386BSD 0.1 development
authorWilliam F. Jolitz <wjolitz@soda.berkeley.edu>
Mon, 11 May 1992 00:56:53 +0000 (16:56 -0800)
committerWilliam F. Jolitz <wjolitz@soda.berkeley.edu>
Mon, 11 May 1992 00:56:53 +0000 (16:56 -0800)
Work on file usr/src/sys.386bsd/vm/vm_page.c

Co-Authored-By: Lynne Greer Jolitz <ljolitz@cardio.ucsf.edu>
Synthesized-from: 386BSD-0.1

usr/src/sys.386bsd/vm/vm_page.c [new file with mode: 0644]

diff --git a/usr/src/sys.386bsd/vm/vm_page.c b/usr/src/sys.386bsd/vm/vm_page.c
new file mode 100644 (file)
index 0000000..0e794db
--- /dev/null
@@ -0,0 +1,766 @@
+/* 
+ * Copyright (c) 1991 Regents of the University of California.
+ * All rights reserved.
+ *
+ * This code is derived from software contributed to Berkeley by
+ * The Mach Operating System project at Carnegie-Mellon University.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. All advertising materials mentioning features or use of this software
+ *    must display the following acknowledgement:
+ *     This product includes software developed by the University of
+ *     California, Berkeley and its contributors.
+ * 4. Neither the name of the University nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ *     @(#)vm_page.c   7.4 (Berkeley) 5/7/91
+ *
+ *
+ * Copyright (c) 1987, 1990 Carnegie-Mellon University.
+ * All rights reserved.
+ *
+ * Authors: Avadis Tevanian, Jr., Michael Wayne Young
+ * 
+ * Permission to use, copy, modify and distribute this software and
+ * its documentation is hereby granted, provided that both the copyright
+ * notice and this permission notice appear in all copies of the
+ * software, derivative works or modified versions, and any portions
+ * thereof, and that both notices appear in supporting documentation.
+ * 
+ * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" 
+ * CONDITION.  CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND 
+ * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
+ * 
+ * Carnegie Mellon requests users of this software to return to
+ *
+ *  Software Distribution Coordinator  or  Software.Distribution@CS.CMU.EDU
+ *  School of Computer Science
+ *  Carnegie Mellon University
+ *  Pittsburgh PA 15213-3890
+ *
+ * any improvements or extensions that they make and grant Carnegie the
+ * rights to redistribute these changes.
+ */
+
+/*
+ *     Resident memory management module.
+ */
+
+#include "param.h"
+
+#include "vm.h"
+#include "vm_map.h"
+#include "vm_page.h"
+#include "vm_pageout.h"
+
+/*
+ *     Associated with page of user-allocatable memory is a
+ *     page structure.
+ */
+
+queue_head_t   *vm_page_buckets;               /* Array of buckets */
+int            vm_page_bucket_count = 0;       /* How big is array? */
+int            vm_page_hash_mask;              /* Mask for hash function */
+simple_lock_data_t     bucket_lock;            /* lock for all buckets XXX */
+
+vm_size_t      page_size  = 4096;
+vm_size_t      page_mask  = 4095;
+int            page_shift = 12;
+
+queue_head_t   vm_page_queue_free;
+queue_head_t   vm_page_queue_active;
+queue_head_t   vm_page_queue_inactive;
+simple_lock_data_t     vm_page_queue_lock;
+simple_lock_data_t     vm_page_queue_free_lock;
+
+vm_page_t      vm_page_array;
+long           first_page;
+long           last_page;
+vm_offset_t    first_phys_addr;
+vm_offset_t    last_phys_addr;
+
+int    vm_page_free_count;
+int    vm_page_active_count;
+int    vm_page_inactive_count;
+int    vm_page_wire_count;
+int    vm_page_laundry_count;
+
+int    vm_page_free_target = 0;
+int    vm_page_free_min = 0;
+int    vm_page_inactive_target = 0;
+int    vm_page_free_reserved = 0;
+
+/*
+ *     vm_set_page_size:
+ *
+ *     Sets the page size, perhaps based upon the memory
+ *     size.  Must be called before any use of page-size
+ *     dependent functions.
+ *
+ *     Sets page_shift and page_mask from page_size.
+ */
+void vm_set_page_size()
+{
+       page_mask = page_size - 1;
+
+       if ((page_mask & page_size) != 0)
+               panic("vm_set_page_size: page size not a power of two");
+
+       for (page_shift = 0; ; page_shift++)
+               if ((1 << page_shift) == page_size)
+                       break;
+}
+
+
+/*
+ *     vm_page_startup:
+ *
+ *     Initializes the resident memory module.
+ *
+ *     Allocates memory for the page cells, and
+ *     for the object/offset-to-page hash table headers.
+ *     Each page cell is initialized and placed on the free list.
+ */
+vm_offset_t vm_page_startup(start, end, vaddr)
+       register vm_offset_t    start;
+       vm_offset_t     end;
+       register vm_offset_t    vaddr;
+{
+       register vm_offset_t    mapped;
+       register vm_page_t      m;
+       register queue_t        bucket;
+       vm_size_t               npages;
+       register vm_offset_t    new_start;
+       int                     i;
+       vm_offset_t             pa;
+
+       extern  vm_offset_t     kentry_data;
+       extern  vm_size_t       kentry_data_size;
+
+
+       /*
+        *      Initialize the locks
+        */
+
+       simple_lock_init(&vm_page_queue_free_lock);
+       simple_lock_init(&vm_page_queue_lock);
+
+       /*
+        *      Initialize the queue headers for the free queue,
+        *      the active queue and the inactive queue.
+        */
+
+       queue_init(&vm_page_queue_free);
+       queue_init(&vm_page_queue_active);
+       queue_init(&vm_page_queue_inactive);
+
+       /*
+        *      Allocate (and initialize) the hash table buckets.
+        *
+        *      The number of buckets MUST BE a power of 2, and
+        *      the actual value is the next power of 2 greater
+        *      than the number of physical pages in the system.
+        *
+        *      Note:
+        *              This computation can be tweaked if desired.
+        */
+
+       vm_page_buckets = (queue_t) vaddr;
+       bucket = vm_page_buckets;
+       if (vm_page_bucket_count == 0) {
+               vm_page_bucket_count = 1;
+               while (vm_page_bucket_count < atop(end - start))
+                       vm_page_bucket_count <<= 1;
+       }
+
+       vm_page_hash_mask = vm_page_bucket_count - 1;
+
+       /*
+        *      Validate these addresses.
+        */
+
+       new_start = round_page(((queue_t)start) + vm_page_bucket_count);
+       mapped = vaddr;
+       vaddr = pmap_map(mapped, start, new_start,
+                       VM_PROT_READ|VM_PROT_WRITE);
+       start = new_start;
+       bzero((caddr_t) mapped, vaddr - mapped);
+       mapped = vaddr;
+
+       for (i = vm_page_bucket_count; i--;) {
+               queue_init(bucket);
+               bucket++;
+       }
+
+       simple_lock_init(&bucket_lock);
+
+       /*
+        *      round (or truncate) the addresses to our page size.
+        */
+
+       end = trunc_page(end);
+
+       /*
+        *      Pre-allocate maps and map entries that cannot be dynamically
+        *      allocated via malloc().  The maps include the kernel_map and
+        *      kmem_map which must be initialized before malloc() will
+        *      work (obviously).  Also could include pager maps which would
+        *      be allocated before kmeminit.
+        *
+        *      Allow some kernel map entries... this should be plenty
+        *      since people shouldn't be cluttering up the kernel
+        *      map (they should use their own maps).
+        */
+
+       kentry_data_size = MAX_KMAP * sizeof(struct vm_map) +
+                          MAX_KMAPENT * sizeof(struct vm_map_entry);
+       kentry_data_size = round_page(kentry_data_size);
+       kentry_data = (vm_offset_t) vaddr;
+       vaddr += kentry_data_size;
+
+       /*
+        *      Validate these zone addresses.
+        */
+
+       new_start = start + (vaddr - mapped);
+       pmap_map(mapped, start, new_start, VM_PROT_READ|VM_PROT_WRITE);
+       bzero((caddr_t) mapped, (vaddr - mapped));
+       mapped = vaddr;
+       start = new_start;
+
+       /*
+        *      Compute the number of pages of memory that will be
+        *      available for use (taking into account the overhead
+        *      of a page structure per page).
+        */
+
+       vm_page_free_count = npages =
+               (end - start)/(PAGE_SIZE + sizeof(struct vm_page));
+
+       /*
+        *      Initialize the mem entry structures now, and
+        *      put them in the free queue.
+        */
+
+       m = vm_page_array = (vm_page_t) vaddr;
+       first_page = start;
+       first_page += npages*sizeof(struct vm_page);
+       first_page = atop(round_page(first_page));
+       last_page  = first_page + npages - 1;
+
+       first_phys_addr = ptoa(first_page);
+       last_phys_addr  = ptoa(last_page) + page_mask;
+
+       /*
+        *      Validate these addresses.
+        */
+
+       new_start = start + (round_page(m + npages) - mapped);
+       mapped = pmap_map(mapped, start, new_start,
+                       VM_PROT_READ|VM_PROT_WRITE);
+       start = new_start;
+
+       /*
+        *      Clear all of the page structures
+        */
+       bzero((caddr_t)m, npages * sizeof(*m));
+
+       pa = first_phys_addr;
+       while (npages--) {
+               m->copy_on_write = FALSE;
+               m->wanted = FALSE;
+               m->inactive = FALSE;
+               m->active = FALSE;
+               m->busy = FALSE;
+               m->object = NULL;
+               m->phys_addr = pa;
+               queue_enter(&vm_page_queue_free, m, vm_page_t, pageq);
+               m++;
+               pa += PAGE_SIZE;
+       }
+
+       /*
+        *      Initialize vm_pages_needed lock here - don't wait for pageout
+        *      daemon  XXX
+        */
+       simple_lock_init(&vm_pages_needed_lock);
+
+       return(mapped);
+}
+
+/*
+ *     vm_page_hash:
+ *
+ *     Distributes the object/offset key pair among hash buckets.
+ *
+ *     NOTE:  This macro depends on vm_page_bucket_count being a power of 2.
+ */
+#define vm_page_hash(object, offset) \
+       (((unsigned)object+(unsigned)atop(offset))&vm_page_hash_mask)
+
+/*
+ *     vm_page_insert:         [ internal use only ]
+ *
+ *     Inserts the given mem entry into the object/object-page
+ *     table and object list.
+ *
+ *     The object and page must be locked.
+ */
+
+void vm_page_insert(mem, object, offset)
+       register vm_page_t      mem;
+       register vm_object_t    object;
+       register vm_offset_t    offset;
+{
+       register queue_t        bucket;
+       int                     spl;
+
+       VM_PAGE_CHECK(mem);
+
+       if (mem->tabled)
+               panic("vm_page_insert: already inserted");
+
+       /*
+        *      Record the object/offset pair in this page
+        */
+
+       mem->object = object;
+       mem->offset = offset;
+
+       /*
+        *      Insert it into the object_object/offset hash table
+        */
+
+       bucket = &vm_page_buckets[vm_page_hash(object, offset)];
+       spl = splimp();
+       simple_lock(&bucket_lock);
+       queue_enter(bucket, mem, vm_page_t, hashq);
+       simple_unlock(&bucket_lock);
+       (void) splx(spl);
+
+       /*
+        *      Now link into the object's list of backed pages.
+        */
+
+       queue_enter(&object->memq, mem, vm_page_t, listq);
+       mem->tabled = TRUE;
+
+       /*
+        *      And show that the object has one more resident
+        *      page.
+        */
+
+       object->resident_page_count++;
+}
+
+/*
+ *     vm_page_remove:         [ internal use only ]
+ *
+ *     Removes the given mem entry from the object/offset-page
+ *     table and the object page list.
+ *
+ *     The object and page must be locked.
+ */
+
+void vm_page_remove(mem)
+       register vm_page_t      mem;
+{
+       register queue_t        bucket;
+       int                     spl;
+
+       VM_PAGE_CHECK(mem);
+
+       if (!mem->tabled)
+               return;
+
+       /*
+        *      Remove from the object_object/offset hash table
+        */
+
+       bucket = &vm_page_buckets[vm_page_hash(mem->object, mem->offset)];
+       spl = splimp();
+       simple_lock(&bucket_lock);
+       queue_remove(bucket, mem, vm_page_t, hashq);
+       simple_unlock(&bucket_lock);
+       (void) splx(spl);
+
+       /*
+        *      Now remove from the object's list of backed pages.
+        */
+
+       queue_remove(&mem->object->memq, mem, vm_page_t, listq);
+
+       /*
+        *      And show that the object has one fewer resident
+        *      page.
+        */
+
+       mem->object->resident_page_count--;
+
+       mem->tabled = FALSE;
+}
+
+/*
+ *     vm_page_lookup:
+ *
+ *     Returns the page associated with the object/offset
+ *     pair specified; if none is found, NULL is returned.
+ *
+ *     The object must be locked.  No side effects.
+ */
+
+vm_page_t vm_page_lookup(object, offset)
+       register vm_object_t    object;
+       register vm_offset_t    offset;
+{
+       register vm_page_t      mem;
+       register queue_t        bucket;
+       int                     spl;
+
+       /*
+        *      Search the hash table for this object/offset pair
+        */
+
+       bucket = &vm_page_buckets[vm_page_hash(object, offset)];
+
+       spl = splimp();
+       simple_lock(&bucket_lock);
+       mem = (vm_page_t) queue_first(bucket);
+       while (!queue_end(bucket, (queue_entry_t) mem)) {
+               VM_PAGE_CHECK(mem);
+               if ((mem->object == object) && (mem->offset == offset)) {
+                       simple_unlock(&bucket_lock);
+                       splx(spl);
+                       return(mem);
+               }
+               mem = (vm_page_t) queue_next(&mem->hashq);
+       }
+
+       simple_unlock(&bucket_lock);
+       splx(spl);
+       return(NULL);
+}
+
+/*
+ *     vm_page_rename:
+ *
+ *     Move the given memory entry from its
+ *     current object to the specified target object/offset.
+ *
+ *     The object must be locked.
+ */
+void vm_page_rename(mem, new_object, new_offset)
+       register vm_page_t      mem;
+       register vm_object_t    new_object;
+       vm_offset_t             new_offset;
+{
+       if (mem->object == new_object)
+               return;
+
+       vm_page_lock_queues();  /* keep page from moving out from
+                                  under pageout daemon */
+       vm_page_remove(mem);
+       vm_page_insert(mem, new_object, new_offset);
+       vm_page_unlock_queues();
+}
+
+void           vm_page_init(mem, object, offset)
+       vm_page_t       mem;
+       vm_object_t     object;
+       vm_offset_t     offset;
+{
+#ifdef DEBUG
+#define        vm_page_init(mem, object, offset)  {\
+               (mem)->busy = TRUE; \
+               (mem)->tabled = FALSE; \
+               vm_page_insert((mem), (object), (offset)); \
+               (mem)->absent = FALSE; \
+               (mem)->fictitious = FALSE; \
+               (mem)->page_lock = VM_PROT_NONE; \
+               (mem)->unlock_request = VM_PROT_NONE; \
+               (mem)->laundry = FALSE; \
+               (mem)->active = FALSE; \
+               (mem)->inactive = FALSE; \
+               (mem)->wire_count = 0; \
+               (mem)->clean = TRUE; \
+               (mem)->copy_on_write = FALSE; \
+               (mem)->fake = TRUE; \
+               (mem)->pagerowned = FALSE; \
+               (mem)->ptpage = FALSE; \
+       }
+#else
+#define        vm_page_init(mem, object, offset)  {\
+               (mem)->busy = TRUE; \
+               (mem)->tabled = FALSE; \
+               vm_page_insert((mem), (object), (offset)); \
+               (mem)->absent = FALSE; \
+               (mem)->fictitious = FALSE; \
+               (mem)->page_lock = VM_PROT_NONE; \
+               (mem)->unlock_request = VM_PROT_NONE; \
+               (mem)->laundry = FALSE; \
+               (mem)->active = FALSE; \
+               (mem)->inactive = FALSE; \
+               (mem)->wire_count = 0; \
+               (mem)->clean = TRUE; \
+               (mem)->copy_on_write = FALSE; \
+               (mem)->fake = TRUE; \
+       }
+#endif
+
+       vm_page_init(mem, object, offset);
+}
+
+/*
+ *     vm_page_alloc:
+ *
+ *     Allocate and return a memory cell associated
+ *     with this VM object/offset pair.
+ *
+ *     Object must be locked.
+ */
+vm_page_t vm_page_alloc(object, offset)
+       vm_object_t     object;
+       vm_offset_t     offset;
+{
+       register vm_page_t      mem;
+       int             spl;
+
+       spl = splimp();                         /* XXX */
+       simple_lock(&vm_page_queue_free_lock);
+       if (queue_empty(&vm_page_queue_free)) {
+               simple_unlock(&vm_page_queue_free_lock);
+               splx(spl);
+               return(NULL);
+       }
+
+       queue_remove_first(&vm_page_queue_free, mem, vm_page_t, pageq);
+
+       vm_page_free_count--;
+       simple_unlock(&vm_page_queue_free_lock);
+       splx(spl);
+
+       vm_page_init(mem, object, offset);
+
+       /*
+        *      Decide if we should poke the pageout daemon.
+        *      We do this if the free count is less than the low
+        *      water mark, or if the free count is less than the high
+        *      water mark (but above the low water mark) and the inactive
+        *      count is less than its target.
+        *
+        *      We don't have the counts locked ... if they change a little,
+        *      it doesn't really matter.
+        */
+
+       if ((vm_page_free_count < vm_page_free_min) ||
+                       ((vm_page_free_count < vm_page_free_target) &&
+                       (vm_page_inactive_count < vm_page_inactive_target)))
+               thread_wakeup(&vm_pages_needed);
+       return(mem);
+}
+
+/*
+ *     vm_page_free:
+ *
+ *     Returns the given page to the free list,
+ *     disassociating it with any VM object.
+ *
+ *     Object and page must be locked prior to entry.
+ */
+void vm_page_free(mem)
+       register vm_page_t      mem;
+{
+       vm_page_remove(mem);
+       if (mem->active) {
+               queue_remove(&vm_page_queue_active, mem, vm_page_t, pageq);
+               mem->active = FALSE;
+               vm_page_active_count--;
+       }
+
+       if (mem->inactive) {
+               queue_remove(&vm_page_queue_inactive, mem, vm_page_t, pageq);
+               mem->inactive = FALSE;
+               vm_page_inactive_count--;
+       }
+
+       if (!mem->fictitious) {
+               int     spl;
+
+               spl = splimp();
+               simple_lock(&vm_page_queue_free_lock);
+               queue_enter(&vm_page_queue_free, mem, vm_page_t, pageq);
+
+               vm_page_free_count++;
+               simple_unlock(&vm_page_queue_free_lock);
+               splx(spl);
+       }
+}
+
+/*
+ *     vm_page_wire:
+ *
+ *     Mark this page as wired down by yet
+ *     another map, removing it from paging queues
+ *     as necessary.
+ *
+ *     The page queues must be locked.
+ */
+void vm_page_wire(mem)
+       register vm_page_t      mem;
+{
+       VM_PAGE_CHECK(mem);
+
+       if (mem->wire_count == 0) {
+               if (mem->active) {
+                       queue_remove(&vm_page_queue_active, mem, vm_page_t,
+                                               pageq);
+                       vm_page_active_count--;
+                       mem->active = FALSE;
+               }
+               if (mem->inactive) {
+                       queue_remove(&vm_page_queue_inactive, mem, vm_page_t,
+                                               pageq);
+                       vm_page_inactive_count--;
+                       mem->inactive = FALSE;
+               }
+               vm_page_wire_count++;
+       }
+       mem->wire_count++;
+}
+
+/*
+ *     vm_page_unwire:
+ *
+ *     Release one wiring of this page, potentially
+ *     enabling it to be paged again.
+ *
+ *     The page queues must be locked.
+ */
+void vm_page_unwire(mem)
+       register vm_page_t      mem;
+{
+       VM_PAGE_CHECK(mem);
+
+       mem->wire_count--;
+       if (mem->wire_count == 0) {
+               queue_enter(&vm_page_queue_active, mem, vm_page_t, pageq);
+               vm_page_active_count++;
+               mem->active = TRUE;
+               vm_page_wire_count--;
+       }
+}
+
+/*
+ *     vm_page_deactivate:
+ *
+ *     Returns the given page to the inactive list,
+ *     indicating that no physical maps have access
+ *     to this page.  [Used by the physical mapping system.]
+ *
+ *     The page queues must be locked.
+ */
+void vm_page_deactivate(m)
+       register vm_page_t      m;
+{
+       VM_PAGE_CHECK(m);
+
+       /*
+        *      Only move active pages -- ignore locked or already
+        *      inactive ones.
+        */
+
+       if (m->active) {
+               pmap_clear_reference(VM_PAGE_TO_PHYS(m));
+               queue_remove(&vm_page_queue_active, m, vm_page_t, pageq);
+               queue_enter(&vm_page_queue_inactive, m, vm_page_t, pageq);
+               m->active = FALSE;
+               m->inactive = TRUE;
+               vm_page_active_count--;
+               vm_page_inactive_count++;
+               if (pmap_is_modified(VM_PAGE_TO_PHYS(m)))
+                       m->clean = FALSE;
+               m->laundry = !m->clean;
+       }
+}
+
+/*
+ *     vm_page_activate:
+ *
+ *     Put the specified page on the active list (if appropriate).
+ *
+ *     The page queues must be locked.
+ */
+
+void vm_page_activate(m)
+       register vm_page_t      m;
+{
+       VM_PAGE_CHECK(m);
+
+       if (m->inactive) {
+               queue_remove(&vm_page_queue_inactive, m, vm_page_t,
+                                               pageq);
+               vm_page_inactive_count--;
+               m->inactive = FALSE;
+       }
+       if (m->wire_count == 0) {
+               if (m->active)
+                       panic("vm_page_activate: already active");
+
+               queue_enter(&vm_page_queue_active, m, vm_page_t, pageq);
+               m->active = TRUE;
+               vm_page_active_count++;
+       }
+}
+
+/*
+ *     vm_page_zero_fill:
+ *
+ *     Zero-fill the specified page.
+ *     Written as a standard pagein routine, to
+ *     be used by the zero-fill object.
+ */
+
+boolean_t vm_page_zero_fill(m)
+       vm_page_t       m;
+{
+       VM_PAGE_CHECK(m);
+
+       pmap_zero_page(VM_PAGE_TO_PHYS(m));
+       return(TRUE);
+}
+
+/*
+ *     vm_page_copy:
+ *
+ *     Copy one page to another
+ */
+
+void vm_page_copy(src_m, dest_m)
+       vm_page_t       src_m;
+       vm_page_t       dest_m;
+{
+       VM_PAGE_CHECK(src_m);
+       VM_PAGE_CHECK(dest_m);
+
+       pmap_copy_page(VM_PAGE_TO_PHYS(src_m), VM_PAGE_TO_PHYS(dest_m));
+}