pps_fetch: introduce a helper to handle timeouts

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

/*
 * Copyright (c) 2003, 2004 Jeffrey M. Hsu.  All rights reserved.
 * Copyright (c) 2003, 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, 1993
 *      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.
 *
 *      From: @(#)tcp_usrreq.c  8.2 (Berkeley) 1/3/94
 * $FreeBSD: src/sys/netinet/tcp_usrreq.c,v 1.51.2.17 2002/10/11 11:46:44 ume Exp $
 */

#include "opt_inet.h"
#include "opt_inet6.h"
#include "opt_tcpdebug.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/sysctl.h>
#include <sys/globaldata.h>
#include <sys/thread.h>

#include <sys/mbuf.h>
#ifdef INET6
#include <sys/domain.h>
#endif /* INET6 */
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/socketops.h>
#include <sys/protosw.h>
#include <sys/jail.h>

#include <sys/msgport2.h>
#include <sys/socketvar2.h>

#include <net/if.h>
#include <net/netisr.h>
#include <net/route.h>

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

#include <netinet/in.h>
#include <netinet/in_systm.h>
#ifdef INET6
#include <netinet/ip6.h>
#endif
#include <netinet/in_pcb.h>
#ifdef INET6
#include <netinet6/in6_pcb.h>
#endif
#include <netinet/in_var.h>
#include <netinet/ip_var.h>
#ifdef INET6
#include <netinet6/ip6_var.h>
#include <netinet6/tcp6_var.h>
#endif
#include <netinet/tcp.h>
#include <netinet/tcp_fsm.h>
#include <netinet/tcp_seq.h>
#include <netinet/tcp_timer.h>
#include <netinet/tcp_timer2.h>
#include <netinet/tcp_var.h>
#include <netinet/tcpip.h>
#ifdef TCPDEBUG
#include <netinet/tcp_debug.h>
#endif
#include <machine/limits.h>

/*
 * Limits for TCP_KEEP* options (we will adopt the same limits that linux
 * uses).
 */
#define MAXKEEPALIVE            32767
#define MAXKEEPCNT              127

/*
 * TCP protocol interface to socket abstraction.
 */
extern  char *tcpstates[];      /* XXX ??? */

static int      tcp_attach (struct socket *, struct pru_attach_info *);
static void     tcp_connect (netmsg_t msg);
#ifdef INET6
static void     tcp6_connect (netmsg_t msg);
static int      tcp6_connect_oncpu(struct tcpcb *tp, int flags,
                                struct mbuf **mp,
                                struct sockaddr_in6 *sin6,
                                struct in6_addr *addr6);
#endif /* INET6 */
static struct tcpcb *
                tcp_disconnect (struct tcpcb *);
static struct tcpcb *
                tcp_usrclosed (struct tcpcb *);

#ifdef TCPDEBUG
#define TCPDEBUG0       int ostate = 0
#define TCPDEBUG1()     ostate = tp ? tp->t_state : 0
#define TCPDEBUG2(req)  if (tp && (so->so_options & SO_DEBUG)) \
                                tcp_trace(TA_USER, ostate, tp, 0, 0, req)
#else
#define TCPDEBUG0
#define TCPDEBUG1()
#define TCPDEBUG2(req)
#endif

/*
 * For some ill optimized programs, which try to use TCP_NOPUSH
 * to improve performance, will have small amount of data sits
 * in the sending buffer.  These small amount of data will _not_
 * be pushed into the network until more data are written into
 * the socket or the socket write side is shutdown.
 */ 
static int      tcp_disable_nopush = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, disable_nopush, CTLFLAG_RW,
    &tcp_disable_nopush, 0, "TCP_NOPUSH socket option will have no effect");

/*
 * Allocate socket buffer space.
 */
static int
tcp_usr_preattach(struct socket *so, int proto __unused,
    struct pru_attach_info *ai)
{
        int error;

        if (so->so_snd.ssb_hiwat == 0 || so->so_rcv.ssb_hiwat == 0) {
                error = soreserve(so, tcp_sendspace, tcp_recvspace,
                                  ai->sb_rlimit);
                if (error)
                        return (error);
        }
        atomic_set_int(&so->so_rcv.ssb_flags, SSB_AUTOSIZE);
        atomic_set_int(&so->so_snd.ssb_flags, SSB_AUTOSIZE | SSB_PREALLOC);

        return 0;
}

/*
 * TCP attaches to socket via pru_attach(), reserving space,
 * and an internet control block.  This socket may move to
 * other CPU later when we bind/connect.
 */
static void
tcp_usr_attach(netmsg_t msg)
{
        struct socket *so = msg->base.nm_so;
        struct pru_attach_info *ai = msg->attach.nm_ai;
        int error;
        struct inpcb *inp;
        struct tcpcb *tp = NULL;
        TCPDEBUG0;

        inp = so->so_pcb;
        KASSERT(inp == NULL, ("tcp socket attached"));
        TCPDEBUG1();

        error = tcp_attach(so, ai);
        if (error)
                goto out;

        if ((so->so_options & SO_LINGER) && so->so_linger == 0)
                so->so_linger = TCP_LINGERTIME;
        tp = sototcpcb(so);
out:
        TCPDEBUG2(PRU_ATTACH);
        lwkt_replymsg(&msg->lmsg, error);
}

/*
 * pru_detach() detaches the TCP protocol from the socket.
 * If the protocol state is non-embryonic, then can't
 * do this directly: have to initiate a pru_disconnect(),
 * which may finish later; embryonic TCB's can just
 * be discarded here.
 */
static void
tcp_usr_detach(netmsg_t msg)
{
        struct socket *so = msg->base.nm_so;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;
        TCPDEBUG0;

        inp = so->so_pcb;

        /*
         * If the inp is already detached or never attached, it may have
         * been due to an async close or async attach failure.  Just return
         * as if no error occured.
         */
        if (inp) {
                tp = intotcpcb(inp);
                KASSERT(tp != NULL, ("tcp_usr_detach: tp is NULL"));
                TCPDEBUG1();
                tp = tcp_disconnect(tp);
                TCPDEBUG2(PRU_DETACH);
        }
        lwkt_replymsg(&msg->lmsg, error);
}

/*
 * NOTE: ignore_error is non-zero for certain disconnection races
 * which we want to silently allow, otherwise close() may return
 * an unexpected error.
 *
 * NOTE: The variables (msg) and (tp) are assumed.
 */
#define COMMON_START(so, inp, ignore_error)                     \
        TCPDEBUG0;                                              \
                                                                \
        inp = so->so_pcb;                                       \
        do {                                                    \
                if (inp == NULL) {                              \
                        error = ignore_error ? 0 : EINVAL;      \
                        tp = NULL;                              \
                        goto out;                               \
                }                                               \
                tp = intotcpcb(inp);                            \
                TCPDEBUG1();                                    \
        } while(0)

#define COMMON_END1(req, noreply)                               \
        out: do {                                               \
                TCPDEBUG2(req);                                 \
                if (!(noreply))                                 \
                        lwkt_replymsg(&msg->lmsg, error);       \
                return;                                         \
        } while(0)

#define COMMON_END(req)         COMMON_END1((req), 0)

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

/*
 * Give the socket an address.
 */
static void
tcp_usr_bind(netmsg_t msg)
{
        struct socket *so = msg->bind.base.nm_so;
        struct sockaddr *nam = msg->bind.nm_nam;
        struct thread *td = msg->bind.nm_td;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;
        struct sockaddr_in *sinp;
        lwkt_port_t port0 = netisr_cpuport(0);

        COMMON_START(so, inp, 0);

        /*
         * Must check for multicast addresses and disallow binding
         * to them.
         */
        sinp = (struct sockaddr_in *)nam;
        if (sinp->sin_family == AF_INET &&
            IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) {
                error = EAFNOSUPPORT;
                goto out;
        }

        /*
         * Check "already bound" here (in_pcbbind() does the same check
         * though), so we don't forward a connected socket to netisr0,
         * 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 out;
        }

        /*
         * Use netisr0 to serialize in_pcbbind(), so that pru_detach and
         * pru_bind for different sockets on the same local port could be
         * properly ordered.  The original race is illustrated here for
         * reference.
         *
         * s1 = socket();
         * bind(s1, *.PORT);
         * close(s1);  <----- asynchronous
         * s2 = socket();
         * bind(s2, *.PORT);
         *
         * All will expect bind(s2, *.PORT) to succeed.  However, it will
         * fail, if following sequence happens due to random socket initial
         * msgport and asynchronous close(2):
         *
         *    netisrN                  netisrM
         *       :                        :
         *       :                    pru_bind(s2) [*.PORT is used by s1]
         *  pru_detach(s1)                :
         */
        if (&curthread->td_msgport != port0) {
                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, &tcbinfo[mycpuid]);
                msg->bind.nm_flags |= PRUB_RELINK;

                TCP_STATE_MIGRATE_START(tp);

                /* See the related comment in tcp_connect() */
                lwkt_setmsg_receipt(lmsg, tcp_sosetport);
                lwkt_forwardmsg(port0, lmsg);
                /* msg invalid now */
                return;
        }
        KASSERT(so->so_port == port0, ("so_port is not netisr0"));

        if (msg->bind.nm_flags & PRUB_RELINK) {
                msg->bind.nm_flags &= ~PRUB_RELINK;
                TCP_STATE_MIGRATE_END(tp);
                in_pcblink(so->so_pcb, &tcbinfo[mycpuid]);
        }
        KASSERT(inp->inp_pcbinfo == &tcbinfo[0], ("pcbinfo is not tcbinfo0"));

        error = in_pcbbind(inp, nam, td);
        if (error)
                goto out;

        COMMON_END(PRU_BIND);
}

#ifdef INET6

static void
tcp6_usr_bind(netmsg_t msg)
{
        struct socket *so = msg->bind.base.nm_so;
        struct sockaddr *nam = msg->bind.nm_nam;
        struct thread *td = msg->bind.nm_td;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;
        struct sockaddr_in6 *sin6p;

        COMMON_START(so, inp, 0);

        /*
         * Must check for multicast addresses and disallow binding
         * to them.
         */
        sin6p = (struct sockaddr_in6 *)nam;
        if (sin6p->sin6_family == AF_INET6 &&
            IN6_IS_ADDR_MULTICAST(&sin6p->sin6_addr)) {
                error = EAFNOSUPPORT;
                goto out;
        }
        error = in6_pcbbind(inp, nam, td);
        if (error)
                goto out;
        COMMON_END(PRU_BIND);
}
#endif /* INET6 */

struct netmsg_inswildcard {
        struct netmsg_base      base;
        struct inpcb            *nm_inp;
};

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

        in_pcbinswildcardhash_oncpu(nm->nm_inp, &tcbinfo[cpu]);

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

/*
 * Prepare to accept connections.
 */
static void
tcp_usr_listen(netmsg_t msg)
{
        struct socket *so = msg->listen.base.nm_so;
        struct thread *td = msg->listen.nm_td;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;
        struct netmsg_inswildcard nm;
        lwkt_port_t port0 = netisr_cpuport(0);

        COMMON_START(so, inp, 0);

        if (&curthread->td_msgport != port0) {
                lwkt_msg_t lmsg = &msg->listen.base.lmsg;

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

                in_pcbunlink(so->so_pcb, &tcbinfo[mycpuid]);
                msg->listen.nm_flags |= PRUL_RELINK;

                TCP_STATE_MIGRATE_START(tp);

                /* See the related comment in tcp_connect() */
                lwkt_setmsg_receipt(lmsg, tcp_sosetport);
                lwkt_forwardmsg(port0, lmsg);
                /* msg invalid now */
                return;
        }
        KASSERT(so->so_port == port0, ("so_port is not netisr0"));

        if (msg->listen.nm_flags & PRUL_RELINK) {
                msg->listen.nm_flags &= ~PRUL_RELINK;
                TCP_STATE_MIGRATE_END(tp);
                in_pcblink(so->so_pcb, &tcbinfo[mycpuid]);
        }
        KASSERT(inp->inp_pcbinfo == &tcbinfo[0], ("pcbinfo is not tcbinfo0"));

        if (tp->t_flags & TF_LISTEN)
                goto out;

        if (inp->inp_lport == 0) {
                error = in_pcbbind(inp, NULL, td);
                if (error)
                        goto out;
        }

        TCP_STATE_CHANGE(tp, TCPS_LISTEN);
        tp->t_flags |= TF_LISTEN;
        tp->tt_msg = NULL; /* Catch any invalid timer usage */

        /*
         * Create tcpcb per-cpu port cache
         *
         * NOTE:
         * This _must_ be done before installing this inpcb into
         * wildcard hash.
         */
        tcp_pcbport_create(tp);

        if (netisr_ncpus > 1) {
                /*
                 * Put this inpcb into wildcard hash on other cpus.
                 */
                ASSERT_INP_NOTINHASH(inp);
                netmsg_init(&nm.base, NULL, &curthread->td_msgport,
                            MSGF_PRIORITY, in_pcbinswildcardhash_handler);
                nm.nm_inp = inp;
                lwkt_domsg(netisr_cpuport(1), &nm.base.lmsg, 0);
        }
        in_pcbinswildcardhash(inp);
        COMMON_END(PRU_LISTEN);
}

#ifdef INET6

static void
tcp6_usr_listen(netmsg_t msg)
{
        struct socket *so = msg->listen.base.nm_so;
        struct thread *td = msg->listen.nm_td;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;
        struct netmsg_inswildcard nm;

        COMMON_START(so, inp, 0);

        if (tp->t_flags & TF_LISTEN)
                goto out;

        if (inp->inp_lport == 0) {
                error = in6_pcbbind(inp, NULL, td);
                if (error)
                        goto out;
        }

        TCP_STATE_CHANGE(tp, TCPS_LISTEN);
        tp->t_flags |= TF_LISTEN;
        tp->tt_msg = NULL; /* Catch any invalid timer usage */

        /*
         * Create tcpcb per-cpu port cache
         *
         * NOTE:
         * This _must_ be done before installing this inpcb into
         * wildcard hash.
         */
        tcp_pcbport_create(tp);

        if (netisr_ncpus > 1) {
                /*
                 * Put this inpcb into wildcard hash on other cpus.
                 */
                KKASSERT(so->so_port == netisr_cpuport(0));
                ASSERT_NETISR0;
                KKASSERT(inp->inp_pcbinfo == &tcbinfo[0]);
                ASSERT_INP_NOTINHASH(inp);

                netmsg_init(&nm.base, NULL, &curthread->td_msgport,
                            MSGF_PRIORITY, in_pcbinswildcardhash_handler);
                nm.nm_inp = inp;
                lwkt_domsg(netisr_cpuport(1), &nm.base.lmsg, 0);
        }
        in_pcbinswildcardhash(inp);
        COMMON_END(PRU_LISTEN);
}
#endif /* INET6 */

/*
 * Initiate connection to peer.
 * Create a template for use in transmissions on this connection.
 * Enter SYN_SENT state, and mark socket as connecting.
 * Start keep-alive timer, and seed output sequence space.
 * Send initial segment on connection.
 */
static void
tcp_usr_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;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;
        struct sockaddr_in *sinp;

        ASSERT_NETISR_NCPUS(mycpuid);

        COMMON_START(so, inp, 0);

        /*
         * Must disallow TCP ``connections'' to multicast addresses.
         */
        sinp = (struct sockaddr_in *)nam;
        if (sinp->sin_family == AF_INET
            && IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) {
                error = EAFNOSUPPORT;
                goto out;
        }
        tcp_connect(msg);
        /* msg is invalid now */
        return;
out:
        if (msg->connect.nm_m) {
                m_freem(msg->connect.nm_m);
                msg->connect.nm_m = NULL;
        }
        if (msg->connect.nm_flags & PRUC_HELDTD)
                lwkt_rele(td);
        if (error && (msg->connect.nm_flags & PRUC_ASYNC)) {
                so->so_error = error;
                soisdisconnected(so);
        }
        lwkt_replymsg(&msg->lmsg, error);
}

#ifdef INET6

static void
tcp6_usr_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;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;
        struct sockaddr_in6 *sin6p;

        ASSERT_NETISR_NCPUS(mycpuid);

        COMMON_START(so, inp, 0);

        /*
         * Must disallow TCP ``connections'' to multicast addresses.
         */
        sin6p = (struct sockaddr_in6 *)nam;
        if (sin6p->sin6_family == AF_INET6
            && IN6_IS_ADDR_MULTICAST(&sin6p->sin6_addr)) {
                error = EAFNOSUPPORT;
                goto out;
        }

        if (!prison_remote_ip(td, nam)) {
                error = EAFNOSUPPORT;   /* Illegal jail IP */
                goto out;
        }

        /* Reject v4-mapped address */
        if (IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
                error = EADDRNOTAVAIL;
                goto out;
        }

        inp->inp_inc.inc_isipv6 = 1;
        tcp6_connect(msg);
        /* msg is invalid now */
        return;
out:
        if (msg->connect.nm_m) {
                m_freem(msg->connect.nm_m);
                msg->connect.nm_m = NULL;
        }
        lwkt_replymsg(&msg->lmsg, error);
}

#endif /* INET6 */

/*
 * Initiate disconnect from peer.
 * If connection never passed embryonic stage, just drop;
 * else if don't need to let data drain, then can just drop anyways,
 * else have to begin TCP shutdown process: mark socket disconnecting,
 * drain unread data, state switch to reflect user close, and
 * send segment (e.g. FIN) to peer.  Socket will be really disconnected
 * when peer sends FIN and acks ours.
 *
 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB.
 */
static void
tcp_usr_disconnect(netmsg_t msg)
{
        struct socket *so = msg->disconnect.base.nm_so;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;

        COMMON_START(so, inp, 1);
        tp = tcp_disconnect(tp);
        COMMON_END(PRU_DISCONNECT);
}

/*
 * Accept a connection.  Essentially all the work is
 * done at higher levels; just return the address
 * of the peer, storing through addr.
 */
static void
tcp_usr_accept(netmsg_t msg)
{
        struct socket *so = msg->accept.base.nm_so;
        struct sockaddr **nam = msg->accept.nm_nam;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp = NULL;
        TCPDEBUG0;

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

        tp = intotcpcb(inp);
        TCPDEBUG1();
        in_setpeeraddr(so, nam);
        COMMON_END(PRU_ACCEPT);
}

#ifdef INET6
static void
tcp6_usr_accept(netmsg_t msg)
{
        struct socket *so = msg->accept.base.nm_so;
        struct sockaddr **nam = msg->accept.nm_nam;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp = NULL;
        TCPDEBUG0;

        inp = so->so_pcb;

        if (so->so_state & SS_ISDISCONNECTED) {
                error = ECONNABORTED;
                goto out;
        }
        if (inp == NULL) {
                error = EINVAL;
                goto out;
        }
        tp = intotcpcb(inp);
        TCPDEBUG1();
        in6_setpeeraddr(so, nam);
        COMMON_END(PRU_ACCEPT);
}
#endif /* INET6 */

/*
 * Mark the connection as being incapable of further output.
 */
static void
tcp_usr_shutdown(netmsg_t msg)
{
        struct socket *so = msg->shutdown.base.nm_so;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;

        COMMON_START(so, inp, 0);
        socantsendmore(so);
        tp = tcp_usrclosed(tp);
        if (tp)
                error = tcp_output(tp);
        COMMON_END(PRU_SHUTDOWN);
}

/*
 * After a receive, possibly send window update to peer.
 */
static void
tcp_usr_rcvd(netmsg_t msg)
{
        struct socket *so = msg->rcvd.base.nm_so;
        int error = 0, noreply = 0;
        struct inpcb *inp;
        struct tcpcb *tp;

        COMMON_START(so, inp, 0);

        if (msg->rcvd.nm_pru_flags & PRUR_ASYNC) {
                noreply = 1;
                so_async_rcvd_reply(so);
        }
        tcp_output(tp);

        COMMON_END1(PRU_RCVD, noreply);
}

/*
 * Do a send by putting data in output queue and updating urgent
 * marker if URG set.  Possibly send more data.  Unlike the other
 * pru_*() routines, the mbuf chains are our responsibility.  We
 * must either enqueue them or free them.  The other pru_* routines
 * generally are caller-frees.
 */
static void
tcp_usr_send(netmsg_t msg)
{
        struct socket *so = msg->send.base.nm_so;
        int flags = msg->send.nm_flags;
        struct mbuf *m = msg->send.nm_m;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;
        TCPDEBUG0;

        KKASSERT(msg->send.nm_control == NULL);
        KKASSERT(msg->send.nm_addr == NULL);
        KKASSERT((flags & PRUS_FREEADDR) == 0);

        inp = so->so_pcb;

        if (inp == NULL) {
                /*
                 * OOPS! we lost a race, the TCP session got reset after
                 * we checked SS_CANTSENDMORE, eg: while doing uiomove or a
                 * network interrupt in the non-critical section of sosend().
                 */
                m_freem(m);
                error = ECONNRESET;     /* XXX EPIPE? */
                tp = NULL;
                TCPDEBUG1();
                goto out;
        }
        tp = intotcpcb(inp);
        TCPDEBUG1();

#ifdef foo
        /*
         * This is no longer necessary, since:
         * - sosendtcp() has already checked it for us
         * - It does not work with asynchronized send
         */

        /*
         * Don't let too much OOB data build up
         */
        if (flags & PRUS_OOB) {
                if (ssb_space(&so->so_snd) < -512) {
                        m_freem(m);
                        error = ENOBUFS;
                        goto out;
                }
        }
#endif

        /*
         * Pump the data into the socket.
         */
        if (m) {
                ssb_appendstream(&so->so_snd, m);
                sowwakeup(so);
        }
        if (flags & PRUS_OOB) {
                /*
                 * According to RFC961 (Assigned Protocols),
                 * the urgent pointer points to the last octet
                 * of urgent data.  We continue, however,
                 * to consider it to indicate the first octet
                 * of data past the urgent section.
                 * Otherwise, snd_up should be one lower.
                 */
                tp->snd_up = tp->snd_una + so->so_snd.ssb_cc;
                tp->t_flags |= TF_FORCE;
                error = tcp_output(tp);
                tp->t_flags &= ~TF_FORCE;
        } else {
                if (flags & PRUS_EOF) {
                        /*
                         * Close the send side of the connection after
                         * the data is sent.
                         */
                        socantsendmore(so);
                        tp = tcp_usrclosed(tp);
                }
                if (tp != NULL && !tcp_output_pending(tp)) {
                        if (flags & PRUS_MORETOCOME)
                                tp->t_flags |= TF_MORETOCOME;
                        error = tcp_output_fair(tp);
                        if (flags & PRUS_MORETOCOME)
                                tp->t_flags &= ~TF_MORETOCOME;
                }
        }
        COMMON_END1((flags & PRUS_OOB) ? PRU_SENDOOB :
                   ((flags & PRUS_EOF) ? PRU_SEND_EOF : PRU_SEND),
                   (flags & PRUS_NOREPLY));
}

/*
 * NOTE: (so) is referenced from soabort*() and netmsg_pru_abort()
 *       will sofree() it when we return.
 */
static void
tcp_usr_abort(netmsg_t msg)
{
        struct socket *so = msg->abort.base.nm_so;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;

        COMMON_START(so, inp, 1);
        tp = tcp_drop(tp, ECONNABORTED);
        COMMON_END(PRU_ABORT);
}

/*
 * Receive out-of-band data.
 */
static void
tcp_usr_rcvoob(netmsg_t msg)
{
        struct socket *so = msg->rcvoob.base.nm_so;
        struct mbuf *m = msg->rcvoob.nm_m;
        int flags = msg->rcvoob.nm_flags;
        int error = 0;
        struct inpcb *inp;
        struct tcpcb *tp;

        COMMON_START(so, inp, 0);
        if ((so->so_oobmark == 0 &&
             (so->so_state & SS_RCVATMARK) == 0) ||
            so->so_options & SO_OOBINLINE ||
            tp->t_oobflags & TCPOOB_HADDATA) {
                error = EINVAL;
                goto out;
        }
        if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) {
                error = EWOULDBLOCK;
                goto out;
        }
        m->m_len = 1;
        *mtod(m, caddr_t) = tp->t_iobc;
        if ((flags & MSG_PEEK) == 0)
                tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA);
        COMMON_END(PRU_RCVOOB);
}

static void
tcp_usr_savefaddr(struct socket *so, const struct sockaddr *faddr)
{
        in_savefaddr(so, faddr);
}

#ifdef INET6
static void
tcp6_usr_savefaddr(struct socket *so, const struct sockaddr *faddr)
{
        in6_savefaddr(so, faddr);
}
#endif

static int
tcp_usr_preconnect(struct socket *so, const struct sockaddr *nam,
    struct thread *td __unused)
{
        const struct sockaddr_in *sinp;

        sinp = (const struct sockaddr_in *)nam;
        if (sinp->sin_family == AF_INET &&
            IN_MULTICAST(ntohl(sinp->sin_addr.s_addr)))
                return EAFNOSUPPORT;

        soisconnecting(so);
        return 0;
}

/* xxx - should be const */
struct pr_usrreqs tcp_usrreqs = {
        .pru_abort = tcp_usr_abort,
        .pru_accept = tcp_usr_accept,
        .pru_attach = tcp_usr_attach,
        .pru_bind = tcp_usr_bind,
        .pru_connect = tcp_usr_connect,
        .pru_connect2 = pr_generic_notsupp,
        .pru_control = in_control_dispatch,
        .pru_detach = tcp_usr_detach,
        .pru_disconnect = tcp_usr_disconnect,
        .pru_listen = tcp_usr_listen,
        .pru_peeraddr = in_setpeeraddr_dispatch,
        .pru_rcvd = tcp_usr_rcvd,
        .pru_rcvoob = tcp_usr_rcvoob,
        .pru_send = tcp_usr_send,
        .pru_sense = pru_sense_null,
        .pru_shutdown = tcp_usr_shutdown,
        .pru_sockaddr = in_setsockaddr_dispatch,
        .pru_sosend = sosendtcp,
        .pru_soreceive = sorecvtcp,
        .pru_savefaddr = tcp_usr_savefaddr,
        .pru_preconnect = tcp_usr_preconnect,
        .pru_preattach = tcp_usr_preattach
};

#ifdef INET6
struct pr_usrreqs tcp6_usrreqs = {
        .pru_abort = tcp_usr_abort,
        .pru_accept = tcp6_usr_accept,
        .pru_attach = tcp_usr_attach,
        .pru_bind = tcp6_usr_bind,
        .pru_connect = tcp6_usr_connect,
        .pru_connect2 = pr_generic_notsupp,
        .pru_control = in6_control_dispatch,
        .pru_detach = tcp_usr_detach,
        .pru_disconnect = tcp_usr_disconnect,
        .pru_listen = tcp6_usr_listen,
        .pru_peeraddr = in6_setpeeraddr_dispatch,
        .pru_rcvd = tcp_usr_rcvd,
        .pru_rcvoob = tcp_usr_rcvoob,
        .pru_send = tcp_usr_send,
        .pru_sense = pru_sense_null,
        .pru_shutdown = tcp_usr_shutdown,
        .pru_sockaddr = in6_setsockaddr_dispatch,
        .pru_sosend = sosendtcp,
        .pru_soreceive = sorecvtcp,
        .pru_savefaddr = tcp6_usr_savefaddr
};
#endif /* INET6 */

static int
tcp_connect_oncpu(struct tcpcb *tp, int flags, struct mbuf *m,
                  const struct sockaddr_in *sin, struct sockaddr_in *if_sin,
                  uint16_t hash)
{
        struct inpcb *inp = tp->t_inpcb, *oinp;
        struct socket *so = inp->inp_socket;
        struct route *ro = &inp->inp_route;

        KASSERT(inp->inp_pcbinfo == &tcbinfo[mycpu->gd_cpuid],
            ("pcbinfo mismatch"));

        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, 0, NULL);
        if (oinp != NULL) {
                m_freem(m);
                return (EADDRINUSE);
        }
        if (inp->inp_laddr.s_addr == INADDR_ANY)
                inp->inp_laddr = if_sin->sin_addr;
        KASSERT(inp->inp_faddr.s_addr == sin->sin_addr.s_addr,
            ("faddr mismatch for reconnect"));
        KASSERT(inp->inp_fport == sin->sin_port,
            ("fport mismatch for reconnect"));
        in_pcbinsconnhash(inp);

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

        /*
         * We are now on the inpcb's owner CPU, if the cached route was
         * freed because the rtentry's owner CPU is not the current CPU
         * (e.g. in tcp_connect()), then we try to reallocate it here with
         * the hope that a rtentry may be cloned from a RTF_PRCLONING
         * rtentry.
         */
        if (!(inp->inp_socket->so_options & SO_DONTROUTE) && /*XXX*/
            ro->ro_rt == NULL) {
                bzero(&ro->ro_dst, sizeof(struct sockaddr_in));
                ro->ro_dst.sa_family = AF_INET;
                ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
                ((struct sockaddr_in *)&ro->ro_dst)->sin_addr =
                        sin->sin_addr;
                rtalloc(ro);
        }

        /*
         * Now that no more errors can occur, change the protocol processing
         * port to the current thread (which is the correct thread).
         *
         * Create TCP timer message now; we are on the tcpcb's owner
         * CPU/thread.
         */
        tcp_create_timermsg(tp, &curthread->td_msgport);

        /*
         * Compute window scaling to request.  Use a larger scaling then
         * needed for the initial receive buffer in case the receive buffer
         * gets expanded.
         */
        if (tp->request_r_scale < TCP_MIN_WINSHIFT)
                tp->request_r_scale = TCP_MIN_WINSHIFT;
        while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
               (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.ssb_hiwat
        ) {
                tp->request_r_scale++;
        }

        soisconnecting(so);
        tcpstat.tcps_connattempt++;
        TCP_STATE_CHANGE(tp, TCPS_SYN_SENT);
        tcp_callout_reset(tp, tp->tt_keep, tp->t_keepinit, tcp_timer_keep);
        tp->iss = tcp_new_isn(tp);
        tcp_sendseqinit(tp);
        if (m) {
                ssb_appendstream(&so->so_snd, m);
                m = NULL;
                if (flags & PRUS_OOB)
                        tp->snd_up = tp->snd_una + so->so_snd.ssb_cc;
        }

        /*
         * Close the send side of the connection after
         * the data is sent if flagged.
         */
        if ((flags & (PRUS_OOB|PRUS_EOF)) == PRUS_EOF) {
                socantsendmore(so);
                tp = tcp_usrclosed(tp);
        }
        return (tcp_output(tp));
}

/*
 * Common subroutine to open a TCP connection to remote host specified
 * by struct sockaddr_in in mbuf *nam.  Call in_pcbbind to assign a local
 * port number if needed.  Call in_pcbladdr to do the routing and to choose
 * a local host address (interface).
 * Initialize connection parameters and enter SYN-SENT state.
 */
static void
tcp_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 sockaddr_in *sin = (struct sockaddr_in *)nam;
        struct sockaddr_in *if_sin = NULL;
        struct inpcb *inp;
        struct tcpcb *tp;
        int error;
        uint16_t hash;
        lwkt_port_t port;

        COMMON_START(so, inp, 0);

        /*
         * Reconnect our pcb if we have to
         */
        if (msg->connect.nm_flags & PRUC_RECONNECT) {
                msg->connect.nm_flags &= ~PRUC_RECONNECT;
                TCP_STATE_MIGRATE_END(tp);
                in_pcblink(so->so_pcb, &tcbinfo[mycpu->gd_cpuid]);
        } else {
                if (inp->inp_faddr.s_addr != INADDR_ANY) {
                        error = EISCONN;
                        if (so->so_state & SS_ISCONNECTING)
                                error = EALREADY;
                        goto out;
                }
                KASSERT(inp->inp_fport == 0, ("invalid fport"));
        }

        /*
         * Select local port, if it is not yet selected.
         */
        if (inp->inp_lport == 0) {
                KKASSERT(inp->inp_laddr.s_addr == INADDR_ANY);

                error = in_pcbladdr(inp, nam, &if_sin, td);
                if (error)
                        goto out;
                inp->inp_laddr.s_addr = if_sin->sin_addr.s_addr;
                msg->connect.nm_flags |= PRUC_HASLADDR;

                /*
                 * Install faddr/fport earlier, so that when this
                 * inpcb is installed on to the lport hash, the
                 * 4-tuple contains correct value.
                 *
                 * NOTE: The faddr/fport will have to be installed
                 * after the in_pcbladdr(), which may change them.
                 */
                inp->inp_faddr = sin->sin_addr;
                inp->inp_fport = sin->sin_port;

                error = in_pcbbind_remote(inp, nam, td);
                if (error)
                        goto out;
        }

        if ((msg->connect.nm_flags & PRUC_HASLADDR) == 0) {
                /*
                 * Rarely used path:
                 * This inpcb was bound before this connect.
                 */
                error = in_pcbladdr(inp, nam, &if_sin, td);
                if (error)
                        goto out;

                /*
                 * Save or refresh the faddr/fport, since they may
                 * be changed by in_pcbladdr().
                 */
                inp->inp_faddr = sin->sin_addr;
                inp->inp_fport = sin->sin_port;
        }
#ifdef INVARIANTS
        else {
                KASSERT(inp->inp_faddr.s_addr == sin->sin_addr.s_addr,
                    ("faddr mismatch for reconnect"));
                KASSERT(inp->inp_fport == sin->sin_port,
                    ("fport mismatch for reconnect"));
        }
#endif
        KKASSERT(inp->inp_socket == so);

        hash = tcp_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;

                /*
                 * 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);

                /*
                 * 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(so->so_pcb, &tcbinfo[mycpu->gd_cpuid]);
                msg->connect.nm_flags |= PRUC_RECONNECT;
                msg->connect.base.nm_dispatch = tcp_connect;

                TCP_STATE_MIGRATE_START(tp);

                /*
                 * Use message put done receipt to change this socket's
                 * so_port, i.e. _after_ this message was put onto the
                 * target netisr's msgport but _before_ the message could
                 * be pulled from the target netisr's msgport, so that:
                 * - The upper half (socket code) will not see the new
                 *   msgport before this message reaches the new msgport
                 *   and messages for this socket will be ordered.
                 * - This message will see the new msgport, when its
                 *   handler is called in the target netisr.
                 *
                 * NOTE:
                 * We MUST use messege put done receipt to change this
                 * socket's so_port:
                 * If we changed the so_port in this netisr after the
                 * lwkt_forwardmsg (so messages for this socket will be
                 * ordered) and changed the so_port in the target netisr
                 * at the very beginning of this message's handler, we
                 * would suffer so_port overwritten race, given this
                 * message might be forwarded again.
                 *
                 * NOTE:
                 * This mechanism depends on that the netisr's msgport
                 * is spin msgport (currently it is :).
                 *
                 * If the upper half saw the new msgport before this
                 * message reached the target netisr's msgport, the
                 * messages sent from the upper half could reach the new
                 * msgport before this message, thus there would be
                 * message reordering.  The worst case could be soclose()
                 * saw the new msgport and the detach message could reach
                 * the new msgport before this message, i.e. the inpcb
                 * could have been destroyed when this message was still
                 * pending on or on its way to the new msgport.  Other
                 * weird cases could also happen, e.g. inpcb->inp_pcbinfo,
                 * since we have unlinked this inpcb from the current
                 * pcbinfo first.
                 */
                lwkt_setmsg_receipt(lmsg, tcp_sosetport);
                lwkt_forwardmsg(port, lmsg);
                /* msg invalid now */
                return;
        } else if (msg->connect.nm_flags & PRUC_HELDTD) {
                /*
                 * The original thread is no longer needed; release it.
                 */
                lwkt_rele(td);
                msg->connect.nm_flags &= ~PRUC_HELDTD;
        }
        error = tcp_connect_oncpu(tp, msg->connect.nm_sndflags,
                                  msg->connect.nm_m, sin, if_sin, hash);
        msg->connect.nm_m = NULL;
out:
        if (msg->connect.nm_m) {
                m_freem(msg->connect.nm_m);
                msg->connect.nm_m = NULL;
        }
        if (msg->connect.nm_flags & PRUC_HELDTD)
                lwkt_rele(td);
        if (error && (msg->connect.nm_flags & PRUC_ASYNC)) {
                so->so_error = error;
                soisdisconnected(so);
        }
        lwkt_replymsg(&msg->connect.base.lmsg, error);
        /* msg invalid now */
}

#ifdef INET6

static void
tcp6_connect(netmsg_t msg)
{
        struct tcpcb *tp;
        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_in6 *sin6 = (struct sockaddr_in6 *)nam;
        struct in6_addr *addr6;
        lwkt_port_t port;
        int error;

        COMMON_START(so, inp, 0);

        /*
         * Reconnect our pcb if we have to
         */
        if (msg->connect.nm_flags & PRUC_RECONNECT) {
                msg->connect.nm_flags &= ~PRUC_RECONNECT;
                TCP_STATE_MIGRATE_END(tp);
                in_pcblink(so->so_pcb, &tcbinfo[mycpu->gd_cpuid]);
        }

        /*
         * Bind if we have to
         */
        if (inp->inp_lport == 0) {
                error = in6_pcbbind(inp, NULL, td);
                if (error)
                        goto out;
        }

        /*
         * Cannot simply call in_pcbconnect, because there might be an
         * earlier incarnation of this same connection still in
         * TIME_WAIT state, creating an ADDRINUSE error.
         */
        error = in6_pcbladdr(inp, nam, &addr6, td);
        if (error)
                goto out;

        port = tcp6_addrport(); /* XXX hack for now, always cpu0 */

        if (port != &curthread->td_msgport) {
                lwkt_msg_t lmsg = &msg->connect.base.lmsg;

                /*
                 * 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);

                in_pcbunlink(so->so_pcb, &tcbinfo[mycpu->gd_cpuid]);
                msg->connect.nm_flags |= PRUC_RECONNECT;
                msg->connect.base.nm_dispatch = tcp6_connect;

                TCP_STATE_MIGRATE_START(tp);

                /* See the related comment in tcp_connect() */
                lwkt_setmsg_receipt(lmsg, tcp_sosetport);
                lwkt_forwardmsg(port, lmsg);
                /* msg invalid now */
                return;
        }
        error = tcp6_connect_oncpu(tp, msg->connect.nm_sndflags,
                                   &msg->connect.nm_m, sin6, addr6);
        /* nm_m may still be intact */
out:
        if (msg->connect.nm_m) {
                m_freem(msg->connect.nm_m);
                msg->connect.nm_m = NULL;
        }
        lwkt_replymsg(&msg->connect.base.lmsg, error);
        /* msg invalid now */
}

static int
tcp6_connect_oncpu(struct tcpcb *tp, int flags, struct mbuf **mp,
                   struct sockaddr_in6 *sin6, struct in6_addr *addr6)
{
        struct mbuf *m = *mp;
        struct inpcb *inp = tp->t_inpcb;
        struct socket *so = inp->inp_socket;
        struct inpcb *oinp;

        /*
         * Cannot simply call in_pcbconnect, because there might be an
         * earlier incarnation of this same connection still in
         * TIME_WAIT state, creating an ADDRINUSE error.
         */
        oinp = in6_pcblookup_hash(inp->inp_pcbinfo,
                                  &sin6->sin6_addr, sin6->sin6_port,
                                  (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) ?
                                      addr6 : &inp->in6p_laddr),
                                  inp->inp_lport,  0, NULL);
        if (oinp)
                return (EADDRINUSE);

        if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
                inp->in6p_laddr = *addr6;
        inp->in6p_faddr = sin6->sin6_addr;
        inp->inp_fport = sin6->sin6_port;
        if ((sin6->sin6_flowinfo & IPV6_FLOWINFO_MASK) != 0)
                inp->in6p_flowinfo = sin6->sin6_flowinfo;
        in_pcbinsconnhash(inp);

        /*
         * Now that no more errors can occur, change the protocol processing
         * port to the current thread (which is the correct thread).
         *
         * Create TCP timer message now; we are on the tcpcb's owner
         * CPU/thread.
         */
        tcp_create_timermsg(tp, &curthread->td_msgport);

        /* Compute window scaling to request.  */
        if (tp->request_r_scale < TCP_MIN_WINSHIFT)
                tp->request_r_scale = TCP_MIN_WINSHIFT;
        while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
            (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.ssb_hiwat) {
                tp->request_r_scale++;
        }

        soisconnecting(so);
        tcpstat.tcps_connattempt++;
        TCP_STATE_CHANGE(tp, TCPS_SYN_SENT);
        tcp_callout_reset(tp, tp->tt_keep, tp->t_keepinit, tcp_timer_keep);
        tp->iss = tcp_new_isn(tp);
        tcp_sendseqinit(tp);
        if (m) {
                ssb_appendstream(&so->so_snd, m);
                *mp = NULL;
                if (flags & PRUS_OOB)
                        tp->snd_up = tp->snd_una + so->so_snd.ssb_cc;
        }

        /*
         * Close the send side of the connection after
         * the data is sent if flagged.
         */
        if ((flags & (PRUS_OOB|PRUS_EOF)) == PRUS_EOF) {
                socantsendmore(so);
                tp = tcp_usrclosed(tp);
        }
        return (tcp_output(tp));
}

#endif /* INET6 */

/*
 * The new sockopt interface makes it possible for us to block in the
 * copyin/out step (if we take a page fault).  Taking a page fault while
 * in a critical section is probably a Bad Thing.  (Since sockets and pcbs
 * both now use TSM, there probably isn't any need for this function to 
 * run in a critical section any more.  This needs more examination.)
 */
void
tcp_ctloutput(netmsg_t msg)
{
        struct socket *so = msg->base.nm_so;
        struct sockopt *sopt = msg->ctloutput.nm_sopt;
        struct thread *td = NULL;
        int     error, opt, optval, opthz;
        struct  inpcb *inp;
        struct  tcpcb *tp;

        if (msg->ctloutput.nm_flags & PRCO_HELDTD)
                td = sopt->sopt_td;

        error = 0;
        inp = so->so_pcb;
        if (inp == NULL) {
                error = ECONNRESET;
                goto done;
        }
        tp = intotcpcb(inp);

        /* Get socket's owner cpuid hint */
        if (sopt->sopt_level == SOL_SOCKET &&
            sopt->sopt_dir == SOPT_GET &&
            sopt->sopt_name == SO_CPUHINT) {
                if (tp->t_flags & TF_LISTEN) {
                        /*
                         * Listen sockets owner cpuid is always 0,
                         * which does not make sense if SO_REUSEPORT
                         * is not set.
                         *
                         * NOTE: inp_lgrpindex is _not_ assigned in jail.
                         */
                        if ((so->so_options & SO_REUSEPORT) &&
                            inp->inp_lgrpindex >= 0)
                                optval = inp->inp_lgrpindex % netisr_ncpus;
                        else
                                optval = -1; /* no hint */
                } else {
                        optval = mycpuid;
                }
                soopt_from_kbuf(sopt, &optval, sizeof(optval));
                goto done;
        }

        if (sopt->sopt_level != IPPROTO_TCP) {
                if (sopt->sopt_level == IPPROTO_IP) {
                        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:
                                /*
                                 * Multicast does not make sense on
                                 * TCP sockets.
                                 */
                                error = EOPNOTSUPP;
                                goto done;
                        }
                }
#ifdef INET6
                if (INP_CHECK_SOCKAF(so, AF_INET6))
                        ip6_ctloutput_dispatch(msg);
                else
#endif /* INET6 */
                ip_ctloutput(msg);
                /* msg invalid now */
                if (td != NULL)
                        lwkt_rele(td);
                return;
        }

        switch (sopt->sopt_dir) {
        case SOPT_SET:
                error = soopt_to_kbuf(sopt, &optval, sizeof optval,
                                      sizeof optval);
                if (error)
                        break;
                switch (sopt->sopt_name) {
                case TCP_FASTKEEP:
                        if (optval > 0)
                                tp->t_keepidle = tp->t_keepintvl;
                        else
                                tp->t_keepidle = tcp_keepidle;
                        tcp_timer_keep_activity(tp, 0);
                        break;
#ifdef TCP_SIGNATURE
                case TCP_SIGNATURE_ENABLE:
                        if (tp->t_state == TCPS_CLOSED) {
                                /*
                                 * This is the only safe state that this
                                 * option could be changed.  Some segments
                                 * could already have been sent in other
                                 * states.
                                 */
                                if (optval > 0)
                                        tp->t_flags |= TF_SIGNATURE;
                                else
                                        tp->t_flags &= ~TF_SIGNATURE;
                        } else {
                                error = EOPNOTSUPP;
                        }
                        break;
#endif /* TCP_SIGNATURE */
                case TCP_NODELAY:
                case TCP_NOOPT:
                        switch (sopt->sopt_name) {
                        case TCP_NODELAY:
                                opt = TF_NODELAY;
                                break;
                        case TCP_NOOPT:
                                opt = TF_NOOPT;
                                break;
                        default:
                                opt = 0; /* dead code to fool gcc */
                                break;
                        }

                        if (optval)
                                tp->t_flags |= opt;
                        else
                                tp->t_flags &= ~opt;
                        break;

                case TCP_NOPUSH:
                        if (tcp_disable_nopush)
                                break;
                        if (optval)
                                tp->t_flags |= TF_NOPUSH;
                        else {
                                tp->t_flags &= ~TF_NOPUSH;
                                error = tcp_output(tp);
                        }
                        break;

                case TCP_MAXSEG:
                        /*
                         * Must be between 0 and maxseg.  If the requested
                         * maxseg is too small to satisfy the desired minmss,
                         * pump it up (silently so sysctl modifications of
                         * minmss do not create unexpected program failures).
                         * Handle degenerate cases.
                         */
                        if (optval > 0 && optval <= tp->t_maxseg) {
                                if (optval + 40 < tcp_minmss) {
                                        optval = tcp_minmss - 40;
                                        if (optval < 0)
                                                optval = 1;
                                }
                                tp->t_maxseg = optval;
                        } else {
                                error = EINVAL;
                        }
                        break;

                case TCP_KEEPINIT:
                case TCP_KEEPIDLE:
                case TCP_KEEPINTVL:
                        if (optval < 1 || optval > MAXKEEPALIVE) {
                                error = EINVAL;
                                break;
                        }
                        opthz = optval * hz;

                        switch (sopt->sopt_name) {
                        case TCP_KEEPINIT:
                                tp->t_keepinit = opthz;
                                break;
                        case TCP_KEEPIDLE:
                                tp->t_keepidle = opthz;
                                tcp_timer_keep_activity(tp, 0);
                                break;
                        case TCP_KEEPINTVL:
                                tp->t_keepintvl = opthz;
                                tp->t_maxidle = tp->t_keepintvl * tp->t_keepcnt;
                                break;
                        }
                        break;

                case TCP_KEEPCNT:
                        if (optval < 1 || optval > MAXKEEPCNT) {
                                error = EINVAL;
                                break;
                        }
                        tp->t_keepcnt = optval;
                        tp->t_maxidle = tp->t_keepintvl * tp->t_keepcnt;
                        break;

                default:
                        error = ENOPROTOOPT;
                        break;
                }
                break;

        case SOPT_GET:
                switch (sopt->sopt_name) {
#ifdef TCP_SIGNATURE
                case TCP_SIGNATURE_ENABLE:
                        optval = (tp->t_flags & TF_SIGNATURE) ? 1 : 0;
                        break;
#endif /* TCP_SIGNATURE */
                case TCP_NODELAY:
                        optval = tp->t_flags & TF_NODELAY;
                        break;
                case TCP_MAXSEG:
                        optval = tp->t_maxseg;
                        break;
                case TCP_NOOPT:
                        optval = tp->t_flags & TF_NOOPT;
                        break;
                case TCP_NOPUSH:
                        optval = tp->t_flags & TF_NOPUSH;
                        break;
                case TCP_KEEPINIT:
                        optval = tp->t_keepinit / hz;
                        break;
                case TCP_KEEPIDLE:
                        optval = tp->t_keepidle / hz;
                        break;
                case TCP_KEEPINTVL:
                        optval = tp->t_keepintvl / hz;
                        break;
                case TCP_KEEPCNT:
                        optval = tp->t_keepcnt;
                        break;
                default:
                        error = ENOPROTOOPT;
                        break;
                }
                if (error == 0)
                        soopt_from_kbuf(sopt, &optval, sizeof optval);
                break;
        }
done:
        if (td != NULL)
                lwkt_rele(td);
        lwkt_replymsg(&msg->lmsg, error);
}

struct netmsg_tcp_ctloutput {
        struct netmsg_pr_ctloutput ctloutput;
        struct sockopt          sopt;
        int                     sopt_val;
};

/*
 * Allocate netmsg_pr_ctloutput for asynchronous tcp_ctloutput.
 */
struct netmsg_pr_ctloutput *
tcp_ctloutmsg(struct sockopt *sopt)
{
        struct netmsg_tcp_ctloutput *msg;
        int flags = 0, error;

        KASSERT(sopt->sopt_dir == SOPT_SET, ("not from ctloutput"));

        /* Only small set of options allows asynchronous setting. */
        if (sopt->sopt_level != IPPROTO_TCP)
                return NULL;
        switch (sopt->sopt_name) {
        case TCP_NODELAY:
        case TCP_NOOPT:
        case TCP_NOPUSH:
        case TCP_FASTKEEP:
                break;
        default:
                return NULL;
        }

        msg = kmalloc(sizeof(*msg), M_LWKTMSG, M_WAITOK | M_NULLOK);
        if (msg == NULL) {
                /* Fallback to synchronous tcp_ctloutput */
                return NULL;
        }

        /* Save the sockopt */
        msg->sopt = *sopt;

        /* Fixup the sopt.sopt_val ptr */
        error = sooptcopyin(sopt, &msg->sopt_val,
            sizeof(msg->sopt_val), sizeof(msg->sopt_val));
        if (error) {
                kfree(msg, M_LWKTMSG);
                return NULL;
        }
        msg->sopt.sopt_val = &msg->sopt_val;

        /* Hold the current thread */
        if (msg->sopt.sopt_td != NULL) {
                flags |= PRCO_HELDTD;
                lwkt_hold(msg->sopt.sopt_td);
        }

        msg->ctloutput.nm_flags = flags;
        msg->ctloutput.nm_sopt = &msg->sopt;

        return &msg->ctloutput;
}

/*
 * tcp_sendspace and tcp_recvspace are the default send and receive window
 * sizes, respectively.  These are obsolescent (this information should
 * be set by the route).
 *
 * Use a default that does not require tcp window scaling to be turned
 * on.  Individual programs or the administrator can increase the default.
 */
u_long  tcp_sendspace = 57344;  /* largest multiple of PAGE_SIZE < 64k */
SYSCTL_INT(_net_inet_tcp, TCPCTL_SENDSPACE, sendspace, CTLFLAG_RW,
    &tcp_sendspace , 0, "Maximum outgoing TCP datagram size");
u_long  tcp_recvspace = 57344;  /* largest multiple of PAGE_SIZE < 64k */
SYSCTL_INT(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
    &tcp_recvspace , 0, "Maximum incoming TCP datagram size");

/*
 * Attach TCP protocol to socket, allocating internet protocol control
 * block, tcp control block, buffer space, and entering CLOSED state.
 */
static int
tcp_attach(struct socket *so, struct pru_attach_info *ai)
{
        struct inpcb *inp;
        int error;
        int cpu;
#ifdef INET6
        boolean_t isipv6 = INP_CHECK_SOCKAF(so, AF_INET6);
#endif

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

        cpu = mycpu->gd_cpuid;

        /*
         * Set the default pcbinfo.  This will likely change when we
         * bind/connect.
         */
        error = in_pcballoc(so, &tcbinfo[cpu]);
        if (error)
                return (error);
        inp = so->so_pcb;
#ifdef INET6
        if (isipv6)
                inp->in6p_hops = -1;    /* use kernel default */
#endif
        tcp_newtcpcb(inp);
        /* Keep a reference for asynchronized pru_rcvd */
        soreference(so);
        return (0);
}

/*
 * Initiate (or continue) disconnect.
 * If embryonic state, just send reset (once).
 * If in ``let data drain'' option and linger null, just drop.
 * Otherwise (hard), mark socket disconnecting and drop
 * current input data; switch states based on user close, and
 * send segment to peer (with FIN).
 */
static struct tcpcb *
tcp_disconnect(struct tcpcb *tp)
{
        struct socket *so = tp->t_inpcb->inp_socket;

        if (tp->t_state < TCPS_ESTABLISHED) {
                tp = tcp_close(tp);
        } else if ((so->so_options & SO_LINGER) && so->so_linger == 0) {
                tp = tcp_drop(tp, 0);
        } else {
                lwkt_gettoken(&so->so_rcv.ssb_token);
                soisdisconnecting(so);
                sbflush(&so->so_rcv.sb);
                tp = tcp_usrclosed(tp);
                if (tp)
                        tcp_output(tp);
                lwkt_reltoken(&so->so_rcv.ssb_token);
        }
        return (tp);
}

/*
 * User issued close, and wish to trail through shutdown states:
 * if never received SYN, just forget it.  If got a SYN from peer,
 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN.
 * If already got a FIN from peer, then almost done; go to LAST_ACK
 * state.  In all other cases, have already sent FIN to peer (e.g.
 * after PRU_SHUTDOWN), and just have to play tedious game waiting
 * for peer to send FIN or not respond to keep-alives, etc.
 * We can let the user exit from the close as soon as the FIN is acked.
 */
static struct tcpcb *
tcp_usrclosed(struct tcpcb *tp)
{

        switch (tp->t_state) {

        case TCPS_CLOSED:
        case TCPS_LISTEN:
                TCP_STATE_CHANGE(tp, TCPS_CLOSED);
                tp = tcp_close(tp);
                break;

        case TCPS_SYN_SENT:
        case TCPS_SYN_RECEIVED:
                tp->t_flags |= TF_NEEDFIN;
                break;

        case TCPS_ESTABLISHED:
                TCP_STATE_CHANGE(tp, TCPS_FIN_WAIT_1);
                break;

        case TCPS_CLOSE_WAIT:
                TCP_STATE_CHANGE(tp, TCPS_LAST_ACK);
                break;
        }
        if (tp && tp->t_state >= TCPS_FIN_WAIT_2) {
                soisdisconnected(tp->t_inpcb->inp_socket);
                /* To prevent the connection hanging in FIN_WAIT_2 forever. */
                if (tp->t_state == TCPS_FIN_WAIT_2) {
                        tcp_callout_reset(tp, tp->tt_2msl, tp->t_maxidle,
                            tcp_timer_2msl);
                }
        }
        return (tp);
}