Annotation of 43BSDReno/sys/kern/vm_sched.c, revision 1.1.1.1

1.1       root        1: /*
                      2:  * Copyright (c) 1982, 1986, 1989 Regents of the University of California.
                      3:  * All rights reserved.
                      4:  *
                      5:  * Redistribution is only permitted until one year after the first shipment
                      6:  * of 4.4BSD by the Regents.  Otherwise, redistribution and use in source and
                      7:  * binary forms are permitted provided that: (1) source distributions retain
                      8:  * this entire copyright notice and comment, and (2) distributions including
                      9:  * binaries display the following acknowledgement:  This product includes
                     10:  * software developed by the University of California, Berkeley and its
                     11:  * contributors'' in the documentation or other materials provided with the
                     12:  * distribution and in all advertising materials mentioning features or use
                     13:  * of this software.  Neither the name of the University nor the names of
                     14:  * its contributors may be used to endorse or promote products derived from
                     15:  * this software without specific prior written permission.
                     16:  * THIS SOFTWARE IS PROVIDED AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED
                     17:  * WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
                     18:  * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
                     19:  *
                     20:  *     @(#)vm_sched.c  7.9 (Berkeley) 6/28/90
                     21:  */
                     22: 
                     23: #include "param.h"
                     24: #include "systm.h"
                     25: #include "user.h"
                     26: #include "proc.h"
                     27: #include "text.h"
                     28: #include "vm.h"
                     29: #include "kernel.h"
                     30: 
                     31: int    maxslp = MAXSLP;
                     32: int    saferss = SAFERSS;
                     33: 
                     34: /*
                     35:  * The following parameters control operation of the page replacement
                     36:  * algorithm.  They are initialized to 0, and then computed at boot time
                     37:  * based on the size of the system.  If they are patched non-zero in
                     38:  * a loaded vmunix they are left alone and may thus be changed per system
                     39:  * using adb on the loaded system.
                     40:  */
                     41: int    maxpgio = 0;
                     42: int    minfree = 0;
                     43: int    desfree = 0;
                     44: int    lotsfree = 0;
                     45: int    slowscan = 0;
                     46: int    fastscan = 0;
                     47: int    klin = KLIN;
                     48: int    klseql = KLSEQL;
                     49: int    klsdist = KLSDIST;
                     50: int    kltxt = KLTXT;
                     51: int    klout = KLOUT;
                     52: int    multprog = -1;          /* so we don't count process 2 */
                     53: 
                     54: fixpt_t        averunnable[3];         /* load average, of runnable procs */
                     55: 
                     56: /*
                     57:  * The main loop of the scheduling (swapping) process.
                     58:  *
                     59:  * The basic idea is:
                     60:  *     see if anyone wants to be swapped in;
                     61:  *     swap out processes until there is room;
                     62:  *     swap him in;
                     63:  *     repeat.
                     64:  * If the paging rate is too high, or the average free memory
                     65:  * is very low, then we do not consider swapping anyone in,
                     66:  * but rather look for someone to swap out.
                     67:  *
                     68:  * The runout flag is set whenever someone is swapped out.
                     69:  * Sched sleeps on it awaiting work.
                     70:  *
                     71:  * Sched sleeps on runin whenever it cannot find enough
                     72:  * core (by swapping out or otherwise) to fit the
                     73:  * selected swapped process.  It is awakened when the
                     74:  * core situation changes and in any case once per second.
                     75:  *
                     76:  * sched DOESN'T ACCOUNT FOR PAGE TABLE SIZE IN CALCULATIONS.
                     77:  */
                     78: 
                     79: #define        swappable(p) \
                     80:        (((p)->p_flag&(SSYS|SLOCK|SULOCK|SLOAD|SPAGE|SKEEP|SWEXIT|SPHYSIO))==SLOAD)
                     81: 
                     82: /* insure non-zero */
                     83: #define        nz(x)   (x != 0 ? x : 1)
                     84: 
                     85: #define        NBIG    4
                     86: #define        MAXNBIG 10
                     87: int    nbig = NBIG;
                     88: 
                     89: struct bigp {
                     90:        struct  proc *bp_proc;
                     91:        int     bp_pri;
                     92:        struct  bigp *bp_link;
                     93: } bigp[MAXNBIG], bplist;
                     94: 
                     95: sched()
                     96: {
                     97:        register struct proc *rp, *p, *inp;
                     98:        int outpri, inpri, rppri;
                     99:        int sleeper, desperate, deservin, needs, divisor;
                    100:        register struct bigp *bp, *nbp;
                    101:        int biggot, gives;
                    102: 
                    103: loop:
                    104:        wantin = 0;
                    105:        deservin = 0;
                    106:        sleeper = 0;
                    107:        p = 0;
                    108:        /*
                    109:         * See if paging system is overloaded; if so swap someone out.
                    110:         * Conditions for hard outswap are:
                    111:         *      if need kernel map (mix it up).
                    112:         * or
                    113:         *      1. if there are at least 2 runnable processes (on the average)
                    114:         * and  2. the paging rate is excessive or memory is now VERY low.
                    115:         * and  3. the short (5-second) and longer (30-second) average
                    116:         *         memory is less than desirable.
                    117:         */
                    118:        if (kmapwnt ||
                    119:            (averunnable[0] >= 2 * FSCALE &&
                    120:             imax(avefree, avefree30) < desfree &&
                    121:             (rate.v_pgin + rate.v_pgout > maxpgio || avefree < minfree))) {
                    122:                desperate = 1;
                    123:                goto hardswap;
                    124:        }
                    125:        desperate = 0;
                    126:        /*
                    127:         * Not desperate for core,
                    128:         * look for someone who deserves to be brought in.
                    129:         */
                    130:        outpri = -20000;
                    131:        for (rp = allproc; rp != NULL; rp = rp->p_nxt) switch(rp->p_stat) {
                    132: 
                    133:        case SRUN:
                    134:                if ((rp->p_flag&SLOAD) == 0) {
                    135:                        rppri = rp->p_time -
                    136:                            rp->p_swrss / nz((maxpgio/2) * (klin * CLSIZE)) +
                    137:                            rp->p_slptime - (rp->p_nice-NZERO)*8;
                    138:                        if (rppri > outpri) {
                    139:                                if (rp->p_poip)
                    140:                                        continue;
                    141:                                if (rp->p_textp && rp->p_textp->x_poip)
                    142:                                        continue;
                    143:                                p = rp;
                    144:                                outpri = rppri;
                    145:                        }
                    146:                }
                    147:                continue;
                    148: 
                    149:        case SSLEEP:
                    150:        case SSTOP:
                    151:                if ((freemem < desfree || rp->p_rssize == 0) &&
                    152:                    rp->p_slptime > maxslp &&
                    153:                    (!rp->p_textp || (rp->p_textp->x_flag&XLOCK)==0) &&
                    154:                    swappable(rp)) {
                    155:                        /*
                    156:                         * Kick out deadwood.
                    157:                         */
                    158:                        rp->p_flag &= ~SLOAD;
                    159:                        (void) swapout(rp, rp->p_dsize, rp->p_mmsize, rp->p_ssize);
                    160:                }
                    161:                continue;
                    162:        }
                    163: 
                    164:        /*
                    165:         * If something came ready after we checked it,
                    166:         * wantin will be set.  Otherwise,
                    167:         * no one wants in, so nothing to do.
                    168:         */
                    169:        if (outpri == -20000) {
                    170:                (void) splhigh();
                    171:                if (wantin == 0) {
                    172:                        runout++;
                    173:                        sleep((caddr_t)&runout, PSWP);
                    174:                }
                    175:                (void) spl0();
                    176:                goto loop;
                    177:        }
                    178:        /*
                    179:         * Decide how deserving this guy is.  If he is deserving
                    180:         * we will be willing to work harder to bring him in.
                    181:         * Needs is an estimate of how much core he will need.
                    182:         * If he has been out for a while, then we will
                    183:         * bring him in with 1/2 the core he will need, otherwise
                    184:         * we are conservative.
                    185:         */
                    186:        deservin = 0;
                    187:        divisor = 1;
                    188:        if (outpri > maxslp/2) {
                    189:                deservin = 1;
                    190:                divisor = 2;
                    191:        }
                    192:        needs = p->p_swrss;
                    193:        if (p->p_textp && p->p_textp->x_ccount == 0)
                    194:                needs += p->p_textp->x_swrss;
                    195:        needs = imin(needs, lotsfree);
                    196:        if (freemem - deficit > needs / divisor) {
                    197:                deficit += needs;
                    198:                if (swapin(p))
                    199:                        goto loop;
                    200:                deficit -= imin(needs, deficit);
                    201:        }
                    202: 
                    203: hardswap:
                    204:        /*
                    205:         * Need resources (kernel map or memory), swap someone out.
                    206:         * Select the nbig largest jobs, then the oldest of these
                    207:         * is ``most likely to get booted.''
                    208:         */
                    209:        inp = p;
                    210:        sleeper = 0;
                    211:        if (nbig > MAXNBIG)
                    212:                nbig = MAXNBIG;
                    213:        if (nbig < 1)
                    214:                nbig = 1;
                    215:        biggot = 0;
                    216:        bplist.bp_link = 0;
                    217:        for (rp = allproc; rp != NULL; rp = rp->p_nxt) {
                    218:                if (!swappable(rp))
                    219:                        continue;
                    220:                if (rp == inp)
                    221:                        continue;
                    222:                if (rp->p_textp && rp->p_textp->x_flag&XLOCK)
                    223:                        continue;
                    224:                if (rp->p_slptime > maxslp &&
                    225:                    (rp->p_stat==SSLEEP&&rp->p_pri>PZERO||rp->p_stat==SSTOP)) {
                    226:                        if (sleeper < rp->p_slptime) {
                    227:                                p = rp;
                    228:                                sleeper = rp->p_slptime;
                    229:                        }
                    230:                } else if (!sleeper && (rp->p_stat==SRUN||rp->p_stat==SSLEEP)) {
                    231:                        rppri = rp->p_rssize;
                    232:                        if (rp->p_textp)
                    233:                                rppri += rp->p_textp->x_rssize/rp->p_textp->x_ccount;
                    234:                        if (biggot < nbig)
                    235:                                nbp = &bigp[biggot++];
                    236:                        else {
                    237:                                nbp = bplist.bp_link;
                    238:                                if (nbp->bp_pri > rppri)
                    239:                                        continue;
                    240:                                bplist.bp_link = nbp->bp_link;
                    241:                        }
                    242:                        for (bp = &bplist; bp->bp_link; bp = bp->bp_link)
                    243:                                if (rppri < bp->bp_link->bp_pri)
                    244:                                        break;
                    245:                        nbp->bp_link = bp->bp_link;
                    246:                        bp->bp_link = nbp;
                    247:                        nbp->bp_pri = rppri;
                    248:                        nbp->bp_proc = rp;
                    249:                }
                    250:        }
                    251:        if (!sleeper) {
                    252:                p = NULL;
                    253:                inpri = -1000;
                    254:                for (bp = bplist.bp_link; bp; bp = bp->bp_link) {
                    255:                        rp = bp->bp_proc;
                    256:                        rppri = rp->p_time+rp->p_nice-NZERO;
                    257:                        if (rppri >= inpri) {
                    258:                                p = rp;
                    259:                                inpri = rppri;
                    260:                        }
                    261:                }
                    262:        }
                    263:        /*
                    264:         * If we found a long-time sleeper, or we are desperate and
                    265:         * found anyone to swap out, or if someone deserves to come
                    266:         * in and we didn't find a sleeper, but found someone who
                    267:         * has been in core for a reasonable length of time, then
                    268:         * we kick the poor luser out.
                    269:         */
                    270:        if (sleeper || desperate && p || deservin && inpri > maxslp) {
                    271:                (void) splhigh();
                    272:                p->p_flag &= ~SLOAD;
                    273:                if (p->p_stat == SRUN)
                    274:                        remrq(p);
                    275:                (void) spl0();
                    276:                if (desperate) {
                    277:                        /*
                    278:                         * Want to give this space to the rest of
                    279:                         * the processes in core so give them a chance
                    280:                         * by increasing the deficit.
                    281:                         */
                    282:                        gives = p->p_rssize;
                    283:                        if (p->p_textp)
                    284:                                gives += p->p_textp->x_rssize / p->p_textp->x_ccount;
                    285:                        gives = imin(gives, lotsfree);
                    286:                        deficit += gives;
                    287:                } else
                    288:                        gives = 0;      /* someone else taketh away */
                    289:                if (swapout(p, p->p_dsize, p->p_mmsize, p->p_ssize) == 0)
                    290:                        deficit -= imin(gives, deficit);
                    291:                else
                    292:                        goto loop;
                    293:        }
                    294:        /*
                    295:         * Want to swap someone in, but can't
                    296:         * so wait on runin.
                    297:         */
                    298:        (void) splhigh();
                    299:        runin++;
                    300:        sleep((caddr_t)&runin, PSWP);
                    301:        (void) spl0();
                    302:        goto loop;
                    303: }
                    304: 
                    305: vmmeter()
                    306: {
                    307:        register unsigned *cp, *rp, *sp;
                    308: 
                    309:        deficit -= imin(deficit,
                    310:            imax(deficit / 10, ((klin * CLSIZE) / 2) * maxpgio / 2));
                    311:        ave(avefree, freemem, 5);
                    312:        ave(avefree30, freemem, 30);
                    313:        /* v_pgin is maintained by clock.c */
                    314:        cp = &cnt.v_first; rp = &rate.v_first; sp = &sum.v_first;
                    315:        while (cp <= &cnt.v_last) {
                    316:                ave(*rp, *cp, 5);
                    317:                *sp += *cp;
                    318:                *cp = 0;
                    319:                rp++, cp++, sp++;
                    320:        }
                    321:        if (time.tv_sec % 5 == 0) {
                    322:                vmtotal();
                    323:                rate.v_swpin = cnt.v_swpin;
                    324:                sum.v_swpin += cnt.v_swpin;
                    325:                cnt.v_swpin = 0;
                    326:                rate.v_swpout = cnt.v_swpout;
                    327:                sum.v_swpout += cnt.v_swpout;
                    328:                cnt.v_swpout = 0;
                    329:        }
                    330:        if (avefree < minfree && runout || proc[0].p_slptime > maxslp/2) {
                    331:                runout = 0;
                    332:                runin = 0;
                    333:                wakeup((caddr_t)&runin);
                    334:                wakeup((caddr_t)&runout);
                    335:        }
                    336: }
                    337: 
                    338: /*
                    339:  * Schedule rate for paging.
                    340:  * Rate is linear interpolation between
                    341:  * slowscan with lotsfree and fastscan when out of memory.
                    342:  */
                    343: schedpaging()
                    344: {
                    345:        register int vavail;
                    346: 
                    347:        nscan = desscan = 0;
                    348:        vavail = freemem - deficit;
                    349:        if (vavail < 0)
                    350:                vavail = 0;
                    351:        if (freemem < lotsfree) {
                    352:                desscan = (slowscan * vavail + fastscan * (lotsfree - vavail)) /
                    353:                        nz(lotsfree) / RATETOSCHEDPAGING;
                    354:                wakeup((caddr_t)&proc[2]);
                    355:        }
                    356:        timeout(schedpaging, (caddr_t)0, hz / RATETOSCHEDPAGING);
                    357: }
                    358: 
                    359: vmtotal()
                    360: {
                    361:        register struct proc *p;
                    362:        register struct text *xp;
                    363:        int nrun = 0;
                    364: 
                    365:        total.t_vmtxt = 0;
                    366:        total.t_avmtxt = 0;
                    367:        total.t_rmtxt = 0;
                    368:        total.t_armtxt = 0;
                    369:        for (xp = text; xp < textNTEXT; xp++)
                    370:                if (xp->x_vptr) {
                    371:                        total.t_vmtxt += xp->x_size;
                    372:                        total.t_rmtxt += xp->x_rssize;
                    373:                        for (p = xp->x_caddr; p; p = p->p_xlink)
                    374:                        switch (p->p_stat) {
                    375: 
                    376:                        case SSTOP:
                    377:                        case SSLEEP:
                    378:                                if (p->p_slptime >= maxslp)
                    379:                                        continue;
                    380:                                /* fall into... */
                    381: 
                    382:                        case SRUN:
                    383:                        case SIDL:
                    384:                                total.t_avmtxt += xp->x_size;
                    385:                                total.t_armtxt += xp->x_rssize;
                    386:                                goto next;
                    387:                        }
                    388: next:
                    389:                        ;
                    390:                }
                    391:        total.t_vm = 0;
                    392:        total.t_avm = 0;
                    393:        total.t_rm = 0;
                    394:        total.t_arm = 0;
                    395:        total.t_rq = 0;
                    396:        total.t_dw = 0;
                    397:        total.t_pw = 0;
                    398:        total.t_sl = 0;
                    399:        total.t_sw = 0;
                    400:        for (p = allproc; p != NULL; p = p->p_nxt) {
                    401:                if (p->p_flag & SSYS)
                    402:                        continue;
                    403:                if (p->p_stat) {
                    404:                        if (p->p_stat != SZOMB) {
                    405:                                total.t_vm += p->p_dsize + p->p_ssize;
                    406:                                total.t_rm += p->p_rssize;
                    407:                        }
                    408:                        switch (p->p_stat) {
                    409: 
                    410:                        case SSLEEP:
                    411:                                if (p->p_pri <= PZERO && p->p_slptime == 0)
                    412:                                        nrun++;
                    413:                                /* fall through */
                    414:                        case SSTOP:
                    415:                                if (p->p_flag & SPAGE)
                    416:                                        total.t_pw++;
                    417:                                else if (p->p_flag & SLOAD) {
                    418:                                        if (p->p_pri <= PZERO)
                    419:                                                total.t_dw++;
                    420:                                        else if (p->p_slptime < maxslp)
                    421:                                                total.t_sl++;
                    422:                                } else if (p->p_slptime < maxslp)
                    423:                                        total.t_sw++;
                    424:                                if (p->p_slptime < maxslp)
                    425:                                        goto active;
                    426:                                break;
                    427: 
                    428:                        case SRUN:
                    429:                        case SIDL:
                    430:                                nrun++;
                    431:                                if (p->p_flag & SLOAD)
                    432:                                        total.t_rq++;
                    433:                                else
                    434:                                        total.t_sw++;
                    435: active:
                    436:                                total.t_avm += p->p_dsize + p->p_ssize;
                    437:                                total.t_arm += p->p_rssize;
                    438:                                break;
                    439:                        }
                    440:                }
                    441:        }
                    442:        total.t_vm += total.t_vmtxt;
                    443:        total.t_avm += total.t_avmtxt;
                    444:        total.t_rm += total.t_rmtxt;
                    445:        total.t_arm += total.t_armtxt;
                    446:        total.t_free = avefree;
                    447:        loadav(averunnable, nrun);
                    448: }
                    449: 
                    450: /*
                    451:  * Constants for averages over 1, 5, and 15 minutes
                    452:  * when sampling at 5 second intervals.
                    453:  */
                    454: fixpt_t        cexp[3] = {
                    455:        0.9200444146293232 * FSCALE,    /* exp(-1/12) */
                    456:        0.9834714538216174 * FSCALE,    /* exp(-1/60) */
                    457:        0.9944598480048967 * FSCALE,    /* exp(-1/180) */
                    458: };
                    459: 
                    460: /*
                    461:  * Compute a tenex style load average of a quantity on
                    462:  * 1, 5 and 15 minute intervals.
                    463:  */
                    464: loadav(avg, n)
                    465:        register fixpt_t *avg;
                    466:        int n;
                    467: {
                    468:        register int i;
                    469: 
                    470:        for (i = 0; i < 3; i++)
                    471:                avg[i] = (cexp[i] * avg[i] + n * FSCALE * (FSCALE - cexp[i]))
                    472:                         >> FSHIFT;
                    473: #if defined(COMPAT_43) && (defined(vax) || defined(tahoe))
                    474:        for (i = 0; i < 3; i++)
                    475:                avenrun[i] = (double) averunnable[i] / FSCALE;
                    476: #endif /* COMPAT_43 */
                    477: }

unix.superglobalmegacorp.com

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.