From 97d4d394b58777f7056ebba8ffdb4002d0563259 Mon Sep 17 00:00:00 2001 From: Stefano Brivio Date: Thu, 24 Nov 2022 13:04:37 +0100 Subject: [PATCH 1/4] netfilter: nft_set_pipapo: Actually validate intervals in fields after the first one Embarrassingly, nft_pipapo_insert() checked for interval validity in the first field only. The start_p and end_p pointers were reset to key data from the first field at every iteration of the loop which was supposed to go over the set fields. Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges") Reported-by: Pablo Neira Ayuso Signed-off-by: Stefano Brivio Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_set_pipapo.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/net/netfilter/nft_set_pipapo.c b/net/netfilter/nft_set_pipapo.c index 4f9299b9dcdd..06d46d182634 100644 --- a/net/netfilter/nft_set_pipapo.c +++ b/net/netfilter/nft_set_pipapo.c @@ -1162,6 +1162,7 @@ static int nft_pipapo_insert(const struct net *net, const struct nft_set *set, struct nft_pipapo_match *m = priv->clone; u8 genmask = nft_genmask_next(net); struct nft_pipapo_field *f; + const u8 *start_p, *end_p; int i, bsize_max, err = 0; if (nft_set_ext_exists(ext, NFT_SET_EXT_KEY_END)) @@ -1202,9 +1203,9 @@ static int nft_pipapo_insert(const struct net *net, const struct nft_set *set, } /* Validate */ + start_p = start; + end_p = end; nft_pipapo_for_each_field(f, i, m) { - const u8 *start_p = start, *end_p = end; - if (f->rules >= (unsigned long)NFT_PIPAPO_RULE0_MAX) return -ENOSPC; From a81047154e7ce4eb8769d5d21adcbc9693542a79 Mon Sep 17 00:00:00 2001 From: Xin Long Date: Thu, 24 Nov 2022 12:54:10 -0500 Subject: [PATCH 2/4] netfilter: flowtable_offload: fix using __this_cpu_add in preemptible flow_offload_queue_work() can be called in workqueue without bh disabled, like the call trace showed in my act_ct testing, calling NF_FLOW_TABLE_STAT_INC() there would cause a call trace: BUG: using __this_cpu_add() in preemptible [00000000] code: kworker/u4:0/138560 caller is flow_offload_queue_work+0xec/0x1b0 [nf_flow_table] Workqueue: act_ct_workqueue tcf_ct_flow_table_cleanup_work [act_ct] Call Trace: dump_stack_lvl+0x33/0x46 check_preemption_disabled+0xc3/0xf0 flow_offload_queue_work+0xec/0x1b0 [nf_flow_table] nf_flow_table_iterate+0x138/0x170 [nf_flow_table] nf_flow_table_free+0x140/0x1a0 [nf_flow_table] tcf_ct_flow_table_cleanup_work+0x2f/0x2b0 [act_ct] process_one_work+0x6a3/0x1030 worker_thread+0x8a/0xdf0 This patch fixes it by using NF_FLOW_TABLE_STAT_INC_ATOMIC() instead in flow_offload_queue_work(). Note that for FLOW_CLS_REPLACE branch in flow_offload_queue_work(), it may not be called in preemptible path, but it's good to use NF_FLOW_TABLE_STAT_INC_ATOMIC() for all cases in flow_offload_queue_work(). Fixes: b038177636f8 ("netfilter: nf_flow_table: count pending offload workqueue tasks") Signed-off-by: Xin Long Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_flow_table_offload.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/net/netfilter/nf_flow_table_offload.c b/net/netfilter/nf_flow_table_offload.c index 00b522890d77..0fdcdb2c9ae4 100644 --- a/net/netfilter/nf_flow_table_offload.c +++ b/net/netfilter/nf_flow_table_offload.c @@ -997,13 +997,13 @@ static void flow_offload_queue_work(struct flow_offload_work *offload) struct net *net = read_pnet(&offload->flowtable->net); if (offload->cmd == FLOW_CLS_REPLACE) { - NF_FLOW_TABLE_STAT_INC(net, count_wq_add); + NF_FLOW_TABLE_STAT_INC_ATOMIC(net, count_wq_add); queue_work(nf_flow_offload_add_wq, &offload->work); } else if (offload->cmd == FLOW_CLS_DESTROY) { - NF_FLOW_TABLE_STAT_INC(net, count_wq_del); + NF_FLOW_TABLE_STAT_INC_ATOMIC(net, count_wq_del); queue_work(nf_flow_offload_del_wq, &offload->work); } else { - NF_FLOW_TABLE_STAT_INC(net, count_wq_stats); + NF_FLOW_TABLE_STAT_INC_ATOMIC(net, count_wq_stats); queue_work(nf_flow_offload_stats_wq, &offload->work); } } From 9464d0b68f11a9bc768370c3260ec02b3550447b Mon Sep 17 00:00:00 2001 From: Xin Long Date: Thu, 24 Nov 2022 12:21:46 -0500 Subject: [PATCH 3/4] netfilter: conntrack: fix using __this_cpu_add in preemptible Currently in nf_conntrack_hash_check_insert(), when it fails in nf_ct_ext_valid_pre/post(), NF_CT_STAT_INC() will be called in the preemptible context, a call trace can be triggered: BUG: using __this_cpu_add() in preemptible [00000000] code: conntrack/1636 caller is nf_conntrack_hash_check_insert+0x45/0x430 [nf_conntrack] Call Trace: dump_stack_lvl+0x33/0x46 check_preemption_disabled+0xc3/0xf0 nf_conntrack_hash_check_insert+0x45/0x430 [nf_conntrack] ctnetlink_create_conntrack+0x3cd/0x4e0 [nf_conntrack_netlink] ctnetlink_new_conntrack+0x1c0/0x450 [nf_conntrack_netlink] nfnetlink_rcv_msg+0x277/0x2f0 [nfnetlink] netlink_rcv_skb+0x50/0x100 nfnetlink_rcv+0x65/0x144 [nfnetlink] netlink_unicast+0x1ae/0x290 netlink_sendmsg+0x257/0x4f0 sock_sendmsg+0x5f/0x70 This patch is to fix it by changing to use NF_CT_STAT_INC_ATOMIC() for nf_ct_ext_valid_pre/post() check in nf_conntrack_hash_check_insert(), as well as nf_ct_ext_valid_post() in __nf_conntrack_confirm(). Note that nf_ct_ext_valid_pre() check in __nf_conntrack_confirm() is safe to use NF_CT_STAT_INC(), as it's under local_bh_disable(). Fixes: c56716c69ce1 ("netfilter: extensions: introduce extension genid count") Signed-off-by: Xin Long Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_conntrack_core.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index 2692139ce417..23b3fedd619a 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -891,7 +891,7 @@ nf_conntrack_hash_check_insert(struct nf_conn *ct) zone = nf_ct_zone(ct); if (!nf_ct_ext_valid_pre(ct->ext)) { - NF_CT_STAT_INC(net, insert_failed); + NF_CT_STAT_INC_ATOMIC(net, insert_failed); return -ETIMEDOUT; } @@ -938,7 +938,7 @@ nf_conntrack_hash_check_insert(struct nf_conn *ct) if (!nf_ct_ext_valid_post(ct->ext)) { nf_ct_kill(ct); - NF_CT_STAT_INC(net, drop); + NF_CT_STAT_INC_ATOMIC(net, drop); return -ETIMEDOUT; } @@ -1275,7 +1275,7 @@ chaintoolong: */ if (!nf_ct_ext_valid_post(ct->ext)) { nf_ct_kill(ct); - NF_CT_STAT_INC(net, drop); + NF_CT_STAT_INC_ATOMIC(net, drop); return NF_DROP; } From 1feeae071507ad65cf9f462a1bdd543a4bf89e71 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Mon, 28 Nov 2022 10:58:53 +0100 Subject: [PATCH 4/4] netfilter: ctnetlink: fix compilation warning after data race fixes in ct mark All warnings (new ones prefixed by >>): net/netfilter/nf_conntrack_netlink.c: In function '__ctnetlink_glue_build': >> net/netfilter/nf_conntrack_netlink.c:2674:13: warning: unused variable 'mark' [-Wunused-variable] 2674 | u32 mark; | ^~~~ Fixes: 52d1aa8b8249 ("netfilter: conntrack: Fix data-races around ct mark") Reported-by: kernel test robot Tested-by: Ivan Babrou Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_conntrack_netlink.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c index d71150a40fb0..1286ae7d4609 100644 --- a/net/netfilter/nf_conntrack_netlink.c +++ b/net/netfilter/nf_conntrack_netlink.c @@ -328,8 +328,13 @@ nla_put_failure: } #ifdef CONFIG_NF_CONNTRACK_MARK -static int ctnetlink_dump_mark(struct sk_buff *skb, u32 mark) +static int ctnetlink_dump_mark(struct sk_buff *skb, const struct nf_conn *ct) { + u32 mark = READ_ONCE(ct->mark); + + if (!mark) + return 0; + if (nla_put_be32(skb, CTA_MARK, htonl(mark))) goto nla_put_failure; return 0; @@ -543,7 +548,7 @@ static int ctnetlink_dump_extinfo(struct sk_buff *skb, static int ctnetlink_dump_info(struct sk_buff *skb, struct nf_conn *ct) { if (ctnetlink_dump_status(skb, ct) < 0 || - ctnetlink_dump_mark(skb, READ_ONCE(ct->mark)) < 0 || + ctnetlink_dump_mark(skb, ct) < 0 || ctnetlink_dump_secctx(skb, ct) < 0 || ctnetlink_dump_id(skb, ct) < 0 || ctnetlink_dump_use(skb, ct) < 0 || @@ -722,7 +727,6 @@ ctnetlink_conntrack_event(unsigned int events, const struct nf_ct_event *item) struct sk_buff *skb; unsigned int type; unsigned int flags = 0, group; - u32 mark; int err; if (events & (1 << IPCT_DESTROY)) { @@ -827,9 +831,8 @@ ctnetlink_conntrack_event(unsigned int events, const struct nf_ct_event *item) } #ifdef CONFIG_NF_CONNTRACK_MARK - mark = READ_ONCE(ct->mark); - if ((events & (1 << IPCT_MARK) || mark) && - ctnetlink_dump_mark(skb, mark) < 0) + if (events & (1 << IPCT_MARK) && + ctnetlink_dump_mark(skb, ct) < 0) goto nla_put_failure; #endif nlmsg_end(skb, nlh); @@ -2671,7 +2674,6 @@ static int __ctnetlink_glue_build(struct sk_buff *skb, struct nf_conn *ct) { const struct nf_conntrack_zone *zone; struct nlattr *nest_parms; - u32 mark; zone = nf_ct_zone(ct); @@ -2733,8 +2735,7 @@ static int __ctnetlink_glue_build(struct sk_buff *skb, struct nf_conn *ct) goto nla_put_failure; #ifdef CONFIG_NF_CONNTRACK_MARK - mark = READ_ONCE(ct->mark); - if (mark && ctnetlink_dump_mark(skb, mark) < 0) + if (ctnetlink_dump_mark(skb, ct) < 0) goto nla_put_failure; #endif if (ctnetlink_dump_labels(skb, ct) < 0)