@@ -29,8 +29,6 @@
#include <crypto/b128ops.h>
#include <crypto/gf128mul.h>
-#define LRW_BUFFER_SIZE 128u
-
#define LRW_BLOCK_SIZE 16
struct priv {
@@ -56,19 +54,7 @@ struct priv {
};
struct rctx {
- be128 buf[LRW_BUFFER_SIZE / sizeof(be128)];
-
- be128 t;
-
- be128 *ext;
-
- struct scatterlist srcbuf[2];
- struct scatterlist dstbuf[2];
- struct scatterlist *src;
- struct scatterlist *dst;
-
- unsigned int left;
-
+ be128 t, orig_iv;
struct skcipher_request subreq;
};
@@ -135,85 +121,26 @@ static int next_index(u32 *counter)
return res;
}
-static int post_crypt(struct skcipher_request *req)
+/*
+ * We compute the tweak masks twice (both before and after the ECB encryption or
+ * decryption) to avoid having to allocate a temporary buffer and/or make
+ * mutliple calls to the 'ecb(..)' instance, which usually would be slower than
+ * just doing the gf128mul_x_ble() calls again.
+ */
+static int xor_tweak(struct skcipher_request *req, struct skcipher_request *subreq)
{
- struct rctx *rctx = skcipher_request_ctx(req);
- be128 *buf = rctx->ext ?: rctx->buf;
- struct skcipher_request *subreq;
const int bs = LRW_BLOCK_SIZE;
- struct skcipher_walk w;
- struct scatterlist *sg;
- unsigned offset;
- int err;
-
- subreq = &rctx->subreq;
- err = skcipher_walk_virt(&w, subreq, false);
-
- while (w.nbytes) {
- unsigned int avail = w.nbytes;
- be128 *wdst;
-
- wdst = w.dst.virt.addr;
-
- do {
- be128_xor(wdst, buf++, wdst);
- wdst++;
- } while ((avail -= bs) >= bs);
-
- err = skcipher_walk_done(&w, avail);
- }
-
- rctx->left -= subreq->cryptlen;
-
- if (err || !rctx->left)
- goto out;
-
- rctx->dst = rctx->dstbuf;
-
- scatterwalk_done(&w.out, 0, 1);
- sg = w.out.sg;
- offset = w.out.offset;
-
- if (rctx->dst != sg) {
- rctx->dst[0] = *sg;
- sg_unmark_end(rctx->dst);
- scatterwalk_crypto_chain(rctx->dst, sg_next(sg), 2);
- }
- rctx->dst[0].length -= offset - sg->offset;
- rctx->dst[0].offset = offset;
-
-out:
- return err;
-}
-
-static int pre_crypt(struct skcipher_request *req)
-{
struct crypto_skcipher *tfm = crypto_skcipher_reqtfm(req);
- struct rctx *rctx = skcipher_request_ctx(req);
struct priv *ctx = crypto_skcipher_ctx(tfm);
- be128 *buf = rctx->ext ?: rctx->buf;
- struct skcipher_request *subreq;
- const int bs = LRW_BLOCK_SIZE;
+ struct rctx *rctx = skcipher_request_ctx(req);
+ be128 t = rctx->t;
struct skcipher_walk w;
- struct scatterlist *sg;
- unsigned cryptlen;
- unsigned offset;
- bool more;
__u32 *iv;
u32 counter[4];
int err;
- subreq = &rctx->subreq;
skcipher_request_set_tfm(subreq, tfm);
- cryptlen = subreq->cryptlen;
- more = rctx->left > cryptlen;
- if (!more)
- cryptlen = rctx->left;
-
- skcipher_request_set_crypt(subreq, rctx->src, rctx->dst,
- cryptlen, req->iv);
-
err = skcipher_walk_virt(&w, subreq, false);
iv = w.iv;
@@ -231,13 +158,11 @@ static int pre_crypt(struct skcipher_request *req)
wdst = w.dst.virt.addr;
do {
- *buf++ = rctx->t;
- be128_xor(wdst++, &rctx->t, wsrc++);
+ be128_xor(wdst++, &t, wsrc++);
/* T <- I*Key2, using the optimization
* discussed in the specification */
- be128_xor(&rctx->t, &rctx->t,
- &ctx->mulinc[next_index(counter)]);
+ be128_xor(&t, &t, &ctx->mulinc[next_index(counter)]);
} while ((avail -= bs) >= bs);
if (w.nbytes == w.total) {
@@ -250,175 +175,60 @@ static int pre_crypt(struct skcipher_request *req)
err = skcipher_walk_done(&w, avail);
}
- skcipher_request_set_tfm(subreq, ctx->child);
- skcipher_request_set_crypt(subreq, rctx->dst, rctx->dst,
- cryptlen, NULL);
-
- if (err || !more)
- goto out;
-
- rctx->src = rctx->srcbuf;
-
- scatterwalk_done(&w.in, 0, 1);
- sg = w.in.sg;
- offset = w.in.offset;
-
- if (rctx->src != sg) {
- rctx->src[0] = *sg;
- sg_unmark_end(rctx->src);
- scatterwalk_crypto_chain(rctx->src, sg_next(sg), 2);
- }
- rctx->src[0].length -= offset - sg->offset;
- rctx->src[0].offset = offset;
-
-out:
return err;
}
-static int init_crypt(struct skcipher_request *req, crypto_completion_t done)
-{
- struct priv *ctx = crypto_skcipher_ctx(crypto_skcipher_reqtfm(req));
- struct rctx *rctx = skcipher_request_ctx(req);
- struct skcipher_request *subreq;
- gfp_t gfp;
-
- subreq = &rctx->subreq;
- skcipher_request_set_callback(subreq, req->base.flags, done, req);
-
- gfp = req->base.flags & CRYPTO_TFM_REQ_MAY_SLEEP ? GFP_KERNEL :
- GFP_ATOMIC;
- rctx->ext = NULL;
-
- subreq->cryptlen = LRW_BUFFER_SIZE;
- if (req->cryptlen > LRW_BUFFER_SIZE) {
- unsigned int n = min(req->cryptlen, (unsigned int)PAGE_SIZE);
-
- rctx->ext = kmalloc(n, gfp);
- if (rctx->ext)
- subreq->cryptlen = n;
- }
-
- rctx->src = req->src;
- rctx->dst = req->dst;
- rctx->left = req->cryptlen;
-
- /* calculate first value of T */
- memcpy(&rctx->t, req->iv, sizeof(rctx->t));
-
- /* T <- I*Key2 */
- gf128mul_64k_bbe(&rctx->t, ctx->table);
-
- return 0;
-}
-
-static void exit_crypt(struct skcipher_request *req)
+static void crypt_done(struct crypto_async_request *areq, int err)
{
+ struct skcipher_request *req = areq->data;
struct rctx *rctx = skcipher_request_ctx(req);
+ struct skcipher_request *subreq = &rctx->subreq;
- rctx->left = 0;
+ if (!err)
+ err = xor_tweak(req, subreq);
- if (rctx->ext)
- kzfree(rctx->ext);
+ skcipher_request_complete(req, err);
}
-static int do_encrypt(struct skcipher_request *req, int err)
+static void init_crypt(struct skcipher_request *req)
{
+ struct priv *ctx = crypto_skcipher_ctx(crypto_skcipher_reqtfm(req));
struct rctx *rctx = skcipher_request_ctx(req);
- struct skcipher_request *subreq;
+ struct skcipher_request *subreq = &rctx->subreq;
- subreq = &rctx->subreq;
-
- while (!err && rctx->left) {
- err = pre_crypt(req) ?:
- crypto_skcipher_encrypt(subreq) ?:
- post_crypt(req);
-
- if (err == -EINPROGRESS || err == -EBUSY)
- return err;
- }
-
- exit_crypt(req);
- return err;
-}
-
-static void encrypt_done(struct crypto_async_request *areq, int err)
-{
- struct skcipher_request *req = areq->data;
- struct skcipher_request *subreq;
- struct rctx *rctx;
-
- rctx = skcipher_request_ctx(req);
-
- if (err == -EINPROGRESS) {
- if (rctx->left != req->cryptlen)
- return;
- goto out;
- }
-
- subreq = &rctx->subreq;
- subreq->base.flags &= CRYPTO_TFM_REQ_MAY_BACKLOG;
+ skcipher_request_set_tfm(subreq, ctx->child);
+ skcipher_request_set_callback(subreq, req->base.flags, crypt_done, req);
+ skcipher_request_set_crypt(subreq, req->dst, req->dst,
+ req->cryptlen, &rctx->orig_iv);
- err = do_encrypt(req, err ?: post_crypt(req));
- if (rctx->left)
- return;
+ /* calculate first value of T */
+ memcpy(&rctx->orig_iv, req->iv, sizeof(rctx->t));
+ rctx->t = rctx->orig_iv;
-out:
- skcipher_request_complete(req, err);
+ /* T <- I*Key2 */
+ gf128mul_64k_bbe(&rctx->t, ctx->table);
}
static int encrypt(struct skcipher_request *req)
-{
- return do_encrypt(req, init_crypt(req, encrypt_done));
-}
-
-static int do_decrypt(struct skcipher_request *req, int err)
{
struct rctx *rctx = skcipher_request_ctx(req);
- struct skcipher_request *subreq;
-
- subreq = &rctx->subreq;
+ struct skcipher_request *subreq = &rctx->subreq;
- while (!err && rctx->left) {
- err = pre_crypt(req) ?:
- crypto_skcipher_decrypt(subreq) ?:
- post_crypt(req);
-
- if (err == -EINPROGRESS || err == -EBUSY)
- return err;
- }
-
- exit_crypt(req);
- return err;
-}
-
-static void decrypt_done(struct crypto_async_request *areq, int err)
-{
- struct skcipher_request *req = areq->data;
- struct skcipher_request *subreq;
- struct rctx *rctx;
-
- rctx = skcipher_request_ctx(req);
-
- if (err == -EINPROGRESS) {
- if (rctx->left != req->cryptlen)
- return;
- goto out;
- }
-
- subreq = &rctx->subreq;
- subreq->base.flags &= CRYPTO_TFM_REQ_MAY_BACKLOG;
-
- err = do_decrypt(req, err ?: post_crypt(req));
- if (rctx->left)
- return;
-
-out:
- skcipher_request_complete(req, err);
+ init_crypt(req);
+ return xor_tweak(req, req) ?:
+ crypto_skcipher_encrypt(subreq) ?:
+ xor_tweak(req, subreq);
}
static int decrypt(struct skcipher_request *req)
{
- return do_decrypt(req, init_crypt(req, decrypt_done));
+ struct rctx *rctx = skcipher_request_ctx(req);
+ struct skcipher_request *subreq = &rctx->subreq;
+
+ init_crypt(req);
+ return xor_tweak(req, req) ?:
+ crypto_skcipher_decrypt(subreq) ?:
+ xor_tweak(req, subreq);
}
static int init_tfm(struct crypto_skcipher *tfm)
This patch simplifies the LRW template to recompute the LRW tweaks from scratch in the second pass and thus also removes the need to allocate a dynamic buffer using kmalloc(). As discussed at [1], the use of kmalloc causes deadlocks with dm-crypt. PERFORMANCE MEASUREMENTS (x86_64) Performed using: https://gitlab.com/omos/linux-crypto-bench Crypto driver used: lrw(ecb-aes-aesni) The results show that the new code has about the same performance as the old code. For 512-byte message it seems to be even slightly faster, but that might be just noise. Before: ALGORITHM KEY (b) DATA (B) TIME ENC (ns) TIME DEC (ns) lrw(aes) 256 64 200 203 lrw(aes) 320 64 202 204 lrw(aes) 384 64 204 205 lrw(aes) 256 512 415 415 lrw(aes) 320 512 432 440 lrw(aes) 384 512 449 451 lrw(aes) 256 4096 1838 1995 lrw(aes) 320 4096 2123 1980 lrw(aes) 384 4096 2100 2119 lrw(aes) 256 16384 7183 6954 lrw(aes) 320 16384 7844 7631 lrw(aes) 384 16384 8256 8126 lrw(aes) 256 32768 14772 14484 lrw(aes) 320 32768 15281 15431 lrw(aes) 384 32768 16469 16293 After: ALGORITHM KEY (b) DATA (B) TIME ENC (ns) TIME DEC (ns) lrw(aes) 256 64 197 196 lrw(aes) 320 64 200 197 lrw(aes) 384 64 203 199 lrw(aes) 256 512 385 380 lrw(aes) 320 512 401 395 lrw(aes) 384 512 415 415 lrw(aes) 256 4096 1869 1846 lrw(aes) 320 4096 2080 1981 lrw(aes) 384 4096 2160 2109 lrw(aes) 256 16384 7077 7127 lrw(aes) 320 16384 7807 7766 lrw(aes) 384 16384 8108 8357 lrw(aes) 256 32768 14111 14454 lrw(aes) 320 32768 15268 15082 lrw(aes) 384 32768 16581 16250 [1] https://lkml.org/lkml/2018/8/23/1315 Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com> --- crypto/lrw.c | 274 ++++++++------------------------------------------- 1 file changed, 42 insertions(+), 232 deletions(-)