[PATCH nf] netfilter: nf_tables: move hardware offload step after building the chain blob
Pablo Neira Ayuso <[email protected]>
| Newsgroups | gmane.comp.security.firewalls.netfilter.devel |
|---|---|
| Message-ID | <[email protected]> |
Allocate the chain blob before the ruleset offload to reduce chances of
entering an inconsistent state where the offloaded ruleset in the nic
and the software ruleset differ.
Fixes: c9626a2cbdb2 ("netfilter: nf_tables: add hardware offload support")
Signed-off-by: Pablo Neira Ayuso <[email protected]>
---
net/netfilter/nf_tables_api.c | 14 ++++++++++----
1 file changed, 10 insertions(+), 4 deletions(-)
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 9b6050f100d0..7c33b523f678 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -10972,10 +10972,6 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
return -EAGAIN;
}
- err = nft_flow_rule_offload_commit(net);
- if (err < 0)
- return err;
-
/* 1. Allocate space for next generation rules_gen_X[] */
list_for_each_entry_safe(trans, next, &nft_net->commit_list, list) {
struct nft_table *table = trans->table;
@@ -11000,6 +10996,16 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
}
}
+ /* must be last, so audit and chain blob set up does not leave hardware
+ * in consistent state.
+ */
+ err = nft_flow_rule_offload_commit(net);
+ if (err < 0) {
+ nf_tables_commit_chain_prepare_cancel(net);
+ nf_tables_commit_audit_free(&adl);
+ return err;
+ }
+
/* step 2. Make rules_gen_X visible to packet path */
nft_set_commit_update(&ctx, nft_net);
--
2.47.3