bsd/kern/kern_acct.c

   1 /*
   2  * Copyright (c) 2000-2010 Apple 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  * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
  75  * support for mandatory and extensible security protections.  This notice
  76  * is included in support of clause 2.2 (b) of the Apple Public License,
  77  * Version 2.0.
  78  */
  79
  80
  81 #include <sys/param.h>
  82 #include <sys/proc_internal.h>
  83 #include <sys/kauth.h>
  84 #include <sys/mount_internal.h>
  85 #include <sys/vnode_internal.h>
  86 #include <sys/file_internal.h>
  87 #include <sys/syslog.h>
  88 #include <sys/kernel.h>
  89 #include <sys/namei.h>
  90 #include <sys/errno.h>
  91 #include <sys/acct.h>
  92 #include <sys/resourcevar.h>
  93 #include <sys/ioctl.h>
  94 #include <sys/tty.h>
  95 #include <sys/sysproto.h>
  96 #if CONFIG_MACF
  97 #include <security/mac_framework.h>
  98 #endif
  99
 100 /*
 101  * The routines implemented in this file are described in:
 102  *      Leffler, et al.: The Design and Implementation of the 4.3BSD
 103  *          UNIX Operating System (Addison Welley, 1989)
 104  * on pages 62-63.
 105  *
 106  * Arguably, to simplify accounting operations, this mechanism should
 107  * be replaced by one in which an accounting log file (similar to /dev/klog)
 108  * is read by a user process, etc.  However, that has its own problems.
 109  */
 110
 111 /*
 112  * Internal accounting functions.
 113  * The former's operation is described in Leffler, et al., and the latter
 114  * was provided by UCB with the 4.4BSD-Lite release
 115  */
 116 comp_t  encode_comp_t(uint32_t, uint32_t);
 117 void    acctwatch(void *);
 118 void    acct_init(void);
 119
 120 /*
 121  * Accounting vnode pointer, and suspended accounting vnode pointer.  States
 122  * are as follows:
 123  *
 124  *      acctp           suspend_acctp   state
 125  *      -------------   ------------    ------------------------------
 126  *      NULL            NULL            Accounting disabled
 127  *      !NULL           NULL            Accounting enabled
 128  *      NULL            !NULL           Accounting enabled, but suspended
 129  *      !NULL           !NULL           <not allowed>
 130  */
 131 struct  vnode *acctp;
 132 struct  vnode *suspend_acctp;
 133
 134 /*
 135  * Values associated with enabling and disabling accounting
 136  */
 137 int     acctsuspend = 2;        /* stop accounting when < 2% free space left */
 138 int     acctresume = 4;         /* resume when free space risen to > 4% */
 139 int     acctchkfreq = 15;       /* frequency (in seconds) to check space */
 140
 141
 142 static lck_grp_t       *acct_subsys_lck_grp;
 143 static lck_mtx_t       *acct_subsys_mutex;
 144
 145 #define ACCT_SUBSYS_LOCK() lck_mtx_lock(acct_subsys_mutex)
 146 #define ACCT_SUBSYS_UNLOCK() lck_mtx_unlock(acct_subsys_mutex)
 147
 148 void
 149 acct_init(void)
 150 {
 151         acct_subsys_lck_grp = lck_grp_alloc_init("acct", NULL);
 152         acct_subsys_mutex = lck_mtx_alloc_init(acct_subsys_lck_grp, NULL);
 153 }
 154
 155
 156 /*
 157  * Accounting system call.  Written based on the specification and
 158  * previous implementation done by Mark Tinguely.
 159  */
 160 int
 161 acct(proc_t p, struct acct_args *uap, __unused int *retval)
 162 {
 163         struct nameidata nd;
 164         int error;
 165         struct vfs_context *ctx;
 166
 167         ctx = vfs_context_current();
 168
 169         /* Make sure that the caller is root. */
 170         if ((error = suser(vfs_context_ucred(ctx), &p->p_acflag)))
 171                 return (error);
 172
 173         /*
 174          * If accounting is to be started to a file, open that file for
 175          * writing and make sure it's a 'normal'.
 176          */
 177         if (uap->path != USER_ADDR_NULL) {
 178                 NDINIT(&nd, LOOKUP, OP_OPEN, NOFOLLOW, UIO_USERSPACE, uap->path, ctx);
 179                 if ((error = vn_open(&nd, FWRITE, 0)))
 180                         return (error);
 181 #if CONFIG_MACF
 182                 error = mac_system_check_acct(vfs_context_ucred(ctx), nd.ni_vp);
 183                 if (error) {
 184                         vnode_put(nd.ni_vp);
 185                         vn_close(nd.ni_vp, FWRITE, ctx);
 186                         return (error);
 187                 }
 188 #endif
 189                 vnode_put(nd.ni_vp);
 190
 191                 if (nd.ni_vp->v_type != VREG) {
 192                         vn_close(nd.ni_vp, FWRITE, ctx);
 193                         return (EACCES);
 194                 }
 195         }
 196 #if CONFIG_MACF
 197         else {
 198                 error = mac_system_check_acct(vfs_context_ucred(ctx), NULL);
 199                 if (error)
 200                         return (error);
 201         }
 202 #endif
 203
 204         /*
 205          * If accounting was previously enabled, kill the old space-watcher,
 206          * close the file, and (if no new file was specified, leave).
 207          */
 208         ACCT_SUBSYS_LOCK();
 209         if (acctp != NULLVP || suspend_acctp != NULLVP) {
 210                 untimeout(acctwatch, NULL);
 211                 error = vn_close((acctp != NULLVP ? acctp : suspend_acctp),
 212                                 FWRITE, vfs_context_current());
 213
 214                 acctp = suspend_acctp = NULLVP;
 215         }
 216         if (uap->path == USER_ADDR_NULL) {
 217                 ACCT_SUBSYS_UNLOCK();
 218                 return (error);
 219         }
 220
 221         /*
 222          * Save the new accounting file vnode, and schedule the new
 223          * free space watcher.
 224          */
 225         acctp = nd.ni_vp;
 226         ACCT_SUBSYS_UNLOCK();
 227
 228         acctwatch(NULL);
 229         return (error);
 230 }
 231
 232 /*
 233  * Write out process accounting information, on process exit.
 234  * Data to be written out is specified in Leffler, et al.
 235  * and are enumerated below.  (They're also noted in the system
 236  * "acct.h" header file.)
 237  */
 238 int
 239 acct_process(proc_t p)
 240 {
 241         struct acct an_acct;
 242         struct rusage rup, *r;
 243         struct timeval ut, st, tmp;
 244         int t;
 245         int error;
 246         struct vnode *vp;
 247         kauth_cred_t safecred;
 248         struct session * sessp;
 249         struct  tty *tp;
 250
 251         /* If accounting isn't enabled, don't bother */
 252         ACCT_SUBSYS_LOCK();
 253         vp = acctp;
 254         if (vp == NULLVP) {
 255                 ACCT_SUBSYS_UNLOCK();
 256                 return (0);
 257         }
 258
 259         /*
 260          * Get process accounting information.
 261          */
 262
 263         /* (1) The name of the command that ran */
 264         bcopy(p->p_comm, an_acct.ac_comm, sizeof an_acct.ac_comm);
 265
 266         /* (2) The amount of user and system time that was used */
 267         calcru(p, &ut, &st, NULL);
 268         an_acct.ac_utime = encode_comp_t(ut.tv_sec, ut.tv_usec);
 269         an_acct.ac_stime = encode_comp_t(st.tv_sec, st.tv_usec);
 270
 271         /* (3) The elapsed time the commmand ran (and its starting time) */
 272         an_acct.ac_btime = p->p_start.tv_sec;
 273         microtime(&tmp);
 274         timevalsub(&tmp, &p->p_start);
 275         an_acct.ac_etime = encode_comp_t(tmp.tv_sec, tmp.tv_usec);
 276
 277         /* (4) The average amount of memory used */
 278         proc_lock(p);
 279         rup = p->p_stats->p_ru;
 280         proc_unlock(p);
 281         r = &rup;
 282         tmp = ut;
 283         timevaladd(&tmp, &st);
 284         t = tmp.tv_sec * hz + tmp.tv_usec / tick;
 285         if (t)
 286                 an_acct.ac_mem = (r->ru_ixrss + r->ru_idrss + r->ru_isrss) / t;
 287         else
 288                 an_acct.ac_mem = 0;
 289
 290         /* (5) The number of disk I/O operations done */
 291         an_acct.ac_io = encode_comp_t(r->ru_inblock + r->ru_oublock, 0);
 292
 293         /* (6) The UID and GID of the process */
 294         safecred = kauth_cred_proc_ref(p);
 295
 296         an_acct.ac_uid = kauth_cred_getruid(safecred);
 297         an_acct.ac_gid = kauth_cred_getrgid(safecred);
 298
 299         /* (7) The terminal from which the process was started */
 300
 301         sessp = proc_session(p);
 302         if ((p->p_flag & P_CONTROLT) && (sessp != SESSION_NULL) && ((tp = SESSION_TP(sessp)) != TTY_NULL)) {
 303                 tty_lock(tp);
 304                 an_acct.ac_tty = tp->t_dev;
 305                 tty_unlock(tp);
 306          }else
 307                 an_acct.ac_tty = NODEV;
 308
 309         if (sessp != SESSION_NULL)
 310                 session_rele(sessp);
 311
 312         /* (8) The boolean flags that tell how the process terminated, etc. */
 313         an_acct.ac_flag = p->p_acflag;
 314
 315         /*
 316          * Now, just write the accounting information to the file.
 317          */
 318         if ((error = vnode_getwithref(vp)) == 0) {
 319                 error = vn_rdwr(UIO_WRITE, vp, (caddr_t)&an_acct, sizeof (an_acct),
 320                                 (off_t)0, UIO_SYSSPACE, IO_APPEND|IO_UNIT, safecred,
 321                                 (int *)0, p);
 322                 vnode_put(vp);
 323         }
 324
 325         kauth_cred_unref(&safecred);
 326         ACCT_SUBSYS_UNLOCK();
 327
 328         return (error);
 329 }
 330
 331 /*
 332  * Encode_comp_t converts from ticks in seconds and microseconds
 333  * to ticks in 1/AHZ seconds.  The encoding is described in
 334  * Leffler, et al., on page 63.
 335  */
 336
 337 #define MANTSIZE        13                      /* 13 bit mantissa. */
 338 #define EXPSIZE         3                       /* Base 8 (3 bit) exponent. */
 339 #define MAXFRACT        ((1 << MANTSIZE) - 1)   /* Maximum fractional value. */
 340
 341 comp_t
 342 encode_comp_t(uint32_t s, uint32_t us)
 343 {
 344         int exp, rnd;
 345
 346         exp = 0;
 347         rnd = 0;
 348         s *= AHZ;
 349         s += us / (1000000 / AHZ);      /* Maximize precision. */
 350
 351         while (s > MAXFRACT) {
 352         rnd = s & (1 << (EXPSIZE - 1)); /* Round up? */
 353                 s >>= EXPSIZE;          /* Base 8 exponent == 3 bit shift. */
 354                 exp++;
 355         }
 356
 357         /* If we need to round up, do it (and handle overflow correctly). */
 358         if (rnd && (++s > MAXFRACT)) {
 359                 s >>= EXPSIZE;
 360                 exp++;
 361         }
 362
 363         /* Clean it up and polish it off. */
 364         exp <<= MANTSIZE;               /* Shift the exponent into place */
 365         exp += s;                       /* and add on the mantissa. */
 366         return (exp);
 367 }
 368
 369 /*
 370  * Periodically check the file system to see if accounting
 371  * should be turned on or off.  Beware the case where the vnode
 372  * has been vgone()'d out from underneath us, e.g. when the file
 373  * system containing the accounting file has been forcibly unmounted.
 374  */
 375 /* ARGSUSED */
 376 void
 377 acctwatch(__unused void *a)
 378 {
 379         vfs_context_t ctx = vfs_context_current();
 380         struct vfs_attr va;
 381
 382         VFSATTR_INIT(&va);
 383         VFSATTR_WANTED(&va, f_blocks);
 384         VFSATTR_WANTED(&va, f_bavail);
 385
 386         ACCT_SUBSYS_LOCK();
 387         if (suspend_acctp != NULLVP) {
 388                 /*
 389                  * Resuming accounting when accounting is suspended, and the
 390                  * filesystem containing the suspended accounting file goes
 391                  * below a low watermark
 392                  */
 393                 if (suspend_acctp->v_type == VBAD) {
 394                         (void) vn_close(suspend_acctp, FWRITE, vfs_context_kernel());
 395                         suspend_acctp = NULLVP;
 396                         ACCT_SUBSYS_UNLOCK();
 397                         return;
 398                 }
 399                 (void)vfs_getattr(suspend_acctp->v_mount, &va, ctx);
 400                 if (va.f_bavail > acctresume * va.f_blocks / 100) {
 401                         acctp = suspend_acctp;
 402                         suspend_acctp = NULLVP;
 403                         log(LOG_NOTICE, "Accounting resumed\n");
 404                 }
 405         } else if (acctp != NULLVP) {
 406                 /*
 407                  * Suspending accounting when accounting is currently active,
 408                  * and the filesystem containing the active accounting file
 409                  * goes over a high watermark
 410                  */
 411                 if (acctp->v_type == VBAD) {
 412                         (void) vn_close(acctp, FWRITE, vfs_context_kernel());
 413                         acctp = NULLVP;
 414                         ACCT_SUBSYS_UNLOCK();
 415                         return;
 416                 }
 417                 (void)vfs_getattr(acctp->v_mount, &va, ctx);
 418                 if (va.f_bavail <= acctsuspend * va.f_blocks / 100) {
 419                         suspend_acctp = acctp;
 420                         acctp = NULLVP;
 421                         log(LOG_NOTICE, "Accounting suspended\n");
 422                 }
 423         } else {
 424                 ACCT_SUBSYS_UNLOCK();
 425                 return;
 426         }
 427         ACCT_SUBSYS_UNLOCK();
 428
 429         timeout(acctwatch, NULL, acctchkfreq * hz);
 430 }