Received-SPF: pass (google.com: best guess record for domain of linux-crypto-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 27 Jan 2020 18:55:51 -0800
From:   Eric Biggers <ebiggers@kernel.org>
To:     Gilad Ben-Yossef <gilad@benyossef.com>
Cc:     Herbert Xu <herbert@gondor.apana.org.au>,
        "David S. Miller" <davem@davemloft.net>,
        Ofir Drang <ofir.drang@arm.com>,
        Geert Uytterhoeven <geert@linux-m68k.org>,
        linux-crypto@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [RFC v3] crypto: ccree - protect against short scatterlists
Message-ID: <20200128025551.GE960@sol.localdomain>
References: <20200127150822.12126-1-gilad@benyossef.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200127150822.12126-1-gilad@benyossef.com>
Sender: linux-crypto-owner@vger.kernel.org
Precedence: bulk

On Mon, Jan 27, 2020 at 05:08:21PM +0200, Gilad Ben-Yossef wrote:
> Deal gracefully with the event of being handed a scatterlist
> which is shorter than expected.
> 
> This mitigates a crash in some cases due to
> attempt to map empty (but not NULL) scatterlists with none
> zero lengths.
> 
> Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
> Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>

It's definitely wrong use of the crypto API to pass a scatterlist that's too
short.  Note that this is *not* what the test code is doing.

So I don't think you should be hacking around it here.

It is possible the bug is actually in cc_aead_chain_data()?  It looks like it's
adding the authentication tag size to the source data size for encryption, which
is not correct.  The authentication tag is part of the destination only.

	size_for_map += (direct == DRV_CRYPTO_DIRECTION_ENCRYPT) ?
			authsize : 0;
	src_mapped_nents = cc_get_sgl_nents(dev, req->src, size_for_map,
					    &src_last_bytes);

- Eric