* Add a flag to track an in-transit socket abort to avoid races when closing

[dragonfly.git] / sys / kern / uipc_socket.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
 *      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. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *      This product includes software developed by the University of
 *      California, Berkeley and its contributors.
 * 4. 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.
 *
 *      @(#)uipc_socket.c       8.3 (Berkeley) 4/15/94
 * $FreeBSD: src/sys/kern/uipc_socket.c,v 1.68.2.24 2003/11/11 17:18:18 silby Exp $
 * $DragonFly: src/sys/kern/uipc_socket.c,v 1.54 2008/08/28 23:15:43 dillon Exp $
 */

#include "opt_inet.h"
#include "opt_sctp.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/fcntl.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/domain.h>
#include <sys/file.h>                   /* for struct knote */
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/event.h>
#include <sys/poll.h>
#include <sys/proc.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/socketops.h>
#include <sys/resourcevar.h>
#include <sys/signalvar.h>
#include <sys/sysctl.h>
#include <sys/uio.h>
#include <sys/jail.h>
#include <vm/vm_zone.h>
#include <vm/pmap.h>

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

#include <machine/limits.h>

#ifdef INET
static int       do_setopt_accept_filter(struct socket *so, struct sockopt *sopt);
#endif /* INET */

static void     filt_sordetach(struct knote *kn);
static int      filt_soread(struct knote *kn, long hint);
static void     filt_sowdetach(struct knote *kn);
static int      filt_sowrite(struct knote *kn, long hint);
static int      filt_solisten(struct knote *kn, long hint);

static struct filterops solisten_filtops = 
        { 1, NULL, filt_sordetach, filt_solisten };
static struct filterops soread_filtops =
        { 1, NULL, filt_sordetach, filt_soread };
static struct filterops sowrite_filtops = 
        { 1, NULL, filt_sowdetach, filt_sowrite };

MALLOC_DEFINE(M_SOCKET, "socket", "socket struct");
MALLOC_DEFINE(M_SONAME, "soname", "socket name");
MALLOC_DEFINE(M_PCB, "pcb", "protocol control block");


static int somaxconn = SOMAXCONN;
SYSCTL_INT(_kern_ipc, KIPC_SOMAXCONN, somaxconn, CTLFLAG_RW,
    &somaxconn, 0, "Maximum pending socket connection queue size");

/*
 * Socket operation routines.
 * These routines are called by the routines in
 * sys_socket.c or from a system process, and
 * implement the semantics of socket operations by
 * switching out to the protocol specific routines.
 */

/*
 * Get a socket structure, and initialize it.
 * Note that it would probably be better to allocate socket
 * and PCB at the same time, but I'm not convinced that all
 * the protocols can be easily modified to do this.
 */
struct socket *
soalloc(int waitok)
{
        struct socket *so;
        unsigned waitmask;

        waitmask = waitok ? M_WAITOK : M_NOWAIT;
        so = kmalloc(sizeof(struct socket), M_SOCKET, M_ZERO|waitmask);
        if (so) {
                /* XXX race condition for reentrant kernel */
                TAILQ_INIT(&so->so_aiojobq);
                TAILQ_INIT(&so->so_rcv.ssb_sel.si_mlist);
                TAILQ_INIT(&so->so_snd.ssb_sel.si_mlist);
        }
        return so;
}

int
socreate(int dom, struct socket **aso, int type,
        int proto, struct thread *td)
{
        struct proc *p = td->td_proc;
        struct protosw *prp;
        struct socket *so;
        struct pru_attach_info ai;
        int error;

        if (proto)
                prp = pffindproto(dom, proto, type);
        else
                prp = pffindtype(dom, type);

        if (prp == 0 || prp->pr_usrreqs->pru_attach == 0)
                return (EPROTONOSUPPORT);

        if (p->p_ucred->cr_prison && jail_socket_unixiproute_only &&
            prp->pr_domain->dom_family != PF_LOCAL &&
            prp->pr_domain->dom_family != PF_INET &&
            prp->pr_domain->dom_family != PF_INET6 &&
            prp->pr_domain->dom_family != PF_ROUTE) {
                return (EPROTONOSUPPORT);
        }

        if (prp->pr_type != type)
                return (EPROTOTYPE);
        so = soalloc(p != 0);
        if (so == 0)
                return (ENOBUFS);

        TAILQ_INIT(&so->so_incomp);
        TAILQ_INIT(&so->so_comp);
        so->so_type = type;
        so->so_cred = crhold(p->p_ucred);
        so->so_proto = prp;
        ai.sb_rlimit = &p->p_rlimit[RLIMIT_SBSIZE];
        ai.p_ucred = p->p_ucred;
        ai.fd_rdir = p->p_fd->fd_rdir;
        error = so_pru_attach(so, proto, &ai);
        if (error) {
                so->so_state |= SS_NOFDREF;
                sofree(so);
                return (error);
        }
        *aso = so;
        return (0);
}

int
sobind(struct socket *so, struct sockaddr *nam, struct thread *td)
{
        int error;

        crit_enter();
        error = so_pru_bind(so, nam, td);
        crit_exit();
        return (error);
}

void
sodealloc(struct socket *so)
{
        if (so->so_rcv.ssb_hiwat)
                (void)chgsbsize(so->so_cred->cr_uidinfo,
                    &so->so_rcv.ssb_hiwat, 0, RLIM_INFINITY);
        if (so->so_snd.ssb_hiwat)
                (void)chgsbsize(so->so_cred->cr_uidinfo,
                    &so->so_snd.ssb_hiwat, 0, RLIM_INFINITY);
#ifdef INET
        /* remove accept filter if present */
        if (so->so_accf != NULL)
                do_setopt_accept_filter(so, NULL);
#endif /* INET */
        crfree(so->so_cred);
        kfree(so, M_SOCKET);
}

int
solisten(struct socket *so, int backlog, struct thread *td)
{
        int error;
#ifdef SCTP
        short oldopt, oldqlimit;
#endif /* SCTP */

        crit_enter();
        if (so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING)) {
                crit_exit();
                return (EINVAL);
        }

#ifdef SCTP
        oldopt = so->so_options;
        oldqlimit = so->so_qlimit;
#endif /* SCTP */

        if (TAILQ_EMPTY(&so->so_comp))
                so->so_options |= SO_ACCEPTCONN;
        if (backlog < 0 || backlog > somaxconn)
                backlog = somaxconn;
        so->so_qlimit = backlog;
        /* SCTP needs to look at tweak both the inbound backlog parameter AND
         * the so_options (UDP model both connect's and gets inbound
         * connections .. implicitly).
         */
        error = so_pru_listen(so, td);
        if (error) {
#ifdef SCTP
                /* Restore the params */
                so->so_options = oldopt;
                so->so_qlimit = oldqlimit;
#endif /* SCTP */
                crit_exit();
                return (error);
        }
        crit_exit();
        return (0);
}

/*
 * Destroy a disconnected socket.  This routine is a NOP if entities
 * still have a reference on the socket:
 *
 *      so_pcb -        The protocol stack still has a reference
 *      SS_NOFDREF -    There is no longer a file pointer reference
 *      SS_ABORTING -   An abort netmsg is in-flight
 */
void
sofree(struct socket *so)
{
        struct socket *head = so->so_head;

        if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0)
                return;
        if (so->so_state & SS_ABORTING)
                return;
        if (head != NULL) {
                if (so->so_state & SS_INCOMP) {
                        TAILQ_REMOVE(&head->so_incomp, so, so_list);
                        head->so_incqlen--;
                } else if (so->so_state & SS_COMP) {
                        /*
                         * We must not decommission a socket that's
                         * on the accept(2) queue.  If we do, then
                         * accept(2) may hang after select(2) indicated
                         * that the listening socket was ready.
                         */
                        return;
                } else {
                        panic("sofree: not queued");
                }
                so->so_state &= ~SS_INCOMP;
                so->so_head = NULL;
        }
        ssb_release(&so->so_snd, so);
        sorflush(so);
        sodealloc(so);
}

/*
 * Close a socket on last file table reference removal.
 * Initiate disconnect if connected.
 * Free socket when disconnect complete.
 */
int
soclose(struct socket *so, int fflag)
{
        int error = 0;

        crit_enter();
        funsetown(so->so_sigio);
        if (so->so_pcb == NULL)
                goto discard;
        if (so->so_state & SS_ISCONNECTED) {
                if ((so->so_state & SS_ISDISCONNECTING) == 0) {
                        error = sodisconnect(so);
                        if (error)
                                goto drop;
                }
                if (so->so_options & SO_LINGER) {
                        if ((so->so_state & SS_ISDISCONNECTING) &&
                            (fflag & FNONBLOCK))
                                goto drop;
                        while (so->so_state & SS_ISCONNECTED) {
                                error = tsleep((caddr_t)&so->so_timeo,
                                    PCATCH, "soclos", so->so_linger * hz);
                                if (error)
                                        break;
                        }
                }
        }
drop:
        if (so->so_pcb) {
                int error2;

                error2 = so_pru_detach(so);
                if (error == 0)
                        error = error2;
        }
discard:
        if (so->so_options & SO_ACCEPTCONN) {
                struct socket *sp;

                while ((sp = TAILQ_FIRST(&so->so_incomp)) != NULL) {
                        TAILQ_REMOVE(&so->so_incomp, sp, so_list);
                        sp->so_state &= ~SS_INCOMP;
                        sp->so_head = NULL;
                        so->so_incqlen--;
                        if ((sp->so_state & SS_ABORTING) == 0) {
                                sp->so_state |= SS_ABORTING;
                                soaborta(sp);
                        }
                }
                while ((sp = TAILQ_FIRST(&so->so_comp)) != NULL) {
                        TAILQ_REMOVE(&so->so_comp, sp, so_list);
                        sp->so_state &= ~SS_COMP;
                        sp->so_head = NULL;
                        so->so_qlen--;
                        if ((sp->so_state & SS_ABORTING) == 0) {
                                sp->so_state |= SS_ABORTING;
                                soaborta(sp);
                        }
                }
        }
        if (so->so_state & SS_NOFDREF)
                panic("soclose: NOFDREF");
        so->so_state |= SS_NOFDREF;
        sofree(so);
        crit_exit();
        return (error);
}

/*
 * Abort and destroy a socket.
 */
void
soabort(struct socket *so)
{
        so_pru_abort(so);
}

void
soaborta(struct socket *so)
{
        so_pru_aborta(so);
}

int
soaccept(struct socket *so, struct sockaddr **nam)
{
        int error;

        crit_enter();
        if ((so->so_state & SS_NOFDREF) == 0)
                panic("soaccept: !NOFDREF");
        so->so_state &= ~SS_NOFDREF;
        error = so_pru_accept(so, nam);
        crit_exit();
        return (error);
}

int
soconnect(struct socket *so, struct sockaddr *nam, struct thread *td)
{
        int error;

        if (so->so_options & SO_ACCEPTCONN)
                return (EOPNOTSUPP);
        crit_enter();
        /*
         * If protocol is connection-based, can only connect once.
         * Otherwise, if connected, try to disconnect first.
         * This allows user to disconnect by connecting to, e.g.,
         * a null address.
         */
        if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) &&
            ((so->so_proto->pr_flags & PR_CONNREQUIRED) ||
            (error = sodisconnect(so)))) {
                error = EISCONN;
        } else {
                /*
                 * Prevent accumulated error from previous connection
                 * from biting us.
                 */
                so->so_error = 0;
                error = so_pru_connect(so, nam, td);
        }
        crit_exit();
        return (error);
}

int
soconnect2(struct socket *so1, struct socket *so2)
{
        int error;

        crit_enter();
        error = so_pru_connect2(so1, so2);
        crit_exit();
        return (error);
}

int
sodisconnect(struct socket *so)
{
        int error;

        crit_enter();
        if ((so->so_state & SS_ISCONNECTED) == 0) {
                error = ENOTCONN;
                goto bad;
        }
        if (so->so_state & SS_ISDISCONNECTING) {
                error = EALREADY;
                goto bad;
        }
        error = so_pru_disconnect(so);
bad:
        crit_exit();
        return (error);
}

#define SBLOCKWAIT(f)   (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK)
/*
 * Send on a socket.
 * If send must go all at once and message is larger than
 * send buffering, then hard error.
 * Lock against other senders.
 * If must go all at once and not enough room now, then
 * inform user that this would block and do nothing.
 * Otherwise, if nonblocking, send as much as possible.
 * The data to be sent is described by "uio" if nonzero,
 * otherwise by the mbuf chain "top" (which must be null
 * if uio is not).  Data provided in mbuf chain must be small
 * enough to send all at once.
 *
 * Returns nonzero on error, timeout or signal; callers
 * must check for short counts if EINTR/ERESTART are returned.
 * Data and control buffers are freed on return.
 */
int
sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
        struct mbuf *top, struct mbuf *control, int flags,
        struct thread *td)
{
        struct mbuf **mp;
        struct mbuf *m;
        long space, len, resid;
        int clen = 0, error, dontroute, mlen;
        int atomic = sosendallatonce(so) || top;
        int pru_flags;

        if (uio)
                resid = uio->uio_resid;
        else
                resid = top->m_pkthdr.len;
        /*
         * In theory resid should be unsigned.
         * However, space must be signed, as it might be less than 0
         * if we over-committed, and we must use a signed comparison
         * of space and resid.  On the other hand, a negative resid
         * causes us to loop sending 0-length segments to the protocol.
         *
         * Also check to make sure that MSG_EOR isn't used on SOCK_STREAM
         * type sockets since that's an error.
         */
        if (resid < 0 || (so->so_type == SOCK_STREAM && (flags & MSG_EOR))) {
                error = EINVAL;
                goto out;
        }

        dontroute =
            (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 &&
            (so->so_proto->pr_flags & PR_ATOMIC);
        if (td->td_lwp != NULL)
                td->td_lwp->lwp_ru.ru_msgsnd++;
        if (control)
                clen = control->m_len;
#define gotoerr(errcode)        { error = errcode; crit_exit(); goto release; }

restart:
        error = ssb_lock(&so->so_snd, SBLOCKWAIT(flags));
        if (error)
                goto out;
        do {
                crit_enter();
                if (so->so_state & SS_CANTSENDMORE)
                        gotoerr(EPIPE);
                if (so->so_error) {
                        error = so->so_error;
                        so->so_error = 0;
                        crit_exit();
                        goto release;
                }
                if ((so->so_state & SS_ISCONNECTED) == 0) {
                        /*
                         * `sendto' and `sendmsg' is allowed on a connection-
                         * based socket if it supports implied connect.
                         * Return ENOTCONN if not connected and no address is
                         * supplied.
                         */
                        if ((so->so_proto->pr_flags & PR_CONNREQUIRED) &&
                            (so->so_proto->pr_flags & PR_IMPLOPCL) == 0) {
                                if ((so->so_state & SS_ISCONFIRMING) == 0 &&
                                    !(resid == 0 && clen != 0))
                                        gotoerr(ENOTCONN);
                        } else if (addr == 0)
                            gotoerr(so->so_proto->pr_flags & PR_CONNREQUIRED ?
                                   ENOTCONN : EDESTADDRREQ);
                }
                if ((atomic && resid > so->so_snd.ssb_hiwat) ||
                    clen > so->so_snd.ssb_hiwat) {
                        gotoerr(EMSGSIZE);
                }
                space = ssb_space(&so->so_snd);
                if (flags & MSG_OOB)
                        space += 1024;
                if (space < resid + clen && uio &&
                    (atomic || space < so->so_snd.ssb_lowat || space < clen)) {
                        if (flags & (MSG_FNONBLOCKING|MSG_DONTWAIT))
                                gotoerr(EWOULDBLOCK);
                        ssb_unlock(&so->so_snd);
                        error = ssb_wait(&so->so_snd);
                        crit_exit();
                        if (error)
                                goto out;
                        goto restart;
                }
                crit_exit();
                mp = &top;
                space -= clen;
                do {
                    if (uio == NULL) {
                        /*
                         * Data is prepackaged in "top".
                         */
                        resid = 0;
                        if (flags & MSG_EOR)
                                top->m_flags |= M_EOR;
                    } else do {
                        m = m_getl(resid, MB_WAIT, MT_DATA,
                                   top == NULL ? M_PKTHDR : 0, &mlen);
                        if (top == NULL) {
                                m->m_pkthdr.len = 0;
                                m->m_pkthdr.rcvif = (struct ifnet *)0;
                        }
                        len = min(min(mlen, resid), space);
                        if (resid < MINCLSIZE) {
                                /*
                                 * For datagram protocols, leave room
                                 * for protocol headers in first mbuf.
                                 */
                                if (atomic && top == 0 && len < mlen)
                                        MH_ALIGN(m, len);
                        }
                        space -= len;
                        error = uiomove(mtod(m, caddr_t), (int)len, uio);
                        resid = uio->uio_resid;
                        m->m_len = len;
                        *mp = m;
                        top->m_pkthdr.len += len;
                        if (error)
                                goto release;
                        mp = &m->m_next;
                        if (resid <= 0) {
                                if (flags & MSG_EOR)
                                        top->m_flags |= M_EOR;
                                break;
                        }
                    } while (space > 0 && atomic);
                    if (dontroute)
                            so->so_options |= SO_DONTROUTE;
                    if (flags & MSG_OOB) {
                            pru_flags = PRUS_OOB;
                    } else if ((flags & MSG_EOF) &&
                               (so->so_proto->pr_flags & PR_IMPLOPCL) &&
                               (resid <= 0)) {
                            /*
                             * If the user set MSG_EOF, the protocol
                             * understands this flag and nothing left to
                             * send then use PRU_SEND_EOF instead of PRU_SEND.
                             */
                            pru_flags = PRUS_EOF;
                    } else if (resid > 0 && space > 0) {
                            /* If there is more to send, set PRUS_MORETOCOME */
                            pru_flags = PRUS_MORETOCOME;
                    } else {
                            pru_flags = 0;
                    }
                    crit_enter();
                    /*
                     * XXX all the SS_CANTSENDMORE checks previously
                     * done could be out of date.  We could have recieved
                     * a reset packet in an interrupt or maybe we slept
                     * while doing page faults in uiomove() etc. We could
                     * probably recheck again inside the splnet() protection
                     * here, but there are probably other places that this
                     * also happens.  We must rethink this.
                     */
                    error = so_pru_send(so, pru_flags, top, addr, control, td);
                    crit_exit();
                    if (dontroute)
                            so->so_options &= ~SO_DONTROUTE;
                    clen = 0;
                    control = 0;
                    top = 0;
                    mp = &top;
                    if (error)
                            goto release;
                } while (resid && space > 0);
        } while (resid);

release:
        ssb_unlock(&so->so_snd);
out:
        if (top)
                m_freem(top);
        if (control)
                m_freem(control);
        return (error);
}

/*
 * A specialization of sosend() for UDP based on protocol-specific knowledge:
 *   so->so_proto->pr_flags has the PR_ATOMIC field set.  This means that
 *      sosendallatonce() returns true,
 *      the "atomic" variable is true,
 *      and sosendudp() blocks until space is available for the entire send.
 *   so->so_proto->pr_flags does not have the PR_CONNREQUIRED or
 *      PR_IMPLOPCL flags set.
 *   UDP has no out-of-band data.
 *   UDP has no control data.
 *   UDP does not support MSG_EOR.
 */
int
sosendudp(struct socket *so, struct sockaddr *addr, struct uio *uio,
          struct mbuf *top, struct mbuf *control, int flags, struct thread *td)
{
        int resid, error;
        boolean_t dontroute;            /* temporary SO_DONTROUTE setting */

        if (td->td_lwp != NULL)
                td->td_lwp->lwp_ru.ru_msgsnd++;
        if (control)
                m_freem(control);

        KASSERT((uio && !top) || (top && !uio), ("bad arguments to sosendudp"));
        resid = uio ? uio->uio_resid : top->m_pkthdr.len;

restart:
        error = ssb_lock(&so->so_snd, SBLOCKWAIT(flags));
        if (error)
                goto out;

        crit_enter();
        if (so->so_state & SS_CANTSENDMORE)
                gotoerr(EPIPE);
        if (so->so_error) {
                error = so->so_error;
                so->so_error = 0;
                crit_exit();
                goto release;
        }
        if (!(so->so_state & SS_ISCONNECTED) && addr == NULL)
                gotoerr(EDESTADDRREQ);
        if (resid > so->so_snd.ssb_hiwat)
                gotoerr(EMSGSIZE);
        if (uio && ssb_space(&so->so_snd) < resid) {
                if (flags & (MSG_FNONBLOCKING|MSG_DONTWAIT))
                        gotoerr(EWOULDBLOCK);
                ssb_unlock(&so->so_snd);
                error = ssb_wait(&so->so_snd);
                crit_exit();
                if (error)
                        goto out;
                goto restart;
        }
        crit_exit();

        if (uio) {
                top = m_uiomove(uio);
                if (top == NULL)
                        goto release;
        }

        dontroute = (flags & MSG_DONTROUTE) && !(so->so_options & SO_DONTROUTE);
        if (dontroute)
                so->so_options |= SO_DONTROUTE;

        error = so_pru_send(so, 0, top, addr, NULL, td);
        top = NULL;             /* sent or freed in lower layer */

        if (dontroute)
                so->so_options &= ~SO_DONTROUTE;

release:
        ssb_unlock(&so->so_snd);
out:
        if (top)
                m_freem(top);
        return (error);
}

/*
 * Implement receive operations on a socket.
 * We depend on the way that records are added to the signalsockbuf
 * by sbappend*.  In particular, each record (mbufs linked through m_next)
 * must begin with an address if the protocol so specifies,
 * followed by an optional mbuf or mbufs containing ancillary data,
 * and then zero or more mbufs of data.
 * In order to avoid blocking network interrupts for the entire time here,
 * we exit the critical section while doing the actual copy to user space.
 * Although the signalsockbuf is locked, new data may still be appended,
 * and thus we must maintain consistency of the signalsockbuf during that time.
 *
 * The caller may receive the data as a single mbuf chain by supplying
 * an mbuf **mp0 for use in returning the chain.  The uio is then used
 * only for the count in uio_resid.
 */
int
soreceive(struct socket *so, struct sockaddr **psa, struct uio *uio,
          struct sockbuf *sio, struct mbuf **controlp, int *flagsp)
{
        struct mbuf *m, *n;
        struct mbuf *free_chain = NULL;
        int flags, len, error, offset;
        struct protosw *pr = so->so_proto;
        int moff, type = 0;
        int resid, orig_resid;

        if (uio)
                resid = uio->uio_resid;
        else
                resid = (int)(sio->sb_climit - sio->sb_cc);
        orig_resid = resid;

        if (psa)
                *psa = NULL;
        if (controlp)
                *controlp = NULL;
        if (flagsp)
                flags = *flagsp &~ MSG_EOR;
        else
                flags = 0;
        if (flags & MSG_OOB) {
                m = m_get(MB_WAIT, MT_DATA);
                if (m == NULL)
                        return (ENOBUFS);
                error = so_pru_rcvoob(so, m, flags & MSG_PEEK);
                if (error)
                        goto bad;
                if (sio) {
                        do {
                                sbappend(sio, m);
                                resid -= m->m_len;
                        } while (resid > 0 && m);
                } else {
                        do {
                                uio->uio_resid = resid;
                                error = uiomove(mtod(m, caddr_t),
                                                (int)min(resid, m->m_len), uio);
                                resid = uio->uio_resid;
                                m = m_free(m);
                        } while (uio->uio_resid && error == 0 && m);
                }
bad:
                if (m)
                        m_freem(m);
                return (error);
        }
        if (so->so_state & SS_ISCONFIRMING && resid)
                so_pru_rcvd(so, 0);

restart:
        crit_enter();
        error = ssb_lock(&so->so_rcv, SBLOCKWAIT(flags));
        if (error)
                goto done;

        m = so->so_rcv.ssb_mb;
        /*
         * If we have less data than requested, block awaiting more
         * (subject to any timeout) if:
         *   1. the current count is less than the low water mark, or
         *   2. MSG_WAITALL is set, and it is possible to do the entire
         *      receive operation at once if we block (resid <= hiwat).
         *   3. MSG_DONTWAIT is not set
         * If MSG_WAITALL is set but resid is larger than the receive buffer,
         * we have to do the receive in sections, and thus risk returning
         * a short count if a timeout or signal occurs after we start.
         */
        if (m == NULL || (((flags & MSG_DONTWAIT) == 0 &&
            so->so_rcv.ssb_cc < resid) &&
            (so->so_rcv.ssb_cc < so->so_rcv.ssb_lowat ||
            ((flags & MSG_WAITALL) && resid <= so->so_rcv.ssb_hiwat)) &&
            m->m_nextpkt == 0 && (pr->pr_flags & PR_ATOMIC) == 0)) {
                KASSERT(m != NULL || !so->so_rcv.ssb_cc, ("receive 1"));
                if (so->so_error) {
                        if (m)
                                goto dontblock;
                        error = so->so_error;
                        if ((flags & MSG_PEEK) == 0)
                                so->so_error = 0;
                        goto release;
                }
                if (so->so_state & SS_CANTRCVMORE) {
                        if (m)
                                goto dontblock;
                        else
                                goto release;
                }
                for (; m; m = m->m_next) {
                        if (m->m_type == MT_OOBDATA  || (m->m_flags & M_EOR)) {
                                m = so->so_rcv.ssb_mb;
                                goto dontblock;
                        }
                }
                if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 &&
                    (pr->pr_flags & PR_CONNREQUIRED)) {
                        error = ENOTCONN;
                        goto release;
                }
                if (resid == 0)
                        goto release;
                if (flags & (MSG_FNONBLOCKING|MSG_DONTWAIT)) {
                        error = EWOULDBLOCK;
                        goto release;
                }
                ssb_unlock(&so->so_rcv);
                error = ssb_wait(&so->so_rcv);
                if (error)
                        goto done;
                crit_exit();
                goto restart;
        }
dontblock:
        if (uio && uio->uio_td && uio->uio_td->td_proc)
                uio->uio_td->td_lwp->lwp_ru.ru_msgrcv++;

        /*
         * note: m should be == sb_mb here.  Cache the next record while
         * cleaning up.  Note that calling m_free*() will break out critical
         * section.
         */
        KKASSERT(m == so->so_rcv.ssb_mb);

        /*
         * Skip any address mbufs prepending the record.
         */
        if (pr->pr_flags & PR_ADDR) {
                KASSERT(m->m_type == MT_SONAME, ("receive 1a"));
                orig_resid = 0;
                if (psa)
                        *psa = dup_sockaddr(mtod(m, struct sockaddr *));
                if (flags & MSG_PEEK)
                        m = m->m_next;
                else
                        m = sbunlinkmbuf(&so->so_rcv.sb, m, &free_chain);
        }

        /*
         * Skip any control mbufs prepending the record.
         */
#ifdef SCTP
        if (pr->pr_flags & PR_ADDR_OPT) {
                /*
                 * For SCTP we may be getting a
                 * whole message OR a partial delivery.
                 */
                if (m && m->m_type == MT_SONAME) {
                        orig_resid = 0;
                        if (psa)
                                *psa = dup_sockaddr(mtod(m, struct sockaddr *));
                        if (flags & MSG_PEEK)
                                m = m->m_next;
                        else
                                m = sbunlinkmbuf(&so->so_rcv.sb, m, &free_chain);
                }
        }
#endif /* SCTP */
        while (m && m->m_type == MT_CONTROL && error == 0) {
                if (flags & MSG_PEEK) {
                        if (controlp)
                                *controlp = m_copy(m, 0, m->m_len);
                        m = m->m_next;  /* XXX race */
                } else {
                        if (controlp) {
                                n = sbunlinkmbuf(&so->so_rcv.sb, m, NULL);
                                if (pr->pr_domain->dom_externalize &&
                                    mtod(m, struct cmsghdr *)->cmsg_type ==
                                    SCM_RIGHTS)
                                   error = (*pr->pr_domain->dom_externalize)(m);
                                *controlp = m;
                                m = n;
                        } else {
                                m = sbunlinkmbuf(&so->so_rcv.sb, m, &free_chain);
                        }
                }
                if (controlp && *controlp) {
                        orig_resid = 0;
                        controlp = &(*controlp)->m_next;
                }
        }

        /*
         * flag OOB data.
         */
        if (m) {
                type = m->m_type;
                if (type == MT_OOBDATA)
                        flags |= MSG_OOB;
        }

        /*
         * Copy to the UIO or mbuf return chain (*mp).
         */
        moff = 0;
        offset = 0;
        while (m && resid > 0 && error == 0) {
                if (m->m_type == MT_OOBDATA) {
                        if (type != MT_OOBDATA)
                                break;
                } else if (type == MT_OOBDATA)
                        break;
                else
                    KASSERT(m->m_type == MT_DATA || m->m_type == MT_HEADER,
                        ("receive 3"));
                so->so_state &= ~SS_RCVATMARK;
                len = resid;
                if (so->so_oobmark && len > so->so_oobmark - offset)
                        len = so->so_oobmark - offset;
                if (len > m->m_len - moff)
                        len = m->m_len - moff;

                /*
                 * Copy out to the UIO or pass the mbufs back to the SIO.
                 * The SIO is dealt with when we eat the mbuf, but deal
                 * with the resid here either way.
                 */
                if (uio) {
                        crit_exit();
                        uio->uio_resid = resid;
                        error = uiomove(mtod(m, caddr_t) + moff, len, uio);
                        resid = uio->uio_resid;
                        crit_enter();
                        if (error)
                                goto release;
                } else {
                        resid -= len;
                }

                /*
                 * Eat the entire mbuf or just a piece of it
                 */
                if (len == m->m_len - moff) {
                        if (m->m_flags & M_EOR)
                                flags |= MSG_EOR;
#ifdef SCTP
                        if (m->m_flags & M_NOTIFICATION)
                                flags |= MSG_NOTIFICATION;
#endif /* SCTP */
                        if (flags & MSG_PEEK) {
                                m = m->m_next;
                                moff = 0;
                        } else {
                                if (sio) {
                                        n = sbunlinkmbuf(&so->so_rcv.sb, m, NULL);
                                        sbappend(sio, m);
                                        m = n;
                                } else {
                                        m = sbunlinkmbuf(&so->so_rcv.sb, m, &free_chain);
                                }
                        }
                } else {
                        if (flags & MSG_PEEK) {
                                moff += len;
                        } else {
                                if (sio) {
                                        n = m_copym(m, 0, len, MB_WAIT);
                                        if (n)
                                                sbappend(sio, n);
                                }
                                m->m_data += len;
                                m->m_len -= len;
                                so->so_rcv.ssb_cc -= len;
                        }
                }
                if (so->so_oobmark) {
                        if ((flags & MSG_PEEK) == 0) {
                                so->so_oobmark -= len;
                                if (so->so_oobmark == 0) {
                                        so->so_state |= SS_RCVATMARK;
                                        break;
                                }
                        } else {
                                offset += len;
                                if (offset == so->so_oobmark)
                                        break;
                        }
                }
                if (flags & MSG_EOR)
                        break;
                /*
                 * If the MSG_WAITALL flag is set (for non-atomic socket),
                 * we must not quit until resid == 0 or an error
                 * termination.  If a signal/timeout occurs, return
                 * with a short count but without error.
                 * Keep signalsockbuf locked against other readers.
                 */
                while ((flags & MSG_WAITALL) && m == NULL && 
                       resid > 0 && !sosendallatonce(so) && 
                       so->so_rcv.ssb_mb == NULL) {
                        if (so->so_error || so->so_state & SS_CANTRCVMORE)
                                break;
                        /*
                         * The window might have closed to zero, make
                         * sure we send an ack now that we've drained
                         * the buffer or we might end up blocking until
                         * the idle takes over (5 seconds).
                         */
                        if (pr->pr_flags & PR_WANTRCVD && so->so_pcb)
                                so_pru_rcvd(so, flags);
                        error = ssb_wait(&so->so_rcv);
                        if (error) {
                                ssb_unlock(&so->so_rcv);
                                error = 0;
                                goto done;
                        }
                        m = so->so_rcv.ssb_mb;
                }
        }

        /*
         * If an atomic read was requested but unread data still remains
         * in the record, set MSG_TRUNC.
         */
        if (m && pr->pr_flags & PR_ATOMIC)
                flags |= MSG_TRUNC;

        /*
         * Cleanup.  If an atomic read was requested drop any unread data.
         */
        if ((flags & MSG_PEEK) == 0) {
                if (m && (pr->pr_flags & PR_ATOMIC))
                        sbdroprecord(&so->so_rcv.sb);
                if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb)
                        so_pru_rcvd(so, flags);
        }

        if (orig_resid == resid && orig_resid &&
            (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) {
                ssb_unlock(&so->so_rcv);
                crit_exit();
                goto restart;
        }

        if (flagsp)
                *flagsp |= flags;
release:
        ssb_unlock(&so->so_rcv);
done:
        crit_exit();
        if (free_chain)
                m_freem(free_chain);
        return (error);
}

int
soshutdown(struct socket *so, int how)
{
        if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
                return (EINVAL);

        if (how != SHUT_WR)
                sorflush(so);
        if (how != SHUT_RD)
                return (so_pru_shutdown(so));
        return (0);
}

void
sorflush(struct socket *so)
{
        struct signalsockbuf *ssb = &so->so_rcv;
        struct protosw *pr = so->so_proto;
        struct signalsockbuf asb;

        ssb->ssb_flags |= SSB_NOINTR;
        (void) ssb_lock(ssb, M_WAITOK);

        crit_enter();
        socantrcvmore(so);
        ssb_unlock(ssb);
        asb = *ssb;
        bzero((caddr_t)ssb, sizeof (*ssb));
        if (asb.ssb_flags & SSB_KNOTE) {
                ssb->ssb_sel.si_note = asb.ssb_sel.si_note;
                ssb->ssb_flags = SSB_KNOTE;
        }
        crit_exit();

        if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose)
                (*pr->pr_domain->dom_dispose)(asb.ssb_mb);
        ssb_release(&asb, so);
}

#ifdef INET
static int
do_setopt_accept_filter(struct socket *so, struct sockopt *sopt)
{
        struct accept_filter_arg        *afap = NULL;
        struct accept_filter    *afp;
        struct so_accf  *af = so->so_accf;
        int     error = 0;

        /* do not set/remove accept filters on non listen sockets */
        if ((so->so_options & SO_ACCEPTCONN) == 0) {
                error = EINVAL;
                goto out;
        }

        /* removing the filter */
        if (sopt == NULL) {
                if (af != NULL) {
                        if (af->so_accept_filter != NULL && 
                                af->so_accept_filter->accf_destroy != NULL) {
                                af->so_accept_filter->accf_destroy(so);
                        }
                        if (af->so_accept_filter_str != NULL) {
                                FREE(af->so_accept_filter_str, M_ACCF);
                        }
                        FREE(af, M_ACCF);
                        so->so_accf = NULL;
                }
                so->so_options &= ~SO_ACCEPTFILTER;
                return (0);
        }
        /* adding a filter */
        /* must remove previous filter first */
        if (af != NULL) {
                error = EINVAL;
                goto out;
        }
        /* don't put large objects on the kernel stack */
        MALLOC(afap, struct accept_filter_arg *, sizeof(*afap), M_TEMP, M_WAITOK);
        error = sooptcopyin(sopt, afap, sizeof *afap, sizeof *afap);
        afap->af_name[sizeof(afap->af_name)-1] = '\0';
        afap->af_arg[sizeof(afap->af_arg)-1] = '\0';
        if (error)
                goto out;
        afp = accept_filt_get(afap->af_name);
        if (afp == NULL) {
                error = ENOENT;
                goto out;
        }
        MALLOC(af, struct so_accf *, sizeof(*af), M_ACCF, M_WAITOK | M_ZERO);
        if (afp->accf_create != NULL) {
                if (afap->af_name[0] != '\0') {
                        int len = strlen(afap->af_name) + 1;

                        MALLOC(af->so_accept_filter_str, char *, len, M_ACCF, M_WAITOK);
                        strcpy(af->so_accept_filter_str, afap->af_name);
                }
                af->so_accept_filter_arg = afp->accf_create(so, afap->af_arg);
                if (af->so_accept_filter_arg == NULL) {
                        FREE(af->so_accept_filter_str, M_ACCF);
                        FREE(af, M_ACCF);
                        so->so_accf = NULL;
                        error = EINVAL;
                        goto out;
                }
        }
        af->so_accept_filter = afp;
        so->so_accf = af;
        so->so_options |= SO_ACCEPTFILTER;
out:
        if (afap != NULL)
                FREE(afap, M_TEMP);
        return (error);
}
#endif /* INET */

/*
 * Perhaps this routine, and sooptcopyout(), below, ought to come in
 * an additional variant to handle the case where the option value needs
 * to be some kind of integer, but not a specific size.
 * In addition to their use here, these functions are also called by the
 * protocol-level pr_ctloutput() routines.
 */
int
sooptcopyin(struct sockopt *sopt, void *buf, size_t len, size_t minlen)
{
        return soopt_to_kbuf(sopt, buf, len, minlen);
}

int
soopt_to_kbuf(struct sockopt *sopt, void *buf, size_t len, size_t minlen)
{
        size_t  valsize;

        KKASSERT(!sopt->sopt_val || kva_p(sopt->sopt_val));
        KKASSERT(kva_p(buf));

        /*
         * If the user gives us more than we wanted, we ignore it,
         * but if we don't get the minimum length the caller
         * wants, we return EINVAL.  On success, sopt->sopt_valsize
         * is set to however much we actually retrieved.
         */
        if ((valsize = sopt->sopt_valsize) < minlen)
                return EINVAL;
        if (valsize > len)
                sopt->sopt_valsize = valsize = len;

        bcopy(sopt->sopt_val, buf, valsize);
        return 0;
}


int
sosetopt(struct socket *so, struct sockopt *sopt)
{
        int     error, optval;
        struct  linger l;
        struct  timeval tv;
        u_long  val;

        error = 0;
        sopt->sopt_dir = SOPT_SET;
        if (sopt->sopt_level != SOL_SOCKET) {
                if (so->so_proto && so->so_proto->pr_ctloutput) {
                        return (so_pru_ctloutput(so, sopt));
                }
                error = ENOPROTOOPT;
        } else {
                switch (sopt->sopt_name) {
#ifdef INET
                case SO_ACCEPTFILTER:
                        error = do_setopt_accept_filter(so, sopt);
                        if (error)
                                goto bad;
                        break;
#endif /* INET */
                case SO_LINGER:
                        error = sooptcopyin(sopt, &l, sizeof l, sizeof l);
                        if (error)
                                goto bad;

                        so->so_linger = l.l_linger;
                        if (l.l_onoff)
                                so->so_options |= SO_LINGER;
                        else
                                so->so_options &= ~SO_LINGER;
                        break;

                case SO_DEBUG:
                case SO_KEEPALIVE:
                case SO_DONTROUTE:
                case SO_USELOOPBACK:
                case SO_BROADCAST:
                case SO_REUSEADDR:
                case SO_REUSEPORT:
                case SO_OOBINLINE:
                case SO_TIMESTAMP:
                        error = sooptcopyin(sopt, &optval, sizeof optval,
                                            sizeof optval);
                        if (error)
                                goto bad;
                        if (optval)
                                so->so_options |= sopt->sopt_name;
                        else
                                so->so_options &= ~sopt->sopt_name;
                        break;

                case SO_SNDBUF:
                case SO_RCVBUF:
                case SO_SNDLOWAT:
                case SO_RCVLOWAT:
                        error = sooptcopyin(sopt, &optval, sizeof optval,
                                            sizeof optval);
                        if (error)
                                goto bad;

                        /*
                         * Values < 1 make no sense for any of these
                         * options, so disallow them.
                         */
                        if (optval < 1) {
                                error = EINVAL;
                                goto bad;
                        }

                        switch (sopt->sopt_name) {
                        case SO_SNDBUF:
                        case SO_RCVBUF:
                                if (ssb_reserve(sopt->sopt_name == SO_SNDBUF ?
                                    &so->so_snd : &so->so_rcv, (u_long)optval,
                                    so,
                                    &curproc->p_rlimit[RLIMIT_SBSIZE]) == 0) {
                                        error = ENOBUFS;
                                        goto bad;
                                }
                                break;

                        /*
                         * Make sure the low-water is never greater than
                         * the high-water.
                         */
                        case SO_SNDLOWAT:
                                so->so_snd.ssb_lowat =
                                    (optval > so->so_snd.ssb_hiwat) ?
                                    so->so_snd.ssb_hiwat : optval;
                                break;
                        case SO_RCVLOWAT:
                                so->so_rcv.ssb_lowat =
                                    (optval > so->so_rcv.ssb_hiwat) ?
                                    so->so_rcv.ssb_hiwat : optval;
                                break;
                        }
                        break;

                case SO_SNDTIMEO:
                case SO_RCVTIMEO:
                        error = sooptcopyin(sopt, &tv, sizeof tv,
                                            sizeof tv);
                        if (error)
                                goto bad;

                        /* assert(hz > 0); */
                        if (tv.tv_sec < 0 || tv.tv_sec > SHRT_MAX / hz ||
                            tv.tv_usec < 0 || tv.tv_usec >= 1000000) {
                                error = EDOM;
                                goto bad;
                        }
                        /* assert(tick > 0); */
                        /* assert(ULONG_MAX - SHRT_MAX >= 1000000); */
                        val = (u_long)(tv.tv_sec * hz) + tv.tv_usec / tick;
                        if (val > SHRT_MAX) {
                                error = EDOM;
                                goto bad;
                        }
                        if (val == 0 && tv.tv_usec != 0)
                                val = 1;

                        switch (sopt->sopt_name) {
                        case SO_SNDTIMEO:
                                so->so_snd.ssb_timeo = val;
                                break;
                        case SO_RCVTIMEO:
                                so->so_rcv.ssb_timeo = val;
                                break;
                        }
                        break;
                default:
                        error = ENOPROTOOPT;
                        break;
                }
                if (error == 0 && so->so_proto && so->so_proto->pr_ctloutput) {
                        (void) so_pru_ctloutput(so, sopt);
                }
        }
bad:
        return (error);
}

/* Helper routine for getsockopt */
int
sooptcopyout(struct sockopt *sopt, const void *buf, size_t len)
{
        soopt_from_kbuf(sopt, buf, len);
        return 0;
}

void
soopt_from_kbuf(struct sockopt *sopt, const void *buf, size_t len)
{
        size_t  valsize;

        KKASSERT(!sopt->sopt_val || kva_p(sopt->sopt_val));
        KKASSERT(kva_p(buf));

        /*
         * Documented get behavior is that we always return a value,
         * possibly truncated to fit in the user's buffer.
         * Traditional behavior is that we always tell the user
         * precisely how much we copied, rather than something useful
         * like the total amount we had available for her.
         * Note that this interface is not idempotent; the entire answer must
         * generated ahead of time.
         */
        valsize = min(len, sopt->sopt_valsize);
        sopt->sopt_valsize = valsize;
        if (sopt->sopt_val != 0) {
                bcopy(buf, sopt->sopt_val, valsize);
        }
}

int
sogetopt(struct socket *so, struct sockopt *sopt)
{
        int     error, optval;
        struct  linger l;
        struct  timeval tv;
#ifdef INET
        struct accept_filter_arg *afap;
#endif

        error = 0;
        sopt->sopt_dir = SOPT_GET;
        if (sopt->sopt_level != SOL_SOCKET) {
                if (so->so_proto && so->so_proto->pr_ctloutput) {
                        return (so_pru_ctloutput(so, sopt));
                } else
                        return (ENOPROTOOPT);
        } else {
                switch (sopt->sopt_name) {
#ifdef INET
                case SO_ACCEPTFILTER:
                        if ((so->so_options & SO_ACCEPTCONN) == 0)
                                return (EINVAL);
                        MALLOC(afap, struct accept_filter_arg *, sizeof(*afap),
                                M_TEMP, M_WAITOK | M_ZERO);
                        if ((so->so_options & SO_ACCEPTFILTER) != 0) {
                                strcpy(afap->af_name, so->so_accf->so_accept_filter->accf_name);
                                if (so->so_accf->so_accept_filter_str != NULL)
                                        strcpy(afap->af_arg, so->so_accf->so_accept_filter_str);
                        }
                        error = sooptcopyout(sopt, afap, sizeof(*afap));
                        FREE(afap, M_TEMP);
                        break;
#endif /* INET */
                        
                case SO_LINGER:
                        l.l_onoff = so->so_options & SO_LINGER;
                        l.l_linger = so->so_linger;
                        error = sooptcopyout(sopt, &l, sizeof l);
                        break;

                case SO_USELOOPBACK:
                case SO_DONTROUTE:
                case SO_DEBUG:
                case SO_KEEPALIVE:
                case SO_REUSEADDR:
                case SO_REUSEPORT:
                case SO_BROADCAST:
                case SO_OOBINLINE:
                case SO_TIMESTAMP:
                        optval = so->so_options & sopt->sopt_name;
integer:
                        error = sooptcopyout(sopt, &optval, sizeof optval);
                        break;

                case SO_TYPE:
                        optval = so->so_type;
                        goto integer;

                case SO_ERROR:
                        optval = so->so_error;
                        so->so_error = 0;
                        goto integer;

                case SO_SNDBUF:
                        optval = so->so_snd.ssb_hiwat;
                        goto integer;

                case SO_RCVBUF:
                        optval = so->so_rcv.ssb_hiwat;
                        goto integer;

                case SO_SNDLOWAT:
                        optval = so->so_snd.ssb_lowat;
                        goto integer;

                case SO_RCVLOWAT:
                        optval = so->so_rcv.ssb_lowat;
                        goto integer;

                case SO_SNDTIMEO:
                case SO_RCVTIMEO:
                        optval = (sopt->sopt_name == SO_SNDTIMEO ?
                                  so->so_snd.ssb_timeo : so->so_rcv.ssb_timeo);

                        tv.tv_sec = optval / hz;
                        tv.tv_usec = (optval % hz) * tick;
                        error = sooptcopyout(sopt, &tv, sizeof tv);
                        break;                  

                default:
                        error = ENOPROTOOPT;
                        break;
                }
                return (error);
        }
}

/* XXX; prepare mbuf for (__FreeBSD__ < 3) routines. */
int
soopt_getm(struct sockopt *sopt, struct mbuf **mp)
{
        struct mbuf *m, *m_prev;
        int sopt_size = sopt->sopt_valsize, msize;

        m = m_getl(sopt_size, sopt->sopt_td ? MB_WAIT : MB_DONTWAIT, MT_DATA,
                   0, &msize);
        if (m == NULL)
                return (ENOBUFS);
        m->m_len = min(msize, sopt_size);
        sopt_size -= m->m_len;
        *mp = m;
        m_prev = m;

        while (sopt_size > 0) {
                m = m_getl(sopt_size, sopt->sopt_td ? MB_WAIT : MB_DONTWAIT,
                           MT_DATA, 0, &msize);
                if (m == NULL) {
                        m_freem(*mp);
                        return (ENOBUFS);
                }
                m->m_len = min(msize, sopt_size);
                sopt_size -= m->m_len;
                m_prev->m_next = m;
                m_prev = m;
        }
        return (0);
}

/* XXX; copyin sopt data into mbuf chain for (__FreeBSD__ < 3) routines. */
int
soopt_mcopyin(struct sockopt *sopt, struct mbuf *m)
{
        soopt_to_mbuf(sopt, m);
        return 0;
}

void
soopt_to_mbuf(struct sockopt *sopt, struct mbuf *m)
{
        size_t valsize;
        void *val;

        KKASSERT(!sopt->sopt_val || kva_p(sopt->sopt_val));
        KKASSERT(kva_p(m));
        if (sopt->sopt_val == NULL)
                return;
        val = sopt->sopt_val;
        valsize = sopt->sopt_valsize;
        while (m != NULL && valsize >= m->m_len) {
                bcopy(val, mtod(m, char *), m->m_len);
                valsize -= m->m_len;
                val = (caddr_t)val + m->m_len;
                m = m->m_next;
        }
        if (m != NULL) /* should be allocated enoughly at ip6_sooptmcopyin() */
                panic("ip6_sooptmcopyin");
}

/* XXX; copyout mbuf chain data into soopt for (__FreeBSD__ < 3) routines. */
int
soopt_mcopyout(struct sockopt *sopt, struct mbuf *m)
{
        return soopt_from_mbuf(sopt, m);
}

int
soopt_from_mbuf(struct sockopt *sopt, struct mbuf *m)
{
        struct mbuf *m0 = m;
        size_t valsize = 0;
        size_t maxsize;
        void *val;

        KKASSERT(!sopt->sopt_val || kva_p(sopt->sopt_val));
        KKASSERT(kva_p(m));
        if (sopt->sopt_val == NULL)
                return 0;
        val = sopt->sopt_val;
        maxsize = sopt->sopt_valsize;
        while (m != NULL && maxsize >= m->m_len) {
                bcopy(mtod(m, char *), val, m->m_len);
               maxsize -= m->m_len;
               val = (caddr_t)val + m->m_len;
               valsize += m->m_len;
               m = m->m_next;
        }
        if (m != NULL) {
                /* enough soopt buffer should be given from user-land */
                m_freem(m0);
                return (EINVAL);
        }
        sopt->sopt_valsize = valsize;
        return 0;
}

void
sohasoutofband(struct socket *so)
{
        if (so->so_sigio != NULL)
                pgsigio(so->so_sigio, SIGURG, 0);
        selwakeup(&so->so_rcv.ssb_sel);
}

int
sopoll(struct socket *so, int events, struct ucred *cred, struct thread *td)
{
        int revents = 0;

        crit_enter();

        if (events & (POLLIN | POLLRDNORM))
                if (soreadable(so))
                        revents |= events & (POLLIN | POLLRDNORM);

        if (events & POLLINIGNEOF)
                if (so->so_rcv.ssb_cc >= so->so_rcv.ssb_lowat ||
                        !TAILQ_EMPTY(&so->so_comp) || so->so_error)
                        revents |= POLLINIGNEOF;

        if (events & (POLLOUT | POLLWRNORM))
                if (sowriteable(so))
                        revents |= events & (POLLOUT | POLLWRNORM);

        if (events & (POLLPRI | POLLRDBAND))
                if (so->so_oobmark || (so->so_state & SS_RCVATMARK))
                        revents |= events & (POLLPRI | POLLRDBAND);

        if (revents == 0) {
                if (events &
                        (POLLIN | POLLINIGNEOF | POLLPRI | POLLRDNORM |
                         POLLRDBAND)) {
                        selrecord(td, &so->so_rcv.ssb_sel);
                        so->so_rcv.ssb_flags |= SSB_SEL;
                }

                if (events & (POLLOUT | POLLWRNORM)) {
                        selrecord(td, &so->so_snd.ssb_sel);
                        so->so_snd.ssb_flags |= SSB_SEL;
                }
        }

        crit_exit();
        return (revents);
}

int
sokqfilter(struct file *fp, struct knote *kn)
{
        struct socket *so = (struct socket *)kn->kn_fp->f_data;
        struct signalsockbuf *ssb;

        switch (kn->kn_filter) {
        case EVFILT_READ:
                if (so->so_options & SO_ACCEPTCONN)
                        kn->kn_fop = &solisten_filtops;
                else
                        kn->kn_fop = &soread_filtops;
                ssb = &so->so_rcv;
                break;
        case EVFILT_WRITE:
                kn->kn_fop = &sowrite_filtops;
                ssb = &so->so_snd;
                break;
        default:
                return (1);
        }

        crit_enter();
        SLIST_INSERT_HEAD(&ssb->ssb_sel.si_note, kn, kn_selnext);
        ssb->ssb_flags |= SSB_KNOTE;
        crit_exit();
        return (0);
}

static void
filt_sordetach(struct knote *kn)
{
        struct socket *so = (struct socket *)kn->kn_fp->f_data;

        crit_enter();
        SLIST_REMOVE(&so->so_rcv.ssb_sel.si_note, kn, knote, kn_selnext);
        if (SLIST_EMPTY(&so->so_rcv.ssb_sel.si_note))
                so->so_rcv.ssb_flags &= ~SSB_KNOTE;
        crit_exit();
}

/*ARGSUSED*/
static int
filt_soread(struct knote *kn, long hint)
{
        struct socket *so = (struct socket *)kn->kn_fp->f_data;

        kn->kn_data = so->so_rcv.ssb_cc;
        if (so->so_state & SS_CANTRCVMORE) {
                kn->kn_flags |= EV_EOF; 
                kn->kn_fflags = so->so_error;
                return (1);
        }
        if (so->so_error)       /* temporary udp error */
                return (1);
        if (kn->kn_sfflags & NOTE_LOWAT)
                return (kn->kn_data >= kn->kn_sdata);
        return (kn->kn_data >= so->so_rcv.ssb_lowat);
}

static void
filt_sowdetach(struct knote *kn)
{
        struct socket *so = (struct socket *)kn->kn_fp->f_data;

        crit_enter();
        SLIST_REMOVE(&so->so_snd.ssb_sel.si_note, kn, knote, kn_selnext);
        if (SLIST_EMPTY(&so->so_snd.ssb_sel.si_note))
                so->so_snd.ssb_flags &= ~SSB_KNOTE;
        crit_exit();
}

/*ARGSUSED*/
static int
filt_sowrite(struct knote *kn, long hint)
{
        struct socket *so = (struct socket *)kn->kn_fp->f_data;

        kn->kn_data = ssb_space(&so->so_snd);
        if (so->so_state & SS_CANTSENDMORE) {
                kn->kn_flags |= EV_EOF; 
                kn->kn_fflags = so->so_error;
                return (1);
        }
        if (so->so_error)       /* temporary udp error */
                return (1);
        if (((so->so_state & SS_ISCONNECTED) == 0) &&
            (so->so_proto->pr_flags & PR_CONNREQUIRED))
                return (0);
        if (kn->kn_sfflags & NOTE_LOWAT)
                return (kn->kn_data >= kn->kn_sdata);
        return (kn->kn_data >= so->so_snd.ssb_lowat);
}

/*ARGSUSED*/
static int
filt_solisten(struct knote *kn, long hint)
{
        struct socket *so = (struct socket *)kn->kn_fp->f_data;

        kn->kn_data = so->so_qlen;
        return (! TAILQ_EMPTY(&so->so_comp));
}