]> git.saurik.com Git - apple/xnu.git/blobdiff - osfmk/ipc/ipc_port.c
xnu-6153.61.1.tar.gz
[apple/xnu.git] / osfmk / ipc / ipc_port.c
index 9bb7ed78115ef3517e56250371fe83ea01d5bdfc..b8cddf28ae5d75998d6242b1592076b04a510291 100644 (file)
@@ -1,52 +1,64 @@
 /*
- * Copyright (c) 2002,2000 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2019 Apple Inc. All rights reserved.
  *
- * @APPLE_LICENSE_HEADER_START@
- * 
- * The contents of this file constitute Original Code as defined in and
- * are subject to the Apple Public Source License Version 1.1 (the
- * "License").  You may not use this file except in compliance with the
- * License.  Please obtain a copy of the License at
- * http://www.apple.com/publicsource and read it before using this file.
- * 
- * This Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * @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
+ * 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
  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
- * License for the specific language governing rights and limitations
- * under the License.
- * 
- * @APPLE_LICENSE_HEADER_END@
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
+ * limitations under the License.
+ *
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  */
 /*
  * @OSF_FREE_COPYRIGHT@
  */
-/* 
+/*
  * Mach Operating System
  * Copyright (c) 1991,1990,1989 Carnegie Mellon University
  * All Rights Reserved.
- * 
+ *
  * Permission to use, copy, modify and distribute this software and its
  * documentation is hereby granted, provided that both the copyright
  * notice and this permission notice appear in all copies of the
  * software, derivative works or modified versions, and any portions
  * thereof, and that both notices appear in supporting documentation.
- * 
+ *
  * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
  * CONDITION.  CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR
  * ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
- * 
+ *
  * Carnegie Mellon requests users of this software to return to
- * 
+ *
  *  Software Distribution Coordinator  or  Software.Distribution@CS.CMU.EDU
  *  School of Computer Science
  *  Carnegie Mellon University
  *  Pittsburgh PA 15213-3890
- * 
+ *
  * any improvements or extensions that they make and grant Carnegie Mellon
  * the rights to redistribute these changes.
  */
+/*
+ * NOTICE: This file was modified by McAfee Research in 2004 to introduce
+ * support for mandatory and extensible security protections.  This notice
+ * is included in support of clause 2.2 (b) of the Apple Public License,
+ * Version 2.0.
+ */
 /*
  */
 /*
  *     Functions to manipulate IPC ports.
  */
 
-#include <norma_vm.h>
-#include <mach_kdb.h>
 #include <zone_debug.h>
 #include <mach_assert.h>
 
 #include <mach/port.h>
 #include <mach/kern_return.h>
-#include <kern/lock.h>
 #include <kern/ipc_kobject.h>
 #include <kern/thread.h>
 #include <kern/misc_protos.h>
-#include <kern/wait_queue.h>
+#include <kern/waitq.h>
+#include <kern/policy_internal.h>
+#include <kern/debug.h>
+#include <kern/kcdata.h>
 #include <ipc/ipc_entry.h>
 #include <ipc/ipc_space.h>
 #include <ipc/ipc_object.h>
 #include <ipc/ipc_kmsg.h>
 #include <ipc/ipc_mqueue.h>
 #include <ipc/ipc_notify.h>
-#include <ipc/ipc_print.h>
 #include <ipc/ipc_table.h>
+#include <ipc/ipc_importance.h>
+#include <machine/limits.h>
+#include <kern/turnstile.h>
 
-#if    MACH_KDB
-#include <machine/db_machdep.h>
-#include <ddb/db_command.h>
-#include <ddb/db_expr.h>
-#endif /* MACH_KDB */
+#include <security/mac_mach_internal.h>
 
 #include <string.h>
 
-decl_mutex_data(,      ipc_port_multiple_lock_data)
-decl_mutex_data(,      ipc_port_timestamp_lock_data)
-ipc_port_timestamp_t   ipc_port_timestamp_data;
+decl_lck_spin_data(, ipc_port_multiple_lock_data);
+ipc_port_timestamp_t    ipc_port_timestamp_data;
+int ipc_portbt;
+extern int prioritize_launch;
+
+#if     MACH_ASSERT
+void    ipc_port_init_debug(
+       ipc_port_t      port,
+       uintptr_t       *callstack,
+       unsigned int    callstack_max);
+
+void    ipc_port_callstack_init_debug(
+       uintptr_t       *callstack,
+       unsigned int    callstack_max);
+
+#endif  /* MACH_ASSERT */
+
+static void
+ipc_port_send_turnstile_recompute_push_locked(
+       ipc_port_t port);
 
-#if    MACH_ASSERT
-void   ipc_port_init_debug(
-               ipc_port_t      port);
-#endif /* MACH_ASSERT */
+static thread_t
+ipc_port_get_watchport_inheritor(
+       ipc_port_t port);
 
-#if    MACH_KDB && ZONE_DEBUG
-/* Forwards */
-void   print_type_ports(unsigned, unsigned);
-void   print_ports(void);
-#endif /* MACH_KDB && ZONE_DEBUG */
+void
+ipc_port_release(ipc_port_t port)
+{
+       ip_release(port);
+}
+
+void
+ipc_port_reference(ipc_port_t port)
+{
+       ip_reference(port);
+}
 
 /*
  *     Routine:        ipc_port_timestamp
@@ -112,19 +144,13 @@ void      print_ports(void);
 ipc_port_timestamp_t
 ipc_port_timestamp(void)
 {
-       ipc_port_timestamp_t timestamp;
-
-       ipc_port_timestamp_lock();
-       timestamp = ipc_port_timestamp_data++;
-       ipc_port_timestamp_unlock();
-
-       return timestamp;
+       return OSIncrementAtomic(&ipc_port_timestamp_data);
 }
 
 /*
- *     Routine:        ipc_port_dnrequest
+ *     Routine:        ipc_port_request_alloc
  *     Purpose:
- *             Try to allocate a dead-name request slot.
+ *             Try to allocate a request slot.
  *             If successful, returns the request index.
  *             Otherwise returns zero.
  *     Conditions:
@@ -134,43 +160,84 @@ ipc_port_timestamp(void)
  *             KERN_NO_SPACE           No index allocated.
  */
 
+#if IMPORTANCE_INHERITANCE
+kern_return_t
+ipc_port_request_alloc(
+       ipc_port_t                      port,
+       mach_port_name_t                name,
+       ipc_port_t                      soright,
+       boolean_t                       send_possible,
+       boolean_t                       immediate,
+       ipc_port_request_index_t        *indexp,
+       boolean_t                       *importantp)
+#else
 kern_return_t
-ipc_port_dnrequest(
-       ipc_port_t                      port,
-       mach_port_name_t                name,
-       ipc_port_t                      soright,
-       ipc_port_request_index_t        *indexp)
+ipc_port_request_alloc(
+       ipc_port_t                      port,
+       mach_port_name_t                name,
+       ipc_port_t                      soright,
+       boolean_t                       send_possible,
+       boolean_t                       immediate,
+       ipc_port_request_index_t        *indexp)
+#endif /* IMPORTANCE_INHERITANCE */
 {
        ipc_port_request_t ipr, table;
        ipc_port_request_index_t index;
+       uintptr_t mask = 0;
 
-       assert(ip_active(port));
+#if IMPORTANCE_INHERITANCE
+       *importantp = FALSE;
+#endif /* IMPORTANCE_INHERITANCE */
+
+       require_ip_active(port);
        assert(name != MACH_PORT_NULL);
        assert(soright != IP_NULL);
 
-       table = port->ip_dnrequests;
-       if (table == IPR_NULL)
+       table = port->ip_requests;
+
+       if (table == IPR_NULL) {
                return KERN_NO_SPACE;
+       }
 
        index = table->ipr_next;
-       if (index == 0)
+       if (index == 0) {
                return KERN_NO_SPACE;
+       }
 
        ipr = &table[index];
        assert(ipr->ipr_name == MACH_PORT_NULL);
 
        table->ipr_next = ipr->ipr_next;
        ipr->ipr_name = name;
-       ipr->ipr_soright = soright;
+
+       if (send_possible) {
+               mask |= IPR_SOR_SPREQ_MASK;
+               if (immediate) {
+                       mask |= IPR_SOR_SPARM_MASK;
+                       if (port->ip_sprequests == 0) {
+                               port->ip_sprequests = 1;
+#if IMPORTANCE_INHERITANCE
+                               /* TODO: Live importance support in send-possible */
+                               if (port->ip_impdonation != 0 &&
+                                   port->ip_spimportant == 0 &&
+                                   (task_is_importance_donor(current_task()))) {
+                                       *importantp = TRUE;
+                               }
+#endif /* IMPORTANCE_INHERTANCE */
+                       }
+               }
+       }
+       ipr->ipr_soright = IPR_SOR_MAKE(soright, mask);
 
        *indexp = index;
+
        return KERN_SUCCESS;
 }
 
 /*
- *     Routine:        ipc_port_dngrow
+ *     Routine:        ipc_port_request_grow
  *     Purpose:
- *             Grow a port's table of dead-name requests.
+ *             Grow a port's table of requests.
  *     Conditions:
  *             The port must be locked and active.
  *             Nothing else locked; will allocate memory.
@@ -184,27 +251,27 @@ ipc_port_dnrequest(
  */
 
 kern_return_t
-ipc_port_dngrow(
-       ipc_port_t      port,
-       int             target_size)
+ipc_port_request_grow(
+       ipc_port_t              port,
+       ipc_table_elems_t       target_size)
 {
        ipc_table_size_t its;
        ipc_port_request_t otable, ntable;
+       require_ip_active(port);
 
-       assert(ip_active(port));
-
-       otable = port->ip_dnrequests;
-       if (otable == IPR_NULL)
-               its = &ipc_table_dnrequests[0];
-       else
+       otable = port->ip_requests;
+       if (otable == IPR_NULL) {
+               its = &ipc_table_requests[0];
+       } else {
                its = otable->ipr_size + 1;
+       }
 
        if (target_size != ITS_SIZE_NONE) {
                if ((otable != IPR_NULL) &&
                    (target_size <= otable->ipr_size->its_size)) {
                        ip_unlock(port);
                        return KERN_SUCCESS;
-               }
+               }
                while ((its->its_size) && (its->its_size < target_size)) {
                        its++;
                }
@@ -218,24 +285,22 @@ ipc_port_dngrow(
        ip_unlock(port);
 
        if ((its->its_size == 0) ||
-           ((ntable = it_dnrequests_alloc(its)) == IPR_NULL)) {
-               ipc_port_release(port);
+           ((ntable = it_requests_alloc(its)) == IPR_NULL)) {
+               ip_release(port);
                return KERN_RESOURCE_SHORTAGE;
        }
 
        ip_lock(port);
-       ip_release(port);
 
        /*
         *      Check that port is still active and that nobody else
         *      has slipped in and grown the table on us.  Note that
-        *      just checking port->ip_dnrequests == otable isn't
-        *      sufficient; must check ipr_size.
+        *      just checking if the current table pointer == otable
+        *      isn't sufficient; must check ipr_size.
         */
 
-       if (ip_active(port) &&
-           (port->ip_dnrequests == otable) &&
-           ((otable == IPR_NULL) || (otable->ipr_size+1 == its))) {
+       if (ip_active(port) && (port->ip_requests == otable) &&
+           ((otable == IPR_NULL) || (otable->ipr_size + 1 == its))) {
                ipc_table_size_t oits;
                ipc_table_elems_t osize, nsize;
                ipc_port_request_index_t free, i;
@@ -248,10 +313,11 @@ ipc_port_dngrow(
                        free = otable->ipr_next;
 
                        (void) memcpy((void *)(ntable + 1),
-                             (const void *)(otable + 1),
-                             (osize - 1) * sizeof(struct ipc_port_request));
+                           (const void *)(otable + 1),
+                           (osize - 1) * sizeof(struct ipc_port_request));
                } else {
                        osize = 1;
+                       oits = 0;
                        free = 0;
                }
 
@@ -270,55 +336,149 @@ ipc_port_dngrow(
 
                ntable->ipr_next = free;
                ntable->ipr_size = its;
-               port->ip_dnrequests = ntable;
+               port->ip_requests = ntable;
                ip_unlock(port);
+               ip_release(port);
 
                if (otable != IPR_NULL) {
-                       it_dnrequests_free(oits, otable);
-               }
+                       it_requests_free(oits, otable);
+               }
        } else {
-               ip_check_unlock(port);
-               it_dnrequests_free(its, ntable);
+               ip_unlock(port);
+               ip_release(port);
+               it_requests_free(its, ntable);
        }
 
        return KERN_SUCCESS;
 }
+
 /*
- *     Routine:        ipc_port_dncancel
+ *     Routine:        ipc_port_request_sparm
  *     Purpose:
- *             Cancel a dead-name request and return the send-once right.
+ *             Arm delayed send-possible request.
  *     Conditions:
- *             The port must locked and active.
+ *             The port must be locked and active.
+ *
+ *             Returns TRUE if the request was armed
+ *             (or armed with importance in that version).
  */
 
-ipc_port_t
-ipc_port_dncancel(
-       ipc_port_t                      port,
-       mach_port_name_t                name,
-       ipc_port_request_index_t        index)
+boolean_t
+ipc_port_request_sparm(
+       ipc_port_t                      port,
+       __assert_only mach_port_name_t  name,
+       ipc_port_request_index_t        index,
+       mach_msg_option_t       option,
+       mach_msg_priority_t override)
+{
+       if (index != IE_REQ_NONE) {
+               ipc_port_request_t ipr, table;
+
+               require_ip_active(port);
+
+               table = port->ip_requests;
+               assert(table != IPR_NULL);
+
+               ipr = &table[index];
+               assert(ipr->ipr_name == name);
+
+               /* Is there a valid destination? */
+               if (IPR_SOR_SPREQ(ipr->ipr_soright)) {
+                       ipr->ipr_soright = IPR_SOR_MAKE(ipr->ipr_soright, IPR_SOR_SPARM_MASK);
+                       port->ip_sprequests = 1;
+
+                       if (option & MACH_SEND_OVERRIDE) {
+                               /* apply override to message queue */
+                               ipc_mqueue_override_send(&port->ip_messages, override);
+                       }
+
+#if IMPORTANCE_INHERITANCE
+                       if (((option & MACH_SEND_NOIMPORTANCE) == 0) &&
+                           (port->ip_impdonation != 0) &&
+                           (port->ip_spimportant == 0) &&
+                           (((option & MACH_SEND_IMPORTANCE) != 0) ||
+                           (task_is_importance_donor(current_task())))) {
+                               return TRUE;
+                       }
+#else
+                       return TRUE;
+#endif /* IMPORTANCE_INHERITANCE */
+               }
+       }
+       return FALSE;
+}
+
+/*
+ *     Routine:        ipc_port_request_type
+ *     Purpose:
+ *             Determine the type(s) of port requests enabled for a name.
+ *     Conditions:
+ *             The port must be locked or inactive (to avoid table growth).
+ *             The index must not be IE_REQ_NONE and for the name in question.
+ */
+mach_port_type_t
+ipc_port_request_type(
+       ipc_port_t                      port,
+       __assert_only mach_port_name_t  name,
+       ipc_port_request_index_t        index)
 {
        ipc_port_request_t ipr, table;
-       ipc_port_t dnrequest;
+       mach_port_type_t type = 0;
 
-       assert(ip_active(port));
-       assert(name != MACH_PORT_NULL);
-       assert(index != 0);
+       table = port->ip_requests;
+       assert(table != IPR_NULL);
+
+       assert(index != IE_REQ_NONE);
+       ipr = &table[index];
+       assert(ipr->ipr_name == name);
+
+       if (IP_VALID(IPR_SOR_PORT(ipr->ipr_soright))) {
+               type |= MACH_PORT_TYPE_DNREQUEST;
+
+               if (IPR_SOR_SPREQ(ipr->ipr_soright)) {
+                       type |= MACH_PORT_TYPE_SPREQUEST;
+
+                       if (!IPR_SOR_SPARMED(ipr->ipr_soright)) {
+                               type |= MACH_PORT_TYPE_SPREQUEST_DELAYED;
+                       }
+               }
+       }
+       return type;
+}
 
-       table = port->ip_dnrequests;
+/*
+ *     Routine:        ipc_port_request_cancel
+ *     Purpose:
+ *             Cancel a dead-name/send-possible request and return the send-once right.
+ *     Conditions:
+ *             The port must be locked and active.
+ *             The index must not be IPR_REQ_NONE and must correspond with name.
+ */
+
+ipc_port_t
+ipc_port_request_cancel(
+       ipc_port_t                      port,
+       __assert_only mach_port_name_t  name,
+       ipc_port_request_index_t        index)
+{
+       ipc_port_request_t ipr, table;
+       ipc_port_t request = IP_NULL;
+
+       require_ip_active(port);
+       table = port->ip_requests;
        assert(table != IPR_NULL);
 
+       assert(index != IE_REQ_NONE);
        ipr = &table[index];
-       dnrequest = ipr->ipr_soright;
        assert(ipr->ipr_name == name);
+       request = IPR_SOR_PORT(ipr->ipr_soright);
 
        /* return ipr to the free list inside the table */
-
        ipr->ipr_name = MACH_PORT_NULL;
        ipr->ipr_next = table->ipr_next;
        table->ipr_next = index;
 
-       return dnrequest;
+       return request;
 }
 
 /*
@@ -335,13 +495,12 @@ ipc_port_dncancel(
 
 void
 ipc_port_pdrequest(
-       ipc_port_t      port,
-       ipc_port_t      notify,
-       ipc_port_t      *previousp)
+       ipc_port_t      port,
+       ipc_port_t      notify,
+       ipc_port_t      *previousp)
 {
        ipc_port_t previous;
-
-       assert(ip_active(port));
+       require_ip_active(port);
 
        previous = port->ip_pdrequest;
        port->ip_pdrequest = notify;
@@ -364,15 +523,14 @@ ipc_port_pdrequest(
 
 void
 ipc_port_nsrequest(
-       ipc_port_t              port,
-       mach_port_mscount_t     sync,
-       ipc_port_t              notify,
-       ipc_port_t              *previousp)
+       ipc_port_t              port,
+       mach_port_mscount_t     sync,
+       ipc_port_t              notify,
+       ipc_port_t              *previousp)
 {
        ipc_port_t previous;
        mach_port_mscount_t mscount;
-
-       assert(ip_active(port));
+       require_ip_active(port);
 
        previous = port->ip_nsrequest;
        mscount = port->ip_mscount;
@@ -394,38 +552,79 @@ ipc_port_nsrequest(
 /*
  *     Routine:        ipc_port_clear_receiver
  *     Purpose:
- *             Prepares a receive right for transmission/destruction.
+ *             Prepares a receive right for transmission/destruction,
+ *             optionally performs mqueue destruction (with port lock held)
+ *
  *     Conditions:
  *             The port is locked and active.
+ *     Returns:
+ *             If should_destroy is TRUE, then the return value indicates
+ *             whether the caller needs to reap kmsg structures that should
+ *             be destroyed (by calling ipc_kmsg_reap_delayed)
+ *
+ *              If should_destroy is FALSE, this always returns FALSE
  */
 
-void
+boolean_t
 ipc_port_clear_receiver(
-       ipc_port_t      port)
+       ipc_port_t      port,
+       boolean_t       should_destroy)
 {
-       spl_t           s;
-
-       assert(ip_active(port));
+       ipc_mqueue_t    mqueue = &port->ip_messages;
+       boolean_t       reap_messages = FALSE;
 
        /*
-        * pull ourselves from any sets.
+        * Pull ourselves out of any sets to which we belong.
+        * We hold the port locked, so even though this acquires and releases
+        * the mqueue lock, we know we won't be added to any other sets.
         */
-       if (port->ip_pset_count != 0) {
+       if (port->ip_in_pset != 0) {
                ipc_pset_remove_from_all(port);
-               assert(port->ip_pset_count == 0);
+               assert(port->ip_in_pset == 0);
        }
 
        /*
         * Send anyone waiting on the port's queue directly away.
-        * Also clear the mscount and seqno.
+        * Also clear the mscount, seqno, guard bits
         */
-       s = splsched();
-       imq_lock(&port->ip_messages);
-       ipc_mqueue_changed(&port->ip_messages);
-       ipc_port_set_mscount(port, 0);
-       port->ip_messages.imq_seqno = 0;
+       imq_lock(mqueue);
+       if (port->ip_receiver_name) {
+               ipc_mqueue_changed(port->ip_receiver, mqueue);
+       } else {
+               ipc_mqueue_changed(NULL, mqueue);
+       }
+       port->ip_mscount = 0;
+       mqueue->imq_seqno = 0;
+       port->ip_context = port->ip_guarded = port->ip_strict_guard = 0;
+       /*
+        * clear the immovable bit so the port can move back to anyone listening
+        * for the port destroy notification
+        */
+       port->ip_immovable_receive = 0;
+
+       if (should_destroy) {
+               /*
+                * Mark the port and mqueue invalid, preventing further send/receive
+                * operations from succeeding. It's important for this to be
+                * done under the same lock hold as the ipc_mqueue_changed
+                * call to avoid additional threads blocking on an mqueue
+                * that's being destroyed.
+                *
+                * The port active bit needs to be guarded under mqueue lock for
+                * turnstiles
+                */
+               port->ip_object.io_bits &= ~IO_BITS_ACTIVE;
+               port->ip_timestamp = ipc_port_timestamp();
+               reap_messages = ipc_mqueue_destroy_locked(mqueue);
+       } else {
+               /* make port be in limbo */
+               port->ip_receiver_name = MACH_PORT_NULL;
+               port->ip_destination = IP_NULL;
+       }
+
        imq_unlock(&port->ip_messages);
-       splx(s);
+
+       return reap_messages;
 }
 
 /*
@@ -437,9 +636,10 @@ ipc_port_clear_receiver(
 
 void
 ipc_port_init(
-       ipc_port_t              port,
-       ipc_space_t             space,
-       mach_port_name_t        name)
+       ipc_port_t              port,
+       ipc_space_t             space,
+       ipc_port_init_flags_t   flags,
+       mach_port_name_t        name)
 {
        /* port->ip_kobject doesn't have to be initialized */
 
@@ -449,19 +649,44 @@ ipc_port_init(
        port->ip_mscount = 0;
        port->ip_srights = 0;
        port->ip_sorights = 0;
+       if (flags & IPC_PORT_INIT_MAKE_SEND_RIGHT) {
+               port->ip_srights = 1;
+               port->ip_mscount = 1;
+       }
 
        port->ip_nsrequest = IP_NULL;
        port->ip_pdrequest = IP_NULL;
-       port->ip_dnrequests = IPR_NULL;
+       port->ip_requests = IPR_NULL;
 
-       port->ip_pset_count = 0;
        port->ip_premsg = IKM_NULL;
+       port->ip_context = 0;
+       port->ip_reply_context = 0;
+
+       port->ip_sprequests  = 0;
+       port->ip_spimportant = 0;
+       port->ip_impdonation = 0;
+       port->ip_tempowner   = 0;
+
+       port->ip_guarded      = 0;
+       port->ip_strict_guard = 0;
+       port->ip_immovable_receive = 0;
+       port->ip_no_grant    = 0;
+       port->ip_immovable_send = 0;
+       port->ip_impcount    = 0;
+
+       port->ip_specialreply = (flags & IPC_PORT_INIT_SPECIAL_REPLY) != 0;
+       port->ip_sync_link_state = PORT_SYNC_LINK_ANY;
+       port->ip_sync_bootstrap_checkin = 0;
 
-#if    MACH_ASSERT
-       ipc_port_init_debug(port);
-#endif /* MACH_ASSERT */
+       ipc_special_reply_port_bits_reset(port);
 
-       ipc_mqueue_init(&port->ip_messages, FALSE /* set */);
+       port->ip_send_turnstile = TURNSTILE_NULL;
+
+       ipc_mqueue_kind_t kind = IPC_MQUEUE_KIND_NONE;
+       if (flags & IPC_PORT_INIT_MESSAGE_QUEUE) {
+               kind = IPC_MQUEUE_KIND_PORT;
+       }
+       ipc_mqueue_init(&port->ip_messages, kind);
 }
 
 /*
@@ -480,23 +705,41 @@ ipc_port_init(
 
 kern_return_t
 ipc_port_alloc(
-       ipc_space_t             space,
-       mach_port_name_t        *namep,
-       ipc_port_t              *portp)
+       ipc_space_t             space,
+       ipc_port_init_flags_t   flags,
+       mach_port_name_t        *namep,
+       ipc_port_t              *portp)
 {
        ipc_port_t port;
        mach_port_name_t name;
        kern_return_t kr;
+       mach_port_type_t type = MACH_PORT_TYPE_RECEIVE;
+       mach_port_urefs_t urefs = 0;
 
-       kr = ipc_object_alloc(space, IOT_PORT,
-                             MACH_PORT_TYPE_RECEIVE, 0,
-                             &name, (ipc_object_t *) &port);
-       if (kr != KERN_SUCCESS)
+#if     MACH_ASSERT
+       uintptr_t buf[IP_CALLSTACK_MAX];
+       ipc_port_callstack_init_debug(&buf[0], IP_CALLSTACK_MAX);
+#endif /* MACH_ASSERT */
+
+       if (flags & IPC_PORT_INIT_MAKE_SEND_RIGHT) {
+               type |= MACH_PORT_TYPE_SEND;
+               urefs = 1;
+       }
+       kr = ipc_object_alloc(space, IOT_PORT, type, urefs,
+           &name, (ipc_object_t *) &port);
+       if (kr != KERN_SUCCESS) {
                return kr;
+       }
 
-       /* port is locked */
+       /* port and space are locked */
+       ipc_port_init(port, space, flags, name);
 
-       ipc_port_init(port, space, name);
+#if     MACH_ASSERT
+       ipc_port_init_debug(port, &buf[0], IP_CALLSTACK_MAX);
+#endif  /* MACH_ASSERT */
+
+       /* unlock space after init */
+       is_write_unlock(space);
 
        *namep = name;
        *portp = port;
@@ -520,22 +763,38 @@ ipc_port_alloc(
 
 kern_return_t
 ipc_port_alloc_name(
-       ipc_space_t             space,
-       mach_port_name_t        name,
-       ipc_port_t              *portp)
+       ipc_space_t             space,
+       ipc_port_init_flags_t   flags,
+       mach_port_name_t        name,
+       ipc_port_t              *portp)
 {
        ipc_port_t port;
        kern_return_t kr;
+       mach_port_type_t type = MACH_PORT_TYPE_RECEIVE;
+       mach_port_urefs_t urefs = 0;
+
+#if     MACH_ASSERT
+       uintptr_t buf[IP_CALLSTACK_MAX];
+       ipc_port_callstack_init_debug(&buf[0], IP_CALLSTACK_MAX);
+#endif /* MACH_ASSERT */
 
-       kr = ipc_object_alloc_name(space, IOT_PORT,
-                                  MACH_PORT_TYPE_RECEIVE, 0,
-                                  name, (ipc_object_t *) &port);
-       if (kr != KERN_SUCCESS)
+       if (flags & IPC_PORT_INIT_MAKE_SEND_RIGHT) {
+               type |= MACH_PORT_TYPE_SEND;
+               urefs = 1;
+       }
+       kr = ipc_object_alloc_name(space, IOT_PORT, type, urefs,
+           name, (ipc_object_t *) &port);
+       if (kr != KERN_SUCCESS) {
                return kr;
+       }
 
        /* port is locked */
 
-       ipc_port_init(port, space, name);
+       ipc_port_init(port, space, flags, name);
+
+#if     MACH_ASSERT
+       ipc_port_init_debug(port, &buf[0], IP_CALLSTACK_MAX);
+#endif  /* MACH_ASSERT */
 
        *portp = port;
 
@@ -543,37 +802,118 @@ ipc_port_alloc_name(
 }
 
 /*
- * Generate dead name notifications.  Called from ipc_port_destroy.
- * Port is unlocked but still has reference(s);
- * dnrequests was taken from port while the port
- * was locked but the port now has port->ip_dnrequests set to IPR_NULL.
+ *      Routine:       ipc_port_spnotify
+ *     Purpose:
+ *             Generate send-possible port notifications.
+ *     Conditions:
+ *             Nothing locked, reference held on port.
  */
 void
-ipc_port_dnnotify(
-       ipc_port_t              port,
-       ipc_port_request_t      dnrequests)
+ipc_port_spnotify(
+       ipc_port_t      port)
 {
-       ipc_table_size_t        its = dnrequests->ipr_size;
-       ipc_table_elems_t       size = its->its_size;
-       ipc_port_request_index_t index;
+       ipc_port_request_index_t index = 0;
+       ipc_table_elems_t size = 0;
+
+       /*
+        * If the port has no send-possible request
+        * armed, don't bother to lock the port.
+        */
+       if (port->ip_sprequests == 0) {
+               return;
+       }
+
+       ip_lock(port);
+
+#if IMPORTANCE_INHERITANCE
+       if (port->ip_spimportant != 0) {
+               port->ip_spimportant = 0;
+               if (ipc_port_importance_delta(port, IPID_OPTION_NORMAL, -1) == TRUE) {
+                       ip_lock(port);
+               }
+       }
+#endif /* IMPORTANCE_INHERITANCE */
+
+       if (port->ip_sprequests == 0) {
+               ip_unlock(port);
+               return;
+       }
+       port->ip_sprequests = 0;
+
+revalidate:
+       if (ip_active(port)) {
+               ipc_port_request_t requests;
+
+               /* table may change each time port unlocked (reload) */
+               requests = port->ip_requests;
+               assert(requests != IPR_NULL);
 
-       for (index = 1; index < size; index++) {
-               ipc_port_request_t      ipr = &dnrequests[index];
-               mach_port_name_t                name = ipr->ipr_name;
-               ipc_port_t              soright;
+               /*
+                * no need to go beyond table size when first
+                * we entered - those are future notifications.
+                */
+               if (size == 0) {
+                       size = requests->ipr_size->its_size;
+               }
+
+               /* no need to backtrack either */
+               while (++index < size) {
+                       ipc_port_request_t ipr = &requests[index];
+                       mach_port_name_t name = ipr->ipr_name;
+                       ipc_port_t soright = IPR_SOR_PORT(ipr->ipr_soright);
+                       boolean_t armed = IPR_SOR_SPARMED(ipr->ipr_soright);
 
-               if (name == MACH_PORT_NULL)
-                       continue;
+                       if (MACH_PORT_VALID(name) && armed && IP_VALID(soright)) {
+                               /* claim send-once right - slot still inuse */
+                               ipr->ipr_soright = IP_NULL;
+                               ip_unlock(port);
 
-               soright = ipr->ipr_soright;
-               assert(soright != IP_NULL);
+                               ipc_notify_send_possible(soright, name);
 
-               ipc_notify_dead_name(soright, name);
+                               ip_lock(port);
+                               goto revalidate;
+                       }
+               }
        }
+       ip_unlock(port);
+       return;
+}
 
-       it_dnrequests_free(its, dnrequests);
+/*
+ *      Routine:       ipc_port_dnnotify
+ *     Purpose:
+ *             Generate dead name notifications for
+ *             all outstanding dead-name and send-
+ *             possible requests.
+ *     Conditions:
+ *             Nothing locked.
+ *             Port must be inactive.
+ *             Reference held on port.
+ */
+void
+ipc_port_dnnotify(
+       ipc_port_t      port)
+{
+       ipc_port_request_t requests = port->ip_requests;
+
+       assert(!ip_active(port));
+       if (requests != IPR_NULL) {
+               ipc_table_size_t its = requests->ipr_size;
+               ipc_table_elems_t size = its->its_size;
+               ipc_port_request_index_t index;
+               for (index = 1; index < size; index++) {
+                       ipc_port_request_t ipr = &requests[index];
+                       mach_port_name_t name = ipr->ipr_name;
+                       ipc_port_t soright = IPR_SOR_PORT(ipr->ipr_soright);
+
+                       if (MACH_PORT_VALID(name) && IP_VALID(soright)) {
+                               ipc_notify_dead_name(soright, name);
+                       }
+               }
+       }
 }
 
+
 /*
  *     Routine:        ipc_port_destroy
  *     Purpose:
@@ -588,95 +928,190 @@ ipc_port_dnnotify(
  */
 
 void
-ipc_port_destroy(
-       ipc_port_t      port)
+ipc_port_destroy(ipc_port_t port)
 {
        ipc_port_t pdrequest, nsrequest;
        ipc_mqueue_t mqueue;
-       ipc_kmsg_queue_t kmqueue;
        ipc_kmsg_t kmsg;
-       ipc_port_request_t dnrequests;
+       boolean_t special_reply = port->ip_specialreply;
+       struct task_watchport_elem *watchport_elem = NULL;
 
-       assert(ip_active(port));
+#if IMPORTANCE_INHERITANCE
+       ipc_importance_task_t release_imp_task = IIT_NULL;
+       thread_t self = current_thread();
+       boolean_t top = (self->ith_assertions == 0);
+       natural_t assertcnt = 0;
+#endif /* IMPORTANCE_INHERITANCE */
+
+       require_ip_active(port);
        /* port->ip_receiver_name is garbage */
        /* port->ip_receiver/port->ip_destination is garbage */
-       assert(port->ip_pset_count == 0);
-       assert(port->ip_mscount == 0);
 
-       /* first check for a backup port */
+       /* clear any reply-port context */
+       port->ip_reply_context = 0;
 
+       /* check for a backup port */
        pdrequest = port->ip_pdrequest;
+
+#if IMPORTANCE_INHERITANCE
+       /* determine how many assertions to drop and from whom */
+       if (port->ip_tempowner != 0) {
+               assert(top);
+               release_imp_task = port->ip_imp_task;
+               if (IIT_NULL != release_imp_task) {
+                       port->ip_imp_task = IIT_NULL;
+                       assertcnt = port->ip_impcount;
+               }
+               /* Otherwise, nothing to drop */
+       } else {
+               assertcnt = port->ip_impcount;
+               if (pdrequest != IP_NULL) {
+                       /* mark in limbo for the journey */
+                       port->ip_tempowner = 1;
+               }
+       }
+
+       if (top) {
+               self->ith_assertions = assertcnt;
+       }
+#endif /* IMPORTANCE_INHERITANCE */
+
        if (pdrequest != IP_NULL) {
+               /* clear receiver, don't destroy the port */
+               (void)ipc_port_clear_receiver(port, FALSE);
+               assert(port->ip_in_pset == 0);
+               assert(port->ip_mscount == 0);
+
                /* we assume the ref for pdrequest */
                port->ip_pdrequest = IP_NULL;
 
-               /* make port be in limbo */
-               port->ip_receiver_name = MACH_PORT_NULL;
-               port->ip_destination = IP_NULL;
-               ip_unlock(port);
+               imq_lock(&port->ip_messages);
+               watchport_elem = ipc_port_clear_watchport_elem_internal(port);
+               ipc_port_send_turnstile_recompute_push_locked(port);
+               /* mqueue and port unlocked */
 
-               if (!ipc_port_check_circularity(port, pdrequest)) {
-                       /* consumes our refs for port and pdrequest */
-                       ipc_notify_port_destroyed(pdrequest, port);
-                       return;
-               } else {
-                       /* consume pdrequest and destroy port */
-                       ipc_port_release_sonce(pdrequest);
+               if (special_reply) {
+                       ipc_port_adjust_special_reply_port(port,
+                           IPC_PORT_ADJUST_SR_ALLOW_SYNC_LINKAGE);
                }
 
-               ip_lock(port);
-               assert(ip_active(port));
-               assert(port->ip_pset_count == 0);
-               assert(port->ip_mscount == 0);
-               assert(port->ip_pdrequest == IP_NULL);
-               assert(port->ip_receiver_name == MACH_PORT_NULL);
-               assert(port->ip_destination == IP_NULL);
+               if (watchport_elem) {
+                       task_watchport_elem_deallocate(watchport_elem);
+                       watchport_elem = NULL;
+               }
+               /* consumes our refs for port and pdrequest */
+               ipc_notify_port_destroyed(pdrequest, port);
 
-               /* fall through and destroy the port */
+               goto drop_assertions;
        }
 
-       /* once port is dead, we don't need to keep it locked */
-
-       port->ip_object.io_bits &= ~IO_BITS_ACTIVE;
-       port->ip_timestamp = ipc_port_timestamp();
+       /*
+        * The mach_msg_* paths don't hold a port lock, they only hold a
+        * reference to the port object. If a thread raced us and is now
+        * blocked waiting for message reception on this mqueue (or waiting
+        * for ipc_mqueue_full), it will never be woken up. We call
+        * ipc_port_clear_receiver() here, _after_ the port has been marked
+        * inactive, to wakeup any threads which may be blocked and ensure
+        * that no other thread can get lost waiting for a wake up on a
+        * port/mqueue that's been destroyed.
+        */
+       boolean_t reap_msgs = FALSE;
+       reap_msgs = ipc_port_clear_receiver(port, TRUE); /* marks port and mqueue inactive */
+       assert(port->ip_in_pset == 0);
+       assert(port->ip_mscount == 0);
 
-       /* save for later */
-       dnrequests = port->ip_dnrequests;
-       port->ip_dnrequests = IPR_NULL;
+       imq_lock(&port->ip_messages);
+       watchport_elem = ipc_port_clear_watchport_elem_internal(port);
+       imq_unlock(&port->ip_messages);
+       nsrequest = port->ip_nsrequest;
 
        /*
         * If the port has a preallocated message buffer and that buffer
         * is not inuse, free it.  If it has an inuse one, then the kmsg
         * free will detect that we freed the association and it can free it
         * like a normal buffer.
+        *
+        * Once the port is marked inactive we don't need to keep it locked.
         */
        if (IP_PREALLOC(port)) {
+               ipc_port_t inuse_port;
+
                kmsg = port->ip_premsg;
                assert(kmsg != IKM_NULL);
-               IP_CLEAR_PREALLOC(port, kmsg);
-               if (!ikm_prealloc_inuse(kmsg))
+               inuse_port = ikm_prealloc_inuse_port(kmsg);
+               ipc_kmsg_clear_prealloc(kmsg, port);
+
+               imq_lock(&port->ip_messages);
+               ipc_port_send_turnstile_recompute_push_locked(port);
+               /* mqueue and port unlocked */
+
+               if (inuse_port != IP_NULL) {
+                       assert(inuse_port == port);
+               } else {
                        ipc_kmsg_free(kmsg);
+               }
+       } else {
+               imq_lock(&port->ip_messages);
+               ipc_port_send_turnstile_recompute_push_locked(port);
+               /* mqueue and port unlocked */
        }
-       ip_unlock(port);
 
-       /* throw away no-senders request */
+       /* Deallocate the watchport element */
+       if (watchport_elem) {
+               task_watchport_elem_deallocate(watchport_elem);
+               watchport_elem = NULL;
+       }
 
-       nsrequest = port->ip_nsrequest;
-       if (nsrequest != IP_NULL)
+       /* unlink the kmsg from special reply port */
+       if (special_reply) {
+               ipc_port_adjust_special_reply_port(port,
+                   IPC_PORT_ADJUST_SR_ALLOW_SYNC_LINKAGE);
+       }
+
+       /* throw away no-senders request */
+       if (nsrequest != IP_NULL) {
                ipc_notify_send_once(nsrequest); /* consumes ref */
+       }
+       /*
+        * Reap any kmsg objects waiting to be destroyed.
+        * This must be done after we've released the port lock.
+        */
+       if (reap_msgs) {
+               ipc_kmsg_reap_delayed();
+       }
 
-       /* destroy any queued messages */
        mqueue = &port->ip_messages;
-       ipc_mqueue_destroy(mqueue);
+
+       /* cleanup waitq related resources */
+       ipc_mqueue_deinit(mqueue);
 
        /* generate dead-name notifications */
-       if (dnrequests != IPR_NULL) {
-               ipc_port_dnnotify(port, dnrequests);
-       }
+       ipc_port_dnnotify(port);
 
        ipc_kobject_destroy(port);
 
-       ipc_port_release(port); /* consume caller's ref */
+       ip_release(port); /* consume caller's ref */
+
+drop_assertions:
+#if IMPORTANCE_INHERITANCE
+       if (release_imp_task != IIT_NULL) {
+               if (assertcnt > 0) {
+                       assert(top);
+                       self->ith_assertions = 0;
+                       assert(ipc_importance_task_is_any_receiver_type(release_imp_task));
+                       ipc_importance_task_drop_internal_assertion(release_imp_task, assertcnt);
+               }
+               ipc_importance_task_release(release_imp_task);
+       } else if (assertcnt > 0) {
+               if (top) {
+                       self->ith_assertions = 0;
+                       release_imp_task = current_task()->task_imp_base;
+                       if (ipc_importance_task_is_any_receiver_type(release_imp_task)) {
+                               ipc_importance_task_drop_internal_assertion(release_imp_task, assertcnt);
+                       }
+               }
+       }
+#endif /* IMPORTANCE_INHERITANCE */
 }
 
 /*
@@ -691,35 +1126,45 @@ ipc_port_destroy(
  *             That is, we want to set port->ip_destination == dest,
  *             but guaranteeing that this doesn't create a circle
  *             port->ip_destination->ip_destination->... == port
+ *
  *     Conditions:
  *             No ports locked.  References held for "port" and "dest".
  */
 
 boolean_t
 ipc_port_check_circularity(
-       ipc_port_t      port,
-       ipc_port_t      dest)
+       ipc_port_t      port,
+       ipc_port_t      dest)
 {
+#if IMPORTANCE_INHERITANCE
+       /* adjust importance counts at the same time */
+       return ipc_importance_check_circularity(port, dest);
+#else
        ipc_port_t base;
+       struct task_watchport_elem *watchport_elem = NULL;
 
        assert(port != IP_NULL);
        assert(dest != IP_NULL);
 
-       if (port == dest)
+       if (port == dest) {
                return TRUE;
+       }
        base = dest;
 
+       /* Check if destination needs a turnstile */
+       ipc_port_send_turnstile_prepare(dest);
+
        /*
         *      First try a quick check that can run in parallel.
         *      No circularity if dest is not in transit.
         */
-
        ip_lock(port);
        if (ip_lock_try(dest)) {
                if (!ip_active(dest) ||
                    (dest->ip_receiver_name != MACH_PORT_NULL) ||
-                   (dest->ip_destination == IP_NULL))
+                   (dest->ip_destination == IP_NULL)) {
                        goto not_circular;
+               }
 
                /* dest is in transit; further checking necessary */
 
@@ -739,8 +1184,9 @@ ipc_port_check_circularity(
 
                if (!ip_active(base) ||
                    (base->ip_receiver_name != MACH_PORT_NULL) ||
-                   (base->ip_destination == IP_NULL))
+                   (base->ip_destination == IP_NULL)) {
                        break;
+               }
 
                base = base->ip_destination;
        }
@@ -753,24 +1199,24 @@ ipc_port_check_circularity(
                ipc_port_multiple_unlock();
 
                /* port (== base) is in limbo */
-
-               assert(ip_active(port));
+               require_ip_active(port);
                assert(port->ip_receiver_name == MACH_PORT_NULL);
                assert(port->ip_destination == IP_NULL);
 
-               while (dest != IP_NULL) {
+               base = dest;
+               while (base != IP_NULL) {
                        ipc_port_t next;
 
                        /* dest is in transit or in limbo */
+                       require_ip_active(base);
+                       assert(base->ip_receiver_name == MACH_PORT_NULL);
 
-                       assert(ip_active(dest));
-                       assert(dest->ip_receiver_name == MACH_PORT_NULL);
-
-                       next = dest->ip_destination;
-                       ip_unlock(dest);
-                       dest = next;
+                       next = base->ip_destination;
+                       ip_unlock(base);
+                       base = next;
                }
 
+               ipc_port_send_turnstile_complete(dest);
                return TRUE;
        }
 
@@ -783,86 +1229,1172 @@ ipc_port_check_circularity(
        ip_lock(port);
        ipc_port_multiple_unlock();
 
-    not_circular:
+not_circular:
+       imq_lock(&port->ip_messages);
 
        /* port is in limbo */
-
-       assert(ip_active(port));
+       require_ip_active(port);
        assert(port->ip_receiver_name == MACH_PORT_NULL);
        assert(port->ip_destination == IP_NULL);
 
+       /* Clear the watchport boost */
+       watchport_elem = ipc_port_clear_watchport_elem_internal(port);
+
+       /* Check if the port is being enqueued as a part of sync bootstrap checkin */
+       if (dest->ip_specialreply && dest->ip_sync_bootstrap_checkin) {
+               port->ip_sync_bootstrap_checkin = 1;
+       }
+
        ip_reference(dest);
        port->ip_destination = dest;
 
-       /* now unlock chain */
-
-       while (port != base) {
-               ipc_port_t next;
+       /* Setup linkage for source port if it has sync ipc push */
+       struct turnstile *send_turnstile = TURNSTILE_NULL;
+       if (port_send_turnstile(port)) {
+               send_turnstile = turnstile_prepare((uintptr_t)port,
+                   port_send_turnstile_address(port),
+                   TURNSTILE_NULL, TURNSTILE_SYNC_IPC);
 
-               /* port is in transit */
+               /*
+                * What ipc_port_adjust_port_locked would do,
+                * but we need to also drop even more locks before
+                * calling turnstile_update_inheritor_complete().
+                */
+               ipc_port_adjust_sync_link_state_locked(port, PORT_SYNC_LINK_ANY, NULL);
 
-               assert(ip_active(port));
-               assert(port->ip_receiver_name == MACH_PORT_NULL);
-               assert(port->ip_destination != IP_NULL);
+               turnstile_update_inheritor(send_turnstile, port_send_turnstile(dest),
+                   (TURNSTILE_INHERITOR_TURNSTILE | TURNSTILE_IMMEDIATE_UPDATE));
 
-               next = port->ip_destination;
-               ip_unlock(port);
-               port = next;
+               /* update complete and turnstile complete called after dropping all locks */
        }
+       imq_unlock(&port->ip_messages);
 
-       /* base is not in transit */
+       /* now unlock chain */
 
-       assert(!ip_active(base) ||
-              (base->ip_receiver_name != MACH_PORT_NULL) ||
-              (base->ip_destination == IP_NULL));
-       ip_unlock(base);
+       ip_unlock(port);
 
-       return FALSE;
+       for (;;) {
+               ipc_port_t next;
+
+               if (dest == base) {
+                       break;
+               }
+
+               /* port is in transit */
+               require_ip_active(dest);
+               assert(dest->ip_receiver_name == MACH_PORT_NULL);
+               assert(dest->ip_destination != IP_NULL);
+
+               next = dest->ip_destination;
+               ip_unlock(dest);
+               dest = next;
+       }
+
+       /* base is not in transit */
+       assert(!ip_active(base) ||
+           (base->ip_receiver_name != MACH_PORT_NULL) ||
+           (base->ip_destination == IP_NULL));
+
+       ip_unlock(base);
+
+       /* All locks dropped, call turnstile_update_inheritor_complete for source port's turnstile */
+       if (send_turnstile) {
+               turnstile_update_inheritor_complete(send_turnstile, TURNSTILE_INTERLOCK_NOT_HELD);
+
+               /* Take the mq lock to call turnstile complete */
+               imq_lock(&port->ip_messages);
+               turnstile_complete((uintptr_t)port, port_send_turnstile_address(port), NULL, TURNSTILE_SYNC_IPC);
+               send_turnstile = TURNSTILE_NULL;
+               imq_unlock(&port->ip_messages);
+               turnstile_cleanup();
+       }
+
+       if (watchport_elem) {
+               task_watchport_elem_deallocate(watchport_elem);
+       }
+
+       return FALSE;
+#endif /* !IMPORTANCE_INHERITANCE */
 }
 
 /*
- *     Routine:        ipc_port_lookup_notify
+ *     Routine:        ipc_port_watchport_elem
  *     Purpose:
- *             Make a send-once notify port from a receive right.
- *             Returns IP_NULL if name doesn't denote a receive right.
+ *             Get the port's watchport elem field
+ *
  *     Conditions:
- *             The space must be locked (read or write) and active.
- *             Being the active space, we can rely on thread server_id
- *             context to give us the proper server level sub-order
- *             within the space.
+ *             mqueue locked
  */
+static struct task_watchport_elem *
+ipc_port_watchport_elem(ipc_port_t port)
+{
+       return port->ip_messages.imq_wait_queue.waitq_tspriv;
+}
 
-ipc_port_t
-ipc_port_lookup_notify(
-       ipc_space_t             space,
-       mach_port_name_t        name)
+/*
+ *     Routine:        ipc_port_update_watchport_elem
+ *     Purpose:
+ *             Set the port's watchport elem field
+ *
+ *     Conditions:
+ *             mqueue locked
+ */
+static inline struct task_watchport_elem *
+ipc_port_update_watchport_elem(ipc_port_t port, struct task_watchport_elem *we)
 {
-       ipc_port_t port;
-       ipc_entry_t entry;
+       struct task_watchport_elem *old_we = ipc_port_watchport_elem(port);
+       port->ip_messages.imq_wait_queue.waitq_tspriv = we;
+       return old_we;
+}
 
-       assert(space->is_active);
+/*
+ * Update the recv turnstile inheritor for a port.
+ *
+ * Sync IPC through the port receive turnstile only happens for the special
+ * reply port case. It has three sub-cases:
+ *
+ * 1. a send-once right is in transit, and pushes on the send turnstile of its
+ *    destination mqueue.
+ *
+ * 2. a send-once right has been stashed on a knote it was copied out "through",
+ *    as the first such copied out port.
+ *
+ * 3. a send-once right has been stashed on a knote it was copied out "through",
+ *    as the second or more copied out port.
+ */
+void
+ipc_port_recv_update_inheritor(
+       ipc_port_t port,
+       struct turnstile *rcv_turnstile,
+       turnstile_update_flags_t flags)
+{
+       struct turnstile *inheritor = TURNSTILE_NULL;
+       struct knote *kn;
 
-       entry = ipc_entry_lookup(space, name);
-       if (entry == IE_NULL)
-               return IP_NULL;
-       if ((entry->ie_bits & MACH_PORT_TYPE_RECEIVE) == 0)
-               return IP_NULL;
+       if (ip_active(port) && port->ip_specialreply) {
+               imq_held(&port->ip_messages);
 
-       port = (ipc_port_t) entry->ie_object;
-       assert(port != IP_NULL);
+               switch (port->ip_sync_link_state) {
+               case PORT_SYNC_LINK_PORT:
+                       if (port->ip_sync_inheritor_port != NULL) {
+                               inheritor = port_send_turnstile(port->ip_sync_inheritor_port);
+                       }
+                       break;
 
-       ip_lock(port);
-       assert(ip_active(port));
-       assert(port->ip_receiver_name == name);
-       assert(port->ip_receiver == space);
+               case PORT_SYNC_LINK_WORKLOOP_KNOTE:
+                       kn = port->ip_sync_inheritor_knote;
+                       inheritor = filt_ipc_kqueue_turnstile(kn);
+                       break;
 
-       ip_reference(port);
-       port->ip_sorights++;
+               case PORT_SYNC_LINK_WORKLOOP_STASH:
+                       inheritor = port->ip_sync_inheritor_ts;
+                       break;
+               }
+       }
+
+       turnstile_update_inheritor(rcv_turnstile, inheritor,
+           flags | TURNSTILE_INHERITOR_TURNSTILE);
+}
+
+/*
+ * Update the send turnstile inheritor for a port.
+ *
+ * Sync IPC through the port send turnstile has 7 possible reasons to be linked:
+ *
+ * 1. a special reply port is part of sync ipc for bootstrap checkin and needs
+ *    to push on thread doing the sync ipc.
+ *
+ * 2. a receive right is in transit, and pushes on the send turnstile of its
+ *    destination mqueue.
+ *
+ * 3. port was passed as an exec watchport and port is pushing on main thread
+ *    of the task.
+ *
+ * 4. a receive right has been stashed on a knote it was copied out "through",
+ *    as the first such copied out port (same as PORT_SYNC_LINK_WORKLOOP_KNOTE
+ *    for the special reply port)
+ *
+ * 5. a receive right has been stashed on a knote it was copied out "through",
+ *    as the second or more copied out port (same as
+ *    PORT_SYNC_LINK_WORKLOOP_STASH for the special reply port)
+ *
+ * 6. a receive right has been copied out as a part of sync bootstrap checkin
+ *    and needs to push on thread doing the sync bootstrap checkin.
+ *
+ * 7. the receive right is monitored by a knote, and pushes on any that is
+ *    registered on a workloop. filt_machport makes sure that if such a knote
+ *    exists, it is kept as the first item in the knote list, so we never need
+ *    to walk.
+ */
+void
+ipc_port_send_update_inheritor(
+       ipc_port_t port,
+       struct turnstile *send_turnstile,
+       turnstile_update_flags_t flags)
+{
+       ipc_mqueue_t mqueue = &port->ip_messages;
+       turnstile_inheritor_t inheritor = TURNSTILE_INHERITOR_NULL;
+       struct knote *kn;
+       turnstile_update_flags_t inheritor_flags = TURNSTILE_INHERITOR_TURNSTILE;
+
+       assert(imq_held(mqueue));
+
+       if (!ip_active(port)) {
+               /* this port is no longer active, it should not push anywhere */
+       } else if (port->ip_specialreply) {
+               /* Case 1. */
+               if (port->ip_sync_bootstrap_checkin && prioritize_launch) {
+                       inheritor = port->ip_messages.imq_srp_owner_thread;
+                       inheritor_flags = TURNSTILE_INHERITOR_THREAD;
+               }
+       } else if (port->ip_receiver_name == MACH_PORT_NULL &&
+           port->ip_destination != NULL) {
+               /* Case 2. */
+               inheritor = port_send_turnstile(port->ip_destination);
+       } else if (ipc_port_watchport_elem(port) != NULL) {
+               /* Case 3. */
+               if (prioritize_launch) {
+                       assert(port->ip_sync_link_state == PORT_SYNC_LINK_ANY);
+                       inheritor = ipc_port_get_watchport_inheritor(port);
+                       inheritor_flags = TURNSTILE_INHERITOR_THREAD;
+               }
+       } else if (port->ip_sync_link_state == PORT_SYNC_LINK_WORKLOOP_KNOTE) {
+               /* Case 4. */
+               inheritor = filt_ipc_kqueue_turnstile(mqueue->imq_inheritor_knote);
+       } else if (port->ip_sync_link_state == PORT_SYNC_LINK_WORKLOOP_STASH) {
+               /* Case 5. */
+               inheritor = mqueue->imq_inheritor_turnstile;
+       } else if (port->ip_sync_link_state == PORT_SYNC_LINK_RCV_THREAD) {
+               /* Case 6. */
+               if (prioritize_launch) {
+                       inheritor = port->ip_messages.imq_inheritor_thread_ref;
+                       inheritor_flags = TURNSTILE_INHERITOR_THREAD;
+               }
+       } else if ((kn = SLIST_FIRST(&mqueue->imq_klist))) {
+               /* Case 7. Push on a workloop that is interested */
+               if (filt_machport_kqueue_has_turnstile(kn)) {
+                       assert(port->ip_sync_link_state == PORT_SYNC_LINK_ANY);
+                       inheritor = filt_ipc_kqueue_turnstile(kn);
+               }
+       }
+
+       turnstile_update_inheritor(send_turnstile, inheritor,
+           flags | inheritor_flags);
+}
+
+/*
+ *     Routine:        ipc_port_send_turnstile_prepare
+ *     Purpose:
+ *             Get a reference on port's send turnstile, if
+ *             port does not have a send turnstile then allocate one.
+ *
+ *     Conditions:
+ *             Nothing is locked.
+ */
+void
+ipc_port_send_turnstile_prepare(ipc_port_t port)
+{
+       struct turnstile *turnstile = TURNSTILE_NULL;
+       struct turnstile *send_turnstile = TURNSTILE_NULL;
+
+retry_alloc:
+       imq_lock(&port->ip_messages);
+
+       if (port_send_turnstile(port) == NULL ||
+           port_send_turnstile(port)->ts_port_ref == 0) {
+               if (turnstile == TURNSTILE_NULL) {
+                       imq_unlock(&port->ip_messages);
+                       turnstile = turnstile_alloc();
+                       goto retry_alloc;
+               }
+
+               send_turnstile = turnstile_prepare((uintptr_t)port,
+                   port_send_turnstile_address(port),
+                   turnstile, TURNSTILE_SYNC_IPC);
+               turnstile = TURNSTILE_NULL;
+
+               ipc_port_send_update_inheritor(port, send_turnstile,
+                   TURNSTILE_IMMEDIATE_UPDATE);
+
+               /* turnstile complete will be called in ipc_port_send_turnstile_complete */
+       }
+
+       /* Increment turnstile counter */
+       port_send_turnstile(port)->ts_port_ref++;
+       imq_unlock(&port->ip_messages);
+
+       if (send_turnstile) {
+               turnstile_update_inheritor_complete(send_turnstile,
+                   TURNSTILE_INTERLOCK_NOT_HELD);
+       }
+       if (turnstile != TURNSTILE_NULL) {
+               turnstile_deallocate(turnstile);
+       }
+}
+
+
+/*
+ *     Routine:        ipc_port_send_turnstile_complete
+ *     Purpose:
+ *             Drop a ref on the port's send turnstile, if the
+ *             ref becomes zero, deallocate the turnstile.
+ *
+ *     Conditions:
+ *             The space might be locked, use safe deallocate.
+ */
+void
+ipc_port_send_turnstile_complete(ipc_port_t port)
+{
+       struct turnstile *turnstile = TURNSTILE_NULL;
+
+       /* Drop turnstile count on dest port */
+       imq_lock(&port->ip_messages);
+
+       port_send_turnstile(port)->ts_port_ref--;
+       if (port_send_turnstile(port)->ts_port_ref == 0) {
+               turnstile_complete((uintptr_t)port, port_send_turnstile_address(port),
+                   &turnstile, TURNSTILE_SYNC_IPC);
+               assert(turnstile != TURNSTILE_NULL);
+       }
+       imq_unlock(&port->ip_messages);
+       turnstile_cleanup();
+
+       if (turnstile != TURNSTILE_NULL) {
+               turnstile_deallocate_safe(turnstile);
+               turnstile = TURNSTILE_NULL;
+       }
+}
+
+/*
+ *     Routine:        ipc_port_rcv_turnstile
+ *     Purpose:
+ *             Get the port's receive turnstile
+ *
+ *     Conditions:
+ *             mqueue locked or thread waiting on turnstile is locked.
+ */
+static struct turnstile *
+ipc_port_rcv_turnstile(ipc_port_t port)
+{
+       return *port_rcv_turnstile_address(port);
+}
+
+
+/*
+ *     Routine:        ipc_port_link_special_reply_port
+ *     Purpose:
+ *             Link the special reply port with the destination port.
+ *              Allocates turnstile to dest port.
+ *
+ *     Conditions:
+ *             Nothing is locked.
+ */
+void
+ipc_port_link_special_reply_port(
+       ipc_port_t special_reply_port,
+       ipc_port_t dest_port,
+       boolean_t sync_bootstrap_checkin)
+{
+       boolean_t drop_turnstile_ref = FALSE;
+
+       /* Check if dest_port needs a turnstile */
+       ipc_port_send_turnstile_prepare(dest_port);
+
+       /* Lock the special reply port and establish the linkage */
+       ip_lock(special_reply_port);
+       imq_lock(&special_reply_port->ip_messages);
+
+       if (sync_bootstrap_checkin && special_reply_port->ip_specialreply) {
+               special_reply_port->ip_sync_bootstrap_checkin = 1;
+       }
+
+       /* Check if we need to drop the acquired turnstile ref on dest port */
+       if (!special_reply_port->ip_specialreply ||
+           special_reply_port->ip_sync_link_state != PORT_SYNC_LINK_ANY ||
+           special_reply_port->ip_sync_inheritor_port != IPC_PORT_NULL) {
+               drop_turnstile_ref = TRUE;
+       } else {
+               /* take a reference on dest_port */
+               ip_reference(dest_port);
+               special_reply_port->ip_sync_inheritor_port = dest_port;
+               special_reply_port->ip_sync_link_state = PORT_SYNC_LINK_PORT;
+       }
+
+       imq_unlock(&special_reply_port->ip_messages);
+       ip_unlock(special_reply_port);
+
+       if (drop_turnstile_ref) {
+               ipc_port_send_turnstile_complete(dest_port);
+       }
+
+       return;
+}
+
+#if DEVELOPMENT || DEBUG
+inline void
+ipc_special_reply_port_bits_reset(ipc_port_t special_reply_port)
+{
+       special_reply_port->ip_srp_lost_link = 0;
+       special_reply_port->ip_srp_msg_sent = 0;
+}
+
+static inline void
+ipc_special_reply_port_msg_sent_reset(ipc_port_t special_reply_port)
+{
+       if (special_reply_port->ip_specialreply == 1) {
+               special_reply_port->ip_srp_msg_sent = 0;
+       }
+}
+
+inline void
+ipc_special_reply_port_msg_sent(ipc_port_t special_reply_port)
+{
+       if (special_reply_port->ip_specialreply == 1) {
+               special_reply_port->ip_srp_msg_sent = 1;
+       }
+}
+
+static inline void
+ipc_special_reply_port_lost_link(ipc_port_t special_reply_port)
+{
+       if (special_reply_port->ip_specialreply == 1 && special_reply_port->ip_srp_msg_sent == 0) {
+               special_reply_port->ip_srp_lost_link = 1;
+       }
+}
+
+#else /* DEVELOPMENT || DEBUG */
+inline void
+ipc_special_reply_port_bits_reset(__unused ipc_port_t special_reply_port)
+{
+       return;
+}
+
+static inline void
+ipc_special_reply_port_msg_sent_reset(__unused ipc_port_t special_reply_port)
+{
+       return;
+}
+
+inline void
+ipc_special_reply_port_msg_sent(__unused ipc_port_t special_reply_port)
+{
+       return;
+}
+
+static inline void
+ipc_special_reply_port_lost_link(__unused ipc_port_t special_reply_port)
+{
+       return;
+}
+#endif /* DEVELOPMENT || DEBUG */
+
+/*
+ *     Routine:        ipc_port_adjust_special_reply_port_locked
+ *     Purpose:
+ *             If the special port has a turnstile, update its inheritor.
+ *     Condition:
+ *             Special reply port locked on entry.
+ *             Special reply port unlocked on return.
+ *             The passed in port is a special reply port.
+ *     Returns:
+ *             None.
+ */
+void
+ipc_port_adjust_special_reply_port_locked(
+       ipc_port_t special_reply_port,
+       struct knote *kn,
+       uint8_t flags,
+       boolean_t get_turnstile)
+{
+       ipc_port_t dest_port = IPC_PORT_NULL;
+       int sync_link_state = PORT_SYNC_LINK_NO_LINKAGE;
+       turnstile_inheritor_t inheritor = TURNSTILE_INHERITOR_NULL;
+       struct turnstile *ts = TURNSTILE_NULL;
+
+       ip_lock_held(special_reply_port); // ip_sync_link_state is touched
+       imq_lock(&special_reply_port->ip_messages);
+
+       if (!special_reply_port->ip_specialreply) {
+               // only mach_msg_receive_results_complete() calls this with any port
+               assert(get_turnstile);
+               goto not_special;
+       }
+
+       if (flags & IPC_PORT_ADJUST_SR_RECEIVED_MSG) {
+               ipc_special_reply_port_msg_sent_reset(special_reply_port);
+       }
+
+       if (flags & IPC_PORT_ADJUST_UNLINK_THREAD) {
+               special_reply_port->ip_messages.imq_srp_owner_thread = NULL;
+       }
+
+       if (flags & IPC_PORT_ADJUST_RESET_BOOSTRAP_CHECKIN) {
+               special_reply_port->ip_sync_bootstrap_checkin = 0;
+       }
+
+       /* Check if the special reply port is marked non-special */
+       if (special_reply_port->ip_sync_link_state == PORT_SYNC_LINK_ANY) {
+not_special:
+               if (get_turnstile) {
+                       turnstile_complete((uintptr_t)special_reply_port,
+                           port_rcv_turnstile_address(special_reply_port), NULL, TURNSTILE_SYNC_IPC);
+               }
+               imq_unlock(&special_reply_port->ip_messages);
+               ip_unlock(special_reply_port);
+               if (get_turnstile) {
+                       turnstile_cleanup();
+               }
+               return;
+       }
+
+       if (flags & IPC_PORT_ADJUST_SR_LINK_WORKLOOP) {
+               if (ITH_KNOTE_VALID(kn, MACH_MSG_TYPE_PORT_SEND_ONCE)) {
+                       inheritor = filt_machport_stash_port(kn, special_reply_port,
+                           &sync_link_state);
+               }
+       } else if (flags & IPC_PORT_ADJUST_SR_ALLOW_SYNC_LINKAGE) {
+               sync_link_state = PORT_SYNC_LINK_ANY;
+       }
+
+       /* Check if need to break linkage */
+       if (!get_turnstile && sync_link_state == PORT_SYNC_LINK_NO_LINKAGE &&
+           special_reply_port->ip_sync_link_state == PORT_SYNC_LINK_NO_LINKAGE) {
+               imq_unlock(&special_reply_port->ip_messages);
+               ip_unlock(special_reply_port);
+               return;
+       }
+
+       switch (special_reply_port->ip_sync_link_state) {
+       case PORT_SYNC_LINK_PORT:
+               dest_port = special_reply_port->ip_sync_inheritor_port;
+               special_reply_port->ip_sync_inheritor_port = IPC_PORT_NULL;
+               break;
+       case PORT_SYNC_LINK_WORKLOOP_KNOTE:
+               special_reply_port->ip_sync_inheritor_knote = NULL;
+               break;
+       case PORT_SYNC_LINK_WORKLOOP_STASH:
+               special_reply_port->ip_sync_inheritor_ts = NULL;
+               break;
+       }
+
+       special_reply_port->ip_sync_link_state = sync_link_state;
+
+       switch (sync_link_state) {
+       case PORT_SYNC_LINK_WORKLOOP_KNOTE:
+               special_reply_port->ip_sync_inheritor_knote = kn;
+               break;
+       case PORT_SYNC_LINK_WORKLOOP_STASH:
+               special_reply_port->ip_sync_inheritor_ts = inheritor;
+               break;
+       case PORT_SYNC_LINK_NO_LINKAGE:
+               if (flags & IPC_PORT_ADJUST_SR_ENABLE_EVENT) {
+                       ipc_special_reply_port_lost_link(special_reply_port);
+               }
+               break;
+       }
+
+       /* Get thread's turnstile donated to special reply port */
+       if (get_turnstile) {
+               turnstile_complete((uintptr_t)special_reply_port,
+                   port_rcv_turnstile_address(special_reply_port), NULL, TURNSTILE_SYNC_IPC);
+       } else {
+               ts = ipc_port_rcv_turnstile(special_reply_port);
+               if (ts) {
+                       turnstile_reference(ts);
+                       ipc_port_recv_update_inheritor(special_reply_port, ts,
+                           TURNSTILE_IMMEDIATE_UPDATE);
+               }
+       }
+
+       imq_unlock(&special_reply_port->ip_messages);
+       ip_unlock(special_reply_port);
+
+       if (get_turnstile) {
+               turnstile_cleanup();
+       } else if (ts) {
+               /* Call turnstile cleanup after dropping the interlock */
+               turnstile_update_inheritor_complete(ts, TURNSTILE_INTERLOCK_NOT_HELD);
+               turnstile_deallocate_safe(ts);
+       }
+
+       /* Release the ref on the dest port and its turnstile */
+       if (dest_port) {
+               ipc_port_send_turnstile_complete(dest_port);
+               /* release the reference on the dest port */
+               ip_release(dest_port);
+       }
+}
+
+/*
+ *     Routine:        ipc_port_adjust_special_reply_port
+ *     Purpose:
+ *             If the special port has a turnstile, update its inheritor.
+ *     Condition:
+ *             Nothing locked.
+ *     Returns:
+ *             None.
+ */
+void
+ipc_port_adjust_special_reply_port(
+       ipc_port_t port,
+       uint8_t flags)
+{
+       if (port->ip_specialreply) {
+               ip_lock(port);
+               ipc_port_adjust_special_reply_port_locked(port, NULL, flags, FALSE);
+       }
+}
+
+/*
+ *     Routine:        ipc_port_adjust_sync_link_state_locked
+ *     Purpose:
+ *             Update the sync link state of the port and the
+ *             turnstile inheritor.
+ *     Condition:
+ *             Port and mqueue locked on entry.
+ *             Port and mqueue locked on return.
+ *     Returns:
+ *              None.
+ */
+void
+ipc_port_adjust_sync_link_state_locked(
+       ipc_port_t port,
+       int sync_link_state,
+       turnstile_inheritor_t inheritor)
+{
+       switch (port->ip_sync_link_state) {
+       case PORT_SYNC_LINK_RCV_THREAD:
+               /* deallocate the thread reference for the inheritor */
+               thread_deallocate_safe(port->ip_messages.imq_inheritor_thread_ref);
+       /* Fall through */
+
+       default:
+               klist_init(&port->ip_messages.imq_klist);
+       }
+
+       switch (sync_link_state) {
+       case PORT_SYNC_LINK_WORKLOOP_KNOTE:
+               port->ip_messages.imq_inheritor_knote = inheritor;
+               break;
+       case PORT_SYNC_LINK_WORKLOOP_STASH:
+               port->ip_messages.imq_inheritor_turnstile = inheritor;
+               break;
+       case PORT_SYNC_LINK_RCV_THREAD:
+               /* The thread could exit without clearing port state, take a thread ref */
+               thread_reference((thread_t)inheritor);
+               port->ip_messages.imq_inheritor_thread_ref = inheritor;
+               break;
+       default:
+               klist_init(&port->ip_messages.imq_klist);
+               sync_link_state = PORT_SYNC_LINK_ANY;
+       }
+
+       port->ip_sync_link_state = sync_link_state;
+}
+
+
+/*
+ *     Routine:        ipc_port_adjust_port_locked
+ *     Purpose:
+ *             If the port has a turnstile, update its inheritor.
+ *     Condition:
+ *             Port locked on entry.
+ *             Port unlocked on return.
+ *     Returns:
+ *             None.
+ */
+void
+ipc_port_adjust_port_locked(
+       ipc_port_t port,
+       struct knote *kn,
+       boolean_t sync_bootstrap_checkin)
+{
+       int sync_link_state = PORT_SYNC_LINK_ANY;
+       turnstile_inheritor_t inheritor = TURNSTILE_INHERITOR_NULL;
+
+       ip_lock_held(port); // ip_sync_link_state is touched
+       imq_held(&port->ip_messages);
+
+       assert(!port->ip_specialreply);
+
+       if (kn) {
+               inheritor = filt_machport_stash_port(kn, port, &sync_link_state);
+               if (sync_link_state == PORT_SYNC_LINK_WORKLOOP_KNOTE) {
+                       inheritor = kn;
+               }
+       } else if (sync_bootstrap_checkin) {
+               inheritor = current_thread();
+               sync_link_state = PORT_SYNC_LINK_RCV_THREAD;
+       }
+
+       ipc_port_adjust_sync_link_state_locked(port, sync_link_state, inheritor);
+       port->ip_sync_bootstrap_checkin = 0;
+
+       ipc_port_send_turnstile_recompute_push_locked(port);
+       /* port and mqueue unlocked */
+}
+
+/*
+ *     Routine:        ipc_port_clear_sync_rcv_thread_boost_locked
+ *     Purpose:
+ *             If the port is pushing on rcv thread, clear it.
+ *     Condition:
+ *             Port locked on entry
+ *             mqueue is not locked.
+ *             Port unlocked on return.
+ *     Returns:
+ *             None.
+ */
+void
+ipc_port_clear_sync_rcv_thread_boost_locked(
+       ipc_port_t port)
+{
+       ip_lock_held(port); // ip_sync_link_state is touched
+
+       if (port->ip_sync_link_state != PORT_SYNC_LINK_RCV_THREAD) {
+               ip_unlock(port);
+               return;
+       }
+
+       imq_lock(&port->ip_messages);
+       ipc_port_adjust_sync_link_state_locked(port, PORT_SYNC_LINK_ANY, NULL);
+
+       ipc_port_send_turnstile_recompute_push_locked(port);
+       /* port and mqueue unlocked */
+}
+
+/*
+ *     Routine:        ipc_port_add_watchport_elem_locked
+ *     Purpose:
+ *             Transfer the turnstile boost of watchport to task calling exec.
+ *     Condition:
+ *             Port locked on entry.
+ *             Port unlocked on return.
+ *     Returns:
+ *             KERN_SUCESS on success.
+ *             KERN_FAILURE otherwise.
+ */
+kern_return_t
+ipc_port_add_watchport_elem_locked(
+       ipc_port_t                 port,
+       struct task_watchport_elem *watchport_elem,
+       struct task_watchport_elem **old_elem)
+{
+       ip_lock_held(port);
+       imq_held(&port->ip_messages);
+
+       /* Watchport boost only works for non-special active ports mapped in an ipc space */
+       if (!ip_active(port) || port->ip_specialreply ||
+           port->ip_receiver_name == MACH_PORT_NULL) {
+               imq_unlock(&port->ip_messages);
+               ip_unlock(port);
+               return KERN_FAILURE;
+       }
+
+       if (port->ip_sync_link_state != PORT_SYNC_LINK_ANY) {
+               /* Sever the linkage if the port was pushing on knote */
+               ipc_port_adjust_sync_link_state_locked(port, PORT_SYNC_LINK_ANY, NULL);
+       }
+
+       *old_elem = ipc_port_update_watchport_elem(port, watchport_elem);
+
+       ipc_port_send_turnstile_recompute_push_locked(port);
+       /* port and mqueue unlocked */
+       return KERN_SUCCESS;
+}
+
+/*
+ *     Routine:        ipc_port_clear_watchport_elem_internal_conditional_locked
+ *     Purpose:
+ *             Remove the turnstile boost of watchport and recompute the push.
+ *     Condition:
+ *             Port locked on entry.
+ *             Port unlocked on return.
+ *     Returns:
+ *             KERN_SUCESS on success.
+ *             KERN_FAILURE otherwise.
+ */
+kern_return_t
+ipc_port_clear_watchport_elem_internal_conditional_locked(
+       ipc_port_t                 port,
+       struct task_watchport_elem *watchport_elem)
+{
+       ip_lock_held(port);
+       imq_held(&port->ip_messages);
+
+       if (ipc_port_watchport_elem(port) != watchport_elem) {
+               imq_unlock(&port->ip_messages);
+               ip_unlock(port);
+               return KERN_FAILURE;
+       }
+
+       ipc_port_clear_watchport_elem_internal(port);
+       ipc_port_send_turnstile_recompute_push_locked(port);
+       /* port and mqueue unlocked */
+       return KERN_SUCCESS;
+}
+
+/*
+ *     Routine:        ipc_port_replace_watchport_elem_conditional_locked
+ *     Purpose:
+ *             Replace the turnstile boost of watchport and recompute the push.
+ *     Condition:
+ *             Port locked on entry.
+ *             Port unlocked on return.
+ *     Returns:
+ *             KERN_SUCESS on success.
+ *             KERN_FAILURE otherwise.
+ */
+kern_return_t
+ipc_port_replace_watchport_elem_conditional_locked(
+       ipc_port_t                 port,
+       struct task_watchport_elem *old_watchport_elem,
+       struct task_watchport_elem *new_watchport_elem)
+{
+       ip_lock_held(port);
+       imq_held(&port->ip_messages);
+
+       if (ipc_port_watchport_elem(port) != old_watchport_elem) {
+               imq_unlock(&port->ip_messages);
+               ip_unlock(port);
+               return KERN_FAILURE;
+       }
+
+       ipc_port_update_watchport_elem(port, new_watchport_elem);
+       ipc_port_send_turnstile_recompute_push_locked(port);
+       /* port and mqueue unlocked */
+       return KERN_SUCCESS;
+}
+
+/*
+ *     Routine:        ipc_port_clear_watchport_elem_internal
+ *     Purpose:
+ *             Remove the turnstile boost of watchport.
+ *     Condition:
+ *             Port locked on entry.
+ *             Port locked on return.
+ *     Returns:
+ *             Old task_watchport_elem returned.
+ */
+struct task_watchport_elem *
+ipc_port_clear_watchport_elem_internal(
+       ipc_port_t                 port)
+{
+       ip_lock_held(port);
+       imq_held(&port->ip_messages);
+
+       return ipc_port_update_watchport_elem(port, NULL);
+}
+
+/*
+ *     Routine:        ipc_port_send_turnstile_recompute_push_locked
+ *     Purpose:
+ *             Update send turnstile inheritor of port and recompute the push.
+ *     Condition:
+ *             Port locked on entry.
+ *             Port unlocked on return.
+ *     Returns:
+ *             None.
+ */
+static void
+ipc_port_send_turnstile_recompute_push_locked(
+       ipc_port_t port)
+{
+       struct turnstile *send_turnstile = port_send_turnstile(port);
+       if (send_turnstile) {
+               turnstile_reference(send_turnstile);
+               ipc_port_send_update_inheritor(port, send_turnstile,
+                   TURNSTILE_IMMEDIATE_UPDATE);
+       }
+       imq_unlock(&port->ip_messages);
        ip_unlock(port);
 
-       return port;
+       if (send_turnstile) {
+               turnstile_update_inheritor_complete(send_turnstile,
+                   TURNSTILE_INTERLOCK_NOT_HELD);
+               turnstile_deallocate_safe(send_turnstile);
+       }
 }
 
+/*
+ *     Routine:        ipc_port_get_watchport_inheritor
+ *     Purpose:
+ *             Returns inheritor for watchport.
+ *
+ *     Conditions:
+ *             mqueue locked.
+ *     Returns:
+ *             watchport inheritor.
+ */
+static thread_t
+ipc_port_get_watchport_inheritor(
+       ipc_port_t port)
+{
+       imq_held(&port->ip_messages);
+       return ipc_port_watchport_elem(port)->twe_task->watchports->tw_thread;
+}
+
+/*
+ *     Routine:        ipc_port_impcount_delta
+ *     Purpose:
+ *             Adjust only the importance count associated with a port.
+ *             If there are any adjustments to be made to receiver task,
+ *             those are handled elsewhere.
+ *
+ *             For now, be defensive during deductions to make sure the
+ *             impcount for the port doesn't underflow zero.  This will
+ *             go away when the port boost addition is made atomic (see
+ *             note in ipc_port_importance_delta()).
+ *     Conditions:
+ *             The port is referenced and locked.
+ *             Nothing else is locked.
+ */
+mach_port_delta_t
+ipc_port_impcount_delta(
+       ipc_port_t        port,
+       mach_port_delta_t delta,
+       ipc_port_t        __unused base)
+{
+       mach_port_delta_t absdelta;
+
+       if (!ip_active(port)) {
+               return 0;
+       }
+
+       /* adding/doing nothing is easy */
+       if (delta >= 0) {
+               port->ip_impcount += delta;
+               return delta;
+       }
+
+       absdelta = 0 - delta;
+       if (port->ip_impcount >= absdelta) {
+               port->ip_impcount -= absdelta;
+               return delta;
+       }
+
+#if (DEVELOPMENT || DEBUG)
+       if (port->ip_receiver_name != MACH_PORT_NULL) {
+               task_t target_task = port->ip_receiver->is_task;
+               ipc_importance_task_t target_imp = target_task->task_imp_base;
+               const char *target_procname;
+               int target_pid;
+
+               if (target_imp != IIT_NULL) {
+                       target_procname = target_imp->iit_procname;
+                       target_pid = target_imp->iit_bsd_pid;
+               } else {
+                       target_procname = "unknown";
+                       target_pid = -1;
+               }
+               printf("Over-release of importance assertions for port 0x%x receiver pid %d (%s), "
+                   "dropping %d assertion(s) but port only has %d remaining.\n",
+                   port->ip_receiver_name,
+                   target_pid, target_procname,
+                   absdelta, port->ip_impcount);
+       } else if (base != IP_NULL) {
+               task_t target_task = base->ip_receiver->is_task;
+               ipc_importance_task_t target_imp = target_task->task_imp_base;
+               const char *target_procname;
+               int target_pid;
+
+               if (target_imp != IIT_NULL) {
+                       target_procname = target_imp->iit_procname;
+                       target_pid = target_imp->iit_bsd_pid;
+               } else {
+                       target_procname = "unknown";
+                       target_pid = -1;
+               }
+               printf("Over-release of importance assertions for port 0x%lx "
+                   "enqueued on port 0x%x with receiver pid %d (%s), "
+                   "dropping %d assertion(s) but port only has %d remaining.\n",
+                   (unsigned long)VM_KERNEL_UNSLIDE_OR_PERM((uintptr_t)port),
+                   base->ip_receiver_name,
+                   target_pid, target_procname,
+                   absdelta, port->ip_impcount);
+       }
+#endif
+
+       delta = 0 - port->ip_impcount;
+       port->ip_impcount = 0;
+       return delta;
+}
+
+/*
+ *     Routine:        ipc_port_importance_delta_internal
+ *     Purpose:
+ *             Adjust the importance count through the given port.
+ *             If the port is in transit, apply the delta throughout
+ *             the chain. Determine if the there is a task at the
+ *             base of the chain that wants/needs to be adjusted,
+ *             and if so, apply the delta.
+ *     Conditions:
+ *             The port is referenced and locked on entry.
+ *             Importance may be locked.
+ *             Nothing else is locked.
+ *             The lock may be dropped on exit.
+ *             Returns TRUE if lock was dropped.
+ */
+#if IMPORTANCE_INHERITANCE
+
+boolean_t
+ipc_port_importance_delta_internal(
+       ipc_port_t              port,
+       natural_t               options,
+       mach_port_delta_t       *deltap,
+       ipc_importance_task_t   *imp_task)
+{
+       ipc_port_t next, base;
+       boolean_t dropped = FALSE;
+
+       *imp_task = IIT_NULL;
+
+       if (*deltap == 0) {
+               return FALSE;
+       }
+
+       assert(options == IPID_OPTION_NORMAL || options == IPID_OPTION_SENDPOSSIBLE);
+
+       base = port;
+
+       /* if port is in transit, have to search for end of chain */
+       if (ip_active(port) &&
+           port->ip_destination != IP_NULL &&
+           port->ip_receiver_name == MACH_PORT_NULL) {
+               dropped = TRUE;
+
+               ip_unlock(port);
+               ipc_port_multiple_lock(); /* massive serialization */
+               ip_lock(base);
+
+               while (ip_active(base) &&
+                   base->ip_destination != IP_NULL &&
+                   base->ip_receiver_name == MACH_PORT_NULL) {
+                       base = base->ip_destination;
+                       ip_lock(base);
+               }
+               ipc_port_multiple_unlock();
+       }
+
+       /*
+        * If the port lock is dropped b/c the port is in transit, there is a
+        * race window where another thread can drain messages and/or fire a
+        * send possible notification before we get here.
+        *
+        * We solve this race by checking to see if our caller armed the send
+        * possible notification, whether or not it's been fired yet, and
+        * whether or not we've already set the port's ip_spimportant bit. If
+        * we don't need a send-possible boost, then we'll just apply a
+        * harmless 0-boost to the port.
+        */
+       if (options & IPID_OPTION_SENDPOSSIBLE) {
+               assert(*deltap == 1);
+               if (port->ip_sprequests && port->ip_spimportant == 0) {
+                       port->ip_spimportant = 1;
+               } else {
+                       *deltap = 0;
+               }
+       }
+
+       /* unlock down to the base, adjusting boost(s) at each level */
+       for (;;) {
+               *deltap = ipc_port_impcount_delta(port, *deltap, base);
+
+               if (port == base) {
+                       break;
+               }
+
+               /* port is in transit */
+               assert(port->ip_tempowner == 0);
+               next = port->ip_destination;
+               ip_unlock(port);
+               port = next;
+       }
+
+       /* find the task (if any) to boost according to the base */
+       if (ip_active(base)) {
+               if (base->ip_tempowner != 0) {
+                       if (IIT_NULL != base->ip_imp_task) {
+                               *imp_task = base->ip_imp_task;
+                       }
+                       /* otherwise don't boost */
+               } else if (base->ip_receiver_name != MACH_PORT_NULL) {
+                       ipc_space_t space = base->ip_receiver;
+
+                       /* only spaces with boost-accepting tasks */
+                       if (space->is_task != TASK_NULL &&
+                           ipc_importance_task_is_any_receiver_type(space->is_task->task_imp_base)) {
+                               *imp_task = space->is_task->task_imp_base;
+                       }
+               }
+       }
+
+       /*
+        * Only the base is locked.  If we have to hold or drop task
+        * importance assertions, we'll have to drop that lock as well.
+        */
+       if (*imp_task != IIT_NULL) {
+               /* take a reference before unlocking base */
+               ipc_importance_task_reference(*imp_task);
+       }
+
+       if (dropped == TRUE) {
+               ip_unlock(base);
+       }
+
+       return dropped;
+}
+#endif /* IMPORTANCE_INHERITANCE */
+
+/*
+ *     Routine:        ipc_port_importance_delta
+ *     Purpose:
+ *             Adjust the importance count through the given port.
+ *             If the port is in transit, apply the delta throughout
+ *             the chain.
+ *
+ *             If there is a task at the base of the chain that wants/needs
+ *             to be adjusted, apply the delta.
+ *     Conditions:
+ *             The port is referenced and locked on entry.
+ *             Nothing else is locked.
+ *             The lock may be dropped on exit.
+ *             Returns TRUE if lock was dropped.
+ */
+#if IMPORTANCE_INHERITANCE
+
+boolean_t
+ipc_port_importance_delta(
+       ipc_port_t              port,
+       natural_t               options,
+       mach_port_delta_t       delta)
+{
+       ipc_importance_task_t imp_task = IIT_NULL;
+       boolean_t dropped;
+
+       dropped = ipc_port_importance_delta_internal(port, options, &delta, &imp_task);
+
+       if (IIT_NULL == imp_task || delta == 0) {
+               return dropped;
+       }
+
+       if (!dropped) {
+               ip_unlock(port);
+       }
+
+       assert(ipc_importance_task_is_any_receiver_type(imp_task));
+
+       if (delta > 0) {
+               ipc_importance_task_hold_internal_assertion(imp_task, delta);
+       } else {
+               ipc_importance_task_drop_internal_assertion(imp_task, -delta);
+       }
+
+       ipc_importance_task_release(imp_task);
+       return TRUE;
+}
+#endif /* IMPORTANCE_INHERITANCE */
+
 /*
  *     Routine:        ipc_port_make_send_locked
  *     Purpose:
@@ -873,13 +2405,12 @@ ipc_port_lookup_notify(
  */
 ipc_port_t
 ipc_port_make_send_locked(
-       ipc_port_t      port)
+       ipc_port_t      port)
 {
-       assert(ip_active(port));
+       require_ip_active(port);
        port->ip_mscount++;
        port->ip_srights++;
        ip_reference(port);
-       ip_unlock(port);
        return port;
 }
 
@@ -891,17 +2422,15 @@ ipc_port_make_send_locked(
 
 ipc_port_t
 ipc_port_make_send(
-       ipc_port_t      port)
+       ipc_port_t      port)
 {
-       
-       if (!IP_VALID(port))
+       if (!IP_VALID(port)) {
                return port;
+       }
 
        ip_lock(port);
        if (ip_active(port)) {
-               port->ip_mscount++;
-               port->ip_srights++;
-               ip_reference(port);
+               ipc_port_make_send_locked(port);
                ip_unlock(port);
                return port;
        }
@@ -909,6 +2438,22 @@ ipc_port_make_send(
        return IP_DEAD;
 }
 
+/*
+ *     Routine:        ipc_port_copy_send_locked
+ *     Purpose:
+ *             Make a naked send right from another naked send right.
+ *     Conditions:
+ *             port locked and active.
+ */
+void
+ipc_port_copy_send_locked(
+       ipc_port_t      port)
+{
+       assert(port->ip_srights > 0);
+       port->ip_srights++;
+       ip_reference(port);
+}
+
 /*
  *     Routine:        ipc_port_copy_send
  *     Purpose:
@@ -923,22 +2468,21 @@ ipc_port_make_send(
 
 ipc_port_t
 ipc_port_copy_send(
-       ipc_port_t      port)
+       ipc_port_t      port)
 {
        ipc_port_t sright;
 
-       if (!IP_VALID(port))
+       if (!IP_VALID(port)) {
                return port;
+       }
 
        ip_lock(port);
        if (ip_active(port)) {
-               assert(port->ip_srights > 0);
-
-               ip_reference(port);
-               port->ip_srights++;
+               ipc_port_copy_send_locked(port);
                sright = port;
-       } else
+       } else {
                sright = IP_DEAD;
+       }
        ip_unlock(port);
 
        return sright;
@@ -955,26 +2499,28 @@ ipc_port_copy_send(
 
 mach_port_name_t
 ipc_port_copyout_send(
-       ipc_port_t      sright,
-       ipc_space_t     space)
+       ipc_port_t      sright,
+       ipc_space_t     space)
 {
        mach_port_name_t name;
 
        if (IP_VALID(sright)) {
                kern_return_t kr;
 
-               kr = ipc_object_copyout(space, (ipc_object_t) sright,
-                                       MACH_MSG_TYPE_PORT_SEND, TRUE, &name);
+               kr = ipc_object_copyout(space, ip_to_object(sright),
+                   MACH_MSG_TYPE_PORT_SEND, NULL, NULL, &name);
                if (kr != KERN_SUCCESS) {
                        ipc_port_release_send(sright);
 
-                       if (kr == KERN_INVALID_CAPABILITY)
+                       if (kr == KERN_INVALID_CAPABILITY) {
                                name = MACH_PORT_DEAD;
-                       else
+                       } else {
                                name = MACH_PORT_NULL;
+                       }
                }
-       } else
-               name = (mach_port_name_t) sright;
+       } else {
+               name = CAST_MACH_PORT_TO_NAME(sright);
+       }
 
        return name;
 }
@@ -982,7 +2528,7 @@ ipc_port_copyout_send(
 /*
  *     Routine:        ipc_port_release_send
  *     Purpose:
- *             Release a (valid) naked send right.
+ *             Release a naked send right.
  *             Consumes a ref for the port.
  *     Conditions:
  *             Nothing locked.
@@ -990,37 +2536,60 @@ ipc_port_copyout_send(
 
 void
 ipc_port_release_send(
-       ipc_port_t      port)
+       ipc_port_t      port)
 {
        ipc_port_t nsrequest = IP_NULL;
        mach_port_mscount_t mscount;
 
-       assert(IP_VALID(port));
+       if (!IP_VALID(port)) {
+               return;
+       }
 
        ip_lock(port);
-       ip_release(port);
+
+       assert(port->ip_srights > 0);
+       if (port->ip_srights == 0) {
+               panic("Over-release of port %p send right!", port);
+       }
+
+       port->ip_srights--;
 
        if (!ip_active(port)) {
-               ip_check_unlock(port);
+               ip_unlock(port);
+               ip_release(port);
                return;
        }
 
-       assert(port->ip_srights > 0);
-
-       if (--port->ip_srights == 0 &&
+       if (port->ip_srights == 0 &&
            port->ip_nsrequest != IP_NULL) {
                nsrequest = port->ip_nsrequest;
                port->ip_nsrequest = IP_NULL;
                mscount = port->ip_mscount;
                ip_unlock(port);
+               ip_release(port);
                ipc_notify_no_senders(nsrequest, mscount);
-               /*
-                * Check that there are no other locks taken, because
-                * [norma_]ipc_notify_no_senders routines may block.
-                */
-               check_simple_locks();
-       } else
+       } else {
                ip_unlock(port);
+               ip_release(port);
+       }
+}
+
+/*
+ *     Routine:        ipc_port_make_sonce_locked
+ *     Purpose:
+ *             Make a naked send-once right from a receive right.
+ *     Conditions:
+ *             The port is locked and active.
+ */
+
+ipc_port_t
+ipc_port_make_sonce_locked(
+       ipc_port_t      port)
+{
+       require_ip_active(port);
+       port->ip_sorights++;
+       ip_reference(port);
+       return port;
 }
 
 /*
@@ -1028,22 +2597,25 @@ ipc_port_release_send(
  *     Purpose:
  *             Make a naked send-once right from a receive right.
  *     Conditions:
- *             The port is not locked but it is active.
+ *             The port is not locked.
  */
 
 ipc_port_t
 ipc_port_make_sonce(
-       ipc_port_t      port)
+       ipc_port_t      port)
 {
-       assert(IP_VALID(port));
+       if (!IP_VALID(port)) {
+               return port;
+       }
 
        ip_lock(port);
-       assert(ip_active(port));
-       port->ip_sorights++;
-       ip_reference(port);
+       if (ip_active(port)) {
+               ipc_port_make_sonce_locked(port);
+               ip_unlock(port);
+               return port;
+       }
        ip_unlock(port);
-
-       return port;
+       return IP_DEAD;
 }
 
 /*
@@ -1062,24 +2634,25 @@ ipc_port_make_sonce(
 
 void
 ipc_port_release_sonce(
-       ipc_port_t      port)
+       ipc_port_t      port)
 {
-       assert(IP_VALID(port));
+       if (!IP_VALID(port)) {
+               return;
+       }
+
+       ipc_port_adjust_special_reply_port(port, IPC_PORT_ADJUST_RESET_BOOSTRAP_CHECKIN);
 
        ip_lock(port);
 
        assert(port->ip_sorights > 0);
+       if (port->ip_sorights == 0) {
+               panic("Over-release of port %p send-once right!", port);
+       }
 
        port->ip_sorights--;
 
-       ip_release(port);
-
-       if (!ip_active(port)) {
-               ip_check_unlock(port);
-               return;
-       }
-
        ip_unlock(port);
+       ip_release(port);
 }
 
 /*
@@ -1093,21 +2666,25 @@ ipc_port_release_sonce(
 
 void
 ipc_port_release_receive(
-       ipc_port_t      port)
+       ipc_port_t      port)
 {
        ipc_port_t dest;
 
-       assert(IP_VALID(port));
+       if (!IP_VALID(port)) {
+               return;
+       }
 
        ip_lock(port);
-       assert(ip_active(port));
+       require_ip_active(port);
        assert(port->ip_receiver_name == MACH_PORT_NULL);
        dest = port->ip_destination;
 
        ipc_port_destroy(port); /* consumes ref, unlocks */
 
-       if (dest != IP_NULL)
-               ipc_port_release(dest);
+       if (dest != IP_NULL) {
+               ipc_port_send_turnstile_complete(dest);
+               ip_release(dest);
+       }
 }
 
 /*
@@ -1122,20 +2699,31 @@ ipc_port_release_receive(
 
 ipc_port_t
 ipc_port_alloc_special(
-       ipc_space_t     space)
+       ipc_space_t             space,
+       ipc_port_init_flags_t   flags)
 {
        ipc_port_t port;
 
-       port = (ipc_port_t) io_alloc(IOT_PORT);
-       if (port == IP_NULL)
+       port = ip_object_to_port(io_alloc(IOT_PORT));
+       if (port == IP_NULL) {
                return IP_NULL;
+       }
+
+#if     MACH_ASSERT
+       uintptr_t buf[IP_CALLSTACK_MAX];
+       ipc_port_callstack_init_debug(&buf[0], IP_CALLSTACK_MAX);
+#endif /* MACH_ASSERT */
 
        bzero((char *)port, sizeof(*port));
-       io_lock_init(&port->ip_object);
+       io_lock_init(ip_to_object(port));
        port->ip_references = 1;
        port->ip_object.io_bits = io_makebits(TRUE, IOT_PORT, 0);
 
-       ipc_port_init(port, space, 1);
+       ipc_port_init(port, space, flags, 1);
+
+#if     MACH_ASSERT
+       ipc_port_init_debug(port, &buf[0], IP_CALLSTACK_MAX);
+#endif  /* MACH_ASSERT */
 
        return port;
 }
@@ -1151,12 +2739,12 @@ ipc_port_alloc_special(
 
 void
 ipc_port_dealloc_special(
-       ipc_port_t      port,
-       ipc_space_t     space)
+       ipc_port_t                      port,
+       __assert_only ipc_space_t       space)
 {
        ip_lock(port);
-       assert(ip_active(port));
-       assert(port->ip_receiver_name != MACH_PORT_NULL);
+       require_ip_active(port);
+//     assert(port->ip_receiver_name != MACH_PORT_NULL);
        assert(port->ip_receiver == space);
 
        /*
@@ -1164,765 +2752,310 @@ ipc_port_dealloc_special(
         *      the ipc_space_kernel check in ipc_mqueue_send.
         */
 
+       imq_lock(&port->ip_messages);
        port->ip_receiver_name = MACH_PORT_NULL;
        port->ip_receiver = IS_NULL;
+       imq_unlock(&port->ip_messages);
 
        /* relevant part of ipc_port_clear_receiver */
-       ipc_port_set_mscount(port, 0);
+       port->ip_mscount = 0;
        port->ip_messages.imq_seqno = 0;
 
        ipc_port_destroy(port);
 }
 
-
-#if    MACH_ASSERT
-/*
- *     Keep a list of all allocated ports.
- *     Allocation is intercepted via ipc_port_init;
- *     deallocation is intercepted via io_free.
- */
-queue_head_t   port_alloc_queue;
-decl_mutex_data(,port_alloc_queue_lock)
-
-unsigned long  port_count = 0;
-unsigned long  port_count_warning = 20000;
-unsigned long  port_timestamp = 0;
-
-void           db_port_stack_trace(
-                       ipc_port_t      port);
-void           db_ref(
-                       int             refs);
-int            db_port_walk(
-                       unsigned int    verbose,
-                       unsigned int    display,
-                       unsigned int    ref_search,
-                       unsigned int    ref_target);
-
-/*
- *     Initialize global state needed for run-time
- *     port debugging.
- */
-void
-ipc_port_debug_init(void)
-{
-       queue_init(&port_alloc_queue);
-       mutex_init(&port_alloc_queue_lock, ETAP_IPC_PORT_ALLOCQ);
-}
-
-
-/*
- *     Initialize all of the debugging state in a port.
- *     Insert the port into a global list of all allocated ports.
- */
-void
-ipc_port_init_debug(
-       ipc_port_t      port)
-{
-       unsigned int    i;
-
-       port->ip_thread = (unsigned long) current_thread();
-       port->ip_timetrack = port_timestamp++;
-       for (i = 0; i < IP_CALLSTACK_MAX; ++i)
-               port->ip_callstack[i] = 0;
-       for (i = 0; i < IP_NSPARES; ++i)
-               port->ip_spares[i] = 0;
-
-       /*
-        *      Machine-dependent routine to fill in an
-        *      array with up to IP_CALLSTACK_MAX levels
-        *      of return pc information.
-        */
-       machine_callstack(&port->ip_callstack[0], IP_CALLSTACK_MAX);
-
-#if 0
-       mutex_lock(&port_alloc_queue_lock);
-       ++port_count;
-       if (port_count_warning > 0 && port_count >= port_count_warning)
-               assert(port_count < port_count_warning);
-       queue_enter(&port_alloc_queue, port, ipc_port_t, ip_port_links);
-       mutex_unlock(&port_alloc_queue_lock);
-#endif
-}
-
-
 /*
- *     Remove a port from the queue of allocated ports.
- *     This routine should be invoked JUST prior to
- *     deallocating the actual memory occupied by the port.
+ *     Routine:        ipc_port_finalize
+ *     Purpose:
+ *             Called on last reference deallocate to
+ *             free any remaining data associated with the
+ *             port.
+ *     Conditions:
+ *             Nothing locked.
  */
 void
-ipc_port_track_dealloc(
-       ipc_port_t      port)
+ipc_port_finalize(
+       ipc_port_t              port)
 {
-#if 0
-       mutex_lock(&port_alloc_queue_lock);
-       assert(port_count > 0);
-       --port_count;
-       queue_remove(&port_alloc_queue, port, ipc_port_t, ip_port_links);
-       mutex_unlock(&port_alloc_queue_lock);
-#endif
-}
-
-#endif /* MACH_ASSERT */
-
-
-#if    MACH_KDB
-
-#include <ddb/db_output.h>
-#include <ddb/db_print.h>
+       ipc_port_request_t requests = port->ip_requests;
 
-#define        printf  kdbprintf
-extern int db_indent;
+       assert(port_send_turnstile(port) == TURNSTILE_NULL);
+       if (imq_is_turnstile_proxy(&port->ip_messages)) {
+               assert(ipc_port_rcv_turnstile(port) == TURNSTILE_NULL);
+       }
 
-int
-db_port_queue_print(
-       ipc_port_t      port);
+       if (ip_active(port)) {
+               panic("Trying to free an active port. port %p", port);
+       }
 
-/*
- * ipc_entry_print - pretty-print an ipc_entry
- */
-static void ipc_entry_print(struct ipc_entry *, char *); /* forward */
+       if (requests != IPR_NULL) {
+               ipc_table_size_t its = requests->ipr_size;
+               it_requests_free(its, requests);
+               port->ip_requests = IPR_NULL;
+       }
 
-static void ipc_entry_print(struct ipc_entry *iep, char *tag)
-{
-       ipc_entry_bits_t bits = iep->ie_bits;
+       ipc_mqueue_deinit(&port->ip_messages);
 
-       iprintf("%s @", tag);
-       printf(" 0x%x, bits=%x object=%x\n", iep, bits, iep->ie_object);
-       db_indent += 2;
-       iprintf("urefs=%x ", IE_BITS_UREFS(bits));
-       printf("type=%x gen=%x\n", IE_BITS_TYPE(bits), IE_BITS_GEN(bits));
-       db_indent -= 2;
+#if     MACH_ASSERT
+       ipc_port_track_dealloc(port);
+#endif  /* MACH_ASSERT */
 }
 
 /*
- *     Routine:        ipc_port_print
+ *     Routine:        kdp_mqueue_send_find_owner
  *     Purpose:
- *             Pretty-print a port for kdb.
+ *             Discover the owner of the ipc_mqueue that contains the input
+ *             waitq object. The thread blocked on the waitq should be
+ *             waiting for an IPC_MQUEUE_FULL event.
+ *     Conditions:
+ *             The 'waitinfo->wait_type' value should already be set to
+ *             kThreadWaitPortSend.
+ *     Note:
+ *             If we find out that the containing port is actually in
+ *             transit, we reset the wait_type field to reflect this.
  */
-int    ipc_port_print_long = 0;        /* set for more detail */
-
 void
-ipc_port_print(
-       ipc_port_t      port,
-       boolean_t       have_addr,
-       db_expr_t       count,
-       char            *modif)
+kdp_mqueue_send_find_owner(struct waitq * waitq, __assert_only event64_t event, thread_waitinfo_t * waitinfo)
 {
-       extern int      db_indent;
-       db_addr_t       task;
-       int             task_id;
-       int             nmsgs;
-       int             verbose = 0;
-#if    MACH_ASSERT
-       int             i, needs_db_indent, items_printed;
-#endif /* MACH_ASSERT */
-       
-       if (db_option(modif, 'l') || db_option(modif, 'v'))
-               ++verbose;
-
-       printf("port 0x%x\n", port);
-
-       db_indent += 2;
-
-       ipc_object_print(&port->ip_object);
-
-       if (ipc_port_print_long) {
-               printf("\n");
+       struct turnstile *turnstile;
+       assert(waitinfo->wait_type == kThreadWaitPortSend);
+       assert(event == IPC_MQUEUE_FULL);
+       assert(waitq_is_turnstile_queue(waitq));
+
+       turnstile = waitq_to_turnstile(waitq);
+       ipc_port_t port = (ipc_port_t)turnstile->ts_proprietor; /* we are blocking on send */
+       assert(kdp_is_in_zone(port, "ipc ports"));
+
+       waitinfo->owner = 0;
+       waitinfo->context  = VM_KERNEL_UNSLIDE_OR_PERM(port);
+       if (ip_lock_held_kdp(port)) {
+               /*
+                * someone has the port locked: it may be in an
+                * inconsistent state: bail
+                */
+               waitinfo->owner = STACKSHOT_WAITOWNER_PORT_LOCKED;
+               return;
        }
 
-       if (!ip_active(port)) {
-               iprintf("timestamp=0x%x", port->ip_timestamp);
-       } else if (port->ip_receiver_name == MACH_PORT_NULL) {
-               iprintf("destination=0x%x (", port->ip_destination);
-               if (port->ip_destination != MACH_PORT_NULL &&
-                   (task = db_task_from_space(port->ip_destination->
-                                              ip_receiver, &task_id)))
-                       printf("task%d at 0x%x", task_id, task);
-               else
-                       printf("unknown");
-               printf(")");
-       } else {
-               iprintf("receiver=0x%x (", port->ip_receiver);
-               if (port->ip_receiver == ipc_space_kernel)
-                       printf("kernel");
-               else if (port->ip_receiver == ipc_space_reply)
-                       printf("reply");
-               else if (port->ip_receiver == default_pager_space)
-                       printf("default_pager");
-               else if (task = db_task_from_space(port->ip_receiver, &task_id))
-                       printf("task%d at 0x%x", task_id, task);
-               else
-                       printf("unknown");
-               printf(")");
-       }
-       printf(", receiver_name=0x%x\n", port->ip_receiver_name);
-
-       iprintf("mscount=%d", port->ip_mscount);
-       printf(", srights=%d", port->ip_srights);
-       printf(", sorights=%d\n", port->ip_sorights);
-
-       iprintf("nsrequest=0x%x", port->ip_nsrequest);
-       printf(", pdrequest=0x%x", port->ip_pdrequest);
-       printf(", dnrequests=0x%x\n", port->ip_dnrequests);
-
-       iprintf("pset_count=0x%x", port->ip_pset_count);
-       printf(", seqno=%d", port->ip_messages.imq_seqno);
-       printf(", msgcount=%d", port->ip_messages.imq_msgcount);
-       printf(", qlimit=%d\n", port->ip_messages.imq_qlimit);
-
-       iprintf("kmsgs=0x%x", port->ip_messages.imq_messages.ikmq_base);
-       printf(", rcvrs queue=0x%x", port->ip_messages.imq_wait_queue);
-       printf(", kobj=0x%x\n", port->ip_kobject);
-
-       iprintf("premsg=0x%x", port->ip_premsg);
-
-#if    MACH_ASSERT
-       /* don't bother printing callstack or queue links */
-       iprintf("ip_thread=0x%x, ip_timetrack=0x%x\n",
-               port->ip_thread, port->ip_timetrack);
-       items_printed = 0;
-       needs_db_indent = 1;
-       for (i = 0; i < IP_NSPARES; ++i) {
-               if (port->ip_spares[i] != 0) {
-                       if (needs_db_indent) {
-                               iprintf("");
-                               needs_db_indent = 0;
+       if (ip_active(port)) {
+               if (port->ip_tempowner) {
+                       if (port->ip_imp_task != IIT_NULL && port->ip_imp_task->iit_task != NULL) {
+                               /* port is held by a tempowner */
+                               waitinfo->owner = pid_from_task(port->ip_imp_task->iit_task);
+                       } else {
+                               waitinfo->owner = STACKSHOT_WAITOWNER_INTRANSIT;
                        }
-                       printf("%sip_spares[%d] = %d",
-                              items_printed ? ", " : "", i, 
-                              port->ip_spares[i]);
-                       if (++items_printed >= 4) {
-                               needs_db_indent = 1;
-                               printf("\n");
-                               items_printed = 0;
+               } else if (port->ip_receiver_name) {
+                       /* port in a space */
+                       if (port->ip_receiver == ipc_space_kernel) {
+                               /*
+                                * The kernel pid is 0, make this
+                                * distinguishable from no-owner and
+                                * inconsistent port state.
+                                */
+                               waitinfo->owner = STACKSHOT_WAITOWNER_KERNEL;
+                       } else {
+                               waitinfo->owner = pid_from_task(port->ip_receiver->is_task);
                        }
+               } else if (port->ip_destination != IP_NULL) {
+                       /* port in transit */
+                       waitinfo->wait_type = kThreadWaitPortSendInTransit;
+                       waitinfo->owner     = VM_KERNEL_UNSLIDE_OR_PERM(port->ip_destination);
                }
        }
-#endif /* MACH_ASSERT */
-
-       if (verbose) {
-               iprintf("kmsg queue contents:\n");
-               db_indent += 2;
-               nmsgs = db_port_queue_print(port);
-               db_indent -= 2;
-               iprintf("...total kmsgs:  %d\n", nmsgs);
-       }
-
-       db_indent -=2;
-}
-
-ipc_port_t
-ipc_name_to_data(
-       task_t                  task,
-       mach_port_name_t        name)
-{
-       ipc_space_t     space;
-       ipc_entry_t     entry;
-
-       if (task == TASK_NULL) {
-               db_printf("port_name_to_data: task is null\n");
-               return (0);
-       }
-       if ((space = task->itk_space) == 0) {
-               db_printf("port_name_to_data: task->itk_space is null\n");
-               return (0);
-       }
-       if (!space->is_active) {
-               db_printf("port_name_to_data: task->itk_space not active\n");
-               return (0);
-       }
-       if ((entry = ipc_entry_lookup(space, name)) == 0) {
-               db_printf("port_name_to_data: lookup yields zero\n");
-               return (0);
-       }
-       return ((ipc_port_t)entry->ie_object);
 }
 
-#if    ZONE_DEBUG
+/*
+ *     Routine:        kdp_mqueue_recv_find_owner
+ *     Purpose:
+ *             Discover the "owner" of the ipc_mqueue that contains the input
+ *             waitq object. The thread blocked on the waitq is trying to
+ *             receive on the mqueue.
+ *     Conditions:
+ *             The 'waitinfo->wait_type' value should already be set to
+ *             kThreadWaitPortReceive.
+ *     Note:
+ *             If we find that we are actualy waiting on a port set, we reset
+ *             the wait_type field to reflect this.
+ */
 void
-print_type_ports(type, dead)
-       unsigned type;
-       unsigned dead;
+kdp_mqueue_recv_find_owner(struct waitq * waitq, __assert_only event64_t event, thread_waitinfo_t * waitinfo)
 {
-       ipc_port_t port;
-       int n;
-
-       n = 0;
-       for (port = (ipc_port_t)first_element(ipc_object_zones[IOT_PORT]);
-            port;
-            port = (ipc_port_t)next_element(ipc_object_zones[IOT_PORT], 
-                                            (vm_offset_t)port))
-               if (ip_kotype(port) == type &&
-                   (!dead || !ip_active(port))) {
-                       if (++n % 5)
-                               printf("0x%x\t", port);
-                       else
-                               printf("0x%x\n", port);
+       assert(waitinfo->wait_type == kThreadWaitPortReceive);
+       assert(event == IPC_MQUEUE_RECEIVE);
+
+       ipc_mqueue_t mqueue = imq_from_waitq(waitq);
+       waitinfo->owner     = 0;
+       if (imq_is_set(mqueue)) { /* we are waiting on a port set */
+               ipc_pset_t set = ips_from_mq(mqueue);
+               assert(kdp_is_in_zone(set, "ipc port sets"));
+
+               /* Reset wait type to specify waiting on port set receive */
+               waitinfo->wait_type = kThreadWaitPortSetReceive;
+               waitinfo->context   = VM_KERNEL_UNSLIDE_OR_PERM(set);
+               if (ips_lock_held_kdp(set)) {
+                       waitinfo->owner = STACKSHOT_WAITOWNER_PSET_LOCKED;
                }
-       if (n % 5)
-               printf("\n");
-}
+               /* There is no specific owner "at the other end" of a port set, so leave unset. */
+       } else {
+               ipc_port_t port   = ip_from_mq(mqueue);
+               assert(kdp_is_in_zone(port, "ipc ports"));
 
-void
-print_ports(void)
-{
-       ipc_port_t port;
-       int total_port_count;
-       int space_null_count;
-       int space_kernel_count;
-       int space_reply_count;
-       int space_pager_count;
-       int space_other_count;
-
-       struct {
-               int total_count;
-               int dead_count;
-       } port_types[IKOT_MAX_TYPE];
-
-       total_port_count = 0;
-
-       bzero((char *)&port_types[0], sizeof(port_types));
-       space_null_count = 0;
-       space_kernel_count = 0;
-       space_reply_count = 0;
-       space_pager_count = 0;
-       space_other_count = 0;
-
-       for (port = (ipc_port_t)first_element(ipc_object_zones[IOT_PORT]);
-            port;
-            port = (ipc_port_t)next_element(ipc_object_zones[IOT_PORT], 
-                                            (vm_offset_t)port)) {
-               total_port_count++;
-               if (ip_kotype(port) >= IKOT_MAX_TYPE) {
-                       port_types[IKOT_UNKNOWN].total_count++;
-                       if (!io_active(&port->ip_object))
-                               port_types[IKOT_UNKNOWN].dead_count++;
-               } else {
-                       port_types[ip_kotype(port)].total_count++;
-                       if (!io_active(&port->ip_object))
-                               port_types[ip_kotype(port)].dead_count++;
+               waitinfo->context = VM_KERNEL_UNSLIDE_OR_PERM(port);
+               if (ip_lock_held_kdp(port)) {
+                       waitinfo->owner = STACKSHOT_WAITOWNER_PORT_LOCKED;
+                       return;
                }
 
-               if (!port->ip_receiver)
-                       space_null_count++;
-               else if (port->ip_receiver == ipc_space_kernel)
-                       space_kernel_count++;
-               else if (port->ip_receiver == ipc_space_reply)
-                       space_reply_count++;
-               else if (port->ip_receiver == default_pager_space)
-                       space_pager_count++;
-               else
-                       space_other_count++;
-       }
-       printf("\n%7d   total ports\n\n", total_port_count);
-
-#define PRINT_ONE_PORT_TYPE(name) \
-       printf("%7d     %s", port_types[IKOT_##name].total_count, # name); \
-       if (port_types[IKOT_##name].dead_count) \
-            printf(" (%d dead ports)", port_types[IKOT_##name].dead_count);\
-       printf("\n");
-
-       PRINT_ONE_PORT_TYPE(NONE);
-       PRINT_ONE_PORT_TYPE(THREAD);
-       PRINT_ONE_PORT_TYPE(TASK);
-       PRINT_ONE_PORT_TYPE(HOST);
-       PRINT_ONE_PORT_TYPE(HOST_PRIV);
-       PRINT_ONE_PORT_TYPE(PROCESSOR);
-       PRINT_ONE_PORT_TYPE(PSET);
-       PRINT_ONE_PORT_TYPE(PSET_NAME);
-       PRINT_ONE_PORT_TYPE(PAGING_REQUEST);
-       PRINT_ONE_PORT_TYPE(MEMORY_OBJECT);
-       PRINT_ONE_PORT_TYPE(MIG);
-       PRINT_ONE_PORT_TYPE(XMM_PAGER);
-       PRINT_ONE_PORT_TYPE(XMM_KERNEL);
-       PRINT_ONE_PORT_TYPE(XMM_REPLY);
-       PRINT_ONE_PORT_TYPE(CLOCK);
-       PRINT_ONE_PORT_TYPE(CLOCK_CTRL);
-       PRINT_ONE_PORT_TYPE(MASTER_DEVICE);
-       PRINT_ONE_PORT_TYPE(UNKNOWN);
-       printf("\nipc_space:\n\n");
-       printf("NULL    KERNEL  REPLY   PAGER   OTHER\n");
-       printf("%d      %d      %d      %d      %d\n",
-              space_null_count,
-              space_kernel_count,
-              space_reply_count,
-              space_pager_count,
-              space_other_count
-       );
-}
-
-#endif /* ZONE_DEBUG */
-
-
-/*
- *     Print out all the kmsgs in a queue.  Aggregate kmsgs with
- *     identical message ids into a single entry.  Count up the
- *     amount of inline and out-of-line data consumed by each
- *     and every kmsg.
- *
- */
-
-#define        KMSG_MATCH_FIELD(kmsg)  ((unsigned int) kmsg->ikm_header.msgh_id)
-#define        DKQP_LONG(kmsg) FALSE
-char   *dkqp_long_format = "(%3d) <%10d> 0x%x   %10d %10d\n";
-char   *dkqp_format = "(%3d) <%10d> 0x%x   %10d %10d\n";
-
-int
-db_kmsg_queue_print(
-       ipc_kmsg_t      kmsg);
-int
-db_kmsg_queue_print(
-       ipc_kmsg_t      kmsg)
-{
-       ipc_kmsg_t      ikmsg, first_kmsg;
-       register int    icount;
-       mach_msg_id_t   cur_id;
-       unsigned int    inline_total, ool_total;
-       int             nmsgs;
-
-       iprintf("Count      msgh_id  kmsg addr inline bytes   ool bytes\n");
-       inline_total = ool_total = (vm_size_t) 0;
-       cur_id = KMSG_MATCH_FIELD(kmsg);
-       for (icount = 0, nmsgs = 0, first_kmsg = ikmsg = kmsg;
-            kmsg != IKM_NULL && (kmsg != first_kmsg || nmsgs == 0);
-            kmsg = kmsg->ikm_next) {
-               ++nmsgs;
-               if (!(KMSG_MATCH_FIELD(kmsg) == cur_id)) {
-                       iprintf(DKQP_LONG(kmsg) ? dkqp_long_format:dkqp_format,
-                               icount, cur_id, ikmsg, inline_total,ool_total);
-                       cur_id = KMSG_MATCH_FIELD(kmsg);
-                       icount = 1;
-                       ikmsg = kmsg;
-                       inline_total = ool_total = 0;
-               } else {
-                       icount++;
+               if (ip_active(port)) {
+                       if (port->ip_receiver_name != MACH_PORT_NULL) {
+                               waitinfo->owner = port->ip_receiver_name;
+                       } else {
+                               waitinfo->owner = STACKSHOT_WAITOWNER_INTRANSIT;
+                       }
                }
-               if (DKQP_LONG(kmsg))
-                       inline_total += kmsg->ikm_size;
-               else
-                       inline_total += kmsg->ikm_header.msgh_size;
        }
-       iprintf(DKQP_LONG(kmsg) ? dkqp_long_format : dkqp_format,
-               icount, cur_id, ikmsg, inline_total, ool_total);
-       return nmsgs;
 }
 
+#if     MACH_ASSERT
+#include <kern/machine.h>
 
 /*
- *     Process all of the messages on a port - prints out the
- *     number of occurences of each message type, and the first
- *     kmsg with a particular msgh_id.
+ *     Keep a list of all allocated ports.
+ *     Allocation is intercepted via ipc_port_init;
+ *     deallocation is intercepted via io_free.
  */
-int
-db_port_queue_print(
-       ipc_port_t      port)
-{
-       ipc_kmsg_t      kmsg;
-
-       if (ipc_kmsg_queue_empty(&port->ip_messages.imq_messages))
-               return 0;
-       kmsg = ipc_kmsg_queue_first(&port->ip_messages.imq_messages);
-       return db_kmsg_queue_print(kmsg);
-}
-
+#if 0
+queue_head_t    port_alloc_queue;
+lck_spin_t      port_alloc_queue_lock;
+#endif
 
-#if    MACH_ASSERT
-#include <ddb/db_sym.h>
-#include <ddb/db_access.h>
+unsigned long   port_count = 0;
+unsigned long   port_count_warning = 20000;
+unsigned long   port_timestamp = 0;
 
-#define        FUNC_NULL       ((void (*)) 0)
-#define        MAX_REFS        5               /* bins for tracking ref counts */
+void            db_port_stack_trace(
+       ipc_port_t      port);
+void            db_ref(
+       int             refs);
+int             db_port_walk(
+       unsigned int    verbose,
+       unsigned int    display,
+       unsigned int    ref_search,
+       unsigned int    ref_target);
 
 /*
- *     Translate port's cache of call stack pointers
- *     into symbolic names.
+ *     Initialize global state needed for run-time
+ *     port debugging.
  */
 void
-db_port_stack_trace(
-       ipc_port_t      port)
+ipc_port_debug_init(void)
 {
-       unsigned int    i;
+#if 0
+       queue_init(&port_alloc_queue);
+       lck_spin_init(&port_alloc_queue_lock, &ipc_lck_grp, &ipc_lck_attr);
+#endif
 
-       for (i = 0; i < IP_CALLSTACK_MAX; ++i) {
-               iprintf("[%d] 0x%x\t", i, port->ip_callstack[i]);
-               if (port->ip_callstack[i] != 0 &&
-                   DB_VALID_KERN_ADDR(port->ip_callstack[i]))
-                       db_printsym(port->ip_callstack[i], DB_STGY_PROC);
-               printf("\n");
+       if (!PE_parse_boot_argn("ipc_portbt", &ipc_portbt, sizeof(ipc_portbt))) {
+               ipc_portbt = 0;
        }
 }
 
+#ifdef MACH_BSD
+extern int proc_pid(struct proc*);
+#endif /* MACH_BSD */
 
-typedef struct port_item {
-       unsigned long   item;
-       unsigned long   count;
-} port_item;
-
-
-#define        ITEM_MAX        400
-typedef struct port_track {
-       char            *name;
-       unsigned long   max;
-       unsigned long   warning;
-       port_item       items[ITEM_MAX];
-} port_track;
-
-port_track     port_callers;           /* match against calling addresses */
-port_track     port_threads;           /* match against allocating threads */
-port_track     port_spaces;            /* match against ipc spaces */
-
-void           port_track_init(
-                       port_track      *trackp,
-                       char            *name);
-void           port_item_add(
-                       port_track      *trackp,
-                       unsigned long   item);
-void           port_track_sort(
-                       port_track      *trackp);
-void           port_track_print(
-                       port_track      *trackp,
-                       void            (*func)(port_item *));
-void           port_callers_print(
-                       port_item       *p);
-
+/*
+ *     Initialize all of the debugging state in a port.
+ *     Insert the port into a global list of all allocated ports.
+ */
 void
-port_track_init(
-       port_track      *trackp,
-       char            *name)
+ipc_port_init_debug(
+       ipc_port_t      port,
+       uintptr_t       *callstack,
+       unsigned int    callstack_max)
 {
-       port_item       *i;
-
-       trackp->max = trackp->warning = 0;
-       trackp->name = name;
-       for (i = trackp->items; i < trackp->items + ITEM_MAX; ++i)
-               i->item = i->count = 0;
-}
-
+       unsigned int    i;
 
-void
-port_item_add(
-       port_track      *trackp,
-       unsigned long   item)
-{
-       port_item       *limit, *i;
+       port->ip_thread = current_thread();
+       port->ip_timetrack = port_timestamp++;
+       for (i = 0; i < callstack_max; ++i) {
+               port->ip_callstack[i] = callstack[i];
+       }
+       for (i = 0; i < IP_NSPARES; ++i) {
+               port->ip_spares[i] = 0;
+       }
 
-       limit = trackp->items + trackp->max;
-       for (i = trackp->items; i < limit; ++i)
-               if (i->item == item) {
-                       i->count++;
-                       return;
+#ifdef MACH_BSD
+       task_t task = current_task();
+       if (task != TASK_NULL) {
+               struct proc* proc = (struct proc*) get_bsdtask_info(task);
+               if (proc) {
+                       port->ip_spares[0] = proc_pid(proc);
                }
-       if (trackp->max >= ITEM_MAX) {
-               if (trackp->warning++ == 0)
-                       iprintf("%s:  no room\n", trackp->name);
-               return;
        }
-       i->item = item;
-       i->count = 1;
-       trackp->max++;
-}
+#endif /* MACH_BSD */
 
+#if 0
+       lck_spin_lock(&port_alloc_queue_lock);
+       ++port_count;
+       if (port_count_warning > 0 && port_count >= port_count_warning) {
+               assert(port_count < port_count_warning);
+       }
+       queue_enter(&port_alloc_queue, port, ipc_port_t, ip_port_links);
+       lck_spin_unlock(&port_alloc_queue_lock);
+#endif
+}
 
 /*
- *     Simple (and slow) bubble sort.
+ *     Routine:        ipc_port_callstack_init_debug
+ *     Purpose:
+ *             Calls the machine-dependent routine to
+ *             fill in an array with up to IP_CALLSTACK_MAX
+ *             levels of return pc information
+ *     Conditions:
+ *             May block (via copyin)
  */
 void
-port_track_sort(
-       port_track      *trackp)
-{
-       port_item       *limit, *p;
-       port_item       temp;
-       boolean_t       unsorted;
-
-       limit = trackp->items + trackp->max - 1;
-       do {
-               unsorted = FALSE;
-               for (p = trackp->items; p < limit - 1; ++p) {
-                       if (p->count < (p+1)->count) {
-                               temp = *p;
-                               *p = *(p+1);
-                               *(p+1) = temp;
-                               unsorted = TRUE;
-                       }
-               }
-       } while (unsorted == TRUE);
-}
-
-
-void
-port_track_print(
-       port_track      *trackp,
-       void            (*func)(port_item *))
+ipc_port_callstack_init_debug(
+       uintptr_t       *callstack,
+       unsigned int    callstack_max)
 {
-       port_item       *limit, *p;
+       unsigned int    i;
 
-       limit = trackp->items + trackp->max;
-       iprintf("%s:\n", trackp->name);
-       for (p = trackp->items; p < limit; ++p) {
-               if (func != FUNC_NULL)
-                       (*func)(p);
-               else
-                       iprintf("0x%x\t%8d\n", p->item, p->count);
+       /* guarantee the callstack is initialized */
+       for (i = 0; i < callstack_max; i++) {
+               callstack[i] = 0;
        }
-}
 
-
-void
-port_callers_print(
-       port_item       *p)
-{
-       iprintf("0x%x\t%8d\t", p->item, p->count);
-       db_printsym(p->item, DB_STGY_PROC);
-       printf("\n");
+       if (ipc_portbt) {
+               machine_callstack(callstack, callstack_max);
+       }
 }
 
-
 /*
- *     Show all ports with a given reference count.
+ *     Remove a port from the queue of allocated ports.
+ *     This routine should be invoked JUST prior to
+ *     deallocating the actual memory occupied by the port.
  */
+#if 1
 void
-db_ref(
-       int             refs)
-{
-       db_port_walk(1, 1, 1, refs);
-}
-
-
-/*
- *     Examine all currently allocated ports.
- *     Options:
- *             verbose         display suspicious ports
- *             display         print out each port encountered
- *             ref_search      restrict examination to ports with
- *                             a specified reference count
- *             ref_target      reference count for ref_search
- */
-int
-db_port_walk(
-       unsigned int    verbose,
-       unsigned int    display,
-       unsigned int    ref_search,
-       unsigned int    ref_target)
-{
-       ipc_port_t      port;
-       unsigned int    ref_overflow, refs, i, ref_inactive_overflow;
-       unsigned int    no_receiver, no_match;
-       unsigned int    ref_counts[MAX_REFS];
-       unsigned int    inactive[MAX_REFS];
-       unsigned int    ipc_ports = 0;
-       unsigned int    proxies = 0, principals = 0;
-
-       iprintf("Allocated port count is %d\n", port_count);
-       no_receiver = no_match = ref_overflow = 0;
-       ref_inactive_overflow = 0;
-       for (i = 0; i < MAX_REFS; ++i) {
-               ref_counts[i] = 0;
-               inactive[i] = 0;
-       }
-       port_track_init(&port_callers, "port callers");
-       port_track_init(&port_threads, "port threads");
-       port_track_init(&port_spaces, "port spaces");
-       if (ref_search)
-               iprintf("Walking ports of ref_count=%d.\n", ref_target);
-       else
-               iprintf("Walking all ports.\n");
-
-       queue_iterate(&port_alloc_queue, port, ipc_port_t, ip_port_links) {
-               char    *port_type;
-
-               port_type = " IPC port";
-               if (ip_active(port))
-                 ipc_ports++;
-
-               refs = port->ip_references;
-               if (ref_search && refs != ref_target)
-                       continue;
-
-               if (refs >= MAX_REFS) {
-                       if (ip_active(port))
-                               ++ref_overflow;
-                       else
-                               ++ref_inactive_overflow;
-               } else {
-                       if (refs == 0 && verbose)
-                               iprintf("%s 0x%x has ref count of zero!\n",
-                                       port_type, port);
-                       if (ip_active(port))
-                               ref_counts[refs]++;
-                       else
-                               inactive[refs]++;
-               }
-               port_item_add(&port_threads, (unsigned long) port->ip_thread);
-               for (i = 0; i < IP_CALLSTACK_MAX; ++i) {
-                       if (port->ip_callstack[i] != 0 &&
-                           DB_VALID_KERN_ADDR(port->ip_callstack[i]))
-                               port_item_add(&port_callers,
-                                             port->ip_callstack[i]);
-               }
-               if (!ip_active(port)) {
-                       if (verbose)
-                               iprintf("%s 0x%x, inactive, refcnt %d\n",
-                                       port_type, port, refs);
-                       continue;
-               }
-
-               if (port->ip_receiver_name == MACH_PORT_NULL) {
-                       iprintf("%s  0x%x, no receiver, refcnt %d\n",
-                               port, refs);
-                       ++no_receiver;
-                       continue;
-               }
-               if (port->ip_receiver == ipc_space_kernel ||
-                   port->ip_receiver == ipc_space_reply ||
-                   ipc_entry_lookup(port->ip_receiver,
-                                       port->ip_receiver_name) 
-                                       != IE_NULL) {
-                       port_item_add(&port_spaces,
-                                     (unsigned long)port->ip_receiver);
-                       if (display) {
-                               iprintf( "%s 0x%x time 0x%x ref_cnt %d\n",
-                                               port_type, port,
-                                               port->ip_timetrack, refs);
-                       }
-                       continue;
-               }
-               iprintf("%s 0x%x, rcvr 0x%x, name 0x%x, ref %d, no match\n",
-                               port_type, port, port->ip_receiver,
-                               port->ip_receiver_name, refs);
-               ++no_match;
-       }
-       iprintf("Active port type summary:\n");
-       iprintf("\tlocal  IPC %6d\n", ipc_ports);
-       iprintf("summary:\tcallers %d threads %d spaces %d\n",
-               port_callers.max, port_threads.max, port_spaces.max);
-
-       iprintf("\tref_counts:\n");
-       for (i = 0; i < MAX_REFS; ++i)
-               iprintf("\t  ref_counts[%d] = %d\n", i, ref_counts[i]);
-
-       iprintf("\t%d ports w/o receivers, %d w/o matches\n",
-               no_receiver, no_match);
-
-       iprintf("\tinactives:");
-       if ( ref_inactive_overflow || inactive[0] || inactive[1] ||
-            inactive[2] || inactive[3] || inactive[4] )
-               printf(" [0]=%d [1]=%d [2]=%d [3]=%d [4]=%d [5+]=%d\n",
-                       inactive[0], inactive[1], inactive[2],
-                       inactive[3], inactive[4], ref_inactive_overflow);
-       else
-               printf(" No inactive ports.\n");
-
-       port_track_sort(&port_spaces);
-       port_track_print(&port_spaces, FUNC_NULL);
-       port_track_sort(&port_threads);
-       port_track_print(&port_threads, FUNC_NULL);
-       port_track_sort(&port_callers);
-       port_track_print(&port_callers, port_callers_print);
-       return 0;
+ipc_port_track_dealloc(
+       __unused ipc_port_t     port)
+{
 }
+#else
+void
+ipc_port_track_dealloc(
+       ipc_port_t              port)
+{
+       lck_spin_lock(&port_alloc_queue_lock);
+       assert(port_count > 0);
+       --port_count;
+       queue_remove(&port_alloc_queue, port, ipc_port_t, ip_port_links);
+       lck_spin_unlock(&port_alloc_queue_lock);
+}
+#endif
 
 
-#endif /* MACH_ASSERT */
-
-#endif /* MACH_KDB */
+#endif  /* MACH_ASSERT */