Merge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
[linux-drm-fsl-dcu.git] / net / dccp / ipv6.c
index 7171a78671aa9b0e73732db88e5bee406b5ec445..7f51e8db3967280e7698fb1c916a64c6c4a5b24c 100644 (file)
 /* Socket used for sending RSTs and ACKs */
 static struct socket *dccp_v6_ctl_socket;
 
-static void dccp_v6_ctl_send_reset(struct sk_buff *skb);
-static void dccp_v6_reqsk_send_ack(struct sk_buff *skb,
-                                  struct request_sock *req);
-static void dccp_v6_send_check(struct sock *sk, int len, struct sk_buff *skb);
-
-static int dccp_v6_do_rcv(struct sock *sk, struct sk_buff *skb);
-
 static struct inet_connection_sock_af_ops dccp_ipv6_mapped;
 static struct inet_connection_sock_af_ops dccp_ipv6_af_ops;
 
@@ -65,205 +58,37 @@ static void dccp_v6_hash(struct sock *sk)
        }
 }
 
-static inline u16 dccp_v6_check(struct dccp_hdr *dh, int len,
-                               struct in6_addr *saddr,
-                               struct in6_addr *daddr,
-                               unsigned long base)
+/* add pseudo-header to DCCP checksum stored in skb->csum */
+static inline __sum16 dccp_v6_csum_finish(struct sk_buff *skb,
+                                     struct in6_addr *saddr,
+                                     struct in6_addr *daddr)
 {
-       return csum_ipv6_magic(saddr, daddr, len, IPPROTO_DCCP, base);
+       return csum_ipv6_magic(saddr, daddr, skb->len, IPPROTO_DCCP, skb->csum);
 }
 
-static __u32 dccp_v6_init_sequence(struct sock *sk, struct sk_buff *skb)
+static inline void dccp_v6_send_check(struct sock *sk, int unused_value,
+                                     struct sk_buff *skb)
 {
-       const struct dccp_hdr *dh = dccp_hdr(skb);
-
-       if (skb->protocol == htons(ETH_P_IPV6))
-               return secure_tcpv6_sequence_number(skb->nh.ipv6h->daddr.s6_addr32,
-                                                   skb->nh.ipv6h->saddr.s6_addr32,
-                                                   dh->dccph_dport,
-                                                   dh->dccph_sport);
+       struct ipv6_pinfo *np = inet6_sk(sk);
+       struct dccp_hdr *dh = dccp_hdr(skb);
 
-       return secure_dccp_sequence_number(skb->nh.iph->daddr,
-                                          skb->nh.iph->saddr,
-                                          dh->dccph_dport,
-                                          dh->dccph_sport);
+       dccp_csum_outgoing(skb);
+       dh->dccph_checksum = dccp_v6_csum_finish(skb, &np->saddr, &np->daddr);
 }
 
-static int dccp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
-                          int addr_len)
+static inline __u32 secure_dccpv6_sequence_number(__be32 *saddr, __be32 *daddr,
+                                                 __be16 sport, __be16 dport   )
 {
-       struct sockaddr_in6 *usin = (struct sockaddr_in6 *)uaddr;
-       struct inet_connection_sock *icsk = inet_csk(sk);
-       struct inet_sock *inet = inet_sk(sk);
-       struct ipv6_pinfo *np = inet6_sk(sk);
-       struct dccp_sock *dp = dccp_sk(sk);
-       struct in6_addr *saddr = NULL, *final_p = NULL, final;
-       struct flowi fl;
-       struct dst_entry *dst;
-       int addr_type;
-       int err;
-
-       dp->dccps_role = DCCP_ROLE_CLIENT;
-
-       if (addr_len < SIN6_LEN_RFC2133)
-               return -EINVAL;
-
-       if (usin->sin6_family != AF_INET6)
-               return -EAFNOSUPPORT;
-
-       memset(&fl, 0, sizeof(fl));
-
-       if (np->sndflow) {
-               fl.fl6_flowlabel = usin->sin6_flowinfo & IPV6_FLOWINFO_MASK;
-               IP6_ECN_flow_init(fl.fl6_flowlabel);
-               if (fl.fl6_flowlabel & IPV6_FLOWLABEL_MASK) {
-                       struct ip6_flowlabel *flowlabel;
-                       flowlabel = fl6_sock_lookup(sk, fl.fl6_flowlabel);
-                       if (flowlabel == NULL)
-                               return -EINVAL;
-                       ipv6_addr_copy(&usin->sin6_addr, &flowlabel->dst);
-                       fl6_sock_release(flowlabel);
-               }
-       }
-       /*
-        * connect() to INADDR_ANY means loopback (BSD'ism).
-        */
-       if (ipv6_addr_any(&usin->sin6_addr))
-               usin->sin6_addr.s6_addr[15] = 1;
-
-       addr_type = ipv6_addr_type(&usin->sin6_addr);
-
-       if (addr_type & IPV6_ADDR_MULTICAST)
-               return -ENETUNREACH;
-
-       if (addr_type & IPV6_ADDR_LINKLOCAL) {
-               if (addr_len >= sizeof(struct sockaddr_in6) &&
-                   usin->sin6_scope_id) {
-                       /* If interface is set while binding, indices
-                        * must coincide.
-                        */
-                       if (sk->sk_bound_dev_if &&
-                           sk->sk_bound_dev_if != usin->sin6_scope_id)
-                               return -EINVAL;
-
-                       sk->sk_bound_dev_if = usin->sin6_scope_id;
-               }
-
-               /* Connect to link-local address requires an interface */
-               if (!sk->sk_bound_dev_if)
-                       return -EINVAL;
-       }
-
-       ipv6_addr_copy(&np->daddr, &usin->sin6_addr);
-       np->flow_label = fl.fl6_flowlabel;
-
-       /*
-        * DCCP over IPv4
-        */
-       if (addr_type == IPV6_ADDR_MAPPED) {
-               u32 exthdrlen = icsk->icsk_ext_hdr_len;
-               struct sockaddr_in sin;
-
-               SOCK_DEBUG(sk, "connect: ipv4 mapped\n");
-
-               if (__ipv6_only_sock(sk))
-                       return -ENETUNREACH;
-
-               sin.sin_family = AF_INET;
-               sin.sin_port = usin->sin6_port;
-               sin.sin_addr.s_addr = usin->sin6_addr.s6_addr32[3];
-
-               icsk->icsk_af_ops = &dccp_ipv6_mapped;
-               sk->sk_backlog_rcv = dccp_v4_do_rcv;
-
-               err = dccp_v4_connect(sk, (struct sockaddr *)&sin, sizeof(sin));
-               if (err) {
-                       icsk->icsk_ext_hdr_len = exthdrlen;
-                       icsk->icsk_af_ops = &dccp_ipv6_af_ops;
-                       sk->sk_backlog_rcv = dccp_v6_do_rcv;
-                       goto failure;
-               } else {
-                       ipv6_addr_set(&np->saddr, 0, 0, htonl(0x0000FFFF),
-                                     inet->saddr);
-                       ipv6_addr_set(&np->rcv_saddr, 0, 0, htonl(0x0000FFFF),
-                                     inet->rcv_saddr);
-               }
-
-               return err;
-       }
-
-       if (!ipv6_addr_any(&np->rcv_saddr))
-               saddr = &np->rcv_saddr;
-
-       fl.proto = IPPROTO_DCCP;
-       ipv6_addr_copy(&fl.fl6_dst, &np->daddr);
-       ipv6_addr_copy(&fl.fl6_src, saddr ? saddr : &np->saddr);
-       fl.oif = sk->sk_bound_dev_if;
-       fl.fl_ip_dport = usin->sin6_port;
-       fl.fl_ip_sport = inet->sport;
-       security_sk_classify_flow(sk, &fl);
-
-       if (np->opt != NULL && np->opt->srcrt != NULL) {
-               const struct rt0_hdr *rt0 = (struct rt0_hdr *)np->opt->srcrt;
-
-               ipv6_addr_copy(&final, &fl.fl6_dst);
-               ipv6_addr_copy(&fl.fl6_dst, rt0->addr);
-               final_p = &final;
-       }
-
-       err = ip6_dst_lookup(sk, &dst, &fl);
-       if (err)
-               goto failure;
-
-       if (final_p)
-               ipv6_addr_copy(&fl.fl6_dst, final_p);
-
-       err = xfrm_lookup(&dst, &fl, sk, 0);
-       if (err < 0)
-               goto failure;
-
-       if (saddr == NULL) {
-               saddr = &fl.fl6_src;
-               ipv6_addr_copy(&np->rcv_saddr, saddr);
-       }
-
-       /* set the source address */
-       ipv6_addr_copy(&np->saddr, saddr);
-       inet->rcv_saddr = LOOPBACK4_IPV6;
-
-       __ip6_dst_store(sk, dst, NULL, NULL);
-
-       icsk->icsk_ext_hdr_len = 0;
-       if (np->opt != NULL)
-               icsk->icsk_ext_hdr_len = (np->opt->opt_flen +
-                                         np->opt->opt_nflen);
-
-       inet->dport = usin->sin6_port;
-
-       dccp_set_state(sk, DCCP_REQUESTING);
-       err = inet6_hash_connect(&dccp_death_row, sk);
-       if (err)
-               goto late_failure;
-       /* FIXME */
-#if 0
-       dp->dccps_gar = secure_dccp_v6_sequence_number(np->saddr.s6_addr32,
-                                                      np->daddr.s6_addr32,
-                                                      inet->sport,
-                                                      inet->dport);
-#endif
-       err = dccp_connect(sk);
-       if (err)
-               goto late_failure;
+       return secure_tcpv6_sequence_number(saddr, daddr, sport, dport);
+}
 
-       return 0;
+static inline __u32 dccp_v6_init_sequence(struct sk_buff *skb)
+{
+       return secure_dccpv6_sequence_number(skb->nh.ipv6h->daddr.s6_addr32,
+                                            skb->nh.ipv6h->saddr.s6_addr32,
+                                            dccp_hdr(skb)->dccph_dport,
+                                            dccp_hdr(skb)->dccph_sport     );
 
-late_failure:
-       dccp_set_state(sk, DCCP_CLOSED);
-       __sk_dst_reset(sk);
-failure:
-       inet->dport = 0;
-       sk->sk_route_caps = 0;
-       return err;
 }
 
 static void dccp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
@@ -277,7 +102,7 @@ static void dccp_v6_err(struct sk_buff *skb, struct inet6_skb_parm *opt,
        __u64 seq;
 
        sk = inet6_lookup(&dccp_hashinfo, &hdr->daddr, dh->dccph_dport,
-                         &hdr->saddr, dh->dccph_sport, skb->dev->ifindex);
+                         &hdr->saddr, dh->dccph_sport, inet6_iif(skb));
 
        if (sk == NULL) {
                ICMP6_INC_STATS_BH(__in6_dev_get(skb->dev), ICMP6_MIB_INERRORS);
@@ -464,16 +289,12 @@ static int dccp_v6_send_response(struct sock *sk, struct request_sock *req,
        if (skb != NULL) {
                struct dccp_hdr *dh = dccp_hdr(skb);
 
-               dh->dccph_checksum = dccp_v6_check(dh, skb->len,
-                                                  &ireq6->loc_addr,
-                                                  &ireq6->rmt_addr,
-                                                  csum_partial((char *)dh,
-                                                               skb->len,
-                                                               skb->csum));
+               dh->dccph_checksum = dccp_v6_csum_finish(skb,
+                                                        &ireq6->loc_addr,
+                                                        &ireq6->rmt_addr);
                ipv6_addr_copy(&fl.fl6_dst, &ireq6->rmt_addr);
                err = ip6_xmit(sk, skb, &fl, opt, 0);
-               if (err == NET_XMIT_CN)
-                       err = 0;
+               err = net_xmit_eval(err);
        }
 
 done:
@@ -489,32 +310,7 @@ static void dccp_v6_reqsk_destructor(struct request_sock *req)
                kfree_skb(inet6_rsk(req)->pktopts);
 }
 
-static struct request_sock_ops dccp6_request_sock_ops = {
-       .family         = AF_INET6,
-       .obj_size       = sizeof(struct dccp6_request_sock),
-       .rtx_syn_ack    = dccp_v6_send_response,
-       .send_ack       = dccp_v6_reqsk_send_ack,
-       .destructor     = dccp_v6_reqsk_destructor,
-       .send_reset     = dccp_v6_ctl_send_reset,
-};
-
-static struct timewait_sock_ops dccp6_timewait_sock_ops = {
-       .twsk_obj_size  = sizeof(struct dccp6_timewait_sock),
-};
-
-static void dccp_v6_send_check(struct sock *sk, int len, struct sk_buff *skb)
-{
-       struct ipv6_pinfo *np = inet6_sk(sk);
-       struct dccp_hdr *dh = dccp_hdr(skb);
-
-       dh->dccph_checksum = csum_ipv6_magic(&np->saddr, &np->daddr,
-                                            len, IPPROTO_DCCP,
-                                            csum_partial((char *)dh,
-                                                         dh->dccph_doff << 2,
-                                                         skb->csum));
-}
-
-static void dccp_v6_ctl_send_reset(struct sk_buff *rxskb)
+static void dccp_v6_ctl_send_reset(struct sock *sk, struct sk_buff *rxskb)
 {
        struct dccp_hdr *rxdh = dccp_hdr(rxskb), *dh;
        const u32 dccp_hdr_reset_len = sizeof(struct dccp_hdr) +
@@ -522,7 +318,7 @@ static void dccp_v6_ctl_send_reset(struct sk_buff *rxskb)
                                       sizeof(struct dccp_hdr_reset);
        struct sk_buff *skb;
        struct flowi fl;
-       u64 seqno;
+       u64 seqno = 0;
 
        if (rxdh->dccph_type == DCCP_PKT_RESET)
                return;
@@ -533,13 +329,11 @@ static void dccp_v6_ctl_send_reset(struct sk_buff *rxskb)
        skb = alloc_skb(dccp_v6_ctl_socket->sk->sk_prot->max_header,
                        GFP_ATOMIC);
        if (skb == NULL)
-               return;
+               return;
 
        skb_reserve(skb, dccp_v6_ctl_socket->sk->sk_prot->max_header);
 
-       skb->h.raw = skb_push(skb, dccp_hdr_reset_len);
-       dh = dccp_hdr(skb);
-       memset(dh, 0, dccp_hdr_reset_len);
+       dh = dccp_zeroed_hdr(skb, dccp_hdr_reset_len);
 
        /* Swap the send and the receive. */
        dh->dccph_type  = DCCP_PKT_RESET;
@@ -550,21 +344,21 @@ static void dccp_v6_ctl_send_reset(struct sk_buff *rxskb)
        dccp_hdr_reset(skb)->dccph_reset_code =
                                DCCP_SKB_CB(rxskb)->dccpd_reset_code;
 
-       /* See "8.3.1. Abnormal Termination" in draft-ietf-dccp-spec-11 */
-       seqno = 0;
+       /* See "8.3.1. Abnormal Termination" in RFC 4340 */
        if (DCCP_SKB_CB(rxskb)->dccpd_ack_seq != DCCP_PKT_WITHOUT_ACK_SEQ)
                dccp_set_seqno(&seqno, DCCP_SKB_CB(rxskb)->dccpd_ack_seq + 1);
 
        dccp_hdr_set_seq(dh, seqno);
-       dccp_hdr_set_ack(dccp_hdr_ack_bits(skb),
-                        DCCP_SKB_CB(rxskb)->dccpd_seq);
+       dccp_hdr_set_ack(dccp_hdr_ack_bits(skb), DCCP_SKB_CB(rxskb)->dccpd_seq);
+
+       dccp_csum_outgoing(skb);
+       dh->dccph_checksum = dccp_v6_csum_finish(skb, &rxskb->nh.ipv6h->saddr,
+                                                     &rxskb->nh.ipv6h->daddr);
 
        memset(&fl, 0, sizeof(fl));
        ipv6_addr_copy(&fl.fl6_dst, &rxskb->nh.ipv6h->saddr);
        ipv6_addr_copy(&fl.fl6_src, &rxskb->nh.ipv6h->daddr);
-       dh->dccph_checksum = csum_ipv6_magic(&fl.fl6_src, &fl.fl6_dst,
-                                            sizeof(*dh), IPPROTO_DCCP,
-                                            skb->csum);
+
        fl.proto = IPPROTO_DCCP;
        fl.oif = inet6_iif(rxskb);
        fl.fl_ip_dport = dh->dccph_dport;
@@ -584,60 +378,14 @@ static void dccp_v6_ctl_send_reset(struct sk_buff *rxskb)
        kfree_skb(skb);
 }
 
-static void dccp_v6_reqsk_send_ack(struct sk_buff *rxskb,
-                                  struct request_sock *req)
-{
-       struct flowi fl;
-       struct dccp_hdr *rxdh = dccp_hdr(rxskb), *dh;
-       const u32 dccp_hdr_ack_len = sizeof(struct dccp_hdr) +
-                                    sizeof(struct dccp_hdr_ext) +
-                                    sizeof(struct dccp_hdr_ack_bits);
-       struct sk_buff *skb;
-
-       skb = alloc_skb(dccp_v6_ctl_socket->sk->sk_prot->max_header,
-                       GFP_ATOMIC);
-       if (skb == NULL)
-               return;
-
-       skb_reserve(skb, dccp_v6_ctl_socket->sk->sk_prot->max_header);
-
-       skb->h.raw = skb_push(skb, dccp_hdr_ack_len);
-       dh = dccp_hdr(skb);
-       memset(dh, 0, dccp_hdr_ack_len);
-
-       /* Build DCCP header and checksum it. */
-       dh->dccph_type  = DCCP_PKT_ACK;
-       dh->dccph_sport = rxdh->dccph_dport;
-       dh->dccph_dport = rxdh->dccph_sport;
-       dh->dccph_doff  = dccp_hdr_ack_len / 4;
-       dh->dccph_x     = 1;
-
-       dccp_hdr_set_seq(dh, DCCP_SKB_CB(rxskb)->dccpd_ack_seq);
-       dccp_hdr_set_ack(dccp_hdr_ack_bits(skb),
-                        DCCP_SKB_CB(rxskb)->dccpd_seq);
-
-       memset(&fl, 0, sizeof(fl));
-       ipv6_addr_copy(&fl.fl6_dst, &rxskb->nh.ipv6h->saddr);
-       ipv6_addr_copy(&fl.fl6_src, &rxskb->nh.ipv6h->daddr);
-
-       /* FIXME: calculate checksum, IPv4 also should... */
-
-       fl.proto = IPPROTO_DCCP;
-       fl.oif = inet6_iif(rxskb);
-       fl.fl_ip_dport = dh->dccph_dport;
-       fl.fl_ip_sport = dh->dccph_sport;
-       security_req_classify_flow(req, &fl);
-
-       if (!ip6_dst_lookup(NULL, &skb->dst, &fl)) {
-               if (xfrm_lookup(&skb->dst, &fl, NULL, 0) >= 0) {
-                       ip6_xmit(dccp_v6_ctl_socket->sk, skb, &fl, NULL, 0);
-                       DCCP_INC_STATS_BH(DCCP_MIB_OUTSEGS);
-                       return;
-               }
-       }
-
-       kfree_skb(skb);
-}
+static struct request_sock_ops dccp6_request_sock_ops = {
+       .family         = AF_INET6,
+       .obj_size       = sizeof(struct dccp6_request_sock),
+       .rtx_syn_ack    = dccp_v6_send_response,
+       .send_ack       = dccp_reqsk_send_ack,
+       .destructor     = dccp_v6_reqsk_destructor,
+       .send_reset     = dccp_v6_ctl_send_reset,
+};
 
 static struct sock *dccp_v6_hnd_req(struct sock *sk,struct sk_buff *skb)
 {
@@ -672,13 +420,11 @@ static struct sock *dccp_v6_hnd_req(struct sock *sk,struct sk_buff *skb)
 
 static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
 {
-       struct inet_request_sock *ireq;
-       struct dccp_sock dp;
        struct request_sock *req;
        struct dccp_request_sock *dreq;
        struct inet6_request_sock *ireq6;
        struct ipv6_pinfo *np = inet6_sk(sk);
-       const __be32 service = dccp_hdr_request(skb)->dccph_req_service;
+       const __be32 service = dccp_hdr_request(skb)->dccph_req_service;
        struct dccp_skb_cb *dcb = DCCP_SKB_CB(skb);
        __u8 reset_code = DCCP_RESET_CODE_TOO_BUSY;
 
@@ -691,7 +437,7 @@ static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
        if (dccp_bad_service_code(sk, service)) {
                reset_code = DCCP_RESET_CODE_BAD_SERVICE_CODE;
                goto drop;
-       }
+       }
        /*
         * There are no SYN attacks on IPv6, yet...
         */
@@ -701,22 +447,21 @@ static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
        if (sk_acceptq_is_full(sk) && inet_csk_reqsk_queue_young(sk) > 1)
                goto drop;
 
-       req = inet6_reqsk_alloc(sk->sk_prot->rsk_prot);
+       req = inet6_reqsk_alloc(&dccp6_request_sock_ops);
        if (req == NULL)
                goto drop;
 
-       /* FIXME: process options */
+       if (dccp_parse_options(sk, skb))
+               goto drop_and_free;
 
-       dccp_openreq_init(req, &dp, skb);
+       dccp_reqsk_init(req, skb);
 
        if (security_inet_conn_request(sk, skb, req))
                goto drop_and_free;
 
        ireq6 = inet6_rsk(req);
-       ireq = inet_rsk(req);
        ipv6_addr_copy(&ireq6->rmt_addr, &skb->nh.ipv6h->saddr);
        ipv6_addr_copy(&ireq6->loc_addr, &skb->nh.ipv6h->daddr);
-       req->rcv_wnd    = dccp_feat_default_sequence_window;
        ireq6->pktopts  = NULL;
 
        if (ipv6_opt_accepted(sk, skb) ||
@@ -735,14 +480,14 @@ static int dccp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
        /*
         * Step 3: Process LISTEN state
         *
-        * Set S.ISR, S.GSR, S.SWL, S.SWH from packet or Init Cookie
+        *   Set S.ISR, S.GSR, S.SWL, S.SWH from packet or Init Cookie
         *
-        * In fact we defer setting S.GSR, S.SWL, S.SWH to
-        * dccp_create_openreq_child.
+        *   In fact we defer setting S.GSR, S.SWL, S.SWH to
+        *   dccp_create_openreq_child.
         */
        dreq = dccp_rsk(req);
        dreq->dreq_isr     = dcb->dccpd_seq;
-       dreq->dreq_iss     = dccp_v6_init_sequence(sk, skb);
+       dreq->dreq_iss     = dccp_v6_init_sequence(skb);
        dreq->dreq_service = service;
 
        if (dccp_v6_send_response(sk, req, NULL))
@@ -989,17 +734,49 @@ static int dccp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
           looks not very well thought. For now we latch
           options, received in the last packet, enqueued
           by tcp. Feel free to propose better solution.
-                                              --ANK (980728)
+                                              --ANK (980728)
         */
        if (np->rxopt.all)
+       /*
+        * FIXME: Add handling of IPV6_PKTOPTIONS skb. See the comments below
+        *        (wrt ipv6_pktopions) and net/ipv6/tcp_ipv6.c for an example.
+        */
                opt_skb = skb_clone(skb, GFP_ATOMIC);
 
        if (sk->sk_state == DCCP_OPEN) { /* Fast path */
                if (dccp_rcv_established(sk, skb, dccp_hdr(skb), skb->len))
                        goto reset;
+               if (opt_skb) {
+                       /* XXX This is where we would goto ipv6_pktoptions. */
+                       __kfree_skb(opt_skb);
+               }
                return 0;
        }
 
+       /*
+        *  Step 3: Process LISTEN state
+        *     If S.state == LISTEN,
+        *       If P.type == Request or P contains a valid Init Cookie option,
+        *            (* Must scan the packet's options to check for Init
+        *               Cookies.  Only Init Cookies are processed here,
+        *               however; other options are processed in Step 8.  This
+        *               scan need only be performed if the endpoint uses Init
+        *               Cookies *)
+        *            (* Generate a new socket and switch to that socket *)
+        *            Set S := new socket for this port pair
+        *            S.state = RESPOND
+        *            Choose S.ISS (initial seqno) or set from Init Cookies
+        *            Initialize S.GAR := S.ISS
+        *            Set S.ISR, S.GSR, S.SWL, S.SWH from packet or Init Cookies
+        *            Continue with S.state == RESPOND
+        *            (* A Response packet will be generated in Step 11 *)
+        *       Otherwise,
+        *            Generate Reset(No Connection) unless P.type == Reset
+        *            Drop packet and return
+        *
+        * NOTE: the check for the packet types is done in
+        *       dccp_rcv_state_process
+        */
        if (sk->sk_state == DCCP_LISTEN) {
                struct sock *nsk = dccp_v6_hnd_req(sk, skb);
 
@@ -1010,7 +787,7 @@ static int dccp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
                 * otherwise we just shortcircuit this and continue with
                 * the new socket..
                 */
-               if (nsk != sk) {
+               if (nsk != sk) {
                        if (dccp_child_process(sk, nsk, skb))
                                goto reset;
                        if (opt_skb != NULL)
@@ -1021,10 +798,14 @@ static int dccp_v6_do_rcv(struct sock *sk, struct sk_buff *skb)
 
        if (dccp_rcv_state_process(sk, skb, dccp_hdr(skb), skb->len))
                goto reset;
+       if (opt_skb) {
+               /* XXX This is where we would goto ipv6_pktoptions. */
+               __kfree_skb(opt_skb);
+       }
        return 0;
 
 reset:
-       dccp_v6_ctl_send_reset(skb);
+       dccp_v6_ctl_send_reset(sk, skb);
 discard:
        if (opt_skb != NULL)
                __kfree_skb(opt_skb);
@@ -1037,12 +818,20 @@ static int dccp_v6_rcv(struct sk_buff **pskb)
        const struct dccp_hdr *dh;
        struct sk_buff *skb = *pskb;
        struct sock *sk;
+       int min_cov;
 
-       /* Step 1: Check header basics: */
+       /* Step 1: Check header basics */
 
        if (dccp_invalid_packet(skb))
                goto discard_it;
 
+       /* Step 1: If header checksum is incorrect, drop packet and return. */
+       if (dccp_v6_csum_finish(skb, &skb->nh.ipv6h->saddr,
+                                    &skb->nh.ipv6h->daddr)) {
+               DCCP_WARN("dropped packet with invalid checksum\n");
+               goto discard_it;
+       }
+
        dh = dccp_hdr(skb);
 
        DCCP_SKB_CB(skb)->dccpd_seq  = dccp_hdr_seq(skb);
@@ -1054,63 +843,247 @@ static int dccp_v6_rcv(struct sk_buff **pskb)
                DCCP_SKB_CB(skb)->dccpd_ack_seq = dccp_hdr_ack_seq(skb);
 
        /* Step 2:
-        *      Look up flow ID in table and get corresponding socket */
+        *      Look up flow ID in table and get corresponding socket */
        sk = __inet6_lookup(&dccp_hashinfo, &skb->nh.ipv6h->saddr,
                            dh->dccph_sport,
                            &skb->nh.ipv6h->daddr, ntohs(dh->dccph_dport),
                            inet6_iif(skb));
        /*
         * Step 2:
-        *      If no socket ...
-        *              Generate Reset(No Connection) unless P.type == Reset
-        *              Drop packet and return
+        *      If no socket ...
         */
-       if (sk == NULL)
+       if (sk == NULL) {
+               dccp_pr_debug("failed to look up flow ID in table and "
+                             "get corresponding socket\n");
                goto no_dccp_socket;
+       }
 
        /*
         * Step 2:
-        *      ... or S.state == TIMEWAIT,
+        *      ... or S.state == TIMEWAIT,
         *              Generate Reset(No Connection) unless P.type == Reset
         *              Drop packet and return
         */
-       if (sk->sk_state == DCCP_TIME_WAIT)
-               goto do_time_wait;
+       if (sk->sk_state == DCCP_TIME_WAIT) {
+               dccp_pr_debug("sk->sk_state == DCCP_TIME_WAIT: do_time_wait\n");
+               inet_twsk_put(inet_twsk(sk));
+               goto no_dccp_socket;
+       }
+
+       /*
+        * RFC 4340, sec. 9.2.1: Minimum Checksum Coverage
+        *      o if MinCsCov = 0, only packets with CsCov = 0 are accepted
+        *      o if MinCsCov > 0, also accept packets with CsCov >= MinCsCov
+        */
+       min_cov = dccp_sk(sk)->dccps_pcrlen;
+       if (dh->dccph_cscov  &&  (min_cov == 0 || dh->dccph_cscov < min_cov))  {
+               dccp_pr_debug("Packet CsCov %d does not satisfy MinCsCov %d\n",
+                             dh->dccph_cscov, min_cov);
+               /* FIXME: send Data Dropped option (see also dccp_v4_rcv) */
+               goto discard_and_relse;
+       }
 
        if (!xfrm6_policy_check(sk, XFRM_POLICY_IN, skb))
                goto discard_and_relse;
 
-       return sk_receive_skb(sk, skb) ? -1 : 0;
+       return sk_receive_skb(sk, skb, 1) ? -1 : 0;
 
 no_dccp_socket:
        if (!xfrm6_policy_check(NULL, XFRM_POLICY_IN, skb))
                goto discard_it;
        /*
         * Step 2:
+        *      If no socket ...
         *              Generate Reset(No Connection) unless P.type == Reset
         *              Drop packet and return
         */
        if (dh->dccph_type != DCCP_PKT_RESET) {
                DCCP_SKB_CB(skb)->dccpd_reset_code =
                                        DCCP_RESET_CODE_NO_CONNECTION;
-               dccp_v6_ctl_send_reset(skb);
+               dccp_v6_ctl_send_reset(sk, skb);
        }
-discard_it:
-
-       /*
-        *      Discard frame
-        */
 
+discard_it:
        kfree_skb(skb);
        return 0;
 
 discard_and_relse:
        sock_put(sk);
        goto discard_it;
+}
+
+static int dccp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
+                          int addr_len)
+{
+       struct sockaddr_in6 *usin = (struct sockaddr_in6 *)uaddr;
+       struct inet_connection_sock *icsk = inet_csk(sk);
+       struct inet_sock *inet = inet_sk(sk);
+       struct ipv6_pinfo *np = inet6_sk(sk);
+       struct dccp_sock *dp = dccp_sk(sk);
+       struct in6_addr *saddr = NULL, *final_p = NULL, final;
+       struct flowi fl;
+       struct dst_entry *dst;
+       int addr_type;
+       int err;
+
+       dp->dccps_role = DCCP_ROLE_CLIENT;
+
+       if (addr_len < SIN6_LEN_RFC2133)
+               return -EINVAL;
+
+       if (usin->sin6_family != AF_INET6)
+               return -EAFNOSUPPORT;
+
+       memset(&fl, 0, sizeof(fl));
+
+       if (np->sndflow) {
+               fl.fl6_flowlabel = usin->sin6_flowinfo & IPV6_FLOWINFO_MASK;
+               IP6_ECN_flow_init(fl.fl6_flowlabel);
+               if (fl.fl6_flowlabel & IPV6_FLOWLABEL_MASK) {
+                       struct ip6_flowlabel *flowlabel;
+                       flowlabel = fl6_sock_lookup(sk, fl.fl6_flowlabel);
+                       if (flowlabel == NULL)
+                               return -EINVAL;
+                       ipv6_addr_copy(&usin->sin6_addr, &flowlabel->dst);
+                       fl6_sock_release(flowlabel);
+               }
+       }
+       /*
+        * connect() to INADDR_ANY means loopback (BSD'ism).
+        */
+       if (ipv6_addr_any(&usin->sin6_addr))
+               usin->sin6_addr.s6_addr[15] = 1;
+
+       addr_type = ipv6_addr_type(&usin->sin6_addr);
+
+       if (addr_type & IPV6_ADDR_MULTICAST)
+               return -ENETUNREACH;
+
+       if (addr_type & IPV6_ADDR_LINKLOCAL) {
+               if (addr_len >= sizeof(struct sockaddr_in6) &&
+                   usin->sin6_scope_id) {
+                       /* If interface is set while binding, indices
+                        * must coincide.
+                        */
+                       if (sk->sk_bound_dev_if &&
+                           sk->sk_bound_dev_if != usin->sin6_scope_id)
+                               return -EINVAL;
+
+                       sk->sk_bound_dev_if = usin->sin6_scope_id;
+               }
+
+               /* Connect to link-local address requires an interface */
+               if (!sk->sk_bound_dev_if)
+                       return -EINVAL;
+       }
+
+       ipv6_addr_copy(&np->daddr, &usin->sin6_addr);
+       np->flow_label = fl.fl6_flowlabel;
+
+       /*
+        * DCCP over IPv4
+        */
+       if (addr_type == IPV6_ADDR_MAPPED) {
+               u32 exthdrlen = icsk->icsk_ext_hdr_len;
+               struct sockaddr_in sin;
+
+               SOCK_DEBUG(sk, "connect: ipv4 mapped\n");
+
+               if (__ipv6_only_sock(sk))
+                       return -ENETUNREACH;
+
+               sin.sin_family = AF_INET;
+               sin.sin_port = usin->sin6_port;
+               sin.sin_addr.s_addr = usin->sin6_addr.s6_addr32[3];
+
+               icsk->icsk_af_ops = &dccp_ipv6_mapped;
+               sk->sk_backlog_rcv = dccp_v4_do_rcv;
+
+               err = dccp_v4_connect(sk, (struct sockaddr *)&sin, sizeof(sin));
+               if (err) {
+                       icsk->icsk_ext_hdr_len = exthdrlen;
+                       icsk->icsk_af_ops = &dccp_ipv6_af_ops;
+                       sk->sk_backlog_rcv = dccp_v6_do_rcv;
+                       goto failure;
+               } else {
+                       ipv6_addr_set(&np->saddr, 0, 0, htonl(0x0000FFFF),
+                                     inet->saddr);
+                       ipv6_addr_set(&np->rcv_saddr, 0, 0, htonl(0x0000FFFF),
+                                     inet->rcv_saddr);
+               }
+
+               return err;
+       }
+
+       if (!ipv6_addr_any(&np->rcv_saddr))
+               saddr = &np->rcv_saddr;
+
+       fl.proto = IPPROTO_DCCP;
+       ipv6_addr_copy(&fl.fl6_dst, &np->daddr);
+       ipv6_addr_copy(&fl.fl6_src, saddr ? saddr : &np->saddr);
+       fl.oif = sk->sk_bound_dev_if;
+       fl.fl_ip_dport = usin->sin6_port;
+       fl.fl_ip_sport = inet->sport;
+       security_sk_classify_flow(sk, &fl);
+
+       if (np->opt != NULL && np->opt->srcrt != NULL) {
+               const struct rt0_hdr *rt0 = (struct rt0_hdr *)np->opt->srcrt;
 
-do_time_wait:
-       inet_twsk_put(inet_twsk(sk));
-       goto no_dccp_socket;
+               ipv6_addr_copy(&final, &fl.fl6_dst);
+               ipv6_addr_copy(&fl.fl6_dst, rt0->addr);
+               final_p = &final;
+       }
+
+       err = ip6_dst_lookup(sk, &dst, &fl);
+       if (err)
+               goto failure;
+
+       if (final_p)
+               ipv6_addr_copy(&fl.fl6_dst, final_p);
+
+       err = xfrm_lookup(&dst, &fl, sk, 1);
+       if (err < 0)
+               goto failure;
+
+       if (saddr == NULL) {
+               saddr = &fl.fl6_src;
+               ipv6_addr_copy(&np->rcv_saddr, saddr);
+       }
+
+       /* set the source address */
+       ipv6_addr_copy(&np->saddr, saddr);
+       inet->rcv_saddr = LOOPBACK4_IPV6;
+
+       __ip6_dst_store(sk, dst, NULL, NULL);
+
+       icsk->icsk_ext_hdr_len = 0;
+       if (np->opt != NULL)
+               icsk->icsk_ext_hdr_len = (np->opt->opt_flen +
+                                         np->opt->opt_nflen);
+
+       inet->dport = usin->sin6_port;
+
+       dccp_set_state(sk, DCCP_REQUESTING);
+       err = inet6_hash_connect(&dccp_death_row, sk);
+       if (err)
+               goto late_failure;
+
+       dp->dccps_iss = secure_dccpv6_sequence_number(np->saddr.s6_addr32,
+                                                     np->daddr.s6_addr32,
+                                                     inet->sport, inet->dport);
+       err = dccp_connect(sk);
+       if (err)
+               goto late_failure;
+
+       return 0;
+
+late_failure:
+       dccp_set_state(sk, DCCP_CLOSED);
+       __sk_dst_reset(sk);
+failure:
+       inet->dport = 0;
+       sk->sk_route_caps = 0;
+       return err;
 }
 
 static struct inet_connection_sock_af_ops dccp_ipv6_af_ops = {
@@ -1173,6 +1146,10 @@ static int dccp_v6_destroy_sock(struct sock *sk)
        return inet6_destroy_sock(sk);
 }
 
+static struct timewait_sock_ops dccp6_timewait_sock_ops = {
+       .twsk_obj_size  = sizeof(struct dccp6_timewait_sock),
+};
+
 static struct proto dccp_v6_prot = {
        .name              = "DCCPv6",
        .owner             = THIS_MODULE,