freebsd-dev/sys/vm/vm_pageq.c
John Baldwin 9bdaa43379 Move the code to handle the vm.blacklist tunable up a layer into
vm_page_startup().  As a result, we now only lookup the tunable once
instead of looking it up once for every physical page of memory in the
system.  This cuts out about a 1 second or so delay in boot on x86
systems.  The delay is much larger and more noticable on sun4v apparently.

Reported by:	kmacy
MFC after:	1 week
2006-06-23 16:44:24 +00:00

333 lines
8.5 KiB
C

/*-
* Copyright (c) 1998 Matthew Dillon. All Rights Reserved.
* 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.
* 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 AUTHOR ``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 AUTHOR 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_vmpage.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/linker_set.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/sysctl.h>
#include <sys/proc.h>
#include <sys/vmmeter.h>
#include <sys/vnode.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/vm_kern.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_pageout.h>
#include <vm/vm_pager.h>
#include <vm/vm_extern.h>
static void vm_coloring_init(void);
void setPQL2(int *const size, int *const ways);
struct vpgqueues vm_page_queues[PQ_MAXCOUNT];
struct pq_coloring page_queue_coloring;
static int pq_cachesize = 0; /* size of the cache in KB */
static int pq_cachenways = 0; /* associativity of the cache */
SYSCTL_DECL(_vm_stats);
SYSCTL_NODE(_vm_stats, OID_AUTO, pagequeue, CTLFLAG_RW, 0, "VM meter stats");
SYSCTL_INT(_vm_stats_pagequeue, OID_AUTO, page_colors, CTLFLAG_RD,
&(PQ_NUMCOLORS), 0, "Number of colors in the page queue");
SYSCTL_INT(_vm_stats_pagequeue, OID_AUTO, cachesize, CTLFLAG_RD,
&pq_cachesize, 0, "Size of the processor cache in KB");
SYSCTL_INT(_vm_stats_pagequeue, OID_AUTO, cachenways, CTLFLAG_RD,
&pq_cachenways, 0, "Associativity of the processor cache");
SYSCTL_INT(_vm_stats_pagequeue, OID_AUTO, prime1, CTLFLAG_RD,
&(PQ_PRIME1), 0, "Cache tuning value");
SYSCTL_INT(_vm_stats_pagequeue, OID_AUTO, prime2, CTLFLAG_RD,
&(PQ_PRIME2), 0, "Cache tuning value");
static void
vm_coloring_init(void)
{
#ifdef PQ_NOOPT
PQ_NUMCOLORS = PQ_PRIME1 = PQ_PRIME2 = 1;
#else
setPQL2(&pq_cachesize, &pq_cachenways);
CTASSERT(PAGE_SIZE/1024 > 0);
if (pq_cachesize > 0 && pq_cachenways > 0)
PQ_NUMCOLORS = pq_cachesize / (PAGE_SIZE/1024) / \
pq_cachenways;
else
PQ_NUMCOLORS = 32;
if (PQ_MAXCOLORS < PQ_NUMCOLORS) {
printf("VM-PQ color limit (PQ_MAXCOLORS=%u) exceeded (%u), see vm_page.h", PQ_MAXCOLORS, PQ_NUMCOLORS);
PQ_NUMCOLORS = PQ_MAXCOLORS;
}
if (PQ_NUMCOLORS >= 128) {
PQ_PRIME1 = 31;
PQ_PRIME2 = 23;
} else if (PQ_NUMCOLORS >= 64) {
PQ_PRIME1 = 13;
PQ_PRIME2 = 7;
} else if (PQ_NUMCOLORS >= 32) {
PQ_PRIME1 = 9;
PQ_PRIME2 = 5;
} else if (PQ_NUMCOLORS >= 16) {
PQ_PRIME1 = 5;
PQ_PRIME2 = 3;
} else
PQ_NUMCOLORS = PQ_PRIME1 = PQ_PRIME2 = 1;
#endif
/*
* PQ_CACHE represents a
* PQ_NUMCOLORS consecutive queue.
*/
PQ_COLORMASK = PQ_NUMCOLORS - 1;
PQ_INACTIVE = 1 + PQ_NUMCOLORS;
PQ_ACTIVE = 2 + PQ_NUMCOLORS;
PQ_CACHE = 3 + PQ_NUMCOLORS;
PQ_HOLD = 3 + 2 * PQ_NUMCOLORS;
PQ_COUNT = 4 + 2 * PQ_NUMCOLORS;
PQ_MAXLENGTH = PQ_NUMCOLORS / 3 + PQ_PRIME1;
#if 0
/* XXX: is it possible to allocate vm_page_queues[PQ_COUNT] here? */
#error XXX: vm_page_queues = malloc(PQ_COUNT * sizeof(struct vpgqueues));
#endif
if (bootverbose)
if (PQ_NUMCOLORS > 1)
printf("Using %d colors for the VM-PQ tuning (%d, %d)\n",
PQ_NUMCOLORS, pq_cachesize, pq_cachenways);
}
void
vm_pageq_init(void)
{
int i;
vm_coloring_init();
for (i = 0; i < PQ_NUMCOLORS; ++i) {
vm_page_queues[PQ_FREE+i].cnt = &cnt.v_free_count;
}
for (i = 0; i < PQ_NUMCOLORS; ++i) {
vm_page_queues[PQ_CACHE + i].cnt = &cnt.v_cache_count;
}
vm_page_queues[PQ_INACTIVE].cnt = &cnt.v_inactive_count;
vm_page_queues[PQ_ACTIVE].cnt = &cnt.v_active_count;
vm_page_queues[PQ_HOLD].cnt = &cnt.v_active_count;
for (i = 0; i < PQ_COUNT; i++) {
TAILQ_INIT(&vm_page_queues[i].pl);
}
}
void
vm_pageq_requeue(vm_page_t m)
{
int queue = VM_PAGE_GETQUEUE(m);
struct vpgqueues *vpq;
if (queue != PQ_NONE) {
vpq = &vm_page_queues[queue];
TAILQ_REMOVE(&vpq->pl, m, pageq);
TAILQ_INSERT_TAIL(&vpq->pl, m, pageq);
}
}
/*
* vm_pageq_enqueue:
*/
void
vm_pageq_enqueue(int queue, vm_page_t m)
{
struct vpgqueues *vpq;
vpq = &vm_page_queues[queue];
VM_PAGE_SETQUEUE2(m, queue);
TAILQ_INSERT_TAIL(&vpq->pl, m, pageq);
++*vpq->cnt;
++vpq->lcnt;
}
/*
* vm_add_new_page:
*
* Add a new page to the freelist for use by the system.
*/
vm_page_t
vm_pageq_add_new_page(vm_paddr_t pa)
{
vm_page_t m;
atomic_add_int(&cnt.v_page_count, 1);
m = PHYS_TO_VM_PAGE(pa);
m->phys_addr = pa;
m->flags = 0;
m->pc = (pa >> PAGE_SHIFT) & PQ_COLORMASK;
pmap_page_init(m);
mtx_lock_spin(&vm_page_queue_free_mtx);
vm_pageq_enqueue(m->pc + PQ_FREE, m);
mtx_unlock_spin(&vm_page_queue_free_mtx);
return (m);
}
/*
* vm_pageq_remove_nowakeup:
*
* vm_page_unqueue() without any wakeup
*
* The queue containing the given page must be locked.
* This routine may not block.
*/
void
vm_pageq_remove_nowakeup(vm_page_t m)
{
int queue = VM_PAGE_GETQUEUE(m);
struct vpgqueues *pq;
if (queue != PQ_NONE) {
pq = &vm_page_queues[queue];
VM_PAGE_SETQUEUE2(m, PQ_NONE);
TAILQ_REMOVE(&pq->pl, m, pageq);
(*pq->cnt)--;
pq->lcnt--;
}
}
/*
* vm_pageq_remove:
*
* Remove a page from its queue.
*
* The queue containing the given page must be locked.
* This routine may not block.
*/
void
vm_pageq_remove(vm_page_t m)
{
int queue = VM_PAGE_GETQUEUE(m);
struct vpgqueues *pq;
if (queue != PQ_NONE) {
VM_PAGE_SETQUEUE2(m, PQ_NONE);
pq = &vm_page_queues[queue];
TAILQ_REMOVE(&pq->pl, m, pageq);
(*pq->cnt)--;
pq->lcnt--;
if (VM_PAGE_RESOLVEQUEUE(m, queue) == PQ_CACHE) {
if (vm_paging_needed())
pagedaemon_wakeup();
}
}
}
#ifndef PQ_NOOPT
/*
* vm_pageq_find:
*
* Find a page on the specified queue with color optimization.
*
* The page coloring optimization attempts to locate a page
* that does not overload other nearby pages in the object in
* the cpu's L2 cache. We need this optimization because cpu
* caches tend to be physical caches, while object spaces tend
* to be virtual.
*
* The specified queue must be locked.
* This routine may not block.
*
* This routine may only be called from the vm_pageq_find()
* function in this file.
*/
static inline vm_page_t
_vm_pageq_find(int basequeue, int index)
{
int i;
vm_page_t m = NULL;
struct vpgqueues *pq;
pq = &vm_page_queues[basequeue];
/*
* Note that for the first loop, index+i and index-i wind up at the
* same place. Even though this is not totally optimal, we've already
* blown it by missing the cache case so we do not care.
*/
for (i = PQ_NUMCOLORS / 2; i > 0; --i) {
if ((m = TAILQ_FIRST(&pq[(index + i) & PQ_COLORMASK].pl)) \
!= NULL)
break;
if ((m = TAILQ_FIRST(&pq[(index - i) & PQ_COLORMASK].pl)) \
!= NULL)
break;
}
return (m);
}
#endif /* PQ_NOOPT */
vm_page_t
vm_pageq_find(int basequeue, int index, boolean_t prefer_zero)
{
vm_page_t m;
#ifndef PQ_NOOPT
if (PQ_NUMCOLORS > 1) {
if (prefer_zero) {
m = TAILQ_LAST(&vm_page_queues[basequeue+index].pl, \
pglist);
} else {
m = TAILQ_FIRST(&vm_page_queues[basequeue+index].pl);
}
if (m == NULL) {
m = _vm_pageq_find(basequeue, index);
}
} else {
#endif
if (prefer_zero) {
m = TAILQ_LAST(&vm_page_queues[basequeue].pl, pglist);
} else {
m = TAILQ_FIRST(&vm_page_queues[basequeue].pl);
}
#ifndef PQ_NOOPT
}
#endif
return (m);
}