The origin skip check needs to re-test the zone. Else, we might skip a colliding tuple in the reply direction. This only occurs when using 'directional zones' where origin tuples reside in different zones but the reply tuples share the same zone. This causes the new conntrack entry to be dropped at confirmation time because NAT clash resolution was elided. Fixes: 4e35c1cb9460240 ("netfilter: nf_nat: skip nat clash resolution for same-origin entries") Signed-off-by: Florian Westphal <fw@xxxxxxxxx> --- I have a selftest to trigger this bug, but it depends on https://patchwork.ozlabs.org/project/netfilter-devel/patch/20210203165707.21781-4-fw@xxxxxxxxx/ and https://patchwork.ozlabs.org/project/netfilter-devel/patch/20210203165707.21781-5-fw@xxxxxxxxx/ so I will only send it once a new nft release with those patches is out. 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 234b7cab37c3..ff0168736f6e 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -1229,7 +1229,8 @@ nf_conntrack_tuple_taken(const struct nf_conntrack_tuple *tuple, * Let nf_ct_resolve_clash() deal with this later. */ if (nf_ct_tuple_equal(&ignored_conntrack->tuplehash[IP_CT_DIR_ORIGINAL].tuple, - &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple)) + &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple) && + nf_ct_zone_equal(ct, zone, IP_CT_DIR_ORIGINAL)) continue; NF_CT_STAT_INC_ATOMIC(net, found); -- 2.26.2