]> git.proxmox.com Git - mirror_ubuntu-kernels.git/commitdiff
mpls: ip tunnel support
authorRoopa Prabhu <roopa@cumulusnetworks.com>
Tue, 21 Jul 2015 08:43:53 +0000 (10:43 +0200)
committerDavid S. Miller <davem@davemloft.net>
Tue, 21 Jul 2015 17:39:05 +0000 (10:39 -0700)
This implementation uses lwtunnel infrastructure to register
hooks for mpls tunnel encaps.

It picks cues from iptunnel_encaps infrastructure and previous
mpls iptunnel RFC patches from Eric W. Biederman and Robert Shearman

Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/linux/mpls_iptunnel.h [new file with mode: 0644]
include/net/mpls_iptunnel.h [new file with mode: 0644]
include/uapi/linux/mpls_iptunnel.h [new file with mode: 0644]
net/mpls/Kconfig
net/mpls/Makefile
net/mpls/mpls_iptunnel.c [new file with mode: 0644]

diff --git a/include/linux/mpls_iptunnel.h b/include/linux/mpls_iptunnel.h
new file mode 100644 (file)
index 0000000..ef29eb2
--- /dev/null
@@ -0,0 +1,6 @@
+#ifndef _LINUX_MPLS_IPTUNNEL_H
+#define _LINUX_MPLS_IPTUNNEL_H
+
+#include <uapi/linux/mpls_iptunnel.h>
+
+#endif  /* _LINUX_MPLS_IPTUNNEL_H */
diff --git a/include/net/mpls_iptunnel.h b/include/net/mpls_iptunnel.h
new file mode 100644 (file)
index 0000000..4757997
--- /dev/null
@@ -0,0 +1,29 @@
+/*
+ * Copyright (c) 2015 Cumulus Networks, Inc.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of version 2 of the GNU General Public
+ * License as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ */
+
+#ifndef _NET_MPLS_IPTUNNEL_H
+#define _NET_MPLS_IPTUNNEL_H 1
+
+#define MAX_NEW_LABELS 2
+
+struct mpls_iptunnel_encap {
+       u32     label[MAX_NEW_LABELS];
+       u32     labels;
+};
+
+static inline struct mpls_iptunnel_encap *mpls_lwtunnel_encap(struct lwtunnel_state *lwtstate)
+{
+       return (struct mpls_iptunnel_encap *)lwtstate->data;
+}
+
+#endif
diff --git a/include/uapi/linux/mpls_iptunnel.h b/include/uapi/linux/mpls_iptunnel.h
new file mode 100644 (file)
index 0000000..d80a049
--- /dev/null
@@ -0,0 +1,28 @@
+/*
+ *     mpls tunnel api
+ *
+ *     Authors:
+ *             Roopa Prabhu <roopa@cumulusnetworks.com>
+ *
+ *     This program is free software; you can redistribute it and/or
+ *     modify it under the terms of the GNU General Public License
+ *     as published by the Free Software Foundation; either version
+ *     2 of the License, or (at your option) any later version.
+ */
+
+#ifndef _UAPI_LINUX_MPLS_IPTUNNEL_H
+#define _UAPI_LINUX_MPLS_IPTUNNEL_H
+
+/* MPLS tunnel attributes
+ * [RTA_ENCAP] = {
+ *     [MPLS_IPTUNNEL_DST]
+ * }
+ */
+enum {
+       MPLS_IPTUNNEL_UNSPEC,
+       MPLS_IPTUNNEL_DST,
+       __MPLS_IPTUNNEL_MAX,
+};
+#define MPLS_IPTUNNEL_MAX (__MPLS_IPTUNNEL_MAX - 1)
+
+#endif /* _UAPI_LINUX_MPLS_IPTUNNEL_H */
index 17bde799c8548e46a791fabc8b9619c57f2b2f99..5c467ef973114c49b352ac64ed248e4199d3d0f3 100644 (file)
@@ -24,7 +24,13 @@ config NET_MPLS_GSO
 
 config MPLS_ROUTING
        tristate "MPLS: routing support"
-       help
+       ---help---
         Add support for forwarding of mpls packets.
 
+config MPLS_IPTUNNEL
+       tristate "MPLS: IP over MPLS tunnel support"
+       depends on LWTUNNEL && MPLS_ROUTING
+       ---help---
+        mpls ip tunnel support.
+
 endif # MPLS
index 65bbe68c72e66ad44463ed465eaafe01e4c2bd10..9ca92362501653463429b0254d999782f2b6c450 100644 (file)
@@ -3,5 +3,6 @@
 #
 obj-$(CONFIG_NET_MPLS_GSO) += mpls_gso.o
 obj-$(CONFIG_MPLS_ROUTING) += mpls_router.o
+obj-$(CONFIG_MPLS_IPTUNNEL) += mpls_iptunnel.o
 
 mpls_router-y := af_mpls.o
diff --git a/net/mpls/mpls_iptunnel.c b/net/mpls/mpls_iptunnel.c
new file mode 100644 (file)
index 0000000..eea096f
--- /dev/null
@@ -0,0 +1,233 @@
+/*
+ * mpls tunnels        An implementation mpls tunnels using the light weight tunnel
+ *             infrastructure
+ *
+ * Authors:    Roopa Prabhu, <roopa@cumulusnetworks.com>
+ *
+ *             This program is free software; you can redistribute it and/or
+ *             modify it under the terms of the GNU General Public License
+ *             as published by the Free Software Foundation; either version
+ *             2 of the License, or (at your option) any later version.
+ *
+ */
+#include <linux/types.h>
+#include <linux/skbuff.h>
+#include <linux/net.h>
+#include <linux/module.h>
+#include <linux/mpls.h>
+#include <linux/vmalloc.h>
+#include <net/ip.h>
+#include <net/dst.h>
+#include <net/lwtunnel.h>
+#include <net/netevent.h>
+#include <net/netns/generic.h>
+#include <net/ip6_fib.h>
+#include <net/route.h>
+#include <net/mpls_iptunnel.h>
+#include <linux/mpls_iptunnel.h>
+#include "internal.h"
+
+static const struct nla_policy mpls_iptunnel_policy[MPLS_IPTUNNEL_MAX + 1] = {
+       [MPLS_IPTUNNEL_DST]     = { .type = NLA_U32 },
+};
+
+static unsigned int mpls_encap_size(struct mpls_iptunnel_encap *en)
+{
+       /* The size of the layer 2.5 labels to be added for this route */
+       return en->labels * sizeof(struct mpls_shim_hdr);
+}
+
+int mpls_output(struct sock *sk, struct sk_buff *skb)
+{
+       struct mpls_iptunnel_encap *tun_encap_info;
+       struct mpls_shim_hdr *hdr;
+       struct net_device *out_dev;
+       unsigned int hh_len;
+       unsigned int new_header_size;
+       unsigned int mtu;
+       struct dst_entry *dst = skb_dst(skb);
+       struct rtable *rt = NULL;
+       struct rt6_info *rt6 = NULL;
+       struct lwtunnel_state *lwtstate = NULL;
+       int err = 0;
+       bool bos;
+       int i;
+       unsigned int ttl;
+
+       /* Obtain the ttl */
+       if (skb->protocol == htons(ETH_P_IP)) {
+               ttl = ip_hdr(skb)->ttl;
+               rt = (struct rtable *)dst;
+               lwtstate = rt->rt_lwtstate;
+       } else if (skb->protocol == htons(ETH_P_IPV6)) {
+               ttl = ipv6_hdr(skb)->hop_limit;
+               rt6 = (struct rt6_info *)dst;
+               lwtstate = rt6->rt6i_lwtstate;
+       } else {
+               goto drop;
+       }
+
+       skb_orphan(skb);
+
+       /* Find the output device */
+       out_dev = rcu_dereference(dst->dev);
+       if (!mpls_output_possible(out_dev) ||
+           !lwtstate || skb_warn_if_lro(skb))
+               goto drop;
+
+       skb_forward_csum(skb);
+
+       tun_encap_info = mpls_lwtunnel_encap(lwtstate);
+
+       /* Verify the destination can hold the packet */
+       new_header_size = mpls_encap_size(tun_encap_info);
+       mtu = mpls_dev_mtu(out_dev);
+       if (mpls_pkt_too_big(skb, mtu - new_header_size))
+               goto drop;
+
+       hh_len = LL_RESERVED_SPACE(out_dev);
+       if (!out_dev->header_ops)
+               hh_len = 0;
+
+       /* Ensure there is enough space for the headers in the skb */
+       if (skb_cow(skb, hh_len + new_header_size))
+               goto drop;
+
+       skb_push(skb, new_header_size);
+       skb_reset_network_header(skb);
+
+       skb->dev = out_dev;
+       skb->protocol = htons(ETH_P_MPLS_UC);
+
+       /* Push the new labels */
+       hdr = mpls_hdr(skb);
+       bos = true;
+       for (i = tun_encap_info->labels - 1; i >= 0; i--) {
+               hdr[i] = mpls_entry_encode(tun_encap_info->label[i],
+                                          ttl, 0, bos);
+               bos = false;
+       }
+
+       if (rt)
+               err = neigh_xmit(NEIGH_ARP_TABLE, out_dev, &rt->rt_gateway,
+                                skb);
+       else if (rt6)
+               err = neigh_xmit(NEIGH_ND_TABLE, out_dev, &rt6->rt6i_gateway,
+                                skb);
+       if (err)
+               net_dbg_ratelimited("%s: packet transmission failed: %d\n",
+                                   __func__, err);
+
+       return 0;
+
+drop:
+       kfree_skb(skb);
+       return -EINVAL;
+}
+
+static int mpls_build_state(struct net_device *dev, struct nlattr *nla,
+                           struct lwtunnel_state **ts)
+{
+       struct mpls_iptunnel_encap *tun_encap_info;
+       struct nlattr *tb[MPLS_IPTUNNEL_MAX + 1];
+       struct lwtunnel_state *newts;
+       int tun_encap_info_len;
+       int ret;
+
+       ret = nla_parse_nested(tb, MPLS_IPTUNNEL_MAX, nla,
+                              mpls_iptunnel_policy);
+       if (ret < 0)
+               return ret;
+
+       if (!tb[MPLS_IPTUNNEL_DST])
+               return -EINVAL;
+
+       tun_encap_info_len = sizeof(*tun_encap_info);
+
+       newts = lwtunnel_state_alloc(tun_encap_info_len);
+       if (!newts)
+               return -ENOMEM;
+
+       newts->len = tun_encap_info_len;
+       tun_encap_info = mpls_lwtunnel_encap(newts);
+       ret = nla_get_labels(tb[MPLS_IPTUNNEL_DST], MAX_NEW_LABELS,
+                            &tun_encap_info->labels, tun_encap_info->label);
+       if (ret)
+               goto errout;
+       newts->type = LWTUNNEL_ENCAP_MPLS;
+       newts->flags |= LWTUNNEL_STATE_OUTPUT_REDIRECT;
+
+       *ts = newts;
+
+       return 0;
+
+errout:
+       kfree(newts);
+       *ts = NULL;
+
+       return ret;
+}
+
+static int mpls_fill_encap_info(struct sk_buff *skb,
+                               struct lwtunnel_state *lwtstate)
+{
+       struct mpls_iptunnel_encap *tun_encap_info;
+       
+       tun_encap_info = mpls_lwtunnel_encap(lwtstate);
+
+       if (nla_put_labels(skb, MPLS_IPTUNNEL_DST, tun_encap_info->labels,
+                          tun_encap_info->label))
+               goto nla_put_failure;
+
+       return 0;
+
+nla_put_failure:
+       return -EMSGSIZE;
+}
+
+static int mpls_encap_nlsize(struct lwtunnel_state *lwtstate)
+{
+       struct mpls_iptunnel_encap *tun_encap_info;
+
+       tun_encap_info = mpls_lwtunnel_encap(lwtstate);
+
+       return nla_total_size(tun_encap_info->labels * 4);
+}
+
+static int mpls_encap_cmp(struct lwtunnel_state *a, struct lwtunnel_state *b)
+{
+       struct mpls_iptunnel_encap *a_hdr = mpls_lwtunnel_encap(a);
+       struct mpls_iptunnel_encap *b_hdr = mpls_lwtunnel_encap(b);
+       int l;
+
+       if (a_hdr->labels != b_hdr->labels)
+               return 1;
+
+       for (l = 0; l < MAX_NEW_LABELS; l++)
+               if (a_hdr->label[l] != b_hdr->label[l])
+                       return 1;
+       return 0;
+}
+
+static const struct lwtunnel_encap_ops mpls_iptun_ops = {
+       .build_state = mpls_build_state,
+       .output = mpls_output,
+       .fill_encap = mpls_fill_encap_info,
+       .get_encap_size = mpls_encap_nlsize,
+       .cmp_encap = mpls_encap_cmp,
+};
+
+static int __init mpls_iptunnel_init(void)
+{
+       return lwtunnel_encap_add_ops(&mpls_iptun_ops, LWTUNNEL_ENCAP_MPLS);
+}
+module_init(mpls_iptunnel_init);
+
+static void __exit mpls_iptunnel_exit(void)
+{
+       lwtunnel_encap_del_ops(&mpls_iptun_ops, LWTUNNEL_ENCAP_MPLS);
+}
+module_exit(mpls_iptunnel_exit);
+
+MODULE_DESCRIPTION("MultiProtocol Label Switching IP Tunnels");
+MODULE_LICENSE("GPL v2");