inet: frags: drop fraglist conntrack references
Subsystems (3)
PATCH 1/1 inet: frags: drop fraglist conntrack references
Jakub added a warning in nf_conntrack_cleanup_net_list() to make debugging leaked skbs/conntrack references more obvious.
syzbot reports this as triggering, and I can also reproduce this via ip_defrag.sh selftest:
conntrack cleanup blocked for 60s
WARNING: net/netfilter/nf_conntrack_core.c:2512 [..]
conntrack clenups gets stuck because there are skbs with still hold nf_conn references via their frag_list.
net.core.skb_defer_max=0 makes the hang disappear.
Eric Dumazet points out that skb_release_head_state() doesn't follow the fraglist.
ip_defrag.sh can only reproduce this problem since
commit 6471658dc66c ("udp: use skb_attempt_defer_free()"), but AFAICS this problem could happen with TCP as well if pmtu discovery is off.
The relevant problem path for udp is:
1. netns emits fragmented packets
2. nf_defrag_v6_hook reassembles them (in output hook)
3. reassembled skb is tracked (skb owns nf_conn reference)
4. ip6_output refragments
5. refragmented packets also own nf_conn reference (ip6_fragment calls ip6_copy_metadata())
6. on input path, nf_defrag_v6_hook skips defragmentation: the fragments already have skb->nf_conn attached
7. skbs are reassembled via ipv6_frag_rcv()
8. skb_consume_udp -> skb_attempt_defer_free() -> skb ends up in pcpu freelist, but still has nf_conn reference.
Possible solutions:
1 let defrag engine drop nf_conn entry, OR
2 export kick_defer_list_purge() and call it from the conntrack
netns exit callback, OR
3 add skb_has_frag_list() check to skb_attempt_defer_free()
2 & 3 also solve ip_defrag.sh hang but share same drawback:
Such reassembled skbs, queued to socket, can prevent conntrack module removal until userspace has consumed the packet. While both tcp and udp stack do call nf_reset_ct() before placing skb on socket queue, that function doesn't iterate frag_list skbs.
Therefore drop nf_conn entries when they are placed in defrag queue. Keep the nf_conn entry of the first (offset 0) skb so that reassembled skb retains nf_conn entry for sake of TX path.
Note that fixes tag is incorrect; it points to the commit introducing the 'ip_defrag.sh reproducible problem': no need to backport this patch to every stable kernel.
--- net/ipv4/inet_fragment.c | 2 ++ 1 file changed, 2 insertions(+)
net/ipv4/inet_fragment.c +2 -0
@@ -488,6 +488,8 @@ int inet_frag_queue_insert(struct inet_frag_queue *q, struct sk_buff *skb, }FRAG_CB (skb)->ip_defrag_offset = offset; +if (offset) +nf_reset_ct (skb);return IPFRAG_OK ; }
Signature · 1 line
2.52.0
Quoted 57 lines · Florian Westphal
Jakub added a warning in
nf_conntrack_cleanup_net_list()to make debugging leaked skbs/conntrack references more obvious.syzbot reports this as triggering, and I can also reproduce this via
ip_defrag.shselftest:conntrack cleanup blocked for 60s
WARNING: net/netfilter/nf_conntrack_core.c:2512 [..]conntrack clenups gets stuck because there are skbs with still hold
nf_connreferences via theirfrag_list.net.core.skb_defer_max=0 makes the hang disappear.
Eric Dumazet points out that
skb_release_head_state()doesn't follow the fraglist.
ip_defrag.shcan only reproduce this problem since
commit6471658dc66c("udp: useskb_attempt_defer_free()"), but AFAICS this problem could happen with TCP as well if pmtu discovery is off.The relevant problem path for udp is:
1. netns emits fragmented packets
2.nf_defrag_v6_hookreassembles them (in output hook)
3. reassembled skb is tracked (skb ownsnf_connreference)
4.ip6_outputrefragments
5. refragmented packets also ownnf_connreference (ip6_fragmentcallsip6_copy_metadata())
6. on input path,nf_defrag_v6_hookskips defragmentation: the fragments already haveskb->nf_connattached
7. skbs are reassembled viaipv6_frag_rcv()
8.skb_consume_udp->skb_attempt_defer_free()-> skb ends up in pcpu freelist, but still hasnf_connreference.Possible solutions:
1 let defrag engine dropnf_connentry, OR
2 exportkick_defer_list_purge()and call it from the conntrack
netns exit callback, OR
3 addskb_has_frag_list()check toskb_attempt_defer_free()2 & 3 also solve
ip_defrag.shhang but share same drawback:Such reassembled skbs, queued to socket, can prevent conntrack module removal until userspace has consumed the packet. While both tcp and udp stack do call
nf_reset_ct()before placing skb on socket queue, that function doesn't iteratefrag_listskbs.Therefore drop
nf_connentries when they are placed in defrag queue. Keep thenf_connentry of the first (offset 0) skb so that reassembled skb retainsnf_connentry for sake of TX path.Note that fixes tag is incorrect; it points to the commit introducing the '
ip_defrag.shreproducible problem': no need to backport this patch to every stable kernel.
Quoted 5 lines
Reported-by [email protected]Fixes 6471658dc66c ("udp: use skb_attempt_defer_free()")Signed-off-by Florian Westphal <[email protected]>---
Thanks a lot Florian for taking care of this.
This patch was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:
Hello:
This patch was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:
Quoted 11 lines · On Fri, 2 Jan 2026 15:00:07 +0100 you wrote:
Jakub added a warning in
nf_conntrack_cleanup_net_list()to make debugging leaked skbs/conntrack references more obvious.syzbot reports this as triggering, and I can also reproduce this via
ip_defrag.shselftest:conntrack cleanup blocked for 60s
WARNING: net/netfilter/nf_conntrack_core.c:2512 [..][...]
Here is the summary with links:
- [net] inet: frags: drop fraglist conntrack references
https://git.kernel.org/netdev/net/c/2ef02ac38d3c
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html