| blob | 54d078a1330b451d29c73ac2254a7ae96ef6d141 |
1 .section ".text",#alloc,#execinstr
3 .global bn_mul_mont_fpu
4 .align 32
5 bn_mul_mont_fpu:
6 save %sp,-128-64,%sp
8 cmp %i5,4
9 bl,a,pn %icc,.Lret
10 clr %i0
11 andcc %i5,1,%g0 ! %i5 has to be even...
12 bnz,a,pn %icc,.Lret
13 clr %i0 ! signal "unsupported input value"
15 srl %i5,1,%i5
16 sethi %hi(0xffff),%l7
17 ld [%i4+0],%g4 ! %g4 reassigned, remember?
18 or %l7,%lo(0xffff),%l7
19 ld [%i4+4],%o0
20 sllx %o0,32,%o0
21 or %o0,%g4,%g4 ! %g4=n0[1].n0[0]
23 sll %i5,3,%i5 ! num*=8
25 add %sp,0,%o0 ! real top of stack
26 sll %i5,2,%o1
27 add %o1,%i5,%o1 ! %o1=num*5
28 sub %o0,%o1,%o0
29 and %o0,-2048,%o0 ! optimize TLB utilization
30 sub %o0,0,%sp ! alloca(5*num*8)
32 rd %asi,%o7 ! save %asi
33 add %sp,0+128+64,%l0
34 add %l0,%i5,%l1
35 add %l1,%i5,%l1 ! [an]p_[lh] point at the vectors' ends !
36 add %l1,%i5,%l2
37 add %l2,%i5,%l3
38 add %l3,%i5,%l4
40 wr %g0,210,%asi ! setup %asi for 16-bit FP loads
42 add %i0,%i5,%i0 ! readjust input pointers to point
43 add %i1,%i5,%i1 ! at the ends too...
44 add %i2,%i5,%i2
45 add %i3,%i5,%i3
47 stx %o7,[%sp+0+128+48] ! save %asi
48 \f
49 sub %g0,%i5,%l5 ! i=-num
50 sub %g0,%i5,%l6 ! j=-num
52 add %i1,%l6,%o3
53 add %i2,%l5,%o4
55 ld [%o3+4],%g1 ! bp[0]
56 ld [%o3+0],%o0
57 ld [%o4+4],%g5 ! ap[0]
58 sllx %g1,32,%g1
59 ld [%o4+0],%o1
60 sllx %g5,32,%g5
61 or %g1,%o0,%o0
62 or %g5,%o1,%o1
64 add %i3,%l6,%o5
66 mulx %o1,%o0,%o0 ! ap[0]*bp[0]
67 mulx %g4,%o0,%o0 ! ap[0]*bp[0]*n0
68 stx %o0,[%sp+0+128+0]
70 ld [%o3+0],%f17 ! load a[j] as pair of 32-bit words
71 .word 0xa1b00c20 ! fzeros %f16
72 ld [%o3+4],%f19
73 .word 0xa5b00c20 ! fzeros %f18
74 ld [%o5+0],%f21 ! load n[j] as pair of 32-bit words
75 .word 0xa9b00c20 ! fzeros %f20
76 ld [%o5+4],%f23
77 .word 0xadb00c20 ! fzeros %f22
79 ! transfer b[i] to FPU as 4x16-bit values
80 ldda [%o4+2]%asi,%f0
81 fxtod %f16,%f16
82 ldda [%o4+0]%asi,%f2
83 fxtod %f18,%f18
84 ldda [%o4+6]%asi,%f4
85 fxtod %f20,%f20
86 ldda [%o4+4]%asi,%f6
87 fxtod %f22,%f22
89 ! transfer ap[0]*b[0]*n0 to FPU as 4x16-bit values
90 ldda [%sp+0+128+6]%asi,%f8
91 fxtod %f0,%f0
92 ldda [%sp+0+128+4]%asi,%f10
93 fxtod %f2,%f2
94 ldda [%sp+0+128+2]%asi,%f12
95 fxtod %f4,%f4
96 ldda [%sp+0+128+0]%asi,%f14
97 fxtod %f6,%f6
99 std %f16,[%l1+%l6] ! save smashed ap[j] in double format
100 fxtod %f8,%f8
101 std %f18,[%l2+%l6]
102 fxtod %f10,%f10
103 std %f20,[%l3+%l6] ! save smashed np[j] in double format
104 fxtod %f12,%f12
105 std %f22,[%l4+%l6]
106 fxtod %f14,%f14
108 fmuld %f16,%f0,%f32
109 fmuld %f20,%f8,%f48
110 fmuld %f16,%f2,%f34
111 fmuld %f20,%f10,%f50
112 fmuld %f16,%f4,%f36
113 faddd %f32,%f48,%f48
114 fmuld %f20,%f12,%f52
115 fmuld %f16,%f6,%f38
116 faddd %f34,%f50,%f50
117 fmuld %f20,%f14,%f54
118 fmuld %f18,%f0,%f40
119 faddd %f36,%f52,%f52
120 fmuld %f22,%f8,%f56
121 fmuld %f18,%f2,%f42
122 faddd %f38,%f54,%f54
123 fmuld %f22,%f10,%f58
124 fmuld %f18,%f4,%f44
125 faddd %f40,%f56,%f56
126 fmuld %f22,%f12,%f60
127 fmuld %f18,%f6,%f46
128 faddd %f42,%f58,%f58
129 fmuld %f22,%f14,%f62
131 faddd %f44,%f60,%f24 ! %f60
132 faddd %f46,%f62,%f26 ! %f62
134 faddd %f52,%f56,%f52
135 faddd %f54,%f58,%f54
137 fdtox %f48,%f48
138 fdtox %f50,%f50
139 fdtox %f52,%f52
140 fdtox %f54,%f54
142 std %f48,[%sp+0+128+0]
143 add %l6,8,%l6
144 std %f50,[%sp+0+128+8]
145 add %i1,%l6,%o4
146 std %f52,[%sp+0+128+16]
147 add %i3,%l6,%o5
148 std %f54,[%sp+0+128+24]
149 \f
150 ld [%o4+0],%f17 ! load a[j] as pair of 32-bit words
151 .word 0xa1b00c20 ! fzeros %f16
152 ld [%o4+4],%f19
153 .word 0xa5b00c20 ! fzeros %f18
154 ld [%o5+0],%f21 ! load n[j] as pair of 32-bit words
155 .word 0xa9b00c20 ! fzeros %f20
156 ld [%o5+4],%f23
157 .word 0xadb00c20 ! fzeros %f22
159 fxtod %f16,%f16
160 fxtod %f18,%f18
161 fxtod %f20,%f20
162 fxtod %f22,%f22
164 ldx [%sp+0+128+0],%o0
165 fmuld %f16,%f0,%f32
166 ldx [%sp+0+128+8],%o1
167 fmuld %f20,%f8,%f48
168 ldx [%sp+0+128+16],%o2
169 fmuld %f16,%f2,%f34
170 ldx [%sp+0+128+24],%o3
171 fmuld %f20,%f10,%f50
173 srlx %o0,16,%o7
174 std %f16,[%l1+%l6] ! save smashed ap[j] in double format
175 fmuld %f16,%f4,%f36
176 add %o7,%o1,%o1
177 std %f18,[%l2+%l6]
178 faddd %f32,%f48,%f48
179 fmuld %f20,%f12,%f52
180 srlx %o1,16,%o7
181 std %f20,[%l3+%l6] ! save smashed np[j] in double format
182 fmuld %f16,%f6,%f38
183 add %o7,%o2,%o2
184 std %f22,[%l4+%l6]
185 faddd %f34,%f50,%f50
186 fmuld %f20,%f14,%f54
187 srlx %o2,16,%o7
188 fmuld %f18,%f0,%f40
189 add %o7,%o3,%o3 ! %o3.%o2[0..15].%o1[0..15].%o0[0..15]
190 faddd %f36,%f52,%f52
191 fmuld %f22,%f8,%f56
192 !and %o0,%l7,%o0
193 !and %o1,%l7,%o1
194 !and %o2,%l7,%o2
195 !sllx %o1,16,%o1
196 !sllx %o2,32,%o2
197 !sllx %o3,48,%o7
198 !or %o1,%o0,%o0
199 !or %o2,%o0,%o0
200 !or %o7,%o0,%o0 ! 64-bit result
201 srlx %o3,16,%g1 ! 34-bit carry
202 fmuld %f18,%f2,%f42
204 faddd %f38,%f54,%f54
205 fmuld %f22,%f10,%f58
206 fmuld %f18,%f4,%f44
207 faddd %f40,%f56,%f56
208 fmuld %f22,%f12,%f60
209 fmuld %f18,%f6,%f46
210 faddd %f42,%f58,%f58
211 fmuld %f22,%f14,%f62
213 faddd %f24,%f48,%f48
214 faddd %f26,%f50,%f50
215 faddd %f44,%f60,%f24 ! %f60
216 faddd %f46,%f62,%f26 ! %f62
218 faddd %f52,%f56,%f52
219 faddd %f54,%f58,%f54
221 fdtox %f48,%f48
222 fdtox %f50,%f50
223 fdtox %f52,%f52
224 fdtox %f54,%f54
226 std %f48,[%sp+0+128+0]
227 std %f50,[%sp+0+128+8]
228 addcc %l6,8,%l6
229 std %f52,[%sp+0+128+16]
230 bz,pn %icc,.L1stskip
231 std %f54,[%sp+0+128+24]
232 \f
233 .align 32 ! incidentally already aligned !
234 .L1st:
235 add %i1,%l6,%o4
236 add %i3,%l6,%o5
237 ld [%o4+0],%f17 ! load a[j] as pair of 32-bit words
238 .word 0xa1b00c20 ! fzeros %f16
239 ld [%o4+4],%f19
240 .word 0xa5b00c20 ! fzeros %f18
241 ld [%o5+0],%f21 ! load n[j] as pair of 32-bit words
242 .word 0xa9b00c20 ! fzeros %f20
243 ld [%o5+4],%f23
244 .word 0xadb00c20 ! fzeros %f22
246 fxtod %f16,%f16
247 fxtod %f18,%f18
248 fxtod %f20,%f20
249 fxtod %f22,%f22
251 ldx [%sp+0+128+0],%o0
252 fmuld %f16,%f0,%f32
253 ldx [%sp+0+128+8],%o1
254 fmuld %f20,%f8,%f48
255 ldx [%sp+0+128+16],%o2
256 fmuld %f16,%f2,%f34
257 ldx [%sp+0+128+24],%o3
258 fmuld %f20,%f10,%f50
260 srlx %o0,16,%o7
261 std %f16,[%l1+%l6] ! save smashed ap[j] in double format
262 fmuld %f16,%f4,%f36
263 add %o7,%o1,%o1
264 std %f18,[%l2+%l6]
265 faddd %f32,%f48,%f48
266 fmuld %f20,%f12,%f52
267 srlx %o1,16,%o7
268 std %f20,[%l3+%l6] ! save smashed np[j] in double format
269 fmuld %f16,%f6,%f38
270 add %o7,%o2,%o2
271 std %f22,[%l4+%l6]
272 faddd %f34,%f50,%f50
273 fmuld %f20,%f14,%f54
274 srlx %o2,16,%o7
275 fmuld %f18,%f0,%f40
276 add %o7,%o3,%o3 ! %o3.%o2[0..15].%o1[0..15].%o0[0..15]
277 and %o0,%l7,%o0
278 faddd %f36,%f52,%f52
279 fmuld %f22,%f8,%f56
280 and %o1,%l7,%o1
281 and %o2,%l7,%o2
282 fmuld %f18,%f2,%f42
283 sllx %o1,16,%o1
284 faddd %f38,%f54,%f54
285 fmuld %f22,%f10,%f58
286 sllx %o2,32,%o2
287 fmuld %f18,%f4,%f44
288 sllx %o3,48,%o7
289 or %o1,%o0,%o0
290 faddd %f40,%f56,%f56
291 fmuld %f22,%f12,%f60
292 or %o2,%o0,%o0
293 fmuld %f18,%f6,%f46
294 or %o7,%o0,%o0 ! 64-bit result
295 faddd %f42,%f58,%f58
296 fmuld %f22,%f14,%f62
297 addcc %g1,%o0,%o0
298 faddd %f24,%f48,%f48
299 srlx %o3,16,%g1 ! 34-bit carry
300 faddd %f26,%f50,%f50
301 bcs,a %xcc,.+8
302 add %g1,1,%g1
304 stx %o0,[%l0] ! tp[j-1]=
306 faddd %f44,%f60,%f24 ! %f60
307 faddd %f46,%f62,%f26 ! %f62
309 faddd %f52,%f56,%f52
310 faddd %f54,%f58,%f54
312 fdtox %f48,%f48
313 fdtox %f50,%f50
314 fdtox %f52,%f52
315 fdtox %f54,%f54
317 std %f48,[%sp+0+128+0]
318 std %f50,[%sp+0+128+8]
319 std %f52,[%sp+0+128+16]
320 std %f54,[%sp+0+128+24]
322 addcc %l6,8,%l6
323 bnz,pt %icc,.L1st
324 add %l0,8,%l0
325 \f
326 .L1stskip:
327 fdtox %f24,%f24
328 fdtox %f26,%f26
330 ldx [%sp+0+128+0],%o0
331 ldx [%sp+0+128+8],%o1
332 ldx [%sp+0+128+16],%o2
333 ldx [%sp+0+128+24],%o3
335 srlx %o0,16,%o7
336 std %f24,[%sp+0+128+32]
337 add %o7,%o1,%o1
338 std %f26,[%sp+0+128+40]
339 srlx %o1,16,%o7
340 add %o7,%o2,%o2
341 srlx %o2,16,%o7
342 add %o7,%o3,%o3 ! %o3.%o2[0..15].%o1[0..15].%o0[0..15]
343 and %o0,%l7,%o0
344 and %o1,%l7,%o1
345 and %o2,%l7,%o2
346 sllx %o1,16,%o1
347 sllx %o2,32,%o2
348 sllx %o3,48,%o7
349 or %o1,%o0,%o0
350 or %o2,%o0,%o0
351 or %o7,%o0,%o0 ! 64-bit result
352 ldx [%sp+0+128+32],%o4
353 addcc %g1,%o0,%o0
354 ldx [%sp+0+128+40],%o5
355 srlx %o3,16,%g1 ! 34-bit carry
356 bcs,a %xcc,.+8
357 add %g1,1,%g1
359 stx %o0,[%l0] ! tp[j-1]=
360 add %l0,8,%l0
362 srlx %o4,16,%o7
363 add %o7,%o5,%o5
364 and %o4,%l7,%o4
365 sllx %o5,16,%o7
366 or %o7,%o4,%o4
367 addcc %g1,%o4,%o4
368 srlx %o5,48,%g1
369 bcs,a %xcc,.+8
370 add %g1,1,%g1
372 mov %g1,%i4
373 stx %o4,[%l0] ! tp[num-1]=
374 \f
375 ba .Louter
376 add %l5,8,%l5
377 .align 32
378 .Louter:
379 sub %g0,%i5,%l6 ! j=-num
380 add %sp,0+128+64,%l0
382 add %i1,%l6,%o3
383 add %i2,%l5,%o4
385 ld [%o3+4],%g1 ! bp[i]
386 ld [%o3+0],%o0
387 ld [%o4+4],%g5 ! ap[0]
388 sllx %g1,32,%g1
389 ld [%o4+0],%o1
390 sllx %g5,32,%g5
391 or %g1,%o0,%o0
392 or %g5,%o1,%o1
394 ldx [%l0],%o2 ! tp[0]
395 mulx %o1,%o0,%o0
396 addcc %o2,%o0,%o0
397 mulx %g4,%o0,%o0 ! (ap[0]*bp[i]+t[0])*n0
398 stx %o0,[%sp+0+128+0]
400 ! transfer b[i] to FPU as 4x16-bit values
401 ldda [%o4+2]%asi,%f0
402 ldda [%o4+0]%asi,%f2
403 ldda [%o4+6]%asi,%f4
404 ldda [%o4+4]%asi,%f6
406 ! transfer (ap[0]*b[i]+t[0])*n0 to FPU as 4x16-bit values
407 ldda [%sp+0+128+6]%asi,%f8
408 fxtod %f0,%f0
409 ldda [%sp+0+128+4]%asi,%f10
410 fxtod %f2,%f2
411 ldda [%sp+0+128+2]%asi,%f12
412 fxtod %f4,%f4
413 ldda [%sp+0+128+0]%asi,%f14
414 fxtod %f6,%f6
415 ldd [%l1+%l6],%f16 ! load a[j] in double format
416 fxtod %f8,%f8
417 ldd [%l2+%l6],%f18
418 fxtod %f10,%f10
419 ldd [%l3+%l6],%f20 ! load n[j] in double format
420 fxtod %f12,%f12
421 ldd [%l4+%l6],%f22
422 fxtod %f14,%f14
424 fmuld %f16,%f0,%f32
425 fmuld %f20,%f8,%f48
426 fmuld %f16,%f2,%f34
427 fmuld %f20,%f10,%f50
428 fmuld %f16,%f4,%f36
429 faddd %f32,%f48,%f48
430 fmuld %f20,%f12,%f52
431 fmuld %f16,%f6,%f38
432 faddd %f34,%f50,%f50
433 fmuld %f20,%f14,%f54
434 fmuld %f18,%f0,%f40
435 faddd %f36,%f52,%f52
436 fmuld %f22,%f8,%f56
437 fmuld %f18,%f2,%f42
438 faddd %f38,%f54,%f54
439 fmuld %f22,%f10,%f58
440 fmuld %f18,%f4,%f44
441 faddd %f40,%f56,%f56
442 fmuld %f22,%f12,%f60
443 fmuld %f18,%f6,%f46
444 faddd %f42,%f58,%f58
445 fmuld %f22,%f14,%f62
447 faddd %f44,%f60,%f24 ! %f60
448 faddd %f46,%f62,%f26 ! %f62
450 faddd %f52,%f56,%f52
451 faddd %f54,%f58,%f54
453 fdtox %f48,%f48
454 fdtox %f50,%f50
455 fdtox %f52,%f52
456 fdtox %f54,%f54
458 std %f48,[%sp+0+128+0]
459 std %f50,[%sp+0+128+8]
460 std %f52,[%sp+0+128+16]
461 add %l6,8,%l6
462 std %f54,[%sp+0+128+24]
463 \f
464 ldd [%l1+%l6],%f16 ! load a[j] in double format
465 ldd [%l2+%l6],%f18
466 ldd [%l3+%l6],%f20 ! load n[j] in double format
467 ldd [%l4+%l6],%f22
469 fmuld %f16,%f0,%f32
470 fmuld %f20,%f8,%f48
471 fmuld %f16,%f2,%f34
472 fmuld %f20,%f10,%f50
473 fmuld %f16,%f4,%f36
474 ldx [%sp+0+128+0],%o0
475 faddd %f32,%f48,%f48
476 fmuld %f20,%f12,%f52
477 ldx [%sp+0+128+8],%o1
478 fmuld %f16,%f6,%f38
479 ldx [%sp+0+128+16],%o2
480 faddd %f34,%f50,%f50
481 fmuld %f20,%f14,%f54
482 ldx [%sp+0+128+24],%o3
483 fmuld %f18,%f0,%f40
485 srlx %o0,16,%o7
486 faddd %f36,%f52,%f52
487 fmuld %f22,%f8,%f56
488 add %o7,%o1,%o1
489 fmuld %f18,%f2,%f42
490 srlx %o1,16,%o7
491 faddd %f38,%f54,%f54
492 fmuld %f22,%f10,%f58
493 add %o7,%o2,%o2
494 fmuld %f18,%f4,%f44
495 srlx %o2,16,%o7
496 faddd %f40,%f56,%f56
497 fmuld %f22,%f12,%f60
498 add %o7,%o3,%o3 ! %o3.%o2[0..15].%o1[0..15].%o0[0..15]
499 ! why?
500 and %o0,%l7,%o0
501 fmuld %f18,%f6,%f46
502 and %o1,%l7,%o1
503 and %o2,%l7,%o2
504 faddd %f42,%f58,%f58
505 fmuld %f22,%f14,%f62
506 sllx %o1,16,%o1
507 faddd %f24,%f48,%f48
508 sllx %o2,32,%o2
509 faddd %f26,%f50,%f50
510 sllx %o3,48,%o7
511 or %o1,%o0,%o0
512 faddd %f44,%f60,%f24 ! %f60
513 or %o2,%o0,%o0
514 faddd %f46,%f62,%f26 ! %f62
515 or %o7,%o0,%o0 ! 64-bit result
516 ldx [%l0],%o7
517 faddd %f52,%f56,%f52
518 addcc %o7,%o0,%o0
519 ! end-of-why?
520 faddd %f54,%f58,%f54
521 srlx %o3,16,%g1 ! 34-bit carry
522 fdtox %f48,%f48
523 bcs,a %xcc,.+8
524 add %g1,1,%g1
526 fdtox %f50,%f50
527 fdtox %f52,%f52
528 fdtox %f54,%f54
530 std %f48,[%sp+0+128+0]
531 std %f50,[%sp+0+128+8]
532 addcc %l6,8,%l6
533 std %f52,[%sp+0+128+16]
534 bz,pn %icc,.Linnerskip
535 std %f54,[%sp+0+128+24]
536 \f
537 ba .Linner
538 nop
539 .align 32
540 .Linner:
541 ldd [%l1+%l6],%f16 ! load a[j] in double format
542 ldd [%l2+%l6],%f18
543 ldd [%l3+%l6],%f20 ! load n[j] in double format
544 ldd [%l4+%l6],%f22
546 fmuld %f16,%f0,%f32
547 fmuld %f20,%f8,%f48
548 fmuld %f16,%f2,%f34
549 fmuld %f20,%f10,%f50
550 fmuld %f16,%f4,%f36
551 ldx [%sp+0+128+0],%o0
552 faddd %f32,%f48,%f48
553 fmuld %f20,%f12,%f52
554 ldx [%sp+0+128+8],%o1
555 fmuld %f16,%f6,%f38
556 ldx [%sp+0+128+16],%o2
557 faddd %f34,%f50,%f50
558 fmuld %f20,%f14,%f54
559 ldx [%sp+0+128+24],%o3
560 fmuld %f18,%f0,%f40
562 srlx %o0,16,%o7
563 faddd %f36,%f52,%f52
564 fmuld %f22,%f8,%f56
565 add %o7,%o1,%o1
566 fmuld %f18,%f2,%f42
567 srlx %o1,16,%o7
568 faddd %f38,%f54,%f54
569 fmuld %f22,%f10,%f58
570 add %o7,%o2,%o2
571 fmuld %f18,%f4,%f44
572 srlx %o2,16,%o7
573 faddd %f40,%f56,%f56
574 fmuld %f22,%f12,%f60
575 add %o7,%o3,%o3 ! %o3.%o2[0..15].%o1[0..15].%o0[0..15]
576 and %o0,%l7,%o0
577 fmuld %f18,%f6,%f46
578 and %o1,%l7,%o1
579 and %o2,%l7,%o2
580 faddd %f42,%f58,%f58
581 fmuld %f22,%f14,%f62
582 sllx %o1,16,%o1
583 faddd %f24,%f48,%f48
584 sllx %o2,32,%o2
585 faddd %f26,%f50,%f50
586 sllx %o3,48,%o7
587 or %o1,%o0,%o0
588 faddd %f44,%f60,%f24 ! %f60
589 or %o2,%o0,%o0
590 faddd %f46,%f62,%f26 ! %f62
591 or %o7,%o0,%o0 ! 64-bit result
592 faddd %f52,%f56,%f52
593 addcc %g1,%o0,%o0
594 ldx [%l0+8],%o7 ! tp[j]
595 faddd %f54,%f58,%f54
596 srlx %o3,16,%g1 ! 34-bit carry
597 fdtox %f48,%f48
598 bcs,a %xcc,.+8
599 add %g1,1,%g1
600 fdtox %f50,%f50
601 addcc %o7,%o0,%o0
602 fdtox %f52,%f52
603 bcs,a %xcc,.+8
604 add %g1,1,%g1
606 stx %o0,[%l0] ! tp[j-1]
607 fdtox %f54,%f54
609 std %f48,[%sp+0+128+0]
610 std %f50,[%sp+0+128+8]
611 std %f52,[%sp+0+128+16]
612 addcc %l6,8,%l6
613 std %f54,[%sp+0+128+24]
614 bnz,pt %icc,.Linner
615 add %l0,8,%l0
616 \f
617 .Linnerskip:
618 fdtox %f24,%f24
619 fdtox %f26,%f26
621 ldx [%sp+0+128+0],%o0
622 ldx [%sp+0+128+8],%o1
623 ldx [%sp+0+128+16],%o2
624 ldx [%sp+0+128+24],%o3
626 srlx %o0,16,%o7
627 std %f24,[%sp+0+128+32]
628 add %o7,%o1,%o1
629 std %f26,[%sp+0+128+40]
630 srlx %o1,16,%o7
631 add %o7,%o2,%o2
632 srlx %o2,16,%o7
633 add %o7,%o3,%o3 ! %o3.%o2[0..15].%o1[0..15].%o0[0..15]
634 and %o0,%l7,%o0
635 and %o1,%l7,%o1
636 and %o2,%l7,%o2
637 sllx %o1,16,%o1
638 sllx %o2,32,%o2
639 sllx %o3,48,%o7
640 or %o1,%o0,%o0
641 or %o2,%o0,%o0
642 ldx [%sp+0+128+32],%o4
643 or %o7,%o0,%o0 ! 64-bit result
644 ldx [%sp+0+128+40],%o5
645 addcc %g1,%o0,%o0
646 ldx [%l0+8],%o7 ! tp[j]
647 srlx %o3,16,%g1 ! 34-bit carry
648 bcs,a %xcc,.+8
649 add %g1,1,%g1
651 addcc %o7,%o0,%o0
652 bcs,a %xcc,.+8
653 add %g1,1,%g1
655 stx %o0,[%l0] ! tp[j-1]
656 add %l0,8,%l0
658 srlx %o4,16,%o7
659 add %o7,%o5,%o5
660 and %o4,%l7,%o4
661 sllx %o5,16,%o7
662 or %o7,%o4,%o4
663 addcc %g1,%o4,%o4
664 srlx %o5,48,%g1
665 bcs,a %xcc,.+8
666 add %g1,1,%g1
668 addcc %i4,%o4,%o4
669 stx %o4,[%l0] ! tp[num-1]
670 mov %g1,%i4
671 bcs,a %xcc,.+8
672 add %i4,1,%i4
674 addcc %l5,8,%l5
675 bnz %icc,.Louter
676 nop
677 \f
678 add %l0,8,%l0 ! adjust tp to point at the end
679 orn %g0,%g0,%g4
680 sub %g0,%i5,%o7 ! n=-num
681 ba .Lsub
682 subcc %g0,%g0,%g0 ! clear %icc.c
684 .align 32
685 .Lsub:
686 ldx [%l0+%o7],%o0
687 add %i3,%o7,%g1
688 ld [%g1+0],%o2
689 ld [%g1+4],%o3
690 srlx %o0,32,%o1
691 subccc %o0,%o2,%o2
692 add %i0,%o7,%g1
693 subccc %o1,%o3,%o3
694 st %o2,[%g1+0]
695 add %o7,8,%o7
696 brnz,pt %o7,.Lsub
697 st %o3,[%g1+4]
698 subc %i4,0,%g4
699 sub %g0,%i5,%o7 ! n=-num
700 ba .Lcopy
701 nop
703 .align 32
704 .Lcopy:
705 ldx [%l0+%o7],%o0
706 add %i0,%o7,%g1
707 ld [%g1+0],%o2
708 ld [%g1+4],%o3
709 stx %g0,[%l0+%o7]
710 and %o0,%g4,%o0
711 srlx %o0,32,%o1
712 andn %o2,%g4,%o2
713 andn %o3,%g4,%o3
714 or %o2,%o0,%o0
715 or %o3,%o1,%o1
716 st %o0,[%g1+0]
717 add %o7,8,%o7
718 brnz,pt %o7,.Lcopy
719 st %o1,[%g1+4]
720 sub %g0,%i5,%o7 ! n=-num
722 .Lzap:
723 stx %g0,[%l1+%o7]
724 stx %g0,[%l2+%o7]
725 stx %g0,[%l3+%o7]
726 stx %g0,[%l4+%o7]
727 add %o7,8,%o7
728 brnz,pt %o7,.Lzap
729 nop
731 ldx [%sp+0+128+48],%o7
732 wr %g0,%o7,%asi ! restore %asi
734 mov 1,%i0
735 .Lret:
736 ret
737 restore
738 .type bn_mul_mont_fpu,#function
739 .size bn_mul_mont_fpu,(.-bn_mul_mont_fpu)
740 .asciz "Montgomery Multipltication for UltraSPARC, CRYPTOGAMS by <appro@openssl.org>"
741 .align 32