This commit was manufactured by cvs2svn to create branch 'vserver'.
[linux-2.6.git] / net / ipv6 / netfilter / nf_conntrack_proto_icmpv6.c
diff --git a/net/ipv6/netfilter/nf_conntrack_proto_icmpv6.c b/net/ipv6/netfilter/nf_conntrack_proto_icmpv6.c
new file mode 100644 (file)
index 0000000..86c6703
--- /dev/null
@@ -0,0 +1,323 @@
+/*
+ * Copyright (C)2003,2004 USAGI/WIDE Project
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Author:
+ *     Yasuyuki Kozakai @USAGI <yasuyuki.kozakai@toshiba.co.jp>
+ *
+ * 16 Dec 2003: Yasuyuki Kozakai @USAGI <yasuyuki.kozakai@toshiba.co.jp>
+ *     - ICMPv6 tracking support. Derived from the original ip_conntrack code
+ *       net/ipv4/netfilter/ip_conntrack_proto_icmp.c which had the following
+ *       copyright information:
+ *             (C) 1999-2001 Paul `Rusty' Russell
+ *             (C) 2002-2004 Netfilter Core Team <coreteam@netfilter.org>
+ */
+
+#include <linux/types.h>
+#include <linux/sched.h>
+#include <linux/timer.h>
+#include <linux/module.h>
+#include <linux/netfilter.h>
+#include <linux/in6.h>
+#include <linux/icmpv6.h>
+#include <linux/ipv6.h>
+#include <net/ipv6.h>
+#include <net/ip6_checksum.h>
+#include <linux/seq_file.h>
+#include <linux/netfilter_ipv6.h>
+#include <net/netfilter/nf_conntrack_tuple.h>
+#include <net/netfilter/nf_conntrack_protocol.h>
+#include <net/netfilter/nf_conntrack_core.h>
+#include <net/netfilter/ipv6/nf_conntrack_icmpv6.h>
+
+unsigned long nf_ct_icmpv6_timeout = 30*HZ;
+
+#if 0
+#define DEBUGP printk
+#else
+#define DEBUGP(format, args...)
+#endif
+
+static int icmpv6_pkt_to_tuple(const struct sk_buff *skb,
+                              unsigned int dataoff,
+                              struct nf_conntrack_tuple *tuple)
+{
+       struct icmp6hdr _hdr, *hp;
+
+       hp = skb_header_pointer(skb, dataoff, sizeof(_hdr), &_hdr);
+       if (hp == NULL)
+               return 0;
+       tuple->dst.u.icmp.type = hp->icmp6_type;
+       tuple->src.u.icmp.id = hp->icmp6_identifier;
+       tuple->dst.u.icmp.code = hp->icmp6_code;
+
+       return 1;
+}
+
+/* Add 1; spaces filled with 0. */
+static u_int8_t invmap[] = {
+       [ICMPV6_ECHO_REQUEST - 128]     = ICMPV6_ECHO_REPLY + 1,
+       [ICMPV6_ECHO_REPLY - 128]       = ICMPV6_ECHO_REQUEST + 1,
+       [ICMPV6_NI_QUERY - 128]         = ICMPV6_NI_QUERY + 1,
+       [ICMPV6_NI_REPLY - 128]         = ICMPV6_NI_REPLY +1
+};
+
+static int icmpv6_invert_tuple(struct nf_conntrack_tuple *tuple,
+                              const struct nf_conntrack_tuple *orig)
+{
+       int type = orig->dst.u.icmp.type - 128;
+       if (type < 0 || type >= sizeof(invmap) || !invmap[type])
+               return 0;
+
+       tuple->src.u.icmp.id   = orig->src.u.icmp.id;
+       tuple->dst.u.icmp.type = invmap[type] - 1;
+       tuple->dst.u.icmp.code = orig->dst.u.icmp.code;
+       return 1;
+}
+
+/* Print out the per-protocol part of the tuple. */
+static int icmpv6_print_tuple(struct seq_file *s,
+                             const struct nf_conntrack_tuple *tuple)
+{
+       return seq_printf(s, "type=%u code=%u id=%u ",
+                         tuple->dst.u.icmp.type,
+                         tuple->dst.u.icmp.code,
+                         ntohs(tuple->src.u.icmp.id));
+}
+
+/* Print out the private part of the conntrack. */
+static int icmpv6_print_conntrack(struct seq_file *s,
+                                 const struct nf_conn *conntrack)
+{
+       return 0;
+}
+
+/* Returns verdict for packet, or -1 for invalid. */
+static int icmpv6_packet(struct nf_conn *ct,
+                      const struct sk_buff *skb,
+                      unsigned int dataoff,
+                      enum ip_conntrack_info ctinfo,
+                      int pf,
+                      unsigned int hooknum)
+{
+       /* Try to delete connection immediately after all replies:
+           won't actually vanish as we still have skb, and del_timer
+           means this will only run once even if count hits zero twice
+           (theoretically possible with SMP) */
+       if (CTINFO2DIR(ctinfo) == IP_CT_DIR_REPLY) {
+               if (atomic_dec_and_test(&ct->proto.icmp.count)
+                   && del_timer(&ct->timeout))
+                       ct->timeout.function((unsigned long)ct);
+       } else {
+               atomic_inc(&ct->proto.icmp.count);
+               nf_conntrack_event_cache(IPCT_PROTOINFO_VOLATILE, skb);
+               nf_ct_refresh_acct(ct, ctinfo, skb, nf_ct_icmpv6_timeout);
+       }
+
+       return NF_ACCEPT;
+}
+
+/* Called when a new connection for this protocol found. */
+static int icmpv6_new(struct nf_conn *conntrack,
+                     const struct sk_buff *skb,
+                     unsigned int dataoff)
+{
+       static u_int8_t valid_new[] = {
+               [ICMPV6_ECHO_REQUEST - 128] = 1,
+               [ICMPV6_NI_QUERY - 128] = 1
+       };
+       int type = conntrack->tuplehash[0].tuple.dst.u.icmp.type - 128;
+
+       if (type < 0 || type >= sizeof(valid_new) || !valid_new[type]) {
+               /* Can't create a new ICMPv6 `conn' with this. */
+               DEBUGP("icmpv6: can't create new conn with type %u\n",
+                      type + 128);
+               NF_CT_DUMP_TUPLE(&conntrack->tuplehash[0].tuple);
+               return 0;
+       }
+       atomic_set(&conntrack->proto.icmp.count, 0);
+       return 1;
+}
+
+extern int
+nf_ct_ipv6_skip_exthdr(struct sk_buff *skb, int start, u8 *nexthdrp, int len);
+extern struct nf_conntrack_l3proto nf_conntrack_l3proto_ipv6;
+static int
+icmpv6_error_message(struct sk_buff *skb,
+                    unsigned int icmp6off,
+                    enum ip_conntrack_info *ctinfo,
+                    unsigned int hooknum)
+{
+       struct nf_conntrack_tuple intuple, origtuple;
+       struct nf_conntrack_tuple_hash *h;
+       struct icmp6hdr _hdr, *hp;
+       unsigned int inip6off;
+       struct nf_conntrack_protocol *inproto;
+       u_int8_t inprotonum;
+       unsigned int inprotoff;
+
+       NF_CT_ASSERT(skb->nfct == NULL);
+
+       hp = skb_header_pointer(skb, icmp6off, sizeof(_hdr), &_hdr);
+       if (hp == NULL) {
+               DEBUGP("icmpv6_error: Can't get ICMPv6 hdr.\n");
+               return -NF_ACCEPT;
+       }
+
+       inip6off = icmp6off + sizeof(_hdr);
+       if (skb_copy_bits(skb, inip6off+offsetof(struct ipv6hdr, nexthdr),
+                         &inprotonum, sizeof(inprotonum)) != 0) {
+               DEBUGP("icmpv6_error: Can't get nexthdr in inner IPv6 header.\n");
+               return -NF_ACCEPT;
+       }
+       inprotoff = nf_ct_ipv6_skip_exthdr(skb,
+                                          inip6off + sizeof(struct ipv6hdr),
+                                          &inprotonum,
+                                          skb->len - inip6off
+                                                   - sizeof(struct ipv6hdr));
+
+       if ((inprotoff < 0) || (inprotoff > skb->len) ||
+           (inprotonum == NEXTHDR_FRAGMENT)) {
+               DEBUGP("icmpv6_error: Can't get protocol header in ICMPv6 payload.\n");
+               return -NF_ACCEPT;
+       }
+
+       inproto = __nf_ct_proto_find(PF_INET6, inprotonum);
+
+       /* Are they talking about one of our connections? */
+       if (!nf_ct_get_tuple(skb, inip6off, inprotoff, PF_INET6, inprotonum,
+                            &origtuple, &nf_conntrack_l3proto_ipv6, inproto)) {
+               DEBUGP("icmpv6_error: Can't get tuple\n");
+               return -NF_ACCEPT;
+       }
+
+       /* Ordinarily, we'd expect the inverted tupleproto, but it's
+          been preserved inside the ICMP. */
+       if (!nf_ct_invert_tuple(&intuple, &origtuple,
+                               &nf_conntrack_l3proto_ipv6, inproto)) {
+               DEBUGP("icmpv6_error: Can't invert tuple\n");
+               return -NF_ACCEPT;
+       }
+
+       *ctinfo = IP_CT_RELATED;
+
+       h = nf_conntrack_find_get(&intuple, NULL);
+       if (!h) {
+               DEBUGP("icmpv6_error: no match\n");
+               return -NF_ACCEPT;
+       } else {
+               if (NF_CT_DIRECTION(h) == IP_CT_DIR_REPLY)
+                       *ctinfo += IP_CT_IS_REPLY;
+       }
+
+       /* Update skb to refer to this connection */
+       skb->nfct = &nf_ct_tuplehash_to_ctrack(h)->ct_general;
+       skb->nfctinfo = *ctinfo;
+       return -NF_ACCEPT;
+}
+
+static int
+icmpv6_error(struct sk_buff *skb, unsigned int dataoff,
+            enum ip_conntrack_info *ctinfo, int pf, unsigned int hooknum)
+{
+       struct icmp6hdr _ih, *icmp6h;
+
+       icmp6h = skb_header_pointer(skb, dataoff, sizeof(_ih), &_ih);
+       if (icmp6h == NULL) {
+               if (LOG_INVALID(IPPROTO_ICMPV6))
+               nf_log_packet(PF_INET6, 0, skb, NULL, NULL, NULL,
+                             "nf_ct_icmpv6: short packet ");
+               return -NF_ACCEPT;
+       }
+
+       if (hooknum == NF_IP6_PRE_ROUTING &&
+           nf_ip6_checksum(skb, hooknum, dataoff, IPPROTO_ICMPV6)) {
+               nf_log_packet(PF_INET6, 0, skb, NULL, NULL, NULL,
+                             "nf_ct_icmpv6: ICMPv6 checksum failed\n");
+               return -NF_ACCEPT;
+       }
+
+       /* is not error message ? */
+       if (icmp6h->icmp6_type >= 128)
+               return NF_ACCEPT;
+
+       return icmpv6_error_message(skb, dataoff, ctinfo, hooknum);
+}
+
+#if defined(CONFIG_NF_CT_NETLINK) || \
+    defined(CONFIG_NF_CT_NETLINK_MODULE)
+
+#include <linux/netfilter/nfnetlink.h>
+#include <linux/netfilter/nfnetlink_conntrack.h>
+static int icmpv6_tuple_to_nfattr(struct sk_buff *skb,
+                                 const struct nf_conntrack_tuple *t)
+{
+       NFA_PUT(skb, CTA_PROTO_ICMPV6_ID, sizeof(u_int16_t),
+               &t->src.u.icmp.id);
+       NFA_PUT(skb, CTA_PROTO_ICMPV6_TYPE, sizeof(u_int8_t),
+               &t->dst.u.icmp.type);
+       NFA_PUT(skb, CTA_PROTO_ICMPV6_CODE, sizeof(u_int8_t),
+               &t->dst.u.icmp.code);
+
+       return 0;
+
+nfattr_failure:
+       return -1;
+}
+
+static const size_t cta_min_proto[CTA_PROTO_MAX] = {
+       [CTA_PROTO_ICMPV6_TYPE-1] = sizeof(u_int8_t),
+       [CTA_PROTO_ICMPV6_CODE-1] = sizeof(u_int8_t),
+       [CTA_PROTO_ICMPV6_ID-1]   = sizeof(u_int16_t)
+};
+
+static int icmpv6_nfattr_to_tuple(struct nfattr *tb[],
+                               struct nf_conntrack_tuple *tuple)
+{
+       if (!tb[CTA_PROTO_ICMPV6_TYPE-1]
+           || !tb[CTA_PROTO_ICMPV6_CODE-1]
+           || !tb[CTA_PROTO_ICMPV6_ID-1])
+               return -EINVAL;
+
+       if (nfattr_bad_size(tb, CTA_PROTO_MAX, cta_min_proto))
+               return -EINVAL;
+
+       tuple->dst.u.icmp.type =
+                       *(u_int8_t *)NFA_DATA(tb[CTA_PROTO_ICMPV6_TYPE-1]);
+       tuple->dst.u.icmp.code =
+                       *(u_int8_t *)NFA_DATA(tb[CTA_PROTO_ICMPV6_CODE-1]);
+       tuple->src.u.icmp.id =
+                       *(u_int16_t *)NFA_DATA(tb[CTA_PROTO_ICMPV6_ID-1]);
+
+       if (tuple->dst.u.icmp.type < 128
+           || tuple->dst.u.icmp.type - 128 >= sizeof(invmap)
+           || !invmap[tuple->dst.u.icmp.type - 128])
+               return -EINVAL;
+
+       return 0;
+}
+#endif
+
+struct nf_conntrack_protocol nf_conntrack_protocol_icmpv6 =
+{
+       .l3proto                = PF_INET6,
+       .proto                  = IPPROTO_ICMPV6,
+       .name                   = "icmpv6",
+       .pkt_to_tuple           = icmpv6_pkt_to_tuple,
+       .invert_tuple           = icmpv6_invert_tuple,
+       .print_tuple            = icmpv6_print_tuple,
+       .print_conntrack        = icmpv6_print_conntrack,
+       .packet                 = icmpv6_packet,
+       .new                    = icmpv6_new,
+       .error                  = icmpv6_error,
+#if defined(CONFIG_NF_CT_NETLINK) || \
+    defined(CONFIG_NF_CT_NETLINK_MODULE)
+       .tuple_to_nfattr        = icmpv6_tuple_to_nfattr,
+       .nfattr_to_tuple        = icmpv6_nfattr_to_tuple,
+#endif
+};
+
+EXPORT_SYMBOL(nf_conntrack_protocol_icmpv6);