kernel: backport flow offload pppoe fix
authorFelix Fietkau <nbd@nbd.name>
Tue, 23 Apr 2024 10:08:45 +0000 (12:08 +0200)
committerFelix Fietkau <nbd@nbd.name>
Tue, 23 Apr 2024 10:31:17 +0000 (12:31 +0200)
Signed-off-by: Felix Fietkau <nbd@nbd.name>
target/linux/generic/backport-5.15/741-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch [new file with mode: 0644]
target/linux/generic/backport-5.15/741-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch [new file with mode: 0644]
target/linux/generic/backport-6.1/740-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch [new file with mode: 0644]
target/linux/generic/backport-6.1/740-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch [new file with mode: 0644]
target/linux/generic/backport-6.6/740-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch [new file with mode: 0644]
target/linux/generic/backport-6.6/740-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch [new file with mode: 0644]
target/linux/generic/hack-5.15/650-netfilter-add-xt_FLOWOFFLOAD-target.patch
target/linux/generic/hack-6.1/650-netfilter-add-xt_FLOWOFFLOAD-target.patch
target/linux/generic/hack-6.6/650-netfilter-add-xt_FLOWOFFLOAD-target.patch

diff --git a/target/linux/generic/backport-5.15/741-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch b/target/linux/generic/backport-5.15/741-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch
new file mode 100644 (file)
index 0000000..02407da
--- /dev/null
@@ -0,0 +1,85 @@
+From: Pablo Neira Ayuso <pablo@netfilter.org>
+Date: Thu, 11 Apr 2024 13:28:59 +0200
+Subject: [PATCH] netfilter: flowtable: validate pppoe header
+
+Ensure there is sufficient room to access the protocol field of the
+PPPoe header. Validate it once before the flowtable lookup, then use a
+helper function to access protocol field.
+
+Reported-by: syzbot+b6f07e1c07ef40199081@syzkaller.appspotmail.com
+Fixes: 72efd585f714 ("netfilter: flowtable: add pppoe support")
+Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
+---
+
+--- a/include/net/netfilter/nf_flow_table.h
++++ b/include/net/netfilter/nf_flow_table.h
+@@ -318,7 +318,7 @@ int nf_flow_rule_route_ipv6(struct net *
+ int nf_flow_table_offload_init(void);
+ void nf_flow_table_offload_exit(void);
+-static inline __be16 nf_flow_pppoe_proto(const struct sk_buff *skb)
++static inline __be16 __nf_flow_pppoe_proto(const struct sk_buff *skb)
+ {
+       __be16 proto;
+@@ -334,4 +334,14 @@ static inline __be16 nf_flow_pppoe_proto
+       return 0;
+ }
++static inline bool nf_flow_pppoe_proto(struct sk_buff *skb, __be16 *inner_proto)
++{
++      if (!pskb_may_pull(skb, PPPOE_SES_HLEN))
++              return false;
++
++      *inner_proto = __nf_flow_pppoe_proto(skb);
++
++      return true;
++}
++
+ #endif /* _NF_FLOW_TABLE_H */
+--- a/net/netfilter/nf_flow_table_inet.c
++++ b/net/netfilter/nf_flow_table_inet.c
+@@ -21,7 +21,8 @@ nf_flow_offload_inet_hook(void *priv, st
+               proto = veth->h_vlan_encapsulated_proto;
+               break;
+       case htons(ETH_P_PPP_SES):
+-              proto = nf_flow_pppoe_proto(skb);
++              if (!nf_flow_pppoe_proto(skb, &proto))
++                      return NF_ACCEPT;
+               break;
+       default:
+               proto = skb->protocol;
+--- a/net/netfilter/nf_flow_table_ip.c
++++ b/net/netfilter/nf_flow_table_ip.c
+@@ -246,10 +246,11 @@ static unsigned int nf_flow_xmit_xfrm(st
+       return NF_STOLEN;
+ }
+-static bool nf_flow_skb_encap_protocol(const struct sk_buff *skb, __be16 proto,
++static bool nf_flow_skb_encap_protocol(struct sk_buff *skb, __be16 proto,
+                                      u32 *offset)
+ {
+       struct vlan_ethhdr *veth;
++      __be16 inner_proto;
+       switch (skb->protocol) {
+       case htons(ETH_P_8021Q):
+@@ -260,7 +261,8 @@ static bool nf_flow_skb_encap_protocol(c
+               }
+               break;
+       case htons(ETH_P_PPP_SES):
+-              if (nf_flow_pppoe_proto(skb) == proto) {
++              if (nf_flow_pppoe_proto(skb, &inner_proto) &&
++                  inner_proto == proto) {
+                       *offset += PPPOE_SES_HLEN;
+                       return true;
+               }
+@@ -289,7 +291,7 @@ static void nf_flow_encap_pop(struct sk_
+                       skb_reset_network_header(skb);
+                       break;
+               case htons(ETH_P_PPP_SES):
+-                      skb->protocol = nf_flow_pppoe_proto(skb);
++                      skb->protocol = __nf_flow_pppoe_proto(skb);
+                       skb_pull(skb, PPPOE_SES_HLEN);
+                       skb_reset_network_header(skb);
+                       break;
diff --git a/target/linux/generic/backport-5.15/741-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch b/target/linux/generic/backport-5.15/741-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch
new file mode 100644 (file)
index 0000000..3b822b1
--- /dev/null
@@ -0,0 +1,24 @@
+From: Pablo Neira Ayuso <pablo@netfilter.org>
+Date: Thu, 11 Apr 2024 13:29:00 +0200
+Subject: [PATCH] netfilter: flowtable: incorrect pppoe tuple
+
+pppoe traffic reaching ingress path does not match the flowtable entry
+because the pppoe header is expected to be at the network header offset.
+This bug causes a mismatch in the flow table lookup, so pppoe packets
+enter the classical forwarding path.
+
+Fixes: 72efd585f714 ("netfilter: flowtable: add pppoe support")
+Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
+---
+
+--- a/net/netfilter/nf_flow_table_ip.c
++++ b/net/netfilter/nf_flow_table_ip.c
+@@ -156,7 +156,7 @@ static void nf_flow_tuple_encap(struct s
+               tuple->encap[i].proto = skb->protocol;
+               break;
+       case htons(ETH_P_PPP_SES):
+-              phdr = (struct pppoe_hdr *)skb_mac_header(skb);
++              phdr = (struct pppoe_hdr *)skb_network_header(skb);
+               tuple->encap[i].id = ntohs(phdr->sid);
+               tuple->encap[i].proto = skb->protocol;
+               break;
diff --git a/target/linux/generic/backport-6.1/740-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch b/target/linux/generic/backport-6.1/740-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch
new file mode 100644 (file)
index 0000000..29f211e
--- /dev/null
@@ -0,0 +1,87 @@
+From: Pablo Neira Ayuso <pablo@netfilter.org>
+Date: Thu, 11 Apr 2024 13:28:59 +0200
+Subject: [PATCH] netfilter: flowtable: validate pppoe header
+
+Ensure there is sufficient room to access the protocol field of the
+PPPoe header. Validate it once before the flowtable lookup, then use a
+helper function to access protocol field.
+
+Reported-by: syzbot+b6f07e1c07ef40199081@syzkaller.appspotmail.com
+Fixes: 72efd585f714 ("netfilter: flowtable: add pppoe support")
+Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
+---
+
+--- a/include/net/netfilter/nf_flow_table.h
++++ b/include/net/netfilter/nf_flow_table.h
+@@ -335,7 +335,7 @@ int nf_flow_rule_route_ipv6(struct net *
+ int nf_flow_table_offload_init(void);
+ void nf_flow_table_offload_exit(void);
+-static inline __be16 nf_flow_pppoe_proto(const struct sk_buff *skb)
++static inline __be16 __nf_flow_pppoe_proto(const struct sk_buff *skb)
+ {
+       __be16 proto;
+@@ -351,6 +351,16 @@ static inline __be16 nf_flow_pppoe_proto
+       return 0;
+ }
++static inline bool nf_flow_pppoe_proto(struct sk_buff *skb, __be16 *inner_proto)
++{
++      if (!pskb_may_pull(skb, PPPOE_SES_HLEN))
++              return false;
++
++      *inner_proto = __nf_flow_pppoe_proto(skb);
++
++      return true;
++}
++
+ #define NF_FLOW_TABLE_STAT_INC(net, count) __this_cpu_inc((net)->ft.stat->count)
+ #define NF_FLOW_TABLE_STAT_DEC(net, count) __this_cpu_dec((net)->ft.stat->count)
+ #define NF_FLOW_TABLE_STAT_INC_ATOMIC(net, count)     \
+--- a/net/netfilter/nf_flow_table_inet.c
++++ b/net/netfilter/nf_flow_table_inet.c
+@@ -21,7 +21,8 @@ nf_flow_offload_inet_hook(void *priv, st
+               proto = veth->h_vlan_encapsulated_proto;
+               break;
+       case htons(ETH_P_PPP_SES):
+-              proto = nf_flow_pppoe_proto(skb);
++              if (!nf_flow_pppoe_proto(skb, &proto))
++                      return NF_ACCEPT;
+               break;
+       default:
+               proto = skb->protocol;
+--- a/net/netfilter/nf_flow_table_ip.c
++++ b/net/netfilter/nf_flow_table_ip.c
+@@ -267,10 +267,11 @@ static unsigned int nf_flow_xmit_xfrm(st
+       return NF_STOLEN;
+ }
+-static bool nf_flow_skb_encap_protocol(const struct sk_buff *skb, __be16 proto,
++static bool nf_flow_skb_encap_protocol(struct sk_buff *skb, __be16 proto,
+                                      u32 *offset)
+ {
+       struct vlan_ethhdr *veth;
++      __be16 inner_proto;
+       switch (skb->protocol) {
+       case htons(ETH_P_8021Q):
+@@ -281,7 +282,8 @@ static bool nf_flow_skb_encap_protocol(c
+               }
+               break;
+       case htons(ETH_P_PPP_SES):
+-              if (nf_flow_pppoe_proto(skb) == proto) {
++              if (nf_flow_pppoe_proto(skb, &inner_proto) &&
++                  inner_proto == proto) {
+                       *offset += PPPOE_SES_HLEN;
+                       return true;
+               }
+@@ -310,7 +312,7 @@ static void nf_flow_encap_pop(struct sk_
+                       skb_reset_network_header(skb);
+                       break;
+               case htons(ETH_P_PPP_SES):
+-                      skb->protocol = nf_flow_pppoe_proto(skb);
++                      skb->protocol = __nf_flow_pppoe_proto(skb);
+                       skb_pull(skb, PPPOE_SES_HLEN);
+                       skb_reset_network_header(skb);
+                       break;
diff --git a/target/linux/generic/backport-6.1/740-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch b/target/linux/generic/backport-6.1/740-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch
new file mode 100644 (file)
index 0000000..3b822b1
--- /dev/null
@@ -0,0 +1,24 @@
+From: Pablo Neira Ayuso <pablo@netfilter.org>
+Date: Thu, 11 Apr 2024 13:29:00 +0200
+Subject: [PATCH] netfilter: flowtable: incorrect pppoe tuple
+
+pppoe traffic reaching ingress path does not match the flowtable entry
+because the pppoe header is expected to be at the network header offset.
+This bug causes a mismatch in the flow table lookup, so pppoe packets
+enter the classical forwarding path.
+
+Fixes: 72efd585f714 ("netfilter: flowtable: add pppoe support")
+Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
+---
+
+--- a/net/netfilter/nf_flow_table_ip.c
++++ b/net/netfilter/nf_flow_table_ip.c
+@@ -156,7 +156,7 @@ static void nf_flow_tuple_encap(struct s
+               tuple->encap[i].proto = skb->protocol;
+               break;
+       case htons(ETH_P_PPP_SES):
+-              phdr = (struct pppoe_hdr *)skb_mac_header(skb);
++              phdr = (struct pppoe_hdr *)skb_network_header(skb);
+               tuple->encap[i].id = ntohs(phdr->sid);
+               tuple->encap[i].proto = skb->protocol;
+               break;
diff --git a/target/linux/generic/backport-6.6/740-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch b/target/linux/generic/backport-6.6/740-v6.9-01-netfilter-flowtable-validate-pppoe-header.patch
new file mode 100644 (file)
index 0000000..8a8773b
--- /dev/null
@@ -0,0 +1,87 @@
+From: Pablo Neira Ayuso <pablo@netfilter.org>
+Date: Thu, 11 Apr 2024 13:28:59 +0200
+Subject: [PATCH] netfilter: flowtable: validate pppoe header
+
+Ensure there is sufficient room to access the protocol field of the
+PPPoe header. Validate it once before the flowtable lookup, then use a
+helper function to access protocol field.
+
+Reported-by: syzbot+b6f07e1c07ef40199081@syzkaller.appspotmail.com
+Fixes: 72efd585f714 ("netfilter: flowtable: add pppoe support")
+Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
+---
+
+--- a/include/net/netfilter/nf_flow_table.h
++++ b/include/net/netfilter/nf_flow_table.h
+@@ -335,7 +335,7 @@ int nf_flow_rule_route_ipv6(struct net *
+ int nf_flow_table_offload_init(void);
+ void nf_flow_table_offload_exit(void);
+-static inline __be16 nf_flow_pppoe_proto(const struct sk_buff *skb)
++static inline __be16 __nf_flow_pppoe_proto(const struct sk_buff *skb)
+ {
+       __be16 proto;
+@@ -351,6 +351,16 @@ static inline __be16 nf_flow_pppoe_proto
+       return 0;
+ }
++static inline bool nf_flow_pppoe_proto(struct sk_buff *skb, __be16 *inner_proto)
++{
++      if (!pskb_may_pull(skb, PPPOE_SES_HLEN))
++              return false;
++
++      *inner_proto = __nf_flow_pppoe_proto(skb);
++
++      return true;
++}
++
+ #define NF_FLOW_TABLE_STAT_INC(net, count) __this_cpu_inc((net)->ft.stat->count)
+ #define NF_FLOW_TABLE_STAT_DEC(net, count) __this_cpu_dec((net)->ft.stat->count)
+ #define NF_FLOW_TABLE_STAT_INC_ATOMIC(net, count)     \
+--- a/net/netfilter/nf_flow_table_inet.c
++++ b/net/netfilter/nf_flow_table_inet.c
+@@ -21,7 +21,8 @@ nf_flow_offload_inet_hook(void *priv, st
+               proto = veth->h_vlan_encapsulated_proto;
+               break;
+       case htons(ETH_P_PPP_SES):
+-              proto = nf_flow_pppoe_proto(skb);
++              if (!nf_flow_pppoe_proto(skb, &proto))
++                      return NF_ACCEPT;
+               break;
+       default:
+               proto = skb->protocol;
+--- a/net/netfilter/nf_flow_table_ip.c
++++ b/net/netfilter/nf_flow_table_ip.c
+@@ -273,10 +273,11 @@ static unsigned int nf_flow_xmit_xfrm(st
+       return NF_STOLEN;
+ }
+-static bool nf_flow_skb_encap_protocol(const struct sk_buff *skb, __be16 proto,
++static bool nf_flow_skb_encap_protocol(struct sk_buff *skb, __be16 proto,
+                                      u32 *offset)
+ {
+       struct vlan_ethhdr *veth;
++      __be16 inner_proto;
+       switch (skb->protocol) {
+       case htons(ETH_P_8021Q):
+@@ -287,7 +288,8 @@ static bool nf_flow_skb_encap_protocol(c
+               }
+               break;
+       case htons(ETH_P_PPP_SES):
+-              if (nf_flow_pppoe_proto(skb) == proto) {
++              if (nf_flow_pppoe_proto(skb, &inner_proto) &&
++                  inner_proto == proto) {
+                       *offset += PPPOE_SES_HLEN;
+                       return true;
+               }
+@@ -316,7 +318,7 @@ static void nf_flow_encap_pop(struct sk_
+                       skb_reset_network_header(skb);
+                       break;
+               case htons(ETH_P_PPP_SES):
+-                      skb->protocol = nf_flow_pppoe_proto(skb);
++                      skb->protocol = __nf_flow_pppoe_proto(skb);
+                       skb_pull(skb, PPPOE_SES_HLEN);
+                       skb_reset_network_header(skb);
+                       break;
diff --git a/target/linux/generic/backport-6.6/740-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch b/target/linux/generic/backport-6.6/740-v6.9-02-netfilter-flowtable-incorrect-pppoe-tuple.patch
new file mode 100644 (file)
index 0000000..20ac222
--- /dev/null
@@ -0,0 +1,24 @@
+From: Pablo Neira Ayuso <pablo@netfilter.org>
+Date: Thu, 11 Apr 2024 13:29:00 +0200
+Subject: [PATCH] netfilter: flowtable: incorrect pppoe tuple
+
+pppoe traffic reaching ingress path does not match the flowtable entry
+because the pppoe header is expected to be at the network header offset.
+This bug causes a mismatch in the flow table lookup, so pppoe packets
+enter the classical forwarding path.
+
+Fixes: 72efd585f714 ("netfilter: flowtable: add pppoe support")
+Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
+---
+
+--- a/net/netfilter/nf_flow_table_ip.c
++++ b/net/netfilter/nf_flow_table_ip.c
+@@ -157,7 +157,7 @@ static void nf_flow_tuple_encap(struct s
+               tuple->encap[i].proto = skb->protocol;
+               break;
+       case htons(ETH_P_PPP_SES):
+-              phdr = (struct pppoe_hdr *)skb_mac_header(skb);
++              phdr = (struct pppoe_hdr *)skb_network_header(skb);
+               tuple->encap[i].id = ntohs(phdr->sid);
+               tuple->encap[i].proto = skb->protocol;
+               break;
index ec887539d5446f5762cbc2d9792951f84fbf27ed..49f339bddcab99ec107561015c261a6abaa50b79 100644 (file)
@@ -98,7 +98,7 @@ Signed-off-by: Felix Fietkau <nbd@nbd.name>
  obj-$(CONFIG_NETFILTER_XT_TARGET_LED) += xt_LED.o
 --- /dev/null
 +++ b/net/netfilter/xt_FLOWOFFLOAD.c
-@@ -0,0 +1,701 @@
+@@ -0,0 +1,702 @@
 +/*
 + * Copyright (C) 2018-2021 Felix Fietkau <nbd@nbd.name>
 + *
@@ -163,7 +163,8 @@ Signed-off-by: Felix Fietkau <nbd@nbd.name>
 +              proto = veth->h_vlan_encapsulated_proto;
 +              break;
 +      case htons(ETH_P_PPP_SES):
-+              proto = nf_flow_pppoe_proto(skb);
++              if (!nf_flow_pppoe_proto(skb, &proto))
++                      return NF_ACCEPT;
 +              break;
 +      default:
 +              proto = skb->protocol;
index 6fdfc7920700dcacf04e53dc1550d26e6e83bf16..0822b1a2ddeb8f330bc68d1ffad341787baa9e47 100644 (file)
@@ -44,7 +44,7 @@ Signed-off-by: Felix Fietkau <nbd@nbd.name>
  obj-$(CONFIG_NETFILTER_XT_TARGET_LED) += xt_LED.o
 --- /dev/null
 +++ b/net/netfilter/xt_FLOWOFFLOAD.c
-@@ -0,0 +1,702 @@
+@@ -0,0 +1,703 @@
 +/*
 + * Copyright (C) 2018-2021 Felix Fietkau <nbd@nbd.name>
 + *
@@ -109,7 +109,8 @@ Signed-off-by: Felix Fietkau <nbd@nbd.name>
 +              proto = veth->h_vlan_encapsulated_proto;
 +              break;
 +      case htons(ETH_P_PPP_SES):
-+              proto = nf_flow_pppoe_proto(skb);
++              if (!nf_flow_pppoe_proto(skb, &proto))
++                      return NF_ACCEPT;
 +              break;
 +      default:
 +              proto = skb->protocol;
index 973598321299d888a058b9a415871c86a273c8cb..eca611da7e837551ad8e91497d7222d16eb9670e 100644 (file)
@@ -44,7 +44,7 @@ Signed-off-by: Felix Fietkau <nbd@nbd.name>
  obj-$(CONFIG_NETFILTER_XT_TARGET_LED) += xt_LED.o
 --- /dev/null
 +++ b/net/netfilter/xt_FLOWOFFLOAD.c
-@@ -0,0 +1,702 @@
+@@ -0,0 +1,703 @@
 +/*
 + * Copyright (C) 2018-2021 Felix Fietkau <nbd@nbd.name>
 + *
@@ -109,7 +109,8 @@ Signed-off-by: Felix Fietkau <nbd@nbd.name>
 +              proto = veth->h_vlan_encapsulated_proto;
 +              break;
 +      case htons(ETH_P_PPP_SES):
-+              proto = nf_flow_pppoe_proto(skb);
++              if (!nf_flow_pppoe_proto(skb, &proto))
++                      return NF_ACCEPT;
 +              break;
 +      default:
 +              proto = skb->protocol;