xref: /netbsd-src/sys/kern/init_sysctl.c (revision a4ddc2c8fb9af816efe3b1c375a5530aef0e89e9)
1 /*	$NetBSD: init_sysctl.c,v 1.197 2013/03/18 13:36:22 para 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 1.197 2013/03/18 13:36:22 para Exp $");
34 
35 #include "opt_sysv.h"
36 #include "opt_compat_netbsd.h"
37 #include "opt_modular.h"
38 #include "pty.h"
39 
40 #define SYSCTL_PRIVATE
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 #include <sys/cpu.h>
73 
74 int security_setidcore_dump;
75 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
76 uid_t security_setidcore_owner = 0;
77 gid_t security_setidcore_group = 0;
78 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
79 
80 static const u_int sysctl_flagmap[] = {
81 	PK_ADVLOCK, P_ADVLOCK,
82 	PK_EXEC, P_EXEC,
83 	PK_NOCLDWAIT, P_NOCLDWAIT,
84 	PK_32, P_32,
85 	PK_CLDSIGIGN, P_CLDSIGIGN,
86 	PK_SUGID, P_SUGID,
87 	0
88 };
89 
90 static const u_int sysctl_sflagmap[] = {
91 	PS_NOCLDSTOP, P_NOCLDSTOP,
92 	PS_WEXIT, P_WEXIT,
93 	PS_STOPFORK, P_STOPFORK,
94 	PS_STOPEXEC, P_STOPEXEC,
95 	PS_STOPEXIT, P_STOPEXIT,
96 	0
97 };
98 
99 static const u_int sysctl_slflagmap[] = {
100 	PSL_TRACED, P_TRACED,
101 	PSL_FSTRACE, P_FSTRACE,
102 	PSL_CHTRACED, P_CHTRACED,
103 	PSL_SYSCALL, P_SYSCALL,
104 	0
105 };
106 
107 static const u_int sysctl_lflagmap[] = {
108 	PL_CONTROLT, P_CONTROLT,
109 	PL_PPWAIT, P_PPWAIT,
110 	0
111 };
112 
113 static const u_int sysctl_stflagmap[] = {
114 	PST_PROFIL, P_PROFIL,
115 	0
116 
117 };
118 
119 static const u_int sysctl_lwpprflagmap[] = {
120 	LPR_DETACHED, L_DETACHED,
121 	0
122 };
123 
124 /*
125  * try over estimating by 5 procs/lwps
126  */
127 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
128 
129 static int dcopyout(struct lwp *, const void *, void *, size_t);
130 
131 static int
132 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
133 {
134 	int error;
135 
136 	error = copyout(kaddr, uaddr, len);
137 	ktrmibio(-1, UIO_READ, uaddr, len, error);
138 
139 	return error;
140 }
141 
142 #ifdef DIAGNOSTIC
143 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
144 #endif
145 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
146 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
147 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
148 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
149 static int sysctl_setlen(SYSCTLFN_PROTO);
150 static int sysctl_kern_clockrate(SYSCTLFN_PROTO);
151 static int sysctl_msgbuf(SYSCTLFN_PROTO);
152 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
153 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
154 #if NPTY > 0
155 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
156 #endif /* NPTY > 0 */
157 static int sysctl_kern_urnd(SYSCTLFN_PROTO);
158 static int sysctl_kern_arnd(SYSCTLFN_PROTO);
159 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
160 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
161 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
162 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
163 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
164 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
165 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
166 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
167 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
168 
169 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
170 
171 /*
172  * ********************************************************************
173  * section 1: setup routines
174  * ********************************************************************
175  * These functions are stuffed into a link set for sysctl setup
176  * functions. They're never called or referenced from anywhere else.
177  * ********************************************************************
178  */
179 
180 /*
181  * this setup routine is a replacement for kern_sysctl()
182  */
183 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
184 {
185 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
186 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
187 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
188 	const struct sysctlnode *rnode;
189 
190 	sysctl_createv(clog, 0, NULL, NULL,
191 		       CTLFLAG_PERMANENT,
192 		       CTLTYPE_NODE, "kern", NULL,
193 		       NULL, 0, NULL, 0,
194 		       CTL_KERN, CTL_EOL);
195 
196 	sysctl_createv(clog, 0, NULL, NULL,
197 		       CTLFLAG_PERMANENT,
198 		       CTLTYPE_STRING, "ostype",
199 		       SYSCTL_DESCR("Operating system type"),
200 		       NULL, 0, __UNCONST(&ostype), 0,
201 		       CTL_KERN, KERN_OSTYPE, CTL_EOL);
202 	sysctl_createv(clog, 0, NULL, NULL,
203 		       CTLFLAG_PERMANENT,
204 		       CTLTYPE_STRING, "osrelease",
205 		       SYSCTL_DESCR("Operating system release"),
206 		       NULL, 0, __UNCONST(&osrelease), 0,
207 		       CTL_KERN, KERN_OSRELEASE, CTL_EOL);
208 	sysctl_createv(clog, 0, NULL, NULL,
209 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
210 		       CTLTYPE_INT, "osrevision",
211 		       SYSCTL_DESCR("Operating system revision"),
212 		       NULL, __NetBSD_Version__, NULL, 0,
213 		       CTL_KERN, KERN_OSREV, CTL_EOL);
214 	sysctl_createv(clog, 0, NULL, NULL,
215 		       CTLFLAG_PERMANENT,
216 		       CTLTYPE_STRING, "version",
217 		       SYSCTL_DESCR("Kernel version"),
218 		       NULL, 0, __UNCONST(&version), 0,
219 		       CTL_KERN, KERN_VERSION, CTL_EOL);
220 	sysctl_createv(clog, 0, NULL, NULL,
221 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
222 		       CTLTYPE_INT, "maxvnodes",
223 		       SYSCTL_DESCR("Maximum number of vnodes"),
224 		       sysctl_kern_maxvnodes, 0, NULL, 0,
225 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
226 	sysctl_createv(clog, 0, NULL, NULL,
227 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
228 		       CTLTYPE_INT, "maxproc",
229 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
230 		       sysctl_kern_maxproc, 0, NULL, 0,
231 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
232 	sysctl_createv(clog, 0, NULL, NULL,
233 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
234 		       CTLTYPE_INT, "maxfiles",
235 		       SYSCTL_DESCR("Maximum number of open files"),
236 		       NULL, 0, &maxfiles, 0,
237 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
238 	sysctl_createv(clog, 0, NULL, NULL,
239 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
240 		       CTLTYPE_INT, "argmax",
241 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
242 				    "execve(2)"),
243 		       NULL, ARG_MAX, NULL, 0,
244 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
245 	sysctl_createv(clog, 0, NULL, NULL,
246 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
247 		       CTLTYPE_STRING, "hostname",
248 		       SYSCTL_DESCR("System hostname"),
249 		       sysctl_setlen, 0, hostname, MAXHOSTNAMELEN,
250 		       CTL_KERN, KERN_HOSTNAME, CTL_EOL);
251 	sysctl_createv(clog, 0, NULL, NULL,
252 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
253 		       CTLTYPE_INT, "hostid",
254 		       SYSCTL_DESCR("System host ID number"),
255 		       sysctl_kern_hostid, 0, NULL, 0,
256 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
257 	sysctl_createv(clog, 0, NULL, NULL,
258 		       CTLFLAG_PERMANENT,
259 		       CTLTYPE_STRUCT, "clockrate",
260 		       SYSCTL_DESCR("Kernel clock rates"),
261 		       sysctl_kern_clockrate, 0, NULL,
262 		       sizeof(struct clockinfo),
263 		       CTL_KERN, KERN_CLOCKRATE, CTL_EOL);
264 	sysctl_createv(clog, 0, NULL, NULL,
265 		       CTLFLAG_PERMANENT,
266 		       CTLTYPE_INT, "hardclock_ticks",
267 		       SYSCTL_DESCR("Number of hardclock ticks"),
268 		       NULL, 0, &hardclock_ticks, sizeof(hardclock_ticks),
269 		       CTL_KERN, KERN_HARDCLOCK_TICKS, CTL_EOL);
270 	sysctl_createv(clog, 0, NULL, NULL,
271 		       CTLFLAG_PERMANENT,
272 		       CTLTYPE_STRUCT, "vnode",
273 		       SYSCTL_DESCR("System vnode table"),
274 		       sysctl_kern_vnode, 0, NULL, 0,
275 		       CTL_KERN, KERN_VNODE, CTL_EOL);
276 #ifndef GPROF
277 	sysctl_createv(clog, 0, NULL, NULL,
278 		       CTLFLAG_PERMANENT,
279 		       CTLTYPE_NODE, "profiling",
280 		       SYSCTL_DESCR("Profiling information (not available)"),
281 		       sysctl_notavail, 0, NULL, 0,
282 		       CTL_KERN, KERN_PROF, CTL_EOL);
283 #endif
284 	sysctl_createv(clog, 0, NULL, NULL,
285 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
286 		       CTLTYPE_INT, "posix1version",
287 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
288 				    "with which the operating system attempts "
289 				    "to comply"),
290 		       NULL, _POSIX_VERSION, NULL, 0,
291 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
292 	sysctl_createv(clog, 0, NULL, NULL,
293 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
294 		       CTLTYPE_INT, "ngroups",
295 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
296 		       NULL, NGROUPS_MAX, NULL, 0,
297 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
298 	sysctl_createv(clog, 0, NULL, NULL,
299 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
300 		       CTLTYPE_INT, "job_control",
301 		       SYSCTL_DESCR("Whether job control is available"),
302 		       NULL, 1, NULL, 0,
303 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
304 	sysctl_createv(clog, 0, NULL, NULL,
305 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
306 		       CTLTYPE_INT, "saved_ids",
307 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
308 				    "available"), NULL,
309 #ifdef _POSIX_SAVED_IDS
310 		       1,
311 #else /* _POSIX_SAVED_IDS */
312 		       0,
313 #endif /* _POSIX_SAVED_IDS */
314 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
315 	sysctl_createv(clog, 0, NULL, NULL,
316 		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
317 		       CTLTYPE_INT, "boothowto",
318 		       SYSCTL_DESCR("Flags from boot loader"),
319 		       NULL, 0, &boothowto, sizeof(boothowto),
320 		       CTL_KERN, CTL_CREATE, CTL_EOL);
321 	sysctl_createv(clog, 0, NULL, NULL,
322 		       CTLFLAG_PERMANENT,
323 		       CTLTYPE_STRUCT, "boottime",
324 		       SYSCTL_DESCR("System boot time"),
325 		       NULL, 0, &boottime, sizeof(boottime),
326 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
327 	sysctl_createv(clog, 0, NULL, NULL,
328 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
329 		       CTLTYPE_STRING, "domainname",
330 		       SYSCTL_DESCR("YP domain name"),
331 		       sysctl_setlen, 0, domainname, MAXHOSTNAMELEN,
332 		       CTL_KERN, KERN_DOMAINNAME, CTL_EOL);
333 	sysctl_createv(clog, 0, NULL, NULL,
334 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
335 		       CTLTYPE_INT, "maxpartitions",
336 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
337 				    "disk"),
338 		       NULL, MAXPARTITIONS, NULL, 0,
339 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
340 	sysctl_createv(clog, 0, NULL, NULL,
341 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
342 		       CTLTYPE_INT, "rawpartition",
343 		       SYSCTL_DESCR("Raw partition of a disk"),
344 		       NULL, RAW_PART, NULL, 0,
345 		       CTL_KERN, KERN_RAWPARTITION, CTL_EOL);
346 	sysctl_createv(clog, 0, NULL, NULL,
347 		       CTLFLAG_PERMANENT,
348 		       CTLTYPE_STRUCT, "timex", NULL,
349 		       sysctl_notavail, 0, NULL, 0,
350 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
351 	sysctl_createv(clog, 0, NULL, NULL,
352 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
353 		       CTLTYPE_INT, "rtc_offset",
354 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
355 				    "minutes"),
356 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
357 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
358 	sysctl_createv(clog, 0, NULL, NULL,
359 		       CTLFLAG_PERMANENT,
360 		       CTLTYPE_STRING, "root_device",
361 		       SYSCTL_DESCR("Name of the root device"),
362 		       sysctl_root_device, 0, NULL, 0,
363 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
364 	sysctl_createv(clog, 0, NULL, NULL,
365 		       CTLFLAG_PERMANENT,
366 		       CTLTYPE_INT, "msgbufsize",
367 		       SYSCTL_DESCR("Size of the kernel message buffer"),
368 		       sysctl_msgbuf, 0, NULL, 0,
369 		       CTL_KERN, KERN_MSGBUFSIZE, CTL_EOL);
370 	sysctl_createv(clog, 0, NULL, NULL,
371 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
372 		       CTLTYPE_INT, "fsync",
373 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
374 				    "Synchronization Option is available on "
375 				    "this system"),
376 		       NULL, 1, NULL, 0,
377 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
378 	sysctl_createv(clog, 0, NULL, NULL,
379 		       CTLFLAG_PERMANENT,
380 		       CTLTYPE_NODE, "ipc",
381 		       SYSCTL_DESCR("SysV IPC options"),
382 		       NULL, 0, NULL, 0,
383 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
384 	sysctl_createv(clog, 0, NULL, NULL,
385 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
386 		       CTLTYPE_INT, "sysvmsg",
387 		       SYSCTL_DESCR("System V style message support available"),
388 		       NULL,
389 #ifdef SYSVMSG
390 		       1,
391 #else /* SYSVMSG */
392 		       0,
393 #endif /* SYSVMSG */
394 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
395 	sysctl_createv(clog, 0, NULL, NULL,
396 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
397 		       CTLTYPE_INT, "sysvsem",
398 		       SYSCTL_DESCR("System V style semaphore support "
399 				    "available"), NULL,
400 #ifdef SYSVSEM
401 		       1,
402 #else /* SYSVSEM */
403 		       0,
404 #endif /* SYSVSEM */
405 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
406 	sysctl_createv(clog, 0, NULL, NULL,
407 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
408 		       CTLTYPE_INT, "sysvshm",
409 		       SYSCTL_DESCR("System V style shared memory support "
410 				    "available"), NULL,
411 #ifdef SYSVSHM
412 		       1,
413 #else /* SYSVSHM */
414 		       0,
415 #endif /* SYSVSHM */
416 		       NULL, 0, CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
417 	sysctl_createv(clog, 0, NULL, NULL,
418 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
419 		       CTLTYPE_INT, "synchronized_io",
420 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
421 				    "I/O Option is available on this system"),
422 		       NULL, 1, NULL, 0,
423 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
424 	sysctl_createv(clog, 0, NULL, NULL,
425 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
426 		       CTLTYPE_INT, "iov_max",
427 		       SYSCTL_DESCR("Maximum number of iovec structures per "
428 				    "process"),
429 		       NULL, IOV_MAX, NULL, 0,
430 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
431 	sysctl_createv(clog, 0, NULL, NULL,
432 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
433 		       CTLTYPE_INT, "mapped_files",
434 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
435 				    "Files Option is available on this system"),
436 		       NULL, 1, NULL, 0,
437 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
438 	sysctl_createv(clog, 0, NULL, NULL,
439 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
440 		       CTLTYPE_INT, "memlock",
441 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
442 				    "Locking Option is available on this "
443 				    "system"),
444 		       NULL, 1, NULL, 0,
445 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
446 	sysctl_createv(clog, 0, NULL, NULL,
447 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
448 		       CTLTYPE_INT, "memlock_range",
449 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
450 				    "Locking Option is available on this "
451 				    "system"),
452 		       NULL, 1, NULL, 0,
453 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
454 	sysctl_createv(clog, 0, NULL, NULL,
455 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
456 		       CTLTYPE_INT, "memory_protection",
457 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
458 				    "Protection Option is available on this "
459 				    "system"),
460 		       NULL, 1, NULL, 0,
461 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
462 	sysctl_createv(clog, 0, NULL, NULL,
463 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
464 		       CTLTYPE_INT, "login_name_max",
465 		       SYSCTL_DESCR("Maximum login name length"),
466 		       NULL, LOGIN_NAME_MAX, NULL, 0,
467 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
468 	sysctl_createv(clog, 0, NULL, NULL,
469 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
470 		       CTLTYPE_STRING, "defcorename",
471 		       SYSCTL_DESCR("Default core file name"),
472 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
473 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
474 	sysctl_createv(clog, 0, NULL, NULL,
475 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
476 		       CTLTYPE_INT, "logsigexit",
477 		       SYSCTL_DESCR("Log process exit when caused by signals"),
478 		       NULL, 0, &kern_logsigexit, 0,
479 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
480 	sysctl_createv(clog, 0, NULL, NULL,
481 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
482 		       CTLTYPE_INT, "fscale",
483 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
484 		       NULL, FSCALE, NULL, 0,
485 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
486 	sysctl_createv(clog, 0, NULL, NULL,
487 		       CTLFLAG_PERMANENT,
488 		       CTLTYPE_INT, "ccpu",
489 		       SYSCTL_DESCR("Scheduler exponential decay value"),
490 		       NULL, 0, &ccpu, 0,
491 		       CTL_KERN, KERN_CCPU, CTL_EOL);
492 	sysctl_createv(clog, 0, NULL, NULL,
493 		       CTLFLAG_PERMANENT,
494 		       CTLTYPE_STRUCT, "cp_time",
495 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
496 		       sysctl_kern_cptime, 0, NULL, 0,
497 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
498 	sysctl_createv(clog, 0, NULL, NULL,
499 		       CTLFLAG_PERMANENT,
500 		       CTLTYPE_INT, "msgbuf",
501 		       SYSCTL_DESCR("Kernel message buffer"),
502 		       sysctl_msgbuf, 0, NULL, 0,
503 		       CTL_KERN, KERN_MSGBUF, CTL_EOL);
504 	sysctl_createv(clog, 0, NULL, NULL,
505 		       CTLFLAG_PERMANENT,
506 		       CTLTYPE_STRUCT, "consdev",
507 		       SYSCTL_DESCR("Console device"),
508 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
509 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
510 #if NPTY > 0
511 	sysctl_createv(clog, 0, NULL, NULL,
512 		       CTLFLAG_PERMANENT,
513 		       CTLTYPE_INT, "maxptys",
514 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
515 		       sysctl_kern_maxptys, 0, NULL, 0,
516 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
517 #endif /* NPTY > 0 */
518 	sysctl_createv(clog, 0, NULL, NULL,
519 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
520 		       CTLTYPE_INT, "maxphys",
521 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
522 		       NULL, MAXPHYS, NULL, 0,
523 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
524 	sysctl_createv(clog, 0, NULL, NULL,
525 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
526 		       CTLTYPE_INT, "monotonic_clock",
527 		       SYSCTL_DESCR("Implementation version of the POSIX "
528 				    "1003.1b Monotonic Clock Option"),
529 		       /* XXX _POSIX_VERSION */
530 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
531 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
532 	sysctl_createv(clog, 0, NULL, NULL,
533 		       CTLFLAG_PERMANENT,
534 		       CTLTYPE_INT, "urandom",
535 		       SYSCTL_DESCR("Random integer value"),
536 		       sysctl_kern_urnd, 0, NULL, 0,
537 		       CTL_KERN, KERN_URND, CTL_EOL);
538 	sysctl_createv(clog, 0, NULL, NULL,
539 		       CTLFLAG_PERMANENT,
540 		       CTLTYPE_INT, "arandom",
541 		       SYSCTL_DESCR("n bytes of random data"),
542 		       sysctl_kern_arnd, 0, NULL, 0,
543 		       CTL_KERN, KERN_ARND, CTL_EOL);
544 	sysctl_createv(clog, 0, NULL, NULL,
545 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
546 		       CTLTYPE_INT, "labelsector",
547 		       SYSCTL_DESCR("Sector number containing the disklabel"),
548 		       NULL, LABELSECTOR, NULL, 0,
549 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
550 	sysctl_createv(clog, 0, NULL, NULL,
551 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
552 		       CTLTYPE_INT, "labeloffset",
553 		       SYSCTL_DESCR("Offset of the disklabel within the "
554 				    "sector"),
555 		       NULL, LABELOFFSET, NULL, 0,
556 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
557 	sysctl_createv(clog, 0, NULL, NULL,
558 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
559 		       CTLTYPE_INT, "labelusesmbr",
560 		       SYSCTL_DESCR("disklabel is inside MBR partition"),
561 		       NULL, LABELUSESMBR, NULL, 0,
562 		       CTL_KERN, CTL_CREATE, CTL_EOL);
563 	sysctl_createv(clog, 0, NULL, NULL,
564 		       CTLFLAG_PERMANENT,
565 		       CTLTYPE_NODE, "lwp",
566 		       SYSCTL_DESCR("System-wide LWP information"),
567 		       sysctl_kern_lwp, 0, NULL, 0,
568 		       CTL_KERN, KERN_LWP, CTL_EOL);
569 	sysctl_createv(clog, 0, NULL, NULL,
570 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
571 		       CTLTYPE_INT, "forkfsleep",
572 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
573 				    "to process limits"),
574 		       sysctl_kern_forkfsleep, 0, NULL, 0,
575 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
576 	sysctl_createv(clog, 0, NULL, NULL,
577 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
578 		       CTLTYPE_INT, "posix_threads",
579 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
580 				    "Threads option to which the system "
581 				    "attempts to conform"),
582 		       /* XXX _POSIX_VERSION */
583 		       NULL, _POSIX_THREADS, NULL, 0,
584 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
585 	sysctl_createv(clog, 0, NULL, NULL,
586 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
587 		       CTLTYPE_INT, "posix_semaphores",
588 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
589 				    "Semaphores option to which the system "
590 				    "attempts to conform"), NULL,
591 		       200112, NULL, 0,
592 		       CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
593 	sysctl_createv(clog, 0, NULL, NULL,
594 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
595 		       CTLTYPE_INT, "posix_barriers",
596 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
597 				    "Barriers option to which the system "
598 				    "attempts to conform"),
599 		       /* XXX _POSIX_VERSION */
600 		       NULL, _POSIX_BARRIERS, NULL, 0,
601 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
602 	sysctl_createv(clog, 0, NULL, NULL,
603 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
604 		       CTLTYPE_INT, "posix_timers",
605 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
606 				    "Timers option to which the system "
607 				    "attempts to conform"),
608 		       /* XXX _POSIX_VERSION */
609 		       NULL, _POSIX_TIMERS, NULL, 0,
610 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
611 	sysctl_createv(clog, 0, NULL, NULL,
612 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
613 		       CTLTYPE_INT, "posix_spin_locks",
614 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
615 				    "Locks option to which the system attempts "
616 				    "to conform"),
617 		       /* XXX _POSIX_VERSION */
618 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
619 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
620 	sysctl_createv(clog, 0, NULL, NULL,
621 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
622 		       CTLTYPE_INT, "posix_reader_writer_locks",
623 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
624 				    "Read-Write Locks option to which the "
625 				    "system attempts to conform"),
626 		       /* XXX _POSIX_VERSION */
627 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
628 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
629 	sysctl_createv(clog, 0, NULL, NULL,
630 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
631 		       CTLTYPE_INT, "dump_on_panic",
632 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
633 		       NULL, 0, &dumponpanic, 0,
634 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
635 #ifdef DIAGNOSTIC
636 	sysctl_createv(clog, 0, NULL, NULL,
637 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
638 		       CTLTYPE_INT, "panic_now",
639 		       SYSCTL_DESCR("Trigger a panic"),
640 		       sysctl_kern_trigger_panic, 0, NULL, 0,
641 		       CTL_KERN, CTL_CREATE, CTL_EOL);
642 #endif
643 	sysctl_createv(clog, 0, NULL, NULL,
644 		       CTLFLAG_PERMANENT,
645 		       CTLTYPE_INT, "root_partition",
646 		       SYSCTL_DESCR("Root partition on the root device"),
647 		       sysctl_kern_root_partition, 0, NULL, 0,
648 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
649 	sysctl_createv(clog, 0, NULL, NULL,
650 		       CTLFLAG_PERMANENT,
651 		       CTLTYPE_STRUCT, "drivers",
652 		       SYSCTL_DESCR("List of all drivers with block and "
653 				    "character device numbers"),
654 		       sysctl_kern_drivers, 0, NULL, 0,
655 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
656 	sysctl_createv(clog, 0, NULL, NULL,
657 		       CTLFLAG_PERMANENT,
658 		       CTLTYPE_STRUCT, "cp_id",
659 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
660 		       sysctl_kern_cpid, 0, NULL, 0,
661 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
662 	sysctl_createv(clog, 0, NULL, &rnode,
663 		       CTLFLAG_PERMANENT,
664 		       CTLTYPE_NODE, "coredump",
665 		       SYSCTL_DESCR("Coredump settings."),
666 		       NULL, 0, NULL, 0,
667 		       CTL_KERN, CTL_CREATE, CTL_EOL);
668 	sysctl_createv(clog, 0, &rnode, &rnode,
669 		       CTLFLAG_PERMANENT,
670 		       CTLTYPE_NODE, "setid",
671 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
672 		       NULL, 0, NULL, 0,
673 		       CTL_CREATE, CTL_EOL);
674 	sysctl_createv(clog, 0, &rnode, NULL,
675 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
676 		       CTLTYPE_INT, "dump",
677 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
678 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
679 		       sizeof(security_setidcore_dump),
680 		       CTL_CREATE, CTL_EOL);
681 	sysctl_createv(clog, 0, &rnode, NULL,
682 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
683 		       CTLTYPE_STRING, "path",
684 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
685 		       sysctl_security_setidcorename, 0,
686 		       security_setidcore_path,
687 		       sizeof(security_setidcore_path),
688 		       CTL_CREATE, CTL_EOL);
689 	sysctl_createv(clog, 0, &rnode, NULL,
690 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
691 		       CTLTYPE_INT, "owner",
692 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
693 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
694 		       0,
695 		       CTL_CREATE, CTL_EOL);
696 	sysctl_createv(clog, 0, &rnode, NULL,
697 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
698 		       CTLTYPE_INT, "group",
699 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
700 		       sysctl_security_setidcore, 0, &security_setidcore_group,
701 		       0,
702 		       CTL_CREATE, CTL_EOL);
703 	sysctl_createv(clog, 0, &rnode, NULL,
704 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
705 		       CTLTYPE_INT, "mode",
706 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
707 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
708 		       0,
709 		       CTL_CREATE, CTL_EOL);
710 	sysctl_createv(clog, 0, NULL, NULL,
711 		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
712 		       CTLTYPE_INT, "no_sa_support",
713 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
714 		       "it doesn't"),
715 		       NULL, 1, NULL, 0,
716 		       CTL_KERN, CTL_CREATE, CTL_EOL);
717 	/* kern.posix. */
718 	sysctl_createv(clog, 0, NULL, &rnode,
719 			CTLFLAG_PERMANENT,
720 			CTLTYPE_NODE, "posix",
721 			SYSCTL_DESCR("POSIX options"),
722 			NULL, 0, NULL, 0,
723 			CTL_KERN, CTL_CREATE, CTL_EOL);
724 	sysctl_createv(clog, 0, &rnode, NULL,
725 			CTLFLAG_PERMANENT | CTLFLAG_READWRITE,
726 			CTLTYPE_INT, "semmax",
727 			SYSCTL_DESCR("Maximal number of semaphores"),
728 			NULL, 0, &ksem_max, 0,
729 			CTL_CREATE, CTL_EOL);
730 	sysctl_createv(clog, 0, NULL, NULL,
731 			CTLFLAG_PERMANENT,
732 			CTLTYPE_STRING, "configname",
733 			SYSCTL_DESCR("Name of config file"),
734 			NULL, 0, __UNCONST(kernel_ident), 0,
735 			CTL_KERN, CTL_CREATE, CTL_EOL);
736 }
737 
738 SYSCTL_SETUP(sysctl_hw_setup, "sysctl hw subtree setup")
739 {
740 	u_int u;
741 	u_quad_t q;
742 
743 	sysctl_createv(clog, 0, NULL, NULL,
744 		       CTLFLAG_PERMANENT,
745 		       CTLTYPE_NODE, "hw", NULL,
746 		       NULL, 0, NULL, 0,
747 		       CTL_HW, CTL_EOL);
748 
749 	sysctl_createv(clog, 0, NULL, NULL,
750 		       CTLFLAG_PERMANENT,
751 		       CTLTYPE_STRING, "machine",
752 		       SYSCTL_DESCR("Machine class"),
753 		       NULL, 0, machine, 0,
754 		       CTL_HW, HW_MACHINE, CTL_EOL);
755 	sysctl_createv(clog, 0, NULL, NULL,
756 		       CTLFLAG_PERMANENT,
757 		       CTLTYPE_STRING, "model",
758 		       SYSCTL_DESCR("Machine model"),
759 		       NULL, 0, cpu_model, 0,
760 		       CTL_HW, HW_MODEL, CTL_EOL);
761 	sysctl_createv(clog, 0, NULL, NULL,
762 		       CTLFLAG_PERMANENT,
763 		       CTLTYPE_INT, "ncpu",
764 		       SYSCTL_DESCR("Number of CPUs configured"),
765 		       NULL, 0, &ncpu, 0,
766 		       CTL_HW, HW_NCPU, CTL_EOL);
767 	sysctl_createv(clog, 0, NULL, NULL,
768 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
769 		       CTLTYPE_INT, "byteorder",
770 		       SYSCTL_DESCR("System byte order"),
771 		       NULL, BYTE_ORDER, NULL, 0,
772 		       CTL_HW, HW_BYTEORDER, CTL_EOL);
773 	u = ((u_int)physmem > (UINT_MAX / PAGE_SIZE)) ?
774 		UINT_MAX : physmem * PAGE_SIZE;
775 	sysctl_createv(clog, 0, NULL, NULL,
776 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
777 		       CTLTYPE_INT, "physmem",
778 		       SYSCTL_DESCR("Bytes of physical memory"),
779 		       NULL, u, NULL, 0,
780 		       CTL_HW, HW_PHYSMEM, CTL_EOL);
781 	sysctl_createv(clog, 0, NULL, NULL,
782 		       CTLFLAG_PERMANENT,
783 		       CTLTYPE_INT, "usermem",
784 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
785 		       sysctl_hw_usermem, 0, NULL, 0,
786 		       CTL_HW, HW_USERMEM, CTL_EOL);
787 	sysctl_createv(clog, 0, NULL, NULL,
788 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
789 		       CTLTYPE_INT, "pagesize",
790 		       SYSCTL_DESCR("Software page size"),
791 		       NULL, PAGE_SIZE, NULL, 0,
792 		       CTL_HW, HW_PAGESIZE, CTL_EOL);
793 	sysctl_createv(clog, 0, NULL, NULL,
794 		       CTLFLAG_PERMANENT,
795 		       CTLTYPE_STRING, "machine_arch",
796 		       SYSCTL_DESCR("Machine CPU class"),
797 		       NULL, 0, machine_arch, 0,
798 		       CTL_HW, HW_MACHINE_ARCH, CTL_EOL);
799 	sysctl_createv(clog, 0, NULL, NULL,
800 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
801 		       CTLTYPE_INT, "alignbytes",
802 		       SYSCTL_DESCR("Alignment constraint for all possible "
803 				    "data types"),
804 		       NULL, ALIGNBYTES, NULL, 0,
805 		       CTL_HW, HW_ALIGNBYTES, CTL_EOL);
806 	sysctl_createv(clog, 0, NULL, NULL,
807 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
808 		       CTLTYPE_STRING, "cnmagic",
809 		       SYSCTL_DESCR("Console magic key sequence"),
810 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
811 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
812 	q = (u_quad_t)physmem * PAGE_SIZE;
813 	sysctl_createv(clog, 0, NULL, NULL,
814 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
815 		       CTLTYPE_QUAD, "physmem64",
816 		       SYSCTL_DESCR("Bytes of physical memory"),
817 		       NULL, q, NULL, 0,
818 		       CTL_HW, HW_PHYSMEM64, CTL_EOL);
819 	sysctl_createv(clog, 0, NULL, NULL,
820 		       CTLFLAG_PERMANENT,
821 		       CTLTYPE_QUAD, "usermem64",
822 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
823 		       sysctl_hw_usermem, 0, NULL, 0,
824 		       CTL_HW, HW_USERMEM64, CTL_EOL);
825 	sysctl_createv(clog, 0, NULL, NULL,
826 		       CTLFLAG_PERMANENT,
827 		       CTLTYPE_INT, "ncpuonline",
828 		       SYSCTL_DESCR("Number of CPUs online"),
829 		       NULL, 0, &ncpuonline, 0,
830 		       CTL_HW, HW_NCPUONLINE, CTL_EOL);
831 }
832 
833 #ifdef DEBUG
834 /*
835  * Debugging related system variables.
836  */
837 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
838 struct ctldebug debug5, debug6, debug7, debug8, debug9;
839 struct ctldebug debug10, debug11, debug12, debug13, debug14;
840 struct ctldebug debug15, debug16, debug17, debug18, debug19;
841 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
842 	&debug0, &debug1, &debug2, &debug3, &debug4,
843 	&debug5, &debug6, &debug7, &debug8, &debug9,
844 	&debug10, &debug11, &debug12, &debug13, &debug14,
845 	&debug15, &debug16, &debug17, &debug18, &debug19,
846 };
847 
848 /*
849  * this setup routine is a replacement for debug_sysctl()
850  *
851  * note that it creates several nodes per defined debug variable
852  */
853 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
854 {
855 	struct ctldebug *cdp;
856 	char nodename[20];
857 	int i;
858 
859 	/*
860 	 * two ways here:
861 	 *
862 	 * the "old" way (debug.name -> value) which was emulated by
863 	 * the sysctl(8) binary
864 	 *
865 	 * the new way, which the sysctl(8) binary was actually using
866 
867 	 node	debug
868 	 node	debug.0
869 	 string debug.0.name
870 	 int	debug.0.value
871 	 int	debug.name
872 
873 	 */
874 
875 	sysctl_createv(clog, 0, NULL, NULL,
876 		       CTLFLAG_PERMANENT,
877 		       CTLTYPE_NODE, "debug", NULL,
878 		       NULL, 0, NULL, 0,
879 		       CTL_DEBUG, CTL_EOL);
880 
881 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
882 		cdp = debugvars[i];
883 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
884 			continue;
885 
886 		snprintf(nodename, sizeof(nodename), "debug%d", i);
887 		sysctl_createv(clog, 0, NULL, NULL,
888 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
889 			       CTLTYPE_NODE, nodename, NULL,
890 			       NULL, 0, NULL, 0,
891 			       CTL_DEBUG, i, CTL_EOL);
892 		sysctl_createv(clog, 0, NULL, NULL,
893 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
894 			       CTLTYPE_STRING, "name", NULL,
895 			       /*XXXUNCONST*/
896 			       NULL, 0, __UNCONST(cdp->debugname), 0,
897 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
898 		sysctl_createv(clog, 0, NULL, NULL,
899 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
900 			       CTLTYPE_INT, "value", NULL,
901 			       NULL, 0, cdp->debugvar, 0,
902 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
903 		sysctl_createv(clog, 0, NULL, NULL,
904 			       CTLFLAG_PERMANENT,
905 			       CTLTYPE_INT, cdp->debugname, NULL,
906 			       NULL, 0, cdp->debugvar, 0,
907 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
908 	}
909 }
910 #endif /* DEBUG */
911 
912 /*
913  * ********************************************************************
914  * section 2: private node-specific helper routines.
915  * ********************************************************************
916  */
917 
918 #ifdef DIAGNOSTIC
919 static int
920 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
921 {
922 	int newtrig, error;
923 	struct sysctlnode node;
924 
925 	newtrig = 0;
926 	node = *rnode;
927 	node.sysctl_data = &newtrig;
928 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
929 	if (error || newp == NULL)
930 		return (error);
931 
932 	if (newtrig != 0)
933 		panic("Panic triggered");
934 
935 	return (error);
936 }
937 #endif
938 
939 /*
940  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
941  * new value is lower than desiredvnodes and then calls reinit
942  * routines that needs to adjust to the new value.
943  */
944 static int
945 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
946 {
947 	int error, new_vnodes, old_vnodes, new_max;
948 	struct sysctlnode node;
949 
950 	new_vnodes = desiredvnodes;
951 	node = *rnode;
952 	node.sysctl_data = &new_vnodes;
953 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
954 	if (error || newp == NULL)
955 		return (error);
956 
957 	/*
958 	 * sysctl passes down unsigned values, require them
959 	 * to be positive
960 	 */
961 	if (new_vnodes <= 0)
962 		return (EINVAL);
963 
964 	/* Limits: 75% of kmem and physical memory. */
965 	new_max = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC),
966 	    75, 75) / VNODE_COST;
967 	if (new_vnodes > new_max)
968 		new_vnodes = new_max;
969 
970 	old_vnodes = desiredvnodes;
971 	desiredvnodes = new_vnodes;
972 	if (new_vnodes < old_vnodes) {
973 		error = vfs_drainvnodes(new_vnodes);
974 		if (error) {
975 			desiredvnodes = old_vnodes;
976 			return (error);
977 		}
978 	}
979 	vfs_reinit();
980 	nchreinit();
981 
982 	return (0);
983 }
984 
985 /*
986  * sysctl helper routine for rtc_offset - set time after changes
987  */
988 static int
989 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
990 {
991 	struct timespec ts, delta;
992 	int error, new_rtc_offset;
993 	struct sysctlnode node;
994 
995 	new_rtc_offset = rtc_offset;
996 	node = *rnode;
997 	node.sysctl_data = &new_rtc_offset;
998 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
999 	if (error || newp == NULL)
1000 		return (error);
1001 
1002 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
1003 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
1004 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
1005 		return (EPERM);
1006 	if (rtc_offset == new_rtc_offset)
1007 		return (0);
1008 
1009 	/* if we change the offset, adjust the time */
1010 	nanotime(&ts);
1011 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
1012 	delta.tv_nsec = 0;
1013 	timespecadd(&ts, &delta, &ts);
1014 	rtc_offset = new_rtc_offset;
1015 	return (settime(l->l_proc, &ts));
1016 }
1017 
1018 /*
1019  * sysctl helper routine for kern.maxproc. Ensures that the new
1020  * values are not too low or too high.
1021  */
1022 static int
1023 sysctl_kern_maxproc(SYSCTLFN_ARGS)
1024 {
1025 	int error, nmaxproc;
1026 	struct sysctlnode node;
1027 
1028 	nmaxproc = maxproc;
1029 	node = *rnode;
1030 	node.sysctl_data = &nmaxproc;
1031 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1032 	if (error || newp == NULL)
1033 		return (error);
1034 
1035 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
1036 		return (EINVAL);
1037 #ifdef __HAVE_CPU_MAXPROC
1038 	if (nmaxproc > cpu_maxproc())
1039 		return (EINVAL);
1040 #endif
1041 	maxproc = nmaxproc;
1042 
1043 	return (0);
1044 }
1045 
1046 /*
1047  * sysctl helper function for kern.hostid. The hostid is a long, but
1048  * we export it as an int, so we need to give it a little help.
1049  */
1050 static int
1051 sysctl_kern_hostid(SYSCTLFN_ARGS)
1052 {
1053 	int error, inthostid;
1054 	struct sysctlnode node;
1055 
1056 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
1057 	node = *rnode;
1058 	node.sysctl_data = &inthostid;
1059 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1060 	if (error || newp == NULL)
1061 		return (error);
1062 
1063 	hostid = (unsigned)inthostid;
1064 
1065 	return (0);
1066 }
1067 
1068 /*
1069  * sysctl helper function for kern.hostname and kern.domainnname.
1070  * resets the relevant recorded length when the underlying name is
1071  * changed.
1072  */
1073 static int
1074 sysctl_setlen(SYSCTLFN_ARGS)
1075 {
1076 	int error;
1077 
1078 	error = sysctl_lookup(SYSCTLFN_CALL(rnode));
1079 	if (error || newp == NULL)
1080 		return (error);
1081 
1082 	switch (rnode->sysctl_num) {
1083 	case KERN_HOSTNAME:
1084 		hostnamelen = strlen((const char*)rnode->sysctl_data);
1085 		break;
1086 	case KERN_DOMAINNAME:
1087 		domainnamelen = strlen((const char*)rnode->sysctl_data);
1088 		break;
1089 	}
1090 
1091 	return (0);
1092 }
1093 
1094 /*
1095  * sysctl helper routine for kern.clockrate. Assembles a struct on
1096  * the fly to be returned to the caller.
1097  */
1098 static int
1099 sysctl_kern_clockrate(SYSCTLFN_ARGS)
1100 {
1101 	struct clockinfo clkinfo;
1102 	struct sysctlnode node;
1103 
1104 	clkinfo.tick = tick;
1105 	clkinfo.tickadj = tickadj;
1106 	clkinfo.hz = hz;
1107 	clkinfo.profhz = profhz;
1108 	clkinfo.stathz = stathz ? stathz : hz;
1109 
1110 	node = *rnode;
1111 	node.sysctl_data = &clkinfo;
1112 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1113 }
1114 
1115 /*
1116  * sysctl helper routine for kern.msgbufsize and kern.msgbuf. For the
1117  * former it merely checks the message buffer is set up. For the latter,
1118  * it also copies out the data if necessary.
1119  */
1120 static int
1121 sysctl_msgbuf(SYSCTLFN_ARGS)
1122 {
1123 	char *where = oldp;
1124 	size_t len, maxlen;
1125 	long beg, end;
1126 	extern kmutex_t log_lock;
1127 	int error;
1128 
1129 	if (!msgbufenabled || msgbufp->msg_magic != MSG_MAGIC) {
1130 		msgbufenabled = 0;
1131 		return (ENXIO);
1132 	}
1133 
1134 	switch (rnode->sysctl_num) {
1135 	case KERN_MSGBUFSIZE: {
1136 		struct sysctlnode node = *rnode;
1137 		int msg_bufs = (int)msgbufp->msg_bufs;
1138 		node.sysctl_data = &msg_bufs;
1139 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1140 	}
1141 	case KERN_MSGBUF:
1142 		break;
1143 	default:
1144 		return (EOPNOTSUPP);
1145 	}
1146 
1147 	if (newp != NULL)
1148 		return (EPERM);
1149 
1150 	if (oldp == NULL) {
1151 		/* always return full buffer size */
1152 		*oldlenp = msgbufp->msg_bufs;
1153 		return (0);
1154 	}
1155 
1156 	sysctl_unlock();
1157 
1158 	/*
1159 	 * First, copy from the write pointer to the end of
1160 	 * message buffer.
1161 	 */
1162 	error = 0;
1163 	mutex_spin_enter(&log_lock);
1164 	maxlen = MIN(msgbufp->msg_bufs, *oldlenp);
1165 	beg = msgbufp->msg_bufx;
1166 	end = msgbufp->msg_bufs;
1167 	mutex_spin_exit(&log_lock);
1168 
1169 	while (maxlen > 0) {
1170 		len = MIN(end - beg, maxlen);
1171 		if (len == 0)
1172 			break;
1173 		/* XXX unlocked, but hardly matters. */
1174 		error = dcopyout(l, &msgbufp->msg_bufc[beg], where, len);
1175 		if (error)
1176 			break;
1177 		where += len;
1178 		maxlen -= len;
1179 
1180 		/*
1181 		 * ... then, copy from the beginning of message buffer to
1182 		 * the write pointer.
1183 		 */
1184 		beg = 0;
1185 		end = msgbufp->msg_bufx;
1186 	}
1187 
1188 	sysctl_relock();
1189 	return (error);
1190 }
1191 
1192 /*
1193  * sysctl helper routine for kern.defcorename. In the case of a new
1194  * string being assigned, check that it's not a zero-length string.
1195  * (XXX the check in -current doesn't work, but do we really care?)
1196  */
1197 static int
1198 sysctl_kern_defcorename(SYSCTLFN_ARGS)
1199 {
1200 	int error;
1201 	char *newcorename;
1202 	struct sysctlnode node;
1203 
1204 	newcorename = PNBUF_GET();
1205 	node = *rnode;
1206 	node.sysctl_data = &newcorename[0];
1207 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1208 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1209 	if (error || newp == NULL) {
1210 		goto done;
1211 	}
1212 
1213 	/*
1214 	 * when sysctl_lookup() deals with a string, it's guaranteed
1215 	 * to come back nul terminated. So there.  :)
1216 	 */
1217 	if (strlen(newcorename) == 0) {
1218 		error = EINVAL;
1219 	} else {
1220 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1221 		error = 0;
1222 	}
1223 done:
1224 	PNBUF_PUT(newcorename);
1225 	return error;
1226 }
1227 
1228 /*
1229  * sysctl helper routine for kern.cp_time node. Adds up cpu time
1230  * across all cpus.
1231  */
1232 static int
1233 sysctl_kern_cptime(SYSCTLFN_ARGS)
1234 {
1235 	struct sysctlnode node = *rnode;
1236 	uint64_t *cp_time = NULL;
1237 	int error, n = ncpu, i;
1238 	struct cpu_info *ci;
1239 	CPU_INFO_ITERATOR cii;
1240 
1241 	/*
1242 	 * if you specifically pass a buffer that is the size of the
1243 	 * sum, or if you are probing for the size, you get the "sum"
1244 	 * of cp_time (and the size thereof) across all processors.
1245 	 *
1246 	 * alternately, you can pass an additional mib number and get
1247 	 * cp_time for that particular processor.
1248 	 */
1249 	switch (namelen) {
1250 	case 0:
1251 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
1252 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1253 			n = -1; /* SUM */
1254 		}
1255 		else {
1256 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
1257 			n = -2; /* ALL */
1258 		}
1259 		break;
1260 	case 1:
1261 		if (name[0] < 0 || name[0] >= n)
1262 			return (ENOENT); /* ENOSUCHPROCESSOR */
1263 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
1264 		n = name[0];
1265 		/*
1266 		 * adjust these so that sysctl_lookup() will be happy
1267 		 */
1268 		name++;
1269 		namelen--;
1270 		break;
1271 	default:
1272 		return (EINVAL);
1273 	}
1274 
1275 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
1276 	if (cp_time == NULL)
1277 		return (ENOMEM);
1278 	node.sysctl_data = cp_time;
1279 	memset(cp_time, 0, node.sysctl_size);
1280 
1281 	for (CPU_INFO_FOREACH(cii, ci)) {
1282 		if (n <= 0) {
1283 			for (i = 0; i < CPUSTATES; i++) {
1284 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1285 			}
1286 		}
1287 		/*
1288 		 * if a specific processor was requested and we just
1289 		 * did it, we're done here
1290 		 */
1291 		if (n == 0)
1292 			break;
1293 		/*
1294 		 * if doing "all", skip to next cp_time set for next processor
1295 		 */
1296 		if (n == -2)
1297 			cp_time += CPUSTATES;
1298 		/*
1299 		 * if we're doing a specific processor, we're one
1300 		 * processor closer
1301 		 */
1302 		if (n > 0)
1303 			n--;
1304 	}
1305 
1306 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1307 	kmem_free(node.sysctl_data, node.sysctl_size);
1308 	return (error);
1309 }
1310 
1311 #if NPTY > 0
1312 /*
1313  * sysctl helper routine for kern.maxptys. Ensures that any new value
1314  * is acceptable to the pty subsystem.
1315  */
1316 static int
1317 sysctl_kern_maxptys(SYSCTLFN_ARGS)
1318 {
1319 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1320 	int error, xmax;
1321 	struct sysctlnode node;
1322 
1323 	/* get current value of maxptys */
1324 	xmax = pty_maxptys(0, 0);
1325 
1326 	node = *rnode;
1327 	node.sysctl_data = &xmax;
1328 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1329 	if (error || newp == NULL)
1330 		return (error);
1331 
1332 	if (xmax != pty_maxptys(xmax, 1))
1333 		return (EINVAL);
1334 
1335 	return (0);
1336 }
1337 #endif /* NPTY > 0 */
1338 
1339 /*
1340  * sysctl helper routine for kern.urandom node. Picks a random number
1341  * for you.
1342  */
1343 static int
1344 sysctl_kern_urnd(SYSCTLFN_ARGS)
1345 {
1346 	int v, rv;
1347 
1348 	rv = cprng_strong(sysctl_prng, &v, sizeof(v), 0);
1349 	if (rv == sizeof(v)) {
1350 		struct sysctlnode node = *rnode;
1351 		node.sysctl_data = &v;
1352 		return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1353 	}
1354 	else
1355 		return (EIO);	/*XXX*/
1356 }
1357 
1358 /*
1359  * sysctl helper routine for kern.arandom node. Picks a random number
1360  * for you.
1361  */
1362 static int
1363 sysctl_kern_arnd(SYSCTLFN_ARGS)
1364 {
1365 	int error;
1366 	void *v;
1367 	struct sysctlnode node = *rnode;
1368 
1369 	if (*oldlenp == 0)
1370 		return 0;
1371 	/*
1372 	 * This code used to allow sucking 8192 bytes at a time out
1373 	 * of the kernel arc4random generator.  Evidently there is some
1374 	 * very old OpenBSD application code that may try to do this.
1375 	 *
1376 	 * Note that this node is documented as type "INT" -- 4 or 8
1377 	 * bytes, not 8192.
1378 	 *
1379 	 * We continue to support this abuse of the "len" pointer here
1380 	 * but only 256 bytes at a time, as, anecdotally, the actual
1381 	 * application use here was to generate RC4 keys in userspace.
1382 	 *
1383 	 * Support for such large requests will probably be removed
1384 	 * entirely in the future.
1385 	 */
1386 	if (*oldlenp > 256)
1387 		return E2BIG;
1388 
1389 	v = kmem_alloc(*oldlenp, KM_SLEEP);
1390 	cprng_fast(v, *oldlenp);
1391 	node.sysctl_data = v;
1392 	node.sysctl_size = *oldlenp;
1393 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1394 	kmem_free(v, *oldlenp);
1395 	return error;
1396 }
1397 /*
1398  * sysctl helper routine to do kern.lwp.* work.
1399  */
1400 static int
1401 sysctl_kern_lwp(SYSCTLFN_ARGS)
1402 {
1403 	struct kinfo_lwp klwp;
1404 	struct proc *p;
1405 	struct lwp *l2, *l3;
1406 	char *where, *dp;
1407 	int pid, elem_size, elem_count;
1408 	int buflen, needed, error;
1409 	bool gotit;
1410 
1411 	if (namelen == 1 && name[0] == CTL_QUERY)
1412 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1413 
1414 	dp = where = oldp;
1415 	buflen = where != NULL ? *oldlenp : 0;
1416 	error = needed = 0;
1417 
1418 	if (newp != NULL || namelen != 3)
1419 		return (EINVAL);
1420 	pid = name[0];
1421 	elem_size = name[1];
1422 	elem_count = name[2];
1423 
1424 	sysctl_unlock();
1425 	if (pid == -1) {
1426 		mutex_enter(proc_lock);
1427 		PROCLIST_FOREACH(p, &allproc) {
1428 			/* Grab a hold on the process. */
1429 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1430 				continue;
1431 			}
1432 			mutex_exit(proc_lock);
1433 
1434 			mutex_enter(p->p_lock);
1435 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1436 				if (buflen >= elem_size && elem_count > 0) {
1437 					lwp_lock(l2);
1438 					fill_lwp(l2, &klwp);
1439 					lwp_unlock(l2);
1440 					mutex_exit(p->p_lock);
1441 
1442 					/*
1443 					 * Copy out elem_size, but not
1444 					 * larger than the size of a
1445 					 * struct kinfo_proc2.
1446 					 */
1447 					error = dcopyout(l, &klwp, dp,
1448 					    min(sizeof(klwp), elem_size));
1449 					if (error) {
1450 						rw_exit(&p->p_reflock);
1451 						goto cleanup;
1452 					}
1453 					mutex_enter(p->p_lock);
1454 					LIST_FOREACH(l3, &p->p_lwps,
1455 					    l_sibling) {
1456 						if (l2 == l3)
1457 							break;
1458 					}
1459 					if (l3 == NULL) {
1460 						mutex_exit(p->p_lock);
1461 						rw_exit(&p->p_reflock);
1462 						error = EAGAIN;
1463 						goto cleanup;
1464 					}
1465 					dp += elem_size;
1466 					buflen -= elem_size;
1467 					elem_count--;
1468 				}
1469 				needed += elem_size;
1470 			}
1471 			mutex_exit(p->p_lock);
1472 
1473 			/* Drop reference to process. */
1474 			mutex_enter(proc_lock);
1475 			rw_exit(&p->p_reflock);
1476 		}
1477 		mutex_exit(proc_lock);
1478 	} else {
1479 		mutex_enter(proc_lock);
1480 		p = proc_find(pid);
1481 		if (p == NULL) {
1482 			error = ESRCH;
1483 			mutex_exit(proc_lock);
1484 			goto cleanup;
1485 		}
1486 		/* Grab a hold on the process. */
1487 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1488 		mutex_exit(proc_lock);
1489 		if (!gotit) {
1490 			error = ESRCH;
1491 			goto cleanup;
1492 		}
1493 
1494 		mutex_enter(p->p_lock);
1495 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1496 			if (buflen >= elem_size && elem_count > 0) {
1497 				lwp_lock(l2);
1498 				fill_lwp(l2, &klwp);
1499 				lwp_unlock(l2);
1500 				mutex_exit(p->p_lock);
1501 				/*
1502 				 * Copy out elem_size, but not larger than
1503 				 * the size of a struct kinfo_proc2.
1504 				 */
1505 				error = dcopyout(l, &klwp, dp,
1506 				    min(sizeof(klwp), elem_size));
1507 				if (error) {
1508 					rw_exit(&p->p_reflock);
1509 					goto cleanup;
1510 				}
1511 				mutex_enter(p->p_lock);
1512 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1513 					if (l2 == l3)
1514 						break;
1515 				}
1516 				if (l3 == NULL) {
1517 					mutex_exit(p->p_lock);
1518 					rw_exit(&p->p_reflock);
1519 					error = EAGAIN;
1520 					goto cleanup;
1521 				}
1522 				dp += elem_size;
1523 				buflen -= elem_size;
1524 				elem_count--;
1525 			}
1526 			needed += elem_size;
1527 		}
1528 		mutex_exit(p->p_lock);
1529 
1530 		/* Drop reference to process. */
1531 		rw_exit(&p->p_reflock);
1532 	}
1533 
1534 	if (where != NULL) {
1535 		*oldlenp = dp - where;
1536 		if (needed > *oldlenp) {
1537 			sysctl_relock();
1538 			return (ENOMEM);
1539 		}
1540 	} else {
1541 		needed += KERN_LWPSLOP;
1542 		*oldlenp = needed;
1543 	}
1544 	error = 0;
1545  cleanup:
1546 	sysctl_relock();
1547 	return (error);
1548 }
1549 
1550 /*
1551  * sysctl helper routine for kern.forkfsleep node. Ensures that the
1552  * given value is not too large or two small, and is at least one
1553  * timer tick if not zero.
1554  */
1555 static int
1556 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1557 {
1558 	/* userland sees value in ms, internally is in ticks */
1559 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1560 	int error, timo, lsleep;
1561 	struct sysctlnode node;
1562 
1563 	lsleep = forkfsleep * 1000 / hz;
1564 	node = *rnode;
1565 	node.sysctl_data = &lsleep;
1566 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1567 	if (error || newp == NULL)
1568 		return (error);
1569 
1570 	/* refuse negative values, and overly 'long time' */
1571 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1572 		return (EINVAL);
1573 
1574 	timo = mstohz(lsleep);
1575 
1576 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1577 	if (lsleep != 0 && timo == 0)
1578 		forkfsleep = 1;
1579 	else
1580 		forkfsleep = timo;
1581 
1582 	return (0);
1583 }
1584 
1585 /*
1586  * sysctl helper routine for kern.root_partition
1587  */
1588 static int
1589 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1590 {
1591 	int rootpart = DISKPART(rootdev);
1592 	struct sysctlnode node = *rnode;
1593 
1594 	node.sysctl_data = &rootpart;
1595 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1596 }
1597 
1598 /*
1599  * sysctl helper function for kern.drivers
1600  */
1601 static int
1602 sysctl_kern_drivers(SYSCTLFN_ARGS)
1603 {
1604 	int error;
1605 	size_t buflen;
1606 	struct kinfo_drivers kd;
1607 	char *start, *where;
1608 	const char *dname;
1609 	int i;
1610 	extern struct devsw_conv *devsw_conv;
1611 	extern int max_devsw_convs;
1612 
1613 	start = where = oldp;
1614 	buflen = *oldlenp;
1615 	if (where == NULL) {
1616 		*oldlenp = max_devsw_convs * sizeof kd;
1617 		return 0;
1618 	}
1619 
1620 	/*
1621 	 * An array of kinfo_drivers structures
1622 	 */
1623 	error = 0;
1624 	sysctl_unlock();
1625 	mutex_enter(&device_lock);
1626 	for (i = 0; i < max_devsw_convs; i++) {
1627 		dname = devsw_conv[i].d_name;
1628 		if (dname == NULL)
1629 			continue;
1630 		if (buflen < sizeof kd) {
1631 			error = ENOMEM;
1632 			break;
1633 		}
1634 		memset(&kd, 0, sizeof(kd));
1635 		kd.d_bmajor = devsw_conv[i].d_bmajor;
1636 		kd.d_cmajor = devsw_conv[i].d_cmajor;
1637 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1638 		mutex_exit(&device_lock);
1639 		error = dcopyout(l, &kd, where, sizeof kd);
1640 		mutex_enter(&device_lock);
1641 		if (error != 0)
1642 			break;
1643 		buflen -= sizeof kd;
1644 		where += sizeof kd;
1645 	}
1646 	mutex_exit(&device_lock);
1647 	sysctl_relock();
1648 	*oldlenp = where - start;
1649 	return error;
1650 }
1651 
1652 static int
1653 sysctl_security_setidcore(SYSCTLFN_ARGS)
1654 {
1655 	int newsize, error;
1656 	struct sysctlnode node;
1657 
1658 	node = *rnode;
1659 	node.sysctl_data = &newsize;
1660 	newsize = *(int *)rnode->sysctl_data;
1661 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1662 	if (error || newp == NULL)
1663 		return error;
1664 
1665 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1666 	    0, NULL, NULL, NULL))
1667 		return (EPERM);
1668 
1669 	*(int *)rnode->sysctl_data = newsize;
1670 
1671 	return 0;
1672 }
1673 
1674 static int
1675 sysctl_security_setidcorename(SYSCTLFN_ARGS)
1676 {
1677 	int error;
1678 	char *newsetidcorename;
1679 	struct sysctlnode node;
1680 
1681 	newsetidcorename = PNBUF_GET();
1682 	node = *rnode;
1683 	node.sysctl_data = newsetidcorename;
1684 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1685 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1686 	if (error || newp == NULL) {
1687 		goto out;
1688 	}
1689 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1690 	    0, NULL, NULL, NULL)) {
1691 		error = EPERM;
1692 		goto out;
1693 	}
1694 	if (strlen(newsetidcorename) == 0) {
1695 		error = EINVAL;
1696 		goto out;
1697 	}
1698 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1699 out:
1700 	PNBUF_PUT(newsetidcorename);
1701 	return error;
1702 }
1703 
1704 /*
1705  * sysctl helper routine for kern.cp_id node. Maps cpus to their
1706  * cpuids.
1707  */
1708 static int
1709 sysctl_kern_cpid(SYSCTLFN_ARGS)
1710 {
1711 	struct sysctlnode node = *rnode;
1712 	uint64_t *cp_id = NULL;
1713 	int error, n = ncpu;
1714 	struct cpu_info *ci;
1715 	CPU_INFO_ITERATOR cii;
1716 
1717 	/*
1718 	 * Here you may either retrieve a single cpu id or the whole
1719 	 * set. The size you get back when probing depends on what
1720 	 * you ask for.
1721 	 */
1722 	switch (namelen) {
1723 	case 0:
1724 		node.sysctl_size = n * sizeof(uint64_t);
1725 		n = -2; /* ALL */
1726 		break;
1727 	case 1:
1728 		if (name[0] < 0 || name[0] >= n)
1729 			return (ENOENT); /* ENOSUCHPROCESSOR */
1730 		node.sysctl_size = sizeof(uint64_t);
1731 		n = name[0];
1732 		/*
1733 		 * adjust these so that sysctl_lookup() will be happy
1734 		 */
1735 		name++;
1736 		namelen--;
1737 		break;
1738 	default:
1739 		return (EINVAL);
1740 	}
1741 
1742 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
1743 	if (cp_id == NULL)
1744 		return (ENOMEM);
1745 	node.sysctl_data = cp_id;
1746 	memset(cp_id, 0, node.sysctl_size);
1747 
1748 	for (CPU_INFO_FOREACH(cii, ci)) {
1749 		if (n <= 0)
1750 			cp_id[0] = cpu_index(ci);
1751 		/*
1752 		 * if a specific processor was requested and we just
1753 		 * did it, we're done here
1754 		 */
1755 		if (n == 0)
1756 			break;
1757 		/*
1758 		 * if doing "all", skip to next cp_id slot for next processor
1759 		 */
1760 		if (n == -2)
1761 			cp_id++;
1762 		/*
1763 		 * if we're doing a specific processor, we're one
1764 		 * processor closer
1765 		 */
1766 		if (n > 0)
1767 			n--;
1768 	}
1769 
1770 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1771 	kmem_free(node.sysctl_data, node.sysctl_size);
1772 	return (error);
1773 }
1774 
1775 /*
1776  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
1777  * calculate on the fly taking into account integer overflow and the
1778  * current wired count.
1779  */
1780 static int
1781 sysctl_hw_usermem(SYSCTLFN_ARGS)
1782 {
1783 	u_int ui;
1784 	u_quad_t uq;
1785 	struct sysctlnode node;
1786 
1787 	node = *rnode;
1788 	switch (rnode->sysctl_num) {
1789 	case HW_USERMEM:
1790 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
1791 			ui = UINT_MAX;
1792 		else
1793 			ui *= PAGE_SIZE;
1794 		node.sysctl_data = &ui;
1795 		break;
1796 	case HW_USERMEM64:
1797 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
1798 		node.sysctl_data = &uq;
1799 		break;
1800 	default:
1801 		return (EINVAL);
1802 	}
1803 
1804 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1805 }
1806 
1807 /*
1808  * sysctl helper routine for kern.cnmagic node. Pulls the old value
1809  * out, encoded, and stuffs the new value in for decoding.
1810  */
1811 static int
1812 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
1813 {
1814 	char magic[CNS_LEN];
1815 	int error;
1816 	struct sysctlnode node;
1817 
1818 	if (oldp)
1819 		cn_get_magic(magic, CNS_LEN);
1820 	node = *rnode;
1821 	node.sysctl_data = &magic[0];
1822 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1823 	if (error || newp == NULL)
1824 		return (error);
1825 
1826 	return (cn_set_magic(magic));
1827 }
1828 
1829 /*
1830  * ********************************************************************
1831  * section 3: public helper routines that are used for more than one
1832  * node
1833  * ********************************************************************
1834  */
1835 
1836 /*
1837  * sysctl helper routine for the kern.root_device node and some ports'
1838  * machdep.root_device nodes.
1839  */
1840 int
1841 sysctl_root_device(SYSCTLFN_ARGS)
1842 {
1843 	struct sysctlnode node;
1844 
1845 	node = *rnode;
1846 	node.sysctl_data = __UNCONST(device_xname(root_device));
1847 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
1848 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1849 }
1850 
1851 /*
1852  * sysctl helper routine for kern.consdev, dependent on the current
1853  * state of the console. Also used for machdep.console_device on some
1854  * ports.
1855  */
1856 int
1857 sysctl_consdev(SYSCTLFN_ARGS)
1858 {
1859 	dev_t consdev;
1860 	uint32_t oconsdev;
1861 	struct sysctlnode node;
1862 
1863 	if (cn_tab != NULL)
1864 		consdev = cn_tab->cn_dev;
1865 	else
1866 		consdev = NODEV;
1867 	node = *rnode;
1868 	switch (*oldlenp) {
1869 	case sizeof(consdev):
1870 		node.sysctl_data = &consdev;
1871 		node.sysctl_size = sizeof(consdev);
1872 		break;
1873 	case sizeof(oconsdev):
1874 		oconsdev = (uint32_t)consdev;
1875 		node.sysctl_data = &oconsdev;
1876 		node.sysctl_size = sizeof(oconsdev);
1877 		break;
1878 	default:
1879 		return EINVAL;
1880 	}
1881 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1882 }
1883 
1884 /*
1885  * ********************************************************************
1886  * section 4: support for some helpers
1887  * ********************************************************************
1888  */
1889 
1890 
1891 /*
1892  * Fill in a kinfo_lwp structure for the specified lwp.
1893  */
1894 static void
1895 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
1896 {
1897 	struct proc *p = l->l_proc;
1898 	struct timeval tv;
1899 
1900 	KASSERT(lwp_locked(l, NULL));
1901 
1902 	memset(kl, 0, sizeof(*kl));
1903 
1904 	kl->l_forw = 0;
1905 	kl->l_back = 0;
1906 	kl->l_laddr = PTRTOUINT64(l);
1907 	kl->l_addr = PTRTOUINT64(l->l_addr);
1908 	kl->l_stat = l->l_stat;
1909 	kl->l_lid = l->l_lid;
1910 	kl->l_flag = L_INMEM;
1911 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
1912 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
1913 
1914 	kl->l_swtime = l->l_swtime;
1915 	kl->l_slptime = l->l_slptime;
1916 	if (l->l_stat == LSONPROC)
1917 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1918 	else
1919 		kl->l_schedflags = 0;
1920 	kl->l_priority = lwp_eprio(l);
1921 	kl->l_usrpri = l->l_priority;
1922 	if (l->l_wchan)
1923 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
1924 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
1925 	kl->l_cpuid = cpu_index(l->l_cpu);
1926 	bintime2timeval(&l->l_rtime, &tv);
1927 	kl->l_rtime_sec = tv.tv_sec;
1928 	kl->l_rtime_usec = tv.tv_usec;
1929 	kl->l_cpticks = l->l_cpticks;
1930 	kl->l_pctcpu = l->l_pctcpu;
1931 	kl->l_pid = p->p_pid;
1932 	if (l->l_name == NULL)
1933 		kl->l_name[0] = '\0';
1934 	else
1935 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
1936 }
1937