Date:   Fri, 17 Dec 2021 15:51:50 +1100
From:   Herbert Xu <herbert@gondor.apana.org.au>
To:     Kees Cook <keescook@chromium.org>
Cc:     Boris Brezillon <bbrezillon@kernel.org>,
        Arnaud Ebalard <arno@natisbad.org>,
        Srujana Challa <schalla@marvell.com>,
        "David S. Miller" <davem@davemloft.net>,
        Suheil Chandran <schandran@marvell.com>,
        Shijith Thotton <sthotton@marvell.com>,
        Lukasz Bartosik <lbartosik@marvell.com>,
        linux-kernel@vger.kernel.org, linux-crypto@vger.kernel.org,
        linux-hardening@vger.kernel.org
Subject: Re: [PATCH] crypto: octeontx2 - Avoid stack variable overflow
Message-ID: <20211217045150.GA20345@gondor.apana.org.au>
References: <20211215225558.1995027-1-keescook@chromium.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20211215225558.1995027-1-keescook@chromium.org>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On Wed, Dec 15, 2021 at 02:55:58PM -0800, Kees Cook wrote:
>> @@ -1762,7 +1762,7 @@ void otx2_cpt_print_uc_dbg_info(struct otx2_cptpf_dev *cptpf)
>  	char engs_info[2 * OTX2_CPT_NAME_LENGTH];
>  	struct otx2_cpt_eng_grp_info *grp;
>  	struct otx2_cpt_engs_rsvd *engs;
> -	u32 mask[4];
> +	u32 mask[5];

Are you sure 144 bits will overflow u32[4]? If not then shouldn't
the fix be to remove the bogus print on mask[4]?

On another note, the debug code looks like crap if engs_num is
less than the maximum of 144 as it will print random data from
the kernel stack.

Thanks,
-- 
Email: Herbert Xu <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt