<html>
<head>
<base href="https://bugzilla.netfilter.org/" />
</head>
<body><table border="1" cellspacing="0" cellpadding="8">
<tr>
<th>Bug ID</th>
<td><a class="bz_bug_link
bz_status_NEW "
title="NEW - nf_conntrack_reasm.c : Silent discard of overlapping fragments is not silent"
href="https://bugzilla.netfilter.org/show_bug.cgi?id=987">987</a>
</td>
</tr>
<tr>
<th>Summary</th>
<td>nf_conntrack_reasm.c : Silent discard of overlapping fragments is not silent
</td>
</tr>
<tr>
<th>Product</th>
<td>netfilter/iptables
</td>
</tr>
<tr>
<th>Version</th>
<td>unspecified
</td>
</tr>
<tr>
<th>Hardware</th>
<td>x86_64
</td>
</tr>
<tr>
<th>OS</th>
<td>Debian GNU/Linux
</td>
</tr>
<tr>
<th>Status</th>
<td>NEW
</td>
</tr>
<tr>
<th>Severity</th>
<td>normal
</td>
</tr>
<tr>
<th>Priority</th>
<td>P5
</td>
</tr>
<tr>
<th>Component</th>
<td>nf_conntrack
</td>
</tr>
<tr>
<th>Assignee</th>
<td>netfilter-buglog@lists.netfilter.org
</td>
</tr>
<tr>
<th>Reporter</th>
<td>netfilter@kiffie.de
</td>
</tr></table>
<p>
<div>
<pre>When conntrack collects IPv6 packet fragments and encounters an overlapping (or
a replayed identical) fragment, then the overlapping fragment gets forwarded
and is not silently dropped with all the others from the fragment queue.
Possible explanation:
The function nf_ct_frag6_queue contains checks for overlapping fragments:
/* Check for overlap with preceding fragment. */
if (prev &&
(NFCT_FRAG6_CB(prev)->offset + prev->len) > offset)
goto discard_fq;
/* Look for overlap with succeeding segment. */
if (next && NFCT_FRAG6_CB(next)->offset < end)
goto discard_fq;
This causes the destruction of the fragment queue and returns an error code:
discard_fq:
inet_frag_kill(&fq->q, &nf_frags);
err:
return -1;
}
The function nf_ct_frag6_queue is called from nf_ct_frag6_gather,
that falls back to ret_orig on error:
if (nf_ct_frag6_queue(fq, clone, fhdr, nhoff) < 0) {
spin_unlock_bh(&fq->q.lock);
pr_debug("Can't insert skb to queue\n");
inet_frag_put(&fq->q, &nf_frags);
goto ret_orig;
}
where the skb clone is freed and the original skb is returned.
ret_orig:
kfree_skb(clone);
return skb;
}
In case of an overlapping fragment this seems to lead to the emission
of said fragment instead of completely and silently discarding all fragments.
I observed this behavior while replaying sequences of IP fragments with
conntrack active on Debian Wheezy (Kernel 3.2). When I double-send any fragment
the queue is dropped and the double-sent fragment is emitted by the
firewall/router.
The code I cited is the current mainline kernel code from kernel.org.</pre>
</div>
</p>
<hr>
<span>You are receiving this mail because:</span>
<ul>
<li>You are watching all bug changes.</li>
</ul>
</body>
</html>