2016-05-03 18:58:08

by Muhammad Falak R Wani

Subject: [PATCH] crypto: algif_skcipher: replace sg++ with sg_next()

Never use sg++, always use sg = sg_next(sg). Scatterlist entries can
be combined if the memory is contiguous but sg++ won't know about
that. It sure would run on the slower side.
But regardless, sg++ should never be used, only sg_next is safe.

Signed-off-by: Muhammad Falak R Wani <[email protected]>
---
crypto/algif_skcipher.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/crypto/algif_skcipher.c b/crypto/algif_skcipher.c
index 28556fc..a668867 100644
--- a/crypto/algif_skcipher.c
+++ b/crypto/algif_skcipher.c
@@ -489,7 +489,7 @@ static int skcipher_all_sg_nents(struct skcipher_ctx *ctx)
sg = sgl->sg;

while (!sg->length)
- sg++;
+ sg = sg_next(sg);

nents += sg_nents(sg);
}
@@ -553,7 +553,7 @@ static int skcipher_recvmsg_async(struct socket *sock, struct msghdr *msg,
sg = sgl->sg;

while (!sg->length)
- sg++;
+ sg = sg_next(sg);

used = min_t(unsigned long, ctx->used,
iov_iter_count(&msg->msg_iter));
@@ -676,7 +676,7 @@ static int skcipher_recvmsg_sync(struct socket *sock, struct msghdr *msg,
sg = sgl->sg;

while (!sg->length)
- sg++;
+ sg = sg_next(sg);

skcipher_request_set_crypt(&ctx->req, sg, ctx->rsgl.sg, used,
ctx->iv);
--
1.9.1

2016-05-04 04:09:47

by Herbert Xu

Subject: Re: [PATCH] crypto: algif_skcipher: replace sg++ with sg_next()

Muhammad Falak R Wani <[email protected]> wrote:
> Never use sg++, always use sg = sg_next(sg). Scatterlist entries can
> be combined if the memory is contiguous but sg++ won't know about
> that. It sure would run on the slower side.
> But regardless, sg++ should never be used, only sg_next is safe.
>
> Signed-off-by: Muhammad Falak R Wani <[email protected]>

No please the only difference between sg_next and sg++ is chaining,
which is irrelevant here.

Patch rejected.
--
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt