1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
|
/*
* mpi_x86_asm.c - MSVC inline assembly implementation of s_mpv_ functions.
*
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#include "mpi-priv.h"
static int is_sse = -1;
extern unsigned long s_mpi_is_sse2();
/*
* ebp - 36: caller's esi
* ebp - 32: caller's edi
* ebp - 28:
* ebp - 24:
* ebp - 20:
* ebp - 16:
* ebp - 12:
* ebp - 8:
* ebp - 4:
* ebp + 0: caller's ebp
* ebp + 4: return address
* ebp + 8: a argument
* ebp + 12: a_len argument
* ebp + 16: b argument
* ebp + 20: c argument
* registers:
* eax:
* ebx: carry
* ecx: a_len
* edx:
* esi: a ptr
* edi: c ptr
*/
__declspec(naked) void s_mpv_mul_d(const mp_digit *a, mp_size a_len, mp_digit b, mp_digit *c)
{
__asm {
mov eax, is_sse
cmp eax, 0
je s_mpv_mul_d_x86
jg s_mpv_mul_d_sse2
call s_mpi_is_sse2
mov is_sse, eax
cmp eax, 0
jg s_mpv_mul_d_sse2
s_mpv_mul_d_x86:
push ebp
mov ebp,esp
sub esp,28
push edi
push esi
push ebx
mov ebx,0 ; carry = 0
mov ecx,[ebp+12] ; ecx = a_len
mov edi,[ebp+20]
cmp ecx,0
je L_2 ; jmp if a_len == 0
mov esi,[ebp+8] ; esi = a
cld
L_1:
lodsd ; eax = [ds:esi]; esi += 4
mov edx,[ebp+16] ; edx = b
mul edx ; edx:eax = Phi:Plo = a_i * b
add eax,ebx ; add carry (ebx) to edx:eax
adc edx,0
mov ebx,edx ; high half of product becomes next carry
stosd ; [es:edi] = ax; edi += 4;
dec ecx ; --a_len
jnz L_1 ; jmp if a_len != 0
L_2:
mov [edi],ebx ; *c = carry
pop ebx
pop esi
pop edi
leave
ret
nop
s_mpv_mul_d_sse2:
push ebp
mov ebp, esp
push edi
push esi
psubq mm2, mm2 ; carry = 0
mov ecx, [ebp+12] ; ecx = a_len
movd mm1, [ebp+16] ; mm1 = b
mov edi, [ebp+20]
cmp ecx, 0
je L_6 ; jmp if a_len == 0
mov esi, [ebp+8] ; esi = a
cld
L_5:
movd mm0, [esi] ; mm0 = *a++
add esi, 4
pmuludq mm0, mm1 ; mm0 = b * *a++
paddq mm2, mm0 ; add the carry
movd [edi], mm2 ; store the 32bit result
add edi, 4
psrlq mm2, 32 ; save the carry
dec ecx ; --a_len
jnz L_5 ; jmp if a_len != 0
L_6:
movd [edi], mm2 ; *c = carry
emms
pop esi
pop edi
leave
ret
nop
}
}
/*
* ebp - 36: caller's esi
* ebp - 32: caller's edi
* ebp - 28:
* ebp - 24:
* ebp - 20:
* ebp - 16:
* ebp - 12:
* ebp - 8:
* ebp - 4:
* ebp + 0: caller's ebp
* ebp + 4: return address
* ebp + 8: a argument
* ebp + 12: a_len argument
* ebp + 16: b argument
* ebp + 20: c argument
* registers:
* eax:
* ebx: carry
* ecx: a_len
* edx:
* esi: a ptr
* edi: c ptr
*/
__declspec(naked) void s_mpv_mul_d_add(const mp_digit *a, mp_size a_len, mp_digit b, mp_digit *c)
{
__asm {
mov eax, is_sse
cmp eax, 0
je s_mpv_mul_d_add_x86
jg s_mpv_mul_d_add_sse2
call s_mpi_is_sse2
mov is_sse, eax
cmp eax, 0
jg s_mpv_mul_d_add_sse2
s_mpv_mul_d_add_x86:
push ebp
mov ebp,esp
sub esp,28
push edi
push esi
push ebx
mov ebx,0 ; carry = 0
mov ecx,[ebp+12] ; ecx = a_len
mov edi,[ebp+20]
cmp ecx,0
je L_11 ; jmp if a_len == 0
mov esi,[ebp+8] ; esi = a
cld
L_10:
lodsd ; eax = [ds:esi]; esi += 4
mov edx,[ebp+16] ; edx = b
mul edx ; edx:eax = Phi:Plo = a_i * b
add eax,ebx ; add carry (ebx) to edx:eax
adc edx,0
mov ebx,[edi] ; add in current word from *c
add eax,ebx
adc edx,0
mov ebx,edx ; high half of product becomes next carry
stosd ; [es:edi] = ax; edi += 4;
dec ecx ; --a_len
jnz L_10 ; jmp if a_len != 0
L_11:
mov [edi],ebx ; *c = carry
pop ebx
pop esi
pop edi
leave
ret
nop
s_mpv_mul_d_add_sse2:
push ebp
mov ebp, esp
push edi
push esi
psubq mm2, mm2 ; carry = 0
mov ecx, [ebp+12] ; ecx = a_len
movd mm1, [ebp+16] ; mm1 = b
mov edi, [ebp+20]
cmp ecx, 0
je L_16 ; jmp if a_len == 0
mov esi, [ebp+8] ; esi = a
cld
L_15:
movd mm0, [esi] ; mm0 = *a++
add esi, 4
pmuludq mm0, mm1 ; mm0 = b * *a++
paddq mm2, mm0 ; add the carry
movd mm0, [edi]
paddq mm2, mm0 ; add the carry
movd [edi], mm2 ; store the 32bit result
add edi, 4
psrlq mm2, 32 ; save the carry
dec ecx ; --a_len
jnz L_15 ; jmp if a_len != 0
L_16:
movd [edi], mm2 ; *c = carry
emms
pop esi
pop edi
leave
ret
nop
}
}
/*
* ebp - 36: caller's esi
* ebp - 32: caller's edi
* ebp - 28:
* ebp - 24:
* ebp - 20:
* ebp - 16:
* ebp - 12:
* ebp - 8:
* ebp - 4:
* ebp + 0: caller's ebp
* ebp + 4: return address
* ebp + 8: a argument
* ebp + 12: a_len argument
* ebp + 16: b argument
* ebp + 20: c argument
* registers:
* eax:
* ebx: carry
* ecx: a_len
* edx:
* esi: a ptr
* edi: c ptr
*/
__declspec(naked) void s_mpv_mul_d_add_prop(const mp_digit *a, mp_size a_len, mp_digit b, mp_digit *c)
{
__asm {
mov eax, is_sse
cmp eax, 0
je s_mpv_mul_d_add_prop_x86
jg s_mpv_mul_d_add_prop_sse2
call s_mpi_is_sse2
mov is_sse, eax
cmp eax, 0
jg s_mpv_mul_d_add_prop_sse2
s_mpv_mul_d_add_prop_x86:
push ebp
mov ebp,esp
sub esp,28
push edi
push esi
push ebx
mov ebx,0 ; carry = 0
mov ecx,[ebp+12] ; ecx = a_len
mov edi,[ebp+20]
cmp ecx,0
je L_21 ; jmp if a_len == 0
cld
mov esi,[ebp+8] ; esi = a
L_20:
lodsd ; eax = [ds:esi]; esi += 4
mov edx,[ebp+16] ; edx = b
mul edx ; edx:eax = Phi:Plo = a_i * b
add eax,ebx ; add carry (ebx) to edx:eax
adc edx,0
mov ebx,[edi] ; add in current word from *c
add eax,ebx
adc edx,0
mov ebx,edx ; high half of product becomes next carry
stosd ; [es:edi] = ax; edi += 4;
dec ecx ; --a_len
jnz L_20 ; jmp if a_len != 0
L_21:
cmp ebx,0 ; is carry zero?
jz L_23
mov eax,[edi] ; add in current word from *c
add eax,ebx
stosd ; [es:edi] = ax; edi += 4;
jnc L_23
L_22:
mov eax,[edi] ; add in current word from *c
adc eax,0
stosd ; [es:edi] = ax; edi += 4;
jc L_22
L_23:
pop ebx
pop esi
pop edi
leave
ret
nop
s_mpv_mul_d_add_prop_sse2:
push ebp
mov ebp, esp
push edi
push esi
push ebx
psubq mm2, mm2 ; carry = 0
mov ecx, [ebp+12] ; ecx = a_len
movd mm1, [ebp+16] ; mm1 = b
mov edi, [ebp+20]
cmp ecx, 0
je L_26 ; jmp if a_len == 0
mov esi, [ebp+8] ; esi = a
cld
L_25:
movd mm0, [esi] ; mm0 = *a++
movd mm3, [edi] ; fetch the sum
add esi, 4
pmuludq mm0, mm1 ; mm0 = b * *a++
paddq mm2, mm0 ; add the carry
paddq mm2, mm3 ; add *c++
movd [edi], mm2 ; store the 32bit result
add edi, 4
psrlq mm2, 32 ; save the carry
dec ecx ; --a_len
jnz L_25 ; jmp if a_len != 0
L_26:
movd ebx, mm2
cmp ebx, 0 ; is carry zero?
jz L_28
mov eax, [edi]
add eax, ebx
stosd
jnc L_28
L_27:
mov eax, [edi] ; add in current word from *c
adc eax, 0
stosd ; [es:edi] = ax; edi += 4;
jc L_27
L_28:
emms
pop ebx
pop esi
pop edi
leave
ret
nop
}
}
/*
* ebp - 20: caller's esi
* ebp - 16: caller's edi
* ebp - 12:
* ebp - 8: carry
* ebp - 4: a_len local
* ebp + 0: caller's ebp
* ebp + 4: return address
* ebp + 8: pa argument
* ebp + 12: a_len argument
* ebp + 16: ps argument
* ebp + 20:
* registers:
* eax:
* ebx: carry
* ecx: a_len
* edx:
* esi: a ptr
* edi: c ptr
*/
__declspec(naked) void s_mpv_sqr_add_prop(const mp_digit *a, mp_size a_len, mp_digit *sqrs)
{
__asm {
mov eax, is_sse
cmp eax, 0
je s_mpv_sqr_add_prop_x86
jg s_mpv_sqr_add_prop_sse2
call s_mpi_is_sse2
mov is_sse, eax
cmp eax, 0
jg s_mpv_sqr_add_prop_sse2
s_mpv_sqr_add_prop_x86:
push ebp
mov ebp,esp
sub esp,12
push edi
push esi
push ebx
mov ebx,0 ; carry = 0
mov ecx,[ebp+12] ; a_len
mov edi,[ebp+16] ; edi = ps
cmp ecx,0
je L_31 ; jump if a_len == 0
cld
mov esi,[ebp+8] ; esi = pa
L_30:
lodsd ; eax = [ds:si]; si += 4;
mul eax
add eax,ebx ; add "carry"
adc edx,0
mov ebx,[edi]
add eax,ebx ; add low word from result
mov ebx,[edi+4]
stosd ; [es:di] = eax; di += 4;
adc edx,ebx ; add high word from result
mov ebx,0
mov eax,edx
adc ebx,0
stosd ; [es:di] = eax; di += 4;
dec ecx ; --a_len
jnz L_30 ; jmp if a_len != 0
L_31:
cmp ebx,0 ; is carry zero?
jz L_34
mov eax,[edi] ; add in current word from *c
add eax,ebx
stosd ; [es:edi] = ax; edi += 4;
jnc L_34
L_32:
mov eax,[edi] ; add in current word from *c
adc eax,0
stosd ; [es:edi] = ax; edi += 4;
jc L_32
L_34:
pop ebx
pop esi
pop edi
leave
ret
nop
s_mpv_sqr_add_prop_sse2:
push ebp
mov ebp, esp
push edi
push esi
push ebx
psubq mm2, mm2 ; carry = 0
mov ecx, [ebp+12] ; ecx = a_len
mov edi, [ebp+16]
cmp ecx, 0
je L_36 ; jmp if a_len == 0
mov esi, [ebp+8] ; esi = a
cld
L_35:
movd mm0, [esi] ; mm0 = *a
movd mm3, [edi] ; fetch the sum
add esi, 4
pmuludq mm0, mm0 ; mm0 = sqr(a)
paddq mm2, mm0 ; add the carry
paddq mm2, mm3 ; add the low word
movd mm3, [edi+4]
movd [edi], mm2 ; store the 32bit result
psrlq mm2, 32
paddq mm2, mm3 ; add the high word
movd [edi+4], mm2 ; store the 32bit result
psrlq mm2, 32 ; save the carry.
add edi, 8
dec ecx ; --a_len
jnz L_35 ; jmp if a_len != 0
L_36:
movd ebx, mm2
cmp ebx, 0 ; is carry zero?
jz L_38
mov eax, [edi]
add eax, ebx
stosd
jnc L_38
L_37:
mov eax, [edi] ; add in current word from *c
adc eax, 0
stosd ; [es:edi] = ax; edi += 4;
jc L_37
L_38:
emms
pop ebx
pop esi
pop edi
leave
ret
nop
}
}
/*
* Divide 64-bit (Nhi,Nlo) by 32-bit divisor, which must be normalized
* so its high bit is 1. This code is from NSPR.
*
* Dump of assembler code for function s_mpv_div_2dx1d:
*
* esp + 0: Caller's ebx
* esp + 4: return address
* esp + 8: Nhi argument
* esp + 12: Nlo argument
* esp + 16: divisor argument
* esp + 20: qp argument
* esp + 24: rp argument
* registers:
* eax:
* ebx: carry
* ecx: a_len
* edx:
* esi: a ptr
* edi: c ptr
*/
__declspec(naked) mp_err
s_mpv_div_2dx1d(mp_digit Nhi, mp_digit Nlo, mp_digit divisor,
mp_digit *qp, mp_digit *rp)
{
__asm {
push ebx
mov edx,[esp+8]
mov eax,[esp+12]
mov ebx,[esp+16]
div ebx
mov ebx,[esp+20]
mov [ebx],eax
mov ebx,[esp+24]
mov [ebx],edx
xor eax,eax ; return zero
pop ebx
ret
nop
}
}
|
