ipv6: remove dependency of nf_defrag_ipv6 on ipv6 module
authorFlorian Westphal <fw@strlen.de>
Fri, 13 Jul 2018 23:14:01 +0000 (01:14 +0200)
committerPablo Neira Ayuso <pablo@netfilter.org>
Wed, 18 Jul 2018 09:26:53 +0000 (11:26 +0200)
IPV6=m
DEFRAG_IPV6=m
CONNTRACK=y yields:

net/netfilter/nf_conntrack_proto.o: In function `nf_ct_netns_do_get':
net/netfilter/nf_conntrack_proto.c:802: undefined reference to `nf_defrag_ipv6_enable'
net/netfilter/nf_conntrack_proto.o:(.rodata+0x640): undefined reference to `nf_conntrack_l4proto_icmpv6'

Setting DEFRAG_IPV6=y causes undefined references to ip6_rhash_params
ip6_frag_init and ip6_expire_frag_queue so it would be needed to force
IPV6=y too.

This patch gets rid of the 'followup linker error' by removing
the dependency of ipv6.ko symbols from netfilter ipv6 defrag.

Shared code is placed into a header, then used from both.

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
include/net/ipv6.h
include/net/ipv6_frag.h [new file with mode: 0644]
net/ieee802154/6lowpan/reassembly.c
net/ipv6/netfilter/nf_conntrack_reasm.c
net/ipv6/netfilter/nf_defrag_ipv6_hooks.c
net/ipv6/reassembly.c
net/openvswitch/conntrack.c

index aa6fd11..3720958 100644 (file)
@@ -581,34 +581,6 @@ static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
 }
 #endif
 
-struct inet_frag_queue;
-
-enum ip6_defrag_users {
-       IP6_DEFRAG_LOCAL_DELIVER,
-       IP6_DEFRAG_CONNTRACK_IN,
-       __IP6_DEFRAG_CONNTRACK_IN       = IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX,
-       IP6_DEFRAG_CONNTRACK_OUT,
-       __IP6_DEFRAG_CONNTRACK_OUT      = IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
-       IP6_DEFRAG_CONNTRACK_BRIDGE_IN,
-       __IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
-};
-
-void ip6_frag_init(struct inet_frag_queue *q, const void *a);
-extern const struct rhashtable_params ip6_rhash_params;
-
-/*
- *     Equivalent of ipv4 struct ip
- */
-struct frag_queue {
-       struct inet_frag_queue  q;
-
-       int                     iif;
-       __u16                   nhoffset;
-       u8                      ecn;
-};
-
-void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq);
-
 static inline bool ipv6_addr_any(const struct in6_addr *a)
 {
 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
diff --git a/include/net/ipv6_frag.h b/include/net/ipv6_frag.h
new file mode 100644 (file)
index 0000000..6ced1e6
--- /dev/null
@@ -0,0 +1,104 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _IPV6_FRAG_H
+#define _IPV6_FRAG_H
+#include <linux/kernel.h>
+#include <net/addrconf.h>
+#include <net/ipv6.h>
+#include <net/inet_frag.h>
+
+enum ip6_defrag_users {
+       IP6_DEFRAG_LOCAL_DELIVER,
+       IP6_DEFRAG_CONNTRACK_IN,
+       __IP6_DEFRAG_CONNTRACK_IN       = IP6_DEFRAG_CONNTRACK_IN + USHRT_MAX,
+       IP6_DEFRAG_CONNTRACK_OUT,
+       __IP6_DEFRAG_CONNTRACK_OUT      = IP6_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
+       IP6_DEFRAG_CONNTRACK_BRIDGE_IN,
+       __IP6_DEFRAG_CONNTRACK_BRIDGE_IN = IP6_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
+};
+
+/*
+ *     Equivalent of ipv4 struct ip
+ */
+struct frag_queue {
+       struct inet_frag_queue  q;
+
+       int                     iif;
+       __u16                   nhoffset;
+       u8                      ecn;
+};
+
+#if IS_ENABLED(CONFIG_IPV6)
+static inline void ip6frag_init(struct inet_frag_queue *q, const void *a)
+{
+       struct frag_queue *fq = container_of(q, struct frag_queue, q);
+       const struct frag_v6_compare_key *key = a;
+
+       q->key.v6 = *key;
+       fq->ecn = 0;
+}
+
+static inline u32 ip6frag_key_hashfn(const void *data, u32 len, u32 seed)
+{
+       return jhash2(data,
+                     sizeof(struct frag_v6_compare_key) / sizeof(u32), seed);
+}
+
+static inline u32 ip6frag_obj_hashfn(const void *data, u32 len, u32 seed)
+{
+       const struct inet_frag_queue *fq = data;
+
+       return jhash2((const u32 *)&fq->key.v6,
+                     sizeof(struct frag_v6_compare_key) / sizeof(u32), seed);
+}
+
+static inline int
+ip6frag_obj_cmpfn(struct rhashtable_compare_arg *arg, const void *ptr)
+{
+       const struct frag_v6_compare_key *key = arg->key;
+       const struct inet_frag_queue *fq = ptr;
+
+       return !!memcmp(&fq->key, key, sizeof(*key));
+}
+
+static inline void
+ip6frag_expire_frag_queue(struct net *net, struct frag_queue *fq)
+{
+       struct net_device *dev = NULL;
+       struct sk_buff *head;
+
+       rcu_read_lock();
+       spin_lock(&fq->q.lock);
+
+       if (fq->q.flags & INET_FRAG_COMPLETE)
+               goto out;
+
+       inet_frag_kill(&fq->q);
+
+       dev = dev_get_by_index_rcu(net, fq->iif);
+       if (!dev)
+               goto out;
+
+       __IP6_INC_STATS(net, __in6_dev_get(dev), IPSTATS_MIB_REASMFAILS);
+       __IP6_INC_STATS(net, __in6_dev_get(dev), IPSTATS_MIB_REASMTIMEOUT);
+
+       /* Don't send error if the first segment did not arrive. */
+       head = fq->q.fragments;
+       if (!(fq->q.flags & INET_FRAG_FIRST_IN) || !head)
+               goto out;
+
+       head->dev = dev;
+       skb_get(head);
+       spin_unlock(&fq->q.lock);
+
+       icmpv6_send(head, ICMPV6_TIME_EXCEED, ICMPV6_EXC_FRAGTIME, 0);
+       kfree_skb(head);
+       goto out_rcu_unlock;
+
+out:
+       spin_unlock(&fq->q.lock);
+out_rcu_unlock:
+       rcu_read_unlock();
+       inet_frag_put(&fq->q);
+}
+#endif
+#endif
index 2cc2241..ec7a5da 100644 (file)
@@ -25,7 +25,7 @@
 
 #include <net/ieee802154_netdev.h>
 #include <net/6lowpan.h>
-#include <net/ipv6.h>
+#include <net/ipv6_frag.h>
 #include <net/inet_frag.h>
 
 #include "6lowpan_i.h"
index a452d99..333ee32 100644 (file)
@@ -33,9 +33,8 @@
 
 #include <net/sock.h>
 #include <net/snmp.h>
-#include <net/inet_frag.h>
+#include <net/ipv6_frag.h>
 
-#include <net/ipv6.h>
 #include <net/protocol.h>
 #include <net/transp_v6.h>
 #include <net/rawv6.h>
@@ -151,7 +150,7 @@ static void nf_ct_frag6_expire(struct timer_list *t)
        fq = container_of(frag, struct frag_queue, q);
        net = container_of(fq->q.net, struct net, nf_frag.frags);
 
-       ip6_expire_frag_queue(net, fq);
+       ip6frag_expire_frag_queue(net, fq);
 }
 
 /* Creation primitives. */
@@ -622,16 +621,24 @@ static struct pernet_operations nf_ct_net_ops = {
        .exit = nf_ct_net_exit,
 };
 
+static const struct rhashtable_params nfct_rhash_params = {
+       .head_offset            = offsetof(struct inet_frag_queue, node),
+       .hashfn                 = ip6frag_key_hashfn,
+       .obj_hashfn             = ip6frag_obj_hashfn,
+       .obj_cmpfn              = ip6frag_obj_cmpfn,
+       .automatic_shrinking    = true,
+};
+
 int nf_ct_frag6_init(void)
 {
        int ret = 0;
 
-       nf_frags.constructor = ip6_frag_init;
+       nf_frags.constructor = ip6frag_init;
        nf_frags.destructor = NULL;
        nf_frags.qsize = sizeof(struct frag_queue);
        nf_frags.frag_expire = nf_ct_frag6_expire;
        nf_frags.frags_cache_name = nf_frags_cache_name;
-       nf_frags.rhash_params = ip6_rhash_params;
+       nf_frags.rhash_params = nfct_rhash_params;
        ret = inet_frags_init(&nf_frags);
        if (ret)
                goto out;
index e631be2..72dd3e2 100644 (file)
@@ -14,8 +14,7 @@
 #include <linux/skbuff.h>
 #include <linux/icmp.h>
 #include <linux/sysctl.h>
-#include <net/ipv6.h>
-#include <net/inet_frag.h>
+#include <net/ipv6_frag.h>
 
 #include <linux/netfilter_ipv6.h>
 #include <linux/netfilter_bridge.h>
index b939b94..6edd2ac 100644 (file)
@@ -57,7 +57,7 @@
 #include <net/rawv6.h>
 #include <net/ndisc.h>
 #include <net/addrconf.h>
-#include <net/inet_frag.h>
+#include <net/ipv6_frag.h>
 #include <net/inet_ecn.h>
 
 static const char ip6_frag_cache_name[] = "ip6-frags";
@@ -72,61 +72,6 @@ static struct inet_frags ip6_frags;
 static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev,
                          struct net_device *dev);
 
-void ip6_frag_init(struct inet_frag_queue *q, const void *a)
-{
-       struct frag_queue *fq = container_of(q, struct frag_queue, q);
-       const struct frag_v6_compare_key *key = a;
-
-       q->key.v6 = *key;
-       fq->ecn = 0;
-}
-EXPORT_SYMBOL(ip6_frag_init);
-
-void ip6_expire_frag_queue(struct net *net, struct frag_queue *fq)
-{
-       struct net_device *dev = NULL;
-       struct sk_buff *head;
-
-       rcu_read_lock();
-       spin_lock(&fq->q.lock);
-
-       if (fq->q.flags & INET_FRAG_COMPLETE)
-               goto out;
-
-       inet_frag_kill(&fq->q);
-
-       dev = dev_get_by_index_rcu(net, fq->iif);
-       if (!dev)
-               goto out;
-
-       __IP6_INC_STATS(net, __in6_dev_get(dev), IPSTATS_MIB_REASMFAILS);
-       __IP6_INC_STATS(net, __in6_dev_get(dev), IPSTATS_MIB_REASMTIMEOUT);
-
-       /* Don't send error if the first segment did not arrive. */
-       head = fq->q.fragments;
-       if (!(fq->q.flags & INET_FRAG_FIRST_IN) || !head)
-               goto out;
-
-       /* But use as source device on which LAST ARRIVED
-        * segment was received. And do not use fq->dev
-        * pointer directly, device might already disappeared.
-        */
-       head->dev = dev;
-       skb_get(head);
-       spin_unlock(&fq->q.lock);
-
-       icmpv6_send(head, ICMPV6_TIME_EXCEED, ICMPV6_EXC_FRAGTIME, 0);
-       kfree_skb(head);
-       goto out_rcu_unlock;
-
-out:
-       spin_unlock(&fq->q.lock);
-out_rcu_unlock:
-       rcu_read_unlock();
-       inet_frag_put(&fq->q);
-}
-EXPORT_SYMBOL(ip6_expire_frag_queue);
-
 static void ip6_frag_expire(struct timer_list *t)
 {
        struct inet_frag_queue *frag = from_timer(frag, t, timer);
@@ -136,7 +81,7 @@ static void ip6_frag_expire(struct timer_list *t)
        fq = container_of(frag, struct frag_queue, q);
        net = container_of(fq->q.net, struct net, ipv6.frags);
 
-       ip6_expire_frag_queue(net, fq);
+       ip6frag_expire_frag_queue(net, fq);
 }
 
 static struct frag_queue *
@@ -696,42 +641,19 @@ static struct pernet_operations ip6_frags_ops = {
        .exit = ipv6_frags_exit_net,
 };
 
-static u32 ip6_key_hashfn(const void *data, u32 len, u32 seed)
-{
-       return jhash2(data,
-                     sizeof(struct frag_v6_compare_key) / sizeof(u32), seed);
-}
-
-static u32 ip6_obj_hashfn(const void *data, u32 len, u32 seed)
-{
-       const struct inet_frag_queue *fq = data;
-
-       return jhash2((const u32 *)&fq->key.v6,
-                     sizeof(struct frag_v6_compare_key) / sizeof(u32), seed);
-}
-
-static int ip6_obj_cmpfn(struct rhashtable_compare_arg *arg, const void *ptr)
-{
-       const struct frag_v6_compare_key *key = arg->key;
-       const struct inet_frag_queue *fq = ptr;
-
-       return !!memcmp(&fq->key, key, sizeof(*key));
-}
-
-const struct rhashtable_params ip6_rhash_params = {
+static const struct rhashtable_params ip6_rhash_params = {
        .head_offset            = offsetof(struct inet_frag_queue, node),
-       .hashfn                 = ip6_key_hashfn,
-       .obj_hashfn             = ip6_obj_hashfn,
-       .obj_cmpfn              = ip6_obj_cmpfn,
+       .hashfn                 = ip6frag_key_hashfn,
+       .obj_hashfn             = ip6frag_obj_hashfn,
+       .obj_cmpfn              = ip6frag_obj_cmpfn,
        .automatic_shrinking    = true,
 };
-EXPORT_SYMBOL(ip6_rhash_params);
 
 int __init ipv6_frag_init(void)
 {
        int ret;
 
-       ip6_frags.constructor = ip6_frag_init;
+       ip6_frags.constructor = ip6frag_init;
        ip6_frags.destructor = NULL;
        ip6_frags.qsize = sizeof(struct frag_queue);
        ip6_frags.frag_expire = ip6_frag_expire;
index 3e33c38..86a7510 100644 (file)
@@ -26,6 +26,7 @@
 #include <net/netfilter/nf_conntrack_seqadj.h>
 #include <net/netfilter/nf_conntrack_zones.h>
 #include <net/netfilter/ipv6/nf_defrag_ipv6.h>
+#include <net/ipv6_frag.h>
 
 #ifdef CONFIG_NF_NAT_NEEDED
 #include <linux/netfilter/nf_nat.h>