]> git.saurik.com Git - apple/xnu.git/blobdiff - osfmk/kern/task.c
xnu-2050.22.13.tar.gz
[apple/xnu.git] / osfmk / kern / task.c
index 2a2fde2d817ef06099385731d0f32ca479a6d1bd..1378bf4150bc3c606fd76dda90ebeb7483e30b68 100644 (file)
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
+ * Copyright (c) 2000-2010 Apple Inc. All rights reserved.
  *
  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
  * 
@@ -86,7 +86,6 @@
  * Copyright (c) 2005 SPARTA, Inc.
  */
 
-#include <mach_kdb.h>
 #include <fast_tas.h>
 #include <platforms.h>
 
 #include <kern/processor.h>
 #include <kern/sched_prim.h>   /* for thread_wakeup */
 #include <kern/ipc_tt.h>
-#include <kern/ledger.h>
 #include <kern/host.h>
 #include <kern/clock.h>
 #include <kern/timer.h>
 #include <vm/vm_pageout.h>
 #include <vm/vm_protos.h>
 
-#if    MACH_KDB
-#include <ddb/db_sym.h>
-#endif /* MACH_KDB */
-
-#ifdef __ppc__
-#include <ppc/exception.h>
-#include <ppc/hw_perfmon.h>
-#endif
-
 /*
  * Exported interfaces
  */
 #include <security/mac_mach_internal.h>
 #endif
 
-task_t kernel_task;
-zone_t task_zone;
+#if CONFIG_COUNTERS
+#include <pmc/pmc.h>
+#endif /* CONFIG_COUNTERS */
+
+task_t                 kernel_task;
+zone_t                 task_zone;
+lck_attr_t      task_lck_attr;
+lck_grp_t       task_lck_grp;
+lck_grp_attr_t  task_lck_grp_attr;
+#if CONFIG_EMBEDDED
+lck_mtx_t      task_watch_mtx;
+#endif /* CONFIG_EMBEDDED */
+
+zinfo_usage_store_t tasks_tkm_private;
+zinfo_usage_store_t tasks_tkm_shared;
+
+static ledger_template_t task_ledger_template = NULL;
+struct _task_ledger_indices task_ledgers = {-1, -1, -1, -1, -1};
+void init_task_ledgers(void);
+
+
+int task_max = CONFIG_TASK_MAX; /* Max number of tasks */
+
+/* externs for BSD kernel */
+extern void proc_getexecutableuuid(void *, unsigned char *, unsigned long);
 
 /* Forwards */
 
 void           task_hold_locked(
                        task_t          task);
 void           task_wait_locked(
-                       task_t          task);
+                       task_t          task,
+                       boolean_t       until_not_runnable);
 void           task_release_locked(
                        task_t          task);
 void           task_free(
@@ -168,10 +181,8 @@ void               task_free(
 void           task_synchronizer_destroy_all(
                        task_t          task);
 
-kern_return_t  task_set_ledger(
-                       task_t          task,
-                       ledger_t        wired,
-                       ledger_t        paged);
+int check_for_tasksuspend(
+                       task_t task);
 
 void
 task_backing_store_privileged(
@@ -189,7 +200,7 @@ task_set_64bit(
                task_t task,
                boolean_t is64bit)
 {
-#ifdef __i386__
+#if defined(__i386__) || defined(__x86_64__)
        thread_t thread;
 #endif /* __i386__ */
        int     vm_flags = 0;
@@ -213,18 +224,6 @@ task_set_64bit(
                                     (vm_map_offset_t) VM_MAX_ADDRESS,
                                     MACH_VM_MAX_ADDRESS,
                                     0);
-#ifdef __ppc__
-               /* LP64todo - make this clean */
-               /*
-                * PPC51: ppc64 is limited to 51-bit addresses.
-                * Memory mapped above that limit is handled specially
-                * at the pmap level, so let pmap clean the commpage mapping
-                * explicitly...
-                */
-               pmap_unmap_sharedpage(task->map->pmap); /* Unmap commpage */
-               /* ... and avoid regular pmap cleanup */
-               vm_flags |= VM_MAP_REMOVE_NO_PMAP_CLEANUP;
-#endif /* __ppc__ */
                /* remove the higher VM mappings */
                (void) vm_map_remove(task->map,
                                     MACH_VM_MAX_ADDRESS,
@@ -238,30 +237,62 @@ task_set_64bit(
         * certain routines may observe the thread as being in an inconsistent
         * state with respect to its task's 64-bitness.
         */
-#ifdef __i386__
+#if defined(__i386__) || defined(__x86_64__)
+       task_lock(task);
        queue_iterate(&task->threads, thread, thread_t, task_threads) {
+               thread_mtx_lock(thread);
                machine_thread_switch_addrmode(thread);
+               thread_mtx_unlock(thread);
        }
+       task_unlock(task);
 #endif /* __i386__ */
 }
 
+
+void
+task_set_dyld_info(task_t task, mach_vm_address_t addr, mach_vm_size_t size)
+{
+       task_lock(task);
+       task->all_image_info_addr = addr;
+       task->all_image_info_size = size;
+       task_unlock(task);
+}
+
 void
 task_init(void)
 {
+
+       lck_grp_attr_setdefault(&task_lck_grp_attr);
+       lck_grp_init(&task_lck_grp, "task", &task_lck_grp_attr);
+       lck_attr_setdefault(&task_lck_attr);
+       lck_mtx_init(&tasks_threads_lock, &task_lck_grp, &task_lck_attr);
+#if CONFIG_EMBEDDED
+       lck_mtx_init(&task_watch_mtx, &task_lck_grp, &task_lck_attr);
+#endif /* CONFIG_EMBEDDED */
+
        task_zone = zinit(
                        sizeof(struct task),
-                       TASK_MAX * sizeof(struct task),
+                       task_max * sizeof(struct task),
                        TASK_CHUNK * sizeof(struct task),
                        "tasks");
 
+       zone_change(task_zone, Z_NOENCRYPT, TRUE);
+
+       init_task_ledgers();
+
        /*
         * Create the kernel task as the first task.
         */
+#ifdef __LP64__
+       if (task_create_internal(TASK_NULL, FALSE, TRUE, &kernel_task) != KERN_SUCCESS)
+#else
        if (task_create_internal(TASK_NULL, FALSE, FALSE, &kernel_task) != KERN_SUCCESS)
+#endif
                panic("task_init\n");
 
        vm_map_deallocate(kernel_task->map);
        kernel_task->map = kernel_map;
+
 }
 
 /*
@@ -320,6 +351,36 @@ host_security_create_task_token(
        return(KERN_FAILURE);
 }
 
+void
+init_task_ledgers(void)
+{
+       ledger_template_t t;
+       
+       assert(task_ledger_template == NULL);
+       assert(kernel_task == TASK_NULL);
+
+       if ((t = ledger_template_create("Per-task ledger")) == NULL)
+               panic("couldn't create task ledger template");
+
+       task_ledgers.cpu_time = ledger_entry_add(t, "cpu_time", "sched", "ns");
+       task_ledgers.tkm_private = ledger_entry_add(t, "tkm_private",
+           "physmem", "bytes");
+       task_ledgers.tkm_shared = ledger_entry_add(t, "tkm_shared", "physmem",
+           "bytes");
+       task_ledgers.phys_mem = ledger_entry_add(t, "phys_mem", "physmem",
+           "bytes");
+       task_ledgers.wired_mem = ledger_entry_add(t, "wired_mem", "physmem",
+           "bytes");
+
+       if ((task_ledgers.cpu_time < 0) || (task_ledgers.tkm_private < 0) ||
+           (task_ledgers.tkm_shared < 0) || (task_ledgers.phys_mem < 0) ||
+           (task_ledgers.wired_mem < 0)) {
+               panic("couldn't create entries for task ledger template");
+       }
+
+       task_ledger_template = t;
+}
+
 kern_return_t
 task_create_internal(
        task_t          parent_task,
@@ -329,6 +390,7 @@ task_create_internal(
 {
        task_t                  new_task;
        vm_shared_region_t      shared_region;
+       ledger_t                ledger = NULL;
 
        new_task = (task_t) zalloc(task_zone);
 
@@ -338,18 +400,28 @@ task_create_internal(
        /* one ref for just being alive; one for our caller */
        new_task->ref_count = 2;
 
+       /* allocate with active entries */
+       assert(task_ledger_template != NULL);
+       if ((ledger = ledger_instantiate(task_ledger_template,
+                       LEDGER_CREATE_ACTIVE_ENTRIES)) == NULL) {
+               zfree(task_zone, new_task);
+               return(KERN_RESOURCE_SHORTAGE);
+       }
+       new_task->ledger = ledger;
+
+       /* if inherit_memory is true, parent_task MUST not be NULL */
        if (inherit_memory)
-               new_task->map = vm_map_fork(parent_task->map);
+               new_task->map = vm_map_fork(ledger, parent_task->map);
        else
-               new_task->map = vm_map_create(pmap_create(0, is_64bit),
-                                       (vm_map_offset_t)(VM_MIN_ADDRESS),
-                                       (vm_map_offset_t)(VM_MAX_ADDRESS), TRUE);
+               new_task->map = vm_map_create(pmap_create(ledger, 0, is_64bit),
+                               (vm_map_offset_t)(VM_MIN_ADDRESS),
+                               (vm_map_offset_t)(VM_MAX_ADDRESS), TRUE);
 
        /* Inherit memlock limit from parent */
        if (parent_task)
-               vm_map_set_user_wire_limit(new_task->map, parent_task->map->user_wire_limit);
+               vm_map_set_user_wire_limit(new_task->map, (vm_size_t)parent_task->map->user_wire_limit);
 
-       mutex_init(&new_task->lock, 0);
+       lck_mtx_init(&new_task->lock, &task_lck_grp, &task_lck_attr);
        queue_init(&new_task->threads);
        new_task->suspend_count = 0;
        new_task->thread_count = 0;
@@ -357,6 +429,7 @@ task_create_internal(
        new_task->user_stop_count = 0;
        new_task->role = TASK_UNSPECIFIED;
        new_task->active = TRUE;
+       new_task->halting = FALSE;
        new_task->user_data = NULL;
        new_task->faults = 0;
        new_task->cow_faults = 0;
@@ -370,17 +443,17 @@ task_create_internal(
        new_task->taskFeatures[0] = 0;                          /* Init task features */
        new_task->taskFeatures[1] = 0;                          /* Init task features */
 
+       zinfo_task_init(new_task);
+
 #ifdef MACH_BSD
        new_task->bsd_info = NULL;
 #endif /* MACH_BSD */
 
-#ifdef __i386__
+#if defined(__i386__) || defined(__x86_64__)
        new_task->i386_ldt = 0;
+       new_task->task_debug = NULL;
 #endif
 
-#ifdef __ppc__
-       if(BootProcInfo.pf.Available & pf64Bit) new_task->taskFeatures[0] |= tf64BitData;       /* If 64-bit machine, show we have 64-bit registers at least */
-#endif
 
        queue_init(&new_task->semaphore_list);
        queue_init(&new_task->lock_set_list);
@@ -388,7 +461,6 @@ task_create_internal(
        new_task->lock_sets_owned = 0;
 
 #if CONFIG_MACF_MACH
-       /*mutex_init(&new_task->labellock, ETAP_NO_TRACE);*/
        new_task->label = labelh_new(1);
        mac_task_label_init (&new_task->maclabel);
 #endif
@@ -404,6 +476,27 @@ task_create_internal(
 
        new_task->affinity_space = NULL;
 
+#if CONFIG_COUNTERS
+       new_task->t_chud = 0U;
+#endif
+
+       new_task->pidsuspended = FALSE;
+       new_task->frozen = FALSE;
+       new_task->rusage_cpu_flags = 0;
+       new_task->rusage_cpu_percentage = 0;
+       new_task->rusage_cpu_interval = 0;
+       new_task->rusage_cpu_deadline = 0;
+       new_task->rusage_cpu_callt = NULL;
+       new_task->proc_terminate = 0;
+#if CONFIG_EMBEDDED
+       queue_init(&new_task->task_watchers);
+       new_task->appstate = TASK_APPSTATE_ACTIVE;
+       new_task->num_taskwatchers  = 0;
+       new_task->watchapplying  = 0;
+#endif /* CONFIG_EMBEDDED */
+
+       new_task->uexc_range_start = new_task->uexc_range_size = new_task->uexc_handler = 0;
+
        if (parent_task != TASK_NULL) {
                new_task->sec_token = parent_task->sec_token;
                new_task->audit_token = parent_task->audit_token;
@@ -412,25 +505,45 @@ task_create_internal(
                shared_region = vm_shared_region_get(parent_task);
                vm_shared_region_set(new_task, shared_region);
 
-               new_task->wired_ledger_port = ledger_copy(
-                       convert_port_to_ledger(parent_task->wired_ledger_port));
-               new_task->paged_ledger_port = ledger_copy(
-                       convert_port_to_ledger(parent_task->paged_ledger_port));
                if(task_has_64BitAddr(parent_task))
                        task_set_64BitAddr(new_task);
+               new_task->all_image_info_addr = parent_task->all_image_info_addr;
+               new_task->all_image_info_size = parent_task->all_image_info_size;
 
-#ifdef __i386__
+#if defined(__i386__) || defined(__x86_64__)
                if (inherit_memory && parent_task->i386_ldt)
                        new_task->i386_ldt = user_ldt_copy(parent_task->i386_ldt);
 #endif
                if (inherit_memory && parent_task->affinity_space)
                        task_affinity_create(parent_task, new_task);
+
+               new_task->pset_hint = parent_task->pset_hint = task_choose_pset(parent_task);
+               new_task->policystate = parent_task->policystate;
+               /* inherit the self action state */
+               new_task->appliedstate = parent_task->appliedstate;
+               new_task->ext_policystate = parent_task->ext_policystate;
+#if NOTYET
+               /* till the child lifecycle is cleared do not inherit external action */
+               new_task->ext_appliedstate = parent_task->ext_appliedstate;
+#else
+               new_task->ext_appliedstate = default_task_null_policy;
+#endif
        }
        else {
                new_task->sec_token = KERNEL_SECURITY_TOKEN;
                new_task->audit_token = KERNEL_AUDIT_TOKEN;
-               new_task->wired_ledger_port = ledger_copy(root_wired_ledger);
-               new_task->paged_ledger_port = ledger_copy(root_paged_ledger);
+#ifdef __LP64__
+               if(is_64bit)
+                       task_set_64BitAddr(new_task);
+#endif
+               new_task->all_image_info_addr = (mach_vm_address_t)0;
+               new_task->all_image_info_size = (mach_vm_size_t)0;
+
+               new_task->pset_hint = PROCESSOR_SET_NULL;
+               new_task->policystate = default_task_proc_policy;
+               new_task->ext_policystate = default_task_proc_policy;
+               new_task->appliedstate = default_task_null_policy;
+               new_task->ext_appliedstate = default_task_null_policy;
        }
 
        if (kernel_task == TASK_NULL) {
@@ -441,11 +554,13 @@ task_create_internal(
                new_task->priority = BASEPRI_DEFAULT;
                new_task->max_priority = MAXPRI_USER;
        }
+
+       bzero(&new_task->extmod_statistics, sizeof(new_task->extmod_statistics));
        
-       mutex_lock(&tasks_threads_lock);
+       lck_mtx_lock(&tasks_threads_lock);
        queue_enter(&tasks, new_task, task_t, tasks);
        tasks_count++;
-       mutex_unlock(&tasks_threads_lock);
+       lck_mtx_unlock(&tasks_threads_lock);
 
        if (vm_backing_store_low && parent_task != NULL)
                new_task->priv_flags |= (parent_task->priv_flags&VM_BACKING_STORE_PRIV);
@@ -465,12 +580,25 @@ void
 task_deallocate(
        task_t          task)
 {
+       ledger_amount_t credit, debit;
+
        if (task == TASK_NULL)
            return;
 
        if (task_deallocate_internal(task) > 0)
                return;
 
+       lck_mtx_lock(&tasks_threads_lock);
+       queue_remove(&terminated_tasks, task, task_t, tasks);
+       lck_mtx_unlock(&tasks_threads_lock);
+
+       /*
+        *      Give the machine dependent code a chance
+        *      to perform cleanup before ripping apart
+        *      the task.
+        */
+       machine_task_terminate(task);
+
        ipc_task_terminate(task);
 
        if (task->affinity_space)
@@ -479,9 +607,24 @@ task_deallocate(
        vm_map_deallocate(task->map);
        is_release(task->itk_space);
 
+       lck_mtx_destroy(&task->lock, &task_lck_grp);
+
 #if CONFIG_MACF_MACH
        labelh_release(task->label);
 #endif
+
+       if (!ledger_get_entries(task->ledger, task_ledgers.tkm_private, &credit,
+           &debit)) {
+               OSAddAtomic64(credit, (int64_t *)&tasks_tkm_private.alloc);
+               OSAddAtomic64(debit, (int64_t *)&tasks_tkm_private.free);
+       }
+       if (!ledger_get_entries(task->ledger, task_ledgers.tkm_shared, &credit,
+           &debit)) {
+               OSAddAtomic64(credit, (int64_t *)&tasks_tkm_shared.alloc);
+               OSAddAtomic64(debit, (int64_t *)&tasks_tkm_shared.free);
+       }
+       ledger_dereference(task->ledger);
+       zinfo_task_free(task);
        zfree(task_zone, task);
 }
 
@@ -547,9 +690,9 @@ task_terminate_internal(
                task_lock(task);
        }
 
-       if (!task->active || !self->active) {
+       if (!task->active) {
                /*
-                *      Task or current act is already being terminated.
+                *      Task is already being terminated.
                 *      Just return an error. If we are dying, this will
                 *      just get us to our AST special handler and that
                 *      will get us to finalize the termination of ourselves.
@@ -589,15 +732,14 @@ task_terminate_internal(
                        thread_terminate_internal(thread);
        }
 
+       task_unlock(task);
+
+#if CONFIG_EMBEDDED
        /*
-        *      Give the machine dependent code a chance
-        *      to perform cleanup before ripping apart
-        *      the task.
+        * remove all task watchers 
         */
-       if (self_task == task)
-               machine_thread_terminate_self();
-
-       task_unlock(task);
+       task_removewatchers(task);
+#endif /* CONFIG_EMBEDDED */
 
        /*
         *      Destroy all synchronizers owned by the task.
@@ -607,15 +749,7 @@ task_terminate_internal(
        /*
         *      Destroy the IPC space, leaving just a reference for it.
         */
-       ipc_space_destroy(task->itk_space);
-
-#ifdef __ppc__
-       /* LP64todo - make this clean */
-       /*
-        * PPC51: ppc64 is limited to 51-bit addresses.
-        */
-       pmap_unmap_sharedpage(task->map->pmap);         /* Unmap commpage */
-#endif /* __ppc__ */
+       ipc_space_terminate(task->itk_space);
 
        if (vm_map_has_4GB_pagezero(task->map))
                vm_map_clear_4GB_pagezero(task->map);
@@ -636,10 +770,11 @@ task_terminate_internal(
        /* release our shared region */
        vm_shared_region_set(task, NULL);
 
-       mutex_lock(&tasks_threads_lock);
+       lck_mtx_lock(&tasks_threads_lock);
        queue_remove(&tasks, task, task_t, tasks);
+       queue_enter(&terminated_tasks, task, task_t, tasks);
        tasks_count--;
-       mutex_unlock(&tasks_threads_lock);
+       lck_mtx_unlock(&tasks_threads_lock);
 
        /*
         * We no longer need to guard against being aborted, so restore
@@ -647,10 +782,6 @@ task_terminate_internal(
         */
        thread_interrupt_level(interrupt_save);
 
-#if __ppc__
-    perfmon_release_facility(task); // notify the perfmon facility
-#endif
-
        /*
         * Get rid of the task active reference on itself.
         */
@@ -660,16 +791,15 @@ task_terminate_internal(
 }
 
 /*
- * task_halt:
+ * task_start_halt:
  *
  *     Shut the current task down (except for the current thread) in
  *     preparation for dramatic changes to the task (probably exec).
- *     We hold the task, terminate all other threads in the task and
- *     wait for them to terminate, clean up the portspace, and when
- *     all done, let the current thread go.
+ *     We hold the task and mark all other threads in the task for
+ *     termination.
  */
 kern_return_t
-task_halt(
+task_start_halt(
        task_t          task)
 {
        thread_t        thread, self;
@@ -683,7 +813,7 @@ task_halt(
 
        task_lock(task);
 
-       if (!task->active || !self->active) {
+       if (task->halting || !task->active || !self->active) {
                /*
                 *      Task or current thread is already being terminated.
                 *      Hurry up and return out of the current kernel context
@@ -695,7 +825,10 @@ task_halt(
                return (KERN_FAILURE);
        }
 
+       task->halting = TRUE;
+
        if (task->thread_count > 1) {
+
                /*
                 * Mark all the threads to keep them from starting any more
                 * user-level execution.  The thread_terminate_internal code
@@ -714,15 +847,45 @@ task_halt(
 
                task_release_locked(task);
        }
+       task_unlock(task);
+       return KERN_SUCCESS;
+}
+
+
+/*
+ * task_complete_halt:
+ *
+ *     Complete task halt by waiting for threads to terminate, then clean
+ *     up task resources (VM, port namespace, etc...) and then let the
+ *     current thread go in the (practically empty) task context.
+ */
+void
+task_complete_halt(task_t task)
+{
+       task_lock(task);
+       assert(task->halting);
+       assert(task == current_task());
 
        /*
-        *      Give the machine dependent code a chance
-        *      to perform cleanup before ripping apart
-        *      the task.
+        *      Wait for the other threads to get shut down.
+        *      When the last other thread is reaped, we'll be
+        *      woken up.
         */
-       machine_thread_terminate_self();
+       if (task->thread_count > 1) {
+               assert_wait((event_t)&task->halting, THREAD_UNINT);
+               task_unlock(task);
+               thread_block(THREAD_CONTINUE_NULL);
+       } else {
+               task_unlock(task);
+       }
 
-       task_unlock(task);
+       /*
+        *      Give the machine dependent code a chance
+        *      to perform cleanup of task-level resources
+        *      associated with the current thread before
+        *      ripping apart the task.
+        */
+       machine_task_terminate(task);
 
        /*
         *      Destroy all synchronizers owned by the task.
@@ -742,7 +905,7 @@ task_halt(
        vm_map_remove(task->map, task->map->min_offset,
                      task->map->max_offset, VM_MAP_NO_FLAGS);
 
-       return (KERN_SUCCESS);
+       task->halting = FALSE;
 }
 
 /*
@@ -807,6 +970,28 @@ task_hold(
        return (KERN_SUCCESS);
 }
 
+kern_return_t
+task_wait(
+               task_t          task,
+               boolean_t       until_not_runnable)
+{
+       if (task == TASK_NULL)
+               return (KERN_INVALID_ARGUMENT);
+
+       task_lock(task);
+
+       if (!task->active) {
+               task_unlock(task);
+
+               return (KERN_FAILURE);
+       }
+
+       task_wait_locked(task, until_not_runnable);
+       task_unlock(task);
+
+       return (KERN_SUCCESS);
+}
+
 /*
  *     task_wait_locked:
  *
@@ -817,7 +1002,8 @@ task_hold(
  */
 void
 task_wait_locked(
-       register task_t         task)
+       register task_t         task,
+       boolean_t               until_not_runnable)
 {
        register thread_t       thread, self;
 
@@ -833,7 +1019,7 @@ task_wait_locked(
         */
        queue_iterate(&task->threads, thread, thread_t, task_threads) {
                if (thread != self)
-                       thread_wait(thread);
+                       thread_wait(thread, until_not_runnable);
        }
 }
 
@@ -1001,26 +1187,11 @@ task_threads(
        return (KERN_SUCCESS);
 }
 
-/*
- *     task_suspend:
- *
- *     Implement a user-level suspension on a task.
- *
- * Conditions:
- *     The caller holds a reference to the task
- */
-kern_return_t
-task_suspend(
+static kern_return_t
+place_task_hold    (
        register task_t         task)
-{
-       if (task == TASK_NULL || task == kernel_task)
-               return (KERN_INVALID_ARGUMENT);
-
-       task_lock(task);
-
+{    
        if (!task->active) {
-               task_unlock(task);
-
                return (KERN_FAILURE);
        }
 
@@ -1029,8 +1200,6 @@ task_suspend(
                 *      If the stop count was positive, the task is
                 *      already stopped and we can exit.
                 */
-               task_unlock(task);
-
                return (KERN_SUCCESS);
        }
 
@@ -1041,11 +1210,71 @@ task_suspend(
         * to stop executing user code.
         */
        task_hold_locked(task);
-       task_wait_locked(task);
+       task_wait_locked(task, TRUE);
+       
+       return (KERN_SUCCESS);
+}
+
+static kern_return_t
+release_task_hold    (
+       register task_t         task,
+       boolean_t           pidresume)
+{
+       register boolean_t release = FALSE;
+    
+       if (!task->active) {
+               return (KERN_FAILURE);
+       }
+       
+       if (pidresume) {
+           if (task->pidsuspended == FALSE) {
+            return (KERN_FAILURE);
+           }
+           task->pidsuspended = FALSE;
+       }
+
+    if (task->user_stop_count > (task->pidsuspended ? 1 : 0)) {
+               if (--task->user_stop_count == 0) {
+                       release = TRUE;
+               }
+       }
+       else {
+               return (KERN_FAILURE);
+       }
+
+       /*
+        *      Release the task if necessary.
+        */
+       if (release)
+               task_release_locked(task);
+               
+    return (KERN_SUCCESS);
+}
+
+/*
+ *     task_suspend:
+ *
+ *     Implement a user-level suspension on a task.
+ *
+ * Conditions:
+ *     The caller holds a reference to the task
+ */
+kern_return_t
+task_suspend(
+       register task_t         task)
+{
+       kern_return_t    kr;
+       
+       if (task == TASK_NULL || task == kernel_task)
+               return (KERN_INVALID_ARGUMENT);
+
+       task_lock(task);
+
+       kr = place_task_hold(task);
 
        task_unlock(task);
 
-       return (KERN_SUCCESS);
+       return (kr);
 }
 
 /*
@@ -1059,40 +1288,192 @@ kern_return_t
 task_resume(
        register task_t task)
 {
-       register boolean_t      release = FALSE;
+       kern_return_t    kr;
 
        if (task == TASK_NULL || task == kernel_task)
                return (KERN_INVALID_ARGUMENT);
 
        task_lock(task);
 
-       if (!task->active) {
-               task_unlock(task);
+       kr = release_task_hold(task, FALSE);
 
-               return (KERN_FAILURE);
+       task_unlock(task);
+
+       return (kr);
+}
+
+kern_return_t
+task_pidsuspend_locked(task_t task)
+{
+       kern_return_t kr;
+
+       if (task->pidsuspended) {
+               kr = KERN_FAILURE;
+               goto out;
        }
 
-       if (task->user_stop_count > 0) {
-               if (--task->user_stop_count == 0)
-                       release = TRUE;
+       task->pidsuspended = TRUE;
+
+       kr = place_task_hold(task);
+       if (kr != KERN_SUCCESS) {
+               task->pidsuspended = FALSE;
        }
-       else {
-               task_unlock(task);
+out:
+       return(kr);
+}
 
-               return (KERN_FAILURE);
+
+/*
+ *     task_pidsuspend:
+ *
+ *     Suspends a task by placing a hold on its threads.
+ *
+ * Conditions:
+ *     The caller holds a reference to the task
+ */
+kern_return_t
+task_pidsuspend(
+       register task_t         task)
+{
+       kern_return_t    kr;
+    
+       if (task == TASK_NULL || task == kernel_task)
+               return (KERN_INVALID_ARGUMENT);
+
+       task_lock(task);
+
+       kr = task_pidsuspend_locked(task);
+
+       task_unlock(task);
+
+       return (kr);
+}
+
+/* If enabled, we bring all the frozen pages back in prior to resumption; otherwise, they're faulted back in on demand */
+#define THAW_ON_RESUME 1
+
+/*
+ *     task_pidresume:
+ *             Resumes a previously suspended task.
+ *             
+ * Conditions:
+ *             The caller holds a reference to the task
+ */
+kern_return_t 
+task_pidresume(
+       register task_t task)
+{
+       kern_return_t    kr;
+#if (CONFIG_FREEZE && THAW_ON_RESUME)
+    boolean_t frozen;
+#endif
+
+       if (task == TASK_NULL || task == kernel_task)
+               return (KERN_INVALID_ARGUMENT);
+
+       task_lock(task);
+       
+#if (CONFIG_FREEZE && THAW_ON_RESUME)
+    frozen = task->frozen;
+    task->frozen = FALSE;
+#endif
+
+       kr = release_task_hold(task, TRUE);
+
+       task_unlock(task);
+
+#if (CONFIG_FREEZE && THAW_ON_RESUME)
+       if ((kr == KERN_SUCCESS) && (frozen == TRUE)) {
+           kr = vm_map_thaw(task->map);
        }
+#endif
 
-       /*
-        *      Release the task if necessary.
-        */
-       if (release)
-               task_release_locked(task);
+       return (kr);
+}
+
+#if CONFIG_FREEZE
+
+/*
+ *     task_freeze:
+ *
+ *     Freeze a task.
+ *
+ * Conditions:
+ *     The caller holds a reference to the task
+ */
+kern_return_t
+task_freeze(
+       register task_t    task,
+       uint32_t           *purgeable_count,
+       uint32_t           *wired_count,
+       uint32_t           *clean_count,
+       uint32_t           *dirty_count,
+       uint32_t           dirty_budget,
+       boolean_t          *shared,
+       boolean_t          walk_only)
+{
+       kern_return_t kr;
+    
+       if (task == TASK_NULL || task == kernel_task)
+               return (KERN_INVALID_ARGUMENT);
+
+       task_lock(task);
+
+       if (task->frozen) {
+           task_unlock(task);
+           return (KERN_FAILURE);
+       }
+
+    if (walk_only == FALSE) {
+           task->frozen = TRUE;
+    }
 
        task_unlock(task);
 
-       return (KERN_SUCCESS);
+       if (walk_only) {
+               kr = vm_map_freeze_walk(task->map, purgeable_count, wired_count, clean_count, dirty_count, dirty_budget, shared);               
+       } else {
+               kr = vm_map_freeze(task->map, purgeable_count, wired_count, clean_count, dirty_count, dirty_budget, shared);
+       }
+
+       return (kr);
 }
 
+/*
+ *     task_thaw:
+ *
+ *     Thaw a currently frozen task.
+ *
+ * Conditions:
+ *     The caller holds a reference to the task
+ */
+kern_return_t
+task_thaw(
+       register task_t         task)
+{
+       kern_return_t kr;
+    
+       if (task == TASK_NULL || task == kernel_task)
+               return (KERN_INVALID_ARGUMENT);
+
+       task_lock(task);
+       
+       if (!task->frozen) {
+           task_unlock(task);
+           return (KERN_FAILURE);
+       }
+
+       task->frozen = FALSE;
+
+       task_unlock(task);
+
+       kr = vm_map_thaw(task->map);
+
+       return (kr);
+}
+
+#endif /* CONFIG_FREEZE */
+
 kern_return_t
 host_security_set_task_token(
         host_security_t  host_security,
@@ -1125,32 +1506,6 @@ host_security_set_task_token(
         return(kr);
 }
 
-/*
- * Utility routine to set a ledger
- */
-kern_return_t
-task_set_ledger(
-        task_t         task,
-        ledger_t       wired,
-        ledger_t       paged)
-{
-       if (task == TASK_NULL)
-               return(KERN_INVALID_ARGUMENT);
-
-        task_lock(task);
-        if (wired) {
-                ipc_port_release_send(task->wired_ledger_port);
-                task->wired_ledger_port = ledger_copy(wired);
-        }                
-        if (paged) {
-                ipc_port_release_send(task->paged_ledger_port);
-                task->paged_ledger_port = ledger_copy(paged);
-        }                
-        task_unlock(task);
-
-        return(KERN_SUCCESS);
-}
-
 /*
  * This routine was added, pretty much exclusively, for registering the
  * RPC glue vector for in-kernel short circuited tasks.  Rather than
@@ -1183,24 +1538,37 @@ task_info(
        task_info_t                             task_info_out,
        mach_msg_type_number_t  *task_info_count)
 {
+       kern_return_t error = KERN_SUCCESS;
+
        if (task == TASK_NULL)
                return (KERN_INVALID_ARGUMENT);
 
+       task_lock(task);
+
+       if ((task != current_task()) && (!task->active)) {
+               task_unlock(task);
+               return (KERN_INVALID_ARGUMENT);
+       }
+
        switch (flavor) {
 
        case TASK_BASIC_INFO_32:
        case TASK_BASIC2_INFO_32:
        {
                task_basic_info_32_t    basic_info;
-               vm_map_t                        map;
+               vm_map_t                                map;
+               clock_sec_t                             secs;
+               clock_usec_t                    usecs;
 
-               if (*task_info_count < TASK_BASIC_INFO_32_COUNT)
-                   return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < TASK_BASIC_INFO_32_COUNT) {
+                   error = KERN_INVALID_ARGUMENT;
+                   break;
+               }
 
                basic_info = (task_basic_info_32_t)task_info_out;
 
                map = (task == kernel_task)? kernel_map: task->map;
-               basic_info->virtual_size  = CAST_DOWN(vm_offset_t,map->size);
+               basic_info->virtual_size = (typeof(basic_info->virtual_size))map->size;
                if (flavor == TASK_BASIC2_INFO_32) {
                        /*
                         * The "BASIC2" flavor gets the maximum resident
@@ -1212,18 +1580,19 @@ task_info(
                }
                basic_info->resident_size *= PAGE_SIZE;
 
-               task_lock(task);
                basic_info->policy = ((task != kernel_task)?
                                                                                  POLICY_TIMESHARE: POLICY_RR);
                basic_info->suspend_count = task->user_stop_count;
 
-               absolutetime_to_microtime(task->total_user_time,
-                                         (unsigned *)&basic_info->user_time.seconds,
-                                         (unsigned *)&basic_info->user_time.microseconds);
-               absolutetime_to_microtime(task->total_system_time,
-                                         (unsigned *)&basic_info->system_time.seconds,
-                                         (unsigned *)&basic_info->system_time.microseconds);
-               task_unlock(task);
+               absolutetime_to_microtime(task->total_user_time, &secs, &usecs);
+               basic_info->user_time.seconds = 
+                       (typeof(basic_info->user_time.seconds))secs;
+               basic_info->user_time.microseconds = usecs;
+
+               absolutetime_to_microtime(task->total_system_time, &secs, &usecs);
+               basic_info->system_time.seconds = 
+                       (typeof(basic_info->system_time.seconds))secs;
+               basic_info->system_time.microseconds = usecs;
 
                *task_info_count = TASK_BASIC_INFO_32_COUNT;
                break;
@@ -1232,10 +1601,14 @@ task_info(
        case TASK_BASIC_INFO_64:
        {
                task_basic_info_64_t    basic_info;
-               vm_map_t                        map;
+               vm_map_t                                map;
+               clock_sec_t                             secs;
+               clock_usec_t                    usecs;
 
-               if (*task_info_count < TASK_BASIC_INFO_64_COUNT)
-                   return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < TASK_BASIC_INFO_64_COUNT) {
+                   error = KERN_INVALID_ARGUMENT;
+                   break;
+               }
 
                basic_info = (task_basic_info_64_t)task_info_out;
 
@@ -1245,30 +1618,78 @@ task_info(
                        (mach_vm_size_t)(pmap_resident_count(map->pmap))
                        * PAGE_SIZE_64;
 
-               task_lock(task);
                basic_info->policy = ((task != kernel_task)?
                                                                                  POLICY_TIMESHARE: POLICY_RR);
                basic_info->suspend_count = task->user_stop_count;
 
-               absolutetime_to_microtime(task->total_user_time,
-                                         (unsigned *)&basic_info->user_time.seconds,
-                                         (unsigned *)&basic_info->user_time.microseconds);
-               absolutetime_to_microtime(task->total_system_time,
-                                         (unsigned *)&basic_info->system_time.seconds,
-                                         (unsigned *)&basic_info->system_time.microseconds);
-               task_unlock(task);
+               absolutetime_to_microtime(task->total_user_time, &secs, &usecs);
+               basic_info->user_time.seconds = 
+                       (typeof(basic_info->user_time.seconds))secs;
+               basic_info->user_time.microseconds = usecs;
+
+               absolutetime_to_microtime(task->total_system_time, &secs, &usecs);
+               basic_info->system_time.seconds =
+                       (typeof(basic_info->system_time.seconds))secs;
+               basic_info->system_time.microseconds = usecs;
 
                *task_info_count = TASK_BASIC_INFO_64_COUNT;
                break;
        }
 
+       case MACH_TASK_BASIC_INFO:
+       {
+               mach_task_basic_info_t  basic_info;
+               vm_map_t                map;
+               clock_sec_t             secs;
+               clock_usec_t            usecs;
+
+               if (*task_info_count < MACH_TASK_BASIC_INFO_COUNT) {
+                   error = KERN_INVALID_ARGUMENT;
+                   break;
+               }
+
+               basic_info = (mach_task_basic_info_t)task_info_out;
+
+               map = (task == kernel_task) ? kernel_map : task->map;
+
+               basic_info->virtual_size  = map->size;
+
+               basic_info->resident_size =
+                   (mach_vm_size_t)(pmap_resident_count(map->pmap));
+               basic_info->resident_size *= PAGE_SIZE_64;
+
+               basic_info->resident_size_max =
+                   (mach_vm_size_t)(pmap_resident_max(map->pmap));
+               basic_info->resident_size_max *= PAGE_SIZE_64;
+
+               basic_info->policy = ((task != kernel_task) ? 
+                                     POLICY_TIMESHARE : POLICY_RR);
+
+               basic_info->suspend_count = task->user_stop_count;
+
+               absolutetime_to_microtime(task->total_user_time, &secs, &usecs);
+               basic_info->user_time.seconds = 
+                   (typeof(basic_info->user_time.seconds))secs;
+               basic_info->user_time.microseconds = usecs;
+
+               absolutetime_to_microtime(task->total_system_time, &secs, &usecs);
+               basic_info->system_time.seconds =
+                   (typeof(basic_info->system_time.seconds))secs;
+               basic_info->system_time.microseconds = usecs;
+
+               *task_info_count = MACH_TASK_BASIC_INFO_COUNT;
+               break;
+       }
+
        case TASK_THREAD_TIMES_INFO:
        {
                register task_thread_times_info_t       times_info;
                register thread_t                                       thread;
 
-               if (*task_info_count < TASK_THREAD_TIMES_INFO_COUNT)
-                   return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < TASK_THREAD_TIMES_INFO_COUNT) {
+                   error = KERN_INVALID_ARGUMENT;
+                   break;
+               }
 
                times_info = (task_thread_times_info_t) task_info_out;
                times_info->user_time.seconds = 0;
@@ -1276,7 +1697,6 @@ task_info(
                times_info->system_time.seconds = 0;
                times_info->system_time.microseconds = 0;
 
-               task_lock(task);
 
                queue_iterate(&task->threads, thread, thread_t, task_threads) {
                    time_value_t        user_time, system_time;
@@ -1287,7 +1707,6 @@ task_info(
                    time_value_add(&times_info->system_time, &system_time);
                }
 
-               task_unlock(task);
 
                *task_info_count = TASK_THREAD_TIMES_INFO_COUNT;
                break;
@@ -1298,65 +1717,210 @@ task_info(
                task_absolutetime_info_t        info;
                register thread_t                       thread;
 
-               if (*task_info_count < TASK_ABSOLUTETIME_INFO_COUNT)
-                       return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < TASK_ABSOLUTETIME_INFO_COUNT) {
+                       error = KERN_INVALID_ARGUMENT;
+                       break;
+               }
 
                info = (task_absolutetime_info_t)task_info_out;
                info->threads_user = info->threads_system = 0;
 
-               task_lock(task);
 
                info->total_user = task->total_user_time;
                info->total_system = task->total_system_time;
 
                queue_iterate(&task->threads, thread, thread_t, task_threads) {
                        uint64_t        tval;
+                       spl_t           x;
+
+                       x = splsched();
+                       thread_lock(thread);
 
                        tval = timer_grab(&thread->user_timer);
                        info->threads_user += tval;
                        info->total_user += tval;
 
                        tval = timer_grab(&thread->system_timer);
-                       info->threads_system += tval;
-                       info->total_system += tval;
+                       if (thread->precise_user_kernel_time) {
+                               info->threads_system += tval;
+                               info->total_system += tval;
+                       } else {
+                               /* system_timer may represent either sys or user */
+                               info->threads_user += tval;
+                               info->total_user += tval;
+                       }
+
+                       thread_unlock(thread);
+                       splx(x);
                }
 
-               task_unlock(task);
 
                *task_info_count = TASK_ABSOLUTETIME_INFO_COUNT;
                break;
        }
 
+       case TASK_DYLD_INFO:
+       {
+               task_dyld_info_t info;
+
+               /*
+                * We added the format field to TASK_DYLD_INFO output.  For
+                * temporary backward compatibility, accept the fact that
+                * clients may ask for the old version - distinquished by the
+                * size of the expected result structure.
+                */
+#define TASK_LEGACY_DYLD_INFO_COUNT \
+               offsetof(struct task_dyld_info, all_image_info_format)/sizeof(natural_t)
+
+               if (*task_info_count < TASK_LEGACY_DYLD_INFO_COUNT) {
+                       error = KERN_INVALID_ARGUMENT;
+                       break;
+               }
+
+               info = (task_dyld_info_t)task_info_out;
+               info->all_image_info_addr = task->all_image_info_addr;
+               info->all_image_info_size = task->all_image_info_size;
+
+               /* only set format on output for those expecting it */
+               if (*task_info_count >= TASK_DYLD_INFO_COUNT) {
+                       info->all_image_info_format = task_has_64BitAddr(task) ?
+                                                TASK_DYLD_ALL_IMAGE_INFO_64 : 
+                                                TASK_DYLD_ALL_IMAGE_INFO_32 ;
+                       *task_info_count = TASK_DYLD_INFO_COUNT;
+               } else {
+                       *task_info_count = TASK_LEGACY_DYLD_INFO_COUNT;
+               }
+               break;
+       }
+
+       case TASK_EXTMOD_INFO:
+       {
+               task_extmod_info_t info;
+               void *p;
+
+               if (*task_info_count < TASK_EXTMOD_INFO_COUNT) {
+                       error = KERN_INVALID_ARGUMENT;
+                       break;
+               }
+
+               info = (task_extmod_info_t)task_info_out;
+
+               p = get_bsdtask_info(task);
+               if (p) {
+                       proc_getexecutableuuid(p, info->task_uuid, sizeof(info->task_uuid));
+               } else {
+                       bzero(info->task_uuid, sizeof(info->task_uuid));
+               }
+               info->extmod_statistics = task->extmod_statistics;
+               *task_info_count = TASK_EXTMOD_INFO_COUNT;
+
+               break;
+       }
+
+       case TASK_KERNELMEMORY_INFO:
+       {
+               task_kernelmemory_info_t        tkm_info;
+               ledger_amount_t                 credit, debit;
+
+               if (*task_info_count < TASK_KERNELMEMORY_INFO_COUNT) {
+                  error = KERN_INVALID_ARGUMENT;
+                  break;
+               }
+
+               tkm_info = (task_kernelmemory_info_t) task_info_out;
+               tkm_info->total_palloc = 0;
+               tkm_info->total_pfree = 0;
+               tkm_info->total_salloc = 0;
+               tkm_info->total_sfree = 0;
+
+               if (task == kernel_task) {
+                       /*
+                        * All shared allocs/frees from other tasks count against
+                        * the kernel private memory usage.  If we are looking up
+                        * info for the kernel task, gather from everywhere.
+                        */
+                       task_unlock(task);
+
+                       /* start by accounting for all the terminated tasks against the kernel */
+                       tkm_info->total_palloc = tasks_tkm_private.alloc + tasks_tkm_shared.alloc;
+                       tkm_info->total_pfree = tasks_tkm_private.free + tasks_tkm_shared.free;
+
+                       /* count all other task/thread shared alloc/free against the kernel */
+                       lck_mtx_lock(&tasks_threads_lock);
+
+                       /* XXX this really shouldn't be using the function parameter 'task' as a local var! */
+                       queue_iterate(&tasks, task, task_t, tasks) {
+                               if (task == kernel_task) {
+                                       if (ledger_get_entries(task->ledger,
+                                           task_ledgers.tkm_private, &credit,
+                                           &debit) == KERN_SUCCESS) {
+                                               tkm_info->total_palloc += credit;
+                                               tkm_info->total_pfree += debit;
+                                       }
+                               }
+                               if (!ledger_get_entries(task->ledger,
+                                   task_ledgers.tkm_shared, &credit, &debit)) {
+                                       tkm_info->total_palloc += credit;
+                                       tkm_info->total_pfree += debit;
+                               }
+                       }
+                       lck_mtx_unlock(&tasks_threads_lock);
+               } else {
+                       if (!ledger_get_entries(task->ledger,
+                           task_ledgers.tkm_private, &credit, &debit)) {
+                               tkm_info->total_palloc = credit;
+                               tkm_info->total_pfree = debit;
+                       }
+                       if (!ledger_get_entries(task->ledger,
+                           task_ledgers.tkm_shared, &credit, &debit)) {
+                               tkm_info->total_salloc = credit;
+                               tkm_info->total_sfree = debit;
+                       }
+                       task_unlock(task);
+               }
+
+               *task_info_count = TASK_KERNELMEMORY_INFO_COUNT;
+               return KERN_SUCCESS;
+       }
+
        /* OBSOLETE */
        case TASK_SCHED_FIFO_INFO:
        {
 
-               if (*task_info_count < POLICY_FIFO_BASE_COUNT)
-                       return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < POLICY_FIFO_BASE_COUNT) {
+                       error = KERN_INVALID_ARGUMENT;
+                       break;
+               }
 
-               return (KERN_INVALID_POLICY);
+               error = KERN_INVALID_POLICY;
+               break;
        }
 
        /* OBSOLETE */
        case TASK_SCHED_RR_INFO:
        {
                register policy_rr_base_t       rr_base;
+               uint32_t quantum_time;
+               uint64_t quantum_ns;
 
-               if (*task_info_count < POLICY_RR_BASE_COUNT)
-                       return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < POLICY_RR_BASE_COUNT) {
+                       error = KERN_INVALID_ARGUMENT;
+                       break;
+               }
 
                rr_base = (policy_rr_base_t) task_info_out;
 
-               task_lock(task);
                if (task != kernel_task) {
-                       task_unlock(task);
-                       return (KERN_INVALID_POLICY);
+                       error = KERN_INVALID_POLICY;
+                       break;
                }
 
                rr_base->base_priority = task->priority;
-               task_unlock(task);
 
-               rr_base->quantum = std_quantum_us / 1000;
+               quantum_time = SCHED(initial_quantum_size)(THREAD_NULL);
+               absolutetime_to_nanoseconds(quantum_time, &quantum_ns);
+               
+               rr_base->quantum = (uint32_t)(quantum_ns / 1000 / 1000);
 
                *task_info_count = POLICY_RR_BASE_COUNT;
                break;
@@ -1367,19 +1931,19 @@ task_info(
        {
                register policy_timeshare_base_t        ts_base;
 
-               if (*task_info_count < POLICY_TIMESHARE_BASE_COUNT)
-                       return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < POLICY_TIMESHARE_BASE_COUNT) {
+                       error = KERN_INVALID_ARGUMENT;
+                       break;
+               }
 
                ts_base = (policy_timeshare_base_t) task_info_out;
 
-               task_lock(task);
                if (task == kernel_task) {
-                       task_unlock(task);
-                       return (KERN_INVALID_POLICY);
+                       error = KERN_INVALID_POLICY;
+                       break;
                }
 
                ts_base->base_priority = task->priority;
-               task_unlock(task);
 
                *task_info_count = POLICY_TIMESHARE_BASE_COUNT;
                break;
@@ -1389,14 +1953,14 @@ task_info(
        {
                register security_token_t       *sec_token_p;
 
-               if (*task_info_count < TASK_SECURITY_TOKEN_COUNT)
-                   return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < TASK_SECURITY_TOKEN_COUNT) {
+                   error = KERN_INVALID_ARGUMENT;
+                   break;
+               }
 
                sec_token_p = (security_token_t *) task_info_out;
 
-               task_lock(task);
                *sec_token_p = task->sec_token;
-               task_unlock(task);
 
                *task_info_count = TASK_SECURITY_TOKEN_COUNT;
                break;
@@ -1406,33 +1970,35 @@ task_info(
        {
                register audit_token_t  *audit_token_p;
 
-               if (*task_info_count < TASK_AUDIT_TOKEN_COUNT)
-                   return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < TASK_AUDIT_TOKEN_COUNT) {
+                   error = KERN_INVALID_ARGUMENT;
+                   break;
+               }
 
                audit_token_p = (audit_token_t *) task_info_out;
 
-               task_lock(task);
                *audit_token_p = task->audit_token;
-               task_unlock(task);
 
                *task_info_count = TASK_AUDIT_TOKEN_COUNT;
                break;
        }
             
        case TASK_SCHED_INFO:
-               return (KERN_INVALID_ARGUMENT);
+               error = KERN_INVALID_ARGUMENT;
+               break;
 
        case TASK_EVENTS_INFO:
        {
                register task_events_info_t     events_info;
                register thread_t                       thread;
 
-               if (*task_info_count < TASK_EVENTS_INFO_COUNT)
-                   return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < TASK_EVENTS_INFO_COUNT) {
+                  error = KERN_INVALID_ARGUMENT;
+                  break;
+               }
 
                events_info = (task_events_info_t) task_info_out;
 
-               task_lock(task);
 
                events_info->faults = task->faults;
                events_info->pageins = task->pageins;
@@ -1445,27 +2011,31 @@ task_info(
                events_info->csw = task->c_switch;
 
                queue_iterate(&task->threads, thread, thread_t, task_threads) {
-                       events_info->csw += thread->c_switch;
+                       events_info->csw           += thread->c_switch;
+                       events_info->syscalls_mach += thread->syscalls_mach;
+                       events_info->syscalls_unix += thread->syscalls_unix;
                }
 
-               task_unlock(task);
 
                *task_info_count = TASK_EVENTS_INFO_COUNT;
                break;
        }
        case TASK_AFFINITY_TAG_INFO:
        {
-               if (*task_info_count < TASK_AFFINITY_TAG_INFO_COUNT)
-                   return (KERN_INVALID_ARGUMENT);
+               if (*task_info_count < TASK_AFFINITY_TAG_INFO_COUNT) {
+                   error = KERN_INVALID_ARGUMENT;
+                   break;
+               }
 
-               return task_affinity_info(task, task_info_out, task_info_count);
+               error = task_affinity_info(task, task_info_out, task_info_count);
+               break;
        }
-
        default:
-               return (KERN_INVALID_ARGUMENT);
+               error = KERN_INVALID_ARGUMENT;
        }
 
-       return (KERN_SUCCESS);
+       task_unlock(task);
+       return (error);
 }
 
 void
@@ -1474,6 +2044,7 @@ task_vtimer_set(
        integer_t       which)
 {
        thread_t        thread;
+       spl_t           x;
 
        /* assert(task == current_task()); */ /* bogus assert 4803227 4807483 */
 
@@ -1485,21 +2056,36 @@ task_vtimer_set(
 
        case TASK_VTIMER_USER:
                queue_iterate(&task->threads, thread, thread_t, task_threads) {
-                       thread->vtimer_user_save = timer_grab(&thread->user_timer);
+                       x = splsched();
+                       thread_lock(thread);
+                       if (thread->precise_user_kernel_time)
+                               thread->vtimer_user_save = timer_grab(&thread->user_timer);
+                       else
+                               thread->vtimer_user_save = timer_grab(&thread->system_timer);
+                       thread_unlock(thread);
+                       splx(x);
                }
                break;
 
        case TASK_VTIMER_PROF:
                queue_iterate(&task->threads, thread, thread_t, task_threads) {
+                       x = splsched();
+                       thread_lock(thread);
                        thread->vtimer_prof_save = timer_grab(&thread->user_timer);
                        thread->vtimer_prof_save += timer_grab(&thread->system_timer);
+                       thread_unlock(thread);
+                       splx(x);
                }
                break;
 
        case TASK_VTIMER_RLIM:
                queue_iterate(&task->threads, thread, thread_t, task_threads) {
+                       x = splsched();
+                       thread_lock(thread);
                        thread->vtimer_rlim_save = timer_grab(&thread->user_timer);
                        thread->vtimer_rlim_save += timer_grab(&thread->system_timer);
+                       thread_unlock(thread);
+                       splx(x);
                }
                break;
        }
@@ -1529,38 +2115,48 @@ __unused
        uint32_t        *microsecs)
 {
        thread_t        thread = current_thread();
-       uint32_t        tdelt, secs;
+       uint32_t        tdelt;
+       clock_sec_t     secs;
        uint64_t        tsum;
 
        assert(task == current_task());
 
        assert(task->vtimers & which);
 
-       tdelt = secs = 0;
+       secs = tdelt = 0;
 
        switch (which) {
 
        case TASK_VTIMER_USER:
-               tdelt = timer_delta(&thread->user_timer,
+               if (thread->precise_user_kernel_time) {
+                       tdelt = (uint32_t)timer_delta(&thread->user_timer,
+                                                               &thread->vtimer_user_save);
+               } else {
+                       tdelt = (uint32_t)timer_delta(&thread->system_timer,
                                                                &thread->vtimer_user_save);
+               }
+               absolutetime_to_microtime(tdelt, &secs, microsecs);
                break;
 
        case TASK_VTIMER_PROF:
                tsum = timer_grab(&thread->user_timer);
                tsum += timer_grab(&thread->system_timer);
-               tdelt = tsum - thread->vtimer_prof_save;
-               thread->vtimer_prof_save = tsum;
+               tdelt = (uint32_t)(tsum - thread->vtimer_prof_save);
+               absolutetime_to_microtime(tdelt, &secs, microsecs);
+               /* if the time delta is smaller than a usec, ignore */
+               if (*microsecs != 0)
+                       thread->vtimer_prof_save = tsum;
                break;
 
        case TASK_VTIMER_RLIM:
                tsum = timer_grab(&thread->user_timer);
                tsum += timer_grab(&thread->system_timer);
-               tdelt = tsum - thread->vtimer_rlim_save;
+               tdelt = (uint32_t)(tsum - thread->vtimer_rlim_save);
                thread->vtimer_rlim_save = tsum;
+               absolutetime_to_microtime(tdelt, &secs, microsecs);
                break;
        }
 
-       absolutetime_to_microtime(tdelt, &secs, microsecs);
 }
 
 /*
@@ -1705,6 +2301,70 @@ task_synchronizer_destroy_all(task_t task)
        }
 }
 
+/*
+ * Install default (machine-dependent) initial thread state 
+ * on the task.  Subsequent thread creation will have this initial
+ * state set on the thread by machine_thread_inherit_taskwide().
+ * Flavors and structures are exactly the same as those to thread_set_state()
+ */
+kern_return_t 
+task_set_state(
+       task_t task, 
+       int flavor, 
+       thread_state_t state, 
+       mach_msg_type_number_t state_count)
+{
+       kern_return_t ret;
+
+       if (task == TASK_NULL) {
+               return (KERN_INVALID_ARGUMENT);
+       }
+
+       task_lock(task);
+
+       if (!task->active) {
+               task_unlock(task);
+               return (KERN_FAILURE);
+       }
+
+       ret = machine_task_set_state(task, flavor, state, state_count);
+
+       task_unlock(task);
+       return ret;
+}
+
+/*
+ * Examine the default (machine-dependent) initial thread state 
+ * on the task, as set by task_set_state().  Flavors and structures
+ * are exactly the same as those passed to thread_get_state().
+ */
+kern_return_t 
+task_get_state(
+       task_t  task, 
+       int     flavor,
+       thread_state_t state,
+       mach_msg_type_number_t *state_count)
+{
+       kern_return_t ret;
+
+       if (task == TASK_NULL) {
+               return (KERN_INVALID_ARGUMENT);
+       }
+
+       task_lock(task);
+
+       if (!task->active) {
+               task_unlock(task);
+               return (KERN_FAILURE);
+       }
+
+       ret = machine_task_get_state(task, flavor, state, state_count);
+
+       task_unlock(task);
+       return ret;
+}
+
+
 /*
  * We need to export some functions to other components that
  * are currently implemented in macros within the osfmk
@@ -1718,6 +2378,16 @@ boolean_t is_kerneltask(task_t t)
        return (FALSE);
 }
 
+int
+check_for_tasksuspend(task_t task)
+{
+
+       if (task == TASK_NULL)
+               return (0);
+
+       return (task->suspend_count > 0);
+}
+
 #undef current_task
 task_t current_task(void);
 task_t current_task(void)
@@ -1735,6 +2405,24 @@ task_reference(
                task_reference_internal(task);
 }
 
+/* 
+ * This routine is called always with task lock held.
+ * And it returns a thread handle without reference as the caller
+ * operates on it under the task lock held.
+ */
+thread_t
+task_findtid(task_t task, uint64_t tid)
+{
+       thread_t thread= THREAD_NULL;
+
+       queue_iterate(&task->threads, thread, thread_t, task_threads) {
+                       if (thread->thread_id == tid)
+                               return(thread);
+       }
+       return(THREAD_NULL);
+}
+
+
 #if CONFIG_MACF_MACH
 /*
  * Protect 2 task labels against modification by adding a reference on