xnu-3248.60.10.tar.gz

[apple/xnu.git] / bsd / netinet / ip_divert.c

diff --git a/bsd/netinet/ip_divert.c b/bsd/netinet/ip_divert.c
index 512c65ca78c857f068d7f5d7abec49dc3bfa7c19..614c0c7c150aae2756f55ee1c939e73a0c90f078 100644 (file)
--- a/bsd/netinet/ip_divert.c
+++ b/bsd/netinet/ip_divert.c
@@ -1,16 +1,19 @@
 /*
 /*

- * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2013 Apple Inc. All rights reserved.

  *
  *

- * @APPLE_LICENSE_HEADER_START@
- * 
- * Copyright (c) 1999-2003 Apple Computer, Inc.  All Rights Reserved.
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@

  * 
  * This file contains Original Code and/or Modifications of Original Code
  * as defined in and that are subject to the Apple Public Source License
  * Version 2.0 (the 'License'). You may not use this file except in
  * 
  * This file contains Original Code and/or Modifications of Original Code
  * as defined in and that are subject to the Apple Public Source License
  * Version 2.0 (the 'License'). You may not use this file except in

- * compliance with the License. Please obtain a copy of the License at
- * http://www.opensource.apple.com/apsl/ and read it before using this
- * file.
+ * compliance with the License. The rights granted to you under the License
+ * may not be used to create, or enable the creation or redistribution of,
+ * unlawful or unlicensed copies of an Apple operating system, or to
+ * circumvent, violate, or enable the circumvention or violation of, any
+ * terms of an Apple operating system software license agreement.
+ * 
+ * Please obtain a copy of the License at
+ * http://www.opensource.apple.com/apsl/ and read it before using this file.

  * 
  * The Original Code and all software distributed under the License are
  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  * 
  * The Original Code and all software distributed under the License are
  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER


@@ -20,7 +23,7 @@
  * Please see the License for the specific language governing rights and
  * limitations under the License.
  * 
  * Please see the License for the specific language governing rights and
  * limitations under the License.
  * 

- * @APPLE_LICENSE_HEADER_END@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@

  */
 /*
  * Copyright (c) 1982, 1986, 1988, 1993
  */
 /*
  * Copyright (c) 1982, 1986, 1988, 1993


@@ -54,11 +57,10 @@
  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  * SUCH DAMAGE.
  *
  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  * SUCH DAMAGE.
  *

- * $FreeBSD: src/sys/netinet/ip_divert.c,v 1.42.2.4 2001/07/29 19:32:40 ume Exp $
+ * $FreeBSD: src/sys/netinet/ip_divert.c,v 1.98 2004/08/17 22:05:54 andre Exp $

  */
 
  */
 

-
-#ifndef INET
+#if !INET

 #error "IPDIVERT requires INET."
 #endif
 
 #error "IPDIVERT requires INET."
 #endif
 


@@ -67,15 +69,18 @@
 #include <sys/malloc.h>
 #include <sys/mbuf.h>
 #include <sys/socket.h>
 #include <sys/malloc.h>
 #include <sys/mbuf.h>
 #include <sys/socket.h>

+#include <sys/domain.h>

 #include <sys/protosw.h>
 #include <sys/socketvar.h>
 #include <sys/sysctl.h>
 #include <sys/systm.h>
 #include <sys/proc.h>
 
 #include <sys/protosw.h>
 #include <sys/socketvar.h>
 #include <sys/sysctl.h>
 #include <sys/systm.h>
 #include <sys/proc.h>
 

+#include <machine/endian.h>

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

+#include <net/kpi_protocol.h>

 
 #include <netinet/in.h>
 #include <netinet/in_systm.h>
 
 #include <netinet/in.h>
 #include <netinet/in_systm.h>


@@ -83,6 +88,11 @@
 #include <netinet/in_pcb.h>
 #include <netinet/in_var.h>
 #include <netinet/ip_var.h>
 #include <netinet/in_pcb.h>
 #include <netinet/in_var.h>
 #include <netinet/ip_var.h>

+#include <netinet/ip_fw.h>
+#include <netinet/ip_divert.h>
+
+#include <kern/zalloc.h>
+#include <libkern/OSAtomic.h>

 
 /*
  * Divert sockets
 
 /*
  * Divert sockets


@@ -95,58 +105,96 @@
 #define        DIVRCVQ         (65536 + 100)
 
 /*
 #define        DIVRCVQ         (65536 + 100)
 
 /*

- * A 16 bit cookie is passed to and from the user process.
- * The user process can send it back to help the caller know
- * something about where the packet originally came from.
+ * Divert sockets work in conjunction with ipfw, see the divert(4)
+ * manpage for features.
+ * Internally, packets selected by ipfw in ip_input() or ip_output(),
+ * and never diverted before, are passed to the input queue of the
+ * divert socket with a given 'divert_port' number (as specified in
+ * the matching ipfw rule), and they are tagged with a 16 bit cookie
+ * (representing the rule number of the matching ipfw rule), which
+ * is passed to process reading from the socket.

  *
  *

- * In the case of ipfw, then the cookie is the rule that sent
- * us here. On reinjection is is the rule after which processing
- * should continue. Leaving it the same will make processing start
- * at the rule number after that which sent it here. Setting it to
- * 0 will restart processing at the beginning. 
+ * Packets written to the divert socket are again tagged with a cookie
+ * (usually the same as above) and a destination address.
+ * If the destination address is INADDR_ANY then the packet is
+ * treated as outgoing and sent to ip_output(), otherwise it is
+ * treated as incoming and sent to ip_input().
+ * In both cases, the packet is tagged with the cookie.

  *
  *

- * For divert_packet(), ip_divert_cookie is an input value only.
- * For div_output(), ip_divert_cookie is an output value only.
+ * On reinjection, processing in ip_input() and ip_output()
+ * will be exactly the same as for the original packet, except that
+ * ipfw processing will start at the rule number after the one
+ * written in the cookie (so, tagging a packet with a cookie of 0
+ * will cause it to be effectively considered as a standard packet).

  */
  */

-u_int16_t ip_divert_cookie;

 
 /* Internal variables */
 static struct inpcbhead divcb;
 static struct inpcbinfo divcbinfo;
 
 
 /* Internal variables */
 static struct inpcbhead divcb;
 static struct inpcbinfo divcbinfo;
 

-static u_long  div_sendspace = DIVSNDQ;        /* XXX sysctl ? */
-static u_long  div_recvspace = DIVRCVQ;        /* XXX sysctl ? */
+static u_int32_t       div_sendspace = DIVSNDQ;        /* XXX sysctl ? */
+static u_int32_t       div_recvspace = DIVRCVQ;        /* XXX sysctl ? */

 
 /* Optimization: have this preinitialized */
 
 /* Optimization: have this preinitialized */

-static struct sockaddr_in divsrc = { sizeof(divsrc), AF_INET };
+static struct sockaddr_in divsrc = { sizeof(divsrc), AF_INET, 0, { 0 }, { 0,0,0,0,0,0,0,0 } };

 
 /* Internal functions */
 static int div_output(struct socket *so,
 
 /* Internal functions */
 static int div_output(struct socket *so,

-               struct mbuf *m, struct sockaddr *addr, struct mbuf *control);
+               struct mbuf *m, struct sockaddr_in *addr, struct mbuf *control);

 
 

+extern int load_ipfw(void);

 /*
  * Initialize divert connection block queue.
  */
 void
 /*
  * Initialize divert connection block queue.
  */
 void

-div_init(void)
+div_init(struct protosw *pp, struct domain *dp)

 {
 {

+#pragma unused(dp)
+       static int div_initialized = 0;
+       struct inpcbinfo *pcbinfo;
+
+       VERIFY((pp->pr_flags & (PR_INITIALIZED|PR_ATTACHED)) == PR_ATTACHED);
+
+       if (div_initialized)
+               return;
+       div_initialized = 1;
+

        LIST_INIT(&divcb);
        LIST_INIT(&divcb);

-       divcbinfo.listhead = &divcb;
+       divcbinfo.ipi_listhead = &divcb;

        /*
         * XXX We don't use the hash list for divert IP, but it's easier
         * to allocate a one entry hash list than it is to check all
        /*
         * XXX We don't use the hash list for divert IP, but it's easier
         * to allocate a one entry hash list than it is to check all

-        * over the place for hashbase == NULL.
+        * over the place for ipi_hashbase == NULL.

         */
         */

-       divcbinfo.hashbase = hashinit(1, M_PCB, &divcbinfo.hashmask);
-       divcbinfo.porthashbase = hashinit(1, M_PCB, &divcbinfo.porthashmask);
-       divcbinfo.ipi_zone = (void *) zinit(sizeof(struct inpcb),(maxsockets * sizeof(struct inpcb)),
+       divcbinfo.ipi_hashbase = hashinit(1, M_PCB, &divcbinfo.ipi_hashmask);
+       divcbinfo.ipi_porthashbase = hashinit(1, M_PCB, &divcbinfo.ipi_porthashmask);
+       divcbinfo.ipi_zone = zinit(sizeof(struct inpcb),(512 * sizeof(struct inpcb)),

                                   4096, "divzone");
                                   4096, "divzone");

+       pcbinfo = &divcbinfo;
+        /*
+        * allocate lock group attribute and group for udp pcb mutexes
+        */
+       pcbinfo->ipi_lock_grp_attr = lck_grp_attr_alloc_init();

 
 

-/*
- * ### LD 08/03: init IP forwarding at this point [ipfw is not a module yet]
- */
-#if !IPFIREWALL_KEXT
-       ip_fw_init();
+       pcbinfo->ipi_lock_grp = lck_grp_alloc_init("divcb", pcbinfo->ipi_lock_grp_attr);
+
+       /*
+        * allocate the lock attribute for divert pcb mutexes
+        */
+       pcbinfo->ipi_lock_attr = lck_attr_alloc_init();
+
+       if ((pcbinfo->ipi_lock = lck_rw_alloc_init(pcbinfo->ipi_lock_grp,
+           pcbinfo->ipi_lock_attr)) == NULL) {
+               panic("%s: unable to allocate PCB lock\n", __func__);
+               /* NOTREACHED */
+       }
+
+       in_pcbinfo_attach(&divcbinfo);
+
+#if IPFIREWALL
+       if (!IPFW_LOADED) {
+               load_ipfw();
+       }

 #endif
 }
 
 #endif
 }
 


@@ -155,9 +203,9 @@ div_init(void)
  * with that protocol number to enter the system from the outside.
  */
 void
  * with that protocol number to enter the system from the outside.
  */
 void

-div_input(struct mbuf *m, int off)
+div_input(struct mbuf *m, __unused int off)

 {
 {

-       ipstat.ips_noproto++;
+       OSAddAtomic(1, &ipstat.ips_noproto);

        m_freem(m);
 }
 
        m_freem(m);
 }
 


@@ -168,7 +216,7 @@ div_input(struct mbuf *m, int off)
  * then pass them along with mbuf chain.
  */
 void
  * then pass them along with mbuf chain.
  */
 void

-divert_packet(struct mbuf *m, int incoming, int port)
+divert_packet(struct mbuf *m, int incoming, int port, int rule)

 {
        struct ip *ip;
        struct inpcb *inp;
 {
        struct ip *ip;
        struct inpcb *inp;


@@ -178,9 +226,7 @@ divert_packet(struct mbuf *m, int incoming, int port)
        /* Sanity check */
        KASSERT(port != 0, ("%s: port=0", __FUNCTION__));
 
        /* Sanity check */
        KASSERT(port != 0, ("%s: port=0", __FUNCTION__));
 

-       /* Record and reset divert cookie */
-       divsrc.sin_port = ip_divert_cookie;
-       ip_divert_cookie = 0;
+       divsrc.sin_port = rule;         /* record matching rule */

 
        /* Assure header */
        if (m->m_len < sizeof(struct ip) &&
 
        /* Assure header */
        if (m->m_len < sizeof(struct ip) &&


@@ -201,15 +247,19 @@ divert_packet(struct mbuf *m, int incoming, int port)
                KASSERT((m->m_flags & M_PKTHDR), ("%s: !PKTHDR", __FUNCTION__));
 
                /* Find IP address for receive interface */
                KASSERT((m->m_flags & M_PKTHDR), ("%s: !PKTHDR", __FUNCTION__));
 
                /* Find IP address for receive interface */

+               ifnet_lock_shared(m->m_pkthdr.rcvif);

                TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) {
                TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) {

-                       if (ifa->ifa_addr == NULL)
-                               continue;
-                       if (ifa->ifa_addr->sa_family != AF_INET)
+                       IFA_LOCK(ifa);
+                       if (ifa->ifa_addr->sa_family != AF_INET) {
+                               IFA_UNLOCK(ifa);

                                continue;
                                continue;

+                       }

                        divsrc.sin_addr =
                        divsrc.sin_addr =

-                           ((struct sockaddr_in *) ifa->ifa_addr)->sin_addr;
+                           ((struct sockaddr_in *)(void *) ifa->ifa_addr)->sin_addr;
+                       IFA_UNLOCK(ifa);

                        break;
                }
                        break;
                }

+               ifnet_lock_done(m->m_pkthdr.rcvif);

        }
        /*
         * Record the incoming interface name whenever we have one.
        }
        /*
         * Record the incoming interface name whenever we have one.


@@ -235,28 +285,31 @@ divert_packet(struct mbuf *m, int incoming, int port)
                 * (see div_output for the other half of this.)
                 */ 
                snprintf(divsrc.sin_zero, sizeof(divsrc.sin_zero),
                 * (see div_output for the other half of this.)
                 */ 
                snprintf(divsrc.sin_zero, sizeof(divsrc.sin_zero),

-                       "%s%d", m->m_pkthdr.rcvif->if_name,
-                       m->m_pkthdr.rcvif->if_unit);
+                       "%s", if_name(m->m_pkthdr.rcvif));

        }
 
        /* Put packet on socket queue, if any */
        sa = NULL;
        nport = htons((u_int16_t)port);
        }
 
        /* Put packet on socket queue, if any */
        sa = NULL;
        nport = htons((u_int16_t)port);

+       lck_rw_lock_shared(divcbinfo.ipi_lock);         

        LIST_FOREACH(inp, &divcb, inp_list) {
                if (inp->inp_lport == nport)
                        sa = inp->inp_socket;
        }
        if (sa) {
        LIST_FOREACH(inp, &divcb, inp_list) {
                if (inp->inp_lport == nport)
                        sa = inp->inp_socket;
        }
        if (sa) {

+               int error = 0;
+               
+               socket_lock(sa, 1);

                if (sbappendaddr(&sa->so_rcv, (struct sockaddr *)&divsrc,
                if (sbappendaddr(&sa->so_rcv, (struct sockaddr *)&divsrc,

-                               m, (struct mbuf *)0) == 0)
-                       m_freem(m);
-               else
+                               m, (struct mbuf *)0, &error) != 0)

                        sorwakeup(sa);
                        sorwakeup(sa);

+               socket_unlock(sa, 1);

        } else {
                m_freem(m);
        } else {
                m_freem(m);

-               ipstat.ips_noproto++;
-               ipstat.ips_delivered--;
+               OSAddAtomic(1, &ipstat.ips_noproto);
+               OSAddAtomic(-1, &ipstat.ips_delivered);

         }
         }

+       lck_rw_done(divcbinfo.ipi_lock);        

 }
 
 /*
 }
 
 /*


@@ -265,28 +318,40 @@ divert_packet(struct mbuf *m, int incoming, int port)
  * If no address specified, or address is 0.0.0.0, send to ip_output();
  * otherwise, send to ip_input() and mark as having been received on
  * the interface with that address.
  * If no address specified, or address is 0.0.0.0, send to ip_output();
  * otherwise, send to ip_input() and mark as having been received on
  * the interface with that address.

+ * ###LOCK  called in inet_proto mutex when from div_send. 

  */
 static int
  */
 static int

-div_output(so, m, addr, control)
-       struct socket *so;
-       register struct mbuf *m;
-       struct sockaddr *addr;
-       struct mbuf *control;
+div_output(struct socket *so, struct mbuf *m, struct sockaddr_in *sin,
+          struct mbuf *control)

 {
 {

-       register struct inpcb *const inp = sotoinpcb(so);
-       register struct ip *const ip = mtod(m, struct ip *);
-       struct sockaddr_in *sin = (struct sockaddr_in *)addr;
+       struct inpcb *const inp = sotoinpcb(so);
+       struct ip *const ip = mtod(m, struct ip *);

        int error = 0;
        int error = 0;

+       mbuf_svc_class_t msc = MBUF_SC_UNSPEC;

 
 

-       if (control)
-               m_freem(control);               /* XXX */
+       if (control != NULL) {
+               msc = mbuf_service_class_from_control(control);

 
 

+               m_freem(control);               /* XXX */
+               control = NULL;
+       }

        /* Loopback avoidance and state recovery */
        if (sin) {
        /* Loopback avoidance and state recovery */
        if (sin) {

+               struct m_tag *mtag;
+               struct divert_tag *dt;

                int     len = 0;
                char    *c = sin->sin_zero;
 
                int     len = 0;
                char    *c = sin->sin_zero;
 

-               ip_divert_cookie = sin->sin_port;
+               mtag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_DIVERT,
+                               sizeof(struct divert_tag), M_NOWAIT, m);
+               if (mtag == NULL) {
+                       error = ENOBUFS;
+                       goto cantsend;
+               }
+               dt = (struct divert_tag *)(mtag+1);
+               dt->info = 0;
+               dt->cookie = sin->sin_port;
+               m_tag_prepend(m, mtag);

 
                /*
                 * Find receive interface with the given name or IP address.
 
                /*
                 * Find receive interface with the given name or IP address.


@@ -300,12 +365,15 @@ div_output(so, m, addr, control)
                while (*c++ && (len++ < sizeof(sin->sin_zero)));
                if ((len > 0) && (len < sizeof(sin->sin_zero)))
                        m->m_pkthdr.rcvif = ifunit(sin->sin_zero);
                while (*c++ && (len++ < sizeof(sin->sin_zero)));
                if ((len > 0) && (len < sizeof(sin->sin_zero)))
                        m->m_pkthdr.rcvif = ifunit(sin->sin_zero);

-       } else {
-               ip_divert_cookie = 0;

        }
 
        /* Reinject packet into the system as incoming or outgoing */
        if (!sin || sin->sin_addr.s_addr == 0) {
        }
 
        /* Reinject packet into the system as incoming or outgoing */
        if (!sin || sin->sin_addr.s_addr == 0) {

+               struct ip_out_args ipoa =
+                   { IFSCOPE_NONE, { 0 }, IPOAF_SELECT_SRCIF, 0 };
+               struct route ro;
+               struct ip_moptions *imo;
+

                /*
                 * Don't allow both user specified and setsockopt options,
                 * and don't allow packet length sizes that will crash
                /*
                 * Don't allow both user specified and setsockopt options,
                 * and don't allow packet length sizes that will crash


@@ -317,47 +385,67 @@ div_output(so, m, addr, control)
                }
 
                /* Convert fields to host order for ip_output() */
                }
 
                /* Convert fields to host order for ip_output() */

+#if BYTE_ORDER != BIG_ENDIAN

                NTOHS(ip->ip_len);
                NTOHS(ip->ip_off);
                NTOHS(ip->ip_len);
                NTOHS(ip->ip_off);

+#endif
+
+               OSAddAtomic(1, &ipstat.ips_rawout);
+               /* Copy the cached route and take an extra reference */
+               inp_route_copyout(inp, &ro);
+
+               set_packet_service_class(m, so, msc, 0);

 
 

+               imo = inp->inp_moptions;
+               if (imo != NULL)
+                       IMO_ADDREF(imo);
+               socket_unlock(so, 0);
+#if CONFIG_MACF_NET
+               mac_mbuf_label_associate_inpcb(inp, m);
+#endif

                /* Send packet to output processing */
                /* Send packet to output processing */

-               ipstat.ips_rawout++;                    /* XXX */
-               error = ip_output(m, inp->inp_options, &inp->inp_route,
+               error = ip_output(m, inp->inp_options, &ro,

                        (so->so_options & SO_DONTROUTE) |
                        (so->so_options & SO_DONTROUTE) |

-                       IP_ALLOWBROADCAST | IP_RAWOUTPUT,
-                       inp->inp_moptions);
+                       IP_ALLOWBROADCAST | IP_RAWOUTPUT | IP_OUTARGS,
+                       imo, &ipoa);
+
+               socket_lock(so, 0);
+               if (imo != NULL)
+                       IMO_REMREF(imo);
+               /* Synchronize cached PCB route */
+               inp_route_copyin(inp, &ro);

        } else {
                struct  ifaddr *ifa;
 
                /* If no luck with the name above. check by IP address.  */
                if (m->m_pkthdr.rcvif == NULL) {
        } else {
                struct  ifaddr *ifa;
 
                /* If no luck with the name above. check by IP address.  */
                if (m->m_pkthdr.rcvif == NULL) {

+                       struct sockaddr_in _sin;

                        /*
                        /*

-                        * Make sure there are no distractions
-                        * for ifa_ifwithaddr. Clear the port and the ifname.
-                        * Maybe zap all 8 bytes at once using a 64bit write?
+                        * Make sure there are no distractions for
+                        * ifa_ifwithaddr; use sanitized version.

                         */
                         */

-                       bzero(sin->sin_zero, sizeof(sin->sin_zero));
-                       /* *((u_int64_t *)sin->sin_zero) = 0; */ /* XXX ?? */
-                       sin->sin_port = 0;
-                       if (!(ifa = ifa_ifwithaddr((struct sockaddr *) sin))) {
+                       bzero(&_sin, sizeof (_sin));
+                       _sin.sin_family = AF_INET;
+                       _sin.sin_len = sizeof (struct sockaddr_in);
+                       _sin.sin_addr.s_addr = sin->sin_addr.s_addr;
+                       if (!(ifa = ifa_ifwithaddr(SA(&_sin)))) {

                                error = EADDRNOTAVAIL;
                                goto cantsend;
                        }
                        m->m_pkthdr.rcvif = ifa->ifa_ifp;
                                error = EADDRNOTAVAIL;
                                goto cantsend;
                        }
                        m->m_pkthdr.rcvif = ifa->ifa_ifp;

+                       IFA_REMREF(ifa);

                }
                }

-
+#if CONFIG_MACF_NET
+               mac_mbuf_label_associate_socket(so, m);
+#endif

                /* Send packet to input processing */
                /* Send packet to input processing */

-               ip_input(m);
+               proto_inject(PF_INET, m);

        }
 
        }
 

-       /* paranoid: Reset for next time (and other packets) */
-       /* almost definitly already done in the ipfw filter but.. */
-       ip_divert_cookie = 0;

        return error;
 
 cantsend:
        m_freem(m);
        return error;
 
 cantsend:
        m_freem(m);

-       ip_divert_cookie = 0;

        return error;
 }
 
        return error;
 }
 


@@ -365,20 +453,19 @@ static int
 div_attach(struct socket *so, int proto, struct proc *p)
 {
        struct inpcb *inp;
 div_attach(struct socket *so, int proto, struct proc *p)
 {
        struct inpcb *inp;

-       int error, s;
+       int error;
+

 
        inp  = sotoinpcb(so);
        if (inp)
                panic("div_attach");
 
        inp  = sotoinpcb(so);
        if (inp)
                panic("div_attach");

-       if (p && (error = suser(p->p_ucred, &p->p_acflag)) != 0)
+       if ((error = proc_suser(p)) != 0)

                return error;
 
        error = soreserve(so, div_sendspace, div_recvspace);
        if (error)
                return error;
                return error;
 
        error = soreserve(so, div_sendspace, div_recvspace);
        if (error)
                return error;

-       s = splnet();

        error = in_pcballoc(so, &divcbinfo, p);
        error = in_pcballoc(so, &divcbinfo, p);

-       splx(s);

        if (error)
                return error;
        inp = (struct inpcb *)so->so_pcb;
        if (error)
                return error;
        inp = (struct inpcb *)so->so_pcb;


@@ -388,6 +475,14 @@ div_attach(struct socket *so, int proto, struct proc *p)
        /* The socket is always "connected" because
           we always know "where" to send the packet */
        so->so_state |= SS_ISCONNECTED;
        /* The socket is always "connected" because
           we always know "where" to send the packet */
        so->so_state |= SS_ISCONNECTED;

+
+#ifdef MORE_DICVLOCK_DEBUG
+       printf("div_attach: so=0x%llx sopcb=0x%llx lock=0x%llx ref=%x\n",
+           (uint64_t)VM_KERNEL_ADDRPERM(so),
+           (uint64_t)VM_KERNEL_ADDRPERM(so->so_pcb),
+           (uint64_t)VM_KERNEL_ADDRPERM(&(sotoinpcb(so)->inpcb_mtx)),
+           so->so_usecount);
+#endif

        return 0;
 }
 
        return 0;
 }
 


@@ -396,10 +491,18 @@ div_detach(struct socket *so)
 {
        struct inpcb *inp;
 
 {
        struct inpcb *inp;
 

+#ifdef MORE_DICVLOCK_DEBUG
+       printf("div_detach: so=0x%llx sopcb=0x%llx lock=0x%llx ref=%x\n",
+           (uint64_t)VM_KERNEL_ADDRPERM(so),
+           (uint64_t)VM_KERNEL_ADDRPERM(so->so_pcb),
+           (uint64_t)VM_KERNEL_ADDRPERM(&(sotoinpcb(so)->inpcb_mtx)),
+           so->so_usecount);
+#endif

        inp = sotoinpcb(so);
        if (inp == 0)
        inp = sotoinpcb(so);
        if (inp == 0)

-               panic("div_detach");
+               panic("div_detach: so=%p null inp\n", so);

        in_pcbdetach(inp);
        in_pcbdetach(inp);

+       inp->inp_state = INPCB_STATE_DEAD;

        return 0;
 }
 
        return 0;
 }
 


@@ -422,10 +525,8 @@ static int
 div_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 {
        struct inpcb *inp;
 div_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
 {
        struct inpcb *inp;

-       int s;

        int error;
 
        int error;
 

-       s = splnet();

        inp = sotoinpcb(so);
        /* in_pcbbind assumes that the socket is a sockaddr_in
        * and in_pcbbind requires a valid address. Since divert
        inp = sotoinpcb(so);
        /* in_pcbbind assumes that the socket is a sockaddr_in
        * and in_pcbbind requires a valid address. Since divert


@@ -437,10 +538,9 @@ div_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
        if (nam->sa_family != AF_INET) {
                error = EAFNOSUPPORT;
        } else {
        if (nam->sa_family != AF_INET) {
                error = EAFNOSUPPORT;
        } else {

-               ((struct sockaddr_in *)nam)->sin_addr.s_addr = INADDR_ANY;
+               ((struct sockaddr_in *)(void *)nam)->sin_addr.s_addr = INADDR_ANY;

                error = in_pcbbind(inp, nam, p);
        }
                error = in_pcbbind(inp, nam, p);
        }

-       splx(s);

        return error;
 }
 
        return error;
 }
 


@@ -452,25 +552,27 @@ div_shutdown(struct socket *so)
 }
 
 static int
 }
 
 static int

-div_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
-        struct mbuf *control, struct proc *p)
+div_send(struct socket *so, __unused int flags, struct mbuf *m, struct sockaddr *nam,
+        struct mbuf *control, __unused struct proc *p)

 {
        /* Packet must have a header (but that's about it) */
        if (m->m_len < sizeof (struct ip) &&
            (m = m_pullup(m, sizeof (struct ip))) == 0) {
 {
        /* Packet must have a header (but that's about it) */
        if (m->m_len < sizeof (struct ip) &&
            (m = m_pullup(m, sizeof (struct ip))) == 0) {

-               ipstat.ips_toosmall++;
+               OSAddAtomic(1, &ipstat.ips_toosmall);

                m_freem(m);
                return EINVAL;
        }
 
        /* Send packet */
                m_freem(m);
                return EINVAL;
        }
 
        /* Send packet */

-       return div_output(so, m, nam, control);
+       return div_output(so, m, SIN(nam), control);

 }
 
 }
 

+#if 0

 static int
 div_pcblist SYSCTL_HANDLER_ARGS
 {
 static int
 div_pcblist SYSCTL_HANDLER_ARGS
 {

-       int error, i, n, s;
+#pragma unused(oidp, arg1, arg2)
+       int error, i, n;

        struct inpcb *inp, **inp_list;
        inp_gen_t gencnt;
        struct xinpgen xig;
        struct inpcb *inp, **inp_list;
        inp_gen_t gencnt;
        struct xinpgen xig;


@@ -479,57 +581,64 @@ div_pcblist SYSCTL_HANDLER_ARGS
         * The process of preparing the TCB list is too time-consuming and
         * resource-intensive to repeat twice on every request.
         */
         * The process of preparing the TCB list is too time-consuming and
         * resource-intensive to repeat twice on every request.
         */

-       if (req->oldptr == 0) {
+       lck_rw_lock_exclusive(divcbinfo.ipi_lock);
+       if (req->oldptr == USER_ADDR_NULL) {

                n = divcbinfo.ipi_count;
                req->oldidx = 2 * (sizeof xig)
                        + (n + n/8) * sizeof(struct xinpcb);
                n = divcbinfo.ipi_count;
                req->oldidx = 2 * (sizeof xig)
                        + (n + n/8) * sizeof(struct xinpcb);

+               lck_rw_done(divcbinfo.ipi_lock);

                return 0;
        }
 
                return 0;
        }
 

-       if (req->newptr != 0)
+       if (req->newptr != USER_ADDR_NULL) {
+               lck_rw_done(divcbinfo.ipi_lock);

                return EPERM;
                return EPERM;

+       }

 
        /*
         * OK, now we're committed to doing something.
         */
 
        /*
         * OK, now we're committed to doing something.
         */

-       s = splnet();

        gencnt = divcbinfo.ipi_gencnt;
        n = divcbinfo.ipi_count;
        gencnt = divcbinfo.ipi_gencnt;
        n = divcbinfo.ipi_count;

-       splx(s);

 
 

+       bzero(&xig, sizeof(xig));

        xig.xig_len = sizeof xig;
        xig.xig_count = n;
        xig.xig_gen = gencnt;
        xig.xig_sogen = so_gencnt;
        error = SYSCTL_OUT(req, &xig, sizeof xig);
        xig.xig_len = sizeof xig;
        xig.xig_count = n;
        xig.xig_gen = gencnt;
        xig.xig_sogen = so_gencnt;
        error = SYSCTL_OUT(req, &xig, sizeof xig);

-       if (error)
+       if (error) {
+               lck_rw_done(divcbinfo.ipi_lock);

                return error;
                return error;

+       }

 
        inp_list = _MALLOC(n * sizeof *inp_list, M_TEMP, M_WAITOK);
 
        inp_list = _MALLOC(n * sizeof *inp_list, M_TEMP, M_WAITOK);

-       if (inp_list == 0)
+       if (inp_list == 0) {
+               lck_rw_done(divcbinfo.ipi_lock);

                return ENOMEM;
                return ENOMEM;

+       }

        
        

-       s = splnet();
-       for (inp = LIST_FIRST(divcbinfo.listhead), i = 0; inp && i < n;
+       for (inp = LIST_FIRST(divcbinfo.ipi_listhead), i = 0; inp && i < n;

             inp = LIST_NEXT(inp, inp_list)) {
 #ifdef __APPLE__
             inp = LIST_NEXT(inp, inp_list)) {
 #ifdef __APPLE__

-               if (inp->inp_gencnt <= gencnt)
+               if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD)

 #else
                if (inp->inp_gencnt <= gencnt && !prison_xinpcb(req->p, inp))
 #endif
                        inp_list[i++] = inp;
        }
 #else
                if (inp->inp_gencnt <= gencnt && !prison_xinpcb(req->p, inp))
 #endif
                        inp_list[i++] = inp;
        }

-       splx(s);

        n = i;
 
        error = 0;
        for (i = 0; i < n; i++) {
                inp = inp_list[i];
        n = i;
 
        error = 0;
        for (i = 0; i < n; i++) {
                inp = inp_list[i];

-               if (inp->inp_gencnt <= gencnt) {
+               if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD) {

                        struct xinpcb xi;
                        struct xinpcb xi;

+
+                       bzero(&xi, sizeof(xi));

                        xi.xi_len = sizeof xi;
                        /* XXX should avoid extra copy */
                        xi.xi_len = sizeof xi;
                        /* XXX should avoid extra copy */

-                       bcopy(inp, &xi.xi_inp, sizeof *inp);
+                       inpcb_to_compat(inp, &xi.xi_inp);

                        if (inp->inp_socket)
                                sotoxsocket(inp->inp_socket, &xi.xi_socket);
                        error = SYSCTL_OUT(req, &xi, sizeof xi);
                        if (inp->inp_socket)
                                sotoxsocket(inp->inp_socket, &xi.xi_socket);
                        error = SYSCTL_OUT(req, &xi, sizeof xi);


@@ -543,28 +652,137 @@ div_pcblist SYSCTL_HANDLER_ARGS
                 * while we were processing this request, and it
                 * might be necessary to retry.
                 */
                 * while we were processing this request, and it
                 * might be necessary to retry.
                 */

-               s = splnet();
+               bzero(&xig, sizeof(xig));
+               xig.xig_len = sizeof xig;

                xig.xig_gen = divcbinfo.ipi_gencnt;
                xig.xig_sogen = so_gencnt;
                xig.xig_count = divcbinfo.ipi_count;
                xig.xig_gen = divcbinfo.ipi_gencnt;
                xig.xig_sogen = so_gencnt;
                xig.xig_count = divcbinfo.ipi_count;

-               splx(s);

                error = SYSCTL_OUT(req, &xig, sizeof xig);
        }
        FREE(inp_list, M_TEMP);
                error = SYSCTL_OUT(req, &xig, sizeof xig);
        }
        FREE(inp_list, M_TEMP);

+       lck_rw_done(divcbinfo.ipi_lock);

        return error;
 }
        return error;
 }

+#endif

 
 

-#warning Fix SYSCTL net_inet_divert
-#if 0
-SYSCTL_DECL(_net_inet_divert);
-SYSCTL_PROC(_net_inet_divert, OID_AUTO, pcblist, CTLFLAG_RD, 0, 0,
-           div_pcblist, "S,xinpcb", "List of active divert sockets");
+__private_extern__ int
+div_lock(struct socket *so, int refcount, void *lr)
+{
+       void *lr_saved;
+
+       if (lr == NULL)
+               lr_saved = __builtin_return_address(0);
+       else
+               lr_saved = lr;
+
+#ifdef MORE_DICVLOCK_DEBUG
+       printf("div_lock: so=0x%llx sopcb=0x%llx lock=0x%llx ref=%x "
+           "lr=0x%llx\n", (uint64_t)VM_KERNEL_ADDRPERM(so),
+           (uint64_t)VM_KERNEL_ADDRPERM(so->so_pcb), so->so_pcb ?
+           (uint64_t)VM_KERNEL_ADDRPERM(&(sotoinpcb(so)->inpcb_mtx)) : NULL,
+           so->so_usecount, (uint64_t)VM_KERNEL_ADDRPERM(lr_saved));
+#endif
+       if (so->so_pcb) {
+               lck_mtx_lock(&((struct inpcb *)so->so_pcb)->inpcb_mtx);
+       } else  {
+               panic("div_lock: so=%p NO PCB! lr=%p lrh= lrh= %s\n", 
+                   so, lr_saved, solockhistory_nr(so));
+               /* NOTREACHED */
+       }
+
+       if (so->so_usecount < 0) {
+               panic("div_lock: so=%p so_pcb=%p lr=%p ref=%x lrh= %s\n",
+                   so, so->so_pcb, lr_saved, so->so_usecount,
+                   solockhistory_nr(so));
+               /* NOTREACHED */
+       }
+
+       if (refcount)
+               so->so_usecount++;
+       so->lock_lr[so->next_lock_lr] = lr_saved;
+       so->next_lock_lr = (so->next_lock_lr+1) % SO_LCKDBG_MAX;
+
+       return (0);
+}
+
+__private_extern__ int
+div_unlock(struct socket *so, int refcount, void *lr)
+{
+       void *lr_saved;
+       lck_mtx_t * mutex_held;
+       struct inpcb *inp = sotoinpcb(so);
+
+       if (lr == NULL)
+               lr_saved = __builtin_return_address(0);
+       else
+               lr_saved = lr;
+
+#ifdef MORE_DICVLOCK_DEBUG
+       printf("div_unlock: so=0x%llx sopcb=0x%llx lock=0x%llx ref=%x "
+           "lr=0x%llx\n", (uint64_t)VM_KERNEL_ADDRPERM(so),
+           (uint64_t)VM_KERNEL_ADDRPERM(so->so_pcb), so->so_pcb ?
+           (uint64_t)VM_KERNEL_ADDRPERM(&(sotoinpcb(so)->inpcb_mtx)) : NULL,
+           so->so_usecount, lr_saved);

 #endif
 #endif

+       if (refcount)
+               so->so_usecount--;
+
+       if (so->so_usecount < 0) {
+               panic("div_unlock: so=%p usecount=%x lrh= %s\n", 
+                   so, so->so_usecount, solockhistory_nr(so));
+               /* NOTREACHED */
+       }
+       if (so->so_pcb == NULL) {
+               panic("div_unlock: so=%p NO PCB usecount=%x lr=%p lrh= %s\n",
+                   so, so->so_usecount, lr_saved, solockhistory_nr(so));
+               /* NOTREACHED */
+       }
+       mutex_held = &((struct inpcb *)so->so_pcb)->inpcb_mtx;
+
+       if (so->so_usecount == 0 && (inp->inp_wantcnt == WNT_STOPUSING)) {
+               lck_rw_lock_exclusive(divcbinfo.ipi_lock);
+               if (inp->inp_state != INPCB_STATE_DEAD)
+                       in_pcbdetach(inp);
+               in_pcbdispose(inp);
+               lck_rw_done(divcbinfo.ipi_lock);
+               return (0);
+       }
+       lck_mtx_assert(mutex_held, LCK_MTX_ASSERT_OWNED);
+       so->unlock_lr[so->next_unlock_lr] = lr_saved;
+       so->next_unlock_lr = (so->next_unlock_lr+1) % SO_LCKDBG_MAX;
+       lck_mtx_unlock(mutex_held);
+       return (0);
+}
+
+__private_extern__ lck_mtx_t *
+div_getlock(struct socket *so, __unused int locktype)
+{
+       struct inpcb *inpcb = (struct inpcb *)so->so_pcb;
+       
+       if (so->so_pcb)  {
+               if (so->so_usecount < 0)
+                       panic("div_getlock: so=%p usecount=%x lrh= %s\n", 
+                           so, so->so_usecount, solockhistory_nr(so));
+               return(&inpcb->inpcb_mtx);
+       } else {
+               panic("div_getlock: so=%p NULL NO PCB lrh= %s\n", 
+                   so, solockhistory_nr(so));
+               return (so->so_proto->pr_domain->dom_mtx);
+       }
+}

 
 struct pr_usrreqs div_usrreqs = {
 
 struct pr_usrreqs div_usrreqs = {

-       div_abort, pru_accept_notsupp, div_attach, div_bind,
-       pru_connect_notsupp, pru_connect2_notsupp, in_control, div_detach,
-       div_disconnect, pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
-       pru_rcvoob_notsupp, div_send, pru_sense_null, div_shutdown,
-       in_setsockaddr, sosend, soreceive, sopoll
+       .pru_abort =            div_abort,
+       .pru_attach =           div_attach,
+       .pru_bind =             div_bind,
+       .pru_control =          in_control,
+       .pru_detach =           div_detach,
+       .pru_disconnect =       div_disconnect,
+       .pru_peeraddr =         in_getpeeraddr,
+       .pru_send =             div_send,
+       .pru_shutdown =         div_shutdown,
+       .pru_sockaddr =         in_getsockaddr,
+       .pru_sosend =           sosend,
+       .pru_soreceive =        soreceive,

 };
 };

+