diff options
Diffstat (limited to 'net/bridge/br_netfilter_hooks.c')
| -rw-r--r-- | net/bridge/br_netfilter_hooks.c | 138 |
1 files changed, 130 insertions, 8 deletions
diff --git a/net/bridge/br_netfilter_hooks.c b/net/bridge/br_netfilter_hooks.c index 6adcb45bca75..35e10c5a766d 100644 --- a/net/bridge/br_netfilter_hooks.c +++ b/net/bridge/br_netfilter_hooks.c @@ -43,6 +43,10 @@ #include <linux/sysctl.h> #endif +#if IS_ENABLED(CONFIG_NF_CONNTRACK) +#include <net/netfilter/nf_conntrack_core.h> +#endif + static unsigned int brnf_net_id __read_mostly; struct brnf_net { @@ -279,8 +283,17 @@ int br_nf_pre_routing_finish_bridge(struct net *net, struct sock *sk, struct sk_ if ((READ_ONCE(neigh->nud_state) & NUD_CONNECTED) && READ_ONCE(neigh->hh.hh_len)) { + struct net_device *br_indev; + + br_indev = nf_bridge_get_physindev(skb, net); + if (!br_indev) { + neigh_release(neigh); + goto free_skb; + } + neigh_hh_bridge(&neigh->hh, skb); - skb->dev = nf_bridge->physindev; + skb->dev = br_indev; + ret = br_handle_frame_finish(net, sk, skb); } else { /* the neighbour function below overwrites the complete @@ -352,12 +365,18 @@ br_nf_ipv4_daddr_was_changed(const struct sk_buff *skb, */ static int br_nf_pre_routing_finish(struct net *net, struct sock *sk, struct sk_buff *skb) { - struct net_device *dev = skb->dev; + struct net_device *dev = skb->dev, *br_indev; struct iphdr *iph = ip_hdr(skb); struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); struct rtable *rt; int err; + br_indev = nf_bridge_get_physindev(skb, net); + if (!br_indev) { + kfree_skb(skb); + return 0; + } + nf_bridge->frag_max_size = IPCB(skb)->frag_max_size; if (nf_bridge->pkt_otherhost) { @@ -397,7 +416,7 @@ free_skb: } else { if (skb_dst(skb)->dev == dev) { bridged_dnat: - skb->dev = nf_bridge->physindev; + skb->dev = br_indev; nf_bridge_update_protocol(skb); nf_bridge_push_encap_header(skb); br_nf_hook_thresh(NF_BR_PRE_ROUTING, @@ -410,7 +429,7 @@ bridged_dnat: skb->pkt_type = PACKET_HOST; } } else { - rt = bridge_parent_rtable(nf_bridge->physindev); + rt = bridge_parent_rtable(br_indev); if (!rt) { kfree_skb(skb); return 0; @@ -419,7 +438,7 @@ bridged_dnat: skb_dst_set_noref(skb, &rt->dst); } - skb->dev = nf_bridge->physindev; + skb->dev = br_indev; nf_bridge_update_protocol(skb); nf_bridge_push_encap_header(skb); br_nf_hook_thresh(NF_BR_PRE_ROUTING, net, sk, skb, skb->dev, NULL, @@ -456,7 +475,7 @@ struct net_device *setup_pre_routing(struct sk_buff *skb, const struct net *net) } nf_bridge->in_prerouting = 1; - nf_bridge->physindev = skb->dev; + nf_bridge->physinif = skb->dev->ifindex; skb->dev = brnf_get_logical_dev(skb, skb->dev, net); if (skb->protocol == htons(ETH_P_8021Q)) @@ -538,6 +557,90 @@ static unsigned int br_nf_pre_routing(void *priv, return NF_STOLEN; } +#if IS_ENABLED(CONFIG_NF_CONNTRACK) +/* conntracks' nf_confirm logic cannot handle cloned skbs referencing + * the same nf_conn entry, which will happen for multicast (broadcast) + * Frames on bridges. + * + * Example: + * macvlan0 + * br0 + * ethX ethY + * + * ethX (or Y) receives multicast or broadcast packet containing + * an IP packet, not yet in conntrack table. + * + * 1. skb passes through bridge and fake-ip (br_netfilter)Prerouting. + * -> skb->_nfct now references a unconfirmed entry + * 2. skb is broad/mcast packet. bridge now passes clones out on each bridge + * interface. + * 3. skb gets passed up the stack. + * 4. In macvlan case, macvlan driver retains clone(s) of the mcast skb + * and schedules a work queue to send them out on the lower devices. + * + * The clone skb->_nfct is not a copy, it is the same entry as the + * original skb. The macvlan rx handler then returns RX_HANDLER_PASS. + * 5. Normal conntrack hooks (in NF_INET_LOCAL_IN) confirm the orig skb. + * + * The Macvlan broadcast worker and normal confirm path will race. + * + * This race will not happen if step 2 already confirmed a clone. In that + * case later steps perform skb_clone() with skb->_nfct already confirmed (in + * hash table). This works fine. + * + * But such confirmation won't happen when eb/ip/nftables rules dropped the + * packets before they reached the nf_confirm step in postrouting. + * + * Work around this problem by explicit confirmation of the entry at + * LOCAL_IN time, before upper layer has a chance to clone the unconfirmed + * entry. + * + */ +static unsigned int br_nf_local_in(void *priv, + struct sk_buff *skb, + const struct nf_hook_state *state) +{ + struct nf_conntrack *nfct = skb_nfct(skb); + const struct nf_ct_hook *ct_hook; + struct nf_conn *ct; + int ret; + + if (!nfct || skb->pkt_type == PACKET_HOST) + return NF_ACCEPT; + + ct = container_of(nfct, struct nf_conn, ct_general); + if (likely(nf_ct_is_confirmed(ct))) + return NF_ACCEPT; + + WARN_ON_ONCE(skb_shared(skb)); + WARN_ON_ONCE(refcount_read(&nfct->use) != 1); + + /* We can't call nf_confirm here, it would create a dependency + * on nf_conntrack module. + */ + ct_hook = rcu_dereference(nf_ct_hook); + if (!ct_hook) { + skb->_nfct = 0ul; + nf_conntrack_put(nfct); + return NF_ACCEPT; + } + + nf_bridge_pull_encap_header(skb); + ret = ct_hook->confirm(skb); + switch (ret & NF_VERDICT_MASK) { + case NF_STOLEN: + return NF_STOLEN; + default: + nf_bridge_push_encap_header(skb); + break; + } + + ct = container_of(nfct, struct nf_conn, ct_general); + WARN_ON_ONCE(!nf_ct_is_confirmed(ct)); + + return ret; +} +#endif /* PF_BRIDGE/FORWARD *************************************************/ static int br_nf_forward_finish(struct net *net, struct sock *sk, struct sk_buff *skb) @@ -553,7 +656,11 @@ static int br_nf_forward_finish(struct net *net, struct sock *sk, struct sk_buff if (skb->protocol == htons(ETH_P_IPV6)) nf_bridge->frag_max_size = IP6CB(skb)->frag_max_size; - in = nf_bridge->physindev; + in = nf_bridge_get_physindev(skb, net); + if (!in) { + kfree_skb(skb); + return 0; + } if (nf_bridge->pkt_otherhost) { skb->pkt_type = PACKET_OTHERHOST; nf_bridge->pkt_otherhost = false; @@ -899,6 +1006,13 @@ static unsigned int ip_sabotage_in(void *priv, static void br_nf_pre_routing_finish_bridge_slow(struct sk_buff *skb) { struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); + struct net_device *br_indev; + + br_indev = nf_bridge_get_physindev(skb, dev_net(skb->dev)); + if (!br_indev) { + kfree_skb(skb); + return; + } skb_pull(skb, ETH_HLEN); nf_bridge->bridged_dnat = 0; @@ -908,7 +1022,7 @@ static void br_nf_pre_routing_finish_bridge_slow(struct sk_buff *skb) skb_copy_to_linear_data_offset(skb, -(ETH_HLEN - ETH_ALEN), nf_bridge->neigh_header, ETH_HLEN - ETH_ALEN); - skb->dev = nf_bridge->physindev; + skb->dev = br_indev; nf_bridge->physoutdev = NULL; br_handle_frame_finish(dev_net(skb->dev), NULL, skb); @@ -938,6 +1052,14 @@ static const struct nf_hook_ops br_nf_ops[] = { .hooknum = NF_BR_PRE_ROUTING, .priority = NF_BR_PRI_BRNF, }, +#if IS_ENABLED(CONFIG_NF_CONNTRACK) + { + .hook = br_nf_local_in, + .pf = NFPROTO_BRIDGE, + .hooknum = NF_BR_LOCAL_IN, + .priority = NF_BR_PRI_LAST, + }, +#endif { .hook = br_nf_forward, .pf = NFPROTO_BRIDGE, |