Home | History | Annotate | Line # | Download | only in kern
init_sysctl.c revision 1.125.2.1
      1 /*	$NetBSD: init_sysctl.c,v 1.125.2.1 2008/03/29 20:47:00 christos Exp $ */
      2 
      3 /*-
      4  * Copyright (c) 2003, 2007, 2008 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by Andrew Brown, and by Andrew Doran.
      9  *
     10  * Redistribution and use in source and binary forms, with or without
     11  * modification, are permitted provided that the following conditions
     12  * are met:
     13  * 1. Redistributions of source code must retain the above copyright
     14  *    notice, this list of conditions and the following disclaimer.
     15  * 2. Redistributions in binary form must reproduce the above copyright
     16  *    notice, this list of conditions and the following disclaimer in the
     17  *    documentation and/or other materials provided with the distribution.
     18  * 3. All advertising materials mentioning features or use of this software
     19  *    must display the following acknowledgement:
     20  *	This product includes software developed by the NetBSD
     21  *	Foundation, Inc. and its contributors.
     22  * 4. Neither the name of The NetBSD Foundation nor the names of its
     23  *    contributors may be used to endorse or promote products derived
     24  *    from this software without specific prior written permission.
     25  *
     26  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     27  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     28  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     29  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     30  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     31  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     32  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     33  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     34  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     35  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     36  * POSSIBILITY OF SUCH DAMAGE.
     37  */
     38 
     39 #include <sys/cdefs.h>
     40 __KERNEL_RCSID(0, "$NetBSD: init_sysctl.c,v 1.125.2.1 2008/03/29 20:47:00 christos Exp $");
     41 
     42 #include "opt_sysv.h"
     43 #include "opt_posix.h"
     44 #include "opt_compat_netbsd32.h"
     45 #include "opt_compat_netbsd.h"
     46 #include "pty.h"
     47 #include "rnd.h"
     48 
     49 #include <sys/types.h>
     50 #include <sys/param.h>
     51 #include <sys/sysctl.h>
     52 #include <sys/cpu.h>
     53 #include <sys/errno.h>
     54 #include <sys/systm.h>
     55 #include <sys/kernel.h>
     56 #include <sys/unistd.h>
     57 #include <sys/disklabel.h>
     58 #include <sys/rnd.h>
     59 #include <sys/vnode.h>
     60 #include <sys/mount.h>
     61 #include <sys/namei.h>
     62 #include <sys/msgbuf.h>
     63 #include <dev/cons.h>
     64 #include <sys/socketvar.h>
     65 #include <sys/file.h>
     66 #include <sys/filedesc.h>
     67 #include <sys/tty.h>
     68 #include <sys/malloc.h>
     69 #include <sys/resource.h>
     70 #include <sys/resourcevar.h>
     71 #include <sys/exec.h>
     72 #include <sys/conf.h>
     73 #include <sys/device.h>
     74 #include <sys/stat.h>
     75 #include <sys/kauth.h>
     76 #include <sys/ktrace.h>
     77 
     78 #ifdef COMPAT_NETBSD32
     79 #include <compat/netbsd32/netbsd32.h>
     80 #endif
     81 #ifdef COMPAT_50
     82 #include <compat/sys/time.h>
     83 #endif
     84 
     85 #include <sys/cpu.h>
     86 
     87 /* XXX this should not be here */
     88 int security_setidcore_dump;
     89 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
     90 uid_t security_setidcore_owner = 0;
     91 gid_t security_setidcore_group = 0;
     92 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
     93 
     94 static const u_int sysctl_flagmap[] = {
     95 	PK_ADVLOCK, P_ADVLOCK,
     96 	PK_EXEC, P_EXEC,
     97 	PK_NOCLDWAIT, P_NOCLDWAIT,
     98 	PK_32, P_32,
     99 	PK_CLDSIGIGN, P_CLDSIGIGN,
    100 	PK_SUGID, P_SUGID,
    101 	0
    102 };
    103 
    104 static const u_int sysctl_sflagmap[] = {
    105 	PS_NOCLDSTOP, P_NOCLDSTOP,
    106 	PS_PPWAIT, P_PPWAIT,
    107 	PS_WEXIT, P_WEXIT,
    108 	PS_STOPFORK, P_STOPFORK,
    109 	PS_STOPEXEC, P_STOPEXEC,
    110 	PS_STOPEXIT, P_STOPEXIT,
    111 	0
    112 };
    113 
    114 static const u_int sysctl_slflagmap[] = {
    115 	PSL_TRACED, P_TRACED,
    116 	PSL_FSTRACE, P_FSTRACE,
    117 	PSL_CHTRACED, P_CHTRACED,
    118 	PSL_SYSCALL, P_SYSCALL,
    119 	0
    120 };
    121 
    122 static const u_int sysctl_lflagmap[] = {
    123 	PL_CONTROLT, P_CONTROLT,
    124 	0
    125 };
    126 
    127 static const u_int sysctl_stflagmap[] = {
    128 	PST_PROFIL, P_PROFIL,
    129 	0
    130 
    131 };
    132 
    133 static const u_int sysctl_lwpflagmap[] = {
    134 	LW_INMEM, P_INMEM,
    135 	LW_SINTR, P_SINTR,
    136 	LW_SYSTEM, P_SYSTEM,
    137 	0
    138 };
    139 
    140 static const u_int sysctl_lwpprflagmap[] = {
    141 	LPR_DETACHED, L_DETACHED,
    142 	0
    143 };
    144 
    145 /*
    146  * try over estimating by 5 procs/lwps
    147  */
    148 #define KERN_PROCSLOP	(5 * sizeof(struct kinfo_proc))
    149 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
    150 
    151 static int dcopyout(struct lwp *, const void *, void *, size_t);
    152 
    153 static int
    154 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
    155 {
    156 	int error;
    157 
    158 	error = copyout(kaddr, uaddr, len);
    159 	ktrmibio(-1, UIO_READ, uaddr, len, error);
    160 
    161 	return error;
    162 }
    163 
    164 #ifdef DIAGNOSTIC
    165 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
    166 #endif
    167 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
    168 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
    169 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
    170 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
    171 static int sysctl_setlen(SYSCTLFN_PROTO);
    172 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
    173 static int sysctl_kern_file(SYSCTLFN_PROTO);
    174 static int sysctl_msgbuf(SYSCTLFN_PROTO);
    175 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
    176 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
    177 #if NPTY > 0
    178 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
    179 #endif /* NPTY > 0 */
    180 static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
    181 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
    182 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
    183 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
    184 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
    185 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
    186 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
    187 static int sysctl_kern_file2(SYSCTLFN_PROTO);
    188 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
    189 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
    190 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
    191 static int sysctl_doeproc(SYSCTLFN_PROTO);
    192 static int sysctl_kern_proc_args(SYSCTLFN_PROTO);
    193 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
    194 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
    195 
    196 static u_int sysctl_map_flags(const u_int *, u_int);
    197 static void fill_kproc2(struct proc *, struct kinfo_proc2 *);
    198 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
    199 static void fill_file(struct kinfo_file *, const file_t *, const fdfile_t *,
    200 		      int, pid_t);
    201 
    202 /*
    203  * ********************************************************************
    204  * section 1: setup routines
    205  * ********************************************************************
    206  * These functions are stuffed into a link set for sysctl setup
    207  * functions. They're never called or referenced from anywhere else.
    208  * ********************************************************************
    209  */
    210 
    211 /*
    212  * sets up the base nodes...
    213  */
    214 SYSCTL_SETUP(sysctl_root_setup, "sysctl base setup")
    215 {
    216 
    217 	sysctl_createv(clog, 0, NULL, NULL,
    218 		       CTLFLAG_PERMANENT,
    219 		       CTLTYPE_NODE, "kern",
    220 		       SYSCTL_DESCR("High kernel"),
    221 		       NULL, 0, NULL, 0,
    222 		       CTL_KERN, CTL_EOL);
    223 	sysctl_createv(clog, 0, NULL, NULL,
    224 		       CTLFLAG_PERMANENT,
    225 		       CTLTYPE_NODE, "vm",
    226 		       SYSCTL_DESCR("Virtual memory"),
    227 		       NULL, 0, NULL, 0,
    228 		       CTL_VM, CTL_EOL);
    229 	sysctl_createv(clog, 0, NULL, NULL,
    230 		       CTLFLAG_PERMANENT,
    231 		       CTLTYPE_NODE, "vfs",
    232 		       SYSCTL_DESCR("Filesystem"),
    233 		       NULL, 0, NULL, 0,
    234 		       CTL_VFS, CTL_EOL);
    235 	sysctl_createv(clog, 0, NULL, NULL,
    236 		       CTLFLAG_PERMANENT,
    237 		       CTLTYPE_NODE, "net",
    238 		       SYSCTL_DESCR("Networking"),
    239 		       NULL, 0, NULL, 0,
    240 		       CTL_NET, CTL_EOL);
    241 	sysctl_createv(clog, 0, NULL, NULL,
    242 		       CTLFLAG_PERMANENT,
    243 		       CTLTYPE_NODE, "debug",
    244 		       SYSCTL_DESCR("Debugging"),
    245 		       NULL, 0, NULL, 0,
    246 		       CTL_DEBUG, CTL_EOL);
    247 	sysctl_createv(clog, 0, NULL, NULL,
    248 		       CTLFLAG_PERMANENT,
    249 		       CTLTYPE_NODE, "hw",
    250 		       SYSCTL_DESCR("Generic CPU, I/O"),
    251 		       NULL, 0, NULL, 0,
    252 		       CTL_HW, CTL_EOL);
    253 	sysctl_createv(clog, 0, NULL, NULL,
    254 		       CTLFLAG_PERMANENT,
    255 		       CTLTYPE_NODE, "machdep",
    256 		       SYSCTL_DESCR("Machine dependent"),
    257 		       NULL, 0, NULL, 0,
    258 		       CTL_MACHDEP, CTL_EOL);
    259 	/*
    260 	 * this node is inserted so that the sysctl nodes in libc can
    261 	 * operate.
    262 	 */
    263 	sysctl_createv(clog, 0, NULL, NULL,
    264 		       CTLFLAG_PERMANENT,
    265 		       CTLTYPE_NODE, "user",
    266 		       SYSCTL_DESCR("User-level"),
    267 		       NULL, 0, NULL, 0,
    268 		       CTL_USER, CTL_EOL);
    269 	sysctl_createv(clog, 0, NULL, NULL,
    270 		       CTLFLAG_PERMANENT,
    271 		       CTLTYPE_NODE, "ddb",
    272 		       SYSCTL_DESCR("In-kernel debugger"),
    273 		       NULL, 0, NULL, 0,
    274 		       CTL_DDB, CTL_EOL);
    275 	sysctl_createv(clog, 0, NULL, NULL,
    276 		       CTLFLAG_PERMANENT,
    277 		       CTLTYPE_NODE, "proc",
    278 		       SYSCTL_DESCR("Per-process"),
    279 		       NULL, 0, NULL, 0,
    280 		       CTL_PROC, CTL_EOL);
    281 	sysctl_createv(clog, 0, NULL, NULL,
    282 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    283 		       CTLTYPE_NODE, "vendor",
    284 		       SYSCTL_DESCR("Vendor specific"),
    285 		       NULL, 0, NULL, 0,
    286 		       CTL_VENDOR, CTL_EOL);
    287 	sysctl_createv(clog, 0, NULL, NULL,
    288 		       CTLFLAG_PERMANENT,
    289 		       CTLTYPE_NODE, "emul",
    290 		       SYSCTL_DESCR("Emulation settings"),
    291 		       NULL, 0, NULL, 0,
    292 		       CTL_EMUL, CTL_EOL);
    293 	sysctl_createv(clog, 0, NULL, NULL,
    294 		       CTLFLAG_PERMANENT,
    295 		       CTLTYPE_NODE, "security",
    296 		       SYSCTL_DESCR("Security"),
    297 		       NULL, 0, NULL, 0,
    298 		       CTL_SECURITY, CTL_EOL);
    299 }
    300 
    301 /*
    302  * this setup routine is a replacement for kern_sysctl()
    303  */
    304 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
    305 {
    306 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
    307 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
    308 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
    309 	const struct sysctlnode *rnode;
    310 
    311 	sysctl_createv(clog, 0, NULL, NULL,
    312 		       CTLFLAG_PERMANENT,
    313 		       CTLTYPE_NODE, "kern", NULL,
    314 		       NULL, 0, NULL, 0,
    315 		       CTL_KERN, CTL_EOL);
    316 
    317 	sysctl_createv(clog, 0, NULL, NULL,
    318 		       CTLFLAG_PERMANENT,
    319 		       CTLTYPE_STRING, "ostype",
    320 		       SYSCTL_DESCR("Operating system type"),
    321 		       NULL, 0, &ostype, 0,
    322 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
    323 	sysctl_createv(clog, 0, NULL, NULL,
    324 		       CTLFLAG_PERMANENT,
    325 		       CTLTYPE_STRING, "osrelease",
    326 		       SYSCTL_DESCR("Operating system release"),
    327 		       NULL, 0, &osrelease, 0,
    328 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
    329 	sysctl_createv(clog, 0, NULL, NULL,
    330 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    331 		       CTLTYPE_INT, "osrevision",
    332 		       SYSCTL_DESCR("Operating system revision"),
    333 		       NULL, __NetBSD_Version__, NULL, 0,
    334 		       CTL_KERN, KERN_OSREV, CTL_EOL);
    335 	sysctl_createv(clog, 0, NULL, NULL,
    336 		       CTLFLAG_PERMANENT,
    337 		       CTLTYPE_STRING, "version",
    338 		       SYSCTL_DESCR("Kernel version"),
    339 		       NULL, 0, &version, 0,
    340 		       CTL_KERN, KERN_VERSION, CTL_EOL);
    341 	sysctl_createv(clog, 0, NULL, NULL,
    342 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    343 		       CTLTYPE_INT, "maxvnodes",
    344 		       SYSCTL_DESCR("Maximum number of vnodes"),
    345 		       sysctl_kern_maxvnodes, 0, NULL, 0,
    346 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
    347 	sysctl_createv(clog, 0, NULL, NULL,
    348 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    349 		       CTLTYPE_INT, "maxproc",
    350 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
    351 		       sysctl_kern_maxproc, 0, NULL, 0,
    352 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
    353 	sysctl_createv(clog, 0, NULL, NULL,
    354 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    355 		       CTLTYPE_INT, "maxfiles",
    356 		       SYSCTL_DESCR("Maximum number of open files"),
    357 		       NULL, 0, &maxfiles, 0,
    358 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
    359 	sysctl_createv(clog, 0, NULL, NULL,
    360 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    361 		       CTLTYPE_INT, "argmax",
    362 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
    363 				    "execve(2)"),
    364 		       NULL, ARG_MAX, NULL, 0,
    365 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
    366 	sysctl_createv(clog, 0, NULL, NULL,
    367 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    368 		       CTLTYPE_STRING, "hostname",
    369 		       SYSCTL_DESCR("System hostname"),
    370 		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
    371 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
    372 	sysctl_createv(clog, 0, NULL, NULL,
    373 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    374 		       CTLTYPE_INT, "hostid",
    375 		       SYSCTL_DESCR("System host ID number"),
    376 		       sysctl_kern_hostid, 0, NULL, 0,
    377 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
    378 	sysctl_createv(clog, 0, NULL, NULL,
    379 		       CTLFLAG_PERMANENT,
    380 		       CTLTYPE_STRUCT, "clockrate",
    381 		       SYSCTL_DESCR("Kernel clock rates"),
    382 		       sysctl_kern_clockrate, 0, NULL,
    383 		       sizeof(struct clockinfo),
    384 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
    385 	sysctl_createv(clog, 0, NULL, NULL,
    386 		       CTLFLAG_PERMANENT,
    387 		       CTLTYPE_INT, "hardclock_ticks",
    388 		       SYSCTL_DESCR("Number of hardclock ticks"),
    389 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
    390 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
    391 	sysctl_createv(clog, 0, NULL, NULL,
    392 		       CTLFLAG_PERMANENT,
    393 		       CTLTYPE_STRUCT, "vnode",
    394 		       SYSCTL_DESCR("System vnode table"),
    395 		       sysctl_kern_vnode, 0, NULL, 0,
    396 		       CTL_KERN, KERN_VNODE, CTL_EOL);
    397 	sysctl_createv(clog, 0, NULL, NULL,
    398 		       CTLFLAG_PERMANENT,
    399 		       CTLTYPE_STRUCT, "file",
    400 		       SYSCTL_DESCR("System open file table"),
    401 		       sysctl_kern_file, 0, NULL, 0,
    402 		       CTL_KERN, KERN_FILE, CTL_EOL);
    403 #ifndef GPROF
    404 	sysctl_createv(clog, 0, NULL, NULL,
    405 		       CTLFLAG_PERMANENT,
    406 		       CTLTYPE_NODE, "profiling",
    407 		       SYSCTL_DESCR("Profiling information (not available)"),
    408 		       sysctl_notavail, 0, NULL, 0,
    409 		       CTL_KERN, KERN_PROF, CTL_EOL);
    410 #endif
    411 	sysctl_createv(clog, 0, NULL, NULL,
    412 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    413 		       CTLTYPE_INT, "posix1version",
    414 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
    415 				    "with which the operating system attempts "
    416 				    "to comply"),
    417 		       NULL, _POSIX_VERSION, NULL, 0,
    418 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
    419 	sysctl_createv(clog, 0, NULL, NULL,
    420 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    421 		       CTLTYPE_INT, "ngroups",
    422 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
    423 		       NULL, NGROUPS_MAX, NULL, 0,
    424 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
    425 	sysctl_createv(clog, 0, NULL, NULL,
    426 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    427 		       CTLTYPE_INT, "job_control",
    428 		       SYSCTL_DESCR("Whether job control is available"),
    429 		       NULL, 1, NULL, 0,
    430 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
    431 	sysctl_createv(clog, 0, NULL, NULL,
    432 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    433 		       CTLTYPE_INT, "saved_ids",
    434 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
    435 				    "available"), NULL,
    436 #ifdef _POSIX_SAVED_IDS
    437 		       1,
    438 #else /* _POSIX_SAVED_IDS */
    439 		       0,
    440 #endif /* _POSIX_SAVED_IDS */
    441 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
    442 	sysctl_createv(clog, 0, NULL, NULL,
    443 		       CTLFLAG_PERMANENT,
    444 		       CTLTYPE_STRUCT, "boottime",
    445 		       SYSCTL_DESCR("System boot time"),
    446 		       NULL, 0, &boottime, sizeof(boottime),
    447 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
    448 #ifdef COMPAT_50
    449 	{
    450 		extern struct timeval50 boottime50;
    451 		sysctl_createv(clog, 0, NULL, NULL,
    452 			       CTLFLAG_PERMANENT,
    453 			       CTLTYPE_STRUCT, "oboottime",
    454 			       SYSCTL_DESCR("System boot time"),
    455 			       NULL, 0, &boottime50, sizeof(boottime50),
    456 			       CTL_KERN, KERN_OBOOTTIME, CTL_EOL);
    457 	}
    458 #endif
    459 	sysctl_createv(clog, 0, NULL, NULL,
    460 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    461 		       CTLTYPE_STRING, "domainname",
    462 		       SYSCTL_DESCR("YP domain name"),
    463 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
    464 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
    465 	sysctl_createv(clog, 0, NULL, NULL,
    466 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    467 		       CTLTYPE_INT, "maxpartitions",
    468 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
    469 				    "disk"),
    470 		       NULL, MAXPARTITIONS, NULL, 0,
    471 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
    472 	sysctl_createv(clog, 0, NULL, NULL,
    473 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    474 		       CTLTYPE_INT, "rawpartition",
    475 		       SYSCTL_DESCR("Raw partition of a disk"),
    476 		       NULL, RAW_PART, NULL, 0,
    477 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
    478 	sysctl_createv(clog, 0, NULL, NULL,
    479 		       CTLFLAG_PERMANENT,
    480 		       CTLTYPE_STRUCT, "timex", NULL,
    481 		       sysctl_notavail, 0, NULL, 0,
    482 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
    483 	sysctl_createv(clog, 0, NULL, NULL,
    484 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    485 		       CTLTYPE_INT, "rtc_offset",
    486 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
    487 				    "minutes"),
    488 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
    489 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
    490 	sysctl_createv(clog, 0, NULL, NULL,
    491 		       CTLFLAG_PERMANENT,
    492 		       CTLTYPE_STRING, "root_device",
    493 		       SYSCTL_DESCR("Name of the root device"),
    494 		       sysctl_root_device, 0, NULL, 0,
    495 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
    496 	sysctl_createv(clog, 0, NULL, NULL,
    497 		       CTLFLAG_PERMANENT,
    498 		       CTLTYPE_INT, "msgbufsize",
    499 		       SYSCTL_DESCR("Size of the kernel message buffer"),
    500 		       sysctl_msgbuf, 0, NULL, 0,
    501 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
    502 	sysctl_createv(clog, 0, NULL, NULL,
    503 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    504 		       CTLTYPE_INT, "fsync",
    505 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
    506 				    "Synchronization Option is available on "
    507 				    "this system"),
    508 		       NULL, 1, NULL, 0,
    509 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
    510 	sysctl_createv(clog, 0, NULL, NULL,
    511 		       CTLFLAG_PERMANENT,
    512 		       CTLTYPE_NODE, "ipc",
    513 		       SYSCTL_DESCR("SysV IPC options"),
    514 		       NULL, 0, NULL, 0,
    515 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
    516 	sysctl_createv(clog, 0, NULL, NULL,
    517 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    518 		       CTLTYPE_INT, "sysvmsg",
    519 		       SYSCTL_DESCR("System V style message support available"),
    520 		       NULL,
    521 #ifdef SYSVMSG
    522 		       1,
    523 #else /* SYSVMSG */
    524 		       0,
    525 #endif /* SYSVMSG */
    526 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
    527 	sysctl_createv(clog, 0, NULL, NULL,
    528 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    529 		       CTLTYPE_INT, "sysvsem",
    530 		       SYSCTL_DESCR("System V style semaphore support "
    531 				    "available"), NULL,
    532 #ifdef SYSVSEM
    533 		       1,
    534 #else /* SYSVSEM */
    535 		       0,
    536 #endif /* SYSVSEM */
    537 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
    538 	sysctl_createv(clog, 0, NULL, NULL,
    539 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    540 		       CTLTYPE_INT, "sysvshm",
    541 		       SYSCTL_DESCR("System V style shared memory support "
    542 				    "available"), NULL,
    543 #ifdef SYSVSHM
    544 		       1,
    545 #else /* SYSVSHM */
    546 		       0,
    547 #endif /* SYSVSHM */
    548 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
    549 	sysctl_createv(clog, 0, NULL, NULL,
    550 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    551 		       CTLTYPE_INT, "synchronized_io",
    552 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
    553 				    "I/O Option is available on this system"),
    554 		       NULL, 1, NULL, 0,
    555 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
    556 	sysctl_createv(clog, 0, NULL, NULL,
    557 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    558 		       CTLTYPE_INT, "iov_max",
    559 		       SYSCTL_DESCR("Maximum number of iovec structures per "
    560 				    "process"),
    561 		       NULL, IOV_MAX, NULL, 0,
    562 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
    563 	sysctl_createv(clog, 0, NULL, NULL,
    564 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    565 		       CTLTYPE_INT, "mapped_files",
    566 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
    567 				    "Files Option is available on this system"),
    568 		       NULL, 1, NULL, 0,
    569 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
    570 	sysctl_createv(clog, 0, NULL, NULL,
    571 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    572 		       CTLTYPE_INT, "memlock",
    573 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
    574 				    "Locking Option is available on this "
    575 				    "system"),
    576 		       NULL, 1, NULL, 0,
    577 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
    578 	sysctl_createv(clog, 0, NULL, NULL,
    579 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    580 		       CTLTYPE_INT, "memlock_range",
    581 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
    582 				    "Locking Option is available on this "
    583 				    "system"),
    584 		       NULL, 1, NULL, 0,
    585 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
    586 	sysctl_createv(clog, 0, NULL, NULL,
    587 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    588 		       CTLTYPE_INT, "memory_protection",
    589 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
    590 				    "Protection Option is available on this "
    591 				    "system"),
    592 		       NULL, 1, NULL, 0,
    593 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
    594 	sysctl_createv(clog, 0, NULL, NULL,
    595 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    596 		       CTLTYPE_INT, "login_name_max",
    597 		       SYSCTL_DESCR("Maximum login name length"),
    598 		       NULL, LOGIN_NAME_MAX, NULL, 0,
    599 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
    600 	sysctl_createv(clog, 0, NULL, NULL,
    601 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    602 		       CTLTYPE_STRING, "defcorename",
    603 		       SYSCTL_DESCR("Default core file name"),
    604 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
    605 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
    606 	sysctl_createv(clog, 0, NULL, NULL,
    607 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    608 		       CTLTYPE_INT, "logsigexit",
    609 		       SYSCTL_DESCR("Log process exit when caused by signals"),
    610 		       NULL, 0, &kern_logsigexit, 0,
    611 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
    612 	sysctl_createv(clog, 0, NULL, NULL,
    613 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    614 		       CTLTYPE_INT, "fscale",
    615 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
    616 		       NULL, FSCALE, NULL, 0,
    617 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
    618 	sysctl_createv(clog, 0, NULL, NULL,
    619 		       CTLFLAG_PERMANENT,
    620 		       CTLTYPE_INT, "ccpu",
    621 		       SYSCTL_DESCR("Scheduler exponential decay value"),
    622 		       NULL, 0, &ccpu, 0,
    623 		       CTL_KERN, KERN_CCPU, CTL_EOL);
    624 	sysctl_createv(clog, 0, NULL, NULL,
    625 		       CTLFLAG_PERMANENT,
    626 		       CTLTYPE_STRUCT, "cp_time",
    627 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
    628 		       sysctl_kern_cptime, 0, NULL, 0,
    629 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
    630 	sysctl_createv(clog, 0, NULL, NULL,
    631 		       CTLFLAG_PERMANENT,
    632 		       CTLTYPE_INT, "msgbuf",
    633 		       SYSCTL_DESCR("Kernel message buffer"),
    634 		       sysctl_msgbuf, 0, NULL, 0,
    635 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
    636 	sysctl_createv(clog, 0, NULL, NULL,
    637 		       CTLFLAG_PERMANENT,
    638 		       CTLTYPE_STRUCT, "consdev",
    639 		       SYSCTL_DESCR("Console device"),
    640 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
    641 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
    642 #if NPTY > 0
    643 	sysctl_createv(clog, 0, NULL, NULL,
    644 		       CTLFLAG_PERMANENT,
    645 		       CTLTYPE_INT, "maxptys",
    646 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
    647 		       sysctl_kern_maxptys, 0, NULL, 0,
    648 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
    649 #endif /* NPTY > 0 */
    650 	sysctl_createv(clog, 0, NULL, NULL,
    651 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    652 		       CTLTYPE_INT, "maxphys",
    653 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
    654 		       NULL, MAXPHYS, NULL, 0,
    655 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
    656 	sysctl_createv(clog, 0, NULL, NULL,
    657 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    658 		       CTLTYPE_INT, "sbmax",
    659 		       SYSCTL_DESCR("Maximum socket buffer size"),
    660 		       sysctl_kern_sbmax, 0, NULL, 0,
    661 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
    662 	sysctl_createv(clog, 0, NULL, NULL,
    663 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    664 		       CTLTYPE_INT, "monotonic_clock",
    665 		       SYSCTL_DESCR("Implementation version of the POSIX "
    666 				    "1003.1b Monotonic Clock Option"),
    667 		       /* XXX _POSIX_VERSION */
    668 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
    669 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
    670 	sysctl_createv(clog, 0, NULL, NULL,
    671 		       CTLFLAG_PERMANENT,
    672 		       CTLTYPE_INT, "urandom",
    673 		       SYSCTL_DESCR("Random integer value"),
    674 		       sysctl_kern_urnd, 0, NULL, 0,
    675 		       CTL_KERN, KERN_URND, CTL_EOL);
    676 	sysctl_createv(clog, 0, NULL, NULL,
    677 		       CTLFLAG_PERMANENT,
    678 		       CTLTYPE_INT, "arandom",
    679 		       SYSCTL_DESCR("n bytes of random data"),
    680 		       sysctl_kern_arnd, 0, NULL, 0,
    681 		       CTL_KERN, KERN_ARND, CTL_EOL);
    682 	sysctl_createv(clog, 0, NULL, NULL,
    683 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    684 		       CTLTYPE_INT, "labelsector",
    685 		       SYSCTL_DESCR("Sector number containing the disklabel"),
    686 		       NULL, LABELSECTOR, NULL, 0,
    687 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
    688 	sysctl_createv(clog, 0, NULL, NULL,
    689 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    690 		       CTLTYPE_INT, "labeloffset",
    691 		       SYSCTL_DESCR("Offset of the disklabel within the "
    692 				    "sector"),
    693 		       NULL, LABELOFFSET, NULL, 0,
    694 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
    695 	sysctl_createv(clog, 0, NULL, NULL,
    696 		       CTLFLAG_PERMANENT,
    697 		       CTLTYPE_NODE, "lwp",
    698 		       SYSCTL_DESCR("System-wide LWP information"),
    699 		       sysctl_kern_lwp, 0, NULL, 0,
    700 		       CTL_KERN, KERN_LWP, CTL_EOL);
    701 	sysctl_createv(clog, 0, NULL, NULL,
    702 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    703 		       CTLTYPE_INT, "forkfsleep",
    704 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
    705 				    "to process limits"),
    706 		       sysctl_kern_forkfsleep, 0, NULL, 0,
    707 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
    708 	sysctl_createv(clog, 0, NULL, NULL,
    709 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    710 		       CTLTYPE_INT, "posix_threads",
    711 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    712 				    "Threads option to which the system "
    713 				    "attempts to conform"),
    714 		       /* XXX _POSIX_VERSION */
    715 		       NULL, _POSIX_THREADS, NULL, 0,
    716 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
    717 	sysctl_createv(clog, 0, NULL, NULL,
    718 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    719 		       CTLTYPE_INT, "posix_semaphores",
    720 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    721 				    "Semaphores option to which the system "
    722 				    "attempts to conform"), NULL,
    723 #ifdef P1003_1B_SEMAPHORE
    724 		       200112,
    725 #else /* P1003_1B_SEMAPHORE */
    726 		       0,
    727 #endif /* P1003_1B_SEMAPHORE */
    728 		       NULL, 0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
    729 	sysctl_createv(clog, 0, NULL, NULL,
    730 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    731 		       CTLTYPE_INT, "posix_barriers",
    732 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    733 				    "Barriers option to which the system "
    734 				    "attempts to conform"),
    735 		       /* XXX _POSIX_VERSION */
    736 		       NULL, _POSIX_BARRIERS, NULL, 0,
    737 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
    738 	sysctl_createv(clog, 0, NULL, NULL,
    739 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    740 		       CTLTYPE_INT, "posix_timers",
    741 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    742 				    "Timers option to which the system "
    743 				    "attempts to conform"),
    744 		       /* XXX _POSIX_VERSION */
    745 		       NULL, _POSIX_TIMERS, NULL, 0,
    746 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
    747 	sysctl_createv(clog, 0, NULL, NULL,
    748 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    749 		       CTLTYPE_INT, "posix_spin_locks",
    750 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
    751 				    "Locks option to which the system attempts "
    752 				    "to conform"),
    753 		       /* XXX _POSIX_VERSION */
    754 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
    755 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
    756 	sysctl_createv(clog, 0, NULL, NULL,
    757 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    758 		       CTLTYPE_INT, "posix_reader_writer_locks",
    759 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    760 				    "Read-Write Locks option to which the "
    761 				    "system attempts to conform"),
    762 		       /* XXX _POSIX_VERSION */
    763 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
    764 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
    765 	sysctl_createv(clog, 0, NULL, NULL,
    766 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    767 		       CTLTYPE_INT, "dump_on_panic",
    768 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
    769 		       NULL, 0, &dumponpanic, 0,
    770 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
    771 #ifdef DIAGNOSTIC
    772 	sysctl_createv(clog, 0, NULL, NULL,
    773 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    774 		       CTLTYPE_INT, "panic_now",
    775 		       SYSCTL_DESCR("Trigger a panic"),
    776 		       sysctl_kern_trigger_panic, 0, NULL, 0,
    777 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    778 #endif
    779 	sysctl_createv(clog, 0, NULL, NULL,
    780 		       CTLFLAG_PERMANENT,
    781 		       CTLTYPE_INT, "root_partition",
    782 		       SYSCTL_DESCR("Root partition on the root device"),
    783 		       sysctl_kern_root_partition, 0, NULL, 0,
    784 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
    785 	sysctl_createv(clog, 0, NULL, NULL,
    786 		       CTLFLAG_PERMANENT,
    787 		       CTLTYPE_STRUCT, "drivers",
    788 		       SYSCTL_DESCR("List of all drivers with block and "
    789 				    "character device numbers"),
    790 		       sysctl_kern_drivers, 0, NULL, 0,
    791 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
    792 	sysctl_createv(clog, 0, NULL, NULL,
    793 		       CTLFLAG_PERMANENT,
    794 		       CTLTYPE_STRUCT, "file2",
    795 		       SYSCTL_DESCR("System open file table"),
    796 		       sysctl_kern_file2, 0, NULL, 0,
    797 		       CTL_KERN, KERN_FILE2, CTL_EOL);
    798 	sysctl_createv(clog, 0, NULL, NULL,
    799 		       CTLFLAG_PERMANENT,
    800 		       CTLTYPE_STRUCT, "cp_id",
    801 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
    802 		       sysctl_kern_cpid, 0, NULL, 0,
    803 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
    804 	sysctl_createv(clog, 0, NULL, &rnode,
    805 		       CTLFLAG_PERMANENT,
    806 		       CTLTYPE_NODE, "coredump",
    807 		       SYSCTL_DESCR("Coredump settings."),
    808 		       NULL, 0, NULL, 0,
    809 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    810 	sysctl_createv(clog, 0, &rnode, &rnode,
    811 		       CTLFLAG_PERMANENT,
    812 		       CTLTYPE_NODE, "setid",
    813 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
    814 		       NULL, 0, NULL, 0,
    815 		       CTL_CREATE, CTL_EOL);
    816 	sysctl_createv(clog, 0, &rnode, NULL,
    817 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    818 		       CTLTYPE_INT, "dump",
    819 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
    820 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
    821 		       sizeof(security_setidcore_dump),
    822 		       CTL_CREATE, CTL_EOL);
    823 	sysctl_createv(clog, 0, &rnode, NULL,
    824 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    825 		       CTLTYPE_STRING, "path",
    826 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
    827 		       sysctl_security_setidcorename, 0,
    828 		       &security_setidcore_path,
    829 		       sizeof(security_setidcore_path),
    830 		       CTL_CREATE, CTL_EOL);
    831 	sysctl_createv(clog, 0, &rnode, NULL,
    832 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    833 		       CTLTYPE_INT, "owner",
    834 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
    835 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
    836 		       0,
    837 		       CTL_CREATE, CTL_EOL);
    838 	sysctl_createv(clog, 0, &rnode, NULL,
    839 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    840 		       CTLTYPE_INT, "group",
    841 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
    842 		       sysctl_security_setidcore, 0, &security_setidcore_group,
    843 		       0,
    844 		       CTL_CREATE, CTL_EOL);
    845 	sysctl_createv(clog, 0, &rnode, NULL,
    846 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    847 		       CTLTYPE_INT, "mode",
    848 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
    849 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
    850 		       0,
    851 		       CTL_CREATE, CTL_EOL);
    852 	sysctl_createv(clog, 0, NULL, NULL,
    853 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    854 		       CTLTYPE_INT, "no_sa_support",
    855 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise it doesn't"),
    856 		       NULL, 1, NULL, 0,
    857 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    858 }
    859 
    860 SYSCTL_SETUP(sysctl_kern_proc_setup,
    861 	     "sysctl kern.proc/proc2/proc_args subtree setup")
    862 {
    863 
    864 	sysctl_createv(clog, 0, NULL, NULL,
    865 		       CTLFLAG_PERMANENT,
    866 		       CTLTYPE_NODE, "kern", NULL,
    867 		       NULL, 0, NULL, 0,
    868 		       CTL_KERN, CTL_EOL);
    869 
    870 	sysctl_createv(clog, 0, NULL, NULL,
    871 		       CTLFLAG_PERMANENT,
    872 		       CTLTYPE_NODE, "proc",
    873 		       SYSCTL_DESCR("System-wide process information"),
    874 		       sysctl_doeproc, 0, NULL, 0,
    875 		       CTL_KERN, KERN_PROC, CTL_EOL);
    876 	sysctl_createv(clog, 0, NULL, NULL,
    877 		       CTLFLAG_PERMANENT,
    878 		       CTLTYPE_NODE, "proc2",
    879 		       SYSCTL_DESCR("Machine-independent process information"),
    880 		       sysctl_doeproc, 0, NULL, 0,
    881 		       CTL_KERN, KERN_PROC2, CTL_EOL);
    882 	sysctl_createv(clog, 0, NULL, NULL,
    883 		       CTLFLAG_PERMANENT,
    884 		       CTLTYPE_NODE, "proc_args",
    885 		       SYSCTL_DESCR("Process argument information"),
    886 		       sysctl_kern_proc_args, 0, NULL, 0,
    887 		       CTL_KERN, KERN_PROC_ARGS, CTL_EOL);
    888 
    889 	/*
    890 	  "nodes" under these:
    891 
    892 	  KERN_PROC_ALL
    893 	  KERN_PROC_PID pid
    894 	  KERN_PROC_PGRP pgrp
    895 	  KERN_PROC_SESSION sess
    896 	  KERN_PROC_TTY tty
    897 	  KERN_PROC_UID uid
    898 	  KERN_PROC_RUID uid
    899 	  KERN_PROC_GID gid
    900 	  KERN_PROC_RGID gid
    901 
    902 	  all in all, probably not worth the effort...
    903 	*/
    904 }
    905 
    906 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
    907 {
    908 	u_int u;
    909 	u_quad_t q;
    910 
    911 	sysctl_createv(clog, 0, NULL, NULL,
    912 		       CTLFLAG_PERMANENT,
    913 		       CTLTYPE_NODE, "hw", NULL,
    914 		       NULL, 0, NULL, 0,
    915 		       CTL_HW, CTL_EOL);
    916 
    917 	sysctl_createv(clog, 0, NULL, NULL,
    918 		       CTLFLAG_PERMANENT,
    919 		       CTLTYPE_STRING, "machine",
    920 		       SYSCTL_DESCR("Machine class"),
    921 		       NULL, 0, machine, 0,
    922 		       CTL_HW, HW_MACHINE, CTL_EOL);
    923 	sysctl_createv(clog, 0, NULL, NULL,
    924 		       CTLFLAG_PERMANENT,
    925 		       CTLTYPE_STRING, "model",
    926 		       SYSCTL_DESCR("Machine model"),
    927 		       NULL, 0, cpu_model, 0,
    928 		       CTL_HW, HW_MODEL, CTL_EOL);
    929 	sysctl_createv(clog, 0, NULL, NULL,
    930 		       CTLFLAG_PERMANENT,
    931 		       CTLTYPE_INT, "ncpu",
    932 		       SYSCTL_DESCR("Number of CPUs configured"),
    933 		       NULL, 0, &ncpu, 0,
    934 		       CTL_HW, HW_NCPU, CTL_EOL);
    935 	sysctl_createv(clog, 0, NULL, NULL,
    936 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    937 		       CTLTYPE_INT, "byteorder",
    938 		       SYSCTL_DESCR("System byte order"),
    939 		       NULL, BYTE_ORDER, NULL, 0,
    940 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
    941 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
    942 		UINT_MAX : physmem * PAGE_SIZE;
    943 	sysctl_createv(clog, 0, NULL, NULL,
    944 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    945 		       CTLTYPE_INT, "physmem",
    946 		       SYSCTL_DESCR("Bytes of physical memory"),
    947 		       NULL, u, NULL, 0,
    948 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
    949 	sysctl_createv(clog, 0, NULL, NULL,
    950 		       CTLFLAG_PERMANENT,
    951 		       CTLTYPE_INT, "usermem",
    952 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    953 		       sysctl_hw_usermem, 0, NULL, 0,
    954 		       CTL_HW, HW_USERMEM, CTL_EOL);
    955 	sysctl_createv(clog, 0, NULL, NULL,
    956 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    957 		       CTLTYPE_INT, "pagesize",
    958 		       SYSCTL_DESCR("Software page size"),
    959 		       NULL, PAGE_SIZE, NULL, 0,
    960 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
    961 	sysctl_createv(clog, 0, NULL, NULL,
    962 		       CTLFLAG_PERMANENT,
    963 		       CTLTYPE_STRING, "machine_arch",
    964 		       SYSCTL_DESCR("Machine CPU class"),
    965 		       NULL, 0, machine_arch, 0,
    966 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
    967 	sysctl_createv(clog, 0, NULL, NULL,
    968 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    969 		       CTLTYPE_INT, "alignbytes",
    970 		       SYSCTL_DESCR("Alignment constraint for all possible "
    971 				    "data types"),
    972 		       NULL, ALIGNBYTES, NULL, 0,
    973 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
    974 	sysctl_createv(clog, 0, NULL, NULL,
    975 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    976 		       CTLTYPE_STRING, "cnmagic",
    977 		       SYSCTL_DESCR("Console magic key sequence"),
    978 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
    979 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
    980 	q = (u_quad_t)physmem * PAGE_SIZE;
    981 	sysctl_createv(clog, 0, NULL, NULL,
    982 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    983 		       CTLTYPE_QUAD, "physmem64",
    984 		       SYSCTL_DESCR("Bytes of physical memory"),
    985 		       NULL, q, NULL, 0,
    986 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
    987 	sysctl_createv(clog, 0, NULL, NULL,
    988 		       CTLFLAG_PERMANENT,
    989 		       CTLTYPE_QUAD, "usermem64",
    990 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    991 		       sysctl_hw_usermem, 0, NULL, 0,
    992 		       CTL_HW, HW_USERMEM64, CTL_EOL);
    993 	sysctl_createv(clog, 0, NULL, NULL,
    994 		       CTLFLAG_PERMANENT,
    995 		       CTLTYPE_INT, "ncpuonline",
    996 		       SYSCTL_DESCR("Number of CPUs online"),
    997 		       NULL, 0, &ncpuonline, 0,
    998 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
    999 }
   1000 
   1001 #ifdef DEBUG
   1002 /*
   1003  * Debugging related system variables.
   1004  */
   1005 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
   1006 struct ctldebug debug5, debug6, debug7, debug8, debug9;
   1007 struct ctldebug debug10, debug11, debug12, debug13, debug14;
   1008 struct ctldebug debug15, debug16, debug17, debug18, debug19;
   1009 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
   1010 	&debug0, &debug1, &debug2, &debug3, &debug4,
   1011 	&debug5, &debug6, &debug7, &debug8, &debug9,
   1012 	&debug10, &debug11, &debug12, &debug13, &debug14,
   1013 	&debug15, &debug16, &debug17, &debug18, &debug19,
   1014 };
   1015 
   1016 /*
   1017  * this setup routine is a replacement for debug_sysctl()
   1018  *
   1019  * note that it creates several nodes per defined debug variable
   1020  */
   1021 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
   1022 {
   1023 	struct ctldebug *cdp;
   1024 	char nodename[20];
   1025 	int i;
   1026 
   1027 	/*
   1028 	 * two ways here:
   1029 	 *
   1030 	 * the "old" way (debug.name -> value) which was emulated by
   1031 	 * the sysctl(8) binary
   1032 	 *
   1033 	 * the new way, which the sysctl(8) binary was actually using
   1034 
   1035 	 node	debug
   1036 	 node	debug.0
   1037 	 string debug.0.name
   1038 	 int	debug.0.value
   1039 	 int	debug.name
   1040 
   1041 	 */
   1042 
   1043 	sysctl_createv(clog, 0, NULL, NULL,
   1044 		       CTLFLAG_PERMANENT,
   1045 		       CTLTYPE_NODE, "debug", NULL,
   1046 		       NULL, 0, NULL, 0,
   1047 		       CTL_DEBUG, CTL_EOL);
   1048 
   1049 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
   1050 		cdp = debugvars[i];
   1051 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
   1052 			continue;
   1053 
   1054 		snprintf(nodename, sizeof(nodename), "debug%d", i);
   1055 		sysctl_createv(clog, 0, NULL, NULL,
   1056 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1057 			       CTLTYPE_NODE, nodename, NULL,
   1058 			       NULL, 0, NULL, 0,
   1059 			       CTL_DEBUG, i, CTL_EOL);
   1060 		sysctl_createv(clog, 0, NULL, NULL,
   1061 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1062 			       CTLTYPE_STRING, "name", NULL,
   1063 			       /*XXXUNCONST*/
   1064 			       NULL, 0, __UNCONST(cdp->debugname), 0,
   1065 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
   1066 		sysctl_createv(clog, 0, NULL, NULL,
   1067 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
   1068 			       CTLTYPE_INT, "value", NULL,
   1069 			       NULL, 0, cdp->debugvar, 0,
   1070 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
   1071 		sysctl_createv(clog, 0, NULL, NULL,
   1072 			       CTLFLAG_PERMANENT,
   1073 			       CTLTYPE_INT, cdp->debugname, NULL,
   1074 			       NULL, 0, cdp->debugvar, 0,
   1075 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
   1076 	}
   1077 }
   1078 #endif /* DEBUG */
   1079 
   1080 /*
   1081  * ********************************************************************
   1082  * section 2: private node-specific helper routines.
   1083  * ********************************************************************
   1084  */
   1085 
   1086 #ifdef DIAGNOSTIC
   1087 static int
   1088 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
   1089 {
   1090 	int newtrig, error;
   1091 	struct sysctlnode node;
   1092 
   1093 	newtrig = 0;
   1094 	node = *rnode;
   1095 	node.sysctl_data = &newtrig;
   1096 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1097 	if (error || newp == NULL)
   1098 		return (error);
   1099 
   1100 	if (newtrig != 0)
   1101 		panic("Panic triggered");
   1102 
   1103 	return (error);
   1104 }
   1105 #endif
   1106 
   1107 /*
   1108  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
   1109  * new value is lower than desiredvnodes and then calls reinit
   1110  * routines that needs to adjust to the new value.
   1111  */
   1112 static int
   1113 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
   1114 {
   1115 	int error, new_vnodes, old_vnodes;
   1116 	struct sysctlnode node;
   1117 
   1118 	new_vnodes = desiredvnodes;
   1119 	node = *rnode;
   1120 	node.sysctl_data = &new_vnodes;
   1121 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1122 	if (error || newp == NULL)
   1123 		return (error);
   1124 
   1125 	old_vnodes = desiredvnodes;
   1126 	desiredvnodes = new_vnodes;
   1127 	if (new_vnodes < old_vnodes) {
   1128 		error = vfs_drainvnodes(new_vnodes, l);
   1129 		if (error) {
   1130 			desiredvnodes = old_vnodes;
   1131 			return (error);
   1132 		}
   1133 	}
   1134 	vfs_reinit();
   1135 	nchreinit();
   1136 
   1137 	return (0);
   1138 }
   1139 
   1140 /*
   1141  * sysctl helper routine for rtc_offset - set time after changes
   1142  */
   1143 static int
   1144 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
   1145 {
   1146 	struct timespec ts, delta;
   1147 	int error, new_rtc_offset;
   1148 	struct sysctlnode node;
   1149 
   1150 	new_rtc_offset = rtc_offset;
   1151 	node = *rnode;
   1152 	node.sysctl_data = &new_rtc_offset;
   1153 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1154 	if (error || newp == NULL)
   1155 		return (error);
   1156 
   1157 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
   1158 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
   1159 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
   1160 		return (EPERM);
   1161 	if (rtc_offset == new_rtc_offset)
   1162 		return (0);
   1163 
   1164 	/* if we change the offset, adjust the time */
   1165 	nanotime(&ts);
   1166 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
   1167 	delta.tv_nsec = 0;
   1168 	timespecadd(&ts, &delta, &ts);
   1169 	rtc_offset = new_rtc_offset;
   1170 	return (settime(l->l_proc, &ts));
   1171 }
   1172 
   1173 /*
   1174  * sysctl helper routine for kern.maxproc. Ensures that the new
   1175  * values are not too low or too high.
   1176  */
   1177 static int
   1178 sysctl_kern_maxproc(SYSCTLFN_ARGS)
   1179 {
   1180 	int error, nmaxproc;
   1181 	struct sysctlnode node;
   1182 
   1183 	nmaxproc = maxproc;
   1184 	node = *rnode;
   1185 	node.sysctl_data = &nmaxproc;
   1186 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1187 	if (error || newp == NULL)
   1188 		return (error);
   1189 
   1190 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
   1191 		return (EINVAL);
   1192 #ifdef __HAVE_CPU_MAXPROC
   1193 	if (nmaxproc > cpu_maxproc())
   1194 		return (EINVAL);
   1195 #endif
   1196 	maxproc = nmaxproc;
   1197 
   1198 	return (0);
   1199 }
   1200 
   1201 /*
   1202  * sysctl helper function for kern.hostid. The hostid is a long, but
   1203  * we export it as an int, so we need to give it a little help.
   1204  */
   1205 static int
   1206 sysctl_kern_hostid(SYSCTLFN_ARGS)
   1207 {
   1208 	int error, inthostid;
   1209 	struct sysctlnode node;
   1210 
   1211 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
   1212 	node = *rnode;
   1213 	node.sysctl_data = &inthostid;
   1214 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1215 	if (error || newp == NULL)
   1216 		return (error);
   1217 
   1218 	hostid = (unsigned)inthostid;
   1219 
   1220 	return (0);
   1221 }
   1222 
   1223 /*
   1224  * sysctl helper function for kern.hostname and kern.domainnname.
   1225  * resets the relevant recorded length when the underlying name is
   1226  * changed.
   1227  */
   1228 static int
   1229 sysctl_setlen(SYSCTLFN_ARGS)
   1230 {
   1231 	int error;
   1232 
   1233 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
   1234 	if (error || newp == NULL)
   1235 		return (error);
   1236 
   1237 	switch (rnode->sysctl_num) {
   1238 	case KERN_HOSTNAME:
   1239 		hostnamelen = strlen((const char*)rnode->sysctl_data);
   1240 		break;
   1241 	case KERN_DOMAINNAME:
   1242 		domainnamelen = strlen((const char*)rnode->sysctl_data);
   1243 		break;
   1244 	}
   1245 
   1246 	return (0);
   1247 }
   1248 
   1249 /*
   1250  * sysctl helper routine for kern.clockrate. Assembles a struct on
   1251  * the fly to be returned to the caller.
   1252  */
   1253 static int
   1254 sysctl_kern_clockrate(SYSCTLFN_ARGS)
   1255 {
   1256 	struct clockinfo clkinfo;
   1257 	struct sysctlnode node;
   1258 
   1259 	clkinfo.tick = tick;
   1260 	clkinfo.tickadj = tickadj;
   1261 	clkinfo.hz = hz;
   1262 	clkinfo.profhz = profhz;
   1263 	clkinfo.stathz = stathz ? stathz : hz;
   1264 
   1265 	node = *rnode;
   1266 	node.sysctl_data = &clkinfo;
   1267 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1268 }
   1269 
   1270 
   1271 /*
   1272  * sysctl helper routine for kern.file pseudo-subtree.
   1273  */
   1274 static int
   1275 sysctl_kern_file(SYSCTLFN_ARGS)
   1276 {
   1277 	int error;
   1278 	size_t buflen;
   1279 	struct file *fp, *dp, *np, fbuf;
   1280 	char *start, *where;
   1281 
   1282 	start = where = oldp;
   1283 	buflen = *oldlenp;
   1284 	dp = NULL;
   1285 
   1286 	if (where == NULL) {
   1287 		/*
   1288 		 * overestimate by 10 files
   1289 		 */
   1290 		*oldlenp = sizeof(filehead) + (nfiles + 10) *
   1291 		    sizeof(struct file);
   1292 		return (0);
   1293 	}
   1294 
   1295 	/*
   1296 	 * first dcopyout filehead
   1297 	 */
   1298 	if (buflen < sizeof(filehead)) {
   1299 		*oldlenp = 0;
   1300 		return (0);
   1301 	}
   1302 	sysctl_unlock();
   1303 	error = dcopyout(l, &filehead, where, sizeof(filehead));
   1304 	if (error) {
   1305 	 	sysctl_relock();
   1306 		return error;
   1307 	}
   1308 	buflen -= sizeof(filehead);
   1309 	where += sizeof(filehead);
   1310 
   1311 	/*
   1312 	 * allocate dummy file descriptor to make position in list
   1313 	 */
   1314 	if ((dp = fgetdummy()) == NULL) {
   1315 	 	sysctl_relock();
   1316 		return ENOMEM;
   1317 	}
   1318 
   1319 	/*
   1320 	 * followed by an array of file structures
   1321 	 */
   1322 	mutex_enter(&filelist_lock);
   1323 	for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
   1324 	    	np = LIST_NEXT(fp, f_list);
   1325 	    	mutex_enter(&fp->f_lock);
   1326 	    	if (fp->f_count == 0) {
   1327 		    	mutex_exit(&fp->f_lock);
   1328 	    		continue;
   1329 		}
   1330 		/*
   1331 		 * XXX Need to prevent that from being an alternative way
   1332 		 * XXX to getting process information.
   1333 		 */
   1334 		if (kauth_authorize_generic(l->l_cred,
   1335 		    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
   1336 		    	mutex_exit(&fp->f_lock);
   1337 			continue;
   1338 		}
   1339 		if (buflen < sizeof(struct file)) {
   1340 			*oldlenp = where - start;
   1341 		    	mutex_exit(&fp->f_lock);
   1342 			error = ENOMEM;
   1343 			break;
   1344 		}
   1345 		memcpy(&fbuf, fp, sizeof(fbuf));
   1346 		LIST_INSERT_AFTER(fp, dp, f_list);
   1347 	    	mutex_exit(&fp->f_lock);
   1348 		mutex_exit(&filelist_lock);
   1349 		error = dcopyout(l, &fbuf, where, sizeof(fbuf));
   1350 		if (error) {
   1351 			mutex_enter(&filelist_lock);
   1352 			LIST_REMOVE(dp, f_list);
   1353 			break;
   1354 		}
   1355 		buflen -= sizeof(struct file);
   1356 		where += sizeof(struct file);
   1357 		mutex_enter(&filelist_lock);
   1358 		np = LIST_NEXT(dp, f_list);
   1359 		LIST_REMOVE(dp, f_list);
   1360 	}
   1361 	mutex_exit(&filelist_lock);
   1362 	*oldlenp = where - start;
   1363  	if (dp != NULL)
   1364 		fputdummy(dp);
   1365  	sysctl_relock();
   1366 	return (error);
   1367 }
   1368 
   1369 /*
   1370  * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
   1371  * former it merely checks the message buffer is set up. For the latter,
   1372  * it also copies out the data if necessary.
   1373  */
   1374 static int
   1375 sysctl_msgbuf(SYSCTLFN_ARGS)
   1376 {
   1377 	char *where = oldp;
   1378 	size_t len, maxlen;
   1379 	long beg, end;
   1380 	extern kmutex_t log_lock;
   1381 	int error;
   1382 
   1383 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
   1384 		msgbufenabled = 0;
   1385 		return (ENXIO);
   1386 	}
   1387 
   1388 	switch (rnode->sysctl_num) {
   1389 	case KERN_MSGBUFSIZE: {
   1390 		struct sysctlnode node = *rnode;
   1391 		int msg_bufs = (int)msgbufp->msg_bufs;
   1392 		node.sysctl_data = &msg_bufs;
   1393 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1394 	}
   1395 	case KERN_MSGBUF:
   1396 		break;
   1397 	default:
   1398 		return (EOPNOTSUPP);
   1399 	}
   1400 
   1401 	if (newp != NULL)
   1402 		return (EPERM);
   1403 
   1404 	if (oldp == NULL) {
   1405 		/* always return full buffer size */
   1406 		*oldlenp = msgbufp->msg_bufs;
   1407 		return (0);
   1408 	}
   1409 
   1410 	sysctl_unlock();
   1411 
   1412 	/*
   1413 	 * First, copy from the write pointer to the end of
   1414 	 * message buffer.
   1415 	 */
   1416 	error = 0;
   1417 	mutex_spin_enter(&log_lock);
   1418 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
   1419 	beg = msgbufp->msg_bufx;
   1420 	end = msgbufp->msg_bufs;
   1421 	mutex_spin_exit(&log_lock);
   1422 
   1423 	while (maxlen > 0) {
   1424 		len = MIN(end - beg, maxlen);
   1425 		if (len == 0)
   1426 			break;
   1427 		/* XXX unlocked, but hardly matters. */
   1428 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
   1429 		if (error)
   1430 			break;
   1431 		where += len;
   1432 		maxlen -= len;
   1433 
   1434 		/*
   1435 		 * ... then, copy from the beginning of message buffer to
   1436 		 * the write pointer.
   1437 		 */
   1438 		beg = 0;
   1439 		end = msgbufp->msg_bufx;
   1440 	}
   1441 
   1442 	sysctl_relock();
   1443 	return (error);
   1444 }
   1445 
   1446 /*
   1447  * sysctl helper routine for kern.defcorename. In the case of a new
   1448  * string being assigned, check that it's not a zero-length string.
   1449  * (XXX the check in -current doesn't work, but do we really care?)
   1450  */
   1451 static int
   1452 sysctl_kern_defcorename(SYSCTLFN_ARGS)
   1453 {
   1454 	int error;
   1455 	char *newcorename;
   1456 	struct sysctlnode node;
   1457 
   1458 	newcorename = PNBUF_GET();
   1459 	node = *rnode;
   1460 	node.sysctl_data = &newcorename[0];
   1461 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   1462 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1463 	if (error || newp == NULL) {
   1464 		goto done;
   1465 	}
   1466 
   1467 	/*
   1468 	 * when sysctl_lookup() deals with a string, it's guaranteed
   1469 	 * to come back nul terminated. So there.  :)
   1470 	 */
   1471 	if (strlen(newcorename) == 0) {
   1472 		error = EINVAL;
   1473 	} else {
   1474 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   1475 		error = 0;
   1476 	}
   1477 done:
   1478 	PNBUF_PUT(newcorename);
   1479 	return error;
   1480 }
   1481 
   1482 /*
   1483  * sysctl helper routine for kern.cp_time node. Adds up cpu time
   1484  * across all cpus.
   1485  */
   1486 static int
   1487 sysctl_kern_cptime(SYSCTLFN_ARGS)
   1488 {
   1489 	struct sysctlnode node = *rnode;
   1490 	uint64_t *cp_time = NULL;
   1491 	int error, n = ncpu, i;
   1492 	struct cpu_info *ci;
   1493 	CPU_INFO_ITERATOR cii;
   1494 
   1495 	/*
   1496 	 * if you specifically pass a buffer that is the size of the
   1497 	 * sum, or if you are probing for the size, you get the "sum"
   1498 	 * of cp_time (and the size thereof) across all processors.
   1499 	 *
   1500 	 * alternately, you can pass an additional mib number and get
   1501 	 * cp_time for that particular processor.
   1502 	 */
   1503 	switch (namelen) {
   1504 	case 0:
   1505 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
   1506 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
   1507 			n = -1; /* SUM */
   1508 		}
   1509 		else {
   1510 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
   1511 			n = -2; /* ALL */
   1512 		}
   1513 		break;
   1514 	case 1:
   1515 		if (name[0] < 0 || name[0] >= n)
   1516 			return (ENOENT); /* ENOSUCHPROCESSOR */
   1517 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
   1518 		n = name[0];
   1519 		/*
   1520 		 * adjust these so that sysctl_lookup() will be happy
   1521 		 */
   1522 		name++;
   1523 		namelen--;
   1524 		break;
   1525 	default:
   1526 		return (EINVAL);
   1527 	}
   1528 
   1529 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
   1530 	if (cp_time == NULL)
   1531 		return (ENOMEM);
   1532 	node.sysctl_data = cp_time;
   1533 	memset(cp_time, 0, node.sysctl_size);
   1534 
   1535 	for (CPU_INFO_FOREACH(cii, ci)) {
   1536 		if (n <= 0) {
   1537 			for (i = 0; i < CPUSTATES; i++) {
   1538 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
   1539 			}
   1540 		}
   1541 		/*
   1542 		 * if a specific processor was requested and we just
   1543 		 * did it, we're done here
   1544 		 */
   1545 		if (n == 0)
   1546 			break;
   1547 		/*
   1548 		 * if doing "all", skip to next cp_time set for next processor
   1549 		 */
   1550 		if (n == -2)
   1551 			cp_time += CPUSTATES;
   1552 		/*
   1553 		 * if we're doing a specific processor, we're one
   1554 		 * processor closer
   1555 		 */
   1556 		if (n > 0)
   1557 			n--;
   1558 	}
   1559 
   1560 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1561 	kmem_free(node.sysctl_data, node.sysctl_size);
   1562 	return (error);
   1563 }
   1564 
   1565 #if NPTY > 0
   1566 /*
   1567  * sysctl helper routine for kern.maxptys. Ensures that any new value
   1568  * is acceptable to the pty subsystem.
   1569  */
   1570 static int
   1571 sysctl_kern_maxptys(SYSCTLFN_ARGS)
   1572 {
   1573 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
   1574 	int error, xmax;
   1575 	struct sysctlnode node;
   1576 
   1577 	/* get current value of maxptys */
   1578 	xmax = pty_maxptys(0, 0);
   1579 
   1580 	node = *rnode;
   1581 	node.sysctl_data = &xmax;
   1582 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1583 	if (error || newp == NULL)
   1584 		return (error);
   1585 
   1586 	if (xmax != pty_maxptys(xmax, 1))
   1587 		return (EINVAL);
   1588 
   1589 	return (0);
   1590 }
   1591 #endif /* NPTY > 0 */
   1592 
   1593 /*
   1594  * sysctl helper routine for kern.sbmax. Basically just ensures that
   1595  * any new value is not too small.
   1596  */
   1597 static int
   1598 sysctl_kern_sbmax(SYSCTLFN_ARGS)
   1599 {
   1600 	int error, new_sbmax;
   1601 	struct sysctlnode node;
   1602 
   1603 	new_sbmax = sb_max;
   1604 	node = *rnode;
   1605 	node.sysctl_data = &new_sbmax;
   1606 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1607 	if (error || newp == NULL)
   1608 		return (error);
   1609 
   1610 	KERNEL_LOCK(1, NULL);
   1611 	error = sb_max_set(new_sbmax);
   1612 	KERNEL_UNLOCK_ONE(NULL);
   1613 
   1614 	return (error);
   1615 }
   1616 
   1617 /*
   1618  * sysctl helper routine for kern.urandom node. Picks a random number
   1619  * for you.
   1620  */
   1621 static int
   1622 sysctl_kern_urnd(SYSCTLFN_ARGS)
   1623 {
   1624 #if NRND > 0
   1625 	int v, rv;
   1626 
   1627 	KERNEL_LOCK(1, NULL);
   1628 	rv = rnd_extract_data(&v, sizeof(v), RND_EXTRACT_ANY);
   1629 	KERNEL_UNLOCK_ONE(NULL);
   1630 	if (rv == sizeof(v)) {
   1631 		struct sysctlnode node = *rnode;
   1632 		node.sysctl_data = &v;
   1633 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1634 	}
   1635 	else
   1636 		return (EIO);	/*XXX*/
   1637 #else
   1638 	return (EOPNOTSUPP);
   1639 #endif
   1640 }
   1641 
   1642 /*
   1643  * sysctl helper routine for kern.arandom node. Picks a random number
   1644  * for you.
   1645  */
   1646 static int
   1647 sysctl_kern_arnd(SYSCTLFN_ARGS)
   1648 {
   1649 #if NRND > 0
   1650 	int error;
   1651 	void *v;
   1652 	struct sysctlnode node = *rnode;
   1653 
   1654 	if (*oldlenp == 0)
   1655 		return 0;
   1656 	if (*oldlenp > 8192)
   1657 		return E2BIG;
   1658 
   1659 	v = kmem_alloc(*oldlenp, KM_SLEEP);
   1660 	arc4randbytes(v, *oldlenp);
   1661 	node.sysctl_data = v;
   1662 	node.sysctl_size = *oldlenp;
   1663 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1664 	kmem_free(v, *oldlenp);
   1665 	return error;
   1666 #else
   1667 	return (EOPNOTSUPP);
   1668 #endif
   1669 }
   1670 /*
   1671  * sysctl helper routine to do kern.lwp.* work.
   1672  */
   1673 static int
   1674 sysctl_kern_lwp(SYSCTLFN_ARGS)
   1675 {
   1676 	struct kinfo_lwp klwp;
   1677 	struct proc *p;
   1678 	struct lwp *l2, *l3;
   1679 	char *where, *dp;
   1680 	int pid, elem_size, elem_count;
   1681 	int buflen, needed, error;
   1682 	bool gotit;
   1683 
   1684 	if (namelen == 1 && name[0] == CTL_QUERY)
   1685 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   1686 
   1687 	dp = where = oldp;
   1688 	buflen = where != NULL ? *oldlenp : 0;
   1689 	error = needed = 0;
   1690 
   1691 	if (newp != NULL || namelen != 3)
   1692 		return (EINVAL);
   1693 	pid = name[0];
   1694 	elem_size = name[1];
   1695 	elem_count = name[2];
   1696 
   1697 	sysctl_unlock();
   1698 	if (pid == -1) {
   1699 		mutex_enter(&proclist_lock);
   1700 		LIST_FOREACH(p, &allproc, p_list) {
   1701 			/* Grab a hold on the process. */
   1702 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   1703 				continue;
   1704 			}
   1705 			mutex_exit(&proclist_lock);
   1706 
   1707 			mutex_enter(&p->p_smutex);
   1708 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1709 				if (buflen >= elem_size && elem_count > 0) {
   1710 					lwp_lock(l2);
   1711 					fill_lwp(l2, &klwp);
   1712 					lwp_unlock(l2);
   1713 					mutex_exit(&p->p_smutex);
   1714 
   1715 					/*
   1716 					 * Copy out elem_size, but not
   1717 					 * larger than the size of a
   1718 					 * struct kinfo_proc2.
   1719 					 */
   1720 					error = dcopyout(l, &klwp, dp,
   1721 					    min(sizeof(klwp), elem_size));
   1722 					if (error) {
   1723 						rw_exit(&p->p_reflock);
   1724 						goto cleanup;
   1725 					}
   1726 					mutex_enter(&p->p_smutex);
   1727 					LIST_FOREACH(l3, &p->p_lwps,
   1728 					    l_sibling) {
   1729 						if (l2 == l3)
   1730 							break;
   1731 					}
   1732 					if (l3 == NULL) {
   1733 						mutex_exit(&p->p_smutex);
   1734 						rw_exit(&p->p_reflock);
   1735 						error = EAGAIN;
   1736 						goto cleanup;
   1737 					}
   1738 					dp += elem_size;
   1739 					buflen -= elem_size;
   1740 					elem_count--;
   1741 				}
   1742 				needed += elem_size;
   1743 			}
   1744 			mutex_exit(&p->p_smutex);
   1745 
   1746 			/* Drop reference to process. */
   1747 			mutex_enter(&proclist_lock);
   1748 			rw_exit(&p->p_reflock);
   1749 		}
   1750 		mutex_exit(&proclist_lock);
   1751 	} else {
   1752 		mutex_enter(&proclist_lock);
   1753 		p = p_find(pid, PFIND_LOCKED);
   1754 		if (p == NULL) {
   1755 			error = ESRCH;
   1756 			mutex_exit(&proclist_lock);
   1757 			goto cleanup;
   1758 		}
   1759 		/* Grab a hold on the process. */
   1760 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
   1761 		mutex_exit(&proclist_lock);
   1762 		if (!gotit) {
   1763 			error = ESRCH;
   1764 			goto cleanup;
   1765 		}
   1766 
   1767 		mutex_enter(&p->p_smutex);
   1768 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1769 			if (buflen >= elem_size && elem_count > 0) {
   1770 				lwp_lock(l2);
   1771 				fill_lwp(l2, &klwp);
   1772 				lwp_unlock(l2);
   1773 				mutex_exit(&p->p_smutex);
   1774 				/*
   1775 				 * Copy out elem_size, but not larger than
   1776 				 * the size of a struct kinfo_proc2.
   1777 				 */
   1778 				error = dcopyout(l, &klwp, dp,
   1779 				    min(sizeof(klwp), elem_size));
   1780 				if (error) {
   1781 					rw_exit(&p->p_reflock);
   1782 					goto cleanup;
   1783 				}
   1784 				mutex_enter(&p->p_smutex);
   1785 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
   1786 					if (l2 == l3)
   1787 						break;
   1788 				}
   1789 				if (l3 == NULL) {
   1790 					mutex_exit(&p->p_smutex);
   1791 					rw_exit(&p->p_reflock);
   1792 					error = EAGAIN;
   1793 					goto cleanup;
   1794 				}
   1795 				dp += elem_size;
   1796 				buflen -= elem_size;
   1797 				elem_count--;
   1798 			}
   1799 			needed += elem_size;
   1800 		}
   1801 		mutex_exit(&p->p_smutex);
   1802 
   1803 		/* Drop reference to process. */
   1804 		rw_exit(&p->p_reflock);
   1805 	}
   1806 
   1807 	if (where != NULL) {
   1808 		*oldlenp = dp - where;
   1809 		if (needed > *oldlenp) {
   1810 			sysctl_relock();
   1811 			return (ENOMEM);
   1812 		}
   1813 	} else {
   1814 		needed += KERN_LWPSLOP;
   1815 		*oldlenp = needed;
   1816 	}
   1817 	error = 0;
   1818  cleanup:
   1819 	sysctl_relock();
   1820 	return (error);
   1821 }
   1822 
   1823 /*
   1824  * sysctl helper routine for kern.forkfsleep node. Ensures that the
   1825  * given value is not too large or two small, and is at least one
   1826  * timer tick if not zero.
   1827  */
   1828 static int
   1829 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
   1830 {
   1831 	/* userland sees value in ms, internally is in ticks */
   1832 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
   1833 	int error, timo, lsleep;
   1834 	struct sysctlnode node;
   1835 
   1836 	lsleep = forkfsleep * 1000 / hz;
   1837 	node = *rnode;
   1838 	node.sysctl_data = &lsleep;
   1839 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1840 	if (error || newp == NULL)
   1841 		return (error);
   1842 
   1843 	/* refuse negative values, and overly 'long time' */
   1844 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
   1845 		return (EINVAL);
   1846 
   1847 	timo = mstohz(lsleep);
   1848 
   1849 	/* if the interval is >0 ms && <1 tick, use 1 tick */
   1850 	if (lsleep != 0 && timo == 0)
   1851 		forkfsleep = 1;
   1852 	else
   1853 		forkfsleep = timo;
   1854 
   1855 	return (0);
   1856 }
   1857 
   1858 /*
   1859  * sysctl helper routine for kern.root_partition
   1860  */
   1861 static int
   1862 sysctl_kern_root_partition(SYSCTLFN_ARGS)
   1863 {
   1864 	int rootpart = DISKPART(rootdev);
   1865 	struct sysctlnode node = *rnode;
   1866 
   1867 	node.sysctl_data = &rootpart;
   1868 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1869 }
   1870 
   1871 /*
   1872  * sysctl helper function for kern.drivers
   1873  */
   1874 static int
   1875 sysctl_kern_drivers(SYSCTLFN_ARGS)
   1876 {
   1877 	int error;
   1878 	size_t buflen;
   1879 	struct kinfo_drivers kd;
   1880 	char *start, *where;
   1881 	const char *dname;
   1882 	int i;
   1883 	extern struct devsw_conv *devsw_conv;
   1884 	extern int max_devsw_convs;
   1885 	extern kmutex_t devsw_lock;
   1886 
   1887 	if (newp != NULL || namelen != 0)
   1888 		return (EINVAL);
   1889 
   1890 	start = where = oldp;
   1891 	buflen = *oldlenp;
   1892 	if (where == NULL) {
   1893 		*oldlenp = max_devsw_convs * sizeof kd;
   1894 		return 0;
   1895 	}
   1896 
   1897 	/*
   1898 	 * An array of kinfo_drivers structures
   1899 	 */
   1900 	error = 0;
   1901 	sysctl_unlock();
   1902 	mutex_enter(&devsw_lock);
   1903 	for (i = 0; i < max_devsw_convs; i++) {
   1904 		dname = devsw_conv[i].d_name;
   1905 		if (dname == NULL)
   1906 			continue;
   1907 		if (buflen < sizeof kd) {
   1908 			error = ENOMEM;
   1909 			break;
   1910 		}
   1911 		memset(&kd, 0, sizeof(kd));
   1912 		kd.d_bmajor = devsw_conv[i].d_bmajor;
   1913 		kd.d_cmajor = devsw_conv[i].d_cmajor;
   1914 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
   1915 		mutex_exit(&devsw_lock);
   1916 		error = dcopyout(l, &kd, where, sizeof kd);
   1917 		mutex_enter(&devsw_lock);
   1918 		if (error != 0)
   1919 			break;
   1920 		buflen -= sizeof kd;
   1921 		where += sizeof kd;
   1922 	}
   1923 	mutex_exit(&devsw_lock);
   1924 	sysctl_relock();
   1925 	*oldlenp = where - start;
   1926 	return error;
   1927 }
   1928 
   1929 /*
   1930  * sysctl helper function for kern.file2
   1931  */
   1932 static int
   1933 sysctl_kern_file2(SYSCTLFN_ARGS)
   1934 {
   1935 	struct proc *p;
   1936 	struct file *fp, *tp, *np;
   1937 	struct filedesc *fd;
   1938 	struct kinfo_file kf;
   1939 	char *dp;
   1940 	u_int i, op;
   1941 	size_t len, needed, elem_size, out_size;
   1942 	int error, arg, elem_count;
   1943 	fdfile_t *ff;
   1944 
   1945 	if (namelen == 1 && name[0] == CTL_QUERY)
   1946 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   1947 
   1948 	if (namelen != 4)
   1949 		return (EINVAL);
   1950 
   1951 	error = 0;
   1952 	dp = oldp;
   1953 	len = (oldp != NULL) ? *oldlenp : 0;
   1954 	op = name[0];
   1955 	arg = name[1];
   1956 	elem_size = name[2];
   1957 	elem_count = name[3];
   1958 	out_size = MIN(sizeof(kf), elem_size);
   1959 	needed = 0;
   1960 
   1961 	if (elem_size < 1 || elem_count < 0)
   1962 		return (EINVAL);
   1963 
   1964 	switch (op) {
   1965 	case KERN_FILE_BYFILE:
   1966 		/*
   1967 		 * doesn't use arg so it must be zero
   1968 		 */
   1969 		if (arg != 0)
   1970 			return (EINVAL);
   1971 		sysctl_unlock();
   1972 		/*
   1973 		 * allocate dummy file descriptor to make position in list
   1974 		 */
   1975 		if ((tp = fgetdummy()) == NULL) {
   1976 		 	sysctl_relock();
   1977 			return ENOMEM;
   1978 		}
   1979 		mutex_enter(&filelist_lock);
   1980 		for (fp = LIST_FIRST(&filehead); fp != NULL; fp = np) {
   1981 			np = LIST_NEXT(fp, f_list);
   1982 			mutex_enter(&fp->f_lock);
   1983 			if (fp->f_count == 0) {
   1984 				mutex_exit(&fp->f_lock);
   1985 				continue;
   1986 			}
   1987 			/*
   1988 			 * XXX Need to prevent that from being an alternative
   1989 			 * XXX way for getting process information.
   1990 			 */
   1991 			if (kauth_authorize_generic(l->l_cred,
   1992 			    KAUTH_GENERIC_CANSEE, fp->f_cred) != 0) {
   1993 				mutex_exit(&fp->f_lock);
   1994 				continue;
   1995 			}
   1996 			if (len >= elem_size && elem_count > 0) {
   1997 				fill_file(&kf, fp, NULL, 0, 0);
   1998 				LIST_INSERT_AFTER(fp, tp, f_list);
   1999 				mutex_exit(&fp->f_lock);
   2000 				mutex_exit(&filelist_lock);
   2001 				error = dcopyout(l, &kf, dp, out_size);
   2002 				mutex_enter(&filelist_lock);
   2003 				np = LIST_NEXT(tp, f_list);
   2004 				LIST_REMOVE(tp, f_list);
   2005 				if (error) {
   2006 					break;
   2007 				}
   2008 				dp += elem_size;
   2009 				len -= elem_size;
   2010 			} else {
   2011 				mutex_exit(&fp->f_lock);
   2012 			}
   2013 			if (elem_count > 0) {
   2014 				needed += elem_size;
   2015 				if (elem_count != INT_MAX)
   2016 					elem_count--;
   2017 			}
   2018 		}
   2019 		mutex_exit(&filelist_lock);
   2020 		fputdummy(tp);
   2021 		sysctl_relock();
   2022 		break;
   2023 	case KERN_FILE_BYPID:
   2024 		if (arg < -1)
   2025 			/* -1 means all processes */
   2026 			return (EINVAL);
   2027 		sysctl_unlock();
   2028 		mutex_enter(&proclist_lock);
   2029 		LIST_FOREACH(p, &allproc, p_list) {
   2030 			if (p->p_stat == SIDL) {
   2031 				/* skip embryonic processes */
   2032 				continue;
   2033 			}
   2034 			if (arg > 0 && p->p_pid != arg) {
   2035 				/* pick only the one we want */
   2036 				/* XXX want 0 to mean "kernel files" */
   2037 				continue;
   2038 			}
   2039 			mutex_enter(&p->p_mutex);
   2040 			error = kauth_authorize_process(l->l_cred,
   2041 			    KAUTH_PROCESS_CANSEE, p,
   2042 			    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES),
   2043 			    NULL, NULL);
   2044 			mutex_exit(&p->p_mutex);
   2045 			if (error != 0) {
   2046 				continue;
   2047 			}
   2048 
   2049 			/*
   2050 			 * Grab a hold on the process.
   2051 			 */
   2052 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   2053 				continue;
   2054 			}
   2055 			mutex_exit(&proclist_lock);
   2056 
   2057 			/* XXX Do we need to check permission per file? */
   2058 			fd = p->p_fd;
   2059 			mutex_enter(&fd->fd_lock);
   2060 			for (i = 0; i < fd->fd_nfiles; i++) {
   2061 				if ((ff = fd->fd_ofiles[i]) == NULL) {
   2062 					continue;
   2063 				}
   2064 				mutex_enter(&ff->ff_lock);
   2065 				if ((fp = ff->ff_file) == NULL) {
   2066 					mutex_exit(&ff->ff_lock);
   2067 					continue;
   2068 				}
   2069 				if (len >= elem_size && elem_count > 0) {
   2070 					fill_file(&kf, fp, ff, i, p->p_pid);
   2071 					mutex_exit(&fp->f_lock);
   2072 					mutex_exit(&ff->ff_lock);
   2073 					mutex_exit(&fd->fd_lock);
   2074 					error = dcopyout(l, &kf, dp, out_size);
   2075 					mutex_enter(&fd->fd_lock);
   2076 					if (error)
   2077 						break;
   2078 					dp += elem_size;
   2079 					len -= elem_size;
   2080 				} else {
   2081 					mutex_exit(&fp->f_lock);
   2082 					mutex_exit(&ff->ff_lock);
   2083 				}
   2084 				if (elem_count > 0) {
   2085 					needed += elem_size;
   2086 					if (elem_count != INT_MAX)
   2087 						elem_count--;
   2088 				}
   2089 			}
   2090 			mutex_exit(&fd->fd_lock);
   2091 
   2092 			/*
   2093 			 * Release reference to process.
   2094 			 */
   2095 			mutex_enter(&proclist_lock);
   2096 			rw_exit(&p->p_reflock);
   2097 		}
   2098 		mutex_exit(&proclist_lock);
   2099 		sysctl_relock();
   2100 		break;
   2101 	default:
   2102 		return (EINVAL);
   2103 	}
   2104 
   2105 	if (oldp == NULL)
   2106 		needed += KERN_FILESLOP * elem_size;
   2107 	*oldlenp = needed;
   2108 
   2109 	return (error);
   2110 }
   2111 
   2112 static void
   2113 fill_file(struct kinfo_file *kp, const file_t *fp, const fdfile_t *ff,
   2114 	  int i, pid_t pid)
   2115 {
   2116 
   2117 	memset(kp, 0, sizeof(*kp));
   2118 
   2119 	kp->ki_fileaddr =	PTRTOUINT64(fp);
   2120 	kp->ki_flag =		fp->f_flag;
   2121 	kp->ki_iflags =		fp->f_iflags;
   2122 	kp->ki_ftype =		fp->f_type;
   2123 	kp->ki_count =		fp->f_count;
   2124 	kp->ki_msgcount =	fp->f_msgcount;
   2125 	kp->ki_fucred =		PTRTOUINT64(fp->f_cred);
   2126 	kp->ki_fuid =		kauth_cred_geteuid(fp->f_cred);
   2127 	kp->ki_fgid =		kauth_cred_getegid(fp->f_cred);
   2128 	kp->ki_fops =		PTRTOUINT64(fp->f_ops);
   2129 	kp->ki_foffset =	fp->f_offset;
   2130 	kp->ki_fdata =		PTRTOUINT64(fp->f_data);
   2131 
   2132 	/* vnode information to glue this file to something */
   2133 	if (fp->f_type == DTYPE_VNODE) {
   2134 		struct vnode *vp = (struct vnode *)fp->f_data;
   2135 
   2136 		kp->ki_vun =	PTRTOUINT64(vp->v_un.vu_socket);
   2137 		kp->ki_vsize =	vp->v_size;
   2138 		kp->ki_vtype =	vp->v_type;
   2139 		kp->ki_vtag =	vp->v_tag;
   2140 		kp->ki_vdata =	PTRTOUINT64(vp->v_data);
   2141 	}
   2142 
   2143 	/* process information when retrieved via KERN_FILE_BYPID */
   2144 	if (ff != NULL) {
   2145 		kp->ki_pid =		pid;
   2146 		kp->ki_fd =		i;
   2147 		kp->ki_ofileflags =	ff->ff_exclose;
   2148 		kp->ki_usecount =	ff->ff_refcnt;
   2149 	}
   2150 }
   2151 
   2152 static int
   2153 sysctl_doeproc(SYSCTLFN_ARGS)
   2154 {
   2155 	struct eproc *eproc;
   2156 	struct kinfo_proc2 *kproc2;
   2157 	struct kinfo_proc *dp;
   2158 	struct proc *p;
   2159 	char *where, *dp2;
   2160 	int type, op, arg;
   2161 	u_int elem_size, elem_count;
   2162 	size_t buflen, needed;
   2163 	bool match;
   2164 	int error;
   2165 
   2166 	if (namelen == 1 && name[0] == CTL_QUERY)
   2167 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   2168 
   2169 	dp = oldp;
   2170 	dp2 = where = oldp;
   2171 	buflen = where != NULL ? *oldlenp : 0;
   2172 	error = 0;
   2173 	needed = 0;
   2174 	type = rnode->sysctl_num;
   2175 
   2176 	if (type == KERN_PROC) {
   2177 		if (namelen != 2 && !(namelen == 1 && name[0] == KERN_PROC_ALL))
   2178 			return (EINVAL);
   2179 		op = name[0];
   2180 		if (op != KERN_PROC_ALL)
   2181 			arg = name[1];
   2182 		else
   2183 			arg = 0;		/* Quell compiler warning */
   2184 		elem_size = elem_count = 0;	/* Ditto */
   2185 	} else {
   2186 		if (namelen != 4)
   2187 			return (EINVAL);
   2188 		op = name[0];
   2189 		arg = name[1];
   2190 		elem_size = name[2];
   2191 		elem_count = name[3];
   2192 	}
   2193 
   2194 	sysctl_unlock();
   2195 
   2196 	if (type == KERN_PROC) {
   2197 		eproc = kmem_alloc(sizeof(*eproc), KM_SLEEP);
   2198 		kproc2 = NULL;
   2199 	} else {
   2200 		eproc = NULL;
   2201 		kproc2 = kmem_alloc(sizeof(*kproc2), KM_SLEEP);
   2202 	}
   2203 
   2204 	mutex_enter(&proclist_lock);
   2205 	LIST_FOREACH(p, &allproc, p_list) {
   2206 		/*
   2207 		 * Skip embryonic processes.
   2208 		 */
   2209 		if (p->p_stat == SIDL)
   2210 			continue;
   2211 
   2212 		mutex_enter(&p->p_mutex);
   2213 		error = kauth_authorize_process(l->l_cred,
   2214 		    KAUTH_PROCESS_CANSEE, p,
   2215 		    KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL);
   2216 		if (error != 0) {
   2217 			mutex_exit(&p->p_mutex);
   2218 			continue;
   2219 		}
   2220 
   2221 		/*
   2222 		 * TODO - make more efficient (see notes below).
   2223 		 * do by session.
   2224 		 */
   2225 		switch (op) {
   2226 		case KERN_PROC_PID:
   2227 			/* could do this with just a lookup */
   2228 			match = (p->p_pid == (pid_t)arg);
   2229 			break;
   2230 
   2231 		case KERN_PROC_PGRP:
   2232 			/* could do this by traversing pgrp */
   2233 			match = (p->p_pgrp->pg_id == (pid_t)arg);
   2234 			break;
   2235 
   2236 		case KERN_PROC_SESSION:
   2237 			match = (p->p_session->s_sid == (pid_t)arg);
   2238 			break;
   2239 
   2240 		case KERN_PROC_TTY:
   2241 			match = true;
   2242 			if (arg == (int) KERN_PROC_TTY_REVOKE) {
   2243 				if ((p->p_lflag & PL_CONTROLT) == 0 ||
   2244 				    p->p_session->s_ttyp == NULL ||
   2245 				    p->p_session->s_ttyvp != NULL) {
   2246 				    	match = false;
   2247 				}
   2248 			} else if ((p->p_lflag & PL_CONTROLT) == 0 ||
   2249 			    p->p_session->s_ttyp == NULL) {
   2250 				if ((dev_t)arg != KERN_PROC_TTY_NODEV) {
   2251 					match = false;
   2252 				}
   2253 			} else if (p->p_session->s_ttyp->t_dev != (dev_t)arg) {
   2254 				match = false;
   2255 			}
   2256 			break;
   2257 
   2258 		case KERN_PROC_UID:
   2259 			match = (kauth_cred_geteuid(p->p_cred) == (uid_t)arg);
   2260 			break;
   2261 
   2262 		case KERN_PROC_RUID:
   2263 			match = (kauth_cred_getuid(p->p_cred) == (uid_t)arg);
   2264 			break;
   2265 
   2266 		case KERN_PROC_GID:
   2267 			match = (kauth_cred_getegid(p->p_cred) == (uid_t)arg);
   2268 			break;
   2269 
   2270 		case KERN_PROC_RGID:
   2271 			match = (kauth_cred_getgid(p->p_cred) == (uid_t)arg);
   2272 			break;
   2273 
   2274 		case KERN_PROC_ALL:
   2275 			match = true;
   2276 			/* allow everything */
   2277 			break;
   2278 
   2279 		default:
   2280 			error = EINVAL;
   2281 			mutex_exit(&p->p_mutex);
   2282 			goto cleanup;
   2283 		}
   2284 		if (!match) {
   2285 			mutex_exit(&p->p_mutex);
   2286 			continue;
   2287 		}
   2288 
   2289 		/*
   2290 		 * Grab a hold on the process.
   2291 		 */
   2292 		if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   2293 			mutex_exit(&p->p_mutex);
   2294 			continue;
   2295 		}
   2296 
   2297 		if (type == KERN_PROC) {
   2298 			if (buflen >= sizeof(struct kinfo_proc)) {
   2299 				fill_eproc(p, eproc);
   2300 				mutex_exit(&p->p_mutex);
   2301 				mutex_exit(&proclist_lock);
   2302 				error = dcopyout(l, p, &dp->kp_proc,
   2303 				    sizeof(struct proc));
   2304 				mutex_enter(&proclist_lock);
   2305 				if (error) {
   2306 					rw_exit(&p->p_reflock);
   2307 					goto cleanup;
   2308 				}
   2309 				error = dcopyout(l, eproc, &dp->kp_eproc,
   2310 				    sizeof(*eproc));
   2311 				if (error) {
   2312 					rw_exit(&p->p_reflock);
   2313 					goto cleanup;
   2314 				}
   2315 				dp++;
   2316 				buflen -= sizeof(struct kinfo_proc);
   2317 			} else {
   2318 				mutex_exit(&p->p_mutex);
   2319 			}
   2320 			needed += sizeof(struct kinfo_proc);
   2321 		} else { /* KERN_PROC2 */
   2322 			if (buflen >= elem_size && elem_count > 0) {
   2323 				fill_kproc2(p, kproc2);
   2324 				mutex_exit(&p->p_mutex);
   2325 				mutex_exit(&proclist_lock);
   2326 				/*
   2327 				 * Copy out elem_size, but not larger than
   2328 				 * the size of a struct kinfo_proc2.
   2329 				 */
   2330 				error = dcopyout(l, kproc2, dp2,
   2331 				    min(sizeof(*kproc2), elem_size));
   2332 				mutex_enter(&proclist_lock);
   2333 				if (error) {
   2334 					rw_exit(&p->p_reflock);
   2335 					goto cleanup;
   2336 				}
   2337 				dp2 += elem_size;
   2338 				buflen -= elem_size;
   2339 				elem_count--;
   2340 			} else {
   2341 				mutex_exit(&p->p_mutex);
   2342 			}
   2343 			needed += elem_size;
   2344 		}
   2345 
   2346 		/*
   2347 		 * Release reference to process.
   2348 		 */
   2349 		rw_exit(&p->p_reflock);
   2350 	}
   2351 	mutex_exit(&proclist_lock);
   2352 
   2353 	if (where != NULL) {
   2354 		if (type == KERN_PROC)
   2355 			*oldlenp = (char *)dp - where;
   2356 		else
   2357 			*oldlenp = dp2 - where;
   2358 		if (needed > *oldlenp) {
   2359 			error = ENOMEM;
   2360 			goto out;
   2361 		}
   2362 	} else {
   2363 		needed += KERN_PROCSLOP;
   2364 		*oldlenp = needed;
   2365 	}
   2366 	if (kproc2)
   2367 		kmem_free(kproc2, sizeof(*kproc2));
   2368 	if (eproc)
   2369 		kmem_free(eproc, sizeof(*eproc));
   2370 	sysctl_relock();
   2371 	return 0;
   2372  cleanup:
   2373 	mutex_exit(&proclist_lock);
   2374  out:
   2375 	if (kproc2)
   2376 		kmem_free(kproc2, sizeof(*kproc2));
   2377 	if (eproc)
   2378 		kmem_free(eproc, sizeof(*eproc));
   2379 	sysctl_relock();
   2380 	return error;
   2381 }
   2382 
   2383 /*
   2384  * sysctl helper routine for kern.proc_args pseudo-subtree.
   2385  */
   2386 static int
   2387 sysctl_kern_proc_args(SYSCTLFN_ARGS)
   2388 {
   2389 	struct ps_strings pss;
   2390 	struct proc *p;
   2391 	size_t len, i;
   2392 	struct uio auio;
   2393 	struct iovec aiov;
   2394 	pid_t pid;
   2395 	int nargv, type, error, argvlen;
   2396 	char *arg;
   2397 	char **argv = NULL;
   2398 	char *tmp;
   2399 	struct vmspace *vmspace;
   2400 	vaddr_t psstr_addr;
   2401 	vaddr_t offsetn;
   2402 	vaddr_t offsetv;
   2403 
   2404 	if (namelen == 1 && name[0] == CTL_QUERY)
   2405 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   2406 
   2407 	if (newp != NULL || namelen != 2)
   2408 		return (EINVAL);
   2409 	pid = name[0];
   2410 	type = name[1];
   2411 	argv = NULL;
   2412 	argvlen = 0;
   2413 
   2414 	switch (type) {
   2415 	case KERN_PROC_ARGV:
   2416 	case KERN_PROC_NARGV:
   2417 	case KERN_PROC_ENV:
   2418 	case KERN_PROC_NENV:
   2419 		/* ok */
   2420 		break;
   2421 	default:
   2422 		return (EINVAL);
   2423 	}
   2424 
   2425 	sysctl_unlock();
   2426 
   2427 	/* check pid */
   2428 	mutex_enter(&proclist_lock);
   2429 	if ((p = p_find(pid, PFIND_LOCKED)) == NULL) {
   2430 		error = EINVAL;
   2431 		goto out_locked;
   2432 	}
   2433 	mutex_enter(&p->p_mutex);
   2434 
   2435 	/* Check permission. */
   2436 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV)
   2437 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
   2438 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ARGS), NULL, NULL);
   2439 	else if (type == KERN_PROC_ENV || type == KERN_PROC_NENV)
   2440 		error = kauth_authorize_process(l->l_cred, KAUTH_PROCESS_CANSEE,
   2441 		    p, KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENV), NULL, NULL);
   2442 	else
   2443 		error = EINVAL; /* XXXGCC */
   2444 	if (error) {
   2445 		mutex_exit(&p->p_mutex);
   2446 		goto out_locked;
   2447 	}
   2448 
   2449 	if (oldp == NULL) {
   2450 		if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV)
   2451 			*oldlenp = sizeof (int);
   2452 		else
   2453 			*oldlenp = ARG_MAX;	/* XXX XXX XXX */
   2454 		error = 0;
   2455 		mutex_exit(&p->p_mutex);
   2456 		goto out_locked;
   2457 	}
   2458 
   2459 	/*
   2460 	 * Zombies don't have a stack, so we can't read their psstrings.
   2461 	 * System processes also don't have a user stack.
   2462 	 */
   2463 	if (P_ZOMBIE(p) || (p->p_flag & PK_SYSTEM) != 0) {
   2464 		error = EINVAL;
   2465 		mutex_exit(&p->p_mutex);
   2466 		goto out_locked;
   2467 	}
   2468 
   2469 	/*
   2470 	 * Lock the process down in memory.
   2471 	 */
   2472 	psstr_addr = (vaddr_t)p->p_psstr;
   2473 	if (type == KERN_PROC_ARGV || type == KERN_PROC_NARGV) {
   2474 		offsetn = p->p_psnargv;
   2475 		offsetv = p->p_psargv;
   2476 	} else {
   2477 		offsetn = p->p_psnenv;
   2478 		offsetv = p->p_psenv;
   2479 	}
   2480 	vmspace = p->p_vmspace;
   2481 	uvmspace_addref(vmspace);
   2482 	mutex_exit(&p->p_mutex);
   2483 	mutex_exit(&proclist_lock);
   2484 
   2485 	/*
   2486 	 * Allocate a temporary buffer to hold the arguments.
   2487 	 */
   2488 	arg = kmem_alloc(PAGE_SIZE, KM_SLEEP);
   2489 
   2490 	/*
   2491 	 * Read in the ps_strings structure.
   2492 	 */
   2493 	aiov.iov_base = &pss;
   2494 	aiov.iov_len = sizeof(pss);
   2495 	auio.uio_iov = &aiov;
   2496 	auio.uio_iovcnt = 1;
   2497 	auio.uio_offset = psstr_addr;
   2498 	auio.uio_resid = sizeof(pss);
   2499 	auio.uio_rw = UIO_READ;
   2500 	UIO_SETUP_SYSSPACE(&auio);
   2501 	error = uvm_io(&vmspace->vm_map, &auio);
   2502 	if (error)
   2503 		goto done;
   2504 
   2505 	memcpy(&nargv, (char *)&pss + offsetn, sizeof(nargv));
   2506 	if (type == KERN_PROC_NARGV || type == KERN_PROC_NENV) {
   2507 		error = dcopyout(l, &nargv, oldp, sizeof(nargv));
   2508 		*oldlenp = sizeof(nargv);
   2509 		goto done;
   2510 	}
   2511 	/*
   2512 	 * Now read the address of the argument vector.
   2513 	 */
   2514 	switch (type) {
   2515 	case KERN_PROC_ARGV:
   2516 		/* FALLTHROUGH */
   2517 	case KERN_PROC_ENV:
   2518 		memcpy(&tmp, (char *)&pss + offsetv, sizeof(tmp));
   2519 		break;
   2520 	default:
   2521 		error = EINVAL;
   2522 		goto done;
   2523 	}
   2524 
   2525 #ifdef COMPAT_NETBSD32
   2526 	if (p->p_flag & PK_32)
   2527 		len = sizeof(netbsd32_charp) * nargv;
   2528 	else
   2529 #endif
   2530 		len = sizeof(char *) * nargv;
   2531 
   2532 	if ((argvlen = len) != 0)
   2533 		argv = kmem_alloc(len, KM_SLEEP);
   2534 
   2535 	aiov.iov_base = argv;
   2536 	aiov.iov_len = len;
   2537 	auio.uio_iov = &aiov;
   2538 	auio.uio_iovcnt = 1;
   2539 	auio.uio_offset = (off_t)(unsigned long)tmp;
   2540 	auio.uio_resid = len;
   2541 	auio.uio_rw = UIO_READ;
   2542 	UIO_SETUP_SYSSPACE(&auio);
   2543 	error = uvm_io(&vmspace->vm_map, &auio);
   2544 	if (error)
   2545 		goto done;
   2546 
   2547 	/*
   2548 	 * Now copy each string.
   2549 	 */
   2550 	len = 0; /* bytes written to user buffer */
   2551 	for (i = 0; i < nargv; i++) {
   2552 		int finished = 0;
   2553 		vaddr_t base;
   2554 		size_t xlen;
   2555 		int j;
   2556 
   2557 #ifdef COMPAT_NETBSD32
   2558 		if (p->p_flag & PK_32) {
   2559 			netbsd32_charp *argv32;
   2560 
   2561 			argv32 = (netbsd32_charp *)argv;
   2562 			base = (vaddr_t)NETBSD32PTR64(argv32[i]);
   2563 		} else
   2564 #endif
   2565 			base = (vaddr_t)argv[i];
   2566 
   2567 		/*
   2568 		 * The program has messed around with its arguments,
   2569 		 * possibly deleting some, and replacing them with
   2570 		 * NULL's. Treat this as the last argument and not
   2571 		 * a failure.
   2572 		 */
   2573 		if (base == 0)
   2574 			break;
   2575 
   2576 		while (!finished) {
   2577 			xlen = PAGE_SIZE - (base & PAGE_MASK);
   2578 
   2579 			aiov.iov_base = arg;
   2580 			aiov.iov_len = PAGE_SIZE;
   2581 			auio.uio_iov = &aiov;
   2582 			auio.uio_iovcnt = 1;
   2583 			auio.uio_offset = base;
   2584 			auio.uio_resid = xlen;
   2585 			auio.uio_rw = UIO_READ;
   2586 			UIO_SETUP_SYSSPACE(&auio);
   2587 			error = uvm_io(&vmspace->vm_map, &auio);
   2588 			if (error)
   2589 				goto done;
   2590 
   2591 			/* Look for the end of the string */
   2592 			for (j = 0; j < xlen; j++) {
   2593 				if (arg[j] == '\0') {
   2594 					xlen = j + 1;
   2595 					finished = 1;
   2596 					break;
   2597 				}
   2598 			}
   2599 
   2600 			/* Check for user buffer overflow */
   2601 			if (len + xlen > *oldlenp) {
   2602 				finished = 1;
   2603 				if (len > *oldlenp)
   2604 					xlen = 0;
   2605 				else
   2606 					xlen = *oldlenp - len;
   2607 			}
   2608 
   2609 			/* Copyout the page */
   2610 			error = dcopyout(l, arg, (char *)oldp + len, xlen);
   2611 			if (error)
   2612 				goto done;
   2613 
   2614 			len += xlen;
   2615 			base += xlen;
   2616 		}
   2617 	}
   2618 	*oldlenp = len;
   2619 
   2620 done:
   2621 	if (argvlen != 0)
   2622 		kmem_free(argv, argvlen);
   2623 	uvmspace_free(vmspace);
   2624 	kmem_free(arg, PAGE_SIZE);
   2625 	sysctl_relock();
   2626 	return error;
   2627 
   2628 out_locked:
   2629 	mutex_exit(&proclist_lock);
   2630 	sysctl_relock();
   2631 	return error;
   2632 }
   2633 
   2634 static int
   2635 sysctl_security_setidcore(SYSCTLFN_ARGS)
   2636 {
   2637 	int newsize, error;
   2638 	struct sysctlnode node;
   2639 
   2640 	node = *rnode;
   2641 	node.sysctl_data = &newsize;
   2642 	newsize = *(int *)rnode->sysctl_data;
   2643 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2644 	if (error || newp == NULL)
   2645 		return error;
   2646 
   2647 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   2648 	    0, NULL, NULL, NULL))
   2649 		return (EPERM);
   2650 
   2651 	*(int *)rnode->sysctl_data = newsize;
   2652 
   2653 	return 0;
   2654 }
   2655 
   2656 static int
   2657 sysctl_security_setidcorename(SYSCTLFN_ARGS)
   2658 {
   2659 	int error;
   2660 	char *newsetidcorename;
   2661 	struct sysctlnode node;
   2662 
   2663 	newsetidcorename = PNBUF_GET();
   2664 	node = *rnode;
   2665 	node.sysctl_data = newsetidcorename;
   2666 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   2667 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2668 	if (error || newp == NULL) {
   2669 		goto out;
   2670 	}
   2671 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   2672 	    0, NULL, NULL, NULL)) {
   2673 		error = EPERM;
   2674 		goto out;
   2675 	}
   2676 	if (strlen(newsetidcorename) == 0) {
   2677 		error = EINVAL;
   2678 		goto out;
   2679 	}
   2680 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   2681 out:
   2682 	PNBUF_PUT(newsetidcorename);
   2683 	return error;
   2684 }
   2685 
   2686 /*
   2687  * sysctl helper routine for kern.cp_id node. Maps cpus to their
   2688  * cpuids.
   2689  */
   2690 static int
   2691 sysctl_kern_cpid(SYSCTLFN_ARGS)
   2692 {
   2693 	struct sysctlnode node = *rnode;
   2694 	uint64_t *cp_id = NULL;
   2695 	int error, n = ncpu;
   2696 	struct cpu_info *ci;
   2697 	CPU_INFO_ITERATOR cii;
   2698 
   2699 	/*
   2700 	 * Here you may either retrieve a single cpu id or the whole
   2701 	 * set. The size you get back when probing depends on what
   2702 	 * you ask for.
   2703 	 */
   2704 	switch (namelen) {
   2705 	case 0:
   2706 		node.sysctl_size = n * sizeof(uint64_t);
   2707 		n = -2; /* ALL */
   2708 		break;
   2709 	case 1:
   2710 		if (name[0] < 0 || name[0] >= n)
   2711 			return (ENOENT); /* ENOSUCHPROCESSOR */
   2712 		node.sysctl_size = sizeof(uint64_t);
   2713 		n = name[0];
   2714 		/*
   2715 		 * adjust these so that sysctl_lookup() will be happy
   2716 		 */
   2717 		name++;
   2718 		namelen--;
   2719 		break;
   2720 	default:
   2721 		return (EINVAL);
   2722 	}
   2723 
   2724 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
   2725 	if (cp_id == NULL)
   2726 		return (ENOMEM);
   2727 	node.sysctl_data = cp_id;
   2728 	memset(cp_id, 0, node.sysctl_size);
   2729 
   2730 	for (CPU_INFO_FOREACH(cii, ci)) {
   2731 		if (n <= 0)
   2732 			cp_id[0] = ci->ci_cpuid;
   2733 		/*
   2734 		 * if a specific processor was requested and we just
   2735 		 * did it, we're done here
   2736 		 */
   2737 		if (n == 0)
   2738 			break;
   2739 		/*
   2740 		 * if doing "all", skip to next cp_id slot for next processor
   2741 		 */
   2742 		if (n == -2)
   2743 			cp_id++;
   2744 		/*
   2745 		 * if we're doing a specific processor, we're one
   2746 		 * processor closer
   2747 		 */
   2748 		if (n > 0)
   2749 			n--;
   2750 	}
   2751 
   2752 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2753 	kmem_free(node.sysctl_data, node.sysctl_size);
   2754 	return (error);
   2755 }
   2756 
   2757 /*
   2758  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
   2759  * calculate on the fly taking into account integer overflow and the
   2760  * current wired count.
   2761  */
   2762 static int
   2763 sysctl_hw_usermem(SYSCTLFN_ARGS)
   2764 {
   2765 	u_int ui;
   2766 	u_quad_t uq;
   2767 	struct sysctlnode node;
   2768 
   2769 	node = *rnode;
   2770 	switch (rnode->sysctl_num) {
   2771 	    case HW_USERMEM:
   2772 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
   2773 			ui = UINT_MAX;
   2774 		else
   2775 			ui *= PAGE_SIZE;
   2776 		node.sysctl_data = &ui;
   2777 		break;
   2778 	case HW_USERMEM64:
   2779 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
   2780 		node.sysctl_data = &uq;
   2781 		break;
   2782 	default:
   2783 		return (EINVAL);
   2784 	}
   2785 
   2786 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2787 }
   2788 
   2789 /*
   2790  * sysctl helper routine for kern.cnmagic node. Pulls the old value
   2791  * out, encoded, and stuffs the new value in for decoding.
   2792  */
   2793 static int
   2794 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
   2795 {
   2796 	char magic[CNS_LEN];
   2797 	int error;
   2798 	struct sysctlnode node;
   2799 
   2800 	if (oldp)
   2801 		cn_get_magic(magic, CNS_LEN);
   2802 	node = *rnode;
   2803 	node.sysctl_data = &magic[0];
   2804 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   2805 	if (error || newp == NULL)
   2806 		return (error);
   2807 
   2808 	return (cn_set_magic(magic));
   2809 }
   2810 
   2811 /*
   2812  * ********************************************************************
   2813  * section 3: public helper routines that are used for more than one
   2814  * node
   2815  * ********************************************************************
   2816  */
   2817 
   2818 /*
   2819  * sysctl helper routine for the kern.root_device node and some ports'
   2820  * machdep.root_device nodes.
   2821  */
   2822 int
   2823 sysctl_root_device(SYSCTLFN_ARGS)
   2824 {
   2825 	struct sysctlnode node;
   2826 
   2827 	node = *rnode;
   2828 	node.sysctl_data = root_device->dv_xname;
   2829 	node.sysctl_size = strlen(root_device->dv_xname) + 1;
   2830 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2831 }
   2832 
   2833 /*
   2834  * sysctl helper routine for kern.consdev, dependent on the current
   2835  * state of the console. Also used for machdep.console_device on some
   2836  * ports.
   2837  */
   2838 int
   2839 sysctl_consdev(SYSCTLFN_ARGS)
   2840 {
   2841 	dev_t consdev;
   2842 	struct sysctlnode node;
   2843 
   2844 	if (cn_tab != NULL)
   2845 		consdev = cn_tab->cn_dev;
   2846 	else
   2847 		consdev = NODEV;
   2848 	node = *rnode;
   2849 	node.sysctl_data = &consdev;
   2850 	node.sysctl_size = sizeof(consdev);
   2851 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   2852 }
   2853 
   2854 /*
   2855  * ********************************************************************
   2856  * section 4: support for some helpers
   2857  * ********************************************************************
   2858  */
   2859 
   2860 /*
   2861  * Fill in a kinfo_proc2 structure for the specified process.
   2862  */
   2863 static void
   2864 fill_kproc2(struct proc *p, struct kinfo_proc2 *ki)
   2865 {
   2866 	struct tty *tp;
   2867 	struct lwp *l, *l2;
   2868 	struct timeval ut, st, rt;
   2869 	sigset_t ss1, ss2;
   2870 	struct rusage ru;
   2871 
   2872 	KASSERT(mutex_owned(&proclist_lock));
   2873 	KASSERT(mutex_owned(&p->p_mutex));
   2874 
   2875 	memset(ki, 0, sizeof(*ki));
   2876 
   2877 	ki->p_paddr = PTRTOUINT64(p);
   2878 	ki->p_fd = PTRTOUINT64(p->p_fd);
   2879 	ki->p_cwdi = PTRTOUINT64(p->p_cwdi);
   2880 	ki->p_stats = PTRTOUINT64(p->p_stats);
   2881 	ki->p_limit = PTRTOUINT64(p->p_limit);
   2882 	ki->p_vmspace = PTRTOUINT64(p->p_vmspace);
   2883 	ki->p_sigacts = PTRTOUINT64(p->p_sigacts);
   2884 	ki->p_sess = PTRTOUINT64(p->p_session);
   2885 	ki->p_tsess = 0;	/* may be changed if controlling tty below */
   2886 	ki->p_ru = PTRTOUINT64(&p->p_stats->p_ru);
   2887 
   2888 	ki->p_eflag = 0;
   2889 	ki->p_exitsig = p->p_exitsig;
   2890 
   2891 	ki->p_flag = sysctl_map_flags(sysctl_flagmap, p->p_flag);
   2892 	ki->p_flag |= sysctl_map_flags(sysctl_sflagmap, p->p_sflag);
   2893 	ki->p_flag |= sysctl_map_flags(sysctl_slflagmap, p->p_slflag);
   2894 	ki->p_flag |= sysctl_map_flags(sysctl_lflagmap, p->p_lflag);
   2895 	ki->p_flag |= sysctl_map_flags(sysctl_stflagmap, p->p_stflag);
   2896 
   2897 	ki->p_pid = p->p_pid;
   2898 	if (p->p_pptr)
   2899 		ki->p_ppid = p->p_pptr->p_pid;
   2900 	else
   2901 		ki->p_ppid = 0;
   2902 	ki->p_sid = p->p_session->s_sid;
   2903 	ki->p__pgid = p->p_pgrp->pg_id;
   2904 
   2905 	ki->p_tpgid = NO_PGID;	/* may be changed if controlling tty below */
   2906 
   2907 	ki->p_uid = kauth_cred_geteuid(p->p_cred);
   2908 	ki->p_ruid = kauth_cred_getuid(p->p_cred);
   2909 	ki->p_gid = kauth_cred_getegid(p->p_cred);
   2910 	ki->p_rgid = kauth_cred_getgid(p->p_cred);
   2911 	ki->p_svuid = kauth_cred_getsvuid(p->p_cred);
   2912 	ki->p_svgid = kauth_cred_getsvgid(p->p_cred);
   2913 
   2914 	ki->p_ngroups = kauth_cred_ngroups(p->p_cred);
   2915 	kauth_cred_getgroups(p->p_cred, ki->p_groups,
   2916 	    min(ki->p_ngroups, sizeof(ki->p_groups) / sizeof(ki->p_groups[0])),
   2917 	    UIO_SYSSPACE);
   2918 
   2919 	ki->p_jobc = p->p_pgrp->pg_jobc;
   2920 	if ((p->p_lflag & PL_CONTROLT) && (tp = p->p_session->s_ttyp)) {
   2921 		ki->p_tdev = tp->t_dev;
   2922 		ki->p_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
   2923 		ki->p_tsess = PTRTOUINT64(tp->t_session);
   2924 	} else {
   2925 		ki->p_tdev = (uint32_t)NODEV;
   2926 	}
   2927 
   2928 	mutex_enter(&p->p_smutex);
   2929 
   2930 	ki->p_uticks = p->p_uticks;
   2931 	ki->p_sticks = p->p_sticks;
   2932 	ki->p_iticks = p->p_iticks;
   2933 
   2934 	ki->p_tracep = PTRTOUINT64(p->p_tracep);
   2935 	ki->p_traceflag = p->p_traceflag;
   2936 
   2937 	memcpy(&ki->p_sigignore, &p->p_sigctx.ps_sigignore,sizeof(ki_sigset_t));
   2938 	memcpy(&ki->p_sigcatch, &p->p_sigctx.ps_sigcatch, sizeof(ki_sigset_t));
   2939 
   2940 	ki->p_cpticks = 0;
   2941 	ki->p_pctcpu = p->p_pctcpu;
   2942 	ki->p_estcpu = 0;
   2943 	ss1 = p->p_sigpend.sp_set;
   2944 	LIST_FOREACH(l, &p->p_lwps, l_sibling) {
   2945 		/* This is hardly correct, but... */
   2946 		sigplusset(&l->l_sigpend.sp_set, &ss1);
   2947 		sigplusset(&l->l_sigmask, &ss2);
   2948 		ki->p_cpticks += l->l_cpticks;
   2949 		ki->p_pctcpu += l->l_pctcpu;
   2950 		ki->p_estcpu += l->l_estcpu;
   2951 	}
   2952 	memcpy(&ki->p_siglist, &ss1, sizeof(ki_sigset_t));
   2953 	memcpy(&ki->p_sigmask, &ss2, sizeof(ki_sigset_t));
   2954 
   2955 	ki->p_stat = p->p_stat; /* Will likely be overridden by LWP status */
   2956 	ki->p_realstat = p->p_stat;
   2957 	ki->p_nice = p->p_nice;
   2958 
   2959 	ki->p_xstat = p->p_xstat;
   2960 	ki->p_acflag = p->p_acflag;
   2961 
   2962 	strncpy(ki->p_comm, p->p_comm,
   2963 	    min(sizeof(ki->p_comm), sizeof(p->p_comm)));
   2964 
   2965 	strncpy(ki->p_login, p->p_session->s_login,
   2966 	    min(sizeof ki->p_login - 1, sizeof p->p_session->s_login));
   2967 
   2968 	ki->p_nlwps = p->p_nlwps;
   2969 	ki->p_realflag = ki->p_flag;
   2970 
   2971 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
   2972 		ki->p_vm_rssize = 0;
   2973 		ki->p_vm_tsize = 0;
   2974 		ki->p_vm_dsize = 0;
   2975 		ki->p_vm_ssize = 0;
   2976 		ki->p_nrlwps = 0;
   2977 		l = NULL;
   2978 	} else {
   2979 		struct vmspace *vm = p->p_vmspace;
   2980 		int tmp;
   2981 
   2982 		ki->p_vm_rssize = vm_resident_count(vm);
   2983 		ki->p_vm_tsize = vm->vm_tsize;
   2984 		ki->p_vm_dsize = vm->vm_dsize;
   2985 		ki->p_vm_ssize = vm->vm_ssize;
   2986 
   2987 		/* Pick a "representative" LWP */
   2988 		l = proc_representative_lwp(p, &tmp, 1);
   2989 		lwp_lock(l);
   2990 		ki->p_nrlwps = tmp;
   2991 		ki->p_forw = 0;
   2992 		ki->p_back = 0;
   2993 		ki->p_addr = PTRTOUINT64(l->l_addr);
   2994 		ki->p_stat = l->l_stat;
   2995 		ki->p_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
   2996 		ki->p_swtime = l->l_swtime;
   2997 		ki->p_slptime = l->l_slptime;
   2998 		if (l->l_stat == LSONPROC)
   2999 			ki->p_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   3000 		else
   3001 			ki->p_schedflags = 0;
   3002 		ki->p_holdcnt = l->l_holdcnt;
   3003 		ki->p_priority = lwp_eprio(l);
   3004 		ki->p_usrpri = l->l_priority;
   3005 		if (l->l_wmesg)
   3006 			strncpy(ki->p_wmesg, l->l_wmesg, sizeof(ki->p_wmesg));
   3007 		ki->p_wchan = PTRTOUINT64(l->l_wchan);
   3008 		lwp_unlock(l);
   3009 	}
   3010 	if (p->p_session->s_ttyvp)
   3011 		ki->p_eflag |= EPROC_CTTY;
   3012 	if (SESS_LEADER(p))
   3013 		ki->p_eflag |= EPROC_SLEADER;
   3014 
   3015 	/* XXX Is this double check necessary? */
   3016 	if (P_ZOMBIE(p)) {
   3017 		ki->p_uvalid = 0;
   3018 		ki->p_rtime_sec = 0;
   3019 		ki->p_rtime_usec = 0;
   3020 	} else {
   3021 		ki->p_uvalid = 1;
   3022 
   3023 		ki->p_ustart_sec = p->p_stats->p_start.tv_sec;
   3024 		ki->p_ustart_usec = p->p_stats->p_start.tv_usec;
   3025 
   3026 		calcru(p, &ut, &st, NULL, &rt);
   3027 		ki->p_rtime_sec = rt.tv_sec;
   3028 		ki->p_rtime_usec = rt.tv_usec;
   3029 		ki->p_uutime_sec = ut.tv_sec;
   3030 		ki->p_uutime_usec = ut.tv_usec;
   3031 		ki->p_ustime_sec = st.tv_sec;
   3032 		ki->p_ustime_usec = st.tv_usec;
   3033 
   3034 		memcpy(&ru, &p->p_stats->p_ru, sizeof(ru));
   3035 		ki->p_uru_nvcsw = 0;
   3036 		ki->p_uru_nivcsw = 0;
   3037 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   3038 			ki->p_uru_nvcsw += (l->l_ncsw - l->l_nivcsw);
   3039 			ki->p_uru_nivcsw += l->l_nivcsw;
   3040 			ruadd(&ru, &l2->l_ru);
   3041 		}
   3042 		ki->p_uru_maxrss = ru.ru_maxrss;
   3043 		ki->p_uru_ixrss = ru.ru_ixrss;
   3044 		ki->p_uru_idrss = ru.ru_idrss;
   3045 		ki->p_uru_isrss = ru.ru_isrss;
   3046 		ki->p_uru_minflt = ru.ru_minflt;
   3047 		ki->p_uru_majflt = ru.ru_majflt;
   3048 		ki->p_uru_nswap = ru.ru_nswap;
   3049 		ki->p_uru_inblock = ru.ru_inblock;
   3050 		ki->p_uru_oublock = ru.ru_oublock;
   3051 		ki->p_uru_msgsnd = ru.ru_msgsnd;
   3052 		ki->p_uru_msgrcv = ru.ru_msgrcv;
   3053 		ki->p_uru_nsignals = ru.ru_nsignals;
   3054 
   3055 		timeradd(&p->p_stats->p_cru.ru_utime,
   3056 			 &p->p_stats->p_cru.ru_stime, &ut);
   3057 		ki->p_uctime_sec = ut.tv_sec;
   3058 		ki->p_uctime_usec = ut.tv_usec;
   3059 	}
   3060 	if (l != NULL)
   3061 		ki->p_cpuid = l->l_cpu->ci_cpuid;
   3062 
   3063 	mutex_exit(&p->p_smutex);
   3064 }
   3065 
   3066 /*
   3067  * Fill in a kinfo_lwp structure for the specified lwp.
   3068  */
   3069 static void
   3070 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
   3071 {
   3072 	struct proc *p = l->l_proc;
   3073 	struct timeval tv;
   3074 
   3075 	kl->l_forw = 0;
   3076 	kl->l_back = 0;
   3077 	kl->l_laddr = PTRTOUINT64(l);
   3078 	kl->l_addr = PTRTOUINT64(l->l_addr);
   3079 	kl->l_stat = l->l_stat;
   3080 	kl->l_lid = l->l_lid;
   3081 	kl->l_flag = sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
   3082 
   3083 	kl->l_swtime = l->l_swtime;
   3084 	kl->l_slptime = l->l_slptime;
   3085 	if (l->l_stat == LSONPROC)
   3086 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   3087 	else
   3088 		kl->l_schedflags = 0;
   3089 	kl->l_holdcnt = l->l_holdcnt;
   3090 	kl->l_priority = lwp_eprio(l);
   3091 	kl->l_usrpri = l->l_priority;
   3092 	if (l->l_wmesg)
   3093 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
   3094 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
   3095 	kl->l_cpuid = l->l_cpu->ci_cpuid;
   3096 	bintime2timeval(&l->l_rtime, &tv);
   3097 	kl->l_rtime_sec = tv.tv_sec;
   3098 	kl->l_rtime_usec = tv.tv_usec;
   3099 	kl->l_cpticks = l->l_cpticks;
   3100 	kl->l_pctcpu = l->l_pctcpu;
   3101 	kl->l_pid = p->p_pid;
   3102 	if (l->l_name == NULL)
   3103 		kl->l_name[0] = '\0';
   3104 	else
   3105 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
   3106 }
   3107 
   3108 /*
   3109  * Fill in an eproc structure for the specified process.
   3110  */
   3111 void
   3112 fill_eproc(struct proc *p, struct eproc *ep)
   3113 {
   3114 	struct tty *tp;
   3115 	struct lwp *l;
   3116 
   3117 	KASSERT(mutex_owned(&proclist_lock));
   3118 	KASSERT(mutex_owned(&p->p_mutex));
   3119 
   3120 	ep->e_paddr = p;
   3121 	ep->e_sess = p->p_session;
   3122 	kauth_cred_topcred(p->p_cred, &ep->e_pcred);
   3123 	kauth_cred_toucred(p->p_cred, &ep->e_ucred);
   3124 	if (p->p_stat == SIDL || P_ZOMBIE(p)) {
   3125 		ep->e_vm.vm_rssize = 0;
   3126 		ep->e_vm.vm_tsize = 0;
   3127 		ep->e_vm.vm_dsize = 0;
   3128 		ep->e_vm.vm_ssize = 0;
   3129 		/* ep->e_vm.vm_pmap = XXX; */
   3130 	} else {
   3131 		struct vmspace *vm = p->p_vmspace;
   3132 
   3133 		ep->e_vm.vm_rssize = vm_resident_count(vm);
   3134 		ep->e_vm.vm_tsize = vm->vm_tsize;
   3135 		ep->e_vm.vm_dsize = vm->vm_dsize;
   3136 		ep->e_vm.vm_ssize = vm->vm_ssize;
   3137 
   3138 		/* Pick a "representative" LWP */
   3139 		mutex_enter(&p->p_smutex);
   3140 		l = proc_representative_lwp(p, NULL, 1);
   3141 		lwp_lock(l);
   3142 		if (l->l_wmesg)
   3143 			strncpy(ep->e_wmesg, l->l_wmesg, WMESGLEN);
   3144 		lwp_unlock(l);
   3145 		mutex_exit(&p->p_smutex);
   3146 	}
   3147 	if (p->p_pptr)
   3148 		ep->e_ppid = p->p_pptr->p_pid;
   3149 	else
   3150 		ep->e_ppid = 0;
   3151 	ep->e_pgid = p->p_pgrp->pg_id;
   3152 	ep->e_sid = ep->e_sess->s_sid;
   3153 	ep->e_jobc = p->p_pgrp->pg_jobc;
   3154 	if ((p->p_lflag & PL_CONTROLT) &&
   3155 	    (tp = ep->e_sess->s_ttyp)) {
   3156 		ep->e_tdev = tp->t_dev;
   3157 		ep->e_tpgid = tp->t_pgrp ? tp->t_pgrp->pg_id : NO_PGID;
   3158 		ep->e_tsess = tp->t_session;
   3159 	} else
   3160 		ep->e_tdev = NODEV;
   3161 
   3162 	ep->e_xsize = ep->e_xrssize = 0;
   3163 	ep->e_xccount = ep->e_xswrss = 0;
   3164 	ep->e_flag = ep->e_sess->s_ttyvp ? EPROC_CTTY : 0;
   3165 	if (SESS_LEADER(p))
   3166 		ep->e_flag |= EPROC_SLEADER;
   3167 	strncpy(ep->e_login, ep->e_sess->s_login, MAXLOGNAME);
   3168 }
   3169 
   3170 u_int
   3171 sysctl_map_flags(const u_int *map, u_int word)
   3172 {
   3173 	u_int rv;
   3174 
   3175 	for (rv = 0; *map != 0; map += 2)
   3176 		if ((word & map[0]) != 0)
   3177 			rv |= map[1];
   3178 
   3179 	return rv;
   3180 }
   3181