crypto: twofish-avx - remove duplicated glue code and use shared glue code from glue_helper

Now that shared glue code is available, convert twofish-avx to use it.

Cc: Johannes Goetzfried <Johannes.Goetzfried@informatik.stud.uni-erlangen.de>
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

4 files changed, 162 insertions, 487 deletions

diff --git a/arch/x86/crypto/twofish_avx_glue.c b/arch/x86/crypto/twofish_avx_glue.c
index cabe058eba14..782b67ddaf6a 100644
--- a/arch/x86/crypto/twofish_avx_glue.c
+++ b/arch/x86/crypto/twofish_avx_glue.c
@@ -4,9 +4,6 @@
  * 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>
- *
  * 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
@@ -39,38 +36,21 @@
 #include <asm/i387.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>
 #include <linux/workqueue.h>
 #include <linux/spinlock.h>
 
-
 #define TWOFISH_PARALLEL_BLOCKS 8
 
-/* regular block cipher functions from twofish_x86_64 module */
-asmlinkage void twofish_enc_blk(struct twofish_ctx *ctx, u8 *dst,
-                                const u8 *src);
-asmlinkage void twofish_dec_blk(struct twofish_ctx *ctx, u8 *dst,
-                                const u8 *src);
-
-/* 3-way parallel cipher functions from twofish_x86_64-3way module */
-asmlinkage void __twofish_enc_blk_3way(struct twofish_ctx *ctx, u8 *dst,
-                                       const u8 *src, bool xor);
-asmlinkage void twofish_dec_blk_3way(struct twofish_ctx *ctx, u8 *dst,
-                                     const u8 *src);
-
 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 inline void twofish_enc_blk_3way_xor(struct twofish_ctx *ctx, u8 *dst,
-                                            const u8 *src)
-{
-        __twofish_enc_blk_3way(ctx, dst, src, true);
-}
-
 /* 8-way parallel cipher functions */
 asmlinkage void __twofish_enc_blk_8way(struct twofish_ctx *ctx, u8 *dst,
                                        const u8 *src, bool xor);
@@ -95,423 +75,142 @@ static inline void twofish_dec_blk_xway(struct twofish_ctx *ctx, u8 *dst,
         twofish_dec_blk_8way(ctx, dst, src);
 }
 
-
-static inline bool twofish_fpu_begin(bool fpu_enabled, unsigned int nbytes)
+static void twofish_dec_blk_cbc_xway(void *ctx, u128 *dst, const u128 *src)
 {
-        if (fpu_enabled)
-                return true;
+        u128 ivs[TWOFISH_PARALLEL_BLOCKS - 1];
+        unsigned int j;
 
-        /* AVX is only used when chunk to be processed is large enough, so
-         * do not enable FPU until it is necessary.
-         */
-        if (nbytes < TF_BLOCK_SIZE * TWOFISH_PARALLEL_BLOCKS)
-                return false;
+        for (j = 0; j < TWOFISH_PARALLEL_BLOCKS - 1; j++)
+                ivs[j] = src[j];
 
-        kernel_fpu_begin();
-        return true;
-}
+        twofish_dec_blk_xway(ctx, (u8 *)dst, (u8 *)src);
 
-static inline void twofish_fpu_end(bool fpu_enabled)
-{
-        if (fpu_enabled)
-                kernel_fpu_end();
+        for (j = 0; j < TWOFISH_PARALLEL_BLOCKS - 1; j++)
+                u128_xor(dst + (j + 1), dst + (j + 1), ivs + j);
 }
 
-static int ecb_crypt(struct blkcipher_desc *desc, struct blkcipher_walk *walk,
-                     bool enc)
+static void twofish_enc_blk_ctr_xway(void *ctx, u128 *dst, const u128 *src,
+                                     u128 *iv)
 {
-        bool fpu_enabled = false;
-        struct twofish_ctx *ctx = crypto_blkcipher_ctx(desc->tfm);
-        const unsigned int bsize = TF_BLOCK_SIZE;
-        unsigned int nbytes;
-        int err;
+        be128 ctrblks[TWOFISH_PARALLEL_BLOCKS];
+        unsigned int i;
 
-        err = blkcipher_walk_virt(desc, walk);
-        desc->flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+        for (i = 0; i < TWOFISH_PARALLEL_BLOCKS; i++) {
+                if (dst != src)
+                        dst[i] = src[i];
 
-        while ((nbytes = walk->nbytes)) {
-                u8 *wsrc = walk->src.virt.addr;
-                u8 *wdst = walk->dst.virt.addr;
-
-                fpu_enabled = twofish_fpu_begin(fpu_enabled, nbytes);
-
-                /* Process multi-block batch */
-                if (nbytes >= bsize * TWOFISH_PARALLEL_BLOCKS) {
-                        do {
-                                if (enc)
-                                        twofish_enc_blk_xway(ctx, wdst, wsrc);
-                                else
-                                        twofish_dec_blk_xway(ctx, wdst, wsrc);
-
-                                wsrc += bsize * TWOFISH_PARALLEL_BLOCKS;
-                                wdst += bsize * TWOFISH_PARALLEL_BLOCKS;
-                                nbytes -= bsize * TWOFISH_PARALLEL_BLOCKS;
-                        } while (nbytes >= bsize * TWOFISH_PARALLEL_BLOCKS);
-
-                        if (nbytes < bsize)
-                                goto done;
-                }
-
-                /* Process three block batch */
-                if (nbytes >= bsize * 3) {
-                        do {
-                                if (enc)
-                                        twofish_enc_blk_3way(ctx, wdst, wsrc);
-                                else
-                                        twofish_dec_blk_3way(ctx, wdst, wsrc);
-
-                                wsrc += bsize * 3;
-                                wdst += bsize * 3;
-                                nbytes -= bsize * 3;
-                        } while (nbytes >= bsize * 3);
-
-                        if (nbytes < bsize)
-                                goto done;
-                }
-
-                /* Handle leftovers */
-                do {
-                        if (enc)
-                                twofish_enc_blk(ctx, wdst, wsrc);
-                        else
-                                twofish_dec_blk(ctx, wdst, wsrc);
-
-                        wsrc += bsize;
-                        wdst += bsize;
-                        nbytes -= bsize;
-                } while (nbytes >= bsize);
-
-done:
-                err = blkcipher_walk_done(desc, walk, nbytes);
+                u128_to_be128(&ctrblks[i], iv);
+                u128_inc(iv);
         }
 
-        twofish_fpu_end(fpu_enabled);
-        return err;
+        twofish_enc_blk_xway_xor(ctx, (u8 *)dst, (u8 *)ctrblks);
 }
 
+static const struct common_glue_ctx twofish_enc = {
+        .num_funcs = 3,
+        .fpu_blocks_limit = TWOFISH_PARALLEL_BLOCKS,
+
+        .funcs = { {
+                .num_blocks = TWOFISH_PARALLEL_BLOCKS,
+                .fn_u = { .ecb = GLUE_FUNC_CAST(twofish_enc_blk_xway) }
+        }, {
+                .num_blocks = 3,
+                .fn_u = { .ecb = GLUE_FUNC_CAST(twofish_enc_blk_3way) }
+        }, {