Home | History | Annotate | Line # | Download | only in kern
init_sysctl.c revision 1.187
      1 /*	$NetBSD: init_sysctl.c,v 1.187 2012/02/19 21:06:49 rmind 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 1.187 2012/02/19 21:06:49 rmind Exp $");
     34 
     35 #include "opt_sysv.h"
     36 #include "opt_compat_netbsd.h"
     37 #include "opt_modular.h"
     38 #include "opt_posix.h"
     39 #include "pty.h"
     40 
     41 #include <sys/types.h>
     42 #include <sys/param.h>
     43 #include <sys/sysctl.h>
     44 #include <sys/cpu.h>
     45 #include <sys/errno.h>
     46 #include <sys/systm.h>
     47 #include <sys/kernel.h>
     48 #include <sys/unistd.h>
     49 #include <sys/disklabel.h>
     50 #include <sys/cprng.h>
     51 #include <sys/vnode.h>
     52 #include <sys/mount.h>
     53 #include <sys/namei.h>
     54 #include <sys/msgbuf.h>
     55 #include <dev/cons.h>
     56 #include <sys/socketvar.h>
     57 #include <sys/file.h>
     58 #include <sys/filedesc.h>
     59 #include <sys/tty.h>
     60 #include <sys/kmem.h>
     61 #include <sys/resource.h>
     62 #include <sys/resourcevar.h>
     63 #include <sys/exec.h>
     64 #include <sys/conf.h>
     65 #include <sys/device.h>
     66 #include <sys/stat.h>
     67 #include <sys/kauth.h>
     68 #include <sys/ktrace.h>
     69 #include <sys/ksem.h>
     70 
     71 #ifdef COMPAT_50
     72 #include <compat/sys/time.h>
     73 #endif
     74 
     75 #include <sys/cpu.h>
     76 
     77 #if defined(MODULAR) || defined(P1003_1B_SEMAPHORE)
     78 int posix_semaphores = 200112;
     79 #else
     80 int posix_semaphores;
     81 #endif
     82 
     83 int security_setidcore_dump;
     84 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
     85 uid_t security_setidcore_owner = 0;
     86 gid_t security_setidcore_group = 0;
     87 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
     88 
     89 static const u_int sysctl_flagmap[] = {
     90 	PK_ADVLOCK, P_ADVLOCK,
     91 	PK_EXEC, P_EXEC,
     92 	PK_NOCLDWAIT, P_NOCLDWAIT,
     93 	PK_32, P_32,
     94 	PK_CLDSIGIGN, P_CLDSIGIGN,
     95 	PK_SUGID, P_SUGID,
     96 	0
     97 };
     98 
     99 static const u_int sysctl_sflagmap[] = {
    100 	PS_NOCLDSTOP, P_NOCLDSTOP,
    101 	PS_WEXIT, P_WEXIT,
    102 	PS_STOPFORK, P_STOPFORK,
    103 	PS_STOPEXEC, P_STOPEXEC,
    104 	PS_STOPEXIT, P_STOPEXIT,
    105 	0
    106 };
    107 
    108 static const u_int sysctl_slflagmap[] = {
    109 	PSL_TRACED, P_TRACED,
    110 	PSL_FSTRACE, P_FSTRACE,
    111 	PSL_CHTRACED, P_CHTRACED,
    112 	PSL_SYSCALL, P_SYSCALL,
    113 	0
    114 };
    115 
    116 static const u_int sysctl_lflagmap[] = {
    117 	PL_CONTROLT, P_CONTROLT,
    118 	PL_PPWAIT, P_PPWAIT,
    119 	0
    120 };
    121 
    122 static const u_int sysctl_stflagmap[] = {
    123 	PST_PROFIL, P_PROFIL,
    124 	0
    125 
    126 };
    127 
    128 static const u_int sysctl_lwpprflagmap[] = {
    129 	LPR_DETACHED, L_DETACHED,
    130 	0
    131 };
    132 
    133 /*
    134  * try over estimating by 5 procs/lwps
    135  */
    136 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
    137 
    138 static int dcopyout(struct lwp *, const void *, void *, size_t);
    139 
    140 static int
    141 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
    142 {
    143 	int error;
    144 
    145 	error = copyout(kaddr, uaddr, len);
    146 	ktrmibio(-1, UIO_READ, uaddr, len, error);
    147 
    148 	return error;
    149 }
    150 
    151 #ifdef DIAGNOSTIC
    152 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
    153 #endif
    154 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
    155 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
    156 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
    157 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
    158 static int sysctl_setlen(SYSCTLFN_PROTO);
    159 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
    160 static int sysctl_msgbuf(SYSCTLFN_PROTO);
    161 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
    162 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
    163 #if NPTY > 0
    164 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
    165 #endif /* NPTY > 0 */
    166 static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
    167 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
    168 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
    169 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
    170 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
    171 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
    172 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
    173 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
    174 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
    175 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
    176 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
    177 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
    178 
    179 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
    180 
    181 /*
    182  * ********************************************************************
    183  * section 1: setup routines
    184  * ********************************************************************
    185  * These functions are stuffed into a link set for sysctl setup
    186  * functions. They're never called or referenced from anywhere else.
    187  * ********************************************************************
    188  */
    189 
    190 /*
    191  * this setup routine is a replacement for kern_sysctl()
    192  */
    193 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
    194 {
    195 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
    196 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
    197 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
    198 	const struct sysctlnode *rnode;
    199 
    200 	sysctl_createv(clog, 0, NULL, NULL,
    201 		       CTLFLAG_PERMANENT,
    202 		       CTLTYPE_NODE, "kern", NULL,
    203 		       NULL, 0, NULL, 0,
    204 		       CTL_KERN, CTL_EOL);
    205 
    206 	sysctl_createv(clog, 0, NULL, NULL,
    207 		       CTLFLAG_PERMANENT,
    208 		       CTLTYPE_STRING, "ostype",
    209 		       SYSCTL_DESCR("Operating system type"),
    210 		       NULL, 0, __UNCONST(&ostype), 0,
    211 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
    212 	sysctl_createv(clog, 0, NULL, NULL,
    213 		       CTLFLAG_PERMANENT,
    214 		       CTLTYPE_STRING, "osrelease",
    215 		       SYSCTL_DESCR("Operating system release"),
    216 		       NULL, 0, __UNCONST(&osrelease), 0,
    217 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
    218 	sysctl_createv(clog, 0, NULL, NULL,
    219 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    220 		       CTLTYPE_INT, "osrevision",
    221 		       SYSCTL_DESCR("Operating system revision"),
    222 		       NULL, __NetBSD_Version__, NULL, 0,
    223 		       CTL_KERN, KERN_OSREV, CTL_EOL);
    224 	sysctl_createv(clog, 0, NULL, NULL,
    225 		       CTLFLAG_PERMANENT,
    226 		       CTLTYPE_STRING, "version",
    227 		       SYSCTL_DESCR("Kernel version"),
    228 		       NULL, 0, __UNCONST(&version), 0,
    229 		       CTL_KERN, KERN_VERSION, CTL_EOL);
    230 	sysctl_createv(clog, 0, NULL, NULL,
    231 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    232 		       CTLTYPE_INT, "maxvnodes",
    233 		       SYSCTL_DESCR("Maximum number of vnodes"),
    234 		       sysctl_kern_maxvnodes, 0, NULL, 0,
    235 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
    236 	sysctl_createv(clog, 0, NULL, NULL,
    237 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    238 		       CTLTYPE_INT, "maxproc",
    239 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
    240 		       sysctl_kern_maxproc, 0, NULL, 0,
    241 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
    242 	sysctl_createv(clog, 0, NULL, NULL,
    243 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    244 		       CTLTYPE_INT, "maxfiles",
    245 		       SYSCTL_DESCR("Maximum number of open files"),
    246 		       NULL, 0, &maxfiles, 0,
    247 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
    248 	sysctl_createv(clog, 0, NULL, NULL,
    249 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    250 		       CTLTYPE_INT, "argmax",
    251 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
    252 				    "execve(2)"),
    253 		       NULL, ARG_MAX, NULL, 0,
    254 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
    255 	sysctl_createv(clog, 0, NULL, NULL,
    256 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    257 		       CTLTYPE_STRING, "hostname",
    258 		       SYSCTL_DESCR("System hostname"),
    259 		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
    260 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
    261 	sysctl_createv(clog, 0, NULL, NULL,
    262 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    263 		       CTLTYPE_INT, "hostid",
    264 		       SYSCTL_DESCR("System host ID number"),
    265 		       sysctl_kern_hostid, 0, NULL, 0,
    266 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
    267 	sysctl_createv(clog, 0, NULL, NULL,
    268 		       CTLFLAG_PERMANENT,
    269 		       CTLTYPE_STRUCT, "clockrate",
    270 		       SYSCTL_DESCR("Kernel clock rates"),
    271 		       sysctl_kern_clockrate, 0, NULL,
    272 		       sizeof(struct clockinfo),
    273 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
    274 	sysctl_createv(clog, 0, NULL, NULL,
    275 		       CTLFLAG_PERMANENT,
    276 		       CTLTYPE_INT, "hardclock_ticks",
    277 		       SYSCTL_DESCR("Number of hardclock ticks"),
    278 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
    279 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
    280 	sysctl_createv(clog, 0, NULL, NULL,
    281 		       CTLFLAG_PERMANENT,
    282 		       CTLTYPE_STRUCT, "vnode",
    283 		       SYSCTL_DESCR("System vnode table"),
    284 		       sysctl_kern_vnode, 0, NULL, 0,
    285 		       CTL_KERN, KERN_VNODE, CTL_EOL);
    286 #ifndef GPROF
    287 	sysctl_createv(clog, 0, NULL, NULL,
    288 		       CTLFLAG_PERMANENT,
    289 		       CTLTYPE_NODE, "profiling",
    290 		       SYSCTL_DESCR("Profiling information (not available)"),
    291 		       sysctl_notavail, 0, NULL, 0,
    292 		       CTL_KERN, KERN_PROF, CTL_EOL);
    293 #endif
    294 	sysctl_createv(clog, 0, NULL, NULL,
    295 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    296 		       CTLTYPE_INT, "posix1version",
    297 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
    298 				    "with which the operating system attempts "
    299 				    "to comply"),
    300 		       NULL, _POSIX_VERSION, NULL, 0,
    301 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
    302 	sysctl_createv(clog, 0, NULL, NULL,
    303 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    304 		       CTLTYPE_INT, "ngroups",
    305 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
    306 		       NULL, NGROUPS_MAX, NULL, 0,
    307 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
    308 	sysctl_createv(clog, 0, NULL, NULL,
    309 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    310 		       CTLTYPE_INT, "job_control",
    311 		       SYSCTL_DESCR("Whether job control is available"),
    312 		       NULL, 1, NULL, 0,
    313 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
    314 	sysctl_createv(clog, 0, NULL, NULL,
    315 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    316 		       CTLTYPE_INT, "saved_ids",
    317 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
    318 				    "available"), NULL,
    319 #ifdef _POSIX_SAVED_IDS
    320 		       1,
    321 #else /* _POSIX_SAVED_IDS */
    322 		       0,
    323 #endif /* _POSIX_SAVED_IDS */
    324 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
    325 	sysctl_createv(clog, 0, NULL, NULL,
    326 		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
    327 		       CTLTYPE_INT, "boothowto",
    328 		       SYSCTL_DESCR("Flags from boot loader"),
    329 		       NULL, 0, &boothowto, sizeof(boothowto),
    330 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    331 	sysctl_createv(clog, 0, NULL, NULL,
    332 		       CTLFLAG_PERMANENT,
    333 		       CTLTYPE_STRUCT, "boottime",
    334 		       SYSCTL_DESCR("System boot time"),
    335 		       NULL, 0, &boottime, sizeof(boottime),
    336 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
    337 #ifdef COMPAT_50
    338 	{
    339 		extern struct timeval50 boottime50;
    340 		sysctl_createv(clog, 0, NULL, NULL,
    341 			       CTLFLAG_PERMANENT,
    342 			       CTLTYPE_STRUCT, "oboottime",
    343 			       SYSCTL_DESCR("System boot time"),
    344 			       NULL, 0, &boottime50, sizeof(boottime50),
    345 			       CTL_KERN, KERN_OBOOTTIME, CTL_EOL);
    346 	}
    347 #endif
    348 	sysctl_createv(clog, 0, NULL, NULL,
    349 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    350 		       CTLTYPE_STRING, "domainname",
    351 		       SYSCTL_DESCR("YP domain name"),
    352 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
    353 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
    354 	sysctl_createv(clog, 0, NULL, NULL,
    355 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    356 		       CTLTYPE_INT, "maxpartitions",
    357 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
    358 				    "disk"),
    359 		       NULL, MAXPARTITIONS, NULL, 0,
    360 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
    361 	sysctl_createv(clog, 0, NULL, NULL,
    362 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    363 		       CTLTYPE_INT, "rawpartition",
    364 		       SYSCTL_DESCR("Raw partition of a disk"),
    365 		       NULL, RAW_PART, NULL, 0,
    366 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
    367 	sysctl_createv(clog, 0, NULL, NULL,
    368 		       CTLFLAG_PERMANENT,
    369 		       CTLTYPE_STRUCT, "timex", NULL,
    370 		       sysctl_notavail, 0, NULL, 0,
    371 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
    372 	sysctl_createv(clog, 0, NULL, NULL,
    373 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    374 		       CTLTYPE_INT, "rtc_offset",
    375 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
    376 				    "minutes"),
    377 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
    378 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
    379 	sysctl_createv(clog, 0, NULL, NULL,
    380 		       CTLFLAG_PERMANENT,
    381 		       CTLTYPE_STRING, "root_device",
    382 		       SYSCTL_DESCR("Name of the root device"),
    383 		       sysctl_root_device, 0, NULL, 0,
    384 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
    385 	sysctl_createv(clog, 0, NULL, NULL,
    386 		       CTLFLAG_PERMANENT,
    387 		       CTLTYPE_INT, "msgbufsize",
    388 		       SYSCTL_DESCR("Size of the kernel message buffer"),
    389 		       sysctl_msgbuf, 0, NULL, 0,
    390 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
    391 	sysctl_createv(clog, 0, NULL, NULL,
    392 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    393 		       CTLTYPE_INT, "fsync",
    394 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
    395 				    "Synchronization Option is available on "
    396 				    "this system"),
    397 		       NULL, 1, NULL, 0,
    398 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
    399 	sysctl_createv(clog, 0, NULL, NULL,
    400 		       CTLFLAG_PERMANENT,
    401 		       CTLTYPE_NODE, "ipc",
    402 		       SYSCTL_DESCR("SysV IPC options"),
    403 		       NULL, 0, NULL, 0,
    404 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
    405 	sysctl_createv(clog, 0, NULL, NULL,
    406 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    407 		       CTLTYPE_INT, "sysvmsg",
    408 		       SYSCTL_DESCR("System V style message support available"),
    409 		       NULL,
    410 #ifdef SYSVMSG
    411 		       1,
    412 #else /* SYSVMSG */
    413 		       0,
    414 #endif /* SYSVMSG */
    415 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
    416 	sysctl_createv(clog, 0, NULL, NULL,
    417 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    418 		       CTLTYPE_INT, "sysvsem",
    419 		       SYSCTL_DESCR("System V style semaphore support "
    420 				    "available"), NULL,
    421 #ifdef SYSVSEM
    422 		       1,
    423 #else /* SYSVSEM */
    424 		       0,
    425 #endif /* SYSVSEM */
    426 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
    427 	sysctl_createv(clog, 0, NULL, NULL,
    428 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    429 		       CTLTYPE_INT, "sysvshm",
    430 		       SYSCTL_DESCR("System V style shared memory support "
    431 				    "available"), NULL,
    432 #ifdef SYSVSHM
    433 		       1,
    434 #else /* SYSVSHM */
    435 		       0,
    436 #endif /* SYSVSHM */
    437 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
    438 	sysctl_createv(clog, 0, NULL, NULL,
    439 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    440 		       CTLTYPE_INT, "synchronized_io",
    441 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
    442 				    "I/O Option is available on this system"),
    443 		       NULL, 1, NULL, 0,
    444 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
    445 	sysctl_createv(clog, 0, NULL, NULL,
    446 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    447 		       CTLTYPE_INT, "iov_max",
    448 		       SYSCTL_DESCR("Maximum number of iovec structures per "
    449 				    "process"),
    450 		       NULL, IOV_MAX, NULL, 0,
    451 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
    452 	sysctl_createv(clog, 0, NULL, NULL,
    453 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    454 		       CTLTYPE_INT, "mapped_files",
    455 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
    456 				    "Files Option is available on this system"),
    457 		       NULL, 1, NULL, 0,
    458 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
    459 	sysctl_createv(clog, 0, NULL, NULL,
    460 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    461 		       CTLTYPE_INT, "memlock",
    462 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
    463 				    "Locking Option is available on this "
    464 				    "system"),
    465 		       NULL, 1, NULL, 0,
    466 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
    467 	sysctl_createv(clog, 0, NULL, NULL,
    468 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    469 		       CTLTYPE_INT, "memlock_range",
    470 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
    471 				    "Locking Option is available on this "
    472 				    "system"),
    473 		       NULL, 1, NULL, 0,
    474 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
    475 	sysctl_createv(clog, 0, NULL, NULL,
    476 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    477 		       CTLTYPE_INT, "memory_protection",
    478 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
    479 				    "Protection Option is available on this "
    480 				    "system"),
    481 		       NULL, 1, NULL, 0,
    482 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
    483 	sysctl_createv(clog, 0, NULL, NULL,
    484 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    485 		       CTLTYPE_INT, "login_name_max",
    486 		       SYSCTL_DESCR("Maximum login name length"),
    487 		       NULL, LOGIN_NAME_MAX, NULL, 0,
    488 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
    489 	sysctl_createv(clog, 0, NULL, NULL,
    490 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    491 		       CTLTYPE_STRING, "defcorename",
    492 		       SYSCTL_DESCR("Default core file name"),
    493 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
    494 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
    495 	sysctl_createv(clog, 0, NULL, NULL,
    496 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    497 		       CTLTYPE_INT, "logsigexit",
    498 		       SYSCTL_DESCR("Log process exit when caused by signals"),
    499 		       NULL, 0, &kern_logsigexit, 0,
    500 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
    501 	sysctl_createv(clog, 0, NULL, NULL,
    502 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    503 		       CTLTYPE_INT, "fscale",
    504 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
    505 		       NULL, FSCALE, NULL, 0,
    506 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
    507 	sysctl_createv(clog, 0, NULL, NULL,
    508 		       CTLFLAG_PERMANENT,
    509 		       CTLTYPE_INT, "ccpu",
    510 		       SYSCTL_DESCR("Scheduler exponential decay value"),
    511 		       NULL, 0, &ccpu, 0,
    512 		       CTL_KERN, KERN_CCPU, CTL_EOL);
    513 	sysctl_createv(clog, 0, NULL, NULL,
    514 		       CTLFLAG_PERMANENT,
    515 		       CTLTYPE_STRUCT, "cp_time",
    516 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
    517 		       sysctl_kern_cptime, 0, NULL, 0,
    518 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
    519 	sysctl_createv(clog, 0, NULL, NULL,
    520 		       CTLFLAG_PERMANENT,
    521 		       CTLTYPE_INT, "msgbuf",
    522 		       SYSCTL_DESCR("Kernel message buffer"),
    523 		       sysctl_msgbuf, 0, NULL, 0,
    524 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
    525 	sysctl_createv(clog, 0, NULL, NULL,
    526 		       CTLFLAG_PERMANENT,
    527 		       CTLTYPE_STRUCT, "consdev",
    528 		       SYSCTL_DESCR("Console device"),
    529 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
    530 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
    531 #if NPTY > 0
    532 	sysctl_createv(clog, 0, NULL, NULL,
    533 		       CTLFLAG_PERMANENT,
    534 		       CTLTYPE_INT, "maxptys",
    535 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
    536 		       sysctl_kern_maxptys, 0, NULL, 0,
    537 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
    538 #endif /* NPTY > 0 */
    539 	sysctl_createv(clog, 0, NULL, NULL,
    540 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    541 		       CTLTYPE_INT, "maxphys",
    542 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
    543 		       NULL, MAXPHYS, NULL, 0,
    544 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
    545 	sysctl_createv(clog, 0, NULL, NULL,
    546 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    547 		       CTLTYPE_INT, "sbmax",
    548 		       SYSCTL_DESCR("Maximum socket buffer size"),
    549 		       sysctl_kern_sbmax, 0, NULL, 0,
    550 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
    551 	sysctl_createv(clog, 0, NULL, NULL,
    552 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    553 		       CTLTYPE_INT, "monotonic_clock",
    554 		       SYSCTL_DESCR("Implementation version of the POSIX "
    555 				    "1003.1b Monotonic Clock Option"),
    556 		       /* XXX _POSIX_VERSION */
    557 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
    558 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
    559 	sysctl_createv(clog, 0, NULL, NULL,
    560 		       CTLFLAG_PERMANENT,
    561 		       CTLTYPE_INT, "urandom",
    562 		       SYSCTL_DESCR("Random integer value"),
    563 		       sysctl_kern_urnd, 0, NULL, 0,
    564 		       CTL_KERN, KERN_URND, CTL_EOL);
    565 	sysctl_createv(clog, 0, NULL, NULL,
    566 		       CTLFLAG_PERMANENT,
    567 		       CTLTYPE_INT, "arandom",
    568 		       SYSCTL_DESCR("n bytes of random data"),
    569 		       sysctl_kern_arnd, 0, NULL, 0,
    570 		       CTL_KERN, KERN_ARND, CTL_EOL);
    571 	sysctl_createv(clog, 0, NULL, NULL,
    572 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    573 		       CTLTYPE_INT, "labelsector",
    574 		       SYSCTL_DESCR("Sector number containing the disklabel"),
    575 		       NULL, LABELSECTOR, NULL, 0,
    576 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
    577 	sysctl_createv(clog, 0, NULL, NULL,
    578 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    579 		       CTLTYPE_INT, "labeloffset",
    580 		       SYSCTL_DESCR("Offset of the disklabel within the "
    581 				    "sector"),
    582 		       NULL, LABELOFFSET, NULL, 0,
    583 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
    584 	sysctl_createv(clog, 0, NULL, NULL,
    585 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    586 		       CTLTYPE_INT, "labelusesmbr",
    587 		       SYSCTL_DESCR("disklabel is inside MBR partition"),
    588 		       NULL, LABELUSESMBR, NULL, 0,
    589 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    590 	sysctl_createv(clog, 0, NULL, NULL,
    591 		       CTLFLAG_PERMANENT,
    592 		       CTLTYPE_NODE, "lwp",
    593 		       SYSCTL_DESCR("System-wide LWP information"),
    594 		       sysctl_kern_lwp, 0, NULL, 0,
    595 		       CTL_KERN, KERN_LWP, CTL_EOL);
    596 	sysctl_createv(clog, 0, NULL, NULL,
    597 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    598 		       CTLTYPE_INT, "forkfsleep",
    599 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
    600 				    "to process limits"),
    601 		       sysctl_kern_forkfsleep, 0, NULL, 0,
    602 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
    603 	sysctl_createv(clog, 0, NULL, NULL,
    604 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    605 		       CTLTYPE_INT, "posix_threads",
    606 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    607 				    "Threads option to which the system "
    608 				    "attempts to conform"),
    609 		       /* XXX _POSIX_VERSION */
    610 		       NULL, _POSIX_THREADS, NULL, 0,
    611 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
    612 	sysctl_createv(clog, 0, NULL, NULL,
    613 		       CTLFLAG_PERMANENT,
    614 		       CTLTYPE_INT, "posix_semaphores",
    615 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    616 				    "Semaphores option to which the system "
    617 				    "attempts to conform"), NULL,
    618 		       0, &posix_semaphores,
    619 		       0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
    620 	sysctl_createv(clog, 0, NULL, NULL,
    621 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    622 		       CTLTYPE_INT, "posix_barriers",
    623 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    624 				    "Barriers option to which the system "
    625 				    "attempts to conform"),
    626 		       /* XXX _POSIX_VERSION */
    627 		       NULL, _POSIX_BARRIERS, NULL, 0,
    628 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
    629 	sysctl_createv(clog, 0, NULL, NULL,
    630 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    631 		       CTLTYPE_INT, "posix_timers",
    632 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    633 				    "Timers option to which the system "
    634 				    "attempts to conform"),
    635 		       /* XXX _POSIX_VERSION */
    636 		       NULL, _POSIX_TIMERS, NULL, 0,
    637 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
    638 	sysctl_createv(clog, 0, NULL, NULL,
    639 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    640 		       CTLTYPE_INT, "posix_spin_locks",
    641 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
    642 				    "Locks option to which the system attempts "
    643 				    "to conform"),
    644 		       /* XXX _POSIX_VERSION */
    645 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
    646 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
    647 	sysctl_createv(clog, 0, NULL, NULL,
    648 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    649 		       CTLTYPE_INT, "posix_reader_writer_locks",
    650 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
    651 				    "Read-Write Locks option to which the "
    652 				    "system attempts to conform"),
    653 		       /* XXX _POSIX_VERSION */
    654 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
    655 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
    656 	sysctl_createv(clog, 0, NULL, NULL,
    657 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    658 		       CTLTYPE_INT, "dump_on_panic",
    659 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
    660 		       NULL, 0, &dumponpanic, 0,
    661 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
    662 #ifdef DIAGNOSTIC
    663 	sysctl_createv(clog, 0, NULL, NULL,
    664 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    665 		       CTLTYPE_INT, "panic_now",
    666 		       SYSCTL_DESCR("Trigger a panic"),
    667 		       sysctl_kern_trigger_panic, 0, NULL, 0,
    668 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    669 #endif
    670 	sysctl_createv(clog, 0, NULL, NULL,
    671 		       CTLFLAG_PERMANENT,
    672 		       CTLTYPE_INT, "root_partition",
    673 		       SYSCTL_DESCR("Root partition on the root device"),
    674 		       sysctl_kern_root_partition, 0, NULL, 0,
    675 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
    676 	sysctl_createv(clog, 0, NULL, NULL,
    677 		       CTLFLAG_PERMANENT,
    678 		       CTLTYPE_STRUCT, "drivers",
    679 		       SYSCTL_DESCR("List of all drivers with block and "
    680 				    "character device numbers"),
    681 		       sysctl_kern_drivers, 0, NULL, 0,
    682 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
    683 	sysctl_createv(clog, 0, NULL, NULL,
    684 		       CTLFLAG_PERMANENT,
    685 		       CTLTYPE_STRUCT, "cp_id",
    686 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
    687 		       sysctl_kern_cpid, 0, NULL, 0,
    688 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
    689 	sysctl_createv(clog, 0, NULL, &rnode,
    690 		       CTLFLAG_PERMANENT,
    691 		       CTLTYPE_NODE, "coredump",
    692 		       SYSCTL_DESCR("Coredump settings."),
    693 		       NULL, 0, NULL, 0,
    694 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    695 	sysctl_createv(clog, 0, &rnode, &rnode,
    696 		       CTLFLAG_PERMANENT,
    697 		       CTLTYPE_NODE, "setid",
    698 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
    699 		       NULL, 0, NULL, 0,
    700 		       CTL_CREATE, CTL_EOL);
    701 	sysctl_createv(clog, 0, &rnode, NULL,
    702 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    703 		       CTLTYPE_INT, "dump",
    704 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
    705 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
    706 		       sizeof(security_setidcore_dump),
    707 		       CTL_CREATE, CTL_EOL);
    708 	sysctl_createv(clog, 0, &rnode, NULL,
    709 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    710 		       CTLTYPE_STRING, "path",
    711 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
    712 		       sysctl_security_setidcorename, 0,
    713 		       &security_setidcore_path,
    714 		       sizeof(security_setidcore_path),
    715 		       CTL_CREATE, CTL_EOL);
    716 	sysctl_createv(clog, 0, &rnode, NULL,
    717 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    718 		       CTLTYPE_INT, "owner",
    719 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
    720 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
    721 		       0,
    722 		       CTL_CREATE, CTL_EOL);
    723 	sysctl_createv(clog, 0, &rnode, NULL,
    724 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    725 		       CTLTYPE_INT, "group",
    726 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
    727 		       sysctl_security_setidcore, 0, &security_setidcore_group,
    728 		       0,
    729 		       CTL_CREATE, CTL_EOL);
    730 	sysctl_createv(clog, 0, &rnode, NULL,
    731 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    732 		       CTLTYPE_INT, "mode",
    733 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
    734 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
    735 		       0,
    736 		       CTL_CREATE, CTL_EOL);
    737 	sysctl_createv(clog, 0, NULL, NULL,
    738 		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
    739 		       CTLTYPE_INT, "no_sa_support",
    740 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
    741 		       "it doesn't"),
    742 		       NULL, 1, NULL, 0,
    743 		       CTL_KERN, CTL_CREATE, CTL_EOL);
    744 	/* kern.posix. */
    745 	sysctl_createv(clog, 0, NULL, &rnode,
    746 			CTLFLAG_PERMANENT,
    747 			CTLTYPE_NODE, "posix",
    748 			SYSCTL_DESCR("POSIX options"),
    749 			NULL, 0, NULL, 0,
    750 			CTL_KERN, CTL_CREATE, CTL_EOL);
    751 	sysctl_createv(clog, 0, &rnode, NULL,
    752 			CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
    753 			CTLTYPE_INT, "semmax",
    754 			SYSCTL_DESCR("Maximal number of semaphores"),
    755 			NULL, 0, &ksem_max, 0,
    756 			CTL_CREATE, CTL_EOL);
    757 }
    758 
    759 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
    760 {
    761 	u_int u;
    762 	u_quad_t q;
    763 
    764 	sysctl_createv(clog, 0, NULL, NULL,
    765 		       CTLFLAG_PERMANENT,
    766 		       CTLTYPE_NODE, "hw", NULL,
    767 		       NULL, 0, NULL, 0,
    768 		       CTL_HW, CTL_EOL);
    769 
    770 	sysctl_createv(clog, 0, NULL, NULL,
    771 		       CTLFLAG_PERMANENT,
    772 		       CTLTYPE_STRING, "machine",
    773 		       SYSCTL_DESCR("Machine class"),
    774 		       NULL, 0, machine, 0,
    775 		       CTL_HW, HW_MACHINE, CTL_EOL);
    776 	sysctl_createv(clog, 0, NULL, NULL,
    777 		       CTLFLAG_PERMANENT,
    778 		       CTLTYPE_STRING, "model",
    779 		       SYSCTL_DESCR("Machine model"),
    780 		       NULL, 0, cpu_model, 0,
    781 		       CTL_HW, HW_MODEL, CTL_EOL);
    782 	sysctl_createv(clog, 0, NULL, NULL,
    783 		       CTLFLAG_PERMANENT,
    784 		       CTLTYPE_INT, "ncpu",
    785 		       SYSCTL_DESCR("Number of CPUs configured"),
    786 		       NULL, 0, &ncpu, 0,
    787 		       CTL_HW, HW_NCPU, CTL_EOL);
    788 	sysctl_createv(clog, 0, NULL, NULL,
    789 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    790 		       CTLTYPE_INT, "byteorder",
    791 		       SYSCTL_DESCR("System byte order"),
    792 		       NULL, BYTE_ORDER, NULL, 0,
    793 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
    794 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
    795 		UINT_MAX : physmem * PAGE_SIZE;
    796 	sysctl_createv(clog, 0, NULL, NULL,
    797 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    798 		       CTLTYPE_INT, "physmem",
    799 		       SYSCTL_DESCR("Bytes of physical memory"),
    800 		       NULL, u, NULL, 0,
    801 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
    802 	sysctl_createv(clog, 0, NULL, NULL,
    803 		       CTLFLAG_PERMANENT,
    804 		       CTLTYPE_INT, "usermem",
    805 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    806 		       sysctl_hw_usermem, 0, NULL, 0,
    807 		       CTL_HW, HW_USERMEM, CTL_EOL);
    808 	sysctl_createv(clog, 0, NULL, NULL,
    809 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    810 		       CTLTYPE_INT, "pagesize",
    811 		       SYSCTL_DESCR("Software page size"),
    812 		       NULL, PAGE_SIZE, NULL, 0,
    813 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
    814 	sysctl_createv(clog, 0, NULL, NULL,
    815 		       CTLFLAG_PERMANENT,
    816 		       CTLTYPE_STRING, "machine_arch",
    817 		       SYSCTL_DESCR("Machine CPU class"),
    818 		       NULL, 0, machine_arch, 0,
    819 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
    820 	sysctl_createv(clog, 0, NULL, NULL,
    821 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    822 		       CTLTYPE_INT, "alignbytes",
    823 		       SYSCTL_DESCR("Alignment constraint for all possible "
    824 				    "data types"),
    825 		       NULL, ALIGNBYTES, NULL, 0,
    826 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
    827 	sysctl_createv(clog, 0, NULL, NULL,
    828 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
    829 		       CTLTYPE_STRING, "cnmagic",
    830 		       SYSCTL_DESCR("Console magic key sequence"),
    831 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
    832 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
    833 	q = (u_quad_t)physmem * PAGE_SIZE;
    834 	sysctl_createv(clog, 0, NULL, NULL,
    835 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
    836 		       CTLTYPE_QUAD, "physmem64",
    837 		       SYSCTL_DESCR("Bytes of physical memory"),
    838 		       NULL, q, NULL, 0,
    839 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
    840 	sysctl_createv(clog, 0, NULL, NULL,
    841 		       CTLFLAG_PERMANENT,
    842 		       CTLTYPE_QUAD, "usermem64",
    843 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
    844 		       sysctl_hw_usermem, 0, NULL, 0,
    845 		       CTL_HW, HW_USERMEM64, CTL_EOL);
    846 	sysctl_createv(clog, 0, NULL, NULL,
    847 		       CTLFLAG_PERMANENT,
    848 		       CTLTYPE_INT, "ncpuonline",
    849 		       SYSCTL_DESCR("Number of CPUs online"),
    850 		       NULL, 0, &ncpuonline, 0,
    851 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
    852 }
    853 
    854 #ifdef DEBUG
    855 /*
    856  * Debugging related system variables.
    857  */
    858 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
    859 struct ctldebug debug5, debug6, debug7, debug8, debug9;
    860 struct ctldebug debug10, debug11, debug12, debug13, debug14;
    861 struct ctldebug debug15, debug16, debug17, debug18, debug19;
    862 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
    863 	&debug0, &debug1, &debug2, &debug3, &debug4,
    864 	&debug5, &debug6, &debug7, &debug8, &debug9,
    865 	&debug10, &debug11, &debug12, &debug13, &debug14,
    866 	&debug15, &debug16, &debug17, &debug18, &debug19,
    867 };
    868 
    869 /*
    870  * this setup routine is a replacement for debug_sysctl()
    871  *
    872  * note that it creates several nodes per defined debug variable
    873  */
    874 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
    875 {
    876 	struct ctldebug *cdp;
    877 	char nodename[20];
    878 	int i;
    879 
    880 	/*
    881 	 * two ways here:
    882 	 *
    883 	 * the "old" way (debug.name -> value) which was emulated by
    884 	 * the sysctl(8) binary
    885 	 *
    886 	 * the new way, which the sysctl(8) binary was actually using
    887 
    888 	 node	debug
    889 	 node	debug.0
    890 	 string debug.0.name
    891 	 int	debug.0.value
    892 	 int	debug.name
    893 
    894 	 */
    895 
    896 	sysctl_createv(clog, 0, NULL, NULL,
    897 		       CTLFLAG_PERMANENT,
    898 		       CTLTYPE_NODE, "debug", NULL,
    899 		       NULL, 0, NULL, 0,
    900 		       CTL_DEBUG, CTL_EOL);
    901 
    902 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
    903 		cdp = debugvars[i];
    904 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
    905 			continue;
    906 
    907 		snprintf(nodename, sizeof(nodename), "debug%d", i);
    908 		sysctl_createv(clog, 0, NULL, NULL,
    909 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    910 			       CTLTYPE_NODE, nodename, NULL,
    911 			       NULL, 0, NULL, 0,
    912 			       CTL_DEBUG, i, CTL_EOL);
    913 		sysctl_createv(clog, 0, NULL, NULL,
    914 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    915 			       CTLTYPE_STRING, "name", NULL,
    916 			       /*XXXUNCONST*/
    917 			       NULL, 0, __UNCONST(cdp->debugname), 0,
    918 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
    919 		sysctl_createv(clog, 0, NULL, NULL,
    920 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
    921 			       CTLTYPE_INT, "value", NULL,
    922 			       NULL, 0, cdp->debugvar, 0,
    923 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
    924 		sysctl_createv(clog, 0, NULL, NULL,
    925 			       CTLFLAG_PERMANENT,
    926 			       CTLTYPE_INT, cdp->debugname, NULL,
    927 			       NULL, 0, cdp->debugvar, 0,
    928 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
    929 	}
    930 }
    931 #endif /* DEBUG */
    932 
    933 /*
    934  * ********************************************************************
    935  * section 2: private node-specific helper routines.
    936  * ********************************************************************
    937  */
    938 
    939 #ifdef DIAGNOSTIC
    940 static int
    941 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
    942 {
    943 	int newtrig, error;
    944 	struct sysctlnode node;
    945 
    946 	newtrig = 0;
    947 	node = *rnode;
    948 	node.sysctl_data = &newtrig;
    949 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
    950 	if (error || newp == NULL)
    951 		return (error);
    952 
    953 	if (newtrig != 0)
    954 		panic("Panic triggered");
    955 
    956 	return (error);
    957 }
    958 #endif
    959 
    960 /*
    961  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
    962  * new value is lower than desiredvnodes and then calls reinit
    963  * routines that needs to adjust to the new value.
    964  */
    965 static int
    966 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
    967 {
    968 	int error, new_vnodes, old_vnodes, new_max;
    969 	struct sysctlnode node;
    970 
    971 	new_vnodes = desiredvnodes;
    972 	node = *rnode;
    973 	node.sysctl_data = &new_vnodes;
    974 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
    975 	if (error || newp == NULL)
    976 		return (error);
    977 
    978 	/* Limits: 75% of KVA and physical memory. */
    979 	new_max = calc_cache_size(kernel_map, 75, 75) / VNODE_COST;
    980 	if (new_vnodes > new_max)
    981 		new_vnodes = new_max;
    982 
    983 	old_vnodes = desiredvnodes;
    984 	desiredvnodes = new_vnodes;
    985 	if (new_vnodes < old_vnodes) {
    986 		error = vfs_drainvnodes(new_vnodes);
    987 		if (error) {
    988 			desiredvnodes = old_vnodes;
    989 			return (error);
    990 		}
    991 	}
    992 	vfs_reinit();
    993 	nchreinit();
    994 
    995 	return (0);
    996 }
    997 
    998 /*
    999  * sysctl helper routine for rtc_offset - set time after changes
   1000  */
   1001 static int
   1002 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
   1003 {
   1004 	struct timespec ts, delta;
   1005 	int error, new_rtc_offset;
   1006 	struct sysctlnode node;
   1007 
   1008 	new_rtc_offset = rtc_offset;
   1009 	node = *rnode;
   1010 	node.sysctl_data = &new_rtc_offset;
   1011 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1012 	if (error || newp == NULL)
   1013 		return (error);
   1014 
   1015 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
   1016 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
   1017 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
   1018 		return (EPERM);
   1019 	if (rtc_offset == new_rtc_offset)
   1020 		return (0);
   1021 
   1022 	/* if we change the offset, adjust the time */
   1023 	nanotime(&ts);
   1024 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
   1025 	delta.tv_nsec = 0;
   1026 	timespecadd(&ts, &delta, &ts);
   1027 	rtc_offset = new_rtc_offset;
   1028 	return (settime(l->l_proc, &ts));
   1029 }
   1030 
   1031 /*
   1032  * sysctl helper routine for kern.maxproc. Ensures that the new
   1033  * values are not too low or too high.
   1034  */
   1035 static int
   1036 sysctl_kern_maxproc(SYSCTLFN_ARGS)
   1037 {
   1038 	int error, nmaxproc;
   1039 	struct sysctlnode node;
   1040 
   1041 	nmaxproc = maxproc;
   1042 	node = *rnode;
   1043 	node.sysctl_data = &nmaxproc;
   1044 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1045 	if (error || newp == NULL)
   1046 		return (error);
   1047 
   1048 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
   1049 		return (EINVAL);
   1050 #ifdef __HAVE_CPU_MAXPROC
   1051 	if (nmaxproc > cpu_maxproc())
   1052 		return (EINVAL);
   1053 #endif
   1054 	maxproc = nmaxproc;
   1055 
   1056 	return (0);
   1057 }
   1058 
   1059 /*
   1060  * sysctl helper function for kern.hostid. The hostid is a long, but
   1061  * we export it as an int, so we need to give it a little help.
   1062  */
   1063 static int
   1064 sysctl_kern_hostid(SYSCTLFN_ARGS)
   1065 {
   1066 	int error, inthostid;
   1067 	struct sysctlnode node;
   1068 
   1069 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
   1070 	node = *rnode;
   1071 	node.sysctl_data = &inthostid;
   1072 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1073 	if (error || newp == NULL)
   1074 		return (error);
   1075 
   1076 	hostid = (unsigned)inthostid;
   1077 
   1078 	return (0);
   1079 }
   1080 
   1081 /*
   1082  * sysctl helper function for kern.hostname and kern.domainnname.
   1083  * resets the relevant recorded length when the underlying name is
   1084  * changed.
   1085  */
   1086 static int
   1087 sysctl_setlen(SYSCTLFN_ARGS)
   1088 {
   1089 	int error;
   1090 
   1091 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
   1092 	if (error || newp == NULL)
   1093 		return (error);
   1094 
   1095 	switch (rnode->sysctl_num) {
   1096 	case KERN_HOSTNAME:
   1097 		hostnamelen = strlen((const char*)rnode->sysctl_data);
   1098 		break;
   1099 	case KERN_DOMAINNAME:
   1100 		domainnamelen = strlen((const char*)rnode->sysctl_data);
   1101 		break;
   1102 	}
   1103 
   1104 	return (0);
   1105 }
   1106 
   1107 /*
   1108  * sysctl helper routine for kern.clockrate. Assembles a struct on
   1109  * the fly to be returned to the caller.
   1110  */
   1111 static int
   1112 sysctl_kern_clockrate(SYSCTLFN_ARGS)
   1113 {
   1114 	struct clockinfo clkinfo;
   1115 	struct sysctlnode node;
   1116 
   1117 	clkinfo.tick = tick;
   1118 	clkinfo.tickadj = tickadj;
   1119 	clkinfo.hz = hz;
   1120 	clkinfo.profhz = profhz;
   1121 	clkinfo.stathz = stathz ? stathz : hz;
   1122 
   1123 	node = *rnode;
   1124 	node.sysctl_data = &clkinfo;
   1125 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1126 }
   1127 
   1128 /*
   1129  * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
   1130  * former it merely checks the message buffer is set up. For the latter,
   1131  * it also copies out the data if necessary.
   1132  */
   1133 static int
   1134 sysctl_msgbuf(SYSCTLFN_ARGS)
   1135 {
   1136 	char *where = oldp;
   1137 	size_t len, maxlen;
   1138 	long beg, end;
   1139 	extern kmutex_t log_lock;
   1140 	int error;
   1141 
   1142 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
   1143 		msgbufenabled = 0;
   1144 		return (ENXIO);
   1145 	}
   1146 
   1147 	switch (rnode->sysctl_num) {
   1148 	case KERN_MSGBUFSIZE: {
   1149 		struct sysctlnode node = *rnode;
   1150 		int msg_bufs = (int)msgbufp->msg_bufs;
   1151 		node.sysctl_data = &msg_bufs;
   1152 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1153 	}
   1154 	case KERN_MSGBUF:
   1155 		break;
   1156 	default:
   1157 		return (EOPNOTSUPP);
   1158 	}
   1159 
   1160 	if (newp != NULL)
   1161 		return (EPERM);
   1162 
   1163 	if (oldp == NULL) {
   1164 		/* always return full buffer size */
   1165 		*oldlenp = msgbufp->msg_bufs;
   1166 		return (0);
   1167 	}
   1168 
   1169 	sysctl_unlock();
   1170 
   1171 	/*
   1172 	 * First, copy from the write pointer to the end of
   1173 	 * message buffer.
   1174 	 */
   1175 	error = 0;
   1176 	mutex_spin_enter(&log_lock);
   1177 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
   1178 	beg = msgbufp->msg_bufx;
   1179 	end = msgbufp->msg_bufs;
   1180 	mutex_spin_exit(&log_lock);
   1181 
   1182 	while (maxlen > 0) {
   1183 		len = MIN(end - beg, maxlen);
   1184 		if (len == 0)
   1185 			break;
   1186 		/* XXX unlocked, but hardly matters. */
   1187 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
   1188 		if (error)
   1189 			break;
   1190 		where += len;
   1191 		maxlen -= len;
   1192 
   1193 		/*
   1194 		 * ... then, copy from the beginning of message buffer to
   1195 		 * the write pointer.
   1196 		 */
   1197 		beg = 0;
   1198 		end = msgbufp->msg_bufx;
   1199 	}
   1200 
   1201 	sysctl_relock();
   1202 	return (error);
   1203 }
   1204 
   1205 /*
   1206  * sysctl helper routine for kern.defcorename. In the case of a new
   1207  * string being assigned, check that it's not a zero-length string.
   1208  * (XXX the check in -current doesn't work, but do we really care?)
   1209  */
   1210 static int
   1211 sysctl_kern_defcorename(SYSCTLFN_ARGS)
   1212 {
   1213 	int error;
   1214 	char *newcorename;
   1215 	struct sysctlnode node;
   1216 
   1217 	newcorename = PNBUF_GET();
   1218 	node = *rnode;
   1219 	node.sysctl_data = &newcorename[0];
   1220 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   1221 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1222 	if (error || newp == NULL) {
   1223 		goto done;
   1224 	}
   1225 
   1226 	/*
   1227 	 * when sysctl_lookup() deals with a string, it's guaranteed
   1228 	 * to come back nul terminated. So there.  :)
   1229 	 */
   1230 	if (strlen(newcorename) == 0) {
   1231 		error = EINVAL;
   1232 	} else {
   1233 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   1234 		error = 0;
   1235 	}
   1236 done:
   1237 	PNBUF_PUT(newcorename);
   1238 	return error;
   1239 }
   1240 
   1241 /*
   1242  * sysctl helper routine for kern.cp_time node. Adds up cpu time
   1243  * across all cpus.
   1244  */
   1245 static int
   1246 sysctl_kern_cptime(SYSCTLFN_ARGS)
   1247 {
   1248 	struct sysctlnode node = *rnode;
   1249 	uint64_t *cp_time = NULL;
   1250 	int error, n = ncpu, i;
   1251 	struct cpu_info *ci;
   1252 	CPU_INFO_ITERATOR cii;
   1253 
   1254 	/*
   1255 	 * if you specifically pass a buffer that is the size of the
   1256 	 * sum, or if you are probing for the size, you get the "sum"
   1257 	 * of cp_time (and the size thereof) across all processors.
   1258 	 *
   1259 	 * alternately, you can pass an additional mib number and get
   1260 	 * cp_time for that particular processor.
   1261 	 */
   1262 	switch (namelen) {
   1263 	case 0:
   1264 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
   1265 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
   1266 			n = -1; /* SUM */
   1267 		}
   1268 		else {
   1269 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
   1270 			n = -2; /* ALL */
   1271 		}
   1272 		break;
   1273 	case 1:
   1274 		if (name[0] < 0 || name[0] >= n)
   1275 			return (ENOENT); /* ENOSUCHPROCESSOR */
   1276 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
   1277 		n = name[0];
   1278 		/*
   1279 		 * adjust these so that sysctl_lookup() will be happy
   1280 		 */
   1281 		name++;
   1282 		namelen--;
   1283 		break;
   1284 	default:
   1285 		return (EINVAL);
   1286 	}
   1287 
   1288 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
   1289 	if (cp_time == NULL)
   1290 		return (ENOMEM);
   1291 	node.sysctl_data = cp_time;
   1292 	memset(cp_time, 0, node.sysctl_size);
   1293 
   1294 	for (CPU_INFO_FOREACH(cii, ci)) {
   1295 		if (n <= 0) {
   1296 			for (i = 0; i < CPUSTATES; i++) {
   1297 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
   1298 			}
   1299 		}
   1300 		/*
   1301 		 * if a specific processor was requested and we just
   1302 		 * did it, we're done here
   1303 		 */
   1304 		if (n == 0)
   1305 			break;
   1306 		/*
   1307 		 * if doing "all", skip to next cp_time set for next processor
   1308 		 */
   1309 		if (n == -2)
   1310 			cp_time += CPUSTATES;
   1311 		/*
   1312 		 * if we're doing a specific processor, we're one
   1313 		 * processor closer
   1314 		 */
   1315 		if (n > 0)
   1316 			n--;
   1317 	}
   1318 
   1319 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1320 	kmem_free(node.sysctl_data, node.sysctl_size);
   1321 	return (error);
   1322 }
   1323 
   1324 #if NPTY > 0
   1325 /*
   1326  * sysctl helper routine for kern.maxptys. Ensures that any new value
   1327  * is acceptable to the pty subsystem.
   1328  */
   1329 static int
   1330 sysctl_kern_maxptys(SYSCTLFN_ARGS)
   1331 {
   1332 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
   1333 	int error, xmax;
   1334 	struct sysctlnode node;
   1335 
   1336 	/* get current value of maxptys */
   1337 	xmax = pty_maxptys(0, 0);
   1338 
   1339 	node = *rnode;
   1340 	node.sysctl_data = &xmax;
   1341 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1342 	if (error || newp == NULL)
   1343 		return (error);
   1344 
   1345 	if (xmax != pty_maxptys(xmax, 1))
   1346 		return (EINVAL);
   1347 
   1348 	return (0);
   1349 }
   1350 #endif /* NPTY > 0 */
   1351 
   1352 /*
   1353  * sysctl helper routine for kern.sbmax. Basically just ensures that
   1354  * any new value is not too small.
   1355  */
   1356 static int
   1357 sysctl_kern_sbmax(SYSCTLFN_ARGS)
   1358 {
   1359 	int error, new_sbmax;
   1360 	struct sysctlnode node;
   1361 
   1362 	new_sbmax = sb_max;
   1363 	node = *rnode;
   1364 	node.sysctl_data = &new_sbmax;
   1365 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1366 	if (error || newp == NULL)
   1367 		return (error);
   1368 
   1369 	KERNEL_LOCK(1, NULL);
   1370 	error = sb_max_set(new_sbmax);
   1371 	KERNEL_UNLOCK_ONE(NULL);
   1372 
   1373 	return (error);
   1374 }
   1375 
   1376 /*
   1377  * sysctl helper routine for kern.urandom node. Picks a random number
   1378  * for you.
   1379  */
   1380 static int
   1381 sysctl_kern_urnd(SYSCTLFN_ARGS)
   1382 {
   1383 	int v, rv;
   1384 
   1385 	rv = cprng_strong(sysctl_prng, &v, sizeof(v), 0);
   1386 	if (rv == sizeof(v)) {
   1387 		struct sysctlnode node = *rnode;
   1388 		node.sysctl_data = &v;
   1389 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1390 	}
   1391 	else
   1392 		return (EIO);	/*XXX*/
   1393 }
   1394 
   1395 /*
   1396  * sysctl helper routine for kern.arandom node. Picks a random number
   1397  * for you.
   1398  */
   1399 static int
   1400 sysctl_kern_arnd(SYSCTLFN_ARGS)
   1401 {
   1402 	int error;
   1403 	void *v;
   1404 	struct sysctlnode node = *rnode;
   1405 
   1406 	if (*oldlenp == 0)
   1407 		return 0;
   1408 	/*
   1409 	 * This code used to allow sucking 8192 bytes at a time out
   1410 	 * of the kernel arc4random generator.  Evidently there is some
   1411 	 * very old OpenBSD application code that may try to do this.
   1412 	 *
   1413 	 * Note that this node is documented as type "INT" -- 4 or 8
   1414 	 * bytes, not 8192.
   1415 	 *
   1416 	 * We continue to support this abuse of the "len" pointer here
   1417 	 * but only 256 bytes at a time, as, anecdotally, the actual
   1418 	 * application use here was to generate RC4 keys in userspace.
   1419 	 *
   1420 	 * Support for such large requests will probably be removed
   1421 	 * entirely in the future.
   1422 	 */
   1423 	if (*oldlenp > 256)
   1424 		return E2BIG;
   1425 
   1426 	v = kmem_alloc(*oldlenp, KM_SLEEP);
   1427 	cprng_fast(v, *oldlenp);
   1428 	node.sysctl_data = v;
   1429 	node.sysctl_size = *oldlenp;
   1430 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1431 	kmem_free(v, *oldlenp);
   1432 	return error;
   1433 }
   1434 /*
   1435  * sysctl helper routine to do kern.lwp.* work.
   1436  */
   1437 static int
   1438 sysctl_kern_lwp(SYSCTLFN_ARGS)
   1439 {
   1440 	struct kinfo_lwp klwp;
   1441 	struct proc *p;
   1442 	struct lwp *l2, *l3;
   1443 	char *where, *dp;
   1444 	int pid, elem_size, elem_count;
   1445 	int buflen, needed, error;
   1446 	bool gotit;
   1447 
   1448 	if (namelen == 1 && name[0] == CTL_QUERY)
   1449 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
   1450 
   1451 	dp = where = oldp;
   1452 	buflen = where != NULL ? *oldlenp : 0;
   1453 	error = needed = 0;
   1454 
   1455 	if (newp != NULL || namelen != 3)
   1456 		return (EINVAL);
   1457 	pid = name[0];
   1458 	elem_size = name[1];
   1459 	elem_count = name[2];
   1460 
   1461 	sysctl_unlock();
   1462 	if (pid == -1) {
   1463 		mutex_enter(proc_lock);
   1464 		PROCLIST_FOREACH(p, &allproc) {
   1465 			/* Grab a hold on the process. */
   1466 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
   1467 				continue;
   1468 			}
   1469 			mutex_exit(proc_lock);
   1470 
   1471 			mutex_enter(p->p_lock);
   1472 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1473 				if (buflen >= elem_size && elem_count > 0) {
   1474 					lwp_lock(l2);
   1475 					fill_lwp(l2, &klwp);
   1476 					lwp_unlock(l2);
   1477 					mutex_exit(p->p_lock);
   1478 
   1479 					/*
   1480 					 * Copy out elem_size, but not
   1481 					 * larger than the size of a
   1482 					 * struct kinfo_proc2.
   1483 					 */
   1484 					error = dcopyout(l, &klwp, dp,
   1485 					    min(sizeof(klwp), elem_size));
   1486 					if (error) {
   1487 						rw_exit(&p->p_reflock);
   1488 						goto cleanup;
   1489 					}
   1490 					mutex_enter(p->p_lock);
   1491 					LIST_FOREACH(l3, &p->p_lwps,
   1492 					    l_sibling) {
   1493 						if (l2 == l3)
   1494 							break;
   1495 					}
   1496 					if (l3 == NULL) {
   1497 						mutex_exit(p->p_lock);
   1498 						rw_exit(&p->p_reflock);
   1499 						error = EAGAIN;
   1500 						goto cleanup;
   1501 					}
   1502 					dp += elem_size;
   1503 					buflen -= elem_size;
   1504 					elem_count--;
   1505 				}
   1506 				needed += elem_size;
   1507 			}
   1508 			mutex_exit(p->p_lock);
   1509 
   1510 			/* Drop reference to process. */
   1511 			mutex_enter(proc_lock);
   1512 			rw_exit(&p->p_reflock);
   1513 		}
   1514 		mutex_exit(proc_lock);
   1515 	} else {
   1516 		mutex_enter(proc_lock);
   1517 		p = proc_find(pid);
   1518 		if (p == NULL) {
   1519 			error = ESRCH;
   1520 			mutex_exit(proc_lock);
   1521 			goto cleanup;
   1522 		}
   1523 		/* Grab a hold on the process. */
   1524 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
   1525 		mutex_exit(proc_lock);
   1526 		if (!gotit) {
   1527 			error = ESRCH;
   1528 			goto cleanup;
   1529 		}
   1530 
   1531 		mutex_enter(p->p_lock);
   1532 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
   1533 			if (buflen >= elem_size && elem_count > 0) {
   1534 				lwp_lock(l2);
   1535 				fill_lwp(l2, &klwp);
   1536 				lwp_unlock(l2);
   1537 				mutex_exit(p->p_lock);
   1538 				/*
   1539 				 * Copy out elem_size, but not larger than
   1540 				 * the size of a struct kinfo_proc2.
   1541 				 */
   1542 				error = dcopyout(l, &klwp, dp,
   1543 				    min(sizeof(klwp), elem_size));
   1544 				if (error) {
   1545 					rw_exit(&p->p_reflock);
   1546 					goto cleanup;
   1547 				}
   1548 				mutex_enter(p->p_lock);
   1549 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
   1550 					if (l2 == l3)
   1551 						break;
   1552 				}
   1553 				if (l3 == NULL) {
   1554 					mutex_exit(p->p_lock);
   1555 					rw_exit(&p->p_reflock);
   1556 					error = EAGAIN;
   1557 					goto cleanup;
   1558 				}
   1559 				dp += elem_size;
   1560 				buflen -= elem_size;
   1561 				elem_count--;
   1562 			}
   1563 			needed += elem_size;
   1564 		}
   1565 		mutex_exit(p->p_lock);
   1566 
   1567 		/* Drop reference to process. */
   1568 		rw_exit(&p->p_reflock);
   1569 	}
   1570 
   1571 	if (where != NULL) {
   1572 		*oldlenp = dp - where;
   1573 		if (needed > *oldlenp) {
   1574 			sysctl_relock();
   1575 			return (ENOMEM);
   1576 		}
   1577 	} else {
   1578 		needed += KERN_LWPSLOP;
   1579 		*oldlenp = needed;
   1580 	}
   1581 	error = 0;
   1582  cleanup:
   1583 	sysctl_relock();
   1584 	return (error);
   1585 }
   1586 
   1587 /*
   1588  * sysctl helper routine for kern.forkfsleep node. Ensures that the
   1589  * given value is not too large or two small, and is at least one
   1590  * timer tick if not zero.
   1591  */
   1592 static int
   1593 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
   1594 {
   1595 	/* userland sees value in ms, internally is in ticks */
   1596 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
   1597 	int error, timo, lsleep;
   1598 	struct sysctlnode node;
   1599 
   1600 	lsleep = forkfsleep * 1000 / hz;
   1601 	node = *rnode;
   1602 	node.sysctl_data = &lsleep;
   1603 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1604 	if (error || newp == NULL)
   1605 		return (error);
   1606 
   1607 	/* refuse negative values, and overly 'long time' */
   1608 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
   1609 		return (EINVAL);
   1610 
   1611 	timo = mstohz(lsleep);
   1612 
   1613 	/* if the interval is >0 ms && <1 tick, use 1 tick */
   1614 	if (lsleep != 0 && timo == 0)
   1615 		forkfsleep = 1;
   1616 	else
   1617 		forkfsleep = timo;
   1618 
   1619 	return (0);
   1620 }
   1621 
   1622 /*
   1623  * sysctl helper routine for kern.root_partition
   1624  */
   1625 static int
   1626 sysctl_kern_root_partition(SYSCTLFN_ARGS)
   1627 {
   1628 	int rootpart = DISKPART(rootdev);
   1629 	struct sysctlnode node = *rnode;
   1630 
   1631 	node.sysctl_data = &rootpart;
   1632 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1633 }
   1634 
   1635 /*
   1636  * sysctl helper function for kern.drivers
   1637  */
   1638 static int
   1639 sysctl_kern_drivers(SYSCTLFN_ARGS)
   1640 {
   1641 	int error;
   1642 	size_t buflen;
   1643 	struct kinfo_drivers kd;
   1644 	char *start, *where;
   1645 	const char *dname;
   1646 	int i;
   1647 	extern struct devsw_conv *devsw_conv;
   1648 	extern int max_devsw_convs;
   1649 
   1650 	if (newp != NULL || namelen != 0)
   1651 		return (EINVAL);
   1652 
   1653 	start = where = oldp;
   1654 	buflen = *oldlenp;
   1655 	if (where == NULL) {
   1656 		*oldlenp = max_devsw_convs * sizeof kd;
   1657 		return 0;
   1658 	}
   1659 
   1660 	/*
   1661 	 * An array of kinfo_drivers structures
   1662 	 */
   1663 	error = 0;
   1664 	sysctl_unlock();
   1665 	mutex_enter(&device_lock);
   1666 	for (i = 0; i < max_devsw_convs; i++) {
   1667 		dname = devsw_conv[i].d_name;
   1668 		if (dname == NULL)
   1669 			continue;
   1670 		if (buflen < sizeof kd) {
   1671 			error = ENOMEM;
   1672 			break;
   1673 		}
   1674 		memset(&kd, 0, sizeof(kd));
   1675 		kd.d_bmajor = devsw_conv[i].d_bmajor;
   1676 		kd.d_cmajor = devsw_conv[i].d_cmajor;
   1677 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
   1678 		mutex_exit(&device_lock);
   1679 		error = dcopyout(l, &kd, where, sizeof kd);
   1680 		mutex_enter(&device_lock);
   1681 		if (error != 0)
   1682 			break;
   1683 		buflen -= sizeof kd;
   1684 		where += sizeof kd;
   1685 	}
   1686 	mutex_exit(&device_lock);
   1687 	sysctl_relock();
   1688 	*oldlenp = where - start;
   1689 	return error;
   1690 }
   1691 
   1692 static int
   1693 sysctl_security_setidcore(SYSCTLFN_ARGS)
   1694 {
   1695 	int newsize, error;
   1696 	struct sysctlnode node;
   1697 
   1698 	node = *rnode;
   1699 	node.sysctl_data = &newsize;
   1700 	newsize = *(int *)rnode->sysctl_data;
   1701 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1702 	if (error || newp == NULL)
   1703 		return error;
   1704 
   1705 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   1706 	    0, NULL, NULL, NULL))
   1707 		return (EPERM);
   1708 
   1709 	*(int *)rnode->sysctl_data = newsize;
   1710 
   1711 	return 0;
   1712 }
   1713 
   1714 static int
   1715 sysctl_security_setidcorename(SYSCTLFN_ARGS)
   1716 {
   1717 	int error;
   1718 	char *newsetidcorename;
   1719 	struct sysctlnode node;
   1720 
   1721 	newsetidcorename = PNBUF_GET();
   1722 	node = *rnode;
   1723 	node.sysctl_data = newsetidcorename;
   1724 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
   1725 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1726 	if (error || newp == NULL) {
   1727 		goto out;
   1728 	}
   1729 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
   1730 	    0, NULL, NULL, NULL)) {
   1731 		error = EPERM;
   1732 		goto out;
   1733 	}
   1734 	if (strlen(newsetidcorename) == 0) {
   1735 		error = EINVAL;
   1736 		goto out;
   1737 	}
   1738 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
   1739 out:
   1740 	PNBUF_PUT(newsetidcorename);
   1741 	return error;
   1742 }
   1743 
   1744 /*
   1745  * sysctl helper routine for kern.cp_id node. Maps cpus to their
   1746  * cpuids.
   1747  */
   1748 static int
   1749 sysctl_kern_cpid(SYSCTLFN_ARGS)
   1750 {
   1751 	struct sysctlnode node = *rnode;
   1752 	uint64_t *cp_id = NULL;
   1753 	int error, n = ncpu;
   1754 	struct cpu_info *ci;
   1755 	CPU_INFO_ITERATOR cii;
   1756 
   1757 	/*
   1758 	 * Here you may either retrieve a single cpu id or the whole
   1759 	 * set. The size you get back when probing depends on what
   1760 	 * you ask for.
   1761 	 */
   1762 	switch (namelen) {
   1763 	case 0:
   1764 		node.sysctl_size = n * sizeof(uint64_t);
   1765 		n = -2; /* ALL */
   1766 		break;
   1767 	case 1:
   1768 		if (name[0] < 0 || name[0] >= n)
   1769 			return (ENOENT); /* ENOSUCHPROCESSOR */
   1770 		node.sysctl_size = sizeof(uint64_t);
   1771 		n = name[0];
   1772 		/*
   1773 		 * adjust these so that sysctl_lookup() will be happy
   1774 		 */
   1775 		name++;
   1776 		namelen--;
   1777 		break;
   1778 	default:
   1779 		return (EINVAL);
   1780 	}
   1781 
   1782 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
   1783 	if (cp_id == NULL)
   1784 		return (ENOMEM);
   1785 	node.sysctl_data = cp_id;
   1786 	memset(cp_id, 0, node.sysctl_size);
   1787 
   1788 	for (CPU_INFO_FOREACH(cii, ci)) {
   1789 		if (n <= 0)
   1790 			cp_id[0] = cpu_index(ci);
   1791 		/*
   1792 		 * if a specific processor was requested and we just
   1793 		 * did it, we're done here
   1794 		 */
   1795 		if (n == 0)
   1796 			break;
   1797 		/*
   1798 		 * if doing "all", skip to next cp_id slot for next processor
   1799 		 */
   1800 		if (n == -2)
   1801 			cp_id++;
   1802 		/*
   1803 		 * if we're doing a specific processor, we're one
   1804 		 * processor closer
   1805 		 */
   1806 		if (n > 0)
   1807 			n--;
   1808 	}
   1809 
   1810 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1811 	kmem_free(node.sysctl_data, node.sysctl_size);
   1812 	return (error);
   1813 }
   1814 
   1815 /*
   1816  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
   1817  * calculate on the fly taking into account integer overflow and the
   1818  * current wired count.
   1819  */
   1820 static int
   1821 sysctl_hw_usermem(SYSCTLFN_ARGS)
   1822 {
   1823 	u_int ui;
   1824 	u_quad_t uq;
   1825 	struct sysctlnode node;
   1826 
   1827 	node = *rnode;
   1828 	switch (rnode->sysctl_num) {
   1829 	case HW_USERMEM:
   1830 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
   1831 			ui = UINT_MAX;
   1832 		else
   1833 			ui *= PAGE_SIZE;
   1834 		node.sysctl_data = &ui;
   1835 		break;
   1836 	case HW_USERMEM64:
   1837 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
   1838 		node.sysctl_data = &uq;
   1839 		break;
   1840 	default:
   1841 		return (EINVAL);
   1842 	}
   1843 
   1844 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1845 }
   1846 
   1847 /*
   1848  * sysctl helper routine for kern.cnmagic node. Pulls the old value
   1849  * out, encoded, and stuffs the new value in for decoding.
   1850  */
   1851 static int
   1852 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
   1853 {
   1854 	char magic[CNS_LEN];
   1855 	int error;
   1856 	struct sysctlnode node;
   1857 
   1858 	if (oldp)
   1859 		cn_get_magic(magic, CNS_LEN);
   1860 	node = *rnode;
   1861 	node.sysctl_data = &magic[0];
   1862 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
   1863 	if (error || newp == NULL)
   1864 		return (error);
   1865 
   1866 	return (cn_set_magic(magic));
   1867 }
   1868 
   1869 /*
   1870  * ********************************************************************
   1871  * section 3: public helper routines that are used for more than one
   1872  * node
   1873  * ********************************************************************
   1874  */
   1875 
   1876 /*
   1877  * sysctl helper routine for the kern.root_device node and some ports'
   1878  * machdep.root_device nodes.
   1879  */
   1880 int
   1881 sysctl_root_device(SYSCTLFN_ARGS)
   1882 {
   1883 	struct sysctlnode node;
   1884 
   1885 	node = *rnode;
   1886 	node.sysctl_data = root_device->dv_xname;
   1887 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
   1888 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1889 }
   1890 
   1891 /*
   1892  * sysctl helper routine for kern.consdev, dependent on the current
   1893  * state of the console. Also used for machdep.console_device on some
   1894  * ports.
   1895  */
   1896 int
   1897 sysctl_consdev(SYSCTLFN_ARGS)
   1898 {
   1899 	dev_t consdev;
   1900 	uint32_t oconsdev;
   1901 	struct sysctlnode node;
   1902 
   1903 	if (cn_tab != NULL)
   1904 		consdev = cn_tab->cn_dev;
   1905 	else
   1906 		consdev = NODEV;
   1907 	node = *rnode;
   1908 	switch (*oldlenp) {
   1909 	case sizeof(consdev):
   1910 		node.sysctl_data = &consdev;
   1911 		node.sysctl_size = sizeof(consdev);
   1912 		break;
   1913 	case sizeof(oconsdev):
   1914 		oconsdev = (uint32_t)consdev;
   1915 		node.sysctl_data = &oconsdev;
   1916 		node.sysctl_size = sizeof(oconsdev);
   1917 		break;
   1918 	default:
   1919 		return EINVAL;
   1920 	}
   1921 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
   1922 }
   1923 
   1924 /*
   1925  * ********************************************************************
   1926  * section 4: support for some helpers
   1927  * ********************************************************************
   1928  */
   1929 
   1930 
   1931 /*
   1932  * Fill in a kinfo_lwp structure for the specified lwp.
   1933  */
   1934 static void
   1935 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
   1936 {
   1937 	struct proc *p = l->l_proc;
   1938 	struct timeval tv;
   1939 
   1940 	KASSERT(lwp_locked(l, NULL));
   1941 
   1942 	memset(kl, 0, sizeof(*kl));
   1943 
   1944 	kl->l_forw = 0;
   1945 	kl->l_back = 0;
   1946 	kl->l_laddr = PTRTOUINT64(l);
   1947 	kl->l_addr = PTRTOUINT64(l->l_addr);
   1948 	kl->l_stat = l->l_stat;
   1949 	kl->l_lid = l->l_lid;
   1950 	kl->l_flag = L_INMEM;
   1951 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
   1952 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
   1953 
   1954 	kl->l_swtime = l->l_swtime;
   1955 	kl->l_slptime = l->l_slptime;
   1956 	if (l->l_stat == LSONPROC)
   1957 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
   1958 	else
   1959 		kl->l_schedflags = 0;
   1960 	kl->l_priority = lwp_eprio(l);
   1961 	kl->l_usrpri = l->l_priority;
   1962 	if (l->l_wchan)
   1963 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
   1964 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
   1965 	kl->l_cpuid = cpu_index(l->l_cpu);
   1966 	bintime2timeval(&l->l_rtime, &tv);
   1967 	kl->l_rtime_sec = tv.tv_sec;
   1968 	kl->l_rtime_usec = tv.tv_usec;
   1969 	kl->l_cpticks = l->l_cpticks;
   1970 	kl->l_pctcpu = l->l_pctcpu;
   1971 	kl->l_pid = p->p_pid;
   1972 	if (l->l_name == NULL)
   1973 		kl->l_name[0] = '\0';
   1974 	else
   1975 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
   1976 }
   1977