• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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 
poly_step_func(u64 * ahi,u64 * alo,const u64 * kh,const u64 * kl,const u64 * mh,const u64 * ml)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 
vhash_abort(struct vmac_ctx * ctx)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 
l3hash(u64 p1,u64 p2,u64 k1,u64 k2,u64 len)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 
vhash_update(const unsigned char * m,unsigned int mbytes,struct vmac_ctx * ctx)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 
vhash(unsigned char m[],unsigned int mbytes,u64 * tagl,struct vmac_ctx * ctx)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 
vmac(unsigned char m[],unsigned int mbytes,const unsigned char n[16],u64 * tagl,struct vmac_ctx_t * ctx)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 
vmac_set_key(unsigned char user_key[],struct vmac_ctx_t * ctx)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 
vmac_setkey(struct crypto_shash * parent,const u8 * key,unsigned int keylen)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 
vmac_init(struct shash_desc * pdesc)557 static int vmac_init(struct shash_desc *pdesc)
558 {
559 	return 0;
560 }
561 
vmac_update(struct shash_desc * pdesc,const u8 * p,unsigned int len)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 
vmac_final(struct shash_desc * pdesc,u8 * out)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 
vmac_init_tfm(struct crypto_tfm * tfm)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 
vmac_exit_tfm(struct crypto_tfm * tfm)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 
vmac_create(struct crypto_template * tmpl,struct rtattr ** tb)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 
vmac_module_init(void)701 static int __init vmac_module_init(void)
702 {
703 	return crypto_register_template(&vmac_tmpl);
704 }
705 
vmac_module_exit(void)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");
717