[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 10 Oct 2016 10:37:56 -0700
From: Eric Biggers <ebiggers@...gle.com>
To: Joe Perches <joe@...ches.com>
Cc: herbert@...dor.apana.org.au, davem@...emloft.net,
linux-crypto@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] crypto: cmac - fix alignment of 'consts'
On Mon, Oct 10, 2016 at 10:29:55AM -0700, Joe Perches wrote:
> On Mon, 2016-10-10 at 10:15 -0700, Eric Biggers wrote:
> > The per-transform 'consts' array is accessed as __be64 in
> > crypto_cmac_digest_setkey() but was only guaranteed to be aligned to
> > __alignof__(long). Fix this by aligning it to __alignof__(__be64).
> []
> > diff --git a/crypto/cmac.c b/crypto/cmac.c
> []
> > @@ -57,7 +57,8 @@ static int crypto_cmac_digest_setkey(struct crypto_shash *parent,
> > unsigned long alignmask = crypto_shash_alignmask(parent);
> > struct cmac_tfm_ctx *ctx = crypto_shash_ctx(parent);
> > unsigned int bs = crypto_shash_blocksize(parent);
> > - __be64 *consts = PTR_ALIGN((void *)ctx->ctx, alignmask + 1);
> > + __be64 *consts = PTR_ALIGN((void *)ctx->ctx,
> > + (alignmask | (__alignof__(__be64) - 1)) + 1);
>
> Using a bitwise or looks very odd there. Perhaps:
>
> min(alignmask + 1, __alignof__(__be64))
>
Alignment has to be a power of 2. From the code I've read, crypto drivers work
with alignment a lot and use bitwise OR to mean "the more restrictive of these
alignmasks". So I believe the way it's written is the preferred style.
Eric
Powered by blists - more mailing lists