1 |
gbeauche |
1.1 |
/* This should eventually end up in machdep/, but for now, x86 is the |
2 |
|
|
only target, and it's easier this way... */ |
3 |
|
|
|
4 |
|
|
/************************************************************************* |
5 |
|
|
* Some basic information about the the target CPU * |
6 |
|
|
*************************************************************************/ |
7 |
|
|
|
8 |
|
|
#define EAX_INDEX 0 |
9 |
|
|
#define ECX_INDEX 1 |
10 |
|
|
#define EDX_INDEX 2 |
11 |
|
|
#define EBX_INDEX 3 |
12 |
|
|
#define ESP_INDEX 4 |
13 |
|
|
#define EBP_INDEX 5 |
14 |
|
|
#define ESI_INDEX 6 |
15 |
|
|
#define EDI_INDEX 7 |
16 |
|
|
|
17 |
|
|
/* The register in which subroutines return an integer return value */ |
18 |
|
|
#define REG_RESULT 0 |
19 |
|
|
|
20 |
|
|
/* The registers subroutines take their first and second argument in */ |
21 |
|
|
#if defined( _MSC_VER ) && !defined( USE_NORMAL_CALLING_CONVENTION ) |
22 |
|
|
/* Handle the _fastcall parameters of ECX and EDX */ |
23 |
|
|
#define REG_PAR1 1 |
24 |
|
|
#define REG_PAR2 2 |
25 |
|
|
#else |
26 |
|
|
#define REG_PAR1 0 |
27 |
|
|
#define REG_PAR2 2 |
28 |
|
|
#endif |
29 |
|
|
|
30 |
|
|
/* Three registers that are not used for any of the above */ |
31 |
|
|
#define REG_NOPAR1 6 |
32 |
|
|
#define REG_NOPAR2 5 |
33 |
|
|
#define REG_NOPAR3 3 |
34 |
|
|
|
35 |
|
|
#define REG_PC_PRE 0 /* The register we use for preloading regs.pc_p */ |
36 |
|
|
#if defined( _MSC_VER ) && !defined( USE_NORMAL_CALLING_CONVENTION ) |
37 |
|
|
#define REG_PC_TMP 0 |
38 |
|
|
#else |
39 |
|
|
#define REG_PC_TMP 1 /* Another register that is not the above */ |
40 |
|
|
#endif |
41 |
|
|
|
42 |
|
|
#define SHIFTCOUNT_NREG 1 /* Register that can be used for shiftcount. |
43 |
|
|
-1 if any reg will do */ |
44 |
|
|
#define MUL_NREG1 0 /* %eax will hold the low 32 bits after a 32x32 mul */ |
45 |
|
|
#define MUL_NREG2 2 /* %edx will hold the high 32 bits */ |
46 |
|
|
|
47 |
|
|
uae_s8 always_used[]={4,-1}; |
48 |
|
|
uae_s8 can_byte[]={0,1,2,3,-1}; |
49 |
|
|
uae_s8 can_word[]={0,1,2,3,5,6,7,-1}; |
50 |
|
|
|
51 |
|
|
/* cpuopti mutate instruction handlers to assume registers are saved |
52 |
|
|
by the caller */ |
53 |
|
|
uae_u8 call_saved[]={0,0,0,0,1,0,0,0}; |
54 |
|
|
|
55 |
|
|
/* This *should* be the same as call_saved. But: |
56 |
|
|
- We might not really know which registers are saved, and which aren't, |
57 |
|
|
so we need to preserve some, but don't want to rely on everyone else |
58 |
|
|
also saving those registers |
59 |
|
|
- Special registers (such like the stack pointer) should not be "preserved" |
60 |
|
|
by pushing, even though they are "saved" across function calls |
61 |
|
|
*/ |
62 |
|
|
uae_u8 need_to_preserve[]={1,1,1,1,0,1,1,1}; |
63 |
|
|
|
64 |
|
|
/* Whether classes of instructions do or don't clobber the native flags */ |
65 |
|
|
#define CLOBBER_MOV |
66 |
|
|
#define CLOBBER_LEA |
67 |
|
|
#define CLOBBER_CMOV |
68 |
|
|
#define CLOBBER_POP |
69 |
|
|
#define CLOBBER_PUSH |
70 |
|
|
#define CLOBBER_SUB clobber_flags() |
71 |
|
|
#define CLOBBER_SBB clobber_flags() |
72 |
|
|
#define CLOBBER_CMP clobber_flags() |
73 |
|
|
#define CLOBBER_ADD clobber_flags() |
74 |
|
|
#define CLOBBER_ADC clobber_flags() |
75 |
|
|
#define CLOBBER_AND clobber_flags() |
76 |
|
|
#define CLOBBER_OR clobber_flags() |
77 |
|
|
#define CLOBBER_XOR clobber_flags() |
78 |
|
|
|
79 |
|
|
#define CLOBBER_ROL clobber_flags() |
80 |
|
|
#define CLOBBER_ROR clobber_flags() |
81 |
|
|
#define CLOBBER_SHLL clobber_flags() |
82 |
|
|
#define CLOBBER_SHRL clobber_flags() |
83 |
|
|
#define CLOBBER_SHRA clobber_flags() |
84 |
|
|
#define CLOBBER_TEST clobber_flags() |
85 |
|
|
#define CLOBBER_CL16 |
86 |
|
|
#define CLOBBER_CL8 |
87 |
|
|
#define CLOBBER_SE16 |
88 |
|
|
#define CLOBBER_SE8 |
89 |
|
|
#define CLOBBER_ZE16 |
90 |
|
|
#define CLOBBER_ZE8 |
91 |
|
|
#define CLOBBER_SW16 clobber_flags() |
92 |
|
|
#define CLOBBER_SW32 |
93 |
|
|
#define CLOBBER_SETCC |
94 |
|
|
#define CLOBBER_MUL clobber_flags() |
95 |
|
|
#define CLOBBER_BT clobber_flags() |
96 |
|
|
#define CLOBBER_BSF clobber_flags() |
97 |
|
|
|
98 |
|
|
const bool optimize_imm8 = true; |
99 |
|
|
const bool optimize_shift_once = true; |
100 |
|
|
|
101 |
|
|
/************************************************************************* |
102 |
|
|
* Actual encoding of the instructions on the target CPU * |
103 |
|
|
*************************************************************************/ |
104 |
|
|
|
105 |
|
|
static __inline__ int isbyte(uae_s32 x) |
106 |
|
|
{ |
107 |
|
|
return (x>=-128 && x<=127); |
108 |
|
|
} |
109 |
|
|
|
110 |
|
|
static __inline__ int isword(uae_s32 x) |
111 |
|
|
{ |
112 |
|
|
return (x>=-32768 && x<=32767); |
113 |
|
|
} |
114 |
|
|
|
115 |
|
|
LOWFUNC(NONE,WRITE,1,raw_push_l_r,(R4 r)) |
116 |
|
|
{ |
117 |
|
|
emit_byte(0x50+r); |
118 |
|
|
} |
119 |
|
|
LENDFUNC(NONE,WRITE,1,raw_push_l_r,(R4 r)) |
120 |
|
|
|
121 |
|
|
LOWFUNC(NONE,READ,1,raw_pop_l_r,(R4 r)) |
122 |
|
|
{ |
123 |
|
|
emit_byte(0x58+r); |
124 |
|
|
} |
125 |
|
|
LENDFUNC(NONE,READ,1,raw_pop_l_r,(R4 r)) |
126 |
|
|
|
127 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bt_l_ri,(R4 r, IMM i)) |
128 |
|
|
{ |
129 |
|
|
emit_byte(0x0f); |
130 |
|
|
emit_byte(0xba); |
131 |
|
|
emit_byte(0xe0+r); |
132 |
|
|
emit_byte(i); |
133 |
|
|
} |
134 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bt_l_ri,(R4 r, IMM i)) |
135 |
|
|
|
136 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bt_l_rr,(R4 r, R4 b)) |
137 |
|
|
{ |
138 |
|
|
emit_byte(0x0f); |
139 |
|
|
emit_byte(0xa3); |
140 |
|
|
emit_byte(0xc0+8*b+r); |
141 |
|
|
} |
142 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bt_l_rr,(R4 r, R4 b)) |
143 |
|
|
|
144 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btc_l_ri,(RW4 r, IMM i)) |
145 |
|
|
{ |
146 |
|
|
emit_byte(0x0f); |
147 |
|
|
emit_byte(0xba); |
148 |
|
|
emit_byte(0xf8+r); |
149 |
|
|
emit_byte(i); |
150 |
|
|
} |
151 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btc_l_ri,(RW4 r, IMM i)) |
152 |
|
|
|
153 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btc_l_rr,(RW4 r, R4 b)) |
154 |
|
|
{ |
155 |
|
|
emit_byte(0x0f); |
156 |
|
|
emit_byte(0xbb); |
157 |
|
|
emit_byte(0xc0+8*b+r); |
158 |
|
|
} |
159 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btc_l_rr,(RW4 r, R4 b)) |
160 |
|
|
|
161 |
|
|
|
162 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btr_l_ri,(RW4 r, IMM i)) |
163 |
|
|
{ |
164 |
|
|
emit_byte(0x0f); |
165 |
|
|
emit_byte(0xba); |
166 |
|
|
emit_byte(0xf0+r); |
167 |
|
|
emit_byte(i); |
168 |
|
|
} |
169 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btr_l_ri,(RW4 r, IMM i)) |
170 |
|
|
|
171 |
|
|
LOWFUNC(WRITE,NONE,2,raw_btr_l_rr,(RW4 r, R4 b)) |
172 |
|
|
{ |
173 |
|
|
emit_byte(0x0f); |
174 |
|
|
emit_byte(0xb3); |
175 |
|
|
emit_byte(0xc0+8*b+r); |
176 |
|
|
} |
177 |
|
|
LENDFUNC(WRITE,NONE,2,raw_btr_l_rr,(RW4 r, R4 b)) |
178 |
|
|
|
179 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bts_l_ri,(RW4 r, IMM i)) |
180 |
|
|
{ |
181 |
|
|
emit_byte(0x0f); |
182 |
|
|
emit_byte(0xba); |
183 |
|
|
emit_byte(0xe8+r); |
184 |
|
|
emit_byte(i); |
185 |
|
|
} |
186 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bts_l_ri,(RW4 r, IMM i)) |
187 |
|
|
|
188 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bts_l_rr,(RW4 r, R4 b)) |
189 |
|
|
{ |
190 |
|
|
emit_byte(0x0f); |
191 |
|
|
emit_byte(0xab); |
192 |
|
|
emit_byte(0xc0+8*b+r); |
193 |
|
|
} |
194 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bts_l_rr,(RW4 r, R4 b)) |
195 |
|
|
|
196 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_w_ri,(RW2 d, IMM i)) |
197 |
|
|
{ |
198 |
|
|
emit_byte(0x66); |
199 |
|
|
if (isbyte(i)) { |
200 |
|
|
emit_byte(0x83); |
201 |
|
|
emit_byte(0xe8+d); |
202 |
|
|
emit_byte(i); |
203 |
|
|
} |
204 |
|
|
else { |
205 |
|
|
emit_byte(0x81); |
206 |
|
|
emit_byte(0xe8+d); |
207 |
|
|
emit_word(i); |
208 |
|
|
} |
209 |
|
|
} |
210 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_w_ri,(RW2 d, IMM i)) |
211 |
|
|
|
212 |
|
|
|
213 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_l_rm,(W4 d, MEMR s)) |
214 |
|
|
{ |
215 |
|
|
emit_byte(0x8b); |
216 |
|
|
emit_byte(0x05+8*d); |
217 |
|
|
emit_long(s); |
218 |
|
|
} |
219 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_l_rm,(W4 d, MEMR s)) |
220 |
|
|
|
221 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_l_mi,(MEMW d, IMM s)) |
222 |
|
|
{ |
223 |
|
|
emit_byte(0xc7); |
224 |
|
|
emit_byte(0x05); |
225 |
|
|
emit_long(d); |
226 |
|
|
emit_long(s); |
227 |
|
|
} |
228 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_l_mi,(MEMW d, IMM s)) |
229 |
|
|
|
230 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_w_mi,(MEMW d, IMM s)) |
231 |
|
|
{ |
232 |
|
|
emit_byte(0x66); |
233 |
|
|
emit_byte(0xc7); |
234 |
|
|
emit_byte(0x05); |
235 |
|
|
emit_long(d); |
236 |
|
|
emit_word(s); |
237 |
|
|
} |
238 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_w_mi,(MEMW d, IMM s)) |
239 |
|
|
|
240 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_b_mi,(MEMW d, IMM s)) |
241 |
|
|
{ |
242 |
|
|
emit_byte(0xc6); |
243 |
|
|
emit_byte(0x05); |
244 |
|
|
emit_long(d); |
245 |
|
|
emit_byte(s); |
246 |
|
|
} |
247 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_b_mi,(MEMW d, IMM s)) |
248 |
|
|
|
249 |
|
|
LOWFUNC(WRITE,RMW,2,raw_rol_b_mi,(MEMRW d, IMM i)) |
250 |
|
|
{ |
251 |
|
|
if (optimize_shift_once && (i == 1)) { |
252 |
|
|
emit_byte(0xd0); |
253 |
|
|
emit_byte(0x05); |
254 |
|
|
emit_long(d); |
255 |
|
|
} |
256 |
|
|
else { |
257 |
|
|
emit_byte(0xc0); |
258 |
|
|
emit_byte(0x05); |
259 |
|
|
emit_long(d); |
260 |
|
|
emit_byte(i); |
261 |
|
|
} |
262 |
|
|
} |
263 |
|
|
LENDFUNC(WRITE,RMW,2,raw_rol_b_mi,(MEMRW d, IMM i)) |
264 |
|
|
|
265 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_b_ri,(RW1 r, IMM i)) |
266 |
|
|
{ |
267 |
|
|
if (optimize_shift_once && (i == 1)) { |
268 |
|
|
emit_byte(0xd0); |
269 |
|
|
emit_byte(0xc0+r); |
270 |
|
|
} |
271 |
|
|
else { |
272 |
|
|
emit_byte(0xc0); |
273 |
|
|
emit_byte(0xc0+r); |
274 |
|
|
emit_byte(i); |
275 |
|
|
} |
276 |
|
|
} |
277 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_b_ri,(RW1 r, IMM i)) |
278 |
|
|
|
279 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_w_ri,(RW2 r, IMM i)) |
280 |
|
|
{ |
281 |
|
|
emit_byte(0x66); |
282 |
|
|
emit_byte(0xc1); |
283 |
|
|
emit_byte(0xc0+r); |
284 |
|
|
emit_byte(i); |
285 |
|
|
} |
286 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_w_ri,(RW2 r, IMM i)) |
287 |
|
|
|
288 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_l_ri,(RW4 r, IMM i)) |
289 |
|
|
{ |
290 |
|
|
if (optimize_shift_once && (i == 1)) { |
291 |
|
|
emit_byte(0xd1); |
292 |
|
|
emit_byte(0xc0+r); |
293 |
|
|
} |
294 |
|
|
else { |
295 |
|
|
emit_byte(0xc1); |
296 |
|
|
emit_byte(0xc0+r); |
297 |
|
|
emit_byte(i); |
298 |
|
|
} |
299 |
|
|
} |
300 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_l_ri,(RW4 r, IMM i)) |
301 |
|
|
|
302 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_l_rr,(RW4 d, R1 r)) |
303 |
|
|
{ |
304 |
|
|
emit_byte(0xd3); |
305 |
|
|
emit_byte(0xc0+d); |
306 |
|
|
} |
307 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_l_rr,(RW4 d, R1 r)) |
308 |
|
|
|
309 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_w_rr,(RW2 d, R1 r)) |
310 |
|
|
{ |
311 |
|
|
emit_byte(0x66); |
312 |
|
|
emit_byte(0xd3); |
313 |
|
|
emit_byte(0xc0+d); |
314 |
|
|
} |
315 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_w_rr,(RW2 d, R1 r)) |
316 |
|
|
|
317 |
|
|
LOWFUNC(WRITE,NONE,2,raw_rol_b_rr,(RW1 d, R1 r)) |
318 |
|
|
{ |
319 |
|
|
emit_byte(0xd2); |
320 |
|
|
emit_byte(0xc0+d); |
321 |
|
|
} |
322 |
|
|
LENDFUNC(WRITE,NONE,2,raw_rol_b_rr,(RW1 d, R1 r)) |
323 |
|
|
|
324 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_l_rr,(RW4 d, R1 r)) |
325 |
|
|
{ |
326 |
|
|
emit_byte(0xd3); |
327 |
|
|
emit_byte(0xe0+d); |
328 |
|
|
} |
329 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_l_rr,(RW4 d, R1 r)) |
330 |
|
|
|
331 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_w_rr,(RW2 d, R1 r)) |
332 |
|
|
{ |
333 |
|
|
emit_byte(0x66); |
334 |
|
|
emit_byte(0xd3); |
335 |
|
|
emit_byte(0xe0+d); |
336 |
|
|
} |
337 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_w_rr,(RW2 d, R1 r)) |
338 |
|
|
|
339 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_b_rr,(RW1 d, R1 r)) |
340 |
|
|
{ |
341 |
|
|
emit_byte(0xd2); |
342 |
|
|
emit_byte(0xe0+d); |
343 |
|
|
} |
344 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_b_rr,(RW1 d, R1 r)) |
345 |
|
|
|
346 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_b_ri,(RW1 r, IMM i)) |
347 |
|
|
{ |
348 |
|
|
if (optimize_shift_once && (i == 1)) { |
349 |
|
|
emit_byte(0xd0); |
350 |
|
|
emit_byte(0xc8+r); |
351 |
|
|
} |
352 |
|
|
else { |
353 |
|
|
emit_byte(0xc0); |
354 |
|
|
emit_byte(0xc8+r); |
355 |
|
|
emit_byte(i); |
356 |
|
|
} |
357 |
|
|
} |
358 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_b_ri,(RW1 r, IMM i)) |
359 |
|
|
|
360 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_w_ri,(RW2 r, IMM i)) |
361 |
|
|
{ |
362 |
|
|
emit_byte(0x66); |
363 |
|
|
emit_byte(0xc1); |
364 |
|
|
emit_byte(0xc8+r); |
365 |
|
|
emit_byte(i); |
366 |
|
|
} |
367 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_w_ri,(RW2 r, IMM i)) |
368 |
|
|
|
369 |
|
|
// gb-- used for making an fpcr value in compemu_fpp.cpp |
370 |
|
|
LOWFUNC(WRITE,READ,2,raw_or_l_rm,(RW4 d, MEMR s)) |
371 |
|
|
{ |
372 |
|
|
emit_byte(0x0b); |
373 |
|
|
emit_byte(0x05+8*d); |
374 |
|
|
emit_long(s); |
375 |
|
|
} |
376 |
|
|
LENDFUNC(WRITE,READ,2,raw_or_l_rm,(RW4 d, MEMR s)) |
377 |
|
|
|
378 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_l_ri,(RW4 r, IMM i)) |
379 |
|
|
{ |
380 |
|
|
if (optimize_shift_once && (i == 1)) { |
381 |
|
|
emit_byte(0xd1); |
382 |
|
|
emit_byte(0xc8+r); |
383 |
|
|
} |
384 |
|
|
else { |
385 |
|
|
emit_byte(0xc1); |
386 |
|
|
emit_byte(0xc8+r); |
387 |
|
|
emit_byte(i); |
388 |
|
|
} |
389 |
|
|
} |
390 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_l_ri,(RW4 r, IMM i)) |
391 |
|
|
|
392 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_l_rr,(RW4 d, R1 r)) |
393 |
|
|
{ |
394 |
|
|
emit_byte(0xd3); |
395 |
|
|
emit_byte(0xc8+d); |
396 |
|
|
} |
397 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_l_rr,(RW4 d, R1 r)) |
398 |
|
|
|
399 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_w_rr,(RW2 d, R1 r)) |
400 |
|
|
{ |
401 |
|
|
emit_byte(0x66); |
402 |
|
|
emit_byte(0xd3); |
403 |
|
|
emit_byte(0xc8+d); |
404 |
|
|
} |
405 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_w_rr,(RW2 d, R1 r)) |
406 |
|
|
|
407 |
|
|
LOWFUNC(WRITE,NONE,2,raw_ror_b_rr,(RW1 d, R1 r)) |
408 |
|
|
{ |
409 |
|
|
emit_byte(0xd2); |
410 |
|
|
emit_byte(0xc8+d); |
411 |
|
|
} |
412 |
|
|
LENDFUNC(WRITE,NONE,2,raw_ror_b_rr,(RW1 d, R1 r)) |
413 |
|
|
|
414 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_l_rr,(RW4 d, R1 r)) |
415 |
|
|
{ |
416 |
|
|
emit_byte(0xd3); |
417 |
|
|
emit_byte(0xe8+d); |
418 |
|
|
} |
419 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_l_rr,(RW4 d, R1 r)) |
420 |
|
|
|
421 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_w_rr,(RW2 d, R1 r)) |
422 |
|
|
{ |
423 |
|
|
emit_byte(0x66); |
424 |
|
|
emit_byte(0xd3); |
425 |
|
|
emit_byte(0xe8+d); |
426 |
|
|
} |
427 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_w_rr,(RW2 d, R1 r)) |
428 |
|
|
|
429 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_b_rr,(RW1 d, R1 r)) |
430 |
|
|
{ |
431 |
|
|
emit_byte(0xd2); |
432 |
|
|
emit_byte(0xe8+d); |
433 |
|
|
} |
434 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_b_rr,(RW1 d, R1 r)) |
435 |
|
|
|
436 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_l_rr,(RW4 d, R1 r)) |
437 |
|
|
{ |
438 |
|
|
emit_byte(0xd3); |
439 |
|
|
emit_byte(0xf8+d); |
440 |
|
|
} |
441 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_l_rr,(RW4 d, R1 r)) |
442 |
|
|
|
443 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_w_rr,(RW2 d, R1 r)) |
444 |
|
|
{ |
445 |
|
|
emit_byte(0x66); |
446 |
|
|
emit_byte(0xd3); |
447 |
|
|
emit_byte(0xf8+d); |
448 |
|
|
} |
449 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_w_rr,(RW2 d, R1 r)) |
450 |
|
|
|
451 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_b_rr,(RW1 d, R1 r)) |
452 |
|
|
{ |
453 |
|
|
emit_byte(0xd2); |
454 |
|
|
emit_byte(0xf8+d); |
455 |
|
|
} |
456 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_b_rr,(RW1 d, R1 r)) |
457 |
|
|
|
458 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_l_ri,(RW4 r, IMM i)) |
459 |
|
|
{ |
460 |
|
|
if (optimize_shift_once && (i == 1)) { |
461 |
|
|
emit_byte(0xd1); |
462 |
|
|
emit_byte(0xe0+r); |
463 |
|
|
} |
464 |
|
|
else { |
465 |
|
|
emit_byte(0xc1); |
466 |
|
|
emit_byte(0xe0+r); |
467 |
|
|
emit_byte(i); |
468 |
|
|
} |
469 |
|
|
} |
470 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_l_ri,(RW4 r, IMM i)) |
471 |
|
|
|
472 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_w_ri,(RW2 r, IMM i)) |
473 |
|
|
{ |
474 |
|
|
emit_byte(0x66); |
475 |
|
|
emit_byte(0xc1); |
476 |
|
|
emit_byte(0xe0+r); |
477 |
|
|
emit_byte(i); |
478 |
|
|
} |
479 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_w_ri,(RW2 r, IMM i)) |
480 |
|
|
|
481 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shll_b_ri,(RW1 r, IMM i)) |
482 |
|
|
{ |
483 |
|
|
if (optimize_shift_once && (i == 1)) { |
484 |
|
|
emit_byte(0xd0); |
485 |
|
|
emit_byte(0xe0+r); |
486 |
|
|
} |
487 |
|
|
else { |
488 |
|
|
emit_byte(0xc0); |
489 |
|
|
emit_byte(0xe0+r); |
490 |
|
|
emit_byte(i); |
491 |
|
|
} |
492 |
|
|
} |
493 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shll_b_ri,(RW1 r, IMM i)) |
494 |
|
|
|
495 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_l_ri,(RW4 r, IMM i)) |
496 |
|
|
{ |
497 |
|
|
if (optimize_shift_once && (i == 1)) { |
498 |
|
|
emit_byte(0xd1); |
499 |
|
|
emit_byte(0xe8+r); |
500 |
|
|
} |
501 |
|
|
else { |
502 |
|
|
emit_byte(0xc1); |
503 |
|
|
emit_byte(0xe8+r); |
504 |
|
|
emit_byte(i); |
505 |
|
|
} |
506 |
|
|
} |
507 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_l_ri,(RW4 r, IMM i)) |
508 |
|
|
|
509 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_w_ri,(RW2 r, IMM i)) |
510 |
|
|
{ |
511 |
|
|
emit_byte(0x66); |
512 |
|
|
emit_byte(0xc1); |
513 |
|
|
emit_byte(0xe8+r); |
514 |
|
|
emit_byte(i); |
515 |
|
|
} |
516 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_w_ri,(RW2 r, IMM i)) |
517 |
|
|
|
518 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shrl_b_ri,(RW1 r, IMM i)) |
519 |
|
|
{ |
520 |
|
|
if (optimize_shift_once && (i == 1)) { |
521 |
|
|
emit_byte(0xd0); |
522 |
|
|
emit_byte(0xe8+r); |
523 |
|
|
} |
524 |
|
|
else { |
525 |
|
|
emit_byte(0xc0); |
526 |
|
|
emit_byte(0xe8+r); |
527 |
|
|
emit_byte(i); |
528 |
|
|
} |
529 |
|
|
} |
530 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shrl_b_ri,(RW1 r, IMM i)) |
531 |
|
|
|
532 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_l_ri,(RW4 r, IMM i)) |
533 |
|
|
{ |
534 |
|
|
if (optimize_shift_once && (i == 1)) { |
535 |
|
|
emit_byte(0xd1); |
536 |
|
|
emit_byte(0xf8+r); |
537 |
|
|
} |
538 |
|
|
else { |
539 |
|
|
emit_byte(0xc1); |
540 |
|
|
emit_byte(0xf8+r); |
541 |
|
|
emit_byte(i); |
542 |
|
|
} |
543 |
|
|
} |
544 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_l_ri,(RW4 r, IMM i)) |
545 |
|
|
|
546 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_w_ri,(RW2 r, IMM i)) |
547 |
|
|
{ |
548 |
|
|
emit_byte(0x66); |
549 |
|
|
emit_byte(0xc1); |
550 |
|
|
emit_byte(0xf8+r); |
551 |
|
|
emit_byte(i); |
552 |
|
|
} |
553 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_w_ri,(RW2 r, IMM i)) |
554 |
|
|
|
555 |
|
|
LOWFUNC(WRITE,NONE,2,raw_shra_b_ri,(RW1 r, IMM i)) |
556 |
|
|
{ |
557 |
|
|
if (optimize_shift_once && (i == 1)) { |
558 |
|
|
emit_byte(0xd0); |
559 |
|
|
emit_byte(0xf8+r); |
560 |
|
|
} |
561 |
|
|
else { |
562 |
|
|
emit_byte(0xc0); |
563 |
|
|
emit_byte(0xf8+r); |
564 |
|
|
emit_byte(i); |
565 |
|
|
} |
566 |
|
|
} |
567 |
|
|
LENDFUNC(WRITE,NONE,2,raw_shra_b_ri,(RW1 r, IMM i)) |
568 |
|
|
|
569 |
|
|
LOWFUNC(WRITE,NONE,1,raw_sahf,(R2 dummy_ah)) |
570 |
|
|
{ |
571 |
|
|
emit_byte(0x9e); |
572 |
|
|
} |
573 |
|
|
LENDFUNC(WRITE,NONE,1,raw_sahf,(R2 dummy_ah)) |
574 |
|
|
|
575 |
|
|
LOWFUNC(NONE,NONE,1,raw_cpuid,(R4 dummy_eax)) |
576 |
|
|
{ |
577 |
|
|
emit_byte(0x0f); |
578 |
|
|
emit_byte(0xa2); |
579 |
|
|
} |
580 |
|
|
LENDFUNC(NONE,NONE,1,raw_cpuid,(R4 dummy_eax)) |
581 |
|
|
|
582 |
|
|
LOWFUNC(READ,NONE,1,raw_lahf,(W2 dummy_ah)) |
583 |
|
|
{ |
584 |
|
|
emit_byte(0x9f); |
585 |
|
|
} |
586 |
|
|
LENDFUNC(READ,NONE,1,raw_lahf,(W2 dummy_ah)) |
587 |
|
|
|
588 |
|
|
LOWFUNC(READ,NONE,2,raw_setcc,(W1 d, IMM cc)) |
589 |
|
|
{ |
590 |
|
|
emit_byte(0x0f); |
591 |
|
|
emit_byte(0x90+cc); |
592 |
|
|
emit_byte(0xc0+d); |
593 |
|
|
} |
594 |
|
|
LENDFUNC(READ,NONE,2,raw_setcc,(W1 d, IMM cc)) |
595 |
|
|
|
596 |
|
|
LOWFUNC(READ,WRITE,2,raw_setcc_m,(MEMW d, IMM cc)) |
597 |
|
|
{ |
598 |
|
|
emit_byte(0x0f); |
599 |
|
|
emit_byte(0x90+cc); |
600 |
|
|
emit_byte(0x05); |
601 |
|
|
emit_long(d); |
602 |
|
|
} |
603 |
|
|
LENDFUNC(READ,WRITE,2,raw_setcc_m,(MEMW d, IMM cc)) |
604 |
|
|
|
605 |
|
|
LOWFUNC(READ,NONE,3,raw_cmov_l_rr,(RW4 d, R4 s, IMM cc)) |
606 |
|
|
{ |
607 |
|
|
if (have_cmov) { |
608 |
|
|
emit_byte(0x0f); |
609 |
|
|
emit_byte(0x40+cc); |
610 |
|
|
emit_byte(0xc0+8*d+s); |
611 |
|
|
} |
612 |
|
|
else { /* replacement using branch and mov */ |
613 |
|
|
int uncc=(cc^1); |
614 |
|
|
emit_byte(0x70+uncc); |
615 |
|
|
emit_byte(2); /* skip next 2 bytes if not cc=true */ |
616 |
|
|
emit_byte(0x89); |
617 |
|
|
emit_byte(0xc0+8*s+d); |
618 |
|
|
} |
619 |
|
|
} |
620 |
|
|
LENDFUNC(READ,NONE,3,raw_cmov_l_rr,(RW4 d, R4 s, IMM cc)) |
621 |
|
|
|
622 |
|
|
LOWFUNC(WRITE,NONE,2,raw_bsf_l_rr,(W4 d, R4 s)) |
623 |
|
|
{ |
624 |
|
|
emit_byte(0x0f); |
625 |
|
|
emit_byte(0xbc); |
626 |
|
|
emit_byte(0xc0+8*d+s); |
627 |
|
|
} |
628 |
|
|
LENDFUNC(WRITE,NONE,2,raw_bsf_l_rr,(W4 d, R4 s)) |
629 |
|
|
|
630 |
|
|
LOWFUNC(NONE,NONE,2,raw_sign_extend_16_rr,(W4 d, R2 s)) |
631 |
|
|
{ |
632 |
|
|
emit_byte(0x0f); |
633 |
|
|
emit_byte(0xbf); |
634 |
|
|
emit_byte(0xc0+8*d+s); |
635 |
|
|
} |
636 |
|
|
LENDFUNC(NONE,NONE,2,raw_sign_extend_16_rr,(W4 d, R2 s)) |
637 |
|
|
|
638 |
|
|
LOWFUNC(NONE,NONE,2,raw_sign_extend_8_rr,(W4 d, R1 s)) |
639 |
|
|
{ |
640 |
|
|
emit_byte(0x0f); |
641 |
|
|
emit_byte(0xbe); |
642 |
|
|
emit_byte(0xc0+8*d+s); |
643 |
|
|
} |
644 |
|
|
LENDFUNC(NONE,NONE,2,raw_sign_extend_8_rr,(W4 d, R1 s)) |
645 |
|
|
|
646 |
|
|
LOWFUNC(NONE,NONE,2,raw_zero_extend_16_rr,(W4 d, R2 s)) |
647 |
|
|
{ |
648 |
|
|
emit_byte(0x0f); |
649 |
|
|
emit_byte(0xb7); |
650 |
|
|
emit_byte(0xc0+8*d+s); |
651 |
|
|
} |
652 |
|
|
LENDFUNC(NONE,NONE,2,raw_zero_extend_16_rr,(W4 d, R2 s)) |
653 |
|
|
|
654 |
|
|
LOWFUNC(NONE,NONE,2,raw_zero_extend_8_rr,(W4 d, R1 s)) |
655 |
|
|
{ |
656 |
|
|
emit_byte(0x0f); |
657 |
|
|
emit_byte(0xb6); |
658 |
|
|
emit_byte(0xc0+8*d+s); |
659 |
|
|
} |
660 |
|
|
LENDFUNC(NONE,NONE,2,raw_zero_extend_8_rr,(W4 d, R1 s)) |
661 |
|
|
|
662 |
|
|
LOWFUNC(NONE,NONE,2,raw_imul_32_32,(RW4 d, R4 s)) |
663 |
|
|
{ |
664 |
|
|
emit_byte(0x0f); |
665 |
|
|
emit_byte(0xaf); |
666 |
|
|
emit_byte(0xc0+8*d+s); |
667 |
|
|
} |
668 |
|
|
LENDFUNC(NONE,NONE,2,raw_imul_32_32,(RW4 d, R4 s)) |
669 |
|
|
|
670 |
|
|
LOWFUNC(NONE,NONE,2,raw_imul_64_32,(RW4 d, RW4 s)) |
671 |
|
|
{ |
672 |
|
|
if (d!=MUL_NREG1 || s!=MUL_NREG2) |
673 |
|
|
abort(); |
674 |
|
|
emit_byte(0xf7); |
675 |
|
|
emit_byte(0xea); |
676 |
|
|
} |
677 |
|
|
LENDFUNC(NONE,NONE,2,raw_imul_64_32,(RW4 d, RW4 s)) |
678 |
|
|
|
679 |
|
|
LOWFUNC(NONE,NONE,2,raw_mul_64_32,(RW4 d, RW4 s)) |
680 |
|
|
{ |
681 |
|
|
if (d!=MUL_NREG1 || s!=MUL_NREG2) { |
682 |
|
|
printf("Bad register in MUL: d=%d, s=%d\n",d,s); |
683 |
|
|
abort(); |
684 |
|
|
} |
685 |
|
|
emit_byte(0xf7); |
686 |
|
|
emit_byte(0xe2); |
687 |
|
|
} |
688 |
|
|
LENDFUNC(NONE,NONE,2,raw_mul_64_32,(RW4 d, RW4 s)) |
689 |
|
|
|
690 |
|
|
LOWFUNC(NONE,NONE,2,raw_mul_32_32,(RW4 d, R4 s)) |
691 |
|
|
{ |
692 |
|
|
abort(); /* %^$&%^$%#^ x86! */ |
693 |
|
|
emit_byte(0x0f); |
694 |
|
|
emit_byte(0xaf); |
695 |
|
|
emit_byte(0xc0+8*d+s); |
696 |
|
|
} |
697 |
|
|
LENDFUNC(NONE,NONE,2,raw_mul_32_32,(RW4 d, R4 s)) |
698 |
|
|
|
699 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_b_rr,(W1 d, R1 s)) |
700 |
|
|
{ |
701 |
|
|
emit_byte(0x88); |
702 |
|
|
emit_byte(0xc0+8*s+d); |
703 |
|
|
} |
704 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_b_rr,(W1 d, R1 s)) |
705 |
|
|
|
706 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_w_rr,(W2 d, R2 s)) |
707 |
|
|
{ |
708 |
|
|
emit_byte(0x66); |
709 |
|
|
emit_byte(0x89); |
710 |
|
|
emit_byte(0xc0+8*s+d); |
711 |
|
|
} |
712 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_w_rr,(W2 d, R2 s)) |
713 |
|
|
|
714 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_l_rrm_indexed,(W4 d,R4 baser, R4 index, IMM factor)) |
715 |
|
|
{ |
716 |
|
|
int isebp=(baser==5)?0x40:0; |
717 |
|
|
int fi; |
718 |
|
|
|
719 |
|
|
switch(factor) { |
720 |
|
|
case 1: fi=0; break; |
721 |
|
|
case 2: fi=1; break; |
722 |
|
|
case 4: fi=2; break; |
723 |
|
|
case 8: fi=3; break; |
724 |
|
|
default: abort(); |
725 |
|
|
} |
726 |
|
|
|
727 |
|
|
|
728 |
|
|
emit_byte(0x8b); |
729 |
|
|
emit_byte(0x04+8*d+isebp); |
730 |
|
|
emit_byte(baser+8*index+0x40*fi); |
731 |
|
|
if (isebp) |
732 |
|
|
emit_byte(0x00); |
733 |
|
|
} |
734 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_l_rrm_indexed,(W4 d,R4 baser, R4 index, IMM factor)) |
735 |
|
|
|
736 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_w_rrm_indexed,(W2 d, R4 baser, R4 index, IMM factor)) |
737 |
|
|
{ |
738 |
|
|
int fi; |
739 |
|
|
int isebp; |
740 |
|
|
|
741 |
|
|
switch(factor) { |
742 |
|
|
case 1: fi=0; break; |
743 |
|
|
case 2: fi=1; break; |
744 |
|
|
case 4: fi=2; break; |
745 |
|
|
case 8: fi=3; break; |
746 |
|
|
default: abort(); |
747 |
|
|
} |
748 |
|
|
isebp=(baser==5)?0x40:0; |
749 |
|
|
|
750 |
|
|
emit_byte(0x66); |
751 |
|
|
emit_byte(0x8b); |
752 |
|
|
emit_byte(0x04+8*d+isebp); |
753 |
|
|
emit_byte(baser+8*index+0x40*fi); |
754 |
|
|
if (isebp) |
755 |
|
|
emit_byte(0x00); |
756 |
|
|
} |
757 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_w_rrm_indexed,(W2 d, R4 baser, R4 index, IMM factor)) |
758 |
|
|
|
759 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_b_rrm_indexed,(W1 d, R4 baser, R4 index, IMM factor)) |
760 |
|
|
{ |
761 |
|
|
int fi; |
762 |
|
|
int isebp; |
763 |
|
|
|
764 |
|
|
switch(factor) { |
765 |
|
|
case 1: fi=0; break; |
766 |
|
|
case 2: fi=1; break; |
767 |
|
|
case 4: fi=2; break; |
768 |
|
|
case 8: fi=3; break; |
769 |
|
|
default: abort(); |
770 |
|
|
} |
771 |
|
|
isebp=(baser==5)?0x40:0; |
772 |
|
|
|
773 |
|
|
emit_byte(0x8a); |
774 |
|
|
emit_byte(0x04+8*d+isebp); |
775 |
|
|
emit_byte(baser+8*index+0x40*fi); |
776 |
|
|
if (isebp) |
777 |
|
|
emit_byte(0x00); |
778 |
|
|
} |
779 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_b_rrm_indexed,(W1 d, R4 baser, R4 index, IMM factor)) |
780 |
|
|
|
781 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_l_mrr_indexed,(R4 baser, R4 index, IMM factor, R4 s)) |
782 |
|
|
{ |
783 |
|
|
int fi; |
784 |
|
|
int isebp; |
785 |
|
|
|
786 |
|
|
switch(factor) { |
787 |
|
|
case 1: fi=0; break; |
788 |
|
|
case 2: fi=1; break; |
789 |
|
|
case 4: fi=2; break; |
790 |
|
|
case 8: fi=3; break; |
791 |
|
|
default: abort(); |
792 |
|
|
} |
793 |
|
|
|
794 |
|
|
|
795 |
|
|
isebp=(baser==5)?0x40:0; |
796 |
|
|
|
797 |
|
|
emit_byte(0x89); |
798 |
|
|
emit_byte(0x04+8*s+isebp); |
799 |
|
|
emit_byte(baser+8*index+0x40*fi); |
800 |
|
|
if (isebp) |
801 |
|
|
emit_byte(0x00); |
802 |
|
|
} |
803 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_l_mrr_indexed,(R4 baser, R4 index, IMM factor, R4 s)) |
804 |
|
|
|
805 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_w_mrr_indexed,(R4 baser, R4 index, IMM factor, R2 s)) |
806 |
|
|
{ |
807 |
|
|
int fi; |
808 |
|
|
int isebp; |
809 |
|
|
|
810 |
|
|
switch(factor) { |
811 |
|
|
case 1: fi=0; break; |
812 |
|
|
case 2: fi=1; break; |
813 |
|
|
case 4: fi=2; break; |
814 |
|
|
case 8: fi=3; break; |
815 |
|
|
default: abort(); |
816 |
|
|
} |
817 |
|
|
isebp=(baser==5)?0x40:0; |
818 |
|
|
|
819 |
|
|
emit_byte(0x66); |
820 |
|
|
emit_byte(0x89); |
821 |
|
|
emit_byte(0x04+8*s+isebp); |
822 |
|
|
emit_byte(baser+8*index+0x40*fi); |
823 |
|
|
if (isebp) |
824 |
|
|
emit_byte(0x00); |
825 |
|
|
} |
826 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_w_mrr_indexed,(R4 baser, R4 index, IMM factor, R2 s)) |
827 |
|
|
|
828 |
|
|
LOWFUNC(NONE,WRITE,4,raw_mov_b_mrr_indexed,(R4 baser, R4 index, IMM factor, R1 s)) |
829 |
|
|
{ |
830 |
|
|
int fi; |
831 |
|
|
int isebp; |
832 |
|
|
|
833 |
|
|
switch(factor) { |
834 |
|
|
case 1: fi=0; break; |
835 |
|
|
case 2: fi=1; break; |
836 |
|
|
case 4: fi=2; break; |
837 |
|
|
case 8: fi=3; break; |
838 |
|
|
default: abort(); |
839 |
|
|
} |
840 |
|
|
isebp=(baser==5)?0x40:0; |
841 |
|
|
|
842 |
|
|
emit_byte(0x88); |
843 |
|
|
emit_byte(0x04+8*s+isebp); |
844 |
|
|
emit_byte(baser+8*index+0x40*fi); |
845 |
|
|
if (isebp) |
846 |
|
|
emit_byte(0x00); |
847 |
|
|
} |
848 |
|
|
LENDFUNC(NONE,WRITE,4,raw_mov_b_mrr_indexed,(R4 baser, R4 index, IMM factor, R1 s)) |
849 |
|
|
|
850 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_l_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R4 s)) |
851 |
|
|
{ |
852 |
|
|
int fi; |
853 |
|
|
|
854 |
|
|
switch(factor) { |
855 |
|
|
case 1: fi=0; break; |
856 |
|
|
case 2: fi=1; break; |
857 |
|
|
case 4: fi=2; break; |
858 |
|
|
case 8: fi=3; break; |
859 |
|
|
default: abort(); |
860 |
|
|
} |
861 |
|
|
|
862 |
|
|
emit_byte(0x89); |
863 |
|
|
emit_byte(0x84+8*s); |
864 |
|
|
emit_byte(baser+8*index+0x40*fi); |
865 |
|
|
emit_long(base); |
866 |
|
|
} |
867 |
|
|
LENDFUNC(NONE,WRITE,5,raw_mov_l_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R4 s)) |
868 |
|
|
|
869 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_w_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R2 s)) |
870 |
|
|
{ |
871 |
|
|
int fi; |
872 |
|
|
|
873 |
|
|
switch(factor) { |
874 |
|
|
case 1: fi=0; break; |
875 |
|
|
case 2: fi=1; break; |
876 |
|
|
case 4: fi=2; break; |
877 |
|
|
case 8: fi=3; break; |
878 |
|
|
default: abort(); |
879 |
|
|
} |
880 |
|
|
|
881 |
|
|
emit_byte(0x66); |
882 |
|
|
emit_byte(0x89); |
883 |
|
|
emit_byte(0x84+8*s); |
884 |
|
|
emit_byte(baser+8*index+0x40*fi); |
885 |
|
|
emit_long(base); |
886 |
|
|
} |
887 |
|
|
LENDFUNC(NONE,WRITE,5,raw_mov_w_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R2 s)) |
888 |
|
|
|
889 |
|
|
LOWFUNC(NONE,WRITE,5,raw_mov_b_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R1 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(0x88); |
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_b_bmrr_indexed,(IMM base, R4 baser, R4 index, IMM factor, R1 s)) |
907 |
|
|
|
908 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_l_brrm_indexed,(W4 d, IMM base, R4 baser, R4 index, IMM factor)) |
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(0x8b); |
921 |
|
|
emit_byte(0x84+8*d); |
922 |
|
|
emit_byte(baser+8*index+0x40*fi); |
923 |
|
|
emit_long(base); |
924 |
|
|
} |
925 |
|
|
LENDFUNC(NONE,READ,5,raw_mov_l_brrm_indexed,(W4 d, IMM base, R4 baser, R4 index, IMM factor)) |
926 |
|
|
|
927 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_w_brrm_indexed,(W2 d, IMM base, R4 baser, R4 index, IMM factor)) |
928 |
|
|
{ |
929 |
|
|
int fi; |
930 |
|
|
|
931 |
|
|
switch(factor) { |
932 |
|
|
case 1: fi=0; break; |
933 |
|
|
case 2: fi=1; break; |
934 |
|
|
case 4: fi=2; break; |
935 |
|
|
case 8: fi=3; break; |
936 |
|
|
default: abort(); |
937 |
|
|
} |
938 |
|
|
|
939 |
|
|
emit_byte(0x66); |
940 |
|
|
emit_byte(0x8b); |
941 |
|
|
emit_byte(0x84+8*d); |
942 |
|
|
emit_byte(baser+8*index+0x40*fi); |
943 |
|
|
emit_long(base); |
944 |
|
|
} |
945 |
|
|
LENDFUNC(NONE,READ,5,raw_mov_w_brrm_indexed,(W2 d, IMM base, R4 baser, R4 index, IMM factor)) |
946 |
|
|
|
947 |
|
|
LOWFUNC(NONE,READ,5,raw_mov_b_brrm_indexed,(W1 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(0x8a); |
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_b_brrm_indexed,(W1 d, IMM base, R4 baser, R4 index, IMM factor)) |
965 |
|
|
|
966 |
|
|
LOWFUNC(NONE,READ,4,raw_mov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor)) |
967 |
|
|
{ |
968 |
|
|
int fi; |
969 |
|
|
switch(factor) { |
970 |
|
|
case 1: fi=0; break; |
971 |
|
|
case 2: fi=1; break; |
972 |
|
|
case 4: fi=2; break; |
973 |
|
|
case 8: fi=3; break; |
974 |
|
|
default: |
975 |
|
|
fprintf(stderr,"Bad factor %d in mov_l_rm_indexed!\n",factor); |
976 |
|
|
abort(); |
977 |
|
|
} |
978 |
|
|
emit_byte(0x8b); |
979 |
|
|
emit_byte(0x04+8*d); |
980 |
|
|
emit_byte(0x05+8*index+64*fi); |
981 |
|
|
emit_long(base); |
982 |
|
|
} |
983 |
|
|
LENDFUNC(NONE,READ,4,raw_mov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor)) |
984 |
|
|
|
985 |
|
|
LOWFUNC(NONE,READ,5,raw_cmov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor, IMM cond)) |
986 |
|
|
{ |
987 |
|
|
int fi; |
988 |
|
|
switch(factor) { |
989 |
|
|
case 1: fi=0; break; |
990 |
|
|
case 2: fi=1; break; |
991 |
|
|
case 4: fi=2; break; |
992 |
|
|
case 8: fi=3; break; |
993 |
|
|
default: |
994 |
|
|
fprintf(stderr,"Bad factor %d in mov_l_rm_indexed!\n",factor); |
995 |
|
|
abort(); |
996 |
|
|
} |
997 |
|
|
if (have_cmov) { |
998 |
|
|
emit_byte(0x0f); |
999 |
|
|
emit_byte(0x40+cond); |
1000 |
|
|
emit_byte(0x04+8*d); |
1001 |
|
|
emit_byte(0x05+8*index+64*fi); |
1002 |
|
|
emit_long(base); |
1003 |
|
|
} |
1004 |
|
|
else { /* replacement using branch and mov */ |
1005 |
|
|
int uncc=(cond^1); |
1006 |
|
|
emit_byte(0x70+uncc); |
1007 |
|
|
emit_byte(7); /* skip next 7 bytes if not cc=true */ |
1008 |
|
|
emit_byte(0x8b); |
1009 |
|
|
emit_byte(0x04+8*d); |
1010 |
|
|
emit_byte(0x05+8*index+64*fi); |
1011 |
|
|
emit_long(base); |
1012 |
|
|
} |
1013 |
|
|
} |
1014 |
|
|
LENDFUNC(NONE,READ,5,raw_cmov_l_rm_indexed,(W4 d, IMM base, R4 index, IMM factor, IMM cond)) |
1015 |
|
|
|
1016 |
|
|
LOWFUNC(NONE,READ,3,raw_cmov_l_rm,(W4 d, IMM mem, IMM cond)) |
1017 |
|
|
{ |
1018 |
|
|
if (have_cmov) { |
1019 |
|
|
emit_byte(0x0f); |
1020 |
|
|
emit_byte(0x40+cond); |
1021 |
|
|
emit_byte(0x05+8*d); |
1022 |
|
|
emit_long(mem); |
1023 |
|
|
} |
1024 |
|
|
else { /* replacement using branch and mov */ |
1025 |
|
|
int uncc=(cond^1); |
1026 |
|
|
emit_byte(0x70+uncc); |
1027 |
|
|
emit_byte(6); /* skip next 6 bytes if not cc=true */ |
1028 |
|
|
emit_byte(0x8b); |
1029 |
|
|
emit_byte(0x05+8*d); |
1030 |
|
|
emit_long(mem); |
1031 |
|
|
} |
1032 |
|
|
} |
1033 |
|
|
LENDFUNC(NONE,READ,3,raw_cmov_l_rm,(W4 d, IMM mem, IMM cond)) |
1034 |
|
|
|
1035 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
1036 |
|
|
{ |
1037 |
|
|
emit_byte(0x8b); |
1038 |
|
|
emit_byte(0x40+8*d+s); |
1039 |
|
|
emit_byte(offset); |
1040 |
|
|
} |
1041 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_l_rR,(W4 d, R4 s, IMM offset)) |
1042 |
|
|
|
1043 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
1044 |
|
|
{ |
1045 |
|
|
emit_byte(0x66); |
1046 |
|
|
emit_byte(0x8b); |
1047 |
|
|
emit_byte(0x40+8*d+s); |
1048 |
|
|
emit_byte(offset); |
1049 |
|
|
} |
1050 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_w_rR,(W2 d, R4 s, IMM offset)) |
1051 |
|
|
|
1052 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
1053 |
|
|
{ |
1054 |
|
|
emit_byte(0x8a); |
1055 |
|
|
emit_byte(0x40+8*d+s); |
1056 |
|
|
emit_byte(offset); |
1057 |
|
|
} |
1058 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_b_rR,(W1 d, R4 s, IMM offset)) |
1059 |
|
|
|
1060 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_l_brR,(W4 d, R4 s, IMM offset)) |
1061 |
|
|
{ |
1062 |
|
|
emit_byte(0x8b); |
1063 |
|
|
emit_byte(0x80+8*d+s); |
1064 |
|
|
emit_long(offset); |
1065 |
|
|
} |
1066 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_l_brR,(W4 d, R4 s, IMM offset)) |
1067 |
|
|
|
1068 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_w_brR,(W2 d, R4 s, IMM offset)) |
1069 |
|
|
{ |
1070 |
|
|
emit_byte(0x66); |
1071 |
|
|
emit_byte(0x8b); |
1072 |
|
|
emit_byte(0x80+8*d+s); |
1073 |
|
|
emit_long(offset); |
1074 |
|
|
} |
1075 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_w_brR,(W2 d, R4 s, IMM offset)) |
1076 |
|
|
|
1077 |
|
|
LOWFUNC(NONE,READ,3,raw_mov_b_brR,(W1 d, R4 s, IMM offset)) |
1078 |
|
|
{ |
1079 |
|
|
emit_byte(0x8a); |
1080 |
|
|
emit_byte(0x80+8*d+s); |
1081 |
|
|
emit_long(offset); |
1082 |
|
|
} |
1083 |
|
|
LENDFUNC(NONE,READ,3,raw_mov_b_brR,(W1 d, R4 s, IMM offset)) |
1084 |
|
|
|
1085 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
1086 |
|
|
{ |
1087 |
|
|
emit_byte(0xc7); |
1088 |
|
|
emit_byte(0x40+d); |
1089 |
|
|
emit_byte(offset); |
1090 |
|
|
emit_long(i); |
1091 |
|
|
} |
1092 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_Ri,(R4 d, IMM i, IMM offset)) |
1093 |
|
|
|
1094 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
1095 |
|
|
{ |
1096 |
|
|
emit_byte(0x66); |
1097 |
|
|
emit_byte(0xc7); |
1098 |
|
|
emit_byte(0x40+d); |
1099 |
|
|
emit_byte(offset); |
1100 |
|
|
emit_word(i); |
1101 |
|
|
} |
1102 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_Ri,(R4 d, IMM i, IMM offset)) |
1103 |
|
|
|
1104 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
1105 |
|
|
{ |
1106 |
|
|
emit_byte(0xc6); |
1107 |
|
|
emit_byte(0x40+d); |
1108 |
|
|
emit_byte(offset); |
1109 |
|
|
emit_byte(i); |
1110 |
|
|
} |
1111 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_Ri,(R4 d, IMM i, IMM offset)) |
1112 |
|
|
|
1113 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
1114 |
|
|
{ |
1115 |
|
|
emit_byte(0x89); |
1116 |
|
|
emit_byte(0x40+8*s+d); |
1117 |
|
|
emit_byte(offset); |
1118 |
|
|
} |
1119 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_Rr,(R4 d, R4 s, IMM offset)) |
1120 |
|
|
|
1121 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
1122 |
|
|
{ |
1123 |
|
|
emit_byte(0x66); |
1124 |
|
|
emit_byte(0x89); |
1125 |
|
|
emit_byte(0x40+8*s+d); |
1126 |
|
|
emit_byte(offset); |
1127 |
|
|
} |
1128 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_Rr,(R4 d, R2 s, IMM offset)) |
1129 |
|
|
|
1130 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
1131 |
|
|
{ |
1132 |
|
|
emit_byte(0x88); |
1133 |
|
|
emit_byte(0x40+8*s+d); |
1134 |
|
|
emit_byte(offset); |
1135 |
|
|
} |
1136 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_Rr,(R4 d, R1 s, IMM offset)) |
1137 |
|
|
|
1138 |
|
|
LOWFUNC(NONE,NONE,3,raw_lea_l_brr,(W4 d, R4 s, IMM offset)) |
1139 |
|
|
{ |
1140 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1141 |
|
|
emit_byte(0x8d); |
1142 |
|
|
emit_byte(0x40+8*d+s); |
1143 |
|
|
emit_byte(offset); |
1144 |
|
|
} |
1145 |
|
|
else { |
1146 |
|
|
emit_byte(0x8d); |
1147 |
|
|
emit_byte(0x80+8*d+s); |
1148 |
|
|
emit_long(offset); |
1149 |
|
|
} |
1150 |
|
|
} |
1151 |
|
|
LENDFUNC(NONE,NONE,3,raw_lea_l_brr,(W4 d, R4 s, IMM offset)) |
1152 |
|
|
|
1153 |
|
|
LOWFUNC(NONE,NONE,5,raw_lea_l_brr_indexed,(W4 d, R4 s, R4 index, IMM factor, IMM offset)) |
1154 |
|
|
{ |
1155 |
|
|
int fi; |
1156 |
|
|
|
1157 |
|
|
switch(factor) { |
1158 |
|
|
case 1: fi=0; break; |
1159 |
|
|
case 2: fi=1; break; |
1160 |
|
|
case 4: fi=2; break; |
1161 |
|
|
case 8: fi=3; break; |
1162 |
|
|
default: abort(); |
1163 |
|
|
} |
1164 |
|
|
|
1165 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1166 |
|
|
emit_byte(0x8d); |
1167 |
|
|
emit_byte(0x44+8*d); |
1168 |
|
|
emit_byte(0x40*fi+8*index+s); |
1169 |
|
|
emit_byte(offset); |
1170 |
|
|
} |
1171 |
|
|
else { |
1172 |
|
|
emit_byte(0x8d); |
1173 |
|
|
emit_byte(0x84+8*d); |
1174 |
|
|
emit_byte(0x40*fi+8*index+s); |
1175 |
|
|
emit_long(offset); |
1176 |
|
|
} |
1177 |
|
|
} |
1178 |
|
|
LENDFUNC(NONE,NONE,5,raw_lea_l_brr_indexed,(W4 d, R4 s, R4 index, IMM factor, IMM offset)) |
1179 |
|
|
|
1180 |
|
|
LOWFUNC(NONE,NONE,4,raw_lea_l_rr_indexed,(W4 d, R4 s, R4 index, IMM factor)) |
1181 |
|
|
{ |
1182 |
|
|
int isebp=(s==5)?0x40:0; |
1183 |
|
|
int fi; |
1184 |
|
|
|
1185 |
|
|
switch(factor) { |
1186 |
|
|
case 1: fi=0; break; |
1187 |
|
|
case 2: fi=1; break; |
1188 |
|
|
case 4: fi=2; break; |
1189 |
|
|
case 8: fi=3; break; |
1190 |
|
|
default: abort(); |
1191 |
|
|
} |
1192 |
|
|
|
1193 |
|
|
emit_byte(0x8d); |
1194 |
|
|
emit_byte(0x04+8*d+isebp); |
1195 |
|
|
emit_byte(0x40*fi+8*index+s); |
1196 |
|
|
if (isebp) |
1197 |
|
|
emit_byte(0); |
1198 |
|
|
} |
1199 |
|
|
LENDFUNC(NONE,NONE,4,raw_lea_l_rr_indexed,(W4 d, R4 s, R4 index, IMM factor)) |
1200 |
|
|
|
1201 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_l_bRr,(R4 d, R4 s, IMM offset)) |
1202 |
|
|
{ |
1203 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1204 |
|
|
emit_byte(0x89); |
1205 |
|
|
emit_byte(0x40+8*s+d); |
1206 |
|
|
emit_byte(offset); |
1207 |
|
|
} |
1208 |
|
|
else { |
1209 |
|
|
emit_byte(0x89); |
1210 |
|
|
emit_byte(0x80+8*s+d); |
1211 |
|
|
emit_long(offset); |
1212 |
|
|
} |
1213 |
|
|
} |
1214 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_l_bRr,(R4 d, R4 s, IMM offset)) |
1215 |
|
|
|
1216 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_w_bRr,(R4 d, R2 s, IMM offset)) |
1217 |
|
|
{ |
1218 |
|
|
emit_byte(0x66); |
1219 |
|
|
emit_byte(0x89); |
1220 |
|
|
emit_byte(0x80+8*s+d); |
1221 |
|
|
emit_long(offset); |
1222 |
|
|
} |
1223 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_w_bRr,(R4 d, R2 s, IMM offset)) |
1224 |
|
|
|
1225 |
|
|
LOWFUNC(NONE,WRITE,3,raw_mov_b_bRr,(R4 d, R1 s, IMM offset)) |
1226 |
|
|
{ |
1227 |
|
|
if (optimize_imm8 && isbyte(offset)) { |
1228 |
|
|
emit_byte(0x88); |
1229 |
|
|
emit_byte(0x40+8*s+d); |
1230 |
|
|
emit_byte(offset); |
1231 |
|
|
} |
1232 |
|
|
else { |
1233 |
|
|
emit_byte(0x88); |
1234 |
|
|
emit_byte(0x80+8*s+d); |
1235 |
|
|
emit_long(offset); |
1236 |
|
|
} |
1237 |
|
|
} |
1238 |
|
|
LENDFUNC(NONE,WRITE,3,raw_mov_b_bRr,(R4 d, R1 s, IMM offset)) |
1239 |
|
|
|
1240 |
|
|
LOWFUNC(NONE,NONE,1,raw_bswap_32,(RW4 r)) |
1241 |
|
|
{ |
1242 |
|
|
emit_byte(0x0f); |
1243 |
|
|
emit_byte(0xc8+r); |
1244 |
|
|
} |
1245 |
|
|
LENDFUNC(NONE,NONE,1,raw_bswap_32,(RW4 r)) |
1246 |
|
|
|
1247 |
|
|
LOWFUNC(WRITE,NONE,1,raw_bswap_16,(RW2 r)) |
1248 |
|
|
{ |
1249 |
|
|
emit_byte(0x66); |
1250 |
|
|
emit_byte(0xc1); |
1251 |
|
|
emit_byte(0xc0+r); |
1252 |
|
|
emit_byte(0x08); |
1253 |
|
|
} |
1254 |
|
|
LENDFUNC(WRITE,NONE,1,raw_bswap_16,(RW2 r)) |
1255 |
|
|
|
1256 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_l_rr,(W4 d, R4 s)) |
1257 |
|
|
{ |
1258 |
|
|
emit_byte(0x89); |
1259 |
|
|
emit_byte(0xc0+8*s+d); |
1260 |
|
|
} |
1261 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_l_rr,(W4 d, R4 s)) |
1262 |
|
|
|
1263 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_l_mr,(IMM d, R4 s)) |
1264 |
|
|
{ |
1265 |
|
|
emit_byte(0x89); |
1266 |
|
|
emit_byte(0x05+8*s); |
1267 |
|
|
emit_long(d); |
1268 |
|
|
} |
1269 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_l_mr,(IMM d, R4 s)) |
1270 |
|
|
|
1271 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_w_mr,(IMM d, R2 s)) |
1272 |
|
|
{ |
1273 |
|
|
emit_byte(0x66); |
1274 |
|
|
emit_byte(0x89); |
1275 |
|
|
emit_byte(0x05+8*s); |
1276 |
|
|
emit_long(d); |
1277 |
|
|
} |
1278 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_w_mr,(IMM d, R2 s)) |
1279 |
|
|
|
1280 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_w_rm,(W2 d, IMM s)) |
1281 |
|
|
{ |
1282 |
|
|
emit_byte(0x66); |
1283 |
|
|
emit_byte(0x8b); |
1284 |
|
|
emit_byte(0x05+8*d); |
1285 |
|
|
emit_long(s); |
1286 |
|
|
} |
1287 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_w_rm,(W2 d, IMM s)) |
1288 |
|
|
|
1289 |
|
|
LOWFUNC(NONE,WRITE,2,raw_mov_b_mr,(IMM d, R1 s)) |
1290 |
|
|
{ |
1291 |
|
|
emit_byte(0x88); |
1292 |
|
|
emit_byte(0x05+8*s); |
1293 |
|
|
emit_long(d); |
1294 |
|
|
} |
1295 |
|
|
LENDFUNC(NONE,WRITE,2,raw_mov_b_mr,(IMM d, R1 s)) |
1296 |
|
|
|
1297 |
|
|
LOWFUNC(NONE,READ,2,raw_mov_b_rm,(W1 d, IMM s)) |
1298 |
|
|
{ |
1299 |
|
|
emit_byte(0x8a); |
1300 |
|
|
emit_byte(0x05+8*d); |
1301 |
|
|
emit_long(s); |
1302 |
|
|
} |
1303 |
|
|
LENDFUNC(NONE,READ,2,raw_mov_b_rm,(W1 d, IMM s)) |
1304 |
|
|
|
1305 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_l_ri,(W4 d, IMM s)) |
1306 |
|
|
{ |
1307 |
|
|
emit_byte(0xb8+d); |
1308 |
|
|
emit_long(s); |
1309 |
|
|
} |
1310 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_l_ri,(W4 d, IMM s)) |
1311 |
|
|
|
1312 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_w_ri,(W2 d, IMM s)) |
1313 |
|
|
{ |
1314 |
|
|
emit_byte(0x66); |
1315 |
|
|
emit_byte(0xb8+d); |
1316 |
|
|
emit_word(s); |
1317 |
|
|
} |
1318 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_w_ri,(W2 d, IMM s)) |
1319 |
|
|
|
1320 |
|
|
LOWFUNC(NONE,NONE,2,raw_mov_b_ri,(W1 d, IMM s)) |
1321 |
|
|
{ |
1322 |
|
|
emit_byte(0xb0+d); |
1323 |
|
|
emit_byte(s); |
1324 |
|
|
} |
1325 |
|
|
LENDFUNC(NONE,NONE,2,raw_mov_b_ri,(W1 d, IMM s)) |
1326 |
|
|
|
1327 |
|
|
LOWFUNC(RMW,RMW,2,raw_adc_l_mi,(MEMRW d, IMM s)) |
1328 |
|
|
{ |
1329 |
|
|
emit_byte(0x81); |
1330 |
|
|
emit_byte(0x15); |
1331 |
|
|
emit_long(d); |
1332 |
|
|
emit_long(s); |
1333 |
|
|
} |
1334 |
|
|
LENDFUNC(RMW,RMW,2,raw_adc_l_mi,(MEMRW d, IMM s)) |
1335 |
|
|
|
1336 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_l_mi,(IMM d, IMM s)) |
1337 |
|
|
{ |
1338 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1339 |
|
|
emit_byte(0x83); |
1340 |
|
|
emit_byte(0x05); |
1341 |
|
|
emit_long(d); |
1342 |
|
|
emit_byte(s); |
1343 |
|
|
} |
1344 |
|
|
else { |
1345 |
|
|
emit_byte(0x81); |
1346 |
|
|
emit_byte(0x05); |
1347 |
|
|
emit_long(d); |
1348 |
|
|
emit_long(s); |
1349 |
|
|
} |
1350 |
|
|
} |
1351 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_l_mi,(IMM d, IMM s)) |
1352 |
|
|
|
1353 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_w_mi,(IMM d, IMM s)) |
1354 |
|
|
{ |
1355 |
|
|
emit_byte(0x66); |
1356 |
|
|
emit_byte(0x81); |
1357 |
|
|
emit_byte(0x05); |
1358 |
|
|
emit_long(d); |
1359 |
|
|
emit_word(s); |
1360 |
|
|
} |
1361 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_w_mi,(IMM d, IMM s)) |
1362 |
|
|
|
1363 |
|
|
LOWFUNC(WRITE,RMW,2,raw_add_b_mi,(IMM d, IMM s)) |
1364 |
|
|
{ |
1365 |
|
|
emit_byte(0x80); |
1366 |
|
|
emit_byte(0x05); |
1367 |
|
|
emit_long(d); |
1368 |
|
|
emit_byte(s); |
1369 |
|
|
} |
1370 |
|
|
LENDFUNC(WRITE,RMW,2,raw_add_b_mi,(IMM d, IMM s)) |
1371 |
|
|
|
1372 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_l_ri,(R4 d, IMM i)) |
1373 |
|
|
{ |
1374 |
|
|
emit_byte(0xf7); |
1375 |
|
|
emit_byte(0xc0+d); |
1376 |
|
|
emit_long(i); |
1377 |
|
|
} |
1378 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_l_ri,(R4 d, IMM i)) |
1379 |
|
|
|
1380 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_l_rr,(R4 d, R4 s)) |
1381 |
|
|
{ |
1382 |
|
|
emit_byte(0x85); |
1383 |
|
|
emit_byte(0xc0+8*s+d); |
1384 |
|
|
} |
1385 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_l_rr,(R4 d, R4 s)) |
1386 |
|
|
|
1387 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_w_rr,(R2 d, R2 s)) |
1388 |
|
|
{ |
1389 |
|
|
emit_byte(0x66); |
1390 |
|
|
emit_byte(0x85); |
1391 |
|
|
emit_byte(0xc0+8*s+d); |
1392 |
|
|
} |
1393 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_w_rr,(R2 d, R2 s)) |
1394 |
|
|
|
1395 |
|
|
LOWFUNC(WRITE,NONE,2,raw_test_b_rr,(R1 d, R1 s)) |
1396 |
|
|
{ |
1397 |
|
|
emit_byte(0x84); |
1398 |
|
|
emit_byte(0xc0+8*s+d); |
1399 |
|
|
} |
1400 |
|
|
LENDFUNC(WRITE,NONE,2,raw_test_b_rr,(R1 d, R1 s)) |
1401 |
|
|
|
1402 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_l_ri,(RW4 d, IMM i)) |
1403 |
|
|
{ |
1404 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1405 |
|
|
emit_byte(0x83); |
1406 |
|
|
emit_byte(0xe0+d); |
1407 |
|
|
emit_byte(i); |
1408 |
|
|
} |
1409 |
|
|
else { |
1410 |
|
|
emit_byte(0x81); |
1411 |
|
|
emit_byte(0xe0+d); |
1412 |
|
|
emit_long(i); |
1413 |
|
|
} |
1414 |
|
|
} |
1415 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_l_ri,(RW4 d, IMM i)) |
1416 |
|
|
|
1417 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_w_ri,(RW2 d, IMM i)) |
1418 |
|
|
{ |
1419 |
|
|
emit_byte(0x66); |
1420 |
|
|
emit_byte(0x81); |
1421 |
|
|
emit_byte(0xe0+d); |
1422 |
|
|
emit_word(i); |
1423 |
|
|
} |
1424 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_w_ri,(RW2 d, IMM i)) |
1425 |
|
|
|
1426 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_l,(RW4 d, R4 s)) |
1427 |
|
|
{ |
1428 |
|
|
emit_byte(0x21); |
1429 |
|
|
emit_byte(0xc0+8*s+d); |
1430 |
|
|
} |
1431 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_l,(RW4 d, R4 s)) |
1432 |
|
|
|
1433 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_w,(RW2 d, R2 s)) |
1434 |
|
|
{ |
1435 |
|
|
emit_byte(0x66); |
1436 |
|
|
emit_byte(0x21); |
1437 |
|
|
emit_byte(0xc0+8*s+d); |
1438 |
|
|
} |
1439 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_w,(RW2 d, R2 s)) |
1440 |
|
|
|
1441 |
|
|
LOWFUNC(WRITE,NONE,2,raw_and_b,(RW1 d, R1 s)) |
1442 |
|
|
{ |
1443 |
|
|
emit_byte(0x20); |
1444 |
|
|
emit_byte(0xc0+8*s+d); |
1445 |
|
|
} |
1446 |
|
|
LENDFUNC(WRITE,NONE,2,raw_and_b,(RW1 d, R1 s)) |
1447 |
|
|
|
1448 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_l_ri,(RW4 d, IMM i)) |
1449 |
|
|
{ |
1450 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1451 |
|
|
emit_byte(0x83); |
1452 |
|
|
emit_byte(0xc8+d); |
1453 |
|
|
emit_byte(i); |
1454 |
|
|
} |
1455 |
|
|
else { |
1456 |
|
|
emit_byte(0x81); |
1457 |
|
|
emit_byte(0xc8+d); |
1458 |
|
|
emit_long(i); |
1459 |
|
|
} |
1460 |
|
|
} |
1461 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_l_ri,(RW4 d, IMM i)) |
1462 |
|
|
|
1463 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_l,(RW4 d, R4 s)) |
1464 |
|
|
{ |
1465 |
|
|
emit_byte(0x09); |
1466 |
|
|
emit_byte(0xc0+8*s+d); |
1467 |
|
|
} |
1468 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_l,(RW4 d, R4 s)) |
1469 |
|
|
|
1470 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_w,(RW2 d, R2 s)) |
1471 |
|
|
{ |
1472 |
|
|
emit_byte(0x66); |
1473 |
|
|
emit_byte(0x09); |
1474 |
|
|
emit_byte(0xc0+8*s+d); |
1475 |
|
|
} |
1476 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_w,(RW2 d, R2 s)) |
1477 |
|
|
|
1478 |
|
|
LOWFUNC(WRITE,NONE,2,raw_or_b,(RW1 d, R1 s)) |
1479 |
|
|
{ |
1480 |
|
|
emit_byte(0x08); |
1481 |
|
|
emit_byte(0xc0+8*s+d); |
1482 |
|
|
} |
1483 |
|
|
LENDFUNC(WRITE,NONE,2,raw_or_b,(RW1 d, R1 s)) |
1484 |
|
|
|
1485 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_l,(RW4 d, R4 s)) |
1486 |
|
|
{ |
1487 |
|
|
emit_byte(0x11); |
1488 |
|
|
emit_byte(0xc0+8*s+d); |
1489 |
|
|
} |
1490 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_l,(RW4 d, R4 s)) |
1491 |
|
|
|
1492 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_w,(RW2 d, R2 s)) |
1493 |
|
|
{ |
1494 |
|
|
emit_byte(0x66); |
1495 |
|
|
emit_byte(0x11); |
1496 |
|
|
emit_byte(0xc0+8*s+d); |
1497 |
|
|
} |
1498 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_w,(RW2 d, R2 s)) |
1499 |
|
|
|
1500 |
|
|
LOWFUNC(RMW,NONE,2,raw_adc_b,(RW1 d, R1 s)) |
1501 |
|
|
{ |
1502 |
|
|
emit_byte(0x10); |
1503 |
|
|
emit_byte(0xc0+8*s+d); |
1504 |
|
|
} |
1505 |
|
|
LENDFUNC(RMW,NONE,2,raw_adc_b,(RW1 d, R1 s)) |
1506 |
|
|
|
1507 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_l,(RW4 d, R4 s)) |
1508 |
|
|
{ |
1509 |
|
|
emit_byte(0x01); |
1510 |
|
|
emit_byte(0xc0+8*s+d); |
1511 |
|
|
} |
1512 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_l,(RW4 d, R4 s)) |
1513 |
|
|
|
1514 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_w,(RW2 d, R2 s)) |
1515 |
|
|
{ |
1516 |
|
|
emit_byte(0x66); |
1517 |
|
|
emit_byte(0x01); |
1518 |
|
|
emit_byte(0xc0+8*s+d); |
1519 |
|
|
} |
1520 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_w,(RW2 d, R2 s)) |
1521 |
|
|
|
1522 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_b,(RW1 d, R1 s)) |
1523 |
|
|
{ |
1524 |
|
|
emit_byte(0x00); |
1525 |
|
|
emit_byte(0xc0+8*s+d); |
1526 |
|
|
} |
1527 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_b,(RW1 d, R1 s)) |
1528 |
|
|
|
1529 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_l_ri,(RW4 d, IMM i)) |
1530 |
|
|
{ |
1531 |
|
|
if (isbyte(i)) { |
1532 |
|
|
emit_byte(0x83); |
1533 |
|
|
emit_byte(0xe8+d); |
1534 |
|
|
emit_byte(i); |
1535 |
|
|
} |
1536 |
|
|
else { |
1537 |
|
|
emit_byte(0x81); |
1538 |
|
|
emit_byte(0xe8+d); |
1539 |
|
|
emit_long(i); |
1540 |
|
|
} |
1541 |
|
|
} |
1542 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_l_ri,(RW4 d, IMM i)) |
1543 |
|
|
|
1544 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_b_ri,(RW1 d, IMM i)) |
1545 |
|
|
{ |
1546 |
|
|
emit_byte(0x80); |
1547 |
|
|
emit_byte(0xe8+d); |
1548 |
|
|
emit_byte(i); |
1549 |
|
|
} |
1550 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_b_ri,(RW1 d, IMM i)) |
1551 |
|
|
|
1552 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_l_ri,(RW4 d, IMM i)) |
1553 |
|
|
{ |
1554 |
|
|
if (isbyte(i)) { |
1555 |
|
|
emit_byte(0x83); |
1556 |
|
|
emit_byte(0xc0+d); |
1557 |
|
|
emit_byte(i); |
1558 |
|
|
} |
1559 |
|
|
else { |
1560 |
|
|
emit_byte(0x81); |
1561 |
|
|
emit_byte(0xc0+d); |
1562 |
|
|
emit_long(i); |
1563 |
|
|
} |
1564 |
|
|
} |
1565 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_l_ri,(RW4 d, IMM i)) |
1566 |
|
|
|
1567 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_w_ri,(RW2 d, IMM i)) |
1568 |
|
|
{ |
1569 |
|
|
if (isbyte(i)) { |
1570 |
|
|
emit_byte(0x66); |
1571 |
|
|
emit_byte(0x83); |
1572 |
|
|
emit_byte(0xc0+d); |
1573 |
|
|
emit_byte(i); |
1574 |
|
|
} |
1575 |
|
|
else { |
1576 |
|
|
emit_byte(0x66); |
1577 |
|
|
emit_byte(0x81); |
1578 |
|
|
emit_byte(0xc0+d); |
1579 |
|
|
emit_word(i); |
1580 |
|
|
} |
1581 |
|
|
} |
1582 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_w_ri,(RW2 d, IMM i)) |
1583 |
|
|
|
1584 |
|
|
LOWFUNC(WRITE,NONE,2,raw_add_b_ri,(RW1 d, IMM i)) |
1585 |
|
|
{ |
1586 |
|
|
emit_byte(0x80); |
1587 |
|
|
emit_byte(0xc0+d); |
1588 |
|
|
emit_byte(i); |
1589 |
|
|
} |
1590 |
|
|
LENDFUNC(WRITE,NONE,2,raw_add_b_ri,(RW1 d, IMM i)) |
1591 |
|
|
|
1592 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_l,(RW4 d, R4 s)) |
1593 |
|
|
{ |
1594 |
|
|
emit_byte(0x19); |
1595 |
|
|
emit_byte(0xc0+8*s+d); |
1596 |
|
|
} |
1597 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_l,(RW4 d, R4 s)) |
1598 |
|
|
|
1599 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_w,(RW2 d, R2 s)) |
1600 |
|
|
{ |
1601 |
|
|
emit_byte(0x66); |
1602 |
|
|
emit_byte(0x19); |
1603 |
|
|
emit_byte(0xc0+8*s+d); |
1604 |
|
|
} |
1605 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_w,(RW2 d, R2 s)) |
1606 |
|
|
|
1607 |
|
|
LOWFUNC(RMW,NONE,2,raw_sbb_b,(RW1 d, R1 s)) |
1608 |
|
|
{ |
1609 |
|
|
emit_byte(0x18); |
1610 |
|
|
emit_byte(0xc0+8*s+d); |
1611 |
|
|
} |
1612 |
|
|
LENDFUNC(RMW,NONE,2,raw_sbb_b,(RW1 d, R1 s)) |
1613 |
|
|
|
1614 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_l,(RW4 d, R4 s)) |
1615 |
|
|
{ |
1616 |
|
|
emit_byte(0x29); |
1617 |
|
|
emit_byte(0xc0+8*s+d); |
1618 |
|
|
} |
1619 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_l,(RW4 d, R4 s)) |
1620 |
|
|
|
1621 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_w,(RW2 d, R2 s)) |
1622 |
|
|
{ |
1623 |
|
|
emit_byte(0x66); |
1624 |
|
|
emit_byte(0x29); |
1625 |
|
|
emit_byte(0xc0+8*s+d); |
1626 |
|
|
} |
1627 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_w,(RW2 d, R2 s)) |
1628 |
|
|
|
1629 |
|
|
LOWFUNC(WRITE,NONE,2,raw_sub_b,(RW1 d, R1 s)) |
1630 |
|
|
{ |
1631 |
|
|
emit_byte(0x28); |
1632 |
|
|
emit_byte(0xc0+8*s+d); |
1633 |
|
|
} |
1634 |
|
|
LENDFUNC(WRITE,NONE,2,raw_sub_b,(RW1 d, R1 s)) |
1635 |
|
|
|
1636 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_l,(R4 d, R4 s)) |
1637 |
|
|
{ |
1638 |
|
|
emit_byte(0x39); |
1639 |
|
|
emit_byte(0xc0+8*s+d); |
1640 |
|
|
} |
1641 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_l,(R4 d, R4 s)) |
1642 |
|
|
|
1643 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_l_ri,(R4 r, IMM i)) |
1644 |
|
|
{ |
1645 |
|
|
if (optimize_imm8 && isbyte(i)) { |
1646 |
|
|
emit_byte(0x83); |
1647 |
|
|
emit_byte(0xf8+r); |
1648 |
|
|
emit_byte(i); |
1649 |
|
|
} |
1650 |
|
|
else { |
1651 |
|
|
emit_byte(0x81); |
1652 |
|
|
emit_byte(0xf8+r); |
1653 |
|
|
emit_long(i); |
1654 |
|
|
} |
1655 |
|
|
} |
1656 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_l_ri,(R4 r, IMM i)) |
1657 |
|
|
|
1658 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_w,(R2 d, R2 s)) |
1659 |
|
|
{ |
1660 |
|
|
emit_byte(0x66); |
1661 |
|
|
emit_byte(0x39); |
1662 |
|
|
emit_byte(0xc0+8*s+d); |
1663 |
|
|
} |
1664 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_w,(R2 d, R2 s)) |
1665 |
|
|
|
1666 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_b_ri,(R1 d, IMM i)) |
1667 |
|
|
{ |
1668 |
|
|
emit_byte(0x80); |
1669 |
|
|
emit_byte(0xf8+d); |
1670 |
|
|
emit_byte(i); |
1671 |
|
|
} |
1672 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_b_ri,(R1 d, IMM i)) |
1673 |
|
|
|
1674 |
|
|
LOWFUNC(WRITE,NONE,2,raw_cmp_b,(R1 d, R1 s)) |
1675 |
|
|
{ |
1676 |
|
|
emit_byte(0x38); |
1677 |
|
|
emit_byte(0xc0+8*s+d); |
1678 |
|
|
} |
1679 |
|
|
LENDFUNC(WRITE,NONE,2,raw_cmp_b,(R1 d, R1 s)) |
1680 |
|
|
|
1681 |
|
|
LOWFUNC(WRITE,READ,4,raw_cmp_l_rm_indexed,(R4 d, IMM offset, R4 index, IMM factor)) |
1682 |
|
|
{ |
1683 |
|
|
int fi; |
1684 |
|
|
|
1685 |
|
|
switch(factor) { |
1686 |
|
|
case 1: fi=0; break; |
1687 |
|
|
case 2: fi=1; break; |
1688 |
|
|
case 4: fi=2; break; |
1689 |
|
|
case 8: fi=3; break; |
1690 |
|
|
default: abort(); |
1691 |
|
|
} |
1692 |
|
|
emit_byte(0x39); |
1693 |
|
|
emit_byte(0x04+8*d); |
1694 |
|
|
emit_byte(5+8*index+0x40*fi); |
1695 |
|
|
emit_long(offset); |
1696 |
|
|
} |
1697 |
|
|
LENDFUNC(WRITE,READ,4,raw_cmp_l_rm_indexed,(R4 d, IMM offset, R4 index, IMM factor)) |
1698 |
|
|
|
1699 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_l,(RW4 d, R4 s)) |
1700 |
|
|
{ |
1701 |
|
|
emit_byte(0x31); |
1702 |
|
|
emit_byte(0xc0+8*s+d); |
1703 |
|
|
} |
1704 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_l,(RW4 d, R4 s)) |
1705 |
|
|
|
1706 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_w,(RW2 d, R2 s)) |
1707 |
|
|
{ |
1708 |
|
|
emit_byte(0x66); |
1709 |
|
|
emit_byte(0x31); |
1710 |
|
|
emit_byte(0xc0+8*s+d); |
1711 |
|
|
} |
1712 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_w,(RW2 d, R2 s)) |
1713 |
|
|
|
1714 |
|
|
LOWFUNC(WRITE,NONE,2,raw_xor_b,(RW1 d, R1 s)) |
1715 |
|
|
{ |
1716 |
|
|
emit_byte(0x30); |
1717 |
|
|
emit_byte(0xc0+8*s+d); |
1718 |
|
|
} |
1719 |
|
|
LENDFUNC(WRITE,NONE,2,raw_xor_b,(RW1 d, R1 s)) |
1720 |
|
|
|
1721 |
|
|
LOWFUNC(WRITE,RMW,2,raw_sub_l_mi,(MEMRW d, IMM s)) |
1722 |
|
|
{ |
1723 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1724 |
|
|
emit_byte(0x83); |
1725 |
|
|
emit_byte(0x2d); |
1726 |
|
|
emit_long(d); |
1727 |
|
|
emit_byte(s); |
1728 |
|
|
} |
1729 |
|
|
else { |
1730 |
|
|
emit_byte(0x81); |
1731 |
|
|
emit_byte(0x2d); |
1732 |
|
|
emit_long(d); |
1733 |
|
|
emit_long(s); |
1734 |
|
|
} |
1735 |
|
|
} |
1736 |
|
|
LENDFUNC(WRITE,RMW,2,raw_sub_l_mi,(MEMRW d, IMM s)) |
1737 |
|
|
|
1738 |
|
|
LOWFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1739 |
|
|
{ |
1740 |
|
|
if (optimize_imm8 && isbyte(s)) { |
1741 |
|
|
emit_byte(0x83); |
1742 |
|
|
emit_byte(0x3d); |
1743 |
|
|
emit_long(d); |
1744 |
|
|
emit_byte(s); |
1745 |
|
|
} |
1746 |
|
|
else { |
1747 |
|
|
emit_byte(0x81); |
1748 |
|
|
emit_byte(0x3d); |
1749 |
|
|
emit_long(d); |
1750 |
|
|
emit_long(s); |
1751 |
|
|
} |
1752 |
|
|
} |
1753 |
|
|
LENDFUNC(WRITE,READ,2,raw_cmp_l_mi,(MEMR d, IMM s)) |
1754 |
|
|
|
1755 |
|
|
LOWFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
1756 |
|
|
{ |
1757 |
|
|
emit_byte(0x87); |
1758 |
|
|
emit_byte(0xc0+8*r1+r2); |
1759 |
|
|
} |
1760 |
|
|
LENDFUNC(NONE,NONE,2,raw_xchg_l_rr,(RW4 r1, RW4 r2)) |
1761 |
|
|
|
1762 |
|
|
/************************************************************************* |
1763 |
|
|
* FIXME: string-related instructions * |
1764 |
|
|
*************************************************************************/ |
1765 |
|
|
|
1766 |
|
|
LOWFUNC(WRITE,NONE,0,raw_cld,(void)) |
1767 |
|
|
{ |
1768 |
|
|
emit_byte(0xfc); |
1769 |
|
|
} |
1770 |
|
|
LENDFUNC(WRITE,NONE,0,raw_cld,(void)) |
1771 |
|
|
|
1772 |
|
|
LOWFUNC(WRITE,NONE,0,raw_std,(void)) |
1773 |
|
|
{ |
1774 |
|
|
emit_byte(0xfd); |
1775 |
|
|
} |
1776 |
|
|
LENDFUNC(WRITE,NONE,0,raw_std,(void)) |
1777 |
|
|
|
1778 |
|
|
LOWFUNC(NONE,RMW,0,raw_movs_b,(void)) |
1779 |
|
|
{ |
1780 |
|
|
emit_byte(0xa4); |
1781 |
|
|
} |
1782 |
|
|
LENDFUNC(NONE,RMW,0,raw_movs_b,(void)) |
1783 |
|
|
|
1784 |
|
|
LOWFUNC(NONE,RMW,0,raw_movs_l,(void)) |
1785 |
|
|
{ |
1786 |
|
|
emit_byte(0xa5); |
1787 |
|
|
} |
1788 |
|
|
LENDFUNC(NONE,RMW,0,raw_movs_l,(void)) |
1789 |
|
|
|
1790 |
|
|
LOWFUNC(NONE,RMW,0,raw_rep,(void)) |
1791 |
|
|
{ |
1792 |
|
|
emit_byte(0xf3); |
1793 |
|
|
} |
1794 |
|
|
LENDFUNC(NONE,RMW,0,raw_rep,(void)) |
1795 |
|
|
|
1796 |
|
|
LOWFUNC(NONE,RMW,0,raw_rep_movsb,(void)) |
1797 |
|
|
{ |
1798 |
|
|
raw_rep(); |
1799 |
|
|
raw_movs_b(); |
1800 |
|
|
} |
1801 |
|
|
LENDFUNC(NONE,RMW,0,raw_rep_movsb,(void)) |
1802 |
|
|
|
1803 |
|
|
LOWFUNC(NONE,RMW,0,raw_rep_movsl,(void)) |
1804 |
|
|
{ |
1805 |
|
|
raw_rep(); |
1806 |
|
|
raw_movs_l(); |
1807 |
|
|
} |
1808 |
|
|
LENDFUNC(NONE,RMW,0,raw_rep_movsl,(void)) |
1809 |
|
|
|
1810 |
|
|
/************************************************************************* |
1811 |
|
|
* FIXME: mem access modes probably wrong * |
1812 |
|
|
*************************************************************************/ |
1813 |
|
|
|
1814 |
|
|
LOWFUNC(READ,WRITE,0,raw_pushfl,(void)) |
1815 |
|
|
{ |
1816 |
|
|
emit_byte(0x9c); |
1817 |
|
|
} |
1818 |
|
|
LENDFUNC(READ,WRITE,0,raw_pushfl,(void)) |
1819 |
|
|
|
1820 |
|
|
LOWFUNC(WRITE,READ,0,raw_popfl,(void)) |
1821 |
|
|
{ |
1822 |
|
|
emit_byte(0x9d); |
1823 |
|
|
} |
1824 |
|
|
LENDFUNC(WRITE,READ,0,raw_popfl,(void)) |
1825 |
|
|
|
1826 |
|
|
/************************************************************************* |
1827 |
|
|
* Unoptimizable stuff --- jump * |
1828 |
|
|
*************************************************************************/ |
1829 |
|
|
|
1830 |
|
|
static __inline__ void raw_call_r(R4 r) |
1831 |
|
|
{ |
1832 |
|
|
emit_byte(0xff); |
1833 |
|
|
emit_byte(0xd0+r); |
1834 |
|
|
} |
1835 |
|
|
|
1836 |
|
|
static __inline__ void raw_jmp_r(R4 r) |
1837 |
|
|
{ |
1838 |
|
|
emit_byte(0xff); |
1839 |
|
|
emit_byte(0xe0+r); |
1840 |
|
|
} |
1841 |
|
|
|
1842 |
|
|
static __inline__ void raw_jmp_m_indexed(uae_u32 base, uae_u32 r, uae_u32 m) |
1843 |
|
|
{ |
1844 |
|
|
int mu; |
1845 |
|
|
switch(m) { |
1846 |
|
|
case 1: mu=0; break; |
1847 |
|
|
case 2: mu=1; break; |
1848 |
|
|
case 4: mu=2; break; |
1849 |
|
|
case 8: mu=3; break; |
1850 |
|
|
default: abort(); |
1851 |
|
|
} |
1852 |
|
|
emit_byte(0xff); |
1853 |
|
|
emit_byte(0x24); |
1854 |
|
|
emit_byte(0x05+8*r+0x40*mu); |
1855 |
|
|
emit_long(base); |
1856 |
|
|
} |
1857 |
|
|
|
1858 |
|
|
static __inline__ void raw_jmp_m(uae_u32 base) |
1859 |
|
|
{ |
1860 |
|
|
emit_byte(0xff); |
1861 |
|
|
emit_byte(0x25); |
1862 |
|
|
emit_long(base); |
1863 |
|
|
} |
1864 |
|
|
|
1865 |
|
|
|
1866 |
|
|
static __inline__ void raw_call(uae_u32 t) |
1867 |
|
|
{ |
1868 |
|
|
emit_byte(0xe8); |
1869 |
|
|
emit_long(t-(uae_u32)target-4); |
1870 |
|
|
} |
1871 |
|
|
|
1872 |
|
|
static __inline__ void raw_jmp(uae_u32 t) |
1873 |
|
|
{ |
1874 |
|
|
emit_byte(0xe9); |
1875 |
|
|
emit_long(t-(uae_u32)target-4); |
1876 |
|
|
} |
1877 |
|
|
|
1878 |
|
|
static __inline__ void raw_jl(uae_u32 t) |
1879 |
|
|
{ |
1880 |
|
|
emit_byte(0x0f); |
1881 |
|
|
emit_byte(0x8c); |
1882 |
|
|
emit_long(t-(uae_u32)target-4); |
1883 |
|
|
} |
1884 |
|
|
|
1885 |
|
|
static __inline__ void raw_jz(uae_u32 t) |
1886 |
|
|
{ |
1887 |
|
|
emit_byte(0x0f); |
1888 |
|
|
emit_byte(0x84); |
1889 |
|
|
emit_long(t-(uae_u32)target-4); |
1890 |
|
|
} |
1891 |
|
|
|
1892 |
|
|
static __inline__ void raw_jnz(uae_u32 t) |
1893 |
|
|
{ |
1894 |
|
|
emit_byte(0x0f); |
1895 |
|
|
emit_byte(0x85); |
1896 |
|
|
emit_long(t-(uae_u32)target-4); |
1897 |
|
|
} |
1898 |
|
|
|
1899 |
|
|
static __inline__ void raw_jnz_l_oponly(void) |
1900 |
|
|
{ |
1901 |
|
|
emit_byte(0x0f); |
1902 |
|
|
emit_byte(0x85); |
1903 |
|
|
} |
1904 |
|
|
|
1905 |
|
|
static __inline__ void raw_jcc_l_oponly(int cc) |
1906 |
|
|
{ |
1907 |
|
|
emit_byte(0x0f); |
1908 |
|
|
emit_byte(0x80+cc); |
1909 |
|
|
} |
1910 |
|
|
|
1911 |
|
|
static __inline__ void raw_jnz_b_oponly(void) |
1912 |
|
|
{ |
1913 |
|
|
emit_byte(0x75); |
1914 |
|
|
} |
1915 |
|
|
|
1916 |
|
|
static __inline__ void raw_jz_b_oponly(void) |
1917 |
|
|
{ |
1918 |
|
|
emit_byte(0x74); |
1919 |
|
|
} |
1920 |
|
|
|
1921 |
|
|
static __inline__ void raw_jcc_b_oponly(int cc) |
1922 |
|
|
{ |
1923 |
|
|
emit_byte(0x70+cc); |
1924 |
|
|
} |
1925 |
|
|
|
1926 |
|
|
static __inline__ void raw_jmp_l_oponly(void) |
1927 |
|
|
{ |
1928 |
|
|
emit_byte(0xe9); |
1929 |
|
|
} |
1930 |
|
|
|
1931 |
|
|
static __inline__ void raw_jmp_b_oponly(void) |
1932 |
|
|
{ |
1933 |
|
|
emit_byte(0xeb); |
1934 |
|
|
} |
1935 |
|
|
|
1936 |
|
|
static __inline__ void raw_ret(void) |
1937 |
|
|
{ |
1938 |
|
|
emit_byte(0xc3); |
1939 |
|
|
} |
1940 |
|
|
|
1941 |
|
|
static __inline__ void raw_nop(void) |
1942 |
|
|
{ |
1943 |
|
|
emit_byte(0x90); |
1944 |
|
|
} |
1945 |
|
|
|
1946 |
|
|
|
1947 |
|
|
/************************************************************************* |
1948 |
|
|
* Flag handling, to and fro UAE flag register * |
1949 |
|
|
*************************************************************************/ |
1950 |
|
|
|
1951 |
|
|
#ifdef SAHF_SETO_PROFITABLE |
1952 |
|
|
|
1953 |
|
|
#define FLAG_NREG1 0 /* Set to -1 if any register will do */ |
1954 |
|
|
|
1955 |
|
|
static __inline__ void raw_flags_to_reg(int r) |
1956 |
|
|
{ |
1957 |
|
|
raw_lahf(0); /* Most flags in AH */ |
1958 |
|
|
//raw_setcc(r,0); /* V flag in AL */ |
1959 |
|
|
raw_setcc_m((uae_u32)live.state[FLAGTMP].mem,0); |
1960 |
|
|
|
1961 |
|
|
#if 1 /* Let's avoid those nasty partial register stalls */ |
1962 |
|
|
//raw_mov_b_mr((uae_u32)live.state[FLAGTMP].mem,r); |
1963 |
|
|
raw_mov_b_mr(((uae_u32)live.state[FLAGTMP].mem)+1,r+4); |
1964 |
|
|
//live.state[FLAGTMP].status=CLEAN; |
1965 |
|
|
live.state[FLAGTMP].status=INMEM; |
1966 |
|
|
live.state[FLAGTMP].realreg=-1; |
1967 |
|
|
/* We just "evicted" FLAGTMP. */ |
1968 |
|
|
if (live.nat[r].nholds!=1) { |
1969 |
|
|
/* Huh? */ |
1970 |
|
|
abort(); |
1971 |
|
|
} |
1972 |
|
|
live.nat[r].nholds=0; |
1973 |
|
|
#endif |
1974 |
|
|
} |
1975 |
|
|
|
1976 |
|
|
#define FLAG_NREG2 0 /* Set to -1 if any register will do */ |
1977 |
|
|
static __inline__ void raw_reg_to_flags(int r) |
1978 |
|
|
{ |
1979 |
|
|
raw_cmp_b_ri(r,-127); /* set V */ |
1980 |
|
|
raw_sahf(0); |
1981 |
|
|
} |
1982 |
|
|
|
1983 |
|
|
#else |
1984 |
|
|
|
1985 |
|
|
#define FLAG_NREG1 -1 /* Set to -1 if any register will do */ |
1986 |
|
|
static __inline__ void raw_flags_to_reg(int r) |
1987 |
|
|
{ |
1988 |
|
|
raw_pushfl(); |
1989 |
|
|
raw_pop_l_r(r); |
1990 |
|
|
raw_mov_l_mr((uae_u32)live.state[FLAGTMP].mem,r); |
1991 |
|
|
// live.state[FLAGTMP].status=CLEAN; |
1992 |
|
|
live.state[FLAGTMP].status=INMEM; |
1993 |
|
|
live.state[FLAGTMP].realreg=-1; |
1994 |
|
|
/* We just "evicted" FLAGTMP. */ |
1995 |
|
|
if (live.nat[r].nholds!=1) { |
1996 |
|
|
/* Huh? */ |
1997 |
|
|
abort(); |
1998 |
|
|
} |
1999 |
|
|
live.nat[r].nholds=0; |
2000 |
|
|
} |
2001 |
|
|
|
2002 |
|
|
#define FLAG_NREG2 -1 /* Set to -1 if any register will do */ |
2003 |
|
|
static __inline__ void raw_reg_to_flags(int r) |
2004 |
|
|
{ |
2005 |
|
|
raw_push_l_r(r); |
2006 |
|
|
raw_popfl(); |
2007 |
|
|
} |
2008 |
|
|
|
2009 |
|
|
#endif |
2010 |
|
|
|
2011 |
|
|
/* Apparently, there are enough instructions between flag store and |
2012 |
|
|
flag reload to avoid the partial memory stall */ |
2013 |
|
|
static __inline__ void raw_load_flagreg(uae_u32 target, uae_u32 r) |
2014 |
|
|
{ |
2015 |
|
|
#if 1 |
2016 |
|
|
raw_mov_l_rm(target,(uae_u32)live.state[r].mem); |
2017 |
|
|
#else |
2018 |
|
|
raw_mov_b_rm(target,(uae_u32)live.state[r].mem); |
2019 |
|
|
raw_mov_b_rm(target+4,((uae_u32)live.state[r].mem)+1); |
2020 |
|
|
#endif |
2021 |
|
|
} |
2022 |
|
|
|
2023 |
|
|
/* FLAGX is byte sized, and we *do* write it at that size */ |
2024 |
|
|
static __inline__ void raw_load_flagx(uae_u32 target, uae_u32 r) |
2025 |
|
|
{ |
2026 |
|
|
if (live.nat[target].canbyte) |
2027 |
|
|
raw_mov_b_rm(target,(uae_u32)live.state[r].mem); |
2028 |
|
|
else if (live.nat[target].canword) |
2029 |
|
|
raw_mov_w_rm(target,(uae_u32)live.state[r].mem); |
2030 |
|
|
else |
2031 |
|
|
raw_mov_l_rm(target,(uae_u32)live.state[r].mem); |
2032 |
|
|
} |
2033 |
|
|
|
2034 |
|
|
|
2035 |
|
|
static __inline__ void raw_inc_sp(int off) |
2036 |
|
|
{ |
2037 |
|
|
raw_add_l_ri(4,off); |
2038 |
|
|
} |
2039 |
|
|
|
2040 |
|
|
/************************************************************************* |
2041 |
|
|
* Handling mistaken direct memory access * |
2042 |
|
|
*************************************************************************/ |
2043 |
|
|
|
2044 |
|
|
// gb-- I don't need that part for JIT Basilisk II |
2045 |
|
|
#if defined(NATMEM_OFFSET) && 0 |
2046 |
|
|
#include <asm/sigcontext.h> |
2047 |
|
|
#include <signal.h> |
2048 |
|
|
|
2049 |
|
|
#define SIG_READ 1 |
2050 |
|
|
#define SIG_WRITE 2 |
2051 |
|
|
|
2052 |
|
|
static int in_handler=0; |
2053 |
|
|
static uae_u8 veccode[256]; |
2054 |
|
|
|
2055 |
|
|
static void vec(int x, struct sigcontext sc) |
2056 |
|
|
{ |
2057 |
|
|
uae_u8* i=(uae_u8*)sc.eip; |
2058 |
|
|
uae_u32 addr=sc.cr2; |
2059 |
|
|
int r=-1; |
2060 |
|
|
int size=4; |
2061 |
|
|
int dir=-1; |
2062 |
|
|
int len=0; |
2063 |
|
|
int j; |
2064 |
|
|
|
2065 |
|
|
write_log("fault address is %08x at %08x\n",sc.cr2,sc.eip); |
2066 |
|
|
if (!canbang) |
2067 |
|
|
write_log("Not happy! Canbang is 0 in SIGSEGV handler!\n"); |
2068 |
|
|
if (in_handler) |
2069 |
|
|
write_log("Argh --- Am already in a handler. Shouldn't happen!\n"); |
2070 |
|
|
|
2071 |
|
|
if (canbang && i>=compiled_code && i<=current_compile_p) { |
2072 |
|
|
if (*i==0x66) { |
2073 |
|
|
i++; |
2074 |
|
|
size=2; |
2075 |
|
|
len++; |
2076 |
|
|
} |
2077 |
|
|
|
2078 |
|
|
switch(i[0]) { |
2079 |
|
|
case 0x8a: |
2080 |
|
|
if ((i[1]&0xc0)==0x80) { |
2081 |
|
|
r=(i[1]>>3)&7; |
2082 |
|
|
dir=SIG_READ; |
2083 |
|
|
size=1; |
2084 |
|
|
len+=6; |
2085 |
|
|
break; |
2086 |
|
|
} |
2087 |
|
|
break; |
2088 |
|
|
case 0x88: |
2089 |
|
|
if ((i[1]&0xc0)==0x80) { |
2090 |
|
|
r=(i[1]>>3)&7; |
2091 |
|
|
dir=SIG_WRITE; |
2092 |
|
|
size=1; |
2093 |
|
|
len+=6; |
2094 |
|
|
break; |
2095 |
|
|
} |
2096 |
|
|
break; |
2097 |
|
|
case 0x8b: |
2098 |
|
|
if ((i[1]&0xc0)==0x80) { |
2099 |
|
|
r=(i[1]>>3)&7; |
2100 |
|
|
dir=SIG_READ; |
2101 |
|
|
len+=6; |
2102 |
|
|
break; |
2103 |
|
|
} |
2104 |
|
|
if ((i[1]&0xc0)==0x40) { |
2105 |
|
|
r=(i[1]>>3)&7; |
2106 |
|
|
dir=SIG_READ; |
2107 |
|
|
len+=3; |
2108 |
|
|
break; |
2109 |
|
|
} |
2110 |
|
|
break; |
2111 |
|
|
case 0x89: |
2112 |
|
|
if ((i[1]&0xc0)==0x80) { |
2113 |
|
|
r=(i[1]>>3)&7; |
2114 |
|
|
dir=SIG_WRITE; |
2115 |
|
|
len+=6; |
2116 |
|
|
break; |
2117 |
|
|
} |
2118 |
|
|
if ((i[1]&0xc0)==0x40) { |
2119 |
|
|
r=(i[1]>>3)&7; |
2120 |
|
|
dir=SIG_WRITE; |
2121 |
|
|
len+=3; |
2122 |
|
|
break; |
2123 |
|
|
} |
2124 |
|
|
break; |
2125 |
|
|
} |
2126 |
|
|
} |
2127 |
|
|
|
2128 |
|
|
if (r!=-1) { |
2129 |
|
|
void* pr=NULL; |
2130 |
|
|
write_log("register was %d, direction was %d, size was %d\n",r,dir,size); |
2131 |
|
|
|
2132 |
|
|
switch(r) { |
2133 |
|
|
case 0: pr=&(sc.eax); break; |
2134 |
|
|
case 1: pr=&(sc.ecx); break; |
2135 |
|
|
case 2: pr=&(sc.edx); break; |
2136 |
|
|
case 3: pr=&(sc.ebx); break; |
2137 |
|
|
case 4: pr=(size>1)?NULL:(((uae_u8*)&(sc.eax))+1); break; |
2138 |
|
|
case 5: pr=(size>1)? |
2139 |
|
|
(void*)(&(sc.ebp)): |
2140 |
|
|
(void*)(((uae_u8*)&(sc.ecx))+1); break; |
2141 |
|
|
case 6: pr=(size>1)? |
2142 |
|
|
(void*)(&(sc.esi)): |
2143 |
|
|
(void*)(((uae_u8*)&(sc.edx))+1); break; |
2144 |
|
|
case 7: pr=(size>1)? |
2145 |
|
|
(void*)(&(sc.edi)): |
2146 |
|
|
(void*)(((uae_u8*)&(sc.ebx))+1); break; |
2147 |
|
|
default: abort(); |
2148 |
|
|
} |
2149 |
|
|
if (pr) { |
2150 |
|
|
blockinfo* bi; |
2151 |
|
|
|
2152 |
|
|
if (currprefs.comp_oldsegv) { |
2153 |
|
|
addr-=NATMEM_OFFSET; |
2154 |
|
|
|
2155 |
|
|
if ((addr>=0x10000000 && addr<0x40000000) || |
2156 |
|
|
(addr>=0x50000000)) { |
2157 |
|
|
write_log("Suspicious address in %x SEGV handler.\n",addr); |
2158 |
|
|
} |
2159 |
|
|
if (dir==SIG_READ) { |
2160 |
|
|
switch(size) { |
2161 |
|
|
case 1: *((uae_u8*)pr)=get_byte(addr); break; |
2162 |
|
|
case 2: *((uae_u16*)pr)=get_word(addr); break; |
2163 |
|
|
case 4: *((uae_u32*)pr)=get_long(addr); break; |
2164 |
|
|
default: abort(); |
2165 |
|
|
} |
2166 |
|
|
} |
2167 |
|
|
else { /* write */ |
2168 |
|
|
switch(size) { |
2169 |
|
|
case 1: put_byte(addr,*((uae_u8*)pr)); break; |
2170 |
|
|
case 2: put_word(addr,*((uae_u16*)pr)); break; |
2171 |
|
|
case 4: put_long(addr,*((uae_u32*)pr)); break; |
2172 |
|
|
default: abort(); |
2173 |
|
|
} |
2174 |
|
|
} |
2175 |
|
|
write_log("Handled one access!\n"); |
2176 |
|
|
fflush(stdout); |
2177 |
|
|
segvcount++; |
2178 |
|
|
sc.eip+=len; |
2179 |
|
|
} |
2180 |
|
|
else { |
2181 |
|
|
void* tmp=target; |
2182 |
|
|
int i; |
2183 |
|
|
uae_u8 vecbuf[5]; |
2184 |
|
|
|
2185 |
|
|
addr-=NATMEM_OFFSET; |
2186 |
|
|
|
2187 |
|
|
if ((addr>=0x10000000 && addr<0x40000000) || |
2188 |
|
|
(addr>=0x50000000)) { |
2189 |
|
|
write_log("Suspicious address in %x SEGV handler.\n",addr); |
2190 |
|
|
} |
2191 |
|
|
|
2192 |
|
|
target=(uae_u8*)sc.eip; |
2193 |
|
|
for (i=0;i<5;i++) |
2194 |
|
|
vecbuf[i]=target[i]; |
2195 |
|
|
emit_byte(0xe9); |
2196 |
|
|
emit_long((uae_u32)veccode-(uae_u32)target-4); |
2197 |
|
|
write_log("Create jump to %p\n",veccode); |
2198 |
|
|
|
2199 |
|
|
write_log("Handled one access!\n"); |
2200 |
|
|
fflush(stdout); |
2201 |
|
|
segvcount++; |
2202 |
|
|
|
2203 |
|
|
target=veccode; |
2204 |
|
|
|
2205 |
|
|
if (dir==SIG_READ) { |
2206 |
|
|
switch(size) { |
2207 |
|
|
case 1: raw_mov_b_ri(r,get_byte(addr)); break; |
2208 |
|
|
case 2: raw_mov_w_ri(r,get_byte(addr)); break; |
2209 |
|
|
case 4: raw_mov_l_ri(r,get_byte(addr)); break; |
2210 |
|
|
default: abort(); |
2211 |
|
|
} |
2212 |
|
|
} |
2213 |
|
|
else { /* write */ |
2214 |
|
|
switch(size) { |
2215 |
|
|
case 1: put_byte(addr,*((uae_u8*)pr)); break; |
2216 |
|
|
case 2: put_word(addr,*((uae_u16*)pr)); break; |
2217 |
|
|
case 4: put_long(addr,*((uae_u32*)pr)); break; |
2218 |
|
|
default: abort(); |
2219 |
|
|
} |
2220 |
|
|
} |
2221 |
|
|
for (i=0;i<5;i++) |
2222 |
|
|
raw_mov_b_mi(sc.eip+i,vecbuf[i]); |
2223 |
|
|
raw_mov_l_mi((uae_u32)&in_handler,0); |
2224 |
|
|
emit_byte(0xe9); |
2225 |
|
|
emit_long(sc.eip+len-(uae_u32)target-4); |
2226 |
|
|
in_handler=1; |
2227 |
|
|
target=tmp; |
2228 |
|
|
} |
2229 |
|
|
bi=active; |
2230 |
|
|
while (bi) { |
2231 |
|
|
if (bi->handler && |
2232 |
|
|
(uae_u8*)bi->direct_handler<=i && |
2233 |
|
|
(uae_u8*)bi->nexthandler>i) { |
2234 |
|
|
write_log("deleted trigger (%p<%p<%p) %p\n", |
2235 |
|
|
bi->handler, |
2236 |
|
|
i, |
2237 |
|
|
bi->nexthandler, |
2238 |
|
|
bi->pc_p); |
2239 |
|
|
invalidate_block(bi); |
2240 |
|
|
raise_in_cl_list(bi); |
2241 |
|
|
set_special(0); |
2242 |
|
|
return; |
2243 |
|
|
} |
2244 |
|
|
bi=bi->next; |
2245 |
|
|
} |
2246 |
|
|
/* Not found in the active list. Might be a rom routine that |
2247 |
|
|
is in the dormant list */ |
2248 |
|
|
bi=dormant; |
2249 |
|
|
while (bi) { |
2250 |
|
|
if (bi->handler && |
2251 |
|
|
(uae_u8*)bi->direct_handler<=i && |
2252 |
|
|
(uae_u8*)bi->nexthandler>i) { |
2253 |
|
|
write_log("deleted trigger (%p<%p<%p) %p\n", |
2254 |
|
|
bi->handler, |
2255 |
|
|
i, |
2256 |
|
|
bi->nexthandler, |
2257 |
|
|
bi->pc_p); |
2258 |
|
|
invalidate_block(bi); |
2259 |
|
|
raise_in_cl_list(bi); |
2260 |
|
|
set_special(0); |
2261 |
|
|
return; |
2262 |
|
|
} |
2263 |
|
|
bi=bi->next; |
2264 |
|
|
} |
2265 |
|
|
write_log("Huh? Could not find trigger!\n"); |
2266 |
|
|
return; |
2267 |
|
|
} |
2268 |
|
|
} |
2269 |
|
|
write_log("Can't handle access!\n"); |
2270 |
|
|
for (j=0;j<10;j++) { |
2271 |
|
|
write_log("instruction byte %2d is %02x\n",j,i[j]); |
2272 |
|
|
} |
2273 |
|
|
write_log("Please send the above info (starting at \"fault address\") to\n" |
2274 |
|
|
"bmeyer@csse.monash.edu.au\n" |
2275 |
|
|
"This shouldn't happen ;-)\n"); |
2276 |
|
|
fflush(stdout); |
2277 |
|
|
signal(SIGSEGV,SIG_DFL); /* returning here will cause a "real" SEGV */ |
2278 |
|
|
} |
2279 |
|
|
#endif |
2280 |
|
|
|
2281 |
|
|
|
2282 |
|
|
/************************************************************************* |
2283 |
|
|
* Checking for CPU features * |
2284 |
|
|
*************************************************************************/ |
2285 |
|
|
|
2286 |
|
|
typedef struct { |
2287 |
|
|
uae_u32 eax; |
2288 |
|
|
uae_u32 ecx; |
2289 |
|
|
uae_u32 edx; |
2290 |
|
|
uae_u32 ebx; |
2291 |
|
|
} x86_regs; |
2292 |
|
|
|
2293 |
|
|
|
2294 |
|
|
/* This could be so much easier if it could make assumptions about the |
2295 |
|
|
compiler... */ |
2296 |
|
|
|
2297 |
|
|
static uae_u8 cpuid_space[256]; |
2298 |
|
|
static uae_u32 cpuid_ptr; |
2299 |
|
|
static uae_u32 cpuid_level; |
2300 |
|
|
|
2301 |
|
|
static x86_regs cpuid(uae_u32 level) |
2302 |
|
|
{ |
2303 |
|
|
x86_regs answer; |
2304 |
|
|
uae_u8* tmp=get_target(); |
2305 |
|
|
|
2306 |
|
|
cpuid_ptr=(uae_u32)&answer; |
2307 |
|
|
cpuid_level=level; |
2308 |
|
|
|
2309 |
|
|
set_target(cpuid_space); |
2310 |
|
|
raw_push_l_r(0); /* eax */ |
2311 |
|
|
raw_push_l_r(1); /* ecx */ |
2312 |
|
|
raw_push_l_r(2); /* edx */ |
2313 |
|
|
raw_push_l_r(3); /* ebx */ |
2314 |
|
|
raw_push_l_r(7); /* edi */ |
2315 |
|
|
raw_mov_l_rm(0,(uae_u32)&cpuid_level); |
2316 |
|
|
raw_cpuid(0); |
2317 |
|
|
raw_mov_l_rm(7,(uae_u32)&cpuid_ptr); |
2318 |
|
|
raw_mov_l_Rr(7,0,0); |
2319 |
|
|
raw_mov_l_Rr(7,1,4); |
2320 |
|
|
raw_mov_l_Rr(7,2,8); |
2321 |
|
|
raw_mov_l_Rr(7,3,12); |
2322 |
|
|
raw_pop_l_r(7); |
2323 |
|
|
raw_pop_l_r(3); |
2324 |
|
|
raw_pop_l_r(2); |
2325 |
|
|
raw_pop_l_r(1); |
2326 |
|
|
raw_pop_l_r(0); |
2327 |
|
|
raw_ret(); |
2328 |
|
|
set_target(tmp); |
2329 |
|
|
|
2330 |
|
|
((cpuop_func*)cpuid_space)(0); |
2331 |
|
|
return answer; |
2332 |
|
|
} |
2333 |
|
|
|
2334 |
|
|
static void raw_init_cpu(void) |
2335 |
|
|
{ |
2336 |
|
|
x86_regs x; |
2337 |
|
|
uae_u32 maxlev; |
2338 |
|
|
|
2339 |
|
|
x=cpuid(0); |
2340 |
|
|
maxlev=x.eax; |
2341 |
|
|
write_log("Max CPUID level=%d Processor is %c%c%c%c%c%c%c%c%c%c%c%c\n", |
2342 |
|
|
maxlev, |
2343 |
|
|
x.ebx, |
2344 |
|
|
x.ebx>>8, |
2345 |
|
|
x.ebx>>16, |
2346 |
|
|
x.ebx>>24, |
2347 |
|
|
x.edx, |
2348 |
|
|
x.edx>>8, |
2349 |
|
|
x.edx>>16, |
2350 |
|
|
x.edx>>24, |
2351 |
|
|
x.ecx, |
2352 |
|
|
x.ecx>>8, |
2353 |
|
|
x.ecx>>16, |
2354 |
|
|
x.ecx>>24 |
2355 |
|
|
); |
2356 |
|
|
have_rat_stall=(x.ecx==0x6c65746e); |
2357 |
|
|
|
2358 |
|
|
if (maxlev>=1) { |
2359 |
|
|
x=cpuid(1); |
2360 |
|
|
if (x.edx&(1<<15)) |
2361 |
|
|
have_cmov=1; |
2362 |
|
|
} |
2363 |
|
|
if (!have_cmov) |
2364 |
|
|
have_rat_stall=0; |
2365 |
|
|
#if 0 /* For testing of non-cmov code! */ |
2366 |
|
|
have_cmov=0; |
2367 |
|
|
#endif |
2368 |
|
|
#if 1 /* It appears that partial register writes are a bad idea even on |
2369 |
|
|
AMD K7 cores, even though they are not supposed to have the |
2370 |
|
|
dreaded rat stall. Why? Anyway, that's why we lie about it ;-) */ |
2371 |
|
|
if (have_cmov) |
2372 |
|
|
have_rat_stall=1; |
2373 |
|
|
#endif |
2374 |
|
|
} |
2375 |
|
|
|
2376 |
|
|
|
2377 |
|
|
/************************************************************************* |
2378 |
|
|
* FPU stuff * |
2379 |
|
|
*************************************************************************/ |
2380 |
|
|
|
2381 |
|
|
|
2382 |
|
|
static __inline__ void raw_fp_init(void) |
2383 |
|
|
{ |
2384 |
|
|
int i; |
2385 |
|
|
|
2386 |
|
|
for (i=0;i<N_FREGS;i++) |
2387 |
|
|
live.spos[i]=-2; |
2388 |
|
|
live.tos=-1; /* Stack is empty */ |
2389 |
|
|
} |
2390 |
|
|
|
2391 |
|
|
static __inline__ void raw_fp_cleanup_drop(void) |
2392 |
|
|
{ |
2393 |
|
|
#if 0 |
2394 |
|
|
/* using FINIT instead of popping all the entries. |
2395 |
|
|
Seems to have side effects --- there is display corruption in |
2396 |
|
|
Quake when this is used */ |
2397 |
|
|
if (live.tos>1) { |
2398 |
|
|
emit_byte(0x9b); |
2399 |
|
|
emit_byte(0xdb); |
2400 |
|
|
emit_byte(0xe3); |
2401 |
|
|
live.tos=-1; |
2402 |
|
|
} |
2403 |
|
|
#endif |
2404 |
|
|
while (live.tos>=1) { |
2405 |
|
|
emit_byte(0xde); |
2406 |
|
|
emit_byte(0xd9); |
2407 |
|
|
live.tos-=2; |
2408 |
|
|
} |
2409 |
|
|
while (live.tos>=0) { |
2410 |
|
|
emit_byte(0xdd); |
2411 |
|
|
emit_byte(0xd8); |
2412 |
|
|
live.tos--; |
2413 |
|
|
} |
2414 |
|
|
raw_fp_init(); |
2415 |
|
|
} |
2416 |
|
|
|
2417 |
|
|
static __inline__ void make_tos(int r) |
2418 |
|
|
{ |
2419 |
|
|
int p,q; |
2420 |
|
|
|
2421 |
|
|
if (live.spos[r]<0) { /* Register not yet on stack */ |
2422 |
|
|
emit_byte(0xd9); |
2423 |
|
|
emit_byte(0xe8); /* Push '1' on the stack, just to grow it */ |
2424 |
|
|
live.tos++; |
2425 |
|
|
live.spos[r]=live.tos; |
2426 |
|
|
live.onstack[live.tos]=r; |
2427 |
|
|
return; |
2428 |
|
|
} |
2429 |
|
|
/* Register is on stack */ |
2430 |
|
|
if (live.tos==live.spos[r]) |
2431 |
|
|
return; |
2432 |
|
|
p=live.spos[r]; |
2433 |
|
|
q=live.onstack[live.tos]; |
2434 |
|
|
|
2435 |
|
|
emit_byte(0xd9); |
2436 |
|
|
emit_byte(0xc8+live.tos-live.spos[r]); /* exchange it with top of stack */ |
2437 |
|
|
live.onstack[live.tos]=r; |
2438 |
|
|
live.spos[r]=live.tos; |
2439 |
|
|
live.onstack[p]=q; |
2440 |
|
|
live.spos[q]=p; |
2441 |
|
|
} |
2442 |
|
|
|
2443 |
|
|
static __inline__ void make_tos2(int r, int r2) |
2444 |
|
|
{ |
2445 |
|
|
int q; |
2446 |
|
|
|
2447 |
|
|
make_tos(r2); /* Put the reg that's supposed to end up in position2 |
2448 |
|
|
on top */ |
2449 |
|
|
|
2450 |
|
|
if (live.spos[r]<0) { /* Register not yet on stack */ |
2451 |
|
|
make_tos(r); /* This will extend the stack */ |
2452 |
|
|
return; |
2453 |
|
|
} |
2454 |
|
|
/* Register is on stack */ |
2455 |
|
|
emit_byte(0xd9); |
2456 |
|
|
emit_byte(0xc9); /* Move r2 into position 2 */ |
2457 |
|
|
|
2458 |
|
|
q=live.onstack[live.tos-1]; |
2459 |
|
|
live.onstack[live.tos]=q; |
2460 |
|
|
live.spos[q]=live.tos; |
2461 |
|
|
live.onstack[live.tos-1]=r2; |
2462 |
|
|
live.spos[r2]=live.tos-1; |
2463 |
|
|
|
2464 |
|
|
make_tos(r); /* And r into 1 */ |
2465 |
|
|
} |
2466 |
|
|
|
2467 |
|
|
static __inline__ int stackpos(int r) |
2468 |
|
|
{ |
2469 |
|
|
if (live.spos[r]<0) |
2470 |
|
|
abort(); |
2471 |
|
|
if (live.tos<live.spos[r]) { |
2472 |
|
|
printf("Looking for spos for fnreg %d\n",r); |
2473 |
|
|
abort(); |
2474 |
|
|
} |
2475 |
|
|
return live.tos-live.spos[r]; |
2476 |
|
|
} |
2477 |
|
|
|
2478 |
|
|
static __inline__ void usereg(int r) |
2479 |
|
|
{ |
2480 |
|
|
if (live.spos[r]<0) |
2481 |
|
|
make_tos(r); |
2482 |
|
|
} |
2483 |
|
|
|
2484 |
|
|
/* This is called with one FP value in a reg *above* tos, which it will |
2485 |
|
|
pop off the stack if necessary */ |
2486 |
|
|
static __inline__ void tos_make(int r) |
2487 |
|
|
{ |
2488 |
|
|
if (live.spos[r]<0) { |
2489 |
|
|
live.tos++; |
2490 |
|
|
live.spos[r]=live.tos; |
2491 |
|
|
live.onstack[live.tos]=r; |
2492 |
|
|
return; |
2493 |
|
|
} |
2494 |
|
|
emit_byte(0xdd); |
2495 |
|
|
emit_byte(0xd8+(live.tos+1)-live.spos[r]); /* store top of stack in reg, |
2496 |
|
|
and pop it*/ |
2497 |
|
|
} |
2498 |
|
|
|
2499 |
|
|
|
2500 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2501 |
|
|
{ |
2502 |
|
|
make_tos(r); |
2503 |
|
|
emit_byte(0xdd); |
2504 |
|
|
emit_byte(0x15); |
2505 |
|
|
emit_long(m); |
2506 |
|
|
} |
2507 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2508 |
|
|
|
2509 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_mr_drop,(MEMW m, FR r)) |
2510 |
|
|
{ |
2511 |
|
|
make_tos(r); |
2512 |
|
|
emit_byte(0xdd); |
2513 |
|
|
emit_byte(0x1d); |
2514 |
|
|
emit_long(m); |
2515 |
|
|
live.onstack[live.tos]=-1; |
2516 |
|
|
live.tos--; |
2517 |
|
|
live.spos[r]=-2; |
2518 |
|
|
} |
2519 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_mr,(MEMW m, FR r)) |
2520 |
|
|
|
2521 |
|
|
LOWFUNC(NONE,READ,2,raw_fmov_rm,(FW r, MEMR m)) |
2522 |
|
|
{ |
2523 |
|
|
emit_byte(0xdd); |
2524 |
|
|
emit_byte(0x05); |
2525 |
|
|
emit_long(m); |
2526 |
|
|
tos_make(r); |
2527 |
|
|
} |
2528 |
|
|
LENDFUNC(NONE,READ,2,raw_fmov_rm,(FW r, MEMR m)) |
2529 |
|
|
|
2530 |
|
|
LOWFUNC(NONE,READ,2,raw_fmovi_rm,(FW r, MEMR m)) |
2531 |
|
|
{ |
2532 |
|
|
emit_byte(0xdb); |
2533 |
|
|
emit_byte(0x05); |
2534 |
|
|
emit_long(m); |
2535 |
|
|
tos_make(r); |
2536 |
|
|
} |
2537 |
|
|
LENDFUNC(NONE,READ,2,raw_fmovi_rm,(FW r, MEMR m)) |
2538 |
|
|
|
2539 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmovi_mr,(MEMW m, FR r)) |
2540 |
|
|
{ |
2541 |
|
|
make_tos(r); |
2542 |
|
|
emit_byte(0xdb); |
2543 |
|
|
emit_byte(0x15); |
2544 |
|
|
emit_long(m); |
2545 |
|
|
} |
2546 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmovi_mr,(MEMW m, FR r)) |
2547 |
|
|
|
2548 |
|
|
LOWFUNC(NONE,READ,2,raw_fmovs_rm,(FW r, MEMR m)) |
2549 |
|
|
{ |
2550 |
|
|
emit_byte(0xd9); |
2551 |
|
|
emit_byte(0x05); |
2552 |
|
|
emit_long(m); |
2553 |
|
|
tos_make(r); |
2554 |
|
|
} |
2555 |
|
|
LENDFUNC(NONE,READ,2,raw_fmovs_rm,(FW r, MEMR m)) |
2556 |
|
|
|
2557 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmovs_mr,(MEMW m, FR r)) |
2558 |
|
|
{ |
2559 |
|
|
make_tos(r); |
2560 |
|
|
emit_byte(0xd9); |
2561 |
|
|
emit_byte(0x15); |
2562 |
|
|
emit_long(m); |
2563 |
|
|
} |
2564 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmovs_mr,(MEMW m, FR r)) |
2565 |
|
|
|
2566 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2567 |
|
|
{ |
2568 |
|
|
int rs; |
2569 |
|
|
|
2570 |
|
|
/* Stupid x87 can't write a long double to mem without popping the |
2571 |
|
|
stack! */ |
2572 |
|
|
usereg(r); |
2573 |
|
|
rs=stackpos(r); |
2574 |
|
|
emit_byte(0xd9); /* Get a copy to the top of stack */ |
2575 |
|
|
emit_byte(0xc0+rs); |
2576 |
|
|
|
2577 |
|
|
emit_byte(0xdb); /* store and pop it */ |
2578 |
|
|
emit_byte(0x3d); |
2579 |
|
|
emit_long(m); |
2580 |
|
|
} |
2581 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2582 |
|
|
|
2583 |
|
|
LOWFUNC(NONE,WRITE,2,raw_fmov_ext_mr_drop,(MEMW m, FR r)) |
2584 |
|
|
{ |
2585 |
|
|
int rs; |
2586 |
|
|
|
2587 |
|
|
make_tos(r); |
2588 |
|
|
emit_byte(0xdb); /* store and pop it */ |
2589 |
|
|
emit_byte(0x3d); |
2590 |
|
|
emit_long(m); |
2591 |
|
|
live.onstack[live.tos]=-1; |
2592 |
|
|
live.tos--; |
2593 |
|
|
live.spos[r]=-2; |
2594 |
|
|
} |
2595 |
|
|
LENDFUNC(NONE,WRITE,2,raw_fmov_ext_mr,(MEMW m, FR r)) |
2596 |
|
|
|
2597 |
|
|
LOWFUNC(NONE,READ,2,raw_fmov_ext_rm,(FW r, MEMR m)) |
2598 |
|
|
{ |
2599 |
|
|
emit_byte(0xdb); |
2600 |
|
|
emit_byte(0x2d); |
2601 |
|
|
emit_long(m); |
2602 |
|
|
tos_make(r); |
2603 |
|
|
} |
2604 |
|
|
LENDFUNC(NONE,READ,2,raw_fmov_ext_rm,(FW r, MEMR m)) |
2605 |
|
|
|
2606 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_pi,(FW r)) |
2607 |
|
|
{ |
2608 |
|
|
emit_byte(0xd9); |
2609 |
|
|
emit_byte(0xeb); |
2610 |
|
|
tos_make(r); |
2611 |
|
|
} |
2612 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_pi,(FW r)) |
2613 |
|
|
|
2614 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_log10_2,(FW r)) |
2615 |
|
|
{ |
2616 |
|
|
emit_byte(0xd9); |
2617 |
|
|
emit_byte(0xec); |
2618 |
|
|
tos_make(r); |
2619 |
|
|
} |
2620 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_log10_2,(FW r)) |
2621 |
|
|
|
2622 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_log2_e,(FW r)) |
2623 |
|
|
{ |
2624 |
|
|
emit_byte(0xd9); |
2625 |
|
|
emit_byte(0xea); |
2626 |
|
|
tos_make(r); |
2627 |
|
|
} |
2628 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_log2_e,(FW r)) |
2629 |
|
|
|
2630 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_loge_2,(FW r)) |
2631 |
|
|
{ |
2632 |
|
|
emit_byte(0xd9); |
2633 |
|
|
emit_byte(0xed); |
2634 |
|
|
tos_make(r); |
2635 |
|
|
} |
2636 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_loge_2,(FW r)) |
2637 |
|
|
|
2638 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_1,(FW r)) |
2639 |
|
|
{ |
2640 |
|
|
emit_byte(0xd9); |
2641 |
|
|
emit_byte(0xe8); |
2642 |
|
|
tos_make(r); |
2643 |
|
|
} |
2644 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_1,(FW r)) |
2645 |
|
|
|
2646 |
|
|
LOWFUNC(NONE,NONE,1,raw_fmov_0,(FW r)) |
2647 |
|
|
{ |
2648 |
|
|
emit_byte(0xd9); |
2649 |
|
|
emit_byte(0xee); |
2650 |
|
|
tos_make(r); |
2651 |
|
|
} |
2652 |
|
|
LENDFUNC(NONE,NONE,1,raw_fmov_0,(FW r)) |
2653 |
|
|
|
2654 |
|
|
LOWFUNC(NONE,NONE,2,raw_fmov_rr,(FW d, FR s)) |
2655 |
|
|
{ |
2656 |
|
|
int ds; |
2657 |
|
|
|
2658 |
|
|
usereg(s); |
2659 |
|
|
ds=stackpos(s); |
2660 |
|
|
if (ds==0 && live.spos[d]>=0) { |
2661 |
|
|
/* source is on top of stack, and we already have the dest */ |
2662 |
|
|
int dd=stackpos(d); |
2663 |
|
|
emit_byte(0xdd); |
2664 |
|
|
emit_byte(0xd0+dd); |
2665 |
|
|
} |
2666 |
|
|
else { |
2667 |
|
|
emit_byte(0xd9); |
2668 |
|
|
emit_byte(0xc0+ds); /* duplicate source on tos */ |
2669 |
|
|
tos_make(d); /* store to destination, pop if necessary */ |
2670 |
|
|
} |
2671 |
|
|
} |
2672 |
|
|
LENDFUNC(NONE,NONE,2,raw_fmov_rr,(FW d, FR s)) |
2673 |
|
|
|
2674 |
|
|
LOWFUNC(NONE,READ,4,raw_fldcw_m_indexed,(R4 index, IMM base)) |
2675 |
|
|
{ |
2676 |
|
|
emit_byte(0xd9); |
2677 |
|
|
emit_byte(0xa8+index); |
2678 |
|
|
emit_long(base); |
2679 |
|
|
} |
2680 |
|
|
LENDFUNC(NONE,READ,4,raw_fldcw_m_indexed,(R4 index, IMM base)) |
2681 |
|
|
|
2682 |
|
|
|
2683 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsqrt_rr,(FW d, FR s)) |
2684 |
|
|
{ |
2685 |
|
|
int ds; |
2686 |
|
|
|
2687 |
|
|
if (d!=s) { |
2688 |
|
|
usereg(s); |
2689 |
|
|
ds=stackpos(s); |
2690 |
|
|
emit_byte(0xd9); |
2691 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2692 |
|
|
emit_byte(0xd9); |
2693 |
|
|
emit_byte(0xfa); /* take square root */ |
2694 |
|
|
tos_make(d); /* store to destination */ |
2695 |
|
|
} |
2696 |
|
|
else { |
2697 |
|
|
make_tos(d); |
2698 |
|
|
emit_byte(0xd9); |
2699 |
|
|
emit_byte(0xfa); /* take square root */ |
2700 |
|
|
} |
2701 |
|
|
} |
2702 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsqrt_rr,(FW d, FR s)) |
2703 |
|
|
|
2704 |
|
|
LOWFUNC(NONE,NONE,2,raw_fabs_rr,(FW d, FR s)) |
2705 |
|
|
{ |
2706 |
|
|
int ds; |
2707 |
|
|
|
2708 |
|
|
if (d!=s) { |
2709 |
|
|
usereg(s); |
2710 |
|
|
ds=stackpos(s); |
2711 |
|
|
emit_byte(0xd9); |
2712 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2713 |
|
|
emit_byte(0xd9); |
2714 |
|
|
emit_byte(0xe1); /* take fabs */ |
2715 |
|
|
tos_make(d); /* store to destination */ |
2716 |
|
|
} |
2717 |
|
|
else { |
2718 |
|
|
make_tos(d); |
2719 |
|
|
emit_byte(0xd9); |
2720 |
|
|
emit_byte(0xe1); /* take fabs */ |
2721 |
|
|
} |
2722 |
|
|
} |
2723 |
|
|
LENDFUNC(NONE,NONE,2,raw_fabs_rr,(FW d, FR s)) |
2724 |
|
|
|
2725 |
|
|
LOWFUNC(NONE,NONE,2,raw_frndint_rr,(FW d, FR s)) |
2726 |
|
|
{ |
2727 |
|
|
int ds; |
2728 |
|
|
|
2729 |
|
|
if (d!=s) { |
2730 |
|
|
usereg(s); |
2731 |
|
|
ds=stackpos(s); |
2732 |
|
|
emit_byte(0xd9); |
2733 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2734 |
|
|
emit_byte(0xd9); |
2735 |
|
|
emit_byte(0xfc); /* take frndint */ |
2736 |
|
|
tos_make(d); /* store to destination */ |
2737 |
|
|
} |
2738 |
|
|
else { |
2739 |
|
|
make_tos(d); |
2740 |
|
|
emit_byte(0xd9); |
2741 |
|
|
emit_byte(0xfc); /* take frndint */ |
2742 |
|
|
} |
2743 |
|
|
} |
2744 |
|
|
LENDFUNC(NONE,NONE,2,raw_frndint_rr,(FW d, FR s)) |
2745 |
|
|
|
2746 |
|
|
LOWFUNC(NONE,NONE,2,raw_fcos_rr,(FW d, FR s)) |
2747 |
|
|
{ |
2748 |
|
|
int ds; |
2749 |
|
|
|
2750 |
|
|
if (d!=s) { |
2751 |
|
|
usereg(s); |
2752 |
|
|
ds=stackpos(s); |
2753 |
|
|
emit_byte(0xd9); |
2754 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2755 |
|
|
emit_byte(0xd9); |
2756 |
|
|
emit_byte(0xff); /* take cos */ |
2757 |
|
|
tos_make(d); /* store to destination */ |
2758 |
|
|
} |
2759 |
|
|
else { |
2760 |
|
|
make_tos(d); |
2761 |
|
|
emit_byte(0xd9); |
2762 |
|
|
emit_byte(0xff); /* take cos */ |
2763 |
|
|
} |
2764 |
|
|
} |
2765 |
|
|
LENDFUNC(NONE,NONE,2,raw_fcos_rr,(FW d, FR s)) |
2766 |
|
|
|
2767 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsin_rr,(FW d, FR s)) |
2768 |
|
|
{ |
2769 |
|
|
int ds; |
2770 |
|
|
|
2771 |
|
|
if (d!=s) { |
2772 |
|
|
usereg(s); |
2773 |
|
|
ds=stackpos(s); |
2774 |
|
|
emit_byte(0xd9); |
2775 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2776 |
|
|
emit_byte(0xd9); |
2777 |
|
|
emit_byte(0xfe); /* take sin */ |
2778 |
|
|
tos_make(d); /* store to destination */ |
2779 |
|
|
} |
2780 |
|
|
else { |
2781 |
|
|
make_tos(d); |
2782 |
|
|
emit_byte(0xd9); |
2783 |
|
|
emit_byte(0xfe); /* take sin */ |
2784 |
|
|
} |
2785 |
|
|
} |
2786 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsin_rr,(FW d, FR s)) |
2787 |
|
|
|
2788 |
|
|
double one=1; |
2789 |
|
|
LOWFUNC(NONE,NONE,2,raw_ftwotox_rr,(FW d, FR s)) |
2790 |
|
|
{ |
2791 |
|
|
int ds; |
2792 |
|
|
|
2793 |
|
|
usereg(s); |
2794 |
|
|
ds=stackpos(s); |
2795 |
|
|
emit_byte(0xd9); |
2796 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2797 |
|
|
|
2798 |
|
|
emit_byte(0xd9); |
2799 |
|
|
emit_byte(0xc0); /* duplicate top of stack. Now up to 8 high */ |
2800 |
|
|
emit_byte(0xd9); |
2801 |
|
|
emit_byte(0xfc); /* rndint */ |
2802 |
|
|
emit_byte(0xd9); |
2803 |
|
|
emit_byte(0xc9); /* swap top two elements */ |
2804 |
|
|
emit_byte(0xd8); |
2805 |
|
|
emit_byte(0xe1); /* subtract rounded from original */ |
2806 |
|
|
emit_byte(0xd9); |
2807 |
|
|
emit_byte(0xf0); /* f2xm1 */ |
2808 |
|
|
emit_byte(0xdc); |
2809 |
|
|
emit_byte(0x05); |
2810 |
|
|
emit_long((uae_u32)&one); /* Add '1' without using extra stack space */ |
2811 |
|
|
emit_byte(0xd9); |
2812 |
|
|
emit_byte(0xfd); /* and scale it */ |
2813 |
|
|
emit_byte(0xdd); |
2814 |
|
|
emit_byte(0xd9); /* take he rounded value off */ |
2815 |
|
|
tos_make(d); /* store to destination */ |
2816 |
|
|
} |
2817 |
|
|
LENDFUNC(NONE,NONE,2,raw_ftwotox_rr,(FW d, FR s)) |
2818 |
|
|
|
2819 |
|
|
LOWFUNC(NONE,NONE,2,raw_fetox_rr,(FW d, FR s)) |
2820 |
|
|
{ |
2821 |
|
|
int ds; |
2822 |
|
|
|
2823 |
|
|
usereg(s); |
2824 |
|
|
ds=stackpos(s); |
2825 |
|
|
emit_byte(0xd9); |
2826 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2827 |
|
|
emit_byte(0xd9); |
2828 |
|
|
emit_byte(0xea); /* fldl2e */ |
2829 |
|
|
emit_byte(0xde); |
2830 |
|
|
emit_byte(0xc9); /* fmulp --- multiply source by log2(e) */ |
2831 |
|
|
|
2832 |
|
|
emit_byte(0xd9); |
2833 |
|
|
emit_byte(0xc0); /* duplicate top of stack. Now up to 8 high */ |
2834 |
|
|
emit_byte(0xd9); |
2835 |
|
|
emit_byte(0xfc); /* rndint */ |
2836 |
|
|
emit_byte(0xd9); |
2837 |
|
|
emit_byte(0xc9); /* swap top two elements */ |
2838 |
|
|
emit_byte(0xd8); |
2839 |
|
|
emit_byte(0xe1); /* subtract rounded from original */ |
2840 |
|
|
emit_byte(0xd9); |
2841 |
|
|
emit_byte(0xf0); /* f2xm1 */ |
2842 |
|
|
emit_byte(0xdc); |
2843 |
|
|
emit_byte(0x05); |
2844 |
|
|
emit_long((uae_u32)&one); /* Add '1' without using extra stack space */ |
2845 |
|
|
emit_byte(0xd9); |
2846 |
|
|
emit_byte(0xfd); /* and scale it */ |
2847 |
|
|
emit_byte(0xdd); |
2848 |
|
|
emit_byte(0xd9); /* take he rounded value off */ |
2849 |
|
|
tos_make(d); /* store to destination */ |
2850 |
|
|
} |
2851 |
|
|
LENDFUNC(NONE,NONE,2,raw_fetox_rr,(FW d, FR s)) |
2852 |
|
|
|
2853 |
|
|
LOWFUNC(NONE,NONE,2,raw_flog2_rr,(FW d, FR s)) |
2854 |
|
|
{ |
2855 |
|
|
int ds; |
2856 |
|
|
|
2857 |
|
|
usereg(s); |
2858 |
|
|
ds=stackpos(s); |
2859 |
|
|
emit_byte(0xd9); |
2860 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2861 |
|
|
emit_byte(0xd9); |
2862 |
|
|
emit_byte(0xe8); /* push '1' */ |
2863 |
|
|
emit_byte(0xd9); |
2864 |
|
|
emit_byte(0xc9); /* swap top two */ |
2865 |
|
|
emit_byte(0xd9); |
2866 |
|
|
emit_byte(0xf1); /* take 1*log2(x) */ |
2867 |
|
|
tos_make(d); /* store to destination */ |
2868 |
|
|
} |
2869 |
|
|
LENDFUNC(NONE,NONE,2,raw_flog2_rr,(FW d, FR s)) |
2870 |
|
|
|
2871 |
|
|
|
2872 |
|
|
LOWFUNC(NONE,NONE,2,raw_fneg_rr,(FW d, FR s)) |
2873 |
|
|
{ |
2874 |
|
|
int ds; |
2875 |
|
|
|
2876 |
|
|
if (d!=s) { |
2877 |
|
|
usereg(s); |
2878 |
|
|
ds=stackpos(s); |
2879 |
|
|
emit_byte(0xd9); |
2880 |
|
|
emit_byte(0xc0+ds); /* duplicate source */ |
2881 |
|
|
emit_byte(0xd9); |
2882 |
|
|
emit_byte(0xe0); /* take fchs */ |
2883 |
|
|
tos_make(d); /* store to destination */ |
2884 |
|
|
} |
2885 |
|
|
else { |
2886 |
|
|
make_tos(d); |
2887 |
|
|
emit_byte(0xd9); |
2888 |
|
|
emit_byte(0xe0); /* take fchs */ |
2889 |
|
|
} |
2890 |
|
|
} |
2891 |
|
|
LENDFUNC(NONE,NONE,2,raw_fneg_rr,(FW d, FR s)) |
2892 |
|
|
|
2893 |
|
|
LOWFUNC(NONE,NONE,2,raw_fadd_rr,(FRW d, FR s)) |
2894 |
|
|
{ |
2895 |
|
|
int ds; |
2896 |
|
|
|
2897 |
|
|
usereg(s); |
2898 |
|
|
usereg(d); |
2899 |
|
|
|
2900 |
|
|
if (live.spos[s]==live.tos) { |
2901 |
|
|
/* Source is on top of stack */ |
2902 |
|
|
ds=stackpos(d); |
2903 |
|
|
emit_byte(0xdc); |
2904 |
|
|
emit_byte(0xc0+ds); /* add source to dest*/ |
2905 |
|
|
} |
2906 |
|
|
else { |
2907 |
|
|
make_tos(d); |
2908 |
|
|
ds=stackpos(s); |
2909 |
|
|
|
2910 |
|
|
emit_byte(0xd8); |
2911 |
|
|
emit_byte(0xc0+ds); /* add source to dest*/ |
2912 |
|
|
} |
2913 |
|
|
} |
2914 |
|
|
LENDFUNC(NONE,NONE,2,raw_fadd_rr,(FRW d, FR s)) |
2915 |
|
|
|
2916 |
|
|
LOWFUNC(NONE,NONE,2,raw_fsub_rr,(FRW d, FR s)) |
2917 |
|
|
{ |
2918 |
|
|
int ds; |
2919 |
|
|
|
2920 |
|
|
usereg(s); |
2921 |
|
|
usereg(d); |
2922 |
|
|
|
2923 |
|
|
if (live.spos[s]==live.tos) { |
2924 |
|
|
/* Source is on top of stack */ |
2925 |
|
|
ds=stackpos(d); |
2926 |
|
|
emit_byte(0xdc); |
2927 |
|
|
emit_byte(0xe8+ds); /* sub source from dest*/ |
2928 |
|
|
} |
2929 |
|
|
else { |
2930 |
|
|
make_tos(d); |
2931 |
|
|
ds=stackpos(s); |
2932 |
|
|
|
2933 |
|
|
emit_byte(0xd8); |
2934 |
|
|
emit_byte(0xe0+ds); /* sub src from dest */ |
2935 |
|
|
} |
2936 |
|
|
} |
2937 |
|
|
LENDFUNC(NONE,NONE,2,raw_fsub_rr,(FRW d, FR s)) |
2938 |
|
|
|
2939 |
|
|
LOWFUNC(NONE,NONE,2,raw_fcmp_rr,(FR d, FR s)) |
2940 |
|
|
{ |
2941 |
|
|
int ds; |
2942 |
|
|
|
2943 |
|
|
usereg(s); |
2944 |
|
|
usereg(d); |
2945 |
|
|
|
2946 |
|
|
make_tos(d); |
2947 |
|
|
ds=stackpos(s); |
2948 |
|
|
|
2949 |
|
|
emit_byte(0xdd); |
2950 |
|
|
emit_byte(0xe0+ds); /* cmp dest with source*/ |
2951 |
|
|
} |
2952 |
|
|
LENDFUNC(NONE,NONE,2,raw_fcmp_rr,(FR d, FR s)) |
2953 |
|
|
|
2954 |
|
|
LOWFUNC(NONE,NONE,2,raw_fmul_rr,(FRW d, FR s)) |
2955 |
|
|
{ |
2956 |
|
|
int ds; |
2957 |
|
|
|
2958 |
|
|
usereg(s); |
2959 |
|
|
usereg(d); |
2960 |
|
|
|
2961 |
|
|
if (live.spos[s]==live.tos) { |
2962 |
|
|
/* Source is on top of stack */ |
2963 |
|
|
ds=stackpos(d); |
2964 |
|
|
emit_byte(0xdc); |
2965 |
|
|
emit_byte(0xc8+ds); /* mul dest by source*/ |
2966 |
|
|
} |
2967 |
|
|
else { |
2968 |
|
|
make_tos(d); |
2969 |
|
|
ds=stackpos(s); |
2970 |
|
|
|
2971 |
|
|
emit_byte(0xd8); |
2972 |
|
|
emit_byte(0xc8+ds); /* mul dest by source*/ |
2973 |
|
|
} |
2974 |
|
|
} |
2975 |
|
|
LENDFUNC(NONE,NONE,2,raw_fmul_rr,(FRW d, FR s)) |
2976 |
|
|
|
2977 |
|
|
LOWFUNC(NONE,NONE,2,raw_fdiv_rr,(FRW d, FR s)) |
2978 |
|
|
{ |
2979 |
|
|
int ds; |
2980 |
|
|
|
2981 |
|
|
usereg(s); |
2982 |
|
|
usereg(d); |
2983 |
|
|
|
2984 |
|
|
if (live.spos[s]==live.tos) { |
2985 |
|
|
/* Source is on top of stack */ |
2986 |
|
|
ds=stackpos(d); |
2987 |
|
|
emit_byte(0xdc); |
2988 |
|
|
emit_byte(0xf8+ds); /* div dest by source */ |
2989 |
|
|
} |
2990 |
|
|
else { |
2991 |
|
|
make_tos(d); |
2992 |
|
|
ds=stackpos(s); |
2993 |
|
|
|
2994 |
|
|
emit_byte(0xd8); |
2995 |
|
|
emit_byte(0xf0+ds); /* div dest by source*/ |
2996 |
|
|
} |
2997 |
|
|
} |
2998 |
|
|
LENDFUNC(NONE,NONE,2,raw_fdiv_rr,(FRW d, FR s)) |
2999 |
|
|
|
3000 |
|
|
LOWFUNC(NONE,NONE,2,raw_frem_rr,(FRW d, FR s)) |
3001 |
|
|
{ |
3002 |
|
|
int ds; |
3003 |
|
|
|
3004 |
|
|
usereg(s); |
3005 |
|
|
usereg(d); |
3006 |
|
|
|
3007 |
|
|
make_tos2(d,s); |
3008 |
|
|
ds=stackpos(s); |
3009 |
|
|
|
3010 |
|
|
if (ds!=1) { |
3011 |
|
|
printf("Failed horribly in raw_frem_rr! ds is %d\n",ds); |
3012 |
|
|
abort(); |
3013 |
|
|
} |
3014 |
|
|
emit_byte(0xd9); |
3015 |
|
|
emit_byte(0xf8); /* take rem from dest by source */ |
3016 |
|
|
} |
3017 |
|
|
LENDFUNC(NONE,NONE,2,raw_frem_rr,(FRW d, FR s)) |
3018 |
|
|
|
3019 |
|
|
LOWFUNC(NONE,NONE,2,raw_frem1_rr,(FRW d, FR s)) |
3020 |
|
|
{ |
3021 |
|
|
int ds; |
3022 |
|
|
|
3023 |
|
|
usereg(s); |
3024 |
|
|
usereg(d); |
3025 |
|
|
|
3026 |
|
|
make_tos2(d,s); |
3027 |
|
|
ds=stackpos(s); |
3028 |
|
|
|
3029 |
|
|
if (ds!=1) { |
3030 |
|
|
printf("Failed horribly in raw_frem1_rr! ds is %d\n",ds); |
3031 |
|
|
abort(); |
3032 |
|
|
} |
3033 |
|
|
emit_byte(0xd9); |
3034 |
|
|
emit_byte(0xf5); /* take rem1 from dest by source */ |
3035 |
|
|
} |
3036 |
|
|
LENDFUNC(NONE,NONE,2,raw_frem1_rr,(FRW d, FR s)) |
3037 |
|
|
|
3038 |
|
|
|
3039 |
|
|
LOWFUNC(NONE,NONE,1,raw_ftst_r,(FR r)) |
3040 |
|
|
{ |
3041 |
|
|
make_tos(r); |
3042 |
|
|
emit_byte(0xd9); /* ftst */ |
3043 |
|
|
emit_byte(0xe4); |
3044 |
|
|
} |
3045 |
|
|
LENDFUNC(NONE,NONE,1,raw_ftst_r,(FR r)) |
3046 |
|
|
|
3047 |
|
|
/* %eax register is clobbered if target processor doesn't support fucomi */ |
3048 |
|
|
#define FFLAG_NREG_CLOBBER_CONDITION !have_cmov |
3049 |
|
|
#define FFLAG_NREG EAX_INDEX |
3050 |
|
|
|
3051 |
|
|
static __inline__ void raw_fflags_into_flags(int r) |
3052 |
|
|
{ |
3053 |
|
|
int p; |
3054 |
|
|
|
3055 |
|
|
usereg(r); |
3056 |
|
|
p=stackpos(r); |
3057 |
|
|
|
3058 |
|
|
emit_byte(0xd9); |
3059 |
|
|
emit_byte(0xee); /* Push 0 */ |
3060 |
|
|
emit_byte(0xd9); |
3061 |
|
|
emit_byte(0xc9+p); /* swap top two around */ |
3062 |
|
|
if (have_cmov) { |
3063 |
|
|
// gb-- fucomi is for P6 cores only, not K6-2 then... |
3064 |
|
|
emit_byte(0xdb); |
3065 |
|
|
emit_byte(0xe9+p); /* fucomi them */ |
3066 |
|
|
} |
3067 |
|
|
else { |
3068 |
|
|
emit_byte(0xdd); |
3069 |
|
|
emit_byte(0xe1+p); /* fucom them */ |
3070 |
|
|
emit_byte(0x9b); |
3071 |
|
|
emit_byte(0xdf); |
3072 |
|
|
emit_byte(0xe0); /* fstsw ax */ |
3073 |
|
|
raw_sahf(0); /* sahf */ |
3074 |
|
|
} |
3075 |
|
|
emit_byte(0xdd); |
3076 |
|
|
emit_byte(0xd9+p); /* store value back, and get rid of 0 */ |
3077 |
|
|
} |