[PATCH 0/2] Netfilter/IPVS fixes for net

[PATCH 0/2] Netfilter/IPVS fixes for net

[Pablo Neira Ayuso]

Hi,

The following patchset contains Netfilter/IPVS fixes for net:

1) Fix NAT hook deletion when table is dormant, from Florian Westphal.

2) Fix IPVS sync stalls, from guodeqing.

Please, pull these changes from:

  git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf.git

Thank you.

----------------------------------------------------------------

The following changes since commit 1d61e21852d3161f234b9656797669fe185c251b:

  qed: Disable "MFW indication via attention" SPAM every 5 minutes (2020-07-14 15:15:44 -0700)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf.git HEAD

for you to fetch changes up to 8210e344ccb798c672ab237b1a4f241bda08909b:

  ipvs: fix the connection sync failed in some cases (2020-07-22 01:21:34 +0200)

----------------------------------------------------------------
Florian Westphal (1):
      netfilter: nf_tables: fix nat hook table deletion

guodeqing (1):
      ipvs: fix the connection sync failed in some cases

 net/netfilter/ipvs/ip_vs_sync.c | 12 ++++++++----
 net/netfilter/nf_tables_api.c   | 41 ++++++++++++++---------------------------
 2 files changed, 22 insertions(+), 31 deletions(-)

[PATCH 1/2] netfilter: nf_tables: fix nat hook table deletion

[Pablo Neira Ayuso]

From: Florian Westphal <fw@strlen.de>

sybot came up with following transaction:
 add table ip syz0
 add chain ip syz0 syz2 { type nat hook prerouting priority 0; policy accept; }
 add table ip syz0 { flags dormant; }
 delete chain ip syz0 syz2
 delete table ip syz0

which yields:
hook not found, pf 2 num 0
WARNING: CPU: 0 PID: 6775 at net/netfilter/core.c:413 __nf_unregister_net_hook+0x3e6/0x4a0 net/netfilter/core.c:413
[..]
 nft_unregister_basechain_hooks net/netfilter/nf_tables_api.c:206 [inline]
 nft_table_disable net/netfilter/nf_tables_api.c:835 [inline]
 nf_tables_table_disable net/netfilter/nf_tables_api.c:868 [inline]
 nf_tables_commit+0x32d3/0x4d70 net/netfilter/nf_tables_api.c:7550
 nfnetlink_rcv_batch net/netfilter/nfnetlink.c:486 [inline]
 nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:544 [inline]
 nfnetlink_rcv+0x14a5/0x1e50 net/netfilter/nfnetlink.c:562
 netlink_unicast_kernel net/netlink/af_netlink.c:1303 [inline]

Problem is that when I added ability to override base hook registration
to make nat basechains register with the nat core instead of netfilter
core, I forgot to update nft_table_disable() to use that instead of
the 'raw' hook register interface.

In syzbot transaction, the basechain is of 'nat' type. Its registered
with the nat core.  The switch to 'dormant mode' attempts to delete from
netfilter core instead.

After updating nft_table_disable/enable to use the correct helper,
nft_(un)register_basechain_hooks can be folded into the only remaining
caller.

Because nft_trans_table_enable() won't do anything when the DORMANT flag
is set, remove the flag first, then re-add it in case re-enablement
fails, else this patch breaks sequence:

add table ip x { flags dormant; }
/* add base chains */
add table ip x

The last 'add' will remove the dormant flags, but won't have any other
effect -- base chains are not registered.
Then, next 'set dormant flag' will create another 'hook not found'
splat.

Reported-by: syzbot+2570f2c036e3da5db176@syzkaller.appspotmail.com
Fixes: 4e25ceb80b58 ("netfilter: nf_tables: allow chain type to override hook register")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 net/netfilter/nf_tables_api.c | 41 ++++++++++++-----------------------
 1 file changed, 14 insertions(+), 27 deletions(-)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 7647ecfa0d40..88325b264737 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -188,24 +188,6 @@ static void nft_netdev_unregister_hooks(struct net *net,
 		nf_unregister_net_hook(net, &hook->ops);
 }
 
-static int nft_register_basechain_hooks(struct net *net, int family,
-					struct nft_base_chain *basechain)
-{
-	if (family == NFPROTO_NETDEV)
-		return nft_netdev_register_hooks(net, &basechain->hook_list);
-
-	return nf_register_net_hook(net, &basechain->ops);
-}
-
-static void nft_unregister_basechain_hooks(struct net *net, int family,
-					   struct nft_base_chain *basechain)
-{
-	if (family == NFPROTO_NETDEV)
-		nft_netdev_unregister_hooks(net, &basechain->hook_list);
-	else
-		nf_unregister_net_hook(net, &basechain->ops);
-}
-
 static int nf_tables_register_hook(struct net *net,
 				   const struct nft_table *table,
 				   struct nft_chain *chain)
@@ -223,7 +205,10 @@ static int nf_tables_register_hook(struct net *net,
 	if (basechain->type->ops_register)
 		return basechain->type->ops_register(net, ops);
 
-	return nft_register_basechain_hooks(net, table->family, basechain);
+	if (table->family == NFPROTO_NETDEV)
+		return nft_netdev_register_hooks(net, &basechain->hook_list);
+
+	return nf_register_net_hook(net, &basechain->ops);
 }
 
 static void nf_tables_unregister_hook(struct net *net,
@@ -242,7 +227,10 @@ static void nf_tables_unregister_hook(struct net *net,
 	if (basechain->type->ops_unregister)
 		return basechain->type->ops_unregister(net, ops);
 
-	nft_unregister_basechain_hooks(net, table->family, basechain);
+	if (table->family == NFPROTO_NETDEV)
+		nft_netdev_unregister_hooks(net, &basechain->hook_list);
+	else
+		nf_unregister_net_hook(net, &basechain->ops);
 }
 
 static int nft_trans_table_add(struct nft_ctx *ctx, int msg_type)
@@ -832,8 +820,7 @@ static void nft_table_disable(struct net *net, struct nft_table *table, u32 cnt)
 		if (cnt && i++ == cnt)
 			break;
 
-		nft_unregister_basechain_hooks(net, table->family,
-					       nft_base_chain(chain));
+		nf_tables_unregister_hook(net, table, chain);
 	}
 }
 
@@ -848,8 +835,7 @@ static int nf_tables_table_enable(struct net *net, struct nft_table *table)
 		if (!nft_is_base_chain(chain))
 			continue;
 
-		err = nft_register_basechain_hooks(net, table->family,
-						   nft_base_chain(chain));
+		err = nf_tables_register_hook(net, table, chain);
 		if (err < 0)
 			goto err_register_hooks;
 
@@ -894,11 +880,12 @@ static int nf_tables_updtable(struct nft_ctx *ctx)
 		nft_trans_table_enable(trans) = false;
 	} else if (!(flags & NFT_TABLE_F_DORMANT) &&
 		   ctx->table->flags & NFT_TABLE_F_DORMANT) {
+		ctx->table->flags &= ~NFT_TABLE_F_DORMANT;
 		ret = nf_tables_table_enable(ctx->net, ctx->table);
-		if (ret >= 0) {
-			ctx->table->flags &= ~NFT_TABLE_F_DORMANT;
+		if (ret >= 0)
 			nft_trans_table_enable(trans) = true;
-		}
+		else
+			ctx->table->flags |= NFT_TABLE_F_DORMANT;
 	}
 	if (ret < 0)
 		goto err;
-- 
2.20.1

[PATCH 2/2] ipvs: fix the connection sync failed in some cases

[Pablo Neira Ayuso]

From: guodeqing <geffrey.guo@huawei.com>

The sync_thread_backup only checks sk_receive_queue is empty or not,
there is a situation which cannot sync the connection entries when
sk_receive_queue is empty and sk_rmem_alloc is larger than sk_rcvbuf,
the sync packets are dropped in __udp_enqueue_schedule_skb, this is
because the packets in reader_queue is not read, so the rmem is
not reclaimed.

Here I add the check of whether the reader_queue of the udp sock is
empty or not to solve this problem.

Fixes: 2276f58ac589 ("udp: use a separate rx queue for packet reception")
Reported-by: zhouxudong <zhouxudong8@huawei.com>
Signed-off-by: guodeqing <geffrey.guo@huawei.com>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 net/netfilter/ipvs/ip_vs_sync.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/net/netfilter/ipvs/ip_vs_sync.c b/net/netfilter/ipvs/ip_vs_sync.c
index 605e0f68f8bd..2b8abbfe018c 100644
--- a/net/netfilter/ipvs/ip_vs_sync.c
+++ b/net/netfilter/ipvs/ip_vs_sync.c
@@ -1717,6 +1717,8 @@ static int sync_thread_backup(void *data)
 {
 	struct ip_vs_sync_thread_data *tinfo = data;
 	struct netns_ipvs *ipvs = tinfo->ipvs;
+	struct sock *sk = tinfo->sock->sk;
+	struct udp_sock *up = udp_sk(sk);
 	int len;
 
 	pr_info("sync thread started: state = BACKUP, mcast_ifn = %s, "
@@ -1724,12 +1726,14 @@ static int sync_thread_backup(void *data)
 		ipvs->bcfg.mcast_ifn, ipvs->bcfg.syncid, tinfo->id);
 
 	while (!kthread_should_stop()) {
-		wait_event_interruptible(*sk_sleep(tinfo->sock->sk),
-			 !skb_queue_empty(&tinfo->sock->sk->sk_receive_queue)
-			 || kthread_should_stop());
+		wait_event_interruptible(*sk_sleep(sk),
+					 !skb_queue_empty_lockless(&sk->sk_receive_queue) ||
+					 !skb_queue_empty_lockless(&up->reader_queue) ||
+					 kthread_should_stop());
 
 		/* do we have data now? */
-		while (!skb_queue_empty(&(tinfo->sock->sk->sk_receive_queue))) {
+		while (!skb_queue_empty_lockless(&sk->sk_receive_queue) ||
+		       !skb_queue_empty_lockless(&up->reader_queue)) {
 			len = ip_vs_receive(tinfo->sock, tinfo->buf,
 					ipvs->bcfg.sync_maxlen);
 			if (len <= 0) {
-- 
2.20.1

Re: [PATCH 0/2] Netfilter/IPVS fixes for net

[David Miller]

From: Pablo Neira Ayuso <pablo@netfilter.org>
Date: Fri, 24 Jul 2020 00:35:06 +0200

> The following patchset contains Netfilter/IPVS fixes for net:
> 
> 1) Fix NAT hook deletion when table is dormant, from Florian Westphal.
> 
> 2) Fix IPVS sync stalls, from guodeqing.
> 
> Please, pull these changes from:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf.git

Pulled, thank you.