bsd/kern/kern_acct.c

   1 /*
   2  * Copyright (c) 2000-2004 Apple Computer, Inc. All rights reserved.
   3  *
   4  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. The rights granted to you under the License
  10  * may not be used to create, or enable the creation or redistribution of,
  11  * unlawful or unlicensed copies of an Apple operating system, or to
  12  * circumvent, violate, or enable the circumvention or violation of, any
  13  * terms of an Apple operating system software license agreement.
  14  *
  15  * Please obtain a copy of the License at
  16  * http://www.opensource.apple.com/apsl/ and read it before using this file.
  17  *
  18  * The Original Code and all software distributed under the License are
  19  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  20  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  21  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  23  * Please see the License for the specific language governing rights and
  24  * limitations under the License.
  25  *
  26  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  27  */
  28 /* Copyright (c) 1995 NeXT Computer, Inc. All Rights Reserved */
  29 /*-
  30  * Copyright (c) 1982, 1986, 1989, 1993
  31  *      The Regents of the University of California.  All rights reserved.
  32  * (c) UNIX System Laboratories, Inc.
  33  * All or some portions of this file are derived from material licensed
  34  * to the University of California by American Telephone and Telegraph
  35  * Co. or Unix System Laboratories, Inc. and are reproduced herein with
  36  * the permission of UNIX System Laboratories, Inc.
  37  *
  38  * Redistribution and use in source and binary forms, with or without
  39  * modification, are permitted provided that the following conditions
  40  * are met:
  41  * 1. Redistributions of source code must retain the above copyright
  42  *    notice, this list of conditions and the following disclaimer.
  43  * 2. Redistributions in binary form must reproduce the above copyright
  44  *    notice, this list of conditions and the following disclaimer in the
  45  *    documentation and/or other materials provided with the distribution.
  46  * 3. All advertising materials mentioning features or use of this software
  47  *    must display the following acknowledgement:
  48  *      This product includes software developed by the University of
  49  *      California, Berkeley and its contributors.
  50  * 4. Neither the name of the University nor the names of its contributors
  51  *    may be used to endorse or promote products derived from this software
  52  *    without specific prior written permission.
  53  *
  54  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  55  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  56  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  57  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  58  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  59  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  60  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  61  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  62  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  63  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  64  * SUCH DAMAGE.
  65  *
  66  *      @(#)kern_acct.c 8.1 (Berkeley) 6/14/93
  67  */
  68 /* HISTORY
  69  * 08-May-95  Mac Gillon (mgillon) at NeXT
  70  *      Purged old history
  71  *      New version based on 4.4
  72  */
  73
  74
  75 #include <sys/param.h>
  76 #include <sys/proc_internal.h>
  77 #include <sys/kauth.h>
  78 #include <sys/mount_internal.h>
  79 #include <sys/vnode_internal.h>
  80 #include <sys/file_internal.h>
  81 #include <sys/syslog.h>
  82 #include <sys/kernel.h>
  83 #include <sys/namei.h>
  84 #include <sys/errno.h>
  85 #include <sys/acct.h>
  86 #include <sys/resourcevar.h>
  87 #include <sys/ioctl.h>
  88 #include <sys/tty.h>
  89 #include <sys/sysproto.h>
  90 #include <machine/spl.h>
  91
  92 /*
  93  * The routines implemented in this file are described in:
  94  *      Leffler, et al.: The Design and Implementation of the 4.3BSD
  95  *          UNIX Operating System (Addison Welley, 1989)
  96  * on pages 62-63.
  97  *
  98  * Arguably, to simplify accounting operations, this mechanism should
  99  * be replaced by one in which an accounting log file (similar to /dev/klog)
 100  * is read by a user process, etc.  However, that has its own problems.
 101  */
 102
 103 /*
 104  * Internal accounting functions.
 105  * The former's operation is described in Leffler, et al., and the latter
 106  * was provided by UCB with the 4.4BSD-Lite release
 107  */
 108 comp_t  encode_comp_t(u_long, u_long);
 109 void    acctwatch(void *);
 110 void    acctwatch_funnel(void *);
 111
 112 /*
 113  * Accounting vnode pointer, and suspended accounting vnode pointer.  States
 114  * are as follows:
 115  *
 116  *      acctp           suspend_acctp   state
 117  *      -------------   ------------    ------------------------------
 118  *      NULL            NULL            Accounting disabled
 119  *      !NULL           NULL            Accounting enabled
 120  *      NULL            !NULL           Accounting enabled, but suspended
 121  *      !NULL           !NULL           <not allowed>
 122  */
 123 struct  vnode *acctp;
 124 struct  vnode *suspend_acctp;
 125
 126 /*
 127  * Values associated with enabling and disabling accounting
 128  */
 129 int     acctsuspend = 2;        /* stop accounting when < 2% free space left */
 130 int     acctresume = 4;         /* resume when free space risen to > 4% */
 131 int     acctchkfreq = 15;       /* frequency (in seconds) to check space */
 132
 133 /*
 134  * Accounting system call.  Written based on the specification and
 135  * previous implementation done by Mark Tinguely.
 136  */
 137 int
 138 acct(struct proc *p, struct acct_args *uap, __unused int *retval)
 139 {
 140         struct nameidata nd;
 141         int error;
 142         struct vfs_context context;
 143
 144         context.vc_proc = p;
 145         context.vc_ucred = kauth_cred_get();
 146
 147         /* Make sure that the caller is root. */
 148         if ((error = suser(kauth_cred_get(), &p->p_acflag)))
 149                 return (error);
 150
 151         /*
 152          * If accounting is to be started to a file, open that file for
 153          * writing and make sure it's a 'normal'.
 154          */
 155         if (uap->path != USER_ADDR_NULL) {
 156                 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, uap->path, &context);
 157                 if ((error = vn_open(&nd, FWRITE, 0)))
 158                         return (error);
 159                 vnode_put(nd.ni_vp);
 160
 161                 if (nd.ni_vp->v_type != VREG) {
 162                         vn_close(nd.ni_vp, FWRITE, kauth_cred_get(), p);
 163                         return (EACCES);
 164                 }
 165         }
 166
 167         /*
 168          * If accounting was previously enabled, kill the old space-watcher,
 169          * close the file, and (if no new file was specified, leave).
 170          */
 171         if (acctp != NULLVP || suspend_acctp != NULLVP) {
 172                 untimeout(acctwatch_funnel, NULL);
 173                 error = vn_close((acctp != NULLVP ? acctp : suspend_acctp), FWRITE,
 174                     kauth_cred_get(), p);
 175
 176                 acctp = suspend_acctp = NULLVP;
 177         }
 178         if (uap->path == USER_ADDR_NULL)
 179                 return (error);
 180
 181         /*
 182          * Save the new accounting file vnode, and schedule the new
 183          * free space watcher.
 184          */
 185         acctp = nd.ni_vp;
 186         acctwatch(NULL);
 187         return (error);
 188 }
 189
 190 /*
 191  * Write out process accounting information, on process exit.
 192  * Data to be written out is specified in Leffler, et al.
 193  * and are enumerated below.  (They're also noted in the system
 194  * "acct.h" header file.)
 195  */
 196 int
 197 acct_process(p)
 198         struct proc *p;
 199 {
 200         struct acct an_acct;
 201         struct rusage *r;
 202         struct timeval ut, st, tmp;
 203         int t;
 204         int error;
 205         struct vnode *vp;
 206
 207         /* If accounting isn't enabled, don't bother */
 208         vp = acctp;
 209         if (vp == NULLVP)
 210                 return (0);
 211
 212         /*
 213          * Get process accounting information.
 214          */
 215
 216         /* (1) The name of the command that ran */
 217         bcopy(p->p_comm, an_acct.ac_comm, sizeof an_acct.ac_comm);
 218
 219         /* (2) The amount of user and system time that was used */
 220         calcru(p, &ut, &st, NULL);
 221         an_acct.ac_utime = encode_comp_t(ut.tv_sec, ut.tv_usec);
 222         an_acct.ac_stime = encode_comp_t(st.tv_sec, st.tv_usec);
 223
 224         /* (3) The elapsed time the commmand ran (and its starting time) */
 225         an_acct.ac_btime = p->p_stats->p_start.tv_sec;
 226         microtime(&tmp);
 227         timevalsub(&tmp, &p->p_stats->p_start);
 228         an_acct.ac_etime = encode_comp_t(tmp.tv_sec, tmp.tv_usec);
 229
 230         /* (4) The average amount of memory used */
 231         r = &p->p_stats->p_ru;
 232         tmp = ut;
 233         timevaladd(&tmp, &st);
 234         t = tmp.tv_sec * hz + tmp.tv_usec / tick;
 235         if (t)
 236                 an_acct.ac_mem = (r->ru_ixrss + r->ru_idrss + r->ru_isrss) / t;
 237         else
 238                 an_acct.ac_mem = 0;
 239
 240         /* (5) The number of disk I/O operations done */
 241         an_acct.ac_io = encode_comp_t(r->ru_inblock + r->ru_oublock, 0);
 242
 243         /* (6) The UID and GID of the process */
 244         an_acct.ac_uid = p->p_ucred->cr_ruid;
 245         an_acct.ac_gid = p->p_ucred->cr_rgid;
 246
 247         /* (7) The terminal from which the process was started */
 248         if ((p->p_flag & P_CONTROLT) && p->p_pgrp->pg_session->s_ttyp)
 249                 an_acct.ac_tty = p->p_pgrp->pg_session->s_ttyp->t_dev;
 250         else
 251                 an_acct.ac_tty = NODEV;
 252
 253         /* (8) The boolean flags that tell how the process terminated, etc. */
 254         an_acct.ac_flag = p->p_acflag;
 255
 256         /*
 257          * Now, just write the accounting information to the file.
 258          */
 259         if ((error = vnode_getwithref(vp)) == 0) {
 260                 error = vn_rdwr(UIO_WRITE, vp, (caddr_t)&an_acct, sizeof (an_acct),
 261                                 (off_t)0, UIO_SYSSPACE32, IO_APPEND|IO_UNIT, p->p_ucred,
 262                                 (int *)0, p);
 263                 vnode_put(vp);
 264         }
 265         return (error);
 266 }
 267
 268 /*
 269  * Encode_comp_t converts from ticks in seconds and microseconds
 270  * to ticks in 1/AHZ seconds.  The encoding is described in
 271  * Leffler, et al., on page 63.
 272  */
 273
 274 #define MANTSIZE        13                      /* 13 bit mantissa. */
 275 #define EXPSIZE         3                       /* Base 8 (3 bit) exponent. */
 276 #define MAXFRACT        ((1 << MANTSIZE) - 1)   /* Maximum fractional value. */
 277
 278 comp_t
 279 encode_comp_t(s, us)
 280         u_long s, us;
 281 {
 282         int exp, rnd;
 283
 284         exp = 0;
 285         rnd = 0;
 286         s *= AHZ;
 287         s += us / (1000000 / AHZ);      /* Maximize precision. */
 288
 289         while (s > MAXFRACT) {
 290         rnd = s & (1 << (EXPSIZE - 1)); /* Round up? */
 291                 s >>= EXPSIZE;          /* Base 8 exponent == 3 bit shift. */
 292                 exp++;
 293         }
 294
 295         /* If we need to round up, do it (and handle overflow correctly). */
 296         if (rnd && (++s > MAXFRACT)) {
 297                 s >>= EXPSIZE;
 298                 exp++;
 299         }
 300
 301         /* Clean it up and polish it off. */
 302         exp <<= MANTSIZE;               /* Shift the exponent into place */
 303         exp += s;                       /* and add on the mantissa. */
 304         return (exp);
 305 }
 306
 307 void
 308 acctwatch_funnel(a)
 309         void *a;
 310 {
 311         thread_funnel_set(kernel_flock, TRUE);
 312         acctwatch(a);
 313         thread_funnel_set(kernel_flock, FALSE);
 314 }
 315
 316
 317 /*
 318  * Periodically check the file system to see if accounting
 319  * should be turned on or off.  Beware the case where the vnode
 320  * has been vgone()'d out from underneath us, e.g. when the file
 321  * system containing the accounting file has been forcibly unmounted.
 322  */
 323 /* ARGSUSED */
 324 void
 325 acctwatch(__unused void *a)
 326 {
 327         struct vfs_context context;
 328         struct vfs_attr va;
 329
 330         VFSATTR_INIT(&va);
 331         VFSATTR_WANTED(&va, f_blocks);
 332         VFSATTR_WANTED(&va, f_bavail);
 333         context.vc_proc = current_proc();
 334         context.vc_ucred = kauth_cred_get();
 335
 336         if (suspend_acctp != NULLVP) {
 337                 /*
 338                  * Resuming accounting when accounting is suspended, and the
 339                  * filesystem containing the suspended accounting file goes
 340                  * below a low watermark
 341                  */
 342                 if (suspend_acctp->v_type == VBAD) {
 343                         (void) vn_close(suspend_acctp, FWRITE, NOCRED, NULL);
 344                         suspend_acctp = NULLVP;
 345                         return;
 346                 }
 347                 (void)vfs_getattr(suspend_acctp->v_mount, &va, &context);
 348                 if (va.f_bavail > acctresume * va.f_blocks / 100) {
 349                         acctp = suspend_acctp;
 350                         suspend_acctp = NULLVP;
 351                         log(LOG_NOTICE, "Accounting resumed\n");
 352                 }
 353         } else if (acctp != NULLVP) {
 354                 /*
 355                  * Suspending accounting when accounting is currently active,
 356                  * and the filesystem containing the active accounting file
 357                  * goes over a high watermark
 358                  */
 359                 if (acctp->v_type == VBAD) {
 360                         (void) vn_close(acctp, FWRITE, NOCRED, NULL);
 361                         acctp = NULLVP;
 362                         return;
 363                 }
 364                 (void)vfs_getattr(acctp->v_mount, &va, &context);
 365                 if (va.f_bavail <= acctsuspend * va.f_blocks / 100) {
 366                         suspend_acctp = acctp;
 367                         acctp = NULLVP;
 368                         log(LOG_NOTICE, "Accounting suspended\n");
 369                 }
 370         } else {
 371                 return;
 372         }
 373
 374         timeout(acctwatch_funnel, NULL, acctchkfreq * hz);
 375 }