crypto: aesni-intel - RFC4106 AES-GCM Driver Using Intel New Instructions

This patch adds an optimized RFC4106 AES-GCM implementation for 64-bit
kernels. It supports 128-bit AES key size. This leverages the crypto
AEAD interface type to facilitate a combined AES & GCM operation to
be implemented in assembly code. The assembly code leverages Intel(R)
AES New Instructions and the PCLMULQDQ instruction.

Signed-off-by: Adrian Hoban <adrian.hoban@intel.com>
Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Gabriele Paoloni <gabriele.paoloni@intel.com>
Signed-off-by: Aidan O'Mahony <aidan.o.mahony@intel.com>
Signed-off-by: Erdinc Ozturk <erdinc.ozturk@intel.com>
Signed-off-by: James Guilford <james.guilford@intel.com>
Signed-off-by: Wajdi Feghali <wajdi.k.feghali@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

1 files changed, 516 insertions, 2 deletions

diff --git a/arch/x86/crypto/aesni-intel_glue.c b/arch/x86/crypto/aesni-intel_glue.c
index 2cb3dcc4490a..02d349d64423 100644
--- a/arch/x86/crypto/aesni-intel_glue.c
+++ b/arch/x86/crypto/aesni-intel_glue.c
@@ -5,6 +5,14 @@
  * Copyright (C) 2008, Intel Corp.
  *    Author: Huang Ying <ying.huang@intel.com>
  *
+ * Added RFC4106 AES-GCM support for 128-bit keys under the AEAD
+ * interface for 64-bit kernels.
+ *    Authors: Adrian Hoban <adrian.hoban@intel.com>
+ *             Gabriele Paoloni <gabriele.paoloni@intel.com>
+ *             Tadeusz Struk (tadeusz.struk@intel.com)
+ *             Aidan O'Mahony (aidan.o.mahony@intel.com)
+ *    Copyright (c) 2010, Intel Corporation.
+ *
  * 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
@@ -21,6 +29,10 @@
 #include <crypto/ctr.h>
 #include <asm/i387.h>
 #include <asm/aes.h>
+#include <crypto/scatterwalk.h>
+#include <crypto/internal/aead.h>
+#include <linux/workqueue.h>
+#include <linux/spinlock.h>
 
 #if defined(CONFIG_CRYPTO_CTR) || defined(CONFIG_CRYPTO_CTR_MODULE)
 #define HAS_CTR
@@ -42,8 +54,31 @@ struct async_aes_ctx {
 	struct cryptd_ablkcipher *cryptd_tfm;
 };
 
-#define AESNI_ALIGN	16
+/* This data is stored at the end of the crypto_tfm struct.
+ * It's a type of per "session" data storage location.
+ * This needs to be 16 byte aligned.
+ */
+struct aesni_rfc4106_gcm_ctx {
+	u8 hash_subkey[16];
+	struct crypto_aes_ctx aes_key_expanded;
+	u8 nonce[4];
+	struct cryptd_aead *cryptd_tfm;
+};
+
+struct aesni_gcm_set_hash_subkey_result {
+	int err;
+	struct completion completion;
+};
+
+struct aesni_hash_subkey_req_data {
+	u8 iv[16];
+	struct aesni_gcm_set_hash_subkey_result result;
+	struct scatterlist sg;
+};
+
+#define AESNI_ALIGN	(16)
 #define AES_BLOCK_MASK	(~(AES_BLOCK_SIZE-1))
+#define RFC4106_HASH_SUBKEY_SIZE 16
 
 asmlinkage int aesni_set_key(struct crypto_aes_ctx *ctx, const u8 *in_key,
 			     unsigned int key_len);
@@ -62,6 +97,57 @@ asmlinkage void aesni_cbc_dec(struct crypto_aes_ctx *ctx, u8 *out,
 asmlinkage void aesni_ctr_enc(struct crypto_aes_ctx *ctx, u8 *out,
 			      const u8 *in, unsigned int len, u8 *iv);
 
+/* asmlinkage void aesni_gcm_enc()
+ * void *ctx,  AES Key schedule. Starts on a 16 byte boundary.
+ * u8 *out, Ciphertext output. Encrypt in-place is allowed.
+ * const u8 *in, Plaintext input
+ * unsigned long plaintext_len, Length of data in bytes for encryption.
+ * u8 *iv, Pre-counter block j0: 4 byte salt (from Security Association)
+ *         concatenated with 8 byte Initialisation Vector (from IPSec ESP
+ *         Payload) concatenated with 0x00000001. 16-byte aligned pointer.
+ * u8 *hash_subkey, the Hash sub key input. Data starts on a 16-byte boundary.
+ * const u8 *aad, Additional Authentication Data (AAD)
+ * unsigned long aad_len, Length of AAD in bytes. With RFC4106 this
+ *          is going to be 8 or 12 bytes
+ * u8 *auth_tag, Authenticated Tag output.
+ * unsigned long auth_tag_len), Authenticated Tag Length in bytes.
+ *          Valid values are 16 (most likely), 12 or 8.
+ */
+asmlinkage void aesni_gcm_enc(void *ctx, u8 *out,
+			const u8 *in, unsigned long plaintext_len, u8 *iv,
+			u8 *hash_subkey, const u8 *aad, unsigned long aad_len,
+			u8 *auth_tag, unsigned long auth_tag_len);
+
+/* asmlinkage void aesni_gcm_dec()
+ * void *ctx, AES Key schedule. Starts on a 16 byte boundary.
+ * u8 *out, Plaintext output. Decrypt in-place is allowed.
+ * const u8 *in, Ciphertext input
+ * unsigned long ciphertext_len, Length of data in bytes for decryption.
+ * u8 *iv, Pre-counter block j0: 4 byte salt (from Security Association)
+ *         concatenated with 8 byte Initialisation Vector (from IPSec ESP
+ *         Payload) concatenated with 0x00000001. 16-byte aligned pointer.
+ * u8 *hash_subkey, the Hash sub key input. Data starts on a 16-byte boundary.
+ * const u8 *aad, Additional Authentication Data (AAD)
+ * unsigned long aad_len, Length of AAD in bytes. With RFC4106 this is going
+ * to be 8 or 12 bytes
+ * u8 *auth_tag, Authenticated Tag output.
+ * unsigned long auth_tag_len) Authenticated Tag Length in bytes.
+ * Valid values are 16 (most likely), 12 or 8.
+ */
+asmlinkage void aesni_gcm_dec(void *ctx, u8 *out,
+			const u8 *in, unsigned long ciphertext_len, u8 *iv,
+			u8 *hash_subkey, const u8 *aad, unsigned long aad_len,
+			u8 *auth_tag, unsigned long auth_tag_len);
+
+static inline struct
+aesni_rfc4106_gcm_ctx *aesni_rfc4106_gcm_ctx_get(struct crypto_aead *tfm)
+{
+	return
+		(struct aesni_rfc4106_gcm_ctx *)
+		PTR_ALIGN((u8 *)
+		crypto_tfm_ctx(crypto_aead_tfm(tfm)), AESNI_ALIGN);
+}
+
 static inline struct crypto_aes_ctx *aes_ctx(void *raw_ctx)
 {
 	unsigned long addr = (unsigned long)raw_ctx;
@@ -730,6 +816,422 @@ static struct crypto_alg ablk_xts_alg = {
 };
 #endif
 
+static int rfc4106_init(struct crypto_tfm *tfm)
+{
+	struct cryptd_aead *cryptd_tfm;
+	struct aesni_rfc4106_gcm_ctx *ctx = (struct aesni_rfc4106_gcm_ctx *)
+		PTR_ALIGN((u8 *)crypto_tfm_ctx(tfm), AESNI_ALIGN);
+	cryptd_tfm = cryptd_alloc_aead("__driver-gcm-aes-aesni", 0, 0);
+	if (IS_ERR(cryptd_tfm))
+		return PTR_ERR(cryptd_tfm);
+	ctx->cryptd_tfm = cryptd_tfm;
+	tfm->crt_aead.reqsize = sizeof(struct aead_request)
+		+ crypto_aead_reqsize(&cryptd_tfm->base);
+	return 0;
+}
+
+static void rfc4106_exit(struct crypto_tfm *tfm)
+{
+	struct aesni_rfc4106_gcm_ctx *ctx =
+		(struct aesni_rfc4106_gcm_ctx *)
+		PTR_ALIGN((u8 *)crypto_tfm_ctx(tfm), AESNI_ALIGN);
+	if (!IS_ERR(ctx->cryptd_tfm))
+		cryptd_free_aead(ctx->cryptd_tfm);
+	return;
+}
+
+static void
+rfc4106_set_hash_subkey_done(struct crypto_async_request *req, int err)
+{
+	struct aesni_gcm_set_hash_subkey_result *result = req->data;
+
+	if (err == -EINPROGRESS)
+		return;
+	result->err = err;
+	complete(&result->completion);
+}
+
+static int
+rfc4106_set_hash_subkey(u8 *hash_subkey, const u8 *key, unsigned int key_len)
+{
+	struct crypto_ablkcipher *ctr_tfm;
+	struct ablkcipher_request *req;
+	int ret = -EINVAL;
+	struct aesni_hash_subkey_req_data *req_data;
+
+	ctr_tfm = crypto_alloc_ablkcipher("ctr(aes)", 0, 0);
+	if (IS_ERR(ctr_tfm))
+		return PTR_ERR(ctr_tfm);
+
+	crypto_ablkcipher_clear_flags(ctr_tfm, ~0);
+
+	ret = crypto_ablkcipher_setkey(ctr_tfm, key, key_len);
+	if (ret) {
+		crypto_free_ablkcipher(ctr_tfm);
+		return ret;
+	}
+
+	req = ablkcipher_request_alloc(ctr_tfm, GFP_KERNEL);
+	if (!req) {
+		crypto_free_ablkcipher(ctr_tfm);
+		return -EINVAL;
+	}
+
+	req_data = kmalloc(sizeof(*req_data), GFP_KERNEL);
+	if (!req_data) {
+		crypto_free_ablkcipher(ctr_tfm);
+		return -ENOMEM;
+	}
+	memset(req_data->iv, 0, sizeof(req_data->iv));
+
+	/* Clear the data in the hash sub key container to zero.*/
+	/* We want to cipher all zeros to create the hash sub key. */
+	memset(hash_subkey, 0, RFC4106_HASH_SUBKEY_SIZE);
+
+	init_completion(&req_data->result.completion);
+	sg_init_one(&req_data->sg, hash_subkey, RFC4106_HASH_SUBKEY_SIZE);
+	ablkcipher_request_set_tfm(req, ctr_tfm);
+	ablkcipher_request_set_callback(req, CRYPTO_TFM_REQ_MAY_SLEEP |
+					CRYPTO_TFM_REQ_MAY_BACKLOG,
+					rfc4106_set_hash_subkey_done,
+					&req_data->result);
+
+	ablkcipher_request_set_crypt(req, &req_data->sg,
+		&req_data->sg, RFC4106_HASH_SUBKEY_SIZE, req_data->iv);
+
+	ret = crypto_ablkcipher_encrypt(req);
+	if (ret == -EINPROGRESS || ret == -EBUSY) {
+		ret = wait_for_completion_interruptible
+			(&req_data->result.completion);
+		if (!ret)
+			ret = req_data->result.err;
+	}
+	ablkcipher_request_free(req);
+	kfree(req_data);
+	crypto_free_ablkcipher(ctr_tfm);
+	return ret;
+}
+
+static int rfc4106_set_key(struct crypto_aead *parent, const u8 *key,
+						   unsigned int key_len)
+{
+	int ret = 0;
+	struct crypto_tfm *tfm = crypto_aead_tfm(parent);
+	struct aesni_rfc4106_gcm_ctx *ctx = aesni_rfc4106_gcm_ctx_get(parent);
+	u8 *new_key_mem = NULL;
+
+	if (key_len < 4) {
+		crypto_tfm_set_flags(tfm, CRYPTO_TFM_RES_BAD_KEY_LEN);
+		return -EINVAL;
+	}
+	/*Account for 4 byte nonce at the end.*/
+	key_len -= 4;
+	if (key_len != AES_KEYSIZE_128) {
+		crypto_tfm_set_flags(tfm, CRYPTO_TFM_RES_BAD_KEY_LEN);
+		return -EINVAL;
+	}
+
+	memcpy(ctx->nonce, key + key_len, sizeof(ctx->nonce));
+	/*This must be on a 16 byte boundary!*/
+	if ((unsigned long)(&(ctx->aes_key_expanded.key_enc[0])) % AESNI_ALIGN)
+		return -EINVAL;
+
+	if ((unsigned long)key % AESNI_ALIGN) {
+		/*key is not aligned: use an auxuliar aligned pointer*/
+		new_key_mem = kmalloc(key_len+AESNI_ALIGN, GFP_KERNEL);
+		if (!new_key_mem)
+			return -ENOMEM;
+
+		new_key_mem = PTR_ALIGN(new_key_mem, AESNI_ALIGN);
+		memcpy(new_key_mem, key, key_len);
+		key = new_key_mem;
+	}
+
+	if (!irq_fpu_usable())
+		ret = crypto_aes_expand_key(&(ctx->aes_key_expanded),
+		key, key_len);
+	else {
+		kernel_fpu_begin();
+		ret = aesni_set_key(&(ctx->aes_key_expanded), key, key_len);
+		kernel_fpu_end();
+	}
+	/*This must be on a 16 byte boundary!*/
+	if ((unsigned long)(&(ctx->hash_subkey[0])) % AESNI_ALIGN) {
+		ret = -EINVAL;
+		goto exit;
+	}
+	ret = rfc4106_set_hash_subkey(ctx->hash_subkey, key, key_len);
+exit:
+	kfree(new_key_mem);
+	return ret;
+}
+
+/* This is the Integrity Check Value (aka the authentication tag length and can
+ * be 8, 12 or 16 bytes long. */
+static int rfc4106_set_authsize(struct crypto_aead *parent,
+				unsigned int authsize)
+{
+	struct aesni_rfc4106_gcm_ctx *ctx = aesni_rfc4106_gcm_ctx_get(parent);
+	struct crypto_aead *cryptd_child = cryptd_aead_child(ctx->cryptd_tfm);
+
+	switch (authsize) {
+	case 8:
+	case 12:
+	case 16:
+		break;
+	default:
+		return -EINVAL;
+	}
+	crypto_aead_crt(parent)->authsize = authsize;
+	crypto_aead_crt(cryptd_child)->authsize = authsize;
+	return 0;
+}
+
+static int rfc4106_encrypt(struct aead_request *req)
+{
+	int ret;
+	struct crypto_aead *tfm = crypto_aead_reqtfm(req);
+	struct aesni_rfc4106_gcm_ctx *ctx = aesni_rfc4106_gcm_ctx_get(tfm);
+	struct crypto_aead *cryptd_child = cryptd_aead_child(ctx->cryptd_tfm);
+
+	if (!irq_fpu_usable()) {
+		struct aead_request *cryptd_req =
+			(struct aead_request *) aead_request_ctx(req);
+		memcpy(cryptd_req, req, sizeof(*req));
+		aead_request_set_tfm(cryptd_req, &ctx->cryptd_tfm->base);
+		return crypto_aead_encrypt(cryptd_req);
+	} else {
+		kernel_fpu_begin();
+		ret = cryptd_child->base.crt_aead.encrypt(req);
+		kernel_fpu_end();
+		return ret;
+	}
+}
+
+static int rfc4106_decrypt(struct aead_request *req)
+{
+	int ret;
+	struct crypto_aead *tfm = crypto_aead_reqtfm(req);
+	struct aesni_rfc4106_gcm_ctx *ctx = aesni_rfc4106_gcm_ctx_get(tfm);
+	struct crypto_aead *cryptd_child = cryptd_aead_child(ctx->cryptd_tfm);
+
+	if (!irq_fpu_usable()) {
+		struct aead_request *cryptd_req =
+			(struct aead_request *) aead_request_ctx(req);
+		memcpy(cryptd_req, req, sizeof(*req));
+		aead_request_set_tfm(cryptd_req, &ctx->cryptd_tfm->base);
+		return crypto_aead_decrypt(cryptd_req);
+	} else {
+		kernel_fpu_begin();
+		ret = cryptd_child->base.crt_aead.decrypt(req);
+		kernel_fpu_end();
+		return ret;
+	}
+}
+
+static struct crypto_alg rfc4106_alg = {
+	.cra_name = "rfc4106(gcm(aes))",
+	.cra_driver_name = "rfc4106-gcm-aesni",
+	.cra_priority = 400,
+	.cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC,
+	.cra_blocksize = 1,
+	.cra_ctxsize = sizeof(struct aesni_rfc4106_gcm_ctx) + AESNI_ALIGN,
+	.cra_alignmask = 0,
+	.cra_type = &crypto_nivaead_type,
+	.cra_module = THIS_MODULE,
+	.cra_list = LIST_HEAD_INIT(rfc4106_alg.cra_list),
+	.cra_init = rfc4106_init,
+	.cra_exit = rfc4106_exit,
+	.cra_u = {
+		.aead = {
+			.setkey = rfc4106_set_key,
+			.setauthsize = rfc4106_set_authsize,
+			.encrypt = rfc4106_encrypt,
+			.decrypt = rfc4106_decrypt,
+			.geniv = "seqiv",
+			.ivsize = 8,
+			.maxauthsize = 16,
+		},
+	},
+};
+
+static int __driver_rfc4106_encrypt(struct aead_request *req)
+{
+	u8 one_entry_in_sg = 0;
+	u8 *src, *dst, *assoc;
+	__be32 counter = cpu_to_be32(1);
+	struct crypto_aead *tfm = crypto_aead_reqtfm(req);
+	struct aesni_rfc4106_gcm_ctx *ctx = aesni_rfc4106_gcm_ctx_get(tfm);
+	void *aes_ctx = &(ctx->aes_key_expanded);
+	unsigned long auth_tag_len = crypto_aead_authsize(tfm);
+	u8 iv_tab[16+AESNI_ALIGN];
+	u8* iv = (u8 *) PTR_ALIGN((u8 *)iv_tab, AESNI_ALIGN);
+	struct scatter_walk src_sg_walk;
+	struct scatter_walk assoc_sg_walk;
+	struct scatter_walk dst_sg_walk;
+	unsigned int i;
+
+	/* Assuming we are supporting rfc4106 64-bit extended */
+	/* sequence numbers We need to have the AAD length equal */
+	/* to 8 or 12 bytes */
+	if (unlikely(req->assoclen != 8 && req->assoclen != 12))
+		return -EINVAL;
+	/* IV below built */
+	for (i = 0; i < 4; i++)
+		*(iv+i) = ctx->nonce[i];
+	for (i = 0; i < 8; i++)
+		*(iv+4+i) = req->iv[i];
+	*((__be32 *)(iv+12)) = counter;
+
+	if ((sg_is_last(req->src)) && (sg_is_last(req->assoc))) {
+		one_entry_in_sg = 1;
+		scatterwalk_start(&src_sg_walk, req->src);
+		scatterwalk_start(&assoc_sg_walk, req->assoc);
+		src = scatterwalk_map(&src_sg_walk, 0);
+		assoc = scatterwalk_map(&assoc_sg_walk, 0);
+		dst = src;
+		if (unlikely(req->src != req->dst)) {
+			scatterwalk_start(&dst_sg_walk, req->dst);
+			dst = scatterwalk_map(&dst_sg_walk, 0);
+		}
+
+	} else {
+		/* Allocate memory for src, dst, assoc */
+		src = kmalloc(req->cryptlen + auth_tag_len + req->assoclen,
+			GFP_ATOMIC);
+		if (unlikely(!src))
+			return -ENOMEM;
+		assoc = (src + req->cryptlen + auth_tag_len);
+		scatterwalk_map_and_copy(src, req->src, 0, req->cryptlen, 0);
+		scatterwalk_map_and_copy(assoc, req->assoc, 0,
+					req->assoclen, 0);
+		dst = src;
+	}
+
+	aesni_gcm_enc(aes_ctx, dst, src, (unsigned long)req->cryptlen, iv,
+		ctx->hash_subkey, assoc, (unsigned long)req->assoclen, dst
+		+ ((unsigned long)req->cryptlen), auth_tag_len);
+
+	/* The authTag (aka the Integrity Check Value) needs to be written
+	 * back to the packet. */
+	if (one_entry_in_sg) {
+		if (unlikely(req->src != req->dst)) {
+			scatterwalk_unmap(dst, 0);
+			scatterwalk_done(&dst_sg_walk, 0, 0);
+		}
+		scatterwalk_unmap(src, 0);
+		scatterwalk_unmap(assoc, 0);
+		scatterwalk_done(&src_sg_walk, 0, 0);
+		scatterwalk_done(&assoc_sg_walk, 0, 0);
+	} else {
+		scatterwalk_map_and_copy(dst, req->dst, 0,
+			req->cryptlen + auth_tag_len, 1);
+		kfree(src);
+	}
+	return 0;
+}
+
+static int __driver_rfc4106_decrypt(struct aead_request *req)
+{
+	u8 one_entry_in_sg = 0;
+	u8 *src, *dst, *assoc;
+	unsigned long tempCipherLen = 0;
+	__be32 counter = cpu_to_be32(1);
+	int retval = 0;
+	struct crypto_aead *tfm = crypto_aead_reqtfm(req);
+	struct aesni_rfc4106_gcm_ctx *ctx = aesni_rfc4106_gcm_ctx_get(tfm);
+	void *aes_ctx = &(ctx->aes_key_expanded);
+	unsigned long auth_tag_len = crypto_aead_authsize(tfm);
+	u8 iv_and_authTag[32+AESNI_ALIGN];
+	u8 *iv = (u8 *) PTR_ALIGN((u8 *)iv_and_authTag, AESNI_ALIGN);
+	u8 *authTag = iv + 16;
+	struct scatter_walk src_sg_walk;
+	struct scatter_walk assoc_sg_walk;
+	struct scatter_walk dst_sg_walk;
+	unsigned int i;
+
+	if (unlikely((req->cryptlen < auth_tag_len) ||
+		(req->assoclen != 8 && req->assoclen != 12)))
+		return -EINVAL;
+	/* Assuming we are supporting rfc4106 64-bit extended */
+	/* sequence numbers We need to have the AAD length */
+	/* equal to 8 or 12 bytes */
+
+	tempCipherLen = (unsigned long)(req->cryptlen - auth_tag_len);
+	/* IV below built */
+	for (i = 0; i < 4; i++)
+		*(iv+i) = ctx->nonce[i];
+	for (i = 0; i < 8; i++)
+		*(iv+4+i) = req->iv[i];
+	*((__be32 *)(iv+12)) = counter;
+
+	if ((sg_is_last(req->src)) && (sg_is_last(req->assoc))) {
+		one_entry_in_sg = 1;
+		scatterwalk_start(&src_sg_walk, req->src);
+		scatterwalk_start(&assoc_sg_walk, req->assoc);
+		src = scatterwalk_map(&src_sg_walk, 0);
+		assoc = scatterwalk_map(&assoc_sg_walk, 0);
+		dst = src;
+		if (unlikely(req->src != req->dst)) {
+			scatterwalk_start(&dst_sg_walk, req->dst);
+			dst = scatterwalk_map(&dst_sg_walk, 0);
+		}
+
+	} else {
+		/* Allocate memory for src, dst, assoc */
+		src = kmalloc(req->cryptlen + req->assoclen, GFP_ATOMIC);
+		if (!src)
+			return -ENOMEM;
+		assoc = (src + req->cryptlen + auth_tag_len);
+		scatterwalk_map_and_copy(src, req->src, 0, req->cryptlen, 0);
+		scatterwalk_map_and_copy(assoc, req->assoc, 0,
+			req->assoclen, 0);
+		dst = src;
+	}
+
+	aesni_gcm_dec(aes_ctx, dst, src, tempCipherLen, iv,
+		ctx->hash_subkey, assoc, (unsigned long)req->assoclen,
+		authTag, auth_tag_len);
+
+	/* Compare generated tag with passed in tag. */
+	retval = memcmp(src + tempCipherLen, authTag, auth_tag_len) ?
+		-EBADMSG : 0;
+
+	if (one_entry_in_sg) {
+		if (unlikely(req->src != req->dst)) {
+			scatterwalk_unmap(dst, 0);
+			scatterwalk_done(&dst_sg_walk, 0, 0);
+		}
+		scatterwalk_unmap(src, 0);
+		scatterwalk_unmap(assoc, 0);
+		scatterwalk_done(&src_sg_walk, 0, 0);
+		scatterwalk_done(&assoc_sg_walk, 0, 0);
+	} else {
+		scatterwalk_map_and_copy(dst, req->dst, 0, req->cryptlen, 1);
+		kfree(src);
+	}
+	return retval;
+}
+
+static struct crypto_alg __rfc4106_alg = {
+	.cra_name		= "__gcm-aes-aesni",
+	.cra_driver_name	= "__driver-gcm-aes-aesni",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_AEAD,
+	.cra_blocksize		= 1,
+	.cra_ctxsize	= sizeof(struct aesni_rfc4106_gcm_ctx) + AESNI_ALIGN,
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_aead_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(__rfc4106_alg.cra_list),
+	.cra_u = {
+		.aead = {
+			.encrypt	= __driver_rfc4106_encrypt,
+			.decrypt	= __driver_rfc4106_decrypt,
+		},
+	},
+};
+
 static int __init aesni_init(void)
 {
 	int err;
@@ -738,6 +1240,7 @@ static int __init aesni_init(void)
 		printk(KERN_INFO "Intel AES-NI instructions are not detected.\n");
 		return -ENODEV;
 	}
+
 	if ((err = crypto_register_alg(&aesni_alg)))
 		goto aes_err;
 	if ((err = crypto_register_alg(&__aesni_alg)))
@@ -770,10 +1273,19 @@ static int __init aesni_init(void)
 	if ((err = crypto_register_alg(&ablk_xts_alg)))
 		goto ablk_xts_err;
 #endif
-
+	err = crypto_register_alg(&__rfc4106_alg);
+	if (err)
+		goto __aead_gcm_err;
+	err = crypto_register_alg(&rfc4106_alg);
+	if (err)
+		goto aead_gcm_err;
 	return err;
 
+aead_gcm_err:
+	crypto_unregister_alg(&__rfc4106_alg);
+__aead_gcm_err:
 #ifdef HAS_XTS
+	crypto_unregister_alg(&ablk_xts_alg);
 ablk_xts_err:
 #endif
 #ifdef HAS_PCBC
@@ -809,6 +1321,8 @@ aes_err:
 
 static void __exit aesni_exit(void)
 {
+	crypto_unregister_alg(&__rfc4106_alg);
+	crypto_unregister_alg(&rfc4106_alg);
 #ifdef HAS_XTS
 	crypto_unregister_alg(&ablk_xts_alg);
 #endif