freebsd-skq/sys/kern/kern_acct.c
Julian Elischer b40ce4165d KSE Milestone 2
Note ALL MODULES MUST BE RECOMPILED
make the kernel aware that there are smaller units of scheduling than the
process. (but only allow one thread per process at this time).
This is functionally equivalent to teh previousl -current except
that there is a thread associated with each process.

Sorry john! (your next MFC will be a doosie!)

Reviewed by: peter@freebsd.org, dillon@freebsd.org

X-MFC after:    ha ha ha ha
2001-09-12 08:38:13 +00:00

344 lines
10 KiB
C

/*-
* Copyright (c) 1994 Christopher G. Demetriou
* Copyright (c) 1982, 1986, 1989, 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_acct.c 8.1 (Berkeley) 6/14/93
* $FreeBSD$
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/sysproto.h>
#include <sys/proc.h>
#include <sys/mount.h>
#include <sys/vnode.h>
#include <sys/fcntl.h>
#include <sys/syslog.h>
#include <sys/kernel.h>
#include <sys/sysent.h>
#include <sys/sysctl.h>
#include <sys/namei.h>
#include <sys/acct.h>
#include <sys/resourcevar.h>
#include <sys/tty.h>
/*
* The routines implemented in this file are described in:
* Leffler, et al.: The Design and Implementation of the 4.3BSD
* UNIX Operating System (Addison Welley, 1989)
* on pages 62-63.
*
* Arguably, to simplify accounting operations, this mechanism should
* be replaced by one in which an accounting log file (similar to /dev/klog)
* is read by a user process, etc. However, that has its own problems.
*/
/*
* Internal accounting functions.
* The former's operation is described in Leffler, et al., and the latter
* was provided by UCB with the 4.4BSD-Lite release
*/
static comp_t encode_comp_t __P((u_long, u_long));
static void acctwatch __P((void *));
/*
* Accounting callout used for periodic scheduling of acctwatch.
*/
static struct callout acctwatch_callout;
/*
* Accounting vnode pointer, and saved vnode pointer.
*/
static struct vnode *acctp;
static struct vnode *savacctp;
/*
* Values associated with enabling and disabling accounting
*/
static int acctsuspend = 2; /* stop accounting when < 2% free space left */
SYSCTL_INT(_kern, OID_AUTO, acct_suspend, CTLFLAG_RW,
&acctsuspend, 0, "percentage of free disk space below which accounting stops");
static int acctresume = 4; /* resume when free space risen to > 4% */
SYSCTL_INT(_kern, OID_AUTO, acct_resume, CTLFLAG_RW,
&acctresume, 0, "percentage of free disk space above which accounting resumes");
static int acctchkfreq = 15; /* frequency (in seconds) to check space */
SYSCTL_INT(_kern, OID_AUTO, acct_chkfreq, CTLFLAG_RW,
&acctchkfreq, 0, "frequency for checking the free space");
/*
* Accounting system call. Written based on the specification and
* previous implementation done by Mark Tinguely.
*
* MPSAFE
*/
int
acct(td, uap)
struct thread *td;
struct acct_args /* {
syscallarg(char *) path;
} */ *uap;
{
struct nameidata nd;
int error, flags;
mtx_lock(&Giant);
if (td != curthread)
panic("acct"); /* XXXKSE DIAGNOSTIC */
/* Make sure that the caller is root. */
error = suser(td->td_proc);
if (error)
goto done2;
/*
* If accounting is to be started to a file, open that file for
* writing and make sure it's a 'normal'.
*/
if (SCARG(uap, path) != NULL) {
NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, SCARG(uap, path),
td);
flags = FWRITE;
error = vn_open(&nd, &flags, 0);
if (error)
goto done2;
NDFREE(&nd, NDF_ONLY_PNBUF);
VOP_UNLOCK(nd.ni_vp, 0, td);
if (nd.ni_vp->v_type != VREG) {
vn_close(nd.ni_vp, FWRITE, td->td_proc->p_ucred, td);
error = EACCES;
goto done2;
}
}
/*
* If accounting was previously enabled, kill the old space-watcher,
* close the file, and (if no new file was specified, leave).
*/
if (acctp != NULLVP || savacctp != NULLVP) {
callout_stop(&acctwatch_callout);
error = vn_close((acctp != NULLVP ? acctp : savacctp), FWRITE,
td->td_proc->p_ucred, td);
acctp = savacctp = NULLVP;
}
if (SCARG(uap, path) == NULL)
goto done2;
/*
* Save the new accounting file vnode, and schedule the new
* free space watcher.
*/
acctp = nd.ni_vp;
callout_init(&acctwatch_callout, 0);
acctwatch(NULL);
done2:
mtx_unlock(&Giant);
return (error);
}
/*
* Write out process accounting information, on process exit.
* Data to be written out is specified in Leffler, et al.
* and are enumerated below. (They're also noted in the system
* "acct.h" header file.)
*/
int
acct_process(td)
struct thread *td;
{
struct proc *p = td->td_proc;
struct acct acct;
struct rusage *r;
struct timeval ut, st, tmp;
int t;
struct vnode *vp;
/* If accounting isn't enabled, don't bother */
vp = acctp;
if (vp == NULLVP)
return (0);
/*
* Get process accounting information.
*/
/* (1) The name of the command that ran */
bcopy(p->p_comm, acct.ac_comm, sizeof acct.ac_comm);
/* (2) The amount of user and system time that was used */
mtx_lock_spin(&sched_lock);
calcru(p, &ut, &st, NULL);
mtx_unlock_spin(&sched_lock);
acct.ac_utime = encode_comp_t(ut.tv_sec, ut.tv_usec);
acct.ac_stime = encode_comp_t(st.tv_sec, st.tv_usec);
/* (3) The elapsed time the commmand ran (and its starting time) */
acct.ac_btime = p->p_stats->p_start.tv_sec;
microtime(&tmp);
timevalsub(&tmp, &p->p_stats->p_start);
acct.ac_etime = encode_comp_t(tmp.tv_sec, tmp.tv_usec);
/* (4) The average amount of memory used */
r = &p->p_stats->p_ru;
tmp = ut;
timevaladd(&tmp, &st);
t = tmp.tv_sec * hz + tmp.tv_usec / tick;
if (t)
acct.ac_mem = (r->ru_ixrss + r->ru_idrss + r->ru_isrss) / t;
else
acct.ac_mem = 0;
/* (5) The number of disk I/O operations done */
acct.ac_io = encode_comp_t(r->ru_inblock + r->ru_oublock, 0);
/* (6) The UID and GID of the process */
acct.ac_uid = p->p_ucred->cr_ruid;
acct.ac_gid = p->p_ucred->cr_rgid;
/* (7) The terminal from which the process was started */
if ((p->p_flag & P_CONTROLT) && p->p_pgrp->pg_session->s_ttyp)
acct.ac_tty = dev2udev(p->p_pgrp->pg_session->s_ttyp->t_dev);
else
acct.ac_tty = NOUDEV;
/* (8) The boolean flags that tell how the process terminated, etc. */
acct.ac_flag = p->p_acflag;
/*
* Eliminate any file size rlimit.
*/
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);
}
p->p_rlimit[RLIMIT_FSIZE].rlim_cur = RLIM_INFINITY;
/*
* Write the accounting information to the file.
*/
VOP_LEASE(vp, td, td->td_proc->p_ucred, LEASE_WRITE);
return (vn_rdwr(UIO_WRITE, vp, (caddr_t)&acct, sizeof (acct),
(off_t)0, UIO_SYSSPACE, IO_APPEND|IO_UNIT, td->td_proc->p_ucred,
(int *)0, td));
}
/*
* Encode_comp_t converts from ticks in seconds and microseconds
* to ticks in 1/AHZ seconds. The encoding is described in
* Leffler, et al., on page 63.
*/
#define MANTSIZE 13 /* 13 bit mantissa. */
#define EXPSIZE 3 /* Base 8 (3 bit) exponent. */
#define MAXFRACT ((1 << MANTSIZE) - 1) /* Maximum fractional value. */
static comp_t
encode_comp_t(s, us)
u_long s, us;
{
int exp, rnd;
exp = 0;
rnd = 0;
s *= AHZ;
s += us / (1000000 / AHZ); /* Maximize precision. */
while (s > MAXFRACT) {
rnd = s & (1 << (EXPSIZE - 1)); /* Round up? */
s >>= EXPSIZE; /* Base 8 exponent == 3 bit shift. */
exp++;
}
/* If we need to round up, do it (and handle overflow correctly). */
if (rnd && (++s > MAXFRACT)) {
s >>= EXPSIZE;
exp++;
}
/* Clean it up and polish it off. */
exp <<= MANTSIZE; /* Shift the exponent into place */
exp += s; /* and add on the mantissa. */
return (exp);
}
/*
* Periodically check the file system to see if accounting
* should be turned on or off. Beware the case where the vnode
* has been vgone()'d out from underneath us, e.g. when the file
* system containing the accounting file has been forcibly unmounted.
*/
/* ARGSUSED */
static void
acctwatch(a)
void *a;
{
struct statfs sb;
if (savacctp != NULLVP) {
if (savacctp->v_type == VBAD) {
(void) vn_close(savacctp, FWRITE, NOCRED, NULL);
savacctp = NULLVP;
return;
}
(void)VFS_STATFS(savacctp->v_mount, &sb, (struct thread *)0);
if (sb.f_bavail > acctresume * sb.f_blocks / 100) {
acctp = savacctp;
savacctp = NULLVP;
log(LOG_NOTICE, "Accounting resumed\n");
}
} else {
if (acctp == NULLVP)
return;
if (acctp->v_type == VBAD) {
(void) vn_close(acctp, FWRITE, NOCRED, NULL);
acctp = NULLVP;
return;
}
(void)VFS_STATFS(acctp->v_mount, &sb, (struct thread *)0);
if (sb.f_bavail <= acctsuspend * sb.f_blocks / 100) {
savacctp = acctp;
acctp = NULLVP;
log(LOG_NOTICE, "Accounting suspended\n");
}
}
callout_reset(&acctwatch_callout, acctchkfreq * hz, acctwatch, NULL);
}