Home | History | Annotate | Line # | Download | only in kern
init_sysctl.c revision 1.193
      1 /*	$NetBSD: init_sysctl.c,v 1.193 2012/10/27 17:18:39 chs Exp $ */
      2 
      3 /*-
      4  * Copyright (c) 2003, 2007, 2008, 2009 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  *
     19  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     20  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     21  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     22  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     23  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     24  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     25  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     26  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     27  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     28  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     29  * POSSIBILITY OF SUCH DAMAGE.
     30  */
     31 
     32 #include <sys/cdefs.h>
     33 __KERNEL_RCSID(0, "$NetBSD: init_sysctl.c,v 1.193 2012/10/27 17:18:39 chs Exp $");
     34 
     35 #include "opt_sysv.h"
     36 #include "opt_compat_netbsd.h"
     37 #include "opt_modular.h"
     38 #include "pty.h"
     39 
     40 #include <sys/types.h>
     41 #include <sys/param.h>
     42 #include <sys/sysctl.h>
     43 #include <sys/cpu.h>
     44 #include <sys/errno.h>
     45 #include <sys/systm.h>
     46 #include <sys/kernel.h>
     47 #include <sys/unistd.h>
     48 #include <sys/disklabel.h>
     49 #include <sys/cprng.h>
     50 #include <sys/vnode.h>
     51 #include <sys/mount.h>
     52 #include <sys/namei.h>
     53 #include <sys/msgbuf.h>
     54 #include <dev/cons.h>
     55 #include <sys/socketvar.h>
     56 #include <sys/file.h>
     57 #include <sys/filedesc.h>
     58 #include <sys/tty.h>
     59 #include <sys/kmem.h>
     60 #include <sys/resource.h>
     61 #include <sys/resourcevar.h>
     62 #include <sys/exec.h>
     63 #include <sys/conf.h>
     64 #include <sys/device.h>
     65 #include <sys/stat.h>
     66 #include <sys/kauth.h>
     67 #include <sys/ktrace.h>
     68 #include <sys/ksem.h>
     69 
     70 #ifdef COMPAT_50
     71 #include <compat/sys/time.h>
     72 #endif
     73 
     74 #include <sys/cpu.h>
     75 
     76 int security_setidcore_dump;
     77 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
     78 uid_t security_setidcore_owner = 0;
     79 gid_t security_setidcore_group = 0;
     80 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
     81 
     82 static const u_int sysctl_flagmap[] = {
     83 	PK_ADVLOCK, P_ADVLOCK,
     84 	PK_EXEC, P_EXEC,
     85 	PK_NOCLDWAIT, P_NOCLDWAIT,
     86 	PK_32, P_32,
     87 	PK_CLDSIGIGN, P_CLDSIGIGN,
     88 	PK_SUGID, P_SUGID,
     89 	0
     90 };
     91 
     92 static const u_int sysctl_sflagmap[] = {
     93 	PS_NOCLDSTOP, P_NOCLDSTOP,
     94 	PS_WEXIT, P_WEXIT,
     95 	PS_STOPFORK, P_STOPFORK,
     96 	PS_STOPEXEC, P_STOPEXEC,
     97 	PS_STOPEXIT, P_STOPEXIT,
     98 	0
     99 };
    100 
    101 static const u_int sysctl_slflagmap[] = {
    102 	PSL_TRACED, P_TRACED,
    103 	PSL_FSTRACE, P_FSTRACE,
    104 	PSL_CHTRACED, P_CHTRACED,
    105 	PSL_SYSCALL, P_SYSCALL,
    106 	0
    107 };
    108 
    109 static const u_int sysctl_lflagmap[] = {
    110 	PL_CONTROLT, P_CONTROLT,
    111 	PL_PPWAIT, P_PPWAIT,
    112 	0
    113 };
    114 
    115 static const u_int sysctl_stflagmap[] = {
    116 	PST_PROFIL, P_PROFIL,
    117 	0
    118 
    119 };
    120 
    121 static const u_int sysctl_lwpprflagmap[] = {
    122 	LPR_DETACHED, L_DETACHED,
    123 	0
    124 };
    125 
    126 /*
    127  * try over estimating by 5 procs/lwps
    128  */
    129 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
    130 
    131 static int dcopyout(struct lwp *, const void *, void *, size_t);
    132 
    133 static int
    134 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
    135 {
    136 	int error;
    137 
    138 	error = copyout(kaddr, uaddr, len);
    139 	ktrmibio(-1, UIO_READ, uaddr, len, error);
    140 
    141 	return error;
    142 }
    143 
    144 #ifdef DIAGNOSTIC
    145 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
    146 #endif
    147 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
    148 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
    149 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
    150 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
    151 static int sysctl_setlen(SYSCTLFN_PROTO);
    152 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
    153 static int sysctl_msgbuf(SYSCTLFN_PROTO);
    154 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
    155 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
    156 #if NPTY > 0
    157 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
    158 #endif /* NPTY > 0 */
    159 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
    160 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
    161 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
    162 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
    163 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
    164 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
    165 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
    166 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
    167 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
    168 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
    169 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
    170 
    171 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
    172 
    173 /*
    174  * ********************************************************************
    175  * section 1: setup routines
    176  * ********************************************************************
    177  * These functions are stuffed into a link set for sysctl setup
    178  * functions. They're never called or referenced from anywhere else.
    179  * ********************************************************************
    180  */
    181 
    182 /*
    183  * this setup routine is a replacement for kern_sysctl()
    184  */
    185 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
    186 {
    187 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
    188 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
    189 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
    190 	const struct sysctlnode *rnode;
    191 
    192 	sysctl_createv(clog, 0, NULL, NULL,
    193 		       CTLFLAG_PERMANENT,
    194 		       CTLTYPE_NODE, "kern", NULL,
    195 		       NULL, 0, NULL, 0,
    196 		       CTL_KERN, CTL_EOL);
    197 
    198 	sysctl_createv(clog, 0, NULL, NULL,
    199 		       CTLFLAG_PERMANENT,
    200 		       CTLTYPE_STRING, "ostype",
    201 		       SYSCTL_DESCR("Operating system type"),
    202 		       NULL, 0, __UNCONST(&ostype), 0,
    203 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
    204 	sysctl_createv(clog, 0, NULL, NULL,
    205 		       CTLFLAG_PERMANENT,
    206 		       CTLTYPE_STRING, "osrelease",
    207 		       SYSCTL_DESCR("Operating system release"),
    208 		       NULL, 0, __UNCONST(&osrelease), 0,
    209 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
    210 	sysctl_createv(clog, 0, NULL, NULL,
    211 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    212 		       CTLTYPE_INT, "osrevision",
    213 		       SYSCTL_DESCR("Operating system revision"),
    214 		       NULL, __NetBSD_Version__, NULL, 0,
    215 		       CTL_KERN, KERN_OSREV, CTL_EOL);
    216 	sysctl_createv(clog, 0, NULL, NULL,
    217 		       CTLFLAG_PERMANENT,
    218 		       CTLTYPE_STRING, "version",
    219 		       SYSCTL_DESCR("Kernel version"),
    220 		       NULL, 0, __UNCONST(&version), 0,
    221 		       CTL_KERN, KERN_VERSION, CTL_EOL);
    222 	sysctl_createv(clog, 0, NULL, NULL,
    223 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    224 		       CTLTYPE_INT, "maxvnodes",
    225 		       SYSCTL_DESCR("Maximum number of vnodes"),
    226 		       sysctl_kern_maxvnodes, 0, NULL, 0,
    227 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
    228 	sysctl_createv(clog, 0, NULL, NULL,
    229 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    230 		       CTLTYPE_INT, "maxproc",
    231 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
    232 		       sysctl_kern_maxproc, 0, NULL, 0,
    233 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
    234 	sysctl_createv(clog, 0, NULL, NULL,
    235 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    236 		       CTLTYPE_INT, "maxfiles",
    237 		       SYSCTL_DESCR("Maximum number of open files"),
    238 		       NULL, 0, &maxfiles, 0,
    239 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
    240 	sysctl_createv(clog, 0, NULL, NULL,
    241 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    242 		       CTLTYPE_INT, "argmax",
    243 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
    244 				    "execve(2)"),
    245 		       NULL, ARG_MAX, NULL, 0,
    246 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
    247 	sysctl_createv(clog, 0, NULL, NULL,
    248 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    249 		       CTLTYPE_STRING, "hostname",
    250 		       SYSCTL_DESCR("System hostname"),
    251 		       sysctl_setlen, 0, hostname, MAXHOSTNAMELEN,
    252 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
    253 	sysctl_createv(clog, 0, NULL, NULL,
    254 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    255 		       CTLTYPE_INT, "hostid",
    256 		       SYSCTL_DESCR("System host ID number"),
    257 		       sysctl_kern_hostid, 0, NULL, 0,
    258 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
    259 	sysctl_createv(clog, 0, NULL, NULL,
    260 		       CTLFLAG_PERMANENT,
    261 		       CTLTYPE_STRUCT, "clockrate",
    262 		       SYSCTL_DESCR("Kernel clock rates"),
    263 		       sysctl_kern_clockrate, 0, NULL,
    264 		       sizeof(struct clockinfo),
    265 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
    266 	sysctl_createv(clog, 0, NULL, NULL,
    267 		       CTLFLAG_PERMANENT,
    268 		       CTLTYPE_INT, "hardclock_ticks",
    269 		       SYSCTL_DESCR("Number of hardclock ticks"),
    270 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
    271 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
    272 	sysctl_createv(clog, 0, NULL, NULL,
    273 		       CTLFLAG_PERMANENT,
    274 		       CTLTYPE_STRUCT, "vnode",
    275 		       SYSCTL_DESCR("System vnode table"),
    276 		       sysctl_kern_vnode, 0, NULL, 0,
    277 		       CTL_KERN, KERN_VNODE, CTL_EOL);
    278 #ifndef GPROF
    279 	sysctl_createv(clog, 0, NULL, NULL,
    280 		       CTLFLAG_PERMANENT,
    281 		       CTLTYPE_NODE, "profiling",
    282 		       SYSCTL_DESCR("Profiling information (not available)"),
    283 		       sysctl_notavail, 0, NULL, 0,
    284 		       CTL_KERN, KERN_PROF, CTL_EOL);
    285 #endif
    286 	sysctl_createv(clog, 0, NULL, NULL,
    287 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    288 		       CTLTYPE_INT, "posix1version",
    289 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
    290 				    "with which the operating system attempts "
    291 				    "to comply"),
    292 		       NULL, _POSIX_VERSION, NULL, 0,
    293 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
    294 	sysctl_createv(clog, 0, NULL, NULL,
    295 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    296 		       CTLTYPE_INT, "ngroups",
    297 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
    298 		       NULL, NGROUPS_MAX, NULL, 0,
    299 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
    300 	sysctl_createv(clog, 0, NULL, NULL,
    301 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    302 		       CTLTYPE_INT, "job_control",
    303 		       SYSCTL_DESCR("Whether job control is available"),
    304 		       NULL, 1, NULL, 0,
    305 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
    306 	sysctl_createv(clog, 0, NULL, NULL,
    307 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    308 		       CTLTYPE_INT, "saved_ids",
    309 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
    310 				    "available"), NULL,
    311 #ifdef _POSIX_SAVED_IDS
    312 		       1,
    313 #else /* _POSIX_SAVED_IDS */
    314 		       0,
    315 #endif /* _POSIX_SAVED_IDS */
    316 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
    317 	sysctl_createv(clog, 0, NULL, NULL,
    318 		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
    319 		       CTLTYPE_INT, "boothowto",
    320 		       SYSCTL_DESCR("Flags from boot loader"),
    321 		       NULL, 0, &boothowto, sizeof(boothowto),
    322 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    323 	sysctl_createv(clog, 0, NULL, NULL,
    324 		       CTLFLAG_PERMANENT,
    325 		       CTLTYPE_STRUCT, "boottime",
    326 		       SYSCTL_DESCR("System boot time"),
    327 		       NULL, 0, &boottime, sizeof(boottime),
    328 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
    329 #ifdef COMPAT_50
    330 	{
    331 		extern struct timeval50 boottime50;
    332 		sysctl_createv(clog, 0, NULL, NULL,
    333 			       CTLFLAG_PERMANENT,
    334 			       CTLTYPE_STRUCT, "oboottime",
    335 			       SYSCTL_DESCR("System boot time"),
    336 			       NULL, 0, &boottime50, sizeof(boottime50),
    337 			       CTL_KERN, KERN_OBOOTTIME, CTL_EOL);
    338 	}
    339 #endif
    340 	sysctl_createv(clog, 0, NULL, NULL,
    341 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    342 		       CTLTYPE_STRING, "domainname",
    343 		       SYSCTL_DESCR("YP domain name"),
    344 		       sysctl_setlen, 0, domainname, MAXHOSTNAMELEN,
    345 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
    346 	sysctl_createv(clog, 0, NULL, NULL,
    347 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    348 		       CTLTYPE_INT, "maxpartitions",
    349 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
    350 				    "disk"),
    351 		       NULL, MAXPARTITIONS, NULL, 0,
    352 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
    353 	sysctl_createv(clog, 0, NULL, NULL,
    354 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    355 		       CTLTYPE_INT, "rawpartition",
    356 		       SYSCTL_DESCR("Raw partition of a disk"),
    357 		       NULL, RAW_PART, NULL, 0,
    358 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
    359 	sysctl_createv(clog, 0, NULL, NULL,
    360 		       CTLFLAG_PERMANENT,
    361 		       CTLTYPE_STRUCT, "timex", NULL,
    362 		       sysctl_notavail, 0, NULL, 0,
    363 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
    364 	sysctl_createv(clog, 0, NULL, NULL,
    365 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    366 		       CTLTYPE_INT, "rtc_offset",
    367 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
    368 				    "minutes"),
    369 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
    370 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
    371 	sysctl_createv(clog, 0, NULL, NULL,
    372 		       CTLFLAG_PERMANENT,
    373 		       CTLTYPE_STRING, "root_device",
    374 		       SYSCTL_DESCR("Name of the root device"),
    375 		       sysctl_root_device, 0, NULL, 0,
    376 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
    377 	sysctl_createv(clog, 0, NULL, NULL,
    378 		       CTLFLAG_PERMANENT,
    379 		       CTLTYPE_INT, "msgbufsize",
    380 		       SYSCTL_DESCR("Size of the kernel message buffer"),
    381 		       sysctl_msgbuf, 0, NULL, 0,
    382 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
    383 	sysctl_createv(clog, 0, NULL, NULL,
    384 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    385 		       CTLTYPE_INT, "fsync",
    386 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
    387 				    "Synchronization Option is available on "
    388 				    "this system"),
    389 		       NULL, 1, NULL, 0,
    390 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
    391 	sysctl_createv(clog, 0, NULL, NULL,
    392 		       CTLFLAG_PERMANENT,
    393 		       CTLTYPE_NODE, "ipc",
    394 		       SYSCTL_DESCR("SysV IPC options"),
    395 		       NULL, 0, NULL, 0,
    396 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
    397 	sysctl_createv(clog, 0, NULL, NULL,
    398 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    399 		       CTLTYPE_INT, "sysvmsg",
    400 		       SYSCTL_DESCR("System V style message support available"),
    401 		       NULL,
    402 #ifdef SYSVMSG
    403 		       1,
    404 #else /* SYSVMSG */
    405 		       0,
    406 #endif /* SYSVMSG */
    407 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
    408 	sysctl_createv(clog, 0, NULL, NULL,
    409 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    410 		       CTLTYPE_INT, "sysvsem",
    411 		       SYSCTL_DESCR("System V style semaphore support "
    412 				    "available"), NULL,
    413 #ifdef SYSVSEM
    414 		       1,
    415 #else /* SYSVSEM */
    416 		       0,
    417 #endif /* SYSVSEM */
    418 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
    419 	sysctl_createv(clog, 0, NULL, NULL,
    420 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    421 		       CTLTYPE_INT, "sysvshm",
    422 		       SYSCTL_DESCR("System V style shared memory support "
    423 				    "available"), NULL,
    424 #ifdef SYSVSHM
    425 		       1,
    426 #else /* SYSVSHM */
    427 		       0,
    428 #endif /* SYSVSHM */
    429 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
    430 	sysctl_createv(clog, 0, NULL, NULL,
    431 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    432 		       CTLTYPE_INT, "synchronized_io",
    433 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
    434 				    "I/O Option is available on this system"),
    435 		       NULL, 1, NULL, 0,
    436 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
    437 	sysctl_createv(clog, 0, NULL, NULL,
    438 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    439 		       CTLTYPE_INT, "iov_max",
    440 		       SYSCTL_DESCR("Maximum number of iovec structures per "
    441 				    "process"),
    442 		       NULL, IOV_MAX, NULL, 0,
    443 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
    444 	sysctl_createv(clog, 0, NULL, NULL,
    445 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    446 		       CTLTYPE_INT, "mapped_files",
    447 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
    448 				    "Files Option is available on this system"),
    449 		       NULL, 1, NULL, 0,
    450 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
    451 	sysctl_createv(clog, 0, NULL, NULL,
    452 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    453 		       CTLTYPE_INT, "memlock",
    454 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
    455 				    "Locking Option is available on this "
    456 				    "system"),
    457 		       NULL, 1, NULL, 0,
    458 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
    459 	sysctl_createv(clog, 0, NULL, NULL,
    460 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    461 		       CTLTYPE_INT, "memlock_range",
    462 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
    463 				    "Locking Option is available on this "
    464 				    "system"),
    465 		       NULL, 1, NULL, 0,
    466 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
    467 	sysctl_createv(clog, 0, NULL, NULL,
    468 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    469 		       CTLTYPE_INT, "memory_protection",
    470 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
    471 				    "Protection Option is available on this "
    472 				    "system"),
    473 		       NULL, 1, NULL, 0,
    474 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
    475 	sysctl_createv(clog, 0, NULL, NULL,
    476 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    477 		       CTLTYPE_INT, "login_name_max",
    478 		       SYSCTL_DESCR("Maximum login name length"),
    479 		       NULL, LOGIN_NAME_MAX, NULL, 0,
    480 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
    481 	sysctl_createv(clog, 0, NULL, NULL,
    482 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    483 		       CTLTYPE_STRING, "defcorename",
    484 		       SYSCTL_DESCR("Default core file name"),
    485 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
    486 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
    487 	sysctl_createv(clog, 0, NULL, NULL,
    488 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    489 		       CTLTYPE_INT, "logsigexit",
    490 		       SYSCTL_DESCR("Log process exit when caused by signals"),
    491 		       NULL, 0, &kern_logsigexit, 0,
    492 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
    493 	sysctl_createv(clog, 0, NULL, NULL,
    494 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    495 		       CTLTYPE_INT, "fscale",
    496 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
    497 		       NULL, FSCALE, NULL, 0,
    498 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
    499 	sysctl_createv(clog, 0, NULL, NULL,
    500 		       CTLFLAG_PERMANENT,
    501 		       CTLTYPE_INT, "ccpu",
    502 		       SYSCTL_DESCR("Scheduler exponential decay value"),
    503 		       NULL, 0, &ccpu, 0,
    504 		       CTL_KERN, KERN_CCPU, CTL_EOL);
    505 	sysctl_createv(clog, 0, NULL, NULL,
    506 		       CTLFLAG_PERMANENT,
    507 		       CTLTYPE_STRUCT, "cp_time",
    508 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
    509 		       sysctl_kern_cptime, 0, NULL, 0,
    510 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
    511 	sysctl_createv(clog, 0, NULL, NULL,
    512 		       CTLFLAG_PERMANENT,
    513 		       CTLTYPE_INT, "msgbuf",
    514 		       SYSCTL_DESCR("Kernel message buffer"),
    515 		       sysctl_msgbuf, 0, NULL, 0,
    516 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
    517 	sysctl_createv(clog, 0, NULL, NULL,
    518 		       CTLFLAG_PERMANENT,
    519 		       CTLTYPE_STRUCT, "consdev",
    520 		       SYSCTL_DESCR("Console device"),
    521 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
    522 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
    523 #if NPTY > 0
    524 	sysctl_createv(clog, 0, NULL, NULL,
    525 		       CTLFLAG_PERMANENT,
    526 		       CTLTYPE_INT, "maxptys",
    527 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
    528 		       sysctl_kern_maxptys, 0, NULL, 0,
    529 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
    530 #endif /* NPTY > 0 */
    531 	sysctl_createv(clog, 0, NULL, NULL,
    532 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    533 		       CTLTYPE_INT, "maxphys",
    534 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
    535 		       NULL, MAXPHYS, NULL, 0,
    536 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
    537 	sysctl_createv(clog, 0, NULL, NULL,
    538 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    539 		       CTLTYPE_INT, "monotonic_clock",
    540 		       SYSCTL_DESCR("Implementation version of the POSIX "
    541 				    "1003.1b Monotonic Clock Option"),
    542 		       /* XXX _POSIX_VERSION */
    543 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
    544 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
    545 	sysctl_createv(clog, 0, NULL, NULL,
    546 		       CTLFLAG_PERMANENT,
    547 		       CTLTYPE_INT, "urandom",
    548 		       SYSCTL_DESCR("Random integer value"),
    549 		       sysctl_kern_urnd, 0, NULL, 0,
    550 		       CTL_KERN, KERN_URND, CTL_EOL);
    551 	sysctl_createv(clog, 0, NULL, NULL,
    552 		       CTLFLAG_PERMANENT,
    553 		       CTLTYPE_INT, "arandom",
    554 		       SYSCTL_DESCR("n bytes of random data"),
    555 		       sysctl_kern_arnd, 0, NULL, 0,
    556 		       CTL_KERN, KERN_ARND, CTL_EOL);
    557 	sysctl_createv(clog, 0, NULL, NULL,
    558 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    559 		       CTLTYPE_INT, "labelsector",
    560 		       SYSCTL_DESCR("Sector number containing the disklabel"),
    561 		       NULL, LABELSECTOR, NULL, 0,
    562 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
    563 	sysctl_createv(clog, 0, NULL, NULL,
    564 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    565 		       CTLTYPE_INT, "labeloffset",
    566 		       SYSCTL_DESCR("Offset of the disklabel within the "
    567 				    "sector"),
    568 		       NULL, LABELOFFSET, NULL, 0,
    569 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
    570 	sysctl_createv(clog, 0, NULL, NULL,
    571 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    572 		       CTLTYPE_INT, "labelusesmbr",
    573 		       SYSCTL_DESCR("disklabel is inside MBR partition"),
    574 		       NULL, LABELUSESMBR, NULL, 0,
    575 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    576 	sysctl_createv(clog, 0, NULL, NULL,
    577 		       CTLFLAG_PERMANENT,
    578 		       CTLTYPE_NODE, "lwp",
    579 		       SYSCTL_DESCR("System-wide LWP information"),
    580 		       sysctl_kern_lwp, 0, NULL, 0,
    581 		       CTL_KERN, KERN_LWP, CTL_EOL);
    582 	sysctl_createv(clog, 0, NULL, NULL,
    583 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    584 		       CTLTYPE_INT, "forkfsleep",
    585 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
    586 				    "to process limits"),
    587 		       sysctl_kern_forkfsleep, 0, NULL, 0,
    588 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
    589 	sysctl_createv(clog, 0, NULL, NULL,
    590 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    591 		       CTLTYPE_INT, "posix_threads",
    592 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    593 				    "Threads option to which the system "
    594 				    "attempts to conform"),
    595 		       /* XXX _POSIX_VERSION */
    596 		       NULL, _POSIX_THREADS, NULL, 0,
    597 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
    598 	sysctl_createv(clog, 0, NULL, NULL,
    599 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    600 		       CTLTYPE_INT, "posix_semaphores",
    601 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    602 				    "Semaphores option to which the system "
    603 				    "attempts to conform"), NULL,
    604 		       200112, NULL, 0,
    605 		       CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
    606 	sysctl_createv(clog, 0, NULL, NULL,
    607 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    608 		       CTLTYPE_INT, "posix_barriers",
    609 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    610 				    "Barriers option to which the system "
    611 				    "attempts to conform"),
    612 		       /* XXX _POSIX_VERSION */
    613 		       NULL, _POSIX_BARRIERS, NULL, 0,
    614 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
    615 	sysctl_createv(clog, 0, NULL, NULL,
    616 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    617 		       CTLTYPE_INT, "posix_timers",
    618 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    619 				    "Timers option to which the system "
    620 				    "attempts to conform"),
    621 		       /* XXX _POSIX_VERSION */
    622 		       NULL, _POSIX_TIMERS, NULL, 0,
    623 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
    624 	sysctl_createv(clog, 0, NULL, NULL,
    625 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    626 		       CTLTYPE_INT, "posix_spin_locks",
    627 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
    628 				    "Locks option to which the system attempts "
    629 				    "to conform"),
    630 		       /* XXX _POSIX_VERSION */
    631 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
    632 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
    633 	sysctl_createv(clog, 0, NULL, NULL,
    634 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    635 		       CTLTYPE_INT, "posix_reader_writer_locks",
    636 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    637 				    "Read-Write Locks option to which the "
    638 				    "system attempts to conform"),
    639 		       /* XXX _POSIX_VERSION */
    640 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
    641 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
    642 	sysctl_createv(clog, 0, NULL, NULL,
    643 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    644 		       CTLTYPE_INT, "dump_on_panic",
    645 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
    646 		       NULL, 0, &dumponpanic, 0,
    647 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
    648 #ifdef DIAGNOSTIC
    649 	sysctl_createv(clog, 0, NULL, NULL,
    650 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    651 		       CTLTYPE_INT, "panic_now",
    652 		       SYSCTL_DESCR("Trigger a panic"),
    653 		       sysctl_kern_trigger_panic, 0, NULL, 0,
    654 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    655 #endif
    656 	sysctl_createv(clog, 0, NULL, NULL,
    657 		       CTLFLAG_PERMANENT,
    658 		       CTLTYPE_INT, "root_partition",
    659 		       SYSCTL_DESCR("Root partition on the root device"),
    660 		       sysctl_kern_root_partition, 0, NULL, 0,
    661 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
    662 	sysctl_createv(clog, 0, NULL, NULL,
    663 		       CTLFLAG_PERMANENT,
    664 		       CTLTYPE_STRUCT, "drivers",
    665 		       SYSCTL_DESCR("List of all drivers with block and "
    666 				    "character device numbers"),
    667 		       sysctl_kern_drivers, 0, NULL, 0,
    668 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
    669 	sysctl_createv(clog, 0, NULL, NULL,
    670 		       CTLFLAG_PERMANENT,
    671 		       CTLTYPE_STRUCT, "cp_id",
    672 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
    673 		       sysctl_kern_cpid, 0, NULL, 0,
    674 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
    675 	sysctl_createv(clog, 0, NULL, &rnode,
    676 		       CTLFLAG_PERMANENT,
    677 		       CTLTYPE_NODE, "coredump",
    678 		       SYSCTL_DESCR("Coredump settings."),
    679 		       NULL, 0, NULL, 0,
    680 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    681 	sysctl_createv(clog, 0, &rnode, &rnode,
    682 		       CTLFLAG_PERMANENT,
    683 		       CTLTYPE_NODE, "setid",
    684 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
    685 		       NULL, 0, NULL, 0,
    686 		       CTL_CREATE, CTL_EOL);
    687 	sysctl_createv(clog, 0, &rnode, NULL,
    688 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    689 		       CTLTYPE_INT, "dump",
    690 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
    691 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
    692 		       sizeof(security_setidcore_dump),
    693 		       CTL_CREATE, CTL_EOL);
    694 	sysctl_createv(clog, 0, &rnode, NULL,
    695 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    696 		       CTLTYPE_STRING, "path",
    697 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
    698 		       sysctl_security_setidcorename, 0,
    699 		       security_setidcore_path,
    700 		       sizeof(security_setidcore_path),
    701 		       CTL_CREATE, CTL_EOL);
    702 	sysctl_createv(clog, 0, &rnode, NULL,
    703 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    704 		       CTLTYPE_INT, "owner",
    705 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
    706 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
    707 		       0,
    708 		       CTL_CREATE, CTL_EOL);
    709 	sysctl_createv(clog, 0, &rnode, NULL,
    710 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    711 		       CTLTYPE_INT, "group",
    712 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
    713 		       sysctl_security_setidcore, 0, &security_setidcore_group,
    714 		       0,
    715 		       CTL_CREATE, CTL_EOL);
    716 	sysctl_createv(clog, 0, &rnode, NULL,
    717 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    718 		       CTLTYPE_INT, "mode",
    719 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
    720 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
    721 		       0,
    722 		       CTL_CREATE, CTL_EOL);
    723 	sysctl_createv(clog, 0, NULL, NULL,
    724 		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    725 		       CTLTYPE_INT, "no_sa_support",
    726 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
    727 		       "it doesn't"),
    728 		       NULL, 1, NULL, 0,
    729 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    730 	/* kern.posix. */
    731 	sysctl_createv(clog, 0, NULL, &rnode,
    732 			CTLFLAG_PERMANENT,
    733 			CTLTYPE_NODE, "posix",
    734 			SYSCTL_DESCR("POSIX options"),
    735 			NULL, 0, NULL, 0,
    736 			CTL_KERN, CTL_CREATE, CTL_EOL);
    737 	sysctl_createv(clog, 0, &rnode, NULL,
    738 			CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
    739 			CTLTYPE_INT, "semmax",
    740 			SYSCTL_DESCR("Maximal number of semaphores"),
    741 			NULL, 0, &ksem_max, 0,
    742 			CTL_CREATE, CTL_EOL);
    743 }
    744 
    745 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
    746 {
    747 	u_int u;
    748 	u_quad_t q;
    749 
    750 	sysctl_createv(clog, 0, NULL, NULL,
    751 		       CTLFLAG_PERMANENT,
    752 		       CTLTYPE_NODE, "hw", NULL,
    753 		       NULL, 0, NULL, 0,
    754 		       CTL_HW, CTL_EOL);
    755 
    756 	sysctl_createv(clog, 0, NULL, NULL,
    757 		       CTLFLAG_PERMANENT,
    758 		       CTLTYPE_STRING, "machine",
    759 		       SYSCTL_DESCR("Machine class"),
    760 		       NULL, 0, machine, 0,
    761 		       CTL_HW, HW_MACHINE, CTL_EOL);
    762 	sysctl_createv(clog, 0, NULL, NULL,
    763 		       CTLFLAG_PERMANENT,
    764 		       CTLTYPE_STRING, "model",
    765 		       SYSCTL_DESCR("Machine model"),
    766 		       NULL, 0, cpu_model, 0,
    767 		       CTL_HW, HW_MODEL, CTL_EOL);
    768 	sysctl_createv(clog, 0, NULL, NULL,
    769 		       CTLFLAG_PERMANENT,
    770 		       CTLTYPE_INT, "ncpu",
    771 		       SYSCTL_DESCR("Number of CPUs configured"),
    772 		       NULL, 0, &ncpu, 0,
    773 		       CTL_HW, HW_NCPU, CTL_EOL);
    774 	sysctl_createv(clog, 0, NULL, NULL,
    775 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    776 		       CTLTYPE_INT, "byteorder",
    777 		       SYSCTL_DESCR("System byte order"),
    778 		       NULL, BYTE_ORDER, NULL, 0,
    779 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
    780 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
    781 		UINT_MAX : physmem * PAGE_SIZE;
    782 	sysctl_createv(clog, 0, NULL, NULL,
    783 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    784 		       CTLTYPE_INT, "physmem",
    785 		       SYSCTL_DESCR("Bytes of physical memory"),
    786 		       NULL, u, NULL, 0,
    787 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
    788 	sysctl_createv(clog, 0, NULL, NULL,
    789 		       CTLFLAG_PERMANENT,
    790 		       CTLTYPE_INT, "usermem",
    791 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    792 		       sysctl_hw_usermem, 0, NULL, 0,
    793 		       CTL_HW, HW_USERMEM, CTL_EOL);
    794 	sysctl_createv(clog, 0, NULL, NULL,
    795 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    796 		       CTLTYPE_INT, "pagesize",
    797 		       SYSCTL_DESCR("Software page size"),
    798 		       NULL, PAGE_SIZE, NULL, 0,
    799 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
    800 	sysctl_createv(clog, 0, NULL, NULL,
    801 		       CTLFLAG_PERMANENT,
    802 		       CTLTYPE_STRING, "machine_arch",
    803 		       SYSCTL_DESCR("Machine CPU class"),
    804 		       NULL, 0, machine_arch, 0,
    805 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
    806 	sysctl_createv(clog, 0, NULL, NULL,
    807 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    808 		       CTLTYPE_INT, "alignbytes",
    809 		       SYSCTL_DESCR("Alignment constraint for all possible "
    810 				    "data types"),
    811 		       NULL, ALIGNBYTES, NULL, 0,
    812 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
    813 	sysctl_createv(clog, 0, NULL, NULL,
    814 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    815 		       CTLTYPE_STRING, "cnmagic",
    816 		       SYSCTL_DESCR("Console magic key sequence"),
    817 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
    818 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
    819 	q = (u_quad_t)physmem * PAGE_SIZE;
    820 	sysctl_createv(clog, 0, NULL, NULL,
    821 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    822 		       CTLTYPE_QUAD, "physmem64",
    823 		       SYSCTL_DESCR("Bytes of physical memory"),
    824 		       NULL, q, NULL, 0,
    825 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
    826 	sysctl_createv(clog, 0, NULL, NULL,
    827 		       CTLFLAG_PERMANENT,
    828 		       CTLTYPE_QUAD, "usermem64",
    829 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    830 		       sysctl_hw_usermem, 0, NULL, 0,
    831 		       CTL_HW, HW_USERMEM64, CTL_EOL);
    832 	sysctl_createv(clog, 0, NULL, NULL,
    833 		       CTLFLAG_PERMANENT,
    834 		       CTLTYPE_INT, "ncpuonline",
    835 		       SYSCTL_DESCR("Number of CPUs online"),
    836 		       NULL, 0, &ncpuonline, 0,
    837 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
    838 }
    839 
    840 #ifdef DEBUG
    841 /*
    842  * Debugging related system variables.
    843  */
    844 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
    845 struct ctldebug debug5, debug6, debug7, debug8, debug9;
    846 struct ctldebug debug10, debug11, debug12, debug13, debug14;
    847 struct ctldebug debug15, debug16, debug17, debug18, debug19;
    848 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
    849 	&debug0, &debug1, &debug2, &debug3, &debug4,
    850 	&debug5, &debug6, &debug7, &debug8, &debug9,
    851 	&debug10, &debug11, &debug12, &debug13, &debug14,
    852 	&debug15, &debug16, &debug17, &debug18, &debug19,
    853 };
    854 
    855 /*
    856  * this setup routine is a replacement for debug_sysctl()
    857  *
    858  * note that it creates several nodes per defined debug variable
    859  */
    860 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
    861 {
    862 	struct ctldebug *cdp;
    863 	char nodename[20];
    864 	int i;
    865 
    866 	/*
    867 	 * two ways here:
    868 	 *
    869 	 * the "old" way (debug.name -> value) which was emulated by
    870 	 * the sysctl(8) binary
    871 	 *
    872 	 * the new way, which the sysctl(8) binary was actually using
    873 
    874 	 node	debug
    875 	 node	debug.0
    876 	 string debug.0.name
    877 	 int	debug.0.value
    878 	 int	debug.name
    879 
    880 	 */
    881 
    882 	sysctl_createv(clog, 0, NULL, NULL,
    883 		       CTLFLAG_PERMANENT,
    884 		       CTLTYPE_NODE, "debug", NULL,
    885 		       NULL, 0, NULL, 0,
    886 		       CTL_DEBUG, CTL_EOL);
    887 
    888 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
    889 		cdp = debugvars[i];
    890 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
    891 			continue;
    892 
    893 		snprintf(nodename, sizeof(nodename), "debug%d", i);
    894 		sysctl_createv(clog, 0, NULL, NULL,
    895 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    896 			       CTLTYPE_NODE, nodename, NULL,
    897 			       NULL, 0, NULL, 0,
    898 			       CTL_DEBUG, i, CTL_EOL);
    899 		sysctl_createv(clog, 0, NULL, NULL,
    900 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    901 			       CTLTYPE_STRING, "name", NULL,
    902 			       /*XXXUNCONST*/
    903 			       NULL, 0, __UNCONST(cdp->debugname), 0,
    904 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
    905 		sysctl_createv(clog, 0, NULL, NULL,
    906 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    907 			       CTLTYPE_INT, "value", NULL,
    908 			       NULL, 0, cdp->debugvar, 0,
    909 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
    910 		sysctl_createv(clog, 0, NULL, NULL,
    911 			       CTLFLAG_PERMANENT,
    912 			       CTLTYPE_INT, cdp->debugname, NULL,
    913 			       NULL, 0, cdp->debugvar, 0,
    914 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
    915 	}
    916 }
    917 #endif /* DEBUG */
    918 
    919 /*
    920  * ********************************************************************
    921  * section 2: private node-specific helper routines.
    922  * ********************************************************************
    923  */
    924 
    925 #ifdef DIAGNOSTIC
    926 static int
    927 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
    928 {
    929 	int newtrig, error;
    930 	struct sysctlnode node;
    931 
    932 	newtrig = 0;
    933 	node = *rnode;
    934 	node.sysctl_data = &newtrig;
    935 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
    936 	if (error || newp == NULL)
    937 		return (error);
    938 
    939 	if (newtrig != 0)
    940 		panic("Panic triggered");
    941 
    942 	return (error);
    943 }
    944 #endif
    945 
    946 /*
    947  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
    948  * new value is lower than desiredvnodes and then calls reinit
    949  * routines that needs to adjust to the new value.
    950  */
    951 static int
    952 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
    953 {
    954 	int error, new_vnodes, old_vnodes, new_max;
    955 	struct sysctlnode node;
    956 
    957 	new_vnodes = desiredvnodes;
    958 	node = *rnode;
    959 	node.sysctl_data = &new_vnodes;
    960 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
    961 	if (error || newp == NULL)
    962 		return (error);
    963 
    964 	/*
    965 	 * sysctl passes down unsigned values, require them
    966 	 * to be positive
    967 	 */
    968 	if (new_vnodes <= 0)
    969 		return (EINVAL);
    970 
    971 	/* Limits: 75% of KVA and physical memory. */
    972 	new_max = calc_cache_size(kernel_map, 75, 75) / VNODE_COST;
    973 	if (new_vnodes > new_max)
    974 		new_vnodes = new_max;
    975 
    976 	old_vnodes = desiredvnodes;
    977 	desiredvnodes = new_vnodes;
    978 	if (new_vnodes < old_vnodes) {
    979 		error = vfs_drainvnodes(new_vnodes);
    980 		if (error) {
    981 			desiredvnodes = old_vnodes;
    982 			return (error);
    983 		}
    984 	}
    985 	vfs_reinit();
    986 	nchreinit();
    987 
    988 	return (0);
    989 }
    990 
    991 /*
    992  * sysctl helper routine for rtc_offset - set time after changes
    993  */
    994 static int
    995 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
    996 {
    997 	struct timespec ts, delta;
    998 	int error, new_rtc_offset;
    999 	struct sysctlnode node;
   1000 
   1001 	new_rtc_offset = rtc_offset;
   1002 	node = *rnode;
   1003 	node.sysctl_data = &new_rtc_offset;
   1004 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1005 	if (error || newp == NULL)
   1006 		return (error);
   1007 
   1008 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
   1009 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
   1010 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
   1011 		return (EPERM);
   1012 	if (rtc_offset == new_rtc_offset)
   1013 		return (0);
   1014 
   1015 	/* if we change the offset, adjust the time */
   1016 	nanotime(&ts);
   1017 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
   1018 	delta.tv_nsec = 0;
   1019 	timespecadd(&ts, &delta, &ts);
   1020 	rtc_offset = new_rtc_offset;
   1021 	return (settime(l->l_proc, &ts));
   1022 }
   1023 
   1024 /*
   1025  * sysctl helper routine for kern.maxproc. Ensures that the new
   1026  * values are not too low or too high.
   1027  */
   1028 static int
   1029 sysctl_kern_maxproc(SYSCTLFN_ARGS)
   1030 {
   1031 	int error, nmaxproc;
   1032 	struct sysctlnode node;
   1033 
   1034 	nmaxproc = maxproc;
   1035 	node = *rnode;
   1036 	node.sysctl_data = &nmaxproc;
   1037 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1038 	if (error || newp == NULL)
   1039 		return (error);
   1040 
   1041 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
   1042 		return (EINVAL);
   1043 #ifdef __HAVE_CPU_MAXPROC
   1044 	if (nmaxproc > cpu_maxproc())
   1045 		return (EINVAL);
   1046 #endif
   1047 	maxproc = nmaxproc;
   1048 
   1049 	return (0);
   1050 }
   1051 
   1052 /*
   1053  * sysctl helper function for kern.hostid. The hostid is a long, but
   1054  * we export it as an int, so we need to give it a little help.
   1055  */
   1056 static int
   1057 sysctl_kern_hostid(SYSCTLFN_ARGS)
   1058 {
   1059 	int error, inthostid;
   1060 	struct sysctlnode node;
   1061 
   1062 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
   1063 	node = *rnode;
   1064 	node.sysctl_data = &inthostid;
   1065 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1066 	if (error || newp == NULL)
   1067 		return (error);
   1068 
   1069 	hostid = (unsigned)inthostid;
   1070 
   1071 	return (0);
   1072 }
   1073 
   1074 /*
   1075  * sysctl helper function for kern.hostname and kern.domainnname.
   1076  * resets the relevant recorded length when the underlying name is
   1077  * changed.
   1078  */
   1079 static int
   1080 sysctl_setlen(SYSCTLFN_ARGS)
   1081 {
   1082 	int error;
   1083 
   1084 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
   1085 	if (error || newp == NULL)
   1086 		return (error);
   1087 
   1088 	switch (rnode->sysctl_num) {
   1089 	case KERN_HOSTNAME:
   1090 		hostnamelen = strlen((const char*)rnode->sysctl_data);
   1091 		break;
   1092 	case KERN_DOMAINNAME:
   1093 		domainnamelen = strlen((const char*)rnode->sysctl_data);
   1094 		break;
   1095 	}
   1096 
   1097 	return (0);
   1098 }
   1099 
   1100 /*
   1101  * sysctl helper routine for kern.clockrate. Assembles a struct on
   1102  * the fly to be returned to the caller.
   1103  */
   1104 static int
   1105 sysctl_kern_clockrate(SYSCTLFN_ARGS)
   1106 {
   1107 	struct clockinfo clkinfo;
   1108 	struct sysctlnode node;
   1109 
   1110 	clkinfo.tick = tick;
   1111 	clkinfo.tickadj = tickadj;
   1112 	clkinfo.hz = hz;
   1113 	clkinfo.profhz = profhz;
   1114 	clkinfo.stathz = stathz ? stathz : hz;
   1115 
   1116 	node = *rnode;
   1117 	node.sysctl_data = &clkinfo;
   1118 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1119 }
   1120 
   1121 /*
   1122  * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
   1123  * former it merely checks the message buffer is set up. For the latter,
   1124  * it also copies out the data if necessary.
   1125  */
   1126 static int
   1127 sysctl_msgbuf(SYSCTLFN_ARGS)
   1128 {
   1129 	char *where = oldp;
   1130 	size_t len, maxlen;
   1131 	long beg, end;
   1132 	extern kmutex_t log_lock;
   1133 	int error;
   1134 
   1135 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
   1136 		msgbufenabled = 0;
   1137 		return (ENXIO);
   1138 	}
   1139 
   1140 	switch (rnode->sysctl_num) {
   1141 	case KERN_MSGBUFSIZE: {
   1142 		struct sysctlnode node = *rnode;
   1143 		int msg_bufs = (int)msgbufp->msg_bufs;
   1144 		node.sysctl_data = &msg_bufs;
   1145 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1146 	}
   1147 	case KERN_MSGBUF:
   1148 		break;
   1149 	default:
   1150 		return (EOPNOTSUPP);
   1151 	}
   1152 
   1153 	if (newp != NULL)
   1154 		return (EPERM);
   1155 
   1156 	if (oldp == NULL) {
   1157 		/* always return full buffer size */
   1158 		*oldlenp = msgbufp->msg_bufs;
   1159 		return (0);
   1160 	}
   1161 
   1162 	sysctl_unlock();
   1163 
   1164 	/*
   1165 	 * First, copy from the write pointer to the end of
   1166 	 * message buffer.
   1167 	 */
   1168 	error = 0;
   1169 	mutex_spin_enter(&log_lock);
   1170 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
   1171 	beg = msgbufp->msg_bufx;
   1172 	end = msgbufp->msg_bufs;
   1173 	mutex_spin_exit(&log_lock);
   1174 
   1175 	while (maxlen > 0) {
   1176 		len = MIN(end - beg, maxlen);
   1177 		if (len == 0)
   1178 			break;
   1179 		/* XXX unlocked, but hardly matters. */
   1180 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
   1181 		if (error)
   1182 			break;
   1183 		where += len;
   1184 		maxlen -= len;
   1185 
   1186 		/*
   1187 		 * ... then, copy from the beginning of message buffer to
   1188 		 * the write pointer.
   1189 		 */
   1190 		beg = 0;
   1191 		end = msgbufp->msg_bufx;
   1192 	}
   1193 
   1194 	sysctl_relock();
   1195 	return (error);
   1196 }
   1197 
   1198 /*
   1199  * sysctl helper routine for kern.defcorename. In the case of a new
   1200  * string being assigned, check that it's not a zero-length string.
   1201  * (XXX the check in -current doesn't work, but do we really care?)
   1202  */
   1203 static int
   1204 sysctl_kern_defcorename(SYSCTLFN_ARGS)
   1205 {
   1206 	int error;
   1207 	char *newcorename;
   1208 	struct sysctlnode node;
   1209 
   1210 	newcorename = PNBUF_GET();
   1211 	node = *rnode;
   1212 	node.sysctl_data = &newcorename[0];
   1213 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   1214 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1215 	if (error || newp == NULL) {
   1216 		goto done;
   1217 	}
   1218 
   1219 	/*
   1220 	 * when sysctl_lookup() deals with a string, it's guaranteed
   1221 	 * to come back nul terminated. So there.  :)
   1222 	 */
   1223 	if (strlen(newcorename) == 0) {
   1224 		error = EINVAL;
   1225 	} else {
   1226 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   1227 		error = 0;
   1228 	}
   1229 done:
   1230 	PNBUF_PUT(newcorename);
   1231 	return error;
   1232 }
   1233 
   1234 /*
   1235  * sysctl helper routine for kern.cp_time node. Adds up cpu time
   1236  * across all cpus.
   1237  */
   1238 static int
   1239 sysctl_kern_cptime(SYSCTLFN_ARGS)
   1240 {
   1241 	struct sysctlnode node = *rnode;
   1242 	uint64_t *cp_time = NULL;
   1243 	int error, n = ncpu, i;
   1244 	struct cpu_info *ci;
   1245 	CPU_INFO_ITERATOR cii;
   1246 
   1247 	/*
   1248 	 * if you specifically pass a buffer that is the size of the
   1249 	 * sum, or if you are probing for the size, you get the "sum"
   1250 	 * of cp_time (and the size thereof) across all processors.
   1251 	 *
   1252 	 * alternately, you can pass an additional mib number and get
   1253 	 * cp_time for that particular processor.
   1254 	 */
   1255 	switch (namelen) {
   1256 	case 0:
   1257 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
   1258 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
   1259 			n = -1; /* SUM */
   1260 		}
   1261 		else {
   1262 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
   1263 			n = -2; /* ALL */
   1264 		}
   1265 		break;
   1266 	case 1:
   1267 		if (name[0] < 0 || name[0] >= n)
   1268 			return (ENOENT); /* ENOSUCHPROCESSOR */
   1269 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
   1270 		n = name[0];
   1271 		/*
   1272 		 * adjust these so that sysctl_lookup() will be happy
   1273 		 */
   1274 		name++;
   1275 		namelen--;
   1276 		break;
   1277 	default:
   1278 		return (EINVAL);
   1279 	}
   1280 
   1281 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
   1282 	if (cp_time == NULL)
   1283 		return (ENOMEM);
   1284 	node.sysctl_data = cp_time;
   1285 	memset(cp_time, 0, node.sysctl_size);
   1286 
   1287 	for (CPU_INFO_FOREACH(cii, ci)) {
   1288 		if (n <= 0) {
   1289 			for (i = 0; i < CPUSTATES; i++) {
   1290 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
   1291 			}
   1292 		}
   1293 		/*
   1294 		 * if a specific processor was requested and we just
   1295 		 * did it, we're done here
   1296 		 */
   1297 		if (n == 0)
   1298 			break;
   1299 		/*
   1300 		 * if doing "all", skip to next cp_time set for next processor
   1301 		 */
   1302 		if (n == -2)
   1303 			cp_time += CPUSTATES;
   1304 		/*
   1305 		 * if we're doing a specific processor, we're one
   1306 		 * processor closer
   1307 		 */
   1308 		if (n > 0)
   1309 			n--;
   1310 	}
   1311 
   1312 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1313 	kmem_free(node.sysctl_data, node.sysctl_size);
   1314 	return (error);
   1315 }
   1316 
   1317 #if NPTY > 0
   1318 /*
   1319  * sysctl helper routine for kern.maxptys. Ensures that any new value
   1320  * is acceptable to the pty subsystem.
   1321  */
   1322 static int
   1323 sysctl_kern_maxptys(SYSCTLFN_ARGS)
   1324 {
   1325 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
   1326 	int error, xmax;
   1327 	struct sysctlnode node;
   1328 
   1329 	/* get current value of maxptys */
   1330 	xmax = pty_maxptys(0, 0);
   1331 
   1332 	node = *rnode;
   1333 	node.sysctl_data = &xmax;
   1334 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1335 	if (error || newp == NULL)
   1336 		return (error);
   1337 
   1338 	if (xmax != pty_maxptys(xmax, 1))
   1339 		return (EINVAL);
   1340 
   1341 	return (0);
   1342 }
   1343 #endif /* NPTY > 0 */
   1344 
   1345 /*
   1346  * sysctl helper routine for kern.urandom node. Picks a random number
   1347  * for you.
   1348  */
   1349 static int
   1350 sysctl_kern_urnd(SYSCTLFN_ARGS)
   1351 {
   1352 	int v, rv;
   1353 
   1354 	rv = cprng_strong(sysctl_prng, &v, sizeof(v), 0);
   1355 	if (rv == sizeof(v)) {
   1356 		struct sysctlnode node = *rnode;
   1357 		node.sysctl_data = &v;
   1358 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1359 	}
   1360 	else
   1361 		return (EIO);	/*XXX*/
   1362 }
   1363 
   1364 /*
   1365  * sysctl helper routine for kern.arandom node. Picks a random number
   1366  * for you.
   1367  */
   1368 static int
   1369 sysctl_kern_arnd(SYSCTLFN_ARGS)
   1370 {
   1371 	int error;
   1372 	void *v;
   1373 	struct sysctlnode node = *rnode;
   1374 
   1375 	if (*oldlenp == 0)
   1376 		return 0;
   1377 	/*
   1378 	 * This code used to allow sucking 8192 bytes at a time out
   1379 	 * of the kernel arc4random generator.  Evidently there is some
   1380 	 * very old OpenBSD application code that may try to do this.
   1381 	 *
   1382 	 * Note that this node is documented as type "INT" -- 4 or 8
   1383 	 * bytes, not 8192.
   1384 	 *
   1385 	 * We continue to support this abuse of the "len" pointer here
   1386 	 * but only 256 bytes at a time, as, anecdotally, the actual
   1387 	 * application use here was to generate RC4 keys in userspace.
   1388 	 *
   1389 	 * Support for such large requests will probably be removed
   1390 	 * entirely in the future.
   1391 	 */
   1392 	if (*oldlenp > 256)
   1393 		return E2BIG;
   1394 
   1395 	v = kmem_alloc(*oldlenp, KM_SLEEP);
   1396 	cprng_fast(v, *oldlenp);
   1397 	node.sysctl_data = v;
   1398 	node.sysctl_size = *oldlenp;
   1399 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1400 	kmem_free(v, *oldlenp);
   1401 	return error;
   1402 }
   1403 /*
   1404  * sysctl helper routine to do kern.lwp.* work.
   1405  */
   1406 static int
   1407 sysctl_kern_lwp(SYSCTLFN_ARGS)
   1408 {
   1409 	struct kinfo_lwp klwp;
   1410 	struct proc *p;
   1411 	struct lwp *l2, *l3;
   1412 	char *where, *dp;
   1413 	int pid, elem_size, elem_count;
   1414 	int buflen, needed, error;
   1415 	bool gotit;
   1416 
   1417 	if (namelen == 1 && name[0] == CTL_QUERY)
   1418 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   1419 
   1420 	dp = where = oldp;
   1421 	buflen = where != NULL ? *oldlenp : 0;
   1422 	error = needed = 0;
   1423 
   1424 	if (newp != NULL || namelen != 3)
   1425 		return (EINVAL);
   1426 	pid = name[0];
   1427 	elem_size = name[1];
   1428 	elem_count = name[2];
   1429 
   1430 	sysctl_unlock();
   1431 	if (pid == -1) {
   1432 		mutex_enter(proc_lock);
   1433 		PROCLIST_FOREACH(p, &allproc) {
   1434 			/* Grab a hold on the process. */
   1435 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   1436 				continue;
   1437 			}
   1438 			mutex_exit(proc_lock);
   1439 
   1440 			mutex_enter(p->p_lock);
   1441 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1442 				if (buflen >= elem_size && elem_count > 0) {
   1443 					lwp_lock(l2);
   1444 					fill_lwp(l2, &klwp);
   1445 					lwp_unlock(l2);
   1446 					mutex_exit(p->p_lock);
   1447 
   1448 					/*
   1449 					 * Copy out elem_size, but not
   1450 					 * larger than the size of a
   1451 					 * struct kinfo_proc2.
   1452 					 */
   1453 					error = dcopyout(l, &klwp, dp,
   1454 					    min(sizeof(klwp), elem_size));
   1455 					if (error) {
   1456 						rw_exit(&p->p_reflock);
   1457 						goto cleanup;
   1458 					}
   1459 					mutex_enter(p->p_lock);
   1460 					LIST_FOREACH(l3, &p->p_lwps,
   1461 					    l_sibling) {
   1462 						if (l2 == l3)
   1463 							break;
   1464 					}
   1465 					if (l3 == NULL) {
   1466 						mutex_exit(p->p_lock);
   1467 						rw_exit(&p->p_reflock);
   1468 						error = EAGAIN;
   1469 						goto cleanup;
   1470 					}
   1471 					dp += elem_size;
   1472 					buflen -= elem_size;
   1473 					elem_count--;
   1474 				}
   1475 				needed += elem_size;
   1476 			}
   1477 			mutex_exit(p->p_lock);
   1478 
   1479 			/* Drop reference to process. */
   1480 			mutex_enter(proc_lock);
   1481 			rw_exit(&p->p_reflock);
   1482 		}
   1483 		mutex_exit(proc_lock);
   1484 	} else {
   1485 		mutex_enter(proc_lock);
   1486 		p = proc_find(pid);
   1487 		if (p == NULL) {
   1488 			error = ESRCH;
   1489 			mutex_exit(proc_lock);
   1490 			goto cleanup;
   1491 		}
   1492 		/* Grab a hold on the process. */
   1493 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
   1494 		mutex_exit(proc_lock);
   1495 		if (!gotit) {
   1496 			error = ESRCH;
   1497 			goto cleanup;
   1498 		}
   1499 
   1500 		mutex_enter(p->p_lock);
   1501 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1502 			if (buflen >= elem_size && elem_count > 0) {
   1503 				lwp_lock(l2);
   1504 				fill_lwp(l2, &klwp);
   1505 				lwp_unlock(l2);
   1506 				mutex_exit(p->p_lock);
   1507 				/*
   1508 				 * Copy out elem_size, but not larger than
   1509 				 * the size of a struct kinfo_proc2.
   1510 				 */
   1511 				error = dcopyout(l, &klwp, dp,
   1512 				    min(sizeof(klwp), elem_size));
   1513 				if (error) {
   1514 					rw_exit(&p->p_reflock);
   1515 					goto cleanup;
   1516 				}
   1517 				mutex_enter(p->p_lock);
   1518 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
   1519 					if (l2 == l3)
   1520 						break;
   1521 				}
   1522 				if (l3 == NULL) {
   1523 					mutex_exit(p->p_lock);
   1524 					rw_exit(&p->p_reflock);
   1525 					error = EAGAIN;
   1526 					goto cleanup;
   1527 				}
   1528 				dp += elem_size;
   1529 				buflen -= elem_size;
   1530 				elem_count--;
   1531 			}
   1532 			needed += elem_size;
   1533 		}
   1534 		mutex_exit(p->p_lock);
   1535 
   1536 		/* Drop reference to process. */
   1537 		rw_exit(&p->p_reflock);
   1538 	}
   1539 
   1540 	if (where != NULL) {
   1541 		*oldlenp = dp - where;
   1542 		if (needed > *oldlenp) {
   1543 			sysctl_relock();
   1544 			return (ENOMEM);
   1545 		}
   1546 	} else {
   1547 		needed += KERN_LWPSLOP;
   1548 		*oldlenp = needed;
   1549 	}
   1550 	error = 0;
   1551  cleanup:
   1552 	sysctl_relock();
   1553 	return (error);
   1554 }
   1555 
   1556 /*
   1557  * sysctl helper routine for kern.forkfsleep node. Ensures that the
   1558  * given value is not too large or two small, and is at least one
   1559  * timer tick if not zero.
   1560  */
   1561 static int
   1562 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
   1563 {
   1564 	/* userland sees value in ms, internally is in ticks */
   1565 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
   1566 	int error, timo, lsleep;
   1567 	struct sysctlnode node;
   1568 
   1569 	lsleep = forkfsleep * 1000 / hz;
   1570 	node = *rnode;
   1571 	node.sysctl_data = &lsleep;
   1572 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1573 	if (error || newp == NULL)
   1574 		return (error);
   1575 
   1576 	/* refuse negative values, and overly 'long time' */
   1577 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
   1578 		return (EINVAL);
   1579 
   1580 	timo = mstohz(lsleep);
   1581 
   1582 	/* if the interval is >0 ms && <1 tick, use 1 tick */
   1583 	if (lsleep != 0 && timo == 0)
   1584 		forkfsleep = 1;
   1585 	else
   1586 		forkfsleep = timo;
   1587 
   1588 	return (0);
   1589 }
   1590 
   1591 /*
   1592  * sysctl helper routine for kern.root_partition
   1593  */
   1594 static int
   1595 sysctl_kern_root_partition(SYSCTLFN_ARGS)
   1596 {
   1597 	int rootpart = DISKPART(rootdev);
   1598 	struct sysctlnode node = *rnode;
   1599 
   1600 	node.sysctl_data = &rootpart;
   1601 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1602 }
   1603 
   1604 /*
   1605  * sysctl helper function for kern.drivers
   1606  */
   1607 static int
   1608 sysctl_kern_drivers(SYSCTLFN_ARGS)
   1609 {
   1610 	int error;
   1611 	size_t buflen;
   1612 	struct kinfo_drivers kd;
   1613 	char *start, *where;
   1614 	const char *dname;
   1615 	int i;
   1616 	extern struct devsw_conv *devsw_conv;
   1617 	extern int max_devsw_convs;
   1618 
   1619 	start = where = oldp;
   1620 	buflen = *oldlenp;
   1621 	if (where == NULL) {
   1622 		*oldlenp = max_devsw_convs * sizeof kd;
   1623 		return 0;
   1624 	}
   1625 
   1626 	/*
   1627 	 * An array of kinfo_drivers structures
   1628 	 */
   1629 	error = 0;
   1630 	sysctl_unlock();
   1631 	mutex_enter(&device_lock);
   1632 	for (i = 0; i < max_devsw_convs; i++) {
   1633 		dname = devsw_conv[i].d_name;
   1634 		if (dname == NULL)
   1635 			continue;
   1636 		if (buflen < sizeof kd) {
   1637 			error = ENOMEM;
   1638 			break;
   1639 		}
   1640 		memset(&kd, 0, sizeof(kd));
   1641 		kd.d_bmajor = devsw_conv[i].d_bmajor;
   1642 		kd.d_cmajor = devsw_conv[i].d_cmajor;
   1643 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
   1644 		mutex_exit(&device_lock);
   1645 		error = dcopyout(l, &kd, where, sizeof kd);
   1646 		mutex_enter(&device_lock);
   1647 		if (error != 0)
   1648 			break;
   1649 		buflen -= sizeof kd;
   1650 		where += sizeof kd;
   1651 	}
   1652 	mutex_exit(&device_lock);
   1653 	sysctl_relock();
   1654 	*oldlenp = where - start;
   1655 	return error;
   1656 }
   1657 
   1658 static int
   1659 sysctl_security_setidcore(SYSCTLFN_ARGS)
   1660 {
   1661 	int newsize, error;
   1662 	struct sysctlnode node;
   1663 
   1664 	node = *rnode;
   1665 	node.sysctl_data = &newsize;
   1666 	newsize = *(int *)rnode->sysctl_data;
   1667 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1668 	if (error || newp == NULL)
   1669 		return error;
   1670 
   1671 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   1672 	    0, NULL, NULL, NULL))
   1673 		return (EPERM);
   1674 
   1675 	*(int *)rnode->sysctl_data = newsize;
   1676 
   1677 	return 0;
   1678 }
   1679 
   1680 static int
   1681 sysctl_security_setidcorename(SYSCTLFN_ARGS)
   1682 {
   1683 	int error;
   1684 	char *newsetidcorename;
   1685 	struct sysctlnode node;
   1686 
   1687 	newsetidcorename = PNBUF_GET();
   1688 	node = *rnode;
   1689 	node.sysctl_data = newsetidcorename;
   1690 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   1691 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1692 	if (error || newp == NULL) {
   1693 		goto out;
   1694 	}
   1695 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   1696 	    0, NULL, NULL, NULL)) {
   1697 		error = EPERM;
   1698 		goto out;
   1699 	}
   1700 	if (strlen(newsetidcorename) == 0) {
   1701 		error = EINVAL;
   1702 		goto out;
   1703 	}
   1704 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   1705 out:
   1706 	PNBUF_PUT(newsetidcorename);
   1707 	return error;
   1708 }
   1709 
   1710 /*
   1711  * sysctl helper routine for kern.cp_id node. Maps cpus to their
   1712  * cpuids.
   1713  */
   1714 static int
   1715 sysctl_kern_cpid(SYSCTLFN_ARGS)
   1716 {
   1717 	struct sysctlnode node = *rnode;
   1718 	uint64_t *cp_id = NULL;
   1719 	int error, n = ncpu;
   1720 	struct cpu_info *ci;
   1721 	CPU_INFO_ITERATOR cii;
   1722 
   1723 	/*
   1724 	 * Here you may either retrieve a single cpu id or the whole
   1725 	 * set. The size you get back when probing depends on what
   1726 	 * you ask for.
   1727 	 */
   1728 	switch (namelen) {
   1729 	case 0:
   1730 		node.sysctl_size = n * sizeof(uint64_t);
   1731 		n = -2; /* ALL */
   1732 		break;
   1733 	case 1:
   1734 		if (name[0] < 0 || name[0] >= n)
   1735 			return (ENOENT); /* ENOSUCHPROCESSOR */
   1736 		node.sysctl_size = sizeof(uint64_t);
   1737 		n = name[0];
   1738 		/*
   1739 		 * adjust these so that sysctl_lookup() will be happy
   1740 		 */
   1741 		name++;
   1742 		namelen--;
   1743 		break;
   1744 	default:
   1745 		return (EINVAL);
   1746 	}
   1747 
   1748 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
   1749 	if (cp_id == NULL)
   1750 		return (ENOMEM);
   1751 	node.sysctl_data = cp_id;
   1752 	memset(cp_id, 0, node.sysctl_size);
   1753 
   1754 	for (CPU_INFO_FOREACH(cii, ci)) {
   1755 		if (n <= 0)
   1756 			cp_id[0] = cpu_index(ci);
   1757 		/*
   1758 		 * if a specific processor was requested and we just
   1759 		 * did it, we're done here
   1760 		 */
   1761 		if (n == 0)
   1762 			break;
   1763 		/*
   1764 		 * if doing "all", skip to next cp_id slot for next processor
   1765 		 */
   1766 		if (n == -2)
   1767 			cp_id++;
   1768 		/*
   1769 		 * if we're doing a specific processor, we're one
   1770 		 * processor closer
   1771 		 */
   1772 		if (n > 0)
   1773 			n--;
   1774 	}
   1775 
   1776 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1777 	kmem_free(node.sysctl_data, node.sysctl_size);
   1778 	return (error);
   1779 }
   1780 
   1781 /*
   1782  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
   1783  * calculate on the fly taking into account integer overflow and the
   1784  * current wired count.
   1785  */
   1786 static int
   1787 sysctl_hw_usermem(SYSCTLFN_ARGS)
   1788 {
   1789 	u_int ui;
   1790 	u_quad_t uq;
   1791 	struct sysctlnode node;
   1792 
   1793 	node = *rnode;
   1794 	switch (rnode->sysctl_num) {
   1795 	case HW_USERMEM:
   1796 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
   1797 			ui = UINT_MAX;
   1798 		else
   1799 			ui *= PAGE_SIZE;
   1800 		node.sysctl_data = &ui;
   1801 		break;
   1802 	case HW_USERMEM64:
   1803 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
   1804 		node.sysctl_data = &uq;
   1805 		break;
   1806 	default:
   1807 		return (EINVAL);
   1808 	}
   1809 
   1810 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1811 }
   1812 
   1813 /*
   1814  * sysctl helper routine for kern.cnmagic node. Pulls the old value
   1815  * out, encoded, and stuffs the new value in for decoding.
   1816  */
   1817 static int
   1818 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
   1819 {
   1820 	char magic[CNS_LEN];
   1821 	int error;
   1822 	struct sysctlnode node;
   1823 
   1824 	if (oldp)
   1825 		cn_get_magic(magic, CNS_LEN);
   1826 	node = *rnode;
   1827 	node.sysctl_data = &magic[0];
   1828 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1829 	if (error || newp == NULL)
   1830 		return (error);
   1831 
   1832 	return (cn_set_magic(magic));
   1833 }
   1834 
   1835 /*
   1836  * ********************************************************************
   1837  * section 3: public helper routines that are used for more than one
   1838  * node
   1839  * ********************************************************************
   1840  */
   1841 
   1842 /*
   1843  * sysctl helper routine for the kern.root_device node and some ports'
   1844  * machdep.root_device nodes.
   1845  */
   1846 int
   1847 sysctl_root_device(SYSCTLFN_ARGS)
   1848 {
   1849 	struct sysctlnode node;
   1850 
   1851 	node = *rnode;
   1852 	node.sysctl_data = __UNCONST(device_xname(root_device));
   1853 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
   1854 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1855 }
   1856 
   1857 /*
   1858  * sysctl helper routine for kern.consdev, dependent on the current
   1859  * state of the console. Also used for machdep.console_device on some
   1860  * ports.
   1861  */
   1862 int
   1863 sysctl_consdev(SYSCTLFN_ARGS)
   1864 {
   1865 	dev_t consdev;
   1866 	uint32_t oconsdev;
   1867 	struct sysctlnode node;
   1868 
   1869 	if (cn_tab != NULL)
   1870 		consdev = cn_tab->cn_dev;
   1871 	else
   1872 		consdev = NODEV;
   1873 	node = *rnode;
   1874 	switch (*oldlenp) {
   1875 	case sizeof(consdev):
   1876 		node.sysctl_data = &consdev;
   1877 		node.sysctl_size = sizeof(consdev);
   1878 		break;
   1879 	case sizeof(oconsdev):
   1880 		oconsdev = (uint32_t)consdev;
   1881 		node.sysctl_data = &oconsdev;
   1882 		node.sysctl_size = sizeof(oconsdev);
   1883 		break;
   1884 	default:
   1885 		return EINVAL;
   1886 	}
   1887 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1888 }
   1889 
   1890 /*
   1891  * ********************************************************************
   1892  * section 4: support for some helpers
   1893  * ********************************************************************
   1894  */
   1895 
   1896 
   1897 /*
   1898  * Fill in a kinfo_lwp structure for the specified lwp.
   1899  */
   1900 static void
   1901 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
   1902 {
   1903 	struct proc *p = l->l_proc;
   1904 	struct timeval tv;
   1905 
   1906 	KASSERT(lwp_locked(l, NULL));
   1907 
   1908 	memset(kl, 0, sizeof(*kl));
   1909 
   1910 	kl->l_forw = 0;
   1911 	kl->l_back = 0;
   1912 	kl->l_laddr = PTRTOUINT64(l);
   1913 	kl->l_addr = PTRTOUINT64(l->l_addr);
   1914 	kl->l_stat = l->l_stat;
   1915 	kl->l_lid = l->l_lid;
   1916 	kl->l_flag = L_INMEM;
   1917 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
   1918 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
   1919 
   1920 	kl->l_swtime = l->l_swtime;
   1921 	kl->l_slptime = l->l_slptime;
   1922 	if (l->l_stat == LSONPROC)
   1923 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   1924 	else
   1925 		kl->l_schedflags = 0;
   1926 	kl->l_priority = lwp_eprio(l);
   1927 	kl->l_usrpri = l->l_priority;
   1928 	if (l->l_wchan)
   1929 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
   1930 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
   1931 	kl->l_cpuid = cpu_index(l->l_cpu);
   1932 	bintime2timeval(&l->l_rtime, &tv);
   1933 	kl->l_rtime_sec = tv.tv_sec;
   1934 	kl->l_rtime_usec = tv.tv_usec;
   1935 	kl->l_cpticks = l->l_cpticks;
   1936 	kl->l_pctcpu = l->l_pctcpu;
   1937 	kl->l_pid = p->p_pid;
   1938 	if (l->l_name == NULL)
   1939 		kl->l_name[0] = '\0';
   1940 	else
   1941 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
   1942 }
   1943