The Design and Implementation of the FreeBSD Operating System, Second Edition
Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]

FreeBSD/Linux Kernel Cross Reference
sys/kern/kern_acct.c

Version: -  FREEBSD  -  FREEBSD-13-STABLE  -  FREEBSD-13-0  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  l41  -  OPENBSD  -  linux-2.6  -  MK84  -  PLAN9  -  xnu-8792 
SearchContext: -  none  -  3  -  10 

    1 /*-
    2  * Copyright (c) 1994 Christopher G. Demetriou
    3  * Copyright (c) 1982, 1986, 1989, 1993
    4  *      The Regents of the University of California.  All rights reserved.
    5  * (c) UNIX System Laboratories, Inc.
    6  * All or some portions of this file are derived from material licensed
    7  * to the University of California by American Telephone and Telegraph
    8  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
    9  * the permission of UNIX System Laboratories, Inc.
   10  *
   11  * Redistribution and use in source and binary forms, with or without
   12  * modification, are permitted provided that the following conditions
   13  * are met:
   14  * 1. Redistributions of source code must retain the above copyright
   15  *    notice, this list of conditions and the following disclaimer.
   16  * 2. Redistributions in binary form must reproduce the above copyright
   17  *    notice, this list of conditions and the following disclaimer in the
   18  *    documentation and/or other materials provided with the distribution.
   19  * 3. All advertising materials mentioning features or use of this software
   20  *    must display the following acknowledgement:
   21  *      This product includes software developed by the University of
   22  *      California, Berkeley and its contributors.
   23  * 4. Neither the name of the University nor the names of its contributors
   24  *    may be used to endorse or promote products derived from this software
   25  *    without specific prior written permission.
   26  *
   27  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
   28  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
   29  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
   30  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
   31  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
   32  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
   33  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
   34  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
   35  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
   36  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
   37  * SUCH DAMAGE.
   38  *
   39  *      @(#)kern_acct.c 8.1 (Berkeley) 6/14/93
   40  * $FreeBSD: releng/5.1/sys/kern/kern_acct.c 114434 2003-05-01 16:59:23Z des $
   41  */
   42 
   43 #include "opt_mac.h"
   44 
   45 #include <sys/param.h>
   46 #include <sys/systm.h>
   47 #include <sys/lock.h>
   48 #include <sys/mutex.h>
   49 #include <sys/sysproto.h>
   50 #include <sys/proc.h>
   51 #include <sys/mac.h>
   52 #include <sys/mount.h>
   53 #include <sys/vnode.h>
   54 #include <sys/fcntl.h>
   55 #include <sys/syslog.h>
   56 #include <sys/kernel.h>
   57 #include <sys/sysent.h>
   58 #include <sys/sysctl.h>
   59 #include <sys/namei.h>
   60 #include <sys/acct.h>
   61 #include <sys/resourcevar.h>
   62 #include <sys/tty.h>
   63 
   64 /*
   65  * The routines implemented in this file are described in:
   66  *      Leffler, et al.: The Design and Implementation of the 4.3BSD
   67  *          UNIX Operating System (Addison Welley, 1989)
   68  * on pages 62-63.
   69  *
   70  * Arguably, to simplify accounting operations, this mechanism should
   71  * be replaced by one in which an accounting log file (similar to /dev/klog)
   72  * is read by a user process, etc.  However, that has its own problems.
   73  */
   74 
   75 /*
   76  * Internal accounting functions.
   77  * The former's operation is described in Leffler, et al., and the latter
   78  * was provided by UCB with the 4.4BSD-Lite release
   79  */
   80 static comp_t   encode_comp_t(u_long, u_long);
   81 static void     acctwatch(void *);
   82 
   83 /*
   84  * Accounting callout used for periodic scheduling of acctwatch.
   85  */
   86 static struct   callout acctwatch_callout;
   87 
   88 /*
   89  * Accounting vnode pointer, saved vnode pointer, and flags for each.
   90  */
   91 static struct   vnode *acctp;
   92 static struct   ucred *acctcred;
   93 static int      acctflags;
   94 static struct   vnode *savacctp;
   95 static struct   ucred *savacctcred;
   96 static int      savacctflags;
   97 
   98 static struct mtx       acct_mtx;
   99 MTX_SYSINIT(acct, &acct_mtx, "accounting", MTX_DEF);
  100 
  101 /*
  102  * Values associated with enabling and disabling accounting
  103  */
  104 static int acctsuspend = 2;     /* stop accounting when < 2% free space left */
  105 SYSCTL_INT(_kern, OID_AUTO, acct_suspend, CTLFLAG_RW,
  106         &acctsuspend, 0, "percentage of free disk space below which accounting stops");
  107 
  108 static int acctresume = 4;      /* resume when free space risen to > 4% */
  109 SYSCTL_INT(_kern, OID_AUTO, acct_resume, CTLFLAG_RW,
  110         &acctresume, 0, "percentage of free disk space above which accounting resumes");
  111 
  112 static int acctchkfreq = 15;    /* frequency (in seconds) to check space */
  113 SYSCTL_INT(_kern, OID_AUTO, acct_chkfreq, CTLFLAG_RW,
  114         &acctchkfreq, 0, "frequency for checking the free space");
  115 
  116 /*
  117  * Accounting system call.  Written based on the specification and
  118  * previous implementation done by Mark Tinguely.
  119  *
  120  * MPSAFE
  121  */
  122 int
  123 acct(td, uap)
  124         struct thread *td;
  125         struct acct_args /* {
  126                 char *path;
  127         } */ *uap;
  128 {
  129         struct nameidata nd;
  130         int error, flags;
  131 
  132         /* Make sure that the caller is root. */
  133         error = suser(td);
  134         if (error)
  135                 return (error);
  136 
  137         mtx_lock(&Giant);
  138         /*
  139          * If accounting is to be started to a file, open that file for
  140          * appending and make sure it's a 'normal'.
  141          */
  142         if (uap->path != NULL) {
  143                 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, uap->path, td);
  144                 flags = FWRITE | O_APPEND;
  145                 error = vn_open(&nd, &flags, 0);
  146                 if (error)
  147                         goto done2;
  148                 NDFREE(&nd, NDF_ONLY_PNBUF);
  149 #ifdef MAC
  150                 error = mac_check_system_acct(td->td_ucred, nd.ni_vp);
  151                 if (error) {
  152                         vn_close(nd.ni_vp, flags, td->td_ucred, td);
  153                         goto done2;
  154                 }
  155 #endif
  156                 VOP_UNLOCK(nd.ni_vp, 0, td);
  157                 if (nd.ni_vp->v_type != VREG) {
  158                         vn_close(nd.ni_vp, flags, td->td_ucred, td);
  159                         error = EACCES;
  160                         goto done2;
  161                 }
  162 #ifdef MAC
  163         } else {
  164                 error = mac_check_system_acct(td->td_ucred, NULL);
  165                 if (error)
  166                         goto done2;
  167 #endif
  168         }
  169 
  170         /*
  171          * If accounting was previously enabled, kill the old space-watcher,
  172          * close the file, and (if no new file was specified, leave).
  173          */
  174 
  175         /*
  176          * XXX arr: Should not hold lock over vnode operation.
  177          */
  178 
  179         mtx_lock(&acct_mtx);
  180         if (acctp != NULLVP || savacctp != NULLVP) {
  181                 callout_stop(&acctwatch_callout);
  182                 error = vn_close((acctp != NULLVP ? acctp : savacctp),
  183                     (acctp != NULLVP ? acctflags : savacctflags),
  184                     (acctcred != NOCRED ? acctcred : savacctcred), td);
  185                 acctp = savacctp = NULLVP;
  186                 crfree(acctcred != NOCRED ? acctcred : savacctcred);
  187                 acctcred = savacctcred = NOCRED;
  188         }
  189         if (uap->path == NULL) {
  190                 mtx_unlock(&acct_mtx);
  191                 goto done2;
  192         }
  193 
  194         /*
  195          * Save the new accounting file vnode, and schedule the new
  196          * free space watcher.
  197          */
  198         acctp = nd.ni_vp;
  199         acctcred = crhold(td->td_ucred);
  200         acctflags = flags;
  201         callout_init(&acctwatch_callout, 0);
  202         mtx_unlock(&acct_mtx);
  203         acctwatch(NULL);
  204 done2:
  205         mtx_unlock(&Giant);
  206         return (error);
  207 }
  208 
  209 /*
  210  * Write out process accounting information, on process exit.
  211  * Data to be written out is specified in Leffler, et al.
  212  * and are enumerated below.  (They're also noted in the system
  213  * "acct.h" header file.)
  214  */
  215 
  216 int
  217 acct_process(td)
  218         struct thread *td;
  219 {
  220         struct proc *p = td->td_proc;
  221         struct acct acct;
  222         struct rusage *r;
  223         struct timeval ut, st, tmp;
  224         int t, ret;
  225         struct vnode *vp;
  226         struct ucred *uc;
  227 
  228         mtx_lock(&acct_mtx);
  229 
  230         /* If accounting isn't enabled, don't bother */
  231         vp = acctp;
  232         if (vp == NULLVP) {
  233                 mtx_unlock(&acct_mtx);
  234                 return (0);
  235         }
  236 
  237         /*
  238          * Get process accounting information.
  239          */
  240 
  241         PROC_LOCK(p);
  242         /* (1) The name of the command that ran */
  243         bcopy(p->p_comm, acct.ac_comm, sizeof acct.ac_comm);
  244 
  245         /* (2) The amount of user and system time that was used */
  246         mtx_lock_spin(&sched_lock);
  247         calcru(p, &ut, &st, NULL);
  248         mtx_unlock_spin(&sched_lock);
  249         acct.ac_utime = encode_comp_t(ut.tv_sec, ut.tv_usec);
  250         acct.ac_stime = encode_comp_t(st.tv_sec, st.tv_usec);
  251 
  252         /* (3) The elapsed time the command ran (and its starting time) */
  253         tmp = boottime;
  254         timevaladd(&tmp, &p->p_stats->p_start);
  255         acct.ac_btime = tmp.tv_sec;
  256         microuptime(&tmp);
  257         timevalsub(&tmp, &p->p_stats->p_start);
  258         acct.ac_etime = encode_comp_t(tmp.tv_sec, tmp.tv_usec);
  259 
  260         /* (4) The average amount of memory used */
  261         r = &p->p_stats->p_ru;
  262         tmp = ut;
  263         timevaladd(&tmp, &st);
  264         t = tmp.tv_sec * hz + tmp.tv_usec / tick;
  265         if (t)
  266                 acct.ac_mem = (r->ru_ixrss + r->ru_idrss + r->ru_isrss) / t;
  267         else
  268                 acct.ac_mem = 0;
  269 
  270         /* (5) The number of disk I/O operations done */
  271         acct.ac_io = encode_comp_t(r->ru_inblock + r->ru_oublock, 0);
  272 
  273         /* (6) The UID and GID of the process */
  274         acct.ac_uid = p->p_ucred->cr_ruid;
  275         acct.ac_gid = p->p_ucred->cr_rgid;
  276 
  277         /* (7) The terminal from which the process was started */
  278         SESS_LOCK(p->p_session);
  279         if ((p->p_flag & P_CONTROLT) && p->p_pgrp->pg_session->s_ttyp)
  280                 acct.ac_tty = dev2udev(p->p_pgrp->pg_session->s_ttyp->t_dev);
  281         else
  282                 acct.ac_tty = NOUDEV;
  283         SESS_UNLOCK(p->p_session);
  284 
  285         /* (8) The boolean flags that tell how the process terminated, etc. */
  286         acct.ac_flag = p->p_acflag;
  287         PROC_UNLOCK(p);
  288 
  289         /*
  290          * Write the accounting information to the file.
  291          */
  292         uc = crhold(acctcred);
  293         vref(vp);
  294         mtx_unlock(&acct_mtx);
  295 
  296         /*
  297          * Eliminate any file size rlimit.
  298          */
  299         if (p->p_limit->p_refcnt > 1) {
  300                 p->p_limit->p_refcnt--;
  301                 p->p_limit = limcopy(p->p_limit);
  302         }
  303         p->p_rlimit[RLIMIT_FSIZE].rlim_cur = RLIM_INFINITY;
  304 
  305         VOP_LEASE(vp, td, uc, LEASE_WRITE);
  306         ret = vn_rdwr(UIO_WRITE, vp, (caddr_t)&acct, sizeof (acct),
  307             (off_t)0, UIO_SYSSPACE, IO_APPEND|IO_UNIT, uc, NOCRED,
  308             (int *)0, td);
  309         vrele(vp);
  310         crfree(uc);
  311         return (ret);
  312 }
  313 
  314 /*
  315  * Encode_comp_t converts from ticks in seconds and microseconds
  316  * to ticks in 1/AHZ seconds.  The encoding is described in
  317  * Leffler, et al., on page 63.
  318  */
  319 
  320 #define MANTSIZE        13                      /* 13 bit mantissa. */
  321 #define EXPSIZE         3                       /* Base 8 (3 bit) exponent. */
  322 #define MAXFRACT        ((1 << MANTSIZE) - 1)   /* Maximum fractional value. */
  323 
  324 static comp_t
  325 encode_comp_t(s, us)
  326         u_long s, us;
  327 {
  328         int exp, rnd;
  329 
  330         exp = 0;
  331         rnd = 0;
  332         s *= AHZ;
  333         s += us / (1000000 / AHZ);      /* Maximize precision. */
  334 
  335         while (s > MAXFRACT) {
  336         rnd = s & (1 << (EXPSIZE - 1)); /* Round up? */
  337                 s >>= EXPSIZE;          /* Base 8 exponent == 3 bit shift. */
  338                 exp++;
  339         }
  340 
  341         /* If we need to round up, do it (and handle overflow correctly). */
  342         if (rnd && (++s > MAXFRACT)) {
  343                 s >>= EXPSIZE;
  344                 exp++;
  345         }
  346 
  347         /* Clean it up and polish it off. */
  348         exp <<= MANTSIZE;               /* Shift the exponent into place */
  349         exp += s;                       /* and add on the mantissa. */
  350         return (exp);
  351 }
  352 
  353 /*
  354  * Periodically check the filesystem to see if accounting
  355  * should be turned on or off.  Beware the case where the vnode
  356  * has been vgone()'d out from underneath us, e.g. when the file
  357  * system containing the accounting file has been forcibly unmounted.
  358  */
  359 /* ARGSUSED */
  360 static void
  361 acctwatch(a)
  362         void *a;
  363 {
  364         struct statfs sb;
  365 
  366         mtx_lock(&acct_mtx);
  367 
  368         /*
  369          * XXX arr: Need to fix the issue of holding acct_mtx over
  370          * the below vnode operations.
  371          */
  372 
  373         if (savacctp != NULLVP) {
  374                 if (savacctp->v_type == VBAD) {
  375                         (void) vn_close(savacctp, savacctflags, savacctcred,
  376                             NULL);
  377                         savacctp = NULLVP;
  378                         savacctcred = NOCRED;
  379                         mtx_unlock(&acct_mtx);
  380                         return;
  381                 }
  382                 (void)VFS_STATFS(savacctp->v_mount, &sb, (struct thread *)0);
  383                 if (sb.f_bavail > acctresume * sb.f_blocks / 100) {
  384                         acctp = savacctp;
  385                         acctcred = savacctcred;
  386                         acctflags = savacctflags;
  387                         savacctp = NULLVP;
  388                         savacctcred = NOCRED;
  389                         log(LOG_NOTICE, "Accounting resumed\n");
  390                 }
  391         } else {
  392                 if (acctp == NULLVP) {
  393                         mtx_unlock(&acct_mtx);
  394                         return;
  395                 }
  396                 if (acctp->v_type == VBAD) {
  397                         (void) vn_close(acctp, acctflags, acctcred, NULL);
  398                         acctp = NULLVP;
  399                         crfree(acctcred);
  400                         acctcred = NOCRED;
  401                         mtx_unlock(&acct_mtx);
  402                         return;
  403                 }
  404                 (void)VFS_STATFS(acctp->v_mount, &sb, (struct thread *)0);
  405                 if (sb.f_bavail <= acctsuspend * sb.f_blocks / 100) {
  406                         savacctp = acctp;
  407                         savacctflags = acctflags;
  408                         savacctcred = acctcred;
  409                         acctp = NULLVP;
  410                         acctcred = NOCRED;
  411                         log(LOG_NOTICE, "Accounting suspended\n");
  412                 }
  413         }
  414         callout_reset(&acctwatch_callout, acctchkfreq * hz, acctwatch, NULL);
  415         mtx_unlock(&acct_mtx);
  416 }

Cache object: 17fa6b13b3542e1d77472e274c760cb6


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]


This page is part of the FreeBSD/Linux Linux Kernel Cross-Reference, and was automatically generated using a modified version of the LXR engine.