
|
switch(factor) { |
781 |
|
|
case 1: fi=0; break; |
782 |
|
|
case 2: fi=1; break; |
783 |
|
|
case 4: fi=2; break; |
784 |
|
|
case 8: fi=3; break; |
785 |
|
|
default: abort(); |
786 |
|
|
} |
787 |
|
|
isebp=(baser==5)?0x40:0; |
788 |
|
|
|
789 |
|
|
emit_byte(0x66); |
790 |
|
|
emit_byte(0x8b); |
791 |
|
|
emit_byte(0x04+8*d+isebp); |
792 |
|
|
emit_byte(baser+8*index+0x40*fi); |
793 |
|
|
if (isebp) |
794 |
|
|
emit_byte(0x00); |
795 |
|
|
} |
796 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_w_rrm_indexed,(W2 d, R4 baser, R4 index, IMM factor)) |
797 |
|
|
|
798 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_b_rrm_indexed,(W1 d, R4 baser, R4 index, IMM factor)) |
799 |
|
|
{ |
800 |
|
|
int fi; |
801 |
|
|
int isebp; |
802 |
|
|
|
803 |
|
|
switch(factor) { |
804 |
|
|
case 1: fi=0; break; |
805 |
|
|
case 2: fi=1; break; |
806 |
|
|
case 4: fi=2; break; |
807 |
|
|
case 8: fi=3; break; |
808 |
|
|
default: abort(); |
809 |
|
|
} |
810 |
|
|
isebp=(baser==5)?0x40:0; |
811 |
|
|
|
812 |
|
|
emit_byte(0x8a); |
813 |
|
|
emit_byte(0x04+8*d+isebp); |
814 |
|
|
emit_byte(baser+8*index+0x40*fi); |
815 |
|
|
if (isebp) |
816 |
|
|
emit_byte(0x00); |
817 |
|
|
} |
818 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_b_rrm_indexed,(W1 d, R4 baser, R4 index, IMM factor)) |
819 |
|
|
|
820 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_l_mrr_indexed,(R4 baser, R4 index, IMM factor, R4 s)) |
821 |
|
|
{ |
822 |
|
|
int fi; |
823 |
|
|
int isebp; |
824 |
|
|
|
825 |
|
|
switch(factor) { |
826 |
|
|
case 1: fi=0; break; |
827 |
|
|
case 2: fi=1; break; |
828 |
|
|
case 4: fi=2; break; |
829 |
|
|
case 8: fi=3; break; |
830 |
|
|
default: abort(); |
831 |
|
|
} |
832 |
|
|
|
833 |
|
|
|
834 |
|
|
isebp=(baser==5)?0x40:0; |
835 |
|
|
|
836 |
|
|
emit_byte(0x89); |
837 |
|
|
emit_byte(0x04+8*s+isebp); |
838 |
|
|
emit_byte(baser+8*index+0x40*fi); |
839 |
|
|
if (isebp) |
840 |
|
|
emit_byte(0x00); |
841 |
|
|
} |
842 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_l_mrr_indexed,(R4 baser, R4 index, IMM factor, R4 s)) |
843 |
|
|
|
844 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_w_mrr_indexed,(R4 baser, R4 index, IMM factor, R2 s)) |
845 |
|
|
{ |
846 |
|
|
int fi; |
847 |
|
|
int isebp; |
848 |
|
|
|
849 |
|
|
switch(factor) { |
850 |
|
|
case 1: fi=0; break; |
851 |
|
|
case 2: fi=1; break; |
852 |
|
|
case 4: fi=2; break; |
853 |
|
|
case 8: fi=3; break; |
854 |
|
|
default: abort(); |
855 |
|
|
} |
856 |
|
|
isebp=(baser==5)?0x40:0; |
857 |
|
|
|
858 |
|
|
emit_byte(0x66); |
859 |
|
|
emit_byte(0x89); |
860 |
|
|
emit_byte(0x04+8*s+isebp); |
861 |
|
|
emit_byte(baser+8*index+0x40*fi); |
862 |
|
|
if (isebp) |
863 |
|
|
emit_byte(0x00); |
864 |
|
|
} |
865 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_w_mrr_indexed,(R4 baser, R4 index, IMM factor, R2 s)) |
866 |
|
|
|
867 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_b_mrr_indexed,(R4 baser, R4 index, IMM factor, R1 s)) |
868 |
|
|
{ |
869 |
|
|
int fi; |
870 |
|
|
int isebp; |
871 |
|
|
|
872 |
|
|
switch(factor) { |
873 |
|
|
case 1: fi=0; break; |
874 |
|
|
case 2: fi=1; break; |
875 |
|
|
case 4: fi=2; break; |
876 |
|
|
case 8: fi=3; break; |
877 |
|
|
default: abort(); |
878 |
|
|
} |
879 |
|
|
isebp=(baser==5)?0x40:0; |
880 |
|
|
|
881 |
|
|
emit_byte(0x88); |
882 |
|
|
emit_byte(0x04+8*s+isebp); |
883 |
|
|
emit_byte(baser+8*index+0x40*fi); |
884 |
|
|
if (isebp) |
885 |
|
|
emit_byte(0x00); |
886 |
|
|
} |
887 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_b_mrr_indexed,(R4 baser, R4 index, IMM factor, R1 s)) |
888 |
|
|
|
889 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_l_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R4 s)) |
890 |
|
|
{ |
891 |
|
|
int fi; |
892 |
|
|
|
893 |
|
|
switch(factor) { |
894 |
|
|
case 1: fi=0; break; |
895 |
|
|
case 2: fi=1; break; |
896 |
|
|
case 4: fi=2; break; |
897 |
|
|
case 8: fi=3; break; |
898 |
|
|
default: abort(); |
899 |
|
|
} |
900 |
|
|
|
901 |
|
|
emit_byte(0x89); |
902 |
|
|
emit_byte(0x84+8*s); |
903 |
|
|
emit_byte(baser+8*index+0x40*fi); |
904 |
|
|
emit_long(base); |
905 |
|
|
} |
906 |
|
|
LENDFUNC(NONE,WRITE,5,raw_mov_l_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R4 s)) |
907 |
|
|
|
908 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_w_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R2 s)) |
909 |
|
|
{ |
910 |
|
|
int fi; |
911 |
|
|
|
912 |
|
|
switch(factor) { |
913 |
|
|
case 1: fi=0; break; |
914 |
|
|
case 2: fi=1; break; |
915 |
|
|
case 4: fi=2; break; |
916 |
|
|
case 8: fi=3; break; |
917 |
|
|
default: abort(); |
918 |
|
|
} |
919 |
|
|
|
920 |
|
|
emit_byte(0x66); |
921 |
|
|
emit_byte(0x89); |
922 |
|
|
emit_byte(0x84+8*s); |
923 |
|
|
emit_byte(baser+8*index+0x40*fi); |
924 |
|
|
emit_long(base); |
925 |
|
|
} |
926 |
|
|
LENDFUNC(NONE,WRITE,5,raw_mov_w_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R2 s)) |
927 |
|
|
|
928 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_b_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R1 s)) |
929 |
|
|
{ |
930 |
|
|
int fi; |
931 |
|
|
|
932 |
|
|
switch(factor) { |
933 |
|
|
case 1: fi=0; break; |
934 |
|
|
case 2: fi=1; break; |
935 |
|
|
case 4: fi=2; break; |
936 |
|
|
case 8: fi=3; break; |
937 |
|
|
default: abort(); |
938 |
|
|
} |
939 |
|
|
|
940 |
|
|
emit_byte(0x88); |
941 |
|
|
emit_byte(0x84+8*s); |
942 |
|
|
emit_byte(baser+8*index+0x40*fi); |
943 |
|
|
emit_long(base); |
944 |
|
|
} |
945 |
|
|
LENDFUNC(NONE,WRITE,5,raw_mov_b_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R1 s)) |
946 |
|
|
|
947 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_l_brrm_indexed,(W4 d, IMM base, R4 baser, R4 index, IMM factor)) |
948 |
|
|
{ |
949 |
|
|
int fi; |
950 |
|
|
|
951 |
|
|
switch(factor) { |
952 |
|
|
case 1: fi=0; break; |
953 |
|
|
case 2: fi=1; break; |
954 |
|
|
case 4: fi=2; break; |
955 |
|
|
case 8: fi=3; break; |
956 |
|
|
default: abort(); |
957 |
|
|
} |
958 |
|
|
|
959 |
|
|
emit_byte(0x8b); |
960 |
|
|
emit_byte(0x84+8*d); |
961 |
|
|
emit_byte(baser+8*index+0x40*fi); |
962 |
|
|
emit_long(base); |
963 |
|
|
} |
964 |
|
|
LENDFUNC(NONE,READ,5,raw_mov_l_brrm_indexed,(W4 d, IMM base, R4 baser, R4 index, IMM factor)) |
965 |
|
|
|
966 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_w_brrm_indexed,(W2 d, IMM base, R4 baser, R4 index, IMM factor)) |
967 |
|
|
{ |
968 |
|
|
int fi; |
969 |
|
|
|
970 |
|
|
switch(factor) { |
971 |
|
|
case 1: fi=0; break; |
972 |
|
|
case 2: fi=1; break; |
973 |
|
|
case 4: fi=2; break; |
974 |
|
|
case 8: fi=3; break; |
975 |
|
|
default: abort(); |
976 |
|
|
} |
977 |
|
|
|
978 |
|
|
emit_byte(0x66); |
979 |
|
|
emit_byte(0x8b); |
980 |
|
|
emit_byte(0x84+8*d); |
981 |
|
|
emit_byte(baser+8*index+0x40*fi); |
982 |
|
|
emit_long(base); |
983 |
|
|
} |
984 |
|
|
LENDFUNC(NONE,READ,5,raw_mov_w_brrm_indexed,(W2 d, IMM base, R4 baser, R4 index, IMM factor)) |
985 |
|
|
|
986 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_b_brrm_indexed,(W1 d, IMM base, R4 baser, R4 index, IMM factor)) |
987 |
|
|
{ |
988 |
|
|
int fi; |
989 |
|
|
|
990 |
|
|
switch(factor) { |
991 |
|
|
case 1: fi=0; break; |
992 |
|
|
case 2: fi=1; break; |
993 |
|
|
case 4: fi=2; break; |
994 |
|
|
case 8: fi=3; break; |
995 |
|
|
default: abort(); |
996 |
|
|
} |
997 |
|
|
|
998 |
|
|
emit_byte(0x8a); |
999 |
|
|
emit_byte(0x84+8*d); |
1000 |
|
|
emit_byte(baser+8*index+0x40*fi); |
1001 |
|
|
emit_long(base); |
1002 |
|
|
} |
1003 |
|
|
LENDFUNC(NONE,READ,5,raw_mov_b_brrm_indexed,(W1 d, IMM base, R4 baser, R4 index, IMM factor)) |
1004 |
|
|
|
1005 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor)) |
1006 |
|
|
{ |
1007 |
|
|
int fi; |
1008 |
|
|
switch(factor) { |
1009 |
|
|
case 1: fi=0; break; |
1010 |
|
|
case 2: fi=1; break; |
1011 |
|
|
case 4: fi=2; break; |
1012 |
|
|
case 8: fi=3; break; |
1013 |
|
|
default: |
1014 |
|
|
fprintf(stderr,"Bad factor %d in mov_l_rm_indexed!\n",factor); |
1015 |
|
|
abort(); |
1016 |
|
|
} |
1017 |
|
|
emit_byte(0x8b); |
1018 |
|
|
emit_byte(0x04+8*d); |
1019 |
|
|
emit_byte(0x05+8*index+64*fi); |
1020 |
|
|
emit_long(base); |
1021 |
|
|
} |
1022 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor)) |
1023 |
|
|
|
1024 |
|
|
LOWFUNC(NONE,READ,5,raw_cmov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor, IMM cond)) |
1025 |
|
|
{ |
1026 |
|
|
int fi; |
1027 |
|
|
switch(factor) { |
1028 |
|
|
case 1: fi=0; break; |
1029 |
|
|
case 2: fi=1; break; |
1030 |
|
|
case 4: fi=2; break; |
1031 |
|
|
case 8: fi=3; break; |
1032 |
|
|
default: |
1033 |
|
|
fprintf(stderr,"Bad factor %d in mov_l_rm_indexed!\n",factor); |
1034 |
|
|
abort(); |
1035 |
|
|
} |
1036 |
|
|
if (have_cmov) { |
1037 |
|
|
emit_byte(0x0f); |
1038 |
|
|
emit_byte(0x40+cond); |
1039 |
|
|
emit_byte(0x04+8*d); |
1040 |
|
|
emit_byte(0x05+8*index+64*fi); |
1041 |
|
|
emit_long(base); |
1042 |
|
|
} |
1043 |
|
|
else { /* replacement using branch and mov */ |
1044 |
|
|
int uncc=(cond^1); |
1045 |
|
|
emit_byte(0x70+uncc); |
1046 |
|
|
emit_byte(7); /* skip next 7 bytes if not cc=true */ |
1047 |
|
|
emit_byte(0x8b); |
1048 |
|
|
emit_byte(0x04+8*d); |
1049 |
|
|
emit_byte(0x05+8*index+64*fi); |
1050 |
|
|
emit_long(base); |
1051 |
|
|
} |
1052 |
|
|
} |
1053 |
|
|
LENDFUNC(NONE,READ,5,raw_cmov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor, IMM cond)) |
1054 |
|
|
|
1055 |
|
|
LOWFUNC(NONE,READ,3,raw_cmov_l_rm,(W4 d, IMM mem, IMM cond)) |
1056 |
|
|
{ |
1057 |
|
|
if (have_cmov) { |
1058 |
|
|
emit_byte(0x0f); |
1059 |
|
|
emit_byte(0x40+cond); |
1060 |
|
|
emit_byte(0x05+8*d); |
1061 |
|
|
emit_long(mem); |
1062 |
|
|
} |
1063 |
|
|
else { /* replacement using branch and mov */ |
1064 |
|
|
int uncc=(cond^1); |
1065 |
|
|
emit_byte(0x70+uncc); |
1066 |
|
|
emit_byte(6); /* skip next 6 bytes if not cc=true */ |
1067 |
|
|
emit_byte(0x8b); |
1068 |
|
|
emit_byte(0x05+8*d); |
1069 |
|
|
emit_long(mem); |
1070 |
|
|
} |
1071 |
|
|
} |
1072 |
|
|
LENDFUNC(NONE,READ,3,raw_cmov_l_rm,(W4 d, IMM mem, IMM cond)) |
1073 |
|
|
|
1074 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
1075 |
|
|
{ |
1076 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1077 |
gbeauche |
1.1 |
emit_byte(0x8b); |
1078 |
|
|
emit_byte(0x40+8*d+s); |
1079 |
|
|
emit_byte(offset); |
1080 |
|
|
} |
1081 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
1082 |
|
|
|
1083 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
1084 |
|
|
{ |
1085 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1086 |
gbeauche |
1.1 |
emit_byte(0x66); |
1087 |
|
|
emit_byte(0x8b); |
1088 |
|
|
emit_byte(0x40+8*d+s); |
1089 |
|
|
emit_byte(offset); |
1090 |
|
|
} |
1091 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
1092 |
|
|
|
1093 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
1094 |
|
|
{ |
1095 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1096 |
gbeauche |
1.1 |
emit_byte(0x8a); |
1097 |
|
|
emit_byte(0x40+8*d+s); |
1098 |
|
|
emit_byte(offset); |
1099 |
|
|
} |
1100 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
1101 |
|
|
|
1102 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_l_brR,(W4 d, R4 s, IMM offset)) |
1103 |
|
|
{ |
1104 |
|
|
emit_byte(0x8b); |
1105 |
|
|
emit_byte(0x80+8*d+s); |
1106 |
|
|
emit_long(offset); |
1107 |
|
|
} |
1108 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_l_brR,(W4 d, R4 s, IMM offset)) |
1109 |
|
|
|
1110 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_w_brR,(W2 d, R4 s, IMM offset)) |
1111 |
|
|
{ |
1112 |
|
|
emit_byte(0x66); |
1113 |
|
|
emit_byte(0x8b); |
1114 |
|
|
emit_byte(0x80+8*d+s); |
1115 |
|
|
emit_long(offset); |
1116 |
|
|
} |
1117 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_w_brR,(W2 d, R4 s, IMM offset)) |
1118 |
|
|
|
1119 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_b_brR,(W1 d, R4 s, IMM offset)) |
1120 |
|
|
{ |
1121 |
|
|
emit_byte(0x8a); |
1122 |
|
|
emit_byte(0x80+8*d+s); |
1123 |
|
|
emit_long(offset); |
1124 |
|
|
} |
1125 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_b_brR,(W1 d, R4 s, IMM offset)) |
1126 |
|
|
|
1127 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
1128 |
|
|
{ |
1129 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1130 |
gbeauche |
1.1 |
emit_byte(0xc7); |
1131 |
|
|
emit_byte(0x40+d); |
1132 |
|
|
emit_byte(offset); |
1133 |
|
|
emit_long(i); |
1134 |
|
|
} |
1135 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
1136 |
|
|
|
1137 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
1138 |
|
|
{ |
1139 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1140 |
gbeauche |
1.1 |
emit_byte(0x66); |
1141 |
|
|
emit_byte(0xc7); |
1142 |
|
|
emit_byte(0x40+d); |
1143 |
|
|
emit_byte(offset); |
1144 |
|
|
emit_word(i); |
1145 |
|
|
} |
1146 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
1147 |
|
|
|
1148 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
1149 |
|
|
{ |
1150 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1151 |
gbeauche |
1.1 |
emit_byte(0xc6); |
1152 |
|
|
emit_byte(0x40+d); |
1153 |
|
|
emit_byte(offset); |
1154 |
|
|
emit_byte(i); |
1155 |
|
|
} |
1156 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
1157 |
|
|
|
1158 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
1159 |
|
|
{ |
1160 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1161 |
gbeauche |
1.1 |
emit_byte(0x89); |
1162 |
|
|
emit_byte(0x40+8*s+d); |
1163 |
|
|
emit_byte(offset); |
1164 |
|
|
} |
1165 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
1166 |
|
|
|
1167 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
1168 |
|
|
{ |
1169 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1170 |
gbeauche |
1.1 |
emit_byte(0x66); |
1171 |
|
|
emit_byte(0x89); |
1172 |
|
|
emit_byte(0x40+8*s+d); |
1173 |
|
|
emit_byte(offset); |
1174 |
|
|
} |
1175 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
1176 |
|
|
|
1177 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
1178 |
|
|
{ |
1179 |
gbeauche |
1.9 |
Dif(!isbyte(offset)) abort(); |
1180 |
gbeauche |
1.1 |
emit_byte(0x88); |
1181 |
|
|
emit_byte(0x40+8*s+d); |
1182 |
|
|
emit_byte(offset); |
1183 |
|
|
} |
1184 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
1185 |
|
|
|
1186 |
|
|
LOWFUNC(NONE,NONE,3,raw_lea_l_brr,(W4 d, R4 s, IMM offset)) |
1187 |
|
|
{ |
1188 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1189 |
|
|
emit_byte(0x8d); |
1190 |
|
|
emit_byte(0x40+8*d+s); |
1191 |
|
|
emit_byte(offset); |
1192 |
|
|
} |
1193 |
|
|
else { |
1194 |
|
|
emit_byte(0x8d); |
1195 |
|
|
emit_byte(0x80+8*d+s); |
1196 |
|
|
emit_long(offset); |
1197 |
|
|
} |
1198 |
|
|
} |
1199 |
|
|
LENDFUNC(NONE,NONE,3,raw_lea_l_brr,(W4 d, R4 s, IMM offset)) |
1200 |
|
|
|
1201 |
|
|
LOWFUNC(NONE,NONE,5,raw_lea_l_brr_indexed,(W4 d, R4 s, R4 index, IMM factor, IMM offset)) |
1202 |
|
|
{ |
1203 |
|
|
int fi; |
1204 |
|
|
|
1205 |
|
|
switch(factor) { |
1206 |
|
|
case 1: fi=0; break; |
1207 |
|
|
case 2: fi=1; break; |
1208 |
|
|
case 4: fi=2; break; |
1209 |
|
|
case 8: fi=3; break; |
1210 |
|
|
default: abort(); |
1211 |
|
|
} |
1212 |
|
|
|
1213 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1214 |
|
|
emit_byte(0x8d); |
1215 |
|
|
emit_byte(0x44+8*d); |
1216 |
|
|
emit_byte(0x40*fi+8*index+s); |
1217 |
|
|
emit_byte(offset); |
1218 |
|
|
} |
1219 |
|
|
else { |
1220 |
|
|
emit_byte(0x8d); |
1221 |
|
|
emit_byte(0x84+8*d); |
1222 |
|
|
emit_byte(0x40*fi+8*index+s); |
1223 |
|
|
emit_long(offset); |
1224 |
|
|
} |
1225 |
|
|
} |
1226 |
|
|
LENDFUNC(NONE,NONE,5,raw_lea_l_brr_indexed,(W4 d, R4 s, R4 index, IMM factor, IMM offset)) |
1227 |
|
|
|
1228 |
|
|
LOWFUNC(NONE,NONE,4,raw_lea_l_rr_indexed,(W4 d, R4 s, R4 index, IMM factor)) |
1229 |
|
|
{ |
1230 |
|
|
int isebp=(s==5)?0x40:0; |
1231 |
|
|
int fi; |
1232 |
|
|
|
1233 |
|
|
switch(factor) { |
1234 |
|
|
case 1: fi=0; break; |
1235 |
|
|
case 2: fi=1; break; |
1236 |
|
|
case 4: fi=2; break; |
1237 |
|
|
case 8: fi=3; break; |
1238 |
|
|
default: abort(); |
1239 |
|
|
} |
1240 |
|
|
|
1241 |
|
|
emit_byte(0x8d); |
1242 |
|
|
emit_byte(0x04+8*d+isebp); |
1243 |
|
|
emit_byte(0x40*fi+8*index+s); |
1244 |
|
|
if (isebp) |
1245 |
|
|
emit_byte(0); |
1246 |
|
|
} |
1247 |
|
|
LENDFUNC(NONE,NONE,4,raw_lea_l_rr_indexed,(W4 d, R4 s, R4 index, IMM factor)) |
1248 |
|
|
|
1249 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_bRr,(R4 d, R4 s, IMM offset)) |
1250 |
|
|
{ |
1251 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1252 |
|
|
emit_byte(0x89); |
1253 |
|
|
emit_byte(0x40+8*s+d); |
1254 |
|
|
emit_byte(offset); |
1255 |
|
|
} |
1256 |
|
|
else { |
1257 |
|
|
emit_byte(0x89); |
1258 |
|
|
emit_byte(0x80+8*s+d); |
1259 |
|
|
emit_long(offset); |
1260 |
|
|
} |
1261 |
|
|
} |
1262 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_bRr,(R4 d, R4 s, IMM offset)) |
1263 |
|
|
|
1264 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_bRr,(R4 d, R2 s, IMM offset)) |
1265 |
|
|
{ |
1266 |
|
|
emit_byte(0x66); |
1267 |
|
|
emit_byte(0x89); |
1268 |
|
|
emit_byte(0x80+8*s+d); |
1269 |
|
|
emit_long(offset); |
1270 |
|
|
} |
1271 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_bRr,(R4 d, R2 s, IMM offset)) |
1272 |
|
|
|
1273 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_bRr,(R4 d, R1 s, IMM offset)) |
1274 |
|
|
{ |
1275 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1276 |
|
|
emit_byte(0x88); |
1277 |
|
|
emit_byte(0x40+8*s+d); |
1278 |
|
|
emit_byte(offset); |
1279 |
|
|
} |
1280 |
|
|
else { |
1281 |
|
|
emit_byte(0x88); |
1282 |
|
|
emit_byte(0x80+8*s+d); |
1283 |
|
|
emit_long(offset); |
1284 |
|
|
} |
1285 |
|
|
} |
1286 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_bRr,(R4 d, R1 s, IMM offset)) |
1287 |
|
|
|
1288 |
|
|
LOWFUNC(NONE,NONE,1,raw_bswap_32,(RW4 r)) |
1289 |
|
|
{ |
1290 |
|
|
emit_byte(0x0f); |
1291 |
|
|
emit_byte(0xc8+r); |
1292 |
|
|
} |
1293 |
|
|
LENDFUNC(NONE,NONE,1,raw_bswap_32,(RW4 r)) |
1294 |
|
|
|
1295 |
|
|
LOWFUNC(WRITE,NONE,1,raw_bswap_16,(RW2 r)) |
1296 |
|
|
{ |
1297 |
|
|
emit_byte(0x66); |
1298 |
|
|
emit_byte(0xc1); |
1299 |
|
|
emit_byte(0xc0+r); |
1300 |
|
|
emit_byte(0x08); |
1301 |
|
|
} |
1302 |
|
|
LENDFUNC(WRITE,NONE,1,raw_bswap_16,(RW2 r)) |
1303 |
|
|
|
1304 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_l_rr,(W4 d, R4 s)) |
1305 |
|
|
{ |
1306 |
|
|
emit_byte(0x89); |
1307 |
|
|
emit_byte(0xc0+8*s+d); |
1308 |
|
|
} |
1309 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_l_rr,(W4 d, R4 s)) |
1310 |
|
|
|
1311 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_l_mr,(IMM d, R4 s)) |
1312 |
|
|
{ |
1313 |
|
|
emit_byte(0x89); |
1314 |
|
|
emit_byte(0x05+8*s); |
1315 |
|
|
emit_long(d); |
1316 |
|
|
} |
1317 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_l_mr,(IMM d, R4 s)) |
1318 |
|
|
|
1319 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_w_mr,(IMM d, R2 s)) |
1320 |
|
|
{ |
1321 |
|
|
emit_byte(0x66); |
1322 |
|
|
emit_byte(0x89); |
1323 |
|
|
emit_byte(0x05+8*s); |
1324 |
|
|
emit_long(d); |
1325 |
|
|
} |
1326 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_w_mr,(IMM d, R2 s)) |
1327 |
|
|
|
1328 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_w_rm,(W2 d, IMM s)) |
1329 |
|
|
{ |
1330 |
|
|
emit_byte(0x66); |
1331 |
|
|
emit_byte(0x8b); |
1332 |
|
|
emit_byte(0x05+8*d); |
1333 |
|
|
emit_long(s); |
1334 |
|
|
} |
1335 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_w_rm,(W2 d, IMM s)) |
1336 |
|
|
|
1337 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_b_mr,(IMM d, R1 s)) |
1338 |
|
|
{ |
1339 |
|
|
emit_byte(0x88); |
1340 |
|
|
emit_byte(0x05+8*s); |
1341 |
|
|
emit_long(d); |
1342 |
|
|
} |
1343 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_b_mr,(IMM d, R1 s)) |
1344 |
|
|
|
1345 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_b_rm,(W1 d, IMM s)) |
1346 |
|
|
{ |
1347 |
|
|
emit_byte(0x8a); |
1348 |
|
|
emit_byte(0x05+8*d); |
1349 |
|
|
emit_long(s); |
1350 |
|
|
} |
1351 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_b_rm,(W1 d, IMM s)) |
1352 |
|
|
|
1353 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_l_ri,(W4 d, IMM s)) |
1354 |
|
|
{ |
1355 |
|
|
emit_byte(0xb8+d); |
1356 |
|
|
emit_long(s); |
1357 |
|
|
} |
1358 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_l_ri,(W4 d, IMM s)) |
1359 |
|
|
|
1360 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_w_ri,(W2 d, IMM s)) |
1361 |
|
|
{ |
1362 |
|
|
emit_byte(0x66); |
1363 |
|
|
emit_byte(0xb8+d); |
1364 |
|
|
emit_word(s); |
1365 |
|
|
} |
1366 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_w_ri,(W2 d, IMM s)) |
1367 |
|
|
|
1368 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_b_ri,(W1 d, IMM s)) |
1369 |
|
|
{ |
1370 |
|
|
emit_byte(0xb0+d); |
1371 |
|
|
emit_byte(s); |
1372 |
|
|
} |
1373 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_b_ri,(W1 d, IMM s)) |
1374 |
|
|
|
1375 |
|
|
LOWFUNC(RMW,RMW,2,raw_adc_l_mi,(MEMRW d, IMM s)) |
1376 |
|
|
{ |
1377 |
|
|
emit_byte(0x81); |
1378 |
|
|
emit_byte(0x15); |
1379 |
|
|
emit_long(d); |
1380 |
|
|
emit_long(s); |
1381 |
|
|
} |
1382 |
|
|
LENDFUNC(RMW,RMW,2,raw_adc_l_mi,(MEMRW d, IMM s)) |
1383 |
|
|
|
1384 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_l_mi,(IMM d, IMM s)) |
1385 |
|
|
{ |
1386 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1387 |
|
|
emit_byte(0x83); |
1388 |
|
|
emit_byte(0x05); |
1389 |
|
|
emit_long(d); |
1390 |
|
|
emit_byte(s); |
1391 |
|
|
} |
1392 |
|
|
else { |
1393 |
|
|
emit_byte(0x81); |
1394 |
|
|
emit_byte(0x05); |
1395 |
|
|
emit_long(d); |
1396 |
|
|
emit_long(s); |
1397 |
|
|
} |
1398 |
|
|
} |
1399 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_l_mi,(IMM d, IMM s)) |
1400 |
|
|
|
1401 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_w_mi,(IMM d, IMM s)) |
1402 |
|
|
{ |
1403 |
|
|
emit_byte(0x66); |
1404 |
|
|
emit_byte(0x81); |
1405 |
|
|
emit_byte(0x05); |
1406 |
|
|
emit_long(d); |
1407 |
|
|
emit_word(s); |
1408 |
|
|
} |
1409 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_w_mi,(IMM d, IMM s)) |
1410 |
|
|
|
1411 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_b_mi,(IMM d, IMM s)) |
1412 |
|
|
{ |
1413 |
|
|
emit_byte(0x80); |
1414 |
|
|
emit_byte(0x05); |
1415 |
|
|
emit_long(d); |
1416 |
|
|
emit_byte(s); |
1417 |
|
|
} |
1418 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_b_mi,(IMM d, IMM s)) |
1419 |
|
|
|
1420 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_l_ri,(R4 d, IMM i)) |
1421 |
|
|
{ |
1422 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1423 |
|
|
emit_byte(0xa9); |
1424 |
|
|
else { |
1425 |
gbeauche |
1.1 |
emit_byte(0xf7); |
1426 |
|
|
emit_byte(0xc0+d); |
1427 |
gbeauche |
1.2 |
} |
1428 |
gbeauche |
1.1 |
emit_long(i); |
1429 |
|
|
} |
1430 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_l_ri,(R4 d, IMM i)) |
1431 |
|
|
|
1432 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_l_rr,(R4 d, R4 s)) |
1433 |
|
|
{ |
1434 |
|
|
emit_byte(0x85); |
1435 |
|
|
emit_byte(0xc0+8*s+d); |
1436 |
|
|
} |
1437 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_l_rr,(R4 d, R4 s)) |
1438 |
|
|
|
1439 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_w_rr,(R2 d, R2 s)) |
1440 |
|
|
{ |
1441 |
|
|
emit_byte(0x66); |
1442 |
|
|
emit_byte(0x85); |
1443 |
|
|
emit_byte(0xc0+8*s+d); |
1444 |
|
|
} |
1445 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_w_rr,(R2 d, R2 s)) |
1446 |
|
|
|
1447 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_b_rr,(R1 d, R1 s)) |
1448 |
|
|
{ |
1449 |
|
|
emit_byte(0x84); |
1450 |
|
|
emit_byte(0xc0+8*s+d); |
1451 |
|
|
} |
1452 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_b_rr,(R1 d, R1 s)) |
1453 |
|
|
|
1454 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_l_ri,(RW4 d, IMM i)) |
1455 |
|
|
{ |
1456 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1457 |
gbeauche |
1.2 |
emit_byte(0x83); |
1458 |
|
|
emit_byte(0xe0+d); |
1459 |
|
|
emit_byte(i); |
1460 |
gbeauche |
1.1 |
} |
1461 |
|
|
else { |
1462 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1463 |
|
|
emit_byte(0x25); |
1464 |
|
|
else { |
1465 |
|
|
emit_byte(0x81); |
1466 |
|
|
emit_byte(0xe0+d); |
1467 |
|
|
} |
1468 |
|
|
emit_long(i); |
1469 |
gbeauche |
1.1 |
} |
1470 |
|
|
} |
1471 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_l_ri,(RW4 d, IMM i)) |
1472 |
|
|
|
1473 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_w_ri,(RW2 d, IMM i)) |
1474 |
|
|
{ |
1475 |
gbeauche |
1.2 |
emit_byte(0x66); |
1476 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1477 |
|
|
emit_byte(0x83); |
1478 |
|
|
emit_byte(0xe0+d); |
1479 |
|
|
emit_byte(i); |
1480 |
|
|
} |
1481 |
|
|
else { |
1482 |
|
|
if (optimize_accum && isaccum(d)) |
1483 |
|
|
emit_byte(0x25); |
1484 |
|
|
else { |
1485 |
|
|
emit_byte(0x81); |
1486 |
|
|
emit_byte(0xe0+d); |
1487 |
|
|
} |
1488 |
|
|
emit_word(i); |
1489 |
|
|
} |
1490 |
gbeauche |
1.1 |
} |
1491 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_w_ri,(RW2 d, IMM i)) |
1492 |
|
|
|
1493 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_l,(RW4 d, R4 s)) |
1494 |
|
|
{ |
1495 |
|
|
emit_byte(0x21); |
1496 |
|
|
emit_byte(0xc0+8*s+d); |
1497 |
|
|
} |
1498 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_l,(RW4 d, R4 s)) |
1499 |
|
|
|
1500 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_w,(RW2 d, R2 s)) |
1501 |
|
|
{ |
1502 |
|
|
emit_byte(0x66); |
1503 |
|
|
emit_byte(0x21); |
1504 |
|
|
emit_byte(0xc0+8*s+d); |
1505 |
|
|
} |
1506 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_w,(RW2 d, R2 s)) |
1507 |
|
|
|
1508 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_b,(RW1 d, R1 s)) |
1509 |
|
|
{ |
1510 |
|
|
emit_byte(0x20); |
1511 |
|
|
emit_byte(0xc0+8*s+d); |
1512 |
|
|
} |
1513 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_b,(RW1 d, R1 s)) |
1514 |
|
|
|
1515 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_l_ri,(RW4 d, IMM i)) |
1516 |
|
|
{ |
1517 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1518 |
|
|
emit_byte(0x83); |
1519 |
|
|
emit_byte(0xc8+d); |
1520 |
|
|
emit_byte(i); |
1521 |
|
|
} |
1522 |
|
|
else { |
1523 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1524 |
|
|
emit_byte(0x0d); |
1525 |
|
|
else { |
1526 |
gbeauche |
1.1 |
emit_byte(0x81); |
1527 |
|
|
emit_byte(0xc8+d); |
1528 |
gbeauche |
1.2 |
} |
1529 |
gbeauche |
1.1 |
emit_long(i); |
1530 |
|
|
} |
1531 |
|
|
} |
1532 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_l_ri,(RW4 d, IMM i)) |
1533 |
|
|
|
1534 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_l,(RW4 d, R4 s)) |
1535 |
|
|
{ |
1536 |
|
|
emit_byte(0x09); |
1537 |
|
|
emit_byte(0xc0+8*s+d); |
1538 |
|
|
} |
1539 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_l,(RW4 d, R4 s)) |
1540 |
|
|
|
1541 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_w,(RW2 d, R2 s)) |
1542 |
|
|
{ |
1543 |
|
|
emit_byte(0x66); |
1544 |
|
|
emit_byte(0x09); |
1545 |
|
|
emit_byte(0xc0+8*s+d); |
1546 |
|
|
} |
1547 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_w,(RW2 d, R2 s)) |
1548 |
|
|
|
1549 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_b,(RW1 d, R1 s)) |
1550 |
|
|
{ |
1551 |
|
|
emit_byte(0x08); |
1552 |
|
|
emit_byte(0xc0+8*s+d); |
1553 |
|
|
} |
1554 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_b,(RW1 d, R1 s)) |
1555 |
|
|
|
1556 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_l,(RW4 d, R4 s)) |
1557 |
|
|
{ |
1558 |
|
|
emit_byte(0x11); |
1559 |
|
|
emit_byte(0xc0+8*s+d); |
1560 |
|
|
} |
1561 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_l,(RW4 d, R4 s)) |
1562 |
|
|
|
1563 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_w,(RW2 d, R2 s)) |
1564 |
|
|
{ |
1565 |
|
|
emit_byte(0x66); |
1566 |
|
|
emit_byte(0x11); |
1567 |
|
|
emit_byte(0xc0+8*s+d); |
1568 |
|
|
} |
1569 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_w,(RW2 d, R2 s)) |
1570 |
|
|
|
1571 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_b,(RW1 d, R1 s)) |
1572 |
|
|
{ |
1573 |
|
|
emit_byte(0x10); |
1574 |
|
|
emit_byte(0xc0+8*s+d); |
1575 |
|
|
} |
1576 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_b,(RW1 d, R1 s)) |
1577 |
|
|
|
1578 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_l,(RW4 d, R4 s)) |
1579 |
|
|
{ |
1580 |
|
|
emit_byte(0x01); |
1581 |
|
|
emit_byte(0xc0+8*s+d); |
1582 |
|
|
} |
1583 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_l,(RW4 d, R4 s)) |
1584 |
|
|
|
1585 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_w,(RW2 d, R2 s)) |
1586 |
|
|
{ |
1587 |
|
|
emit_byte(0x66); |
1588 |
|
|
emit_byte(0x01); |
1589 |
|
|
emit_byte(0xc0+8*s+d); |
1590 |
|
|
} |
1591 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_w,(RW2 d, R2 s)) |
1592 |
|
|
|
1593 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_b,(RW1 d, R1 s)) |
1594 |
|
|
{ |
1595 |
|
|
emit_byte(0x00); |
1596 |
|
|
emit_byte(0xc0+8*s+d); |
1597 |
|
|
} |
1598 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_b,(RW1 d, R1 s)) |
1599 |
|
|
|
1600 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_l_ri,(RW4 d, IMM i)) |
1601 |
|
|
{ |
1602 |
|
|
if (isbyte(i)) { |
1603 |
|
|
emit_byte(0x83); |
1604 |
|
|
emit_byte(0xe8+d); |
1605 |
|
|
emit_byte(i); |
1606 |
|
|
} |
1607 |
|
|
else { |
1608 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1609 |
|
|
emit_byte(0x2d); |
1610 |
|
|
else { |
1611 |
gbeauche |
1.1 |
emit_byte(0x81); |
1612 |
|
|
emit_byte(0xe8+d); |
1613 |
gbeauche |
1.2 |
} |
1614 |
gbeauche |
1.1 |
emit_long(i); |
1615 |
|
|
} |
1616 |
|
|
} |
1617 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_l_ri,(RW4 d, IMM i)) |
1618 |
|
|
|
1619 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_b_ri,(RW1 d, IMM i)) |
1620 |
|
|
{ |
1621 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1622 |
|
|
emit_byte(0x2c); |
1623 |
|
|
else { |
1624 |
gbeauche |
1.1 |
emit_byte(0x80); |
1625 |
|
|
emit_byte(0xe8+d); |
1626 |
gbeauche |
1.2 |
} |
1627 |
gbeauche |
1.1 |
emit_byte(i); |
1628 |
|
|
} |
1629 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_b_ri,(RW1 d, IMM i)) |
1630 |
|
|
|
1631 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_l_ri,(RW4 d, IMM i)) |
1632 |
|
|
{ |
1633 |
|
|
if (isbyte(i)) { |
1634 |
|
|
emit_byte(0x83); |
1635 |
|
|
emit_byte(0xc0+d); |
1636 |
|
|
emit_byte(i); |
1637 |
|
|
} |
1638 |
|
|
else { |
1639 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1640 |
|
|
emit_byte(0x05); |
1641 |
|
|
else { |
1642 |
gbeauche |
1.1 |
emit_byte(0x81); |
1643 |
|
|
emit_byte(0xc0+d); |
1644 |
gbeauche |
1.2 |
} |
1645 |
gbeauche |
1.1 |
emit_long(i); |
1646 |
|
|
} |
1647 |
|
|
} |
1648 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_l_ri,(RW4 d, IMM i)) |
1649 |
|
|
|
1650 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_w_ri,(RW2 d, IMM i)) |
1651 |
|
|
{ |
1652 |
gbeauche |
1.2 |
emit_byte(0x66); |
1653 |
gbeauche |
1.1 |
if (isbyte(i)) { |
1654 |
|
|
emit_byte(0x83); |
1655 |
|
|
emit_byte(0xc0+d); |
1656 |
|
|
emit_byte(i); |
1657 |
|
|
} |
1658 |
|
|
else { |
1659 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1660 |
|
|
emit_byte(0x05); |
1661 |
|
|
else { |
1662 |
gbeauche |
1.1 |
emit_byte(0x81); |
1663 |
|
|
emit_byte(0xc0+d); |
1664 |
gbeauche |
1.2 |
} |
1665 |
gbeauche |
1.1 |
emit_word(i); |
1666 |
|
|
} |
1667 |
|
|
} |
1668 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_w_ri,(RW2 d, IMM i)) |
1669 |
|
|
|
1670 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_b_ri,(RW1 d, IMM i)) |
1671 |
|
|
{ |
1672 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1673 |
|
|
emit_byte(0x04); |
1674 |
|
|
else { |
1675 |
|
|
emit_byte(0x80); |
1676 |
|
|
emit_byte(0xc0+d); |
1677 |
|
|
} |
1678 |
gbeauche |
1.1 |
emit_byte(i); |
1679 |
|
|
} |
1680 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_b_ri,(RW1 d, IMM i)) |
1681 |
|
|
|
1682 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_l,(RW4 d, R4 s)) |
1683 |
|
|
{ |
1684 |
|
|
emit_byte(0x19); |
1685 |
|
|
emit_byte(0xc0+8*s+d); |
1686 |
|
|
} |
1687 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_l,(RW4 d, R4 s)) |
1688 |
|
|
|
1689 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_w,(RW2 d, R2 s)) |
1690 |
|
|
{ |
1691 |
|
|
emit_byte(0x66); |
1692 |
|
|
emit_byte(0x19); |
1693 |
|
|
emit_byte(0xc0+8*s+d); |
1694 |
|
|
} |
1695 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_w,(RW2 d, R2 s)) |
1696 |
|
|
|
1697 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_b,(RW1 d, R1 s)) |
1698 |
|
|
{ |
1699 |
|
|
emit_byte(0x18); |
1700 |
|
|
emit_byte(0xc0+8*s+d); |
1701 |
|
|
} |
1702 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_b,(RW1 d, R1 s)) |
1703 |
|
|
|
1704 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_l,(RW4 d, R4 s)) |
1705 |
|
|
{ |
1706 |
|
|
emit_byte(0x29); |
1707 |
|
|
emit_byte(0xc0+8*s+d); |
1708 |
|
|
} |
1709 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_l,(RW4 d, R4 s)) |
1710 |
|
|
|
1711 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_w,(RW2 d, R2 s)) |
1712 |
|
|
{ |
1713 |
|
|
emit_byte(0x66); |
1714 |
|
|
emit_byte(0x29); |
1715 |
|
|
emit_byte(0xc0+8*s+d); |
1716 |
|
|
} |
1717 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_w,(RW2 d, R2 s)) |
1718 |
|
|
|
1719 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_b,(RW1 d, R1 s)) |
1720 |
|
|
{ |
1721 |
|
|
emit_byte(0x28); |
1722 |
|
|
emit_byte(0xc0+8*s+d); |
1723 |
|
|
} |
1724 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_b,(RW1 d, R1 s)) |
1725 |
|
|
|
1726 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_l,(R4 d, R4 s)) |
1727 |
|
|
{ |
1728 |
|
|
emit_byte(0x39); |
1729 |
|
|
emit_byte(0xc0+8*s+d); |
1730 |
|
|
} |
1731 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_l,(R4 d, R4 s)) |
1732 |
|
|
|
1733 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_l_ri,(R4 r, IMM i)) |
1734 |
|
|
{ |
1735 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1736 |
|
|
emit_byte(0x83); |
1737 |
|
|
emit_byte(0xf8+r); |
1738 |
|
|
emit_byte(i); |
1739 |
|
|
} |
1740 |
|
|
else { |
1741 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(r)) |
1742 |
|
|
emit_byte(0x3d); |
1743 |
|
|
else { |
1744 |
gbeauche |
1.1 |
emit_byte(0x81); |
1745 |
|
|
emit_byte(0xf8+r); |
1746 |
gbeauche |
1.2 |
} |
1747 |
gbeauche |
1.1 |
emit_long(i); |
1748 |
|
|
} |
1749 |
|
|
} |
1750 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_l_ri,(R4 r, IMM i)) |
1751 |
|
|
|
1752 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_w,(R2 d, R2 s)) |
1753 |
|
|
{ |
1754 |
|
|
emit_byte(0x66); |
1755 |
|
|
emit_byte(0x39); |
1756 |
|
|
emit_byte(0xc0+8*s+d); |
1757 |
|
|
} |
1758 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_w,(R2 d, R2 s)) |
1759 |
|
|
|
1760 |
gbeauche |
1.5 |
LOWFUNC(WRITE,READ,2,raw_cmp_b_mi,(MEMR d, IMM s)) |
1761 |
|
|
{ |
1762 |
|
|
emit_byte(0x80); |
1763 |
|
|
emit_byte(0x3d); |
1764 |
|
|
emit_long(d); |
1765 |
|
|
emit_byte(s); |
1766 |
|
|
} |
1767 |
|
|
LENDFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1768 |
|
|
|
1769 |
gbeauche |
1.1 |
LOWFUNC(WRITE,NONE,2,raw_cmp_b_ri,(R1 d, IMM i)) |
1770 |
|
|
{ |
1771 |
gbeauche |
1.2 |
if (optimize_accum && isaccum(d)) |
1772 |
|
|
emit_byte(0x3c); |
1773 |
|
|
else { |
1774 |
gbeauche |
1.1 |
emit_byte(0x80); |
1775 |
|
|
emit_byte(0xf8+d); |
1776 |
gbeauche |
1.2 |
} |
1777 |
gbeauche |
1.1 |
emit_byte(i); |
1778 |
|
|
} |
1779 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_b_ri,(R1 d, IMM i)) |
1780 |
|
|
|
1781 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_b,(R1 d, R1 s)) |
1782 |
|
|
{ |
1783 |
|
|
emit_byte(0x38); |
1784 |
|
|
emit_byte(0xc0+8*s+d); |
1785 |
|
|
} |
1786 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_b,(R1 d, R1 s)) |
1787 |
|
|
|
1788 |
|
|
LOWFUNC(WRITE,READ,4,raw_cmp_l_rm_indexed,(R4 d, IMM offset, R4 index, IMM factor)) |
1789 |
|
|
{ |
1790 |
|
|
int fi; |
1791 |
|
|
|
1792 |
|
|
switch(factor) { |
1793 |
|
|
case 1: fi=0; break; |
1794 |
|
|
case 2: fi=1; break; |
1795 |
|
|
case 4: fi=2; break; |
1796 |
|
|
case 8: fi=3; break; |
1797 |
|
|
default: abort(); |
1798 |
|
|
} |
1799 |
|
|
emit_byte(0x39); |
1800 |
|
|
emit_byte(0x04+8*d); |
1801 |
|
|
emit_byte(5+8*index+0x40*fi); |
1802 |
|
|
emit_long(offset); |
1803 |
|
|
} |
1804 |
|
|
LENDFUNC(WRITE,READ,4,raw_cmp_l_rm_indexed,(R4 d, IMM offset, R4 index, IMM factor)) |
1805 |
|
|
|
1806 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_l,(RW4 d, R4 s)) |
1807 |
|
|
{ |
1808 |
|
|
emit_byte(0x31); |
1809 |
|
|
emit_byte(0xc0+8*s+d); |
1810 |
|
|
} |
1811 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_l,(RW4 d, R4 s)) |
1812 |
|
|
|
1813 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_w,(RW2 d, R2 s)) |
1814 |
|
|
{ |
1815 |
|
|
emit_byte(0x66); |
1816 |
|
|
emit_byte(0x31); |
1817 |
|
|
emit_byte(0xc0+8*s+d); |
1818 |
|
|
} |
1819 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_w,(RW2 d, R2 s)) |
1820 |
|
|
|
1821 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_b,(RW1 d, R1 s)) |
1822 |
|
|
{ |
1823 |
|
|
emit_byte(0x30); |
1824 |
|
|
emit_byte(0xc0+8*s+d); |
1825 |
|
|
} |
1826 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_b,(RW1 d, R1 s)) |
1827 |
|
|
|
1828 |
|
|
LOWFUNC(WRITE,RMW,2,raw_sub_l_mi,(MEMRW d, IMM s)) |
1829 |
|
|
{ |
1830 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1831 |
|
|
emit_byte(0x83); |
1832 |
|
|
emit_byte(0x2d); |
1833 |
|
|
emit_long(d); |
1834 |
|
|
emit_byte(s); |
1835 |
|
|
} |
1836 |
|
|
else { |
1837 |
|
|
emit_byte(0x81); |
1838 |
|
|
emit_byte(0x2d); |
1839 |
|
|
emit_long(d); |
1840 |
|
|
emit_long(s); |
1841 |
|
|
} |
1842 |
|
|
} |
1843 |
|
|
LENDFUNC(WRITE,RMW,2,raw_sub_l_mi,(MEMRW d, IMM s)) |
1844 |
|
|
|
1845 |
|
|
LOWFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1846 |
|
|
{ |
1847 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1848 |
|
|
emit_byte(0x83); |
1849 |
|
|
emit_byte(0x3d); |
1850 |
|
|
emit_long(d); |
1851 |
|
|
emit_byte(s); |
1852 |
|
|
} |
1853 |
|
|
else { |
1854 |
|
|
emit_byte(0x81); |
1855 |
|
|
emit_byte(0x3d); |
1856 |
|
|
emit_long(d); |
1857 |
|
|
emit_long(s); |
1858 |
|
|
} |
1859 |
|
|
} |
1860 |
|
|
LENDFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1861 |
|
|
|
1862 |
|
|
LOWFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
1863 |
|
|
{ |
1864 |
|
|
emit_byte(0x87); |
1865 |
|
|
emit_byte(0xc0+8*r1+r2); |
1866 |
|
|
} |
1867 |
|
|
LENDFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
1868 |
|
|
|
1869 |
|
|
/************************************************************************* |
1870 |
|
|
* FIXME: string-related instructions * |
1871 |
|
|
*************************************************************************/ |
1872 |
|
|
|
1873 |
|
|
LOWFUNC(WRITE,NONE,0,raw_cld,(void)) |
1874 |
|
|
{ |
1875 |
|
|
emit_byte(0xfc); |
1876 |
|
|
} |
1877 |
|
|
LENDFUNC(WRITE,NONE,0,raw_cld,(void)) |
1878 |
|
|
|
1879 |
|
|
LOWFUNC(WRITE,NONE,0,raw_std,(void)) |
1880 |
|
|
{ |
1881 |
|
|
emit_byte(0xfd); |
1882 |
|
|
} |
1883 |
|
|
LENDFUNC(WRITE,NONE,0,raw_std,(void)) |
1884 |
|
|
|
1885 |
|
|
LOWFUNC(NONE,RMW,0,raw_movs_b,(void)) |
1886 |
|
|
{ |
1887 |
|
|
emit_byte(0xa4); |
1888 |
|
|
} |
1889 |
|
|
LENDFUNC(NONE,RMW,0,raw_movs_b,(void)) |
1890 |
|
|
|
1891 |
|
|
LOWFUNC(NONE,RMW,0,raw_movs_l,(void)) |
1892 |
|
|
{ |
1893 |
|
|
emit_byte(0xa5); |
1894 |
|
|
} |
1895 |
|
|
LENDFUNC(NONE,RMW,0,raw_movs_l,(void)) |
1896 |
|
|
|
1897 |
|
|
LOWFUNC(NONE,RMW,0,raw_rep,(void)) |
1898 |
|
|
{ |
1899 |
|
|
emit_byte(0xf3); |
1900 |
|
|
} |
1901 |
|
|
LENDFUNC(NONE,RMW,0,raw_rep,(void)) |
1902 |
|
|
|
1903 |
|
|
LOWFUNC(NONE,RMW,0,raw_rep_movsb,(void)) |
1904 |
|
|
{ |
1905 |
|
|
raw_rep(); |
1906 |
|
|
raw_movs_b(); |
1907 |
|
|
} |
1908 |
|
|
LENDFUNC(NONE,RMW,0,raw_rep_movsb,(void)) |
1909 |
|
|
|
1910 |
|
|
LOWFUNC(NONE,RMW,0,raw_rep_movsl,(void)) |
1911 |
|
|
{ |
1912 |
|
|
raw_rep(); |
1913 |
|
|
raw_movs_l(); |
1914 |
|
|
} |
1915 |
|
|
LENDFUNC(NONE,RMW,0,raw_rep_movsl,(void)) |
1916 |
|
|
|
1917 |
|
|
/************************************************************************* |
1918 |
|
|
* FIXME: mem access modes probably wrong * |
1919 |
|
|
*************************************************************************/ |
1920 |
|
|
|
1921 |
|
|
LOWFUNC(READ,WRITE,0,raw_pushfl,(void)) |
1922 |
|
|
{ |
1923 |
|
|
emit_byte(0x9c); |
1924 |
|
|
} |
1925 |
|
|
LENDFUNC(READ,WRITE,0,raw_pushfl,(void)) |
1926 |
|
|
|
1927 |
|
|
LOWFUNC(WRITE,READ,0,raw_popfl,(void)) |
1928 |
|
|
{ |
1929 |
|
|
emit_byte(0x9d); |
1930 |
|
|
} |
1931 |
|
|
LENDFUNC(WRITE,READ,0,raw_popfl,(void)) |
1932 |
|
|
|
1933 |
|
|
/************************************************************************* |
1934 |
|
|
* Unoptimizable stuff --- jump * |
1935 |
|
|
*************************************************************************/ |
1936 |
|
|
|
1937 |
|
|
static __inline__ void raw_call_r(R4 r) |
1938 |
|
|
{ |
1939 |
|
|
emit_byte(0xff); |
1940 |
|
|
emit_byte(0xd0+r); |
1941 |
gbeauche |
1.5 |
} |
1942 |
|
|
|
1943 |
|
|
static __inline__ void raw_call_m_indexed(uae_u32 base, uae_u32 r, uae_u32 m) |
1944 |
|
|
{ |
1945 |
|
|
int mu; |
1946 |
|
|
switch(m) { |
1947 |
|
|
case 1: mu=0; break; |
1948 |
|
|
case 2: mu=1; break; |
1949 |
|
|
case 4: mu=2; break; |
1950 |
|
|
case 8: mu=3; break; |
1951 |
|
|
default: abort(); |
1952 |
|
|
} |
1953 |
|
|
emit_byte(0xff); |
1954 |
|
|
emit_byte(0x14); |
1955 |
|
|
emit_byte(0x05+8*r+0x40*mu); |
1956 |
|
|
emit_long(base); |
1957 |
gbeauche |
1.1 |
} |
1958 |
|
|
|
1959 |
|
|
static __inline__ void raw_jmp_r(R4 r) |
1960 |
|
|
{ |
1961 |
|
|
emit_byte(0xff); |
1962 |
|
|
emit_byte(0xe0+r); |
1963 |
|
|
} |
1964 |
|
|
|
1965 |
|
|
static __inline__ void raw_jmp_m_indexed(uae_u32 base, uae_u32 r, uae_u32 m) |
1966 |
|
|
{ |
1967 |
|
|
int mu; |
1968 |
|
|
switch(m) { |
1969 |
|
|
case 1: mu=0; break; |
1970 |
|
|
case 2: mu=1; break; |
1971 |
|
|
case 4: mu=2; break; |
1972 |
|
|
case 8: mu=3; break; |
1973 |
|
|
default: abort(); |
1974 |
|
|
} |
1975 |
|
|
emit_byte(0xff); |
1976 |
|
|
emit_byte(0x24); |
1977 |
|
|
emit_byte(0x05+8*r+0x40*mu); |
1978 |
|
|
emit_long(base); |
1979 |
|
|
} |
1980 |
|
|
|
1981 |
|
|
static __inline__ void raw_jmp_m(uae_u32 base) |
1982 |
|
|
{ |
1983 |
|
|
emit_byte(0xff); |
1984 |
|
|
emit_byte(0x25); |
1985 |
|
|
emit_long(base); |
1986 |
|
|
} |
1987 |
|
|
|
1988 |
|
|
|
1989 |
|
|
static __inline__ void raw_call(uae_u32 t) |
1990 |
|
|
{ |
1991 |
|
|
emit_byte(0xe8); |
1992 |
|
|
emit_long(t-(uae_u32)target-4); |
1993 |
|
|
} |
1994 |
|
|
|
1995 |
|
|
static __inline__ void raw_jmp(uae_u32 t) |
1996 |
|
|
{ |
1997 |
|
|
emit_byte(0xe9); |
1998 |
|
|
emit_long(t-(uae_u32)target-4); |
1999 |
|
|
} |
2000 |
|
|
|
2001 |
|
|
static __inline__ void raw_jl(uae_u32 t) |
2002 |
|
|
{ |
2003 |
|
|
emit_byte(0x0f); |
2004 |
|
|
emit_byte(0x8c); |
2005 |
|
|
emit_long(t-(uae_u32)target-4); |
2006 |
|
|
} |
2007 |
|
|
|
2008 |
|
|
static __inline__ void raw_jz(uae_u32 t) |
2009 |
|
|
{ |
2010 |
|
|
emit_byte(0x0f); |
2011 |
|
|
emit_byte(0x84); |
2012 |
|
|
emit_long(t-(uae_u32)target-4); |
2013 |
|
|
} |
2014 |
|
|
|
2015 |
|
|
static __inline__ void raw_jnz(uae_u32 t) |
2016 |
|
|
{ |
2017 |
|
|
emit_byte(0x0f); |
2018 |
|
|
emit_byte(0x85); |
2019 |
|
|
emit_long(t-(uae_u32)target-4); |
2020 |
|
|
} |
2021 |
|
|
|
2022 |
|
|
static __inline__ void raw_jnz_l_oponly(void) |
2023 |
|
|
{ |
2024 |
|
|
emit_byte(0x0f); |
2025 |
|
|
emit_byte(0x85); |
2026 |
|
|
} |
2027 |
|
|
|
2028 |
|
|
static __inline__ void raw_jcc_l_oponly(int cc) |
2029 |
|
|
{ |
2030 |
|
|
emit_byte(0x0f); |
2031 |
|
|
emit_byte(0x80+cc); |
2032 |
|
|
} |
2033 |
|
|
|
2034 |
|
|
static __inline__ void raw_jnz_b_oponly(void) |
2035 |
|
|
{ |
2036 |
|
|
emit_byte(0x75); |
2037 |
|
|
} |
2038 |
|
|
|
2039 |
|
|
static __inline__ void raw_jz_b_oponly(void) |
2040 |
|
|
{ |
2041 |
|
|
emit_byte(0x74); |
2042 |
|
|
} |
2043 |
|
|
|
2044 |
|
|
static __inline__ void raw_jcc_b_oponly(int cc) |
2045 |
|
|
{ |
2046 |
|
|
emit_byte(0x70+cc); |
2047 |
|
|
} |
2048 |
|
|
|
2049 |
|
|
static __inline__ void raw_jmp_l_oponly(void) |
2050 |
|
|
{ |
2051 |
|
|
emit_byte(0xe9); |
2052 |
|
|
} |
2053 |
|
|
|
2054 |
|
|
static __inline__ void raw_jmp_b_oponly(void) |
2055 |
|
|
{ |
2056 |
|
|
emit_byte(0xeb); |
2057 |
|
|
} |
2058 |
|
|
|
2059 |
|
|
static __inline__ void raw_ret(void) |
2060 |
|
|
{ |
2061 |
|
|
emit_byte(0xc3); |
2062 |
|
|
} |
2063 |
|
|
|
2064 |
|
|
static __inline__ void raw_nop(void) |
2065 |
|
|
{ |
2066 |
|
|
emit_byte(0x90); |
2067 |
|
|
} |
2068 |
|
|
|
2069 |
gbeauche |
1.8 |
static __inline__ void raw_emit_nop_filler(int nbytes) |
2070 |
|
|
{ |
2071 |
|
|
/* Source: GNU Binutils 2.12.90.0.15 */ |
2072 |
|
|
/* Various efficient no-op patterns for aligning code labels. |
2073 |
|
|
Note: Don't try to assemble the instructions in the comments. |
2074 |
|
|
0L and 0w are not legal. */ |
2075 |
|
|
static const uae_u8 f32_1[] = |
2076 |
|
|
{0x90}; /* nop */ |
2077 |
|
|
static const uae_u8 f32_2[] = |
2078 |
|
|
{0x89,0xf6}; /* movl %esi,%esi */ |
2079 |
|
|
static const uae_u8 f32_3[] = |
2080 |
|
|
{0x8d,0x76,0x00}; /* leal 0(%esi),%esi */ |
2081 |
|
|
static const uae_u8 f32_4[] = |
2082 |
|
|
{0x8d,0x74,0x26,0x00}; /* leal 0(%esi,1),%esi */ |
2083 |
|
|
static const uae_u8 f32_5[] = |
2084 |
|
|
{0x90, /* nop */ |
2085 |
|
|
0x8d,0x74,0x26,0x00}; /* leal 0(%esi,1),%esi */ |
2086 |
|
|
static const uae_u8 f32_6[] = |
2087 |
|
|
{0x8d,0xb6,0x00,0x00,0x00,0x00}; /* leal 0L(%esi),%esi */ |
2088 |
|
|
static const uae_u8 f32_7[] = |
2089 |
|
|
{0x8d,0xb4,0x26,0x00,0x00,0x00,0x00}; /* leal 0L(%esi,1),%esi */ |
2090 |
|
|
static const uae_u8 f32_8[] = |
2091 |
|
|
{0x90, /* nop */ |
2092 |
|
|
0x8d,0xb4,0x26,0x00,0x00,0x00,0x00}; /* leal 0L(%esi,1),%esi */ |
2093 |
|
|
static const uae_u8 f32_9[] = |
2094 |
|
|
{0x89,0xf6, /* movl %esi,%esi */ |
2095 |
|
|
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
2096 |
|
|
static const uae_u8 f32_10[] = |
2097 |
|
|
{0x8d,0x76,0x00, /* leal 0(%esi),%esi */ |
2098 |
|
|
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
2099 |
|
|
static const uae_u8 f32_11[] = |
2100 |
|
|
{0x8d,0x74,0x26,0x00, /* leal 0(%esi,1),%esi */ |
2101 |
|
|
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
2102 |
|
|
static const uae_u8 f32_12[] = |
2103 |
|
|
{0x8d,0xb6,0x00,0x00,0x00,0x00, /* leal 0L(%esi),%esi */ |
2104 |
|
|
0x8d,0xbf,0x00,0x00,0x00,0x00}; /* leal 0L(%edi),%edi */ |
2105 |
|
|
static const uae_u8 f32_13[] = |
2106 |
|
|
{0x8d,0xb6,0x00,0x00,0x00,0x00, /* leal 0L(%esi),%esi */ |
2107 |
|
|
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
2108 |
|
|
static const uae_u8 f32_14[] = |
2109 |
|
|
{0x8d,0xb4,0x26,0x00,0x00,0x00,0x00, /* leal 0L(%esi,1),%esi */ |
2110 |
|
|
0x8d,0xbc,0x27,0x00,0x00,0x00,0x00}; /* leal 0L(%edi,1),%edi */ |
2111 |
|
|
static const uae_u8 f32_15[] = |
2112 |
|
|
{0xeb,0x0d,0x90,0x90,0x90,0x90,0x90, /* jmp .+15; lotsa nops */ |
2113 |
|
|
0x90,0x90,0x90,0x90,0x90,0x90,0x90,0x90}; |
2114 |
|
|
static const uae_u8 f32_16[] = |
2115 |
|
|
{0xeb,0x0d,0x90,0x90,0x90,0x90,0x90, /* jmp .+15; lotsa nops */ |
2116 |
|
|
0x90,0x90,0x90,0x90,0x90,0x90,0x90,0x90,0x90}; |
2117 |
|
|
static const uae_u8 *const f32_patt[] = { |
2118 |
|
|
f32_1, f32_2, f32_3, f32_4, f32_5, f32_6, f32_7, f32_8, |
2119 |
|
|
f32_9, f32_10, f32_11, f32_12, f32_13, f32_14, f32_15 |
2120 |
|
|
}; |
2121 |
|
|
|
2122 |
|
|
int nloops = nbytes / 16; |
2123 |
|
|
while (nloops-- > 0) |
2124 |
|
|
emit_block(f32_16, sizeof(f32_16)); |
2125 |
|
|
|
2126 |
|
|
nbytes %= 16; |
2127 |
|
|
if (nbytes) |
2128 |
|
|
emit_block(f32_patt[nbytes - 1], nbytes); |
2129 |
|
|
} |
2130 |
|
|
|
2131 |
gbeauche |
1.1 |
|
2132 |
|
|
/************************************************************************* |
2133 |
|
|
* Flag handling, to and fro UAE flag register * |
2134 |
|
|
*************************************************************************/ |
2135 |
|
|
|
2136 |
|
|
#ifdef SAHF_SETO_PROFITABLE |
2137 |
|
|
|
2138 |
|
|
#define FLAG_NREG1 0 /* Set to -1 if any register will do */ |
2139 |
|
|
|
2140 |
|
|
static __inline__ void raw_flags_to_reg(int r) |
2141 |
|
|
{ |
2142 |
|
|
raw_lahf(0); /* Most flags in AH */ |
2143 |
|
|
//raw_setcc(r,0); /* V flag in AL */ |
2144 |
|
|
raw_setcc_m((uae_u32)live.state[FLAGTMP].mem,0); |
2145 |
|
|
|
2146 |
|
|
#if 1 /* Let's avoid those nasty partial register stalls */ |
2147 |
|
|
//raw_mov_b_mr((uae_u32)live.state[FLAGTMP].mem,r); |
2148 |
|
|
raw_mov_b_mr(((uae_u32)live.state[FLAGTMP].mem)+1,r+4); |
2149 |
|
|
//live.state[FLAGTMP].status=CLEAN; |
2150 |
|
|
live.state[FLAGTMP].status=INMEM; |
2151 |
|
|
live.state[FLAGTMP].realreg=-1; |
2152 |
|
|
/* We just "evicted" FLAGTMP. */ |
2153 |
|
|
if (live.nat[r].nholds!=1) { |
2154 |
|
|
/* Huh? */ |
2155 |
|
|
abort(); |
2156 |
|
|
} |
2157 |
|
|
live.nat[r].nholds=0; |
2158 |
|
|
#endif |
2159 |
|
|
} |
2160 |
|
|
|
2161 |
|
|
#define FLAG_NREG2 0 /* Set to -1 if any register will do */ |
2162 |
|
|
static __inline__ void raw_reg_to_flags(int r) |
2163 |
|
|
{ |
2164 |
|
|
raw_cmp_b_ri(r,-127); /* set V */ |
2165 |
|
|
raw_sahf(0); |
2166 |
|
|
} |
2167 |
|
|
|
2168 |
|
|
#else |
2169 |
|
|
|
2170 |
|
|
#define FLAG_NREG1 -1 /* Set to -1 if any register will do */ |
2171 |
|
|
static __inline__ void raw_flags_to_reg(int r) |
2172 |
|
|
{ |
2173 |
|
|
raw_pushfl(); |
2174 |
|
|
raw_pop_l_r(r); |
2175 |
|
|
raw_mov_l_mr((uae_u32)live.state[FLAGTMP].mem,r); |
2176 |
|
|
// live.state[FLAGTMP].status=CLEAN; |
2177 |
|
|
live.state[FLAGTMP].status=INMEM; |
2178 |
|
|
live.state[FLAGTMP].realreg=-1; |
2179 |
|
|
/* We just "evicted" FLAGTMP. */ |
2180 |
|
|
if (live.nat[r].nholds!=1) { |
2181 |
|
|
/* Huh? */ |
2182 |
|
|
abort(); |
2183 |
|
|
} |
2184 |
|
|
live.nat[r].nholds=0; |
2185 |
|
|
} |
2186 |
|
|
|
2187 |
|
|
#define FLAG_NREG2 -1 /* Set to -1 if any register will do */ |
2188 |
|
|
static __inline__ void raw_reg_to_flags(int r) |
2189 |
|
|
{ |
2190 |
|
|
raw_push_l_r(r); |
2191 |
|
|
raw_popfl(); |
2192 |
|
|
} |
2193 |
|
|
|
2194 |
|
|
#endif |
2195 |
|
|
|
2196 |
|
|
/* Apparently, there are enough instructions between flag store and |
2197 |
|
|
flag reload to avoid the partial memory stall */ |
2198 |
|
|
static __inline__ void raw_load_flagreg(uae_u32 target, uae_u32 r) |
2199 |
|
|
{ |
2200 |
|
|
#if 1 |
2201 |
|
|
raw_mov_l_rm(target,(uae_u32)live.state[r].mem); |
2202 |
|
|
#else |
2203 |
|
|
raw_mov_b_rm(target,(uae_u32)live.state[r].mem); |
2204 |
|
|
raw_mov_b_rm(target+4,((uae_u32)live.state[r].mem)+1); |
2205 |
|
|
#endif |
2206 |
|
|
} |
2207 |
|
|
|
2208 |
|
|
/* FLAGX is byte sized, and we *do* write it at that size */ |
2209 |
|
|
static __inline__ void raw_load_flagx(uae_u32 target, uae_u32 r) |
2210 |
|
|
{ |
2211 |
|
|
if (live.nat[target].canbyte) |
2212 |
|
|
raw_mov_b_rm(target,(uae_u32)live.state[r].mem); |
2213 |
|
|
else if (live.nat[target].canword) |
2214 |
|
|
raw_mov_w_rm(target,(uae_u32)live.state[r].mem); |
2215 |
|
|
else |
2216 |
|
|
raw_mov_l_rm(target,(uae_u32)live.state[r].mem); |
2217 |
|
|
} |
2218 |
|
|
|
2219 |
|
|
|
2220 |
|
|
static __inline__ void raw_inc_sp(int off) |
2221 |
|
|
{ |
2222 |
gbeauche |
1.2 |
raw_add_l_ri(ESP_INDEX,off); |
2223 |
gbeauche |
1.1 |
} |
2224 |
|
|
|
2225 |
|
|
/************************************************************************* |
2226 |
|
|
* Handling mistaken direct memory access * |
2227 |
|
|
*************************************************************************/ |
2228 |
|
|
|
2229 |
|
|
// gb-- I don't need that part for JIT Basilisk II |
2230 |
|
|
#if defined(NATMEM_OFFSET) && 0 |
2231 |
|
|
#include <asm/sigcontext.h> |
2232 |
|
|
#include <signal.h> |
2233 |
|
|
|
2234 |
|
|
#define SIG_READ 1 |
2235 |
|
|
#define SIG_WRITE 2 |
2236 |
|
|
|
2237 |
|
|
static int in_handler=0; |
2238 |
|
|
static uae_u8 veccode[256]; |
2239 |
|
|
|
2240 |
|
|
static void vec(int x, struct sigcontext sc) |
2241 |
|
|
{ |
2242 |
|
|
uae_u8* i=(uae_u8*)sc.eip; |
2243 |
|
|
uae_u32 addr=sc.cr2; |
2244 |
|
|
int r=-1; |
2245 |
|
|
int size=4; |
2246 |
|
|
int dir=-1; |
2247 |
|
|
int len=0; |
2248 |
|
|
int j; |
2249 |
|
|
|
2250 |
|
|
write_log("fault address is %08x at %08x\n",sc.cr2,sc.eip); |
2251 |
|
|
if (!canbang) |
2252 |
|
|
write_log("Not happy! Canbang is 0 in SIGSEGV handler!\n"); |
2253 |
|
|
if (in_handler) |
2254 |
|
|
write_log("Argh --- Am already in a handler. Shouldn't happen!\n"); |
2255 |
|
|
|
2256 |
|
|
if (canbang && i>=compiled_code && i<=current_compile_p) { |
2257 |
|
|
if (*i==0x66) { |
2258 |
|
|
i++; |
2259 |
|
|
size=2; |
2260 |
|
|
len++; |
2261 |
|
|
} |
2262 |
|
|
|
2263 |
|
|
switch(i[0]) { |
2264 |
|
|
case 0x8a: |
2265 |
|
|
if ((i[1]&0xc0)==0x80) { |
2266 |
|
|
r=(i[1]>>3)&7; |
2267 |
|
|
dir=SIG_READ; |
2268 |
|
|
size=1; |
2269 |
|
|
len+=6; |
2270 |
|
|
break; |
2271 |
|
|
} |
2272 |
|
|
break; |
2273 |
|
|
case 0x88: |
2274 |
|
|
if ((i[1]&0xc0)==0x80) { |
2275 |
|
|
r=(i[1]>>3)&7; |
2276 |
|
|
dir=SIG_WRITE; |
2277 |
|
|
size=1; |
2278 |
|
|
len+=6; |
2279 |
|
|
break; |
2280 |
|
|
} |
2281 |
|
|
break; |
2282 |
|
|
case 0x8b: |
2283 |
|
|
if ((i[1]&0xc0)==0x80) { |
2284 |
|
|
r=(i[1]>>3)&7; |
2285 |
|
|
dir=SIG_READ; |
2286 |
|
|
len+=6; |
2287 |
|
|
break; |
2288 |
|
|
} |
2289 |
|
|
if ((i[1]&0xc0)==0x40) { |
2290 |
|
|
r=(i[1]>>3)&7; |
2291 |
|
|
dir=SIG_READ; |
2292 |
|
|
len+=3; |
2293 |
|
|
break; |
2294 |
|
|
} |
2295 |
|
|
break; |
2296 |
|
|
case 0x89: |
2297 |
|
|
if ((i[1]&0xc0)==0x80) { |
2298 |
|
|
r=(i[1]>>3)&7; |
2299 |
|
|
dir=SIG_WRITE; |
2300 |
|
|
len+=6; |
2301 |
|
|
break; |
2302 |
|
|
} |
2303 |
|
|
if ((i[1]&0xc0)==0x40) { |
2304 |
|
|
r=(i[1]>>3)&7; |
2305 |
|
|
dir=SIG_WRITE; |
2306 |
|
|
len+=3; |
2307 |
|
|
break; |
2308 |
|
|
} |
2309 |
|
|
break; |
2310 |
|
|
} |
2311 |
|
|
} |
2312 |
|
|
|
2313 |
|
|
if (r!=-1) { |
2314 |
|
|
void* pr=NULL; |
2315 |
|
|
write_log("register was %d, direction was %d, size was %d\n",r,dir,size); |
2316 |
|
|
|
2317 |
|
|
switch(r) { |
2318 |
|
|
case 0: pr=&(sc.eax); break; |
2319 |
|
|
case 1: pr=&(sc.ecx); break; |
2320 |
|
|
case 2: pr=&(sc.edx); break; |
2321 |
|
|
case 3: pr=&(sc.ebx); break; |
2322 |
|
|
case 4: pr=(size>1)?NULL:(((uae_u8*)&(sc.eax))+1); break; |
2323 |
|
|
case 5: pr=(size>1)? |
2324 |
|
|
(void*)(&(sc.ebp)): |
2325 |
|
|
(void*)(((uae_u8*)&(sc.ecx))+1); break; |
2326 |
|
|
case 6: pr=(size>1)? |
2327 |
|
|
(void*)(&(sc.esi)): |
2328 |
|
|
(void*)(((uae_u8*)&(sc.edx))+1); break; |
2329 |
|
|
case 7: pr=(size>1)? |
2330 |
|
|
(void*)(&(sc.edi)): |
2331 |
|
|
(void*)(((uae_u8*)&(sc.ebx))+1); break; |
2332 |
|
|
default: abort(); |
2333 |
|
|
} |
2334 |
|
|
if (pr) { |
2335 |
|
|
blockinfo* bi; |
2336 |
|
|
|
2337 |
|
|
if (currprefs.comp_oldsegv) { |
2338 |
|
|
addr-=NATMEM_OFFSET; |
2339 |
|
|
|
2340 |
|
|
if ((addr>=0x10000000 && addr<0x40000000) || |
2341 |
|
|
(addr>=0x50000000)) { |
2342 |
|
|
write_log("Suspicious address in %x SEGV handler.\n",addr); |
2343 |
|
|
} |
2344 |
|
|
if (dir==SIG_READ) { |
2345 |
|
|
switch(size) { |
2346 |
|
|
case 1: *((uae_u8*)pr)=get_byte(addr); break; |
2347 |
|
|
case 2: *((uae_u16*)pr)=get_word(addr); break; |
2348 |
|
|
case 4: *((uae_u32*)pr)=get_long(addr); break; |
2349 |
|
|
default: abort(); |
2350 |
|
|
} |
2351 |
|
|
} |
2352 |
|
|
else { /* write */ |
2353 |
|
|
switch(size) { |
2354 |
|
|
case 1: put_byte(addr,*((uae_u8*)pr)); break; |
2355 |
|
|
case 2: put_word(addr,*((uae_u16*)pr)); break; |
2356 |
|
|
case 4: put_long(addr,*((uae_u32*)pr)); break; |
2357 |
|
|
default: abort(); |
2358 |
|
|
} |
2359 |
|
|
} |
2360 |
|
|
write_log("Handled one access!\n"); |
2361 |
|
|
fflush(stdout); |
2362 |
|
|
segvcount++; |
2363 |
|
|
sc.eip+=len; |
2364 |
|
|
} |
2365 |
|
|
else { |
2366 |
|
|
void* tmp=target; |
2367 |
|
|
int i; |
2368 |
|
|
uae_u8 vecbuf[5]; |
2369 |
|
|
|
2370 |
|
|
addr-=NATMEM_OFFSET; |
2371 |
|
|
|
2372 |
|
|
if ((addr>=0x10000000 && addr<0x40000000) || |
2373 |
|
|
(addr>=0x50000000)) { |
2374 |
|
|
write_log("Suspicious address in %x SEGV handler.\n",addr); |
2375 |
|
|
} |
2376 |
|
|
|
2377 |
|
|
target=(uae_u8*)sc.eip; |
2378 |
|
|
for (i=0;i<5;i++) |
2379 |
|
|
vecbuf[i]=target[i]; |
2380 |
|
|
emit_byte(0xe9); |
2381 |
|
|
emit_long((uae_u32)veccode-(uae_u32)target-4); |
2382 |
|
|
write_log("Create jump to %p\n",veccode); |
2383 |
|
|
|
2384 |
|
|
write_log("Handled one access!\n"); |
2385 |
|
|
fflush(stdout); |
2386 |
|
|
segvcount++; |
2387 |
|
|
|
2388 |
|
|
target=veccode; |
2389 |
|
|
|
2390 |
|
|
if (dir==SIG_READ) { |
2391 |
|
|
switch(size) { |
2392 |
|
|
case 1: raw_mov_b_ri(r,get_byte(addr)); break; |
2393 |
|
|
case 2: raw_mov_w_ri(r,get_byte(addr)); break; |
2394 |
|
|
case 4: raw_mov_l_ri(r,get_byte(addr)); break; |
2395 |
|
|
default: abort(); |
2396 |
|
|
} |
2397 |
|
|
} |
2398 |
|
|
else { /* write */ |
2399 |
|
|
switch(size) { |
2400 |
|
|
case 1: put_byte(addr,*((uae_u8*)pr)); break; |
2401 |
|
|
case 2: put_word(addr,*((uae_u16*)pr)); break; |
2402 |
|
|
case 4: put_long(addr,*((uae_u32*)pr)); break; |
2403 |
|
|
default: abort(); |
2404 |
|
|
} |
2405 |
|
|
} |
2406 |
|
|
for (i=0;i<5;i++) |
2407 |
|
|
raw_mov_b_mi(sc.eip+i,vecbuf[i]); |
2408 |
|
|
raw_mov_l_mi((uae_u32)&in_handler,0); |
2409 |
|
|
emit_byte(0xe9); |
2410 |
|
|
emit_long(sc.eip+len-(uae_u32)target-4); |
2411 |
|
|
in_handler=1; |
2412 |
|
|
target=tmp; |
2413 |
|
|
} |
2414 |
|
|
bi=active; |
2415 |
|
|
while (bi) { |
2416 |
|
|
if (bi->handler && |
2417 |
|
|
(uae_u8*)bi->direct_handler<=i && |
2418 |
|
|
(uae_u8*)bi->nexthandler>i) { |
2419 |
|
|
write_log("deleted trigger (%p<%p<%p) %p\n", |
2420 |
|
|
bi->handler, |
2421 |
|
|
i, |
2422 |
|
|
bi->nexthandler, |
2423 |
|
|
bi->pc_p); |
2424 |
|
|
invalidate_block(bi); |
2425 |
|
|
raise_in_cl_list(bi); |
2426 |
|
|
set_special(0); |
2427 |
|
|
return; |
2428 |
|
|
} |
2429 |
|
|
bi=bi->next; |
2430 |
|
|
} |
2431 |
|
|
/* Not found in the active list. Might be a rom routine that |
2432 |
|
|
is in the dormant list */ |
2433 |
|
|
bi=dormant; |
2434 |
|
|
while (bi) { |
2435 |
|
|
if (bi->handler && |
2436 |
|
|
(uae_u8*)bi->direct_handler<=i && |
2437 |
|
|
(uae_u8*)bi->nexthandler>i) { |
2438 |
|
|
write_log("deleted trigger (%p<%p<%p) %p\n", |
2439 |
|
|
bi->handler, |
2440 |
|
|
i, |
2441 |
|
|
bi->nexthandler, |
2442 |
|
|
bi->pc_p); |
2443 |
|
|
invalidate_block(bi); |
2444 |
|
|
raise_in_cl_list(bi); |
2445 |
|
|
set_special(0); |
2446 |
|
|
return; |
2447 |
|
|
} |
2448 |
|
|
bi=bi->next; |
2449 |
|
|
} |
2450 |
|
|
write_log("Huh? Could not find trigger!\n"); |
2451 |
|
|
return; |
2452 |
|
|
} |
2453 |
|
|
} |
2454 |
|
|
write_log("Can't handle access!\n"); |
2455 |
|
|
for (j=0;j<10;j++) { |
2456 |
|
|
write_log("instruction byte %2d is %02x\n",j,i[j]); |
2457 |
|
|
} |
2458 |
|
|
write_log("Please send the above info (starting at \"fault address\") to\n" |
2459 |
|
|
"bmeyer@csse.monash.edu.au\n" |
2460 |
|
|
"This shouldn't happen ;-)\n"); |
2461 |
|
|
fflush(stdout); |
2462 |
|
|
signal(SIGSEGV,SIG_DFL); /* returning here will cause a "real" SEGV */ |
2463 |
|
|
} |
2464 |
|
|
#endif |
2465 |
|
|
|
2466 |
|
|
|
2467 |
|
|
/************************************************************************* |
2468 |
|
|
* Checking for CPU features * |
2469 |
|
|
*************************************************************************/ |
2470 |
|
|
|
2471 |
gbeauche |
1.3 |
struct cpuinfo_x86 { |
2472 |
|
|
uae_u8 x86; // CPU family |
2473 |
|
|
uae_u8 x86_vendor; // CPU vendor |
2474 |
|
|
uae_u8 x86_processor; // CPU canonical processor type |
2475 |
|
|
uae_u8 x86_brand_id; // CPU BrandID if supported, yield 0 otherwise |
2476 |
|
|
uae_u32 x86_hwcap; |
2477 |
|
|
uae_u8 x86_model; |
2478 |
|
|
uae_u8 x86_mask; |
2479 |
|
|
int cpuid_level; // Maximum supported CPUID level, -1=no CPUID |
2480 |
|
|
char x86_vendor_id[16]; |
2481 |
|
|
}; |
2482 |
|
|
struct cpuinfo_x86 cpuinfo; |
2483 |
|
|
|
2484 |
|
|
enum { |
2485 |
|
|
X86_VENDOR_INTEL = 0, |
2486 |
|
|
X86_VENDOR_CYRIX = 1, |
2487 |
|
|
X86_VENDOR_AMD = 2, |
2488 |
|
|
X86_VENDOR_UMC = 3, |
2489 |
|
|
X86_VENDOR_NEXGEN = 4, |
2490 |
|
|
X86_VENDOR_CENTAUR = 5, |
2491 |
|
|
X86_VENDOR_RISE = 6, |
2492 |
|
|
X86_VENDOR_TRANSMETA = 7, |
2493 |
|
|
X86_VENDOR_NSC = 8, |
2494 |
|
|
X86_VENDOR_UNKNOWN = 0xff |
2495 |
|
|
}; |
2496 |
|
|
|
2497 |
|
|
enum { |
2498 |
|
|
X86_PROCESSOR_I386, /* 80386 */ |
2499 |
|
|
X86_PROCESSOR_I486, /* 80486DX, 80486SX, 80486DX[24] */ |
2500 |
|
|
X86_PROCESSOR_PENTIUM, |
2501 |
|
|
X86_PROCESSOR_PENTIUMPRO, |
2502 |
|
|
X86_PROCESSOR_K6, |
2503 |
|
|
X86_PROCESSOR_ATHLON, |
2504 |
|
|
X86_PROCESSOR_PENTIUM4, |
2505 |
|
|
X86_PROCESSOR_max |
2506 |
|
|
}; |
2507 |
|
|
|
2508 |
|
|
static const char * x86_processor_string_table[X86_PROCESSOR_max] = { |
2509 |
|
|
"80386", |
2510 |
|
|
"80486", |
2511 |
|
|
"Pentium", |
2512 |
|
|
"PentiumPro", |
2513 |
|
|
"K6", |
2514 |
|
|
"Athlon", |
2515 |
|
|
"Pentium4" |
2516 |
|
|
}; |
2517 |
|
|
|
2518 |
|
|
static struct ptt { |
2519 |
|
|
const int align_loop; |
2520 |
|
|
const int align_loop_max_skip; |
2521 |
|
|
const int align_jump; |
2522 |
|
|
const int align_jump_max_skip; |
2523 |
|
|
const int align_func; |
2524 |
|
|
} |
2525 |
|
|
x86_alignments[X86_PROCESSOR_max] = { |
2526 |
|
|
{ 4, 3, 4, 3, 4 }, |
2527 |
|
|
{ 16, 15, 16, 15, 16 }, |
2528 |
|
|
{ 16, 7, 16, 7, 16 }, |
2529 |
|
|
{ 16, 15, 16, 7, 16 }, |
2530 |
|
|
{ 32, 7, 32, 7, 32 }, |
2531 |
gbeauche |
1.4 |
{ 16, 7, 16, 7, 16 }, |
2532 |
gbeauche |
1.3 |
{ 0, 0, 0, 0, 0 } |
2533 |
|
|
}; |
2534 |
gbeauche |
1.1 |
|
2535 |
gbeauche |
1.3 |
static void |
2536 |
|
|
x86_get_cpu_vendor(struct cpuinfo_x86 *c) |
2537 |
gbeauche |
1.1 |
{ |
2538 |
gbeauche |
1.3 |
char *v = c->x86_vendor_id; |
2539 |
|
|
|
2540 |
|
|
if (!strcmp(v, "GenuineIntel")) |
2541 |
|
|
c->x86_vendor = X86_VENDOR_INTEL; |
2542 |
|
|
else if (!strcmp(v, "AuthenticAMD")) |
2543 |
|
|
c->x86_vendor = X86_VENDOR_AMD; |
2544 |
|
|
else if (!strcmp(v, "CyrixInstead")) |
2545 |
|
|
c->x86_vendor = X86_VENDOR_CYRIX; |
2546 |
|
|
else if (!strcmp(v, "Geode by NSC")) |
2547 |
|
|
c->x86_vendor = X86_VENDOR_NSC; |
2548 |
|
|
else if (!strcmp(v, "UMC UMC UMC ")) |
2549 |
|
|
c->x86_vendor = X86_VENDOR_UMC; |
2550 |
|
|
else if (!strcmp(v, "CentaurHauls")) |
2551 |
|
|
c->x86_vendor = X86_VENDOR_CENTAUR; |
2552 |
|
|
else if (!strcmp(v, "NexGenDriven")) |
2553 |
|
|
c->x86_vendor = X86_VENDOR_NEXGEN; |
2554 |
|
|
else if (!strcmp(v, "RiseRiseRise")) |
2555 |
|
|
c->x86_vendor = X86_VENDOR_RISE; |
2556 |
|
|
else if (!strcmp(v, "GenuineTMx86") || |
2557 |
|
|
!strcmp(v, "TransmetaCPU")) |
2558 |
|
|
c->x86_vendor = X86_VENDOR_TRANSMETA; |
2559 |
|
|
else |
2560 |
|
|
c->x86_vendor = X86_VENDOR_UNKNOWN; |
2561 |
|
|
} |
2562 |
gbeauche |
1.1 |
|
2563 |
gbeauche |
1.3 |
static void |
2564 |
|
|
cpuid(uae_u32 op, uae_u32 *eax, uae_u32 *ebx, uae_u32 *ecx, uae_u32 *edx) |
2565 |
|
|
{ |
2566 |
|
|
static uae_u8 cpuid_space[256]; |
2567 |
|
|
uae_u8* tmp=get_target(); |
2568 |
gbeauche |
1.1 |
|
2569 |
gbeauche |
1.3 |
set_target(cpuid_space); |
2570 |
|
|
raw_push_l_r(0); /* eax */ |
2571 |
|
|
raw_push_l_r(1); /* ecx */ |
2572 |
|
|
raw_push_l_r(2); /* edx */ |
2573 |
|
|
raw_push_l_r(3); /* ebx */ |
2574 |
|
|
raw_mov_l_rm(0,(uae_u32)&op); |
2575 |
|
|
raw_cpuid(0); |
2576 |
|
|
if (eax != NULL) raw_mov_l_mr((uae_u32)eax,0); |
2577 |
|
|
if (ebx != NULL) raw_mov_l_mr((uae_u32)ebx,3); |
2578 |
|
|
if (ecx != NULL) raw_mov_l_mr((uae_u32)ecx,1); |
2579 |
|
|
if (edx != NULL) raw_mov_l_mr((uae_u32)edx,2); |
2580 |
|
|
raw_pop_l_r(3); |
2581 |
|
|
raw_pop_l_r(2); |
2582 |
|
|
raw_pop_l_r(1); |
2583 |
|
|
raw_pop_l_r(0); |
2584 |
|
|
raw_ret(); |
2585 |
|
|
set_target(tmp); |
2586 |
gbeauche |
1.1 |
|
2587 |
gbeauche |
1.3 |
((cpuop_func*)cpuid_space)(0); |
2588 |
gbeauche |
1.1 |
} |
2589 |
|
|
|
2590 |
gbeauche |
1.3 |
static void |
2591 |
|
|
raw_init_cpu(void) |
2592 |
gbeauche |
1.1 |
{ |
2593 |
gbeauche |
1.3 |
struct cpuinfo_x86 *c = &cpuinfo; |
2594 |
|
|
|
2595 |
|
|
/* Defaults */ |
2596 |
|
|
c->x86_vendor = X86_VENDOR_UNKNOWN; |
2597 |
|
|
c->cpuid_level = -1; /* CPUID not detected */ |
2598 |
|
|
c->x86_model = c->x86_mask = 0; /* So far unknown... */ |
2599 |
|
|
c->x86_vendor_id[0] = '\0'; /* Unset */ |
2600 |
|
|
c->x86_hwcap = 0; |
2601 |
|
|
|
2602 |
|
|
/* Get vendor name */ |
2603 |
|
|
c->x86_vendor_id[12] = '\0'; |
2604 |
|
|
cpuid(0x00000000, |
2605 |
|
|
(uae_u32 *)&c->cpuid_level, |
2606 |
|
|
(uae_u32 *)&c->x86_vendor_id[0], |
2607 |
|
|
(uae_u32 *)&c->x86_vendor_id[8], |
2608 |
|
|
(uae_u32 *)&c->x86_vendor_id[4]); |
2609 |
|
|
x86_get_cpu_vendor(c); |
2610 |
|
|
|
2611 |
|
|
/* Intel-defined flags: level 0x00000001 */ |
2612 |
|
|
c->x86_brand_id = 0; |
2613 |
|
|
if ( c->cpuid_level >= 0x00000001 ) { |
2614 |
|
|
uae_u32 tfms, brand_id; |
2615 |
|
|
cpuid(0x00000001, &tfms, &brand_id, NULL, &c->x86_hwcap); |
2616 |
|
|
c->x86 = (tfms >> 8) & 15; |
2617 |
|
|
c->x86_model = (tfms >> 4) & 15; |
2618 |
|
|
c->x86_brand_id = brand_id & 0xff; |
2619 |
|
|
if ( (c->x86_vendor == X86_VENDOR_AMD) && |
2620 |
|
|
(c->x86 == 0xf)) { |
2621 |
|
|
/* AMD Extended Family and Model Values */ |
2622 |
|
|
c->x86 += (tfms >> 20) & 0xff; |
2623 |
|
|
c->x86_model += (tfms >> 12) & 0xf0; |
2624 |
|
|
} |
2625 |
|
|
c->x86_mask = tfms & 15; |
2626 |
|
|
} else { |
2627 |
|
|
/* Have CPUID level 0 only - unheard of */ |
2628 |
|
|
c->x86 = 4; |
2629 |
|
|
} |
2630 |
|
|
|
2631 |
|
|
/* Canonicalize processor ID */ |
2632 |
|
|
c->x86_processor = X86_PROCESSOR_max; |
2633 |
|
|
switch (c->x86) { |
2634 |
|
|
case 3: |
2635 |
|
|
c->x86_processor = X86_PROCESSOR_I386; |
2636 |
|
|
break; |
2637 |
|
|
case 4: |
2638 |
|
|
c->x86_processor = X86_PROCESSOR_I486; |
2639 |
|
|
break; |
2640 |
|
|
case 5: |
2641 |
|
|
if (c->x86_vendor == X86_VENDOR_AMD) |
2642 |
|
|
c->x86_processor = X86_PROCESSOR_K6; |
2643 |
|
|
else |
2644 |
|
|
c->x86_processor = X86_PROCESSOR_PENTIUM; |
2645 |
|
|
break; |
2646 |
|
|
case 6: |
2647 |
|
|
if (c->x86_vendor == X86_VENDOR_AMD) |
2648 |
|
|
c->x86_processor = X86_PROCESSOR_ATHLON; |
2649 |
|
|
else |
2650 |
|
|
c->x86_processor = X86_PROCESSOR_PENTIUMPRO; |
2651 |
|
|
break; |
2652 |
|
|
case 15: |
2653 |
|
|
if (c->x86_vendor == X86_VENDOR_INTEL) { |
2654 |
|
|
/* Assume any BranID >= 8 and family == 15 yields a Pentium 4 */ |
2655 |
|
|
if (c->x86_brand_id >= 8) |
2656 |
|
|
c->x86_processor = X86_PROCESSOR_PENTIUM4; |
2657 |
|
|
} |
2658 |
|
|
break; |
2659 |
|
|
} |
2660 |
|
|
if (c->x86_processor == X86_PROCESSOR_max) { |
2661 |
|
|
fprintf(stderr, "Error: unknown processor type\n"); |
2662 |
|
|
fprintf(stderr, " Family : %d\n", c->x86); |
2663 |
|
|
fprintf(stderr, " Model : %d\n", c->x86_model); |
2664 |
|
|
fprintf(stderr, " Mask : %d\n", c->x86_mask); |
2665 |
|
|
if (c->x86_brand_id) |
2666 |
|
|
fprintf(stderr, " BrandID : %02x\n", c->x86_brand_id); |
2667 |
|
|
abort(); |
2668 |
|
|
} |
2669 |
|
|
|
2670 |
|
|
/* Have CMOV support? */ |
2671 |
|
|
have_cmov = (c->x86_hwcap & (1 << 15)) && true; |
2672 |
|
|
|
2673 |
|
|
/* Can the host CPU suffer from partial register stalls? */ |
2674 |
|
|
have_rat_stall = (c->x86_vendor == X86_VENDOR_INTEL); |
2675 |
|
|
#if 1 |
2676 |
|
|
/* It appears that partial register writes are a bad idea even on |
2677 |
gbeauche |
1.1 |
AMD K7 cores, even though they are not supposed to have the |
2678 |
|
|
dreaded rat stall. Why? Anyway, that's why we lie about it ;-) */ |
2679 |
gbeauche |
1.3 |
if (c->x86_processor == X86_PROCESSOR_ATHLON) |
2680 |
|
|
have_rat_stall = true; |
2681 |
gbeauche |
1.1 |
#endif |
2682 |
gbeauche |
1.3 |
|
2683 |
|
|
/* Alignments */ |
2684 |
|
|
if (tune_alignment) { |
2685 |
|
|
align_loops = x86_alignments[c->x86_processor].align_loop; |
2686 |
|
|
align_jumps = x86_alignments[c->x86_processor].align_jump; |
2687 |
|
|
} |
2688 |
|
|
|
2689 |
|
|
write_log("Max CPUID level=%d Processor is %s [%s]\n", |
2690 |
|
|
c->cpuid_level, c->x86_vendor_id, |
2691 |
|
|
x86_processor_string_table[c->x86_processor]); |
2692 |
gbeauche |
1.1 |
} |
2693 |
|
|
|
2694 |
gbeauche |
1.10 |
static bool target_check_bsf(void) |
2695 |
|
|
{ |
2696 |
|
|
bool mismatch = false; |
2697 |
|
|
for (int g_ZF = 0; g_ZF <= 1; g_ZF++) { |
2698 |
|
|
for (int g_CF = 0; g_CF <= 1; g_CF++) { |
2699 |
|
|
for (int g_OF = 0; g_OF <= 1; g_OF++) { |
2700 |
|
|
for (int g_SF = 0; g_SF <= 1; g_SF++) { |
2701 |
|
|
for (int value = -1; value <= 1; value++) { |
2702 |
|
|
int flags = (g_SF << 7) | (g_OF << 11) | (g_ZF << 6) | g_CF; |
2703 |
|
|
int tmp = value; |
2704 |
|
|
__asm__ __volatile__ ("push %0; popf; bsf %1,%1; pushf; pop %0" |
2705 |
|
|
: "+r" (flags), "+r" (tmp) : : "flags"); |
2706 |
|
|
int OF = (flags >> 11) & 1; |
2707 |
|
|
int SF = (flags >> 7) & 1; |
2708 |
|
|
int ZF = (flags >> 6) & 1; |
2709 |
|
|
int CF = flags & 1; |
2710 |
|
|
tmp = (value == 0); |
2711 |
|
|
if (ZF != tmp || SF != g_SF || OF != g_OF || CF != g_CF) |
2712 |
|
|
mismatch = true; |
2713 |
|
|
} |
2714 |
|
|
}}}} |
2715 |
|
|
if (mismatch) |
2716 |
|
|
write_log("Target CPU defines all flags on BSF instruction\n"); |
2717 |
|
|
return !mismatch; |
2718 |
|
|
} |
2719 |
|
|
|
2720 |
gbeauche |
1.1 |
|
2721 |
|
|
/************************************************************************* |
2722 |
|
|
* FPU stuff * |
2723 |
|
|
*************************************************************************/ |
2724 |
|
|
|
2725 |
|
|
|
2726 |
|
|
static __inline__ void raw_fp_init(void) |
2727 |
|
|
{ |
2728 |
|
|
int i; |
2729 |
|
|
|
2730 |
|
|
for (i=0;i<N_FREGS;i++) |
2731 |
|
|
live.spos[i]=-2; |
2732 |
|
|
live.tos=-1; /* Stack is empty */ |
2733 |
|
|
} |
2734 |
|
|
|
2735 |
|
|
static __inline__ void raw_fp_cleanup_drop(void) |
2736 |
|
|
{ |
2737 |
|
|
#if 0 |
2738 |
|
|
/* using FINIT instead of popping all the entries. |
2739 |
|
|
Seems to have side effects --- there is display corruption in |
2740 |
|
|
Quake when this is used */ |
2741 |
|
|
if (live.tos>1) { |
2742 |
|
|
emit_byte(0x9b); |
2743 |
|
|
emit_byte(0xdb); |
2744 |
|
|
emit_byte(0xe3); |
2745 |
|
|
live.tos=-1; |
2746 |
|
|
} |
2747 |
|
|
#endif |
2748 |
|
|
while (live.tos>=1) { |
2749 |
|
|
emit_byte(0xde); |
2750 |
|
|
emit_byte(0xd9); |
2751 |
|
|
live.tos-=2; |
2752 |
|
|
} |
2753 |
|
|
while (live.tos>=0) { |
2754 |
|
|
emit_byte(0xdd); |
2755 |
|
|
emit_byte(0xd8); |
2756 |
|
|
live.tos--; |
2757 |
|
|
} |
2758 |
|
|
raw_fp_init(); |
2759 |
|
|
} |
2760 |
|
|
|
2761 |
|
|
static __inline__ void make_tos(int r) |
2762 |
|
|
{ |
2763 |
|
|
int p,q; |
2764 |
|
|
|
2765 |
|
|
if (live.spos[r]<0) { /* Register not yet on stack */ |
2766 |
|
|
emit_byte(0xd9); |
2767 |
|
|
emit_byte(0xe8); /* Push '1' on the stack, just to grow it */ |
2768 |
|
|
live.tos++; |
2769 |
|
|
live.spos[r]=live.tos; |
2770 |
|
|
live.onstack[live.tos]=r; |
2771 |
|
|
return; |
2772 |
|
|
} |
2773 |
|
|
/* Register is on stack */ |
2774 |
|
|
if (live.tos==live.spos[r]) |
2775 |
|
|
return; |
2776 |
|
|
p=live.spos[r]; |
2777 |
|
|
q=live.onstack[live.tos]; |
2778 |
|
|
|
2779 |
|
|
emit_byte(0xd9); |
2780 |
|
|
emit_byte(0xc8+live.tos-live.spos[r]); /* exchange it with top of stack */ |
2781 |
|
|
live.onstack[live.tos]=r; |
2782 |
|
|
live.spos[r]=live.tos; |
2783 |
|
|
live.onstack[p]=q; |
2784 |
|
|
live.spos[q]=p; |
2785 |
|
|
} |
2786 |
|
|
|
2787 |
|
|
static __inline__ void make_tos2(int r, int r2) |
2788 |
|
|
{ |
2789 |
|
|
int q; |
2790 |
|
|
|
2791 |
|
|
make_tos(r2); /* Put the reg that's supposed to end up in position2 |
2792 |
|
|
on top */ |
2793 |
|
|
|
2794 |
|
|
if (live.spos[r]<0) { /* Register not yet on stack */ |
2795 |
|
|
make_tos(r); /* This will extend the stack */ |
2796 |
|
|
return; |
2797 |
|
|
} |
2798 |
|
|
/* Register is on stack */ |
2799 |
|
|
emit_byte(0xd9); |
2800 |
|
|
emit_byte(0xc9); /* Move r2 into position 2 */ |
2801 |
|
|
|
2802 |
|
|
q=live.onstack[live.tos-1]; |
2803 |
|
|
live.onstack[live.tos]=q; |
2804 |
|
|
live.spos[q]=live.tos; |
2805 |
|
|
live.onstack[live.tos-1]=r2; |
2806 |
|
|
live.spos[r2]=live.tos-1; |
2807 |
|
|
|
2808 |
|
|
make_tos(r); /* And r into 1 */ |
2809 |
|
|
} |
2810 |
|
|
|
2811 |
|
|
static __inline__ int stackpos(int r) |
2812 |
|
|
{ |
2813 |
|
|
if (live.spos[r]<0) |
2814 |
|
|
abort(); |
2815 |
|
|
if (live.tos<live.spos[r]) { |
2816 |
|
|
printf("Looking for spos for fnreg %d\n",r); |
2817 |
|
|
abort(); |
2818 |
|
|
} |
2819 |
|
|
return live.tos-live.spos[r]; |
2820 |
|
|
} |
2821 |
|
|
|
2822 |
|
|
static __inline__ void usereg(int r) |
2823 |
|
|
{ |
2824 |
|
|
if (live.spos[r]<0) |
2825 |
|
|
make_tos(r); |
2826 |
|
|
} |
2827 |
|
|
|
2828 |
|
|
/* This is called with one FP value in a reg *above* tos, which it will |
2829 |
|
|
pop off the stack if necessary */ |
2830 |
|
|
static __inline__ void tos_make(int r) |
2831 |
|
|
{ |
2832 |
|
|
if (live.spos[r]<0) { |
2833 |
|
|
live.tos++; |
2834 |
|
|
live.spos[r]=live.tos; |
2835 |
|
|
live.onstack[live.tos]=r; |
2836 |
|
|
return; |
2837 |
|
|
} |
2838 |
|
|
emit_byte(0xdd); |
2839 |
|
|
emit_byte(0xd8+(live.tos+1)-live.spos[r]); /* store top of stack in reg, |
2840 |
|
|
and pop it*/ |
2841 |
|
|
} |
2842 |
|
|
|
2843 |
|
|
|
2844 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2845 |
|
|
{ |
2846 |
|
|
make_tos(r); |
2847 |
|
|
emit_byte(0xdd); |
2848 |
|
|
emit_byte(0x15); |
2849 |
|
|
emit_long(m); |
2850 |
|
|
} |
2851 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2852 |
|
|
|
2853 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_mr_drop,(MEMW m, FR r)) |
2854 |
|
|
{ |
2855 |
|
|
make_tos(r); |
2856 |
|
|
emit_byte(0xdd); |
2857 |
|
|
emit_byte(0x1d); |
2858 |
|
|
emit_long(m); |
2859 |
|
|
live.onstack[live.tos]=-1; |
2860 |
|
|
live.tos--; |
2861 |
|
|
live.spos[r]=-2; |
2862 |
|
|
} |
2863 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2864 |
|
|
|
2865 |
|
|
LOWFUNC(NONE,READ,2,raw_fmov_rm,(FW r, MEMR m)) |
2866 |
|
|
{ |
2867 |
|
|
emit_byte(0xdd); |
2868 |
|
|
emit_byte(0x05); |
2869 |
|
|
emit_long(m); |
2870 |
|
|
tos_make(r); |
2871 |
|
|
} |
2872 |
|
|
LENDFUNC(NONE,READ,2,raw_fmov_rm,(FW r, MEMR m)) |
2873 |
|
|
|
2874 |
|
|
LOWFUNC(NONE,READ,2,raw_fmovi_rm,(FW r, MEMR m)) |
2875 |
|
|
{ |
2876 |
|
|
emit_byte(0xdb); |
2877 |
|
|
emit_byte(0x05); |
2878 |
|
|
emit_long(m); |
2879 |
|
|
tos_make(r); |
2880 |
|
|
} |
2881 |
|
|
LENDFUNC(NONE,READ,2,raw_fmovi_rm,(FW r, MEMR m)) |
2882 |
|
|
|
2883 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmovi_mr,(MEMW m, FR r)) |
2884 |
|
|
{ |
2885 |
|
|
make_tos(r); |
2886 |
|
|
emit_byte(0xdb); |
2887 |
|
|
emit_byte(0x15); |
2888 |
|
|
emit_long(m); |
2889 |
|
|
} |
2890 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmovi_mr,(MEMW m, FR r)) |
2891 |
|
|
|
2892 |
|
|
LOWFUNC(NONE,READ,2,raw_fmovs_rm,(FW r, MEMR m)) |
2893 |
|
|
{ |
2894 |
|
|
emit_byte(0xd9); |
2895 |
|
|
emit_byte(0x05); |
2896 |
|
|
emit_long(m); |
2897 |
|
|
tos_make(r); |
2898 |
|
|
} |
2899 |
|
|
LENDFUNC(NONE,READ,2,raw_fmovs_rm,(FW r, MEMR m)) |
2900 |
|
|
|
2901 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmovs_mr,(MEMW m, FR r)) |
2902 |
|
|
{ |
2903 |
|
|
make_tos(r); |
2904 |
|
|
emit_byte(0xd9); |
2905 |
|
|
emit_byte(0x15); |
2906 |
|
|
emit_long(m); |
2907 |
|
|
} |
2908 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmovs_mr,(MEMW m, FR r)) |
2909 |
|
|
|
2910 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2911 |
|
|
{ |
2912 |
|
|
int rs; |
2913 |
|
|
|
2914 |
|
|
/* Stupid x87 can't write a long double to mem without popping the |
2915 |
|
|
stack! */ |
2916 |
|
|
usereg(r); |
2917 |
|
|
rs=stackpos(r); |
2918 |
|
|
emit_byte(0xd9); /* Get a copy to the top of stack */ |
2919 |
|
|
emit_byte(0xc0+rs); |
2920 |
|
|
|
2921 |
|
|
emit_byte(0xdb); /* store and pop it */ |
2922 |
|
|
emit_byte(0x3d); |
2923 |
|
|
emit_long(m); |
2924 |
|
|
} |
2925 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2926 |
|
|
|
2927 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_ext_mr_drop,(MEMW m, FR r)) |
2928 |
|
|
{ |
2929 |
|
|
int rs; |
2930 |
|
|
|
2931 |
|
|
make_tos(r); |
2932 |
|
|
emit_byte(0xdb); /* store and pop it */ |
2933 |
|
|
emit_byte(0x3d); |
2934 |
|
|
emit_long(m); |
2935 |
|
|
live.onstack[live.tos]=-1; |
2936 |
|
|
live.tos--; |
2937 |
|
|
live.spos[r]=-2; |
2938 |
|
|
} |
2939 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2940 |
|
|
|
2941 |
|
|
LOWFUNC(NONE,READ,2,raw_fmov_ext_rm,(FW r, MEMR m)) |
2942 |
|
|
{ |
2943 |
|
|
emit_byte(0xdb); |
2944 |
|
|
emit_byte(0x2d); |
2945 |
|
|
emit_long(m); |
2946 |
|
|
tos_make(r); |
2947 |
|
|
} |
2948 |
|
|
LENDFUNC(NONE,READ,2,raw_fmov_ext_rm,(FW r, MEMR m)) |
2949 |
|
|
|
2950 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_pi,(FW r)) |
2951 |
|
|
{ |
2952 |
|
|
emit_byte(0xd9); |
2953 |
|
|
emit_byte(0xeb); |
2954 |
|
|
tos_make(r); |
2955 |
|
|
} |
2956 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_pi,(FW r)) |
2957 |
|
|
|
2958 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_log10_2,(FW r)) |
2959 |
|
|
{ |
2960 |
|
|
emit_byte(0xd9); |
2961 |
|
|
emit_byte(0xec); |
2962 |
|
|
tos_make(r); |
2963 |
|
|
} |
2964 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_log10_2,(FW r)) |
2965 |
|
|
|
2966 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_log2_e,(FW r)) |
2967 |
|
|
{ |
2968 |
|
|
emit_byte(0xd9); |
2969 |
|
|
emit_byte(0xea); |
2970 |
|
|
tos_make(r); |
2971 |
|
|
} |
2972 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_log2_e,(FW r)) |
2973 |
|
|
|
2974 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_loge_2,(FW r)) |
2975 |
|
|
{ |
2976 |
|
|
emit_byte(0xd9); |
2977 |
|
|
emit_byte(0xed); |
2978 |
|
|
tos_make(r); |
2979 |
|
|
} |
2980 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_loge_2,(FW r)) |
2981 |
|
|
|
2982 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_1,(FW r)) |
2983 |
|
|
{ |
2984 |
|
|
emit_byte(0xd9); |
2985 |
|
|
emit_byte(0xe8); |
2986 |
|
|
tos_make(r); |
2987 |
|
|
} |
2988 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_1,(FW r)) |
2989 |
|
|
|
2990 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_0,(FW r)) |
2991 |
|
|
{ |
2992 |
|
|
emit_byte(0xd9); |
2993 |
|
|
emit_byte(0xee); |
2994 |
|
|
tos_make(r); |
2995 |
|
|
} |
2996 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_0,(FW r)) |
2997 |
|
|
|
2998 |
|
|
LOWFUNC(NONE,NONE,2,raw_fmov_rr,(FW d, FR s)) |
2999 |
|
|
{ |
3000 |
|
|
int ds; |
3001 |
|
|
|
3002 |
|
|
usereg(s); |
3003 |
|
|
ds=stackpos(s); |
3004 |
|
|
if (ds==0 && live.spos[d]>=0) { |
3005 |
|
|
/* source is on top of stack, and we already have the dest */ |
3006 |
|
|
int dd=stackpos(d); |
3007 |
|
|
emit_byte(0xdd); |
3008 |
|
|
emit_byte(0xd0+dd); |
3009 |
|
|
} |
3010 |
|
|
else { |
3011 |
|
|
emit_byte(0xd9); |
3012 |
|
|
emit_byte(0xc0+ds); /* duplicate source on tos */ |
3013 |
|
|
tos_make(d); /* store to destination, pop if necessary */ |
3014 |
|
|
} |
3015 |
|
|
} |
3016 |
|
|
LENDFUNC(NONE,NONE,2,raw_fmov_rr,(FW d, FR s)) |
3017 |
|
|
|
3018 |
|
|
LOWFUNC(NONE,READ,4,raw_fldcw_m_indexed,(R4 index, IMM base)) |
3019 |
|
|
{ |
3020 |
|
|
emit_byte(0xd9); |
3021 |
|
|
emit_byte(0xa8+index); |
3022 |
|
|
emit_long(base); |
3023 |
|
|
} |
3024 |
|
|
LENDFUNC(NONE,READ,4,raw_fldcw_m_indexed,(R4 index, IMM base)) |
3025 |
|
|
|
3026 |
|
|
|
3027 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsqrt_rr,(FW d, FR s)) |
3028 |
|
|
{ |
3029 |
|
|
int ds; |
3030 |
|
|
|
3031 |
|
|
if (d!=s) { |
3032 |
|
|
usereg(s); |
3033 |
|
|
ds=stackpos(s); |
3034 |
|
|
emit_byte(0xd9); |
3035 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3036 |
|
|
emit_byte(0xd9); |
3037 |
|
|
emit_byte(0xfa); /* take square root */ |
3038 |
|
|
tos_make(d); /* store to destination */ |
3039 |
|
|
} |
3040 |
|
|
else { |
3041 |
|
|
make_tos(d); |
3042 |
|
|
emit_byte(0xd9); |
3043 |
|
|
emit_byte(0xfa); /* take square root */ |
3044 |
|
|
} |
3045 |
|
|
} |
3046 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsqrt_rr,(FW d, FR s)) |
3047 |
|
|
|
3048 |
|
|
LOWFUNC(NONE,NONE,2,raw_fabs_rr,(FW d, FR s)) |
3049 |
|
|
{ |
3050 |
|
|
int ds; |
3051 |
|
|
|
3052 |
|
|
if (d!=s) { |
3053 |
|
|
usereg(s); |
3054 |
|
|
ds=stackpos(s); |
3055 |
|
|
emit_byte(0xd9); |
3056 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3057 |
|
|
emit_byte(0xd9); |
3058 |
|
|
emit_byte(0xe1); /* take fabs */ |
3059 |
|
|
tos_make(d); /* store to destination */ |
3060 |
|
|
} |
3061 |
|
|
else { |
3062 |
|
|
make_tos(d); |
3063 |
|
|
emit_byte(0xd9); |
3064 |
|
|
emit_byte(0xe1); /* take fabs */ |
3065 |
|
|
} |
3066 |
|
|
} |
3067 |
|
|
LENDFUNC(NONE,NONE,2,raw_fabs_rr,(FW d, FR s)) |
3068 |
|
|
|
3069 |
|
|
LOWFUNC(NONE,NONE,2,raw_frndint_rr,(FW d, FR s)) |
3070 |
|
|
{ |
3071 |
|
|
int ds; |
3072 |
|
|
|
3073 |
|
|
if (d!=s) { |
3074 |
|
|
usereg(s); |
3075 |
|
|
ds=stackpos(s); |
3076 |
|
|
emit_byte(0xd9); |
3077 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3078 |
|
|
emit_byte(0xd9); |
3079 |
|
|
emit_byte(0xfc); /* take frndint */ |
3080 |
|
|
tos_make(d); /* store to destination */ |
3081 |
|
|
} |
3082 |
|
|
else { |
3083 |
|
|
make_tos(d); |
3084 |
|
|
emit_byte(0xd9); |
3085 |
|
|
emit_byte(0xfc); /* take frndint */ |
3086 |
|
|
} |
3087 |
|
|
} |
3088 |
|
|
LENDFUNC(NONE,NONE,2,raw_frndint_rr,(FW d, FR s)) |
3089 |
|
|
|
3090 |
|
|
LOWFUNC(NONE,NONE,2,raw_fcos_rr,(FW d, FR s)) |
3091 |
|
|
{ |
3092 |
|
|
int ds; |
3093 |
|
|
|
3094 |
|
|
if (d!=s) { |
3095 |
|
|
usereg(s); |
3096 |
|
|
ds=stackpos(s); |
3097 |
|
|
emit_byte(0xd9); |
3098 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3099 |
|
|
emit_byte(0xd9); |
3100 |
|
|
emit_byte(0xff); /* take cos */ |
3101 |
|
|
tos_make(d); /* store to destination */ |
3102 |
|
|
} |
3103 |
|
|
else { |
3104 |
|
|
make_tos(d); |
3105 |
|
|
emit_byte(0xd9); |
3106 |
|
|
emit_byte(0xff); /* take cos */ |
3107 |
|
|
} |
3108 |
|
|
} |
3109 |
|
|
LENDFUNC(NONE,NONE,2,raw_fcos_rr,(FW d, FR s)) |
3110 |
|
|
|
3111 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsin_rr,(FW d, FR s)) |
3112 |
|
|
{ |
3113 |
|
|
int ds; |
3114 |
|
|
|
3115 |
|
|
if (d!=s) { |
3116 |
|
|
usereg(s); |
3117 |
|
|
ds=stackpos(s); |
3118 |
|
|
emit_byte(0xd9); |
3119 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3120 |
|
|
emit_byte(0xd9); |
3121 |
|
|
emit_byte(0xfe); /* take sin */ |
3122 |
|
|
tos_make(d); /* store to destination */ |
3123 |
|
|
} |
3124 |
|
|
else { |
3125 |
|
|
make_tos(d); |
3126 |
|
|
emit_byte(0xd9); |
3127 |
|
|
emit_byte(0xfe); /* take sin */ |
3128 |
|
|
} |
3129 |
|
|
} |
3130 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsin_rr,(FW d, FR s)) |
3131 |
|
|
|
3132 |
|
|
double one=1; |
3133 |
|
|
LOWFUNC(NONE,NONE,2,raw_ftwotox_rr,(FW d, FR s)) |
3134 |
|
|
{ |
3135 |
|
|
int ds; |
3136 |
|
|
|
3137 |
|
|
usereg(s); |
3138 |
|
|
ds=stackpos(s); |
3139 |
|
|
emit_byte(0xd9); |
3140 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3141 |
|
|
|
3142 |
|
|
emit_byte(0xd9); |
3143 |
|
|
emit_byte(0xc0); /* duplicate top of stack. Now up to 8 high */ |
3144 |
|
|
emit_byte(0xd9); |
3145 |
|
|
emit_byte(0xfc); /* rndint */ |
3146 |
|
|
emit_byte(0xd9); |
3147 |
|
|
emit_byte(0xc9); /* swap top two elements */ |
3148 |
|
|
emit_byte(0xd8); |
3149 |
|
|
emit_byte(0xe1); /* subtract rounded from original */ |
3150 |
|
|
emit_byte(0xd9); |
3151 |
|
|
emit_byte(0xf0); /* f2xm1 */ |
3152 |
|
|
emit_byte(0xdc); |
3153 |
|
|
emit_byte(0x05); |
3154 |
|
|
emit_long((uae_u32)&one); /* Add '1' without using extra stack space */ |
3155 |
|
|
emit_byte(0xd9); |
3156 |
|
|
emit_byte(0xfd); /* and scale it */ |
3157 |
|
|
emit_byte(0xdd); |
3158 |
|
|
emit_byte(0xd9); /* take he rounded value off */ |
3159 |
|
|
tos_make(d); /* store to destination */ |
3160 |
|
|
} |
3161 |
|
|
LENDFUNC(NONE,NONE,2,raw_ftwotox_rr,(FW d, FR s)) |
3162 |
|
|
|
3163 |
|
|
LOWFUNC(NONE,NONE,2,raw_fetox_rr,(FW d, FR s)) |
3164 |
|
|
{ |
3165 |
|
|
int ds; |
3166 |
|
|
|
3167 |
|
|
usereg(s); |
3168 |
|
|
ds=stackpos(s); |
3169 |
|
|
emit_byte(0xd9); |
3170 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3171 |
|
|
emit_byte(0xd9); |
3172 |
|
|
emit_byte(0xea); /* fldl2e */ |
3173 |
|
|
emit_byte(0xde); |
3174 |
|
|
emit_byte(0xc9); /* fmulp --- multiply source by log2(e) */ |
3175 |
|
|
|
3176 |
|
|
emit_byte(0xd9); |
3177 |
|
|
emit_byte(0xc0); /* duplicate top of stack. Now up to 8 high */ |
3178 |
|
|
emit_byte(0xd9); |
3179 |
|
|
emit_byte(0xfc); /* rndint */ |
3180 |
|
|
emit_byte(0xd9); |
3181 |
|
|
emit_byte(0xc9); /* swap top two elements */ |
3182 |
|
|
emit_byte(0xd8); |
3183 |
|
|
emit_byte(0xe1); /* subtract rounded from original */ |
3184 |
|
|
emit_byte(0xd9); |
3185 |
|
|
emit_byte(0xf0); /* f2xm1 */ |
3186 |
|
|
emit_byte(0xdc); |
3187 |
|
|
emit_byte(0x05); |
3188 |
|
|
emit_long((uae_u32)&one); /* Add '1' without using extra stack space */ |
3189 |
|
|
emit_byte(0xd9); |
3190 |
|
|
emit_byte(0xfd); /* and scale it */ |
3191 |
|
|
emit_byte(0xdd); |
3192 |
|
|
emit_byte(0xd9); /* take he rounded value off */ |
3193 |
|
|
tos_make(d); /* store to destination */ |
3194 |
|
|
} |
3195 |
|
|
LENDFUNC(NONE,NONE,2,raw_fetox_rr,(FW d, FR s)) |
3196 |
|
|
|
3197 |
|
|
LOWFUNC(NONE,NONE,2,raw_flog2_rr,(FW d, FR s)) |
3198 |
|
|
{ |
3199 |
|
|
int ds; |
3200 |
|
|
|
3201 |
|
|
usereg(s); |
3202 |
|
|
ds=stackpos(s); |
3203 |
|
|
emit_byte(0xd9); |
3204 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3205 |
|
|
emit_byte(0xd9); |
3206 |
|
|
emit_byte(0xe8); /* push '1' */ |
3207 |
|
|
emit_byte(0xd9); |
3208 |
|
|
emit_byte(0xc9); /* swap top two */ |
3209 |
|
|
emit_byte(0xd9); |
3210 |
|
|
emit_byte(0xf1); /* take 1*log2(x) */ |
3211 |
|
|
tos_make(d); /* store to destination */ |
3212 |
|
|
} |
3213 |
|
|
LENDFUNC(NONE,NONE,2,raw_flog2_rr,(FW d, FR s)) |
3214 |
|
|
|
3215 |
|
|
|
3216 |
|
|
LOWFUNC(NONE,NONE,2,raw_fneg_rr,(FW d, FR s)) |
3217 |
|
|
{ |
3218 |
|
|
int ds; |
3219 |
|
|
|
3220 |
|
|
if (d!=s) { |
3221 |
|
|
usereg(s); |
3222 |
|
|
ds=stackpos(s); |
3223 |
|
|
emit_byte(0xd9); |
3224 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
3225 |
|
|
emit_byte(0xd9); |
3226 |
|
|
emit_byte(0xe0); /* take fchs */ |
3227 |
|
|
tos_make(d); /* store to destination */ |
3228 |
|
|
} |
3229 |
|
|
else { |
3230 |
|
|
make_tos(d); |
3231 |
|
|
emit_byte(0xd9); |
3232 |
|
|
emit_byte(0xe0); /* take fchs */ |
3233 |
|
|
} |
3234 |
|
|
} |
3235 |
|
|
LENDFUNC(NONE,NONE,2,raw_fneg_rr,(FW d, FR s)) |
3236 |
|
|
|
3237 |
|
|
LOWFUNC(NONE,NONE,2,raw_fadd_rr,(FRW d, FR s)) |
3238 |
|
|
{ |
3239 |
|
|
int ds; |
3240 |
|
|
|
3241 |
|
|
usereg(s); |
3242 |
|
|
usereg(d); |
3243 |
|
|
|
3244 |
|
|
if (live.spos[s]==live.tos) { |
3245 |
|
|
/* Source is on top of stack */ |
3246 |
|
|
ds=stackpos(d); |
3247 |
|
|
emit_byte(0xdc); |
3248 |
|
|
emit_byte(0xc0+ds); /* add source to dest*/ |
3249 |
|
|
} |
3250 |
|
|
else { |
3251 |
|
|
make_tos(d); |
3252 |
|
|
ds=stackpos(s); |
3253 |
|
|
|
3254 |
|
|
emit_byte(0xd8); |
3255 |
|
|
emit_byte(0xc0+ds); /* add source to dest*/ |
3256 |
|
|
} |
3257 |
|
|
} |
3258 |
|
|
LENDFUNC(NONE,NONE,2,raw_fadd_rr,(FRW d, FR s)) |
3259 |
|
|
|
3260 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsub_rr,(FRW d, FR s)) |
3261 |
|
|
{ |
3262 |
|
|
int ds; |
3263 |
|
|
|
3264 |
|
|
usereg(s); |
3265 |
|
|
usereg(d); |
3266 |
|
|
|
3267 |
|
|
if (live.spos[s]==live.tos) { |
3268 |
|
|
/* Source is on top of stack */ |
3269 |
|
|
ds=stackpos(d); |
3270 |
|
|
emit_byte(0xdc); |
3271 |
|
|
emit_byte(0xe8+ds); /* sub source from dest*/ |
3272 |
|
|
} |
3273 |
|
|
else { |
3274 |
|
|
make_tos(d); |
3275 |
|
|
ds=stackpos(s); |
3276 |
|
|
|
3277 |
|
|
emit_byte(0xd8); |
3278 |
|
|
emit_byte(0xe0+ds); /* sub src from dest */ |
3279 |
|
|
} |
3280 |
|
|
} |
3281 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsub_rr,(FRW d, FR s)) |
3282 |
|
|
|
3283 |
|
|
LOWFUNC(NONE,NONE,2,raw_fcmp_rr,(FR d, FR s)) |
3284 |
|
|
{ |
3285 |
|
|
int ds; |
3286 |
|
|
|
3287 |
|
|
usereg(s); |
3288 |
|
|
usereg(d); |
3289 |
|
|
|
3290 |
|
|
make_tos(d); |
3291 |
|
|
ds=stackpos(s); |
3292 |
|
|
|
3293 |
|
|
emit_byte(0xdd); |
3294 |
|
|
emit_byte(0xe0+ds); /* cmp dest with source*/ |
3295 |
|
|
} |
3296 |
|
|
LENDFUNC(NONE,NONE,2,raw_fcmp_rr,(FR d, FR s)) |
3297 |
|
|
|
3298 |
|
|
LOWFUNC(NONE,NONE,2,raw_fmul_rr,(FRW d, FR s)) |
3299 |
|
|
{ |
3300 |
|
|
int ds; |
3301 |
|
|
|
3302 |
|
|
usereg(s); |
3303 |
|
|
usereg(d); |
3304 |
|
|
|
3305 |
|
|
if (live.spos[s]==live.tos) { |
3306 |
|
|
/* Source is on top of stack */ |
3307 |
|
|
ds=stackpos(d); |
3308 |
|
|
emit_byte(0xdc); |
3309 |
|
|
emit_byte(0xc8+ds); /* mul dest by source*/ |
3310 |
|
|
} |
3311 |
|
|
else { |
3312 |
|
|
make_tos(d); |
3313 |
|
|
ds=stackpos(s); |
3314 |
|
|
|
3315 |
|
|
emit_byte(0xd8); |
3316 |
|
|
emit_byte(0xc8+ds); /* mul dest by source*/ |
3317 |
|
|
} |
3318 |
|
|
} |
3319 |
|
|
LENDFUNC(NONE,NONE,2,raw_fmul_rr,(FRW d, FR s)) |
3320 |
|
|
|
3321 |
|
|
LOWFUNC(NONE,NONE,2,raw_fdiv_rr,(FRW d, FR s)) |
3322 |
|
|
{ |
3323 |
|
|
int ds; |
3324 |
|
|
|
3325 |
|
|
usereg(s); |
3326 |
|
|
usereg(d); |
3327 |
|
|
|
3328 |
|
|
if (live.spos[s]==live.tos) { |
3329 |
|
|
/* Source is on top of stack */ |
3330 |
|
|
ds=stackpos(d); |
3331 |
|
|
emit_byte(0xdc); |
3332 |
|
|
emit_byte(0xf8+ds); /* div dest by source */ |
3333 |
|
|
} |
3334 |
|
|
else { |
3335 |
|
|
make_tos(d); |
3336 |
|
|
ds=stackpos(s); |
3337 |
|
|
|
3338 |
|
|
emit_byte(0xd8); |
3339 |
|
|
emit_byte(0xf0+ds); /* div dest by source*/ |
3340 |
|
|
} |
3341 |
|
|
} |
3342 |
|
|
LENDFUNC(NONE,NONE,2,raw_fdiv_rr,(FRW d, FR s)) |
3343 |
|
|
|
3344 |
|
|
LOWFUNC(NONE,NONE,2,raw_frem_rr,(FRW d, FR s)) |
3345 |
|
|
{ |
3346 |
|
|
int ds; |
3347 |
|
|
|
3348 |
|
|
usereg(s); |
3349 |
|
|
usereg(d); |
3350 |
|
|
|
3351 |
|
|
make_tos2(d,s); |
3352 |
|
|
ds=stackpos(s); |
3353 |
|
|
|
3354 |
|
|
if (ds!=1) { |
3355 |
|
|
printf("Failed horribly in raw_frem_rr! ds is %d\n",ds); |
3356 |
|
|
abort(); |
3357 |
|
|
} |
3358 |
|
|
emit_byte(0xd9); |
3359 |
|
|
emit_byte(0xf8); /* take rem from dest by source */ |
3360 |
|
|
} |
3361 |
|
|
LENDFUNC(NONE,NONE,2,raw_frem_rr,(FRW d, FR s)) |
3362 |
|
|
|
3363 |
|
|
LOWFUNC(NONE,NONE,2,raw_frem1_rr,(FRW d, FR s)) |
3364 |
|
|
{ |
3365 |
|
|
int ds; |
3366 |
|
|
|
3367 |
|
|
usereg(s); |
3368 |
|
|
usereg(d); |
3369 |
|
|
|
3370 |
|
|
make_tos2(d,s); |
3371 |
|
|
ds=stackpos(s); |
3372 |
|
|
|
3373 |
|
|
if (ds!=1) { |
3374 |
|
|
printf("Failed horribly in raw_frem1_rr! ds is %d\n",ds); |
3375 |
|
|
abort(); |
3376 |
|
|
} |
3377 |
|
|
emit_byte(0xd9); |
3378 |
|
|
emit_byte(0xf5); /* take rem1 from dest by source */ |
3379 |
|
|
} |
3380 |
|
|
LENDFUNC(NONE,NONE,2,raw_frem1_rr,(FRW d, FR s)) |
3381 |
|
|
|
3382 |
|
|
|
3383 |
|
|
LOWFUNC(NONE,NONE,1,raw_ftst_r,(FR r)) |
3384 |
|
|
{ |
3385 |
|
|
make_tos(r); |
3386 |
|
|
emit_byte(0xd9); /* ftst */ |
3387 |
|
|
emit_byte(0xe4); |
3388 |
|
|
} |
3389 |
|
|
LENDFUNC(NONE,NONE,1,raw_ftst_r,(FR r)) |
3390 |
|
|
|
3391 |
|
|
/* %eax register is clobbered if target processor doesn't support fucomi */ |
3392 |
|
|
#define FFLAG_NREG_CLOBBER_CONDITION !have_cmov |
3393 |
|
|
#define FFLAG_NREG EAX_INDEX |
3394 |
|
|
|
3395 |
|
|
static __inline__ void raw_fflags_into_flags(int r) |
3396 |
|
|
{ |
3397 |
|
|
int p; |
3398 |
|
|
|
3399 |
|
|
usereg(r); |
3400 |
|
|
p=stackpos(r); |
3401 |
|
|
|
3402 |
|
|
emit_byte(0xd9); |
3403 |
|
|
emit_byte(0xee); /* Push 0 */ |
3404 |
|
|
emit_byte(0xd9); |
3405 |
|
|
emit_byte(0xc9+p); /* swap top two around */ |
3406 |
|
|
if (have_cmov) { |
3407 |
|
|
// gb-- fucomi is for P6 cores only, not K6-2 then... |
3408 |
|
|
emit_byte(0xdb); |
3409 |
|
|
emit_byte(0xe9+p); /* fucomi them */ |
3410 |
|
|
} |
3411 |
|
|
else { |
3412 |
|
|
emit_byte(0xdd); |
3413 |
|
|
emit_byte(0xe1+p); /* fucom them */ |
3414 |
|
|
emit_byte(0x9b); |
3415 |
|
|
emit_byte(0xdf); |
3416 |
|
|
emit_byte(0xe0); /* fstsw ax */ |
3417 |
|
|
raw_sahf(0); /* sahf */ |
3418 |
|
|
} |
3419 |
|
|
emit_byte(0xdd); |
3420 |
|
|
emit_byte(0xd9+p); /* store value back, and get rid of 0 */ |
3421 |
|
|
} |