lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:	Fri, 17 May 2013 21:58:04 +0800
From:	Herbert Xu <herbert@...dor.apana.org.au>
To:	Eric Dumazet <eric.dumazet@...il.com>
Cc:	David Miller <davem@...emloft.net>,
	netdev <netdev@...r.kernel.org>,
	Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
	Neal Cardwell <ncardwell@...gle.com>
Subject: Re: [PATCH net-next] ipv6: use ipv6_dup_options() from
 ip6_append_data()

On Thu, May 16, 2013 at 05:27:32PM -0700, Eric Dumazet wrote:
> On Thu, 2013-05-16 at 15:23 -0700, Eric Dumazet wrote:
> > Hi Herbert
> > 
> > Looking at the code added in commit 0178b695fd6b40a62a215cb
> > ("ipv6: Copy cork options in ip6_append_data") it looks like we can have
> > either a memleak or corruption (later in ip6_cork_release()) in case one
> > of the sub-allocation (ip6_opt_dup()/ip6_rthdr_dup()) fails.
> > 
> > I would at least use a kzalloc() instead of kmalloc() in 
> > 
> > np->cork.opt = kmalloc(opt->tot_len, sk->sk_allocation);
> > 
> > Or maybe better, reuse the code in  ipv6_dup_options() so that we
> > perform a single memory allocation ?
> 
> Something like following maybe ?
> 
> [PATCH net-next] ipv6: use ipv6_dup_options() from ip6_append_data()
> 
> commit 0178b695fd6b4 ("ipv6: Copy cork options in ip6_append_data")
> added some code duplication and bad error recovery, leading to potential
> crash.
> 
> Allow ipv6_dup_options() to be called with a NULL socket argument
> so that we can reuse it.

Yes you're quite right, my code was definitely buggy.

> @@ -721,7 +721,11 @@ ipv6_dup_options(struct sock *sk, struct ipv6_txoptions *opt)
>  {
>  	struct ipv6_txoptions *opt2;
>  
> -	opt2 = sock_kmalloc(sk, opt->tot_len, GFP_ATOMIC);
> +	if (sk)
> +		opt2 = sock_kmalloc(sk, opt->tot_len, GFP_ATOMIC);
> +	else
> +		opt2 = kmalloc(opt->tot_len, GFP_ATOMIC);
> +

However, I think this function is just as buggy as the original
code that I replaced.  If you look at the code that fills in the
options in ip6_datagram_send_ctl, you'll find that the options do
not lie in the memory area of the opt + opt->tot_len.  They instead
point to data in the cmsg.

So I think we should

1) fix ipv6_dup_options to do what I tried do but in a non-buggy way;
2) make the UDP path use it.

BTW, in the UDP path we also have a socket so we can just charge the
memory to it and avoid using kmalloc at all.

Cheers,
-- 
Email: Herbert Xu <herbert@...dor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists