xref: /netbsd-src/sys/kern/init_sysctl.c (revision 1897181a7231d5fc7ab48994d1447fcbc4e13a49)
1 /*	$NetBSD: init_sysctl.c,v 1.186 2011/12/17 20:05:39 tls 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 1.186 2011/12/17 20:05:39 tls Exp $");
34 
35 #include "opt_sysv.h"
36 #include "opt_compat_netbsd.h"
37 #include "opt_modular.h"
38 #include "opt_sa.h"
39 #include "opt_posix.h"
40 #include "pty.h"
41 
42 #include <sys/types.h>
43 #include <sys/param.h>
44 #include <sys/sysctl.h>
45 #include <sys/cpu.h>
46 #include <sys/errno.h>
47 #include <sys/systm.h>
48 #include <sys/kernel.h>
49 #include <sys/unistd.h>
50 #include <sys/disklabel.h>
51 #include <sys/cprng.h>
52 #include <sys/vnode.h>
53 #include <sys/mount.h>
54 #include <sys/namei.h>
55 #include <sys/msgbuf.h>
56 #include <dev/cons.h>
57 #include <sys/socketvar.h>
58 #include <sys/file.h>
59 #include <sys/filedesc.h>
60 #include <sys/tty.h>
61 #include <sys/kmem.h>
62 #include <sys/resource.h>
63 #include <sys/resourcevar.h>
64 #include <sys/exec.h>
65 #include <sys/conf.h>
66 #include <sys/device.h>
67 #include <sys/stat.h>
68 #include <sys/kauth.h>
69 #include <sys/ktrace.h>
70 #include <sys/ksem.h>
71 
72 #ifdef COMPAT_50
73 #include <compat/sys/time.h>
74 #endif
75 
76 #ifdef KERN_SA
77 #include <sys/sa.h>
78 #endif
79 
80 #include <sys/cpu.h>
81 
82 #if defined(MODULAR) || defined(P1003_1B_SEMAPHORE)
83 int posix_semaphores = 200112;
84 #else
85 int posix_semaphores;
86 #endif
87 
88 int security_setidcore_dump;
89 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
90 uid_t security_setidcore_owner = 0;
91 gid_t security_setidcore_group = 0;
92 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
93 
94 static const u_int sysctl_flagmap[] = {
95 	PK_ADVLOCK, P_ADVLOCK,
96 	PK_EXEC, P_EXEC,
97 	PK_NOCLDWAIT, P_NOCLDWAIT,
98 	PK_32, P_32,
99 	PK_CLDSIGIGN, P_CLDSIGIGN,
100 	PK_SUGID, P_SUGID,
101 	0
102 };
103 
104 static const u_int sysctl_sflagmap[] = {
105 	PS_NOCLDSTOP, P_NOCLDSTOP,
106 	PS_WEXIT, P_WEXIT,
107 	PS_STOPFORK, P_STOPFORK,
108 	PS_STOPEXEC, P_STOPEXEC,
109 	PS_STOPEXIT, P_STOPEXIT,
110 	0
111 };
112 
113 static const u_int sysctl_slflagmap[] = {
114 	PSL_TRACED, P_TRACED,
115 	PSL_FSTRACE, P_FSTRACE,
116 	PSL_CHTRACED, P_CHTRACED,
117 	PSL_SYSCALL, P_SYSCALL,
118 	0
119 };
120 
121 static const u_int sysctl_lflagmap[] = {
122 	PL_CONTROLT, P_CONTROLT,
123 	PL_PPWAIT, P_PPWAIT,
124 	0
125 };
126 
127 static const u_int sysctl_stflagmap[] = {
128 	PST_PROFIL, P_PROFIL,
129 	0
130 
131 };
132 
133 static const u_int sysctl_lwpprflagmap[] = {
134 	LPR_DETACHED, L_DETACHED,
135 	0
136 };
137 
138 /*
139  * try over estimating by 5 procs/lwps
140  */
141 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
142 
143 static int dcopyout(struct lwp *, const void *, void *, size_t);
144 
145 static int
146 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
147 {
148 	int error;
149 
150 	error = copyout(kaddr, uaddr, len);
151 	ktrmibio(-1, UIO_READ, uaddr, len, error);
152 
153 	return error;
154 }
155 
156 #ifdef DIAGNOSTIC
157 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
158 #endif
159 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
160 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
161 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
162 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
163 static int sysctl_setlen(SYSCTLFN_PROTO);
164 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
165 static int sysctl_msgbuf(SYSCTLFN_PROTO);
166 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
167 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
168 #if NPTY > 0
169 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
170 #endif /* NPTY > 0 */
171 static int sysctl_kern_sbmax(SYSCTLFN_PROTO);
172 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
173 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
174 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
175 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
176 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
177 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
178 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
179 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
180 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
181 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
182 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
183 
184 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
185 
186 /*
187  * ********************************************************************
188  * section 1: setup routines
189  * ********************************************************************
190  * These functions are stuffed into a link set for sysctl setup
191  * functions. They're never called or referenced from anywhere else.
192  * ********************************************************************
193  */
194 
195 /*
196  * this setup routine is a replacement for kern_sysctl()
197  */
198 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
199 {
200 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
201 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
202 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
203 	const struct sysctlnode *rnode;
204 
205 	sysctl_createv(clog, 0, NULL, NULL,
206 		       CTLFLAG_PERMANENT,
207 		       CTLTYPE_NODE, "kern", NULL,
208 		       NULL, 0, NULL, 0,
209 		       CTL_KERN, CTL_EOL);
210 
211 	sysctl_createv(clog, 0, NULL, NULL,
212 		       CTLFLAG_PERMANENT,
213 		       CTLTYPE_STRING, "ostype",
214 		       SYSCTL_DESCR("Operating system type"),
215 		       NULL, 0, __UNCONST(&ostype), 0,
216 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
217 	sysctl_createv(clog, 0, NULL, NULL,
218 		       CTLFLAG_PERMANENT,
219 		       CTLTYPE_STRING, "osrelease",
220 		       SYSCTL_DESCR("Operating system release"),
221 		       NULL, 0, __UNCONST(&osrelease), 0,
222 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
223 	sysctl_createv(clog, 0, NULL, NULL,
224 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
225 		       CTLTYPE_INT, "osrevision",
226 		       SYSCTL_DESCR("Operating system revision"),
227 		       NULL, __NetBSD_Version__, NULL, 0,
228 		       CTL_KERN, KERN_OSREV, CTL_EOL);
229 	sysctl_createv(clog, 0, NULL, NULL,
230 		       CTLFLAG_PERMANENT,
231 		       CTLTYPE_STRING, "version",
232 		       SYSCTL_DESCR("Kernel version"),
233 		       NULL, 0, __UNCONST(&version), 0,
234 		       CTL_KERN, KERN_VERSION, CTL_EOL);
235 	sysctl_createv(clog, 0, NULL, NULL,
236 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
237 		       CTLTYPE_INT, "maxvnodes",
238 		       SYSCTL_DESCR("Maximum number of vnodes"),
239 		       sysctl_kern_maxvnodes, 0, NULL, 0,
240 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
241 	sysctl_createv(clog, 0, NULL, NULL,
242 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
243 		       CTLTYPE_INT, "maxproc",
244 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
245 		       sysctl_kern_maxproc, 0, NULL, 0,
246 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
247 	sysctl_createv(clog, 0, NULL, NULL,
248 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
249 		       CTLTYPE_INT, "maxfiles",
250 		       SYSCTL_DESCR("Maximum number of open files"),
251 		       NULL, 0, &maxfiles, 0,
252 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
253 	sysctl_createv(clog, 0, NULL, NULL,
254 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
255 		       CTLTYPE_INT, "argmax",
256 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
257 				    "execve(2)"),
258 		       NULL, ARG_MAX, NULL, 0,
259 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
260 	sysctl_createv(clog, 0, NULL, NULL,
261 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
262 		       CTLTYPE_STRING, "hostname",
263 		       SYSCTL_DESCR("System hostname"),
264 		       sysctl_setlen, 0, &hostname, MAXHOSTNAMELEN,
265 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
266 	sysctl_createv(clog, 0, NULL, NULL,
267 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
268 		       CTLTYPE_INT, "hostid",
269 		       SYSCTL_DESCR("System host ID number"),
270 		       sysctl_kern_hostid, 0, NULL, 0,
271 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
272 	sysctl_createv(clog, 0, NULL, NULL,
273 		       CTLFLAG_PERMANENT,
274 		       CTLTYPE_STRUCT, "clockrate",
275 		       SYSCTL_DESCR("Kernel clock rates"),
276 		       sysctl_kern_clockrate, 0, NULL,
277 		       sizeof(struct clockinfo),
278 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
279 	sysctl_createv(clog, 0, NULL, NULL,
280 		       CTLFLAG_PERMANENT,
281 		       CTLTYPE_INT, "hardclock_ticks",
282 		       SYSCTL_DESCR("Number of hardclock ticks"),
283 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
284 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
285 	sysctl_createv(clog, 0, NULL, NULL,
286 		       CTLFLAG_PERMANENT,
287 		       CTLTYPE_STRUCT, "vnode",
288 		       SYSCTL_DESCR("System vnode table"),
289 		       sysctl_kern_vnode, 0, NULL, 0,
290 		       CTL_KERN, KERN_VNODE, CTL_EOL);
291 #ifndef GPROF
292 	sysctl_createv(clog, 0, NULL, NULL,
293 		       CTLFLAG_PERMANENT,
294 		       CTLTYPE_NODE, "profiling",
295 		       SYSCTL_DESCR("Profiling information (not available)"),
296 		       sysctl_notavail, 0, NULL, 0,
297 		       CTL_KERN, KERN_PROF, CTL_EOL);
298 #endif
299 	sysctl_createv(clog, 0, NULL, NULL,
300 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
301 		       CTLTYPE_INT, "posix1version",
302 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
303 				    "with which the operating system attempts "
304 				    "to comply"),
305 		       NULL, _POSIX_VERSION, NULL, 0,
306 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
307 	sysctl_createv(clog, 0, NULL, NULL,
308 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
309 		       CTLTYPE_INT, "ngroups",
310 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
311 		       NULL, NGROUPS_MAX, NULL, 0,
312 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
313 	sysctl_createv(clog, 0, NULL, NULL,
314 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
315 		       CTLTYPE_INT, "job_control",
316 		       SYSCTL_DESCR("Whether job control is available"),
317 		       NULL, 1, NULL, 0,
318 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
319 	sysctl_createv(clog, 0, NULL, NULL,
320 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
321 		       CTLTYPE_INT, "saved_ids",
322 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
323 				    "available"), NULL,
324 #ifdef _POSIX_SAVED_IDS
325 		       1,
326 #else /* _POSIX_SAVED_IDS */
327 		       0,
328 #endif /* _POSIX_SAVED_IDS */
329 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
330 	sysctl_createv(clog, 0, NULL, NULL,
331 		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
332 		       CTLTYPE_INT, "boothowto",
333 		       SYSCTL_DESCR("Flags from boot loader"),
334 		       NULL, 0, &boothowto, sizeof(boothowto),
335 		       CTL_KERN, CTL_CREATE, CTL_EOL);
336 	sysctl_createv(clog, 0, NULL, NULL,
337 		       CTLFLAG_PERMANENT,
338 		       CTLTYPE_STRUCT, "boottime",
339 		       SYSCTL_DESCR("System boot time"),
340 		       NULL, 0, &boottime, sizeof(boottime),
341 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
342 #ifdef COMPAT_50
343 	{
344 		extern struct timeval50 boottime50;
345 		sysctl_createv(clog, 0, NULL, NULL,
346 			       CTLFLAG_PERMANENT,
347 			       CTLTYPE_STRUCT, "oboottime",
348 			       SYSCTL_DESCR("System boot time"),
349 			       NULL, 0, &boottime50, sizeof(boottime50),
350 			       CTL_KERN, KERN_OBOOTTIME, CTL_EOL);
351 	}
352 #endif
353 	sysctl_createv(clog, 0, NULL, NULL,
354 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
355 		       CTLTYPE_STRING, "domainname",
356 		       SYSCTL_DESCR("YP domain name"),
357 		       sysctl_setlen, 0, &domainname, MAXHOSTNAMELEN,
358 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
359 	sysctl_createv(clog, 0, NULL, NULL,
360 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
361 		       CTLTYPE_INT, "maxpartitions",
362 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
363 				    "disk"),
364 		       NULL, MAXPARTITIONS, NULL, 0,
365 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
366 	sysctl_createv(clog, 0, NULL, NULL,
367 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
368 		       CTLTYPE_INT, "rawpartition",
369 		       SYSCTL_DESCR("Raw partition of a disk"),
370 		       NULL, RAW_PART, NULL, 0,
371 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
372 	sysctl_createv(clog, 0, NULL, NULL,
373 		       CTLFLAG_PERMANENT,
374 		       CTLTYPE_STRUCT, "timex", NULL,
375 		       sysctl_notavail, 0, NULL, 0,
376 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
377 	sysctl_createv(clog, 0, NULL, NULL,
378 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
379 		       CTLTYPE_INT, "rtc_offset",
380 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
381 				    "minutes"),
382 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
383 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
384 	sysctl_createv(clog, 0, NULL, NULL,
385 		       CTLFLAG_PERMANENT,
386 		       CTLTYPE_STRING, "root_device",
387 		       SYSCTL_DESCR("Name of the root device"),
388 		       sysctl_root_device, 0, NULL, 0,
389 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
390 	sysctl_createv(clog, 0, NULL, NULL,
391 		       CTLFLAG_PERMANENT,
392 		       CTLTYPE_INT, "msgbufsize",
393 		       SYSCTL_DESCR("Size of the kernel message buffer"),
394 		       sysctl_msgbuf, 0, NULL, 0,
395 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
396 	sysctl_createv(clog, 0, NULL, NULL,
397 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
398 		       CTLTYPE_INT, "fsync",
399 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
400 				    "Synchronization Option is available on "
401 				    "this system"),
402 		       NULL, 1, NULL, 0,
403 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
404 	sysctl_createv(clog, 0, NULL, NULL,
405 		       CTLFLAG_PERMANENT,
406 		       CTLTYPE_NODE, "ipc",
407 		       SYSCTL_DESCR("SysV IPC options"),
408 		       NULL, 0, NULL, 0,
409 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
410 	sysctl_createv(clog, 0, NULL, NULL,
411 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
412 		       CTLTYPE_INT, "sysvmsg",
413 		       SYSCTL_DESCR("System V style message support available"),
414 		       NULL,
415 #ifdef SYSVMSG
416 		       1,
417 #else /* SYSVMSG */
418 		       0,
419 #endif /* SYSVMSG */
420 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
421 	sysctl_createv(clog, 0, NULL, NULL,
422 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
423 		       CTLTYPE_INT, "sysvsem",
424 		       SYSCTL_DESCR("System V style semaphore support "
425 				    "available"), NULL,
426 #ifdef SYSVSEM
427 		       1,
428 #else /* SYSVSEM */
429 		       0,
430 #endif /* SYSVSEM */
431 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
432 	sysctl_createv(clog, 0, NULL, NULL,
433 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
434 		       CTLTYPE_INT, "sysvshm",
435 		       SYSCTL_DESCR("System V style shared memory support "
436 				    "available"), NULL,
437 #ifdef SYSVSHM
438 		       1,
439 #else /* SYSVSHM */
440 		       0,
441 #endif /* SYSVSHM */
442 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
443 	sysctl_createv(clog, 0, NULL, NULL,
444 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
445 		       CTLTYPE_INT, "synchronized_io",
446 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
447 				    "I/O Option is available on this system"),
448 		       NULL, 1, NULL, 0,
449 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
450 	sysctl_createv(clog, 0, NULL, NULL,
451 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
452 		       CTLTYPE_INT, "iov_max",
453 		       SYSCTL_DESCR("Maximum number of iovec structures per "
454 				    "process"),
455 		       NULL, IOV_MAX, NULL, 0,
456 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
457 	sysctl_createv(clog, 0, NULL, NULL,
458 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
459 		       CTLTYPE_INT, "mapped_files",
460 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
461 				    "Files Option is available on this system"),
462 		       NULL, 1, NULL, 0,
463 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
464 	sysctl_createv(clog, 0, NULL, NULL,
465 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
466 		       CTLTYPE_INT, "memlock",
467 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
468 				    "Locking Option is available on this "
469 				    "system"),
470 		       NULL, 1, NULL, 0,
471 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
472 	sysctl_createv(clog, 0, NULL, NULL,
473 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
474 		       CTLTYPE_INT, "memlock_range",
475 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
476 				    "Locking Option is available on this "
477 				    "system"),
478 		       NULL, 1, NULL, 0,
479 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
480 	sysctl_createv(clog, 0, NULL, NULL,
481 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
482 		       CTLTYPE_INT, "memory_protection",
483 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
484 				    "Protection Option is available on this "
485 				    "system"),
486 		       NULL, 1, NULL, 0,
487 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
488 	sysctl_createv(clog, 0, NULL, NULL,
489 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
490 		       CTLTYPE_INT, "login_name_max",
491 		       SYSCTL_DESCR("Maximum login name length"),
492 		       NULL, LOGIN_NAME_MAX, NULL, 0,
493 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
494 	sysctl_createv(clog, 0, NULL, NULL,
495 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
496 		       CTLTYPE_STRING, "defcorename",
497 		       SYSCTL_DESCR("Default core file name"),
498 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
499 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
500 	sysctl_createv(clog, 0, NULL, NULL,
501 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
502 		       CTLTYPE_INT, "logsigexit",
503 		       SYSCTL_DESCR("Log process exit when caused by signals"),
504 		       NULL, 0, &kern_logsigexit, 0,
505 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
506 	sysctl_createv(clog, 0, NULL, NULL,
507 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
508 		       CTLTYPE_INT, "fscale",
509 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
510 		       NULL, FSCALE, NULL, 0,
511 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
512 	sysctl_createv(clog, 0, NULL, NULL,
513 		       CTLFLAG_PERMANENT,
514 		       CTLTYPE_INT, "ccpu",
515 		       SYSCTL_DESCR("Scheduler exponential decay value"),
516 		       NULL, 0, &ccpu, 0,
517 		       CTL_KERN, KERN_CCPU, CTL_EOL);
518 	sysctl_createv(clog, 0, NULL, NULL,
519 		       CTLFLAG_PERMANENT,
520 		       CTLTYPE_STRUCT, "cp_time",
521 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
522 		       sysctl_kern_cptime, 0, NULL, 0,
523 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
524 	sysctl_createv(clog, 0, NULL, NULL,
525 		       CTLFLAG_PERMANENT,
526 		       CTLTYPE_INT, "msgbuf",
527 		       SYSCTL_DESCR("Kernel message buffer"),
528 		       sysctl_msgbuf, 0, NULL, 0,
529 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
530 	sysctl_createv(clog, 0, NULL, NULL,
531 		       CTLFLAG_PERMANENT,
532 		       CTLTYPE_STRUCT, "consdev",
533 		       SYSCTL_DESCR("Console device"),
534 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
535 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
536 #if NPTY > 0
537 	sysctl_createv(clog, 0, NULL, NULL,
538 		       CTLFLAG_PERMANENT,
539 		       CTLTYPE_INT, "maxptys",
540 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
541 		       sysctl_kern_maxptys, 0, NULL, 0,
542 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
543 #endif /* NPTY > 0 */
544 	sysctl_createv(clog, 0, NULL, NULL,
545 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
546 		       CTLTYPE_INT, "maxphys",
547 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
548 		       NULL, MAXPHYS, NULL, 0,
549 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
550 	sysctl_createv(clog, 0, NULL, NULL,
551 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
552 		       CTLTYPE_INT, "sbmax",
553 		       SYSCTL_DESCR("Maximum socket buffer size"),
554 		       sysctl_kern_sbmax, 0, NULL, 0,
555 		       CTL_KERN, KERN_SBMAX, CTL_EOL);
556 	sysctl_createv(clog, 0, NULL, NULL,
557 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
558 		       CTLTYPE_INT, "monotonic_clock",
559 		       SYSCTL_DESCR("Implementation version of the POSIX "
560 				    "1003.1b Monotonic Clock Option"),
561 		       /* XXX _POSIX_VERSION */
562 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
563 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
564 	sysctl_createv(clog, 0, NULL, NULL,
565 		       CTLFLAG_PERMANENT,
566 		       CTLTYPE_INT, "urandom",
567 		       SYSCTL_DESCR("Random integer value"),
568 		       sysctl_kern_urnd, 0, NULL, 0,
569 		       CTL_KERN, KERN_URND, CTL_EOL);
570 	sysctl_createv(clog, 0, NULL, NULL,
571 		       CTLFLAG_PERMANENT,
572 		       CTLTYPE_INT, "arandom",
573 		       SYSCTL_DESCR("n bytes of random data"),
574 		       sysctl_kern_arnd, 0, NULL, 0,
575 		       CTL_KERN, KERN_ARND, CTL_EOL);
576 	sysctl_createv(clog, 0, NULL, NULL,
577 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
578 		       CTLTYPE_INT, "labelsector",
579 		       SYSCTL_DESCR("Sector number containing the disklabel"),
580 		       NULL, LABELSECTOR, NULL, 0,
581 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
582 	sysctl_createv(clog, 0, NULL, NULL,
583 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
584 		       CTLTYPE_INT, "labeloffset",
585 		       SYSCTL_DESCR("Offset of the disklabel within the "
586 				    "sector"),
587 		       NULL, LABELOFFSET, NULL, 0,
588 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
589 	sysctl_createv(clog, 0, NULL, NULL,
590 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
591 		       CTLTYPE_INT, "labelusesmbr",
592 		       SYSCTL_DESCR("disklabel is inside MBR partition"),
593 		       NULL, LABELUSESMBR, NULL, 0,
594 		       CTL_KERN, CTL_CREATE, CTL_EOL);
595 	sysctl_createv(clog, 0, NULL, NULL,
596 		       CTLFLAG_PERMANENT,
597 		       CTLTYPE_NODE, "lwp",
598 		       SYSCTL_DESCR("System-wide LWP information"),
599 		       sysctl_kern_lwp, 0, NULL, 0,
600 		       CTL_KERN, KERN_LWP, CTL_EOL);
601 	sysctl_createv(clog, 0, NULL, NULL,
602 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
603 		       CTLTYPE_INT, "forkfsleep",
604 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
605 				    "to process limits"),
606 		       sysctl_kern_forkfsleep, 0, NULL, 0,
607 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
608 	sysctl_createv(clog, 0, NULL, NULL,
609 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
610 		       CTLTYPE_INT, "posix_threads",
611 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
612 				    "Threads option to which the system "
613 				    "attempts to conform"),
614 		       /* XXX _POSIX_VERSION */
615 		       NULL, _POSIX_THREADS, NULL, 0,
616 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
617 	sysctl_createv(clog, 0, NULL, NULL,
618 		       CTLFLAG_PERMANENT,
619 		       CTLTYPE_INT, "posix_semaphores",
620 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
621 				    "Semaphores option to which the system "
622 				    "attempts to conform"), NULL,
623 		       0, &posix_semaphores,
624 		       0, CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
625 	sysctl_createv(clog, 0, NULL, NULL,
626 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
627 		       CTLTYPE_INT, "posix_barriers",
628 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
629 				    "Barriers option to which the system "
630 				    "attempts to conform"),
631 		       /* XXX _POSIX_VERSION */
632 		       NULL, _POSIX_BARRIERS, NULL, 0,
633 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
634 	sysctl_createv(clog, 0, NULL, NULL,
635 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
636 		       CTLTYPE_INT, "posix_timers",
637 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
638 				    "Timers option to which the system "
639 				    "attempts to conform"),
640 		       /* XXX _POSIX_VERSION */
641 		       NULL, _POSIX_TIMERS, NULL, 0,
642 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
643 	sysctl_createv(clog, 0, NULL, NULL,
644 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
645 		       CTLTYPE_INT, "posix_spin_locks",
646 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
647 				    "Locks option to which the system attempts "
648 				    "to conform"),
649 		       /* XXX _POSIX_VERSION */
650 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
651 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
652 	sysctl_createv(clog, 0, NULL, NULL,
653 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
654 		       CTLTYPE_INT, "posix_reader_writer_locks",
655 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
656 				    "Read-Write Locks option to which the "
657 				    "system attempts to conform"),
658 		       /* XXX _POSIX_VERSION */
659 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
660 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
661 	sysctl_createv(clog, 0, NULL, NULL,
662 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
663 		       CTLTYPE_INT, "dump_on_panic",
664 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
665 		       NULL, 0, &dumponpanic, 0,
666 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
667 #ifdef DIAGNOSTIC
668 	sysctl_createv(clog, 0, NULL, NULL,
669 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
670 		       CTLTYPE_INT, "panic_now",
671 		       SYSCTL_DESCR("Trigger a panic"),
672 		       sysctl_kern_trigger_panic, 0, NULL, 0,
673 		       CTL_KERN, CTL_CREATE, CTL_EOL);
674 #endif
675 	sysctl_createv(clog, 0, NULL, NULL,
676 		       CTLFLAG_PERMANENT,
677 		       CTLTYPE_INT, "root_partition",
678 		       SYSCTL_DESCR("Root partition on the root device"),
679 		       sysctl_kern_root_partition, 0, NULL, 0,
680 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
681 	sysctl_createv(clog, 0, NULL, NULL,
682 		       CTLFLAG_PERMANENT,
683 		       CTLTYPE_STRUCT, "drivers",
684 		       SYSCTL_DESCR("List of all drivers with block and "
685 				    "character device numbers"),
686 		       sysctl_kern_drivers, 0, NULL, 0,
687 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
688 	sysctl_createv(clog, 0, NULL, NULL,
689 		       CTLFLAG_PERMANENT,
690 		       CTLTYPE_STRUCT, "cp_id",
691 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
692 		       sysctl_kern_cpid, 0, NULL, 0,
693 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
694 	sysctl_createv(clog, 0, NULL, &rnode,
695 		       CTLFLAG_PERMANENT,
696 		       CTLTYPE_NODE, "coredump",
697 		       SYSCTL_DESCR("Coredump settings."),
698 		       NULL, 0, NULL, 0,
699 		       CTL_KERN, CTL_CREATE, CTL_EOL);
700 	sysctl_createv(clog, 0, &rnode, &rnode,
701 		       CTLFLAG_PERMANENT,
702 		       CTLTYPE_NODE, "setid",
703 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
704 		       NULL, 0, NULL, 0,
705 		       CTL_CREATE, CTL_EOL);
706 	sysctl_createv(clog, 0, &rnode, NULL,
707 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
708 		       CTLTYPE_INT, "dump",
709 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
710 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
711 		       sizeof(security_setidcore_dump),
712 		       CTL_CREATE, CTL_EOL);
713 	sysctl_createv(clog, 0, &rnode, NULL,
714 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
715 		       CTLTYPE_STRING, "path",
716 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
717 		       sysctl_security_setidcorename, 0,
718 		       &security_setidcore_path,
719 		       sizeof(security_setidcore_path),
720 		       CTL_CREATE, CTL_EOL);
721 	sysctl_createv(clog, 0, &rnode, NULL,
722 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
723 		       CTLTYPE_INT, "owner",
724 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
725 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
726 		       0,
727 		       CTL_CREATE, CTL_EOL);
728 	sysctl_createv(clog, 0, &rnode, NULL,
729 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
730 		       CTLTYPE_INT, "group",
731 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
732 		       sysctl_security_setidcore, 0, &security_setidcore_group,
733 		       0,
734 		       CTL_CREATE, CTL_EOL);
735 	sysctl_createv(clog, 0, &rnode, NULL,
736 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
737 		       CTLTYPE_INT, "mode",
738 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
739 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
740 		       0,
741 		       CTL_CREATE, CTL_EOL);
742 	sysctl_createv(clog, 0, NULL, NULL,
743 #ifndef KERN_SA
744 		       CTLFLAG_IMMEDIATE|
745 #endif
746 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
747 		       CTLTYPE_INT, "no_sa_support",
748 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
749 		       "it doesn't"),
750 		       NULL,
751 #ifndef KERN_SA
752 		       1, NULL,
753 #else
754 		       0, &sa_system_disabled,
755 #endif
756 		       0,
757 		       CTL_KERN, CTL_CREATE, CTL_EOL);
758 	/* kern.posix. */
759 	sysctl_createv(clog, 0, NULL, &rnode,
760 			CTLFLAG_PERMANENT,
761 			CTLTYPE_NODE, "posix",
762 			SYSCTL_DESCR("POSIX options"),
763 			NULL, 0, NULL, 0,
764 			CTL_KERN, CTL_CREATE, CTL_EOL);
765 	sysctl_createv(clog, 0, &rnode, NULL,
766 			CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
767 			CTLTYPE_INT, "semmax",
768 			SYSCTL_DESCR("Maximal number of semaphores"),
769 			NULL, 0, &ksem_max, 0,
770 			CTL_CREATE, CTL_EOL);
771 }
772 
773 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
774 {
775 	u_int u;
776 	u_quad_t q;
777 
778 	sysctl_createv(clog, 0, NULL, NULL,
779 		       CTLFLAG_PERMANENT,
780 		       CTLTYPE_NODE, "hw", NULL,
781 		       NULL, 0, NULL, 0,
782 		       CTL_HW, CTL_EOL);
783 
784 	sysctl_createv(clog, 0, NULL, NULL,
785 		       CTLFLAG_PERMANENT,
786 		       CTLTYPE_STRING, "machine",
787 		       SYSCTL_DESCR("Machine class"),
788 		       NULL, 0, machine, 0,
789 		       CTL_HW, HW_MACHINE, CTL_EOL);
790 	sysctl_createv(clog, 0, NULL, NULL,
791 		       CTLFLAG_PERMANENT,
792 		       CTLTYPE_STRING, "model",
793 		       SYSCTL_DESCR("Machine model"),
794 		       NULL, 0, cpu_model, 0,
795 		       CTL_HW, HW_MODEL, CTL_EOL);
796 	sysctl_createv(clog, 0, NULL, NULL,
797 		       CTLFLAG_PERMANENT,
798 		       CTLTYPE_INT, "ncpu",
799 		       SYSCTL_DESCR("Number of CPUs configured"),
800 		       NULL, 0, &ncpu, 0,
801 		       CTL_HW, HW_NCPU, CTL_EOL);
802 	sysctl_createv(clog, 0, NULL, NULL,
803 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
804 		       CTLTYPE_INT, "byteorder",
805 		       SYSCTL_DESCR("System byte order"),
806 		       NULL, BYTE_ORDER, NULL, 0,
807 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
808 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
809 		UINT_MAX : physmem * PAGE_SIZE;
810 	sysctl_createv(clog, 0, NULL, NULL,
811 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
812 		       CTLTYPE_INT, "physmem",
813 		       SYSCTL_DESCR("Bytes of physical memory"),
814 		       NULL, u, NULL, 0,
815 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
816 	sysctl_createv(clog, 0, NULL, NULL,
817 		       CTLFLAG_PERMANENT,
818 		       CTLTYPE_INT, "usermem",
819 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
820 		       sysctl_hw_usermem, 0, NULL, 0,
821 		       CTL_HW, HW_USERMEM, CTL_EOL);
822 	sysctl_createv(clog, 0, NULL, NULL,
823 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
824 		       CTLTYPE_INT, "pagesize",
825 		       SYSCTL_DESCR("Software page size"),
826 		       NULL, PAGE_SIZE, NULL, 0,
827 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
828 	sysctl_createv(clog, 0, NULL, NULL,
829 		       CTLFLAG_PERMANENT,
830 		       CTLTYPE_STRING, "machine_arch",
831 		       SYSCTL_DESCR("Machine CPU class"),
832 		       NULL, 0, machine_arch, 0,
833 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
834 	sysctl_createv(clog, 0, NULL, NULL,
835 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
836 		       CTLTYPE_INT, "alignbytes",
837 		       SYSCTL_DESCR("Alignment constraint for all possible "
838 				    "data types"),
839 		       NULL, ALIGNBYTES, NULL, 0,
840 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
841 	sysctl_createv(clog, 0, NULL, NULL,
842 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
843 		       CTLTYPE_STRING, "cnmagic",
844 		       SYSCTL_DESCR("Console magic key sequence"),
845 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
846 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
847 	q = (u_quad_t)physmem * PAGE_SIZE;
848 	sysctl_createv(clog, 0, NULL, NULL,
849 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
850 		       CTLTYPE_QUAD, "physmem64",
851 		       SYSCTL_DESCR("Bytes of physical memory"),
852 		       NULL, q, NULL, 0,
853 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
854 	sysctl_createv(clog, 0, NULL, NULL,
855 		       CTLFLAG_PERMANENT,
856 		       CTLTYPE_QUAD, "usermem64",
857 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
858 		       sysctl_hw_usermem, 0, NULL, 0,
859 		       CTL_HW, HW_USERMEM64, CTL_EOL);
860 	sysctl_createv(clog, 0, NULL, NULL,
861 		       CTLFLAG_PERMANENT,
862 		       CTLTYPE_INT, "ncpuonline",
863 		       SYSCTL_DESCR("Number of CPUs online"),
864 		       NULL, 0, &ncpuonline, 0,
865 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
866 }
867 
868 #ifdef DEBUG
869 /*
870  * Debugging related system variables.
871  */
872 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
873 struct ctldebug debug5, debug6, debug7, debug8, debug9;
874 struct ctldebug debug10, debug11, debug12, debug13, debug14;
875 struct ctldebug debug15, debug16, debug17, debug18, debug19;
876 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
877 	&debug0, &debug1, &debug2, &debug3, &debug4,
878 	&debug5, &debug6, &debug7, &debug8, &debug9,
879 	&debug10, &debug11, &debug12, &debug13, &debug14,
880 	&debug15, &debug16, &debug17, &debug18, &debug19,
881 };
882 
883 /*
884  * this setup routine is a replacement for debug_sysctl()
885  *
886  * note that it creates several nodes per defined debug variable
887  */
888 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
889 {
890 	struct ctldebug *cdp;
891 	char nodename[20];
892 	int i;
893 
894 	/*
895 	 * two ways here:
896 	 *
897 	 * the "old" way (debug.name -> value) which was emulated by
898 	 * the sysctl(8) binary
899 	 *
900 	 * the new way, which the sysctl(8) binary was actually using
901 
902 	 node	debug
903 	 node	debug.0
904 	 string debug.0.name
905 	 int	debug.0.value
906 	 int	debug.name
907 
908 	 */
909 
910 	sysctl_createv(clog, 0, NULL, NULL,
911 		       CTLFLAG_PERMANENT,
912 		       CTLTYPE_NODE, "debug", NULL,
913 		       NULL, 0, NULL, 0,
914 		       CTL_DEBUG, CTL_EOL);
915 
916 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
917 		cdp = debugvars[i];
918 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
919 			continue;
920 
921 		snprintf(nodename, sizeof(nodename), "debug%d", i);
922 		sysctl_createv(clog, 0, NULL, NULL,
923 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
924 			       CTLTYPE_NODE, nodename, NULL,
925 			       NULL, 0, NULL, 0,
926 			       CTL_DEBUG, i, CTL_EOL);
927 		sysctl_createv(clog, 0, NULL, NULL,
928 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
929 			       CTLTYPE_STRING, "name", NULL,
930 			       /*XXXUNCONST*/
931 			       NULL, 0, __UNCONST(cdp->debugname), 0,
932 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
933 		sysctl_createv(clog, 0, NULL, NULL,
934 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
935 			       CTLTYPE_INT, "value", NULL,
936 			       NULL, 0, cdp->debugvar, 0,
937 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
938 		sysctl_createv(clog, 0, NULL, NULL,
939 			       CTLFLAG_PERMANENT,
940 			       CTLTYPE_INT, cdp->debugname, NULL,
941 			       NULL, 0, cdp->debugvar, 0,
942 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
943 	}
944 }
945 #endif /* DEBUG */
946 
947 /*
948  * ********************************************************************
949  * section 2: private node-specific helper routines.
950  * ********************************************************************
951  */
952 
953 #ifdef DIAGNOSTIC
954 static int
955 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
956 {
957 	int newtrig, error;
958 	struct sysctlnode node;
959 
960 	newtrig = 0;
961 	node = *rnode;
962 	node.sysctl_data = &newtrig;
963 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
964 	if (error || newp == NULL)
965 		return (error);
966 
967 	if (newtrig != 0)
968 		panic("Panic triggered");
969 
970 	return (error);
971 }
972 #endif
973 
974 /*
975  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
976  * new value is lower than desiredvnodes and then calls reinit
977  * routines that needs to adjust to the new value.
978  */
979 static int
980 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
981 {
982 	int error, new_vnodes, old_vnodes, new_max;
983 	struct sysctlnode node;
984 
985 	new_vnodes = desiredvnodes;
986 	node = *rnode;
987 	node.sysctl_data = &new_vnodes;
988 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
989 	if (error || newp == NULL)
990 		return (error);
991 
992 	/* Limits: 75% of KVA and physical memory. */
993 	new_max = calc_cache_size(kernel_map, 75, 75) / VNODE_COST;
994 	if (new_vnodes > new_max)
995 		new_vnodes = new_max;
996 
997 	old_vnodes = desiredvnodes;
998 	desiredvnodes = new_vnodes;
999 	if (new_vnodes < old_vnodes) {
1000 		error = vfs_drainvnodes(new_vnodes);
1001 		if (error) {
1002 			desiredvnodes = old_vnodes;
1003 			return (error);
1004 		}
1005 	}
1006 	vfs_reinit();
1007 	nchreinit();
1008 
1009 	return (0);
1010 }
1011 
1012 /*
1013  * sysctl helper routine for rtc_offset - set time after changes
1014  */
1015 static int
1016 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
1017 {
1018 	struct timespec ts, delta;
1019 	int error, new_rtc_offset;
1020 	struct sysctlnode node;
1021 
1022 	new_rtc_offset = rtc_offset;
1023 	node = *rnode;
1024 	node.sysctl_data = &new_rtc_offset;
1025 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1026 	if (error || newp == NULL)
1027 		return (error);
1028 
1029 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
1030 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
1031 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
1032 		return (EPERM);
1033 	if (rtc_offset == new_rtc_offset)
1034 		return (0);
1035 
1036 	/* if we change the offset, adjust the time */
1037 	nanotime(&ts);
1038 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
1039 	delta.tv_nsec = 0;
1040 	timespecadd(&ts, &delta, &ts);
1041 	rtc_offset = new_rtc_offset;
1042 	return (settime(l->l_proc, &ts));
1043 }
1044 
1045 /*
1046  * sysctl helper routine for kern.maxproc. Ensures that the new
1047  * values are not too low or too high.
1048  */
1049 static int
1050 sysctl_kern_maxproc(SYSCTLFN_ARGS)
1051 {
1052 	int error, nmaxproc;
1053 	struct sysctlnode node;
1054 
1055 	nmaxproc = maxproc;
1056 	node = *rnode;
1057 	node.sysctl_data = &nmaxproc;
1058 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1059 	if (error || newp == NULL)
1060 		return (error);
1061 
1062 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
1063 		return (EINVAL);
1064 #ifdef __HAVE_CPU_MAXPROC
1065 	if (nmaxproc > cpu_maxproc())
1066 		return (EINVAL);
1067 #endif
1068 	maxproc = nmaxproc;
1069 
1070 	return (0);
1071 }
1072 
1073 /*
1074  * sysctl helper function for kern.hostid. The hostid is a long, but
1075  * we export it as an int, so we need to give it a little help.
1076  */
1077 static int
1078 sysctl_kern_hostid(SYSCTLFN_ARGS)
1079 {
1080 	int error, inthostid;
1081 	struct sysctlnode node;
1082 
1083 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
1084 	node = *rnode;
1085 	node.sysctl_data = &inthostid;
1086 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1087 	if (error || newp == NULL)
1088 		return (error);
1089 
1090 	hostid = (unsigned)inthostid;
1091 
1092 	return (0);
1093 }
1094 
1095 /*
1096  * sysctl helper function for kern.hostname and kern.domainnname.
1097  * resets the relevant recorded length when the underlying name is
1098  * changed.
1099  */
1100 static int
1101 sysctl_setlen(SYSCTLFN_ARGS)
1102 {
1103 	int error;
1104 
1105 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1106 	if (error || newp == NULL)
1107 		return (error);
1108 
1109 	switch (rnode->sysctl_num) {
1110 	case KERN_HOSTNAME:
1111 		hostnamelen = strlen((const char*)rnode->sysctl_data);
1112 		break;
1113 	case KERN_DOMAINNAME:
1114 		domainnamelen = strlen((const char*)rnode->sysctl_data);
1115 		break;
1116 	}
1117 
1118 	return (0);
1119 }
1120 
1121 /*
1122  * sysctl helper routine for kern.clockrate. Assembles a struct on
1123  * the fly to be returned to the caller.
1124  */
1125 static int
1126 sysctl_kern_clockrate(SYSCTLFN_ARGS)
1127 {
1128 	struct clockinfo clkinfo;
1129 	struct sysctlnode node;
1130 
1131 	clkinfo.tick = tick;
1132 	clkinfo.tickadj = tickadj;
1133 	clkinfo.hz = hz;
1134 	clkinfo.profhz = profhz;
1135 	clkinfo.stathz = stathz ? stathz : hz;
1136 
1137 	node = *rnode;
1138 	node.sysctl_data = &clkinfo;
1139 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1140 }
1141 
1142 /*
1143  * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
1144  * former it merely checks the message buffer is set up. For the latter,
1145  * it also copies out the data if necessary.
1146  */
1147 static int
1148 sysctl_msgbuf(SYSCTLFN_ARGS)
1149 {
1150 	char *where = oldp;
1151 	size_t len, maxlen;
1152 	long beg, end;
1153 	extern kmutex_t log_lock;
1154 	int error;
1155 
1156 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1157 		msgbufenabled = 0;
1158 		return (ENXIO);
1159 	}
1160 
1161 	switch (rnode->sysctl_num) {
1162 	case KERN_MSGBUFSIZE: {
1163 		struct sysctlnode node = *rnode;
1164 		int msg_bufs = (int)msgbufp->msg_bufs;
1165 		node.sysctl_data = &msg_bufs;
1166 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1167 	}
1168 	case KERN_MSGBUF:
1169 		break;
1170 	default:
1171 		return (EOPNOTSUPP);
1172 	}
1173 
1174 	if (newp != NULL)
1175 		return (EPERM);
1176 
1177 	if (oldp == NULL) {
1178 		/* always return full buffer size */
1179 		*oldlenp = msgbufp->msg_bufs;
1180 		return (0);
1181 	}
1182 
1183 	sysctl_unlock();
1184 
1185 	/*
1186 	 * First, copy from the write pointer to the end of
1187 	 * message buffer.
1188 	 */
1189 	error = 0;
1190 	mutex_spin_enter(&log_lock);
1191 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
1192 	beg = msgbufp->msg_bufx;
1193 	end = msgbufp->msg_bufs;
1194 	mutex_spin_exit(&log_lock);
1195 
1196 	while (maxlen > 0) {
1197 		len = MIN(end - beg, maxlen);
1198 		if (len == 0)
1199 			break;
1200 		/* XXX unlocked, but hardly matters. */
1201 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
1202 		if (error)
1203 			break;
1204 		where += len;
1205 		maxlen -= len;
1206 
1207 		/*
1208 		 * ... then, copy from the beginning of message buffer to
1209 		 * the write pointer.
1210 		 */
1211 		beg = 0;
1212 		end = msgbufp->msg_bufx;
1213 	}
1214 
1215 	sysctl_relock();
1216 	return (error);
1217 }
1218 
1219 /*
1220  * sysctl helper routine for kern.defcorename. In the case of a new
1221  * string being assigned, check that it's not a zero-length string.
1222  * (XXX the check in -current doesn't work, but do we really care?)
1223  */
1224 static int
1225 sysctl_kern_defcorename(SYSCTLFN_ARGS)
1226 {
1227 	int error;
1228 	char *newcorename;
1229 	struct sysctlnode node;
1230 
1231 	newcorename = PNBUF_GET();
1232 	node = *rnode;
1233 	node.sysctl_data = &newcorename[0];
1234 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1235 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1236 	if (error || newp == NULL) {
1237 		goto done;
1238 	}
1239 
1240 	/*
1241 	 * when sysctl_lookup() deals with a string, it's guaranteed
1242 	 * to come back nul terminated. So there.  :)
1243 	 */
1244 	if (strlen(newcorename) == 0) {
1245 		error = EINVAL;
1246 	} else {
1247 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1248 		error = 0;
1249 	}
1250 done:
1251 	PNBUF_PUT(newcorename);
1252 	return error;
1253 }
1254 
1255 /*
1256  * sysctl helper routine for kern.cp_time node. Adds up cpu time
1257  * across all cpus.
1258  */
1259 static int
1260 sysctl_kern_cptime(SYSCTLFN_ARGS)
1261 {
1262 	struct sysctlnode node = *rnode;
1263 	uint64_t *cp_time = NULL;
1264 	int error, n = ncpu, i;
1265 	struct cpu_info *ci;
1266 	CPU_INFO_ITERATOR cii;
1267 
1268 	/*
1269 	 * if you specifically pass a buffer that is the size of the
1270 	 * sum, or if you are probing for the size, you get the "sum"
1271 	 * of cp_time (and the size thereof) across all processors.
1272 	 *
1273 	 * alternately, you can pass an additional mib number and get
1274 	 * cp_time for that particular processor.
1275 	 */
1276 	switch (namelen) {
1277 	case 0:
1278 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
1279 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1280 			n = -1; /* SUM */
1281 		}
1282 		else {
1283 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
1284 			n = -2; /* ALL */
1285 		}
1286 		break;
1287 	case 1:
1288 		if (name[0] < 0 || name[0] >= n)
1289 			return (ENOENT); /* ENOSUCHPROCESSOR */
1290 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1291 		n = name[0];
1292 		/*
1293 		 * adjust these so that sysctl_lookup() will be happy
1294 		 */
1295 		name++;
1296 		namelen--;
1297 		break;
1298 	default:
1299 		return (EINVAL);
1300 	}
1301 
1302 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
1303 	if (cp_time == NULL)
1304 		return (ENOMEM);
1305 	node.sysctl_data = cp_time;
1306 	memset(cp_time, 0, node.sysctl_size);
1307 
1308 	for (CPU_INFO_FOREACH(cii, ci)) {
1309 		if (n <= 0) {
1310 			for (i = 0; i < CPUSTATES; i++) {
1311 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1312 			}
1313 		}
1314 		/*
1315 		 * if a specific processor was requested and we just
1316 		 * did it, we're done here
1317 		 */
1318 		if (n == 0)
1319 			break;
1320 		/*
1321 		 * if doing "all", skip to next cp_time set for next processor
1322 		 */
1323 		if (n == -2)
1324 			cp_time += CPUSTATES;
1325 		/*
1326 		 * if we're doing a specific processor, we're one
1327 		 * processor closer
1328 		 */
1329 		if (n > 0)
1330 			n--;
1331 	}
1332 
1333 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1334 	kmem_free(node.sysctl_data, node.sysctl_size);
1335 	return (error);
1336 }
1337 
1338 #if NPTY > 0
1339 /*
1340  * sysctl helper routine for kern.maxptys. Ensures that any new value
1341  * is acceptable to the pty subsystem.
1342  */
1343 static int
1344 sysctl_kern_maxptys(SYSCTLFN_ARGS)
1345 {
1346 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1347 	int error, xmax;
1348 	struct sysctlnode node;
1349 
1350 	/* get current value of maxptys */
1351 	xmax = pty_maxptys(0, 0);
1352 
1353 	node = *rnode;
1354 	node.sysctl_data = &xmax;
1355 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1356 	if (error || newp == NULL)
1357 		return (error);
1358 
1359 	if (xmax != pty_maxptys(xmax, 1))
1360 		return (EINVAL);
1361 
1362 	return (0);
1363 }
1364 #endif /* NPTY > 0 */
1365 
1366 /*
1367  * sysctl helper routine for kern.sbmax. Basically just ensures that
1368  * any new value is not too small.
1369  */
1370 static int
1371 sysctl_kern_sbmax(SYSCTLFN_ARGS)
1372 {
1373 	int error, new_sbmax;
1374 	struct sysctlnode node;
1375 
1376 	new_sbmax = sb_max;
1377 	node = *rnode;
1378 	node.sysctl_data = &new_sbmax;
1379 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1380 	if (error || newp == NULL)
1381 		return (error);
1382 
1383 	KERNEL_LOCK(1, NULL);
1384 	error = sb_max_set(new_sbmax);
1385 	KERNEL_UNLOCK_ONE(NULL);
1386 
1387 	return (error);
1388 }
1389 
1390 /*
1391  * sysctl helper routine for kern.urandom node. Picks a random number
1392  * for you.
1393  */
1394 static int
1395 sysctl_kern_urnd(SYSCTLFN_ARGS)
1396 {
1397 	int v, rv;
1398 
1399 	rv = cprng_strong(sysctl_prng, &v, sizeof(v), 0);
1400 	if (rv == sizeof(v)) {
1401 		struct sysctlnode node = *rnode;
1402 		node.sysctl_data = &v;
1403 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1404 	}
1405 	else
1406 		return (EIO);	/*XXX*/
1407 }
1408 
1409 /*
1410  * sysctl helper routine for kern.arandom node. Picks a random number
1411  * for you.
1412  */
1413 static int
1414 sysctl_kern_arnd(SYSCTLFN_ARGS)
1415 {
1416 	int error;
1417 	void *v;
1418 	struct sysctlnode node = *rnode;
1419 
1420 	if (*oldlenp == 0)
1421 		return 0;
1422 	/*
1423 	 * This code used to allow sucking 8192 bytes at a time out
1424 	 * of the kernel arc4random generator.  Evidently there is some
1425 	 * very old OpenBSD application code that may try to do this.
1426 	 *
1427 	 * Note that this node is documented as type "INT" -- 4 or 8
1428 	 * bytes, not 8192.
1429 	 *
1430 	 * We continue to support this abuse of the "len" pointer here
1431 	 * but only 256 bytes at a time, as, anecdotally, the actual
1432 	 * application use here was to generate RC4 keys in userspace.
1433 	 *
1434 	 * Support for such large requests will probably be removed
1435 	 * entirely in the future.
1436 	 */
1437 	if (*oldlenp > 256)
1438 		return E2BIG;
1439 
1440 	v = kmem_alloc(*oldlenp, KM_SLEEP);
1441 	cprng_fast(v, *oldlenp);
1442 	node.sysctl_data = v;
1443 	node.sysctl_size = *oldlenp;
1444 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1445 	kmem_free(v, *oldlenp);
1446 	return error;
1447 }
1448 /*
1449  * sysctl helper routine to do kern.lwp.* work.
1450  */
1451 static int
1452 sysctl_kern_lwp(SYSCTLFN_ARGS)
1453 {
1454 	struct kinfo_lwp klwp;
1455 	struct proc *p;
1456 	struct lwp *l2, *l3;
1457 	char *where, *dp;
1458 	int pid, elem_size, elem_count;
1459 	int buflen, needed, error;
1460 	bool gotit;
1461 
1462 	if (namelen == 1 && name[0] == CTL_QUERY)
1463 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1464 
1465 	dp = where = oldp;
1466 	buflen = where != NULL ? *oldlenp : 0;
1467 	error = needed = 0;
1468 
1469 	if (newp != NULL || namelen != 3)
1470 		return (EINVAL);
1471 	pid = name[0];
1472 	elem_size = name[1];
1473 	elem_count = name[2];
1474 
1475 	sysctl_unlock();
1476 	if (pid == -1) {
1477 		mutex_enter(proc_lock);
1478 		PROCLIST_FOREACH(p, &allproc) {
1479 			/* Grab a hold on the process. */
1480 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1481 				continue;
1482 			}
1483 			mutex_exit(proc_lock);
1484 
1485 			mutex_enter(p->p_lock);
1486 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1487 				if (buflen >= elem_size && elem_count > 0) {
1488 					lwp_lock(l2);
1489 					fill_lwp(l2, &klwp);
1490 					lwp_unlock(l2);
1491 					mutex_exit(p->p_lock);
1492 
1493 					/*
1494 					 * Copy out elem_size, but not
1495 					 * larger than the size of a
1496 					 * struct kinfo_proc2.
1497 					 */
1498 					error = dcopyout(l, &klwp, dp,
1499 					    min(sizeof(klwp), elem_size));
1500 					if (error) {
1501 						rw_exit(&p->p_reflock);
1502 						goto cleanup;
1503 					}
1504 					mutex_enter(p->p_lock);
1505 					LIST_FOREACH(l3, &p->p_lwps,
1506 					    l_sibling) {
1507 						if (l2 == l3)
1508 							break;
1509 					}
1510 					if (l3 == NULL) {
1511 						mutex_exit(p->p_lock);
1512 						rw_exit(&p->p_reflock);
1513 						error = EAGAIN;
1514 						goto cleanup;
1515 					}
1516 					dp += elem_size;
1517 					buflen -= elem_size;
1518 					elem_count--;
1519 				}
1520 				needed += elem_size;
1521 			}
1522 			mutex_exit(p->p_lock);
1523 
1524 			/* Drop reference to process. */
1525 			mutex_enter(proc_lock);
1526 			rw_exit(&p->p_reflock);
1527 		}
1528 		mutex_exit(proc_lock);
1529 	} else {
1530 		mutex_enter(proc_lock);
1531 		p = proc_find(pid);
1532 		if (p == NULL) {
1533 			error = ESRCH;
1534 			mutex_exit(proc_lock);
1535 			goto cleanup;
1536 		}
1537 		/* Grab a hold on the process. */
1538 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1539 		mutex_exit(proc_lock);
1540 		if (!gotit) {
1541 			error = ESRCH;
1542 			goto cleanup;
1543 		}
1544 
1545 		mutex_enter(p->p_lock);
1546 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1547 			if (buflen >= elem_size && elem_count > 0) {
1548 				lwp_lock(l2);
1549 				fill_lwp(l2, &klwp);
1550 				lwp_unlock(l2);
1551 				mutex_exit(p->p_lock);
1552 				/*
1553 				 * Copy out elem_size, but not larger than
1554 				 * the size of a struct kinfo_proc2.
1555 				 */
1556 				error = dcopyout(l, &klwp, dp,
1557 				    min(sizeof(klwp), elem_size));
1558 				if (error) {
1559 					rw_exit(&p->p_reflock);
1560 					goto cleanup;
1561 				}
1562 				mutex_enter(p->p_lock);
1563 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1564 					if (l2 == l3)
1565 						break;
1566 				}
1567 				if (l3 == NULL) {
1568 					mutex_exit(p->p_lock);
1569 					rw_exit(&p->p_reflock);
1570 					error = EAGAIN;
1571 					goto cleanup;
1572 				}
1573 				dp += elem_size;
1574 				buflen -= elem_size;
1575 				elem_count--;
1576 			}
1577 			needed += elem_size;
1578 		}
1579 		mutex_exit(p->p_lock);
1580 
1581 		/* Drop reference to process. */
1582 		rw_exit(&p->p_reflock);
1583 	}
1584 
1585 	if (where != NULL) {
1586 		*oldlenp = dp - where;
1587 		if (needed > *oldlenp) {
1588 			sysctl_relock();
1589 			return (ENOMEM);
1590 		}
1591 	} else {
1592 		needed += KERN_LWPSLOP;
1593 		*oldlenp = needed;
1594 	}
1595 	error = 0;
1596  cleanup:
1597 	sysctl_relock();
1598 	return (error);
1599 }
1600 
1601 /*
1602  * sysctl helper routine for kern.forkfsleep node. Ensures that the
1603  * given value is not too large or two small, and is at least one
1604  * timer tick if not zero.
1605  */
1606 static int
1607 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1608 {
1609 	/* userland sees value in ms, internally is in ticks */
1610 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1611 	int error, timo, lsleep;
1612 	struct sysctlnode node;
1613 
1614 	lsleep = forkfsleep * 1000 / hz;
1615 	node = *rnode;
1616 	node.sysctl_data = &lsleep;
1617 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1618 	if (error || newp == NULL)
1619 		return (error);
1620 
1621 	/* refuse negative values, and overly 'long time' */
1622 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1623 		return (EINVAL);
1624 
1625 	timo = mstohz(lsleep);
1626 
1627 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1628 	if (lsleep != 0 && timo == 0)
1629 		forkfsleep = 1;
1630 	else
1631 		forkfsleep = timo;
1632 
1633 	return (0);
1634 }
1635 
1636 /*
1637  * sysctl helper routine for kern.root_partition
1638  */
1639 static int
1640 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1641 {
1642 	int rootpart = DISKPART(rootdev);
1643 	struct sysctlnode node = *rnode;
1644 
1645 	node.sysctl_data = &rootpart;
1646 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1647 }
1648 
1649 /*
1650  * sysctl helper function for kern.drivers
1651  */
1652 static int
1653 sysctl_kern_drivers(SYSCTLFN_ARGS)
1654 {
1655 	int error;
1656 	size_t buflen;
1657 	struct kinfo_drivers kd;
1658 	char *start, *where;
1659 	const char *dname;
1660 	int i;
1661 	extern struct devsw_conv *devsw_conv;
1662 	extern int max_devsw_convs;
1663 
1664 	if (newp != NULL || namelen != 0)
1665 		return (EINVAL);
1666 
1667 	start = where = oldp;
1668 	buflen = *oldlenp;
1669 	if (where == NULL) {
1670 		*oldlenp = max_devsw_convs * sizeof kd;
1671 		return 0;
1672 	}
1673 
1674 	/*
1675 	 * An array of kinfo_drivers structures
1676 	 */
1677 	error = 0;
1678 	sysctl_unlock();
1679 	mutex_enter(&device_lock);
1680 	for (i = 0; i < max_devsw_convs; i++) {
1681 		dname = devsw_conv[i].d_name;
1682 		if (dname == NULL)
1683 			continue;
1684 		if (buflen < sizeof kd) {
1685 			error = ENOMEM;
1686 			break;
1687 		}
1688 		memset(&kd, 0, sizeof(kd));
1689 		kd.d_bmajor = devsw_conv[i].d_bmajor;
1690 		kd.d_cmajor = devsw_conv[i].d_cmajor;
1691 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1692 		mutex_exit(&device_lock);
1693 		error = dcopyout(l, &kd, where, sizeof kd);
1694 		mutex_enter(&device_lock);
1695 		if (error != 0)
1696 			break;
1697 		buflen -= sizeof kd;
1698 		where += sizeof kd;
1699 	}
1700 	mutex_exit(&device_lock);
1701 	sysctl_relock();
1702 	*oldlenp = where - start;
1703 	return error;
1704 }
1705 
1706 static int
1707 sysctl_security_setidcore(SYSCTLFN_ARGS)
1708 {
1709 	int newsize, error;
1710 	struct sysctlnode node;
1711 
1712 	node = *rnode;
1713 	node.sysctl_data = &newsize;
1714 	newsize = *(int *)rnode->sysctl_data;
1715 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1716 	if (error || newp == NULL)
1717 		return error;
1718 
1719 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1720 	    0, NULL, NULL, NULL))
1721 		return (EPERM);
1722 
1723 	*(int *)rnode->sysctl_data = newsize;
1724 
1725 	return 0;
1726 }
1727 
1728 static int
1729 sysctl_security_setidcorename(SYSCTLFN_ARGS)
1730 {
1731 	int error;
1732 	char *newsetidcorename;
1733 	struct sysctlnode node;
1734 
1735 	newsetidcorename = PNBUF_GET();
1736 	node = *rnode;
1737 	node.sysctl_data = newsetidcorename;
1738 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1739 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1740 	if (error || newp == NULL) {
1741 		goto out;
1742 	}
1743 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1744 	    0, NULL, NULL, NULL)) {
1745 		error = EPERM;
1746 		goto out;
1747 	}
1748 	if (strlen(newsetidcorename) == 0) {
1749 		error = EINVAL;
1750 		goto out;
1751 	}
1752 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1753 out:
1754 	PNBUF_PUT(newsetidcorename);
1755 	return error;
1756 }
1757 
1758 /*
1759  * sysctl helper routine for kern.cp_id node. Maps cpus to their
1760  * cpuids.
1761  */
1762 static int
1763 sysctl_kern_cpid(SYSCTLFN_ARGS)
1764 {
1765 	struct sysctlnode node = *rnode;
1766 	uint64_t *cp_id = NULL;
1767 	int error, n = ncpu;
1768 	struct cpu_info *ci;
1769 	CPU_INFO_ITERATOR cii;
1770 
1771 	/*
1772 	 * Here you may either retrieve a single cpu id or the whole
1773 	 * set. The size you get back when probing depends on what
1774 	 * you ask for.
1775 	 */
1776 	switch (namelen) {
1777 	case 0:
1778 		node.sysctl_size = n * sizeof(uint64_t);
1779 		n = -2; /* ALL */
1780 		break;
1781 	case 1:
1782 		if (name[0] < 0 || name[0] >= n)
1783 			return (ENOENT); /* ENOSUCHPROCESSOR */
1784 		node.sysctl_size = sizeof(uint64_t);
1785 		n = name[0];
1786 		/*
1787 		 * adjust these so that sysctl_lookup() will be happy
1788 		 */
1789 		name++;
1790 		namelen--;
1791 		break;
1792 	default:
1793 		return (EINVAL);
1794 	}
1795 
1796 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
1797 	if (cp_id == NULL)
1798 		return (ENOMEM);
1799 	node.sysctl_data = cp_id;
1800 	memset(cp_id, 0, node.sysctl_size);
1801 
1802 	for (CPU_INFO_FOREACH(cii, ci)) {
1803 		if (n <= 0)
1804 			cp_id[0] = cpu_index(ci);
1805 		/*
1806 		 * if a specific processor was requested and we just
1807 		 * did it, we're done here
1808 		 */
1809 		if (n == 0)
1810 			break;
1811 		/*
1812 		 * if doing "all", skip to next cp_id slot for next processor
1813 		 */
1814 		if (n == -2)
1815 			cp_id++;
1816 		/*
1817 		 * if we're doing a specific processor, we're one
1818 		 * processor closer
1819 		 */
1820 		if (n > 0)
1821 			n--;
1822 	}
1823 
1824 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1825 	kmem_free(node.sysctl_data, node.sysctl_size);
1826 	return (error);
1827 }
1828 
1829 /*
1830  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
1831  * calculate on the fly taking into account integer overflow and the
1832  * current wired count.
1833  */
1834 static int
1835 sysctl_hw_usermem(SYSCTLFN_ARGS)
1836 {
1837 	u_int ui;
1838 	u_quad_t uq;
1839 	struct sysctlnode node;
1840 
1841 	node = *rnode;
1842 	switch (rnode->sysctl_num) {
1843 	case HW_USERMEM:
1844 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
1845 			ui = UINT_MAX;
1846 		else
1847 			ui *= PAGE_SIZE;
1848 		node.sysctl_data = &ui;
1849 		break;
1850 	case HW_USERMEM64:
1851 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
1852 		node.sysctl_data = &uq;
1853 		break;
1854 	default:
1855 		return (EINVAL);
1856 	}
1857 
1858 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1859 }
1860 
1861 /*
1862  * sysctl helper routine for kern.cnmagic node. Pulls the old value
1863  * out, encoded, and stuffs the new value in for decoding.
1864  */
1865 static int
1866 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
1867 {
1868 	char magic[CNS_LEN];
1869 	int error;
1870 	struct sysctlnode node;
1871 
1872 	if (oldp)
1873 		cn_get_magic(magic, CNS_LEN);
1874 	node = *rnode;
1875 	node.sysctl_data = &magic[0];
1876 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1877 	if (error || newp == NULL)
1878 		return (error);
1879 
1880 	return (cn_set_magic(magic));
1881 }
1882 
1883 /*
1884  * ********************************************************************
1885  * section 3: public helper routines that are used for more than one
1886  * node
1887  * ********************************************************************
1888  */
1889 
1890 /*
1891  * sysctl helper routine for the kern.root_device node and some ports'
1892  * machdep.root_device nodes.
1893  */
1894 int
1895 sysctl_root_device(SYSCTLFN_ARGS)
1896 {
1897 	struct sysctlnode node;
1898 
1899 	node = *rnode;
1900 	node.sysctl_data = root_device->dv_xname;
1901 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
1902 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1903 }
1904 
1905 /*
1906  * sysctl helper routine for kern.consdev, dependent on the current
1907  * state of the console. Also used for machdep.console_device on some
1908  * ports.
1909  */
1910 int
1911 sysctl_consdev(SYSCTLFN_ARGS)
1912 {
1913 	dev_t consdev;
1914 	uint32_t oconsdev;
1915 	struct sysctlnode node;
1916 
1917 	if (cn_tab != NULL)
1918 		consdev = cn_tab->cn_dev;
1919 	else
1920 		consdev = NODEV;
1921 	node = *rnode;
1922 	switch (*oldlenp) {
1923 	case sizeof(consdev):
1924 		node.sysctl_data = &consdev;
1925 		node.sysctl_size = sizeof(consdev);
1926 		break;
1927 	case sizeof(oconsdev):
1928 		oconsdev = (uint32_t)consdev;
1929 		node.sysctl_data = &oconsdev;
1930 		node.sysctl_size = sizeof(oconsdev);
1931 		break;
1932 	default:
1933 		return EINVAL;
1934 	}
1935 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1936 }
1937 
1938 /*
1939  * ********************************************************************
1940  * section 4: support for some helpers
1941  * ********************************************************************
1942  */
1943 
1944 
1945 /*
1946  * Fill in a kinfo_lwp structure for the specified lwp.
1947  */
1948 static void
1949 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
1950 {
1951 	struct proc *p = l->l_proc;
1952 	struct timeval tv;
1953 
1954 	KASSERT(lwp_locked(l, NULL));
1955 
1956 	memset(kl, 0, sizeof(*kl));
1957 
1958 	kl->l_forw = 0;
1959 	kl->l_back = 0;
1960 	kl->l_laddr = PTRTOUINT64(l);
1961 	kl->l_addr = PTRTOUINT64(l->l_addr);
1962 	kl->l_stat = l->l_stat;
1963 	kl->l_lid = l->l_lid;
1964 	kl->l_flag = L_INMEM;
1965 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
1966 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
1967 
1968 	kl->l_swtime = l->l_swtime;
1969 	kl->l_slptime = l->l_slptime;
1970 	if (l->l_stat == LSONPROC)
1971 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1972 	else
1973 		kl->l_schedflags = 0;
1974 	kl->l_priority = lwp_eprio(l);
1975 	kl->l_usrpri = l->l_priority;
1976 	if (l->l_wchan)
1977 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
1978 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
1979 	kl->l_cpuid = cpu_index(l->l_cpu);
1980 	bintime2timeval(&l->l_rtime, &tv);
1981 	kl->l_rtime_sec = tv.tv_sec;
1982 	kl->l_rtime_usec = tv.tv_usec;
1983 	kl->l_cpticks = l->l_cpticks;
1984 	kl->l_pctcpu = l->l_pctcpu;
1985 	kl->l_pid = p->p_pid;
1986 	if (l->l_name == NULL)
1987 		kl->l_name[0] = '\0';
1988 	else
1989 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
1990 }
1991