xref: /netbsd-src/sys/kern/init_sysctl.c (revision 796c32c94f6e154afc9de0f63da35c91bb739b45)
1 /*	$NetBSD: init_sysctl.c,v 1.213 2017/06/01 02:45:13 chs Exp $ */
2 
3 /*-
4  * Copyright (c) 2003, 2007, 2008, 2009 The NetBSD Foundation, Inc.
5  * All rights reserved.
6  *
7  * This code is derived from software contributed to The NetBSD Foundation
8  * by Andrew Brown, and by Andrew Doran.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29  * POSSIBILITY OF SUCH DAMAGE.
30  */
31 
32 #include <sys/cdefs.h>
33 __KERNEL_RCSID(0, "$NetBSD: init_sysctl.c,v 1.213 2017/06/01 02:45:13 chs Exp $");
34 
35 #include "opt_sysv.h"
36 #include "opt_compat_netbsd.h"
37 #include "opt_modular.h"
38 #include "pty.h"
39 
40 #include <sys/types.h>
41 #include <sys/param.h>
42 #include <sys/sysctl.h>
43 #include <sys/cpu.h>
44 #include <sys/errno.h>
45 #include <sys/systm.h>
46 #include <sys/kernel.h>
47 #include <sys/unistd.h>
48 #include <sys/disklabel.h>
49 #include <sys/cprng.h>
50 #include <sys/vnode_impl.h>	/* For vfs_drainvnodes(). */
51 #include <sys/mount.h>
52 #include <sys/namei.h>
53 #include <dev/cons.h>
54 #include <sys/socketvar.h>
55 #include <sys/file.h>
56 #include <sys/filedesc.h>
57 #include <sys/tty.h>
58 #include <sys/kmem.h>
59 #include <sys/reboot.h>
60 #include <sys/resource.h>
61 #include <sys/resourcevar.h>
62 #include <sys/exec.h>
63 #include <sys/conf.h>
64 #include <sys/device.h>
65 #include <sys/stat.h>
66 #include <sys/kauth.h>
67 #include <sys/ktrace.h>
68 
69 #include <sys/cpu.h>
70 
71 int security_setidcore_dump;
72 char security_setidcore_path[MAXPATHLEN] = "/var/crash/%n.core";
73 uid_t security_setidcore_owner = 0;
74 gid_t security_setidcore_group = 0;
75 mode_t security_setidcore_mode = (S_IRUSR|S_IWUSR);
76 
77 /*
78  * Current status of SysV IPC capability.  Initially, these are
79  * 0 if the capability is not built-in to the kernel, but can
80  * be updated if the appropriate kernel module is (auto)loaded.
81  */
82 
83 int kern_has_sysvmsg = 0;
84 int kern_has_sysvshm = 0;
85 int kern_has_sysvsem = 0;
86 
87 static const u_int sysctl_lwpprflagmap[] = {
88 	LPR_DETACHED, L_DETACHED,
89 	0
90 };
91 
92 /*
93  * try over estimating by 5 procs/lwps
94  */
95 #define KERN_LWPSLOP	(5 * sizeof(struct kinfo_lwp))
96 
97 static int dcopyout(struct lwp *, const void *, void *, size_t);
98 
99 static int
100 dcopyout(struct lwp *l, const void *kaddr, void *uaddr, size_t len)
101 {
102 	int error;
103 
104 	error = copyout(kaddr, uaddr, len);
105 	ktrmibio(-1, UIO_READ, uaddr, len, error);
106 
107 	return error;
108 }
109 
110 #ifdef DIAGNOSTIC
111 static int sysctl_kern_trigger_panic(SYSCTLFN_PROTO);
112 #endif
113 static int sysctl_kern_maxvnodes(SYSCTLFN_PROTO);
114 static int sysctl_kern_messages(SYSCTLFN_PROTO);
115 static int sysctl_kern_rtc_offset(SYSCTLFN_PROTO);
116 static int sysctl_kern_maxproc(SYSCTLFN_PROTO);
117 static int sysctl_kern_hostid(SYSCTLFN_PROTO);
118 static int sysctl_kern_defcorename(SYSCTLFN_PROTO);
119 static int sysctl_kern_cptime(SYSCTLFN_PROTO);
120 #if NPTY > 0
121 static int sysctl_kern_maxptys(SYSCTLFN_PROTO);
122 #endif /* NPTY > 0 */
123 static int sysctl_kern_lwp(SYSCTLFN_PROTO);
124 static int sysctl_kern_forkfsleep(SYSCTLFN_PROTO);
125 static int sysctl_kern_root_partition(SYSCTLFN_PROTO);
126 static int sysctl_kern_drivers(SYSCTLFN_PROTO);
127 static int sysctl_security_setidcore(SYSCTLFN_PROTO);
128 static int sysctl_security_setidcorename(SYSCTLFN_PROTO);
129 static int sysctl_kern_cpid(SYSCTLFN_PROTO);
130 static int sysctl_hw_usermem(SYSCTLFN_PROTO);
131 static int sysctl_hw_cnmagic(SYSCTLFN_PROTO);
132 
133 static void fill_lwp(struct lwp *l, struct kinfo_lwp *kl);
134 
135 /*
136  * ********************************************************************
137  * section 1: setup routines
138  * ********************************************************************
139  * These functions are stuffed into a link set for sysctl setup
140  * functions. They're never called or referenced from anywhere else.
141  * ********************************************************************
142  */
143 
144 /*
145  * this setup routine is a replacement for kern_sysctl()
146  */
147 SYSCTL_SETUP(sysctl_kern_setup, "sysctl kern subtree setup")
148 {
149 	extern int kern_logsigexit;	/* defined in kern/kern_sig.c */
150 	extern fixpt_t ccpu;		/* defined in kern/kern_synch.c */
151 	extern int dumponpanic;		/* defined in kern/subr_prf.c */
152 	const struct sysctlnode *rnode;
153 
154 	sysctl_createv(clog, 0, NULL, NULL,
155 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
156 		       CTLTYPE_INT, "maxvnodes",
157 		       SYSCTL_DESCR("Maximum number of vnodes"),
158 		       sysctl_kern_maxvnodes, 0, NULL, 0,
159 		       CTL_KERN, KERN_MAXVNODES, CTL_EOL);
160 	sysctl_createv(clog, 0, NULL, NULL,
161 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
162 		       CTLTYPE_INT, "maxproc",
163 		       SYSCTL_DESCR("Maximum number of simultaneous processes"),
164 		       sysctl_kern_maxproc, 0, NULL, 0,
165 		       CTL_KERN, KERN_MAXPROC, CTL_EOL);
166 	sysctl_createv(clog, 0, NULL, NULL,
167 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
168 		       CTLTYPE_INT, "maxfiles",
169 		       SYSCTL_DESCR("Maximum number of open files"),
170 		       NULL, 0, &maxfiles, 0,
171 		       CTL_KERN, KERN_MAXFILES, CTL_EOL);
172 	sysctl_createv(clog, 0, NULL, NULL,
173 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
174 		       CTLTYPE_INT, "argmax",
175 		       SYSCTL_DESCR("Maximum number of bytes of arguments to "
176 				    "execve(2)"),
177 		       NULL, ARG_MAX, NULL, 0,
178 		       CTL_KERN, KERN_ARGMAX, CTL_EOL);
179 	sysctl_createv(clog, 0, NULL, NULL,
180 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
181 		       CTLTYPE_INT, "hostid",
182 		       SYSCTL_DESCR("System host ID number"),
183 		       sysctl_kern_hostid, 0, NULL, 0,
184 		       CTL_KERN, KERN_HOSTID, CTL_EOL);
185 	sysctl_createv(clog, 0, NULL, NULL,
186 		       CTLFLAG_PERMANENT,
187 		       CTLTYPE_STRUCT, "vnode",
188 		       SYSCTL_DESCR("System vnode table"),
189 		       sysctl_kern_vnode, 0, NULL, 0,
190 		       CTL_KERN, KERN_VNODE, CTL_EOL);
191 #ifndef GPROF
192 	sysctl_createv(clog, 0, NULL, NULL,
193 		       CTLFLAG_PERMANENT,
194 		       CTLTYPE_NODE, "profiling",
195 		       SYSCTL_DESCR("Profiling information (not available)"),
196 		       sysctl_notavail, 0, NULL, 0,
197 		       CTL_KERN, KERN_PROF, CTL_EOL);
198 #endif
199 	sysctl_createv(clog, 0, NULL, NULL,
200 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
201 		       CTLTYPE_INT, "posix1version",
202 		       SYSCTL_DESCR("Version of ISO/IEC 9945 (POSIX 1003.1) "
203 				    "with which the operating system attempts "
204 				    "to comply"),
205 		       NULL, _POSIX_VERSION, NULL, 0,
206 		       CTL_KERN, KERN_POSIX1, CTL_EOL);
207 	sysctl_createv(clog, 0, NULL, NULL,
208 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
209 		       CTLTYPE_INT, "ngroups",
210 		       SYSCTL_DESCR("Maximum number of supplemental groups"),
211 		       NULL, NGROUPS_MAX, NULL, 0,
212 		       CTL_KERN, KERN_NGROUPS, CTL_EOL);
213 	sysctl_createv(clog, 0, NULL, NULL,
214 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
215 		       CTLTYPE_INT, "job_control",
216 		       SYSCTL_DESCR("Whether job control is available"),
217 		       NULL, 1, NULL, 0,
218 		       CTL_KERN, KERN_JOB_CONTROL, CTL_EOL);
219 	sysctl_createv(clog, 0, NULL, NULL,
220 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
221 		       CTLTYPE_INT, "saved_ids",
222 		       SYSCTL_DESCR("Whether POSIX saved set-group/user ID is "
223 				    "available"), NULL,
224 #ifdef _POSIX_SAVED_IDS
225 		       1,
226 #else /* _POSIX_SAVED_IDS */
227 		       0,
228 #endif /* _POSIX_SAVED_IDS */
229 		       NULL, 0, CTL_KERN, KERN_SAVED_IDS, CTL_EOL);
230 	sysctl_createv(clog, 0, NULL, NULL,
231 		       CTLFLAG_PERMANENT|CTLFLAG_HEX,
232 		       CTLTYPE_INT, "boothowto",
233 		       SYSCTL_DESCR("Flags from boot loader"),
234 		       NULL, 0, &boothowto, sizeof(boothowto),
235 		       CTL_KERN, CTL_CREATE, CTL_EOL);
236 	sysctl_createv(clog, 0, NULL, NULL,
237 		       CTLFLAG_PERMANENT,
238 		       CTLTYPE_STRUCT, "boottime",
239 		       SYSCTL_DESCR("System boot time"),
240 		       NULL, 0, &boottime, sizeof(boottime),
241 		       CTL_KERN, KERN_BOOTTIME, CTL_EOL);
242 	sysctl_createv(clog, 0, NULL, NULL,
243 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
244 		       CTLTYPE_INT, "maxpartitions",
245 		       SYSCTL_DESCR("Maximum number of partitions allowed per "
246 				    "disk"),
247 		       NULL, MAXPARTITIONS, NULL, 0,
248 		       CTL_KERN, KERN_MAXPARTITIONS, CTL_EOL);
249 	sysctl_createv(clog, 0, NULL, NULL,
250 		       CTLFLAG_PERMANENT,
251 		       CTLTYPE_STRUCT, "timex", NULL,
252 		       sysctl_notavail, 0, NULL, 0,
253 		       CTL_KERN, KERN_TIMEX, CTL_EOL);
254 	sysctl_createv(clog, 0, NULL, NULL,
255 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
256 		       CTLTYPE_INT, "rtc_offset",
257 		       SYSCTL_DESCR("Offset of real time clock from UTC in "
258 				    "minutes"),
259 		       sysctl_kern_rtc_offset, 0, &rtc_offset, 0,
260 		       CTL_KERN, KERN_RTC_OFFSET, CTL_EOL);
261 	sysctl_createv(clog, 0, NULL, NULL,
262 		       CTLFLAG_PERMANENT,
263 		       CTLTYPE_STRING, "root_device",
264 		       SYSCTL_DESCR("Name of the root device"),
265 		       sysctl_root_device, 0, NULL, 0,
266 		       CTL_KERN, KERN_ROOT_DEVICE, CTL_EOL);
267 	sysctl_createv(clog, 0, NULL, NULL,
268 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
269 		       CTLTYPE_INT, "fsync",
270 		       SYSCTL_DESCR("Whether the POSIX 1003.1b File "
271 				    "Synchronization Option is available on "
272 				    "this system"),
273 		       NULL, 1, NULL, 0,
274 		       CTL_KERN, KERN_FSYNC, CTL_EOL);
275 	sysctl_createv(clog, 0, NULL, NULL,
276 		       CTLFLAG_PERMANENT,
277 		       CTLTYPE_NODE, "ipc",
278 		       SYSCTL_DESCR("SysV IPC options"),
279 		       NULL, 0, NULL, 0,
280 		       CTL_KERN, KERN_SYSVIPC, CTL_EOL);
281 	sysctl_createv(clog, 0, NULL, NULL,
282 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
283 		       CTLTYPE_INT, "sysvmsg",
284 		       SYSCTL_DESCR("System V style message support available"),
285 		       NULL, 0, &kern_has_sysvmsg, sizeof(int),
286 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_MSG, CTL_EOL);
287 	sysctl_createv(clog, 0, NULL, NULL,
288 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
289 		       CTLTYPE_INT, "sysvsem",
290 		       SYSCTL_DESCR("System V style semaphore support "
291 				    "available"),
292 		       NULL, 0, &kern_has_sysvsem, sizeof(int),
293 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SEM, CTL_EOL);
294 	sysctl_createv(clog, 0, NULL, NULL,
295 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
296 		       CTLTYPE_INT, "sysvshm",
297 		       SYSCTL_DESCR("System V style shared memory support "
298 				    "available"),
299 		       NULL, 0, &kern_has_sysvshm, sizeof(int),
300 		       CTL_KERN, KERN_SYSVIPC, KERN_SYSVIPC_SHM, CTL_EOL);
301 	sysctl_createv(clog, 0, NULL, NULL,
302 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
303 		       CTLTYPE_INT, "synchronized_io",
304 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Synchronized "
305 				    "I/O Option is available on this system"),
306 		       NULL, 1, NULL, 0,
307 		       CTL_KERN, KERN_SYNCHRONIZED_IO, CTL_EOL);
308 	sysctl_createv(clog, 0, NULL, NULL,
309 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
310 		       CTLTYPE_INT, "iov_max",
311 		       SYSCTL_DESCR("Maximum number of iovec structures per "
312 				    "process"),
313 		       NULL, IOV_MAX, NULL, 0,
314 		       CTL_KERN, KERN_IOV_MAX, CTL_EOL);
315 	sysctl_createv(clog, 0, NULL, NULL,
316 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
317 		       CTLTYPE_INT, "mapped_files",
318 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory Mapped "
319 				    "Files Option is available on this system"),
320 		       NULL, 1, NULL, 0,
321 		       CTL_KERN, KERN_MAPPED_FILES, CTL_EOL);
322 	sysctl_createv(clog, 0, NULL, NULL,
323 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
324 		       CTLTYPE_INT, "memlock",
325 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Process Memory "
326 				    "Locking Option is available on this "
327 				    "system"),
328 		       NULL, 1, NULL, 0,
329 		       CTL_KERN, KERN_MEMLOCK, CTL_EOL);
330 	sysctl_createv(clog, 0, NULL, NULL,
331 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
332 		       CTLTYPE_INT, "memlock_range",
333 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Range Memory "
334 				    "Locking Option is available on this "
335 				    "system"),
336 		       NULL, 1, NULL, 0,
337 		       CTL_KERN, KERN_MEMLOCK_RANGE, CTL_EOL);
338 	sysctl_createv(clog, 0, NULL, NULL,
339 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
340 		       CTLTYPE_INT, "memory_protection",
341 		       SYSCTL_DESCR("Whether the POSIX 1003.1b Memory "
342 				    "Protection Option is available on this "
343 				    "system"),
344 		       NULL, 1, NULL, 0,
345 		       CTL_KERN, KERN_MEMORY_PROTECTION, CTL_EOL);
346 	sysctl_createv(clog, 0, NULL, NULL,
347 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
348 		       CTLTYPE_INT, "login_name_max",
349 		       SYSCTL_DESCR("Maximum login name length"),
350 		       NULL, LOGIN_NAME_MAX, NULL, 0,
351 		       CTL_KERN, KERN_LOGIN_NAME_MAX, CTL_EOL);
352 	sysctl_createv(clog, 0, NULL, NULL,
353 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
354 		       CTLTYPE_STRING, "defcorename",
355 		       SYSCTL_DESCR("Default core file name"),
356 		       sysctl_kern_defcorename, 0, defcorename, MAXPATHLEN,
357 		       CTL_KERN, KERN_DEFCORENAME, CTL_EOL);
358 	sysctl_createv(clog, 0, NULL, NULL,
359 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
360 		       CTLTYPE_INT, "logsigexit",
361 		       SYSCTL_DESCR("Log process exit when caused by signals"),
362 		       NULL, 0, &kern_logsigexit, 0,
363 		       CTL_KERN, KERN_LOGSIGEXIT, CTL_EOL);
364 	sysctl_createv(clog, 0, NULL, NULL,
365 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
366 		       CTLTYPE_INT, "fscale",
367 		       SYSCTL_DESCR("Kernel fixed-point scale factor"),
368 		       NULL, FSCALE, NULL, 0,
369 		       CTL_KERN, KERN_FSCALE, CTL_EOL);
370 	sysctl_createv(clog, 0, NULL, NULL,
371 		       CTLFLAG_PERMANENT,
372 		       CTLTYPE_INT, "ccpu",
373 		       SYSCTL_DESCR("Scheduler exponential decay value"),
374 		       NULL, 0, &ccpu, 0,
375 		       CTL_KERN, KERN_CCPU, CTL_EOL);
376 	sysctl_createv(clog, 0, NULL, NULL,
377 		       CTLFLAG_PERMANENT,
378 		       CTLTYPE_STRUCT, "cp_time",
379 		       SYSCTL_DESCR("Clock ticks spent in different CPU states"),
380 		       sysctl_kern_cptime, 0, NULL, 0,
381 		       CTL_KERN, KERN_CP_TIME, CTL_EOL);
382 	sysctl_createv(clog, 0, NULL, NULL,
383 		       CTLFLAG_PERMANENT,
384 		       CTLTYPE_STRUCT, "consdev",
385 		       SYSCTL_DESCR("Console device"),
386 		       sysctl_consdev, 0, NULL, sizeof(dev_t),
387 		       CTL_KERN, KERN_CONSDEV, CTL_EOL);
388 #if NPTY > 0
389 	sysctl_createv(clog, 0, NULL, NULL,
390 		       CTLFLAG_PERMANENT,
391 		       CTLTYPE_INT, "maxptys",
392 		       SYSCTL_DESCR("Maximum number of pseudo-ttys"),
393 		       sysctl_kern_maxptys, 0, NULL, 0,
394 		       CTL_KERN, KERN_MAXPTYS, CTL_EOL);
395 #endif /* NPTY > 0 */
396 	sysctl_createv(clog, 0, NULL, NULL,
397 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
398 		       CTLTYPE_INT, "maxphys",
399 		       SYSCTL_DESCR("Maximum raw I/O transfer size"),
400 		       NULL, MAXPHYS, NULL, 0,
401 		       CTL_KERN, KERN_MAXPHYS, CTL_EOL);
402 	sysctl_createv(clog, 0, NULL, NULL,
403 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
404 		       CTLTYPE_INT, "monotonic_clock",
405 		       SYSCTL_DESCR("Implementation version of the POSIX "
406 				    "1003.1b Monotonic Clock Option"),
407 		       /* XXX _POSIX_VERSION */
408 		       NULL, _POSIX_MONOTONIC_CLOCK, NULL, 0,
409 		       CTL_KERN, KERN_MONOTONIC_CLOCK, CTL_EOL);
410 	sysctl_createv(clog, 0, NULL, NULL,
411 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
412 		       CTLTYPE_INT, "labelsector",
413 		       SYSCTL_DESCR("Sector number containing the disklabel"),
414 		       NULL, LABELSECTOR, NULL, 0,
415 		       CTL_KERN, KERN_LABELSECTOR, CTL_EOL);
416 	sysctl_createv(clog, 0, NULL, NULL,
417 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
418 		       CTLTYPE_INT, "labeloffset",
419 		       SYSCTL_DESCR("Offset of the disklabel within the "
420 				    "sector"),
421 		       NULL, LABELOFFSET, NULL, 0,
422 		       CTL_KERN, KERN_LABELOFFSET, CTL_EOL);
423 	sysctl_createv(clog, 0, NULL, NULL,
424 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
425 		       CTLTYPE_INT, "labelusesmbr",
426 		       SYSCTL_DESCR("disklabel is inside MBR partition"),
427 		       NULL, LABELUSESMBR, NULL, 0,
428 		       CTL_KERN, CTL_CREATE, CTL_EOL);
429 	sysctl_createv(clog, 0, NULL, NULL,
430 		       CTLFLAG_PERMANENT,
431 		       CTLTYPE_NODE, "lwp",
432 		       SYSCTL_DESCR("System-wide LWP information"),
433 		       sysctl_kern_lwp, 0, NULL, 0,
434 		       CTL_KERN, KERN_LWP, CTL_EOL);
435 	sysctl_createv(clog, 0, NULL, NULL,
436 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
437 		       CTLTYPE_INT, "forkfsleep",
438 		       SYSCTL_DESCR("Milliseconds to sleep on fork failure due "
439 				    "to process limits"),
440 		       sysctl_kern_forkfsleep, 0, NULL, 0,
441 		       CTL_KERN, KERN_FORKFSLEEP, CTL_EOL);
442 	sysctl_createv(clog, 0, NULL, NULL,
443 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
444 		       CTLTYPE_INT, "posix_threads",
445 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
446 				    "Threads option to which the system "
447 				    "attempts to conform"),
448 		       /* XXX _POSIX_VERSION */
449 		       NULL, _POSIX_THREADS, NULL, 0,
450 		       CTL_KERN, KERN_POSIX_THREADS, CTL_EOL);
451 	sysctl_createv(clog, 0, NULL, NULL,
452 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
453 		       CTLTYPE_INT, "posix_semaphores",
454 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
455 				    "Semaphores option to which the system "
456 				    "attempts to conform"), NULL,
457 		       200112, NULL, 0,
458 		       CTL_KERN, KERN_POSIX_SEMAPHORES, CTL_EOL);
459 	sysctl_createv(clog, 0, NULL, NULL,
460 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
461 		       CTLTYPE_INT, "posix_barriers",
462 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
463 				    "Barriers option to which the system "
464 				    "attempts to conform"),
465 		       /* XXX _POSIX_VERSION */
466 		       NULL, _POSIX_BARRIERS, NULL, 0,
467 		       CTL_KERN, KERN_POSIX_BARRIERS, CTL_EOL);
468 	sysctl_createv(clog, 0, NULL, NULL,
469 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
470 		       CTLTYPE_INT, "posix_timers",
471 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
472 				    "Timers option to which the system "
473 				    "attempts to conform"),
474 		       /* XXX _POSIX_VERSION */
475 		       NULL, _POSIX_TIMERS, NULL, 0,
476 		       CTL_KERN, KERN_POSIX_TIMERS, CTL_EOL);
477 	sysctl_createv(clog, 0, NULL, NULL,
478 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
479 		       CTLTYPE_INT, "posix_spin_locks",
480 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its Spin "
481 				    "Locks option to which the system attempts "
482 				    "to conform"),
483 		       /* XXX _POSIX_VERSION */
484 		       NULL, _POSIX_SPIN_LOCKS, NULL, 0,
485 		       CTL_KERN, KERN_POSIX_SPIN_LOCKS, CTL_EOL);
486 	sysctl_createv(clog, 0, NULL, NULL,
487 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
488 		       CTLTYPE_INT, "posix_reader_writer_locks",
489 		       SYSCTL_DESCR("Version of IEEE Std 1003.1 and its "
490 				    "Read-Write Locks option to which the "
491 				    "system attempts to conform"),
492 		       /* XXX _POSIX_VERSION */
493 		       NULL, _POSIX_READER_WRITER_LOCKS, NULL, 0,
494 		       CTL_KERN, KERN_POSIX_READER_WRITER_LOCKS, CTL_EOL);
495 	sysctl_createv(clog, 0, NULL, NULL,
496 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
497 		       CTLTYPE_INT, "dump_on_panic",
498 		       SYSCTL_DESCR("Perform a crash dump on system panic"),
499 		       NULL, 0, &dumponpanic, 0,
500 		       CTL_KERN, KERN_DUMP_ON_PANIC, CTL_EOL);
501 #ifdef DIAGNOSTIC
502 	sysctl_createv(clog, 0, NULL, NULL,
503 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
504 		       CTLTYPE_INT, "panic_now",
505 		       SYSCTL_DESCR("Trigger a panic"),
506 		       sysctl_kern_trigger_panic, 0, NULL, 0,
507 		       CTL_KERN, CTL_CREATE, CTL_EOL);
508 #endif
509 	sysctl_createv(clog, 0, NULL, NULL,
510 		       CTLFLAG_PERMANENT,
511 		       CTLTYPE_INT, "root_partition",
512 		       SYSCTL_DESCR("Root partition on the root device"),
513 		       sysctl_kern_root_partition, 0, NULL, 0,
514 		       CTL_KERN, KERN_ROOT_PARTITION, CTL_EOL);
515 	sysctl_createv(clog, 0, NULL, NULL,
516 		       CTLFLAG_PERMANENT,
517 		       CTLTYPE_STRUCT, "drivers",
518 		       SYSCTL_DESCR("List of all drivers with block and "
519 				    "character device numbers"),
520 		       sysctl_kern_drivers, 0, NULL, 0,
521 		       CTL_KERN, KERN_DRIVERS, CTL_EOL);
522 	sysctl_createv(clog, 0, NULL, NULL,
523 		       CTLFLAG_PERMANENT,
524 		       CTLTYPE_STRUCT, "cp_id",
525 		       SYSCTL_DESCR("Mapping of CPU number to CPU id"),
526 		       sysctl_kern_cpid, 0, NULL, 0,
527 		       CTL_KERN, KERN_CP_ID, CTL_EOL);
528 	sysctl_createv(clog, 0, NULL, &rnode,
529 		       CTLFLAG_PERMANENT,
530 		       CTLTYPE_NODE, "coredump",
531 		       SYSCTL_DESCR("Coredump settings."),
532 		       NULL, 0, NULL, 0,
533 		       CTL_KERN, CTL_CREATE, CTL_EOL);
534 	sysctl_createv(clog, 0, &rnode, &rnode,
535 		       CTLFLAG_PERMANENT,
536 		       CTLTYPE_NODE, "setid",
537 		       SYSCTL_DESCR("Set-id processes' coredump settings."),
538 		       NULL, 0, NULL, 0,
539 		       CTL_CREATE, CTL_EOL);
540 	sysctl_createv(clog, 0, &rnode, NULL,
541 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
542 		       CTLTYPE_INT, "dump",
543 		       SYSCTL_DESCR("Allow set-id processes to dump core."),
544 		       sysctl_security_setidcore, 0, &security_setidcore_dump,
545 		       sizeof(security_setidcore_dump),
546 		       CTL_CREATE, CTL_EOL);
547 	sysctl_createv(clog, 0, &rnode, NULL,
548 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
549 		       CTLTYPE_STRING, "path",
550 		       SYSCTL_DESCR("Path pattern for set-id coredumps."),
551 		       sysctl_security_setidcorename, 0,
552 		       security_setidcore_path,
553 		       sizeof(security_setidcore_path),
554 		       CTL_CREATE, CTL_EOL);
555 	sysctl_createv(clog, 0, &rnode, NULL,
556 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
557 		       CTLTYPE_INT, "owner",
558 		       SYSCTL_DESCR("Owner id for set-id processes' cores."),
559 		       sysctl_security_setidcore, 0, &security_setidcore_owner,
560 		       0,
561 		       CTL_CREATE, CTL_EOL);
562 	sysctl_createv(clog, 0, &rnode, NULL,
563 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
564 		       CTLTYPE_INT, "group",
565 		       SYSCTL_DESCR("Group id for set-id processes' cores."),
566 		       sysctl_security_setidcore, 0, &security_setidcore_group,
567 		       0,
568 		       CTL_CREATE, CTL_EOL);
569 	sysctl_createv(clog, 0, &rnode, NULL,
570 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
571 		       CTLTYPE_INT, "mode",
572 		       SYSCTL_DESCR("Mode for set-id processes' cores."),
573 		       sysctl_security_setidcore, 0, &security_setidcore_mode,
574 		       0,
575 		       CTL_CREATE, CTL_EOL);
576 	sysctl_createv(clog, 0, NULL, NULL,
577 		       CTLFLAG_IMMEDIATE|CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
578 		       CTLTYPE_INT, "no_sa_support",
579 		       SYSCTL_DESCR("0 if the kernel supports SA, otherwise "
580 		       "it doesn't"),
581 		       NULL, 1, NULL, 0,
582 		       CTL_KERN, CTL_CREATE, CTL_EOL);
583 	sysctl_createv(clog, 0, NULL, NULL,
584 			CTLFLAG_PERMANENT,
585 			CTLTYPE_STRING, "configname",
586 			SYSCTL_DESCR("Name of config file"),
587 			NULL, 0, __UNCONST(kernel_ident), 0,
588 			CTL_KERN, CTL_CREATE, CTL_EOL);
589 	sysctl_createv(clog, 0, NULL, NULL,
590 			CTLFLAG_PERMANENT,
591 			CTLTYPE_STRING, "buildinfo",
592 			SYSCTL_DESCR("Information from build environment"),
593 			NULL, 0, __UNCONST(buildinfo), 0,
594 			CTL_KERN, CTL_CREATE, CTL_EOL);
595 	sysctl_createv(clog, 0, NULL, NULL,
596 			CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
597 			CTLTYPE_INT, "messages",
598 			SYSCTL_DESCR("Kernel message verbosity"),
599 			sysctl_kern_messages, 0, NULL, 0,
600 			CTL_KERN, CTL_CREATE, CTL_EOL);
601 }
602 
603 SYSCTL_SETUP(sysctl_hw_misc_setup, "sysctl hw subtree misc setup")
604 {
605 
606 	sysctl_createv(clog, 0, NULL, NULL,
607 		       CTLFLAG_PERMANENT,
608 		       CTLTYPE_INT, "usermem",
609 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
610 		       sysctl_hw_usermem, 0, NULL, 0,
611 		       CTL_HW, HW_USERMEM, CTL_EOL);
612 	sysctl_createv(clog, 0, NULL, NULL,
613 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE|CTLFLAG_HEX,
614 		       CTLTYPE_STRING, "cnmagic",
615 		       SYSCTL_DESCR("Console magic key sequence"),
616 		       sysctl_hw_cnmagic, 0, NULL, CNS_LEN,
617 		       CTL_HW, HW_CNMAGIC, CTL_EOL);
618 	sysctl_createv(clog, 0, NULL, NULL,
619 		       CTLFLAG_PERMANENT,
620 		       CTLTYPE_QUAD, "usermem64",
621 		       SYSCTL_DESCR("Bytes of non-kernel memory"),
622 		       sysctl_hw_usermem, 0, NULL, 0,
623 		       CTL_HW, HW_USERMEM64, CTL_EOL);
624 }
625 
626 #ifdef DEBUG
627 /*
628  * Debugging related system variables.
629  */
630 struct ctldebug /* debug0, */ /* debug1, */ debug2, debug3, debug4;
631 struct ctldebug debug5, debug6, debug7, debug8, debug9;
632 struct ctldebug debug10, debug11, debug12, debug13, debug14;
633 struct ctldebug debug15, debug16, debug17, debug18, debug19;
634 static struct ctldebug *debugvars[CTL_DEBUG_MAXID] = {
635 	&debug0, &debug1, &debug2, &debug3, &debug4,
636 	&debug5, &debug6, &debug7, &debug8, &debug9,
637 	&debug10, &debug11, &debug12, &debug13, &debug14,
638 	&debug15, &debug16, &debug17, &debug18, &debug19,
639 };
640 
641 /*
642  * this setup routine is a replacement for debug_sysctl()
643  *
644  * note that it creates several nodes per defined debug variable
645  */
646 SYSCTL_SETUP(sysctl_debug_setup, "sysctl debug subtree setup")
647 {
648 	struct ctldebug *cdp;
649 	char nodename[20];
650 	int i;
651 
652 	/*
653 	 * two ways here:
654 	 *
655 	 * the "old" way (debug.name -> value) which was emulated by
656 	 * the sysctl(8) binary
657 	 *
658 	 * the new way, which the sysctl(8) binary was actually using
659 
660 	 node	debug
661 	 node	debug.0
662 	 string debug.0.name
663 	 int	debug.0.value
664 	 int	debug.name
665 
666 	 */
667 
668 	for (i = 0; i < CTL_DEBUG_MAXID; i++) {
669 		cdp = debugvars[i];
670 		if (cdp->debugname == NULL || cdp->debugvar == NULL)
671 			continue;
672 
673 		snprintf(nodename, sizeof(nodename), "debug%d", i);
674 		sysctl_createv(clog, 0, NULL, NULL,
675 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
676 			       CTLTYPE_NODE, nodename, NULL,
677 			       NULL, 0, NULL, 0,
678 			       CTL_DEBUG, i, CTL_EOL);
679 		sysctl_createv(clog, 0, NULL, NULL,
680 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
681 			       CTLTYPE_STRING, "name", NULL,
682 			       /*XXXUNCONST*/
683 			       NULL, 0, __UNCONST(cdp->debugname), 0,
684 			       CTL_DEBUG, i, CTL_DEBUG_NAME, CTL_EOL);
685 		sysctl_createv(clog, 0, NULL, NULL,
686 			       CTLFLAG_PERMANENT|CTLFLAG_HIDDEN,
687 			       CTLTYPE_INT, "value", NULL,
688 			       NULL, 0, cdp->debugvar, 0,
689 			       CTL_DEBUG, i, CTL_DEBUG_VALUE, CTL_EOL);
690 		sysctl_createv(clog, 0, NULL, NULL,
691 			       CTLFLAG_PERMANENT,
692 			       CTLTYPE_INT, cdp->debugname, NULL,
693 			       NULL, 0, cdp->debugvar, 0,
694 			       CTL_DEBUG, CTL_CREATE, CTL_EOL);
695 	}
696 }
697 #endif /* DEBUG */
698 
699 /*
700  * ********************************************************************
701  * section 2: private node-specific helper routines.
702  * ********************************************************************
703  */
704 
705 #ifdef DIAGNOSTIC
706 static int
707 sysctl_kern_trigger_panic(SYSCTLFN_ARGS)
708 {
709 	int newtrig, error;
710 	struct sysctlnode node;
711 
712 	newtrig = 0;
713 	node = *rnode;
714 	node.sysctl_data = &newtrig;
715 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
716 	if (error || newp == NULL)
717 		return (error);
718 
719 	if (newtrig != 0)
720 		panic("Panic triggered");
721 
722 	return (error);
723 }
724 #endif
725 
726 /*
727  * sysctl helper routine for kern.maxvnodes.  Drain vnodes if
728  * new value is lower than desiredvnodes and then calls reinit
729  * routines that needs to adjust to the new value.
730  */
731 static int
732 sysctl_kern_maxvnodes(SYSCTLFN_ARGS)
733 {
734 	int error, new_vnodes, old_vnodes, new_max;
735 	struct sysctlnode node;
736 
737 	new_vnodes = desiredvnodes;
738 	node = *rnode;
739 	node.sysctl_data = &new_vnodes;
740 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
741 	if (error || newp == NULL)
742 		return (error);
743 
744 	/*
745 	 * sysctl passes down unsigned values, require them
746 	 * to be positive
747 	 */
748 	if (new_vnodes <= 0)
749 		return (EINVAL);
750 
751 	/* Limits: 75% of kmem and physical memory. */
752 	new_max = calc_cache_size(vmem_size(kmem_arena, VMEM_FREE|VMEM_ALLOC),
753 	    75, 75) / VNODE_COST;
754 	if (new_vnodes > new_max)
755 		new_vnodes = new_max;
756 
757 	old_vnodes = desiredvnodes;
758 	desiredvnodes = new_vnodes;
759 	error = vfs_drainvnodes();
760 	if (error) {
761 		desiredvnodes = old_vnodes;
762 		return (error);
763 	}
764 	vfs_reinit();
765 	nchreinit();
766 
767 	return (0);
768 }
769 
770 /*
771  * sysctl helper routine for kern.messages.
772  * Alters boothowto to display kernel messages in increasing verbosity
773  * from 0 to 4.
774  */
775 
776 #define MAXMESSAGES            4
777 static int
778 sysctl_kern_messages(SYSCTLFN_ARGS)
779 {
780 	int error, messageverbose, messagemask, newboothowto;
781 	struct sysctlnode node;
782 
783 	messagemask = (AB_NORMAL|AB_QUIET|AB_SILENT|AB_VERBOSE|AB_DEBUG);
784 	switch (boothowto & messagemask) {
785 	case AB_SILENT:
786 		messageverbose = 0;
787 		break;
788 	case AB_QUIET:
789 		messageverbose = 1;
790 		break;
791 	case AB_VERBOSE:
792 		messageverbose = 3;
793 		break;
794 	case AB_DEBUG:
795 		messageverbose = 4;
796 		break;
797 	case AB_NORMAL:
798 	default:
799 		messageverbose = 2;
800 }
801 
802 	node = *rnode;
803 	node.sysctl_data = &messageverbose;
804 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
805 	if (error || newp == NULL)
806 		return (error);
807 	if (messageverbose < 0 || messageverbose > MAXMESSAGES)
808 		return EINVAL;
809 
810 	/* Set boothowto */
811 	newboothowto = boothowto & ~messagemask;
812 
813 	switch (messageverbose) {
814 	case 0:
815 		newboothowto |= AB_SILENT;
816 		break;
817 	case 1:
818 		newboothowto |= AB_QUIET;
819 		break;
820 	case 3:
821 		newboothowto |= AB_VERBOSE;
822 		break;
823 	case 4:
824 		newboothowto |= AB_DEBUG;
825 		break;
826 	case 2:
827 	default:                /* Messages default to normal. */
828 		break;
829 	}
830 
831 	boothowto = newboothowto;
832 
833 	return (0);
834 }
835 
836 /*
837  * sysctl helper routine for rtc_offset - set time after changes
838  */
839 static int
840 sysctl_kern_rtc_offset(SYSCTLFN_ARGS)
841 {
842 	struct timespec ts, delta;
843 	int error, new_rtc_offset;
844 	struct sysctlnode node;
845 
846 	new_rtc_offset = rtc_offset;
847 	node = *rnode;
848 	node.sysctl_data = &new_rtc_offset;
849 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
850 	if (error || newp == NULL)
851 		return (error);
852 
853 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_TIME,
854 	    KAUTH_REQ_SYSTEM_TIME_RTCOFFSET,
855 	    KAUTH_ARG(new_rtc_offset), NULL, NULL))
856 		return (EPERM);
857 	if (rtc_offset == new_rtc_offset)
858 		return (0);
859 
860 	/* if we change the offset, adjust the time */
861 	nanotime(&ts);
862 	delta.tv_sec = 60 * (new_rtc_offset - rtc_offset);
863 	delta.tv_nsec = 0;
864 	timespecadd(&ts, &delta, &ts);
865 	rtc_offset = new_rtc_offset;
866 	return (settime(l->l_proc, &ts));
867 }
868 
869 /*
870  * sysctl helper routine for kern.maxproc. Ensures that the new
871  * values are not too low or too high.
872  */
873 static int
874 sysctl_kern_maxproc(SYSCTLFN_ARGS)
875 {
876 	int error, nmaxproc;
877 	struct sysctlnode node;
878 
879 	nmaxproc = maxproc;
880 	node = *rnode;
881 	node.sysctl_data = &nmaxproc;
882 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
883 	if (error || newp == NULL)
884 		return (error);
885 
886 	if (nmaxproc < 0 || nmaxproc >= PID_MAX)
887 		return (EINVAL);
888 #ifdef __HAVE_CPU_MAXPROC
889 	if (nmaxproc > cpu_maxproc())
890 		return (EINVAL);
891 #endif
892 	maxproc = nmaxproc;
893 
894 	return (0);
895 }
896 
897 /*
898  * sysctl helper function for kern.hostid. The hostid is a long, but
899  * we export it as an int, so we need to give it a little help.
900  */
901 static int
902 sysctl_kern_hostid(SYSCTLFN_ARGS)
903 {
904 	int error, inthostid;
905 	struct sysctlnode node;
906 
907 	inthostid = hostid;  /* XXX assumes sizeof int <= sizeof long */
908 	node = *rnode;
909 	node.sysctl_data = &inthostid;
910 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
911 	if (error || newp == NULL)
912 		return (error);
913 
914 	hostid = (unsigned)inthostid;
915 
916 	return (0);
917 }
918 
919 /*
920  * sysctl helper routine for kern.defcorename. In the case of a new
921  * string being assigned, check that it's not a zero-length string.
922  * (XXX the check in -current doesn't work, but do we really care?)
923  */
924 static int
925 sysctl_kern_defcorename(SYSCTLFN_ARGS)
926 {
927 	int error;
928 	char *newcorename;
929 	struct sysctlnode node;
930 
931 	newcorename = PNBUF_GET();
932 	node = *rnode;
933 	node.sysctl_data = &newcorename[0];
934 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
935 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
936 	if (error || newp == NULL) {
937 		goto done;
938 	}
939 
940 	/*
941 	 * when sysctl_lookup() deals with a string, it's guaranteed
942 	 * to come back nul terminated. So there.  :)
943 	 */
944 	if (strlen(newcorename) == 0) {
945 		error = EINVAL;
946 	} else {
947 		memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
948 		error = 0;
949 	}
950 done:
951 	PNBUF_PUT(newcorename);
952 	return error;
953 }
954 
955 /*
956  * sysctl helper routine for kern.cp_time node. Adds up cpu time
957  * across all cpus.
958  */
959 static int
960 sysctl_kern_cptime(SYSCTLFN_ARGS)
961 {
962 	struct sysctlnode node = *rnode;
963 	uint64_t *cp_time = NULL;
964 	int error, n = ncpu, i;
965 	struct cpu_info *ci;
966 	CPU_INFO_ITERATOR cii;
967 
968 	/*
969 	 * if you specifically pass a buffer that is the size of the
970 	 * sum, or if you are probing for the size, you get the "sum"
971 	 * of cp_time (and the size thereof) across all processors.
972 	 *
973 	 * alternately, you can pass an additional mib number and get
974 	 * cp_time for that particular processor.
975 	 */
976 	switch (namelen) {
977 	case 0:
978 		if (*oldlenp == sizeof(uint64_t) * CPUSTATES || oldp == NULL) {
979 			node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
980 			n = -1; /* SUM */
981 		}
982 		else {
983 			node.sysctl_size = n * sizeof(uint64_t) * CPUSTATES;
984 			n = -2; /* ALL */
985 		}
986 		break;
987 	case 1:
988 		if (name[0] < 0 || name[0] >= n)
989 			return (ENOENT); /* ENOSUCHPROCESSOR */
990 		node.sysctl_size = sizeof(uint64_t) * CPUSTATES;
991 		n = name[0];
992 		/*
993 		 * adjust these so that sysctl_lookup() will be happy
994 		 */
995 		name++;
996 		namelen--;
997 		break;
998 	default:
999 		return (EINVAL);
1000 	}
1001 
1002 	cp_time = kmem_alloc(node.sysctl_size, KM_SLEEP);
1003 	node.sysctl_data = cp_time;
1004 	memset(cp_time, 0, node.sysctl_size);
1005 
1006 	for (CPU_INFO_FOREACH(cii, ci)) {
1007 		if (n <= 0) {
1008 			for (i = 0; i < CPUSTATES; i++) {
1009 				cp_time[i] += ci->ci_schedstate.spc_cp_time[i];
1010 			}
1011 		}
1012 		/*
1013 		 * if a specific processor was requested and we just
1014 		 * did it, we're done here
1015 		 */
1016 		if (n == 0)
1017 			break;
1018 		/*
1019 		 * if doing "all", skip to next cp_time set for next processor
1020 		 */
1021 		if (n == -2)
1022 			cp_time += CPUSTATES;
1023 		/*
1024 		 * if we're doing a specific processor, we're one
1025 		 * processor closer
1026 		 */
1027 		if (n > 0)
1028 			n--;
1029 	}
1030 
1031 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1032 	kmem_free(node.sysctl_data, node.sysctl_size);
1033 	return (error);
1034 }
1035 
1036 #if NPTY > 0
1037 /*
1038  * sysctl helper routine for kern.maxptys. Ensures that any new value
1039  * is acceptable to the pty subsystem.
1040  */
1041 static int
1042 sysctl_kern_maxptys(SYSCTLFN_ARGS)
1043 {
1044 	int pty_maxptys(int, int);		/* defined in kern/tty_pty.c */
1045 	int error, xmax;
1046 	struct sysctlnode node;
1047 
1048 	/* get current value of maxptys */
1049 	xmax = pty_maxptys(0, 0);
1050 
1051 	node = *rnode;
1052 	node.sysctl_data = &xmax;
1053 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1054 	if (error || newp == NULL)
1055 		return (error);
1056 
1057 	if (xmax != pty_maxptys(xmax, 1))
1058 		return (EINVAL);
1059 
1060 	return (0);
1061 }
1062 #endif /* NPTY > 0 */
1063 
1064 /*
1065  * sysctl helper routine to do kern.lwp.* work.
1066  */
1067 static int
1068 sysctl_kern_lwp(SYSCTLFN_ARGS)
1069 {
1070 	struct kinfo_lwp klwp;
1071 	struct proc *p;
1072 	struct lwp *l2, *l3;
1073 	char *where, *dp;
1074 	int pid, elem_size, elem_count;
1075 	int buflen, needed, error;
1076 	bool gotit;
1077 
1078 	if (namelen == 1 && name[0] == CTL_QUERY)
1079 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1080 
1081 	dp = where = oldp;
1082 	buflen = where != NULL ? *oldlenp : 0;
1083 	error = needed = 0;
1084 
1085 	if (newp != NULL || namelen != 3)
1086 		return (EINVAL);
1087 	pid = name[0];
1088 	elem_size = name[1];
1089 	elem_count = name[2];
1090 
1091 	sysctl_unlock();
1092 	if (pid == -1) {
1093 		mutex_enter(proc_lock);
1094 		PROCLIST_FOREACH(p, &allproc) {
1095 			/* Grab a hold on the process. */
1096 			if (!rw_tryenter(&p->p_reflock, RW_READER)) {
1097 				continue;
1098 			}
1099 			mutex_exit(proc_lock);
1100 
1101 			mutex_enter(p->p_lock);
1102 			LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1103 				if (buflen >= elem_size && elem_count > 0) {
1104 					lwp_lock(l2);
1105 					fill_lwp(l2, &klwp);
1106 					lwp_unlock(l2);
1107 					mutex_exit(p->p_lock);
1108 
1109 					/*
1110 					 * Copy out elem_size, but not
1111 					 * larger than the size of a
1112 					 * struct kinfo_proc2.
1113 					 */
1114 					error = dcopyout(l, &klwp, dp,
1115 					    min(sizeof(klwp), elem_size));
1116 					if (error) {
1117 						rw_exit(&p->p_reflock);
1118 						goto cleanup;
1119 					}
1120 					mutex_enter(p->p_lock);
1121 					LIST_FOREACH(l3, &p->p_lwps,
1122 					    l_sibling) {
1123 						if (l2 == l3)
1124 							break;
1125 					}
1126 					if (l3 == NULL) {
1127 						mutex_exit(p->p_lock);
1128 						rw_exit(&p->p_reflock);
1129 						error = EAGAIN;
1130 						goto cleanup;
1131 					}
1132 					dp += elem_size;
1133 					buflen -= elem_size;
1134 					elem_count--;
1135 				}
1136 				needed += elem_size;
1137 			}
1138 			mutex_exit(p->p_lock);
1139 
1140 			/* Drop reference to process. */
1141 			mutex_enter(proc_lock);
1142 			rw_exit(&p->p_reflock);
1143 		}
1144 		mutex_exit(proc_lock);
1145 	} else {
1146 		mutex_enter(proc_lock);
1147 		p = proc_find(pid);
1148 		if (p == NULL) {
1149 			error = ESRCH;
1150 			mutex_exit(proc_lock);
1151 			goto cleanup;
1152 		}
1153 		/* Grab a hold on the process. */
1154 		gotit = rw_tryenter(&p->p_reflock, RW_READER);
1155 		mutex_exit(proc_lock);
1156 		if (!gotit) {
1157 			error = ESRCH;
1158 			goto cleanup;
1159 		}
1160 
1161 		mutex_enter(p->p_lock);
1162 		LIST_FOREACH(l2, &p->p_lwps, l_sibling) {
1163 			if (buflen >= elem_size && elem_count > 0) {
1164 				lwp_lock(l2);
1165 				fill_lwp(l2, &klwp);
1166 				lwp_unlock(l2);
1167 				mutex_exit(p->p_lock);
1168 				/*
1169 				 * Copy out elem_size, but not larger than
1170 				 * the size of a struct kinfo_proc2.
1171 				 */
1172 				error = dcopyout(l, &klwp, dp,
1173 				    min(sizeof(klwp), elem_size));
1174 				if (error) {
1175 					rw_exit(&p->p_reflock);
1176 					goto cleanup;
1177 				}
1178 				mutex_enter(p->p_lock);
1179 				LIST_FOREACH(l3, &p->p_lwps, l_sibling) {
1180 					if (l2 == l3)
1181 						break;
1182 				}
1183 				if (l3 == NULL) {
1184 					mutex_exit(p->p_lock);
1185 					rw_exit(&p->p_reflock);
1186 					error = EAGAIN;
1187 					goto cleanup;
1188 				}
1189 				dp += elem_size;
1190 				buflen -= elem_size;
1191 				elem_count--;
1192 			}
1193 			needed += elem_size;
1194 		}
1195 		mutex_exit(p->p_lock);
1196 
1197 		/* Drop reference to process. */
1198 		rw_exit(&p->p_reflock);
1199 	}
1200 
1201 	if (where != NULL) {
1202 		*oldlenp = dp - where;
1203 		if (needed > *oldlenp) {
1204 			sysctl_relock();
1205 			return (ENOMEM);
1206 		}
1207 	} else {
1208 		needed += KERN_LWPSLOP;
1209 		*oldlenp = needed;
1210 	}
1211 	error = 0;
1212  cleanup:
1213 	sysctl_relock();
1214 	return (error);
1215 }
1216 
1217 /*
1218  * sysctl helper routine for kern.forkfsleep node. Ensures that the
1219  * given value is not too large or two small, and is at least one
1220  * timer tick if not zero.
1221  */
1222 static int
1223 sysctl_kern_forkfsleep(SYSCTLFN_ARGS)
1224 {
1225 	/* userland sees value in ms, internally is in ticks */
1226 	extern int forkfsleep;		/* defined in kern/kern_fork.c */
1227 	int error, timo, lsleep;
1228 	struct sysctlnode node;
1229 
1230 	lsleep = forkfsleep * 1000 / hz;
1231 	node = *rnode;
1232 	node.sysctl_data = &lsleep;
1233 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1234 	if (error || newp == NULL)
1235 		return (error);
1236 
1237 	/* refuse negative values, and overly 'long time' */
1238 	if (lsleep < 0 || lsleep > MAXSLP * 1000)
1239 		return (EINVAL);
1240 
1241 	timo = mstohz(lsleep);
1242 
1243 	/* if the interval is >0 ms && <1 tick, use 1 tick */
1244 	if (lsleep != 0 && timo == 0)
1245 		forkfsleep = 1;
1246 	else
1247 		forkfsleep = timo;
1248 
1249 	return (0);
1250 }
1251 
1252 /*
1253  * sysctl helper routine for kern.root_partition
1254  */
1255 static int
1256 sysctl_kern_root_partition(SYSCTLFN_ARGS)
1257 {
1258 	int rootpart = DISKPART(rootdev);
1259 	struct sysctlnode node = *rnode;
1260 
1261 	node.sysctl_data = &rootpart;
1262 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1263 }
1264 
1265 /*
1266  * sysctl helper function for kern.drivers
1267  */
1268 static int
1269 sysctl_kern_drivers(SYSCTLFN_ARGS)
1270 {
1271 	int error;
1272 	size_t buflen;
1273 	struct kinfo_drivers kd;
1274 	char *start, *where;
1275 	const char *dname;
1276 	int i;
1277 	extern struct devsw_conv *devsw_conv;
1278 	extern int max_devsw_convs;
1279 
1280 	start = where = oldp;
1281 	buflen = *oldlenp;
1282 	if (where == NULL) {
1283 		*oldlenp = max_devsw_convs * sizeof kd;
1284 		return 0;
1285 	}
1286 
1287 	/*
1288 	 * An array of kinfo_drivers structures
1289 	 */
1290 	error = 0;
1291 	sysctl_unlock();
1292 	mutex_enter(&device_lock);
1293 	for (i = 0; i < max_devsw_convs; i++) {
1294 		dname = devsw_conv[i].d_name;
1295 		if (dname == NULL)
1296 			continue;
1297 		if (buflen < sizeof kd) {
1298 			error = ENOMEM;
1299 			break;
1300 		}
1301 		memset(&kd, 0, sizeof(kd));
1302 		kd.d_bmajor = devsw_conv[i].d_bmajor;
1303 		kd.d_cmajor = devsw_conv[i].d_cmajor;
1304 		strlcpy(kd.d_name, dname, sizeof kd.d_name);
1305 		mutex_exit(&device_lock);
1306 		error = dcopyout(l, &kd, where, sizeof kd);
1307 		mutex_enter(&device_lock);
1308 		if (error != 0)
1309 			break;
1310 		buflen -= sizeof kd;
1311 		where += sizeof kd;
1312 	}
1313 	mutex_exit(&device_lock);
1314 	sysctl_relock();
1315 	*oldlenp = where - start;
1316 	return error;
1317 }
1318 
1319 static int
1320 sysctl_security_setidcore(SYSCTLFN_ARGS)
1321 {
1322 	int newsize, error;
1323 	struct sysctlnode node;
1324 
1325 	node = *rnode;
1326 	node.sysctl_data = &newsize;
1327 	newsize = *(int *)rnode->sysctl_data;
1328 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1329 	if (error || newp == NULL)
1330 		return error;
1331 
1332 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1333 	    0, NULL, NULL, NULL))
1334 		return (EPERM);
1335 
1336 	*(int *)rnode->sysctl_data = newsize;
1337 
1338 	return 0;
1339 }
1340 
1341 static int
1342 sysctl_security_setidcorename(SYSCTLFN_ARGS)
1343 {
1344 	int error;
1345 	char *newsetidcorename;
1346 	struct sysctlnode node;
1347 
1348 	newsetidcorename = PNBUF_GET();
1349 	node = *rnode;
1350 	node.sysctl_data = newsetidcorename;
1351 	memcpy(node.sysctl_data, rnode->sysctl_data, MAXPATHLEN);
1352 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1353 	if (error || newp == NULL) {
1354 		goto out;
1355 	}
1356 	if (kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_SETIDCORE,
1357 	    0, NULL, NULL, NULL)) {
1358 		error = EPERM;
1359 		goto out;
1360 	}
1361 	if (strlen(newsetidcorename) == 0) {
1362 		error = EINVAL;
1363 		goto out;
1364 	}
1365 	memcpy(rnode->sysctl_data, node.sysctl_data, MAXPATHLEN);
1366 out:
1367 	PNBUF_PUT(newsetidcorename);
1368 	return error;
1369 }
1370 
1371 /*
1372  * sysctl helper routine for kern.cp_id node. Maps cpus to their
1373  * cpuids.
1374  */
1375 static int
1376 sysctl_kern_cpid(SYSCTLFN_ARGS)
1377 {
1378 	struct sysctlnode node = *rnode;
1379 	uint64_t *cp_id = NULL;
1380 	int error, n = ncpu;
1381 	struct cpu_info *ci;
1382 	CPU_INFO_ITERATOR cii;
1383 
1384 	/*
1385 	 * Here you may either retrieve a single cpu id or the whole
1386 	 * set. The size you get back when probing depends on what
1387 	 * you ask for.
1388 	 */
1389 	switch (namelen) {
1390 	case 0:
1391 		node.sysctl_size = n * sizeof(uint64_t);
1392 		n = -2; /* ALL */
1393 		break;
1394 	case 1:
1395 		if (name[0] < 0 || name[0] >= n)
1396 			return (ENOENT); /* ENOSUCHPROCESSOR */
1397 		node.sysctl_size = sizeof(uint64_t);
1398 		n = name[0];
1399 		/*
1400 		 * adjust these so that sysctl_lookup() will be happy
1401 		 */
1402 		name++;
1403 		namelen--;
1404 		break;
1405 	default:
1406 		return (EINVAL);
1407 	}
1408 
1409 	cp_id = kmem_alloc(node.sysctl_size, KM_SLEEP);
1410 	node.sysctl_data = cp_id;
1411 	memset(cp_id, 0, node.sysctl_size);
1412 
1413 	for (CPU_INFO_FOREACH(cii, ci)) {
1414 		if (n <= 0)
1415 			cp_id[0] = cpu_index(ci);
1416 		/*
1417 		 * if a specific processor was requested and we just
1418 		 * did it, we're done here
1419 		 */
1420 		if (n == 0)
1421 			break;
1422 		/*
1423 		 * if doing "all", skip to next cp_id slot for next processor
1424 		 */
1425 		if (n == -2)
1426 			cp_id++;
1427 		/*
1428 		 * if we're doing a specific processor, we're one
1429 		 * processor closer
1430 		 */
1431 		if (n > 0)
1432 			n--;
1433 	}
1434 
1435 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1436 	kmem_free(node.sysctl_data, node.sysctl_size);
1437 	return (error);
1438 }
1439 
1440 /*
1441  * sysctl helper routine for hw.usermem and hw.usermem64. Values are
1442  * calculate on the fly taking into account integer overflow and the
1443  * current wired count.
1444  */
1445 static int
1446 sysctl_hw_usermem(SYSCTLFN_ARGS)
1447 {
1448 	u_int ui;
1449 	u_quad_t uq;
1450 	struct sysctlnode node;
1451 
1452 	node = *rnode;
1453 	switch (rnode->sysctl_num) {
1454 	case HW_USERMEM:
1455 		if ((ui = physmem - uvmexp.wired) > (UINT_MAX / PAGE_SIZE))
1456 			ui = UINT_MAX;
1457 		else
1458 			ui *= PAGE_SIZE;
1459 		node.sysctl_data = &ui;
1460 		break;
1461 	case HW_USERMEM64:
1462 		uq = (u_quad_t)(physmem - uvmexp.wired) * PAGE_SIZE;
1463 		node.sysctl_data = &uq;
1464 		break;
1465 	default:
1466 		return (EINVAL);
1467 	}
1468 
1469 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1470 }
1471 
1472 /*
1473  * sysctl helper routine for kern.cnmagic node. Pulls the old value
1474  * out, encoded, and stuffs the new value in for decoding.
1475  */
1476 static int
1477 sysctl_hw_cnmagic(SYSCTLFN_ARGS)
1478 {
1479 	char magic[CNS_LEN];
1480 	int error;
1481 	struct sysctlnode node;
1482 
1483 	if (oldp)
1484 		cn_get_magic(magic, CNS_LEN);
1485 	node = *rnode;
1486 	node.sysctl_data = &magic[0];
1487 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1488 	if (error || newp == NULL)
1489 		return (error);
1490 
1491 	return (cn_set_magic(magic));
1492 }
1493 
1494 /*
1495  * ********************************************************************
1496  * section 3: public helper routines that are used for more than one
1497  * node
1498  * ********************************************************************
1499  */
1500 
1501 /*
1502  * sysctl helper routine for the kern.root_device node and some ports'
1503  * machdep.root_device nodes.
1504  */
1505 int
1506 sysctl_root_device(SYSCTLFN_ARGS)
1507 {
1508 	struct sysctlnode node;
1509 
1510 	node = *rnode;
1511 	node.sysctl_data = __UNCONST(device_xname(root_device));
1512 	node.sysctl_size = strlen(device_xname(root_device)) + 1;
1513 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1514 }
1515 
1516 /*
1517  * sysctl helper routine for kern.consdev, dependent on the current
1518  * state of the console. Also used for machdep.console_device on some
1519  * ports.
1520  */
1521 int
1522 sysctl_consdev(SYSCTLFN_ARGS)
1523 {
1524 	dev_t consdev;
1525 	uint32_t oconsdev;
1526 	struct sysctlnode node;
1527 
1528 	if (cn_tab != NULL)
1529 		consdev = cn_tab->cn_dev;
1530 	else
1531 		consdev = NODEV;
1532 	node = *rnode;
1533 	switch (*oldlenp) {
1534 	case sizeof(consdev):
1535 		node.sysctl_data = &consdev;
1536 		node.sysctl_size = sizeof(consdev);
1537 		break;
1538 	case sizeof(oconsdev):
1539 		oconsdev = (uint32_t)consdev;
1540 		node.sysctl_data = &oconsdev;
1541 		node.sysctl_size = sizeof(oconsdev);
1542 		break;
1543 	default:
1544 		return EINVAL;
1545 	}
1546 	return (sysctl_lookup(SYSCTLFN_CALL(&node)));
1547 }
1548 
1549 /*
1550  * ********************************************************************
1551  * section 4: support for some helpers
1552  * ********************************************************************
1553  */
1554 
1555 
1556 /*
1557  * Fill in a kinfo_lwp structure for the specified lwp.
1558  */
1559 static void
1560 fill_lwp(struct lwp *l, struct kinfo_lwp *kl)
1561 {
1562 	struct proc *p = l->l_proc;
1563 	struct timeval tv;
1564 
1565 	KASSERT(lwp_locked(l, NULL));
1566 
1567 	memset(kl, 0, sizeof(*kl));
1568 
1569 	kl->l_forw = 0;
1570 	kl->l_back = 0;
1571 	kl->l_laddr = PTRTOUINT64(l);
1572 	kl->l_addr = PTRTOUINT64(l->l_addr);
1573 	kl->l_stat = l->l_stat;
1574 	kl->l_lid = l->l_lid;
1575 	kl->l_flag = L_INMEM;
1576 	kl->l_flag |= sysctl_map_flags(sysctl_lwpprflagmap, l->l_prflag);
1577 	kl->l_flag |= sysctl_map_flags(sysctl_lwpflagmap, l->l_flag);
1578 
1579 	kl->l_swtime = l->l_swtime;
1580 	kl->l_slptime = l->l_slptime;
1581 	if (l->l_stat == LSONPROC)
1582 		kl->l_schedflags = l->l_cpu->ci_schedstate.spc_flags;
1583 	else
1584 		kl->l_schedflags = 0;
1585 	kl->l_priority = lwp_eprio(l);
1586 	kl->l_usrpri = l->l_priority;
1587 	if (l->l_wchan)
1588 		strncpy(kl->l_wmesg, l->l_wmesg, sizeof(kl->l_wmesg));
1589 	kl->l_wchan = PTRTOUINT64(l->l_wchan);
1590 	kl->l_cpuid = cpu_index(l->l_cpu);
1591 	bintime2timeval(&l->l_rtime, &tv);
1592 	kl->l_rtime_sec = tv.tv_sec;
1593 	kl->l_rtime_usec = tv.tv_usec;
1594 	kl->l_cpticks = l->l_cpticks;
1595 	kl->l_pctcpu = l->l_pctcpu;
1596 	kl->l_pid = p->p_pid;
1597 	if (l->l_name == NULL)
1598 		kl->l_name[0] = '\0';
1599 	else
1600 		strlcpy(kl->l_name, l->l_name, sizeof(kl->l_name));
1601 }
1602