From 735795f68b37e9bb49f642407a0d49b1631ea1c7 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Tue, 24 Oct 2023 21:09:47 +0200 Subject: netfilter: flowtable: GC pushes back packets to classic path Since 41f2c7c342d3 ("net/sched: act_ct: Fix promotion of offloaded unreplied tuple"), flowtable GC pushes back flows with IPS_SEEN_REPLY back to classic path in every run, ie. every second. This is because of a new check for NF_FLOW_HW_ESTABLISHED which is specific of sched/act_ct. In Netfilter's flowtable case, NF_FLOW_HW_ESTABLISHED never gets set on and IPS_SEEN_REPLY is unreliable since users decide when to offload the flow before, such bit might be set on at a later stage. Fix it by adding a custom .gc handler that sched/act_ct can use to deal with its NF_FLOW_HW_ESTABLISHED bit. Fixes: 41f2c7c342d3 ("net/sched: act_ct: Fix promotion of offloaded unreplied tuple") Reported-by: Vladimir Smelhaus Reviewed-by: Paul Blakey Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_flow_table.h | 1 + net/netfilter/nf_flow_table_core.c | 14 +++++++------- net/sched/act_ct.c | 7 +++++++ 3 files changed, 15 insertions(+), 7 deletions(-) diff --git a/include/net/netfilter/nf_flow_table.h b/include/net/netfilter/nf_flow_table.h index d466e1a3b0b1..fe1507c1db82 100644 --- a/include/net/netfilter/nf_flow_table.h +++ b/include/net/netfilter/nf_flow_table.h @@ -53,6 +53,7 @@ struct nf_flowtable_type { struct list_head list; int family; int (*init)(struct nf_flowtable *ft); + bool (*gc)(const struct flow_offload *flow); int (*setup)(struct nf_flowtable *ft, struct net_device *dev, enum flow_block_command cmd); diff --git a/net/netfilter/nf_flow_table_core.c b/net/netfilter/nf_flow_table_core.c index 1d34d700bd09..920a5a29ae1d 100644 --- a/net/netfilter/nf_flow_table_core.c +++ b/net/netfilter/nf_flow_table_core.c @@ -316,12 +316,6 @@ void flow_offload_refresh(struct nf_flowtable *flow_table, } EXPORT_SYMBOL_GPL(flow_offload_refresh); -static bool nf_flow_is_outdated(const struct flow_offload *flow) -{ - return test_bit(IPS_SEEN_REPLY_BIT, &flow->ct->status) && - !test_bit(NF_FLOW_HW_ESTABLISHED, &flow->flags); -} - static inline bool nf_flow_has_expired(const struct flow_offload *flow) { return nf_flow_timeout_delta(flow->timeout) <= 0; @@ -407,12 +401,18 @@ nf_flow_table_iterate(struct nf_flowtable *flow_table, return err; } +static bool nf_flow_custom_gc(struct nf_flowtable *flow_table, + const struct flow_offload *flow) +{ + return flow_table->type->gc && flow_table->type->gc(flow); +} + static void nf_flow_offload_gc_step(struct nf_flowtable *flow_table, struct flow_offload *flow, void *data) { if (nf_flow_has_expired(flow) || nf_ct_is_dying(flow->ct) || - nf_flow_is_outdated(flow)) + nf_flow_custom_gc(flow_table, flow)) flow_offload_teardown(flow); if (test_bit(NF_FLOW_TEARDOWN, &flow->flags)) { diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c index 7c652d14528b..0d44da4e8c8e 100644 --- a/net/sched/act_ct.c +++ b/net/sched/act_ct.c @@ -278,7 +278,14 @@ err_nat: return err; } +static bool tcf_ct_flow_is_outdated(const struct flow_offload *flow) +{ + return test_bit(IPS_SEEN_REPLY_BIT, &flow->ct->status) && + !test_bit(NF_FLOW_HW_ESTABLISHED, &flow->flags); +} + static struct nf_flowtable_type flowtable_ct = { + .gc = tcf_ct_flow_is_outdated, .action = tcf_ct_flow_table_fill_actions, .owner = THIS_MODULE, }; -- cgit v1.2.3-58-ga151 From a63b6622120cd03a304796dbccb80655b3a21798 Mon Sep 17 00:00:00 2001 From: Vlad Buslov Date: Tue, 24 Oct 2023 21:58:57 +0200 Subject: net/sched: act_ct: additional checks for outdated flows Current nf_flow_is_outdated() implementation considers any flow table flow which state diverged from its underlying CT connection status for teardown which can be problematic in the following cases: - Flow has never been offloaded to hardware in the first place either because flow table has hardware offload disabled (flag NF_FLOWTABLE_HW_OFFLOAD is not set) or because it is still pending on 'add' workqueue to be offloaded for the first time. The former is incorrect, the later generates excessive deletions and additions of flows. - Flow is already pending to be updated on the workqueue. Tearing down such flows will also generate excessive removals from the flow table, especially on highly loaded system where the latency to re-offload a flow via 'add' workqueue can be quite high. When considering a flow for teardown as outdated verify that it is both offloaded to hardware and doesn't have any pending updates. Fixes: 41f2c7c342d3 ("net/sched: act_ct: Fix promotion of offloaded unreplied tuple") Reviewed-by: Paul Blakey Signed-off-by: Vlad Buslov Signed-off-by: Pablo Neira Ayuso --- net/sched/act_ct.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/net/sched/act_ct.c b/net/sched/act_ct.c index 0d44da4e8c8e..fb52d6f9aff9 100644 --- a/net/sched/act_ct.c +++ b/net/sched/act_ct.c @@ -281,6 +281,8 @@ err_nat: static bool tcf_ct_flow_is_outdated(const struct flow_offload *flow) { return test_bit(IPS_SEEN_REPLY_BIT, &flow->ct->status) && + test_bit(IPS_HW_OFFLOAD_BIT, &flow->ct->status) && + !test_bit(NF_FLOW_HW_PENDING, &flow->flags) && !test_bit(NF_FLOW_HW_ESTABLISHED, &flow->flags); } -- cgit v1.2.3-58-ga151