Merge tag 'media/v4.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
[platform/kernel/linux-rpi.git] / crypto / vmac.c
1 /*
2  * Modified to interface to the Linux kernel
3  * Copyright (c) 2009, Intel Corporation.
4  *
5  * This program is free software; you can redistribute it and/or modify it
6  * under the terms and conditions of the GNU General Public License,
7  * version 2, as published by the Free Software Foundation.
8  *
9  * This program is distributed in the hope it will be useful, but WITHOUT
10  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
12  * more details.
13  *
14  * You should have received a copy of the GNU General Public License along with
15  * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
16  * Place - Suite 330, Boston, MA 02111-1307 USA.
17  */
18
19 /* --------------------------------------------------------------------------
20  * VMAC and VHASH Implementation by Ted Krovetz (tdk@acm.org) and Wei Dai.
21  * This implementation is herby placed in the public domain.
22  * The authors offers no warranty. Use at your own risk.
23  * Please send bug reports to the authors.
24  * Last modified: 17 APR 08, 1700 PDT
25  * ----------------------------------------------------------------------- */
26
27 #include <linux/init.h>
28 #include <linux/types.h>
29 #include <linux/crypto.h>
30 #include <linux/module.h>
31 #include <linux/scatterlist.h>
32 #include <asm/byteorder.h>
33 #include <crypto/scatterwalk.h>
34 #include <crypto/vmac.h>
35 #include <crypto/internal/hash.h>
36
37 /*
38  * Constants and masks
39  */
40 #define UINT64_C(x) x##ULL
41 static const u64 p64   = UINT64_C(0xfffffffffffffeff);  /* 2^64 - 257 prime  */
42 static const u64 m62   = UINT64_C(0x3fffffffffffffff);  /* 62-bit mask       */
43 static const u64 m63   = UINT64_C(0x7fffffffffffffff);  /* 63-bit mask       */
44 static const u64 m64   = UINT64_C(0xffffffffffffffff);  /* 64-bit mask       */
45 static const u64 mpoly = UINT64_C(0x1fffffff1fffffff);  /* Poly key mask     */
46
47 #define pe64_to_cpup le64_to_cpup               /* Prefer little endian */
48
49 #ifdef __LITTLE_ENDIAN
50 #define INDEX_HIGH 1
51 #define INDEX_LOW 0
52 #else
53 #define INDEX_HIGH 0
54 #define INDEX_LOW 1
55 #endif
56
57 /*
58  * The following routines are used in this implementation. They are
59  * written via macros to simulate zero-overhead call-by-reference.
60  *
61  * MUL64: 64x64->128-bit multiplication
62  * PMUL64: assumes top bits cleared on inputs
63  * ADD128: 128x128->128-bit addition
64  */
65
66 #define ADD128(rh, rl, ih, il)                                          \
67         do {                                                            \
68                 u64 _il = (il);                                         \
69                 (rl) += (_il);                                          \
70                 if ((rl) < (_il))                                       \
71                         (rh)++;                                         \
72                 (rh) += (ih);                                           \
73         } while (0)
74
75 #define MUL32(i1, i2)   ((u64)(u32)(i1)*(u32)(i2))
76
77 #define PMUL64(rh, rl, i1, i2)  /* Assumes m doesn't overflow */        \
78         do {                                                            \
79                 u64 _i1 = (i1), _i2 = (i2);                             \
80                 u64 m = MUL32(_i1, _i2>>32) + MUL32(_i1>>32, _i2);      \
81                 rh = MUL32(_i1>>32, _i2>>32);                           \
82                 rl = MUL32(_i1, _i2);                                   \
83                 ADD128(rh, rl, (m >> 32), (m << 32));                   \
84         } while (0)
85
86 #define MUL64(rh, rl, i1, i2)                                           \
87         do {                                                            \
88                 u64 _i1 = (i1), _i2 = (i2);                             \
89                 u64 m1 = MUL32(_i1, _i2>>32);                           \
90                 u64 m2 = MUL32(_i1>>32, _i2);                           \
91                 rh = MUL32(_i1>>32, _i2>>32);                           \
92                 rl = MUL32(_i1, _i2);                                   \
93                 ADD128(rh, rl, (m1 >> 32), (m1 << 32));                 \
94                 ADD128(rh, rl, (m2 >> 32), (m2 << 32));                 \
95         } while (0)
96
97 /*
98  * For highest performance the L1 NH and L2 polynomial hashes should be
99  * carefully implemented to take advantage of one's target architecture.
100  * Here these two hash functions are defined multiple time; once for
101  * 64-bit architectures, once for 32-bit SSE2 architectures, and once
102  * for the rest (32-bit) architectures.
103  * For each, nh_16 *must* be defined (works on multiples of 16 bytes).
104  * Optionally, nh_vmac_nhbytes can be defined (for multiples of
105  * VMAC_NHBYTES), and nh_16_2 and nh_vmac_nhbytes_2 (versions that do two
106  * NH computations at once).
107  */
108
109 #ifdef CONFIG_64BIT
110
111 #define nh_16(mp, kp, nw, rh, rl)                                       \
112         do {                                                            \
113                 int i; u64 th, tl;                                      \
114                 rh = rl = 0;                                            \
115                 for (i = 0; i < nw; i += 2) {                           \
116                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i],     \
117                                 pe64_to_cpup((mp)+i+1)+(kp)[i+1]);      \
118                         ADD128(rh, rl, th, tl);                         \
119                 }                                                       \
120         } while (0)
121
122 #define nh_16_2(mp, kp, nw, rh, rl, rh1, rl1)                           \
123         do {                                                            \
124                 int i; u64 th, tl;                                      \
125                 rh1 = rl1 = rh = rl = 0;                                \
126                 for (i = 0; i < nw; i += 2) {                           \
127                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i],     \
128                                 pe64_to_cpup((mp)+i+1)+(kp)[i+1]);      \
129                         ADD128(rh, rl, th, tl);                         \
130                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i+2],   \
131                                 pe64_to_cpup((mp)+i+1)+(kp)[i+3]);      \
132                         ADD128(rh1, rl1, th, tl);                       \
133                 }                                                       \
134         } while (0)
135
136 #if (VMAC_NHBYTES >= 64) /* These versions do 64-bytes of message at a time */
137 #define nh_vmac_nhbytes(mp, kp, nw, rh, rl)                             \
138         do {                                                            \
139                 int i; u64 th, tl;                                      \
140                 rh = rl = 0;                                            \
141                 for (i = 0; i < nw; i += 8) {                           \
142                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i],     \
143                                 pe64_to_cpup((mp)+i+1)+(kp)[i+1]);      \
144                         ADD128(rh, rl, th, tl);                         \
145                         MUL64(th, tl, pe64_to_cpup((mp)+i+2)+(kp)[i+2], \
146                                 pe64_to_cpup((mp)+i+3)+(kp)[i+3]);      \
147                         ADD128(rh, rl, th, tl);                         \
148                         MUL64(th, tl, pe64_to_cpup((mp)+i+4)+(kp)[i+4], \
149                                 pe64_to_cpup((mp)+i+5)+(kp)[i+5]);      \
150                         ADD128(rh, rl, th, tl);                         \
151                         MUL64(th, tl, pe64_to_cpup((mp)+i+6)+(kp)[i+6], \
152                                 pe64_to_cpup((mp)+i+7)+(kp)[i+7]);      \
153                         ADD128(rh, rl, th, tl);                         \
154                 }                                                       \
155         } while (0)
156
157 #define nh_vmac_nhbytes_2(mp, kp, nw, rh, rl, rh1, rl1)                 \
158         do {                                                            \
159                 int i; u64 th, tl;                                      \
160                 rh1 = rl1 = rh = rl = 0;                                \
161                 for (i = 0; i < nw; i += 8) {                           \
162                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i],     \
163                                 pe64_to_cpup((mp)+i+1)+(kp)[i+1]);      \
164                         ADD128(rh, rl, th, tl);                         \
165                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i+2],   \
166                                 pe64_to_cpup((mp)+i+1)+(kp)[i+3]);      \
167                         ADD128(rh1, rl1, th, tl);                       \
168                         MUL64(th, tl, pe64_to_cpup((mp)+i+2)+(kp)[i+2], \
169                                 pe64_to_cpup((mp)+i+3)+(kp)[i+3]);      \
170                         ADD128(rh, rl, th, tl);                         \
171                         MUL64(th, tl, pe64_to_cpup((mp)+i+2)+(kp)[i+4], \
172                                 pe64_to_cpup((mp)+i+3)+(kp)[i+5]);      \
173                         ADD128(rh1, rl1, th, tl);                       \
174                         MUL64(th, tl, pe64_to_cpup((mp)+i+4)+(kp)[i+4], \
175                                 pe64_to_cpup((mp)+i+5)+(kp)[i+5]);      \
176                         ADD128(rh, rl, th, tl);                         \
177                         MUL64(th, tl, pe64_to_cpup((mp)+i+4)+(kp)[i+6], \
178                                 pe64_to_cpup((mp)+i+5)+(kp)[i+7]);      \
179                         ADD128(rh1, rl1, th, tl);                       \
180                         MUL64(th, tl, pe64_to_cpup((mp)+i+6)+(kp)[i+6], \
181                                 pe64_to_cpup((mp)+i+7)+(kp)[i+7]);      \
182                         ADD128(rh, rl, th, tl);                         \
183                         MUL64(th, tl, pe64_to_cpup((mp)+i+6)+(kp)[i+8], \
184                                 pe64_to_cpup((mp)+i+7)+(kp)[i+9]);      \
185                         ADD128(rh1, rl1, th, tl);                       \
186                 }                                                       \
187         } while (0)
188 #endif
189
190 #define poly_step(ah, al, kh, kl, mh, ml)                               \
191         do {                                                            \
192                 u64 t1h, t1l, t2h, t2l, t3h, t3l, z = 0;                \
193                 /* compute ab*cd, put bd into result registers */       \
194                 PMUL64(t3h, t3l, al, kh);                               \
195                 PMUL64(t2h, t2l, ah, kl);                               \
196                 PMUL64(t1h, t1l, ah, 2*kh);                             \
197                 PMUL64(ah, al, al, kl);                                 \
198                 /* add 2 * ac to result */                              \
199                 ADD128(ah, al, t1h, t1l);                               \
200                 /* add together ad + bc */                              \
201                 ADD128(t2h, t2l, t3h, t3l);                             \
202                 /* now (ah,al), (t2l,2*t2h) need summing */             \
203                 /* first add the high registers, carrying into t2h */   \
204                 ADD128(t2h, ah, z, t2l);                                \
205                 /* double t2h and add top bit of ah */                  \
206                 t2h = 2 * t2h + (ah >> 63);                             \
207                 ah &= m63;                                              \
208                 /* now add the low registers */                         \
209                 ADD128(ah, al, mh, ml);                                 \
210                 ADD128(ah, al, z, t2h);                                 \
211         } while (0)
212
213 #else /* ! CONFIG_64BIT */
214
215 #ifndef nh_16
216 #define nh_16(mp, kp, nw, rh, rl)                                       \
217         do {                                                            \
218                 u64 t1, t2, m1, m2, t;                                  \
219                 int i;                                                  \
220                 rh = rl = t = 0;                                        \
221                 for (i = 0; i < nw; i += 2)  {                          \
222                         t1 = pe64_to_cpup(mp+i) + kp[i];                \
223                         t2 = pe64_to_cpup(mp+i+1) + kp[i+1];            \
224                         m2 = MUL32(t1 >> 32, t2);                       \
225                         m1 = MUL32(t1, t2 >> 32);                       \
226                         ADD128(rh, rl, MUL32(t1 >> 32, t2 >> 32),       \
227                                 MUL32(t1, t2));                         \
228                         rh += (u64)(u32)(m1 >> 32)                      \
229                                 + (u32)(m2 >> 32);                      \
230                         t += (u64)(u32)m1 + (u32)m2;                    \
231                 }                                                       \
232                 ADD128(rh, rl, (t >> 32), (t << 32));                   \
233         } while (0)
234 #endif
235
236 static void poly_step_func(u64 *ahi, u64 *alo,
237                         const u64 *kh, const u64 *kl,
238                         const u64 *mh, const u64 *ml)
239 {
240 #define a0 (*(((u32 *)alo)+INDEX_LOW))
241 #define a1 (*(((u32 *)alo)+INDEX_HIGH))
242 #define a2 (*(((u32 *)ahi)+INDEX_LOW))
243 #define a3 (*(((u32 *)ahi)+INDEX_HIGH))
244 #define k0 (*(((u32 *)kl)+INDEX_LOW))
245 #define k1 (*(((u32 *)kl)+INDEX_HIGH))
246 #define k2 (*(((u32 *)kh)+INDEX_LOW))
247 #define k3 (*(((u32 *)kh)+INDEX_HIGH))
248
249         u64 p, q, t;
250         u32 t2;
251
252         p = MUL32(a3, k3);
253         p += p;
254         p += *(u64 *)mh;
255         p += MUL32(a0, k2);
256         p += MUL32(a1, k1);
257         p += MUL32(a2, k0);
258         t = (u32)(p);
259         p >>= 32;
260         p += MUL32(a0, k3);
261         p += MUL32(a1, k2);
262         p += MUL32(a2, k1);
263         p += MUL32(a3, k0);
264         t |= ((u64)((u32)p & 0x7fffffff)) << 32;
265         p >>= 31;
266         p += (u64)(((u32 *)ml)[INDEX_LOW]);
267         p += MUL32(a0, k0);
268         q =  MUL32(a1, k3);
269         q += MUL32(a2, k2);
270         q += MUL32(a3, k1);
271         q += q;
272         p += q;
273         t2 = (u32)(p);
274         p >>= 32;
275         p += (u64)(((u32 *)ml)[INDEX_HIGH]);
276         p += MUL32(a0, k1);
277         p += MUL32(a1, k0);
278         q =  MUL32(a2, k3);
279         q += MUL32(a3, k2);
280         q += q;
281         p += q;
282         *(u64 *)(alo) = (p << 32) | t2;
283         p >>= 32;
284         *(u64 *)(ahi) = p + t;
285
286 #undef a0
287 #undef a1
288 #undef a2
289 #undef a3
290 #undef k0
291 #undef k1
292 #undef k2
293 #undef k3
294 }
295
296 #define poly_step(ah, al, kh, kl, mh, ml)                               \
297         poly_step_func(&(ah), &(al), &(kh), &(kl), &(mh), &(ml))
298
299 #endif  /* end of specialized NH and poly definitions */
300
301 /* At least nh_16 is defined. Defined others as needed here */
302 #ifndef nh_16_2
303 #define nh_16_2(mp, kp, nw, rh, rl, rh2, rl2)                           \
304         do {                                                            \
305                 nh_16(mp, kp, nw, rh, rl);                              \
306                 nh_16(mp, ((kp)+2), nw, rh2, rl2);                      \
307         } while (0)
308 #endif
309 #ifndef nh_vmac_nhbytes
310 #define nh_vmac_nhbytes(mp, kp, nw, rh, rl)                             \
311         nh_16(mp, kp, nw, rh, rl)
312 #endif
313 #ifndef nh_vmac_nhbytes_2
314 #define nh_vmac_nhbytes_2(mp, kp, nw, rh, rl, rh2, rl2)                 \
315         do {                                                            \
316                 nh_vmac_nhbytes(mp, kp, nw, rh, rl);                    \
317                 nh_vmac_nhbytes(mp, ((kp)+2), nw, rh2, rl2);            \
318         } while (0)
319 #endif
320
321 static void vhash_abort(struct vmac_ctx *ctx)
322 {
323         ctx->polytmp[0] = ctx->polykey[0] ;
324         ctx->polytmp[1] = ctx->polykey[1] ;
325         ctx->first_block_processed = 0;
326 }
327
328 static u64 l3hash(u64 p1, u64 p2, u64 k1, u64 k2, u64 len)
329 {
330         u64 rh, rl, t, z = 0;
331
332         /* fully reduce (p1,p2)+(len,0) mod p127 */
333         t = p1 >> 63;
334         p1 &= m63;
335         ADD128(p1, p2, len, t);
336         /* At this point, (p1,p2) is at most 2^127+(len<<64) */
337         t = (p1 > m63) + ((p1 == m63) && (p2 == m64));
338         ADD128(p1, p2, z, t);
339         p1 &= m63;
340
341         /* compute (p1,p2)/(2^64-2^32) and (p1,p2)%(2^64-2^32) */
342         t = p1 + (p2 >> 32);
343         t += (t >> 32);
344         t += (u32)t > 0xfffffffeu;
345         p1 += (t >> 32);
346         p2 += (p1 << 32);
347
348         /* compute (p1+k1)%p64 and (p2+k2)%p64 */
349         p1 += k1;
350         p1 += (0 - (p1 < k1)) & 257;
351         p2 += k2;
352         p2 += (0 - (p2 < k2)) & 257;
353
354         /* compute (p1+k1)*(p2+k2)%p64 */
355         MUL64(rh, rl, p1, p2);
356         t = rh >> 56;
357         ADD128(t, rl, z, rh);
358         rh <<= 8;
359         ADD128(t, rl, z, rh);
360         t += t << 8;
361         rl += t;
362         rl += (0 - (rl < t)) & 257;
363         rl += (0 - (rl > p64-1)) & 257;
364         return rl;
365 }
366
367 static void vhash_update(const unsigned char *m,
368                         unsigned int mbytes, /* Pos multiple of VMAC_NHBYTES */
369                         struct vmac_ctx *ctx)
370 {
371         u64 rh, rl, *mptr;
372         const u64 *kptr = (u64 *)ctx->nhkey;
373         int i;
374         u64 ch, cl;
375         u64 pkh = ctx->polykey[0];
376         u64 pkl = ctx->polykey[1];
377
378         if (!mbytes)
379                 return;
380
381         BUG_ON(mbytes % VMAC_NHBYTES);
382
383         mptr = (u64 *)m;
384         i = mbytes / VMAC_NHBYTES;  /* Must be non-zero */
385
386         ch = ctx->polytmp[0];
387         cl = ctx->polytmp[1];
388
389         if (!ctx->first_block_processed) {
390                 ctx->first_block_processed = 1;
391                 nh_vmac_nhbytes(mptr, kptr, VMAC_NHBYTES/8, rh, rl);
392                 rh &= m62;
393                 ADD128(ch, cl, rh, rl);
394                 mptr += (VMAC_NHBYTES/sizeof(u64));
395                 i--;
396         }
397
398         while (i--) {
399                 nh_vmac_nhbytes(mptr, kptr, VMAC_NHBYTES/8, rh, rl);
400                 rh &= m62;
401                 poly_step(ch, cl, pkh, pkl, rh, rl);
402                 mptr += (VMAC_NHBYTES/sizeof(u64));
403         }
404
405         ctx->polytmp[0] = ch;
406         ctx->polytmp[1] = cl;
407 }
408
409 static u64 vhash(unsigned char m[], unsigned int mbytes,
410                         u64 *tagl, struct vmac_ctx *ctx)
411 {
412         u64 rh, rl, *mptr;
413         const u64 *kptr = (u64 *)ctx->nhkey;
414         int i, remaining;
415         u64 ch, cl;
416         u64 pkh = ctx->polykey[0];
417         u64 pkl = ctx->polykey[1];
418
419         mptr = (u64 *)m;
420         i = mbytes / VMAC_NHBYTES;
421         remaining = mbytes % VMAC_NHBYTES;
422
423         if (ctx->first_block_processed) {
424                 ch = ctx->polytmp[0];
425                 cl = ctx->polytmp[1];
426         } else if (i) {
427                 nh_vmac_nhbytes(mptr, kptr, VMAC_NHBYTES/8, ch, cl);
428                 ch &= m62;
429                 ADD128(ch, cl, pkh, pkl);
430                 mptr += (VMAC_NHBYTES/sizeof(u64));
431                 i--;
432         } else if (remaining) {
433                 nh_16(mptr, kptr, 2*((remaining+15)/16), ch, cl);
434                 ch &= m62;
435                 ADD128(ch, cl, pkh, pkl);
436                 mptr += (VMAC_NHBYTES/sizeof(u64));
437                 goto do_l3;
438         } else {/* Empty String */
439                 ch = pkh; cl = pkl;
440                 goto do_l3;
441         }
442
443         while (i--) {
444                 nh_vmac_nhbytes(mptr, kptr, VMAC_NHBYTES/8, rh, rl);
445                 rh &= m62;
446                 poly_step(ch, cl, pkh, pkl, rh, rl);
447                 mptr += (VMAC_NHBYTES/sizeof(u64));
448         }
449         if (remaining) {
450                 nh_16(mptr, kptr, 2*((remaining+15)/16), rh, rl);
451                 rh &= m62;
452                 poly_step(ch, cl, pkh, pkl, rh, rl);
453         }
454
455 do_l3:
456         vhash_abort(ctx);
457         remaining *= 8;
458         return l3hash(ch, cl, ctx->l3key[0], ctx->l3key[1], remaining);
459 }
460
461 static u64 vmac(unsigned char m[], unsigned int mbytes,
462                         const unsigned char n[16], u64 *tagl,
463                         struct vmac_ctx_t *ctx)
464 {
465         u64 *in_n, *out_p;
466         u64 p, h;
467         int i;
468
469         in_n = ctx->__vmac_ctx.cached_nonce;
470         out_p = ctx->__vmac_ctx.cached_aes;
471
472         i = n[15] & 1;
473         if ((*(u64 *)(n+8) != in_n[1]) || (*(u64 *)(n) != in_n[0])) {
474                 in_n[0] = *(u64 *)(n);
475                 in_n[1] = *(u64 *)(n+8);
476                 ((unsigned char *)in_n)[15] &= 0xFE;
477                 crypto_cipher_encrypt_one(ctx->child,
478                         (unsigned char *)out_p, (unsigned char *)in_n);
479
480                 ((unsigned char *)in_n)[15] |= (unsigned char)(1-i);
481         }
482         p = be64_to_cpup(out_p + i);
483         h = vhash(m, mbytes, (u64 *)0, &ctx->__vmac_ctx);
484         return le64_to_cpu(p + h);
485 }
486
487 static int vmac_set_key(unsigned char user_key[], struct vmac_ctx_t *ctx)
488 {
489         u64 in[2] = {0}, out[2];
490         unsigned i;
491         int err = 0;
492
493         err = crypto_cipher_setkey(ctx->child, user_key, VMAC_KEY_LEN);
494         if (err)
495                 return err;
496
497         /* Fill nh key */
498         ((unsigned char *)in)[0] = 0x80;
499         for (i = 0; i < sizeof(ctx->__vmac_ctx.nhkey)/8; i += 2) {
500                 crypto_cipher_encrypt_one(ctx->child,
501                         (unsigned char *)out, (unsigned char *)in);
502                 ctx->__vmac_ctx.nhkey[i] = be64_to_cpup(out);
503                 ctx->__vmac_ctx.nhkey[i+1] = be64_to_cpup(out+1);
504                 ((unsigned char *)in)[15] += 1;
505         }
506
507         /* Fill poly key */
508         ((unsigned char *)in)[0] = 0xC0;
509         in[1] = 0;
510         for (i = 0; i < sizeof(ctx->__vmac_ctx.polykey)/8; i += 2) {
511                 crypto_cipher_encrypt_one(ctx->child,
512                         (unsigned char *)out, (unsigned char *)in);
513                 ctx->__vmac_ctx.polytmp[i] =
514                         ctx->__vmac_ctx.polykey[i] =
515                                 be64_to_cpup(out) & mpoly;
516                 ctx->__vmac_ctx.polytmp[i+1] =
517                         ctx->__vmac_ctx.polykey[i+1] =
518                                 be64_to_cpup(out+1) & mpoly;
519                 ((unsigned char *)in)[15] += 1;
520         }
521
522         /* Fill ip key */
523         ((unsigned char *)in)[0] = 0xE0;
524         in[1] = 0;
525         for (i = 0; i < sizeof(ctx->__vmac_ctx.l3key)/8; i += 2) {
526                 do {
527                         crypto_cipher_encrypt_one(ctx->child,
528                                 (unsigned char *)out, (unsigned char *)in);
529                         ctx->__vmac_ctx.l3key[i] = be64_to_cpup(out);
530                         ctx->__vmac_ctx.l3key[i+1] = be64_to_cpup(out+1);
531                         ((unsigned char *)in)[15] += 1;
532                 } while (ctx->__vmac_ctx.l3key[i] >= p64
533                         || ctx->__vmac_ctx.l3key[i+1] >= p64);
534         }
535
536         /* Invalidate nonce/aes cache and reset other elements */
537         ctx->__vmac_ctx.cached_nonce[0] = (u64)-1; /* Ensure illegal nonce */
538         ctx->__vmac_ctx.cached_nonce[1] = (u64)0;  /* Ensure illegal nonce */
539         ctx->__vmac_ctx.first_block_processed = 0;
540
541         return err;
542 }
543
544 static int vmac_setkey(struct crypto_shash *parent,
545                 const u8 *key, unsigned int keylen)
546 {
547         struct vmac_ctx_t *ctx = crypto_shash_ctx(parent);
548
549         if (keylen != VMAC_KEY_LEN) {
550                 crypto_shash_set_flags(parent, CRYPTO_TFM_RES_BAD_KEY_LEN);
551                 return -EINVAL;
552         }
553
554         return vmac_set_key((u8 *)key, ctx);
555 }
556
557 static int vmac_init(struct shash_desc *pdesc)
558 {
559         return 0;
560 }
561
562 static int vmac_update(struct shash_desc *pdesc, const u8 *p,
563                 unsigned int len)
564 {
565         struct crypto_shash *parent = pdesc->tfm;
566         struct vmac_ctx_t *ctx = crypto_shash_ctx(parent);
567         int expand;
568         int min;
569
570         expand = VMAC_NHBYTES - ctx->partial_size > 0 ?
571                         VMAC_NHBYTES - ctx->partial_size : 0;
572
573         min = len < expand ? len : expand;
574
575         memcpy(ctx->partial + ctx->partial_size, p, min);
576         ctx->partial_size += min;
577
578         if (len < expand)
579                 return 0;
580
581         vhash_update(ctx->partial, VMAC_NHBYTES, &ctx->__vmac_ctx);
582         ctx->partial_size = 0;
583
584         len -= expand;
585         p += expand;
586
587         if (len % VMAC_NHBYTES) {
588                 memcpy(ctx->partial, p + len - (len % VMAC_NHBYTES),
589                         len % VMAC_NHBYTES);
590                 ctx->partial_size = len % VMAC_NHBYTES;
591         }
592
593         vhash_update(p, len - len % VMAC_NHBYTES, &ctx->__vmac_ctx);
594
595         return 0;
596 }
597
598 static int vmac_final(struct shash_desc *pdesc, u8 *out)
599 {
600         struct crypto_shash *parent = pdesc->tfm;
601         struct vmac_ctx_t *ctx = crypto_shash_ctx(parent);
602         vmac_t mac;
603         u8 nonce[16] = {};
604
605         /* vmac() ends up accessing outside the array bounds that
606          * we specify.  In appears to access up to the next 2-word
607          * boundary.  We'll just be uber cautious and zero the
608          * unwritten bytes in the buffer.
609          */
610         if (ctx->partial_size) {
611                 memset(ctx->partial + ctx->partial_size, 0,
612                         VMAC_NHBYTES - ctx->partial_size);
613         }
614         mac = vmac(ctx->partial, ctx->partial_size, nonce, NULL, ctx);
615         memcpy(out, &mac, sizeof(vmac_t));
616         memzero_explicit(&mac, sizeof(vmac_t));
617         memset(&ctx->__vmac_ctx, 0, sizeof(struct vmac_ctx));
618         ctx->partial_size = 0;
619         return 0;
620 }
621
622 static int vmac_init_tfm(struct crypto_tfm *tfm)
623 {
624         struct crypto_cipher *cipher;
625         struct crypto_instance *inst = (void *)tfm->__crt_alg;
626         struct crypto_spawn *spawn = crypto_instance_ctx(inst);
627         struct vmac_ctx_t *ctx = crypto_tfm_ctx(tfm);
628
629         cipher = crypto_spawn_cipher(spawn);
630         if (IS_ERR(cipher))
631                 return PTR_ERR(cipher);
632
633         ctx->child = cipher;
634         return 0;
635 }
636
637 static void vmac_exit_tfm(struct crypto_tfm *tfm)
638 {
639         struct vmac_ctx_t *ctx = crypto_tfm_ctx(tfm);
640         crypto_free_cipher(ctx->child);
641 }
642
643 static int vmac_create(struct crypto_template *tmpl, struct rtattr **tb)
644 {
645         struct shash_instance *inst;
646         struct crypto_alg *alg;
647         int err;
648
649         err = crypto_check_attr_type(tb, CRYPTO_ALG_TYPE_SHASH);
650         if (err)
651                 return err;
652
653         alg = crypto_get_attr_alg(tb, CRYPTO_ALG_TYPE_CIPHER,
654                         CRYPTO_ALG_TYPE_MASK);
655         if (IS_ERR(alg))
656                 return PTR_ERR(alg);
657
658         inst = shash_alloc_instance("vmac", alg);
659         err = PTR_ERR(inst);
660         if (IS_ERR(inst))
661                 goto out_put_alg;
662
663         err = crypto_init_spawn(shash_instance_ctx(inst), alg,
664                         shash_crypto_instance(inst),
665                         CRYPTO_ALG_TYPE_MASK);
666         if (err)
667                 goto out_free_inst;
668
669         inst->alg.base.cra_priority = alg->cra_priority;
670         inst->alg.base.cra_blocksize = alg->cra_blocksize;
671         inst->alg.base.cra_alignmask = alg->cra_alignmask;
672
673         inst->alg.digestsize = sizeof(vmac_t);
674         inst->alg.base.cra_ctxsize = sizeof(struct vmac_ctx_t);
675         inst->alg.base.cra_init = vmac_init_tfm;
676         inst->alg.base.cra_exit = vmac_exit_tfm;
677
678         inst->alg.init = vmac_init;
679         inst->alg.update = vmac_update;
680         inst->alg.final = vmac_final;
681         inst->alg.setkey = vmac_setkey;
682
683         err = shash_register_instance(tmpl, inst);
684         if (err) {
685 out_free_inst:
686                 shash_free_instance(shash_crypto_instance(inst));
687         }
688
689 out_put_alg:
690         crypto_mod_put(alg);
691         return err;
692 }
693
694 static struct crypto_template vmac_tmpl = {
695         .name = "vmac",
696         .create = vmac_create,
697         .free = shash_free_instance,
698         .module = THIS_MODULE,
699 };
700
701 static int __init vmac_module_init(void)
702 {
703         return crypto_register_template(&vmac_tmpl);
704 }
705
706 static void __exit vmac_module_exit(void)
707 {
708         crypto_unregister_template(&vmac_tmpl);
709 }
710
711 module_init(vmac_module_init);
712 module_exit(vmac_module_exit);
713
714 MODULE_LICENSE("GPL");
715 MODULE_DESCRIPTION("VMAC hash algorithm");
716 MODULE_ALIAS_CRYPTO("vmac");