Home | History | Annotate | Line # | Download | only in kern
init_sysctl.c revision 1.188
      1 /*	$NetBSD: init_sysctl.c,v 1.188 2012/03/10 21:51:59 joerg 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 1.188 2012/03/10 21:51:59 joerg 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_sbmax(SYSCTLFN_PROTO);
    160 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
    161 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
    162 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
    163 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
    164 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
    165 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
    166 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
    167 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
    168 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
    169 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
    170 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
    171 
    172 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
    173 
    174 /*
    175  * ********************************************************************
    176  * section 1: setup routines
    177  * ********************************************************************
    178  * These functions are stuffed into a link set for sysctl setup
    179  * functions. They're never called or referenced from anywhere else.
    180  * ********************************************************************
    181  */
    182 
    183 /*
    184  * this setup routine is a replacement for kern_sysctl()
    185  */
    186 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
    187 {
    188 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
    189 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
    190 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
    191 	const struct sysctlnode *rnode;
    192 
    193 	sysctl_createv(clog, 0, NULL, NULL,
    194 		       CTLFLAG_PERMANENT,
    195 		       CTLTYPE_NODE, "kern", NULL,
    196 		       NULL, 0, NULL, 0,
    197 		       CTL_KERN, CTL_EOL);
    198 
    199 	sysctl_createv(clog, 0, NULL, NULL,
    200 		       CTLFLAG_PERMANENT,
    201 		       CTLTYPE_STRING, "ostype",
    202 		       SYSCTL_DESCR("Operating system type"),
    203 		       NULL, 0, __UNCONST(&ostype), 0,
    204 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
    205 	sysctl_createv(clog, 0, NULL, NULL,
    206 		       CTLFLAG_PERMANENT,
    207 		       CTLTYPE_STRING, "osrelease",
    208 		       SYSCTL_DESCR("Operating system release"),
    209 		       NULL, 0, __UNCONST(&osrelease), 0,
    210 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
    211 	sysctl_createv(clog, 0, NULL, NULL,
    212 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    213 		       CTLTYPE_INT, "osrevision",
    214 		       SYSCTL_DESCR("Operating system revision"),
    215 		       NULL, __NetBSD_Version__, NULL, 0,
    216 		       CTL_KERN, KERN_OSREV, CTL_EOL);
    217 	sysctl_createv(clog, 0, NULL, NULL,
    218 		       CTLFLAG_PERMANENT,
    219 		       CTLTYPE_STRING, "version",
    220 		       SYSCTL_DESCR("Kernel version"),
    221 		       NULL, 0, __UNCONST(&version), 0,
    222 		       CTL_KERN, KERN_VERSION, CTL_EOL);
    223 	sysctl_createv(clog, 0, NULL, NULL,
    224 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    225 		       CTLTYPE_INT, "maxvnodes",
    226 		       SYSCTL_DESCR("Maximum number of vnodes"),
    227 		       sysctl_kern_maxvnodes, 0, NULL, 0,
    228 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
    229 	sysctl_createv(clog, 0, NULL, NULL,
    230 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    231 		       CTLTYPE_INT, "maxproc",
    232 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
    233 		       sysctl_kern_maxproc, 0, NULL, 0,
    234 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
    235 	sysctl_createv(clog, 0, NULL, NULL,
    236 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    237 		       CTLTYPE_INT, "maxfiles",
    238 		       SYSCTL_DESCR("Maximum number of open files"),
    239 		       NULL, 0, &maxfiles, 0,
    240 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
    241 	sysctl_createv(clog, 0, NULL, NULL,
    242 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    243 		       CTLTYPE_INT, "argmax",
    244 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
    245 				    "execve(2)"),
    246 		       NULL, ARG_MAX, NULL, 0,
    247 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
    248 	sysctl_createv(clog, 0, NULL, NULL,
    249 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    250 		       CTLTYPE_STRING, "hostname",
    251 		       SYSCTL_DESCR("System hostname"),
    252 		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
    253 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
    254 	sysctl_createv(clog, 0, NULL, NULL,
    255 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    256 		       CTLTYPE_INT, "hostid",
    257 		       SYSCTL_DESCR("System host ID number"),
    258 		       sysctl_kern_hostid, 0, NULL, 0,
    259 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
    260 	sysctl_createv(clog, 0, NULL, NULL,
    261 		       CTLFLAG_PERMANENT,
    262 		       CTLTYPE_STRUCT, "clockrate",
    263 		       SYSCTL_DESCR("Kernel clock rates"),
    264 		       sysctl_kern_clockrate, 0, NULL,
    265 		       sizeof(struct clockinfo),
    266 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
    267 	sysctl_createv(clog, 0, NULL, NULL,
    268 		       CTLFLAG_PERMANENT,
    269 		       CTLTYPE_INT, "hardclock_ticks",
    270 		       SYSCTL_DESCR("Number of hardclock ticks"),
    271 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
    272 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
    273 	sysctl_createv(clog, 0, NULL, NULL,
    274 		       CTLFLAG_PERMANENT,
    275 		       CTLTYPE_STRUCT, "vnode",
    276 		       SYSCTL_DESCR("System vnode table"),
    277 		       sysctl_kern_vnode, 0, NULL, 0,
    278 		       CTL_KERN, KERN_VNODE, CTL_EOL);
    279 #ifndef GPROF
    280 	sysctl_createv(clog, 0, NULL, NULL,
    281 		       CTLFLAG_PERMANENT,
    282 		       CTLTYPE_NODE, "profiling",
    283 		       SYSCTL_DESCR("Profiling information (not available)"),
    284 		       sysctl_notavail, 0, NULL, 0,
    285 		       CTL_KERN, KERN_PROF, CTL_EOL);
    286 #endif
    287 	sysctl_createv(clog, 0, NULL, NULL,
    288 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    289 		       CTLTYPE_INT, "posix1version",
    290 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
    291 				    "with which the operating system attempts "
    292 				    "to comply"),
    293 		       NULL, _POSIX_VERSION, NULL, 0,
    294 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
    295 	sysctl_createv(clog, 0, NULL, NULL,
    296 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    297 		       CTLTYPE_INT, "ngroups",
    298 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
    299 		       NULL, NGROUPS_MAX, NULL, 0,
    300 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
    301 	sysctl_createv(clog, 0, NULL, NULL,
    302 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    303 		       CTLTYPE_INT, "job_control",
    304 		       SYSCTL_DESCR("Whether job control is available"),
    305 		       NULL, 1, NULL, 0,
    306 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
    307 	sysctl_createv(clog, 0, NULL, NULL,
    308 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    309 		       CTLTYPE_INT, "saved_ids",
    310 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
    311 				    "available"), NULL,
    312 #ifdef _POSIX_SAVED_IDS
    313 		       1,
    314 #else /* _POSIX_SAVED_IDS */
    315 		       0,
    316 #endif /* _POSIX_SAVED_IDS */
    317 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
    318 	sysctl_createv(clog, 0, NULL, NULL,
    319 		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
    320 		       CTLTYPE_INT, "boothowto",
    321 		       SYSCTL_DESCR("Flags from boot loader"),
    322 		       NULL, 0, &boothowto, sizeof(boothowto),
    323 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    324 	sysctl_createv(clog, 0, NULL, NULL,
    325 		       CTLFLAG_PERMANENT,
    326 		       CTLTYPE_STRUCT, "boottime",
    327 		       SYSCTL_DESCR("System boot time"),
    328 		       NULL, 0, &boottime, sizeof(boottime),
    329 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
    330 #ifdef COMPAT_50
    331 	{
    332 		extern struct timeval50 boottime50;
    333 		sysctl_createv(clog, 0, NULL, NULL,
    334 			       CTLFLAG_PERMANENT,
    335 			       CTLTYPE_STRUCT, "oboottime",
    336 			       SYSCTL_DESCR("System boot time"),
    337 			       NULL, 0, &boottime50, sizeof(boottime50),
    338 			       CTL_KERN, KERN_OBOOTTIME, CTL_EOL);
    339 	}
    340 #endif
    341 	sysctl_createv(clog, 0, NULL, NULL,
    342 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    343 		       CTLTYPE_STRING, "domainname",
    344 		       SYSCTL_DESCR("YP domain name"),
    345 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
    346 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
    347 	sysctl_createv(clog, 0, NULL, NULL,
    348 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    349 		       CTLTYPE_INT, "maxpartitions",
    350 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
    351 				    "disk"),
    352 		       NULL, MAXPARTITIONS, NULL, 0,
    353 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
    354 	sysctl_createv(clog, 0, NULL, NULL,
    355 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    356 		       CTLTYPE_INT, "rawpartition",
    357 		       SYSCTL_DESCR("Raw partition of a disk"),
    358 		       NULL, RAW_PART, NULL, 0,
    359 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
    360 	sysctl_createv(clog, 0, NULL, NULL,
    361 		       CTLFLAG_PERMANENT,
    362 		       CTLTYPE_STRUCT, "timex", NULL,
    363 		       sysctl_notavail, 0, NULL, 0,
    364 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
    365 	sysctl_createv(clog, 0, NULL, NULL,
    366 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    367 		       CTLTYPE_INT, "rtc_offset",
    368 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
    369 				    "minutes"),
    370 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
    371 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
    372 	sysctl_createv(clog, 0, NULL, NULL,
    373 		       CTLFLAG_PERMANENT,
    374 		       CTLTYPE_STRING, "root_device",
    375 		       SYSCTL_DESCR("Name of the root device"),
    376 		       sysctl_root_device, 0, NULL, 0,
    377 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
    378 	sysctl_createv(clog, 0, NULL, NULL,
    379 		       CTLFLAG_PERMANENT,
    380 		       CTLTYPE_INT, "msgbufsize",
    381 		       SYSCTL_DESCR("Size of the kernel message buffer"),
    382 		       sysctl_msgbuf, 0, NULL, 0,
    383 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
    384 	sysctl_createv(clog, 0, NULL, NULL,
    385 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    386 		       CTLTYPE_INT, "fsync",
    387 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
    388 				    "Synchronization Option is available on "
    389 				    "this system"),
    390 		       NULL, 1, NULL, 0,
    391 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
    392 	sysctl_createv(clog, 0, NULL, NULL,
    393 		       CTLFLAG_PERMANENT,
    394 		       CTLTYPE_NODE, "ipc",
    395 		       SYSCTL_DESCR("SysV IPC options"),
    396 		       NULL, 0, NULL, 0,
    397 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
    398 	sysctl_createv(clog, 0, NULL, NULL,
    399 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    400 		       CTLTYPE_INT, "sysvmsg",
    401 		       SYSCTL_DESCR("System V style message support available"),
    402 		       NULL,
    403 #ifdef SYSVMSG
    404 		       1,
    405 #else /* SYSVMSG */
    406 		       0,
    407 #endif /* SYSVMSG */
    408 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
    409 	sysctl_createv(clog, 0, NULL, NULL,
    410 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    411 		       CTLTYPE_INT, "sysvsem",
    412 		       SYSCTL_DESCR("System V style semaphore support "
    413 				    "available"), NULL,
    414 #ifdef SYSVSEM
    415 		       1,
    416 #else /* SYSVSEM */
    417 		       0,
    418 #endif /* SYSVSEM */
    419 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
    420 	sysctl_createv(clog, 0, NULL, NULL,
    421 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    422 		       CTLTYPE_INT, "sysvshm",
    423 		       SYSCTL_DESCR("System V style shared memory support "
    424 				    "available"), NULL,
    425 #ifdef SYSVSHM
    426 		       1,
    427 #else /* SYSVSHM */
    428 		       0,
    429 #endif /* SYSVSHM */
    430 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
    431 	sysctl_createv(clog, 0, NULL, NULL,
    432 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    433 		       CTLTYPE_INT, "synchronized_io",
    434 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
    435 				    "I/O Option is available on this system"),
    436 		       NULL, 1, NULL, 0,
    437 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
    438 	sysctl_createv(clog, 0, NULL, NULL,
    439 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    440 		       CTLTYPE_INT, "iov_max",
    441 		       SYSCTL_DESCR("Maximum number of iovec structures per "
    442 				    "process"),
    443 		       NULL, IOV_MAX, NULL, 0,
    444 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
    445 	sysctl_createv(clog, 0, NULL, NULL,
    446 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    447 		       CTLTYPE_INT, "mapped_files",
    448 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
    449 				    "Files Option is available on this system"),
    450 		       NULL, 1, NULL, 0,
    451 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
    452 	sysctl_createv(clog, 0, NULL, NULL,
    453 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    454 		       CTLTYPE_INT, "memlock",
    455 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
    456 				    "Locking Option is available on this "
    457 				    "system"),
    458 		       NULL, 1, NULL, 0,
    459 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
    460 	sysctl_createv(clog, 0, NULL, NULL,
    461 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    462 		       CTLTYPE_INT, "memlock_range",
    463 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
    464 				    "Locking Option is available on this "
    465 				    "system"),
    466 		       NULL, 1, NULL, 0,
    467 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
    468 	sysctl_createv(clog, 0, NULL, NULL,
    469 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    470 		       CTLTYPE_INT, "memory_protection",
    471 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
    472 				    "Protection Option is available on this "
    473 				    "system"),
    474 		       NULL, 1, NULL, 0,
    475 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
    476 	sysctl_createv(clog, 0, NULL, NULL,
    477 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    478 		       CTLTYPE_INT, "login_name_max",
    479 		       SYSCTL_DESCR("Maximum login name length"),
    480 		       NULL, LOGIN_NAME_MAX, NULL, 0,
    481 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
    482 	sysctl_createv(clog, 0, NULL, NULL,
    483 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    484 		       CTLTYPE_STRING, "defcorename",
    485 		       SYSCTL_DESCR("Default core file name"),
    486 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
    487 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
    488 	sysctl_createv(clog, 0, NULL, NULL,
    489 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    490 		       CTLTYPE_INT, "logsigexit",
    491 		       SYSCTL_DESCR("Log process exit when caused by signals"),
    492 		       NULL, 0, &kern_logsigexit, 0,
    493 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
    494 	sysctl_createv(clog, 0, NULL, NULL,
    495 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    496 		       CTLTYPE_INT, "fscale",
    497 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
    498 		       NULL, FSCALE, NULL, 0,
    499 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
    500 	sysctl_createv(clog, 0, NULL, NULL,
    501 		       CTLFLAG_PERMANENT,
    502 		       CTLTYPE_INT, "ccpu",
    503 		       SYSCTL_DESCR("Scheduler exponential decay value"),
    504 		       NULL, 0, &ccpu, 0,
    505 		       CTL_KERN, KERN_CCPU, CTL_EOL);
    506 	sysctl_createv(clog, 0, NULL, NULL,
    507 		       CTLFLAG_PERMANENT,
    508 		       CTLTYPE_STRUCT, "cp_time",
    509 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
    510 		       sysctl_kern_cptime, 0, NULL, 0,
    511 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
    512 	sysctl_createv(clog, 0, NULL, NULL,
    513 		       CTLFLAG_PERMANENT,
    514 		       CTLTYPE_INT, "msgbuf",
    515 		       SYSCTL_DESCR("Kernel message buffer"),
    516 		       sysctl_msgbuf, 0, NULL, 0,
    517 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
    518 	sysctl_createv(clog, 0, NULL, NULL,
    519 		       CTLFLAG_PERMANENT,
    520 		       CTLTYPE_STRUCT, "consdev",
    521 		       SYSCTL_DESCR("Console device"),
    522 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
    523 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
    524 #if NPTY > 0
    525 	sysctl_createv(clog, 0, NULL, NULL,
    526 		       CTLFLAG_PERMANENT,
    527 		       CTLTYPE_INT, "maxptys",
    528 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
    529 		       sysctl_kern_maxptys, 0, NULL, 0,
    530 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
    531 #endif /* NPTY > 0 */
    532 	sysctl_createv(clog, 0, NULL, NULL,
    533 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    534 		       CTLTYPE_INT, "maxphys",
    535 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
    536 		       NULL, MAXPHYS, NULL, 0,
    537 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
    538 	sysctl_createv(clog, 0, NULL, NULL,
    539 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    540 		       CTLTYPE_INT, "sbmax",
    541 		       SYSCTL_DESCR("Maximum socket buffer size"),
    542 		       sysctl_kern_sbmax, 0, NULL, 0,
    543 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
    544 	sysctl_createv(clog, 0, NULL, NULL,
    545 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    546 		       CTLTYPE_INT, "monotonic_clock",
    547 		       SYSCTL_DESCR("Implementation version of the POSIX "
    548 				    "1003.1b Monotonic Clock Option"),
    549 		       /* XXX _POSIX_VERSION */
    550 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
    551 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
    552 	sysctl_createv(clog, 0, NULL, NULL,
    553 		       CTLFLAG_PERMANENT,
    554 		       CTLTYPE_INT, "urandom",
    555 		       SYSCTL_DESCR("Random integer value"),
    556 		       sysctl_kern_urnd, 0, NULL, 0,
    557 		       CTL_KERN, KERN_URND, CTL_EOL);
    558 	sysctl_createv(clog, 0, NULL, NULL,
    559 		       CTLFLAG_PERMANENT,
    560 		       CTLTYPE_INT, "arandom",
    561 		       SYSCTL_DESCR("n bytes of random data"),
    562 		       sysctl_kern_arnd, 0, NULL, 0,
    563 		       CTL_KERN, KERN_ARND, CTL_EOL);
    564 	sysctl_createv(clog, 0, NULL, NULL,
    565 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    566 		       CTLTYPE_INT, "labelsector",
    567 		       SYSCTL_DESCR("Sector number containing the disklabel"),
    568 		       NULL, LABELSECTOR, NULL, 0,
    569 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
    570 	sysctl_createv(clog, 0, NULL, NULL,
    571 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    572 		       CTLTYPE_INT, "labeloffset",
    573 		       SYSCTL_DESCR("Offset of the disklabel within the "
    574 				    "sector"),
    575 		       NULL, LABELOFFSET, NULL, 0,
    576 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
    577 	sysctl_createv(clog, 0, NULL, NULL,
    578 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    579 		       CTLTYPE_INT, "labelusesmbr",
    580 		       SYSCTL_DESCR("disklabel is inside MBR partition"),
    581 		       NULL, LABELUSESMBR, NULL, 0,
    582 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    583 	sysctl_createv(clog, 0, NULL, NULL,
    584 		       CTLFLAG_PERMANENT,
    585 		       CTLTYPE_NODE, "lwp",
    586 		       SYSCTL_DESCR("System-wide LWP information"),
    587 		       sysctl_kern_lwp, 0, NULL, 0,
    588 		       CTL_KERN, KERN_LWP, CTL_EOL);
    589 	sysctl_createv(clog, 0, NULL, NULL,
    590 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    591 		       CTLTYPE_INT, "forkfsleep",
    592 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
    593 				    "to process limits"),
    594 		       sysctl_kern_forkfsleep, 0, NULL, 0,
    595 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
    596 	sysctl_createv(clog, 0, NULL, NULL,
    597 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    598 		       CTLTYPE_INT, "posix_threads",
    599 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    600 				    "Threads option to which the system "
    601 				    "attempts to conform"),
    602 		       /* XXX _POSIX_VERSION */
    603 		       NULL, _POSIX_THREADS, NULL, 0,
    604 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
    605 	sysctl_createv(clog, 0, NULL, NULL,
    606 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    607 		       CTLTYPE_INT, "posix_semaphores",
    608 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    609 				    "Semaphores option to which the system "
    610 				    "attempts to conform"), NULL,
    611 		       200112, NULL, 0,
    612 		       CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
    613 	sysctl_createv(clog, 0, NULL, NULL,
    614 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    615 		       CTLTYPE_INT, "posix_barriers",
    616 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    617 				    "Barriers option to which the system "
    618 				    "attempts to conform"),
    619 		       /* XXX _POSIX_VERSION */
    620 		       NULL, _POSIX_BARRIERS, NULL, 0,
    621 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
    622 	sysctl_createv(clog, 0, NULL, NULL,
    623 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    624 		       CTLTYPE_INT, "posix_timers",
    625 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    626 				    "Timers option to which the system "
    627 				    "attempts to conform"),
    628 		       /* XXX _POSIX_VERSION */
    629 		       NULL, _POSIX_TIMERS, NULL, 0,
    630 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
    631 	sysctl_createv(clog, 0, NULL, NULL,
    632 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    633 		       CTLTYPE_INT, "posix_spin_locks",
    634 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
    635 				    "Locks option to which the system attempts "
    636 				    "to conform"),
    637 		       /* XXX _POSIX_VERSION */
    638 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
    639 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
    640 	sysctl_createv(clog, 0, NULL, NULL,
    641 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    642 		       CTLTYPE_INT, "posix_reader_writer_locks",
    643 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    644 				    "Read-Write Locks option to which the "
    645 				    "system attempts to conform"),
    646 		       /* XXX _POSIX_VERSION */
    647 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
    648 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
    649 	sysctl_createv(clog, 0, NULL, NULL,
    650 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    651 		       CTLTYPE_INT, "dump_on_panic",
    652 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
    653 		       NULL, 0, &dumponpanic, 0,
    654 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
    655 #ifdef DIAGNOSTIC
    656 	sysctl_createv(clog, 0, NULL, NULL,
    657 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    658 		       CTLTYPE_INT, "panic_now",
    659 		       SYSCTL_DESCR("Trigger a panic"),
    660 		       sysctl_kern_trigger_panic, 0, NULL, 0,
    661 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    662 #endif
    663 	sysctl_createv(clog, 0, NULL, NULL,
    664 		       CTLFLAG_PERMANENT,
    665 		       CTLTYPE_INT, "root_partition",
    666 		       SYSCTL_DESCR("Root partition on the root device"),
    667 		       sysctl_kern_root_partition, 0, NULL, 0,
    668 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
    669 	sysctl_createv(clog, 0, NULL, NULL,
    670 		       CTLFLAG_PERMANENT,
    671 		       CTLTYPE_STRUCT, "drivers",
    672 		       SYSCTL_DESCR("List of all drivers with block and "
    673 				    "character device numbers"),
    674 		       sysctl_kern_drivers, 0, NULL, 0,
    675 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
    676 	sysctl_createv(clog, 0, NULL, NULL,
    677 		       CTLFLAG_PERMANENT,
    678 		       CTLTYPE_STRUCT, "cp_id",
    679 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
    680 		       sysctl_kern_cpid, 0, NULL, 0,
    681 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
    682 	sysctl_createv(clog, 0, NULL, &rnode,
    683 		       CTLFLAG_PERMANENT,
    684 		       CTLTYPE_NODE, "coredump",
    685 		       SYSCTL_DESCR("Coredump settings."),
    686 		       NULL, 0, NULL, 0,
    687 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    688 	sysctl_createv(clog, 0, &rnode, &rnode,
    689 		       CTLFLAG_PERMANENT,
    690 		       CTLTYPE_NODE, "setid",
    691 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
    692 		       NULL, 0, NULL, 0,
    693 		       CTL_CREATE, CTL_EOL);
    694 	sysctl_createv(clog, 0, &rnode, NULL,
    695 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    696 		       CTLTYPE_INT, "dump",
    697 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
    698 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
    699 		       sizeof(security_setidcore_dump),
    700 		       CTL_CREATE, CTL_EOL);
    701 	sysctl_createv(clog, 0, &rnode, NULL,
    702 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    703 		       CTLTYPE_STRING, "path",
    704 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
    705 		       sysctl_security_setidcorename, 0,
    706 		       &security_setidcore_path,
    707 		       sizeof(security_setidcore_path),
    708 		       CTL_CREATE, CTL_EOL);
    709 	sysctl_createv(clog, 0, &rnode, NULL,
    710 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    711 		       CTLTYPE_INT, "owner",
    712 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
    713 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
    714 		       0,
    715 		       CTL_CREATE, CTL_EOL);
    716 	sysctl_createv(clog, 0, &rnode, NULL,
    717 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    718 		       CTLTYPE_INT, "group",
    719 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
    720 		       sysctl_security_setidcore, 0, &security_setidcore_group,
    721 		       0,
    722 		       CTL_CREATE, CTL_EOL);
    723 	sysctl_createv(clog, 0, &rnode, NULL,
    724 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    725 		       CTLTYPE_INT, "mode",
    726 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
    727 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
    728 		       0,
    729 		       CTL_CREATE, CTL_EOL);
    730 	sysctl_createv(clog, 0, NULL, NULL,
    731 		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    732 		       CTLTYPE_INT, "no_sa_support",
    733 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
    734 		       "it doesn't"),
    735 		       NULL, 1, NULL, 0,
    736 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    737 	/* kern.posix. */
    738 	sysctl_createv(clog, 0, NULL, &rnode,
    739 			CTLFLAG_PERMANENT,
    740 			CTLTYPE_NODE, "posix",
    741 			SYSCTL_DESCR("POSIX options"),
    742 			NULL, 0, NULL, 0,
    743 			CTL_KERN, CTL_CREATE, CTL_EOL);
    744 	sysctl_createv(clog, 0, &rnode, NULL,
    745 			CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
    746 			CTLTYPE_INT, "semmax",
    747 			SYSCTL_DESCR("Maximal number of semaphores"),
    748 			NULL, 0, &ksem_max, 0,
    749 			CTL_CREATE, CTL_EOL);
    750 }
    751 
    752 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
    753 {
    754 	u_int u;
    755 	u_quad_t q;
    756 
    757 	sysctl_createv(clog, 0, NULL, NULL,
    758 		       CTLFLAG_PERMANENT,
    759 		       CTLTYPE_NODE, "hw", NULL,
    760 		       NULL, 0, NULL, 0,
    761 		       CTL_HW, CTL_EOL);
    762 
    763 	sysctl_createv(clog, 0, NULL, NULL,
    764 		       CTLFLAG_PERMANENT,
    765 		       CTLTYPE_STRING, "machine",
    766 		       SYSCTL_DESCR("Machine class"),
    767 		       NULL, 0, machine, 0,
    768 		       CTL_HW, HW_MACHINE, CTL_EOL);
    769 	sysctl_createv(clog, 0, NULL, NULL,
    770 		       CTLFLAG_PERMANENT,
    771 		       CTLTYPE_STRING, "model",
    772 		       SYSCTL_DESCR("Machine model"),
    773 		       NULL, 0, cpu_model, 0,
    774 		       CTL_HW, HW_MODEL, CTL_EOL);
    775 	sysctl_createv(clog, 0, NULL, NULL,
    776 		       CTLFLAG_PERMANENT,
    777 		       CTLTYPE_INT, "ncpu",
    778 		       SYSCTL_DESCR("Number of CPUs configured"),
    779 		       NULL, 0, &ncpu, 0,
    780 		       CTL_HW, HW_NCPU, CTL_EOL);
    781 	sysctl_createv(clog, 0, NULL, NULL,
    782 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    783 		       CTLTYPE_INT, "byteorder",
    784 		       SYSCTL_DESCR("System byte order"),
    785 		       NULL, BYTE_ORDER, NULL, 0,
    786 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
    787 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
    788 		UINT_MAX : physmem * PAGE_SIZE;
    789 	sysctl_createv(clog, 0, NULL, NULL,
    790 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    791 		       CTLTYPE_INT, "physmem",
    792 		       SYSCTL_DESCR("Bytes of physical memory"),
    793 		       NULL, u, NULL, 0,
    794 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
    795 	sysctl_createv(clog, 0, NULL, NULL,
    796 		       CTLFLAG_PERMANENT,
    797 		       CTLTYPE_INT, "usermem",
    798 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    799 		       sysctl_hw_usermem, 0, NULL, 0,
    800 		       CTL_HW, HW_USERMEM, CTL_EOL);
    801 	sysctl_createv(clog, 0, NULL, NULL,
    802 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    803 		       CTLTYPE_INT, "pagesize",
    804 		       SYSCTL_DESCR("Software page size"),
    805 		       NULL, PAGE_SIZE, NULL, 0,
    806 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
    807 	sysctl_createv(clog, 0, NULL, NULL,
    808 		       CTLFLAG_PERMANENT,
    809 		       CTLTYPE_STRING, "machine_arch",
    810 		       SYSCTL_DESCR("Machine CPU class"),
    811 		       NULL, 0, machine_arch, 0,
    812 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
    813 	sysctl_createv(clog, 0, NULL, NULL,
    814 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    815 		       CTLTYPE_INT, "alignbytes",
    816 		       SYSCTL_DESCR("Alignment constraint for all possible "
    817 				    "data types"),
    818 		       NULL, ALIGNBYTES, NULL, 0,
    819 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
    820 	sysctl_createv(clog, 0, NULL, NULL,
    821 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    822 		       CTLTYPE_STRING, "cnmagic",
    823 		       SYSCTL_DESCR("Console magic key sequence"),
    824 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
    825 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
    826 	q = (u_quad_t)physmem * PAGE_SIZE;
    827 	sysctl_createv(clog, 0, NULL, NULL,
    828 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    829 		       CTLTYPE_QUAD, "physmem64",
    830 		       SYSCTL_DESCR("Bytes of physical memory"),
    831 		       NULL, q, NULL, 0,
    832 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
    833 	sysctl_createv(clog, 0, NULL, NULL,
    834 		       CTLFLAG_PERMANENT,
    835 		       CTLTYPE_QUAD, "usermem64",
    836 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    837 		       sysctl_hw_usermem, 0, NULL, 0,
    838 		       CTL_HW, HW_USERMEM64, CTL_EOL);
    839 	sysctl_createv(clog, 0, NULL, NULL,
    840 		       CTLFLAG_PERMANENT,
    841 		       CTLTYPE_INT, "ncpuonline",
    842 		       SYSCTL_DESCR("Number of CPUs online"),
    843 		       NULL, 0, &ncpuonline, 0,
    844 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
    845 }
    846 
    847 #ifdef DEBUG
    848 /*
    849  * Debugging related system variables.
    850  */
    851 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
    852 struct ctldebug debug5, debug6, debug7, debug8, debug9;
    853 struct ctldebug debug10, debug11, debug12, debug13, debug14;
    854 struct ctldebug debug15, debug16, debug17, debug18, debug19;
    855 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
    856 	&debug0, &debug1, &debug2, &debug3, &debug4,
    857 	&debug5, &debug6, &debug7, &debug8, &debug9,
    858 	&debug10, &debug11, &debug12, &debug13, &debug14,
    859 	&debug15, &debug16, &debug17, &debug18, &debug19,
    860 };
    861 
    862 /*
    863  * this setup routine is a replacement for debug_sysctl()
    864  *
    865  * note that it creates several nodes per defined debug variable
    866  */
    867 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
    868 {
    869 	struct ctldebug *cdp;
    870 	char nodename[20];
    871 	int i;
    872 
    873 	/*
    874 	 * two ways here:
    875 	 *
    876 	 * the "old" way (debug.name -> value) which was emulated by
    877 	 * the sysctl(8) binary
    878 	 *
    879 	 * the new way, which the sysctl(8) binary was actually using
    880 
    881 	 node	debug
    882 	 node	debug.0
    883 	 string debug.0.name
    884 	 int	debug.0.value
    885 	 int	debug.name
    886 
    887 	 */
    888 
    889 	sysctl_createv(clog, 0, NULL, NULL,
    890 		       CTLFLAG_PERMANENT,
    891 		       CTLTYPE_NODE, "debug", NULL,
    892 		       NULL, 0, NULL, 0,
    893 		       CTL_DEBUG, CTL_EOL);
    894 
    895 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
    896 		cdp = debugvars[i];
    897 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
    898 			continue;
    899 
    900 		snprintf(nodename, sizeof(nodename), "debug%d", i);
    901 		sysctl_createv(clog, 0, NULL, NULL,
    902 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    903 			       CTLTYPE_NODE, nodename, NULL,
    904 			       NULL, 0, NULL, 0,
    905 			       CTL_DEBUG, i, CTL_EOL);
    906 		sysctl_createv(clog, 0, NULL, NULL,
    907 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    908 			       CTLTYPE_STRING, "name", NULL,
    909 			       /*XXXUNCONST*/
    910 			       NULL, 0, __UNCONST(cdp->debugname), 0,
    911 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
    912 		sysctl_createv(clog, 0, NULL, NULL,
    913 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    914 			       CTLTYPE_INT, "value", NULL,
    915 			       NULL, 0, cdp->debugvar, 0,
    916 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
    917 		sysctl_createv(clog, 0, NULL, NULL,
    918 			       CTLFLAG_PERMANENT,
    919 			       CTLTYPE_INT, cdp->debugname, NULL,
    920 			       NULL, 0, cdp->debugvar, 0,
    921 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
    922 	}
    923 }
    924 #endif /* DEBUG */
    925 
    926 /*
    927  * ********************************************************************
    928  * section 2: private node-specific helper routines.
    929  * ********************************************************************
    930  */
    931 
    932 #ifdef DIAGNOSTIC
    933 static int
    934 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
    935 {
    936 	int newtrig, error;
    937 	struct sysctlnode node;
    938 
    939 	newtrig = 0;
    940 	node = *rnode;
    941 	node.sysctl_data = &newtrig;
    942 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
    943 	if (error || newp == NULL)
    944 		return (error);
    945 
    946 	if (newtrig != 0)
    947 		panic("Panic triggered");
    948 
    949 	return (error);
    950 }
    951 #endif
    952 
    953 /*
    954  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
    955  * new value is lower than desiredvnodes and then calls reinit
    956  * routines that needs to adjust to the new value.
    957  */
    958 static int
    959 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
    960 {
    961 	int error, new_vnodes, old_vnodes, new_max;
    962 	struct sysctlnode node;
    963 
    964 	new_vnodes = desiredvnodes;
    965 	node = *rnode;
    966 	node.sysctl_data = &new_vnodes;
    967 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
    968 	if (error || newp == NULL)
    969 		return (error);
    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.sbmax. Basically just ensures that
   1347  * any new value is not too small.
   1348  */
   1349 static int
   1350 sysctl_kern_sbmax(SYSCTLFN_ARGS)
   1351 {
   1352 	int error, new_sbmax;
   1353 	struct sysctlnode node;
   1354 
   1355 	new_sbmax = sb_max;
   1356 	node = *rnode;
   1357 	node.sysctl_data = &new_sbmax;
   1358 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1359 	if (error || newp == NULL)
   1360 		return (error);
   1361 
   1362 	KERNEL_LOCK(1, NULL);
   1363 	error = sb_max_set(new_sbmax);
   1364 	KERNEL_UNLOCK_ONE(NULL);
   1365 
   1366 	return (error);
   1367 }
   1368 
   1369 /*
   1370  * sysctl helper routine for kern.urandom node. Picks a random number
   1371  * for you.
   1372  */
   1373 static int
   1374 sysctl_kern_urnd(SYSCTLFN_ARGS)
   1375 {
   1376 	int v, rv;
   1377 
   1378 	rv = cprng_strong(sysctl_prng, &v, sizeof(v), 0);
   1379 	if (rv == sizeof(v)) {
   1380 		struct sysctlnode node = *rnode;
   1381 		node.sysctl_data = &v;
   1382 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1383 	}
   1384 	else
   1385 		return (EIO);	/*XXX*/
   1386 }
   1387 
   1388 /*
   1389  * sysctl helper routine for kern.arandom node. Picks a random number
   1390  * for you.
   1391  */
   1392 static int
   1393 sysctl_kern_arnd(SYSCTLFN_ARGS)
   1394 {
   1395 	int error;
   1396 	void *v;
   1397 	struct sysctlnode node = *rnode;
   1398 
   1399 	if (*oldlenp == 0)
   1400 		return 0;
   1401 	/*
   1402 	 * This code used to allow sucking 8192 bytes at a time out
   1403 	 * of the kernel arc4random generator.  Evidently there is some
   1404 	 * very old OpenBSD application code that may try to do this.
   1405 	 *
   1406 	 * Note that this node is documented as type "INT" -- 4 or 8
   1407 	 * bytes, not 8192.
   1408 	 *
   1409 	 * We continue to support this abuse of the "len" pointer here
   1410 	 * but only 256 bytes at a time, as, anecdotally, the actual
   1411 	 * application use here was to generate RC4 keys in userspace.
   1412 	 *
   1413 	 * Support for such large requests will probably be removed
   1414 	 * entirely in the future.
   1415 	 */
   1416 	if (*oldlenp > 256)
   1417 		return E2BIG;
   1418 
   1419 	v = kmem_alloc(*oldlenp, KM_SLEEP);
   1420 	cprng_fast(v, *oldlenp);
   1421 	node.sysctl_data = v;
   1422 	node.sysctl_size = *oldlenp;
   1423 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1424 	kmem_free(v, *oldlenp);
   1425 	return error;
   1426 }
   1427 /*
   1428  * sysctl helper routine to do kern.lwp.* work.
   1429  */
   1430 static int
   1431 sysctl_kern_lwp(SYSCTLFN_ARGS)
   1432 {
   1433 	struct kinfo_lwp klwp;
   1434 	struct proc *p;
   1435 	struct lwp *l2, *l3;
   1436 	char *where, *dp;
   1437 	int pid, elem_size, elem_count;
   1438 	int buflen, needed, error;
   1439 	bool gotit;
   1440 
   1441 	if (namelen == 1 && name[0] == CTL_QUERY)
   1442 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   1443 
   1444 	dp = where = oldp;
   1445 	buflen = where != NULL ? *oldlenp : 0;
   1446 	error = needed = 0;
   1447 
   1448 	if (newp != NULL || namelen != 3)
   1449 		return (EINVAL);
   1450 	pid = name[0];
   1451 	elem_size = name[1];
   1452 	elem_count = name[2];
   1453 
   1454 	sysctl_unlock();
   1455 	if (pid == -1) {
   1456 		mutex_enter(proc_lock);
   1457 		PROCLIST_FOREACH(p, &allproc) {
   1458 			/* Grab a hold on the process. */
   1459 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   1460 				continue;
   1461 			}
   1462 			mutex_exit(proc_lock);
   1463 
   1464 			mutex_enter(p->p_lock);
   1465 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1466 				if (buflen >= elem_size && elem_count > 0) {
   1467 					lwp_lock(l2);
   1468 					fill_lwp(l2, &klwp);
   1469 					lwp_unlock(l2);
   1470 					mutex_exit(p->p_lock);
   1471 
   1472 					/*
   1473 					 * Copy out elem_size, but not
   1474 					 * larger than the size of a
   1475 					 * struct kinfo_proc2.
   1476 					 */
   1477 					error = dcopyout(l, &klwp, dp,
   1478 					    min(sizeof(klwp), elem_size));
   1479 					if (error) {
   1480 						rw_exit(&p->p_reflock);
   1481 						goto cleanup;
   1482 					}
   1483 					mutex_enter(p->p_lock);
   1484 					LIST_FOREACH(l3, &p->p_lwps,
   1485 					    l_sibling) {
   1486 						if (l2 == l3)
   1487 							break;
   1488 					}
   1489 					if (l3 == NULL) {
   1490 						mutex_exit(p->p_lock);
   1491 						rw_exit(&p->p_reflock);
   1492 						error = EAGAIN;
   1493 						goto cleanup;
   1494 					}
   1495 					dp += elem_size;
   1496 					buflen -= elem_size;
   1497 					elem_count--;
   1498 				}
   1499 				needed += elem_size;
   1500 			}
   1501 			mutex_exit(p->p_lock);
   1502 
   1503 			/* Drop reference to process. */
   1504 			mutex_enter(proc_lock);
   1505 			rw_exit(&p->p_reflock);
   1506 		}
   1507 		mutex_exit(proc_lock);
   1508 	} else {
   1509 		mutex_enter(proc_lock);
   1510 		p = proc_find(pid);
   1511 		if (p == NULL) {
   1512 			error = ESRCH;
   1513 			mutex_exit(proc_lock);
   1514 			goto cleanup;
   1515 		}
   1516 		/* Grab a hold on the process. */
   1517 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
   1518 		mutex_exit(proc_lock);
   1519 		if (!gotit) {
   1520 			error = ESRCH;
   1521 			goto cleanup;
   1522 		}
   1523 
   1524 		mutex_enter(p->p_lock);
   1525 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1526 			if (buflen >= elem_size && elem_count > 0) {
   1527 				lwp_lock(l2);
   1528 				fill_lwp(l2, &klwp);
   1529 				lwp_unlock(l2);
   1530 				mutex_exit(p->p_lock);
   1531 				/*
   1532 				 * Copy out elem_size, but not larger than
   1533 				 * the size of a struct kinfo_proc2.
   1534 				 */
   1535 				error = dcopyout(l, &klwp, dp,
   1536 				    min(sizeof(klwp), elem_size));
   1537 				if (error) {
   1538 					rw_exit(&p->p_reflock);
   1539 					goto cleanup;
   1540 				}
   1541 				mutex_enter(p->p_lock);
   1542 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
   1543 					if (l2 == l3)
   1544 						break;
   1545 				}
   1546 				if (l3 == NULL) {
   1547 					mutex_exit(p->p_lock);
   1548 					rw_exit(&p->p_reflock);
   1549 					error = EAGAIN;
   1550 					goto cleanup;
   1551 				}
   1552 				dp += elem_size;
   1553 				buflen -= elem_size;
   1554 				elem_count--;
   1555 			}
   1556 			needed += elem_size;
   1557 		}
   1558 		mutex_exit(p->p_lock);
   1559 
   1560 		/* Drop reference to process. */
   1561 		rw_exit(&p->p_reflock);
   1562 	}
   1563 
   1564 	if (where != NULL) {
   1565 		*oldlenp = dp - where;
   1566 		if (needed > *oldlenp) {
   1567 			sysctl_relock();
   1568 			return (ENOMEM);
   1569 		}
   1570 	} else {
   1571 		needed += KERN_LWPSLOP;
   1572 		*oldlenp = needed;
   1573 	}
   1574 	error = 0;
   1575  cleanup:
   1576 	sysctl_relock();
   1577 	return (error);
   1578 }
   1579 
   1580 /*
   1581  * sysctl helper routine for kern.forkfsleep node. Ensures that the
   1582  * given value is not too large or two small, and is at least one
   1583  * timer tick if not zero.
   1584  */
   1585 static int
   1586 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
   1587 {
   1588 	/* userland sees value in ms, internally is in ticks */
   1589 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
   1590 	int error, timo, lsleep;
   1591 	struct sysctlnode node;
   1592 
   1593 	lsleep = forkfsleep * 1000 / hz;
   1594 	node = *rnode;
   1595 	node.sysctl_data = &lsleep;
   1596 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1597 	if (error || newp == NULL)
   1598 		return (error);
   1599 
   1600 	/* refuse negative values, and overly 'long time' */
   1601 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
   1602 		return (EINVAL);
   1603 
   1604 	timo = mstohz(lsleep);
   1605 
   1606 	/* if the interval is >0 ms && <1 tick, use 1 tick */
   1607 	if (lsleep != 0 && timo == 0)
   1608 		forkfsleep = 1;
   1609 	else
   1610 		forkfsleep = timo;
   1611 
   1612 	return (0);
   1613 }
   1614 
   1615 /*
   1616  * sysctl helper routine for kern.root_partition
   1617  */
   1618 static int
   1619 sysctl_kern_root_partition(SYSCTLFN_ARGS)
   1620 {
   1621 	int rootpart = DISKPART(rootdev);
   1622 	struct sysctlnode node = *rnode;
   1623 
   1624 	node.sysctl_data = &rootpart;
   1625 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1626 }
   1627 
   1628 /*
   1629  * sysctl helper function for kern.drivers
   1630  */
   1631 static int
   1632 sysctl_kern_drivers(SYSCTLFN_ARGS)
   1633 {
   1634 	int error;
   1635 	size_t buflen;
   1636 	struct kinfo_drivers kd;
   1637 	char *start, *where;
   1638 	const char *dname;
   1639 	int i;
   1640 	extern struct devsw_conv *devsw_conv;
   1641 	extern int max_devsw_convs;
   1642 
   1643 	if (newp != NULL || namelen != 0)
   1644 		return (EINVAL);
   1645 
   1646 	start = where = oldp;
   1647 	buflen = *oldlenp;
   1648 	if (where == NULL) {
   1649 		*oldlenp = max_devsw_convs * sizeof kd;
   1650 		return 0;
   1651 	}
   1652 
   1653 	/*
   1654 	 * An array of kinfo_drivers structures
   1655 	 */
   1656 	error = 0;
   1657 	sysctl_unlock();
   1658 	mutex_enter(&device_lock);
   1659 	for (i = 0; i < max_devsw_convs; i++) {
   1660 		dname = devsw_conv[i].d_name;
   1661 		if (dname == NULL)
   1662 			continue;
   1663 		if (buflen < sizeof kd) {
   1664 			error = ENOMEM;
   1665 			break;
   1666 		}
   1667 		memset(&kd, 0, sizeof(kd));
   1668 		kd.d_bmajor = devsw_conv[i].d_bmajor;
   1669 		kd.d_cmajor = devsw_conv[i].d_cmajor;
   1670 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
   1671 		mutex_exit(&device_lock);
   1672 		error = dcopyout(l, &kd, where, sizeof kd);
   1673 		mutex_enter(&device_lock);
   1674 		if (error != 0)
   1675 			break;
   1676 		buflen -= sizeof kd;
   1677 		where += sizeof kd;
   1678 	}
   1679 	mutex_exit(&device_lock);
   1680 	sysctl_relock();
   1681 	*oldlenp = where - start;
   1682 	return error;
   1683 }
   1684 
   1685 static int
   1686 sysctl_security_setidcore(SYSCTLFN_ARGS)
   1687 {
   1688 	int newsize, error;
   1689 	struct sysctlnode node;
   1690 
   1691 	node = *rnode;
   1692 	node.sysctl_data = &newsize;
   1693 	newsize = *(int *)rnode->sysctl_data;
   1694 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1695 	if (error || newp == NULL)
   1696 		return error;
   1697 
   1698 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   1699 	    0, NULL, NULL, NULL))
   1700 		return (EPERM);
   1701 
   1702 	*(int *)rnode->sysctl_data = newsize;
   1703 
   1704 	return 0;
   1705 }
   1706 
   1707 static int
   1708 sysctl_security_setidcorename(SYSCTLFN_ARGS)
   1709 {
   1710 	int error;
   1711 	char *newsetidcorename;
   1712 	struct sysctlnode node;
   1713 
   1714 	newsetidcorename = PNBUF_GET();
   1715 	node = *rnode;
   1716 	node.sysctl_data = newsetidcorename;
   1717 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   1718 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1719 	if (error || newp == NULL) {
   1720 		goto out;
   1721 	}
   1722 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   1723 	    0, NULL, NULL, NULL)) {
   1724 		error = EPERM;
   1725 		goto out;
   1726 	}
   1727 	if (strlen(newsetidcorename) == 0) {
   1728 		error = EINVAL;
   1729 		goto out;
   1730 	}
   1731 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   1732 out:
   1733 	PNBUF_PUT(newsetidcorename);
   1734 	return error;
   1735 }
   1736 
   1737 /*
   1738  * sysctl helper routine for kern.cp_id node. Maps cpus to their
   1739  * cpuids.
   1740  */
   1741 static int
   1742 sysctl_kern_cpid(SYSCTLFN_ARGS)
   1743 {
   1744 	struct sysctlnode node = *rnode;
   1745 	uint64_t *cp_id = NULL;
   1746 	int error, n = ncpu;
   1747 	struct cpu_info *ci;
   1748 	CPU_INFO_ITERATOR cii;
   1749 
   1750 	/*
   1751 	 * Here you may either retrieve a single cpu id or the whole
   1752 	 * set. The size you get back when probing depends on what
   1753 	 * you ask for.
   1754 	 */
   1755 	switch (namelen) {
   1756 	case 0:
   1757 		node.sysctl_size = n * sizeof(uint64_t);
   1758 		n = -2; /* ALL */
   1759 		break;
   1760 	case 1:
   1761 		if (name[0] < 0 || name[0] >= n)
   1762 			return (ENOENT); /* ENOSUCHPROCESSOR */
   1763 		node.sysctl_size = sizeof(uint64_t);
   1764 		n = name[0];
   1765 		/*
   1766 		 * adjust these so that sysctl_lookup() will be happy
   1767 		 */
   1768 		name++;
   1769 		namelen--;
   1770 		break;
   1771 	default:
   1772 		return (EINVAL);
   1773 	}
   1774 
   1775 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
   1776 	if (cp_id == NULL)
   1777 		return (ENOMEM);
   1778 	node.sysctl_data = cp_id;
   1779 	memset(cp_id, 0, node.sysctl_size);
   1780 
   1781 	for (CPU_INFO_FOREACH(cii, ci)) {
   1782 		if (n <= 0)
   1783 			cp_id[0] = cpu_index(ci);
   1784 		/*
   1785 		 * if a specific processor was requested and we just
   1786 		 * did it, we're done here
   1787 		 */
   1788 		if (n == 0)
   1789 			break;
   1790 		/*
   1791 		 * if doing "all", skip to next cp_id slot for next processor
   1792 		 */
   1793 		if (n == -2)
   1794 			cp_id++;
   1795 		/*
   1796 		 * if we're doing a specific processor, we're one
   1797 		 * processor closer
   1798 		 */
   1799 		if (n > 0)
   1800 			n--;
   1801 	}
   1802 
   1803 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1804 	kmem_free(node.sysctl_data, node.sysctl_size);
   1805 	return (error);
   1806 }
   1807 
   1808 /*
   1809  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
   1810  * calculate on the fly taking into account integer overflow and the
   1811  * current wired count.
   1812  */
   1813 static int
   1814 sysctl_hw_usermem(SYSCTLFN_ARGS)
   1815 {
   1816 	u_int ui;
   1817 	u_quad_t uq;
   1818 	struct sysctlnode node;
   1819 
   1820 	node = *rnode;
   1821 	switch (rnode->sysctl_num) {
   1822 	case HW_USERMEM:
   1823 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
   1824 			ui = UINT_MAX;
   1825 		else
   1826 			ui *= PAGE_SIZE;
   1827 		node.sysctl_data = &ui;
   1828 		break;
   1829 	case HW_USERMEM64:
   1830 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
   1831 		node.sysctl_data = &uq;
   1832 		break;
   1833 	default:
   1834 		return (EINVAL);
   1835 	}
   1836 
   1837 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1838 }
   1839 
   1840 /*
   1841  * sysctl helper routine for kern.cnmagic node. Pulls the old value
   1842  * out, encoded, and stuffs the new value in for decoding.
   1843  */
   1844 static int
   1845 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
   1846 {
   1847 	char magic[CNS_LEN];
   1848 	int error;
   1849 	struct sysctlnode node;
   1850 
   1851 	if (oldp)
   1852 		cn_get_magic(magic, CNS_LEN);
   1853 	node = *rnode;
   1854 	node.sysctl_data = &magic[0];
   1855 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1856 	if (error || newp == NULL)
   1857 		return (error);
   1858 
   1859 	return (cn_set_magic(magic));
   1860 }
   1861 
   1862 /*
   1863  * ********************************************************************
   1864  * section 3: public helper routines that are used for more than one
   1865  * node
   1866  * ********************************************************************
   1867  */
   1868 
   1869 /*
   1870  * sysctl helper routine for the kern.root_device node and some ports'
   1871  * machdep.root_device nodes.
   1872  */
   1873 int
   1874 sysctl_root_device(SYSCTLFN_ARGS)
   1875 {
   1876 	struct sysctlnode node;
   1877 
   1878 	node = *rnode;
   1879 	node.sysctl_data = root_device->dv_xname;
   1880 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
   1881 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1882 }
   1883 
   1884 /*
   1885  * sysctl helper routine for kern.consdev, dependent on the current
   1886  * state of the console. Also used for machdep.console_device on some
   1887  * ports.
   1888  */
   1889 int
   1890 sysctl_consdev(SYSCTLFN_ARGS)
   1891 {
   1892 	dev_t consdev;
   1893 	uint32_t oconsdev;
   1894 	struct sysctlnode node;
   1895 
   1896 	if (cn_tab != NULL)
   1897 		consdev = cn_tab->cn_dev;
   1898 	else
   1899 		consdev = NODEV;
   1900 	node = *rnode;
   1901 	switch (*oldlenp) {
   1902 	case sizeof(consdev):
   1903 		node.sysctl_data = &consdev;
   1904 		node.sysctl_size = sizeof(consdev);
   1905 		break;
   1906 	case sizeof(oconsdev):
   1907 		oconsdev = (uint32_t)consdev;
   1908 		node.sysctl_data = &oconsdev;
   1909 		node.sysctl_size = sizeof(oconsdev);
   1910 		break;
   1911 	default:
   1912 		return EINVAL;
   1913 	}
   1914 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1915 }
   1916 
   1917 /*
   1918  * ********************************************************************
   1919  * section 4: support for some helpers
   1920  * ********************************************************************
   1921  */
   1922 
   1923 
   1924 /*
   1925  * Fill in a kinfo_lwp structure for the specified lwp.
   1926  */
   1927 static void
   1928 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
   1929 {
   1930 	struct proc *p = l->l_proc;
   1931 	struct timeval tv;
   1932 
   1933 	KASSERT(lwp_locked(l, NULL));
   1934 
   1935 	memset(kl, 0, sizeof(*kl));
   1936 
   1937 	kl->l_forw = 0;
   1938 	kl->l_back = 0;
   1939 	kl->l_laddr = PTRTOUINT64(l);
   1940 	kl->l_addr = PTRTOUINT64(l->l_addr);
   1941 	kl->l_stat = l->l_stat;
   1942 	kl->l_lid = l->l_lid;
   1943 	kl->l_flag = L_INMEM;
   1944 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
   1945 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
   1946 
   1947 	kl->l_swtime = l->l_swtime;
   1948 	kl->l_slptime = l->l_slptime;
   1949 	if (l->l_stat == LSONPROC)
   1950 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   1951 	else
   1952 		kl->l_schedflags = 0;
   1953 	kl->l_priority = lwp_eprio(l);
   1954 	kl->l_usrpri = l->l_priority;
   1955 	if (l->l_wchan)
   1956 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
   1957 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
   1958 	kl->l_cpuid = cpu_index(l->l_cpu);
   1959 	bintime2timeval(&l->l_rtime, &tv);
   1960 	kl->l_rtime_sec = tv.tv_sec;
   1961 	kl->l_rtime_usec = tv.tv_usec;
   1962 	kl->l_cpticks = l->l_cpticks;
   1963 	kl->l_pctcpu = l->l_pctcpu;
   1964 	kl->l_pid = p->p_pid;
   1965 	if (l->l_name == NULL)
   1966 		kl->l_name[0] = '\0';
   1967 	else
   1968 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
   1969 }
   1970