crypto/vmac.c

   1 /*
   2  * VMAC: Message Authentication Code using Universal Hashing
   3  *
   4  * Reference: https://tools.ietf.org/html/draft-krovetz-vmac-01
   5  *
   6  * Copyright (c) 2009, Intel Corporation.
   7  * Copyright (c) 2018, Google Inc.
   8  *
   9  * This program is free software; you can redistribute it and/or modify it
  10  * under the terms and conditions of the GNU General Public License,
  11  * version 2, as published by the Free Software Foundation.
  12  *
  13  * This program is distributed in the hope it will be useful, but WITHOUT
  14  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  15  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
  16  * more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along with
  19  * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
  20  * Place - Suite 330, Boston, MA 02111-1307 USA.
  21  */
  22
  23 /*
  24  * Derived from:
  25  *      VMAC and VHASH Implementation by Ted Krovetz (tdk@acm.org) and Wei Dai.
  26  *      This implementation is herby placed in the public domain.
  27  *      The authors offers no warranty. Use at your own risk.
  28  *      Last modified: 17 APR 08, 1700 PDT
  29  */
  30
  31 #include <asm/unaligned.h>
  32 #include <linux/init.h>
  33 #include <linux/types.h>
  34 #include <linux/crypto.h>
  35 #include <linux/module.h>
  36 #include <linux/scatterlist.h>
  37 #include <asm/byteorder.h>
  38 #include <crypto/scatterwalk.h>
  39 #include <crypto/internal/hash.h>
  40
  41 /*
  42  * User definable settings.
  43  */
  44 #define VMAC_TAG_LEN    64
  45 #define VMAC_KEY_SIZE   128/* Must be 128, 192 or 256                   */
  46 #define VMAC_KEY_LEN    (VMAC_KEY_SIZE/8)
  47 #define VMAC_NHBYTES    128/* Must 2^i for any 3 < i < 13 Standard = 128*/
  48 #define VMAC_NONCEBYTES 16
  49
  50 /* per-transform (per-key) context */
  51 struct vmac_tfm_ctx {
  52         struct crypto_cipher *cipher;
  53         u64 nhkey[(VMAC_NHBYTES/8)+2*(VMAC_TAG_LEN/64-1)];
  54         u64 polykey[2*VMAC_TAG_LEN/64];
  55         u64 l3key[2*VMAC_TAG_LEN/64];
  56 };
  57
  58 /* per-request context */
  59 struct vmac_desc_ctx {
  60         union {
  61                 u8 partial[VMAC_NHBYTES];       /* partial block */
  62                 __le64 partial_words[VMAC_NHBYTES / 8];
  63         };
  64         unsigned int partial_size;      /* size of the partial block */
  65         bool first_block_processed;
  66         u64 polytmp[2*VMAC_TAG_LEN/64]; /* running total of L2-hash */
  67         union {
  68                 u8 bytes[VMAC_NONCEBYTES];
  69                 __be64 pads[VMAC_NONCEBYTES / 8];
  70         } nonce;
  71         unsigned int nonce_size; /* nonce bytes filled so far */
  72 };
  73
  74 /*
  75  * Constants and masks
  76  */
  77 #define UINT64_C(x) x##ULL
  78 static const u64 p64   = UINT64_C(0xfffffffffffffeff);  /* 2^64 - 257 prime  */
  79 static const u64 m62   = UINT64_C(0x3fffffffffffffff);  /* 62-bit mask       */
  80 static const u64 m63   = UINT64_C(0x7fffffffffffffff);  /* 63-bit mask       */
  81 static const u64 m64   = UINT64_C(0xffffffffffffffff);  /* 64-bit mask       */
  82 static const u64 mpoly = UINT64_C(0x1fffffff1fffffff);  /* Poly key mask     */
  83
  84 #define pe64_to_cpup le64_to_cpup               /* Prefer little endian */
  85
  86 #ifdef __LITTLE_ENDIAN
  87 #define INDEX_HIGH 1
  88 #define INDEX_LOW 0
  89 #else
  90 #define INDEX_HIGH 0
  91 #define INDEX_LOW 1
  92 #endif
  93
  94 /*
  95  * The following routines are used in this implementation. They are
  96  * written via macros to simulate zero-overhead call-by-reference.
  97  *
  98  * MUL64: 64x64->128-bit multiplication
  99  * PMUL64: assumes top bits cleared on inputs
 100  * ADD128: 128x128->128-bit addition
 101  */
 102
 103 #define ADD128(rh, rl, ih, il)                                          \
 104         do {                                                            \
 105                 u64 _il = (il);                                         \
 106                 (rl) += (_il);                                          \
 107                 if ((rl) < (_il))                                       \
 108                         (rh)++;                                         \
 109                 (rh) += (ih);                                           \
 110         } while (0)
 111
 112 #define MUL32(i1, i2)   ((u64)(u32)(i1)*(u32)(i2))
 113
 114 #define PMUL64(rh, rl, i1, i2)  /* Assumes m doesn't overflow */        \
 115         do {                                                            \
 116                 u64 _i1 = (i1), _i2 = (i2);                             \
 117                 u64 m = MUL32(_i1, _i2>>32) + MUL32(_i1>>32, _i2);      \
 118                 rh = MUL32(_i1>>32, _i2>>32);                           \
 119                 rl = MUL32(_i1, _i2);                                   \
 120                 ADD128(rh, rl, (m >> 32), (m << 32));                   \
 121         } while (0)
 122
 123 #define MUL64(rh, rl, i1, i2)                                           \
 124         do {                                                            \
 125                 u64 _i1 = (i1), _i2 = (i2);                             \
 126                 u64 m1 = MUL32(_i1, _i2>>32);                           \
 127                 u64 m2 = MUL32(_i1>>32, _i2);                           \
 128                 rh = MUL32(_i1>>32, _i2>>32);                           \
 129                 rl = MUL32(_i1, _i2);                                   \
 130                 ADD128(rh, rl, (m1 >> 32), (m1 << 32));                 \
 131                 ADD128(rh, rl, (m2 >> 32), (m2 << 32));                 \
 132         } while (0)
 133
 134 /*
 135  * For highest performance the L1 NH and L2 polynomial hashes should be
 136  * carefully implemented to take advantage of one's target architecture.
 137  * Here these two hash functions are defined multiple time; once for
 138  * 64-bit architectures, once for 32-bit SSE2 architectures, and once
 139  * for the rest (32-bit) architectures.
 140  * For each, nh_16 *must* be defined (works on multiples of 16 bytes).
 141  * Optionally, nh_vmac_nhbytes can be defined (for multiples of
 142  * VMAC_NHBYTES), and nh_16_2 and nh_vmac_nhbytes_2 (versions that do two
 143  * NH computations at once).
 144  */
 145
 146 #ifdef CONFIG_64BIT
 147
 148 #define nh_16(mp, kp, nw, rh, rl)                                       \
 149         do {                                                            \
 150                 int i; u64 th, tl;                                      \
 151                 rh = rl = 0;                                            \
 152                 for (i = 0; i < nw; i += 2) {                           \
 153                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i],     \
 154                                 pe64_to_cpup((mp)+i+1)+(kp)[i+1]);      \
 155                         ADD128(rh, rl, th, tl);                         \
 156                 }                                                       \
 157         } while (0)
 158
 159 #define nh_16_2(mp, kp, nw, rh, rl, rh1, rl1)                           \
 160         do {                                                            \
 161                 int i; u64 th, tl;                                      \
 162                 rh1 = rl1 = rh = rl = 0;                                \
 163                 for (i = 0; i < nw; i += 2) {                           \
 164                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i],     \
 165                                 pe64_to_cpup((mp)+i+1)+(kp)[i+1]);      \
 166                         ADD128(rh, rl, th, tl);                         \
 167                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i+2],   \
 168                                 pe64_to_cpup((mp)+i+1)+(kp)[i+3]);      \
 169                         ADD128(rh1, rl1, th, tl);                       \
 170                 }                                                       \
 171         } while (0)
 172
 173 #if (VMAC_NHBYTES >= 64) /* These versions do 64-bytes of message at a time */
 174 #define nh_vmac_nhbytes(mp, kp, nw, rh, rl)                             \
 175         do {                                                            \
 176                 int i; u64 th, tl;                                      \
 177                 rh = rl = 0;                                            \
 178                 for (i = 0; i < nw; i += 8) {                           \
 179                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i],     \
 180                                 pe64_to_cpup((mp)+i+1)+(kp)[i+1]);      \
 181                         ADD128(rh, rl, th, tl);                         \
 182                         MUL64(th, tl, pe64_to_cpup((mp)+i+2)+(kp)[i+2], \
 183                                 pe64_to_cpup((mp)+i+3)+(kp)[i+3]);      \
 184                         ADD128(rh, rl, th, tl);                         \
 185                         MUL64(th, tl, pe64_to_cpup((mp)+i+4)+(kp)[i+4], \
 186                                 pe64_to_cpup((mp)+i+5)+(kp)[i+5]);      \
 187                         ADD128(rh, rl, th, tl);                         \
 188                         MUL64(th, tl, pe64_to_cpup((mp)+i+6)+(kp)[i+6], \
 189                                 pe64_to_cpup((mp)+i+7)+(kp)[i+7]);      \
 190                         ADD128(rh, rl, th, tl);                         \
 191                 }                                                       \
 192         } while (0)
 193
 194 #define nh_vmac_nhbytes_2(mp, kp, nw, rh, rl, rh1, rl1)                 \
 195         do {                                                            \
 196                 int i; u64 th, tl;                                      \
 197                 rh1 = rl1 = rh = rl = 0;                                \
 198                 for (i = 0; i < nw; i += 8) {                           \
 199                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i],     \
 200                                 pe64_to_cpup((mp)+i+1)+(kp)[i+1]);      \
 201                         ADD128(rh, rl, th, tl);                         \
 202                         MUL64(th, tl, pe64_to_cpup((mp)+i)+(kp)[i+2],   \
 203                                 pe64_to_cpup((mp)+i+1)+(kp)[i+3]);      \
 204                         ADD128(rh1, rl1, th, tl);                       \
 205                         MUL64(th, tl, pe64_to_cpup((mp)+i+2)+(kp)[i+2], \
 206                                 pe64_to_cpup((mp)+i+3)+(kp)[i+3]);      \
 207                         ADD128(rh, rl, th, tl);                         \
 208                         MUL64(th, tl, pe64_to_cpup((mp)+i+2)+(kp)[i+4], \
 209                                 pe64_to_cpup((mp)+i+3)+(kp)[i+5]);      \
 210                         ADD128(rh1, rl1, th, tl);                       \
 211                         MUL64(th, tl, pe64_to_cpup((mp)+i+4)+(kp)[i+4], \
 212                                 pe64_to_cpup((mp)+i+5)+(kp)[i+5]);      \
 213                         ADD128(rh, rl, th, tl);                         \
 214                         MUL64(th, tl, pe64_to_cpup((mp)+i+4)+(kp)[i+6], \
 215                                 pe64_to_cpup((mp)+i+5)+(kp)[i+7]);      \
 216                         ADD128(rh1, rl1, th, tl);                       \
 217                         MUL64(th, tl, pe64_to_cpup((mp)+i+6)+(kp)[i+6], \
 218                                 pe64_to_cpup((mp)+i+7)+(kp)[i+7]);      \
 219                         ADD128(rh, rl, th, tl);                         \
 220                         MUL64(th, tl, pe64_to_cpup((mp)+i+6)+(kp)[i+8], \
 221                                 pe64_to_cpup((mp)+i+7)+(kp)[i+9]);      \
 222                         ADD128(rh1, rl1, th, tl);                       \
 223                 }                                                       \
 224         } while (0)
 225 #endif
 226
 227 #define poly_step(ah, al, kh, kl, mh, ml)                               \
 228         do {                                                            \
 229                 u64 t1h, t1l, t2h, t2l, t3h, t3l, z = 0;                \
 230                 /* compute ab*cd, put bd into result registers */       \
 231                 PMUL64(t3h, t3l, al, kh);                               \
 232                 PMUL64(t2h, t2l, ah, kl);                               \
 233                 PMUL64(t1h, t1l, ah, 2*kh);                             \
 234                 PMUL64(ah, al, al, kl);                                 \
 235                 /* add 2 * ac to result */                              \
 236                 ADD128(ah, al, t1h, t1l);                               \
 237                 /* add together ad + bc */                              \
 238                 ADD128(t2h, t2l, t3h, t3l);                             \
 239                 /* now (ah,al), (t2l,2*t2h) need summing */             \
 240                 /* first add the high registers, carrying into t2h */   \
 241                 ADD128(t2h, ah, z, t2l);                                \
 242                 /* double t2h and add top bit of ah */                  \
 243                 t2h = 2 * t2h + (ah >> 63);                             \
 244                 ah &= m63;                                              \
 245                 /* now add the low registers */                         \
 246                 ADD128(ah, al, mh, ml);                                 \
 247                 ADD128(ah, al, z, t2h);                                 \
 248         } while (0)
 249
 250 #else /* ! CONFIG_64BIT */
 251
 252 #ifndef nh_16
 253 #define nh_16(mp, kp, nw, rh, rl)                                       \
 254         do {                                                            \
 255                 u64 t1, t2, m1, m2, t;                                  \
 256                 int i;                                                  \
 257                 rh = rl = t = 0;                                        \
 258                 for (i = 0; i < nw; i += 2)  {                          \
 259                         t1 = pe64_to_cpup(mp+i) + kp[i];                \
 260                         t2 = pe64_to_cpup(mp+i+1) + kp[i+1];            \
 261                         m2 = MUL32(t1 >> 32, t2);                       \
 262                         m1 = MUL32(t1, t2 >> 32);                       \
 263                         ADD128(rh, rl, MUL32(t1 >> 32, t2 >> 32),       \
 264                                 MUL32(t1, t2));                         \
 265                         rh += (u64)(u32)(m1 >> 32)                      \
 266                                 + (u32)(m2 >> 32);                      \
 267                         t += (u64)(u32)m1 + (u32)m2;                    \
 268                 }                                                       \
 269                 ADD128(rh, rl, (t >> 32), (t << 32));                   \
 270         } while (0)
 271 #endif
 272
 273 static void poly_step_func(u64 *ahi, u64 *alo,
 274                         const u64 *kh, const u64 *kl,
 275                         const u64 *mh, const u64 *ml)
 276 {
 277 #define a0 (*(((u32 *)alo)+INDEX_LOW))
 278 #define a1 (*(((u32 *)alo)+INDEX_HIGH))
 279 #define a2 (*(((u32 *)ahi)+INDEX_LOW))
 280 #define a3 (*(((u32 *)ahi)+INDEX_HIGH))
 281 #define k0 (*(((u32 *)kl)+INDEX_LOW))
 282 #define k1 (*(((u32 *)kl)+INDEX_HIGH))
 283 #define k2 (*(((u32 *)kh)+INDEX_LOW))
 284 #define k3 (*(((u32 *)kh)+INDEX_HIGH))
 285
 286         u64 p, q, t;
 287         u32 t2;
 288
 289         p = MUL32(a3, k3);
 290         p += p;
 291         p += *(u64 *)mh;
 292         p += MUL32(a0, k2);
 293         p += MUL32(a1, k1);
 294         p += MUL32(a2, k0);
 295         t = (u32)(p);
 296         p >>= 32;
 297         p += MUL32(a0, k3);
 298         p += MUL32(a1, k2);
 299         p += MUL32(a2, k1);
 300         p += MUL32(a3, k0);
 301         t |= ((u64)((u32)p & 0x7fffffff)) << 32;
 302         p >>= 31;
 303         p += (u64)(((u32 *)ml)[INDEX_LOW]);
 304         p += MUL32(a0, k0);
 305         q =  MUL32(a1, k3);
 306         q += MUL32(a2, k2);
 307         q += MUL32(a3, k1);
 308         q += q;
 309         p += q;
 310         t2 = (u32)(p);
 311         p >>= 32;
 312         p += (u64)(((u32 *)ml)[INDEX_HIGH]);
 313         p += MUL32(a0, k1);
 314         p += MUL32(a1, k0);
 315         q =  MUL32(a2, k3);
 316         q += MUL32(a3, k2);
 317         q += q;
 318         p += q;
 319         *(u64 *)(alo) = (p << 32) | t2;
 320         p >>= 32;
 321         *(u64 *)(ahi) = p + t;
 322
 323 #undef a0
 324 #undef a1
 325 #undef a2
 326 #undef a3
 327 #undef k0
 328 #undef k1
 329 #undef k2
 330 #undef k3
 331 }
 332
 333 #define poly_step(ah, al, kh, kl, mh, ml)                               \
 334         poly_step_func(&(ah), &(al), &(kh), &(kl), &(mh), &(ml))
 335
 336 #endif  /* end of specialized NH and poly definitions */
 337
 338 /* At least nh_16 is defined. Defined others as needed here */
 339 #ifndef nh_16_2
 340 #define nh_16_2(mp, kp, nw, rh, rl, rh2, rl2)                           \
 341         do {                                                            \
 342                 nh_16(mp, kp, nw, rh, rl);                              \
 343                 nh_16(mp, ((kp)+2), nw, rh2, rl2);                      \
 344         } while (0)
 345 #endif
 346 #ifndef nh_vmac_nhbytes
 347 #define nh_vmac_nhbytes(mp, kp, nw, rh, rl)                             \
 348         nh_16(mp, kp, nw, rh, rl)
 349 #endif
 350 #ifndef nh_vmac_nhbytes_2
 351 #define nh_vmac_nhbytes_2(mp, kp, nw, rh, rl, rh2, rl2)                 \
 352         do {                                                            \
 353                 nh_vmac_nhbytes(mp, kp, nw, rh, rl);                    \
 354                 nh_vmac_nhbytes(mp, ((kp)+2), nw, rh2, rl2);            \
 355         } while (0)
 356 #endif
 357
 358 static u64 l3hash(u64 p1, u64 p2, u64 k1, u64 k2, u64 len)
 359 {
 360         u64 rh, rl, t, z = 0;
 361
 362         /* fully reduce (p1,p2)+(len,0) mod p127 */
 363         t = p1 >> 63;
 364         p1 &= m63;
 365         ADD128(p1, p2, len, t);
 366         /* At this point, (p1,p2) is at most 2^127+(len<<64) */
 367         t = (p1 > m63) + ((p1 == m63) && (p2 == m64));
 368         ADD128(p1, p2, z, t);
 369         p1 &= m63;
 370
 371         /* compute (p1,p2)/(2^64-2^32) and (p1,p2)%(2^64-2^32) */
 372         t = p1 + (p2 >> 32);
 373         t += (t >> 32);
 374         t += (u32)t > 0xfffffffeu;
 375         p1 += (t >> 32);
 376         p2 += (p1 << 32);
 377
 378         /* compute (p1+k1)%p64 and (p2+k2)%p64 */
 379         p1 += k1;
 380         p1 += (0 - (p1 < k1)) & 257;
 381         p2 += k2;
 382         p2 += (0 - (p2 < k2)) & 257;
 383
 384         /* compute (p1+k1)*(p2+k2)%p64 */
 385         MUL64(rh, rl, p1, p2);
 386         t = rh >> 56;
 387         ADD128(t, rl, z, rh);
 388         rh <<= 8;
 389         ADD128(t, rl, z, rh);
 390         t += t << 8;
 391         rl += t;
 392         rl += (0 - (rl < t)) & 257;
 393         rl += (0 - (rl > p64-1)) & 257;
 394         return rl;
 395 }
 396
 397 /* L1 and L2-hash one or more VMAC_NHBYTES-byte blocks */
 398 static void vhash_blocks(const struct vmac_tfm_ctx *tctx,
 399                          struct vmac_desc_ctx *dctx,
 400                          const __le64 *mptr, unsigned int blocks)
 401 {
 402         const u64 *kptr = tctx->nhkey;
 403         const u64 pkh = tctx->polykey[0];
 404         const u64 pkl = tctx->polykey[1];
 405         u64 ch = dctx->polytmp[0];
 406         u64 cl = dctx->polytmp[1];
 407         u64 rh, rl;
 408
 409         if (!dctx->first_block_processed) {
 410                 dctx->first_block_processed = true;
 411                 nh_vmac_nhbytes(mptr, kptr, VMAC_NHBYTES/8, rh, rl);
 412                 rh &= m62;
 413                 ADD128(ch, cl, rh, rl);
 414                 mptr += (VMAC_NHBYTES/sizeof(u64));
 415                 blocks--;
 416         }
 417
 418         while (blocks--) {
 419                 nh_vmac_nhbytes(mptr, kptr, VMAC_NHBYTES/8, rh, rl);
 420                 rh &= m62;
 421                 poly_step(ch, cl, pkh, pkl, rh, rl);
 422                 mptr += (VMAC_NHBYTES/sizeof(u64));
 423         }
 424
 425         dctx->polytmp[0] = ch;
 426         dctx->polytmp[1] = cl;
 427 }
 428
 429 static int vmac_setkey(struct crypto_shash *tfm,
 430                        const u8 *key, unsigned int keylen)
 431 {
 432         struct vmac_tfm_ctx *tctx = crypto_shash_ctx(tfm);
 433         __be64 out[2];
 434         u8 in[16] = { 0 };
 435         unsigned int i;
 436         int err;
 437
 438         if (keylen != VMAC_KEY_LEN) {
 439                 crypto_shash_set_flags(tfm, CRYPTO_TFM_RES_BAD_KEY_LEN);
 440                 return -EINVAL;
 441         }
 442
 443         err = crypto_cipher_setkey(tctx->cipher, key, keylen);
 444         if (err)
 445                 return err;
 446
 447         /* Fill nh key */
 448         in[0] = 0x80;
 449         for (i = 0; i < ARRAY_SIZE(tctx->nhkey); i += 2) {
 450                 crypto_cipher_encrypt_one(tctx->cipher, (u8 *)out, in);
 451                 tctx->nhkey[i] = be64_to_cpu(out[0]);
 452                 tctx->nhkey[i+1] = be64_to_cpu(out[1]);
 453                 in[15]++;
 454         }
 455
 456         /* Fill poly key */
 457         in[0] = 0xC0;
 458         in[15] = 0;
 459         for (i = 0; i < ARRAY_SIZE(tctx->polykey); i += 2) {
 460                 crypto_cipher_encrypt_one(tctx->cipher, (u8 *)out, in);
 461                 tctx->polykey[i] = be64_to_cpu(out[0]) & mpoly;
 462                 tctx->polykey[i+1] = be64_to_cpu(out[1]) & mpoly;
 463                 in[15]++;
 464         }
 465
 466         /* Fill ip key */
 467         in[0] = 0xE0;
 468         in[15] = 0;
 469         for (i = 0; i < ARRAY_SIZE(tctx->l3key); i += 2) {
 470                 do {
 471                         crypto_cipher_encrypt_one(tctx->cipher, (u8 *)out, in);
 472                         tctx->l3key[i] = be64_to_cpu(out[0]);
 473                         tctx->l3key[i+1] = be64_to_cpu(out[1]);
 474                         in[15]++;
 475                 } while (tctx->l3key[i] >= p64 || tctx->l3key[i+1] >= p64);
 476         }
 477
 478         return 0;
 479 }
 480
 481 static int vmac_init(struct shash_desc *desc)
 482 {
 483         const struct vmac_tfm_ctx *tctx = crypto_shash_ctx(desc->tfm);
 484         struct vmac_desc_ctx *dctx = shash_desc_ctx(desc);
 485
 486         dctx->partial_size = 0;
 487         dctx->first_block_processed = false;
 488         memcpy(dctx->polytmp, tctx->polykey, sizeof(dctx->polytmp));
 489         dctx->nonce_size = 0;
 490         return 0;
 491 }
 492
 493 static int vmac_update(struct shash_desc *desc, const u8 *p, unsigned int len)
 494 {
 495         const struct vmac_tfm_ctx *tctx = crypto_shash_ctx(desc->tfm);
 496         struct vmac_desc_ctx *dctx = shash_desc_ctx(desc);
 497         unsigned int n;
 498
 499         /* Nonce is passed as first VMAC_NONCEBYTES bytes of data */
 500         if (dctx->nonce_size < VMAC_NONCEBYTES) {
 501                 n = min(len, VMAC_NONCEBYTES - dctx->nonce_size);
 502                 memcpy(&dctx->nonce.bytes[dctx->nonce_size], p, n);
 503                 dctx->nonce_size += n;
 504                 p += n;
 505                 len -= n;
 506         }
 507
 508         if (dctx->partial_size) {
 509                 n = min(len, VMAC_NHBYTES - dctx->partial_size);
 510                 memcpy(&dctx->partial[dctx->partial_size], p, n);
 511                 dctx->partial_size += n;
 512                 p += n;
 513                 len -= n;
 514                 if (dctx->partial_size == VMAC_NHBYTES) {
 515                         vhash_blocks(tctx, dctx, dctx->partial_words, 1);
 516                         dctx->partial_size = 0;
 517                 }
 518         }
 519
 520         if (len >= VMAC_NHBYTES) {
 521                 n = round_down(len, VMAC_NHBYTES);
 522                 /* TODO: 'p' may be misaligned here */
 523                 vhash_blocks(tctx, dctx, (const __le64 *)p, n / VMAC_NHBYTES);
 524                 p += n;
 525                 len -= n;
 526         }
 527
 528         if (len) {
 529                 memcpy(dctx->partial, p, len);
 530                 dctx->partial_size = len;
 531         }
 532
 533         return 0;
 534 }
 535
 536 static u64 vhash_final(const struct vmac_tfm_ctx *tctx,
 537                        struct vmac_desc_ctx *dctx)
 538 {
 539         unsigned int partial = dctx->partial_size;
 540         u64 ch = dctx->polytmp[0];
 541         u64 cl = dctx->polytmp[1];
 542
 543         /* L1 and L2-hash the final block if needed */
 544         if (partial) {
 545                 /* Zero-pad to next 128-bit boundary */
 546                 unsigned int n = round_up(partial, 16);
 547                 u64 rh, rl;
 548
 549                 memset(&dctx->partial[partial], 0, n - partial);
 550                 nh_16(dctx->partial_words, tctx->nhkey, n / 8, rh, rl);
 551                 rh &= m62;
 552                 if (dctx->first_block_processed)
 553                         poly_step(ch, cl, tctx->polykey[0], tctx->polykey[1],
 554                                   rh, rl);
 555                 else
 556                         ADD128(ch, cl, rh, rl);
 557         }
 558
 559         /* L3-hash the 128-bit output of L2-hash */
 560         return l3hash(ch, cl, tctx->l3key[0], tctx->l3key[1], partial * 8);
 561 }
 562
 563 static int vmac_final(struct shash_desc *desc, u8 *out)
 564 {
 565         const struct vmac_tfm_ctx *tctx = crypto_shash_ctx(desc->tfm);
 566         struct vmac_desc_ctx *dctx = shash_desc_ctx(desc);
 567         int index;
 568         u64 hash, pad;
 569
 570         if (dctx->nonce_size != VMAC_NONCEBYTES)
 571                 return -EINVAL;
 572
 573         /*
 574          * The VMAC specification requires a nonce at least 1 bit shorter than
 575          * the block cipher's block length, so we actually only accept a 127-bit
 576          * nonce.  We define the unused bit to be the first one and require that
 577          * it be 0, so the needed prepending of a 0 bit is implicit.
 578          */
 579         if (dctx->nonce.bytes[0] & 0x80)
 580                 return -EINVAL;
 581
 582         /* Finish calculating the VHASH of the message */
 583         hash = vhash_final(tctx, dctx);
 584
 585         /* Generate pseudorandom pad by encrypting the nonce */
 586         BUILD_BUG_ON(VMAC_NONCEBYTES != 2 * (VMAC_TAG_LEN / 8));
 587         index = dctx->nonce.bytes[VMAC_NONCEBYTES - 1] & 1;
 588         dctx->nonce.bytes[VMAC_NONCEBYTES - 1] &= ~1;
 589         crypto_cipher_encrypt_one(tctx->cipher, dctx->nonce.bytes,
 590                                   dctx->nonce.bytes);
 591         pad = be64_to_cpu(dctx->nonce.pads[index]);
 592
 593         /* The VMAC is the sum of VHASH and the pseudorandom pad */
 594         put_unaligned_be64(hash + pad, out);
 595         return 0;
 596 }
 597
 598 static int vmac_init_tfm(struct crypto_tfm *tfm)
 599 {
 600         struct crypto_instance *inst = crypto_tfm_alg_instance(tfm);
 601         struct crypto_spawn *spawn = crypto_instance_ctx(inst);
 602         struct vmac_tfm_ctx *tctx = crypto_tfm_ctx(tfm);
 603         struct crypto_cipher *cipher;
 604
 605         cipher = crypto_spawn_cipher(spawn);
 606         if (IS_ERR(cipher))
 607                 return PTR_ERR(cipher);
 608
 609         tctx->cipher = cipher;
 610         return 0;
 611 }
 612
 613 static void vmac_exit_tfm(struct crypto_tfm *tfm)
 614 {
 615         struct vmac_tfm_ctx *tctx = crypto_tfm_ctx(tfm);
 616
 617         crypto_free_cipher(tctx->cipher);
 618 }
 619
 620 static int vmac_create(struct crypto_template *tmpl, struct rtattr **tb)
 621 {
 622         struct shash_instance *inst;
 623         struct crypto_alg *alg;
 624         int err;
 625
 626         err = crypto_check_attr_type(tb, CRYPTO_ALG_TYPE_SHASH);
 627         if (err)
 628                 return err;
 629
 630         alg = crypto_get_attr_alg(tb, CRYPTO_ALG_TYPE_CIPHER,
 631                         CRYPTO_ALG_TYPE_MASK);
 632         if (IS_ERR(alg))
 633                 return PTR_ERR(alg);
 634
 635         err = -EINVAL;
 636         if (alg->cra_blocksize != VMAC_NONCEBYTES)
 637                 goto out_put_alg;
 638
 639         inst = shash_alloc_instance(tmpl->name, alg);
 640         err = PTR_ERR(inst);
 641         if (IS_ERR(inst))
 642                 goto out_put_alg;
 643
 644         err = crypto_init_spawn(shash_instance_ctx(inst), alg,
 645                         shash_crypto_instance(inst),
 646                         CRYPTO_ALG_TYPE_MASK);
 647         if (err)
 648                 goto out_free_inst;
 649
 650         inst->alg.base.cra_priority = alg->cra_priority;
 651         inst->alg.base.cra_blocksize = alg->cra_blocksize;
 652         inst->alg.base.cra_alignmask = alg->cra_alignmask;
 653
 654         inst->alg.base.cra_ctxsize = sizeof(struct vmac_tfm_ctx);
 655         inst->alg.base.cra_init = vmac_init_tfm;
 656         inst->alg.base.cra_exit = vmac_exit_tfm;
 657
 658         inst->alg.descsize = sizeof(struct vmac_desc_ctx);
 659         inst->alg.digestsize = VMAC_TAG_LEN / 8;
 660         inst->alg.init = vmac_init;
 661         inst->alg.update = vmac_update;
 662         inst->alg.final = vmac_final;
 663         inst->alg.setkey = vmac_setkey;
 664
 665         err = shash_register_instance(tmpl, inst);
 666         if (err) {
 667 out_free_inst:
 668                 shash_free_instance(shash_crypto_instance(inst));
 669         }
 670
 671 out_put_alg:
 672         crypto_mod_put(alg);
 673         return err;
 674 }
 675
 676 static struct crypto_template vmac64_tmpl = {
 677         .name = "vmac64",
 678         .create = vmac_create,
 679         .free = shash_free_instance,
 680         .module = THIS_MODULE,
 681 };
 682
 683 static int __init vmac_module_init(void)
 684 {
 685         return crypto_register_template(&vmac64_tmpl);
 686 }
 687
 688 static void __exit vmac_module_exit(void)
 689 {
 690         crypto_unregister_template(&vmac64_tmpl);
 691 }
 692
 693 subsys_initcall(vmac_module_init);
 694 module_exit(vmac_module_exit);
 695
 696 MODULE_LICENSE("GPL");
 697 MODULE_DESCRIPTION("VMAC hash algorithm");
 698 MODULE_ALIAS_CRYPTO("vmac64");