Re: [PATCH net-next] inet: defrag: drop non-last frags smaller than min mtu

[Eric Dumazet]

On 08/02/2018 04:43 PM, Florian Westphal wrote:
> don't bother with pathological cases, they only waste cycles.
> IPv6 requires a minimum MTU of 1280 so we should never see fragments
> smaller than this (except last frag).
> 
> For IPv4, in practice, we could probably also adopt a higher limit,
> but for now use ipv4 min mtu (68).

...

> + if (-skb_network_offset(skb) + skb->len < IPV4_MIN_MTU &&
> + ip_hdr(skb)->frag_off & htons(IP_MF))
> + goto drop;
> +
>

I am not totally sure this is legit for IPv4.

Some intermediate nodes can try to be smart and could decide to further split 
fragments.

I am pretty sure I have seen this behavior on some radio environments :/

Eventually we could add a sysctl to allow an admin to set the threshold ?

[PATCH net-next] inet: defrag: drop non-last frags smaller than min mtu

[Florian Westphal]

don't bother with pathological cases, they only waste cycles.
IPv6 requires a minimum MTU of 1280 so we should never see fragments
smaller than this (except last frag).

For IPv4, in practice, we could probably also adopt a higher limit,
but for now use ipv4 min mtu (68).

Cc: Peter Oskolkov 
Cc: Eric Dumazet 
Signed-off-by: Florian Westphal 
---
 net/ipv4/ip_fragment.c  | 5 +
 net/ipv6/netfilter/nf_conntrack_reasm.c | 4 
 net/ipv6/reassembly.c   | 4 
 3 files changed, 13 insertions(+)

diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index 8e9528ebaa8e..19aa10abc6ab 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -605,6 +605,10 @@ int ip_defrag(struct net *net, struct sk_buff *skb, u32 
user)
int vif = l3mdev_master_ifindex_rcu(dev);
struct ipq *qp;
 
+   if (-skb_network_offset(skb) + skb->len < IPV4_MIN_MTU &&
+   ip_hdr(skb)->frag_off & htons(IP_MF))
+   goto drop;
+
__IP_INC_STATS(net, IPSTATS_MIB_REASMREQDS);
skb_orphan(skb);
 
@@ -622,6 +626,7 @@ int ip_defrag(struct net *net, struct sk_buff *skb, u32 
user)
return ret;
}
 
+drop:
__IP_INC_STATS(net, IPSTATS_MIB_REASMFAILS);
kfree_skb(skb);
return -ENOMEM;
diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c 
b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 0610bdab721c..c121d534d321 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -557,6 +557,10 @@ int nf_ct_frag6_gather(struct net *net, struct sk_buff 
*skb, u32 user)
hdr = ipv6_hdr(skb);
fhdr = (struct frag_hdr *)skb_transport_header(skb);
 
+   if (-skb_network_offset(skb) + skb->len < IPV6_MIN_MTU &&
+   fhdr->frag_off & htons(IP6_MF))
+   return -EINVAL;
+
skb_orphan(skb);
fq = fq_find(net, fhdr->identification, user, hdr,
 skb->dev ? skb->dev->ifindex : 0);
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 6edd2ac8ae4b..ff00ada6128f 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -455,6 +455,10 @@ static int ipv6_frag_rcv(struct sk_buff *skb)
return 1;
}
 
+   if (-skb_network_offset(skb) + skb->len < IPV6_MIN_MTU &&
+   fhdr->frag_off & htons(IP6_MF))
+   goto fail_hdr;
+
iif = skb->dev ? skb->dev->ifindex : 0;
fq = fq_find(net, fhdr->identification, hdr, iif);
if (fq) {
-- 
2.16.4