NetBSD/sys/uvm/uvm_meter.c
chs 83d071a318 add UBC memory-usage balancing. we track the number of pages in use for
each of the basic types (anonymous data, executable image, cached files)
and prevent the pagedaemon from reusing a given page if that would reduce
the count of that type of page below a sysctl-setable minimum threshold.
the thresholds are controlled via three new sysctl tunables:
vm.anonmin, vm.vnodemin, and vm.vtextmin.  these tunables are the
percentages of pageable memory reserved for each usage, and we do not allow
the sum of the minimums to be more than 95% so that there's always some
memory that can be reused.
2001-03-09 01:02:10 +00:00

381 lines
9.6 KiB
C

/* $NetBSD: uvm_meter.c,v 1.17 2001/03/09 01:02:12 chs Exp $ */
/*
* Copyright (c) 1997 Charles D. Cranor and Washington University.
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California.
*
* 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.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Charles D. Cranor,
* Washington University, and 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_meter.c 8.4 (Berkeley) 1/4/94
* from: Id: uvm_meter.c,v 1.1.2.1 1997/08/14 19:10:35 chuck Exp
*/
#include <sys/param.h>
#include <sys/proc.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <uvm/uvm_extern.h>
#include <sys/sysctl.h>
/*
* maxslp: ???? XXXCDC
*/
int maxslp = MAXSLP; /* patchable ... */
struct loadavg averunnable;
/*
* constants for averages over 1, 5, and 15 minutes when sampling at
* 5 second intervals.
*/
static fixpt_t cexp[3] = {
0.9200444146293232 * FSCALE, /* exp(-1/12) */
0.9834714538216174 * FSCALE, /* exp(-1/60) */
0.9944598480048967 * FSCALE, /* exp(-1/180) */
};
/*
* prototypes
*/
static void uvm_loadav __P((struct loadavg *));
static void uvm_total __P((struct vmtotal *));
static int sysctl_uvmexp __P((void *, size_t *));
/*
* uvm_meter: calculate load average and wake up the swapper (if needed)
*/
void
uvm_meter()
{
if ((time.tv_sec % 5) == 0)
uvm_loadav(&averunnable);
if (proc0.p_slptime > (maxslp / 2))
wakeup(&proc0);
}
/*
* uvm_loadav: compute a tenex style load average of a quantity on
* 1, 5, and 15 minute internvals.
*/
static void
uvm_loadav(avg)
struct loadavg *avg;
{
int i, nrun;
struct proc *p;
proclist_lock_read();
nrun = 0;
LIST_FOREACH(p, &allproc, p_list) {
switch (p->p_stat) {
case SSLEEP:
if (p->p_priority > PZERO || p->p_slptime > 1)
continue;
/* fall through */
case SRUN:
case SONPROC:
case SIDL:
nrun++;
}
}
proclist_unlock_read();
for (i = 0; i < 3; i++)
avg->ldavg[i] = (cexp[i] * avg->ldavg[i] +
nrun * FSCALE * (FSCALE - cexp[i])) >> FSHIFT;
}
/*
* uvm_sysctl: sysctl hook into UVM system.
*/
int
uvm_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p)
int *name;
u_int namelen;
void *oldp;
size_t *oldlenp;
void *newp;
size_t newlen;
struct proc *p;
{
struct vmtotal vmtotals;
int rv, t;
/* all sysctl names at this level are terminal */
if (namelen != 1)
return (ENOTDIR); /* overloaded */
switch (name[0]) {
case VM_LOADAVG:
return (sysctl_rdstruct(oldp, oldlenp, newp, &averunnable,
sizeof(averunnable)));
case VM_METER:
uvm_total(&vmtotals);
return (sysctl_rdstruct(oldp, oldlenp, newp, &vmtotals,
sizeof(vmtotals)));
case VM_UVMEXP:
return (sysctl_rdminstruct(oldp, oldlenp, newp, &uvmexp,
sizeof(uvmexp)));
case VM_UVMEXP2:
if (newp)
return (EPERM);
return (sysctl_uvmexp(oldp, oldlenp));
case VM_NKMEMPAGES:
return (sysctl_rdint(oldp, oldlenp, newp, nkmempages));
case VM_ANONMIN:
t = uvmexp.anonminpct;
rv = sysctl_int(oldp, oldlenp, newp, newlen, &t);
if (rv) {
return rv;
}
if (t + uvmexp.vtextminpct + uvmexp.vnodeminpct > 95 || t < 0) {
return EINVAL;
}
uvmexp.anonminpct = t;
uvmexp.anonmin = t * 256 / 100;
return rv;
case VM_VTEXTMIN:
t = uvmexp.vtextminpct;
rv = sysctl_int(oldp, oldlenp, newp, newlen, &t);
if (rv) {
return rv;
}
if (uvmexp.anonminpct + t + uvmexp.vnodeminpct > 95 || t < 0) {
return EINVAL;
}
uvmexp.vtextminpct = t;
uvmexp.vtextmin = t * 256 / 100;
return rv;
case VM_VNODEMIN:
t = uvmexp.vnodeminpct;
rv = sysctl_int(oldp, oldlenp, newp, newlen, &t);
if (rv) {
return rv;
}
if (uvmexp.anonminpct + uvmexp.vtextminpct + t > 95 || t < 0) {
return EINVAL;
}
uvmexp.vnodeminpct = t;
uvmexp.vnodemin = t * 256 / 100;
return rv;
default:
return (EOPNOTSUPP);
}
/* NOTREACHED */
}
static int
sysctl_uvmexp(oldp, oldlenp)
void *oldp;
size_t *oldlenp;
{
struct uvmexp_sysctl u;
memset(&u, 0, sizeof(u));
/* Entries here are in order of uvmexp_sysctl, not uvmexp */
u.pagesize = uvmexp.pagesize;
u.pagemask = uvmexp.pagemask;
u.pageshift = uvmexp.pageshift;
u.npages = uvmexp.npages;
u.free = uvmexp.free;
u.active = uvmexp.active;
u.inactive = uvmexp.inactive;
u.paging = uvmexp.paging;
u.wired = uvmexp.wired;
u.zeropages = uvmexp.zeropages;
u.reserve_pagedaemon = uvmexp.reserve_pagedaemon;
u.reserve_kernel = uvmexp.reserve_kernel;
u.freemin = uvmexp.freemin;
u.freetarg = uvmexp.freetarg;
u.inactarg = uvmexp.inactarg;
u.wiredmax = uvmexp.wiredmax;
u.nswapdev = uvmexp.nswapdev;
u.swpages = uvmexp.swpages;
u.swpginuse = uvmexp.swpginuse;
u.swpgonly = uvmexp.swpgonly;
u.nswget = uvmexp.nswget;
u.nanon = uvmexp.nanon;
u.nanonneeded = uvmexp.nanonneeded;
u.nfreeanon = uvmexp.nfreeanon;
u.faults = uvmexp.faults;
u.traps = uvmexp.traps;
u.intrs = uvmexp.intrs;
u.swtch = uvmexp.swtch;
u.softs = uvmexp.softs;
u.syscalls = uvmexp.syscalls;
u.pageins = uvmexp.pageins;
u.swapins = uvmexp.swapins;
u.swapouts = uvmexp.swapouts;
u.pgswapin = uvmexp.pgswapin;
u.pgswapout = uvmexp.pgswapout;
u.forks = uvmexp.forks;
u.forks_ppwait = uvmexp.forks_ppwait;
u.forks_sharevm = uvmexp.forks_sharevm;
u.pga_zerohit = uvmexp.pga_zerohit;
u.pga_zeromiss = uvmexp.pga_zeromiss;
u.zeroaborts = uvmexp.zeroaborts;
u.fltnoram = uvmexp.fltnoram;
u.fltnoanon = uvmexp.fltnoanon;
u.fltpgwait = uvmexp.fltpgwait;
u.fltpgrele = uvmexp.fltpgrele;
u.fltrelck = uvmexp.fltrelck;
u.fltrelckok = uvmexp.fltrelckok;
u.fltanget = uvmexp.fltanget;
u.fltanretry = uvmexp.fltanretry;
u.fltamcopy = uvmexp.fltamcopy;
u.fltnamap = uvmexp.fltnamap;
u.fltnomap = uvmexp.fltnomap;
u.fltlget = uvmexp.fltlget;
u.fltget = uvmexp.fltget;
u.flt_anon = uvmexp.flt_anon;
u.flt_acow = uvmexp.flt_acow;
u.flt_obj = uvmexp.flt_obj;
u.flt_prcopy = uvmexp.flt_prcopy;
u.flt_przero = uvmexp.flt_przero;
u.pdwoke = uvmexp.pdwoke;
u.pdrevs = uvmexp.pdrevs;
u.pdswout = uvmexp.pdswout;
u.pdfreed = uvmexp.pdfreed;
u.pdscans = uvmexp.pdscans;
u.pdanscan = uvmexp.pdanscan;
u.pdobscan = uvmexp.pdobscan;
u.pdreact = uvmexp.pdreact;
u.pdbusy = uvmexp.pdbusy;
u.pdpageouts = uvmexp.pdpageouts;
u.pdpending = uvmexp.pdpending;
u.pddeact = uvmexp.pddeact;
u.anonpages = uvmexp.anonpages;
u.vnodepages = uvmexp.vnodepages;
u.vtextpages = uvmexp.vtextpages;
return (sysctl_rdminstruct(oldp, oldlenp, NULL, &u, sizeof(u)));
}
/*
* uvm_total: calculate the current state of the system.
*/
static void
uvm_total(totalp)
struct vmtotal *totalp;
{
struct proc *p;
#if 0
vm_map_entry_t entry;
vm_map_t map;
int paging;
#endif
memset(totalp, 0, sizeof *totalp);
/*
* calculate process statistics
*/
proclist_lock_read();
LIST_FOREACH(p, &allproc, p_list) {
if (p->p_flag & P_SYSTEM)
continue;
switch (p->p_stat) {
case 0:
continue;
case SSLEEP:
case SSTOP:
if (p->p_flag & P_INMEM) {
if (p->p_priority <= PZERO)
totalp->t_dw++;
else if (p->p_slptime < maxslp)
totalp->t_sl++;
} else if (p->p_slptime < maxslp)
totalp->t_sw++;
if (p->p_slptime >= maxslp)
continue;
break;
case SRUN:
case SONPROC:
case SIDL:
if (p->p_flag & P_INMEM)
totalp->t_rq++;
else
totalp->t_sw++;
if (p->p_stat == SIDL)
continue;
break;
}
/*
* note active objects
*/
#if 0
/*
* XXXCDC: BOGUS! rethink this. in the mean time
* don't do it.
*/
paging = 0;
vm_map_lock(map);
for (map = &p->p_vmspace->vm_map, entry = map->header.next;
entry != &map->header; entry = entry->next) {
if (entry->is_a_map || entry->is_sub_map ||
entry->object.uvm_obj == NULL)
continue;
/* XXX how to do this with uvm */
}
vm_map_unlock(map);
if (paging)
totalp->t_pw++;
#endif
}
proclist_unlock_read();
/*
* Calculate object memory usage statistics.
*/
totalp->t_free = uvmexp.free;
totalp->t_vm = uvmexp.npages - uvmexp.free + uvmexp.swpginuse;
totalp->t_avm = uvmexp.active + uvmexp.swpginuse; /* XXX */
totalp->t_rm = uvmexp.npages - uvmexp.free;
totalp->t_arm = uvmexp.active;
totalp->t_vmshr = 0; /* XXX */
totalp->t_avmshr = 0; /* XXX */
totalp->t_rmshr = 0; /* XXX */
totalp->t_armshr = 0; /* XXX */
}