hammer2 - Work on concurrent bulkfree stability

[dragonfly.git] / sys / netinet / udp_usrreq.c

/*
 * Copyright (c) 2004 Jeffrey M. Hsu.  All rights reserved.
 * Copyright (c) 2004 The DragonFly Project.  All rights reserved.
 *
 * This code is derived from software contributed to The DragonFly Project
 * by Jeffrey M. Hsu.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of The DragonFly Project nor the names of its
 *    contributors may be used to endorse or promote products derived
 *    from this software without specific, prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

/*
 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
 *      The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *      @(#)udp_usrreq.c        8.6 (Berkeley) 5/23/95
 * $FreeBSD: src/sys/netinet/udp_usrreq.c,v 1.64.2.18 2003/01/24 05:11:34 sam Exp $
 */

#include "opt_ipsec.h"
#include "opt_inet6.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/domain.h>
#include <sys/proc.h>
#include <sys/priv.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sysctl.h>
#include <sys/syslog.h>
#include <sys/in_cksum.h>
#include <sys/ktr.h>

#include <sys/thread2.h>
#include <sys/socketvar2.h>
#include <sys/serialize.h>

#include <machine/stdarg.h>

#include <net/if.h>
#include <net/route.h>
#include <net/netmsg2.h>
#include <net/netisr2.h>

#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#ifdef INET6
#include <netinet/ip6.h>
#endif
#include <netinet/in_pcb.h>
#include <netinet/in_var.h>
#include <netinet/ip_var.h>
#ifdef INET6
#include <netinet6/ip6_var.h>
#endif
#include <netinet/ip_icmp.h>
#include <netinet/icmp_var.h>
#include <netinet/udp.h>
#include <netinet/udp_var.h>

#ifdef FAST_IPSEC
#include <netproto/ipsec/ipsec.h>
#endif

#ifdef IPSEC
#include <netinet6/ipsec.h>
#endif

#define MSGF_UDP_SEND           MSGF_PROTO1

#define INP_DIRECT_DETACH       INP_FLAG_PROTO2

#define UDP_KTR_STRING          "inp=%p"
#define UDP_KTR_ARGS            struct inpcb *inp

#ifndef KTR_UDP
#define KTR_UDP                 KTR_ALL
#endif

KTR_INFO_MASTER(udp);
KTR_INFO(KTR_UDP, udp, send_beg, 0, UDP_KTR_STRING, UDP_KTR_ARGS);
KTR_INFO(KTR_UDP, udp, send_end, 1, UDP_KTR_STRING, UDP_KTR_ARGS);
KTR_INFO(KTR_UDP, udp, send_ipout, 2, UDP_KTR_STRING, UDP_KTR_ARGS);
KTR_INFO(KTR_UDP, udp, redisp_ipout_beg, 3, UDP_KTR_STRING, UDP_KTR_ARGS);
KTR_INFO(KTR_UDP, udp, redisp_ipout_end, 4, UDP_KTR_STRING, UDP_KTR_ARGS);
KTR_INFO(KTR_UDP, udp, send_redisp, 5, UDP_KTR_STRING, UDP_KTR_ARGS);
KTR_INFO(KTR_UDP, udp, send_inswildcard, 6, UDP_KTR_STRING, UDP_KTR_ARGS);

#define logudp(name, inp)       KTR_LOG(udp_##name, inp)

/*
 * UDP protocol implementation.
 * Per RFC 768, August, 1980.
 */
#ifndef COMPAT_42
static int      udpcksum = 1;
#else
static int      udpcksum = 0;           /* XXX */
#endif
SYSCTL_INT(_net_inet_udp, UDPCTL_CHECKSUM, checksum, CTLFLAG_RW,
    &udpcksum, 0, "Enable checksumming of UDP packets");

int     log_in_vain = 0;
SYSCTL_INT(_net_inet_udp, OID_AUTO, log_in_vain, CTLFLAG_RW,
    &log_in_vain, 0, "Log all incoming UDP packets");

static int      blackhole = 0;
SYSCTL_INT(_net_inet_udp, OID_AUTO, blackhole, CTLFLAG_RW,
        &blackhole, 0, "Do not send port unreachables for refused connects");

static int      strict_mcast_mship = 1;
SYSCTL_INT(_net_inet_udp, OID_AUTO, strict_mcast_mship, CTLFLAG_RW,
        &strict_mcast_mship, 0, "Only send multicast to member sockets");

int     udp_sosend_async = 1;
SYSCTL_INT(_net_inet_udp, OID_AUTO, sosend_async, CTLFLAG_RW,
        &udp_sosend_async, 0, "UDP asynchronized pru_send");

int     udp_sosend_prepend = 1;
SYSCTL_INT(_net_inet_udp, OID_AUTO, sosend_prepend, CTLFLAG_RW,
        &udp_sosend_prepend, 0,
        "Prepend enough space for proto and link header in pru_send");

static int udp_reuseport_ext = 1;
SYSCTL_INT(_net_inet_udp, OID_AUTO, reuseport_ext, CTLFLAG_RW,
        &udp_reuseport_ext, 0, "SO_REUSEPORT extension");

struct  inpcbinfo udbinfo[MAXCPU];

#ifndef UDBHASHSIZE
#define UDBHASHSIZE 16
#endif

struct  udpstat udpstat_percpu[MAXCPU] __cachealign;

static void udp_append(struct inpcb *last, struct ip *ip,
    struct mbuf *n, int off, struct sockaddr_in *udp_in);

static int udp_connect_oncpu(struct inpcb *inp, struct sockaddr_in *sin,
    struct sockaddr_in *if_sin, uint16_t hash);

static boolean_t udp_inswildcardhash(struct inpcb *inp,
    struct netmsg_base *msg, int error);
static void udp_remwildcardhash(struct inpcb *inp);

static __inline int
udp_lportcpu(short lport)
{
        return (ntohs(lport) % netisr_ncpus);
}

void
udp_init(void)
{
        struct inpcbportinfo *portinfo;
        int cpu;

        portinfo = kmalloc_cachealign(sizeof(*portinfo) * netisr_ncpus, M_PCB,
            M_WAITOK);

        for (cpu = 0; cpu < netisr_ncpus; cpu++) {
                struct inpcbinfo *uicb = &udbinfo[cpu];

                /*
                 * NOTE:
                 * UDP pcb list, wildcard hash table and localgroup hash
                 * table are shared.
                 */
                in_pcbinfo_init(uicb, cpu, TRUE);
                uicb->hashbase = hashinit(UDBHASHSIZE, M_PCB, &uicb->hashmask);

                in_pcbportinfo_init(&portinfo[cpu], UDBHASHSIZE, cpu);
                in_pcbportinfo_set(uicb, portinfo, netisr_ncpus);

                uicb->wildcardhashbase = hashinit(UDBHASHSIZE, M_PCB,
                    &uicb->wildcardhashmask);
                uicb->localgrphashbase = hashinit(UDBHASHSIZE, M_PCB,
                    &uicb->localgrphashmask);

                uicb->ipi_size = sizeof(struct inpcb);
        }

        /*
         * Initialize UDP statistics counters for each CPU.
         */
        for (cpu = 0; cpu < netisr_ncpus; ++cpu)
                bzero(&udpstat_percpu[cpu], sizeof(struct udpstat));
}

static int
sysctl_udpstat(SYSCTL_HANDLER_ARGS)
{
        int cpu, error = 0;

        for (cpu = 0; cpu < netisr_ncpus; ++cpu) {
                if ((error = SYSCTL_OUT(req, &udpstat_percpu[cpu],
                                        sizeof(struct udpstat))))
                        break;
                if ((error = SYSCTL_IN(req, &udpstat_percpu[cpu],
                                       sizeof(struct udpstat))))
                        break;
        }

        return (error);
}
SYSCTL_PROC(_net_inet_udp, UDPCTL_STATS, stats, (CTLTYPE_OPAQUE | CTLFLAG_RW),
    0, 0, sysctl_udpstat, "S,udpstat", "UDP statistics");

void
udp_ctloutput(netmsg_t msg)
{
        struct socket *so = msg->base.nm_so;
        struct sockopt *sopt = msg->ctloutput.nm_sopt;
        struct inpcb *inp = so->so_pcb;

        if (inp == NULL) {
                lwkt_replymsg(&msg->lmsg, EINVAL);
                return;
        }

        if (sopt->sopt_level == IPPROTO_IP && sopt->sopt_dir == SOPT_SET) {
                switch (sopt->sopt_name) {
                case IP_MULTICAST_IF:
                case IP_MULTICAST_VIF:
                case IP_MULTICAST_TTL:
                case IP_MULTICAST_LOOP:
                case IP_ADD_MEMBERSHIP:
                case IP_DROP_MEMBERSHIP:
                        /*
                         * This pr_ctloutput msg will be forwarded
                         * to netisr0 to run; we can't do direct
                         * detaching anymore.
                         *
                         * NOTE:
                         * Don't optimize for the sockets whose
                         * current so_port is netisr0's msgport.
                         * These sockets could be connect(2)'ed
                         * later and the so_port will be changed.
                         */
                        inp->inp_flags &= ~INP_DIRECT_DETACH;
                        break;
                }
        }
        return ip_ctloutput(msg);
}

/*
 * Check multicast packets to make sure they are only sent to sockets with
 * multicast memberships for the packet's destination address and arrival
 * interface.  Multicast packets to multicast-unaware sockets are also
 * disallowed.
 *
 * Returns 0 if the packet is acceptable, -1 if it is not.
 */
static __inline int
check_multicast_membership(const struct ip *ip, const struct inpcb *inp,
    const struct mbuf *m)
{
        const struct ip_moptions *mopt;
        int mshipno;

        if (strict_mcast_mship == 0 ||
            !IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
                return (0);
        }

        ASSERT_NETISR0;

        mopt = inp->inp_moptions;
        if (mopt == NULL)
                return (-1);
        for (mshipno = 0; mshipno < mopt->imo_num_memberships; ++mshipno) {
                const struct in_multi *maddr = mopt->imo_membership[mshipno];

                if (ip->ip_dst.s_addr == maddr->inm_addr.s_addr &&
                    m->m_pkthdr.rcvif == maddr->inm_ifp) {
                        return (0);
                }
        }
        return (-1);
}

struct udp_mcast_arg {
        struct inpcb    *inp;
        struct inpcb    *last;
        struct ip       *ip;
        struct mbuf     *m;
        int             iphlen;
        struct sockaddr_in *udp_in;
};

static int
udp_mcast_input(struct udp_mcast_arg *arg)
{
        struct inpcb *inp = arg->inp;
        struct inpcb *last = arg->last;
        struct ip *ip = arg->ip;
        struct mbuf *m = arg->m;

        if (check_multicast_membership(ip, inp, m) < 0)
                return ERESTART; /* caller continue */

        if (last != NULL) {
                struct mbuf *n;

#ifdef IPSEC
                /* check AH/ESP integrity. */
                if (ipsec4_in_reject_so(m, last->inp_socket))
                        ipsecstat.in_polvio++;
                        /* do not inject data to pcb */
                else
#endif /*IPSEC*/
#ifdef FAST_IPSEC
                /* check AH/ESP integrity. */
                if (ipsec4_in_reject(m, last))
                        ;
                else
#endif /*FAST_IPSEC*/
                if ((n = m_copypacket(m, M_NOWAIT)) != NULL)
                        udp_append(last, ip, n,
                            arg->iphlen + sizeof(struct udphdr),
                            arg->udp_in);
        }
        arg->last = last = inp;

        /*
         * Don't look for additional matches if this one does
         * not have either the SO_REUSEPORT or SO_REUSEADDR
         * socket options set.  This heuristic avoids searching
         * through all pcbs in the common case of a non-shared
         * port.  It * assumes that an application will never
         * clear these options after setting them.
         */
        if (!(last->inp_socket->so_options &
            (SO_REUSEPORT | SO_REUSEADDR)))
                return EJUSTRETURN; /* caller stop */
        return 0;
}

int
udp_input(struct mbuf **mp, int *offp, int proto)
{
        struct sockaddr_in udp_in = { sizeof udp_in, AF_INET };
        int iphlen;
        struct ip *ip;
        struct udphdr *uh;
        struct inpcb *inp;
        struct mbuf *m;
        struct mbuf *opts = NULL;
        int len, off;
        struct ip save_ip;
        struct inpcbinfo *pcbinfo = &udbinfo[mycpuid];

        off = *offp;
        m = *mp;
        *mp = NULL;

        iphlen = off;
        udp_stat.udps_ipackets++;

        /*
         * Strip IP options, if any; should skip this,
         * make available to user, and use on returned packets,
         * but we don't yet have a way to check the checksum
         * with options still present.
         */
        if (iphlen > sizeof(struct ip)) {
                ip_stripoptions(m);
                iphlen = sizeof(struct ip);
        }

        /*
         * IP and UDP headers are together in first mbuf.
         * Already checked and pulled up in ip_demux().
         */
        KASSERT(m->m_len >= iphlen + sizeof(struct udphdr),
            ("UDP header not in one mbuf"));

        ip = mtod(m, struct ip *);
        uh = (struct udphdr *)((caddr_t)ip + iphlen);

        /* destination port of 0 is illegal, based on RFC768. */
        if (uh->uh_dport == 0)
                goto bad;

        /*
         * Make mbuf data length reflect UDP length.
         * If not enough data to reflect UDP length, drop.
         */
        len = ntohs((u_short)uh->uh_ulen);
        if (ip->ip_len != len) {
                if (len > ip->ip_len || len < sizeof(struct udphdr)) {
                        udp_stat.udps_badlen++;
                        goto bad;
                }
                m_adj(m, len - ip->ip_len);
                /* ip->ip_len = len; */
        }
        /*
         * Save a copy of the IP header in case we want restore it
         * for sending an ICMP error message in response.
         */
        save_ip = *ip;

        /*
         * Checksum extended UDP header and data.
         */
        if (uh->uh_sum) {
                if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
                        if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
                                uh->uh_sum = m->m_pkthdr.csum_data;
                        else
                                uh->uh_sum = in_pseudo(ip->ip_src.s_addr,
                                    ip->ip_dst.s_addr, htonl((u_short)len +
                                    m->m_pkthdr.csum_data + IPPROTO_UDP));
                        uh->uh_sum ^= 0xffff;
                } else {
                        char b[9];

                        bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
                        bzero(((struct ipovly *)ip)->ih_x1, 9);
                        ((struct ipovly *)ip)->ih_len = uh->uh_ulen;
                        uh->uh_sum = in_cksum(m, len + sizeof(struct ip));
                        bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
                }
                if (uh->uh_sum) {
                        udp_stat.udps_badsum++;
                        m_freem(m);
                        return(IPPROTO_DONE);
                }
        } else
                udp_stat.udps_nosum++;

        if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
            in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
                struct inpcbhead *connhead;
                struct inpcontainer *ic, *ic_marker;
                struct inpcontainerhead *ichead;
                struct udp_mcast_arg arg;
                struct inpcb *last;
                int error;

                /*
                 * Deliver a multicast or broadcast datagram to *all* sockets
                 * for which the local and remote addresses and ports match
                 * those of the incoming datagram.  This allows more than
                 * one process to receive multi/broadcasts on the same port.
                 * (This really ought to be done for unicast datagrams as
                 * well, but that would cause problems with existing
                 * applications that open both address-specific sockets and
                 * a wildcard socket listening to the same port -- they would
                 * end up receiving duplicates of every unicast datagram.
                 * Those applications open the multiple sockets to overcome an
                 * inadequacy of the UDP socket interface, but for backwards
                 * compatibility we avoid the problem here rather than
                 * fixing the interface.  Maybe 4.5BSD will remedy this?)
                 */

                /*
                 * Construct sockaddr format source address.
                 */
                udp_in.sin_port = uh->uh_sport;
                udp_in.sin_addr = ip->ip_src;
                arg.udp_in = &udp_in;
                /*
                 * Locate pcb(s) for datagram.
                 * (Algorithm copied from raw_intr().)
                 */
                last = NULL;
                arg.iphlen = iphlen;

                connhead = &pcbinfo->hashbase[
                    INP_PCBCONNHASH(ip->ip_src.s_addr, uh->uh_sport,
                    ip->ip_dst.s_addr, uh->uh_dport, pcbinfo->hashmask)];
                LIST_FOREACH(inp, connhead, inp_hash) {
#ifdef INET6
                        if (!INP_ISIPV4(inp))
                                continue;
#endif
                        if (!in_hosteq(inp->inp_faddr, ip->ip_src) ||
                            !in_hosteq(inp->inp_laddr, ip->ip_dst) ||
                            inp->inp_fport != uh->uh_sport ||
                            inp->inp_lport != uh->uh_dport)
                                continue;

                        arg.inp = inp;
                        arg.last = last;
                        arg.ip = ip;
                        arg.m = m;

                        error = udp_mcast_input(&arg);
                        if (error == ERESTART)
                                continue;
                        last = arg.last;

                        if (error == EJUSTRETURN)
                                goto done;
                }

                ichead = &pcbinfo->wildcardhashbase[
                    INP_PCBWILDCARDHASH(uh->uh_dport,
                    pcbinfo->wildcardhashmask)];
                ic_marker = in_pcbcontainer_marker();

                GET_PCBINFO_TOKEN(pcbinfo);
                LIST_INSERT_HEAD(ichead, ic_marker, ic_list);
                while ((ic = LIST_NEXT(ic_marker, ic_list)) != NULL) {
                        LIST_REMOVE(ic_marker, ic_list);
                        LIST_INSERT_AFTER(ic, ic_marker, ic_list);

                        inp = ic->ic_inp;
                        if (inp->inp_flags & INP_PLACEMARKER)
                                continue;
#ifdef INET6
                        if (!INP_ISIPV4(inp))
                                continue;
#endif
                        if (inp->inp_lport != uh->uh_dport)
                                continue;
                        if (inp->inp_laddr.s_addr != INADDR_ANY &&
                            inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
                                continue;

                        arg.inp = inp;
                        arg.last = last;
                        arg.ip = ip;
                        arg.m = m;

                        error = udp_mcast_input(&arg);
                        if (error == ERESTART)
                                continue;
                        last = arg.last;

                        if (error == EJUSTRETURN)
                                break;
                }
                LIST_REMOVE(ic_marker, ic_list);
                REL_PCBINFO_TOKEN(pcbinfo);
done:
                if (last == NULL) {
                        /*
                         * No matching pcb found; discard datagram.
                         * (No need to send an ICMP Port Unreachable
                         * for a broadcast or multicast datgram.)
                         */
                        udp_stat.udps_noportbcast++;
                        goto bad;
                }
#ifdef IPSEC
                /* check AH/ESP integrity. */
                if (ipsec4_in_reject_so(m, last->inp_socket)) {
                        ipsecstat.in_polvio++;
                        goto bad;
                }
#endif /*IPSEC*/
#ifdef FAST_IPSEC
                /* check AH/ESP integrity. */
                if (ipsec4_in_reject(m, last))
                        goto bad;
#endif /*FAST_IPSEC*/
                udp_append(last, ip, m, iphlen + sizeof(struct udphdr),
                    &udp_in);
                return(IPPROTO_DONE);
        }
        /*
         * Locate pcb for datagram.
         */
        inp = in_pcblookup_pkthash(pcbinfo, ip->ip_src, uh->uh_sport,
            ip->ip_dst, uh->uh_dport, TRUE, m->m_pkthdr.rcvif,
            udp_reuseport_ext ? m : NULL);
        if (inp == NULL) {
                if (log_in_vain) {
                        char src[INET_ADDRSTRLEN], dst[INET_ADDRSTRLEN];

                        log(LOG_INFO,
                            "Connection attempt to UDP %s:%d from %s:%d\n",
                            kinet_ntoa(ip->ip_dst, dst), ntohs(uh->uh_dport),
                            kinet_ntoa(ip->ip_src, src), ntohs(uh->uh_sport));
                }
                udp_stat.udps_noport++;
                if (m->m_flags & (M_BCAST | M_MCAST)) {
                        udp_stat.udps_noportbcast++;
                        goto bad;
                }
                if (blackhole)
                        goto bad;
#ifdef ICMP_BANDLIM
                if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
                        goto bad;
#endif
                *ip = save_ip;
                ip->ip_len += iphlen;
                icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
                return(IPPROTO_DONE);
        }
        KASSERT(INP_ISIPV4(inp), ("not inet inpcb"));
#ifdef IPSEC
        if (ipsec4_in_reject_so(m, inp->inp_socket)) {
                ipsecstat.in_polvio++;
                goto bad;
        }
#endif /*IPSEC*/
#ifdef FAST_IPSEC
        if (ipsec4_in_reject(m, inp))
                goto bad;
#endif /*FAST_IPSEC*/
        /*
         * Check the minimum TTL for socket.
         */
        if (ip->ip_ttl < inp->inp_ip_minttl)
                goto bad;

        /*
         * Construct sockaddr format source address.
         * Stuff source address and datagram in user buffer.
         */
        udp_in.sin_port = uh->uh_sport;
        udp_in.sin_addr = ip->ip_src;
        if ((inp->inp_flags & INP_CONTROLOPTS) ||
            (inp->inp_socket->so_options & SO_TIMESTAMP))
                ip_savecontrol(inp, &opts, ip, m);
        m_adj(m, iphlen + sizeof(struct udphdr));

        lwkt_gettoken(&inp->inp_socket->so_rcv.ssb_token);
        if (ssb_appendaddr(&inp->inp_socket->so_rcv,
            (struct sockaddr *)&udp_in, m, opts) == 0) {
                lwkt_reltoken(&inp->inp_socket->so_rcv.ssb_token);
                udp_stat.udps_fullsock++;
                goto bad;
        }
        lwkt_reltoken(&inp->inp_socket->so_rcv.ssb_token);
        sorwakeup(inp->inp_socket);
        return(IPPROTO_DONE);
bad:
        m_freem(m);
        if (opts)
                m_freem(opts);
        return(IPPROTO_DONE);
}

/*
 * subroutine of udp_input(), mainly for source code readability.
 * caller must properly init udp_ip6 and udp_in6 beforehand.
 */
static void
udp_append(struct inpcb *last, struct ip *ip, struct mbuf *n, int off,
    struct sockaddr_in *udp_in)
{
        struct mbuf *opts = NULL;
        int ret;

        KASSERT(INP_ISIPV4(last), ("not inet inpcb"));

        if (last->inp_flags & INP_CONTROLOPTS ||
            last->inp_socket->so_options & SO_TIMESTAMP)
                ip_savecontrol(last, &opts, ip, n);
        m_adj(n, off);

        lwkt_gettoken(&last->inp_socket->so_rcv.ssb_token);
        ret = ssb_appendaddr(&last->inp_socket->so_rcv,
            (struct sockaddr *)udp_in, n, opts);
        lwkt_reltoken(&last->inp_socket->so_rcv.ssb_token);
        if (ret == 0) {
                m_freem(n);
                if (opts)
                        m_freem(opts);
                udp_stat.udps_fullsock++;
        } else {
                sorwakeup(last->inp_socket);
        }
}

/*
 * Notify a udp user of an asynchronous error;
 * just wake up so that he can collect error status.
 */
void
udp_notify(struct inpcb *inp, int error)
{
        inp->inp_socket->so_error = error;
        sorwakeup(inp->inp_socket);
        sowwakeup(inp->inp_socket);
}

struct netmsg_udp_notify {
        struct netmsg_base base;
        inp_notify_t    nm_notify;
        struct in_addr  nm_faddr;
        int             nm_arg;
};

static void
udp_notifyall_oncpu(netmsg_t msg)
{
        struct netmsg_udp_notify *nm = (struct netmsg_udp_notify *)msg;
        int nextcpu, cpu = mycpuid;

        ASSERT_NETISR_NCPUS(cpu);

        in_pcbnotifyall(&udbinfo[cpu], nm->nm_faddr, nm->nm_arg, nm->nm_notify);

        nextcpu = cpu + 1;
        if (nextcpu < netisr_ncpus)
                lwkt_forwardmsg(netisr_cpuport(nextcpu), &nm->base.lmsg);
        else
                lwkt_replymsg(&nm->base.lmsg, 0);
}

inp_notify_t
udp_get_inpnotify(int cmd, const struct sockaddr *sa,
    struct ip **ip0, int *cpuid)
{
        struct in_addr faddr;
        struct ip *ip = *ip0;
        inp_notify_t notify = udp_notify;

        faddr = ((const struct sockaddr_in *)sa)->sin_addr;
        if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY)
                return NULL;

        if (PRC_IS_REDIRECT(cmd)) {
                ip = NULL;
                notify = in_rtchange;
        } else if (cmd == PRC_HOSTDEAD) {
                ip = NULL;
        } else if ((unsigned)cmd >= PRC_NCMDS || inetctlerrmap[cmd] == 0) {
                return NULL;
        }

        if (cpuid != NULL) {
                if (ip == NULL) {
                        /* Go through all effective netisr CPUs. */
                        *cpuid = netisr_ncpus;
                } else {
                        const struct udphdr *uh;

                        uh = (const struct udphdr *)
                            ((caddr_t)ip + (ip->ip_hl << 2));
                        *cpuid = udp_addrcpu(faddr.s_addr, uh->uh_dport,
                            ip->ip_src.s_addr, uh->uh_sport);
                }
        }

        *ip0 = ip;
        return notify;
}

void
udp_ctlinput(netmsg_t msg)
{
        struct sockaddr *sa = msg->ctlinput.nm_arg;
        struct ip *ip = msg->ctlinput.nm_extra;
        int cmd = msg->ctlinput.nm_cmd, cpuid;
        inp_notify_t notify;
        struct in_addr faddr;

        ASSERT_NETISR_NCPUS(mycpuid);

        notify = udp_get_inpnotify(cmd, sa, &ip, &cpuid);
        if (notify == NULL)
                goto done;

        faddr = ((struct sockaddr_in *)sa)->sin_addr;
        if (ip) {
                const struct udphdr *uh;
                struct inpcb *inp;

                if (cpuid != mycpuid)
                        goto done;

                uh = (const struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2));
                inp = in_pcblookup_hash(&udbinfo[mycpuid], faddr, uh->uh_dport,
                                        ip->ip_src, uh->uh_sport, 0, NULL);
                if (inp != NULL && inp->inp_socket != NULL)
                        notify(inp, inetctlerrmap[cmd]);
        } else if (msg->ctlinput.nm_direct) {
                if (cpuid != netisr_ncpus && cpuid != mycpuid)
                        goto done;

                in_pcbnotifyall(&udbinfo[mycpuid], faddr, inetctlerrmap[cmd],
                    notify);
        } else {
                struct netmsg_udp_notify *nm;

                ASSERT_NETISR0;
                nm = kmalloc(sizeof(*nm), M_LWKTMSG, M_INTWAIT);
                netmsg_init(&nm->base, NULL, &netisr_afree_rport,
                            0, udp_notifyall_oncpu);
                nm->nm_faddr = faddr;
                nm->nm_arg = inetctlerrmap[cmd];
                nm->nm_notify = notify;
                lwkt_sendmsg(netisr_cpuport(0), &nm->base.lmsg);
        }
done:
        lwkt_replymsg(&msg->lmsg, 0);
}

SYSCTL_PROC(_net_inet_udp, UDPCTL_PCBLIST, pcblist, CTLFLAG_RD, udbinfo, 0,
            in_pcblist_ncpus, "S,xinpcb", "List of active UDP sockets");

static int
udp_getcred(SYSCTL_HANDLER_ARGS)
{
        struct sockaddr_in addrs[2];
        struct ucred cred0, *cred = NULL;
        struct inpcb *inp;
        int error, cpu, origcpu;

        error = priv_check(req->td, PRIV_ROOT);
        if (error)
                return (error);
        error = SYSCTL_IN(req, addrs, sizeof addrs);
        if (error)
                return (error);

        origcpu = mycpuid;
        cpu = udp_addrcpu(addrs[1].sin_addr.s_addr, addrs[1].sin_port,
            addrs[0].sin_addr.s_addr, addrs[0].sin_port);

        lwkt_migratecpu(cpu);

        inp = in_pcblookup_hash(&udbinfo[cpu],
            addrs[1].sin_addr, addrs[1].sin_port,
            addrs[0].sin_addr, addrs[0].sin_port, TRUE, NULL);
        if (inp == NULL || inp->inp_socket == NULL) {
                error = ENOENT;
        } else if (inp->inp_socket->so_cred != NULL) {
                cred0 = *(inp->inp_socket->so_cred);
                cred = &cred0;
        }

        lwkt_migratecpu(origcpu);

        if (error)
                return error;

        return SYSCTL_OUT(req, cred, sizeof(struct ucred));
}
SYSCTL_PROC(_net_inet_udp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW,
    0, 0, udp_getcred, "S,ucred", "Get the ucred of a UDP connection");

static void
udp_send_redispatch(netmsg_t msg)
{
        struct mbuf *m = msg->send.nm_m;
        int pru_flags = msg->send.nm_flags;
        struct inpcb *inp = msg->send.base.nm_so->so_pcb;
        struct mbuf *m_opt = msg->send.nm_control; /* XXX save ipopt */
        int flags = msg->send.nm_priv; /* ip_output flags */
        int error;

        logudp(redisp_ipout_beg, inp);

        /*
         * - Don't use inp route cache.  It should only be used in the
         *   inp owner netisr.
         * - Access to inp_moptions should be safe, since multicast UDP
         *   datagrams are redispatched to netisr0 and inp_moptions is
         *   changed only in netisr0.
         */
        error = ip_output(m, m_opt, NULL, flags, inp->inp_moptions, inp);
        if ((pru_flags & PRUS_NOREPLY) == 0)
                lwkt_replymsg(&msg->send.base.lmsg, error);

        if (m_opt != NULL) {
                /* Free saved ip options, if any */
                m_freem(m_opt);
        }

        logudp(redisp_ipout_end, inp);
}

static void
udp_send(netmsg_t msg)
{
        struct socket *so = msg->send.base.nm_so;
        struct mbuf *m = msg->send.nm_m;
        struct sockaddr *dstaddr = msg->send.nm_addr;
        int pru_flags = msg->send.nm_flags;
        struct inpcb *inp = so->so_pcb;
        struct thread *td = msg->send.nm_td;
        uint16_t hash;
        int flags;

        struct udpiphdr *ui;
        int len = m->m_pkthdr.len;
        struct sockaddr_in *sin;        /* really is initialized before use */
        int error = 0, cpu;

        KKASSERT(msg->send.nm_control == NULL);

        logudp(send_beg, inp);

        if (inp == NULL) {
                error = EINVAL;
                goto release;
        }

        if (len + sizeof(struct udpiphdr) > IP_MAXPACKET) {
                error = EMSGSIZE;
                goto release;
        }

        if (inp->inp_lport == 0) {      /* unbound socket */
                boolean_t forwarded;

                error = in_pcbbind(inp, NULL, td);
                if (error)
                        goto release;

                /*
                 * Need to call udp_send again, after this inpcb is
                 * inserted into wildcard hash table.
                 */
                msg->send.base.lmsg.ms_flags |= MSGF_UDP_SEND;
                forwarded = udp_inswildcardhash(inp, &msg->send.base, 0);
                if (forwarded) {
                        /*
                         * The message is further forwarded, so we are
                         * done here.
                         */
                        logudp(send_inswildcard, inp);
                        return;
                }
        }

        if (dstaddr != NULL) {          /* destination address specified */
                if (inp->inp_faddr.s_addr != INADDR_ANY) {
                        /* already connected */
                        error = EISCONN;
                        goto release;
                }
                sin = (struct sockaddr_in *)dstaddr;
                if (!prison_remote_ip(td, (struct sockaddr *)&sin)) {
                        error = EAFNOSUPPORT; /* IPv6 only jail */
                        goto release;
                }
        } else {
                if (inp->inp_faddr.s_addr == INADDR_ANY) {
                        /* no destination specified and not already connected */
                        error = ENOTCONN;
                        goto release;
                }
                sin = NULL;
        }

        /*
         * Calculate data length and get a mbuf
         * for UDP and IP headers.
         */
        M_PREPEND(m, sizeof(struct udpiphdr), M_NOWAIT);
        if (m == NULL) {
                error = ENOBUFS;
                goto release;
        }

        /*
         * Fill in mbuf with extended UDP header
         * and addresses and length put into network format.
         */
        ui = mtod(m, struct udpiphdr *);
        bzero(ui->ui_x1, sizeof ui->ui_x1);     /* XXX still needed? */
        ui->ui_pr = IPPROTO_UDP;

        /*
         * Set destination address.
         */
        if (dstaddr != NULL) {                  /* use specified destination */
                ui->ui_dst = sin->sin_addr;
                ui->ui_dport = sin->sin_port;
        } else {                                /* use connected destination */
                ui->ui_dst = inp->inp_faddr;
                ui->ui_dport = inp->inp_fport;
        }

        /*
         * Set source address.
         */
        if (inp->inp_laddr.s_addr == INADDR_ANY ||
            IN_MULTICAST(ntohl(inp->inp_laddr.s_addr))) {
                struct sockaddr_in *if_sin;

                if (dstaddr == NULL) {  
                        /*
                         * connect() had (or should have) failed because
                         * the interface had no IP address, but the
                         * application proceeded to call send() anyways.
                         */
                        error = ENOTCONN;
                        goto release;
                }

                /* Look up outgoing interface. */
                error = in_pcbladdr_find(inp, dstaddr, &if_sin, td, 1);
                if (error)
                        goto release;
                ui->ui_src = if_sin->sin_addr;  /* use address of interface */
        } else {
                ui->ui_src = inp->inp_laddr;    /* use non-null bound address */
        }
        ui->ui_sport = inp->inp_lport;
        KASSERT(inp->inp_lport != 0, ("inp lport should have been bound"));

        /*
         * Release the original thread, since it is no longer used
         */
        if (pru_flags & PRUS_HELDTD) {
                lwkt_rele(td);
                pru_flags &= ~PRUS_HELDTD;
        }
        /*
         * Free the dest address, since it is no longer needed
         */
        if (pru_flags & PRUS_FREEADDR) {
                kfree(dstaddr, M_SONAME);
                pru_flags &= ~PRUS_FREEADDR;
        }

        ui->ui_ulen = htons((u_short)len + sizeof(struct udphdr));

        /*
         * Set up checksum and output datagram.
         */
        if (udpcksum) {
                ui->ui_sum = in_pseudo(ui->ui_src.s_addr, ui->ui_dst.s_addr,
                    htons((u_short)len + sizeof(struct udphdr) + IPPROTO_UDP));
                m->m_pkthdr.csum_flags = CSUM_UDP;
                m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum);
                m->m_pkthdr.csum_thlen = sizeof(struct udphdr);
        } else {
                ui->ui_sum = 0;
        }
        ((struct ip *)ui)->ip_len = sizeof(struct udpiphdr) + len;
        ((struct ip *)ui)->ip_ttl = inp->inp_ip_ttl;    /* XXX */
        ((struct ip *)ui)->ip_tos = inp->inp_ip_tos;    /* XXX */
        udp_stat.udps_opackets++;

        flags = IP_DEBUGROUTE |
            (inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST));
        if (pru_flags & PRUS_DONTROUTE)
                flags |= SO_DONTROUTE;

        if (inp->inp_flags & INP_CONNECTED) {
                /*
                 * For connected socket, this datagram has already
                 * been in the correct netisr; no need to rehash.
                 */
                KASSERT(inp->inp_flags & INP_HASH, ("inpcb has no hash"));
                m_sethash(m, inp->inp_hashval);
                goto sendit;
        }

        hash = udp_addrhash(ui->ui_dst.s_addr, ui->ui_dport,
            ui->ui_src.s_addr, ui->ui_sport);
        m_sethash(m, hash);

        cpu = netisr_hashcpu(hash);
        if (cpu != mycpuid) {
                struct mbuf *m_opt = NULL;
                struct netmsg_pru_send *smsg;
                struct lwkt_port *port = netisr_cpuport(cpu);

                /*
                 * Not on the CPU that matches this UDP datagram hash;
                 * redispatch to the correct CPU to do the ip_output().
                 */
                if (inp->inp_options != NULL) {
                        /*
                         * If there are ip options, then save a copy,
                         * since accessing inp_options on other CPUs'
                         * is not safe.
                         *
                         * XXX optimize this?
                         */
                        m_opt = m_copym(inp->inp_options, 0, M_COPYALL,
                            M_WAITOK);
                }
                if ((pru_flags & PRUS_NOREPLY) == 0) {
                        /*
                         * Change some parts of the original netmsg and
                         * forward it to the target netisr.
                         *
                         * NOTE: so_port MUST NOT be checked in the target
                         * netisr.
                         */
                        smsg = &msg->send;
                        smsg->nm_priv = flags; /* ip_output flags */
                        smsg->nm_m = m;
                        smsg->nm_control = m_opt; /* XXX save ipopt */
                        smsg->base.lmsg.ms_flags |= MSGF_IGNSOPORT;
                        smsg->base.nm_dispatch = udp_send_redispatch;
                        lwkt_forwardmsg(port, &smsg->base.lmsg);
                } else {
                        /*
                         * Recreate the netmsg, since the original mbuf
                         * could have been changed.  And send it to the
                         * target netisr.
                         *
                         * NOTE: so_port MUST NOT be checked in the target
                         * netisr.
                         */
                        smsg = &m->m_hdr.mh_sndmsg;
                        netmsg_init(&smsg->base, so, &netisr_apanic_rport,
                            MSGF_IGNSOPORT, udp_send_redispatch);
                        smsg->nm_priv = flags; /* ip_output flags */
                        smsg->nm_flags = pru_flags;
                        smsg->nm_m = m;
                        smsg->nm_control = m_opt; /* XXX save ipopt */
                        lwkt_sendmsg(port, &smsg->base.lmsg);
                }

                /* This UDP datagram is redispatched; done */
                logudp(send_redisp, inp);
                return;
        }

sendit:
        logudp(send_ipout, inp);
        error = ip_output(m, inp->inp_options, &inp->inp_route, flags,
            inp->inp_moptions, inp);
        m = NULL;

release:
        if (m != NULL)
                m_freem(m);

        if (pru_flags & PRUS_HELDTD)
                lwkt_rele(td);
        if (pru_flags & PRUS_FREEADDR)
                kfree(dstaddr, M_SONAME);
        if ((pru_flags & PRUS_NOREPLY) == 0)
                lwkt_replymsg(&msg->send.base.lmsg, error);

        logudp(send_end, inp);
}

u_long  udp_sendspace = 9216;           /* really max datagram size */
                                        /* 40 1K datagrams */
SYSCTL_INT(_net_inet_udp, UDPCTL_MAXDGRAM, maxdgram, CTLFLAG_RW,
    &udp_sendspace, 0, "Maximum outgoing UDP datagram size");

u_long  udp_recvspace = 40 * (1024 +
#ifdef INET6
                                      sizeof(struct sockaddr_in6)
#else
                                      sizeof(struct sockaddr_in)
#endif
                                      );
SYSCTL_INT(_net_inet_udp, UDPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
    &udp_recvspace, 0, "Maximum incoming UDP datagram size");

/*
 * This should never happen, since UDP socket does not support
 * connection acception (SO_ACCEPTCONN, i.e. listen(2)).
 */
static void
udp_abort(netmsg_t msg __unused)
{
        panic("udp_abort is called");
}

static int
udp_preattach(struct socket *so, int proto __unused, struct pru_attach_info *ai)
{
        return soreserve(so, udp_sendspace, udp_recvspace, ai->sb_rlimit);
}

static void
udp_attach(netmsg_t msg)
{
        struct socket *so = msg->attach.base.nm_so;
        struct pru_attach_info *ai = msg->attach.nm_ai;
        struct inpcb *inp;
        int error;

        KASSERT(so->so_pcb == NULL, ("udp socket attached"));

        if (ai != NULL) {
                error = udp_preattach(so, 0 /* don't care */, ai);
                if (error)
                        goto out;
        } else {
                /* Post attach; do nothing */
        }

        error = in_pcballoc(so, &udbinfo[mycpuid]);
        if (error)
                goto out;

        inp = so->so_pcb;
        inp->inp_flags |= INP_DIRECT_DETACH;
        inp->inp_ip_ttl = ip_defttl;
        error = 0;
out:
        lwkt_replymsg(&msg->attach.base.lmsg, error);
}

static void
udp_inswildcard_replymsg(netmsg_t msg)
{
        lwkt_msg_t lmsg = &msg->lmsg;

        if (lmsg->ms_flags & MSGF_UDP_SEND) {
                udp_send(msg);
                /* msg is replied by udp_send() */
        } else {
                lwkt_replymsg(lmsg, lmsg->ms_error);
        }
}

static void
udp_soreuseport_dispatch(netmsg_t msg)
{
        /* This inpcb has already been in the wildcard hash. */
        in_pcblink_flags(msg->base.nm_so->so_pcb, &udbinfo[mycpuid], 0);
        udp_inswildcard_replymsg(msg);
}

static void
udp_sosetport(struct lwkt_msg *msg, lwkt_port_t port)
{
        sosetport(((struct netmsg_base *)msg)->nm_so, port);
}

static boolean_t
udp_inswildcardhash_oncpu(struct inpcb *inp, struct netmsg_base *msg)
{
        int cpu;

        KASSERT(inp->inp_pcbinfo == &udbinfo[mycpuid],
            ("not on owner cpu"));

        in_pcbinswildcardhash(inp);
        for (cpu = 0; cpu < netisr_ncpus; ++cpu) {
                if (cpu == mycpuid) {
                        /*
                         * This inpcb has been inserted by the above
                         * in_pcbinswildcardhash().
                         */
                        continue;
                }
                in_pcbinswildcardhash_oncpu(inp, &udbinfo[cpu]);
        }

        /* NOTE: inp_lgrpindex is _not_ assigned in jail. */
        if ((inp->inp_socket->so_options & SO_REUSEPORT) &&
            inp->inp_lgrpindex >= 0) {
                /*
                 * For SO_REUSEPORT socket, redistribute it based on its
                 * local group index.
                 */
                cpu = inp->inp_lgrpindex % netisr_ncpus;
                if (cpu != mycpuid) {
                        struct lwkt_port *port = netisr_cpuport(cpu);
                        lwkt_msg_t lmsg = &msg->lmsg;

                        /*
                         * We are moving the protocol processing port the
                         * socket is on, we have to unlink here and re-link
                         * on the target cpu (this inpcb is still left in
                         * the wildcard hash).
                         */
                        in_pcbunlink_flags(inp, &udbinfo[mycpuid], 0);
                        msg->nm_dispatch = udp_soreuseport_dispatch;

                        /*
                         * See the related comment in tcp_usrreq.c
                         * tcp_connect()
                         */
                        lwkt_setmsg_receipt(lmsg, udp_sosetport);
                        lwkt_forwardmsg(port, lmsg);
                        return TRUE; /* forwarded */
                }
        }
        return FALSE;
}

static void
udp_inswildcardhash_dispatch(netmsg_t msg)
{
        struct inpcb *inp = msg->base.nm_so->so_pcb;
        boolean_t forwarded;

        KASSERT(inp->inp_lport != 0, ("local port not set yet"));
        KASSERT(udp_lportcpu(inp->inp_lport) == mycpuid, ("not target cpu"));

        in_pcblink(inp, &udbinfo[mycpuid]);

        forwarded = udp_inswildcardhash_oncpu(inp, &msg->base);
        if (forwarded) {
                /* The message is further forwarded, so we are done here. */
                return;
        }
        udp_inswildcard_replymsg(msg);
}

static boolean_t
udp_inswildcardhash(struct inpcb *inp, struct netmsg_base *msg, int error)
{
        lwkt_msg_t lmsg = &msg->lmsg;
        int cpu;

        ASSERT_INP_NOTINHASH(inp);

        /* This inpcb could no longer be directly detached */
        inp->inp_flags &= ~INP_DIRECT_DETACH;

        /*
         * Always clear the route cache, so we don't need to
         * worry about any owner CPU changes later.
         */
        in_pcbresetroute(inp);

        KASSERT(inp->inp_lport != 0, ("local port not set yet"));
        cpu = udp_lportcpu(inp->inp_lport);

        lmsg->ms_error = error;
        if (cpu != mycpuid) {
                struct lwkt_port *port = netisr_cpuport(cpu);

                /*
                 * We are moving the protocol processing port the socket
                 * is on, we have to unlink here and re-link on the
                 * target cpu.
                 */
                in_pcbunlink(inp, &udbinfo[mycpuid]);
                msg->nm_dispatch = udp_inswildcardhash_dispatch;

                /* See the related comment in tcp_usrreq.c tcp_connect() */
                lwkt_setmsg_receipt(lmsg, udp_sosetport);
                lwkt_forwardmsg(port, lmsg);
                return TRUE; /* forwarded */
        }

        return udp_inswildcardhash_oncpu(inp, msg);
}

static void
udp_bind(netmsg_t msg)
{
        struct socket *so = msg->bind.base.nm_so;
        struct inpcb *inp;
        int error;

        inp = so->so_pcb;
        if (inp) {
                struct sockaddr *nam = msg->bind.nm_nam;
                struct thread *td = msg->bind.nm_td;
                struct sockaddr_in *sin;
                lwkt_port_t port;
                int cpu;

                /*
                 * Check "already bound" here (in_pcbbind() does the same
                 * check though), so we don't forward a connected/bound
                 * socket randomly which would panic in the following
                 * in_pcbunlink().
                 */
                if (inp->inp_lport != 0 ||
                    inp->inp_laddr.s_addr != INADDR_ANY) {
                        error = EINVAL; /* already bound */
                        goto done;
                }

                if (nam->sa_len != sizeof(*sin)) {
                        error = EINVAL;
                        goto done;
                }
                sin = (struct sockaddr_in *)nam;

                cpu = udp_lportcpu(sin->sin_port);
                port = netisr_cpuport(cpu);

                /*
                 * See the related comment in tcp_usrreq.c tcp_usr_bind().
                 * The exception is that we use local port based netisr
                 * to serialize in_pcbbind().
                 */
                if (&curthread->td_msgport != port) {
                        lwkt_msg_t lmsg = &msg->bind.base.lmsg;

                        KASSERT((msg->bind.nm_flags & PRUB_RELINK) == 0,
                            ("already asked to relink"));

                        in_pcbunlink(so->so_pcb, &udbinfo[mycpuid]);
                        msg->bind.nm_flags |= PRUB_RELINK;

                        /*
                         * See the related comment in tcp_usrreq.c
                         * tcp_connect().
                         */
                        lwkt_setmsg_receipt(lmsg, udp_sosetport);
                        lwkt_forwardmsg(port, lmsg);
                        /* msg invalid now */
                        return;
                }
                KASSERT(so->so_port == port, ("so_port is not netisr%d", cpu));

                if (msg->bind.nm_flags & PRUB_RELINK) {
                        msg->bind.nm_flags &= ~PRUB_RELINK;
                        in_pcblink(so->so_pcb, &udbinfo[mycpuid]);
                }
                KASSERT(inp->inp_pcbinfo == &udbinfo[cpu],
                    ("pcbinfo is not udbinfo%d", cpu));

                error = in_pcbbind(inp, nam, td);
                if (error == 0) {
                        boolean_t forwarded;

                        if (sin->sin_addr.s_addr != INADDR_ANY)
                                inp->inp_flags |= INP_WASBOUND_NOTANY;

                        forwarded = udp_inswildcardhash(inp,
                            &msg->bind.base, 0);
                        if (forwarded) {
                                /*
                                 * The message is further forwarded, so
                                 * we are done here.
                                 */
                                return;
                        }
                }
        } else {
                error = EINVAL;
        }
done:
        lwkt_replymsg(&msg->bind.base.lmsg, error);
}

static int
udp_preconnect(struct socket *so, const struct sockaddr *nam __unused,
    struct thread *td __unused)
{
        sosetstate(so, SS_ISCONNECTED);         /* XXX */
        return 0;
}

static void
udp_connect(netmsg_t msg)
{
        struct socket *so = msg->connect.base.nm_so;
        struct sockaddr *nam = msg->connect.nm_nam;
        struct thread *td = msg->connect.nm_td;
        struct inpcb *inp;
        struct sockaddr_in *sin = (struct sockaddr_in *)nam;
        struct sockaddr_in *if_sin;
        struct lwkt_port *port;
        uint16_t hash;
        int error;

        KKASSERT(msg->connect.nm_m == NULL);

        inp = so->so_pcb;
        if (inp == NULL) {
                error = EINVAL;
                goto out;
        }

        if (msg->connect.nm_flags & PRUC_RECONNECT) {
                msg->connect.nm_flags &= ~PRUC_RECONNECT;
                in_pcblink(inp, &udbinfo[mycpuid]);
        }

        if (inp->inp_faddr.s_addr != INADDR_ANY) {
                error = EISCONN;
                goto out;
        }
        error = 0;

        /*
         * Bind if we have to
         */
        if (inp->inp_lport == 0) {
                error = in_pcbbind(inp, NULL, td);
                if (error)
                        goto out;
        }

        /*
         * Calculate the correct protocol processing thread.  The connect
         * operation must run there.
         */
        error = in_pcbladdr(inp, nam, &if_sin, td);
        if (error)
                goto out;
        if (!prison_remote_ip(td, nam)) {
                error = EAFNOSUPPORT; /* IPv6 only jail */
                goto out;
        }

        hash = udp_addrhash(sin->sin_addr.s_addr, sin->sin_port,
            inp->inp_laddr.s_addr != INADDR_ANY ?
            inp->inp_laddr.s_addr : if_sin->sin_addr.s_addr, inp->inp_lport);
        port = netisr_hashport(hash);
        if (port != &curthread->td_msgport) {
                lwkt_msg_t lmsg = &msg->connect.base.lmsg;
                int nm_flags = PRUC_RECONNECT;

                /*
                 * in_pcbladdr() may have allocated a route entry for us
                 * on the current CPU, but we need a route entry on the
                 * inpcb's owner CPU, so free it here.
                 */
                in_pcbresetroute(inp);

                if (inp->inp_flags & INP_WILDCARD) {
                        /*
                         * Remove this inpcb from the wildcard hash before
                         * the socket's msgport changes.
                         */
                        udp_remwildcardhash(inp);
                }

                if (so->so_orig_port == NULL) {
                        /*
                         * First time change protocol processing port.
                         * Save the current port for synchronization upon
                         * udp_detach.
                         */
                        so->so_orig_port = &curthread->td_msgport;
                } else {
                        /*
                         * We have changed protocol processing port more
                         * than once.  We could not do direct detach
                         * anymore, because we lose the track of the
                         * original protocol processing ports to perform
                         * synchronization upon udp_detach.  This should
                         * be rare though.
                         */
                        inp->inp_flags &= ~INP_DIRECT_DETACH;
                }

                /*
                 * We are moving the protocol processing port the socket
                 * is on, we have to unlink here and re-link on the
                 * target cpu.
                 */
                in_pcbunlink(inp, &udbinfo[mycpuid]);
                msg->connect.nm_flags |= nm_flags;

                /* See the related comment in tcp_usrreq.c tcp_connect() */
                lwkt_setmsg_receipt(lmsg, udp_sosetport);
                lwkt_forwardmsg(port, lmsg);
                /* msg invalid now */
                return;
        }
        error = udp_connect_oncpu(inp, sin, if_sin, hash);
out:
        if (msg->connect.nm_flags & PRUC_HELDTD)
                lwkt_rele(td);
        if (error && (msg->connect.nm_flags & PRUC_ASYNC)) {
                if (inp->inp_lport == 0) {
                        /*
                         * As long as we have the local port, it is fine
                         * for connect to fail, e.g. disconnect.
                         */
                        so->so_error = error;
                }
                soclrstate(so, SS_ISCONNECTED);
                /*
                 * Wake up callers blocked on this socket to make sure
                 * that they can see this error.
                 *
                 * NOTE:
                 * sodisconnected() can't be used here, which bricks
                 * sending and receiving.
                 */
                wakeup(&so->so_timeo);
                sowwakeup(so);
                sorwakeup(so);
        }
        if (error && inp != NULL && inp->inp_lport != 0 &&
            (inp->inp_flags & INP_WILDCARD) == 0) {
                boolean_t forwarded;

                /* Connect failed; put it to wildcard hash. */
                forwarded = udp_inswildcardhash(inp, &msg->connect.base,
                    error);
                if (forwarded) {
                        /*
                         * The message is further forwarded, so we are done
                         * here.
                         */
                        return;
                }
        }
        lwkt_replymsg(&msg->connect.base.lmsg, error);
}

static void
udp_remwildcardhash(struct inpcb *inp)
{
        int cpu;

        KASSERT(inp->inp_pcbinfo == &udbinfo[mycpuid],
            ("not on owner cpu"));

        for (cpu = 0; cpu < netisr_ncpus; ++cpu) {
                if (cpu == mycpuid) {
                        /*
                         * This inpcb will be removed by the later
                         * in_pcbremwildcardhash().
                         */
                        continue;
                }
                in_pcbremwildcardhash_oncpu(inp, &udbinfo[cpu]);
        }
        in_pcbremwildcardhash(inp);
}

static int
udp_connect_oncpu(struct inpcb *inp, struct sockaddr_in *sin,
    struct sockaddr_in *if_sin, uint16_t hash)
{
        struct socket *so = inp->inp_socket;
        struct inpcb *oinp;

        oinp = in_pcblookup_hash(inp->inp_pcbinfo,
            sin->sin_addr, sin->sin_port,
            inp->inp_laddr.s_addr != INADDR_ANY ?
            inp->inp_laddr : if_sin->sin_addr, inp->inp_lport, FALSE, NULL);
        if (oinp != NULL)
                return EADDRINUSE;

        /*
         * No more errors can occur, finish adjusting the socket
         * and change the processing port to reflect the connected
         * socket.  Once set we can no longer safely mess with the
         * socket.
         */

        if (inp->inp_flags & INP_WILDCARD)
                udp_remwildcardhash(inp);

        if (inp->inp_laddr.s_addr == INADDR_ANY)
                inp->inp_laddr = if_sin->sin_addr;
        inp->inp_faddr = sin->sin_addr;
        inp->inp_fport = sin->sin_port;
        in_pcbinsconnhash(inp);

        inp->inp_flags |= INP_HASH;
        inp->inp_hashval = hash;

        soisconnected(so);

        return 0;
}

static void
udp_detach2(struct socket *so)
{
        in_pcbdetach(so->so_pcb);
        sodiscard(so);
        sofree(so);
}

static void
udp_detach_final_dispatch(netmsg_t msg)
{
        udp_detach2(msg->base.nm_so);
}

static void
udp_detach_oncpu_dispatch(netmsg_t msg)
{
        struct netmsg_base *clomsg = &msg->base;
        struct socket *so = clomsg->nm_so;
        struct inpcb *inp = so->so_pcb;
        struct thread *td = curthread;
        int nextcpu, cpuid = mycpuid;

        KASSERT(td->td_type == TD_TYPE_NETISR, ("not in netisr"));

        if (inp->inp_flags & INP_WILDCARD) {
                /*
                 * This inp will be removed on the inp's
                 * owner CPU later, so don't do it now.
                 */
                if (&td->td_msgport != so->so_port)
                        in_pcbremwildcardhash_oncpu(inp, &udbinfo[cpuid]);
        }

        if (cpuid == 0) {
                /*
                 * Free and clear multicast socket option,
                 * which is only accessed in netisr0.
                 */
                ip_freemoptions(inp->inp_moptions);
                inp->inp_moptions = NULL;
        }

        nextcpu = cpuid + 1;
        if (nextcpu < netisr_ncpus) {
                lwkt_forwardmsg(netisr_cpuport(nextcpu), &clomsg->lmsg);
        } else {
                /*
                 * No one could see this inpcb now; destroy this
                 * inpcb in its owner netisr.
                 */
                netmsg_init(clomsg, so, &netisr_apanic_rport, 0,
                    udp_detach_final_dispatch);
                lwkt_sendmsg(so->so_port, &clomsg->lmsg);
        }
}

static void
udp_detach_syncorig_dispatch(netmsg_t msg)
{
        struct netmsg_base *clomsg = &msg->base;
        struct socket *so = clomsg->nm_so;

        /*
         * Original protocol processing port is synchronized;
         * destroy this inpcb in its owner netisr.
         */
        netmsg_init(clomsg, so, &netisr_apanic_rport, 0,
            udp_detach_final_dispatch);
        lwkt_sendmsg(so->so_port, &clomsg->lmsg);
}

static void
udp_detach(netmsg_t msg)
{
        struct socket *so = msg->detach.base.nm_so;
        struct netmsg_base *clomsg;
        struct inpcb *inp;

        inp = so->so_pcb;
        if (inp == NULL) {
                lwkt_replymsg(&msg->detach.base.lmsg, EINVAL);
                return;
        }

        /*
         * Reply EJUSTRETURN ASAP, we will call sodiscard() and
         * sofree() later.
         */
        lwkt_replymsg(&msg->detach.base.lmsg, EJUSTRETURN);

        if (netisr_ncpus == 1) {
                /* Only one CPU, detach the inpcb directly. */
                udp_detach2(so);
                return;
        }

        /*
         * Remove this inpcb from the inpcb list first, so that
         * no one could find this inpcb from the inpcb list.
         */
        in_pcbofflist(inp);

        /*
         * Remove this inpcb from the local port hash directly
         * here, so that its bound local port could be recycled
         * timely.
         */
        in_pcbremporthash(inp);

        if (inp->inp_flags & INP_DIRECT_DETACH) {
                /*
                 * Direct detaching is allowed
                 */
                KASSERT((inp->inp_flags & INP_WILDCARD) == 0,
                    ("in the wildcardhash"));
                KASSERT(inp->inp_moptions == NULL, ("has mcast options"));
                if (so->so_orig_port == NULL) {
                        udp_detach2(so);
                } else {
                        /*
                         * Protocol processing port changed once, so
                         * we need to make sure that there are nothing
                         * left on the original protocol processing
                         * port before we destroy this socket and inpcb.
                         * This is more lightweight than going through
                         * all UDP processing netisrs.
                         */
                        clomsg = &so->so_clomsg;
                        netmsg_init(clomsg, so, &netisr_apanic_rport,
                            MSGF_IGNSOPORT, udp_detach_syncorig_dispatch);
                        lwkt_sendmsg(so->so_orig_port, &clomsg->lmsg);
                }
                return;
        }

        /*
         * Go through netisrs which process UDP to make sure
         * no one could find this inpcb anymore.
         */
        clomsg = &so->so_clomsg;
        netmsg_init(clomsg, so, &netisr_apanic_rport, MSGF_IGNSOPORT,
            udp_detach_oncpu_dispatch);
        lwkt_sendmsg(netisr_cpuport(0), &clomsg->lmsg);
}

static void
udp_disconnect(netmsg_t msg)
{
        struct socket *so = msg->disconnect.base.nm_so;
        struct inpcb *inp;
        boolean_t forwarded;
        int error = 0;

        inp = so->so_pcb;
        if (inp == NULL) {
                error = EINVAL;
                goto out;
        }
        if (inp->inp_faddr.s_addr == INADDR_ANY) {
                error = ENOTCONN;
                goto out;
        }

        soclrstate(so, SS_ISCONNECTED);         /* XXX */

        in_pcbdisconnect(inp);
        inp->inp_flags &= ~INP_HASH;

        /*
         * Follow traditional BSD behavior and retain the local port
         * binding.  But, fix the old misbehavior of overwriting any
         * previously bound local address.
         */
        if (!(inp->inp_flags & INP_WASBOUND_NOTANY))
                inp->inp_laddr.s_addr = INADDR_ANY;

        if (so->so_state & SS_ISCLOSING) {
                /*
                 * If this socket is being closed, there is no need
                 * to put this socket back into wildcard hash table.
                 */
                error = 0;
                goto out;
        }

        forwarded = udp_inswildcardhash(inp, &msg->disconnect.base, 0);
        if (forwarded) {
                /*
                 * The message is further forwarded, so we are done
                 * here.
                 */
                return;
        }
out:
        lwkt_replymsg(&msg->disconnect.base.lmsg, error);
}

void
udp_shutdown(netmsg_t msg)
{
        struct socket *so = msg->shutdown.base.nm_so;
        struct inpcb *inp;
        int error;

        inp = so->so_pcb;
        if (inp) {
                socantsendmore(so);
                error = 0;
        } else {
                error = EINVAL;
        }
        lwkt_replymsg(&msg->shutdown.base.lmsg, error);
}

struct pr_usrreqs udp_usrreqs = {
        .pru_abort = udp_abort,
        .pru_accept = pr_generic_notsupp,
        .pru_attach = udp_attach,
        .pru_bind = udp_bind,
        .pru_connect = udp_connect,
        .pru_connect2 = pr_generic_notsupp,
        .pru_control = in_control_dispatch,
        .pru_detach = udp_detach,
        .pru_disconnect = udp_disconnect,
        .pru_listen = pr_generic_notsupp,
        .pru_peeraddr = in_setpeeraddr_dispatch,
        .pru_rcvd = pr_generic_notsupp,
        .pru_rcvoob = pr_generic_notsupp,
        .pru_send = udp_send,
        .pru_sense = pru_sense_null,
        .pru_shutdown = udp_shutdown,
        .pru_sockaddr = in_setsockaddr_dispatch,
        .pru_sosend = sosendudp,
        .pru_soreceive = soreceive,
        .pru_preconnect = udp_preconnect,
        .pru_preattach = udp_preattach
};