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