Received-SPF: pass (google.com: domain of linux-crypto-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 20 Jul 2022 10:41:16 +0100
From:   Will Deacon <will@kernel.org>
To:     GUO Zihua <guozihua@huawei.com>
Cc:     linux-crypto@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
        herbert@gondor.apana.org.au, davem@davemloft.net,
        catalin.marinas@arm.com, ebiggers@kernel.org
Subject: Re: [PATCH v2] arm64/crypto: poly1305 fix a read out-of-bound
Message-ID: <20220720094116.GC15752@willie-the-truck>
References: <20220712075031.29061-1-guozihua@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220712075031.29061-1-guozihua@huawei.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On Tue, Jul 12, 2022 at 03:50:31PM +0800, GUO Zihua wrote:
> A kasan error was reported during fuzzing:

[...]

> This patch fixes the issue by calling poly1305_init_arm64() instead of
> poly1305_init_arch(). This is also the implementation for the same
> algorithm on arm platform.
> 
> Fixes: f569ca164751 ("crypto: arm64/poly1305 - incorporate OpenSSL/CRYPTOGAMS NEON implementation")
> Cc: stable@vger.kernel.org
> Signed-off-by: GUO Zihua <guozihua@huawei.com>
> ---
>  arch/arm64/crypto/poly1305-glue.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)

I'm not a crypto guy by any stretch of the imagination, but Ard is out
at the moment and this looks like an important fix so I had a crack at
reviewing it.

> diff --git a/arch/arm64/crypto/poly1305-glue.c b/arch/arm64/crypto/poly1305-glue.c
> index 9c3d86e397bf..1fae18ba11ed 100644
> --- a/arch/arm64/crypto/poly1305-glue.c
> +++ b/arch/arm64/crypto/poly1305-glue.c
> @@ -52,7 +52,7 @@ static void neon_poly1305_blocks(struct poly1305_desc_ctx *dctx, const u8 *src,
>  {
>  	if (unlikely(!dctx->sset)) {
>  		if (!dctx->rset) {
> -			poly1305_init_arch(dctx, src);
> +			poly1305_init_arm64(&dctx->h, src);
>  			src += POLY1305_BLOCK_SIZE;
>  			len -= POLY1305_BLOCK_SIZE;
>  			dctx->rset = 1;

With this change, we no longer initialise dctx->buflen to 0 as part of the
initialisation. Looking at neon_poly1305_do_update(), I'm a bit worried
that we could land in the 'if (likely(len >= POLY1305_BLOCK_SIZE))' block,
end up with len == 0 and fail to set dctx->buflen. Is this a problem, or is
my ignorance showing?

Will