Revert "crypto: twofish - add AVX2/x86_64 assembler implementation of twofish cipher"

This reverts commit cf1521a1.

Instruction (vpgatherdd) that this implementation relied on turned out to be
slow performer on real hardware (i5-4570). The previous 8-way twofish/AVX
implementation is therefore faster and this implementation should be removed.

Converting this implementation to use the same method as in twofish/AVX for
table look-ups would give additional ~3% speed up vs twofish/AVX, but would
hardly be worth of the added code and binary size.
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@iki.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

arch/x86/crypto/Makefile

@@ -43,7 +43,6 @@ endif
 ifeq ($(avx2_supported),yes)
 	obj-$(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
@@ -73,7 +72,6 @@ endif
 ifeq ($(avx2_supported),yes)
 	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

arch/x86/crypto/twofish_avx_glue.c

@@ -50,26 +50,18 @@
 /* 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)
@@ -77,19 +69,17 @@ 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)
+static 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)
+static 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 = {

arch/x86/include/asm/crypto/twofish.h

@@ -28,20 +28,6 @@ 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,
@@ -57,8 +43,4 @@ 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 */

crypto/Kconfig

@@ -1297,30 +1297,6 @@ config CRYPTO_TWOFISH_AVX_X86_64
 	  See also:
 	  <http://www.schneier.com/twofish.html>
 
-config CRYPTO_TWOFISH_AVX2_X86_64
-	tristate "Twofish cipher algorithm (x86_64/AVX2)"
-	depends on X86 && 64BIT
-	select CRYPTO_ALGAPI
-	select CRYPTO_CRYPTD
-	select CRYPTO_ABLK_HELPER_X86
-	select CRYPTO_GLUE_HELPER_X86
-	select CRYPTO_TWOFISH_COMMON
-	select CRYPTO_TWOFISH_X86_64
-	select CRYPTO_TWOFISH_X86_64_3WAY
-	select CRYPTO_TWOFISH_AVX_X86_64
-	select CRYPTO_LRW
-	select CRYPTO_XTS
-	help
-	  Twofish cipher algorithm (x86_64/AVX2).
-
-	  Twofish was submitted as an AES (Advanced Encryption Standard)
-	  candidate cipher by researchers at CounterPane Systems.  It is a
-	  16 round block cipher supporting key sizes of 128, 192, and 256
-	  bits.
-
-	  See also:
-	  <http://www.schneier.com/twofish.html>
-
 comment "Compression"
 
 config CRYPTO_DEFLATE

crypto/testmgr.c

@@ -1653,9 +1653,6 @@ static const struct alg_test_desc alg_test_descs[] = {
 	}, {
 		.alg = "__cbc-twofish-avx",
 		.test = alg_test_null,
-	}, {
-		.alg = "__cbc-twofish-avx2",
-		.test = alg_test_null,
 	}, {
 		.alg = "__driver-cbc-aes-aesni",
 		.test = alg_test_null,
@@ -1684,9 +1681,6 @@ static const struct alg_test_desc alg_test_descs[] = {
 	}, {
 		.alg = "__driver-cbc-twofish-avx",
 		.test = alg_test_null,
-	}, {
-		.alg = "__driver-cbc-twofish-avx2",
-		.test = alg_test_null,
 	}, {
 		.alg = "__driver-ecb-aes-aesni",
 		.test = alg_test_null,
@@ -1715,9 +1709,6 @@ static const struct alg_test_desc alg_test_descs[] = {
 	}, {
 		.alg = "__driver-ecb-twofish-avx",
 		.test = alg_test_null,
-	}, {
-		.alg = "__driver-ecb-twofish-avx2",
-		.test = alg_test_null,
 	}, {
 		.alg = "__ghash-pclmulqdqni",
 		.test = alg_test_null,
@@ -2018,9 +2009,6 @@ static const struct alg_test_desc alg_test_descs[] = {
 	}, {
 		.alg = "cryptd(__driver-ecb-twofish-avx)",
 		.test = alg_test_null,
-	}, {
-		.alg = "cryptd(__driver-ecb-twofish-avx2)",
-		.test = alg_test_null,
 	}, {
 		.alg = "cryptd(__driver-gcm-aes-aesni)",
 		.test = alg_test_null,