Netfilter updates for net-next
Subsystems (7)
Pablo Neira Ayuso
maintainercover Netfilter updates for net-next
Hi,
The following patchset contains Netfilter updates for net-next. This includes one patch to update ovs and act_ct to use nf_ct_put() instead of nf_conntrack_put().
1) Add netns_tracker to nfnetlink_log and masquerade, from Eric Dumazet.
2) Remove redundant rcu read-size lock in nf_tables packet path.
3) Replace BUG() by WARN_ON_ONCE() in nft_payload.
4) Consolidate rule verdict tracing.
5) Replace WARN_ON() by WARN_ON_ONCE() in nf_tables core.
6) Make counter support built-in in nf_tables.
7) Add new field to conntrack object to identify locally generated traffic, from Florian Westphal.
8) Prevent NAT from shadowing well-known ports, from Florian Westphal.
9) Merge nf_flow_table_{ipv4,ipv6} into nf_flow_table_inet, also from Florian.
10) Remove redundant pointer in nft_pipapo AVX2 support, from Colin Ian King.
11) Replace opencoded max() in conntrack, from Jiapeng Chong.
12) Update conntrack to use refcount_t API, from Florian Westphal.
13) Move ip_ct_attach indirection into the nf_ct_hook structure.
14) Constify several pointer object in the netfilter codebase, from Florian Westphal.
15) Tree-wide replacement of nf_conntrack_put() by nf_ct_put(), also from Florian.
16) Fix egress splat due to incorrect rcu notation, from Florian.
17) Move stateful fields of connlimit, last, quota, numgen and limit out of the expression data area.
18) Build a blob to represent the ruleset in nf_tables, this is a requirement of the new register tracking infrastructure.
19) Add NFT_REG32_NUM to define the maximum number of 32-bit registers.
20) Add register tracking infrastructure to skip redundant store-to-register operations, this includes support for payload, meta and bitwise expresssions.
Please, pull these changes from:
git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next.git
Thanks.
----------------------------------------------------------------
The following changes since commit 604ba230902d23c6e85c7dba9cfcb6a37661cb12:
net: prestera: flower template support (2021-12-16 10:52:53 +0000)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf-next.git HEAD
for you to fetch changes up to 4a80e026981b791da3937470ace84796490c7796:
netfilter: nft_meta: cancel register tracking after meta update (2022-01-09 23:35:17 +0100)
----------------------------------------------------------------
Colin Ian King (1):
netfilter: nft_set_pipapo_avx2: remove redundant pointer lt
Eric Dumazet (2):
netfilter: nfnetlink: add netns refcount tracker to struct nfulnl_instance
netfilter: nf_nat_masquerade: add netns refcount tracker to masq_dev_work
Florian Westphal (9):
netfilter: conntrack: tag conntracks picked up in local out hook
netfilter: nat: force port remap to prevent shadowing well-known ports
netfilter: flowtable: remove ipv4/ipv6 modules
netfilter: conntrack: convert to refcount_t api
netfilter: core: move ip_ct_attach indirection to struct nf_ct_hook
netfilter: make function op structures const
netfilter: conntrack: avoid useless indirection during conntrack destruction
net: prefer nf_ct_put instead of nf_conntrack_put
netfilter: egress: avoid a lockdep splat
Jiapeng Chong (1):
netfilter: conntrack: Use max() instead of doing it manually
Pablo Neira Ayuso (19):
netfilter: nf_tables: remove rcu read-size lock
netfilter: nft_payload: WARN_ON_ONCE instead of BUG
netfilter: nf_tables: consolidate rule verdict trace call
netfilter: nf_tables: replace WARN_ON by WARN_ON_ONCE for unknown verdicts
netfilter: nf_tables: make counter support built-in
netfilter: nft_connlimit: move stateful fields out of expression data
netfilter: nft_last: move stateful fields out of expression data
netfilter: nft_quota: move stateful fields out of expression data
netfilter: nft_numgen: move stateful fields out of expression data
netfilter: nft_limit: rename stateful structure
netfilter: nft_limit: move stateful fields out of expression data
netfilter: nf_tables: add rule blob layout
netfilter: nf_tables: add NFT_REG32_NUM
netfilter: nf_tables: add register tracking infrastructure
netfilter: nft_payload: track register operations
netfilter: nft_meta: track register operations
netfilter: nft_bitwise: track register operations
netfilter: nft_payload: cancel register tracking after payload update
netfilter: nft_meta: cancel register tracking after meta update
include/linux/netfilter.h | 10 +- include/linux/netfilter/nf_conntrack_common.h | 10 +- include/linux/netfilter_netdev.h | 2 +- include/net/netfilter/nf_conntrack.h | 11 +- include/net/netfilter/nf_tables.h | 40 +++++- include/net/netfilter/nf_tables_core.h | 6 + net/bridge/netfilter/nft_meta_bridge.c | 20 +++ net/ipv4/netfilter/Kconfig | 8 +- net/ipv4/netfilter/Makefile | 3 - net/ipv4/netfilter/nf_flow_table_ipv4.c | 37 ------ net/ipv6/netfilter/Kconfig | 8 +- net/ipv6/netfilter/nf_flow_table_ipv6.c | 38 ------ net/netfilter/Kconfig | 6 - net/netfilter/Makefile | 3 +- net/netfilter/core.c | 29 ++--- net/netfilter/nf_conntrack_core.c | 53 ++++---- net/netfilter/nf_conntrack_expect.c | 4 +- net/netfilter/nf_conntrack_netlink.c | 10 +- net/netfilter/nf_conntrack_standalone.c | 4 +- net/netfilter/nf_flow_table_core.c | 2 +- net/netfilter/nf_flow_table_inet.c | 26 ++++ net/netfilter/nf_nat_core.c | 45 ++++++- net/netfilter/nf_nat_masquerade.c | 4 +- net/netfilter/nf_synproxy_core.c | 1 - net/netfilter/nf_tables_api.c | 160 +++++++++++++++++------- net/netfilter/nf_tables_core.c | 87 +++++++++---- net/netfilter/nf_tables_trace.c | 2 +- net/netfilter/nfnetlink_log.c | 5 +- net/netfilter/nfnetlink_queue.c | 8 +- net/netfilter/nft_bitwise.c | 95 ++++++++++++++ net/netfilter/nft_connlimit.c | 26 ++-- net/netfilter/nft_counter.c | 58 +++------ net/netfilter/nft_ct.c | 4 +- net/netfilter/nft_last.c | 69 ++++++++--- net/netfilter/nft_limit.c | 172 ++++++++++++++++++-------- net/netfilter/nft_meta.c | 48 +++++++ net/netfilter/nft_numgen.c | 34 ++++- net/netfilter/nft_payload.c | 57 ++++++++- net/netfilter/nft_quota.c | 52 +++++++- net/netfilter/nft_set_pipapo_avx2.c | 4 +- net/netfilter/xt_CT.c | 3 +- net/openvswitch/conntrack.c | 15 ++- net/sched/act_ct.c | 7 +- tools/testing/selftests/netfilter/nft_nat.sh | 5 +- 44 files changed, 887 insertions(+), 404 deletions(-)
Pablo Neira Ayuso
maintainerPATCH 1/32 netfilter: nfnetlink: add netns refcount tracker to struct nfulnl_instance
From: Eric Dumazet <[email protected]>
If compiled with CONFIG_NET_NS_REFCNT_TRACKER=y, using put_net_track() in nfulnl_instance_free_rcu() and get_net_track() in instance_create() might help us finding netns refcount imbalances.
--- net/netfilter/nfnetlink_log.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
net/netfilter/nfnetlink_log.c +3 -2
@@ -66,6 +66,7 @@ struct nfulnl_instance {struct sk_buff *skb;/* pre-allocatd skb */ struct timer_list timer;struct net *net; +netns_tracker ns_tracker;struct user_namespace *peer_user_ns;/* User namespace of the peer process */ u32 peer_portid;/* PORTID of the peer process */
@@ -140,7 +141,7 @@ static void nfulnl_instance_free_rcu(struct rcu_head *head)struct nfulnl_instance *inst = container_of(head, struct nfulnl_instance, rcu); -put_net (inst->net); +put_net_track (inst->net, &inst->ns_tracker);kfree (inst);module_put (THIS_MODULE ); }
@@ -187,7 +188,7 @@ instance_create(struct net *net, u_int16_t group_num,timer_setup (&inst->timer, nfulnl_timer,0 ); - inst->net =get_net (net); + inst->net =get_net_track (net, &inst->ns_tracker,GFP_ATOMIC ); inst->peer_user_ns = user_ns; inst->peer_portid = portid; inst->group_num = group_num;
Signature · 1 line
2.30.2
Re: [PATCH net-next 01/32] netfilter: nfnetlink: add netns refcount tracker to struct nfulnl_instance
This series was applied to netdev/net-next.git (master) by Pablo Neira Ayuso <[email protected]>:
Hello:
This series was applied to netdev/net-next.git (master) by Pablo Neira Ayuso <[email protected]>:
Quoted 8 lines · On Mon, 10 Jan 2022 00:16:09 +0100 you wrote:
From: Eric Dumazet <[email protected]>
If compiled with
CONFIG_NET_NS_REFCNT_TRACKER=y, usingput_net_track()innfulnl_instance_free_rcu()andget_net_track()ininstance_create()might help us finding netns refcount imbalances.[...]
Here is the summary with links:
- [net-next,01/32] netfilter: nfnetlink: add netns refcount tracker to struct nfulnl_instance
https://git.kernel.org/netdev/net-next/c/a9382d9389a0 - [net-next,02/32] netfilter: nf_nat_masquerade: add netns refcount tracker to masq_dev_work https://git.kernel.org/netdev/net-next/c/fc0d026a2fad - [net-next,03/32] netfilter: nf_tables: remove rcu read-size lock https://git.kernel.org/netdev/net-next/c/0d1873a52289 - [net-next,04/32] netfilter: nft_payload: WARN_ON_ONCE instead of BUG https://git.kernel.org/netdev/net-next/c/8801d791b487 - [net-next,05/32] netfilter: nf_tables: consolidate rule verdict trace call https://git.kernel.org/netdev/net-next/c/4765473fefd4 - [net-next,06/32] netfilter: nf_tables: replace WARN_ON by WARN_ON_ONCE for unknown verdicts https://git.kernel.org/netdev/net-next/c/690d541739a3 - [net-next,07/32] netfilter: nf_tables: make counter support built-in https://git.kernel.org/netdev/net-next/c/023223dfbfb3 - [net-next,08/32] netfilter: conntrack: tag conntracks picked up in local out hook https://git.kernel.org/netdev/net-next/c/4a6fbdd801e8 - [net-next,09/32] netfilter: nat: force port remap to prevent shadowing well-known ports https://git.kernel.org/netdev/net-next/c/878aed8db324 - [net-next,10/32] netfilter: flowtable: remove ipv4/ipv6 modules https://git.kernel.org/netdev/net-next/c/c42ba4290b21 - [net-next,11/32] netfilter: nft_set_pipapo_avx2: remove redundant pointer lt https://git.kernel.org/netdev/net-next/c/2b71e2c7b56c - [net-next,12/32] netfilter: conntrack: Use max() instead of doing it manually https://git.kernel.org/netdev/net-next/c/613a0c67d12f - [net-next,13/32] netfilter: conntrack: convert to refcount_t api https://git.kernel.org/netdev/net-next/c/719774377622 - [net-next,14/32] netfilter: core: move ip_ct_attach indirection to struct nf_ct_hook https://git.kernel.org/netdev/net-next/c/3fce16493dc1 - [net-next,15/32] netfilter: make function op structures const https://git.kernel.org/netdev/net-next/c/285c8a7a5815 - [net-next,16/32] netfilter: conntrack: avoid useless indirection during conntrack destruction https://git.kernel.org/netdev/net-next/c/6ae7989c9af0 - [net-next,17/32] net: prefer nf_ct_put instead of nf_conntrack_put https://git.kernel.org/netdev/net-next/c/408bdcfce8df - [net-next,18/32] netfilter: egress: avoid a lockdep splat https://git.kernel.org/netdev/net-next/c/6316136ec6e3 - [net-next,19/32] netfilter: nft_connlimit: move stateful fields out of expression data https://git.kernel.org/netdev/net-next/c/37f319f37d90 - [net-next,20/32] netfilter: nft_last: move stateful fields out of expression data https://git.kernel.org/netdev/net-next/c/33a24de37e81 - [net-next,21/32] netfilter: nft_quota: move stateful fields out of expression data https://git.kernel.org/netdev/net-next/c/ed0a0c60f0e5 - [net-next,22/32] netfilter: nft_numgen: move stateful fields out of expression data https://git.kernel.org/netdev/net-next/c/567882eb3d44 - [net-next,23/32] netfilter: nft_limit: rename stateful structure https://git.kernel.org/netdev/net-next/c/369b6cb5d391 - [net-next,24/32] netfilter: nft_limit: move stateful fields out of expression data https://git.kernel.org/netdev/net-next/c/3b9e2ea6c11b - [net-next,25/32] netfilter: nf_tables: add rule blob layout https://git.kernel.org/netdev/net-next/c/2c865a8a28a1 - [net-next,26/32] netfilter: nf_tables: add NFT_REG32_NUM https://git.kernel.org/netdev/net-next/c/642c8eff5c60 - [net-next,27/32] netfilter: nf_tables: add register tracking infrastructure https://git.kernel.org/netdev/net-next/c/12e4ecfa244b - [net-next,28/32] netfilter: nft_payload: track register operations https://git.kernel.org/netdev/net-next/c/a7c176bf9f0e - [net-next,29/32] netfilter: nft_meta: track register operations https://git.kernel.org/netdev/net-next/c/9b17afb2c88b - [net-next,30/32] netfilter: nft_bitwise: track register operations https://git.kernel.org/netdev/net-next/c/be5650f8f47e - [net-next,31/32] netfilter: nft_payload: cancel register tracking after payload update https://git.kernel.org/netdev/net-next/c/cc003c7ee609 - [net-next,32/32] netfilter: nft_meta: cancel register tracking after meta update https://git.kernel.org/netdev/net-next/c/4a80e026981b
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html
Pablo Neira Ayuso
maintainerPATCH 2/32 netfilter: nf_nat_masquerade: add netns refcount tracker to masq_dev_work
In reply to Pablo Neira AyusoFrom: Eric Dumazet <[email protected]>
If compiled with CONFIG_NET_NS_REFCNT_TRACKER=y, using put_net_track() in iterate_cleanup_work() and netns_tracker_alloc() in nf_nat_masq_schedule() might help us finding netns refcount imbalances.
--- net/netfilter/nf_nat_masquerade.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
net/netfilter/nf_nat_masquerade.c +3 -1
@@ -12,6 +12,7 @@struct masq_dev_work {struct work_struct work;struct net *net; +netns_tracker ns_tracker;union nf_inet_addr addr;int ifindex;int (*iter)(struct nf_conn *i,void *data);
@@ -82,7 +83,7 @@ static void iterate_cleanup_work(struct work_struct *work)nf_ct_iterate_cleanup_net (w->net, w->iter, (void *)w,0 ,0 ); -put_net (w->net); +put_net_track (w->net, &w->ns_tracker);kfree (w);atomic_dec (&masq_worker_count);module_put (THIS_MODULE );
@@ -119,6 +120,7 @@ static void nf_nat_masq_schedule(struct net *net, union nf_inet_addr *addr,INIT_WORK (&w->work, iterate_cleanup_work); w->ifindex = ifindex; w->net = net; +netns_tracker_alloc (net, &w->ns_tracker, gfp_flags); w->iter = iter;if (addr) w->addr = *addr;
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 3/32 netfilter: nf_tables: remove rcu read-size lock
In reply to Pablo Neira AyusoChain stats are updated from the Netfilter hook path which already run under rcu read-size lock section.
--- net/netfilter/nf_tables_core.c | 2 -- 1 file changed, 2 deletions(-)
net/netfilter/nf_tables_core.c +0 -2
@@ -110,7 +110,6 @@ static noinline void nft_update_chain_stats(const struct nft_chain *chain, base_chain =nft_base_chain (chain); -rcu_read_lock (); pstats =READ_ONCE (base_chain->stats);if (pstats) {local_bh_disable ();
@@ -121,7 +120,6 @@ static noinline void nft_update_chain_stats(const struct nft_chain *chain,u64_stats_update_end (&stats->syncp);local_bh_enable (); } -rcu_read_unlock (); }struct nft_jumpstack {
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 4/32 netfilter: nft_payload: WARN_ON_ONCE instead of BUG
In reply to Pablo Neira AyusoBUG() is too harsh for unknown payload base, use WARN_ON_ONCE() instead.
--- net/netfilter/nft_payload.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
net/netfilter/nft_payload.c +4 -2
@@ -157,7 +157,8 @@ void nft_payload_eval(const struct nft_expr *expr, goto err;break ;default : -BUG (); +WARN_ON_ONCE (1 ); + goto err; } offset += priv->offset;
@@ -664,7 +665,8 @@ static void nft_payload_set_eval(const struct nft_expr *expr, goto err;break ;default : -BUG (); +WARN_ON_ONCE (1 ); + goto err; } csum_offset = offset + priv->csum_offset;
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 5/32 netfilter: nf_tables: consolidate rule verdict trace call
In reply to Pablo Neira AyusoAdd function to consolidate verdict tracing.
--- net/netfilter/nf_tables_core.c | 39 ++++++++++++++++++++++++++++------ 1 file changed, 32 insertions(+), 7 deletions(-)
net/netfilter/nf_tables_core.c +32 -7
@@ -67,6 +67,36 @@ static void nft_cmp_fast_eval(const struct nft_expr *expr, regs->verdict.code =NFT_BREAK ; } +static noinline void __nft_trace_verdict(struct nft_traceinfo *info, +const struct nft_chain *chain, +const struct nft_regs *regs) +{ +enum nft_trace_types type; + +switch (regs->verdict.code) { +case NFT_CONTINUE : +case NFT_RETURN : + type =NFT_TRACETYPE_RETURN ; +break ; +default : + type =NFT_TRACETYPE_RULE ; +break ; + } + +__nft_trace_packet (info, chain, type); +} + +static inline void nft_trace_verdict (struct nft_traceinfo *info, +const struct nft_chain *chain, +const struct nft_rule *rule, +const struct nft_regs *regs) +{ +if (static_branch_unlikely (&nft_trace_enabled)) { + info->rule = rule; +__nft_trace_verdict (info, chain, regs); + } +} +static bool nft_payload_fast_eval (const struct nft_expr *expr,struct nft_regs *regs,const struct nft_pktinfo *pkt)
@@ -205,13 +235,13 @@ nft_do_chain(struct nft_pktinfo *pkt, void *priv)break ; } +nft_trace_verdict (&info, chain, rule, ®s); +switch (regs.verdict.code &NF_VERDICT_MASK ) {case NF_ACCEPT :case NF_DROP :case NF_QUEUE :case NF_STOLEN : -nft_trace_packet (&info, chain, rule, -NFT_TRACETYPE_RULE );return regs.verdict.code; }
@@ -224,15 +254,10 @@ nft_do_chain(struct nft_pktinfo *pkt, void *priv) stackptr++; fallthrough;case NFT_GOTO : -nft_trace_packet (&info, chain, rule, -NFT_TRACETYPE_RULE ); - chain = regs.verdict.chain; goto do_chain;case NFT_CONTINUE :case NFT_RETURN : -nft_trace_packet (&info, chain, rule, -NFT_TRACETYPE_RETURN );break ;default :WARN_ON (1 );
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 6/32 netfilter: nf_tables: replace WARN_ON by WARN_ON_ONCE for unknown verdicts
In reply to Pablo Neira AyusoBug might trigger warning for each packet, call WARN_ON_ONCE instead.
--- net/netfilter/nf_tables_core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
net/netfilter/nf_tables_core.c +1 -1
@@ -260,7 +260,7 @@ nft_do_chain(struct nft_pktinfo *pkt, void *priv)case NFT_RETURN :break ;default : -WARN_ON (1 ); +WARN_ON_ONCE (1 ); }if (stackptr >0 ) {
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 7/32 netfilter: nf_tables: make counter support built-in
In reply to Pablo Neira AyusoMake counter support built-in to allow for direct call in case of CONFIG_RETPOLINE.
--- include/net/netfilter/nf_tables_core.h | 6 +++ net/netfilter/Kconfig | 6 --- net/netfilter/Makefile | 3 +- net/netfilter/nf_tables_core.c | 5 +++ net/netfilter/nft_counter.c | 58 +++++++------------------- 5 files changed, 27 insertions(+), 51 deletions(-)
include/net/netfilter/nf_tables_core.h +6 -0
@@ -7,6 +7,7 @@extern struct nft_expr_type nft_imm_type;extern struct nft_expr_type nft_cmp_type; +extern struct nft_expr_type nft_counter_type;extern struct nft_expr_type nft_lookup_type;extern struct nft_expr_type nft_bitwise_type;extern struct nft_expr_type nft_byteorder_type;
@@ -21,6 +22,7 @@ extern struct nft_expr_type nft_last_type;#ifdef CONFIG_NETWORK_SECMARK extern struct nft_object_type nft_secmark_obj_type;#endif +extern struct nft_object_type nft_counter_obj_type;int nf_tables_core_module_init (void );void nf_tables_core_module_exit (void );
@@ -120,6 +122,8 @@ bool nft_pipapo_lookup(const struct net *net, const struct nft_set *set,bool nft_pipapo_avx2_lookup (const struct net *net,const struct nft_set *set,const u32 *key,const struct nft_set_ext **ext); +void nft_counter_init_seqcount (void ); +struct nft_expr ;struct nft_regs ;struct nft_pktinfo ;
@@ -143,4 +147,6 @@ void nft_dynset_eval(const struct nft_expr *expr, struct nft_regs *regs, const struct nft_pktinfo *pkt);void nft_rt_get_eval (const struct nft_expr *expr,struct nft_regs *regs,const struct nft_pktinfo *pkt); +void nft_counter_eval (const struct nft_expr *expr,struct nft_regs *regs, +const struct nft_pktinfo *pkt);#endif /* _NET_NF_TABLES_CORE_H */
net/netfilter/Kconfig +0 -6
@@ -515,12 +515,6 @@ config NFT_FLOW_OFFLOAD This option adds the "flow_offload" expression that you can use to choose what flows are placed into the hardware. -config NFT_COUNTER - tristate "Netfilter nf_tables counter module" - help - This option adds the "counter" expression that you can use to - include packet and byte counters in a rule. - config NFT_CONNLIMIT tristate "Netfilter nf_tables connlimit module" depends on NF_CONNTRACK
net/netfilter/Makefile +1 -2
@@ -75,7 +75,7 @@ nf_tables-objs := nf_tables_core.o nf_tables_api.o nft_chain_filter.o \ nf_tables_trace.o nft_immediate.o nft_cmp.o nft_range.o \ nft_bitwise.o nft_byteorder.o nft_payload.o nft_lookup.o \ nft_dynset.o nft_meta.o nft_rt.o nft_exthdr.o nft_last.o \ - nft_chain_route.o nf_tables_offload.o \ + nft_counter.o nft_chain_route.o nf_tables_offload.o \ nft_set_hash.o nft_set_bitmap.o nft_set_rbtree.o \ nft_set_pipapo.o
@@ -100,7 +100,6 @@ obj-$(CONFIG_NFT_REJECT) += nft_reject.o obj-$(CONFIG_NFT_REJECT_INET) += nft_reject_inet.o obj-$(CONFIG_NFT_REJECT_NETDEV) += nft_reject_netdev.o obj-$(CONFIG_NFT_TUNNEL) += nft_tunnel.o -obj-$(CONFIG_NFT_COUNTER) += nft_counter.o obj-$(CONFIG_NFT_LOG) += nft_log.o obj-$(CONFIG_NFT_MASQ) += nft_masq.o obj-$(CONFIG_NFT_REDIR) += nft_redir.o
net/netfilter/nf_tables_core.c +5 -0
@@ -169,6 +169,7 @@ static void expr_call_ops_eval(const struct nft_expr *expr,X (e, nft_payload_eval);X (e, nft_cmp_eval); +X (e, nft_counter_eval);X (e, nft_meta_get_eval);X (e, nft_lookup_eval);X (e, nft_range_eval);
@@ -292,18 +293,22 @@ static struct nft_expr_type *nft_basic_types[] = { &nft_rt_type, &nft_exthdr_type, &nft_last_type, + &nft_counter_type, };static struct nft_object_type *nft_basic_objects[] = { #ifdef CONFIG_NETWORK_SECMARK &nft_secmark_obj_type, #endif + &nft_counter_obj_type, };int __init nf_tables_core_module_init (void ) {int err, i, j =0 ; +nft_counter_init_seqcount (); +for (i =0 ; i <ARRAY_SIZE (nft_basic_objects); i++) { err =nft_register_obj (nft_basic_objects[i]);if (err)
net/netfilter/nft_counter.c +15 -43
97 lines not on this page · raw message
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 8/32 netfilter: conntrack: tag conntracks picked up in local out hook
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
This allows to identify flows that originate from local machine in a followup patch.
It would be possible to make this a ->status bit instead. For now I did not do that yet because I don't have a use-case for exposing this info to userspace.
If one comes up the toggle can be replaced with a status bit.
--- include/net/netfilter/nf_conntrack.h | 1 + net/netfilter/nf_conntrack_core.c | 3 +++ 2 files changed, 4 insertions(+)
include/net/netfilter/nf_conntrack.h +1 -0
@@ -95,6 +95,7 @@ struct nf_conn {unsigned long status;u16 cpu; + u16 local_origin:1;possible_net_t ct_net;#if IS_ENABLED (CONFIG_NF_NAT )
net/netfilter/nf_conntrack_core.c +3 -0
@@ -1747,6 +1747,9 @@ resolve_normal_ct(struct nf_conn *tmpl,return 0 ;if (IS_ERR (h))return PTR_ERR (h); + + ct =nf_ct_tuplehash_to_ctrack (h); + ct->local_origin = state->hook ==NF_INET_LOCAL_OUT ; } ct =nf_ct_tuplehash_to_ctrack (h);
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 9/32 netfilter: nat: force port remap to prevent shadowing well-known ports
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
If destination port is above 32k and source port below 16k assume this might cause 'port shadowing' where a 'new' inbound connection matches an existing one, e.g.
inbound X:41234 -> Y:53 matches existing conntrack entry
Z:53 -> X:4123, where Z got natted to X.
In this case, new packet is natted to Z:53 which is likely unwanted.
We avoid the rewrite for connections that originate from local host:
port-shadowing is only possible with forwarded connections.
Also adjust test case.
v3: no need to call tuple_force_port_remap if already in random mode (Phil)
--- net/netfilter/nf_nat_core.c | 43 ++++++++++++++++++-- tools/testing/selftests/netfilter/nft_nat.sh | 5 ++- 2 files changed, 43 insertions(+), 5 deletions(-)
net/netfilter/nf_nat_core.c +40 -3
@@ -494,6 +494,38 @@ static void nf_nat_l4proto_unique_tuple(struct nf_conntrack_tuple *tuple, goto another_round; } +static bool tuple_force_port_remap (const struct nf_conntrack_tuple *tuple) +{ +u16 sp, dp; + +switch (tuple->dst.protonum) { +case IPPROTO_TCP : + sp =ntohs (tuple->src.u.tcp.port); + dp =ntohs (tuple->dst.u.tcp.port); +break ; +case IPPROTO_UDP : +case IPPROTO_UDPLITE : + sp =ntohs (tuple->src.u.udp.port); + dp =ntohs (tuple->dst.u.udp.port); +break ; +default : +return false; + } + +/* IANA: System port range: 1-1023, +* user port range: 1024-49151, +* private port range: 49152-65535. +* +* Linux default ephemeral port range is 32768-60999. +* +* Enforce port remapping if sport is significantly lower +* than dport to prevent NAT port shadowing, i.e. +* accidental match of 'new' inbound connection vs. +* existing outbound one. +*/ +return sp <16384 && dp >=32768 ; +} +/* Manipulate the tuple into the range given. For NF_INET_POST_ROUTING, * we change the source to map into the range. For NF_INET_PRE_ROUTING * and NF_INET_LOCAL_OUT, we change the destination to map into the
@@ -507,11 +539,17 @@ get_unique_tuple(struct nf_conntrack_tuple *tuple,struct nf_conn *ct, enum nf_nat_manip_type maniptype) { + bool random_port = range->flags & NF_NAT_RANGE_PROTO_RANDOM_ALL;const struct nf_conntrack_zone *zone;struct net *net =nf_ct_net (ct); zone =nf_ct_zone (ct); +if (maniptype ==NF_NAT_MANIP_SRC && + !random_port && + !ct->local_origin) + random_port =tuple_force_port_remap (orig_tuple); +/* 1) If this srcip/proto/src-proto-part is currently mapped, * and that same mapping gives a unique tuple within the given * range, use that.
@@ -520,8 +558,7 @@ get_unique_tuple(struct nf_conntrack_tuple *tuple,* So far, we don't do local source mappings, so multiple * manips not an issue. */ -if (maniptype ==NF_NAT_MANIP_SRC && - !(range->flags &NF_NAT_RANGE_PROTO_RANDOM_ALL )) { +if (maniptype ==NF_NAT_MANIP_SRC && !random_port) {/* try the original tuple first */ if (in_range (orig_tuple, range)) {if (!nf_nat_used_tuple (orig_tuple, ct)) {
@@ -545,7 +582,7 @@ get_unique_tuple(struct nf_conntrack_tuple *tuple,*/ /* Only bother mapping if it's not already in range and unique */ - if (!(range->flags & NF_NAT_RANGE_PROTO_RANDOM_ALL)) { + if (!random_port) { if (range->flags & NF_NAT_RANGE_PROTO_SPECIFIED) { if (!(range->flags & NF_NAT_RANGE_PROTO_OFFSET) && l4proto_in_range(tuple, maniptype,
tools/testing/selftests/netfilter/nft_nat.sh +3 -2
@@ -880,8 +880,9 @@ EOF return $ksft_skip fi - # test default behaviour. Packet from ns1 to ns0 is redirected to ns2. - test_port_shadow "default" "CLIENT" + # test default behaviour. Packet from ns1 to ns0 is not redirected + # due to automatic port translation. + test_port_shadow "default" "ROUTER" # test packet filter based mitigation: prevent forwarding of # packets claiming to come from the service port.
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 10/32 netfilter: flowtable: remove ipv4/ipv6 modules
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
Just place the structs and registration in the inet module. nf_flow_table_ipv6, nf_flow_table_ipv4 and nf_flow_table_inet share same module dependencies: nf_flow_table, nf_tables.
before:
text data bss dec hex filename 2278 1480 0 3758 eae nf_flow_table_inet.ko 1159 1352 0 2511 9cf nf_flow_table_ipv6.ko 1154 1352 0 2506 9ca nf_flow_table_ipv4.ko
after:
2369 1672 0 4041 fc9 nf_flow_table_inet.ko
--- net/ipv4/netfilter/Kconfig | 8 ++---- net/ipv4/netfilter/Makefile | 3 -- net/ipv4/netfilter/nf_flow_table_ipv4.c | 37 ------------------------ net/ipv6/netfilter/Kconfig | 8 ++---- net/ipv6/netfilter/nf_flow_table_ipv6.c | 38 ------------------------- net/netfilter/nf_flow_table_inet.c | 26 +++++++++++++++++ 6 files changed, 30 insertions(+), 90 deletions(-)
net/ipv4/netfilter/Kconfig +2 -6
@@ -59,12 +59,8 @@ config NF_TABLES_ARP endif # NF_TABLES config NF_FLOW_TABLE_IPV4 - tristate "Netfilter flow table IPv4 module" - depends on NF_FLOW_TABLE - help - This option adds the flow table IPv4 support. - - To compile it as a module, choose M here. + tristate + select NF_FLOW_TABLE_INET config NF_DUP_IPV4 tristate "Netfilter IPv4 packet duplication to alternate destination"
net/ipv4/netfilter/Makefile +0 -3
@@ -24,9 +24,6 @@ obj-$(CONFIG_NFT_REJECT_IPV4) += nft_reject_ipv4.o obj-$(CONFIG_NFT_FIB_IPV4) += nft_fib_ipv4.o obj-$(CONFIG_NFT_DUP_IPV4) += nft_dup_ipv4.o -# flow table support -obj-$(CONFIG_NF_FLOW_TABLE_IPV4) += nf_flow_table_ipv4.o - # generic IP tables obj-$(CONFIG_IP_NF_IPTABLES) += ip_tables.o
net/ipv4/netfilter/nf_flow_table_ipv4.c +0 -37
@@ -1,37 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0-only -#include <linux/kernel.h> -#include <linux/init.h> -#include <linux/module.h> -#include <linux/netfilter.h> -#include <net/netfilter/nf_flow_table.h> -#include <net/netfilter/nf_tables.h> - -static struct nf_flowtable_type flowtable_ipv4 = { - .family =NFPROTO_IPV4 , - .init = nf_flow_table_init, - .setup = nf_flow_table_offload_setup, - .action = nf_flow_rule_route_ipv4, - .free = nf_flow_table_free, - .hook = nf_flow_offload_ip_hook, - .owner =THIS_MODULE , -}; - -static int __init nf_flow_ipv4_module_init (void ) -{ -nft_register_flowtable_type (&flowtable_ipv4); - -return 0 ; -} - -static void __exit nf_flow_ipv4_module_exit(void) -{ -nft_unregister_flowtable_type (&flowtable_ipv4); -} - -module_init (nf_flow_ipv4_module_init); -module_exit (nf_flow_ipv4_module_exit); - -MODULE_LICENSE ("GPL" ); -MODULE_AUTHOR ("Pablo Neira Ayuso <[email protected]>" ); -MODULE_ALIAS_NF_FLOWTABLE (AF_INET ); -MODULE_DESCRIPTION ("Netfilter flow table support" );
net/ipv6/netfilter/Kconfig +2 -6
@@ -48,12 +48,8 @@ endif # NF_TABLES_IPV6 endif # NF_TABLES config NF_FLOW_TABLE_IPV6 - tristate "Netfilter flow table IPv6 module" - depends on NF_FLOW_TABLE - help - This option adds the flow table IPv6 support. - - To compile it as a module, choose M here. + tristate + select NF_FLOW_TABLE_INET config NF_DUP_IPV6 tristate "Netfilter IPv6 packet duplication to alternate destination"
net/ipv6/netfilter/nf_flow_table_ipv6.c +0 -38
@@ -1,38 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0-only -#include <linux/kernel.h> -#include <linux/init.h> -#include <linux/module.h> -#include <linux/netfilter.h> -#include <linux/rhashtable.h> -#include <net/netfilter/nf_flow_table.h> -#include <net/netfilter/nf_tables.h> - -static struct nf_flowtable_type flowtable_ipv6 = { - .family =NFPROTO_IPV6 , - .init = nf_flow_table_init, - .setup = nf_flow_table_offload_setup, - .action = nf_flow_rule_route_ipv6, - .free = nf_flow_table_free, - .hook = nf_flow_offload_ipv6_hook, - .owner =THIS_MODULE , -}; - -static int __init nf_flow_ipv6_module_init (void ) -{ -nft_register_flowtable_type (&flowtable_ipv6); - -return 0 ; -} - -static void __exit nf_flow_ipv6_module_exit(void) -{ -nft_unregister_flowtable_type (&flowtable_ipv6); -} - -module_init (nf_flow_ipv6_module_init); -module_exit (nf_flow_ipv6_module_exit); - -MODULE_LICENSE ("GPL" ); -MODULE_AUTHOR ("Pablo Neira Ayuso <[email protected]>" ); -MODULE_ALIAS_NF_FLOWTABLE (AF_INET6 ); -MODULE_DESCRIPTION ("Netfilter flow table IPv6 module" );
net/netfilter/nf_flow_table_inet.c +26 -0
@@ -54,8 +54,30 @@ static struct nf_flowtable_type flowtable_inet = { .owner = THIS_MODULE, }; +static struct nf_flowtable_type flowtable_ipv4 = { + .family =NFPROTO_IPV4 , + .init = nf_flow_table_init, + .setup = nf_flow_table_offload_setup, + .action = nf_flow_rule_route_ipv4, + .free = nf_flow_table_free, + .hook = nf_flow_offload_ip_hook, + .owner =THIS_MODULE , +}; + +static struct nf_flowtable_type flowtable_ipv6 = { + .family =NFPROTO_IPV6 , + .init = nf_flow_table_init, + .setup = nf_flow_table_offload_setup, + .action = nf_flow_rule_route_ipv6, + .free = nf_flow_table_free, + .hook = nf_flow_offload_ipv6_hook, + .owner =THIS_MODULE , +}; +static int __init nf_flow_inet_module_init (void ) { +nft_register_flowtable_type (&flowtable_ipv4); +nft_register_flowtable_type (&flowtable_ipv6);nft_register_flowtable_type (&flowtable_inet);return 0 ;
@@ -64,6 +86,8 @@ static int __init nf_flow_inet_module_init(void) static void __exit nf_flow_inet_module_exit(void) {nft_unregister_flowtable_type (&flowtable_inet); +nft_unregister_flowtable_type (&flowtable_ipv6); +nft_unregister_flowtable_type (&flowtable_ipv4); }module_init (nf_flow_inet_module_init);
@@ -71,5 +95,7 @@ module_exit(nf_flow_inet_module_exit);MODULE_LICENSE ("GPL" );MODULE_AUTHOR ("Pablo Neira Ayuso <[email protected]>" ); +MODULE_ALIAS_NF_FLOWTABLE (AF_INET ); +MODULE_ALIAS_NF_FLOWTABLE (AF_INET6 );MODULE_ALIAS_NF_FLOWTABLE (1 );/* NFPROTO_INET */ MODULE_DESCRIPTION ("Netfilter flow table mixed IPv4/IPv6 module" );
Signature · 1 line
2.30.2
Re: [PATCH net-next 10/32] netfilter: flowtable: remove ipv4/ipv6 modules
Hi Pablo, Florian,
Quoted 12 lines · Pablo Neira Ayuso
From: Florian Westphal <[email protected]>
Just place the structs and registration in the inet module.
nf_flow_table_ipv6,nf_flow_table_ipv4andnf_flow_table_inetshare same module dependencies:nf_flow_table,nf_tables.before:
text data bss dec hex filename 2278 1480 0 3758 eae nf_flow_table_inet.ko 1159 1352 0 2511 9cf nf_flow_table_ipv6.ko 1154 1352 0 2506 9ca nf_flow_table_ipv4.ko
Quoted 5 lines
after:
2369 1672 0 4041 fc9 nf_flow_table_inet.koSigned-off-by Florian Westphal <[email protected]>Signed-off-by Pablo Neira Ayuso <[email protected]>
Thanks for your patch, which is now commit c42ba4290b2147aa ("netfilter: flowtable: remove ipv4/ipv6 modules") upstream.
Quoted 6 lines
net/ipv4/netfilter/Kconfig +0
-0@@ -59,12 +59,8 @@ config NF_TABLES_ARPendif #
NF_TABLESconfig
NF_FLOW_TABLE_IPV4
net/ipv4/netfilter/Kconfig config NF_TABLES_ARP · line 63
Quoted 8 lines
net/ipv4/netfilter/Kconfig +2
-6- tristate "Netfilter flow table IPv4 module" - depends on NF_FLOW_TABLE - help - This option adds the flow table IPv4 support. - - To compile it as a module, choose M here. + tristate + select NF_FLOW_TABLE_INET
What is the point in keeping this symbol? It is invisble, selected by nothing (so it can no longer be enabled), and its last user is removed below.
Is there a mistake, or should this symbol just be removed?
Quoted 21 lines
config
NF_DUP_IPV4
tristate "Netfilter IPv4 packet duplication to alternate destination"net/ipv4/netfilter/Makefile +0
-3@@ -24,9 +24,6 @@ obj-$(CONFIG_NFT_REJECT_IPV4) += nft_reject_ipv4.o obj-$(CONFIG_NFT_FIB_IPV4) += nft_fib_ipv4.o obj-$(CONFIG_NFT_DUP_IPV4) += nft_dup_ipv4.o -# flow table support -obj-$(CONFIG_NF_FLOW_TABLE_IPV4) += nf_flow_table_ipv4.o - # generic IP tables obj-$(CONFIG_IP_NF_IPTABLES) += ip_tables.onet/ipv6/netfilter/Kconfig +0
-0@@ -48,12 +48,8 @@ endif # NF_TABLES_IPV6endif #
NF_TABLESconfig
NF_FLOW_TABLE_IPV6
net/ipv6/netfilter/Kconfig endif # NF_TABLES_IPV6 · line 52
Quoted 8 lines
net/ipv6/netfilter/Kconfig +2
-6- tristate "Netfilter flow table IPv6 module" - depends on NF_FLOW_TABLE - help - This option adds the flow table IPv6 support. - - To compile it as a module, choose M here. + tristate + select NF_FLOW_TABLE_INET
Likewise, except that its last user was not removed:
net/ipv6/netfilter/Makefile:obj-$(CONFIG_NF_FLOW_TABLE_IPV6) += nf_flow_table_ipv6.o
Quoted 2 lines
config
NF_DUP_IPV6
tristate "Netfilter IPv6 packet duplication to alternate destination"
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]
In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Pablo Neira Ayuso
maintainerPATCH 11/32 netfilter: nft_set_pipapo_avx2: remove redundant pointer lt
In reply to Pablo Neira AyusoFrom: Colin Ian King <[email protected]>
The pointer lt is being assigned a value and then later updated but that value is never read. The pointer is redundant and can be removed.
--- net/netfilter/nft_set_pipapo_avx2.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-)
net/netfilter/nft_set_pipapo_avx2.c +1 -3
@@ -1048,11 +1048,9 @@ static int nft_pipapo_avx2_lookup_slow(unsigned long *map, unsigned long *fill, struct nft_pipapo_field *f, int offset, const u8 *pkt, bool first, bool last) { - unsigned long *lt = f->lt, bsize = f->bsize; + unsigned long bsize = f->bsize;int i, ret =-1 , b; - lt += offset *NFT_PIPAPO_LONGS_PER_M256 ; -if (first)memset (map,0xff , bsize *sizeof (*map));
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 12/32 netfilter: conntrack: Use max() instead of doing it manually
In reply to Pablo Neira AyusoFrom: Jiapeng Chong <[email protected]>
Fix following coccicheck warning:
./include/net/netfilter/nf_conntrack.h:282:16-17: WARNING opportunity for max().
--- include/net/netfilter/nf_conntrack.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
include/net/netfilter/nf_conntrack.h +1 -1
@@ -279,7 +279,7 @@ static inline unsigned long nf_ct_expires(const struct nf_conn *ct) {s32 timeout =READ_ONCE (ct->timeout) - nfct_time_stamp; -return timeout >0 ? timeout :0 ; +return max (timeout,0 ); }static inline bool nf_ct_is_expired (const struct nf_conn *ct)
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 13/32 netfilter: conntrack: convert to refcount_t api
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
Convert nf_conn reference counting from atomic_t to refcount_t based api. refcount_t api provides more runtime sanity checks and will warn on certain constructs, e.g. refcount_inc() on a zero reference count, which usually indicates use-after-free.
For this reason template allocation is changed to init the refcount to 1, the subsequenct add operations are removed.
Likewise, init_conntrack() is changed to set the initial refcount to 1 instead refcount_inc().
This is safe because the new entry is not (yet) visible to other cpus.
--- include/linux/netfilter/nf_conntrack_common.h | 8 +++--- net/netfilter/nf_conntrack_core.c | 26 +++++++++---------- net/netfilter/nf_conntrack_expect.c | 4 +-- net/netfilter/nf_conntrack_netlink.c | 6 ++--- net/netfilter/nf_conntrack_standalone.c | 4 +-- net/netfilter/nf_flow_table_core.c | 2 +- net/netfilter/nf_synproxy_core.c | 1 - net/netfilter/nft_ct.c | 4 +-- net/netfilter/xt_CT.c | 3 +-- net/openvswitch/conntrack.c | 1 - net/sched/act_ct.c | 1 - 11 files changed, 27 insertions(+), 33 deletions(-)
include/linux/netfilter/nf_conntrack_common.h +4 -4
@@ -2,7 +2,7 @@#ifndef _NF_CONNTRACK_COMMON_H#define _NF_CONNTRACK_COMMON_H -#include <linux/atomic.h> +#include <linux/refcount.h> #include <uapi/linux/netfilter/nf_conntrack_common.h> struct ip_conntrack_stat {
@@ -25,19 +25,19 @@ struct ip_conntrack_stat {#define NFCT_PTRMASK ~(NFCT_INFOMASK)struct nf_conntrack { -atomic_t use; +refcount_t use; };void nf_conntrack_destroy (struct nf_conntrack *nfct);static inline void nf_conntrack_put (struct nf_conntrack *nfct) { -if (nfct &&atomic_dec_and_test (&nfct->use)) +if (nfct &&refcount_dec_and_test (&nfct->use))nf_conntrack_destroy (nfct); }static inline void nf_conntrack_get (struct nf_conntrack *nfct) {if (nfct) -atomic_inc (&nfct->use); +refcount_inc (&nfct->use); }#endif /* _NF_CONNTRACK_COMMON_H */
net/netfilter/nf_conntrack_core.c +13 -13
@@ -585,7 +585,7 @@ struct nf_conn *nf_ct_tmpl_alloc(struct net *net, tmpl->status =IPS_TEMPLATE ;write_pnet (&tmpl->ct_net, net);nf_ct_zone_add (tmpl, zone); -atomic_set (&tmpl->ct_general.use,0 ); +refcount_set (&tmpl->ct_general.use,1 );return tmpl; }
@@ -618,7 +618,7 @@ destroy_conntrack(struct nf_conntrack *nfct)struct nf_conn *ct = (struct nf_conn *)nfct;pr_debug ("destroy_conntrack(%p)\n" , ct); -WARN_ON (atomic_read (&nfct->use) !=0 ); +WARN_ON (refcount_read (&nfct->use) !=0 );if (unlikely (nf_ct_is_template (ct))) {nf_ct_tmpl_free (ct);
@@ -742,7 +742,7 @@ nf_ct_match(const struct nf_conn *ct1, const struct nf_conn *ct2)/* caller must hold rcu readlock and none of the nf_conntrack_locks */ static void nf_ct_gc_expired (struct nf_conn *ct) { -if (!atomic_inc_not_zero (&ct->ct_general.use)) +if (!refcount_inc_not_zero (&ct->ct_general.use))return ;if (nf_ct_should_gc (ct))
@@ -810,7 +810,7 @@ __nf_conntrack_find_get(struct net *net, const struct nf_conntrack_zone *zone,* in, try to obtain a reference and re-check tuple */ ct =nf_ct_tuplehash_to_ctrack (h); -if (likely (atomic_inc_not_zero (&ct->ct_general.use))) { +if (likely (refcount_inc_not_zero (&ct->ct_general.use))) {if (likely (nf_ct_key_equal (h, tuple, zone, net))) goto found;
@@ -907,7 +907,7 @@ nf_conntrack_hash_check_insert(struct nf_conn *ct)smp_wmb ();/* The caller holds a reference to this object */ -atomic_set (&ct->ct_general.use,2 ); +refcount_set (&ct->ct_general.use,2 );__nf_conntrack_hash_insert (ct, hash, reply_hash);nf_conntrack_double_unlock (hash, reply_hash);NF_CT_STAT_INC (net, insert);
@@ -958,7 +958,7 @@ static void __nf_conntrack_insert_prepare(struct nf_conn *ct) {struct nf_conn_tstamp *tstamp; -atomic_inc (&ct->ct_general.use); +refcount_inc (&ct->ct_general.use); ct->status |=IPS_CONFIRMED ;/* set conntrack timestamp, if enabled. */
@@ -1351,7 +1351,7 @@ static unsigned int early_drop_list(struct net *net, nf_ct_is_dying(tmp))continue ; -if (!atomic_inc_not_zero (&tmp->ct_general.use)) +if (!refcount_inc_not_zero (&tmp->ct_general.use))continue ;/* kill only if still in same netns -- might have moved due to
@@ -1469,7 +1469,7 @@ static void gc_worker(struct work_struct *work)continue ;/* need to take reference to avoid possible races */ -if (!atomic_inc_not_zero (&tmp->ct_general.use)) +if (!refcount_inc_not_zero (&tmp->ct_general.use))continue ;if (gc_worker_skip_ct (tmp)) {
@@ -1569,7 +1569,7 @@ __nf_conntrack_alloc(struct net *net,/* Because we use RCU lookups, we set ct_general.use to zero before * this is inserted in any list. */ -atomic_set (&ct->ct_general.use,0 ); +refcount_set (&ct->ct_general.use,0 );return ct; out:atomic_dec (&cnet->count);
@@ -1594,7 +1594,7 @@ void nf_conntrack_free(struct nf_conn *ct)/* A freed object has refcnt == 0, that's * the golden rule for SLAB_TYPESAFE_BY_RCU */ -WARN_ON (atomic_read (&ct->ct_general.use) !=0 ); +WARN_ON (refcount_read (&ct->ct_general.use) !=0 );nf_ct_ext_destroy (ct);kmem_cache_free (nf_conntrack_cachep, ct);
@@ -1686,8 +1686,8 @@ init_conntrack(struct net *net, struct nf_conn *tmpl,if (!exp)__nf_ct_try_assign_helper (ct, tmpl,GFP_ATOMIC ); -/* Now it is inserted into the unconfirmed list, bump refcount */ -nf_conntrack_get (&ct->ct_general); +/* Now it is inserted into the unconfirmed list, set refcount to 1. */ +refcount_set (&ct->ct_general.use,1 );nf_ct_add_to_unconfirmed_list (ct);local_bh_enable ();
@@ -2300,7 +2300,7 @@ get_next_corpse(int (*iter)(struct nf_conn *i, void *data),return NULL ; found: -atomic_inc (&ct->ct_general.use); +refcount_inc (&ct->ct_general.use);spin_unlock (lockp);local_bh_enable ();return ct;
net/netfilter/nf_conntrack_expect.c +2 -2
@@ -203,12 +203,12 @@ nf_ct_find_expectation(struct net *net,* about to invoke ->destroy(), or nf_ct_delete() via timeout * or early_drop(). * -* The atomic_inc_not_zero() check tells: If that fails, we +* The refcount_inc_not_zero() check tells: If that fails, we * know that the ct is being destroyed. If it succeeds, we * can be sure the ct cannot disappear underneath. */ if (unlikely (nf_ct_is_dying (exp->master) || - !atomic_inc_not_zero (&exp->master->ct_general.use))) + !refcount_inc_not_zero (&exp->master->ct_general.use)))return NULL ;if (exp->flags &NF_CT_EXPECT_PERMANENT ) {
net/netfilter/nf_conntrack_netlink.c +3 -3
@@ -508,7 +508,7 @@ static int ctnetlink_dump_id(struct sk_buff *skb, const struct nf_conn *ct)static int ctnetlink_dump_use (struct sk_buff *skb,const struct nf_conn *ct) { -if (nla_put_be32 (skb,CTA_USE ,htonl (atomic_read (&ct->ct_general.use)))) +if (nla_put_be32 (skb,CTA_USE ,htonl (refcount_read (&ct->ct_general.use)))) goto nla_put_failure;return 0 ;
@@ -1200,7 +1200,7 @@ ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb) ct =nf_ct_tuplehash_to_ctrack (h);if (nf_ct_is_expired (ct)) {if (i <ARRAY_SIZE (nf_ct_evict) && -atomic_inc_not_zero (&ct->ct_general.use)) +refcount_inc_not_zero (&ct->ct_general.use)) nf_ct_evict[i++] = ct;continue ; }
@@ -1748,7 +1748,7 @@ ctnetlink_dump_list(struct sk_buff *skb, struct netlink_callback *cb, bool dyingNFNL_MSG_TYPE (cb->nlh->nlmsg_type), ct, dying, 0);if (res <0 ) { -if (!atomic_inc_not_zero (&ct->ct_general.use)) +if (!refcount_inc_not_zero (&ct->ct_general.use))continue ; cb->args[0 ] = cpu; cb->args[1 ] = (unsigned long )ct;
net/netfilter/nf_conntrack_standalone.c +2 -2
@@ -303,7 +303,7 @@ static int ct_seq_show(struct seq_file *s, void *v)int ret =0 ;WARN_ON (!ct); -if (unlikely (!atomic_inc_not_zero (&ct->ct_general.use))) +if (unlikely (!refcount_inc_not_zero (&ct->ct_general.use)))return 0 ;if (nf_ct_should_gc (ct)) {
@@ -370,7 +370,7 @@ static int ct_seq_show(struct seq_file *s, void *v)ct_show_zone (s, ct,NF_CT_DEFAULT_ZONE_DIR );ct_show_delta_time (s, ct); -seq_printf (s,"use=%u\n" ,atomic_read (&ct->ct_general.use)); +seq_printf (s,"use=%u\n" ,refcount_read (&ct->ct_general.use));if (seq_has_overflowed (s)) goto release;
net/netfilter/nf_flow_table_core.c +1 -1
@@ -48,7 +48,7 @@ struct flow_offload *flow_offload_alloc(struct nf_conn *ct)struct flow_offload *flow;if (unlikely (nf_ct_is_dying (ct) || - !atomic_inc_not_zero (&ct->ct_general.use))) + !refcount_inc_not_zero (&ct->ct_general.use)))return NULL ; flow =kzalloc (sizeof (*flow),GFP_ATOMIC );
net/netfilter/nf_synproxy_core.c +0 -1
@@ -349,7 +349,6 @@ static int __net_init synproxy_net_init(struct net *net) goto err2;__set_bit (IPS_CONFIRMED_BIT , &ct->status); -nf_conntrack_get (&ct->ct_general); snet->tmpl = ct; snet->stats = alloc_percpu(struct synproxy_stats);
net/netfilter/nft_ct.c +1 -3
@@ -259,7 +259,7 @@ static void nft_ct_set_zone_eval(const struct nft_expr *expr, ct =this_cpu_read (nft_ct_pcpu_template); -if (likely (atomic_read (&ct->ct_general.use) ==1 )) { +if (likely (refcount_read (&ct->ct_general.use) ==1 )) {nf_ct_zone_add (ct, &zone); }else {/* previous skb got queued to userspace */
@@ -270,7 +270,6 @@ static void nft_ct_set_zone_eval(const struct nft_expr *expr, } } -atomic_inc (&ct->ct_general.use);nf_ct_set (skb, ct,IP_CT_NEW ); }#endif
@@ -375,7 +374,6 @@ static bool nft_ct_tmpl_alloc_pcpu(void)return false; } -atomic_set (&tmp->ct_general.use,1 );per_cpu (nft_ct_pcpu_template, cpu) = tmp; }
net/netfilter/xt_CT.c +1 -2
@@ -24,7 +24,7 @@ static inline int xt_ct_target(struct sk_buff *skb, struct nf_conn *ct)return XT_CONTINUE ;if (ct) { -atomic_inc (&ct->ct_general.use); +refcount_inc (&ct->ct_general.use);nf_ct_set (skb, ct,IP_CT_NEW ); }else {nf_ct_set (skb, ct,IP_CT_UNTRACKED );
@@ -201,7 +201,6 @@ static int xt_ct_tg_check(const struct xt_tgchk_param *par, goto err4; }__set_bit (IPS_CONFIRMED_BIT , &ct->status); -nf_conntrack_get (&ct->ct_general); out: info->ct = ct;return 0 ;
net/openvswitch/conntrack.c +0 -1
@@ -1716,7 +1716,6 @@ int ovs_ct_copy_action(struct net *net, const struct nlattr *attr, goto err_free_ct;__set_bit (IPS_CONFIRMED_BIT , &ct_info.ct->status); -nf_conntrack_get (&ct_info.ct->ct_general);return 0 ; err_free_ct:__ovs_ct_free_action (&ct_info);
net/sched/act_ct.c +0 -1
@@ -1228,7 +1228,6 @@ static int tcf_ct_fill_params(struct net *net,return -ENOMEM ; }__set_bit (IPS_CONFIRMED_BIT , &tmpl->status); -nf_conntrack_get (&tmpl->ct_general); p->tmpl = tmpl;return 0 ;
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 14/32 netfilter: core: move ip_ct_attach indirection to struct nf_ct_hook
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
ip_ct_attach predates struct nf_ct_hook, we can place it there and remove the exported symbol.
--- include/linux/netfilter.h | 2 +- net/netfilter/core.c | 19 ++++++++----------- net/netfilter/nf_conntrack_core.c | 4 +--- 3 files changed, 10 insertions(+), 15 deletions(-)
include/linux/netfilter.h +1 -1
@@ -440,7 +440,6 @@ nf_nat_decode_session(struct sk_buff *skb, struct flowi *fl, u_int8_t family)#if IS_ENABLED (CONFIG_NF_CONNTRACK )#include <linux/netfilter/nf_conntrack_zones_common.h> -extern void (*ip_ct_attach)(struct sk_buff *,const struct sk_buff *)__rcu ;void nf_ct_attach (struct sk_buff *,const struct sk_buff *);struct nf_conntrack_tuple ; bool nf_ct_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple,
@@ -463,6 +462,7 @@ struct nf_ct_hook {void (*destroy)(struct nf_conntrack *);bool (*get_tuple_skb)(struct nf_conntrack_tuple *,const struct sk_buff *); +void (*attach)(struct sk_buff *nskb,const struct sk_buff *skb); };extern struct nf_ct_hook __rcu *nf_ct_hook;
net/netfilter/core.c +8 -11
@@ -673,25 +673,22 @@ struct nf_ct_hook __rcu *nf_ct_hook __read_mostly;EXPORT_SYMBOL_GPL (nf_ct_hook);#if IS_ENABLED (CONFIG_NF_CONNTRACK ) -/* This does not belong here, but locally generated errors need it if connection -tracking in use: without this, connection may not be in hash table, and hence -manufactured ICMP or RST packets will not be associated with it. */ -void (*ip_ct_attach)(struct sk_buff *,const struct sk_buff *) -__rcu __read_mostly ; -EXPORT_SYMBOL (ip_ct_attach); -struct nf_nat_hook __rcu *nf_nat_hook__read_mostly ;EXPORT_SYMBOL_GPL (nf_nat_hook); +/* This does not belong here, but locally generated errors need it if connection +* tracking in use: without this, connection may not be in hash table, and hence +* manufactured ICMP or RST packets will not be associated with it. +*/ void nf_ct_attach (struct sk_buff *new,const struct sk_buff *skb) { -void (*attach)(struct sk_buff *,const struct sk_buff *); +const struct nf_ct_hook *ct_hook;if (skb->_nfct) {rcu_read_lock (); - attach =rcu_dereference (ip_ct_attach); -if (attach) -attach (new, skb); + ct_hook =rcu_dereference (nf_ct_hook); +if (ct_hook) + ct_hook->attach (new, skb);rcu_read_unlock (); } }
net/netfilter/nf_conntrack_core.c +1 -3
@@ -2455,7 +2455,6 @@ static int kill_all(struct nf_conn *i, void *data)void nf_conntrack_cleanup_start (void ) { conntrack_gc_work.exiting = true; -RCU_INIT_POINTER (ip_ct_attach,NULL ); }void nf_conntrack_cleanup_end (void )
@@ -2774,12 +2773,11 @@ static struct nf_ct_hook nf_conntrack_hook = { .update = nf_conntrack_update, .destroy = destroy_conntrack, .get_tuple_skb = nf_conntrack_get_tuple_skb, + .attach = nf_conntrack_attach, };void nf_conntrack_init_end (void ) { -/* For use by REJECT target */ -RCU_INIT_POINTER (ip_ct_attach, nf_conntrack_attach);RCU_INIT_POINTER (nf_ct_hook, &nf_conntrack_hook); }
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 15/32 netfilter: make function op structures const
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
No functional changes, these structures should be const.
--- include/linux/netfilter.h | 8 ++++---- net/netfilter/core.c | 10 +++++----- net/netfilter/nf_conntrack_core.c | 4 ++-- net/netfilter/nf_conntrack_netlink.c | 4 ++-- net/netfilter/nf_nat_core.c | 2 +- net/netfilter/nfnetlink_queue.c | 8 ++++---- 6 files changed, 18 insertions(+), 18 deletions(-)
include/linux/netfilter.h +4 -4
@@ -381,13 +381,13 @@ struct nf_nat_hook { enum ip_conntrack_dir dir); }; -extern struct nf_nat_hook __rcu *nf_nat_hook; +extern const struct nf_nat_hook __rcu *nf_nat_hook;static inline void nf_nat_decode_session (struct sk_buff *skb,struct flowi *fl,u_int8_t family) {#if IS_ENABLED (CONFIG_NF_NAT ) -struct nf_nat_hook *nat_hook; +const struct nf_nat_hook *nat_hook;rcu_read_lock (); nat_hook =rcu_dereference (nf_nat_hook);
@@ -464,7 +464,7 @@ struct nf_ct_hook { const struct sk_buff *); void (*attach)(struct sk_buff *nskb, const struct sk_buff *skb); }; -extern struct nf_ct_hook __rcu *nf_ct_hook; +extern const struct nf_ct_hook __rcu *nf_ct_hook;struct nlattr ;
@@ -479,7 +479,7 @@ struct nfnl_ct_hook {void (*seq_adjust)(struct sk_buff *skb,struct nf_conn *ct,enum ip_conntrack_info ctinfo,s32 off); }; -extern struct nfnl_ct_hook __rcu *nfnl_ct_hook; +extern const struct nfnl_ct_hook __rcu *nfnl_ct_hook;/** * nf_skb_duplicated - TEE target has sent a packet
net/netfilter/core.c +5 -5
@@ -666,14 +666,14 @@ EXPORT_SYMBOL(nf_hook_slow_list);/* This needs to be compiled in any case to avoid dependencies between the * nfnetlink_queue code and nf_conntrack. */ -struct nfnl_ct_hook __rcu *nfnl_ct_hook__read_mostly ; +const struct nfnl_ct_hook __rcu *nfnl_ct_hook__read_mostly ;EXPORT_SYMBOL_GPL (nfnl_ct_hook); -struct nf_ct_hook __rcu *nf_ct_hook__read_mostly ; +const struct nf_ct_hook __rcu *nf_ct_hook__read_mostly ;EXPORT_SYMBOL_GPL (nf_ct_hook);#if IS_ENABLED (CONFIG_NF_CONNTRACK ) -struct nf_nat_hook __rcu *nf_nat_hook__read_mostly ; +const struct nf_nat_hook __rcu *nf_nat_hook__read_mostly ;EXPORT_SYMBOL_GPL (nf_nat_hook);/* This does not belong here, but locally generated errors need it if connection
@@ -696,7 +696,7 @@ EXPORT_SYMBOL(nf_ct_attach);void nf_conntrack_destroy (struct nf_conntrack *nfct) { -struct nf_ct_hook *ct_hook; +const struct nf_ct_hook *ct_hook;rcu_read_lock (); ct_hook =rcu_dereference (nf_ct_hook);
@@ -709,7 +709,7 @@ EXPORT_SYMBOL(nf_conntrack_destroy);bool nf_ct_get_tuple_skb (struct nf_conntrack_tuple *dst_tuple,const struct sk_buff *skb) { -struct nf_ct_hook *ct_hook; +const struct nf_ct_hook *ct_hook;bool ret = false;rcu_read_lock ();
net/netfilter/nf_conntrack_core.c +2 -2
@@ -2085,9 +2085,9 @@ static int __nf_conntrack_update(struct net *net, struct sk_buff *skb,struct nf_conn *ct, enum ip_conntrack_info ctinfo) { + const struct nf_nat_hook *nat_hook;struct nf_conntrack_tuple_hash *h;struct nf_conntrack_tuple tuple; -struct nf_nat_hook *nat_hook;unsigned int status;int dataoff;u16 l3num;
@@ -2769,7 +2769,7 @@ int nf_conntrack_init_start(void)return ret; } -static struct nf_ct_hook nf_conntrack_hook = { +static const struct nf_ct_hook nf_conntrack_hook = { .update = nf_conntrack_update, .destroy = destroy_conntrack, .get_tuple_skb = nf_conntrack_get_tuple_skb,
net/netfilter/nf_conntrack_netlink.c +2 -2
@@ -1819,7 +1819,7 @@ ctnetlink_parse_nat_setup(struct nf_conn *ct, const struct nlattr *attr)__must_hold (RCU ) { -struct nf_nat_hook *nat_hook; +const struct nf_nat_hook *nat_hook;int err; nat_hook =rcu_dereference (nf_nat_hook);
@@ -2921,7 +2921,7 @@ static void ctnetlink_glue_seqadj(struct sk_buff *skb, struct nf_conn *ct,nf_ct_tcp_seqadj_set (skb, ct, ctinfo, diff); } -static struct nfnl_ct_hook ctnetlink_glue_hook = { +static const struct nfnl_ct_hook ctnetlink_glue_hook = { .build_size = ctnetlink_glue_build_size, .build = ctnetlink_glue_build, .parse = ctnetlink_glue_parse,
net/netfilter/nf_nat_core.c +1 -1
@@ -1167,7 +1167,7 @@ static struct pernet_operations nat_net_ops = { .size = sizeof(struct nat_net), }; -static struct nf_nat_hook nat_hook = { +static const struct nf_nat_hook nat_hook = { .parse_nat_setup = nfnetlink_parse_nat_setup, #ifdef CONFIG_XFRM .decode_session = __nf_nat_decode_session,
net/netfilter/nfnetlink_queue.c +4 -4
@@ -225,7 +225,7 @@ find_dequeue_entry(struct nfqnl_instance *queue, unsigned int id)static void nfqnl_reinject (struct nf_queue_entry *entry,unsigned int verdict) { -struct nf_ct_hook *ct_hook; +const struct nf_ct_hook *ct_hook;int err; if (verdict == NF_ACCEPT ||
@@ -388,7 +388,7 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue,struct net_device *outdev;struct nf_conn *ct =NULL ;enum ip_conntrack_info ctinfo =0 ; -struct nfnl_ct_hook *nfnl_ct; +const struct nfnl_ct_hook *nfnl_ct;bool csum_verify;char *secdata =NULL ;u32 seclen =0 ;
@@ -1103,7 +1103,7 @@ static int nfqnl_recv_verdict_batch(struct sk_buff *skb,return 0 ; } -static struct nf_conn *nfqnl_ct_parse(struct nfnl_ct_hook *nfnl_ct, +static struct nf_conn *nfqnl_ct_parse(const struct nfnl_ct_hook *nfnl_ct, const struct nlmsghdr *nlh, const struct nlattr * const nfqa[], struct nf_queue_entry *entry,
@@ -1170,11 +1170,11 @@ static int nfqnl_recv_verdict(struct sk_buff *skb, const struct nfnl_info *info, {struct nfnl_queue_net *q =nfnl_queue_pernet (info->net);u_int16_t queue_num =ntohs (info->nfmsg->res_id); +const struct nfnl_ct_hook *nfnl_ct;struct nfqnl_msg_verdict_hdr *vhdr;enum ip_conntrack_info ctinfo;struct nfqnl_instance *queue;struct nf_queue_entry *entry; -struct nfnl_ct_hook *nfnl_ct;struct nf_conn *ct =NULL ;unsigned int verdict;int err;
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 16/32 netfilter: conntrack: avoid useless indirection during conntrack destruction
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
nf_ct_put() results in a usesless indirection:
nf_ct_put -> nf_conntrack_put -> nf_conntrack_destroy -> rcu readlock + indirect call of ct_hooks->destroy().
There are two _put helpers:nf_ct_put and nf_conntrack_put. The latter is what should be used in code that MUST NOT cause a linker dependency on the conntrack module (e.g. calls from core network stack).
Everyone else should call nf_ct_put() instead.
A followup patch will convert a few nf_conntrack_put() calls to nf_ct_put(), in particular from modules that already have a conntrack dependency such as act_ct or even nf_conntrack itself.
--- include/linux/netfilter/nf_conntrack_common.h | 2 ++ include/net/netfilter/nf_conntrack.h | 8 ++++++-- net/netfilter/nf_conntrack_core.c | 12 ++++++------ 3 files changed, 14 insertions(+), 8 deletions(-)
include/linux/netfilter/nf_conntrack_common.h +2 -0
@@ -29,6 +29,8 @@ struct nf_conntrack { };void nf_conntrack_destroy (struct nf_conntrack *nfct); + +/* like nf_ct_put, but without module dependency on nf_conntrack */ static inline void nf_conntrack_put (struct nf_conntrack *nfct) {if (nfct &&refcount_dec_and_test (&nfct->use))
include/net/netfilter/nf_conntrack.h +6 -2
@@ -76,6 +76,8 @@ struct nf_conn {* Hint, SKB address this struct and refcnt via skb->_nfct and * helpers nf_conntrack_get() and nf_conntrack_put(). * Helper nf_ct_put() equals nf_conntrack_put() by dec refcnt, +* except that the latter uses internal indirection and does not +* result in a conntrack module dependency. * beware nf_ct_get() is different and don't inc refcnt. */ struct nf_conntrack ct_general;
@@ -170,11 +172,13 @@ nf_ct_get(const struct sk_buff *skb, enum ip_conntrack_info *ctinfo)return (struct nf_conn *)(nfct &NFCT_PTRMASK ); } +void nf_ct_destroy (struct nf_conntrack *nfct); +/* decrement reference count on a conntrack */ static inline void nf_ct_put (struct nf_conn *ct) { -WARN_ON (!ct); -nf_conntrack_put (&ct->ct_general); +if (ct &&refcount_dec_and_test (&ct->ct_general.use)) +nf_ct_destroy (&ct->ct_general); }/* Protocol module loading */
net/netfilter/nf_conntrack_core.c +6 -6
@@ -558,7 +558,7 @@ static void nf_ct_del_from_dying_or_unconfirmed_list(struct nf_conn *ct)#define NFCT_ALIGN (len) (((len) + NFCT_INFOMASK) & ~NFCT_INFOMASK) -/* Released via destroy_conntrack() */ +/* Released via nf_ct_destroy() */ struct nf_conn *nf_ct_tmpl_alloc (struct net *net,const struct nf_conntrack_zone *zone,gfp_t flags)
@@ -612,12 +612,11 @@ static void destroy_gre_conntrack(struct nf_conn *ct)#endif } -static void -destroy_conntrack (struct nf_conntrack *nfct) +void nf_ct_destroy (struct nf_conntrack *nfct) {struct nf_conn *ct = (struct nf_conn *)nfct; -pr_debug ("destroy_conntrack(%p)\n" , ct); +pr_debug ("%s(%p)\n" , __func__, ct);WARN_ON (refcount_read (&nfct->use) !=0 );if (unlikely (nf_ct_is_template (ct))) {
@@ -643,9 +642,10 @@ destroy_conntrack(struct nf_conntrack *nfct)if (ct->master)nf_ct_put (ct->master); -pr_debug ("destroy_conntrack: returning ct=%p to slab\n" , ct); +pr_debug ("%s: returning ct=%p to slab\n" , __func__, ct);nf_conntrack_free (ct); } +EXPORT_SYMBOL (nf_ct_destroy);static void nf_ct_delete_from_lists (struct nf_conn *ct) {
@@ -2771,7 +2771,7 @@ int nf_conntrack_init_start(void)static const struct nf_ct_hook nf_conntrack_hook = { .update = nf_conntrack_update, - .destroy = destroy_conntrack, + .destroy = nf_ct_destroy, .get_tuple_skb = nf_conntrack_get_tuple_skb, .attach = nf_conntrack_attach, };
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 17/32 net: prefer nf_ct_put instead of nf_conntrack_put
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
Its the same as nf_conntrack_put(), but without the need for an indirect call. The downside is a module dependency on nf_conntrack, but all of these already depend on conntrack anyway.
--- net/netfilter/nf_conntrack_core.c | 4 ++-- net/openvswitch/conntrack.c | 14 ++++++++++---- net/sched/act_ct.c | 6 +++--- 3 files changed, 15 insertions(+), 9 deletions(-)
net/netfilter/nf_conntrack_core.c +2 -2
@@ -989,7 +989,7 @@ static int __nf_ct_resolve_clash(struct sk_buff *skb,nf_ct_acct_merge (ct, ctinfo, loser_ct);nf_ct_add_to_dying_list (loser_ct); -nf_conntrack_put (&loser_ct->ct_general); +nf_ct_put (loser_ct);nf_ct_set (skb, ct, ctinfo);NF_CT_STAT_INC (net, clash_resolve);
@@ -1921,7 +1921,7 @@ nf_conntrack_in(struct sk_buff *skb, const struct nf_hook_state *state)/* Invalid: inverse of the return code tells * the netfilter core what to do */ pr_debug ("nf_conntrack_in: Can't track with proto module\n" ); -nf_conntrack_put (&ct->ct_general); +nf_ct_put (ct); skb->_nfct =0 ;NF_CT_STAT_INC_ATOMIC (state->net, invalid);if (ret == -NF_DROP )
net/openvswitch/conntrack.c +10 -4
@@ -574,7 +574,7 @@ ovs_ct_expect_find(struct net *net, const struct nf_conntrack_zone *zone,struct nf_conn *ct =nf_ct_tuplehash_to_ctrack (h);nf_ct_delete (ct,0 ,0 ); -nf_conntrack_put (&ct->ct_general); +nf_ct_put (ct); } }
@@ -723,7 +723,7 @@ static bool skb_nfct_cached(struct net *net,if (nf_ct_is_confirmed (ct))nf_ct_delete (ct,0 ,0 ); -nf_conntrack_put (&ct->ct_general); +nf_ct_put (ct);nf_ct_set (skb,NULL ,0 );return false; }
@@ -967,7 +967,8 @@ static int __ovs_ct_lookup(struct net *net, struct sw_flow_key *key,/* Associate skb with specified zone. */ if (tmpl) { -nf_conntrack_put (skb_nfct (skb)); + ct =nf_ct_get (skb, &ctinfo); +nf_ct_put (ct);nf_conntrack_get (&tmpl->ct_general);nf_ct_set (skb, tmpl,IP_CT_NEW ); }
@@ -1328,7 +1329,12 @@ int ovs_ct_execute(struct net *net, struct sk_buff *skb,int ovs_ct_clear (struct sk_buff *skb,struct sw_flow_key *key) { -nf_conntrack_put (skb_nfct (skb)); +enum ip_conntrack_info ctinfo; +struct nf_conn *ct; + + ct =nf_ct_get (skb, &ctinfo); + +nf_ct_put (ct);nf_ct_set (skb,NULL ,IP_CT_UNTRACKED );ovs_ct_fill_key (skb, key, false);
net/sched/act_ct.c +3 -3
@@ -598,7 +598,7 @@ static bool tcf_ct_skb_nfct_cached(struct net *net, struct sk_buff *skb,if (nf_ct_is_confirmed (ct))nf_ct_kill (ct); -nf_conntrack_put (&ct->ct_general); +nf_ct_put (ct);nf_ct_set (skb,NULL ,IP_CT_UNTRACKED );return false;
@@ -763,7 +763,7 @@ static void tcf_ct_params_free(struct rcu_head *head)tcf_ct_flow_table_put (params);if (params->tmpl) -nf_conntrack_put (¶ms->tmpl->ct_general); +nf_ct_put (params->tmpl);kfree (params); }
@@ -967,7 +967,7 @@ static int tcf_ct_act(struct sk_buff *skb, const struct tc_action *a,qdisc_skb_cb (skb)->post_ct = false; ct =nf_ct_get (skb, &ctinfo);if (ct) { -nf_conntrack_put (&ct->ct_general); +nf_ct_put (ct);nf_ct_set (skb,NULL ,IP_CT_UNTRACKED ); }
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 18/32 netfilter: egress: avoid a lockdep splat
In reply to Pablo Neira AyusoFrom: Florian Westphal <[email protected]>
include/linux/netfilter_netdev.h:97 suspicious rcu_dereference_check() usage! 2 locks held by sd-resolve/1100:
0: ..(rcu_read_lock_bh){1:3}, at: ip_finish_output2
1: ..(rcu_read_lock_bh){1:3}, at: __dev_queue_xmit __dev_queue_xmit+0 ..
The helper has two callers, one uses rcu_read_lock, the other rcu_read_lock_bh(). Annotate the dereference to reflect this.
--- include/linux/netfilter_netdev.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
include/linux/netfilter_netdev.h +1 -1
@@ -94,7 +94,7 @@ static inline struct sk_buff *nf_hook_egress(struct sk_buff *skb, int *rc,return skb;#endif - e =rcu_dereference (dev->nf_hooks_egress); + e =rcu_dereference_check (dev->nf_hooks_egress,rcu_read_lock_bh_held ());if (!e)return skb;
Signature · 1 line
2.30.2
Re: [PATCH net-next 18/32] netfilter: egress: avoid a lockdep splat
Quoted 23 lines · Pablo Neira Ayuso
From: Florian Westphal <[email protected]>
include/linux/netfilter_netdev.h:97suspiciousrcu_dereference_check()usage! 2 locks held by sd-resolve/1100:
0: ..(rcu_read_lock_bh){1:3}, at:ip_finish_output2
1: ..(rcu_read_lock_bh){1:3}, at:__dev_queue_xmit__dev_queue_xmit+0 ..The helper has two callers, one uses
rcu_read_lock, the otherrcu_read_lock_bh(). Annotate the dereference to reflect this.Fixes 42df6e1d221dd ("netfilter: Introduce egress hook")Signed-off-by Florian Westphal <[email protected]>Signed-off-by Pablo Neira Ayuso <[email protected]>--- include/linux/netfilter_netdev.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/linux/netfilter_netdev.h +0
-0
include/linux/netfilter_netdev.h static inline struct sk_buff *nf_hook_egress(struct sk_buff *skb, int *rc, · line 100
Quoted 8 lines
include/linux/netfilter_netdev.h +1
-1@@ -94,7 +94,7 @@ static inline struct sk_buff *nf_hook_egress(struct sk_buff *skb, int *rc,return skb;#endif - e =rcu_dereference (dev->nf_hooks_egress); + e =rcu_dereference_check (dev->nf_hooks_egress,rcu_read_lock_bh_held ());if (!e)return skb;
It seems other rcu_dereference() uses will also trigger lockdep splat.
nft_do_chain()
...
if (genbit)
blob = rcu_dereference(chain->blob_gen_1);
else
blob = rcu_dereference(chain->blob_gen_0);
I wonder how many other places will need a fix ?
Florian Westphal
maintainerRe: [PATCH net-next 18/32] netfilter: egress: avoid a lockdep splat
Quoted 14 lines · Eric Dumazet
include/linux/netfilter_netdev.h static inline struct sk_buff *nf_hook_egress(struct sk_buff *skb, int *rc, · line 98
Quoted 10 lines
index
b71b57a83bb4..b4dd96e4dc8d100644include/linux/netfilter_netdev.h +1
-1@@ -94,7 +94,7 @@ static inline struct sk_buff *nf_hook_egress(struct sk_buff *skb, int *rc,return skb;#endif - e =rcu_dereference (dev->nf_hooks_egress); + e =rcu_dereference_check (dev->nf_hooks_egress,rcu_read_lock_bh_held ());if (!e)return skb;It seems other
rcu_dereference()uses will also trigger lockdep splat.
Quoted 6 lines
if (genbit)
blob =rcu_dereference(chain->blob_gen_1);
else
blob =rcu_dereference(chain->blob_gen_0);I wonder how many other places will need a fix ?
I don't like that, all nf hooks assume rcu_read_lock is held. nf_hook_egress() call to nf_hook_slow() needs to be wrapped in rcu_read_(un)lock pair.
Pablo Neira Ayuso
maintainerPATCH 19/32 netfilter: nft_connlimit: move stateful fields out of expression data
In reply to Pablo Neira AyusoIn preparation for the rule blob representation.
--- net/netfilter/nft_connlimit.c | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-)
net/netfilter/nft_connlimit.c +18 -8
@@ -14,7 +14,7 @@#include <net/netfilter/nf_conntrack_zones.h> struct nft_connlimit { -struct nf_conncount_list list; +struct nf_conncount_list *list;u32 limit;bool invert; };
@@ -43,12 +43,12 @@ static inline void nft_connlimit_do_eval(struct nft_connlimit *priv,return ; } -if (nf_conncount_add (nft_net (pkt), &priv->list, tuple_ptr, zone)) { +if (nf_conncount_add (nft_net (pkt), priv->list, tuple_ptr, zone)) { regs->verdict.code =NF_DROP ;return ; } - count = priv->list.count; + count = priv->list->count;if ((count > priv->limit) ^ priv->invert) { regs->verdict.code =NFT_BREAK ;
@@ -76,7 +76,11 @@ static int nft_connlimit_do_init(const struct nft_ctx *ctx, invert = true; } -nf_conncount_list_init (&priv->list); + priv->list =kmalloc (sizeof (*priv->list),GFP_KERNEL ); +if (!priv->list) +return -ENOMEM ; + +nf_conncount_list_init (priv->list); priv->limit = limit; priv->invert = invert;
@@ -87,7 +91,8 @@ static void nft_connlimit_do_destroy(const struct nft_ctx *ctx, struct nft_connlimit *priv) {nf_ct_netns_put (ctx->net, ctx->family); -nf_conncount_cache_free (&priv->list); +nf_conncount_cache_free (priv->list); +kfree (priv->list); } static int nft_connlimit_do_dump(struct sk_buff *skb,
@@ -200,7 +205,11 @@ static int nft_connlimit_clone(struct nft_expr *dst, const struct nft_expr *src)struct nft_connlimit *priv_dst =nft_expr_priv (dst);struct nft_connlimit *priv_src =nft_expr_priv (src); -nf_conncount_list_init (&priv_dst->list); + priv_dst->list =kmalloc (sizeof (*priv_dst->list),GFP_ATOMIC ); +if (priv_dst->list) +return -ENOMEM ; + +nf_conncount_list_init (priv_dst->list); priv_dst->limit = priv_src->limit; priv_dst->invert = priv_src->invert;
@@ -212,7 +221,8 @@ static void nft_connlimit_destroy_clone(const struct nft_ctx *ctx, {struct nft_connlimit *priv =nft_expr_priv (expr); -nf_conncount_cache_free (&priv->list); +nf_conncount_cache_free (priv->list); +kfree (priv->list); }static bool nft_connlimit_gc (struct net *net,const struct nft_expr *expr)
@@ -221,7 +231,7 @@ static bool nft_connlimit_gc(struct net *net, const struct nft_expr *expr)bool ret;local_bh_disable (); - ret =nf_conncount_gc_list (net, &priv->list); + ret =nf_conncount_gc_list (net, priv->list);local_bh_enable ();return ret;
Signature · 1 line
2.30.2
Re: [PATCH net-next 19/32] netfilter: nft_connlimit: move stateful fields out of expression data
Quoted 11 lines · Pablo Neira Ayuso
In preparation for the rule blob representation.
Signed-off-by Pablo Neira Ayuso <[email protected]>--- net/netfilter/nft_connlimit.c | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-)net/netfilter/nft_connlimit.c +0
-0
[...]
Quoted 5 lines
static int nft_connlimit_do_dump(struct
sk_buff*skb,net/netfilter/nft_connlimit.c +0
-0@@ -200,7 +205,11 @@ static int nft_connlimit_clone(struct nft_expr *dst, const struct nft_expr *src)struct nft_connlimit *priv_dst =nft_expr_priv (dst);struct nft_connlimit *priv_src =nft_expr_priv (src);
net/netfilter/nft_connlimit.c static int nft_connlimit_clone(struct nft_expr *dst, const struct nft_expr *src) · line 215
Quoted 10 lines
net/netfilter/nft_connlimit.c +5
-1-nf_conncount_list_init (&priv_dst->list); + priv_dst->list =kmalloc (sizeof (*priv_dst->list),GFP_ATOMIC ); +if (priv_dst->list) +return -ENOMEM ; + +nf_conncount_list_init (priv_dst->list); priv_dst->limit = priv_src->limit; priv_dst->invert = priv_src->invert;
Hi Pablo,
Coverity (CID 1510697) spotted a typo in this NULL check, it should be
if (!priv_dst->list) return -ENOMEM;
Looks like the following patches also have this bug.
Pablo Neira Ayuso
maintainerRe: [PATCH net-next 19/32] netfilter: nft_connlimit: move stateful fields out of expression data
Hi,
Quoted 33 lines · Julian Wiedmann
Quoted 11 lines · Pablo Neira Ayuso
In preparation for the rule blob representation.
Signed-off-by Pablo Neira Ayuso <[email protected]>--- net/netfilter/nft_connlimit.c | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-)net/netfilter/nft_connlimit.c +0
-0[...]
Quoted 5 lines
static int nft_connlimit_do_dump(struct
sk_buff*skb,net/netfilter/nft_connlimit.c +0
-0@@ -200,7 +205,11 @@ static int nft_connlimit_clone(struct nft_expr *dst, const struct nft_expr *src)struct nft_connlimit *priv_dst =nft_expr_priv (dst);struct nft_connlimit *priv_src =nft_expr_priv (src);net/netfilter/nft_connlimit.c static int nft_connlimit_clone(struct nft_expr *dst, const struct nft_expr *src) · line 215
Quoted 10 lines
net/netfilter/nft_connlimit.c +5
-1-nf_conncount_list_init (&priv_dst->list); + priv_dst->list =kmalloc (sizeof (*priv_dst->list),GFP_ATOMIC ); +if (priv_dst->list) +return -ENOMEM ; + +nf_conncount_list_init (priv_dst->list); priv_dst->limit = priv_src->limit; priv_dst->invert = priv_src->invert;Hi Pablo,
Quoted 7 lines
Coverity (CID 1510697) spotted a typo in this NULL check, it should be
if (!priv_dst->list) return -ENOMEM;Looks like the following patches also have this bug.
Thanks, I'll post a patchset to fix this.
Pablo Neira Ayuso
maintainerPATCH 20/32 netfilter: nft_last: move stateful fields out of expression data
In reply to Pablo Neira AyusoIn preparation for the rule blob representation.
--- net/netfilter/nft_last.c | 69 +++++++++++++++++++++++++++++----------- 1 file changed, 51 insertions(+), 18 deletions(-)
net/netfilter/nft_last.c +51 -18
@@ -8,9 +8,13 @@#include <net/netfilter/nf_tables_core.h> #include <net/netfilter/nf_tables.h> +struct nft_last { +unsigned long jiffies; +unsigned int set; +}; +struct nft_last_priv { -unsigned long last_jiffies; -unsigned int last_set; +struct nft_last *last; };static const struct nla_policy nft_last_policy[NFTA_LAST_MAX +1 ] = {
@@ -22,47 +26,55 @@ static int nft_last_init(const struct nft_ctx *ctx, const struct nft_expr *expr, const struct nlattr * const tb[]) {struct nft_last_priv *priv =nft_expr_priv (expr); +struct nft_last *last;u64 last_jiffies; -u32 last_set =0 ;int err; -if (tb[NFTA_LAST_SET ]) { - last_set =ntohl (nla_get_be32 (tb[NFTA_LAST_SET ])); -if (last_set ==1 ) - priv->last_set =1 ; - } + last =kzalloc (sizeof (*last),GFP_KERNEL ); +if (!last) +return -ENOMEM ; + +if (tb[NFTA_LAST_SET ]) + last->set =ntohl (nla_get_be32 (tb[NFTA_LAST_SET ])); -if (last_set && tb[NFTA_LAST_MSECS ]) { +if (last->set && tb[NFTA_LAST_MSECS ]) { err =nf_msecs_to_jiffies64 (tb[NFTA_LAST_MSECS ], &last_jiffies);if (err <0 ) -return err; + goto err; - priv->last_jiffies = jiffies - (unsigned long )last_jiffies; + last->jiffies = jiffies - (unsigned long )last_jiffies; } + priv->last = last;return 0 ; +err: +kfree (last); + +return err; }static void nft_last_eval (const struct nft_expr *expr,struct nft_regs *regs,const struct nft_pktinfo *pkt) {struct nft_last_priv *priv =nft_expr_priv (expr); +struct nft_last *last = priv->last; -if (READ_ONCE (priv->last_jiffies) != jiffies) -WRITE_ONCE (priv->last_jiffies, jiffies); -if (READ_ONCE (priv->last_set) ==0 ) -WRITE_ONCE (priv->last_set,1 ); +if (READ_ONCE (last->jiffies) != jiffies) +WRITE_ONCE (last->jiffies, jiffies); +if (READ_ONCE (last->set) ==0 ) +WRITE_ONCE (last->set,1 ); }static int nft_last_dump (struct sk_buff *skb,const struct nft_expr *expr) {struct nft_last_priv *priv =nft_expr_priv (expr); -unsigned long last_jiffies =READ_ONCE (priv->last_jiffies); -u32 last_set =READ_ONCE (priv->last_set); +struct nft_last *last = priv->last; +unsigned long last_jiffies =READ_ONCE (last->jiffies); +u32 last_set =READ_ONCE (last->set);__be64 msecs;if (time_before (jiffies, last_jiffies)) { -WRITE_ONCE (priv->last_set,0 ); +WRITE_ONCE (last->set,0 ); last_set =0 ; }
@@ -81,11 +93,32 @@ static int nft_last_dump(struct sk_buff *skb, const struct nft_expr *expr)return -1 ; } +static void nft_last_destroy (const struct nft_ctx *ctx, +const struct nft_expr *expr) +{ +struct nft_last_priv *priv =nft_expr_priv (expr); + +kfree (priv->last); +} + +static int nft_last_clone (struct nft_expr *dst,const struct nft_expr *src) +{ +struct nft_last_priv *priv_dst =nft_expr_priv (dst); + + priv_dst->last =kzalloc (sizeof (*priv_dst->last),GFP_ATOMIC ); +if (priv_dst->last) +return -ENOMEM ; + +return 0 ; +} +static const struct nft_expr_ops nft_last_ops = { .type = &nft_last_type, .size =NFT_EXPR_SIZE (sizeof (struct nft_last_priv )), .eval = nft_last_eval, .init = nft_last_init, + .destroy = nft_last_destroy, + .clone = nft_last_clone, .dump = nft_last_dump, };
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 21/32 netfilter: nft_quota: move stateful fields out of expression data
In reply to Pablo Neira AyusoIn preparation for the rule blob representation.
--- net/netfilter/nft_quota.c | 52 +++++++++++++++++++++++++++++++++++---- 1 file changed, 47 insertions(+), 5 deletions(-)
net/netfilter/nft_quota.c +47 -5
@@ -15,13 +15,13 @@struct nft_quota {atomic64_t quota;unsigned long flags; -atomic64_t consumed; +atomic64_t *consumed; };static inline bool nft_overquota (struct nft_quota *priv,const struct sk_buff *skb) { -return atomic64_add_return (skb->len, &priv->consumed) >= +return atomic64_add_return (skb->len, priv->consumed) >=atomic64_read (&priv->quota); }
@@ -90,13 +90,23 @@ static int nft_quota_do_init(const struct nlattr * const tb[],return -EOPNOTSUPP ; } + priv->consumed =kmalloc (sizeof (*priv->consumed),GFP_KERNEL ); +if (!priv->consumed) +return -ENOMEM ; +atomic64_set (&priv->quota, quota); priv->flags = flags; -atomic64_set (&priv->consumed, consumed); +atomic64_set (priv->consumed, consumed);return 0 ; } +static void nft_quota_do_destroy (const struct nft_ctx *ctx, +struct nft_quota *priv) +{ +kfree (priv->consumed); +} +static int nft_quota_obj_init (const struct nft_ctx *ctx,const struct nlattr *const tb[],struct nft_object *obj)
@@ -128,7 +138,7 @@ static int nft_quota_do_dump(struct sk_buff *skb, struct nft_quota *priv,* that we see, don't go over the quota boundary in what we send to * userspace. */ - consumed =atomic64_read (&priv->consumed); + consumed =atomic64_read (priv->consumed); quota =atomic64_read (&priv->quota);if (consumed >= quota) { consumed_cap = quota;
@@ -145,7 +155,7 @@ static int nft_quota_do_dump(struct sk_buff *skb, struct nft_quota *priv, goto nla_put_failure;if (reset) { -atomic64_sub (consumed, &priv->consumed); +atomic64_sub (consumed, priv->consumed);clear_bit (NFT_QUOTA_DEPLETED_BIT , &priv->flags); }return 0 ;
@@ -162,11 +172,20 @@ static int nft_quota_obj_dump(struct sk_buff *skb, struct nft_object *obj,return nft_quota_do_dump (skb, priv, reset); } +static void nft_quota_obj_destroy (const struct nft_ctx *ctx, +struct nft_object *obj) +{ +struct nft_quota *priv =nft_obj_data (obj); + +return nft_quota_do_destroy (ctx, priv); +} +static struct nft_object_type nft_quota_obj_type;static const struct nft_object_ops nft_quota_obj_ops = { .type = &nft_quota_obj_type, .size =sizeof (struct nft_quota ), .init = nft_quota_obj_init, + .destroy = nft_quota_obj_destroy, .eval = nft_quota_obj_eval, .dump = nft_quota_obj_dump, .update = nft_quota_obj_update,
@@ -205,12 +224,35 @@ static int nft_quota_dump(struct sk_buff *skb, const struct nft_expr *expr)return nft_quota_do_dump (skb, priv, false); } +static void nft_quota_destroy (const struct nft_ctx *ctx, +const struct nft_expr *expr) +{ +struct nft_quota *priv =nft_expr_priv (expr); + +return nft_quota_do_destroy (ctx, priv); +} + +static int nft_quota_clone (struct nft_expr *dst,const struct nft_expr *src) +{ +struct nft_quota *priv_dst =nft_expr_priv (dst); + + priv_dst->consumed =kmalloc (sizeof (*priv_dst->consumed),GFP_ATOMIC ); +if (priv_dst->consumed) +return -ENOMEM ; + +atomic64_set (priv_dst->consumed,0 ); + +return 0 ; +} +static struct nft_expr_type nft_quota_type;static const struct nft_expr_ops nft_quota_ops = { .type = &nft_quota_type, .size =NFT_EXPR_SIZE (sizeof (struct nft_quota )), .eval = nft_quota_eval, .init = nft_quota_init, + .destroy = nft_quota_destroy, + .clone = nft_quota_clone, .dump = nft_quota_dump, };
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 22/32 netfilter: nft_numgen: move stateful fields out of expression data
In reply to Pablo Neira AyusoIn preparation for the rule blob representation.
--- net/netfilter/nft_numgen.c | 34 ++++++++++++++++++++++++++++------ 1 file changed, 28 insertions(+), 6 deletions(-)
net/netfilter/nft_numgen.c +28 -6
@@ -18,7 +18,7 @@ static DEFINE_PER_CPU(struct rnd_state, nft_numgen_prandom_state);struct nft_ng_inc {u8 dreg;u32 modulus; -atomic_t counter; +atomic_t *counter;u32 offset; };
@@ -27,9 +27,9 @@ static u32 nft_ng_inc_gen(struct nft_ng_inc *priv)u32 nval, oval;do { - oval =atomic_read (&priv->counter); + oval =atomic_read (priv->counter); nval = (oval +1 < priv->modulus) ? oval +1 :0 ; - }while (atomic_cmpxchg (&priv->counter, oval, nval) != oval); + }while (atomic_cmpxchg (priv->counter, oval, nval) != oval);return nval + priv->offset; }
@@ -55,6 +55,7 @@ static int nft_ng_inc_init(const struct nft_ctx *ctx, const struct nlattr * const tb[]) {struct nft_ng_inc *priv =nft_expr_priv (expr); +int err;if (tb[NFTA_NG_OFFSET ]) priv->offset =ntohl (nla_get_be32 (tb[NFTA_NG_OFFSET ]));
@@ -66,10 +67,22 @@ static int nft_ng_inc_init(const struct nft_ctx *ctx,if (priv->offset + priv->modulus -1 < priv->offset)return -EOVERFLOW ; -atomic_set (&priv->counter, priv->modulus -1 ); + priv->counter =kmalloc (sizeof (*priv->counter),GFP_KERNEL ); +if (!priv->counter) +return -ENOMEM ; -return nft_parse_register_store (ctx, tb[NFTA_NG_DREG ], &priv->dreg, -NULL ,NFT_DATA_VALUE ,sizeof (u32)); +atomic_set (priv->counter, priv->modulus -1 ); + + err =nft_parse_register_store (ctx, tb[NFTA_NG_DREG ], &priv->dreg, +NULL ,NFT_DATA_VALUE ,sizeof (u32)); +if (err <0 ) + goto err; + +return 0 ; +err: +kfree (priv->counter); + +return err; } static int nft_ng_dump(struct sk_buff *skb, enum nft_registers dreg,
@@ -98,6 +111,14 @@ static int nft_ng_inc_dump(struct sk_buff *skb, const struct nft_expr *expr) priv->offset); } +static void nft_ng_inc_destroy (const struct nft_ctx *ctx, +const struct nft_expr *expr) +{ +const struct nft_ng_inc *priv =nft_expr_priv (expr); + +kfree (priv->counter); +} +struct nft_ng_random {u8 dreg;u32 modulus;
@@ -157,6 +178,7 @@ static const struct nft_expr_ops nft_ng_inc_ops = { .size = NFT_EXPR_SIZE(sizeof(struct nft_ng_inc)), .eval = nft_ng_inc_eval, .init = nft_ng_inc_init, + .destroy = nft_ng_inc_destroy, .dump = nft_ng_inc_dump, };
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 23/32 netfilter: nft_limit: rename stateful structure
In reply to Pablo Neira AyusoFrom struct nft_limit to nft_limit_priv.
--- net/netfilter/nft_limit.c | 104 +++++++++++++++++++------------------- 1 file changed, 52 insertions(+), 52 deletions(-)
net/netfilter/nft_limit.c +52 -52
@@ -14,7 +14,7 @@#include <linux/netfilter/nf_tables.h> #include <net/netfilter/nf_tables.h> -struct nft_limit { +struct nft_limit_priv {spinlock_t lock;u64 last;u64 tokens;
@@ -25,33 +25,33 @@ struct nft_limit {bool invert; }; -static inline bool nft_limit_eval (struct nft_limit *limit,u64 cost) +static inline bool nft_limit_eval (struct nft_limit_priv *priv,u64 cost) {u64 now, tokens;s64 delta; -spin_lock_bh (&limit->lock); +spin_lock_bh (&priv->lock); now =ktime_get_ns (); - tokens = limit->tokens + now - limit->last; -if (tokens > limit->tokens_max) - tokens = limit->tokens_max; + tokens = priv->tokens + now - priv->last; +if (tokens > priv->tokens_max) + tokens = priv->tokens_max; - limit->last = now; + priv->last = now; delta = tokens - cost;if (delta >=0 ) { - limit->tokens = delta; -spin_unlock_bh (&limit->lock); -return limit->invert; + priv->tokens = delta; +spin_unlock_bh (&priv->lock); +return priv->invert; } - limit->tokens = tokens; -spin_unlock_bh (&limit->lock); -return !limit->invert; + priv->tokens = tokens; +spin_unlock_bh (&priv->lock); +return !priv->invert; }/* Use same default as in iptables. */ #define NFT_LIMIT_PKT_BURST_DEFAULT 5 -static int nft_limit_init (struct nft_limit *limit, +static int nft_limit_init (struct nft_limit_priv *priv,const struct nlattr *const tb[],bool pkts) {u64 unit, tokens;
@@ -60,58 +60,58 @@ static int nft_limit_init(struct nft_limit *limit, tb[NFTA_LIMIT_UNIT] == NULL)return -EINVAL ; - limit->rate =be64_to_cpu (nla_get_be64 (tb[NFTA_LIMIT_RATE ])); + priv->rate =be64_to_cpu (nla_get_be64 (tb[NFTA_LIMIT_RATE ])); unit =be64_to_cpu (nla_get_be64 (tb[NFTA_LIMIT_UNIT ])); - limit->nsecs = unit *NSEC_PER_SEC ; -if (limit->rate ==0 || limit->nsecs < unit) + priv->nsecs = unit *NSEC_PER_SEC ; +if (priv->rate ==0 || priv->nsecs < unit)return -EOVERFLOW ;if (tb[NFTA_LIMIT_BURST ]) - limit->burst =ntohl (nla_get_be32 (tb[NFTA_LIMIT_BURST ])); + priv->burst =ntohl (nla_get_be32 (tb[NFTA_LIMIT_BURST ])); -if (pkts && limit->burst ==0 ) - limit->burst =NFT_LIMIT_PKT_BURST_DEFAULT ; +if (pkts && priv->burst ==0 ) + priv->burst =NFT_LIMIT_PKT_BURST_DEFAULT ; -if (limit->rate + limit->burst < limit->rate) +if (priv->rate + priv->burst < priv->rate)return -EOVERFLOW ;if (pkts) { - tokens =div64_u64 (limit->nsecs, limit->rate) * limit->burst; + tokens =div64_u64 (priv->nsecs, priv->rate) * priv->burst; }else {/* The token bucket size limits the number of tokens can be * accumulated. tokens_max specifies the bucket size. * tokens_max = unit * (rate + burst) / rate. */ - tokens =div64_u64 (limit->nsecs * (limit->rate + limit->burst), - limit->rate); + tokens =div64_u64 (priv->nsecs * (priv->rate + priv->burst), + priv->rate); } - limit->tokens = tokens; - limit->tokens_max = limit->tokens; + priv->tokens = tokens; + priv->tokens_max = priv->tokens;if (tb[NFTA_LIMIT_FLAGS ]) {u32 flags =ntohl (nla_get_be32 (tb[NFTA_LIMIT_FLAGS ]));if (flags &NFT_LIMIT_F_INV ) - limit->invert = true; + priv->invert = true; } - limit->last =ktime_get_ns (); -spin_lock_init (&limit->lock); + priv->last =ktime_get_ns (); +spin_lock_init (&priv->lock);return 0 ; } -static int nft_limit_dump (struct sk_buff *skb,const struct nft_limit *limit, +static int nft_limit_dump (struct sk_buff *skb,const struct nft_limit_priv *priv,enum nft_limit_type type) { -u32 flags = limit->invert ?NFT_LIMIT_F_INV :0 ; -u64 secs =div_u64 (limit->nsecs,NSEC_PER_SEC ); +u32 flags = priv->invert ?NFT_LIMIT_F_INV :0 ; +u64 secs =div_u64 (priv->nsecs,NSEC_PER_SEC ); -if (nla_put_be64 (skb,NFTA_LIMIT_RATE ,cpu_to_be64 (limit->rate), +if (nla_put_be64 (skb,NFTA_LIMIT_RATE ,cpu_to_be64 (priv->rate),NFTA_LIMIT_PAD ) ||nla_put_be64 (skb,NFTA_LIMIT_UNIT ,cpu_to_be64 (secs),NFTA_LIMIT_PAD ) || -nla_put_be32 (skb,NFTA_LIMIT_BURST ,htonl (limit->burst)) || +nla_put_be32 (skb,NFTA_LIMIT_BURST ,htonl (priv->burst)) ||nla_put_be32 (skb,NFTA_LIMIT_TYPE ,htonl (type)) ||nla_put_be32 (skb,NFTA_LIMIT_FLAGS ,htonl (flags))) goto nla_put_failure;
@@ -121,8 +121,8 @@ static int nft_limit_dump(struct sk_buff *skb, const struct nft_limit *limit,return -1 ; } -struct nft_limit_pkts { -struct nft_limit limit; +struct nft_limit_priv_pkts { +struct nft_limit_priv limit;u64 cost; };
@@ -130,7 +130,7 @@ static void nft_limit_pkts_eval(const struct nft_expr *expr,struct nft_regs *regs, const struct nft_pktinfo *pkt) { - struct nft_limit_pkts *priv = nft_expr_priv(expr); + struct nft_limit_priv_pkts *priv = nft_expr_priv(expr);if (nft_limit_eval (&priv->limit, priv->cost)) regs->verdict.code =NFT_BREAK ;
@@ -148,7 +148,7 @@ static int nft_limit_pkts_init(const struct nft_ctx *ctx,const struct nft_expr *expr, const struct nlattr * const tb[]) { - struct nft_limit_pkts *priv = nft_expr_priv(expr); + struct nft_limit_priv_pkts *priv = nft_expr_priv(expr);int err; err =nft_limit_init (&priv->limit, tb, true);
@@ -161,7 +161,7 @@ static int nft_limit_pkts_init(const struct nft_ctx *ctx,static int nft_limit_pkts_dump (struct sk_buff *skb,const struct nft_expr *expr) { -const struct nft_limit_pkts *priv =nft_expr_priv (expr); +const struct nft_limit_priv_pkts *priv =nft_expr_priv (expr);return nft_limit_dump (skb, &priv->limit,NFT_LIMIT_PKTS ); }
@@ -169,7 +169,7 @@ static int nft_limit_pkts_dump(struct sk_buff *skb, const struct nft_expr *expr)static struct nft_expr_type nft_limit_type;static const struct nft_expr_ops nft_limit_pkts_ops = { .type = &nft_limit_type, - .size =NFT_EXPR_SIZE (sizeof (struct nft_limit_pkts )), + .size =NFT_EXPR_SIZE (sizeof (struct nft_limit_priv_pkts )), .eval = nft_limit_pkts_eval, .init = nft_limit_pkts_init, .dump = nft_limit_pkts_dump,
@@ -179,7 +179,7 @@ static void nft_limit_bytes_eval(const struct nft_expr *expr,struct nft_regs *regs, const struct nft_pktinfo *pkt) { - struct nft_limit *priv = nft_expr_priv(expr); + struct nft_limit_priv *priv = nft_expr_priv(expr);u64 cost =div64_u64 (priv->nsecs * pkt->skb->len, priv->rate);if (nft_limit_eval (priv, cost))
@@ -190,7 +190,7 @@ static int nft_limit_bytes_init(const struct nft_ctx *ctx,const struct nft_expr *expr, const struct nlattr * const tb[]) { - struct nft_limit *priv = nft_expr_priv(expr); + struct nft_limit_priv *priv = nft_expr_priv(expr);return nft_limit_init (priv, tb, false); }
@@ -198,14 +198,14 @@ static int nft_limit_bytes_init(const struct nft_ctx *ctx,static int nft_limit_bytes_dump (struct sk_buff *skb,const struct nft_expr *expr) { -const struct nft_limit *priv =nft_expr_priv (expr); +const struct nft_limit_priv *priv =nft_expr_priv (expr);return nft_limit_dump (skb, priv,NFT_LIMIT_PKT_BYTES ); }static const struct nft_expr_ops nft_limit_bytes_ops = { .type = &nft_limit_type, - .size =NFT_EXPR_SIZE (sizeof (struct nft_limit )), + .size =NFT_EXPR_SIZE (sizeof (struct nft_limit_priv )), .eval = nft_limit_bytes_eval, .init = nft_limit_bytes_init, .dump = nft_limit_bytes_dump,
@@ -240,7 +240,7 @@ static void nft_limit_obj_pkts_eval(struct nft_object *obj,struct nft_regs *regs, const struct nft_pktinfo *pkt) { - struct nft_limit_pkts *priv = nft_obj_data(obj); + struct nft_limit_priv_pkts *priv = nft_obj_data(obj);if (nft_limit_eval (&priv->limit, priv->cost)) regs->verdict.code =NFT_BREAK ;
@@ -250,7 +250,7 @@ static int nft_limit_obj_pkts_init(const struct nft_ctx *ctx,const struct nlattr *const tb[], struct nft_object *obj) { - struct nft_limit_pkts *priv = nft_obj_data(obj); + struct nft_limit_priv_pkts *priv = nft_obj_data(obj);int err; err =nft_limit_init (&priv->limit, tb, true);
@@ -265,7 +265,7 @@ static int nft_limit_obj_pkts_dump(struct sk_buff *skb, struct nft_object *obj, bool reset) { -const struct nft_limit_pkts *priv =nft_obj_data (obj); +const struct nft_limit_priv_pkts *priv =nft_obj_data (obj);return nft_limit_dump (skb, &priv->limit,NFT_LIMIT_PKTS ); }
@@ -273,7 +273,7 @@ static int nft_limit_obj_pkts_dump(struct sk_buff *skb,static struct nft_object_type nft_limit_obj_type;static const struct nft_object_ops nft_limit_obj_pkts_ops = { .type = &nft_limit_obj_type, - .size =NFT_EXPR_SIZE (sizeof (struct nft_limit_pkts )), + .size =NFT_EXPR_SIZE (sizeof (struct nft_limit_priv_pkts )), .init = nft_limit_obj_pkts_init, .eval = nft_limit_obj_pkts_eval, .dump = nft_limit_obj_pkts_dump,
@@ -283,7 +283,7 @@ static void nft_limit_obj_bytes_eval(struct nft_object *obj,struct nft_regs *regs, const struct nft_pktinfo *pkt) { - struct nft_limit *priv = nft_obj_data(obj); + struct nft_limit_priv *priv = nft_obj_data(obj);u64 cost =div64_u64 (priv->nsecs * pkt->skb->len, priv->rate);if (nft_limit_eval (priv, cost))
@@ -294,7 +294,7 @@ static int nft_limit_obj_bytes_init(const struct nft_ctx *ctx,const struct nlattr *const tb[], struct nft_object *obj) { - struct nft_limit *priv = nft_obj_data(obj); + struct nft_limit_priv *priv = nft_obj_data(obj);return nft_limit_init (priv, tb, false); }
@@ -303,7 +303,7 @@ static int nft_limit_obj_bytes_dump(struct sk_buff *skb, struct nft_object *obj, bool reset) { -const struct nft_limit *priv =nft_obj_data (obj); +const struct nft_limit_priv *priv =nft_obj_data (obj);return nft_limit_dump (skb, priv,NFT_LIMIT_PKT_BYTES ); }
@@ -311,7 +311,7 @@ static int nft_limit_obj_bytes_dump(struct sk_buff *skb,static struct nft_object_type nft_limit_obj_type;static const struct nft_object_ops nft_limit_obj_bytes_ops = { .type = &nft_limit_obj_type, - .size =sizeof (struct nft_limit ), + .size =sizeof (struct nft_limit_priv ), .init = nft_limit_obj_bytes_init, .eval = nft_limit_obj_bytes_eval, .dump = nft_limit_obj_bytes_dump,
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 24/32 netfilter: nft_limit: move stateful fields out of expression data
In reply to Pablo Neira AyusoIn preparation for the rule blob representation.
--- net/netfilter/nft_limit.c | 94 ++++++++++++++++++++++++++++++++++----- 1 file changed, 82 insertions(+), 12 deletions(-)
net/netfilter/nft_limit.c +82 -12
@@ -14,10 +14,14 @@#include <linux/netfilter/nf_tables.h> #include <net/netfilter/nf_tables.h> -struct nft_limit_priv { +struct nft_limit {spinlock_t lock;u64 last;u64 tokens; +}; + +struct nft_limit_priv { +struct nft_limit *limit;u64 tokens_max;u64 rate;u64 nsecs;
@@ -30,21 +34,21 @@ static inline bool nft_limit_eval(struct nft_limit_priv *priv, u64 cost)u64 now, tokens;s64 delta; -spin_lock_bh (&priv->lock); +spin_lock_bh (&priv->limit->lock); now =ktime_get_ns (); - tokens = priv->tokens + now - priv->last; + tokens = priv->limit->tokens + now - priv->limit->last;if (tokens > priv->tokens_max) tokens = priv->tokens_max; - priv->last = now; + priv->limit->last = now; delta = tokens - cost;if (delta >=0 ) { - priv->tokens = delta; -spin_unlock_bh (&priv->lock); + priv->limit->tokens = delta; +spin_unlock_bh (&priv->limit->lock);return priv->invert; } - priv->tokens = tokens; -spin_unlock_bh (&priv->lock); + priv->limit->tokens = tokens; +spin_unlock_bh (&priv->limit->lock);return !priv->invert; }
@@ -86,8 +90,12 @@ static int nft_limit_init(struct nft_limit_priv *priv, priv->rate); } - priv->tokens = tokens; - priv->tokens_max = priv->tokens; + priv->limit =kmalloc (sizeof (*priv->limit),GFP_KERNEL ); +if (!priv->limit) +return -ENOMEM ; + + priv->limit->tokens = tokens; + priv->tokens_max = priv->limit->tokens;if (tb[NFTA_LIMIT_FLAGS ]) {u32 flags =ntohl (nla_get_be32 (tb[NFTA_LIMIT_FLAGS ]));
@@ -95,8 +103,8 @@ static int nft_limit_init(struct nft_limit_priv *priv,if (flags &NFT_LIMIT_F_INV ) priv->invert = true; } - priv->last =ktime_get_ns (); -spin_lock_init (&priv->lock); + priv->limit->last =ktime_get_ns (); +spin_lock_init (&priv->limit->lock);return 0 ; }
@@ -121,6 +129,32 @@ static int nft_limit_dump(struct sk_buff *skb, const struct nft_limit_priv *privreturn -1 ; } +static void nft_limit_destroy (const struct nft_ctx *ctx, +const struct nft_limit_priv *priv) +{ +kfree (priv->limit); +} + +static int nft_limit_clone (struct nft_limit_priv *priv_dst, +const struct nft_limit_priv *priv_src) +{ + priv_dst->tokens_max = priv_src->tokens_max; + priv_dst->rate = priv_src->rate; + priv_dst->nsecs = priv_src->nsecs; + priv_dst->burst = priv_src->burst; + priv_dst->invert = priv_src->invert; + + priv_dst->limit =kmalloc (sizeof (*priv_dst->limit),GFP_ATOMIC ); +if (priv_dst->limit) +return -ENOMEM ; + +spin_lock_init (&priv_dst->limit->lock); + priv_dst->limit->tokens = priv_src->tokens_max; + priv_dst->limit->last =ktime_get_ns (); + +return 0 ; +} +struct nft_limit_priv_pkts {struct nft_limit_priv limit;u64 cost;
@@ -166,12 +200,30 @@ static int nft_limit_pkts_dump(struct sk_buff *skb, const struct nft_expr *expr)return nft_limit_dump (skb, &priv->limit,NFT_LIMIT_PKTS ); } +static void nft_limit_pkts_destroy (const struct nft_ctx *ctx, +const struct nft_expr *expr) +{ +const struct nft_limit_priv_pkts *priv =nft_expr_priv (expr); + +nft_limit_destroy (ctx, &priv->limit); +} + +static int nft_limit_pkts_clone (struct nft_expr *dst,const struct nft_expr *src) +{ +struct nft_limit_priv_pkts *priv_dst =nft_expr_priv (dst); +struct nft_limit_priv_pkts *priv_src =nft_expr_priv (src); + +return nft_limit_clone (&priv_dst->limit, &priv_src->limit); +} +static struct nft_expr_type nft_limit_type;static const struct nft_expr_ops nft_limit_pkts_ops = { .type = &nft_limit_type, .size =NFT_EXPR_SIZE (sizeof (struct nft_limit_priv_pkts )), .eval = nft_limit_pkts_eval, .init = nft_limit_pkts_init, + .destroy = nft_limit_pkts_destroy, + .clone = nft_limit_pkts_clone, .dump = nft_limit_pkts_dump, };
@@ -203,12 +255,30 @@ static int nft_limit_bytes_dump(struct sk_buff *skb,return nft_limit_dump (skb, priv,NFT_LIMIT_PKT_BYTES ); } +static void nft_limit_bytes_destroy (const struct nft_ctx *ctx, +const struct nft_expr *expr) +{ +const struct nft_limit_priv *priv =nft_expr_priv (expr); + +nft_limit_destroy (ctx, priv); +} + +static int nft_limit_bytes_clone (struct nft_expr *dst,const struct nft_expr *src) +{ +struct nft_limit_priv *priv_dst =nft_expr_priv (dst); +struct nft_limit_priv *priv_src =nft_expr_priv (src); + +return nft_limit_clone (priv_dst, priv_src); +} +static const struct nft_expr_ops nft_limit_bytes_ops = { .type = &nft_limit_type, .size =NFT_EXPR_SIZE (sizeof (struct nft_limit_priv )), .eval = nft_limit_bytes_eval, .init = nft_limit_bytes_init, .dump = nft_limit_bytes_dump, + .clone = nft_limit_bytes_clone, + .destroy = nft_limit_bytes_destroy, }; static const struct nft_expr_ops *
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 25/32 netfilter: nf_tables: add rule blob layout
In reply to Pablo Neira AyusoThis patch adds a blob layout per chain to represent the ruleset in the packet datapath.
size (unsigned long)
struct nft_rule_dp
struct nft_expr
...
struct nft_rule_dp
struct nft_expr
...
struct nft_rule_dp (is_last=1)The new structure nft_rule_dp represents the rule in a more compact way (smaller memory footprint) compared to the control-plane nft_rule structure.
The ruleset blob is a read-only data structure. The first field contains the blob size, then the rules containing expressions. There is a trailing rule which is used by the tracing infrastructure which is equivalent to the NULL rule marker in the previous representation. The blob size field does not include the size of this trailing rule marker.
The ruleset blob is generated from the commit path.
This patch reuses the infrastructure available since 0cbc06b3faba ("netfilter: nf_tables: remove synchronize_rcu in commit phase") to build the array of rules per chain.
--- include/net/netfilter/nf_tables.h | 22 ++++- net/netfilter/nf_tables_api.c | 149 ++++++++++++++++++++---------- net/netfilter/nf_tables_core.c | 41 +++++--- net/netfilter/nf_tables_trace.c | 2 +- 4 files changed, 147 insertions(+), 67 deletions(-)
include/net/netfilter/nf_tables.h +18 -4
@@ -974,6 +974,20 @@ static inline void nft_set_elem_update_expr(const struct nft_set_ext *ext,#define NFT_CHAIN_POLICY_UNSET U8_MAX +struct nft_rule_dp { +u64 is_last:1 , + dlen:12 , + handle:42 ;/* for tracing */ +unsigned char data[] + __attribute__((aligned (__alignof__(struct nft_expr )))); +}; + +struct nft_rule_blob { +unsigned long size; +unsigned char data[] + __attribute__((aligned (__alignof__(struct nft_rule_dp )))); +}; +/** * struct nft_chain - nf_tables chain *
@@ -987,8 +1001,8 @@ static inline void nft_set_elem_update_expr(const struct nft_set_ext *ext,* @name: name of the chain */ struct nft_chain { -struct nft_rule *__rcu *rules_gen_0; -struct nft_rule *__rcu *rules_gen_1; +struct nft_rule_blob __rcu *blob_gen_0; +struct nft_rule_blob __rcu *blob_gen_1;struct list_head rules;struct list_head list;struct rhlist_head rhlhead;
@@ -1003,7 +1017,7 @@ struct nft_chain {u8 *udata;/* Only used during control plane commit phase: */ -struct nft_rule **rules_next; +struct nft_rule_blob *blob_next; };int nft_chain_validate (const struct nft_ctx *ctx,const struct nft_chain *chain);
@@ -1321,7 +1335,7 @@ struct nft_traceinfo {const struct nft_pktinfo *pkt;const struct nft_base_chain *basechain;const struct nft_chain *chain; -const struct nft_rule *rule; +const struct nft_rule_dp *rule;const struct nft_verdict *verdict;enum nft_trace_types type;bool packet_dumped;
net/netfilter/nf_tables_api.c +101 -48
263 lines not on this page · raw message
net/netfilter/nf_tables_core.c +27 -14
97 lines not on this page · raw message
net/netfilter/nf_tables_trace.c +1 -1
@@ -142,7 +142,7 @@ static int nf_trace_fill_pkt_info(struct sk_buff *nlskb,static int nf_trace_fill_rule_info (struct sk_buff *nlskb,const struct nft_traceinfo *info) { -if (!info->rule) +if (!info->rule || info->rule->is_last)return 0 ;/* a continue verdict with ->type == RETURN means that this is
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 26/32 netfilter: nf_tables: add NFT_REG32_NUM
In reply to Pablo Neira AyusoAdd a definition including the maximum number of 32-bits registers that are used a scratchpad memory area to store data.
--- include/net/netfilter/nf_tables.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
include/net/netfilter/nf_tables.h +3 -1
@@ -105,6 +105,8 @@ struct nft_data { }; } __attribute__((aligned(__alignof__(u64)))); +#define NFT_REG32_NUM 20 +/** * struct nft_regs - nf_tables register set *
@@ -115,7 +117,7 @@ struct nft_data {*/ struct nft_regs {union { -u32 data[20 ]; +u32 data[NFT_REG32_NUM ];struct nft_verdict verdict; }; };
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 27/32 netfilter: nf_tables: add register tracking infrastructure
In reply to Pablo Neira AyusoThis patch adds new infrastructure to skip redundant selector store operations on the same register to achieve a performance boost from the packet path.
This is particularly noticeable in pure linear rulesets but it also helps in rulesets which are already heaving relying in maps to avoid ruleset linear inspection.
The idea is to keep data of the most recurrent store operations on register to reuse them with cmp and lookup expressions.
This infrastructure allows for dynamic ruleset updates since the ruleset blob reduction happens from the kernel.
Userspace still needs to be updated to maximize register utilization to cooperate to improve register data reuse / reduce number of store on register operations.
--- include/net/netfilter/nf_tables.h | 12 ++++++++++++ net/netfilter/nf_tables_api.c | 11 +++++++++++ 2 files changed, 23 insertions(+)
include/net/netfilter/nf_tables.h +12 -0
@@ -122,6 +122,16 @@ struct nft_regs { }; }; +struct nft_regs_track { +struct { +const struct nft_expr *selector; +const struct nft_expr *bitwise; + } regs[NFT_REG32_NUM ]; + +const struct nft_expr *cur; +const struct nft_expr *last; +}; +/* Store/load an u8, u16 or u64 integer to/from the u32 data register. * * Note, when using concatenations, register allocation happens at 32-bit
@@ -886,6 +896,8 @@ struct nft_expr_ops {int (*validate)(const struct nft_ctx *ctx,const struct nft_expr *expr,const struct nft_data **data); +bool (*reduce)(struct nft_regs_track *track, +const struct nft_expr *expr);bool (*gc)(struct net *net,const struct nft_expr *expr); int (*offload)(struct nft_offload_ctx *ctx,
net/netfilter/nf_tables_api.c +11 -0
@@ -8259,6 +8259,7 @@ EXPORT_SYMBOL_GPL(nf_tables_trans_destroy_flush_work);static int nf_tables_commit_chain_prepare (struct net *net,struct nft_chain *chain) {const struct nft_expr *expr, *last; +struct nft_regs_track track = {};unsigned int size, data_size;void *data, *data_boundary;struct nft_rule_dp *prule;
@@ -8298,7 +8299,17 @@ static int nf_tables_commit_chain_prepare(struct net *net, struct nft_chain *chaif (WARN_ON_ONCE (data > data_boundary))return -ENOMEM ; + size =0 ; + track.last = last;nft_rule_for_each_expr (expr, last, rule) { + track.cur = expr; + +if (expr->ops->reduce && + expr->ops->reduce (&track, expr)) { + expr = track.cur; +continue ; + } +if (WARN_ON_ONCE (data + expr->ops->size > data_boundary))return -ENOMEM ;
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 28/32 netfilter: nft_payload: track register operations
In reply to Pablo Neira AyusoCheck if the destination register already contains the data that this payload store expression performs. This allows to skip this redundant operation. If the destination contains a different selector, update the register tracking information.
--- net/netfilter/nft_payload.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+)
net/netfilter/nft_payload.c +30 -0
@@ -210,6 +210,34 @@ static int nft_payload_dump(struct sk_buff *skb, const struct nft_expr *expr)return -1 ; } +static bool nft_payload_reduce (struct nft_regs_track *track, +const struct nft_expr *expr) +{ +const struct nft_payload *priv =nft_expr_priv (expr); +const struct nft_payload *payload; + +if (!track->regs[priv->dreg].selector || + track->regs[priv->dreg].selector->ops != expr->ops) { + track->regs[priv->dreg].selector = expr; + track->regs[priv->dreg].bitwise =NULL ; +return false; + } + + payload =nft_expr_priv (track->regs[priv->dreg].selector); +if (priv->base != payload->base || + priv->offset != payload->offset || + priv->len != payload->len) { + track->regs[priv->dreg].selector = expr; + track->regs[priv->dreg].bitwise =NULL ; +return false; + } + +if (!track->regs[priv->dreg].bitwise) +return true; + +return false; +} +static bool nft_payload_offload_mask (struct nft_offload_reg *reg,u32 priv_len,u32 field_len) {
@@ -513,6 +541,7 @@ static const struct nft_expr_ops nft_payload_ops = { .eval = nft_payload_eval, .init = nft_payload_init, .dump = nft_payload_dump, + .reduce = nft_payload_reduce, .offload = nft_payload_offload, };
@@ -522,6 +551,7 @@ const struct nft_expr_ops nft_payload_fast_ops = { .eval = nft_payload_eval, .init = nft_payload_init, .dump = nft_payload_dump, + .reduce = nft_payload_reduce, .offload = nft_payload_offload, };
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 29/32 netfilter: nft_meta: track register operations
In reply to Pablo Neira AyusoCheck if the destination register already contains the data that this meta store expression performs. This allows to skip this redundant operation. If the destination contains a different selector, update the register tracking information.
--- net/netfilter/nft_meta.c | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+)
net/netfilter/nft_meta.c +28 -0
@@ -750,12 +750,40 @@ static int nft_meta_get_offload(struct nft_offload_ctx *ctx,return 0 ; } +static bool nft_meta_get_reduce (struct nft_regs_track *track, +const struct nft_expr *expr) +{ +const struct nft_meta *priv =nft_expr_priv (expr); +const struct nft_meta *meta; + +if (!track->regs[priv->dreg].selector || + track->regs[priv->dreg].selector->ops != expr->ops) { + track->regs[priv->dreg].selector = expr; + track->regs[priv->dreg].bitwise =NULL ; +return false; + } + + meta =nft_expr_priv (track->regs[priv->dreg].selector); +if (priv->key != meta->key || + priv->dreg != meta->dreg) { + track->regs[priv->dreg].selector = expr; + track->regs[priv->dreg].bitwise =NULL ; +return false; + } + +if (!track->regs[priv->dreg].bitwise) +return true; + +return false; +} +static const struct nft_expr_ops nft_meta_get_ops = { .type = &nft_meta_type, .size =NFT_EXPR_SIZE (sizeof (struct nft_meta )), .eval = nft_meta_get_eval, .init = nft_meta_get_init, .dump = nft_meta_get_dump, + .reduce = nft_meta_get_reduce, .validate = nft_meta_get_validate, .offload = nft_meta_get_offload, };
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 30/32 netfilter: nft_bitwise: track register operations
In reply to Pablo Neira AyusoCheck if the destination register already contains the data that this bitwise expression performs. This allows to skip this redundant operation.
If the destination contains a different bitwise operation, cancel the register tracking information. If the destination contains no bitwise operation, update the register tracking information.
Update the payload and meta expression to check if this bitwise operation has been already performed on the register. Hence, both the payload/meta and the bitwise expressions are reduced.
There is also a special case: If source register != destination register and source register is not updated by a previous bitwise operation, then transfer selector from the source register to the destination register.
--- include/net/netfilter/nf_tables.h | 2 + net/netfilter/nft_bitwise.c | 95 +++++++++++++++++++++++++++++++ net/netfilter/nft_meta.c | 2 +- net/netfilter/nft_payload.c | 2 +- 4 files changed, 99 insertions(+), 2 deletions(-)
include/net/netfilter/nf_tables.h +2 -0
@@ -358,6 +358,8 @@ int nft_expr_clone(struct nft_expr *dst, struct nft_expr *src);void nft_expr_destroy (const struct nft_ctx *ctx,struct nft_expr *expr);int nft_expr_dump (struct sk_buff *skb,unsigned int attr,const struct nft_expr *expr); +bool nft_expr_reduce_bitwise (struct nft_regs_track *track, +const struct nft_expr *expr);struct nft_set_ext ;
net/netfilter/nft_bitwise.c +95 -0
122 lines not on this page · raw message
net/netfilter/nft_meta.c +1 -1
@@ -774,7 +774,7 @@ static bool nft_meta_get_reduce(struct nft_regs_track *track,if (!track->regs[priv->dreg].bitwise)return true; -return false; +return nft_expr_reduce_bitwise (track, expr); }static const struct nft_expr_ops nft_meta_get_ops = {
net/netfilter/nft_payload.c +1 -1
@@ -235,7 +235,7 @@ static bool nft_payload_reduce(struct nft_regs_track *track,if (!track->regs[priv->dreg].bitwise)return true; -return false; +return nft_expr_reduce_bitwise (track, expr); } static bool nft_payload_offload_mask(struct nft_offload_reg *reg,
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 31/32 netfilter: nft_payload: cancel register tracking after payload update
In reply to Pablo Neira AyusoThe payload expression might mangle the packet, cancel register tracking since any payload data in the registers is stale.
Finer grain register tracking cancellation by inspecting the payload base, offset and length on the register is also possible.
--- net/netfilter/nft_payload.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+)
net/netfilter/nft_payload.c +21 -0
@@ -798,12 +798,33 @@ static int nft_payload_set_dump(struct sk_buff *skb, const struct nft_expr *exprreturn -1 ; } +static bool nft_payload_set_reduce (struct nft_regs_track *track, +const struct nft_expr *expr) +{ +int i; + +for (i =0 ; i <NFT_REG32_NUM ; i++) { +if (!track->regs[i].selector) +continue ; + +if (track->regs[i].selector->ops != &nft_payload_ops && + track->regs[i].selector->ops != &nft_payload_fast_ops) +continue ; + + track->regs[i].selector =NULL ; + track->regs[i].bitwise =NULL ; + } + +return false; +} +static const struct nft_expr_ops nft_payload_set_ops = { .type = &nft_payload_type, .size =NFT_EXPR_SIZE (sizeof (struct nft_payload_set )), .eval = nft_payload_set_eval, .init = nft_payload_set_init, .dump = nft_payload_set_dump, + .reduce = nft_payload_set_reduce, }; static const struct nft_expr_ops *
Signature · 1 line
2.30.2
Pablo Neira Ayuso
maintainerPATCH 32/32 netfilter: nft_meta: cancel register tracking after meta update
In reply to Pablo Neira AyusoThe meta expression might mangle the packet metadata, cancel register tracking since any metadata in the registers is stale.
Finer grain register tracking cancellation by inspecting the meta type on the register is also possible.
--- net/bridge/netfilter/nft_meta_bridge.c | 20 ++++++++++++++++++++ net/netfilter/nft_meta.c | 20 ++++++++++++++++++++ 2 files changed, 40 insertions(+)
net/bridge/netfilter/nft_meta_bridge.c +20 -0
@@ -100,6 +100,25 @@ static const struct nft_expr_ops nft_meta_bridge_get_ops = { .dump = nft_meta_get_dump, }; +static bool nft_meta_bridge_set_reduce (struct nft_regs_track *track, +const struct nft_expr *expr) +{ +int i; + +for (i =0 ; i <NFT_REG32_NUM ; i++) { +if (!track->regs[i].selector) +continue ; + +if (track->regs[i].selector->ops != &nft_meta_bridge_get_ops) +continue ; + + track->regs[i].selector =NULL ; + track->regs[i].bitwise =NULL ; + } + +return false; +} +static const struct nft_expr_ops nft_meta_bridge_set_ops = { .type = &nft_meta_bridge_type, .size =NFT_EXPR_SIZE (sizeof (struct nft_meta )),
@@ -107,6 +126,7 @@ static const struct nft_expr_ops nft_meta_bridge_set_ops = { .init = nft_meta_set_init, .destroy = nft_meta_set_destroy, .dump = nft_meta_set_dump, + .reduce = nft_meta_bridge_set_reduce, .validate = nft_meta_set_validate, };
net/netfilter/nft_meta.c +20 -0
@@ -788,6 +788,25 @@ static const struct nft_expr_ops nft_meta_get_ops = { .offload = nft_meta_get_offload, }; +static bool nft_meta_set_reduce (struct nft_regs_track *track, +const struct nft_expr *expr) +{ +int i; + +for (i =0 ; i <NFT_REG32_NUM ; i++) { +if (!track->regs[i].selector) +continue ; + +if (track->regs[i].selector->ops != &nft_meta_get_ops) +continue ; + + track->regs[i].selector =NULL ; + track->regs[i].bitwise =NULL ; + } + +return false; +} +static const struct nft_expr_ops nft_meta_set_ops = { .type = &nft_meta_type, .size =NFT_EXPR_SIZE (sizeof (struct nft_meta )),
@@ -795,6 +814,7 @@ static const struct nft_expr_ops nft_meta_set_ops = { .init = nft_meta_set_init, .destroy = nft_meta_set_destroy, .dump = nft_meta_set_dump, + .reduce = nft_meta_set_reduce, .validate = nft_meta_set_validate, };
Signature · 1 line
2.30.2