crypto: x86/glue_helper - use le128 instead of u128 for CTR mode

'u128' currently used for CTR mode is on little-endian 'long long' swapped
and would require extra swap operations by SSE/AVX code. Use of le128
instead of u128 allows IV calculations to be done with vector registers
easier.

Signed-off-by: Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

2 files changed, 19 insertions, 13 deletions

diff --git a/arch/x86/include/asm/crypto/glue_helper.h b/arch/x86/include/asm/crypto/glue_helper.h
index 3e408bd..e2d65b0 100644
--- a/arch/x86/include/asm/crypto/glue_helper.h
+++ b/arch/x86/include/asm/crypto/glue_helper.h
@@ -13,7 +13,7 @@
 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,
-				       u128 *iv);
+				       le128 *iv);
 
 #define GLUE_FUNC_CAST(fn) ((common_glue_func_t)(fn))
 #define GLUE_CBC_FUNC_CAST(fn) ((common_glue_cbc_func_t)(fn))
@@ -71,23 +71,29 @@ static inline void glue_fpu_end(bool fpu_enabled)
 		kernel_fpu_end();
 }
 
-static inline void u128_to_be128(be128 *dst, const u128 *src)
+static inline void le128_to_be128(be128 *dst, const le128 *src)
 {
-	dst->a = cpu_to_be64(src->a);
-	dst->b = cpu_to_be64(src->b);
+	dst->a = cpu_to_be64(le64_to_cpu(src->a));
+	dst->b = cpu_to_be64(le64_to_cpu(src->b));
 }
 
-static inline void be128_to_u128(u128 *dst, const be128 *src)
+static inline void be128_to_le128(le128 *dst, const be128 *src)
 {
-	dst->a = be64_to_cpu(src->a);
-	dst->b = be64_to_cpu(src->b);
+	dst->a = cpu_to_le64(be64_to_cpu(src->a));
+	dst->b = cpu_to_le64(be64_to_cpu(src->b));
 }
 
-static inline void u128_inc(u128 *i)
+static inline void le128_inc(le128 *i)
 {
-	i->b++;
-	if (!i->b)
-		i->a++;
+	u64 a = le64_to_cpu(i->a);
+	u64 b = le64_to_cpu(i->b);
+
+	b++;
+	if (!b)
+		a++;
+
+	i->a = cpu_to_le64(a);
+	i->b = cpu_to_le64(b);
 }
 
 extern int glue_ecb_crypt_128bit(const struct common_glue_ctx *gctx,
diff --git a/arch/x86/include/asm/crypto/twofish.h b/arch/x86/include/asm/crypto/twofish.h
index 9d2c514..878c51c 100644
--- a/arch/x86/include/asm/crypto/twofish.h
+++ b/arch/x86/include/asm/crypto/twofish.h
@@ -31,9 +31,9 @@ asmlinkage void twofish_dec_blk_3way(struct twofish_ctx *ctx, u8 *dst,
 /* 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,
-				u128 *iv);
+				le128 *iv);
 extern void twofish_enc_blk_ctr_3way(void *ctx, u128 *dst, const u128 *src,
-				     u128 *iv);
+				     le128 *iv);
 
 extern int lrw_twofish_setkey(struct crypto_tfm *tfm, const u8 *key,
 			      unsigned int keylen);