net: ipv4: don't allow setting net.ipv4.route.min_pmtu below 68
authorSabrina Dubroca <sd@queasysnail.net>
Mon, 26 Feb 2018 15:13:43 +0000 (16:13 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Fri, 9 Mar 2018 06:41:09 +0000 (22:41 -0800)
commit3bcf69f8e7869e00babd3cc11f8c3acca9f79cc9
tree5c78cfc6164ad159bafe637a20d00b5b4606a766
parentf80c28a585b57bfc0c33cfac2a10b0c93eede575
net: ipv4: don't allow setting net.ipv4.route.min_pmtu below 68

[ Upstream commit c7272c2f1229125f74f22dcdd59de9bbd804f1c8 ]

According to RFC 1191 sections 3 and 4, ICMP frag-needed messages
indicating an MTU below 68 should be rejected:

    A host MUST never reduce its estimate of the Path MTU below 68
    octets.

and (talking about ICMP frag-needed's Next-Hop MTU field):

    This field will never contain a value less than 68, since every
    router "must be able to forward a datagram of 68 octets without
    fragmentation".

Furthermore, by letting net.ipv4.route.min_pmtu be set to negative
values, we can end up with a very large PMTU when (-1) is cast into u32.

Let's also make ip_rt_min_pmtu a u32, since it's only ever compared to
unsigned ints.

Reported-by: Jianlin Shi <jishi@redhat.com>
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net/ipv4/route.c