freebsd-skq/sys/kern/kern_resource.c
bde d3ce1542c8 Fixed double counting of runtime after a process exits. The last
timeslice of the exiting process was counted for both the exiting
process and the next process to run if the next process runs
immediately.

Broken in:	mostly in kern_clock.c rev.1.70 (1998/05/28)
1998-07-27 19:16:21 +00:00

624 lines
14 KiB
C

/*-
* Copyright (c) 1982, 1986, 1991, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* 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 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.
*
* @(#)kern_resource.c 8.5 (Berkeley) 1/21/94
* $Id: kern_resource.c,v 1.37 1998/05/28 09:30:18 phk Exp $
*/
#include "opt_compat.h"
#include "opt_rlimit.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/sysproto.h>
#include <sys/kernel.h>
#include <sys/file.h>
#include <sys/resourcevar.h>
#include <sys/malloc.h>
#include <sys/proc.h>
#include <vm/vm.h>
#include <vm/vm_param.h>
#include <vm/vm_prot.h>
#include <sys/lock.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
static int donice __P((struct proc *curp, struct proc *chgp, int n));
static int dosetrlimit __P((struct proc *p, u_int which, struct rlimit *limp));
/*
* Resource controls and accounting.
*/
#ifndef _SYS_SYSPROTO_H_
struct getpriority_args {
int which;
int who;
};
#endif
int
getpriority(curp, uap)
struct proc *curp;
register struct getpriority_args *uap;
{
register struct proc *p;
register int low = PRIO_MAX + 1;
switch (uap->which) {
case PRIO_PROCESS:
if (uap->who == 0)
p = curp;
else
p = pfind(uap->who);
if (p == 0)
break;
low = p->p_nice;
break;
case PRIO_PGRP: {
register struct pgrp *pg;
if (uap->who == 0)
pg = curp->p_pgrp;
else if ((pg = pgfind(uap->who)) == NULL)
break;
for (p = pg->pg_members.lh_first; p != 0;
p = p->p_pglist.le_next) {
if (p->p_nice < low)
low = p->p_nice;
}
break;
}
case PRIO_USER:
if (uap->who == 0)
uap->who = curp->p_ucred->cr_uid;
for (p = allproc.lh_first; p != 0; p = p->p_list.le_next)
if (p->p_ucred->cr_uid == uap->who &&
p->p_nice < low)
low = p->p_nice;
break;
default:
return (EINVAL);
}
if (low == PRIO_MAX + 1)
return (ESRCH);
curp->p_retval[0] = low;
return (0);
}
#ifndef _SYS_SYSPROTO_H_
struct setpriority_args {
int which;
int who;
int prio;
};
#endif
/* ARGSUSED */
int
setpriority(curp, uap)
struct proc *curp;
register struct setpriority_args *uap;
{
register struct proc *p;
int found = 0, error = 0;
switch (uap->which) {
case PRIO_PROCESS:
if (uap->who == 0)
p = curp;
else
p = pfind(uap->who);
if (p == 0)
break;
error = donice(curp, p, uap->prio);
found++;
break;
case PRIO_PGRP: {
register struct pgrp *pg;
if (uap->who == 0)
pg = curp->p_pgrp;
else if ((pg = pgfind(uap->who)) == NULL)
break;
for (p = pg->pg_members.lh_first; p != 0;
p = p->p_pglist.le_next) {
error = donice(curp, p, uap->prio);
found++;
}
break;
}
case PRIO_USER:
if (uap->who == 0)
uap->who = curp->p_ucred->cr_uid;
for (p = allproc.lh_first; p != 0; p = p->p_list.le_next)
if (p->p_ucred->cr_uid == uap->who) {
error = donice(curp, p, uap->prio);
found++;
}
break;
default:
return (EINVAL);
}
if (found == 0)
return (ESRCH);
return (error);
}
static int
donice(curp, chgp, n)
register struct proc *curp, *chgp;
register int n;
{
register struct pcred *pcred = curp->p_cred;
if (pcred->pc_ucred->cr_uid && pcred->p_ruid &&
pcred->pc_ucred->cr_uid != chgp->p_ucred->cr_uid &&
pcred->p_ruid != chgp->p_ucred->cr_uid)
return (EPERM);
if (n > PRIO_MAX)
n = PRIO_MAX;
if (n < PRIO_MIN)
n = PRIO_MIN;
if (n < chgp->p_nice && suser(pcred->pc_ucred, &curp->p_acflag))
return (EACCES);
chgp->p_nice = n;
(void)resetpriority(chgp);
return (0);
}
/* rtprio system call */
#ifndef _SYS_SYSPROTO_H_
struct rtprio_args {
int function;
pid_t pid;
struct rtprio *rtp;
};
#endif
/*
* Set realtime priority
*/
/* ARGSUSED */
int
rtprio(curp, uap)
struct proc *curp;
register struct rtprio_args *uap;
{
register struct proc *p;
register struct pcred *pcred = curp->p_cred;
struct rtprio rtp;
int error;
error = copyin(uap->rtp, &rtp, sizeof(struct rtprio));
if (error)
return (error);
if (uap->pid == 0)
p = curp;
else
p = pfind(uap->pid);
if (p == 0)
return (ESRCH);
switch (uap->function) {
case RTP_LOOKUP:
return (copyout(&p->p_rtprio, uap->rtp, sizeof(struct rtprio)));
case RTP_SET:
if (pcred->pc_ucred->cr_uid && pcred->p_ruid &&
pcred->pc_ucred->cr_uid != p->p_ucred->cr_uid &&
pcred->p_ruid != p->p_ucred->cr_uid)
return (EPERM);
/* disallow setting rtprio in most cases if not superuser */
if (suser(pcred->pc_ucred, &curp->p_acflag)) {
/* can't set someone else's */
if (uap->pid)
return (EPERM);
/* can't set realtime priority */
/*
* Realtime priority has to be restricted for reasons which should be
* obvious. However, for idle priority, there is a potential for
* system deadlock if an idleprio process gains a lock on a resource
* that other processes need (and the idleprio process can't run
* due to a CPU-bound normal process). Fix me! XXX
*/
#if 0
if (RTP_PRIO_IS_REALTIME(rtp.type))
#endif
if (rtp.type != RTP_PRIO_NORMAL)
return (EPERM);
}
switch (rtp.type) {
#ifdef RTP_PRIO_FIFO
case RTP_PRIO_FIFO:
#endif
case RTP_PRIO_REALTIME:
case RTP_PRIO_NORMAL:
case RTP_PRIO_IDLE:
if (rtp.prio > RTP_PRIO_MAX)
return (EINVAL);
p->p_rtprio = rtp;
return (0);
default:
return (EINVAL);
}
default:
return (EINVAL);
}
}
#if defined(COMPAT_43) || defined(COMPAT_SUNOS)
#ifndef _SYS_SYSPROTO_H_
struct osetrlimit_args {
u_int which;
struct orlimit *rlp;
};
#endif
/* ARGSUSED */
int
osetrlimit(p, uap)
struct proc *p;
register struct osetrlimit_args *uap;
{
struct orlimit olim;
struct rlimit lim;
int error;
if ((error =
copyin((caddr_t)uap->rlp, (caddr_t)&olim, sizeof(struct orlimit))))
return (error);
lim.rlim_cur = olim.rlim_cur;
lim.rlim_max = olim.rlim_max;
return (dosetrlimit(p, uap->which, &lim));
}
#ifndef _SYS_SYSPROTO_H_
struct ogetrlimit_args {
u_int which;
struct orlimit *rlp;
};
#endif
/* ARGSUSED */
int
ogetrlimit(p, uap)
struct proc *p;
register struct ogetrlimit_args *uap;
{
struct orlimit olim;
if (uap->which >= RLIM_NLIMITS)
return (EINVAL);
olim.rlim_cur = p->p_rlimit[uap->which].rlim_cur;
if (olim.rlim_cur == -1)
olim.rlim_cur = 0x7fffffff;
olim.rlim_max = p->p_rlimit[uap->which].rlim_max;
if (olim.rlim_max == -1)
olim.rlim_max = 0x7fffffff;
return (copyout((caddr_t)&olim, (caddr_t)uap->rlp, sizeof(olim)));
}
#endif /* COMPAT_43 || COMPAT_SUNOS */
#ifndef _SYS_SYSPROTO_H_
struct __setrlimit_args {
u_int which;
struct rlimit *rlp;
};
#endif
/* ARGSUSED */
int
setrlimit(p, uap)
struct proc *p;
register struct __setrlimit_args *uap;
{
struct rlimit alim;
int error;
if ((error =
copyin((caddr_t)uap->rlp, (caddr_t)&alim, sizeof (struct rlimit))))
return (error);
return (dosetrlimit(p, uap->which, &alim));
}
static int
dosetrlimit(p, which, limp)
struct proc *p;
u_int which;
struct rlimit *limp;
{
register struct rlimit *alimp;
int error;
if (which >= RLIM_NLIMITS)
return (EINVAL);
alimp = &p->p_rlimit[which];
/*
* Preserve historical bugs by treating negative limits as unsigned.
*/
if (limp->rlim_cur < 0)
limp->rlim_cur = RLIM_INFINITY;
if (limp->rlim_max < 0)
limp->rlim_max = RLIM_INFINITY;
if (limp->rlim_cur > alimp->rlim_max ||
limp->rlim_max > alimp->rlim_max)
if ((error = suser(p->p_ucred, &p->p_acflag)))
return (error);
if (limp->rlim_cur > limp->rlim_max)
limp->rlim_cur = limp->rlim_max;
if (p->p_limit->p_refcnt > 1 &&
(p->p_limit->p_lflags & PL_SHAREMOD) == 0) {
p->p_limit->p_refcnt--;
p->p_limit = limcopy(p->p_limit);
alimp = &p->p_rlimit[which];
}
switch (which) {
case RLIMIT_CPU:
if (limp->rlim_cur > RLIM_INFINITY / (rlim_t)1000000)
p->p_limit->p_cpulimit = RLIM_INFINITY;
else
p->p_limit->p_cpulimit =
(rlim_t)1000000 * limp->rlim_cur;
break;
case RLIMIT_DATA:
if (limp->rlim_cur > MAXDSIZ)
limp->rlim_cur = MAXDSIZ;
if (limp->rlim_max > MAXDSIZ)
limp->rlim_max = MAXDSIZ;
break;
case RLIMIT_STACK:
if (limp->rlim_cur > MAXSSIZ)
limp->rlim_cur = MAXSSIZ;
if (limp->rlim_max > MAXSSIZ)
limp->rlim_max = MAXSSIZ;
/*
* Stack is allocated to the max at exec time with only
* "rlim_cur" bytes accessible. If stack limit is going
* up make more accessible, if going down make inaccessible.
*/
if (limp->rlim_cur != alimp->rlim_cur) {
vm_offset_t addr;
vm_size_t size;
vm_prot_t prot;
if (limp->rlim_cur > alimp->rlim_cur) {
prot = VM_PROT_ALL;
size = limp->rlim_cur - alimp->rlim_cur;
addr = USRSTACK - limp->rlim_cur;
} else {
prot = VM_PROT_NONE;
size = alimp->rlim_cur - limp->rlim_cur;
addr = USRSTACK - alimp->rlim_cur;
}
addr = trunc_page(addr);
size = round_page(size);
(void) vm_map_protect(&p->p_vmspace->vm_map,
addr, addr+size, prot, FALSE);
}
break;
case RLIMIT_NOFILE:
if (limp->rlim_cur > maxfilesperproc)
limp->rlim_cur = maxfilesperproc;
if (limp->rlim_max > maxfilesperproc)
limp->rlim_max = maxfilesperproc;
break;
case RLIMIT_NPROC:
if (limp->rlim_cur > maxprocperuid)
limp->rlim_cur = maxprocperuid;
if (limp->rlim_max > maxprocperuid)
limp->rlim_max = maxprocperuid;
break;
}
*alimp = *limp;
return (0);
}
#ifndef _SYS_SYSPROTO_H_
struct __getrlimit_args {
u_int which;
struct rlimit *rlp;
};
#endif
/* ARGSUSED */
int
getrlimit(p, uap)
struct proc *p;
register struct __getrlimit_args *uap;
{
if (uap->which >= RLIM_NLIMITS)
return (EINVAL);
return (copyout((caddr_t)&p->p_rlimit[uap->which], (caddr_t)uap->rlp,
sizeof (struct rlimit)));
}
/*
* Transform the running time and tick information in proc p into user,
* system, and interrupt time usage.
*/
void
calcru(p, up, sp, ip)
struct proc *p;
struct timeval *up;
struct timeval *sp;
struct timeval *ip;
{
int64_t totusec;
u_int64_t u, st, ut, it, tot;
int s;
struct timeval tv;
/* XXX: why spl-protect ? worst case is an off-by-one report */
s = splstatclock();
st = p->p_sticks;
ut = p->p_uticks;
it = p->p_iticks;
splx(s);
tot = st + ut + it;
if (tot == 0) {
st = 1;
tot = 1;
}
totusec = p->p_runtime;
#ifdef SMP
if (p->p_oncpu != (char)0xff) {
#else
if (p == curproc) {
#endif
/*
* Adjust for the current time slice. This is actually fairly
* important since the error here is on the order of a time
* quantum, which is much greater than the sampling error.
*/
microuptime(&tv);
totusec += (tv.tv_usec - p->p_switchtime.tv_usec) +
(tv.tv_sec - p->p_switchtime.tv_sec) * (int64_t)1000000;
/*
* Copy the time that was just read to `switchtime' in case
* we are being called from exit1(). Exits don't go through
* mi_switch(), so `switchtime' doesn't get set in the normal
* way. We set it here instead of more cleanly in exit1()
* to avoid losing track of the time between the calls to
* microuptime().
*/
switchtime = tv;
}
if (totusec < 0) {
/* XXX no %qd in kernel. Truncate. */
printf("calcru: negative time of %ld usec for pid %d (%s)\n",
(long)totusec, p->p_pid, p->p_comm);
totusec = 0;
}
u = totusec;
st = (u * st) / tot;
sp->tv_sec = st / 1000000;
sp->tv_usec = st % 1000000;
ut = (u * ut) / tot;
up->tv_sec = ut / 1000000;
up->tv_usec = ut % 1000000;
if (ip != NULL) {
it = (u * it) / tot;
ip->tv_sec = it / 1000000;
ip->tv_usec = it % 1000000;
}
}
#ifndef _SYS_SYSPROTO_H_
struct getrusage_args {
int who;
struct rusage *rusage;
};
#endif
/* ARGSUSED */
int
getrusage(p, uap)
register struct proc *p;
register struct getrusage_args *uap;
{
register struct rusage *rup;
switch (uap->who) {
case RUSAGE_SELF:
rup = &p->p_stats->p_ru;
calcru(p, &rup->ru_utime, &rup->ru_stime, NULL);
break;
case RUSAGE_CHILDREN:
rup = &p->p_stats->p_cru;
break;
default:
return (EINVAL);
}
return (copyout((caddr_t)rup, (caddr_t)uap->rusage,
sizeof (struct rusage)));
}
void
ruadd(ru, ru2)
register struct rusage *ru, *ru2;
{
register long *ip, *ip2;
register int i;
timevaladd(&ru->ru_utime, &ru2->ru_utime);
timevaladd(&ru->ru_stime, &ru2->ru_stime);
if (ru->ru_maxrss < ru2->ru_maxrss)
ru->ru_maxrss = ru2->ru_maxrss;
ip = &ru->ru_first; ip2 = &ru2->ru_first;
for (i = &ru->ru_last - &ru->ru_first; i >= 0; i--)
*ip++ += *ip2++;
}
/*
* Make a copy of the plimit structure.
* We share these structures copy-on-write after fork,
* and copy when a limit is changed.
*/
struct plimit *
limcopy(lim)
struct plimit *lim;
{
register struct plimit *copy;
MALLOC(copy, struct plimit *, sizeof(struct plimit),
M_SUBPROC, M_WAITOK);
bcopy(lim->pl_rlimit, copy->pl_rlimit, sizeof(struct plimit));
copy->p_lflags = 0;
copy->p_refcnt = 1;
return (copy);
}