[PATCH 5/6] crypto: LRW, Liskov Rivest Wagner, a tweakable narrow block cipher mode

@@ -156,6 +156,19 @@ config CRYPTO_CBC
 	  CBC: Cipher Block Chaining mode
 	  This block cipher algorithm is required for IPSec.
 
+config CRYPTO_LRW
+	tristate "LRW support (EXPERIMENTAL)"
+	depends on EXPERIMENTAL
+	select CRYPTO_BLKCIPHER
+	select CRYPTO_GF128MUL
+	default n
+	help
+	  LRW: Liskov Rivest Wagner, a tweakable, non malleable, non movable
+	  narrow block cipher mode. Use it with cipher specification string
+	  aes-lrw-benbi, the key must be 256, 320 or 384. The first 128, 192
+	  or 256 bits in the key are used for AES and the rest is used to tie
+	  each cipher block to its logical position.
+
 config CRYPTO_DES
 	tristate "DES and Triple DES EDE cipher algorithms"
 	select CRYPTO_ALGAPI

@@ -26,6 +26,7 @@ obj-$(CONFIG_CRYPTO_TGR192) += tgr192.o
 obj-$(CONFIG_CRYPTO_GF128MUL) += gf128mul.o
 obj-$(CONFIG_CRYPTO_ECB) += ecb.o
 obj-$(CONFIG_CRYPTO_CBC) += cbc.o
+obj-$(CONFIG_CRYPTO_LRW) += lrw.o
 obj-$(CONFIG_CRYPTO_DES) += des.o
 obj-$(CONFIG_CRYPTO_BLOWFISH) += blowfish.o
 obj-$(CONFIG_CRYPTO_TWOFISH) += twofish.o

@@ -0,0 +1,297 @@
+/* LRW: as defined by Cyril Guyot in
+ *	http://grouper.ieee.org/groups/1619/email/pdf00017.pdf
+ *
+ * Copyright (c) 2006 Rik Snel <rsnel@cube.dyndns.org>
+ *
+ * Based om ecb.c
+ * Copyright (c) 2006 Herbert Xu <herbert@gondor.apana.org.au>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation; either version 2 of the License, or (at your option)
+ * any later version.
+ */
+
+/* This implementation is checked against the test vectors in the above 
+ * document and by a test vector provided by Ken Buchanan at
+ * http://www.mail-archive.com/stds-p1619@listserv.ieee.org/msg00173.html 
+ *
+ * The vectors can be found in Documentation/crypto they are in the form
+ * of a script and can therefore easily be checked, just run lrw-32-aes
+ * with sufficient permissions after reading it and also testvector.fun */
+#include <crypto/algapi.h>
+#include <linux/err.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/scatterlist.h>
+#include <linux/slab.h>
+
+#include "b128ops.h"
+#include "gf128mul.h"
+
+struct priv {
+	struct crypto_cipher *child;
+	/* optimizes multiplying a random (non incrementing, as at the 
+	 * start of a new sector) value with key2, we could also have 
+	 * used 4k optimization tables or no optimization at all. In the 
+	 * latter case we would have to store key2 here */
+	struct gf128mul_64k table;	
+	/* stores:
+	 *  key2*{ 0,0,...0,0,0,0,1 }, key2*{ 0,0,...0,0,0,1,1 },
+	 *  key2*{ 0,0,...0,0,1,1,1 }, key2*{ 0,0,...0,1,1,1,1 }
+	 *  key2*{ 0,0,...1,1,1,1,1 }, etc 
+	 * needed for optimized multiplication of incrementing values
+	 * with key2 */
+	u64 mulinc[128][GF128MUL_BYTES >> 3]; 
+};
+
+static inline void setbit128(void *b, int bit)
+{
+	int index = 15 - bit/8;
+	((u8*)b)[index] |= 1<<(bit%8);
+}
+
+static int setkey(struct crypto_tfm *parent, const u8 *key,
+			     unsigned int keylen)
+{
+	struct priv *ctx = crypto_tfm_ctx(parent);
+	struct crypto_cipher *child = ctx->child;
+	int err, i;
+	u64 tmp[2] = { 0, }, scratch[2];
+	int bsize = crypto_cipher_blocksize(child);
+
+	crypto_cipher_clear_flags(child, CRYPTO_TFM_REQ_MASK);
+	crypto_cipher_set_flags(child, crypto_tfm_get_flags(parent) &
+				       CRYPTO_TFM_REQ_MASK);
+	if ((err = crypto_cipher_setkey(child, key, keylen - bsize))) 
+		return err;
+	crypto_tfm_set_flags(parent, crypto_cipher_get_flags(child) &
+				     CRYPTO_TFM_RES_MASK);
+
+
+	/* initialize multiplication table for Key2 */
+	gf128mul_init_64k_bbe(&ctx->table, (u64*)(key + keylen - bsize));
+
+	/* initialize optimization table */
+	for (i = 0; i < 128; i++) {
+		setbit128(tmp, i);
+		b128ops_mov(ctx->mulinc[i], tmp);
+		gf128mul_64k_bbe(ctx->mulinc[i], &ctx->table, scratch);
+	}
+
+	return 0;
+}
+
+struct sinfo {
+	u64 b1[2], b2[2];
+	struct crypto_tfm *tfm;
+	void (*fn)(struct crypto_tfm*, u8*, const u8*);
+};
+
+static inline void inc(u64 *iv)
+{
+	if (!(iv[1] = cpu_to_be64(be64_to_cpu(iv[1]) + 1)))
+		iv[0] = cpu_to_be64(be64_to_cpu(iv[0]) + 1);
+}
+
+static inline void round(struct sinfo *s, u8 *dst, const u8 *src)
+{
+	b128ops_xor(s->b2, src);	/* PP <- T xor P */
+	s->fn(s->tfm, dst, (u8*)s->b2);	/* CC <- E(Key2,PP) */
+	b128ops_xor(dst, s->b1);	/* C <- T xor CC */
+}
+
+/* this returns the number of consequative 1 bits 
+ * starting from the right in i */
+static inline int get_index8(u8 i) 
+{
+	int j = 1;
+
+	if (i&1) {
+		while ((i >>= 1)&1) j++;
+		return j;
+	}
+
+	return 0;
+}
+
+/* this returns the number of consequative 1 bits starting
+ * from the right, get_index128(00 00 00 00 00 00 ... 00 00 10 FB) = 2 */
+static inline int get_index128(u8 *block)
+{
+	int inc, ret = 0, len = 16;
+	while ((inc = get_index8(block[--len])) == 8) ret += 8;
+	return ret + inc;
+}
+
+static int crypt(struct blkcipher_desc *d,
+			    struct scatterlist *dst,
+			    struct scatterlist *src,
+			    unsigned int nbytes, struct priv *ctx,
+			    void (*fn)(struct crypto_tfm*, u8*, const u8*))
+{
+	struct blkcipher_walk w;
+	int err;
+	unsigned int avail;
+	const int bs = crypto_cipher_blocksize(ctx->child);
+	u8 *wsrc, *wdst;
+	struct sinfo s = { 
+		.tfm = crypto_cipher_tfm(ctx->child), 
+		.fn = fn
+	};
+
+	blkcipher_walk_init(&w, dst, src, nbytes);
+
+	/* start of the loop unrolled to be able to
+	 * handle the first block diffently from the others */
+	if ((err = blkcipher_walk_virt(d, &w))) return err;
+
+	wsrc = w.src.virt.addr;
+	wdst = w.dst.virt.addr;
+
+	/* calculate first value of T */
+	b128ops_mov(s.b1, w.iv);
+	gf128mul_64k_bbe(s.b1, &ctx->table, s.b2); /* T <- I*Key2 */
+
+	round(&s, wdst, wsrc);
+	
+	while ((avail = w.nbytes)) {
+		while ((avail -= bs) >= bs) {
+			wsrc += bs;
+			wdst += bs;
+
+			/* old T is available in s.b1; new one
+			 * must be made available in b1 and b2 */
+
+			/* T <- I*Key2, using the optimization
+			 * discussed in the specification */
+			b128ops_xor(s.b1, ctx->mulinc[get_index128(w.iv)]);
+			inc((u64*)w.iv);
+			b128ops_mov(s.b2, s.b1);	
+
+			round(&s, wdst, wsrc);
+		}
+
+		if ((err = blkcipher_walk_done(d, &w, avail))) return err;
+
+		wsrc = w.src.virt.addr;
+		wdst = w.dst.virt.addr;
+	}
+	return err;
+}
+
+static int encrypt(struct blkcipher_desc *desc, struct scatterlist *dst, 
+		struct scatterlist *src, unsigned int nbytes)
+{
+	struct priv *ctx = crypto_blkcipher_ctx(desc->tfm);
+	return crypt(desc, dst, src, nbytes, ctx,
+			crypto_cipher_alg(ctx->child)->cia_encrypt);
+}
+
+static int decrypt(struct blkcipher_desc *desc, struct scatterlist *dst, 
+		struct scatterlist *src, unsigned int nbytes)
+{
+	struct priv *ctx = crypto_blkcipher_ctx(desc->tfm);
+	return crypt(desc, dst, src, nbytes, ctx,
+			crypto_cipher_alg(ctx->child)->cia_decrypt);
+}
+
+static int init_tfm(struct crypto_tfm *tfm)
+{
+	struct crypto_instance *inst = (void *)tfm->__crt_alg;
+	struct crypto_spawn *spawn = crypto_instance_ctx(inst);
+	struct priv *ctx = crypto_tfm_ctx(tfm);
+	u32 *flags = &tfm->crt_flags;
+
+	tfm = crypto_spawn_tfm(spawn);
+	if (IS_ERR(tfm))
+		return PTR_ERR(tfm);
+	
+	if (crypto_tfm_alg_blocksize(tfm) != 16) {
+		*flags |= CRYPTO_TFM_RES_BAD_BLOCK_LEN;
+		return -EINVAL;
+	}
+
+	ctx->child = crypto_cipher_cast(tfm);
+	return 0;
+}
+
+static void exit_tfm(struct crypto_tfm *tfm)
+{
+	struct priv *ctx = crypto_tfm_ctx(tfm);
+	crypto_free_cipher(ctx->child);
+}
+
+static struct crypto_instance *alloc(void *param, unsigned int len)
+{
+	struct crypto_instance *inst;
+	struct crypto_alg *alg;
+
+	alg = crypto_get_attr_alg(param, len, CRYPTO_ALG_TYPE_CIPHER,
+				  CRYPTO_ALG_TYPE_MASK | CRYPTO_ALG_ASYNC);
+	if (IS_ERR(alg))
+		return ERR_PTR(PTR_ERR(alg));
+
+	inst = crypto_alloc_instance("lrw", alg);
+	if (IS_ERR(inst))
+		goto out_put_alg;
+
+	inst->alg.cra_flags = CRYPTO_ALG_TYPE_BLKCIPHER;
+	inst->alg.cra_priority = alg->cra_priority;
+	inst->alg.cra_blocksize = alg->cra_blocksize;
+
+	if (alg->cra_alignmask < 7) inst->alg.cra_alignmask = 7;
+	else inst->alg.cra_alignmask = alg->cra_alignmask;
+	inst->alg.cra_type = &crypto_blkcipher_type;
+
+	if (!(alg->cra_blocksize % 4))
+		inst->alg.cra_alignmask |= 3;
+	inst->alg.cra_blkcipher.ivsize = alg->cra_blocksize;
+	inst->alg.cra_blkcipher.min_keysize = 
+		alg->cra_cipher.cia_min_keysize + alg->cra_blocksize;
+	inst->alg.cra_blkcipher.max_keysize = 
+		alg->cra_cipher.cia_max_keysize + alg->cra_blocksize;
+
+	inst->alg.cra_ctxsize = sizeof(struct priv);
+
+	inst->alg.cra_init = init_tfm;
+	inst->alg.cra_exit = exit_tfm;
+
+	inst->alg.cra_blkcipher.setkey = setkey;
+	inst->alg.cra_blkcipher.encrypt = encrypt;
+	inst->alg.cra_blkcipher.decrypt = decrypt;
+
+out_put_alg:
+	crypto_mod_put(alg);
+	return inst;
+}
+
+static void free(struct crypto_instance *inst)
+{
+	crypto_drop_spawn(crypto_instance_ctx(inst));
+	kfree(inst);
+}
+
+static struct crypto_template crypto_tmpl = {
+	.name = "lrw",
+	.alloc = alloc,
+	.free = free,
+	.module = THIS_MODULE,
+};
+
+static int __init crypto_module_init(void)
+{
+	return crypto_register_template(&crypto_tmpl);
+}
+
+static void __exit crypto_module_exit(void)
+{
+	crypto_unregister_template(&crypto_tmpl);
+}
+
+module_init(crypto_module_init);
+module_exit(crypto_module_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("LRW block cipher mode");