]> git.saurik.com Git - apple/xnu.git/blobdiff - osfmk/vm/vm_object.c
xnu-1699.24.23.tar.gz
[apple/xnu.git] / osfmk / vm / vm_object.c
index c71627a4f204d3e3469b85596c20e396e8350ffc..2f7d54e3c1d05b002742f6367b3b89505ad6adef 100644 (file)
@@ -80,6 +80,7 @@
 #include <kern/lock.h>
 #include <kern/queue.h>
 #include <kern/xpr.h>
+#include <kern/kalloc.h>
 #include <kern/zalloc.h>
 #include <kern/host.h>
 #include <kern/host_statistics.h>
@@ -261,28 +262,40 @@ unsigned int vm_page_purged_others = 0;
 static vm_object_t     vm_object_cache_trim(
                                boolean_t called_from_vm_object_deallocate);
 
-static queue_head_t    vm_object_cached_list;
-static int             vm_object_cached_count=0;
+static void            vm_object_deactivate_all_pages(
+                               vm_object_t     object);
+
 static int             vm_object_cached_high;  /* highest # cached objects */
 static int             vm_object_cached_max = 512;     /* may be patched*/
 
-static lck_mtx_t       vm_object_cached_lock_data;
-static lck_mtx_ext_t   vm_object_cached_lock_data_ext;
-
 #define vm_object_cache_lock()         \
                lck_mtx_lock(&vm_object_cached_lock_data)
 #define vm_object_cache_lock_try()             \
                lck_mtx_try_lock(&vm_object_cached_lock_data)
+
+#endif /* VM_OBJECT_CACHE */
+
+static queue_head_t    vm_object_cached_list;
+static uint32_t                vm_object_cache_pages_freed = 0;
+static uint32_t                vm_object_cache_pages_moved = 0;
+static uint32_t                vm_object_cache_pages_skipped = 0;
+static uint32_t                vm_object_cache_adds = 0;
+static uint32_t                vm_object_cached_count = 0;
+static lck_mtx_t       vm_object_cached_lock_data;
+static lck_mtx_ext_t   vm_object_cached_lock_data_ext;
+
+static uint32_t                vm_object_page_grab_failed = 0;
+static uint32_t                vm_object_page_grab_skipped = 0;
+static uint32_t                vm_object_page_grab_returned = 0;
+static uint32_t                vm_object_page_grab_pmapped = 0;
+static uint32_t                vm_object_page_grab_reactivations = 0;
+
 #define vm_object_cache_lock_spin()            \
                lck_mtx_lock_spin(&vm_object_cached_lock_data)
 #define vm_object_cache_unlock()       \
                lck_mtx_unlock(&vm_object_cached_lock_data)
 
-#endif /* VM_OBJECT_CACHE */
-
-
-static void            vm_object_deactivate_all_pages(
-                               vm_object_t     object);
+static void    vm_object_cache_remove_locked(vm_object_t);
 
 
 #define        VM_OBJECT_HASH_COUNT            1024
@@ -333,6 +346,10 @@ unsigned int vm_object_reap_count_async = 0;
 #define vm_object_reaper_unlock()      \
                lck_mtx_unlock(&vm_object_reaper_lock_data)
 
+#if 0
+#undef KERNEL_DEBUG
+#define KERNEL_DEBUG KERNEL_DEBUG_CONSTANT
+#endif
 
 
 static lck_mtx_t *
@@ -446,7 +463,7 @@ _vm_object_allocate(
        queue_init(&object->uplq);
 #endif /* UPL_DEBUG */
        vm_object_lock_init(object);
-       object->size = size;
+       object->vo_size = size;
 }
 
 __private_extern__ vm_object_t
@@ -467,7 +484,8 @@ vm_object_allocate(
 
 
 lck_grp_t              vm_object_lck_grp;
-lck_grp_attr_t vm_object_lck_grp_attr;
+lck_grp_t              vm_object_cache_lck_grp;
+lck_grp_attr_t         vm_object_lck_grp_attr;
 lck_attr_t             vm_object_lck_attr;
 lck_attr_t             kernel_object_lck_attr;
 
@@ -485,18 +503,18 @@ vm_object_bootstrap(void)
                                round_page(512*1024),
                                round_page(12*1024),
                                "vm objects");
+       zone_change(vm_object_zone, Z_CALLERACCT, FALSE); /* don't charge caller */
        zone_change(vm_object_zone, Z_NOENCRYPT, TRUE);
 
        vm_object_init_lck_grp();
 
-#if VM_OBJECT_CACHE
        queue_init(&vm_object_cached_list);
 
        lck_mtx_init_ext(&vm_object_cached_lock_data,
                &vm_object_cached_lock_data_ext,
-               &vm_object_lck_grp,
+               &vm_object_cache_lck_grp,
                &vm_object_lck_attr);
-#endif
+
        queue_init(&vm_object_reaper_queue);
 
        for (i = 0; i < VM_OBJECT_HASH_LOCK_COUNT; i++) {
@@ -515,6 +533,7 @@ vm_object_bootstrap(void)
                              round_page(512*1024),
                              round_page(12*1024),
                              "vm object hash entries");
+       zone_change(vm_object_hash_zone, Z_CALLERACCT, FALSE);
        zone_change(vm_object_hash_zone, Z_NOENCRYPT, TRUE);
 
        for (i = 0; i < VM_OBJECT_HASH_COUNT; i++)
@@ -538,7 +557,7 @@ vm_object_bootstrap(void)
         */
        vm_object_lock_init(&vm_object_template);
 #endif
-       vm_object_template.size = 0;
+       vm_object_template.vo_size = 0;
        vm_object_template.memq_hint = VM_PAGE_NULL;
        vm_object_template.ref_count = 1;
 #if    TASK_SWAPPER
@@ -549,7 +568,7 @@ vm_object_bootstrap(void)
        vm_object_template.reusable_page_count = 0;
        vm_object_template.copy = VM_OBJECT_NULL;
        vm_object_template.shadow = VM_OBJECT_NULL;
-       vm_object_template.shadow_offset = (vm_object_offset_t) 0;
+       vm_object_template.vo_shadow_offset = (vm_object_offset_t) 0;
        vm_object_template.pager = MEMORY_OBJECT_NULL;
        vm_object_template.paging_offset = 0;
        vm_object_template.pager_control = MEMORY_OBJECT_CONTROL_NULL;
@@ -590,6 +609,7 @@ vm_object_bootstrap(void)
        vm_object_template.sequential = (vm_object_offset_t) 0;
        vm_object_template.pages_created = 0;
        vm_object_template.pages_used = 0;
+       vm_object_template.scan_collisions = 0;
 
 #if    MACH_PAGEMAP
        vm_object_template.existence_map = VM_EXTERNAL_NULL;
@@ -600,7 +620,8 @@ vm_object_bootstrap(void)
 #endif /* MACH_ASSERT */
 
        /* cache bitfields */
-       vm_object_template.wimg_bits = VM_WIMG_DEFAULT;
+       vm_object_template.wimg_bits = VM_WIMG_USE_DEFAULT;
+       vm_object_template.set_cache_attr = FALSE;
        vm_object_template.code_signed = FALSE;
        vm_object_template.hashed = FALSE;
        vm_object_template.transposed = FALSE;
@@ -622,6 +643,7 @@ vm_object_bootstrap(void)
        vm_object_template.objq.next=NULL;
        vm_object_template.objq.prev=NULL;
 
+       vm_object_template.vo_cache_ts = 0;
        
        /*
         *      Initialize the "kernel object"
@@ -704,6 +726,7 @@ vm_object_init_lck_grp(void)
         */
        lck_grp_attr_setdefault(&vm_object_lck_grp_attr);
        lck_grp_init(&vm_object_lck_grp, "vm_object", &vm_object_lck_grp_attr);
+       lck_grp_init(&vm_object_cache_lck_grp, "vm_object_cache", &vm_object_lck_grp_attr);
        lck_attr_setdefault(&vm_object_lck_attr);
        lck_attr_setdefault(&kernel_object_lck_attr);
        lck_attr_cleardebug(&kernel_object_lck_attr);
@@ -838,17 +861,6 @@ vm_object_deallocate(
                                vm_object_lock(object);
                                vm_object_mapping_end(object);
                        }
-                       /*
-                        * recheck the ref_count since we dropped the object lock
-                        * to call 'memory_object_last_unmap'... it's possible
-                        * additional references got taken and we only want
-                        * to deactivate the pages if this 'named' object will only
-                        * referenced by the backing pager once we drop our reference
-                        * below
-                        */
-                       if (!object->terminating && object->ref_count == 2)
-                               vm_object_deactivate_all_pages(object);
-
                        assert(object->ref_count > 0);
                }
 
@@ -1073,6 +1085,360 @@ vm_object_deallocate(
 }
 
 
+
+vm_page_t
+vm_object_page_grab(
+       vm_object_t     object)
+{
+       vm_page_t       p, next_p;
+       int             p_limit = 0;
+       int             p_skipped = 0;
+
+       vm_object_lock_assert_exclusive(object);
+
+       next_p = (vm_page_t)queue_first(&object->memq);
+       p_limit = MIN(50, object->resident_page_count);
+
+       while (!queue_end(&object->memq, (queue_entry_t)next_p) && --p_limit > 0) {
+
+               p = next_p;
+               next_p = (vm_page_t)queue_next(&next_p->listq);
+
+               if (VM_PAGE_WIRED(p) || p->busy || p->cleaning || p->fictitious)
+                       goto move_page_in_obj;
+
+               if (p->pmapped || p->dirty || p->precious) {
+                       vm_page_lockspin_queues();
+
+                       if (p->pmapped) {
+                               int refmod_state;
+
+                               vm_object_page_grab_pmapped++;
+
+                               if (p->reference == FALSE || p->dirty == FALSE) {
+
+                                       refmod_state = pmap_get_refmod(p->phys_page);
+
+                                       if (refmod_state & VM_MEM_REFERENCED)
+                                               p->reference = TRUE;
+                                       if (refmod_state & VM_MEM_MODIFIED)
+                                               p->dirty = TRUE;
+                               }
+                               if (p->dirty == FALSE && p->precious == FALSE) {
+
+                                       refmod_state = pmap_disconnect(p->phys_page);
+
+                                       if (refmod_state & VM_MEM_REFERENCED)
+                                               p->reference = TRUE;
+                                       if (refmod_state & VM_MEM_MODIFIED)
+                                               p->dirty = TRUE;
+
+                                       if (p->dirty == FALSE)
+                                               goto take_page;
+                               }
+                       }
+                       if (p->inactive && p->reference == TRUE) {
+                               vm_page_activate(p);
+
+                               VM_STAT_INCR(reactivations);
+                               vm_object_page_grab_reactivations++;
+                       }
+                       vm_page_unlock_queues();
+move_page_in_obj:
+                       queue_remove(&object->memq, p, vm_page_t, listq);
+                       queue_enter(&object->memq, p, vm_page_t, listq);
+
+                       p_skipped++;
+                       continue;
+               }
+               vm_page_lockspin_queues();
+take_page:
+               vm_page_free_prepare_queues(p);
+               vm_object_page_grab_returned++;
+               vm_object_page_grab_skipped += p_skipped;
+
+               vm_page_unlock_queues();
+
+               vm_page_free_prepare_object(p, TRUE);
+               
+               return (p);
+       }
+       vm_object_page_grab_skipped += p_skipped;
+       vm_object_page_grab_failed++;
+
+       return (NULL);
+}
+
+
+
+#define EVICT_PREPARE_LIMIT    64
+#define EVICT_AGE              10
+
+static clock_sec_t     vm_object_cache_aging_ts = 0;
+
+static void
+vm_object_cache_remove_locked(
+       vm_object_t     object)
+{
+       queue_remove(&vm_object_cached_list, object, vm_object_t, objq);
+       object->objq.next = NULL;
+       object->objq.prev = NULL;
+
+       vm_object_cached_count--;
+}
+
+void
+vm_object_cache_remove(
+       vm_object_t     object)
+{
+       vm_object_cache_lock_spin();
+
+       if (object->objq.next || object->objq.prev)
+               vm_object_cache_remove_locked(object);
+
+       vm_object_cache_unlock();
+}
+
+void
+vm_object_cache_add(
+       vm_object_t     object)
+{
+       clock_sec_t sec;
+       clock_nsec_t nsec;
+
+       if (object->resident_page_count == 0)
+               return;
+       clock_get_system_nanotime(&sec, &nsec);
+
+       vm_object_cache_lock_spin();
+
+       if (object->objq.next == NULL && object->objq.prev == NULL) {
+               queue_enter(&vm_object_cached_list, object, vm_object_t, objq);
+               object->vo_cache_ts = sec + EVICT_AGE;
+               object->vo_cache_pages_to_scan = object->resident_page_count;
+
+               vm_object_cached_count++;
+               vm_object_cache_adds++;
+       }
+       vm_object_cache_unlock();
+}
+
+int
+vm_object_cache_evict(
+       int     num_to_evict,
+       int     max_objects_to_examine)
+{
+       vm_object_t     object = VM_OBJECT_NULL;
+       vm_object_t     next_obj = VM_OBJECT_NULL;
+       vm_page_t       local_free_q = VM_PAGE_NULL;
+       vm_page_t       p;
+       vm_page_t       next_p;
+       int             object_cnt = 0;
+       vm_page_t       ep_array[EVICT_PREPARE_LIMIT];
+       int             ep_count;
+       int             ep_limit;
+       int             ep_index;
+       int             ep_freed = 0;
+       int             ep_moved = 0;
+       uint32_t        ep_skipped = 0;
+       clock_sec_t     sec;
+       clock_nsec_t    nsec;
+
+       KERNEL_DEBUG(0x13001ec | DBG_FUNC_START, 0, 0, 0, 0, 0);
+       /*
+        * do a couple of quick checks to see if it's 
+        * worthwhile grabbing the lock
+        */
+       if (queue_empty(&vm_object_cached_list)) {
+               KERNEL_DEBUG(0x13001ec | DBG_FUNC_END, 0, 0, 0, 0, 0);
+               return (0);
+       }
+       clock_get_system_nanotime(&sec, &nsec);
+
+       /*
+        * the object on the head of the queue has not
+        * yet sufficiently aged
+        */
+       if (sec < vm_object_cache_aging_ts) {
+               KERNEL_DEBUG(0x13001ec | DBG_FUNC_END, 0, 0, 0, 0, 0);
+               return (0);
+       }
+       /*
+        * don't need the queue lock to find 
+        * and lock an object on the cached list
+        */
+       vm_page_unlock_queues();
+
+       vm_object_cache_lock_spin();
+
+       for (;;) {
+               next_obj = (vm_object_t)queue_first(&vm_object_cached_list);
+
+               while (!queue_end(&vm_object_cached_list, (queue_entry_t)next_obj) && object_cnt++ < max_objects_to_examine) {
+
+                       object = next_obj;
+                       next_obj = (vm_object_t)queue_next(&next_obj->objq);
+                       
+                       if (sec < object->vo_cache_ts) {
+                               KERNEL_DEBUG(0x130020c, object, object->resident_page_count, object->vo_cache_ts, sec, 0);
+
+                               vm_object_cache_aging_ts = object->vo_cache_ts;
+                               object = VM_OBJECT_NULL;
+                               break;
+                       }
+                       if (!vm_object_lock_try_scan(object)) {
+                               /*
+                                * just skip over this guy for now... if we find
+                                * an object to steal pages from, we'll revist in a bit...
+                                * hopefully, the lock will have cleared
+                                */
+                               KERNEL_DEBUG(0x13001f8, object, object->resident_page_count, 0, 0, 0);
+
+                               object = VM_OBJECT_NULL;
+                               continue;
+                       }
+                       if (queue_empty(&object->memq) || object->vo_cache_pages_to_scan == 0) {
+                               /*
+                                * this case really shouldn't happen, but it's not fatal
+                                * so deal with it... if we don't remove the object from
+                                * the list, we'll never move past it.
+                                */
+                               KERNEL_DEBUG(0x13001fc, object, object->resident_page_count, ep_freed, ep_moved, 0);
+                               
+                               vm_object_cache_remove_locked(object);
+                               vm_object_unlock(object);
+                               object = VM_OBJECT_NULL;
+                               continue;
+                       }
+                       /*
+                        * we have a locked object with pages...
+                        * time to start harvesting
+                        */
+                       break;
+               }
+               vm_object_cache_unlock();
+
+               if (object == VM_OBJECT_NULL)
+                       break;
+
+               /*
+                * object is locked at this point and
+                * has resident pages
+                */
+               next_p = (vm_page_t)queue_first(&object->memq);
+
+               /*
+                * break the page scan into 2 pieces to minimize the time spent
+                * behind the page queue lock...
+                * the list of pages on these unused objects is likely to be cold
+                * w/r to the cpu cache which increases the time to scan the list
+                * tenfold...  and we may have a 'run' of pages we can't utilize that
+                * needs to be skipped over...
+                */
+               if ((ep_limit = num_to_evict - (ep_freed + ep_moved)) > EVICT_PREPARE_LIMIT)
+                       ep_limit = EVICT_PREPARE_LIMIT;
+               ep_count = 0;
+
+               while (!queue_end(&object->memq, (queue_entry_t)next_p) && object->vo_cache_pages_to_scan && ep_count < ep_limit) {
+
+                       p = next_p;
+                       next_p = (vm_page_t)queue_next(&next_p->listq);
+
+                       object->vo_cache_pages_to_scan--;
+
+                       if (VM_PAGE_WIRED(p) || p->busy || p->cleaning) {
+                               queue_remove(&object->memq, p, vm_page_t, listq);
+                               queue_enter(&object->memq, p, vm_page_t, listq);
+
+                               ep_skipped++;
+                               continue;
+                       }
+                       if (p->wpmapped || p->dirty || p->precious) {
+                               queue_remove(&object->memq, p, vm_page_t, listq);
+                               queue_enter(&object->memq, p, vm_page_t, listq);
+
+                               pmap_clear_reference(p->phys_page);
+                       }
+                       ep_array[ep_count++] = p;
+               }
+               KERNEL_DEBUG(0x13001f4 | DBG_FUNC_START, object, object->resident_page_count, ep_freed, ep_moved, 0);
+
+               vm_page_lockspin_queues();
+
+               for (ep_index = 0; ep_index < ep_count; ep_index++) {
+
+                       p = ep_array[ep_index];
+
+                       if (p->wpmapped || p->dirty || p->precious) {
+                               p->reference = FALSE;
+                               p->no_cache = FALSE;
+
+                               VM_PAGE_QUEUES_REMOVE(p);
+                               VM_PAGE_ENQUEUE_INACTIVE(p, TRUE);
+
+                               ep_moved++;
+                       } else {
+                               vm_page_free_prepare_queues(p);
+
+                               assert(p->pageq.next == NULL && p->pageq.prev == NULL);
+                               /*
+                                * Add this page to our list of reclaimed pages,
+                                * to be freed later.
+                                */
+                               p->pageq.next = (queue_entry_t) local_free_q;
+                               local_free_q = p;
+
+                               ep_freed++;
+                       }
+               }
+               vm_page_unlock_queues();
+
+               KERNEL_DEBUG(0x13001f4 | DBG_FUNC_END, object, object->resident_page_count, ep_freed, ep_moved, 0);
+
+               if (local_free_q) {
+                       vm_page_free_list(local_free_q, TRUE);
+                       local_free_q = VM_PAGE_NULL;
+               }
+               if (object->vo_cache_pages_to_scan == 0) {
+                       KERNEL_DEBUG(0x1300208, object, object->resident_page_count, ep_freed, ep_moved, 0);
+
+                       vm_object_cache_remove(object);
+
+                       KERNEL_DEBUG(0x13001fc, object, object->resident_page_count, ep_freed, ep_moved, 0);
+               }
+               /*
+                * done with this object
+                */
+               vm_object_unlock(object);
+               object = VM_OBJECT_NULL;
+
+               /*
+                * at this point, we are not holding any locks
+                */
+               if ((ep_freed + ep_moved) >= num_to_evict) {
+                       /*
+                        * we've reached our target for the
+                        * number of pages to evict
+                        */
+                       break;
+               }
+               vm_object_cache_lock_spin();
+       }
+       /*
+        * put the page queues lock back to the caller's
+        * idea of it 
+        */
+       vm_page_lock_queues();
+
+       vm_object_cache_pages_freed += ep_freed;
+       vm_object_cache_pages_moved += ep_moved;
+       vm_object_cache_pages_skipped += ep_skipped;
+
+       KERNEL_DEBUG(0x13001ec | DBG_FUNC_END, ep_freed, 0, 0, 0, 0);
+       return (ep_freed);
+}
+
+
 #if VM_OBJECT_CACHE
 /*
  *     Check to see whether we really need to trim
@@ -1233,6 +1599,9 @@ vm_object_terminate(
        object->terminating = TRUE;
        object->alive = FALSE;
 
+       if ( !object->internal && (object->objq.next || object->objq.prev))
+               vm_object_cache_remove(object);
+
        if (object->hashed) {
                lck_mtx_t       *lck;
 
@@ -1344,7 +1713,7 @@ vm_object_reap(
        /*
         * remove from purgeable queue if it's on
         */
-       if (object->objq.next || object->objq.prev) {
+       if (object->internal && (object->objq.next || object->objq.prev)) {
                purgeable_q_t queue = vm_purgeable_object_remove(object);
                assert(queue);
 
@@ -1393,7 +1762,7 @@ vm_object_reap(
        vm_object_unlock(object);
 
 #if    MACH_PAGEMAP
-       vm_external_destroy(object->existence_map, object->size);
+       vm_external_destroy(object->existence_map, object->vo_size);
 #endif /* MACH_PAGEMAP */
 
        object->shadow = VM_OBJECT_NULL;
@@ -1407,9 +1776,12 @@ vm_object_reap(
 }
 
 
+unsigned int vm_max_batch = 256;
 
 #define V_O_R_MAX_BATCH 128
 
+#define BATCH_LIMIT(max)       (vm_max_batch >= max ? max : vm_max_batch)
+
 
 #define VM_OBJ_REAP_FREELIST(_local_free_q, do_disconnect)             \
        MACRO_BEGIN                                                     \
@@ -1461,7 +1833,7 @@ vm_object_reap_pages(
 restart_after_sleep:
        if (queue_empty(&object->memq))
                return;
-       loop_count = V_O_R_MAX_BATCH + 1;
+       loop_count = BATCH_LIMIT(V_O_R_MAX_BATCH) + 1;
 
        vm_page_lockspin_queues();
 
@@ -1487,7 +1859,7 @@ restart_after_sleep:
                        } else
                                mutex_pause(0);
 
-                       loop_count = V_O_R_MAX_BATCH + 1;
+                       loop_count = BATCH_LIMIT(V_O_R_MAX_BATCH) + 1;
 
                        vm_page_lockspin_queues();
                }
@@ -1505,12 +1877,8 @@ restart_after_sleep:
                                 *
                                 * the laundry and pageout_queue flags are cleared...
                                 */
-#if CONFIG_EMBEDDED
-                               if (p->laundry) 
-                                       vm_pageout_throttle_up(p);
-#else
                                vm_pageout_throttle_up(p);
-#endif
+
                                if (p->pageout == TRUE) {
                                        /*
                                         * toss the wire count we picked up
@@ -1624,6 +1992,11 @@ restart_after_sleep:
                                p->busy = TRUE;
 
                                VM_PAGE_QUEUES_REMOVE(p);
+                               /*
+                                * flush page... page will be freed
+                                * upon completion of I/O
+                                */
+                               vm_pageout_cluster(p);
 
                                vm_page_unlock_queues();
                                /*
@@ -1632,11 +2005,6 @@ restart_after_sleep:
                                VM_OBJ_REAP_FREELIST(local_free_q,
                                                     disconnect_on_release);
 
-                               /*
-                                * flush page... page will be freed
-                                * upon completion of I/O
-                                */
-                               vm_pageout_cluster(p);
                                vm_object_paging_wait(object, THREAD_UNINT);
 
                                goto restart_after_sleep;
@@ -1885,6 +2253,8 @@ vm_object_destroy(
 }
 
 
+#if VM_OBJECT_CACHE
+
 #define VM_OBJ_DEACT_ALL_STATS DEBUG
 #if VM_OBJ_DEACT_ALL_STATS
 uint32_t vm_object_deactivate_all_pages_batches = 0;
@@ -1909,7 +2279,7 @@ vm_object_deactivate_all_pages(
 #endif /* VM_OBJ_DEACT_ALL_STATS */
 #define V_O_D_A_P_MAX_BATCH    256
 
-       loop_count = V_O_D_A_P_MAX_BATCH;
+       loop_count = BATCH_LIMIT(V_O_D_A_P_MAX_BATCH);
 #if VM_OBJ_DEACT_ALL_STATS
        pages_count = 0;
 #endif /* VM_OBJ_DEACT_ALL_STATS */
@@ -1924,7 +2294,7 @@ vm_object_deactivate_all_pages(
                        pages_count = 0;
 #endif /* VM_OBJ_DEACT_ALL_STATS */
                        lck_mtx_yield(&vm_page_queue_lock);
-                       loop_count = V_O_D_A_P_MAX_BATCH;
+                       loop_count = BATCH_LIMIT(V_O_D_A_P_MAX_BATCH);
                }
                if (!p->busy && !p->throttled) {
 #if VM_OBJ_DEACT_ALL_STATS
@@ -1943,133 +2313,7 @@ vm_object_deactivate_all_pages(
 #endif /* VM_OBJ_DEACT_ALL_STATS */
        vm_page_unlock_queues();
 }
-
-
-
-/*
- * when deallocating pages it is necessary to hold 
- * the vm_page_queue_lock (a hot global lock) for certain operations
- * on the page... however, the majority of the work can be done
- * while merely holding the object lock... to mitigate the time spent behind the
- * global lock, go to a 2 pass algorithm... collect pages up to DELAYED_WORK_LIMIT
- * while doing all of the work that doesn't require the vm_page_queue_lock...
- * them call dw_do_work to acquire the vm_page_queue_lock and do the
- * necessary work for each page... we will grab the busy bit on the page
- * so that dw_do_work can drop the object lock if it can't immediately take the
- * vm_page_queue_lock in order to compete for the locks in the same order that
- * vm_pageout_scan takes them.
- */
-
-#define DELAYED_WORK_LIMIT     32
-
-#define DW_clear_reference     0x01
-#define DW_move_page           0x02
-#define DW_clear_busy          0x04
-#define DW_PAGE_WAKEUP         0x08
-
-
-struct dw {
-       vm_page_t       dw_m;
-       int             dw_mask;
-};
-
-static void dw_do_work(vm_object_t object, struct dw *dwp, int dw_count);
-
-
-static void
-dw_do_work(
-       vm_object_t     object,
-       struct dw       *dwp,
-       int             dw_count)
-{
-       vm_page_t       m;
-       int             j;
-
-       /*
-        * pageout_scan takes the vm_page_lock_queues first
-        * then tries for the object lock... to avoid what
-        * is effectively a lock inversion, we'll go to the
-        * trouble of taking them in that same order... otherwise
-        * if this object contains the majority of the pages resident
-        * in the UBC (or a small set of large objects actively being
-        * worked on contain the majority of the pages), we could
-        * cause the pageout_scan thread to 'starve' in its attempt
-        * to find pages to move to the free queue, since it has to
-        * successfully acquire the object lock of any candidate page
-        * before it can steal/clean it.
-        */
-       if (!vm_page_trylockspin_queues()) {
-               vm_object_unlock(object);
-
-               vm_page_lockspin_queues();
-
-               for (j = 0; ; j++) {
-                       if (!vm_object_lock_avoid(object) &&
-                           _vm_object_lock_try(object))
-                               break;
-                       vm_page_unlock_queues();
-                       mutex_pause(j);
-                       vm_page_lockspin_queues();
-               }
-       }
-       for (j = 0; j < dw_count; j++, dwp++) {
-
-               m = dwp->dw_m;
-
-               if (dwp->dw_mask & DW_clear_reference)
-                       m->reference = FALSE;
-
-               if (dwp->dw_mask & DW_move_page) {
-                       VM_PAGE_QUEUES_REMOVE(m);
-
-                       assert(!m->laundry);
-                       assert(m->object != kernel_object);
-                       assert(m->pageq.next == NULL &&
-                              m->pageq.prev == NULL);
-                                       
-                       if (m->zero_fill) {
-                               queue_enter_first(&vm_page_queue_zf, m, vm_page_t, pageq);
-                               vm_zf_queue_count++;
-                       } else {
-                               queue_enter_first(&vm_page_queue_inactive, m, vm_page_t, pageq);
-                       }
-                       m->inactive = TRUE;
-
-                       if (!m->fictitious) {
-                               vm_page_inactive_count++;
-                               token_new_pagecount++;
-                       } else {
-                               assert(m->phys_page == vm_page_fictitious_addr);
-                       }
-               }
-               if (dwp->dw_mask & DW_clear_busy)
-                       dwp->dw_m->busy = FALSE;
-
-               if (dwp->dw_mask & DW_PAGE_WAKEUP)
-                       PAGE_WAKEUP(dwp->dw_m);
-       }
-       vm_page_unlock_queues();
-
-#if CONFIG_EMBEDDED
-       {
-       int percent_avail;
-
-       /*
-        * Decide if we need to send a memory status notification.
-        */
-       percent_avail = 
-               (vm_page_active_count + vm_page_inactive_count + 
-                vm_page_speculative_count + vm_page_free_count +
-                (IP_VALID(memory_manager_default)?0:vm_page_purgeable_count) ) * 100 /
-               atop_64(max_mem);
-       if (percent_avail >= (kern_memorystatus_level + 5) || 
-           percent_avail <= (kern_memorystatus_level - 5)) {
-               kern_memorystatus_level = percent_avail;
-               thread_wakeup((event_t)&kern_memorystatus_wakeup);
-       }
-       }
-#endif
-}
+#endif /* VM_OBJECT_CACHE */
 
 
 
@@ -2121,6 +2365,7 @@ typedef uint64_t  chunk_state_t;
                MARK_PAGE_HANDLED(c, p);                                \
        MACRO_END
 
+
 /*
  * Return true if all pages in the chunk have not yet been processed.
  */
@@ -2211,6 +2456,7 @@ page_is_paged_out(
 }
 
 
+
 /*
  * Deactivate the pages in the specified object and range.  If kill_page is set, also discard any
  * page modified state from the pmap.  Update the chunk_state as we go along.  The caller must specify
@@ -2232,9 +2478,10 @@ deactivate_pages_in_object(
 {
        vm_page_t       m;
        int             p;
-       struct  dw      dw_array[DELAYED_WORK_LIMIT];
-       struct  dw      *dwp;
+       struct vm_page_delayed_work     dw_array[DEFAULT_DELAYED_WORK_LIMIT];
+       struct vm_page_delayed_work     *dwp;
        int             dw_count;
+       int             dw_limit;
        unsigned int    reusable = 0;
 
 
@@ -2247,6 +2494,7 @@ deactivate_pages_in_object(
 
        dwp = &dw_array[0];
        dw_count = 0;
+       dw_limit = DELAYED_WORK_LIMIT(DEFAULT_DELAYED_WORK_LIMIT);
 
        for(p = 0; size && CHUNK_NOT_COMPLETE(*chunk_state); p++, size -= PAGE_SIZE_64, offset += PAGE_SIZE_64) {
 
@@ -2307,39 +2555,23 @@ deactivate_pages_in_object(
                                                object->reusable_page_count++;
                                                assert(object->resident_page_count >= object->reusable_page_count);
                                                reusable++;
-#if CONFIG_EMBEDDED
-                                       } else {
-                                               if (m->reusable) {
-                                                       m->reusable = FALSE;
-                                                       object->reusable_page_count--;
-                                               }
-#endif
                                        }
                                }
                                pmap_clear_refmod(m->phys_page, clear_refmod);
 
                                if (!m->throttled && !(reusable_page || all_reusable))
                                        dwp->dw_mask |= DW_move_page;
-                               /*
-                                * dw_do_work may need to drop the object lock
-                                * if it does, we need the pages its looking at to
-                                * be held stable via the busy bit.
-                                */
-                               m->busy = TRUE;
-                               dwp->dw_mask |= (DW_clear_busy | DW_PAGE_WAKEUP);
-
-                               dwp->dw_m = m;
-                               dwp++;
-                               dw_count++;
+                               
+                               VM_PAGE_ADD_DELAYED_WORK(dwp, m, dw_count);
 
-                               if (dw_count >= DELAYED_WORK_LIMIT) {
+                               if (dw_count >= dw_limit) {
                                        if (reusable) {
                                                OSAddAtomic(reusable,
                                                            &vm_page_stats_reusable.reusable_count);
                                                vm_page_stats_reusable.reusable += reusable;
                                                reusable = 0;
                                        }
-                                       dw_do_work(object, &dw_array[0], dw_count);
+                                       vm_page_do_delayed_work(object, &dw_array[0], dw_count);
 
                                        dwp = &dw_array[0];
                                        dw_count = 0;
@@ -2378,7 +2610,7 @@ deactivate_pages_in_object(
        }
                
        if (dw_count)
-               dw_do_work(object, &dw_array[0], dw_count);
+               vm_page_do_delayed_work(object, &dw_array[0], dw_count);
 }
 
 
@@ -2447,7 +2679,7 @@ deactivate_a_chunk(
                        kill_page = FALSE;
                        reusable_page = FALSE;
                        all_reusable = FALSE;
-                       offset += object->shadow_offset;
+                       offset += object->vo_shadow_offset;
                        vm_object_lock(tmp_object);
                }
 
@@ -2492,20 +2724,19 @@ vm_object_deactivate_pages(
 
        all_reusable = FALSE;
        if (reusable_page &&
-           object->size != 0 &&
-           object->size == size &&
+           object->internal &&
+           object->vo_size != 0 &&
+           object->vo_size == size &&
            object->reusable_page_count == 0) {
                all_reusable = TRUE;
                reusable_page = FALSE;
        }
 
-#if CONFIG_EMBEDDED
        if ((reusable_page || all_reusable) && object->all_reusable) {
                /* This means MADV_FREE_REUSABLE has been called twice, which 
                 * is probably illegal. */
                return;
        }
-#endif
 
        while (size) {
                length = deactivate_a_chunk(object, offset, size, kill_page, reusable_page, all_reusable);
@@ -2564,7 +2795,7 @@ vm_object_reuse_pages(
        if (object->all_reusable) {
                assert(object->reusable_page_count == 0);
                object->all_reusable = FALSE;
-               if (end_offset - start_offset == object->size ||
+               if (end_offset - start_offset == object->vo_size ||
                    !allow_partial_reuse) {
                        vm_page_stats_reusable.all_reuse_calls++;
                        reused = object->resident_page_count;
@@ -2662,10 +2893,10 @@ vm_object_pmap_protect(
                } else {
                        vm_object_offset_t phys_start, phys_end, phys_addr;
 
-                       phys_start = object->shadow_offset + offset;
+                       phys_start = object->vo_shadow_offset + offset;
                        phys_end = phys_start + size;
                        assert(phys_start <= phys_end);
-                       assert(phys_end <= object->shadow_offset + object->size);
+                       assert(phys_end <= object->vo_shadow_offset + object->vo_size);
                        vm_object_unlock(object);
 
                        for (phys_addr = phys_start;
@@ -2754,7 +2985,7 @@ vm_object_pmap_protect(
 
                next_object = object->shadow;
                if (next_object != VM_OBJECT_NULL) {
-                   offset += object->shadow_offset;
+                   offset += object->vo_shadow_offset;
                    vm_object_lock(next_object);
                    vm_object_unlock(object);
                    object = next_object;
@@ -2858,6 +3089,8 @@ vm_object_copy_slowly(
        fault_info.hi_offset = src_offset + size;
        fault_info.no_cache  = FALSE;
        fault_info.stealth = TRUE;
+       fault_info.io_sync = FALSE;
+       fault_info.cs_bypass = FALSE;
        fault_info.mark_zf_absent = FALSE;
 
        for ( ;
@@ -2965,10 +3198,6 @@ vm_object_copy_slowly(
                        case VM_FAULT_RETRY:
                                break;
 
-                       case VM_FAULT_FICTITIOUS_SHORTAGE:
-                               vm_page_more_fictitious();
-                               break;
-
                        case VM_FAULT_MEMORY_SHORTAGE:
                                if (vm_page_wait(interruptible))
                                        break;
@@ -3198,8 +3427,8 @@ Retry:
                vm_object_lock(src_object);
                goto Retry;
        }
-       if (copy->size < src_offset+size)
-               copy->size = src_offset+size;
+       if (copy->vo_size < src_offset+size)
+               copy->vo_size = src_offset+size;
 
        if (!copy->pager_ready)
                check_ready = TRUE;
@@ -3365,7 +3594,7 @@ vm_object_copy_delayed(
                         *      needed).
                         */
 
-                       if (old_copy->size < copy_size) {
+                       if (old_copy->vo_size < copy_size) {
                                if (src_object_shared == TRUE) {
                                        vm_object_unlock(old_copy);
                                        vm_object_unlock(src_object);
@@ -3385,7 +3614,7 @@ vm_object_copy_delayed(
 
                                queue_iterate(&src_object->memq, p, vm_page_t, listq) {
                                        if (!p->fictitious && 
-                                           p->offset >= old_copy->size && 
+                                           p->offset >= old_copy->vo_size && 
                                            p->offset < copy_size) {
                                                if (VM_PAGE_WIRED(p)) {
                                                        vm_object_unlock(old_copy);
@@ -3403,7 +3632,7 @@ vm_object_copy_delayed(
                                                }
                                        }
                                }
-                               old_copy->size = copy_size;
+                               old_copy->vo_size = copy_size;
                        }
                        if (src_object_shared == TRUE)
                                vm_object_reference_shared(old_copy);
@@ -3426,8 +3655,8 @@ vm_object_copy_delayed(
                 * copy object will be large enough to back either the
                 * old copy object or the new mapping.
                 */
-               if (old_copy->size > copy_size)
-                       copy_size = old_copy->size;
+               if (old_copy->vo_size > copy_size)
+                       copy_size = old_copy->vo_size;
 
                if (new_copy == VM_OBJECT_NULL) {
                        vm_object_unlock(old_copy);
@@ -3439,7 +3668,7 @@ vm_object_copy_delayed(
                        src_object_shared = FALSE;
                        goto Retry;
                }
-               new_copy->size = copy_size;     
+               new_copy->vo_size = copy_size;  
 
                /*
                 *      The copy-object is always made large enough to
@@ -3449,7 +3678,7 @@ vm_object_copy_delayed(
                 */
 
                assert((old_copy->shadow == src_object) &&
-                   (old_copy->shadow_offset == (vm_object_offset_t) 0));
+                   (old_copy->vo_shadow_offset == (vm_object_offset_t) 0));
 
        } else if (new_copy == VM_OBJECT_NULL) {
                vm_object_unlock(src_object);
@@ -3521,7 +3750,7 @@ vm_object_copy_delayed(
         */
        vm_object_lock_assert_exclusive(new_copy);
        new_copy->shadow = src_object;
-       new_copy->shadow_offset = 0;
+       new_copy->vo_shadow_offset = 0;
        new_copy->shadowed = TRUE;      /* caller must set needs_copy */
 
        vm_object_lock_assert_exclusive(src_object);
@@ -3653,7 +3882,7 @@ vm_object_copy_strategically(
  *     The new object and offset into that object
  *     are returned in the source parameters.
  */
-boolean_t vm_object_shadow_check = FALSE;
+boolean_t vm_object_shadow_check = TRUE;
 
 __private_extern__ boolean_t
 vm_object_shadow(
@@ -3665,6 +3894,10 @@ vm_object_shadow(
        register vm_object_t    result;
 
        source = *object;
+       assert(source != VM_OBJECT_NULL);
+       if (source == VM_OBJECT_NULL)
+               return FALSE;
+
 #if 0
        /*
         * XXX FBDP
@@ -3684,11 +3917,19 @@ vm_object_shadow(
 
        /*
         *      Determine if we really need a shadow.
+        *
+        *      If the source object is larger than what we are trying
+        *      to create, then force the shadow creation even if the
+        *      ref count is 1.  This will allow us to [potentially]
+        *      collapse the underlying object away in the future
+        *      (freeing up the extra data it might contain and that
+        *      we don't need).
         */
-
-       if (vm_object_shadow_check && source->ref_count == 1 &&
+       if (vm_object_shadow_check &&
+           source->vo_size == length &&
+           source->ref_count == 1 &&
            (source->shadow == VM_OBJECT_NULL ||
-            source->shadow->copy == VM_OBJECT_NULL))
+            source->shadow->copy == VM_OBJECT_NULL) )
        {
                source->shadowed = FALSE;
                return FALSE;
@@ -3715,7 +3956,7 @@ vm_object_shadow(
         *      and fix up the offset into the new object.
         */
 
-       result->shadow_offset = *offset;
+       result->vo_shadow_offset = *offset;
 
        /*
         *      Return the new things
@@ -4088,21 +4329,21 @@ vm_object_pager_create(
        object->paging_offset = 0;
                
 #if    MACH_PAGEMAP
-       size = object->size;
+       size = object->vo_size;
 #endif /* MACH_PAGEMAP */
        vm_object_unlock(object);
 
 #if    MACH_PAGEMAP
        map = vm_external_create(size);
        vm_object_lock(object);
-       assert(object->size == size);
+       assert(object->vo_size == size);
        object->existence_map = map;
        vm_object_unlock(object);
 #endif /* MACH_PAGEMAP */
 
-       if ((uint32_t) object->size != object->size) {
+       if ((uint32_t) object->vo_size != object->vo_size) {
                panic("vm_object_pager_create(): object size 0x%llx >= 4GB\n",
-                     (uint64_t) object->size);
+                     (uint64_t) object->vo_size);
        }
 
        /*
@@ -4121,8 +4362,8 @@ vm_object_pager_create(
                assert(object->temporary);
 
                /* create our new memory object */
-               assert((vm_size_t) object->size == object->size);
-               (void) memory_object_create(dmm, (vm_size_t) object->size,
+               assert((vm_size_t) object->vo_size == object->vo_size);
+               (void) memory_object_create(dmm, (vm_size_t) object->vo_size,
                                            &pager);
 
                memory_object_default_deallocate(dmm);
@@ -4140,7 +4381,7 @@ vm_object_pager_create(
         *      copied by vm_object_enter().
         */
 
-       if (vm_object_enter(pager, object->size, TRUE, TRUE, FALSE) != object)
+       if (vm_object_enter(pager, object->vo_size, TRUE, TRUE, FALSE) != object)
                panic("vm_object_pager_create: mismatch");
 
        /*
@@ -4222,8 +4463,8 @@ vm_object_do_collapse(
        vm_object_lock_assert_exclusive(object);
        vm_object_lock_assert_exclusive(backing_object);
 
-       backing_offset = object->shadow_offset;
-       size = object->size;
+       backing_offset = object->vo_shadow_offset;
+       size = object->vo_size;
 
        /*
         *      Move all in-memory pages from backing_object
@@ -4363,10 +4604,10 @@ vm_object_do_collapse(
         *      this code should be fixed to salvage the map.
         */
        assert(object->existence_map == VM_EXTERNAL_NULL);
-       if (backing_offset || (size != backing_object->size)) {
+       if (backing_offset || (size != backing_object->vo_size)) {
                vm_external_discarded++;
                vm_external_destroy(backing_object->existence_map,
-                       backing_object->size);
+                       backing_object->vo_size);
        }
        else {
                vm_external_collapsed++;
@@ -4385,10 +4626,10 @@ vm_object_do_collapse(
        assert(!backing_object->phys_contiguous);
        object->shadow = backing_object->shadow;
        if (object->shadow) {
-               object->shadow_offset += backing_object->shadow_offset;
+               object->vo_shadow_offset += backing_object->vo_shadow_offset;
        } else {
                /* no shadow, therefore no shadow offset... */
-               object->shadow_offset = 0;
+               object->vo_shadow_offset = 0;
        }
        assert((object->shadow == VM_OBJECT_NULL) ||
               (object->shadow->copy != backing_object));
@@ -4456,10 +4697,10 @@ vm_object_do_bypass(
        assert(!backing_object->phys_contiguous);
        object->shadow = backing_object->shadow;
        if (object->shadow) {
-               object->shadow_offset += backing_object->shadow_offset;
+               object->vo_shadow_offset += backing_object->vo_shadow_offset;
        } else {
                /* no shadow, therefore no shadow offset... */
-               object->shadow_offset = 0;
+               object->vo_shadow_offset = 0;
        }
        
        /*
@@ -4762,7 +5003,7 @@ retry:
                 *      we have to make sure no pages in the backing object
                 *      "show through" before bypassing it.
                 */
-               size = atop(object->size);
+               size = atop(object->vo_size);
                rcount = object->resident_page_count;
                if (rcount != size) {
                        vm_object_offset_t      offset;
@@ -4821,7 +5062,7 @@ retry:
                         *
                         */
 
-                       backing_offset = object->shadow_offset;
+                       backing_offset = object->vo_shadow_offset;
                        backing_rcount = backing_object->resident_page_count;
 
 #if    MACH_PAGEMAP
@@ -4894,7 +5135,7 @@ retry:
                                        }
 
                                        offset = (p->offset - backing_offset);
-                                       if (offset < object->size &&
+                                       if (offset < object->vo_size &&
                                            offset != hint_offset &&
                                            !EXISTS_IN_OBJECT(object, offset, rc)) {
                                                /* found a dependency */
@@ -4928,7 +5169,7 @@ retry:
                                offset = hint_offset;
                                
                                while((offset =
-                                     (offset + PAGE_SIZE_64 < object->size) ?
+                                     (offset + PAGE_SIZE_64 < object->vo_size) ?
                                      (offset + PAGE_SIZE_64) : 0) != hint_offset) {
 
                                        /* Until we get more than one lookup lock */
@@ -5148,7 +5389,7 @@ vm_object_coalesce(
         *      Extend the object if necessary.
         */
        newsize = prev_offset + prev_size + next_size;
-       if (newsize > prev_object->size) {
+       if (newsize > prev_object->vo_size) {
 #if    MACH_PAGEMAP
                /*
                 *      We cannot extend an object that has existence info,
@@ -5161,7 +5402,7 @@ vm_object_coalesce(
                 */
                assert(prev_object->existence_map == VM_EXTERNAL_NULL);
 #endif /* MACH_PAGEMAP */
-               prev_object->size = newsize;
+               prev_object->vo_size = newsize;
        }
 
        vm_object_unlock(prev_object);
@@ -5346,7 +5587,7 @@ vm_object_print(db_expr_t db_addr, __unused boolean_t have_addr,
 
        db_indent += 2;
 
-       iprintf("size=0x%x", object->size);
+       iprintf("size=0x%x", object->vo_size);
        printf(", memq_hint=%p", object->memq_hint);
        printf(", ref_count=%d\n", object->ref_count);
        iprintf("");
@@ -5364,7 +5605,7 @@ vm_object_print(db_expr_t db_addr, __unused boolean_t have_addr,
                printf(" (depth %d)", i);
        }
        printf(", copy=0x%x", object->copy);
-       printf(", shadow_offset=0x%x", object->shadow_offset);
+       printf(", shadow_offset=0x%x", object->vo_shadow_offset);
        printf(", last_alloc=0x%x\n", object->last_alloc);
 
        iprintf("pager=0x%x", object->pager);
@@ -5455,7 +5696,7 @@ vm_object_print(db_expr_t db_addr, __unused boolean_t have_addr,
 
 #if    MACH_PAGEMAP
        iprintf("existence_map=");
-       vm_external_print(object->existence_map, object->size);
+       vm_external_print(object->existence_map, object->vo_size);
 #endif /* MACH_PAGEMAP */
 #if    MACH_ASSERT
        iprintf("paging_object=0x%x\n", object->paging_object);
@@ -5642,8 +5883,8 @@ vm_object_populate_with_private(
                
                /* shadows on contiguous memory are not allowed */
                /* we therefore can use the offset field */
-               object->shadow_offset = (vm_object_offset_t)phys_page << PAGE_SHIFT;
-               object->size = size;
+               object->vo_shadow_offset = (vm_object_offset_t)phys_page << PAGE_SHIFT;
+               object->vo_size = size;
        }
        vm_object_unlock(object);
        return KERN_SUCCESS;
@@ -6585,8 +6826,8 @@ vm_object_transpose(
        vm_object_paging_only_wait(object2, THREAD_UNINT);
 
 
-       if (object1->size != object2->size ||
-           object1->size != transpose_size) {
+       if (object1->vo_size != object2->vo_size ||
+           object1->vo_size != transpose_size) {
                /*
                 * If the 2 objects don't have the same size, we can't
                 * exchange their backing stores or one would overflow.
@@ -6661,7 +6902,7 @@ MACRO_END
 
        /* "Lock" refers to the object not its contents */
        /* "size" should be identical */
-       assert(object1->size == object2->size);
+       assert(object1->vo_size == object2->vo_size);
        /* "memq_hint" was updated above when transposing pages */
        /* "ref_count" refers to the object not its contents */
 #if TASK_SWAPPER
@@ -6676,7 +6917,7 @@ MACRO_END
        /* there should be no "shadow" */
        assert(!object1->shadow);
        assert(!object2->shadow);
-       __TRANSPOSE_FIELD(shadow_offset); /* used by phys_contiguous objects */
+       __TRANSPOSE_FIELD(vo_shadow_offset); /* used by phys_contiguous objects */
        __TRANSPOSE_FIELD(pager);
        __TRANSPOSE_FIELD(paging_offset);
        __TRANSPOSE_FIELD(pager_control);
@@ -6735,6 +6976,7 @@ MACRO_END
        __TRANSPOSE_FIELD(sequential);
        __TRANSPOSE_FIELD(pages_created);
        __TRANSPOSE_FIELD(pages_used);
+       __TRANSPOSE_FIELD(scan_collisions);
 #if MACH_PAGEMAP
        __TRANSPOSE_FIELD(existence_map);
 #endif
@@ -6743,6 +6985,7 @@ MACRO_END
        __TRANSPOSE_FIELD(paging_object);
 #endif
        __TRANSPOSE_FIELD(wimg_bits);
+       __TRANSPOSE_FIELD(set_cache_attr);
        __TRANSPOSE_FIELD(code_signed);
        if (object1->hashed) {
                hash_lck = vm_object_hash_lock_spin(object2->pager);
@@ -6825,14 +7068,14 @@ done:
  *
  */
 extern int speculative_reads_disabled;
+extern int ignore_is_ssd;
+
 #if CONFIG_EMBEDDED
 unsigned int preheat_pages_max = MAX_UPL_TRANSFER;
 unsigned int preheat_pages_min = 8;
-unsigned int preheat_pages_mult = 4;
 #else
 unsigned int preheat_pages_max = MAX_UPL_TRANSFER;
 unsigned int preheat_pages_min = 8;
-unsigned int preheat_pages_mult = 4;
 #endif
 
 uint32_t pre_heat_scaling[MAX_UPL_TRANSFER + 1];
@@ -6855,26 +7098,21 @@ vm_object_cluster_size(vm_object_t object, vm_object_offset_t *start,
        vm_behavior_t           behavior;
        boolean_t               look_behind = TRUE;
        boolean_t               look_ahead  = TRUE;
+       boolean_t               isSSD = FALSE;
        uint32_t                throttle_limit;
        int                     sequential_run;
        int                     sequential_behavior = VM_BEHAVIOR_SEQUENTIAL;
        unsigned int            max_ph_size;
        unsigned int            min_ph_size;
-       unsigned int            ph_mult;
+       unsigned int            min_ph_size_in_bytes;
 
        assert( !(*length & PAGE_MASK));
        assert( !(*start & PAGE_MASK_64));
 
-       if ( (ph_mult = preheat_pages_mult) < 1 ) 
-               ph_mult = 1;
-       if ( (min_ph_size = preheat_pages_min) < 1 ) 
-               min_ph_size = 1;
-       if ( (max_ph_size = preheat_pages_max) > MAX_UPL_TRANSFER ) 
-               max_ph_size = MAX_UPL_TRANSFER;
-       
-       if ( (max_length = *length) > (max_ph_size * PAGE_SIZE) ) 
-               max_length = (max_ph_size * PAGE_SIZE);
-
+       /*
+        * remember maxiumum length of run requested
+        */
+       max_length = *length;
        /*
         * we'll always return a cluster size of at least
         * 1 page, since the original fault must always
@@ -6883,7 +7121,7 @@ vm_object_cluster_size(vm_object_t object, vm_object_offset_t *start,
        *length = PAGE_SIZE;
        *io_streaming = 0;
 
-       if (speculative_reads_disabled || fault_info == NULL || max_length == 0) {
+       if (speculative_reads_disabled || fault_info == NULL) {
                /*
                 * no cluster... just fault the page in
                 */
@@ -6896,12 +7134,39 @@ vm_object_cluster_size(vm_object_t object, vm_object_offset_t *start,
 
        vm_object_lock(object);
 
+       if (object->pager == MEMORY_OBJECT_NULL)
+               goto out;       /* pager is gone for this object, nothing more to do */
+
+       if (!ignore_is_ssd)
+               vnode_pager_get_isSSD(object->pager, &isSSD);
+
+       min_ph_size = preheat_pages_min;
+       max_ph_size = preheat_pages_max;
+
+       if (isSSD) {
+               min_ph_size /= 2;
+               max_ph_size /= 8;
+       }
+       if (min_ph_size < 1)
+               min_ph_size = 1;
+
+       if (max_ph_size < 1)
+               max_ph_size = 1;
+       else if (max_ph_size > MAX_UPL_TRANSFER)
+               max_ph_size = MAX_UPL_TRANSFER;
+
+       if (max_length > (max_ph_size * PAGE_SIZE)) 
+               max_length = max_ph_size * PAGE_SIZE;
+
+       if (max_length <= PAGE_SIZE)
+               goto out;
+
+       min_ph_size_in_bytes = min_ph_size * PAGE_SIZE;
+
        if (object->internal)
-               object_size = object->size;
-       else if (object->pager != MEMORY_OBJECT_NULL)
-               vnode_pager_get_object_size(object->pager, &object_size);
+               object_size = object->vo_size;
        else
-               goto out;       /* pager is gone for this object, nothing more to do */
+               vnode_pager_get_object_size(object->pager, &object_size);
 
        object_size = round_page_64(object_size);
 
@@ -6929,7 +7194,7 @@ vm_object_cluster_size(vm_object_t object, vm_object_offset_t *start,
                  }
 
        }
-       switch(behavior) {
+       switch (behavior) {
 
        default:
                behavior = VM_BEHAVIOR_DEFAULT;
@@ -6949,25 +7214,25 @@ vm_object_cluster_size(vm_object_t object, vm_object_offset_t *start,
                        *io_streaming = 1;
                } else {
 
-                       if (object->pages_created < 32 * ph_mult) {
+                       if (object->pages_created < (20 * min_ph_size)) {
                                /*
                                 * prime the pump
                                 */
-                               pre_heat_size = PAGE_SIZE * 8 * ph_mult;
-                               break;
+                               pre_heat_size = min_ph_size_in_bytes;
+                       } else {
+                               /*
+                                * Linear growth in PH size: The maximum size is max_length...
+                                * this cacluation will result in a size that is neither a 
+                                * power of 2 nor a multiple of PAGE_SIZE... so round
+                                * it up to the nearest PAGE_SIZE boundary
+                                */
+                               pre_heat_size = (max_length * object->pages_used) / object->pages_created;
+                               
+                               if (pre_heat_size < min_ph_size_in_bytes)
+                                       pre_heat_size = min_ph_size_in_bytes;
+                               else
+                                       pre_heat_size = round_page(pre_heat_size);
                        }
-                       /*
-                        * Linear growth in PH size: The maximum size is max_length...
-                        * this cacluation will result in a size that is neither a 
-                        * power of 2 nor a multiple of PAGE_SIZE... so round
-                        * it up to the nearest PAGE_SIZE boundary
-                        */
-                       pre_heat_size = (ph_mult * (max_length * object->pages_used) / object->pages_created);
-                       
-                       if (pre_heat_size < PAGE_SIZE * min_ph_size)
-                               pre_heat_size = PAGE_SIZE * min_ph_size;
-                       else
-                               pre_heat_size = round_page(pre_heat_size);
                }
                break;
 
@@ -7003,14 +7268,14 @@ vm_object_cluster_size(vm_object_t object, vm_object_offset_t *start,
        if (pre_heat_size > max_length)
                pre_heat_size = max_length;
 
-       if (behavior == VM_BEHAVIOR_DEFAULT) {
+       if (behavior == VM_BEHAVIOR_DEFAULT && (pre_heat_size > min_ph_size_in_bytes)) {
                if (vm_page_free_count < vm_page_throttle_limit)
-                       pre_heat_size = trunc_page(pre_heat_size / 8);
+                       pre_heat_size = trunc_page(pre_heat_size / 16);
                else if (vm_page_free_count < vm_page_free_target)
-                       pre_heat_size = trunc_page(pre_heat_size / 2);
+                       pre_heat_size = trunc_page(pre_heat_size / 4);
 
-               if (pre_heat_size <= PAGE_SIZE)
-                       goto out;
+               if (pre_heat_size < min_ph_size_in_bytes)
+                       pre_heat_size = min_ph_size_in_bytes;
        }
        if (look_ahead == TRUE) {
                if (look_behind == TRUE) { 
@@ -7047,8 +7312,14 @@ vm_object_cluster_size(vm_object_t object, vm_object_offset_t *start,
                assert((vm_size_t)(orig_start - target_start) == (orig_start - target_start));
                tail_size = pre_heat_size - (vm_size_t)(orig_start - target_start) - PAGE_SIZE;
        } else {
-               if (pre_heat_size > target_start)
-                       pre_heat_size = (vm_size_t) target_start; /* XXX: 32-bit vs 64-bit ? Joe ? */
+               if (pre_heat_size > target_start) {
+                       /*
+                        * since pre_heat_size is always smaller then 2^32,
+                        * if it is larger then target_start (a 64 bit value)
+                        * it is safe to clip target_start to 32 bits
+                        */
+                       pre_heat_size = (vm_size_t) target_start;
+               }
                tail_size = 0;
        }
        assert( !(target_start & PAGE_MASK_64));
@@ -7159,7 +7430,7 @@ vm_object_page_op(
                if(object->phys_contiguous) {
                        if (phys_entry) {
                                *phys_entry = (ppnum_t)
-                                       (object->shadow_offset >> PAGE_SHIFT);
+                                       (object->vo_shadow_offset >> PAGE_SHIFT);
                        }
                        vm_object_unlock(object);
                        return KERN_SUCCESS;
@@ -7340,8 +7611,13 @@ vm_object_range_op(
                dst_page = vm_page_lookup(object, offset);
                if (dst_page != VM_PAGE_NULL) {
                        if (ops & UPL_ROP_DUMP) {
-                               if (dst_page->busy || dst_page->cleaning) {
-                                       /*
+                               if (dst_page->list_req_pending) {
+                                       /*
+                                        * This page isn't on a UPL yet.
+                                        * So it's safe to steal it here and dump it.
+                                        */
+                               } else if (dst_page->busy || dst_page->cleaning) {
+                                       /*
                                         * someone else is playing with the 
                                         * page, we will have to wait
                                         */
@@ -7413,12 +7689,15 @@ _vm_object_lock_try(vm_object_t object)
 boolean_t
 vm_object_lock_try(vm_object_t object)
 {
-    // called from hibernate path so check before blocking
-       if (vm_object_lock_avoid(object) && ml_get_interrupts_enabled()) {
+       /*
+        * Called from hibernate path so check before blocking.
+        */
+       if (vm_object_lock_avoid(object) && ml_get_interrupts_enabled() && get_preemption_level()==0) {
                mutex_pause(2);
        }
        return _vm_object_lock_try(object);
 }
+
 void
 vm_object_lock_shared(vm_object_t object)
 {
@@ -7436,3 +7715,264 @@ vm_object_lock_try_shared(vm_object_t object)
        }
        return (lck_rw_try_lock_shared(&object->Lock));
 }
+
+
+unsigned int vm_object_change_wimg_mode_count = 0;
+
+/*
+ * The object must be locked
+ */
+void
+vm_object_change_wimg_mode(vm_object_t object, unsigned int wimg_mode)
+{
+       vm_page_t p;
+
+       vm_object_lock_assert_exclusive(object);
+
+       vm_object_paging_wait(object, THREAD_UNINT);
+
+       queue_iterate(&object->memq, p, vm_page_t, listq) {
+
+               if (!p->fictitious)
+                       pmap_set_cache_attributes(p->phys_page, wimg_mode);
+       }
+       if (wimg_mode == VM_WIMG_USE_DEFAULT)
+               object->set_cache_attr = FALSE;
+       else
+               object->set_cache_attr = TRUE;
+
+       object->wimg_bits = wimg_mode;
+
+       vm_object_change_wimg_mode_count++;
+}
+
+#if CONFIG_FREEZE
+
+__private_extern__ void        default_freezer_pack_page(vm_page_t , vm_object_t , vm_object_offset_t, void**);
+__private_extern__ void        default_freezer_unpack(vm_object_t , void**);
+
+kern_return_t vm_object_pack(
+       unsigned int       *purgeable_count,
+       unsigned int       *wired_count,
+       unsigned int       *clean_count,
+       unsigned int       *dirty_count,
+       boolean_t          *shared,
+       vm_object_t         src_object,
+       vm_object_t         compact_object,
+       void                  **table,
+       vm_object_offset_t *offset)
+{
+       kern_return_t   kr = KERN_SUCCESS;
+       
+       vm_object_lock(src_object);
+
+       *purgeable_count = *wired_count = *clean_count = *dirty_count = 0;
+       *shared = FALSE;
+
+       if (!src_object->alive || src_object->terminating){
+               kr = KERN_FAILURE;
+               goto done;
+       }
+
+       if (src_object->purgable == VM_PURGABLE_VOLATILE) {
+               *purgeable_count = src_object->resident_page_count;
+               
+               /* If the destination object is null, we're just walking the pages to discover how many can be hibernated */
+               if (VM_OBJECT_NULL != compact_object) {
+                       purgeable_q_t queue;
+                       /* object should be on a queue */
+                       assert(src_object->objq.next != NULL &&
+                              src_object->objq.prev != NULL);
+                       queue = vm_purgeable_object_remove(src_object);
+                       assert(queue);
+                       vm_page_lock_queues();
+                       vm_purgeable_token_delete_first(queue);
+                       vm_page_unlock_queues();
+                       vm_object_purge(src_object);
+               }
+               goto done;
+       }
+
+       if (src_object->ref_count == 1) {
+               vm_object_pack_pages(wired_count, clean_count, dirty_count, src_object, compact_object, table, offset);
+       } else {
+               if (src_object->internal) {
+                       *shared = TRUE;
+               }
+       }
+done:
+       vm_object_unlock(src_object);
+       
+       return kr;
+}
+
+
+void
+vm_object_pack_pages(
+       unsigned int       *wired_count,
+       unsigned int       *clean_count,
+       unsigned int       *dirty_count,
+       vm_object_t         src_object,
+       vm_object_t         compact_object,
+       void                  **table,
+       vm_object_offset_t *offset)
+{
+       vm_page_t p, next;
+
+       next = (vm_page_t)queue_first(&src_object->memq);
+
+       /* Since this function is dual purpose in order that we can count
+        * the freezable pages as well as prepare them, assert that our 
+        * arguments are sane. Gnarly, but avoids code duplication. 
+        */
+       if (VM_OBJECT_NULL == compact_object){
+               assert(!table);
+               assert(!offset);
+       } else {
+               assert(table);
+               assert(offset);
+       }
+
+       while (!queue_end(&src_object->memq, (queue_entry_t)next)) {
+               p = next;
+               next = (vm_page_t)queue_next(&next->listq);
+               
+               if (p->fictitious || p->busy ) 
+                       continue;
+               
+               if (p->absent || p->unusual || p->error)
+                       continue;
+               
+               if (VM_PAGE_WIRED(p)) {
+                       (*wired_count)++;
+                       continue;
+               }
+               
+               if (VM_OBJECT_NULL == compact_object) {
+                       if (p->dirty || pmap_is_modified(p->phys_page)) {
+                               (*dirty_count)++;
+                       } else {
+                               (*clean_count)++;                               
+                       }
+                       continue;
+               }
+               
+               if (p->cleaning) {
+                       p->busy = TRUE;
+                       p->pageout = TRUE;
+                       p->dump_cleaning = TRUE;
+
+                       vm_page_lockspin_queues();
+                       vm_page_wire(p);
+                       vm_page_unlock_queues();
+
+                       continue;
+               }
+
+               if (p->pmapped == TRUE) {
+                       int refmod_state;
+                       refmod_state = pmap_disconnect(p->phys_page);
+                       if (refmod_state & VM_MEM_MODIFIED) {
+                               p->dirty = TRUE;
+                       }
+               }
+               
+               if (p->dirty) {
+                       p->busy = TRUE;
+               
+                       default_freezer_pack_page(p, compact_object, *offset, table);   
+                       *offset += PAGE_SIZE;
+
+                       (*dirty_count)++;
+               }
+               else {
+                       VM_PAGE_FREE(p);
+                       (*clean_count)++;
+               }
+       }
+}
+
+void
+vm_object_pageout(
+       vm_object_t object)
+{
+       vm_page_t p, next;
+       
+       assert(object != VM_OBJECT_NULL );
+       
+       vm_object_lock(object);
+       
+       next = (vm_page_t)queue_first(&object->memq);
+
+       while (!queue_end(&object->memq, (queue_entry_t)next)) {
+               p = next;
+               next = (vm_page_t)queue_next(&next->listq);
+               
+               /* Throw to the pageout queue */
+               vm_page_lockspin_queues();
+
+               VM_PAGE_QUEUES_REMOVE(p);
+               vm_pageout_cluster(p);
+
+               vm_page_unlock_queues();
+       }
+
+       vm_object_unlock(object);
+}
+
+kern_return_t
+vm_object_pagein(
+       vm_object_t object)
+{
+       memory_object_t pager;
+       kern_return_t   kr;
+
+       vm_object_lock(object);
+
+       pager = object->pager;
+
+       if (!object->pager_ready || pager == MEMORY_OBJECT_NULL) {
+               vm_object_unlock(object);
+               return KERN_FAILURE;
+       }
+       
+       vm_object_paging_wait(object, THREAD_UNINT);
+       vm_object_paging_begin(object);
+
+       object->blocked_access = TRUE;
+       vm_object_unlock(object);
+       
+       kr = memory_object_data_reclaim(pager, TRUE);
+
+       vm_object_lock(object);
+
+       object->blocked_access = FALSE;
+       vm_object_paging_end(object);
+
+       vm_object_unlock(object);
+       
+       return kr;
+}
+
+void
+vm_object_unpack(
+       vm_object_t compact_object,
+       void    **table)
+{
+       /*
+        * Future Work:
+        * Right now we treat the default freezer much like
+        * the default pager with respect to when it is
+        * created and terminated.
+        * But, in the future, we may want to terminate the
+        * default freezer at the very instant that an object
+        * has been completely re-filled with all it's previously
+        * paged-out pages.
+        * At that time we'll need to reset the object fields like
+        * "pager" and the associated "pager_{created,initialized,trusted}"
+        * fields right here.
+        */
+       default_freezer_unpack(compact_object, table);
+}
+
+#endif /* CONFIG_FREEZE */