Checkpoint of a new physical memory backed object type, that does not

have pv_entries.  This is intended for very special circumstances,
eg: a certain database that has a 1GB shm segment mapped into 300
processes.  That would consume 2GB of kvm just to hold the pv_entries
alone.  This would not be used on systems unless the physical ram was
available, as it's not pageable.

This is a work-in-progress, but is a useful and functional checkpoint.
Matt has got some more fixes for it that will be committed soon.

Reviewed by:	dillon

sys/conf/files

@@ -880,6 +880,7 @@ ufs/ufs/ufs_vfsops.c	standard
 ufs/ufs/ufs_vnops.c	standard
 vm/default_pager.c	standard
 vm/device_pager.c	standard
+vm/phys_pager.c		standard
 vm/swap_pager.c		standard
 vm/vm_fault.c		standard
 vm/vm_glue.c		standard

sys/vm/phys_pager.c

@@ -0,0 +1,222 @@
+/*
+ * Copyright (c) 2000 Peter Wemm
+ *
+ * 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.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHORS 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 AUTHORS 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.
+ *
+ * $FreeBSD$
+ */
+
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/linker_set.h>
+#include <sys/conf.h>
+#include <sys/mman.h>
+#include <sys/sysctl.h>
+
+#include <vm/vm.h>
+#include <vm/vm_object.h>
+#include <vm/vm_page.h>
+#include <vm/vm_pager.h>
+#include <vm/vm_zone.h>
+
+static void phys_pager_init __P((void));
+static vm_object_t phys_pager_alloc __P((void *, vm_ooffset_t, vm_prot_t,
+		vm_ooffset_t));
+static void phys_pager_dealloc __P((vm_object_t));
+static int phys_pager_getpages __P((vm_object_t, vm_page_t *, int, int));
+static void phys_pager_putpages __P((vm_object_t, vm_page_t *, int, 
+		boolean_t, int *));
+static boolean_t phys_pager_haspage __P((vm_object_t, vm_pindex_t, int *,
+		int *));
+
+/* list of device pager objects */
+static struct pagerlst phys_pager_object_list;
+
+static int phys_pager_alloc_lock, phys_pager_alloc_lock_want;
+
+struct pagerops physpagerops = {
+	phys_pager_init,
+	phys_pager_alloc,
+	phys_pager_dealloc,
+	phys_pager_getpages,
+	phys_pager_putpages,
+	phys_pager_haspage,
+	NULL
+};
+
+static void
+phys_pager_init()
+{
+	TAILQ_INIT(&phys_pager_object_list);
+}
+
+static vm_object_t
+phys_pager_alloc(void *handle, vm_ooffset_t size, vm_prot_t prot,
+		 vm_ooffset_t foff)
+{
+	vm_object_t object;
+
+	/*
+	 * Offset should be page aligned.
+	 */
+	if (foff & PAGE_MASK)
+		return (NULL);
+
+	size = round_page(size);
+
+	/*
+	 * Lock to prevent object creation race condition.
+	 */
+	while (phys_pager_alloc_lock) {
+		phys_pager_alloc_lock_want++;
+		tsleep(&phys_pager_alloc_lock, PVM, "ppall", 0);
+		phys_pager_alloc_lock_want--;
+	}
+	phys_pager_alloc_lock = 1;
+
+	/*
+	 * Look up pager, creating as necessary.
+	 */
+	object = vm_pager_object_lookup(&phys_pager_object_list, handle);
+	if (object == NULL) {
+		/*
+		 * Allocate object and associate it with the pager.
+		 */
+		object = vm_object_allocate(OBJT_PHYS,
+			OFF_TO_IDX(foff + size));
+		object->handle = handle;
+		TAILQ_INIT(&object->un_pager.physp.physp_pglist);
+		TAILQ_INSERT_TAIL(&phys_pager_object_list, object,
+		    pager_object_list);
+	} else {
+		/*
+		 * Gain a reference to the object.
+		 */
+		vm_object_reference(object);
+		if (OFF_TO_IDX(foff + size) > object->size)
+			object->size = OFF_TO_IDX(foff + size);
+	}
+
+	phys_pager_alloc_lock = 0;
+	if (phys_pager_alloc_lock_want)
+		wakeup(&phys_pager_alloc_lock);
+
+	return (object);
+}
+
+static void
+phys_pager_dealloc(object)
+	vm_object_t object;
+{
+	vm_page_t m;
+	int s;
+
+	TAILQ_REMOVE(&phys_pager_object_list, object, pager_object_list);
+	/*
+	 * Free up our fake pages.
+	 */
+	s = splvm();
+	while ((m = TAILQ_FIRST(&object->un_pager.physp.physp_pglist)) != 0) {
+		TAILQ_REMOVE(&object->un_pager.physp.physp_pglist, m, pageq);
+		/* return the page back to normal */
+		m->flags &= ~PG_FICTITIOUS;
+		m->dirty = 0;
+		vm_page_unwire(m, 0);
+		vm_page_flag_clear(m, PG_ZERO);
+		vm_page_free(m);
+	}
+	splx(s);
+}
+
+static int
+phys_pager_getpages(object, m, count, reqpage)
+	vm_object_t object;
+	vm_page_t *m;
+	int count;
+	int reqpage;
+{
+	int i, s;
+
+	s = splvm();
+	/*
+	 * Fill as many pages as vm_fault has allocated for us.
+	 */
+	for (i = 0; i < count; i++) {
+		if ((m[i]->flags & PG_ZERO) == 0)
+			vm_page_zero_fill(m[i]);
+		vm_page_flag_set(m[i], PG_ZERO);
+		/* Switch off pv_entries */
+		vm_page_wire(m[i]);
+		vm_page_flag_set(m[i], PG_FICTITIOUS);
+		m[i]->valid = VM_PAGE_BITS_ALL;
+		m[i]->dirty = 0;
+		/* The requested page must remain busy, the others not. */
+		if (reqpage != i) {
+			vm_page_flag_clear(m[i], PG_BUSY);
+			m[i]->busy = 0;
+		}
+		TAILQ_INSERT_TAIL(&object->un_pager.physp.physp_pglist, m[i],
+		    pageq);
+	}
+	splx(s);
+
+	return (VM_PAGER_OK);
+}
+
+static void
+phys_pager_putpages(object, m, count, sync, rtvals)
+	vm_object_t object;
+	vm_page_t *m;
+	int count;
+	boolean_t sync;
+	int *rtvals;
+{
+	panic("phys_pager_putpage called");
+}
+
+/*
+ * Implement a pretty aggressive clustered getpages strategy.  Hint that
+ * everything in an entire 4MB window should be prefaulted at once.
+ *
+ * XXX 4MB (1024 slots per page table page) is convenient for x86,
+ * but may not be for other arches.
+ */
+#ifndef PHYSCLUSTER
+#define PHYSCLUSTER 1024
+#endif
+static boolean_t
+phys_pager_haspage(object, pindex, before, after)
+	vm_object_t object;
+	vm_pindex_t pindex;
+	int *before;
+	int *after;
+{
+	vm_pindex_t base, end;
+
+	base = pindex & (~(PHYSCLUSTER - 1));
+	end = base + (PHYSCLUSTER - 1);
+	if (before != NULL)
+		*before = pindex - base;
+	if (after != NULL)
+		*after = end - pindex;
+	return (TRUE);
+}

sys/vm/vm_mmap.c

@@ -1096,7 +1096,7 @@ vm_mmap(vm_map_t map, vm_offset_t *addr, vm_size_t size, vm_prot_t prot,
 	/*
 	 * Force device mappings to be shared.
 	 */
-	if (type == OBJT_DEVICE) {
+	if (type == OBJT_DEVICE || type == OBJT_PHYS) {
 		flags &= ~(MAP_PRIVATE|MAP_COPY);
 		flags |= MAP_SHARED;
 	}

sys/vm/vm_object.h

@@ -74,7 +74,8 @@
 #include <sys/queue.h>
 #include <machine/atomic.h>
 
-enum obj_type { OBJT_DEFAULT, OBJT_SWAP, OBJT_VNODE, OBJT_DEVICE, OBJT_DEAD };
+enum obj_type { OBJT_DEFAULT, OBJT_SWAP, OBJT_VNODE, OBJT_DEVICE, OBJT_PHYS,
+		OBJT_DEAD };
 typedef u_char objtype_t;
 
 /*
@@ -122,6 +123,15 @@ struct vm_object {
 			TAILQ_HEAD(, vm_page) devp_pglist;
 		} devp;
 
+		/*
+		 * Physmem pager
+		 *
+		 *	physp_pglist - list of allocated pages
+		 */
+		struct {
+			TAILQ_HEAD(, vm_page) physp_pglist;
+		} physp;
+
 		/*
 		 * Swap pager
 		 *

sys/vm/vm_pageout.c

@@ -448,7 +448,7 @@ vm_pageout_object_deactivate_pages(map, object, desired, map_remove_only)
 	int remove_mode;
 	int s;
 
-	if (object->type == OBJT_DEVICE)
+	if (object->type == OBJT_DEVICE || object->type == OBJT_PHYS)
 		return;
 
 	while (object) {

sys/vm/vm_pager.c

@@ -92,6 +92,7 @@ extern struct pagerops defaultpagerops;
 extern struct pagerops swappagerops;
 extern struct pagerops vnodepagerops;
 extern struct pagerops devicepagerops;
+extern struct pagerops physpagerops;
 
 int cluster_pbuf_freecnt = -1;	/* unlimited to begin with */
 
@@ -173,6 +174,7 @@ struct pagerops *pagertab[] = {
 	&swappagerops,		/* OBJT_SWAP */
 	&vnodepagerops,		/* OBJT_VNODE */
 	&devicepagerops,	/* OBJT_DEVICE */
+	&physpagerops,		/* OBJT_PHYS */
 	&deadpagerops		/* OBJT_DEAD */
 };