Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 24 Oct 2018 17:57:39 +0200
From:   Willy Tarreau <w@1wt.eu>
To:     Wang Hai <wanghaifine@gmail.com>
Cc:     edumazet@google.com, davem@davemloft.net, kuznet@ms2.inr.ac.ru,
        yoshfuji@linux-ipv6.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] Change judgment len position
Message-ID: <20181024155739.GA25314@1wt.eu>
References: <20181024154729.5312-1-wanghaifine@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20181024154729.5312-1-wanghaifine@gmail.com>
User-Agent: Mutt/1.6.1 (2016-04-27)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Oct 24, 2018 at 11:47:29PM +0800, Wang Hai wrote:
> To determine whether len is less than zero, it should be put before 
> the function min_t, because the return value of min_t is not likely 
> to be less than zero.

Huh? First, the <0 test is made on "len", not "min_t", so it still
is signed. Second, you're in fact completely removing the test here,
look :

>  	struct net *net = sock_net(sk);
>  	int val, len;
>  
> +	len = min_t(unsigned int, len, sizeof(int));
> +

len is used uninitialized here, so the result is undefined.

>  	if (get_user(len, optlen))
>  		return -EFAULT;

Then it gets overridden by get_user()

> -	len = min_t(unsigned int, len, sizeof(int));
> -

Then its positive values are not bounded anymore since you moved the test.

>  	if (len < 0)
>  		return -EINVAL;

Then only negative values are dropped. So unless I'm missing something
obvious, you're just allowing len to be as large as 2GB-1 based on the
user's fed optlen.

Am I wrong ?

Willy