From: Florian Westphal <fw@xxxxxxxxx>
Its a kernel implementation detail, at least at this time:
We can later decide to revert this patch if there is a compelling
reason, but then we should also remove the ifdef that prevents exposure
of ip_conntrack_status enum IPS_NAT_CLASH value in the uapi header.
Clash entries are not included in dumps (true for both old /proc
and ctnetlink) either. So for now exclude the clash bit when dumping.
Fixes: 7e5c6aa67e6f ("netfilter: nf_tables: add packets conntrack state to debug trace info")
Link: https://lore.kernel.org/netfilter-devel/aGwf3dCggwBlRKKC@xxxxxxxxx/
Signed-off-by: Florian Westphal <fw@xxxxxxxxx>
Signed-off-by: Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx>
---
net/netfilter/nf_tables_trace.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/net/netfilter/nf_tables_trace.c b/net/netfilter/nf_tables_trace.c
index ae3fe87195ab..a88abae5a9de 100644
--- a/net/netfilter/nf_tables_trace.c
+++ b/net/netfilter/nf_tables_trace.c
@@ -127,6 +127,9 @@ static int nf_trace_fill_ct_info(struct sk_buff *nlskb,
if (nla_put_be32(nlskb, NFTA_TRACE_CT_ID, (__force __be32)id))
return -1;
+ /* Kernel implementation detail, withhold this from userspace for now */
+ status &= ~IPS_NAT_CLASH;
+
if (status && nla_put_be32(nlskb, NFTA_TRACE_CT_STATUS, htonl(status)))
return -1;
}
--
2.39.5
