summaryrefslogtreecommitdiffstats
path: root/arch/x86/crypto/salsa20_glue.c
diff options
context:
space:
mode:
authorEric Biggers <ebiggers@google.com>2018-05-26 09:08:58 +0200
committerHerbert Xu <herbert@gondor.apana.org.au>2018-05-30 18:13:57 +0200
commitb7b73cd5d74694ed59abcdb4974dacb4ff8b2a2a (patch)
treeee9c6e6e31bbda855a25f76052c0922b5f4fe094 /arch/x86/crypto/salsa20_glue.c
parentcrypto: ccp - Add GET_ID SEV command (diff)
downloadlinux-b7b73cd5d74694ed59abcdb4974dacb4ff8b2a2a.tar.xz
linux-b7b73cd5d74694ed59abcdb4974dacb4ff8b2a2a.zip
crypto: x86/salsa20 - remove x86 salsa20 implementations
The x86 assembly implementations of Salsa20 use the frame base pointer register (%ebp or %rbp), which breaks frame pointer convention and breaks stack traces when unwinding from an interrupt in the crypto code. Recent (v4.10+) kernels will warn about this, e.g. WARNING: kernel stack regs at 00000000a8291e69 in syzkaller047086:4677 has bad 'bp' value 000000001077994c [...] But after looking into it, I believe there's very little reason to still retain the x86 Salsa20 code. First, these are *not* vectorized (SSE2/SSSE3/AVX2) implementations, which would be needed to get anywhere close to the best Salsa20 performance on any remotely modern x86 processor; they're just regular x86 assembly. Second, it's still unclear that anyone is actually using the kernel's Salsa20 at all, especially given that now ChaCha20 is supported too, and with much more efficient SSSE3 and AVX2 implementations. Finally, in benchmarks I did on both Intel and AMD processors with both gcc 8.1.0 and gcc 4.9.4, the x86_64 salsa20-asm is actually slightly *slower* than salsa20-generic (~3% slower on Skylake, ~10% slower on Zen), while the i686 salsa20-asm is only slightly faster than salsa20-generic (~15% faster on Skylake, ~20% faster on Zen). The gcc version made little difference. So, the x86_64 salsa20-asm is pretty clearly useless. That leaves just the i686 salsa20-asm, which based on my tests provides a 15-20% speed boost. But that's without updating the code to not use %ebp. And given the maintenance cost, the small speed difference vs. salsa20-generic, the fact that few people still use i686 kernels, the doubt that anyone is even using the kernel's Salsa20 at all, and the fact that a SSE2 implementation would almost certainly be much faster on any remotely modern x86 processor yet no one has cared enough to add one yet, I don't think it's worthwhile to keep. Thus, just remove both the x86_64 and i686 salsa20-asm implementations. Reported-by: syzbot+ffa3a158337bbc01ff09@syzkaller.appspotmail.com Signed-off-by: Eric Biggers <ebiggers@google.com> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Diffstat (limited to 'arch/x86/crypto/salsa20_glue.c')
-rw-r--r--arch/x86/crypto/salsa20_glue.c91
1 files changed, 0 insertions, 91 deletions
diff --git a/arch/x86/crypto/salsa20_glue.c b/arch/x86/crypto/salsa20_glue.c
deleted file mode 100644
index b07d7d959806..000000000000
--- a/arch/x86/crypto/salsa20_glue.c
+++ /dev/null
@@ -1,91 +0,0 @@
-/*
- * Glue code for optimized assembly version of Salsa20.
- *
- * Copyright (c) 2007 Tan Swee Heng <thesweeheng@gmail.com>
- *
- * The assembly codes are public domain assembly codes written by Daniel. J.
- * Bernstein <djb@cr.yp.to>. The codes are modified to include indentation
- * and to remove extraneous comments and functions that are not needed.
- * - i586 version, renamed as salsa20-i586-asm_32.S
- * available from <http://cr.yp.to/snuffle/salsa20/x86-pm/salsa20.s>
- * - x86-64 version, renamed as salsa20-x86_64-asm_64.S
- * available from <http://cr.yp.to/snuffle/salsa20/amd64-3/salsa20.s>
- *
- * Also modified to set up the initial state using the generic C code rather
- * than in assembly.
- *
- * 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 <asm/unaligned.h>
-#include <crypto/internal/skcipher.h>
-#include <crypto/salsa20.h>
-#include <linux/module.h>
-
-asmlinkage void salsa20_encrypt_bytes(u32 state[16], const u8 *src, u8 *dst,
- u32 bytes);
-
-static int salsa20_asm_crypt(struct skcipher_request *req)
-{
- struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
- const struct salsa20_ctx *ctx = crypto_skcipher_ctx(tfm);
- struct skcipher_walk walk;
- u32 state[16];
- int err;
-
- err = skcipher_walk_virt(&walk, req, true);
-
- crypto_salsa20_init(state, ctx, walk.iv);
-
- while (walk.nbytes > 0) {
- unsigned int nbytes = walk.nbytes;
-
- if (nbytes < walk.total)
- nbytes = round_down(nbytes, walk.stride);
-
- salsa20_encrypt_bytes(state, walk.src.virt.addr,
- walk.dst.virt.addr, nbytes);
- err = skcipher_walk_done(&walk, walk.nbytes - nbytes);
- }
-
- return err;
-}
-
-static struct skcipher_alg alg = {
- .base.cra_name = "salsa20",
- .base.cra_driver_name = "salsa20-asm",
- .base.cra_priority = 200,
- .base.cra_blocksize = 1,
- .base.cra_ctxsize = sizeof(struct salsa20_ctx),
- .base.cra_module = THIS_MODULE,
-
- .min_keysize = SALSA20_MIN_KEY_SIZE,
- .max_keysize = SALSA20_MAX_KEY_SIZE,
- .ivsize = SALSA20_IV_SIZE,
- .chunksize = SALSA20_BLOCK_SIZE,
- .setkey = crypto_salsa20_setkey,
- .encrypt = salsa20_asm_crypt,
- .decrypt = salsa20_asm_crypt,
-};
-
-static int __init init(void)
-{
- return crypto_register_skcipher(&alg);
-}
-
-static void __exit fini(void)
-{
- crypto_unregister_skcipher(&alg);
-}
-
-module_init(init);
-module_exit(fini);
-
-MODULE_LICENSE("GPL");
-MODULE_DESCRIPTION ("Salsa20 stream cipher algorithm (optimized assembly version)");
-MODULE_ALIAS_CRYPTO("salsa20");
-MODULE_ALIAS_CRYPTO("salsa20-asm");