]> git.saurik.com Git - apple/xnu.git/blobdiff - osfmk/kern/ipc_kobject.c
xnu-7195.101.1.tar.gz
[apple/xnu.git] / osfmk / kern / ipc_kobject.c
index 95fbf38856bf276e23cb54c696c0c92337dbd63c..dc196835adba23a969238a35914614c12bb4cc6a 100644 (file)
@@ -1,52 +1,65 @@
 /*
- * Copyright (c) 2000 Apple Computer, 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
+ * Copyright (c) 2000-2020 Apple Inc. All rights reserved.
+ *
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
+ *
+ * This file contains Original Code and/or Modifications of Original Code
+ * as defined in and that are subject to the Apple Public Source License
+ * Version 2.0 (the 'License'). You may not use this file except in
+ * 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_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.
+ * Copyright (c) 2005 SPARTA, Inc.
+ */
 /*
  */
 /*
 
 #include <mach_debug.h>
 #include <mach_ipc_test.h>
-#include <mach_machine_routines.h>
-#include <norma_task.h>
-#include <mach_rt.h>
-#include <platforms.h>
-
 #include <mach/mig.h>
 #include <mach/port.h>
 #include <mach/kern_return.h>
 #include <mach/message.h>
 #include <mach/mig_errors.h>
 #include <mach/notify.h>
+#include <mach/ndr.h>
+#include <mach/vm_param.h>
 
-#include <kern/etap_macros.h>
+#include <mach/mach_vm_server.h>
+#include <mach/mach_port_server.h>
+#include <mach/mach_host_server.h>
+#include <mach/host_priv_server.h>
+#include <mach/host_security_server.h>
+#include <mach/clock_server.h>
+#include <mach/clock_priv_server.h>
+#include <mach/lock_set_server.h>
+#include <mach/memory_entry_server.h>
+#include <mach/memory_object_control_server.h>
+#include <mach/memory_object_default_server.h>
+#include <mach/processor_server.h>
+#include <mach/processor_set_server.h>
+#include <mach/task_server.h>
+#include <mach/mach_voucher_server.h>
+#include <mach/mach_voucher_attr_control_server.h>
+#ifdef VM32_SUPPORT
+#include <mach/vm32_map_server.h>
+#endif
+#include <mach/thread_act_server.h>
+#include <mach/restartable_server.h>
+
+#include <mach/exc_server.h>
+#include <mach/mach_exc_server.h>
+#include <mach/mach_eventlink_server.h>
+
+#include <device/device_types.h>
+#include <device/device_server.h>
+
+#if     CONFIG_USER_NOTIFICATION
+#include <UserNotification/UNDReplyServer.h>
+#endif
+
+#if     CONFIG_ARCADE
+#include <mach/arcade_register_server.h>
+#endif
+
+#if     CONFIG_AUDIT
+#include <kern/audit_sessionport.h>
+#endif
+
+#if     MACH_MACHINE_ROUTINES
+#include <machine/machine_routines.h>
+#endif  /* MACH_MACHINE_ROUTINES */
+#if     XK_PROXY
+#include <uk_xkern/xk_uproxy_server.h>
+#endif  /* XK_PROXY */
+
+#include <kern/counter.h>
+#include <kern/ipc_tt.h>
 #include <kern/ipc_mig.h>
+#include <kern/ipc_misc.h>
 #include <kern/ipc_kobject.h>
-#include <kern/misc_protos.h>
+#include <kern/host_notify.h>
 #include <kern/mk_timer.h>
+#include <kern/misc_protos.h>
+
+#if CONFIG_ARCADE
+#include <kern/arcade.h>
+#endif /* CONFIG_ARCADE */
+
 #include <ipc/ipc_kmsg.h>
 #include <ipc/ipc_port.h>
-#include <kern/counters.h>
+#include <ipc/ipc_voucher.h>
+#include <kern/sync_sema.h>
+#include <kern/work_interval.h>
+#include <kern/suid_cred.h>
+#include <kern/task_ident.h>
+
+#if HYPERVISOR
+#include <kern/hv_support.h>
+#endif
 
+#include <vm/vm_protos.h>
+
+#include <security/mac_mach_internal.h>
+
+extern char *proc_name_address(void *p);
+struct proc;
+extern int proc_pid(struct proc *p);
 
 /*
  *     Routine:        ipc_kobject_notify
  */
 boolean_t
 ipc_kobject_notify(
-        mach_msg_header_t *request_header,
-        mach_msg_header_t *reply_header);
-
-#include <mach/ndr.h>
+       mach_msg_header_t *request_header,
+       mach_msg_header_t *reply_header);
 
 typedef struct {
-        mach_msg_id_t num;
-        mig_routine_t routine;
+       mach_msg_id_t num;
+       mig_routine_t routine;
        int size;
-#if    MACH_COUNTERS
-       mach_counter_t callcount;
-#endif
+       int kobjidx;
 } mig_hash_t;
 
-#define MAX_MIG_ENTRIES 1024
+#define MAX_MIG_ENTRIES 1031
 #define MIG_HASH(x) (x)
 
+#define KOBJ_IDX_NOT_SET (-1)
+
 #ifndef max
-#define max(a,b)        (((a) > (b)) ? (a) : (b))
+#define max(a, b)        (((a) > (b)) ? (a) : (b))
 #endif /* max */
 
-mig_hash_t mig_buckets[MAX_MIG_ENTRIES];
-int mig_table_max_displ;
-mach_msg_size_t mig_reply_size;
-
-
-#include <mach/mach_port_server.h>
-#include <mach/mach_host_server.h>
-#include <mach/host_priv_server.h>
-#include <mach/host_security_server.h>
-#include <mach/clock_server.h>
-#include <mach/clock_priv_server.h>
-#include <mach/ledger_server.h>
-#include <mach/lock_set_server.h>
-#include <default_pager/default_pager_object_server.h>
-#include <mach/memory_object_server.h>
-#include <mach/memory_object_control_server.h>
-#include <mach/memory_object_default_server.h>
-#include <mach/memory_object_name_server.h>
-#include <mach/processor_server.h>
-#include <mach/processor_set_server.h>
-#include <mach/semaphore_server.h>
-#include <mach/task_server.h>
-#include <mach/vm_map_server.h>
-#include <mach/thread_act_server.h>
-#include <device/device_server.h>
-#include <UserNotification/UNDReplyServer.h>
-
-#if     MACH_MACHINE_ROUTINES
-#include <machine/machine_routines.h>
-#endif /* MACH_MACHINE_ROUTINES */
-#if    XK_PROXY
-#include <uk_xkern/xk_uproxy_server.h>
-#endif /* XK_PROXY */
-
-
-mig_subsystem_t mig_e[] = {
-        (mig_subsystem_t)&mach_port_subsystem,
-        (mig_subsystem_t)&mach_host_subsystem,
-        (mig_subsystem_t)&host_priv_subsystem,
-        (mig_subsystem_t)&host_security_subsystem,
-        (mig_subsystem_t)&clock_subsystem,
-        (mig_subsystem_t)&clock_priv_subsystem,
-        (mig_subsystem_t)&processor_subsystem,
-        (mig_subsystem_t)&processor_set_subsystem,
-        (mig_subsystem_t)&is_iokit_subsystem,
-        (mig_subsystem_t)&memory_object_name_subsystem,
-       (mig_subsystem_t)&lock_set_subsystem,
-       (mig_subsystem_t)&ledger_subsystem,
-       (mig_subsystem_t)&semaphore_subsystem,
-       (mig_subsystem_t)&task_subsystem,
-       (mig_subsystem_t)&thread_act_subsystem,
-       (mig_subsystem_t)&vm_map_subsystem,
-       (mig_subsystem_t)&UNDReply_subsystem,
+static SECURITY_READ_ONLY_LATE(mig_hash_t) mig_buckets[MAX_MIG_ENTRIES];
+static SECURITY_READ_ONLY_LATE(int) mig_table_max_displ;
+SECURITY_READ_ONLY_LATE(int) mach_kobj_count; /* count of total number of kobjects */
+
+static ZONE_DECLARE(ipc_kobject_label_zone, "ipc kobject labels",
+    sizeof(struct ipc_kobject_label), ZC_NONE);
+
+__startup_data
+static const struct mig_subsystem *mig_e[] = {
+       (const struct mig_subsystem *)&mach_vm_subsystem,
+       (const struct mig_subsystem *)&mach_port_subsystem,
+       (const struct mig_subsystem *)&mach_host_subsystem,
+       (const struct mig_subsystem *)&host_priv_subsystem,
+       (const struct mig_subsystem *)&host_security_subsystem,
+       (const struct mig_subsystem *)&clock_subsystem,
+       (const struct mig_subsystem *)&clock_priv_subsystem,
+       (const struct mig_subsystem *)&processor_subsystem,
+       (const struct mig_subsystem *)&processor_set_subsystem,
+       (const struct mig_subsystem *)&is_iokit_subsystem,
+       (const struct mig_subsystem *)&lock_set_subsystem,
+       (const struct mig_subsystem *)&task_subsystem,
+       (const struct mig_subsystem *)&thread_act_subsystem,
+#ifdef VM32_SUPPORT
+       (const struct mig_subsystem *)&vm32_map_subsystem,
+#endif
+#if CONFIG_USER_NOTIFICATION
+       (const struct mig_subsystem *)&UNDReply_subsystem,
+#endif
+       (const struct mig_subsystem *)&mach_voucher_subsystem,
+       (const struct mig_subsystem *)&mach_voucher_attr_control_subsystem,
+       (const struct mig_subsystem *)&memory_entry_subsystem,
+       (const struct mig_subsystem *)&task_restartable_subsystem,
 
 #if     XK_PROXY
-        (mig_subsystem_t)&do_uproxy_xk_uproxy_subsystem,
+       (const struct mig_subsystem *)&do_uproxy_xk_uproxy_subsystem,
 #endif /* XK_PROXY */
 #if     MACH_MACHINE_ROUTINES
-        (mig_subsystem_t)&MACHINE_SUBSYSTEM,
+       (const struct mig_subsystem *)&MACHINE_SUBSYSTEM,
 #endif  /* MACH_MACHINE_ROUTINES */
 #if     MCMSG && iPSC860
-       (mig_subsystem_t)&mcmsg_info_subsystem,
+       (const struct mig_subsystem *)&mcmsg_info_subsystem,
 #endif  /* MCMSG && iPSC860 */
+       (const struct mig_subsystem *)&catch_exc_subsystem,
+       (const struct mig_subsystem *)&catch_mach_exc_subsystem,
+#if CONFIG_ARCADE
+       (const struct mig_subsystem *)&arcade_register_subsystem,
+#endif
+       (const struct mig_subsystem *)&mach_eventlink_subsystem,
 };
 
-void
+static void
 mig_init(void)
 {
-    register unsigned int i, n = sizeof(mig_e)/sizeof(mig_subsystem_t);
-    register unsigned int howmany;
-    register mach_msg_id_t j, pos, nentry, range;
-       
-    for (i = 0; i < n; i++) {
-       range = mig_e[i]->end - mig_e[i]->start;
-       if (!mig_e[i]->start || range < 0)
-           panic("the msgh_ids in mig_e[] aren't valid!");
-       mig_reply_size = max(mig_reply_size, mig_e[i]->maxsize);
-
-       for  (j = 0; j < range; j++) {
-         if (mig_e[i]->routine[j].stub_routine) { 
-           /* Only put real entries in the table */
-           nentry = j + mig_e[i]->start;       
-           for (pos = MIG_HASH(nentry) % MAX_MIG_ENTRIES, howmany = 1;
-                mig_buckets[pos].num;
-                pos = ++pos % MAX_MIG_ENTRIES, howmany++) {
-                if (mig_buckets[pos].num == nentry) {
-                       printf("message id = %d\n", nentry);
-                       panic("multiple entries with the same msgh_id");
-                }
-                if (howmany == MAX_MIG_ENTRIES)
-                      panic("the mig dispatch table is too small");
-           }
-               
-           mig_buckets[pos].num = nentry;
-           mig_buckets[pos].routine = mig_e[i]->routine[j].stub_routine;
-           if (mig_e[i]->routine[j].max_reply_msg)
-                   mig_buckets[pos].size = mig_e[i]->routine[j].max_reply_msg;
-           else
-                   mig_buckets[pos].size = mig_e[i]->maxsize;
-
-           mig_table_max_displ = max(howmany, mig_table_max_displ);
-         }
-       }
-    }
-    printf("mig_table_max_displ = %d\n", mig_table_max_displ);
+       unsigned int i, n = sizeof(mig_e) / sizeof(const struct mig_subsystem *);
+       int howmany;
+       mach_msg_id_t j, pos, nentry, range;
+
+       for (i = 0; i < n; i++) {
+               range = mig_e[i]->end - mig_e[i]->start;
+               if (!mig_e[i]->start || range < 0) {
+                       panic("the msgh_ids in mig_e[] aren't valid!");
+               }
+
+               for (j = 0; j < range; j++) {
+                       if (mig_e[i]->routine[j].stub_routine) {
+                               /* Only put real entries in the table */
+                               nentry = j + mig_e[i]->start;
+                               for (pos = MIG_HASH(nentry) % MAX_MIG_ENTRIES, howmany = 1;
+                                   mig_buckets[pos].num;
+                                   pos++, pos = pos % MAX_MIG_ENTRIES, howmany++) {
+                                       if (mig_buckets[pos].num == nentry) {
+                                               printf("message id = %d\n", nentry);
+                                               panic("multiple entries with the same msgh_id");
+                                       }
+                                       if (howmany == MAX_MIG_ENTRIES) {
+                                               panic("the mig dispatch table is too small");
+                                       }
+                               }
+
+                               mig_buckets[pos].num = nentry;
+                               mig_buckets[pos].routine = mig_e[i]->routine[j].stub_routine;
+                               if (mig_e[i]->routine[j].max_reply_msg) {
+                                       mig_buckets[pos].size = mig_e[i]->routine[j].max_reply_msg;
+                               } else {
+                                       mig_buckets[pos].size = mig_e[i]->maxsize;
+                               }
+                               mig_buckets[pos].kobjidx = KOBJ_IDX_NOT_SET;
+
+                               mig_table_max_displ = max(howmany, mig_table_max_displ);
+                               mach_kobj_count++;
+                       }
+               }
+       }
+       printf("mig_table_max_displ = %d mach_kobj_count = %d\n",
+           mig_table_max_displ, mach_kobj_count);
 }
+STARTUP(MACH_IPC, STARTUP_RANK_FIRST, mig_init);
+
+/*
+ * Do a hash table lookup for given msgh_id. Return 0
+ * if not found.
+ */
+static mig_hash_t *
+find_mig_hash_entry(int msgh_id)
+{
+       unsigned int i = (unsigned int)MIG_HASH(msgh_id);
+       int max_iter = mig_table_max_displ;
+       mig_hash_t *ptr;
+
+       do {
+               ptr = &mig_buckets[i++ % MAX_MIG_ENTRIES];
+       } while (msgh_id != ptr->num && ptr->num && --max_iter);
+
+       if (!ptr->routine || msgh_id != ptr->num) {
+               ptr = (mig_hash_t *)0;
+       }
+
+       return ptr;
+}
+
+/*
+ *      Routine:       ipc_kobject_set_kobjidx
+ *      Purpose:
+ *              Set the index for the kobject filter
+ *              mask for a given message ID.
+ */
+kern_return_t
+ipc_kobject_set_kobjidx(
+       int       msgh_id,
+       int       index)
+{
+       mig_hash_t *ptr = find_mig_hash_entry(msgh_id);
+
+       if (ptr == (mig_hash_t *)0) {
+               return KERN_INVALID_ARGUMENT;
+       }
 
+       assert(index < mach_kobj_count);
+       ptr->kobjidx = index;
+
+       return KERN_SUCCESS;
+}
 
 /*
  *     Routine:        ipc_kobject_server
@@ -229,53 +343,54 @@ mig_init(void)
 
 ipc_kmsg_t
 ipc_kobject_server(
-       ipc_kmsg_t      request)
+       ipc_kmsg_t      request,
+       mach_msg_option_t __unused option)
 {
        mach_msg_size_t reply_size;
        ipc_kmsg_t reply;
        kern_return_t kr;
-       mig_routine_t routine;
-       ipc_port_t *destp;
-       mach_msg_format_0_trailer_t *trailer;
-       register mig_hash_t *ptr;
-       unsigned int th;
-
-       /* Only fetch current thread if ETAP is configured */
-       ETAP_DATA_LOAD(th, current_thread());
-        ETAP_PROBE_DATA(ETAP_P_SYSCALL_MACH,
-                        EVENT_BEGIN,
-                       ((thread_t) th),
-                        &request->ikm_header.msgh_id,
-                        sizeof(int));
-       /*
-         * Find out corresponding mig_hash entry if any
-         */
-       {
-           register int key = request->ikm_header.msgh_id;
-           register int i = MIG_HASH(key);
-           register int max_iter = mig_table_max_displ;
-       
-           do
-               ptr = &mig_buckets[i++ % MAX_MIG_ENTRIES];
-           while (key != ptr->num && ptr->num && --max_iter);
+       ipc_port_t  replyp = IPC_PORT_NULL;
+       mach_msg_max_trailer_t *trailer;
+       mig_hash_t *ptr;
+       task_t task = TASK_NULL;
+       uint32_t exec_token;
+       boolean_t exec_token_changed = FALSE;
+       int request_msgh_id = request->ikm_header->msgh_id;
+       natural_t ikot;
+       ipc_port_t port;
+
+       reply = NULL;
+       port = request->ikm_header->msgh_remote_port;
+       if (IP_VALID(port)) {
+               ikot = ip_kotype(port);
+       } else {
+               ikot = IKOT_UNKNOWN;
+       }
+       if (ikot == IKOT_UEXT_OBJECT) {
+               kr = uext_server(request, &reply);
+               if ((MIG_NO_REPLY == kr) || (KERN_SUCCESS == kr)) {
+                       ipc_kmsg_trace_send(request, option);
+                       goto msgdone;
+               }
+       }
 
-           if (!ptr->routine || key != ptr->num) {
-               ptr = (mig_hash_t *)0;
-               reply_size = mig_reply_size;
-           } else {
+       /* Find corresponding mig_hash entry, if any */
+       ptr = find_mig_hash_entry(request_msgh_id);
+
+       /* Get the reply_size. */
+       if (ptr == (mig_hash_t *)0) {
+               reply_size = sizeof(mig_reply_error_t);
+       } else {
                reply_size = ptr->size;
-#if    MACH_COUNTER
-               ptr->callcount++;
-#endif
-           }
        }
 
        /* round up for trailer size */
-        reply_size += MAX_TRAILER_SIZE;
+       reply_size += MAX_TRAILER_SIZE;
        reply = ipc_kmsg_alloc(reply_size);
 
        if (reply == IKM_NULL) {
                printf("ipc_kobject_server: dropping request\n");
+               ipc_kmsg_trace_send(request, option);
                ipc_kmsg_destroy(request);
                return IKM_NULL;
        }
@@ -284,79 +399,141 @@ ipc_kobject_server(
         * Initialize reply message.
         */
        {
-#define        InP     ((mach_msg_header_t *) &request->ikm_header)
-#define        OutP    ((mig_reply_error_t *) &reply->ikm_header)
+#define InP     ((mach_msg_header_t *) request->ikm_header)
+#define OutP    ((mig_reply_error_t *) reply->ikm_header)
 
-           OutP->NDR = NDR_record;
-           OutP->Head.msgh_size = sizeof(mig_reply_error_t);
+               /*
+                * MIG should really assure no data leakage -
+                * but until it does, pessimistically zero the
+                * whole reply buffer.
+                */
+               bzero((void *)OutP, reply_size);
+
+               OutP->NDR = NDR_record;
+               OutP->Head.msgh_size = sizeof(mig_reply_error_t);
 
-           OutP->Head.msgh_bits =
-               MACH_MSGH_BITS(MACH_MSGH_BITS_LOCAL(InP->msgh_bits), 0);
-           OutP->Head.msgh_remote_port = InP->msgh_local_port;
-           OutP->Head.msgh_local_port  = MACH_PORT_NULL;
-           OutP->Head.msgh_id = InP->msgh_id + 100;
+               OutP->Head.msgh_bits =
+                   MACH_MSGH_BITS_SET(MACH_MSGH_BITS_LOCAL(InP->msgh_bits), 0, 0, 0);
+               OutP->Head.msgh_remote_port = InP->msgh_local_port;
+               OutP->Head.msgh_local_port = MACH_PORT_NULL;
+               OutP->Head.msgh_voucher_port = MACH_PORT_NULL;
+               OutP->Head.msgh_id = InP->msgh_id + 100;
 
-#undef InP
-#undef OutP
+#undef  InP
+#undef  OutP
        }
 
        /*
         * Find the routine to call, and call it
         * to perform the kernel function
         */
+       ipc_kmsg_trace_send(request, option);
        {
-           if (ptr) {  
-               (*ptr->routine)(&request->ikm_header, &reply->ikm_header);
-               kernel_task->messages_received++;
-           }
-           else {
-               if (!ipc_kobject_notify(&request->ikm_header, &reply->ikm_header)){
-#if    MACH_IPC_TEST
-                   printf("ipc_kobject_server: bogus kernel message, id=%d\n",
-                       request->ikm_header.msgh_id);
-#endif /* MACH_IPC_TEST */
-                   _MIG_MSGID_INVALID(request->ikm_header.msgh_id);
-
-                   ((mig_reply_error_t *) &reply->ikm_header)->RetCode
-                       = MIG_BAD_ID;
+               if (ptr) {
+                       /*
+                        * Check if the port is a task port, if its a task port then
+                        * snapshot the task exec token before the mig routine call.
+                        */
+                       if (ikot == IKOT_TASK_CONTROL) {
+                               task = convert_port_to_task_with_exec_token(port, &exec_token, TRUE);
+                       }
+
+#if CONFIG_MACF
+                       int idx = ptr->kobjidx;
+                       task_t curtask = current_task();
+                       uint8_t *filter_mask = curtask->mach_kobj_filter_mask;
+
+                       /* Check kobject mig filter mask, if exists. */
+                       if (__improbable(filter_mask != NULL && idx != KOBJ_IDX_NOT_SET &&
+                           !bitstr_test(filter_mask, idx))) {
+                               /* Not in filter mask, evaluate policy. */
+                               if (mac_task_kobj_msg_evaluate != NULL) {
+                                       kr = mac_task_kobj_msg_evaluate(get_bsdtask_info(curtask),
+                                           request_msgh_id, idx);
+                                       if (kr != KERN_SUCCESS) {
+                                               ((mig_reply_error_t *) reply->ikm_header)->RetCode = kr;
+                                               goto skip_kobjcall;
+                                       }
+                               }
+                       }
+#endif /* CONFIG_MACF */
+
+                       (*ptr->routine)(request->ikm_header, reply->ikm_header);
+
+#if CONFIG_MACF
+skip_kobjcall:
+#endif
+
+                       /* Check if the exec token changed during the mig routine */
+                       if (task != TASK_NULL) {
+                               if (exec_token != task->exec_token) {
+                                       exec_token_changed = TRUE;
+                               }
+                               task_deallocate(task);
+                       }
+
+                       kernel_task->messages_received++;
+               } else {
+                       if (!ipc_kobject_notify(request->ikm_header, reply->ikm_header)) {
+#if DEVELOPMENT || DEBUG
+                               printf("ipc_kobject_server: bogus kernel message, id=%d\n",
+                                   request->ikm_header->msgh_id);
+#endif  /* DEVELOPMENT || DEBUG */
+                               _MIG_MSGID_INVALID(request->ikm_header->msgh_id);
+
+                               ((mig_reply_error_t *) reply->ikm_header)->RetCode
+                                       = MIG_BAD_ID;
+                       } else {
+                               kernel_task->messages_received++;
+                       }
                }
-               else
-                 kernel_task->messages_received++;
-           }
-           kernel_task->messages_sent++;
+               kernel_task->messages_sent++;
        }
 
+       if (!(reply->ikm_header->msgh_bits & MACH_MSGH_BITS_COMPLEX) &&
+           ((mig_reply_error_t *) reply->ikm_header)->RetCode != KERN_SUCCESS) {
+               kr = ((mig_reply_error_t *) reply->ikm_header)->RetCode;
+       } else {
+               kr = KERN_SUCCESS;
+       }
+
+msgdone:
        /*
         *      Destroy destination. The following code differs from
         *      ipc_object_destroy in that we release the send-once
         *      right instead of generating a send-once notification
-        *      (which would bring us here again, creating a loop).
+        *      (which would bring us here again, creating a loop).
         *      It also differs in that we only expect send or
         *      send-once rights, never receive rights.
         *
         *      We set msgh_remote_port to IP_NULL so that the kmsg
         *      destroy routines don't try to destroy the port twice.
         */
-       destp = (ipc_port_t *) &request->ikm_header.msgh_remote_port;
-       switch (MACH_MSGH_BITS_REMOTE(request->ikm_header.msgh_bits)) {
-               case MACH_MSG_TYPE_PORT_SEND:
-                   ipc_port_release_send(*destp);
-                   break;
-               
-               case MACH_MSG_TYPE_PORT_SEND_ONCE:
-                   ipc_port_release_sonce(*destp);
-                   break;
-               
-               default:
-                   panic("ipc_kobject_server: strange destination rights");
+       switch (MACH_MSGH_BITS_REMOTE(request->ikm_header->msgh_bits)) {
+       case MACH_MSG_TYPE_PORT_SEND:
+               ipc_port_release_send(request->ikm_header->msgh_remote_port);
+               request->ikm_header->msgh_remote_port = IP_NULL;
+               break;
+
+       case MACH_MSG_TYPE_PORT_SEND_ONCE:
+               ipc_port_release_sonce(request->ikm_header->msgh_remote_port);
+               request->ikm_header->msgh_remote_port = IP_NULL;
+               break;
+
+       default:
+               panic("ipc_kobject_server: strange destination rights");
        }
-       *destp = IP_NULL;
 
-        if (!(reply->ikm_header.msgh_bits & MACH_MSGH_BITS_COMPLEX) &&
-           ((mig_reply_error_t *) &reply->ikm_header)->RetCode != KERN_SUCCESS)
-               kr = ((mig_reply_error_t *) &reply->ikm_header)->RetCode;
-       else
-               kr = KERN_SUCCESS;
+       /*
+        *      Destroy voucher.  The kernel MIG servers never take ownership
+        *      of vouchers sent in messages.  Swallow any such rights here.
+        */
+       if (IP_VALID(request->ikm_voucher)) {
+               assert(MACH_MSG_TYPE_PORT_SEND ==
+                   MACH_MSGH_BITS_VOUCHER(request->ikm_header->msgh_bits));
+               ipc_port_release_send(request->ikm_voucher);
+               request->ikm_voucher = IP_NULL;
+       }
 
        if ((kr == KERN_SUCCESS) || (kr == MIG_NO_REPLY)) {
                /*
@@ -367,14 +544,13 @@ ipc_kobject_server(
                 *      to free the kmsg.
                 */
                ipc_kmsg_free(request);
-
        } else {
                /*
                 *      The message contents of the request are intact.
                 *      Destroy everthing except the reply port right,
                 *      which is needed in the reply message.
                 */
-               request->ikm_header.msgh_local_port = MACH_PORT_NULL;
+               request->ikm_header->msgh_local_port = MACH_PORT_NULL;
                ipc_kmsg_destroy(request);
        }
 
@@ -384,16 +560,17 @@ ipc_kobject_server(
                 *      using the reply port right, which it has saved.
                 */
 
-               ipc_kmsg_free(reply);
+               if (reply) {
+                       ipc_kmsg_free(reply);
+               }
+               return IKM_NULL;
+       }
 
-               ETAP_PROBE_DATA(ETAP_P_SYSCALL_MACH,
-                               EVENT_END,
-                               ((thread_t) th),
-                               &request->ikm_header.msgh_id,
-                               sizeof(int));
+       if (reply) {
+               replyp = reply->ikm_header->msgh_remote_port;
+       }
 
-               return IKM_NULL;
-       } else if (!IP_VALID((ipc_port_t)reply->ikm_header.msgh_remote_port)) {
+       if (!IP_VALID(replyp)) {
                /*
                 *      Can't queue the reply message if the destination
                 *      (the reply port) isn't valid.
@@ -401,26 +578,72 @@ ipc_kobject_server(
 
                ipc_kmsg_destroy(reply);
 
-               ETAP_PROBE_DATA(ETAP_P_SYSCALL_MACH,
-                               EVENT_END,
-                               ((thread_t) th),
-                               &request->ikm_header.msgh_id,
-                               sizeof(int));
-
+               return IKM_NULL;
+       } else if (replyp->ip_receiver == ipc_space_kernel) {
+               /*
+                * Don't send replies to kobject kernel ports
+                */
+#if DEVELOPMENT || DEBUG
+               printf("%s: refusing to send reply to kobject %d port (id:%d)\n",
+                   __func__, ip_kotype(replyp), request_msgh_id);
+#endif  /* DEVELOPMENT || DEBUG */
+               ipc_kmsg_destroy(reply);
                return IKM_NULL;
        }
 
-       trailer = (mach_msg_format_0_trailer_t *)
-               ((vm_offset_t)&reply->ikm_header + (int)reply->ikm_header.msgh_size);                
-       trailer->msgh_sender = KERNEL_SECURITY_TOKEN;
-       trailer->msgh_trailer_type = MACH_MSG_TRAILER_FORMAT_0;
-       trailer->msgh_trailer_size = MACH_MSG_TRAILER_MINIMUM_SIZE;
+       /* Fail the MIG call if the task exec token changed during the call */
+       if (kr == KERN_SUCCESS && exec_token_changed) {
+               /*
+                *      Create a new reply msg with error and destroy the old reply msg.
+                */
+               ipc_kmsg_t new_reply = ipc_kmsg_alloc(reply_size);
 
-        ETAP_PROBE_DATA(ETAP_P_SYSCALL_MACH,
-                        EVENT_END,
-                       ((thread_t) th),
-                        &request->ikm_header.msgh_id,
-                        sizeof(int));
+               if (new_reply == IKM_NULL) {
+                       printf("ipc_kobject_server: dropping request\n");
+                       ipc_kmsg_destroy(reply);
+                       return IKM_NULL;
+               }
+               /*
+                *      Initialize the new reply message.
+                */
+               {
+#define OutP_new        ((mig_reply_error_t *) new_reply->ikm_header)
+#define OutP_old        ((mig_reply_error_t *) reply->ikm_header)
+
+                       bzero((void *)OutP_new, reply_size);
+
+                       OutP_new->NDR = OutP_old->NDR;
+                       OutP_new->Head.msgh_size = sizeof(mig_reply_error_t);
+                       OutP_new->Head.msgh_bits = OutP_old->Head.msgh_bits & ~MACH_MSGH_BITS_COMPLEX;
+                       OutP_new->Head.msgh_remote_port = OutP_old->Head.msgh_remote_port;
+                       OutP_new->Head.msgh_local_port = MACH_PORT_NULL;
+                       OutP_new->Head.msgh_voucher_port = MACH_PORT_NULL;
+                       OutP_new->Head.msgh_id = OutP_old->Head.msgh_id;
+
+                       /* Set the error as KERN_INVALID_TASK */
+                       OutP_new->RetCode = KERN_INVALID_TASK;
+
+#undef  OutP_new
+#undef  OutP_old
+               }
+
+               /*
+                *      Destroy everything in reply except the reply port right,
+                *      which is needed in the new reply message.
+                */
+               reply->ikm_header->msgh_remote_port = MACH_PORT_NULL;
+               ipc_kmsg_destroy(reply);
+
+               reply = new_reply;
+       }
+
+       trailer = (mach_msg_max_trailer_t *)
+           ((vm_offset_t)reply->ikm_header + (int)reply->ikm_header->msgh_size);
+       bzero(trailer, sizeof(*trailer));
+       trailer->msgh_sender = KERNEL_SECURITY_TOKEN;
+       trailer->msgh_audit = KERNEL_AUDIT_TOKEN;
+       trailer->msgh_trailer_type = MACH_MSG_TRAILER_FORMAT_0;
+       trailer->msgh_trailer_size = MACH_MSG_TRAILER_MINIMUM_SIZE;
 
        return reply;
 }
@@ -438,9 +661,9 @@ ipc_kobject_server(
  */
 void
 ipc_kobject_set(
-       ipc_port_t                      port,
-       ipc_kobject_t           kobject,
-       ipc_kobject_type_t      type)
+       ipc_port_t                      port,
+       ipc_kobject_t           kobject,
+       ipc_kobject_type_t      type)
 {
        ip_lock(port);
        ipc_kobject_set_atomically(port, kobject, type);
@@ -449,38 +672,349 @@ ipc_kobject_set(
 
 void
 ipc_kobject_set_atomically(
-       ipc_port_t                      port,
-       ipc_kobject_t           kobject,
-       ipc_kobject_type_t      type)
+       ipc_port_t                      port,
+       ipc_kobject_t           kobject,
+       ipc_kobject_type_t      type)
 {
        assert(type == IKOT_NONE || ip_active(port));
-#if    MACH_ASSERT
-       port->ip_spares[2] = (port->ip_bits & IO_BITS_KOTYPE);
-#endif /* MACH_ASSERT */
-       port->ip_bits = (port->ip_bits &~ IO_BITS_KOTYPE) | type;
-       port->ip_kobject = kobject;
+#if     MACH_ASSERT
+       port->ip_spares[2] = (port->ip_object.io_bits & IO_BITS_KOTYPE);
+#endif  /* MACH_ASSERT */
+       port->ip_object.io_bits = (port->ip_object.io_bits & ~IO_BITS_KOTYPE) | type;
+       if (ip_is_kolabeled(port)) {
+               ipc_kobject_label_t labelp = port->ip_kolabel;
+               labelp->ikol_kobject = kobject;
+       } else {
+               port->ip_kobject = kobject;
+       }
+       if (type != IKOT_NONE) {
+               /* Once set, this bit can never be unset */
+               port->ip_object.io_bits |= IO_BITS_KOBJECT;
+       }
+}
+
+/*
+ *     Routine:        ipc_kobject_init_port
+ *     Purpose:
+ *             Initialize a kobject port with the given types and options.
+ *
+ *             This function never fails.
+ */
+static inline void
+ipc_kobject_init_port(
+       ipc_port_t port,
+       ipc_kobject_t kobject,
+       ipc_kobject_type_t type,
+       ipc_kobject_alloc_options_t options)
+{
+       ipc_kobject_set_atomically(port, kobject, type);
+
+       if (options & IPC_KOBJECT_ALLOC_MAKE_SEND) {
+               ipc_port_make_send_locked(port);
+       }
+       if (options & IPC_KOBJECT_ALLOC_NSREQUEST) {
+               ipc_port_make_sonce_locked(port);
+               port->ip_nsrequest = port;
+       }
+       if (options & IPC_KOBJECT_ALLOC_NO_GRANT) {
+               port->ip_no_grant = 1;
+       }
+       if (options & IPC_KOBJECT_ALLOC_IMMOVABLE_SEND) {
+               port->ip_immovable_send = 1;
+       }
+       if (options & IPC_KOBJECT_ALLOC_PINNED) {
+               port->ip_pinned = 1;
+       }
+}
+
+/*
+ *     Routine:        ipc_kobject_alloc_port
+ *     Purpose:
+ *             Allocate a kobject port in the kernel space of the specified type.
+ *
+ *             This function never fails.
+ *
+ *     Conditions:
+ *             No locks held (memory is allocated)
+ */
+ipc_port_t
+ipc_kobject_alloc_port(
+       ipc_kobject_t           kobject,
+       ipc_kobject_type_t      type,
+       ipc_kobject_alloc_options_t     options)
+{
+       ipc_port_t port = ipc_port_alloc_kernel();
+
+       if (port == IP_NULL) {
+               panic("ipc_kobject_alloc_port(): failed to allocate port");
+       }
+
+       ipc_kobject_init_port(port, kobject, type, options);
+       return port;
+}
+
+/*
+ *     Routine:        ipc_kobject_alloc_labeled_port
+ *     Purpose:
+ *             Allocate a kobject port and associated mandatory access label
+ *             in the kernel space of the specified type.
+ *
+ *             This function never fails.
+ *
+ *     Conditions:
+ *             No locks held (memory is allocated)
+ */
+
+ipc_port_t
+ipc_kobject_alloc_labeled_port(
+       ipc_kobject_t           kobject,
+       ipc_kobject_type_t      type,
+       ipc_label_t             label,
+       ipc_kobject_alloc_options_t     options)
+{
+       ipc_port_t port;
+       ipc_kobject_label_t labelp;
+
+       port = ipc_port_alloc_kernel();
+       if (port == IP_NULL) {
+               panic("ipc_kobject_alloc_port(): failed to allocate port");
+       }
+
+       labelp = (ipc_kobject_label_t)zalloc(ipc_kobject_label_zone);
+       if (labelp == NULL) {
+               panic("ipc_kobject_alloc_labeled_port(): failed to allocate label");
+       }
+       labelp->ikol_label = label;
+       port->ip_kolabel = labelp;
+       port->ip_object.io_bits |= IO_BITS_KOLABEL;
+
+       ipc_kobject_init_port(port, kobject, type, options);
+       return port;
+}
+
+static void
+ipc_kobject_subst_once_notify(mach_msg_header_t *msg)
+{
+       mach_no_senders_notification_t *notification = (void *)msg;
+       ipc_port_t port = notification->not_header.msgh_remote_port;
+
+       require_ip_active(port);
+       assert(IKOT_PORT_SUBST_ONCE == ip_kotype(port));
+
+       ip_release((ipc_port_t)ip_get_kobject(port));
+       ipc_port_dealloc_kernel(port);
 }
 
+/*
+ *     Routine:        ipc_kobject_alloc_subst_once
+ *     Purpose:
+ *             Make a port that will be substituted by the kolabel
+ *             rules once, preventing the next substitution (of its target)
+ *             to happen if any.
+ *
+ *     Returns:
+ *             A port with a send right, that will substitute to its "kobject".
+ *
+ *     Conditions:
+ *             No locks held (memory is allocated)
+ *             `target` has a refcount that this function consumes
+ */
+ipc_port_t
+ipc_kobject_alloc_subst_once(
+       ipc_port_t                  target)
+{
+       return ipc_kobject_alloc_labeled_port(target,
+                  IKOT_PORT_SUBST_ONCE, IPC_LABEL_SUBST_ONCE,
+                  IPC_KOBJECT_ALLOC_MAKE_SEND | IPC_KOBJECT_ALLOC_NSREQUEST);
+}
+
+/*
+ *     Routine:        ipc_kobject_make_send_lazy_alloc_port
+ *     Purpose:
+ *             Make a send once for a kobject port.
+ *
+ *             A location owning this port is passed in port_store.
+ *             If no port exists, a port is made lazily.
+ *
+ *             A send right is made for the port, and if this is the first one
+ *             (possibly not for the first time), then the no-more-senders
+ *             notification is rearmed.
+ *
+ *             When a notification is armed, the kobject must donate
+ *             one of its references to the port. It is expected
+ *             the no-more-senders notification will consume this reference.
+ *
+ *     Returns:
+ *             TRUE if a notification was armed
+ *             FALSE else
+ *
+ *     Conditions:
+ *             Nothing is locked, memory can be allocated.
+ *             The caller must be able to donate a kobject reference to the port.
+ */
+boolean_t
+ipc_kobject_make_send_lazy_alloc_port(
+       ipc_port_t              *port_store,
+       ipc_kobject_t           kobject,
+       ipc_kobject_type_t      type,
+       ipc_kobject_alloc_options_t alloc_opts,
+       boolean_t               __ptrauth_only should_ptrauth,
+       uint64_t                __ptrauth_only ptrauth_discriminator)
+{
+       ipc_port_t port, previous, __ptrauth_only port_addr;
+       boolean_t rc = FALSE;
+
+       port = os_atomic_load(port_store, dependency);
+
+#if __has_feature(ptrauth_calls)
+       /* If we're on a ptrauth system and this port is signed, authenticate and strip the pointer */
+       if (should_ptrauth && IP_VALID(port)) {
+               port = ptrauth_auth_data(port,
+                   ptrauth_key_process_independent_data,
+                   ptrauth_blend_discriminator(port_store, ptrauth_discriminator));
+       }
+#endif // __has_feature(ptrauth_calls)
+
+       if (!IP_VALID(port)) {
+               port = ipc_kobject_alloc_port(kobject, type,
+                   IPC_KOBJECT_ALLOC_MAKE_SEND | IPC_KOBJECT_ALLOC_NSREQUEST | alloc_opts);
+
+#if __has_feature(ptrauth_calls)
+               if (should_ptrauth) {
+                       port_addr = ptrauth_sign_unauthenticated(port,
+                           ptrauth_key_process_independent_data,
+                           ptrauth_blend_discriminator(port_store, ptrauth_discriminator));
+               } else {
+                       port_addr = port;
+               }
+#else
+               port_addr = port;
+#endif // __has_feature(ptrauth_calls)
+
+               if (os_atomic_cmpxchgv(port_store, IP_NULL, port_addr, &previous, release)) {
+                       return TRUE;
+               }
+
+               // undo what ipc_kobject_alloc_port() did above
+               port->ip_nsrequest = IP_NULL;
+               port->ip_mscount = 0;
+               port->ip_sorights = 0;
+               port->ip_srights = 0;
+               ip_release(port);
+               ip_release(port);
+               ipc_port_dealloc_kernel(port);
+
+               port = previous;
+       }
+
+       ip_lock(port);
+       ipc_port_make_send_locked(port);
+       if (port->ip_srights == 1) {
+               ipc_port_make_sonce_locked(port);
+               assert(port->ip_nsrequest == IP_NULL);
+               port->ip_nsrequest = port;
+               rc = TRUE;
+       }
+       ip_unlock(port);
+
+       return rc;
+}
+
+/*
+ *     Routine:        ipc_kobject_make_send_lazy_alloc_labeled_port
+ *     Purpose:
+ *             Make a send once for a kobject port.
+ *
+ *             A location owning this port is passed in port_store.
+ *             If no port exists, a port is made lazily.
+ *
+ *             A send right is made for the port, and if this is the first one
+ *             (possibly not for the first time), then the no-more-senders
+ *             notification is rearmed.
+ *
+ *             When a notification is armed, the kobject must donate
+ *             one of its references to the port. It is expected
+ *             the no-more-senders notification will consume this reference.
+ *
+ *     Returns:
+ *             TRUE if a notification was armed
+ *             FALSE else
+ *
+ *     Conditions:
+ *             Nothing is locked, memory can be allocated.
+ *             The caller must be able to donate a kobject reference to the port.
+ */
+boolean_t
+ipc_kobject_make_send_lazy_alloc_labeled_port(
+       ipc_port_t              *port_store,
+       ipc_kobject_t           kobject,
+       ipc_kobject_type_t      type,
+       ipc_label_t             label)
+{
+       ipc_port_t port, previous;
+       boolean_t rc = FALSE;
+
+       port = os_atomic_load(port_store, dependency);
+
+       if (!IP_VALID(port)) {
+               port = ipc_kobject_alloc_labeled_port(kobject, type, label,
+                   IPC_KOBJECT_ALLOC_MAKE_SEND | IPC_KOBJECT_ALLOC_NSREQUEST);
+               if (os_atomic_cmpxchgv(port_store, IP_NULL, port, &previous, release)) {
+                       return TRUE;
+               }
+
+               // undo what ipc_kobject_alloc_port() did above
+               port->ip_nsrequest = IP_NULL;
+               port->ip_mscount = 0;
+               port->ip_sorights = 0;
+               port->ip_srights = 0;
+               ip_release(port);
+               ip_release(port);
+               zfree(ipc_kobject_label_zone, port->ip_kolabel);
+               port->ip_object.io_bits &= ~IO_BITS_KOLABEL;
+               port->ip_kolabel = NULL;
+               ipc_port_dealloc_kernel(port);
+
+               port = previous;
+               assert(ip_is_kolabeled(port));
+       }
+
+       ip_lock(port);
+       ipc_port_make_send_locked(port);
+       if (port->ip_srights == 1) {
+               ipc_port_make_sonce_locked(port);
+               assert(port->ip_nsrequest == IP_NULL);
+               port->ip_nsrequest = port;
+               rc = TRUE;
+       }
+       ip_unlock(port);
+
+       return rc;
+}
+
+
 /*
  *     Routine:        ipc_kobject_destroy
  *     Purpose:
  *             Release any kernel object resources associated
  *             with the port, which is being destroyed.
  *
- *             This should only be needed when resources are
- *             associated with a user's port.  In the normal case,
- *             when the kernel is the receiver, the code calling
- *             ipc_port_dealloc_kernel should clean up the resources.
+ *             This path to free object resources should only be
+ *      needed when resources are associated with a user's port.
+ *      In the normal case, when the kernel is the receiver,
+ *      the code calling ipc_port_dealloc_kernel should clean
+ *      up the object resources.
+ *
+ *      Cleans up any kobject label that might be present.
  *     Conditions:
  *             The port is not locked, but it is dead.
  */
 
 void
 ipc_kobject_destroy(
-       ipc_port_t              port)
+       ipc_port_t              port)
 {
        switch (ip_kotype(port)) {
-
        case IKOT_TIMER:
                mk_timer_port_destroy(port);
                break;
@@ -489,27 +1023,231 @@ ipc_kobject_destroy(
                mach_destroy_memory_entry(port);
                break;
 
-       default:        /* XXX (bogon) */
+       case IKOT_HOST_NOTIFY:
+               host_notify_port_destroy(port);
+               break;
+
+       case IKOT_SUID_CRED:
+               suid_cred_destroy(port);
                break;
+
+       default:
+               break;
+       }
+
+       if (ip_is_kolabeled(port)) {
+               ipc_kobject_label_t labelp = port->ip_kolabel;
+
+               assert(labelp != NULL);
+               assert(ip_is_kobject(port));
+               port->ip_kolabel = NULL;
+               port->ip_object.io_bits &= ~IO_BITS_KOLABEL;
+               zfree(ipc_kobject_label_zone, labelp);
+       }
+}
+
+/*
+ *     Routine:        ipc_kobject_label_substitute_task
+ *     Purpose:
+ *             Substitute a task control port for its immovable
+ *             equivalent when the receiver is that task.
+ *     Conditions:
+ *             Space is write locked and active.
+ *             Port is locked and active.
+ *     Returns:
+ *             - IP_NULL port if no substitution is to be done
+ *             - a valid port if a substitution needs to happen
+ */
+static ipc_port_t
+ipc_kobject_label_substitute_task(
+       ipc_space_t             space,
+       ipc_port_t              port)
+{
+       ipc_port_t subst = IP_NULL;
+       task_t task = ipc_kobject_get(port);
+
+       if (task != TASK_NULL && task == space->is_task) {
+               if ((subst = port->ip_alt_port)) {
+                       return subst;
+               }
+       }
+
+       return IP_NULL;
+}
+
+/*
+ *     Routine:        ipc_kobject_label_substitute_thread
+ *     Purpose:
+ *             Substitute a thread control port for its immovable
+ *             equivalent when it belongs to the receiver task.
+ *     Conditions:
+ *             Space is write locked and active.
+ *             Port is locked and active.
+ *     Returns:
+ *             - IP_NULL port if no substitution is to be done
+ *             - a valid port if a substitution needs to happen
+ */
+static ipc_port_t
+ipc_kobject_label_substitute_thread(
+       ipc_space_t             space,
+       ipc_port_t              port)
+{
+       ipc_port_t subst = IP_NULL;
+       thread_t thread = ipc_kobject_get(port);
+
+       if (thread != THREAD_NULL && space->is_task == thread->task) {
+               if ((subst = port->ip_alt_port) != IP_NULL) {
+                       return subst;
+               }
        }
+
+       return IP_NULL;
 }
 
+/*
+ *     Routine:        ipc_kobject_label_check
+ *     Purpose:
+ *             Check to see if the space is allowed to possess
+ *             a right for the given port. In order to qualify,
+ *             the space label must contain all the privileges
+ *             listed in the port/kobject label.
+ *
+ *     Conditions:
+ *             Space is write locked and active.
+ *             Port is locked and active.
+ *
+ *     Returns:
+ *             Whether the copyout is authorized.
+ *
+ *             If a port substitution is requested, the space is unlocked,
+ *             the port is unlocked and its "right" consumed.
+ *
+ *             As of now, substituted ports only happen for send rights.
+ */
+bool
+ipc_kobject_label_check(
+       ipc_space_t                     space,
+       ipc_port_t                      port,
+       mach_msg_type_name_t            msgt_name,
+       ipc_object_copyout_flags_t     *flags,
+       ipc_port_t                     *subst_portp)
+{
+       ipc_kobject_label_t labelp;
+       ipc_label_t label;
+
+       assert(is_active(space));
+       assert(ip_active(port));
+
+       *subst_portp = IP_NULL;
+
+       /* Unlabled ports/kobjects are always allowed */
+       if (!ip_is_kolabeled(port)) {
+               return true;
+       }
+
+       /* Never OK to copyout the receive right for a labeled kobject */
+       if (msgt_name == MACH_MSG_TYPE_PORT_RECEIVE) {
+               panic("ipc_kobject_label_check: attempted receive right "
+                   "copyout for labeled kobject");
+       }
 
-extern int vnode_pager_workaround;
+       labelp = port->ip_kolabel;
+       label = labelp->ikol_label;
+
+       if ((*flags & IPC_OBJECT_COPYOUT_FLAGS_NO_LABEL_CHECK) == 0 &&
+           (label & IPC_LABEL_SUBST_MASK)) {
+               ipc_port_t subst = IP_NULL;
+
+               if (msgt_name != MACH_MSG_TYPE_PORT_SEND) {
+                       return false;
+               }
+
+               switch (label & IPC_LABEL_SUBST_MASK) {
+               case IPC_LABEL_SUBST_TASK:
+                       subst = ipc_kobject_label_substitute_task(space, port);
+                       break;
+               case IPC_LABEL_SUBST_THREAD:
+                       subst = ipc_kobject_label_substitute_thread(space, port);
+                       break;
+               case IPC_LABEL_SUBST_ONCE:
+                       /* the next check will _not_ substitute */
+                       *flags |= IPC_OBJECT_COPYOUT_FLAGS_NO_LABEL_CHECK;
+                       subst = ip_get_kobject(port);
+                       break;
+               default:
+                       panic("unexpected label: %llx\n", label);
+               }
+
+               if (subst != IP_NULL) {
+                       ip_reference(subst);
+                       is_write_unlock(space);
+                       ipc_port_release_send_and_unlock(port);
+                       port = ipc_port_make_send(subst);
+                       ip_release(subst);
+                       *subst_portp = port;
+                       return true;
+               }
+       }
+
+       return (label & space->is_label & IPC_LABEL_SPACE_MASK) ==
+              (label & IPC_LABEL_SPACE_MASK);
+}
 
 boolean_t
 ipc_kobject_notify(
        mach_msg_header_t *request_header,
        mach_msg_header_t *reply_header)
 {
-       ipc_port_t port = (ipc_port_t) request_header->msgh_remote_port;
-       mig_subsystem_t         paging_subsystem_object;
-       mach_port_seqno_t       seqno;
+       mach_msg_max_trailer_t * trailer;
+       ipc_port_t port = request_header->msgh_remote_port;
 
        ((mig_reply_error_t *) reply_header)->RetCode = MIG_NO_REPLY;
+
+       trailer = (mach_msg_max_trailer_t *)
+           ((vm_offset_t)request_header + request_header->msgh_size);
+
+       /*
+        * The kobject notification is privileged and can change the
+        * refcount on kernel-internal objects - make sure
+        * that the message wasn't faked!
+        */
+       if (0 != bcmp(&trailer->msgh_audit, &KERNEL_AUDIT_TOKEN,
+           sizeof(trailer->msgh_audit))) {
+               return FALSE;
+       }
+       if (0 != bcmp(&trailer->msgh_sender, &KERNEL_SECURITY_TOKEN,
+           sizeof(trailer->msgh_sender))) {
+               return FALSE;
+       }
+
        switch (request_header->msgh_id) {
-               case MACH_NOTIFY_NO_SENDERS:
-                  if(ip_kotype(port) == IKOT_NAMED_ENTRY) {
+       case MACH_NOTIFY_NO_SENDERS:
+               switch (ip_kotype(port)) {
+               case IKOT_VOUCHER:
+                       ipc_voucher_notify(request_header);
+                       return TRUE;
+
+               case IKOT_VOUCHER_ATTR_CONTROL:
+                       ipc_voucher_attr_control_notify(request_header);
+                       return TRUE;
+
+               case IKOT_PORT_SUBST_ONCE:
+                       ipc_kobject_subst_once_notify(request_header);
+                       return TRUE;
+
+               case IKOT_SEMAPHORE:
+                       semaphore_notify(request_header);
+                       return TRUE;
+
+               case IKOT_EVENTLINK:
+                       ipc_eventlink_notify(request_header);
+                       return TRUE;
+
+               case IKOT_TASK_CONTROL:
+                       task_port_notify(request_header);
+                       return TRUE;
+
+               case IKOT_NAMED_ENTRY:
                        ip_lock(port);
 
                        /*
@@ -518,109 +1256,78 @@ ipc_kobject_notify(
                         */
                        port->ip_mscount = 0;
                        port->ip_messages.imq_seqno = 0;
-                       ipc_port_destroy(port); /* releases lock */
+                       ipc_port_destroy(port);         /* releases lock */
                        return TRUE;
-                  }
-                  if (ip_kotype(port) == IKOT_UPL) {
-                          upl_no_senders(
-                               (ipc_port_t)request_header->msgh_remote_port, 
-                               (mach_port_mscount_t) 
-                               ((mach_no_senders_notification_t *) 
-                                request_header)->not_count);
-                          (ipc_port_t)reply_header->msgh_remote_port
-                                  = MACH_PORT_NULL;
-                          return TRUE;
-                  }
-                               
-                  break;
-
-               case MACH_NOTIFY_PORT_DELETED:
-               case MACH_NOTIFY_PORT_DESTROYED:
-               case MACH_NOTIFY_SEND_ONCE:
-               case MACH_NOTIFY_DEAD_NAME:
-               break;
-
-               default:
-               return FALSE;
-       }
-       switch (ip_kotype(port)) {
 
-#ifdef IOKIT
-               case IKOT_IOKIT_OBJECT:
-               case IKOT_IOKIT_CONNECT:
-               case IKOT_IOKIT_SPARE:
-               {
-                extern boolean_t iokit_notify( mach_msg_header_t *msg);
+               case IKOT_UPL:
+                       upl_no_senders(
+                               request_header->msgh_remote_port,
+                               (mach_port_mscount_t)
+                               ((mach_no_senders_notification_t *)
+                               request_header)->not_count);
+                       reply_header->msgh_remote_port = MACH_PORT_NULL;
+                       return TRUE;
 
-                return iokit_notify(request_header);
-               }
+#if     CONFIG_AUDIT
+               case IKOT_AU_SESSIONPORT:
+                       audit_session_nosenders(request_header);
+                       return TRUE;
 #endif
-               default:
-                return FALSE;
-        }
-}
-
-
-
-#include <mach_kdb.h>
-#if    MACH_COUNTERS && MACH_KDB
-
-#include <ddb/db_output.h>
-#include <ddb/db_sym.h>
-
-#define printf  kdbprintf
+               case IKOT_FILEPORT:
+                       fileport_notify(request_header);
+                       return TRUE;
 
-extern void kobjserver_stats(void);
-extern void bucket_stats_print(mig_hash_t *bucket);
+               case IKOT_WORK_INTERVAL:
+                       work_interval_port_notify(request_header);
+                       return TRUE;
+               case IKOT_TASK_READ:
+               case IKOT_TASK_INSPECT:
+                       task_port_with_flavor_notify(request_header);
+                       return TRUE;
+               case IKOT_THREAD_READ:
+               case IKOT_THREAD_INSPECT:
+                       thread_port_with_flavor_notify(request_header);
+                       return TRUE;
+               case IKOT_SUID_CRED:
+                       suid_cred_notify(request_header);
+                       return TRUE;
+               case IKOT_TASK_ID_TOKEN:
+                       task_id_token_notify(request_header);
+                       return TRUE;
+#if HYPERVISOR
+               case IKOT_HYPERVISOR:
+                       hv_port_notify(request_header);
+                       return TRUE;
+#endif
+               }
 
-extern void kobjserver_stats_clear(void);
+               break;
 
+       case MACH_NOTIFY_PORT_DELETED:
+       case MACH_NOTIFY_PORT_DESTROYED:
+       case MACH_NOTIFY_SEND_ONCE:
+       case MACH_NOTIFY_DEAD_NAME:
+               break;
 
-void
-kobjserver_stats_clear(void)
-{
-       int i;
-       for (i = 0; i < MAX_MIG_ENTRIES; i++) {
-               mig_buckets[i].callcount = 0;
+       default:
+               return FALSE;
+       }
+       switch (ip_kotype(port)) {
+#ifdef IOKIT
+       case IKOT_IOKIT_OBJECT:
+       case IKOT_IOKIT_CONNECT:
+       case IKOT_IOKIT_IDENT:
+       case IKOT_UEXT_OBJECT:
+       {
+               return iokit_notify(request_header);
+       }
+#endif
+       case IKOT_TASK_RESUME:
+       {
+               return task_suspension_notify(request_header);
        }
-}
-
-void
-kobjserver_stats(void)
-{
-    register unsigned int i, n = sizeof(mig_e)/sizeof(mig_subsystem_t);
-    register unsigned int howmany;
-    register mach_msg_id_t j, pos, nentry, range;
-       
-    db_printf("Kobject server call counts:\n");
-    for (i = 0; i < n; i++) {
-       db_printf("  ");
-       db_printsym((vm_offset_t)mig_e[i], DB_STGY_ANY);
-       db_printf(":\n");
-       range = mig_e[i]->end - mig_e[i]->start;
-       if (!mig_e[i]->start || range < 0) continue;
-
-       for  (j = 0; j < range; j++) {
-           nentry = j + mig_e[i]->start;       
-           for (pos = MIG_HASH(nentry) % MAX_MIG_ENTRIES, howmany = 1;
-                mig_buckets[pos].num;
-                pos = ++pos % MAX_MIG_ENTRIES, howmany++) {
-                   if (mig_buckets[pos].num == nentry)
-                       bucket_stats_print(&mig_buckets[pos]);
-           }
-       }
-    }
-}
 
-void
-bucket_stats_print(mig_hash_t *bucket)
-{
-       if (bucket->callcount) {
-               db_printf("    ");
-               db_printsym((vm_offset_t)bucket->routine, DB_STGY_ANY);
-               db_printf(" (%d):\t%d\n", bucket->num, bucket->callcount);
+       default:
+               return FALSE;
        }
 }
-
-
-#endif /* MACH_COUNTERS && MACH_KDB */