Merge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6

Pull crypto update from Herbert Xu:

 - XTS mode optimisation for twofish/cast6/camellia/aes on x86

 - AVX2/x86_64 implementation for blowfish/twofish/serpent/camellia

 - SSSE3/AVX/AVX2 optimisations for sha256/sha512

 - Added driver for SAHARA2 crypto accelerator

 - Fix for GMAC when used in non-IPsec secnarios

 - Added generic CMAC implementation (including IPsec glue)

 - IP update for crypto/atmel

 - Support for more than one device in hwrng/timeriomem

 - Added Broadcom BCM2835 RNG driver

 - Misc fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (59 commits)
  crypto: caam - fix job ring cleanup code
  crypto: camellia - add AVX2/AES-NI/x86_64 assembler implementation of camellia cipher
  crypto: serpent - add AVX2/x86_64 assembler implementation of serpent cipher
  crypto: twofish - add AVX2/x86_64 assembler implementation of twofish cipher
  crypto: blowfish - add AVX2/x86_64 implementation of blowfish cipher
  crypto: tcrypt - add async cipher speed tests for blowfish
  crypto: testmgr - extend camellia test-vectors for camellia-aesni/avx2
  crypto: aesni_intel - fix Kconfig problem with CRYPTO_GLUE_HELPER_X86
  crypto: aesni_intel - add more optimized XTS mode for x86-64
  crypto: x86/camellia-aesni-avx - add more optimized XTS code
  crypto: cast6-avx: use new optimized XTS code
  crypto: x86/twofish-avx - use optimized XTS code
  crypto: x86 - add more optimized XTS-mode for serpent-avx
  xfrm: add rfc4494 AES-CMAC-96 support
  crypto: add CMAC support to CryptoAPI
  crypto: testmgr - add empty test vectors for null ciphers
  crypto: testmgr - add AES GMAC test vectors
  crypto: gcm - fix rfc4543 to handle async crypto correctly
  crypto: gcm - make GMAC work when dst and src are different
  hwrng: timeriomem - added devicetree hooks
  ...

40 files changed, 10762 insertions, 240 deletions

diff --git a/arch/arm/mach-at91/at91sam9g45_devices.c b/arch/arm/mach-at91/at91sam9g45_devices.c
index 827c9f2a70fb..f0bf68268ca2 100644
--- a/arch/arm/mach-at91/at91sam9g45_devices.c
+++ b/arch/arm/mach-at91/at91sam9g45_devices.c
@@ -18,7 +18,7 @@
 #include <linux/platform_device.h>
 #include <linux/i2c-gpio.h>
 #include <linux/atmel-mci.h>
-#include <linux/platform_data/atmel-aes.h>
+#include <linux/platform_data/crypto-atmel.h>
 
 #include <linux/platform_data/at91_adc.h>
 
@@ -1900,7 +1900,8 @@ static void __init at91_add_device_tdes(void) {}
  * -------------------------------------------------------------------- */
 
 #if defined(CONFIG_CRYPTO_DEV_ATMEL_AES) || defined(CONFIG_CRYPTO_DEV_ATMEL_AES_MODULE)
-static struct aes_platform_data aes_data;
+static struct crypto_platform_data aes_data;
+static struct crypto_dma_data alt_atslave;
 static u64 aes_dmamask = DMA_BIT_MASK(32);
 
 static struct resource aes_resources[] = {
@@ -1931,23 +1932,20 @@ static struct platform_device at91sam9g45_aes_device = {
 static void __init at91_add_device_aes(void)
 {
 	struct at_dma_slave	*atslave;
-	struct aes_dma_data	*alt_atslave;
-
-	alt_atslave = kzalloc(sizeof(struct aes_dma_data), GFP_KERNEL);
 
 	/* DMA TX slave channel configuration */
-	atslave = &alt_atslave->txdata;
+	atslave = &alt_atslave.txdata;
 	atslave->dma_dev = &at_hdmac_device.dev;
 	atslave->cfg = ATC_FIFOCFG_ENOUGHSPACE	| ATC_SRC_H2SEL_HW |
 						ATC_SRC_PER(AT_DMA_ID_AES_RX);
 
 	/* DMA RX slave channel configuration */
-	atslave = &alt_atslave->rxdata;
+	atslave = &alt_atslave.rxdata;
 	atslave->dma_dev = &at_hdmac_device.dev;
 	atslave->cfg = ATC_FIFOCFG_ENOUGHSPACE	| ATC_DST_H2SEL_HW |
 						ATC_DST_PER(AT_DMA_ID_AES_TX);
 
-	aes_data.dma_slave = alt_atslave;
+	aes_data.dma_slave = &alt_atslave;
 	platform_device_register(&at91sam9g45_aes_device);
 }
 #else
diff --git a/arch/x86/crypto/Makefile b/arch/x86/crypto/Makefile
index 63947a8f9f0f..a3a0ed80f17c 100644
--- a/arch/x86/crypto/Makefile
+++ b/arch/x86/crypto/Makefile
@@ -2,6 +2,10 @@
 # Arch-specific CryptoAPI modules.
 #
 
+avx_supported := $(call as-instr,vpxor %xmm0$(comma)%xmm0$(comma)%xmm0,yes,no)
+avx2_supported := $(call as-instr,vpgatherdd %ymm0$(comma)(%eax$(comma)%ymm1\
+					$(comma)4)$(comma)%ymm2,yes,no)
+
 obj-$(CONFIG_CRYPTO_ABLK_HELPER_X86) += ablk_helper.o
 obj-$(CONFIG_CRYPTO_GLUE_HELPER_X86) += glue_helper.o
 
@@ -12,22 +16,37 @@ obj-$(CONFIG_CRYPTO_SERPENT_SSE2_586) += serpent-sse2-i586.o
 
 obj-$(CONFIG_CRYPTO_AES_X86_64) += aes-x86_64.o
 obj-$(CONFIG_CRYPTO_CAMELLIA_X86_64) += camellia-x86_64.o
-obj-$(CONFIG_CRYPTO_CAMELLIA_AESNI_AVX_X86_64) += camellia-aesni-avx-x86_64.o
-obj-$(CONFIG_CRYPTO_CAST5_AVX_X86_64) += cast5-avx-x86_64.o
-obj-$(CONFIG_CRYPTO_CAST6_AVX_X86_64) += cast6-avx-x86_64.o
 obj-$(CONFIG_CRYPTO_BLOWFISH_X86_64) += blowfish-x86_64.o
 obj-$(CONFIG_CRYPTO_TWOFISH_X86_64) += twofish-x86_64.o
 obj-$(CONFIG_CRYPTO_TWOFISH_X86_64_3WAY) += twofish-x86_64-3way.o
-obj-$(CONFIG_CRYPTO_TWOFISH_AVX_X86_64) += twofish-avx-x86_64.o
 obj-$(CONFIG_CRYPTO_SALSA20_X86_64) += salsa20-x86_64.o
 obj-$(CONFIG_CRYPTO_SERPENT_SSE2_X86_64) += serpent-sse2-x86_64.o
-obj-$(CONFIG_CRYPTO_SERPENT_AVX_X86_64) += serpent-avx-x86_64.o
 obj-$(CONFIG_CRYPTO_AES_NI_INTEL) += aesni-intel.o
 obj-$(CONFIG_CRYPTO_GHASH_CLMUL_NI_INTEL) += ghash-clmulni-intel.o
 
 obj-$(CONFIG_CRYPTO_CRC32C_INTEL) += crc32c-intel.o
 obj-$(CONFIG_CRYPTO_SHA1_SSSE3) += sha1-ssse3.o
 obj-$(CONFIG_CRYPTO_CRC32_PCLMUL) += crc32-pclmul.o
+obj-$(CONFIG_CRYPTO_SHA256_SSSE3) += sha256-ssse3.o
+obj-$(CONFIG_CRYPTO_SHA512_SSSE3) += sha512-ssse3.o
+
+# These modules require assembler to support AVX.
+ifeq ($(avx_supported),yes)
+	obj-$(CONFIG_CRYPTO_CAMELLIA_AESNI_AVX_X86_64) += \
+						camellia-aesni-avx-x86_64.o
+	obj-$(CONFIG_CRYPTO_CAST5_AVX_X86_64) += cast5-avx-x86_64.o
+	obj-$(CONFIG_CRYPTO_CAST6_AVX_X86_64) += cast6-avx-x86_64.o
+	obj-$(CONFIG_CRYPTO_TWOFISH_AVX_X86_64) += twofish-avx-x86_64.o
+	obj-$(CONFIG_CRYPTO_SERPENT_AVX_X86_64) += serpent-avx-x86_64.o
+endif
+
+# These modules require assembler to support AVX2.
+ifeq ($(avx2_supported),yes)
+	obj-$(CONFIG_CRYPTO_BLOWFISH_AVX2_X86_64) += blowfish-avx2.o
+	obj-$(CONFIG_CRYPTO_CAMELLIA_AESNI_AVX2_X86_64) += camellia-aesni-avx2.o
+	obj-$(CONFIG_CRYPTO_SERPENT_AVX2_X86_64) += serpent-avx2.o
+	obj-$(CONFIG_CRYPTO_TWOFISH_AVX2_X86_64) += twofish-avx2.o
+endif
 
 aes-i586-y := aes-i586-asm_32.o aes_glue.o
 twofish-i586-y := twofish-i586-asm_32.o twofish_glue.o
@@ -36,21 +55,35 @@ serpent-sse2-i586-y := serpent-sse2-i586-asm_32.o serpent_sse2_glue.o
 
 aes-x86_64-y := aes-x86_64-asm_64.o aes_glue.o
 camellia-x86_64-y := camellia-x86_64-asm_64.o camellia_glue.o
-camellia-aesni-avx-x86_64-y := camellia-aesni-avx-asm_64.o \
-			       camellia_aesni_avx_glue.o
-cast5-avx-x86_64-y := cast5-avx-x86_64-asm_64.o cast5_avx_glue.o
-cast6-avx-x86_64-y := cast6-avx-x86_64-asm_64.o cast6_avx_glue.o
 blowfish-x86_64-y := blowfish-x86_64-asm_64.o blowfish_glue.o
 twofish-x86_64-y := twofish-x86_64-asm_64.o twofish_glue.o
 twofish-x86_64-3way-y := twofish-x86_64-asm_64-3way.o twofish_glue_3way.o
-twofish-avx-x86_64-y := twofish-avx-x86_64-asm_64.o twofish_avx_glue.o
 salsa20-x86_64-y := salsa20-x86_64-asm_64.o salsa20_glue.o
 serpent-sse2-x86_64-y := serpent-sse2-x86_64-asm_64.o serpent_sse2_glue.o
-serpent-avx-x86_64-y := serpent-avx-x86_64-asm_64.o serpent_avx_glue.o
+
+ifeq ($(avx_supported),yes)
+	camellia-aesni-avx-x86_64-y := camellia-aesni-avx-asm_64.o \
+					camellia_aesni_avx_glue.o
+	cast5-avx-x86_64-y := cast5-avx-x86_64-asm_64.o cast5_avx_glue.o
+	cast6-avx-x86_64-y := cast6-avx-x86_64-asm_64.o cast6_avx_glue.o
+	twofish-avx-x86_64-y := twofish-avx-x86_64-asm_64.o \
+				twofish_avx_glue.o
+	serpent-avx-x86_64-y := serpent-avx-x86_64-asm_64.o \
+				serpent_avx_glue.o
+endif
+
+ifeq ($(avx2_supported),yes)
+	blowfish-avx2-y := blowfish-avx2-asm_64.o blowfish_avx2_glue.o
+	camellia-aesni-avx2-y := camellia-aesni-avx2-asm_64.o camellia_aesni_avx2_glue.o
+	serpent-avx2-y := serpent-avx2-asm_64.o serpent_avx2_glue.o
+	twofish-avx2-y := twofish-avx2-asm_64.o twofish_avx2_glue.o
+endif
 
 aesni-intel-y := aesni-intel_asm.o aesni-intel_glue.o fpu.o
 ghash-clmulni-intel-y := ghash-clmulni-intel_asm.o ghash-clmulni-intel_glue.o
 sha1-ssse3-y := sha1_ssse3_asm.o sha1_ssse3_glue.o
 crc32c-intel-y := crc32c-intel_glue.o
-crc32c-intel-$(CONFIG_CRYPTO_CRC32C_X86_64) += crc32c-pcl-intel-asm_64.o
+crc32c-intel-$(CONFIG_64BIT) += crc32c-pcl-intel-asm_64.o
 crc32-pclmul-y := crc32-pclmul_asm.o crc32-pclmul_glue.o
+sha256-ssse3-y := sha256-ssse3-asm.o sha256-avx-asm.o sha256-avx2-asm.o sha256_ssse3_glue.o
+sha512-ssse3-y := sha512-ssse3-asm.o sha512-avx-asm.o sha512-avx2-asm.o sha512_ssse3_glue.o
diff --git a/arch/x86/crypto/aesni-intel_asm.S b/arch/x86/crypto/aesni-intel_asm.S
index 04b797767b9e..62fe22cd4cba 100644
--- a/arch/x86/crypto/aesni-intel_asm.S
+++ b/arch/x86/crypto/aesni-intel_asm.S
@@ -34,6 +34,10 @@
 
 #ifdef __x86_64__
 .data
+.align 16
+.Lgf128mul_x_ble_mask:
+	.octa 0x00000000000000010000000000000087
+
 POLY:   .octa 0xC2000000000000000000000000000001
 TWOONE: .octa 0x00000001000000000000000000000001
 
@@ -105,6 +109,8 @@ enc:        .octa 0x2
 #define CTR	%xmm11
 #define INC	%xmm12
 
+#define GF128MUL_MASK %xmm10
+
 #ifdef __x86_64__
 #define AREG	%rax
 #define KEYP	%rdi
@@ -2636,4 +2642,115 @@ ENTRY(aesni_ctr_enc)
 .Lctr_enc_just_ret:
 	ret
 ENDPROC(aesni_ctr_enc)
+
+/*
+ * _aesni_gf128mul_x_ble:		internal ABI
+ *	Multiply in GF(2^128) for XTS IVs
+ * input:
+ *	IV:	current IV
+ *	GF128MUL_MASK == mask with 0x87 and 0x01
+ * output:
+ *	IV:	next IV
+ * changed:
+ *	CTR:	== temporary value
+ */
+#define _aesni_gf128mul_x_ble() \
+	pshufd $0x13, IV, CTR; \
+	paddq IV, IV; \
+	psrad $31, CTR; \
+	pand GF128MUL_MASK, CTR; \
+	pxor CTR, IV;
+
+/*
+ * void aesni_xts_crypt8(struct crypto_aes_ctx *ctx, const u8 *dst, u8 *src,
+ *			 bool enc, u8 *iv)
+ */
+ENTRY(aesni_xts_crypt8)
+	cmpb $0, %cl
+	movl $0, %ecx
+	movl $240, %r10d
+	leaq _aesni_enc4, %r11
+	leaq _aesni_dec4, %rax
+	cmovel %r10d, %ecx
+	cmoveq %rax, %r11
+
+	movdqa .Lgf128mul_x_ble_mask, GF128MUL_MASK
+	movups (IVP), IV
+
+	mov 480(KEYP), KLEN
+	addq %rcx, KEYP
+
+	movdqa IV, STATE1
+	pxor 0x00(INP), STATE1
+	movdqu IV, 0x00(OUTP)
+
+	_aesni_gf128mul_x_ble()
+	movdqa IV, STATE2
+	pxor 0x10(INP), STATE2
+	movdqu IV, 0x10(OUTP)
+
+	_aesni_gf128mul_x_ble()
+	movdqa IV, STATE3
+	pxor 0x20(INP), STATE3
+	movdqu IV, 0x20(OUTP)
+
+	_aesni_gf128mul_x_ble()
+	movdqa IV, STATE4
+	pxor 0x30(INP), STATE4
+	movdqu IV, 0x30(OUTP)
+
+	call *%r11
+
+	pxor 0x00(OUTP), STATE1
+	movdqu STATE1, 0x00(OUTP)
+
+	_aesni_gf128mul_x_ble()
+	movdqa IV, STATE1
+	pxor 0x40(INP), STATE1
+	movdqu IV, 0x40(OUTP)
+
+	pxor 0x10(OUTP), STATE2
+	movdqu STATE2, 0x10(OUTP)
+
+	_aesni_gf128mul_x_ble()
+	movdqa IV, STATE2
+	pxor 0x50(INP), STATE2
+	movdqu IV, 0x50(OUTP)
+
+	pxor 0x20(OUTP), STATE3
+	movdqu STATE3, 0x20(OUTP)
+
+	_aesni_gf128mul_x_ble()
+	movdqa IV, STATE3
+	pxor 0x60(INP), STATE3
+	movdqu IV, 0x60(OUTP)
+
+	pxor 0x30(OUTP), STATE4
+	movdqu STATE4, 0x30(OUTP)
+
+	_aesni_gf128mul_x_ble()
+	movdqa IV, STATE4
+	pxor 0x70(INP), STATE4
+	movdqu IV, 0x70(OUTP)
+
+	_aesni_gf128mul_x_ble()
+	movups IV, (IVP)
+
+	call *%r11
+
+	pxor 0x40(OUTP), STATE1
+	movdqu STATE1, 0x40(OUTP)
+
+	pxor 0x50(OUTP), STATE2
+	movdqu STATE2, 0x50(OUTP)
+
+	pxor 0x60(OUTP), STATE3
+	movdqu STATE3, 0x60(OUTP)
+
+	pxor 0x70(OUTP), STATE4
+	movdqu STATE4, 0x70(OUTP)
+
+	ret
+ENDPROC(aesni_xts_crypt8)
+
 #endif
diff --git a/arch/x86/crypto/aesni-intel_glue.c b/arch/x86/crypto/aesni-intel_glue.c
index a0795da22c02..f80e668785c0 100644
--- a/arch/x86/crypto/aesni-intel_glue.c
+++ b/arch/x86/crypto/aesni-intel_glue.c
@@ -39,6 +39,9 @@
 #include <crypto/internal/aead.h>
 #include <linux/workqueue.h>
 #include <linux/spinlock.h>
+#ifdef CONFIG_X86_64
+#include <asm/crypto/glue_helper.h>
+#endif
 
 #if defined(CONFIG_CRYPTO_PCBC) || defined(CONFIG_CRYPTO_PCBC_MODULE)
 #define HAS_PCBC
@@ -102,6 +105,9 @@ void crypto_fpu_exit(void);
 asmlinkage void aesni_ctr_enc(struct crypto_aes_ctx *ctx, u8 *out,
 			      const u8 *in, unsigned int len, u8 *iv);
 
+asmlinkage void aesni_xts_crypt8(struct crypto_aes_ctx *ctx, u8 *out,
+				 const u8 *in, bool enc, 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.
@@ -510,6 +516,78 @@ static void aesni_xts_tweak(void *ctx, u8 *out, const u8 *in)
 	aesni_enc(ctx, out, in);
 }
 
+#ifdef CONFIG_X86_64
+
+static void aesni_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv, GLUE_FUNC_CAST(aesni_enc));
+}
+
+static void aesni_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv, GLUE_FUNC_CAST(aesni_dec));
+}
+
+static void aesni_xts_enc8(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	aesni_xts_crypt8(ctx, (u8 *)dst, (const u8 *)src, true, (u8 *)iv);
+}
+
+static void aesni_xts_dec8(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	aesni_xts_crypt8(ctx, (u8 *)dst, (const u8 *)src, false, (u8 *)iv);
+}
+
+static const struct common_glue_ctx aesni_enc_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = 1,
+
+	.funcs = { {
+		.num_blocks = 8,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(aesni_xts_enc8) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(aesni_xts_enc) }
+	} }
+};
+
+static const struct common_glue_ctx aesni_dec_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = 1,
+
+	.funcs = { {
+		.num_blocks = 8,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(aesni_xts_dec8) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(aesni_xts_dec) }
+	} }
+};
+
+static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct aesni_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+
+	return glue_xts_crypt_128bit(&aesni_enc_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(aesni_xts_tweak),
+				     aes_ctx(ctx->raw_tweak_ctx),
+				     aes_ctx(ctx->raw_crypt_ctx));
+}
+
+static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct aesni_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+
+	return glue_xts_crypt_128bit(&aesni_dec_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(aesni_xts_tweak),
+				     aes_ctx(ctx->raw_tweak_ctx),
+				     aes_ctx(ctx->raw_crypt_ctx));
+}
+
+#else
+
 static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
@@ -560,6 +638,8 @@ static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 	return ret;
 }
 
+#endif
+
 #ifdef CONFIG_X86_64
 static int rfc4106_init(struct crypto_tfm *tfm)
 {
diff --git a/arch/x86/crypto/blowfish-avx2-asm_64.S b/arch/x86/crypto/blowfish-avx2-asm_64.S
new file mode 100644
index 000000000000..784452e0d05d
--- /dev/null
+++ b/arch/x86/crypto/blowfish-avx2-asm_64.S
@@ -0,0 +1,449 @@
+/*
+ * x86_64/AVX2 assembler optimized version of Blowfish
+ *
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
+ *
+ * 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.
+ *
+ */
+
+#include <linux/linkage.h>
+
+.file "blowfish-avx2-asm_64.S"
+
+.data
+.align 32
+
+.Lprefetch_mask:
+.long 0*64
+.long 1*64
+.long 2*64
+.long 3*64
+.long 4*64
+.long 5*64
+.long 6*64
+.long 7*64
+
+.Lbswap32_mask:
+.long 0x00010203
+.long 0x04050607
+.long 0x08090a0b
+.long 0x0c0d0e0f
+
+.Lbswap128_mask:
+	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
+.Lbswap_iv_mask:
+	.byte 7, 6, 5, 4, 3, 2, 1, 0, 7, 6, 5, 4, 3, 2, 1, 0
+
+.text
+/* structure of crypto context */
+#define p	0
+#define s0	((16 + 2) * 4)
+#define s1	((16 + 2 + (1 * 256)) * 4)
+#define s2	((16 + 2 + (2 * 256)) * 4)
+#define s3	((16 + 2 + (3 * 256)) * 4)
+
+/* register macros */
+#define CTX	%rdi
+#define RIO	 %rdx
+
+#define RS0	%rax
+#define RS1	%r8
+#define RS2	%r9
+#define RS3	%r10
+
+#define RLOOP	%r11
+#define RLOOPd	%r11d
+
+#define RXr0	%ymm8
+#define RXr1	%ymm9
+#define RXr2	%ymm10
+#define RXr3	%ymm11
+#define RXl0	%ymm12
+#define RXl1	%ymm13
+#define RXl2	%ymm14
+#define RXl3	%ymm15
+
+/* temp regs */
+#define RT0	%ymm0
+#define RT0x	%xmm0
+#define RT1	%ymm1
+#define RT1x	%xmm1
+#define RIDX0	%ymm2
+#define RIDX1	%ymm3
+#define RIDX1x	%xmm3
+#define RIDX2	%ymm4
+#define RIDX3	%ymm5
+
+/* vpgatherdd mask and '-1' */
+#define RNOT	%ymm6
+
+/* byte mask, (-1 >> 24) */
+#define RBYTE	%ymm7
+
+/***********************************************************************
+ * 32-way AVX2 blowfish
+ ***********************************************************************/
+#define F(xl, xr) \
+	vpsrld $24, xl, RIDX0; \
+	vpsrld $16, xl, RIDX1; \
+	vpsrld $8, xl, RIDX2; \
+	vpand RBYTE, RIDX1, RIDX1; \
+	vpand RBYTE, RIDX2, RIDX2; \
+	vpand RBYTE, xl, RIDX3; \
+	\
+	vpgatherdd RNOT, (RS0, RIDX0, 4), RT0; \
+	vpcmpeqd RNOT, RNOT, RNOT; \
+	vpcmpeqd RIDX0, RIDX0, RIDX0; \
+	\
+	vpgatherdd RNOT, (RS1, RIDX1, 4), RT1; \
+	vpcmpeqd RIDX1, RIDX1, RIDX1; \
+	vpaddd RT0, RT1, RT0; \
+	\
+	vpgatherdd RIDX0, (RS2, RIDX2, 4), RT1; \
+	vpxor RT0, RT1, RT0; \
+	\
+	vpgatherdd RIDX1, (RS3, RIDX3, 4), RT1; \
+	vpcmpeqd RNOT, RNOT, RNOT; \
+	vpaddd RT0, RT1, RT0; \
+	\
+	vpxor RT0, xr, xr;
+
+#define add_roundkey(xl, nmem) \
+	vpbroadcastd nmem, RT0; \
+	vpxor RT0, xl ## 0, xl ## 0; \
+	vpxor RT0, xl ## 1, xl ## 1; \
+	vpxor RT0, xl ## 2, xl ## 2; \
+	vpxor RT0, xl ## 3, xl ## 3;
+
+#define round_enc() \
+	add_roundkey(RXr, p(CTX,RLOOP,4)); \
+	F(RXl0, RXr0); \
+	F(RXl1, RXr1); \
+	F(RXl2, RXr2); \
+	F(RXl3, RXr3); \
+	\
+	add_roundkey(RXl, p+4(CTX,RLOOP,4)); \
+	F(RXr0, RXl0); \
+	F(RXr1, RXl1); \
+	F(RXr2, RXl2); \
+	F(RXr3, RXl3);
+
+#define round_dec() \
+	add_roundkey(RXr, p+4*2(CTX,RLOOP,4)); \
+	F(RXl0, RXr0); \
+	F(RXl1, RXr1); \
+	F(RXl2, RXr2); \
+	F(RXl3, RXr3); \
+	\
+	add_roundkey(RXl, p+4(CTX,RLOOP,4)); \
+	F(RXr0, RXl0); \
+	F(RXr1, RXl1); \
+	F(RXr2, RXl2); \
+	F(RXr3, RXl3);
+
+#define init_round_constants() \
+	vpcmpeqd RNOT, RNOT, RNOT; \
+	leaq s0(CTX), RS0; \
+	leaq s1(CTX), RS1; \
+	leaq s2(CTX), RS2; \
+	leaq s3(CTX), RS3; \
+	vpsrld $24, RNOT, RBYTE;
+
+#define transpose_2x2(x0, x1, t0) \
+	vpunpckldq x0, x1, t0; \
+	vpunpckhdq x0, x1, x1; \
+	\
+	vpunpcklqdq t0, x1, x0; \
+	vpunpckhqdq t0, x1, x1;
+
+#define read_block(xl, xr) \
+	vbroadcasti128 .Lbswap32_mask, RT1; \
+	\
+	vpshufb RT1, xl ## 0, xl ## 0; \
+	vpshufb RT1, xr ## 0, xr ## 0; \
+	vpshufb RT1, xl ## 1, xl ## 1; \
+	vpshufb RT1, xr ## 1, xr ## 1; \
+	vpshufb RT1, xl ## 2, xl ## 2; \
+	vpshufb RT1, xr ## 2, xr ## 2; \
+	vpshufb RT1, xl ## 3, xl ## 3; \
+	vpshufb RT1, xr ## 3, xr ## 3; \
+	\
+	transpose_2x2(xl ## 0, xr ## 0, RT0); \
+	transpose_2x2(xl ## 1, xr ## 1, RT0); \
+	transpose_2x2(xl ## 2, xr ## 2, RT0); \
+	transpose_2x2(xl ## 3, xr ## 3, RT0);
+
+#define write_block(xl, xr) \
+	vbroadcasti128 .Lbswap32_mask, RT1; \
+	\
+	transpose_2x2(xl ## 0, xr ## 0, RT0); \
+	transpose_2x2(xl ## 1, xr ## 1, RT0); \
+	transpose_2x2(xl ## 2, xr ## 2, RT0); \
+	transpose_2x2(xl ## 3, xr ## 3, RT0); \
+	\
+	vpshufb RT1, xl ## 0, xl ## 0; \
+	vpshufb RT1, xr ## 0, xr ## 0; \
+	vpshufb RT1, xl ## 1, xl ## 1; \
+	vpshufb RT1, xr ## 1, xr ## 1; \
+	vpshufb RT1, xl ## 2, xl ## 2; \
+	vpshufb RT1, xr ## 2, xr ## 2; \
+	vpshufb RT1, xl ## 3, xl ## 3; \
+	vpshufb RT1, xr ## 3, xr ## 3;
+
+.align 8
+__blowfish_enc_blk32:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	RXl0..4, RXr0..4: plaintext
+	 * output:
+	 *	RXl0..4, RXr0..4: ciphertext (RXl <=> RXr swapped)
+	 */
+	init_round_constants();
+
+	read_block(RXl, RXr);
+
+	movl $1, RLOOPd;
+	add_roundkey(RXl, p+4*(0)(CTX));
+
+.align 4
+.L__enc_loop:
+	round_enc();
+
+	leal 2(RLOOPd), RLOOPd;
+	cmpl $17, RLOOPd;
+	jne .L__enc_loop;
+
+	add_roundkey(RXr, p+4*(17)(CTX));
+
+	write_block(RXl, RXr);
+
+	ret;
+ENDPROC(__blowfish_enc_blk32)
+
+.align 8
+__blowfish_dec_blk32:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	RXl0..4, RXr0..4: ciphertext
+	 * output:
+	 *	RXl0..4, RXr0..4: plaintext (RXl <=> RXr swapped)
+	 */
+	init_round_constants();
+
+	read_block(RXl, RXr);
+
+	movl $14, RLOOPd;
+	add_roundkey(RXl, p+4*(17)(CTX));
+
+.align 4
+.L__dec_loop:
+	round_dec();
+
+	addl $-2, RLOOPd;
+	jns .L__dec_loop;
+
+	add_roundkey(RXr, p+4*(0)(CTX));
+
+	write_block(RXl, RXr);
+
+	ret;
+ENDPROC(__blowfish_dec_blk32)
+
+ENTRY(blowfish_ecb_enc_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+
+	vmovdqu 0*32(%rdx), RXl0;
+	vmovdqu 1*32(%rdx), RXr0;
+	vmovdqu 2*32(%rdx), RXl1;
+	vmovdqu 3*32(%rdx), RXr1;
+	vmovdqu 4*32(%rdx), RXl2;
+	vmovdqu 5*32(%rdx), RXr2;
+	vmovdqu 6*32(%rdx), RXl3;
+	vmovdqu 7*32(%rdx), RXr3;
+
+	call __blowfish_enc_blk32;
+
+	vmovdqu RXr0, 0*32(%rsi);
+	vmovdqu RXl0, 1*32(%rsi);
+	vmovdqu RXr1, 2*32(%rsi);
+	vmovdqu RXl1, 3*32(%rsi);
+	vmovdqu RXr2, 4*32(%rsi);
+	vmovdqu RXl2, 5*32(%rsi);
+	vmovdqu RXr3, 6*32(%rsi);
+	vmovdqu RXl3, 7*32(%rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(blowfish_ecb_enc_32way)
+
+ENTRY(blowfish_ecb_dec_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+
+	vmovdqu 0*32(%rdx), RXl0;
+	vmovdqu 1*32(%rdx), RXr0;
+	vmovdqu 2*32(%rdx), RXl1;
+	vmovdqu 3*32(%rdx), RXr1;
+	vmovdqu 4*32(%rdx), RXl2;
+	vmovdqu 5*32(%rdx), RXr2;
+	vmovdqu 6*32(%rdx), RXl3;
+	vmovdqu 7*32(%rdx), RXr3;
+
+	call __blowfish_dec_blk32;
+
+	vmovdqu RXr0, 0*32(%rsi);
+	vmovdqu RXl0, 1*32(%rsi);
+	vmovdqu RXr1, 2*32(%rsi);
+	vmovdqu RXl1, 3*32(%rsi);
+	vmovdqu RXr2, 4*32(%rsi);
+	vmovdqu RXl2, 5*32(%rsi);
+	vmovdqu RXr3, 6*32(%rsi);
+	vmovdqu RXl3, 7*32(%rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(blowfish_ecb_dec_32way)
+
+ENTRY(blowfish_cbc_dec_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+
+	vmovdqu 0*32(%rdx), RXl0;
+	vmovdqu 1*32(%rdx), RXr0;
+	vmovdqu 2*32(%rdx), RXl1;
+	vmovdqu 3*32(%rdx), RXr1;
+	vmovdqu 4*32(%rdx), RXl2;
+	vmovdqu 5*32(%rdx), RXr2;
+	vmovdqu 6*32(%rdx), RXl3;
+	vmovdqu 7*32(%rdx), RXr3;
+
+	call __blowfish_dec_blk32;
+
+	/* xor with src */
+	vmovq (%rdx), RT0x;
+	vpshufd $0x4f, RT0x, RT0x;
+	vinserti128 $1, 8(%rdx), RT0, RT0;
+	vpxor RT0, RXr0, RXr0;
+	vpxor 0*32+24(%rdx), RXl0, RXl0;
+	vpxor 1*32+24(%rdx), RXr1, RXr1;
+	vpxor 2*32+24(%rdx), RXl1, RXl1;
+	vpxor 3*32+24(%rdx), RXr2, RXr2;
+	vpxor 4*32+24(%rdx), RXl2, RXl2;
+	vpxor 5*32+24(%rdx), RXr3, RXr3;
+	vpxor 6*32+24(%rdx), RXl3, RXl3;
+
+	vmovdqu RXr0, (0*32)(%rsi);
+	vmovdqu RXl0, (1*32)(%rsi);
+	vmovdqu RXr1, (2*32)(%rsi);
+	vmovdqu RXl1, (3*32)(%rsi);
+	vmovdqu RXr2, (4*32)(%rsi);
+	vmovdqu RXl2, (5*32)(%rsi);
+	vmovdqu RXr3, (6*32)(%rsi);
+	vmovdqu RXl3, (7*32)(%rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(blowfish_cbc_dec_32way)
+
+ENTRY(blowfish_ctr_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 *	%rcx: iv (big endian, 64bit)
+	 */
+
+	vzeroupper;
+
+	vpcmpeqd RT0, RT0, RT0;
+	vpsrldq $8, RT0, RT0; /* a: -1, b: 0, c: -1, d: 0 */
+
+	vpcmpeqd RT1x, RT1x, RT1x;
+	vpaddq RT1x, RT1x, RT1x; /* a: -2, b: -2 */
+	vpxor RIDX0, RIDX0, RIDX0;
+	vinserti128 $1, RT1x, RIDX0, RIDX0; /* a: 0, b: 0, c: -2, d: -2 */
+
+	vpaddq RIDX0, RT0, RT0; /* a: -1, b: 0, c: -3, d: -2 */
+
+	vpcmpeqd RT1, RT1, RT1;
+	vpaddq RT1, RT1, RT1; /* a: -2, b: -2, c: -2, d: -2 */
+	vpaddq RT1, RT1, RIDX2; /* a: -4, b: -4, c: -4, d: -4 */
+
+	vbroadcasti128 .Lbswap_iv_mask, RIDX0;
+	vbroadcasti128 .Lbswap128_mask, RIDX1;
+
+	/* load IV and byteswap */
+	vmovq (%rcx), RT1x;
+	vinserti128 $1, RT1x, RT1, RT1; /* a: BE, b: 0, c: BE, d: 0 */
+	vpshufb RIDX0, RT1, RT1; /* a: LE, b: LE, c: LE, d: LE */
+
+	/* construct IVs */
+	vpsubq RT0, RT1, RT1;		/* a: le1, b: le0, c: le3, d: le2 */
+	vpshufb RIDX1, RT1, RXl0;	/* a: be0, b: be1, c: be2, d: be3 */
+	vpsubq RIDX2, RT1, RT1;		/* le5, le4, le7, le6 */
+	vpshufb RIDX1, RT1, RXr0;	/* be4, be5, be6, be7 */
+	vpsubq RIDX2, RT1, RT1;
+	vpshufb RIDX1, RT1, RXl1;
+	vpsubq RIDX2, RT1, RT1;
+	vpshufb RIDX1, RT1, RXr1;
+	vpsubq RIDX2, RT1, RT1;
+	vpshufb RIDX1, RT1, RXl2;
+	vpsubq RIDX2, RT1, RT1;
+	vpshufb RIDX1, RT1, RXr2;
+	vpsubq RIDX2, RT1, RT1;
+	vpshufb RIDX1, RT1, RXl3;
+	vpsubq RIDX2, RT1, RT1;
+	vpshufb RIDX1, RT1, RXr3;
+
+	/* store last IV */
+	vpsubq RIDX2, RT1, RT1; /* a: le33, b: le32, ... */
+	vpshufb RIDX1x, RT1x, RT1x; /* a: be32, ... */
+	vmovq RT1x, (%rcx);
+
+	call __blowfish_enc_blk32;
+
+	/* dst = src ^ iv */
+	vpxor 0*32(%rdx), RXr0, RXr0;
+	vpxor 1*32(%rdx), RXl0, RXl0;
+	vpxor 2*32(%rdx), RXr1, RXr1;
+	vpxor 3*32(%rdx), RXl1, RXl1;
+	vpxor 4*32(%rdx), RXr2, RXr2;
+	vpxor 5*32(%rdx), RXl2, RXl2;
+	vpxor 6*32(%rdx), RXr3, RXr3;
+	vpxor 7*32(%rdx), RXl3, RXl3;
+	vmovdqu RXr0, (0*32)(%rsi);
+	vmovdqu RXl0, (1*32)(%rsi);
+	vmovdqu RXr1, (2*32)(%rsi);
+	vmovdqu RXl1, (3*32)(%rsi);
+	vmovdqu RXr2, (4*32)(%rsi);
+	vmovdqu RXl2, (5*32)(%rsi);
+	vmovdqu RXr3, (6*32)(%rsi);
+	vmovdqu RXl3, (7*32)(%rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(blowfish_ctr_32way)
diff --git a/arch/x86/crypto/blowfish_avx2_glue.c b/arch/x86/crypto/blowfish_avx2_glue.c
new file mode 100644
index 000000000000..4417e9aea78d
--- /dev/null
+++ b/arch/x86/crypto/blowfish_avx2_glue.c
@@ -0,0 +1,585 @@
+/*
+ * Glue Code for x86_64/AVX2 assembler optimized version of Blowfish
+ *
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ *
+ * CBC & ECB parts based on code (crypto/cbc.c,ecb.c) by:
+ *   Copyright (c) 2006 Herbert Xu <herbert@gondor.apana.org.au>
+ * CTR part based on code (crypto/ctr.c) by:
+ *   (C) Copyright IBM Corp. 2007 - Joy Latten <latten@us.ibm.com>
+ *
+ * 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 program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/crypto.h>
+#include <linux/err.h>
+#include <crypto/algapi.h>
+#include <crypto/blowfish.h>
+#include <crypto/cryptd.h>
+#include <crypto/ctr.h>
+#include <asm/i387.h>
+#include <asm/xcr.h>
+#include <asm/xsave.h>
+#include <asm/crypto/blowfish.h>
+#include <asm/crypto/ablk_helper.h>
+#include <crypto/scatterwalk.h>
+
+#define BF_AVX2_PARALLEL_BLOCKS 32
+
+/* 32-way AVX2 parallel cipher functions */
+asmlinkage void blowfish_ecb_enc_32way(struct bf_ctx *ctx, u8 *dst,
+				       const u8 *src);
+asmlinkage void blowfish_ecb_dec_32way(struct bf_ctx *ctx, u8 *dst,
+				       const u8 *src);
+asmlinkage void blowfish_cbc_dec_32way(struct bf_ctx *ctx, u8 *dst,
+				       const u8 *src);
+asmlinkage void blowfish_ctr_32way(struct bf_ctx *ctx, u8 *dst, const u8 *src,
+				   __be64 *iv);
+
+static inline bool bf_fpu_begin(bool fpu_enabled, unsigned int nbytes)
+{
+	if (fpu_enabled)
+		return true;
+
+	/* FPU is only used when chunk to be processed is large enough, so
+	 * do not enable FPU until it is necessary.
+	 */
+	if (nbytes < BF_BLOCK_SIZE * BF_AVX2_PARALLEL_BLOCKS)
+		return false;
+
+	kernel_fpu_begin();
+	return true;
+}
+
+static inline void bf_fpu_end(bool fpu_enabled)
+{
+	if (fpu_enabled)
+		kernel_fpu_end();
+}
+
+static int ecb_crypt(struct blkcipher_desc *desc, struct blkcipher_walk *walk,
+		     bool enc)
+{
+	bool fpu_enabled = false;
+	struct bf_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	const unsigned int bsize = BF_BLOCK_SIZE;
+	unsigned int nbytes;
+	int err;
+
+	err = blkcipher_walk_virt(desc, walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+
+	while ((nbytes = walk->nbytes)) {
+		u8 *wsrc = walk->src.virt.addr;
+		u8 *wdst = walk->dst.virt.addr;
+
+		fpu_enabled = bf_fpu_begin(fpu_enabled, nbytes);
+
+		/* Process multi-block AVX2 batch */
+		if (nbytes >= bsize * BF_AVX2_PARALLEL_BLOCKS) {
+			do {
+				if (enc)
+					blowfish_ecb_enc_32way(ctx, wdst, wsrc);
+				else
+					blowfish_ecb_dec_32way(ctx, wdst, wsrc);
+
+				wsrc += bsize * BF_AVX2_PARALLEL_BLOCKS;
+				wdst += bsize * BF_AVX2_PARALLEL_BLOCKS;
+				nbytes -= bsize * BF_AVX2_PARALLEL_BLOCKS;
+			} while (nbytes >= bsize * BF_AVX2_PARALLEL_BLOCKS);
+
+			if (nbytes < bsize)
+				goto done;
+		}
+
+		/* Process multi-block batch */
+		if (nbytes >= bsize * BF_PARALLEL_BLOCKS) {
+			do {
+				if (enc)
+					blowfish_enc_blk_4way(ctx, wdst, wsrc);
+				else
+					blowfish_dec_blk_4way(ctx, wdst, wsrc);
+
+				wsrc += bsize * BF_PARALLEL_BLOCKS;
+				wdst += bsize * BF_PARALLEL_BLOCKS;
+				nbytes -= bsize * BF_PARALLEL_BLOCKS;
+			} while (nbytes >= bsize * BF_PARALLEL_BLOCKS);
+
+			if (nbytes < bsize)
+				goto done;
+		}
+
+		/* Handle leftovers */
+		do {
+			if (enc)
+				blowfish_enc_blk(ctx, wdst, wsrc);
+			else
+				blowfish_dec_blk(ctx, wdst, wsrc);
+
+			wsrc += bsize;
+			wdst += bsize;
+			nbytes -= bsize;
+		} while (nbytes >= bsize);
+
+done:
+		err = blkcipher_walk_done(desc, walk, nbytes);
+	}
+
+	bf_fpu_end(fpu_enabled);
+	return err;
+}
+
+static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct blkcipher_walk walk;
+
+	blkcipher_walk_init(&walk, dst, src, nbytes);
+	return ecb_crypt(desc, &walk, true);
+}
+
+static int ecb_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct blkcipher_walk walk;
+
+	blkcipher_walk_init(&walk, dst, src, nbytes);
+	return ecb_crypt(desc, &walk, false);
+}
+
+static unsigned int __cbc_encrypt(struct blkcipher_desc *desc,
+				  struct blkcipher_walk *walk)
+{
+	struct bf_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	unsigned int bsize = BF_BLOCK_SIZE;
+	unsigned int nbytes = walk->nbytes;
+	u64 *src = (u64 *)walk->src.virt.addr;
+	u64 *dst = (u64 *)walk->dst.virt.addr;
+	u64 *iv = (u64 *)walk->iv;
+
+	do {
+		*dst = *src ^ *iv;
+		blowfish_enc_blk(ctx, (u8 *)dst, (u8 *)dst);
+		iv = dst;
+
+		src += 1;
+		dst += 1;
+		nbytes -= bsize;
+	} while (nbytes >= bsize);
+
+	*(u64 *)walk->iv = *iv;
+	return nbytes;
+}
+
+static int cbc_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct blkcipher_walk walk;
+	int err;
+
+	blkcipher_walk_init(&walk, dst, src, nbytes);
+	err = blkcipher_walk_virt(desc, &walk);
+
+	while ((nbytes = walk.nbytes)) {
+		nbytes = __cbc_encrypt(desc, &walk);
+		err = blkcipher_walk_done(desc, &walk, nbytes);
+	}
+
+	return err;
+}
+
+static unsigned int __cbc_decrypt(struct blkcipher_desc *desc,
+				  struct blkcipher_walk *walk)
+{
+	struct bf_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	const unsigned int bsize = BF_BLOCK_SIZE;
+	unsigned int nbytes = walk->nbytes;
+	u64 *src = (u64 *)walk->src.virt.addr;
+	u64 *dst = (u64 *)walk->dst.virt.addr;
+	u64 last_iv;
+	int i;
+
+	/* Start of the last block. */
+	src += nbytes / bsize - 1;
+	dst += nbytes / bsize - 1;
+
+	last_iv = *src;
+
+	/* Process multi-block AVX2 batch */
+	if (nbytes >= bsize * BF_AVX2_PARALLEL_BLOCKS) {
+		do {
+			nbytes -= bsize * (BF_AVX2_PARALLEL_BLOCKS - 1);
+			src -= BF_AVX2_PARALLEL_BLOCKS - 1;
+			dst -= BF_AVX2_PARALLEL_BLOCKS - 1;
+
+			blowfish_cbc_dec_32way(ctx, (u8 *)dst, (u8 *)src);
+
+			nbytes -= bsize;
+			if (nbytes < bsize)
+				goto done;
+
+			*dst ^= *(src - 1);
+			src -= 1;
+			dst -= 1;
+		} while (nbytes >= bsize * BF_AVX2_PARALLEL_BLOCKS);
+
+		if (nbytes < bsize)
+			goto done;
+	}
+
+	/* Process multi-block batch */
+	if (nbytes >= bsize * BF_PARALLEL_BLOCKS) {
+		u64 ivs[BF_PARALLEL_BLOCKS - 1];
+
+		do {
+			nbytes -= bsize * (BF_PARALLEL_BLOCKS - 1);
+			src -= BF_PARALLEL_BLOCKS - 1;
+			dst -= BF_PARALLEL_BLOCKS - 1;
+
+			for (i = 0; i < BF_PARALLEL_BLOCKS - 1; i++)
+				ivs[i] = src[i];
+
+			blowfish_dec_blk_4way(ctx, (u8 *)dst, (u8 *)src);
+
+			for (i = 0; i < BF_PARALLEL_BLOCKS - 1; i++)
+				dst[i + 1] ^= ivs[i];
+
+			nbytes -= bsize;
+			if (nbytes < bsize)
+				goto done;
+
+			*dst ^= *(src - 1);
+			src -= 1;
+			dst -= 1;
+		} while (nbytes >= bsize * BF_PARALLEL_BLOCKS);
+
+		if (nbytes < bsize)
+			goto done;
+	}
+
+	/* Handle leftovers */
+	for (;;) {
+		blowfish_dec_blk(ctx, (u8 *)dst, (u8 *)src);
+
+		nbytes -= bsize;
+		if (nbytes < bsize)
+			break;
+
+		*dst ^= *(src - 1);
+		src -= 1;
+		dst -= 1;
+	}
+
+done:
+	*dst ^= *(u64 *)walk->iv;
+	*(u64 *)walk->iv = last_iv;
+
+	return nbytes;
+}
+
+static int cbc_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	bool fpu_enabled = false;
+	struct blkcipher_walk walk;
+	int err;
+
+	blkcipher_walk_init(&walk, dst, src, nbytes);
+	err = blkcipher_walk_virt(desc, &walk);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+
+	while ((nbytes = walk.nbytes)) {
+		fpu_enabled = bf_fpu_begin(fpu_enabled, nbytes);
+		nbytes = __cbc_decrypt(desc, &walk);
+		err = blkcipher_walk_done(desc, &walk, nbytes);
+	}
+
+	bf_fpu_end(fpu_enabled);
+	return err;
+}
+
+static void ctr_crypt_final(struct blkcipher_desc *desc,
+			    struct blkcipher_walk *walk)
+{
+	struct bf_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	u8 *ctrblk = walk->iv;
+	u8 keystream[BF_BLOCK_SIZE];
+	u8 *src = walk->src.virt.addr;
+	u8 *dst = walk->dst.virt.addr;
+	unsigned int nbytes = walk->nbytes;
+
+	blowfish_enc_blk(ctx, keystream, ctrblk);
+	crypto_xor(keystream, src, nbytes);
+	memcpy(dst, keystream, nbytes);
+
+	crypto_inc(ctrblk, BF_BLOCK_SIZE);
+}
+
+static unsigned int __ctr_crypt(struct blkcipher_desc *desc,
+				struct blkcipher_walk *walk)
+{
+	struct bf_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	unsigned int bsize = BF_BLOCK_SIZE;
+	unsigned int nbytes = walk->nbytes;
+	u64 *src = (u64 *)walk->src.virt.addr;
+	u64 *dst = (u64 *)walk->dst.virt.addr;
+	int i;
+
+	/* Process multi-block AVX2 batch */
+	if (nbytes >= bsize * BF_AVX2_PARALLEL_BLOCKS) {
+		do {
+			blowfish_ctr_32way(ctx, (u8 *)dst, (u8 *)src,
+					   (__be64 *)walk->iv);
+
+			src += BF_AVX2_PARALLEL_BLOCKS;
+			dst += BF_AVX2_PARALLEL_BLOCKS;
+			nbytes -= bsize * BF_AVX2_PARALLEL_BLOCKS;
+		} while (nbytes >= bsize * BF_AVX2_PARALLEL_BLOCKS);
+
+		if (nbytes < bsize)
+			goto done;
+	}
+
+	/* Process four block batch */
+	if (nbytes >= bsize * BF_PARALLEL_BLOCKS) {
+		__be64 ctrblocks[BF_PARALLEL_BLOCKS];
+		u64 ctrblk = be64_to_cpu(*(__be64 *)walk->iv);
+
+		do {
+			/* create ctrblks for parallel encrypt */
+			for (i = 0; i < BF_PARALLEL_BLOCKS; i++) {
+				if (dst != src)
+					dst[i] = src[i];
+
+				ctrblocks[i] = cpu_to_be64(ctrblk++);
+			}
+
+			blowfish_enc_blk_xor_4way(ctx, (u8 *)dst,
+						  (u8 *)ctrblocks);
+
+			src += BF_PARALLEL_BLOCKS;
+			dst += BF_PARALLEL_BLOCKS;
+			nbytes -= bsize * BF_PARALLEL_BLOCKS;
+		} while (nbytes >= bsize * BF_PARALLEL_BLOCKS);
+
+		*(__be64 *)walk->iv = cpu_to_be64(ctrblk);
+
+		if (nbytes < bsize)
+			goto done;
+	}
+
+	/* Handle leftovers */
+	do {
+		u64 ctrblk;
+
+		if (dst != src)
+			*dst = *src;
+
+		ctrblk = *(u64 *)walk->iv;
+		be64_add_cpu((__be64 *)walk->iv, 1);
+
+		blowfish_enc_blk_xor(ctx, (u8 *)dst, (u8 *)&ctrblk);
+
+		src += 1;
+		dst += 1;
+	} while ((nbytes -= bsize) >= bsize);
+
+done:
+	return nbytes;
+}
+
+static int ctr_crypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		     struct scatterlist *src, unsigned int nbytes)
+{
+	bool fpu_enabled = false;
+	struct blkcipher_walk walk;
+	int err;
+
+	blkcipher_walk_init(&walk, dst, src, nbytes);
+	err = blkcipher_walk_virt_block(desc, &walk, BF_BLOCK_SIZE);
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+
+	while ((nbytes = walk.nbytes) >= BF_BLOCK_SIZE) {
+		fpu_enabled = bf_fpu_begin(fpu_enabled, nbytes);
+		nbytes = __ctr_crypt(desc, &walk);
+		err = blkcipher_walk_done(desc, &walk, nbytes);
+	}
+
+	bf_fpu_end(fpu_enabled);
+
+	if (walk.nbytes) {
+		ctr_crypt_final(desc, &walk);
+		err = blkcipher_walk_done(desc, &walk, 0);
+	}
+
+	return err;
+}
+
+static struct crypto_alg bf_algs[6] = { {
+	.cra_name		= "__ecb-blowfish-avx2",
+	.cra_driver_name	= "__driver-ecb-blowfish-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= BF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct bf_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= BF_MIN_KEY_SIZE,
+			.max_keysize	= BF_MAX_KEY_SIZE,
+			.setkey		= blowfish_setkey,
+			.encrypt	= ecb_encrypt,
+			.decrypt	= ecb_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__cbc-blowfish-avx2",
+	.cra_driver_name	= "__driver-cbc-blowfish-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= BF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct bf_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= BF_MIN_KEY_SIZE,
+			.max_keysize	= BF_MAX_KEY_SIZE,
+			.setkey		= blowfish_setkey,
+			.encrypt	= cbc_encrypt,
+			.decrypt	= cbc_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__ctr-blowfish-avx2",
+	.cra_driver_name	= "__driver-ctr-blowfish-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= 1,
+	.cra_ctxsize		= sizeof(struct bf_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= BF_MIN_KEY_SIZE,
+			.max_keysize	= BF_MAX_KEY_SIZE,
+			.ivsize		= BF_BLOCK_SIZE,
+			.setkey		= blowfish_setkey,
+			.encrypt	= ctr_crypt,
+			.decrypt	= ctr_crypt,
+		},
+	},
+}, {
+	.cra_name		= "ecb(blowfish)",
+	.cra_driver_name	= "ecb-blowfish-avx2",
+	.cra_priority		= 400,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= BF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= BF_MIN_KEY_SIZE,
+			.max_keysize	= BF_MAX_KEY_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "cbc(blowfish)",
+	.cra_driver_name	= "cbc-blowfish-avx2",
+	.cra_priority		= 400,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= BF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= BF_MIN_KEY_SIZE,
+			.max_keysize	= BF_MAX_KEY_SIZE,
+			.ivsize		= BF_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= __ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "ctr(blowfish)",
+	.cra_driver_name	= "ctr-blowfish-avx2",
+	.cra_priority		= 400,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= 1,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= BF_MIN_KEY_SIZE,
+			.max_keysize	= BF_MAX_KEY_SIZE,
+			.ivsize		= BF_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_encrypt,
+			.geniv		= "chainiv",
+		},
+	},
+} };
+
+
+static int __init init(void)
+{
+	u64 xcr0;
+
+	if (!cpu_has_avx2 || !cpu_has_osxsave) {
+		pr_info("AVX2 instructions are not detected.\n");
+		return -ENODEV;
+	}
+
+	xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
+	if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
+		pr_info("AVX detected but unusable.\n");
+		return -ENODEV;
+	}
+
+	return crypto_register_algs(bf_algs, ARRAY_SIZE(bf_algs));
+}
+
+static void __exit fini(void)
+{
+	crypto_unregister_algs(bf_algs, ARRAY_SIZE(bf_algs));
+}
+
+module_init(init);
+module_exit(fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Blowfish Cipher Algorithm, AVX2 optimized");
+MODULE_ALIAS("blowfish");
+MODULE_ALIAS("blowfish-asm");
diff --git a/arch/x86/crypto/blowfish_glue.c b/arch/x86/crypto/blowfish_glue.c
index 50ec333b70e6..3548d76dbaa9 100644
--- a/arch/x86/crypto/blowfish_glue.c
+++ b/arch/x86/crypto/blowfish_glue.c
@@ -1,7 +1,7 @@
 /*
  * Glue Code for assembler optimized version of Blowfish
  *
- * Copyright (c) 2011 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2011-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * CBC & ECB parts based on code (crypto/cbc.c,ecb.c) by:
  *   Copyright (c) 2006 Herbert Xu <herbert@gondor.apana.org.au>
@@ -32,40 +32,24 @@
 #include <linux/module.h>
 #include <linux/types.h>
 #include <crypto/algapi.h>
+#include <asm/crypto/blowfish.h>
 
 /* regular block cipher functions */
 asmlinkage void __blowfish_enc_blk(struct bf_ctx *ctx, u8 *dst, const u8 *src,
 				   bool xor);
+EXPORT_SYMBOL_GPL(__blowfish_enc_blk);
+
 asmlinkage void blowfish_dec_blk(struct bf_ctx *ctx, u8 *dst, const u8 *src);
+EXPORT_SYMBOL_GPL(blowfish_dec_blk);
 
 /* 4-way parallel cipher functions */
 asmlinkage void __blowfish_enc_blk_4way(struct bf_ctx *ctx, u8 *dst,
 					const u8 *src, bool xor);
+EXPORT_SYMBOL_GPL(__blowfish_enc_blk_4way);
+
 asmlinkage void blowfish_dec_blk_4way(struct bf_ctx *ctx, u8 *dst,
 				      const u8 *src);
-
-static inline void blowfish_enc_blk(struct bf_ctx *ctx, u8 *dst, const u8 *src)
-{
-	__blowfish_enc_blk(ctx, dst, src, false);
-}
-
-static inline void blowfish_enc_blk_xor(struct bf_ctx *ctx, u8 *dst,
-					const u8 *src)
-{
-	__blowfish_enc_blk(ctx, dst, src, true);
-}
-
-static inline void blowfish_enc_blk_4way(struct bf_ctx *ctx, u8 *dst,
-					 const u8 *src)
-{
-	__blowfish_enc_blk_4way(ctx, dst, src, false);
-}
-
-static inline void blowfish_enc_blk_xor_4way(struct bf_ctx *ctx, u8 *dst,
-				      const u8 *src)
-{
-	__blowfish_enc_blk_4way(ctx, dst, src, true);
-}
+EXPORT_SYMBOL_GPL(blowfish_dec_blk_4way);
 
 static void blowfish_encrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
 {
diff --git a/arch/x86/crypto/camellia-aesni-avx-asm_64.S b/arch/x86/crypto/camellia-aesni-avx-asm_64.S
index cfc163469c71..ce71f9212409 100644
--- a/arch/x86/crypto/camellia-aesni-avx-asm_64.S
+++ b/arch/x86/crypto/camellia-aesni-avx-asm_64.S
@@ -1,7 +1,7 @@
 /*
  * x86_64/AVX/AES-NI assembler implementation of Camellia
  *
- * Copyright © 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * 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
@@ -589,6 +589,10 @@ ENDPROC(roundsm16_x4_x5_x6_x7_x0_x1_x2_x3_y4_y5_y6_y7_y0_y1_y2_y3_ab)
 .Lbswap128_mask:
 	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
 
+/* For XTS mode IV generation */
+.Lxts_gf128mul_and_shl1_mask:
+	.byte 0x87, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
+
 /*
  * pre-SubByte transform
  *
@@ -1090,3 +1094,177 @@ ENTRY(camellia_ctr_16way)
 
 	ret;
 ENDPROC(camellia_ctr_16way)
+
+#define gf128mul_x_ble(iv, mask, tmp) \
+	vpsrad $31, iv, tmp; \
+	vpaddq iv, iv, iv; \
+	vpshufd $0x13, tmp, tmp; \
+	vpand mask, tmp, tmp; \
+	vpxor tmp, iv, iv;
+
+.align 8
+camellia_xts_crypt_16way:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 *	%r8: index for input whitening key
+	 *	%r9: pointer to  __camellia_enc_blk16 or __camellia_dec_blk16
+	 */
+
+	subq $(16 * 16), %rsp;
+	movq %rsp, %rax;
+
+	vmovdqa .Lxts_gf128mul_and_shl1_mask, %xmm14;
+
+	/* load IV */
+	vmovdqu (%rcx), %xmm0;
+	vpxor 0 * 16(%rdx), %xmm0, %xmm15;
+	vmovdqu %xmm15, 15 * 16(%rax);
+	vmovdqu %xmm0, 0 * 16(%rsi);
+
+	/* construct IVs */
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 1 * 16(%rdx), %xmm0, %xmm15;
+	vmovdqu %xmm15, 14 * 16(%rax);
+	vmovdqu %xmm0, 1 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 2 * 16(%rdx), %xmm0, %xmm13;
+	vmovdqu %xmm0, 2 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 3 * 16(%rdx), %xmm0, %xmm12;
+	vmovdqu %xmm0, 3 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 4 * 16(%rdx), %xmm0, %xmm11;
+	vmovdqu %xmm0, 4 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 5 * 16(%rdx), %xmm0, %xmm10;
+	vmovdqu %xmm0, 5 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 6 * 16(%rdx), %xmm0, %xmm9;
+	vmovdqu %xmm0, 6 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 7 * 16(%rdx), %xmm0, %xmm8;
+	vmovdqu %xmm0, 7 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 8 * 16(%rdx), %xmm0, %xmm7;
+	vmovdqu %xmm0, 8 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 9 * 16(%rdx), %xmm0, %xmm6;
+	vmovdqu %xmm0, 9 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 10 * 16(%rdx), %xmm0, %xmm5;
+	vmovdqu %xmm0, 10 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 11 * 16(%rdx), %xmm0, %xmm4;
+	vmovdqu %xmm0, 11 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 12 * 16(%rdx), %xmm0, %xmm3;
+	vmovdqu %xmm0, 12 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 13 * 16(%rdx), %xmm0, %xmm2;
+	vmovdqu %xmm0, 13 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 14 * 16(%rdx), %xmm0, %xmm1;
+	vmovdqu %xmm0, 14 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vpxor 15 * 16(%rdx), %xmm0, %xmm15;
+	vmovdqu %xmm15, 0 * 16(%rax);
+	vmovdqu %xmm0, 15 * 16(%rsi);
+
+	gf128mul_x_ble(%xmm0, %xmm14, %xmm15);
+	vmovdqu %xmm0, (%rcx);
+
+	/* inpack16_pre: */
+	vmovq (key_table)(CTX, %r8, 8), %xmm15;
+	vpshufb .Lpack_bswap, %xmm15, %xmm15;
+	vpxor 0 * 16(%rax), %xmm15, %xmm0;
+	vpxor %xmm1, %xmm15, %xmm1;
+	vpxor %xmm2, %xmm15, %xmm2;
+	vpxor %xmm3, %xmm15, %xmm3;
+	vpxor %xmm4, %xmm15, %xmm4;
+	vpxor %xmm5, %xmm15, %xmm5;
+	vpxor %xmm6, %xmm15, %xmm6;
+	vpxor %xmm7, %xmm15, %xmm7;
+	vpxor %xmm8, %xmm15, %xmm8;
+	vpxor %xmm9, %xmm15, %xmm9;
+	vpxor %xmm10, %xmm15, %xmm10;
+	vpxor %xmm11, %xmm15, %xmm11;
+	vpxor %xmm12, %xmm15, %xmm12;
+	vpxor %xmm13, %xmm15, %xmm13;
+	vpxor 14 * 16(%rax), %xmm15, %xmm14;
+	vpxor 15 * 16(%rax), %xmm15, %xmm15;
+
+	call *%r9;
+
+	addq $(16 * 16), %rsp;
+
+	vpxor 0 * 16(%rsi), %xmm7, %xmm7;
+	vpxor 1 * 16(%rsi), %xmm6, %xmm6;
+	vpxor 2 * 16(%rsi), %xmm5, %xmm5;
+	vpxor 3 * 16(%rsi), %xmm4, %xmm4;
+	vpxor 4 * 16(%rsi), %xmm3, %xmm3;
+	vpxor 5 * 16(%rsi), %xmm2, %xmm2;
+	vpxor 6 * 16(%rsi), %xmm1, %xmm1;
+	vpxor 7 * 16(%rsi), %xmm0, %xmm0;
+	vpxor 8 * 16(%rsi), %xmm15, %xmm15;
+	vpxor 9 * 16(%rsi), %xmm14, %xmm14;
+	vpxor 10 * 16(%rsi), %xmm13, %xmm13;
+	vpxor 11 * 16(%rsi), %xmm12, %xmm12;
+	vpxor 12 * 16(%rsi), %xmm11, %xmm11;
+	vpxor 13 * 16(%rsi), %xmm10, %xmm10;
+	vpxor 14 * 16(%rsi), %xmm9, %xmm9;
+	vpxor 15 * 16(%rsi), %xmm8, %xmm8;
+	write_output(%xmm7, %xmm6, %xmm5, %xmm4, %xmm3, %xmm2, %xmm1, %xmm0,
+		     %xmm15, %xmm14, %xmm13, %xmm12, %xmm11, %xmm10, %xmm9,
+		     %xmm8, %rsi);
+
+	ret;
+ENDPROC(camellia_xts_crypt_16way)
+
+ENTRY(camellia_xts_enc_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+	xorl %r8d, %r8d; /* input whitening key, 0 for enc */
+
+	leaq __camellia_enc_blk16, %r9;
+
+	jmp camellia_xts_crypt_16way;
+ENDPROC(camellia_xts_enc_16way)
+
+ENTRY(camellia_xts_dec_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	cmpl $16, key_length(CTX);
+	movl $32, %r8d;
+	movl $24, %eax;
+	cmovel %eax, %r8d;  /* input whitening key, last for dec */
+
+	leaq __camellia_dec_blk16, %r9;
+
+	jmp camellia_xts_crypt_16way;
+ENDPROC(camellia_xts_dec_16way)
diff --git a/arch/x86/crypto/camellia-aesni-avx2-asm_64.S b/arch/x86/crypto/camellia-aesni-avx2-asm_64.S
new file mode 100644
index 000000000000..91a1878fcc3e
--- /dev/null
+++ b/arch/x86/crypto/camellia-aesni-avx2-asm_64.S
@@ -0,0 +1,1368 @@
+/*
+ * x86_64/AVX2/AES-NI assembler implementation of Camellia
+ *
+ * Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
+ *
+ * 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.
+ *
+ */
+
+#include <linux/linkage.h>
+
+#define CAMELLIA_TABLE_BYTE_LEN 272
+
+/* struct camellia_ctx: */
+#define key_table 0
+#define key_length CAMELLIA_TABLE_BYTE_LEN
+
+/* register macros */
+#define CTX %rdi
+#define RIO %r8
+
+/**********************************************************************
+  helper macros
+ **********************************************************************/
+#define filter_8bit(x, lo_t, hi_t, mask4bit, tmp0) \
+	vpand x, mask4bit, tmp0; \
+	vpandn x, mask4bit, x; \
+	vpsrld $4, x, x; \
+	\
+	vpshufb tmp0, lo_t, tmp0; \
+	vpshufb x, hi_t, x; \
+	vpxor tmp0, x, x;
+
+#define ymm0_x xmm0
+#define ymm1_x xmm1
+#define ymm2_x xmm2
+#define ymm3_x xmm3
+#define ymm4_x xmm4
+#define ymm5_x xmm5
+#define ymm6_x xmm6
+#define ymm7_x xmm7
+#define ymm8_x xmm8
+#define ymm9_x xmm9
+#define ymm10_x xmm10
+#define ymm11_x xmm11
+#define ymm12_x xmm12
+#define ymm13_x xmm13
+#define ymm14_x xmm14
+#define ymm15_x xmm15
+
+/*
+ * AES-NI instructions do not support ymmX registers, so we need splitting and
+ * merging.
+ */
+#define vaesenclast256(zero, yreg, tmp) \
+	vextracti128 $1, yreg, tmp##_x; \
+	vaesenclast zero##_x, yreg##_x, yreg##_x; \
+	vaesenclast zero##_x, tmp##_x, tmp##_x; \
+	vinserti128 $1, tmp##_x, yreg, yreg;
+
+/**********************************************************************
+  32-way camellia
+ **********************************************************************/
+
+/*
+ * IN:
+ *   x0..x7: byte-sliced AB state
+ *   mem_cd: register pointer storing CD state
+ *   key: index for key material
+ * OUT:
+ *   x0..x7: new byte-sliced CD state
+ */
+#define roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, t0, t1, t2, t3, t4, t5, t6, \
+		  t7, mem_cd, key) \
+	/* \
+	 * S-function with AES subbytes \
+	 */ \
+	vbroadcasti128 .Linv_shift_row, t4; \
+	vpbroadcastb .L0f0f0f0f, t7; \
+	vbroadcasti128 .Lpre_tf_lo_s1, t0; \
+	vbroadcasti128 .Lpre_tf_hi_s1, t1; \
+	\
+	/* AES inverse shift rows */ \
+	vpshufb t4, x0, x0; \
+	vpshufb t4, x7, x7; \
+	vpshufb t4, x1, x1; \
+	vpshufb t4, x4, x4; \
+	vpshufb t4, x2, x2; \
+	vpshufb t4, x5, x5; \
+	vpshufb t4, x3, x3; \
+	vpshufb t4, x6, x6; \
+	\
+	/* prefilter sboxes 1, 2 and 3 */ \
+	vbroadcasti128 .Lpre_tf_lo_s4, t2; \
+	vbroadcasti128 .Lpre_tf_hi_s4, t3; \
+	filter_8bit(x0, t0, t1, t7, t6); \
+	filter_8bit(x7, t0, t1, t7, t6); \
+	filter_8bit(x1, t0, t1, t7, t6); \
+	filter_8bit(x4, t0, t1, t7, t6); \
+	filter_8bit(x2, t0, t1, t7, t6); \
+	filter_8bit(x5, t0, t1, t7, t6); \
+	\
+	/* prefilter sbox 4 */ \
+	vpxor t4##_x, t4##_x, t4##_x; \
+	filter_8bit(x3, t2, t3, t7, t6); \
+	filter_8bit(x6, t2, t3, t7, t6); \
+	\
+	/* AES subbytes + AES shift rows */ \
+	vbroadcasti128 .Lpost_tf_lo_s1, t0; \
+	vbroadcasti128 .Lpost_tf_hi_s1, t1; \
+	vaesenclast256(t4, x0, t5); \
+	vaesenclast256(t4, x7, t5); \
+	vaesenclast256(t4, x1, t5); \
+	vaesenclast256(t4, x4, t5); \
+	vaesenclast256(t4, x2, t5); \
+	vaesenclast256(t4, x5, t5); \
+	vaesenclast256(t4, x3, t5); \
+	vaesenclast256(t4, x6, t5); \
+	\
+	/* postfilter sboxes 1 and 4 */ \
+	vbroadcasti128 .Lpost_tf_lo_s3, t2; \
+	vbroadcasti128 .Lpost_tf_hi_s3, t3; \
+	filter_8bit(x0, t0, t1, t7, t6); \
+	filter_8bit(x7, t0, t1, t7, t6); \
+	filter_8bit(x3, t0, t1, t7, t6); \
+	filter_8bit(x6, t0, t1, t7, t6); \
+	\
+	/* postfilter sbox 3 */ \
+	vbroadcasti128 .Lpost_tf_lo_s2, t4; \
+	vbroadcasti128 .Lpost_tf_hi_s2, t5; \
+	filter_8bit(x2, t2, t3, t7, t6); \
+	filter_8bit(x5, t2, t3, t7, t6); \
+	\
+	vpbroadcastq key, t0; /* higher 64-bit duplicate ignored */ \
+	\
+	/* postfilter sbox 2 */ \
+	filter_8bit(x1, t4, t5, t7, t2); \
+	filter_8bit(x4, t4, t5, t7, t2); \
+	\
+	vpsrldq $1, t0, t1; \
+	vpsrldq $2, t0, t2; \
+	vpsrldq $3, t0, t3; \
+	vpsrldq $4, t0, t4; \
+	vpsrldq $5, t0, t5; \
+	vpsrldq $6, t0, t6; \
+	vpsrldq $7, t0, t7; \
+	vpbroadcastb t0##_x, t0; \
+	vpbroadcastb t1##_x, t1; \
+	vpbroadcastb t2##_x, t2; \
+	vpbroadcastb t3##_x, t3; \
+	vpbroadcastb t4##_x, t4; \
+	vpbroadcastb t6##_x, t6; \
+	vpbroadcastb t5##_x, t5; \
+	vpbroadcastb t7##_x, t7; \
+	\
+	/* P-function */ \
+	vpxor x5, x0, x0; \
+	vpxor x6, x1, x1; \
+	vpxor x7, x2, x2; \
+	vpxor x4, x3, x3; \
+	\
+	vpxor x2, x4, x4; \
+	vpxor x3, x5, x5; \
+	vpxor x0, x6, x6; \
+	vpxor x1, x7, x7; \
+	\
+	vpxor x7, x0, x0; \
+	vpxor x4, x1, x1; \
+	vpxor x5, x2, x2; \
+	vpxor x6, x3, x3; \
+	\
+	vpxor x3, x4, x4; \
+	vpxor x0, x5, x5; \
+	vpxor x1, x6, x6; \
+	vpxor x2, x7, x7; /* note: high and low parts swapped */ \
+	\
+	/* Add key material and result to CD (x becomes new CD) */ \
+	\
+	vpxor t7, x0, x0; \
+	vpxor 4 * 32(mem_cd), x0, x0; \
+	\
+	vpxor t6, x1, x1; \
+	vpxor 5 * 32(mem_cd), x1, x1; \
+	\
+	vpxor t5, x2, x2; \
+	vpxor 6 * 32(mem_cd), x2, x2; \
+	\
+	vpxor t4, x3, x3; \
+	vpxor 7 * 32(mem_cd), x3, x3; \
+	\
+	vpxor t3, x4, x4; \
+	vpxor 0 * 32(mem_cd), x4, x4; \
+	\
+	vpxor t2, x5, x5; \
+	vpxor 1 * 32(mem_cd), x5, x5; \
+	\
+	vpxor t1, x6, x6; \
+	vpxor 2 * 32(mem_cd), x6, x6; \
+	\
+	vpxor t0, x7, x7; \
+	vpxor 3 * 32(mem_cd), x7, x7;
+
+/*
+ * Size optimization... with inlined roundsm16 binary would be over 5 times
+ * larger and would only marginally faster.
+ */
+.align 8
+roundsm32_x0_x1_x2_x3_x4_x5_x6_x7_y0_y1_y2_y3_y4_y5_y6_y7_cd:
+	roundsm32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		  %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14, %ymm15,
+		  %rcx, (%r9));
+	ret;
+ENDPROC(roundsm32_x0_x1_x2_x3_x4_x5_x6_x7_y0_y1_y2_y3_y4_y5_y6_y7_cd)
+
+.align 8
+roundsm32_x4_x5_x6_x7_x0_x1_x2_x3_y4_y5_y6_y7_y0_y1_y2_y3_ab:
+	roundsm32(%ymm4, %ymm5, %ymm6, %ymm7, %ymm0, %ymm1, %ymm2, %ymm3,
+		  %ymm12, %ymm13, %ymm14, %ymm15, %ymm8, %ymm9, %ymm10, %ymm11,
+		  %rax, (%r9));
+	ret;
+ENDPROC(roundsm32_x4_x5_x6_x7_x0_x1_x2_x3_y4_y5_y6_y7_y0_y1_y2_y3_ab)
+
+/*
+ * IN/OUT:
+ *  x0..x7: byte-sliced AB state preloaded
+ *  mem_ab: byte-sliced AB state in memory
+ *  mem_cb: byte-sliced CD state in memory
+ */
+#define two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, i, dir, store_ab) \
+	leaq (key_table + (i) * 8)(CTX), %r9; \
+	call roundsm32_x0_x1_x2_x3_x4_x5_x6_x7_y0_y1_y2_y3_y4_y5_y6_y7_cd; \
+	\
+	vmovdqu x0, 4 * 32(mem_cd); \
+	vmovdqu x1, 5 * 32(mem_cd); \
+	vmovdqu x2, 6 * 32(mem_cd); \
+	vmovdqu x3, 7 * 32(mem_cd); \
+	vmovdqu x4, 0 * 32(mem_cd); \
+	vmovdqu x5, 1 * 32(mem_cd); \
+	vmovdqu x6, 2 * 32(mem_cd); \
+	vmovdqu x7, 3 * 32(mem_cd); \
+	\
+	leaq (key_table + ((i) + (dir)) * 8)(CTX), %r9; \
+	call roundsm32_x4_x5_x6_x7_x0_x1_x2_x3_y4_y5_y6_y7_y0_y1_y2_y3_ab; \
+	\
+	store_ab(x0, x1, x2, x3, x4, x5, x6, x7, mem_ab);
+
+#define dummy_store(x0, x1, x2, x3, x4, x5, x6, x7, mem_ab) /* do nothing */
+
+#define store_ab_state(x0, x1, x2, x3, x4, x5, x6, x7, mem_ab) \
+	/* Store new AB state */ \
+	vmovdqu x4, 4 * 32(mem_ab); \
+	vmovdqu x5, 5 * 32(mem_ab); \
+	vmovdqu x6, 6 * 32(mem_ab); \
+	vmovdqu x7, 7 * 32(mem_ab); \
+	vmovdqu x0, 0 * 32(mem_ab); \
+	vmovdqu x1, 1 * 32(mem_ab); \
+	vmovdqu x2, 2 * 32(mem_ab); \
+	vmovdqu x3, 3 * 32(mem_ab);
+
+#define enc_rounds32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, i) \
+	two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, (i) + 2, 1, store_ab_state); \
+	two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, (i) + 4, 1, store_ab_state); \
+	two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, (i) + 6, 1, dummy_store);
+
+#define dec_rounds32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, i) \
+	two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, (i) + 7, -1, store_ab_state); \
+	two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, (i) + 5, -1, store_ab_state); \
+	two_roundsm32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd, (i) + 3, -1, dummy_store);
+
+/*
+ * IN:
+ *  v0..3: byte-sliced 32-bit integers
+ * OUT:
+ *  v0..3: (IN <<< 1)
+ */
+#define rol32_1_32(v0, v1, v2, v3, t0, t1, t2, zero) \
+	vpcmpgtb v0, zero, t0; \
+	vpaddb v0, v0, v0; \
+	vpabsb t0, t0; \
+	\
+	vpcmpgtb v1, zero, t1; \
+	vpaddb v1, v1, v1; \
+	vpabsb t1, t1; \
+	\
+	vpcmpgtb v2, zero, t2; \
+	vpaddb v2, v2, v2; \
+	vpabsb t2, t2; \
+	\
+	vpor t0, v1, v1; \
+	\
+	vpcmpgtb v3, zero, t0; \
+	vpaddb v3, v3, v3; \
+	vpabsb t0, t0; \
+	\
+	vpor t1, v2, v2; \
+	vpor t2, v3, v3; \
+	vpor t0, v0, v0;
+
+/*
+ * IN:
+ *   r: byte-sliced AB state in memory
+ *   l: byte-sliced CD state in memory
+ * OUT:
+ *   x0..x7: new byte-sliced CD state
+ */
+#define fls32(l, l0, l1, l2, l3, l4, l5, l6, l7, r, t0, t1, t2, t3, tt0, \
+	      tt1, tt2, tt3, kll, klr, krl, krr) \
+	/* \
+	 * t0 = kll; \
+	 * t0 &= ll; \
+	 * lr ^= rol32(t0, 1); \
+	 */ \
+	vpbroadcastd kll, t0; /* only lowest 32-bit used */ \
+	vpxor tt0, tt0, tt0; \
+	vpbroadcastb t0##_x, t3; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t2; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t1; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t0; \
+	\
+	vpand l0, t0, t0; \
+	vpand l1, t1, t1; \
+	vpand l2, t2, t2; \
+	vpand l3, t3, t3; \
+	\
+	rol32_1_32(t3, t2, t1, t0, tt1, tt2, tt3, tt0); \
+	\
+	vpxor l4, t0, l4; \
+	vmovdqu l4, 4 * 32(l); \
+	vpxor l5, t1, l5; \
+	vmovdqu l5, 5 * 32(l); \
+	vpxor l6, t2, l6; \
+	vmovdqu l6, 6 * 32(l); \
+	vpxor l7, t3, l7; \
+	vmovdqu l7, 7 * 32(l); \
+	\
+	/* \
+	 * t2 = krr; \
+	 * t2 |= rr; \
+	 * rl ^= t2; \
+	 */ \
+	\
+	vpbroadcastd krr, t0; /* only lowest 32-bit used */ \
+	vpbroadcastb t0##_x, t3; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t2; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t1; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t0; \
+	\
+	vpor 4 * 32(r), t0, t0; \
+	vpor 5 * 32(r), t1, t1; \
+	vpor 6 * 32(r), t2, t2; \
+	vpor 7 * 32(r), t3, t3; \
+	\
+	vpxor 0 * 32(r), t0, t0; \
+	vpxor 1 * 32(r), t1, t1; \
+	vpxor 2 * 32(r), t2, t2; \
+	vpxor 3 * 32(r), t3, t3; \
+	vmovdqu t0, 0 * 32(r); \
+	vmovdqu t1, 1 * 32(r); \
+	vmovdqu t2, 2 * 32(r); \
+	vmovdqu t3, 3 * 32(r); \
+	\
+	/* \
+	 * t2 = krl; \
+	 * t2 &= rl; \
+	 * rr ^= rol32(t2, 1); \
+	 */ \
+	vpbroadcastd krl, t0; /* only lowest 32-bit used */ \
+	vpbroadcastb t0##_x, t3; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t2; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t1; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t0; \
+	\
+	vpand 0 * 32(r), t0, t0; \
+	vpand 1 * 32(r), t1, t1; \
+	vpand 2 * 32(r), t2, t2; \
+	vpand 3 * 32(r), t3, t3; \
+	\
+	rol32_1_32(t3, t2, t1, t0, tt1, tt2, tt3, tt0); \
+	\
+	vpxor 4 * 32(r), t0, t0; \
+	vpxor 5 * 32(r), t1, t1; \
+	vpxor 6 * 32(r), t2, t2; \
+	vpxor 7 * 32(r), t3, t3; \
+	vmovdqu t0, 4 * 32(r); \
+	vmovdqu t1, 5 * 32(r); \
+	vmovdqu t2, 6 * 32(r); \
+	vmovdqu t3, 7 * 32(r); \
+	\
+	/* \
+	 * t0 = klr; \
+	 * t0 |= lr; \
+	 * ll ^= t0; \
+	 */ \
+	\
+	vpbroadcastd klr, t0; /* only lowest 32-bit used */ \
+	vpbroadcastb t0##_x, t3; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t2; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t1; \
+	vpsrldq $1, t0, t0; \
+	vpbroadcastb t0##_x, t0; \
+	\
+	vpor l4, t0, t0; \
+	vpor l5, t1, t1; \
+	vpor l6, t2, t2; \
+	vpor l7, t3, t3; \
+	\
+	vpxor l0, t0, l0; \
+	vmovdqu l0, 0 * 32(l); \
+	vpxor l1, t1, l1; \
+	vmovdqu l1, 1 * 32(l); \
+	vpxor l2, t2, l2; \
+	vmovdqu l2, 2 * 32(l); \
+	vpxor l3, t3, l3; \
+	vmovdqu l3, 3 * 32(l);
+
+#define transpose_4x4(x0, x1, x2, x3, t1, t2) \
+	vpunpckhdq x1, x0, t2; \
+	vpunpckldq x1, x0, x0; \
+	\
+	vpunpckldq x3, x2, t1; \
+	vpunpckhdq x3, x2, x2; \
+	\
+	vpunpckhqdq t1, x0, x1; \
+	vpunpcklqdq t1, x0, x0; \
+	\
+	vpunpckhqdq x2, t2, x3; \
+	vpunpcklqdq x2, t2, x2;
+
+#define byteslice_16x16b_fast(a0, b0, c0, d0, a1, b1, c1, d1, a2, b2, c2, d2, \
+			      a3, b3, c3, d3, st0, st1) \
+	vmovdqu d2, st0; \
+	vmovdqu d3, st1; \
+	transpose_4x4(a0, a1, a2, a3, d2, d3); \
+	transpose_4x4(b0, b1, b2, b3, d2, d3); \
+	vmovdqu st0, d2; \
+	vmovdqu st1, d3; \
+	\
+	vmovdqu a0, st0; \
+	vmovdqu a1, st1; \
+	transpose_4x4(c0, c1, c2, c3, a0, a1); \
+	transpose_4x4(d0, d1, d2, d3, a0, a1); \
+	\
+	vbroadcasti128 .Lshufb_16x16b, a0; \
+	vmovdqu st1, a1; \
+	vpshufb a0, a2, a2; \
+	vpshufb a0, a3, a3; \
+	vpshufb a0, b0, b0; \
+	vpshufb a0, b1, b1; \
+	vpshufb a0, b2, b2; \
+	vpshufb a0, b3, b3; \
+	vpshufb a0, a1, a1; \
+	vpshufb a0, c0, c0; \
+	vpshufb a0, c1, c1; \
+	vpshufb a0, c2, c2; \
+	vpshufb a0, c3, c3; \
+	vpshufb a0, d0, d0; \
+	vpshufb a0, d1, d1; \
+	vpshufb a0, d2, d2; \
+	vpshufb a0, d3, d3; \
+	vmovdqu d3, st1; \
+	vmovdqu st0, d3; \
+	vpshufb a0, d3, a0; \
+	vmovdqu d2, st0; \
+	\
+	transpose_4x4(a0, b0, c0, d0, d2, d3); \
+	transpose_4x4(a1, b1, c1, d1, d2, d3); \
+	vmovdqu st0, d2; \
+	vmovdqu st1, d3; \
+	\
+	vmovdqu b0, st0; \
+	vmovdqu b1, st1; \
+	transpose_4x4(a2, b2, c2, d2, b0, b1); \
+	transpose_4x4(a3, b3, c3, d3, b0, b1); \
+	vmovdqu st0, b0; \
+	vmovdqu st1, b1; \
+	/* does not adjust output bytes inside vectors */
+
+/* load blocks to registers and apply pre-whitening */
+#define inpack32_pre(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		     y6, y7, rio, key) \
+	vpbroadcastq key, x0; \
+	vpshufb .Lpack_bswap, x0, x0; \
+	\
+	vpxor 0 * 32(rio), x0, y7; \
+	vpxor 1 * 32(rio), x0, y6; \
+	vpxor 2 * 32(rio), x0, y5; \
+	vpxor 3 * 32(rio), x0, y4; \
+	vpxor 4 * 32(rio), x0, y3; \
+	vpxor 5 * 32(rio), x0, y2; \
+	vpxor 6 * 32(rio), x0, y1; \
+	vpxor 7 * 32(rio), x0, y0; \
+	vpxor 8 * 32(rio), x0, x7; \
+	vpxor 9 * 32(rio), x0, x6; \
+	vpxor 10 * 32(rio), x0, x5; \
+	vpxor 11 * 32(rio), x0, x4; \
+	vpxor 12 * 32(rio), x0, x3; \
+	vpxor 13 * 32(rio), x0, x2; \
+	vpxor 14 * 32(rio), x0, x1; \
+	vpxor 15 * 32(rio), x0, x0;
+
+/* byteslice pre-whitened blocks and store to temporary memory */
+#define inpack32_post(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		      y6, y7, mem_ab, mem_cd) \
+	byteslice_16x16b_fast(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, \
+			      y4, y5, y6, y7, (mem_ab), (mem_cd)); \
+	\
+	vmovdqu x0, 0 * 32(mem_ab); \
+	vmovdqu x1, 1 * 32(mem_ab); \
+	vmovdqu x2, 2 * 32(mem_ab); \
+	vmovdqu x3, 3 * 32(mem_ab); \
+	vmovdqu x4, 4 * 32(mem_ab); \
+	vmovdqu x5, 5 * 32(mem_ab); \
+	vmovdqu x6, 6 * 32(mem_ab); \
+	vmovdqu x7, 7 * 32(mem_ab); \
+	vmovdqu y0, 0 * 32(mem_cd); \
+	vmovdqu y1, 1 * 32(mem_cd); \
+	vmovdqu y2, 2 * 32(mem_cd); \
+	vmovdqu y3, 3 * 32(mem_cd); \
+	vmovdqu y4, 4 * 32(mem_cd); \
+	vmovdqu y5, 5 * 32(mem_cd); \
+	vmovdqu y6, 6 * 32(mem_cd); \
+	vmovdqu y7, 7 * 32(mem_cd);
+
+/* de-byteslice, apply post-whitening and store blocks */
+#define outunpack32(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, \
+		    y5, y6, y7, key, stack_tmp0, stack_tmp1) \
+	byteslice_16x16b_fast(y0, y4, x0, x4, y1, y5, x1, x5, y2, y6, x2, x6, \
+			      y3, y7, x3, x7, stack_tmp0, stack_tmp1); \
+	\
+	vmovdqu x0, stack_tmp0; \
+	\
+	vpbroadcastq key, x0; \
+	vpshufb .Lpack_bswap, x0, x0; \
+	\
+	vpxor x0, y7, y7; \
+	vpxor x0, y6, y6; \
+	vpxor x0, y5, y5; \
+	vpxor x0, y4, y4; \
+	vpxor x0, y3, y3; \
+	vpxor x0, y2, y2; \
+	vpxor x0, y1, y1; \
+	vpxor x0, y0, y0; \
+	vpxor x0, x7, x7; \
+	vpxor x0, x6, x6; \
+	vpxor x0, x5, x5; \
+	vpxor x0, x4, x4; \
+	vpxor x0, x3, x3; \
+	vpxor x0, x2, x2; \
+	vpxor x0, x1, x1; \
+	vpxor stack_tmp0, x0, x0;
+
+#define write_output(x0, x1, x2, x3, x4, x5, x6, x7, y0, y1, y2, y3, y4, y5, \
+		     y6, y7, rio) \
+	vmovdqu x0, 0 * 32(rio); \
+	vmovdqu x1, 1 * 32(rio); \
+	vmovdqu x2, 2 * 32(rio); \
+	vmovdqu x3, 3 * 32(rio); \
+	vmovdqu x4, 4 * 32(rio); \
+	vmovdqu x5, 5 * 32(rio); \
+	vmovdqu x6, 6 * 32(rio); \
+	vmovdqu x7, 7 * 32(rio); \
+	vmovdqu y0, 8 * 32(rio); \
+	vmovdqu y1, 9 * 32(rio); \
+	vmovdqu y2, 10 * 32(rio); \
+	vmovdqu y3, 11 * 32(rio); \
+	vmovdqu y4, 12 * 32(rio); \
+	vmovdqu y5, 13 * 32(rio); \
+	vmovdqu y6, 14 * 32(rio); \
+	vmovdqu y7, 15 * 32(rio);
+
+.data
+.align 32
+
+#define SHUFB_BYTES(idx) \
+	0 + (idx), 4 + (idx), 8 + (idx), 12 + (idx)
+
+.Lshufb_16x16b:
+	.byte SHUFB_BYTES(0), SHUFB_BYTES(1), SHUFB_BYTES(2), SHUFB_BYTES(3)
+	.byte SHUFB_BYTES(0), SHUFB_BYTES(1), SHUFB_BYTES(2), SHUFB_BYTES(3)
+
+.Lpack_bswap:
+	.long 0x00010203, 0x04050607, 0x80808080, 0x80808080
+	.long 0x00010203, 0x04050607, 0x80808080, 0x80808080
+
+/* For CTR-mode IV byteswap */
+.Lbswap128_mask:
+	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
+
+/* For XTS mode */
+.Lxts_gf128mul_and_shl1_mask_0:
+	.byte 0x87, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
+.Lxts_gf128mul_and_shl1_mask_1:
+	.byte 0x0e, 1, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0
+
+/*
+ * pre-SubByte transform
+ *
+ * pre-lookup for sbox1, sbox2, sbox3:
+ *   swap_bitendianness(
+ *       isom_map_camellia_to_aes(
+ *           camellia_f(
+ *               swap_bitendianess(in)
+ *           )
+ *       )
+ *   )
+ *
+ * (note: '⊕ 0xc5' inside camellia_f())
+ */
+.Lpre_tf_lo_s1:
+	.byte 0x45, 0xe8, 0x40, 0xed, 0x2e, 0x83, 0x2b, 0x86
+	.byte 0x4b, 0xe6, 0x4e, 0xe3, 0x20, 0x8d, 0x25, 0x88
+.Lpre_tf_hi_s1:
+	.byte 0x00, 0x51, 0xf1, 0xa0, 0x8a, 0xdb, 0x7b, 0x2a
+	.byte 0x09, 0x58, 0xf8, 0xa9, 0x83, 0xd2, 0x72, 0x23
+
+/*
+ * pre-SubByte transform
+ *
+ * pre-lookup for sbox4:
+ *   swap_bitendianness(
+ *       isom_map_camellia_to_aes(
+ *           camellia_f(
+ *               swap_bitendianess(in <<< 1)
+ *           )
+ *       )
+ *   )
+ *
+ * (note: '⊕ 0xc5' inside camellia_f())
+ */
+.Lpre_tf_lo_s4:
+	.byte 0x45, 0x40, 0x2e, 0x2b, 0x4b, 0x4e, 0x20, 0x25
+	.byte 0x14, 0x11, 0x7f, 0x7a, 0x1a, 0x1f, 0x71, 0x74
+.Lpre_tf_hi_s4:
+	.byte 0x00, 0xf1, 0x8a, 0x7b, 0x09, 0xf8, 0x83, 0x72
+	.byte 0xad, 0x5c, 0x27, 0xd6, 0xa4, 0x55, 0x2e, 0xdf
+
+/*
+ * post-SubByte transform
+ *
+ * post-lookup for sbox1, sbox4:
+ *  swap_bitendianness(
+ *      camellia_h(
+ *          isom_map_aes_to_camellia(
+ *              swap_bitendianness(
+ *                  aes_inverse_affine_transform(in)
+ *              )
+ *          )
+ *      )
+ *  )
+ *
+ * (note: '⊕ 0x6e' inside camellia_h())
+ */
+.Lpost_tf_lo_s1:
+	.byte 0x3c, 0xcc, 0xcf, 0x3f, 0x32, 0xc2, 0xc1, 0x31
+	.byte 0xdc, 0x2c, 0x2f, 0xdf, 0xd2, 0x22, 0x21, 0xd1
+.Lpost_tf_hi_s1:
+	.byte 0x00, 0xf9, 0x86, 0x7f, 0xd7, 0x2e, 0x51, 0xa8
+	.byte 0xa4, 0x5d, 0x22, 0xdb, 0x73, 0x8a, 0xf5, 0x0c
+
+/*
+ * post-SubByte transform
+ *
+ * post-lookup for sbox2:
+ *  swap_bitendianness(
+ *      camellia_h(
+ *          isom_map_aes_to_camellia(
+ *              swap_bitendianness(
+ *                  aes_inverse_affine_transform(in)
+ *              )
+ *          )
+ *      )
+ *  ) <<< 1
+ *
+ * (note: '⊕ 0x6e' inside camellia_h())
+ */
+.Lpost_tf_lo_s2:
+	.byte 0x78, 0x99, 0x9f, 0x7e, 0x64, 0x85, 0x83, 0x62
+	.byte 0xb9, 0x58, 0x5e, 0xbf, 0xa5, 0x44, 0x42, 0xa3
+.Lpost_tf_hi_s2:
+	.byte 0x00, 0xf3, 0x0d, 0xfe, 0xaf, 0x5c, 0xa2, 0x51
+	.byte 0x49, 0xba, 0x44, 0xb7, 0xe6, 0x15, 0xeb, 0x18
+
+/*
+ * post-SubByte transform
+ *
+ * post-lookup for sbox3:
+ *  swap_bitendianness(
+ *      camellia_h(
+ *          isom_map_aes_to_camellia(
+ *              swap_bitendianness(
+ *                  aes_inverse_affine_transform(in)
+ *              )
+ *          )
+ *      )
+ *  ) >>> 1
+ *
+ * (note: '⊕ 0x6e' inside camellia_h())
+ */
+.Lpost_tf_lo_s3:
+	.byte 0x1e, 0x66, 0xe7, 0x9f, 0x19, 0x61, 0xe0, 0x98
+	.byte 0x6e, 0x16, 0x97, 0xef, 0x69, 0x11, 0x90, 0xe8
+.Lpost_tf_hi_s3:
+	.byte 0x00, 0xfc, 0x43, 0xbf, 0xeb, 0x17, 0xa8, 0x54
+	.byte 0x52, 0xae, 0x11, 0xed, 0xb9, 0x45, 0xfa, 0x06
+
+/* For isolating SubBytes from AESENCLAST, inverse shift row */
+.Linv_shift_row:
+	.byte 0x00, 0x0d, 0x0a, 0x07, 0x04, 0x01, 0x0e, 0x0b
+	.byte 0x08, 0x05, 0x02, 0x0f, 0x0c, 0x09, 0x06, 0x03
+
+.align 4
+/* 4-bit mask */
+.L0f0f0f0f:
+	.long 0x0f0f0f0f
+
+.text
+
+.align 8
+__camellia_enc_blk32:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rax: temporary storage, 512 bytes
+	 *	%ymm0..%ymm15: 32 plaintext blocks
+	 * output:
+	 *	%ymm0..%ymm15: 32 encrypted blocks, order swapped:
+	 *       7, 8, 6, 5, 4, 3, 2, 1, 0, 15, 14, 13, 12, 11, 10, 9, 8
+	 */
+
+	leaq 8 * 32(%rax), %rcx;
+
+	inpack32_post(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		      %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		      %ymm15, %rax, %rcx);
+
+	enc_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rax, %rcx, 0);
+
+	fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+	      %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+	      %ymm15,
+	      ((key_table + (8) * 8) + 0)(CTX),
+	      ((key_table + (8) * 8) + 4)(CTX),
+	      ((key_table + (8) * 8) + 8)(CTX),
+	      ((key_table + (8) * 8) + 12)(CTX));
+
+	enc_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rax, %rcx, 8);
+
+	fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+	      %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+	      %ymm15,
+	      ((key_table + (16) * 8) + 0)(CTX),
+	      ((key_table + (16) * 8) + 4)(CTX),
+	      ((key_table + (16) * 8) + 8)(CTX),
+	      ((key_table + (16) * 8) + 12)(CTX));
+
+	enc_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rax, %rcx, 16);
+
+	movl $24, %r8d;
+	cmpl $16, key_length(CTX);
+	jne .Lenc_max32;
+
+.Lenc_done:
+	/* load CD for output */
+	vmovdqu 0 * 32(%rcx), %ymm8;
+	vmovdqu 1 * 32(%rcx), %ymm9;
+	vmovdqu 2 * 32(%rcx), %ymm10;
+	vmovdqu 3 * 32(%rcx), %ymm11;
+	vmovdqu 4 * 32(%rcx), %ymm12;
+	vmovdqu 5 * 32(%rcx), %ymm13;
+	vmovdqu 6 * 32(%rcx), %ymm14;
+	vmovdqu 7 * 32(%rcx), %ymm15;
+
+	outunpack32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		    %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		    %ymm15, (key_table)(CTX, %r8, 8), (%rax), 1 * 32(%rax));
+
+	ret;
+
+.align 8
+.Lenc_max32:
+	movl $32, %r8d;
+
+	fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+	      %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+	      %ymm15,
+	      ((key_table + (24) * 8) + 0)(CTX),
+	      ((key_table + (24) * 8) + 4)(CTX),
+	      ((key_table + (24) * 8) + 8)(CTX),
+	      ((key_table + (24) * 8) + 12)(CTX));
+
+	enc_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rax, %rcx, 24);
+
+	jmp .Lenc_done;
+ENDPROC(__camellia_enc_blk32)
+
+.align 8
+__camellia_dec_blk32:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rax: temporary storage, 512 bytes
+	 *	%r8d: 24 for 16 byte key, 32 for larger
+	 *	%ymm0..%ymm15: 16 encrypted blocks
+	 * output:
+	 *	%ymm0..%ymm15: 16 plaintext blocks, order swapped:
+	 *       7, 8, 6, 5, 4, 3, 2, 1, 0, 15, 14, 13, 12, 11, 10, 9, 8
+	 */
+
+	leaq 8 * 32(%rax), %rcx;
+
+	inpack32_post(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		      %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		      %ymm15, %rax, %rcx);
+
+	cmpl $32, %r8d;
+	je .Ldec_max32;
+
+.Ldec_max24:
+	dec_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rax, %rcx, 16);
+
+	fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+	      %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+	      %ymm15,
+	      ((key_table + (16) * 8) + 8)(CTX),
+	      ((key_table + (16) * 8) + 12)(CTX),
+	      ((key_table + (16) * 8) + 0)(CTX),
+	      ((key_table + (16) * 8) + 4)(CTX));
+
+	dec_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rax, %rcx, 8);
+
+	fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+	      %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+	      %ymm15,
+	      ((key_table + (8) * 8) + 8)(CTX),
+	      ((key_table + (8) * 8) + 12)(CTX),
+	      ((key_table + (8) * 8) + 0)(CTX),
+	      ((key_table + (8) * 8) + 4)(CTX));
+
+	dec_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rax, %rcx, 0);
+
+	/* load CD for output */
+	vmovdqu 0 * 32(%rcx), %ymm8;
+	vmovdqu 1 * 32(%rcx), %ymm9;
+	vmovdqu 2 * 32(%rcx), %ymm10;
+	vmovdqu 3 * 32(%rcx), %ymm11;
+	vmovdqu 4 * 32(%rcx), %ymm12;
+	vmovdqu 5 * 32(%rcx), %ymm13;
+	vmovdqu 6 * 32(%rcx), %ymm14;
+	vmovdqu 7 * 32(%rcx), %ymm15;
+
+	outunpack32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		    %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		    %ymm15, (key_table)(CTX), (%rax), 1 * 32(%rax));
+
+	ret;
+
+.align 8
+.Ldec_max32:
+	dec_rounds32(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rax, %rcx, 24);
+
+	fls32(%rax, %ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+	      %rcx, %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+	      %ymm15,
+	      ((key_table + (24) * 8) + 8)(CTX),
+	      ((key_table + (24) * 8) + 12)(CTX),
+	      ((key_table + (24) * 8) + 0)(CTX),
+	      ((key_table + (24) * 8) + 4)(CTX));
+
+	jmp .Ldec_max24;
+ENDPROC(__camellia_dec_blk32)
+
+ENTRY(camellia_ecb_enc_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (32 blocks)
+	 *	%rdx: src (32 blocks)
+	 */
+
+	vzeroupper;
+
+	inpack32_pre(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rdx, (key_table)(CTX));
+
+	/* now dst can be used as temporary buffer (even in src == dst case) */
+	movq	%rsi, %rax;
+
+	call __camellia_enc_blk32;
+
+	write_output(%ymm7, %ymm6, %ymm5, %ymm4, %ymm3, %ymm2, %ymm1, %ymm0,
+		     %ymm15, %ymm14, %ymm13, %ymm12, %ymm11, %ymm10, %ymm9,
+		     %ymm8, %rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(camellia_ecb_enc_32way)
+
+ENTRY(camellia_ecb_dec_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (32 blocks)
+	 *	%rdx: src (32 blocks)
+	 */
+
+	vzeroupper;
+
+	cmpl $16, key_length(CTX);
+	movl $32, %r8d;
+	movl $24, %eax;
+	cmovel %eax, %r8d; /* max */
+
+	inpack32_pre(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rdx, (key_table)(CTX, %r8, 8));
+
+	/* now dst can be used as temporary buffer (even in src == dst case) */
+	movq	%rsi, %rax;
+
+	call __camellia_dec_blk32;
+
+	write_output(%ymm7, %ymm6, %ymm5, %ymm4, %ymm3, %ymm2, %ymm1, %ymm0,
+		     %ymm15, %ymm14, %ymm13, %ymm12, %ymm11, %ymm10, %ymm9,
+		     %ymm8, %rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(camellia_ecb_dec_32way)
+
+ENTRY(camellia_cbc_dec_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (32 blocks)
+	 *	%rdx: src (32 blocks)
+	 */
+
+	vzeroupper;
+
+	cmpl $16, key_length(CTX);
+	movl $32, %r8d;
+	movl $24, %eax;
+	cmovel %eax, %r8d; /* max */
+
+	inpack32_pre(%ymm0, %ymm1, %ymm2, %ymm3, %ymm4, %ymm5, %ymm6, %ymm7,
+		     %ymm8, %ymm9, %ymm10, %ymm11, %ymm12, %ymm13, %ymm14,
+		     %ymm15, %rdx, (key_table)(CTX, %r8, 8));
+
+	movq %rsp, %r10;
+	cmpq %rsi, %rdx;
+	je .Lcbc_dec_use_stack;
+
+	/* dst can be used as temporary storage, src is not overwritten. */
+	movq %rsi, %rax;
+	jmp .Lcbc_dec_continue;
+
+.Lcbc_dec_use_stack:
+	/*
+	 * dst still in-use (because dst == src), so use stack for temporary
+	 * storage.
+	 */
+	subq $(16 * 32), %rsp;
+	movq %rsp, %rax;
+
+.Lcbc_dec_continue:
+	call __camellia_dec_blk32;
+
+	vmovdqu %ymm7, (%rax);
+	vpxor %ymm7, %ymm7, %ymm7;
+	vinserti128 $1, (%rdx), %ymm7, %ymm7;
+	vpxor (%rax), %ymm7, %ymm7;
+	movq %r10, %rsp;
+	vpxor (0 * 32 + 16)(%rdx), %ymm6, %ymm6;
+	vpxor (1 * 32 + 16)(%rdx), %ymm5, %ymm5;
+	vpxor (2 * 32 + 16)(%rdx), %ymm4, %ymm4;
+	vpxor (3 * 32 + 16)(%rdx), %ymm3, %ymm3;
+	vpxor (4 * 32 + 16)(%rdx), %ymm2, %ymm2;
+	vpxor (5 * 32 + 16)(%rdx), %ymm1, %ymm1;
+	vpxor (6 * 32 + 16)(%rdx), %ymm0, %ymm0;
+	vpxor (7 * 32 + 16)(%rdx), %ymm15, %ymm15;
+	vpxor (8 * 32 + 16)(%rdx), %ymm14, %ymm14;
+	vpxor (9 * 32 + 16)(%rdx), %ymm13, %ymm13;
+	vpxor (10 * 32 + 16)(%rdx), %ymm12, %ymm12;
+	vpxor (11 * 32 + 16)(%rdx), %ymm11, %ymm11;
+	vpxor (12 * 32 + 16)(%rdx), %ymm10, %ymm10;
+	vpxor (13 * 32 + 16)(%rdx), %ymm9, %ymm9;
+	vpxor (14 * 32 + 16)(%rdx), %ymm8, %ymm8;
+	write_output(%ymm7, %ymm6, %ymm5, %ymm4, %ymm3, %ymm2, %ymm1, %ymm0,
+		     %ymm15, %ymm14, %ymm13, %ymm12, %ymm11, %ymm10, %ymm9,
+		     %ymm8, %rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(camellia_cbc_dec_32way)
+
+#define inc_le128(x, minus_one, tmp) \
+	vpcmpeqq minus_one, x, tmp; \
+	vpsubq minus_one, x, x; \
+	vpslldq $8, tmp, tmp; \
+	vpsubq tmp, x, x;
+
+#define add2_le128(x, minus_one, minus_two, tmp1, tmp2) \
+	vpcmpeqq minus_one, x, tmp1; \
+	vpcmpeqq minus_two, x, tmp2; \
+	vpsubq minus_two, x, x; \
+	vpor tmp2, tmp1, tmp1; \
+	vpslldq $8, tmp1, tmp1; \
+	vpsubq tmp1, x, x;
+
+ENTRY(camellia_ctr_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (32 blocks)
+	 *	%rdx: src (32 blocks)
+	 *	%rcx: iv (little endian, 128bit)
+	 */
+
+	vzeroupper;
+
+	movq %rsp, %r10;
+	cmpq %rsi, %rdx;
+	je .Lctr_use_stack;
+
+	/* dst can be used as temporary storage, src is not overwritten. */
+	movq %rsi, %rax;
+	jmp .Lctr_continue;
+
+.Lctr_use_stack:
+	subq $(16 * 32), %rsp;
+	movq %rsp, %rax;
+
+.Lctr_continue:
+	vpcmpeqd %ymm15, %ymm15, %ymm15;
+	vpsrldq $8, %ymm15, %ymm15; /* ab: -1:0 ; cd: -1:0 */
+	vpaddq %ymm15, %ymm15, %ymm12; /* ab: -2:0 ; cd: -2:0 */
+
+	/* load IV and byteswap */
+	vmovdqu (%rcx), %xmm0;
+	vmovdqa %xmm0, %xmm1;
+	inc_le128(%xmm0, %xmm15, %xmm14);
+	vbroadcasti128 .Lbswap128_mask, %ymm14;
+	vinserti128 $1, %xmm0, %ymm1, %ymm0;
+	vpshufb %ymm14, %ymm0, %ymm13;
+	vmovdqu %ymm13, 15 * 32(%rax);
+
+	/* construct IVs */
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13); /* ab:le2 ; cd:le3 */
+	vpshufb %ymm14, %ymm0, %ymm13;
+	vmovdqu %ymm13, 14 * 32(%rax);
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm13;
+	vmovdqu %ymm13, 13 * 32(%rax);
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm13;
+	vmovdqu %ymm13, 12 * 32(%rax);
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm13;
+	vmovdqu %ymm13, 11 * 32(%rax);
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm10;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm9;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm8;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm7;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm6;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm5;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm4;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm3;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm2;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vpshufb %ymm14, %ymm0, %ymm1;
+	add2_le128(%ymm0, %ymm15, %ymm12, %ymm11, %ymm13);
+	vextracti128 $1, %ymm0, %xmm13;
+	vpshufb %ymm14, %ymm0, %ymm0;
+	inc_le128(%xmm13, %xmm15, %xmm14);
+	vmovdqu %xmm13, (%rcx);
+
+	/* inpack32_pre: */
+	vpbroadcastq (key_table)(CTX), %ymm15;
+	vpshufb .Lpack_bswap, %ymm15, %ymm15;
+	vpxor %ymm0, %ymm15, %ymm0;
+	vpxor %ymm1, %ymm15, %ymm1;
+	vpxor %ymm2, %ymm15, %ymm2;
+	vpxor %ymm3, %ymm15, %ymm3;
+	vpxor %ymm4, %ymm15, %ymm4;
+	vpxor %ymm5, %ymm15, %ymm5;
+	vpxor %ymm6, %ymm15, %ymm6;
+	vpxor %ymm7, %ymm15, %ymm7;
+	vpxor %ymm8, %ymm15, %ymm8;
+	vpxor %ymm9, %ymm15, %ymm9;
+	vpxor %ymm10, %ymm15, %ymm10;
+	vpxor 11 * 32(%rax), %ymm15, %ymm11;
+	vpxor 12 * 32(%rax), %ymm15, %ymm12;
+	vpxor 13 * 32(%rax), %ymm15, %ymm13;
+	vpxor 14 * 32(%rax), %ymm15, %ymm14;
+	vpxor 15 * 32(%rax), %ymm15, %ymm15;
+
+	call __camellia_enc_blk32;
+
+	movq %r10, %rsp;
+
+	vpxor 0 * 32(%rdx), %ymm7, %ymm7;
+	vpxor 1 * 32(%rdx), %ymm6, %ymm6;
+	vpxor 2 * 32(%rdx), %ymm5, %ymm5;
+	vpxor 3 * 32(%rdx), %ymm4, %ymm4;
+	vpxor 4 * 32(%rdx), %ymm3, %ymm3;
+	vpxor 5 * 32(%rdx), %ymm2, %ymm2;
+	vpxor 6 * 32(%rdx), %ymm1, %ymm1;
+	vpxor 7 * 32(%rdx), %ymm0, %ymm0;
+	vpxor 8 * 32(%rdx), %ymm15, %ymm15;
+	vpxor 9 * 32(%rdx), %ymm14, %ymm14;
+	vpxor 10 * 32(%rdx), %ymm13, %ymm13;
+	vpxor 11 * 32(%rdx), %ymm12, %ymm12;
+	vpxor 12 * 32(%rdx), %ymm11, %ymm11;
+	vpxor 13 * 32(%rdx), %ymm10, %ymm10;
+	vpxor 14 * 32(%rdx), %ymm9, %ymm9;
+	vpxor 15 * 32(%rdx), %ymm8, %ymm8;
+	write_output(%ymm7, %ymm6, %ymm5, %ymm4, %ymm3, %ymm2, %ymm1, %ymm0,
+		     %ymm15, %ymm14, %ymm13, %ymm12, %ymm11, %ymm10, %ymm9,
+		     %ymm8, %rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(camellia_ctr_32way)
+
+#define gf128mul_x_ble(iv, mask, tmp) \
+	vpsrad $31, iv, tmp; \
+	vpaddq iv, iv, iv; \
+	vpshufd $0x13, tmp, tmp; \
+	vpand mask, tmp, tmp; \
+	vpxor tmp, iv, iv;
+
+#define gf128mul_x2_ble(iv, mask1, mask2, tmp0, tmp1) \
+	vpsrad $31, iv, tmp0; \
+	vpaddq iv, iv, tmp1; \
+	vpsllq $2, iv, iv; \
+	vpshufd $0x13, tmp0, tmp0; \
+	vpsrad $31, tmp1, tmp1; \
+	vpand mask2, tmp0, tmp0; \
+	vpshufd $0x13, tmp1, tmp1; \
+	vpxor tmp0, iv, iv; \
+	vpand mask1, tmp1, tmp1; \
+	vpxor tmp1, iv, iv;
+
+.align 8
+camellia_xts_crypt_32way:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (32 blocks)
+	 *	%rdx: src (32 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 *	%r8: index for input whitening key
+	 *	%r9: pointer to  __camellia_enc_blk32 or __camellia_dec_blk32
+	 */
+
+	vzeroupper;
+
+	subq $(16 * 32), %rsp;
+	movq %rsp, %rax;
+
+	vbroadcasti128 .Lxts_gf128mul_and_shl1_mask_0, %ymm12;
+
+	/* load IV and construct second IV */
+	vmovdqu (%rcx), %xmm0;
+	vmovdqa %xmm0, %xmm15;
+	gf128mul_x_ble(%xmm0, %xmm12, %xmm13);
+	vbroadcasti128 .Lxts_gf128mul_and_shl1_mask_1, %ymm13;
+	vinserti128 $1, %xmm0, %ymm15, %ymm0;
+	vpxor 0 * 32(%rdx), %ymm0, %ymm15;
+	vmovdqu %ymm15, 15 * 32(%rax);
+	vmovdqu %ymm0, 0 * 32(%rsi);
+
+	/* construct IVs */
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 1 * 32(%rdx), %ymm0, %ymm15;
+	vmovdqu %ymm15, 14 * 32(%rax);
+	vmovdqu %ymm0, 1 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 2 * 32(%rdx), %ymm0, %ymm15;
+	vmovdqu %ymm15, 13 * 32(%rax);
+	vmovdqu %ymm0, 2 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 3 * 32(%rdx), %ymm0, %ymm15;
+	vmovdqu %ymm15, 12 * 32(%rax);
+	vmovdqu %ymm0, 3 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 4 * 32(%rdx), %ymm0, %ymm11;
+	vmovdqu %ymm0, 4 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 5 * 32(%rdx), %ymm0, %ymm10;
+	vmovdqu %ymm0, 5 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 6 * 32(%rdx), %ymm0, %ymm9;
+	vmovdqu %ymm0, 6 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 7 * 32(%rdx), %ymm0, %ymm8;
+	vmovdqu %ymm0, 7 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 8 * 32(%rdx), %ymm0, %ymm7;
+	vmovdqu %ymm0, 8 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 9 * 32(%rdx), %ymm0, %ymm6;
+	vmovdqu %ymm0, 9 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 10 * 32(%rdx), %ymm0, %ymm5;
+	vmovdqu %ymm0, 10 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 11 * 32(%rdx), %ymm0, %ymm4;
+	vmovdqu %ymm0, 11 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 12 * 32(%rdx), %ymm0, %ymm3;
+	vmovdqu %ymm0, 12 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 13 * 32(%rdx), %ymm0, %ymm2;
+	vmovdqu %ymm0, 13 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 14 * 32(%rdx), %ymm0, %ymm1;
+	vmovdqu %ymm0, 14 * 32(%rsi);
+
+	gf128mul_x2_ble(%ymm0, %ymm12, %ymm13, %ymm14, %ymm15);
+	vpxor 15 * 32(%rdx), %ymm0, %ymm15;
+	vmovdqu %ymm15, 0 * 32(%rax);
+	vmovdqu %ymm0, 15 * 32(%rsi);
+
+	vextracti128 $1, %ymm0, %xmm0;
+	gf128mul_x_ble(%xmm0, %xmm12, %xmm15);
+	vmovdqu %xmm0, (%rcx);
+
+	/* inpack32_pre: */
+	vpbroadcastq (key_table)(CTX, %r8, 8), %ymm15;
+	vpshufb .Lpack_bswap, %ymm15, %ymm15;
+	vpxor 0 * 32(%rax), %ymm15, %ymm0;
+	vpxor %ymm1, %ymm15, %ymm1;
+	vpxor %ymm2, %ymm15, %ymm2;
+	vpxor %ymm3, %ymm15, %ymm3;
+	vpxor %ymm4, %ymm15, %ymm4;
+	vpxor %ymm5, %ymm15, %ymm5;
+	vpxor %ymm6, %ymm15, %ymm6;
+	vpxor %ymm7, %ymm15, %ymm7;
+	vpxor %ymm8, %ymm15, %ymm8;
+	vpxor %ymm9, %ymm15, %ymm9;
+	vpxor %ymm10, %ymm15, %ymm10;
+	vpxor %ymm11, %ymm15, %ymm11;
+	vpxor 12 * 32(%rax), %ymm15, %ymm12;
+	vpxor 13 * 32(%rax), %ymm15, %ymm13;
+	vpxor 14 * 32(%rax), %ymm15, %ymm14;
+	vpxor 15 * 32(%rax), %ymm15, %ymm15;
+
+	call *%r9;
+
+	addq $(16 * 32), %rsp;
+
+	vpxor 0 * 32(%rsi), %ymm7, %ymm7;
+	vpxor 1 * 32(%rsi), %ymm6, %ymm6;
+	vpxor 2 * 32(%rsi), %ymm5, %ymm5;
+	vpxor 3 * 32(%rsi), %ymm4, %ymm4;
+	vpxor 4 * 32(%rsi), %ymm3, %ymm3;
+	vpxor 5 * 32(%rsi), %ymm2, %ymm2;
+	vpxor 6 * 32(%rsi), %ymm1, %ymm1;
+	vpxor 7 * 32(%rsi), %ymm0, %ymm0;
+	vpxor 8 * 32(%rsi), %ymm15, %ymm15;
+	vpxor 9 * 32(%rsi), %ymm14, %ymm14;
+	vpxor 10 * 32(%rsi), %ymm13, %ymm13;
+	vpxor 11 * 32(%rsi), %ymm12, %ymm12;
+	vpxor 12 * 32(%rsi), %ymm11, %ymm11;
+	vpxor 13 * 32(%rsi), %ymm10, %ymm10;
+	vpxor 14 * 32(%rsi), %ymm9, %ymm9;
+	vpxor 15 * 32(%rsi), %ymm8, %ymm8;
+	write_output(%ymm7, %ymm6, %ymm5, %ymm4, %ymm3, %ymm2, %ymm1, %ymm0,
+		     %ymm15, %ymm14, %ymm13, %ymm12, %ymm11, %ymm10, %ymm9,
+		     %ymm8, %rsi);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(camellia_xts_crypt_32way)
+
+ENTRY(camellia_xts_enc_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (32 blocks)
+	 *	%rdx: src (32 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	xorl %r8d, %r8d; /* input whitening key, 0 for enc */
+
+	leaq __camellia_enc_blk32, %r9;
+
+	jmp camellia_xts_crypt_32way;
+ENDPROC(camellia_xts_enc_32way)
+
+ENTRY(camellia_xts_dec_32way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (32 blocks)
+	 *	%rdx: src (32 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	cmpl $16, key_length(CTX);
+	movl $32, %r8d;
+	movl $24, %eax;
+	cmovel %eax, %r8d;  /* input whitening key, last for dec */
+
+	leaq __camellia_dec_blk32, %r9;
+
+	jmp camellia_xts_crypt_32way;
+ENDPROC(camellia_xts_dec_32way)
diff --git a/arch/x86/crypto/camellia_aesni_avx2_glue.c b/arch/x86/crypto/camellia_aesni_avx2_glue.c
new file mode 100644
index 000000000000..414fe5d7946b
--- /dev/null
+++ b/arch/x86/crypto/camellia_aesni_avx2_glue.c
@@ -0,0 +1,586 @@
+/*
+ * Glue Code for x86_64/AVX2/AES-NI assembler optimized version of Camellia
+ *
+ * Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ *
+ * 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.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/crypto.h>
+#include <linux/err.h>
+#include <crypto/algapi.h>
+#include <crypto/ctr.h>
+#include <crypto/lrw.h>
+#include <crypto/xts.h>
+#include <asm/xcr.h>
+#include <asm/xsave.h>
+#include <asm/crypto/camellia.h>
+#include <asm/crypto/ablk_helper.h>
+#include <asm/crypto/glue_helper.h>
+
+#define CAMELLIA_AESNI_PARALLEL_BLOCKS 16
+#define CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS 32
+
+/* 32-way AVX2/AES-NI parallel cipher functions */
+asmlinkage void camellia_ecb_enc_32way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src);
+asmlinkage void camellia_ecb_dec_32way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src);
+
+asmlinkage void camellia_cbc_dec_32way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src);
+asmlinkage void camellia_ctr_32way(struct camellia_ctx *ctx, u8 *dst,
+				   const u8 *src, le128 *iv);
+
+asmlinkage void camellia_xts_enc_32way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src, le128 *iv);
+asmlinkage void camellia_xts_dec_32way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src, le128 *iv);
+
+static const struct common_glue_ctx camellia_enc = {
+	.num_funcs = 4,
+	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_enc_32way) }
+	}, {
+		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_enc_16way) }
+	}, {
+		.num_blocks = 2,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_enc_blk_2way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_enc_blk) }
+	} }
+};
+
+static const struct common_glue_ctx camellia_ctr = {
+	.num_funcs = 4,
+	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_ctr_32way) }
+	}, {
+		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_ctr_16way) }
+	}, {
+		.num_blocks = 2,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_crypt_ctr_2way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(camellia_crypt_ctr) }
+	} }
+};
+
+static const struct common_glue_ctx camellia_enc_xts = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc_32way) }
+	}, {
+		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc_16way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc) }
+	} }
+};
+
+static const struct common_glue_ctx camellia_dec = {
+	.num_funcs = 4,
+	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_dec_32way) }
+	}, {
+		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_ecb_dec_16way) }
+	}, {
+		.num_blocks = 2,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_dec_blk_2way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(camellia_dec_blk) }
+	} }
+};
+
+static const struct common_glue_ctx camellia_dec_cbc = {
+	.num_funcs = 4,
+	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_cbc_dec_32way) }
+	}, {
+		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_cbc_dec_16way) }
+	}, {
+		.num_blocks = 2,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_decrypt_cbc_2way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(camellia_dec_blk) }
+	} }
+};
+
+static const struct common_glue_ctx camellia_dec_xts = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec_32way) }
+	}, {
+		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec_16way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec) }
+	} }
+};
+
+static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ecb_crypt_128bit(&camellia_enc, desc, dst, src, nbytes);
+}
+
+static int ecb_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ecb_crypt_128bit(&camellia_dec, desc, dst, src, nbytes);
+}
+
+static int cbc_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_cbc_encrypt_128bit(GLUE_FUNC_CAST(camellia_enc_blk), desc,
+				       dst, src, nbytes);
+}
+
+static int cbc_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_cbc_decrypt_128bit(&camellia_dec_cbc, desc, dst, src,
+				       nbytes);
+}
+
+static int ctr_crypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		     struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ctr_crypt_128bit(&camellia_ctr, desc, dst, src, nbytes);
+}
+
+static inline bool camellia_fpu_begin(bool fpu_enabled, unsigned int nbytes)
+{
+	return glue_fpu_begin(CAMELLIA_BLOCK_SIZE,
+			      CAMELLIA_AESNI_PARALLEL_BLOCKS, NULL, fpu_enabled,
+			      nbytes);
+}
+
+static inline void camellia_fpu_end(bool fpu_enabled)
+{
+	glue_fpu_end(fpu_enabled);
+}
+
+static int camellia_setkey(struct crypto_tfm *tfm, const u8 *in_key,
+			   unsigned int key_len)
+{
+	return __camellia_setkey(crypto_tfm_ctx(tfm), in_key, key_len,
+				 &tfm->crt_flags);
+}
+
+struct crypt_priv {
+	struct camellia_ctx *ctx;
+	bool fpu_enabled;
+};
+
+static void encrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
+{
+	const unsigned int bsize = CAMELLIA_BLOCK_SIZE;
+	struct crypt_priv *ctx = priv;
+	int i;
+
+	ctx->fpu_enabled = camellia_fpu_begin(ctx->fpu_enabled, nbytes);
+
+	if (nbytes >= CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS * bsize) {
+		camellia_ecb_enc_32way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
+		nbytes -= bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
+	}
+
+	if (nbytes >= CAMELLIA_AESNI_PARALLEL_BLOCKS * bsize) {
+		camellia_ecb_enc_16way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
+		nbytes -= bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
+	}
+
+	while (nbytes >= CAMELLIA_PARALLEL_BLOCKS * bsize) {
+		camellia_enc_blk_2way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * CAMELLIA_PARALLEL_BLOCKS;
+		nbytes -= bsize * CAMELLIA_PARALLEL_BLOCKS;
+	}
+
+	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
+		camellia_enc_blk(ctx->ctx, srcdst, srcdst);
+}
+
+static void decrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
+{
+	const unsigned int bsize = CAMELLIA_BLOCK_SIZE;
+	struct crypt_priv *ctx = priv;
+	int i;
+
+	ctx->fpu_enabled = camellia_fpu_begin(ctx->fpu_enabled, nbytes);
+
+	if (nbytes >= CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS * bsize) {
+		camellia_ecb_dec_32way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
+		nbytes -= bsize * CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS;
+	}
+
+	if (nbytes >= CAMELLIA_AESNI_PARALLEL_BLOCKS * bsize) {
+		camellia_ecb_dec_16way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
+		nbytes -= bsize * CAMELLIA_AESNI_PARALLEL_BLOCKS;
+	}
+
+	while (nbytes >= CAMELLIA_PARALLEL_BLOCKS * bsize) {
+		camellia_dec_blk_2way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * CAMELLIA_PARALLEL_BLOCKS;
+		nbytes -= bsize * CAMELLIA_PARALLEL_BLOCKS;
+	}
+
+	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
+		camellia_dec_blk(ctx->ctx, srcdst, srcdst);
+}
+
+static int lrw_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct camellia_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	be128 buf[CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS];
+	struct crypt_priv crypt_ctx = {
+		.ctx = &ctx->camellia_ctx,
+		.fpu_enabled = false,
+	};
+	struct lrw_crypt_req req = {
+		.tbuf = buf,
+		.tbuflen = sizeof(buf),
+
+		.table_ctx = &ctx->lrw_table,
+		.crypt_ctx = &crypt_ctx,
+		.crypt_fn = encrypt_callback,
+	};
+	int ret;
+
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+	ret = lrw_crypt(desc, dst, src, nbytes, &req);
+	camellia_fpu_end(crypt_ctx.fpu_enabled);
+
+	return ret;
+}
+
+static int lrw_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct camellia_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	be128 buf[CAMELLIA_AESNI_AVX2_PARALLEL_BLOCKS];
+	struct crypt_priv crypt_ctx = {
+		.ctx = &ctx->camellia_ctx,
+		.fpu_enabled = false,
+	};
+	struct lrw_crypt_req req = {
+		.tbuf = buf,
+		.tbuflen = sizeof(buf),
+
+		.table_ctx = &ctx->lrw_table,
+		.crypt_ctx = &crypt_ctx,
+		.crypt_fn = decrypt_callback,
+	};
+	int ret;
+
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+	ret = lrw_crypt(desc, dst, src, nbytes, &req);
+	camellia_fpu_end(crypt_ctx.fpu_enabled);
+
+	return ret;
+}
+
+static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct camellia_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+
+	return glue_xts_crypt_128bit(&camellia_enc_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(camellia_enc_blk),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
+}
+
+static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct camellia_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+
+	return glue_xts_crypt_128bit(&camellia_dec_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(camellia_enc_blk),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
+}
+
+static struct crypto_alg cmll_algs[10] = { {
+	.cra_name		= "__ecb-camellia-aesni-avx2",
+	.cra_driver_name	= "__driver-ecb-camellia-aesni-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct camellia_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
+			.setkey		= camellia_setkey,
+			.encrypt	= ecb_encrypt,
+			.decrypt	= ecb_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__cbc-camellia-aesni-avx2",
+	.cra_driver_name	= "__driver-cbc-camellia-aesni-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct camellia_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
+			.setkey		= camellia_setkey,
+			.encrypt	= cbc_encrypt,
+			.decrypt	= cbc_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__ctr-camellia-aesni-avx2",
+	.cra_driver_name	= "__driver-ctr-camellia-aesni-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= 1,
+	.cra_ctxsize		= sizeof(struct camellia_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
+			.ivsize		= CAMELLIA_BLOCK_SIZE,
+			.setkey		= camellia_setkey,
+			.encrypt	= ctr_crypt,
+			.decrypt	= ctr_crypt,
+		},
+	},
+}, {
+	.cra_name		= "__lrw-camellia-aesni-avx2",
+	.cra_driver_name	= "__driver-lrw-camellia-aesni-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct camellia_lrw_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_exit		= lrw_camellia_exit_tfm,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE +
+					  CAMELLIA_BLOCK_SIZE,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE +
+					  CAMELLIA_BLOCK_SIZE,
+			.ivsize		= CAMELLIA_BLOCK_SIZE,
+			.setkey		= lrw_camellia_setkey,
+			.encrypt	= lrw_encrypt,
+			.decrypt	= lrw_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__xts-camellia-aesni-avx2",
+	.cra_driver_name	= "__driver-xts-camellia-aesni-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct camellia_xts_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE * 2,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE * 2,
+			.ivsize		= CAMELLIA_BLOCK_SIZE,
+			.setkey		= xts_camellia_setkey,
+			.encrypt	= xts_encrypt,
+			.decrypt	= xts_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "ecb(camellia)",
+	.cra_driver_name	= "ecb-camellia-aesni-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "cbc(camellia)",
+	.cra_driver_name	= "cbc-camellia-aesni-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
+			.ivsize		= CAMELLIA_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= __ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "ctr(camellia)",
+	.cra_driver_name	= "ctr-camellia-aesni-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= 1,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE,
+			.ivsize		= CAMELLIA_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_encrypt,
+			.geniv		= "chainiv",
+		},
+	},
+}, {
+	.cra_name		= "lrw(camellia)",
+	.cra_driver_name	= "lrw-camellia-aesni-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE +
+					  CAMELLIA_BLOCK_SIZE,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE +
+					  CAMELLIA_BLOCK_SIZE,
+			.ivsize		= CAMELLIA_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "xts(camellia)",
+	.cra_driver_name	= "xts-camellia-aesni-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= CAMELLIA_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= CAMELLIA_MIN_KEY_SIZE * 2,
+			.max_keysize	= CAMELLIA_MAX_KEY_SIZE * 2,
+			.ivsize		= CAMELLIA_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+} };
+
+static int __init camellia_aesni_init(void)
+{
+	u64 xcr0;
+
+	if (!cpu_has_avx2 || !cpu_has_avx || !cpu_has_aes || !cpu_has_osxsave) {
+		pr_info("AVX2 or AES-NI instructions are not detected.\n");
+		return -ENODEV;
+	}
+
+	xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
+	if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
+		pr_info("AVX2 detected but unusable.\n");
+		return -ENODEV;
+	}
+
+	return crypto_register_algs(cmll_algs, ARRAY_SIZE(cmll_algs));
+}
+
+static void __exit camellia_aesni_fini(void)
+{
+	crypto_unregister_algs(cmll_algs, ARRAY_SIZE(cmll_algs));
+}
+
+module_init(camellia_aesni_init);
+module_exit(camellia_aesni_fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Camellia Cipher Algorithm, AES-NI/AVX2 optimized");
+MODULE_ALIAS("camellia");
+MODULE_ALIAS("camellia-asm");
diff --git a/arch/x86/crypto/camellia_aesni_avx_glue.c b/arch/x86/crypto/camellia_aesni_avx_glue.c
index 96cbb6068fce..37fd0c0a81ea 100644
--- a/arch/x86/crypto/camellia_aesni_avx_glue.c
+++ b/arch/x86/crypto/camellia_aesni_avx_glue.c
@@ -1,7 +1,7 @@
 /*
  * Glue Code for x86_64/AVX/AES-NI assembler optimized version of Camellia
  *
- * Copyright © 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * 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
@@ -26,16 +26,44 @@
 
 #define CAMELLIA_AESNI_PARALLEL_BLOCKS 16
 
-/* 16-way AES-NI parallel cipher functions */
+/* 16-way parallel cipher functions (avx/aes-ni) */
 asmlinkage void camellia_ecb_enc_16way(struct camellia_ctx *ctx, u8 *dst,
 				       const u8 *src);
+EXPORT_SYMBOL_GPL(camellia_ecb_enc_16way);
+
 asmlinkage void camellia_ecb_dec_16way(struct camellia_ctx *ctx, u8 *dst,
 				       const u8 *src);
+EXPORT_SYMBOL_GPL(camellia_ecb_dec_16way);
 
 asmlinkage void camellia_cbc_dec_16way(struct camellia_ctx *ctx, u8 *dst,
 				       const u8 *src);
+EXPORT_SYMBOL_GPL(camellia_cbc_dec_16way);
+
 asmlinkage void camellia_ctr_16way(struct camellia_ctx *ctx, u8 *dst,
 				   const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(camellia_ctr_16way);
+
+asmlinkage void camellia_xts_enc_16way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(camellia_xts_enc_16way);
+
+asmlinkage void camellia_xts_dec_16way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(camellia_xts_dec_16way);
+
+void camellia_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv,
+				  GLUE_FUNC_CAST(camellia_enc_blk));
+}
+EXPORT_SYMBOL_GPL(camellia_xts_enc);
+
+void camellia_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv,
+				  GLUE_FUNC_CAST(camellia_dec_blk));
+}
+EXPORT_SYMBOL_GPL(camellia_xts_dec);
 
 static const struct common_glue_ctx camellia_enc = {
 	.num_funcs = 3,
@@ -69,6 +97,19 @@ static const struct common_glue_ctx camellia_ctr = {
 	} }
 };
 
+static const struct common_glue_ctx camellia_enc_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc_16way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_enc) }
+	} }
+};
+
 static const struct common_glue_ctx camellia_dec = {
 	.num_funcs = 3,
 	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
@@ -101,6 +142,19 @@ static const struct common_glue_ctx camellia_dec_cbc = {
 	} }
 };
 
+static const struct common_glue_ctx camellia_dec_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAMELLIA_AESNI_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec_16way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(camellia_xts_dec) }
+	} }
+};
+
 static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
@@ -261,54 +315,20 @@ static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
 	struct camellia_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-	be128 buf[CAMELLIA_AESNI_PARALLEL_BLOCKS];
-	struct crypt_priv crypt_ctx = {
-		.ctx = &ctx->crypt_ctx,
-		.fpu_enabled = false,
-	};
-	struct xts_crypt_req req = {
-		.tbuf = buf,
-		.tbuflen = sizeof(buf),
 
-		.tweak_ctx = &ctx->tweak_ctx,
-		.tweak_fn = XTS_TWEAK_CAST(camellia_enc_blk),
-		.crypt_ctx = &crypt_ctx,
-		.crypt_fn = encrypt_callback,
-	};
-	int ret;
-
-	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
-	ret = xts_crypt(desc, dst, src, nbytes, &req);
-	camellia_fpu_end(crypt_ctx.fpu_enabled);
-
-	return ret;
+	return glue_xts_crypt_128bit(&camellia_enc_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(camellia_enc_blk),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
 }
 
 static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
 	struct camellia_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-	be128 buf[CAMELLIA_AESNI_PARALLEL_BLOCKS];
-	struct crypt_priv crypt_ctx = {
-		.ctx = &ctx->crypt_ctx,
-		.fpu_enabled = false,
-	};
-	struct xts_crypt_req req = {
-		.tbuf = buf,
-		.tbuflen = sizeof(buf),
-
-		.tweak_ctx = &ctx->tweak_ctx,
-		.tweak_fn = XTS_TWEAK_CAST(camellia_enc_blk),
-		.crypt_ctx = &crypt_ctx,
-		.crypt_fn = decrypt_callback,
-	};
-	int ret;
 
-	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
-	ret = xts_crypt(desc, dst, src, nbytes, &req);
-	camellia_fpu_end(crypt_ctx.fpu_enabled);
-
-	return ret;
+	return glue_xts_crypt_128bit(&camellia_dec_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(camellia_enc_blk),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
 }
 
 static struct crypto_alg cmll_algs[10] = { {
diff --git a/arch/x86/crypto/cast6-avx-x86_64-asm_64.S b/arch/x86/crypto/cast6-avx-x86_64-asm_64.S
index f93b6105a0ce..e3531f833951 100644
--- a/arch/x86/crypto/cast6-avx-x86_64-asm_64.S
+++ b/arch/x86/crypto/cast6-avx-x86_64-asm_64.S
@@ -4,7 +4,7 @@
  * Copyright (C) 2012 Johannes Goetzfried
  *     <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
  *
- * Copyright © 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * 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
@@ -227,6 +227,8 @@
 .data
 
 .align 16
+.Lxts_gf128mul_and_shl1_mask:
+	.byte 0x87, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
 .Lbswap_mask:
 	.byte 3, 2, 1, 0, 7, 6, 5, 4, 11, 10, 9, 8, 15, 14, 13, 12
 .Lbswap128_mask:
@@ -424,3 +426,47 @@ ENTRY(cast6_ctr_8way)
 
 	ret;
 ENDPROC(cast6_ctr_8way)
+
+ENTRY(cast6_xts_enc_8way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	movq %rsi, %r11;
+
+	/* regs <= src, dst <= IVs, regs <= regs xor IVs */
+	load_xts_8way(%rcx, %rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2,
+		      RX, RKR, RKM, .Lxts_gf128mul_and_shl1_mask);
+
+	call __cast6_enc_blk8;
+
+	/* dst <= regs xor IVs(in dst) */
+	store_xts_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	ret;
+ENDPROC(cast6_xts_enc_8way)
+
+ENTRY(cast6_xts_dec_8way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	movq %rsi, %r11;
+
+	/* regs <= src, dst <= IVs, regs <= regs xor IVs */
+	load_xts_8way(%rcx, %rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2,
+		      RX, RKR, RKM, .Lxts_gf128mul_and_shl1_mask);
+
+	call __cast6_dec_blk8;
+
+	/* dst <= regs xor IVs(in dst) */
+	store_xts_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	ret;
+ENDPROC(cast6_xts_dec_8way)
diff --git a/arch/x86/crypto/cast6_avx_glue.c b/arch/x86/crypto/cast6_avx_glue.c
index 92f7ca24790a..8d0dfb86a559 100644
--- a/arch/x86/crypto/cast6_avx_glue.c
+++ b/arch/x86/crypto/cast6_avx_glue.c
@@ -4,6 +4,8 @@
  * Copyright (C) 2012 Johannes Goetzfried
  *     <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
  *
+ * Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
+ *
  * 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
@@ -50,6 +52,23 @@ asmlinkage void cast6_cbc_dec_8way(struct cast6_ctx *ctx, u8 *dst,
 asmlinkage void cast6_ctr_8way(struct cast6_ctx *ctx, u8 *dst, const u8 *src,
 			       le128 *iv);
 
+asmlinkage void cast6_xts_enc_8way(struct cast6_ctx *ctx, u8 *dst,
+				   const u8 *src, le128 *iv);
+asmlinkage void cast6_xts_dec_8way(struct cast6_ctx *ctx, u8 *dst,
+				   const u8 *src, le128 *iv);
+
+static void cast6_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv,
+				  GLUE_FUNC_CAST(__cast6_encrypt));
+}
+
+static void cast6_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv,
+				  GLUE_FUNC_CAST(__cast6_decrypt));
+}
+
 static void cast6_crypt_ctr(void *ctx, u128 *dst, const u128 *src, le128 *iv)
 {
 	be128 ctrblk;
@@ -87,6 +106,19 @@ static const struct common_glue_ctx cast6_ctr = {
 	} }
 };
 
+static const struct common_glue_ctx cast6_enc_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAST6_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(cast6_xts_enc_8way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(cast6_xts_enc) }
+	} }
+};
+
 static const struct common_glue_ctx cast6_dec = {
 	.num_funcs = 2,
 	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
@@ -113,6 +145,19 @@ static const struct common_glue_ctx cast6_dec_cbc = {
 	} }
 };
 
+static const struct common_glue_ctx cast6_dec_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = CAST6_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = CAST6_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(cast6_xts_dec_8way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(cast6_xts_dec) }
+	} }
+};
+
 static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
@@ -307,54 +352,20 @@ static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
 	struct cast6_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-	be128 buf[CAST6_PARALLEL_BLOCKS];
-	struct crypt_priv crypt_ctx = {
-		.ctx = &ctx->crypt_ctx,
-		.fpu_enabled = false,
-	};
-	struct xts_crypt_req req = {
-		.tbuf = buf,
-		.tbuflen = sizeof(buf),
 
-		.tweak_ctx = &ctx->tweak_ctx,
-		.tweak_fn = XTS_TWEAK_CAST(__cast6_encrypt),
-		.crypt_ctx = &crypt_ctx,
-		.crypt_fn = encrypt_callback,
-	};
-	int ret;
-
-	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
-	ret = xts_crypt(desc, dst, src, nbytes, &req);
-	cast6_fpu_end(crypt_ctx.fpu_enabled);
-
-	return ret;
+	return glue_xts_crypt_128bit(&cast6_enc_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(__cast6_encrypt),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
 }
 
 static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
 	struct cast6_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-	be128 buf[CAST6_PARALLEL_BLOCKS];
-	struct crypt_priv crypt_ctx = {
-		.ctx = &ctx->crypt_ctx,
-		.fpu_enabled = false,
-	};
-	struct xts_crypt_req req = {
-		.tbuf = buf,
-		.tbuflen = sizeof(buf),
-
-		.tweak_ctx = &ctx->tweak_ctx,
-		.tweak_fn = XTS_TWEAK_CAST(__cast6_encrypt),
-		.crypt_ctx = &crypt_ctx,
-		.crypt_fn = decrypt_callback,
-	};
-	int ret;
 
-	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
-	ret = xts_crypt(desc, dst, src, nbytes, &req);
-	cast6_fpu_end(crypt_ctx.fpu_enabled);
-
-	return ret;
+	return glue_xts_crypt_128bit(&cast6_dec_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(__cast6_encrypt),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
 }
 
 static struct crypto_alg cast6_algs[10] = { {
diff --git a/arch/x86/crypto/crc32-pclmul_asm.S b/arch/x86/crypto/crc32-pclmul_asm.S
index c8335014a044..94c27df8a549 100644
--- a/arch/x86/crypto/crc32-pclmul_asm.S
+++ b/arch/x86/crypto/crc32-pclmul_asm.S
@@ -101,9 +101,8 @@
  *      uint crc32_pclmul_le_16(unsigned char const *buffer,
  *	                     size_t len, uint crc32)
  */
-.globl crc32_pclmul_le_16
-.align 4, 0x90
-crc32_pclmul_le_16:/* buffer and buffer size are 16 bytes aligned */
+
+ENTRY(crc32_pclmul_le_16) /* buffer and buffer size are 16 bytes aligned */
 	movdqa  (BUF), %xmm1
 	movdqa  0x10(BUF), %xmm2
 	movdqa  0x20(BUF), %xmm3
@@ -244,3 +243,4 @@ fold_64:
 	pextrd  $0x01, %xmm1, %eax
 
 	ret
+ENDPROC(crc32_pclmul_le_16)
diff --git a/arch/x86/crypto/crc32c-pcl-intel-asm_64.S b/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
index cf1a7ec4cc3a..dbc4339b5417 100644
--- a/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
+++ b/arch/x86/crypto/crc32c-pcl-intel-asm_64.S
@@ -1,9 +1,10 @@
 /*
  * Implement fast CRC32C with PCLMULQDQ instructions. (x86_64)
  *
- * The white paper on CRC32C calculations with PCLMULQDQ instruction can be
+ * The white papers on CRC32C calculations with PCLMULQDQ instruction can be
  * downloaded from:
- * http://download.intel.com/design/intarch/papers/323405.pdf
+ * http://www.intel.com/content/dam/www/public/us/en/documents/white-papers/crc-iscsi-polynomial-crc32-instruction-paper.pdf
+ * http://www.intel.com/content/dam/www/public/us/en/documents/white-papers/fast-crc-computation-paper.pdf
  *
  * Copyright (C) 2012 Intel Corporation.
  *
@@ -42,6 +43,7 @@
  * SOFTWARE.
  */
 
+#include <asm/inst.h>
 #include <linux/linkage.h>
 
 ## ISCSI CRC 32 Implementation with crc32 and pclmulqdq Instruction
@@ -225,10 +227,10 @@ LABEL crc_ %i
 	movdqa  (bufp), %xmm0			# 2 consts: K1:K2
 
 	movq    crc_init, %xmm1			# CRC for block 1
-	pclmulqdq $0x00,%xmm0,%xmm1		# Multiply by K2
+	PCLMULQDQ 0x00,%xmm0,%xmm1		# Multiply by K2
 
 	movq    crc1, %xmm2			# CRC for block 2
-	pclmulqdq $0x10, %xmm0, %xmm2		# Multiply by K1
+	PCLMULQDQ 0x10, %xmm0, %xmm2		# Multiply by K1
 
 	pxor    %xmm2,%xmm1
 	movq    %xmm1, %rax
diff --git a/arch/x86/crypto/glue_helper-asm-avx.S b/arch/x86/crypto/glue_helper-asm-avx.S
index f7b6ea2ddfdb..02ee2308fb38 100644
--- a/arch/x86/crypto/glue_helper-asm-avx.S
+++ b/arch/x86/crypto/glue_helper-asm-avx.S
@@ -1,7 +1,7 @@
 /*
  * Shared glue code for 128bit block ciphers, AVX assembler macros
  *
- * Copyright (c) 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * 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
@@ -89,3 +89,62 @@
 	vpxor (6*16)(src), x6, x6; \
 	vpxor (7*16)(src), x7, x7; \
 	store_8way(dst, x0, x1, x2, x3, x4, x5, x6, x7);
+
+#define gf128mul_x_ble(iv, mask, tmp) \
+	vpsrad $31, iv, tmp; \
+	vpaddq iv, iv, iv; \
+	vpshufd $0x13, tmp, tmp; \
+	vpand mask, tmp, tmp; \
+	vpxor tmp, iv, iv;
+
+#define load_xts_8way(iv, src, dst, x0, x1, x2, x3, x4, x5, x6, x7, tiv, t0, \
+		      t1, xts_gf128mul_and_shl1_mask) \
+	vmovdqa xts_gf128mul_and_shl1_mask, t0; \
+	\
+	/* load IV */ \
+	vmovdqu (iv), tiv; \
+	vpxor (0*16)(src), tiv, x0; \
+	vmovdqu tiv, (0*16)(dst); \
+	\
+	/* construct and store IVs, also xor with source */ \
+	gf128mul_x_ble(tiv, t0, t1); \
+	vpxor (1*16)(src), tiv, x1; \
+	vmovdqu tiv, (1*16)(dst); \
+	\
+	gf128mul_x_ble(tiv, t0, t1); \
+	vpxor (2*16)(src), tiv, x2; \
+	vmovdqu tiv, (2*16)(dst); \
+	\
+	gf128mul_x_ble(tiv, t0, t1); \
+	vpxor (3*16)(src), tiv, x3; \
+	vmovdqu tiv, (3*16)(dst); \
+	\
+	gf128mul_x_ble(tiv, t0, t1); \
+	vpxor (4*16)(src), tiv, x4; \
+	vmovdqu tiv, (4*16)(dst); \
+	\
+	gf128mul_x_ble(tiv, t0, t1); \
+	vpxor (5*16)(src), tiv, x5; \
+	vmovdqu tiv, (5*16)(dst); \
+	\
+	gf128mul_x_ble(tiv, t0, t1); \
+	vpxor (6*16)(src), tiv, x6; \
+	vmovdqu tiv, (6*16)(dst); \
+	\
+	gf128mul_x_ble(tiv, t0, t1); \
+	vpxor (7*16)(src), tiv, x7; \
+	vmovdqu tiv, (7*16)(dst); \
+	\
+	gf128mul_x_ble(tiv, t0, t1); \
+	vmovdqu tiv, (iv);
+
+#define store_xts_8way(dst, x0, x1, x2, x3, x4, x5, x6, x7) \
+	vpxor (0*16)(dst), x0, x0; \
+	vpxor (1*16)(dst), x1, x1; \
+	vpxor (2*16)(dst), x2, x2; \
+	vpxor (3*16)(dst), x3, x3; \
+	vpxor (4*16)(dst), x4, x4; \
+	vpxor (5*16)(dst), x5, x5; \
+	vpxor (6*16)(dst), x6, x6; \
+	vpxor (7*16)(dst), x7, x7; \
+	store_8way(dst, x0, x1, x2, x3, x4, x5, x6, x7);
diff --git a/arch/x86/crypto/glue_helper-asm-avx2.S b/arch/x86/crypto/glue_helper-asm-avx2.S
new file mode 100644
index 000000000000..a53ac11dd385
--- /dev/null
+++ b/arch/x86/crypto/glue_helper-asm-avx2.S
@@ -0,0 +1,180 @@
+/*
+ * Shared glue code for 128bit block ciphers, AVX2 assembler macros
+ *
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ *
+ * 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.
+ *
+ */
+
+#define load_16way(src, x0, x1, x2, x3, x4, x5, x6, x7) \
+	vmovdqu (0*32)(src), x0; \
+	vmovdqu (1*32)(src), x1; \
+	vmovdqu (2*32)(src), x2; \
+	vmovdqu (3*32)(src), x3; \
+	vmovdqu (4*32)(src), x4; \
+	vmovdqu (5*32)(src), x5; \
+	vmovdqu (6*32)(src), x6; \
+	vmovdqu (7*32)(src), x7;
+
+#define store_16way(dst, x0, x1, x2, x3, x4, x5, x6, x7) \
+	vmovdqu x0, (0*32)(dst); \
+	vmovdqu x1, (1*32)(dst); \
+	vmovdqu x2, (2*32)(dst); \
+	vmovdqu x3, (3*32)(dst); \
+	vmovdqu x4, (4*32)(dst); \
+	vmovdqu x5, (5*32)(dst); \
+	vmovdqu x6, (6*32)(dst); \
+	vmovdqu x7, (7*32)(dst);
+
+#define store_cbc_16way(src, dst, x0, x1, x2, x3, x4, x5, x6, x7, t0) \
+	vpxor t0, t0, t0; \
+	vinserti128 $1, (src), t0, t0; \
+	vpxor t0, x0, x0; \
+	vpxor (0*32+16)(src), x1, x1; \
+	vpxor (1*32+16)(src), x2, x2; \
+	vpxor (2*32+16)(src), x3, x3; \
+	vpxor (3*32+16)(src), x4, x4; \
+	vpxor (4*32+16)(src), x5, x5; \
+	vpxor (5*32+16)(src), x6, x6; \
+	vpxor (6*32+16)(src), x7, x7; \
+	store_16way(dst, x0, x1, x2, x3, x4, x5, x6, x7);
+
+#define inc_le128(x, minus_one, tmp) \
+	vpcmpeqq minus_one, x, tmp; \
+	vpsubq minus_one, x, x; \
+	vpslldq $8, tmp, tmp; \
+	vpsubq tmp, x, x;
+
+#define add2_le128(x, minus_one, minus_two, tmp1, tmp2) \
+	vpcmpeqq minus_one, x, tmp1; \
+	vpcmpeqq minus_two, x, tmp2; \
+	vpsubq minus_two, x, x; \
+	vpor tmp2, tmp1, tmp1; \
+	vpslldq $8, tmp1, tmp1; \
+	vpsubq tmp1, x, x;
+
+#define load_ctr_16way(iv, bswap, x0, x1, x2, x3, x4, x5, x6, x7, t0, t0x, t1, \
+		       t1x, t2, t2x, t3, t3x, t4, t5) \
+	vpcmpeqd t0, t0, t0; \
+	vpsrldq $8, t0, t0; /* ab: -1:0 ; cd: -1:0 */ \
+	vpaddq t0, t0, t4; /* ab: -2:0 ; cd: -2:0 */\
+	\
+	/* load IV and byteswap */ \
+	vmovdqu (iv), t2x; \
+	vmovdqa t2x, t3x; \
+	inc_le128(t2x, t0x, t1x); \
+	vbroadcasti128 bswap, t1; \
+	vinserti128 $1, t2x, t3, t2; /* ab: le0 ; cd: le1 */ \
+	vpshufb t1, t2, x0; \
+	\
+	/* construct IVs */ \
+	add2_le128(t2, t0, t4, t3, t5); /* ab: le2 ; cd: le3 */ \
+	vpshufb t1, t2, x1; \
+	add2_le128(t2, t0, t4, t3, t5); \
+	vpshufb t1, t2, x2; \
+	add2_le128(t2, t0, t4, t3, t5); \
+	vpshufb t1, t2, x3; \
+	add2_le128(t2, t0, t4, t3, t5); \
+	vpshufb t1, t2, x4; \
+	add2_le128(t2, t0, t4, t3, t5); \
+	vpshufb t1, t2, x5; \
+	add2_le128(t2, t0, t4, t3, t5); \
+	vpshufb t1, t2, x6; \
+	add2_le128(t2, t0, t4, t3, t5); \
+	vpshufb t1, t2, x7; \
+	vextracti128 $1, t2, t2x; \
+	inc_le128(t2x, t0x, t3x); \
+	vmovdqu t2x, (iv);
+
+#define store_ctr_16way(src, dst, x0, x1, x2, x3, x4, x5, x6, x7) \
+	vpxor (0*32)(src), x0, x0; \
+	vpxor (1*32)(src), x1, x1; \
+	vpxor (2*32)(src), x2, x2; \
+	vpxor (3*32)(src), x3, x3; \
+	vpxor (4*32)(src), x4, x4; \
+	vpxor (5*32)(src), x5, x5; \
+	vpxor (6*32)(src), x6, x6; \
+	vpxor (7*32)(src), x7, x7; \
+	store_16way(dst, x0, x1, x2, x3, x4, x5, x6, x7);
+
+#define gf128mul_x_ble(iv, mask, tmp) \
+	vpsrad $31, iv, tmp; \
+	vpaddq iv, iv, iv; \
+	vpshufd $0x13, tmp, tmp; \
+	vpand mask, tmp, tmp; \
+	vpxor tmp, iv, iv;
+
+#define gf128mul_x2_ble(iv, mask1, mask2, tmp0, tmp1) \
+	vpsrad $31, iv, tmp0; \
+	vpaddq iv, iv, tmp1; \
+	vpsllq $2, iv, iv; \
+	vpshufd $0x13, tmp0, tmp0; \
+	vpsrad $31, tmp1, tmp1; \
+	vpand mask2, tmp0, tmp0; \
+	vpshufd $0x13, tmp1, tmp1; \
+	vpxor tmp0, iv, iv; \
+	vpand mask1, tmp1, tmp1; \
+	vpxor tmp1, iv, iv;
+
+#define load_xts_16way(iv, src, dst, x0, x1, x2, x3, x4, x5, x6, x7, tiv, \
+		       tivx, t0, t0x, t1, t1x, t2, t2x, t3, \
+		       xts_gf128mul_and_shl1_mask_0, \
+		       xts_gf128mul_and_shl1_mask_1) \
+	vbroadcasti128 xts_gf128mul_and_shl1_mask_0, t1; \
+	\
+	/* load IV and construct second IV */ \
+	vmovdqu (iv), tivx; \
+	vmovdqa tivx, t0x; \
+	gf128mul_x_ble(tivx, t1x, t2x); \
+	vbroadcasti128 xts_gf128mul_and_shl1_mask_1, t2; \
+	vinserti128 $1, tivx, t0, tiv; \
+	vpxor (0*32)(src), tiv, x0; \
+	vmovdqu tiv, (0*32)(dst); \
+	\
+	/* construct and store IVs, also xor with source */ \
+	gf128mul_x2_ble(tiv, t1, t2, t0, t3); \
+	vpxor (1*32)(src), tiv, x1; \
+	vmovdqu tiv, (1*32)(dst); \
+	\
+	gf128mul_x2_ble(tiv, t1, t2, t0, t3); \
+	vpxor (2*32)(src), tiv, x2; \
+	vmovdqu tiv, (2*32)(dst); \
+	\
+	gf128mul_x2_ble(tiv, t1, t2, t0, t3); \
+	vpxor (3*32)(src), tiv, x3; \
+	vmovdqu tiv, (3*32)(dst); \
+	\
+	gf128mul_x2_ble(tiv, t1, t2, t0, t3); \
+	vpxor (4*32)(src), tiv, x4; \
+	vmovdqu tiv, (4*32)(dst); \
+	\
+	gf128mul_x2_ble(tiv, t1, t2, t0, t3); \
+	vpxor (5*32)(src), tiv, x5; \
+	vmovdqu tiv, (5*32)(dst); \
+	\
+	gf128mul_x2_ble(tiv, t1, t2, t0, t3); \
+	vpxor (6*32)(src), tiv, x6; \
+	vmovdqu tiv, (6*32)(dst); \
+	\
+	gf128mul_x2_ble(tiv, t1, t2, t0, t3); \
+	vpxor (7*32)(src), tiv, x7; \
+	vmovdqu tiv, (7*32)(dst); \
+	\
+	vextracti128 $1, tiv, tivx; \
+	gf128mul_x_ble(tivx, t1x, t2x); \
+	vmovdqu tivx, (iv);
+
+#define store_xts_16way(dst, x0, x1, x2, x3, x4, x5, x6, x7) \
+	vpxor (0*32)(dst), x0, x0; \
+	vpxor (1*32)(dst), x1, x1; \
+	vpxor (2*32)(dst), x2, x2; \
+	vpxor (3*32)(dst), x3, x3; \
+	vpxor (4*32)(dst), x4, x4; \
+	vpxor (5*32)(dst), x5, x5; \
+	vpxor (6*32)(dst), x6, x6; \
+	vpxor (7*32)(dst), x7, x7; \
+	store_16way(dst, x0, x1, x2, x3, x4, x5, x6, x7);
diff --git a/arch/x86/crypto/glue_helper.c b/arch/x86/crypto/glue_helper.c
index 22ce4f683e55..432f1d76ceb8 100644
--- a/arch/x86/crypto/glue_helper.c
+++ b/arch/x86/crypto/glue_helper.c
@@ -1,7 +1,7 @@
 /*
  * Shared glue code for 128bit block ciphers
  *
- * Copyright (c) 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * CBC & ECB parts based on code (crypto/cbc.c,ecb.c) by:
  *   Copyright (c) 2006 Herbert Xu <herbert@gondor.apana.org.au>
@@ -304,4 +304,99 @@ int glue_ctr_crypt_128bit(const struct common_glue_ctx *gctx,
 }
 EXPORT_SYMBOL_GPL(glue_ctr_crypt_128bit);
 
+static unsigned int __glue_xts_crypt_128bit(const struct common_glue_ctx *gctx,
+					    void *ctx,
+					    struct blkcipher_desc *desc,
+					    struct blkcipher_walk *walk)
+{
+	const unsigned int bsize = 128 / 8;
+	unsigned int nbytes = walk->nbytes;
+	u128 *src = (u128 *)walk->src.virt.addr;
+	u128 *dst = (u128 *)walk->dst.virt.addr;
+	unsigned int num_blocks, func_bytes;
+	unsigned int i;
+
+	/* Process multi-block batch */
+	for (i = 0; i < gctx->num_funcs; i++) {
+		num_blocks = gctx->funcs[i].num_blocks;
+		func_bytes = bsize * num_blocks;
+
+		if (nbytes >= func_bytes) {
+			do {
+				gctx->funcs[i].fn_u.xts(ctx, dst, src,
+							(le128 *)walk->iv);
+
+				src += num_blocks;
+				dst += num_blocks;
+				nbytes -= func_bytes;
+			} while (nbytes >= func_bytes);
+
+			if (nbytes < bsize)
+				goto done;
+		}
+	}
+
+done:
+	return nbytes;
+}
+
+/* for implementations implementing faster XTS IV generator */
+int glue_xts_crypt_128bit(const struct common_glue_ctx *gctx,
+			  struct blkcipher_desc *desc, struct scatterlist *dst,
+			  struct scatterlist *src, unsigned int nbytes,
+			  void (*tweak_fn)(void *ctx, u8 *dst, const u8 *src),
+			  void *tweak_ctx, void *crypt_ctx)
+{
+	const unsigned int bsize = 128 / 8;
+	bool fpu_enabled = false;
+	struct blkcipher_walk walk;
+	int err;
+
+	blkcipher_walk_init(&walk, dst, src, nbytes);
+
+	err = blkcipher_walk_virt(desc, &walk);
+	nbytes = walk.nbytes;
+	if (!nbytes)
+		return err;
+
+	/* set minimum length to bsize, for tweak_fn */
+	fpu_enabled = glue_fpu_begin(bsize, gctx->fpu_blocks_limit,
+				     desc, fpu_enabled,
+				     nbytes < bsize ? bsize : nbytes);
+
+	/* calculate first value of T */
+	tweak_fn(tweak_ctx, walk.iv, walk.iv);
+
+	while (nbytes) {
+		nbytes = __glue_xts_crypt_128bit(gctx, crypt_ctx, desc, &walk);
+
+		err = blkcipher_walk_done(desc, &walk, nbytes);
+		nbytes = walk.nbytes;
+	}
+
+	glue_fpu_end(fpu_enabled);
+
+	return err;
+}
+EXPORT_SYMBOL_GPL(glue_xts_crypt_128bit);
+
+void glue_xts_crypt_128bit_one(void *ctx, u128 *dst, const u128 *src, le128 *iv,
+			       common_glue_func_t fn)
+{
+	le128 ivblk = *iv;
+
+	/* generate next IV */
+	le128_gf128mul_x_ble(iv, &ivblk);
+
+	/* CC <- T xor C */
+	u128_xor(dst, src, (u128 *)&ivblk);
+
+	/* PP <- D(Key2,CC) */
+	fn(ctx, (u8 *)dst, (u8 *)dst);
+
+	/* P <- T xor PP */
+	u128_xor(dst, dst, (u128 *)&ivblk);
+}
+EXPORT_SYMBOL_GPL(glue_xts_crypt_128bit_one);
+
 MODULE_LICENSE("GPL");
diff --git a/arch/x86/crypto/serpent-avx-x86_64-asm_64.S b/arch/x86/crypto/serpent-avx-x86_64-asm_64.S
index 43c938612b74..2f202f49872b 100644
--- a/arch/x86/crypto/serpent-avx-x86_64-asm_64.S
+++ b/arch/x86/crypto/serpent-avx-x86_64-asm_64.S
@@ -4,8 +4,7 @@
  * Copyright (C) 2012 Johannes Goetzfried
  *     <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
  *
- * Based on arch/x86/crypto/serpent-sse2-x86_64-asm_64.S by
- *  Copyright (C) 2011 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2011-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * 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
@@ -34,6 +33,8 @@
 
 .Lbswap128_mask:
 	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
+.Lxts_gf128mul_and_shl1_mask:
+	.byte 0x87, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
 
 .text
 
@@ -739,3 +740,43 @@ ENTRY(serpent_ctr_8way_avx)
 
 	ret;
 ENDPROC(serpent_ctr_8way_avx)
+
+ENTRY(serpent_xts_enc_8way_avx)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	/* regs <= src, dst <= IVs, regs <= regs xor IVs */
+	load_xts_8way(%rcx, %rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2,
+		      RK0, RK1, RK2, .Lxts_gf128mul_and_shl1_mask);
+
+	call __serpent_enc_blk8_avx;
+
+	/* dst <= regs xor IVs(in dst) */
+	store_xts_8way(%rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	ret;
+ENDPROC(serpent_xts_enc_8way_avx)
+
+ENTRY(serpent_xts_dec_8way_avx)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	/* regs <= src, dst <= IVs, regs <= regs xor IVs */
+	load_xts_8way(%rcx, %rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2,
+		      RK0, RK1, RK2, .Lxts_gf128mul_and_shl1_mask);
+
+	call __serpent_dec_blk8_avx;
+
+	/* dst <= regs xor IVs(in dst) */
+	store_xts_8way(%rsi, RC1, RD1, RB1, RE1, RC2, RD2, RB2, RE2);
+
+	ret;
+ENDPROC(serpent_xts_dec_8way_avx)
diff --git a/arch/x86/crypto/serpent-avx2-asm_64.S b/arch/x86/crypto/serpent-avx2-asm_64.S
new file mode 100644
index 000000000000..b222085cccac
--- /dev/null
+++ b/arch/x86/crypto/serpent-avx2-asm_64.S
@@ -0,0 +1,800 @@
+/*
+ * x86_64/AVX2 assembler optimized version of Serpent
+ *
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ *
+ * Based on AVX assembler implementation of Serpent by:
+ *  Copyright © 2012 Johannes Goetzfried
+ *      <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
+ *
+ * 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.
+ *
+ */
+
+#include <linux/linkage.h>
+#include "glue_helper-asm-avx2.S"
+
+.file "serpent-avx2-asm_64.S"
+
+.data
+.align 16
+
+.Lbswap128_mask:
+	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
+.Lxts_gf128mul_and_shl1_mask_0:
+	.byte 0x87, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
+.Lxts_gf128mul_and_shl1_mask_1:
+	.byte 0x0e, 1, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0
+
+.text
+
+#define CTX %rdi
+
+#define RNOT %ymm0
+#define tp  %ymm1
+
+#define RA1 %ymm2
+#define RA2 %ymm3
+#define RB1 %ymm4
+#define RB2 %ymm5
+#define RC1 %ymm6
+#define RC2 %ymm7
+#define RD1 %ymm8
+#define RD2 %ymm9
+#define RE1 %ymm10
+#define RE2 %ymm11
+
+#define RK0 %ymm12
+#define RK1 %ymm13
+#define RK2 %ymm14
+#define RK3 %ymm15
+
+#define RK0x %xmm12
+#define RK1x %xmm13
+#define RK2x %xmm14
+#define RK3x %xmm15
+
+#define S0_1(x0, x1, x2, x3, x4)      \
+	vpor		x0,   x3, tp; \
+	vpxor		x3,   x0, x0; \
+	vpxor		x2,   x3, x4; \
+	vpxor		RNOT, x4, x4; \
+	vpxor		x1,   tp, x3; \
+	vpand		x0,   x1, x1; \
+	vpxor		x4,   x1, x1; \
+	vpxor		x0,   x2, x2;
+#define S0_2(x0, x1, x2, x3, x4)      \
+	vpxor		x3,   x0, x0; \
+	vpor		x0,   x4, x4; \
+	vpxor		x2,   x0, x0; \
+	vpand		x1,   x2, x2; \
+	vpxor		x2,   x3, x3; \
+	vpxor		RNOT, x1, x1; \
+	vpxor		x4,   x2, x2; \
+	vpxor		x2,   x1, x1;
+
+#define S1_1(x0, x1, x2, x3, x4)      \
+	vpxor		x0,   x1, tp; \
+	vpxor		x3,   x0, x0; \
+	vpxor		RNOT, x3, x3; \
+	vpand		tp,   x1, x4; \
+	vpor		tp,   x0, x0; \
+	vpxor		x2,   x3, x3; \
+	vpxor		x3,   x0, x0; \
+	vpxor		x3,   tp, x1;
+#define S1_2(x0, x1, x2, x3, x4)      \
+	vpxor		x4,   x3, x3; \
+	vpor		x4,   x1, x1; \
+	vpxor		x2,   x4, x4; \
+	vpand		x0,   x2, x2; \
+	vpxor		x1,   x2, x2; \
+	vpor		x0,   x1, x1; \
+	vpxor		RNOT, x0, x0; \
+	vpxor		x2,   x0, x0; \
+	vpxor		x1,   x4, x4;
+
+#define S2_1(x0, x1, x2, x3, x4)      \
+	vpxor		RNOT, x3, x3; \
+	vpxor		x0,   x1, x1; \
+	vpand		x2,   x0, tp; \
+	vpxor		x3,   tp, tp; \
+	vpor		x0,   x3, x3; \
+	vpxor		x1,   x2, x2; \
+	vpxor		x1,   x3, x3; \
+	vpand		tp,   x1, x1;
+#define S2_2(x0, x1, x2, x3, x4)      \
+	vpxor		x2,   tp, tp; \
+	vpand		x3,   x2, x2; \
+	vpor		x1,   x3, x3; \
+	vpxor		RNOT, tp, tp; \
+	vpxor		tp,   x3, x3; \
+	vpxor		tp,   x0, x4; \
+	vpxor		x2,   tp, x0; \
+	vpor		x2,   x1, x1;
+
+#define S3_1(x0, x1, x2, x3, x4)      \
+	vpxor		x3,   x1, tp; \
+	vpor		x0,   x3, x3; \
+	vpand		x0,   x1, x4; \
+	vpxor		x2,   x0, x0; \
+	vpxor		tp,   x2, x2; \
+	vpand		x3,   tp, x1; \
+	vpxor		x3,   x2, x2; \
+	vpor		x4,   x0, x0; \
+	vpxor		x3,   x4, x4;
+#define S3_2(x0, x1, x2, x3, x4)      \
+	vpxor		x0,   x1, x1; \
+	vpand		x3,   x0, x0; \
+	vpand		x4,   x3, x3; \
+	vpxor		x2,   x3, x3; \
+	vpor		x1,   x4, x4; \
+	vpand		x1,   x2, x2; \
+	vpxor		x3,   x4, x4; \
+	vpxor		x3,   x0, x0; \
+	vpxor		x2,   x3, x3;
+
+#define S4_1(x0, x1, x2, x3, x4)      \
+	vpand		x0,   x3, tp; \
+	vpxor		x3,   x0, x0; \
+	vpxor		x2,   tp, tp; \
+	vpor		x3,   x2, x2; \
+	vpxor		x1,   x0, x0; \
+	vpxor		tp,   x3, x4; \
+	vpor		x0,   x2, x2; \
+	vpxor		x1,   x2, x2;
+#define S4_2(x0, x1, x2, x3, x4)      \
+	vpand		x0,   x1, x1; \
+	vpxor		x4,   x1, x1; \
+	vpand		x2,   x4, x4; \
+	vpxor		tp,   x2, x2; \
+	vpxor		x0,   x4, x4; \
+	vpor		x1,   tp, x3; \
+	vpxor		RNOT, x1, x1; \
+	vpxor		x0,   x3, x3;
+
+#define S5_1(x0, x1, x2, x3, x4)      \
+	vpor		x0,   x1, tp; \
+	vpxor		tp,   x2, x2; \
+	vpxor		RNOT, x3, x3; \
+	vpxor		x0,   x1, x4; \
+	vpxor		x2,   x0, x0; \
+	vpand		x4,   tp, x1; \
+	vpor		x3,   x4, x4; \
+	vpxor		x0,   x4, x4;
+#define S5_2(x0, x1, x2, x3, x4)      \
+	vpand		x3,   x0, x0; \
+	vpxor		x3,   x1, x1; \
+	vpxor		x2,   x3, x3; \
+	vpxor		x1,   x0, x0; \
+	vpand		x4,   x2, x2; \
+	vpxor		x2,   x1, x1; \
+	vpand		x0,   x2, x2; \
+	vpxor		x2,   x3, x3;
+
+#define S6_1(x0, x1, x2, x3, x4)      \
+	vpxor		x0,   x3, x3; \
+	vpxor		x2,   x1, tp; \
+	vpxor		x0,   x2, x2; \
+	vpand		x3,   x0, x0; \
+	vpor		x3,   tp, tp; \
+	vpxor		RNOT, x1, x4; \
+	vpxor		tp,   x0, x0; \
+	vpxor		x2,   tp, x1;
+#define S6_2(x0, x1, x2, x3, x4)      \
+	vpxor		x4,   x3, x3; \
+	vpxor		x0,   x4, x4; \
+	vpand		x0,   x2, x2; \
+	vpxor		x1,   x4, x4; \
+	vpxor		x3,   x2, x2; \
+	vpand		x1,   x3, x3; \
+	vpxor		x0,   x3, x3; \
+	vpxor		x2,   x1, x1;
+
+#define S7_1(x0, x1, x2, x3, x4)      \
+	vpxor		RNOT, x1, tp; \
+	vpxor		RNOT, x0, x0; \
+	vpand		x2,   tp, x1; \
+	vpxor		x3,   x1, x1; \
+	vpor		tp,   x3, x3; \
+	vpxor		x2,   tp, x4; \
+	vpxor		x3,   x2, x2; \
+	vpxor		x0,   x3, x3; \
+	vpor		x1,   x0, x0;
+#define S7_2(x0, x1, x2, x3, x4)      \
+	vpand		x0,   x2, x2; \
+	vpxor		x4,   x0, x0; \
+	vpxor		x3,   x4, x4; \
+	vpand		x0,   x3, x3; \
+	vpxor		x1,   x4, x4; \
+	vpxor		x4,   x2, x2; \
+	vpxor		x1,   x3, x3; \
+	vpor		x0,   x4, x4; \
+	vpxor		x1,   x4, x4;
+
+#define SI0_1(x0, x1, x2, x3, x4)     \
+	vpxor		x0,   x1, x1; \
+	vpor		x1,   x3, tp; \
+	vpxor		x1,   x3, x4; \
+	vpxor		RNOT, x0, x0; \
+	vpxor		tp,   x2, x2; \
+	vpxor		x0,   tp, x3; \
+	vpand		x1,   x0, x0; \
+	vpxor		x2,   x0, x0;
+#define SI0_2(x0, x1, x2, x3, x4)     \
+	vpand		x3,   x2, x2; \
+	vpxor		x4,   x3, x3; \
+	vpxor		x3,   x2, x2; \
+	vpxor		x3,   x1, x1; \
+	vpand		x0,   x3, x3; \
+	vpxor		x0,   x1, x1; \
+	vpxor		x2,   x0, x0; \
+	vpxor		x3,   x4, x4;
+
+#define SI1_1(x0, x1, x2, x3, x4)     \
+	vpxor		x3,   x1, x1; \
+	vpxor		x2,   x0, tp; \
+	vpxor		RNOT, x2, x2; \
+	vpor		x1,   x0, x4; \
+	vpxor		x3,   x4, x4; \
+	vpand		x1,   x3, x3; \
+	vpxor		x2,   x1, x1; \
+	vpand		x4,   x2, x2;
+#define SI1_2(x0, x1, x2, x3, x4)     \
+	vpxor		x1,   x4, x4; \
+	vpor		x3,   x1, x1; \
+	vpxor		tp,   x3, x3; \
+	vpxor		tp,   x2, x2; \
+	vpor		x4,   tp, x0; \
+	vpxor		x4,   x2, x2; \
+	vpxor		x0,   x1, x1; \
+	vpxor		x1,   x4, x4;
+
+#define SI2_1(x0, x1, x2, x3, x4)     \
+	vpxor		x1,   x2, x2; \
+	vpxor		RNOT, x3, tp; \
+	vpor		x2,   tp, tp; \
+	vpxor		x3,   x2, x2; \
+	vpxor		x0,   x3, x4; \
+	vpxor		x1,   tp, x3; \
+	vpor		x2,   x1, x1; \
+	vpxor		x0,   x2, x2;
+#define SI2_2(x0, x1, x2, x3, x4)     \
+	vpxor		x4,   x1, x1; \
+	vpor		x3,   x4, x4; \
+	vpxor		x3,   x2, x2; \
+	vpxor		x2,   x4, x4; \
+	vpand		x1,   x2, x2; \
+	vpxor		x3,   x2, x2; \
+	vpxor		x4,   x3, x3; \
+	vpxor		x0,   x4, x4;
+
+#define SI3_1(x0, x1, x2, x3, x4)     \
+	vpxor		x1,   x2, x2; \
+	vpand		x2,   x1, tp; \
+	vpxor		x0,   tp, tp; \
+	vpor		x1,   x0, x0; \
+	vpxor		x3,   x1, x4; \
+	vpxor		x3,   x0, x0; \
+	vpor		tp,   x3, x3; \
+	vpxor		x2,   tp, x1;
+#define SI3_2(x0, x1, x2, x3, x4)     \
+	vpxor		x3,   x1, x1; \
+	vpxor		x2,   x0, x0; \
+	vpxor		x3,   x2, x2; \
+	vpand		x1,   x3, x3; \
+	vpxor		x0,   x1, x1; \
+	vpand		x2,   x0, x0; \
+	vpxor		x3,   x4, x4; \
+	vpxor		x0,   x3, x3; \
+	vpxor		x1,   x0, x0;
+
+#define SI4_1(x0, x1, x2, x3, x4)     \
+	vpxor		x3,   x2, x2; \
+	vpand		x1,   x0, tp; \
+	vpxor		x2,   tp, tp; \
+	vpor		x3,   x2, x2; \
+	vpxor		RNOT, x0, x4; \
+	vpxor		tp,   x1, x1; \
+	vpxor		x2,   tp, x0; \
+	vpand		x4,   x2, x2;
+#define SI4_2(x0, x1, x2, x3, x4)     \
+	vpxor		x0,   x2, x2; \
+	vpor		x4,   x0, x0; \
+	vpxor		x3,   x0, x0; \
+	vpand		x2,   x3, x3; \
+	vpxor		x3,   x4, x4; \
+	vpxor		x1,   x3, x3; \
+	vpand		x0,   x1, x1; \
+	vpxor		x1,   x4, x4; \
+	vpxor		x3,   x0, x0;
+
+#define SI5_1(x0, x1, x2, x3, x4)     \
+	vpor		x2,   x1, tp; \
+	vpxor		x1,   x2, x2; \
+	vpxor		x3,   tp, tp; \
+	vpand		x1,   x3, x3; \
+	vpxor		x3,   x2, x2; \
+	vpor		x0,   x3, x3; \
+	vpxor		RNOT, x0, x0; \
+	vpxor		x2,   x3, x3; \
+	vpor		x0,   x2, x2;
+#define SI5_2(x0, x1, x2, x3, x4)     \
+	vpxor		tp,   x1, x4; \
+	vpxor		x4,   x2, x2; \
+	vpand		x0,   x4, x4; \
+	vpxor		tp,   x0, x0; \
+	vpxor		x3,   tp, x1; \
+	vpand		x2,   x0, x0; \
+	vpxor		x3,   x2, x2; \
+	vpxor		x2,   x0, x0; \
+	vpxor		x4,   x2, x2; \
+	vpxor		x3,   x4, x4;
+
+#define SI6_1(x0, x1, x2, x3, x4)     \
+	vpxor		x2,   x0, x0; \
+	vpand		x3,   x0, tp; \
+	vpxor		x3,   x2, x2; \
+	vpxor		x2,   tp, tp; \
+	vpxor		x1,   x3, x3; \
+	vpor		x0,   x2, x2; \
+	vpxor		x3,   x2, x2; \
+	vpand		tp,   x3, x3;
+#define SI6_2(x0, x1, x2, x3, x4)     \
+	vpxor		RNOT, tp, tp; \
+	vpxor		x1,   x3, x3; \
+	vpand		x2,   x1, x1; \
+	vpxor		tp,   x0, x4; \
+	vpxor		x4,   x3, x3; \
+	vpxor		x2,   x4, x4; \
+	vpxor		x1,   tp, x0; \
+	vpxor		x0,   x2, x2;
+
+#define SI7_1(x0, x1, x2, x3, x4)     \
+	vpand		x0,   x3, tp; \
+	vpxor		x2,   x0, x0; \
+	vpor		x3,   x2, x2; \
+	vpxor		x1,   x3, x4; \
+	vpxor		RNOT, x0, x0; \
+	vpor		tp,   x1, x1; \
+	vpxor		x0,   x4, x4; \
+	vpand		x2,   x0, x0; \
+	vpxor		x1,   x0, x0;
+#define SI7_2(x0, x1, x2, x3, x4)     \
+	vpand		x2,   x1, x1; \
+	vpxor		x2,   tp, x3; \
+	vpxor		x3,   x4, x4; \
+	vpand		x3,   x2, x2; \
+	vpor		x0,   x3, x3; \
+	vpxor		x4,   x1, x1; \
+	vpxor		x4,   x3, x3; \
+	vpand		x0,   x4, x4; \
+	vpxor		x2,   x4, x4;
+
+#define get_key(i,j,t) \
+	vpbroadcastd (4*(i)+(j))*4(CTX), t;
+
+#define K2(x0, x1, x2, x3, x4, i) \
+	get_key(i, 0, RK0); \
+	get_key(i, 1, RK1); \
+	get_key(i, 2, RK2); \
+	get_key(i, 3, RK3); \
+	vpxor RK0,	x0 ## 1, x0 ## 1; \
+	vpxor RK1,	x1 ## 1, x1 ## 1; \
+	vpxor RK2,	x2 ## 1, x2 ## 1; \
+	vpxor RK3,	x3 ## 1, x3 ## 1; \
+		vpxor RK0,	x0 ## 2, x0 ## 2; \
+		vpxor RK1,	x1 ## 2, x1 ## 2; \
+		vpxor RK2,	x2 ## 2, x2 ## 2; \
+		vpxor RK3,	x3 ## 2, x3 ## 2;
+
+#define LK2(x0, x1, x2, x3, x4, i) \
+	vpslld $13,		x0 ## 1, x4 ## 1;          \
+	vpsrld $(32 - 13),	x0 ## 1, x0 ## 1;          \
+	vpor			x4 ## 1, x0 ## 1, x0 ## 1; \
+	vpxor			x0 ## 1, x1 ## 1, x1 ## 1; \
+	vpslld $3,		x2 ## 1, x4 ## 1;          \
+	vpsrld $(32 - 3),	x2 ## 1, x2 ## 1;          \
+	vpor			x4 ## 1, x2 ## 1, x2 ## 1; \
+	vpxor			x2 ## 1, x1 ## 1, x1 ## 1; \
+		vpslld $13,		x0 ## 2, x4 ## 2;          \
+		vpsrld $(32 - 13),	x0 ## 2, x0 ## 2;          \
+		vpor			x4 ## 2, x0 ## 2, x0 ## 2; \
+		vpxor			x0 ## 2, x1 ## 2, x1 ## 2; \
+		vpslld $3,		x2 ## 2, x4 ## 2;          \
+		vpsrld $(32 - 3),	x2 ## 2, x2 ## 2;          \
+		vpor			x4 ## 2, x2 ## 2, x2 ## 2; \
+		vpxor			x2 ## 2, x1 ## 2, x1 ## 2; \
+	vpslld $1,		x1 ## 1, x4 ## 1;          \
+	vpsrld $(32 - 1),	x1 ## 1, x1 ## 1;          \
+	vpor			x4 ## 1, x1 ## 1, x1 ## 1; \
+	vpslld $3,		x0 ## 1, x4 ## 1;          \
+	vpxor			x2 ## 1, x3 ## 1, x3 ## 1; \
+	vpxor			x4 ## 1, x3 ## 1, x3 ## 1; \
+	get_key(i, 1, RK1); \
+		vpslld $1,		x1 ## 2, x4 ## 2;          \
+		vpsrld $(32 - 1),	x1 ## 2, x1 ## 2;          \
+		vpor			x4 ## 2, x1 ## 2, x1 ## 2; \
+		vpslld $3,		x0 ## 2, x4 ## 2;          \
+		vpxor			x2 ## 2, x3 ## 2, x3 ## 2; \
+		vpxor			x4 ## 2, x3 ## 2, x3 ## 2; \
+		get_key(i, 3, RK3); \
+	vpslld $7,		x3 ## 1, x4 ## 1;          \
+	vpsrld $(32 - 7),	x3 ## 1, x3 ## 1;          \
+	vpor			x4 ## 1, x3 ## 1, x3 ## 1; \
+	vpslld $7,		x1 ## 1, x4 ## 1;          \
+	vpxor			x1 ## 1, x0 ## 1, x0 ## 1; \
+	vpxor			x3 ## 1, x0 ## 1, x0 ## 1; \
+	vpxor			x3 ## 1, x2 ## 1, x2 ## 1; \
+	vpxor			x4 ## 1, x2 ## 1, x2 ## 1; \
+	get_key(i, 0, RK0); \
+		vpslld $7,		x3 ## 2, x4 ## 2;          \
+		vpsrld $(32 - 7),	x3 ## 2, x3 ## 2;          \
+		vpor			x4 ## 2, x3 ## 2, x3 ## 2; \
+		vpslld $7,		x1 ## 2, x4 ## 2;          \
+		vpxor			x1 ## 2, x0 ## 2, x0 ## 2; \
+		vpxor			x3 ## 2, x0 ## 2, x0 ## 2; \
+		vpxor			x3 ## 2, x2 ## 2, x2 ## 2; \
+		vpxor			x4 ## 2, x2 ## 2, x2 ## 2; \
+		get_key(i, 2, RK2); \
+	vpxor			RK1, x1 ## 1, x1 ## 1;     \
+	vpxor			RK3, x3 ## 1, x3 ## 1;     \
+	vpslld $5,		x0 ## 1, x4 ## 1;          \
+	vpsrld $(32 - 5),	x0 ## 1, x0 ## 1;          \
+	vpor			x4 ## 1, x0 ## 1, x0 ## 1; \
+	vpslld $22,		x2 ## 1, x4 ## 1;          \
+	vpsrld $(32 - 22),	x2 ## 1, x2 ## 1;          \
+	vpor			x4 ## 1, x2 ## 1, x2 ## 1; \
+	vpxor			RK0, x0 ## 1, x0 ## 1;     \
+	vpxor			RK2, x2 ## 1, x2 ## 1;     \
+		vpxor			RK1, x1 ## 2, x1 ## 2;     \
+		vpxor			RK3, x3 ## 2, x3 ## 2;     \
+		vpslld $5,		x0 ## 2, x4 ## 2;          \
+		vpsrld $(32 - 5),	x0 ## 2, x0 ## 2;          \
+		vpor			x4 ## 2, x0 ## 2, x0 ## 2; \
+		vpslld $22,		x2 ## 2, x4 ## 2;          \
+		vpsrld $(32 - 22),	x2 ## 2, x2 ## 2;          \
+		vpor			x4 ## 2, x2 ## 2, x2 ## 2; \
+		vpxor			RK0, x0 ## 2, x0 ## 2;     \
+		vpxor			RK2, x2 ## 2, x2 ## 2;
+
+#define KL2(x0, x1, x2, x3, x4, i) \
+	vpxor			RK0, x0 ## 1, x0 ## 1;     \
+	vpxor			RK2, x2 ## 1, x2 ## 1;     \
+	vpsrld $5,		x0 ## 1, x4 ## 1;          \
+	vpslld $(32 - 5),	x0 ## 1, x0 ## 1;          \
+	vpor			x4 ## 1, x0 ## 1, x0 ## 1; \
+	vpxor			RK3, x3 ## 1, x3 ## 1;     \
+	vpxor			RK1, x1 ## 1, x1 ## 1;     \
+	vpsrld $22,		x2 ## 1, x4 ## 1;          \
+	vpslld $(32 - 22),	x2 ## 1, x2 ## 1;          \
+	vpor			x4 ## 1, x2 ## 1, x2 ## 1; \
+	vpxor			x3 ## 1, x2 ## 1, x2 ## 1; \
+		vpxor			RK0, x0 ## 2, x0 ## 2;     \
+		vpxor			RK2, x2 ## 2, x2 ## 2;     \
+		vpsrld $5,		x0 ## 2, x4 ## 2;          \
+		vpslld $(32 - 5),	x0 ## 2, x0 ## 2;          \
+		vpor			x4 ## 2, x0 ## 2, x0 ## 2; \
+		vpxor			RK3, x3 ## 2, x3 ## 2;     \
+		vpxor			RK1, x1 ## 2, x1 ## 2;     \
+		vpsrld $22,		x2 ## 2, x4 ## 2;          \
+		vpslld $(32 - 22),	x2 ## 2, x2 ## 2;          \
+		vpor			x4 ## 2, x2 ## 2, x2 ## 2; \
+		vpxor			x3 ## 2, x2 ## 2, x2 ## 2; \
+	vpxor			x3 ## 1, x0 ## 1, x0 ## 1; \
+	vpslld $7,		x1 ## 1, x4 ## 1;          \
+	vpxor			x1 ## 1, x0 ## 1, x0 ## 1; \
+	vpxor			x4 ## 1, x2 ## 1, x2 ## 1; \
+	vpsrld $1,		x1 ## 1, x4 ## 1;          \
+	vpslld $(32 - 1),	x1 ## 1, x1 ## 1;          \
+	vpor			x4 ## 1, x1 ## 1, x1 ## 1; \
+		vpxor			x3 ## 2, x0 ## 2, x0 ## 2; \
+		vpslld $7,		x1 ## 2, x4 ## 2;          \
+		vpxor			x1 ## 2, x0 ## 2, x0 ## 2; \
+		vpxor			x4 ## 2, x2 ## 2, x2 ## 2; \
+		vpsrld $1,		x1 ## 2, x4 ## 2;          \
+		vpslld $(32 - 1),	x1 ## 2, x1 ## 2;          \
+		vpor			x4 ## 2, x1 ## 2, x1 ## 2; \
+	vpsrld $7,		x3 ## 1, x4 ## 1;          \
+	vpslld $(32 - 7),	x3 ## 1, x3 ## 1;          \
+	vpor			x4 ## 1, x3 ## 1, x3 ## 1; \
+	vpxor			x0 ## 1, x1 ## 1, x1 ## 1; \
+	vpslld $3,		x0 ## 1, x4 ## 1;          \
+	vpxor			x4 ## 1, x3 ## 1, x3 ## 1; \
+		vpsrld $7,		x3 ## 2, x4 ## 2;          \
+		vpslld $(32 - 7),	x3 ## 2, x3 ## 2;          \
+		vpor			x4 ## 2, x3 ## 2, x3 ## 2; \
+		vpxor			x0 ## 2, x1 ## 2, x1 ## 2; \
+		vpslld $3,		x0 ## 2, x4 ## 2;          \
+		vpxor			x4 ## 2, x3 ## 2, x3 ## 2; \
+	vpsrld $13,		x0 ## 1, x4 ## 1;          \
+	vpslld $(32 - 13),	x0 ## 1, x0 ## 1;          \
+	vpor			x4 ## 1, x0 ## 1, x0 ## 1; \
+	vpxor			x2 ## 1, x1 ## 1, x1 ## 1; \
+	vpxor			x2 ## 1, x3 ## 1, x3 ## 1; \
+	vpsrld $3,		x2 ## 1, x4 ## 1;          \
+	vpslld $(32 - 3),	x2 ## 1, x2 ## 1;          \
+	vpor			x4 ## 1, x2 ## 1, x2 ## 1; \
+		vpsrld $13,		x0 ## 2, x4 ## 2;          \
+		vpslld $(32 - 13),	x0 ## 2, x0 ## 2;          \
+		vpor			x4 ## 2, x0 ## 2, x0 ## 2; \
+		vpxor			x2 ## 2, x1 ## 2, x1 ## 2; \
+		vpxor			x2 ## 2, x3 ## 2, x3 ## 2; \
+		vpsrld $3,		x2 ## 2, x4 ## 2;          \
+		vpslld $(32 - 3),	x2 ## 2, x2 ## 2;          \
+		vpor			x4 ## 2, x2 ## 2, x2 ## 2;
+
+#define S(SBOX, x0, x1, x2, x3, x4) \
+	SBOX ## _1(x0 ## 1, x1 ## 1, x2 ## 1, x3 ## 1, x4 ## 1); \
+	SBOX ## _2(x0 ## 1, x1 ## 1, x2 ## 1, x3 ## 1, x4 ## 1); \
+	SBOX ## _1(x0 ## 2, x1 ## 2, x2 ## 2, x3 ## 2, x4 ## 2); \
+	SBOX ## _2(x0 ## 2, x1 ## 2, x2 ## 2, x3 ## 2, x4 ## 2);
+
+#define SP(SBOX, x0, x1, x2, x3, x4, i) \
+	get_key(i, 0, RK0); \
+	SBOX ## _1(x0 ## 1, x1 ## 1, x2 ## 1, x3 ## 1, x4 ## 1); \
+	get_key(i, 2, RK2); \
+	SBOX ## _2(x0 ## 1, x1 ## 1, x2 ## 1, x3 ## 1, x4 ## 1); \
+	get_key(i, 3, RK3); \
+	SBOX ## _1(x0 ## 2, x1 ## 2, x2 ## 2, x3 ## 2, x4 ## 2); \
+	get_key(i, 1, RK1); \
+	SBOX ## _2(x0 ## 2, x1 ## 2, x2 ## 2, x3 ## 2, x4 ## 2); \
+
+#define transpose_4x4(x0, x1, x2, x3, t0, t1, t2) \
+	vpunpckldq		x1, x0, t0; \
+	vpunpckhdq		x1, x0, t2; \
+	vpunpckldq		x3, x2, t1; \
+	vpunpckhdq		x3, x2, x3; \
+	\
+	vpunpcklqdq		t1, t0, x0; \
+	vpunpckhqdq		t1, t0, x1; \
+	vpunpcklqdq		x3, t2, x2; \
+	vpunpckhqdq		x3, t2, x3;
+
+#define read_blocks(x0, x1, x2, x3, t0, t1, t2) \
+	transpose_4x4(x0, x1, x2, x3, t0, t1, t2)
+
+#define write_blocks(x0, x1, x2, x3, t0, t1, t2) \
+	transpose_4x4(x0, x1, x2, x3, t0, t1, t2)
+
+.align 8
+__serpent_enc_blk16:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: plaintext
+	 * output:
+	 *	RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: ciphertext
+	 */
+
+	vpcmpeqd RNOT, RNOT, RNOT;
+
+	read_blocks(RA1, RB1, RC1, RD1, RK0, RK1, RK2);
+	read_blocks(RA2, RB2, RC2, RD2, RK0, RK1, RK2);
+
+						 K2(RA, RB, RC, RD, RE, 0);
+	S(S0, RA, RB, RC, RD, RE);		LK2(RC, RB, RD, RA, RE, 1);
+	S(S1, RC, RB, RD, RA, RE);		LK2(RE, RD, RA, RC, RB, 2);
+	S(S2, RE, RD, RA, RC, RB);		LK2(RB, RD, RE, RC, RA, 3);
+	S(S3, RB, RD, RE, RC, RA);		LK2(RC, RA, RD, RB, RE, 4);
+	S(S4, RC, RA, RD, RB, RE);		LK2(RA, RD, RB, RE, RC, 5);
+	S(S5, RA, RD, RB, RE, RC);		LK2(RC, RA, RD, RE, RB, 6);
+	S(S6, RC, RA, RD, RE, RB);		LK2(RD, RB, RA, RE, RC, 7);
+	S(S7, RD, RB, RA, RE, RC);		LK2(RC, RA, RE, RD, RB, 8);
+	S(S0, RC, RA, RE, RD, RB);		LK2(RE, RA, RD, RC, RB, 9);
+	S(S1, RE, RA, RD, RC, RB);		LK2(RB, RD, RC, RE, RA, 10);
+	S(S2, RB, RD, RC, RE, RA);		LK2(RA, RD, RB, RE, RC, 11);
+	S(S3, RA, RD, RB, RE, RC);		LK2(RE, RC, RD, RA, RB, 12);
+	S(S4, RE, RC, RD, RA, RB);		LK2(RC, RD, RA, RB, RE, 13);
+	S(S5, RC, RD, RA, RB, RE);		LK2(RE, RC, RD, RB, RA, 14);
+	S(S6, RE, RC, RD, RB, RA);		LK2(RD, RA, RC, RB, RE, 15);
+	S(S7, RD, RA, RC, RB, RE);		LK2(RE, RC, RB, RD, RA, 16);
+	S(S0, RE, RC, RB, RD, RA);		LK2(RB, RC, RD, RE, RA, 17);
+	S(S1, RB, RC, RD, RE, RA);		LK2(RA, RD, RE, RB, RC, 18);
+	S(S2, RA, RD, RE, RB, RC);		LK2(RC, RD, RA, RB, RE, 19);
+	S(S3, RC, RD, RA, RB, RE);		LK2(RB, RE, RD, RC, RA, 20);
+	S(S4, RB, RE, RD, RC, RA);		LK2(RE, RD, RC, RA, RB, 21);
+	S(S5, RE, RD, RC, RA, RB);		LK2(RB, RE, RD, RA, RC, 22);
+	S(S6, RB, RE, RD, RA, RC);		LK2(RD, RC, RE, RA, RB, 23);
+	S(S7, RD, RC, RE, RA, RB);		LK2(RB, RE, RA, RD, RC, 24);
+	S(S0, RB, RE, RA, RD, RC);		LK2(RA, RE, RD, RB, RC, 25);
+	S(S1, RA, RE, RD, RB, RC);		LK2(RC, RD, RB, RA, RE, 26);
+	S(S2, RC, RD, RB, RA, RE);		LK2(RE, RD, RC, RA, RB, 27);
+	S(S3, RE, RD, RC, RA, RB);		LK2(RA, RB, RD, RE, RC, 28);
+	S(S4, RA, RB, RD, RE, RC);		LK2(RB, RD, RE, RC, RA, 29);
+	S(S5, RB, RD, RE, RC, RA);		LK2(RA, RB, RD, RC, RE, 30);
+	S(S6, RA, RB, RD, RC, RE);		LK2(RD, RE, RB, RC, RA, 31);
+	S(S7, RD, RE, RB, RC, RA);		 K2(RA, RB, RC, RD, RE, 32);
+
+	write_blocks(RA1, RB1, RC1, RD1, RK0, RK1, RK2);
+	write_blocks(RA2, RB2, RC2, RD2, RK0, RK1, RK2);
+
+	ret;
+ENDPROC(__serpent_enc_blk16)
+
+.align 8
+__serpent_dec_blk16:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2: ciphertext
+	 * output:
+	 *	RC1, RD1, RB1, RE1, RC2, RD2, RB2, RE2: plaintext
+	 */
+
+	vpcmpeqd RNOT, RNOT, RNOT;
+
+	read_blocks(RA1, RB1, RC1, RD1, RK0, RK1, RK2);
+	read_blocks(RA2, RB2, RC2, RD2, RK0, RK1, RK2);
+
+						 K2(RA, RB, RC, RD, RE, 32);
+	SP(SI7, RA, RB, RC, RD, RE, 31);	KL2(RB, RD, RA, RE, RC, 31);
+	SP(SI6, RB, RD, RA, RE, RC, 30);	KL2(RA, RC, RE, RB, RD, 30);
+	SP(SI5, RA, RC, RE, RB, RD, 29);	KL2(RC, RD, RA, RE, RB, 29);
+	SP(SI4, RC, RD, RA, RE, RB, 28);	KL2(RC, RA, RB, RE, RD, 28);
+	SP(SI3, RC, RA, RB, RE, RD, 27);	KL2(RB, RC, RD, RE, RA, 27);
+	SP(SI2, RB, RC, RD, RE, RA, 26);	KL2(RC, RA, RE, RD, RB, 26);
+	SP(SI1, RC, RA, RE, RD, RB, 25);	KL2(RB, RA, RE, RD, RC, 25);
+	SP(SI0, RB, RA, RE, RD, RC, 24);	KL2(RE, RC, RA, RB, RD, 24);
+	SP(SI7, RE, RC, RA, RB, RD, 23);	KL2(RC, RB, RE, RD, RA, 23);
+	SP(SI6, RC, RB, RE, RD, RA, 22);	KL2(RE, RA, RD, RC, RB, 22);
+	SP(SI5, RE, RA, RD, RC, RB, 21);	KL2(RA, RB, RE, RD, RC, 21);
+	SP(SI4, RA, RB, RE, RD, RC, 20);	KL2(RA, RE, RC, RD, RB, 20);
+	SP(SI3, RA, RE, RC, RD, RB, 19);	KL2(RC, RA, RB, RD, RE, 19);
+	SP(SI2, RC, RA, RB, RD, RE, 18);	KL2(RA, RE, RD, RB, RC, 18);
+	SP(SI1, RA, RE, RD, RB, RC, 17);	KL2(RC, RE, RD, RB, RA, 17);
+	SP(SI0, RC, RE, RD, RB, RA, 16);	KL2(RD, RA, RE, RC, RB, 16);
+	SP(SI7, RD, RA, RE, RC, RB, 15);	KL2(RA, RC, RD, RB, RE, 15);
+	SP(SI6, RA, RC, RD, RB, RE, 14);	KL2(RD, RE, RB, RA, RC, 14);
+	SP(SI5, RD, RE, RB, RA, RC, 13);	KL2(RE, RC, RD, RB, RA, 13);
+	SP(SI4, RE, RC, RD, RB, RA, 12);	KL2(RE, RD, RA, RB, RC, 12);
+	SP(SI3, RE, RD, RA, RB, RC, 11);	KL2(RA, RE, RC, RB, RD, 11);
+	SP(SI2, RA, RE, RC, RB, RD, 10);	KL2(RE, RD, RB, RC, RA, 10);
+	SP(SI1, RE, RD, RB, RC, RA, 9);		KL2(RA, RD, RB, RC, RE, 9);
+	SP(SI0, RA, RD, RB, RC, RE, 8);		KL2(RB, RE, RD, RA, RC, 8);
+	SP(SI7, RB, RE, RD, RA, RC, 7);		KL2(RE, RA, RB, RC, RD, 7);
+	SP(SI6, RE, RA, RB, RC, RD, 6);		KL2(RB, RD, RC, RE, RA, 6);
+	SP(SI5, RB, RD, RC, RE, RA, 5);		KL2(RD, RA, RB, RC, RE, 5);
+	SP(SI4, RD, RA, RB, RC, RE, 4);		KL2(RD, RB, RE, RC, RA, 4);
+	SP(SI3, RD, RB, RE, RC, RA, 3);		KL2(RE, RD, RA, RC, RB, 3);
+	SP(SI2, RE, RD, RA, RC, RB, 2);		KL2(RD, RB, RC, RA, RE, 2);
+	SP(SI1, RD, RB, RC, RA, RE, 1);		KL2(RE, RB, RC, RA, RD, 1);
+	S(SI0, RE, RB, RC, RA, RD);		 K2(RC, RD, RB, RE, RA, 0);
+
+	write_blocks(RC1, RD1, RB1, RE1, RK0, RK1, RK2);
+	write_blocks(RC2, RD2, RB2, RE2, RK0, RK1, RK2);
+
+	ret;
+ENDPROC(__serpent_dec_blk16)
+
+ENTRY(serpent_ecb_enc_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+
+	load_16way(%rdx, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	call __serpent_enc_blk16;
+
+	store_16way(%rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(serpent_ecb_enc_16way)
+
+ENTRY(serpent_ecb_dec_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+
+	load_16way(%rdx, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	call __serpent_dec_blk16;
+
+	store_16way(%rsi, RC1, RD1, RB1, RE1, RC2, RD2, RB2, RE2);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(serpent_ecb_dec_16way)
+
+ENTRY(serpent_cbc_dec_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+
+	load_16way(%rdx, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	call __serpent_dec_blk16;
+
+	store_cbc_16way(%rdx, %rsi, RC1, RD1, RB1, RE1, RC2, RD2, RB2, RE2,
+			RK0);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(serpent_cbc_dec_16way)
+
+ENTRY(serpent_ctr_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (little endian, 128bit)
+	 */
+
+	vzeroupper;
+
+	load_ctr_16way(%rcx, .Lbswap128_mask, RA1, RB1, RC1, RD1, RA2, RB2, RC2,
+		       RD2, RK0, RK0x, RK1, RK1x, RK2, RK2x, RK3, RK3x, RNOT,
+		       tp);
+
+	call __serpent_enc_blk16;
+
+	store_ctr_16way(%rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(serpent_ctr_16way)
+
+ENTRY(serpent_xts_enc_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	vzeroupper;
+
+	load_xts_16way(%rcx, %rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2,
+		       RD2, RK0, RK0x, RK1, RK1x, RK2, RK2x, RK3, RK3x, RNOT,
+		       .Lxts_gf128mul_and_shl1_mask_0,
+		       .Lxts_gf128mul_and_shl1_mask_1);
+
+	call __serpent_enc_blk16;
+
+	store_xts_16way(%rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(serpent_xts_enc_16way)
+
+ENTRY(serpent_xts_dec_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	vzeroupper;
+
+	load_xts_16way(%rcx, %rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2,
+		       RD2, RK0, RK0x, RK1, RK1x, RK2, RK2x, RK3, RK3x, RNOT,
+		       .Lxts_gf128mul_and_shl1_mask_0,
+		       .Lxts_gf128mul_and_shl1_mask_1);
+
+	call __serpent_dec_blk16;
+
+	store_xts_16way(%rsi, RC1, RD1, RB1, RE1, RC2, RD2, RB2, RE2);
+
+	vzeroupper;
+
+	ret;
+ENDPROC(serpent_xts_dec_16way)
diff --git a/arch/x86/crypto/serpent_avx2_glue.c b/arch/x86/crypto/serpent_avx2_glue.c
new file mode 100644
index 000000000000..23aabc6c20a5
--- /dev/null
+++ b/arch/x86/crypto/serpent_avx2_glue.c
@@ -0,0 +1,562 @@
+/*
+ * Glue Code for x86_64/AVX2 assembler optimized version of Serpent
+ *
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ *
+ * 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.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/crypto.h>
+#include <linux/err.h>
+#include <crypto/algapi.h>
+#include <crypto/ctr.h>
+#include <crypto/lrw.h>
+#include <crypto/xts.h>
+#include <crypto/serpent.h>
+#include <asm/xcr.h>
+#include <asm/xsave.h>
+#include <asm/crypto/serpent-avx.h>
+#include <asm/crypto/ablk_helper.h>
+#include <asm/crypto/glue_helper.h>
+
+#define SERPENT_AVX2_PARALLEL_BLOCKS 16
+
+/* 16-way AVX2 parallel cipher functions */
+asmlinkage void serpent_ecb_enc_16way(struct serpent_ctx *ctx, u8 *dst,
+				      const u8 *src);
+asmlinkage void serpent_ecb_dec_16way(struct serpent_ctx *ctx, u8 *dst,
+				      const u8 *src);
+asmlinkage void serpent_cbc_dec_16way(void *ctx, u128 *dst, const u128 *src);
+
+asmlinkage void serpent_ctr_16way(void *ctx, u128 *dst, const u128 *src,
+				  le128 *iv);
+asmlinkage void serpent_xts_enc_16way(struct serpent_ctx *ctx, u8 *dst,
+				      const u8 *src, le128 *iv);
+asmlinkage void serpent_xts_dec_16way(struct serpent_ctx *ctx, u8 *dst,
+				      const u8 *src, le128 *iv);
+
+static const struct common_glue_ctx serpent_enc = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(serpent_ecb_enc_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(serpent_ecb_enc_8way_avx) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(__serpent_encrypt) }
+	} }
+};
+
+static const struct common_glue_ctx serpent_ctr = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(serpent_ctr_16way) }
+	},  {
+		.num_blocks = 8,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(serpent_ctr_8way_avx) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(__serpent_crypt_ctr) }
+	} }
+};
+
+static const struct common_glue_ctx serpent_enc_xts = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_enc_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_enc_8way_avx) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_enc) }
+	} }
+};
+
+static const struct common_glue_ctx serpent_dec = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(serpent_ecb_dec_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(serpent_ecb_dec_8way_avx) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(__serpent_decrypt) }
+	} }
+};
+
+static const struct common_glue_ctx serpent_dec_cbc = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(serpent_cbc_dec_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(serpent_cbc_dec_8way_avx) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(__serpent_decrypt) }
+	} }
+};
+
+static const struct common_glue_ctx serpent_dec_xts = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_dec_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_dec_8way_avx) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_dec) }
+	} }
+};
+
+static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ecb_crypt_128bit(&serpent_enc, desc, dst, src, nbytes);
+}
+
+static int ecb_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ecb_crypt_128bit(&serpent_dec, desc, dst, src, nbytes);
+}
+
+static int cbc_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_cbc_encrypt_128bit(GLUE_FUNC_CAST(__serpent_encrypt), desc,
+				       dst, src, nbytes);
+}
+
+static int cbc_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_cbc_decrypt_128bit(&serpent_dec_cbc, desc, dst, src,
+				       nbytes);
+}
+
+static int ctr_crypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		     struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ctr_crypt_128bit(&serpent_ctr, desc, dst, src, nbytes);
+}
+
+static inline bool serpent_fpu_begin(bool fpu_enabled, unsigned int nbytes)
+{
+	/* since reusing AVX functions, starts using FPU at 8 parallel blocks */
+	return glue_fpu_begin(SERPENT_BLOCK_SIZE, 8, NULL, fpu_enabled, nbytes);
+}
+
+static inline void serpent_fpu_end(bool fpu_enabled)
+{
+	glue_fpu_end(fpu_enabled);
+}
+
+struct crypt_priv {
+	struct serpent_ctx *ctx;
+	bool fpu_enabled;
+};
+
+static void encrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
+{
+	const unsigned int bsize = SERPENT_BLOCK_SIZE;
+	struct crypt_priv *ctx = priv;
+	int i;
+
+	ctx->fpu_enabled = serpent_fpu_begin(ctx->fpu_enabled, nbytes);
+
+	if (nbytes >= SERPENT_AVX2_PARALLEL_BLOCKS * bsize) {
+		serpent_ecb_enc_16way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * SERPENT_AVX2_PARALLEL_BLOCKS;
+		nbytes -= bsize * SERPENT_AVX2_PARALLEL_BLOCKS;
+	}
+
+	while (nbytes >= SERPENT_PARALLEL_BLOCKS * bsize) {
+		serpent_ecb_enc_8way_avx(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * SERPENT_PARALLEL_BLOCKS;
+		nbytes -= bsize * SERPENT_PARALLEL_BLOCKS;
+	}
+
+	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
+		__serpent_encrypt(ctx->ctx, srcdst, srcdst);
+}
+
+static void decrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
+{
+	const unsigned int bsize = SERPENT_BLOCK_SIZE;
+	struct crypt_priv *ctx = priv;
+	int i;
+
+	ctx->fpu_enabled = serpent_fpu_begin(ctx->fpu_enabled, nbytes);
+
+	if (nbytes >= SERPENT_AVX2_PARALLEL_BLOCKS * bsize) {
+		serpent_ecb_dec_16way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * SERPENT_AVX2_PARALLEL_BLOCKS;
+		nbytes -= bsize * SERPENT_AVX2_PARALLEL_BLOCKS;
+	}
+
+	while (nbytes >= SERPENT_PARALLEL_BLOCKS * bsize) {
+		serpent_ecb_dec_8way_avx(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * SERPENT_PARALLEL_BLOCKS;
+		nbytes -= bsize * SERPENT_PARALLEL_BLOCKS;
+	}
+
+	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
+		__serpent_decrypt(ctx->ctx, srcdst, srcdst);
+}
+
+static int lrw_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct serpent_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	be128 buf[SERPENT_AVX2_PARALLEL_BLOCKS];
+	struct crypt_priv crypt_ctx = {
+		.ctx = &ctx->serpent_ctx,
+		.fpu_enabled = false,
+	};
+	struct lrw_crypt_req req = {
+		.tbuf = buf,
+		.tbuflen = sizeof(buf),
+
+		.table_ctx = &ctx->lrw_table,
+		.crypt_ctx = &crypt_ctx,
+		.crypt_fn = encrypt_callback,
+	};
+	int ret;
+
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+	ret = lrw_crypt(desc, dst, src, nbytes, &req);
+	serpent_fpu_end(crypt_ctx.fpu_enabled);
+
+	return ret;
+}
+
+static int lrw_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct serpent_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	be128 buf[SERPENT_AVX2_PARALLEL_BLOCKS];
+	struct crypt_priv crypt_ctx = {
+		.ctx = &ctx->serpent_ctx,
+		.fpu_enabled = false,
+	};
+	struct lrw_crypt_req req = {
+		.tbuf = buf,
+		.tbuflen = sizeof(buf),
+
+		.table_ctx = &ctx->lrw_table,
+		.crypt_ctx = &crypt_ctx,
+		.crypt_fn = decrypt_callback,
+	};
+	int ret;
+
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+	ret = lrw_crypt(desc, dst, src, nbytes, &req);
+	serpent_fpu_end(crypt_ctx.fpu_enabled);
+
+	return ret;
+}
+
+static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct serpent_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+
+	return glue_xts_crypt_128bit(&serpent_enc_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(__serpent_encrypt),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
+}
+
+static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct serpent_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+
+	return glue_xts_crypt_128bit(&serpent_dec_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(__serpent_encrypt),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
+}
+
+static struct crypto_alg srp_algs[10] = { {
+	.cra_name		= "__ecb-serpent-avx2",
+	.cra_driver_name	= "__driver-ecb-serpent-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= SERPENT_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct serpent_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[0].cra_list),
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE,
+			.setkey		= serpent_setkey,
+			.encrypt	= ecb_encrypt,
+			.decrypt	= ecb_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__cbc-serpent-avx2",
+	.cra_driver_name	= "__driver-cbc-serpent-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= SERPENT_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct serpent_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[1].cra_list),
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE,
+			.setkey		= serpent_setkey,
+			.encrypt	= cbc_encrypt,
+			.decrypt	= cbc_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__ctr-serpent-avx2",
+	.cra_driver_name	= "__driver-ctr-serpent-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= 1,
+	.cra_ctxsize		= sizeof(struct serpent_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[2].cra_list),
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE,
+			.ivsize		= SERPENT_BLOCK_SIZE,
+			.setkey		= serpent_setkey,
+			.encrypt	= ctr_crypt,
+			.decrypt	= ctr_crypt,
+		},
+	},
+}, {
+	.cra_name		= "__lrw-serpent-avx2",
+	.cra_driver_name	= "__driver-lrw-serpent-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= SERPENT_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct serpent_lrw_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[3].cra_list),
+	.cra_exit		= lrw_serpent_exit_tfm,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE +
+					  SERPENT_BLOCK_SIZE,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE +
+					  SERPENT_BLOCK_SIZE,
+			.ivsize		= SERPENT_BLOCK_SIZE,
+			.setkey		= lrw_serpent_setkey,
+			.encrypt	= lrw_encrypt,
+			.decrypt	= lrw_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__xts-serpent-avx2",
+	.cra_driver_name	= "__driver-xts-serpent-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= SERPENT_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct serpent_xts_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[4].cra_list),
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE * 2,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE * 2,
+			.ivsize		= SERPENT_BLOCK_SIZE,
+			.setkey		= xts_serpent_setkey,
+			.encrypt	= xts_encrypt,
+			.decrypt	= xts_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "ecb(serpent)",
+	.cra_driver_name	= "ecb-serpent-avx2",
+	.cra_priority		= 600,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= SERPENT_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[5].cra_list),
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "cbc(serpent)",
+	.cra_driver_name	= "cbc-serpent-avx2",
+	.cra_priority		= 600,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= SERPENT_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[6].cra_list),
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE,
+			.ivsize		= SERPENT_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= __ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "ctr(serpent)",
+	.cra_driver_name	= "ctr-serpent-avx2",
+	.cra_priority		= 600,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= 1,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[7].cra_list),
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE,
+			.ivsize		= SERPENT_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_encrypt,
+			.geniv		= "chainiv",
+		},
+	},
+}, {
+	.cra_name		= "lrw(serpent)",
+	.cra_driver_name	= "lrw-serpent-avx2",
+	.cra_priority		= 600,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= SERPENT_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[8].cra_list),
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE +
+					  SERPENT_BLOCK_SIZE,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE +
+					  SERPENT_BLOCK_SIZE,
+			.ivsize		= SERPENT_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "xts(serpent)",
+	.cra_driver_name	= "xts-serpent-avx2",
+	.cra_priority		= 600,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= SERPENT_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_list		= LIST_HEAD_INIT(srp_algs[9].cra_list),
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= SERPENT_MIN_KEY_SIZE * 2,
+			.max_keysize	= SERPENT_MAX_KEY_SIZE * 2,
+			.ivsize		= SERPENT_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+} };
+
+static int __init init(void)
+{
+	u64 xcr0;
+
+	if (!cpu_has_avx2 || !cpu_has_osxsave) {
+		pr_info("AVX2 instructions are not detected.\n");
+		return -ENODEV;
+	}
+
+	xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
+	if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
+		pr_info("AVX detected but unusable.\n");
+		return -ENODEV;
+	}
+
+	return crypto_register_algs(srp_algs, ARRAY_SIZE(srp_algs));
+}
+
+static void __exit fini(void)
+{
+	crypto_unregister_algs(srp_algs, ARRAY_SIZE(srp_algs));
+}
+
+module_init(init);
+module_exit(fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Serpent Cipher Algorithm, AVX2 optimized");
+MODULE_ALIAS("serpent");
+MODULE_ALIAS("serpent-asm");
diff --git a/arch/x86/crypto/serpent_avx_glue.c b/arch/x86/crypto/serpent_avx_glue.c
index 52abaaf28e7f..9ae83cf8d21e 100644
--- a/arch/x86/crypto/serpent_avx_glue.c
+++ b/arch/x86/crypto/serpent_avx_glue.c
@@ -4,8 +4,7 @@
  * Copyright (C) 2012 Johannes Goetzfried
  *     <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
  *
- * Glue code based on serpent_sse2_glue.c by:
- *  Copyright (C) 2011 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2011-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * 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
@@ -42,7 +41,32 @@
 #include <asm/crypto/ablk_helper.h>
 #include <asm/crypto/glue_helper.h>
 
-static void serpent_crypt_ctr(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+/* 8-way parallel cipher functions */
+asmlinkage void serpent_ecb_enc_8way_avx(struct serpent_ctx *ctx, u8 *dst,
+					 const u8 *src);
+EXPORT_SYMBOL_GPL(serpent_ecb_enc_8way_avx);
+
+asmlinkage void serpent_ecb_dec_8way_avx(struct serpent_ctx *ctx, u8 *dst,
+					 const u8 *src);
+EXPORT_SYMBOL_GPL(serpent_ecb_dec_8way_avx);
+
+asmlinkage void serpent_cbc_dec_8way_avx(struct serpent_ctx *ctx, u8 *dst,
+					 const u8 *src);
+EXPORT_SYMBOL_GPL(serpent_cbc_dec_8way_avx);
+
+asmlinkage void serpent_ctr_8way_avx(struct serpent_ctx *ctx, u8 *dst,
+				     const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(serpent_ctr_8way_avx);
+
+asmlinkage void serpent_xts_enc_8way_avx(struct serpent_ctx *ctx, u8 *dst,
+					 const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(serpent_xts_enc_8way_avx);
+
+asmlinkage void serpent_xts_dec_8way_avx(struct serpent_ctx *ctx, u8 *dst,
+					 const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(serpent_xts_dec_8way_avx);
+
+void __serpent_crypt_ctr(void *ctx, u128 *dst, const u128 *src, le128 *iv)
 {
 	be128 ctrblk;
 
@@ -52,6 +76,22 @@ static void serpent_crypt_ctr(void *ctx, u128 *dst, const u128 *src, le128 *iv)
 	__serpent_encrypt(ctx, (u8 *)&ctrblk, (u8 *)&ctrblk);
 	u128_xor(dst, src, (u128 *)&ctrblk);
 }
+EXPORT_SYMBOL_GPL(__serpent_crypt_ctr);
+
+void serpent_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv,
+				  GLUE_FUNC_CAST(__serpent_encrypt));
+}
+EXPORT_SYMBOL_GPL(serpent_xts_enc);
+
+void serpent_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv,
+				  GLUE_FUNC_CAST(__serpent_decrypt));
+}
+EXPORT_SYMBOL_GPL(serpent_xts_dec);
+
 
 static const struct common_glue_ctx serpent_enc = {
 	.num_funcs = 2,
@@ -75,7 +115,20 @@ static const struct common_glue_ctx serpent_ctr = {
 		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(serpent_ctr_8way_avx) }
 	}, {
 		.num_blocks = 1,
-		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(serpent_crypt_ctr) }
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(__serpent_crypt_ctr) }
+	} }
+};
+
+static const struct common_glue_ctx serpent_enc_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = SERPENT_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = SERPENT_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_enc_8way_avx) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_enc) }
 	} }
 };
 
@@ -105,6 +158,19 @@ static const struct common_glue_ctx serpent_dec_cbc = {
 	} }
 };
 
+static const struct common_glue_ctx serpent_dec_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = SERPENT_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = SERPENT_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_dec_8way_avx) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(serpent_xts_dec) }
+	} }
+};
+
 static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
@@ -187,13 +253,8 @@ static void decrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
 		__serpent_decrypt(ctx->ctx, srcdst, srcdst);
 }
 
-struct serpent_lrw_ctx {
-	struct lrw_table_ctx lrw_table;
-	struct serpent_ctx serpent_ctx;
-};
-
-static int lrw_serpent_setkey(struct crypto_tfm *tfm, const u8 *key,
-			      unsigned int keylen)
+int lrw_serpent_setkey(struct crypto_tfm *tfm, const u8 *key,
+		       unsigned int keylen)
 {
 	struct serpent_lrw_ctx *ctx = crypto_tfm_ctx(tfm);
 	int err;
@@ -206,6 +267,7 @@ static int lrw_serpent_setkey(struct crypto_tfm *tfm, const u8 *key,
 	return lrw_init_table(&ctx->lrw_table, key + keylen -
 						SERPENT_BLOCK_SIZE);
 }
+EXPORT_SYMBOL_GPL(lrw_serpent_setkey);
 
 static int lrw_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
@@ -259,20 +321,16 @@ static int lrw_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 	return ret;
 }
 
-static void lrw_exit_tfm(struct crypto_tfm *tfm)
+void lrw_serpent_exit_tfm(struct crypto_tfm *tfm)
 {
 	struct serpent_lrw_ctx *ctx = crypto_tfm_ctx(tfm);
 
 	lrw_free_table(&ctx->lrw_table);
 }
+EXPORT_SYMBOL_GPL(lrw_serpent_exit_tfm);
 
-struct serpent_xts_ctx {
-	struct serpent_ctx tweak_ctx;
-	struct serpent_ctx crypt_ctx;
-};
-
-static int xts_serpent_setkey(struct crypto_tfm *tfm, const u8 *key,
-			      unsigned int keylen)
+int xts_serpent_setkey(struct crypto_tfm *tfm, const u8 *key,
+		       unsigned int keylen)
 {
 	struct serpent_xts_ctx *ctx = crypto_tfm_ctx(tfm);
 	u32 *flags = &tfm->crt_flags;
@@ -294,59 +352,26 @@ static int xts_serpent_setkey(struct crypto_tfm *tfm, const u8 *key,
 	/* second half of xts-key is for tweak */
 	return __serpent_setkey(&ctx->tweak_ctx, key + keylen / 2, keylen / 2);
 }
+EXPORT_SYMBOL_GPL(xts_serpent_setkey);
 
 static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
 	struct serpent_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-	be128 buf[SERPENT_PARALLEL_BLOCKS];
-	struct crypt_priv crypt_ctx = {
-		.ctx = &ctx->crypt_ctx,
-		.fpu_enabled = false,
-	};
-	struct xts_crypt_req req = {
-		.tbuf = buf,
-		.tbuflen = sizeof(buf),
-
-		.tweak_ctx = &ctx->tweak_ctx,
-		.tweak_fn = XTS_TWEAK_CAST(__serpent_encrypt),
-		.crypt_ctx = &crypt_ctx,
-		.crypt_fn = encrypt_callback,
-	};
-	int ret;
-
-	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
-	ret = xts_crypt(desc, dst, src, nbytes, &req);
-	serpent_fpu_end(crypt_ctx.fpu_enabled);
 
-	return ret;
+	return glue_xts_crypt_128bit(&serpent_enc_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(__serpent_encrypt),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
 }
 
 static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
 	struct serpent_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-	be128 buf[SERPENT_PARALLEL_BLOCKS];
-	struct crypt_priv crypt_ctx = {
-		.ctx = &ctx->crypt_ctx,
-		.fpu_enabled = false,
-	};
-	struct xts_crypt_req req = {
-		.tbuf = buf,
-		.tbuflen = sizeof(buf),
-
-		.tweak_ctx = &ctx->tweak_ctx,
-		.tweak_fn = XTS_TWEAK_CAST(__serpent_encrypt),
-		.crypt_ctx = &crypt_ctx,
-		.crypt_fn = decrypt_callback,
-	};
-	int ret;
 
-	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
-	ret = xts_crypt(desc, dst, src, nbytes, &req);
-	serpent_fpu_end(crypt_ctx.fpu_enabled);
-
-	return ret;
+	return glue_xts_crypt_128bit(&serpent_dec_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(__serpent_encrypt),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
 }
 
 static struct crypto_alg serpent_algs[10] = { {
@@ -417,7 +442,7 @@ static struct crypto_alg serpent_algs[10] = { {
 	.cra_alignmask		= 0,
 	.cra_type		= &crypto_blkcipher_type,
 	.cra_module		= THIS_MODULE,
-	.cra_exit		= lrw_exit_tfm,
+	.cra_exit		= lrw_serpent_exit_tfm,
 	.cra_u = {
 		.blkcipher = {
 			.min_keysize	= SERPENT_MIN_KEY_SIZE +
diff --git a/arch/x86/crypto/sha256-avx-asm.S b/arch/x86/crypto/sha256-avx-asm.S
new file mode 100644
index 000000000000..56610c4bf31b
--- /dev/null
+++ b/arch/x86/crypto/sha256-avx-asm.S
@@ -0,0 +1,496 @@
+########################################################################
+# Implement fast SHA-256 with AVX1 instructions. (x86_64)
+#
+# Copyright (C) 2013 Intel Corporation.
+#
+# Authors:
+#     James Guilford <james.guilford@intel.com>
+#     Kirk Yap <kirk.s.yap@intel.com>
+#     Tim Chen <tim.c.chen@linux.intel.com>
+#
+# This software is available to you under a choice of one of two
+# licenses.  You may choose to be licensed under the terms of the GNU
+# General Public License (GPL) Version 2, available from the file
+# COPYING in the main directory of this source tree, or the
+# OpenIB.org BSD license below:
+#
+#     Redistribution and use in source and binary forms, with or
+#     without modification, are permitted provided that the following
+#     conditions are met:
+#
+#      - Redistributions of source code must retain the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer.
+#
+#      - Redistributions in binary form must reproduce the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer in the documentation and/or other materials
+#        provided with the distribution.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+# BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+# ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+########################################################################
+#
+# This code is described in an Intel White-Paper:
+# "Fast SHA-256 Implementations on Intel Architecture Processors"
+#
+# To find it, surf to http://www.intel.com/p/en_US/embedded
+# and search for that title.
+#
+########################################################################
+# This code schedules 1 block at a time, with 4 lanes per block
+########################################################################
+
+#ifdef CONFIG_AS_AVX
+#include <linux/linkage.h>
+
+## assume buffers not aligned
+#define    VMOVDQ vmovdqu
+
+################################ Define Macros
+
+# addm [mem], reg
+# Add reg to mem using reg-mem add and store
+.macro addm p1 p2
+	add     \p1, \p2
+	mov     \p2, \p1
+.endm
+
+
+.macro MY_ROR p1 p2
+	shld    $(32-(\p1)), \p2, \p2
+.endm
+
+################################
+
+# COPY_XMM_AND_BSWAP xmm, [mem], byte_flip_mask
+# Load xmm with mem and byte swap each dword
+.macro COPY_XMM_AND_BSWAP p1 p2 p3
+	VMOVDQ \p2, \p1
+	vpshufb \p3, \p1, \p1
+.endm
+
+################################
+
+X0 = %xmm4
+X1 = %xmm5
+X2 = %xmm6
+X3 = %xmm7
+
+XTMP0 = %xmm0
+XTMP1 = %xmm1
+XTMP2 = %xmm2
+XTMP3 = %xmm3
+XTMP4 = %xmm8
+XFER = %xmm9
+XTMP5 = %xmm11
+
+SHUF_00BA = %xmm10      # shuffle xBxA -> 00BA
+SHUF_DC00 = %xmm12      # shuffle xDxC -> DC00
+BYTE_FLIP_MASK = %xmm13
+
+NUM_BLKS = %rdx   # 3rd arg
+CTX = %rsi        # 2nd arg
+INP = %rdi        # 1st arg
+
+SRND = %rdi       # clobbers INP
+c = %ecx
+d = %r8d
+e = %edx
+TBL = %rbp
+a = %eax
+b = %ebx
+
+f = %r9d
+g = %r10d
+h = %r11d
+
+y0 = %r13d
+y1 = %r14d
+y2 = %r15d
+
+
+_INP_END_SIZE = 8
+_INP_SIZE = 8
+_XFER_SIZE = 8
+_XMM_SAVE_SIZE = 0
+
+_INP_END = 0
+_INP            = _INP_END  + _INP_END_SIZE
+_XFER           = _INP      + _INP_SIZE
+_XMM_SAVE       = _XFER     + _XFER_SIZE
+STACK_SIZE      = _XMM_SAVE + _XMM_SAVE_SIZE
+
+# rotate_Xs
+# Rotate values of symbols X0...X3
+.macro rotate_Xs
+X_ = X0
+X0 = X1
+X1 = X2
+X2 = X3
+X3 = X_
+.endm
+
+# ROTATE_ARGS
+# Rotate values of symbols a...h
+.macro ROTATE_ARGS
+TMP_ = h
+h = g
+g = f
+f = e
+e = d
+d = c
+c = b
+b = a
+a = TMP_
+.endm
+
+.macro FOUR_ROUNDS_AND_SCHED
+	## compute s0 four at a time and s1 two at a time
+	## compute W[-16] + W[-7] 4 at a time
+
+	mov     e, y0			# y0 = e
+	MY_ROR  (25-11), y0             # y0 = e >> (25-11)
+	mov     a, y1                   # y1 = a
+	vpalignr $4, X2, X3, XTMP0      # XTMP0 = W[-7]
+	MY_ROR  (22-13), y1             # y1 = a >> (22-13)
+	xor     e, y0                   # y0 = e ^ (e >> (25-11))
+	mov     f, y2                   # y2 = f
+	MY_ROR  (11-6), y0              # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	xor     a, y1                   # y1 = a ^ (a >> (22-13)
+	xor     g, y2                   # y2 = f^g
+	vpaddd  X0, XTMP0, XTMP0        # XTMP0 = W[-7] + W[-16]
+	xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+	and     e, y2                   # y2 = (f^g)&e
+	MY_ROR  (13-2), y1              # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	## compute s0
+	vpalignr $4, X0, X1, XTMP1      # XTMP1 = W[-15]
+	xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+	MY_ROR  6, y0                   # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+	xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+	MY_ROR  2, y1                   # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+	add     y0, y2                  # y2 = S1 + CH
+	add     _XFER(%rsp), y2         # y2 = k + w + S1 + CH
+	mov     a, y0                   # y0 = a
+	add     y2, h                   # h = h + S1 + CH + k + w
+	mov     a, y2                   # y2 = a
+	vpsrld  $7, XTMP1, XTMP2
+	or      c, y0                   # y0 = a|c
+	add     h, d                    # d = d + h + S1 + CH + k + w
+	and     c, y2                   # y2 = a&c
+	vpslld  $(32-7), XTMP1, XTMP3
+	and     b, y0                   # y0 = (a|c)&b
+	add     y1, h                   # h = h + S1 + CH + k + w + S0
+	vpor    XTMP2, XTMP3, XTMP3     # XTMP1 = W[-15] MY_ROR 7
+	or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+	ROTATE_ARGS
+	mov     e, y0                   # y0 = e
+	mov     a, y1                   # y1 = a
+	MY_ROR  (25-11), y0             # y0 = e >> (25-11)
+	xor     e, y0                   # y0 = e ^ (e >> (25-11))
+	mov     f, y2                   # y2 = f
+	MY_ROR  (22-13), y1             # y1 = a >> (22-13)
+	vpsrld  $18, XTMP1, XTMP2       #
+	xor     a, y1                   # y1 = a ^ (a >> (22-13)
+	MY_ROR  (11-6), y0              # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	xor     g, y2                   # y2 = f^g
+	vpsrld  $3, XTMP1, XTMP4        # XTMP4 = W[-15] >> 3
+	MY_ROR  (13-2), y1              # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+	and     e, y2                   # y2 = (f^g)&e
+	MY_ROR  6, y0                   # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+	vpslld  $(32-18), XTMP1, XTMP1
+	xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+	xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+	vpxor   XTMP1, XTMP3, XTMP3     #
+	add     y0, y2                  # y2 = S1 + CH
+	add     (1*4 + _XFER)(%rsp), y2 # y2 = k + w + S1 + CH
+	MY_ROR  2, y1                   # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+	vpxor   XTMP2, XTMP3, XTMP3     # XTMP1 = W[-15] MY_ROR 7 ^ W[-15] MY_ROR
+	mov     a, y0                   # y0 = a
+	add     y2, h                   # h = h + S1 + CH + k + w
+	mov     a, y2                   # y2 = a
+	vpxor   XTMP4, XTMP3, XTMP1     # XTMP1 = s0
+	or      c, y0                   # y0 = a|c
+	add     h, d                    # d = d + h + S1 + CH + k + w
+	and     c, y2                   # y2 = a&c
+	## compute low s1
+	vpshufd $0b11111010, X3, XTMP2  # XTMP2 = W[-2] {BBAA}
+	and     b, y0                   # y0 = (a|c)&b
+	add     y1, h                   # h = h + S1 + CH + k + w + S0
+	vpaddd  XTMP1, XTMP0, XTMP0     # XTMP0 = W[-16] + W[-7] + s0
+	or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+	ROTATE_ARGS
+	mov     e, y0                   # y0 = e
+	mov     a, y1                   # y1 = a
+	MY_ROR  (25-11), y0             # y0 = e >> (25-11)
+	xor     e, y0                   # y0 = e ^ (e >> (25-11))
+	MY_ROR  (22-13), y1             # y1 = a >> (22-13)
+	mov     f, y2                   # y2 = f
+	xor     a, y1                   # y1 = a ^ (a >> (22-13)
+	MY_ROR  (11-6), y0              # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	vpsrld  $10, XTMP2, XTMP4       # XTMP4 = W[-2] >> 10 {BBAA}
+	xor     g, y2                   # y2 = f^g
+	vpsrlq  $19, XTMP2, XTMP3       # XTMP3 = W[-2] MY_ROR 19 {xBxA}
+	xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+	and     e, y2                   # y2 = (f^g)&e
+	vpsrlq  $17, XTMP2, XTMP2       # XTMP2 = W[-2] MY_ROR 17 {xBxA}
+	MY_ROR  (13-2), y1              # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+	xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+	MY_ROR  6, y0                   # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+	vpxor   XTMP3, XTMP2, XTMP2     #
+	add     y0, y2                  # y2 = S1 + CH
+	MY_ROR  2, y1                   # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+	add     (2*4 + _XFER)(%rsp), y2 # y2 = k + w + S1 + CH
+	vpxor   XTMP2, XTMP4, XTMP4     # XTMP4 = s1 {xBxA}
+	mov     a, y0                   # y0 = a
+	add     y2, h                   # h = h + S1 + CH + k + w
+	mov     a, y2                   # y2 = a
+	vpshufb SHUF_00BA, XTMP4, XTMP4 # XTMP4 = s1 {00BA}
+	or      c, y0                   # y0 = a|c
+	add     h, d                    # d = d + h + S1 + CH + k + w
+	and     c, y2                   # y2 = a&c
+	vpaddd  XTMP4, XTMP0, XTMP0     # XTMP0 = {..., ..., W[1], W[0]}
+	and     b, y0                   # y0 = (a|c)&b
+	add     y1, h                   # h = h + S1 + CH + k + w + S0
+	## compute high s1
+	vpshufd $0b01010000, XTMP0, XTMP2 # XTMP2 = W[-2] {DDCC}
+	or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+	ROTATE_ARGS
+	mov     e, y0                   # y0 = e
+	MY_ROR  (25-11), y0             # y0 = e >> (25-11)
+	mov     a, y1                   # y1 = a
+	MY_ROR  (22-13), y1             # y1 = a >> (22-13)
+	xor     e, y0                   # y0 = e ^ (e >> (25-11))
+	mov     f, y2                   # y2 = f
+	MY_ROR  (11-6), y0              # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	vpsrld  $10, XTMP2, XTMP5       # XTMP5 = W[-2] >> 10 {DDCC}
+	xor     a, y1                   # y1 = a ^ (a >> (22-13)
+	xor     g, y2                   # y2 = f^g
+	vpsrlq  $19, XTMP2, XTMP3       # XTMP3 = W[-2] MY_ROR 19 {xDxC}
+	xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+	and     e, y2                   # y2 = (f^g)&e
+	MY_ROR  (13-2), y1              # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	vpsrlq  $17, XTMP2, XTMP2       # XTMP2 = W[-2] MY_ROR 17 {xDxC}
+	xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+	MY_ROR  6, y0                   # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+	xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+	vpxor   XTMP3, XTMP2, XTMP2
+	MY_ROR  2, y1                   # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+	add     y0, y2                  # y2 = S1 + CH
+	add     (3*4 + _XFER)(%rsp), y2 # y2 = k + w + S1 + CH
+	vpxor   XTMP2, XTMP5, XTMP5     # XTMP5 = s1 {xDxC}
+	mov     a, y0                   # y0 = a
+	add     y2, h                   # h = h + S1 + CH + k + w
+	mov     a, y2                   # y2 = a
+	vpshufb SHUF_DC00, XTMP5, XTMP5 # XTMP5 = s1 {DC00}
+	or      c, y0                   # y0 = a|c
+	add     h, d                    # d = d + h + S1 + CH + k + w
+	and     c, y2                   # y2 = a&c
+	vpaddd  XTMP0, XTMP5, X0        # X0 = {W[3], W[2], W[1], W[0]}
+	and     b, y0                   # y0 = (a|c)&b
+	add     y1, h                   # h = h + S1 + CH + k + w + S0
+	or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+	ROTATE_ARGS
+	rotate_Xs
+.endm
+
+## input is [rsp + _XFER + %1 * 4]
+.macro DO_ROUND round
+	mov	e, y0			# y0 = e
+        MY_ROR  (25-11), y0             # y0 = e >> (25-11)
+        mov     a, y1                   # y1 = a
+        xor     e, y0                   # y0 = e ^ (e >> (25-11))
+        MY_ROR  (22-13), y1             # y1 = a >> (22-13)
+        mov     f, y2                   # y2 = f
+        xor     a, y1                   # y1 = a ^ (a >> (22-13)
+        MY_ROR  (11-6), y0              # y0 = (e >> (11-6)) ^ (e >> (25-6))
+        xor     g, y2                   # y2 = f^g
+        xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+        MY_ROR  (13-2), y1              # y1 = (a >> (13-2)) ^ (a >> (22-2))
+        and     e, y2                   # y2 = (f^g)&e
+        xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+        MY_ROR  6, y0                   # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+        xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+        add     y0, y2                  # y2 = S1 + CH
+        MY_ROR  2, y1                   # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+        offset = \round * 4 + _XFER     #
+        add     offset(%rsp), y2	# y2 = k + w + S1 + CH
+        mov     a, y0			# y0 = a
+        add     y2, h                   # h = h + S1 + CH + k + w
+        mov     a, y2                   # y2 = a
+        or      c, y0                   # y0 = a|c
+        add     h, d                    # d = d + h + S1 + CH + k + w
+        and     c, y2                   # y2 = a&c
+        and     b, y0                   # y0 = (a|c)&b
+        add     y1, h                   # h = h + S1 + CH + k + w + S0
+        or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+        add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+        ROTATE_ARGS
+.endm
+
+########################################################################
+## void sha256_transform_avx(void *input_data, UINT32 digest[8], UINT64 num_blks)
+## arg 1 : pointer to input data
+## arg 2 : pointer to digest
+## arg 3 : Num blocks
+########################################################################
+.text
+ENTRY(sha256_transform_avx)
+.align 32
+	pushq   %rbx
+	pushq   %rbp
+	pushq   %r13
+	pushq   %r14
+	pushq   %r15
+	pushq   %r12
+
+	mov	%rsp, %r12
+	subq    $STACK_SIZE, %rsp	# allocate stack space
+	and	$~15, %rsp		# align stack pointer
+
+	shl     $6, NUM_BLKS		# convert to bytes
+	jz      done_hash
+	add     INP, NUM_BLKS		# pointer to end of data
+	mov     NUM_BLKS, _INP_END(%rsp)
+
+	## load initial digest
+	mov     4*0(CTX), a
+	mov     4*1(CTX), b
+	mov     4*2(CTX), c
+	mov     4*3(CTX), d
+	mov     4*4(CTX), e
+	mov     4*5(CTX), f
+	mov     4*6(CTX), g
+	mov     4*7(CTX), h
+
+	vmovdqa  PSHUFFLE_BYTE_FLIP_MASK(%rip), BYTE_FLIP_MASK
+	vmovdqa  _SHUF_00BA(%rip), SHUF_00BA
+	vmovdqa  _SHUF_DC00(%rip), SHUF_DC00
+loop0:
+	lea     K256(%rip), TBL
+
+	## byte swap first 16 dwords
+	COPY_XMM_AND_BSWAP      X0, 0*16(INP), BYTE_FLIP_MASK
+	COPY_XMM_AND_BSWAP      X1, 1*16(INP), BYTE_FLIP_MASK
+	COPY_XMM_AND_BSWAP      X2, 2*16(INP), BYTE_FLIP_MASK
+	COPY_XMM_AND_BSWAP      X3, 3*16(INP), BYTE_FLIP_MASK
+
+	mov     INP, _INP(%rsp)
+
+	## schedule 48 input dwords, by doing 3 rounds of 16 each
+	mov     $3, SRND
+.align 16
+loop1:
+	vpaddd  (TBL), X0, XFER
+	vmovdqa XFER, _XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	vpaddd  1*16(TBL), X0, XFER
+	vmovdqa XFER, _XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	vpaddd  2*16(TBL), X0, XFER
+	vmovdqa XFER, _XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	vpaddd  3*16(TBL), X0, XFER
+	vmovdqa XFER, _XFER(%rsp)
+	add	$4*16, TBL
+	FOUR_ROUNDS_AND_SCHED
+
+	sub     $1, SRND
+	jne     loop1
+
+	mov     $2, SRND
+loop2:
+	vpaddd  (TBL), X0, XFER
+	vmovdqa XFER, _XFER(%rsp)
+	DO_ROUND        0
+	DO_ROUND        1
+	DO_ROUND        2
+	DO_ROUND        3
+
+	vpaddd  1*16(TBL), X1, XFER
+	vmovdqa XFER, _XFER(%rsp)
+	add     $2*16, TBL
+	DO_ROUND        0
+	DO_ROUND        1
+	DO_ROUND        2
+	DO_ROUND        3
+
+	vmovdqa X2, X0
+	vmovdqa X3, X1
+
+	sub     $1, SRND
+	jne     loop2
+
+	addm    (4*0)(CTX),a
+	addm    (4*1)(CTX),b
+	addm    (4*2)(CTX),c
+	addm    (4*3)(CTX),d
+	addm    (4*4)(CTX),e
+	addm    (4*5)(CTX),f
+	addm    (4*6)(CTX),g
+	addm    (4*7)(CTX),h
+
+	mov     _INP(%rsp), INP
+	add     $64, INP
+	cmp     _INP_END(%rsp), INP
+	jne     loop0
+
+done_hash:
+
+	mov	%r12, %rsp
+
+	popq	%r12
+	popq    %r15
+	popq    %r14
+	popq    %r13
+	popq    %rbp
+	popq    %rbx
+	ret
+ENDPROC(sha256_transform_avx)
+
+.data
+.align 64
+K256:
+	.long 0x428a2f98,0x71374491,0xb5c0fbcf,0xe9b5dba5
+	.long 0x3956c25b,0x59f111f1,0x923f82a4,0xab1c5ed5
+	.long 0xd807aa98,0x12835b01,0x243185be,0x550c7dc3
+	.long 0x72be5d74,0x80deb1fe,0x9bdc06a7,0xc19bf174
+	.long 0xe49b69c1,0xefbe4786,0x0fc19dc6,0x240ca1cc
+	.long 0x2de92c6f,0x4a7484aa,0x5cb0a9dc,0x76f988da
+	.long 0x983e5152,0xa831c66d,0xb00327c8,0xbf597fc7
+	.long 0xc6e00bf3,0xd5a79147,0x06ca6351,0x14292967
+	.long 0x27b70a85,0x2e1b2138,0x4d2c6dfc,0x53380d13
+	.long 0x650a7354,0x766a0abb,0x81c2c92e,0x92722c85
+	.long 0xa2bfe8a1,0xa81a664b,0xc24b8b70,0xc76c51a3
+	.long 0xd192e819,0xd6990624,0xf40e3585,0x106aa070
+	.long 0x19a4c116,0x1e376c08,0x2748774c,0x34b0bcb5
+	.long 0x391c0cb3,0x4ed8aa4a,0x5b9cca4f,0x682e6ff3
+	.long 0x748f82ee,0x78a5636f,0x84c87814,0x8cc70208
+	.long 0x90befffa,0xa4506ceb,0xbef9a3f7,0xc67178f2
+
+PSHUFFLE_BYTE_FLIP_MASK:
+	.octa 0x0c0d0e0f08090a0b0405060700010203
+
+# shuffle xBxA -> 00BA
+_SHUF_00BA:
+	.octa 0xFFFFFFFFFFFFFFFF0b0a090803020100
+
+# shuffle xDxC -> DC00
+_SHUF_DC00:
+	.octa 0x0b0a090803020100FFFFFFFFFFFFFFFF
+#endif
diff --git a/arch/x86/crypto/sha256-avx2-asm.S b/arch/x86/crypto/sha256-avx2-asm.S
new file mode 100644
index 000000000000..9e86944c539d
--- /dev/null
+++ b/arch/x86/crypto/sha256-avx2-asm.S
@@ -0,0 +1,772 @@
+########################################################################
+# Implement fast SHA-256 with AVX2 instructions. (x86_64)
+#
+# Copyright (C) 2013 Intel Corporation.
+#
+# Authors:
+#     James Guilford <james.guilford@intel.com>
+#     Kirk Yap <kirk.s.yap@intel.com>
+#     Tim Chen <tim.c.chen@linux.intel.com>
+#
+# This software is available to you under a choice of one of two
+# licenses.  You may choose to be licensed under the terms of the GNU
+# General Public License (GPL) Version 2, available from the file
+# COPYING in the main directory of this source tree, or the
+# OpenIB.org BSD license below:
+#
+#     Redistribution and use in source and binary forms, with or
+#     without modification, are permitted provided that the following
+#     conditions are met:
+#
+#      - Redistributions of source code must retain the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer.
+#
+#      - Redistributions in binary form must reproduce the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer in the documentation and/or other materials
+#        provided with the distribution.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+# BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+# ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+#
+########################################################################
+#
+# This code is described in an Intel White-Paper:
+# "Fast SHA-256 Implementations on Intel Architecture Processors"
+#
+# To find it, surf to http://www.intel.com/p/en_US/embedded
+# and search for that title.
+#
+########################################################################
+# This code schedules 2 blocks at a time, with 4 lanes per block
+########################################################################
+
+#ifdef CONFIG_AS_AVX2
+#include <linux/linkage.h>
+
+## assume buffers not aligned
+#define	VMOVDQ vmovdqu
+
+################################ Define Macros
+
+# addm [mem], reg
+# Add reg to mem using reg-mem add and store
+.macro addm p1 p2
+	add	\p1, \p2
+	mov	\p2, \p1
+.endm
+
+################################
+
+X0 = %ymm4
+X1 = %ymm5
+X2 = %ymm6
+X3 = %ymm7
+
+# XMM versions of above
+XWORD0 = %xmm4
+XWORD1 = %xmm5
+XWORD2 = %xmm6
+XWORD3 = %xmm7
+
+XTMP0 = %ymm0
+XTMP1 = %ymm1
+XTMP2 = %ymm2
+XTMP3 = %ymm3
+XTMP4 = %ymm8
+XFER  = %ymm9
+XTMP5 = %ymm11
+
+SHUF_00BA =	%ymm10 # shuffle xBxA -> 00BA
+SHUF_DC00 =	%ymm12 # shuffle xDxC -> DC00
+BYTE_FLIP_MASK = %ymm13
+
+X_BYTE_FLIP_MASK = %xmm13 # XMM version of BYTE_FLIP_MASK
+
+NUM_BLKS = %rdx	# 3rd arg
+CTX	= %rsi  # 2nd arg
+INP	= %rdi	# 1st arg
+c	= %ecx
+d	= %r8d
+e       = %edx	# clobbers NUM_BLKS
+y3	= %edi	# clobbers INP
+
+
+TBL	= %rbp
+SRND	= CTX	# SRND is same register as CTX
+
+a = %eax
+b = %ebx
+f = %r9d
+g = %r10d
+h = %r11d
+old_h = %r11d
+
+T1 = %r12d
+y0 = %r13d
+y1 = %r14d
+y2 = %r15d
+
+
+_XFER_SIZE	= 2*64*4	# 2 blocks, 64 rounds, 4 bytes/round
+_XMM_SAVE_SIZE	= 0
+_INP_END_SIZE	= 8
+_INP_SIZE	= 8
+_CTX_SIZE	= 8
+_RSP_SIZE	= 8
+
+_XFER		= 0
+_XMM_SAVE	= _XFER     + _XFER_SIZE
+_INP_END	= _XMM_SAVE + _XMM_SAVE_SIZE
+_INP		= _INP_END  + _INP_END_SIZE
+_CTX		= _INP      + _INP_SIZE
+_RSP		= _CTX      + _CTX_SIZE
+STACK_SIZE	= _RSP      + _RSP_SIZE
+
+# rotate_Xs
+# Rotate values of symbols X0...X3
+.macro rotate_Xs
+	X_ = X0
+	X0 = X1
+	X1 = X2
+	X2 = X3
+	X3 = X_
+.endm
+
+# ROTATE_ARGS
+# Rotate values of symbols a...h
+.macro ROTATE_ARGS
+	old_h = h
+	TMP_ = h
+	h = g
+	g = f
+	f = e
+	e = d
+	d = c
+	c = b
+	b = a
+	a = TMP_
+.endm
+
+.macro FOUR_ROUNDS_AND_SCHED disp
+################################### RND N + 0 ############################
+
+	mov	a, y3		# y3 = a                                # MAJA
+	rorx	$25, e, y0	# y0 = e >> 25				# S1A
+	rorx	$11, e, y1	# y1 = e >> 11				# S1B
+
+	addl	\disp(%rsp, SRND), h		# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+	vpalignr $4, X2, X3, XTMP0 # XTMP0 = W[-7]
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$13, a, T1	# T1 = a >> 13				# S0B
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11)		# S1
+	xor	g, y2		# y2 = f^g                              # CH
+	vpaddd	X0, XTMP0, XTMP0 # XTMP0 = W[-7] + W[-16]# y1 = (e >> 6)# S1
+	rorx	$6, e, y1	# y1 = (e >> 6)				# S1
+
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11) ^ (e>>6)	# S1
+	rorx	$22, a, y1	# y1 = a >> 22				# S0A
+	add	h, d		# d = k + w + h + d                     # --
+
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	vpalignr $4, X0, X1, XTMP1	# XTMP1 = W[-15]
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13)		# S0
+	rorx	$2, a, T1	# T1 = (a >> 2)				# S0
+
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	vpsrld	$7, XTMP1, XTMP2
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13) ^ (a>>2)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	c, T1		# T1 = a&c                              # MAJB
+
+	add	y0, y2		# y2 = S1 + CH                          # --
+	vpslld	$(32-7), XTMP1, XTMP3
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+	vpor	XTMP2, XTMP3, XTMP3	# XTMP3 = W[-15] ror 7
+
+	vpsrld	$18, XTMP1, XTMP2
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+
+	ROTATE_ARGS
+
+################################### RND N + 1 ############################
+
+	mov	a, y3		# y3 = a                                # MAJA
+	rorx	$25, e, y0	# y0 = e >> 25				# S1A
+	rorx	$11, e, y1	# y1 = e >> 11				# S1B
+	offset = \disp + 1*4
+	addl	offset(%rsp, SRND), h	# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+
+	vpsrld	$3, XTMP1, XTMP4 # XTMP4 = W[-15] >> 3
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$13, a, T1	# T1 = a >> 13				# S0B
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11)		# S1
+	xor	g, y2		# y2 = f^g                              # CH
+
+
+	rorx	$6, e, y1	# y1 = (e >> 6)				# S1
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11) ^ (e>>6)	# S1
+	rorx	$22, a, y1	# y1 = a >> 22				# S0A
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	h, d		# d = k + w + h + d                     # --
+
+	vpslld	$(32-18), XTMP1, XTMP1
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13)		# S0
+
+	vpxor	XTMP1, XTMP3, XTMP3
+	rorx	$2, a, T1	# T1 = (a >> 2)				# S0
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+
+	vpxor	XTMP2, XTMP3, XTMP3	# XTMP3 = W[-15] ror 7 ^ W[-15] ror 18
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13) ^ (a>>2)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+	vpxor	XTMP4, XTMP3, XTMP1	# XTMP1 = s0
+	vpshufd	$0b11111010, X3, XTMP2	# XTMP2 = W[-2] {BBAA}
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	vpaddd	XTMP1, XTMP0, XTMP0	# XTMP0 = W[-16] + W[-7] + s0
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+	vpsrld	$10, XTMP2, XTMP4 # XTMP4 = W[-2] >> 10 {BBAA}
+
+
+	ROTATE_ARGS
+
+################################### RND N + 2 ############################
+
+	mov	a, y3		# y3 = a                                # MAJA
+	rorx	$25, e, y0	# y0 = e >> 25				# S1A
+	offset = \disp + 2*4
+	addl	offset(%rsp, SRND), h	# h = k + w + h         # --
+
+	vpsrlq	$19, XTMP2, XTMP3 # XTMP3 = W[-2] ror 19 {xBxA}
+	rorx	$11, e, y1	# y1 = e >> 11				# S1B
+	or	c, y3		# y3 = a|c                              # MAJA
+	mov	f, y2		# y2 = f                                # CH
+	xor	g, y2		# y2 = f^g                              # CH
+
+	rorx	$13, a, T1	# T1 = a >> 13				# S0B
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11)		# S1
+	vpsrlq	$17, XTMP2, XTMP2	# XTMP2 = W[-2] ror 17 {xBxA}
+	and	e, y2		# y2 = (f^g)&e                          # CH
+
+	rorx	$6, e, y1	# y1 = (e >> 6)				# S1
+	vpxor	XTMP3, XTMP2, XTMP2
+	add	h, d		# d = k + w + h + d                     # --
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11) ^ (e>>6)	# S1
+	rorx	$22, a, y1	# y1 = a >> 22				# S0A
+	vpxor	XTMP2, XTMP4, XTMP4	# XTMP4 = s1 {xBxA}
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+
+	vpshufb	SHUF_00BA, XTMP4, XTMP4	# XTMP4 = s1 {00BA}
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13)		# S0
+	rorx	$2, a ,T1	# T1 = (a >> 2)				# S0
+	vpaddd	XTMP4, XTMP0, XTMP0	# XTMP0 = {..., ..., W[1], W[0]}
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13) ^ (a>>2)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+	vpshufd	$0b01010000, XTMP0, XTMP2	# XTMP2 = W[-2] {DDCC}
+
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1,h		# h = k + w + h + S0                    # --
+	add	y2,d		# d = k + w + h + d + S1 + CH = d + t1  # --
+	add	y2,h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+
+	add	y3,h		# h = t1 + S0 + MAJ                     # --
+
+
+	ROTATE_ARGS
+
+################################### RND N + 3 ############################
+
+	mov	a, y3		# y3 = a                                # MAJA
+	rorx	$25, e, y0	# y0 = e >> 25				# S1A
+	rorx	$11, e, y1	# y1 = e >> 11				# S1B
+	offset = \disp + 3*4
+	addl	offset(%rsp, SRND), h	# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+
+	vpsrld	$10, XTMP2, XTMP5	# XTMP5 = W[-2] >> 10 {DDCC}
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$13, a, T1	# T1 = a >> 13				# S0B
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11)		# S1
+	xor	g, y2		# y2 = f^g                              # CH
+
+
+	vpsrlq	$19, XTMP2, XTMP3	# XTMP3 = W[-2] ror 19 {xDxC}
+	rorx	$6, e, y1	# y1 = (e >> 6)				# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	h, d		# d = k + w + h + d                     # --
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+
+	vpsrlq	$17, XTMP2, XTMP2	# XTMP2 = W[-2] ror 17 {xDxC}
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11) ^ (e>>6)	# S1
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+
+	vpxor	XTMP3, XTMP2, XTMP2
+	rorx	$22, a, y1	# y1 = a >> 22				# S0A
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+	vpxor	XTMP2, XTMP5, XTMP5	# XTMP5 = s1 {xDxC}
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13)		# S0
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	rorx	$2, a, T1	# T1 = (a >> 2)				# S0
+	vpshufb	SHUF_DC00, XTMP5, XTMP5	# XTMP5 = s1 {DC00}
+
+	vpaddd	XTMP0, XTMP5, X0	# X0 = {W[3], W[2], W[1], W[0]}
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13) ^ (a>>2)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	c, T1		# T1 = a&c                              # MAJB
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+
+	add	y1, h		# h = k + w + h + S0                    # --
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+	ROTATE_ARGS
+	rotate_Xs
+.endm
+
+.macro DO_4ROUNDS disp
+################################### RND N + 0 ###########################
+
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$25, e, y0	# y0 = e >> 25				# S1A
+	rorx	$11, e, y1	# y1 = e >> 11				# S1B
+	xor	g, y2		# y2 = f^g                              # CH
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11)		# S1
+	rorx	$6, e, y1	# y1 = (e >> 6)				# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11) ^ (e>>6)	# S1
+	rorx	$13, a, T1	# T1 = a >> 13				# S0B
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	rorx	$22, a, y1	# y1 = a >> 22				# S0A
+	mov	a, y3		# y3 = a                                # MAJA
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13)		# S0
+	rorx	$2, a, T1	# T1 = (a >> 2)				# S0
+	addl	\disp(%rsp, SRND), h		# h = k + w + h # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13) ^ (a>>2)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+
+	add	h, d		# d = k + w + h + d                     # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	ROTATE_ARGS
+
+################################### RND N + 1 ###########################
+
+	add	y2, old_h	# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$25, e, y0	# y0 = e >> 25				# S1A
+	rorx	$11, e, y1	# y1 = e >> 11				# S1B
+	xor	g, y2		# y2 = f^g                              # CH
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11)		# S1
+	rorx	$6, e, y1	# y1 = (e >> 6)				# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	y3, old_h	# h = t1 + S0 + MAJ                     # --
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11) ^ (e>>6)	# S1
+	rorx	$13, a, T1	# T1 = a >> 13				# S0B
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	rorx	$22, a, y1	# y1 = a >> 22				# S0A
+	mov	a, y3		# y3 = a                                # MAJA
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13)		# S0
+	rorx	$2, a, T1	# T1 = (a >> 2)				# S0
+	offset = 4*1 + \disp
+	addl	offset(%rsp, SRND), h		# h = k + w + h # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13) ^ (a>>2)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+
+	add	h, d		# d = k + w + h + d                     # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	ROTATE_ARGS
+
+################################### RND N + 2 ##############################
+
+	add	y2, old_h	# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$25, e, y0	# y0 = e >> 25				# S1A
+	rorx	$11, e, y1	# y1 = e >> 11				# S1B
+	xor	g, y2		# y2 = f^g                              # CH
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11)		# S1
+	rorx	$6, e, y1	# y1 = (e >> 6)				# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	y3, old_h	# h = t1 + S0 + MAJ                     # --
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11) ^ (e>>6)	# S1
+	rorx	$13, a, T1	# T1 = a >> 13				# S0B
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	rorx	$22, a, y1	# y1 = a >> 22				# S0A
+	mov	a, y3		# y3 = a                                # MAJA
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13)		# S0
+	rorx	$2, a, T1	# T1 = (a >> 2)				# S0
+	offset = 4*2 + \disp
+	addl	offset(%rsp, SRND), h		# h = k + w + h # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13) ^ (a>>2)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+
+	add	h, d		# d = k + w + h + d                     # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	ROTATE_ARGS
+
+################################### RND N + 3 ###########################
+
+	add	y2, old_h	# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$25, e, y0	# y0 = e >> 25				# S1A
+	rorx	$11, e, y1	# y1 = e >> 11				# S1B
+	xor	g, y2		# y2 = f^g                              # CH
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11)		# S1
+	rorx	$6, e, y1	# y1 = (e >> 6)				# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	y3, old_h	# h = t1 + S0 + MAJ                     # --
+
+	xor	y1, y0		# y0 = (e>>25) ^ (e>>11) ^ (e>>6)	# S1
+	rorx	$13, a, T1	# T1 = a >> 13				# S0B
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	rorx	$22, a, y1	# y1 = a >> 22				# S0A
+	mov	a, y3		# y3 = a                                # MAJA
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13)		# S0
+	rorx	$2, a, T1	# T1 = (a >> 2)				# S0
+	offset = 4*3 + \disp
+	addl	offset(%rsp, SRND), h		# h = k + w + h # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+	xor	T1, y1		# y1 = (a>>22) ^ (a>>13) ^ (a>>2)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+
+	add	h, d		# d = k + w + h + d                     # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+	ROTATE_ARGS
+
+.endm
+
+########################################################################
+## void sha256_transform_rorx(void *input_data, UINT32 digest[8], UINT64 num_blks)
+## arg 1 : pointer to input data
+## arg 2 : pointer to digest
+## arg 3 : Num blocks
+########################################################################
+.text
+ENTRY(sha256_transform_rorx)
+.align 32
+	pushq	%rbx
+	pushq	%rbp
+	pushq	%r12
+	pushq	%r13
+	pushq	%r14
+	pushq	%r15
+
+	mov	%rsp, %rax
+	subq	$STACK_SIZE, %rsp
+	and	$-32, %rsp	# align rsp to 32 byte boundary
+	mov	%rax, _RSP(%rsp)
+
+
+	shl	$6, NUM_BLKS	# convert to bytes
+	jz	done_hash
+	lea	-64(INP, NUM_BLKS), NUM_BLKS # pointer to last block
+	mov	NUM_BLKS, _INP_END(%rsp)
+
+	cmp	NUM_BLKS, INP
+	je	only_one_block
+
+	## load initial digest
+	mov	(CTX), a
+	mov	4*1(CTX), b
+	mov	4*2(CTX), c
+	mov	4*3(CTX), d
+	mov	4*4(CTX), e
+	mov	4*5(CTX), f
+	mov	4*6(CTX), g
+	mov	4*7(CTX), h
+
+	vmovdqa  PSHUFFLE_BYTE_FLIP_MASK(%rip), BYTE_FLIP_MASK
+	vmovdqa  _SHUF_00BA(%rip), SHUF_00BA
+	vmovdqa  _SHUF_DC00(%rip), SHUF_DC00
+
+	mov	CTX, _CTX(%rsp)
+
+loop0:
+	lea     K256(%rip), TBL
+
+	## Load first 16 dwords from two blocks
+	VMOVDQ	0*32(INP),XTMP0
+	VMOVDQ	1*32(INP),XTMP1
+	VMOVDQ	2*32(INP),XTMP2
+	VMOVDQ	3*32(INP),XTMP3
+
+	## byte swap data
+	vpshufb	BYTE_FLIP_MASK, XTMP0, XTMP0
+	vpshufb	BYTE_FLIP_MASK, XTMP1, XTMP1
+	vpshufb	BYTE_FLIP_MASK, XTMP2, XTMP2
+	vpshufb	BYTE_FLIP_MASK, XTMP3, XTMP3
+
+	## transpose data into high/low halves
+	vperm2i128	$0x20, XTMP2, XTMP0, X0
+	vperm2i128	$0x31, XTMP2, XTMP0, X1
+	vperm2i128	$0x20, XTMP3, XTMP1, X2
+	vperm2i128	$0x31, XTMP3, XTMP1, X3
+
+last_block_enter:
+	add	$64, INP
+	mov	INP, _INP(%rsp)
+
+	## schedule 48 input dwords, by doing 3 rounds of 12 each
+	xor	SRND, SRND
+
+.align 16
+loop1:
+	vpaddd	0*32(TBL, SRND), X0, XFER
+	vmovdqa XFER, 0*32+_XFER(%rsp, SRND)
+	FOUR_ROUNDS_AND_SCHED	_XFER + 0*32
+
+	vpaddd	1*32(TBL, SRND), X0, XFER
+	vmovdqa XFER, 1*32+_XFER(%rsp, SRND)
+	FOUR_ROUNDS_AND_SCHED	_XFER + 1*32
+
+	vpaddd	2*32(TBL, SRND), X0, XFER
+	vmovdqa XFER, 2*32+_XFER(%rsp, SRND)
+	FOUR_ROUNDS_AND_SCHED	_XFER + 2*32
+
+	vpaddd	3*32(TBL, SRND), X0, XFER
+	vmovdqa XFER, 3*32+_XFER(%rsp, SRND)
+	FOUR_ROUNDS_AND_SCHED	_XFER + 3*32
+
+	add	$4*32, SRND
+	cmp	$3*4*32, SRND
+	jb	loop1
+
+loop2:
+	## Do last 16 rounds with no scheduling
+	vpaddd	0*32(TBL, SRND), X0, XFER
+	vmovdqa XFER, 0*32+_XFER(%rsp, SRND)
+	DO_4ROUNDS	_XFER + 0*32
+	vpaddd	1*32(TBL, SRND), X1, XFER
+	vmovdqa XFER, 1*32+_XFER(%rsp, SRND)
+	DO_4ROUNDS	_XFER + 1*32
+	add	$2*32, SRND
+
+	vmovdqa	X2, X0
+	vmovdqa	X3, X1
+
+	cmp	$4*4*32, SRND
+	jb	loop2
+
+	mov	_CTX(%rsp), CTX
+	mov	_INP(%rsp), INP
+
+	addm    (4*0)(CTX),a
+	addm    (4*1)(CTX),b
+	addm    (4*2)(CTX),c
+	addm    (4*3)(CTX),d
+	addm    (4*4)(CTX),e
+	addm    (4*5)(CTX),f
+	addm    (4*6)(CTX),g
+	addm    (4*7)(CTX),h
+
+	cmp	_INP_END(%rsp), INP
+	ja	done_hash
+
+	#### Do second block using previously scheduled results
+	xor	SRND, SRND
+.align 16
+loop3:
+	DO_4ROUNDS	 _XFER + 0*32 + 16
+	DO_4ROUNDS	 _XFER + 1*32 + 16
+	add	$2*32, SRND
+	cmp	$4*4*32, SRND
+	jb	loop3
+
+	mov	_CTX(%rsp), CTX
+	mov	_INP(%rsp), INP
+	add	$64, INP
+
+	addm    (4*0)(CTX),a
+	addm    (4*1)(CTX),b
+	addm    (4*2)(CTX),c
+	addm    (4*3)(CTX),d
+	addm    (4*4)(CTX),e
+	addm    (4*5)(CTX),f
+	addm    (4*6)(CTX),g
+	addm    (4*7)(CTX),h
+
+	cmp	_INP_END(%rsp), INP
+	jb	loop0
+	ja	done_hash
+
+do_last_block:
+	#### do last block
+	lea	K256(%rip), TBL
+
+	VMOVDQ	0*16(INP),XWORD0
+	VMOVDQ	1*16(INP),XWORD1
+	VMOVDQ	2*16(INP),XWORD2
+	VMOVDQ	3*16(INP),XWORD3
+
+	vpshufb	X_BYTE_FLIP_MASK, XWORD0, XWORD0
+	vpshufb	X_BYTE_FLIP_MASK, XWORD1, XWORD1
+	vpshufb	X_BYTE_FLIP_MASK, XWORD2, XWORD2
+	vpshufb	X_BYTE_FLIP_MASK, XWORD3, XWORD3
+
+	jmp	last_block_enter
+
+only_one_block:
+
+	## load initial digest
+	mov	(4*0)(CTX),a
+	mov	(4*1)(CTX),b
+	mov	(4*2)(CTX),c
+	mov	(4*3)(CTX),d
+	mov	(4*4)(CTX),e
+	mov	(4*5)(CTX),f
+	mov	(4*6)(CTX),g
+	mov	(4*7)(CTX),h
+
+	vmovdqa	PSHUFFLE_BYTE_FLIP_MASK(%rip), BYTE_FLIP_MASK
+	vmovdqa	_SHUF_00BA(%rip), SHUF_00BA
+	vmovdqa	_SHUF_DC00(%rip), SHUF_DC00
+
+	mov	CTX, _CTX(%rsp)
+	jmp	do_last_block
+
+done_hash:
+
+	mov	_RSP(%rsp), %rsp
+
+	popq	%r15
+	popq	%r14
+	popq	%r13
+	popq	%r12
+	popq	%rbp
+	popq	%rbx
+	ret
+ENDPROC(sha256_transform_rorx)
+
+.data
+.align 64
+K256:
+	.long	0x428a2f98,0x71374491,0xb5c0fbcf,0xe9b5dba5
+	.long	0x428a2f98,0x71374491,0xb5c0fbcf,0xe9b5dba5
+	.long	0x3956c25b,0x59f111f1,0x923f82a4,0xab1c5ed5
+	.long	0x3956c25b,0x59f111f1,0x923f82a4,0xab1c5ed5
+	.long	0xd807aa98,0x12835b01,0x243185be,0x550c7dc3
+	.long	0xd807aa98,0x12835b01,0x243185be,0x550c7dc3
+	.long	0x72be5d74,0x80deb1fe,0x9bdc06a7,0xc19bf174
+	.long	0x72be5d74,0x80deb1fe,0x9bdc06a7,0xc19bf174
+	.long	0xe49b69c1,0xefbe4786,0x0fc19dc6,0x240ca1cc
+	.long	0xe49b69c1,0xefbe4786,0x0fc19dc6,0x240ca1cc
+	.long	0x2de92c6f,0x4a7484aa,0x5cb0a9dc,0x76f988da
+	.long	0x2de92c6f,0x4a7484aa,0x5cb0a9dc,0x76f988da
+	.long	0x983e5152,0xa831c66d,0xb00327c8,0xbf597fc7
+	.long	0x983e5152,0xa831c66d,0xb00327c8,0xbf597fc7
+	.long	0xc6e00bf3,0xd5a79147,0x06ca6351,0x14292967
+	.long	0xc6e00bf3,0xd5a79147,0x06ca6351,0x14292967
+	.long	0x27b70a85,0x2e1b2138,0x4d2c6dfc,0x53380d13
+	.long	0x27b70a85,0x2e1b2138,0x4d2c6dfc,0x53380d13
+	.long	0x650a7354,0x766a0abb,0x81c2c92e,0x92722c85
+	.long	0x650a7354,0x766a0abb,0x81c2c92e,0x92722c85
+	.long	0xa2bfe8a1,0xa81a664b,0xc24b8b70,0xc76c51a3
+	.long	0xa2bfe8a1,0xa81a664b,0xc24b8b70,0xc76c51a3
+	.long	0xd192e819,0xd6990624,0xf40e3585,0x106aa070
+	.long	0xd192e819,0xd6990624,0xf40e3585,0x106aa070
+	.long	0x19a4c116,0x1e376c08,0x2748774c,0x34b0bcb5
+	.long	0x19a4c116,0x1e376c08,0x2748774c,0x34b0bcb5
+	.long	0x391c0cb3,0x4ed8aa4a,0x5b9cca4f,0x682e6ff3
+	.long	0x391c0cb3,0x4ed8aa4a,0x5b9cca4f,0x682e6ff3
+	.long	0x748f82ee,0x78a5636f,0x84c87814,0x8cc70208
+	.long	0x748f82ee,0x78a5636f,0x84c87814,0x8cc70208
+	.long	0x90befffa,0xa4506ceb,0xbef9a3f7,0xc67178f2
+	.long	0x90befffa,0xa4506ceb,0xbef9a3f7,0xc67178f2
+
+PSHUFFLE_BYTE_FLIP_MASK:
+	.octa 0x0c0d0e0f08090a0b0405060700010203,0x0c0d0e0f08090a0b0405060700010203
+
+# shuffle xBxA -> 00BA
+_SHUF_00BA:
+	.octa 0xFFFFFFFFFFFFFFFF0b0a090803020100,0xFFFFFFFFFFFFFFFF0b0a090803020100
+
+# shuffle xDxC -> DC00
+_SHUF_DC00:
+	.octa 0x0b0a090803020100FFFFFFFFFFFFFFFF,0x0b0a090803020100FFFFFFFFFFFFFFFF
+#endif
diff --git a/arch/x86/crypto/sha256-ssse3-asm.S b/arch/x86/crypto/sha256-ssse3-asm.S
new file mode 100644
index 000000000000..98d3c391da81
--- /dev/null
+++ b/arch/x86/crypto/sha256-ssse3-asm.S
@@ -0,0 +1,506 @@
+########################################################################
+# Implement fast SHA-256 with SSSE3 instructions. (x86_64)
+#
+# Copyright (C) 2013 Intel Corporation.
+#
+# Authors:
+#     James Guilford <james.guilford@intel.com>
+#     Kirk Yap <kirk.s.yap@intel.com>
+#     Tim Chen <tim.c.chen@linux.intel.com>
+#
+# This software is available to you under a choice of one of two
+# licenses.  You may choose to be licensed under the terms of the GNU
+# General Public License (GPL) Version 2, available from the file
+# COPYING in the main directory of this source tree, or the
+# OpenIB.org BSD license below:
+#
+#     Redistribution and use in source and binary forms, with or
+#     without modification, are permitted provided that the following
+#     conditions are met:
+#
+#      - Redistributions of source code must retain the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer.
+#
+#      - Redistributions in binary form must reproduce the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer in the documentation and/or other materials
+#        provided with the distribution.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+# BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+# ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+#
+########################################################################
+#
+# This code is described in an Intel White-Paper:
+# "Fast SHA-256 Implementations on Intel Architecture Processors"
+#
+# To find it, surf to http://www.intel.com/p/en_US/embedded
+# and search for that title.
+#
+########################################################################
+
+#include <linux/linkage.h>
+
+## assume buffers not aligned
+#define    MOVDQ movdqu
+
+################################ Define Macros
+
+# addm [mem], reg
+# Add reg to mem using reg-mem add and store
+.macro addm p1 p2
+        add     \p1, \p2
+        mov     \p2, \p1
+.endm
+
+################################
+
+# COPY_XMM_AND_BSWAP xmm, [mem], byte_flip_mask
+# Load xmm with mem and byte swap each dword
+.macro COPY_XMM_AND_BSWAP p1 p2 p3
+        MOVDQ \p2, \p1
+        pshufb \p3, \p1
+.endm
+
+################################
+
+X0 = %xmm4
+X1 = %xmm5
+X2 = %xmm6
+X3 = %xmm7
+
+XTMP0 = %xmm0
+XTMP1 = %xmm1
+XTMP2 = %xmm2
+XTMP3 = %xmm3
+XTMP4 = %xmm8
+XFER = %xmm9
+
+SHUF_00BA = %xmm10      # shuffle xBxA -> 00BA
+SHUF_DC00 = %xmm11      # shuffle xDxC -> DC00
+BYTE_FLIP_MASK = %xmm12
+
+NUM_BLKS = %rdx   # 3rd arg
+CTX = %rsi        # 2nd arg
+INP = %rdi        # 1st arg
+
+SRND = %rdi       # clobbers INP
+c = %ecx
+d = %r8d
+e = %edx
+TBL = %rbp
+a = %eax
+b = %ebx
+
+f = %r9d
+g = %r10d
+h = %r11d
+
+y0 = %r13d
+y1 = %r14d
+y2 = %r15d
+
+
+
+_INP_END_SIZE = 8
+_INP_SIZE = 8
+_XFER_SIZE = 8
+_XMM_SAVE_SIZE = 0
+
+_INP_END = 0
+_INP            = _INP_END  + _INP_END_SIZE
+_XFER           = _INP      + _INP_SIZE
+_XMM_SAVE       = _XFER     + _XFER_SIZE
+STACK_SIZE      = _XMM_SAVE + _XMM_SAVE_SIZE
+
+# rotate_Xs
+# Rotate values of symbols X0...X3
+.macro rotate_Xs
+X_ = X0
+X0 = X1
+X1 = X2
+X2 = X3
+X3 = X_
+.endm
+
+# ROTATE_ARGS
+# Rotate values of symbols a...h
+.macro ROTATE_ARGS
+TMP_ = h
+h = g
+g = f
+f = e
+e = d
+d = c
+c = b
+b = a
+a = TMP_
+.endm
+
+.macro FOUR_ROUNDS_AND_SCHED
+	## compute s0 four at a time and s1 two at a time
+	## compute W[-16] + W[-7] 4 at a time
+	movdqa  X3, XTMP0
+	mov     e, y0			# y0 = e
+	ror     $(25-11), y0            # y0 = e >> (25-11)
+	mov     a, y1                   # y1 = a
+	palignr $4, X2, XTMP0           # XTMP0 = W[-7]
+	ror     $(22-13), y1            # y1 = a >> (22-13)
+	xor     e, y0                   # y0 = e ^ (e >> (25-11))
+	mov     f, y2                   # y2 = f
+	ror     $(11-6), y0             # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	movdqa  X1, XTMP1
+	xor     a, y1                   # y1 = a ^ (a >> (22-13)
+	xor     g, y2                   # y2 = f^g
+	paddd   X0, XTMP0               # XTMP0 = W[-7] + W[-16]
+	xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+	and     e, y2                   # y2 = (f^g)&e
+	ror     $(13-2), y1             # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	## compute s0
+	palignr $4, X0, XTMP1           # XTMP1 = W[-15]
+	xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+	ror     $6, y0                  # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+	xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+	movdqa  XTMP1, XTMP2            # XTMP2 = W[-15]
+	ror     $2, y1                  # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+	add     y0, y2                  # y2 = S1 + CH
+	add     _XFER(%rsp) , y2        # y2 = k + w + S1 + CH
+	movdqa  XTMP1, XTMP3            # XTMP3 = W[-15]
+	mov     a, y0                   # y0 = a
+	add     y2, h                   # h = h + S1 + CH + k + w
+	mov     a, y2                   # y2 = a
+	pslld   $(32-7), XTMP1          #
+	or      c, y0                   # y0 = a|c
+	add     h, d                    # d = d + h + S1 + CH + k + w
+	and     c, y2                   # y2 = a&c
+	psrld   $7, XTMP2               #
+	and     b, y0                   # y0 = (a|c)&b
+	add     y1, h                   # h = h + S1 + CH + k + w + S0
+	por     XTMP2, XTMP1            # XTMP1 = W[-15] ror 7
+	or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+					#
+	ROTATE_ARGS                     #
+	movdqa  XTMP3, XTMP2            # XTMP2 = W[-15]
+	mov     e, y0                   # y0 = e
+	mov     a, y1                   # y1 = a
+	movdqa  XTMP3, XTMP4            # XTMP4 = W[-15]
+	ror     $(25-11), y0            # y0 = e >> (25-11)
+	xor     e, y0                   # y0 = e ^ (e >> (25-11))
+	mov     f, y2                   # y2 = f
+	ror     $(22-13), y1            # y1 = a >> (22-13)
+	pslld   $(32-18), XTMP3         #
+	xor     a, y1                   # y1 = a ^ (a >> (22-13)
+	ror     $(11-6), y0             # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	xor     g, y2                   # y2 = f^g
+	psrld   $18, XTMP2              #
+	ror     $(13-2), y1             # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+	and     e, y2                   # y2 = (f^g)&e
+	ror     $6, y0                  # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+	pxor    XTMP3, XTMP1
+	xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+	xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+	psrld   $3, XTMP4               # XTMP4 = W[-15] >> 3
+	add     y0, y2                  # y2 = S1 + CH
+	add     (1*4 + _XFER)(%rsp), y2 # y2 = k + w + S1 + CH
+	ror     $2, y1                  # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+	pxor    XTMP2, XTMP1            # XTMP1 = W[-15] ror 7 ^ W[-15] ror 18
+	mov     a, y0                   # y0 = a
+	add     y2, h                   # h = h + S1 + CH + k + w
+	mov     a, y2                   # y2 = a
+	pxor    XTMP4, XTMP1            # XTMP1 = s0
+	or      c, y0                   # y0 = a|c
+	add     h, d                    # d = d + h + S1 + CH + k + w
+	and     c, y2                   # y2 = a&c
+	## compute low s1
+	pshufd  $0b11111010, X3, XTMP2   # XTMP2 = W[-2] {BBAA}
+	and     b, y0			# y0 = (a|c)&b
+	add     y1, h                   # h = h + S1 + CH + k + w + S0
+	paddd   XTMP1, XTMP0            # XTMP0 = W[-16] + W[-7] + s0
+	or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+
+	ROTATE_ARGS
+	movdqa  XTMP2, XTMP3            # XTMP3 = W[-2] {BBAA}
+	mov     e, y0                   # y0 = e
+	mov     a, y1                   # y1 = a
+	ror     $(25-11), y0            # y0 = e >> (25-11)
+	movdqa  XTMP2, XTMP4            # XTMP4 = W[-2] {BBAA}
+	xor     e, y0                   # y0 = e ^ (e >> (25-11))
+	ror     $(22-13), y1            # y1 = a >> (22-13)
+	mov     f, y2                   # y2 = f
+	xor     a, y1                   # y1 = a ^ (a >> (22-13)
+	ror     $(11-6), y0             # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	psrlq   $17, XTMP2              # XTMP2 = W[-2] ror 17 {xBxA}
+	xor     g, y2                   # y2 = f^g
+	psrlq   $19, XTMP3              # XTMP3 = W[-2] ror 19 {xBxA}
+	xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+	and     e, y2                   # y2 = (f^g)&e
+	psrld   $10, XTMP4              # XTMP4 = W[-2] >> 10 {BBAA}
+	ror     $(13-2), y1             # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+	xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+	ror     $6, y0                  # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+	pxor    XTMP3, XTMP2
+	add     y0, y2                  # y2 = S1 + CH
+	ror     $2, y1                  # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+	add     (2*4 + _XFER)(%rsp), y2 # y2 = k + w + S1 + CH
+	pxor    XTMP2, XTMP4            # XTMP4 = s1 {xBxA}
+	mov     a, y0                   # y0 = a
+	add     y2, h                   # h = h + S1 + CH + k + w
+	mov     a, y2                   # y2 = a
+	pshufb  SHUF_00BA, XTMP4        # XTMP4 = s1 {00BA}
+	or      c, y0                   # y0 = a|c
+	add     h, d                    # d = d + h + S1 + CH + k + w
+	and     c, y2                   # y2 = a&c
+	paddd   XTMP4, XTMP0            # XTMP0 = {..., ..., W[1], W[0]}
+	and     b, y0                   # y0 = (a|c)&b
+	add     y1, h                   # h = h + S1 + CH + k + w + S0
+	## compute high s1
+	pshufd  $0b01010000, XTMP0, XTMP2 # XTMP2 = W[-2] {BBAA}
+	or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+					#
+	ROTATE_ARGS                     #
+	movdqa  XTMP2, XTMP3            # XTMP3 = W[-2] {DDCC}
+	mov     e, y0                   # y0 = e
+	ror     $(25-11), y0            # y0 = e >> (25-11)
+	mov     a, y1                   # y1 = a
+	movdqa  XTMP2, X0               # X0    = W[-2] {DDCC}
+	ror     $(22-13), y1            # y1 = a >> (22-13)
+	xor     e, y0                   # y0 = e ^ (e >> (25-11))
+	mov     f, y2                   # y2 = f
+	ror     $(11-6), y0             # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	psrlq   $17, XTMP2              # XTMP2 = W[-2] ror 17 {xDxC}
+	xor     a, y1                   # y1 = a ^ (a >> (22-13)
+	xor     g, y2                   # y2 = f^g
+	psrlq   $19, XTMP3              # XTMP3 = W[-2] ror 19 {xDxC}
+	xor     e, y0                   # y0 = e ^ (e >> (11-6)) ^ (e >> (25
+	and     e, y2                   # y2 = (f^g)&e
+	ror     $(13-2), y1             # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	psrld   $10, X0                 # X0 = W[-2] >> 10 {DDCC}
+	xor     a, y1                   # y1 = a ^ (a >> (13-2)) ^ (a >> (22
+	ror     $6, y0                  # y0 = S1 = (e>>6) & (e>>11) ^ (e>>2
+	xor     g, y2                   # y2 = CH = ((f^g)&e)^g
+	pxor    XTMP3, XTMP2            #
+	ror     $2, y1                  # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>2
+	add     y0, y2                  # y2 = S1 + CH
+	add     (3*4 + _XFER)(%rsp), y2 # y2 = k + w + S1 + CH
+	pxor    XTMP2, X0               # X0 = s1 {xDxC}
+	mov     a, y0                   # y0 = a
+	add     y2, h                   # h = h + S1 + CH + k + w
+	mov     a, y2                   # y2 = a
+	pshufb  SHUF_DC00, X0           # X0 = s1 {DC00}
+	or      c, y0                   # y0 = a|c
+	add     h, d                    # d = d + h + S1 + CH + k + w
+	and     c, y2                   # y2 = a&c
+	paddd   XTMP0, X0               # X0 = {W[3], W[2], W[1], W[0]}
+	and     b, y0                   # y0 = (a|c)&b
+	add     y1, h                   # h = h + S1 + CH + k + w + S0
+	or      y2, y0                  # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h                   # h = h + S1 + CH + k + w + S0 + MAJ
+
+	ROTATE_ARGS
+	rotate_Xs
+.endm
+
+## input is [rsp + _XFER + %1 * 4]
+.macro DO_ROUND round
+	mov     e, y0                 # y0 = e
+	ror     $(25-11), y0          # y0 = e >> (25-11)
+	mov     a, y1                 # y1 = a
+	xor     e, y0                 # y0 = e ^ (e >> (25-11))
+	ror     $(22-13), y1          # y1 = a >> (22-13)
+	mov     f, y2                 # y2 = f
+	xor     a, y1                 # y1 = a ^ (a >> (22-13)
+	ror     $(11-6), y0           # y0 = (e >> (11-6)) ^ (e >> (25-6))
+	xor     g, y2                 # y2 = f^g
+	xor     e, y0                 # y0 = e ^ (e >> (11-6)) ^ (e >> (25-6))
+	ror     $(13-2), y1           # y1 = (a >> (13-2)) ^ (a >> (22-2))
+	and     e, y2                 # y2 = (f^g)&e
+	xor     a, y1                 # y1 = a ^ (a >> (13-2)) ^ (a >> (22-2))
+	ror     $6, y0                # y0 = S1 = (e>>6) & (e>>11) ^ (e>>25)
+	xor     g, y2                 # y2 = CH = ((f^g)&e)^g
+	add     y0, y2                # y2 = S1 + CH
+	ror     $2, y1                # y1 = S0 = (a>>2) ^ (a>>13) ^ (a>>22)
+	offset = \round * 4 + _XFER
+	add     offset(%rsp), y2      # y2 = k + w + S1 + CH
+	mov     a, y0                 # y0 = a
+	add     y2, h                 # h = h + S1 + CH + k + w
+	mov     a, y2                 # y2 = a
+	or      c, y0                 # y0 = a|c
+	add     h, d                  # d = d + h + S1 + CH + k + w
+	and     c, y2                 # y2 = a&c
+	and     b, y0                 # y0 = (a|c)&b
+	add     y1, h                 # h = h + S1 + CH + k + w + S0
+	or      y2, y0		      # y0 = MAJ = (a|c)&b)|(a&c)
+	add     y0, h		      # h = h + S1 + CH + k + w + S0 + MAJ
+	ROTATE_ARGS
+.endm
+
+########################################################################
+## void sha256_transform_ssse3(void *input_data, UINT32 digest[8], UINT64 num_blks)
+## arg 1 : pointer to input data
+## arg 2 : pointer to digest
+## arg 3 : Num blocks
+########################################################################
+.text
+ENTRY(sha256_transform_ssse3)
+.align 32
+	pushq   %rbx
+	pushq   %rbp
+	pushq   %r13
+	pushq   %r14
+	pushq   %r15
+	pushq   %r12
+
+	mov	%rsp, %r12
+	subq    $STACK_SIZE, %rsp
+	and	$~15, %rsp
+
+	shl     $6, NUM_BLKS		 # convert to bytes
+	jz      done_hash
+	add     INP, NUM_BLKS
+	mov     NUM_BLKS, _INP_END(%rsp) # pointer to end of data
+
+	## load initial digest
+	mov     4*0(CTX), a
+	mov     4*1(CTX), b
+	mov     4*2(CTX), c
+	mov     4*3(CTX), d
+	mov     4*4(CTX), e
+	mov     4*5(CTX), f
+	mov     4*6(CTX), g
+	mov     4*7(CTX), h
+
+	movdqa  PSHUFFLE_BYTE_FLIP_MASK(%rip), BYTE_FLIP_MASK
+	movdqa  _SHUF_00BA(%rip), SHUF_00BA
+	movdqa  _SHUF_DC00(%rip), SHUF_DC00
+
+loop0:
+	lea     K256(%rip), TBL
+
+	## byte swap first 16 dwords
+	COPY_XMM_AND_BSWAP      X0, 0*16(INP), BYTE_FLIP_MASK
+	COPY_XMM_AND_BSWAP      X1, 1*16(INP), BYTE_FLIP_MASK
+	COPY_XMM_AND_BSWAP      X2, 2*16(INP), BYTE_FLIP_MASK
+	COPY_XMM_AND_BSWAP      X3, 3*16(INP), BYTE_FLIP_MASK
+
+	mov     INP, _INP(%rsp)
+
+	## schedule 48 input dwords, by doing 3 rounds of 16 each
+	mov     $3, SRND
+.align 16
+loop1:
+	movdqa  (TBL), XFER
+	paddd   X0, XFER
+	movdqa  XFER, _XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	movdqa  1*16(TBL), XFER
+	paddd   X0, XFER
+	movdqa  XFER, _XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	movdqa  2*16(TBL), XFER
+	paddd   X0, XFER
+	movdqa  XFER, _XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	movdqa  3*16(TBL), XFER
+	paddd   X0, XFER
+	movdqa  XFER, _XFER(%rsp)
+	add     $4*16, TBL
+	FOUR_ROUNDS_AND_SCHED
+
+	sub     $1, SRND
+	jne     loop1
+
+	mov     $2, SRND
+loop2:
+	paddd   (TBL), X0
+	movdqa  X0, _XFER(%rsp)
+	DO_ROUND        0
+	DO_ROUND        1
+	DO_ROUND        2
+	DO_ROUND        3
+	paddd   1*16(TBL), X1
+	movdqa  X1, _XFER(%rsp)
+	add     $2*16, TBL
+	DO_ROUND        0
+	DO_ROUND        1
+	DO_ROUND        2
+	DO_ROUND        3
+
+	movdqa  X2, X0
+	movdqa  X3, X1
+
+	sub     $1, SRND
+	jne     loop2
+
+	addm    (4*0)(CTX),a
+	addm    (4*1)(CTX),b
+	addm    (4*2)(CTX),c
+	addm    (4*3)(CTX),d
+	addm    (4*4)(CTX),e
+	addm    (4*5)(CTX),f
+	addm    (4*6)(CTX),g
+	addm    (4*7)(CTX),h
+
+	mov     _INP(%rsp), INP
+	add     $64, INP
+	cmp     _INP_END(%rsp), INP
+	jne     loop0
+
+done_hash:
+
+	mov	%r12, %rsp
+
+	popq    %r12
+	popq    %r15
+	popq    %r14
+	popq    %r13
+	popq    %rbp
+	popq    %rbx
+
+	ret
+ENDPROC(sha256_transform_ssse3)
+
+.data
+.align 64
+K256:
+        .long 0x428a2f98,0x71374491,0xb5c0fbcf,0xe9b5dba5
+        .long 0x3956c25b,0x59f111f1,0x923f82a4,0xab1c5ed5
+        .long 0xd807aa98,0x12835b01,0x243185be,0x550c7dc3
+        .long 0x72be5d74,0x80deb1fe,0x9bdc06a7,0xc19bf174
+        .long 0xe49b69c1,0xefbe4786,0x0fc19dc6,0x240ca1cc
+        .long 0x2de92c6f,0x4a7484aa,0x5cb0a9dc,0x76f988da
+        .long 0x983e5152,0xa831c66d,0xb00327c8,0xbf597fc7
+        .long 0xc6e00bf3,0xd5a79147,0x06ca6351,0x14292967
+        .long 0x27b70a85,0x2e1b2138,0x4d2c6dfc,0x53380d13
+        .long 0x650a7354,0x766a0abb,0x81c2c92e,0x92722c85
+        .long 0xa2bfe8a1,0xa81a664b,0xc24b8b70,0xc76c51a3
+        .long 0xd192e819,0xd6990624,0xf40e3585,0x106aa070
+        .long 0x19a4c116,0x1e376c08,0x2748774c,0x34b0bcb5
+        .long 0x391c0cb3,0x4ed8aa4a,0x5b9cca4f,0x682e6ff3
+        .long 0x748f82ee,0x78a5636f,0x84c87814,0x8cc70208
+        .long 0x90befffa,0xa4506ceb,0xbef9a3f7,0xc67178f2
+
+PSHUFFLE_BYTE_FLIP_MASK:
+	.octa 0x0c0d0e0f08090a0b0405060700010203
+
+# shuffle xBxA -> 00BA
+_SHUF_00BA:
+	.octa 0xFFFFFFFFFFFFFFFF0b0a090803020100
+
+# shuffle xDxC -> DC00
+_SHUF_DC00:
+	.octa 0x0b0a090803020100FFFFFFFFFFFFFFFF
diff --git a/arch/x86/crypto/sha256_ssse3_glue.c b/arch/x86/crypto/sha256_ssse3_glue.c
new file mode 100644
index 000000000000..597d4da69656
--- /dev/null
+++ b/arch/x86/crypto/sha256_ssse3_glue.c
@@ -0,0 +1,275 @@
+/*
+ * Cryptographic API.
+ *
+ * Glue code for the SHA256 Secure Hash Algorithm assembler
+ * implementation using supplemental SSE3 / AVX / AVX2 instructions.
+ *
+ * This file is based on sha256_generic.c
+ *
+ * Copyright (C) 2013 Intel Corporation.
+ *
+ * Author:
+ *     Tim Chen <tim.c.chen@linux.intel.com>
+ *
+ * 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.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+
+#define pr_fmt(fmt)	KBUILD_MODNAME ": " fmt
+
+#include <crypto/internal/hash.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/cryptohash.h>
+#include <linux/types.h>
+#include <crypto/sha.h>
+#include <asm/byteorder.h>
+#include <asm/i387.h>
+#include <asm/xcr.h>
+#include <asm/xsave.h>
+#include <linux/string.h>
+
+asmlinkage void sha256_transform_ssse3(const char *data, u32 *digest,
+				     u64 rounds);
+#ifdef CONFIG_AS_AVX
+asmlinkage void sha256_transform_avx(const char *data, u32 *digest,
+				     u64 rounds);
+#endif
+#ifdef CONFIG_AS_AVX2
+asmlinkage void sha256_transform_rorx(const char *data, u32 *digest,
+				     u64 rounds);
+#endif
+
+static asmlinkage void (*sha256_transform_asm)(const char *, u32 *, u64);
+
+
+static int sha256_ssse3_init(struct shash_desc *desc)
+{
+	struct sha256_state *sctx = shash_desc_ctx(desc);
+
+	sctx->state[0] = SHA256_H0;
+	sctx->state[1] = SHA256_H1;
+	sctx->state[2] = SHA256_H2;
+	sctx->state[3] = SHA256_H3;
+	sctx->state[4] = SHA256_H4;
+	sctx->state[5] = SHA256_H5;
+	sctx->state[6] = SHA256_H6;
+	sctx->state[7] = SHA256_H7;
+	sctx->count = 0;
+
+	return 0;
+}
+
+static int __sha256_ssse3_update(struct shash_desc *desc, const u8 *data,
+			       unsigned int len, unsigned int partial)
+{
+	struct sha256_state *sctx = shash_desc_ctx(desc);
+	unsigned int done = 0;
+
+	sctx->count += len;
+
+	if (partial) {
+		done = SHA256_BLOCK_SIZE - partial;
+		memcpy(sctx->buf + partial, data, done);
+		sha256_transform_asm(sctx->buf, sctx->state, 1);
+	}
+
+	if (len - done >= SHA256_BLOCK_SIZE) {
+		const unsigned int rounds = (len - done) / SHA256_BLOCK_SIZE;
+
+		sha256_transform_asm(data + done, sctx->state, (u64) rounds);
+
+		done += rounds * SHA256_BLOCK_SIZE;
+	}
+
+	memcpy(sctx->buf, data + done, len - done);
+
+	return 0;
+}
+
+static int sha256_ssse3_update(struct shash_desc *desc, const u8 *data,
+			     unsigned int len)
+{
+	struct sha256_state *sctx = shash_desc_ctx(desc);
+	unsigned int partial = sctx->count % SHA256_BLOCK_SIZE;
+	int res;
+
+	/* Handle the fast case right here */
+	if (partial + len < SHA256_BLOCK_SIZE) {
+		sctx->count += len;
+		memcpy(sctx->buf + partial, data, len);
+
+		return 0;
+	}
+
+	if (!irq_fpu_usable()) {
+		res = crypto_sha256_update(desc, data, len);
+	} else {
+		kernel_fpu_begin();
+		res = __sha256_ssse3_update(desc, data, len, partial);
+		kernel_fpu_end();
+	}
+
+	return res;
+}
+
+
+/* Add padding and return the message digest. */
+static int sha256_ssse3_final(struct shash_desc *desc, u8 *out)
+{
+	struct sha256_state *sctx = shash_desc_ctx(desc);
+	unsigned int i, index, padlen;
+	__be32 *dst = (__be32 *)out;
+	__be64 bits;
+	static const u8 padding[SHA256_BLOCK_SIZE] = { 0x80, };
+
+	bits = cpu_to_be64(sctx->count << 3);
+
+	/* Pad out to 56 mod 64 and append length */
+	index = sctx->count % SHA256_BLOCK_SIZE;
+	padlen = (index < 56) ? (56 - index) : ((SHA256_BLOCK_SIZE+56)-index);
+
+	if (!irq_fpu_usable()) {
+		crypto_sha256_update(desc, padding, padlen);
+		crypto_sha256_update(desc, (const u8 *)&bits, sizeof(bits));
+	} else {
+		kernel_fpu_begin();
+		/* We need to fill a whole block for __sha256_ssse3_update() */
+		if (padlen <= 56) {
+			sctx->count += padlen;
+			memcpy(sctx->buf + index, padding, padlen);
+		} else {
+			__sha256_ssse3_update(desc, padding, padlen, index);
+		}
+		__sha256_ssse3_update(desc, (const u8 *)&bits,
+					sizeof(bits), 56);
+		kernel_fpu_end();
+	}
+
+	/* Store state in digest */
+	for (i = 0; i < 8; i++)
+		dst[i] = cpu_to_be32(sctx->state[i]);
+
+	/* Wipe context */
+	memset(sctx, 0, sizeof(*sctx));
+
+	return 0;
+}
+
+static int sha256_ssse3_export(struct shash_desc *desc, void *out)
+{
+	struct sha256_state *sctx = shash_desc_ctx(desc);
+
+	memcpy(out, sctx, sizeof(*sctx));
+
+	return 0;
+}
+
+static int sha256_ssse3_import(struct shash_desc *desc, const void *in)
+{
+	struct sha256_state *sctx = shash_desc_ctx(desc);
+
+	memcpy(sctx, in, sizeof(*sctx));
+
+	return 0;
+}
+
+static struct shash_alg alg = {
+	.digestsize	=	SHA256_DIGEST_SIZE,
+	.init		=	sha256_ssse3_init,
+	.update		=	sha256_ssse3_update,
+	.final		=	sha256_ssse3_final,
+	.export		=	sha256_ssse3_export,
+	.import		=	sha256_ssse3_import,
+	.descsize	=	sizeof(struct sha256_state),
+	.statesize	=	sizeof(struct sha256_state),
+	.base		=	{
+		.cra_name	=	"sha256",
+		.cra_driver_name =	"sha256-ssse3",
+		.cra_priority	=	150,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA256_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+};
+
+#ifdef CONFIG_AS_AVX
+static bool __init avx_usable(void)
+{
+	u64 xcr0;
+
+	if (!cpu_has_avx || !cpu_has_osxsave)
+		return false;
+
+	xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
+	if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
+		pr_info("AVX detected but unusable.\n");
+
+		return false;
+	}
+
+	return true;
+}
+#endif
+
+static int __init sha256_ssse3_mod_init(void)
+{
+	/* test for SSE3 first */
+	if (cpu_has_ssse3)
+		sha256_transform_asm = sha256_transform_ssse3;
+
+#ifdef CONFIG_AS_AVX
+	/* allow AVX to override SSSE3, it's a little faster */
+	if (avx_usable()) {
+#ifdef CONFIG_AS_AVX2
+		if (boot_cpu_has(X86_FEATURE_AVX2))
+			sha256_transform_asm = sha256_transform_rorx;
+		else
+#endif
+			sha256_transform_asm = sha256_transform_avx;
+	}
+#endif
+
+	if (sha256_transform_asm) {
+#ifdef CONFIG_AS_AVX
+		if (sha256_transform_asm == sha256_transform_avx)
+			pr_info("Using AVX optimized SHA-256 implementation\n");
+#ifdef CONFIG_AS_AVX2
+		else if (sha256_transform_asm == sha256_transform_rorx)
+			pr_info("Using AVX2 optimized SHA-256 implementation\n");
+#endif
+		else
+#endif
+			pr_info("Using SSSE3 optimized SHA-256 implementation\n");
+		return crypto_register_shash(&alg);
+	}
+	pr_info("Neither AVX nor SSSE3 is available/usable.\n");
+
+	return -ENODEV;
+}
+
+static void __exit sha256_ssse3_mod_fini(void)
+{
+	crypto_unregister_shash(&alg);
+}
+
+module_init(sha256_ssse3_mod_init);
+module_exit(sha256_ssse3_mod_fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("SHA256 Secure Hash Algorithm, Supplemental SSE3 accelerated");
+
+MODULE_ALIAS("sha256");
diff --git a/arch/x86/crypto/sha512-avx-asm.S b/arch/x86/crypto/sha512-avx-asm.S
new file mode 100644
index 000000000000..974dde9bc6cd
--- /dev/null
+++ b/arch/x86/crypto/sha512-avx-asm.S
@@ -0,0 +1,423 @@
+########################################################################
+# Implement fast SHA-512 with AVX instructions. (x86_64)
+#
+# Copyright (C) 2013 Intel Corporation.
+#
+# Authors:
+#     James Guilford <james.guilford@intel.com>
+#     Kirk Yap <kirk.s.yap@intel.com>
+#     David Cote <david.m.cote@intel.com>
+#     Tim Chen <tim.c.chen@linux.intel.com>
+#
+# This software is available to you under a choice of one of two
+# licenses.  You may choose to be licensed under the terms of the GNU
+# General Public License (GPL) Version 2, available from the file
+# COPYING in the main directory of this source tree, or the
+# OpenIB.org BSD license below:
+#
+#     Redistribution and use in source and binary forms, with or
+#     without modification, are permitted provided that the following
+#     conditions are met:
+#
+#      - Redistributions of source code must retain the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer.
+#
+#      - Redistributions in binary form must reproduce the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer in the documentation and/or other materials
+#        provided with the distribution.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+# BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+# ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+#
+########################################################################
+#
+# This code is described in an Intel White-Paper:
+# "Fast SHA-512 Implementations on Intel Architecture Processors"
+#
+# To find it, surf to http://www.intel.com/p/en_US/embedded
+# and search for that title.
+#
+########################################################################
+
+#ifdef CONFIG_AS_AVX
+#include <linux/linkage.h>
+
+.text
+
+# Virtual Registers
+# ARG1
+msg	= %rdi
+# ARG2
+digest	= %rsi
+# ARG3
+msglen	= %rdx
+T1	= %rcx
+T2	= %r8
+a_64	= %r9
+b_64	= %r10
+c_64	= %r11
+d_64	= %r12
+e_64	= %r13
+f_64	= %r14
+g_64	= %r15
+h_64	= %rbx
+tmp0	= %rax
+
+# Local variables (stack frame)
+
+# Message Schedule
+W_SIZE = 80*8
+# W[t] + K[t] | W[t+1] + K[t+1]
+WK_SIZE = 2*8
+RSPSAVE_SIZE = 1*8
+GPRSAVE_SIZE = 5*8
+
+frame_W = 0
+frame_WK = frame_W + W_SIZE
+frame_RSPSAVE = frame_WK + WK_SIZE
+frame_GPRSAVE = frame_RSPSAVE + RSPSAVE_SIZE
+frame_size = frame_GPRSAVE + GPRSAVE_SIZE
+
+# Useful QWORD "arrays" for simpler memory references
+# MSG, DIGEST, K_t, W_t are arrays
+# WK_2(t) points to 1 of 2 qwords at frame.WK depdending on t being odd/even
+
+# Input message (arg1)
+#define MSG(i)    8*i(msg)
+
+# Output Digest (arg2)
+#define DIGEST(i) 8*i(digest)
+
+# SHA Constants (static mem)
+#define K_t(i)    8*i+K512(%rip)
+
+# Message Schedule (stack frame)
+#define W_t(i)    8*i+frame_W(%rsp)
+
+# W[t]+K[t] (stack frame)
+#define WK_2(i)   8*((i%2))+frame_WK(%rsp)
+
+.macro RotateState
+	# Rotate symbols a..h right
+	TMP   = h_64
+	h_64  = g_64
+	g_64  = f_64
+	f_64  = e_64
+	e_64  = d_64
+	d_64  = c_64
+	c_64  = b_64
+	b_64  = a_64
+	a_64  = TMP
+.endm
+
+.macro RORQ p1 p2
+	# shld is faster than ror on Sandybridge
+	shld	$(64-\p2), \p1, \p1
+.endm
+
+.macro SHA512_Round rnd
+	# Compute Round %%t
+	mov     f_64, T1          # T1 = f
+	mov     e_64, tmp0        # tmp = e
+	xor     g_64, T1          # T1 = f ^ g
+	RORQ    tmp0, 23   # 41    # tmp = e ror 23
+	and     e_64, T1          # T1 = (f ^ g) & e
+	xor     e_64, tmp0        # tmp = (e ror 23) ^ e
+	xor     g_64, T1          # T1 = ((f ^ g) & e) ^ g = CH(e,f,g)
+	idx = \rnd
+	add     WK_2(idx), T1     # W[t] + K[t] from message scheduler
+	RORQ    tmp0, 4   # 18    # tmp = ((e ror 23) ^ e) ror 4
+	xor     e_64, tmp0        # tmp = (((e ror 23) ^ e) ror 4) ^ e
+	mov     a_64, T2          # T2 = a
+	add     h_64, T1          # T1 = CH(e,f,g) + W[t] + K[t] + h
+	RORQ    tmp0, 14  # 14    # tmp = ((((e ror23)^e)ror4)^e)ror14 = S1(e)
+	add     tmp0, T1          # T1 = CH(e,f,g) + W[t] + K[t] + S1(e)
+	mov     a_64, tmp0        # tmp = a
+	xor     c_64, T2          # T2 = a ^ c
+	and     c_64, tmp0        # tmp = a & c
+	and     b_64, T2          # T2 = (a ^ c) & b
+	xor     tmp0, T2          # T2 = ((a ^ c) & b) ^ (a & c) = Maj(a,b,c)
+	mov     a_64, tmp0        # tmp = a
+	RORQ    tmp0, 5  # 39     # tmp = a ror 5
+	xor     a_64, tmp0        # tmp = (a ror 5) ^ a
+	add     T1, d_64          # e(next_state) = d + T1
+	RORQ    tmp0, 6  # 34     # tmp = ((a ror 5) ^ a) ror 6
+	xor     a_64, tmp0        # tmp = (((a ror 5) ^ a) ror 6) ^ a
+	lea     (T1, T2), h_64    # a(next_state) = T1 + Maj(a,b,c)
+	RORQ    tmp0, 28  # 28    # tmp = ((((a ror5)^a)ror6)^a)ror28 = S0(a)
+	add     tmp0, h_64        # a(next_state) = T1 + Maj(a,b,c) S0(a)
+	RotateState
+.endm
+
+.macro SHA512_2Sched_2Round_avx rnd
+	# Compute rounds t-2 and t-1
+	# Compute message schedule QWORDS t and t+1
+
+	#   Two rounds are computed based on the values for K[t-2]+W[t-2] and
+	# K[t-1]+W[t-1] which were previously stored at WK_2 by the message
+	# scheduler.
+	#   The two new schedule QWORDS are stored at [W_t(t)] and [W_t(t+1)].
+	# They are then added to their respective SHA512 constants at
+	# [K_t(t)] and [K_t(t+1)] and stored at dqword [WK_2(t)]
+	#   For brievity, the comments following vectored instructions only refer to
+	# the first of a pair of QWORDS.
+	# Eg. XMM4=W[t-2] really means XMM4={W[t-2]|W[t-1]}
+	#   The computation of the message schedule and the rounds are tightly
+	# stitched to take advantage of instruction-level parallelism.
+
+	idx = \rnd - 2
+	vmovdqa	W_t(idx), %xmm4		# XMM4 = W[t-2]
+	idx = \rnd - 15
+	vmovdqu	W_t(idx), %xmm5		# XMM5 = W[t-15]
+	mov	f_64, T1
+	vpsrlq	$61, %xmm4, %xmm0	# XMM0 = W[t-2]>>61
+	mov	e_64, tmp0
+	vpsrlq	$1, %xmm5, %xmm6	# XMM6 = W[t-15]>>1
+	xor	g_64, T1
+	RORQ	tmp0, 23 # 41
+	vpsrlq	$19, %xmm4, %xmm1	# XMM1 = W[t-2]>>19
+	and	e_64, T1
+	xor	e_64, tmp0
+	vpxor	%xmm1, %xmm0, %xmm0	# XMM0 = W[t-2]>>61 ^ W[t-2]>>19
+	xor	g_64, T1
+	idx = \rnd
+	add	WK_2(idx), T1#
+	vpsrlq	$8, %xmm5, %xmm7	# XMM7 = W[t-15]>>8
+	RORQ	tmp0, 4 # 18
+	vpsrlq	$6, %xmm4, %xmm2	# XMM2 = W[t-2]>>6
+	xor	e_64, tmp0
+	mov	a_64, T2
+	add	h_64, T1
+	vpxor	%xmm7, %xmm6, %xmm6	# XMM6 = W[t-15]>>1 ^ W[t-15]>>8
+	RORQ	tmp0, 14 # 14
+	add	tmp0, T1
+	vpsrlq	$7, %xmm5, %xmm8	# XMM8 = W[t-15]>>7
+	mov	a_64, tmp0
+	xor	c_64, T2
+	vpsllq	$(64-61), %xmm4, %xmm3  # XMM3 = W[t-2]<<3
+	and	c_64, tmp0
+	and	b_64, T2
+	vpxor	%xmm3, %xmm2, %xmm2	# XMM2 = W[t-2]>>6 ^ W[t-2]<<3
+	xor	tmp0, T2
+	mov	a_64, tmp0
+	vpsllq	$(64-1), %xmm5, %xmm9	# XMM9 = W[t-15]<<63
+	RORQ	tmp0, 5 # 39
+	vpxor	%xmm9, %xmm8, %xmm8	# XMM8 = W[t-15]>>7 ^ W[t-15]<<63
+	xor	a_64, tmp0
+	add	T1, d_64
+	RORQ	tmp0, 6 # 34
+	xor	a_64, tmp0
+	vpxor	%xmm8, %xmm6, %xmm6	# XMM6 = W[t-15]>>1 ^ W[t-15]>>8 ^
+					#  W[t-15]>>7 ^ W[t-15]<<63
+	lea	(T1, T2), h_64
+	RORQ	tmp0, 28 # 28
+	vpsllq	$(64-19), %xmm4, %xmm4  # XMM4 = W[t-2]<<25
+	add	tmp0, h_64
+	RotateState
+	vpxor	%xmm4, %xmm0, %xmm0     # XMM0 = W[t-2]>>61 ^ W[t-2]>>19 ^
+					#        W[t-2]<<25
+	mov	f_64, T1
+	vpxor	%xmm2, %xmm0, %xmm0     # XMM0 = s1(W[t-2])
+	mov	e_64, tmp0
+	xor	g_64, T1
+	idx = \rnd - 16
+	vpaddq	W_t(idx), %xmm0, %xmm0  # XMM0 = s1(W[t-2]) + W[t-16]
+	idx = \rnd - 7
+	vmovdqu	W_t(idx), %xmm1		# XMM1 = W[t-7]
+	RORQ	tmp0, 23 # 41
+	and	e_64, T1
+	xor	e_64, tmp0
+	xor	g_64, T1
+	vpsllq	$(64-8), %xmm5, %xmm5   # XMM5 = W[t-15]<<56
+	idx = \rnd + 1
+	add	WK_2(idx), T1
+	vpxor	%xmm5, %xmm6, %xmm6     # XMM6 = s0(W[t-15])
+	RORQ	tmp0, 4 # 18
+	vpaddq	%xmm6, %xmm0, %xmm0     # XMM0 = s1(W[t-2]) + W[t-16] + s0(W[t-15])
+	xor	e_64, tmp0
+	vpaddq	%xmm1, %xmm0, %xmm0     # XMM0 = W[t] = s1(W[t-2]) + W[t-7] +
+					#               s0(W[t-15]) + W[t-16]
+	mov	a_64, T2
+	add	h_64, T1
+	RORQ	tmp0, 14 # 14
+	add	tmp0, T1
+	idx = \rnd
+	vmovdqa	%xmm0, W_t(idx)		# Store W[t]
+	vpaddq	K_t(idx), %xmm0, %xmm0  # Compute W[t]+K[t]
+	vmovdqa	%xmm0, WK_2(idx)	# Store W[t]+K[t] for next rounds
+	mov	a_64, tmp0
+	xor	c_64, T2
+	and	c_64, tmp0
+	and	b_64, T2
+	xor	tmp0, T2
+	mov	a_64, tmp0
+	RORQ	tmp0, 5 # 39
+	xor	a_64, tmp0
+	add	T1, d_64
+	RORQ	tmp0, 6 # 34
+	xor	a_64, tmp0
+	lea	(T1, T2), h_64
+	RORQ	tmp0, 28 # 28
+	add	tmp0, h_64
+	RotateState
+.endm
+
+########################################################################
+# void sha512_transform_avx(const void* M, void* D, u64 L)
+# Purpose: Updates the SHA512 digest stored at D with the message stored in M.
+# The size of the message pointed to by M must be an integer multiple of SHA512
+# message blocks.
+# L is the message length in SHA512 blocks
+########################################################################
+ENTRY(sha512_transform_avx)
+	cmp $0, msglen
+	je nowork
+
+	# Allocate Stack Space
+	mov	%rsp, %rax
+	sub     $frame_size, %rsp
+	and	$~(0x20 - 1), %rsp
+	mov	%rax, frame_RSPSAVE(%rsp)
+
+	# Save GPRs
+	mov     %rbx, frame_GPRSAVE(%rsp)
+	mov     %r12, frame_GPRSAVE +8*1(%rsp)
+	mov     %r13, frame_GPRSAVE +8*2(%rsp)
+	mov     %r14, frame_GPRSAVE +8*3(%rsp)
+	mov     %r15, frame_GPRSAVE +8*4(%rsp)
+
+updateblock:
+
+	# Load state variables
+	mov     DIGEST(0), a_64
+	mov     DIGEST(1), b_64
+	mov     DIGEST(2), c_64
+	mov     DIGEST(3), d_64
+	mov     DIGEST(4), e_64
+	mov     DIGEST(5), f_64
+	mov     DIGEST(6), g_64
+	mov     DIGEST(7), h_64
+
+	t = 0
+	.rept 80/2 + 1
+	# (80 rounds) / (2 rounds/iteration) + (1 iteration)
+	# +1 iteration because the scheduler leads hashing by 1 iteration
+		.if t < 2
+			# BSWAP 2 QWORDS
+			vmovdqa  XMM_QWORD_BSWAP(%rip), %xmm1
+			vmovdqu  MSG(t), %xmm0
+			vpshufb  %xmm1, %xmm0, %xmm0    # BSWAP
+			vmovdqa  %xmm0, W_t(t) # Store Scheduled Pair
+			vpaddq   K_t(t), %xmm0, %xmm0 # Compute W[t]+K[t]
+			vmovdqa  %xmm0, WK_2(t) # Store into WK for rounds
+		.elseif t < 16
+			# BSWAP 2 QWORDS# Compute 2 Rounds
+			vmovdqu  MSG(t), %xmm0
+			vpshufb  %xmm1, %xmm0, %xmm0    # BSWAP
+			SHA512_Round t-2    # Round t-2
+			vmovdqa  %xmm0, W_t(t) # Store Scheduled Pair
+			vpaddq   K_t(t), %xmm0, %xmm0 # Compute W[t]+K[t]
+			SHA512_Round t-1    # Round t-1
+			vmovdqa  %xmm0, WK_2(t)# Store W[t]+K[t] into WK
+		.elseif t < 79
+			# Schedule 2 QWORDS# Compute 2 Rounds
+			SHA512_2Sched_2Round_avx t
+		.else
+			# Compute 2 Rounds
+			SHA512_Round t-2
+			SHA512_Round t-1
+		.endif
+		t = t+2
+	.endr
+
+	# Update digest
+	add     a_64, DIGEST(0)
+	add     b_64, DIGEST(1)
+	add     c_64, DIGEST(2)
+	add     d_64, DIGEST(3)
+	add     e_64, DIGEST(4)
+	add     f_64, DIGEST(5)
+	add     g_64, DIGEST(6)
+	add     h_64, DIGEST(7)
+
+	# Advance to next message block
+	add     $16*8, msg
+	dec     msglen
+	jnz     updateblock
+
+	# Restore GPRs
+	mov     frame_GPRSAVE(%rsp),      %rbx
+	mov     frame_GPRSAVE +8*1(%rsp), %r12
+	mov     frame_GPRSAVE +8*2(%rsp), %r13
+	mov     frame_GPRSAVE +8*3(%rsp), %r14
+	mov     frame_GPRSAVE +8*4(%rsp), %r15
+
+	# Restore Stack Pointer
+	mov	frame_RSPSAVE(%rsp), %rsp
+
+nowork:
+	ret
+ENDPROC(sha512_transform_avx)
+
+########################################################################
+### Binary Data
+
+.data
+
+.align 16
+
+# Mask for byte-swapping a couple of qwords in an XMM register using (v)pshufb.
+XMM_QWORD_BSWAP:
+	.octa 0x08090a0b0c0d0e0f0001020304050607
+
+# K[t] used in SHA512 hashing
+K512:
+	.quad 0x428a2f98d728ae22,0x7137449123ef65cd
+	.quad 0xb5c0fbcfec4d3b2f,0xe9b5dba58189dbbc
+	.quad 0x3956c25bf348b538,0x59f111f1b605d019
+	.quad 0x923f82a4af194f9b,0xab1c5ed5da6d8118
+	.quad 0xd807aa98a3030242,0x12835b0145706fbe
+	.quad 0x243185be4ee4b28c,0x550c7dc3d5ffb4e2
+	.quad 0x72be5d74f27b896f,0x80deb1fe3b1696b1
+	.quad 0x9bdc06a725c71235,0xc19bf174cf692694
+	.quad 0xe49b69c19ef14ad2,0xefbe4786384f25e3
+	.quad 0x0fc19dc68b8cd5b5,0x240ca1cc77ac9c65
+	.quad 0x2de92c6f592b0275,0x4a7484aa6ea6e483
+	.quad 0x5cb0a9dcbd41fbd4,0x76f988da831153b5
+	.quad 0x983e5152ee66dfab,0xa831c66d2db43210
+	.quad 0xb00327c898fb213f,0xbf597fc7beef0ee4
+	.quad 0xc6e00bf33da88fc2,0xd5a79147930aa725
+	.quad 0x06ca6351e003826f,0x142929670a0e6e70
+	.quad 0x27b70a8546d22ffc,0x2e1b21385c26c926
+	.quad 0x4d2c6dfc5ac42aed,0x53380d139d95b3df
+	.quad 0x650a73548baf63de,0x766a0abb3c77b2a8
+	.quad 0x81c2c92e47edaee6,0x92722c851482353b
+	.quad 0xa2bfe8a14cf10364,0xa81a664bbc423001
+	.quad 0xc24b8b70d0f89791,0xc76c51a30654be30
+	.quad 0xd192e819d6ef5218,0xd69906245565a910
+	.quad 0xf40e35855771202a,0x106aa07032bbd1b8
+	.quad 0x19a4c116b8d2d0c8,0x1e376c085141ab53
+	.quad 0x2748774cdf8eeb99,0x34b0bcb5e19b48a8
+	.quad 0x391c0cb3c5c95a63,0x4ed8aa4ae3418acb
+	.quad 0x5b9cca4f7763e373,0x682e6ff3d6b2b8a3
+	.quad 0x748f82ee5defb2fc,0x78a5636f43172f60
+	.quad 0x84c87814a1f0ab72,0x8cc702081a6439ec
+	.quad 0x90befffa23631e28,0xa4506cebde82bde9
+	.quad 0xbef9a3f7b2c67915,0xc67178f2e372532b
+	.quad 0xca273eceea26619c,0xd186b8c721c0c207
+	.quad 0xeada7dd6cde0eb1e,0xf57d4f7fee6ed178
+	.quad 0x06f067aa72176fba,0x0a637dc5a2c898a6
+	.quad 0x113f9804bef90dae,0x1b710b35131c471b
+	.quad 0x28db77f523047d84,0x32caab7b40c72493
+	.quad 0x3c9ebe0a15c9bebc,0x431d67c49c100d4c
+	.quad 0x4cc5d4becb3e42b6,0x597f299cfc657e2a
+	.quad 0x5fcb6fab3ad6faec,0x6c44198c4a475817
+#endif
diff --git a/arch/x86/crypto/sha512-avx2-asm.S b/arch/x86/crypto/sha512-avx2-asm.S
new file mode 100644
index 000000000000..568b96105f5c
--- /dev/null
+++ b/arch/x86/crypto/sha512-avx2-asm.S
@@ -0,0 +1,743 @@
+########################################################################
+# Implement fast SHA-512 with AVX2 instructions. (x86_64)
+#
+# Copyright (C) 2013 Intel Corporation.
+#
+# Authors:
+#     James Guilford <james.guilford@intel.com>
+#     Kirk Yap <kirk.s.yap@intel.com>
+#     David Cote <david.m.cote@intel.com>
+#     Tim Chen <tim.c.chen@linux.intel.com>
+#
+# This software is available to you under a choice of one of two
+# licenses.  You may choose to be licensed under the terms of the GNU
+# General Public License (GPL) Version 2, available from the file
+# COPYING in the main directory of this source tree, or the
+# OpenIB.org BSD license below:
+#
+#     Redistribution and use in source and binary forms, with or
+#     without modification, are permitted provided that the following
+#     conditions are met:
+#
+#      - Redistributions of source code must retain the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer.
+#
+#      - Redistributions in binary form must reproduce the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer in the documentation and/or other materials
+#        provided with the distribution.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+# BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+# ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+#
+########################################################################
+#
+# This code is described in an Intel White-Paper:
+# "Fast SHA-512 Implementations on Intel Architecture Processors"
+#
+# To find it, surf to http://www.intel.com/p/en_US/embedded
+# and search for that title.
+#
+########################################################################
+# This code schedules 1 blocks at a time, with 4 lanes per block
+########################################################################
+
+#ifdef CONFIG_AS_AVX2
+#include <linux/linkage.h>
+
+.text
+
+# Virtual Registers
+Y_0 = %ymm4
+Y_1 = %ymm5
+Y_2 = %ymm6
+Y_3 = %ymm7
+
+YTMP0 = %ymm0
+YTMP1 = %ymm1
+YTMP2 = %ymm2
+YTMP3 = %ymm3
+YTMP4 = %ymm8
+XFER  = YTMP0
+
+BYTE_FLIP_MASK  = %ymm9
+
+# 1st arg
+INP         = %rdi
+# 2nd arg
+CTX         = %rsi
+# 3rd arg
+NUM_BLKS    = %rdx
+
+c           = %rcx
+d           = %r8
+e           = %rdx
+y3          = %rdi
+
+TBL   = %rbp
+
+a     = %rax
+b     = %rbx
+
+f     = %r9
+g     = %r10
+h     = %r11
+old_h = %r11
+
+T1    = %r12
+y0    = %r13
+y1    = %r14
+y2    = %r15
+
+y4    = %r12
+
+# Local variables (stack frame)
+XFER_SIZE = 4*8
+SRND_SIZE = 1*8
+INP_SIZE = 1*8
+INPEND_SIZE = 1*8
+RSPSAVE_SIZE = 1*8
+GPRSAVE_SIZE = 6*8
+
+frame_XFER = 0
+frame_SRND = frame_XFER + XFER_SIZE
+frame_INP = frame_SRND + SRND_SIZE
+frame_INPEND = frame_INP + INP_SIZE
+frame_RSPSAVE = frame_INPEND + INPEND_SIZE
+frame_GPRSAVE = frame_RSPSAVE + RSPSAVE_SIZE
+frame_size = frame_GPRSAVE + GPRSAVE_SIZE
+
+## assume buffers not aligned
+#define	VMOVDQ vmovdqu
+
+# addm [mem], reg
+# Add reg to mem using reg-mem add and store
+.macro addm p1 p2
+	add	\p1, \p2
+	mov	\p2, \p1
+.endm
+
+
+# COPY_YMM_AND_BSWAP ymm, [mem], byte_flip_mask
+# Load ymm with mem and byte swap each dword
+.macro COPY_YMM_AND_BSWAP p1 p2 p3
+	VMOVDQ \p2, \p1
+	vpshufb \p3, \p1, \p1
+.endm
+# rotate_Ys
+# Rotate values of symbols Y0...Y3
+.macro rotate_Ys
+	Y_ = Y_0
+	Y_0 = Y_1
+	Y_1 = Y_2
+	Y_2 = Y_3
+	Y_3 = Y_
+.endm
+
+# RotateState
+.macro RotateState
+	# Rotate symbols a..h right
+	old_h  = h
+	TMP_   = h
+	h      = g
+	g      = f
+	f      = e
+	e      = d
+	d      = c
+	c      = b
+	b      = a
+	a      = TMP_
+.endm
+
+# macro MY_VPALIGNR	YDST, YSRC1, YSRC2, RVAL
+# YDST = {YSRC1, YSRC2} >> RVAL*8
+.macro MY_VPALIGNR YDST YSRC1 YSRC2 RVAL
+	vperm2f128      $0x3, \YSRC2, \YSRC1, \YDST     # YDST = {YS1_LO, YS2_HI}
+	vpalignr        $\RVAL, \YSRC2, \YDST, \YDST    # YDST = {YDS1, YS2} >> RVAL*8
+.endm
+
+.macro FOUR_ROUNDS_AND_SCHED
+################################### RND N + 0 #########################################
+
+	# Extract w[t-7]
+	MY_VPALIGNR	YTMP0, Y_3, Y_2, 8		# YTMP0 = W[-7]
+	# Calculate w[t-16] + w[t-7]
+	vpaddq		Y_0, YTMP0, YTMP0		# YTMP0 = W[-7] + W[-16]
+	# Extract w[t-15]
+	MY_VPALIGNR	YTMP1, Y_1, Y_0, 8		# YTMP1 = W[-15]
+
+	# Calculate sigma0
+
+	# Calculate w[t-15] ror 1
+	vpsrlq		$1, YTMP1, YTMP2
+	vpsllq		$(64-1), YTMP1, YTMP3
+	vpor		YTMP2, YTMP3, YTMP3		# YTMP3 = W[-15] ror 1
+	# Calculate w[t-15] shr 7
+	vpsrlq		$7, YTMP1, YTMP4		# YTMP4 = W[-15] >> 7
+
+	mov	a, y3		# y3 = a                                # MAJA
+	rorx	$41, e, y0	# y0 = e >> 41				# S1A
+	rorx	$18, e, y1	# y1 = e >> 18				# S1B
+	add	frame_XFER(%rsp),h		# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$34, a, T1	# T1 = a >> 34				# S0B
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18)		# S1
+	xor	g, y2		# y2 = f^g                              # CH
+	rorx	$14, e, y1	# y1 = (e >> 14)			# S1
+
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18) ^ (e>>14)	# S1
+	rorx	$39, a, y1	# y1 = a >> 39				# S0A
+	add	h, d		# d = k + w + h + d                     # --
+
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34)		# S0
+	rorx	$28, a, T1	# T1 = (a >> 28)			# S0
+
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34) ^ (a>>28)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	c, T1		# T1 = a&c                              # MAJB
+
+	add	y0, y2		# y2 = S1 + CH                          # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+	RotateState
+
+################################### RND N + 1 #########################################
+
+	# Calculate w[t-15] ror 8
+	vpsrlq		$8, YTMP1, YTMP2
+	vpsllq		$(64-8), YTMP1, YTMP1
+	vpor		YTMP2, YTMP1, YTMP1		# YTMP1 = W[-15] ror 8
+	# XOR the three components
+	vpxor		YTMP4, YTMP3, YTMP3		# YTMP3 = W[-15] ror 1 ^ W[-15] >> 7
+	vpxor		YTMP1, YTMP3, YTMP1		# YTMP1 = s0
+
+
+	# Add three components, w[t-16], w[t-7] and sigma0
+	vpaddq		YTMP1, YTMP0, YTMP0		# YTMP0 = W[-16] + W[-7] + s0
+	# Move to appropriate lanes for calculating w[16] and w[17]
+	vperm2f128	$0x0, YTMP0, YTMP0, Y_0		# Y_0 = W[-16] + W[-7] + s0 {BABA}
+	# Move to appropriate lanes for calculating w[18] and w[19]
+	vpand		MASK_YMM_LO(%rip), YTMP0, YTMP0	# YTMP0 = W[-16] + W[-7] + s0 {DC00}
+
+	# Calculate w[16] and w[17] in both 128 bit lanes
+
+	# Calculate sigma1 for w[16] and w[17] on both 128 bit lanes
+	vperm2f128	$0x11, Y_3, Y_3, YTMP2		# YTMP2 = W[-2] {BABA}
+	vpsrlq		$6, YTMP2, YTMP4		# YTMP4 = W[-2] >> 6 {BABA}
+
+
+	mov	a, y3		# y3 = a                                # MAJA
+	rorx	$41, e, y0	# y0 = e >> 41				# S1A
+	rorx	$18, e, y1	# y1 = e >> 18				# S1B
+	add	1*8+frame_XFER(%rsp), h		# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$34, a, T1	# T1 = a >> 34				# S0B
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18)		# S1
+	xor	g, y2		# y2 = f^g                              # CH
+
+
+	rorx	$14, e, y1	# y1 = (e >> 14)			# S1
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18) ^ (e>>14)	# S1
+	rorx	$39, a, y1	# y1 = a >> 39				# S0A
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	h, d		# d = k + w + h + d                     # --
+
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34)		# S0
+
+	rorx	$28, a, T1	# T1 = (a >> 28)			# S0
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34) ^ (a>>28)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+	RotateState
+
+
+################################### RND N + 2 #########################################
+
+	vpsrlq		$19, YTMP2, YTMP3		# YTMP3 = W[-2] >> 19 {BABA}
+	vpsllq		$(64-19), YTMP2, YTMP1		# YTMP1 = W[-2] << 19 {BABA}
+	vpor		YTMP1, YTMP3, YTMP3		# YTMP3 = W[-2] ror 19 {BABA}
+	vpxor		YTMP3, YTMP4, YTMP4		# YTMP4 = W[-2] ror 19 ^ W[-2] >> 6 {BABA}
+	vpsrlq		$61, YTMP2, YTMP3		# YTMP3 = W[-2] >> 61 {BABA}
+	vpsllq		$(64-61), YTMP2, YTMP1		# YTMP1 = W[-2] << 61 {BABA}
+	vpor		YTMP1, YTMP3, YTMP3		# YTMP3 = W[-2] ror 61 {BABA}
+	vpxor		YTMP3, YTMP4, YTMP4		# YTMP4 = s1 = (W[-2] ror 19) ^
+							#  (W[-2] ror 61) ^ (W[-2] >> 6) {BABA}
+
+	# Add sigma1 to the other compunents to get w[16] and w[17]
+	vpaddq		YTMP4, Y_0, Y_0			# Y_0 = {W[1], W[0], W[1], W[0]}
+
+	# Calculate sigma1 for w[18] and w[19] for upper 128 bit lane
+	vpsrlq		$6, Y_0, YTMP4			# YTMP4 = W[-2] >> 6 {DC--}
+
+	mov	a, y3		# y3 = a                                # MAJA
+	rorx	$41, e, y0	# y0 = e >> 41				# S1A
+	add	2*8+frame_XFER(%rsp), h		# h = k + w + h         # --
+
+	rorx	$18, e, y1	# y1 = e >> 18				# S1B
+	or	c, y3		# y3 = a|c                              # MAJA
+	mov	f, y2		# y2 = f                                # CH
+	xor	g, y2		# y2 = f^g                              # CH
+
+	rorx	$34, a, T1	# T1 = a >> 34				# S0B
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18)		# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+
+	rorx	$14, e, y1	# y1 = (e >> 14)			# S1
+	add	h, d		# d = k + w + h + d                     # --
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18) ^ (e>>14)	# S1
+	rorx	$39, a, y1	# y1 = a >> 39				# S0A
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34)		# S0
+	rorx	$28, a, T1	# T1 = (a >> 28)			# S0
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34) ^ (a>>28)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+	RotateState
+
+################################### RND N + 3 #########################################
+
+	vpsrlq		$19, Y_0, YTMP3			# YTMP3 = W[-2] >> 19 {DC--}
+	vpsllq		$(64-19), Y_0, YTMP1		# YTMP1 = W[-2] << 19 {DC--}
+	vpor		YTMP1, YTMP3, YTMP3		# YTMP3 = W[-2] ror 19 {DC--}
+	vpxor		YTMP3, YTMP4, YTMP4		# YTMP4 = W[-2] ror 19 ^ W[-2] >> 6 {DC--}
+	vpsrlq		$61, Y_0, YTMP3			# YTMP3 = W[-2] >> 61 {DC--}
+	vpsllq		$(64-61), Y_0, YTMP1		# YTMP1 = W[-2] << 61 {DC--}
+	vpor		YTMP1, YTMP3, YTMP3		# YTMP3 = W[-2] ror 61 {DC--}
+	vpxor		YTMP3, YTMP4, YTMP4		# YTMP4 = s1 = (W[-2] ror 19) ^
+							#  (W[-2] ror 61) ^ (W[-2] >> 6) {DC--}
+
+	# Add the sigma0 + w[t-7] + w[t-16] for w[18] and w[19]
+	# to newly calculated sigma1 to get w[18] and w[19]
+	vpaddq		YTMP4, YTMP0, YTMP2		# YTMP2 = {W[3], W[2], --, --}
+
+	# Form w[19, w[18], w17], w[16]
+	vpblendd		$0xF0, YTMP2, Y_0, Y_0		# Y_0 = {W[3], W[2], W[1], W[0]}
+
+	mov	a, y3		# y3 = a                                # MAJA
+	rorx	$41, e, y0	# y0 = e >> 41				# S1A
+	rorx	$18, e, y1	# y1 = e >> 18				# S1B
+	add	3*8+frame_XFER(%rsp), h		# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$34, a, T1	# T1 = a >> 34				# S0B
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18)		# S1
+	xor	g, y2		# y2 = f^g                              # CH
+
+
+	rorx	$14, e, y1	# y1 = (e >> 14)			# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	h, d		# d = k + w + h + d                     # --
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18) ^ (e>>14)	# S1
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+
+	rorx	$39, a, y1	# y1 = a >> 39				# S0A
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34)		# S0
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	rorx	$28, a, T1	# T1 = (a >> 28)			# S0
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34) ^ (a>>28)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	c, T1		# T1 = a&c                              # MAJB
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+
+	add	y1, h		# h = k + w + h + S0                    # --
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+	RotateState
+
+	rotate_Ys
+.endm
+
+.macro DO_4ROUNDS
+
+################################### RND N + 0 #########################################
+
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$41, e, y0	# y0 = e >> 41				# S1A
+	rorx	$18, e, y1	# y1 = e >> 18				# S1B
+	xor	g, y2		# y2 = f^g                              # CH
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18)		# S1
+	rorx	$14, e, y1	# y1 = (e >> 14)			# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18) ^ (e>>14)	# S1
+	rorx	$34, a, T1	# T1 = a >> 34				# S0B
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	rorx	$39, a, y1	# y1 = a >> 39				# S0A
+	mov	a, y3		# y3 = a                                # MAJA
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34)		# S0
+	rorx	$28, a, T1	# T1 = (a >> 28)			# S0
+	add	frame_XFER(%rsp), h		# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34) ^ (a>>28)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+	add	h, d		# d = k + w + h + d                     # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	RotateState
+
+################################### RND N + 1 #########################################
+
+	add	y2, old_h	# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$41, e, y0	# y0 = e >> 41				# S1A
+	rorx	$18, e, y1	# y1 = e >> 18				# S1B
+	xor	g, y2		# y2 = f^g                              # CH
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18)		# S1
+	rorx	$14, e, y1	# y1 = (e >> 14)			# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	y3, old_h	# h = t1 + S0 + MAJ                     # --
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18) ^ (e>>14)	# S1
+	rorx	$34, a, T1	# T1 = a >> 34				# S0B
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	rorx	$39, a, y1	# y1 = a >> 39				# S0A
+	mov	a, y3		# y3 = a                                # MAJA
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34)		# S0
+	rorx	$28, a, T1	# T1 = (a >> 28)			# S0
+	add	8*1+frame_XFER(%rsp), h		# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34) ^ (a>>28)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+	add	h, d		# d = k + w + h + d                     # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	RotateState
+
+################################### RND N + 2 #########################################
+
+	add	y2, old_h	# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$41, e, y0	# y0 = e >> 41				# S1A
+	rorx	$18, e, y1	# y1 = e >> 18				# S1B
+	xor	g, y2		# y2 = f^g                              # CH
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18)		# S1
+	rorx	$14, e, y1	# y1 = (e >> 14)			# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	y3, old_h	# h = t1 + S0 + MAJ                     # --
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18) ^ (e>>14)	# S1
+	rorx	$34, a, T1	# T1 = a >> 34				# S0B
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	rorx	$39, a, y1	# y1 = a >> 39				# S0A
+	mov	a, y3		# y3 = a                                # MAJA
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34)		# S0
+	rorx	$28, a, T1	# T1 = (a >> 28)			# S0
+	add	8*2+frame_XFER(%rsp), h		# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34) ^ (a>>28)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+	add	h, d		# d = k + w + h + d                     # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	RotateState
+
+################################### RND N + 3 #########################################
+
+	add	y2, old_h	# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+	mov	f, y2		# y2 = f                                # CH
+	rorx	$41, e, y0	# y0 = e >> 41				# S1A
+	rorx	$18, e, y1	# y1 = e >> 18				# S1B
+	xor	g, y2		# y2 = f^g                              # CH
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18)		# S1
+	rorx	$14, e, y1	# y1 = (e >> 14)			# S1
+	and	e, y2		# y2 = (f^g)&e                          # CH
+	add	y3, old_h	# h = t1 + S0 + MAJ                     # --
+
+	xor	y1, y0		# y0 = (e>>41) ^ (e>>18) ^ (e>>14)	# S1
+	rorx	$34, a, T1	# T1 = a >> 34				# S0B
+	xor	g, y2		# y2 = CH = ((f^g)&e)^g                 # CH
+	rorx	$39, a, y1	# y1 = a >> 39				# S0A
+	mov	a, y3		# y3 = a                                # MAJA
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34)		# S0
+	rorx	$28, a, T1	# T1 = (a >> 28)			# S0
+	add	8*3+frame_XFER(%rsp), h		# h = k + w + h         # --
+	or	c, y3		# y3 = a|c                              # MAJA
+
+	xor	T1, y1		# y1 = (a>>39) ^ (a>>34) ^ (a>>28)	# S0
+	mov	a, T1		# T1 = a                                # MAJB
+	and	b, y3		# y3 = (a|c)&b                          # MAJA
+	and	c, T1		# T1 = a&c                              # MAJB
+	add	y0, y2		# y2 = S1 + CH                          # --
+
+
+	add	h, d		# d = k + w + h + d                     # --
+	or	T1, y3		# y3 = MAJ = (a|c)&b)|(a&c)             # MAJ
+	add	y1, h		# h = k + w + h + S0                    # --
+
+	add	y2, d		# d = k + w + h + d + S1 + CH = d + t1  # --
+
+	add	y2, h		# h = k + w + h + S0 + S1 + CH = t1 + S0# --
+
+	add	y3, h		# h = t1 + S0 + MAJ                     # --
+
+	RotateState
+
+.endm
+
+########################################################################
+# void sha512_transform_rorx(const void* M, void* D, uint64_t L)#
+# Purpose: Updates the SHA512 digest stored at D with the message stored in M.
+# The size of the message pointed to by M must be an integer multiple of SHA512
+#   message blocks.
+# L is the message length in SHA512 blocks
+########################################################################
+ENTRY(sha512_transform_rorx)
+	# Allocate Stack Space
+	mov	%rsp, %rax
+	sub	$frame_size, %rsp
+	and	$~(0x20 - 1), %rsp
+	mov	%rax, frame_RSPSAVE(%rsp)
+
+	# Save GPRs
+	mov	%rbp, frame_GPRSAVE(%rsp)
+	mov	%rbx, 8*1+frame_GPRSAVE(%rsp)
+	mov	%r12, 8*2+frame_GPRSAVE(%rsp)
+	mov	%r13, 8*3+frame_GPRSAVE(%rsp)
+	mov	%r14, 8*4+frame_GPRSAVE(%rsp)
+	mov	%r15, 8*5+frame_GPRSAVE(%rsp)
+
+	shl	$7, NUM_BLKS	# convert to bytes
+	jz	done_hash
+	add	INP, NUM_BLKS	# pointer to end of data
+	mov	NUM_BLKS, frame_INPEND(%rsp)
+
+	## load initial digest
+	mov	8*0(CTX),a
+	mov	8*1(CTX),b
+	mov	8*2(CTX),c
+	mov	8*3(CTX),d
+	mov	8*4(CTX),e
+	mov	8*5(CTX),f
+	mov	8*6(CTX),g
+	mov	8*7(CTX),h
+
+	vmovdqa	PSHUFFLE_BYTE_FLIP_MASK(%rip), BYTE_FLIP_MASK
+
+loop0:
+	lea	K512(%rip), TBL
+
+	## byte swap first 16 dwords
+	COPY_YMM_AND_BSWAP	Y_0, (INP), BYTE_FLIP_MASK
+	COPY_YMM_AND_BSWAP	Y_1, 1*32(INP), BYTE_FLIP_MASK
+	COPY_YMM_AND_BSWAP	Y_2, 2*32(INP), BYTE_FLIP_MASK
+	COPY_YMM_AND_BSWAP	Y_3, 3*32(INP), BYTE_FLIP_MASK
+
+	mov	INP, frame_INP(%rsp)
+
+	## schedule 64 input dwords, by doing 12 rounds of 4 each
+	movq	$4, frame_SRND(%rsp)
+
+.align 16
+loop1:
+	vpaddq	(TBL), Y_0, XFER
+	vmovdqa XFER, frame_XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	vpaddq	1*32(TBL), Y_0, XFER
+	vmovdqa XFER, frame_XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	vpaddq	2*32(TBL), Y_0, XFER
+	vmovdqa XFER, frame_XFER(%rsp)
+	FOUR_ROUNDS_AND_SCHED
+
+	vpaddq	3*32(TBL), Y_0, XFER
+	vmovdqa XFER, frame_XFER(%rsp)
+	add	$(4*32), TBL
+	FOUR_ROUNDS_AND_SCHED
+
+	subq	$1, frame_SRND(%rsp)
+	jne	loop1
+
+	movq	$2, frame_SRND(%rsp)
+loop2:
+	vpaddq	(TBL), Y_0, XFER
+	vmovdqa XFER, frame_XFER(%rsp)
+	DO_4ROUNDS
+	vpaddq	1*32(TBL), Y_1, XFER
+	vmovdqa XFER, frame_XFER(%rsp)
+	add	$(2*32), TBL
+	DO_4ROUNDS
+
+	vmovdqa	Y_2, Y_0
+	vmovdqa	Y_3, Y_1
+
+	subq	$1, frame_SRND(%rsp)
+	jne	loop2
+
+	addm	8*0(CTX),a
+	addm	8*1(CTX),b
+	addm	8*2(CTX),c
+	addm	8*3(CTX),d
+	addm	8*4(CTX),e
+	addm	8*5(CTX),f
+	addm	8*6(CTX),g
+	addm	8*7(CTX),h
+
+	mov	frame_INP(%rsp), INP
+	add	$128, INP
+	cmp	frame_INPEND(%rsp), INP
+	jne	loop0
+
+done_hash:
+
+# Restore GPRs
+	mov	frame_GPRSAVE(%rsp)     ,%rbp
+	mov	8*1+frame_GPRSAVE(%rsp) ,%rbx
+	mov	8*2+frame_GPRSAVE(%rsp) ,%r12
+	mov	8*3+frame_GPRSAVE(%rsp) ,%r13
+	mov	8*4+frame_GPRSAVE(%rsp) ,%r14
+	mov	8*5+frame_GPRSAVE(%rsp) ,%r15
+
+	# Restore Stack Pointer
+	mov	frame_RSPSAVE(%rsp), %rsp
+	ret
+ENDPROC(sha512_transform_rorx)
+
+########################################################################
+### Binary Data
+
+.data
+
+.align 64
+# K[t] used in SHA512 hashing
+K512:
+	.quad	0x428a2f98d728ae22,0x7137449123ef65cd
+	.quad	0xb5c0fbcfec4d3b2f,0xe9b5dba58189dbbc
+	.quad	0x3956c25bf348b538,0x59f111f1b605d019
+	.quad	0x923f82a4af194f9b,0xab1c5ed5da6d8118
+	.quad	0xd807aa98a3030242,0x12835b0145706fbe
+	.quad	0x243185be4ee4b28c,0x550c7dc3d5ffb4e2
+	.quad	0x72be5d74f27b896f,0x80deb1fe3b1696b1
+	.quad	0x9bdc06a725c71235,0xc19bf174cf692694
+	.quad	0xe49b69c19ef14ad2,0xefbe4786384f25e3
+	.quad	0x0fc19dc68b8cd5b5,0x240ca1cc77ac9c65
+	.quad	0x2de92c6f592b0275,0x4a7484aa6ea6e483
+	.quad	0x5cb0a9dcbd41fbd4,0x76f988da831153b5
+	.quad	0x983e5152ee66dfab,0xa831c66d2db43210
+	.quad	0xb00327c898fb213f,0xbf597fc7beef0ee4
+	.quad	0xc6e00bf33da88fc2,0xd5a79147930aa725
+	.quad	0x06ca6351e003826f,0x142929670a0e6e70
+	.quad	0x27b70a8546d22ffc,0x2e1b21385c26c926
+	.quad	0x4d2c6dfc5ac42aed,0x53380d139d95b3df
+	.quad	0x650a73548baf63de,0x766a0abb3c77b2a8
+	.quad	0x81c2c92e47edaee6,0x92722c851482353b
+	.quad	0xa2bfe8a14cf10364,0xa81a664bbc423001
+	.quad	0xc24b8b70d0f89791,0xc76c51a30654be30
+	.quad	0xd192e819d6ef5218,0xd69906245565a910
+	.quad	0xf40e35855771202a,0x106aa07032bbd1b8
+	.quad	0x19a4c116b8d2d0c8,0x1e376c085141ab53
+	.quad	0x2748774cdf8eeb99,0x34b0bcb5e19b48a8
+	.quad	0x391c0cb3c5c95a63,0x4ed8aa4ae3418acb
+	.quad	0x5b9cca4f7763e373,0x682e6ff3d6b2b8a3
+	.quad	0x748f82ee5defb2fc,0x78a5636f43172f60
+	.quad	0x84c87814a1f0ab72,0x8cc702081a6439ec
+	.quad	0x90befffa23631e28,0xa4506cebde82bde9
+	.quad	0xbef9a3f7b2c67915,0xc67178f2e372532b
+	.quad	0xca273eceea26619c,0xd186b8c721c0c207
+	.quad	0xeada7dd6cde0eb1e,0xf57d4f7fee6ed178
+	.quad	0x06f067aa72176fba,0x0a637dc5a2c898a6
+	.quad	0x113f9804bef90dae,0x1b710b35131c471b
+	.quad	0x28db77f523047d84,0x32caab7b40c72493
+	.quad	0x3c9ebe0a15c9bebc,0x431d67c49c100d4c
+	.quad	0x4cc5d4becb3e42b6,0x597f299cfc657e2a
+	.quad	0x5fcb6fab3ad6faec,0x6c44198c4a475817
+
+.align 32
+
+# Mask for byte-swapping a couple of qwords in an XMM register using (v)pshufb.
+PSHUFFLE_BYTE_FLIP_MASK:
+	.octa 0x08090a0b0c0d0e0f0001020304050607
+	.octa 0x18191a1b1c1d1e1f1011121314151617
+
+MASK_YMM_LO:
+	.octa 0x00000000000000000000000000000000
+	.octa 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
+#endif
diff --git a/arch/x86/crypto/sha512-ssse3-asm.S b/arch/x86/crypto/sha512-ssse3-asm.S
new file mode 100644
index 000000000000..fb56855d51f5
--- /dev/null
+++ b/arch/x86/crypto/sha512-ssse3-asm.S
@@ -0,0 +1,421 @@
+########################################################################
+# Implement fast SHA-512 with SSSE3 instructions. (x86_64)
+#
+# Copyright (C) 2013 Intel Corporation.
+#
+# Authors:
+#     James Guilford <james.guilford@intel.com>
+#     Kirk Yap <kirk.s.yap@intel.com>
+#     David Cote <david.m.cote@intel.com>
+#     Tim Chen <tim.c.chen@linux.intel.com>
+#
+# This software is available to you under a choice of one of two
+# licenses.  You may choose to be licensed under the terms of the GNU
+# General Public License (GPL) Version 2, available from the file
+# COPYING in the main directory of this source tree, or the
+# OpenIB.org BSD license below:
+#
+#     Redistribution and use in source and binary forms, with or
+#     without modification, are permitted provided that the following
+#     conditions are met:
+#
+#      - Redistributions of source code must retain the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer.
+#
+#      - Redistributions in binary form must reproduce the above
+#        copyright notice, this list of conditions and the following
+#        disclaimer in the documentation and/or other materials
+#        provided with the distribution.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+# BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+# ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+#
+########################################################################
+#
+# This code is described in an Intel White-Paper:
+# "Fast SHA-512 Implementations on Intel Architecture Processors"
+#
+# To find it, surf to http://www.intel.com/p/en_US/embedded
+# and search for that title.
+#
+########################################################################
+
+#include <linux/linkage.h>
+
+.text
+
+# Virtual Registers
+# ARG1
+msg =		%rdi
+# ARG2
+digest =	%rsi
+# ARG3
+msglen =	%rdx
+T1 =		%rcx
+T2 =		%r8
+a_64 =		%r9
+b_64 =		%r10
+c_64 =		%r11
+d_64 =		%r12
+e_64 =		%r13
+f_64 =		%r14
+g_64 =		%r15
+h_64 =		%rbx
+tmp0 =		%rax
+
+# Local variables (stack frame)
+
+W_SIZE = 80*8
+WK_SIZE = 2*8
+RSPSAVE_SIZE = 1*8
+GPRSAVE_SIZE = 5*8
+
+frame_W = 0
+frame_WK = frame_W + W_SIZE
+frame_RSPSAVE = frame_WK + WK_SIZE
+frame_GPRSAVE = frame_RSPSAVE + RSPSAVE_SIZE
+frame_size = frame_GPRSAVE + GPRSAVE_SIZE
+
+# Useful QWORD "arrays" for simpler memory references
+# MSG, DIGEST, K_t, W_t are arrays
+# WK_2(t) points to 1 of 2 qwords at frame.WK depdending on t being odd/even
+
+# Input message (arg1)
+#define MSG(i)    8*i(msg)
+
+# Output Digest (arg2)
+#define DIGEST(i) 8*i(digest)
+
+# SHA Constants (static mem)
+#define K_t(i)    8*i+K512(%rip)
+
+# Message Schedule (stack frame)
+#define W_t(i)    8*i+frame_W(%rsp)
+
+# W[t]+K[t] (stack frame)
+#define WK_2(i)   8*((i%2))+frame_WK(%rsp)
+
+.macro RotateState
+	# Rotate symbols a..h right
+	TMP   = h_64
+	h_64  = g_64
+	g_64  = f_64
+	f_64  = e_64
+	e_64  = d_64
+	d_64  = c_64
+	c_64  = b_64
+	b_64  = a_64
+	a_64  = TMP
+.endm
+
+.macro SHA512_Round rnd
+
+	# Compute Round %%t
+	mov	f_64, T1          # T1 = f
+	mov	e_64, tmp0        # tmp = e
+	xor	g_64, T1          # T1 = f ^ g
+	ror	$23, tmp0 # 41    # tmp = e ror 23
+	and	e_64, T1          # T1 = (f ^ g) & e
+	xor	e_64, tmp0        # tmp = (e ror 23) ^ e
+	xor	g_64, T1          # T1 = ((f ^ g) & e) ^ g = CH(e,f,g)
+	idx = \rnd
+	add	WK_2(idx), T1     # W[t] + K[t] from message scheduler
+	ror	$4, tmp0  # 18    # tmp = ((e ror 23) ^ e) ror 4
+	xor	e_64, tmp0        # tmp = (((e ror 23) ^ e) ror 4) ^ e
+	mov	a_64, T2          # T2 = a
+	add	h_64, T1          # T1 = CH(e,f,g) + W[t] + K[t] + h
+	ror	$14, tmp0 # 14    # tmp = ((((e ror23)^e)ror4)^e)ror14 = S1(e)
+	add	tmp0, T1          # T1 = CH(e,f,g) + W[t] + K[t] + S1(e)
+	mov	a_64, tmp0        # tmp = a
+	xor	c_64, T2          # T2 = a ^ c
+	and	c_64, tmp0        # tmp = a & c
+	and	b_64, T2          # T2 = (a ^ c) & b
+	xor	tmp0, T2          # T2 = ((a ^ c) & b) ^ (a & c) = Maj(a,b,c)
+	mov	a_64, tmp0        # tmp = a
+	ror	$5, tmp0 # 39     # tmp = a ror 5
+	xor	a_64, tmp0        # tmp = (a ror 5) ^ a
+	add	T1, d_64          # e(next_state) = d + T1
+	ror	$6, tmp0 # 34     # tmp = ((a ror 5) ^ a) ror 6
+	xor	a_64, tmp0        # tmp = (((a ror 5) ^ a) ror 6) ^ a
+	lea	(T1, T2), h_64    # a(next_state) = T1 + Maj(a,b,c)
+	ror	$28, tmp0 # 28    # tmp = ((((a ror5)^a)ror6)^a)ror28 = S0(a)
+	add	tmp0, h_64        # a(next_state) = T1 + Maj(a,b,c) S0(a)
+	RotateState
+.endm
+
+.macro SHA512_2Sched_2Round_sse rnd
+
+	# Compute rounds t-2 and t-1
+	# Compute message schedule QWORDS t and t+1
+
+	#   Two rounds are computed based on the values for K[t-2]+W[t-2] and
+	# K[t-1]+W[t-1] which were previously stored at WK_2 by the message
+	# scheduler.
+	#   The two new schedule QWORDS are stored at [W_t(%%t)] and [W_t(%%t+1)].
+	# They are then added to their respective SHA512 constants at
+	# [K_t(%%t)] and [K_t(%%t+1)] and stored at dqword [WK_2(%%t)]
+	#   For brievity, the comments following vectored instructions only refer to
+	# the first of a pair of QWORDS.
+	# Eg. XMM2=W[t-2] really means XMM2={W[t-2]|W[t-1]}
+	#   The computation of the message schedule and the rounds are tightly
+	# stitched to take advantage of instruction-level parallelism.
+	# For clarity, integer instructions (for the rounds calculation) are indented
+	# by one tab. Vectored instructions (for the message scheduler) are indented
+	# by two tabs.
+
+	mov	f_64, T1
+	idx = \rnd -2
+	movdqa	W_t(idx), %xmm2		    # XMM2 = W[t-2]
+	xor	g_64, T1
+	and	e_64, T1
+	movdqa	%xmm2, %xmm0	            # XMM0 = W[t-2]
+	xor	g_64, T1
+	idx = \rnd
+	add	WK_2(idx), T1
+	idx = \rnd - 15
+	movdqu	W_t(idx), %xmm5		    # XMM5 = W[t-15]
+	mov	e_64, tmp0
+	ror	$23, tmp0 # 41
+	movdqa	%xmm5, %xmm3	            # XMM3 = W[t-15]
+	xor	e_64, tmp0
+	ror	$4, tmp0 # 18
+	psrlq	$61-19, %xmm0		    # XMM0 = W[t-2] >> 42
+	xor	e_64, tmp0
+	ror	$14, tmp0 # 14
+	psrlq	$(8-7), %xmm3		    # XMM3 = W[t-15] >> 1
+	add	tmp0, T1
+	add	h_64, T1
+	pxor	%xmm2, %xmm0                # XMM0 = (W[t-2] >> 42) ^ W[t-2]
+	mov	a_64, T2
+	xor	c_64, T2
+	pxor	%xmm5, %xmm3                # XMM3 = (W[t-15] >> 1) ^ W[t-15]
+	and	b_64, T2
+	mov	a_64, tmp0
+	psrlq	$(19-6), %xmm0		    # XMM0 = ((W[t-2]>>42)^W[t-2])>>13
+	and	c_64, tmp0
+	xor	tmp0, T2
+	psrlq	$(7-1), %xmm3		    # XMM3 = ((W[t-15]>>1)^W[t-15])>>6
+	mov	a_64, tmp0
+	ror	$5, tmp0 # 39
+	pxor	%xmm2, %xmm0	            # XMM0 = (((W[t-2]>>42)^W[t-2])>>13)^W[t-2]
+	xor	a_64, tmp0
+	ror	$6, tmp0 # 34
+	pxor	%xmm5, %xmm3                # XMM3 = (((W[t-15]>>1)^W[t-15])>>6)^W[t-15]
+	xor	a_64, tmp0
+	ror	$28, tmp0 # 28
+	psrlq	$6, %xmm0                   # XMM0 = ((((W[t-2]>>42)^W[t-2])>>13)^W[t-2])>>6
+	add	tmp0, T2
+	add	T1, d_64
+	psrlq	$1, %xmm3                   # XMM3 = (((W[t-15]>>1)^W[t-15])>>6)^W[t-15]>>1
+	lea	(T1, T2), h_64
+	RotateState
+	movdqa	%xmm2, %xmm1	            # XMM1 = W[t-2]
+	mov	f_64, T1
+	xor	g_64, T1
+	movdqa	%xmm5, %xmm4		    # XMM4 = W[t-15]
+	and	e_64, T1
+	xor	g_64, T1
+	psllq	$(64-19)-(64-61) , %xmm1    # XMM1 = W[t-2] << 42
+	idx = \rnd + 1
+	add	WK_2(idx), T1
+	mov	e_64, tmp0
+	psllq	$(64-1)-(64-8), %xmm4	    # XMM4 = W[t-15] << 7
+	ror	$23, tmp0 # 41
+	xor	e_64, tmp0
+	pxor	%xmm2, %xmm1		    # XMM1 = (W[t-2] << 42)^W[t-2]
+	ror	$4, tmp0 # 18
+	xor	e_64, tmp0
+	pxor	%xmm5, %xmm4		    # XMM4 = (W[t-15]<<7)^W[t-15]
+	ror	$14, tmp0 # 14
+	add	tmp0, T1
+	psllq	$(64-61), %xmm1		    # XMM1 = ((W[t-2] << 42)^W[t-2])<<3
+	add	h_64, T1
+	mov	a_64, T2
+	psllq	$(64-8), %xmm4		    # XMM4 = ((W[t-15]<<7)^W[t-15])<<56
+	xor	c_64, T2
+	and	b_64, T2
+	pxor	%xmm1, %xmm0		    # XMM0 = s1(W[t-2])
+	mov	a_64, tmp0
+	and	c_64, tmp0
+	idx = \rnd - 7
+	movdqu	W_t(idx), %xmm1		    # XMM1 = W[t-7]
+	xor	tmp0, T2
+	pxor	%xmm4, %xmm3                # XMM3 = s0(W[t-15])
+	mov	a_64, tmp0
+	paddq	%xmm3, %xmm0		    # XMM0 = s1(W[t-2]) + s0(W[t-15])
+	ror	$5, tmp0 # 39
+	idx =\rnd-16
+	paddq	W_t(idx), %xmm0		    # XMM0 = s1(W[t-2]) + s0(W[t-15]) + W[t-16]
+	xor	a_64, tmp0
+	paddq	%xmm1, %xmm0	            # XMM0 = s1(W[t-2]) + W[t-7] + s0(W[t-15]) + W[t-16]
+	ror	$6, tmp0 # 34
+	movdqa	%xmm0, W_t(\rnd)	    # Store scheduled qwords
+	xor	a_64, tmp0
+	paddq	K_t(\rnd), %xmm0	    # Compute W[t]+K[t]
+	ror	$28, tmp0 # 28
+	idx = \rnd
+	movdqa	%xmm0, WK_2(idx)	    # Store W[t]+K[t] for next rounds
+	add	tmp0, T2
+	add	T1, d_64
+	lea	(T1, T2), h_64
+	RotateState
+.endm
+
+########################################################################
+# void sha512_transform_ssse3(const void* M, void* D, u64 L)#
+# Purpose: Updates the SHA512 digest stored at D with the message stored in M.
+# The size of the message pointed to by M must be an integer multiple of SHA512
+#   message blocks.
+# L is the message length in SHA512 blocks.
+########################################################################
+ENTRY(sha512_transform_ssse3)
+
+	cmp $0, msglen
+	je nowork
+
+	# Allocate Stack Space
+	mov	%rsp, %rax
+	sub	$frame_size, %rsp
+	and	$~(0x20 - 1), %rsp
+	mov	%rax, frame_RSPSAVE(%rsp)
+
+	# Save GPRs
+	mov	%rbx, frame_GPRSAVE(%rsp)
+	mov	%r12, frame_GPRSAVE +8*1(%rsp)
+	mov	%r13, frame_GPRSAVE +8*2(%rsp)
+	mov	%r14, frame_GPRSAVE +8*3(%rsp)
+	mov	%r15, frame_GPRSAVE +8*4(%rsp)
+
+updateblock:
+
+# Load state variables
+	mov	DIGEST(0), a_64
+	mov	DIGEST(1), b_64
+	mov	DIGEST(2), c_64
+	mov	DIGEST(3), d_64
+	mov	DIGEST(4), e_64
+	mov	DIGEST(5), f_64
+	mov	DIGEST(6), g_64
+	mov	DIGEST(7), h_64
+
+	t = 0
+	.rept 80/2 + 1
+	# (80 rounds) / (2 rounds/iteration) + (1 iteration)
+	# +1 iteration because the scheduler leads hashing by 1 iteration
+		.if t < 2
+			# BSWAP 2 QWORDS
+			movdqa	XMM_QWORD_BSWAP(%rip), %xmm1
+			movdqu	MSG(t), %xmm0
+			pshufb	%xmm1, %xmm0	# BSWAP
+			movdqa	%xmm0, W_t(t)	# Store Scheduled Pair
+			paddq	K_t(t), %xmm0	# Compute W[t]+K[t]
+			movdqa	%xmm0, WK_2(t)	# Store into WK for rounds
+		.elseif t < 16
+			# BSWAP 2 QWORDS# Compute 2 Rounds
+			movdqu	MSG(t), %xmm0
+			pshufb	%xmm1, %xmm0	# BSWAP
+			SHA512_Round t-2	# Round t-2
+			movdqa	%xmm0, W_t(t)	# Store Scheduled Pair
+			paddq	K_t(t), %xmm0	# Compute W[t]+K[t]
+			SHA512_Round t-1	# Round t-1
+			movdqa	%xmm0, WK_2(t)	# Store W[t]+K[t] into WK
+		.elseif t < 79
+			# Schedule 2 QWORDS# Compute 2 Rounds
+			SHA512_2Sched_2Round_sse t
+		.else
+			# Compute 2 Rounds
+			SHA512_Round t-2
+			SHA512_Round t-1
+		.endif
+		t = t+2
+	.endr
+
+	# Update digest
+	add	a_64, DIGEST(0)
+	add	b_64, DIGEST(1)
+	add	c_64, DIGEST(2)
+	add	d_64, DIGEST(3)
+	add	e_64, DIGEST(4)
+	add	f_64, DIGEST(5)
+	add	g_64, DIGEST(6)
+	add	h_64, DIGEST(7)
+
+	# Advance to next message block
+	add	$16*8, msg
+	dec	msglen
+	jnz	updateblock
+
+	# Restore GPRs
+	mov	frame_GPRSAVE(%rsp),      %rbx
+	mov	frame_GPRSAVE +8*1(%rsp), %r12
+	mov	frame_GPRSAVE +8*2(%rsp), %r13
+	mov	frame_GPRSAVE +8*3(%rsp), %r14
+	mov	frame_GPRSAVE +8*4(%rsp), %r15
+
+	# Restore Stack Pointer
+	mov	frame_RSPSAVE(%rsp), %rsp
+
+nowork:
+	ret
+ENDPROC(sha512_transform_ssse3)
+
+########################################################################
+### Binary Data
+
+.data
+
+.align 16
+
+# Mask for byte-swapping a couple of qwords in an XMM register using (v)pshufb.
+XMM_QWORD_BSWAP:
+	.octa 0x08090a0b0c0d0e0f0001020304050607
+
+# K[t] used in SHA512 hashing
+K512:
+	.quad 0x428a2f98d728ae22,0x7137449123ef65cd
+	.quad 0xb5c0fbcfec4d3b2f,0xe9b5dba58189dbbc
+	.quad 0x3956c25bf348b538,0x59f111f1b605d019
+	.quad 0x923f82a4af194f9b,0xab1c5ed5da6d8118
+	.quad 0xd807aa98a3030242,0x12835b0145706fbe
+	.quad 0x243185be4ee4b28c,0x550c7dc3d5ffb4e2
+	.quad 0x72be5d74f27b896f,0x80deb1fe3b1696b1
+	.quad 0x9bdc06a725c71235,0xc19bf174cf692694
+	.quad 0xe49b69c19ef14ad2,0xefbe4786384f25e3
+	.quad 0x0fc19dc68b8cd5b5,0x240ca1cc77ac9c65
+	.quad 0x2de92c6f592b0275,0x4a7484aa6ea6e483
+	.quad 0x5cb0a9dcbd41fbd4,0x76f988da831153b5
+	.quad 0x983e5152ee66dfab,0xa831c66d2db43210
+	.quad 0xb00327c898fb213f,0xbf597fc7beef0ee4
+	.quad 0xc6e00bf33da88fc2,0xd5a79147930aa725
+	.quad 0x06ca6351e003826f,0x142929670a0e6e70
+	.quad 0x27b70a8546d22ffc,0x2e1b21385c26c926
+	.quad 0x4d2c6dfc5ac42aed,0x53380d139d95b3df
+	.quad 0x650a73548baf63de,0x766a0abb3c77b2a8
+	.quad 0x81c2c92e47edaee6,0x92722c851482353b
+	.quad 0xa2bfe8a14cf10364,0xa81a664bbc423001
+	.quad 0xc24b8b70d0f89791,0xc76c51a30654be30
+	.quad 0xd192e819d6ef5218,0xd69906245565a910
+	.quad 0xf40e35855771202a,0x106aa07032bbd1b8
+	.quad 0x19a4c116b8d2d0c8,0x1e376c085141ab53
+	.quad 0x2748774cdf8eeb99,0x34b0bcb5e19b48a8
+	.quad 0x391c0cb3c5c95a63,0x4ed8aa4ae3418acb
+	.quad 0x5b9cca4f7763e373,0x682e6ff3d6b2b8a3
+	.quad 0x748f82ee5defb2fc,0x78a5636f43172f60
+	.quad 0x84c87814a1f0ab72,0x8cc702081a6439ec
+	.quad 0x90befffa23631e28,0xa4506cebde82bde9
+	.quad 0xbef9a3f7b2c67915,0xc67178f2e372532b
+	.quad 0xca273eceea26619c,0xd186b8c721c0c207
+	.quad 0xeada7dd6cde0eb1e,0xf57d4f7fee6ed178
+	.quad 0x06f067aa72176fba,0x0a637dc5a2c898a6
+	.quad 0x113f9804bef90dae,0x1b710b35131c471b
+	.quad 0x28db77f523047d84,0x32caab7b40c72493
+	.quad 0x3c9ebe0a15c9bebc,0x431d67c49c100d4c
+	.quad 0x4cc5d4becb3e42b6,0x597f299cfc657e2a
+	.quad 0x5fcb6fab3ad6faec,0x6c44198c4a475817
diff --git a/arch/x86/crypto/sha512_ssse3_glue.c b/arch/x86/crypto/sha512_ssse3_glue.c
new file mode 100644
index 000000000000..6cbd8df348d2
--- /dev/null
+++ b/arch/x86/crypto/sha512_ssse3_glue.c
@@ -0,0 +1,282 @@
+/*
+ * Cryptographic API.
+ *
+ * Glue code for the SHA512 Secure Hash Algorithm assembler
+ * implementation using supplemental SSE3 / AVX / AVX2 instructions.
+ *
+ * This file is based on sha512_generic.c
+ *
+ * Copyright (C) 2013 Intel Corporation
+ * Author: Tim Chen <tim.c.chen@linux.intel.com>
+ *
+ * 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.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#define pr_fmt(fmt)	KBUILD_MODNAME ": " fmt
+
+#include <crypto/internal/hash.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/cryptohash.h>
+#include <linux/types.h>
+#include <crypto/sha.h>
+#include <asm/byteorder.h>
+#include <asm/i387.h>
+#include <asm/xcr.h>
+#include <asm/xsave.h>
+
+#include <linux/string.h>
+
+asmlinkage void sha512_transform_ssse3(const char *data, u64 *digest,
+				     u64 rounds);
+#ifdef CONFIG_AS_AVX
+asmlinkage void sha512_transform_avx(const char *data, u64 *digest,
+				     u64 rounds);
+#endif
+#ifdef CONFIG_AS_AVX2
+asmlinkage void sha512_transform_rorx(const char *data, u64 *digest,
+				     u64 rounds);
+#endif
+
+static asmlinkage void (*sha512_transform_asm)(const char *, u64 *, u64);
+
+
+static int sha512_ssse3_init(struct shash_desc *desc)
+{
+	struct sha512_state *sctx = shash_desc_ctx(desc);
+
+	sctx->state[0] = SHA512_H0;
+	sctx->state[1] = SHA512_H1;
+	sctx->state[2] = SHA512_H2;
+	sctx->state[3] = SHA512_H3;
+	sctx->state[4] = SHA512_H4;
+	sctx->state[5] = SHA512_H5;
+	sctx->state[6] = SHA512_H6;
+	sctx->state[7] = SHA512_H7;
+	sctx->count[0] = sctx->count[1] = 0;
+
+	return 0;
+}
+
+static int __sha512_ssse3_update(struct shash_desc *desc, const u8 *data,
+			       unsigned int len, unsigned int partial)
+{
+	struct sha512_state *sctx = shash_desc_ctx(desc);
+	unsigned int done = 0;
+
+	sctx->count[0] += len;
+	if (sctx->count[0] < len)
+		sctx->count[1]++;
+
+	if (partial) {
+		done = SHA512_BLOCK_SIZE - partial;
+		memcpy(sctx->buf + partial, data, done);
+		sha512_transform_asm(sctx->buf, sctx->state, 1);
+	}
+
+	if (len - done >= SHA512_BLOCK_SIZE) {
+		const unsigned int rounds = (len - done) / SHA512_BLOCK_SIZE;
+
+		sha512_transform_asm(data + done, sctx->state, (u64) rounds);
+
+		done += rounds * SHA512_BLOCK_SIZE;
+	}
+
+	memcpy(sctx->buf, data + done, len - done);
+
+	return 0;
+}
+
+static int sha512_ssse3_update(struct shash_desc *desc, const u8 *data,
+			     unsigned int len)
+{
+	struct sha512_state *sctx = shash_desc_ctx(desc);
+	unsigned int partial = sctx->count[0] % SHA512_BLOCK_SIZE;
+	int res;
+
+	/* Handle the fast case right here */
+	if (partial + len < SHA512_BLOCK_SIZE) {
+		sctx->count[0] += len;
+		if (sctx->count[0] < len)
+			sctx->count[1]++;
+		memcpy(sctx->buf + partial, data, len);
+
+		return 0;
+	}
+
+	if (!irq_fpu_usable()) {
+		res = crypto_sha512_update(desc, data, len);
+	} else {
+		kernel_fpu_begin();
+		res = __sha512_ssse3_update(desc, data, len, partial);
+		kernel_fpu_end();
+	}
+
+	return res;
+}
+
+
+/* Add padding and return the message digest. */
+static int sha512_ssse3_final(struct shash_desc *desc, u8 *out)
+{
+	struct sha512_state *sctx = shash_desc_ctx(desc);
+	unsigned int i, index, padlen;
+	__be64 *dst = (__be64 *)out;
+	__be64 bits[2];
+	static const u8 padding[SHA512_BLOCK_SIZE] = { 0x80, };
+
+	/* save number of bits */
+	bits[1] = cpu_to_be64(sctx->count[0] << 3);
+	bits[0] = cpu_to_be64(sctx->count[1] << 3) | sctx->count[0] >> 61;
+
+	/* Pad out to 112 mod 128 and append length */
+	index = sctx->count[0] & 0x7f;
+	padlen = (index < 112) ? (112 - index) : ((128+112) - index);
+
+	if (!irq_fpu_usable()) {
+		crypto_sha512_update(desc, padding, padlen);
+		crypto_sha512_update(desc, (const u8 *)&bits, sizeof(bits));
+	} else {
+		kernel_fpu_begin();
+		/* We need to fill a whole block for __sha512_ssse3_update() */
+		if (padlen <= 112) {
+			sctx->count[0] += padlen;
+			if (sctx->count[0] < padlen)
+				sctx->count[1]++;
+			memcpy(sctx->buf + index, padding, padlen);
+		} else {
+			__sha512_ssse3_update(desc, padding, padlen, index);
+		}
+		__sha512_ssse3_update(desc, (const u8 *)&bits,
+					sizeof(bits), 112);
+		kernel_fpu_end();
+	}
+
+	/* Store state in digest */
+	for (i = 0; i < 8; i++)
+		dst[i] = cpu_to_be64(sctx->state[i]);
+
+	/* Wipe context */
+	memset(sctx, 0, sizeof(*sctx));
+
+	return 0;
+}
+
+static int sha512_ssse3_export(struct shash_desc *desc, void *out)
+{
+	struct sha512_state *sctx = shash_desc_ctx(desc);
+
+	memcpy(out, sctx, sizeof(*sctx));
+
+	return 0;
+}
+
+static int sha512_ssse3_import(struct shash_desc *desc, const void *in)
+{
+	struct sha512_state *sctx = shash_desc_ctx(desc);
+
+	memcpy(sctx, in, sizeof(*sctx));
+
+	return 0;
+}
+
+static struct shash_alg alg = {
+	.digestsize	=	SHA512_DIGEST_SIZE,
+	.init		=	sha512_ssse3_init,
+	.update		=	sha512_ssse3_update,
+	.final		=	sha512_ssse3_final,
+	.export		=	sha512_ssse3_export,
+	.import		=	sha512_ssse3_import,
+	.descsize	=	sizeof(struct sha512_state),
+	.statesize	=	sizeof(struct sha512_state),
+	.base		=	{
+		.cra_name	=	"sha512",
+		.cra_driver_name =	"sha512-ssse3",
+		.cra_priority	=	150,
+		.cra_flags	=	CRYPTO_ALG_TYPE_SHASH,
+		.cra_blocksize	=	SHA512_BLOCK_SIZE,
+		.cra_module	=	THIS_MODULE,
+	}
+};
+
+#ifdef CONFIG_AS_AVX
+static bool __init avx_usable(void)
+{
+	u64 xcr0;
+
+	if (!cpu_has_avx || !cpu_has_osxsave)
+		return false;
+
+	xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
+	if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
+		pr_info("AVX detected but unusable.\n");
+
+		return false;
+	}
+
+	return true;
+}
+#endif
+
+static int __init sha512_ssse3_mod_init(void)
+{
+	/* test for SSE3 first */
+	if (cpu_has_ssse3)
+		sha512_transform_asm = sha512_transform_ssse3;
+
+#ifdef CONFIG_AS_AVX
+	/* allow AVX to override SSSE3, it's a little faster */
+	if (avx_usable()) {
+#ifdef CONFIG_AS_AVX2
+		if (boot_cpu_has(X86_FEATURE_AVX2))
+			sha512_transform_asm = sha512_transform_rorx;
+		else
+#endif
+			sha512_transform_asm = sha512_transform_avx;
+	}
+#endif
+
+	if (sha512_transform_asm) {
+#ifdef CONFIG_AS_AVX
+		if (sha512_transform_asm == sha512_transform_avx)
+			pr_info("Using AVX optimized SHA-512 implementation\n");
+#ifdef CONFIG_AS_AVX2
+		else if (sha512_transform_asm == sha512_transform_rorx)
+			pr_info("Using AVX2 optimized SHA-512 implementation\n");
+#endif
+		else
+#endif
+			pr_info("Using SSSE3 optimized SHA-512 implementation\n");
+		return crypto_register_shash(&alg);
+	}
+	pr_info("Neither AVX nor SSSE3 is available/usable.\n");
+
+	return -ENODEV;
+}
+
+static void __exit sha512_ssse3_mod_fini(void)
+{
+	crypto_unregister_shash(&alg);
+}
+
+module_init(sha512_ssse3_mod_init);
+module_exit(sha512_ssse3_mod_fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("SHA512 Secure Hash Algorithm, Supplemental SSE3 accelerated");
+
+MODULE_ALIAS("sha512");
diff --git a/arch/x86/crypto/twofish-avx-x86_64-asm_64.S b/arch/x86/crypto/twofish-avx-x86_64-asm_64.S
index 8d3e113b2c95..05058134c443 100644
--- a/arch/x86/crypto/twofish-avx-x86_64-asm_64.S
+++ b/arch/x86/crypto/twofish-avx-x86_64-asm_64.S
@@ -4,7 +4,7 @@
  * Copyright (C) 2012 Johannes Goetzfried
  *     <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
  *
- * Copyright © 2012 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
  *
  * 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
@@ -33,6 +33,8 @@
 
 .Lbswap128_mask:
 	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
+.Lxts_gf128mul_and_shl1_mask:
+	.byte 0x87, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
 
 .text
 
@@ -408,3 +410,47 @@ ENTRY(twofish_ctr_8way)
 
 	ret;
 ENDPROC(twofish_ctr_8way)
+
+ENTRY(twofish_xts_enc_8way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	movq %rsi, %r11;
+
+	/* regs <= src, dst <= IVs, regs <= regs xor IVs */
+	load_xts_8way(%rcx, %rdx, %rsi, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2,
+		      RX0, RX1, RY0, .Lxts_gf128mul_and_shl1_mask);
+
+	call __twofish_enc_blk8;
+
+	/* dst <= regs xor IVs(in dst) */
+	store_xts_8way(%r11, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2);
+
+	ret;
+ENDPROC(twofish_xts_enc_8way)
+
+ENTRY(twofish_xts_dec_8way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+
+	movq %rsi, %r11;
+
+	/* regs <= src, dst <= IVs, regs <= regs xor IVs */
+	load_xts_8way(%rcx, %rdx, %rsi, RC1, RD1, RA1, RB1, RC2, RD2, RA2, RB2,
+		      RX0, RX1, RY0, .Lxts_gf128mul_and_shl1_mask);
+
+	call __twofish_dec_blk8;
+
+	/* dst <= regs xor IVs(in dst) */
+	store_xts_8way(%r11, RA1, RB1, RC1, RD1, RA2, RB2, RC2, RD2);
+
+	ret;
+ENDPROC(twofish_xts_dec_8way)
diff --git a/arch/x86/crypto/twofish-avx2-asm_64.S b/arch/x86/crypto/twofish-avx2-asm_64.S
new file mode 100644
index 000000000000..e1a83b9cd389
--- /dev/null
+++ b/arch/x86/crypto/twofish-avx2-asm_64.S
@@ -0,0 +1,600 @@
+/*
+ * x86_64/AVX2 assembler optimized version of Twofish
+ *
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
+ *
+ * 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.
+ *
+ */
+
+#include <linux/linkage.h>
+#include "glue_helper-asm-avx2.S"
+
+.file "twofish-avx2-asm_64.S"
+
+.data
+.align 16
+
+.Lvpshufb_mask0:
+.long 0x80808000
+.long 0x80808004
+.long 0x80808008
+.long 0x8080800c
+
+.Lbswap128_mask:
+	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
+.Lxts_gf128mul_and_shl1_mask_0:
+	.byte 0x87, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0
+.Lxts_gf128mul_and_shl1_mask_1:
+	.byte 0x0e, 1, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0
+
+.text
+
+/* structure of crypto context */
+#define s0	0
+#define s1	1024
+#define s2	2048
+#define s3	3072
+#define w	4096
+#define	k	4128
+
+/* register macros */
+#define CTX	%rdi
+
+#define RS0	CTX
+#define RS1	%r8
+#define RS2	%r9
+#define RS3	%r10
+#define RK	%r11
+#define RW	%rax
+#define RROUND  %r12
+#define RROUNDd %r12d
+
+#define RA0	%ymm8
+#define RB0	%ymm9
+#define RC0	%ymm10
+#define RD0	%ymm11
+#define RA1	%ymm12
+#define RB1	%ymm13
+#define RC1	%ymm14
+#define RD1	%ymm15
+
+/* temp regs */
+#define RX0	%ymm0
+#define RY0	%ymm1
+#define RX1	%ymm2
+#define RY1	%ymm3
+#define RT0	%ymm4
+#define RIDX	%ymm5
+
+#define RX0x	%xmm0
+#define RY0x	%xmm1
+#define RX1x	%xmm2
+#define RY1x	%xmm3
+#define RT0x	%xmm4
+
+/* vpgatherdd mask and '-1' */
+#define RNOT	%ymm6
+
+/* byte mask, (-1 >> 24) */
+#define RBYTE	%ymm7
+
+/**********************************************************************
+  16-way AVX2 twofish
+ **********************************************************************/
+#define init_round_constants() \
+	vpcmpeqd RNOT, RNOT, RNOT; \
+	vpsrld $24, RNOT, RBYTE; \
+	leaq k(CTX), RK; \
+	leaq w(CTX), RW; \
+	leaq s1(CTX), RS1; \
+	leaq s2(CTX), RS2; \
+	leaq s3(CTX), RS3; \
+
+#define g16(ab, rs0, rs1, rs2, rs3, xy) \
+	vpand RBYTE, ab ## 0, RIDX; \
+	vpgatherdd RNOT, (rs0, RIDX, 4), xy ## 0; \
+	vpcmpeqd RNOT, RNOT, RNOT; \
+		\
+		vpand RBYTE, ab ## 1, RIDX; \
+		vpgatherdd RNOT, (rs0, RIDX, 4), xy ## 1; \
+		vpcmpeqd RNOT, RNOT, RNOT; \
+	\
+	vpsrld $8, ab ## 0, RIDX; \
+	vpand RBYTE, RIDX, RIDX; \
+	vpgatherdd RNOT, (rs1, RIDX, 4), RT0; \
+	vpcmpeqd RNOT, RNOT, RNOT; \
+	vpxor RT0, xy ## 0, xy ## 0; \
+		\
+		vpsrld $8, ab ## 1, RIDX; \
+		vpand RBYTE, RIDX, RIDX; \
+		vpgatherdd RNOT, (rs1, RIDX, 4), RT0; \
+		vpcmpeqd RNOT, RNOT, RNOT; \
+		vpxor RT0, xy ## 1, xy ## 1; \
+	\
+	vpsrld $16, ab ## 0, RIDX; \
+	vpand RBYTE, RIDX, RIDX; \
+	vpgatherdd RNOT, (rs2, RIDX, 4), RT0; \
+	vpcmpeqd RNOT, RNOT, RNOT; \
+	vpxor RT0, xy ## 0, xy ## 0; \
+		\
+		vpsrld $16, ab ## 1, RIDX; \
+		vpand RBYTE, RIDX, RIDX; \
+		vpgatherdd RNOT, (rs2, RIDX, 4), RT0; \
+		vpcmpeqd RNOT, RNOT, RNOT; \
+		vpxor RT0, xy ## 1, xy ## 1; \
+	\
+	vpsrld $24, ab ## 0, RIDX; \
+	vpgatherdd RNOT, (rs3, RIDX, 4), RT0; \
+	vpcmpeqd RNOT, RNOT, RNOT; \
+	vpxor RT0, xy ## 0, xy ## 0; \
+		\
+		vpsrld $24, ab ## 1, RIDX; \
+		vpgatherdd RNOT, (rs3, RIDX, 4), RT0; \
+		vpcmpeqd RNOT, RNOT, RNOT; \
+		vpxor RT0, xy ## 1, xy ## 1;
+
+#define g1_16(a, x) \
+	g16(a, RS0, RS1, RS2, RS3, x);
+
+#define g2_16(b, y) \
+	g16(b, RS1, RS2, RS3, RS0, y);
+
+#define encrypt_round_end16(a, b, c, d, nk) \
+	vpaddd RY0, RX0, RX0; \
+	vpaddd RX0, RY0, RY0; \
+	vpbroadcastd nk(RK,RROUND,8), RT0; \
+	vpaddd RT0, RX0, RX0; \
+	vpbroadcastd 4+nk(RK,RROUND,8), RT0; \
+	vpaddd RT0, RY0, RY0; \
+	\
+	vpxor RY0, d ## 0, d ## 0; \
+	\
+	vpxor RX0, c ## 0, c ## 0; \
+	vpsrld $1, c ## 0, RT0; \
+	vpslld $31, c ## 0, c ## 0; \
+	vpor RT0, c ## 0, c ## 0; \
+	\
+		vpaddd RY1, RX1, RX1; \
+		vpaddd RX1, RY1, RY1; \
+		vpbroadcastd nk(RK,RROUND,8), RT0; \
+		vpaddd RT0, RX1, RX1; \
+		vpbroadcastd 4+nk(RK,RROUND,8), RT0; \
+		vpaddd RT0, RY1, RY1; \
+		\
+		vpxor RY1, d ## 1, d ## 1; \
+		\
+		vpxor RX1, c ## 1, c ## 1; \
+		vpsrld $1, c ## 1, RT0; \
+		vpslld $31, c ## 1, c ## 1; \
+		vpor RT0, c ## 1, c ## 1; \
+
+#define encrypt_round16(a, b, c, d, nk) \
+	g2_16(b, RY); \
+	\
+	vpslld $1, b ## 0, RT0; \
+	vpsrld $31, b ## 0, b ## 0; \
+	vpor RT0, b ## 0, b ## 0; \
+	\
+		vpslld $1, b ## 1, RT0; \
+		vpsrld $31, b ## 1, b ## 1; \
+		vpor RT0, b ## 1, b ## 1; \
+	\
+	g1_16(a, RX); \
+	\
+	encrypt_round_end16(a, b, c, d, nk);
+
+#define encrypt_round_first16(a, b, c, d, nk) \
+	vpslld $1, d ## 0, RT0; \
+	vpsrld $31, d ## 0, d ## 0; \
+	vpor RT0, d ## 0, d ## 0; \
+	\
+		vpslld $1, d ## 1, RT0; \
+		vpsrld $31, d ## 1, d ## 1; \
+		vpor RT0, d ## 1, d ## 1; \
+	\
+	encrypt_round16(a, b, c, d, nk);
+
+#define encrypt_round_last16(a, b, c, d, nk) \
+	g2_16(b, RY); \
+	\
+	g1_16(a, RX); \
+	\
+	encrypt_round_end16(a, b, c, d, nk);
+
+#define decrypt_round_end16(a, b, c, d, nk) \
+	vpaddd RY0, RX0, RX0; \
+	vpaddd RX0, RY0, RY0; \
+	vpbroadcastd nk(RK,RROUND,8), RT0; \
+	vpaddd RT0, RX0, RX0; \
+	vpbroadcastd 4+nk(RK,RROUND,8), RT0; \
+	vpaddd RT0, RY0, RY0; \
+	\
+	vpxor RX0, c ## 0, c ## 0; \
+	\
+	vpxor RY0, d ## 0, d ## 0; \
+	vpsrld $1, d ## 0, RT0; \
+	vpslld $31, d ## 0, d ## 0; \
+	vpor RT0, d ## 0, d ## 0; \
+	\
+		vpaddd RY1, RX1, RX1; \
+		vpaddd RX1, RY1, RY1; \
+		vpbroadcastd nk(RK,RROUND,8), RT0; \
+		vpaddd RT0, RX1, RX1; \
+		vpbroadcastd 4+nk(RK,RROUND,8), RT0; \
+		vpaddd RT0, RY1, RY1; \
+		\
+		vpxor RX1, c ## 1, c ## 1; \
+		\
+		vpxor RY1, d ## 1, d ## 1; \
+		vpsrld $1, d ## 1, RT0; \
+		vpslld $31, d ## 1, d ## 1; \
+		vpor RT0, d ## 1, d ## 1;
+
+#define decrypt_round16(a, b, c, d, nk) \
+	g1_16(a, RX); \
+	\
+	vpslld $1, a ## 0, RT0; \
+	vpsrld $31, a ## 0, a ## 0; \
+	vpor RT0, a ## 0, a ## 0; \
+	\
+		vpslld $1, a ## 1, RT0; \
+		vpsrld $31, a ## 1, a ## 1; \
+		vpor RT0, a ## 1, a ## 1; \
+	\
+	g2_16(b, RY); \
+	\
+	decrypt_round_end16(a, b, c, d, nk);
+
+#define decrypt_round_first16(a, b, c, d, nk) \
+	vpslld $1, c ## 0, RT0; \
+	vpsrld $31, c ## 0, c ## 0; \
+	vpor RT0, c ## 0, c ## 0; \
+	\
+		vpslld $1, c ## 1, RT0; \
+		vpsrld $31, c ## 1, c ## 1; \
+		vpor RT0, c ## 1, c ## 1; \
+	\
+	decrypt_round16(a, b, c, d, nk)
+
+#define decrypt_round_last16(a, b, c, d, nk) \
+	g1_16(a, RX); \
+	\
+	g2_16(b, RY); \
+	\
+	decrypt_round_end16(a, b, c, d, nk);
+
+#define encrypt_cycle16() \
+	encrypt_round16(RA, RB, RC, RD, 0); \
+	encrypt_round16(RC, RD, RA, RB, 8);
+
+#define encrypt_cycle_first16() \
+	encrypt_round_first16(RA, RB, RC, RD, 0); \
+	encrypt_round16(RC, RD, RA, RB, 8);
+
+#define encrypt_cycle_last16() \
+	encrypt_round16(RA, RB, RC, RD, 0); \
+	encrypt_round_last16(RC, RD, RA, RB, 8);
+
+#define decrypt_cycle16(n) \
+	decrypt_round16(RC, RD, RA, RB, 8); \
+	decrypt_round16(RA, RB, RC, RD, 0);
+
+#define decrypt_cycle_first16(n) \
+	decrypt_round_first16(RC, RD, RA, RB, 8); \
+	decrypt_round16(RA, RB, RC, RD, 0);
+
+#define decrypt_cycle_last16(n) \
+	decrypt_round16(RC, RD, RA, RB, 8); \
+	decrypt_round_last16(RA, RB, RC, RD, 0);
+
+#define transpose_4x4(x0,x1,x2,x3,t1,t2) \
+	vpunpckhdq x1, x0, t2; \
+	vpunpckldq x1, x0, x0; \
+	\
+	vpunpckldq x3, x2, t1; \
+	vpunpckhdq x3, x2, x2; \
+	\
+	vpunpckhqdq t1,	x0, x1; \
+	vpunpcklqdq t1,	x0, x0; \
+	\
+	vpunpckhqdq x2, t2, x3; \
+	vpunpcklqdq x2,	t2, x2;
+
+#define read_blocks8(offs,a,b,c,d) \
+	transpose_4x4(a, b, c, d, RX0, RY0);
+
+#define write_blocks8(offs,a,b,c,d) \
+	transpose_4x4(a, b, c, d, RX0, RY0);
+
+#define inpack_enc8(a,b,c,d) \
+	vpbroadcastd 4*0(RW), RT0; \
+	vpxor RT0, a, a; \
+	\
+	vpbroadcastd 4*1(RW), RT0; \
+	vpxor RT0, b, b; \
+	\
+	vpbroadcastd 4*2(RW), RT0; \
+	vpxor RT0, c, c; \
+	\
+	vpbroadcastd 4*3(RW), RT0; \
+	vpxor RT0, d, d;
+
+#define outunpack_enc8(a,b,c,d) \
+	vpbroadcastd 4*4(RW), RX0; \
+	vpbroadcastd 4*5(RW), RY0; \
+	vpxor RX0, c, RX0; \
+	vpxor RY0, d, RY0; \
+	\
+	vpbroadcastd 4*6(RW), RT0; \
+	vpxor RT0, a, c; \
+	vpbroadcastd 4*7(RW), RT0; \
+	vpxor RT0, b, d; \
+	\
+	vmovdqa RX0, a; \
+	vmovdqa RY0, b;
+
+#define inpack_dec8(a,b,c,d) \
+	vpbroadcastd 4*4(RW), RX0; \
+	vpbroadcastd 4*5(RW), RY0; \
+	vpxor RX0, a, RX0; \
+	vpxor RY0, b, RY0; \
+	\
+	vpbroadcastd 4*6(RW), RT0; \
+	vpxor RT0, c, a; \
+	vpbroadcastd 4*7(RW), RT0; \
+	vpxor RT0, d, b; \
+	\
+	vmovdqa RX0, c; \
+	vmovdqa RY0, d;
+
+#define outunpack_dec8(a,b,c,d) \
+	vpbroadcastd 4*0(RW), RT0; \
+	vpxor RT0, a, a; \
+	\
+	vpbroadcastd 4*1(RW), RT0; \
+	vpxor RT0, b, b; \
+	\
+	vpbroadcastd 4*2(RW), RT0; \
+	vpxor RT0, c, c; \
+	\
+	vpbroadcastd 4*3(RW), RT0; \
+	vpxor RT0, d, d;
+
+#define read_blocks16(a,b,c,d) \
+	read_blocks8(0, a ## 0, b ## 0, c ## 0, d ## 0); \
+	read_blocks8(8, a ## 1, b ## 1, c ## 1, d ## 1);
+
+#define write_blocks16(a,b,c,d) \
+	write_blocks8(0, a ## 0, b ## 0, c ## 0, d ## 0); \
+	write_blocks8(8, a ## 1, b ## 1, c ## 1, d ## 1);
+
+#define xor_blocks16(a,b,c,d) \
+	xor_blocks8(0, a ## 0, b ## 0, c ## 0, d ## 0); \
+	xor_blocks8(8, a ## 1, b ## 1, c ## 1, d ## 1);
+
+#define inpack_enc16(a,b,c,d) \
+	inpack_enc8(a ## 0, b ## 0, c ## 0, d ## 0); \
+	inpack_enc8(a ## 1, b ## 1, c ## 1, d ## 1);
+
+#define outunpack_enc16(a,b,c,d) \
+	outunpack_enc8(a ## 0, b ## 0, c ## 0, d ## 0); \
+	outunpack_enc8(a ## 1, b ## 1, c ## 1, d ## 1);
+
+#define inpack_dec16(a,b,c,d) \
+	inpack_dec8(a ## 0, b ## 0, c ## 0, d ## 0); \
+	inpack_dec8(a ## 1, b ## 1, c ## 1, d ## 1);
+
+#define outunpack_dec16(a,b,c,d) \
+	outunpack_dec8(a ## 0, b ## 0, c ## 0, d ## 0); \
+	outunpack_dec8(a ## 1, b ## 1, c ## 1, d ## 1);
+
+.align 8
+__twofish_enc_blk16:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1: plaintext
+	 * output:
+	 *	RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1: ciphertext
+	 */
+	init_round_constants();
+
+	read_blocks16(RA, RB, RC, RD);
+	inpack_enc16(RA, RB, RC, RD);
+
+	xorl RROUNDd, RROUNDd;
+	encrypt_cycle_first16();
+	movl $2, RROUNDd;
+
+.align 4
+.L__enc_loop:
+	encrypt_cycle16();
+
+	addl $2, RROUNDd;
+	cmpl $14, RROUNDd;
+	jne .L__enc_loop;
+
+	encrypt_cycle_last16();
+
+	outunpack_enc16(RA, RB, RC, RD);
+	write_blocks16(RA, RB, RC, RD);
+
+	ret;
+ENDPROC(__twofish_enc_blk16)
+
+.align 8
+__twofish_dec_blk16:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1: ciphertext
+	 * output:
+	 *	RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1: plaintext
+	 */
+	init_round_constants();
+
+	read_blocks16(RA, RB, RC, RD);
+	inpack_dec16(RA, RB, RC, RD);
+
+	movl $14, RROUNDd;
+	decrypt_cycle_first16();
+	movl $12, RROUNDd;
+
+.align 4
+.L__dec_loop:
+	decrypt_cycle16();
+
+	addl $-2, RROUNDd;
+	jnz .L__dec_loop;
+
+	decrypt_cycle_last16();
+
+	outunpack_dec16(RA, RB, RC, RD);
+	write_blocks16(RA, RB, RC, RD);
+
+	ret;
+ENDPROC(__twofish_dec_blk16)
+
+ENTRY(twofish_ecb_enc_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+	pushq %r12;
+
+	load_16way(%rdx, RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1);
+
+	call __twofish_enc_blk16;
+
+	store_16way(%rsi, RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1);
+
+	popq %r12;
+	vzeroupper;
+
+	ret;
+ENDPROC(twofish_ecb_enc_16way)
+
+ENTRY(twofish_ecb_dec_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+	pushq %r12;
+
+	load_16way(%rdx, RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1);
+
+	call __twofish_dec_blk16;
+
+	store_16way(%rsi, RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1);
+
+	popq %r12;
+	vzeroupper;
+
+	ret;
+ENDPROC(twofish_ecb_dec_16way)
+
+ENTRY(twofish_cbc_dec_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst
+	 *	%rdx: src
+	 */
+
+	vzeroupper;
+	pushq %r12;
+
+	load_16way(%rdx, RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1);
+
+	call __twofish_dec_blk16;
+
+	store_cbc_16way(%rdx, %rsi, RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1,
+			RX0);
+
+	popq %r12;
+	vzeroupper;
+
+	ret;
+ENDPROC(twofish_cbc_dec_16way)
+
+ENTRY(twofish_ctr_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (little endian, 128bit)
+	 */
+
+	vzeroupper;
+	pushq %r12;
+
+	load_ctr_16way(%rcx, .Lbswap128_mask, RA0, RB0, RC0, RD0, RA1, RB1, RC1,
+		       RD1, RX0, RX0x, RX1, RX1x, RY0, RY0x, RY1, RY1x, RNOT,
+		       RBYTE);
+
+	call __twofish_enc_blk16;
+
+	store_ctr_16way(%rdx, %rsi, RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1);
+
+	popq %r12;
+	vzeroupper;
+
+	ret;
+ENDPROC(twofish_ctr_16way)
+
+.align 8
+twofish_xts_crypt_16way:
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 *	%r8: pointer to __twofish_enc_blk16 or __twofish_dec_blk16
+	 */
+
+	vzeroupper;
+	pushq %r12;
+
+	load_xts_16way(%rcx, %rdx, %rsi, RA0, RB0, RC0, RD0, RA1, RB1, RC1,
+		       RD1, RX0, RX0x, RX1, RX1x, RY0, RY0x, RY1, RY1x, RNOT,
+		       .Lxts_gf128mul_and_shl1_mask_0,
+		       .Lxts_gf128mul_and_shl1_mask_1);
+
+	call *%r8;
+
+	store_xts_16way(%rsi, RA0, RB0, RC0, RD0, RA1, RB1, RC1, RD1);
+
+	popq %r12;
+	vzeroupper;
+
+	ret;
+ENDPROC(twofish_xts_crypt_16way)
+
+ENTRY(twofish_xts_enc_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+	leaq __twofish_enc_blk16, %r8;
+	jmp twofish_xts_crypt_16way;
+ENDPROC(twofish_xts_enc_16way)
+
+ENTRY(twofish_xts_dec_16way)
+	/* input:
+	 *	%rdi: ctx, CTX
+	 *	%rsi: dst (16 blocks)
+	 *	%rdx: src (16 blocks)
+	 *	%rcx: iv (t ⊕ αⁿ ∈ GF(2¹²⁸))
+	 */
+	leaq __twofish_dec_blk16, %r8;
+	jmp twofish_xts_crypt_16way;
+ENDPROC(twofish_xts_dec_16way)
diff --git a/arch/x86/crypto/twofish_avx2_glue.c b/arch/x86/crypto/twofish_avx2_glue.c
new file mode 100644
index 000000000000..ce33b5be64ee
--- /dev/null
+++ b/arch/x86/crypto/twofish_avx2_glue.c
@@ -0,0 +1,584 @@
+/*
+ * Glue Code for x86_64/AVX2 assembler optimized version of Twofish
+ *
+ * Copyright © 2012-2013 Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
+ *
+ * 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.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/crypto.h>
+#include <linux/err.h>
+#include <crypto/algapi.h>
+#include <crypto/ctr.h>
+#include <crypto/twofish.h>
+#include <crypto/lrw.h>
+#include <crypto/xts.h>
+#include <asm/xcr.h>
+#include <asm/xsave.h>
+#include <asm/crypto/twofish.h>
+#include <asm/crypto/ablk_helper.h>
+#include <asm/crypto/glue_helper.h>
+#include <crypto/scatterwalk.h>
+
+#define TF_AVX2_PARALLEL_BLOCKS 16
+
+/* 16-way AVX2 parallel cipher functions */
+asmlinkage void twofish_ecb_enc_16way(struct twofish_ctx *ctx, u8 *dst,
+				      const u8 *src);
+asmlinkage void twofish_ecb_dec_16way(struct twofish_ctx *ctx, u8 *dst,
+				      const u8 *src);
+asmlinkage void twofish_cbc_dec_16way(void *ctx, u128 *dst, const u128 *src);
+
+asmlinkage void twofish_ctr_16way(void *ctx, u128 *dst, const u128 *src,
+				  le128 *iv);
+
+asmlinkage void twofish_xts_enc_16way(struct twofish_ctx *ctx, u8 *dst,
+				      const u8 *src, le128 *iv);
+asmlinkage void twofish_xts_dec_16way(struct twofish_ctx *ctx, u8 *dst,
+				      const u8 *src, le128 *iv);
+
+static inline void twofish_enc_blk_3way(struct twofish_ctx *ctx, u8 *dst,
+					const u8 *src)
+{
+	__twofish_enc_blk_3way(ctx, dst, src, false);
+}
+
+static const struct common_glue_ctx twofish_enc = {
+	.num_funcs = 4,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(twofish_ecb_enc_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(twofish_ecb_enc_8way) }
+	}, {
+		.num_blocks = 3,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(twofish_enc_blk_3way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(twofish_enc_blk) }
+	} }
+};
+
+static const struct common_glue_ctx twofish_ctr = {
+	.num_funcs = 4,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(twofish_ctr_16way) }
+	},  {
+		.num_blocks = 8,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(twofish_ctr_8way) }
+	}, {
+		.num_blocks = 3,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(twofish_enc_blk_ctr_3way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ctr = GLUE_CTR_FUNC_CAST(twofish_enc_blk_ctr) }
+	} }
+};
+
+static const struct common_glue_ctx twofish_enc_xts = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_enc_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_enc_8way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_enc) }
+	} }
+};
+
+static const struct common_glue_ctx twofish_dec = {
+	.num_funcs = 4,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(twofish_ecb_dec_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(twofish_ecb_dec_8way) }
+	}, {
+		.num_blocks = 3,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(twofish_dec_blk_3way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .ecb = GLUE_FUNC_CAST(twofish_dec_blk) }
+	} }
+};
+
+static const struct common_glue_ctx twofish_dec_cbc = {
+	.num_funcs = 4,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(twofish_cbc_dec_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(twofish_cbc_dec_8way) }
+	}, {
+		.num_blocks = 3,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(twofish_dec_blk_cbc_3way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .cbc = GLUE_CBC_FUNC_CAST(twofish_dec_blk) }
+	} }
+};
+
+static const struct common_glue_ctx twofish_dec_xts = {
+	.num_funcs = 3,
+	.fpu_blocks_limit = 8,
+
+	.funcs = { {
+		.num_blocks = 16,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_dec_16way) }
+	}, {
+		.num_blocks = 8,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_dec_8way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_dec) }
+	} }
+};
+
+static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ecb_crypt_128bit(&twofish_enc, desc, dst, src, nbytes);
+}
+
+static int ecb_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ecb_crypt_128bit(&twofish_dec, desc, dst, src, nbytes);
+}
+
+static int cbc_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_cbc_encrypt_128bit(GLUE_FUNC_CAST(twofish_enc_blk), desc,
+				       dst, src, nbytes);
+}
+
+static int cbc_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_cbc_decrypt_128bit(&twofish_dec_cbc, desc, dst, src,
+				       nbytes);
+}
+
+static int ctr_crypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		     struct scatterlist *src, unsigned int nbytes)
+{
+	return glue_ctr_crypt_128bit(&twofish_ctr, desc, dst, src, nbytes);
+}
+
+static inline bool twofish_fpu_begin(bool fpu_enabled, unsigned int nbytes)
+{
+	/* since reusing AVX functions, starts using FPU at 8 parallel blocks */
+	return glue_fpu_begin(TF_BLOCK_SIZE, 8, NULL, fpu_enabled, nbytes);
+}
+
+static inline void twofish_fpu_end(bool fpu_enabled)
+{
+	glue_fpu_end(fpu_enabled);
+}
+
+struct crypt_priv {
+	struct twofish_ctx *ctx;
+	bool fpu_enabled;
+};
+
+static void encrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
+{
+	const unsigned int bsize = TF_BLOCK_SIZE;
+	struct crypt_priv *ctx = priv;
+	int i;
+
+	ctx->fpu_enabled = twofish_fpu_begin(ctx->fpu_enabled, nbytes);
+
+	while (nbytes >= TF_AVX2_PARALLEL_BLOCKS * bsize) {
+		twofish_ecb_enc_16way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * TF_AVX2_PARALLEL_BLOCKS;
+		nbytes -= bsize * TF_AVX2_PARALLEL_BLOCKS;
+	}
+
+	while (nbytes >= 8 * bsize) {
+		twofish_ecb_enc_8way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * 8;
+		nbytes -= bsize * 8;
+	}
+
+	while (nbytes >= 3 * bsize) {
+		twofish_enc_blk_3way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * 3;
+		nbytes -= bsize * 3;
+	}
+
+	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
+		twofish_enc_blk(ctx->ctx, srcdst, srcdst);
+}
+
+static void decrypt_callback(void *priv, u8 *srcdst, unsigned int nbytes)
+{
+	const unsigned int bsize = TF_BLOCK_SIZE;
+	struct crypt_priv *ctx = priv;
+	int i;
+
+	ctx->fpu_enabled = twofish_fpu_begin(ctx->fpu_enabled, nbytes);
+
+	while (nbytes >= TF_AVX2_PARALLEL_BLOCKS * bsize) {
+		twofish_ecb_dec_16way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * TF_AVX2_PARALLEL_BLOCKS;
+		nbytes -= bsize * TF_AVX2_PARALLEL_BLOCKS;
+	}
+
+	while (nbytes >= 8 * bsize) {
+		twofish_ecb_dec_8way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * 8;
+		nbytes -= bsize * 8;
+	}
+
+	while (nbytes >= 3 * bsize) {
+		twofish_dec_blk_3way(ctx->ctx, srcdst, srcdst);
+		srcdst += bsize * 3;
+		nbytes -= bsize * 3;
+	}
+
+	for (i = 0; i < nbytes / bsize; i++, srcdst += bsize)
+		twofish_dec_blk(ctx->ctx, srcdst, srcdst);
+}
+
+static int lrw_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct twofish_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	be128 buf[TF_AVX2_PARALLEL_BLOCKS];
+	struct crypt_priv crypt_ctx = {
+		.ctx = &ctx->twofish_ctx,
+		.fpu_enabled = false,
+	};
+	struct lrw_crypt_req req = {
+		.tbuf = buf,
+		.tbuflen = sizeof(buf),
+
+		.table_ctx = &ctx->lrw_table,
+		.crypt_ctx = &crypt_ctx,
+		.crypt_fn = encrypt_callback,
+	};
+	int ret;
+
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+	ret = lrw_crypt(desc, dst, src, nbytes, &req);
+	twofish_fpu_end(crypt_ctx.fpu_enabled);
+
+	return ret;
+}
+
+static int lrw_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct twofish_lrw_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+	be128 buf[TF_AVX2_PARALLEL_BLOCKS];
+	struct crypt_priv crypt_ctx = {
+		.ctx = &ctx->twofish_ctx,
+		.fpu_enabled = false,
+	};
+	struct lrw_crypt_req req = {
+		.tbuf = buf,
+		.tbuflen = sizeof(buf),
+
+		.table_ctx = &ctx->lrw_table,
+		.crypt_ctx = &crypt_ctx,
+		.crypt_fn = decrypt_callback,
+	};
+	int ret;
+
+	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+	ret = lrw_crypt(desc, dst, src, nbytes, &req);
+	twofish_fpu_end(crypt_ctx.fpu_enabled);
+
+	return ret;
+}
+
+static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct twofish_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+
+	return glue_xts_crypt_128bit(&twofish_enc_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(twofish_enc_blk),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
+}
+
+static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
+		       struct scatterlist *src, unsigned int nbytes)
+{
+	struct twofish_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
+
+	return glue_xts_crypt_128bit(&twofish_dec_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(twofish_enc_blk),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
+}
+
+static struct crypto_alg tf_algs[10] = { {
+	.cra_name		= "__ecb-twofish-avx2",
+	.cra_driver_name	= "__driver-ecb-twofish-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= TF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct twofish_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE,
+			.max_keysize	= TF_MAX_KEY_SIZE,
+			.setkey		= twofish_setkey,
+			.encrypt	= ecb_encrypt,
+			.decrypt	= ecb_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__cbc-twofish-avx2",
+	.cra_driver_name	= "__driver-cbc-twofish-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= TF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct twofish_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE,
+			.max_keysize	= TF_MAX_KEY_SIZE,
+			.setkey		= twofish_setkey,
+			.encrypt	= cbc_encrypt,
+			.decrypt	= cbc_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__ctr-twofish-avx2",
+	.cra_driver_name	= "__driver-ctr-twofish-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= 1,
+	.cra_ctxsize		= sizeof(struct twofish_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE,
+			.max_keysize	= TF_MAX_KEY_SIZE,
+			.ivsize		= TF_BLOCK_SIZE,
+			.setkey		= twofish_setkey,
+			.encrypt	= ctr_crypt,
+			.decrypt	= ctr_crypt,
+		},
+	},
+}, {
+	.cra_name		= "__lrw-twofish-avx2",
+	.cra_driver_name	= "__driver-lrw-twofish-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= TF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct twofish_lrw_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_exit		= lrw_twofish_exit_tfm,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE +
+					  TF_BLOCK_SIZE,
+			.max_keysize	= TF_MAX_KEY_SIZE +
+					  TF_BLOCK_SIZE,
+			.ivsize		= TF_BLOCK_SIZE,
+			.setkey		= lrw_twofish_setkey,
+			.encrypt	= lrw_encrypt,
+			.decrypt	= lrw_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "__xts-twofish-avx2",
+	.cra_driver_name	= "__driver-xts-twofish-avx2",
+	.cra_priority		= 0,
+	.cra_flags		= CRYPTO_ALG_TYPE_BLKCIPHER,
+	.cra_blocksize		= TF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct twofish_xts_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_blkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_u = {
+		.blkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE * 2,
+			.max_keysize	= TF_MAX_KEY_SIZE * 2,
+			.ivsize		= TF_BLOCK_SIZE,
+			.setkey		= xts_twofish_setkey,
+			.encrypt	= xts_encrypt,
+			.decrypt	= xts_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "ecb(twofish)",
+	.cra_driver_name	= "ecb-twofish-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= TF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE,
+			.max_keysize	= TF_MAX_KEY_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "cbc(twofish)",
+	.cra_driver_name	= "cbc-twofish-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= TF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE,
+			.max_keysize	= TF_MAX_KEY_SIZE,
+			.ivsize		= TF_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= __ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "ctr(twofish)",
+	.cra_driver_name	= "ctr-twofish-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= 1,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE,
+			.max_keysize	= TF_MAX_KEY_SIZE,
+			.ivsize		= TF_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_encrypt,
+			.geniv		= "chainiv",
+		},
+	},
+}, {
+	.cra_name		= "lrw(twofish)",
+	.cra_driver_name	= "lrw-twofish-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= TF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE +
+					  TF_BLOCK_SIZE,
+			.max_keysize	= TF_MAX_KEY_SIZE +
+					  TF_BLOCK_SIZE,
+			.ivsize		= TF_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+}, {
+	.cra_name		= "xts(twofish)",
+	.cra_driver_name	= "xts-twofish-avx2",
+	.cra_priority		= 500,
+	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
+	.cra_blocksize		= TF_BLOCK_SIZE,
+	.cra_ctxsize		= sizeof(struct async_helper_ctx),
+	.cra_alignmask		= 0,
+	.cra_type		= &crypto_ablkcipher_type,
+	.cra_module		= THIS_MODULE,
+	.cra_init		= ablk_init,
+	.cra_exit		= ablk_exit,
+	.cra_u = {
+		.ablkcipher = {
+			.min_keysize	= TF_MIN_KEY_SIZE * 2,
+			.max_keysize	= TF_MAX_KEY_SIZE * 2,
+			.ivsize		= TF_BLOCK_SIZE,
+			.setkey		= ablk_set_key,
+			.encrypt	= ablk_encrypt,
+			.decrypt	= ablk_decrypt,
+		},
+	},
+} };
+
+static int __init init(void)
+{
+	u64 xcr0;
+
+	if (!cpu_has_avx2 || !cpu_has_osxsave) {
+		pr_info("AVX2 instructions are not detected.\n");
+		return -ENODEV;
+	}
+
+	xcr0 = xgetbv(XCR_XFEATURE_ENABLED_MASK);
+	if ((xcr0 & (XSTATE_SSE | XSTATE_YMM)) != (XSTATE_SSE | XSTATE_YMM)) {
+		pr_info("AVX2 detected but unusable.\n");
+		return -ENODEV;
+	}
+
+	return crypto_register_algs(tf_algs, ARRAY_SIZE(tf_algs));
+}
+
+static void __exit fini(void)
+{
+	crypto_unregister_algs(tf_algs, ARRAY_SIZE(tf_algs));
+}
+
+module_init(init);
+module_exit(fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Twofish Cipher Algorithm, AVX2 optimized");
+MODULE_ALIAS("twofish");
+MODULE_ALIAS("twofish-asm");
diff --git a/arch/x86/crypto/twofish_avx_glue.c b/arch/x86/crypto/twofish_avx_glue.c
index 94ac91d26e47..2047a562f6b3 100644
--- a/arch/x86/crypto/twofish_avx_glue.c
+++ b/arch/x86/crypto/twofish_avx_glue.c
@@ -4,6 +4,8 @@
  * Copyright (C) 2012 Johannes Goetzfried
  *     <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
  *
+ * Copyright © 2013 Jussi Kivilinna <jussi.kivilinna@iki.fi>
+ *
  * 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
@@ -48,13 +50,26 @@
 /* 8-way parallel cipher functions */
 asmlinkage void twofish_ecb_enc_8way(struct twofish_ctx *ctx, u8 *dst,
 				     const u8 *src);
+EXPORT_SYMBOL_GPL(twofish_ecb_enc_8way);
+
 asmlinkage void twofish_ecb_dec_8way(struct twofish_ctx *ctx, u8 *dst,
 				     const u8 *src);
+EXPORT_SYMBOL_GPL(twofish_ecb_dec_8way);
 
 asmlinkage void twofish_cbc_dec_8way(struct twofish_ctx *ctx, u8 *dst,
 				     const u8 *src);
+EXPORT_SYMBOL_GPL(twofish_cbc_dec_8way);
+
 asmlinkage void twofish_ctr_8way(struct twofish_ctx *ctx, u8 *dst,
 				 const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(twofish_ctr_8way);
+
+asmlinkage void twofish_xts_enc_8way(struct twofish_ctx *ctx, u8 *dst,
+				     const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(twofish_xts_enc_8way);
+asmlinkage void twofish_xts_dec_8way(struct twofish_ctx *ctx, u8 *dst,
+				     const u8 *src, le128 *iv);
+EXPORT_SYMBOL_GPL(twofish_xts_dec_8way);
 
 static inline void twofish_enc_blk_3way(struct twofish_ctx *ctx, u8 *dst,
 					const u8 *src)
@@ -62,6 +77,20 @@ static inline void twofish_enc_blk_3way(struct twofish_ctx *ctx, u8 *dst,
 	__twofish_enc_blk_3way(ctx, dst, src, false);
 }
 
+void twofish_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv,
+				  GLUE_FUNC_CAST(twofish_enc_blk));
+}
+EXPORT_SYMBOL_GPL(twofish_xts_enc);
+
+void twofish_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv)
+{
+	glue_xts_crypt_128bit_one(ctx, dst, src, iv,
+				  GLUE_FUNC_CAST(twofish_dec_blk));
+}
+EXPORT_SYMBOL_GPL(twofish_xts_dec);
+
 
 static const struct common_glue_ctx twofish_enc = {
 	.num_funcs = 3,
@@ -95,6 +124,19 @@ static const struct common_glue_ctx twofish_ctr = {
 	} }
 };
 
+static const struct common_glue_ctx twofish_enc_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = TWOFISH_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = TWOFISH_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_enc_8way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_enc) }
+	} }
+};
+
 static const struct common_glue_ctx twofish_dec = {
 	.num_funcs = 3,
 	.fpu_blocks_limit = TWOFISH_PARALLEL_BLOCKS,
@@ -127,6 +169,19 @@ static const struct common_glue_ctx twofish_dec_cbc = {
 	} }
 };
 
+static const struct common_glue_ctx twofish_dec_xts = {
+	.num_funcs = 2,
+	.fpu_blocks_limit = TWOFISH_PARALLEL_BLOCKS,
+
+	.funcs = { {
+		.num_blocks = TWOFISH_PARALLEL_BLOCKS,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_dec_8way) }
+	}, {
+		.num_blocks = 1,
+		.fn_u = { .xts = GLUE_XTS_FUNC_CAST(twofish_xts_dec) }
+	} }
+};
+
 static int ecb_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
@@ -275,54 +330,20 @@ static int xts_encrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
 	struct twofish_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-	be128 buf[TWOFISH_PARALLEL_BLOCKS];
-	struct crypt_priv crypt_ctx = {
-		.ctx = &ctx->crypt_ctx,
-		.fpu_enabled = false,
-	};
-	struct xts_crypt_req req = {
-		.tbuf = buf,
-		.tbuflen = sizeof(buf),
 
-		.tweak_ctx = &ctx->tweak_ctx,
-		.tweak_fn = XTS_TWEAK_CAST(twofish_enc_blk),
-		.crypt_ctx = &crypt_ctx,
-		.crypt_fn = encrypt_callback,
-	};
-	int ret;
-
-	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
-	ret = xts_crypt(desc, dst, src, nbytes, &req);
-	twofish_fpu_end(crypt_ctx.fpu_enabled);
-
-	return ret;
+	return glue_xts_crypt_128bit(&twofish_enc_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(twofish_enc_blk),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
 }
 
 static int xts_decrypt(struct blkcipher_desc *desc, struct scatterlist *dst,
 		       struct scatterlist *src, unsigned int nbytes)
 {
 	struct twofish_xts_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-	be128 buf[TWOFISH_PARALLEL_BLOCKS];
-	struct crypt_priv crypt_ctx = {
-		.ctx = &ctx->crypt_ctx,
-		.fpu_enabled = false,
-	};
-	struct xts_crypt_req req = {
-		.tbuf = buf,
-		.tbuflen = sizeof(buf),
-
-		.tweak_ctx = &ctx->tweak_ctx,
-		.tweak_fn = XTS_TWEAK_CAST(twofish_enc_blk),
-		.crypt_ctx = &crypt_ctx,
-		.crypt_fn = decrypt_callback,
-	};
-	int ret;
 
-	desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
-	ret = xts_crypt(desc, dst, src, nbytes, &req);
-	twofish_fpu_end(crypt_ctx.fpu_enabled);
-
-	return ret;
+	return glue_xts_crypt_128bit(&twofish_dec_xts, desc, dst, src, nbytes,
+				     XTS_TWEAK_CAST(twofish_enc_blk),
+				     &ctx->tweak_ctx, &ctx->crypt_ctx);
 }
 
 static struct crypto_alg twofish_algs[10] = { {
diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h
index 8010ebc5705f..e99ac27f95b2 100644
--- a/arch/x86/include/asm/cpufeature.h
+++ b/arch/x86/include/asm/cpufeature.h
@@ -293,6 +293,7 @@ extern const char * const x86_power_flags[32];
 #define cpu_has_ssse3		boot_cpu_has(X86_FEATURE_SSSE3)
 #define cpu_has_aes		boot_cpu_has(X86_FEATURE_AES)
 #define cpu_has_avx		boot_cpu_has(X86_FEATURE_AVX)
+#define cpu_has_avx2		boot_cpu_has(X86_FEATURE_AVX2)
 #define cpu_has_ht		boot_cpu_has(X86_FEATURE_HT)
 #define cpu_has_mp		boot_cpu_has(X86_FEATURE_MP)
 #define cpu_has_nx		boot_cpu_has(X86_FEATURE_NX)
diff --git a/arch/x86/include/asm/crypto/blowfish.h b/arch/x86/include/asm/crypto/blowfish.h
new file mode 100644
index 000000000000..f097b2face10
--- /dev/null
+++ b/arch/x86/include/asm/crypto/blowfish.h
@@ -0,0 +1,43 @@
+#ifndef ASM_X86_BLOWFISH_H
+#define ASM_X86_BLOWFISH_H
+
+#include <linux/crypto.h>
+#include <crypto/blowfish.h>
+
+#define BF_PARALLEL_BLOCKS 4
+
+/* regular block cipher functions */
+asmlinkage void __blowfish_enc_blk(struct bf_ctx *ctx, u8 *dst, const u8 *src,
+				   bool xor);
+asmlinkage void blowfish_dec_blk(struct bf_ctx *ctx, u8 *dst, const u8 *src);
+
+/* 4-way parallel cipher functions */
+asmlinkage void __blowfish_enc_blk_4way(struct bf_ctx *ctx, u8 *dst,
+					const u8 *src, bool xor);
+asmlinkage void blowfish_dec_blk_4way(struct bf_ctx *ctx, u8 *dst,
+				      const u8 *src);
+
+static inline void blowfish_enc_blk(struct bf_ctx *ctx, u8 *dst, const u8 *src)
+{
+	__blowfish_enc_blk(ctx, dst, src, false);
+}
+
+static inline void blowfish_enc_blk_xor(struct bf_ctx *ctx, u8 *dst,
+					const u8 *src)
+{
+	__blowfish_enc_blk(ctx, dst, src, true);
+}
+
+static inline void blowfish_enc_blk_4way(struct bf_ctx *ctx, u8 *dst,
+					 const u8 *src)
+{
+	__blowfish_enc_blk_4way(ctx, dst, src, false);
+}
+
+static inline void blowfish_enc_blk_xor_4way(struct bf_ctx *ctx, u8 *dst,
+				      const u8 *src)
+{
+	__blowfish_enc_blk_4way(ctx, dst, src, true);
+}
+
+#endif
diff --git a/arch/x86/include/asm/crypto/camellia.h b/arch/x86/include/asm/crypto/camellia.h
index 98038add801e..bb93333d9200 100644
--- a/arch/x86/include/asm/crypto/camellia.h
+++ b/arch/x86/include/asm/crypto/camellia.h
@@ -48,6 +48,22 @@ asmlinkage void __camellia_enc_blk_2way(struct camellia_ctx *ctx, u8 *dst,
 asmlinkage void camellia_dec_blk_2way(struct camellia_ctx *ctx, u8 *dst,
 				      const u8 *src);
 
+/* 16-way parallel cipher functions (avx/aes-ni) */
+asmlinkage void camellia_ecb_enc_16way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src);
+asmlinkage void camellia_ecb_dec_16way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src);
+
+asmlinkage void camellia_cbc_dec_16way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src);
+asmlinkage void camellia_ctr_16way(struct camellia_ctx *ctx, u8 *dst,
+				   const u8 *src, le128 *iv);
+
+asmlinkage void camellia_xts_enc_16way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src, le128 *iv);
+asmlinkage void camellia_xts_dec_16way(struct camellia_ctx *ctx, u8 *dst,
+				       const u8 *src, le128 *iv);
+
 static inline void camellia_enc_blk(struct camellia_ctx *ctx, u8 *dst,
 				    const u8 *src)
 {
@@ -79,4 +95,7 @@ extern void camellia_crypt_ctr(void *ctx, u128 *dst, const u128 *src,
 extern void camellia_crypt_ctr_2way(void *ctx, u128 *dst, const u128 *src,
 				    le128 *iv);
 
+extern void camellia_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv);
+extern void camellia_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv);
+
 #endif /* ASM_X86_CAMELLIA_H */
diff --git a/arch/x86/include/asm/crypto/glue_helper.h b/arch/x86/include/asm/crypto/glue_helper.h
index e2d65b061d27..1eef55596e82 100644
--- a/arch/x86/include/asm/crypto/glue_helper.h
+++ b/arch/x86/include/asm/crypto/glue_helper.h
@@ -14,10 +14,13 @@ typedef void (*common_glue_func_t)(void *ctx, u8 *dst, const u8 *src);
 typedef void (*common_glue_cbc_func_t)(void *ctx, u128 *dst, const u128 *src);
 typedef void (*common_glue_ctr_func_t)(void *ctx, u128 *dst, const u128 *src,
 				       le128 *iv);
+typedef void (*common_glue_xts_func_t)(void *ctx, u128 *dst, const u128 *src,
+				       le128 *iv);
 
 #define GLUE_FUNC_CAST(fn) ((common_glue_func_t)(fn))
 #define GLUE_CBC_FUNC_CAST(fn) ((common_glue_cbc_func_t)(fn))
 #define GLUE_CTR_FUNC_CAST(fn) ((common_glue_ctr_func_t)(fn))
+#define GLUE_XTS_FUNC_CAST(fn) ((common_glue_xts_func_t)(fn))
 
 struct common_glue_func_entry {
 	unsigned int num_blocks; /* number of blocks that @fn will process */
@@ -25,6 +28,7 @@ struct common_glue_func_entry {
 		common_glue_func_t ecb;
 		common_glue_cbc_func_t cbc;
 		common_glue_ctr_func_t ctr;
+		common_glue_xts_func_t xts;
 	} fn_u;
 };
 
@@ -96,6 +100,16 @@ static inline void le128_inc(le128 *i)
 	i->b = cpu_to_le64(b);
 }
 
+static inline void le128_gf128mul_x_ble(le128 *dst, const le128 *src)
+{
+	u64 a = le64_to_cpu(src->a);
+	u64 b = le64_to_cpu(src->b);
+	u64 _tt = ((s64)a >> 63) & 0x87;
+
+	dst->a = cpu_to_le64((a << 1) ^ (b >> 63));
+	dst->b = cpu_to_le64((b << 1) ^ _tt);
+}
+
 extern int glue_ecb_crypt_128bit(const struct common_glue_ctx *gctx,
 				 struct blkcipher_desc *desc,
 				 struct scatterlist *dst,
@@ -118,4 +132,14 @@ extern int glue_ctr_crypt_128bit(const struct common_glue_ctx *gctx,
 				 struct scatterlist *dst,
 				 struct scatterlist *src, unsigned int nbytes);
 
+extern int glue_xts_crypt_128bit(const struct common_glue_ctx *gctx,
+				 struct blkcipher_desc *desc,
+				 struct scatterlist *dst,
+				 struct scatterlist *src, unsigned int nbytes,
+				 common_glue_func_t tweak_fn, void *tweak_ctx,
+				 void *crypt_ctx);
+
+extern void glue_xts_crypt_128bit_one(void *ctx, u128 *dst, const u128 *src,
+				      le128 *iv, common_glue_func_t fn);
+
 #endif /* _CRYPTO_GLUE_HELPER_H */
diff --git a/arch/x86/include/asm/crypto/serpent-avx.h b/arch/x86/include/asm/crypto/serpent-avx.h
index 0da1d3e2a55c..33c2b8a435da 100644
--- a/arch/x86/include/asm/crypto/serpent-avx.h
+++ b/arch/x86/include/asm/crypto/serpent-avx.h
@@ -6,6 +6,16 @@
 
 #define SERPENT_PARALLEL_BLOCKS 8
 
+struct serpent_lrw_ctx {
+	struct lrw_table_ctx lrw_table;
+	struct serpent_ctx serpent_ctx;
+};
+
+struct serpent_xts_ctx {
+	struct serpent_ctx tweak_ctx;
+	struct serpent_ctx crypt_ctx;
+};
+
 asmlinkage void serpent_ecb_enc_8way_avx(struct serpent_ctx *ctx, u8 *dst,
 					 const u8 *src);
 asmlinkage void serpent_ecb_dec_8way_avx(struct serpent_ctx *ctx, u8 *dst,
@@ -16,4 +26,23 @@ asmlinkage void serpent_cbc_dec_8way_avx(struct serpent_ctx *ctx, u8 *dst,
 asmlinkage void serpent_ctr_8way_avx(struct serpent_ctx *ctx, u8 *dst,
 				     const u8 *src, le128 *iv);
 
+asmlinkage void serpent_xts_enc_8way_avx(struct serpent_ctx *ctx, u8 *dst,
+					 const u8 *src, le128 *iv);
+asmlinkage void serpent_xts_dec_8way_avx(struct serpent_ctx *ctx, u8 *dst,
+					 const u8 *src, le128 *iv);
+
+extern void __serpent_crypt_ctr(void *ctx, u128 *dst, const u128 *src,
+				le128 *iv);
+
+extern void serpent_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv);
+extern void serpent_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv);
+
+extern int lrw_serpent_setkey(struct crypto_tfm *tfm, const u8 *key,
+			      unsigned int keylen);
+
+extern void lrw_serpent_exit_tfm(struct crypto_tfm *tfm);
+
+extern int xts_serpent_setkey(struct crypto_tfm *tfm, const u8 *key,
+			      unsigned int keylen);
+
 #endif
diff --git a/arch/x86/include/asm/crypto/twofish.h b/arch/x86/include/asm/crypto/twofish.h
index 878c51ceebb5..e655c6029b45 100644
--- a/arch/x86/include/asm/crypto/twofish.h
+++ b/arch/x86/include/asm/crypto/twofish.h
@@ -28,6 +28,20 @@ asmlinkage void __twofish_enc_blk_3way(struct twofish_ctx *ctx, u8 *dst,
 asmlinkage void twofish_dec_blk_3way(struct twofish_ctx *ctx, u8 *dst,
 				     const u8 *src);
 
+/* 8-way parallel cipher functions */
+asmlinkage void twofish_ecb_enc_8way(struct twofish_ctx *ctx, u8 *dst,
+				     const u8 *src);
+asmlinkage void twofish_ecb_dec_8way(struct twofish_ctx *ctx, u8 *dst,
+				     const u8 *src);
+asmlinkage void twofish_cbc_dec_8way(struct twofish_ctx *ctx, u8 *dst,
+				     const u8 *src);
+asmlinkage void twofish_ctr_8way(struct twofish_ctx *ctx, u8 *dst,
+				 const u8 *src, le128 *iv);
+asmlinkage void twofish_xts_enc_8way(struct twofish_ctx *ctx, u8 *dst,
+				     const u8 *src, le128 *iv);
+asmlinkage void twofish_xts_dec_8way(struct twofish_ctx *ctx, u8 *dst,
+				     const u8 *src, le128 *iv);
+
 /* helpers from twofish_x86_64-3way module */
 extern void twofish_dec_blk_cbc_3way(void *ctx, u128 *dst, const u128 *src);
 extern void twofish_enc_blk_ctr(void *ctx, u128 *dst, const u128 *src,
@@ -43,4 +57,8 @@ extern void lrw_twofish_exit_tfm(struct crypto_tfm *tfm);
 extern int xts_twofish_setkey(struct crypto_tfm *tfm, const u8 *key,
 			      unsigned int keylen);
 
+/* helpers from twofish-avx module */
+extern void twofish_xts_enc(void *ctx, u128 *dst, const u128 *src, le128 *iv);
+extern void twofish_xts_dec(void *ctx, u128 *dst, const u128 *src, le128 *iv);
+
 #endif /* ASM_X86_TWOFISH_H */