1 /* SPDX-License-Identifier: GPL-2.0-or-later *
2 /*
3 * Implement AES algorithm in Intel AES-NI ins
4 *
5 * The white paper of AES-NI instructions can
6 * http://softwarecommunity.intel.com/isn/do
7 *
8 * Copyright (C) 2008, Intel Corp.
9 * Author: Huang Ying <ying.huang@intel.com>
10 * Vinodh Gopal <vinodh.gopal@intel.
11 * Kahraman Akdemir
12 *
13 * Copyright (c) 2010, Intel Corporation.
14 *
15 * Ported x86_64 version to x86:
16 * Author: Mathias Krause <minipli@googlemai
17 */
18
19 #include <linux/linkage.h>
20 #include <asm/frame.h>
21
22 #define STATE1 %xmm0
23 #define STATE2 %xmm4
24 #define STATE3 %xmm5
25 #define STATE4 %xmm6
26 #define STATE STATE1
27 #define IN1 %xmm1
28 #define IN2 %xmm7
29 #define IN3 %xmm8
30 #define IN4 %xmm9
31 #define IN IN1
32 #define KEY %xmm2
33 #define IV %xmm3
34
35 #define BSWAP_MASK %xmm10
36 #define CTR %xmm11
37 #define INC %xmm12
38
39 #define GF128MUL_MASK %xmm7
40
41 #ifdef __x86_64__
42 #define AREG %rax
43 #define KEYP %rdi
44 #define OUTP %rsi
45 #define UKEYP OUTP
46 #define INP %rdx
47 #define LEN %rcx
48 #define IVP %r8
49 #define KLEN %r9d
50 #define T1 %r10
51 #define TKEYP T1
52 #define T2 %r11
53 #define TCTR_LOW T2
54 #else
55 #define AREG %eax
56 #define KEYP %edi
57 #define OUTP AREG
58 #define UKEYP OUTP
59 #define INP %edx
60 #define LEN %esi
61 #define IVP %ebp
62 #define KLEN %ebx
63 #define T1 %ecx
64 #define TKEYP T1
65 #endif
66
67 SYM_FUNC_START_LOCAL(_key_expansion_256a)
68 pshufd $0b11111111, %xmm1, %xmm1
69 shufps $0b00010000, %xmm0, %xmm4
70 pxor %xmm4, %xmm0
71 shufps $0b10001100, %xmm0, %xmm4
72 pxor %xmm4, %xmm0
73 pxor %xmm1, %xmm0
74 movaps %xmm0, (TKEYP)
75 add $0x10, TKEYP
76 RET
77 SYM_FUNC_END(_key_expansion_256a)
78 SYM_FUNC_ALIAS_LOCAL(_key_expansion_128, _key_
79
80 SYM_FUNC_START_LOCAL(_key_expansion_192a)
81 pshufd $0b01010101, %xmm1, %xmm1
82 shufps $0b00010000, %xmm0, %xmm4
83 pxor %xmm4, %xmm0
84 shufps $0b10001100, %xmm0, %xmm4
85 pxor %xmm4, %xmm0
86 pxor %xmm1, %xmm0
87
88 movaps %xmm2, %xmm5
89 movaps %xmm2, %xmm6
90 pslldq $4, %xmm5
91 pshufd $0b11111111, %xmm0, %xmm3
92 pxor %xmm3, %xmm2
93 pxor %xmm5, %xmm2
94
95 movaps %xmm0, %xmm1
96 shufps $0b01000100, %xmm0, %xmm6
97 movaps %xmm6, (TKEYP)
98 shufps $0b01001110, %xmm2, %xmm1
99 movaps %xmm1, 0x10(TKEYP)
100 add $0x20, TKEYP
101 RET
102 SYM_FUNC_END(_key_expansion_192a)
103
104 SYM_FUNC_START_LOCAL(_key_expansion_192b)
105 pshufd $0b01010101, %xmm1, %xmm1
106 shufps $0b00010000, %xmm0, %xmm4
107 pxor %xmm4, %xmm0
108 shufps $0b10001100, %xmm0, %xmm4
109 pxor %xmm4, %xmm0
110 pxor %xmm1, %xmm0
111
112 movaps %xmm2, %xmm5
113 pslldq $4, %xmm5
114 pshufd $0b11111111, %xmm0, %xmm3
115 pxor %xmm3, %xmm2
116 pxor %xmm5, %xmm2
117
118 movaps %xmm0, (TKEYP)
119 add $0x10, TKEYP
120 RET
121 SYM_FUNC_END(_key_expansion_192b)
122
123 SYM_FUNC_START_LOCAL(_key_expansion_256b)
124 pshufd $0b10101010, %xmm1, %xmm1
125 shufps $0b00010000, %xmm2, %xmm4
126 pxor %xmm4, %xmm2
127 shufps $0b10001100, %xmm2, %xmm4
128 pxor %xmm4, %xmm2
129 pxor %xmm1, %xmm2
130 movaps %xmm2, (TKEYP)
131 add $0x10, TKEYP
132 RET
133 SYM_FUNC_END(_key_expansion_256b)
134
135 /*
136 * void aesni_set_key(struct crypto_aes_ctx *c
137 * unsigned int key_len)
138 */
139 SYM_FUNC_START(aesni_set_key)
140 FRAME_BEGIN
141 #ifndef __x86_64__
142 pushl KEYP
143 movl (FRAME_OFFSET+8)(%esp), KEYP
144 movl (FRAME_OFFSET+12)(%esp), UKEYP
145 movl (FRAME_OFFSET+16)(%esp), %edx
146 #endif
147 movups (UKEYP), %xmm0 # user
148 movaps %xmm0, (KEYP)
149 lea 0x10(KEYP), TKEYP # key
150 movl %edx, 480(KEYP)
151 pxor %xmm4, %xmm4 # xmm4
152 cmp $24, %dl
153 jb .Lenc_key128
154 je .Lenc_key192
155 movups 0x10(UKEYP), %xmm2 # othe
156 movaps %xmm2, (TKEYP)
157 add $0x10, TKEYP
158 aeskeygenassist $0x1, %xmm2, %xmm1
159 call _key_expansion_256a
160 aeskeygenassist $0x1, %xmm0, %xmm1
161 call _key_expansion_256b
162 aeskeygenassist $0x2, %xmm2, %xmm1
163 call _key_expansion_256a
164 aeskeygenassist $0x2, %xmm0, %xmm1
165 call _key_expansion_256b
166 aeskeygenassist $0x4, %xmm2, %xmm1
167 call _key_expansion_256a
168 aeskeygenassist $0x4, %xmm0, %xmm1
169 call _key_expansion_256b
170 aeskeygenassist $0x8, %xmm2, %xmm1
171 call _key_expansion_256a
172 aeskeygenassist $0x8, %xmm0, %xmm1
173 call _key_expansion_256b
174 aeskeygenassist $0x10, %xmm2, %xmm1
175 call _key_expansion_256a
176 aeskeygenassist $0x10, %xmm0, %xmm1
177 call _key_expansion_256b
178 aeskeygenassist $0x20, %xmm2, %xmm1
179 call _key_expansion_256a
180 aeskeygenassist $0x20, %xmm0, %xmm1
181 call _key_expansion_256b
182 aeskeygenassist $0x40, %xmm2, %xmm1
183 call _key_expansion_256a
184 jmp .Ldec_key
185 .Lenc_key192:
186 movq 0x10(UKEYP), %xmm2 # othe
187 aeskeygenassist $0x1, %xmm2, %xmm1
188 call _key_expansion_192a
189 aeskeygenassist $0x2, %xmm2, %xmm1
190 call _key_expansion_192b
191 aeskeygenassist $0x4, %xmm2, %xmm1
192 call _key_expansion_192a
193 aeskeygenassist $0x8, %xmm2, %xmm1
194 call _key_expansion_192b
195 aeskeygenassist $0x10, %xmm2, %xmm1
196 call _key_expansion_192a
197 aeskeygenassist $0x20, %xmm2, %xmm1
198 call _key_expansion_192b
199 aeskeygenassist $0x40, %xmm2, %xmm1
200 call _key_expansion_192a
201 aeskeygenassist $0x80, %xmm2, %xmm1
202 call _key_expansion_192b
203 jmp .Ldec_key
204 .Lenc_key128:
205 aeskeygenassist $0x1, %xmm0, %xmm1
206 call _key_expansion_128
207 aeskeygenassist $0x2, %xmm0, %xmm1
208 call _key_expansion_128
209 aeskeygenassist $0x4, %xmm0, %xmm1
210 call _key_expansion_128
211 aeskeygenassist $0x8, %xmm0, %xmm1
212 call _key_expansion_128
213 aeskeygenassist $0x10, %xmm0, %xmm1
214 call _key_expansion_128
215 aeskeygenassist $0x20, %xmm0, %xmm1
216 call _key_expansion_128
217 aeskeygenassist $0x40, %xmm0, %xmm1
218 call _key_expansion_128
219 aeskeygenassist $0x80, %xmm0, %xmm1
220 call _key_expansion_128
221 aeskeygenassist $0x1b, %xmm0, %xmm1
222 call _key_expansion_128
223 aeskeygenassist $0x36, %xmm0, %xmm1
224 call _key_expansion_128
225 .Ldec_key:
226 sub $0x10, TKEYP
227 movaps (KEYP), %xmm0
228 movaps (TKEYP), %xmm1
229 movaps %xmm0, 240(TKEYP)
230 movaps %xmm1, 240(KEYP)
231 add $0x10, KEYP
232 lea 240-16(TKEYP), UKEYP
233 .align 4
234 .Ldec_key_loop:
235 movaps (KEYP), %xmm0
236 aesimc %xmm0, %xmm1
237 movaps %xmm1, (UKEYP)
238 add $0x10, KEYP
239 sub $0x10, UKEYP
240 cmp TKEYP, KEYP
241 jb .Ldec_key_loop
242 #ifndef __x86_64__
243 popl KEYP
244 #endif
245 FRAME_END
246 RET
247 SYM_FUNC_END(aesni_set_key)
248
249 /*
250 * void aesni_enc(const void *ctx, u8 *dst, co
251 */
252 SYM_FUNC_START(aesni_enc)
253 FRAME_BEGIN
254 #ifndef __x86_64__
255 pushl KEYP
256 pushl KLEN
257 movl (FRAME_OFFSET+12)(%esp), KEYP
258 movl (FRAME_OFFSET+16)(%esp), OUTP
259 movl (FRAME_OFFSET+20)(%esp), INP
260 #endif
261 movl 480(KEYP), KLEN # key
262 movups (INP), STATE # inpu
263 call _aesni_enc1
264 movups STATE, (OUTP) # outp
265 #ifndef __x86_64__
266 popl KLEN
267 popl KEYP
268 #endif
269 FRAME_END
270 RET
271 SYM_FUNC_END(aesni_enc)
272
273 /*
274 * _aesni_enc1: internal ABI
275 * input:
276 * KEYP: key struct pointer
277 * KLEN: round count
278 * STATE: initial state (input)
279 * output:
280 * STATE: finial state (output)
281 * changed:
282 * KEY
283 * TKEYP (T1)
284 */
285 SYM_FUNC_START_LOCAL(_aesni_enc1)
286 movaps (KEYP), KEY # key
287 mov KEYP, TKEYP
288 pxor KEY, STATE # round 0
289 add $0x30, TKEYP
290 cmp $24, KLEN
291 jb .Lenc128
292 lea 0x20(TKEYP), TKEYP
293 je .Lenc192
294 add $0x20, TKEYP
295 movaps -0x60(TKEYP), KEY
296 aesenc KEY, STATE
297 movaps -0x50(TKEYP), KEY
298 aesenc KEY, STATE
299 .align 4
300 .Lenc192:
301 movaps -0x40(TKEYP), KEY
302 aesenc KEY, STATE
303 movaps -0x30(TKEYP), KEY
304 aesenc KEY, STATE
305 .align 4
306 .Lenc128:
307 movaps -0x20(TKEYP), KEY
308 aesenc KEY, STATE
309 movaps -0x10(TKEYP), KEY
310 aesenc KEY, STATE
311 movaps (TKEYP), KEY
312 aesenc KEY, STATE
313 movaps 0x10(TKEYP), KEY
314 aesenc KEY, STATE
315 movaps 0x20(TKEYP), KEY
316 aesenc KEY, STATE
317 movaps 0x30(TKEYP), KEY
318 aesenc KEY, STATE
319 movaps 0x40(TKEYP), KEY
320 aesenc KEY, STATE
321 movaps 0x50(TKEYP), KEY
322 aesenc KEY, STATE
323 movaps 0x60(TKEYP), KEY
324 aesenc KEY, STATE
325 movaps 0x70(TKEYP), KEY
326 aesenclast KEY, STATE
327 RET
328 SYM_FUNC_END(_aesni_enc1)
329
330 /*
331 * _aesni_enc4: internal ABI
332 * input:
333 * KEYP: key struct pointer
334 * KLEN: round count
335 * STATE1: initial state (input)
336 * STATE2
337 * STATE3
338 * STATE4
339 * output:
340 * STATE1: finial state (output)
341 * STATE2
342 * STATE3
343 * STATE4
344 * changed:
345 * KEY
346 * TKEYP (T1)
347 */
348 SYM_FUNC_START_LOCAL(_aesni_enc4)
349 movaps (KEYP), KEY # key
350 mov KEYP, TKEYP
351 pxor KEY, STATE1 # roun
352 pxor KEY, STATE2
353 pxor KEY, STATE3
354 pxor KEY, STATE4
355 add $0x30, TKEYP
356 cmp $24, KLEN
357 jb .L4enc128
358 lea 0x20(TKEYP), TKEYP
359 je .L4enc192
360 add $0x20, TKEYP
361 movaps -0x60(TKEYP), KEY
362 aesenc KEY, STATE1
363 aesenc KEY, STATE2
364 aesenc KEY, STATE3
365 aesenc KEY, STATE4
366 movaps -0x50(TKEYP), KEY
367 aesenc KEY, STATE1
368 aesenc KEY, STATE2
369 aesenc KEY, STATE3
370 aesenc KEY, STATE4
371 #.align 4
372 .L4enc192:
373 movaps -0x40(TKEYP), KEY
374 aesenc KEY, STATE1
375 aesenc KEY, STATE2
376 aesenc KEY, STATE3
377 aesenc KEY, STATE4
378 movaps -0x30(TKEYP), KEY
379 aesenc KEY, STATE1
380 aesenc KEY, STATE2
381 aesenc KEY, STATE3
382 aesenc KEY, STATE4
383 #.align 4
384 .L4enc128:
385 movaps -0x20(TKEYP), KEY
386 aesenc KEY, STATE1
387 aesenc KEY, STATE2
388 aesenc KEY, STATE3
389 aesenc KEY, STATE4
390 movaps -0x10(TKEYP), KEY
391 aesenc KEY, STATE1
392 aesenc KEY, STATE2
393 aesenc KEY, STATE3
394 aesenc KEY, STATE4
395 movaps (TKEYP), KEY
396 aesenc KEY, STATE1
397 aesenc KEY, STATE2
398 aesenc KEY, STATE3
399 aesenc KEY, STATE4
400 movaps 0x10(TKEYP), KEY
401 aesenc KEY, STATE1
402 aesenc KEY, STATE2
403 aesenc KEY, STATE3
404 aesenc KEY, STATE4
405 movaps 0x20(TKEYP), KEY
406 aesenc KEY, STATE1
407 aesenc KEY, STATE2
408 aesenc KEY, STATE3
409 aesenc KEY, STATE4
410 movaps 0x30(TKEYP), KEY
411 aesenc KEY, STATE1
412 aesenc KEY, STATE2
413 aesenc KEY, STATE3
414 aesenc KEY, STATE4
415 movaps 0x40(TKEYP), KEY
416 aesenc KEY, STATE1
417 aesenc KEY, STATE2
418 aesenc KEY, STATE3
419 aesenc KEY, STATE4
420 movaps 0x50(TKEYP), KEY
421 aesenc KEY, STATE1
422 aesenc KEY, STATE2
423 aesenc KEY, STATE3
424 aesenc KEY, STATE4
425 movaps 0x60(TKEYP), KEY
426 aesenc KEY, STATE1
427 aesenc KEY, STATE2
428 aesenc KEY, STATE3
429 aesenc KEY, STATE4
430 movaps 0x70(TKEYP), KEY
431 aesenclast KEY, STATE1 # last
432 aesenclast KEY, STATE2
433 aesenclast KEY, STATE3
434 aesenclast KEY, STATE4
435 RET
436 SYM_FUNC_END(_aesni_enc4)
437
438 /*
439 * void aesni_dec (const void *ctx, u8 *dst, c
440 */
441 SYM_FUNC_START(aesni_dec)
442 FRAME_BEGIN
443 #ifndef __x86_64__
444 pushl KEYP
445 pushl KLEN
446 movl (FRAME_OFFSET+12)(%esp), KEYP
447 movl (FRAME_OFFSET+16)(%esp), OUTP
448 movl (FRAME_OFFSET+20)(%esp), INP
449 #endif
450 mov 480(KEYP), KLEN # key
451 add $240, KEYP
452 movups (INP), STATE # inpu
453 call _aesni_dec1
454 movups STATE, (OUTP) #outpu
455 #ifndef __x86_64__
456 popl KLEN
457 popl KEYP
458 #endif
459 FRAME_END
460 RET
461 SYM_FUNC_END(aesni_dec)
462
463 /*
464 * _aesni_dec1: internal ABI
465 * input:
466 * KEYP: key struct pointer
467 * KLEN: key length
468 * STATE: initial state (input)
469 * output:
470 * STATE: finial state (output)
471 * changed:
472 * KEY
473 * TKEYP (T1)
474 */
475 SYM_FUNC_START_LOCAL(_aesni_dec1)
476 movaps (KEYP), KEY # key
477 mov KEYP, TKEYP
478 pxor KEY, STATE # round 0
479 add $0x30, TKEYP
480 cmp $24, KLEN
481 jb .Ldec128
482 lea 0x20(TKEYP), TKEYP
483 je .Ldec192
484 add $0x20, TKEYP
485 movaps -0x60(TKEYP), KEY
486 aesdec KEY, STATE
487 movaps -0x50(TKEYP), KEY
488 aesdec KEY, STATE
489 .align 4
490 .Ldec192:
491 movaps -0x40(TKEYP), KEY
492 aesdec KEY, STATE
493 movaps -0x30(TKEYP), KEY
494 aesdec KEY, STATE
495 .align 4
496 .Ldec128:
497 movaps -0x20(TKEYP), KEY
498 aesdec KEY, STATE
499 movaps -0x10(TKEYP), KEY
500 aesdec KEY, STATE
501 movaps (TKEYP), KEY
502 aesdec KEY, STATE
503 movaps 0x10(TKEYP), KEY
504 aesdec KEY, STATE
505 movaps 0x20(TKEYP), KEY
506 aesdec KEY, STATE
507 movaps 0x30(TKEYP), KEY
508 aesdec KEY, STATE
509 movaps 0x40(TKEYP), KEY
510 aesdec KEY, STATE
511 movaps 0x50(TKEYP), KEY
512 aesdec KEY, STATE
513 movaps 0x60(TKEYP), KEY
514 aesdec KEY, STATE
515 movaps 0x70(TKEYP), KEY
516 aesdeclast KEY, STATE
517 RET
518 SYM_FUNC_END(_aesni_dec1)
519
520 /*
521 * _aesni_dec4: internal ABI
522 * input:
523 * KEYP: key struct pointer
524 * KLEN: key length
525 * STATE1: initial state (input)
526 * STATE2
527 * STATE3
528 * STATE4
529 * output:
530 * STATE1: finial state (output)
531 * STATE2
532 * STATE3
533 * STATE4
534 * changed:
535 * KEY
536 * TKEYP (T1)
537 */
538 SYM_FUNC_START_LOCAL(_aesni_dec4)
539 movaps (KEYP), KEY # key
540 mov KEYP, TKEYP
541 pxor KEY, STATE1 # roun
542 pxor KEY, STATE2
543 pxor KEY, STATE3
544 pxor KEY, STATE4
545 add $0x30, TKEYP
546 cmp $24, KLEN
547 jb .L4dec128
548 lea 0x20(TKEYP), TKEYP
549 je .L4dec192
550 add $0x20, TKEYP
551 movaps -0x60(TKEYP), KEY
552 aesdec KEY, STATE1
553 aesdec KEY, STATE2
554 aesdec KEY, STATE3
555 aesdec KEY, STATE4
556 movaps -0x50(TKEYP), KEY
557 aesdec KEY, STATE1
558 aesdec KEY, STATE2
559 aesdec KEY, STATE3
560 aesdec KEY, STATE4
561 .align 4
562 .L4dec192:
563 movaps -0x40(TKEYP), KEY
564 aesdec KEY, STATE1
565 aesdec KEY, STATE2
566 aesdec KEY, STATE3
567 aesdec KEY, STATE4
568 movaps -0x30(TKEYP), KEY
569 aesdec KEY, STATE1
570 aesdec KEY, STATE2
571 aesdec KEY, STATE3
572 aesdec KEY, STATE4
573 .align 4
574 .L4dec128:
575 movaps -0x20(TKEYP), KEY
576 aesdec KEY, STATE1
577 aesdec KEY, STATE2
578 aesdec KEY, STATE3
579 aesdec KEY, STATE4
580 movaps -0x10(TKEYP), KEY
581 aesdec KEY, STATE1
582 aesdec KEY, STATE2
583 aesdec KEY, STATE3
584 aesdec KEY, STATE4
585 movaps (TKEYP), KEY
586 aesdec KEY, STATE1
587 aesdec KEY, STATE2
588 aesdec KEY, STATE3
589 aesdec KEY, STATE4
590 movaps 0x10(TKEYP), KEY
591 aesdec KEY, STATE1
592 aesdec KEY, STATE2
593 aesdec KEY, STATE3
594 aesdec KEY, STATE4
595 movaps 0x20(TKEYP), KEY
596 aesdec KEY, STATE1
597 aesdec KEY, STATE2
598 aesdec KEY, STATE3
599 aesdec KEY, STATE4
600 movaps 0x30(TKEYP), KEY
601 aesdec KEY, STATE1
602 aesdec KEY, STATE2
603 aesdec KEY, STATE3
604 aesdec KEY, STATE4
605 movaps 0x40(TKEYP), KEY
606 aesdec KEY, STATE1
607 aesdec KEY, STATE2
608 aesdec KEY, STATE3
609 aesdec KEY, STATE4
610 movaps 0x50(TKEYP), KEY
611 aesdec KEY, STATE1
612 aesdec KEY, STATE2
613 aesdec KEY, STATE3
614 aesdec KEY, STATE4
615 movaps 0x60(TKEYP), KEY
616 aesdec KEY, STATE1
617 aesdec KEY, STATE2
618 aesdec KEY, STATE3
619 aesdec KEY, STATE4
620 movaps 0x70(TKEYP), KEY
621 aesdeclast KEY, STATE1 # last
622 aesdeclast KEY, STATE2
623 aesdeclast KEY, STATE3
624 aesdeclast KEY, STATE4
625 RET
626 SYM_FUNC_END(_aesni_dec4)
627
628 /*
629 * void aesni_ecb_enc(struct crypto_aes_ctx *c
630 * size_t len)
631 */
632 SYM_FUNC_START(aesni_ecb_enc)
633 FRAME_BEGIN
634 #ifndef __x86_64__
635 pushl LEN
636 pushl KEYP
637 pushl KLEN
638 movl (FRAME_OFFSET+16)(%esp), KEYP
639 movl (FRAME_OFFSET+20)(%esp), OUTP
640 movl (FRAME_OFFSET+24)(%esp), INP
641 movl (FRAME_OFFSET+28)(%esp), LEN
642 #endif
643 test LEN, LEN # check length
644 jz .Lecb_enc_ret
645 mov 480(KEYP), KLEN
646 cmp $16, LEN
647 jb .Lecb_enc_ret
648 cmp $64, LEN
649 jb .Lecb_enc_loop1
650 .align 4
651 .Lecb_enc_loop4:
652 movups (INP), STATE1
653 movups 0x10(INP), STATE2
654 movups 0x20(INP), STATE3
655 movups 0x30(INP), STATE4
656 call _aesni_enc4
657 movups STATE1, (OUTP)
658 movups STATE2, 0x10(OUTP)
659 movups STATE3, 0x20(OUTP)
660 movups STATE4, 0x30(OUTP)
661 sub $64, LEN
662 add $64, INP
663 add $64, OUTP
664 cmp $64, LEN
665 jge .Lecb_enc_loop4
666 cmp $16, LEN
667 jb .Lecb_enc_ret
668 .align 4
669 .Lecb_enc_loop1:
670 movups (INP), STATE1
671 call _aesni_enc1
672 movups STATE1, (OUTP)
673 sub $16, LEN
674 add $16, INP
675 add $16, OUTP
676 cmp $16, LEN
677 jge .Lecb_enc_loop1
678 .Lecb_enc_ret:
679 #ifndef __x86_64__
680 popl KLEN
681 popl KEYP
682 popl LEN
683 #endif
684 FRAME_END
685 RET
686 SYM_FUNC_END(aesni_ecb_enc)
687
688 /*
689 * void aesni_ecb_dec(struct crypto_aes_ctx *c
690 * size_t len);
691 */
692 SYM_FUNC_START(aesni_ecb_dec)
693 FRAME_BEGIN
694 #ifndef __x86_64__
695 pushl LEN
696 pushl KEYP
697 pushl KLEN
698 movl (FRAME_OFFSET+16)(%esp), KEYP
699 movl (FRAME_OFFSET+20)(%esp), OUTP
700 movl (FRAME_OFFSET+24)(%esp), INP
701 movl (FRAME_OFFSET+28)(%esp), LEN
702 #endif
703 test LEN, LEN
704 jz .Lecb_dec_ret
705 mov 480(KEYP), KLEN
706 add $240, KEYP
707 cmp $16, LEN
708 jb .Lecb_dec_ret
709 cmp $64, LEN
710 jb .Lecb_dec_loop1
711 .align 4
712 .Lecb_dec_loop4:
713 movups (INP), STATE1
714 movups 0x10(INP), STATE2
715 movups 0x20(INP), STATE3
716 movups 0x30(INP), STATE4
717 call _aesni_dec4
718 movups STATE1, (OUTP)
719 movups STATE2, 0x10(OUTP)
720 movups STATE3, 0x20(OUTP)
721 movups STATE4, 0x30(OUTP)
722 sub $64, LEN
723 add $64, INP
724 add $64, OUTP
725 cmp $64, LEN
726 jge .Lecb_dec_loop4
727 cmp $16, LEN
728 jb .Lecb_dec_ret
729 .align 4
730 .Lecb_dec_loop1:
731 movups (INP), STATE1
732 call _aesni_dec1
733 movups STATE1, (OUTP)
734 sub $16, LEN
735 add $16, INP
736 add $16, OUTP
737 cmp $16, LEN
738 jge .Lecb_dec_loop1
739 .Lecb_dec_ret:
740 #ifndef __x86_64__
741 popl KLEN
742 popl KEYP
743 popl LEN
744 #endif
745 FRAME_END
746 RET
747 SYM_FUNC_END(aesni_ecb_dec)
748
749 /*
750 * void aesni_cbc_enc(struct crypto_aes_ctx *c
751 * size_t len, u8 *iv)
752 */
753 SYM_FUNC_START(aesni_cbc_enc)
754 FRAME_BEGIN
755 #ifndef __x86_64__
756 pushl IVP
757 pushl LEN
758 pushl KEYP
759 pushl KLEN
760 movl (FRAME_OFFSET+20)(%esp), KEYP
761 movl (FRAME_OFFSET+24)(%esp), OUTP
762 movl (FRAME_OFFSET+28)(%esp), INP
763 movl (FRAME_OFFSET+32)(%esp), LEN
764 movl (FRAME_OFFSET+36)(%esp), IVP
765 #endif
766 cmp $16, LEN
767 jb .Lcbc_enc_ret
768 mov 480(KEYP), KLEN
769 movups (IVP), STATE # load iv as i
770 .align 4
771 .Lcbc_enc_loop:
772 movups (INP), IN # load input
773 pxor IN, STATE
774 call _aesni_enc1
775 movups STATE, (OUTP) # store output
776 sub $16, LEN
777 add $16, INP
778 add $16, OUTP
779 cmp $16, LEN
780 jge .Lcbc_enc_loop
781 movups STATE, (IVP)
782 .Lcbc_enc_ret:
783 #ifndef __x86_64__
784 popl KLEN
785 popl KEYP
786 popl LEN
787 popl IVP
788 #endif
789 FRAME_END
790 RET
791 SYM_FUNC_END(aesni_cbc_enc)
792
793 /*
794 * void aesni_cbc_dec(struct crypto_aes_ctx *c
795 * size_t len, u8 *iv)
796 */
797 SYM_FUNC_START(aesni_cbc_dec)
798 FRAME_BEGIN
799 #ifndef __x86_64__
800 pushl IVP
801 pushl LEN
802 pushl KEYP
803 pushl KLEN
804 movl (FRAME_OFFSET+20)(%esp), KEYP
805 movl (FRAME_OFFSET+24)(%esp), OUTP
806 movl (FRAME_OFFSET+28)(%esp), INP
807 movl (FRAME_OFFSET+32)(%esp), LEN
808 movl (FRAME_OFFSET+36)(%esp), IVP
809 #endif
810 cmp $16, LEN
811 jb .Lcbc_dec_just_ret
812 mov 480(KEYP), KLEN
813 add $240, KEYP
814 movups (IVP), IV
815 cmp $64, LEN
816 jb .Lcbc_dec_loop1
817 .align 4
818 .Lcbc_dec_loop4:
819 movups (INP), IN1
820 movaps IN1, STATE1
821 movups 0x10(INP), IN2
822 movaps IN2, STATE2
823 #ifdef __x86_64__
824 movups 0x20(INP), IN3
825 movaps IN3, STATE3
826 movups 0x30(INP), IN4
827 movaps IN4, STATE4
828 #else
829 movups 0x20(INP), IN1
830 movaps IN1, STATE3
831 movups 0x30(INP), IN2
832 movaps IN2, STATE4
833 #endif
834 call _aesni_dec4
835 pxor IV, STATE1
836 #ifdef __x86_64__
837 pxor IN1, STATE2
838 pxor IN2, STATE3
839 pxor IN3, STATE4
840 movaps IN4, IV
841 #else
842 pxor IN1, STATE4
843 movaps IN2, IV
844 movups (INP), IN1
845 pxor IN1, STATE2
846 movups 0x10(INP), IN2
847 pxor IN2, STATE3
848 #endif
849 movups STATE1, (OUTP)
850 movups STATE2, 0x10(OUTP)
851 movups STATE3, 0x20(OUTP)
852 movups STATE4, 0x30(OUTP)
853 sub $64, LEN
854 add $64, INP
855 add $64, OUTP
856 cmp $64, LEN
857 jge .Lcbc_dec_loop4
858 cmp $16, LEN
859 jb .Lcbc_dec_ret
860 .align 4
861 .Lcbc_dec_loop1:
862 movups (INP), IN
863 movaps IN, STATE
864 call _aesni_dec1
865 pxor IV, STATE
866 movups STATE, (OUTP)
867 movaps IN, IV
868 sub $16, LEN
869 add $16, INP
870 add $16, OUTP
871 cmp $16, LEN
872 jge .Lcbc_dec_loop1
873 .Lcbc_dec_ret:
874 movups IV, (IVP)
875 .Lcbc_dec_just_ret:
876 #ifndef __x86_64__
877 popl KLEN
878 popl KEYP
879 popl LEN
880 popl IVP
881 #endif
882 FRAME_END
883 RET
884 SYM_FUNC_END(aesni_cbc_dec)
885
886 /*
887 * void aesni_cts_cbc_enc(struct crypto_aes_ct
888 * size_t len, u8 *iv)
889 */
890 SYM_FUNC_START(aesni_cts_cbc_enc)
891 FRAME_BEGIN
892 #ifndef __x86_64__
893 pushl IVP
894 pushl LEN
895 pushl KEYP
896 pushl KLEN
897 movl (FRAME_OFFSET+20)(%esp), KEYP
898 movl (FRAME_OFFSET+24)(%esp), OUTP
899 movl (FRAME_OFFSET+28)(%esp), INP
900 movl (FRAME_OFFSET+32)(%esp), LEN
901 movl (FRAME_OFFSET+36)(%esp), IVP
902 lea .Lcts_permute_table, T1
903 #else
904 lea .Lcts_permute_table(%rip), T1
905 #endif
906 mov 480(KEYP), KLEN
907 movups (IVP), STATE
908 sub $16, LEN
909 mov T1, IVP
910 add $32, IVP
911 add LEN, T1
912 sub LEN, IVP
913 movups (T1), %xmm4
914 movups (IVP), %xmm5
915
916 movups (INP), IN1
917 add LEN, INP
918 movups (INP), IN2
919
920 pxor IN1, STATE
921 call _aesni_enc1
922
923 pshufb %xmm5, IN2
924 pxor STATE, IN2
925 pshufb %xmm4, STATE
926 add OUTP, LEN
927 movups STATE, (LEN)
928
929 movaps IN2, STATE
930 call _aesni_enc1
931 movups STATE, (OUTP)
932
933 #ifndef __x86_64__
934 popl KLEN
935 popl KEYP
936 popl LEN
937 popl IVP
938 #endif
939 FRAME_END
940 RET
941 SYM_FUNC_END(aesni_cts_cbc_enc)
942
943 /*
944 * void aesni_cts_cbc_dec(struct crypto_aes_ct
945 * size_t len, u8 *iv)
946 */
947 SYM_FUNC_START(aesni_cts_cbc_dec)
948 FRAME_BEGIN
949 #ifndef __x86_64__
950 pushl IVP
951 pushl LEN
952 pushl KEYP
953 pushl KLEN
954 movl (FRAME_OFFSET+20)(%esp), KEYP
955 movl (FRAME_OFFSET+24)(%esp), OUTP
956 movl (FRAME_OFFSET+28)(%esp), INP
957 movl (FRAME_OFFSET+32)(%esp), LEN
958 movl (FRAME_OFFSET+36)(%esp), IVP
959 lea .Lcts_permute_table, T1
960 #else
961 lea .Lcts_permute_table(%rip), T1
962 #endif
963 mov 480(KEYP), KLEN
964 add $240, KEYP
965 movups (IVP), IV
966 sub $16, LEN
967 mov T1, IVP
968 add $32, IVP
969 add LEN, T1
970 sub LEN, IVP
971 movups (T1), %xmm4
972
973 movups (INP), STATE
974 add LEN, INP
975 movups (INP), IN1
976
977 call _aesni_dec1
978 movaps STATE, IN2
979 pshufb %xmm4, STATE
980 pxor IN1, STATE
981
982 add OUTP, LEN
983 movups STATE, (LEN)
984
985 movups (IVP), %xmm0
986 pshufb %xmm0, IN1
987 pblendvb IN2, IN1
988 movaps IN1, STATE
989 call _aesni_dec1
990
991 pxor IV, STATE
992 movups STATE, (OUTP)
993
994 #ifndef __x86_64__
995 popl KLEN
996 popl KEYP
997 popl LEN
998 popl IVP
999 #endif
1000 FRAME_END
1001 RET
1002 SYM_FUNC_END(aesni_cts_cbc_dec)
1003
1004 .pushsection .rodata
1005 .align 16
1006 .Lcts_permute_table:
1007 .byte 0x80, 0x80, 0x80, 0x8
1008 .byte 0x80, 0x80, 0x80, 0x8
1009 .byte 0x00, 0x01, 0x02, 0x0
1010 .byte 0x08, 0x09, 0x0a, 0x0
1011 .byte 0x80, 0x80, 0x80, 0x8
1012 .byte 0x80, 0x80, 0x80, 0x8
1013 #ifdef __x86_64__
1014 .Lbswap_mask:
1015 .byte 15, 14, 13, 12, 11, 10, 9, 8, 7
1016 #endif
1017 .popsection
1018
1019 #ifdef __x86_64__
1020 /*
1021 * _aesni_inc_init: internal ABI
1022 * setup registers used by _aesni_inc
1023 * input:
1024 * IV
1025 * output:
1026 * CTR: == IV, in little endian
1027 * TCTR_LOW: == lower qword of CTR
1028 * INC: == 1, in little endian
1029 * BSWAP_MASK == endian swapping mask
1030 */
1031 SYM_FUNC_START_LOCAL(_aesni_inc_init)
1032 movaps .Lbswap_mask(%rip), BSWAP_MASK
1033 movaps IV, CTR
1034 pshufb BSWAP_MASK, CTR
1035 mov $1, TCTR_LOW
1036 movq TCTR_LOW, INC
1037 movq CTR, TCTR_LOW
1038 RET
1039 SYM_FUNC_END(_aesni_inc_init)
1040
1041 /*
1042 * _aesni_inc: internal ABI
1043 * Increase IV by 1, IV is in big endian
1044 * input:
1045 * IV
1046 * CTR: == IV, in little endian
1047 * TCTR_LOW: == lower qword of CTR
1048 * INC: == 1, in little endian
1049 * BSWAP_MASK == endian swapping mask
1050 * output:
1051 * IV: Increase by 1
1052 * changed:
1053 * CTR: == output IV, in little endia
1054 * TCTR_LOW: == lower qword of CTR
1055 */
1056 SYM_FUNC_START_LOCAL(_aesni_inc)
1057 paddq INC, CTR
1058 add $1, TCTR_LOW
1059 jnc .Linc_low
1060 pslldq $8, INC
1061 paddq INC, CTR
1062 psrldq $8, INC
1063 .Linc_low:
1064 movaps CTR, IV
1065 pshufb BSWAP_MASK, IV
1066 RET
1067 SYM_FUNC_END(_aesni_inc)
1068
1069 /*
1070 * void aesni_ctr_enc(struct crypto_aes_ctx *
1071 * size_t len, u8 *iv)
1072 */
1073 SYM_FUNC_START(aesni_ctr_enc)
1074 FRAME_BEGIN
1075 cmp $16, LEN
1076 jb .Lctr_enc_just_ret
1077 mov 480(KEYP), KLEN
1078 movups (IVP), IV
1079 call _aesni_inc_init
1080 cmp $64, LEN
1081 jb .Lctr_enc_loop1
1082 .align 4
1083 .Lctr_enc_loop4:
1084 movaps IV, STATE1
1085 call _aesni_inc
1086 movups (INP), IN1
1087 movaps IV, STATE2
1088 call _aesni_inc
1089 movups 0x10(INP), IN2
1090 movaps IV, STATE3
1091 call _aesni_inc
1092 movups 0x20(INP), IN3
1093 movaps IV, STATE4
1094 call _aesni_inc
1095 movups 0x30(INP), IN4
1096 call _aesni_enc4
1097 pxor IN1, STATE1
1098 movups STATE1, (OUTP)
1099 pxor IN2, STATE2
1100 movups STATE2, 0x10(OUTP)
1101 pxor IN3, STATE3
1102 movups STATE3, 0x20(OUTP)
1103 pxor IN4, STATE4
1104 movups STATE4, 0x30(OUTP)
1105 sub $64, LEN
1106 add $64, INP
1107 add $64, OUTP
1108 cmp $64, LEN
1109 jge .Lctr_enc_loop4
1110 cmp $16, LEN
1111 jb .Lctr_enc_ret
1112 .align 4
1113 .Lctr_enc_loop1:
1114 movaps IV, STATE
1115 call _aesni_inc
1116 movups (INP), IN
1117 call _aesni_enc1
1118 pxor IN, STATE
1119 movups STATE, (OUTP)
1120 sub $16, LEN
1121 add $16, INP
1122 add $16, OUTP
1123 cmp $16, LEN
1124 jge .Lctr_enc_loop1
1125 .Lctr_enc_ret:
1126 movups IV, (IVP)
1127 .Lctr_enc_just_ret:
1128 FRAME_END
1129 RET
1130 SYM_FUNC_END(aesni_ctr_enc)
1131
1132 #endif
1133
1134 .section .rodata.cst16.gf128mul_x_ble_
1135 .align 16
1136 .Lgf128mul_x_ble_mask:
1137 .octa 0x00000000000000010000000000000
1138 .previous
1139
1140 /*
1141 * _aesni_gf128mul_x_ble: Multiply in GF(2^12
1142 * input:
1143 * IV: current IV
1144 * GF128MUL_MASK == mask with 0x87 and 0
1145 * output:
1146 * IV: next IV
1147 * changed:
1148 * KEY: == temporary value
1149 */
1150 .macro _aesni_gf128mul_x_ble
1151 pshufd $0x13, IV, KEY
1152 paddq IV, IV
1153 psrad $31, KEY
1154 pand GF128MUL_MASK, KEY
1155 pxor KEY, IV
1156 .endm
1157
1158 .macro _aesni_xts_crypt enc
1159 FRAME_BEGIN
1160 #ifndef __x86_64__
1161 pushl IVP
1162 pushl LEN
1163 pushl KEYP
1164 pushl KLEN
1165 movl (FRAME_OFFSET+20)(%esp), KEYP
1166 movl (FRAME_OFFSET+24)(%esp), OUTP
1167 movl (FRAME_OFFSET+28)(%esp), INP
1168 movl (FRAME_OFFSET+32)(%esp), LEN
1169 movl (FRAME_OFFSET+36)(%esp), IVP
1170 movdqa .Lgf128mul_x_ble_mask, GF128MU
1171 #else
1172 movdqa .Lgf128mul_x_ble_mask(%rip), G
1173 #endif
1174 movups (IVP), IV
1175
1176 mov 480(KEYP), KLEN
1177 .if !\enc
1178 add $240, KEYP
1179
1180 test $15, LEN
1181 jz .Lxts_loop4\@
1182 sub $16, LEN
1183 .endif
1184
1185 .Lxts_loop4\@:
1186 sub $64, LEN
1187 jl .Lxts_1x\@
1188
1189 movdqa IV, STATE1
1190 movdqu 0x00(INP), IN
1191 pxor IN, STATE1
1192 movdqu IV, 0x00(OUTP)
1193
1194 _aesni_gf128mul_x_ble
1195 movdqa IV, STATE2
1196 movdqu 0x10(INP), IN
1197 pxor IN, STATE2
1198 movdqu IV, 0x10(OUTP)
1199
1200 _aesni_gf128mul_x_ble
1201 movdqa IV, STATE3
1202 movdqu 0x20(INP), IN
1203 pxor IN, STATE3
1204 movdqu IV, 0x20(OUTP)
1205
1206 _aesni_gf128mul_x_ble
1207 movdqa IV, STATE4
1208 movdqu 0x30(INP), IN
1209 pxor IN, STATE4
1210 movdqu IV, 0x30(OUTP)
1211
1212 .if \enc
1213 call _aesni_enc4
1214 .else
1215 call _aesni_dec4
1216 .endif
1217
1218 movdqu 0x00(OUTP), IN
1219 pxor IN, STATE1
1220 movdqu STATE1, 0x00(OUTP)
1221
1222 movdqu 0x10(OUTP), IN
1223 pxor IN, STATE2
1224 movdqu STATE2, 0x10(OUTP)
1225
1226 movdqu 0x20(OUTP), IN
1227 pxor IN, STATE3
1228 movdqu STATE3, 0x20(OUTP)
1229
1230 movdqu 0x30(OUTP), IN
1231 pxor IN, STATE4
1232 movdqu STATE4, 0x30(OUTP)
1233
1234 _aesni_gf128mul_x_ble
1235
1236 add $64, INP
1237 add $64, OUTP
1238 test LEN, LEN
1239 jnz .Lxts_loop4\@
1240
1241 .Lxts_ret_iv\@:
1242 movups IV, (IVP)
1243
1244 .Lxts_ret\@:
1245 #ifndef __x86_64__
1246 popl KLEN
1247 popl KEYP
1248 popl LEN
1249 popl IVP
1250 #endif
1251 FRAME_END
1252 RET
1253
1254 .Lxts_1x\@:
1255 add $64, LEN
1256 jz .Lxts_ret_iv\@
1257 .if \enc
1258 sub $16, LEN
1259 jl .Lxts_cts4\@
1260 .endif
1261
1262 .Lxts_loop1\@:
1263 movdqu (INP), STATE
1264 .if \enc
1265 pxor IV, STATE
1266 call _aesni_enc1
1267 .else
1268 add $16, INP
1269 sub $16, LEN
1270 jl .Lxts_cts1\@
1271 pxor IV, STATE
1272 call _aesni_dec1
1273 .endif
1274 pxor IV, STATE
1275 _aesni_gf128mul_x_ble
1276
1277 test LEN, LEN
1278 jz .Lxts_out\@
1279
1280 .if \enc
1281 add $16, INP
1282 sub $16, LEN
1283 jl .Lxts_cts1\@
1284 .endif
1285
1286 movdqu STATE, (OUTP)
1287 add $16, OUTP
1288 jmp .Lxts_loop1\@
1289
1290 .Lxts_out\@:
1291 movdqu STATE, (OUTP)
1292 jmp .Lxts_ret_iv\@
1293
1294 .if \enc
1295 .Lxts_cts4\@:
1296 movdqa STATE4, STATE
1297 sub $16, OUTP
1298 .Lxts_cts1\@:
1299 .else
1300 .Lxts_cts1\@:
1301 movdqa IV, STATE4
1302 _aesni_gf128mul_x_ble
1303
1304 pxor IV, STATE
1305 call _aesni_dec1
1306 pxor IV, STATE
1307 .endif
1308 #ifndef __x86_64__
1309 lea .Lcts_permute_table, T1
1310 #else
1311 lea .Lcts_permute_table(%rip), T1
1312 #endif
1313 add LEN, INP /* rewind inp
1314 add $16, LEN /* # bytes in
1315 movups (INP), IN1
1316
1317 mov T1, IVP
1318 add $32, IVP
1319 add LEN, T1
1320 sub LEN, IVP
1321 add OUTP, LEN
1322
1323 movups (T1), %xmm4
1324 movaps STATE, IN2
1325 pshufb %xmm4, STATE
1326 movups STATE, (LEN)
1327
1328 movups (IVP), %xmm0
1329 pshufb %xmm0, IN1
1330 pblendvb IN2, IN1
1331 movaps IN1, STATE
1332
1333 .if \enc
1334 pxor IV, STATE
1335 call _aesni_enc1
1336 pxor IV, STATE
1337 .else
1338 pxor STATE4, STATE
1339 call _aesni_dec1
1340 pxor STATE4, STATE
1341 .endif
1342
1343 movups STATE, (OUTP)
1344 jmp .Lxts_ret\@
1345 .endm
1346
1347 /*
1348 * void aesni_xts_enc(const struct crypto_aes
1349 * const u8 *src, unsigned
1350 */
1351 SYM_FUNC_START(aesni_xts_enc)
1352 _aesni_xts_crypt 1
1353 SYM_FUNC_END(aesni_xts_enc)
1354
1355 /*
1356 * void aesni_xts_dec(const struct crypto_aes
1357 * const u8 *src, unsigned
1358 */
1359 SYM_FUNC_START(aesni_xts_dec)
1360 _aesni_xts_crypt 0
1361 SYM_FUNC_END(aesni_xts_dec)
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.