2005-01-07 02:29:27 +00:00
|
|
|
/*-
|
2000-05-21 13:41:29 +00:00
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
|
2003-06-11 23:50:51 +00:00
|
|
|
#include <sys/cdefs.h>
|
|
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
|
2000-05-21 13:41:29 +00:00
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/systm.h>
|
|
|
|
#include <sys/conf.h>
|
2001-04-18 20:24:16 +00:00
|
|
|
#include <sys/kernel.h>
|
2001-05-01 08:13:21 +00:00
|
|
|
#include <sys/lock.h>
|
2001-07-04 16:20:28 +00:00
|
|
|
#include <sys/proc.h>
|
2001-05-01 08:13:21 +00:00
|
|
|
#include <sys/mutex.h>
|
2000-05-21 13:41:29 +00:00
|
|
|
#include <sys/mman.h>
|
2013-03-09 02:32:23 +00:00
|
|
|
#include <sys/rwlock.h>
|
2000-05-21 13:41:29 +00:00
|
|
|
#include <sys/sysctl.h>
|
|
|
|
|
|
|
|
#include <vm/vm.h>
|
2013-08-09 11:11:11 +00:00
|
|
|
#include <vm/vm_param.h>
|
2000-05-21 13:41:29 +00:00
|
|
|
#include <vm/vm_object.h>
|
|
|
|
#include <vm/vm_page.h>
|
2016-12-08 11:35:53 +00:00
|
|
|
#include <vm/vm_pageout.h>
|
2000-05-21 13:41:29 +00:00
|
|
|
#include <vm/vm_pager.h>
|
|
|
|
|
2007-10-30 14:48:13 +00:00
|
|
|
/* list of phys pager objects */
|
2000-05-21 13:41:29 +00:00
|
|
|
static struct pagerlst phys_pager_object_list;
|
2001-04-18 20:24:16 +00:00
|
|
|
/* protect access to phys_pager_object_list */
|
|
|
|
static struct mtx phys_pager_mtx;
|
2000-05-21 13:41:29 +00:00
|
|
|
|
|
|
|
static void
|
2000-07-28 22:03:08 +00:00
|
|
|
phys_pager_init(void)
|
2000-05-21 13:41:29 +00:00
|
|
|
{
|
2000-07-28 22:03:08 +00:00
|
|
|
|
2000-05-21 13:41:29 +00:00
|
|
|
TAILQ_INIT(&phys_pager_object_list);
|
2002-04-04 21:03:38 +00:00
|
|
|
mtx_init(&phys_pager_mtx, "phys_pager list", NULL, MTX_DEF);
|
2000-05-21 13:41:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static vm_object_t
|
|
|
|
phys_pager_alloc(void *handle, vm_ooffset_t size, vm_prot_t prot,
|
Implement global and per-uid accounting of the anonymous memory. Add
rlimit RLIMIT_SWAP that limits the amount of swap that may be reserved
for the uid.
The accounting information (charge) is associated with either map entry,
or vm object backing the entry, assuming the object is the first one
in the shadow chain and entry does not require COW. Charge is moved
from entry to object on allocation of the object, e.g. during the mmap,
assuming the object is allocated, or on the first page fault on the
entry. It moves back to the entry on forks due to COW setup.
The per-entry granularity of accounting makes the charge process fair
for processes that change uid during lifetime, and decrements charge
for proper uid when region is unmapped.
The interface of vm_pager_allocate(9) is extended by adding struct ucred *,
that is used to charge appropriate uid when allocation if performed by
kernel, e.g. md(4).
Several syscalls, among them is fork(2), may now return ENOMEM when
global or per-uid limits are enforced.
In collaboration with: pho
Reviewed by: alc
Approved by: re (kensmith)
2009-06-23 20:45:22 +00:00
|
|
|
vm_ooffset_t foff, struct ucred *cred)
|
2000-05-21 13:41:29 +00:00
|
|
|
{
|
2007-08-18 16:40:33 +00:00
|
|
|
vm_object_t object, object1;
|
2004-01-04 20:55:15 +00:00
|
|
|
vm_pindex_t pindex;
|
2000-05-21 13:41:29 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Offset should be page aligned.
|
|
|
|
*/
|
|
|
|
if (foff & PAGE_MASK)
|
|
|
|
return (NULL);
|
|
|
|
|
2004-01-04 20:55:15 +00:00
|
|
|
pindex = OFF_TO_IDX(foff + PAGE_MASK + size);
|
2000-05-21 13:41:29 +00:00
|
|
|
|
2000-12-06 21:52:23 +00:00
|
|
|
if (handle != NULL) {
|
2007-08-18 16:40:33 +00:00
|
|
|
mtx_lock(&phys_pager_mtx);
|
2000-05-21 13:41:29 +00:00
|
|
|
/*
|
2000-12-06 21:52:23 +00:00
|
|
|
* Look up pager, creating as necessary.
|
2000-05-21 13:41:29 +00:00
|
|
|
*/
|
2007-08-18 16:40:33 +00:00
|
|
|
object1 = NULL;
|
2000-12-06 21:52:23 +00:00
|
|
|
object = vm_pager_object_lookup(&phys_pager_object_list, handle);
|
|
|
|
if (object == NULL) {
|
|
|
|
/*
|
|
|
|
* Allocate object and associate it with the pager.
|
|
|
|
*/
|
2001-04-18 20:24:16 +00:00
|
|
|
mtx_unlock(&phys_pager_mtx);
|
2007-08-18 16:40:33 +00:00
|
|
|
object1 = vm_object_allocate(OBJT_PHYS, pindex);
|
|
|
|
mtx_lock(&phys_pager_mtx);
|
|
|
|
object = vm_pager_object_lookup(&phys_pager_object_list,
|
|
|
|
handle);
|
|
|
|
if (object != NULL) {
|
|
|
|
/*
|
|
|
|
* We raced with other thread while
|
|
|
|
* allocating object.
|
|
|
|
*/
|
|
|
|
if (pindex > object->size)
|
|
|
|
object->size = pindex;
|
|
|
|
} else {
|
|
|
|
object = object1;
|
|
|
|
object1 = NULL;
|
|
|
|
object->handle = handle;
|
2016-12-08 11:35:53 +00:00
|
|
|
vm_object_set_flag(object, OBJ_POPULATE);
|
2016-12-08 10:28:51 +00:00
|
|
|
TAILQ_INSERT_TAIL(&phys_pager_object_list,
|
|
|
|
object, pager_object_list);
|
2007-08-18 16:40:33 +00:00
|
|
|
}
|
2000-12-06 21:52:23 +00:00
|
|
|
} else {
|
2004-01-04 20:55:15 +00:00
|
|
|
if (pindex > object->size)
|
|
|
|
object->size = pindex;
|
2000-12-06 21:52:23 +00:00
|
|
|
}
|
2007-08-18 16:40:33 +00:00
|
|
|
mtx_unlock(&phys_pager_mtx);
|
|
|
|
vm_object_deallocate(object1);
|
2000-12-06 21:52:23 +00:00
|
|
|
} else {
|
2004-01-04 20:55:15 +00:00
|
|
|
object = vm_object_allocate(OBJT_PHYS, pindex);
|
2016-12-08 11:35:53 +00:00
|
|
|
vm_object_set_flag(object, OBJ_POPULATE);
|
2000-05-21 13:41:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return (object);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
2000-07-28 22:03:08 +00:00
|
|
|
phys_pager_dealloc(vm_object_t object)
|
2000-05-21 13:41:29 +00:00
|
|
|
{
|
|
|
|
|
2001-04-18 20:24:16 +00:00
|
|
|
if (object->handle != NULL) {
|
2013-03-09 02:32:23 +00:00
|
|
|
VM_OBJECT_WUNLOCK(object);
|
2001-04-18 20:24:16 +00:00
|
|
|
mtx_lock(&phys_pager_mtx);
|
2000-12-06 21:52:23 +00:00
|
|
|
TAILQ_REMOVE(&phys_pager_object_list, object, pager_object_list);
|
2001-04-18 20:24:16 +00:00
|
|
|
mtx_unlock(&phys_pager_mtx);
|
2013-03-09 02:32:23 +00:00
|
|
|
VM_OBJECT_WLOCK(object);
|
2001-04-18 20:24:16 +00:00
|
|
|
}
|
2015-05-08 19:43:37 +00:00
|
|
|
object->handle = NULL;
|
|
|
|
object->type = OBJT_DEAD;
|
2000-05-21 13:41:29 +00:00
|
|
|
}
|
|
|
|
|
2004-04-24 23:23:36 +00:00
|
|
|
/*
|
|
|
|
* Fill as many pages as vm_fault has allocated for us.
|
|
|
|
*/
|
2000-05-21 13:41:29 +00:00
|
|
|
static int
|
A change to KPI of vm_pager_get_pages() and underlying VOP_GETPAGES().
o With new KPI consumers can request contiguous ranges of pages, and
unlike before, all pages will be kept busied on return, like it was
done before with the 'reqpage' only. Now the reqpage goes away. With
new interface it is easier to implement code protected from race
conditions.
Such arrayed requests for now should be preceeded by a call to
vm_pager_haspage() to make sure that request is possible. This
could be improved later, making vm_pager_haspage() obsolete.
Strenghtening the promises on the business of the array of pages
allows us to remove such hacks as swp_pager_free_nrpage() and
vm_pager_free_nonreq().
o New KPI accepts two integer pointers that may optionally point at
values for read ahead and read behind, that a pager may do, if it
can. These pages are completely owned by pager, and not controlled
by the caller.
This shifts the UFS-specific readahead logic from vm_fault.c, which
should be file system agnostic, into vnode_pager.c. It also removes
one VOP_BMAP() request per hard fault.
Discussed with: kib, alc, jeff, scottl
Sponsored by: Nginx, Inc.
Sponsored by: Netflix
2015-12-16 21:30:45 +00:00
|
|
|
phys_pager_getpages(vm_object_t object, vm_page_t *m, int count, int *rbehind,
|
|
|
|
int *rahead)
|
2000-05-21 13:41:29 +00:00
|
|
|
{
|
2004-04-24 23:23:36 +00:00
|
|
|
int i;
|
2000-05-21 13:41:29 +00:00
|
|
|
|
2013-03-09 02:32:23 +00:00
|
|
|
VM_OBJECT_ASSERT_WLOCKED(object);
|
2000-05-21 13:41:29 +00:00
|
|
|
for (i = 0; i < count; i++) {
|
2004-04-25 07:58:59 +00:00
|
|
|
if (m[i]->valid == 0) {
|
|
|
|
if ((m[i]->flags & PG_ZERO) == 0)
|
|
|
|
pmap_zero_page(m[i]);
|
|
|
|
m[i]->valid = VM_PAGE_BITS_ALL;
|
2002-12-27 06:09:56 +00:00
|
|
|
}
|
2004-04-25 07:58:59 +00:00
|
|
|
KASSERT(m[i]->valid == VM_PAGE_BITS_ALL,
|
|
|
|
("phys_pager_getpages: partially valid page %p", m[i]));
|
2009-06-13 20:58:12 +00:00
|
|
|
KASSERT(m[i]->dirty == 0,
|
|
|
|
("phys_pager_getpages: dirty page %p", m[i]));
|
2000-05-21 13:41:29 +00:00
|
|
|
}
|
A change to KPI of vm_pager_get_pages() and underlying VOP_GETPAGES().
o With new KPI consumers can request contiguous ranges of pages, and
unlike before, all pages will be kept busied on return, like it was
done before with the 'reqpage' only. Now the reqpage goes away. With
new interface it is easier to implement code protected from race
conditions.
Such arrayed requests for now should be preceeded by a call to
vm_pager_haspage() to make sure that request is possible. This
could be improved later, making vm_pager_haspage() obsolete.
Strenghtening the promises on the business of the array of pages
allows us to remove such hacks as swp_pager_free_nrpage() and
vm_pager_free_nonreq().
o New KPI accepts two integer pointers that may optionally point at
values for read ahead and read behind, that a pager may do, if it
can. These pages are completely owned by pager, and not controlled
by the caller.
This shifts the UFS-specific readahead logic from vm_fault.c, which
should be file system agnostic, into vnode_pager.c. It also removes
one VOP_BMAP() request per hard fault.
Discussed with: kib, alc, jeff, scottl
Sponsored by: Nginx, Inc.
Sponsored by: Netflix
2015-12-16 21:30:45 +00:00
|
|
|
if (rbehind)
|
|
|
|
*rbehind = 0;
|
|
|
|
if (rahead)
|
|
|
|
*rahead = 0;
|
2000-05-21 13:41:29 +00:00
|
|
|
return (VM_PAGER_OK);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Implement a pretty aggressive clustered getpages strategy. Hint that
|
|
|
|
* everything in an entire 4MB window should be prefaulted at once.
|
|
|
|
*
|
2016-12-08 11:35:53 +00:00
|
|
|
* 4MB (1024 slots per page table page) is convenient for x86,
|
2000-05-21 13:41:29 +00:00
|
|
|
* but may not be for other arches.
|
|
|
|
*/
|
|
|
|
#ifndef PHYSCLUSTER
|
|
|
|
#define PHYSCLUSTER 1024
|
|
|
|
#endif
|
2016-12-08 11:35:53 +00:00
|
|
|
static int phys_pager_cluster = PHYSCLUSTER;
|
|
|
|
SYSCTL_INT(_vm, OID_AUTO, phys_pager_cluster, CTLFLAG_RWTUN,
|
|
|
|
&phys_pager_cluster, 0,
|
|
|
|
"prefault window size for phys pager");
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Max hint to vm_page_alloc() about the further allocation needs
|
|
|
|
* inside the phys_pager_populate() loop. The number of bits used to
|
|
|
|
* implement VM_ALLOC_COUNT() determines the hard limit on this value.
|
|
|
|
* That limit is currently 65535.
|
|
|
|
*/
|
|
|
|
#define PHYSALLOC 16
|
|
|
|
|
|
|
|
static int
|
|
|
|
phys_pager_populate(vm_object_t object, vm_pindex_t pidx,
|
|
|
|
int fault_type __unused, vm_prot_t max_prot __unused, vm_pindex_t *first,
|
|
|
|
vm_pindex_t *last)
|
|
|
|
{
|
|
|
|
vm_page_t m;
|
|
|
|
vm_pindex_t base, end, i;
|
|
|
|
int ahead;
|
|
|
|
|
|
|
|
base = rounddown(pidx, phys_pager_cluster);
|
|
|
|
end = base + phys_pager_cluster - 1;
|
|
|
|
if (end >= object->size)
|
|
|
|
end = object->size - 1;
|
|
|
|
if (*first > base)
|
|
|
|
base = *first;
|
|
|
|
if (end > *last)
|
|
|
|
end = *last;
|
|
|
|
*first = base;
|
|
|
|
*last = end;
|
|
|
|
|
|
|
|
for (i = base; i <= end; i++) {
|
|
|
|
retry:
|
|
|
|
m = vm_page_lookup(object, i);
|
|
|
|
if (m == NULL) {
|
|
|
|
ahead = MIN(end - i, PHYSALLOC);
|
|
|
|
m = vm_page_alloc(object, i, VM_ALLOC_NORMAL |
|
2017-11-08 02:39:37 +00:00
|
|
|
VM_ALLOC_ZERO | VM_ALLOC_WAITFAIL |
|
|
|
|
VM_ALLOC_COUNT(ahead));
|
|
|
|
if (m == NULL)
|
2016-12-08 11:35:53 +00:00
|
|
|
goto retry;
|
|
|
|
if ((m->flags & PG_ZERO) == 0)
|
|
|
|
pmap_zero_page(m);
|
|
|
|
m->valid = VM_PAGE_BITS_ALL;
|
|
|
|
} else if (vm_page_xbusied(m)) {
|
|
|
|
vm_page_lock(m);
|
|
|
|
VM_OBJECT_WUNLOCK(object);
|
|
|
|
vm_page_busy_sleep(m, "physb", true);
|
|
|
|
VM_OBJECT_WLOCK(object);
|
|
|
|
goto retry;
|
|
|
|
} else {
|
|
|
|
vm_page_xbusy(m);
|
|
|
|
if (m->valid != VM_PAGE_BITS_ALL)
|
|
|
|
vm_page_zero_invalid(m, TRUE);
|
|
|
|
}
|
|
|
|
|
|
|
|
KASSERT(m->valid == VM_PAGE_BITS_ALL,
|
|
|
|
("phys_pager_populate: partially valid page %p", m));
|
|
|
|
KASSERT(m->dirty == 0,
|
|
|
|
("phys_pager_populate: dirty page %p", m));
|
|
|
|
}
|
|
|
|
return (VM_PAGER_OK);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
phys_pager_putpages(vm_object_t object, vm_page_t *m, int count, boolean_t sync,
|
|
|
|
int *rtvals)
|
|
|
|
{
|
|
|
|
|
|
|
|
panic("phys_pager_putpage called");
|
|
|
|
}
|
|
|
|
|
2000-05-21 13:41:29 +00:00
|
|
|
static boolean_t
|
2000-07-28 22:03:08 +00:00
|
|
|
phys_pager_haspage(vm_object_t object, vm_pindex_t pindex, int *before,
|
2016-12-08 10:28:51 +00:00
|
|
|
int *after)
|
2000-05-21 13:41:29 +00:00
|
|
|
{
|
|
|
|
vm_pindex_t base, end;
|
|
|
|
|
2016-12-08 11:35:53 +00:00
|
|
|
base = rounddown(pindex, phys_pager_cluster);
|
|
|
|
end = base + phys_pager_cluster - 1;
|
2000-05-21 13:41:29 +00:00
|
|
|
if (before != NULL)
|
|
|
|
*before = pindex - base;
|
|
|
|
if (after != NULL)
|
|
|
|
*after = end - pindex;
|
|
|
|
return (TRUE);
|
|
|
|
}
|
2000-07-28 22:03:08 +00:00
|
|
|
|
|
|
|
struct pagerops physpagerops = {
|
2003-08-05 06:51:26 +00:00
|
|
|
.pgo_init = phys_pager_init,
|
|
|
|
.pgo_alloc = phys_pager_alloc,
|
|
|
|
.pgo_dealloc = phys_pager_dealloc,
|
|
|
|
.pgo_getpages = phys_pager_getpages,
|
|
|
|
.pgo_putpages = phys_pager_putpages,
|
|
|
|
.pgo_haspage = phys_pager_haspage,
|
2016-12-08 11:35:53 +00:00
|
|
|
.pgo_populate = phys_pager_populate,
|
2000-07-28 22:03:08 +00:00
|
|
|
};
|