Re: [PATCH 1/2] netfilter: fix a memory leak in nf_conntrack_in

[Florian Westphal <fw@strlen.de>]

Maciej Żenczykowski <zenczykowski@gmail.com> wrote:
> From: Maciej Żenczykowski <maze@google.com>
> 
> Cc: Cong Wang <xiyou.wangcong@gmail.com>
> Cc: Eric Dumazet <edumazet@google.com>
> Cc: Pablo Neira Ayuso <pablo@netfilter.org>
> Signed-off-by: Maciej Żenczykowski <maze@google.com>
> ---
>  net/netfilter/nf_conntrack_core.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
> index 0c63120b2db2..35459d04a050 100644
> --- a/net/netfilter/nf_conntrack_core.c
> +++ b/net/netfilter/nf_conntrack_core.c
> @@ -1679,7 +1679,8 @@ nf_conntrack_in(struct sk_buff *skb, const struct nf_hook_state *state)
>  		if ((tmpl && !nf_ct_is_template(tmpl)) ||
>  		     ctinfo == IP_CT_UNTRACKED) {
>  			NF_CT_STAT_INC_ATOMIC(state->net, ignore);
> -			return NF_ACCEPT;
> +			ret = NF_ACCEPT;

This looks wrong.

> +			goto out;

This puts tmpl, causing underflow of skb->nfct.
When we enter nf_conntrack_in and this branch, then 'tmpl'
is already assigned to skb->nfct, it will be put when skb
is free'd.

nf_ct_get() doesn't increment the refcnt.

tmpl only needs to be put in case of ...


>  		}
>  		skb->_nfct = 0;

...this.